Merge remote branch 'origin/master' into lp-binning

Conflicts:
	src/gallium/auxiliary/util/u_surface.c
	src/gallium/drivers/llvmpipe/Makefile
	src/gallium/drivers/llvmpipe/SConscript
	src/gallium/drivers/llvmpipe/lp_bld_arit.c
	src/gallium/drivers/llvmpipe/lp_bld_flow.c
	src/gallium/drivers/llvmpipe/lp_bld_interp.c
	src/gallium/drivers/llvmpipe/lp_clear.c
	src/gallium/drivers/llvmpipe/lp_context.c
	src/gallium/drivers/llvmpipe/lp_context.h
	src/gallium/drivers/llvmpipe/lp_draw_arrays.c
	src/gallium/drivers/llvmpipe/lp_jit.c
	src/gallium/drivers/llvmpipe/lp_jit.h
	src/gallium/drivers/llvmpipe/lp_prim_vbuf.c
	src/gallium/drivers/llvmpipe/lp_setup.c
	src/gallium/drivers/llvmpipe/lp_setup_point.c
	src/gallium/drivers/llvmpipe/lp_state.h
	src/gallium/drivers/llvmpipe/lp_state_blend.c
	src/gallium/drivers/llvmpipe/lp_state_derived.c
	src/gallium/drivers/llvmpipe/lp_state_fs.c
	src/gallium/drivers/llvmpipe/lp_state_sampler.c
	src/gallium/drivers/llvmpipe/lp_state_surface.c
	src/gallium/drivers/llvmpipe/lp_tex_cache.c
	src/gallium/drivers/llvmpipe/lp_tex_cache.h
	src/gallium/drivers/llvmpipe/lp_tex_sample.h
	src/gallium/drivers/llvmpipe/lp_tile_cache.c

415 files changed, 11491 insertions, 27777 deletions

diff --git a/src/mesa/drivers/dri/Makefile b/src/mesa/drivers/dri/Makefile
index 32db097861..264648c3fb 100644
--- a/src/mesa/drivers/dri/Makefile
+++ b/src/mesa/drivers/dri/Makefile
@@ -25,7 +25,7 @@ pcedit = sed \
 	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
 	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
 	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
-	-e 's,@DRI_DRIVER_DIR@,$(DRI_DRIVER_SEARCH_DIR),' \
+	-e 's,@DRI_DRIVER_DIR@,$(DRI_DRIVER_INSTALL_DIR),' \
 	-e 's,@DRI_PC_REQ_PRIV@,$(DRI_PC_REQ_PRIV),'
 
 dri.pc: dri.pc.in
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 18dbeba24a..39d25ce3f4 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -60,18 +60,40 @@ SHARED_INCLUDES = \
 
 ##### TARGETS #####
 
-default: symlinks depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+default: subdirs lib
 
 
-$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(PIPE_DRIVERS) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
+.PHONY: lib
+lib: symlinks subdirs depend
+	@$(MAKE) $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
+
+$(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/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c
index e48e10d7c0..0e01d74265 100644
--- a/src/mesa/drivers/dri/common/dri_util.c
+++ b/src/mesa/drivers/dri/common/dri_util.c
@@ -97,7 +97,7 @@ driIntersectArea( drm_clip_rect_t rect1, drm_clip_rect_t rect2 )
  * 
  * \internal
  * This function calls __DriverAPIRec::UnbindContext, and then decrements
- * __DRIdrawablePrivateRec::refcount which must be non-zero for a successful
+ * __DRIdrawableRec::refcount which must be non-zero for a successful
  * return.
  * 
  * While casting the opaque private pointers associated with the parameters
@@ -167,11 +167,12 @@ static int driBindContext(__DRIcontext *pcp,
 			  __DRIdrawable *pdp,
 			  __DRIdrawable *prp)
 {
-    __DRIscreenPrivate *psp = pcp->driScreenPriv;
+    __DRIscreen *psp = NULL;
 
     /* Bind the drawable to the context */
 
     if (pcp) {
+	psp = pcp->driScreenPriv;
 	pcp->driDrawablePriv = pdp;
 	pcp->driReadablePriv = prp;
 	if (pdp) {
@@ -219,7 +220,7 @@ static int driBindContext(__DRIcontext *pcp,
  *
  * \param pdp pointer to the private drawable information to update.
  * 
- * This function basically updates the __DRIdrawablePrivate struct's
+ * This function basically updates the __DRIdrawable struct's
  * cliprect information by calling \c __DRIinterfaceMethods::getDrawableInfo.
  * This is usually called by the DRI_VALIDATE_DRAWABLE_INFO macro which
  * compares the __DRIdrwablePrivate pStamp and lastStamp values.  If
@@ -227,10 +228,10 @@ static int driBindContext(__DRIcontext *pcp,
  * info.
  */
 void
-__driUtilUpdateDrawableInfo(__DRIdrawablePrivate *pdp)
+__driUtilUpdateDrawableInfo(__DRIdrawable *pdp)
 {
-    __DRIscreenPrivate *psp = pdp->driScreenPriv;
-    __DRIcontextPrivate *pcp = pdp->driContextPriv;
+    __DRIscreen *psp = pdp->driScreenPriv;
+    __DRIcontext *pcp = pdp->driContextPriv;
     
     if (!pcp 
 	|| ((pdp != pcp->driDrawablePriv) && (pdp != pcp->driReadablePriv))) {
@@ -308,7 +309,7 @@ static void driReportDamage(__DRIdrawable *pdp,
  * \param drawablePrivate opaque pointer to the per-drawable private info.
  * 
  * \internal
- * This function calls __DRIdrawablePrivate::swapBuffers.
+ * This function calls __DRIdrawable::swapBuffers.
  * 
  * Is called directly from glXSwapBuffers().
  */
@@ -496,13 +497,13 @@ static void dri_get_drawable(__DRIdrawable *pdp)
 	
 static void dri_put_drawable(__DRIdrawable *pdp)
 {
-    __DRIscreenPrivate *psp;
-
-    pdp->refcount--;
-    if (pdp->refcount)
-	return;
+    __DRIscreen *psp;
 
     if (pdp) {
+	pdp->refcount--;
+	if (pdp->refcount)
+	    return;
+
 	psp = pdp->driScreenPriv;
         (*psp->DriverAPI.DestroyBuffer)(pdp);
 	if (pdp->pClipRects) {
@@ -559,7 +560,7 @@ driDestroyContext(__DRIcontext *pcp)
  *          success, or \c NULL on failure.
  * 
  * \internal
- * This function allocates and fills a __DRIcontextPrivateRec structure.  It
+ * This function allocates and fills a __DRIcontextRec structure.  It
  * performs some device independent initialization and passes all the
  * relevent information to __DriverAPIRec::CreateContext to create the
  * context.
@@ -840,7 +841,7 @@ const __DRIlegacyExtension driLegacyExtension = {
     driCreateNewContext,
 };
 
-/** Legacy DRI interface */
+/** DRI2 interface */
 const __DRIdri2Extension driDRI2Extension = {
     { __DRI_DRI2, __DRI_DRI2_VERSION },
     dri2CreateNewScreen,
@@ -848,14 +849,6 @@ const __DRIdri2Extension driDRI2Extension = {
     dri2CreateNewContext,
 };
 
-/* This is the table of extensions that the loader will dlsym() for. */
-PUBLIC const __DRIextension *__driDriverExtensions[] = {
-    &driCoreExtension.base,
-    &driLegacyExtension.base,
-    &driDRI2Extension.base,
-    NULL
-};
-
 static int
 driFrameTracking(__DRIdrawable *drawable, GLboolean enable)
 {
@@ -870,7 +863,7 @@ driQueryFrameTracking(__DRIdrawable *dpriv,
    __DRIswapInfo   sInfo;
    int             status;
    int64_t         ust;
-   __DRIscreenPrivate *psp = dpriv->driScreenPriv;
+   __DRIscreen *psp = dpriv->driScreenPriv;
 
    status = dpriv->driScreenPriv->DriverAPI.GetSwapInfo( dpriv, & sInfo );
    if ( status == 0 ) {
@@ -920,14 +913,14 @@ const __DRIframeTrackingExtension driFrameTrackingExtension = {
  *       be possible to cache the sync rate?
  */
 float
-driCalculateSwapUsage( __DRIdrawablePrivate *dPriv, int64_t last_swap_ust,
+driCalculateSwapUsage( __DRIdrawable *dPriv, int64_t last_swap_ust,
 		       int64_t current_ust )
 {
    int32_t   n;
    int32_t   d;
    int       interval;
    float     usage = 1.0;
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+   __DRIscreen *psp = dPriv->driScreenPriv;
 
    if ( (*psp->systemTime->getMSCRate)(dPriv, &n, &d, dPriv->loaderPrivate) ) {
       interval = (dPriv->swap_interval != 0) ? dPriv->swap_interval : 1;
diff --git a/src/mesa/drivers/dri/common/dri_util.h b/src/mesa/drivers/dri/common/dri_util.h
index c95a5c8299..35d8b8ff93 100644
--- a/src/mesa/drivers/dri/common/dri_util.h
+++ b/src/mesa/drivers/dri/common/dri_util.h
@@ -59,16 +59,12 @@
 
 typedef struct __DRIswapInfoRec        __DRIswapInfo;
 
-/* Typedefs to avoid rewriting the world. */
-typedef struct __DRIscreenRec	__DRIscreenPrivate;
-typedef struct __DRIdrawableRec	__DRIdrawablePrivate;
-typedef struct __DRIcontextRec	__DRIcontextPrivate;
-
 /**
  * Extensions.
  */
 extern const __DRIlegacyExtension driLegacyExtension;
 extern const __DRIcoreExtension driCoreExtension;
+extern const __DRIdri2Extension driDRI2Extension;
 extern const __DRIextension driReadDrawableExtension;
 extern const __DRIcopySubBufferExtension driCopySubBufferExtension;
 extern const __DRIswapControlExtension driSwapControlExtension;
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.c b/src/mesa/drivers/dri/common/drirenderbuffer.c
index 15af99136c..3126ea8476 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.c
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.c
@@ -1,5 +1,6 @@
 
 #include "main/mtypes.h"
+#include "main/formats.h"
 #include "main/framebuffer.h"
 #include "main/renderbuffer.h"
 #include "main/imports.h"
@@ -53,9 +54,9 @@ driDeleteRenderbuffer(struct gl_renderbuffer *rb)
  * \param pitch   pixels per row
  */
 driRenderbuffer *
-driNewRenderbuffer(GLenum format, GLvoid *addr,
+driNewRenderbuffer(gl_format format, GLvoid *addr,
                    GLint cpp, GLint offset, GLint pitch,
-                   __DRIdrawablePrivate *dPriv)
+                   __DRIdrawable *dPriv)
 {
    driRenderbuffer *drb;
 
@@ -79,56 +80,47 @@ driNewRenderbuffer(GLenum format, GLvoid *addr,
       /* Make sure we're using a null-valued GetPointer routine */
       assert(drb->Base.GetPointer(NULL, &drb->Base, 0, 0) == NULL);
 
-      drb->Base.InternalFormat = format;
-
-      if (format == GL_RGBA || format == GL_RGB5 || format == GL_RGBA8) {
-         /* Color */
-         drb->Base._BaseFormat = GL_RGBA;
-         drb->Base.DataType = GL_UNSIGNED_BYTE;
-         if (format == GL_RGB5) {
-            drb->Base.RedBits = 5;
-            drb->Base.GreenBits = 6;
-            drb->Base.BlueBits = 5;
+      switch (format) {
+      case MESA_FORMAT_ARGB8888:
+         if (cpp == 2) {
+            /* override format */
+            format = MESA_FORMAT_RGB565;
          }
-         else {
-            drb->Base.RedBits =
-            drb->Base.GreenBits =
-            drb->Base.BlueBits =
-            drb->Base.AlphaBits = 8;
-         }
-      }
-      else if (format == GL_DEPTH_COMPONENT16) {
-         /* Depth */
-         drb->Base._BaseFormat = GL_DEPTH_COMPONENT;
-         /* we always Get/Put 32-bit Z values */
-         drb->Base.DataType = GL_UNSIGNED_INT;
-         drb->Base.DepthBits = 16;
-      }
-      else if (format == GL_DEPTH_COMPONENT24) {
+         drb->Base.DataType = GL_UNSIGNED_BYTE;
+         break;
+      case MESA_FORMAT_Z16:
          /* Depth */
-         drb->Base._BaseFormat = GL_DEPTH_COMPONENT;
          /* we always Get/Put 32-bit Z values */
          drb->Base.DataType = GL_UNSIGNED_INT;
-         drb->Base.DepthBits = 24;
-      }
-      else if (format == GL_DEPTH_COMPONENT32) {
+         assert(cpp == 2);
+         break;
+      case MESA_FORMAT_Z32:
          /* Depth */
-         drb->Base._BaseFormat = GL_DEPTH_COMPONENT;
          /* we always Get/Put 32-bit Z values */
          drb->Base.DataType = GL_UNSIGNED_INT;
-         drb->Base.DepthBits = 32;
-      }
-      else {
+         assert(cpp == 4);
+         break;
+      case MESA_FORMAT_Z24_S8:
+         drb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
+         assert(cpp == 4);
+         break;
+      case MESA_FORMAT_S8_Z24:
+         drb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
+         assert(cpp == 4);
+         break;
+      case MESA_FORMAT_S8:
          /* Stencil */
-         ASSERT(format == GL_STENCIL_INDEX8_EXT);
-         drb->Base._BaseFormat = GL_STENCIL_INDEX;
          drb->Base.DataType = GL_UNSIGNED_BYTE;
-         drb->Base.StencilBits = 8;
+         break;
+      default:
+         _mesa_problem(NULL, "Bad format 0x%x in driNewRenderbuffer", format);
+         return NULL;
       }
 
-      /* XXX if we were allocating a user-created renderbuffer, we'd have
-       * to fill in the Red/Green/Blue/.../Bits values too.
-       */
+      drb->Base.Format = format;
+
+      drb->Base.InternalFormat =
+      drb->Base._BaseFormat = _mesa_get_format_base_format(format);
 
       drb->Base.AllocStorage = driRenderbufferStorage;
       drb->Base.Delete = driDeleteRenderbuffer;
@@ -204,7 +196,7 @@ driFlipRenderbuffers(struct gl_framebuffer *fb, GLboolean flipped)
  * gl_framebuffer object.
  */
 void
-driUpdateFramebufferSize(GLcontext *ctx, const __DRIdrawablePrivate *dPriv)
+driUpdateFramebufferSize(GLcontext *ctx, const __DRIdrawable *dPriv)
 {
    struct gl_framebuffer *fb = (struct gl_framebuffer *) dPriv->driverPrivate;
    if (fb && (dPriv->w != fb->Width || dPriv->h != fb->Height)) {
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.h b/src/mesa/drivers/dri/common/drirenderbuffer.h
index cf55286b30..677511334d 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.h
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.h
@@ -11,6 +11,7 @@
 #define DRIRENDERBUFFER_H
 
 #include "main/mtypes.h"
+#include "main/formats.h"
 #include "dri_util.h"
 
 
@@ -42,10 +43,10 @@ typedef struct {
    GLint flippedPitch;
    GLvoid *flippedData;  /* mmap'd address of buffer memory, if used */
 
-   /* Pointer to corresponding __DRIdrawablePrivate.  This is used to compute
+   /* Pointer to corresponding __DRIdrawable.  This is used to compute
     * the window's position within the framebuffer.
     */
-   __DRIdrawablePrivate *dPriv;
+   __DRIdrawable *dPriv;
 
    /* XXX this is for radeon/r200 only.  We should really create a new
     * r200Renderbuffer class, derived from this class...  not a huge deal.
@@ -56,23 +57,23 @@ 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;
 
 
 extern driRenderbuffer *
-driNewRenderbuffer(GLenum format, GLvoid *addr,
+driNewRenderbuffer(gl_format format, GLvoid *addr,
                    GLint cpp, GLint offset, GLint pitch,
-                   __DRIdrawablePrivate *dPriv);
+                   __DRIdrawable *dPriv);
 
 extern void
 driFlipRenderbuffers(struct gl_framebuffer *fb, GLboolean flipped);
 
 
 extern void
-driUpdateFramebufferSize(GLcontext *ctx, const __DRIdrawablePrivate *dPriv);
+driUpdateFramebufferSize(GLcontext *ctx, const __DRIdrawable *dPriv);
 
 
 #endif /* DRIRENDERBUFFER_H */
diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h
deleted file mode 100644
index 5e86324eec..0000000000
--- a/src/mesa/drivers/dri/common/extension_helper.h
+++ /dev/null
@@ -1,6647 +0,0 @@
-/* DO NOT EDIT - This file generated automatically by extension_helper.py (from Mesa) script */
-
-/*
- * (C) Copyright IBM Corporation 2005
- * 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
- * IBM,
- * 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 "utils.h"
-#include "glapi/dispatch.h"
-
-#ifndef NULL
-# define NULL 0
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char UniformMatrix3fvARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix3fv\0"
-    "glUniformMatrix3fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_multisample)
-static const char SampleCoverageARB_names[] =
-    "fi\0" /* Parameter signature */
-    "glSampleCoverage\0"
-    "glSampleCoverageARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionFilter1D_names[] =
-    "iiiiip\0" /* Parameter signature */
-    "glConvolutionFilter1D\0"
-    "glConvolutionFilter1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char BeginQueryARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glBeginQuery\0"
-    "glBeginQueryARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_NV_point_sprite)
-static const char PointParameteriNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glPointParameteri\0"
-    "glPointParameteriNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char GetProgramiv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramiv\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3sARB_names[] =
-    "iiii\0" /* Parameter signature */
-    "glMultiTexCoord3s\0"
-    "glMultiTexCoord3sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3iEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3i\0"
-    "glSecondaryColor3iEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3fMESA_names[] =
-    "fff\0" /* Parameter signature */
-    "glWindowPos3f\0"
-    "glWindowPos3fARB\0"
-    "glWindowPos3fMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char PixelTexGenParameterfvSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glPixelTexGenParameterfvSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char ActiveTextureARB_names[] =
-    "i\0" /* Parameter signature */
-    "glActiveTexture\0"
-    "glActiveTextureARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_blit)
-static const char BlitFramebufferEXT_names[] =
-    "iiiiiiiiii\0" /* Parameter signature */
-    "glBlitFramebuffer\0"
-    "glBlitFramebufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4ubvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4ubvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char GetProgramNamedParameterdvNV_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetProgramNamedParameterdvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char Histogram_names[] =
-    "iiii\0" /* Parameter signature */
-    "glHistogram\0"
-    "glHistogramEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_texture4D)
-static const char TexImage4DSGIS_names[] =
-    "iiiiiiiiiip\0" /* Parameter signature */
-    "glTexImage4DSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2dvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos2dv\0"
-    "glWindowPos2dvARB\0"
-    "glWindowPos2dvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor3fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glReplacementCodeuiColor3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_EXT_blend_equation_separate) || defined(need_GL_ATI_blend_equation_separate)
-static const char BlendEquationSeparateEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glBlendEquationSeparate\0"
-    "glBlendEquationSeparateEXT\0"
-    "glBlendEquationSeparateATI\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char ListParameterfSGIX_names[] =
-    "iif\0" /* Parameter signature */
-    "glListParameterfSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3bEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3b\0"
-    "glSecondaryColor3bEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord4fColor4fNormal3fVertex4fvSUN_names[] =
-    "pppp\0" /* Parameter signature */
-    "glTexCoord4fColor4fNormal3fVertex4fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4svNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char GetBufferSubDataARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetBufferSubData\0"
-    "glGetBufferSubDataARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char BufferSubDataARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glBufferSubData\0"
-    "glBufferSubDataARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor4ubVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glTexCoord2fColor4ubVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char AttachShader_names[] =
-    "ii\0" /* Parameter signature */
-    "glAttachShader\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2fARB_names[] =
-    "iff\0" /* Parameter signature */
-    "glVertexAttrib2f\0"
-    "glVertexAttrib2fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_shader_debug)
-static const char GetDebugLogLengthMESA_names[] =
-    "iii\0" /* Parameter signature */
-    "glGetDebugLogLengthMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3fARB_names[] =
-    "ifff\0" /* Parameter signature */
-    "glVertexAttrib3f\0"
-    "glVertexAttrib3fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char GetQueryivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetQueryiv\0"
-    "glGetQueryivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture3D)
-static const char TexImage3D_names[] =
-    "iiiiiiiiip\0" /* Parameter signature */
-    "glTexImage3D\0"
-    "glTexImage3DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiVertex3fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glReplacementCodeuiVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char GetQueryObjectivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetQueryObjectiv\0"
-    "glGetQueryObjectivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexSubImage2DARB_names[] =
-    "iiiiiiiip\0" /* Parameter signature */
-    "glCompressedTexSubImage2D\0"
-    "glCompressedTexSubImage2DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerOutputNV_names[] =
-    "iiiiiiiiii\0" /* Parameter signature */
-    "glCombinerOutputNV\0"
-    "";
-#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 */
-    "glUniform2f\0"
-    "glUniform2fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1sv\0"
-    "glVertexAttrib1svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs1dvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs1dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform2ivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform2iv\0"
-    "glUniform2ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char GetImageTransformParameterfvHP_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetImageTransformParameterfvHP\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightubvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightubvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1fvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char CopyConvolutionFilter1D_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glCopyConvolutionFilter1D\0"
-    "glCopyConvolutionFilter1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiNormal3fVertex3fSUN_names[] =
-    "iffffff\0" /* Parameter signature */
-    "glReplacementCodeuiNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char DeleteSync_names[] =
-    "i\0" /* Parameter signature */
-    "glDeleteSync\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentMaterialfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glFragmentMaterialfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_blend_color)
-static const char BlendColor_names[] =
-    "ffff\0" /* Parameter signature */
-    "glBlendColor\0"
-    "glBlendColorEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char UniformMatrix4fvARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix4fv\0"
-    "glUniformMatrix4fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_array_object) || defined(need_GL_APPLE_vertex_array_object)
-static const char DeleteVertexArraysAPPLE_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteVertexArrays\0"
-    "glDeleteVertexArraysAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char ReadInstrumentsSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glReadInstrumentsSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix2x4fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix2x4fv\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4ubVertex3fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glColor4ubVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_texture_array)
-static const char FramebufferTextureLayerEXT_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glFramebufferTextureLayer\0"
-    "glFramebufferTextureLayerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char GetListParameterfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetListParameterfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NusvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Nusv\0"
-    "glVertexAttrib4NusvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4svMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos4svMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char CreateProgramObjectARB_names[] =
-    "\0" /* Parameter signature */
-    "glCreateProgramObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightModelivSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glFragmentLightModelivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix4x3fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix4x3fv\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char PrioritizeTextures_names[] =
-    "ipp\0" /* Parameter signature */
-    "glPrioritizeTextures\0"
-    "glPrioritizeTexturesEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char AsyncMarkerSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glAsyncMarkerSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactorubSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactorubSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_shader_debug)
-static const char ClearDebugLogMESA_names[] =
-    "iii\0" /* Parameter signature */
-    "glClearDebugLogMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char ResetHistogram_names[] =
-    "i\0" /* Parameter signature */
-    "glResetHistogram\0"
-    "glResetHistogramEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char GetProgramNamedParameterfvNV_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetProgramNamedParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_point_parameters) || defined(need_GL_EXT_point_parameters) || defined(need_GL_SGIS_point_parameters)
-static const char PointParameterfEXT_names[] =
-    "if\0" /* Parameter signature */
-    "glPointParameterf\0"
-    "glPointParameterfARB\0"
-    "glPointParameterfEXT\0"
-    "glPointParameterfSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const char LoadIdentityDeformationMapSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glLoadIdentityDeformationMapSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char GenFencesNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenFencesNV\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char ImageTransformParameterfHP_names[] =
-    "iif\0" /* Parameter signature */
-    "glImageTransformParameterfHP\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const char MatrixIndexusvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMatrixIndexusvARB\0"
-    "";
-#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 */
-    "glDisableVertexAttribArray\0"
-    "glDisableVertexAttribArrayARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char StencilMaskSeparate_names[] =
-    "ii\0" /* Parameter signature */
-    "glStencilMaskSeparate\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char ProgramLocalParameter4dARB_names[] =
-    "iidddd\0" /* Parameter signature */
-    "glProgramLocalParameter4dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexImage3DARB_names[] =
-    "iiiiiiiip\0" /* Parameter signature */
-    "glCompressedTexImage3D\0"
-    "glCompressedTexImage3DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char GetConvolutionParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetConvolutionParameteriv\0"
-    "glGetConvolutionParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1fARB_names[] =
-    "if\0" /* Parameter signature */
-    "glVertexAttrib1f\0"
-    "glVertexAttrib1fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char TestFenceNV_names[] =
-    "i\0" /* Parameter signature */
-    "glTestFenceNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord1fv\0"
-    "glMultiTexCoord1fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char ColorFragmentOp2ATI_names[] =
-    "iiiiiiiiii\0" /* Parameter signature */
-    "glColorFragmentOp2ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char SecondaryColorPointerListIBM_names[] =
-    "iiipi\0" /* Parameter signature */
-    "glSecondaryColorPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char GetPixelTexGenParameterivSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetPixelTexGenParameterivSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4fNV_names[] =
-    "iffff\0" /* Parameter signature */
-    "glVertexAttrib4fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeubSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glReplacementCodeubSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char FinishAsyncSGIX_names[] =
-    "p\0" /* Parameter signature */
-    "glFinishAsyncSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_shader_debug)
-static const char GetDebugLogMESA_names[] =
-    "iiiipp\0" /* Parameter signature */
-    "glGetDebugLogMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
-static const char FogCoorddEXT_names[] =
-    "d\0" /* Parameter signature */
-    "glFogCoordd\0"
-    "glFogCoorddEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4ubVertex3fSUN_names[] =
-    "iiiifff\0" /* Parameter signature */
-    "glColor4ubVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
-static const char FogCoordfEXT_names[] =
-    "f\0" /* Parameter signature */
-    "glFogCoordf\0"
-    "glFogCoordfEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fVertex3fSUN_names[] =
-    "fffff\0" /* Parameter signature */
-    "glTexCoord2fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactoriSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactoriSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2dNV_names[] =
-    "idd\0" /* Parameter signature */
-    "glVertexAttrib2dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char GetProgramInfoLog_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetProgramInfoLog\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NbvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Nbv\0"
-    "glVertexAttrib4NbvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_shader)
-static const char GetActiveAttribARB_names[] =
-    "iiipppp\0" /* Parameter signature */
-    "glGetActiveAttrib\0"
-    "glGetActiveAttribARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4ubNV_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glVertexAttrib4ubNV\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_texture_range)
-static const char TextureRangeAPPLE_names[] =
-    "iip\0" /* Parameter signature */
-    "glTextureRangeAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor4fNormal3fVertex3fSUN_names[] =
-    "ffffffffffff\0" /* Parameter signature */
-    "glTexCoord2fColor4fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerParameterfvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glCombinerParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs3dvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs3dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs4fvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs4fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_array_range)
-static const char VertexArrayRangeNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexArrayRangeNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightiSGIX_names[] =
-    "iii\0" /* Parameter signature */
-    "glFragmentLightiSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_polygon_offset)
-static const char PolygonOffsetEXT_names[] =
-    "ff\0" /* Parameter signature */
-    "glPolygonOffsetEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char PollAsyncSGIX_names[] =
-    "p\0" /* Parameter signature */
-    "glPollAsyncSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char DeleteFragmentShaderATI_names[] =
-    "i\0" /* Parameter signature */
-    "glDeleteFragmentShaderATI\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fNormal3fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glTexCoord2fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_transpose_matrix)
-static const char MultTransposeMatrixdARB_names[] =
-    "p\0" /* Parameter signature */
-    "glMultTransposeMatrixd\0"
-    "glMultTransposeMatrixdARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2svMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos2sv\0"
-    "glWindowPos2svARB\0"
-    "glWindowPos2svMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexImage1DARB_names[] =
-    "iiiiiip\0" /* Parameter signature */
-    "glCompressedTexImage1D\0"
-    "glCompressedTexImage1DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2sNV_names[] =
-    "iii\0" /* Parameter signature */
-    "glVertexAttrib2sNV\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char NormalPointerListIBM_names[] =
-    "iipi\0" /* Parameter signature */
-    "glNormalPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char IndexPointerEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glIndexPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char NormalPointerEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glNormalPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3dARB_names[] =
-    "iddd\0" /* Parameter signature */
-    "glMultiTexCoord3d\0"
-    "glMultiTexCoord3dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2iARB_names[] =
-    "iii\0" /* Parameter signature */
-    "glMultiTexCoord2i\0"
-    "glMultiTexCoord2iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fNormal3fVertex3fSUN_names[] =
-    "iffffffff\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord2sv\0"
-    "glMultiTexCoord2svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeubvSUN_names[] =
-    "p\0" /* Parameter signature */
-    "glReplacementCodeubvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform3iARB_names[] =
-    "iiii\0" /* Parameter signature */
-    "glUniform3i\0"
-    "glUniform3iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char GetFragmentMaterialfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFragmentMaterialfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char GetShaderInfoLog_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetShaderInfoLog\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char PollInstrumentsSGIX_names[] =
-    "p\0" /* Parameter signature */
-    "glPollInstrumentsSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactordSUN_names[] =
-    "d\0" /* Parameter signature */
-    "glGlobalAlphaFactordSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetFinalCombinerInputParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFinalCombinerInputParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char GenerateMipmapEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glGenerateMipmap\0"
-    "glGenerateMipmapEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char SetFragmentShaderConstantATI_names[] =
-    "ip\0" /* Parameter signature */
-    "glSetFragmentShaderConstantATI\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char GetMapAttribParameterivNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetMapAttribParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char CreateShaderObjectARB_names[] =
-    "i\0" /* Parameter signature */
-    "glCreateShaderObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_sharpen_texture)
-static const char GetSharpenTexFuncSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetSharpenTexFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char BufferDataARB_names[] =
-    "iipi\0" /* Parameter signature */
-    "glBufferData\0"
-    "glBufferDataARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_array_range)
-static const char FlushVertexArrayRangeNV_names[] =
-    "\0" /* Parameter signature */
-    "glFlushVertexArrayRangeNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char SampleMapATI_names[] =
-    "iii\0" /* Parameter signature */
-    "glSampleMapATI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char VertexPointerEXT_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glVertexPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_texture_filter4)
-static const char GetTexFilterFuncSGIS_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetTexFilterFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetCombinerOutputParameterfvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetCombinerOutputParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_subtexture)
-static const char TexSubImage1D_names[] =
-    "iiiiiip\0" /* Parameter signature */
-    "glTexSubImage1D\0"
-    "glTexSubImage1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1sARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glVertexAttrib1s\0"
-    "glVertexAttrib1sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char FenceSync_names[] =
-    "ii\0" /* Parameter signature */
-    "glFenceSync\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char FinalCombinerInputNV_names[] =
-    "iiii\0" /* Parameter signature */
-    "glFinalCombinerInputNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_flush_raster)
-static const char FlushRasterSGIX_names[] =
-    "\0" /* Parameter signature */
-    "glFlushRasterSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fVertex3fSUN_names[] =
-    "ifffff\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform1fARB_names[] =
-    "if\0" /* Parameter signature */
-    "glUniform1f\0"
-    "glUniform1fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char AreTexturesResident_names[] =
-    "ipp\0" /* Parameter signature */
-    "glAreTexturesResident\0"
-    "glAreTexturesResidentEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ATI_separate_stencil)
-static const char StencilOpSeparate_names[] =
-    "iiii\0" /* Parameter signature */
-    "glStencilOpSeparate\0"
-    "glStencilOpSeparateATI\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table)
-static const char ColorTableParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glColorTableParameteriv\0"
-    "glColorTableParameterivSGI\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char FogCoordPointerListIBM_names[] =
-    "iipi\0" /* Parameter signature */
-    "glFogCoordPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3dMESA_names[] =
-    "ddd\0" /* Parameter signature */
-    "glWindowPos3d\0"
-    "glWindowPos3dARB\0"
-    "glWindowPos3dMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_point_parameters) || defined(need_GL_EXT_point_parameters) || defined(need_GL_SGIS_point_parameters)
-static const char PointParameterfvEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glPointParameterfv\0"
-    "glPointParameterfvARB\0"
-    "glPointParameterfvEXT\0"
-    "glPointParameterfvSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2fvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos2fv\0"
-    "glWindowPos2fvARB\0"
-    "glWindowPos2fvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3bvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3bv\0"
-    "glSecondaryColor3bvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char VertexPointerListIBM_names[] =
-    "iiipi\0" /* Parameter signature */
-    "glVertexPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramLocalParameterfvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramLocalParameterfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentMaterialfSGIX_names[] =
-    "iif\0" /* Parameter signature */
-    "glFragmentMaterialfSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fNormal3fVertex3fSUN_names[] =
-    "ffffffff\0" /* Parameter signature */
-    "glTexCoord2fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char RenderbufferStorageEXT_names[] =
-    "iiii\0" /* Parameter signature */
-    "glRenderbufferStorage\0"
-    "glRenderbufferStorageEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char IsFenceNV_names[] =
-    "i\0" /* Parameter signature */
-    "glIsFenceNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char AttachObjectARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glAttachObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char GetFragmentLightivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFragmentLightivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char UniformMatrix2fvARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix2fv\0"
-    "glUniformMatrix2fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2fARB_names[] =
-    "iff\0" /* Parameter signature */
-    "glMultiTexCoord2f\0"
-    "glMultiTexCoord2fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table) || defined(need_GL_EXT_paletted_texture)
-static const char ColorTable_names[] =
-    "iiiiip\0" /* Parameter signature */
-    "glColorTable\0"
-    "glColorTableSGI\0"
-    "glColorTableEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char MapControlPointsNV_names[] =
-    "iiiiiiiip\0" /* Parameter signature */
-    "glMapControlPointsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionFilter2D_names[] =
-    "iiiiiip\0" /* Parameter signature */
-    "glConvolutionFilter2D\0"
-    "glConvolutionFilter2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char MapParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glMapParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3dv\0"
-    "glVertexAttrib3dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_PGI_misc_hints)
-static const char HintPGI_names[] =
-    "ii\0" /* Parameter signature */
-    "glHintPGI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glConvolutionParameteriv\0"
-    "glConvolutionParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_cull_vertex)
-static const char CullParameterdvEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glCullParameterdvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char ProgramNamedParameter4fNV_names[] =
-    "iipffff\0" /* Parameter signature */
-    "glProgramNamedParameter4fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color3fVertex3fSUN_names[] =
-    "ffffff\0" /* Parameter signature */
-    "glColor3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char ProgramEnvParameter4fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glProgramEnvParameter4fvARB\0"
-    "glProgramParameter4fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightModeliSGIX_names[] =
-    "ii\0" /* Parameter signature */
-    "glFragmentLightModeliSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glConvolutionParameterfv\0"
-    "glConvolutionParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_3DFX_tbuffer)
-static const char TbufferMask3DFX_names[] =
-    "i\0" /* Parameter signature */
-    "glTbufferMask3DFX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char LoadProgramNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glLoadProgramNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char WaitSync_names[] =
-    "iii\0" /* Parameter signature */
-    "glWaitSync\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4fvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char GetAttachedObjectsARB_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetAttachedObjectsARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform3fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform3fv\0"
-    "glUniform3fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_draw_range_elements)
-static const char DrawRangeElements_names[] =
-    "iiiiip\0" /* Parameter signature */
-    "glDrawRangeElements\0"
-    "glDrawRangeElementsEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_sprite)
-static const char SpriteParameterfvSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glSpriteParameterfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char CheckFramebufferStatusEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glCheckFramebufferStatus\0"
-    "glCheckFramebufferStatusEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactoruiSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactoruiSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char GetHandleARB_names[] =
-    "i\0" /* Parameter signature */
-    "glGetHandleARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char GetVertexAttribivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribiv\0"
-    "glGetVertexAttribivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetCombinerInputParameterfvNV_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glGetCombinerInputParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char CreateProgram_names[] =
-    "\0" /* Parameter signature */
-    "glCreateProgram\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_transpose_matrix)
-static const char LoadTransposeMatrixdARB_names[] =
-    "p\0" /* Parameter signature */
-    "glLoadTransposeMatrixd\0"
-    "glLoadTransposeMatrixdARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetMinmax_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glGetMinmax\0"
-    "glGetMinmaxEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char StencilFuncSeparate_names[] =
-    "iiii\0" /* Parameter signature */
-    "glStencilFuncSeparate\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3sEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3s\0"
-    "glSecondaryColor3sEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color3fVertex3fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glColor3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactorbSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactorbSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char ImageTransformParameterfvHP_names[] =
-    "iip\0" /* Parameter signature */
-    "glImageTransformParameterfvHP\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4ivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4iv\0"
-    "glVertexAttrib4ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3fNV_names[] =
-    "ifff\0" /* Parameter signature */
-    "glVertexAttrib3fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs2dvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs2dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_timer_query)
-static const char GetQueryObjectui64vEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetQueryObjectui64vEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord3fv\0"
-    "glMultiTexCoord3fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3dEXT_names[] =
-    "ddd\0" /* Parameter signature */
-    "glSecondaryColor3d\0"
-    "glSecondaryColor3dEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetProgramParameterfvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetProgramParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char TangentPointerEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glTangentPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4fNormal3fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glColor4fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char GetInstrumentsSGIX_names[] =
-    "\0" /* Parameter signature */
-    "glGetInstrumentsSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char EvalMapsNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glEvalMapsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_subtexture)
-static const char TexSubImage2D_names[] =
-    "iiiiiiiip\0" /* Parameter signature */
-    "glTexSubImage2D\0"
-    "glTexSubImage2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glFragmentLightivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_texture_range)
-static const char GetTexParameterPointervAPPLE_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetTexParameterPointervAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_pixel_transform)
-static const char PixelTransformParameterfvEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glPixelTransformParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4bvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4bv\0"
-    "glVertexAttrib4bvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char AlphaFragmentOp2ATI_names[] =
-    "iiiiiiiii\0" /* Parameter signature */
-    "glAlphaFragmentOp2ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4sARB_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glMultiTexCoord4s\0"
-    "glMultiTexCoord4sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char GetFragmentMaterialivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFragmentMaterialivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4dMESA_names[] =
-    "dddd\0" /* Parameter signature */
-    "glWindowPos4dMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightPointerARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glWeightPointerARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2dMESA_names[] =
-    "dd\0" /* Parameter signature */
-    "glWindowPos2d\0"
-    "glWindowPos2dARB\0"
-    "glWindowPos2dMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char FramebufferTexture3DEXT_names[] =
-    "iiiiii\0" /* Parameter signature */
-    "glFramebufferTexture3D\0"
-    "glFramebufferTexture3DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_blend_minmax)
-static const char BlendEquation_names[] =
-    "i\0" /* Parameter signature */
-    "glBlendEquation\0"
-    "glBlendEquationEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3dNV_names[] =
-    "iddd\0" /* Parameter signature */
-    "glVertexAttrib3dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3dARB_names[] =
-    "iddd\0" /* Parameter signature */
-    "glVertexAttrib3d\0"
-    "glVertexAttrib3dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN_names[] =
-    "ppppp\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4fARB_names[] =
-    "iffff\0" /* Parameter signature */
-    "glVertexAttrib4f\0"
-    "glVertexAttrib4fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_index_func)
-static const char IndexFuncEXT_names[] =
-    "if\0" /* Parameter signature */
-    "glIndexFuncEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char FramebufferTexture2DEXT_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glFramebufferTexture2D\0"
-    "glFramebufferTexture2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord2dv\0"
-    "glMultiTexCoord2dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_cull_vertex)
-static const char CullParameterfvEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glCullParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char ProgramNamedParameter4fvNV_names[] =
-    "iipp\0" /* Parameter signature */
-    "glProgramNamedParameter4fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColorPointerEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glSecondaryColorPointer\0"
-    "glSecondaryColorPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4fv\0"
-    "glVertexAttrib4fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char ColorPointerListIBM_names[] =
-    "iiipi\0" /* Parameter signature */
-    "glColorPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char GetActiveUniformARB_names[] =
-    "iiipppp\0" /* Parameter signature */
-    "glGetActiveUniform\0"
-    "glGetActiveUniformARB\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char ImageTransformParameteriHP_names[] =
-    "iii\0" /* Parameter signature */
-    "glImageTransformParameteriHP\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord1sv\0"
-    "glMultiTexCoord1svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char EndQueryARB_names[] =
-    "i\0" /* Parameter signature */
-    "glEndQuery\0"
-    "glEndQueryARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char DeleteFencesNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteFencesNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const char DeformationMap3dSGIX_names[] =
-    "iddiiddiiddiip\0" /* Parameter signature */
-    "glDeformationMap3dSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char IsShader_names[] =
-    "i\0" /* Parameter signature */
-    "glIsShader\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char GetImageTransformParameterivHP_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetImageTransformParameterivHP\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4ivMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos4ivMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord3sv\0"
-    "glMultiTexCoord3svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4iARB_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glMultiTexCoord4i\0"
-    "glMultiTexCoord4iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3ivEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glBinormal3ivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_resize_buffers)
-static const char ResizeBuffersMESA_names[] =
-    "\0" /* Parameter signature */
-    "glResizeBuffersMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char GetUniformivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetUniformiv\0"
-    "glGetUniformivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char PixelTexGenParameteriSGIS_names[] =
-    "ii\0" /* Parameter signature */
-    "glPixelTexGenParameteriSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_INTEL_parallel_arrays)
-static const char VertexPointervINTEL_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexPointervINTEL\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor4fNormal3fVertex3fvSUN_names[] =
-    "pppp\0" /* Parameter signature */
-    "glReplacementCodeuiColor4fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3uiEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3ui\0"
-    "glSecondaryColor3uiEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char StartInstrumentsSGIX_names[] =
-    "\0" /* Parameter signature */
-    "glStartInstrumentsSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3usvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3usv\0"
-    "glSecondaryColor3usvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2fvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char ProgramLocalParameter4dvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glProgramLocalParameter4dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const char MatrixIndexuivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMatrixIndexuivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object)
-static const char RenderbufferStorageMultisample_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glRenderbufferStorageMultisample\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3sEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glTangent3sEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactorfSUN_names[] =
-    "f\0" /* Parameter signature */
-    "glGlobalAlphaFactorfSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3iARB_names[] =
-    "iiii\0" /* Parameter signature */
-    "glMultiTexCoord3i\0"
-    "glMultiTexCoord3iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char IsProgram_names[] =
-    "i\0" /* Parameter signature */
-    "glIsProgram\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char TexCoordPointerListIBM_names[] =
-    "iiipi\0" /* Parameter signature */
-    "glTexCoordPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactorusSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactorusSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2dvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char FramebufferRenderbufferEXT_names[] =
-    "iiii\0" /* Parameter signature */
-    "glFramebufferRenderbuffer\0"
-    "glFramebufferRenderbufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1dvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char GenTextures_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenTextures\0"
-    "glGenTexturesEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char SetFenceNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glSetFenceNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char FramebufferTexture1DEXT_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glFramebufferTexture1D\0"
-    "glFramebufferTexture1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetCombinerOutputParameterivNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetCombinerOutputParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char PixelTexGenParameterivSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glPixelTexGenParameterivSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_perturb_normal)
-static const char TextureNormalEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glTextureNormalEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char IndexPointerListIBM_names[] =
-    "iipi\0" /* Parameter signature */
-    "glIndexPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightfvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4fMESA_names[] =
-    "ffff\0" /* Parameter signature */
-    "glWindowPos4fMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3dvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos3dv\0"
-    "glWindowPos3dvARB\0"
-    "glWindowPos3dvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_timer_query)
-static const char GetQueryObjecti64vEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetQueryObjecti64vEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1dARB_names[] =
-    "id\0" /* Parameter signature */
-    "glMultiTexCoord1d\0"
-    "glMultiTexCoord1dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_NV_point_sprite)
-static const char PointParameterivNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glPointParameteriv\0"
-    "glPointParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform2fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform2fv\0"
-    "glUniform2fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_flush_buffer_range)
-static const char BufferParameteriAPPLE_names[] =
-    "iii\0" /* Parameter signature */
-    "glBufferParameteriAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord3dv\0"
-    "glMultiTexCoord3dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN_names[] =
-    "pppp\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char DeleteObjectARB_names[] =
-    "i\0" /* Parameter signature */
-    "glDeleteObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const char MatrixIndexPointerARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glMatrixIndexPointerARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char ProgramNamedParameter4dvNV_names[] =
-    "iipp\0" /* Parameter signature */
-    "glProgramNamedParameter4dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3fvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glTangent3fvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_array_object)
-static const char GenVertexArrays_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenVertexArrays\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char BindFramebufferEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glBindFramebuffer\0"
-    "glBindFramebufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_reference_plane)
-static const char ReferencePlaneSGIX_names[] =
-    "p\0" /* Parameter signature */
-    "glReferencePlaneSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char ValidateProgramARB_names[] =
-    "i\0" /* Parameter signature */
-    "glValidateProgram\0"
-    "glValidateProgramARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_compiled_vertex_array)
-static const char UnlockArraysEXT_names[] =
-    "\0" /* Parameter signature */
-    "glUnlockArraysEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor3fVertex3fSUN_names[] =
-    "ffffffff\0" /* Parameter signature */
-    "glTexCoord2fColor3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3fvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos3fv\0"
-    "glWindowPos3fvARB\0"
-    "glWindowPos3fvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1svNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const char CopyTexSubImage3D_names[] =
-    "iiiiiiiii\0" /* Parameter signature */
-    "glCopyTexSubImage3D\0"
-    "glCopyTexSubImage3DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2dARB_names[] =
-    "idd\0" /* Parameter signature */
-    "glVertexAttrib2d\0"
-    "glVertexAttrib2dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char GetInteger64v_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetInteger64v\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_texture_color_mask)
-static const char TextureColorMaskSGIS_names[] =
-    "iiii\0" /* Parameter signature */
-    "glTextureColorMaskSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table) || defined(need_GL_EXT_paletted_texture)
-static const char GetColorTable_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetColorTable\0"
-    "glGetColorTableSGI\0"
-    "glGetColorTableEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table)
-static const char CopyColorTable_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glCopyColorTable\0"
-    "glCopyColorTableSGI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetHistogramParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetHistogramParameterfv\0"
-    "glGetHistogramParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_INTEL_parallel_arrays)
-static const char ColorPointervINTEL_names[] =
-    "iip\0" /* Parameter signature */
-    "glColorPointervINTEL\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char AlphaFragmentOp1ATI_names[] =
-    "iiiiii\0" /* Parameter signature */
-    "glAlphaFragmentOp1ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3ivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord3iv\0"
-    "glMultiTexCoord3ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2sARB_names[] =
-    "iii\0" /* Parameter signature */
-    "glMultiTexCoord2s\0"
-    "glMultiTexCoord2sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1dv\0"
-    "glVertexAttrib1dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char DeleteTextures_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteTextures\0"
-    "glDeleteTexturesEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char TexCoordPointerEXT_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glTexCoordPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_texture4D)
-static const char TexSubImage4DSGIS_names[] =
-    "iiiiiiiiiiiip\0" /* Parameter signature */
-    "glTexSubImage4DSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners2)
-static const char CombinerStageParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glCombinerStageParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char StopInstrumentsSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glStopInstrumentsSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord4fColor4fNormal3fVertex4fSUN_names[] =
-    "fffffffffffffff\0" /* Parameter signature */
-    "glTexCoord4fColor4fNormal3fVertex4fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const char DeformSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glDeformSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char GetVertexAttribfvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribfv\0"
-    "glGetVertexAttribfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3ivEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3iv\0"
-    "glSecondaryColor3ivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix4x2fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix4x2fv\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_detail_texture)
-static const char GetDetailTexFuncSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetDetailTexFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners2)
-static const char GetCombinerStageParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetCombinerStageParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_array_object)
-static const char BindVertexArray_names[] =
-    "i\0" /* Parameter signature */
-    "glBindVertexArray\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4ubVertex2fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glColor4ubVertex2fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_texture_filter4)
-static const char TexFilterFuncSGIS_names[] =
-    "iiip\0" /* Parameter signature */
-    "glTexFilterFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_multisample) || defined(need_GL_EXT_multisample)
-static const char SampleMaskSGIS_names[] =
-    "fi\0" /* Parameter signature */
-    "glSampleMaskSGIS\0"
-    "glSampleMaskEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_shader)
-static const char GetAttribLocationARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetAttribLocation\0"
-    "glGetAttribLocationARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4ubvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4ubv\0"
-    "glVertexAttrib4ubvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_detail_texture)
-static const char DetailTexFuncSGIS_names[] =
-    "iip\0" /* Parameter signature */
-    "glDetailTexFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Normal3fVertex3fSUN_names[] =
-    "ffffff\0" /* Parameter signature */
-    "glNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const char CopyTexImage2D_names[] =
-    "iiiiiiii\0" /* Parameter signature */
-    "glCopyTexImage2D\0"
-    "glCopyTexImage2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char GetBufferPointervARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetBufferPointerv\0"
-    "glGetBufferPointervARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char ProgramEnvParameter4fARB_names[] =
-    "iiffff\0" /* Parameter signature */
-    "glProgramEnvParameter4fARB\0"
-    "glProgramParameter4fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform3ivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform3iv\0"
-    "glUniform3ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char GetFenceivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFenceivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4dvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos4dvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_color_subtable)
-static const char ColorSubTable_names[] =
-    "iiiiip\0" /* Parameter signature */
-    "glColorSubTable\0"
-    "glColorSubTableEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4ivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord4iv\0"
-    "glMultiTexCoord4ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_gpu_program_parameters)
-static const char ProgramLocalParameters4fvEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glProgramLocalParameters4fvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char GetMapAttribParameterfvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetMapAttribParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4sARB_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glVertexAttrib4s\0"
-    "glVertexAttrib4sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char GetQueryObjectuivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetQueryObjectuiv\0"
-    "glGetQueryObjectuivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char MapParameterivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glMapParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char GenRenderbuffersEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenRenderbuffers\0"
-    "glGenRenderbuffersEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2dv\0"
-    "glVertexAttrib2dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char EdgeFlagPointerEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glEdgeFlagPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs2svNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs2svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightbvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightbvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2fv\0"
-    "glVertexAttrib2fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char GetBufferParameterivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetBufferParameteriv\0"
-    "glGetBufferParameterivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char ListParameteriSGIX_names[] =
-    "iii\0" /* Parameter signature */
-    "glListParameteriSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor4fNormal3fVertex3fSUN_names[] =
-    "iffffffffff\0" /* Parameter signature */
-    "glReplacementCodeuiColor4fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const char InstrumentsBufferSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glInstrumentsBufferSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Niv\0"
-    "glVertexAttrib4NivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char GetAttachedShaders_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetAttachedShaders\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_vertex_array_object)
-static const char GenVertexArraysAPPLE_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenVertexArraysAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_gpu_program_parameters)
-static const char ProgramEnvParameters4fvEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glProgramEnvParameters4fvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor4fNormal3fVertex3fvSUN_names[] =
-    "pppp\0" /* Parameter signature */
-    "glTexCoord2fColor4fNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2iMESA_names[] =
-    "ii\0" /* Parameter signature */
-    "glWindowPos2i\0"
-    "glWindowPos2iARB\0"
-    "glWindowPos2iMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3fvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3fv\0"
-    "glSecondaryColor3fvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexSubImage1DARB_names[] =
-    "iiiiiip\0" /* Parameter signature */
-    "glCompressedTexSubImage1D\0"
-    "glCompressedTexSubImage1DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetVertexAttribivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramStringARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramStringARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_envmap_bumpmap)
-static const char TexBumpParameterfvATI_names[] =
-    "ip\0" /* Parameter signature */
-    "glTexBumpParameterfvATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char CompileShaderARB_names[] =
-    "i\0" /* Parameter signature */
-    "glCompileShader\0"
-    "glCompileShaderARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char DeleteShader_names[] =
-    "i\0" /* Parameter signature */
-    "glDeleteShader\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform3fARB_names[] =
-    "ifff\0" /* Parameter signature */
-    "glUniform3f\0"
-    "glUniform3fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char ListParameterfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glListParameterfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3dvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glTangent3dvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetVertexAttribfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3sMESA_names[] =
-    "iii\0" /* Parameter signature */
-    "glWindowPos3s\0"
-    "glWindowPos3sARB\0"
-    "glWindowPos3sMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2svNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs1fvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs1fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fVertex3fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glTexCoord2fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4sMESA_names[] =
-    "iiii\0" /* Parameter signature */
-    "glWindowPos4sMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NuivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Nuiv\0"
-    "glVertexAttrib4NuivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char ClientActiveTextureARB_names[] =
-    "i\0" /* Parameter signature */
-    "glClientActiveTexture\0"
-    "glClientActiveTextureARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_pixel_texture)
-static const char PixelTexGenSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glPixelTexGenSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeusvSUN_names[] =
-    "p\0" /* Parameter signature */
-    "glReplacementCodeusvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform4fARB_names[] =
-    "iffff\0" /* Parameter signature */
-    "glUniform4f\0"
-    "glUniform4fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_map_buffer_range)
-static const char FlushMappedBufferRange_names[] =
-    "iii\0" /* Parameter signature */
-    "glFlushMappedBufferRange\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char IsProgramNV_names[] =
-    "i\0" /* Parameter signature */
-    "glIsProgramARB\0"
-    "glIsProgramNV\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_flush_buffer_range)
-static const char FlushMappedBufferRangeAPPLE_names[] =
-    "iii\0" /* Parameter signature */
-    "glFlushMappedBufferRangeAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodePointerSUN_names[] =
-    "iip\0" /* Parameter signature */
-    "glReplacementCodePointerSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char ProgramEnvParameter4dARB_names[] =
-    "iidddd\0" /* Parameter signature */
-    "glProgramEnvParameter4dARB\0"
-    "glProgramParameter4dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table)
-static const char ColorTableParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glColorTableParameterfv\0"
-    "glColorTableParameterfvSGI\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightModelfSGIX_names[] =
-    "if\0" /* Parameter signature */
-    "glFragmentLightModelfSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3bvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glBinormal3bvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char IsTexture_names[] =
-    "i\0" /* Parameter signature */
-    "glIsTexture\0"
-    "glIsTextureEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_weighting)
-static const char VertexWeightfvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glVertexWeightfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1dARB_names[] =
-    "id\0" /* Parameter signature */
-    "glVertexAttrib1d\0"
-    "glVertexAttrib1dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const char ImageTransformParameterivHP_names[] =
-    "iip\0" /* Parameter signature */
-    "glImageTransformParameterivHP\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char DeleteQueriesARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteQueries\0"
-    "glDeleteQueriesARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4ubVertex2fSUN_names[] =
-    "iiiiff\0" /* Parameter signature */
-    "glColor4ubVertex2fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentColorMaterialSGIX_names[] =
-    "ii\0" /* Parameter signature */
-    "glFragmentColorMaterialSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const char CurrentPaletteMatrixARB_names[] =
-    "i\0" /* Parameter signature */
-    "glCurrentPaletteMatrixARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_multisample) || defined(need_GL_EXT_multisample)
-static const char SamplePatternSGIS_names[] =
-    "i\0" /* Parameter signature */
-    "glSamplePatternSGIS\0"
-    "glSamplePatternEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char IsQueryARB_names[] =
-    "i\0" /* Parameter signature */
-    "glIsQuery\0"
-    "glIsQueryARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor4ubVertex3fSUN_names[] =
-    "iiiiifff\0" /* Parameter signature */
-    "glReplacementCodeuiColor4ubVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4usvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4usv\0"
-    "glVertexAttrib4usvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char LinkProgramARB_names[] =
-    "i\0" /* Parameter signature */
-    "glLinkProgram\0"
-    "glLinkProgramARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib2fNV_names[] =
-    "iff\0" /* Parameter signature */
-    "glVertexAttrib2fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char ShaderSourceARB_names[] =
-    "iipp\0" /* Parameter signature */
-    "glShaderSource\0"
-    "glShaderSourceARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentMaterialiSGIX_names[] =
-    "iii\0" /* Parameter signature */
-    "glFragmentMaterialiSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3sv\0"
-    "glVertexAttrib3svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexSubImage3DARB_names[] =
-    "iiiiiiiiiip\0" /* Parameter signature */
-    "glCompressedTexSubImage3D\0"
-    "glCompressedTexSubImage3DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2ivMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos2iv\0"
-    "glWindowPos2ivARB\0"
-    "glWindowPos2ivMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char IsFramebufferEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glIsFramebuffer\0"
-    "glIsFramebufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform4ivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform4iv\0"
-    "glUniform4ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char GetVertexAttribdvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribdv\0"
-    "glGetVertexAttribdvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_envmap_bumpmap)
-static const char TexBumpParameterivATI_names[] =
-    "ip\0" /* Parameter signature */
-    "glTexBumpParameterivATI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char GetSeparableFilter_names[] =
-    "iiippp\0" /* Parameter signature */
-    "glGetSeparableFilter\0"
-    "glGetSeparableFilterEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3dEXT_names[] =
-    "ddd\0" /* Parameter signature */
-    "glBinormal3dEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_sprite)
-static const char SpriteParameteriSGIX_names[] =
-    "ii\0" /* Parameter signature */
-    "glSpriteParameteriSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char RequestResidentProgramsNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glRequestResidentProgramsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_tag_sample_buffer)
-static const char TagSampleBufferSGIX_names[] =
-    "\0" /* Parameter signature */
-    "glTagSampleBufferSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeusSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glReplacementCodeusSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char ListParameterivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glListParameterivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_multi_draw_arrays)
-static const char MultiDrawElementsEXT_names[] =
-    "ipipi\0" /* Parameter signature */
-    "glMultiDrawElements\0"
-    "glMultiDrawElementsEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform1ivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform1iv\0"
-    "glUniform1ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2sMESA_names[] =
-    "ii\0" /* Parameter signature */
-    "glWindowPos2s\0"
-    "glWindowPos2sARB\0"
-    "glWindowPos2sMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightusvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightusvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
-static const char FogCoordPointerEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glFogCoordPointer\0"
-    "glFogCoordPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_index_material)
-static const char IndexMaterialEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glIndexMaterialEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3ubvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3ubv\0"
-    "glSecondaryColor3ubvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4dv\0"
-    "glVertexAttrib4dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_shader)
-static const char BindAttribLocationARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glBindAttribLocation\0"
-    "glBindAttribLocationARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2dARB_names[] =
-    "idd\0" /* Parameter signature */
-    "glMultiTexCoord2d\0"
-    "glMultiTexCoord2dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char ExecuteProgramNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glExecuteProgramNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char LightEnviSGIX_names[] =
-    "ii\0" /* Parameter signature */
-    "glLightEnviSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeuiSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glReplacementCodeuiSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribPointerNV_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glVertexAttribPointerNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char GetFramebufferAttachmentParameterivEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetFramebufferAttachmentParameteriv\0"
-    "glGetFramebufferAttachmentParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_pixel_transform)
-static const char PixelTransformParameterfEXT_names[] =
-    "iif\0" /* Parameter signature */
-    "glPixelTransformParameterfEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord4dv\0"
-    "glMultiTexCoord4dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_pixel_transform)
-static const char PixelTransformParameteriEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glPixelTransformParameteriEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor4ubVertex3fSUN_names[] =
-    "ffiiiifff\0" /* Parameter signature */
-    "glTexCoord2fColor4ubVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform1iARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glUniform1i\0"
-    "glUniform1iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttribPointerARB_names[] =
-    "iiiiip\0" /* Parameter signature */
-    "glVertexAttribPointer\0"
-    "glVertexAttribPointerARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_sharpen_texture)
-static const char SharpenTexFuncSGIS_names[] =
-    "iip\0" /* Parameter signature */
-    "glSharpenTexFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord4fv\0"
-    "glMultiTexCoord4fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix2x3fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix2x3fv\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char TrackMatrixNV_names[] =
-    "iiii\0" /* Parameter signature */
-    "glTrackMatrixNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerParameteriNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glCombinerParameteriNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char DeleteAsyncMarkersSGIX_names[] =
-    "ii\0" /* Parameter signature */
-    "glDeleteAsyncMarkersSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char IsAsyncMarkerSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glIsAsyncMarkerSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_framezoom)
-static const char FrameZoomSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glFrameZoomSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Normal3fVertex3fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NsvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Nsv\0"
-    "glVertexAttrib4NsvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3fv\0"
-    "glVertexAttrib3fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char GetSynciv_names[] =
-    "iiipp\0" /* Parameter signature */
-    "glGetSynciv\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char DeleteFramebuffersEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteFramebuffers\0"
-    "glDeleteFramebuffersEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const char GlobalAlphaFactorsSUN_names[] =
-    "i\0" /* Parameter signature */
-    "glGlobalAlphaFactorsSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture3D)
-static const char TexSubImage3D_names[] =
-    "iiiiiiiiiip\0" /* Parameter signature */
-    "glTexSubImage3D\0"
-    "glTexSubImage3DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3fEXT_names[] =
-    "fff\0" /* Parameter signature */
-    "glTangent3fEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3uivEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3uiv\0"
-    "glSecondaryColor3uivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const char MatrixIndexubvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMatrixIndexubvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char Color4fNormal3fVertex3fSUN_names[] =
-    "ffffffffff\0" /* Parameter signature */
-    "glColor4fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char PixelTexGenParameterfSGIS_names[] =
-    "if\0" /* Parameter signature */
-    "glPixelTexGenParameterfSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char CreateShader_names[] =
-    "i\0" /* Parameter signature */
-    "glCreateShader\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table) || defined(need_GL_EXT_paletted_texture)
-static const char GetColorTableParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetColorTableParameterfv\0"
-    "glGetColorTableParameterfvSGI\0"
-    "glGetColorTableParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightModelfvSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glFragmentLightModelfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord3fARB_names[] =
-    "ifff\0" /* Parameter signature */
-    "glMultiTexCoord3f\0"
-    "glMultiTexCoord3fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const char GetPixelTexGenParameterfvSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetPixelTexGenParameterfvSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char GenFramebuffersEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenFramebuffers\0"
-    "glGenFramebuffersEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetProgramParameterdvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetProgramParameterdvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_array_object) || defined(need_GL_APPLE_vertex_array_object)
-static const char IsVertexArrayAPPLE_names[] =
-    "i\0" /* Parameter signature */
-    "glIsVertexArray\0"
-    "glIsVertexArrayAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glFragmentLightfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char DetachShader_names[] =
-    "ii\0" /* Parameter signature */
-    "glDetachShader\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NubARB_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glVertexAttrib4Nub\0"
-    "glVertexAttrib4NubARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramEnvParameterfvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramEnvParameterfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetTrackMatrixivNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetTrackMatrixivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3svNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform4fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform4fv\0"
-    "glUniform4fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_transpose_matrix)
-static const char MultTransposeMatrixfARB_names[] =
-    "p\0" /* Parameter signature */
-    "glMultTransposeMatrixf\0"
-    "glMultTransposeMatrixfARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char ColorFragmentOp1ATI_names[] =
-    "iiiiiii\0" /* Parameter signature */
-    "glColorFragmentOp1ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char GetUniformfvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetUniformfv\0"
-    "glGetUniformfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN_names[] =
-    "iffffffffffff\0" /* Parameter signature */
-    "glReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char DetachObjectARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glDetachObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char VertexBlendARB_names[] =
-    "i\0" /* Parameter signature */
-    "glVertexBlendARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3iMESA_names[] =
-    "iii\0" /* Parameter signature */
-    "glWindowPos3i\0"
-    "glWindowPos3iARB\0"
-    "glWindowPos3iMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char SeparableFilter2D_names[] =
-    "iiiiiipp\0" /* Parameter signature */
-    "glSeparableFilter2D\0"
-    "glSeparableFilter2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor4ubVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glReplacementCodeuiColor4ubVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char CompressedTexImage2DARB_names[] =
-    "iiiiiiip\0" /* Parameter signature */
-    "glCompressedTexImage2D\0"
-    "glCompressedTexImage2DARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char ArrayElement_names[] =
-    "i\0" /* Parameter signature */
-    "glArrayElement\0"
-    "glArrayElementEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_depth_bounds_test)
-static const char DepthBoundsEXT_names[] =
-    "dd\0" /* Parameter signature */
-    "glDepthBoundsEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char ProgramParameters4fvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glProgramParameters4fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const char DeformationMap3fSGIX_names[] =
-    "iffiiffiiffiip\0" /* Parameter signature */
-    "glDeformationMap3fSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetProgramivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetMinmaxParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetMinmaxParameteriv\0"
-    "glGetMinmaxParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const char CopyTexImage1D_names[] =
-    "iiiiiii\0" /* Parameter signature */
-    "glCopyTexImage1D\0"
-    "glCopyTexImage1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char AlphaFragmentOp3ATI_names[] =
-    "iiiiiiiiiiii\0" /* Parameter signature */
-    "glAlphaFragmentOp3ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetVertexAttribdvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribdvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3fvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetFinalCombinerInputParameterivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFinalCombinerInputParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char GetMapParameterivNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetMapParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform4iARB_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glUniform4i\0"
-    "glUniform4iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionParameteri_names[] =
-    "iii\0" /* Parameter signature */
-    "glConvolutionParameteri\0"
-    "glConvolutionParameteriEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3sEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glBinormal3sEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char ConvolutionParameterf_names[] =
-    "iif\0" /* Parameter signature */
-    "glConvolutionParameterf\0"
-    "glConvolutionParameterfEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGI_color_table) || defined(need_GL_EXT_paletted_texture)
-static const char GetColorTableParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetColorTableParameteriv\0"
-    "glGetColorTableParameterivSGI\0"
-    "glGetColorTableParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char ProgramEnvParameter4dvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glProgramEnvParameter4dvARB\0"
-    "glProgramParameter4dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs2fvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs2fvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char UseProgramObjectARB_names[] =
-    "i\0" /* Parameter signature */
-    "glUseProgram\0"
-    "glUseProgramObjectARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char GetMapParameterfvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetMapParameterfvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char PassTexCoordATI_names[] =
-    "iii\0" /* Parameter signature */
-    "glPassTexCoordATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char DeleteProgram_names[] =
-    "i\0" /* Parameter signature */
-    "glDeleteProgram\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3ivEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glTangent3ivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3dEXT_names[] =
-    "ddd\0" /* Parameter signature */
-    "glTangent3dEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3dvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3dv\0"
-    "glSecondaryColor3dvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_multi_draw_arrays)
-static const char MultiDrawArraysEXT_names[] =
-    "ippi\0" /* Parameter signature */
-    "glMultiDrawArrays\0"
-    "glMultiDrawArraysEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char BindRenderbufferEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glBindRenderbuffer\0"
-    "glBindRenderbufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4dARB_names[] =
-    "idddd\0" /* Parameter signature */
-    "glMultiTexCoord4d\0"
-    "glMultiTexCoord4dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3usEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3us\0"
-    "glSecondaryColor3usEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char ProgramLocalParameter4fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glProgramLocalParameter4fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char DeleteProgramsNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteProgramsARB\0"
-    "glDeleteProgramsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1sARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glMultiTexCoord1s\0"
-    "glMultiTexCoord1sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiColor3fVertex3fSUN_names[] =
-    "iffffff\0" /* Parameter signature */
-    "glReplacementCodeuiColor3fVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char GetVertexAttribPointervNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribPointerv\0"
-    "glGetVertexAttribPointervARB\0"
-    "glGetVertexAttribPointervNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1dvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord1dv\0"
-    "glMultiTexCoord1dvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform2iARB_names[] =
-    "iii\0" /* Parameter signature */
-    "glUniform2i\0"
-    "glUniform2iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char GetProgramStringNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramStringNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char ColorPointerEXT_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glColorPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char MapBufferARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glMapBuffer\0"
-    "glMapBufferARB\0"
-    "";
-#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 */
-    "glBinormal3svEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_light_texture)
-static const char ApplyTextureEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glApplyTextureEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_light_texture)
-static const char TextureMaterialEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glTextureMaterialEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_light_texture)
-static const char TextureLightEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glTextureLightEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char ResetMinmax_names[] =
-    "i\0" /* Parameter signature */
-    "glResetMinmax\0"
-    "glResetMinmaxEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_sprite)
-static const char SpriteParameterfSGIX_names[] =
-    "if\0" /* Parameter signature */
-    "glSpriteParameterfSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4sNV_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glVertexAttrib4sNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char GetConvolutionParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetConvolutionParameterfv\0"
-    "glGetConvolutionParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs4dvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs4dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_multimode_draw_arrays)
-static const char MultiModeDrawArraysIBM_names[] =
-    "pppii\0" /* Parameter signature */
-    "glMultiModeDrawArraysIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4dARB_names[] =
-    "idddd\0" /* Parameter signature */
-    "glVertexAttrib4d\0"
-    "glVertexAttrib4dARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_envmap_bumpmap)
-static const char GetTexBumpParameterfvATI_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetTexBumpParameterfvATI\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const char ProgramNamedParameter4dNV_names[] =
-    "iipdddd\0" /* Parameter signature */
-    "glProgramNamedParameter4dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_weighting)
-static const char VertexWeightfEXT_names[] =
-    "f\0" /* Parameter signature */
-    "glVertexWeightfEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3fEXT_names[] =
-    "fff\0" /* Parameter signature */
-    "glBinormal3fEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
-static const char FogCoordfvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glFogCoordfv\0"
-    "glFogCoordfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1ivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord1iv\0"
-    "glMultiTexCoord1ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3ubEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glSecondaryColor3ub\0"
-    "glSecondaryColor3ubEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2ivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord2iv\0"
-    "glMultiTexCoord2ivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_fog_function)
-static const char FogFuncSGIS_names[] =
-    "ip\0" /* Parameter signature */
-    "glFogFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const char CopyTexSubImage2D_names[] =
-    "iiiiiiii\0" /* Parameter signature */
-    "glCopyTexSubImage2D\0"
-    "glCopyTexSubImage2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char GetObjectParameterivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetObjectParameterivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord4fVertex4fSUN_names[] =
-    "ffffffff\0" /* Parameter signature */
-    "glTexCoord4fVertex4fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_APPLE_vertex_array_object)
-static const char BindVertexArrayAPPLE_names[] =
-    "i\0" /* Parameter signature */
-    "glBindVertexArrayAPPLE\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramLocalParameterdvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramLocalParameterdvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetHistogramParameteriv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetHistogramParameteriv\0"
-    "glGetHistogramParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1iARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glMultiTexCoord1i\0"
-    "glMultiTexCoord1iARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char GetConvolutionFilter_names[] =
-    "iiip\0" /* Parameter signature */
-    "glGetConvolutionFilter\0"
-    "glGetConvolutionFilterEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramivARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_blend_func_separate) || defined(need_GL_INGR_blend_func_separate)
-static const char BlendFuncSeparateEXT_names[] =
-    "iiii\0" /* Parameter signature */
-    "glBlendFuncSeparate\0"
-    "glBlendFuncSeparateEXT\0"
-    "glBlendFuncSeparateINGR\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_map_buffer_range)
-static const char MapBufferRange_names[] =
-    "iiii\0" /* Parameter signature */
-    "glMapBufferRange\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char ProgramParameters4dvNV_names[] =
-    "iiip\0" /* Parameter signature */
-    "glProgramParameters4dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord2fColor3fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glTexCoord2fColor3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3dvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glBinormal3dvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_fence)
-static const char FinishFenceNV_names[] =
-    "i\0" /* Parameter signature */
-    "glFinishFenceNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIS_fog_function)
-static const char GetFogFuncSGIS_names[] =
-    "p\0" /* Parameter signature */
-    "glGetFogFuncSGIS\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char GetUniformLocationARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetUniformLocation\0"
-    "glGetUniformLocationARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3fEXT_names[] =
-    "fff\0" /* Parameter signature */
-    "glSecondaryColor3f\0"
-    "glSecondaryColor3fEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerInputNV_names[] =
-    "iiiiii\0" /* Parameter signature */
-    "glCombinerInputNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib3sARB_names[] =
-    "iiii\0" /* Parameter signature */
-    "glVertexAttrib3s\0"
-    "glVertexAttrib3sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiNormal3fVertex3fvSUN_names[] =
-    "ppp\0" /* Parameter signature */
-    "glReplacementCodeuiNormal3fVertex3fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char ProgramStringARB_names[] =
-    "iiip\0" /* Parameter signature */
-    "glProgramStringARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char TexCoord4fVertex4fvSUN_names[] =
-    "pp\0" /* Parameter signature */
-    "glTexCoord4fVertex4fvSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3sNV_names[] =
-    "iiii\0" /* Parameter signature */
-    "glVertexAttrib3sNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1fNV_names[] =
-    "if\0" /* Parameter signature */
-    "glVertexAttrib1fNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentLightfSGIX_names[] =
-    "iif\0" /* Parameter signature */
-    "glFragmentLightfSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_texture_compression)
-static const char GetCompressedTexImageARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetCompressedTexImage\0"
-    "glGetCompressedTexImageARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_weighting)
-static const char VertexWeightPointerEXT_names[] =
-    "iiip\0" /* Parameter signature */
-    "glVertexWeightPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetHistogram_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glGetHistogram\0"
-    "glGetHistogramEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_stencil_two_side)
-static const char ActiveStencilFaceEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glActiveStencilFaceEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_separate_stencil)
-static const char StencilFuncSeparateATI_names[] =
-    "iiii\0" /* Parameter signature */
-    "glStencilFuncSeparateATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char GetShaderSourceARB_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetShaderSource\0"
-    "glGetShaderSourceARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_igloo_interface)
-static const char IglooInterfaceSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glIglooInterfaceSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4dNV_names[] =
-    "idddd\0" /* Parameter signature */
-    "glVertexAttrib4dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_multimode_draw_arrays)
-static const char MultiModeDrawElementsIBM_names[] =
-    "ppipii\0" /* Parameter signature */
-    "glMultiModeDrawElementsIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord4sv\0"
-    "glMultiTexCoord4svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_occlusion_query)
-static const char GenQueriesARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenQueries\0"
-    "glGenQueriesARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const char ReplacementCodeuiVertex3fSUN_names[] =
-    "ifff\0" /* Parameter signature */
-    "glReplacementCodeuiVertex3fSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3iEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glTangent3iEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_mesh_array)
-static const char DrawMeshArraysSUN_names[] =
-    "iiii\0" /* Parameter signature */
-    "glDrawMeshArraysSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char IsSync_names[] =
-    "i\0" /* Parameter signature */
-    "glIsSync\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const char GetMapControlPointsNV_names[] =
-    "iiiiiip\0" /* Parameter signature */
-    "glGetMapControlPointsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_draw_buffers) || defined(need_GL_ATI_draw_buffers)
-static const char DrawBuffersARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glDrawBuffers\0"
-    "glDrawBuffersARB\0"
-    "glDrawBuffersATI\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char ProgramLocalParameter4fARB_names[] =
-    "iiffff\0" /* Parameter signature */
-    "glProgramLocalParameter4fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_sprite)
-static const char SpriteParameterivSGIX_names[] =
-    "ip\0" /* Parameter signature */
-    "glSpriteParameterivSGIX\0"
-    "";
-#endif
-
-#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
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord1fARB_names[] =
-    "if\0" /* Parameter signature */
-    "glMultiTexCoord1f\0"
-    "glMultiTexCoord1fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs4ubvNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs4ubvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightsvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightsvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_shader_objects)
-static const char Uniform1fvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glUniform1fv\0"
-    "glUniform1fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const char CopyTexSubImage1D_names[] =
-    "iiiiii\0" /* Parameter signature */
-    "glCopyTexSubImage1D\0"
-    "glCopyTexSubImage1DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const char BindTexture_names[] =
-    "ii\0" /* Parameter signature */
-    "glBindTexture\0"
-    "glBindTextureEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char BeginFragmentShaderATI_names[] =
-    "\0" /* Parameter signature */
-    "glBeginFragmentShaderATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord4fARB_names[] =
-    "iffff\0" /* Parameter signature */
-    "glMultiTexCoord4f\0"
-    "glMultiTexCoord4fARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs3svNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs3svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const char ReplacementCodeuivSUN_names[] =
-    "p\0" /* Parameter signature */
-    "glReplacementCodeuivSUN\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char EnableVertexAttribArrayARB_names[] =
-    "i\0" /* Parameter signature */
-    "glEnableVertexAttribArray\0"
-    "glEnableVertexAttribArrayARB\0"
-    "";
-#endif
-
-#if defined(need_GL_INTEL_parallel_arrays)
-static const char NormalPointervINTEL_names[] =
-    "ip\0" /* Parameter signature */
-    "glNormalPointervINTEL\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const char CopyConvolutionFilter2D_names[] =
-    "iiiiii\0" /* Parameter signature */
-    "glCopyConvolutionFilter2D\0"
-    "glCopyConvolutionFilter2DEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3ivMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos3iv\0"
-    "glWindowPos3ivARB\0"
-    "glWindowPos3ivMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_copy_buffer)
-static const char CopyBufferSubData_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glCopyBufferSubData\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char IsBufferARB_names[] =
-    "i\0" /* Parameter signature */
-    "glIsBuffer\0"
-    "glIsBufferARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4iMESA_names[] =
-    "iiii\0" /* Parameter signature */
-    "glWindowPos4iMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4uivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4uiv\0"
-    "glVertexAttrib4uivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3bvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glTangent3bvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix3x4fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix3x4fv\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_draw_elements_base_vertex)
-static const char DrawRangeElementsBaseVertex_names[] =
-    "iiiiipi\0" /* Parameter signature */
-    "glDrawRangeElementsBaseVertex\0"
-    "";
-#endif
-
-#if defined(need_GL_INTEL_parallel_arrays)
-static const char TexCoordPointervINTEL_names[] =
-    "iip\0" /* Parameter signature */
-    "glTexCoordPointervINTEL\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char DeleteBuffersARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteBuffers\0"
-    "glDeleteBuffersARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const char WindowPos4fvMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos4fvMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1sNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glVertexAttrib1sNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_secondary_color)
-static const char SecondaryColor3svEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glSecondaryColor3sv\0"
-    "glSecondaryColor3svEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3) || defined(need_GL_ARB_transpose_matrix)
-static const char LoadTransposeMatrixfARB_names[] =
-    "p\0" /* Parameter signature */
-    "glLoadTransposeMatrixf\0"
-    "glLoadTransposeMatrixfARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char GetPointerv_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetPointerv\0"
-    "glGetPointervEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3bEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glTangent3bEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerParameterfNV_names[] =
-    "if\0" /* Parameter signature */
-    "glCombinerParameterfNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char BindProgramNV_names[] =
-    "ii\0" /* Parameter signature */
-    "glBindProgramARB\0"
-    "glBindProgramNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4sv\0"
-    "glVertexAttrib4svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_MESA_shader_debug)
-static const char CreateDebugObjectMESA_names[] =
-    "\0" /* Parameter signature */
-    "glCreateDebugObjectMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const char GetShaderiv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetShaderiv\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const char ClientWaitSync_names[] =
-    "iii\0" /* Parameter signature */
-    "glClientWaitSync\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char BindFragmentShaderATI_names[] =
-    "i\0" /* Parameter signature */
-    "glBindFragmentShaderATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char UnmapBufferARB_names[] =
-    "i\0" /* Parameter signature */
-    "glUnmapBuffer\0"
-    "glUnmapBufferARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char Minmax_names[] =
-    "iii\0" /* Parameter signature */
-    "glMinmax\0"
-    "glMinmaxEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
-static const char FogCoorddvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glFogCoorddv\0"
-    "glFogCoorddvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SUNX_constant_data)
-static const char FinishTextureSUNX_names[] =
-    "\0" /* Parameter signature */
-    "glFinishTextureSUNX\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char GetFragmentLightfvSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetFragmentLightfvSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3fvEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glBinormal3fvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char ColorFragmentOp3ATI_names[] =
-    "iiiiiiiiiiiii\0" /* Parameter signature */
-    "glColorFragmentOp3ATI\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2svARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib2sv\0"
-    "glVertexAttrib2svARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char AreProgramsResidentNV_names[] =
-    "ipp\0" /* Parameter signature */
-    "glAreProgramsResidentNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos3svMESA_names[] =
-    "p\0" /* Parameter signature */
-    "glWindowPos3sv\0"
-    "glWindowPos3svARB\0"
-    "glWindowPos3svMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_color_subtable)
-static const char CopyColorSubTable_names[] =
-    "iiiii\0" /* Parameter signature */
-    "glCopyColorSubTable\0"
-    "glCopyColorSubTableEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightdvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightdvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char DeleteRenderbuffersEXT_names[] =
-    "ip\0" /* Parameter signature */
-    "glDeleteRenderbuffers\0"
-    "glDeleteRenderbuffersEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib4NubvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4Nubv\0"
-    "glVertexAttrib4NubvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib3dvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib3dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char GetObjectParameterfvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetObjectParameterfvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const char GetProgramEnvParameterdvARB_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetProgramEnvParameterdvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_compiled_vertex_array)
-static const char LockArraysEXT_names[] =
-    "ii\0" /* Parameter signature */
-    "glLockArraysEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_pixel_transform)
-static const char PixelTransformParameterivEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glPixelTransformParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char BinormalPointerEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glBinormalPointerEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib1dNV_names[] =
-    "id\0" /* Parameter signature */
-    "glVertexAttrib1dNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char GetCombinerInputParameterivNV_names[] =
-    "iiiip\0" /* Parameter signature */
-    "glGetCombinerInputParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const char MultiTexCoord2fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glMultiTexCoord2fv\0"
-    "glMultiTexCoord2fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char GetRenderbufferParameterivEXT_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetRenderbufferParameteriv\0"
-    "glGetRenderbufferParameterivEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-static const char CombinerParameterivNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glCombinerParameterivNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char GenFragmentShadersATI_names[] =
-    "i\0" /* Parameter signature */
-    "glGenFragmentShadersATI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const char DrawArrays_names[] =
-    "iii\0" /* Parameter signature */
-    "glDrawArrays\0"
-    "glDrawArraysEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const char WeightuivARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glWeightuivARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib2sARB_names[] =
-    "iii\0" /* Parameter signature */
-    "glVertexAttrib2s\0"
-    "glVertexAttrib2sARB\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const char GenAsyncMarkersSGIX_names[] =
-    "i\0" /* Parameter signature */
-    "glGenAsyncMarkersSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Tangent3svEXT_names[] =
-    "p\0" /* Parameter signature */
-    "glTangent3svEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const char GetListParameterivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetListParameterivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char BindBufferARB_names[] =
-    "ii\0" /* Parameter signature */
-    "glBindBuffer\0"
-    "glBindBufferARB\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_shader_objects)
-static const char GetInfoLogARB_names[] =
-    "iipp\0" /* Parameter signature */
-    "glGetInfoLogARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs4svNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs4svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const char EdgeFlagPointerListIBM_names[] =
-    "ipi\0" /* Parameter signature */
-    "glEdgeFlagPointerListIBM\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const char UniformMatrix3x2fv_names[] =
-    "iiip\0" /* Parameter signature */
-    "glUniformMatrix3x2fv\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const char GetMinmaxParameterfv_names[] =
-    "iip\0" /* Parameter signature */
-    "glGetMinmaxParameterfv\0"
-    "glGetMinmaxParameterfvEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program)
-static const char VertexAttrib1fvARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib1fv\0"
-    "glVertexAttrib1fvARB\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
-static const char GenBuffersARB_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenBuffers\0"
-    "glGenBuffersARB\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttribs1svNV_names[] =
-    "iip\0" /* Parameter signature */
-    "glVertexAttribs1svNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_envmap_bumpmap)
-static const char GetTexBumpParameterivATI_names[] =
-    "ip\0" /* Parameter signature */
-    "glGetTexBumpParameterivATI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3bEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glBinormal3bEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const char FragmentMaterialivSGIX_names[] =
-    "iip\0" /* Parameter signature */
-    "glFragmentMaterialivSGIX\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
-static const char IsRenderbufferEXT_names[] =
-    "i\0" /* Parameter signature */
-    "glIsRenderbuffer\0"
-    "glIsRenderbufferEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
-static const char GenProgramsNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glGenProgramsARB\0"
-    "glGenProgramsNV\0"
-    "";
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-static const char VertexAttrib4dvNV_names[] =
-    "ip\0" /* Parameter signature */
-    "glVertexAttrib4dvNV\0"
-    "";
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const char EndFragmentShaderATI_names[] =
-    "\0" /* Parameter signature */
-    "glEndFragmentShaderATI\0"
-    "";
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const char Binormal3iEXT_names[] =
-    "iii\0" /* Parameter signature */
-    "glBinormal3iEXT\0"
-    "";
-#endif
-
-#if defined(need_GL_VERSION_1_4) || defined(need_GL_ARB_window_pos) || defined(need_GL_MESA_window_pos)
-static const char WindowPos2fMESA_names[] =
-    "ff\0" /* Parameter signature */
-    "glWindowPos2f\0"
-    "glWindowPos2fARB\0"
-    "glWindowPos2fMESA\0"
-    "";
-#endif
-
-#if defined(need_GL_3DFX_tbuffer)
-static const struct dri_extension_function GL_3DFX_tbuffer_functions[] = {
-    { TbufferMask3DFX_names, TbufferMask3DFX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_APPLE_flush_buffer_range)
-static const struct dri_extension_function GL_APPLE_flush_buffer_range_functions[] = {
-    { BufferParameteriAPPLE_names, BufferParameteriAPPLE_remap_index, -1 },
-    { FlushMappedBufferRangeAPPLE_names, FlushMappedBufferRangeAPPLE_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_APPLE_texture_range)
-static const struct dri_extension_function GL_APPLE_texture_range_functions[] = {
-    { TextureRangeAPPLE_names, TextureRangeAPPLE_remap_index, -1 },
-    { GetTexParameterPointervAPPLE_names, GetTexParameterPointervAPPLE_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_APPLE_vertex_array_object)
-static const struct dri_extension_function GL_APPLE_vertex_array_object_functions[] = {
-    { DeleteVertexArraysAPPLE_names, DeleteVertexArraysAPPLE_remap_index, -1 },
-    { GenVertexArraysAPPLE_names, GenVertexArraysAPPLE_remap_index, -1 },
-    { IsVertexArrayAPPLE_names, IsVertexArrayAPPLE_remap_index, -1 },
-    { BindVertexArrayAPPLE_names, BindVertexArrayAPPLE_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_copy_buffer)
-static const struct dri_extension_function GL_ARB_copy_buffer_functions[] = {
-    { CopyBufferSubData_names, CopyBufferSubData_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_draw_buffers)
-static const struct dri_extension_function GL_ARB_draw_buffers_functions[] = {
-    { DrawBuffersARB_names, DrawBuffersARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#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 },
-    { FramebufferTextureLayerEXT_names, FramebufferTextureLayerEXT_remap_index, -1 },
-    { GenerateMipmapEXT_names, GenerateMipmapEXT_remap_index, -1 },
-    { RenderbufferStorageEXT_names, RenderbufferStorageEXT_remap_index, -1 },
-    { CheckFramebufferStatusEXT_names, CheckFramebufferStatusEXT_remap_index, -1 },
-    { FramebufferTexture3DEXT_names, FramebufferTexture3DEXT_remap_index, -1 },
-    { FramebufferTexture2DEXT_names, FramebufferTexture2DEXT_remap_index, -1 },
-    { RenderbufferStorageMultisample_names, RenderbufferStorageMultisample_remap_index, -1 },
-    { FramebufferRenderbufferEXT_names, FramebufferRenderbufferEXT_remap_index, -1 },
-    { FramebufferTexture1DEXT_names, FramebufferTexture1DEXT_remap_index, -1 },
-    { BindFramebufferEXT_names, BindFramebufferEXT_remap_index, -1 },
-    { GenRenderbuffersEXT_names, GenRenderbuffersEXT_remap_index, -1 },
-    { IsFramebufferEXT_names, IsFramebufferEXT_remap_index, -1 },
-    { GetFramebufferAttachmentParameterivEXT_names, GetFramebufferAttachmentParameterivEXT_remap_index, -1 },
-    { DeleteFramebuffersEXT_names, DeleteFramebuffersEXT_remap_index, -1 },
-    { GenFramebuffersEXT_names, GenFramebuffersEXT_remap_index, -1 },
-    { BindRenderbufferEXT_names, BindRenderbufferEXT_remap_index, -1 },
-    { DeleteRenderbuffersEXT_names, DeleteRenderbuffersEXT_remap_index, -1 },
-    { GetRenderbufferParameterivEXT_names, GetRenderbufferParameterivEXT_remap_index, -1 },
-    { IsRenderbufferEXT_names, IsRenderbufferEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_map_buffer_range)
-static const struct dri_extension_function GL_ARB_map_buffer_range_functions[] = {
-    { FlushMappedBufferRange_names, FlushMappedBufferRange_remap_index, -1 },
-    { MapBufferRange_names, MapBufferRange_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_matrix_palette)
-static const struct dri_extension_function GL_ARB_matrix_palette_functions[] = {
-    { MatrixIndexusvARB_names, MatrixIndexusvARB_remap_index, -1 },
-    { MatrixIndexuivARB_names, MatrixIndexuivARB_remap_index, -1 },
-    { MatrixIndexPointerARB_names, MatrixIndexPointerARB_remap_index, -1 },
-    { CurrentPaletteMatrixARB_names, CurrentPaletteMatrixARB_remap_index, -1 },
-    { MatrixIndexubvARB_names, MatrixIndexubvARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_multisample)
-static const struct dri_extension_function GL_ARB_multisample_functions[] = {
-    { SampleCoverageARB_names, SampleCoverageARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_occlusion_query)
-static const struct dri_extension_function GL_ARB_occlusion_query_functions[] = {
-    { BeginQueryARB_names, BeginQueryARB_remap_index, -1 },
-    { GetQueryivARB_names, GetQueryivARB_remap_index, -1 },
-    { GetQueryObjectivARB_names, GetQueryObjectivARB_remap_index, -1 },
-    { EndQueryARB_names, EndQueryARB_remap_index, -1 },
-    { GetQueryObjectuivARB_names, GetQueryObjectuivARB_remap_index, -1 },
-    { DeleteQueriesARB_names, DeleteQueriesARB_remap_index, -1 },
-    { IsQueryARB_names, IsQueryARB_remap_index, -1 },
-    { GenQueriesARB_names, GenQueriesARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_point_parameters)
-static const struct dri_extension_function GL_ARB_point_parameters_functions[] = {
-    { PointParameterfEXT_names, PointParameterfEXT_remap_index, -1 },
-    { PointParameterfvEXT_names, PointParameterfvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#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 },
-    { Uniform2fARB_names, Uniform2fARB_remap_index, -1 },
-    { Uniform2ivARB_names, Uniform2ivARB_remap_index, -1 },
-    { UniformMatrix4fvARB_names, UniformMatrix4fvARB_remap_index, -1 },
-    { CreateProgramObjectARB_names, CreateProgramObjectARB_remap_index, -1 },
-    { Uniform3iARB_names, Uniform3iARB_remap_index, -1 },
-    { CreateShaderObjectARB_names, CreateShaderObjectARB_remap_index, -1 },
-    { Uniform1fARB_names, Uniform1fARB_remap_index, -1 },
-    { AttachObjectARB_names, AttachObjectARB_remap_index, -1 },
-    { UniformMatrix2fvARB_names, UniformMatrix2fvARB_remap_index, -1 },
-    { GetAttachedObjectsARB_names, GetAttachedObjectsARB_remap_index, -1 },
-    { Uniform3fvARB_names, Uniform3fvARB_remap_index, -1 },
-    { GetHandleARB_names, GetHandleARB_remap_index, -1 },
-    { GetActiveUniformARB_names, GetActiveUniformARB_remap_index, -1 },
-    { GetUniformivARB_names, GetUniformivARB_remap_index, -1 },
-    { Uniform2fvARB_names, Uniform2fvARB_remap_index, -1 },
-    { DeleteObjectARB_names, DeleteObjectARB_remap_index, -1 },
-    { ValidateProgramARB_names, ValidateProgramARB_remap_index, -1 },
-    { Uniform3ivARB_names, Uniform3ivARB_remap_index, -1 },
-    { CompileShaderARB_names, CompileShaderARB_remap_index, -1 },
-    { Uniform3fARB_names, Uniform3fARB_remap_index, -1 },
-    { Uniform4fARB_names, Uniform4fARB_remap_index, -1 },
-    { LinkProgramARB_names, LinkProgramARB_remap_index, -1 },
-    { ShaderSourceARB_names, ShaderSourceARB_remap_index, -1 },
-    { Uniform4ivARB_names, Uniform4ivARB_remap_index, -1 },
-    { Uniform1ivARB_names, Uniform1ivARB_remap_index, -1 },
-    { Uniform1iARB_names, Uniform1iARB_remap_index, -1 },
-    { Uniform4fvARB_names, Uniform4fvARB_remap_index, -1 },
-    { GetUniformfvARB_names, GetUniformfvARB_remap_index, -1 },
-    { DetachObjectARB_names, DetachObjectARB_remap_index, -1 },
-    { Uniform4iARB_names, Uniform4iARB_remap_index, -1 },
-    { UseProgramObjectARB_names, UseProgramObjectARB_remap_index, -1 },
-    { Uniform2iARB_names, Uniform2iARB_remap_index, -1 },
-    { GetObjectParameterivARB_names, GetObjectParameterivARB_remap_index, -1 },
-    { GetUniformLocationARB_names, GetUniformLocationARB_remap_index, -1 },
-    { GetShaderSourceARB_names, GetShaderSourceARB_remap_index, -1 },
-    { Uniform1fvARB_names, Uniform1fvARB_remap_index, -1 },
-    { GetObjectParameterfvARB_names, GetObjectParameterfvARB_remap_index, -1 },
-    { GetInfoLogARB_names, GetInfoLogARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_sync)
-static const struct dri_extension_function GL_ARB_sync_functions[] = {
-    { DeleteSync_names, DeleteSync_remap_index, -1 },
-    { FenceSync_names, FenceSync_remap_index, -1 },
-    { WaitSync_names, WaitSync_remap_index, -1 },
-    { GetInteger64v_names, GetInteger64v_remap_index, -1 },
-    { GetSynciv_names, GetSynciv_remap_index, -1 },
-    { IsSync_names, IsSync_remap_index, -1 },
-    { ClientWaitSync_names, ClientWaitSync_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_texture_compression)
-static const struct dri_extension_function GL_ARB_texture_compression_functions[] = {
-    { CompressedTexSubImage2DARB_names, CompressedTexSubImage2DARB_remap_index, -1 },
-    { CompressedTexImage3DARB_names, CompressedTexImage3DARB_remap_index, -1 },
-    { CompressedTexImage1DARB_names, CompressedTexImage1DARB_remap_index, -1 },
-    { CompressedTexSubImage1DARB_names, CompressedTexSubImage1DARB_remap_index, -1 },
-    { CompressedTexSubImage3DARB_names, CompressedTexSubImage3DARB_remap_index, -1 },
-    { CompressedTexImage2DARB_names, CompressedTexImage2DARB_remap_index, -1 },
-    { GetCompressedTexImageARB_names, GetCompressedTexImageARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_transpose_matrix)
-static const struct dri_extension_function GL_ARB_transpose_matrix_functions[] = {
-    { MultTransposeMatrixdARB_names, MultTransposeMatrixdARB_remap_index, -1 },
-    { LoadTransposeMatrixdARB_names, LoadTransposeMatrixdARB_remap_index, -1 },
-    { MultTransposeMatrixfARB_names, MultTransposeMatrixfARB_remap_index, -1 },
-    { LoadTransposeMatrixfARB_names, LoadTransposeMatrixfARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_vertex_array_object)
-static const struct dri_extension_function GL_ARB_vertex_array_object_functions[] = {
-    { DeleteVertexArraysAPPLE_names, DeleteVertexArraysAPPLE_remap_index, -1 },
-    { GenVertexArrays_names, GenVertexArrays_remap_index, -1 },
-    { BindVertexArray_names, BindVertexArray_remap_index, -1 },
-    { IsVertexArrayAPPLE_names, IsVertexArrayAPPLE_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_vertex_blend)
-static const struct dri_extension_function GL_ARB_vertex_blend_functions[] = {
-    { WeightubvARB_names, WeightubvARB_remap_index, -1 },
-    { WeightivARB_names, WeightivARB_remap_index, -1 },
-    { WeightPointerARB_names, WeightPointerARB_remap_index, -1 },
-    { WeightfvARB_names, WeightfvARB_remap_index, -1 },
-    { WeightbvARB_names, WeightbvARB_remap_index, -1 },
-    { WeightusvARB_names, WeightusvARB_remap_index, -1 },
-    { VertexBlendARB_names, VertexBlendARB_remap_index, -1 },
-    { WeightsvARB_names, WeightsvARB_remap_index, -1 },
-    { WeightdvARB_names, WeightdvARB_remap_index, -1 },
-    { WeightuivARB_names, WeightuivARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_vertex_buffer_object)
-static const struct dri_extension_function GL_ARB_vertex_buffer_object_functions[] = {
-    { GetBufferSubDataARB_names, GetBufferSubDataARB_remap_index, -1 },
-    { BufferSubDataARB_names, BufferSubDataARB_remap_index, -1 },
-    { BufferDataARB_names, BufferDataARB_remap_index, -1 },
-    { GetBufferPointervARB_names, GetBufferPointervARB_remap_index, -1 },
-    { GetBufferParameterivARB_names, GetBufferParameterivARB_remap_index, -1 },
-    { MapBufferARB_names, MapBufferARB_remap_index, -1 },
-    { IsBufferARB_names, IsBufferARB_remap_index, -1 },
-    { DeleteBuffersARB_names, DeleteBuffersARB_remap_index, -1 },
-    { UnmapBufferARB_names, UnmapBufferARB_remap_index, -1 },
-    { BindBufferARB_names, BindBufferARB_remap_index, -1 },
-    { GenBuffersARB_names, GenBuffersARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_vertex_program)
-static const struct dri_extension_function GL_ARB_vertex_program_functions[] = {
-    { VertexAttrib2fARB_names, VertexAttrib2fARB_remap_index, -1 },
-    { VertexAttrib3fARB_names, VertexAttrib3fARB_remap_index, -1 },
-    { VertexAttrib1svARB_names, VertexAttrib1svARB_remap_index, -1 },
-    { VertexAttrib4NusvARB_names, VertexAttrib4NusvARB_remap_index, -1 },
-    { DisableVertexAttribArrayARB_names, DisableVertexAttribArrayARB_remap_index, -1 },
-    { ProgramLocalParameter4dARB_names, ProgramLocalParameter4dARB_remap_index, -1 },
-    { VertexAttrib1fARB_names, VertexAttrib1fARB_remap_index, -1 },
-    { VertexAttrib4NbvARB_names, VertexAttrib4NbvARB_remap_index, -1 },
-    { VertexAttrib1sARB_names, VertexAttrib1sARB_remap_index, -1 },
-    { GetProgramLocalParameterfvARB_names, GetProgramLocalParameterfvARB_remap_index, -1 },
-    { VertexAttrib3dvARB_names, VertexAttrib3dvARB_remap_index, -1 },
-    { ProgramEnvParameter4fvARB_names, ProgramEnvParameter4fvARB_remap_index, -1 },
-    { GetVertexAttribivARB_names, GetVertexAttribivARB_remap_index, -1 },
-    { VertexAttrib4ivARB_names, VertexAttrib4ivARB_remap_index, -1 },
-    { VertexAttrib4bvARB_names, VertexAttrib4bvARB_remap_index, -1 },
-    { VertexAttrib3dARB_names, VertexAttrib3dARB_remap_index, -1 },
-    { VertexAttrib4fARB_names, VertexAttrib4fARB_remap_index, -1 },
-    { VertexAttrib4fvARB_names, VertexAttrib4fvARB_remap_index, -1 },
-    { ProgramLocalParameter4dvARB_names, ProgramLocalParameter4dvARB_remap_index, -1 },
-    { VertexAttrib2dARB_names, VertexAttrib2dARB_remap_index, -1 },
-    { VertexAttrib1dvARB_names, VertexAttrib1dvARB_remap_index, -1 },
-    { GetVertexAttribfvARB_names, GetVertexAttribfvARB_remap_index, -1 },
-    { VertexAttrib4ubvARB_names, VertexAttrib4ubvARB_remap_index, -1 },
-    { ProgramEnvParameter4fARB_names, ProgramEnvParameter4fARB_remap_index, -1 },
-    { VertexAttrib4sARB_names, VertexAttrib4sARB_remap_index, -1 },
-    { VertexAttrib2dvARB_names, VertexAttrib2dvARB_remap_index, -1 },
-    { VertexAttrib2fvARB_names, VertexAttrib2fvARB_remap_index, -1 },
-    { VertexAttrib4NivARB_names, VertexAttrib4NivARB_remap_index, -1 },
-    { GetProgramStringARB_names, GetProgramStringARB_remap_index, -1 },
-    { VertexAttrib4NuivARB_names, VertexAttrib4NuivARB_remap_index, -1 },
-    { IsProgramNV_names, IsProgramNV_remap_index, -1 },
-    { ProgramEnvParameter4dARB_names, ProgramEnvParameter4dARB_remap_index, -1 },
-    { VertexAttrib1dARB_names, VertexAttrib1dARB_remap_index, -1 },
-    { VertexAttrib4usvARB_names, VertexAttrib4usvARB_remap_index, -1 },
-    { VertexAttrib3svARB_names, VertexAttrib3svARB_remap_index, -1 },
-    { GetVertexAttribdvARB_names, GetVertexAttribdvARB_remap_index, -1 },
-    { VertexAttrib4dvARB_names, VertexAttrib4dvARB_remap_index, -1 },
-    { VertexAttribPointerARB_names, VertexAttribPointerARB_remap_index, -1 },
-    { VertexAttrib4NsvARB_names, VertexAttrib4NsvARB_remap_index, -1 },
-    { VertexAttrib3fvARB_names, VertexAttrib3fvARB_remap_index, -1 },
-    { VertexAttrib4NubARB_names, VertexAttrib4NubARB_remap_index, -1 },
-    { GetProgramEnvParameterfvARB_names, GetProgramEnvParameterfvARB_remap_index, -1 },
-    { ProgramEnvParameter4dvARB_names, ProgramEnvParameter4dvARB_remap_index, -1 },
-    { ProgramLocalParameter4fvARB_names, ProgramLocalParameter4fvARB_remap_index, -1 },
-    { DeleteProgramsNV_names, DeleteProgramsNV_remap_index, -1 },
-    { GetVertexAttribPointervNV_names, GetVertexAttribPointervNV_remap_index, -1 },
-    { VertexAttrib4dARB_names, VertexAttrib4dARB_remap_index, -1 },
-    { GetProgramLocalParameterdvARB_names, GetProgramLocalParameterdvARB_remap_index, -1 },
-    { GetProgramivARB_names, GetProgramivARB_remap_index, -1 },
-    { VertexAttrib3sARB_names, VertexAttrib3sARB_remap_index, -1 },
-    { ProgramStringARB_names, ProgramStringARB_remap_index, -1 },
-    { ProgramLocalParameter4fARB_names, ProgramLocalParameter4fARB_remap_index, -1 },
-    { EnableVertexAttribArrayARB_names, EnableVertexAttribArrayARB_remap_index, -1 },
-    { VertexAttrib4uivARB_names, VertexAttrib4uivARB_remap_index, -1 },
-    { BindProgramNV_names, BindProgramNV_remap_index, -1 },
-    { VertexAttrib4svARB_names, VertexAttrib4svARB_remap_index, -1 },
-    { VertexAttrib2svARB_names, VertexAttrib2svARB_remap_index, -1 },
-    { VertexAttrib4NubvARB_names, VertexAttrib4NubvARB_remap_index, -1 },
-    { GetProgramEnvParameterdvARB_names, GetProgramEnvParameterdvARB_remap_index, -1 },
-    { VertexAttrib2sARB_names, VertexAttrib2sARB_remap_index, -1 },
-    { VertexAttrib1fvARB_names, VertexAttrib1fvARB_remap_index, -1 },
-    { GenProgramsNV_names, GenProgramsNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_vertex_shader)
-static const struct dri_extension_function GL_ARB_vertex_shader_functions[] = {
-    { GetActiveAttribARB_names, GetActiveAttribARB_remap_index, -1 },
-    { GetAttribLocationARB_names, GetAttribLocationARB_remap_index, -1 },
-    { BindAttribLocationARB_names, BindAttribLocationARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ARB_window_pos)
-static const struct dri_extension_function GL_ARB_window_pos_functions[] = {
-    { WindowPos3fMESA_names, WindowPos3fMESA_remap_index, -1 },
-    { WindowPos2dvMESA_names, WindowPos2dvMESA_remap_index, -1 },
-    { WindowPos2svMESA_names, WindowPos2svMESA_remap_index, -1 },
-    { WindowPos3dMESA_names, WindowPos3dMESA_remap_index, -1 },
-    { WindowPos2fvMESA_names, WindowPos2fvMESA_remap_index, -1 },
-    { WindowPos2dMESA_names, WindowPos2dMESA_remap_index, -1 },
-    { WindowPos3dvMESA_names, WindowPos3dvMESA_remap_index, -1 },
-    { WindowPos3fvMESA_names, WindowPos3fvMESA_remap_index, -1 },
-    { WindowPos2iMESA_names, WindowPos2iMESA_remap_index, -1 },
-    { WindowPos3sMESA_names, WindowPos3sMESA_remap_index, -1 },
-    { WindowPos2ivMESA_names, WindowPos2ivMESA_remap_index, -1 },
-    { WindowPos2sMESA_names, WindowPos2sMESA_remap_index, -1 },
-    { WindowPos3iMESA_names, WindowPos3iMESA_remap_index, -1 },
-    { WindowPos3ivMESA_names, WindowPos3ivMESA_remap_index, -1 },
-    { WindowPos3svMESA_names, WindowPos3svMESA_remap_index, -1 },
-    { WindowPos2fMESA_names, WindowPos2fMESA_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ATI_blend_equation_separate)
-static const struct dri_extension_function GL_ATI_blend_equation_separate_functions[] = {
-    { BlendEquationSeparateEXT_names, BlendEquationSeparateEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ATI_draw_buffers)
-static const struct dri_extension_function GL_ATI_draw_buffers_functions[] = {
-    { DrawBuffersARB_names, DrawBuffersARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ATI_envmap_bumpmap)
-static const struct dri_extension_function GL_ATI_envmap_bumpmap_functions[] = {
-    { TexBumpParameterfvATI_names, TexBumpParameterfvATI_remap_index, -1 },
-    { TexBumpParameterivATI_names, TexBumpParameterivATI_remap_index, -1 },
-    { GetTexBumpParameterfvATI_names, GetTexBumpParameterfvATI_remap_index, -1 },
-    { GetTexBumpParameterivATI_names, GetTexBumpParameterivATI_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ATI_fragment_shader)
-static const struct dri_extension_function GL_ATI_fragment_shader_functions[] = {
-    { ColorFragmentOp2ATI_names, ColorFragmentOp2ATI_remap_index, -1 },
-    { DeleteFragmentShaderATI_names, DeleteFragmentShaderATI_remap_index, -1 },
-    { SetFragmentShaderConstantATI_names, SetFragmentShaderConstantATI_remap_index, -1 },
-    { SampleMapATI_names, SampleMapATI_remap_index, -1 },
-    { AlphaFragmentOp2ATI_names, AlphaFragmentOp2ATI_remap_index, -1 },
-    { AlphaFragmentOp1ATI_names, AlphaFragmentOp1ATI_remap_index, -1 },
-    { ColorFragmentOp1ATI_names, ColorFragmentOp1ATI_remap_index, -1 },
-    { AlphaFragmentOp3ATI_names, AlphaFragmentOp3ATI_remap_index, -1 },
-    { PassTexCoordATI_names, PassTexCoordATI_remap_index, -1 },
-    { BeginFragmentShaderATI_names, BeginFragmentShaderATI_remap_index, -1 },
-    { BindFragmentShaderATI_names, BindFragmentShaderATI_remap_index, -1 },
-    { ColorFragmentOp3ATI_names, ColorFragmentOp3ATI_remap_index, -1 },
-    { GenFragmentShadersATI_names, GenFragmentShadersATI_remap_index, -1 },
-    { EndFragmentShaderATI_names, EndFragmentShaderATI_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_ATI_separate_stencil)
-static const struct dri_extension_function GL_ATI_separate_stencil_functions[] = {
-    { StencilOpSeparate_names, StencilOpSeparate_remap_index, -1 },
-    { StencilFuncSeparateATI_names, StencilFuncSeparateATI_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_blend_color)
-static const struct dri_extension_function GL_EXT_blend_color_functions[] = {
-    { BlendColor_names, -1, 336 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_blend_equation_separate)
-static const struct dri_extension_function GL_EXT_blend_equation_separate_functions[] = {
-    { BlendEquationSeparateEXT_names, BlendEquationSeparateEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_blend_func_separate)
-static const struct dri_extension_function GL_EXT_blend_func_separate_functions[] = {
-    { BlendFuncSeparateEXT_names, BlendFuncSeparateEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_blend_minmax)
-static const struct dri_extension_function GL_EXT_blend_minmax_functions[] = {
-    { BlendEquation_names, -1, 337 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_color_subtable)
-static const struct dri_extension_function GL_EXT_color_subtable_functions[] = {
-    { ColorSubTable_names, -1, 346 },
-    { CopyColorSubTable_names, -1, 347 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_compiled_vertex_array)
-static const struct dri_extension_function GL_EXT_compiled_vertex_array_functions[] = {
-    { UnlockArraysEXT_names, UnlockArraysEXT_remap_index, -1 },
-    { LockArraysEXT_names, LockArraysEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_convolution)
-static const struct dri_extension_function GL_EXT_convolution_functions[] = {
-    { ConvolutionFilter1D_names, -1, 348 },
-    { CopyConvolutionFilter1D_names, -1, 354 },
-    { GetConvolutionParameteriv_names, -1, 358 },
-    { ConvolutionFilter2D_names, -1, 349 },
-    { ConvolutionParameteriv_names, -1, 353 },
-    { ConvolutionParameterfv_names, -1, 351 },
-    { GetSeparableFilter_names, -1, 359 },
-    { SeparableFilter2D_names, -1, 360 },
-    { ConvolutionParameteri_names, -1, 352 },
-    { ConvolutionParameterf_names, -1, 350 },
-    { GetConvolutionParameterfv_names, -1, 357 },
-    { GetConvolutionFilter_names, -1, 356 },
-    { CopyConvolutionFilter2D_names, -1, 355 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_coordinate_frame)
-static const struct dri_extension_function GL_EXT_coordinate_frame_functions[] = {
-    { TangentPointerEXT_names, TangentPointerEXT_remap_index, -1 },
-    { Binormal3ivEXT_names, Binormal3ivEXT_remap_index, -1 },
-    { Tangent3sEXT_names, Tangent3sEXT_remap_index, -1 },
-    { Tangent3fvEXT_names, Tangent3fvEXT_remap_index, -1 },
-    { Tangent3dvEXT_names, Tangent3dvEXT_remap_index, -1 },
-    { Binormal3bvEXT_names, Binormal3bvEXT_remap_index, -1 },
-    { Binormal3dEXT_names, Binormal3dEXT_remap_index, -1 },
-    { Tangent3fEXT_names, Tangent3fEXT_remap_index, -1 },
-    { Binormal3sEXT_names, Binormal3sEXT_remap_index, -1 },
-    { Tangent3ivEXT_names, Tangent3ivEXT_remap_index, -1 },
-    { Tangent3dEXT_names, Tangent3dEXT_remap_index, -1 },
-    { Binormal3svEXT_names, Binormal3svEXT_remap_index, -1 },
-    { Binormal3fEXT_names, Binormal3fEXT_remap_index, -1 },
-    { Binormal3dvEXT_names, Binormal3dvEXT_remap_index, -1 },
-    { Tangent3iEXT_names, Tangent3iEXT_remap_index, -1 },
-    { Tangent3bvEXT_names, Tangent3bvEXT_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 },
-    { Binormal3iEXT_names, Binormal3iEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_copy_texture)
-static const struct dri_extension_function GL_EXT_copy_texture_functions[] = {
-    { CopyTexSubImage3D_names, -1, 373 },
-    { CopyTexImage2D_names, -1, 324 },
-    { CopyTexImage1D_names, -1, 323 },
-    { CopyTexSubImage2D_names, -1, 326 },
-    { CopyTexSubImage1D_names, -1, 325 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_cull_vertex)
-static const struct dri_extension_function GL_EXT_cull_vertex_functions[] = {
-    { CullParameterdvEXT_names, CullParameterdvEXT_remap_index, -1 },
-    { CullParameterfvEXT_names, CullParameterfvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_depth_bounds_test)
-static const struct dri_extension_function GL_EXT_depth_bounds_test_functions[] = {
-    { DepthBoundsEXT_names, DepthBoundsEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_draw_range_elements)
-static const struct dri_extension_function GL_EXT_draw_range_elements_functions[] = {
-    { DrawRangeElements_names, -1, 338 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_fog_coord)
-static const struct dri_extension_function GL_EXT_fog_coord_functions[] = {
-    { FogCoorddEXT_names, FogCoorddEXT_remap_index, -1 },
-    { FogCoordfEXT_names, FogCoordfEXT_remap_index, -1 },
-    { FogCoordPointerEXT_names, FogCoordPointerEXT_remap_index, -1 },
-    { FogCoordfvEXT_names, FogCoordfvEXT_remap_index, -1 },
-    { FogCoorddvEXT_names, FogCoorddvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_framebuffer_blit)
-static const struct dri_extension_function GL_EXT_framebuffer_blit_functions[] = {
-    { BlitFramebufferEXT_names, BlitFramebufferEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_framebuffer_object)
-static const struct dri_extension_function GL_EXT_framebuffer_object_functions[] = {
-    { GenerateMipmapEXT_names, GenerateMipmapEXT_remap_index, -1 },
-    { RenderbufferStorageEXT_names, RenderbufferStorageEXT_remap_index, -1 },
-    { CheckFramebufferStatusEXT_names, CheckFramebufferStatusEXT_remap_index, -1 },
-    { FramebufferTexture3DEXT_names, FramebufferTexture3DEXT_remap_index, -1 },
-    { FramebufferTexture2DEXT_names, FramebufferTexture2DEXT_remap_index, -1 },
-    { FramebufferRenderbufferEXT_names, FramebufferRenderbufferEXT_remap_index, -1 },
-    { FramebufferTexture1DEXT_names, FramebufferTexture1DEXT_remap_index, -1 },
-    { BindFramebufferEXT_names, BindFramebufferEXT_remap_index, -1 },
-    { GenRenderbuffersEXT_names, GenRenderbuffersEXT_remap_index, -1 },
-    { IsFramebufferEXT_names, IsFramebufferEXT_remap_index, -1 },
-    { GetFramebufferAttachmentParameterivEXT_names, GetFramebufferAttachmentParameterivEXT_remap_index, -1 },
-    { DeleteFramebuffersEXT_names, DeleteFramebuffersEXT_remap_index, -1 },
-    { GenFramebuffersEXT_names, GenFramebuffersEXT_remap_index, -1 },
-    { BindRenderbufferEXT_names, BindRenderbufferEXT_remap_index, -1 },
-    { DeleteRenderbuffersEXT_names, DeleteRenderbuffersEXT_remap_index, -1 },
-    { GetRenderbufferParameterivEXT_names, GetRenderbufferParameterivEXT_remap_index, -1 },
-    { IsRenderbufferEXT_names, IsRenderbufferEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_gpu_program_parameters)
-static const struct dri_extension_function GL_EXT_gpu_program_parameters_functions[] = {
-    { ProgramLocalParameters4fvEXT_names, ProgramLocalParameters4fvEXT_remap_index, -1 },
-    { ProgramEnvParameters4fvEXT_names, ProgramEnvParameters4fvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_histogram)
-static const struct dri_extension_function GL_EXT_histogram_functions[] = {
-    { Histogram_names, -1, 367 },
-    { ResetHistogram_names, -1, 369 },
-    { GetMinmax_names, -1, 364 },
-    { GetHistogramParameterfv_names, -1, 362 },
-    { GetMinmaxParameteriv_names, -1, 366 },
-    { ResetMinmax_names, -1, 370 },
-    { GetHistogramParameteriv_names, -1, 363 },
-    { GetHistogram_names, -1, 361 },
-    { Minmax_names, -1, 368 },
-    { GetMinmaxParameterfv_names, -1, 365 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_index_func)
-static const struct dri_extension_function GL_EXT_index_func_functions[] = {
-    { IndexFuncEXT_names, IndexFuncEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_index_material)
-static const struct dri_extension_function GL_EXT_index_material_functions[] = {
-    { IndexMaterialEXT_names, IndexMaterialEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_light_texture)
-static const struct dri_extension_function GL_EXT_light_texture_functions[] = {
-    { ApplyTextureEXT_names, ApplyTextureEXT_remap_index, -1 },
-    { TextureMaterialEXT_names, TextureMaterialEXT_remap_index, -1 },
-    { TextureLightEXT_names, TextureLightEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_multi_draw_arrays)
-static const struct dri_extension_function GL_EXT_multi_draw_arrays_functions[] = {
-    { MultiDrawElementsEXT_names, MultiDrawElementsEXT_remap_index, -1 },
-    { MultiDrawArraysEXT_names, MultiDrawArraysEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_multisample)
-static const struct dri_extension_function GL_EXT_multisample_functions[] = {
-    { SampleMaskSGIS_names, SampleMaskSGIS_remap_index, -1 },
-    { SamplePatternSGIS_names, SamplePatternSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_paletted_texture)
-static const struct dri_extension_function GL_EXT_paletted_texture_functions[] = {
-    { ColorTable_names, -1, 339 },
-    { GetColorTable_names, -1, 343 },
-    { GetColorTableParameterfv_names, -1, 344 },
-    { GetColorTableParameteriv_names, -1, 345 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_pixel_transform)
-static const struct dri_extension_function GL_EXT_pixel_transform_functions[] = {
-    { PixelTransformParameterfvEXT_names, PixelTransformParameterfvEXT_remap_index, -1 },
-    { PixelTransformParameterfEXT_names, PixelTransformParameterfEXT_remap_index, -1 },
-    { PixelTransformParameteriEXT_names, PixelTransformParameteriEXT_remap_index, -1 },
-    { PixelTransformParameterivEXT_names, PixelTransformParameterivEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_point_parameters)
-static const struct dri_extension_function GL_EXT_point_parameters_functions[] = {
-    { PointParameterfEXT_names, PointParameterfEXT_remap_index, -1 },
-    { PointParameterfvEXT_names, PointParameterfvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_polygon_offset)
-static const struct dri_extension_function GL_EXT_polygon_offset_functions[] = {
-    { PolygonOffsetEXT_names, PolygonOffsetEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_provoking_vertex)
-static const struct dri_extension_function GL_EXT_provoking_vertex_functions[] = {
-    { ProvokingVertexEXT_names, ProvokingVertexEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_secondary_color)
-static const struct dri_extension_function GL_EXT_secondary_color_functions[] = {
-    { SecondaryColor3iEXT_names, SecondaryColor3iEXT_remap_index, -1 },
-    { SecondaryColor3bEXT_names, SecondaryColor3bEXT_remap_index, -1 },
-    { SecondaryColor3bvEXT_names, SecondaryColor3bvEXT_remap_index, -1 },
-    { SecondaryColor3sEXT_names, SecondaryColor3sEXT_remap_index, -1 },
-    { SecondaryColor3dEXT_names, SecondaryColor3dEXT_remap_index, -1 },
-    { SecondaryColorPointerEXT_names, SecondaryColorPointerEXT_remap_index, -1 },
-    { SecondaryColor3uiEXT_names, SecondaryColor3uiEXT_remap_index, -1 },
-    { SecondaryColor3usvEXT_names, SecondaryColor3usvEXT_remap_index, -1 },
-    { SecondaryColor3ivEXT_names, SecondaryColor3ivEXT_remap_index, -1 },
-    { SecondaryColor3fvEXT_names, SecondaryColor3fvEXT_remap_index, -1 },
-    { SecondaryColor3ubvEXT_names, SecondaryColor3ubvEXT_remap_index, -1 },
-    { SecondaryColor3uivEXT_names, SecondaryColor3uivEXT_remap_index, -1 },
-    { SecondaryColor3dvEXT_names, SecondaryColor3dvEXT_remap_index, -1 },
-    { SecondaryColor3usEXT_names, SecondaryColor3usEXT_remap_index, -1 },
-    { SecondaryColor3ubEXT_names, SecondaryColor3ubEXT_remap_index, -1 },
-    { SecondaryColor3fEXT_names, SecondaryColor3fEXT_remap_index, -1 },
-    { SecondaryColor3svEXT_names, SecondaryColor3svEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_stencil_two_side)
-static const struct dri_extension_function GL_EXT_stencil_two_side_functions[] = {
-    { ActiveStencilFaceEXT_names, ActiveStencilFaceEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_subtexture)
-static const struct dri_extension_function GL_EXT_subtexture_functions[] = {
-    { TexSubImage1D_names, -1, 332 },
-    { TexSubImage2D_names, -1, 333 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_texture3D)
-static const struct dri_extension_function GL_EXT_texture3D_functions[] = {
-    { TexImage3D_names, -1, 371 },
-    { TexSubImage3D_names, -1, 372 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_texture_array)
-static const struct dri_extension_function GL_EXT_texture_array_functions[] = {
-    { FramebufferTextureLayerEXT_names, FramebufferTextureLayerEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_texture_object)
-static const struct dri_extension_function GL_EXT_texture_object_functions[] = {
-    { PrioritizeTextures_names, -1, 331 },
-    { AreTexturesResident_names, -1, 322 },
-    { GenTextures_names, -1, 328 },
-    { DeleteTextures_names, -1, 327 },
-    { IsTexture_names, -1, 330 },
-    { BindTexture_names, -1, 307 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_texture_perturb_normal)
-static const struct dri_extension_function GL_EXT_texture_perturb_normal_functions[] = {
-    { TextureNormalEXT_names, TextureNormalEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_timer_query)
-static const struct dri_extension_function GL_EXT_timer_query_functions[] = {
-    { GetQueryObjectui64vEXT_names, GetQueryObjectui64vEXT_remap_index, -1 },
-    { GetQueryObjecti64vEXT_names, GetQueryObjecti64vEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_vertex_array)
-static const struct dri_extension_function GL_EXT_vertex_array_functions[] = {
-    { IndexPointerEXT_names, IndexPointerEXT_remap_index, -1 },
-    { NormalPointerEXT_names, NormalPointerEXT_remap_index, -1 },
-    { VertexPointerEXT_names, VertexPointerEXT_remap_index, -1 },
-    { TexCoordPointerEXT_names, TexCoordPointerEXT_remap_index, -1 },
-    { EdgeFlagPointerEXT_names, EdgeFlagPointerEXT_remap_index, -1 },
-    { ArrayElement_names, -1, 306 },
-    { ColorPointerEXT_names, ColorPointerEXT_remap_index, -1 },
-    { GetPointerv_names, -1, 329 },
-    { DrawArrays_names, -1, 310 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_EXT_vertex_weighting)
-static const struct dri_extension_function GL_EXT_vertex_weighting_functions[] = {
-    { VertexWeightfvEXT_names, VertexWeightfvEXT_remap_index, -1 },
-    { VertexWeightfEXT_names, VertexWeightfEXT_remap_index, -1 },
-    { VertexWeightPointerEXT_names, VertexWeightPointerEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_HP_image_transform)
-static const struct dri_extension_function GL_HP_image_transform_functions[] = {
-    { GetImageTransformParameterfvHP_names, GetImageTransformParameterfvHP_remap_index, -1 },
-    { ImageTransformParameterfHP_names, ImageTransformParameterfHP_remap_index, -1 },
-    { ImageTransformParameterfvHP_names, ImageTransformParameterfvHP_remap_index, -1 },
-    { ImageTransformParameteriHP_names, ImageTransformParameteriHP_remap_index, -1 },
-    { GetImageTransformParameterivHP_names, GetImageTransformParameterivHP_remap_index, -1 },
-    { ImageTransformParameterivHP_names, ImageTransformParameterivHP_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_IBM_multimode_draw_arrays)
-static const struct dri_extension_function GL_IBM_multimode_draw_arrays_functions[] = {
-    { MultiModeDrawArraysIBM_names, MultiModeDrawArraysIBM_remap_index, -1 },
-    { MultiModeDrawElementsIBM_names, MultiModeDrawElementsIBM_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_IBM_vertex_array_lists)
-static const struct dri_extension_function GL_IBM_vertex_array_lists_functions[] = {
-    { SecondaryColorPointerListIBM_names, SecondaryColorPointerListIBM_remap_index, -1 },
-    { NormalPointerListIBM_names, NormalPointerListIBM_remap_index, -1 },
-    { FogCoordPointerListIBM_names, FogCoordPointerListIBM_remap_index, -1 },
-    { VertexPointerListIBM_names, VertexPointerListIBM_remap_index, -1 },
-    { ColorPointerListIBM_names, ColorPointerListIBM_remap_index, -1 },
-    { TexCoordPointerListIBM_names, TexCoordPointerListIBM_remap_index, -1 },
-    { IndexPointerListIBM_names, IndexPointerListIBM_remap_index, -1 },
-    { EdgeFlagPointerListIBM_names, EdgeFlagPointerListIBM_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_INGR_blend_func_separate)
-static const struct dri_extension_function GL_INGR_blend_func_separate_functions[] = {
-    { BlendFuncSeparateEXT_names, BlendFuncSeparateEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_INTEL_parallel_arrays)
-static const struct dri_extension_function GL_INTEL_parallel_arrays_functions[] = {
-    { VertexPointervINTEL_names, VertexPointervINTEL_remap_index, -1 },
-    { ColorPointervINTEL_names, ColorPointervINTEL_remap_index, -1 },
-    { NormalPointervINTEL_names, NormalPointervINTEL_remap_index, -1 },
-    { TexCoordPointervINTEL_names, TexCoordPointervINTEL_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_MESA_resize_buffers)
-static const struct dri_extension_function GL_MESA_resize_buffers_functions[] = {
-    { ResizeBuffersMESA_names, ResizeBuffersMESA_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_MESA_shader_debug)
-static const struct dri_extension_function GL_MESA_shader_debug_functions[] = {
-    { GetDebugLogLengthMESA_names, GetDebugLogLengthMESA_remap_index, -1 },
-    { ClearDebugLogMESA_names, ClearDebugLogMESA_remap_index, -1 },
-    { GetDebugLogMESA_names, GetDebugLogMESA_remap_index, -1 },
-    { CreateDebugObjectMESA_names, CreateDebugObjectMESA_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_MESA_window_pos)
-static const struct dri_extension_function GL_MESA_window_pos_functions[] = {
-    { WindowPos3fMESA_names, WindowPos3fMESA_remap_index, -1 },
-    { WindowPos2dvMESA_names, WindowPos2dvMESA_remap_index, -1 },
-    { WindowPos4svMESA_names, WindowPos4svMESA_remap_index, -1 },
-    { WindowPos2svMESA_names, WindowPos2svMESA_remap_index, -1 },
-    { WindowPos3dMESA_names, WindowPos3dMESA_remap_index, -1 },
-    { WindowPos2fvMESA_names, WindowPos2fvMESA_remap_index, -1 },
-    { WindowPos4dMESA_names, WindowPos4dMESA_remap_index, -1 },
-    { WindowPos2dMESA_names, WindowPos2dMESA_remap_index, -1 },
-    { WindowPos4ivMESA_names, WindowPos4ivMESA_remap_index, -1 },
-    { WindowPos4fMESA_names, WindowPos4fMESA_remap_index, -1 },
-    { WindowPos3dvMESA_names, WindowPos3dvMESA_remap_index, -1 },
-    { WindowPos3fvMESA_names, WindowPos3fvMESA_remap_index, -1 },
-    { WindowPos4dvMESA_names, WindowPos4dvMESA_remap_index, -1 },
-    { WindowPos2iMESA_names, WindowPos2iMESA_remap_index, -1 },
-    { WindowPos3sMESA_names, WindowPos3sMESA_remap_index, -1 },
-    { WindowPos4sMESA_names, WindowPos4sMESA_remap_index, -1 },
-    { WindowPos2ivMESA_names, WindowPos2ivMESA_remap_index, -1 },
-    { WindowPos2sMESA_names, WindowPos2sMESA_remap_index, -1 },
-    { WindowPos3iMESA_names, WindowPos3iMESA_remap_index, -1 },
-    { WindowPos3ivMESA_names, WindowPos3ivMESA_remap_index, -1 },
-    { WindowPos4iMESA_names, WindowPos4iMESA_remap_index, -1 },
-    { WindowPos4fvMESA_names, WindowPos4fvMESA_remap_index, -1 },
-    { WindowPos3svMESA_names, WindowPos3svMESA_remap_index, -1 },
-    { WindowPos2fMESA_names, WindowPos2fMESA_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_evaluators)
-static const struct dri_extension_function GL_NV_evaluators_functions[] = {
-    { GetMapAttribParameterivNV_names, GetMapAttribParameterivNV_remap_index, -1 },
-    { MapControlPointsNV_names, MapControlPointsNV_remap_index, -1 },
-    { MapParameterfvNV_names, MapParameterfvNV_remap_index, -1 },
-    { EvalMapsNV_names, EvalMapsNV_remap_index, -1 },
-    { GetMapAttribParameterfvNV_names, GetMapAttribParameterfvNV_remap_index, -1 },
-    { MapParameterivNV_names, MapParameterivNV_remap_index, -1 },
-    { GetMapParameterivNV_names, GetMapParameterivNV_remap_index, -1 },
-    { GetMapParameterfvNV_names, GetMapParameterfvNV_remap_index, -1 },
-    { GetMapControlPointsNV_names, GetMapControlPointsNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_fence)
-static const struct dri_extension_function GL_NV_fence_functions[] = {
-    { GenFencesNV_names, GenFencesNV_remap_index, -1 },
-    { TestFenceNV_names, TestFenceNV_remap_index, -1 },
-    { IsFenceNV_names, IsFenceNV_remap_index, -1 },
-    { DeleteFencesNV_names, DeleteFencesNV_remap_index, -1 },
-    { SetFenceNV_names, SetFenceNV_remap_index, -1 },
-    { GetFenceivNV_names, GetFenceivNV_remap_index, -1 },
-    { FinishFenceNV_names, FinishFenceNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_fragment_program)
-static const struct dri_extension_function GL_NV_fragment_program_functions[] = {
-    { GetProgramNamedParameterdvNV_names, GetProgramNamedParameterdvNV_remap_index, -1 },
-    { GetProgramNamedParameterfvNV_names, GetProgramNamedParameterfvNV_remap_index, -1 },
-    { ProgramNamedParameter4fNV_names, ProgramNamedParameter4fNV_remap_index, -1 },
-    { ProgramNamedParameter4fvNV_names, ProgramNamedParameter4fvNV_remap_index, -1 },
-    { ProgramNamedParameter4dvNV_names, ProgramNamedParameter4dvNV_remap_index, -1 },
-    { ProgramNamedParameter4dNV_names, ProgramNamedParameter4dNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_point_sprite)
-static const struct dri_extension_function GL_NV_point_sprite_functions[] = {
-    { PointParameteriNV_names, PointParameteriNV_remap_index, -1 },
-    { PointParameterivNV_names, PointParameterivNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_register_combiners)
-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 },
-    { GetCombinerOutputParameterivNV_names, GetCombinerOutputParameterivNV_remap_index, -1 },
-    { CombinerParameteriNV_names, CombinerParameteriNV_remap_index, -1 },
-    { GetFinalCombinerInputParameterivNV_names, GetFinalCombinerInputParameterivNV_remap_index, -1 },
-    { CombinerInputNV_names, CombinerInputNV_remap_index, -1 },
-    { CombinerParameterfNV_names, CombinerParameterfNV_remap_index, -1 },
-    { GetCombinerInputParameterivNV_names, GetCombinerInputParameterivNV_remap_index, -1 },
-    { CombinerParameterivNV_names, CombinerParameterivNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_register_combiners2)
-static const struct dri_extension_function GL_NV_register_combiners2_functions[] = {
-    { CombinerStageParameterfvNV_names, CombinerStageParameterfvNV_remap_index, -1 },
-    { GetCombinerStageParameterfvNV_names, GetCombinerStageParameterfvNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_vertex_array_range)
-static const struct dri_extension_function GL_NV_vertex_array_range_functions[] = {
-    { VertexArrayRangeNV_names, VertexArrayRangeNV_remap_index, -1 },
-    { FlushVertexArrayRangeNV_names, FlushVertexArrayRangeNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_NV_vertex_program)
-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 },
-    { VertexAttrib2dNV_names, VertexAttrib2dNV_remap_index, -1 },
-    { VertexAttrib4ubNV_names, VertexAttrib4ubNV_remap_index, -1 },
-    { VertexAttribs3dvNV_names, VertexAttribs3dvNV_remap_index, -1 },
-    { VertexAttribs4fvNV_names, VertexAttribs4fvNV_remap_index, -1 },
-    { VertexAttrib2sNV_names, VertexAttrib2sNV_remap_index, -1 },
-    { ProgramEnvParameter4fvARB_names, ProgramEnvParameter4fvARB_remap_index, -1 },
-    { LoadProgramNV_names, LoadProgramNV_remap_index, -1 },
-    { VertexAttrib4fvNV_names, VertexAttrib4fvNV_remap_index, -1 },
-    { VertexAttrib3fNV_names, VertexAttrib3fNV_remap_index, -1 },
-    { VertexAttribs2dvNV_names, VertexAttribs2dvNV_remap_index, -1 },
-    { GetProgramParameterfvNV_names, GetProgramParameterfvNV_remap_index, -1 },
-    { VertexAttrib3dNV_names, VertexAttrib3dNV_remap_index, -1 },
-    { VertexAttrib2fvNV_names, VertexAttrib2fvNV_remap_index, -1 },
-    { VertexAttrib2dvNV_names, VertexAttrib2dvNV_remap_index, -1 },
-    { VertexAttrib1dvNV_names, VertexAttrib1dvNV_remap_index, -1 },
-    { VertexAttrib1svNV_names, VertexAttrib1svNV_remap_index, -1 },
-    { ProgramEnvParameter4fARB_names, ProgramEnvParameter4fARB_remap_index, -1 },
-    { VertexAttribs2svNV_names, VertexAttribs2svNV_remap_index, -1 },
-    { GetVertexAttribivNV_names, GetVertexAttribivNV_remap_index, -1 },
-    { GetVertexAttribfvNV_names, GetVertexAttribfvNV_remap_index, -1 },
-    { VertexAttrib2svNV_names, VertexAttrib2svNV_remap_index, -1 },
-    { VertexAttribs1fvNV_names, VertexAttribs1fvNV_remap_index, -1 },
-    { IsProgramNV_names, IsProgramNV_remap_index, -1 },
-    { ProgramEnvParameter4dARB_names, ProgramEnvParameter4dARB_remap_index, -1 },
-    { VertexAttrib2fNV_names, VertexAttrib2fNV_remap_index, -1 },
-    { RequestResidentProgramsNV_names, RequestResidentProgramsNV_remap_index, -1 },
-    { ExecuteProgramNV_names, ExecuteProgramNV_remap_index, -1 },
-    { VertexAttribPointerNV_names, VertexAttribPointerNV_remap_index, -1 },
-    { TrackMatrixNV_names, TrackMatrixNV_remap_index, -1 },
-    { GetProgramParameterdvNV_names, GetProgramParameterdvNV_remap_index, -1 },
-    { GetTrackMatrixivNV_names, GetTrackMatrixivNV_remap_index, -1 },
-    { VertexAttrib3svNV_names, VertexAttrib3svNV_remap_index, -1 },
-    { ProgramParameters4fvNV_names, ProgramParameters4fvNV_remap_index, -1 },
-    { GetProgramivNV_names, GetProgramivNV_remap_index, -1 },
-    { GetVertexAttribdvNV_names, GetVertexAttribdvNV_remap_index, -1 },
-    { VertexAttrib3fvNV_names, VertexAttrib3fvNV_remap_index, -1 },
-    { ProgramEnvParameter4dvARB_names, ProgramEnvParameter4dvARB_remap_index, -1 },
-    { VertexAttribs2fvNV_names, VertexAttribs2fvNV_remap_index, -1 },
-    { DeleteProgramsNV_names, DeleteProgramsNV_remap_index, -1 },
-    { GetVertexAttribPointervNV_names, GetVertexAttribPointervNV_remap_index, -1 },
-    { GetProgramStringNV_names, GetProgramStringNV_remap_index, -1 },
-    { VertexAttrib4sNV_names, VertexAttrib4sNV_remap_index, -1 },
-    { VertexAttribs4dvNV_names, VertexAttribs4dvNV_remap_index, -1 },
-    { ProgramParameters4dvNV_names, ProgramParameters4dvNV_remap_index, -1 },
-    { VertexAttrib3sNV_names, VertexAttrib3sNV_remap_index, -1 },
-    { VertexAttrib1fNV_names, VertexAttrib1fNV_remap_index, -1 },
-    { VertexAttrib4dNV_names, VertexAttrib4dNV_remap_index, -1 },
-    { VertexAttribs4ubvNV_names, VertexAttribs4ubvNV_remap_index, -1 },
-    { VertexAttribs3svNV_names, VertexAttribs3svNV_remap_index, -1 },
-    { VertexAttrib1sNV_names, VertexAttrib1sNV_remap_index, -1 },
-    { BindProgramNV_names, BindProgramNV_remap_index, -1 },
-    { AreProgramsResidentNV_names, AreProgramsResidentNV_remap_index, -1 },
-    { VertexAttrib3dvNV_names, VertexAttrib3dvNV_remap_index, -1 },
-    { VertexAttrib1dNV_names, VertexAttrib1dNV_remap_index, -1 },
-    { VertexAttribs4svNV_names, VertexAttribs4svNV_remap_index, -1 },
-    { VertexAttribs1svNV_names, VertexAttribs1svNV_remap_index, -1 },
-    { GenProgramsNV_names, GenProgramsNV_remap_index, -1 },
-    { VertexAttrib4dvNV_names, VertexAttrib4dvNV_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_PGI_misc_hints)
-static const struct dri_extension_function GL_PGI_misc_hints_functions[] = {
-    { HintPGI_names, HintPGI_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_detail_texture)
-static const struct dri_extension_function GL_SGIS_detail_texture_functions[] = {
-    { GetDetailTexFuncSGIS_names, GetDetailTexFuncSGIS_remap_index, -1 },
-    { DetailTexFuncSGIS_names, DetailTexFuncSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_fog_function)
-static const struct dri_extension_function GL_SGIS_fog_function_functions[] = {
-    { FogFuncSGIS_names, FogFuncSGIS_remap_index, -1 },
-    { GetFogFuncSGIS_names, GetFogFuncSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_multisample)
-static const struct dri_extension_function GL_SGIS_multisample_functions[] = {
-    { SampleMaskSGIS_names, SampleMaskSGIS_remap_index, -1 },
-    { SamplePatternSGIS_names, SamplePatternSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_pixel_texture)
-static const struct dri_extension_function GL_SGIS_pixel_texture_functions[] = {
-    { PixelTexGenParameterfvSGIS_names, PixelTexGenParameterfvSGIS_remap_index, -1 },
-    { GetPixelTexGenParameterivSGIS_names, GetPixelTexGenParameterivSGIS_remap_index, -1 },
-    { PixelTexGenParameteriSGIS_names, PixelTexGenParameteriSGIS_remap_index, -1 },
-    { PixelTexGenParameterivSGIS_names, PixelTexGenParameterivSGIS_remap_index, -1 },
-    { PixelTexGenParameterfSGIS_names, PixelTexGenParameterfSGIS_remap_index, -1 },
-    { GetPixelTexGenParameterfvSGIS_names, GetPixelTexGenParameterfvSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_point_parameters)
-static const struct dri_extension_function GL_SGIS_point_parameters_functions[] = {
-    { PointParameterfEXT_names, PointParameterfEXT_remap_index, -1 },
-    { PointParameterfvEXT_names, PointParameterfvEXT_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_sharpen_texture)
-static const struct dri_extension_function GL_SGIS_sharpen_texture_functions[] = {
-    { GetSharpenTexFuncSGIS_names, GetSharpenTexFuncSGIS_remap_index, -1 },
-    { SharpenTexFuncSGIS_names, SharpenTexFuncSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_texture4D)
-static const struct dri_extension_function GL_SGIS_texture4D_functions[] = {
-    { TexImage4DSGIS_names, TexImage4DSGIS_remap_index, -1 },
-    { TexSubImage4DSGIS_names, TexSubImage4DSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_texture_color_mask)
-static const struct dri_extension_function GL_SGIS_texture_color_mask_functions[] = {
-    { TextureColorMaskSGIS_names, TextureColorMaskSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIS_texture_filter4)
-static const struct dri_extension_function GL_SGIS_texture_filter4_functions[] = {
-    { GetTexFilterFuncSGIS_names, GetTexFilterFuncSGIS_remap_index, -1 },
-    { TexFilterFuncSGIS_names, TexFilterFuncSGIS_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_async)
-static const struct dri_extension_function GL_SGIX_async_functions[] = {
-    { AsyncMarkerSGIX_names, AsyncMarkerSGIX_remap_index, -1 },
-    { FinishAsyncSGIX_names, FinishAsyncSGIX_remap_index, -1 },
-    { PollAsyncSGIX_names, PollAsyncSGIX_remap_index, -1 },
-    { DeleteAsyncMarkersSGIX_names, DeleteAsyncMarkersSGIX_remap_index, -1 },
-    { IsAsyncMarkerSGIX_names, IsAsyncMarkerSGIX_remap_index, -1 },
-    { GenAsyncMarkersSGIX_names, GenAsyncMarkersSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_flush_raster)
-static const struct dri_extension_function GL_SGIX_flush_raster_functions[] = {
-    { FlushRasterSGIX_names, FlushRasterSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_fragment_lighting)
-static const struct dri_extension_function GL_SGIX_fragment_lighting_functions[] = {
-    { FragmentMaterialfvSGIX_names, FragmentMaterialfvSGIX_remap_index, -1 },
-    { FragmentLightModelivSGIX_names, FragmentLightModelivSGIX_remap_index, -1 },
-    { FragmentLightiSGIX_names, FragmentLightiSGIX_remap_index, -1 },
-    { GetFragmentMaterialfvSGIX_names, GetFragmentMaterialfvSGIX_remap_index, -1 },
-    { FragmentMaterialfSGIX_names, FragmentMaterialfSGIX_remap_index, -1 },
-    { GetFragmentLightivSGIX_names, GetFragmentLightivSGIX_remap_index, -1 },
-    { FragmentLightModeliSGIX_names, FragmentLightModeliSGIX_remap_index, -1 },
-    { FragmentLightivSGIX_names, FragmentLightivSGIX_remap_index, -1 },
-    { GetFragmentMaterialivSGIX_names, GetFragmentMaterialivSGIX_remap_index, -1 },
-    { FragmentLightModelfSGIX_names, FragmentLightModelfSGIX_remap_index, -1 },
-    { FragmentColorMaterialSGIX_names, FragmentColorMaterialSGIX_remap_index, -1 },
-    { FragmentMaterialiSGIX_names, FragmentMaterialiSGIX_remap_index, -1 },
-    { LightEnviSGIX_names, LightEnviSGIX_remap_index, -1 },
-    { FragmentLightModelfvSGIX_names, FragmentLightModelfvSGIX_remap_index, -1 },
-    { FragmentLightfvSGIX_names, FragmentLightfvSGIX_remap_index, -1 },
-    { FragmentLightfSGIX_names, FragmentLightfSGIX_remap_index, -1 },
-    { GetFragmentLightfvSGIX_names, GetFragmentLightfvSGIX_remap_index, -1 },
-    { FragmentMaterialivSGIX_names, FragmentMaterialivSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_framezoom)
-static const struct dri_extension_function GL_SGIX_framezoom_functions[] = {
-    { FrameZoomSGIX_names, FrameZoomSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_igloo_interface)
-static const struct dri_extension_function GL_SGIX_igloo_interface_functions[] = {
-    { IglooInterfaceSGIX_names, IglooInterfaceSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_instruments)
-static const struct dri_extension_function GL_SGIX_instruments_functions[] = {
-    { ReadInstrumentsSGIX_names, ReadInstrumentsSGIX_remap_index, -1 },
-    { PollInstrumentsSGIX_names, PollInstrumentsSGIX_remap_index, -1 },
-    { GetInstrumentsSGIX_names, GetInstrumentsSGIX_remap_index, -1 },
-    { StartInstrumentsSGIX_names, StartInstrumentsSGIX_remap_index, -1 },
-    { StopInstrumentsSGIX_names, StopInstrumentsSGIX_remap_index, -1 },
-    { InstrumentsBufferSGIX_names, InstrumentsBufferSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_list_priority)
-static const struct dri_extension_function GL_SGIX_list_priority_functions[] = {
-    { ListParameterfSGIX_names, ListParameterfSGIX_remap_index, -1 },
-    { GetListParameterfvSGIX_names, GetListParameterfvSGIX_remap_index, -1 },
-    { ListParameteriSGIX_names, ListParameteriSGIX_remap_index, -1 },
-    { ListParameterfvSGIX_names, ListParameterfvSGIX_remap_index, -1 },
-    { ListParameterivSGIX_names, ListParameterivSGIX_remap_index, -1 },
-    { GetListParameterivSGIX_names, GetListParameterivSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_pixel_texture)
-static const struct dri_extension_function GL_SGIX_pixel_texture_functions[] = {
-    { PixelTexGenSGIX_names, PixelTexGenSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const struct dri_extension_function GL_SGIX_polynomial_ffd_functions[] = {
-    { LoadIdentityDeformationMapSGIX_names, LoadIdentityDeformationMapSGIX_remap_index, -1 },
-    { DeformationMap3dSGIX_names, DeformationMap3dSGIX_remap_index, -1 },
-    { DeformSGIX_names, DeformSGIX_remap_index, -1 },
-    { DeformationMap3fSGIX_names, DeformationMap3fSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_reference_plane)
-static const struct dri_extension_function GL_SGIX_reference_plane_functions[] = {
-    { ReferencePlaneSGIX_names, ReferencePlaneSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_sprite)
-static const struct dri_extension_function GL_SGIX_sprite_functions[] = {
-    { SpriteParameterfvSGIX_names, SpriteParameterfvSGIX_remap_index, -1 },
-    { SpriteParameteriSGIX_names, SpriteParameteriSGIX_remap_index, -1 },
-    { SpriteParameterfSGIX_names, SpriteParameterfSGIX_remap_index, -1 },
-    { SpriteParameterivSGIX_names, SpriteParameterivSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGIX_tag_sample_buffer)
-static const struct dri_extension_function GL_SGIX_tag_sample_buffer_functions[] = {
-    { TagSampleBufferSGIX_names, TagSampleBufferSGIX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SGI_color_table)
-static const struct dri_extension_function GL_SGI_color_table_functions[] = {
-    { ColorTableParameteriv_names, -1, 341 },
-    { ColorTable_names, -1, 339 },
-    { GetColorTable_names, -1, 343 },
-    { CopyColorTable_names, -1, 342 },
-    { ColorTableParameterfv_names, -1, 340 },
-    { GetColorTableParameterfv_names, -1, 344 },
-    { GetColorTableParameteriv_names, -1, 345 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SUNX_constant_data)
-static const struct dri_extension_function GL_SUNX_constant_data_functions[] = {
-    { FinishTextureSUNX_names, FinishTextureSUNX_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SUN_global_alpha)
-static const struct dri_extension_function GL_SUN_global_alpha_functions[] = {
-    { GlobalAlphaFactorubSUN_names, GlobalAlphaFactorubSUN_remap_index, -1 },
-    { GlobalAlphaFactoriSUN_names, GlobalAlphaFactoriSUN_remap_index, -1 },
-    { GlobalAlphaFactordSUN_names, GlobalAlphaFactordSUN_remap_index, -1 },
-    { GlobalAlphaFactoruiSUN_names, GlobalAlphaFactoruiSUN_remap_index, -1 },
-    { GlobalAlphaFactorbSUN_names, GlobalAlphaFactorbSUN_remap_index, -1 },
-    { GlobalAlphaFactorfSUN_names, GlobalAlphaFactorfSUN_remap_index, -1 },
-    { GlobalAlphaFactorusSUN_names, GlobalAlphaFactorusSUN_remap_index, -1 },
-    { GlobalAlphaFactorsSUN_names, GlobalAlphaFactorsSUN_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SUN_mesh_array)
-static const struct dri_extension_function GL_SUN_mesh_array_functions[] = {
-    { DrawMeshArraysSUN_names, DrawMeshArraysSUN_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SUN_triangle_list)
-static const struct dri_extension_function GL_SUN_triangle_list_functions[] = {
-    { ReplacementCodeubSUN_names, ReplacementCodeubSUN_remap_index, -1 },
-    { ReplacementCodeubvSUN_names, ReplacementCodeubvSUN_remap_index, -1 },
-    { ReplacementCodeusvSUN_names, ReplacementCodeusvSUN_remap_index, -1 },
-    { ReplacementCodePointerSUN_names, ReplacementCodePointerSUN_remap_index, -1 },
-    { ReplacementCodeusSUN_names, ReplacementCodeusSUN_remap_index, -1 },
-    { ReplacementCodeuiSUN_names, ReplacementCodeuiSUN_remap_index, -1 },
-    { ReplacementCodeuivSUN_names, ReplacementCodeuivSUN_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_SUN_vertex)
-static const struct dri_extension_function GL_SUN_vertex_functions[] = {
-    { ReplacementCodeuiColor3fVertex3fvSUN_names, ReplacementCodeuiColor3fVertex3fvSUN_remap_index, -1 },
-    { TexCoord4fColor4fNormal3fVertex4fvSUN_names, TexCoord4fColor4fNormal3fVertex4fvSUN_remap_index, -1 },
-    { TexCoord2fColor4ubVertex3fvSUN_names, TexCoord2fColor4ubVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiVertex3fvSUN_names, ReplacementCodeuiVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fVertex3fvSUN_names, ReplacementCodeuiTexCoord2fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiNormal3fVertex3fSUN_names, ReplacementCodeuiNormal3fVertex3fSUN_remap_index, -1 },
-    { Color4ubVertex3fvSUN_names, Color4ubVertex3fvSUN_remap_index, -1 },
-    { Color4ubVertex3fSUN_names, Color4ubVertex3fSUN_remap_index, -1 },
-    { TexCoord2fVertex3fSUN_names, TexCoord2fVertex3fSUN_remap_index, -1 },
-    { TexCoord2fColor4fNormal3fVertex3fSUN_names, TexCoord2fColor4fNormal3fVertex3fSUN_remap_index, -1 },
-    { TexCoord2fNormal3fVertex3fvSUN_names, TexCoord2fNormal3fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fNormal3fVertex3fSUN_names, ReplacementCodeuiTexCoord2fNormal3fVertex3fSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fVertex3fSUN_names, ReplacementCodeuiTexCoord2fVertex3fSUN_remap_index, -1 },
-    { TexCoord2fNormal3fVertex3fSUN_names, TexCoord2fNormal3fVertex3fSUN_remap_index, -1 },
-    { Color3fVertex3fSUN_names, Color3fVertex3fSUN_remap_index, -1 },
-    { Color3fVertex3fvSUN_names, Color3fVertex3fvSUN_remap_index, -1 },
-    { Color4fNormal3fVertex3fvSUN_names, Color4fNormal3fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN_names, ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiColor4fNormal3fVertex3fvSUN_names, ReplacementCodeuiColor4fNormal3fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN_names, ReplacementCodeuiTexCoord2fNormal3fVertex3fvSUN_remap_index, -1 },
-    { TexCoord2fColor3fVertex3fSUN_names, TexCoord2fColor3fVertex3fSUN_remap_index, -1 },
-    { TexCoord4fColor4fNormal3fVertex4fSUN_names, TexCoord4fColor4fNormal3fVertex4fSUN_remap_index, -1 },
-    { Color4ubVertex2fvSUN_names, Color4ubVertex2fvSUN_remap_index, -1 },
-    { Normal3fVertex3fSUN_names, Normal3fVertex3fSUN_remap_index, -1 },
-    { ReplacementCodeuiColor4fNormal3fVertex3fSUN_names, ReplacementCodeuiColor4fNormal3fVertex3fSUN_remap_index, -1 },
-    { TexCoord2fColor4fNormal3fVertex3fvSUN_names, TexCoord2fColor4fNormal3fVertex3fvSUN_remap_index, -1 },
-    { TexCoord2fVertex3fvSUN_names, TexCoord2fVertex3fvSUN_remap_index, -1 },
-    { Color4ubVertex2fSUN_names, Color4ubVertex2fSUN_remap_index, -1 },
-    { ReplacementCodeuiColor4ubVertex3fSUN_names, ReplacementCodeuiColor4ubVertex3fSUN_remap_index, -1 },
-    { TexCoord2fColor4ubVertex3fSUN_names, TexCoord2fColor4ubVertex3fSUN_remap_index, -1 },
-    { Normal3fVertex3fvSUN_names, Normal3fVertex3fvSUN_remap_index, -1 },
-    { Color4fNormal3fVertex3fSUN_names, Color4fNormal3fVertex3fSUN_remap_index, -1 },
-    { ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN_names, ReplacementCodeuiTexCoord2fColor4fNormal3fVertex3fSUN_remap_index, -1 },
-    { ReplacementCodeuiColor4ubVertex3fvSUN_names, ReplacementCodeuiColor4ubVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiColor3fVertex3fSUN_names, ReplacementCodeuiColor3fVertex3fSUN_remap_index, -1 },
-    { TexCoord4fVertex4fSUN_names, TexCoord4fVertex4fSUN_remap_index, -1 },
-    { TexCoord2fColor3fVertex3fvSUN_names, TexCoord2fColor3fVertex3fvSUN_remap_index, -1 },
-    { ReplacementCodeuiNormal3fVertex3fvSUN_names, ReplacementCodeuiNormal3fVertex3fvSUN_remap_index, -1 },
-    { TexCoord4fVertex4fvSUN_names, TexCoord4fVertex4fvSUN_remap_index, -1 },
-    { ReplacementCodeuiVertex3fSUN_names, ReplacementCodeuiVertex3fSUN_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_VERSION_1_3)
-static const struct dri_extension_function GL_VERSION_1_3_functions[] = {
-    { SampleCoverageARB_names, SampleCoverageARB_remap_index, -1 },
-    { MultiTexCoord3sARB_names, -1, 398 },
-    { ActiveTextureARB_names, -1, 374 },
-    { CompressedTexSubImage2DARB_names, CompressedTexSubImage2DARB_remap_index, -1 },
-    { CompressedTexImage3DARB_names, CompressedTexImage3DARB_remap_index, -1 },
-    { MultiTexCoord1fvARB_names, -1, 379 },
-    { MultTransposeMatrixdARB_names, MultTransposeMatrixdARB_remap_index, -1 },
-    { CompressedTexImage1DARB_names, CompressedTexImage1DARB_remap_index, -1 },
-    { MultiTexCoord3dARB_names, -1, 392 },
-    { MultiTexCoord2iARB_names, -1, 388 },
-    { MultiTexCoord2svARB_names, -1, 391 },
-    { MultiTexCoord2fARB_names, -1, 386 },
-    { LoadTransposeMatrixdARB_names, LoadTransposeMatrixdARB_remap_index, -1 },
-    { MultiTexCoord3fvARB_names, -1, 395 },
-    { MultiTexCoord4sARB_names, -1, 406 },
-    { MultiTexCoord2dvARB_names, -1, 385 },
-    { MultiTexCoord1svARB_names, -1, 383 },
-    { MultiTexCoord3svARB_names, -1, 399 },
-    { MultiTexCoord4iARB_names, -1, 404 },
-    { MultiTexCoord3iARB_names, -1, 396 },
-    { MultiTexCoord1dARB_names, -1, 376 },
-    { MultiTexCoord3dvARB_names, -1, 393 },
-    { MultiTexCoord3ivARB_names, -1, 397 },
-    { MultiTexCoord2sARB_names, -1, 390 },
-    { MultiTexCoord4ivARB_names, -1, 405 },
-    { CompressedTexSubImage1DARB_names, CompressedTexSubImage1DARB_remap_index, -1 },
-    { ClientActiveTextureARB_names, -1, 375 },
-    { CompressedTexSubImage3DARB_names, CompressedTexSubImage3DARB_remap_index, -1 },
-    { MultiTexCoord2dARB_names, -1, 384 },
-    { MultiTexCoord4dvARB_names, -1, 401 },
-    { MultiTexCoord4fvARB_names, -1, 403 },
-    { MultiTexCoord3fARB_names, -1, 394 },
-    { MultTransposeMatrixfARB_names, MultTransposeMatrixfARB_remap_index, -1 },
-    { CompressedTexImage2DARB_names, CompressedTexImage2DARB_remap_index, -1 },
-    { MultiTexCoord4dARB_names, -1, 400 },
-    { MultiTexCoord1sARB_names, -1, 382 },
-    { MultiTexCoord1dvARB_names, -1, 377 },
-    { MultiTexCoord1ivARB_names, -1, 381 },
-    { MultiTexCoord2ivARB_names, -1, 389 },
-    { MultiTexCoord1iARB_names, -1, 380 },
-    { GetCompressedTexImageARB_names, GetCompressedTexImageARB_remap_index, -1 },
-    { MultiTexCoord4svARB_names, -1, 407 },
-    { MultiTexCoord1fARB_names, -1, 378 },
-    { MultiTexCoord4fARB_names, -1, 402 },
-    { LoadTransposeMatrixfARB_names, LoadTransposeMatrixfARB_remap_index, -1 },
-    { MultiTexCoord2fvARB_names, -1, 387 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_VERSION_1_4)
-static const struct dri_extension_function GL_VERSION_1_4_functions[] = {
-    { PointParameteriNV_names, PointParameteriNV_remap_index, -1 },
-    { SecondaryColor3iEXT_names, SecondaryColor3iEXT_remap_index, -1 },
-    { WindowPos3fMESA_names, WindowPos3fMESA_remap_index, -1 },
-    { WindowPos2dvMESA_names, WindowPos2dvMESA_remap_index, -1 },
-    { SecondaryColor3bEXT_names, SecondaryColor3bEXT_remap_index, -1 },
-    { PointParameterfEXT_names, PointParameterfEXT_remap_index, -1 },
-    { FogCoorddEXT_names, FogCoorddEXT_remap_index, -1 },
-    { FogCoordfEXT_names, FogCoordfEXT_remap_index, -1 },
-    { WindowPos2svMESA_names, WindowPos2svMESA_remap_index, -1 },
-    { WindowPos3dMESA_names, WindowPos3dMESA_remap_index, -1 },
-    { PointParameterfvEXT_names, PointParameterfvEXT_remap_index, -1 },
-    { WindowPos2fvMESA_names, WindowPos2fvMESA_remap_index, -1 },
-    { SecondaryColor3bvEXT_names, SecondaryColor3bvEXT_remap_index, -1 },
-    { SecondaryColor3sEXT_names, SecondaryColor3sEXT_remap_index, -1 },
-    { SecondaryColor3dEXT_names, SecondaryColor3dEXT_remap_index, -1 },
-    { WindowPos2dMESA_names, WindowPos2dMESA_remap_index, -1 },
-    { SecondaryColorPointerEXT_names, SecondaryColorPointerEXT_remap_index, -1 },
-    { SecondaryColor3uiEXT_names, SecondaryColor3uiEXT_remap_index, -1 },
-    { SecondaryColor3usvEXT_names, SecondaryColor3usvEXT_remap_index, -1 },
-    { WindowPos3dvMESA_names, WindowPos3dvMESA_remap_index, -1 },
-    { PointParameterivNV_names, PointParameterivNV_remap_index, -1 },
-    { WindowPos3fvMESA_names, WindowPos3fvMESA_remap_index, -1 },
-    { SecondaryColor3ivEXT_names, SecondaryColor3ivEXT_remap_index, -1 },
-    { WindowPos2iMESA_names, WindowPos2iMESA_remap_index, -1 },
-    { SecondaryColor3fvEXT_names, SecondaryColor3fvEXT_remap_index, -1 },
-    { WindowPos3sMESA_names, WindowPos3sMESA_remap_index, -1 },
-    { WindowPos2ivMESA_names, WindowPos2ivMESA_remap_index, -1 },
-    { MultiDrawElementsEXT_names, MultiDrawElementsEXT_remap_index, -1 },
-    { WindowPos2sMESA_names, WindowPos2sMESA_remap_index, -1 },
-    { FogCoordPointerEXT_names, FogCoordPointerEXT_remap_index, -1 },
-    { SecondaryColor3ubvEXT_names, SecondaryColor3ubvEXT_remap_index, -1 },
-    { SecondaryColor3uivEXT_names, SecondaryColor3uivEXT_remap_index, -1 },
-    { WindowPos3iMESA_names, WindowPos3iMESA_remap_index, -1 },
-    { SecondaryColor3dvEXT_names, SecondaryColor3dvEXT_remap_index, -1 },
-    { MultiDrawArraysEXT_names, MultiDrawArraysEXT_remap_index, -1 },
-    { SecondaryColor3usEXT_names, SecondaryColor3usEXT_remap_index, -1 },
-    { FogCoordfvEXT_names, FogCoordfvEXT_remap_index, -1 },
-    { SecondaryColor3ubEXT_names, SecondaryColor3ubEXT_remap_index, -1 },
-    { BlendFuncSeparateEXT_names, BlendFuncSeparateEXT_remap_index, -1 },
-    { SecondaryColor3fEXT_names, SecondaryColor3fEXT_remap_index, -1 },
-    { WindowPos3ivMESA_names, WindowPos3ivMESA_remap_index, -1 },
-    { SecondaryColor3svEXT_names, SecondaryColor3svEXT_remap_index, -1 },
-    { FogCoorddvEXT_names, FogCoorddvEXT_remap_index, -1 },
-    { WindowPos3svMESA_names, WindowPos3svMESA_remap_index, -1 },
-    { WindowPos2fMESA_names, WindowPos2fMESA_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_VERSION_1_5)
-static const struct dri_extension_function GL_VERSION_1_5_functions[] = {
-    { BeginQueryARB_names, BeginQueryARB_remap_index, -1 },
-    { GetBufferSubDataARB_names, GetBufferSubDataARB_remap_index, -1 },
-    { BufferSubDataARB_names, BufferSubDataARB_remap_index, -1 },
-    { GetQueryivARB_names, GetQueryivARB_remap_index, -1 },
-    { GetQueryObjectivARB_names, GetQueryObjectivARB_remap_index, -1 },
-    { BufferDataARB_names, BufferDataARB_remap_index, -1 },
-    { EndQueryARB_names, EndQueryARB_remap_index, -1 },
-    { GetBufferPointervARB_names, GetBufferPointervARB_remap_index, -1 },
-    { GetQueryObjectuivARB_names, GetQueryObjectuivARB_remap_index, -1 },
-    { GetBufferParameterivARB_names, GetBufferParameterivARB_remap_index, -1 },
-    { DeleteQueriesARB_names, DeleteQueriesARB_remap_index, -1 },
-    { IsQueryARB_names, IsQueryARB_remap_index, -1 },
-    { MapBufferARB_names, MapBufferARB_remap_index, -1 },
-    { GenQueriesARB_names, GenQueriesARB_remap_index, -1 },
-    { IsBufferARB_names, IsBufferARB_remap_index, -1 },
-    { DeleteBuffersARB_names, DeleteBuffersARB_remap_index, -1 },
-    { UnmapBufferARB_names, UnmapBufferARB_remap_index, -1 },
-    { BindBufferARB_names, BindBufferARB_remap_index, -1 },
-    { GenBuffersARB_names, GenBuffersARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_VERSION_2_0)
-static const struct dri_extension_function GL_VERSION_2_0_functions[] = {
-    { UniformMatrix3fvARB_names, UniformMatrix3fvARB_remap_index, -1 },
-    { GetProgramiv_names, GetProgramiv_remap_index, -1 },
-    { BlendEquationSeparateEXT_names, BlendEquationSeparateEXT_remap_index, -1 },
-    { AttachShader_names, AttachShader_remap_index, -1 },
-    { VertexAttrib2fARB_names, VertexAttrib2fARB_remap_index, -1 },
-    { VertexAttrib3fARB_names, VertexAttrib3fARB_remap_index, -1 },
-    { Uniform2fARB_names, Uniform2fARB_remap_index, -1 },
-    { VertexAttrib1svARB_names, VertexAttrib1svARB_remap_index, -1 },
-    { Uniform2ivARB_names, Uniform2ivARB_remap_index, -1 },
-    { UniformMatrix4fvARB_names, UniformMatrix4fvARB_remap_index, -1 },
-    { VertexAttrib4NusvARB_names, VertexAttrib4NusvARB_remap_index, -1 },
-    { DisableVertexAttribArrayARB_names, DisableVertexAttribArrayARB_remap_index, -1 },
-    { StencilMaskSeparate_names, StencilMaskSeparate_remap_index, -1 },
-    { VertexAttrib1fARB_names, VertexAttrib1fARB_remap_index, -1 },
-    { GetProgramInfoLog_names, GetProgramInfoLog_remap_index, -1 },
-    { VertexAttrib4NbvARB_names, VertexAttrib4NbvARB_remap_index, -1 },
-    { GetActiveAttribARB_names, GetActiveAttribARB_remap_index, -1 },
-    { Uniform3iARB_names, Uniform3iARB_remap_index, -1 },
-    { GetShaderInfoLog_names, GetShaderInfoLog_remap_index, -1 },
-    { VertexAttrib1sARB_names, VertexAttrib1sARB_remap_index, -1 },
-    { Uniform1fARB_names, Uniform1fARB_remap_index, -1 },
-    { StencilOpSeparate_names, StencilOpSeparate_remap_index, -1 },
-    { UniformMatrix2fvARB_names, UniformMatrix2fvARB_remap_index, -1 },
-    { VertexAttrib3dvARB_names, VertexAttrib3dvARB_remap_index, -1 },
-    { Uniform3fvARB_names, Uniform3fvARB_remap_index, -1 },
-    { GetVertexAttribivARB_names, GetVertexAttribivARB_remap_index, -1 },
-    { CreateProgram_names, CreateProgram_remap_index, -1 },
-    { StencilFuncSeparate_names, StencilFuncSeparate_remap_index, -1 },
-    { VertexAttrib4ivARB_names, VertexAttrib4ivARB_remap_index, -1 },
-    { VertexAttrib4bvARB_names, VertexAttrib4bvARB_remap_index, -1 },
-    { VertexAttrib3dARB_names, VertexAttrib3dARB_remap_index, -1 },
-    { VertexAttrib4fARB_names, VertexAttrib4fARB_remap_index, -1 },
-    { VertexAttrib4fvARB_names, VertexAttrib4fvARB_remap_index, -1 },
-    { GetActiveUniformARB_names, GetActiveUniformARB_remap_index, -1 },
-    { IsShader_names, IsShader_remap_index, -1 },
-    { GetUniformivARB_names, GetUniformivARB_remap_index, -1 },
-    { IsProgram_names, IsProgram_remap_index, -1 },
-    { Uniform2fvARB_names, Uniform2fvARB_remap_index, -1 },
-    { ValidateProgramARB_names, ValidateProgramARB_remap_index, -1 },
-    { VertexAttrib2dARB_names, VertexAttrib2dARB_remap_index, -1 },
-    { VertexAttrib1dvARB_names, VertexAttrib1dvARB_remap_index, -1 },
-    { GetVertexAttribfvARB_names, GetVertexAttribfvARB_remap_index, -1 },
-    { GetAttribLocationARB_names, GetAttribLocationARB_remap_index, -1 },
-    { VertexAttrib4ubvARB_names, VertexAttrib4ubvARB_remap_index, -1 },
-    { Uniform3ivARB_names, Uniform3ivARB_remap_index, -1 },
-    { VertexAttrib4sARB_names, VertexAttrib4sARB_remap_index, -1 },
-    { VertexAttrib2dvARB_names, VertexAttrib2dvARB_remap_index, -1 },
-    { VertexAttrib2fvARB_names, VertexAttrib2fvARB_remap_index, -1 },
-    { VertexAttrib4NivARB_names, VertexAttrib4NivARB_remap_index, -1 },
-    { GetAttachedShaders_names, GetAttachedShaders_remap_index, -1 },
-    { CompileShaderARB_names, CompileShaderARB_remap_index, -1 },
-    { DeleteShader_names, DeleteShader_remap_index, -1 },
-    { Uniform3fARB_names, Uniform3fARB_remap_index, -1 },
-    { VertexAttrib4NuivARB_names, VertexAttrib4NuivARB_remap_index, -1 },
-    { Uniform4fARB_names, Uniform4fARB_remap_index, -1 },
-    { VertexAttrib1dARB_names, VertexAttrib1dARB_remap_index, -1 },
-    { VertexAttrib4usvARB_names, VertexAttrib4usvARB_remap_index, -1 },
-    { LinkProgramARB_names, LinkProgramARB_remap_index, -1 },
-    { ShaderSourceARB_names, ShaderSourceARB_remap_index, -1 },
-    { VertexAttrib3svARB_names, VertexAttrib3svARB_remap_index, -1 },
-    { Uniform4ivARB_names, Uniform4ivARB_remap_index, -1 },
-    { GetVertexAttribdvARB_names, GetVertexAttribdvARB_remap_index, -1 },
-    { Uniform1ivARB_names, Uniform1ivARB_remap_index, -1 },
-    { VertexAttrib4dvARB_names, VertexAttrib4dvARB_remap_index, -1 },
-    { BindAttribLocationARB_names, BindAttribLocationARB_remap_index, -1 },
-    { Uniform1iARB_names, Uniform1iARB_remap_index, -1 },
-    { VertexAttribPointerARB_names, VertexAttribPointerARB_remap_index, -1 },
-    { VertexAttrib4NsvARB_names, VertexAttrib4NsvARB_remap_index, -1 },
-    { VertexAttrib3fvARB_names, VertexAttrib3fvARB_remap_index, -1 },
-    { CreateShader_names, CreateShader_remap_index, -1 },
-    { DetachShader_names, DetachShader_remap_index, -1 },
-    { VertexAttrib4NubARB_names, VertexAttrib4NubARB_remap_index, -1 },
-    { Uniform4fvARB_names, Uniform4fvARB_remap_index, -1 },
-    { GetUniformfvARB_names, GetUniformfvARB_remap_index, -1 },
-    { Uniform4iARB_names, Uniform4iARB_remap_index, -1 },
-    { UseProgramObjectARB_names, UseProgramObjectARB_remap_index, -1 },
-    { DeleteProgram_names, DeleteProgram_remap_index, -1 },
-    { GetVertexAttribPointervNV_names, GetVertexAttribPointervNV_remap_index, -1 },
-    { Uniform2iARB_names, Uniform2iARB_remap_index, -1 },
-    { VertexAttrib4dARB_names, VertexAttrib4dARB_remap_index, -1 },
-    { GetUniformLocationARB_names, GetUniformLocationARB_remap_index, -1 },
-    { VertexAttrib3sARB_names, VertexAttrib3sARB_remap_index, -1 },
-    { GetShaderSourceARB_names, GetShaderSourceARB_remap_index, -1 },
-    { DrawBuffersARB_names, DrawBuffersARB_remap_index, -1 },
-    { Uniform1fvARB_names, Uniform1fvARB_remap_index, -1 },
-    { EnableVertexAttribArrayARB_names, EnableVertexAttribArrayARB_remap_index, -1 },
-    { VertexAttrib4uivARB_names, VertexAttrib4uivARB_remap_index, -1 },
-    { VertexAttrib4svARB_names, VertexAttrib4svARB_remap_index, -1 },
-    { GetShaderiv_names, GetShaderiv_remap_index, -1 },
-    { VertexAttrib2svARB_names, VertexAttrib2svARB_remap_index, -1 },
-    { VertexAttrib4NubvARB_names, VertexAttrib4NubvARB_remap_index, -1 },
-    { VertexAttrib2sARB_names, VertexAttrib2sARB_remap_index, -1 },
-    { VertexAttrib1fvARB_names, VertexAttrib1fvARB_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
-#if defined(need_GL_VERSION_2_1)
-static const struct dri_extension_function GL_VERSION_2_1_functions[] = {
-    { UniformMatrix2x4fv_names, UniformMatrix2x4fv_remap_index, -1 },
-    { UniformMatrix4x3fv_names, UniformMatrix4x3fv_remap_index, -1 },
-    { UniformMatrix4x2fv_names, UniformMatrix4x2fv_remap_index, -1 },
-    { UniformMatrix2x3fv_names, UniformMatrix2x3fv_remap_index, -1 },
-    { UniformMatrix3x4fv_names, UniformMatrix3x4fv_remap_index, -1 },
-    { UniformMatrix3x2fv_names, UniformMatrix3x2fv_remap_index, -1 },
-    { NULL, 0, 0 }
-};
-#endif
-
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h
index 89c815722f..447f3d15b9 100644
--- a/src/mesa/drivers/dri/common/spantmp2.h
+++ b/src/mesa/drivers/dri/common/spantmp2.h
@@ -82,6 +82,72 @@
       rgba[3] = 0xff;							\
    } while (0)
 
+#elif (SPANTMP_PIXEL_FMT == GL_RGB)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_5_6_5_REV)
+
+/**
+ ** GL_RGB, GL_UNSIGNED_SHORT_5_6_5_REV
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+  p = PACK_COLOR_565_REV( color[0], color[1], color[2] )
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_565_REV( r, g, b ))
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      p = p << 8 | p >> 8;						\
+      rgba[0] = ((p >> 8) & 0xf8) * 255 / 0xf8;				\
+      rgba[1] = ((p >> 3) & 0xfc) * 255 / 0xfc;				\
+      rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8;				\
+      rgba[3] = 0xff;							\
+   } while (0)
+
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_4_4_4_4)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_SHORT_4_4_4_4
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+   p = PACK_COLOR_4444_REV(color[3], color[0], color[1], color[2])
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_4444_REV(a, r, g, b))			\
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      rgba[0] = ((p >> 0) & 0xf) * 0x11;				\
+      rgba[1] = ((p >> 12) & 0xf) * 0x11;				\
+      rgba[2] = ((p >> 4) & 0xf) * 0x11;				\
+      rgba[3] = ((p >> 8) & 0xf) * 0x11;				\
+   } while (0)
+
+
 #elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_4_4_4_4_REV)
 
 /**
@@ -147,6 +213,39 @@
       rgba[3] = ((p >> 15) & 0x1) * 0xff;				\
    } while (0)
 
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA)  && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_SHORT_1_5_5_5)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (buf + (_x) * 2 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLushort *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLushort *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+#define INIT_MONO_PIXEL(p, color) \
+   p = PACK_COLOR_1555_REV(color[3], color[0], color[1], color[2])
+
+#define WRITE_RGBA( _x, _y, r, g, b, a )				\
+   PUT_VALUE(_x, _y, PACK_COLOR_1555_REV(a, r, g, b))			\
+
+#define WRITE_PIXEL( _x, _y, p ) PUT_VALUE(_x, _y, p)
+
+#define READ_RGBA( rgba, _x, _y )					\
+   do {									\
+      GLushort p = GET_VALUE(_x, _y);					\
+      p = p << 8 | p >> 8;						\
+      rgba[0] = ((p >> 7) & 0xf8) * 255 / 0xf8;				\
+      rgba[1] = ((p >> 2) & 0xf8) * 255 / 0xf8;				\
+      rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8;				\
+      rgba[3] = ((p >> 15) & 0x1) * 0xff;				\
+   } while (0)
+
 #elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
 
 /**
@@ -202,6 +301,118 @@
      } while (0)
 # endif
 
+#elif (SPANTMP_PIXEL_FMT == GL_BGRA) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8)
+
+/**
+ ** GL_BGRA, GL_UNSIGNED_INT_8_8_8_8
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (     buf + (_x) * 4 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLuint *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLuint *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+# define INIT_MONO_PIXEL(p, color)                       \
+     p = PACK_COLOR_8888(color[2], color[1], color[0], color[3]) 
+
+# define WRITE_RGBA(_x, _y, r, g, b, a)                                 \
+   PUT_VALUE(_x, _y, ((r << 8) |					\
+		      (g << 16) |					\
+		      (b << 24) |					\
+		      (a << 0)))
+
+#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p)
+
+# if defined( USE_X86_ASM )
+#  define READ_RGBA(rgba, _x, _y)                                       \
+    do {                                                                \
+       GLuint p = GET_VALUE(_x, _y);					\
+       __asm__ __volatile__( "rorl $8, %0"				\
+				: "=r" (p) : "0" (p) );                 \
+       ((GLuint *)rgba)[0] = p;                                         \
+    } while (0)
+# elif defined( MESA_BIG_ENDIAN )
+    /* On PowerPC with GCC 3.4.2 the shift madness below becomes a single
+     * rotlwi instruction.  It also produces good code on SPARC.
+     */
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = CPU_TO_LE32(GET_VALUE(_x, _y));                      \
+        GLuint t = p;                                                   \
+        *((uint32_t *) rgba) = (t >> 24) | (p << 8);                    \
+     } while (0)
+# else
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = GET_VALUE(_x, _y);					\
+	rgba[0] = (p >>  8) & 0xff;					\
+	rgba[1] = (p >> 16) & 0xff;					\
+	rgba[2] = (p >> 24) & 0xff;					\
+	rgba[3] = (p >>  0) & 0xff;					\
+     } while (0)
+# endif
+
+#elif (SPANTMP_PIXEL_FMT == GL_BGR) && (SPANTMP_PIXEL_TYPE == GL_UNSIGNED_INT_8_8_8_8_REV)
+
+/**
+ ** GL_BGR, GL_UNSIGNED_INT_8_8_8_8_REV
+ **
+ ** This is really for MESA_FORMAT_XRGB8888.  The spantmp code needs to be
+ ** kicked to the curb, and we need to just code-gen this.
+ **/
+
+#ifndef GET_VALUE
+#ifndef GET_PTR
+#define GET_PTR(_x, _y) (     buf + (_x) * 4 + (_y) * pitch)
+#endif
+
+#define GET_VALUE(_x, _y) *(volatile GLuint *)(GET_PTR(_x, _y))
+#define PUT_VALUE(_x, _y, _v) *(volatile GLuint *)(GET_PTR(_x, _y)) = (_v)
+#endif /* GET_VALUE */
+
+# define INIT_MONO_PIXEL(p, color)                       \
+     p = PACK_COLOR_8888(0xff, color[0], color[1], color[2])
+
+# define WRITE_RGBA(_x, _y, r, g, b, a)					\
+   PUT_VALUE(_x, _y, ((r << 16) |					\
+		      (g << 8) |					\
+		      (b << 0) |					\
+		      (0xff << 24)))
+
+#define WRITE_PIXEL(_x, _y, p) PUT_VALUE(_x, _y, p)
+
+# if defined( USE_X86_ASM )
+#  define READ_RGBA(rgba, _x, _y)                                       \
+    do {                                                                \
+       GLuint p = GET_VALUE(_x, _y);					\
+       __asm__ __volatile__( "bswap	%0; rorl $8, %0"                \
+				: "=r" (p) : "0" (p) );                 \
+       ((GLuint *)rgba)[0] = p | 0xff000000;				\
+    } while (0)
+# elif defined( MESA_BIG_ENDIAN )
+    /* On PowerPC with GCC 3.4.2 the shift madness below becomes a single
+     * rotlwi instruction.  It also produces good code on SPARC.
+     */
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = GET_VALUE(_x, _y);					\
+        *((uint32_t *) rgba) = (t << 8) | 0xff;				\
+     } while (0)
+# else
+#  define READ_RGBA( rgba, _x, _y )				        \
+     do {								\
+        GLuint p = GET_VALUE(_x, _y);					\
+	rgba[0] = (p >> 16) & 0xff;					\
+	rgba[1] = (p >>  8) & 0xff;					\
+	rgba[2] = (p >>  0) & 0xff;					\
+	rgba[3] = 0xff;							\
+     } while (0)
+# endif
+
 #else
 #error SPANTMP_PIXEL_FMT must be set to a valid value!
 #endif
diff --git a/src/mesa/drivers/dri/common/texmem.c b/src/mesa/drivers/dri/common/texmem.c
index b64618a03c..895139b55b 100644
--- a/src/mesa/drivers/dri/common/texmem.c
+++ b/src/mesa/drivers/dri/common/texmem.c
@@ -42,14 +42,10 @@
  * application.
  */
 
-#include "texmem.h"
-#include "main/simple_list.h"
 #include "main/imports.h"
 #include "main/macros.h"
-#include "main/texformat.h"
-
-#include <assert.h>
-
+#include "main/simple_list.h"
+#include "texmem.h"
 
 
 static unsigned dummy_swap_counter;
@@ -1302,46 +1298,44 @@ driCalculateTextureFirstLastLevel( driTextureObject * t )
 
 
 /**
- * \name DRI texture formats.  Pointers initialized to either the big- or
- * little-endian Mesa formats.
+ * \name DRI texture formats.  These vars are initialized to either the
+ * big- or little-endian Mesa formats.
  */
 /*@{*/
-const struct gl_texture_format *_dri_texformat_rgba8888 = NULL;
-const struct gl_texture_format *_dri_texformat_argb8888 = NULL;
-const struct gl_texture_format *_dri_texformat_rgb565 = NULL;
-const struct gl_texture_format *_dri_texformat_argb4444 = NULL;
-const struct gl_texture_format *_dri_texformat_argb1555 = NULL;
-const struct gl_texture_format *_dri_texformat_al88 = NULL;
-const struct gl_texture_format *_dri_texformat_a8 = &_mesa_texformat_a8;
-const struct gl_texture_format *_dri_texformat_ci8 = &_mesa_texformat_ci8;
-const struct gl_texture_format *_dri_texformat_i8 = &_mesa_texformat_i8;
-const struct gl_texture_format *_dri_texformat_l8 = &_mesa_texformat_l8;
+gl_format _dri_texformat_rgba8888 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_argb8888 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_rgb565 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_argb4444 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_argb1555 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_al88 = MESA_FORMAT_NONE;
+gl_format _dri_texformat_a8 = MESA_FORMAT_A8;
+gl_format _dri_texformat_ci8 = MESA_FORMAT_CI8;
+gl_format _dri_texformat_i8 = MESA_FORMAT_I8;
+gl_format _dri_texformat_l8 = MESA_FORMAT_L8;
 /*@}*/
 
 
 /**
- * Initialize little endian target, host byte order independent texture formats
+ * Initialize _dri_texformat_* vars according to whether we're on
+ * a big or little endian system.
  */
 void
 driInitTextureFormats(void)
 {
-   const GLuint ui = 1;
-   const GLubyte littleEndian = *((const GLubyte *) &ui);
-
-   if (littleEndian) {
-      _dri_texformat_rgba8888	= &_mesa_texformat_rgba8888;
-      _dri_texformat_argb8888	= &_mesa_texformat_argb8888;
-      _dri_texformat_rgb565	= &_mesa_texformat_rgb565;
-      _dri_texformat_argb4444	= &_mesa_texformat_argb4444;
-      _dri_texformat_argb1555	= &_mesa_texformat_argb1555;
-      _dri_texformat_al88	= &_mesa_texformat_al88;
+   if (_mesa_little_endian()) {
+      _dri_texformat_rgba8888	= MESA_FORMAT_RGBA8888;
+      _dri_texformat_argb8888	= MESA_FORMAT_ARGB8888;
+      _dri_texformat_rgb565	= MESA_FORMAT_RGB565;
+      _dri_texformat_argb4444	= MESA_FORMAT_ARGB4444;
+      _dri_texformat_argb1555	= MESA_FORMAT_ARGB1555;
+      _dri_texformat_al88	= MESA_FORMAT_AL88;
    }
    else {
-      _dri_texformat_rgba8888	= &_mesa_texformat_rgba8888_rev;
-      _dri_texformat_argb8888	= &_mesa_texformat_argb8888_rev;
-      _dri_texformat_rgb565	= &_mesa_texformat_rgb565_rev;
-      _dri_texformat_argb4444	= &_mesa_texformat_argb4444_rev;
-      _dri_texformat_argb1555	= &_mesa_texformat_argb1555_rev;
-      _dri_texformat_al88	= &_mesa_texformat_al88_rev;
+      _dri_texformat_rgba8888	= MESA_FORMAT_RGBA8888_REV;
+      _dri_texformat_argb8888	= MESA_FORMAT_ARGB8888_REV;
+      _dri_texformat_rgb565	= MESA_FORMAT_RGB565_REV;
+      _dri_texformat_argb4444	= MESA_FORMAT_ARGB4444_REV;
+      _dri_texformat_argb1555	= MESA_FORMAT_ARGB1555_REV;
+      _dri_texformat_al88	= MESA_FORMAT_AL88_REV;
    }
 }
diff --git a/src/mesa/drivers/dri/common/texmem.h b/src/mesa/drivers/dri/common/texmem.h
index 9c065da8b4..725ba2e119 100644
--- a/src/mesa/drivers/dri/common/texmem.h
+++ b/src/mesa/drivers/dri/common/texmem.h
@@ -39,6 +39,7 @@
 #define DRI_TEXMEM_H
 
 #include "main/mtypes.h"
+#include "main/formats.h"
 #include "main/mm.h"
 #include "xf86drm.h"
 
@@ -317,16 +318,16 @@ GLboolean driValidateTextureHeaps( driTexHeap * const * texture_heaps,
 extern void driCalculateTextureFirstLastLevel( driTextureObject * t );
 
 
-extern const struct gl_texture_format *_dri_texformat_rgba8888;
-extern const struct gl_texture_format *_dri_texformat_argb8888;
-extern const struct gl_texture_format *_dri_texformat_rgb565;
-extern const struct gl_texture_format *_dri_texformat_argb4444;
-extern const struct gl_texture_format *_dri_texformat_argb1555;
-extern const struct gl_texture_format *_dri_texformat_al88;
-extern const struct gl_texture_format *_dri_texformat_a8;
-extern const struct gl_texture_format *_dri_texformat_ci8;
-extern const struct gl_texture_format *_dri_texformat_i8;
-extern const struct gl_texture_format *_dri_texformat_l8;
+extern gl_format _dri_texformat_rgba8888;
+extern gl_format _dri_texformat_argb8888;
+extern gl_format _dri_texformat_rgb565;
+extern gl_format _dri_texformat_argb4444;
+extern gl_format _dri_texformat_argb1555;
+extern gl_format _dri_texformat_al88;
+extern gl_format _dri_texformat_a8;
+extern gl_format _dri_texformat_ci8;
+extern gl_format _dri_texformat_i8;
+extern gl_format _dri_texformat_l8;
 
 extern void driInitTextureFormats( void );
 
diff --git a/src/mesa/drivers/dri/common/utils.c b/src/mesa/drivers/dri/common/utils.c
index 66f277c10b..b272eb74ea 100644
--- a/src/mesa/drivers/dri/common/utils.c
+++ b/src/mesa/drivers/dri/common/utils.c
@@ -38,9 +38,6 @@
 #include "utils.h"
 
 
-int driDispatchRemapTable[ driDispatchRemapTable_size ];
-
-
 unsigned
 driParseDebugString( const char * debug, 
 		     const struct dri_debug_control * control  )
@@ -142,7 +139,7 @@ driGetRendererString( char * buffer, const char * hardware_name,
 #define need_GL_EXT_blend_func_separate
 #define need_GL_NV_vertex_program
 
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 static const struct dri_extension all_mesa_extensions[] = {
    { "GL_ARB_draw_buffers",          GL_ARB_draw_buffers_functions },
@@ -165,8 +162,12 @@ static const struct dri_extension all_mesa_extensions[] = {
 
 
 /**
- * Enable extensions supported by the driver.
+ * Enable and map extensions supported by the driver.
  * 
+ * When ctx is NULL, extensions are not enabled, but their functions
+ * are still mapped.  When extensions_to_enable is NULL, all static
+ * functions known to mesa core are mapped.
+ *
  * \bug
  * ARB_imaging isn't handled properly.  In Mesa, enabling ARB_imaging also
  * enables all the sub-extensions that are folded into it.  This means that
@@ -181,18 +182,23 @@ void driInitExtensions( GLcontext * ctx,
    unsigned   i;
 
    if ( first_time ) {
-      for ( i = 0 ; i < driDispatchRemapTable_size ; i++ ) {
-	 driDispatchRemapTable[i] = -1;
-      }
-
       first_time = 0;
-      driInitExtensions( ctx, all_mesa_extensions, GL_FALSE );
+      driInitExtensions( NULL, all_mesa_extensions, GL_FALSE );
    }
 
    if ( (ctx != NULL) && enable_imaging ) {
       _mesa_enable_imaging_extensions( ctx );
    }
 
+   /* The caller is too lazy to list any extension */
+   if ( extensions_to_enable == NULL ) {
+      /* Map the static functions.  Together with those mapped by remap
+       * table, this should cover everything mesa core knows.
+       */
+      _mesa_map_static_functions();
+      return;
+   }
+
    for ( i = 0 ; extensions_to_enable[i].name != NULL ; i++ ) {
        driInitSingleExtension( ctx, & extensions_to_enable[i] );
    }
@@ -202,80 +208,18 @@ void driInitExtensions( GLcontext * ctx,
 
 
 /**
- * Enable and add dispatch functions for a single extension
+ * Enable and map functions for a single extension
  * 
  * \param ctx  Context where extension is to be enabled.
  * \param ext  Extension that is to be enabled.
  * 
- * \sa driInitExtensions, _mesa_enable_extension, _glapi_add_entrypoint
- *
- * \todo
- * Determine if it would be better to use \c strlen instead of the hardcoded
- * for-loops.
+ * \sa driInitExtensions, _mesa_enable_extension, _mesa_map_function_array
  */
 void driInitSingleExtension( GLcontext * ctx,
 			     const struct dri_extension * ext )
 {
-    unsigned i;
-
-
     if ( ext->functions != NULL ) {
-	for ( i = 0 ; ext->functions[i].strings != NULL ; i++ ) {
-	    const char * functions[16];
-	    const char * parameter_signature;
-	    const char * str = ext->functions[i].strings;
-	    unsigned j;
-	    unsigned offset;
-
-
-	    /* Separate the parameter signature from the rest of the string.
-	     * If the parameter signature is empty (i.e., the string starts
-	     * with a NUL character), then the function has a void parameter
-	     * list.
-	     */
-	    parameter_signature = str;
-	    while ( str[0] != '\0' ) {
-		str++;
-	    }
-	    str++;
-
-
-	    /* Divide the string into the substrings that name each
-	     * entry-point for the function.
-	     */
-	    for ( j = 0 ; j < 16 ; j++ ) {
-		if ( str[0] == '\0' ) {
-		    functions[j] = NULL;
-		    break;
-		}
-
-		functions[j] = str;
-
-		while ( str[0] != '\0' ) {
-		    str++;
-		}
-		str++;
-	    }
-
-
-	    /* Add each entry-point to the dispatch table.
-	     */
-	    offset = _glapi_add_dispatch( functions, parameter_signature );
-	    if (offset == -1) {
-#if 0 /* this causes noise with egl */
-		fprintf(stderr, "DISPATCH ERROR! _glapi_add_dispatch failed "
-			"to add %s!\n", functions[0]);
-#endif
-	    }
-	    else if (ext->functions[i].remap_index != -1) {
-		driDispatchRemapTable[ ext->functions[i].remap_index ] = 
-		  offset;
-	    }
-	    else if (ext->functions[i].offset != offset) {
-		fprintf(stderr, "DISPATCH ERROR! %s -> %u != %u\n",
-			functions[0], offset, ext->functions[i].offset);
-	    }
-	}
+       _mesa_map_function_array(ext->functions);
     }
 
     if ( ctx != NULL ) {
diff --git a/src/mesa/drivers/dri/common/utils.h b/src/mesa/drivers/dri/common/utils.h
index 9e9e5bc224..2aa6de66c1 100644
--- a/src/mesa/drivers/dri/common/utils.h
+++ b/src/mesa/drivers/dri/common/utils.h
@@ -31,6 +31,7 @@
 #include <GL/gl.h>
 #include <GL/internal/dri_interface.h>
 #include "main/context.h"
+#include "main/remap.h"
 
 typedef struct __DRIutilversionRec2    __DRIutilversion2;
 
@@ -40,35 +41,6 @@ struct dri_debug_control {
 };
 
 /**
- * Description of the entry-points and parameters for an OpenGL function.
- */
-struct dri_extension_function {
-    /**
-     * \brief
-     * Packed string describing the parameter signature and the entry-point
-     * names.
-     * 
-     * The parameter signature and the names of the entry-points for this
-     * function are packed into a single string.  The substrings are
-     * separated by NUL characters.  The whole string is terminated by
-     * two consecutive NUL characters.
-     */
-    const char * strings;
-
-
-    /**
-     * Location in the remap table where the dispatch offset should be
-     * stored.
-     */
-    int remap_index;
-
-    /**
-     * Offset of the function in the dispatch table.
-     */
-    int offset;
-};
-
-/**
  * Description of the API for an extension to OpenGL.
  */
 struct dri_extension {
@@ -83,7 +55,7 @@ struct dri_extension {
      * is terminated by a structure with a \c NULL
      * \c dri_extension_function::strings pointer.
      */
-    const struct dri_extension_function * functions;
+    const struct gl_function_remap * functions;
 };
 
 /**
diff --git a/src/mesa/drivers/dri/common/vblank.c b/src/mesa/drivers/dri/common/vblank.c
index 12aeaa108f..49b22a2dc7 100644
--- a/src/mesa/drivers/dri/common/vblank.c
+++ b/src/mesa/drivers/dri/common/vblank.c
@@ -34,12 +34,12 @@
 #include "vblank.h"
 #include "xmlpool.h"
 
-static unsigned int msc_to_vblank(__DRIdrawablePrivate * dPriv, int64_t msc)
+static unsigned int msc_to_vblank(__DRIdrawable * dPriv, int64_t msc)
 {
    return (unsigned int)(msc - dPriv->msc_base + dPriv->vblank_base);
 }
 
-static int64_t vblank_to_msc(__DRIdrawablePrivate * dPriv, unsigned int vblank)
+static int64_t vblank_to_msc(__DRIdrawable * dPriv, unsigned int vblank)
 {
    return (int64_t)(vblank - dPriv->vblank_base + dPriv->msc_base);
 }
@@ -64,8 +64,8 @@ static int64_t vblank_to_msc(__DRIdrawablePrivate * dPriv, unsigned int vblank)
  * \return       Zero is returned on success.  A negative errno value
  *               is returned on failure.
  */
-int driDrawableGetMSC32( __DRIscreenPrivate * priv,
-			 __DRIdrawablePrivate * dPriv,
+int driDrawableGetMSC32( __DRIscreen * priv,
+			 __DRIdrawable * dPriv,
 			 int64_t * count)
 {
    drmVBlank vbl;
@@ -122,7 +122,7 @@ int driDrawableGetMSC32( __DRIscreenPrivate * priv,
  * \return            Zero on success or \c GLX_BAD_CONTEXT on failure.
  */
 
-int driWaitForMSC32( __DRIdrawablePrivate *priv,
+int driWaitForMSC32( __DRIdrawable *priv,
 		     int64_t target_msc, int64_t divisor, int64_t remainder,
 		     int64_t * msc )
 {
@@ -278,7 +278,7 @@ static int do_wait( drmVBlank * vbl, GLuint * vbl_seq, int fd )
  */
 
 static unsigned
-driGetDefaultVBlankInterval( const  __DRIdrawablePrivate *priv )
+driGetDefaultVBlankInterval( const  __DRIdrawable *priv )
 {
    if ( (priv->vblFlags & (VBLANK_FLAG_THROTTLE | VBLANK_FLAG_SYNC)) != 0 ) {
       return 1;
@@ -295,7 +295,7 @@ driGetDefaultVBlankInterval( const  __DRIdrawablePrivate *priv )
  * direct rendering context.
  */
 
-void driDrawableInitVBlank( __DRIdrawablePrivate *priv )
+void driDrawableInitVBlank( __DRIdrawable *priv )
 {
    if ( priv->swap_interval == (unsigned)-1 &&
 	!( priv->vblFlags & VBLANK_FLAG_NO_IRQ ) ) {
@@ -320,7 +320,7 @@ void driDrawableInitVBlank( __DRIdrawablePrivate *priv )
  */
 
 unsigned
-driGetVBlankInterval( const  __DRIdrawablePrivate *priv )
+driGetVBlankInterval( const  __DRIdrawable *priv )
 {
    if ( (priv->vblFlags & VBLANK_FLAG_INTERVAL) != 0 ) {
       /* this must have been initialized when the drawable was first bound
@@ -340,7 +340,7 @@ driGetVBlankInterval( const  __DRIdrawablePrivate *priv )
  */
 
 void
-driGetCurrentVBlank( __DRIdrawablePrivate *priv )
+driGetCurrentVBlank( __DRIdrawable *priv )
 {
    drmVBlank vbl;
 
@@ -366,7 +366,7 @@ driGetCurrentVBlank( __DRIdrawablePrivate *priv )
  */
 
 int
-driWaitForVBlank( __DRIdrawablePrivate *priv, GLboolean * missed_deadline )
+driWaitForVBlank( __DRIdrawable *priv, GLboolean * missed_deadline )
 {
    drmVBlank vbl;
    unsigned   original_seq;
diff --git a/src/mesa/drivers/dri/common/vblank.h b/src/mesa/drivers/dri/common/vblank.h
index 8b2c761a11..29d1ad8003 100644
--- a/src/mesa/drivers/dri/common/vblank.h
+++ b/src/mesa/drivers/dri/common/vblank.h
@@ -44,17 +44,17 @@
 #define VBLANK_FLAG_SECONDARY (1U << 8)  /* Wait for secondary vblank.
 					  */
 
-extern int driGetMSC32( __DRIscreenPrivate * priv, int64_t * count );
-extern int driDrawableGetMSC32( __DRIscreenPrivate * priv,
-				__DRIdrawablePrivate * drawablePrivate,
+extern int driGetMSC32( __DRIscreen * priv, int64_t * count );
+extern int driDrawableGetMSC32( __DRIscreen * priv,
+				__DRIdrawable * drawablePrivate,
 				int64_t * count);
-extern int driWaitForMSC32( __DRIdrawablePrivate *priv,
+extern int driWaitForMSC32( __DRIdrawable *priv,
     int64_t target_msc, int64_t divisor, int64_t remainder, int64_t * msc );
 extern GLuint driGetDefaultVBlankFlags( const driOptionCache *optionCache );
-extern void driDrawableInitVBlank ( __DRIdrawablePrivate *priv );
-extern unsigned driGetVBlankInterval( const  __DRIdrawablePrivate *priv );
-extern void driGetCurrentVBlank( __DRIdrawablePrivate *priv );
-extern int driWaitForVBlank( __DRIdrawablePrivate *priv,
+extern void driDrawableInitVBlank ( __DRIdrawable *priv );
+extern unsigned driGetVBlankInterval( const  __DRIdrawable *priv );
+extern void driGetCurrentVBlank( __DRIdrawable *priv );
+extern int driWaitForVBlank( __DRIdrawable *priv,
 			     GLboolean * missed_deadline );
 
 #undef usleep
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/fb/fb_dri.c b/src/mesa/drivers/dri/fb/fb_dri.c
index 571b8922d5..f37241dd69 100644
--- a/src/mesa/drivers/dri/fb/fb_dri.c
+++ b/src/mesa/drivers/dri/fb/fb_dri.c
@@ -64,9 +64,9 @@ typedef struct {
    GLcontext *glCtx;		/* Mesa context */
 
    struct {
-      __DRIcontextPrivate *context;	
-      __DRIscreenPrivate *screen;	
-      __DRIdrawablePrivate *drawable; /* drawable bound to this ctx */
+      __DRIcontext *context;	
+      __DRIscreen *screen;	
+      __DRIdrawable *drawable; /* drawable bound to this ctx */
    } dri;
    
 } fbContext, *fbContextPtr;
@@ -313,14 +313,14 @@ fbSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
 /* Initialize the driver specific screen private data.
  */
 static GLboolean
-fbInitDriver( __DRIscreenPrivate *sPriv )
+fbInitDriver( __DRIscreen *sPriv )
 {
    sPriv->private = NULL;
    return GL_TRUE;
 }
 
 static void
-fbDestroyScreen( __DRIscreenPrivate *sPriv )
+fbDestroyScreen( __DRIscreen *sPriv )
 {
 }
 
@@ -329,7 +329,7 @@ fbDestroyScreen( __DRIscreenPrivate *sPriv )
  */
 static GLboolean
 fbCreateContext( const __GLcontextModes *glVisual,
-		 __DRIcontextPrivate *driContextPriv,
+		 __DRIcontext *driContextPriv,
 		 void *sharedContextPrivate)
 {
    fbContextPtr fbmesa;
@@ -384,7 +384,7 @@ fbCreateContext( const __GLcontextModes *glVisual,
 
 
 static void
-fbDestroyContext( __DRIcontextPrivate *driContextPriv )
+fbDestroyContext( __DRIcontext *driContextPriv )
 {
    GET_CURRENT_CONTEXT(ctx);
    fbContextPtr fbmesa = (fbContextPtr) driContextPriv->driverPrivate;
@@ -415,8 +415,8 @@ fbDestroyContext( __DRIcontextPrivate *driContextPriv )
  * data.
  */
 static GLboolean
-fbCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-		__DRIdrawablePrivate *driDrawPriv,
+fbCreateBuffer( __DRIscreen *driScrnPriv,
+		__DRIdrawable *driDrawPriv,
 		const __GLcontextModes *mesaVis,
 		GLboolean isPixmap )
 {
@@ -437,7 +437,7 @@ fbCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       /* XXX double-check these parameters (bpp vs cpp, etc) */
       {
-         driRenderbuffer *drb = driNewRenderbuffer(GL_RGBA,
+         driRenderbuffer *drb = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                                    driScrnPriv->pFB,
                                                    driScrnPriv->fbBPP / 8,
                                                    driScrnPriv->fbOrigin,
@@ -451,7 +451,7 @@ fbCreateBuffer( __DRIscreenPrivate *driScrnPriv,
          /* XXX what are the correct origin/stride values? */
          GLvoid *backBuf = _mesa_malloc(driScrnPriv->fbStride
                                         * driScrnPriv->fbHeight);
-         driRenderbuffer *drb = driNewRenderbuffer(GL_RGBA,
+         driRenderbuffer *drb = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                                    backBuf,
                                                    driScrnPriv->fbBPP /8,
                                                    driScrnPriv->fbOrigin,
@@ -478,7 +478,7 @@ fbCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-fbDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+fbDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
@@ -488,7 +488,7 @@ fbDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
 /* If the backbuffer is on a videocard, this is extraordinarily slow!
  */
 static void
-fbSwapBuffers( __DRIdrawablePrivate *dPriv )
+fbSwapBuffers( __DRIdrawable *dPriv )
 {
    struct gl_framebuffer *mesa_framebuffer = (struct gl_framebuffer *)dPriv->driverPrivate;
    struct gl_renderbuffer * front_renderbuffer = mesa_framebuffer->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
@@ -532,9 +532,9 @@ fbSwapBuffers( __DRIdrawablePrivate *dPriv )
  * buffer `b'.
  */
 static GLboolean
-fbMakeCurrent( __DRIcontextPrivate *driContextPriv,
-	       __DRIdrawablePrivate *driDrawPriv,
-	       __DRIdrawablePrivate *driReadPriv )
+fbMakeCurrent( __DRIcontext *driContextPriv,
+	       __DRIdrawable *driDrawPriv,
+	       __DRIdrawable *driReadPriv )
 {
    if ( driContextPriv ) {
       fbContextPtr newFbCtx = 
@@ -556,7 +556,7 @@ fbMakeCurrent( __DRIcontextPrivate *driContextPriv,
 /* Force the context `c' to be unbound from its buffer.
  */
 static GLboolean
-fbUnbindContext( __DRIcontextPrivate *driContextPriv )
+fbUnbindContext( __DRIcontext *driContextPriv )
 {
    return GL_TRUE;
 }
@@ -657,7 +657,7 @@ struct DRIDriverRec __driDriver = {
 };
 
 static __GLcontextModes *
-fbFillInModes( __DRIscreenPrivate *psp,
+fbFillInModes( __DRIscreen *psp,
 	       unsigned pixel_bits, unsigned depth_bits,
 	       unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -745,7 +745,7 @@ fbFillInModes( __DRIscreenPrivate *psp,
  * with the \c __GLcontextModes that the driver can support for windows or
  * pbuffers.
  * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
+ * \return A pointer to a \c __DRIscreen on success, or \c NULL on 
  *         failure.
  */
 PUBLIC
@@ -759,7 +759,7 @@ void * __driCreateNewScreen( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc
                                    int internal_api_version,
                                    __GLcontextModes ** driver_modes )
 {
-   __DRIscreenPrivate *psp;
+   __DRIscreen *psp;
    static const __DRIversion ddx_expected = { 4, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
    static const __DRIversion drm_expected = { 1, 5, 0 };
@@ -785,3 +785,10 @@ void * __driCreateNewScreen( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc
 
           return (void *) psp;
 }
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/fb/fb_egl.c b/src/mesa/drivers/dri/fb/fb_egl.c
index 4e41860d8c..02e44bb8ee 100644
--- a/src/mesa/drivers/dri/fb/fb_egl.c
+++ b/src/mesa/drivers/dri/fb/fb_egl.c
@@ -84,9 +84,9 @@ typedef struct fb_context
    _EGLContext Base;  /* base class/object */
    GLcontext *glCtx;
    struct {
-      __DRIcontextPrivate *context;	
-      __DRIscreenPrivate *screen;	
-      __DRIdrawablePrivate *drawable; /* drawable bound to this ctx */
+      __DRIcontext *context;	
+      __DRIscreen *screen;	
+      __DRIdrawable *drawable; /* drawable bound to this ctx */
    } dri;
 } fbContext, *fbContextPtr;
 
@@ -692,7 +692,7 @@ fbCreateScreenSurfaceMESA(_EGLDriver *drv, EGLDisplay dpy, EGLConfig cfg,
 
    /* front color renderbuffer */
    {
-      driRenderbuffer *drb = driNewRenderbuffer(GL_RGBA, display->pFB,
+      driRenderbuffer *drb = driNewRenderbuffer(MESA_FORMAT_ARGB8888, display->pFB,
                                                 bytesPerPixel,
                                                 origin, stride, NULL);
       fbSetSpanFunctions(drb, &vis);
@@ -703,7 +703,7 @@ fbCreateScreenSurfaceMESA(_EGLDriver *drv, EGLDisplay dpy, EGLConfig cfg,
    /* back color renderbuffer */
    if (vis.doubleBufferMode) {
       GLubyte *backBuf = _mesa_malloc(stride * height);
-      driRenderbuffer *drb = driNewRenderbuffer(GL_RGBA, backBuf,
+      driRenderbuffer *drb = driNewRenderbuffer(MESA_FORMAT_ARGB8888, backBuf,
                                                 bytesPerPixel,
                                                 origin, stride, NULL);
       fbSetSpanFunctions(drb, &vis);
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_bitmap.c b/src/mesa/drivers/dri/ffb/ffb_bitmap.c
index f89c0412df..611afddfaf 100644
--- a/src/mesa/drivers/dri/ffb/ffb_bitmap.c
+++ b/src/mesa/drivers/dri/ffb/ffb_bitmap.c
@@ -46,7 +46,7 @@ ffb_bitmap(GLcontext *ctx, GLint px, GLint py,
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
 	ffb_fbcPtr ffb = fmesa->regs;
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	unsigned int ppc, pixel;
 	GLint row, col, row_stride;
 	const GLubyte *src;
diff --git a/src/mesa/drivers/dri/ffb/ffb_clear.c b/src/mesa/drivers/dri/ffb/ffb_clear.c
index 776fb487f8..dfe60f36f2 100644
--- a/src/mesa/drivers/dri/ffb/ffb_clear.c
+++ b/src/mesa/drivers/dri/ffb/ffb_clear.c
@@ -123,7 +123,7 @@ CreatorComputePageFillFixups(struct ff_fixups *fixups,
 }
 
 static void
-ffb_do_clear(GLcontext *ctx, __DRIdrawablePrivate *dPriv)
+ffb_do_clear(GLcontext *ctx, __DRIdrawable *dPriv)
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
 	FFBDRIPtr gDRIPriv = (FFBDRIPtr) fmesa->driScreen->pDevPriv;
@@ -252,7 +252,7 @@ ffb_do_clear(GLcontext *ctx, __DRIdrawablePrivate *dPriv)
 void ffbDDClear(GLcontext *ctx, GLbitfield mask)
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	unsigned int stcmask = BUFFER_BIT_STENCIL;
 
 #ifdef CLEAR_TRACE
diff --git a/src/mesa/drivers/dri/ffb/ffb_context.h b/src/mesa/drivers/dri/ffb/ffb_context.h
index 77f87d41c3..4d1d53ff59 100644
--- a/src/mesa/drivers/dri/ffb/ffb_context.h
+++ b/src/mesa/drivers/dri/ffb/ffb_context.h
@@ -273,8 +273,8 @@ do {	if ((STATE_MASK) & ~((FMESA)->state_dirty)) {	\
 	unsigned int		setupnewinputs;
 	unsigned int		new_gl_state;
 
-	__DRIdrawablePrivate	*driDrawable;
-	__DRIscreenPrivate	*driScreen;
+	__DRIdrawable	*driDrawable;
+	__DRIscreen	*driScreen;
 	ffbScreenPrivate	*ffbScreen;
 	ffb_dri_state_t		*ffb_sarea;
 } ffbContextRec, *ffbContextPtr;
diff --git a/src/mesa/drivers/dri/ffb/ffb_depth.c b/src/mesa/drivers/dri/ffb/ffb_depth.c
index 71f204d21e..5d509ff696 100644
--- a/src/mesa/drivers/dri/ffb/ffb_depth.c
+++ b/src/mesa/drivers/dri/ffb/ffb_depth.c
@@ -49,7 +49,7 @@ static void FFBWriteDepthSpan( GLcontext *ctx,
 #endif
 	if (ctx->Depth.Mask) {
 		ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-		__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+		__DRIdrawable *dPriv = fmesa->driDrawable;
 		GLuint *zptr;
 		GLuint i;
 
@@ -110,7 +110,7 @@ static void FFBWriteDepthPixels( GLcontext *ctx,
 #endif
 	if (ctx->Depth.Mask) {
 		ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-		__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+		__DRIdrawable *dPriv = fmesa->driDrawable;
 		char *zbase;
 		GLuint i;
 
@@ -153,7 +153,7 @@ static void FFBReadDepthSpan( GLcontext *ctx,
 {
         GLuint *depth = (GLuint *) values;
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	GLuint *zptr;
 	GLuint i;
 
@@ -194,7 +194,7 @@ static void FFBReadDepthPixels( GLcontext *ctx,
 {
         GLuint *depth = (GLuint *) values;
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	char *zbase;
 	GLuint i;
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_span.c b/src/mesa/drivers/dri/ffb/ffb_span.c
index 0d3d604095..8ec33a11bc 100644
--- a/src/mesa/drivers/dri/ffb/ffb_span.c
+++ b/src/mesa/drivers/dri/ffb/ffb_span.c
@@ -45,7 +45,7 @@
 		UNLOCK_HARDWARE(fmesa); \
 
 #define LOCAL_VARS						\
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;	\
+	__DRIdrawable *dPriv = fmesa->driDrawable;	\
 	GLuint height = dPriv->h;				\
         GLuint p;						\
 	char *buf; 						\
diff --git a/src/mesa/drivers/dri/ffb/ffb_state.c b/src/mesa/drivers/dri/ffb/ffb_state.c
index 5eb8f417ff..6f8a46d1fc 100644
--- a/src/mesa/drivers/dri/ffb/ffb_state.c
+++ b/src/mesa/drivers/dri/ffb/ffb_state.c
@@ -384,7 +384,7 @@ ffbDDStencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
 static void ffbCalcViewportRegs(GLcontext *ctx)
 {
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	GLuint xmin, xmax, ymin, ymax, zmin, zmax;
 	unsigned int vcmin, vcmax;
 
@@ -430,7 +430,7 @@ void ffbCalcViewport(GLcontext *ctx)
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
 	const GLfloat *v = ctx->Viewport._WindowMap.m;
 	GLfloat *m = fmesa->hw_viewport;
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 
 	m[MAT_SX] =   v[MAT_SX];
 	m[MAT_TX] =   v[MAT_TX] + dPriv->x + SUBPIXEL_X;
@@ -762,7 +762,7 @@ static void ffbDDLineStipple(GLcontext *ctx, GLint factor, GLushort pattern)
 
 void ffbXformAreaPattern(ffbContextPtr fmesa, const GLubyte *mask)
 {
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	int i, lines, xoff;
 
 	lines = 0;
diff --git a/src/mesa/drivers/dri/ffb/ffb_stencil.c b/src/mesa/drivers/dri/ffb/ffb_stencil.c
index 921a83d274..ce8ef43c91 100644
--- a/src/mesa/drivers/dri/ffb/ffb_stencil.c
+++ b/src/mesa/drivers/dri/ffb/ffb_stencil.c
@@ -48,7 +48,7 @@ static void FFBWriteStencilSpan( GLcontext *ctx,
 #endif
 	if (ctx->Depth.Mask) {
 		ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-		__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+		__DRIdrawable *dPriv = fmesa->driDrawable;
 		GLuint *zptr;
 		GLuint i;
 
@@ -93,7 +93,7 @@ static void FFBWriteStencilPixels( GLcontext *ctx,
 #endif
 	if (ctx->Depth.Mask) {
 		ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-		__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+		__DRIdrawable *dPriv = fmesa->driDrawable;
 		char *zbase;
 		GLuint i;
 
@@ -136,7 +136,7 @@ static void FFBReadStencilSpan( GLcontext *ctx,
 {
         GLubyte *stencil = (GLubyte *) values;
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	GLuint *zptr;
 	GLuint i;
 
@@ -176,7 +176,7 @@ static void FFBReadStencilPixels( GLcontext *ctx,
 {
         GLubyte *stencil = (GLubyte *) values;
 	ffbContextPtr fmesa = FFB_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
 	char *zbase;
 	GLuint i;
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_tris.c b/src/mesa/drivers/dri/ffb/ffb_tris.c
index d785c15718..8bf5ae498f 100644
--- a/src/mesa/drivers/dri/ffb/ffb_tris.c
+++ b/src/mesa/drivers/dri/ffb/ffb_tris.c
@@ -351,8 +351,8 @@ static struct {
 
 #define LOCAL_VARS(n)				\
    ffbContextPtr fmesa = FFB_CONTEXT(ctx);	\
-   __DRIdrawablePrivate *dPriv = fmesa->driDrawable; \
-   ffb_color color[n];				\
+   __DRIdrawable *dPriv = fmesa->driDrawable; \
+   ffb_color color[n] = { { 0 } };		\
    (void) color; (void) dPriv;
 
 /***********************************************************************
diff --git a/src/mesa/drivers/dri/ffb/ffb_vbtmp.h b/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
index 0495d0e276..c548ef3ad5 100644
--- a/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
+++ b/src/mesa/drivers/dri/ffb/ffb_vbtmp.h
@@ -38,11 +38,11 @@ static void TAG(emit)(GLcontext *ctx, GLuint start, GLuint end)
 #endif
 
 #if (IND & (FFB_VB_RGBA_BIT))
-	col0 = VB->ColorPtr[0]->data;
-	col0_stride = VB->ColorPtr[0]->stride;
+	col0 = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+	col0_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 #if (IND & (FFB_VB_TWOSIDE_BIT))
-	col1 = VB->ColorPtr[1]->data;
-	col1_stride = VB->ColorPtr[1]->stride;
+	col1 = VB->BackfaceColorPtr->data;
+	col1_stride = VB->BackfaceColorPtr->stride;
 #endif
 #endif
 
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.c b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
index 3b9f5c6759..88285f454e 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.c
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
@@ -62,7 +62,7 @@
 #include "drirenderbuffer.h"
 
 static GLboolean
-ffbInitDriver(__DRIscreenPrivate *sPriv)
+ffbInitDriver(__DRIscreen *sPriv)
 {
 	ffbScreenPrivate *ffbScreen;
 	FFBDRIPtr gDRIPriv = (FFBDRIPtr) sPriv->pDevPriv;
@@ -154,7 +154,7 @@ ffbInitDriver(__DRIscreenPrivate *sPriv)
 
 
 static void
-ffbDestroyScreen(__DRIscreenPrivate *sPriv)
+ffbDestroyScreen(__DRIscreen *sPriv)
 {
 	ffbScreenPrivate *ffbScreen = sPriv->private;
 	FFBDRIPtr gDRIPriv = (FFBDRIPtr) sPriv->pDevPriv;
@@ -183,12 +183,12 @@ static const struct tnl_pipeline_stage *ffb_pipeline[] = {
 /* Create and initialize the Mesa and driver specific context data */
 static GLboolean
 ffbCreateContext(const __GLcontextModes *mesaVis,
-                 __DRIcontextPrivate *driContextPriv,
+                 __DRIcontext *driContextPriv,
                  void *sharedContextPrivate)
 {
 	ffbContextPtr fmesa;
 	GLcontext *ctx, *shareCtx;
-	__DRIscreenPrivate *sPriv;
+	__DRIscreen *sPriv;
 	ffbScreenPrivate *ffbScreen;
 	char *debug;
 	struct dd_function_table functions;
@@ -306,7 +306,7 @@ ffbCreateContext(const __GLcontextModes *mesaVis,
 }
 
 static void
-ffbDestroyContext(__DRIcontextPrivate *driContextPriv)
+ffbDestroyContext(__DRIcontext *driContextPriv)
 {
 	ffbContextPtr fmesa = (ffbContextPtr) driContextPriv->driverPrivate;
 
@@ -328,8 +328,8 @@ ffbDestroyContext(__DRIcontextPrivate *driContextPriv)
 
 /* Create and initialize the Mesa and driver specific pixmap buffer data */
 static GLboolean
-ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
-                __DRIdrawablePrivate *driDrawPriv,
+ffbCreateBuffer(__DRIscreen *driScrnPriv,
+                __DRIdrawable *driDrawPriv,
                 const __GLcontextModes *mesaVis,
                 GLboolean isPixmap )
 {
@@ -347,7 +347,7 @@ ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA, NULL, bpp, offset, bogusPitch,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL, bpp, offset, bogusPitch,
                                  driDrawPriv);
          ffbSetSpanFunctions(frontRb, mesaVis);
          _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
@@ -355,7 +355,7 @@ ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA, NULL, bpp, offset, bogusPitch,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL, bpp, offset, bogusPitch,
                                  driDrawPriv);
          ffbSetSpanFunctions(backRb, mesaVis);
          _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
@@ -363,7 +363,7 @@ ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16, NULL, bpp, offset,
+            = driNewRenderbuffer(MESA_FORMAT_Z16, NULL, bpp, offset,
                                  bogusPitch, driDrawPriv);
          ffbSetDepthFunctions(depthRb, mesaVis);
          _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
@@ -371,7 +371,7 @@ ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0 && !swStencil) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT, NULL, bpp, offset,
+            = driNewRenderbuffer(MESA_FORMAT_S8, NULL, bpp, offset,
                                  bogusPitch, driDrawPriv);
          ffbSetStencilFunctions(stencilRb, mesaVis);
          _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
@@ -392,7 +392,7 @@ ffbCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
 
 static void
-ffbDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+ffbDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
@@ -401,7 +401,7 @@ ffbDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
 #define USE_FAST_SWAP
 
 static void
-ffbSwapBuffers( __DRIdrawablePrivate *dPriv )
+ffbSwapBuffers( __DRIdrawable *dPriv )
 {
 	ffbContextPtr fmesa = (ffbContextPtr) dPriv->driContextPriv->driverPrivate;
 	unsigned int fbc, wid, wid_reg_val, dac_db_bit;
@@ -532,9 +532,9 @@ static void ffb_init_wid(ffbContextPtr fmesa, unsigned int wid)
 /* Force the context `c' to be the current context and associate with it
    buffer `b' */
 static GLboolean
-ffbMakeCurrent(__DRIcontextPrivate *driContextPriv,
-               __DRIdrawablePrivate *driDrawPriv,
-               __DRIdrawablePrivate *driReadPriv)
+ffbMakeCurrent(__DRIcontext *driContextPriv,
+               __DRIdrawable *driDrawPriv,
+               __DRIdrawable *driReadPriv)
 {
 	if (driContextPriv) {
 		ffbContextPtr fmesa = (ffbContextPtr) driContextPriv->driverPrivate;
@@ -581,15 +581,15 @@ ffbMakeCurrent(__DRIcontextPrivate *driContextPriv,
 
 /* Force the context `c' to be unbound from its buffer */
 static GLboolean
-ffbUnbindContext(__DRIcontextPrivate *driContextPriv)
+ffbUnbindContext(__DRIcontext *driContextPriv)
 {
 	return GL_TRUE;
 }
 
 void ffbXMesaUpdateState(ffbContextPtr fmesa)
 {
-	__DRIdrawablePrivate *dPriv = fmesa->driDrawable;
-	__DRIscreenPrivate *sPriv = fmesa->driScreen;
+	__DRIdrawable *dPriv = fmesa->driDrawable;
+	__DRIscreen *sPriv = fmesa->driScreen;
 	int stamp = dPriv->lastStamp;
 
 	DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
@@ -607,7 +607,7 @@ void ffbXMesaUpdateState(ffbContextPtr fmesa)
 }
 
 static const __DRIconfig **
-ffbFillInModes( __DRIscreenPrivate *psp,
+ffbFillInModes( __DRIscreen *psp,
 		unsigned pixel_bits, unsigned depth_bits,
 		unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -722,3 +722,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.h b/src/mesa/drivers/dri/ffb/ffb_xmesa.h
index 255da4c5f8..2b1740d221 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.h
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.h
@@ -11,7 +11,7 @@
 #include "ffb_fifo.h"
 
 typedef struct {
-	__DRIscreenPrivate		*sPriv;
+	__DRIscreen		*sPriv;
 	ffb_fbcPtr			regs;
 	ffb_dacPtr			dac;
 	volatile char			*sfb8r;
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/gamma/gamma_context.c b/src/mesa/drivers/dri/gamma/gamma_context.c
index b0ac299daa..bab5b69a8e 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.c
+++ b/src/mesa/drivers/dri/gamma/gamma_context.c
@@ -68,11 +68,11 @@ static const struct tnl_pipeline_stage *gamma_pipeline[] = {
 };
 
 GLboolean gammaCreateContext( const __GLcontextModes *glVisual,
-			     __DRIcontextPrivate *driContextPriv,
+			     __DRIcontext *driContextPriv,
                      	     void *sharedContextPrivate)
 {
    GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    gammaContextPtr gmesa;
    gammaScreenPtr gammascrn;
    GLINTSAREADRIPtr saPriv=(GLINTSAREADRIPtr)(((char*)sPriv->pSAREA)+
diff --git a/src/mesa/drivers/dri/gamma/gamma_context.h b/src/mesa/drivers/dri/gamma/gamma_context.h
index a32ccb6007..c386aa3007 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.h
+++ b/src/mesa/drivers/dri/gamma/gamma_context.h
@@ -58,10 +58,10 @@ typedef union {
 #define MAX_TEXTURE_STACK       2
 
 extern void	  gammaDDUpdateHWState(GLcontext *ctx);
-extern gammaScreenPtr	  gammaCreateScreen(__DRIscreenPrivate *sPriv);
-extern void	  gammaDestroyScreen(__DRIscreenPrivate *sPriv);
+extern gammaScreenPtr	  gammaCreateScreen(__DRIscreen *sPriv);
+extern void	  gammaDestroyScreen(__DRIscreen *sPriv);
 extern GLboolean gammaCreateContext( const __GLcontextModes *glVisual,
-                                     __DRIcontextPrivate *driContextPriv,
+                                     __DRIcontext *driContextPriv,
                                      void *sharedContextPrivate);
 
 #define GAMMA_UPLOAD_ALL	0xffffffff
@@ -230,9 +230,9 @@ typedef void (*gamma_point_func)( gammaContextPtr,
 struct gamma_context {
 	GLcontext 		*glCtx;		/* Mesa context */
 
-	__DRIcontextPrivate	*driContext;
-	__DRIscreenPrivate	*driScreen;
-	__DRIdrawablePrivate	*driDrawable;
+	__DRIcontext	*driContext;
+	__DRIscreen	*driScreen;
+	__DRIdrawable	*driDrawable;
 
 	GLuint 			new_gl_state;
 	GLuint 			new_state;
diff --git a/src/mesa/drivers/dri/gamma/gamma_lock.c b/src/mesa/drivers/dri/gamma/gamma_lock.c
index 8f2d01688c..cd4acef24d 100644
--- a/src/mesa/drivers/dri/gamma/gamma_lock.c
+++ b/src/mesa/drivers/dri/gamma/gamma_lock.c
@@ -19,8 +19,8 @@ int prevLockLine = 0;
  */
 void gammaGetLock( gammaContextPtr gmesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = gmesa->driDrawable;
-   __DRIscreenPrivate *sPriv = gmesa->driScreen;
+   __DRIdrawable *dPriv = gmesa->driDrawable;
+   __DRIscreen *sPriv = gmesa->driScreen;
 
    drmGetLock( gmesa->driFd, gmesa->hHWContext, flags );
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_macros.h b/src/mesa/drivers/dri/gamma/gamma_macros.h
index c15483b770..d962dcdb56 100644
--- a/src/mesa/drivers/dri/gamma/gamma_macros.h
+++ b/src/mesa/drivers/dri/gamma/gamma_macros.h
@@ -245,8 +245,8 @@ do {                                                                       \
 #ifdef DO_VALIDATE
 #define VALIDATE_DRAWABLE_INFO_NO_LOCK(gcp)                                \
 do {                                                                       \
-    /*__DRIscreenPrivate *psp = gcp->driScreen;*/                          \
-    __DRIdrawablePrivate *pdp = gcp->driDrawable;                          \
+    /*__DRIscreen *psp = gcp->driScreen;*/                          \
+    __DRIdrawable *pdp = gcp->driDrawable;                          \
                                                                            \
     if (*(pdp->pStamp) != pdp->lastStamp) {                                \
 	int old_index = pdp->index;                                        \
@@ -301,7 +301,7 @@ do {                                                                       \
 
 #define VALIDATE_DRAWABLE_INFO(gcp)                                    \
 do {                                                                       \
-    __DRIscreenPrivate *psp = gcp->driScreen;                          \
+    __DRIscreen *psp = gcp->driScreen;                          \
 if (gcp->driDrawable) { \
     DRM_SPINLOCK(&psp->pSAREA->drawable_lock, psp->drawLockID);            \
     VALIDATE_DRAWABLE_INFO_NO_LOCK(gcp);                               \
diff --git a/src/mesa/drivers/dri/gamma/gamma_render.c b/src/mesa/drivers/dri/gamma/gamma_render.c
index 1b9fd169f4..a03a93d132 100644
--- a/src/mesa/drivers/dri/gamma/gamma_render.c
+++ b/src/mesa/drivers/dri/gamma/gamma_render.c
@@ -53,13 +53,13 @@ static void gamma_emit( GLcontext *ctx, GLuint start, GLuint end)
    GLfloat (*tc0)[4] = 0;
    GLuint tc0_size = 0;
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    if (ctx->Texture.Unit[0]._ReallyEnabled) {
-      tc0_stride = VB->TexCoordPtr[0]->stride;
-      tc0 = VB->TexCoordPtr[0]->data;
-      tc0_size = VB->TexCoordPtr[0]->size;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0]->data;
+      tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0]->size;
       coord = VB->ClipPtr->data;
       coord_stride = VB->ClipPtr->stride;
    } else {
diff --git a/src/mesa/drivers/dri/gamma/gamma_screen.c b/src/mesa/drivers/dri/gamma/gamma_screen.c
index f899ebec96..f72a4a5696 100644
--- a/src/mesa/drivers/dri/gamma/gamma_screen.c
+++ b/src/mesa/drivers/dri/gamma/gamma_screen.c
@@ -29,7 +29,7 @@
 
 #include "main/imports.h"
 
-gammaScreenPtr gammaCreateScreen( __DRIscreenPrivate *sPriv )
+gammaScreenPtr gammaCreateScreen( __DRIscreen *sPriv )
 {
    gammaScreenPtr gammaScreen;
    GLINTDRIPtr gDRIPriv = (GLINTDRIPtr)sPriv->pDevPriv;
@@ -129,7 +129,7 @@ gammaScreenPtr gammaCreateScreen( __DRIscreenPrivate *sPriv )
 
 /* Destroy the device specific screen private data struct.
  */
-void gammaDestroyScreen( __DRIscreenPrivate *sPriv )
+void gammaDestroyScreen( __DRIscreen *sPriv )
 {
     gammaScreenPtr gammaScreen = (gammaScreenPtr)sPriv->private;
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_screen.h b/src/mesa/drivers/dri/gamma/gamma_screen.h
index 7f0ed6f80e..c716ea89c2 100644
--- a/src/mesa/drivers/dri/gamma/gamma_screen.h
+++ b/src/mesa/drivers/dri/gamma/gamma_screen.h
@@ -11,7 +11,7 @@ typedef struct {
 
     drmBufMapPtr  bufs;              /* Map of DMA buffers */
 
-    __DRIscreenPrivate *driScreen; /* Back pointer to DRI screen */
+    __DRIscreen *driScreen; /* Back pointer to DRI screen */
 
     int		cpp;
     int		frontPitch;
diff --git a/src/mesa/drivers/dri/gamma/gamma_span.c b/src/mesa/drivers/dri/gamma/gamma_span.c
index cdaaac3f3a..3f0b81800c 100644
--- a/src/mesa/drivers/dri/gamma/gamma_span.c
+++ b/src/mesa/drivers/dri/gamma/gamma_span.c
@@ -10,8 +10,8 @@
 #define LOCAL_VARS							\
    gammaContextPtr gmesa = GAMMA_CONTEXT(ctx);				\
    gammaScreenPtr gammascrn = gmesa->gammaScreen;			\
-   __DRIscreenPrivate *sPriv = gmesa->driScreen;			\
-   __DRIdrawablePrivate *dPriv = gmesa->driDrawable;			\
+   __DRIscreen *sPriv = gmesa->driScreen;			\
+   __DRIdrawable *dPriv = gmesa->driDrawable;			\
    GLuint pitch = sPriv->fbWidth * gammascrn->cpp;		\
    GLuint height = dPriv->h;						\
    char *buf = (char *)(sPriv->pFB +					\
@@ -24,8 +24,8 @@
 /* FIXME! Depth/Stencil read/writes don't work ! */
 #define LOCAL_DEPTH_VARS				\
    gammaScreenPtr gammascrn = gmesa->gammaScreen;	\
-   __DRIdrawablePrivate *dPriv = gmesa->driDrawable;	\
-   __DRIscreenPrivate *sPriv = gmesa->driScreen;	\
+   __DRIdrawable *dPriv = gmesa->driDrawable;	\
+   __DRIscreen *sPriv = gmesa->driScreen;	\
    GLuint pitch = gammascrn->depthPitch;		\
    GLuint height = dPriv->h;				\
    char *buf = (char *)(sPriv->pFB +			\
diff --git a/src/mesa/drivers/dri/gamma/gamma_state.c b/src/mesa/drivers/dri/gamma/gamma_state.c
index 59272f9bc9..47df37466d 100644
--- a/src/mesa/drivers/dri/gamma/gamma_state.c
+++ b/src/mesa/drivers/dri/gamma/gamma_state.c
@@ -813,10 +813,10 @@ static void gammaUpdateMasks( GLcontext *ctx )
 
 
    GLuint mask = gammaPackColor( gmesa->gammaScreen->cpp,
-				ctx->Color.ColorMask[RCOMP],
-				ctx->Color.ColorMask[GCOMP],
-				ctx->Color.ColorMask[BCOMP],
-				ctx->Color.ColorMask[ACOMP] );
+				ctx->Color.ColorMask[0][RCOMP],
+				ctx->Color.ColorMask[0][GCOMP],
+				ctx->Color.ColorMask[0][BCOMP],
+				ctx->Color.ColorMask[0][ACOMP] );
 
    if (gmesa->gammaScreen->cpp == 2) mask |= mask << 16;
 
@@ -1070,7 +1070,7 @@ static void gammaDDReadBuffer( GLcontext *ctx, GLenum mode )
 void gammaUpdateWindow( GLcontext *ctx )
 {
    gammaContextPtr gmesa = GAMMA_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = gmesa->driDrawable;
+   __DRIdrawable *dPriv = gmesa->driDrawable;
    GLfloat xoffset = (GLfloat)dPriv->x;
    GLfloat yoffset = gmesa->driScreen->fbHeight - (GLfloat)dPriv->y - dPriv->h;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1109,7 +1109,7 @@ static void gammaDDDepthRange( GLcontext *ctx, GLclampd nearval,
 void gammaUpdateViewportOffset( GLcontext *ctx )
 {
    gammaContextPtr gmesa = GAMMA_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = gmesa->driDrawable;
+   __DRIdrawable *dPriv = gmesa->driDrawable;
    GLfloat xoffset = (GLfloat)dPriv->x;
    GLfloat yoffset = gmesa->driScreen->fbHeight - (GLfloat)dPriv->y - dPriv->h;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
diff --git a/src/mesa/drivers/dri/gamma/gamma_tex.c b/src/mesa/drivers/dri/gamma/gamma_tex.c
index 97797d4788..694e5eba5b 100644
--- a/src/mesa/drivers/dri/gamma/gamma_tex.c
+++ b/src/mesa/drivers/dri/gamma/gamma_tex.c
@@ -1,21 +1,18 @@
 
-#include <stdlib.h>
-#include <stdio.h>
-
 #include "main/glheader.h"
 #include "main/mtypes.h"
+#include "main/colormac.h"
 #include "main/imports.h"
 #include "main/simple_list.h"
 #include "main/enums.h"
+#include "main/mm.h"
 #include "main/texstore.h"
-#include "teximage.h"
-#include "main/texformat.h"
-#include "texobj.h"
+#include "main/teximage.h"
+#include "main/texobj.h"
+
 #include "swrast/swrast.h"
 
-#include "main/mm.h"
 #include "gammacontext.h"
-#include "colormac.h"
 
 
 /*
@@ -148,7 +145,7 @@ static void gammaTexParameter( GLcontext *ctx, GLenum target,
       break;
   
    case GL_TEXTURE_BORDER_COLOR:
-      gammaSetTexBorderColor( gmesa, t, tObj->BorderColor );
+      gammaSetTexBorderColor( gmesa, t, tObj->BorderColor.f );
       break;
 
    case GL_TEXTURE_BASE_LEVEL:
@@ -352,7 +349,7 @@ static void gammaBindTexture( GLcontext *ctx, GLenum target,
 
 	 gammaSetTexWrapping( t, tObj->WrapS, tObj->WrapT );
 	 gammaSetTexFilter( gmesa, t, tObj->MinFilter, tObj->MagFilter, bias );
-	 gammaSetTexBorderColor( gmesa, t, tObj->BorderColor );
+	 gammaSetTexBorderColor( gmesa, t, tObj->BorderColor.f );
       }
 }
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_xmesa.c b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
index 7b5b53589c..e49ab5bae3 100644
--- a/src/mesa/drivers/dri/gamma/gamma_xmesa.c
+++ b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
@@ -36,7 +36,7 @@
 #include "vbo/vbo.h"
 
 static GLboolean 
-gammaInitDriver(__DRIscreenPrivate *sPriv)
+gammaInitDriver(__DRIscreen *sPriv)
 {
     sPriv->private = (void *) gammaCreateScreen( sPriv );
 
@@ -49,7 +49,7 @@ gammaInitDriver(__DRIscreenPrivate *sPriv)
 }
 
 static void 
-gammaDestroyContext(__DRIcontextPrivate *driContextPriv)
+gammaDestroyContext(__DRIcontext *driContextPriv)
 {
     gammaContextPtr gmesa = (gammaContextPtr)driContextPriv->driverPrivate;
 
@@ -72,8 +72,8 @@ gammaDestroyContext(__DRIcontextPrivate *driContextPriv)
 
 
 static GLboolean
-gammaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                   __DRIdrawablePrivate *driDrawPriv,
+gammaCreateBuffer( __DRIscreen *driScrnPriv,
+                   __DRIdrawable *driDrawPriv,
                    const __GLcontextModes *mesaVis,
                    GLboolean isPixmap )
 {
@@ -94,17 +94,17 @@ gammaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-gammaDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+gammaDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
 static void
-gammaSwapBuffers( __DRIdrawablePrivate *dPriv )
+gammaSwapBuffers( __DRIdrawable *dPriv )
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
     gammaContextPtr gmesa;
-    __DRIscreenPrivate *driScrnPriv;
+    __DRIscreen *driScrnPriv;
     GLcontext *ctx;
 
     gmesa = (gammaContextPtr) dPriv->driContextPriv->driverPrivate;
@@ -127,7 +127,7 @@ gammaSwapBuffers( __DRIdrawablePrivate *dPriv )
 	int i;
 	int nRect = dPriv->numClipRects;
 	drm_clip_rect_t *pRect = dPriv->pClipRects;
-	__DRIscreenPrivate *driScrnPriv = gmesa->driScreen;
+	__DRIscreen *driScrnPriv = gmesa->driScreen;
    	GLINTDRIPtr gDRIPriv = (GLINTDRIPtr)driScrnPriv->pDevPriv;
 
 	CHECK_DMA_BUFFER(gmesa, 2);
@@ -193,9 +193,9 @@ gammaSwapBuffers( __DRIdrawablePrivate *dPriv )
 }
 
 static GLboolean 
-gammaMakeCurrent(__DRIcontextPrivate *driContextPriv,
-		 __DRIdrawablePrivate *driDrawPriv,
-		 __DRIdrawablePrivate *driReadPriv)
+gammaMakeCurrent(__DRIcontext *driContextPriv,
+		 __DRIdrawable *driDrawPriv,
+		 __DRIdrawable *driReadPriv)
 {
     if (driContextPriv) {
 	GET_CURRENT_CONTEXT(ctx);
@@ -232,7 +232,7 @@ newGammaCtx->new_state |= GAMMA_NEW_WINDOW; /* FIXME */
 
 
 static GLboolean 
-gammaUnbindContext( __DRIcontextPrivate *driContextPriv )
+gammaUnbindContext( __DRIcontext *driContextPriv )
 {
    return GL_TRUE;
 }
@@ -254,12 +254,19 @@ const struct __DriverAPIRec driDriverAPI = {
 /*
  * This is the bootstrap function for the driver.
  * The __driCreateScreen name is the symbol that libGL.so fetches.
- * Return:  pointer to a __DRIscreenPrivate.
+ * Return:  pointer to a __DRIscreen.
  */
 void *__driCreateScreen(Display *dpy, int scrn, __DRIscreen *psc,
                         int numConfigs, __GLXvisualConfig *config)
 {
-   __DRIscreenPrivate *psp;
+   __DRIscreen *psp;
    psp = __driUtilCreateScreen(dpy, scrn, psc, numConfigs, config, &gammaAPI);
    return (void *) psp;
 }
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
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/i810/i810context.c b/src/mesa/drivers/dri/i810/i810context.c
index 6785655686..bd9cfe5c0f 100644
--- a/src/mesa/drivers/dri/i810/i810context.c
+++ b/src/mesa/drivers/dri/i810/i810context.c
@@ -116,7 +116,7 @@ static void i810BufferSize(GLframebuffer *buffer, GLuint *width, GLuint *height)
 
 /* Extension strings exported by the i810 driver.
  */
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_env_add",            NULL },
@@ -170,12 +170,12 @@ static const struct dri_debug_control debug_control[] =
 
 GLboolean
 i810CreateContext( const __GLcontextModes *mesaVis,
-                   __DRIcontextPrivate *driContextPriv,
+                   __DRIcontext *driContextPriv,
                    void *sharedContextPrivate )
 {
    GLcontext *ctx, *shareCtx;
    i810ContextPtr imesa;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    i810ScreenPrivate *i810Screen = (i810ScreenPrivate *)sPriv->private;
    I810SAREAPtr saPriv = (I810SAREAPtr)
       (((GLubyte *)sPriv->pSAREA) + i810Screen->sarea_priv_offset);
@@ -337,7 +337,7 @@ i810CreateContext( const __GLcontextModes *mesaVis,
 }
 
 void
-i810DestroyContext(__DRIcontextPrivate *driContextPriv)
+i810DestroyContext(__DRIcontext *driContextPriv)
 {
    i810ContextPtr imesa = (i810ContextPtr) driContextPriv->driverPrivate;
 
@@ -378,7 +378,7 @@ i810DestroyContext(__DRIcontextPrivate *driContextPriv)
 
 void i810XMesaSetFrontClipRects( i810ContextPtr imesa )
 {
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
+   __DRIdrawable *dPriv = imesa->driDrawable;
 
    imesa->numClipRects = dPriv->numClipRects;
    imesa->pClipRects = dPriv->pClipRects;
@@ -392,7 +392,7 @@ void i810XMesaSetFrontClipRects( i810ContextPtr imesa )
 
 void i810XMesaSetBackClipRects( i810ContextPtr imesa )
 {
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
+   __DRIdrawable *dPriv = imesa->driDrawable;
 
    if (imesa->sarea->pf_enabled == 0 && dPriv->numBackClipRects == 0)
    {
@@ -430,7 +430,7 @@ static void i810XMesaWindowMoved( i810ContextPtr imesa )
 
 
 GLboolean
-i810UnbindContext(__DRIcontextPrivate *driContextPriv)
+i810UnbindContext(__DRIcontext *driContextPriv)
 {
    i810ContextPtr imesa = (i810ContextPtr) driContextPriv->driverPrivate;
    if (imesa) {
@@ -444,9 +444,9 @@ i810UnbindContext(__DRIcontextPrivate *driContextPriv)
 
 
 GLboolean
-i810MakeCurrent(__DRIcontextPrivate *driContextPriv,
-                __DRIdrawablePrivate *driDrawPriv,
-                __DRIdrawablePrivate *driReadPriv)
+i810MakeCurrent(__DRIcontext *driContextPriv,
+                __DRIdrawable *driDrawPriv,
+                __DRIdrawable *driReadPriv)
 {
    if (driContextPriv) {
       i810ContextPtr imesa = (i810ContextPtr) driContextPriv->driverPrivate;
@@ -504,8 +504,8 @@ i810UpdatePageFlipping( i810ContextPtr imesa )
 
 void i810GetLock( i810ContextPtr imesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
-   __DRIscreenPrivate *sPriv = imesa->driScreen;
+   __DRIdrawable *dPriv = imesa->driDrawable;
+   __DRIscreen *sPriv = imesa->driScreen;
    I810SAREAPtr sarea = imesa->sarea;
    int me = imesa->hHWContext;
    unsigned i;
@@ -551,7 +551,7 @@ void i810GetLock( i810ContextPtr imesa, GLuint flags )
 
 
 void
-i810SwapBuffers( __DRIdrawablePrivate *dPriv )
+i810SwapBuffers( __DRIdrawable *dPriv )
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
       i810ContextPtr imesa;
diff --git a/src/mesa/drivers/dri/i810/i810context.h b/src/mesa/drivers/dri/i810/i810context.h
index 4b8c71d7c6..19529db020 100644
--- a/src/mesa/drivers/dri/i810/i810context.h
+++ b/src/mesa/drivers/dri/i810/i810context.h
@@ -170,8 +170,8 @@ struct i810_context_t {
    drm_hw_lock_t *driHwLock;
    int driFd;
 
-   __DRIdrawablePrivate *driDrawable;
-   __DRIscreenPrivate *driScreen;
+   __DRIdrawable *driDrawable;
+   __DRIscreen *driScreen;
    i810ScreenPrivate *i810Screen; 
    I810SAREAPtr sarea;
 };
diff --git a/src/mesa/drivers/dri/i810/i810ioctl.c b/src/mesa/drivers/dri/i810/i810ioctl.c
index 3df9c2ac47..c631543d93 100644
--- a/src/mesa/drivers/dri/i810/i810ioctl.c
+++ b/src/mesa/drivers/dri/i810/i810ioctl.c
@@ -50,8 +50,8 @@ static drmBufPtr i810_get_buffer_ioctl( i810ContextPtr imesa )
 static void i810Clear( GLcontext *ctx, GLbitfield mask )
 {
    i810ContextPtr imesa = I810_CONTEXT( ctx );
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
-   const GLuint colorMask = *((GLuint *) &ctx->Color.ColorMask);
+   __DRIdrawable *dPriv = imesa->driDrawable;
+   const GLuint colorMask = *((GLuint *) &ctx->Color.ColorMask[0]);
    drmI810Clear clear;
    unsigned int i;
 
@@ -149,7 +149,7 @@ static void i810Clear( GLcontext *ctx, GLbitfield mask )
 /*
  * Copy the back buffer to the front buffer. 
  */
-void i810CopyBuffer( const __DRIdrawablePrivate *dPriv ) 
+void i810CopyBuffer( const __DRIdrawable *dPriv ) 
 {
    i810ContextPtr imesa;
    drm_clip_rect_t *pbox;
@@ -197,7 +197,7 @@ void i810CopyBuffer( const __DRIdrawablePrivate *dPriv )
 /*
  * XXX implement when full-screen extension is done.
  */
-void i810PageFlip( const __DRIdrawablePrivate *dPriv ) 
+void i810PageFlip( const __DRIdrawable *dPriv ) 
 {
   i810ContextPtr imesa;
   int tmp, ret;
diff --git a/src/mesa/drivers/dri/i810/i810ioctl.h b/src/mesa/drivers/dri/i810/i810ioctl.h
index dfd6e21088..926e38ce51 100644
--- a/src/mesa/drivers/dri/i810/i810ioctl.h
+++ b/src/mesa/drivers/dri/i810/i810ioctl.h
@@ -14,8 +14,8 @@ void i810WaitAge( i810ContextPtr imesa, int age );
 void i810DmaFinish( i810ContextPtr imesa );
 void i810RegetLockQuiescent( i810ContextPtr imesa );
 void i810InitIoctlFuncs( struct dd_function_table *functions );
-void i810CopyBuffer( const __DRIdrawablePrivate *dpriv );
-void i810PageFlip( const __DRIdrawablePrivate *dpriv );
+void i810CopyBuffer( const __DRIdrawable *dpriv );
+void i810PageFlip( const __DRIdrawable *dpriv );
 int i810_check_copy(int fd);
 
 #define I810_STATECHANGE(imesa, flag)				\
diff --git a/src/mesa/drivers/dri/i810/i810screen.c b/src/mesa/drivers/dri/i810/i810screen.c
index 6e49f3466c..2a30782afd 100644
--- a/src/mesa/drivers/dri/i810/i810screen.c
+++ b/src/mesa/drivers/dri/i810/i810screen.c
@@ -53,10 +53,8 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "GL/internal/dri_interface.h"
 
-extern const struct dri_extension card_extensions[];
-
 static const __DRIconfig **
-i810FillInModes( __DRIscreenPrivate *psp,
+i810FillInModes( __DRIscreen *psp,
 		 unsigned pixel_bits, unsigned depth_bits,
 		 unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -166,8 +164,6 @@ i810InitScreen(__DRIscreen *sPriv)
       return NULL;
    }
 
-   driInitExtensions( NULL, card_extensions, GL_TRUE );
-
    if (sPriv->devPrivSize != sizeof(I810DRIRec)) {
       fprintf(stderr,"\nERROR!  sizeof(I810DRIRec) does not match passed size from device driver\n");
       return GL_FALSE;
@@ -259,7 +255,7 @@ i810InitScreen(__DRIscreen *sPriv)
 }
 
 static void
-i810DestroyScreen(__DRIscreenPrivate *sPriv)
+i810DestroyScreen(__DRIscreen *sPriv)
 {
    i810ScreenPrivate *i810Screen = (i810ScreenPrivate *)sPriv->private;
 
@@ -278,8 +274,8 @@ i810DestroyScreen(__DRIscreenPrivate *sPriv)
  * Create a buffer which corresponds to the window.
  */
 static GLboolean
-i810CreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                  __DRIdrawablePrivate *driDrawPriv,
+i810CreateBuffer( __DRIscreen *driScrnPriv,
+                  __DRIdrawable *driDrawPriv,
                   const __GLcontextModes *mesaVis,
                   GLboolean isPixmap )
 {
@@ -293,7 +289,7 @@ i810CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  driScrnPriv->pFB,
                                  screen->cpp,
                                  /*screen->frontOffset*/0, screen->backPitch,
@@ -304,7 +300,7 @@ i810CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  screen->back.map,
                                  screen->cpp,
                                  screen->backOffset, screen->backPitch,
@@ -315,7 +311,7 @@ i810CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
+            = driNewRenderbuffer(MESA_FORMAT_Z16,
                                  screen->depth.map,
                                  screen->cpp,
                                  screen->depthOffset, screen->backPitch,
@@ -339,7 +335,7 @@ i810CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-i810DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+i810DestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
@@ -360,3 +356,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/i810/i810screen.h b/src/mesa/drivers/dri/i810/i810screen.h
index b29937665a..734e2fb002 100644
--- a/src/mesa/drivers/dri/i810/i810screen.h
+++ b/src/mesa/drivers/dri/i810/i810screen.h
@@ -71,7 +71,7 @@ typedef struct {
    int textureSize;
    int logTextureGranularity;
 
-   __DRIscreenPrivate *driScrnPriv;
+   __DRIscreen *driScrnPriv;
    drmBufMapPtr  bufs;
    unsigned int sarea_priv_offset;
 } i810ScreenPrivate;
@@ -79,21 +79,21 @@ typedef struct {
 
 extern GLboolean
 i810CreateContext( const __GLcontextModes *mesaVis,
-                   __DRIcontextPrivate *driContextPriv,
+                   __DRIcontext *driContextPriv,
                    void *sharedContextPrivate );
 
 extern void
-i810DestroyContext(__DRIcontextPrivate *driContextPriv);
+i810DestroyContext(__DRIcontext *driContextPriv);
 
 extern GLboolean
-i810UnbindContext(__DRIcontextPrivate *driContextPriv);
+i810UnbindContext(__DRIcontext *driContextPriv);
 
 extern GLboolean
-i810MakeCurrent(__DRIcontextPrivate *driContextPriv,
-                __DRIdrawablePrivate *driDrawPriv,
-                __DRIdrawablePrivate *driReadPriv);
+i810MakeCurrent(__DRIcontext *driContextPriv,
+                __DRIdrawable *driDrawPriv,
+                __DRIdrawable *driReadPriv);
 
 extern void
-i810SwapBuffers(__DRIdrawablePrivate *driDrawPriv);
+i810SwapBuffers(__DRIdrawable *driDrawPriv);
 
 #endif
diff --git a/src/mesa/drivers/dri/i810/i810span.c b/src/mesa/drivers/dri/i810/i810span.c
index 510723f445..6576f6745e 100644
--- a/src/mesa/drivers/dri/i810/i810span.c
+++ b/src/mesa/drivers/dri/i810/i810span.c
@@ -15,7 +15,7 @@
 
 #define LOCAL_VARS					\
    i810ContextPtr imesa = I810_CONTEXT(ctx);	        \
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;	\
+   __DRIdrawable *dPriv = imesa->driDrawable;	\
    driRenderbuffer *drb = (driRenderbuffer *) rb;	\
    GLuint pitch = drb->pitch;				\
    GLuint height = dPriv->h;				\
@@ -27,7 +27,7 @@
 
 #define LOCAL_DEPTH_VARS				\
    i810ContextPtr imesa = I810_CONTEXT(ctx);	        \
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;	\
+   __DRIdrawable *dPriv = imesa->driDrawable;	\
    driRenderbuffer *drb = (driRenderbuffer *) rb;	\
    GLuint pitch = drb->pitch;				\
    GLuint height = dPriv->h;				\
diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c
index 1e7a6cfe47..642245c61c 100644
--- a/src/mesa/drivers/dri/i810/i810state.c
+++ b/src/mesa/drivers/dri/i810/i810state.c
@@ -641,7 +641,7 @@ static void i810Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 
 void i810EmitDrawingRectangle( i810ContextPtr imesa )
 {
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
+   __DRIdrawable *dPriv = imesa->driDrawable;
    i810ScreenPrivate *i810Screen = imesa->i810Screen;
    int x0 = imesa->drawX;
    int y0 = imesa->drawY;
diff --git a/src/mesa/drivers/dri/i810/i810tex.c b/src/mesa/drivers/dri/i810/i810tex.c
index cd6e1a8e6e..e764644a6c 100644
--- a/src/mesa/drivers/dri/i810/i810tex.c
+++ b/src/mesa/drivers/dri/i810/i810tex.c
@@ -28,7 +28,6 @@
 #include "main/simple_list.h"
 #include "main/enums.h"
 #include "main/texstore.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/colormac.h"
@@ -211,7 +210,7 @@ i810AllocTexObj( GLcontext *ctx, struct gl_texture_object *texObj )
       i810SetTexWrapping( t, texObj->WrapS, texObj->WrapT );
       /*i830SetTexMaxAnisotropy( t, texObj->MaxAnisotropy );*/
       i810SetTexFilter( imesa, t, texObj->MinFilter, texObj->MagFilter, bias );
-      i810SetTexBorderColor( t, texObj->BorderColor );
+      i810SetTexBorderColor( t, texObj->BorderColor.f );
    }
 
    return t;
@@ -252,7 +251,7 @@ static void i810TexParameter( GLcontext *ctx, GLenum target,
       break;
   
    case GL_TEXTURE_BORDER_COLOR:
-      i810SetTexBorderColor( t, tObj->BorderColor );
+      i810SetTexBorderColor( t, tObj->BorderColor.f );
       break;
 
    case GL_TEXTURE_BASE_LEVEL:
@@ -440,7 +439,7 @@ static void i810DeleteTexture( GLcontext *ctx, struct gl_texture_object *tObj )
  * The i810 only supports 5 texture modes that are useful to Mesa.  That
  * makes this routine pretty simple.
  */
-static const struct gl_texture_format *
+static gl_format
 i810ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 			 GLenum format, GLenum type )
 {
@@ -458,9 +457,9 @@ i810ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       if ( ((format == GL_BGRA) && (type == GL_UNSIGNED_SHORT_1_5_5_5_REV))
 	   || ((format == GL_RGBA) && (type == GL_UNSIGNED_SHORT_5_5_5_1))
 	   || (internalFormat == GL_RGB5_A1) ) {
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       }
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case 3:
    case GL_RGB:
@@ -472,7 +471,7 @@ i810ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_ALPHA:
    case GL_ALPHA4:
@@ -502,21 +501,21 @@ i810ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY12:
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_al88;
+      return MESA_FORMAT_AL88;
 
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_MESA ||
 	  type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    default:
       fprintf(stderr, "unexpected texture format in %s\n", __FUNCTION__);
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 
-   return NULL; /* never get here */
+   return MESA_FORMAT_NONE; /* never get here */
 }
 
 /**
diff --git a/src/mesa/drivers/dri/i810/i810tex.h b/src/mesa/drivers/dri/i810/i810tex.h
index d980927030..28958dcb4b 100644
--- a/src/mesa/drivers/dri/i810/i810tex.h
+++ b/src/mesa/drivers/dri/i810/i810tex.h
@@ -29,7 +29,6 @@
 #include "main/mtypes.h"
 #include "main/mm.h"
 
-#include "i810context.h"
 #include "i810_3d_reg.h"
 #include "texmem.h"
 
diff --git a/src/mesa/drivers/dri/i810/i810texmem.c b/src/mesa/drivers/dri/i810/i810texmem.c
index 5ad66dbf5c..d93afbf9ef 100644
--- a/src/mesa/drivers/dri/i810/i810texmem.c
+++ b/src/mesa/drivers/dri/i810/i810texmem.c
@@ -30,7 +30,6 @@
 #include "main/enums.h"
 #include "main/colormac.h"
 #include "main/mm.h"
-#include "main/texformat.h"
 
 #include "i810screen.h"
 #include "i810_dri.h"
@@ -92,44 +91,47 @@ static void i810UploadTexLevel( i810ContextPtr imesa,
 {
    const struct gl_texture_image *image = t->image[hwlevel].image;
    int j;
+   GLuint texelBytes;
 
    if (!image || !image->Data)
       return;
 
-   if (image->Width * image->TexFormat->TexelBytes == t->Pitch) {
+   texelBytes = _mesa_get_format_bytes(image->TexFormat);
+
+   if (image->Width * texelBytes == t->Pitch) {
 	 GLubyte *dst = (GLubyte *)(t->BufAddr + t->image[hwlevel].offset);
 	 GLubyte *src = (GLubyte *)image->Data;
 	 
 	 memcpy( dst, src, t->Pitch * image->Height );
    }
-   else switch (image->TexFormat->TexelBytes) {
-   case 1:
-      {
-	 GLubyte *dst = (GLubyte *)(t->BufAddr + t->image[hwlevel].offset);
-	 GLubyte *src = (GLubyte *)image->Data;
-
-	 for (j = 0 ; j < image->Height ; j++, dst += t->Pitch) {
-	    __memcpy(dst, src, image->Width );
-	    src += image->Width;
-	 }
+   else {
+      switch (texelBytes) {
+      case 1:
+         {
+            GLubyte *dst = (GLubyte *)(t->BufAddr + t->image[hwlevel].offset);
+            GLubyte *src = (GLubyte *)image->Data;
+
+            for (j = 0 ; j < image->Height ; j++, dst += t->Pitch) {
+               __memcpy(dst, src, image->Width );
+               src += image->Width;
+            }
+         }
+         break;
+      case 2:
+         {
+            GLushort *dst = (GLushort *)(t->BufAddr + t->image[hwlevel].offset);
+            GLushort *src = (GLushort *)image->Data;
+
+            for (j = 0 ; j < image->Height ; j++, dst += (t->Pitch/2)) {
+               __memcpy(dst, src, image->Width * 2 );
+               src += image->Width;
+            }
+         }
+         break;
+      default:
+         fprintf(stderr, "%s: Not supported texel size %d\n",
+                 __FUNCTION__, texelBytes);
       }
-      break;
-
-   case 2:
-      {
-	 GLushort *dst = (GLushort *)(t->BufAddr + t->image[hwlevel].offset);
-	 GLushort *src = (GLushort *)image->Data;
-
-	 for (j = 0 ; j < image->Height ; j++, dst += (t->Pitch/2)) {
-	    __memcpy(dst, src, image->Width * 2 );
-	    src += image->Width;
-	 }
-      }
-      break;
-
-   default:
-      fprintf(stderr, "%s: Not supported texel size %d\n",
-	      __FUNCTION__, image->TexFormat->TexelBytes);
    }
 }
 
diff --git a/src/mesa/drivers/dri/i810/i810texstate.c b/src/mesa/drivers/dri/i810/i810texstate.c
index 0e09f54c41..bff28c11c8 100644
--- a/src/mesa/drivers/dri/i810/i810texstate.c
+++ b/src/mesa/drivers/dri/i810/i810texstate.c
@@ -25,7 +25,6 @@
 #include "main/glheader.h"
 #include "main/macros.h"
 #include "main/mtypes.h"
-#include "main/texformat.h"
 #include "main/simple_list.h"
 #include "main/enums.h"
 #include "main/mm.h"
@@ -53,7 +52,7 @@ static void i810SetTexImages( i810ContextPtr imesa,
 /*     fprintf(stderr, "%s\n", __FUNCTION__); */
 
    t->texelBytes = 2;
-   switch (baseImage->TexFormat->MesaFormat) {
+   switch (baseImage->TexFormat) {
    case MESA_FORMAT_ARGB1555:
       textureFormat = MI1_FMT_16BPP | MI1_PF_16BPP_ARGB1555;
       break;
diff --git a/src/mesa/drivers/dri/i810/i810tris.c b/src/mesa/drivers/dri/i810/i810tris.c
index b508496fb6..213ba541ce 100644
--- a/src/mesa/drivers/dri/i810/i810tris.c
+++ b/src/mesa/drivers/dri/i810/i810tris.c
@@ -270,7 +270,8 @@ do {							\
 
 #define LOCAL_VARS(n)							\
    i810ContextPtr imesa = I810_CONTEXT(ctx);				\
-   GLuint color[n], spec[n];						\
+   GLuint color[n] = { 0 };						\
+   GLuint spec[n] = { 0 };						\
    GLuint coloroffset = (imesa->vertex_size == 4 ? 3 : 4);		\
    GLboolean havespec = (imesa->vertex_size > 4);			\
    (void) color; (void) spec; (void) coloroffset; (void) havespec;
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 393312e732..cf32476f40 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -34,7 +34,6 @@ DRIVER_SOURCES = \
 	intel_pixel_read.c \
 	intel_buffers.c \
 	intel_blit.c \
-	intel_swapbuffers.c \
 	i915_tex_layout.c \
 	i915_texstate.c \
 	i915_context.c \
@@ -64,7 +63,8 @@ DRIVER_DEFINES = -I../intel -I../intel/server -DI915 \
 	$(shell pkg-config libdrm --atleast-version=2.3.1 \
 				&& echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP")
 
-DRI_LIB_DEPS += -ldrm_intel
+INCLUDES += $(INTEL_CFLAGS)
+DRI_LIB_DEPS += $(INTEL_LIBS)
 
 include ../Makefile.template
 
@@ -72,4 +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_context.c b/src/mesa/drivers/dri/i915/i830_context.c
index 840946f908..4cb6305988 100644
--- a/src/mesa/drivers/dri/i915/i830_context.c
+++ b/src/mesa/drivers/dri/i915/i830_context.c
@@ -53,7 +53,7 @@ extern const struct tnl_pipeline_stage *intel_pipeline[];
 
 GLboolean
 i830CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
+                  __DRIcontext * driContextPriv,
                   void *sharedContextPrivate)
 {
    struct dd_function_table functions;
diff --git a/src/mesa/drivers/dri/i915/i830_context.h b/src/mesa/drivers/dri/i915/i830_context.h
index f73cbbf88b..592ae53976 100644
--- a/src/mesa/drivers/dri/i915/i830_context.h
+++ b/src/mesa/drivers/dri/i915/i830_context.h
@@ -178,7 +178,7 @@ i830_state_draw_region(struct intel_context *intel,
  */
 extern GLboolean
 i830CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
+                  __DRIcontext * driContextPriv,
                   void *sharedContextPrivate);
 
 /* i830_tex.c, i830_texstate.c
diff --git a/src/mesa/drivers/dri/i915/i830_state.c b/src/mesa/drivers/dri/i915/i830_state.c
index 645ebe3057..acda7e70de 100644
--- a/src/mesa/drivers/dri/i915/i830_state.c
+++ b/src/mesa/drivers/dri/i915/i830_state.c
@@ -620,7 +620,7 @@ i830LineWidth(GLcontext * ctx, GLfloat widthf)
    DBG("%s\n", __FUNCTION__);
    
    width = (int) (widthf * 2);
-   CLAMP_SELF(width, 1, 15);
+   width = CLAMP(width, 1, 15);
 
    state5 = i830->state.Ctx[I830_CTXREG_STATE5] & ~FIXED_LINE_WIDTH_MASK;
    state5 |= (ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(width));
@@ -639,7 +639,7 @@ i830PointSize(GLcontext * ctx, GLfloat size)
 
    DBG("%s\n", __FUNCTION__);
    
-   CLAMP_SELF(point_size, 1, 256);
+   point_size = CLAMP(point_size, 1, 256);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE5] &= ~FIXED_POINT_WIDTH_MASK;
    i830->state.Ctx[I830_CTXREG_STATE5] |= (ENABLE_FIXED_POINT_WIDTH |
diff --git a/src/mesa/drivers/dri/i915/i830_texblend.c b/src/mesa/drivers/dri/i915/i830_texblend.c
index 09f7f37e76..3f64be8c96 100644
--- a/src/mesa/drivers/dri/i915/i830_texblend.c
+++ b/src/mesa/drivers/dri/i915/i830_texblend.c
@@ -30,7 +30,6 @@
 #include "main/mtypes.h"
 #include "main/simple_list.h"
 #include "main/enums.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/mm.h"
 
diff --git a/src/mesa/drivers/dri/i915/i830_texstate.c b/src/mesa/drivers/dri/i915/i830_texstate.c
index 6f998fa6f7..7525f9f2e0 100644
--- a/src/mesa/drivers/dri/i915/i830_texstate.c
+++ b/src/mesa/drivers/dri/i915/i830_texstate.c
@@ -27,7 +27,7 @@
 
 #include "main/mtypes.h"
 #include "main/enums.h"
-#include "main/texformat.h"
+#include "main/colormac.h"
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
@@ -56,10 +56,9 @@ translate_texture_format(GLuint mesa_format, GLuint internal_format)
    case MESA_FORMAT_ARGB4444:
       return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      if (internal_format == GL_RGB)
-	 return MAPSURF_32BIT | MT_32BIT_XRGB8888;
-      else
-	 return MAPSURF_32BIT | MT_32BIT_ARGB8888;
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
+   case MESA_FORMAT_XRGB8888:
+      return MAPSURF_32BIT | MT_32BIT_XRGB8888;
    case MESA_FORMAT_YCBCR_REV:
       return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
@@ -160,13 +159,20 @@ i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 
       pitch = intelObj->pitchOverride;
    } else {
+      GLuint dst_x, dst_y;
+
+      intel_miptree_get_image_offset(intelObj->mt, intelObj->firstLevel, 0, 0,
+				     &dst_x, &dst_y);
+
       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);
+      /* XXX: This calculation is probably broken for tiled images with
+       * a non-page-aligned offset.
+       */
+      i830->state.tex_offset[unit] = (dst_x + dst_y * intelObj->mt->pitch) *
+	 intelObj->mt->cpp;
 
-      format = translate_texture_format(firstImage->TexFormat->MesaFormat,
+      format = translate_texture_format(firstImage->TexFormat,
 					firstImage->InternalFormat);
       pitch = intelObj->mt->pitch * intelObj->mt->cpp;
    }
@@ -298,16 +304,15 @@ i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
    }
 
    /* convert border color from float to ubyte */
-   CLAMPED_FLOAT_TO_UBYTE(border[0], tObj->BorderColor[0]);
-   CLAMPED_FLOAT_TO_UBYTE(border[1], tObj->BorderColor[1]);
-   CLAMPED_FLOAT_TO_UBYTE(border[2], tObj->BorderColor[2]);
-   CLAMPED_FLOAT_TO_UBYTE(border[3], tObj->BorderColor[3]);
-
-   state[I830_TEXREG_TM0S4] = INTEL_PACKCOLOR8888(border[0],
-                                                  border[1],
-                                                  border[2],
-                                                  border[3]);
-
+   CLAMPED_FLOAT_TO_UBYTE(border[0], tObj->BorderColor.f[0]);
+   CLAMPED_FLOAT_TO_UBYTE(border[1], tObj->BorderColor.f[1]);
+   CLAMPED_FLOAT_TO_UBYTE(border[2], tObj->BorderColor.f[2]);
+   CLAMPED_FLOAT_TO_UBYTE(border[3], tObj->BorderColor.f[3]);
+
+   state[I830_TEXREG_TM0S4] = PACK_COLOR_8888(border[3],
+					      border[0],
+					      border[1],
+					      border[2]);
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
    /* memcmp was already disabled, but definitely won't work as the
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index 983f6724c9..4471ca2bbb 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -26,7 +26,6 @@
  **************************************************************************/
 
 #include "glapi/glapi.h"
-#include "main/texformat.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
@@ -127,7 +126,7 @@ i830_render_start(struct intel_context *intel)
 
       for (i = 0; i < I830_TEX_UNITS; i++) {
          if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_TEX(i))) {
-            GLuint sz = VB->TexCoordPtr[i]->size;
+            GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
             GLuint emit;
             GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] &
                           ~TEXCOORDTYPE_MASK);
@@ -299,7 +298,7 @@ i830_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH(29, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(29);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -367,7 +366,7 @@ i830_emit_invarient_state(struct intel_context *intel)
 
 
 #define emit( intel, state, size )			\
-   intel_batchbuffer_data(intel->batch, state, size, IGNORE_CLIPRECTS )
+   intel_batchbuffer_data(intel->batch, state, size )
 
 static GLuint
 get_dirty(struct i830_hw_state *state)
@@ -430,13 +429,9 @@ i830_emit_state(struct intel_context *intel)
     * It might be better to talk about explicit places where
     * scheduling is allowed, rather than assume that it is whenever a
     * batchbuffer fills up.
-    *
-    * Set the space as LOOP_CLIPRECTS now, since that's what our primitives
-    * will be emitted under.
     */
    intel_batchbuffer_require_space(intel->batch,
-				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE,
-				   LOOP_CLIPRECTS);
+				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE);
    count = 0;
  again:
    aper_count = 0;
@@ -492,17 +487,14 @@ i830_emit_state(struct intel_context *intel)
    }
 
    if (dirty & I830_UPLOAD_BUFFERS) {
-      GLuint count = 9; 
+      GLuint count = 15;
 
       DBG("I830_UPLOAD_BUFFERS:\n");
 
       if (state->depth_region)
           count += 3;
 
-      if (intel->constant_cliprect)
-          count += 6;
-
-      BEGIN_BATCH(count, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(count);
       OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR0]);
       OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR1]);
       OUT_RELOC(state->draw_region->buffer,
@@ -524,15 +516,13 @@ i830_emit_state(struct intel_context *intel)
       OUT_BATCH(state->Buffer[I830_DESTREG_SR1]);
       OUT_BATCH(state->Buffer[I830_DESTREG_SR2]);
 
-      if (intel->constant_cliprect) {
-	 assert(state->Buffer[I830_DESTREG_DRAWRECT0] != MI_NOOP);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT0]);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT1]);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT2]);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT3]);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT4]);
-	 OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT5]);
-      }
+      assert(state->Buffer[I830_DESTREG_DRAWRECT0] != MI_NOOP);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT1]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT2]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT3]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT4]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DRAWRECT5]);
       ADVANCE_BATCH();
    }
    
@@ -545,7 +535,7 @@ i830_emit_state(struct intel_context *intel)
       if ((dirty & I830_UPLOAD_TEX(i))) {
          DBG("I830_UPLOAD_TEX(%d):\n", i);
 
-         BEGIN_BATCH(I830_TEX_SETUP_SIZE + 1, IGNORE_CLIPRECTS);
+         BEGIN_BATCH(I830_TEX_SETUP_SIZE + 1);
          OUT_BATCH(state->Tex[i][I830_TEXREG_TM0LI]);
 
          if (state->tex_buffer[i]) {
@@ -646,8 +636,9 @@ i830_state_draw_region(struct intel_context *intel,
             DSTORG_VERT_BIAS(0x8) | DEPTH_IS_Z);    /* .5 */
 
    if (irb != NULL) {
-      switch (irb->texformat->MesaFormat) {
+      switch (irb->Base.Format) {
       case MESA_FORMAT_ARGB8888:
+      case MESA_FORMAT_XRGB8888:
 	 value |= DV_PF_8888;
 	 break;
       case MESA_FORMAT_RGB565:
@@ -661,7 +652,7 @@ i830_state_draw_region(struct intel_context *intel,
 	 break;
       default:
 	 _mesa_problem(ctx, "Bad renderbuffer format: %d\n",
-		       irb->texformat->MesaFormat);
+		       irb->Base.Format);
       }
    }
 
@@ -673,23 +664,14 @@ i830_state_draw_region(struct intel_context *intel,
    }
    state->Buffer[I830_DESTREG_DV1] = value;
 
-   if (intel->constant_cliprect) {
-      state->Buffer[I830_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
-      state->Buffer[I830_DESTREG_DRAWRECT1] = 0;
-      state->Buffer[I830_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
-      state->Buffer[I830_DESTREG_DRAWRECT3] =
-	 (ctx->DrawBuffer->Width & 0xffff) |
-	 (ctx->DrawBuffer->Height << 16);
-      state->Buffer[I830_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
-      state->Buffer[I830_DESTREG_DRAWRECT5] = 0;
-   } else {
-      state->Buffer[I830_DESTREG_DRAWRECT0] = MI_NOOP;
-      state->Buffer[I830_DESTREG_DRAWRECT1] = MI_NOOP;
-      state->Buffer[I830_DESTREG_DRAWRECT2] = MI_NOOP;
-      state->Buffer[I830_DESTREG_DRAWRECT3] = MI_NOOP;
-      state->Buffer[I830_DESTREG_DRAWRECT4] = MI_NOOP;
-      state->Buffer[I830_DESTREG_DRAWRECT5] = MI_NOOP;
-   }
+   state->Buffer[I830_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
+   state->Buffer[I830_DESTREG_DRAWRECT1] = 0;
+   state->Buffer[I830_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
+   state->Buffer[I830_DESTREG_DRAWRECT3] =
+      (ctx->DrawBuffer->Width & 0xffff) |
+      (ctx->DrawBuffer->Height << 16);
+   state->Buffer[I830_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
+   state->Buffer[I830_DESTREG_DRAWRECT5] = 0;
 
    I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
 
@@ -714,20 +696,8 @@ i830_new_batch(struct intel_context *intel)
 {
    struct i830_context *i830 = i830_context(&intel->ctx);
    i830->state.emitted = 0;
-
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!intel->no_batch_wrap);
-}
-
-
-
-static GLuint
-i830_flush_cmd(void)
-{
-   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
-
 static void 
 i830_assert_not_dirty( struct intel_context *intel )
 {
@@ -753,7 +723,6 @@ i830InitVtbl(struct i830_context *i830)
    i830->intel.vtbl.reduced_primitive_state = i830_reduced_primitive_state;
    i830->intel.vtbl.set_draw_region = i830_set_draw_region;
    i830->intel.vtbl.update_texture_state = i830UpdateTextureState;
-   i830->intel.vtbl.flush_cmd = i830_flush_cmd;
    i830->intel.vtbl.render_start = i830_render_start;
    i830->intel.vtbl.render_prevalidate = i830_render_prevalidate;
    i830->intel.vtbl.assert_not_dirty = i830_assert_not_dirty;
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index 7d4c7cfbab..7c7711da09 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -100,7 +100,7 @@ extern const struct tnl_pipeline_stage *intel_pipeline[];
 
 GLboolean
 i915CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
+                  __DRIcontext * driContextPriv,
                   void *sharedContextPrivate)
 {
    struct dd_function_table functions;
@@ -143,6 +143,9 @@ i915CreateContext(const __GLcontextModes * mesaVis,
    ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS;
    ctx->Const.MaxVarying = I915_TEX_UNITS;
+   ctx->Const.MaxCombinedTextureImageUnits =
+      ctx->Const.MaxVertexTextureImageUnits +
+      ctx->Const.MaxTextureImageUnits;
 
    /* 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 082d614442..f55b551139 100644
--- a/src/mesa/drivers/dri/i915/i915_context.h
+++ b/src/mesa/drivers/dri/i915/i915_context.h
@@ -39,6 +39,7 @@
 #define I915_FALLBACK_LOGICOP		 0x20000
 #define I915_FALLBACK_POLYGON_SMOOTH	 0x40000
 #define I915_FALLBACK_POINT_SMOOTH	 0x80000
+#define I915_FALLBACK_POINT_SPRITE_COORD_ORIGIN	 0x100000
 
 #define I915_UPLOAD_CTX              0x1
 #define I915_UPLOAD_BUFFERS          0x2
@@ -317,7 +318,7 @@ do {									\
  * i915_context.c
  */
 extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
+                                   __DRIcontext * driContextPriv,
                                    void *sharedContextPrivate);
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_debug.c b/src/mesa/drivers/dri/i915/i915_debug.c
index f7bb7ea44c..fecfac3033 100644
--- a/src/mesa/drivers/dri/i915/i915_debug.c
+++ b/src/mesa/drivers/dri/i915/i915_debug.c
@@ -806,6 +806,7 @@ static GLboolean i915_debug_packet( struct debug_stream *stream )
       default:
 	 return debug(stream, "", 0);
       }
+      break;
    default:
       assert(0);
       return 0;
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index d9c61446f5..a273bd28ea 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -663,7 +663,7 @@ upload_program(struct i915_fragment_program *p)
 			 A0_MOV,
 			 get_result_vector(p, inst),
 			 get_result_flags(inst), 0,
-			 swizzle(src0, ZERO, ZERO, ZERO, ZERO), 0, 0);
+			 swizzle(tmp, ZERO, ZERO, ZERO, ZERO), 0, 0);
 
       case OPCODE_POW:
          src0 = src_vector(p, &inst->SrcReg[0], program);
@@ -1301,7 +1301,7 @@ i915ValidateFragmentProgram(struct i915_context *i915)
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
       if (inputsRead & FRAG_BIT_TEX(i)) {
-         int sz = VB->TexCoordPtr[i]->size;
+         int sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
          s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
          s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
diff --git a/src/mesa/drivers/dri/i915/i915_program.c b/src/mesa/drivers/dri/i915/i915_program.c
index e7908bd48f..3902c69097 100644
--- a/src/mesa/drivers/dri/i915/i915_program.c
+++ b/src/mesa/drivers/dri/i915/i915_program.c
@@ -245,7 +245,7 @@ GLuint i915_emit_texld( struct i915_fragment_program *p,
    }
    else {
       assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
-      assert(dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
+      assert(dest == UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
       /* Can't use unsaved temps for coords, as the phase boundary would result
        * in the contents becoming undefined.
        */
diff --git a/src/mesa/drivers/dri/i915/i915_reg.h b/src/mesa/drivers/dri/i915/i915_reg.h
index b5fa7fddb9..7f31ff674f 100644
--- a/src/mesa/drivers/dri/i915/i915_reg.h
+++ b/src/mesa/drivers/dri/i915/i915_reg.h
@@ -626,9 +626,9 @@
 #define    MT_32BIT_AWVU2101010	           (0xA<<3)
 #define    MT_32BIT_GR1616	           (0xB<<3)
 #define    MT_32BIT_VU1616	           (0xC<<3)
-#define    MT_32BIT_xI824	           (0xD<<3)
-#define    MT_32BIT_xA824	           (0xE<<3)
-#define    MT_32BIT_xL824	           (0xF<<3)
+#define    MT_32BIT_x8I24	           (0xD<<3)
+#define    MT_32BIT_x8L24	           (0xE<<3)
+#define    MT_32BIT_x8A24	           (0xF<<3)
 #define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
 #define    MT_422_YCRCB_NORMAL	           (1<<3)
 #define    MT_422_YCRCB_SWAPUV	           (2<<3)
diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c
index b60efea75b..9d7a9e1dfe 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -571,7 +571,7 @@ i915LineWidth(GLcontext * ctx, GLfloat widthf)
    DBG("%s\n", __FUNCTION__);
    
    width = (int) (widthf * 2);
-   CLAMP_SELF(width, 1, 0xf);
+   width = CLAMP(width, 1, 0xf);
    lis4 |= width << S4_LINE_WIDTH_SHIFT;
 
    if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
@@ -585,11 +585,11 @@ i915PointSize(GLcontext * ctx, GLfloat size)
 {
    struct i915_context *i915 = I915_CONTEXT(ctx);
    int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_POINT_WIDTH_MASK;
-   GLint point_size = (int) size;
+   GLint point_size = (int) round(size);
 
    DBG("%s\n", __FUNCTION__);
    
-   CLAMP_SELF(point_size, 1, 255);
+   point_size = CLAMP(point_size, 1, 255);
    lis4 |= point_size << S4_POINT_WIDTH_SHIFT;
 
    if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
@@ -599,6 +599,24 @@ i915PointSize(GLcontext * ctx, GLfloat size)
 }
 
 
+static void
+i915PointParameterfv(GLcontext * ctx, GLenum pname, const GLfloat *params)
+{
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+
+   switch (pname) {
+   case GL_POINT_SPRITE_COORD_ORIGIN:
+      /* This could be supported, but it would require modifying the fragment
+       * program to invert the y component of the texture coordinate by
+       * inserting a 'SUB tc.y, {1.0}.xxxx, tc' instruction.
+       */
+      FALLBACK(&i915->intel, I915_FALLBACK_POINT_SPRITE_COORD_ORIGIN,
+	       (params[0] != GL_UPPER_LEFT));
+      break;
+   }
+}
+
+
 /* =============================================================
  * Color masks
  */
@@ -939,6 +957,17 @@ i915Enable(GLcontext * ctx, GLenum cap, GLboolean state)
    case GL_POLYGON_SMOOTH:
       break;
 
+   case GL_POINT_SPRITE:
+      /* This state change is handled in i915_reduced_primitive_state because
+       * the hardware bit should only be set when rendering points.
+       */
+      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
+      if (state)
+	 i915->state.Ctx[I915_CTXREG_LIS4] |= S4_SPRITE_POINT_ENABLE;
+      else
+	 i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_SPRITE_POINT_ENABLE;
+      break;
+
    case GL_POINT_SMOOTH:
       break;
 
@@ -1108,6 +1137,7 @@ i915InitStateFunctions(struct dd_function_table *functions)
    functions->LineWidth = i915LineWidth;
    functions->LogicOpcode = i915LogicOp;
    functions->PointSize = i915PointSize;
+   functions->PointParameterfv = i915PointParameterfv;
    functions->PolygonStipple = i915PolygonStipple;
    functions->Scissor = i915Scissor;
    functions->ShadeModel = i915ShadeModel;
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index 32d4b30cf9..3ee4c8653a 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -27,7 +27,8 @@
 
 #include "main/mtypes.h"
 #include "main/enums.h"
-#include "main/texformat.h"
+#include "main/macros.h"
+#include "main/colormac.h"
 
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
@@ -37,7 +38,7 @@
 
 
 static GLuint
-translate_texture_format(GLuint mesa_format, GLuint internal_format,
+translate_texture_format(gl_format mesa_format, GLuint internal_format,
 			 GLenum DepthMode)
 {
    switch (mesa_format) {
@@ -56,10 +57,9 @@ translate_texture_format(GLuint mesa_format, GLuint internal_format,
    case MESA_FORMAT_ARGB4444:
       return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      if (internal_format == GL_RGB)
-	 return MAPSURF_32BIT | MT_32BIT_XRGB8888;
-      else
-	 return MAPSURF_32BIT | MT_32BIT_ARGB8888;
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
+   case MESA_FORMAT_XRGB8888:
+      return MAPSURF_32BIT | MT_32BIT_XRGB8888;
    case MESA_FORMAT_YCBCR_REV:
       return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
@@ -82,7 +82,12 @@ translate_texture_format(GLuint mesa_format, GLuint internal_format,
    case MESA_FORMAT_RGBA_DXT5:
       return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
    case MESA_FORMAT_S8_Z24:
-      return (MAPSURF_32BIT | MT_32BIT_xI824);
+      if (DepthMode == GL_ALPHA)
+	 return (MAPSURF_32BIT | MT_32BIT_x8A24);
+      else if (DepthMode == GL_INTENSITY)
+	 return (MAPSURF_32BIT | MT_32BIT_x8I24);
+      else
+	 return (MAPSURF_32BIT | MT_32BIT_x8L24);
    default:
       fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
       abort();
@@ -134,6 +139,7 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
    GLuint *state = i915->state.Tex[unit], format, pitch;
    GLint lodbias, aniso = 0;
    GLubyte border[4];
+   GLfloat maxlod;
 
    memset(state, 0, sizeof(state));
 
@@ -173,11 +179,9 @@ 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] = 0; /* Always the origin of the miptree */
 
-      format = translate_texture_format(firstImage->TexFormat->MesaFormat, 
+      format = translate_texture_format(firstImage->TexFormat,
 					firstImage->InternalFormat,
 					tObj->DepthMode);
       pitch = intelObj->mt->pitch * intelObj->mt->cpp;
@@ -193,11 +197,16 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 	 state[I915_TEXREG_MS3] |= MS3_TILE_WALK;
    }
 
+   /* We get one field with fraction bits for the maximum addressable
+    * (lowest resolution) LOD.  Use it to cover both MAX_LEVEL and
+    * MAX_LOD.
+    */
+   maxlod = MIN2(tObj->MaxLod, tObj->_MaxLevel - tObj->BaseLevel);
    state[I915_TEXREG_MS4] =
-     ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | MS4_CUBE_FACE_ENA_MASK |
-       ((((intelObj->lastLevel - intelObj->firstLevel) * 4)) <<
-	MS4_MAX_LOD_SHIFT) | ((firstImage->Depth - 1) <<
-			      MS4_VOLUME_DEPTH_SHIFT));
+      ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) |
+       MS4_CUBE_FACE_ENA_MASK |
+       (U_FIXED(CLAMP(maxlod, 0.0, 11.0), 2) << MS4_MAX_LOD_SHIFT) |
+       ((firstImage->Depth - 1) << MS4_VOLUME_DEPTH_SHIFT));
 
 
    {
@@ -263,8 +272,8 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 
       /* YUV conversion:
        */
-      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
-          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
+      if (firstImage->TexFormat == MESA_FORMAT_YCBCR ||
+          firstImage->TexFormat == MESA_FORMAT_YCBCR_REV)
          state[I915_TEXREG_SS2] |= SS2_COLORSPACE_CONVERSION;
 
       /* Shadow:
@@ -293,6 +302,12 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
       GLenum wt = tObj->WrapT;
       GLenum wr = tObj->WrapR;
 
+      /* We program 1D textures as 2D textures, so the 2D texcoord could
+       * result in sampling border values if we don't set the T wrap to
+       * repeat.
+       */
+      if (tObj->Target == GL_TEXTURE_1D)
+	 wt = GL_REPEAT;
 
       /* 3D textures don't seem to respect the border color.
        * Fallback if there's ever a danger that they might refer to
@@ -327,28 +342,31 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
           (translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT));
 
       state[I915_TEXREG_SS3] |= (unit << SS3_TEXTUREMAP_INDEX_SHIFT);
+      state[I915_TEXREG_SS3] |= (U_FIXED(CLAMP(tObj->MinLod, 0.0, 11.0), 4) <<
+				 SS3_MIN_LOD_SHIFT);
+
    }
 
    /* convert border color from float to ubyte */
-   CLAMPED_FLOAT_TO_UBYTE(border[0], tObj->BorderColor[0]);
-   CLAMPED_FLOAT_TO_UBYTE(border[1], tObj->BorderColor[1]);
-   CLAMPED_FLOAT_TO_UBYTE(border[2], tObj->BorderColor[2]);
-   CLAMPED_FLOAT_TO_UBYTE(border[3], tObj->BorderColor[3]);
+   CLAMPED_FLOAT_TO_UBYTE(border[0], tObj->BorderColor.f[0]);
+   CLAMPED_FLOAT_TO_UBYTE(border[1], tObj->BorderColor.f[1]);
+   CLAMPED_FLOAT_TO_UBYTE(border[2], tObj->BorderColor.f[2]);
+   CLAMPED_FLOAT_TO_UBYTE(border[3], tObj->BorderColor.f[3]);
 
    if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
       /* GL specs that border color for depth textures is taken from the
        * R channel, while the hardware uses A.  Spam R into all the channels
        * for safety.
        */
-      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(border[0],
-						   border[0],
-						   border[0],
-						   border[0]);
+      state[I915_TEXREG_SS4] = PACK_COLOR_8888(border[0],
+					       border[0],
+					       border[0],
+					       border[0]);
    } else {
-      state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(border[0],
-						   border[1],
-						   border[2],
-						   border[3]);
+      state[I915_TEXREG_SS4] = PACK_COLOR_8888(border[3],
+					       border[0],
+					       border[1],
+					       border[2]);
    }
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index 9e2523932f..266e6848c3 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -32,7 +32,6 @@
 #include "main/imports.h"
 #include "main/macros.h"
 #include "main/colormac.h"
-#include "main/texformat.h"
 
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
@@ -175,7 +174,7 @@ i915_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH(17, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(17);
 
    OUT_BATCH(_3DSTATE_AA_CMD |
              AA_LINE_ECAAR_WIDTH_ENABLE |
@@ -221,7 +220,7 @@ i915_emit_invarient_state(struct intel_context *intel)
 
 
 #define emit(intel, state, size )		     \
-   intel_batchbuffer_data(intel->batch, state, size, IGNORE_CLIPRECTS )
+   intel_batchbuffer_data(intel->batch, state, size)
 
 static GLuint
 get_dirty(struct i915_hw_state *state)
@@ -302,13 +301,9 @@ i915_emit_state(struct intel_context *intel)
     * It might be better to talk about explicit places where
     * scheduling is allowed, rather than assume that it is whenever a
     * batchbuffer fills up.
-    *
-    * Set the space as LOOP_CLIPRECTS now, since that's what our primitives
-    * will be emitted under.
     */
    intel_batchbuffer_require_space(intel->batch,
-				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE,
-				   LOOP_CLIPRECTS);
+				   get_state_size(state) + INTEL_PRIM_EMIT_SIZE);
    count = 0;
  again:
    aper_count = 0;
@@ -374,7 +369,7 @@ i915_emit_state(struct intel_context *intel)
    }
 
    if (dirty & I915_UPLOAD_BUFFERS) {
-      GLuint count = 9;
+      GLuint count = 15;
 
       if (INTEL_DEBUG & DEBUG_STATE)
          fprintf(stderr, "I915_UPLOAD_BUFFERS:\n");
@@ -382,10 +377,7 @@ i915_emit_state(struct intel_context *intel)
       if (state->depth_region)
           count += 3;
 
-      if (intel->constant_cliprect)
-          count += 6;
-
-      BEGIN_BATCH(count, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(count);
       OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR0]);
       OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR1]);
       OUT_RELOC(state->draw_region->buffer,
@@ -407,15 +399,13 @@ i915_emit_state(struct intel_context *intel)
       OUT_BATCH(state->Buffer[I915_DESTREG_SR1]);
       OUT_BATCH(state->Buffer[I915_DESTREG_SR2]);
 
-      if (intel->constant_cliprect) {
-	 assert(state->Buffer[I915_DESTREG_DRAWRECT0] != MI_NOOP);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT0]);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT1]);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT2]);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT3]);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT4]);
-	 OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT5]);
-      }
+      assert(state->Buffer[I915_DESTREG_DRAWRECT0] != MI_NOOP);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT1]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT2]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT3]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT4]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DRAWRECT5]);
 
       ADVANCE_BATCH();
    }
@@ -442,7 +432,7 @@ i915_emit_state(struct intel_context *intel)
          if (dirty & I915_UPLOAD_TEX(i))
             nr++;
 
-      BEGIN_BATCH(2 + nr * 3, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(2 + nr * 3);
       OUT_BATCH(_3DSTATE_MAP_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
       for (i = 0; i < I915_TEX_UNITS; i++)
@@ -466,7 +456,7 @@ i915_emit_state(struct intel_context *intel)
          }
       ADVANCE_BATCH();
 
-      BEGIN_BATCH(2 + nr * 3, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(2 + nr * 3);
       OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
       for (i = 0; i < I915_TEX_UNITS; i++)
@@ -588,8 +578,9 @@ i915_state_draw_region(struct intel_context *intel,
             DSTORG_VERT_BIAS(0x8) |     /* .5 */
             LOD_PRECLAMP_OGL | TEX_DEFAULT_COLOR_OGL);
    if (irb != NULL) {
-      switch (irb->texformat->MesaFormat) {
+      switch (irb->Base.Format) {
       case MESA_FORMAT_ARGB8888:
+      case MESA_FORMAT_XRGB8888:
 	 value |= DV_PF_8888;
 	 break;
       case MESA_FORMAT_RGB565:
@@ -603,7 +594,7 @@ i915_state_draw_region(struct intel_context *intel,
 	 break;
       default:
 	 _mesa_problem(ctx, "Bad renderbuffer format: %d\n",
-		       irb->texformat->MesaFormat);
+		       irb->Base.Format);
       }
    }
 
@@ -611,7 +602,7 @@ i915_state_draw_region(struct intel_context *intel,
     * the value of this bit, the pipeline needs to be MI_FLUSHed.  And it
     * can only be set when a depth buffer is already defined.
     */
-   if (IS_945(intel->intelScreen->deviceID) && intel->use_early_z &&
+   if (intel->is_945 && intel->use_early_z &&
        depth_region->tiling != I915_TILING_NONE)
       value |= CLASSIC_EARLY_DEPTH;
 
@@ -623,23 +614,14 @@ i915_state_draw_region(struct intel_context *intel,
    }
    state->Buffer[I915_DESTREG_DV1] = value;
 
-   if (intel->constant_cliprect) {
-      state->Buffer[I915_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
-      state->Buffer[I915_DESTREG_DRAWRECT1] = 0;
-      state->Buffer[I915_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
-      state->Buffer[I915_DESTREG_DRAWRECT3] =
-	 (ctx->DrawBuffer->Width & 0xffff) |
-	 (ctx->DrawBuffer->Height << 16);
-      state->Buffer[I915_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
-      state->Buffer[I915_DESTREG_DRAWRECT5] = 0;
-   } else {
-      state->Buffer[I915_DESTREG_DRAWRECT0] = MI_NOOP;
-      state->Buffer[I915_DESTREG_DRAWRECT1] = MI_NOOP;
-      state->Buffer[I915_DESTREG_DRAWRECT2] = MI_NOOP;
-      state->Buffer[I915_DESTREG_DRAWRECT3] = MI_NOOP;
-      state->Buffer[I915_DESTREG_DRAWRECT4] = MI_NOOP;
-      state->Buffer[I915_DESTREG_DRAWRECT5] = MI_NOOP;
-   }
+   state->Buffer[I915_DESTREG_DRAWRECT0] = _3DSTATE_DRAWRECT_INFO;
+   state->Buffer[I915_DESTREG_DRAWRECT1] = 0;
+   state->Buffer[I915_DESTREG_DRAWRECT2] = 0; /* xmin, ymin */
+   state->Buffer[I915_DESTREG_DRAWRECT3] =
+      (ctx->DrawBuffer->Width & 0xffff) |
+      (ctx->DrawBuffer->Height << 16);
+   state->Buffer[I915_DESTREG_DRAWRECT4] = 0; /* xoff, yoff */
+   state->Buffer[I915_DESTREG_DRAWRECT5] = 0;
 
    I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
 }
@@ -667,15 +649,6 @@ i915_new_batch(struct intel_context *intel)
     * difficulties associated with them (physical address requirements).
     */
    i915->state.emitted = 0;
-
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!intel->no_batch_wrap);
-}
-
-static GLuint
-i915_flush_cmd(void)
-{
-   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
 static void 
@@ -699,7 +672,6 @@ i915InitVtbl(struct i915_context *i915)
    i915->intel.vtbl.render_prevalidate = i915_render_prevalidate;
    i915->intel.vtbl.set_draw_region = i915_set_draw_region;
    i915->intel.vtbl.update_texture_state = i915UpdateTextureState;
-   i915->intel.vtbl.flush_cmd = i915_flush_cmd;
    i915->intel.vtbl.assert_not_dirty = i915_assert_not_dirty;
    i915->intel.vtbl.finish_batch = intel_finish_vb;
 }
diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index 410052b3c2..ec209391ab 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -117,7 +117,7 @@ intelDmaPrimitive(struct intel_context *intel, GLenum prim)
    intel_set_prim(intel, hw_prim[prim]);
 }
 
-static inline GLuint intel_get_vb_max(struct intel_context *intel)
+static INLINE GLuint intel_get_vb_max(struct intel_context *intel)
 {
    GLuint ret;
 
@@ -129,7 +129,7 @@ static inline GLuint intel_get_vb_max(struct intel_context *intel)
    return ret;
 }
 
-static inline GLuint intel_get_current_max(struct intel_context *intel)
+static INLINE GLuint intel_get_current_max(struct intel_context *intel)
 {
 
    if (intel->intelScreen->no_vbo)
diff --git a/src/mesa/drivers/dri/i915/intel_swapbuffers.c b/src/mesa/drivers/dri/i915/intel_swapbuffers.c
deleted file mode 120000
index 148d5215aa..0000000000
--- a/src/mesa/drivers/dri/i915/intel_swapbuffers.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_swapbuffers.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 0641e6df9d..e99baf8e0e 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -89,7 +89,6 @@ intel_flush_inline_primitive(struct intel_context *intel)
 
 static void intel_start_inline(struct intel_context *intel, uint32_t prim)
 {
-   uint32_t batch_flags = LOOP_CLIPRECTS;
    BATCH_LOCALS;
 
    intel->vtbl.emit_state(intel);
@@ -101,7 +100,7 @@ static void intel_start_inline(struct intel_context *intel, uint32_t prim)
    /* Emit a slot which will be filled with the inline primitive
     * command later.
     */
-   BEGIN_BATCH(2, batch_flags);
+   BEGIN_BATCH(2);
    OUT_BATCH(0);
 
    assert((intel->batch->dirty_state & (1<<1)) == 0);
@@ -221,7 +220,7 @@ void intel_flush_prim(struct intel_context *intel)
    intel->prim.count = 0;
    offset = intel->prim.start_offset;
    intel->prim.start_offset = intel->prim.current_offset;
-   if (!IS_9XX(intel->intelScreen->deviceID))
+   if (!intel->gen >= 3)
       intel->prim.start_offset = ALIGN(intel->prim.start_offset, 128);
    intel->prim.flush = NULL;
 
@@ -251,8 +250,8 @@ void intel_flush_prim(struct intel_context *intel)
 	  intel->vertex_size * 4);
 #endif
 
-   if (IS_9XX(intel->intelScreen->deviceID)) {
-      BEGIN_BATCH(5, LOOP_CLIPRECTS);
+   if (intel->gen >= 3) {
+      BEGIN_BATCH(5);
       OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
 		I1_LOAD_S(0) | I1_LOAD_S(1) | 1);
       assert((offset & !S0_VB_OFFSET_MASK) == 0);
@@ -270,7 +269,7 @@ void intel_flush_prim(struct intel_context *intel)
    } else {
       struct i830_context *i830 = i830_context(&intel->ctx);
 
-      BEGIN_BATCH(5, LOOP_CLIPRECTS);
+      BEGIN_BATCH(5);
       OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
 		I1_LOAD_S(0) | I1_LOAD_S(2) | 1);
       /* S0 */
@@ -1088,6 +1087,7 @@ intelRenderStart(GLcontext * ctx)
 {
    struct intel_context *intel = intel_context(ctx);
 
+   intel_check_front_buffer_rendering(intel);
    intel->vtbl.render_start(intel_context(ctx));
    intel->vtbl.emit_state(intel);
 }
@@ -1196,12 +1196,16 @@ getFallbackString(GLuint bit)
 
 
 
+/**
+ * Enable/disable a fallback flag.
+ * \param bit  one of INTEL_FALLBACK_x flags.
+ */
 void
-intelFallback(struct intel_context *intel, GLuint bit, GLboolean mode)
+intelFallback(struct intel_context *intel, GLbitfield bit, GLboolean mode)
 {
    GLcontext *ctx = &intel->ctx;
    TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint oldfallback = intel->Fallback;
+   const GLbitfield oldfallback = intel->Fallback;
 
    if (mode) {
       intel->Fallback |= bit;
@@ -1245,81 +1249,6 @@ union fi
    GLint i;
 };
 
-
-/**********************************************************************/
-/*             Used only with the metaops callbacks.                  */
-/**********************************************************************/
-static void
-intel_meta_draw_poly(struct intel_context *intel,
-                     GLuint n,
-                     GLfloat xy[][2],
-                     GLfloat z, GLuint color, GLfloat tex[][2])
-{
-   union fi *vb;
-   GLint i;
-   unsigned int saved_vertex_size = intel->vertex_size;
-
-   LOCK_HARDWARE(intel);
-
-   intel->vertex_size = 6;
-
-   /* All 3d primitives should be emitted with LOOP_CLIPRECTS,
-    * otherwise the drawing origin (DR4) might not be set correctly.
-    */
-   intel_set_prim(intel, PRIM3D_TRIFAN);
-   vb = (union fi *) intel_get_prim_space(intel, n);
-
-   for (i = 0; i < n; i++) {
-      vb[0].f = xy[i][0];
-      vb[1].f = xy[i][1];
-      vb[2].f = z;
-      vb[3].i = color;
-      vb[4].f = tex[i][0];
-      vb[5].f = tex[i][1];
-      vb += 6;
-   }
-
-   INTEL_FIREVERTICES(intel);
-
-   intel->vertex_size = saved_vertex_size;
-
-   UNLOCK_HARDWARE(intel);
-}
-
-static void
-intel_meta_draw_quad(struct intel_context *intel,
-                     GLfloat x0, GLfloat x1,
-                     GLfloat y0, GLfloat y1,
-                     GLfloat z,
-                     GLuint color,
-                     GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1)
-{
-   GLfloat xy[4][2];
-   GLfloat tex[4][2];
-
-   xy[0][0] = x0;
-   xy[0][1] = y0;
-   xy[1][0] = x1;
-   xy[1][1] = y0;
-   xy[2][0] = x1;
-   xy[2][1] = y1;
-   xy[3][0] = x0;
-   xy[3][1] = y1;
-
-   tex[0][0] = s0;
-   tex[0][1] = t0;
-   tex[1][0] = s1;
-   tex[1][1] = t0;
-   tex[2][0] = s1;
-   tex[2][1] = t1;
-   tex[3][0] = s0;
-   tex[3][1] = t1;
-
-   intel_meta_draw_poly(intel, 4, xy, z, color, tex);
-}
-
-
-
 /**********************************************************************/
 /*                            Initialization.                         */
 /**********************************************************************/
@@ -1328,7 +1257,6 @@ intel_meta_draw_quad(struct intel_context *intel,
 void
 intelInitTriFuncs(GLcontext * ctx)
 {
-   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    static int firsttime = 1;
 
@@ -1345,6 +1273,4 @@ intelInitTriFuncs(GLcontext * ctx)
    tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
    tnl->Driver.Render.CopyPV = _tnl_copy_pv;
    tnl->Driver.Render.Interp = _tnl_interp;
-
-   intel->vtbl.meta_draw_quad = intel_meta_draw_quad;
 }
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 57dcc91586..7758a792fd 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -24,7 +24,6 @@ DRIVER_SOURCES = \
 	intel_pixel_draw.c \
 	intel_pixel_read.c \
 	intel_state.c \
-	intel_swapbuffers.c \
 	intel_syncobj.c \
 	intel_tex.c \
 	intel_tex_copy.c \
@@ -96,10 +95,10 @@ ASM_SOURCES =
 
 DRIVER_DEFINES = -I../intel -I../intel/server
 
-DRI_LIB_DEPS += -ldrm_intel
+INCLUDES += $(INTEL_CFLAGS)
+DRI_LIB_DEPS += $(INTEL_LIBS)
 
 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 1088a7a607..bac1c3a49c 100644
--- a/src/mesa/drivers/dri/i965/brw_cc.c
+++ b/src/mesa/drivers/dri/i965/brw_cc.c
@@ -34,6 +34,7 @@
 #include "brw_state.h"
 #include "brw_defines.h"
 #include "brw_util.h"
+#include "intel_fbo.h"
 #include "main/macros.h"
 #include "main/enums.h"
 
@@ -44,17 +45,24 @@ static void prepare_cc_vp( struct brw_context *brw )
 
    memset(&ccv, 0, sizeof(ccv));
 
-   /* _NEW_VIEWPORT */
-   ccv.min_depth = ctx->Viewport.Near;
-   ccv.max_depth = ctx->Viewport.Far;
+   /* _NEW_TRANSOFORM */
+   if (ctx->Transform.DepthClamp) {
+      /* _NEW_VIEWPORT */
+      ccv.min_depth = MIN2(ctx->Viewport.Near, ctx->Viewport.Far);
+      ccv.max_depth = MAX2(ctx->Viewport.Near, ctx->Viewport.Far);
+   } else {
+      ccv.min_depth = 0.0;
+      ccv.max_depth = 1.0;
+   }
 
    dri_bo_unreference(brw->cc.vp_bo);
-   brw->cc.vp_bo = brw_cache_data( &brw->cache, BRW_CC_VP, &ccv, NULL, 0 );
+   brw->cc.vp_bo = brw_cache_data(&brw->cache, BRW_CC_VP, &ccv, sizeof(ccv),
+				  NULL, 0);
 }
 
 const struct brw_tracked_state brw_cc_vp = {
    .dirty = {
-      .mesa = _NEW_VIEWPORT,
+      .mesa = _NEW_VIEWPORT | _NEW_TRANSFORM,
       .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
@@ -82,6 +90,28 @@ struct brw_cc_unit_key {
    GLenum depth_func;
 };
 
+/**
+ * Modify blend function to force destination alpha to 1.0
+ *
+ * If \c function specifies a blend function that uses destination alpha,
+ * replace it with a function that hard-wires destination alpha to 1.0.  This
+ * is used when rendering to xRGB targets.
+ */
+static GLenum
+fix_xRGB_alpha(GLenum function)
+{
+   switch (function) {
+   case GL_DST_ALPHA:
+      return GL_ONE;
+
+   case GL_ONE_MINUS_DST_ALPHA:
+   case GL_SRC_ALPHA_SATURATE:
+      return GL_ZERO;
+   }
+
+   return function;
+}
+
 static void
 cc_unit_populate_key(struct brw_context *brw, struct brw_cc_unit_key *key)
 {
@@ -125,6 +155,17 @@ cc_unit_populate_key(struct brw_context *brw, struct brw_cc_unit_key *key)
       key->blend_dst_rgb = ctx->Color.BlendDstRGB;
       key->blend_src_a = ctx->Color.BlendSrcA;
       key->blend_dst_a = ctx->Color.BlendDstA;
+
+      /* If the renderbuffer is XRGB, we have to frob the blend function to
+       * force the destination alpha to 1.0.  This means replacing GL_DST_ALPHA
+       * with GL_ONE and GL_ONE_MINUS_DST_ALPHA with GL_ZERO.
+       */
+      if (ctx->DrawBuffer->Visual.alphaBits == 0) {
+	 key->blend_src_rgb = fix_xRGB_alpha(key->blend_src_rgb);
+	 key->blend_src_a   = fix_xRGB_alpha(key->blend_src_a);
+	 key->blend_dst_rgb = fix_xRGB_alpha(key->blend_dst_rgb);
+	 key->blend_dst_a   = fix_xRGB_alpha(key->blend_dst_a);
+      }
    }
 
    key->alpha_enabled = ctx->Color.AlphaEnabled;
diff --git a/src/mesa/drivers/dri/i965/brw_clip.c b/src/mesa/drivers/dri/i965/brw_clip.c
index 20a927cf38..af1d975de9 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.c
+++ b/src/mesa/drivers/dri/i965/brw_clip.c
@@ -50,6 +50,7 @@
 static void compile_clip_prog( struct brw_context *brw,
 			     struct brw_clip_prog_key *key )
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_clip_compile c;
    const GLuint *program;
    GLuint program_size;
@@ -65,27 +66,26 @@ static void compile_clip_prog( struct brw_context *brw,
    c.func.single_program_flow = 1;
 
    c.key = *key;
-   c.need_ff_sync = BRW_IS_IGDNG(brw);
 
    /* Need to locate the two positions present in vertex + header.
     * These are currently hardcoded:
     */
    c.header_position_offset = ATTR_SIZE;
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
        delta = 3 * REG_SIZE;
    else
        delta = REG_SIZE;
 
    for (i = 0; i < VERT_RESULT_MAX; i++)
-      if (c.key.attrs & (1<<i)) {
+      if (c.key.attrs & BITFIELD64_BIT(i)) {
 	 c.offset[i] = delta;
 	 delta += ATTR_SIZE;
       }
 
    c.nr_attrs = brw_count_bits(c.key.attrs);
    
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
        c.nr_regs = (c.nr_attrs + 1) / 2 + 3;  /* are vertices packed, or reg-aligned? */
    else
        c.nr_regs = (c.nr_attrs + 1) / 2 + 1;  /* are vertices packed, or reg-aligned? */
@@ -143,7 +143,8 @@ static void compile_clip_prog( struct brw_context *brw,
  */
 static void upload_clip_prog(struct brw_context *brw)
 {
-   GLcontext *ctx = &brw->intel.ctx;
+   struct intel_context *intel = &brw->intel;
+   GLcontext *ctx = &intel->ctx;
    struct brw_clip_prog_key key;
 
    memset(&key, 0, sizeof(key));
@@ -156,10 +157,11 @@ static void upload_clip_prog(struct brw_context *brw)
    key.attrs = brw->vs.prog_data->outputs_written;
    /* _NEW_LIGHT */
    key.do_flat_shading = (ctx->Light.ShadeModel == GL_FLAT);
+   key.pv_first = (ctx->Light.ProvokingVertex == GL_FIRST_VERTEX_CONVENTION);
    /* _NEW_TRANSFORM */
    key.nr_userclip = brw_count_bits(ctx->Transform.ClipPlanesEnabled);
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
        key.clip_mode = BRW_CLIPMODE_KERNEL_CLIP;
    else
        key.clip_mode = BRW_CLIPMODE_NORMAL;
diff --git a/src/mesa/drivers/dri/i965/brw_clip.h b/src/mesa/drivers/dri/i965/brw_clip.h
index 957df441ab..d71bac7f61 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.h
+++ b/src/mesa/drivers/dri/i965/brw_clip.h
@@ -42,22 +42,21 @@
  * up polygon offset and flatshading at this point:
  */
 struct brw_clip_prog_key {
-   GLuint attrs:32;		
+   GLbitfield64 attrs;
    GLuint primitive:4;
    GLuint nr_userclip:3;
    GLuint do_flat_shading:1;
+   GLuint pv_first:1;
    GLuint do_unfilled:1;
    GLuint fill_cw:2;		/* includes cull information */
    GLuint fill_ccw:2;		/* includes cull information */
    GLuint offset_cw:1;
    GLuint offset_ccw:1;
-   GLuint pad0:17;
-
    GLuint copy_bfc_cw:1;
    GLuint copy_bfc_ccw:1;
    GLuint clip_mode:3;
-   GLuint pad1:27;
-   
+   GLuint pad0:11;
+
    GLfloat offset_factor;
    GLfloat offset_units;
 };
@@ -119,7 +118,6 @@ struct brw_clip_compile {
 
    GLuint header_position_offset;
    GLuint offset[VERT_ATTRIB_MAX];
-   GLboolean need_ff_sync;
 };
 
 #define ATTR_SIZE  (4*4)
diff --git a/src/mesa/drivers/dri/i965/brw_clip_line.c b/src/mesa/drivers/dri/i965/brw_clip_line.c
index 048ca620fa..afc0b11049 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_line.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_line.c
@@ -46,6 +46,7 @@
 
 static void brw_clip_line_alloc_regs( struct brw_clip_compile *c )
 {
+   struct intel_context *intel = &c->func.brw->intel;
    GLuint i = 0,j;
 
    /* Register usage is static, precompute here:
@@ -85,7 +86,7 @@ static void brw_clip_line_alloc_regs( struct brw_clip_compile *c )
       i++;
    }
 
-   if (c->need_ff_sync) {
+   if (intel->needs_ff_sync) {
       c->reg.ff_sync = retype(brw_vec1_grf(i, 0), BRW_REGISTER_TYPE_UD);
       i++;
    }
@@ -126,6 +127,7 @@ static void brw_clip_line_alloc_regs( struct brw_clip_compile *c )
 static void clip_and_emit_line( struct brw_clip_compile *c )
 {
    struct brw_compile *p = &c->func;
+   struct brw_context *brw = p->brw;
    struct brw_indirect vtx0     = brw_indirect(0, 0);
    struct brw_indirect vtx1      = brw_indirect(1, 0);
    struct brw_indirect newvtx0   = brw_indirect(2, 0);
@@ -152,7 +154,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
    brw_clip_init_clipmask(c);
 
    /* -ve rhw workaround */
-   if (BRW_IS_965(p->brw)) {
+   if (brw->has_negative_rhw_bug) {
       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));
@@ -189,7 +191,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
               * Both can be negative on GM965/G965 due to RHW workaround
               * if so, this object should be rejected.
               */
-             if (BRW_IS_965(p->brw)) {
+             if (brw->has_negative_rhw_bug) {
                  brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_LE, c->reg.dp0, brw_imm_f(0.0));
                  is_neg2 = brw_IF(p, BRW_EXECUTE_1);
                  {
@@ -214,7 +216,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
 
              /* If both are positive, do nothing */
              /* Only on GM965/G965 */
-             if (BRW_IS_965(p->brw)) {
+             if (brw->has_negative_rhw_bug) {
                  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);
              }
@@ -229,7 +231,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
                  brw_set_predicate_control(p, BRW_PREDICATE_NONE);
              }
 
-             if (BRW_IS_965(p->brw)) {
+             if (brw->has_negative_rhw_bug) {
                  brw_ENDIF(p, is_neg2);
              }
          }
@@ -269,8 +271,12 @@ void brw_emit_line_clip( struct brw_clip_compile *c )
    brw_clip_line_alloc_regs(c);
    brw_clip_init_ff_sync(c);
 
-   if (c->key.do_flat_shading)
-      brw_clip_copy_colors(c, 0, 1);
+   if (c->key.do_flat_shading) {
+      if (c->key.pv_first)
+         brw_clip_copy_colors(c, 1, 0);
+      else
+         brw_clip_copy_colors(c, 0, 1);
+   }
                 
    clip_and_emit_line(c);
 }
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index 234b3744bf..c8f24a94e4 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -74,6 +74,7 @@ static dri_bo *
 clip_unit_create_from_key(struct brw_context *brw,
 			  struct brw_clip_unit_key *key)
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_clip_unit_state clip;
    dri_bo *bo;
 
@@ -105,7 +106,7 @@ clip_unit_create_from_key(struct brw_context *brw,
       /* Although up to 16 concurrent Clip threads are allowed on IGDNG, 
        * only 2 threads can output VUEs at a time.
        */
-      if (BRW_IS_IGDNG(brw))
+      if (intel->is_ironlake)
          clip.thread4.max_threads = 16 - 1;        
       else
          clip.thread4.max_threads = 2 - 1;
@@ -130,7 +131,7 @@ clip_unit_create_from_key(struct brw_context *brw,
    clip.clip5.api_mode = BRW_CLIP_API_OGL;
    clip.clip5.clip_mode = key->clip_mode;
 
-   if (BRW_IS_G4X(brw))
+   if (intel->is_g4x)
       clip.clip5.negative_w_clip_test = 1;
 
    clip.clip6.clipper_viewport_state_ptr = 0;
diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c b/src/mesa/drivers/dri/i965/brw_clip_tri.c
index 0efd77225e..cfbb8f2686 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_tri.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c
@@ -51,6 +51,7 @@ static void release_tmps( struct brw_clip_compile *c )
 void brw_clip_tri_alloc_regs( struct brw_clip_compile *c, 
 			      GLuint nr_verts )
 {
+   struct intel_context *intel = &c->func.brw->intel;
    GLuint i = 0,j;
 
    /* Register usage is static, precompute here:
@@ -78,7 +79,7 @@ void brw_clip_tri_alloc_regs( struct brw_clip_compile *c,
       for (j = 0; j < 3; j++) {
 	 GLuint delta = c->nr_attrs*16 + 32;
 
-         if (BRW_IS_IGDNG(c->func.brw))
+         if (intel->is_ironlake)
              delta = c->nr_attrs * 16 + 32 * 3;
 
 	 brw_MOV(&c->func, byte_offset(c->reg.vertex[j], delta), brw_imm_f(0));
@@ -119,7 +120,7 @@ void brw_clip_tri_alloc_regs( struct brw_clip_compile *c,
       i++;
    }
 
-   if (c->need_ff_sync) {
+   if (intel->needs_ff_sync) {
       c->reg.ff_sync = retype(brw_vec1_grf(i, 0), BRW_REGISTER_TYPE_UD);
       i++;
    }
@@ -188,14 +189,20 @@ void brw_clip_tri_flat_shade( struct brw_clip_compile *c )
 	   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);
+      if (c->key.pv_first) {
+         brw_clip_copy_colors(c, 1, 0);
+         brw_clip_copy_colors(c, 2, 0);
+      }
+      else {
+         brw_clip_copy_colors(c, 0, 2);
+         brw_clip_copy_colors(c, 1, 2);
+      }
    }
    brw_ENDIF(p, is_poly);
 }
@@ -565,6 +572,7 @@ void brw_emit_tri_clip( struct brw_clip_compile *c )
 {
    struct brw_instruction *neg_rhw;
    struct brw_compile *p = &c->func;
+   struct brw_context *brw = p->brw;
    brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6);
    brw_clip_tri_init_vertices(c);
    brw_clip_init_clipmask(c);
@@ -572,7 +580,7 @@ void brw_emit_tri_clip( struct brw_clip_compile *c )
 
    /* if -ve rhw workaround bit is set, 
       do cliptest */
-   if (BRW_IS_965(p->brw)) {
+   if (brw->has_negative_rhw_bug) {
       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));
diff --git a/src/mesa/drivers/dri/i965/brw_clip_util.c b/src/mesa/drivers/dri/i965/brw_clip_util.c
index 5a73abdfee..86fed59fa4 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_util.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_util.c
@@ -135,6 +135,7 @@ void brw_clip_interp_vertex( struct brw_clip_compile *c,
 			     GLboolean force_edgeflag)
 {
    struct brw_compile *p = &c->func;
+   struct intel_context *intel = &p->brw->intel;
    struct brw_reg tmp = get_tmp(c);
    GLuint i;
 
@@ -142,7 +143,7 @@ void brw_clip_interp_vertex( struct brw_clip_compile *c,
     */
    /*
     * After CLIP stage, only first 256 bits of the VUE are read
-    * back on IGDNG, so needn't change it
+    * back on Ironlake, so needn't change it
     */
    brw_copy_indirect_to_indirect(p, dest_ptr, v0_ptr, 1);
       
@@ -151,7 +152,7 @@ void brw_clip_interp_vertex( struct brw_clip_compile *c,
    for (i = 0; i < c->nr_attrs; i++) {
       GLuint delta = i*16 + 32;
 
-      if (BRW_IS_IGDNG(p->brw))
+      if (intel->is_ironlake)
           delta = i * 16 + 32 * 3;
 
       if (delta == c->offset[VERT_RESULT_EDGE]) {
@@ -185,7 +186,7 @@ void brw_clip_interp_vertex( struct brw_clip_compile *c,
    if (i & 1) {
       GLuint delta = i*16 + 32;
 
-      if (BRW_IS_IGDNG(p->brw))
+      if (intel->is_ironlake)
           delta = i * 16 + 32 * 3;
 
       brw_MOV(p, deref_4f(dest_ptr, delta), brw_imm_f(0));
@@ -359,7 +360,9 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c )
 
 void brw_clip_ff_sync(struct brw_clip_compile *c)
 {
-    if (c->need_ff_sync) {
+    struct intel_context *intel = &c->func.brw->intel;
+
+    if (intel->needs_ff_sync) {
         struct brw_compile *p = &c->func;
         struct brw_instruction *need_ff_sync;
 
@@ -388,7 +391,9 @@ void brw_clip_ff_sync(struct brw_clip_compile *c)
 
 void brw_clip_init_ff_sync(struct brw_clip_compile *c)
 {
-    if (c->need_ff_sync) {
+    struct intel_context *intel = &c->func.brw->intel;
+
+    if (intel->needs_ff_sync) {
 	struct brw_compile *p = &c->func;
         
         brw_MOV(p, c->reg.ff_sync, brw_imm_ud(0));
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index c300c33adc..7bb15956b5 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -77,7 +77,7 @@ static void brwInitDriverFunctions( struct dd_function_table *functions )
 }
 
 GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
+			    __DRIcontext *driContextPriv,
 			    void *sharedContextPrivate)
 {
    struct dd_function_table functions;
@@ -105,11 +105,15 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
 
+   ctx->Const.MaxDrawBuffers = BRW_MAX_DRAW_BUFFERS;
    ctx->Const.MaxTextureImageUnits = BRW_MAX_TEX_UNIT;
    ctx->Const.MaxTextureCoordUnits = 8; /* Mesa limit */
    ctx->Const.MaxTextureUnits = MIN2(ctx->Const.MaxTextureCoordUnits,
                                      ctx->Const.MaxTextureImageUnits);
    ctx->Const.MaxVertexTextureImageUnits = 0; /* no vertex shader textures */
+   ctx->Const.MaxCombinedTextureImageUnits =
+      ctx->Const.MaxVertexTextureImageUnits +
+      ctx->Const.MaxTextureImageUnits;
 
    /* Mesa limits textures to 4kx4k; it would be nice to fix that someday
     */
@@ -154,6 +158,38 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
       MIN2(ctx->Const.FragmentProgram.MaxNativeParameters,
 	   ctx->Const.FragmentProgram.MaxEnvParams);
 
+   if (intel->is_ironlake || intel->is_g4x) {
+      brw->CMD_VF_STATISTICS = CMD_VF_STATISTICS_GM45;
+      brw->CMD_PIPELINE_SELECT = CMD_PIPELINE_SELECT_GM45;
+      brw->has_surface_tile_offset = GL_TRUE;
+      brw->has_compr4 = GL_TRUE;
+      brw->has_aa_line_parameters = GL_TRUE;
+  } else {
+      brw->CMD_VF_STATISTICS = CMD_VF_STATISTICS_965;
+      brw->CMD_PIPELINE_SELECT = CMD_PIPELINE_SELECT_965;
+   }
+
+   /* WM maximum threads is number of EUs times number of threads per EU. */
+   if (intel->is_ironlake) {
+      brw->urb.size = 1024;
+      brw->vs_max_threads = 72;
+      brw->wm_max_threads = 12 * 6;
+   } else if (intel->is_g4x) {
+      brw->urb.size = 384;
+      brw->vs_max_threads = 32;
+      brw->wm_max_threads = 10 * 5;
+   } else {
+      brw->urb.size = 256;
+      brw->vs_max_threads = 16;
+      brw->wm_max_threads = 8 * 4;
+      brw->has_negative_rhw_bug = GL_TRUE;
+   }
+
+   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD) {
+      brw->vs_max_threads = 1;
+      brw->wm_max_threads = 1;
+   }
+
    brw_init_state( brw );
 
    brw->state.dirty.mesa = ~0;
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index fa3e32c7ff..0dd3087143 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -115,7 +115,7 @@
  * Handles blending and (presumably) depth and stencil testing.
  */
 
-#define BRW_FALLBACK_TEXTURE		 0x1
+
 #define BRW_MAX_CURBE                    (32*16)
 
 struct brw_context;
@@ -172,8 +172,8 @@ struct brw_fragment_program {
    GLuint id;  /**< serial no. to identify frag progs, never re-used */
    GLboolean isGLSL;  /**< really, any IF/LOOP/CONT/BREAK instructions */
 
-   dri_bo *const_buffer;    /** Program constant buffer/surface */
    GLboolean use_const_buffer;
+   dri_bo *const_buffer;    /** Program constant buffer/surface */
 
    /** for debugging, which texture units are referenced */
    GLbitfield tex_units_used;
@@ -231,7 +231,7 @@ struct brw_vs_prog_data {
    GLuint curb_read_length;
    GLuint urb_read_length;
    GLuint total_grf;
-   GLuint outputs_written;
+   GLbitfield64 outputs_written;
    GLuint nr_params;       /**< number of float params/constants */
 
    GLuint inputs_read;
@@ -252,20 +252,23 @@ struct brw_vs_ouput_sizes {
 /** Number of texture sampler units */
 #define BRW_MAX_TEX_UNIT 16
 
+/** Max number of render targets in a shader */
+#define BRW_MAX_DRAW_BUFFERS 4
+
 /**
  * Size of our surface binding table for the WM.
  * This contains pointers to the drawing surfaces and current texture
  * objects and shader constant buffers (+2).
  */
-#define BRW_WM_MAX_SURF (MAX_DRAW_BUFFERS + BRW_MAX_TEX_UNIT + 1)
+#define BRW_WM_MAX_SURF (BRW_MAX_DRAW_BUFFERS + BRW_MAX_TEX_UNIT + 1)
 
 /**
  * Helpers to convert drawing buffers, textures and constant buffers
  * to surface binding table indexes, for WM.
  */
 #define SURF_INDEX_DRAW(d)           (d)
-#define SURF_INDEX_FRAG_CONST_BUFFER (MAX_DRAW_BUFFERS) 
-#define SURF_INDEX_TEXTURE(t)        (MAX_DRAW_BUFFERS + 1 + (t))
+#define SURF_INDEX_FRAG_CONST_BUFFER (BRW_MAX_DRAW_BUFFERS) 
+#define SURF_INDEX_TEXTURE(t)        (BRW_MAX_DRAW_BUFFERS + 1 + (t))
 
 /**
  * Size of surface binding table for the VS.
@@ -317,7 +320,6 @@ struct brw_cache_item {
    GLuint nr_reloc_bufs;
 
    dri_bo *bo;
-   GLuint data_size;
 
    struct brw_cache_item *next;
 };   
@@ -330,7 +332,6 @@ struct brw_cache {
    struct brw_cache_item **items;
    GLuint size, n_items;
 
-   GLuint key_size[BRW_MAX_CACHE];		/* for fixed-size keys */
    GLuint aux_size[BRW_MAX_CACHE];
    char *name[BRW_MAX_CACHE];
 
@@ -410,23 +411,6 @@ struct brw_vertex_info {
    GLuint sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[VERT_ATTRIB_MAX] */
 };
 
-
-
-
-/* Cache for TNL programs.
- */
-struct brw_tnl_cache_item {
-   GLuint hash;
-   void *key;
-   void *data;
-   struct brw_tnl_cache_item *next;
-};
-
-struct brw_tnl_cache {
-   struct brw_tnl_cache_item **items;
-   GLuint size, n_items;
-};
-
 struct brw_query_object {
    struct gl_query_object Base;
 
@@ -454,9 +438,11 @@ struct brw_context
    GLuint primitive;
 
    GLboolean emit_state_always;
-   GLboolean tmp_fallback;
-   GLboolean no_batch_wrap;
-
+   GLboolean has_surface_tile_offset;
+   GLboolean has_compr4;
+   GLboolean has_negative_rhw_bug;
+   GLboolean has_aa_line_parameters;
+;
    struct {
       struct brw_state_flags dirty;
 
@@ -532,6 +518,12 @@ struct brw_context
     */
    GLuint next_free_page;
 
+   /* hw-dependent 3DSTATE_VF_STATISTICS opcode */
+   uint32_t CMD_VF_STATISTICS;
+   /* hw-dependent 3DSTATE_PIPELINE_SELECT opcode */
+   uint32_t CMD_PIPELINE_SELECT;
+   int vs_max_threads;
+   int wm_max_threads;
 
    /* BRW_NEW_URB_ALLOCATIONS:
     */
@@ -559,6 +551,7 @@ struct brw_context
       GLuint clip_start;
       GLuint sf_start;
       GLuint cs_start;
+      GLuint size; /* Hardware URB size, in KB. */
    } urb;
 
    
@@ -686,7 +679,7 @@ void brwInitVtbl( struct brw_context *brw );
  * brw_context.c
  */
 GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
+			    __DRIcontext *driContextPriv,
 			    void *sharedContextPrivate);
 
 /*======================================================================
@@ -759,9 +752,5 @@ brw_fragment_program_const(const struct gl_fragment_program *p)
    return (const struct brw_fragment_program *) p;
 }
 
-
-
-#define DO_SETUP_BITS ((1<<(FRAG_ATTRIB_MAX)) - 1)
-
 #endif
 
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index 4be6c77aa1..190310afbb 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -130,7 +130,7 @@ static void calculate_curbe_offsets( struct brw_context *brw )
 const struct brw_tracked_state brw_curbe_offsets = {
    .dirty = {
       .mesa = _NEW_TRANSFORM,
-      .brw  = BRW_NEW_VERTEX_PROGRAM,
+      .brw  = BRW_NEW_VERTEX_PROGRAM | BRW_NEW_CONTEXT,
       .cache = CACHE_NEW_WM_PROG
    },
    .prepare = calculate_curbe_offsets
@@ -340,7 +340,7 @@ static void emit_constant_buffer(struct brw_context *brw)
    struct intel_context *intel = &brw->intel;
    GLuint sz = brw->curbe.total_size;
 
-   BEGIN_BATCH(2, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(2);
    if (sz == 0) {
       OUT_BATCH((CMD_CONST_BUFFER << 16) | (2 - 2));
       OUT_BATCH(0);
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index 78d457ad2b..ea0d7e05d4 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -673,18 +673,10 @@
 #define BRW_SAMPLER_MESSAGE_SIMD8_LD                  3
 #define BRW_SAMPLER_MESSAGE_SIMD16_LD                 3
 
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_IGDNG            0
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_IGDNG          0
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_IGDNG           0
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_BIAS_IGDNG       1
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_BIAS_IGDNG     1
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS_IGDNG      1
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_LOD_IGDNG        2
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_LOD_IGDNG      2
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_LOD_IGDNG       2
-#define BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_COMPARE_IGDNG    3
-#define BRW_SAMPLER_MESSAGE_SIMD4X2_SAMPLE_COMPARE_IGDNG  3
-#define BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE_IGDNG   3
+#define BRW_SAMPLER_MESSAGE_SAMPLE_IGDNG            0
+#define BRW_SAMPLER_MESSAGE_SAMPLE_BIAS_IGDNG       1
+#define BRW_SAMPLER_MESSAGE_SAMPLE_LOD_IGDNG        2
+#define BRW_SAMPLER_MESSAGE_SAMPLE_COMPARE_IGDNG    3
 
 /* for IGDNG only */
 #define BRW_SAMPLER_SIMD_MODE_SIMD4X2                   0
@@ -840,12 +832,4 @@
 
 #include "intel_chipset.h"
 
-#define BRW_IS_G4X(brw)         (IS_G4X((brw)->intel.intelScreen->deviceID))
-#define BRW_IS_IGDNG(brw)         (IS_IGDNG((brw)->intel.intelScreen->deviceID))
-#define BRW_IS_965(brw)         (!(BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw)))
-#define CMD_PIPELINE_SELECT(brw)        ((BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw)) ? CMD_PIPELINE_SELECT_GM45 : CMD_PIPELINE_SELECT_965)
-#define CMD_VF_STATISTICS(brw)          ((BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw)) ? CMD_VF_STATISTICS_GM45 : CMD_VF_STATISTICS_965)
-#define URB_SIZES(brw)                  (BRW_IS_IGDNG(brw) ? 1024 : \
-                                         (BRW_IS_G4X(brw) ? 384 : 256))  /* 512 bit units */
-
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_disasm.c b/src/mesa/drivers/dri/i965/brw_disasm.c
index 9fef230507..a8f6b993ac 100644
--- a/src/mesa/drivers/dri/i965/brw_disasm.c
+++ b/src/mesa/drivers/dri/i965/brw_disasm.c
@@ -239,7 +239,7 @@ char *imm_encoding[8] = {
     [2] = "UW",
     [3] = "W",
     [5] = "VF",
-    [5] = "V",
+    [6] = "V",
     [7] = "F"
 };
 
@@ -365,6 +365,7 @@ static int format (FILE *f, char *format, ...)
     va_start (args, format);
 
     vsnprintf (buf, sizeof (buf) - 1, format, args);
+    va_end (args);
     string (f, buf);
     return 0;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 44bb7bd588..df281b27d5 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -145,7 +145,7 @@ static void brw_emit_prim(struct brw_context *brw,
    prim_packet.base_vert_location = prim->basevertex;
 
    /* Can't wrap here, since we rely on the validated state. */
-   brw->no_batch_wrap = GL_TRUE;
+   intel->no_batch_wrap = GL_TRUE;
 
    /* If we're set to always flush, do it before and after the primitive emit.
     * We want to catch both missed flushes that hurt instruction/state cache
@@ -153,21 +153,17 @@ static void brw_emit_prim(struct brw_context *brw,
     * the besides the draw code.
     */
    if (intel->always_flush_cache) {
-      BEGIN_BATCH(1, IGNORE_CLIPRECTS);
-      OUT_BATCH(intel->vtbl.flush_cmd());
-      ADVANCE_BATCH();
+      intel_batchbuffer_emit_mi_flush(intel->batch);
    }
    if (prim_packet.verts_per_instance) {
       intel_batchbuffer_data( brw->intel.batch, &prim_packet,
-			      sizeof(prim_packet), LOOP_CLIPRECTS);
+			      sizeof(prim_packet));
    }
    if (intel->always_flush_cache) {
-      BEGIN_BATCH(1, IGNORE_CLIPRECTS);
-      OUT_BATCH(intel->vtbl.flush_cmd());
-      ADVANCE_BATCH();
+      intel_batchbuffer_emit_mi_flush(intel->batch);
    }
 
-   brw->no_batch_wrap = GL_FALSE;
+   intel->no_batch_wrap = GL_FALSE;
 }
 
 static void brw_merge_inputs( struct brw_context *brw,
@@ -343,13 +339,6 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
     * so can't access it earlier.
     */
 
-   LOCK_HARDWARE(intel);
-
-   if (!intel->constant_cliprect && intel->driDrawable->numClipRects == 0) {
-      UNLOCK_HARDWARE(intel);
-      return GL_TRUE;
-   }
-
    for (i = 0; i < nr_prims; i++) {
       uint32_t hw_prim;
 
@@ -360,8 +349,7 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
        * an upper bound of how much we might emit in a single
        * brw_try_draw_prims().
        */
-      intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
-				      LOOP_CLIPRECTS);
+      intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4);
 
       hw_prim = brw_set_prim(brw, prim[i].mode);
 
@@ -408,7 +396,6 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    if (intel->always_flush_batch)
       intel_batchbuffer_flush(intel->batch);
  out:
-   UNLOCK_HARDWARE(intel);
 
    brw_state_cache_check_size(brw);
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 765ae5a2fe..c773b71507 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -243,14 +243,6 @@ static void wrap_buffers( struct brw_context *brw,
       dri_bo_unreference(brw->vb.upload.bo);
    brw->vb.upload.bo = dri_bo_alloc(brw->intel.bufmgr, "temporary VBO",
 				    size, 1);
-
-   /* Set the internal VBO\ to no-backing-store.  We only use them as a
-    * temporary within a brw_try_draw_prims while the lock is held.
-    */
-   /* DON'T DO THIS AS IF WE HAVE TO RE-ORG MEMORY WE NEED SOMEWHERE WITH
-      FAKE TO PUSH THIS STUFF */
-//   if (!brw->intel.ttm)
-//      dri_bo_fake_disable_backing_store(brw->vb.upload.bo, NULL, NULL);
 }
 
 static void get_space( struct brw_context *brw,
@@ -375,7 +367,7 @@ static void brw_prepare_vertices(struct brw_context *brw)
     * isn't an issue at this point.
     */
    if (brw->vb.nr_enabled >= BRW_VEP_MAX) {
-      intel->Fallback = 1;
+      intel->Fallback = GL_TRUE; /* boolean, not bitfield */
       return;
    }
 
@@ -427,7 +419,7 @@ static void brw_prepare_vertices(struct brw_context *brw)
 	    /* Position array not properly enabled:
 	     */
             if (input->glarray->StrideB == 0) {
-               intel->Fallback = 1;
+               intel->Fallback = GL_TRUE; /* boolean, not bitfield */
                return;
             }
 
@@ -502,7 +494,7 @@ static void brw_emit_vertices(struct brw_context *brw)
     * a VE loads from them.
     */
    if (brw->vb.nr_enabled == 0) {
-      BEGIN_BATCH(3, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(3);
       OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | 1);
       OUT_BATCH((0 << BRW_VE0_INDEX_SHIFT) |
 		BRW_VE0_VALID |
@@ -522,7 +514,7 @@ static void brw_emit_vertices(struct brw_context *brw)
     * are interleaved or from the same VBO.  TBD if this makes a
     * performance difference.
     */
-   BEGIN_BATCH(1 + brw->vb.nr_enabled * 4, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(1 + brw->vb.nr_enabled * 4);
    OUT_BATCH((CMD_VERTEX_BUFFER << 16) |
 	     ((1 + brw->vb.nr_enabled * 4) - 2));
 
@@ -535,24 +527,17 @@ static void brw_emit_vertices(struct brw_context *brw)
       OUT_RELOC(input->bo,
 		I915_GEM_DOMAIN_VERTEX, 0,
 		input->offset);
-      if (BRW_IS_IGDNG(brw)) {
-          if (input->stride) {
-              OUT_RELOC(input->bo,
-                        I915_GEM_DOMAIN_VERTEX, 0,
-                        input->offset + input->stride * input->count);
-          } else {
-              assert(input->count == 1);
-              OUT_RELOC(input->bo,
-                        I915_GEM_DOMAIN_VERTEX, 0,
-                        input->offset + input->element_size);
-          }
+      if (intel->is_ironlake) {
+	 OUT_RELOC(input->bo,
+		   I915_GEM_DOMAIN_VERTEX, 0,
+		   input->bo->size - 1);
       } else
           OUT_BATCH(input->stride ? input->count : 0);
       OUT_BATCH(0); /* Instance data step rate */
    }
    ADVANCE_BATCH();
 
-   BEGIN_BATCH(1 + brw->vb.nr_enabled * 2, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(1 + brw->vb.nr_enabled * 2);
    OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + brw->vb.nr_enabled * 2) - 2));
    for (i = 0; i < brw->vb.nr_enabled; i++) {
       struct brw_vertex_element *input = brw->vb.enabled[i];
@@ -578,7 +563,7 @@ static void brw_emit_vertices(struct brw_context *brw)
 		(format << BRW_VE0_FORMAT_SHIFT) |
 		(0 << BRW_VE0_SRC_OFFSET_SHIFT));
 
-      if (BRW_IS_IGDNG(brw))
+      if (intel->is_ironlake)
           OUT_BATCH((comp0 << BRW_VE1_COMPONENT_0_SHIFT) |
                     (comp1 << BRW_VE1_COMPONENT_1_SHIFT) |
                     (comp2 << BRW_VE1_COMPONENT_2_SHIFT) |
@@ -719,14 +704,14 @@ static void brw_emit_index_buffer(struct brw_context *brw)
       ib.header.bits.index_format = get_index_type(index_buffer->type);
       ib.header.bits.cut_index_enable = 0;
 
-      BEGIN_BATCH(4, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(4);
       OUT_BATCH( ib.header.dword );
       OUT_RELOC(brw->ib.bo,
 		I915_GEM_DOMAIN_VERTEX, 0,
 		brw->ib.offset);
       OUT_RELOC(brw->ib.bo,
 		I915_GEM_DOMAIN_VERTEX, 0,
-		brw->ib.offset + brw->ib.size);
+		brw->ib.offset + brw->ib.size - 1);
       OUT_BATCH( 0 );
       ADVANCE_BATCH();
    }
diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h
index 30603bdd0e..39eb88d7c2 100644
--- a/src/mesa/drivers/dri/i965/brw_eu.h
+++ b/src/mesa/drivers/dri/i965/brw_eu.h
@@ -170,11 +170,11 @@ static INLINE struct brw_reg brw_reg( GLuint file,
                                       GLuint writemask )
 {
    struct brw_reg reg;
-   if (type == BRW_GENERAL_REGISTER_FILE)
+   if (file == BRW_GENERAL_REGISTER_FILE)
       assert(nr < BRW_MAX_GRF);
-   else if (type == BRW_MESSAGE_REGISTER_FILE)
-      assert(nr < BRW_MAX_MRF);
-   else if (type == BRW_ARCHITECTURE_REGISTER_FILE)
+   else if (file == BRW_MESSAGE_REGISTER_FILE)
+      assert((nr & ~(1 << 7)) < BRW_MAX_MRF);
+   else if (file == BRW_ARCHITECTURE_REGISTER_FILE)
       assert(nr <= BRW_ARF_IP);
 
    reg.type = type;
@@ -538,7 +538,7 @@ static INLINE struct brw_reg brw_mask_reg( GLuint subnr )
 
 static INLINE struct brw_reg brw_message_reg( GLuint nr )
 {
-   assert(nr < BRW_MAX_MRF);
+   assert((nr & ~(1 << 7)) < BRW_MAX_MRF);
    return brw_vec8_reg(BRW_MESSAGE_REGISTER_FILE,
 		       nr,
 		       0);
diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c
index 241cdc33f8..8d6ac00839 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c
@@ -55,7 +55,8 @@ static void guess_execution_size( struct brw_instruction *insn,
 static void brw_set_dest( struct brw_instruction *insn,
 			  struct brw_reg dest )
 {
-   if (dest.type != BRW_ARCHITECTURE_REGISTER_FILE)
+   if (dest.file != BRW_ARCHITECTURE_REGISTER_FILE &&
+       dest.file != BRW_MESSAGE_REGISTER_FILE)
       assert(dest.nr < 128);
 
    insn->bits1.da1.dest_reg_file = dest.file;
@@ -198,7 +199,7 @@ void brw_set_src1( struct brw_instruction *insn,
        * in the future:
        */
       assert (reg.address_mode == BRW_ADDRESS_DIRECT);
-      //assert (reg.file == BRW_GENERAL_REGISTER_FILE);
+      /* assert (reg.file == BRW_GENERAL_REGISTER_FILE); */
 
       if (insn->header.access_mode == BRW_ALIGN_1) {
 	 insn->bits3.da1.src1_subreg_nr = reg.subnr;
@@ -251,9 +252,10 @@ static void brw_set_math_message( struct brw_context *brw,
 				  GLboolean saturate,
 				  GLuint dataType )
 {
+   struct intel_context *intel = &brw->intel;
    brw_set_src1(insn, brw_imm_d(0));
 
-   if (BRW_IS_IGDNG(brw)) {
+   if (intel->is_ironlake) {
        insn->bits3.math_igdng.function = function;
        insn->bits3.math_igdng.int_type = integer_type;
        insn->bits3.math_igdng.precision = low_precision;
@@ -318,9 +320,10 @@ static void brw_set_urb_message( struct brw_context *brw,
 				 GLuint offset,
 				 GLuint swizzle_control )
 {
+    struct intel_context *intel = &brw->intel;
     brw_set_src1(insn, brw_imm_d(0));
 
-    if (BRW_IS_IGDNG(brw)) {
+    if (intel->is_ironlake) {
         insn->bits3.urb_igdng.opcode = 0;	/* ? */
         insn->bits3.urb_igdng.offset = offset;
         insn->bits3.urb_igdng.swizzle_control = swizzle_control;
@@ -357,9 +360,10 @@ static void brw_set_dp_write_message( struct brw_context *brw,
 				      GLuint response_length,
 				      GLuint end_of_thread )
 {
+   struct intel_context *intel = &brw->intel;
    brw_set_src1(insn, brw_imm_d(0));
 
-   if (BRW_IS_IGDNG(brw)) {
+   if (intel->is_ironlake) {
        insn->bits3.dp_write_igdng.binding_table_index = binding_table_index;
        insn->bits3.dp_write_igdng.msg_control = msg_control;
        insn->bits3.dp_write_igdng.pixel_scoreboard_clear = pixel_scoreboard_clear;
@@ -394,9 +398,10 @@ static void brw_set_dp_read_message( struct brw_context *brw,
 				      GLuint response_length,
 				      GLuint end_of_thread )
 {
+   struct intel_context *intel = &brw->intel;
    brw_set_src1(insn, brw_imm_d(0));
 
-   if (BRW_IS_IGDNG(brw)) {
+   if (intel->is_ironlake) {
        insn->bits3.dp_read_igdng.binding_table_index = binding_table_index;
        insn->bits3.dp_read_igdng.msg_control = msg_control;
        insn->bits3.dp_read_igdng.msg_type = msg_type;
@@ -432,10 +437,11 @@ static void brw_set_sampler_message(struct brw_context *brw,
                                     GLuint header_present,
                                     GLuint simd_mode)
 {
+   struct intel_context *intel = &brw->intel;
    assert(eot == 0);
    brw_set_src1(insn, brw_imm_d(0));
 
-   if (BRW_IS_IGDNG(brw)) {
+   if (intel->is_ironlake) {
       insn->bits3.sampler_igdng.binding_table_index = binding_table_index;
       insn->bits3.sampler_igdng.sampler = sampler;
       insn->bits3.sampler_igdng.msg_type = msg_type;
@@ -446,7 +452,7 @@ static void brw_set_sampler_message(struct brw_context *brw,
       insn->bits3.sampler_igdng.end_of_thread = eot;
       insn->bits2.send_igdng.sfid = BRW_MESSAGE_TARGET_SAMPLER;
       insn->bits2.send_igdng.end_of_thread = eot;
-   } else if (BRW_IS_G4X(brw)) {
+   } else if (intel->is_g4x) {
       insn->bits3.sampler_g4x.binding_table_index = binding_table_index;
       insn->bits3.sampler_g4x.sampler = sampler;
       insn->bits3.sampler_g4x.msg_type = msg_type;
@@ -647,10 +653,11 @@ struct brw_instruction *brw_IF(struct brw_compile *p, GLuint execute_size)
 struct brw_instruction *brw_ELSE(struct brw_compile *p, 
 				 struct brw_instruction *if_insn)
 {
+   struct intel_context *intel = &p->brw->intel;
    struct brw_instruction *insn;
    GLuint br = 1;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
       br = 2;
 
    if (p->single_program_flow) {
@@ -689,9 +696,10 @@ struct brw_instruction *brw_ELSE(struct brw_compile *p,
 void brw_ENDIF(struct brw_compile *p, 
 	       struct brw_instruction *patch_insn)
 {
+   struct intel_context *intel = &p->brw->intel;
    GLuint br = 1;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
       br = 2; 
  
    if (p->single_program_flow) {
@@ -802,10 +810,11 @@ struct brw_instruction *brw_DO(struct brw_compile *p, GLuint execute_size)
 struct brw_instruction *brw_WHILE(struct brw_compile *p, 
                                   struct brw_instruction *do_insn)
 {
+   struct intel_context *intel = &p->brw->intel;
    struct brw_instruction *insn;
    GLuint br = 1;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
       br = 2;
 
    if (p->single_program_flow)
@@ -845,14 +854,15 @@ struct brw_instruction *brw_WHILE(struct brw_compile *p,
 void brw_land_fwd_jump(struct brw_compile *p, 
 		       struct brw_instruction *jmp_insn)
 {
+   struct intel_context *intel = &p->brw->intel;
    struct brw_instruction *landing = &p->store[p->nr_insn];
    GLuint jmpi = 1;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
        jmpi = 2;
 
    assert(jmp_insn->header.opcode == BRW_OPCODE_JMPI);
-   assert(jmp_insn->bits1.da1.src1_reg_file = BRW_IMMEDIATE_VALUE);
+   assert(jmp_insn->bits1.da1.src1_reg_file == BRW_IMMEDIATE_VALUE);
 
    jmp_insn->bits3.ud = jmpi * ((landing - jmp_insn) - 1);
 }
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.c b/src/mesa/drivers/dri/i965/brw_fallback.c
index d27c6c24ca..fe5c1ae279 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.c
+++ b/src/mesa/drivers/dri/i965/brw_fallback.c
@@ -47,7 +47,6 @@
 
 static GLboolean do_check_fallback(struct brw_context *brw)
 {
-   struct intel_context *intel = &brw->intel;
    GLcontext *ctx = &brw->intel.ctx;
    GLuint i;
 
@@ -86,8 +85,7 @@ static GLboolean do_check_fallback(struct brw_context *brw)
    }
 
    /* _NEW_BUFFERS */
-   if (IS_965(intel->intelScreen->deviceID) &&
-       !IS_G4X(intel->intelScreen->deviceID)) {
+   if (!brw->has_surface_tile_offset) {
       for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
 	 struct gl_renderbuffer *rb = ctx->DrawBuffer->_ColorDrawBuffers[i];
 	 struct intel_renderbuffer *irb = intel_renderbuffer(rb);
@@ -133,7 +131,11 @@ const struct brw_tracked_state brw_check_fallback = {
 
 
 
-/* Not used:
+/**
+ * Called by the INTEL_FALLBACK() macro.
+ * NOTE: this is a no-op for the i965 driver.  The brw->intel.Fallback
+ * field is treated as a boolean, not a bitmask.  It's only set in a
+ * couple of places.
  */
 void intelFallback( struct intel_context *intel, GLuint bit, GLboolean mode )
 {
diff --git a/src/mesa/drivers/dri/i965/brw_gs.c b/src/mesa/drivers/dri/i965/brw_gs.c
index 48c2b9a41c..1bc3eccf49 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.c
+++ b/src/mesa/drivers/dri/i965/brw_gs.c
@@ -47,6 +47,7 @@
 static void compile_gs_prog( struct brw_context *brw,
 			     struct brw_gs_prog_key *key )
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_gs_compile c;
    const GLuint *program;
    GLuint program_size;
@@ -54,13 +55,12 @@ static void compile_gs_prog( struct brw_context *brw,
    memset(&c, 0, sizeof(c));
    
    c.key = *key;
-   c.need_ff_sync = BRW_IS_IGDNG(brw);
    /* Need to locate the two positions present in vertex + header.
     * These are currently hardcoded:
     */
    c.nr_attrs = brw_count_bits(c.key.attrs);
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
        c.nr_regs = (c.nr_attrs + 1) / 2 + 3;  /* are vertices packed, or reg-aligned? */
    else
        c.nr_regs = (c.nr_attrs + 1) / 2 + 1;  /* are vertices packed, or reg-aligned? */
@@ -85,10 +85,10 @@ static void compile_gs_prog( struct brw_context *brw,
     */
    switch (key->primitive) {
    case GL_QUADS:
-      brw_gs_quads( &c ); 
+      brw_gs_quads( &c, key );
       break;
    case GL_QUAD_STRIP:
-      brw_gs_quad_strip( &c );
+      brw_gs_quad_strip( &c, key );
       break;
    case GL_LINE_LOOP:
       brw_gs_lines( &c );
@@ -149,6 +149,7 @@ static const GLenum gs_prim[GL_POLYGON+1] = {
 static void populate_key( struct brw_context *brw,
 			  struct brw_gs_prog_key *key )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_VS_PROG */
@@ -158,6 +159,9 @@ static void populate_key( struct brw_context *brw,
    key->primitive = gs_prim[brw->primitive];
 
    key->hint_gs_always = 0;	/* debug code? */
+   
+   /* _NEW_LIGHT */
+   key->pv_first = (ctx->Light.ProvokingVertex == GL_FIRST_VERTEX_CONVENTION);
 
    key->need_gs_prog = (key->hint_gs_always ||
 			brw->primitive == GL_QUADS ||
@@ -193,7 +197,7 @@ static void prepare_gs_prog(struct brw_context *brw)
 
 const struct brw_tracked_state brw_gs_prog = {
    .dirty = {
-      .mesa  = 0,
+      .mesa  = _NEW_LIGHT,
       .brw   = BRW_NEW_PRIMITIVE,
       .cache = CACHE_NEW_VS_PROG
    },
diff --git a/src/mesa/drivers/dri/i965/brw_gs.h b/src/mesa/drivers/dri/i965/brw_gs.h
index bbb991ea2e..813b8d447a 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.h
+++ b/src/mesa/drivers/dri/i965/brw_gs.h
@@ -40,11 +40,12 @@
 #define MAX_GS_VERTS (4)	     
 
 struct brw_gs_prog_key {
-   GLuint attrs:32;
+   GLbitfield64 attrs;
    GLuint primitive:4;
    GLuint hint_gs_always:1;
+   GLuint pv_first:1;
    GLuint need_gs_prog:1;
-   GLuint pad:26;
+   GLuint pad:25;
 };
 
 struct brw_gs_compile {
@@ -62,13 +63,12 @@ struct brw_gs_compile {
    GLuint nr_attrs;
    GLuint nr_regs;
    GLuint nr_bytes;
-   GLboolean need_ff_sync;
 };
 
 #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_quads( struct brw_gs_compile *c, struct brw_gs_prog_key *key );
+void brw_gs_quad_strip( struct brw_gs_compile *c, struct brw_gs_prog_key *key );
 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 );
diff --git a/src/mesa/drivers/dri/i965/brw_gs_emit.c b/src/mesa/drivers/dri/i965/brw_gs_emit.c
index a9b2aa2eac..a81b972ef4 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_emit.c
@@ -120,38 +120,60 @@ static void brw_gs_ff_sync(struct brw_gs_compile *c, int num_prim)
 }
 
 
-void brw_gs_quads( struct brw_gs_compile *c )
+void brw_gs_quads( struct brw_gs_compile *c, struct brw_gs_prog_key *key )
 {
+   struct intel_context *intel = &c->func.brw->intel;
+
    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:
     */
-   if (c->need_ff_sync)
-	   brw_gs_ff_sync(c, 1);    
-   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));
+   if (intel->needs_ff_sync)
+	   brw_gs_ff_sync(c, 1);
+   if (key->pv_first) {
+      brw_gs_emit_vue(c, c->reg.vertex[0], 0, ((_3DPRIM_POLYGON << 2) | R02_PRIM_START));
+      brw_gs_emit_vue(c, c->reg.vertex[1], 0, (_3DPRIM_POLYGON << 2));
+      brw_gs_emit_vue(c, c->reg.vertex[2], 0, (_3DPRIM_POLYGON << 2));
+      brw_gs_emit_vue(c, c->reg.vertex[3], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
+   }
+   else {
+      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 )
+void brw_gs_quad_strip( struct brw_gs_compile *c, struct brw_gs_prog_key *key )
 {
+   struct intel_context *intel = &c->func.brw->intel;
+
    brw_gs_alloc_regs(c, 4);
    
-   if (c->need_ff_sync)
+   if (intel->needs_ff_sync)
 	   brw_gs_ff_sync(c, 1);      
-   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));
+   if (key->pv_first) {
+      brw_gs_emit_vue(c, c->reg.vertex[0], 0, ((_3DPRIM_POLYGON << 2) | R02_PRIM_START));
+      brw_gs_emit_vue(c, c->reg.vertex[1], 0, (_3DPRIM_POLYGON << 2));
+      brw_gs_emit_vue(c, c->reg.vertex[2], 0, (_3DPRIM_POLYGON << 2));
+      brw_gs_emit_vue(c, c->reg.vertex[3], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
+   }
+   else {
+      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 )
 {
+   struct intel_context *intel = &c->func.brw->intel;
+
    brw_gs_alloc_regs(c, 3);
 
-   if (c->need_ff_sync)
+   if (intel->needs_ff_sync)
 	   brw_gs_ff_sync(c, 1);      
    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));
@@ -160,9 +182,11 @@ void brw_gs_tris( struct brw_gs_compile *c )
 
 void brw_gs_lines( struct brw_gs_compile *c )
 {
+   struct intel_context *intel = &c->func.brw->intel;
+
    brw_gs_alloc_regs(c, 2);
 
-   if (c->need_ff_sync)
+   if (intel->needs_ff_sync)
 	   brw_gs_ff_sync(c, 1);      
    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));
@@ -170,9 +194,11 @@ void brw_gs_lines( struct brw_gs_compile *c )
 
 void brw_gs_points( struct brw_gs_compile *c )
 {
+   struct intel_context *intel = &c->func.brw->intel;
+
    brw_gs_alloc_regs(c, 1);
 
-   if (c->need_ff_sync)
+   if (intel->needs_ff_sync)
 	   brw_gs_ff_sync(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/mesa/drivers/dri/i965/brw_gs_state.c b/src/mesa/drivers/dri/i965/brw_gs_state.c
index ed9d2ffe60..1af5790a67 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_state.c
@@ -72,6 +72,7 @@ gs_unit_populate_key(struct brw_context *brw, struct brw_gs_unit_key *key)
 static dri_bo *
 gs_unit_create_from_key(struct brw_context *brw, struct brw_gs_unit_key *key)
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_gs_unit_state gs;
    dri_bo *bo;
 
@@ -98,7 +99,7 @@ gs_unit_create_from_key(struct brw_context *brw, struct brw_gs_unit_key *key)
    else
       gs.thread4.max_threads = 0;
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
       gs.thread4.rendering_enable = 1;
 
    if (INTEL_DEBUG & DEBUG_STATS)
diff --git a/src/mesa/drivers/dri/i965/brw_misc_state.c b/src/mesa/drivers/dri/i965/brw_misc_state.c
index ea71857548..7b70f787b7 100644
--- a/src/mesa/drivers/dri/i965/brw_misc_state.c
+++ b/src/mesa/drivers/dri/i965/brw_misc_state.c
@@ -66,7 +66,7 @@ static void upload_blend_constant_color(struct brw_context *brw)
 const struct brw_tracked_state brw_blend_constant_color = {
    .dirty = {
       .mesa = _NEW_COLOR,
-      .brw = 0,
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
    .emit = upload_blend_constant_color
@@ -78,10 +78,7 @@ static void upload_drawing_rect(struct brw_context *brw)
    struct intel_context *intel = &brw->intel;
    GLcontext *ctx = &intel->ctx;
 
-   if (!intel->constant_cliprect)
-      return;
-
-   BEGIN_BATCH(4, NO_LOOP_CLIPRECTS);
+   BEGIN_BATCH(4);
    OUT_BATCH(_3DSTATE_DRAWRECT_INFO_I965);
    OUT_BATCH(0); /* xmin, ymin */
    OUT_BATCH(((ctx->DrawBuffer->Width - 1) & 0xffff) |
@@ -93,7 +90,7 @@ static void upload_drawing_rect(struct brw_context *brw)
 const struct brw_tracked_state brw_drawing_rect = {
    .dirty = {
       .mesa = _NEW_BUFFERS,
-      .brw = 0,
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
    .emit = upload_drawing_rect
@@ -116,7 +113,7 @@ static void upload_binding_table_pointers(struct brw_context *brw)
 {
    struct intel_context *intel = &brw->intel;
 
-   BEGIN_BATCH(6, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(6);
    OUT_BATCH(CMD_BINDING_TABLE_PTRS << 16 | (6 - 2));
    if (brw->vs.bind_bo != NULL)
       OUT_RELOC(brw->vs.bind_bo, I915_GEM_DOMAIN_SAMPLER, 0, 0); /* vs */
@@ -150,7 +147,7 @@ static void upload_pipelined_state_pointers(struct brw_context *brw )
 {
    struct intel_context *intel = &brw->intel;
 
-   BEGIN_BATCH(7, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(7);
    OUT_BATCH(CMD_PIPELINED_STATE_POINTERS << 16 | (7 - 2));
    OUT_RELOC(brw->vs.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
    if (brw->gs.prog_active)
@@ -212,10 +209,10 @@ static void emit_depthbuffer(struct brw_context *brw)
 {
    struct intel_context *intel = &brw->intel;
    struct intel_region *region = brw->state.depth_region;
-   unsigned int len = (BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw)) ? 6 : 5;
+   unsigned int len = (intel->is_g4x || intel->is_ironlake) ? 6 : 5;
 
    if (region == NULL) {
-      BEGIN_BATCH(len, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(len);
       OUT_BATCH(CMD_DEPTH_BUFFER << 16 | (len - 2));
       OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
 		(BRW_SURFACE_NULL << 29));
@@ -223,7 +220,7 @@ static void emit_depthbuffer(struct brw_context *brw)
       OUT_BATCH(0);
       OUT_BATCH(0);
 
-      if (BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw))
+      if (intel->is_g4x || intel->is_ironlake)
          OUT_BATCH(0);
 
       ADVANCE_BATCH();
@@ -247,7 +244,7 @@ static void emit_depthbuffer(struct brw_context *brw)
 
       assert(region->tiling != I915_TILING_X);
 
-      BEGIN_BATCH(len, IGNORE_CLIPRECTS);
+      BEGIN_BATCH(len);
       OUT_BATCH(CMD_DEPTH_BUFFER << 16 | (len - 2));
       OUT_BATCH(((region->pitch * region->cpp) - 1) |
 		(format << 18) |
@@ -262,7 +259,7 @@ static void emit_depthbuffer(struct brw_context *brw)
 		((region->height - 1) << 19));
       OUT_BATCH(0);
 
-      if (BRW_IS_G4X(brw) || BRW_IS_IGDNG(brw))
+      if (intel->is_g4x || intel->is_ironlake)
          OUT_BATCH(0);
 
       ADVANCE_BATCH();
@@ -317,7 +314,7 @@ static void upload_polygon_stipple(struct brw_context *brw)
 const struct brw_tracked_state brw_polygon_stipple = {
    .dirty = {
       .mesa = _NEW_POLYGONSTIPPLE,
-      .brw = 0,
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
    .emit = upload_polygon_stipple
@@ -330,7 +327,7 @@ const struct brw_tracked_state brw_polygon_stipple = {
 
 static void upload_polygon_stipple_offset(struct brw_context *brw)
 {
-   __DRIdrawablePrivate *dPriv = brw->intel.driDrawable;
+   __DRIdrawable *dPriv = brw->intel.driDrawable;
    struct brw_polygon_stipple_offset bpso;
 
    memset(&bpso, 0, sizeof(bpso));
@@ -362,7 +359,7 @@ static void upload_polygon_stipple_offset(struct brw_context *brw)
 const struct brw_tracked_state brw_polygon_stipple_offset = {
    .dirty = {
       .mesa = _NEW_WINDOW_POS,
-      .brw = 0,
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
    .emit = upload_polygon_stipple_offset
@@ -374,8 +371,8 @@ const struct brw_tracked_state brw_polygon_stipple_offset = {
 static void upload_aa_line_parameters(struct brw_context *brw)
 {
    struct brw_aa_line_parameters balp;
-   
-   if (BRW_IS_965(brw))
+
+   if (!brw->has_aa_line_parameters)
       return;
 
    /* use legacy aa line coverage computation */
@@ -425,7 +422,7 @@ static void upload_line_stipple(struct brw_context *brw)
 const struct brw_tracked_state brw_line_stipple = {
    .dirty = {
       .mesa = _NEW_LINE,
-      .brw = 0,
+      .brw = BRW_NEW_CONTEXT,
       .cache = 0
    },
    .emit = upload_line_stipple
@@ -444,7 +441,7 @@ static void upload_invarient_state( struct brw_context *brw )
       struct brw_pipeline_select ps;
 
       memset(&ps, 0, sizeof(ps));
-      ps.header.opcode = CMD_PIPELINE_SELECT(brw);
+      ps.header.opcode = brw->CMD_PIPELINE_SELECT;
       ps.header.pipeline_select = 0;
       BRW_BATCH_STRUCT(brw, &ps);
    }
@@ -480,7 +477,7 @@ static void upload_invarient_state( struct brw_context *brw )
       struct brw_vf_statistics vfs;
       memset(&vfs, 0, sizeof(vfs));
 
-      vfs.opcode = CMD_VF_STATISTICS(brw);
+      vfs.opcode = brw->CMD_VF_STATISTICS;
       if (INTEL_DEBUG & DEBUG_STATS)
 	 vfs.statistics_enable = 1; 
 
@@ -512,8 +509,8 @@ 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.
     */
-   if (BRW_IS_IGDNG(brw)) {
-       BEGIN_BATCH(8, IGNORE_CLIPRECTS);
+   if (intel->is_ironlake) {
+       BEGIN_BATCH(8);
        OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (8 - 2));
        OUT_BATCH(1); /* General state base address */
        OUT_BATCH(1); /* Surface state base address */
@@ -524,7 +521,7 @@ static void upload_state_base_address( struct brw_context *brw )
        OUT_BATCH(1); /* Instruction access upper bound */
        ADVANCE_BATCH();
    } else {
-       BEGIN_BATCH(6, IGNORE_CLIPRECTS);
+       BEGIN_BATCH(6);
        OUT_BATCH(CMD_STATE_BASE_ADDRESS << 16 | (6 - 2));
        OUT_BATCH(1); /* General state base address */
        OUT_BATCH(1); /* Surface state base address */
diff --git a/src/mesa/drivers/dri/i965/brw_queryobj.c b/src/mesa/drivers/dri/i965/brw_queryobj.c
index a195bc32b0..5399a74244 100644
--- a/src/mesa/drivers/dri/i965/brw_queryobj.c
+++ b/src/mesa/drivers/dri/i965/brw_queryobj.c
@@ -188,7 +188,7 @@ brw_emit_query_begin(struct brw_context *brw)
    if (brw->query.active || is_empty_list(&brw->query.active_head))
       return;
 
-   BEGIN_BATCH(4, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(4);
    OUT_BATCH(_3DSTATE_PIPE_CONTROL |
 	     PIPE_CONTROL_DEPTH_STALL |
 	     PIPE_CONTROL_WRITE_DEPTH_COUNT);
@@ -227,7 +227,7 @@ brw_emit_query_end(struct brw_context *brw)
    if (!brw->query.active)
       return;
 
-   BEGIN_BATCH(4, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(4);
    OUT_BATCH(_3DSTATE_PIPE_CONTROL |
 	     PIPE_CONTROL_DEPTH_STALL |
 	     PIPE_CONTROL_WRITE_DEPTH_COUNT);
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index e1c2c7777b..968890f7fb 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -61,7 +61,7 @@ static void compile_sf_prog( struct brw_context *brw,
    c.key = *key;
    c.nr_attrs = brw_count_bits(c.key.attrs);
    c.nr_attr_regs = (c.nr_attrs+1)/2;
-   c.nr_setup_attrs = brw_count_bits(c.key.attrs & DO_SETUP_BITS);
+   c.nr_setup_attrs = brw_count_bits(c.key.attrs);
    c.nr_setup_regs = (c.nr_setup_attrs+1)/2;
 
    c.prog_data.urb_read_length = c.nr_attr_regs;
@@ -70,7 +70,7 @@ static void compile_sf_prog( struct brw_context *brw,
    /* Construct map from attribute number to position in the vertex.
     */
    for (i = idx = 0; i < VERT_RESULT_MAX; i++) 
-      if (c.key.attrs & (1<<i)) {
+      if (c.key.attrs & BITFIELD64_BIT(i)) {
 	 c.attr_to_idx[i] = idx;
 	 c.idx_to_attr[idx] = i;
 	 if (i >= VERT_RESULT_TEX0 && i <= VERT_RESULT_TEX7) {
@@ -147,7 +147,7 @@ static void upload_sf_prog(struct brw_context *brw)
        * edgeflag testing here, it is already done in the clip
        * program.
        */
-      if (key.attrs & (1<<VERT_RESULT_EDGE))
+      if (key.attrs & BITFIELD64_BIT(VERT_RESULT_EDGE))
 	 key.primitive = SF_UNFILLED_TRIS;
       else
 	 key.primitive = SF_TRIANGLES;
@@ -161,7 +161,7 @@ static void upload_sf_prog(struct brw_context *brw)
    }
 
    key.do_point_sprite = ctx->Point.PointSprite;
-   key.SpriteOrigin = ctx->Point.SpriteOrigin;
+   key.sprite_origin_lower_left = (ctx->Point.SpriteOrigin == GL_LOWER_LEFT);
    /* _NEW_LIGHT */
    key.do_flat_shading = (ctx->Light.ShadeModel == GL_FLAT);
    key.do_twoside_color = (ctx->Light.Enabled && ctx->Light.Model.TwoSide);
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index 6426b6df9f..0ba731fac9 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -45,19 +45,19 @@
 #define SF_UNFILLED_TRIS   3
 
 struct brw_sf_prog_key {
-   GLuint attrs:32;
+   GLbitfield64 attrs;
    GLuint primitive:2;
    GLuint do_twoside_color:1;
    GLuint do_flat_shading:1;
    GLuint frontface_ccw:1;
    GLuint do_point_sprite:1;
    GLuint linear_color:1;  /**< linear interp vs. perspective interp */
-   GLuint pad:25;
-   GLenum SpriteOrigin;
+   GLuint sprite_origin_lower_left:1;
+   GLuint pad:24;
 };
 
 struct brw_sf_point_tex {
-	GLboolean CoordReplace;	
+   GLboolean CoordReplace;	
 };
 
 struct brw_sf_compile {
diff --git a/src/mesa/drivers/dri/i965/brw_sf_emit.c b/src/mesa/drivers/dri/i965/brw_sf_emit.c
index ca8f97f9f9..bb08055e3b 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_emit.c
@@ -56,7 +56,7 @@ static struct brw_reg get_vert_attr(struct brw_sf_compile *c,
 static GLboolean have_attr(struct brw_sf_compile *c,
 			   GLuint attr)
 {
-   return (c->key.attrs & (1<<attr)) ? 1 : 0;
+   return (c->key.attrs & BITFIELD64_BIT(attr)) ? 1 : 0;
 }
 
 /*********************************************************************** 
@@ -122,8 +122,8 @@ static void do_twoside_color( struct brw_sf_compile *c )
  * Flat shading
  */
 
-#define VERT_RESULT_COLOR_BITS ((1<<VERT_RESULT_COL0) | \
-                                 (1<<VERT_RESULT_COL1))
+#define VERT_RESULT_COLOR_BITS (BITFIELD64_BIT(VERT_RESULT_COL0) | \
+				BITFIELD64_BIT(VERT_RESULT_COL1))
 
 static void copy_colors( struct brw_sf_compile *c,
 		     struct brw_reg dst,
@@ -149,6 +149,7 @@ static void copy_colors( struct brw_sf_compile *c,
 static void do_flatshade_triangle( struct brw_sf_compile *c )
 {
    struct brw_compile *p = &c->func;
+   struct intel_context *intel = &p->brw->intel;
    struct brw_reg ip = brw_ip_reg();
    GLuint nr = brw_count_bits(c->key.attrs & VERT_RESULT_COLOR_BITS);
    GLuint jmpi = 1;
@@ -161,7 +162,7 @@ static void do_flatshade_triangle( struct brw_sf_compile *c )
    if (c->key.primitive == SF_UNFILLED_TRIS)
       return;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
        jmpi = 2;
 
    brw_push_insn_state(p);
@@ -187,6 +188,7 @@ static void do_flatshade_triangle( struct brw_sf_compile *c )
 static void do_flatshade_line( struct brw_sf_compile *c )
 {
    struct brw_compile *p = &c->func;
+   struct intel_context *intel = &p->brw->intel;
    struct brw_reg ip = brw_ip_reg();
    GLuint nr = brw_count_bits(c->key.attrs & VERT_RESULT_COLOR_BITS);
    GLuint jmpi = 1;
@@ -199,7 +201,7 @@ static void do_flatshade_line( struct brw_sf_compile *c )
    if (c->key.primitive == SF_UNFILLED_TRIS)
       return;
 
-   if (BRW_IS_IGDNG(p->brw))
+   if (intel->is_ironlake)
        jmpi = 2;
 
    brw_push_insn_state(p);
@@ -312,8 +314,8 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
 				  GLushort *pc_linear)
 {
    GLboolean is_last_attr = (reg == c->nr_setup_regs - 1);
-   GLuint persp_mask;
-   GLuint linear_mask;
+   GLbitfield64 persp_mask;
+   GLbitfield64 linear_mask;
 
    if (c->key.do_flat_shading || c->key.linear_color)
       persp_mask = c->key.attrs & ~(FRAG_BIT_WPOS |
@@ -331,10 +333,10 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
    *pc_linear = 0;
    *pc = 0xf;
       
-   if (persp_mask & (1 << c->idx_to_attr[reg*2])) 
+   if (persp_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2]))
       *pc_persp = 0xf;
 
-   if (linear_mask & (1 << c->idx_to_attr[reg*2])) 
+   if (linear_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2]))
       *pc_linear = 0xf;
 
    /* Maybe only processs one attribute on the final round:
@@ -342,10 +344,10 @@ static GLboolean calculate_masks( struct brw_sf_compile *c,
    if (reg*2+1 < c->nr_setup_attrs) {
       *pc |= 0xf0;
 
-      if (persp_mask & (1 << c->idx_to_attr[reg*2+1])) 
+      if (persp_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2+1]))
 	 *pc_persp |= 0xf0;
 
-      if (linear_mask & (1 << c->idx_to_attr[reg*2+1])) 
+      if (linear_mask & BITFIELD64_BIT(c->idx_to_attr[reg*2+1]))
 	 *pc_linear |= 0xf0;
    }
 
@@ -551,7 +553,7 @@ void brw_emit_point_sprite_setup( struct brw_sf_compile *c, GLboolean allocate)
 		  BRW_MATH_DATA_SCALAR,
 		  BRW_MATH_PRECISION_FULL);
 
-	  if (c->key.SpriteOrigin == GL_LOWER_LEFT) {
+	  if (c->key.sprite_origin_lower_left) {
 	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
 		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
 	  	brw_MUL(p, c->m2Cy, c->tmp, negate(c->inv_w[0]));
@@ -570,7 +572,7 @@ void brw_emit_point_sprite_setup( struct brw_sf_compile *c, GLboolean allocate)
       {
 	 brw_set_predicate_control_flag_value(p, pc); 
 	 if (tex->CoordReplace) {
-	     if (c->key.SpriteOrigin == GL_LOWER_LEFT) {
+	     if (c->key.sprite_origin_lower_left) {
 		 brw_MUL(p, c->m3C0, c->inv_w[0], brw_imm_f(1.0));
 		 brw_MOV(p, vec1(suboffset(c->m3C0, 0)), brw_imm_f(0.0));
 	     }
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index bc0f076073..09223b7cfb 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -93,7 +93,8 @@ static void upload_sf_vp(struct brw_context *brw)
    }
 
    dri_bo_unreference(brw->sf.vp_bo);
-   brw->sf.vp_bo = brw_cache_data( &brw->cache, BRW_SF_VP, &sfv, NULL, 0 );
+   brw->sf.vp_bo = brw_cache_data(&brw->cache, BRW_SF_VP, &sfv, sizeof(sfv),
+				  NULL, 0);
 }
 
 const struct brw_tracked_state brw_sf_vp = {
@@ -113,7 +114,8 @@ struct brw_sf_unit_key {
 
    unsigned int nr_urb_entries, urb_size, sfsize;
 
-   GLenum front_face, cull_face, provoking_vertex;
+   GLenum front_face, cull_face;
+   unsigned pv_first:1;
    unsigned scissor:1;
    unsigned line_smooth:1;
    unsigned point_sprite:1;
@@ -154,7 +156,7 @@ sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key)
    key->point_attenuated = ctx->Point._Attenuated;
 
    /* _NEW_LIGHT */
-   key->provoking_vertex = ctx->Light.ProvokingVertex;
+   key->pv_first = (ctx->Light.ProvokingVertex == GL_FIRST_VERTEX_CONVENTION);
 
    key->render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
 }
@@ -163,6 +165,7 @@ static dri_bo *
 sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
 			dri_bo **reloc_bufs)
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_sf_unit_state sf;
    dri_bo *bo;
    int chipset_max_threads;
@@ -175,7 +178,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
 
    sf.thread3.dispatch_grf_start_reg = 3;
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
        sf.thread3.urb_entry_read_offset = 3;
    else
        sf.thread3.urb_entry_read_offset = 1;
@@ -185,10 +188,10 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
    sf.thread4.nr_urb_entries = key->nr_urb_entries;
    sf.thread4.urb_entry_allocation_size = key->sfsize - 1;
 
-   /* Each SF thread produces 1 PUE, and there can be up to 24(Pre-IGDNG) or 
-    * 48(IGDNG) threads 
+   /* Each SF thread produces 1 PUE, and there can be up to 24 (Pre-Ironlake) or
+    * 48 (Ironlake) threads.
     */
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
       chipset_max_threads = 48;
    else
       chipset_max_threads = 24;
@@ -287,7 +290,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
 
    /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
     */
-   if (key->provoking_vertex == GL_LAST_VERTEX_CONVENTION) {
+   if (!key->pv_first) {
       sf.sf7.trifan_pv = 2;
       sf.sf7.linestrip_pv = 1;
       sf.sf7.tristrip_pv = 2;
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index d639656b9d..9c9d145c4b 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -35,7 +35,7 @@
 
 #include "brw_context.h"
 
-static inline void
+static INLINE void
 brw_add_validated_bo(struct brw_context *brw, dri_bo *bo)
 {
    assert(brw->state.validated_bo_count < ARRAY_SIZE(brw->state.validated_bos));
@@ -112,6 +112,7 @@ void brw_validate_state(struct brw_context *brw);
 void brw_upload_state(struct brw_context *brw);
 void brw_init_state(struct brw_context *brw);
 void brw_destroy_state(struct brw_context *brw);
+void brw_clear_validated_bos(struct brw_context *brw);
 
 /***********************************************************************
  * brw_state_cache.c
@@ -119,16 +120,10 @@ void brw_destroy_state(struct brw_context *brw);
 dri_bo *brw_cache_data(struct brw_cache *cache,
 		       enum brw_cache_id cache_id,
 		       const void *data,
+		       GLuint size,
 		       dri_bo **reloc_bufs,
 		       GLuint nr_reloc_bufs);
 
-dri_bo *brw_cache_data_sz(struct brw_cache *cache,
-			  enum brw_cache_id cache_id,
-			  const void *data,
-			  GLuint data_size,
-			  dri_bo **reloc_bufs,
-			  GLuint nr_reloc_bufs);
-
 dri_bo *brw_upload_cache( struct brw_cache *cache,
 			  enum brw_cache_id cache_id,
 			  const void *key,
@@ -156,7 +151,7 @@ void brw_state_cache_bo_delete(struct brw_cache *cache, dri_bo *bo);
 /***********************************************************************
  * brw_state_batch.c
  */
-#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data( brw->intel.batch, (s), sizeof(*(s)), IGNORE_CLIPRECTS)
+#define BRW_BATCH_STRUCT(brw, s) intel_batchbuffer_data( brw->intel.batch, (s), sizeof(*(s)))
 #define BRW_CACHED_BATCH_STRUCT(brw, s) brw_cached_batch_struct( brw, (s), sizeof(*(s)) )
 
 GLboolean brw_cached_batch_struct( struct brw_context *brw,
diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c
index 7821898cf9..ed8120d617 100644
--- a/src/mesa/drivers/dri/i965/brw_state_batch.c
+++ b/src/mesa/drivers/dri/i965/brw_state_batch.c
@@ -48,7 +48,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw,
    struct header *newheader = (struct header *)data;
 
    if (brw->emit_state_always) {
-      intel_batchbuffer_data(brw->intel.batch, data, sz, IGNORE_CLIPRECTS);
+      intel_batchbuffer_data(brw->intel.batch, data, sz);
       return GL_TRUE;
    }
 
@@ -75,7 +75,7 @@ GLboolean brw_cached_batch_struct( struct brw_context *brw,
 
  emit:
    memcpy(item->header, newheader, sz);
-   intel_batchbuffer_data(brw->intel.batch, data, sz, IGNORE_CLIPRECTS);
+   intel_batchbuffer_data(brw->intel.batch, data, sz);
    return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index f8e46aacf7..e4c9ba7d87 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -245,7 +245,6 @@ brw_upload_cache( struct brw_cache *cache,
 
    item->bo = bo;
    dri_bo_reference(bo);
-   item->data_size = data_size;
 
    if (cache->n_items > cache->size * 1.5)
       rehash(cache);
@@ -275,15 +274,22 @@ brw_upload_cache( struct brw_cache *cache,
 
 
 /**
- * This doesn't really work with aux data.  Use search/upload instead
+ * Wrapper around brw_cache_data_sz using the cache_id's canonical key size.
+ *
+ * If nr_reloc_bufs is nonzero, brw_search_cache()/brw_upload_cache() would be
+ * better to use, as the potentially changing offsets in the data-used-as-key
+ * will result in excessive cache misses.
+ *
+ * If aux data is involved, use search/upload instead.
+
  */
 dri_bo *
-brw_cache_data_sz(struct brw_cache *cache,
-		  enum brw_cache_id cache_id,
-		  const void *data,
-		  GLuint data_size,
-		  dri_bo **reloc_bufs,
-		  GLuint nr_reloc_bufs)
+brw_cache_data(struct brw_cache *cache,
+	       enum brw_cache_id cache_id,
+	       const void *data,
+	       GLuint data_size,
+	       dri_bo **reloc_bufs,
+	       GLuint nr_reloc_bufs)
 {
    dri_bo *bo;
    struct brw_cache_item *item;
@@ -306,25 +312,6 @@ brw_cache_data_sz(struct brw_cache *cache,
    return bo;
 }
 
-
-/**
- * Wrapper around brw_cache_data_sz using the cache_id's canonical key size.
- *
- * If nr_reloc_bufs is nonzero, brw_search_cache()/brw_upload_cache() would be
- * better to use, as the potentially changing offsets in the data-used-as-key
- * will result in excessive cache misses.
- */
-dri_bo *
-brw_cache_data(struct brw_cache *cache,
-	       enum brw_cache_id cache_id,
-	       const void *data,
-	       dri_bo **reloc_bufs,
-	       GLuint nr_reloc_bufs)
-{
-   return brw_cache_data_sz(cache, cache_id, data, cache->key_size[cache_id],
-			    reloc_bufs, nr_reloc_bufs);
-}
-
 enum pool_type {
    DW_SURFACE_STATE,
    DW_GENERAL_STATE
@@ -335,11 +322,9 @@ static void
 brw_init_cache_id(struct brw_cache *cache,
                   const char *name,
                   enum brw_cache_id id,
-                  GLuint key_size,
                   GLuint aux_size)
 {
    cache->name[id] = strdup(name);
-   cache->key_size[id] = key_size;
    cache->aux_size[id] = aux_size;
 }
 
@@ -359,91 +344,76 @@ brw_init_non_surface_cache(struct brw_context *brw)
    brw_init_cache_id(cache,
 		     "CC_VP",
 		     BRW_CC_VP,
-		     sizeof(struct brw_cc_viewport),
 		     0);
 
    brw_init_cache_id(cache,
 		     "CC_UNIT",
 		     BRW_CC_UNIT,
-		     sizeof(struct brw_cc_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "WM_PROG",
 		     BRW_WM_PROG,
-		     sizeof(struct brw_wm_prog_key),
 		     sizeof(struct brw_wm_prog_data));
 
    brw_init_cache_id(cache,
 		     "SAMPLER_DEFAULT_COLOR",
 		     BRW_SAMPLER_DEFAULT_COLOR,
-		     sizeof(struct brw_sampler_default_color),
 		     0);
 
    brw_init_cache_id(cache,
 		     "SAMPLER",
 		     BRW_SAMPLER,
-		     0,		/* variable key/data size */
 		     0);
 
    brw_init_cache_id(cache,
 		     "WM_UNIT",
 		     BRW_WM_UNIT,
-		     sizeof(struct brw_wm_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "SF_PROG",
 		     BRW_SF_PROG,
-		     sizeof(struct brw_sf_prog_key),
 		     sizeof(struct brw_sf_prog_data));
 
    brw_init_cache_id(cache,
 		     "SF_VP",
 		     BRW_SF_VP,
-		     sizeof(struct brw_sf_viewport),
 		     0);
 
    brw_init_cache_id(cache,
 		     "SF_UNIT",
 		     BRW_SF_UNIT,
-		     sizeof(struct brw_sf_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "VS_UNIT",
 		     BRW_VS_UNIT,
-		     sizeof(struct brw_vs_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "VS_PROG",
 		     BRW_VS_PROG,
-		     sizeof(struct brw_vs_prog_key),
 		     sizeof(struct brw_vs_prog_data));
 
    brw_init_cache_id(cache,
 		     "CLIP_UNIT",
 		     BRW_CLIP_UNIT,
-		     sizeof(struct brw_clip_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "CLIP_PROG",
 		     BRW_CLIP_PROG,
-		     sizeof(struct brw_clip_prog_key),
 		     sizeof(struct brw_clip_prog_data));
 
    brw_init_cache_id(cache,
 		     "GS_UNIT",
 		     BRW_GS_UNIT,
-		     sizeof(struct brw_gs_unit_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "GS_PROG",
 		     BRW_GS_PROG,
-		     sizeof(struct brw_gs_prog_key),
 		     sizeof(struct brw_gs_prog_data));
 }
 
@@ -463,13 +433,11 @@ brw_init_surface_cache(struct brw_context *brw)
    brw_init_cache_id(cache,
 		     "SS_SURFACE",
 		     BRW_SS_SURFACE,
-		     sizeof(struct brw_surface_state),
 		     0);
 
    brw_init_cache_id(cache,
 		     "SS_SURF_BIND",
 		     BRW_SS_SURF_BIND,
-		     0,
 		     0);
 }
 
@@ -534,14 +502,9 @@ 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 +514,8 @@ 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_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index b817b741e7..af8dfb4c15 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -34,6 +34,7 @@
 #include "brw_context.h"
 #include "brw_state.h"
 #include "intel_batchbuffer.h"
+#include "intel_buffers.h"
 
 /* This is used to initialize brw->state.atoms[].  We could use this
  * list directly except for a single atom, brw_constant_buffer, which
@@ -142,7 +143,7 @@ static void xor_states( struct brw_state_flags *result,
    result->cache = a->cache ^ b->cache;
 }
 
-static void
+void
 brw_clear_validated_bos(struct brw_context *brw)
 {
    int i;
@@ -308,7 +309,7 @@ void brw_validate_state( struct brw_context *brw )
    if (brw->state.dirty.brw & BRW_NEW_CONTEXT)
       brw_clear_batch_cache(brw);
 
-   brw->intel.Fallback = 0;
+   brw->intel.Fallback = GL_FALSE; /* boolean, not bitfield */
 
    /* do prepare stage for all atoms */
    for (i = 0; i < Elements(atoms); i++) {
@@ -324,6 +325,8 @@ void brw_validate_state( struct brw_context *brw )
       }
    }
 
+   intel_check_front_buffer_rendering(intel);
+
    /* Make sure that the textures which are referenced by the current
     * brw fragment program are actually present/valid.
     * If this fails, we can experience GPU lock-ups.
diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
index 5986cbffad..64a9535282 100644
--- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
+++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
@@ -49,7 +49,7 @@ GLboolean brw_miptree_layout(struct intel_context *intel,
 
    switch (mt->target) {
    case GL_TEXTURE_CUBE_MAP:
-      if (IS_IGDNG(intel->intelScreen->deviceID)) {
+      if (intel->is_ironlake) {
           GLuint align_h = 2, align_w = 4;
           GLuint level;
           GLuint x = 0;
@@ -86,10 +86,10 @@ GLboolean brw_miptree_layout(struct intel_context *intel,
           mt->pitch = intel_miptree_pitch_align(intel, mt, tiling, mt->pitch);
 
           if (mt->compressed) {
-              qpitch = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h) / 4 * mt->pitch * mt->cpp;
+              qpitch = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h) / 4;
               mt->total_height = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h) / 4 * 6;
           } else {
-              qpitch = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h) * mt->pitch * mt->cpp;
+              qpitch = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h);
               mt->total_height = (y_pitch + ALIGN(minify(y_pitch), align_h) + 11 * align_h) * 6;
           }
 
@@ -102,7 +102,8 @@ GLboolean brw_miptree_layout(struct intel_context *intel,
                                            height, 1);
 
               for (q = 0; q < nr_images; q++)
-                  intel_miptree_set_image_offset_ex(mt, level, q, x, y, q * qpitch);
+                  intel_miptree_set_image_offset(mt, level, q,
+						 x, y + q * qpitch);
 
               if (mt->compressed)
                   img_height = MAX2(1, height/4);
diff --git a/src/mesa/drivers/dri/i965/brw_urb.c b/src/mesa/drivers/dri/i965/brw_urb.c
index 8c6f4355a6..f2cdb203b8 100644
--- a/src/mesa/drivers/dri/i965/brw_urb.c
+++ b/src/mesa/drivers/dri/i965/brw_urb.c
@@ -105,7 +105,8 @@ static GLboolean check_urb_layout( struct brw_context *brw )
    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 <= URB_SIZES(brw);
+   return brw->urb.cs_start + brw->urb.nr_cs_entries *
+      brw->urb.csize <= brw->urb.size;
 }
 
 /* Most minimal update, forces re-emit of URB fence packet after GS
@@ -113,6 +114,7 @@ static GLboolean check_urb_layout( struct brw_context *brw )
  */
 static void recalculate_urb_fence( struct brw_context *brw )
 {
+   struct intel_context *intel = &brw->intel;
    GLuint csize = brw->curbe.total_size;
    GLuint vsize = brw->vs.prog_data->urb_entry_size;
    GLuint sfsize = brw->sf.prog_data->urb_entry_size;
@@ -146,7 +148,7 @@ static void recalculate_urb_fence( struct brw_context *brw )
 
       brw->urb.constrained = 0;
 
-      if (BRW_IS_IGDNG(brw)) {
+      if (intel->is_ironlake) {
          brw->urb.nr_vs_entries = 128;
          brw->urb.nr_sf_entries = 48;
          if (check_urb_layout(brw)) {
@@ -156,7 +158,7 @@ static void recalculate_urb_fence( struct brw_context *brw )
             brw->urb.nr_vs_entries = limits[VS].preferred_nr_entries;
             brw->urb.nr_sf_entries = limits[SF].preferred_nr_entries;
          }
-      } else if (BRW_IS_G4X(brw)) {
+      } else if (intel->is_g4x) {
 	 brw->urb.nr_vs_entries = 64;
 	 if (check_urb_layout(brw)) {
 	    goto done;
@@ -200,7 +202,7 @@ done:
 		      brw->urb.clip_start,
 		      brw->urb.sf_start,
 		      brw->urb.cs_start, 
-		      URB_SIZES(brw));
+		      brw->urb.size);
       
       brw->state.dirty.brw |= BRW_NEW_URB_FENCE;
    }
@@ -244,7 +246,7 @@ void brw_upload_urb_fence(struct brw_context *brw)
    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  = URB_SIZES(brw);
+   uf.bits1.cs_fence  = brw->urb.size;
 
    BRW_BATCH_STRUCT(brw, &uf);
 }
diff --git a/src/mesa/drivers/dri/i965/brw_util.c b/src/mesa/drivers/dri/i965/brw_util.c
index ce21aa4869..bba9249d1b 100644
--- a/src/mesa/drivers/dri/i965/brw_util.c
+++ b/src/mesa/drivers/dri/i965/brw_util.c
@@ -35,7 +35,7 @@
 #include "brw_util.h"
 #include "brw_defines.h"
 
-GLuint brw_count_bits( GLuint val )
+GLuint brw_count_bits(uint64_t val)
 {
    GLuint i;
    for (i = 0; val ; val >>= 1)
diff --git a/src/mesa/drivers/dri/i965/brw_util.h b/src/mesa/drivers/dri/i965/brw_util.h
index 33e7cd87e4..04f3175d3e 100644
--- a/src/mesa/drivers/dri/i965/brw_util.h
+++ b/src/mesa/drivers/dri/i965/brw_util.h
@@ -35,7 +35,7 @@
 
 #include "main/mtypes.h"
 
-extern GLuint brw_count_bits( GLuint val );
+extern GLuint brw_count_bits(uint64_t val);
 extern GLuint brw_parameter_list_state_flags(struct gl_program_parameter_list *paramList);
 extern GLuint brw_translate_blend_factor( GLenum factor );
 extern GLuint brw_translate_blend_equation( GLenum mode );
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index f0c79efbd9..fd055e225e 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -56,7 +56,7 @@ static void do_vs_prog( struct brw_context *brw,
    c.prog_data.inputs_read = vp->program.Base.InputsRead;
 
    if (c.key.copy_edgeflag) {
-      c.prog_data.outputs_written |= 1<<VERT_RESULT_EDGE;
+      c.prog_data.outputs_written |= BITFIELD64_BIT(VERT_RESULT_EDGE);
       c.prog_data.inputs_read |= 1<<VERT_ATTRIB_EDGEFLAG;
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 1638ef8111..1b84dd505f 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -67,6 +67,7 @@ static void release_tmps( struct brw_vs_compile *c )
  */
 static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 {
+   struct intel_context *intel = &c->func.brw->intel;
    GLuint i, reg = 0, mrf;
    int attributes_in_vue;
 
@@ -141,13 +142,13 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
    c->first_output = reg;
    c->first_overflow_output = 0;
 
-   if (BRW_IS_IGDNG(c->func.brw))
+   if (intel->is_ironlake)
        mrf = 8;
    else
        mrf = 4;
 
    for (i = 0; i < VERT_RESULT_MAX; i++) {
-      if (c->prog_data.outputs_written & (1 << i)) {
+      if (c->prog_data.outputs_written & BITFIELD64_BIT(i)) {
 	 c->nr_outputs++;
          assert(i < Elements(c->regs[PROGRAM_OUTPUT]));
 	 if (i == VERT_RESULT_HPOS) {
@@ -238,7 +239,7 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
     */
    attributes_in_vue = MAX2(c->nr_outputs, c->nr_inputs);
 
-   if (BRW_IS_IGDNG(c->func.brw))
+   if (intel->is_ironlake)
        c->prog_data.urb_entry_size = (attributes_in_vue + 6 + 3) / 4;
    else
        c->prog_data.urb_entry_size = (attributes_in_vue + 2 + 3) / 4;
@@ -331,63 +332,76 @@ static void unalias3( struct brw_vs_compile *c,
    }
 }
 
-static void emit_sop( struct brw_compile *p,
+static void emit_sop( struct brw_vs_compile *c,
                       struct brw_reg dst,
                       struct brw_reg arg0,
                       struct brw_reg arg1, 
 		      GLuint cond)
 {
+   struct brw_compile *p = &c->func;
+
    brw_MOV(p, dst, brw_imm_f(0.0f));
    brw_CMP(p, brw_null_reg(), cond, arg0, arg1);
    brw_MOV(p, dst, brw_imm_f(1.0f));
    brw_set_predicate_control_flag_value(p, 0xff);
 }
 
-static void emit_seq( struct brw_compile *p,
+static void emit_seq( struct brw_vs_compile *c,
                       struct brw_reg dst,
                       struct brw_reg arg0,
                       struct brw_reg arg1 )
 {
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_EQ);
+   emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_EQ);
 }
 
-static void emit_sne( struct brw_compile *p,
+static void emit_sne( struct brw_vs_compile *c,
                       struct brw_reg dst,
                       struct brw_reg arg0,
                       struct brw_reg arg1 )
 {
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_NEQ);
+   emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_NEQ);
 }
-static void emit_slt( struct brw_compile *p, 
+static void emit_slt( struct brw_vs_compile *c,
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_L);
+   emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_L);
 }
 
-static void emit_sle( struct brw_compile *p, 
+static void emit_sle( struct brw_vs_compile *c,
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_LE);
+   emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_LE);
 }
 
-static void emit_sgt( struct brw_compile *p, 
+static void emit_sgt( struct brw_vs_compile *c,
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-   emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_G);
+   emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_G);
 }
 
-static void emit_sge( struct brw_compile *p, 
+static void emit_sge( struct brw_vs_compile *c,
 		      struct brw_reg dst,
 		      struct brw_reg arg0,
 		      struct brw_reg arg1 )
 {
-  emit_sop(p, dst, arg0, arg1, BRW_CONDITIONAL_GE);
+  emit_sop(c, dst, arg0, arg1, BRW_CONDITIONAL_GE);
+}
+
+static void emit_cmp( struct brw_compile *p,
+		      struct brw_reg dst,
+		      struct brw_reg arg0,
+		      struct brw_reg arg1,
+		      struct brw_reg arg2 )
+{
+   brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, arg0, brw_imm_f(0));
+   brw_SEL(p, dst, arg1, arg2);
+   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
 }
 
 static void emit_max( struct brw_compile *p, 
@@ -912,6 +926,7 @@ get_src_reg( struct brw_vs_compile *c,
    case PROGRAM_CONSTANT:
    case PROGRAM_UNIFORM:
    case PROGRAM_ENV_PARAM:
+   case PROGRAM_LOCAL_PARAM:
       if (c->vp->use_const_buffer) {
          return get_constant(c, inst, argIndex);
       }
@@ -930,7 +945,6 @@ get_src_reg( struct brw_vs_compile *c,
       /* this is a normal case since we loop over all three src args */
       return brw_null_reg();
 
-   case PROGRAM_LOCAL_PARAM: 
    case PROGRAM_WRITE_ONLY:
    default:
       assert(0);
@@ -1100,6 +1114,8 @@ static void emit_swz( struct brw_vs_compile *c,
 static void emit_vertex_write( struct brw_vs_compile *c)
 {
    struct brw_compile *p = &c->func;
+   struct brw_context *brw = p->brw;
+   struct intel_context *intel = &brw->intel;
    struct brw_reg m0 = brw_message_reg(0);
    struct brw_reg pos = c->regs[PROGRAM_OUTPUT][VERT_RESULT_HPOS];
    struct brw_reg ndc;
@@ -1122,8 +1138,8 @@ static void emit_vertex_write( struct brw_vs_compile *c)
    /* Update the header for point size, user clipping flags, and -ve rhw
     * workaround.
     */
-   if ((c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) ||
-       c->key.nr_userclip || BRW_IS_965(p->brw))
+   if ((c->prog_data.outputs_written & BITFIELD64_BIT(VERT_RESULT_PSIZ)) ||
+       c->key.nr_userclip || brw->has_negative_rhw_bug)
    {
       struct brw_reg header1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
       GLuint i;
@@ -1132,7 +1148,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 
       brw_set_access_mode(p, BRW_ALIGN_16);	
 
-      if (c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) {
+      if (c->prog_data.outputs_written & BITFIELD64_BIT(VERT_RESULT_PSIZ)) {
 	 struct brw_reg psiz = c->regs[PROGRAM_OUTPUT][VERT_RESULT_PSIZ];
 	 brw_MUL(p, brw_writemask(header1, WRITEMASK_W), brw_swizzle1(psiz, 0), brw_imm_f(1<<11));
 	 brw_AND(p, brw_writemask(header1, WRITEMASK_W), header1, brw_imm_ud(0x7ff<<8));
@@ -1154,7 +1170,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
        * Later, clipping will detect ucp[6] and ensure the primitive is
        * clipped against all fixed planes.
        */
-      if (BRW_IS_965(p->brw)) {
+      if (brw->has_negative_rhw_bug) {
 	 brw_CMP(p,
 		 vec8(brw_null_reg()),
 		 BRW_CONDITIONAL_L,
@@ -1182,8 +1198,8 @@ static void emit_vertex_write( struct brw_vs_compile *c)
    brw_set_access_mode(p, BRW_ALIGN_1);
    brw_MOV(p, offset(m0, 2), ndc);
 
-   if (BRW_IS_IGDNG(p->brw)) {
-       /* There are 20 DWs (D0-D19) in VUE vertex header on IGDNG */
+   if (intel->is_ironlake) {
+       /* There are 20 DWs (D0-D19) in VUE vertex header on Ironlake */
        brw_MOV(p, offset(m0, 3), pos); /* a portion of vertex header */
        /* m4, m5 contain the distances from vertex to the user clip planeXXX. 
         * Seems it is useless for us.
@@ -1222,7 +1238,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
        */
       GLuint i, mrf = 0;
       for (i = c->first_overflow_output; i < VERT_RESULT_MAX; i++) {
-         if (c->prog_data.outputs_written & (1 << i)) {
+         if (c->prog_data.outputs_written & BITFIELD64_BIT(i)) {
             /* move from GRF to MRF */
             brw_MOV(p, brw_message_reg(4+mrf), c->regs[PROGRAM_OUTPUT][i]);
             mrf++;
@@ -1269,6 +1285,38 @@ post_vs_emit( struct brw_vs_compile *c,
    }
 }
 
+static GLboolean
+accumulator_contains(struct brw_vs_compile *c, struct brw_reg val)
+{
+   struct brw_compile *p = &c->func;
+   struct brw_instruction *prev_insn = &p->store[p->nr_insn - 1];
+
+   if (p->nr_insn == 0)
+      return GL_FALSE;
+
+   if (val.address_mode != BRW_ADDRESS_DIRECT)
+      return GL_FALSE;
+
+   switch (prev_insn->header.opcode) {
+   case BRW_OPCODE_MOV:
+   case BRW_OPCODE_MAC:
+   case BRW_OPCODE_MUL:
+      if (prev_insn->header.access_mode == BRW_ALIGN_16 &&
+	  prev_insn->header.execution_size == val.width &&
+	  prev_insn->bits1.da1.dest_reg_file == val.file &&
+	  prev_insn->bits1.da1.dest_reg_type == val.type &&
+	  prev_insn->bits1.da1.dest_address_mode == val.address_mode &&
+	  prev_insn->bits1.da1.dest_reg_nr == val.nr &&
+	  prev_insn->bits1.da16.dest_subreg_nr == val.subnr / 16 &&
+	  prev_insn->bits1.da16.dest_writemask == 0xf)
+	 return GL_TRUE;
+      else
+	 return GL_FALSE;
+   default:
+      return GL_FALSE;
+   }
+}
+
 static uint32_t
 get_predicate(const struct prog_instruction *inst)
 {
@@ -1314,6 +1362,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 #define MAX_LOOP_DEPTH 32
    struct brw_compile *p = &c->func;
    struct brw_context *brw = p->brw;
+   struct intel_context *intel = &brw->intel;
    const GLuint nr_insns = c->vp->program.Base.NumInstructions;
    GLuint insn, if_depth = 0, loop_depth = 0;
    GLuint end_offset = 0;
@@ -1447,9 +1496,13 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 unalias3(c, dst, args[0], args[1], args[2], emit_lrp_noalias);
 	 break;
       case OPCODE_MAD:
-	 brw_MOV(p, brw_acc_reg(), args[2]);
+	 if (!accumulator_contains(c, args[2]))
+	    brw_MOV(p, brw_acc_reg(), args[2]);
 	 brw_MAC(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_CMP:
+	 emit_cmp(p, dst, args[0], args[1], args[2]);
+	 break;
       case OPCODE_MAX:
 	 emit_max(p, dst, args[0], args[1]);
 	 break;
@@ -1473,25 +1526,25 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 break;
 
       case OPCODE_SEQ:
-         emit_seq(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_seq);
          break;
       case OPCODE_SIN:
 	 emit_math1(c, BRW_MATH_FUNCTION_SIN, dst, args[0], BRW_MATH_PRECISION_FULL);
 	 break;
       case OPCODE_SNE:
-         emit_sne(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_sne);
          break;
       case OPCODE_SGE:
-	 emit_sge(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_sge);
 	 break;
       case OPCODE_SGT:
-         emit_sgt(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_sgt);
          break;
       case OPCODE_SLT:
-	 emit_slt(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_slt);
 	 break;
       case OPCODE_SLE:
-         emit_sle(p, dst, args[0], args[1]);
+         unalias2(c, dst, args[0], args[1], emit_sle);
          break;
       case OPCODE_SUB:
 	 brw_ADD(p, dst, args[0], negate(args[1]));
@@ -1543,7 +1596,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 
             loop_depth--;
 
-	    if (BRW_IS_IGDNG(brw))
+	    if (intel->is_ironlake)
 	       br = 2;
 
             inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]);
diff --git a/src/mesa/drivers/dri/i965/brw_vs_state.c b/src/mesa/drivers/dri/i965/brw_vs_state.c
index d790ab6555..345ffa7ee1 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_state.c
@@ -82,9 +82,9 @@ vs_unit_populate_key(struct brw_context *brw, struct brw_vs_unit_key *key)
 static dri_bo *
 vs_unit_create_from_key(struct brw_context *brw, struct brw_vs_unit_key *key)
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_vs_unit_state vs;
    dri_bo *bo;
-   int chipset_max_threads;
 
    memset(&vs, 0, sizeof(vs));
 
@@ -98,7 +98,7 @@ vs_unit_create_from_key(struct brw_context *brw, struct brw_vs_unit_key *key)
     */
    vs.thread1.single_program_flow = 0;
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
       vs.thread1.binding_table_entry_count = 0; /* hardware requirement */
    else
       vs.thread1.binding_table_entry_count = key->nr_surfaces;
@@ -109,24 +109,44 @@ vs_unit_create_from_key(struct brw_context *brw, struct brw_vs_unit_key *key)
    vs.thread3.urb_entry_read_offset = 0;
    vs.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;
 
-   if (BRW_IS_IGDNG(brw))
-       vs.thread4.nr_urb_entries = key->nr_urb_entries >> 2;
-   else
-       vs.thread4.nr_urb_entries = key->nr_urb_entries;
+   if (intel->is_ironlake) {
+      switch (key->nr_urb_entries) {
+      case 8:
+      case 12:
+      case 16:
+      case 32:
+      case 64:
+      case 96:
+      case 128:
+      case 168:
+      case 192:
+      case 224:
+      case 256:
+	 vs.thread4.nr_urb_entries = key->nr_urb_entries >> 2;
+	 break;
+      default:
+	 assert(0);
+      }
+   } else {
+      switch (key->nr_urb_entries) {
+      case 8:
+      case 12:
+      case 16:
+      case 32:
+	 break;
+      case 64:
+	 assert(intel->is_g4x);
+	 break;
+      default:
+	 assert(0);
+      }
+      vs.thread4.nr_urb_entries = key->nr_urb_entries;
+   }
 
    vs.thread4.urb_entry_allocation_size = key->urb_size - 1;
 
-   if (BRW_IS_IGDNG(brw))
-      chipset_max_threads = 72;
-   else if (BRW_IS_G4X(brw))
-      chipset_max_threads = 32;
-   else
-      chipset_max_threads = 16;
    vs.thread4.max_threads = CLAMP(key->nr_urb_entries / 2,
-				  1, chipset_max_threads) - 1;
-
-   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
-      vs.thread4.max_threads = 0;
+				  1, brw->vs_max_threads) - 1;
 
    /* No samplers for ARB_vp programs:
     */
diff --git a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
index 89f47522a1..3bc9840a97 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
@@ -30,7 +30,6 @@
   */
 
 #include "main/mtypes.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "shader/prog_parameter.h"
 
@@ -53,6 +52,7 @@ brw_vs_update_constant_buffer(struct brw_context *brw)
    const struct gl_program_parameter_list *params = vp->program.Base.Parameters;
    const int size = params->NumParameters * 4 * sizeof(GLfloat);
    drm_intel_bo *const_buffer;
+   int i;
 
    /* BRW_NEW_VERTEX_PROGRAM */
    if (!vp->use_const_buffer)
@@ -62,7 +62,19 @@ brw_vs_update_constant_buffer(struct brw_context *brw)
 				     size, 64);
 
    /* _NEW_PROGRAM_CONSTANTS */
-   dri_bo_subdata(const_buffer, 0, size, params->ParameterValues);
+
+   /* Updates the ParamaterValues[i] pointers for all parameters of the
+    * basic type of PROGRAM_STATE_VAR.
+    */
+   _mesa_load_state_parameters(&brw->intel.ctx, vp->program.Base.Parameters);
+
+   intel_bo_map_gtt_preferred(intel, const_buffer, GL_TRUE);
+   for (i = 0; i < params->NumParameters; i++) {
+      memcpy(const_buffer->virtual + i * 4 * sizeof(float),
+	     params->ParameterValues[i],
+	     4 * sizeof(float));
+   }
+   intel_bo_unmap_gtt_preferred(intel, const_buffer);
 
    return const_buffer;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 124fde25fe..72749b3859 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -46,7 +46,7 @@
 #include "brw_state.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
-
+#include "brw_wm.h"
 
 static void
 dri_bo_release(dri_bo **bo)
@@ -66,8 +66,14 @@ static void brw_destroy_context( struct intel_context *intel )
 
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
-
-   _mesa_free(brw->wm.compile_data);
+   brw_clear_validated_bos(brw);
+   if (brw->wm.compile_data) {
+      _mesa_free(brw->wm.compile_data->instruction);
+      _mesa_free(brw->wm.compile_data->vreg);
+      _mesa_free(brw->wm.compile_data->refs);
+      _mesa_free(brw->wm.compile_data->prog_instructions);
+      _mesa_free(brw->wm.compile_data);
+   }
 
    for (i = 0; i < brw->state.nr_color_regions; i++)
       intel_region_release(&brw->state.color_regions[i]);
@@ -144,9 +150,6 @@ static void brw_new_batch( struct intel_context *intel )
 {
    struct brw_context *brw = brw_context(&intel->ctx);
 
-   /* Check that we didn't just wrap our batchbuffer at a bad time. */
-   assert(!brw->no_batch_wrap);
-
    brw->curbe.need_new_bo = GL_TRUE;
 
    /* Mark all context state as needing to be re-emitted.
@@ -175,20 +178,6 @@ static void brw_note_fence( struct intel_context *intel, GLuint fence )
    brw_context(&intel->ctx)->state.dirty.brw |= BRW_NEW_FENCE;
 }
 
-/* called from intelWaitForIdle() and intelFlush()
- *
- * For now, just flush everything.  Could be smarter later.
- */
-static GLuint brw_flush_cmd( void )
-{
-   struct brw_mi_flush flush;
-   flush.opcode = CMD_MI_FLUSH;
-   flush.pad = 0;
-   flush.flags = BRW_FLUSH_STATE_CACHE;
-   return *(GLuint *)&flush;
-}
-
-
 static void brw_invalidate_state( struct intel_context *intel, GLuint new_state )
 {
    /* nothing */
@@ -209,6 +198,5 @@ void brwInitVtbl( struct brw_context *brw )
    brw->intel.vtbl.finish_batch = brw_finish_batch;
    brw->intel.vtbl.destroy = brw_destroy_context;
    brw->intel.vtbl.set_draw_region = brw_set_draw_region;
-   brw->intel.vtbl.flush_cmd = brw_flush_cmd;
    brw->intel.vtbl.debug_batch = brw_debug_batch;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 2292de94c4..6895f64410 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -29,7 +29,6 @@
   *   Keith Whitwell <keith@tungstengraphics.com>
   */
              
-#include "main/texformat.h"
 #include "brw_context.h"
 #include "brw_util.h"
 #include "brw_wm.h"
@@ -153,8 +152,21 @@ static void do_wm_prog( struct brw_context *brw,
           */
          return;
       }
+      c->instruction = _mesa_calloc(BRW_WM_MAX_INSN * sizeof(*c->instruction));
+      c->prog_instructions = _mesa_calloc(BRW_WM_MAX_INSN *
+					  sizeof(*c->prog_instructions));
+      c->vreg = _mesa_calloc(BRW_WM_MAX_VREG * sizeof(*c->vreg));
+      c->refs = _mesa_calloc(BRW_WM_MAX_REF * sizeof(*c->refs));
    } else {
+      void *instruction = c->instruction;
+      void *prog_instructions = c->prog_instructions;
+      void *vreg = c->vreg;
+      void *refs = c->refs;
       memset(c, 0, sizeof(*brw->wm.compile_data));
+      c->instruction = instruction;
+      c->prog_instructions = prog_instructions;
+      c->vreg = vreg;
+      c->refs = refs;
    }
    memcpy(&c->key, key, sizeof(*key));
 
@@ -218,7 +230,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
        ctx->Color.AlphaEnabled)
       lookup |= IZ_PS_KILL_ALPHATEST_BIT;
 
-   if (fp->program.Base.OutputsWritten & (1<<FRAG_RESULT_DEPTH))
+   if (fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH))
       lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
 
    /* _NEW_DEPTH */
@@ -288,7 +300,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
          const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
 	 if (img->InternalFormat == GL_YCBCR_MESA) {
 	    key->yuvtex_mask |= 1 << i;
-	    if (img->TexFormat->MesaFormat == MESA_FORMAT_YCBCR)
+	    if (img->TexFormat == MESA_FORMAT_YCBCR)
 		key->yuvtex_swap_mask |= 1 << i;
 	 }
 
@@ -309,6 +321,9 @@ static void brw_wm_populate_key( struct brw_context *brw,
     * from the incoming screen origin relative position we get as part of our
     * payload.
     *
+    * This is only needed for the WM_WPOSXY opcode when the fragment program
+    * uses the gl_FragCoord input.
+    *
     * We could avoid recompiling by including this as a constant referenced by
     * our program, but if we were to do that it would also be nice to handle
     * getting that constant updated at batchbuffer submit time (when we
@@ -317,17 +332,21 @@ static void brw_wm_populate_key( struct brw_context *brw,
     * just avoid using this as key data if the program doesn't use
     * fragment.position.
     *
-    * This pretty much becomes moot with DRI2 and redirected buffers anyway,
-    * as our origins will always be zero then.
+    * For DRI2 the origin_x/y will always be (0,0) but we still need the
+    * drawable height in order to invert the Y axis.
     */
-   if (brw->intel.driDrawable != NULL) {
-      key->origin_x = brw->intel.driDrawable->x;
-      key->origin_y = brw->intel.driDrawable->y;
-      key->drawable_height = brw->intel.driDrawable->h;
+   if (fp->program.Base.InputsRead & FRAG_BIT_WPOS) {
+      if (brw->intel.driDrawable != NULL) {
+         key->origin_x = brw->intel.driDrawable->x;
+         key->origin_y = brw->intel.driDrawable->y;
+         key->drawable_height = brw->intel.driDrawable->h;
+      }
    }
 
+   key->nr_color_regions = brw->state.nr_color_regions;
+
    /* CACHE_NEW_VS_PROG */
-   key->vp_outputs_written = brw->vs.prog_data->outputs_written & DO_SETUP_BITS;
+   key->vp_outputs_written = brw->vs.prog_data->outputs_written;
 
    /* The unique fragment program ID */
    key->program_string_id = fp->id;
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index 872b1f3ecf..b9b987ea70 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -67,18 +67,19 @@ struct brw_wm_prog_key {
    GLuint flat_shade:1;
    GLuint linear_color:1;  /**< linear interpolation vs perspective interp */
    GLuint runtime_check_aads_emit:1;
+   GLuint nr_color_regions:2;
    
    GLbitfield proj_attrib_mask; /**< one bit per fragment program attribute */
    GLuint shadowtex_mask:16;
    GLuint yuvtex_mask:16;
    GLuint yuvtex_swap_mask:16;	/* UV swaped */
 
-   GLuint tex_swizzles[BRW_MAX_TEX_UNIT];
+   GLushort tex_swizzles[BRW_MAX_TEX_UNIT];
 
+   GLushort origin_x, origin_y;
+   GLushort drawable_height;
+   GLbitfield64 vp_outputs_written;
    GLuint program_string_id:32;
-   GLuint origin_x, origin_y;
-   GLuint drawable_height;
-   GLuint vp_outputs_written;
 };
 
 
@@ -153,15 +154,16 @@ struct brw_wm_instruction {
 };
 
 
-#define BRW_WM_MAX_INSN  (MAX_NV_FRAGMENT_PROGRAM_INSTRUCTIONS*3 + FRAG_ATTRIB_MAX + 3)
+#define BRW_WM_MAX_INSN  (MAX_PROGRAM_INSTRUCTIONS*3 + FRAG_ATTRIB_MAX + 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 BRW_WM_MAX_SUBROUTINE 16
 
+/* used in masks next to WRITEMASK_*. */
+#define SATURATE (1<<5)
 
 
 /* New opcodes to track internal operations required for WM unit.
@@ -200,7 +202,7 @@ struct brw_wm_compile {
     * simplifying and adding instructions for interpolation and
     * framebuffer writes.
     */
-   struct prog_instruction prog_instructions[BRW_WM_MAX_INSN];
+   struct prog_instruction *prog_instructions;
    GLuint nr_fp_insns;
    GLuint fp_temp;
    GLuint fp_interp_emitted;
@@ -211,7 +213,7 @@ struct brw_wm_compile {
    struct prog_src_register pixel_w;
 
 
-   struct brw_wm_value vreg[BRW_WM_MAX_VREG];
+   struct brw_wm_value *vreg;
    GLuint nr_vreg;
 
    struct brw_wm_value creg[BRW_WM_MAX_PARAM];
@@ -228,10 +230,10 @@ struct brw_wm_compile {
    struct brw_wm_ref undef_ref;
    struct brw_wm_value undef_value;
 
-   struct brw_wm_ref refs[BRW_WM_MAX_REF];
+   struct brw_wm_ref *refs;
    GLuint nr_refs;
 
-   struct brw_wm_instruction instruction[BRW_WM_MAX_INSN];
+   struct brw_wm_instruction *instruction;
    GLuint nr_insns;
 
    struct brw_wm_constref constref[BRW_WM_MAX_CONST];
@@ -271,6 +273,12 @@ struct brw_wm_compile {
 };
 
 
+/** Bits for prog_instruction::Aux field */
+#define INST_AUX_EOT      0x1
+#define INST_AUX_TARGET(T)  (T << 1)
+#define INST_AUX_GET_TARGET(AUX) ((AUX) >> 1)
+
+
 GLuint brw_wm_nr_args( GLuint opcode );
 GLuint brw_wm_is_scalar_result( GLuint opcode );
 
@@ -300,10 +308,141 @@ 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);
 
+/* brw_wm_emit.c */
+void emit_alu1(struct brw_compile *p,
+	       struct brw_instruction *(*func)(struct brw_compile *,
+					       struct brw_reg,
+					       struct brw_reg),
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       const struct brw_reg *arg0);
+void emit_alu2(struct brw_compile *p,
+	       struct brw_instruction *(*func)(struct brw_compile *,
+					       struct brw_reg,
+					       struct brw_reg,
+					       struct brw_reg),
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       const struct brw_reg *arg0,
+	       const struct brw_reg *arg1);
+void emit_cinterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0);
 void emit_ddxy(struct brw_compile *p,
 	       const struct brw_reg *dst,
 	       GLuint mask,
 	       GLboolean is_ddx,
 	       const struct brw_reg *arg0);
+void emit_delta_xy(struct brw_compile *p,
+		   const struct brw_reg *dst,
+		   GLuint mask,
+		   const struct brw_reg *arg0);
+void emit_dp3(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_dp4(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_dph(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_fb_write(struct brw_wm_compile *c,
+		   struct brw_reg *arg0,
+		   struct brw_reg *arg1,
+		   struct brw_reg *arg2,
+		   GLuint target,
+		   GLuint eot);
+void emit_frontfacing(struct brw_compile *p,
+		      const struct brw_reg *dst,
+		      GLuint mask);
+void emit_linterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas);
+void emit_lrp(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1,
+	      const struct brw_reg *arg2);
+void emit_mad(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1,
+	      const struct brw_reg *arg2);
+void emit_math1(struct brw_wm_compile *c,
+		GLuint function,
+		const struct brw_reg *dst,
+		GLuint mask,
+		const struct brw_reg *arg0);
+void emit_math2(struct brw_wm_compile *c,
+		GLuint function,
+		const struct brw_reg *dst,
+		GLuint mask,
+		const struct brw_reg *arg0,
+		const struct brw_reg *arg1);
+void emit_min(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_max(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_pinterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas,
+		  const struct brw_reg *w);
+void emit_pixel_xy(struct brw_wm_compile *c,
+		   const struct brw_reg *dst,
+		   GLuint mask);
+void emit_pixel_w(struct brw_wm_compile *c,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas);
+void emit_sop(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      GLuint cond,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
+void emit_tex(struct brw_wm_compile *c,
+	      struct brw_reg *dst,
+	      GLuint dst_flags,
+	      struct brw_reg *arg,
+	      struct brw_reg depth_payload,
+	      GLuint tex_idx,
+	      GLuint sampler,
+	      GLboolean shadow);
+void emit_txb(struct brw_wm_compile *c,
+	      struct brw_reg *dst,
+	      GLuint dst_flags,
+	      struct brw_reg *arg,
+	      struct brw_reg depth_payload,
+	      GLuint tex_idx,
+	      GLuint sampler);
+void emit_wpos_xy(struct brw_wm_compile *c,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0);
+void emit_xpd(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1);
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index bf80a2942a..cc1052f757 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -44,6 +44,7 @@ static INLINE struct brw_reg sechalf( struct brw_reg reg )
    return reg;
 }
 
+
 /* Payload R0:
  *
  * R0.0 -- pixel mask, one bit for each of 4 pixels in 4 tiles,
@@ -60,42 +61,50 @@ static INLINE struct brw_reg sechalf( struct brw_reg reg )
  * R1.8 -- ?
  */
 
-
-static void emit_pixel_xy(struct brw_compile *p,
-			  const struct brw_reg *dst,
-			  GLuint mask)
+void emit_pixel_xy(struct brw_wm_compile *c,
+		   const struct brw_reg *dst,
+		   GLuint mask)
 {
+   struct brw_compile *p = &c->func;
    struct brw_reg r1 = brw_vec1_grf(1, 0);
    struct brw_reg r1_uw = retype(r1, BRW_REGISTER_TYPE_UW);
+   struct brw_reg dst0_uw, dst1_uw;
 
+   brw_push_insn_state(p);
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
 
+   if (c->dispatch_width == 16) {
+      dst0_uw = vec16(retype(dst[0], BRW_REGISTER_TYPE_UW));
+      dst1_uw = vec16(retype(dst[1], BRW_REGISTER_TYPE_UW));
+   } else {
+      dst0_uw = vec8(retype(dst[0], BRW_REGISTER_TYPE_UW));
+      dst1_uw = vec8(retype(dst[1], BRW_REGISTER_TYPE_UW));
+   }
+
    /* Calculate pixel centers by adding 1 or 0 to each of the
     * micro-tile coordinates passed in r1.
     */
    if (mask & WRITEMASK_X) {
       brw_ADD(p,
-	      vec16(retype(dst[0], BRW_REGISTER_TYPE_UW)),
+	      dst0_uw,
 	      stride(suboffset(r1_uw, 4), 2, 4, 0),
 	      brw_imm_v(0x10101010));
    }
 
    if (mask & WRITEMASK_Y) {
       brw_ADD(p,
-	      vec16(retype(dst[1], BRW_REGISTER_TYPE_UW)),
+	      dst1_uw,
 	      stride(suboffset(r1_uw,5), 2, 4, 0),
 	      brw_imm_v(0x11001100));
    }
-
-   brw_set_compression_control(p, BRW_COMPRESSION_COMPRESSED);
+   brw_pop_insn_state(p);
 }
 
 
-
-static void emit_delta_xy(struct brw_compile *p,
-			  const struct brw_reg *dst,
-			  GLuint mask,
-			  const struct brw_reg *arg0)
+void emit_delta_xy(struct brw_compile *p,
+		   const struct brw_reg *dst,
+		   GLuint mask,
+		   const struct brw_reg *arg0)
 {
    struct brw_reg r1 = brw_vec1_grf(1, 0);
 
@@ -118,10 +127,10 @@ static void emit_delta_xy(struct brw_compile *p,
    }
 }
 
-static void emit_wpos_xy(struct brw_wm_compile *c,
-			 const struct brw_reg *dst,
-			 GLuint mask,
-			 const struct brw_reg *arg0)
+void emit_wpos_xy(struct brw_wm_compile *c,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0)
 {
    struct brw_compile *p = &c->func;
 
@@ -146,12 +155,14 @@ static void emit_wpos_xy(struct brw_wm_compile *c,
 }
 
 
-static void emit_pixel_w( struct brw_compile *p,
-			  const struct brw_reg *dst,
-			  GLuint mask,
-			  const struct brw_reg *arg0,
-			  const struct brw_reg *deltas)
+void emit_pixel_w(struct brw_wm_compile *c,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas)
 {
+   struct brw_compile *p = &c->func;
+
    /* Don't need this if all you are doing is interpolating color, for
     * instance.
     */
@@ -165,21 +176,29 @@ static void emit_pixel_w( struct brw_compile *p,
       brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), deltas[1]);
 
       /* Calc w */
-      brw_math_16( p, dst[3],
-		   BRW_MATH_FUNCTION_INV,
-		   BRW_MATH_SATURATE_NONE,
-		   2, brw_null_reg(),
-		   BRW_MATH_PRECISION_FULL);
+      if (c->dispatch_width == 16) {
+	 brw_math_16(p, dst[3],
+		     BRW_MATH_FUNCTION_INV,
+		     BRW_MATH_SATURATE_NONE,
+		     2, brw_null_reg(),
+		     BRW_MATH_PRECISION_FULL);
+      } else {
+	 brw_math(p, dst[3],
+		  BRW_MATH_FUNCTION_INV,
+		  BRW_MATH_SATURATE_NONE,
+		  2, brw_null_reg(),
+		  BRW_MATH_DATA_VECTOR,
+		  BRW_MATH_PRECISION_FULL);
+      }
    }
 }
 
 
-
-static void emit_linterp( struct brw_compile *p, 
-			 const struct brw_reg *dst,
-			 GLuint mask,
-			 const struct brw_reg *arg0,
-			 const struct brw_reg *deltas )
+void emit_linterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas)
 {
    struct brw_reg interp[4];
    GLuint nr = arg0[0].nr;
@@ -199,12 +218,12 @@ static void emit_linterp( struct brw_compile *p,
 }
 
 
-static void emit_pinterp( struct brw_compile *p, 
-			  const struct brw_reg *dst,
-			  GLuint mask,
-			  const struct brw_reg *arg0,
-			  const struct brw_reg *deltas,
-			  const struct brw_reg *w)
+void emit_pinterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0,
+		  const struct brw_reg *deltas,
+		  const struct brw_reg *w)
 {
    struct brw_reg interp[4];
    GLuint nr = arg0[0].nr;
@@ -229,10 +248,10 @@ static void emit_pinterp( struct brw_compile *p,
 }
 
 
-static void emit_cinterp( struct brw_compile *p, 
-			 const struct brw_reg *dst,
-			 GLuint mask,
-			 const struct brw_reg *arg0 )
+void emit_cinterp(struct brw_compile *p,
+		  const struct brw_reg *dst,
+		  GLuint mask,
+		  const struct brw_reg *arg0)
 {
    struct brw_reg interp[4];
    GLuint nr = arg0[0].nr;
@@ -251,9 +270,9 @@ static void emit_cinterp( struct brw_compile *p,
 }
 
 /* Sets the destination channels to 1.0 or 0.0 according to glFrontFacing. */
-static void emit_frontfacing( struct brw_compile *p,
-			      const struct brw_reg *dst,
-			      GLuint mask )
+void emit_frontfacing(struct brw_compile *p,
+		      const struct brw_reg *dst,
+		      GLuint mask)
 {
    struct brw_reg r1_6ud = retype(brw_vec1_grf(1, 6), BRW_REGISTER_TYPE_UD);
    GLuint i;
@@ -352,13 +371,13 @@ void emit_ddxy(struct brw_compile *p,
       brw_set_saturate(p, 0);
 }
 
-static void emit_alu1( struct brw_compile *p, 
-		       struct brw_instruction *(*func)(struct brw_compile *, 
-						       struct brw_reg, 
-						       struct brw_reg),
-		       const struct brw_reg *dst,
-		       GLuint mask,
-		       const struct brw_reg *arg0 )
+void emit_alu1(struct brw_compile *p,
+	       struct brw_instruction *(*func)(struct brw_compile *,
+					       struct brw_reg,
+					       struct brw_reg),
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       const struct brw_reg *arg0)
 {
    GLuint i;
 
@@ -376,15 +395,15 @@ static void emit_alu1( struct brw_compile *p,
 }
 
 
-static void emit_alu2( struct brw_compile *p, 
-		       struct brw_instruction *(*func)(struct brw_compile *, 
-						       struct brw_reg, 
-						       struct brw_reg, 
-						       struct brw_reg),
-		       const struct brw_reg *dst,
-		       GLuint mask,
-		       const struct brw_reg *arg0,
-		       const struct brw_reg *arg1 )
+void emit_alu2(struct brw_compile *p,
+	       struct brw_instruction *(*func)(struct brw_compile *,
+					       struct brw_reg,
+					       struct brw_reg,
+					       struct brw_reg),
+	       const struct brw_reg *dst,
+	       GLuint mask,
+	       const struct brw_reg *arg0,
+	       const struct brw_reg *arg1)
 {
    GLuint i;
 
@@ -402,12 +421,12 @@ static void emit_alu2( struct brw_compile *p,
 }
 
 
-static void emit_mad( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1,
-		      const struct brw_reg *arg2 )
+void emit_mad(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1,
+	      const struct brw_reg *arg2)
 {
    GLuint i;
 
@@ -422,26 +441,12 @@ static void emit_mad( struct brw_compile *p,
    }
 }
 
-static void emit_trunc( struct brw_compile *p,
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0)
-{
-   GLuint i;
-
-   for (i = 0; i < 4; i++) {
-      if (mask & (1<<i)) {
-	 brw_RNDZ(p, dst[i], arg0[i]);
-      }
-   }
-}
-
-static void emit_lrp( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1,
-		      const struct brw_reg *arg2 )
+void emit_lrp(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1,
+	      const struct brw_reg *arg2)
 {
    GLuint i;
 
@@ -461,21 +466,24 @@ static void emit_lrp( struct brw_compile *p,
    }
 }
 
-static void emit_sop( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      GLuint cond,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_sop(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      GLuint cond,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    GLuint i;
 
    for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {	
-	 brw_MOV(p, dst[i], brw_imm_f(0));
+	 brw_push_insn_state(p);
 	 brw_CMP(p, brw_null_reg(), cond, arg0[i], arg1[i]);
+	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	 brw_MOV(p, dst[i], brw_imm_f(0));
+	 brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
 	 brw_MOV(p, dst[i], brw_imm_f(1.0));
-	 brw_set_predicate_control_flag_value(p, 0xff);
+	 brw_pop_insn_state(p);
       }
    }
 }
@@ -559,11 +567,11 @@ static void emit_cmp( struct brw_compile *p,
    }
 }
 
-static void emit_max( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_max(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    GLuint i;
 
@@ -583,11 +591,11 @@ static void emit_max( struct brw_compile *p,
    }
 }
 
-static void emit_min( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_min(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    GLuint i;
 
@@ -608,11 +616,11 @@ static void emit_min( struct brw_compile *p,
 }
 
 
-static void emit_dp3( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_dp3(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
 
@@ -630,11 +638,11 @@ static void emit_dp3( struct brw_compile *p,
 }
 
 
-static void emit_dp4( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_dp4(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
 
@@ -653,11 +661,11 @@ static void emit_dp4( struct brw_compile *p,
 }
 
 
-static void emit_dph( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_dph(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    const int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
 
@@ -676,11 +684,11 @@ static void emit_dph( struct brw_compile *p,
 }
 
 
-static void emit_xpd( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0,
-		      const struct brw_reg *arg1 )
+void emit_xpd(struct brw_compile *p,
+	      const struct brw_reg *dst,
+	      GLuint mask,
+	      const struct brw_reg *arg0,
+	      const struct brw_reg *arg1)
 {
    GLuint i;
 
@@ -701,41 +709,68 @@ static void emit_xpd( struct brw_compile *p,
 }
 
 
-static void emit_math1( struct brw_compile *p, 
-			GLuint function,
-			const struct brw_reg *dst,
-			GLuint mask,
-			const struct brw_reg *arg0 )
+void emit_math1(struct brw_wm_compile *c,
+		GLuint function,
+		const struct brw_reg *dst,
+		GLuint mask,
+		const struct brw_reg *arg0)
 {
+   struct brw_compile *p = &c->func;
    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
+   GLuint saturate = ((mask & SATURATE) ?
+		      BRW_MATH_SATURATE_SATURATE :
+		      BRW_MATH_SATURATE_NONE);
 
    if (!(mask & WRITEMASK_XYZW))
       return; /* Do not emit dead code */
 
    assert(is_power_of_two(mask & WRITEMASK_XYZW));
 
+   /* If compressed, this will write message reg 2,3 from arg0.x's 16
+    * channels.
+    */
    brw_MOV(p, brw_message_reg(2), arg0[0]);
 
    /* Send two messages to perform all 16 operations:
     */
-   brw_math_16(p, 
-	       dst[dst_chan],
+   brw_push_insn_state(p);
+   brw_set_compression_control(p, BRW_COMPRESSION_NONE);
+   brw_math(p,
+	    dst[dst_chan],
+	    function,
+	    saturate,
+	    2,
+	    brw_null_reg(),
+	    BRW_MATH_DATA_VECTOR,
+	    BRW_MATH_PRECISION_FULL);
+
+   if (c->dispatch_width == 16) {
+      brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
+      brw_math(p,
+	       offset(dst[dst_chan],1),
 	       function,
-	       (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
-	       2,
+	       saturate,
+	       3,
 	       brw_null_reg(),
+	       BRW_MATH_DATA_VECTOR,
 	       BRW_MATH_PRECISION_FULL);
+   }
+   brw_pop_insn_state(p);
 }
 
 
-static void emit_math2( struct brw_compile *p, 
-			GLuint function,
-			const struct brw_reg *dst,
-			GLuint mask,
-			const struct brw_reg *arg0,
-			const struct brw_reg *arg1)
+void emit_math2(struct brw_wm_compile *c,
+		GLuint function,
+		const struct brw_reg *dst,
+		GLuint mask,
+		const struct brw_reg *arg0,
+		const struct brw_reg *arg1)
 {
+   struct brw_compile *p = &c->func;
    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
+   GLuint saturate = ((mask & SATURATE) ?
+		      BRW_MATH_SATURATE_SATURATE :
+		      BRW_MATH_SATURATE_NONE);
 
    if (!(mask & WRITEMASK_XYZW))
       return; /* Do not emit dead code */
@@ -746,173 +781,233 @@ static void emit_math2( struct brw_compile *p,
 
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
    brw_MOV(p, brw_message_reg(2), arg0[0]);
-   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
-   brw_MOV(p, brw_message_reg(4), sechalf(arg0[0]));
+   if (c->dispatch_width == 16) {
+      brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
+      brw_MOV(p, brw_message_reg(4), sechalf(arg0[0]));
+   }
 
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
    brw_MOV(p, brw_message_reg(3), arg1[0]);
-   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
-   brw_MOV(p, brw_message_reg(5), sechalf(arg1[0]));
+   if (c->dispatch_width == 16) {
+      brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
+      brw_MOV(p, brw_message_reg(5), sechalf(arg1[0]));
+   }
 
-   
-   /* Send two messages to perform all 16 operations:
-    */
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
    brw_math(p, 
 	    dst[dst_chan],
 	    function,
-	    (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
+	    saturate,
 	    2,
 	    brw_null_reg(),
 	    BRW_MATH_DATA_VECTOR,
 	    BRW_MATH_PRECISION_FULL);
 
-   brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
-   brw_math(p, 
-	    offset(dst[dst_chan],1),
-	    function,
-	    (mask & SATURATE) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
-	    4,
-	    brw_null_reg(),
-	    BRW_MATH_DATA_VECTOR,
-	    BRW_MATH_PRECISION_FULL);
-   
+   /* Send two messages to perform all 16 operations:
+    */
+   if (c->dispatch_width == 16) {
+      brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
+      brw_math(p,
+	       offset(dst[dst_chan],1),
+	       function,
+	       saturate,
+	       4,
+	       brw_null_reg(),
+	       BRW_MATH_DATA_VECTOR,
+	       BRW_MATH_PRECISION_FULL);
+   }
    brw_pop_insn_state(p);
 }
-		     
 
 
-static void emit_tex( struct brw_wm_compile *c,
-		      const struct brw_wm_instruction *inst,
-		      struct brw_reg *dst,
-		      GLuint dst_flags,
-		      struct brw_reg *arg )
+void emit_tex(struct brw_wm_compile *c,
+	      struct brw_reg *dst,
+	      GLuint dst_flags,
+	      struct brw_reg *arg,
+	      struct brw_reg depth_payload,
+	      GLuint tex_idx,
+	      GLuint sampler,
+	      GLboolean shadow)
 {
    struct brw_compile *p = &c->func;
-   GLuint msgLength, responseLength;
-   GLuint i, nr;
+   struct intel_context *intel = &p->brw->intel;
+   struct brw_reg dst_retyped;
+   GLuint cur_mrf = 2, response_length;
+   GLuint i, nr_texcoords;
    GLuint emit;
    GLuint msg_type;
+   GLuint mrf_per_channel;
+   GLuint simd_mode;
+
+   if (c->dispatch_width == 16) {
+      mrf_per_channel = 2;
+      response_length = 8;
+      dst_retyped = retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW);
+      simd_mode = BRW_SAMPLER_SIMD_MODE_SIMD16;
+   } else {
+      mrf_per_channel = 1;
+      response_length = 4;
+      dst_retyped = retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW);
+      simd_mode = BRW_SAMPLER_SIMD_MODE_SIMD8;
+   }
 
    /* How many input regs are there?
     */
-   switch (inst->tex_idx) {
+   switch (tex_idx) {
    case TEXTURE_1D_INDEX:
       emit = WRITEMASK_X;
-      nr = 1;
+      nr_texcoords = 1;
       break;
    case TEXTURE_2D_INDEX:
    case TEXTURE_RECT_INDEX:
       emit = WRITEMASK_XY;
-      nr = 2;
+      nr_texcoords = 2;
       break;
    case TEXTURE_3D_INDEX:
    case TEXTURE_CUBE_INDEX:
       emit = WRITEMASK_XYZ;
-      nr = 3;
+      nr_texcoords = 3;
       break;
    default:
       /* unexpected target */
       abort();
    }
 
-   if (inst->tex_shadow) {
-      nr = 4;
-      emit |= WRITEMASK_W;
-   }
+   /* Pre-Ironlake, the 8-wide sampler always took u,v,r. */
+   if (!intel->is_ironlake && c->dispatch_width == 8)
+      nr_texcoords = 3;
 
-   msgLength = 1;
+   /* For shadow comparisons, we have to supply u,v,r. */
+   if (shadow)
+      nr_texcoords = 3;
 
-   for (i = 0; i < nr; i++) {
-      static const GLuint swz[4] = {0,1,2,2};
-      if (emit & (1<<i)) 
-	 brw_MOV(p, brw_message_reg(msgLength+1), arg[swz[i]]);
+   /* Emit the texcoords. */
+   for (i = 0; i < nr_texcoords; i++) {
+      if (emit & (1<<i))
+	 brw_MOV(p, brw_message_reg(cur_mrf), arg[i]);
       else
-	 brw_MOV(p, brw_message_reg(msgLength+1), brw_imm_f(0));
-      msgLength += 2;
+	 brw_MOV(p, brw_message_reg(cur_mrf), brw_imm_f(0));
+      cur_mrf += mrf_per_channel;
    }
 
-   responseLength = 8;		/* always */
+   /* Fill in the shadow comparison reference value. */
+   if (shadow) {
+      if (intel->is_ironlake) {
+	 /* Fill in the cube map array index value. */
+	 brw_MOV(p, brw_message_reg(cur_mrf), brw_imm_f(0));
+	 cur_mrf += mrf_per_channel;
+      } else if (c->dispatch_width == 8) {
+	 /* Fill in the LOD bias value. */
+	 brw_MOV(p, brw_message_reg(cur_mrf), brw_imm_f(0));
+	 cur_mrf += mrf_per_channel;
+      }
+      brw_MOV(p, brw_message_reg(cur_mrf), arg[2]);
+      cur_mrf += mrf_per_channel;
+   }
 
-   if (BRW_IS_IGDNG(p->brw)) {
-       if (inst->tex_shadow)
-           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE_IGDNG;
-       else
-           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_IGDNG;
+   if (intel->is_ironlake) {
+      if (shadow)
+	 msg_type = BRW_SAMPLER_MESSAGE_SAMPLE_COMPARE_IGDNG;
+      else
+	 msg_type = BRW_SAMPLER_MESSAGE_SAMPLE_IGDNG;
    } else {
-       if (inst->tex_shadow)
-           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE;
-       else
-           msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE;
+      /* Note that G45 and older determines shadow compare and dispatch width
+       * from message length for most messages.
+       */
+      if (c->dispatch_width == 16 && shadow)
+	 msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_COMPARE;
+      else
+	 msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE;
    }
 
-   brw_SAMPLE(p, 
-	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
+   brw_SAMPLE(p,
+	      dst_retyped,
 	      1,
-	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
-              SURF_INDEX_TEXTURE(inst->tex_unit),
-	      inst->tex_unit,	  /* sampler */
-	      inst->writemask,
-	      msg_type, 
-	      responseLength,
-	      msgLength,
-	      0,	
+	      retype(depth_payload, BRW_REGISTER_TYPE_UW),
+              SURF_INDEX_TEXTURE(sampler),
+	      sampler,
+	      dst_flags & WRITEMASK_XYZW,
+	      msg_type,
+	      response_length,
+	      cur_mrf - 1,
+	      0,
 	      1,
-	      BRW_SAMPLER_SIMD_MODE_SIMD16);	
+	      simd_mode);
 }
 
 
-static void emit_txb( struct brw_wm_compile *c,
-		      const struct brw_wm_instruction *inst,
-		      struct brw_reg *dst,
-		      GLuint dst_flags,
-		      struct brw_reg *arg )
+void emit_txb(struct brw_wm_compile *c,
+	      struct brw_reg *dst,
+	      GLuint dst_flags,
+	      struct brw_reg *arg,
+	      struct brw_reg depth_payload,
+	      GLuint tex_idx,
+	      GLuint sampler)
 {
    struct brw_compile *p = &c->func;
+   struct intel_context *intel = &p->brw->intel;
    GLuint msgLength;
    GLuint msg_type;
-   /* Shadow ignored for txb.
+   GLuint mrf_per_channel;
+   GLuint response_length;
+   struct brw_reg dst_retyped;
+
+   /* The G45 and older chipsets don't support 8-wide dispatch for LOD biased
+    * samples, so we'll use the 16-wide instruction, leave the second halves
+    * undefined, and trust the execution mask to keep the undefined pixels
+    * from mattering.
     */
-   switch (inst->tex_idx) {
+   if (c->dispatch_width == 16 || !intel->is_ironlake) {
+      if (intel->is_ironlake)
+	 msg_type = BRW_SAMPLER_MESSAGE_SAMPLE_BIAS_IGDNG;
+      else
+	 msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS;
+      mrf_per_channel = 2;
+      dst_retyped = retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW);
+      response_length = 8;
+   } else {
+      msg_type = BRW_SAMPLER_MESSAGE_SAMPLE_BIAS_IGDNG;
+      mrf_per_channel = 1;
+      dst_retyped = retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW);
+      response_length = 4;
+   }
+
+   /* Shadow ignored for txb. */
+   switch (tex_idx) {
    case TEXTURE_1D_INDEX:
-      brw_MOV(p, brw_message_reg(2), arg[0]);
-      brw_MOV(p, brw_message_reg(4), brw_imm_f(0));
-      brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
+      brw_MOV(p, brw_message_reg(2 + 0 * mrf_per_channel), arg[0]);
+      brw_MOV(p, brw_message_reg(2 + 1 * mrf_per_channel), brw_imm_f(0));
+      brw_MOV(p, brw_message_reg(2 + 2 * mrf_per_channel), brw_imm_f(0));
       break;
    case TEXTURE_2D_INDEX:
    case TEXTURE_RECT_INDEX:
-      brw_MOV(p, brw_message_reg(2), arg[0]);
-      brw_MOV(p, brw_message_reg(4), arg[1]);
-      brw_MOV(p, brw_message_reg(6), brw_imm_f(0));
+      brw_MOV(p, brw_message_reg(2 + 0 * mrf_per_channel), arg[0]);
+      brw_MOV(p, brw_message_reg(2 + 1 * mrf_per_channel), arg[1]);
+      brw_MOV(p, brw_message_reg(2 + 2 * mrf_per_channel), brw_imm_f(0));
       break;
    case TEXTURE_3D_INDEX:
    case TEXTURE_CUBE_INDEX:
-      brw_MOV(p, brw_message_reg(2), arg[0]);
-      brw_MOV(p, brw_message_reg(4), arg[1]);
-      brw_MOV(p, brw_message_reg(6), arg[2]);
+      brw_MOV(p, brw_message_reg(2 + 0 * mrf_per_channel), arg[0]);
+      brw_MOV(p, brw_message_reg(2 + 1 * mrf_per_channel), arg[1]);
+      brw_MOV(p, brw_message_reg(2 + 2 * mrf_per_channel), arg[2]);
       break;
    default:
       /* unexpected target */
       abort();
    }
 
-   brw_MOV(p, brw_message_reg(8), arg[3]);
-   msgLength = 9;
-
-   if (BRW_IS_IGDNG(p->brw))
-       msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS_IGDNG;
-   else
-       msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS;
+   brw_MOV(p, brw_message_reg(2 + 3 * mrf_per_channel), arg[3]);
+   msgLength = 2 + 4 * mrf_per_channel - 1;
 
    brw_SAMPLE(p, 
-	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
+	      dst_retyped,
 	      1,
-	      retype(c->payload.depth[0].hw_reg, BRW_REGISTER_TYPE_UW),
-              SURF_INDEX_TEXTURE(inst->tex_unit),
-	      inst->tex_unit,	  /* sampler */
-	      inst->writemask,
+	      retype(depth_payload, BRW_REGISTER_TYPE_UW),
+              SURF_INDEX_TEXTURE(sampler),
+	      sampler,
+	      dst_flags & WRITEMASK_XYZW,
 	      msg_type,
-	      8,		/* responseLength */
+	      response_length,
 	      msgLength,
 	      0,	
 	      1,
@@ -920,11 +1015,13 @@ static void emit_txb( struct brw_wm_compile *c,
 }
 
 
-static void emit_lit( struct brw_compile *p, 
-		      const struct brw_reg *dst,
-		      GLuint mask,
-		      const struct brw_reg *arg0 )
+static void emit_lit(struct brw_wm_compile *c,
+		     const struct brw_reg *dst,
+		     GLuint mask,
+		     const struct brw_reg *arg0)
 {
+   struct brw_compile *p = &c->func;
+
    assert((mask & WRITEMASK_XW) == 0);
 
    if (mask & WRITEMASK_Y) {
@@ -934,7 +1031,7 @@ static void emit_lit( struct brw_compile *p,
    }
 
    if (mask & WRITEMASK_Z) {
-      emit_math2(p, BRW_MATH_FUNCTION_POW,
+      emit_math2(c, BRW_MATH_FUNCTION_POW,
 		 &dst[2],
 		 WRITEMASK_X | (mask & SATURATE),
 		 &arg0[1],
@@ -989,7 +1086,7 @@ static void emit_kil_nv( struct brw_wm_compile *c )
 
    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_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);
 }
@@ -1001,7 +1098,13 @@ static void fire_fb_write( struct brw_wm_compile *c,
 			   GLuint eot )
 {
    struct brw_compile *p = &c->func;
-   
+   struct brw_reg dst;
+
+   if (c->dispatch_width == 16)
+      dst = retype(vec16(brw_null_reg()), BRW_REGISTER_TYPE_UW);
+   else
+      dst = retype(vec8(brw_null_reg()), BRW_REGISTER_TYPE_UW);
+
    /* Pass through control information:
     */
 /*  mov (8) m1.0<1>:ud   r1.0<8;8,1>:ud   { Align1 NoMask } */
@@ -1018,7 +1121,7 @@ static void fire_fb_write( struct brw_wm_compile *c,
    /* Send framebuffer write message: */
 /*  send (16) null.0<1>:uw m0               r0.0<8;8,1>:uw   0x85a04000:ud    { Align1 EOT } */
    brw_fb_WRITE(p,
-		retype(vec16(brw_null_reg()), BRW_REGISTER_TYPE_UW),
+		dst,
 		base_reg,
 		retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW),
 		target,		
@@ -1050,14 +1153,15 @@ static void emit_aa( struct brw_wm_compile *c,
  * \param arg1  the pass-through depth value
  * \param arg2  the shader-computed depth value
  */
-static void emit_fb_write( struct brw_wm_compile *c,
-			   struct brw_reg *arg0,
-			   struct brw_reg *arg1,
-			   struct brw_reg *arg2,
-			   GLuint target,
-			   GLuint eot)
+void emit_fb_write(struct brw_wm_compile *c,
+		   struct brw_reg *arg0,
+		   struct brw_reg *arg1,
+		   struct brw_reg *arg2,
+		   GLuint target,
+		   GLuint eot)
 {
    struct brw_compile *p = &c->func;
+   struct brw_context *brw = p->brw;
    GLuint nr = 2;
    GLuint channel;
 
@@ -1069,30 +1173,37 @@ static void emit_fb_write( struct brw_wm_compile *c,
    /* I don't really understand how this achieves the color interleave
     * (ie RGBARGBA) in the result:  [Do the saturation here]
     */
-   {
-      brw_push_insn_state(p);
-      
-      for (channel = 0; channel < 4; channel++) {
+   brw_push_insn_state(p);
+
+   for (channel = 0; channel < 4; channel++) {
+      if (c->dispatch_width == 16 && brw->has_compr4) {
+	 /* By setting the high bit of the MRF register number, we indicate
+	  * that we want COMPR4 mode - instead of doing the usual destination
+	  * + 1 for the second half we get destination + 4.
+	  */
+	 brw_MOV(p,
+		 brw_message_reg(nr + channel + (1 << 7)),
+		 arg0[channel]);
+      } else {
 	 /*  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_set_compression_control(p, BRW_COMPRESSION_NONE);
 	 brw_MOV(p,
 		 brw_message_reg(nr + channel),
 		 arg0[channel]);
-       
-	 brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
-	 brw_MOV(p,
-		 brw_message_reg(nr + channel + 4),
-		 sechalf(arg0[channel]));
-      }
 
-      /* skip over the regs populated above:
-       */
-      nr += 8;
-   
-      brw_pop_insn_state(p);
+	 if (c->dispatch_width == 16) {
+	    brw_set_compression_control(p, BRW_COMPRESSION_2NDHALF);
+	    brw_MOV(p,
+		    brw_message_reg(nr + channel + 4),
+		    sechalf(arg0[channel]));
+	 }
+      }
    }
+   /* skip over the regs populated above:
+    */
+   nr += 8;
+   brw_pop_insn_state(p);
 
    if (c->key.source_depth_to_render_target)
    {
@@ -1142,7 +1253,7 @@ static void emit_fb_write( struct brw_wm_compile *c,
 	      get_element_ud(brw_vec8_grf(1,0), 6), 
 	      brw_imm_ud(1<<26)); 
 
-      jmp = brw_JMPI(p, ip, ip, brw_imm_d(0));
+      jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
       {
 	 emit_aa(c, arg1, 2);
 	 fire_fb_write(c, 0, nr, target, eot);
@@ -1156,7 +1267,6 @@ static void emit_fb_write( struct brw_wm_compile *c,
    }
 }
 
-
 /**
  * Move a GPR to scratch memory. 
  */
@@ -1294,7 +1404,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 /* Generated instructions for calculating triangle interpolants:
 	  */
       case WM_PIXELXY:
-	 emit_pixel_xy(p, dst, dst_flags);
+	 emit_pixel_xy(c, dst, dst_flags);
 	 break;
 
       case WM_DELTAXY:
@@ -1306,7 +1416,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 break;
 
       case WM_PIXELW:
-	 emit_pixel_w(p, dst, dst_flags, args[0], args[1]);
+	 emit_pixel_w(c, dst, dst_flags, args[0], args[1]);
 	 break;
 
       case WM_LINTERP:
@@ -1364,7 +1474,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 break;
 
       case OPCODE_TRUNC:
-	 emit_trunc(p, dst, dst_flags, args[0]);
+	 emit_alu1(p, brw_RNDZ, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_LRP:
@@ -1391,27 +1501,27 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 /* Higher math functions:
 	  */
       case OPCODE_RCP:
-	 emit_math1(p, BRW_MATH_FUNCTION_INV, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_INV, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_RSQ:
-	 emit_math1(p, BRW_MATH_FUNCTION_RSQ, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_RSQ, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_SIN:
-	 emit_math1(p, BRW_MATH_FUNCTION_SIN, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_SIN, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_COS:
-	 emit_math1(p, BRW_MATH_FUNCTION_COS, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_COS, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_EX2:
-	 emit_math1(p, BRW_MATH_FUNCTION_EXP, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_EXP, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_LG2:
-	 emit_math1(p, BRW_MATH_FUNCTION_LOG, dst, dst_flags, args[0]);
+	 emit_math1(c, BRW_MATH_FUNCTION_LOG, dst, dst_flags, args[0]);
 	 break;
 
       case OPCODE_SCS:
@@ -1419,13 +1529,13 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	  * fixup for 16-element execution.
 	  */
 	 if (dst_flags & WRITEMASK_X)
-	    emit_math1(p, BRW_MATH_FUNCTION_COS, dst, (dst_flags&SATURATE)|WRITEMASK_X, args[0]);
+	    emit_math1(c, BRW_MATH_FUNCTION_COS, dst, (dst_flags&SATURATE)|WRITEMASK_X, args[0]);
 	 if (dst_flags & WRITEMASK_Y)
-	    emit_math1(p, BRW_MATH_FUNCTION_SIN, dst+1, (dst_flags&SATURATE)|WRITEMASK_X, args[0]);
+	    emit_math1(c, BRW_MATH_FUNCTION_SIN, dst+1, (dst_flags&SATURATE)|WRITEMASK_X, args[0]);
 	 break;
 
       case OPCODE_POW:
-	 emit_math2(p, BRW_MATH_FUNCTION_POW, dst, dst_flags, args[0], args[1]);
+	 emit_math2(c, BRW_MATH_FUNCTION_POW, dst, dst_flags, args[0], args[1]);
 	 break;
 
 	 /* Comparisons:
@@ -1463,17 +1573,20 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	break;
 
       case OPCODE_LIT:
-	 emit_lit(p, dst, dst_flags, args[0]);
+	 emit_lit(c, dst, dst_flags, args[0]);
 	 break;
 
 	 /* Texturing operations:
 	  */
       case OPCODE_TEX:
-	 emit_tex(c, inst, dst, dst_flags, args[0]);
+	 emit_tex(c, dst, dst_flags, args[0], c->payload.depth[0].hw_reg,
+		  inst->tex_idx, inst->tex_unit,
+		  inst->tex_shadow);
 	 break;
 
       case OPCODE_TXB:
-	 emit_txb(c, inst, dst, dst_flags, args[0]);
+	 emit_txb(c, dst, dst_flags, args[0], c->payload.depth[0].hw_reg,
+		  inst->tex_idx, inst->tex_unit);
 	 break;
 
       case OPCODE_KIL:
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index 4e3edfbbff..3737faf26f 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -138,7 +138,6 @@ static struct prog_dst_register dst_reg(GLuint file, GLuint idx)
    reg.CondMask = COND_TR;
    reg.CondSwizzle = 0;
    reg.CondSrc = 0;
-   reg.pad = 0;
    return reg;
 }
 
@@ -181,6 +180,9 @@ static void release_temp( struct brw_wm_compile *c, struct prog_dst_register tem
 
 static struct prog_instruction *get_fp_inst(struct brw_wm_compile *c)
 {
+   assert(c->nr_fp_insns < BRW_WM_MAX_INSN);
+   memset(&c->prog_instructions[c->nr_fp_insns], 0,
+	  sizeof(*c->prog_instructions));
    return &c->prog_instructions[c->nr_fp_insns++];
 }
 
@@ -447,7 +449,6 @@ static void emit_interp( struct brw_wm_compile *c,
       break;
 
    case FRAG_ATTRIB_FACE:
-      /* XXX review/test this case */
       emit_op(c,
               WM_FRONTFACING,
               dst_mask(dst, WRITEMASK_X),
@@ -956,7 +957,7 @@ static void precalc_txp( struct brw_wm_compile *c,
 
 
 
-static void emit_fb_write( struct brw_wm_compile *c )
+static void emit_render_target_writes( struct brw_wm_compile *c )
 {
    struct prog_src_register payload_r0_depth = src_reg(PROGRAM_PAYLOAD, PAYLOAD_DEPTH);
    struct prog_src_register outdepth = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DEPTH);
@@ -964,36 +965,34 @@ static void emit_fb_write( struct brw_wm_compile *c )
    GLuint i;
 
    struct prog_instruction *inst, *last_inst;
-   struct brw_context *brw = c->func.brw;
 
    /* The inst->Aux field is used for FB write target and the EOT marker */
 
-   if (brw->state.nr_color_regions > 1) {
-      for (i = 0 ; i < brw->state.nr_color_regions; i++) {
+   if (c->key.nr_color_regions > 1) {
+      for (i = 0 ; i < c->key.nr_color_regions; i++) {
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0 + i);
-         last_inst = inst = emit_op(c,
-                                    WM_FB_WRITE, dst_mask(dst_undef(),0), 0,
-                                    outcolor, payload_r0_depth, outdepth);
-         inst->Aux = (i<<1);
+         last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(), 0),
+                                    0, outcolor, payload_r0_depth, outdepth);
+         inst->Aux = INST_AUX_TARGET(i);
          if (c->fp_fragcolor_emitted) {
             outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
-            last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
+            last_inst = inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(), 0),
                                        0, outcolor, payload_r0_depth, outdepth);
-            inst->Aux = (i<<1);
+            inst->Aux = INST_AUX_TARGET(i);
          }
       }
-      last_inst->Aux |= 1; //eot
+      last_inst->Aux |= INST_AUX_EOT;
    }
    else {
       /* if gl_FragData[0] is written, use it, else use gl_FragColor */
-      if (c->fp->program.Base.OutputsWritten & (1 << FRAG_RESULT_DATA0))
+      if (c->fp->program.Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DATA0))
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0);
       else 
          outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR);
 
       inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
                      0, outcolor, payload_r0_depth, outdepth);
-      inst->Aux = 1|(0<<1);
+      inst->Aux = INST_AUX_EOT | INST_AUX_TARGET(0);
    }
 }
 
@@ -1155,7 +1154,7 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
 	 out->DstReg.WriteMask = 0;
 	 break;
       case OPCODE_END:
-	 emit_fb_write(c);
+	 emit_render_target_writes(c);
 	 break;
       case OPCODE_PRINT:
 	 break;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index c9fe1dd8ad..fde83eea62 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -371,7 +371,7 @@ static void prealloc_reg(struct brw_wm_compile *c)
 	  for (j = 0; j < 4; j++)
 	     set_reg(c, PROGRAM_PAYLOAD, fp_input, j, reg);
        }
-       if (c->key.vp_outputs_written & (1 << i)) {
+       if (c->key.vp_outputs_written & BITFIELD64_BIT(i)) {
 	  reg_index += 2;
        }
     }
@@ -550,42 +550,6 @@ static struct brw_reg get_src_reg(struct brw_wm_compile *c,
     }
 }
 
-
-/**
- * Same as \sa get_src_reg() but if the register is a literal, emit
- * a brw_reg encoding the literal.
- * Note that a brw instruction only allows one src operand to be a literal.
- * For instructions with more than one operand, only the second can be a
- * literal.  This means that we treat some literals as constants/uniforms
- * (which why PROGRAM_CONSTANT is checked in fetch_constants()).
- * 
- */
-static struct brw_reg get_src_reg_imm(struct brw_wm_compile *c, 
-                                      const struct prog_instruction *inst,
-                                      GLuint srcRegIndex, GLuint channel)
-{
-    const struct prog_src_register *src = &inst->SrcReg[srcRegIndex];
-    if (src->File == PROGRAM_CONSTANT) {
-       /* a literal */
-       const int component = GET_SWZ(src->Swizzle, channel);
-       const GLfloat *param =
-          c->fp->program.Base.Parameters->ParameterValues[src->Index];
-       GLfloat value = param[component];
-       if (src->Negate & (1 << channel))
-          value = -value;
-       if (src->Abs)
-          value = FABSF(value);
-#if 0
-       printf("  form immed value %f for chan %d\n", value, channel);
-#endif
-       return brw_imm_f(value);
-    }
-    else {
-       return get_src_reg(c, inst, srcRegIndex, channel);
-    }
-}
-
-
 /**
  * Subroutines are minimal support for resusable instruction sequences.
  * They are implemented as simply as possible to minimise overhead: there
@@ -650,542 +614,110 @@ static void invoke_subroutine( struct brw_wm_compile *c,
     }
 }
 
-static void emit_trunc( struct brw_wm_compile *c,
-                        const struct prog_instruction *inst)
-{
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
-    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, 0, i);
-	    brw_RNDZ(p, dst, src);
-	}
-    }
-    brw_set_saturate(p, 0);
-}
-
-static void emit_mov( struct brw_wm_compile *c,
-                      const struct prog_instruction *inst)
-{
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
-    for (i = 0; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    struct brw_reg src, dst;
-	    dst = get_dst_reg(c, inst, i);
-            /* XXX some moves from immediate value don't work reliably!!! */
-            /*src = get_src_reg_imm(c, inst, 0, i);*/
-            src = get_src_reg(c, inst, 0, i);
-	    brw_MOV(p, dst, src);
-	}
-    }
-    brw_set_saturate(p, 0);
-}
-
-static void emit_pixel_xy(struct brw_wm_compile *c,
-                          const struct prog_instruction *inst)
-{
-    struct brw_reg r1 = brw_vec1_grf(1, 0);
-    struct brw_reg r1_uw = retype(r1, BRW_REGISTER_TYPE_UW);
-
-    struct brw_reg dst0, dst1;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-
-    dst0 = get_dst_reg(c, inst, 0);
-    dst1 = get_dst_reg(c, inst, 1);
-    /* Calculate pixel centers by adding 1 or 0 to each of the
-     * micro-tile coordinates passed in r1.
-     */
-    if (mask & WRITEMASK_X) {
-	brw_ADD(p,
-		vec8(retype(dst0, BRW_REGISTER_TYPE_UW)),
-		stride(suboffset(r1_uw, 4), 2, 4, 0),
-		brw_imm_v(0x10101010));
-    }
-
-    if (mask & WRITEMASK_Y) {
-	brw_ADD(p,
-		vec8(retype(dst1, BRW_REGISTER_TYPE_UW)),
-		stride(suboffset(r1_uw, 5), 2, 4, 0),
-		brw_imm_v(0x11001100));
-    }
-}
-
-static void emit_delta_xy(struct brw_wm_compile *c,
-                          const struct prog_instruction *inst)
-{
-    struct brw_reg r1 = brw_vec1_grf(1, 0);
-    struct brw_reg dst0, dst1, src0, src1;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-
-    dst0 = get_dst_reg(c, inst, 0);
-    dst1 = get_dst_reg(c, inst, 1);
-    src0 = get_src_reg(c, inst, 0, 0);
-    src1 = get_src_reg(c, inst, 0, 1);
-    /* Calc delta X,Y by subtracting origin in r1 from the pixel
-     * centers.
-     */
-    if (mask & WRITEMASK_X) {
-	brw_ADD(p,
-		dst0,
-		retype(src0, BRW_REGISTER_TYPE_UW),
-		negate(r1));
-    }
-
-    if (mask & WRITEMASK_Y) {
-	brw_ADD(p,
-		dst1,
-		retype(src1, BRW_REGISTER_TYPE_UW),
-		negate(suboffset(r1,1)));
-
-    }
-}
-
-static void fire_fb_write( struct brw_wm_compile *c,
-                           GLuint base_reg,
-                           GLuint nr,
-                           GLuint target,
-                           GLuint eot)
-{
-    struct brw_compile *p = &c->func;
-    /* 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),
-	    target,              
-	    nr,
-	    0,
-	    eot);
-}
-
-static void emit_fb_write(struct brw_wm_compile *c,
-                          const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    int nr = 2;
-    int channel;
-    GLuint target, eot;
-    struct brw_reg src0;
-
-    /* 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++) {
-        src0 = get_src_reg(c,  inst, 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);
-
-    if (c->key.source_depth_to_render_target) {
-       if (c->key.computes_depth) {
-          src0 = get_src_reg(c, inst, 2, 2);
-          brw_MOV(p, brw_message_reg(nr), src0);
-       }
-       else {
-          src0 = get_src_reg(c, inst, 1, 1);
-          brw_MOV(p, brw_message_reg(nr), src0);
-       }
-
-       nr += 2;
-    }
-
-    if (c->key.dest_depth_reg) {
-        const GLuint comp = c->key.dest_depth_reg / 2;
-        const GLuint off = c->key.dest_depth_reg % 2;
-
-        if (off != 0) {
-            /* XXX this code needs review/testing */
-            struct brw_reg arg1_0 = get_src_reg(c, inst, 1, comp);
-            struct brw_reg arg1_1 = get_src_reg(c, inst, 1, comp+1);
-
-            brw_push_insn_state(p);
-            brw_set_compression_control(p, BRW_COMPRESSION_NONE);
-
-            brw_MOV(p, brw_message_reg(nr), offset(arg1_0, 1));
-            /* 2nd half? */
-            brw_MOV(p, brw_message_reg(nr+1), arg1_1);
-            brw_pop_insn_state(p);
-        }
-        else
-        {
-            struct brw_reg src =  get_src_reg(c, inst, 1, 1);
-            brw_MOV(p, brw_message_reg(nr), src);
-        }
-        nr += 2;
-   }
-
-    target = inst->Aux >> 1;
-    eot = inst->Aux & 1;
-    fire_fb_write(c, 0, nr, target, eot);
-}
-
-static void emit_pixel_w( struct brw_wm_compile *c,
-                          const struct prog_instruction *inst)
+/* Workaround for using brw_wm_emit.c's emit functions, which expect
+ * destination regs to be uniquely written.  Moves arguments out to
+ * temporaries as necessary for instructions which use their destination as
+ * a temporary.
+ */
+static void
+unalias3(struct brw_wm_compile *c,
+	 void (*func)(struct brw_compile *c,
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1,
+		      const struct brw_reg *arg2),
+	 const struct brw_reg *dst,
+	 GLuint mask,
+	 const struct brw_reg *arg0,
+	 const struct brw_reg *arg1,
+	 const struct brw_reg *arg2)
 {
     struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    if (mask & WRITEMASK_W) {
-	struct brw_reg dst, src0, delta0, delta1;
-	struct brw_reg interp3;
-
-	dst = get_dst_reg(c, inst, 3);
-	src0 = get_src_reg(c, inst, 0, 0);
-	delta0 = get_src_reg(c, inst, 1, 0);
-	delta1 = get_src_reg(c, inst, 1, 1);
-
-	interp3 = brw_vec1_grf(src0.nr+1, 4);
-	/* Calc 1/w - just linterp wpos[3] optimized by putting the
-	 * result straight into a message reg.
-	 */
-	brw_LINE(p, brw_null_reg(), interp3, delta0);
-	brw_MAC(p, brw_message_reg(2), suboffset(interp3, 1), delta1);
-
-	/* Calc w */
-	brw_math_16( p, dst,
-		BRW_MATH_FUNCTION_INV,
-		BRW_MATH_SATURATE_NONE,
-		2, brw_null_reg(),
-		BRW_MATH_PRECISION_FULL);
-    }
-}
+    struct brw_reg tmp_arg0[4], tmp_arg1[4], tmp_arg2[4];
+    int i, j;
+    int mark = mark_tmps(c);
 
-static void emit_linterp(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, delta0, delta1;
-    struct brw_reg src0;
-    GLuint nr, i;
-
-    src0 = get_src_reg(c, inst, 0, 0);
-    delta0 = get_src_reg(c, inst, 1, 0);
-    delta1 = get_src_reg(c, inst, 1, 1);
-    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);
-
-    for(i = 0; i < 4; i++ ) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    brw_LINE(p, brw_null_reg(), interp[i], delta0);
-	    brw_MAC(p, dst, suboffset(interp[i],1), delta1);
-	}
+    for (j = 0; j < 4; j++) {
+	tmp_arg0[j] = arg0[j];
+	tmp_arg1[j] = arg1[j];
+	tmp_arg2[j] = arg2[j];
     }
-}
-
-static void emit_cinterp(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, src0;
-    GLuint nr, i;
 
-    src0 = get_src_reg(c, inst, 0, 0);
-    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);
-
-    for(i = 0; i < 4; i++ ) {
+    for (i = 0; i < 4; i++) {
 	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    brw_MOV(p, dst, suboffset(interp[i],3));
+	    for (j = 0; j < 4; j++) {
+		if (arg0[j].file == dst[i].file &&
+		    dst[i].nr == arg0[j].nr) {
+		    tmp_arg0[j] = alloc_tmp(c);
+		    brw_MOV(p, tmp_arg0[j], arg0[j]);
+		}
+		if (arg1[j].file == dst[i].file &&
+		    dst[i].nr == arg1[j].nr) {
+		    tmp_arg1[j] = alloc_tmp(c);
+		    brw_MOV(p, tmp_arg1[j], arg1[j]);
+		}
+		if (arg2[j].file == dst[i].file &&
+		    dst[i].nr == arg2[j].nr) {
+		    tmp_arg2[j] = alloc_tmp(c);
+		    brw_MOV(p, tmp_arg2[j], arg2[j]);
+		}
+	    }
 	}
     }
-}
-
-static void emit_pinterp(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, delta0, delta1;
-    struct brw_reg src0, w;
-    GLuint nr, i;
+    func(p, dst, mask, tmp_arg0, tmp_arg1, tmp_arg2);
 
-    src0 = get_src_reg(c, inst, 0, 0);
-    delta0 = get_src_reg(c, inst, 1, 0);
-    delta1 = get_src_reg(c, inst, 1, 1);
-    w = get_src_reg(c, inst, 2, 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);
-
-    for(i = 0; i < 4; i++ ) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    brw_LINE(p, brw_null_reg(), interp[i], delta0);
-	    brw_MAC(p, dst, suboffset(interp[i],1), 
-		    delta1);
-	    brw_MUL(p, dst, dst, w);
-	}
-    }
+    release_tmps(c, mark);
 }
 
-/* Sets the destination channels to 1.0 or 0.0 according to glFrontFacing. */
-static void emit_frontfacing(struct brw_wm_compile *c,
-			     const struct prog_instruction *inst)
+/* Workaround for using brw_wm_emit.c's emit functions, which expect
+ * destination regs to be uniquely written.  Moves arguments out to
+ * temporaries as necessary for instructions which use their destination as
+ * a temporary.
+ */
+static void
+unalias2(struct brw_wm_compile *c,
+	 void (*func)(struct brw_compile *c,
+		      const struct brw_reg *dst,
+		      GLuint mask,
+		      const struct brw_reg *arg0,
+		      const struct brw_reg *arg1),
+	 const struct brw_reg *dst,
+	 GLuint mask,
+	 const struct brw_reg *arg0,
+	 const struct brw_reg *arg1)
 {
     struct brw_compile *p = &c->func;
-    struct brw_reg r1_6ud = retype(brw_vec1_grf(1, 6), BRW_REGISTER_TYPE_UD);
-    struct brw_reg dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int i;
-
-    for (i = 0; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    brw_MOV(p, dst, brw_imm_f(0.0));
-	}
-    }
+    struct brw_reg tmp_arg0[4], tmp_arg1[4];
+    int i, j;
+    int mark = mark_tmps(c);
 
-    /* bit 31 is "primitive is back face", so checking < (1 << 31) gives
-     * us front face
-     */
-    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, r1_6ud, brw_imm_ud(1 << 31));
-    for (i = 0; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    brw_MOV(p, dst, brw_imm_f(1.0));
-	}
+    for (j = 0; j < 4; j++) {
+	tmp_arg0[j] = arg0[j];
+	tmp_arg1[j] = arg1[j];
     }
-    brw_set_predicate_control_flag_value(p, 0xff);
-}
 
-static void emit_xpd(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
     for (i = 0; i < 4; i++) {
-	GLuint i2 = (i+2)%3;
-	GLuint 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, 0, i2));
-	    src1 = get_src_reg_imm(c, inst, 1, i1);
-	    brw_MUL(p, brw_null_reg(), src0, src1);
-	    src0 = get_src_reg(c, inst, 0, i1);
-	    src1 = get_src_reg_imm(c, inst, 1, i2);
-	    brw_set_saturate(p, inst->SaturateMode != SATURATE_OFF);
-	    brw_MAC(p, dst, src0, src1);
-	    brw_set_saturate(p, 0);
+	    for (j = 0; j < 4; j++) {
+		if (arg0[j].file == dst[i].file &&
+		    dst[i].nr == arg0[j].nr) {
+		    tmp_arg0[j] = alloc_tmp(c);
+		    brw_MOV(p, tmp_arg0[j], arg0[j]);
+		}
+		if (arg1[j].file == dst[i].file &&
+		    dst[i].nr == arg1[j].nr) {
+		    tmp_arg1[j] = alloc_tmp(c);
+		    brw_MOV(p, tmp_arg1[j], arg1[j]);
+		}
+	    }
 	}
     }
-    brw_set_saturate(p, 0);
-}
 
-static void emit_dp3(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_reg src0[3], src1[3], dst;
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
-
-    if (!(mask & WRITEMASK_XYZW))
-	return;
-
-    assert(is_power_of_two(mask & WRITEMASK_XYZW));
-
-    for (i = 0; i < 3; i++) {
-	src0[i] = get_src_reg(c, inst, 0, i);
-	src1[i] = get_src_reg_imm(c, inst, 1, i);
-    }
-
-    dst = get_dst_reg(c, inst, dst_chan);
-    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->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    brw_MAC(p, dst, src0[2], src1[2]);
-    brw_set_saturate(p, 0);
-}
-
-static void emit_dp4(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_reg src0[4], src1[4], dst;
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
-
-    if (!(mask & WRITEMASK_XYZW))
-	return;
-
-    assert(is_power_of_two(mask & WRITEMASK_XYZW));
-
-    for (i = 0; i < 4; i++) {
-	src0[i] = get_src_reg(c, inst, 0, i);
-	src1[i] = get_src_reg_imm(c, inst, 1, i);
-    }
-    dst = get_dst_reg(c, inst, dst_chan);
-    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->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    brw_MAC(p, dst, src0[3], src1[3]);
-    brw_set_saturate(p, 0);
-}
-
-static void emit_dph(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_reg src0[4], src1[4], dst;
-    int i;
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
-
-    if (!(mask & WRITEMASK_XYZW))
-	return;
-
-    assert(is_power_of_two(mask & WRITEMASK_XYZW));
-
-    for (i = 0; i < 4; i++) {
-	src0[i] = get_src_reg(c, inst, 0, i);
-	src1[i] = get_src_reg_imm(c, inst, 1, i);
-    }
-    dst = get_dst_reg(c, inst, dst_chan);
-    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->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    brw_ADD(p, dst, dst, src1[3]);
-    brw_set_saturate(p, 0);
-}
-
-/**
- * Emit a scalar instruction, like RCP, RSQ, LOG, EXP.
- * Note that the result of the function is smeared across the dest
- * register's X, Y, Z and W channels (subject to writemasking of course).
- */
-static void emit_math1(struct brw_wm_compile *c,
-                       const struct prog_instruction *inst, GLuint func)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg src0, dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
-
-    if (!(mask & WRITEMASK_XYZW))
-	return;
-
-    assert(is_power_of_two(mask & WRITEMASK_XYZW));
-
-    /* Get first component of source register */
-    dst = get_dst_reg(c, inst, dst_chan);
-    src0 = get_src_reg(c, inst, 0, 0);
-
-    brw_MOV(p, brw_message_reg(2), src0);
-    brw_math(p,
-             dst,
-             func,
-             (inst->SaturateMode != SATURATE_OFF) ? BRW_MATH_SATURATE_SATURATE : BRW_MATH_SATURATE_NONE,
-             2,
-             brw_null_reg(),
-             BRW_MATH_DATA_VECTOR,
-             BRW_MATH_PRECISION_FULL);
-}
-
-static void emit_rcp(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_INV);
-}
-
-static void emit_rsq(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_RSQ);
-}
-
-static void emit_sin(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_SIN);
-}
+    func(p, dst, mask, tmp_arg0, tmp_arg1);
 
-static void emit_cos(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_COS);
-}
-
-static void emit_ex2(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_EXP);
-}
-
-static void emit_lg2(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_math1(c, inst, BRW_MATH_FUNCTION_LOG);
-}
-
-static void emit_add(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg src0, src1, dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int i;
-    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 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, 0, i);
-	    src1 = get_src_reg_imm(c, inst, 1, i);
-	    brw_ADD(p, dst, src0, src1);
-	}
-    }
-    brw_set_saturate(p, 0);
+    release_tmps(c, mark);
 }
 
 static void emit_arl(struct brw_wm_compile *c,
@@ -1201,180 +733,6 @@ static void emit_arl(struct brw_wm_compile *c,
     brw_set_saturate(p, 0);
 }
 
-
-static void emit_mul(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg src0, src1, dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int i;
-    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 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, 0, i);
-	    src1 = get_src_reg_imm(c, inst, 1, i);
-	    brw_MUL(p, dst, src0, src1);
-	}
-    }
-    brw_set_saturate(p, 0);
-}
-
-static void emit_frc(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg src0, dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int i;
-    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    for (i = 0 ; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    src0 = get_src_reg_imm(c, inst, 0, i);
-	    brw_FRC(p, dst, src0);
-	}
-    }
-    if (inst->SaturateMode != SATURATE_OFF)
-	brw_set_saturate(p, 0);
-}
-
-static void emit_flr(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg src0, dst;
-    GLuint mask = inst->DstReg.WriteMask;
-    int i;
-    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    for (i = 0 ; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    src0 = get_src_reg_imm(c, inst, 0, i);
-	    brw_RNDD(p, dst, src0);
-	}
-    }
-    brw_set_saturate(p, 0);
-}
-
-
-static void emit_min_max(struct brw_wm_compile *c,
-                         const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    const GLuint mask = inst->DstReg.WriteMask;
-    const int mark = mark_tmps(c);
-    int i;
-    brw_push_insn_state(p);
-    for (i = 0; i < 4; i++) {
-	if (mask & (1<<i)) {
-            struct brw_reg real_dst = get_dst_reg(c, inst, i);
-	    struct brw_reg src0 = get_src_reg(c, inst, 0, i);
-	    struct brw_reg src1 = get_src_reg(c, inst, 1, i);
-            struct brw_reg dst;
-            /* if dst==src0 or dst==src1 we need to use a temp reg */
-            GLboolean use_temp = brw_same_reg(dst, src0) ||
-                                 brw_same_reg(dst, src1);
-            if (use_temp)
-               dst = alloc_tmp(c);
-            else
-               dst = real_dst;
-
-            /*
-            printf("  Min/max: dst %d  src0 %d  src1 %d\n",
-                   dst.nr, src0.nr, src1.nr);
-            */
-	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
-	    brw_MOV(p, dst, src0);
-	    brw_set_saturate(p, 0);
-
-            if (inst->Opcode == OPCODE_MIN)
-               brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_L, src1, src0);
-            else
-               brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_G, src1, src0);
-
-	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 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);
-            if (use_temp)
-               brw_MOV(p, real_dst, dst);
-	}
-    }
-    brw_pop_insn_state(p);
-    release_tmps(c, mark);
-}
-
-static void emit_pow(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg dst, src0, src1;
-    GLuint mask = inst->DstReg.WriteMask;
-    int dst_chan = _mesa_ffs(mask & WRITEMASK_XYZW) - 1;
-
-    if (!(mask & WRITEMASK_XYZW))
-	return;
-
-    assert(is_power_of_two(mask & WRITEMASK_XYZW));
-
-    dst = get_dst_reg(c, inst, dst_chan);
-    src0 = get_src_reg_imm(c, inst, 0, 0);
-    src1 = get_src_reg_imm(c, inst, 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->SaturateMode != SATURATE_OFF) ? 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,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    struct brw_reg dst, tmp1, tmp2, src0, src1, src2;
-    int i;
-    int mark = mark_tmps(c);
-    for (i = 0; i < 4; i++) {
-	if (mask & (1<<i)) {
-	    dst = get_dst_reg(c, inst, i);
-	    src0 = get_src_reg(c, inst, 0, i);
-
-	    src1 = get_src_reg_imm(c, inst, 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, 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->SaturateMode != SATURATE_OFF) ? 1 : 0);
-	    brw_MAC(p, dst, src0, tmp1);
-	    brw_set_saturate(p, 0);
-	}
-	release_tmps(c, mark);
-    }
-}
-
 /**
  * For GLSL shaders, this KIL will be unconditional.
  * It may be contained inside an IF/ENDIF structure of course.
@@ -1385,94 +743,11 @@ static void emit_kil(struct brw_wm_compile *c)
     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_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,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.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, 0, i);
-	    src1 = get_src_reg_imm(c, inst, 1, i);
-	    src2 = get_src_reg_imm(c, inst, 2, i);
-	    brw_MUL(p, dst, src0, src1);
-
-	    brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
-	    brw_ADD(p, dst, dst, src2);
-	    brw_set_saturate(p, 0);
-	}
-    }
-}
-
-static void emit_sop(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst, GLuint cond)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    struct brw_reg dst, src0, src1;
-    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, 0, i);
-	    src1 = get_src_reg_imm(c, inst, 1, i);
-	    brw_push_insn_state(p);
-	    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_slt(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_L);
-}
-
-static void emit_sle(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_LE);
-}
-
-static void emit_sgt(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_G);
-}
-
-static void emit_sge(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_GE);
-}
-
-static void emit_seq(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_EQ);
-}
-
-static void emit_sne(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    emit_sop(c, inst, BRW_CONDITIONAL_NEQ);
-}
-
 static INLINE struct brw_reg high_words( struct brw_reg reg )
 {
     return stride( suboffset( retype( reg, BRW_REGISTER_TYPE_W ), 1 ),
@@ -2525,196 +1800,6 @@ static void emit_noise4( struct brw_wm_compile *c,
     
     release_tmps( c, mark );
 }
-    
-static void emit_wpos_xy(struct brw_wm_compile *c,
-                         const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    GLuint mask = inst->DstReg.WriteMask;
-    struct brw_reg src0[2], dst[2];
-
-    dst[0] = get_dst_reg(c, inst, 0);
-    dst[1] = get_dst_reg(c, inst, 1);
-
-    src0[0] = get_src_reg(c, inst, 0, 0);
-    src0[1] = get_src_reg(c, inst, 0, 1);
-
-    /* Calculate the pixel offset from window bottom left into destination
-     * X and Y channels.
-     */
-    if (mask & WRITEMASK_X) {
-	/* X' = X - origin_x */
-	brw_ADD(p,
-		dst[0],
-		retype(src0[0], BRW_REGISTER_TYPE_W),
-		brw_imm_d(0 - c->key.origin_x));
-    }
-
-    if (mask & WRITEMASK_Y) {
-	/* Y' = height - (Y - origin_y) = height + origin_y - Y */
-	brw_ADD(p,
-		dst[1],
-		negate(retype(src0[1], BRW_REGISTER_TYPE_W)),
-		brw_imm_d(c->key.origin_y + c->key.drawable_height - 1));
-    }
-}
-
-/* TODO
-   BIAS on SIMD8 not working yet...
- */	
-static void emit_txb(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg dst[4], src[4], payload_reg;
-    /* Note: TexSrcUnit was already looked up through SamplerTextures[] */
-    const GLuint unit = inst->TexSrcUnit;
-    GLuint i;
-    GLuint msg_type;
-
-    assert(unit < BRW_MAX_TEX_UNIT);
-
-    payload_reg = get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH, 0, 1, 0, 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, 0, i);
-
-    switch (inst->TexSrcTarget) {
-	case TEXTURE_1D_INDEX:
-	    brw_MOV(p, brw_message_reg(2), src[0]);         /* s coord */
-	    brw_MOV(p, brw_message_reg(3), brw_imm_f(0));   /* t coord */
-	    brw_MOV(p, brw_message_reg(4), brw_imm_f(0));   /* r coord */
-	    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;
-	case TEXTURE_3D_INDEX:
-	case TEXTURE_CUBE_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), src[2]);
-	    break;
-	default:
-            /* invalid target */
-            abort();
-    }
-    brw_MOV(p, brw_message_reg(5), src[3]);          /* bias */
-    brw_MOV(p, brw_message_reg(6), brw_imm_f(0));    /* ref (unused?) */
-
-    if (BRW_IS_IGDNG(p->brw)) {
-        msg_type = BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_BIAS_IGDNG;
-    } else {
-        /* Does it work well on SIMD8? */
-        msg_type = BRW_SAMPLER_MESSAGE_SIMD16_SAMPLE_BIAS;
-    }
-
-    brw_SAMPLE(p,
-               retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW),  /* dest */
-               1,                                           /* msg_reg_nr */
-               retype(payload_reg, BRW_REGISTER_TYPE_UW),   /* src0 */
-               SURF_INDEX_TEXTURE(unit),
-               unit,                                        /* sampler */
-               inst->DstReg.WriteMask,                      /* writemask */
-               msg_type,                                    /* msg_type */
-               4,                                           /* response_length */
-               4,                                           /* msg_length */
-               0,                                           /* eot */
-               1,
-               BRW_SAMPLER_SIMD_MODE_SIMD8);	
-}
-
-
-static void emit_tex(struct brw_wm_compile *c,
-                     const struct prog_instruction *inst)
-{
-    struct brw_compile *p = &c->func;
-    struct brw_reg dst[4], src[4], payload_reg;
-    /* Note: TexSrcUnit was already looked up through SamplerTextures[] */
-    const GLuint unit = inst->TexSrcUnit;
-    GLuint msg_len;
-    GLuint i, nr;
-    GLuint emit;
-    GLboolean shadow = (c->key.shadowtex_mask & (1<<unit)) ? 1 : 0;
-    GLuint msg_type;
-
-    assert(unit < BRW_MAX_TEX_UNIT);
-
-    payload_reg = get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH, 0, 1, 0, 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, 0, i);
-
-    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;
-	case TEXTURE_3D_INDEX:
-	case TEXTURE_CUBE_INDEX:
-	    emit = WRITEMASK_XYZ;
-	    nr = 3;
-	    break;
-	default:
-           /* invalid target */
-           abort();
-    }
-    msg_len = 1;
-
-    /* move/load S, T, R coords */
-    for (i = 0; i < nr; i++) {
-	static const GLuint 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));  /* lod / bias */
-       brw_MOV(p, brw_message_reg(6), src[2]);        /* ref value / R coord */
-    }
-
-    if (BRW_IS_IGDNG(p->brw)) {
-        if (shadow)
-            msg_type = BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_COMPARE_IGDNG;
-        else
-            msg_type = BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE_IGDNG;
-    } else {
-        /* Does it work for shadow on SIMD8 ? */
-        msg_type = BRW_SAMPLER_MESSAGE_SIMD8_SAMPLE;
-    }
-    
-    brw_SAMPLE(p,
-               retype(vec8(dst[0]), BRW_REGISTER_TYPE_UW), /* dest */
-               1,                                          /* msg_reg_nr */
-               retype(payload_reg, BRW_REGISTER_TYPE_UW),  /* src0 */
-               SURF_INDEX_TEXTURE(unit),
-               unit,                                       /* sampler */
-               inst->DstReg.WriteMask,                     /* writemask */
-               msg_type,                                   /* msg_type */
-               4,                                          /* response_length */
-               shadow ? 6 : 4,                             /* msg_length */
-               0,                                          /* eot */
-               1,
-               BRW_SAMPLER_SIMD_MODE_SIMD8);	
-
-    if (shadow)
-	brw_MOV(p, dst[3], brw_imm_f(1.0));
-}
-
 
 /**
  * Resolve subroutine calls after code emit is done.
@@ -2741,6 +1826,7 @@ get_argument_regs(struct brw_wm_compile *c,
 
 static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 {
+   struct intel_context *intel = &brw->intel;
 #define MAX_IF_DEPTH 32
 #define MAX_LOOP_DEPTH 32
     struct brw_instruction *if_inst[MAX_IF_DEPTH], *loop_inst[MAX_LOOP_DEPTH];
@@ -2771,137 +1857,153 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
         if (c->fp->use_const_buffer)
            fetch_constants(c, inst);
 
-	if (inst->CondUpdate)
-	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-	else
-	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
+	if (inst->Opcode != OPCODE_ARL) {
+	   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();
+	   }
+	}
+	for (j = 0; j < brw_wm_nr_args(inst->Opcode); j++)
+	    get_argument_regs(c, inst, j, args[j], WRITEMASK_XYZW);
 
 	dst_flags = inst->DstReg.WriteMask;
 	if (inst->SaturateMode == SATURATE_ZERO_ONE)
 	    dst_flags |= SATURATE;
 
+	if (inst->CondUpdate)
+	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+	else
+	    brw_set_conditionalmod(p, BRW_CONDITIONAL_NONE);
+
 	switch (inst->Opcode) {
 	    case WM_PIXELXY:
-		emit_pixel_xy(c, inst);
+		emit_pixel_xy(c, dst, dst_flags);
 		break;
 	    case WM_DELTAXY: 
-		emit_delta_xy(c, inst);
+		emit_delta_xy(p, dst, dst_flags, args[0]);
 		break;
 	    case WM_PIXELW:
-		emit_pixel_w(c, inst);
+		emit_pixel_w(c, dst, dst_flags, args[0], args[1]);
 		break;	
 	    case WM_LINTERP:
-		emit_linterp(c, inst);
+		emit_linterp(p, dst, dst_flags, args[0], args[1]);
 		break;
 	    case WM_PINTERP:
-		emit_pinterp(c, inst);
+		emit_pinterp(p, dst, dst_flags, args[0], args[1], args[2]);
 		break;
 	    case WM_CINTERP:
-		emit_cinterp(c, inst);
+		emit_cinterp(p, dst, dst_flags, args[0]);
 		break;
 	    case WM_WPOSXY:
-		emit_wpos_xy(c, inst);
+		emit_wpos_xy(c, dst, dst_flags, args[0]);
 		break;
 	    case WM_FB_WRITE:
-		emit_fb_write(c, inst);
+		emit_fb_write(c, args[0], args[1], args[2],
+			      INST_AUX_GET_TARGET(inst->Aux),
+			      inst->Aux & INST_AUX_EOT);
 		break;
 	    case WM_FRONTFACING:
-		emit_frontfacing(c, inst);
+		emit_frontfacing(p, dst, dst_flags);
 		break;
 	    case OPCODE_ADD:
-		emit_add(c, inst);
+		emit_alu2(p, brw_ADD, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_ARL:
 		emit_arl(c, inst);
 		break;
 	    case OPCODE_FRC:
-		emit_frc(c, inst);
+		emit_alu1(p, brw_FRC, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_FLR:
-		emit_flr(c, inst);
+		emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_LRP:
-		emit_lrp(c, inst);
+		unalias3(c, emit_lrp,
+			 dst, dst_flags, args[0], args[1], args[2]);
 		break;
 	    case OPCODE_TRUNC:
-		emit_trunc(c, inst);
+		emit_alu1(p, brw_RNDZ, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_MOV:
 	    case OPCODE_SWZ:
-		emit_mov(c, inst);
+		emit_alu1(p, brw_MOV, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_DP3:
-		emit_dp3(c, inst);
+		emit_dp3(p, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_DP4:
-		emit_dp4(c, inst);
+		emit_dp4(p, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_XPD:
-		emit_xpd(c, inst);
+		emit_xpd(p, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_DPH:
-		emit_dph(c, inst);
+		emit_dph(p, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_RCP:
-		emit_rcp(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_INV, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_RSQ:
-		emit_rsq(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_RSQ, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_SIN:
-		emit_sin(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_SIN, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_COS:
-		emit_cos(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_COS, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_EX2:
-		emit_ex2(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_EXP, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_LG2:
-		emit_lg2(c, inst);
+		emit_math1(c, BRW_MATH_FUNCTION_LOG, dst, dst_flags, args[0]);
 		break;
 	    case OPCODE_MIN:	
+		unalias2(c, emit_min, dst, dst_flags, args[0], args[1]);
+		break;
 	    case OPCODE_MAX:	
-		emit_min_max(c, inst);
+		unalias2(c, emit_max, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_DDX:
 	    case OPCODE_DDY:
-		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);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_L, args[0], args[1]);
 		break;
 	    case OPCODE_SLE:
-		emit_sle(c, inst);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_LE, args[0], args[1]);
 		break;
 	    case OPCODE_SGT:
-		emit_sgt(c, inst);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_G, args[0], args[1]);
 		break;
 	    case OPCODE_SGE:
-		emit_sge(c, inst);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_GE, args[0], args[1]);
 		break;
 	    case OPCODE_SEQ:
-		emit_seq(c, inst);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_EQ, args[0], args[1]);
 		break;
 	    case OPCODE_SNE:
-		emit_sne(c, inst);
+		emit_sop(p, dst, dst_flags,
+			 BRW_CONDITIONAL_NEQ, args[0], args[1]);
 		break;
 	    case OPCODE_MUL:
-		emit_mul(c, inst);
+		emit_alu2(p, brw_MUL, dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_POW:
-		emit_pow(c, inst);
+		emit_math2(c, BRW_MATH_FUNCTION_POW,
+			   dst, dst_flags, args[0], args[1]);
 		break;
 	    case OPCODE_MAD:
-		emit_mad(c, inst);
+		emit_mad(p, dst, dst_flags, args[0], args[1], args[2]);
 		break;
 	    case OPCODE_NOISE1:
 		emit_noise1(c, inst);
@@ -2916,10 +2018,19 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 		emit_noise4(c, inst);
 		break;
 	    case OPCODE_TEX:
-		emit_tex(c, inst);
+		emit_tex(c, dst, dst_flags, args[0],
+			 get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH,
+				 0, 1, 0, 0),
+			 inst->TexSrcTarget,
+			 inst->TexSrcUnit,
+			 (c->key.shadowtex_mask & (1 << inst->TexSrcUnit)) != 0);
 		break;
 	    case OPCODE_TXB:
-		emit_txb(c, inst);
+		emit_txb(c, dst, dst_flags, args[0],
+			 get_reg(c, PROGRAM_PAYLOAD, PAYLOAD_DEPTH,
+				 0, 1, 0, 0),
+			 inst->TexSrcTarget,
+			 c->fp->program.Base.SamplerUnits[inst->TexSrcUnit]);
 		break;
 	    case OPCODE_KIL_NV:
 		emit_kil(c);
@@ -2929,6 +2040,7 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 		if_inst[if_depth++] = brw_IF(p, BRW_EXECUTE_8);
 		break;
 	    case OPCODE_ELSE:
+		assert(if_depth > 0);
 		if_inst[if_depth-1]  = brw_ELSE(p, if_inst[if_depth-1]);
 		break;
 	    case OPCODE_ENDIF:
@@ -2982,9 +2094,10 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
                   struct brw_instruction *inst0, *inst1;
                   GLuint br = 1;
 
-                  if (BRW_IS_IGDNG(brw))
+                  if (intel->is_ironlake)
                      br = 2;
- 
+
+		  assert(loop_depth > 0);
                   loop_depth--;
                   inst0 = inst1 = brw_WHILE(p, loop_inst[loop_depth]);
                   /* patch all the BREAK/CONT instructions from last BGNLOOP */
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
index 6279258339..ff4c082d5e 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
@@ -42,12 +42,14 @@
 static struct brw_wm_ref *get_ref( struct brw_wm_compile *c )
 {
    assert(c->nr_refs < BRW_WM_MAX_REF);
+   memset(&c->refs[c->nr_refs], 0, sizeof(*c->refs));
    return &c->refs[c->nr_refs++];
 }
 
 static struct brw_wm_value *get_value( struct brw_wm_compile *c)
 {
    assert(c->nr_refs < BRW_WM_MAX_VREG);
+   memset(&c->vreg[c->nr_vreg], 0, sizeof(*c->vreg));
    return &c->vreg[c->nr_vreg++];
 }
 
@@ -55,6 +57,7 @@ static struct brw_wm_value *get_value( struct brw_wm_compile *c)
 static struct brw_wm_instruction *get_instruction( struct brw_wm_compile *c )
 {
    assert(c->nr_insns < BRW_WM_MAX_INSN);
+   memset(&c->instruction[c->nr_insns], 0, sizeof(*c->instruction));
    return &c->instruction[c->nr_insns++];
 }
 
@@ -322,8 +325,8 @@ translate_insn(struct brw_wm_compile *c,
    out->tex_unit = inst->TexSrcUnit;
    out->tex_idx = inst->TexSrcTarget;
    out->tex_shadow = inst->TexShadow;
-   out->eot = inst->Aux & 1;
-   out->target = inst->Aux >> 1;
+   out->eot = inst->Aux & INST_AUX_EOT;
+   out->target = INST_AUX_GET_TARGET(inst->Aux);
 
    /* Args:
     */
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass2.c b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
index 6faea018fb..31303febf0 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass2.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
@@ -82,8 +82,8 @@ static void init_registers( struct brw_wm_compile *c )
    for (j = 0; j < c->nr_creg; j++) 
       prealloc_reg(c, &c->creg[j], i++);
 
-   for (j = 0; j < FRAG_ATTRIB_MAX; j++) {
-      if (c->key.vp_outputs_written & (1<<j)) {
+   for (j = 0; j < VERT_RESULT_MAX; j++) {
+      if (c->key.vp_outputs_written & BITFIELD64_BIT(j)) {
 	 int fp_index;
 
 	 if (j >= VERT_RESULT_VAR0)
diff --git a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
index dff466587a..ad267a4e6a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
@@ -66,19 +66,6 @@ static GLuint translate_wrap_mode( GLenum wrap )
    }
 }
 
-
-static GLuint U_FIXED(GLfloat value, GLuint frac_bits)
-{
-   value *= (1<<frac_bits);
-   return value < 0 ? 0 : value;
-}
-
-static GLint S_FIXED(GLfloat value, GLuint frac_bits)
-{
-   return value * (1<<frac_bits);
-}
-
-
 static dri_bo *upload_default_color( struct brw_context *brw,
 				     const GLfloat *color )
 {
@@ -86,8 +73,8 @@ static dri_bo *upload_default_color( struct brw_context *brw,
 
    COPY_4V(sdc.color, color); 
    
-   return brw_cache_data( &brw->cache, BRW_SAMPLER_DEFAULT_COLOR, &sdc,
-			  NULL, 0 );
+   return brw_cache_data(&brw->cache, BRW_SAMPLER_DEFAULT_COLOR,
+			 &sdc, sizeof(sdc), NULL, 0);
 }
 
 
@@ -228,8 +215,8 @@ static void brw_update_sampler_state(struct wm_sampler_entry *key,
     */
    sampler->ss0.base_level = U_FIXED(0, 1);
 
-   sampler->ss1.max_lod = U_FIXED(MIN2(MAX2(key->maxlod, 0), 13), 6);
-   sampler->ss1.min_lod = U_FIXED(MIN2(MAX2(key->minlod, 0), 13), 6);
+   sampler->ss1.max_lod = U_FIXED(CLAMP(key->maxlod, 0, 13), 6);
+   sampler->ss1.min_lod = U_FIXED(CLAMP(key->minlod, 0, 13), 6);
    
    sampler->ss2.default_color_pointer = sdc_bo->offset >> 5; /* reloc */
 }
@@ -275,10 +262,10 @@ brw_wm_sampler_populate_key(struct brw_context *brw,
 	 dri_bo_unreference(brw->wm.sdc_bo[unit]);
 	 if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
 	    float bordercolor[4] = {
-	       texObj->BorderColor[0],
-	       texObj->BorderColor[0],
-	       texObj->BorderColor[0],
-	       texObj->BorderColor[0]
+	       texObj->BorderColor.f[0],
+	       texObj->BorderColor.f[0],
+	       texObj->BorderColor.f[0],
+	       texObj->BorderColor.f[0]
 	    };
 	    /* GL specs that border color for depth textures is taken from the
 	     * R channel, while the hardware uses A.  Spam R into all the
@@ -287,7 +274,7 @@ brw_wm_sampler_populate_key(struct brw_context *brw,
 	    brw->wm.sdc_bo[unit] = upload_default_color(brw, bordercolor);
 	 } else {
 	    brw->wm.sdc_bo[unit] = upload_default_color(brw,
-							texObj->BorderColor);
+							texObj->BorderColor.f);
 	 }
 	 key->sampler_count = unit + 1;
       }
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 361f91292b..d3373ea79e 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -49,8 +49,6 @@ struct brw_wm_unit_key {
    unsigned int curbe_offset;
    unsigned int urb_size;
 
-   unsigned int max_threads;
-
    unsigned int nr_surfaces, sampler_count;
    GLboolean uses_depth, computes_depth, uses_kill, is_glsl;
    GLboolean polygon_stipple, stats_wm, line_stipple, offset_enable;
@@ -67,18 +65,6 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
 
    memset(key, 0, sizeof(*key));
 
-   if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
-      key->max_threads = 1;
-   else {
-      /* WM maximum threads is number of EUs times number of threads per EU. */
-      if (BRW_IS_IGDNG(brw))
-         key->max_threads = 12 * 6;
-      else if (BRW_IS_G4X(brw))
-	 key->max_threads = 10 * 5;
-      else
-	 key->max_threads = 8 * 4;
-   }
-
    /* CACHE_NEW_WM_PROG */
    key->total_grf = brw->wm.prog_data->total_grf;
    key->urb_entry_read_length = brw->wm.prog_data->urb_read_length;
@@ -106,7 +92,7 @@ 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;
+      (fp->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH)) != 0;
    /* BRW_NEW_DEPTH_BUFFER
     * Override for NULL depthbuffer case, required by the Pixel Shader Computed
     * Depth field.
@@ -140,6 +126,7 @@ static dri_bo *
 wm_unit_create_from_key(struct brw_context *brw, struct brw_wm_unit_key *key,
 			dri_bo **reloc_bufs)
 {
+   struct intel_context *intel = &brw->intel;
    struct brw_wm_unit_state wm;
    dri_bo *bo;
 
@@ -150,7 +137,7 @@ wm_unit_create_from_key(struct brw_context *brw, struct brw_wm_unit_key *key,
    wm.thread1.depth_coef_urb_read_offset = 1;
    wm.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
 
-   if (BRW_IS_IGDNG(brw))
+   if (intel->is_ironlake)
       wm.thread1.binding_table_entry_count = 0; /* hardware requirement */
    else
       wm.thread1.binding_table_entry_count = key->nr_surfaces;
@@ -170,7 +157,7 @@ wm_unit_create_from_key(struct brw_context *brw, struct brw_wm_unit_key *key,
    wm.thread3.const_urb_entry_read_length = key->curb_entry_read_length;
    wm.thread3.const_urb_entry_read_offset = key->curbe_offset * 2;
 
-   if (BRW_IS_IGDNG(brw)) 
+   if (intel->is_ironlake)
       wm.wm4.sampler_count = 0; /* hardware requirement */
    else
       wm.wm4.sampler_count = (key->sampler_count + 1) / 4;
@@ -191,7 +178,7 @@ wm_unit_create_from_key(struct brw_context *brw, struct brw_wm_unit_key *key,
    else
       wm.wm5.enable_16_pix = 1;
 
-   wm.wm5.max_threads = key->max_threads - 1;
+   wm.wm5.max_threads = brw->wm_max_threads - 1;
    wm.wm5.thread_dispatch_enable = 1;	/* AKA: color_write */
    wm.wm5.legacy_line_rast = 0;
    wm.wm5.legacy_global_depth_bias = 0;
@@ -268,7 +255,7 @@ static void upload_wm_unit( struct brw_context *brw )
     */
    assert(key.total_scratch <= 12 * 1024);
    if (key.total_scratch) {
-      GLuint total = key.total_scratch * key.max_threads;
+      GLuint total = key.total_scratch * brw->wm_max_threads;
 
       if (brw->wm.scratch_bo && total > brw->wm.scratch_bo->size) {
 	 dri_bo_unreference(brw->wm.scratch_bo);
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 9c28a22a29..f26cfabb7d 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -31,7 +31,6 @@
                    
 
 #include "main/mtypes.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "shader/prog_parameter.h"
 
@@ -70,7 +69,8 @@ static GLuint translate_tex_target( GLenum target )
 }
 
 
-static GLuint translate_tex_format( GLuint mesa_format, GLenum internal_format,
+static GLuint translate_tex_format( gl_format mesa_format,
+                                    GLenum internal_format,
 				    GLenum depth_mode )
 {
    switch( mesa_format ) {
@@ -86,21 +86,22 @@ static GLuint translate_tex_format( GLuint mesa_format, GLenum internal_format,
    case MESA_FORMAT_AL88:
       return BRW_SURFACEFORMAT_L8A8_UNORM;
 
+   case MESA_FORMAT_AL1616:
+      return BRW_SURFACEFORMAT_L16A16_UNORM;
+
    case MESA_FORMAT_RGB888:
       assert(0);		/* not supported for sampling */
       return BRW_SURFACEFORMAT_R8G8B8_UNORM;      
 
    case MESA_FORMAT_ARGB8888:
-      if (internal_format == GL_RGB)
-	 return BRW_SURFACEFORMAT_B8G8R8X8_UNORM;
-      else
-	 return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+      return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+
+   case MESA_FORMAT_XRGB8888:
+      return BRW_SURFACEFORMAT_B8G8R8X8_UNORM;
 
    case MESA_FORMAT_RGBA8888_REV:
-      if (internal_format == GL_RGB)
-	 return BRW_SURFACEFORMAT_R8G8B8X8_UNORM;
-      else
-	 return BRW_SURFACEFORMAT_R8G8B8A8_UNORM;
+      _mesa_problem(NULL, "unexpected format in i965:translate_tex_format()");
+      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM;
 
    case MESA_FORMAT_RGB565:
       return BRW_SURFACEFORMAT_B5G6R5_UNORM;
@@ -287,7 +288,7 @@ brw_update_texture_surface( GLcontext *ctx, GLuint unit )
       key.bo = NULL;
       key.offset = intelObj->textureOffset;
    } else {
-      key.format = firstImage->TexFormat->MesaFormat;
+      key.format = firstImage->TexFormat;
       key.internal_format = firstImage->InternalFormat;
       key.pitch = intelObj->mt->pitch;
       key.depth = firstImage->Depth;
@@ -354,7 +355,10 @@ brw_create_constant_surface( struct brw_context *brw,
 			 NULL, NULL);
 
    if (key->bo) {
-      /* Emit relocation to surface contents */
+      /* Emit relocation to surface contents.  Section 5.1.1 of the gen4
+       * bspec ("Data Cache") says that the data cache does not exist as
+       * a separate cache and is just the sampler cache.
+       */
       dri_bo_emit_reloc(bo,
 			I915_GEM_DOMAIN_SAMPLER, 0,
 			0,
@@ -507,7 +511,8 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
 				struct gl_renderbuffer *rb,
 				unsigned int unit)
 {
-   GLcontext *ctx = &brw->intel.ctx;
+   struct intel_context *intel = &brw->intel;;
+   GLcontext *ctx = &intel->ctx;
    dri_bo *region_bo = NULL;
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);
    struct intel_region *region = irb ? irb->region : NULL;
@@ -518,7 +523,8 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       GLubyte color_mask[4];
       GLboolean color_blend;
       uint32_t tiling;
-      uint32_t draw_offset;
+      uint32_t draw_x;
+      uint32_t draw_y;
    } key;
 
    memset(&key, 0, sizeof(key));
@@ -527,8 +533,15 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       region_bo = region->buffer;
 
       key.surface_type = BRW_SURFACE_2D;
-      switch (irb->texformat->MesaFormat) {
+      switch (irb->Base.Format) {
+      /* XRGB and ARGB are treated the same here because the chips in this
+       * family cannot render to XRGB targets.  This means that we have to
+       * mask writes to alpha (ala glColorMask) and reconfigure the alpha
+       * blending hardware to use GL_ONE (or GL_ZERO) for cases where
+       * GL_DST_ALPHA (or GL_ONE_MINUS_DST_ALPHA) is used.
+       */
       case MESA_FORMAT_ARGB8888:
+      case MESA_FORMAT_XRGB8888:
 	 key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
 	 break;
       case MESA_FORMAT_RGB565:
@@ -541,8 +554,7 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
 	 key.surface_format = BRW_SURFACEFORMAT_B4G4R4A4_UNORM;
 	 break;
       default:
-	 _mesa_problem(ctx, "Bad renderbuffer format: %d\n",
-		       irb->texformat->MesaFormat);
+	 _mesa_problem(ctx, "Bad renderbuffer format: %d\n", irb->Base.Format);
       }
       key.tiling = region->tiling;
       if (brw->intel.intelScreen->driScrnPriv->dri2.enabled) {
@@ -554,7 +566,8 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       }
       key.pitch = region->pitch;
       key.cpp = region->cpp;
-      key.draw_offset = region->draw_offset; /* cur 3d or cube face offset */
+      key.draw_x = region->draw_x;
+      key.draw_y = region->draw_y;
    } else {
       key.surface_type = BRW_SURFACE_NULL;
       key.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
@@ -562,10 +575,19 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       key.width = 1;
       key.height = 1;
       key.cpp = 4;
-      key.draw_offset = 0;
+      key.draw_x = 0;
+      key.draw_y = 0;
    }
-   memcpy(key.color_mask, ctx->Color.ColorMask,
+   /* _NEW_COLOR */
+   memcpy(key.color_mask, ctx->Color.ColorMask[0],
 	  sizeof(key.color_mask));
+
+   /* As mentioned above, disable writes to the alpha component when the
+    * renderbuffer is XRGB.
+    */
+   if (ctx->DrawBuffer->Visual.alphaBits == 0)
+     key.color_mask[3] = GL_FALSE;
+
    key.color_blend = (!ctx->Color._LogicOpEnabled &&
 		      ctx->Color.BlendEnabled);
 
@@ -584,25 +606,32 @@ brw_update_renderbuffer_surface(struct brw_context *brw,
       surf.ss0.surface_format = key.surface_format;
       surf.ss0.surface_type = key.surface_type;
       if (key.tiling == I915_TILING_NONE) {
-	 surf.ss1.base_addr = key.draw_offset;
+	 surf.ss1.base_addr = (key.draw_x + key.draw_y * key.pitch) * key.cpp;
       } else {
-	 uint32_t tile_offset = key.draw_offset % 4096;
-
-	 surf.ss1.base_addr = key.draw_offset - tile_offset;
-
-	 assert(BRW_IS_G4X(brw) || tile_offset == 0);
-	 if (BRW_IS_G4X(brw)) {
-	    if (key.tiling == I915_TILING_X) {
-	       /* Note that the low bits of these fields are missing, so
-		* there's the possibility of getting in trouble.
-		*/
-	       surf.ss5.x_offset = (tile_offset % 512) / key.cpp / 4;
-	       surf.ss5.y_offset = tile_offset / 512 / 2;
-	    } else {
-	       surf.ss5.x_offset = (tile_offset % 128) / key.cpp / 4;
-	       surf.ss5.y_offset = tile_offset / 128 / 2;
-	    }
+	 uint32_t tile_base, tile_x, tile_y;
+	 uint32_t pitch = key.pitch * key.cpp;
+
+	 if (key.tiling == I915_TILING_X) {
+	    tile_x = key.draw_x % (512 / key.cpp);
+	    tile_y = key.draw_y % 8;
+	    tile_base = ((key.draw_y / 8) * (8 * pitch));
+	    tile_base += (key.draw_x - tile_x) / (512 / key.cpp) * 4096;
+	 } else {
+	    /* Y */
+	    tile_x = key.draw_x % (128 / key.cpp);
+	    tile_y = key.draw_y % 32;
+	    tile_base = ((key.draw_y / 32) * (32 * pitch));
+	    tile_base += (key.draw_x - tile_x) / (128 / key.cpp) * 4096;
 	 }
+	 assert(intel->is_g4x || (tile_x == 0 && tile_y == 0));
+	 assert(tile_x % 4 == 0);
+	 assert(tile_y % 2 == 0);
+	 /* Note that the low bits of these fields are missing, so
+	  * there's the possibility of getting in trouble.
+	  */
+	 surf.ss1.base_addr = tile_base;
+	 surf.ss5.x_offset = tile_x / 4;
+	 surf.ss5.y_offset = tile_y / 2;
       }
       if (region_bo != NULL)
 	 surf.ss1.base_addr += region_bo->offset; /* reloc */
@@ -693,11 +722,10 @@ 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;
 
-   /* _NEW_BUFFERS */
+   /* _NEW_BUFFERS | _NEW_COLOR */
    /* Update surfaces for drawing buffers */
    if (ctx->DrawBuffer->_NumColorDrawBuffers >= 1) {
       for (i = 0; i < ctx->DrawBuffer->_NumColorDrawBuffers; i++) {
@@ -710,7 +738,7 @@ static void prepare_wm_surfaces(struct brw_context *brw )
    }
 
    old_nr_surfaces = brw->wm.nr_surfaces;
-   brw->wm.nr_surfaces = MAX_DRAW_BUFFERS;
+   brw->wm.nr_surfaces = BRW_MAX_DRAW_BUFFERS;
 
    if (brw->wm.surf_bo[SURF_INDEX_FRAG_CONST_BUFFER] != NULL)
        brw->wm.nr_surfaces = SURF_INDEX_FRAG_CONST_BUFFER + 1;
diff --git a/src/mesa/drivers/dri/i965/intel_swapbuffers.c b/src/mesa/drivers/dri/i965/intel_swapbuffers.c
deleted file mode 120000
index 148d5215aa..0000000000
--- a/src/mesa/drivers/dri/i965/intel_swapbuffers.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_swapbuffers.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
index 6aa36d10b1..3a4b21a844 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
@@ -80,7 +80,7 @@ intel_batchbuffer_reset(struct intel_batchbuffer *batch)
       batch->buf = NULL;
    }
 
-   if (!batch->buffer && intel->ttm == GL_TRUE)
+   if (!batch->buffer)
       batch->buffer = malloc (intel->maxBatchSize);
 
    batch->buf = dri_bo_alloc(intel->bufmgr, "batchbuffer",
@@ -94,7 +94,6 @@ intel_batchbuffer_reset(struct intel_batchbuffer *batch)
    batch->size = intel->maxBatchSize;
    batch->ptr = batch->map;
    batch->dirty_state = ~0;
-   batch->cliprect_mode = IGNORE_CLIPRECTS;
 }
 
 struct intel_batchbuffer *
@@ -129,13 +128,10 @@ intel_batchbuffer_free(struct intel_batchbuffer *batch)
 /* TODO: Push this whole function into bufmgr.
  */
 static void
-do_flush_locked(struct intel_batchbuffer *batch,
-		GLuint used, GLboolean allow_unlock)
+do_flush_locked(struct intel_batchbuffer *batch, GLuint used)
 {
    struct intel_context *intel = batch->intel;
    int ret = 0;
-   unsigned int num_cliprects = 0;
-   struct drm_clip_rect *cliprects = NULL;
    int x_off = 0, y_off = 0;
 
    if (batch->buffer)
@@ -146,31 +142,7 @@ do_flush_locked(struct intel_batchbuffer *batch,
    batch->map = NULL;
    batch->ptr = NULL;
 
-
-   if (batch->cliprect_mode == LOOP_CLIPRECTS) {
-      intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
-   }
-   /* Dispatch the batchbuffer, if it has some effect (nonzero cliprects).
-    * Can't short-circuit like this once we have hardware contexts, but we
-    * should always be in DRI2 mode by then anyway.
-    */
-   if ((batch->cliprect_mode != LOOP_CLIPRECTS ||
-	num_cliprects != 0) && !intel->no_hw) {
-      dri_bo_exec(batch->buf, used, cliprects, num_cliprects,
-		  (x_off & 0xffff) | (y_off << 16));
-   }
-
-   if (batch->cliprect_mode == LOOP_CLIPRECTS && num_cliprects == 0) {
-      if (allow_unlock) {
-	 /* If we are not doing any actual user-visible rendering,
-	  * do a sched_yield to keep the app from pegging the cpu while
-	  * achieving nothing.
-	  */
-         UNLOCK_HARDWARE(intel);
-         sched_yield();
-         LOCK_HARDWARE(intel);
-      }
-   }
+   dri_bo_exec(batch->buf, used, NULL, 0, (x_off & 0xffff) | (y_off << 16));
 
    if (INTEL_DEBUG & DEBUG_BATCH) {
       dri_bo_map(batch->buf, GL_FALSE);
@@ -183,7 +155,6 @@ do_flush_locked(struct intel_batchbuffer *batch,
    }
 
    if (ret != 0) {
-      UNLOCK_HARDWARE(intel);
       exit(1);
    }
    intel->vtbl.new_batch(intel);
@@ -201,24 +172,17 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
       drm_intel_bo_reference(intel->first_post_swapbuffers_batch);
    }
 
-   if (intel->first_post_swapbuffers_batch == NULL) {
-      intel->first_post_swapbuffers_batch = intel->batch->buf;
-      drm_intel_bo_reference(intel->first_post_swapbuffers_batch);
-   }
-
-   if (used == 0) {
-      batch->cliprect_mode = IGNORE_CLIPRECTS;
+   if (used == 0)
       return;
-   }
 
    if (INTEL_DEBUG & DEBUG_BATCH)
       fprintf(stderr, "%s:%d: Batchbuffer flush with %db used\n", file, line,
 	      used);
 
+   batch->reserved_space = 0;
    /* Emit a flush if the bufmgr doesn't do it for us. */
-   if (intel->always_flush_cache || !intel->ttm) {
-      *(GLuint *) (batch->ptr) = intel->vtbl.flush_cmd();
-      batch->ptr += 4;
+   if (intel->always_flush_cache) {
+      intel_batchbuffer_emit_mi_flush(batch);
       used = batch->ptr - batch->map;
    }
 
@@ -249,12 +213,15 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
    if (intel->vtbl.finish_batch)
       intel->vtbl.finish_batch(intel);
 
+   /* Check that we didn't just wrap our batchbuffer at a bad time. */
+   assert(!intel->no_batch_wrap);
+
+   batch->reserved_space = BATCH_RESERVED;
+
    /* TODO: Just pass the relocation list and dma buffer up to the
     * kernel.
     */
-   LOCK_HARDWARE(intel);
-   do_flush_locked(batch, used, GL_FALSE);
-   UNLOCK_HARDWARE(intel);
+   do_flush_locked(batch, used);
 
    if (INTEL_DEBUG & DEBUG_SYNC) {
       fprintf(stderr, "waiting for idle\n");
@@ -296,11 +263,38 @@ intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
 
 void
 intel_batchbuffer_data(struct intel_batchbuffer *batch,
-                       const void *data, GLuint bytes,
-		       enum cliprect_mode cliprect_mode)
+                       const void *data, GLuint bytes)
 {
    assert((bytes & 3) == 0);
-   intel_batchbuffer_require_space(batch, bytes, cliprect_mode);
+   intel_batchbuffer_require_space(batch, bytes);
    __memcpy(batch->ptr, data, bytes);
    batch->ptr += bytes;
 }
+
+/* Emit a pipelined flush to either flush render and texture cache for
+ * reading from a FBO-drawn texture, or flush so that frontbuffer
+ * render appears on the screen in DRI1.
+ *
+ * This is also used for the always_flush_cache driconf debug option.
+ */
+void
+intel_batchbuffer_emit_mi_flush(struct intel_batchbuffer *batch)
+{
+   struct intel_context *intel = batch->intel;
+
+   if (intel->gen >= 4) {
+      BEGIN_BATCH(4);
+      OUT_BATCH(_3DSTATE_PIPE_CONTROL |
+		PIPE_CONTROL_INSTRUCTION_FLUSH |
+		PIPE_CONTROL_WRITE_FLUSH |
+		PIPE_CONTROL_NO_WRITE);
+      OUT_BATCH(0); /* write address */
+      OUT_BATCH(0); /* write data */
+      OUT_BATCH(0); /* write data */
+      ADVANCE_BATCH();
+   } else {
+      BEGIN_BATCH(1);
+      OUT_BATCH(MI_FLUSH);
+      ADVANCE_BATCH();
+   }
+}
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
index 9a619fbd5c..b052b724d8 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -10,35 +10,6 @@
 #define BATCH_SZ 16384
 #define BATCH_RESERVED 16
 
-enum cliprect_mode {
-   /**
-    * Batchbuffer contents may be looped over per cliprect, but do not
-    * require it.
-    */
-   IGNORE_CLIPRECTS,
-   /**
-    * Batchbuffer contents require looping over per cliprect at batch submit
-    * time.
-    *
-    * This will be upgraded to NO_LOOP_CLIPRECTS when there's a single
-    * constant cliprect, as in DRI2 or FBO rendering.
-    */
-   LOOP_CLIPRECTS,
-   /**
-    * Batchbuffer contents contain drawing that should not be executed multiple
-    * times.
-    */
-   NO_LOOP_CLIPRECTS,
-   /**
-    * Batchbuffer contents contain drawing that already handles cliprects, such
-    * as 2D drawing to front/back/depth that doesn't respect DRAWING_RECTANGLE.
-    *
-    * Equivalent behavior to NO_LOOP_CLIPRECTS, but may not persist in batch
-    * outside of LOCK/UNLOCK.  This is upgraded to just NO_LOOP_CLIPRECTS when
-    * there's a constant cliprect, as in DRI2 or FBO rendering.
-    */
-   REFERENCES_CLIPRECTS
-};
 
 struct intel_batchbuffer
 {
@@ -51,8 +22,6 @@ struct intel_batchbuffer
    GLubyte *map;
    GLubyte *ptr;
 
-   enum cliprect_mode cliprect_mode;
-
    GLuint size;
 
    /** Tracking of BEGIN_BATCH()/OUT_BATCH()/ADVANCE_BATCH() debugging */
@@ -62,6 +31,7 @@ struct intel_batchbuffer
    } emit;
 
    GLuint dirty_state;
+   GLuint reserved_space;
 };
 
 struct intel_batchbuffer *intel_batchbuffer_alloc(struct intel_context
@@ -84,8 +54,7 @@ void intel_batchbuffer_reset(struct intel_batchbuffer *batch);
  * intel_buffer_dword() calls.
  */
 void intel_batchbuffer_data(struct intel_batchbuffer *batch,
-                            const void *data, GLuint bytes,
-			    enum cliprect_mode cliprect_mode);
+                            const void *data, GLuint bytes);
 
 void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
                                      GLuint bytes);
@@ -95,6 +64,7 @@ GLboolean intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
 				       uint32_t read_domains,
 				       uint32_t write_domain,
 				       uint32_t offset);
+void intel_batchbuffer_emit_mi_flush(struct intel_batchbuffer *batch);
 
 /* Inline functions - might actually be better off with these
  * non-inlined.  Certainly better off switching all command packets to
@@ -104,7 +74,7 @@ GLboolean intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
 static INLINE GLint
 intel_batchbuffer_space(struct intel_batchbuffer *batch)
 {
-   return (batch->size - BATCH_RESERVED) - (batch->ptr - batch->map);
+   return (batch->size - batch->reserved_space) - (batch->ptr - batch->map);
 }
 
 
@@ -119,36 +89,19 @@ intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch, GLuint dword)
 
 static INLINE void
 intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
-                                GLuint sz,
-				enum cliprect_mode cliprect_mode)
+                                GLuint sz)
 {
    assert(sz < batch->size - 8);
    if (intel_batchbuffer_space(batch) < sz)
       intel_batchbuffer_flush(batch);
-
-   if ((cliprect_mode == LOOP_CLIPRECTS ||
-	cliprect_mode == REFERENCES_CLIPRECTS) &&
-       batch->intel->constant_cliprect)
-      cliprect_mode = NO_LOOP_CLIPRECTS;
-
-   if (cliprect_mode != IGNORE_CLIPRECTS) {
-      if (batch->cliprect_mode == IGNORE_CLIPRECTS) {
-	 batch->cliprect_mode = cliprect_mode;
-      } else {
-	 if (batch->cliprect_mode != cliprect_mode) {
-	    intel_batchbuffer_flush(batch);
-	    batch->cliprect_mode = cliprect_mode;
-	 }
-      }
-   }
 }
 
 /* Here are the crusty old macros, to be removed:
  */
 #define BATCH_LOCALS
 
-#define BEGIN_BATCH(n, cliprect_mode) do {				\
-   intel_batchbuffer_require_space(intel->batch, (n)*4, cliprect_mode); \
+#define BEGIN_BATCH(n) do {				\
+   intel_batchbuffer_require_space(intel->batch, (n)*4); \
    assert(intel->batch->emit.start_ptr == NULL);			\
    intel->batch->emit.total = (n) * 4;					\
    intel->batch->emit.start_ptr = intel->batch->ptr;			\
@@ -157,7 +110,7 @@ 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((unsigned) (delta) <= buf->size);				\
+   assert((unsigned) (delta) < buf->size);				\
    intel_batchbuffer_emit_reloc(intel->batch, buf,			\
 				read_domains, write_domain, delta);	\
 } while (0)
@@ -173,12 +126,4 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
    intel->batch->emit.start_ptr = NULL;					\
 } while(0)
 
-
-static INLINE void
-intel_batchbuffer_emit_mi_flush(struct intel_batchbuffer *batch)
-{
-   intel_batchbuffer_require_space(batch, 4, IGNORE_CLIPRECTS);
-   intel_batchbuffer_emit_dword(batch, MI_FLUSH);
-}
-
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c
index 9e114db6c7..55bee0084c 100644
--- a/src/mesa/drivers/dri/intel/intel_blit.c
+++ b/src/mesa/drivers/dri/intel/intel_blit.c
@@ -29,7 +29,6 @@
 #include "main/mtypes.h"
 #include "main/context.h"
 #include "main/enums.h"
-#include "main/texformat.h"
 #include "main/colormac.h"
 
 #include "intel_blit.h"
@@ -43,137 +42,6 @@
 
 #define FILE_DEBUG_FLAG DEBUG_BLIT
 
-/**
- * Copy the back color buffer to the front color buffer. 
- * Used for SwapBuffers().
- */
-void
-intelCopyBuffer(const __DRIdrawablePrivate * dPriv,
-                const drm_clip_rect_t * rect)
-{
-
-   struct intel_context *intel;
-   const intelScreenPrivate *intelScreen;
-
-   DBG("%s\n", __FUNCTION__);
-
-   assert(dPriv);
-
-   intel = intelScreenContext(dPriv->driScreenPriv->private);
-   if (!intel)
-      return;
-
-   intelScreen = intel->intelScreen;
-
-   /* The LOCK_HARDWARE is required for the cliprects.  Buffer offsets
-    * should work regardless.
-    */
-   LOCK_HARDWARE(intel);
-
-   if (dPriv && dPriv->numClipRects) {
-      struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-      struct intel_region *src, *dst;
-      int nbox = dPriv->numClipRects;
-      drm_clip_rect_t *pbox = dPriv->pClipRects;
-      int cpp;
-      int src_pitch, dst_pitch;
-      unsigned short src_x, src_y;
-      int BR13, CMD;
-      int i;
-      dri_bo *aper_array[3];
-
-      src = intel_get_rb_region(&intel_fb->Base, BUFFER_BACK_LEFT);
-      dst = intel_get_rb_region(&intel_fb->Base, BUFFER_FRONT_LEFT);
-
-      src_pitch = src->pitch * src->cpp;
-      dst_pitch = dst->pitch * dst->cpp;
-
-      cpp = src->cpp;
-
-      ASSERT(intel_fb);
-      ASSERT(intel_fb->Base.Name == 0);    /* Not a user-created FBO */
-      ASSERT(src);
-      ASSERT(dst);
-      ASSERT(src->cpp == dst->cpp);
-
-      if (cpp == 2) {
-	 BR13 = (0xCC << 16) | BR13_565;
-	 CMD = XY_SRC_COPY_BLT_CMD;
-      }
-      else {
-	 BR13 = (0xCC << 16) | BR13_8888;
-	 CMD = XY_SRC_COPY_BLT_CMD | XY_BLT_WRITE_ALPHA | XY_BLT_WRITE_RGB;
-      }
-
-      assert(src->tiling != I915_TILING_Y);
-      assert(dst->tiling != I915_TILING_Y);
-#ifndef I915
-      if (src->tiling != I915_TILING_NONE) {
-	 CMD |= XY_SRC_TILED;
-	 src_pitch /= 4;
-      }
-      if (dst->tiling != I915_TILING_NONE) {
-	 CMD |= XY_DST_TILED;
-	 dst_pitch /= 4;
-      }
-#endif
-      /* do space/cliprects check before going any further */
-      intel_batchbuffer_require_space(intel->batch, 8 * 4,
-				      REFERENCES_CLIPRECTS);
-   again:
-      aper_array[0] = intel->batch->buf;
-      aper_array[1] = dst->buffer;
-      aper_array[2] = src->buffer;
-
-      if (dri_bufmgr_check_aperture_space(aper_array, 3) != 0) {
-	intel_batchbuffer_flush(intel->batch);
-	goto again;
-      }
-
-      for (i = 0; i < nbox; i++, pbox++) {
-	 drm_clip_rect_t box = *pbox;
-
-	 if (rect) {
-	    if (!intel_intersect_cliprects(&box, &box, rect))
-	       continue;
-	 }
-
-	 if (box.x1 >= box.x2 ||
-	     box.y1 >= box.y2)
-	    continue;
-
-	 assert(box.x1 < box.x2);
-	 assert(box.y1 < box.y2);
-	 src_x = box.x1 - dPriv->x + dPriv->backX;
-	 src_y = box.y1 - dPriv->y + dPriv->backY;
-
-	 BEGIN_BATCH(8, REFERENCES_CLIPRECTS);
-	 OUT_BATCH(CMD);
-	 OUT_BATCH(BR13 | dst_pitch);
-	 OUT_BATCH((box.y1 << 16) | box.x1);
-	 OUT_BATCH((box.y2 << 16) | box.x2);
-
-	 OUT_RELOC(dst->buffer,
-		   I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
-		   0);
-	 OUT_BATCH((src_y << 16) | src_x);
-	 OUT_BATCH(src_pitch);
-	 OUT_RELOC(src->buffer,
-		   I915_GEM_DOMAIN_RENDER, 0,
-		   0);
-	 ADVANCE_BATCH();
-      }
-
-      /* Flush the rendering and the batch so that the results all land on the
-       * screen in a timely fashion.
-       */
-      intel_batchbuffer_emit_mi_flush(intel->batch);
-      intel_batchbuffer_flush(intel->batch);
-   }
-
-   UNLOCK_HARDWARE(intel);
-}
-
 static GLuint translate_raster_op(GLenum logicop)
 {
    switch(logicop) {
@@ -249,7 +117,6 @@ intelEmitCopyBlit(struct intel_context *intel,
    } while (pass < 2);
 
    if (pass >= 2) {
-       LOCK_HARDWARE(intel);
        dri_bo_map(dst_buffer, GL_TRUE);
        dri_bo_map(src_buffer, GL_FALSE);
        _mesa_copy_rect((GLubyte *)dst_buffer->virtual + dst_offset,
@@ -263,12 +130,11 @@ intelEmitCopyBlit(struct intel_context *intel,
        
        dri_bo_unmap(src_buffer);
        dri_bo_unmap(dst_buffer);
-       UNLOCK_HARDWARE(intel);
 
        return GL_TRUE;
    }
 
-   intel_batchbuffer_require_space(intel->batch, 8 * 4, NO_LOOP_CLIPRECTS);
+   intel_batchbuffer_require_space(intel->batch, 8 * 4);
    DBG("%s src:buf(%p)/%d+%d %d,%d dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n",
        __FUNCTION__,
        src_buffer, src_pitch, src_offset, src_x, src_y,
@@ -313,7 +179,7 @@ intelEmitCopyBlit(struct intel_context *intel,
    assert(dst_x < dst_x2);
    assert(dst_y < dst_y2);
 
-   BEGIN_BATCH(8, NO_LOOP_CLIPRECTS);
+   BEGIN_BATCH(8);
    OUT_BATCH(CMD);
    OUT_BATCH(BR13 | (uint16_t)dst_pitch);
    OUT_BATCH((dst_y << 16) | dst_x);
@@ -371,8 +237,6 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
       skipBuffers = BUFFER_BIT_STENCIL;
    }
 
-   LOCK_HARDWARE(intel);
-
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
    if (num_cliprects) {
       GLint cx, cy, cw, ch;
@@ -436,6 +300,10 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
                   intel_region_buffer(intel, irb->region,
                                       all ? INTEL_WRITE_FULL :
                                       INTEL_WRITE_PART);
+	       int x1 = b.x1 + irb->region->draw_x;
+	       int y1 = b.y1 + irb->region->draw_y;
+	       int x2 = b.x2 + irb->region->draw_x;
+	       int y2 = b.y2 + irb->region->draw_y;
 
                GLuint clearVal;
                GLint pitch, cpp;
@@ -444,11 +312,10 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
                pitch = irb->region->pitch;
                cpp = irb->region->cpp;
 
-               DBG("%s dst:buf(%p)/%d+%d %d,%d sz:%dx%d\n",
+               DBG("%s dst:buf(%p)/%d %d,%d sz:%dx%d\n",
                    __FUNCTION__,
                    irb->region->buffer, (pitch * cpp),
-                   irb->region->draw_offset,
-                   b.x1, b.y1, b.x2 - b.x1, b.y2 - b.y1);
+                   x1, y1, x2 - x1, y2 - y1);
 
 	       BR13 = 0xf0 << 16;
 	       CMD = XY_COLOR_BLT_CMD;
@@ -494,12 +361,14 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
 		  CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
 		  CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
 
-		  switch (irb->texformat->MesaFormat) {
+		  switch (irb->Base.Format) {
 		  case MESA_FORMAT_ARGB8888:
-		     clearVal = intel->ClearColor8888;
+		  case MESA_FORMAT_XRGB8888:
+		     clearVal = PACK_COLOR_8888(clear[3], clear[0],
+						clear[1], clear[2]);
 		     break;
 		  case MESA_FORMAT_RGB565:
-		     clearVal = intel->ClearColor565;
+		     clearVal = PACK_COLOR_565(clear[0], clear[1], clear[2]);
 		     break;
 		  case MESA_FORMAT_ARGB4444:
 		     clearVal = PACK_COLOR_4444(clear[3], clear[0],
@@ -511,7 +380,7 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
 		     break;
 		  default:
 		     _mesa_problem(ctx, "Unexpected renderbuffer format: %d\n",
-				   irb->texformat->MesaFormat);
+				   irb->Base.Format);
 		     clearVal = 0;
 		  }
 	       }
@@ -521,17 +390,17 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
                   buf, irb->Base.Name);
                 */
 
-               assert(b.x1 < b.x2);
-               assert(b.y1 < b.y2);
+               assert(x1 < x2);
+               assert(y1 < y2);
 
-               BEGIN_BATCH(6, REFERENCES_CLIPRECTS);
+               BEGIN_BATCH(6);
                OUT_BATCH(CMD);
                OUT_BATCH(BR13);
-               OUT_BATCH((b.y1 << 16) | b.x1);
-               OUT_BATCH((b.y2 << 16) | b.x2);
+               OUT_BATCH((y1 << 16) | x1);
+               OUT_BATCH((y2 << 16) | x2);
                OUT_RELOC(write_buffer,
 			 I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
-                         irb->region->draw_offset);
+                         0);
                OUT_BATCH(clearVal);
                ADVANCE_BATCH();
                clearMask &= ~bufBit;    /* turn off bit, for faster loop exit */
@@ -539,8 +408,6 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
          }
       }
    }
-
-   UNLOCK_HARDWARE(intel);
 }
 
 GLboolean
@@ -582,8 +449,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
    intel_batchbuffer_require_space( intel->batch,
 				    (8 * 4) +
 				    (3 * 4) +
-				    dwords * 4,
-				    REFERENCES_CLIPRECTS );
+				    dwords * 4 );
 
    opcode = XY_SETUP_BLT_CMD;
    if (cpp == 4)
@@ -605,7 +471,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
    if (dst_tiling != I915_TILING_NONE)
       blit_cmd |= XY_DST_TILED;
 
-   BEGIN_BATCH(8 + 3, REFERENCES_CLIPRECTS);
+   BEGIN_BATCH(8 + 3);
    OUT_BATCH(opcode);
    OUT_BATCH(br13);
    OUT_BATCH((0 << 16) | 0); /* clip x1, y1 */
@@ -624,8 +490,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
 
    intel_batchbuffer_data( intel->batch,
 			   src_bits,
-			   dwords * 4,
-			   REFERENCES_CLIPRECTS );
+			   dwords * 4 );
 
    intel_batchbuffer_emit_mi_flush(intel->batch);
 
diff --git a/src/mesa/drivers/dri/intel/intel_blit.h b/src/mesa/drivers/dri/intel/intel_blit.h
index 240cb7cd1b..eb66fe0481 100644
--- a/src/mesa/drivers/dri/intel/intel_blit.h
+++ b/src/mesa/drivers/dri/intel/intel_blit.h
@@ -30,7 +30,7 @@
 
 #include "intel_context.h"
 
-extern void intelCopyBuffer(const __DRIdrawablePrivate * dpriv,
+extern void intelCopyBuffer(const __DRIdrawable * dpriv,
                             const drm_clip_rect_t * rect);
 
 extern void intelClearWithBlit(GLcontext * ctx, GLbitfield mask);
diff --git a/src/mesa/drivers/dri/intel/intel_buffer_objects.c b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
index a0225936c8..3b7015b5ad 100644
--- a/src/mesa/drivers/dri/intel/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/intel/intel_buffer_objects.c
@@ -209,9 +209,23 @@ 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_busy(intel_obj->buffer) ||
+	  drm_intel_bo_references(intel->batch->buf, intel_obj->buffer)) {
+	 drm_intel_bo *temp_bo;
+
+	 temp_bo = drm_intel_bo_alloc(intel->bufmgr, "subdata temp", size, 64);
+
+	 drm_intel_bo_subdata(temp_bo, 0, size, data);
 
-      dri_bo_subdata(intel_obj->buffer, offset, size, data);
+	 intel_emit_linear_blit(intel,
+				intel_obj->buffer, offset,
+				temp_bo, 0,
+				size);
+
+	 drm_intel_bo_unreference(temp_bo);
+      } else {
+	 dri_bo_subdata(intel_obj->buffer, offset, size, data);
+      }
    }
 }
 
@@ -254,13 +268,14 @@ intel_bufferobj_map(GLcontext * ctx,
 
    if (intel_obj->sys_buffer) {
       obj->Pointer = intel_obj->sys_buffer;
+      obj->Length = obj->Size;
+      obj->Offset = 0;
       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 +345,8 @@ 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_buffers.c b/src/mesa/drivers/dri/intel/intel_buffers.c
index e7357e78c5..7c4b79f743 100644
--- a/src/mesa/drivers/dri/intel/intel_buffers.c
+++ b/src/mesa/drivers/dri/intel/intel_buffers.c
@@ -102,32 +102,33 @@ intel_get_cliprects(struct intel_context *intel,
 		    unsigned int *num_cliprects,
 		    int *x_off, int *y_off)
 {
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (intel->constant_cliprect) {
-      /* FBO or DRI2 rendering, which can just use the fb's size. */
-      intel->fboRect.x1 = 0;
-      intel->fboRect.y1 = 0;
-      intel->fboRect.x2 = intel->ctx.DrawBuffer->Width;
-      intel->fboRect.y2 = intel->ctx.DrawBuffer->Height;
-
-      *cliprects = &intel->fboRect;
-      *num_cliprects = 1;
-      *x_off = 0;
-      *y_off = 0;
-   } else if (intel->front_cliprects || dPriv->numBackClipRects == 0) {
-      /* use the front clip rects */
-      *cliprects = dPriv->pClipRects;
-      *num_cliprects = dPriv->numClipRects;
-      *x_off = dPriv->x;
-      *y_off = dPriv->y;
-   }
-   else {
-      /* use the back clip rects */
-      *num_cliprects = dPriv->numBackClipRects;
-      *cliprects = dPriv->pBackClipRects;
-      *x_off = dPriv->backX;
-      *y_off = dPriv->backY;
+   intel->fboRect.x1 = 0;
+   intel->fboRect.y1 = 0;
+   intel->fboRect.x2 = intel->ctx.DrawBuffer->Width;
+   intel->fboRect.y2 = intel->ctx.DrawBuffer->Height;
+
+   *cliprects = &intel->fboRect;
+   *num_cliprects = 1;
+   *x_off = 0;
+   *y_off = 0;
+}
+
+
+/**
+ * Check if we're about to draw into the front color buffer.
+ * If so, set the intel->front_buffer_dirty field to true.
+ */
+void
+intel_check_front_buffer_rendering(struct intel_context *intel)
+{
+   const struct gl_framebuffer *fb = intel->ctx.DrawBuffer;
+   if (fb->Name == 0) {
+      /* drawing to window system buffer */
+      if (fb->_NumColorDrawBuffers > 0) {
+         if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
+	    intel->front_buffer_dirty = GL_TRUE;
+	 }
+      }
    }
 }
 
@@ -172,13 +173,17 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       return;
    }
 
-   /*
-    * How many color buffers are we drawing into?
+   /* How many color buffers are we drawing into?
+    *
+    * If there are zero buffers or the buffer is too big, don't configure any
+    * regions for hardware drawing.  We'll fallback to software below.  Not
+    * having regions set makes some of the software fallback paths faster.
     */
-   if (fb->_NumColorDrawBuffers == 0) {
+   if ((fb->Width > ctx->Const.MaxRenderbufferSize)
+       || (fb->Height > ctx->Const.MaxRenderbufferSize)
+       || (fb->_NumColorDrawBuffers == 0)) {
       /* writing to 0  */
       colorRegions[0] = NULL;
-      intel->constant_cliprect = GL_TRUE;
    }
    else if (fb->_NumColorDrawBuffers > 1) {
        int i;
@@ -188,36 +193,23 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
            irb = intel_renderbuffer(fb->_ColorDrawBuffers[i]);
            colorRegions[i] = irb ? irb->region : NULL;
        }
-       intel->constant_cliprect = GL_TRUE;
    }
    else {
       /* Get the intel_renderbuffer for the single colorbuffer we're drawing
-       * into, and set up cliprects if it's .
+       * into.
        */
       if (fb->Name == 0) {
-	 intel->constant_cliprect = intel->driScreen->dri2.enabled;
 	 /* drawing to window system buffer */
-	 if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
-	    if (!intel->constant_cliprect && !intel->front_cliprects)
-	       intel_batchbuffer_flush(intel->batch);
-	    intel->front_cliprects = GL_TRUE;
+	 if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT)
 	    colorRegions[0] = intel_get_rb_region(fb, BUFFER_FRONT_LEFT);
-
-	    intel->front_buffer_dirty = GL_TRUE;
-	 }
-	 else {
-	    if (!intel->constant_cliprect && intel->front_cliprects)
-	       intel_batchbuffer_flush(intel->batch);
-	    intel->front_cliprects = GL_FALSE;
-	    colorRegions[0]= intel_get_rb_region(fb, BUFFER_BACK_LEFT);
-	 }
+	 else
+	    colorRegions[0] = intel_get_rb_region(fb, BUFFER_BACK_LEFT);
       }
       else {
 	 /* drawing to user-created FBO */
 	 struct intel_renderbuffer *irb;
 	 irb = intel_renderbuffer(fb->_ColorDrawBuffers[0]);
 	 colorRegions[0] = (irb && irb->region) ? irb->region : NULL;
-	 intel->constant_cliprect = GL_TRUE;
       }
    }
 
@@ -257,7 +249,7 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
    if (fb->_StencilBuffer && fb->_StencilBuffer->Wrapped) {
       irbStencil = intel_renderbuffer(fb->_StencilBuffer->Wrapped);
       if (irbStencil && irbStencil->region) {
-         ASSERT(irbStencil->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
+         ASSERT(irbStencil->Base.Format == MESA_FORMAT_S8_Z24);
          FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
       }
       else {
@@ -269,6 +261,12 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
    }
 
+   /* If we have a (packed) stencil buffer attached but no depth buffer,
+    * we still need to set up the shared depth/stencil state so we can use it.
+    */
+   if (depthRegion == NULL && irbStencil && irbStencil->region)
+      depthRegion = irbStencil->region;
+
    /*
     * Update depth and stencil test state
     */
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.h b/src/mesa/drivers/dri/intel/intel_buffers.h
index 6069d38e9e..d7800f2ca2 100644
--- a/src/mesa/drivers/dri/intel/intel_buffers.h
+++ b/src/mesa/drivers/dri/intel/intel_buffers.h
@@ -45,6 +45,8 @@ extern struct intel_region *intel_readbuf_region(struct intel_context *intel);
 
 extern struct intel_region *intel_drawbuf_region(struct intel_context *intel);
 
+extern void intel_check_front_buffer_rendering(struct intel_context *intel);
+
 extern void intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb);
 
 extern void intelInitBufferFuncs(struct dd_function_table *functions);
diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c
index fb62f0f430..956f2339ff 100644
--- a/src/mesa/drivers/dri/intel/intel_clear.c
+++ b/src/mesa/drivers/dri/intel/intel_clear.c
@@ -68,13 +68,17 @@ static void
 intelClear(GLcontext *ctx, GLbitfield mask)
 {
    struct intel_context *intel = intel_context(ctx);
-   const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
+   const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask[0]);
    GLbitfield tri_mask = 0;
    GLbitfield blit_mask = 0;
    GLbitfield swrast_mask = 0;
    struct gl_framebuffer *fb = ctx->DrawBuffer;
    GLuint i;
 
+   if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
+      intel->front_buffer_dirty = GL_TRUE;
+   }
+
    if (0)
       fprintf(stderr, "%s\n", __FUNCTION__);
 
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 934f7aa187..13a28f0419 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -55,10 +55,8 @@
 #include "intel_decode.h"
 #include "intel_bufmgr.h"
 #include "intel_screen.h"
-#include "intel_swapbuffers.h"
 
 #include "drirenderbuffer.h"
-#include "vblank.h"
 #include "utils.h"
 #include "xmlpool.h"            /* for symbolic values of enum-type options */
 
@@ -68,7 +66,7 @@ int INTEL_DEBUG = (0);
 #endif
 
 
-#define DRIVER_DATE                     "20090712 2009Q2 RC3"
+#define DRIVER_DATE                     "20091221 DEVELOPMENT"
 #define DRIVER_DATE_GEM                 "GEM " DRIVER_DATE
 
 
@@ -176,9 +174,7 @@ intelGetString(GLcontext * ctx, GLenum name)
          break;
       }
 
-      (void) driGetRendererString(buffer, chipset, 
-				  (intel->ttm) ? DRIVER_DATE_GEM : DRIVER_DATE,
-				  0);
+      (void) driGetRendererString(buffer, chipset, DRIVER_DATE_GEM, 0);
       return (GLubyte *) buffer;
 
    default:
@@ -189,28 +185,17 @@ intelGetString(GLcontext * ctx, GLenum name)
 static unsigned
 intel_bits_per_pixel(const struct intel_renderbuffer *rb)
 {
-   switch (rb->Base._ActualFormat) {
-   case GL_RGB5:
-   case GL_DEPTH_COMPONENT16:
-      return 16;
-   case GL_RGB8:
-   case GL_RGBA8:
-   case GL_DEPTH_COMPONENT24:
-   case GL_DEPTH24_STENCIL8_EXT:
-   case GL_STENCIL_INDEX8_EXT:
-      return 32;
-   default:
-      return 0;
-   }
+   return _mesa_get_format_bytes(rb->Base.Format) * 8;
 }
 
 void
 intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 {
-   struct intel_framebuffer *intel_fb = drawable->driverPrivate;
+   struct gl_framebuffer *fb = drawable->driverPrivate;
    struct intel_renderbuffer *rb;
    struct intel_region *region, *depth_region;
    struct intel_context *intel = context->driverPrivate;
+   struct intel_renderbuffer *front_rb, *back_rb, *depth_rb, *stencil_rb;
    __DRIbuffer *buffers = NULL;
    __DRIscreen *screen;
    int i, count;
@@ -226,26 +211,25 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
    if (screen->dri2.loader
        && (screen->dri2.loader->base.version > 2)
        && (screen->dri2.loader->getBuffersWithFormat != NULL)) {
-      struct intel_renderbuffer *depth_rb;
-      struct intel_renderbuffer *stencil_rb;
+
+      front_rb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
+      back_rb = intel_get_renderbuffer(fb, BUFFER_BACK_LEFT);
+      depth_rb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
+      stencil_rb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
 
       i = 0;
       if ((intel->is_front_buffer_rendering ||
 	   intel->is_front_buffer_reading ||
-	   !intel_fb->color_rb[1])
-	   && intel_fb->color_rb[0]) {
+	   !back_rb) && front_rb) {
 	 attachments[i++] = __DRI_BUFFER_FRONT_LEFT;
-	 attachments[i++] = intel_bits_per_pixel(intel_fb->color_rb[0]);
+	 attachments[i++] = intel_bits_per_pixel(front_rb);
       }
 
-      if (intel_fb->color_rb[1]) {
+      if (back_rb) {
 	 attachments[i++] = __DRI_BUFFER_BACK_LEFT;
-	 attachments[i++] = intel_bits_per_pixel(intel_fb->color_rb[1]);
+	 attachments[i++] = intel_bits_per_pixel(back_rb);
       }
 
-      depth_rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
-      stencil_rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
-
       if ((depth_rb != NULL) && (stencil_rb != NULL)) {
 	 attachments[i++] = __DRI_BUFFER_DEPTH_STENCIL;
 	 attachments[i++] = intel_bits_per_pixel(depth_rb);
@@ -266,13 +250,13 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 						      drawable->loaderPrivate);
    } else if (screen->dri2.loader) {
       i = 0;
-      if (intel_fb->color_rb[0])
+      if (intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT))
 	 attachments[i++] = __DRI_BUFFER_FRONT_LEFT;
-      if (intel_fb->color_rb[1])
+      if (intel_get_renderbuffer(fb, BUFFER_BACK_LEFT))
 	 attachments[i++] = __DRI_BUFFER_BACK_LEFT;
-      if (intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH))
+      if (intel_get_renderbuffer(fb, BUFFER_DEPTH))
 	 attachments[i++] = __DRI_BUFFER_DEPTH;
-      if (intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL))
+      if (intel_get_renderbuffer(fb, BUFFER_STENCIL))
 	 attachments[i++] = __DRI_BUFFER_STENCIL;
 
       buffers = (*screen->dri2.loader->getBuffers)(drawable,
@@ -305,32 +289,32 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
    for (i = 0; i < count; i++) {
        switch (buffers[i].attachment) {
        case __DRI_BUFFER_FRONT_LEFT:
-	   rb = intel_fb->color_rb[0];
+	   rb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
 	   region_name = "dri2 front buffer";
 	   break;
 
        case __DRI_BUFFER_FAKE_FRONT_LEFT:
-	   rb = intel_fb->color_rb[0];
+	   rb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
 	   region_name = "dri2 fake front buffer";
 	   break;
 
        case __DRI_BUFFER_BACK_LEFT:
-	   rb = intel_fb->color_rb[1];
+	   rb = intel_get_renderbuffer(fb, BUFFER_BACK_LEFT);
 	   region_name = "dri2 back buffer";
 	   break;
 
        case __DRI_BUFFER_DEPTH:
-	   rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
+	   rb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
 	   region_name = "dri2 depth buffer";
 	   break;
 
        case __DRI_BUFFER_DEPTH_STENCIL:
-	   rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
+	   rb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
 	   region_name = "dri2 depth / stencil buffer";
 	   break;
 
        case __DRI_BUFFER_STENCIL:
-	   rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
+	   rb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
 	   region_name = "dri2 stencil buffer";
 	   break;
 
@@ -377,7 +361,7 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
        intel_region_release(&region);
 
        if (buffers[i].attachment == __DRI_BUFFER_DEPTH_STENCIL) {
-	  rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
+	  rb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
 	  if (rb != NULL) {
 	     struct intel_region *stencil_region = NULL;
 
@@ -405,9 +389,6 @@ intel_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
     void (*old_viewport)(GLcontext *ctx, GLint x, GLint y,
 			 GLsizei w, GLsizei h);
 
-    if (!driContext->driScreenPriv->dri2.enabled)
-	return;
-
     if (!intel->meta.internal_viewport_call && ctx->DrawBuffer->Name == 0) {
        /* If we're rendering to the fake front buffer, make sure all the pending
 	* drawing has landed on the real front buffer.  Otherwise when we
@@ -426,7 +407,6 @@ intel_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
     old_viewport = ctx->Driver.Viewport;
     ctx->Driver.Viewport = NULL;
     intel->driDrawable = driContext->driDrawablePriv;
-    intelWindowMoved(intel);
     intel_draw_buffer(ctx, intel->ctx.DrawBuffer);
     ctx->Driver.Viewport = old_viewport;
 }
@@ -489,16 +469,9 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush)
    if (intel->Fallback)
       _swrast_flush(ctx);
 
-   if (!IS_965(intel->intelScreen->deviceID))
+   if (intel->gen < 4)
       INTEL_FIREVERTICES(intel);
 
-   /* Emit a flush so that any frontbuffer rendering that might have occurred
-    * lands onscreen in a timely manner, even if the X Server doesn't trigger
-    * a flush for us.
-    */
-   if (needs_mi_flush)
-      intel_batchbuffer_emit_mi_flush(intel->batch);
-
    if (intel->batch->map != intel->batch->ptr)
       intel_batchbuffer_flush(intel->batch);
 
@@ -604,15 +577,15 @@ intelInitDriverFunctions(struct dd_function_table *functions)
 GLboolean
 intelInitContext(struct intel_context *intel,
                  const __GLcontextModes * mesaVis,
-                 __DRIcontextPrivate * driContextPriv,
+                 __DRIcontext * driContextPriv,
                  void *sharedContextPrivate,
                  struct dd_function_table *functions)
 {
    GLcontext *ctx = &intel->ctx;
    GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-   int fthrottle_mode;
+   int bo_reuse_mode;
 
    if (!_mesa_initialize_context(&intel->ctx, mesaVis, shareCtx,
                                  functions, (void *) intel)) {
@@ -623,35 +596,46 @@ intelInitContext(struct intel_context *intel,
    driContextPriv->driverPrivate = intel;
    intel->intelScreen = intelScreen;
    intel->driScreen = sPriv;
-   intel->sarea = intelScreen->sarea;
    intel->driContext = driContextPriv;
-
-   /* Dri stuff */
-   intel->hHWContext = driContextPriv->hHWContext;
    intel->driFd = sPriv->fd;
-   intel->driHwLock = sPriv->lock;
+
+   if (IS_965(intel->intelScreen->deviceID)) {
+      intel->gen = 4;
+   } else if (IS_9XX(intel->intelScreen->deviceID)) {
+      intel->gen = 3;
+      if (IS_945(intel->intelScreen->deviceID)) {
+	 intel->is_945 = GL_TRUE;
+      }
+   } else {
+      intel->gen = 2;
+   }
+
+   if (IS_IGDNG(intel->intelScreen->deviceID)) {
+      intel->is_ironlake = GL_TRUE;
+      intel->needs_ff_sync = GL_TRUE;
+      intel->has_luminance_srgb = GL_TRUE;
+   } else if (IS_G4X(intel->intelScreen->deviceID)) {
+      intel->has_luminance_srgb = GL_TRUE;
+      intel->is_g4x = GL_TRUE;
+   }
 
    driParseConfigFiles(&intel->optionCache, &intelScreen->optionCache,
                        intel->driScreen->myNum,
-		       IS_965(intelScreen->deviceID) ? "i965" : "i915");
+		       (intel->gen >= 4) ? "i965" : "i915");
    if (intelScreen->deviceID == PCI_CHIP_I865_G)
       intel->maxBatchSize = 4096;
    else
       intel->maxBatchSize = BATCH_SZ;
 
    intel->bufmgr = intelScreen->bufmgr;
-   intel->ttm = intelScreen->ttm;
-   if (intel->ttm) {
-      int bo_reuse_mode;
 
-      bo_reuse_mode = driQueryOptioni(&intel->optionCache, "bo_reuse");
-      switch (bo_reuse_mode) {
-      case DRI_CONF_BO_REUSE_DISABLED:
-	 break;
-      case DRI_CONF_BO_REUSE_ALL:
-	 intel_bufmgr_gem_enable_reuse(intel->bufmgr);
-	 break;
-      }
+   bo_reuse_mode = driQueryOptioni(&intel->optionCache, "bo_reuse");
+   switch (bo_reuse_mode) {
+   case DRI_CONF_BO_REUSE_DISABLED:
+      break;
+   case DRI_CONF_BO_REUSE_ALL:
+      intel_bufmgr_gem_enable_reuse(intel->bufmgr);
+      break;
    }
 
    /* This doesn't yet catch all non-conformant rendering, but it's a
@@ -695,7 +679,7 @@ intelInitContext(struct intel_context *intel,
 
    meta_init_metaops(ctx, &intel->meta);
    ctx->Const.MaxColorAttachments = 4;  /* XXX FBO: review this */
-   if (IS_965(intelScreen->deviceID)) {
+   if (intel->gen >= 4) {
       if (MAX_WIDTH > 8192)
 	 ctx->Const.MaxRenderbufferSize = 8192;
    } else {
@@ -732,33 +716,22 @@ intelInitContext(struct intel_context *intel,
       break;
    }
 
-   if (IS_965(intelScreen->deviceID))
+   if (intel->gen >= 4)
       intel->polygon_offset_scale /= 0xffff;
 
    intel->RenderIndex = ~0;
 
-   fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
-   intel->irqsEmitted = 0;
-
-   intel->do_irqs = (intel->intelScreen->irq_active &&
-                     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
-
-   intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
-
-   if (IS_965(intelScreen->deviceID) && !intel->intelScreen->irq_active) {
+   if (intel->gen >= 4 && !intel->intelScreen->irq_active) {
       _mesa_printf("IRQs not active.  Exiting\n");
       exit(1);
    }
 
-   intelInitExtensions(ctx, GL_FALSE);
+   intelInitExtensions(ctx);
 
    INTEL_DEBUG = driParseDebugString(getenv("INTEL_DEBUG"), debug_control);
    if (INTEL_DEBUG & DEBUG_BUFMGR)
       dri_bufmgr_set_debug(intel->bufmgr, GL_TRUE);
 
-   if (!sPriv->dri2.enabled)
-      intel_recreate_static_regions(intel);
-
    intel->batch = intel_batchbuffer_alloc(intel);
 
    intel_fbo_init(intel);
@@ -807,7 +780,7 @@ intelInitContext(struct intel_context *intel,
 }
 
 void
-intelDestroyContext(__DRIcontextPrivate * driContextPriv)
+intelDestroyContext(__DRIcontext * driContextPriv)
 {
    struct intel_context *intel =
       (struct intel_context *) driContextPriv->driverPrivate;
@@ -830,7 +803,7 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
       _vbo_DestroyContext(&intel->ctx);
 
       _swrast_DestroyContext(&intel->ctx);
-      intel->Fallback = 0;      /* don't call _swrast_Flush later */
+      intel->Fallback = 0x0;      /* don't call _swrast_Flush later */
 
       intel_batchbuffer_free(intel->batch);
       intel->batch = NULL;
@@ -854,57 +827,6 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
           */
       }
 
-      /* XXX In intelMakeCurrent() below, the context's static regions are 
-       * referenced inside the frame buffer; it's listed as a hack,
-       * with a comment of "XXX FBO temporary fix-ups!", but
-       * as long as it's there, we should release the regions here.
-       * The do/while loop around the block is used to allow the
-       * "continue" statements inside the block to exit the block,
-       * to avoid many layers of "if" constructs.
-       */
-      do {
-         __DRIdrawablePrivate * driDrawPriv = intel->driDrawable;
-         struct intel_framebuffer *intel_fb;
-         struct intel_renderbuffer *irbDepth, *irbStencil;
-         if (!driDrawPriv) {
-            /* We're already detached from the drawable; exit this block. */
-            continue;
-         }
-         intel_fb = (struct intel_framebuffer *) driDrawPriv->driverPrivate;
-         if (!intel_fb) {
-            /* The frame buffer is already gone; exit this block. */
-            continue;
-         }
-         irbDepth = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
-         irbStencil = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
-
-         /* If the regions of the frame buffer still match the regions
-          * of the context, release them.  If they've changed somehow,
-          * leave them alone.
-          */
-         if (intel_fb->color_rb[0] && intel_fb->color_rb[0]->region == intel->front_region) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[0], NULL);
-         }
-         if (intel_fb->color_rb[1] && intel_fb->color_rb[1]->region == intel->back_region) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[1], NULL);
-         }
-
-         if (irbDepth && irbDepth->region == intel->depth_region) {
-	    intel_renderbuffer_set_region(irbDepth, NULL);
-         }
-         /* Usually, the stencil buffer is the same as the depth buffer;
-          * but they're handled separately in MakeCurrent, so we'll
-          * handle them separately here.
-          */
-         if (irbStencil && irbStencil->region == intel->depth_region) {
-	    intel_renderbuffer_set_region(irbStencil, NULL);
-         }
-      } while (0);
-
-      intel_region_release(&intel->front_region);
-      intel_region_release(&intel->back_region);
-      intel_region_release(&intel->depth_region);
-
       driDestroyOptionCache(&intel->optionCache);
 
       /* free the Mesa context */
@@ -916,7 +838,7 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
 }
 
 GLboolean
-intelUnbindContext(__DRIcontextPrivate * driContextPriv)
+intelUnbindContext(__DRIcontext * driContextPriv)
 {
    struct intel_context *intel =
       (struct intel_context *) driContextPriv->driverPrivate;
@@ -930,50 +852,33 @@ intelUnbindContext(__DRIcontextPrivate * driContextPriv)
 }
 
 GLboolean
-intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
-                 __DRIdrawablePrivate * driDrawPriv,
-                 __DRIdrawablePrivate * driReadPriv)
+intelMakeCurrent(__DRIcontext * driContextPriv,
+                 __DRIdrawable * driDrawPriv,
+                 __DRIdrawable * driReadPriv)
 {
-   __DRIscreenPrivate *psp = driDrawPriv->driScreenPriv;
+   struct intel_context *intel;
+   GET_CURRENT_CONTEXT(curCtx);
+
+   if (driContextPriv)
+      intel = (struct intel_context *) driContextPriv->driverPrivate;
+   else
+      intel = NULL;
+
+   /* According to the glXMakeCurrent() man page: "Pending commands to
+    * the previous context, if any, are flushed before it is released."
+    * But only flush if we're actually changing contexts.
+    */
+   if (intel_context(curCtx) && intel_context(curCtx) != intel) {
+      _mesa_flush(curCtx);
+   }
 
    if (driContextPriv) {
-      struct intel_context *intel =
-         (struct intel_context *) driContextPriv->driverPrivate;
-      struct intel_framebuffer *intel_fb =
-	 (struct intel_framebuffer *) driDrawPriv->driverPrivate;
-      GLframebuffer *readFb = (GLframebuffer *) driReadPriv->driverPrivate;
+      struct gl_framebuffer *fb = driDrawPriv->driverPrivate;
+      struct gl_framebuffer *readFb = driReadPriv->driverPrivate;
  
-      if (driContextPriv->driScreenPriv->dri2.enabled) {     
-          intel_update_renderbuffers(driContextPriv, driDrawPriv);
-          if (driDrawPriv != driReadPriv)
-              intel_update_renderbuffers(driContextPriv, driReadPriv);
-      } else {
-          /* XXX FBO temporary fix-ups!  These are released in 
-           * intelDextroyContext(), above.  Changes here should be
-           * reflected there.
-           */
-          /* if the renderbuffers don't have regions, init them from the context */
-         struct intel_renderbuffer *irbDepth
-            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
-         struct intel_renderbuffer *irbStencil
-            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
-
-         if (intel_fb->color_rb[0]) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[0],
-					  intel->front_region);
-         }
-         if (intel_fb->color_rb[1]) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[1],
-					  intel->back_region);
-         }
-
-         if (irbDepth) {
-	    intel_renderbuffer_set_region(irbDepth, intel->depth_region);
-         }
-         if (irbStencil) {
-	    intel_renderbuffer_set_region(irbStencil, intel->depth_region);
-         }
-      }
+      intel_update_renderbuffers(driContextPriv, driDrawPriv);
+      if (driDrawPriv != driReadPriv)
+	 intel_update_renderbuffers(driContextPriv, driReadPriv);
 
       /* set GLframebuffer size to match window, if needed */
       driUpdateFramebufferSize(&intel->ctx, driDrawPriv);
@@ -982,37 +887,10 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 	 driUpdateFramebufferSize(&intel->ctx, driReadPriv);
       }
 
-      _mesa_make_current(&intel->ctx, &intel_fb->Base, readFb);
-
+      _mesa_make_current(&intel->ctx, fb, readFb);
       intel->driReadDrawable = driReadPriv;
-
-      if (intel->driDrawable != driDrawPriv) {
-         if (driDrawPriv->swap_interval == (unsigned)-1) {
-            int i;
-
-            driDrawPriv->vblFlags = (intel->intelScreen->irq_active != 0)
-               ? driGetDefaultVBlankFlags(&intel->optionCache)
-               : VBLANK_FLAG_NO_IRQ;
-
-            /* Prevent error printf if one crtc is disabled, this will
-             * be properly calculated in intelWindowMoved() next.
-             */
-            driDrawPriv->vblFlags = intelFixupVblank(intel, driDrawPriv);
-
-            (*psp->systemTime->getUST) (&intel_fb->swap_ust);
-            driDrawableInitVBlank(driDrawPriv);
-            intel_fb->vbl_waited = driDrawPriv->vblSeq;
-
-            for (i = 0; i < 2; i++) {
-               if (intel_fb->color_rb[i])
-                  intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
-            }
-         }
-         intel->driDrawable = driDrawPriv;
-         intelWindowMoved(intel);
-      }
-
-      intel_draw_buffer(&intel->ctx, &intel_fb->Base);
+      intel->driDrawable = driDrawPriv;
+      intel_draw_buffer(&intel->ctx, fb);
    }
    else {
       _mesa_make_current(NULL, NULL, NULL);
@@ -1020,143 +898,3 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 
    return GL_TRUE;
 }
-
-static void
-intelContendedLock(struct intel_context *intel, GLuint flags)
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   __DRIscreenPrivate *sPriv = intel->driScreen;
-   volatile drm_i915_sarea_t *sarea = intel->sarea;
-   int me = intel->hHWContext;
-
-   drmGetLock(intel->driFd, intel->hHWContext, flags);
-
-   if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - got contended lock\n", __progname);
-
-   /* If the window moved, may need to set a new cliprect now.
-    *
-    * NOTE: This releases and regains the hw lock, so all state
-    * checking must be done *after* this call:
-    */
-   if (dPriv)
-       DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
-
-   if (sarea && sarea->ctxOwner != me) {
-      if (INTEL_DEBUG & DEBUG_BUFMGR) {
-	 fprintf(stderr, "Lost Context: sarea->ctxOwner %x me %x\n",
-		 sarea->ctxOwner, me);
-      }
-      sarea->ctxOwner = me;
-   }
-
-   /* If the last consumer of the texture memory wasn't us, notify the fake
-    * bufmgr and record the new owner.  We should have the memory shared
-    * between contexts of a single fake bufmgr, but this will at least make
-    * things correct for now.
-    */
-   if (!intel->ttm && sarea->texAge != intel->hHWContext) {
-      sarea->texAge = intel->hHWContext;
-      intel_bufmgr_fake_contended_lock_take(intel->bufmgr);
-      if (INTEL_DEBUG & DEBUG_BATCH)
-	 intel_decode_context_reset();
-      if (INTEL_DEBUG & DEBUG_BUFMGR)
-	 fprintf(stderr, "Lost Textures: sarea->texAge %x hw context %x\n",
-		 sarea->ctxOwner, intel->hHWContext);
-   }
-
-   /* Drawable changed?
-    */
-   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
-       intelWindowMoved(intel);
-       intel->lastStamp = dPriv->lastStamp;
-   }
-}
-
-
-_glthread_DECLARE_STATIC_MUTEX(lockMutex);
-
-/* Lock the hardware and validate our state.  
- */
-void LOCK_HARDWARE( struct intel_context *intel )
-{
-    __DRIdrawable *dPriv = intel->driDrawable;
-    __DRIscreen *sPriv = intel->driScreen;
-    char __ret = 0;
-    struct intel_framebuffer *intel_fb = NULL;
-    struct intel_renderbuffer *intel_rb = NULL;
-
-    intel->locked++;
-    if (intel->locked >= 2)
-       return;
-
-    if (!sPriv->dri2.enabled)
-       _glthread_LOCK_MUTEX(lockMutex);
-
-    if (intel->driDrawable) {
-       intel_fb = intel->driDrawable->driverPrivate;
-
-       if (intel_fb)
-	  intel_rb =
-	     intel_get_renderbuffer(&intel_fb->Base,
-				    intel_fb->Base._ColorDrawBufferIndexes[0]);
-    }
-
-    if (intel_rb && dPriv->vblFlags &&
-	!(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ) &&
-	(intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
-	drmVBlank vbl;
-
-	vbl.request.type = DRM_VBLANK_ABSOLUTE;
-
-	if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
-	    vbl.request.type |= DRM_VBLANK_SECONDARY;
-	}
-
-	vbl.request.sequence = intel_rb->vbl_pending;
-	drmWaitVBlank(intel->driFd, &vbl);
-	intel_fb->vbl_waited = vbl.reply.sequence;
-    }
-
-    if (!sPriv->dri2.enabled) {
-	DRM_CAS(intel->driHwLock, intel->hHWContext,
-		(DRM_LOCK_HELD|intel->hHWContext), __ret);
-
-	if (__ret)
-	    intelContendedLock( intel, 0 );
-    }
-
-
-    if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - locked\n", __progname);
-}
-
-
-/* Unlock the hardware using the global current context 
- */
-void UNLOCK_HARDWARE( struct intel_context *intel )
-{
-    __DRIscreen *sPriv = intel->driScreen;
-
-   intel->locked--;
-   if (intel->locked > 0)
-      return;
-
-   assert(intel->locked == 0);
-
-   if (!sPriv->dri2.enabled) {
-      DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
-      _glthread_UNLOCK_MUTEX(lockMutex);
-   }
-
-   if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - unlocked\n", __progname);
-
-   /**
-    * Nothing should be left in batch outside of LOCK/UNLOCK which references
-    * cliprects.
-    */
-   if (intel->batch->cliprect_mode == REFERENCES_CLIPRECTS)
-      intel_batchbuffer_flush(intel->batch);
-}
-
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index b104096912..c7b7235836 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -61,6 +61,10 @@ typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
                                  intelVertex *);
 typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
 
+/**
+ * Bits for intel->Fallback field
+ */
+/*@{*/
 #define INTEL_FALLBACK_DRAW_BUFFER	 0x1
 #define INTEL_FALLBACK_READ_BUFFER	 0x2
 #define INTEL_FALLBACK_DEPTH_BUFFER      0x4
@@ -68,8 +72,10 @@ typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
 #define INTEL_FALLBACK_USER		 0x10
 #define INTEL_FALLBACK_RENDERMODE	 0x20
 #define INTEL_FALLBACK_TEXTURE   	 0x40
+#define INTEL_FALLBACK_DRIVER            0x1000  /**< first for drivers */
+/*@}*/
 
-extern void intelFallback(struct intel_context *intel, GLuint bit,
+extern void intelFallback(struct intel_context *intel, GLbitfield bit,
                           GLboolean mode);
 #define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
 
@@ -111,8 +117,6 @@ struct intel_context
                                struct intel_region * depth_region,
 			       GLuint num_regions);
 
-      GLuint (*flush_cmd) (void);
-
       void (*reduced_primitive_state) (struct intel_context * intel,
                                        GLenum rprim);
 
@@ -131,14 +135,6 @@ struct intel_context
                                 struct intel_region * draw_region,
                                 struct intel_region * depth_region);
 
-      void (*meta_draw_quad)(struct intel_context *intel,
-			     GLfloat x0, GLfloat x1,
-			     GLfloat y0, GLfloat y1,
-			     GLfloat z,
-			     GLuint color, /* ARGB32 */
-			     GLfloat s0, GLfloat s1,
-			     GLfloat t0, GLfloat t1);
-
       void (*meta_color_mask) (struct intel_context * intel, GLboolean);
 
       void (*meta_stencil_replace) (struct intel_context * intel,
@@ -170,27 +166,27 @@ struct intel_context
 
    struct dri_metaops meta;
 
-   GLint refcount;
-   GLuint Fallback;
+   GLbitfield Fallback;  /**< mask of INTEL_FALLBACK_x bits */
    GLuint NewGLState;
 
    dri_bufmgr *bufmgr;
    unsigned int maxBatchSize;
 
-   struct intel_region *front_region;
-   struct intel_region *back_region;
-   struct intel_region *depth_region;
-
    /**
-    * This value indicates that the kernel memory manager is being used
-    * instead of the fake client-side memory manager.
+    * Generation number of the hardware: 2 is 8xx, 3 is 9xx pre-965, 4 is 965.
     */
-   GLboolean ttm;
+   int gen;
+   GLboolean needs_ff_sync;
+   GLboolean is_ironlake;
+   GLboolean is_g4x;
+   GLboolean is_945;
+   GLboolean has_luminance_srgb;
+
+   int urb_size;
 
    struct intel_batchbuffer *batch;
    drm_intel_bo *first_post_swapbuffers_batch;
    GLboolean no_batch_wrap;
-   unsigned batch_id;
 
    struct
    {
@@ -210,10 +206,6 @@ struct intel_context
    char *prevLockFile;
    int prevLockLine;
 
-   GLuint ClearColor565;
-   GLuint ClearColor8888;
-
-
    /* Offsets of fields within the current vertex:
     */
    GLuint coloroffset;
@@ -230,6 +222,7 @@ struct intel_context
    GLboolean hw_stipple;
    GLboolean depth_buffer_is_float;
    GLboolean no_rast;
+   GLboolean no_hw;
    GLboolean always_flush_batch;
    GLboolean always_flush_cache;
 
@@ -255,19 +248,6 @@ struct intel_context
    intel_tri_func draw_tri;
 
    /**
-    * Set to true if a single constant cliprect should be used in the
-    * batchbuffer.  Otherwise, cliprects must be calculated at batchbuffer
-    * flush time while the lock is held.
-    */
-   GLboolean constant_cliprect;
-
-   /**
-    * In !constant_cliprect mode, set to true if the front cliprects should be
-    * used instead of back.
-    */
-   GLboolean front_cliprects;
-
-   /**
     * Set if rendering has occured to the drawable's front buffer.
     *
     * This is used in the DRI2 case to detect that glFlush should also copy
@@ -295,48 +275,23 @@ struct intel_context
    GLboolean use_early_z;
    drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
 
-   int perf_boxes;
-
-   GLuint do_usleeps;
-   int do_irqs;
-   GLuint irqsEmitted;
-
-   GLboolean scissor;
    drm_clip_rect_t draw_rect;
    drm_clip_rect_t scissor_rect;
 
-   drm_context_t hHWContext;
-   drmLock *driHwLock;
    int driFd;
 
-   __DRIcontextPrivate *driContext;
-   __DRIdrawablePrivate *driDrawable;
-   __DRIdrawablePrivate *driReadDrawable;
-   __DRIscreenPrivate *driScreen;
+   __DRIcontext *driContext;
+   __DRIdrawable *driDrawable;
+   __DRIdrawable *driReadDrawable;
+   __DRIscreen *driScreen;
    intelScreenPrivate *intelScreen;
-   volatile drm_i915_sarea_t *sarea;
-
-   GLuint lastStamp;
-
-   GLboolean no_hw;
 
    /**
     * Configuration cache
     */
    driOptionCache optionCache;
-
-   int64_t swap_ust;
-   int64_t swap_missed_ust;
-
-   GLuint swap_count;
-   GLuint swap_missed_count;
 };
 
-/* These are functions now:
- */
-void LOCK_HARDWARE( struct intel_context *intel );
-void UNLOCK_HARDWARE( struct intel_context *intel );
-
 extern char *__progname;
 
 
@@ -347,6 +302,19 @@ extern char *__progname;
 #define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
 #define IS_POWER_OF_TWO(val) (((val) & (val - 1)) == 0)
 
+static INLINE uint32_t
+U_FIXED(float value, uint32_t frac_bits)
+{
+   value *= (1 << frac_bits);
+   return value < 0 ? 0 : value;
+}
+
+static INLINE uint32_t
+S_FIXED(float value, uint32_t frac_bits)
+{
+   return value * (1 << frac_bits);
+}
+
 #define INTEL_FIREVERTICES(intel)		\
 do {						\
    if ((intel)->prim.flush)			\
@@ -354,29 +322,6 @@ do {						\
 } while (0)
 
 /* ================================================================
- * Color packing:
- */
-
-#define INTEL_PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define INTEL_PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define INTEL_PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define INTEL_PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
-(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
- (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
-  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
-   0)))
-
-/* ================================================================
  * From linux kernel i386 header files, copes with odd sizes better
  * than COPY_DWORDS would:
  * XXX Put this in src/mesa/main/imports.h ???
@@ -461,12 +406,10 @@ extern int INTEL_DEBUG;
 
 extern GLboolean intelInitContext(struct intel_context *intel,
                                   const __GLcontextModes * mesaVis,
-                                  __DRIcontextPrivate * driContextPriv,
+                                  __DRIcontext * driContextPriv,
                                   void *sharedContextPrivate,
                                   struct dd_function_table *functions);
 
-extern void intelGetLock(struct intel_context *intel, GLuint flags);
-
 extern void intelFinish(GLcontext * ctx);
 extern void intelFlush(GLcontext * ctx);
 
@@ -568,4 +511,25 @@ is_power_of_two(uint32_t value)
    return (value & (value - 1)) == 0;
 }
 
+static INLINE void
+intel_bo_map_gtt_preferred(struct intel_context *intel,
+			   drm_intel_bo *bo,
+			   GLboolean write)
+{
+   if (intel->intelScreen->kernel_exec_fencing)
+      drm_intel_gem_bo_map_gtt(bo);
+   else
+      drm_intel_bo_map(bo, write);
+}
+
+static INLINE void
+intel_bo_unmap_gtt_preferred(struct intel_context *intel,
+			     drm_intel_bo *bo)
+{
+   if (intel->intelScreen->kernel_exec_fencing)
+      drm_intel_gem_bo_unmap_gtt(bo);
+   else
+      drm_intel_bo_unmap(bo);
+}
+
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_depthtmp.h b/src/mesa/drivers/dri/intel/intel_depthtmp.h
index 16d7708453..a9c75d44cf 100644
--- a/src/mesa/drivers/dri/intel/intel_depthtmp.h
+++ b/src/mesa/drivers/dri/intel/intel_depthtmp.h
@@ -31,6 +31,16 @@
  */
 
 #define VALUE_TYPE INTEL_VALUE_TYPE
+#define WRITE_DEPTH(_x, _y, d) \
+   (*(INTEL_VALUE_TYPE *)(irb->region->buffer->virtual +	\
+			  NO_TILE(_x, _y)) = d)
+#define READ_DEPTH(d, _x, _y) \
+   d = *(INTEL_VALUE_TYPE *)(irb->region->buffer->virtual +	\
+			     NO_TILE(_x, _y))
+#define TAG(x) INTEL_TAG(intel_gttmap_##x)
+#include "depthtmp.h"
+
+#define VALUE_TYPE INTEL_VALUE_TYPE
 #define WRITE_DEPTH(_x, _y, d) INTEL_WRITE_DEPTH(NO_TILE(_x, _y), d)
 #define READ_DEPTH(d, _x, _y) d = INTEL_READ_DEPTH(NO_TILE(_x, _y))
 #define TAG(x) INTEL_TAG(intel##x)
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c
index 2eb08a8f05..5ac5ce10af 100644
--- a/src/mesa/drivers/dri/intel/intel_extensions.c
+++ b/src/mesa/drivers/dri/intel/intel_extensions.c
@@ -28,6 +28,7 @@
 #include "intel_chipset.h"
 #include "intel_context.h"
 #include "intel_extensions.h"
+#include "utils.h"
 
 
 #define need_GL_ARB_copy_buffer
@@ -63,7 +64,7 @@
 #define need_GL_VERSION_2_0
 #define need_GL_VERSION_2_1
 
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 
 /**
@@ -78,6 +79,7 @@ static const struct dri_extension card_extensions[] = {
    { "GL_ARB_half_float_pixel",           NULL },
    { "GL_ARB_map_buffer_range",           GL_ARB_map_buffer_range_functions },
    { "GL_ARB_multitexture",               NULL },
+   { "GL_ARB_pixel_buffer_object",      NULL },
    { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
    { "GL_ARB_point_sprite",               NULL },
    { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
@@ -103,6 +105,8 @@ static const struct dri_extension card_extensions[] = {
    { "GL_EXT_blend_logic_op",             NULL },
    { "GL_EXT_blend_subtract",             NULL },
    { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
+   { "GL_EXT_framebuffer_blit",         GL_EXT_framebuffer_blit_functions },
+   { "GL_EXT_framebuffer_object",       GL_EXT_framebuffer_object_functions },
    { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
    { "GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions },
    { "GL_EXT_packed_depth_stencil",       NULL },
@@ -120,7 +124,6 @@ static const struct dri_extension card_extensions[] = {
    { "GL_MESA_pack_invert",               NULL },
    { "GL_MESA_ycbcr_texture",             NULL },
    { "GL_NV_blend_square",                NULL },
-   { "GL_NV_point_sprite",                GL_NV_point_sprite_functions },
    { "GL_NV_vertex_program",              GL_NV_vertex_program_functions },
    { "GL_NV_vertex_program1_1",           NULL },
    { "GL_SGIS_generate_mipmap",           NULL },
@@ -175,14 +178,6 @@ static const struct dri_extension arb_oq_extensions[] = {
    { NULL, NULL }
 };
 
-
-static const struct dri_extension ttm_extensions[] = {
-   { "GL_ARB_pixel_buffer_object",      NULL },
-   { "GL_EXT_framebuffer_blit",         GL_EXT_framebuffer_blit_functions },
-   { "GL_EXT_framebuffer_object",       GL_EXT_framebuffer_object_functions },
-   { NULL, NULL }
-};
-
 static const struct dri_extension fragment_shader_extensions[] = {
    { "GL_ARB_fragment_shader",            NULL },
    { NULL, NULL }
@@ -193,31 +188,24 @@ static const struct dri_extension fragment_shader_extensions[] = {
  * extensions for a context.
  */
 void
-intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
+intelInitExtensions(GLcontext *ctx)
 {
-   struct intel_context *intel = ctx?intel_context(ctx):NULL;
+   struct intel_context *intel = intel_context(ctx);
 
    /* Disable imaging extension until convolution is working in teximage paths.
     */
-   enable_imaging = GL_FALSE;
-
-   driInitExtensions(ctx, card_extensions, enable_imaging);
-
-   if (intel == NULL || intel->ttm)
-      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
+   driInitExtensions(ctx, card_extensions, GL_FALSE);
 
-   if (intel == NULL || IS_965(intel->intelScreen->deviceID))
+   if (intel->gen >= 4)
       driInitExtensions(ctx, brw_extensions, GL_FALSE);
 
-   if (intel == NULL || IS_915(intel->intelScreen->deviceID)
-       || IS_945(intel->intelScreen->deviceID)) {
+   if (intel->gen == 3) {
       driInitExtensions(ctx, i915_extensions, GL_FALSE);
 
-      if (intel == NULL || driQueryOptionb(&intel->optionCache, "fragment_shader"))
+      if (driQueryOptionb(&intel->optionCache, "fragment_shader"))
 	 driInitExtensions(ctx, fragment_shader_extensions, GL_FALSE);
 
-      if (intel == NULL || driQueryOptionb(&intel->optionCache,
-					   "stub_occlusion_query"))
+      if (driQueryOptionb(&intel->optionCache, "stub_occlusion_query"))
 	 driInitExtensions(ctx, arb_oq_extensions, GL_FALSE);
    }
 }
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.h b/src/mesa/drivers/dri/intel/intel_extensions.h
index 97147ecdb0..1d1c97a4a9 100644
--- a/src/mesa/drivers/dri/intel/intel_extensions.h
+++ b/src/mesa/drivers/dri/intel/intel_extensions.h
@@ -30,7 +30,7 @@
 
 
 extern void
-intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging);
+intelInitExtensions(GLcontext *ctx);
 
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c
index 8dfb24290d..d58ffd95fa 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.c
+++ b/src/mesa/drivers/dri/intel/intel_fbo.c
@@ -33,11 +33,11 @@
 #include "main/framebuffer.h"
 #include "main/renderbuffer.h"
 #include "main/context.h"
-#include "main/texformat.h"
 #include "main/texrender.h"
 #include "drivers/common/meta.h"
 
 #include "intel_context.h"
+#include "intel_batchbuffer.h"
 #include "intel_buffers.h"
 #include "intel_fbo.h"
 #include "intel_mipmap_tree.h"
@@ -106,8 +106,8 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
 {
    struct intel_context *intel = intel_context(ctx);
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);
-   GLboolean softwareBuffer = GL_FALSE;
    int cpp;
+   GLuint pitch;
 
    ASSERT(rb->Name != 0);
 
@@ -115,27 +115,16 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_R3_G3_B2:
    case GL_RGB4:
    case GL_RGB5:
-      rb->_ActualFormat = GL_RGB5;
+      rb->Format = MESA_FORMAT_RGB565;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 5;
-      rb->GreenBits = 6;
-      rb->BlueBits = 5;
-      irb->texformat = &_mesa_texformat_rgb565;
-      cpp = 2;
       break;
    case GL_RGB:
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      rb->_ActualFormat = GL_RGB8;
+      rb->Format = MESA_FORMAT_XRGB8888;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 8;
-      rb->GreenBits = 8;
-      rb->BlueBits = 8;
-      rb->AlphaBits = 0;
-      irb->texformat = &_mesa_texformat_argb8888; /* XXX: Need xrgb8888 */
-      cpp = 4;
       break;
    case GL_RGBA:
    case GL_RGBA2:
@@ -145,14 +134,8 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      rb->_ActualFormat = GL_RGBA8;
+      rb->Format = MESA_FORMAT_ARGB8888;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 8;
-      rb->GreenBits = 8;
-      rb->BlueBits = 8;
-      rb->AlphaBits = 8;
-      irb->texformat = &_mesa_texformat_argb8888;
-      cpp = 4;
       break;
    case GL_STENCIL_INDEX:
    case GL_STENCIL_INDEX1_EXT:
@@ -160,36 +143,23 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_STENCIL_INDEX8_EXT:
    case GL_STENCIL_INDEX16_EXT:
       /* alloc a depth+stencil buffer */
-      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->Format = MESA_FORMAT_S8_Z24;
       rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->StencilBits = 8;
-      cpp = 4;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
    case GL_DEPTH_COMPONENT16:
-      rb->_ActualFormat = GL_DEPTH_COMPONENT16;
+      rb->Format = MESA_FORMAT_Z16;
       rb->DataType = GL_UNSIGNED_SHORT;
-      rb->DepthBits = 16;
-      cpp = 2;
-      irb->texformat = &_mesa_texformat_z16;
       break;
    case GL_DEPTH_COMPONENT:
    case GL_DEPTH_COMPONENT24:
    case GL_DEPTH_COMPONENT32:
-      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->Format = MESA_FORMAT_S8_Z24;
       rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->DepthBits = 24;
-      cpp = 4;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
    case GL_DEPTH_STENCIL_EXT:
    case GL_DEPTH24_STENCIL8_EXT:
-      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->Format = MESA_FORMAT_S8_Z24;
       rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->DepthBits = 24;
-      rb->StencilBits = 8;
-      cpp = 4;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
    default:
       _mesa_problem(ctx,
@@ -197,6 +167,9 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
       return GL_FALSE;
    }
 
+   rb->_BaseFormat = _mesa_base_fbo_format(ctx, internalFormat);
+   cpp = _mesa_get_format_bytes(rb->Format);
+
    intelFlush(ctx);
 
    /* free old region */
@@ -205,32 +178,25 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    }
 
    /* allocate new memory region/renderbuffer */
-   if (softwareBuffer) {
-      return _mesa_soft_renderbuffer_storage(ctx, rb, internalFormat,
-                                             width, height);
-   }
-   else {
-      /* Choose a pitch to match hardware requirements:
-       */
-      GLuint pitch = ((cpp * width + 63) & ~63) / cpp;
 
-      /* alloc hardware renderbuffer */
-      DBG("Allocating %d x %d Intel RBO (pitch %d)\n", width,
-	  height, pitch);
+   /* Choose a pitch to match hardware requirements:
+    */
+   pitch = ((cpp * width + 63) & ~63) / cpp;
+
+   /* alloc hardware renderbuffer */
+   DBG("Allocating %d x %d Intel RBO (pitch %d)\n", width, height, pitch);
 
-      irb->region = intel_region_alloc(intel, I915_TILING_NONE,
-				       cpp, width, height, pitch,
-				       GL_TRUE);
-      if (!irb->region)
-         return GL_FALSE;       /* out of memory? */
+   irb->region = intel_region_alloc(intel, I915_TILING_NONE, cpp,
+				    width, height, pitch, GL_TRUE);
+   if (!irb->region)
+      return GL_FALSE;       /* out of memory? */
 
-      ASSERT(irb->region->buffer);
+   ASSERT(irb->region->buffer);
 
-      rb->Width = width;
-      rb->Height = height;
+   rb->Width = width;
+   rb->Height = height;
 
-      return GL_TRUE;
-   }
+   return GL_TRUE;
 }
 
 
@@ -246,7 +212,7 @@ intel_alloc_window_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    ASSERT(rb->Name == 0);
    rb->Width = width;
    rb->Height = height;
-   rb->_ActualFormat = internalFormat;
+   rb->InternalFormat = internalFormat;
 
    return GL_TRUE;
 }
@@ -256,7 +222,6 @@ static void
 intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
 		     GLuint width, GLuint height)
 {
-   struct intel_framebuffer *intel_fb = (struct intel_framebuffer*)fb;
    int i;
 
    _mesa_resize_framebuffer(ctx, fb, width, height);
@@ -267,9 +232,10 @@ intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
       return;
    }
 
+
    /* Make sure all window system renderbuffers are up to date */
-   for (i = 0; i < 2; i++) {
-      struct gl_renderbuffer *rb = &intel_fb->color_rb[i]->Base;
+   for (i = BUFFER_FRONT_LEFT; i <= BUFFER_BACK_RIGHT; i++) {
+      struct gl_renderbuffer *rb = fb->Attachment[i].Renderbuffer;
 
       /* only resize if size is changing */
       if (rb && (rb->Width != width || rb->Height != height)) {
@@ -307,12 +273,11 @@ intel_renderbuffer_set_region(struct intel_renderbuffer *rb,
  * not a user-created renderbuffer.
  */
 struct intel_renderbuffer *
-intel_create_renderbuffer(GLenum intFormat)
+intel_create_renderbuffer(gl_format format)
 {
    GET_CURRENT_CONTEXT(ctx);
 
    struct intel_renderbuffer *irb;
-   const GLuint name = 0;
 
    irb = CALLOC_STRUCT(intel_renderbuffer);
    if (!irb) {
@@ -320,75 +285,43 @@ intel_create_renderbuffer(GLenum intFormat)
       return NULL;
    }
 
-   _mesa_init_renderbuffer(&irb->Base, name);
+   _mesa_init_renderbuffer(&irb->Base, 0);
    irb->Base.ClassID = INTEL_RB_CLASS;
 
-   switch (intFormat) {
-   case GL_RGB5:
-      irb->Base._ActualFormat = GL_RGB5;
-      irb->Base._BaseFormat = GL_RGBA;
-      irb->Base.RedBits = 5;
-      irb->Base.GreenBits = 6;
-      irb->Base.BlueBits = 5;
+   switch (format) {
+   case MESA_FORMAT_RGB565:
+      irb->Base._BaseFormat = GL_RGB;
       irb->Base.DataType = GL_UNSIGNED_BYTE;
-      irb->texformat = &_mesa_texformat_rgb565;
       break;
-   case GL_RGB8:
-      irb->Base._ActualFormat = GL_RGB8;
+   case MESA_FORMAT_XRGB8888:
       irb->Base._BaseFormat = GL_RGB;
-      irb->Base.RedBits = 8;
-      irb->Base.GreenBits = 8;
-      irb->Base.BlueBits = 8;
-      irb->Base.AlphaBits = 0;
       irb->Base.DataType = GL_UNSIGNED_BYTE;
-      irb->texformat = &_mesa_texformat_argb8888; /* XXX: Need xrgb8888 */
       break;
-   case GL_RGBA8:
-      irb->Base._ActualFormat = GL_RGBA8;
+   case MESA_FORMAT_ARGB8888:
       irb->Base._BaseFormat = GL_RGBA;
-      irb->Base.RedBits = 8;
-      irb->Base.GreenBits = 8;
-      irb->Base.BlueBits = 8;
-      irb->Base.AlphaBits = 8;
-      irb->Base.DataType = GL_UNSIGNED_BYTE;
-      irb->texformat = &_mesa_texformat_argb8888;
-      break;
-   case GL_STENCIL_INDEX8_EXT:
-      irb->Base._ActualFormat = GL_STENCIL_INDEX8_EXT;
-      irb->Base._BaseFormat = GL_STENCIL_INDEX;
-      irb->Base.StencilBits = 8;
       irb->Base.DataType = GL_UNSIGNED_BYTE;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
-   case GL_DEPTH_COMPONENT16:
-      irb->Base._ActualFormat = GL_DEPTH_COMPONENT16;
+   case MESA_FORMAT_Z16:
       irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
-      irb->Base.DepthBits = 16;
       irb->Base.DataType = GL_UNSIGNED_SHORT;
-      irb->texformat = &_mesa_texformat_z16;
       break;
-   case GL_DEPTH_COMPONENT24:
-      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+   case MESA_FORMAT_X8_Z24:
       irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
-      irb->Base.DepthBits = 24;
       irb->Base.DataType = GL_UNSIGNED_INT;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
-   case GL_DEPTH24_STENCIL8_EXT:
-      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
-      irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT;
-      irb->Base.DepthBits = 24;
-      irb->Base.StencilBits = 8;
+   case MESA_FORMAT_S8_Z24:
+      irb->Base._BaseFormat = GL_DEPTH_STENCIL;
       irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
-      irb->texformat = &_mesa_texformat_s8_z24;
       break;
    default:
       _mesa_problem(NULL,
                     "Unexpected intFormat in intel_create_renderbuffer");
+      _mesa_free(irb);
       return NULL;
    }
 
-   irb->Base.InternalFormat = intFormat;
+   irb->Base.Format = format;
+   irb->Base.InternalFormat = irb->Base._BaseFormat;
 
    /* intel-specific methods */
    irb->Base.Delete = intel_delete_renderbuffer;
@@ -465,59 +398,46 @@ static GLboolean
 intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb, 
 		     struct gl_texture_image *texImage)
 {
-   irb->texformat = texImage->TexFormat;
-
-   if (texImage->TexFormat == &_mesa_texformat_argb8888) {
-      irb->Base._ActualFormat = GL_RGBA8;
-      irb->Base._BaseFormat = GL_RGBA;
+   if (texImage->TexFormat == MESA_FORMAT_ARGB8888) {
       irb->Base.DataType = GL_UNSIGNED_BYTE;
       DBG("Render to RGBA8 texture OK\n");
    }
-   else if (texImage->TexFormat == &_mesa_texformat_rgb565) {
-      irb->Base._ActualFormat = GL_RGB5;
-      irb->Base._BaseFormat = GL_RGB;
+   else if (texImage->TexFormat == MESA_FORMAT_XRGB8888) {
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
+      DBG("Render to XGBA8 texture OK\n");
+   }
+   else if (texImage->TexFormat == MESA_FORMAT_RGB565) {
       irb->Base.DataType = GL_UNSIGNED_BYTE;
       DBG("Render to RGB5 texture OK\n");
    }
-   else if (texImage->TexFormat == &_mesa_texformat_argb1555) {
-      irb->Base._ActualFormat = GL_RGB5_A1;
-      irb->Base._BaseFormat = GL_RGBA;
+   else if (texImage->TexFormat == MESA_FORMAT_ARGB1555) {
       irb->Base.DataType = GL_UNSIGNED_BYTE;
       DBG("Render to ARGB1555 texture OK\n");
    }
-   else if (texImage->TexFormat == &_mesa_texformat_argb4444) {
-      irb->Base._ActualFormat = GL_RGBA4;
-      irb->Base._BaseFormat = GL_RGBA;
+   else if (texImage->TexFormat == MESA_FORMAT_ARGB4444) {
       irb->Base.DataType = GL_UNSIGNED_BYTE;
       DBG("Render to ARGB4444 texture OK\n");
    }
-   else if (texImage->TexFormat == &_mesa_texformat_z16) {
-      irb->Base._ActualFormat = GL_DEPTH_COMPONENT16;
-      irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
+   else if (texImage->TexFormat == MESA_FORMAT_Z16) {
       irb->Base.DataType = GL_UNSIGNED_SHORT;
       DBG("Render to DEPTH16 texture OK\n");
    }
-   else if (texImage->TexFormat == &_mesa_texformat_s8_z24) {
-      irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
-      irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT;
+   else if (texImage->TexFormat == MESA_FORMAT_S8_Z24) {
       irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
       DBG("Render to DEPTH_STENCIL texture OK\n");
    }
    else {
-      DBG("Render to texture BAD FORMAT %d\n",
-	  texImage->TexFormat->MesaFormat);
+      DBG("Render to texture BAD FORMAT %s\n",
+	  _mesa_get_format_name(texImage->TexFormat));
       return GL_FALSE;
    }
 
-   irb->Base.InternalFormat = irb->Base._ActualFormat;
+   irb->Base.Format = texImage->TexFormat;
+
+   irb->Base.InternalFormat = texImage->InternalFormat;
+   irb->Base._BaseFormat = _mesa_base_fbo_format(ctx, irb->Base.InternalFormat);
    irb->Base.Width = texImage->Width;
    irb->Base.Height = texImage->Height;
-   irb->Base.RedBits = texImage->TexFormat->RedBits;
-   irb->Base.GreenBits = texImage->TexFormat->GreenBits;
-   irb->Base.BlueBits = texImage->TexFormat->BlueBits;
-   irb->Base.AlphaBits = texImage->TexFormat->AlphaBits;
-   irb->Base.DepthBits = texImage->TexFormat->DepthBits;
-   irb->Base.StencilBits = texImage->TexFormat->StencilBits;
 
    irb->Base.Delete = intel_delete_renderbuffer;
    irb->Base.AllocStorage = intel_nop_alloc_storage;
@@ -571,7 +491,7 @@ intel_render_texture(GLcontext * ctx,
       = att->Texture->Image[att->CubeMapFace][att->TextureLevel];
    struct intel_renderbuffer *irb = intel_renderbuffer(att->Renderbuffer);
    struct intel_texture_image *intel_image;
-   GLuint imageOffset;
+   GLuint dst_x, dst_y;
 
    (void) fb;
 
@@ -618,18 +538,17 @@ 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];
-   }
-
-   /* store that offset in the region */
-   intel_image->mt->region->draw_offset = imageOffset;
+   intel_miptree_get_image_offset(intel_image->mt,
+				  att->TextureLevel,
+				  att->CubeMapFace,
+				  att->Zoffset,
+				  &dst_x, &dst_y);
+
+   intel_image->mt->region->draw_offset = (dst_y * intel_image->mt->pitch +
+					   dst_x) * intel_image->mt->cpp;
+   intel_image->mt->region->draw_x = dst_x;
+   intel_image->mt->region->draw_y = dst_y;
+   intel_image->used_as_render_target = GL_TRUE;
 
    /* update drawing region, etc */
    intel_draw_buffer(ctx, fb);
@@ -643,19 +562,23 @@ static void
 intel_finish_render_texture(GLcontext * ctx,
                             struct gl_renderbuffer_attachment *att)
 {
-   /* no-op
-    * Previously we released the renderbuffer's intel_region but
-    * that's not necessary and actually caused problems when trying
-    * to do a glRead/CopyPixels from the renderbuffer later.
-    * The region will be released later if the texture is replaced
-    * or the renderbuffer deleted.
-    *
-    * The intention of this driver hook is more of a "done rendering
-    * to texture, please re-twiddle/etc if necessary".
+   struct intel_context *intel = intel_context(ctx);
+   struct gl_texture_object *tex_obj = att->Texture;
+   struct gl_texture_image *image =
+      tex_obj->Image[att->CubeMapFace][att->TextureLevel];
+   struct intel_texture_image *intel_image = intel_texture_image(image);
+
+   /* Flag that this image may now be validated into the object's miptree. */
+   intel_image->used_as_render_target = GL_FALSE;
+
+   /* Since we've (probably) rendered to the texture and will (likely) use
+    * it in the texture domain later on in this batchbuffer, flush the
+    * batch.  Once again, we wish for a domain tracker in libdrm to cover
+    * usage inside of a batchbuffer like GEM does in the kernel.
     */
+   intel_batchbuffer_emit_mi_flush(intel->batch);
 }
 
-
 /**
  * Do additional "completeness" testing of a framebuffer object.
  */
@@ -668,11 +591,21 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
       intel_get_renderbuffer(fb, BUFFER_STENCIL);
    int i;
 
-   if (stencilRb && stencilRb != depthRb) {
-      /* we only support combined depth/stencil buffers, not separate
-       * stencil buffers.
-       */
-      fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+   if (depthRb && stencilRb && stencilRb != depthRb) {
+      if (ctx->DrawBuffer->Attachment[BUFFER_DEPTH].Type == GL_TEXTURE &&
+	  ctx->DrawBuffer->Attachment[BUFFER_STENCIL].Type == GL_TEXTURE &&
+	  (ctx->DrawBuffer->Attachment[BUFFER_DEPTH].Texture->Name ==
+	   ctx->DrawBuffer->Attachment[BUFFER_STENCIL].Texture->Name)) {
+	 /* OK */
+      } else {
+	 /* we only support combined depth/stencil buffers, not separate
+	  * stencil buffers.
+	  */
+	 DBG("Only supports combined depth/stencil (found %s, %s)\n",
+	     depthRb ? _mesa_get_format_name(depthRb->Base.Format): "NULL",
+	     stencilRb ? _mesa_get_format_name(stencilRb->Base.Format): "NULL");
+	 fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+      }
    }
 
    for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
@@ -683,12 +616,14 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
 	 continue;
 
       if (irb == NULL) {
+	 DBG("software rendering renderbuffer\n");
 	 fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
 	 continue;
       }
 
-      switch (irb->texformat->MesaFormat) {
+      switch (irb->Base.Format) {
       case MESA_FORMAT_ARGB8888:
+      case MESA_FORMAT_XRGB8888:
       case MESA_FORMAT_RGB565:
       case MESA_FORMAT_ARGB1555:
       case MESA_FORMAT_ARGB4444:
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.h b/src/mesa/drivers/dri/intel/intel_fbo.h
index f0665af482..586dbbbb25 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.h
+++ b/src/mesa/drivers/dri/intel/intel_fbo.h
@@ -28,32 +28,12 @@
 #ifndef INTEL_FBO_H
 #define INTEL_FBO_H
 
+#include "main/formats.h"
 #include "intel_screen.h"
 
 struct intel_context;
 
 /**
- * Intel framebuffer, derived from gl_framebuffer.
- */
-struct intel_framebuffer
-{
-   struct gl_framebuffer Base;
-
-   struct intel_renderbuffer *color_rb[2];
-
-   /* VBI
-    */
-   GLuint vbl_waited;
-
-   int64_t swap_ust;
-   int64_t swap_missed_ust;
-
-   GLuint swap_count;
-   GLuint swap_missed_count;
-};
-
-
-/**
  * Intel renderbuffer, derived from gl_renderbuffer.
  */
 struct intel_renderbuffer
@@ -61,10 +41,6 @@ struct intel_renderbuffer
    struct gl_renderbuffer Base;
    struct intel_region *region;
 
-   const struct gl_texture_format *texformat;
-
-   GLuint vbl_pending;   /**< vblank sequence number of pending flip */
-
    uint8_t *span_cache;
    unsigned long span_cache_offset;
 };
@@ -114,7 +90,7 @@ intel_renderbuffer_set_region(struct intel_renderbuffer *irb,
 
 
 extern struct intel_renderbuffer *
-intel_create_renderbuffer(GLenum intFormat);
+intel_create_renderbuffer(gl_format format);
 
 
 extern void
@@ -122,7 +98,7 @@ intel_fbo_init(struct intel_context *intel);
 
 
 extern void
-intel_flip_renderbuffers(struct intel_framebuffer *intel_fb);
+intel_flip_renderbuffers(struct gl_framebuffer *fb);
 
 
 static INLINE struct intel_region *
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
index 4f5101a312..82e4150c6a 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,
@@ -84,7 +87,7 @@ intel_miptree_create_internal(struct intel_context *intel,
    mt->pitch = 0;
 
 #ifdef I915
-   if (IS_945(intel->intelScreen->deviceID))
+   if (intel->is_945)
       ok = i945_miptree_layout(intel, mt, tiling);
    else
       ok = i915_miptree_layout(intel, mt, tiling);
@@ -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,
@@ -118,7 +122,7 @@ intel_miptree_create(struct intel_context *intel,
 
    if (intel->use_texture_tiling && compress_byte == 0 &&
        intel->intelScreen->kernel_exec_fencing) {
-      if (IS_965(intel->intelScreen->deviceID) &&
+      if (intel->gen >= 4 &&
 	  (base_format == GL_DEPTH_COMPONENT ||
 	   base_format == GL_DEPTH_STENCIL_EXT))
 	 tiling = I915_TILING_Y;
@@ -134,8 +138,10 @@ intel_miptree_create(struct intel_context *intel,
    /*
     * pitch == 0 || height == 0  indicates the null texture
     */
-   if (!mt || !mt->pitch || !mt->total_height)
+   if (!mt || !mt->pitch || !mt->total_height) {
+      free(mt);
       return NULL;
+   }
 
    mt->region = intel_region_alloc(intel,
 				   tiling,
@@ -153,6 +159,7 @@ 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 +197,8 @@ intel_miptree_create_for_region(struct intel_context *intel,
    intel_region_reference(&mt->region, region);
 
    return mt;
- }
+}
+
 
 /**
  * intel_miptree_pitch_align:
@@ -204,7 +212,6 @@ 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,
@@ -217,16 +224,12 @@ int intel_miptree_pitch_align (struct intel_context *intel,
    if (!mt->compressed) {
       int pitch_align;
 
-      if (intel->ttm) {
-	 /* XXX: Align pitch to multiple of 64 bytes for now to allow
-	  * render-to-texture to work in all cases. This should probably be
-	  * replaced at some point by some scheme to only do this when really
-	  * necessary.
-	  */
-	 pitch_align = 64;
-      } else {
-	 pitch_align = 4;
-      }
+      /* XXX: Align pitch to multiple of 64 bytes for now to allow
+       * render-to-texture to work in all cases. This should probably be
+       * replaced at some point by some scheme to only do this when really
+       * necessary.
+       */
+      pitch_align = 64;
 
       if (tiling == I915_TILING_X)
 	 pitch_align = 512;
@@ -250,6 +253,7 @@ 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 +263,7 @@ 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)
@@ -287,9 +292,10 @@ intel_miptree_release(struct intel_context *intel,
 
       intel_region_release(&((*mt)->region));
 
-      for (i = 0; i < MAX_TEXTURE_LEVELS; i++)
-         if ((*mt)->level[i].image_offset)
-            free((*mt)->level[i].image_offset);
+      for (i = 0; i < MAX_TEXTURE_LEVELS; i++) {
+	 free((*mt)->level[i].x_offset);
+	 free((*mt)->level[i].y_offset);
+      }
 
       free(*mt);
    }
@@ -297,33 +303,31 @@ 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.
  */
 GLboolean
 intel_miptree_match_image(struct intel_mipmap_tree *mt,
-                          struct gl_texture_image *image,
-                          GLuint face, GLuint level)
+                          struct gl_texture_image *image)
 {
-   /* Images with borders are never pulled into mipmap trees. 
-    */
-   if (image->Border ||
-       ((image->_BaseFormat == GL_DEPTH_COMPONENT) &&
-        ((image->TexObject->WrapS == GL_CLAMP_TO_BORDER) ||
-         (image->TexObject->WrapT == GL_CLAMP_TO_BORDER)))) 
+   GLboolean isCompressed = _mesa_is_format_compressed(image->TexFormat);
+   struct intel_texture_image *intelImage = intel_texture_image(image);
+   GLuint level = intelImage->level;
+
+   /* Images with borders are never pulled into mipmap trees. */
+   if (image->Border)
       return GL_FALSE;
 
    if (image->InternalFormat != mt->internal_format ||
-       image->IsCompressed != mt->compressed)
+       isCompressed != mt->compressed)
       return GL_FALSE;
 
-   if (!image->IsCompressed &&
+   if (!isCompressed &&
        !mt->compressed &&
-       image->TexFormat->TexelBytes != mt->cpp)
+       _mesa_get_format_bytes(image->TexFormat) != mt->cpp)
       return GL_FALSE;
 
    /* Test image dimensions against the base level image adjusted for
@@ -350,82 +354,59 @@ intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
    mt->level[level].height = h;
    mt->level[level].depth = d;
    mt->level[level].level_offset = (x + y * mt->pitch) * mt->cpp;
+   mt->level[level].level_x = x;
+   mt->level[level].level_y = y;
    mt->level[level].nr_images = nr_images;
 
    DBG("%s level %d size: %d,%d,%d offset %d,%d (0x%x)\n", __FUNCTION__,
        level, w, h, d, x, y, mt->level[level].level_offset);
 
-   /* Not sure when this would happen, but anyway: 
-    */
-   if (mt->level[level].image_offset) {
-      free(mt->level[level].image_offset);
-      mt->level[level].image_offset = NULL;
-   }
-
    assert(nr_images);
+   assert(!mt->level[level].x_offset);
 
-   mt->level[level].image_offset = malloc(nr_images * sizeof(GLuint));
-   mt->level[level].image_offset[0] = 0;
+   mt->level[level].x_offset = malloc(nr_images * sizeof(GLuint));
+   mt->level[level].x_offset[0] = mt->level[level].level_x;
+   mt->level[level].y_offset = malloc(nr_images * sizeof(GLuint));
+   mt->level[level].y_offset[0] = mt->level[level].level_y;
 }
 
 
 void
-intel_miptree_set_image_offset_ex(struct intel_mipmap_tree *mt,
-                                  GLuint level, GLuint img,
-                                  GLuint x, GLuint y, 
-                                  GLuint offset)
-{
-   if (img == 0 && level == 0)
-      assert(x == 0 && y == 0);
-
-   assert(img < mt->level[level].nr_images);
-
-   mt->level[level].image_offset[img] = (x + y * mt->pitch) * mt->cpp + offset;
-
-   DBG("%s level %d img %d pos %d,%d image_offset %x\n",
-       __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,
 			       GLuint x, GLuint y)
 {
-    intel_miptree_set_image_offset_ex(mt, level, img, x, y, 0);
-}
+   if (img == 0 && level == 0)
+      assert(x == 0 && y == 0);
 
+   assert(img < mt->level[level].nr_images);
 
-/* 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:
- */
-const GLuint *
-intel_miptree_depth_offsets(struct intel_mipmap_tree *mt, GLuint level)
-{
-   static const GLuint zero = 0;
+   mt->level[level].x_offset[img] = mt->level[level].level_x + x;
+   mt->level[level].y_offset[img] = mt->level[level].level_y + y;
 
-   if (mt->target != GL_TEXTURE_3D || mt->level[level].nr_images == 1)
-      return &zero;
-   else
-      return mt->level[level].image_offset;
+   DBG("%s level %d img %d pos %d,%d\n",
+       __FUNCTION__, level, img,
+       mt->level[level].x_offset[img], mt->level[level].y_offset[img]);
 }
 
 
-GLuint
-intel_miptree_image_offset(struct intel_mipmap_tree *mt,
-			   GLuint face, GLuint level)
+void
+intel_miptree_get_image_offset(struct intel_mipmap_tree *mt,
+			       GLuint level, GLuint face, GLuint depth,
+			       GLuint *x, GLuint *y)
 {
-   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;
+   if (mt->target == GL_TEXTURE_CUBE_MAP_ARB) {
+      *x = mt->level[level].x_offset[face];
+      *y = mt->level[level].y_offset[face];
+   } else if (mt->target == GL_TEXTURE_3D) {
+      *x = mt->level[level].x_offset[depth];
+      *y = mt->level[level].y_offset[depth];
+   } else {
+      *x = mt->level[level].x_offset[0];
+      *y = mt->level[level].y_offset[0];
+   }
 }
 
-
-
 /**
  * Map a teximage in a mipmap tree.
  * \param row_stride  returns row stride in bytes
@@ -441,6 +422,7 @@ intel_miptree_image_map(struct intel_context * intel,
                         GLuint level,
                         GLuint * row_stride, GLuint * image_offsets)
 {
+   GLuint x, y;
    DBG("%s \n", __FUNCTION__);
 
    if (row_stride)
@@ -449,19 +431,26 @@ intel_miptree_image_map(struct intel_context * intel,
    if (mt->target == GL_TEXTURE_3D) {
       int i;
 
-      for (i = 0; i < mt->level[level].depth; i++)
-	 image_offsets[i] = mt->level[level].image_offset[i] / mt->cpp;
+      for (i = 0; i < mt->level[level].depth; i++) {
+
+	 intel_miptree_get_image_offset(mt, level, face, i,
+					&x, &y);
+	 image_offsets[i] = x + y * mt->pitch;
+      }
+
+      return intel_region_map(intel, mt->region);
    } else {
       assert(mt->level[level].depth == 1);
-      assert(mt->target == GL_TEXTURE_CUBE_MAP ||
-	     mt->level[level].image_offset[0] == 0);
+      intel_miptree_get_image_offset(mt, level, face, 0,
+				     &x, &y);
       image_offsets[0] = 0;
-   }
 
-   return (intel_region_map(intel, mt->region) +
-           intel_miptree_image_offset(mt, face, level));
+      return intel_region_map(intel, mt->region) +
+	 (x + y * mt->pitch) * mt->cpp;
+   }
 }
 
+
 void
 intel_miptree_image_unmap(struct intel_context *intel,
                           struct intel_mipmap_tree *mt)
@@ -471,8 +460,8 @@ 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,
@@ -483,21 +472,21 @@ 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++) {
+      GLuint dst_x, dst_y, height;
+
+      intel_miptree_get_image_offset(dst, level, face, i, &dst_x, &dst_y);
+
       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 */
-			0, 0,                             /* dstx, dsty */
+			dst->region, 0, dst_x, dst_y,
 			src,
 			src_row_pitch,
 			0, 0,                             /* source x, y */
@@ -507,8 +496,9 @@ 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,
@@ -519,38 +509,37 @@ 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 src_x, src_y, dst_x, dst_y;
    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++) {
+      intel_miptree_get_image_offset(src, level, face, i, &src_x, &src_y);
+      intel_miptree_get_image_offset(dst, level, face, i, &dst_x, &dst_y);
       success = intel_region_copy(intel,
-				  dst->region, dst_offset + dst_depth_offset[i],
-				  0, 0,
-				  src->region, src_offset + src_depth_offset[i],
-				  0, 0, width, height, GL_COPY);
+				  dst->region, 0, dst_x, dst_y,
+				  src->region, 0, src_x, src_y, width, height,
+				  GL_COPY);
       if (!success) {
 	 GLubyte *src_ptr, *dst_ptr;
 
 	 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->cpp * (dst_x + dst_y * dst->pitch),
 			 dst->cpp,
 			 dst->pitch,
 			 0, 0, width, height,
-			 src_ptr + src_offset + src_depth_offset[i],
+			 src_ptr + src->cpp * (src_x + src_y * src->pitch),
 			 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..b19c548def 100644
--- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.h
@@ -70,6 +70,10 @@ struct intel_mipmap_level
     * always zero in that case.
     */
    GLuint level_offset;
+   /** Offset to this miptree level, used in computing x_offset. */
+   GLuint level_x;
+   /** Offset to this miptree level, used in computing y_offset. */
+   GLuint level_y;
    GLuint width;
    GLuint height;
    /** Depth of the mipmap at this level: 1 for 1D/2D/CUBE, n for 3D. */
@@ -86,7 +90,7 @@ struct intel_mipmap_level
     * compute the offsets of depth/cube images within a mipmap level,
     * so have to store them as a lookup table.
     */
-   GLuint *image_offset;
+   GLuint *x_offset, *y_offset;
 };
 
 struct intel_mipmap_tree
@@ -161,8 +165,7 @@ void intel_miptree_release(struct intel_context *intel,
 /* Check if an image fits an existing mipmap tree layout
  */
 GLboolean intel_miptree_match_image(struct intel_mipmap_tree *mt,
-                                    struct gl_texture_image *image,
-                                    GLuint face, GLuint level);
+                                    struct gl_texture_image *image);
 
 /* Return a pointer to an image within a tree.  Return image stride as
  * well.
@@ -176,19 +179,10 @@ GLubyte *intel_miptree_image_map(struct intel_context *intel,
 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);
-
+void
+intel_miptree_get_image_offset(struct intel_mipmap_tree *mt,
+			       GLuint level, GLuint face, GLuint depth,
+			       GLuint *x, GLuint *y);
 
 void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
                                   GLuint level,
@@ -196,16 +190,10 @@ void intel_miptree_set_level_info(struct intel_mipmap_tree *mt,
                                   GLuint x, GLuint y,
                                   GLuint w, GLuint h, GLuint d);
 
-void intel_miptree_set_image_offset_ex(struct intel_mipmap_tree *mt,
-                                       GLuint level,
-                                       GLuint img, GLuint x, GLuint y,
-                                       GLuint offset);
-
 void intel_miptree_set_image_offset(struct intel_mipmap_tree *mt,
                                     GLuint level,
                                     GLuint img, GLuint x, GLuint y);
 
-
 /* Upload an image into a tree
  */
 void intel_miptree_image_data(struct intel_context *intel,
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
index 993e427a99..5142f3dcd9 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -88,10 +88,10 @@ intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one)
       return GL_FALSE;
    }
 
-   if (!(ctx->Color.ColorMask[0] &&
-	 ctx->Color.ColorMask[1] &&
-	 ctx->Color.ColorMask[2] &&
-	 ctx->Color.ColorMask[3])) {
+   if (!(ctx->Color.ColorMask[0][0] &&
+	 ctx->Color.ColorMask[0][1] &&
+	 ctx->Color.ColorMask[0][2] &&
+	 ctx->Color.ColorMask[0][3])) {
       DBG("fallback due to color masking\n");
       return GL_FALSE;
    }
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index 9a0bcc07a5..85e5ad2cdd 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -32,7 +32,9 @@
 #include "main/mtypes.h"
 #include "main/macros.h"
 #include "main/bufferobj.h"
+#include "main/polygon.h"
 #include "main/pixelstore.h"
+#include "main/polygon.h"
 #include "main/state.h"
 #include "main/teximage.h"
 #include "main/texenv.h"
@@ -164,7 +166,7 @@ static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
  * Returns the low Y value of the vertical range given, flipped according to
  * whether the framebuffer is or not.
  */
-static inline int
+static INLINE int
 y_flip(struct gl_framebuffer *fb, int y, int height)
 {
    if (fb->Name != 0)
@@ -227,16 +229,13 @@ do_blit_bitmap( GLcontext *ctx,
    UNCLAMPED_FLOAT_TO_UBYTE(ubcolor[3], tmpColor[3]);
 
    if (dst->cpp == 2)
-      color = INTEL_PACKCOLOR565(ubcolor[0], ubcolor[1], ubcolor[2]);
+      color = PACK_COLOR_565(ubcolor[0], ubcolor[1], ubcolor[2]);
    else
-      color = INTEL_PACKCOLOR8888(ubcolor[0], ubcolor[1],
-				  ubcolor[2], ubcolor[3]);
+      color = PACK_COLOR_8888(ubcolor[3], ubcolor[0], ubcolor[1], ubcolor[2]);
 
    if (!intel_check_blit_fragment_ops(ctx, tmpColor[3] == 1.0F))
       return GL_FALSE;
 
-   LOCK_HARDWARE(intel);
-
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
    if (num_cliprects != 0) {
       GLuint i;
@@ -324,7 +323,6 @@ do_blit_bitmap( GLcontext *ctx,
       }
    }
 out:
-   UNLOCK_HARDWARE(intel);
 
    if (INTEL_DEBUG & DEBUG_SYNC)
       intel_batchbuffer_flush(intel->batch);
@@ -335,6 +333,8 @@ out:
                               unpack->BufferObj);
    }
 
+   intel_check_front_buffer_rendering(intel);
+
    return GL_TRUE;
 }
 
@@ -435,13 +435,14 @@ intel_texture_bitmap(GLcontext * ctx,
    }
 
    /* Save GL state before we start setting up our drawing */
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_CURRENT_BIT |
-		    GL_VIEWPORT_BIT);
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_CURRENT_BIT | GL_POLYGON_BIT |
+                    GL_TEXTURE_BIT | GL_VIEWPORT_BIT);
    _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT |
 			  GL_CLIENT_PIXEL_STORE_BIT);
    old_active_texture = ctx->Texture.CurrentUnit;
 
    _mesa_Disable(GL_POLYGON_STIPPLE);
+   _mesa_PolygonMode(GL_FRONT_AND_BACK, GL_FILL);
 
    /* Upload our bitmap data to an alpha texture */
    _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
@@ -500,9 +501,8 @@ intel_texture_bitmap(GLcontext * ctx,
    meta_restore_fragment_program(&intel->meta);
    meta_restore_vertex_program(&intel->meta);
 
-   _mesa_PopClientAttrib();
-   _mesa_Disable(GL_TEXTURE_2D); /* asserted that it was disabled at entry */
    _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
+   _mesa_PopClientAttrib();
    _mesa_PopAttrib();
 
    _mesa_DeleteTextures(1, &texname);
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
index f058b3c8e4..e002516cdd 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
@@ -35,28 +35,33 @@
 #include "intel_buffers.h"
 #include "intel_regions.h"
 #include "intel_pixel.h"
+#include "intel_fbo.h"
 
 #define FILE_DEBUG_FLAG DEBUG_PIXEL
 
 static struct intel_region *
 copypix_src_region(struct intel_context *intel, GLenum type)
 {
+   struct intel_renderbuffer *depth;
+
+   depth = (struct intel_renderbuffer *)
+      &intel->ctx.DrawBuffer->Attachment[BUFFER_DEPTH].Renderbuffer;
+
    switch (type) {
    case GL_COLOR:
       return intel_readbuf_region(intel);
    case GL_DEPTH:
-      /* Don't think this is really possible execpt at 16bpp, when we have no stencil.
-       */
-      if (intel->depth_region && intel->depth_region->cpp == 2)
-         return intel->depth_region;
+      /* Don't think this is really possible execpt at 16bpp, when we
+       * have no stencil. */
+      if (depth && depth->region->cpp == 2)
+         return depth->region;
    case GL_STENCIL:
-      /* Don't think this is really possible. 
-       */
+      /* Don't think this is really possible. */
       break;
    case GL_DEPTH_STENCIL_EXT:
       /* Does it matter whether it is stencil/depth or depth/stencil?
        */
-      return intel->depth_region;
+      return depth->region;
    default:
       break;
    }
@@ -83,10 +88,10 @@ intel_check_copypixel_blit_fragment_ops(GLcontext * ctx)
             ctx->Depth.Test ||
             ctx->Fog.Enabled ||
             ctx->Stencil._Enabled ||
-            !ctx->Color.ColorMask[0] ||
-            !ctx->Color.ColorMask[1] ||
-            !ctx->Color.ColorMask[2] ||
-            !ctx->Color.ColorMask[3] ||
+            !ctx->Color.ColorMask[0][0] ||
+            !ctx->Color.ColorMask[0][1] ||
+            !ctx->Color.ColorMask[0][2] ||
+            !ctx->Color.ColorMask[0][3] ||
             ctx->Texture._EnabledUnits ||
 	    ctx->FragmentProgram._Enabled ||
 	    ctx->Color.BlendEnabled);
@@ -134,8 +139,6 @@ do_blit_copypixels(GLcontext * ctx,
 
    intelFlush(&intel->ctx);
 
-   LOCK_HARDWARE(intel);
-
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
    if (num_cliprects != 0) {
       GLint delta_x;
@@ -214,13 +217,13 @@ do_blit_copypixels(GLcontext * ctx,
 				ctx->Color.ColorLogicOpEnabled ?
 				ctx->Color.LogicOp : GL_COPY)) {
 	    DBG("%s: blit failure\n", __FUNCTION__);
-	    UNLOCK_HARDWARE(intel);
 	    return GL_FALSE;
 	 }
       }
    }
 out:
-   UNLOCK_HARDWARE(intel);
+
+   intel_check_front_buffer_rendering(intel);
 
    DBG("%s: success\n", __FUNCTION__);
    return GL_TRUE;
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
index 5ffa847fd4..b870e9315e 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
@@ -69,7 +69,6 @@ intel_stencil_drawpixels(GLcontext * ctx,
    GLfloat vertices[4][2];
    struct intel_renderbuffer *irb;
    struct intel_renderbuffer *depth_irb;
-   struct gl_renderbuffer *rb;
    struct gl_pixelstore_attrib old_unpack;
    GLstencil *stencil_pixels;
    int row, y1, y2;
@@ -169,8 +168,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
     * buffer.
     */
    depth_irb = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_DEPTH);
-   irb = intel_create_renderbuffer(GL_RGBA8);
-   rb = &irb->Base;
+   irb = intel_create_renderbuffer(MESA_FORMAT_ARGB8888);
    irb->Base.Width = depth_irb->Base.Width;
    irb->Base.Height = depth_irb->Base.Height;
    intel_renderbuffer_set_region(irb, depth_irb->region);
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c
index 4707500180..9c0fdc6067 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c
@@ -77,7 +77,7 @@ do_texture_readpixels(GLcontext * ctx,
    struct intel_context *intel = intel_context(ctx);
    intelScreenPrivate *screen = intel->intelScreen;
    GLint pitch = pack->RowLength ? pack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   __DRIdrawable *dPriv = intel->driDrawable;
    int textureFormat;
    GLenum glTextureFormat;
    int destFormat, depthFormat, destPitch;
@@ -105,15 +105,12 @@ do_texture_readpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   LOCK_HARDWARE(intel);
-
    if (intel->driDrawable->numClipRects) {
       intel->vtbl.install_meta_state(intel);
       intel->vtbl.meta_no_depth_write(intel);
       intel->vtbl.meta_no_stencil_write(intel);
 
       if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-         UNLOCK_HARDWARE(intel);
          SET_STATE(i830, state);
          if (INTEL_DEBUG & DEBUG_PIXEL)
             fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
@@ -150,7 +147,6 @@ do_texture_readpixels(GLcontext * ctx,
 
       intel->vtbl.leave_meta_state(intel);
    }
-   UNLOCK_HARDWARE(intel);
 
    intel_region_wait_fence(ctx, dest_region);   /* required by GL */
    return GL_TRUE;
@@ -224,7 +220,6 @@ do_blit_readpixels(GLcontext * ctx,
     * fire with lock held to guarentee cliprects are correct.
     */
    intelFlush(&intel->ctx);
-   LOCK_HARDWARE(intel);
 
    if (intel->driReadDrawable->numClipRects) {
       GLboolean all = (width * height * src->cpp == dst->Base.Size &&
@@ -233,7 +228,7 @@ do_blit_readpixels(GLcontext * ctx,
       dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
 						  all ? INTEL_WRITE_FULL :
 						  INTEL_WRITE_PART);
-      __DRIdrawablePrivate *dPriv = intel->driReadDrawable;
+      __DRIdrawable *dPriv = intel->driReadDrawable;
       int nbox = dPriv->numClipRects;
       drm_clip_rect_t *box = dPriv->pClipRects;
       drm_clip_rect_t rect;
@@ -261,12 +256,10 @@ do_blit_readpixels(GLcontext * ctx,
 				rect.y2 - src_rect.y2,
 				rect.x2 - rect.x1, rect.y2 - rect.y1,
 				GL_COPY)) {
-	    UNLOCK_HARDWARE(intel);
 	    return GL_FALSE;
 	 }
       }
    }
-   UNLOCK_HARDWARE(intel);
 
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s - DONE\n", __FUNCTION__);
@@ -285,11 +278,11 @@ intelReadPixels(GLcontext * ctx,
 
    intelFlush(ctx);
 
-#ifdef I915
    if (do_blit_readpixels
        (ctx, x, y, width, height, format, type, pack, pixels))
       return;
 
+#ifdef I915
    if (do_texture_readpixels
        (ctx, x, y, width, height, format, type, pack, pixels))
       return;
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index a86c66a844..61aefa01b8 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -362,14 +362,12 @@ intel_region_data(struct intel_context *intel,
          intel_region_cow(intel, dst);
    }
 
-   LOCK_HARDWARE(intel);
    _mesa_copy_rect(intel_region_map(intel, dst) + dst_offset,
                    dst->cpp,
                    dst->pitch,
                    dstx, dsty, width, height, src, src_pitch, srcx, srcy);
 
    intel_region_unmap(intel, dst);
-   UNLOCK_HARDWARE(intel);
 }
 
 /* Copy rectangular sub-regions. Need better logic about when to
@@ -485,7 +483,6 @@ intel_region_cow(struct intel_context *intel, struct intel_region *region)
    /* Now blit from the texture buffer to the new buffer: 
     */
 
-   LOCK_HARDWARE(intel);
    ok = intelEmitCopyBlit(intel,
                           region->cpp,
                           region->pitch, pbo->buffer, 0, region->tiling,
@@ -494,7 +491,6 @@ intel_region_cow(struct intel_context *intel, struct intel_region *region)
                           region->pitch, region->height,
                           GL_COPY);
    assert(ok);
-   UNLOCK_HARDWARE(intel);
 }
 
 dri_bo *
@@ -510,126 +506,3 @@ intel_region_buffer(struct intel_context *intel,
 
    return region->buffer;
 }
-
-static struct intel_region *
-intel_recreate_static(struct intel_context *intel,
-		      const char *name,
-		      struct intel_region *region,
-		      intelRegion *region_desc)
-{
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   int ret;
-
-   if (region == NULL) {
-      region = calloc(sizeof(*region), 1);
-      region->refcount = 1;
-      _DBG("%s creating new region %p\n", __FUNCTION__, region);
-   }
-   else {
-      _DBG("%s %p\n", __FUNCTION__, region);
-   }
-
-   if (intel->ctx.Visual.rgbBits == 24)
-      region->cpp = 4;
-   else
-      region->cpp = intel->ctx.Visual.rgbBits / 8;
-   region->pitch = intelScreen->pitch;
-   region->width = intelScreen->width;
-   region->height = intelScreen->height;
-
-   if (region->buffer != NULL) {
-      dri_bo_unreference(region->buffer);
-      region->buffer = NULL;
-   }
-
-   if (intel->ttm) {
-      assert(region_desc->bo_handle != -1);
-      region->buffer = intel_bo_gem_create_from_name(intel->bufmgr,
-						     name,
-						     region_desc->bo_handle);
-
-      ret = dri_bo_get_tiling(region->buffer, &region->tiling,
-			      &region->bit_6_swizzle);
-      if (ret != 0) {
-	 fprintf(stderr, "Couldn't get tiling of buffer %d (%s): %s\n",
-		 region_desc->bo_handle, name, strerror(-ret));
-	 intel_region_release(&region);
-	 return NULL;
-      }
-   } else {
-      if (region->classic_map != NULL) {
-	 drmUnmap(region->classic_map,
-		  region->pitch * region->cpp * region->height);
-	 region->classic_map = NULL;
-      }
-      ret = drmMap(intel->driFd, region_desc->handle,
-		   region->pitch * region->cpp * region->height,
-		   &region->classic_map);
-      if (ret != 0) {
-	 fprintf(stderr, "Failed to drmMap %s buffer\n", name);
-	 free(region);
-	 return NULL;
-      }
-
-      region->buffer = intel_bo_fake_alloc_static(intel->bufmgr,
-						  name,
-						  region_desc->offset,
-						  region->pitch * region->cpp *
-						  region->height,
-						  region->classic_map);
-
-      /* The sarea just gives us a boolean for whether it's tiled or not,
-       * instead of which tiling mode it is.  Guess.
-       */
-      if (region_desc->tiled) {
-	 if (IS_965(intel->intelScreen->deviceID) &&
-	     region_desc == &intelScreen->depth)
-	    region->tiling = I915_TILING_Y;
-	 else
-	    region->tiling = I915_TILING_X;
-      } else {
-	 region->tiling = I915_TILING_NONE;
-      }
-
-      region->bit_6_swizzle = I915_BIT_6_SWIZZLE_NONE;
-   }
-
-   assert(region->buffer != NULL);
-
-   return region;
-}
-
-/**
- * Create intel_region structs to describe the static front, back, and depth
- * buffers created by the xserver.
- *
- * Although FBO's mean we now no longer use these as render targets in
- * all circumstances, they won't go away until the back and depth
- * buffers become private, and the front buffer will remain even then.
- *
- * Note that these don't allocate video memory, just describe
- * allocations alread made by the X server.
- */
-void
-intel_recreate_static_regions(struct intel_context *intel)
-{
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-
-   intel->front_region =
-      intel_recreate_static(intel, "front",
-			    intel->front_region,
-			    &intelScreen->front);
-
-   intel->back_region =
-      intel_recreate_static(intel, "back",
-			    intel->back_region,
-			    &intelScreen->back);
-
-   /* Still assumes front.cpp == depth.cpp.  We can kill this when we move to
-    * private buffers.
-    */
-   intel->depth_region =
-      intel_recreate_static(intel, "depth",
-			    intel->depth_region,
-			    &intelScreen->depth);
-}
diff --git a/src/mesa/drivers/dri/intel/intel_regions.h b/src/mesa/drivers/dri/intel/intel_regions.h
index 0d379bdc6e..535fcd7be0 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.h
+++ b/src/mesa/drivers/dri/intel/intel_regions.h
@@ -62,6 +62,8 @@ struct intel_region
    GLuint map_refcount;  /**< Reference count for mapping */
 
    GLuint draw_offset; /**< Offset of drawing address within the region */
+   GLuint draw_x, draw_y; /**< Offset of drawing within the region */
+
    uint32_t tiling; /**< Which tiling mode the region is in */
    uint32_t bit_6_swizzle; /**< GEM flag for address swizzling requirement */
    drmAddress classic_map; /**< drmMap of the region when not in GEM mode */
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index 24f7fbc992..5165716e09 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -31,7 +31,6 @@
 #include "main/renderbuffer.h"
 
 #include "utils.h"
-#include "vblank.h"
 #include "xmlpool.h"
 
 #include "intel_batchbuffer.h"
@@ -41,7 +40,6 @@
 #include "intel_extensions.h"
 #include "intel_fbo.h"
 #include "intel_regions.h"
-#include "intel_swapbuffers.h"
 #include "intel_screen.h"
 #include "intel_span.h"
 #include "intel_tex.h"
@@ -57,7 +55,6 @@
 PUBLIC const char __driConfigOptions[] =
    DRI_CONF_BEGIN
    DRI_CONF_SECTION_PERFORMANCE
-      DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS)
       DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_ALWAYS_SYNC)
       /* Options correspond to DRI_CONF_BO_REUSE_DISABLED,
        * DRI_CONF_BO_REUSE_ALL
@@ -99,133 +96,12 @@ PUBLIC const char __driConfigOptions[] =
    DRI_CONF_SECTION_END
 DRI_CONF_END;
 
-const GLuint __driNConfigOptions = 12;
+const GLuint __driNConfigOptions = 11;
 
 #ifdef USE_NEW_INTERFACE
 static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
 #endif /*USE_NEW_INTERFACE */
 
-/**
- * Map all the memory regions described by the screen.
- * \return GL_TRUE if success, GL_FALSE if error.
- */
-GLboolean
-intelMapScreenRegions(__DRIscreenPrivate * sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-
-   if (0)
-      _mesa_printf("TEX 0x%08x ", intelScreen->tex.handle);
-   if (intelScreen->tex.size != 0) {
-      if (drmMap(sPriv->fd,
-		 intelScreen->tex.handle,
-		 intelScreen->tex.size,
-		 (drmAddress *) & intelScreen->tex.map) != 0) {
-	 intelUnmapScreenRegions(intelScreen);
-	 return GL_FALSE;
-      }
-   }
-
-   return GL_TRUE;
-}
-
-void
-intelUnmapScreenRegions(intelScreenPrivate * intelScreen)
-{
-   if (intelScreen->tex.map) {
-      drmUnmap(intelScreen->tex.map, intelScreen->tex.size);
-      intelScreen->tex.map = NULL;
-   }
-}
-
-
-static void
-intelPrintDRIInfo(intelScreenPrivate * intelScreen,
-                  __DRIscreenPrivate * sPriv, I830DRIPtr gDRIPriv)
-{
-   fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->front.size, intelScreen->front.offset,
-           intelScreen->pitch);
-   fprintf(stderr, "*** Back size:    0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->back.size, intelScreen->back.offset,
-           intelScreen->pitch);
-   fprintf(stderr, "*** Depth size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->depth.size, intelScreen->depth.offset,
-           intelScreen->pitch);
-   fprintf(stderr, "*** Texture size: 0x%x  offset: 0x%x\n",
-           intelScreen->tex.size, intelScreen->tex.offset);
-   fprintf(stderr, "*** Memory : 0x%x\n", gDRIPriv->mem);
-}
-
-
-static void
-intelPrintSAREA(const drm_i915_sarea_t * sarea)
-{
-   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width,
-           sarea->height);
-   fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
-   fprintf(stderr,
-           "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
-           sarea->front_offset, sarea->front_size,
-           (unsigned) sarea->front_handle, sarea->front_tiled);
-   fprintf(stderr,
-           "SAREA: back  offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
-           sarea->back_offset, sarea->back_size,
-           (unsigned) sarea->back_handle, sarea->back_tiled);
-   fprintf(stderr, "SAREA: depth offset: 0x%08x  size: 0x%x  handle: 0x%x tiled: %d\n",
-           sarea->depth_offset, sarea->depth_size,
-           (unsigned) sarea->depth_handle, sarea->depth_tiled);
-   fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->tex_offset, sarea->tex_size, (unsigned) sarea->tex_handle);
-}
-
-
-/**
- * A number of the screen parameters are obtained/computed from
- * information in the SAREA.  This function updates those parameters.
- */
-static void
-intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
-                           drm_i915_sarea_t * sarea)
-{
-   intelScreen->width = sarea->width;
-   intelScreen->height = sarea->height;
-   intelScreen->pitch = sarea->pitch;
-
-   intelScreen->front.offset = sarea->front_offset;
-   intelScreen->front.handle = sarea->front_handle;
-   intelScreen->front.size = sarea->front_size;
-   intelScreen->front.tiled = sarea->front_tiled;
-
-   intelScreen->back.offset = sarea->back_offset;
-   intelScreen->back.handle = sarea->back_handle;
-   intelScreen->back.size = sarea->back_size;
-   intelScreen->back.tiled = sarea->back_tiled;
-
-   intelScreen->depth.offset = sarea->depth_offset;
-   intelScreen->depth.handle = sarea->depth_handle;
-   intelScreen->depth.size = sarea->depth_size;
-   intelScreen->depth.tiled = sarea->depth_tiled;
-
-   if (intelScreen->driScrnPriv->ddx_version.minor >= 9) {
-      intelScreen->front.bo_handle = sarea->front_bo_handle;
-      intelScreen->back.bo_handle = sarea->back_bo_handle;
-      intelScreen->depth.bo_handle = sarea->depth_bo_handle;
-   } else {
-      intelScreen->front.bo_handle = -1;
-      intelScreen->back.bo_handle = -1;
-      intelScreen->depth.bo_handle = -1;
-   }
-
-   intelScreen->tex.offset = sarea->tex_offset;
-   intelScreen->logTextureGranularity = sarea->log_tex_granularity;
-   intelScreen->tex.handle = sarea->tex_handle;
-   intelScreen->tex.size = sarea->tex_size;
-
-   if (0)
-      intelPrintSAREA(sarea);
-}
-
 static const __DRItexOffsetExtension intelTexOffsetExtension = {
    { __DRI_TEX_OFFSET },
    intelSetTexOffset,
@@ -239,17 +115,13 @@ static const __DRItexBufferExtension intelTexBufferExtension = {
 
 static const __DRIextension *intelScreenExtensions[] = {
     &driReadDrawableExtension,
-    &driCopySubBufferExtension.base,
-    &driSwapControlExtension.base,
-    &driFrameTrackingExtension.base,
-    &driMediaStreamCounterExtension.base,
     &intelTexOffsetExtension.base,
     &intelTexBufferExtension.base,
     NULL
 };
 
 static GLboolean
-intel_get_param(__DRIscreenPrivate *psp, int param, int *value)
+intel_get_param(__DRIscreen *psp, int param, int *value)
 {
    int ret;
    struct drm_i915_getparam gp;
@@ -266,68 +138,12 @@ intel_get_param(__DRIscreenPrivate *psp, int param, int *value)
    return GL_TRUE;
 }
 
-static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
-{
-   intelScreenPrivate *intelScreen;
-   I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
-   drm_i915_sarea_t *sarea;
-
-   if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
-      fprintf(stderr,
-              "\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
-      return GL_FALSE;
-   }
-
-   /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
-   if (!intelScreen) {
-      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
-      return GL_FALSE;
-   }
-   /* parse information in __driConfigOptions */
-   driParseOptionInfo(&intelScreen->optionCache,
-                      __driConfigOptions, __driNConfigOptions);
-
-   intelScreen->driScrnPriv = sPriv;
-   sPriv->private = (void *) intelScreen;
-   sarea = (drm_i915_sarea_t *)
-      (((GLubyte *) sPriv->pSAREA) + gDRIPriv->sarea_priv_offset);
-   intelScreen->sarea = sarea;
-
-   intelScreen->deviceID = gDRIPriv->deviceID;
-
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr, "\nERROR!  mapping regions\n");
-      _mesa_free(intelScreen);
-      sPriv->private = NULL;
-      return GL_FALSE;
-   }
-
-   if (0)
-      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
-
-   intelScreen->drmMinor = sPriv->drm_version.minor;
-
-   /* Determine if IRQs are active? */
-   if (!intel_get_param(sPriv, I915_PARAM_IRQ_ACTIVE,
-			&intelScreen->irq_active))
-      return GL_FALSE;
-
-   sPriv->extensions = intelScreenExtensions;
-
-   return GL_TRUE;
-}
-
-
 static void
-intelDestroyScreen(__DRIscreenPrivate * sPriv)
+intelDestroyScreen(__DRIscreen * sPriv)
 {
    intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
 
    dri_bufmgr_destroy(intelScreen->bufmgr);
-   intelUnmapScreenRegions(intelScreen);
    driDestroyOptionInfo(&intelScreen->optionCache);
 
    FREE(intelScreen);
@@ -339,160 +155,106 @@ intelDestroyScreen(__DRIscreenPrivate * sPriv)
  * This is called when we need to set up GL rendering to a new X window.
  */
 static GLboolean
-intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
-                  __DRIdrawablePrivate * driDrawPriv,
+intelCreateBuffer(__DRIscreen * driScrnPriv,
+                  __DRIdrawable * driDrawPriv,
                   const __GLcontextModes * mesaVis, GLboolean isPixmap)
 {
+   struct intel_renderbuffer *rb;
+
    if (isPixmap) {
       return GL_FALSE;          /* not implemented */
    }
    else {
       GLboolean swStencil = (mesaVis->stencilBits > 0 &&
                              mesaVis->depthBits != 24);
-      GLenum rgbFormat;
+      gl_format rgbFormat;
 
-      struct intel_framebuffer *intel_fb = CALLOC_STRUCT(intel_framebuffer);
+      struct gl_framebuffer *fb = CALLOC_STRUCT(gl_framebuffer);
 
-      if (!intel_fb)
+      if (!fb)
 	 return GL_FALSE;
 
-      _mesa_initialize_framebuffer(&intel_fb->Base, mesaVis);
+      _mesa_initialize_framebuffer(fb, mesaVis);
 
       if (mesaVis->redBits == 5)
-	 rgbFormat = GL_RGB5;
+	 rgbFormat = MESA_FORMAT_RGB565;
       else if (mesaVis->alphaBits == 0)
-	 rgbFormat = GL_RGB8;
+	 rgbFormat = MESA_FORMAT_XRGB8888;
       else
-	 rgbFormat = GL_RGBA8;
+	 rgbFormat = MESA_FORMAT_ARGB8888;
 
       /* setup the hardware-based renderbuffers */
-      intel_fb->color_rb[0] = intel_create_renderbuffer(rgbFormat);
-      _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT,
-			     &intel_fb->color_rb[0]->Base);
+      rb = intel_create_renderbuffer(rgbFormat);
+      _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &rb->Base);
 
       if (mesaVis->doubleBufferMode) {
-	 intel_fb->color_rb[1] = intel_create_renderbuffer(rgbFormat);
-
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT,
-				&intel_fb->color_rb[1]->Base);
-
+	 rb = intel_create_renderbuffer(rgbFormat);
+         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &rb->Base);
       }
 
       if (mesaVis->depthBits == 24) {
 	 if (mesaVis->stencilBits == 8) {
 	    /* combined depth/stencil buffer */
 	    struct intel_renderbuffer *depthStencilRb
-	       = intel_create_renderbuffer(GL_DEPTH24_STENCIL8_EXT);
+	       = intel_create_renderbuffer(MESA_FORMAT_S8_Z24);
 	    /* note: bind RB to two attachment points */
-	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
-				   &depthStencilRb->Base);
-	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_STENCIL,
-				   &depthStencilRb->Base);
+	    _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthStencilRb->Base);
+	    _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &depthStencilRb->Base);
 	 } else {
 	    struct intel_renderbuffer *depthRb
-	       = intel_create_renderbuffer(GL_DEPTH_COMPONENT24);
-	    _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
-				   &depthRb->Base);
+	       = intel_create_renderbuffer(MESA_FORMAT_X8_Z24);
+	    _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
 	 }
       }
       else if (mesaVis->depthBits == 16) {
          /* just 16-bit depth buffer, no hw stencil */
          struct intel_renderbuffer *depthRb
-	    = intel_create_renderbuffer(GL_DEPTH_COMPONENT16);
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH, &depthRb->Base);
+	    = intel_create_renderbuffer(MESA_FORMAT_Z16);
+         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
       }
 
       /* now add any/all software-based renderbuffers we may need */
-      _mesa_add_soft_renderbuffers(&intel_fb->Base,
+      _mesa_add_soft_renderbuffers(fb,
                                    GL_FALSE, /* never sw color */
                                    GL_FALSE, /* never sw depth */
                                    swStencil, mesaVis->accumRedBits > 0,
                                    GL_FALSE, /* never sw alpha */
                                    GL_FALSE  /* never sw aux */ );
-      driDrawPriv->driverPrivate = (void *) intel_fb;
+      driDrawPriv->driverPrivate = fb;
 
       return GL_TRUE;
    }
 }
 
 static void
-intelDestroyBuffer(__DRIdrawablePrivate * driDrawPriv)
-{
-   struct intel_framebuffer *intel_fb = driDrawPriv->driverPrivate;
-   struct intel_renderbuffer *depth_rb;
-   struct intel_renderbuffer *stencil_rb;
-
-   if (intel_fb) {
-      if (intel_fb->color_rb[0]) {
-         intel_renderbuffer_set_region(intel_fb->color_rb[0], NULL);
-      }
-
-      if (intel_fb->color_rb[1]) {
-         intel_renderbuffer_set_region(intel_fb->color_rb[1], NULL);
-      }
-
-      depth_rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
-      if (depth_rb) {
-         intel_renderbuffer_set_region(depth_rb, NULL);
-      }
-
-      stencil_rb = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
-      if (stencil_rb) {
-         intel_renderbuffer_set_region(stencil_rb, NULL);
-      }
-   }
-
-   _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
-}
-
-
-/**
- * Get information about previous buffer swaps.
- */
-static int
-intelGetSwapInfo(__DRIdrawablePrivate * dPriv, __DRIswapInfo * sInfo)
+intelDestroyBuffer(__DRIdrawable * driDrawPriv)
 {
-   struct intel_framebuffer *intel_fb;
-
-   if ((dPriv == NULL) || (dPriv->driverPrivate == NULL)
-       || (sInfo == NULL)) {
-      return -1;
-   }
-
-   intel_fb = dPriv->driverPrivate;
-   sInfo->swap_count = intel_fb->swap_count;
-   sInfo->swap_ust = intel_fb->swap_ust;
-   sInfo->swap_missed_count = intel_fb->swap_missed_count;
-
-   sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
-      ? driCalculateSwapUsage(dPriv, 0, intel_fb->swap_missed_ust)
-      : 0.0;
-
-   return 0;
+    struct gl_framebuffer *fb = driDrawPriv->driverPrivate;
+  
+    _mesa_reference_framebuffer(&fb, NULL);
 }
 
-
 /* There are probably better ways to do this, such as an
  * init-designated function to register chipids and createcontext
  * functions.
  */
 extern GLboolean i830CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
+                                   __DRIcontext * driContextPriv,
                                    void *sharedContextPrivate);
 
 extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
+                                   __DRIcontext * driContextPriv,
                                    void *sharedContextPrivate);
 extern GLboolean brwCreateContext(const __GLcontextModes * mesaVis,
-				  __DRIcontextPrivate * driContextPriv,
+				  __DRIcontext * driContextPriv,
 				  void *sharedContextPrivate);
 
 static GLboolean
 intelCreateContext(const __GLcontextModes * mesaVis,
-                   __DRIcontextPrivate * driContextPriv,
+                   __DRIcontext * driContextPriv,
                    void *sharedContextPrivate)
 {
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
 
 #ifdef I915
@@ -513,152 +275,20 @@ intelCreateContext(const __GLcontextModes * mesaVis,
    return GL_FALSE;
 }
 
-
-static __DRIconfig **
-intelFillInModes(__DRIscreenPrivate *psp,
-		 unsigned pixel_bits, unsigned depth_bits,
-                 unsigned stencil_bits, GLboolean have_back_buffer)
-{
-   __DRIconfig **configs;
-   __GLcontextModes *m;
-   unsigned depth_buffer_factor;
-   unsigned back_buffer_factor;
-   int i;
-
-   /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
-    * support pageflipping at all.
-    */
-   static const GLenum back_buffer_modes[] = {
-      GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
-   };
-
-   uint8_t depth_bits_array[3];
-   uint8_t stencil_bits_array[3];
-   uint8_t msaa_samples_array[1];
-
-   depth_bits_array[0] = 0;
-   depth_bits_array[1] = depth_bits;
-   depth_bits_array[2] = depth_bits;
-
-   /* Just like with the accumulation buffer, always provide some modes
-    * with a stencil buffer.  It will be a sw fallback, but some apps won't
-    * care about that.
-    */
-   stencil_bits_array[0] = 0;
-   stencil_bits_array[1] = 0;
-   if (depth_bits == 24)
-      stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   msaa_samples_array[0] = 0;
-
-   depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
-   back_buffer_factor = (have_back_buffer) ? 3 : 1;
-
-   if (pixel_bits == 16) {
-      configs = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5,
-				 depth_bits_array, stencil_bits_array,
-				 depth_buffer_factor, back_buffer_modes,
-				 back_buffer_factor,
-				 msaa_samples_array, 1);
-   }
-   else {
-      __DRIconfig **configs_a8r8g8b8;
-      __DRIconfig **configs_x8r8g8b8;
-
-      configs_a8r8g8b8 = driCreateConfigs(GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV,
-					  depth_bits_array,
-					  stencil_bits_array,
-					  depth_buffer_factor,
-					  back_buffer_modes,
-					  back_buffer_factor,
-					  msaa_samples_array, 1);
-      configs_x8r8g8b8 = driCreateConfigs(GL_BGR, GL_UNSIGNED_INT_8_8_8_8_REV,
-					  depth_bits_array,
-					  stencil_bits_array,
-					  depth_buffer_factor,
-					  back_buffer_modes,
-					  back_buffer_factor,
-					  msaa_samples_array, 1);
-      configs = driConcatConfigs(configs_a8r8g8b8, configs_x8r8g8b8);
-   }
-
-   if (configs == NULL) {
-    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
-              __LINE__);
-      return NULL;
-   }
-
-   /* Mark the visual as slow if there are "fake" stencil bits.
-    */
-   for (i = 0; configs[i]; i++) {
-      m = &configs[i]->modes;
-      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
-         m->visualRating = GLX_SLOW_CONFIG;
-      }
-   }
-
-   return configs;
-}
-
 static GLboolean
 intel_init_bufmgr(intelScreenPrivate *intelScreen)
 {
-   GLboolean gem_disable = getenv("INTEL_NO_GEM") != NULL;
-   int gem_kernel = 0;
-   GLboolean gem_supported;
-   struct drm_i915_getparam gp;
-   __DRIscreenPrivate *spriv = intelScreen->driScrnPriv;
+   __DRIscreen *spriv = intelScreen->driScrnPriv;
    int num_fences = 0;
 
    intelScreen->no_hw = getenv("INTEL_NO_HW") != NULL;
 
-   gp.param = I915_PARAM_HAS_GEM;
-   gp.value = &gem_kernel;
-
-   (void) drmCommandWriteRead(spriv->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
-
-   /* If we've got a new enough DDX that's initializing GEM and giving us
-    * object handles for the shared buffers, use that.
-    */
-   intelScreen->ttm = GL_FALSE;
-   if (intelScreen->driScrnPriv->dri2.enabled)
-       gem_supported = GL_TRUE;
-   else if (intelScreen->driScrnPriv->ddx_version.minor >= 9 &&
-	    gem_kernel &&
-	    intelScreen->front.bo_handle != -1)
-       gem_supported = GL_TRUE;
-   else
-       gem_supported = GL_FALSE;
-
-   if (!gem_disable && gem_supported) {
-      intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
-      if (intelScreen->bufmgr != NULL)
-	 intelScreen->ttm = GL_TRUE;
-   }
+   intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
    /* Otherwise, use the classic buffer manager. */
    if (intelScreen->bufmgr == NULL) {
-      if (gem_disable) {
-	 _mesa_warning(NULL, "GEM disabled.  Using classic.");
-      } else {
-	 _mesa_warning(NULL,
-                       "Failed to initialize GEM.  Falling back to classic.");
-      }
-
-      if (intelScreen->tex.size == 0) {
-	 fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
-		 __func__, __LINE__);
-	 return GL_FALSE;
-      }
-
-      intelScreen->bufmgr =
-	 intel_bufmgr_fake_init(spriv->fd,
-				intelScreen->tex.offset,
-				intelScreen->tex.map,
-				intelScreen->tex.size,
-				(unsigned int * volatile)
-				&intelScreen->sarea->last_dispatch);
+      fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
+	      __func__, __LINE__);
+      return GL_FALSE;
    }
 
    if (intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences))
@@ -671,81 +301,12 @@ intel_init_bufmgr(intelScreenPrivate *intelScreen)
 
 /**
  * This is the driver specific part of the createNewScreen entry point.
- * Called when using legacy DRI.
- * 
- * \todo maybe fold this into intelInitDriver
- *
- * \return the __GLcontextModes supported by this driver
- */
-static const __DRIconfig **intelInitScreen(__DRIscreenPrivate *psp)
-{
-   intelScreenPrivate *intelScreen;
-#ifdef I915
-   static const __DRIversion ddx_expected = { 1, 5, 0 };
-#else
-   static const __DRIversion ddx_expected = { 1, 6, 0 };
-#endif
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 5, 0 };
-   I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-
-   if (!driCheckDriDdxDrmVersions2("i915",
-                                   &psp->dri_version, &dri_expected,
-                                   &psp->ddx_version, &ddx_expected,
-                                   &psp->drm_version, &drm_expected)) {
-      return NULL;
-   }
-
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   intelInitExtensions(NULL, GL_TRUE);
-	   
-   if (!intelInitDriver(psp))
-       return NULL;
-
-   psp->extensions = intelScreenExtensions;
-
-   intelScreen = psp->private;
-   if (!intel_init_bufmgr(intelScreen))
-       return GL_FALSE;
-
-   return (const __DRIconfig **)
-       intelFillInModes(psp, dri_priv->cpp * 8,
-			(dri_priv->cpp == 2) ? 16 : 24,
-			(dri_priv->cpp == 2) ? 0  : 8, 1);
-}
-
-struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen)
-{
-  /*
-   * This should probably change to have the screen allocate a dummy
-   * context at screen creation. For now just use the current context.
-   */
-
-  GET_CURRENT_CONTEXT(ctx);
-  if (ctx == NULL) {
-     _mesa_problem(NULL, "No current context in intelScreenContext\n");
-     return NULL;
-  }
-  return intel_context(ctx);
-}
-
-/**
- * This is the driver specific part of the createNewScreen entry point.
  * Called when using DRI2.
  *
  * \return the __GLcontextModes supported by this driver
  */
 static const
-__DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
+__DRIconfig **intelInitScreen2(__DRIscreen *psp)
 {
    intelScreenPrivate *intelScreen;
    GLenum fb_format[3];
@@ -760,18 +321,6 @@ __DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
    int color;
    __DRIconfig **configs = NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   intelInitExtensions(NULL, GL_TRUE);
-
    /* Allocate the private area */
    intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
    if (!intelScreen) {
@@ -866,19 +415,19 @@ __DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
 }
 
 const struct __DriverAPIRec driDriverAPI = {
-   .InitScreen		 = intelInitScreen,
    .DestroyScreen	 = intelDestroyScreen,
    .CreateContext	 = intelCreateContext,
    .DestroyContext	 = intelDestroyContext,
    .CreateBuffer	 = intelCreateBuffer,
    .DestroyBuffer	 = intelDestroyBuffer,
-   .SwapBuffers		 = intelSwapBuffers,
    .MakeCurrent		 = intelMakeCurrent,
    .UnbindContext	 = intelUnbindContext,
-   .GetSwapInfo		 = intelGetSwapInfo,
-   .GetDrawableMSC	 = driDrawableGetMSC32,
-   .WaitForMSC		 = driWaitForMSC32,
-   .CopySubBuffer	 = intelCopySubBuffer,
-
    .InitScreen2		 = intelInitScreen2,
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driDRI2Extension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index a9b9e109a6..e87e306d86 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -66,7 +66,7 @@ typedef struct
 
    int logTextureGranularity;
 
-   __DRIscreenPrivate *driScrnPriv;
+   __DRIscreen *driScrnPriv;
 
    volatile drm_i915_sarea_t *sarea;
 
@@ -77,7 +77,6 @@ typedef struct
    GLboolean no_hw;
 
    GLboolean no_vbo;
-   int ttm;
    dri_bufmgr *bufmgr;
    GLboolean kernel_exec_fencing;
 
@@ -89,18 +88,18 @@ typedef struct
 
 
 
-extern GLboolean intelMapScreenRegions(__DRIscreenPrivate * sPriv);
+extern GLboolean intelMapScreenRegions(__DRIscreen * sPriv);
 
 extern void intelUnmapScreenRegions(intelScreenPrivate * intelScreen);
 
-extern void intelDestroyContext(__DRIcontextPrivate * driContextPriv);
+extern void intelDestroyContext(__DRIcontext * driContextPriv);
 
-extern GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv);
+extern GLboolean intelUnbindContext(__DRIcontext * driContextPriv);
 
 extern GLboolean
-intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
-                 __DRIdrawablePrivate * driDrawPriv,
-                 __DRIdrawablePrivate * driReadPriv);
+intelMakeCurrent(__DRIcontext * driContextPriv,
+                 __DRIdrawable * driDrawPriv,
+                 __DRIdrawable * driReadPriv);
 
 extern struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen);
 
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
index dcfcad1d95..605734d8e5 100644
--- a/src/mesa/drivers/dri/intel/intel_span.c
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -29,7 +29,6 @@
 #include "main/macros.h"
 #include "main/mtypes.h"
 #include "main/colormac.h"
-#include "main/texformat.h"
 
 #include "intel_buffers.h"
 #include "intel_fbo.h"
@@ -132,18 +131,6 @@ pwrite_8(struct intel_renderbuffer *irb, uint32_t offset, uint8_t val)
    dri_bo_subdata(irb->region->buffer, offset, 1, &val);
 }
 
-static uint32_t
-z24s8_to_s8z24(uint32_t val)
-{
-   return (val << 24) | (val >> 8);
-}
-
-static uint32_t
-s8z24_to_z24s8(uint32_t val)
-{
-   return (val >> 24) | (val << 8);
-}
-
 static uint32_t no_tile_swizzle(struct intel_renderbuffer *irb,
 				int x, int y)
 {
@@ -163,6 +150,9 @@ static uint32_t x_tile_swizzle(struct intel_renderbuffer *irb,
 	int	x_tile_number, y_tile_number;
 	int	tile_off, tile_base;
 	
+        x += irb->region->draw_x;
+        y += irb->region->draw_y;
+
 	tile_stride = (irb->region->pitch * irb->region->cpp) << 3;
 
 	xbyte = x * irb->region->cpp;
@@ -218,6 +208,9 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 	int	x_tile_number, y_tile_number;
 	int	tile_off, tile_base;
 	
+        x += irb->region->draw_x;
+        y += irb->region->draw_y;
+
 	tile_stride = (irb->region->pitch * irb->region->cpp) << 5;
 
 	xbyte = x * irb->region->cpp;
@@ -273,8 +266,11 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
    unsigned int num_cliprects;						\
    struct drm_clip_rect *cliprects;					\
    int x_off, y_off;							\
+   int pitch = irb->region->pitch * irb->region->cpp;			\
+   void *buf = irb->region->buffer->virtual;				\
    GLuint p;								\
    (void) p;								\
+   (void)buf; (void)pitch; /* unused for non-gttmap. */			\
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
 
 /* XXX FBO: this is identical to the macro in spantmp2.h except we get
@@ -296,7 +292,6 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 
 #define Y_FLIP(_y) ((_y) * yScale + yBias)
 
-/* XXX with GEM, these need to tell the kernel */
 #define HW_LOCK()
 
 #define HW_UNLOCK()
@@ -339,7 +334,7 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 #include "intel_spantmp.h"
 
 /* x8r8g8b8 color span and pixel functions */
-#define INTEL_PIXEL_FMT GL_BGRA
+#define INTEL_PIXEL_FMT GL_BGR
 #define INTEL_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
 #define INTEL_READ_VALUE(offset) pread_xrgb8888(irb, offset)
 #define INTEL_WRITE_VALUE(offset, v) pwrite_xrgb8888(irb, offset, v)
@@ -354,6 +349,9 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
    unsigned int num_cliprects;						\
    struct drm_clip_rect *cliprects;					\
    int x_off, y_off;							\
+   int pitch = irb->region->pitch * irb->region->cpp;			\
+   void *buf = irb->region->buffer->virtual;				\
+   (void)buf; (void)pitch; /* unused for non-gttmap. */			\
    intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
 
 
@@ -366,20 +364,22 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 #define INTEL_TAG(name) name##_z16
 #include "intel_depthtmp.h"
 
-/* z24 depthbuffer functions. */
+/* z24x8 depthbuffer functions. */
 #define INTEL_VALUE_TYPE GLuint
 #define INTEL_WRITE_DEPTH(offset, d) pwrite_32(irb, offset, d)
 #define INTEL_READ_DEPTH(offset) pread_32(irb, offset)
-#define INTEL_TAG(name) name##_z24
+#define INTEL_TAG(name) name##_z24_x8
 #include "intel_depthtmp.h"
 
-/* z24s8 depthbuffer functions. */
-#define INTEL_VALUE_TYPE GLuint
-#define INTEL_WRITE_DEPTH(offset, d) pwrite_32(irb, offset, z24s8_to_s8z24(d))
-#define INTEL_READ_DEPTH(offset) s8z24_to_z24s8(pread_32(irb, offset))
-#define INTEL_TAG(name) name##_z24_s8
-#include "intel_depthtmp.h"
 
+/**
+ ** 8-bit stencil function (XXX FBO: This is obsolete)
+ **/
+/* XXX */
+#define WRITE_STENCIL(_x, _y, d) pwrite_8(irb, NO_TILE(_x, _y) + 3, d)
+#define READ_STENCIL(d, _x, _y) d = pread_8(irb, NO_TILE(_x, _y) + 3);
+#define TAG(x) intel_gttmap_##x##_z24_s8
+#include "stenciltmp.h"
 
 /**
  ** 8-bit stencil function (XXX FBO: This is obsolete)
@@ -413,6 +413,9 @@ intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb)
    if (irb == NULL || irb->region == NULL)
       return;
 
+   if (intel->intelScreen->kernel_exec_fencing)
+      drm_intel_gem_bo_map_gtt(irb->region->buffer);
+
    intel_set_span_functions(intel, rb);
 }
 
@@ -425,7 +428,10 @@ intel_renderbuffer_unmap(struct intel_context *intel,
    if (irb == NULL || irb->region == NULL)
       return;
 
-   clear_span_cache(irb);
+   if (intel->intelScreen->kernel_exec_fencing)
+      drm_intel_gem_bo_unmap_gtt(irb->region->buffer);
+   else
+      clear_span_cache(irb);
 
    rb->GetRow = NULL;
    rb->PutRow = NULL;
@@ -494,6 +500,8 @@ intel_map_unmap_framebuffer(struct intel_context *intel,
       else
          intel_renderbuffer_unmap(intel, fb->_StencilBuffer->Wrapped);
    }
+
+   intel_check_front_buffer_rendering(intel);
 }
 
 /**
@@ -509,7 +517,6 @@ intelSpanRenderStart(GLcontext * ctx)
    GLuint i;
 
    intelFlush(&intel->ctx);
-   LOCK_HARDWARE(intel);
 
    for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
       if (ctx->Texture.Unit[i]._ReallyEnabled) {
@@ -545,8 +552,6 @@ intelSpanRenderFinish(GLcontext * ctx)
    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);
 }
 
 
@@ -605,17 +610,56 @@ intel_set_span_functions(struct intel_context *intel,
 			 struct gl_renderbuffer *rb)
 {
    struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
-   uint32_t tiling;
+   uint32_t tiling = irb->region->tiling;
+
+   if (intel->intelScreen->kernel_exec_fencing) {
+      switch (irb->Base.Format) {
+      case MESA_FORMAT_RGB565:
+	 intel_gttmap_InitPointers_RGB565(rb);
+	 break;
+      case MESA_FORMAT_ARGB4444:
+	 intel_gttmap_InitPointers_ARGB4444(rb);
+	 break;
+      case MESA_FORMAT_ARGB1555:
+	 intel_gttmap_InitPointers_ARGB1555(rb);
+	 break;
+      case MESA_FORMAT_XRGB8888:
+         intel_gttmap_InitPointers_xRGB8888(rb);
+	 break;
+      case MESA_FORMAT_ARGB8888:
+	 intel_gttmap_InitPointers_ARGB8888(rb);
+	 break;
+      case MESA_FORMAT_Z16:
+	 intel_gttmap_InitDepthPointers_z16(rb);
+	 break;
+      case MESA_FORMAT_X8_Z24:
+	 intel_gttmap_InitDepthPointers_z24_x8(rb);
+	 break;
+      case MESA_FORMAT_S8_Z24:
+	 /* There are a few different ways SW asks us to access the S8Z24 data:
+	  * Z24 depth-only depth reads
+	  * S8Z24 depth reads
+	  * S8Z24 stencil reads.
+	  */
+	 if (rb->Format == MESA_FORMAT_S8_Z24) {
+	    intel_gttmap_InitDepthPointers_z24_x8(rb);
+	 } else if (rb->Format == MESA_FORMAT_S8) {
+	    intel_gttmap_InitStencilPointers_z24_s8(rb);
+	 }
+	 break;
+      default:
+	 _mesa_problem(NULL,
+		       "Unexpected MesaFormat %d in intelSetSpanFunctions",
+		       irb->Base.Format);
+	 break;
+      }
+      return;
+   }
 
    /* If in GEM mode, we need to do the tile address swizzling ourselves,
     * instead of the fence registers handling it.
     */
-   if (intel->ttm)
-      tiling = irb->region->tiling;
-   else
-      tiling = I915_TILING_NONE;
-
-   switch (irb->texformat->MesaFormat) {
+   switch (irb->Base.Format) {
    case MESA_FORMAT_RGB565:
       switch (tiling) {
       case I915_TILING_NONE:
@@ -658,35 +702,33 @@ intel_set_span_functions(struct intel_context *intel,
 	 break;
       }
       break;
+   case MESA_FORMAT_XRGB8888:
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+         intelInitPointers_xRGB8888(rb);
+         break;
+      case I915_TILING_X:
+         intel_XTile_InitPointers_xRGB8888(rb);
+         break;
+      case I915_TILING_Y:
+         intel_YTile_InitPointers_xRGB8888(rb);
+         break;
+      }
+      break;
    case MESA_FORMAT_ARGB8888:
-      if (rb->AlphaBits == 0) { /* XXX: Need xRGB8888 Mesa format */
-	 /* 8888 RGBx */
-	 switch (tiling) {
-	 case I915_TILING_NONE:
-	 default:
-	    intelInitPointers_xRGB8888(rb);
-	    break;
-	 case I915_TILING_X:
-	    intel_XTile_InitPointers_xRGB8888(rb);
-	    break;
-	 case I915_TILING_Y:
-	    intel_YTile_InitPointers_xRGB8888(rb);
-	    break;
-	 }
-      } else {
-	 /* 8888 RGBA */
-	 switch (tiling) {
-	 case I915_TILING_NONE:
-	 default:
-	    intelInitPointers_ARGB8888(rb);
-	    break;
-	 case I915_TILING_X:
-	    intel_XTile_InitPointers_ARGB8888(rb);
-	    break;
-	 case I915_TILING_Y:
-	    intel_YTile_InitPointers_ARGB8888(rb);
-	    break;
-	 }
+      /* 8888 RGBA */
+      switch (tiling) {
+      case I915_TILING_NONE:
+      default:
+	 intelInitPointers_ARGB8888(rb);
+	 break;
+      case I915_TILING_X:
+	 intel_XTile_InitPointers_ARGB8888(rb);
+	 break;
+      case I915_TILING_Y:
+	 intel_YTile_InitPointers_ARGB8888(rb);
+	 break;
       }
       break;
    case MESA_FORMAT_Z16:
@@ -703,39 +745,27 @@ intel_set_span_functions(struct intel_context *intel,
 	 break;
       }
       break;
+   case MESA_FORMAT_X8_Z24:
    case MESA_FORMAT_S8_Z24:
       /* There are a few different ways SW asks us to access the S8Z24 data:
        * Z24 depth-only depth reads
        * S8Z24 depth reads
        * S8Z24 stencil reads.
        */
-      if (rb->_ActualFormat == GL_DEPTH_COMPONENT24) {
-	 switch (tiling) {
-	 case I915_TILING_NONE:
-	 default:
-	    intelInitDepthPointers_z24(rb);
-	    break;
-	 case I915_TILING_X:
-	    intel_XTile_InitDepthPointers_z24(rb);
-	    break;
-	 case I915_TILING_Y:
-	    intel_YTile_InitDepthPointers_z24(rb);
-	    break;
-	 }
-      } else if (rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
+      if (rb->Format == MESA_FORMAT_S8_Z24) {
 	 switch (tiling) {
 	 case I915_TILING_NONE:
 	 default:
-	    intelInitDepthPointers_z24_s8(rb);
+	    intelInitDepthPointers_z24_x8(rb);
 	    break;
 	 case I915_TILING_X:
-	    intel_XTile_InitDepthPointers_z24_s8(rb);
+	    intel_XTile_InitDepthPointers_z24_x8(rb);
 	    break;
 	 case I915_TILING_Y:
-	    intel_YTile_InitDepthPointers_z24_s8(rb);
+	    intel_YTile_InitDepthPointers_z24_x8(rb);
 	    break;
 	 }
-      } else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) {
+      } else if (rb->Format == MESA_FORMAT_S8) {
 	 switch (tiling) {
 	 case I915_TILING_NONE:
 	 default:
diff --git a/src/mesa/drivers/dri/intel/intel_spantmp.h b/src/mesa/drivers/dri/intel/intel_spantmp.h
index ead0b1c168..bad03398f6 100644
--- a/src/mesa/drivers/dri/intel/intel_spantmp.h
+++ b/src/mesa/drivers/dri/intel/intel_spantmp.h
@@ -32,6 +32,12 @@
 
 #define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
 #define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
+#define TAG(x) INTEL_TAG(intel_gttmap_##x)
+#define TAG2(x, y) INTEL_TAG(intel_gttmap_##x##y)
+#include "spantmp2.h"
+
+#define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
+#define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
 #define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(NO_TILE(_x, _y), v)
 #define GET_VALUE(_x, _y) INTEL_READ_VALUE(NO_TILE(_x, _y))
 #define TAG(x) INTEL_TAG(intel##x)
@@ -48,8 +54,8 @@
 
 #define SPANTMP_PIXEL_FMT INTEL_PIXEL_FMT
 #define SPANTMP_PIXEL_TYPE INTEL_PIXEL_TYPE
-#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(X_TILE(_x, _y), v)
-#define GET_VALUE(_x, _y) INTEL_READ_VALUE(X_TILE(_x, _y))
+#define PUT_VALUE(_x, _y, v) INTEL_WRITE_VALUE(Y_TILE(_x, _y), v)
+#define GET_VALUE(_x, _y) INTEL_READ_VALUE(Y_TILE(_x, _y))
 #define TAG(x) INTEL_TAG(intel_YTile_##x)
 #define TAG2(x, y) INTEL_TAG(intel_YTile_##x)##y
 #include "spantmp2.h"
diff --git a/src/mesa/drivers/dri/intel/intel_state.c b/src/mesa/drivers/dri/intel/intel_state.c
index 4ee742377d..aefae53eb2 100644
--- a/src/mesa/drivers/dri/intel/intel_state.c
+++ b/src/mesa/drivers/dri/intel/intel_state.c
@@ -196,25 +196,6 @@ intel_translate_logic_op(GLenum opcode)
    }
 }
 
-
-static void
-intelClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLubyte clear[4];
-
-   CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
-
-   /* compute both 32 and 16-bit clear values */
-   intel->ClearColor8888 = INTEL_PACKCOLOR8888(clear[0], clear[1],
-                                               clear[2], clear[3]);
-   intel->ClearColor565 = INTEL_PACKCOLOR565(clear[0], clear[1], clear[2]);
-}
-
-
 /* Fallback to swrast for select and feedback.
  */
 static void
@@ -229,5 +210,4 @@ void
 intelInitStateFuncs(struct dd_function_table *functions)
 {
    functions->RenderMode = intelRenderMode;
-   functions->ClearColor = intelClearColor;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_swapbuffers.c b/src/mesa/drivers/dri/intel/intel_swapbuffers.c
deleted file mode 100644
index 7d035b9f6e..0000000000
--- a/src/mesa/drivers/dri/intel/intel_swapbuffers.c
+++ /dev/null
@@ -1,248 +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 "intel_blit.h"
-#include "intel_buffers.h"
-#include "intel_swapbuffers.h"
-#include "intel_fbo.h"
-#include "intel_batchbuffer.h"
-#include "drirenderbuffer.h"
-#include "vblank.h"
-#include "i915_drm.h"
-
-
-
-/*
- * Correct a drawablePrivate's set of vblank flags WRT the current context.
- * When considering multiple crtcs.
- */
-GLuint
-intelFixupVblank(struct intel_context *intel, __DRIdrawablePrivate *dPriv)
-{
-   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
-       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
-      volatile drm_i915_sarea_t *sarea = intel->sarea;
-      drm_clip_rect_t drw_rect = { .x1 = dPriv->x, .x2 = dPriv->x + dPriv->w,
-				   .y1 = dPriv->y, .y2 = dPriv->y + dPriv->h };
-      drm_clip_rect_t planeA_rect = { .x1 = sarea->planeA_x, .y1 = sarea->planeA_y,
-				     .x2 = sarea->planeA_x + sarea->planeA_w,
-				     .y2 = sarea->planeA_y + sarea->planeA_h };
-      drm_clip_rect_t planeB_rect = { .x1 = sarea->planeB_x, .y1 = sarea->planeB_y,
-				     .x2 = sarea->planeB_x + sarea->planeB_w,
-				     .y2 = sarea->planeB_y + sarea->planeB_h };
-      GLint areaA = driIntersectArea( drw_rect, planeA_rect );
-      GLint areaB = driIntersectArea( drw_rect, planeB_rect );
-      GLuint flags = dPriv->vblFlags;
-
-      /* Update vblank info
-       */
-      if (areaB > areaA || (areaA == areaB && areaB > 0)) {
-	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
-      } else {
-	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
-      }
-
-      /* Do the stupid test: Is one of them actually disabled?
-       */
-      if (sarea->planeA_w == 0 || sarea->planeA_h == 0) {
-	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
-      } else if (sarea->planeB_w == 0 || sarea->planeB_h == 0) {
-	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
-      }
-
-      return flags;
-   } else {
-      return dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
-   }
-}
-
-
-/**
- * Called from driSwapBuffers()
- */
-void
-intelSwapBuffers(__DRIdrawablePrivate * dPriv)
-{
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
-
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      GET_CURRENT_CONTEXT(ctx);
-      struct intel_context *intel;
-
-      if (ctx == NULL)
-	 return;
-
-      intel = intel_context(ctx);
-
-      if (ctx->Visual.doubleBufferMode) {
-	 GLboolean missed_target;
-	 struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-	 int64_t ust;
-         
-	 _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
-
-	/*
-	 * The old swapping ioctl was incredibly racy, just wait for vblank
-	 * and do the swap ourselves.
-	 */
-	 driWaitForVBlank(dPriv, &missed_target);
-
-	 /*
-	  * Update each buffer's vbl_pending so we don't get too out of
-	  * sync
-	  */
-	 intel_get_renderbuffer(&intel_fb->Base,
-		   		BUFFER_BACK_LEFT)->vbl_pending = dPriv->vblSeq;
-         intel_get_renderbuffer(&intel_fb->Base,
-		   		BUFFER_FRONT_LEFT)->vbl_pending = dPriv->vblSeq;
-
-	 intelCopyBuffer(dPriv, NULL);
-
-	 intel_fb->swap_count++;
-	 (*psp->systemTime->getUST) (&ust);
-	 if (missed_target) {
-	    intel_fb->swap_missed_count++;
-	    intel_fb->swap_missed_ust = ust - intel_fb->swap_ust;
-	 }
-
-	 intel_fb->swap_ust = ust;
-      }
-      drmCommandNone(intel->driFd, DRM_I915_GEM_THROTTLE);
-   }
-   else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-
-/**
- * Called from driCopySubBuffer()
- */
-void
-intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h)
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      struct intel_context *intel =
-         (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-      GLcontext *ctx = &intel->ctx;
-
-      if (ctx->Visual.doubleBufferMode) {
-         drm_clip_rect_t rect;
-         rect.x1 = x + dPriv->x;
-         rect.y1 = (dPriv->h - y - h) + dPriv->y;
-         rect.x2 = rect.x1 + w;
-         rect.y2 = rect.y1 + h;
-         _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
-         intelCopyBuffer(dPriv, &rect);
-      }
-   }
-   else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-
-/**
- * This will be called whenever the currently bound window is moved/resized.
- * XXX: actually, it seems to NOT be called when the window is only moved (BP).
- */
-void
-intelWindowMoved(struct intel_context *intel)
-{
-   GLcontext *ctx = &intel->ctx;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-
-   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
-       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
-      GLuint flags = intelFixupVblank(intel, dPriv);
-
-      /* Check to see if we changed pipes */
-      if (flags != dPriv->vblFlags && dPriv->vblFlags &&
-	  !(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ)) {
-	 int64_t count;
-	 drmVBlank vbl;
-	 int i;
-
-	 /*
-	  * Deal with page flipping
-	  */
-	 vbl.request.type = DRM_VBLANK_ABSOLUTE;
-
-	 if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
-	    vbl.request.type |= DRM_VBLANK_SECONDARY;
-	 }
-
-	 for (i = 0; i < 2; i++) {
-	    if (!intel_fb->color_rb[i] ||
-		(intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <=
-		(1<<23))
-	       continue;
-
-	    vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending;
-	    drmWaitVBlank(intel->driFd, &vbl);
-	 }
-
-	 /*
-	  * Update msc_base from old pipe
-	  */
-	 driDrawableGetMSC32(dPriv->driScreenPriv, dPriv, &count);
-	 dPriv->msc_base = count;
-	 /*
-	  * Then get new vblank_base and vblSeq values
-	  */
-	 dPriv->vblFlags = flags;
-	 driGetCurrentVBlank(dPriv);
-	 dPriv->vblank_base = dPriv->vblSeq;
-
-	 intel_fb->vbl_waited = dPriv->vblSeq;
-
-	 for (i = 0; i < 2; i++) {
-	    if (intel_fb->color_rb[i])
-	       intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_waited;
-	 }
-      }
-   } else {
-      dPriv->vblFlags &= ~VBLANK_FLAG_SECONDARY;
-   }
-
-   /* Update Mesa's notion of window size */
-   driUpdateFramebufferSize(ctx, dPriv);
-   intel_fb->Base.Initialized = GL_TRUE; /* XXX remove someday */
-
-   /* Update hardware scissor */
-   if (ctx->Driver.Scissor != NULL) {
-      ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			  ctx->Scissor.Width, ctx->Scissor.Height);
-   }
-
-   /* Re-calculate viewport related state */
-   if (ctx->Driver.DepthRange != NULL)
-      ctx->Driver.DepthRange( ctx, ctx->Viewport.Near, ctx->Viewport.Far );
-}
diff --git a/src/mesa/drivers/dri/intel/intel_swapbuffers.h b/src/mesa/drivers/dri/intel/intel_swapbuffers.h
deleted file mode 100644
index 75bb6242ff..0000000000
--- a/src/mesa/drivers/dri/intel/intel_swapbuffers.h
+++ /dev/null
@@ -1,52 +0,0 @@
-
-/**************************************************************************
- * 
- * Copyright 2006 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 INTEL_SWAPBUFFERS_H
-#define INTEL_SWAPBUFFERS_H
-
-#include "dri_util.h"
-#include "drm.h"
-
-struct intel_context;
-struct intel_framebuffer;
-
-
-extern void
-intelSwapBuffers(__DRIdrawablePrivate * dPriv);
-
-extern void
-intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h);
-
-extern GLuint
-intelFixupVblank(struct intel_context *intel, __DRIdrawablePrivate *dPriv);
-
-extern void
-intelWindowMoved(struct intel_context *intel);
-
-
-#endif /* INTEL_SWAPBUFFERS_H */
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 3cbc379dbd..215a534a5c 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.c
+++ b/src/mesa/drivers/dri/intel/intel_tex.c
@@ -177,6 +177,28 @@ intelGenerateMipmap(GLcontext *ctx, GLenum target,
       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);
diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h
index 57ed0b1aab..f3cc0fff5c 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.h
+++ b/src/mesa/drivers/dri/intel/intel_tex.h
@@ -29,6 +29,7 @@
 #define INTELTEX_INC
 
 #include "main/mtypes.h"
+#include "main/formats.h"
 #include "intel_context.h"
 #include "texmem.h"
 
@@ -41,10 +42,8 @@ void intelInitTextureSubImageFuncs(struct dd_function_table *functions);
 
 void intelInitTextureCopyImageFuncs(struct dd_function_table *functions);
 
-const struct gl_texture_format *intelChooseTextureFormat(GLcontext * ctx,
-                                                         GLint internalFormat,
-                                                         GLenum format,
-                                                         GLenum type);
+gl_format intelChooseTextureFormat(GLcontext *ctx, GLint internalFormat,
+                                   GLenum format, GLenum type);
 
 void intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
 		       unsigned long long offset, GLint depth, GLuint pitch);
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index f3e312420d..d8e71093c4 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -29,6 +29,7 @@
 #include "main/enums.h"
 #include "main/image.h"
 #include "main/teximage.h"
+#include "main/texstate.h"
 #include "main/mipmap.h"
 
 #include "drivers/common/meta.h"
@@ -92,8 +93,7 @@ do_copy_texsubimage(struct intel_context *intel,
                     GLint x, GLint y, GLsizei width, GLsizei height)
 {
    GLcontext *ctx = &intel->ctx;
-   const struct intel_region *src =
-      get_teximage_source(intel, internalFormat);
+   const struct intel_region *src = get_teximage_source(intel, internalFormat);
 
    if (!intelImage->mt || !src) {
       if (INTEL_DEBUG & DEBUG_FALLBACKS)
@@ -109,33 +109,34 @@ do_copy_texsubimage(struct intel_context *intel,
       return GL_FALSE;
    }
 
-   intelFlush(ctx);
-   LOCK_HARDWARE(intel);
+   /* intelFlush(ctx); */
    {
       drm_intel_bo *dst_bo = intel_region_buffer(intel,
 						 intelImage->mt->region,
 						 INTEL_WRITE_PART);
-      GLuint image_offset = intel_miptree_image_offset(intelImage->mt,
-                                                       intelImage->face,
-                                                       intelImage->level);
       const GLint orig_x = x;
       const GLint orig_y = y;
+      GLuint image_x, image_y;
       GLshort src_pitch;
 
+      /* get dest x/y in destination texture */
+      intel_miptree_get_image_offset(intelImage->mt,
+				     intelImage->level,
+				     intelImage->face,
+				     0,
+				     &image_x, &image_y);
       /* Update dst for clipped src.  Need to also clip the source rect. */
       dstx += x - orig_x;
       dsty += y - orig_y;
 
       /* Can't blit to tiled buffers with non-tile-aligned offset. */
-      if (intelImage->mt->region->tiling != I915_TILING_NONE &&
-	  (image_offset & 4095) != 0) {
-	 UNLOCK_HARDWARE(intel);
+      if (intelImage->mt->region->tiling == I915_TILING_Y) {
 	 return GL_FALSE;
       }
 
       if (ctx->ReadBuffer->Name == 0) {
 	 /* reading from a window, adjust x, y */
-	 const __DRIdrawablePrivate *dPriv = intel->driReadDrawable;
+	 const __DRIdrawable *dPriv = intel->driReadDrawable;
 	 y = dPriv->y + (dPriv->h - (y + height));
 	 x += dPriv->x;
 
@@ -152,6 +153,7 @@ do_copy_texsubimage(struct intel_context *intel,
 	 src_pitch = src->pitch;
       }
 
+      /* blit from src buffer to texture */
       if (!intelEmitCopyBlit(intel,
 			     intelImage->mt->cpp,
 			     src_pitch,
@@ -160,17 +162,16 @@ do_copy_texsubimage(struct intel_context *intel,
 			     src->tiling,
 			     intelImage->mt->pitch,
 			     dst_bo,
-			     image_offset,
+			     0,
 			     intelImage->mt->region->tiling,
-			     x, y, dstx, dsty, width, height,
+			     src->draw_x + x, src->draw_y + y,
+			     image_x + dstx, image_y + dsty,
+			     width, height,
 			     GL_COPY)) {
-	 UNLOCK_HARDWARE(intel);
 	 return GL_FALSE;
       }
    }
 
-   UNLOCK_HARDWARE(intel);
-
    return GL_TRUE;
 }
 
@@ -180,8 +181,7 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
                     GLenum internalFormat,
                     GLint x, GLint y, GLsizei width, GLint border)
 {
-   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 =
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
@@ -217,6 +217,8 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y,
                              width, border);
 }
@@ -228,8 +230,7 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
                     GLint x, GLint y, GLsizei width, GLsizei height,
                     GLint border)
 {
-   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 =
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
@@ -244,7 +245,7 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
     */
    ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
                           width, height, border,
-                          GL_RGBA, CHAN_TYPE, NULL,
+                          GL_RGBA, GL_UNSIGNED_BYTE, NULL,
                           &ctx->DefaultPacking, texObj, texImage);
 
    srcx = x;
@@ -265,6 +266,8 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
    return;
 
  fail:
+   if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
                              width, height, border);
 }
@@ -274,8 +277,7 @@ static void
 intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
                        GLint xoffset, GLint x, GLint y, GLsizei width)
 {
-   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 =
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
@@ -290,6 +292,8 @@ 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)) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width);
    }
 }
@@ -300,8 +304,7 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
                        GLint xoffset, GLint yoffset,
                        GLint x, GLint y, GLsizei width, GLsizei height)
 {
-   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 =
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
@@ -316,8 +319,8 @@ intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
                             internalFormat,
                             xoffset, yoffset, x, y, width, height)) {
 
-      DBG("%s - fallback to _mesa_meta_CopyTexSubImage2D\n", __FUNCTION__);
-
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage2D(ctx, target, level,
                                    xoffset, yoffset, x, y, width, height);
    }
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
index 3322a71130..a7c6c45ffe 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_format.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -1,7 +1,6 @@
 #include "intel_context.h"
 #include "intel_tex.h"
 #include "intel_chipset.h"
-#include "main/texformat.h"
 #include "main/enums.h"
 
 
@@ -16,7 +15,7 @@
  * these if we take the step of simply swizzling the colors
  * immediately after sampling...
  */
-const struct gl_texture_format *
+gl_format
 intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
                          GLenum format, GLenum type)
 {
@@ -34,48 +33,48 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_COMPRESSED_RGBA:
       if (format == GL_BGRA) {
          if (type == GL_UNSIGNED_BYTE || type == GL_UNSIGNED_INT_8_8_8_8_REV) {
-            return &_mesa_texformat_argb8888;
+            return MESA_FORMAT_ARGB8888;
          }
          else if (type == GL_UNSIGNED_SHORT_4_4_4_4_REV) {
-            return &_mesa_texformat_argb4444;
+            return MESA_FORMAT_ARGB4444;
          }
          else if (type == GL_UNSIGNED_SHORT_1_5_5_5_REV) {
-            return &_mesa_texformat_argb1555;
+            return MESA_FORMAT_ARGB1555;
          }
       }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case 3:
    case GL_RGB:
    case GL_COMPRESSED_RGB:
       if (format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5) {
-         return &_mesa_texformat_rgb565;
+         return MESA_FORMAT_RGB565;
       }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+      return do32bpt ? MESA_FORMAT_XRGB8888 : MESA_FORMAT_RGB565;
 
    case GL_RGBA8:
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case GL_RGBA4:
    case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
 
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return &_mesa_texformat_argb8888;
+      return MESA_FORMAT_XRGB8888;
 
    case GL_RGB5:
    case GL_RGB4:
    case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_ALPHA:
    case GL_ALPHA4:
@@ -83,7 +82,7 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_ALPHA12:
    case GL_ALPHA16:
    case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
+      return MESA_FORMAT_A8;
 
    case 1:
    case GL_LUMINANCE:
@@ -92,18 +91,24 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
+      return MESA_FORMAT_L8;
+
+   case GL_LUMINANCE12_ALPHA4:
+   case GL_LUMINANCE12_ALPHA12:
+   case GL_LUMINANCE16_ALPHA16:
+#ifndef I915
+      return MESA_FORMAT_AL1616;
+#else
+      /* FALLTHROUGH */
+#endif
 
    case 2:
    case GL_LUMINANCE_ALPHA:
    case GL_LUMINANCE4_ALPHA4:
    case GL_LUMINANCE6_ALPHA2:
    case GL_LUMINANCE8_ALPHA8:
-   case GL_LUMINANCE12_ALPHA4:
-   case GL_LUMINANCE12_ALPHA12:
-   case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
+      return MESA_FORMAT_AL88;
 
    case GL_INTENSITY:
    case GL_INTENSITY4:
@@ -111,41 +116,41 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_INTENSITY12:
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
+      return MESA_FORMAT_I8;
 
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_MESA || type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    case GL_COMPRESSED_RGB_FXT1_3DFX:
-      return &_mesa_texformat_rgb_fxt1;
+      return MESA_FORMAT_RGB_FXT1;
    case GL_COMPRESSED_RGBA_FXT1_3DFX:
-      return &_mesa_texformat_rgba_fxt1;
+      return MESA_FORMAT_RGBA_FXT1;
 
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
    case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgb_dxt1;
+      return MESA_FORMAT_RGB_DXT1;
 
    case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgba_dxt1;
+      return MESA_FORMAT_RGBA_DXT1;
 
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-      return &_mesa_texformat_rgba_dxt3;
+      return MESA_FORMAT_RGBA_DXT3;
 
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
+      return MESA_FORMAT_RGBA_DXT5;
 
    case GL_DEPTH_COMPONENT:
    case GL_DEPTH_COMPONENT16:
    case GL_DEPTH_COMPONENT24:
    case GL_DEPTH_COMPONENT32:
 #if 0
-      return &_mesa_texformat_z16;
+      return MESA_FORMAT_Z16;
 #else
       /* fall-through.
        * 16bpp depth texture can't be paired with a stencil buffer so
@@ -154,7 +159,7 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
 #endif
    case GL_DEPTH_STENCIL_EXT:
    case GL_DEPTH24_STENCIL8_EXT:
-      return &_mesa_texformat_s8_z24;
+      return MESA_FORMAT_S8_Z24;
 
 #ifndef I915
    case GL_SRGB_EXT:
@@ -165,41 +170,41 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_COMPRESSED_SRGB_ALPHA_EXT:
    case GL_COMPRESSED_SLUMINANCE_EXT:
    case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
-      return &_mesa_texformat_sargb8;
+      return MESA_FORMAT_SARGB8;
    case GL_SLUMINANCE_EXT:
    case GL_SLUMINANCE8_EXT:
-      if (IS_G4X(intel->intelScreen->deviceID))
-         return &_mesa_texformat_sl8;
+      if (intel->has_luminance_srgb)
+         return MESA_FORMAT_SL8;
       else
-         return &_mesa_texformat_sargb8;
+         return MESA_FORMAT_SARGB8;
    case GL_SLUMINANCE_ALPHA_EXT:
    case GL_SLUMINANCE8_ALPHA8_EXT:
-      if (IS_G4X(intel->intelScreen->deviceID))
-         return &_mesa_texformat_sla8;
+      if (intel->has_luminance_srgb)
+         return MESA_FORMAT_SLA8;
       else
-         return &_mesa_texformat_sargb8;
+         return MESA_FORMAT_SARGB8;
    case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_srgb_dxt1;
+      return MESA_FORMAT_SRGB_DXT1;
 
    /* i915 could also do this */
    case GL_DUDV_ATI:
    case GL_DU8DV8_ATI:
-      return &_mesa_texformat_dudv8;
+      return MESA_FORMAT_DUDV8;
    case GL_RGBA_SNORM:
    case GL_RGBA8_SNORM:
-      return &_mesa_texformat_signed_rgba8888_rev;
+      return MESA_FORMAT_SIGNED_RGBA8888_REV;
 #endif
 
    default:
       fprintf(stderr, "unexpected texture format %s in %s\n",
               _mesa_lookup_enum_by_nr(internalFormat), __FUNCTION__);
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 
-   return NULL;                 /* never get here */
+   return MESA_FORMAT_NONE;       /* never get here */
 }
 
 int intel_compressed_num_bytes(GLuint mesaFormat)
diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c
index 66201b1f46..307669f87e 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_image.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_image.c
@@ -6,8 +6,10 @@
 #include "main/bufferobj.h"
 #include "main/convolve.h"
 #include "main/context.h"
+#include "main/formats.h"
+#include "main/image.h"
 #include "main/texcompress.h"
-#include "main/texformat.h"
+#include "main/texstore.h"
 #include "main/texgetimage.h"
 #include "main/texobj.h"
 #include "main/texstore.h"
@@ -69,6 +71,7 @@ guess_and_alloc_mipmap_tree(struct intel_context *intel,
    GLuint depth = intelImage->base.Depth;
    GLuint l2width, l2height, l2depth;
    GLuint i, comp_byte = 0;
+   GLuint texelBytes;
 
    DBG("%s\n", __FUNCTION__);
 
@@ -112,7 +115,8 @@ guess_and_alloc_mipmap_tree(struct intel_context *intel,
     */
    if ((intelObj->base.MinFilter == GL_NEAREST ||
         intelObj->base.MinFilter == GL_LINEAR) &&
-       intelImage->level == firstLevel) {
+       intelImage->level == firstLevel &&
+       (intel->gen < 4 || firstLevel == 0)) {
       lastLevel = firstLevel;
    }
    else {
@@ -123,8 +127,11 @@ guess_and_alloc_mipmap_tree(struct intel_context *intel,
    }
 
    assert(!intelObj->mt);
-   if (intelImage->base.IsCompressed)
-      comp_byte = intel_compressed_num_bytes(intelImage->base.TexFormat->MesaFormat);
+   if (_mesa_is_format_compressed(intelImage->base.TexFormat))
+      comp_byte = intel_compressed_num_bytes(intelImage->base.TexFormat);
+
+   texelBytes = _mesa_get_format_bytes(intelImage->base.TexFormat);
+
    intelObj->mt = intel_miptree_create(intel,
                                        intelObj->base.Target,
                                        intelImage->base._BaseFormat,
@@ -134,7 +141,7 @@ guess_and_alloc_mipmap_tree(struct intel_context *intel,
                                        width,
                                        height,
                                        depth,
-                                       intelImage->base.TexFormat->TexelBytes,
+                                       texelBytes,
                                        comp_byte,
 				       expect_accelerated_upload);
 
@@ -166,7 +173,7 @@ target_to_face(GLenum target)
 static GLboolean
 check_pbo_format(GLint internalFormat,
                  GLenum format, GLenum type,
-                 const struct gl_texture_format *mesa_format)
+                 gl_format mesa_format)
 {
    switch (internalFormat) {
    case 4:
@@ -174,12 +181,12 @@ check_pbo_format(GLint internalFormat,
       return (format == GL_BGRA &&
               (type == GL_UNSIGNED_BYTE ||
                type == GL_UNSIGNED_INT_8_8_8_8_REV) &&
-              mesa_format == &_mesa_texformat_argb8888);
+              mesa_format == MESA_FORMAT_ARGB8888);
    case 3:
    case GL_RGB:
       return (format == GL_RGB &&
               type == GL_UNSIGNED_SHORT_5_6_5 &&
-              mesa_format == &_mesa_texformat_rgb565);
+              mesa_format == MESA_FORMAT_RGB565);
    case GL_YCBCR_MESA:
       return (type == GL_UNSIGNED_SHORT_8_8_MESA || type == GL_UNSIGNED_BYTE);
    default:
@@ -200,7 +207,10 @@ 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;
+   GLuint dst_x, dst_y, dst_stride;
+   dri_bo *dst_buffer = intel_region_buffer(intel,
+					    intelImage->mt->region,
+					    INTEL_WRITE_FULL);
 
    if (!_mesa_is_bufferobj(unpack->BufferObj) ||
        intel->ctx._ImageTransferState ||
@@ -217,32 +227,26 @@ try_pbo_upload(struct intel_context *intel,
    else
       src_stride = width;
 
-   dst_offset = intel_miptree_image_offset(intelImage->mt,
-                                           intelImage->face,
-                                           intelImage->level);
+   intel_miptree_get_image_offset(intelImage->mt, intelImage->level,
+				  intelImage->face, 0,
+				  &dst_x, &dst_y);
 
    dst_stride = intelImage->mt->pitch;
 
-   intelFlush(&intel->ctx);
-   LOCK_HARDWARE(intel);
+   if (drm_intel_bo_references(intel->batch->buf, dst_buffer))
+      intelFlush(&intel->ctx);
    {
       dri_bo *src_buffer = intel_bufferobj_buffer(intel, pbo, INTEL_READ);
-      dri_bo *dst_buffer = intel_region_buffer(intel,
-					       intelImage->mt->region,
-					       INTEL_WRITE_FULL);
-
 
       if (!intelEmitCopyBlit(intel,
 			     intelImage->mt->cpp,
 			     src_stride, src_buffer, src_offset, GL_FALSE,
-			     dst_stride, dst_buffer, dst_offset, GL_FALSE,
-			     0, 0, 0, 0, width, height,
+			     dst_stride, dst_buffer, 0, GL_FALSE,
+			     0, 0, dst_x, dst_y, width, height,
 			     GL_COPY)) {
-	 UNLOCK_HARDWARE(intel);
 	 return GL_FALSE;
       }
    }
-   UNLOCK_HARDWARE(intel);
 
    return GL_TRUE;
 }
@@ -258,7 +262,7 @@ try_pbo_zcopy(struct intel_context *intel,
 {
    struct intel_buffer_object *pbo = intel_buffer_object(unpack->BufferObj);
    GLuint src_offset, src_stride;
-   GLuint dst_offset, dst_stride;
+   GLuint dst_x, dst_y, dst_stride;
 
    if (!_mesa_is_bufferobj(unpack->BufferObj) ||
        intel->ctx._ImageTransferState ||
@@ -275,13 +279,14 @@ try_pbo_zcopy(struct intel_context *intel,
    else
       src_stride = width;
 
-   dst_offset = intel_miptree_image_offset(intelImage->mt,
-                                           intelImage->face,
-                                           intelImage->level);
+   intel_miptree_get_image_offset(intelImage->mt, intelImage->level,
+				  intelImage->face, 0,
+				  &dst_x, &dst_y);
 
    dst_stride = intelImage->mt->pitch;
 
-   if (src_stride != dst_stride || dst_offset != 0 || src_offset != 0) {
+   if (src_stride != dst_stride || dst_x != 0 || dst_y != 0 ||
+       src_offset != 0) {
       DBG("%s: failure 2\n", __FUNCTION__);
       return GL_FALSE;
    }
@@ -316,8 +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;
 
@@ -326,22 +329,11 @@ intelTexImage(GLcontext * ctx,
                                          &postConvHeight);
    }
 
-   /* choose the texture format */
-   texImage->TexFormat = intelChooseTextureFormat(ctx, internalFormat,
-                                                  format, type);
-
-   _mesa_set_fetch_functions(texImage, dims);
-
-   if (texImage->TexFormat->TexelBytes == 0) {
-      /* must be a compressed format */
+   if (_mesa_is_format_compressed(texImage->TexFormat)) {
       texelBytes = 0;
-      texImage->IsCompressed = GL_TRUE;
-      texImage->CompressedSize =
-	 ctx->Driver.CompressedTextureSize(ctx, texImage->Width,
-					   texImage->Height, texImage->Depth,
-					   texImage->TexFormat->MesaFormat);
-   } else {
-      texelBytes = texImage->TexFormat->TexelBytes;
+   }
+   else {
+      texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
       
       /* Minimum pitch of 32 bytes */
       if (postConvWidth * texelBytes < 32) {
@@ -374,8 +366,7 @@ intelTexImage(GLcontext * ctx,
        intelObj->mt->first_level == level &&
        intelObj->mt->last_level == level &&
        intelObj->mt->target != GL_TEXTURE_CUBE_MAP_ARB &&
-       !intel_miptree_match_image(intelObj->mt, &intelImage->base,
-                                  intelImage->face, intelImage->level)) {
+       !intel_miptree_match_image(intelObj->mt, &intelImage->base)) {
 
       DBG("release it\n");
       intel_miptree_release(intel, &intelObj->mt);
@@ -392,17 +383,17 @@ intelTexImage(GLcontext * ctx,
    assert(!intelImage->mt);
 
    if (intelObj->mt &&
-       intel_miptree_match_image(intelObj->mt, &intelImage->base,
-                                 intelImage->face, intelImage->level)) {
+       intel_miptree_match_image(intelObj->mt, &intelImage->base)) {
 
       intel_miptree_reference(&intelImage->mt, intelObj->mt);
       assert(intelImage->mt);
    } else if (intelImage->base.Border == 0) {
       int comp_byte = 0;
-
-      if (intelImage->base.IsCompressed) {
+      GLuint texelBytes = _mesa_get_format_bytes(intelImage->base.TexFormat);
+      GLenum baseFormat = _mesa_get_format_base_format(intelImage->base.TexFormat);
+      if (_mesa_is_format_compressed(intelImage->base.TexFormat)) {
 	 comp_byte =
-	    intel_compressed_num_bytes(intelImage->base.TexFormat->MesaFormat);
+	    intel_compressed_num_bytes(intelImage->base.TexFormat);
       }
 
       /* Didn't fit in the object miptree, but it's suitable for inclusion in
@@ -410,11 +401,11 @@ intelTexImage(GLcontext * ctx,
        * It'll get moved into the object miptree at validate time.
        */
       intelImage->mt = intel_miptree_create(intel, target,
-					    intelImage->base.TexFormat->BaseFormat,
+					    baseFormat,
 					    internalFormat,
 					    level, level,
 					    width, height, depth,
-					    intelImage->base.TexFormat->TexelBytes,
+					    texelBytes,
 					    comp_byte, pixels == NULL);
 
    }
@@ -475,24 +466,32 @@ intelTexImage(GLcontext * ctx,
 					   pixels, unpack, "glTexImage");
    }
 
-   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 {
       /* Allocate regular memory and store the image there temporarily.   */
-      if (texImage->IsCompressed) {
-         sizeInBytes = texImage->CompressedSize;
+      if (_mesa_is_format_compressed(texImage->TexFormat)) {
+         sizeInBytes = _mesa_format_image_size(texImage->TexFormat,
+                                               texImage->Width,
+                                               texImage->Height,
+                                               texImage->Depth);
          dstRowStride =
-            _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
+            _mesa_format_row_stride(texImage->TexFormat, width);
          assert(dims != 3);
       }
       else {
@@ -523,17 +522,20 @@ intelTexImage(GLcontext * ctx,
 			       pixels,
 			       srcRowStride,
 			       0, 0);
-	   } else
+	   }
+           else {
 	       memcpy(texImage->Data, pixels, imageSize);
-       } else if (!texImage->TexFormat->StoreImage(ctx, dims, 
-						   texImage->_BaseFormat, 
-						   texImage->TexFormat, 
-						   texImage->Data, 0, 0, 0, /* dstX/Y/Zoffset */
-						   dstRowStride,
-						   texImage->ImageOffsets,
-						   width, height, depth,
-						   format, type, pixels, unpack)) {
-	   _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
+           }
+       }
+       else if (!_mesa_texstore(ctx, dims, 
+                                texImage->_BaseFormat, 
+                                texImage->TexFormat, 
+                                texImage->Data, 0, 0, 0, /* dstX/Y/Zoffset */
+                                dstRowStride,
+                                texImage->ImageOffsets,
+                                width, height, depth,
+                                format, type, pixels, unpack)) {
+          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
        }
    }
 
@@ -544,8 +546,6 @@ intelTexImage(GLcontext * ctx,
          intel_miptree_image_unmap(intel, intelImage->mt);
       texImage->Data = NULL;
    }
-
-   UNLOCK_HARDWARE(intel);
 }
 
 
@@ -661,9 +661,10 @@ intel_get_tex_image(GLcontext * ctx, GLenum target, GLint level,
    if (compressed) {
       _mesa_get_compressed_teximage(ctx, target, level, pixels,
 				    texObj, texImage);
-   } else {
+   }
+   else {
       _mesa_get_teximage(ctx, target, level, format, type, pixels,
-			 texObj, texImage);
+                         texObj, texImage);
    }
      
 
@@ -724,19 +725,18 @@ intelSetTexBuffer2(__DRIcontext *pDRICtx, GLint target,
 		   GLint glx_texture_format,
 		   __DRIdrawable *dPriv)
 {
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+   struct gl_framebuffer *fb = dPriv->driverPrivate;
    struct intel_context *intel = pDRICtx->driverPrivate;
+   GLcontext *ctx = &intel->ctx;
    struct intel_texture_object *intelObj;
    struct intel_texture_image *intelImage;
    struct intel_mipmap_tree *mt;
    struct intel_renderbuffer *rb;
-   struct gl_texture_unit *texUnit;
    struct gl_texture_object *texObj;
    struct gl_texture_image *texImage;
-   int level = 0, type, format, internalFormat;
+   int level = 0, internalFormat;
 
-   texUnit = &intel->ctx.Texture.Unit[intel->ctx.Texture.CurrentUnit];
-   texObj = _mesa_select_tex_object(&intel->ctx, texUnit, target);
+   texObj = _mesa_get_current_tex_object(ctx, target);
    intelObj = intel_texture_object(texObj);
 
    if (!intelObj)
@@ -744,15 +744,13 @@ intelSetTexBuffer2(__DRIcontext *pDRICtx, GLint target,
 
    intel_update_renderbuffers(pDRICtx, dPriv);
 
-   rb = intel_fb->color_rb[0];
+   rb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
    /* If the region isn't set, then intel_update_renderbuffers was unable
     * to get the buffers for the drawable.
     */
    if (rb->region == NULL)
       return;
 
-   type = GL_BGRA;
-   format = GL_UNSIGNED_BYTE;
    if (glx_texture_format == GLX_TEXTURE_FORMAT_RGB_EXT)
       internalFormat = GL_RGB;
    else
@@ -783,14 +781,14 @@ intelSetTexBuffer2(__DRIcontext *pDRICtx, GLint target,
 
    intelImage->face = target_to_face(target);
    intelImage->level = level;
-   texImage->TexFormat = intelChooseTextureFormat(&intel->ctx, internalFormat,
-                                                  type, format);
-   _mesa_set_fetch_functions(texImage, 2);
+   if (glx_texture_format == GLX_TEXTURE_FORMAT_RGB_EXT)
+      texImage->TexFormat = MESA_FORMAT_XRGB8888;
+   else
+      texImage->TexFormat = MESA_FORMAT_ARGB8888;
    texImage->RowStride = rb->region->pitch;
    intel_miptree_reference(&intelImage->mt, intelObj->mt);
 
-   if (!intel_miptree_match_image(intelObj->mt, &intelImage->base,
-				  intelImage->face, intelImage->level)) {
+   if (!intel_miptree_match_image(intelObj->mt, &intelImage->base)) {
 	   fprintf(stderr, "miptree doesn't match image\n");
    }
 
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_obj.h b/src/mesa/drivers/dri/intel/intel_tex_obj.h
index 5a93461525..3ad10d3d23 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_obj.h
+++ b/src/mesa/drivers/dri/intel/intel_tex_obj.h
@@ -66,6 +66,7 @@ struct intel_texture_image
     * Else there is no image data.
     */
    struct intel_mipmap_tree *mt;
+   GLboolean used_as_render_target;
 };
 
 static INLINE struct intel_texture_object *
diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
index 751ec2c98c..7f1dc89022 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
@@ -72,8 +72,6 @@ intelTexSubimage(GLcontext * ctx,
    if (!pixels)
       return;
 
-   LOCK_HARDWARE(intel);
-
    /* Map buffer if necessary.  Need to lock to prevent other contexts
     * from uploading the buffer under us.
     */
@@ -85,13 +83,13 @@ intelTexSubimage(GLcontext * ctx,
                                                &dstRowStride,
                                                texImage->ImageOffsets);
    else {
-      if (texImage->IsCompressed) {
+      if (_mesa_is_format_compressed(texImage->TexFormat)) {
          dstRowStride =
-            _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
+            _mesa_format_row_stride(texImage->TexFormat, width);
          assert(dims != 3);
       }
       else {
-         dstRowStride = texImage->RowStride * texImage->TexFormat->TexelBytes;
+         dstRowStride = texImage->RowStride * _mesa_get_format_bytes(texImage->TexFormat);
       }
    }
 
@@ -105,18 +103,20 @@ intelTexSubimage(GLcontext * ctx,
                          xoffset, yoffset / 4,
                          (width + 3)  & ~3, (height + 3) / 4,
                          pixels, (width + 3) & ~3, 0, 0);
-      } else
+      }
+      else {
         memcpy(texImage->Data, pixels, imageSize);
+      }
    }
    else {
-      if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat,
-                                           texImage->TexFormat,
-                                           texImage->Data,
-                                           xoffset, yoffset, zoffset,
-                                           dstRowStride,
-                                           texImage->ImageOffsets,
-                                           width, height, depth,
-                                           format, type, pixels, packing)) {
+      if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
+                          texImage->TexFormat,
+                          texImage->Data,
+                          xoffset, yoffset, zoffset,
+                          dstRowStride,
+                          texImage->ImageOffsets,
+                          width, height, depth,
+                          format, type, pixels, packing)) {
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "intelTexSubImage");
       }
    }
@@ -127,8 +127,6 @@ intelTexSubimage(GLcontext * ctx,
       intel_miptree_image_unmap(intel, intelImage->mt);
       texImage->Data = NULL;
    }
-
-   UNLOCK_HARDWARE(intel);
 }
 
 
diff --git a/src/mesa/drivers/dri/intel/intel_tex_validate.c b/src/mesa/drivers/dri/intel/intel_tex_validate.c
index a284d5475f..c9a24ac398 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_validate.c
@@ -5,6 +5,7 @@
 #include "intel_batchbuffer.h"
 #include "intel_mipmap_tree.h"
 #include "intel_tex.h"
+#include "intel_chipset.h"
 
 #define FILE_DEBUG_FLAG DEBUG_TEXTURE
 
@@ -14,7 +15,8 @@
  * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
  */
 static void
-intel_calculate_first_last_level(struct intel_texture_object *intelObj)
+intel_calculate_first_last_level(struct intel_context *intel,
+				 struct intel_texture_object *intelObj)
 {
    struct gl_texture_object *tObj = &intelObj->base;
    const struct gl_texture_image *const baseImage =
@@ -40,27 +42,27 @@ intel_calculate_first_last_level(struct intel_texture_object *intelObj)
          firstLevel = lastLevel = tObj->BaseLevel;
       }
       else {
-#ifdef I915
-         firstLevel = tObj->BaseLevel + (GLint) (tObj->MinLod + 0.5);
-         firstLevel = MAX2(firstLevel, tObj->BaseLevel);
-         firstLevel = MIN2(firstLevel, tObj->BaseLevel + baseImage->MaxLog2);
-         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 */
-#else
-	 /* Currently not taking min/max lod into account here, those
-	  * values are programmed as sampler state elsewhere and we
-	  * upload the same mipmap levels regardless.  Not sure if
-	  * this makes sense as it means it isn't possible for the app
-	  * to use min/max lod to reduce texture memory pressure:
-	  */
-	 firstLevel = tObj->BaseLevel;
-	 lastLevel = MIN2(tObj->BaseLevel + baseImage->MaxLog2,
-			  tObj->MaxLevel);
-	 lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
-#endif
+	 if (intel->gen == 2) {
+	    firstLevel = tObj->BaseLevel + (GLint) (tObj->MinLod + 0.5);
+	    firstLevel = MAX2(firstLevel, tObj->BaseLevel);
+	    firstLevel = MIN2(firstLevel, tObj->BaseLevel + baseImage->MaxLog2);
+	    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 */
+	 } else {
+	    /* Min/max LOD are taken into account in sampler state.  We don't
+	     * want to re-layout textures just because clamping has been applied
+	     * since it means a bunch of blitting around and probably no memory
+	     * savings (since we have to keep the other levels around anyway).
+	     */
+	    firstLevel = tObj->BaseLevel;
+	    lastLevel = MIN2(tObj->BaseLevel + baseImage->MaxLog2,
+			     tObj->MaxLevel);
+	    /* need at least one level */
+	    lastLevel = MAX2(firstLevel, lastLevel);
+	 }
       }
       break;
    case GL_TEXTURE_RECTANGLE_NV:
@@ -135,9 +137,8 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
 
    /* What levels must the tree include at a minimum?
     */
-   intel_calculate_first_last_level(intelObj);
-   firstImage =
-      intel_texture_image(intelObj->base.Image[0][intelObj->firstLevel]);
+   intel_calculate_first_last_level(intel, intelObj);
+   firstImage = intel_texture_image(tObj->Image[0][intelObj->firstLevel]);
 
    /* Fallback case:
     */
@@ -165,11 +166,12 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
       intel_miptree_reference(&intelObj->mt, firstImage->mt);
    }
 
-   if (firstImage->base.IsCompressed) {
-      comp_byte = intel_compressed_num_bytes(firstImage->base.TexFormat->MesaFormat);
+   if (_mesa_is_format_compressed(firstImage->base.TexFormat)) {
+      comp_byte = intel_compressed_num_bytes(firstImage->base.TexFormat);
       cpp = comp_byte;
    }
-   else cpp = firstImage->base.TexFormat->TexelBytes;
+   else
+      cpp = _mesa_get_format_bytes(firstImage->base.TexFormat);
 
    /* Check tree can hold all active levels.  Check tree matches
     * target, imageFormat, etc.
@@ -189,7 +191,7 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
 	intelObj->mt->height0 != firstImage->base.Height ||
 	intelObj->mt->depth0 != firstImage->base.Depth ||
 	intelObj->mt->cpp != cpp ||
-	intelObj->mt->compressed != firstImage->base.IsCompressed)) {
+	intelObj->mt->compressed != _mesa_is_format_compressed(firstImage->base.TexFormat))) {
       intel_miptree_release(intel, &intelObj->mt);
    }
 
@@ -220,8 +222,13 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
             intel_texture_image(intelObj->base.Image[face][i]);
 
          /* Need to import images in main memory or held in other trees.
+	  * If it's a render target, then its data isn't needed to be in
+	  * the object tree (otherwise we'd be FBO incomplete), and we need
+	  * to keep track of the image's MT as needing to be pulled in still,
+	  * or we'll lose the rendering that's done to it.
           */
-         if (intelObj->mt != intelImage->mt) {
+         if (intelObj->mt != intelImage->mt &&
+	     !intelImage->used_as_render_target) {
             copy_image_data_to_tree(intel, intelObj, intelImage);
          }
       }
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_context.c b/src/mesa/drivers/dri/mach64/mach64_context.c
index 9c7f513c6f..3b4ef7ffd8 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.c
+++ b/src/mesa/drivers/dri/mach64/mach64_context.c
@@ -76,7 +76,7 @@ static const struct dri_debug_control debug_control[] =
     { NULL,    0 }
 };
 
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_EXT_texture_edge_clamp",         NULL },
@@ -89,11 +89,11 @@ const struct dri_extension card_extensions[] =
 /* Create the device specific context.
   */
 GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
-			       __DRIcontextPrivate *driContextPriv,
+			       __DRIcontext *driContextPriv,
                                void *sharedContextPrivate )
 {
    GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *driScreen = driContextPriv->driScreenPriv;
+   __DRIscreen *driScreen = driContextPriv->driScreenPriv;
    struct dd_function_table functions;
    mach64ContextPtr mmesa;
    mach64ScreenPtr mach64Screen;
@@ -260,7 +260,7 @@ GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
 
 /* Destroy the device specific context.
  */
-void mach64DestroyContext( __DRIcontextPrivate *driContextPriv  )
+void mach64DestroyContext( __DRIcontext *driContextPriv  )
 {
    mach64ContextPtr mmesa = (mach64ContextPtr) driContextPriv->driverPrivate;
 
@@ -307,9 +307,9 @@ void mach64DestroyContext( __DRIcontextPrivate *driContextPriv  )
  * buffer `b'.
  */
 GLboolean
-mach64MakeCurrent( __DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv )
+mach64MakeCurrent( __DRIcontext *driContextPriv,
+                 __DRIdrawable *driDrawPriv,
+                 __DRIdrawable *driReadPriv )
 {
    if ( driContextPriv ) {
       GET_CURRENT_CONTEXT(ctx);
@@ -352,7 +352,7 @@ mach64MakeCurrent( __DRIcontextPrivate *driContextPriv,
 /* Force the context `c' to be unbound from its buffer.
  */
 GLboolean
-mach64UnbindContext( __DRIcontextPrivate *driContextPriv )
+mach64UnbindContext( __DRIcontext *driContextPriv )
 {
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.h b/src/mesa/drivers/dri/mach64/mach64_context.h
index 854751626d..18fc859d01 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.h
+++ b/src/mesa/drivers/dri/mach64/mach64_context.h
@@ -232,9 +232,9 @@ struct mach64_context {
 
    /* Mirrors of some DRI state
     */
-   __DRIcontextPrivate	*driContext;	/* DRI context */
-   __DRIscreenPrivate	*driScreen;	/* DRI screen */
-   __DRIdrawablePrivate	*driDrawable;	/* DRI drawable bound to this ctx */
+   __DRIcontext	*driContext;	/* DRI context */
+   __DRIscreen	*driScreen;	/* DRI screen */
+   __DRIdrawable	*driDrawable;	/* DRI drawable bound to this ctx */
 
    unsigned int lastStamp;		/* mirror driDrawable->lastStamp */
 
@@ -274,16 +274,16 @@ struct mach64_context {
 
 
 extern GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
-				      __DRIcontextPrivate *driContextPriv,
+				      __DRIcontext *driContextPriv,
                                       void *sharedContextPrivate );
 
-extern void mach64DestroyContext( __DRIcontextPrivate * );
+extern void mach64DestroyContext( __DRIcontext * );
 
-extern GLboolean mach64MakeCurrent( __DRIcontextPrivate *driContextPriv,
-                                    __DRIdrawablePrivate *driDrawPriv,
-                                    __DRIdrawablePrivate *driReadPriv );
+extern GLboolean mach64MakeCurrent( __DRIcontext *driContextPriv,
+                                    __DRIdrawable *driDrawPriv,
+                                    __DRIdrawable *driReadPriv );
 
-extern GLboolean mach64UnbindContext( __DRIcontextPrivate *driContextPriv );
+extern GLboolean mach64UnbindContext( __DRIcontext *driContextPriv );
 
 /* ================================================================
  * Byte ordering
diff --git a/src/mesa/drivers/dri/mach64/mach64_ioctl.c b/src/mesa/drivers/dri/mach64/mach64_ioctl.c
index ef5c0625c3..03587c44fd 100644
--- a/src/mesa/drivers/dri/mach64/mach64_ioctl.c
+++ b/src/mesa/drivers/dri/mach64/mach64_ioctl.c
@@ -279,7 +279,7 @@ static int mach64WaitForFrameCompletion( mach64ContextPtr mmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void mach64CopyBuffer( __DRIdrawablePrivate *dPriv )
+void mach64CopyBuffer( __DRIdrawable *dPriv )
 {
    mach64ContextPtr mmesa;
    GLint nbox, i, ret;
@@ -668,7 +668,7 @@ void mach64PerformanceBoxesLocked( mach64ContextPtr mmesa )
 static void mach64DDClear( GLcontext *ctx, GLbitfield mask )
 {
    mach64ContextPtr mmesa = MACH64_CONTEXT( ctx );
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
    drm_mach64_clear_t clear;
    GLuint flags = 0;
    GLint i;
diff --git a/src/mesa/drivers/dri/mach64/mach64_ioctl.h b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
index 6ef9bc0bca..1ffda1932f 100644
--- a/src/mesa/drivers/dri/mach64/mach64_ioctl.h
+++ b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
@@ -78,7 +78,7 @@ extern void mach64FireBlitLocked( mach64ContextPtr mmesa, void *buffer,
 				  GLint offset, GLint pitch, GLint format,
 				  GLint x, GLint y, GLint width, GLint height );
 
-extern void mach64CopyBuffer( __DRIdrawablePrivate *dPriv );
+extern void mach64CopyBuffer( __DRIdrawable *dPriv );
 #if ENABLE_PERF_BOXES
 extern void mach64PerformanceCounters( mach64ContextPtr mmesa );
 extern void mach64PerformanceBoxesLocked( mach64ContextPtr mmesa );
diff --git a/src/mesa/drivers/dri/mach64/mach64_lock.c b/src/mesa/drivers/dri/mach64/mach64_lock.c
index d018ba4174..8653c77da5 100644
--- a/src/mesa/drivers/dri/mach64/mach64_lock.c
+++ b/src/mesa/drivers/dri/mach64/mach64_lock.c
@@ -51,8 +51,8 @@ int   prevLockLine = 0;
  */
 void mach64GetLock( mach64ContextPtr mmesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
-   __DRIscreenPrivate *sPriv = mmesa->driScreen;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
+   __DRIscreen *sPriv = mmesa->driScreen;
    drm_mach64_sarea_t *sarea = mmesa->sarea;
    int i;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vb.c b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
index 99f1a14e17..816682ec5f 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vb.c
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vb.c
@@ -207,19 +207,19 @@ INTERP_QUALIFIER void TAG(interp_extras)( GLcontext *ctx,
    LOCALVARS
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      assert(VB->ColorPtr[1]->stride == 4 * sizeof(GLfloat));
+   if (VB->BackfaceColorPtr) {
+      assert(VB->BackfaceColorPtr->stride == 4 * sizeof(GLfloat));
       
       INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		 GET_COLOR(VB->BackfaceColorPtr, dst),
+		 GET_COLOR(VB->BackfaceColorPtr, out),
+		 GET_COLOR(VB->BackfaceColorPtr, in) );
 
-      if (VB->SecondaryColorPtr[1]) {
+      if (VB->BackfaceSecondaryColorPtr) {
 	 INTERP_3F( t,
-		       GET_COLOR(VB->SecondaryColorPtr[1], dst),
-		       GET_COLOR(VB->SecondaryColorPtr[1], out),
-		       GET_COLOR(VB->SecondaryColorPtr[1], in) );
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, out),
+		    GET_COLOR(VB->BackfaceSecondaryColorPtr, in) );
       }
    }
 
@@ -236,13 +236,13 @@ INTERP_QUALIFIER void TAG(copy_pv_extras)( GLcontext *ctx,
    LOCALVARS
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-      COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst), 
-		GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+      COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		GET_COLOR(VB->BackfaceColorPtr, src) );
 
-      if (VB->SecondaryColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->SecondaryColorPtr[1], dst), 
-		   GET_COLOR(VB->SecondaryColorPtr[1], src) );
+      if (VB->BackfaceSecondaryColorPtr) {
+	 COPY_4FV( GET_COLOR(VB->BackfaceSecondaryColorPtr, dst),
+		   GET_COLOR(VB->BackfaceSecondaryColorPtr, src) );
       }
    }
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h b/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
index 684f2acc89..6e5fa3520e 100644
--- a/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
+++ b/src/mesa/drivers/dri/mach64/mach64_native_vbtmp.h
@@ -103,10 +103,10 @@ static void TAG(emit)( GLcontext *ctx,
 #if DO_TEX1
    {
       const GLuint t1 = GET_TEXSOURCE(1);
-      tc1 = VB->TexCoordPtr[t1]->data;
-      tc1_stride = VB->TexCoordPtr[t1]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
 #if DO_PTEX
-      tc1_size = VB->TexCoordPtr[t1]->size;
+      tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size;
 #endif
    }
 #endif
@@ -114,18 +114,18 @@ static void TAG(emit)( GLcontext *ctx,
 #if DO_TEX0
    {
       const GLuint t0 = GET_TEXSOURCE(0);
-      tc0 = VB->TexCoordPtr[t0]->data;
-      tc0_stride = VB->TexCoordPtr[t0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
 #if DO_PTEX
-      tc0_size = VB->TexCoordPtr[t0]->size;
+      tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size;
 #endif
    }
 #endif
 
 #if DO_SPEC
-   if (VB->SecondaryColorPtr[0]) {
-      spec = VB->SecondaryColorPtr[0]->data;
-      spec_stride = VB->SecondaryColorPtr[0]->stride;
+   if (VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
+      spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+      spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
    } else {
       spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
       spec_stride = 0;
@@ -133,9 +133,9 @@ static void TAG(emit)( GLcontext *ctx,
 #endif
 
 #if DO_FOG
-   if (VB->FogCoordPtr) {
-      fog = VB->FogCoordPtr->data;
-      fog_stride = VB->FogCoordPtr->stride;
+   if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+      fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+      fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
    } else {
       static GLfloat tmp[4] = {0, 0, 0, 0};
       fog = &tmp;
@@ -144,8 +144,8 @@ static void TAG(emit)( GLcontext *ctx,
 #endif
 
 #if DO_RGBA
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 #endif
 
    coord = VB->NdcPtr->data;
@@ -319,8 +319,8 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
@@ -328,12 +328,12 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
    /* No hardware support for projective texture.  Can fake it for
     * TEX0 only.
     */
-   if ((DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
+   if ((DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4)) {
       PTEX_FALLBACK();
       return GL_FALSE;
    }
 
-   if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
+   if (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4) {
       if (DO_TEX1) {
 	 PTEX_FALLBACK();
       }
diff --git a/src/mesa/drivers/dri/mach64/mach64_screen.c b/src/mesa/drivers/dri/mach64/mach64_screen.c
index 6440027ca4..1ed3b0b70e 100644
--- a/src/mesa/drivers/dri/mach64/mach64_screen.c
+++ b/src/mesa/drivers/dri/mach64/mach64_screen.c
@@ -67,10 +67,8 @@ static const GLuint __driNConfigOptions = 3;
 static const GLuint __driNConfigOptions = 2;
 #endif
 
-extern const struct dri_extension card_extensions[];
-
 static const __DRIconfig **
-mach64FillInModes( __DRIscreenPrivate *psp,
+mach64FillInModes( __DRIscreen *psp,
 		   unsigned pixel_bits, unsigned depth_bits,
 		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -146,7 +144,7 @@ mach64FillInModes( __DRIscreenPrivate *psp,
 /* Create the device specific screen private data struct.
  */
 static mach64ScreenRec *
-mach64CreateScreen( __DRIscreenPrivate *sPriv )
+mach64CreateScreen( __DRIscreen *sPriv )
 {
    mach64ScreenPtr mach64Screen;
    ATIDRIPtr serverInfo = (ATIDRIPtr)sPriv->pDevPriv;
@@ -274,7 +272,7 @@ mach64CreateScreen( __DRIscreenPrivate *sPriv )
 /* Destroy the device specific screen private data struct.
  */
 static void
-mach64DestroyScreen( __DRIscreenPrivate *driScreen )
+mach64DestroyScreen( __DRIscreen *driScreen )
 {
    mach64ScreenRec *mach64Screen = (mach64ScreenRec *) driScreen->private;
 
@@ -301,8 +299,8 @@ mach64DestroyScreen( __DRIscreenPrivate *driScreen )
  * data.
  */
 static GLboolean
-mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
-		    __DRIdrawablePrivate *driDrawPriv,
+mach64CreateBuffer( __DRIscreen *driScrnPriv,
+		    __DRIdrawable *driDrawPriv,
 		    const __GLcontextModes *mesaVis,
 		    GLboolean isPixmap )
 {
@@ -316,7 +314,7 @@ mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->frontOffset, screen->frontPitch,
@@ -327,7 +325,7 @@ mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->backOffset, screen->backPitch,
@@ -338,7 +336,7 @@ mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
+            = driNewRenderbuffer(MESA_FORMAT_Z16,
                                  NULL, screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
                                  driDrawPriv);
@@ -348,7 +346,7 @@ mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
       else if (mesaVis->depthBits == 24) {
          /* XXX I don't think 24-bit Z is supported - so this isn't used */
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
+            = driNewRenderbuffer(MESA_FORMAT_Z24_S8,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -372,7 +370,7 @@ mach64CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-mach64DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+mach64DestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
@@ -380,7 +378,7 @@ mach64DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
 
 /* Copy the back color buffer to the front color buffer */
 static void
-mach64SwapBuffers(__DRIdrawablePrivate *dPriv)
+mach64SwapBuffers(__DRIdrawable *dPriv)
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
       mach64ContextPtr mmesa;
@@ -402,7 +400,7 @@ mach64SwapBuffers(__DRIdrawablePrivate *dPriv)
 /* Initialize the driver specific screen private data.
  */
 static GLboolean
-mach64InitDriver( __DRIscreenPrivate *driScreen )
+mach64InitDriver( __DRIscreen *driScreen )
 {
    driScreen->private = (void *) mach64CreateScreen( driScreen );
 
@@ -422,7 +420,7 @@ mach64InitDriver( __DRIscreenPrivate *driScreen )
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-mach64InitScreen(__DRIscreenPrivate *psp)
+mach64InitScreen(__DRIscreen *psp)
 {
    static const __DRIversion ddx_expected = { 6, 4, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
@@ -436,18 +434,6 @@ mach64InitScreen(__DRIscreenPrivate *psp)
       return NULL;
    }
    
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-
    if (!mach64InitDriver(psp))
       return NULL;
 
@@ -471,3 +457,9 @@ const struct __DriverAPIRec driDriverAPI = {
    .SwapBuffersMSC  = NULL
 };
 
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/mach64/mach64_screen.h b/src/mesa/drivers/dri/mach64/mach64_screen.h
index be5e29a3e5..1966809c03 100644
--- a/src/mesa/drivers/dri/mach64/mach64_screen.h
+++ b/src/mesa/drivers/dri/mach64/mach64_screen.h
@@ -70,7 +70,7 @@ typedef struct {
 
    drmBufMapPtr buffers;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
 
    driOptionCache optionCache;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_span.c b/src/mesa/drivers/dri/mach64/mach64_span.c
index 91d46ce32e..b4ba2a41c9 100644
--- a/src/mesa/drivers/dri/mach64/mach64_span.c
+++ b/src/mesa/drivers/dri/mach64/mach64_span.c
@@ -40,8 +40,8 @@
 
 #define LOCAL_VARS							\
    mach64ContextPtr mmesa = MACH64_CONTEXT(ctx);			\
-   __DRIscreenPrivate *sPriv = mmesa->driScreen;			\
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;			\
+   __DRIscreen *sPriv = mmesa->driScreen;			\
+   __DRIdrawable *dPriv = mmesa->driDrawable;			\
    driRenderbuffer *drb = (driRenderbuffer *) rb;			\
    GLuint height = dPriv->h;						\
    GLushort p;								\
@@ -49,8 +49,8 @@
 
 #define LOCAL_DEPTH_VARS						\
    mach64ContextPtr mmesa = MACH64_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;			\
-   __DRIscreenPrivate *driScreen = mmesa->driScreen;			\
+   __DRIdrawable *dPriv = mmesa->driDrawable;			\
+   __DRIscreen *driScreen = mmesa->driScreen;			\
    driRenderbuffer *drb = (driRenderbuffer *) rb;			\
    GLuint height = dPriv->h;						\
    char *buf = (char *)(driScreen->pFB + drb->offset +			\
@@ -157,15 +157,13 @@ void mach64DDInitSpanFuncs( GLcontext *ctx )
 void
 mach64SetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
 {
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         mach64InitPointers_RGB565(&drb->Base);
-      }
-      else {
-         mach64InitPointers_ARGB8888(&drb->Base);
-      }
+   if (drb->Base.Format == MESA_FORMAT_RGB565) {
+      mach64InitPointers_RGB565(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (drb->Base.Format == MESA_FORMAT_ARGB8888) {
+      mach64InitPointers_ARGB8888(&drb->Base);
+   }
+   else if (drb->Base.Format == MESA_FORMAT_Z16) {
       mach64InitDepthPointers_z16(&drb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/mach64/mach64_state.c b/src/mesa/drivers/dri/mach64/mach64_state.c
index 3a023187ce..df7cbc8670 100644
--- a/src/mesa/drivers/dri/mach64/mach64_state.c
+++ b/src/mesa/drivers/dri/mach64/mach64_state.c
@@ -388,7 +388,7 @@ static void mach64UpdateClipping( GLcontext *ctx )
    mach64ScreenPtr mach64Screen = mmesa->mach64Screen;
 
    if ( mmesa->driDrawable ) {
-      __DRIdrawablePrivate *drawable = mmesa->driDrawable;
+      __DRIdrawable *drawable = mmesa->driDrawable;
       int x1 = 0;
       int y1 = 0;
       int x2 = drawable->w - 1;
@@ -527,10 +527,10 @@ static void mach64UpdateMasks( GLcontext *ctx )
    /* mach64 can't color mask with alpha blending enabled */
    if ( !ctx->Color.BlendEnabled ) {
       mask = mach64PackColor( mmesa->mach64Screen->cpp,
-			      ctx->Color.ColorMask[RCOMP],
-			      ctx->Color.ColorMask[GCOMP],
-			      ctx->Color.ColorMask[BCOMP],
-			      ctx->Color.ColorMask[ACOMP] );
+			      ctx->Color.ColorMask[0][RCOMP],
+			      ctx->Color.ColorMask[0][GCOMP],
+			      ctx->Color.ColorMask[0][BCOMP],
+			      ctx->Color.ColorMask[0][ACOMP] );
    }
 
    if ( mmesa->setup.dp_write_mask != mask ) {
@@ -689,7 +689,7 @@ static void mach64DDLogicOpCode( GLcontext *ctx, GLenum opcode )
 void mach64SetCliprects( GLcontext *ctx, GLenum mode )
 {
    mach64ContextPtr mmesa = MACH64_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
 
    switch ( mode ) {
    case GL_FRONT_LEFT:
diff --git a/src/mesa/drivers/dri/mach64/mach64_tex.c b/src/mesa/drivers/dri/mach64/mach64_tex.c
index 225d23179e..6627d3c38a 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tex.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tex.c
@@ -41,7 +41,6 @@
 #include "main/simple_list.h"
 #include "main/enums.h"
 #include "main/texstore.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/imports.h"
@@ -131,14 +130,14 @@ mach64AllocTexObj( struct gl_texture_object *texObj )
 
    mach64SetTexWrap( t, texObj->WrapS, texObj->WrapT );
    mach64SetTexFilter( t, texObj->MinFilter, texObj->MagFilter );
-   mach64SetTexBorderColor( t, texObj->BorderColor );
+   mach64SetTexBorderColor( t, texObj->BorderColor.f );
 
    return t;
 }
 
 
 /* Called by the _mesa_store_teximage[123]d() functions. */
-static const struct gl_texture_format *
+static gl_format
 mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 			   GLenum format, GLenum type )
 {
@@ -167,15 +166,15 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGBA2:
    case GL_COMPRESSED_RGBA:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888;
+         return MESA_FORMAT_ARGB8888;
       else
-         return &_mesa_texformat_argb4444;
+         return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888;
+         return MESA_FORMAT_ARGB8888;
       else
-         return &_mesa_texformat_argb1555;
+         return MESA_FORMAT_ARGB1555;
 
    case GL_RGBA8:
    case GL_RGB10_A2:
@@ -183,9 +182,9 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGBA16:
    case GL_RGBA4:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888;
+         return MESA_FORMAT_ARGB8888;
       else
-         return &_mesa_texformat_argb4444;
+         return MESA_FORMAT_ARGB4444;
 
    case 3:
    case GL_RGB:
@@ -198,9 +197,9 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGB16:
    case GL_COMPRESSED_RGB:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888;
+         return MESA_FORMAT_ARGB8888;
       else
-         return &_mesa_texformat_rgb565;
+         return MESA_FORMAT_RGB565;
 
    case 1:
    case GL_LUMINANCE:
@@ -210,9 +209,9 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888; /* inefficient but accurate */
+         return MESA_FORMAT_ARGB8888; /* inefficient but accurate */
       else
-         return &_mesa_texformat_argb1555;
+         return MESA_FORMAT_ARGB1555;
 
    case GL_INTENSITY4:
    case GL_INTENSITY:
@@ -221,9 +220,9 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
       if (mmesa->mach64Screen->cpp == 4)
-         return &_mesa_texformat_argb8888; /* inefficient but accurate */
+         return MESA_FORMAT_ARGB8888; /* inefficient but accurate */
       else
-         return &_mesa_texformat_argb4444;
+         return MESA_FORMAT_ARGB4444;
 
    case GL_COLOR_INDEX:
    case GL_COLOR_INDEX1_EXT:
@@ -232,18 +231,18 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_COLOR_INDEX8_EXT:
    case GL_COLOR_INDEX12_EXT:
    case GL_COLOR_INDEX16_EXT:
-      return &_mesa_texformat_ci8;
+      return MESA_FORMAT_CI8;
 
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_APPLE ||
           type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    default:
       _mesa_problem( ctx, "unexpected format in %s", __FUNCTION__ );
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 }
 
@@ -471,7 +470,7 @@ static void mach64DDTexParameter( GLcontext *ctx, GLenum target,
 
    case GL_TEXTURE_BORDER_COLOR:
       if ( t->base.bound ) FLUSH_BATCH( mmesa );
-      mach64SetTexBorderColor( t, tObj->BorderColor );
+      mach64SetTexBorderColor( t, tObj->BorderColor.f );
       break;
 
    case GL_TEXTURE_BASE_LEVEL:
@@ -566,8 +565,6 @@ void mach64InitTextureFuncs( struct dd_function_table *functions )
    functions->IsTextureResident		= driIsTextureResident;
 
    functions->UpdateTexturePalette	= NULL;
-   functions->ActiveTexture		= NULL;
-   functions->PrioritizeTexture		= NULL;
 
    driInitTextureFormats();
 }
diff --git a/src/mesa/drivers/dri/mach64/mach64_texmem.c b/src/mesa/drivers/dri/mach64/mach64_texmem.c
index 734e547952..b97e9eec25 100644
--- a/src/mesa/drivers/dri/mach64/mach64_texmem.c
+++ b/src/mesa/drivers/dri/mach64/mach64_texmem.c
@@ -31,6 +31,11 @@
  *   Jose Fonseca <j_r_fonseca@yahoo.co.uk>
  */
 
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+
 #include "mach64_context.h"
 #include "mach64_state.h"
 #include "mach64_ioctl.h"
@@ -38,12 +43,6 @@
 #include "mach64_tris.h"
 #include "mach64_tex.h"
 
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/simple_list.h"
-#include "main/texformat.h"
-#include "main/imports.h"
-
 
 /* Destroy hardware state associated with texture `t'.
  */
@@ -76,6 +75,7 @@ static void mach64UploadAGPSubImage( mach64ContextPtr mmesa,
    struct gl_texture_image *image;
    int texelsPerDword = 0;
    int dwords;
+   GLuint texelBytes;
 
    /* Ensure we have a valid texture to upload */
    if ( ( level < 0 ) || ( level > mmesa->glCtx->Const.MaxTextureLevels ) )
@@ -85,7 +85,9 @@ static void mach64UploadAGPSubImage( mach64ContextPtr mmesa,
    if ( !image )
       return;
 
-   switch ( image->TexFormat->TexelBytes ) {
+   texelBytes = _mesa_get_format_bytes(image->TexFormat);
+
+   switch ( texelBytes ) {
    case 1: texelsPerDword = 4; break;
    case 2: texelsPerDword = 2; break;
    case 4: texelsPerDword = 1; break;
@@ -118,8 +120,8 @@ static void mach64UploadAGPSubImage( mach64ContextPtr mmesa,
    {
       CARD32 *dst = (CARD32 *)((char *)mach64Screen->agpTextures.map + t->base.memBlock->ofs);
       const GLubyte *src = (const GLubyte *) image->Data +
-	 (y * image->Width + x) * image->TexFormat->TexelBytes;
-      const GLuint bytes = width * height * image->TexFormat->TexelBytes;
+	 (y * image->Width + x) * texelBytes;
+      const GLuint bytes = width * height * texelBytes;
       memcpy(dst, src, bytes);
    }
 
@@ -140,6 +142,7 @@ static void mach64UploadLocalSubImage( mach64ContextPtr mmesa,
    const int maxdwords = (MACH64_BUFFER_MAX_DWORDS - (MACH64_HOSTDATA_BLIT_OFFSET / 4));
    CARD32 pitch, offset;
    int i;
+   GLuint texelBytes;
 
    /* Ensure we have a valid texture to upload */
    if ( ( level < 0 ) || ( level > mmesa->glCtx->Const.MaxTextureLevels ) )
@@ -149,7 +152,9 @@ static void mach64UploadLocalSubImage( mach64ContextPtr mmesa,
    if ( !image )
       return;
 
-   switch ( image->TexFormat->TexelBytes ) {
+   texelBytes = _mesa_get_format_bytes(image->TexFormat);
+
+   switch ( texelBytes ) {
    case 1: texelsPerDword = 4; break;
    case 2: texelsPerDword = 2; break;
    case 4: texelsPerDword = 1; break;
@@ -259,7 +264,7 @@ static void mach64UploadLocalSubImage( mach64ContextPtr mmesa,
 
        {
           const GLubyte *src = (const GLubyte *) image->Data +
-             (y * image->Width + x) * image->TexFormat->TexelBytes;
+             (y * image->Width + x) * texelBytes;
 
           mach64FireBlitLocked( mmesa, (void *)src, offset, pitch, format,
 				x, y, width, height );
diff --git a/src/mesa/drivers/dri/mach64/mach64_texstate.c b/src/mesa/drivers/dri/mach64/mach64_texstate.c
index fd2369dd88..df0a09a5c1 100644
--- a/src/mesa/drivers/dri/mach64/mach64_texstate.c
+++ b/src/mesa/drivers/dri/mach64/mach64_texstate.c
@@ -33,7 +33,6 @@
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 
 #include "mach64_context.h"
 #include "mach64_ioctl.h"
@@ -55,7 +54,7 @@ static void mach64SetTexImages( mach64ContextPtr mmesa,
    if ( MACH64_DEBUG & DEBUG_VERBOSE_API )
       fprintf( stderr, "%s( %p )\n", __FUNCTION__, tObj );
 
-   switch (baseImage->TexFormat->MesaFormat) {
+   switch (baseImage->TexFormat) {
    case MESA_FORMAT_ARGB8888:
       t->textureFormat = MACH64_DATATYPE_ARGB8888;
       break;
@@ -89,7 +88,7 @@ static void mach64SetTexImages( mach64ContextPtr mmesa,
 
    totalSize = ( baseImage->Height *
 		 baseImage->Width *
-		 baseImage->TexFormat->TexelBytes );
+		 _mesa_get_format_bytes(baseImage->TexFormat) );
 
    totalSize = (totalSize + 31) & ~31;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_tris.c b/src/mesa/drivers/dri/mach64/mach64_tris.c
index f2e8e2e3ae..c2a0adfef0 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tris.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tris.c
@@ -1297,7 +1297,8 @@ do {						\
 
 #define LOCAL_VARS(n)						\
    mach64ContextPtr mmesa = MACH64_CONTEXT(ctx);		\
-   GLuint color[n], spec[n];					\
+   GLuint color[n] = { 0 };					\
+   GLuint spec[n] = { 0 };					\
    GLuint vertex_size = mmesa->vertex_size;			\
    const GLuint xyoffset = 9;					\
    const GLuint coloroffset = 8;				\
diff --git a/src/mesa/drivers/dri/mach64/mach64_vbtmp.h b/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
index 938804af9e..60bfab8f6d 100644
--- a/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
+++ b/src/mesa/drivers/dri/mach64/mach64_vbtmp.h
@@ -156,53 +156,53 @@ static void TAG(emit)( GLcontext *ctx,
 
    if (DO_TEX3) {
       const GLuint t3 = GET_TEXSOURCE(3);
-      tc3 = VB->TexCoordPtr[t3]->data;
-      tc3_stride = VB->TexCoordPtr[t3]->stride;
+      tc3 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->data;
+      tc3_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->stride;
       if (DO_PTEX)
-	 tc3_size = VB->TexCoordPtr[t3]->size;
+	 tc3_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t3]->size;
    }
 
    if (DO_TEX2) {
       const GLuint t2 = GET_TEXSOURCE(2);
-      tc2 = VB->TexCoordPtr[t2]->data;
-      tc2_stride = VB->TexCoordPtr[t2]->stride;
+      tc2 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->data;
+      tc2_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->stride;
       if (DO_PTEX)
-	 tc2_size = VB->TexCoordPtr[t2]->size;
+	 tc2_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size;
    }
 
    if (DO_TEX1) {
       const GLuint t1 = GET_TEXSOURCE(1);
-      tc1 = VB->TexCoordPtr[t1]->data;
-      tc1_stride = VB->TexCoordPtr[t1]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
       if (DO_PTEX)
-	 tc1_size = VB->TexCoordPtr[t1]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size;
    }
 
    if (DO_TEX0) {
       const GLuint t0 = GET_TEXSOURCE(0);
-      tc0_stride = VB->TexCoordPtr[t0]->stride;
-      tc0 = VB->TexCoordPtr[t0]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
       if (DO_PTEX) 
-	 tc0_size = VB->TexCoordPtr[t0]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size;
    }
 
    if (DO_RGBA) {
-      col = VB->ColorPtr[0]->data;
-      col_stride = VB->ColorPtr[0]->stride;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
    }
 
    if (DO_SPEC) {
-      spec = VB->SecondaryColorPtr[0]->data;
-      spec_stride = VB->SecondaryColorPtr[0]->stride;
+      spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+      spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
    } else {
       spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
       spec_stride = 0;
    }
 
    if (DO_FOG) {
-      if (VB->FogCoordPtr) {
-	 fog = VB->FogCoordPtr->data;
-	 fog_stride = VB->FogCoordPtr->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+	 fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+	 fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
       } else {
 	 static GLfloat tmp[4] = {0, 0, 0, 0};
 	 fog = &tmp;
@@ -384,8 +384,8 @@ static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
 
    ASSERT(stride == 4);
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    /* Pack what's left into a 4-dword vertex.  Color is in a different
     * place, and there is no 'w' coordinate.
@@ -432,8 +432,8 @@ static void TAG(emit)( GLcontext *ctx, GLuint start, GLuint end,
    GLfloat *v = (GLfloat *)dest;
    int i;
 
-   col = VB->ColorPtr[0]->data;
-   col_stride = VB->ColorPtr[0]->stride;
+   col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+   col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
 
    if (start)
       STRIDE_4F(col, col_stride * start);
@@ -473,22 +473,22 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX2] = VB->AttribPtr[_TNL_ATTRIB_TEX3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX1] = VB->AttribPtr[_TNL_ATTRIB_TEX2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
    
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4) ||
-       (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4))
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4) ||
+       (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4))
       return GL_FALSE;
 
    return GL_TRUE;
@@ -501,14 +501,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
 
    /* Force 'missing' texcoords to something valid.
     */
-   if (DO_TEX3 && VB->TexCoordPtr[2] == 0)
-      VB->TexCoordPtr[2] = VB->TexCoordPtr[3];
+   if (DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX2] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX2] = VB->AttribPtr[_TNL_ATTRIB_TEX3];
 
-   if (DO_TEX2 && VB->TexCoordPtr[1] == 0)
-      VB->TexCoordPtr[1] = VB->TexCoordPtr[2];
+   if (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX1] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX1] = VB->AttribPtr[_TNL_ATTRIB_TEX2];
 
-   if (DO_TEX1 && VB->TexCoordPtr[0] == 0)
-      VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+   if (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+      VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 
    if (DO_PTEX)
       return GL_TRUE;
@@ -516,14 +516,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
    /* No hardware support for projective texture.  Can fake it for
     * TEX0 only.
     */
-   if ((DO_TEX3 && VB->TexCoordPtr[GET_TEXSOURCE(3)]->size == 4) ||
-       (DO_TEX2 && VB->TexCoordPtr[GET_TEXSOURCE(2)]->size == 4) ||
-       (DO_TEX1 && VB->TexCoordPtr[GET_TEXSOURCE(1)]->size == 4)) {
+   if ((DO_TEX3 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(3)]->size == 4) ||
+       (DO_TEX2 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(2)]->size == 4) ||
+       (DO_TEX1 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(1)]->size == 4)) {
       PTEX_FALLBACK();
       return GL_FALSE;
    }
 
-   if (DO_TEX0 && VB->TexCoordPtr[GET_TEXSOURCE(0)]->size == 4) {
+   if (DO_TEX0 && VB->AttribPtr[_TNL_ATTRIB_TEX0 + GET_TEXSOURCE(0)]->size == 4) {
       if (DO_TEX1 || DO_TEX2 || DO_TEX3) {
 	 PTEX_FALLBACK();
       }
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/mga/mga_texstate.c b/src/mesa/drivers/dri/mga/mga_texstate.c
index ad765d1dd7..54eda62a96 100644
--- a/src/mesa/drivers/dri/mga/mga_texstate.c
+++ b/src/mesa/drivers/dri/mga/mga_texstate.c
@@ -27,22 +27,19 @@
  *    Keith Whitwell <keithw@tungstengraphics.com>
  */
 
-#include <stdlib.h>
+#include "main/context.h"
+#include "main/enums.h"
+#include "main/macros.h"
 #include "main/mm.h"
+#include "main/imports.h"
+#include "main/simple_list.h"
+
 #include "mgacontext.h"
 #include "mgatex.h"
 #include "mgaregs.h"
 #include "mgatris.h"
 #include "mgaioctl.h"
 
-#include "main/context.h"
-#include "main/enums.h"
-#include "main/macros.h"
-#include "main/imports.h"
-
-#include "main/simple_list.h"
-#include "main/texformat.h"
-
 #define MGA_USE_TABLE_FOR_FORMAT
 #ifdef MGA_USE_TABLE_FOR_FORMAT
 #define TMC_nr_tformat (MESA_FORMAT_YCBCR_REV + 1)
@@ -94,14 +91,14 @@ mgaSetTexImages( mgaContextPtr mmesa,
 	return;
     }
 #else
-    if ( (baseImage->TexFormat->MesaFormat >= TMC_nr_tformat)
-	 || (TMC_tformat[ baseImage->TexFormat->MesaFormat ] == 0) )
+    if ( (baseImage->TexFormat >= TMC_nr_tformat)
+	 || (TMC_tformat[ baseImage->TexFormat ] == 0) )
     {
 	_mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
 	return;
     }
 
-    txformat = TMC_tformat[ baseImage->TexFormat->MesaFormat ];
+    txformat = TMC_tformat[ baseImage->TexFormat ];
 
 #endif /* MGA_USE_TABLE_FOR_FORMAT */
 
@@ -131,7 +128,7 @@ mgaSetTexImages( mgaContextPtr mmesa,
 	 break;
 
       size = texImage->Width * texImage->Height *
-         baseImage->TexFormat->TexelBytes;
+         _mesa_get_format_bytes(baseImage->TexFormat);
 
       t->offsets[i] = totalSize;
       t->base.dirty_images[0] |= (1<<i);
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.c b/src/mesa/drivers/dri/mga/mga_xmesa.c
index 0dc76fea50..f835cb8bd6 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.c
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.c
@@ -78,7 +78,7 @@
 #endif
 #define need_GL_APPLE_vertex_array_object
 #define need_GL_NV_vertex_program
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 /* MGA configuration
  */
@@ -108,7 +108,7 @@ int MGA_DEBUG = 0;
 #endif
 
 static const __DRIconfig **
-mgaFillInModes( __DRIscreenPrivate *psp,
+mgaFillInModes( __DRIscreen *psp,
 		unsigned pixel_bits, unsigned depth_bits,
 		unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -190,7 +190,7 @@ const __DRIextension *mgaScreenExtensions[] = {
 };
 
 static GLboolean
-mgaInitDriver(__DRIscreenPrivate *sPriv)
+mgaInitDriver(__DRIscreen *sPriv)
 {
    mgaScreenPrivate *mgaScreen;
    MGADRIPtr         serverInfo = (MGADRIPtr)sPriv->pDevPriv;
@@ -332,7 +332,7 @@ mgaInitDriver(__DRIscreenPrivate *sPriv)
 
 
 static void
-mgaDestroyScreen(__DRIscreenPrivate *sPriv)
+mgaDestroyScreen(__DRIscreen *sPriv)
 {
    mgaScreenPrivate *mgaScreen = (mgaScreenPrivate *) sPriv->private;
 
@@ -426,14 +426,14 @@ static const struct dri_debug_control debug_control[] =
 
 static GLboolean
 mgaCreateContext( const __GLcontextModes *mesaVis,
-                  __DRIcontextPrivate *driContextPriv,
+                  __DRIcontext *driContextPriv,
                   void *sharedContextPrivate )
 {
    int i;
    unsigned   maxlevels;
    GLcontext *ctx, *shareCtx;
    mgaContextPtr mmesa;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    mgaScreenPrivate *mgaScreen = (mgaScreenPrivate *)sPriv->private;
    drm_mga_sarea_t *saPriv = (drm_mga_sarea_t *)(((char*)sPriv->pSAREA)+
 					      mgaScreen->sarea_priv_offset);
@@ -645,7 +645,7 @@ mgaCreateContext( const __GLcontextModes *mesaVis,
 }
 
 static void
-mgaDestroyContext(__DRIcontextPrivate *driContextPriv)
+mgaDestroyContext(__DRIcontext *driContextPriv)
 {
    mgaContextPtr mmesa = (mgaContextPtr) driContextPriv->driverPrivate;
 
@@ -697,8 +697,8 @@ mgaDestroyContext(__DRIcontextPrivate *driContextPriv)
 
 
 static GLboolean
-mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
+mgaCreateBuffer( __DRIscreen *driScrnPriv,
+                 __DRIdrawable *driDrawPriv,
                  const __GLcontextModes *mesaVis,
                  GLboolean isPixmap )
 {
@@ -723,7 +723,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->frontOffset, screen->frontPitch,
@@ -734,7 +734,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->backOffset, screen->backPitch,
@@ -745,7 +745,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
+            = driNewRenderbuffer(MESA_FORMAT_Z16,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -757,7 +757,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
          /* XXX is this right? */
          if (mesaVis->stencilBits) {
             driRenderbuffer *depthRb
-               = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
+               = driNewRenderbuffer(MESA_FORMAT_Z24_S8,
                                     NULL,
                                     screen->cpp,
                                     screen->depthOffset, screen->depthPitch,
@@ -767,7 +767,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
          }
          else {
             driRenderbuffer *depthRb
-               = driNewRenderbuffer(GL_DEPTH_COMPONENT32,
+               = driNewRenderbuffer(MESA_FORMAT_Z32,
                                     NULL,
                                     screen->cpp,
                                     screen->depthOffset, screen->depthPitch,
@@ -778,7 +778,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 32) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT32,
+            = driNewRenderbuffer(MESA_FORMAT_Z32,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -789,7 +789,7 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0 && !swStencil) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
+            = driNewRenderbuffer(MESA_FORMAT_S8,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -814,13 +814,13 @@ mgaCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-mgaDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+mgaDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
 static void
-mgaSwapBuffers(__DRIdrawablePrivate *dPriv)
+mgaSwapBuffers(__DRIdrawable *dPriv)
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
       mgaContextPtr mmesa;
@@ -839,7 +839,7 @@ mgaSwapBuffers(__DRIdrawablePrivate *dPriv)
 }
 
 static GLboolean
-mgaUnbindContext(__DRIcontextPrivate *driContextPriv)
+mgaUnbindContext(__DRIcontext *driContextPriv)
 {
    mgaContextPtr mmesa = (mgaContextPtr) driContextPriv->driverPrivate;
    if (mmesa)
@@ -855,9 +855,9 @@ mgaUnbindContext(__DRIcontextPrivate *driContextPriv)
  * But why are we doing context initialization here???
  */
 static GLboolean
-mgaMakeCurrent(__DRIcontextPrivate *driContextPriv,
-               __DRIdrawablePrivate *driDrawPriv,
-               __DRIdrawablePrivate *driReadPriv)
+mgaMakeCurrent(__DRIcontext *driContextPriv,
+               __DRIdrawable *driDrawPriv,
+               __DRIdrawable *driReadPriv)
 {
    if (driContextPriv) {
       mgaContextPtr mmesa = (mgaContextPtr) driContextPriv->driverPrivate;
@@ -892,7 +892,7 @@ mgaMakeCurrent(__DRIcontextPrivate *driContextPriv,
 
 void mgaGetLock( mgaContextPtr mmesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
    drm_mga_sarea_t *sarea = mmesa->sarea;
    int me = mmesa->hHWContext;
    int i;
@@ -945,22 +945,6 @@ static const __DRIconfig **mgaInitScreen(__DRIscreen *psp)
       return NULL;
 
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-   driInitExtensions( NULL, g400_extensions, GL_FALSE );
-   driInitExtensions(NULL, ARB_vp_extensions, GL_FALSE);
-   driInitExtensions( NULL, NV_vp_extensions, GL_FALSE );
-
    if (!mgaInitDriver(psp))
        return NULL;
 
@@ -976,7 +960,7 @@ static const __DRIconfig **mgaInitScreen(__DRIscreen *psp)
  * Get information about previous buffer swaps.
  */
 static int
-getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
+getSwapInfo( __DRIdrawable *dPriv, __DRIswapInfo * sInfo )
 {
    mgaContextPtr  mmesa;
 
@@ -1014,3 +998,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.h b/src/mesa/drivers/dri/mga/mga_xmesa.h
index 07c22bd596..aee146090c 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.h
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.h
@@ -67,7 +67,7 @@ typedef struct mga_screen_private_s {
    char *texVirtual[MGA_NR_TEX_HEAPS];
 
 
-   __DRIscreenPrivate *sPriv;
+   __DRIscreen *sPriv;
    drmBufMapPtr  bufs;
 
    drmRegion mmio;
diff --git a/src/mesa/drivers/dri/mga/mgacontext.h b/src/mesa/drivers/dri/mga/mgacontext.h
index 30640a29b3..4141565931 100644
--- a/src/mesa/drivers/dri/mga/mgacontext.h
+++ b/src/mesa/drivers/dri/mga/mgacontext.h
@@ -294,10 +294,10 @@ struct mga_context_t {
    drm_context_t hHWContext;
    drm_hw_lock_t *driHwLock;
    int driFd;
-   __DRIdrawablePrivate *driDrawable;
-   __DRIdrawablePrivate *driReadable;
+   __DRIdrawable *driDrawable;
+   __DRIdrawable *driReadable;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
    struct mga_screen_private_s *mgaScreen;
    drm_mga_sarea_t *sarea;
 
diff --git a/src/mesa/drivers/dri/mga/mgaioctl.c b/src/mesa/drivers/dri/mga/mgaioctl.c
index 4438bad920..8ce5d802ab 100644
--- a/src/mesa/drivers/dri/mga/mgaioctl.c
+++ b/src/mesa/drivers/dri/mga/mgaioctl.c
@@ -207,7 +207,7 @@ static void
 mgaClear( GLcontext *ctx, GLbitfield mask )
 {
    mgaContextPtr mmesa = MGA_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
    GLuint flags = 0;
    GLuint clear_color = mmesa->ClearColor;
    GLuint clear_depth = 0;
@@ -409,7 +409,7 @@ static void mgaWaitForFrameCompletion( mgaContextPtr mmesa )
 /*
  * Copy the back buffer to the front buffer.
  */
-void mgaCopyBuffer( __DRIdrawablePrivate *dPriv )
+void mgaCopyBuffer( __DRIdrawable *dPriv )
 {
    mgaContextPtr mmesa;
    drm_clip_rect_t *pbox;
@@ -417,7 +417,7 @@ void mgaCopyBuffer( __DRIdrawablePrivate *dPriv )
    GLint ret;
    GLint i;
    GLboolean   missed_target;
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+   __DRIscreen *psp = dPriv->driScreenPriv;
 
    assert(dPriv);
    assert(dPriv->driContextPriv);
diff --git a/src/mesa/drivers/dri/mga/mgaioctl.h b/src/mesa/drivers/dri/mga/mgaioctl.h
index dbc823de80..7a8660d203 100644
--- a/src/mesa/drivers/dri/mga/mgaioctl.h
+++ b/src/mesa/drivers/dri/mga/mgaioctl.h
@@ -32,7 +32,7 @@
 #include "mgacontext.h"
 #include "mga_xmesa.h"
 
-void mgaCopyBuffer( __DRIdrawablePrivate *dPriv );
+void mgaCopyBuffer( __DRIdrawable *dPriv );
 void mgaWaitForVBlank( mgaContextPtr mmesa );
 
 void mgaGetILoadBufferLocked( mgaContextPtr mmesa );
diff --git a/src/mesa/drivers/dri/mga/mgapixel.c b/src/mesa/drivers/dri/mga/mgapixel.c
index 977dfa0b76..69415f8a83 100644
--- a/src/mesa/drivers/dri/mga/mgapixel.c
+++ b/src/mesa/drivers/dri/mga/mgapixel.c
@@ -134,10 +134,10 @@ check_color_per_fragment_ops( const GLcontext *ctx )
 		    ctx->Fog.Enabled ||
 		    ctx->Scissor.Enabled ||
 		    ctx->Stencil._Enabled ||
-		    !ctx->Color.ColorMask[0] ||
-		    !ctx->Color.ColorMask[1] ||
-		    !ctx->Color.ColorMask[2] ||
-		    !ctx->Color.ColorMask[3] ||
+		    !ctx->Color.ColorMask[0][0] ||
+		    !ctx->Color.ColorMask[0][1] ||
+		    !ctx->Color.ColorMask[0][2] ||
+		    !ctx->Color.ColorMask[0][3] ||
 		    ctx->Color.ColorLogicOpEnabled ||
 		    ctx->Texture._EnabledUnits
            ) &&
@@ -150,10 +150,10 @@ static GLboolean
 check_depth_per_fragment_ops( const GLcontext *ctx )
 {
    return ( ctx->Current.RasterPosValid &&
-	    ctx->Color.ColorMask[RCOMP] == 0 &&
-	    ctx->Color.ColorMask[BCOMP] == 0 &&
-	    ctx->Color.ColorMask[GCOMP] == 0 &&
-	    ctx->Color.ColorMask[ACOMP] == 0 &&
+	    ctx->Color.ColorMask[0][RCOMP] == 0 &&
+	    ctx->Color.ColorMask[0][BCOMP] == 0 &&
+	    ctx->Color.ColorMask[0][GCOMP] == 0 &&
+	    ctx->Color.ColorMask[0][ACOMP] == 0 &&
 	    ctx->Pixel.ZoomX == 1.0F &&
 	    ( ctx->Pixel.ZoomY == 1.0F || ctx->Pixel.ZoomY == -1.0F ) );
 }
@@ -299,7 +299,7 @@ mgaTryReadPixels( GLcontext *ctx,
 
 #if 0
    {
-      __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+      __DRIdrawable *dPriv = mmesa->driDrawable;
       int nbox, retcode, i;
 
       UPDATE_LOCK( mmesa, DRM_LOCK_FLUSH | DRM_LOCK_QUIESCENT );
@@ -399,7 +399,7 @@ static void do_draw_pix( GLcontext *ctx,
 #if 0
    mgaContextPtr mmesa = MGA_CONTEXT(ctx);
    drmMGABlit blit;
-   __DRIdrawablePrivate *dPriv = mmesa->driDrawable;
+   __DRIdrawable *dPriv = mmesa->driDrawable;
    drm_clip_rect_t pbox = dPriv->pClipRects;
    int nbox = dPriv->numClipRects;
    int retcode, i;
@@ -525,10 +525,10 @@ mgaTryDrawPixels( GLcontext *ctx,
 	      mmesa->mgaScreen->backOffset);
 
       planemask = mgaPackColor(cpp,
-			       ctx->Color.ColorMask[RCOMP],
-			       ctx->Color.ColorMask[GCOMP],
-			       ctx->Color.ColorMask[BCOMP],
-			       ctx->Color.ColorMask[ACOMP]);
+			       ctx->Color.ColorMask[0][RCOMP],
+			       ctx->Color.ColorMask[0][GCOMP],
+			       ctx->Color.ColorMask[0][BCOMP],
+			       ctx->Color.ColorMask[0][ACOMP]);
 
       if (cpp == 2)
 	 planemask |= planemask << 16;
diff --git a/src/mesa/drivers/dri/mga/mgaspan.c b/src/mesa/drivers/dri/mga/mgaspan.c
index 5b6d323ca9..10606c152c 100644
--- a/src/mesa/drivers/dri/mga/mgaspan.c
+++ b/src/mesa/drivers/dri/mga/mgaspan.c
@@ -36,9 +36,9 @@
 
 #define LOCAL_VARS					\
    mgaContextPtr mmesa = MGA_CONTEXT(ctx);		\
-   __DRIscreenPrivate *sPriv = mmesa->driScreen;	\
+   __DRIscreen *sPriv = mmesa->driScreen;	\
    driRenderbuffer *drb = (driRenderbuffer *) rb;	\
-   const __DRIdrawablePrivate *dPriv = drb->dPriv;	\
+   const __DRIdrawable *dPriv = drb->dPriv;	\
    GLuint pitch = drb->pitch;				\
    GLuint height = dPriv->h;				\
    char *buf = (char *)(sPriv->pFB +			\
@@ -52,9 +52,9 @@
 
 #define LOCAL_DEPTH_VARS						\
    mgaContextPtr mmesa = MGA_CONTEXT(ctx);				\
-   __DRIscreenPrivate *sPriv = mmesa->driScreen;			\
+   __DRIscreen *sPriv = mmesa->driScreen;			\
    driRenderbuffer *drb = (driRenderbuffer *) rb;			\
-   const __DRIdrawablePrivate *dPriv = drb->dPriv;			\
+   const __DRIdrawable *dPriv = drb->dPriv;			\
    GLuint pitch = drb->pitch;						\
    GLuint height = dPriv->h;						\
    char *buf = (char *)(sPriv->pFB +					\
@@ -206,24 +206,22 @@ void mgaDDInitSpanFuncs( GLcontext *ctx )
 void
 mgaSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
 {
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         mgaInitPointers_565(&drb->Base);
-      }
-      else {
-         mgaInitPointers_8888(&drb->Base);
-      }
+   if (drb->Base.Format == MESA_FORMAT_RGB565) {
+      mgaInitPointers_565(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (drb->Base.Format == MESA_FORMAT_ARGB8888) {
+      mgaInitPointers_8888(&drb->Base);
+   }
+   else if (drb->Base.Format == MESA_FORMAT_Z16) {
       mgaInitDepthPointers_z16(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
+   else if (drb->Base.Format == MESA_FORMAT_Z24_S8) {
       mgaInitDepthPointers_z24_s8(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT32) {
+   else if (drb->Base.Format == MESA_FORMAT_Z32) {
       mgaInitDepthPointers_z32(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
+   else if (drb->Base.Format == MESA_FORMAT_S8) {
       mgaInitStencilPointers_z24_s8(&drb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/mga/mgastate.c b/src/mesa/drivers/dri/mga/mgastate.c
index 7c830ec097..0253044761 100644
--- a/src/mesa/drivers/dri/mga/mgastate.c
+++ b/src/mesa/drivers/dri/mga/mgastate.c
@@ -374,13 +374,11 @@ static void mgaDDColorMask(GLcontext *ctx,
 {
    mgaContextPtr mmesa = MGA_CONTEXT( ctx );
    mgaScreenPrivate *mgaScreen = mmesa->mgaScreen;
-
-
    GLuint mask = mgaPackColor(mgaScreen->cpp,
-			      ctx->Color.ColorMask[RCOMP],
-			      ctx->Color.ColorMask[GCOMP],
-			      ctx->Color.ColorMask[BCOMP],
-			      ctx->Color.ColorMask[ACOMP]);
+			      ctx->Color.ColorMask[0][RCOMP],
+			      ctx->Color.ColorMask[0][GCOMP],
+			      ctx->Color.ColorMask[0][BCOMP],
+			      ctx->Color.ColorMask[0][ACOMP]);
 
    if (mgaScreen->cpp == 2)
       mask = mask | (mask << 16);
@@ -748,7 +746,7 @@ static void mgaDDLogicOp( GLcontext *ctx, GLenum opcode )
 
 static void mga_set_cliprects(mgaContextPtr mmesa)
 {
-   __DRIdrawablePrivate *driDrawable = mmesa->driDrawable;
+   __DRIdrawable *driDrawable = mmesa->driDrawable;
 
    if ((mmesa->draw_buffer != MGA_FRONT)
        || (driDrawable->numBackClipRects == 0)) {
@@ -776,8 +774,8 @@ static void mga_set_cliprects(mgaContextPtr mmesa)
 
 void mgaUpdateRects( mgaContextPtr mmesa, GLuint buffers )
 {
-   __DRIdrawablePrivate *const driDrawable = mmesa->driDrawable;
-   __DRIdrawablePrivate *const driReadable = mmesa->driReadable;
+   __DRIdrawable *const driDrawable = mmesa->driDrawable;
+   __DRIdrawable *const driReadable = mmesa->driReadable;
 
    mmesa->dirty_cliprects = 0;	
 
diff --git a/src/mesa/drivers/dri/mga/mgatex.c b/src/mesa/drivers/dri/mga/mgatex.c
index 33eb0be449..62a9317cd4 100644
--- a/src/mesa/drivers/dri/mga/mgatex.c
+++ b/src/mesa/drivers/dri/mga/mgatex.c
@@ -27,23 +27,22 @@
 
 #include "main/glheader.h"
 #include "main/mm.h"
-#include "mgacontext.h"
-#include "mgatex.h"
-#include "mgaregs.h"
-#include "mgatris.h"
-#include "mgaioctl.h"
-
 #include "main/colormac.h"
 #include "main/context.h"
 #include "main/enums.h"
 #include "main/simple_list.h"
 #include "main/imports.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 
+#include "mgacontext.h"
+#include "mgatex.h"
+#include "mgaregs.h"
+#include "mgatris.h"
+#include "mgaioctl.h"
+
 #include "swrast/swrast.h"
 
 #include "xmlpool.h"
@@ -164,7 +163,7 @@ static void mgaSetTexBorderColor(mgaTextureObjectPtr t, const GLfloat color[4])
 }
 
 
-static const struct gl_texture_format *
+static gl_format
 mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 		        GLenum format, GLenum type )
 {
@@ -182,15 +181,15 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_INT_10_10_10_2:
       case GL_UNSIGNED_INT_2_10_10_10_REV:
-	 return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555;
+	 return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
       }
 
    case 3:
@@ -199,15 +198,15 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_5_6_5:
       case GL_UNSIGNED_SHORT_5_6_5_REV:
-	 return &_mesa_texformat_rgb565;
+	 return MESA_FORMAT_RGB565;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
       }
 
    case GL_RGBA8:
@@ -215,25 +214,25 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGBA12:
    case GL_RGBA16:
       return !force16bpt ?
-	  &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+	  MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case GL_RGBA4:
    case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
 
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+      return !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
 
    case GL_RGB5:
    case GL_RGB4:
    case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_ALPHA:
    case GL_ALPHA4:
@@ -242,7 +241,7 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_ALPHA16:
    case GL_COMPRESSED_ALPHA:
       /* FIXME: This will report incorrect component sizes... */
-      return MGA_IS_G400(mmesa) ? &_mesa_texformat_al88 : &_mesa_texformat_argb4444;
+      return MGA_IS_G400(mmesa) ? MESA_FORMAT_AL88 : MESA_FORMAT_ARGB4444;
 
    case 1:
    case GL_LUMINANCE:
@@ -252,7 +251,7 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
       /* FIXME: This will report incorrect component sizes... */
-      return MGA_IS_G400(mmesa) ? &_mesa_texformat_al88 : &_mesa_texformat_rgb565;
+      return MGA_IS_G400(mmesa) ? MESA_FORMAT_AL88 : MESA_FORMAT_RGB565;
 
    case 2:
    case GL_LUMINANCE_ALPHA:
@@ -264,7 +263,7 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
       /* FIXME: This will report incorrect component sizes... */
-      return MGA_IS_G400(mmesa) ? &_mesa_texformat_al88 : &_mesa_texformat_argb4444;
+      return MGA_IS_G400(mmesa) ? MESA_FORMAT_AL88 : MESA_FORMAT_ARGB4444;
 
    case GL_INTENSITY:
    case GL_INTENSITY4:
@@ -273,15 +272,15 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
       /* FIXME: This will report incorrect component sizes... */
-      return MGA_IS_G400(mmesa) ? &_mesa_texformat_i8 : &_mesa_texformat_argb4444;
+      return MGA_IS_G400(mmesa) ? MESA_FORMAT_I8 : MESA_FORMAT_ARGB4444;
 
    case GL_YCBCR_MESA:
       if (MGA_IS_G400(mmesa) &&
           (type == GL_UNSIGNED_SHORT_8_8_APPLE ||
            type == GL_UNSIGNED_BYTE))
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    case GL_COLOR_INDEX:
    case GL_COLOR_INDEX1_EXT:
@@ -290,14 +289,14 @@ mgaChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_COLOR_INDEX8_EXT:
    case GL_COLOR_INDEX12_EXT:
    case GL_COLOR_INDEX16_EXT:
-      return &_mesa_texformat_ci8;
+      return MESA_FORMAT_CI8;
 
    default:
       _mesa_problem( ctx, "unexpected texture format in %s", __FUNCTION__ );
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 
-   return NULL; /* never get here */
+   return MESA_FORMAT_NONE; /* never get here */
 }
 
 
@@ -333,7 +332,7 @@ mgaAllocTexObj( struct gl_texture_object *tObj )
 
       mgaSetTexWrapping( t, tObj->WrapS, tObj->WrapT );
       mgaSetTexFilter( t, tObj->MinFilter, tObj->MagFilter );
-      mgaSetTexBorderColor( t, tObj->BorderColor );
+      mgaSetTexBorderColor( t, tObj->BorderColor.f );
    }
 
    return( t );
@@ -462,7 +461,7 @@ mgaTexParameter( GLcontext *ctx, GLenum target,
 
    case GL_TEXTURE_BORDER_COLOR:
       FLUSH_BATCH(mmesa);
-      mgaSetTexBorderColor(t, tObj->BorderColor);
+      mgaSetTexBorderColor(t, tObj->BorderColor.f);
       break;
 
    case GL_TEXTURE_BASE_LEVEL:
diff --git a/src/mesa/drivers/dri/mga/mgatexmem.c b/src/mesa/drivers/dri/mga/mgatexmem.c
index 9a2d62b53b..47be6f3c18 100644
--- a/src/mesa/drivers/dri/mga/mgatexmem.c
+++ b/src/mesa/drivers/dri/mga/mgatexmem.c
@@ -137,7 +137,7 @@ static void mgaUploadSubImage( mgaContextPtr mmesa,
     * directly used by the hardware for texturing.
     */
 
-   texelBytes = texImage->TexFormat->TexelBytes;
+   texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
    length = texImage->Width * texImage->Height * texelBytes;
    if ( t->base.heap->heapId == MGA_CARD_HEAP ) {
       unsigned  tex_offset = 0;
diff --git a/src/mesa/drivers/dri/mga/mgatris.c b/src/mesa/drivers/dri/mga/mgatris.c
index b93a21c3ac..c1bcd4b853 100644
--- a/src/mesa/drivers/dri/mga/mgatris.c
+++ b/src/mesa/drivers/dri/mga/mgatris.c
@@ -397,7 +397,8 @@ do {						\
 
 #define LOCAL_VARS(n)					\
    mgaContextPtr mmesa = MGA_CONTEXT(ctx);		\
-   GLuint color[n], spec[n];				\
+   GLuint color[n] = { 0 };				\
+   GLuint spec[n] = { 0 };				\
    (void) color; (void) spec;
 
 
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/r128/r128_context.c b/src/mesa/drivers/dri/r128/r128_context.c
index f511a67bad..e389e1c87b 100644
--- a/src/mesa/drivers/dri/r128/r128_context.c
+++ b/src/mesa/drivers/dri/r128/r128_context.c
@@ -68,9 +68,9 @@ int R128_DEBUG = 0;
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_env_add",            NULL },
@@ -101,11 +101,11 @@ static const struct dri_debug_control debug_control[] =
 /* Create the device specific context.
  */
 GLboolean r128CreateContext( const __GLcontextModes *glVisual,
-			     __DRIcontextPrivate *driContextPriv,
+			     __DRIcontext *driContextPriv,
                              void *sharedContextPrivate )
 {
    GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    struct dd_function_table functions;
    r128ContextPtr rmesa;
    r128ScreenPtr r128scrn;
@@ -274,7 +274,7 @@ GLboolean r128CreateContext( const __GLcontextModes *glVisual,
 
 /* Destroy the device specific context.
  */
-void r128DestroyContext( __DRIcontextPrivate *driContextPriv  )
+void r128DestroyContext( __DRIcontext *driContextPriv  )
 {
    r128ContextPtr rmesa = (r128ContextPtr) driContextPriv->driverPrivate;
 
@@ -325,9 +325,9 @@ void r128DestroyContext( __DRIcontextPrivate *driContextPriv  )
  * buffer `b'.
  */
 GLboolean
-r128MakeCurrent( __DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv )
+r128MakeCurrent( __DRIcontext *driContextPriv,
+                 __DRIdrawable *driDrawPriv,
+                 __DRIdrawable *driReadPriv )
 {
    if ( driContextPriv ) {
       GET_CURRENT_CONTEXT(ctx);
@@ -364,7 +364,7 @@ r128MakeCurrent( __DRIcontextPrivate *driContextPriv,
 /* Force the context `c' to be unbound from its buffer.
  */
 GLboolean
-r128UnbindContext( __DRIcontextPrivate *driContextPriv )
+r128UnbindContext( __DRIcontext *driContextPriv )
 {
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/r128/r128_context.h b/src/mesa/drivers/dri/r128/r128_context.h
index 0e10209a6a..65f845c115 100644
--- a/src/mesa/drivers/dri/r128/r128_context.h
+++ b/src/mesa/drivers/dri/r128/r128_context.h
@@ -186,9 +186,9 @@ struct r128_context {
 
    /* Mirrors of some DRI state
     */
-   __DRIcontextPrivate	*driContext;	/* DRI context */
-   __DRIscreenPrivate	*driScreen;	/* DRI screen */
-   __DRIdrawablePrivate	*driDrawable;	/* DRI drawable bound to this ctx */
+   __DRIcontext	*driContext;	/* DRI context */
+   __DRIscreen	*driScreen;	/* DRI screen */
+   __DRIdrawable	*driDrawable;	/* DRI drawable bound to this ctx */
 
    unsigned int lastStamp;	        /* mirror driDrawable->lastStamp */
 
@@ -225,16 +225,16 @@ struct r128_context {
 
 
 extern GLboolean r128CreateContext( const __GLcontextModes *glVisual,
-				    __DRIcontextPrivate *driContextPriv,
+				    __DRIcontext *driContextPriv,
                                     void *sharedContextPrivate );
 
-extern void r128DestroyContext( __DRIcontextPrivate * );
+extern void r128DestroyContext( __DRIcontext * );
 
-extern GLboolean r128MakeCurrent( __DRIcontextPrivate *driContextPriv,
-                                  __DRIdrawablePrivate *driDrawPriv,
-                                  __DRIdrawablePrivate *driReadPriv );
+extern GLboolean r128MakeCurrent( __DRIcontext *driContextPriv,
+                                  __DRIdrawable *driDrawPriv,
+                                  __DRIdrawable *driReadPriv );
 
-extern GLboolean r128UnbindContext( __DRIcontextPrivate *driContextPriv );
+extern GLboolean r128UnbindContext( __DRIcontext *driContextPriv );
 
 /* ================================================================
  * Debugging:
diff --git a/src/mesa/drivers/dri/r128/r128_ioctl.c b/src/mesa/drivers/dri/r128/r128_ioctl.c
index 84ac3d9f79..56758d971c 100644
--- a/src/mesa/drivers/dri/r128/r128_ioctl.c
+++ b/src/mesa/drivers/dri/r128/r128_ioctl.c
@@ -248,7 +248,7 @@ static int r128WaitForFrameCompletion( r128ContextPtr rmesa )
 
 /* Copy the back color buffer to the front color buffer.
  */
-void r128CopyBuffer( __DRIdrawablePrivate *dPriv )
+void r128CopyBuffer( __DRIdrawable *dPriv )
 {
    r128ContextPtr rmesa;
    GLint nbox, i, ret;
@@ -327,7 +327,7 @@ void r128CopyBuffer( __DRIdrawablePrivate *dPriv )
 #endif
 }
 
-void r128PageFlip( __DRIdrawablePrivate *dPriv )
+void r128PageFlip( __DRIdrawable *dPriv )
 {
    r128ContextPtr rmesa;
    GLint ret;
@@ -401,7 +401,7 @@ void r128PageFlip( __DRIdrawablePrivate *dPriv )
 static void r128Clear( GLcontext *ctx, GLbitfield mask )
 {
    r128ContextPtr rmesa = R128_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = rmesa->driDrawable;
+   __DRIdrawable *dPriv = rmesa->driDrawable;
    drm_r128_clear_t clear;
    GLuint flags = 0;
    GLint i;
diff --git a/src/mesa/drivers/dri/r128/r128_ioctl.h b/src/mesa/drivers/dri/r128/r128_ioctl.h
index 4b0c9cdc7f..84ace900ee 100644
--- a/src/mesa/drivers/dri/r128/r128_ioctl.h
+++ b/src/mesa/drivers/dri/r128/r128_ioctl.h
@@ -85,8 +85,8 @@ extern void r128ReadDepthSpanLocked( r128ContextPtr rmesa,
 extern void r128ReadDepthPixelsLocked( r128ContextPtr rmesa, GLuint n,
 				       const GLint x[], const GLint y[] );
 
-extern void r128CopyBuffer( __DRIdrawablePrivate *dPriv );
-extern void r128PageFlip( __DRIdrawablePrivate *dPriv );
+extern void r128CopyBuffer( __DRIdrawable *dPriv );
+extern void r128PageFlip( __DRIdrawable *dPriv );
 void r128WaitForVBlank( r128ContextPtr rmesa );
 
 extern void r128WaitForIdleLocked( r128ContextPtr rmesa );
diff --git a/src/mesa/drivers/dri/r128/r128_lock.c b/src/mesa/drivers/dri/r128/r128_lock.c
index 81488a2742..9bc3515b5a 100644
--- a/src/mesa/drivers/dri/r128/r128_lock.c
+++ b/src/mesa/drivers/dri/r128/r128_lock.c
@@ -68,8 +68,8 @@ r128UpdatePageFlipping( r128ContextPtr rmesa )
  */
 void r128GetLock( r128ContextPtr rmesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = rmesa->driDrawable;
-   __DRIscreenPrivate *sPriv = rmesa->driScreen;
+   __DRIdrawable *dPriv = rmesa->driDrawable;
+   __DRIscreen *sPriv = rmesa->driScreen;
    drm_r128_sarea_t *sarea = rmesa->sarea;
    int i;
 
diff --git a/src/mesa/drivers/dri/r128/r128_screen.c b/src/mesa/drivers/dri/r128/r128_screen.c
index f5bcc2f290..80b265811e 100644
--- a/src/mesa/drivers/dri/r128/r128_screen.c
+++ b/src/mesa/drivers/dri/r128/r128_screen.c
@@ -74,8 +74,6 @@ static const GLuint __driNConfigOptions = 4;
 static const GLuint __driNConfigOptions = 3;
 #endif
 
-extern const struct dri_extension card_extensions[];
-
 #if 1
 /* Including xf86PciInfo.h introduces a bunch of errors...
  */
@@ -93,7 +91,7 @@ extern const struct dri_extension card_extensions[];
 /* Create the device specific screen private data struct.
  */
 static r128ScreenPtr
-r128CreateScreen( __DRIscreenPrivate *sPriv )
+r128CreateScreen( __DRIscreen *sPriv )
 {
    r128ScreenPtr r128Screen;
    R128DRIPtr r128DRIPriv = (R128DRIPtr)sPriv->pDevPriv;
@@ -238,7 +236,7 @@ r128CreateScreen( __DRIscreenPrivate *sPriv )
 /* Destroy the device specific screen private data struct.
  */
 static void
-r128DestroyScreen( __DRIscreenPrivate *sPriv )
+r128DestroyScreen( __DRIscreen *sPriv )
 {
    r128ScreenPtr r128Screen = (r128ScreenPtr)sPriv->private;
 
@@ -264,8 +262,8 @@ r128DestroyScreen( __DRIscreenPrivate *sPriv )
  * data.
  */
 static GLboolean
-r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                  __DRIdrawablePrivate *driDrawPriv,
+r128CreateBuffer( __DRIscreen *driScrnPriv,
+                  __DRIdrawable *driDrawPriv,
                   const __GLcontextModes *mesaVis,
                   GLboolean isPixmap )
 {
@@ -284,7 +282,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->frontOffset, screen->frontPitch,
@@ -295,7 +293,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  NULL,
                                  screen->cpp,
                                  screen->backOffset, screen->backPitch,
@@ -306,7 +304,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
+            = driNewRenderbuffer(MESA_FORMAT_Z16,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -316,7 +314,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 24) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
+            = driNewRenderbuffer(MESA_FORMAT_S8_Z24,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -327,7 +325,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0 && !swStencil) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
+            = driNewRenderbuffer(MESA_FORMAT_S8,
                                  NULL,
                                  screen->cpp,
                                  screen->depthOffset, screen->depthPitch,
@@ -351,7 +349,7 @@ r128CreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-r128DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+r128DestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
@@ -359,7 +357,7 @@ r128DestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
 
 /* Copy the back color buffer to the front color buffer */
 static void
-r128SwapBuffers(__DRIdrawablePrivate *dPriv)
+r128SwapBuffers(__DRIdrawable *dPriv)
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
       r128ContextPtr rmesa;
@@ -386,7 +384,7 @@ r128SwapBuffers(__DRIdrawablePrivate *dPriv)
 /* Initialize the driver specific screen private data.
  */
 static GLboolean
-r128InitDriver( __DRIscreenPrivate *sPriv )
+r128InitDriver( __DRIscreen *sPriv )
 {
    sPriv->private = (void *) r128CreateScreen( sPriv );
 
@@ -399,7 +397,7 @@ r128InitDriver( __DRIscreenPrivate *sPriv )
 }
 
 static const __DRIconfig **
-r128FillInModes( __DRIscreenPrivate *psp,
+r128FillInModes( __DRIscreen *psp,
 		 unsigned pixel_bits, unsigned depth_bits,
 		 unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -480,7 +478,7 @@ r128FillInModes( __DRIscreenPrivate *psp,
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-r128InitScreen(__DRIscreenPrivate *psp)
+r128InitScreen(__DRIscreen *psp)
 {
    static const __DRIversion ddx_expected = { 4, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
@@ -493,18 +491,6 @@ r128InitScreen(__DRIscreenPrivate *psp)
 				      &psp->drm_version, & drm_expected ) )
       return NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-
    if (!r128InitDriver(psp))
        return NULL;
 
@@ -531,3 +517,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/r128/r128_screen.h b/src/mesa/drivers/dri/r128/r128_screen.h
index e2fa1677c9..8d450adff3 100644
--- a/src/mesa/drivers/dri/r128/r128_screen.h
+++ b/src/mesa/drivers/dri/r128/r128_screen.h
@@ -71,7 +71,7 @@ typedef struct {
 
    drmBufMapPtr buffers;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
    unsigned int sarea_priv_offset;
 
    /* Configuration cache with default values for all contexts */
diff --git a/src/mesa/drivers/dri/r128/r128_span.c b/src/mesa/drivers/dri/r128/r128_span.c
index dd177e0def..0413e5b4f1 100644
--- a/src/mesa/drivers/dri/r128/r128_span.c
+++ b/src/mesa/drivers/dri/r128/r128_span.c
@@ -50,8 +50,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define LOCAL_VARS							\
    r128ContextPtr rmesa = R128_CONTEXT(ctx);				\
-   __DRIscreenPrivate *sPriv = rmesa->driScreen;			\
-   __DRIdrawablePrivate *dPriv = rmesa->driDrawable;			\
+   __DRIscreen *sPriv = rmesa->driScreen;			\
+   __DRIdrawable *dPriv = rmesa->driDrawable;			\
    driRenderbuffer *drb = (driRenderbuffer *) rb;			\
    GLuint height = dPriv->h;						\
    GLuint p;								\
@@ -60,8 +60,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define LOCAL_DEPTH_VARS						\
    r128ContextPtr rmesa = R128_CONTEXT(ctx);				\
    r128ScreenPtr r128scrn = rmesa->r128Screen;				\
-   __DRIscreenPrivate *sPriv = rmesa->driScreen;			\
-   __DRIdrawablePrivate *dPriv = rmesa->driDrawable;			\
+   __DRIscreen *sPriv = rmesa->driScreen;			\
+   __DRIdrawable *dPriv = rmesa->driDrawable;			\
    GLuint height = dPriv->h;						\
    (void) r128scrn; (void) sPriv; (void) height
 
@@ -433,21 +433,19 @@ void r128DDInitSpanFuncs( GLcontext *ctx )
 void
 r128SetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
 {
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         r128InitPointers_RGB565(&drb->Base);
-      }
-      else {
-         r128InitPointers_ARGB8888(&drb->Base);
-      }
+   if (drb->Base.Format == MESA_FORMAT_RGB565) {
+      r128InitPointers_RGB565(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (drb->Base.Format == MESA_FORMAT_ARGB8888) {
+      r128InitPointers_ARGB8888(&drb->Base);
+   }
+   else if (drb->Base.Format == MESA_FORMAT_Z16) {
       r128InitDepthPointers_z16(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
+   else if (drb->Base.Format == MESA_FORMAT_S8_Z24) {
       r128InitDepthPointers_z24_s8(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
+   else if (drb->Base.Format == MESA_FORMAT_S8) {
       radeonInitStencilPointers_z24_s8(&drb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/r128/r128_state.c b/src/mesa/drivers/dri/r128/r128_state.c
index 4ae7bf5b97..2254a7a4ff 100644
--- a/src/mesa/drivers/dri/r128/r128_state.c
+++ b/src/mesa/drivers/dri/r128/r128_state.c
@@ -572,7 +572,7 @@ static void r128UpdateClipping( GLcontext *ctx )
    r128ContextPtr rmesa = R128_CONTEXT(ctx);
 
    if ( rmesa->driDrawable ) {
-      __DRIdrawablePrivate *drawable = rmesa->driDrawable;
+      __DRIdrawable *drawable = rmesa->driDrawable;
       int x1 = 0;
       int y1 = 0;
       int x2 = drawable->w - 1;
@@ -702,10 +702,10 @@ static void r128UpdateMasks( GLcontext *ctx )
    r128ContextPtr rmesa = R128_CONTEXT(ctx);
 
    GLuint mask = r128PackColor( rmesa->r128Screen->cpp,
-				ctx->Color.ColorMask[RCOMP],
-				ctx->Color.ColorMask[GCOMP],
-				ctx->Color.ColorMask[BCOMP],
-				ctx->Color.ColorMask[ACOMP] );
+				ctx->Color.ColorMask[0][RCOMP],
+				ctx->Color.ColorMask[0][GCOMP],
+				ctx->Color.ColorMask[0][BCOMP],
+				ctx->Color.ColorMask[0][ACOMP] );
 
    if ( rmesa->setup.plane_3d_mask_c != mask ) {
       rmesa->setup.plane_3d_mask_c = mask;
diff --git a/src/mesa/drivers/dri/r128/r128_tex.c b/src/mesa/drivers/dri/r128/r128_tex.c
index 0920270d7b..f1be7cc1c4 100644
--- a/src/mesa/drivers/dri/r128/r128_tex.c
+++ b/src/mesa/drivers/dri/r128/r128_tex.c
@@ -44,7 +44,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/simple_list.h"
 #include "main/enums.h"
 #include "main/texstore.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/imports.h"
@@ -170,7 +169,7 @@ static r128TexObjPtr r128AllocTexObj( struct gl_texture_object *texObj )
 
       r128SetTexWrap( t, texObj->WrapS, texObj->WrapT );
       r128SetTexFilter( t, texObj->MinFilter, texObj->MagFilter );
-      r128SetTexBorderColor( t, texObj->BorderColor );
+      r128SetTexBorderColor( t, texObj->BorderColor.f );
    }
 
    return t;
@@ -178,7 +177,7 @@ static r128TexObjPtr r128AllocTexObj( struct gl_texture_object *texObj )
 
 
 /* Called by the _mesa_store_teximage[123]d() functions. */
-static const struct gl_texture_format *
+static gl_format
 r128ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
                          GLenum format, GLenum type )
 {
@@ -282,13 +281,13 @@ r128ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_APPLE ||
           type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    default:
       _mesa_problem( ctx, "unexpected format in %s", __FUNCTION__ );
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 }
 
@@ -536,7 +535,7 @@ static void r128TexParameter( GLcontext *ctx, GLenum target,
 
    case GL_TEXTURE_BORDER_COLOR:
       if ( t->base.bound ) FLUSH_BATCH( rmesa );
-      r128SetTexBorderColor( t, tObj->BorderColor );
+      r128SetTexBorderColor( t, tObj->BorderColor.f );
       break;
 
    case GL_TEXTURE_BASE_LEVEL:
diff --git a/src/mesa/drivers/dri/r128/r128_texmem.c b/src/mesa/drivers/dri/r128/r128_texmem.c
index 111fe1fd74..4ddcb86bcd 100644
--- a/src/mesa/drivers/dri/r128/r128_texmem.c
+++ b/src/mesa/drivers/dri/r128/r128_texmem.c
@@ -41,7 +41,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/context.h"
 #include "main/macros.h"
 #include "main/simple_list.h"
-#include "main/texformat.h"
 #include "main/imports.h"
 
 #define TEX_0	1
@@ -95,7 +94,7 @@ static void uploadSubImage( r128ContextPtr rmesa, r128TexObjPtr t,
    if ( !image )
       return;
 
-   switch ( image->TexFormat->TexelBytes ) {
+   switch ( _mesa_get_format_bytes(image->TexFormat) ) {
    case 1: texelsPerDword = 4; break;
    case 2: texelsPerDword = 2; break;
    case 4: texelsPerDword = 1; break;
@@ -215,9 +214,11 @@ static void uploadSubImage( r128ContextPtr rmesa, r128TexObjPtr t,
 
       /* Copy the next chunck of the texture image into the blit buffer */
       {
+         const GLuint texelBytes =
+            _mesa_get_format_bytes(image->TexFormat);
          const GLubyte *src = (const GLubyte *) image->Data +
-            (y * image->Width + x) * image->TexFormat->TexelBytes;
-         const GLuint bytes = width * height * image->TexFormat->TexelBytes;
+            (y * image->Width + x) * texelBytes;            
+         const GLuint bytes = width * height * texelBytes;
          memcpy(dst, src, bytes);
       }
 
diff --git a/src/mesa/drivers/dri/r128/r128_texstate.c b/src/mesa/drivers/dri/r128/r128_texstate.c
index a9c9568003..cb2b5f9536 100644
--- a/src/mesa/drivers/dri/r128/r128_texstate.c
+++ b/src/mesa/drivers/dri/r128/r128_texstate.c
@@ -36,7 +36,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 
 #include "r128_context.h"
 #include "r128_state.h"
@@ -61,7 +60,7 @@ static void r128SetTexImages( r128ContextPtr rmesa,
    if ( R128_DEBUG & DEBUG_VERBOSE_API )
       fprintf( stderr, "%s( %p )\n", __FUNCTION__, (void *) tObj );
 
-   switch (baseImage->TexFormat->MesaFormat) {
+   switch (baseImage->TexFormat) {
    case MESA_FORMAT_ARGB8888:
    case MESA_FORMAT_ARGB8888_REV:
       t->textureFormat = R128_DATATYPE_ARGB8888;
@@ -123,7 +122,7 @@ static void r128SetTexImages( r128ContextPtr rmesa,
 
       totalSize += (tObj->Image[0][i]->Height *
 		    tObj->Image[0][i]->Width *
-		    tObj->Image[0][i]->TexFormat->TexelBytes);
+		    _mesa_get_format_bytes(tObj->Image[0][i]->TexFormat));
 
       /* Offsets must be 32-byte aligned for host data blits and tiling */
       totalSize = (totalSize + 31) & ~31;
diff --git a/src/mesa/drivers/dri/r128/r128_tris.c b/src/mesa/drivers/dri/r128/r128_tris.c
index 5b91271d74..86d4717b05 100644
--- a/src/mesa/drivers/dri/r128/r128_tris.c
+++ b/src/mesa/drivers/dri/r128/r128_tris.c
@@ -216,7 +216,8 @@ do {							\
 
 #define LOCAL_VARS(n)						\
    r128ContextPtr rmesa = R128_CONTEXT(ctx);			\
-   GLuint color[n], spec[n];					\
+   GLuint color[n] = { 0 };					\
+   GLuint spec[n] = { 0 };					\
    GLuint coloroffset = rmesa->coloroffset;			\
    GLuint specoffset = rmesa->specoffset;			\
    GLboolean havespec = (rmesa->specoffset != 0);		\
@@ -650,12 +651,12 @@ static void r128RenderStart( GLcontext *ctx )
    }
 
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(rmesa->tmu_source[0]) )) {
-      if ( VB->TexCoordPtr[rmesa->tmu_source[0]]->size > 2 )
+      if ( VB->AttribPtr[_TNL_ATTRIB_TEX0 + rmesa->tmu_source[0]]->size > 2 )
 	 fallback_projtex = GL_TRUE;
       EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_2F, R128_CCE_VC_FRMT_S_T, 8 );
    }
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(rmesa->tmu_source[1]) )) {
-      if ( VB->TexCoordPtr[rmesa->tmu_source[1]]->size > 2 )
+      if ( VB->AttribPtr[_TNL_ATTRIB_TEX0 + rmesa->tmu_source[1]]->size > 2 )
 	 fallback_projtex = GL_TRUE;
       EMIT_ATTR( _TNL_ATTRIB_TEX1, EMIT_2F, R128_CCE_VC_FRMT_S2_T2, 8 );
    }
diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile
index fbce70c37b..8212dc1203 100644
--- a/src/mesa/drivers/dri/r200/Makefile
+++ b/src/mesa/drivers/dri/r200/Makefile
@@ -14,7 +14,7 @@ EGL_SOURCES = server/radeon_egl.c
 endif
 
 ifeq ($(RADEON_LDFLAGS),)
-CS_SOURCES = radeon_cs_space_drm.c
+CS_SOURCES = radeon_cs_space_drm.c radeon_bo.c radeon_cs.c
 endif
 
 RADEON_COMMON_SOURCES = \
@@ -66,4 +66,3 @@ include ../Makefile.template
 
 #INCLUDES += -I../radeon/server
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c
index 3ddb5bf7d6..f34e319222 100644
--- a/src/mesa/drivers/dri/r200/r200_context.c
+++ b/src/mesa/drivers/dri/r200/r200_context.c
@@ -75,7 +75,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_NV_vertex_program
 #define need_GL_ARB_point_parameters
 #define need_GL_EXT_framebuffer_object
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 #define DRIVER_DATE	"20060602"
 
@@ -115,7 +115,7 @@ static const GLubyte *r200GetString( GLcontext *ctx, GLenum name )
 
 /* Extension strings exported by the R200 driver.
  */
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_occlusion_query",		   GL_ARB_occlusion_query_functions},
@@ -146,31 +146,31 @@ const struct dri_extension card_extensions[] =
     { NULL,                                NULL }
 };
 
-const struct dri_extension blend_extensions[] = {
+static const struct dri_extension blend_extensions[] = {
     { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
     { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
     { NULL,                                NULL }
 };
 
-const struct dri_extension ARB_vp_extension[] = {
+static const struct dri_extension ARB_vp_extension[] = {
     { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions }
 };
 
-const struct dri_extension NV_vp_extension[] = {
+static const struct dri_extension NV_vp_extension[] = {
     { "GL_NV_vertex_program",              GL_NV_vertex_program_functions }
 };
 
-const struct dri_extension ATI_fs_extension[] = {
+static const struct dri_extension ATI_fs_extension[] = {
     { "GL_ATI_fragment_shader",            GL_ATI_fragment_shader_functions }
 };
 
-const struct dri_extension point_extensions[] = {
+static const struct dri_extension point_extensions[] = {
     { "GL_ARB_point_sprite",               NULL },
     { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
     { NULL,                                NULL }
 };
 
-const struct dri_extension mm_extensions[] = {
+static const struct dri_extension mm_extensions[] = {
   { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions },
   { NULL, NULL }
 };
@@ -274,10 +274,10 @@ static void r200_init_vtbl(radeonContextPtr radeon)
 /* Create the device specific rendering context.
  */
 GLboolean r200CreateContext( const __GLcontextModes *glVisual,
-			     __DRIcontextPrivate *driContextPriv,
+			     __DRIcontext *driContextPriv,
 			     void *sharedContextPrivate)
 {
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    radeonScreenPtr screen = (radeonScreenPtr)(sPriv->private);
    struct dd_function_table functions;
    r200ContextPtr rmesa;
@@ -325,9 +325,9 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
    _mesa_init_driver_functions(&functions);
    r200InitDriverFuncs(&functions);
    r200InitIoctlFuncs(&functions);
-   r200InitStateFuncs(&functions, screen->kernel_mm);
+   r200InitStateFuncs(&functions);
    r200InitTextureFuncs(&functions);
-   r200InitShaderFuncs(&functions); 
+   r200InitShaderFuncs(&functions);
    radeonInitQueryObjFunctions(&functions);
 
    if (!radeonInitContext(&rmesa->radeon, &functions,
@@ -496,7 +496,7 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
 }
 
 
-void r200DestroyContext( __DRIcontextPrivate *driContextPriv )
+void r200DestroyContext( __DRIcontext *driContextPriv )
 {
 	int i;
 	r200ContextPtr rmesa = (r200ContextPtr)driContextPriv->driverPrivate;
diff --git a/src/mesa/drivers/dri/r200/r200_context.h b/src/mesa/drivers/dri/r200/r200_context.h
index 246f98c6dc..17e4d8962e 100644
--- a/src/mesa/drivers/dri/r200/r200_context.h
+++ b/src/mesa/drivers/dri/r200/r200_context.h
@@ -636,14 +636,14 @@ struct r200_context {
 #define R200_CONTEXT(ctx)		((r200ContextPtr)(ctx->DriverCtx))
 
 
-extern void r200DestroyContext( __DRIcontextPrivate *driContextPriv );
+extern void r200DestroyContext( __DRIcontext *driContextPriv );
 extern GLboolean r200CreateContext( const __GLcontextModes *glVisual,
-				    __DRIcontextPrivate *driContextPriv,
+				    __DRIcontext *driContextPriv,
 				    void *sharedContextPrivate);
-extern GLboolean r200MakeCurrent( __DRIcontextPrivate *driContextPriv,
-				  __DRIdrawablePrivate *driDrawPriv,
-				  __DRIdrawablePrivate *driReadPriv );
-extern GLboolean r200UnbindContext( __DRIcontextPrivate *driContextPriv );
+extern GLboolean r200MakeCurrent( __DRIcontext *driContextPriv,
+				  __DRIdrawable *driDrawPriv,
+				  __DRIdrawable *driReadPriv );
+extern GLboolean r200UnbindContext( __DRIcontext *driContextPriv );
 
 /* ================================================================
  * Debugging:
diff --git a/src/mesa/drivers/dri/r200/r200_ioctl.c b/src/mesa/drivers/dri/r200/r200_ioctl.c
index b238adb972..66c5d3655a 100644
--- a/src/mesa/drivers/dri/r200/r200_ioctl.c
+++ b/src/mesa/drivers/dri/r200/r200_ioctl.c
@@ -61,7 +61,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 static void r200KernelClear(GLcontext *ctx, GLuint flags)
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLint cx, cy, cw, ch, ret;
    GLuint i;
 
@@ -185,7 +185,7 @@ static void r200KernelClear(GLcontext *ctx, GLuint flags)
 static void r200Clear( GLcontext *ctx, GLbitfield mask )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLuint flags = 0;
    GLuint color_mask = 0;
    GLuint orig_mask = mask;
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index 383a0c4b0d..249c0bbc11 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -90,12 +90,14 @@ static void r200_emit_vecfog(GLcontext *ctx, struct radeon_aos *aos,
 	aos->components = size;
 	aos->count = count;
 
+	radeon_bo_map(aos->bo, 1);
 	out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
 	for (i = 0; i < count; i++) {
 	  out[0] = r200ComputeFogBlendFactor( ctx, *(GLfloat *)data );
 	  out++;
 	  data += stride;
 	}
+	radeon_bo_unmap(aos->bo);
 }
 
 /* Emit any changed arrays to new GART memory, re-emit a packet to
diff --git a/src/mesa/drivers/dri/r200/r200_pixel.c b/src/mesa/drivers/dri/r200/r200_pixel.c
index 95773871e0..bfb7e2a2ed 100644
--- a/src/mesa/drivers/dri/r200/r200_pixel.c
+++ b/src/mesa/drivers/dri/r200/r200_pixel.c
@@ -88,10 +88,10 @@ check_color_per_fragment_ops( const GLcontext *ctx )
 		    ctx->Fog.Enabled ||
 		    ctx->Scissor.Enabled ||
 		    ctx->Stencil._Enabled ||
-		    !ctx->Color.ColorMask[0] ||
-		    !ctx->Color.ColorMask[1] ||
-		    !ctx->Color.ColorMask[2] ||
-		    !ctx->Color.ColorMask[3] ||
+		    !ctx->Color.ColorMask[0][0] ||
+		    !ctx->Color.ColorMask[0][1] ||
+		    !ctx->Color.ColorMask[0][2] ||
+		    !ctx->Color.ColorMask[0][3] ||
 		    ctx->Color.ColorLogicOpEnabled ||
 		    ctx->Texture._EnabledUnits
            ) &&
@@ -214,7 +214,7 @@ r200TryReadPixels( GLcontext *ctx,
    }
 
    {
-      __DRIdrawablePrivate *dPriv = rmesa->radeon.dri.drawable;
+      __DRIdrawable *dPriv = rmesa->radeon.dri.drawable;
       driRenderbuffer *drb = (driRenderbuffer *) ctx->ReadBuffer->_ColorReadBuffer;
       int nbox = dPriv->numClipRects;
       int src_offset = drb->offset
@@ -298,7 +298,7 @@ static void do_draw_pix( GLcontext *ctx,
 
 #if 0
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    drm_clip_rect_t *box = dPriv->pClipRects;
    struct gl_renderbuffer *rb = ctx->ReadBuffer->_ColorDrawBuffers[0];
    driRenderbuffer *drb = (driRenderbuffer *) rb;
@@ -400,10 +400,10 @@ r200TryDrawPixels( GLcontext *ctx,
    case GL_RGBA:
    case GL_BGRA:
       planemask = radeonPackColor(cpp,
-				ctx->Color.ColorMask[RCOMP],
-				ctx->Color.ColorMask[GCOMP],
-				ctx->Color.ColorMask[BCOMP],
-				ctx->Color.ColorMask[ACOMP]);
+				ctx->Color.ColorMask[0][RCOMP],
+				ctx->Color.ColorMask[0][GCOMP],
+				ctx->Color.ColorMask[0][BCOMP],
+				ctx->Color.ColorMask[0][ACOMP]);
 
       if (cpp == 2)
 	 planemask |= planemask << 16;
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index d28e96d9d9..7fe482fe15 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -721,10 +721,10 @@ static void r200ColorMask( GLcontext *ctx,
    if (!rrb)
      return;
    mask = radeonPackColor( rrb->cpp,
-			   ctx->Color.ColorMask[RCOMP],
-			   ctx->Color.ColorMask[GCOMP],
-			   ctx->Color.ColorMask[BCOMP],
-			   ctx->Color.ColorMask[ACOMP] );
+			   ctx->Color.ColorMask[0][RCOMP],
+			   ctx->Color.ColorMask[0][GCOMP],
+			   ctx->Color.ColorMask[0][BCOMP],
+			   ctx->Color.ColorMask[0][ACOMP] );
 
 
    if (!(r && g && b && a))
@@ -1585,7 +1585,7 @@ static void r200ClearStencil( GLcontext *ctx, GLint s )
 void r200UpdateWindow( GLcontext *ctx )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLfloat xoffset = dPriv ? (GLfloat) dPriv->x : 0;
    GLfloat yoffset = dPriv ? (GLfloat) dPriv->y + dPriv->h : 0;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1665,7 +1665,7 @@ static void r200DepthRange( GLcontext *ctx, GLclampd nearval,
 void r200UpdateViewportOffset( GLcontext *ctx )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLfloat xoffset = (GLfloat)dPriv->x;
    GLfloat yoffset = (GLfloat)dPriv->y + dPriv->h;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -2476,7 +2476,7 @@ static void r200PolygonStipple( GLcontext *ctx, const GLubyte *mask )
 }
 /* Initialize the driver's state functions.
  */
-void r200InitStateFuncs( struct dd_function_table *functions, GLboolean dri2 )
+void r200InitStateFuncs( struct dd_function_table *functions )
 {
    functions->UpdateState		= r200InvalidateState;
    functions->LightingSpaceChange	= r200LightingSpaceChange;
@@ -2510,10 +2510,7 @@ void r200InitStateFuncs( struct dd_function_table *functions, GLboolean dri2 )
    functions->LogicOpcode		= r200LogicOpCode;
    functions->PolygonMode		= r200PolygonMode;
    functions->PolygonOffset		= r200PolygonOffset;
-   if (dri2)
-      functions->PolygonStipple		= r200PolygonStipple;
-   else
-      functions->PolygonStipple		= radeonPolygonStipplePreKMS;
+   functions->PolygonStipple		= r200PolygonStipple;
    functions->PointParameterfv		= r200PointParameter;
    functions->PointSize			= r200PointSize;
    functions->RenderMode		= r200RenderMode;
diff --git a/src/mesa/drivers/dri/r200/r200_state.h b/src/mesa/drivers/dri/r200/r200_state.h
index 9c62f0a644..7b9b0c106a 100644
--- a/src/mesa/drivers/dri/r200/r200_state.h
+++ b/src/mesa/drivers/dri/r200/r200_state.h
@@ -38,7 +38,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_context.h"
 
 extern void r200InitState( r200ContextPtr rmesa );
-extern void r200InitStateFuncs( struct dd_function_table *functions, GLboolean dri2 );
+extern void r200InitStateFuncs( struct dd_function_table *functions );
 extern void r200InitTnlFuncs( GLcontext *ctx );
 
 extern void r200UpdateMaterial( GLcontext *ctx );
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index 7697306d88..6c5a0b79ee 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -529,16 +529,18 @@ static void ctx_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom)
    atom->cmd[CTX_RB3D_CNTL] &= ~(0xf << 10);
    if (rrb->cpp == 4)
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB8888;
-   else switch (rrb->base._ActualFormat) {
-   case GL_RGB5:
+   else switch (rrb->base.Format) {
+   case MESA_FORMAT_RGB565:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_RGB565;
 	break;
-   case GL_RGBA4:
+   case MESA_FORMAT_ARGB4444:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB4444;
 	break;
-   case GL_RGB5_A1:
+   case MESA_FORMAT_ARGB1555:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB1555;
 	break;
+   default:
+	_mesa_problem(ctx, "Unexpected format in ctx_emit_cs");
    }
 
    cbpitch = (rrb->pitch / rrb->cpp);
@@ -638,7 +640,7 @@ static void tex_emit(GLcontext *ctx, struct radeon_state_atom *atom)
    OUT_BATCH_TABLE(atom->cmd, 10);
 
    if (t && t->mt && !t->image_override) {
-     OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+     OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 		  RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
    } else if (!t) {
      /* workaround for old CS mechanism */
@@ -885,10 +887,8 @@ void r200InitState( r200ContextPtr rmesa )
          }
       }
    }
-   /* polygon stipple is done with irq for non-kms */
-   if (rmesa->radeon.radeonScreen->kernel_mm) {
-       ALLOC_STATE( stp, always, STP_STATE_SIZE, "STP/stp", 0 );
-   }
+
+   ALLOC_STATE( stp, always, STP_STATE_SIZE, "STP/stp", 0 );
 
    for (i = 0; i < 6; i++)
       if (rmesa->radeon.radeonScreen->kernel_mm)
@@ -1120,12 +1120,11 @@ void r200InitState( r200ContextPtr rmesa )
    rmesa->hw.sci.cmd[SCI_CMD_1] = CP_PACKET0(R200_RE_TOP_LEFT, 0);
    rmesa->hw.sci.cmd[SCI_CMD_2] = CP_PACKET0(R200_RE_WIDTH_HEIGHT, 0);
 
-   if (rmesa->radeon.radeonScreen->kernel_mm) {
-
-	rmesa->hw.stp.cmd[STP_CMD_0] = CP_PACKET0(RADEON_RE_STIPPLE_ADDR, 0);
-	rmesa->hw.stp.cmd[STP_DATA_0] = 0;
-	rmesa->hw.stp.cmd[STP_CMD_1] = CP_PACKET0_ONE(RADEON_RE_STIPPLE_DATA, 31);
+   rmesa->hw.stp.cmd[STP_CMD_0] = CP_PACKET0(RADEON_RE_STIPPLE_ADDR, 0);
+   rmesa->hw.stp.cmd[STP_DATA_0] = 0;
+   rmesa->hw.stp.cmd[STP_CMD_1] = CP_PACKET0_ONE(RADEON_RE_STIPPLE_DATA, 31);
 
+   if (rmesa->radeon.radeonScreen->kernel_mm) {
         rmesa->hw.mtl[0].emit = mtl_emit;
         rmesa->hw.mtl[1].emit = mtl_emit;
 
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index 240fb45078..4596912ddc 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -168,7 +168,7 @@ static void r200SetVertexFormat( GLcontext *ctx )
 
       for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    GLuint sz = VB->TexCoordPtr[i]->size;
+	    GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
 	    fmt_1 |= sz << (3 * i);
 	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_1F + sz - 1, 0 );
@@ -297,7 +297,7 @@ void r200_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
    radeonEmitState(&rmesa->radeon);
    r200EmitVertexAOS( rmesa,
 		      rmesa->radeon.swtcl.vertex_size,
-		      first_elem(&rmesa->radeon.dma.reserved)->bo,
+		      rmesa->radeon.swtcl.bo,
 		      current_offset);
 
 
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index c702910ef2..e7d48a7f29 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -509,25 +509,26 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
 	 prog to a not enabled output however, so just don't mess with it.
 	 We only need to change compsel. */
       GLuint out_compsel = 0;
-      GLuint vp_out = rmesa->curr_vp_hw->mesa_program.Base.OutputsWritten;
+      const GLbitfield64 vp_out =
+	 rmesa->curr_vp_hw->mesa_program.Base.OutputsWritten;
 
       vimap_rev = &rmesa->curr_vp_hw->inputmap_rev[0];
-      assert(vp_out & (1 << VERT_RESULT_HPOS));
+      assert(vp_out & BITFIELD64_BIT(VERT_RESULT_HPOS));
       out_compsel = R200_OUTPUT_XYZW;
-      if (vp_out & (1 << VERT_RESULT_COL0)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_COL0)) {
 	 out_compsel |= R200_OUTPUT_COLOR_0;
       }
-      if (vp_out & (1 << VERT_RESULT_COL1)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_COL1)) {
 	 out_compsel |= R200_OUTPUT_COLOR_1;
       }
-      if (vp_out & (1 << VERT_RESULT_FOGC)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_FOGC)) {
          out_compsel |= R200_OUTPUT_DISCRETE_FOG;
       }
-      if (vp_out & (1 << VERT_RESULT_PSIZ)) {
+      if (vp_out & BITFIELD64_BIT(VERT_RESULT_PSIZ)) {
 	 out_compsel |= R200_OUTPUT_PT_SIZE;
       }
       for (i = VERT_RESULT_TEX0; i < VERT_RESULT_TEX6; i++) {
-	 if (vp_out & (1 << i)) {
+	 if (vp_out & BITFIELD64_BIT(i)) {
 	    out_compsel |= R200_OUTPUT_TEX_0 << (i - VERT_RESULT_TEX0);
 	 }
       }
diff --git a/src/mesa/drivers/dri/r200/r200_tex.c b/src/mesa/drivers/dri/r200/r200_tex.c
index 36d9e37d87..5b87ba6ccd 100644
--- a/src/mesa/drivers/dri/r200/r200_tex.c
+++ b/src/mesa/drivers/dri/r200/r200_tex.c
@@ -38,7 +38,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/enums.h"
 #include "main/image.h"
 #include "main/simple_list.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
@@ -379,23 +378,14 @@ static void r200TexParameter( GLcontext *ctx, GLenum target,
       break;
 
    case GL_TEXTURE_BORDER_COLOR:
-      r200SetTexBorderColor( t, texObj->BorderColor );
+      r200SetTexBorderColor( t, texObj->BorderColor.f );
       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.  Since there's no LOD clamping,
-       * we just have to rely on loading the right subset of mipmap levels
-       * to simulate a clamped LOD.
-       */
-      if (t->mt) {
-         radeon_miptree_unreference(t->mt);
-	 t->mt = 0;
-	 t->validated = GL_FALSE;
-      }
+      t->validated = GL_FALSE;
       break;
 
    default:
@@ -414,7 +404,7 @@ static void r200DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 	      (void *)texObj,
 	      _mesa_lookup_enum_by_nr(texObj->Target));
    }
-   
+
    if (rmesa) {
       int i;
       radeon_firevertices(&rmesa->radeon);
@@ -426,11 +416,9 @@ static void r200DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 	 }
       }      
    }
-   
-   if (t->mt) {
-      radeon_miptree_unreference(t->mt);
-      t->mt = 0;
-   }
+
+   radeon_miptree_unreference(&t->mt);
+
    _mesa_delete_texture_object(ctx, texObj);
 }
 
@@ -482,7 +470,7 @@ static struct gl_texture_object *r200NewTextureObject(GLcontext * ctx,
    r200SetTexWrap( t, t->base.WrapS, t->base.WrapT, t->base.WrapR );
    r200SetTexMaxAnisotropy( t, t->base.MaxAnisotropy );
    r200SetTexFilter(t, t->base.MinFilter, t->base.MagFilter);
-   r200SetTexBorderColor(t, t->base.BorderColor);
+   r200SetTexBorderColor(t, t->base.BorderColor.f);
 
    return &t->base;
 }
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index c94834752e..e2f9cf0ea8 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -36,7 +36,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/enums.h"
@@ -798,24 +797,13 @@ void r200SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
     	    return;
     	}
 
-	radeon_update_renderbuffers(pDRICtx, dPriv);
-	/* back & depth buffer are useless free them right away */
-	rb = (void*)rfb->base.Attachment[BUFFER_DEPTH].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-        rb->bo = NULL;
-	}
-	rb = (void*)rfb->base.Attachment[BUFFER_BACK_LEFT].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-		rb->bo = NULL;
-	}
+	radeon_update_renderbuffers(pDRICtx, dPriv, GL_TRUE);
 	rb = rfb->color_rb[0];
 	if (rb->bo == NULL) {
 		/* Failed to BO for the buffer */
 		return;
 	}
-	
+
 	_mesa_lock_texture(radeon->glCtx, texObj);
 	if (t->bo) {
 		radeon_bo_unref(t->bo);
@@ -825,20 +813,14 @@ void r200SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
-	texImage->TexFormat = radeonChooseTextureFormat(radeon->glCtx,
-							internalFormat,
-							type, format, 0);
+
 	rImage->bo = rb->bo;
 	radeon_bo_ref(rImage->bo);
 	t->bo = rb->bo;
@@ -1426,10 +1408,9 @@ void set_re_cntl_d3d( GLcontext *ctx, int unit, GLboolean use_d3d )
  */
 static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
 {
-   int firstlevel = t->mt ? t->mt->firstLevel : 0;
-   const struct gl_texture_image *firstImage = t->base.Image[0][firstlevel];
+   const struct gl_texture_image *firstImage = t->base.Image[0][t->minLod];
    GLint log2Width, log2Height, log2Depth, texelBytes;
-   
+
    if ( t->bo ) {
        return;
    }
@@ -1437,11 +1418,11 @@ static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
    log2Width  = firstImage->WidthLog2;
    log2Height = firstImage->HeightLog2;
    log2Depth  = firstImage->DepthLog2;
-   texelBytes = firstImage->TexFormat->TexelBytes;
+   texelBytes = _mesa_get_format_bytes(firstImage->TexFormat);
 
 
    if (!t->image_override) {
-      if (VALID_FORMAT(firstImage->TexFormat->MesaFormat)) {
+      if (VALID_FORMAT(firstImage->TexFormat)) {
 	 const struct tx_table *table = _mesa_little_endian() ? tx_table_le :
 	    tx_table_be;
 	 
@@ -1449,17 +1430,17 @@ static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
 			     R200_TXFORMAT_ALPHA_IN_MAP);
 	 t->pp_txfilter &= ~R200_YUV_TO_RGB;
 	 
-	 t->pp_txformat |= table[ firstImage->TexFormat->MesaFormat ].format;
-	 t->pp_txfilter |= table[ firstImage->TexFormat->MesaFormat ].filter;
+	 t->pp_txformat |= table[ firstImage->TexFormat ].format;
+	 t->pp_txfilter |= table[ firstImage->TexFormat ].filter;
       } else {
 	 _mesa_problem(NULL, "unexpected texture format in %s",
 		       __FUNCTION__);
 	 return;
       }
    }
-   
+
    t->pp_txfilter &= ~R200_MAX_MIP_LEVEL_MASK;
-   t->pp_txfilter |= (t->mt->lastLevel - t->mt->firstLevel) << R200_MAX_MIP_LEVEL_SHIFT;
+   t->pp_txfilter |= (t->maxLod - t->minLod) << R200_MAX_MIP_LEVEL_SHIFT;
 	
    t->pp_txformat &= ~(R200_TXFORMAT_WIDTH_MASK |
 		       R200_TXFORMAT_HEIGHT_MASK |
@@ -1504,7 +1485,7 @@ static void setup_hardware_state(r200ContextPtr rmesa, radeonTexObj *t)
 		   | ((firstImage->Height - 1) << R200_PP_TX_HEIGHTMASK_SHIFT));
 
    if ( !t->image_override ) {
-      if (firstImage->IsCompressed)
+      if (_mesa_is_format_compressed(firstImage->TexFormat))
          t->pp_txpitch = (firstImage->Width + 63) & ~(63);
       else
          t->pp_txpitch = ((firstImage->Width * texelBytes) + 63) & ~(63);
diff --git a/src/mesa/drivers/dri/r200/radeon_bo.c b/src/mesa/drivers/dri/r200/radeon_bo.c
new file mode 120000
index 0000000000..9448ffee54
--- /dev/null
+++ b/src/mesa/drivers/dri/r200/radeon_bo.c
@@ -0,0 +1 @@
+../radeon/radeon_bo.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r200/radeon_bo_int_drm.h b/src/mesa/drivers/dri/r200/radeon_bo_int_drm.h
new file mode 120000
index 0000000000..029450928b
--- /dev/null
+++ b/src/mesa/drivers/dri/r200/radeon_bo_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_bo_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r200/radeon_cs.c b/src/mesa/drivers/dri/r200/radeon_cs.c
new file mode 120000
index 0000000000..66b7ad1eb0
--- /dev/null
+++ b/src/mesa/drivers/dri/r200/radeon_cs.c
@@ -0,0 +1 @@
+../radeon/radeon_cs.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r200/radeon_cs_int_drm.h b/src/mesa/drivers/dri/r200/radeon_cs_int_drm.h
new file mode 120000
index 0000000000..462f5245d0
--- /dev/null
+++ b/src/mesa/drivers/dri/r200/radeon_cs_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_cs_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile
index c64f940623..be005bd164 100644
--- a/src/mesa/drivers/dri/r300/Makefile
+++ b/src/mesa/drivers/dri/r300/Makefile
@@ -14,7 +14,7 @@ EGL_SOURCES = server/radeon_egl.c
 endif
 
 ifeq ($(RADEON_LDFLAGS),)
-CS_SOURCES = radeon_cs_space_drm.c
+CS_SOURCES = radeon_cs_space_drm.c radeon_bo.c radeon_cs.c
 endif
 
 COMMON_SOURCES = \
@@ -43,13 +43,14 @@ RADEON_COMMON_SOURCES = \
 
 DRIVER_SOURCES = \
 		 radeon_screen.c \
+		 r300_blit.c \
 		 r300_context.c \
 		 r300_draw.c \
-		 r300_ioctl.c \
 		 r300_cmdbuf.c \
 		 r300_state.c \
 		 r300_render.c \
 		 r300_tex.c \
+		 r300_texcopy.c \
 		 r300_texstate.c \
 		 r300_vertprog.c \
 		 r300_fragprog_common.c \
@@ -69,16 +70,12 @@ DRIVER_DEFINES = -DRADEON_R300
 
 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/memory_pool.c b/src/mesa/drivers/dri/r300/compiler/memory_pool.c
index 37aa2b6579..76c7c60d8f 100644
--- a/src/mesa/drivers/dri/r300/compiler/memory_pool.c
+++ b/src/mesa/drivers/dri/r300/compiler/memory_pool.c
@@ -71,12 +71,14 @@ static void refill_pool(struct memory_pool * pool)
 void * memory_pool_malloc(struct memory_pool * pool, unsigned int bytes)
 {
 	if (bytes < POOL_LARGE_ALLOC) {
+		void * ptr;
+
 		if (pool->head + bytes > pool->end)
 			refill_pool(pool);
 
 		assert(pool->head + bytes <= pool->end);
 
-		void * ptr = pool->head;
+		ptr = pool->head;
 
 		pool->head += bytes;
 		pool->head = (unsigned char*)(((unsigned long)pool->head + POOL_ALIGN - 1) & ~(POOL_ALIGN - 1));
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.c b/src/mesa/drivers/dri/r300/compiler/radeon_code.c
index 1a3d8bb641..853b2becd1 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_code.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.c
@@ -143,7 +143,8 @@ unsigned rc_constants_add_immediate_scalar(struct rc_constant_list * c, float da
 
 	for(index = 0; index < c->Count; ++index) {
 		if (c->Constants[index].Type == RC_CONSTANT_IMMEDIATE) {
-			for(unsigned comp = 0; comp < c->Constants[index].Size; ++comp) {
+			unsigned comp;
+			for(comp = 0; comp < c->Constants[index].Size; ++comp) {
 				if (c->Constants[index].u.Immediate[comp] == data) {
 					*swizzle = RC_MAKE_SWIZZLE(comp, comp, comp, comp);
 					return index;
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.h b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
index 902b7cfa53..6d979bbaec 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_code.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
@@ -59,7 +59,9 @@ enum {
 	RC_STATE_SHADOW_AMBIENT = 0,
 
 	RC_STATE_R300_WINDOW_DIMENSION,
-	RC_STATE_R300_TEXRECT_FACTOR
+	RC_STATE_R300_TEXRECT_FACTOR,
+	RC_STATE_R300_VIEWPORT_SCALE,
+	RC_STATE_R300_VIEWPORT_OFFSET
 };
 
 struct rc_constant {
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
index c0e7a7f7a0..272f9072d4 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.c
@@ -229,15 +229,20 @@ void rc_copy_output(struct radeon_compiler * c, unsigned output, unsigned dup_ou
 /**
  * Introduce standard code fragment to deal with fragment.position.
  */
-void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input)
+void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input,
+                                int full_vtransform)
 {
 	unsigned tempregi = rc_find_free_temporary(c);
+	struct rc_instruction * inst_rcp;
+	struct rc_instruction * inst_mul;
+	struct rc_instruction * inst_mad;
+	struct rc_instruction * inst;
 
 	c->Program.InputsRead &= ~(1 << wpos);
 	c->Program.InputsRead |= 1 << new_input;
 
 	/* perspective divide */
-	struct rc_instruction * inst_rcp = rc_insert_new_instruction(c, &c->Program.Instructions);
+	inst_rcp = rc_insert_new_instruction(c, &c->Program.Instructions);
 	inst_rcp->U.I.Opcode = RC_OPCODE_RCP;
 
 	inst_rcp->U.I.DstReg.File = RC_FILE_TEMPORARY;
@@ -248,7 +253,7 @@ void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsig
 	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 = rc_insert_new_instruction(c, inst_rcp);
 	inst_mul->U.I.Opcode = RC_OPCODE_MUL;
 
 	inst_mul->U.I.DstReg.File = RC_FILE_TEMPORARY;
@@ -263,7 +268,7 @@ void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsig
 	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 = rc_insert_new_instruction(c, inst_mul);
 	inst_mad->U.I.Opcode = RC_OPCODE_MAD;
 
 	inst_mad->U.I.DstReg.File = RC_FILE_TEMPORARY;
@@ -275,14 +280,19 @@ void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsig
 	inst_mad->U.I.SrcReg[0].Swizzle = RC_MAKE_SWIZZLE(RC_SWIZZLE_X, RC_SWIZZLE_Y, RC_SWIZZLE_Z, RC_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->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;
+	if (full_vtransform) {
+		inst_mad->U.I.SrcReg[1].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_VIEWPORT_SCALE, 0);
+		inst_mad->U.I.SrcReg[2].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_VIEWPORT_OFFSET, 0);
+	} else {
+		inst_mad->U.I.SrcReg[1].Index =
+		inst_mad->U.I.SrcReg[2].Index = rc_constants_add_state(&c->Program.Constants, RC_STATE_R300_WINDOW_DIMENSION, 0);
+	}
+
 	for (inst = inst_mad->Next; inst != &c->Program.Instructions; inst = inst->Next) {
 		const struct rc_opcode_info * opcode = rc_get_opcode_info(inst->U.I.Opcode);
 		unsigned i;
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
index 87a732cd90..731adc1af2 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_compiler.h
@@ -73,7 +73,8 @@ void rc_calculate_inputs_outputs(struct radeon_compiler * c);
 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);
+void rc_transform_fragment_wpos(struct radeon_compiler * c, unsigned wpos, unsigned new_input,
+                                int full_vtransform);
 
 struct r300_fragment_program_compiler {
 	struct radeon_compiler Base;
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c b/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c
index 828d0c8e28..b2fe7f76b2 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_pair_regalloc.c
@@ -49,7 +49,7 @@ struct register_info {
 
 	unsigned int Used:1;
 	unsigned int Allocated:1;
-	rc_register_file File:3;
+	unsigned int File:3;
 	unsigned int Index:RC_REGISTER_INDEX_BITS;
 };
 
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_program.h b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
index 33db3ea0ff..e318867696 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program.h
@@ -39,7 +39,7 @@
 struct radeon_compiler;
 
 struct rc_src_register {
-	rc_register_file File:3;
+	unsigned int File:3;
 
 	/** Negative values may be used for relative addressing. */
 	signed int Index:(RC_REGISTER_INDEX_BITS+1);
@@ -55,7 +55,7 @@ struct rc_src_register {
 };
 
 struct rc_dst_register {
-	rc_register_file File:3;
+	unsigned int File:3;
 
 	/** Negative values may be used for relative addressing. */
 	signed int Index:(RC_REGISTER_INDEX_BITS+1);
@@ -79,20 +79,20 @@ struct rc_sub_instruction {
 	/**
 	 * Opcode of this instruction, according to \ref rc_opcode enums.
 	 */
-	rc_opcode Opcode:8;
+	unsigned int 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;
+	unsigned int SaturateMode:2;
 
 	/**
 	 * Writing to the special register RC_SPECIAL_ALU_RESULT
 	 */
 	/*@{*/
-	rc_write_aluresult WriteALUResult:2;
-	rc_compare_func ALUResultCompare:3;
+	unsigned int WriteALUResult:2;
+	unsigned int ALUResultCompare:3;
 	/*@}*/
 
 	/**
@@ -103,7 +103,7 @@ struct rc_sub_instruction {
 	unsigned int TexSrcUnit:5;
 
 	/** Source texture target, one of the \ref rc_texture_target enums */
-	rc_texture_target TexSrcTarget:3;
+	unsigned int TexSrcTarget:3;
 
 	/** True if tex instruction should do shadow comparison */
 	unsigned int TexShadow:1;
@@ -191,7 +191,7 @@ struct rc_src_register lmul_swizzle(unsigned int swizzle, struct rc_src_register
 
 static inline void reset_srcreg(struct rc_src_register* reg)
 {
-	memset(reg, 0, sizeof(reg));
+	memset(reg, 0, sizeof(struct rc_src_register));
 	reg->Swizzle = RC_SWIZZLE_XYZW;
 }
 
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 0326d25233..b5c08aea49 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
@@ -267,9 +267,9 @@ static void transform_LIT(struct radeon_compiler* c,
 	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);
+	/* 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, RC_OPCODE_MAX, 0,
 		dstregtmpmask(temp, RC_MASK_XYW),
 		inst->U.I.SrcReg[0],
@@ -280,7 +280,7 @@ static void transform_LIT(struct radeon_compiler* c,
 		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)
+	/* tmp.w = Pow(tmp.y, tmp.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));
@@ -292,14 +292,14 @@ static void transform_LIT(struct radeon_compiler* c,
 		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
+	/* tmp.z = (tmp.x > 0) ? tmp.w : 0.0 */
 	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
+	/* tmp.x, tmp.y, tmp.w = 1.0, tmp.x, 1.0 */
 	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));
@@ -533,16 +533,16 @@ static void sincos_constants(struct radeon_compiler* c, unsigned int *constants)
 {
 	static const float SinCosConsts[2][4] = {
 		{
-			1.273239545,		// 4/PI
-			-0.405284735,		// -4/(PI*PI)
-			3.141592654,		// PI
-			0.2225			// weight
+			1.273239545,		/* 4/PI */
+			-0.405284735,		/* -4/(PI*PI) */
+			3.141592654,		/* PI */
+			0.2225			/* weight */
 		},
 		{
 			0.75,
 			0.5,
-			0.159154943,		// 1/(2*PI)
-			6.283185307		// 2*PI
+			0.159154943,		/* 1/(2*PI) */
+			6.283185307		/* 2*PI */
 		}
 	};
 	int i;
@@ -560,23 +560,23 @@ static void sincos_constants(struct radeon_compiler* c, unsigned int *constants)
  * MAD dest, tmp.y, weight, tmp.x
  */
 static void sin_approx(
-	struct radeon_compiler* c, struct rc_instruction * before,
+	struct radeon_compiler* c, struct rc_instruction * inst,
 	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),
+	emit2(c, inst->Prev, 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),
+	emit3(c, inst->Prev, 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),
+	emit3(c, inst->Prev, 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,
+	emit3(c, inst->Prev, 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));
@@ -602,9 +602,9 @@ int radeonTransformTrigSimple(struct radeon_compiler* c,
 	sincos_constants(c, constants);
 
 	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
+		/* 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, 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),
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 1600598428..6685ade3ea 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_pair.h
@@ -52,12 +52,12 @@ struct r300_fragment_program_compiler;
 
 struct radeon_pair_instruction_source {
 	unsigned int Used:1;
-	rc_register_file File:3;
+	unsigned int File:3;
 	unsigned int Index:RC_REGISTER_INDEX_BITS;
 };
 
 struct radeon_pair_instruction_rgb {
-	rc_opcode Opcode:8;
+	unsigned int Opcode:8;
 	unsigned int DestIndex:RC_REGISTER_INDEX_BITS;
 	unsigned int WriteMask:3;
 	unsigned int OutputWriteMask:3;
@@ -74,7 +74,7 @@ struct radeon_pair_instruction_rgb {
 };
 
 struct radeon_pair_instruction_alpha {
-	rc_opcode Opcode:8;
+	unsigned int Opcode:8;
 	unsigned int DestIndex:RC_REGISTER_INDEX_BITS;
 	unsigned int WriteMask:1;
 	unsigned int OutputWriteMask:1;
@@ -95,8 +95,8 @@ struct rc_pair_instruction {
 	struct radeon_pair_instruction_rgb RGB;
 	struct radeon_pair_instruction_alpha Alpha;
 
-	rc_write_aluresult WriteALUResult:2;
-	rc_compare_func ALUResultCompare:3;
+	unsigned int WriteALUResult:2;
+	unsigned int ALUResultCompare:3;
 };
 
 
diff --git a/src/mesa/drivers/dri/r300/r300_blit.c b/src/mesa/drivers/dri/r300/r300_blit.c
new file mode 100644
index 0000000000..ea626d942d
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r300_blit.c
@@ -0,0 +1,607 @@
+/*
+ * Copyright (C) 2009 Maciej Cencora <m.cencora@gmail.com>
+ *
+ * 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_common.h"
+#include "r300_context.h"
+
+#include "r300_blit.h"
+#include "r300_cmdbuf.h"
+#include "r300_emit.h"
+#include "r300_tex.h"
+#include "compiler/radeon_compiler.h"
+#include "compiler/radeon_opcodes.h"
+
+static void vp_ins_outs(struct r300_vertex_program_compiler *c)
+{
+    c->code->inputs[VERT_ATTRIB_POS] = 0;
+    c->code->inputs[VERT_ATTRIB_TEX0] = 1;
+    c->code->outputs[VERT_RESULT_HPOS] = 0;
+    c->code->outputs[VERT_RESULT_TEX0] = 1;
+}
+
+static void fp_allocate_hw_inputs(
+    struct r300_fragment_program_compiler * c,
+    void (*allocate)(void * data, unsigned input, unsigned hwreg),
+    void * mydata)
+{
+    allocate(mydata, FRAG_ATTRIB_TEX0, 0);
+}
+
+static void create_vertex_program(struct r300_context *r300)
+{
+    struct r300_vertex_program_compiler compiler;
+    struct rc_instruction *inst;
+
+    rc_init(&compiler.Base);
+
+    inst = rc_insert_new_instruction(&compiler.Base, compiler.Base.Program.Instructions.Prev);
+    inst->U.I.Opcode = RC_OPCODE_MOV;
+    inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+    inst->U.I.DstReg.Index = VERT_RESULT_HPOS;
+    inst->U.I.DstReg.RelAddr = 0;
+    inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+    inst->U.I.SrcReg[0].Abs = 0;
+    inst->U.I.SrcReg[0].File = RC_FILE_INPUT;
+    inst->U.I.SrcReg[0].Index = VERT_ATTRIB_POS;
+    inst->U.I.SrcReg[0].Negate = 0;
+    inst->U.I.SrcReg[0].RelAddr = 0;
+    inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
+
+    inst = rc_insert_new_instruction(&compiler.Base, compiler.Base.Program.Instructions.Prev);
+    inst->U.I.Opcode = RC_OPCODE_MOV;
+    inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+    inst->U.I.DstReg.Index = VERT_RESULT_TEX0;
+    inst->U.I.DstReg.RelAddr = 0;
+    inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+    inst->U.I.SrcReg[0].Abs = 0;
+    inst->U.I.SrcReg[0].File = RC_FILE_INPUT;
+    inst->U.I.SrcReg[0].Index = VERT_ATTRIB_TEX0;
+    inst->U.I.SrcReg[0].Negate = 0;
+    inst->U.I.SrcReg[0].RelAddr = 0;
+    inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
+
+    compiler.Base.Program.InputsRead = (1 << VERT_ATTRIB_POS) | (1 << VERT_ATTRIB_TEX0);
+    compiler.RequiredOutputs = compiler.Base.Program.OutputsWritten = (1 << VERT_RESULT_HPOS) | (1 << VERT_RESULT_TEX0);
+    compiler.SetHwInputOutput = vp_ins_outs;
+    compiler.code = &r300->blit.vp_code;
+
+    r3xx_compile_vertex_program(&compiler);
+}
+
+static void create_fragment_program(struct r300_context *r300)
+{
+    struct r300_fragment_program_compiler compiler;
+    struct rc_instruction *inst;
+
+    rc_init(&compiler.Base);
+
+    inst = rc_insert_new_instruction(&compiler.Base, compiler.Base.Program.Instructions.Prev);
+    inst->U.I.Opcode = RC_OPCODE_TEX;
+    inst->U.I.TexSrcTarget = RC_TEXTURE_2D;
+    inst->U.I.TexSrcUnit = 0;
+    inst->U.I.DstReg.File = RC_FILE_OUTPUT;
+    inst->U.I.DstReg.Index = FRAG_RESULT_COLOR;
+    inst->U.I.DstReg.WriteMask = RC_MASK_XYZW;
+    inst->U.I.SrcReg[0].Abs = 0;
+    inst->U.I.SrcReg[0].File = RC_FILE_INPUT;
+    inst->U.I.SrcReg[0].Index = FRAG_ATTRIB_TEX0;
+    inst->U.I.SrcReg[0].Negate = 0;
+    inst->U.I.SrcReg[0].RelAddr = 0;
+    inst->U.I.SrcReg[0].Swizzle = RC_SWIZZLE_XYZW;
+
+    compiler.Base.Program.InputsRead = (1 << FRAG_ATTRIB_TEX0);
+    compiler.OutputColor = FRAG_RESULT_COLOR;
+    compiler.OutputDepth = FRAG_RESULT_DEPTH;
+    compiler.is_r500 = (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515);
+    compiler.code = &r300->blit.fp_code;
+    compiler.AllocateHwInputs = fp_allocate_hw_inputs;
+
+    r3xx_compile_fragment_program(&compiler);
+}
+
+void r300_blit_init(struct r300_context *r300)
+{
+    create_vertex_program(r300);
+    create_fragment_program(r300);
+}
+
+static void r300_emit_tx_setup(struct r300_context *r300,
+                               gl_format mesa_format,
+                               struct radeon_bo *bo,
+                               intptr_t offset,
+                               unsigned width,
+                               unsigned height,
+                               unsigned pitch)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    assert(width <= 2048);
+    assert(height <= 2048);
+    assert(r300TranslateTexFormat(mesa_format) >= 0);
+    assert(offset % 32 == 0);
+
+    BEGIN_BATCH(17);
+    OUT_BATCH_REGVAL(R300_TX_FILTER0_0,
+                     (R300_TX_CLAMP_TO_EDGE  << R300_TX_WRAP_S_SHIFT) |
+                     (R300_TX_CLAMP_TO_EDGE  << R300_TX_WRAP_T_SHIFT) |
+                     (R300_TX_CLAMP_TO_EDGE  << R300_TX_WRAP_R_SHIFT) |
+                     R300_TX_MIN_FILTER_MIP_NONE |
+                     R300_TX_MIN_FILTER_LINEAR |
+                     R300_TX_MAG_FILTER_LINEAR |
+                     (0 << 28));
+    OUT_BATCH_REGVAL(R300_TX_FILTER1_0, 0);
+    OUT_BATCH_REGVAL(R300_TX_SIZE_0,
+                     ((width-1) << R300_TX_WIDTHMASK_SHIFT) |
+                     ((height-1) << R300_TX_HEIGHTMASK_SHIFT) |
+                     (0 << R300_TX_DEPTHMASK_SHIFT) |
+                     (0 << R300_TX_MAX_MIP_LEVEL_SHIFT) |
+                     R300_TX_SIZE_TXPITCH_EN);
+
+    OUT_BATCH_REGVAL(R300_TX_FORMAT_0, r300TranslateTexFormat(mesa_format));
+    OUT_BATCH_REGVAL(R300_TX_FORMAT2_0, pitch - 1);
+    OUT_BATCH_REGSEQ(R300_TX_OFFSET_0, 1);
+    OUT_BATCH_RELOC(0, bo, offset, RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
+
+    OUT_BATCH_REGSEQ(R300_TX_INVALTAGS, 2);
+    OUT_BATCH(0);
+    OUT_BATCH(1);
+
+    END_BATCH();
+}
+
+#define EASY_US_FORMAT(FMT, C0, C1, C2, C3, SIGN) \
+    (FMT  | R500_C0_SEL_##C0 | R500_C1_SEL_##C1 | \
+    R500_C2_SEL_##C2 | R500_C3_SEL_##C3 | R500_OUT_SIGN(SIGN))
+
+static uint32_t mesa_format_to_us_format(gl_format mesa_format)
+{
+    switch(mesa_format)
+    {
+        case MESA_FORMAT_S8_Z24:
+        case MESA_FORMAT_X8_Z24:
+        case MESA_FORMAT_RGBA8888: // x
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, A, B, G, R, 0);
+        case MESA_FORMAT_RGB565: // x
+        case MESA_FORMAT_ARGB1555: // x
+        case MESA_FORMAT_RGBA8888_REV: // x
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, R, G, B, A, 0);
+        case MESA_FORMAT_ARGB8888: // x
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, B, G, R, A, 0);
+        case MESA_FORMAT_ARGB8888_REV:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, A, R, G, B, 0);
+        case MESA_FORMAT_XRGB8888:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, A, R, G, B, 0);
+
+        case MESA_FORMAT_RGB332:
+            return EASY_US_FORMAT(R500_OUT_FMT_C_3_3_2, A, R, G, B, 0);
+
+        case MESA_FORMAT_RGBA_FLOAT32:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_32_FP, R, G, B, A, 0);
+        case MESA_FORMAT_RGBA_FLOAT16:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_16_FP, R, G, B, A, 0);
+        case MESA_FORMAT_ALPHA_FLOAT32:
+            return EASY_US_FORMAT(R500_OUT_FMT_C_32_FP, A, A, A, A, 0);
+        case MESA_FORMAT_ALPHA_FLOAT16:
+            return EASY_US_FORMAT(R500_OUT_FMT_C_16_FP, A, A, A, A, 0);
+
+        case MESA_FORMAT_SIGNED_RGBA8888:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, R, G, B, A, 0xf);
+        case MESA_FORMAT_SIGNED_RGBA8888_REV:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_8, A, B, G, R, 0xf);
+        case MESA_FORMAT_SIGNED_RGBA_16:
+            return EASY_US_FORMAT(R500_OUT_FMT_C4_16, R, G, B, A, 0xf);
+
+        default:
+            assert(!"Invalid format for US output\n");
+            return 0;
+    }
+}
+#undef EASY_US_FORMAT
+
+static void r500_emit_fp_setup(struct r300_context *r300,
+                               struct r500_fragment_program_code *fp,
+                               gl_format dst_format)
+{
+    r500_emit_fp(r300, (uint32_t *)fp->inst, (fp->inst_end + 1) * 6, 0, 0, 0);
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(10);
+    OUT_BATCH_REGSEQ(R500_US_CODE_ADDR, 3);
+    OUT_BATCH(R500_US_CODE_START_ADDR(0) | R500_US_CODE_END_ADDR(fp->inst_end));
+    OUT_BATCH(R500_US_CODE_RANGE_ADDR(0) | R500_US_CODE_RANGE_SIZE(fp->inst_end));
+    OUT_BATCH(0);
+    OUT_BATCH_REGVAL(R500_US_CONFIG, 0);
+    OUT_BATCH_REGVAL(R500_US_OUT_FMT_0, mesa_format_to_us_format(dst_format));
+    OUT_BATCH_REGVAL(R500_US_PIXSIZE, fp->max_temp_idx);
+    END_BATCH();
+}
+
+static void r500_emit_rs_setup(struct r300_context *r300)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(7);
+    OUT_BATCH_REGSEQ(R300_RS_COUNT, 2);
+    OUT_BATCH((4 << R300_IT_COUNT_SHIFT) | R300_HIRES_EN);
+    OUT_BATCH(0);
+    OUT_BATCH_REGVAL(R500_RS_INST_0,
+                     (0 << R500_RS_INST_TEX_ID_SHIFT) |
+                     (0 << R500_RS_INST_TEX_ADDR_SHIFT) |
+                     R500_RS_INST_TEX_CN_WRITE |
+                     R500_RS_INST_COL_CN_NO_WRITE);
+    OUT_BATCH_REGVAL(R500_RS_IP_0,
+                     (0 << R500_RS_IP_TEX_PTR_S_SHIFT) |
+                     (1 << R500_RS_IP_TEX_PTR_T_SHIFT) |
+                     (2 << R500_RS_IP_TEX_PTR_R_SHIFT) |
+                     (3 << R500_RS_IP_TEX_PTR_Q_SHIFT));
+    END_BATCH();
+}
+
+static void r300_emit_fp_setup(struct r300_context *r300,
+                               struct r300_fragment_program_code *code,
+                               gl_format dst_format)
+{
+    unsigned i;
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH((code->alu.length + 1) * 4 + code->tex.length + 1 + 11);
+
+    OUT_BATCH_REGSEQ(R300_US_ALU_RGB_INST_0, code->alu.length);
+    for (i = 0; i < code->alu.length; i++) {
+        OUT_BATCH(code->alu.inst[i].rgb_inst);
+    }
+    OUT_BATCH_REGSEQ(R300_US_ALU_RGB_ADDR_0, code->alu.length);
+    for (i = 0; i < code->alu.length; i++) {
+        OUT_BATCH(code->alu.inst[i].rgb_addr);
+    }
+    OUT_BATCH_REGSEQ(R300_US_ALU_ALPHA_INST_0, code->alu.length);
+    for (i = 0; i < code->alu.length; i++) {
+        OUT_BATCH(code->alu.inst[i].alpha_inst);
+    }
+    OUT_BATCH_REGSEQ(R300_US_ALU_ALPHA_ADDR_0, code->alu.length);
+    for (i = 0; i < code->alu.length; i++) {
+        OUT_BATCH(code->alu.inst[i].alpha_addr);
+    }
+
+    OUT_BATCH_REGSEQ(R300_US_TEX_INST_0, code->tex.length);
+    OUT_BATCH_TABLE(code->tex.inst, code->tex.length);
+
+    OUT_BATCH_REGSEQ(R300_US_CONFIG, 3);
+    OUT_BATCH(R300_PFS_CNTL_FIRST_NODE_HAS_TEX);
+    OUT_BATCH(code->pixsize);
+    OUT_BATCH(code->code_offset);
+    OUT_BATCH_REGSEQ(R300_US_CODE_ADDR_0, 4);
+    OUT_BATCH_TABLE(code->code_addr, 4);
+    OUT_BATCH_REGVAL(R500_US_OUT_FMT_0, mesa_format_to_us_format(dst_format));
+    END_BATCH();
+}
+
+static void r300_emit_rs_setup(struct r300_context *r300)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(7);
+    OUT_BATCH_REGSEQ(R300_RS_COUNT, 2);
+    OUT_BATCH((4 << R300_IT_COUNT_SHIFT) | R300_HIRES_EN);
+    OUT_BATCH(0);
+    OUT_BATCH_REGVAL(R300_RS_INST_0,
+                     R300_RS_INST_TEX_ID(0) |
+                     R300_RS_INST_TEX_ADDR(0) |
+                     R300_RS_INST_TEX_CN_WRITE);
+    OUT_BATCH_REGVAL(R300_RS_IP_0,
+                     R300_RS_TEX_PTR(0) |
+                     R300_RS_SEL_S(R300_RS_SEL_C0) |
+                     R300_RS_SEL_T(R300_RS_SEL_C1) |
+                     R300_RS_SEL_R(R300_RS_SEL_K0) |
+                     R300_RS_SEL_Q(R300_RS_SEL_K1));
+    END_BATCH();
+}
+
+static void emit_pvs_setup(struct r300_context *r300,
+                           uint32_t *vp_code,
+                           unsigned vp_len)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    r300_emit_vpu(r300, vp_code, vp_len * 4, R300_PVS_CODE_START);
+
+    BEGIN_BATCH(4);
+    OUT_BATCH_REGSEQ(R300_VAP_PVS_CODE_CNTL_0, 3);
+    OUT_BATCH((0 << R300_PVS_FIRST_INST_SHIFT) |
+              ((vp_len - 1)  << R300_PVS_XYZW_VALID_INST_SHIFT) |
+              ((vp_len - 1)<< R300_PVS_LAST_INST_SHIFT));
+    OUT_BATCH(0);
+    OUT_BATCH((vp_len - 1) << R300_PVS_LAST_VTX_SRC_INST_SHIFT);
+    END_BATCH();
+}
+
+static void emit_vap_setup(struct r300_context *r300)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(12);
+    OUT_BATCH_REGSEQ(R300_SE_VTE_CNTL, 2);
+    OUT_BATCH(R300_VTX_XY_FMT | R300_VTX_Z_FMT);
+    OUT_BATCH(4);
+
+    OUT_BATCH_REGVAL(R300_VAP_PSC_SGN_NORM_CNTL, 0xaaaaaaaa);
+    OUT_BATCH_REGVAL(R300_VAP_PROG_STREAM_CNTL_0,
+                     ((R300_DATA_TYPE_FLOAT_2 | (0 << R300_DST_VEC_LOC_SHIFT)) << 0) |
+                     (((1 << R300_DST_VEC_LOC_SHIFT) | R300_DATA_TYPE_FLOAT_2 | R300_LAST_VEC) << 16));
+    OUT_BATCH_REGVAL(R300_VAP_PROG_STREAM_CNTL_EXT_0,
+                    ((((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) ) << 0) |
+                     (((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) ) << 16) ) );
+    OUT_BATCH_REGSEQ(R300_VAP_OUTPUT_VTX_FMT_0, 2);
+    OUT_BATCH(R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT);
+    OUT_BATCH(R300_VAP_OUTPUT_VTX_FMT_1__4_COMPONENTS);
+    END_BATCH();
+}
+
+static GLboolean validate_buffers(struct r300_context *r300,
+                                  struct radeon_bo *src_bo,
+                                  struct radeon_bo *dst_bo)
+{
+    int ret;
+    radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
+                                      src_bo, RADEON_GEM_DOMAIN_VRAM, 0);
+
+    radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
+                                      dst_bo, 0, RADEON_GEM_DOMAIN_VRAM);
+
+    ret = radeon_cs_space_check_with_bo(r300->radeon.cmdbuf.cs,
+                                        first_elem(&r300->radeon.dma.reserved)->bo,
+                                        RADEON_GEM_DOMAIN_GTT, 0);
+    if (ret)
+        return GL_FALSE;
+
+    return GL_TRUE;
+}
+
+/**
+ * Calculate texcoords for given image region.
+ * Output values are [minx, maxx, miny, maxy]
+ */
+static void calc_tex_coords(float img_width, float img_height,
+                            float x, float y,
+                            float reg_width, float reg_height,
+                            unsigned flip_y, float *buf)
+{
+    buf[0] = x / img_width;
+    buf[1] = buf[0] + reg_width / img_width;
+    buf[2] = y / img_height;
+    buf[3] = buf[2] + reg_height / img_height;
+    if (flip_y)
+    {
+        float tmp = buf[2];
+        buf[2] = 1.0 - buf[3];
+        buf[3] = 1.0 - tmp;
+    }
+}
+
+static void emit_draw_packet(struct r300_context *r300,
+                             unsigned src_width, unsigned src_height,
+                             unsigned src_x_offset, unsigned src_y_offset,
+                             unsigned dst_x_offset, unsigned dst_y_offset,
+                             unsigned reg_width, unsigned reg_height,
+                             unsigned flip_y)
+{
+    float texcoords[4];
+
+    calc_tex_coords(src_width, src_height,
+                    src_x_offset, src_y_offset,
+                    reg_width, reg_height,
+                    flip_y, texcoords);
+
+    float verts[] = { dst_x_offset, dst_y_offset,
+                      texcoords[0], texcoords[3],
+                      dst_x_offset, dst_y_offset + reg_height,
+                      texcoords[0], texcoords[2],
+                      dst_x_offset + reg_width, dst_y_offset + reg_height,
+                      texcoords[1], texcoords[2],
+                      dst_x_offset + reg_width, dst_y_offset,
+                      texcoords[1], texcoords[3] };
+
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(19);
+    OUT_BATCH_PACKET3(R300_PACKET3_3D_DRAW_IMMD_2, 16);
+    OUT_BATCH(R300_VAP_VF_CNTL__PRIM_WALK_VERTEX_EMBEDDED |
+              (4 << 16) | R300_VAP_VF_CNTL__PRIM_QUADS);
+    OUT_BATCH_TABLE(verts, 16);
+    END_BATCH();
+}
+
+static void other_stuff(struct r300_context *r300)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH(15);
+    OUT_BATCH_REGVAL(R300_GA_POLY_MODE,
+                     R300_GA_POLY_MODE_FRONT_PTYPE_TRI | R300_GA_POLY_MODE_BACK_PTYPE_TRI);
+    OUT_BATCH_REGVAL(R300_SU_CULL_MODE, R300_FRONT_FACE_CCW);
+    OUT_BATCH_REGVAL(R300_FG_FOG_BLEND, 0);
+    OUT_BATCH_REGVAL(R300_FG_ALPHA_FUNC, 0);
+    OUT_BATCH_REGSEQ(R300_RB3D_CBLEND, 2);
+    OUT_BATCH(0x0);
+    OUT_BATCH(0x0);
+    OUT_BATCH_REGVAL(R300_VAP_CLIP_CNTL, R300_CLIP_DISABLE);
+    OUT_BATCH_REGVAL(R300_ZB_CNTL, 0);
+    END_BATCH();
+}
+
+static void emit_cb_setup(struct r300_context *r300,
+                          struct radeon_bo *bo,
+                          intptr_t offset,
+                          gl_format mesa_format,
+                          unsigned pitch,
+                          unsigned width,
+                          unsigned height)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    unsigned x1, y1, x2, y2;
+    x1 = 0;
+    y1 = 0;
+    x2 = width - 1;
+    y2 = height - 1;
+
+    if (r300->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV515) {
+        x1 += R300_SCISSORS_OFFSET;
+        y1 += R300_SCISSORS_OFFSET;
+        x2 += R300_SCISSORS_OFFSET;
+        y2 += R300_SCISSORS_OFFSET;
+    }
+
+    r300_emit_cb_setup(r300, bo, offset, mesa_format,
+                       _mesa_get_format_bytes(mesa_format),
+                       _mesa_format_row_stride(mesa_format, pitch));
+
+    BEGIN_BATCH_NO_AUTOSTATE(5);
+    OUT_BATCH_REGSEQ(R300_SC_SCISSORS_TL, 2);
+    OUT_BATCH((x1 << R300_SCISSORS_X_SHIFT)|(y1 << R300_SCISSORS_Y_SHIFT));
+    OUT_BATCH((x2 << R300_SCISSORS_X_SHIFT)|(y2 << R300_SCISSORS_Y_SHIFT));
+    OUT_BATCH_REGVAL(R300_RB3D_CCTL, 0);
+    END_BATCH();
+}
+
+/**
+ * Copy a region of [@a width x @a height] pixels from source buffer
+ * to destination buffer.
+ * @param[in] r300 r300 context
+ * @param[in] src_bo source radeon buffer object
+ * @param[in] src_offset offset of the source image in the @a src_bo
+ * @param[in] src_mesaformat source image format
+ * @param[in] src_pitch aligned source image width
+ * @param[in] src_width source image width
+ * @param[in] src_height source image height
+ * @param[in] src_x_offset x offset in the source image
+ * @param[in] src_y_offset y offset in the source image
+ * @param[in] dst_bo destination radeon buffer object
+ * @param[in] dst_offset offset of the destination image in the @a dst_bo
+ * @param[in] dst_mesaformat destination image format
+ * @param[in] dst_pitch aligned destination image width
+ * @param[in] dst_width destination image width
+ * @param[in] dst_height destination image height
+ * @param[in] dst_x_offset x offset in the destination image
+ * @param[in] dst_y_offset y offset in the destination image
+ * @param[in] width region width
+ * @param[in] height region height
+ * @param[in] flip_y set if y coords of the source image need to be flipped
+ */
+GLboolean r300_blit(struct r300_context *r300,
+                    struct radeon_bo *src_bo,
+                    intptr_t src_offset,
+                    gl_format src_mesaformat,
+                    unsigned src_pitch,
+                    unsigned src_width,
+                    unsigned src_height,
+                    unsigned src_x_offset,
+                    unsigned src_y_offset,
+                    struct radeon_bo *dst_bo,
+                    intptr_t dst_offset,
+                    gl_format dst_mesaformat,
+                    unsigned dst_pitch,
+                    unsigned dst_width,
+                    unsigned dst_height,
+                    unsigned dst_x_offset,
+                    unsigned dst_y_offset,
+                    unsigned reg_width,
+                    unsigned reg_height,
+                    unsigned flip_y)
+{
+    /* Need to clamp the region size to make sure
+     * we don't read outside of the source buffer
+     * or write outside of the destination buffer.
+     */
+    if (reg_width + src_x_offset > src_width)
+        reg_width = src_width - src_x_offset;
+    if (reg_height + src_y_offset > src_height)
+        reg_height = src_height - src_y_offset;
+    if (reg_width + dst_x_offset > dst_width)
+        reg_width = dst_width - dst_x_offset;
+    if (reg_height + dst_y_offset > dst_height)
+        reg_height = dst_height - dst_y_offset;
+
+    if (src_bo == dst_bo) {
+        return GL_FALSE;
+    }
+
+    if (0) {
+        fprintf(stderr, "src: size [%d x %d], pitch %d, "
+                "offset [%d x %d], format %s, bo %p\n",
+                src_width, src_height, src_pitch,
+                src_offset, src_y_offset,
+                _mesa_get_format_name(src_mesaformat),
+                src_bo);
+        fprintf(stderr, "dst: pitch %d, offset[%d x %d], format %s, bo %p\n",
+                dst_pitch, dst_x_offset, dst_y_offset,
+                _mesa_get_format_name(dst_mesaformat), dst_bo);
+        fprintf(stderr, "region: %d x %d\n", reg_width, reg_height);
+    }
+
+    if (!validate_buffers(r300, src_bo, dst_bo))
+        return GL_FALSE;
+
+    rcommonEnsureCmdBufSpace(&r300->radeon, 200, __FUNCTION__);
+
+    other_stuff(r300);
+
+    r300_emit_tx_setup(r300, src_mesaformat, src_bo, src_offset, src_width, src_height, src_pitch);
+
+    if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+        r500_emit_fp_setup(r300, &r300->blit.fp_code.code.r500, dst_mesaformat);
+        r500_emit_rs_setup(r300);
+    } else {
+        r300_emit_fp_setup(r300, &r300->blit.fp_code.code.r300, dst_mesaformat);
+        r300_emit_rs_setup(r300);
+    }
+
+    emit_pvs_setup(r300, r300->blit.vp_code.body.d, 2);
+    emit_vap_setup(r300);
+
+    emit_cb_setup(r300, dst_bo, dst_offset, dst_mesaformat, dst_pitch, dst_width, dst_height);
+
+    emit_draw_packet(r300, src_width, src_height,
+                     src_x_offset, src_y_offset,
+                     dst_x_offset, dst_y_offset,
+                     reg_width, reg_height,
+                     flip_y);
+
+    r300EmitCacheFlush(r300);
+
+    radeonFlush(r300->radeon.glCtx);
+
+    return GL_TRUE;
+}
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/r300_blit.h b/src/mesa/drivers/dri/r300/r300_blit.h
new file mode 100644
index 0000000000..dc21e88098
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r300_blit.h
@@ -0,0 +1,54 @@
+/*
+ * Copyright (C) 2009 Maciej Cencora <m.cencora@gmail.com>
+ *
+ * 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 R300_BLIT_H
+#define R300_BLIT_H
+
+void r300_blit_init(struct r300_context *r300);
+
+GLboolean r300_blit(struct r300_context *r300,
+                    struct radeon_bo *src_bo,
+                    intptr_t src_offset,
+                    gl_format src_mesaformat,
+                    unsigned src_pitch,
+                    unsigned src_width,
+                    unsigned src_height,
+                    unsigned src_x_offset,
+                    unsigned src_y_offset,
+                    struct radeon_bo *dst_bo,
+                    intptr_t dst_offset,
+                    gl_format dst_mesaformat,
+                    unsigned dst_pitch,
+                    unsigned dst_width,
+                    unsigned dst_height,
+                    unsigned dst_x_offset,
+                    unsigned dst_y_offset,
+                    unsigned width,
+                    unsigned height,
+                    unsigned flip_y);
+
+#endif // R300_BLIT_H
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.c b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
index da5b7ba642..e1c33bbb2c 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.c
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.c
@@ -45,15 +45,12 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_drm.h"
 
 #include "r300_context.h"
-#include "r300_ioctl.h"
-#include "radeon_reg.h"
 #include "r300_reg.h"
 #include "r300_cmdbuf.h"
 #include "r300_emit.h"
 #include "radeon_bocs_wrapper.h"
 #include "radeon_mipmap_tree.h"
 #include "r300_state.h"
-#include "radeon_reg.h"
 #include "radeon_queryobj.h"
 
 /** # of dwords reserved for additional instructions that may need to be written
@@ -74,7 +71,7 @@ static unsigned packet0_count(r300ContextPtr r300, uint32_t *pkt)
 #define vpu_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->vpu.count)
 #define r500fp_count(ptr) (((drm_r300_cmd_header_t*)(ptr))->r500fp.count)
 
-int check_vpu(GLcontext *ctx, struct radeon_state_atom *atom)
+static int check_vpu(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
 	int cnt;
@@ -88,54 +85,73 @@ int check_vpu(GLcontext *ctx, struct radeon_state_atom *atom)
 	return cnt ? (cnt * 4) + extra : 0;
 }
 
+void r300_emit_vpu(struct r300_context *r300,
+                   uint32_t *data,
+                   unsigned len,
+                   uint32_t addr)
+{
+    BATCH_LOCALS(&r300->radeon);
+
+    BEGIN_BATCH_NO_AUTOSTATE(5 + len);
+    OUT_BATCH_REGVAL(R300_VAP_PVS_STATE_FLUSH_REG, 0);
+    OUT_BATCH_REGVAL(R300_VAP_PVS_VECTOR_INDX_REG, addr);
+    OUT_BATCH(CP_PACKET0(R300_VAP_PVS_UPLOAD_DATA, len-1) | RADEON_ONE_REG_WR);
+    OUT_BATCH_TABLE(data, len);
+    END_BATCH();
+}
 
-void emit_vpu(GLcontext *ctx, struct radeon_state_atom * atom)
+static void emit_vpu_state(GLcontext *ctx, struct radeon_state_atom * atom)
 {
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	BATCH_LOCALS(&r300->radeon);
-	drm_r300_cmd_header_t cmd;
-	uint32_t addr, ndw;
+    r300ContextPtr r300 = R300_CONTEXT(ctx);
+    drm_r300_cmd_header_t cmd;
+    uint32_t addr, ndw;
+
+    cmd.u = atom->cmd[0];
+    addr = (cmd.vpu.adrhi << 8) | cmd.vpu.adrlo;
+    ndw = atom->check(ctx, atom);
 
-	cmd.u = atom->cmd[0];
-	addr = (cmd.vpu.adrhi << 8) | cmd.vpu.adrlo;
-	ndw = atom->check(ctx, atom);
+    r300_emit_vpu(r300, &atom->cmd[1], vpu_count(atom->cmd) * 4, addr);
+}
+
+void r500_emit_fp(struct r300_context *r300,
+                  uint32_t *data,
+                  unsigned len,
+                  uint32_t addr,
+                  unsigned type,
+                  unsigned clamp)
+{
+    BATCH_LOCALS(&r300->radeon);
 
-	BEGIN_BATCH_NO_AUTOSTATE(ndw);
+    addr |= (type << 16);
+    addr |= (clamp << 17);
 
-	ndw -= 5;
-	OUT_BATCH_REGVAL(R300_VAP_PVS_VECTOR_INDX_REG, addr);
-	OUT_BATCH(CP_PACKET0(R300_VAP_PVS_UPLOAD_DATA, ndw-1) | RADEON_ONE_REG_WR);
-	OUT_BATCH_TABLE(&atom->cmd[1], ndw);
-	OUT_BATCH_REGVAL(R300_VAP_PVS_STATE_FLUSH_REG, 0);
-	END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(len + 3);
+    OUT_BATCH(CP_PACKET0(R500_GA_US_VECTOR_INDEX, 0));
+    OUT_BATCH(addr);
+    OUT_BATCH(CP_PACKET0(R500_GA_US_VECTOR_DATA, len-1) | RADEON_ONE_REG_WR);
+    OUT_BATCH_TABLE(data, len);
+    END_BATCH();
 }
 
-void emit_r500fp(GLcontext *ctx, struct radeon_state_atom * atom)
+static void emit_r500fp_atom(GLcontext *ctx, struct radeon_state_atom * atom)
 {
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	BATCH_LOCALS(&r300->radeon);
-	drm_r300_cmd_header_t cmd;
-	uint32_t addr, ndw, sz;
-	int type, clamp;
-
-	ndw = atom->check(ctx, atom);
-
-	cmd.u = atom->cmd[0];
-	sz = cmd.r500fp.count;
-	addr = ((cmd.r500fp.adrhi_flags & 1) << 8) | cmd.r500fp.adrlo;
-	type = !!(cmd.r500fp.adrhi_flags & R500FP_CONSTANT_TYPE);
-	clamp = !!(cmd.r500fp.adrhi_flags & R500FP_CONSTANT_CLAMP);
-
-	addr |= (type << 16);
-	addr |= (clamp << 17);
-
-	BEGIN_BATCH_NO_AUTOSTATE(ndw);
-	OUT_BATCH(CP_PACKET0(R500_GA_US_VECTOR_INDEX, 0));
-	OUT_BATCH(addr);
-	ndw-=3;
-	OUT_BATCH(CP_PACKET0(R500_GA_US_VECTOR_DATA, ndw-1) | RADEON_ONE_REG_WR);
-	OUT_BATCH_TABLE(&atom->cmd[1], ndw);
-	END_BATCH();
+    r300ContextPtr r300 = R300_CONTEXT(ctx);
+    drm_r300_cmd_header_t cmd;
+    uint32_t addr, count;
+    int type, clamp;
+
+    cmd.u = atom->cmd[0];
+    addr = ((cmd.r500fp.adrhi_flags & 1) << 8) | cmd.r500fp.adrlo;
+    type = !!(cmd.r500fp.adrhi_flags & R500FP_CONSTANT_TYPE);
+    clamp = !!(cmd.r500fp.adrhi_flags & R500FP_CONSTANT_CLAMP);
+
+    if (type) {
+        count = r500fp_count(atom->cmd) * 4;
+    } else {
+        count = r500fp_count(atom->cmd) * 6;
+    }
+
+    r500_emit_fp(r300, &atom->cmd[1], count, addr, type, clamp);
 }
 
 static int check_tex_offsets(GLcontext *ctx, struct radeon_state_atom * atom)
@@ -171,7 +187,7 @@ static void emit_tex_offsets(GLcontext *ctx, struct radeon_state_atom * atom)
 		if (t && !t->image_override) {
 			BEGIN_BATCH_NO_AUTOSTATE(4);
 			OUT_BATCH_REGSEQ(R300_TX_OFFSET_0 + (i * 4), 1);
-			OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+			OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 					RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
 			END_BATCH();
 		} else if (!t) {
@@ -258,93 +274,136 @@ static int check_cb_offset(GLcontext *ctx, struct radeon_state_atom * atom)
 	return dw;
 }
 
-static void emit_cb_offset(GLcontext *ctx, struct radeon_state_atom * atom)
+static void emit_scissor(struct r300_context *r300,
+                         unsigned width,
+                         unsigned height)
 {
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	BATCH_LOCALS(&r300->radeon);
-	struct radeon_renderbuffer *rrb;
-	uint32_t cbpitch;
-	uint32_t offset = r300->radeon.state.color.draw_offset;
-	uint32_t dw = 6;
-	int i;
+    int i;
+    BATCH_LOCALS(&r300->radeon);
+    if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
+        BEGIN_BATCH_NO_AUTOSTATE(3);
+        OUT_BATCH_REGSEQ(R300_SC_SCISSORS_TL, 2);
+        OUT_BATCH(0);
+        OUT_BATCH(((width - 1) << R300_SCISSORS_X_SHIFT) |
+                ((height - 1) << R300_SCISSORS_Y_SHIFT));
+        END_BATCH();
+        BEGIN_BATCH_NO_AUTOSTATE(16);
+        for (i = 0; i < 4; i++) {
+            OUT_BATCH_REGSEQ(R300_SC_CLIPRECT_TL_0 + (i * 8), 2);
+            OUT_BATCH((0 << R300_CLIPRECT_X_SHIFT) | (0 << R300_CLIPRECT_Y_SHIFT));
+            OUT_BATCH(((width - 1) << R300_CLIPRECT_X_SHIFT) | ((height - 1) << R300_CLIPRECT_Y_SHIFT));
+        }
+        OUT_BATCH_REGSEQ(R300_SC_CLIP_RULE, 1);
+        OUT_BATCH(0xAAAA);
+        OUT_BATCH_REGSEQ(R300_SC_SCREENDOOR, 1);
+        OUT_BATCH(0xffffff);
+        END_BATCH();
+    } else {
+        BEGIN_BATCH_NO_AUTOSTATE(3);
+        OUT_BATCH_REGSEQ(R300_SC_SCISSORS_TL, 2);
+        OUT_BATCH((R300_SCISSORS_OFFSET << R300_SCISSORS_X_SHIFT) |
+                (R300_SCISSORS_OFFSET << R300_SCISSORS_Y_SHIFT));
+        OUT_BATCH(((width + R300_SCISSORS_OFFSET - 1) << R300_SCISSORS_X_SHIFT) |
+                ((height + R300_SCISSORS_OFFSET - 1) << R300_SCISSORS_Y_SHIFT));
+        END_BATCH();
+        BEGIN_BATCH_NO_AUTOSTATE(16);
+        for (i = 0; i < 4; i++) {
+            OUT_BATCH_REGSEQ(R300_SC_CLIPRECT_TL_0 + (i * 8), 2);
+            OUT_BATCH((R300_SCISSORS_OFFSET << R300_CLIPRECT_X_SHIFT) | (R300_SCISSORS_OFFSET << R300_CLIPRECT_Y_SHIFT));
+            OUT_BATCH(((R300_SCISSORS_OFFSET + width - 1) << R300_CLIPRECT_X_SHIFT) |
+                        ((R300_SCISSORS_OFFSET + height - 1) << R300_CLIPRECT_Y_SHIFT));
+        }
+        OUT_BATCH_REGSEQ(R300_SC_CLIP_RULE, 1);
+        OUT_BATCH(0xAAAA);
+        OUT_BATCH_REGSEQ(R300_SC_SCREENDOOR, 1);
+        OUT_BATCH(0xffffff);
+        END_BATCH();
+    }
+}
 
-	rrb = radeon_get_colorbuffer(&r300->radeon);
-	if (!rrb || !rrb->bo) {
-		fprintf(stderr, "no rrb\n");
-		return;
-	}
+void r300_emit_cb_setup(struct r300_context *r300,
+                        struct radeon_bo *bo,
+                        uint32_t offset,
+                        GLuint format,
+                        unsigned cpp,
+                        unsigned pitch)
+{
+    BATCH_LOCALS(&r300->radeon);
+    uint32_t cbpitch = pitch / cpp;
+    uint32_t dw = 6;
+
+    assert(offset % 32 == 0);
+
+    switch (format) {
+        case MESA_FORMAT_RGB565:
+            assert(_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_RGB565;
+            break;
+        case MESA_FORMAT_RGB565_REV:
+            assert(!_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_RGB565;
+            break;
+        case MESA_FORMAT_ARGB4444:
+            assert(_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_ARGB4444;
+            break;
+        case MESA_FORMAT_ARGB4444_REV:
+            assert(!_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_ARGB4444;
+            break;
+        case MESA_FORMAT_ARGB1555:
+            assert(_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_ARGB1555;
+            break;
+        case MESA_FORMAT_ARGB1555_REV:
+            assert(!_mesa_little_endian());
+            cbpitch |= R300_COLOR_FORMAT_ARGB1555;
+            break;
+        default:
+            if (cpp == 4) {
+                cbpitch |= R300_COLOR_FORMAT_ARGB8888;
+            } else {
+                _mesa_problem(r300->radeon.glCtx, "unexpected format in emit_cb_offset()");;
+            }
+            break;
+    }
 
-        if (RADEON_DEBUG & RADEON_STATE)
-           fprintf(stderr,"rrb is %p %d %dx%d\n", rrb, offset, rrb->base.Width, rrb->base.Height);
-	cbpitch = (rrb->pitch / rrb->cpp);
-	if (rrb->cpp == 4)
-		cbpitch |= R300_COLOR_FORMAT_ARGB8888;
-	else switch (rrb->base._ActualFormat) {
-	case GL_RGB5:
-		cbpitch |= R300_COLOR_FORMAT_RGB565;
-		break;
-	case GL_RGBA4:
-		cbpitch |= R300_COLOR_FORMAT_ARGB4444;
-		break;
-	case GL_RGB5_A1:
-		cbpitch |= R300_COLOR_FORMAT_ARGB1555;
-		break;
-	}
+    if (bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
+        cbpitch |= R300_COLOR_TILE_ENABLE;
 
-	if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
-		cbpitch |= R300_COLOR_TILE_ENABLE;
+    if (r300->radeon.radeonScreen->kernel_mm)
+        dw += 2;
+
+    BEGIN_BATCH_NO_AUTOSTATE(dw);
+    OUT_BATCH_REGSEQ(R300_RB3D_COLOROFFSET0, 1);
+    OUT_BATCH_RELOC(offset, bo, offset, 0, RADEON_GEM_DOMAIN_VRAM, 0);
+    OUT_BATCH_REGSEQ(R300_RB3D_COLORPITCH0, 1);
+    if (!r300->radeon.radeonScreen->kernel_mm)
+        OUT_BATCH(cbpitch);
+    else
+        OUT_BATCH_RELOC(cbpitch, bo, cbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0);
+    END_BATCH();
+}
+
+static void emit_cb_offset_atom(GLcontext *ctx, struct radeon_state_atom * atom)
+{
+    r300ContextPtr r300 = R300_CONTEXT(ctx);
+    struct radeon_renderbuffer *rrb;
+    uint32_t offset = r300->radeon.state.color.draw_offset;
+
+    rrb = radeon_get_colorbuffer(&r300->radeon);
+    if (!rrb || !rrb->bo) {
+        fprintf(stderr, "no rrb\n");
+        return;
+    }
+
+    if (RADEON_DEBUG & RADEON_STATE)
+        fprintf(stderr,"rrb is %p %d %dx%d\n", rrb, offset, rrb->base.Width, rrb->base.Height);
+
+    r300_emit_cb_setup(r300, rrb->bo, offset, rrb->base.Format, rrb->cpp, rrb->pitch);
 
-    	if (r300->radeon.radeonScreen->kernel_mm)
-		dw += 2;
-	BEGIN_BATCH_NO_AUTOSTATE(dw);
-	OUT_BATCH_REGSEQ(R300_RB3D_COLOROFFSET0, 1);
-	OUT_BATCH_RELOC(offset, rrb->bo, offset, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-	OUT_BATCH_REGSEQ(R300_RB3D_COLORPITCH0, 1);
-    	if (!r300->radeon.radeonScreen->kernel_mm)
-		OUT_BATCH(cbpitch);
-	else
-		OUT_BATCH_RELOC(cbpitch, rrb->bo, cbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-	END_BATCH();
     if (r300->radeon.radeonScreen->driScreen->dri2.enabled) {
-        if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
-            BEGIN_BATCH_NO_AUTOSTATE(3);
-            OUT_BATCH_REGSEQ(R300_SC_SCISSORS_TL, 2);
-            OUT_BATCH(0);
-            OUT_BATCH(((rrb->base.Width - 1) << R300_SCISSORS_X_SHIFT) |
-                    ((rrb->base.Height - 1) << R300_SCISSORS_Y_SHIFT));
-            END_BATCH();
-            BEGIN_BATCH_NO_AUTOSTATE(16);
-            for (i = 0; i < 4; i++) {
-                OUT_BATCH_REGSEQ(R300_SC_CLIPRECT_TL_0 + (i * 8), 2);
-                OUT_BATCH((0 << R300_CLIPRECT_X_SHIFT) | (0 << R300_CLIPRECT_Y_SHIFT));
-                OUT_BATCH(((rrb->base.Width - 1) << R300_CLIPRECT_X_SHIFT) | ((rrb->base.Height - 1) << R300_CLIPRECT_Y_SHIFT));
-            }
-            OUT_BATCH_REGSEQ(R300_SC_CLIP_RULE, 1);
-            OUT_BATCH(0xAAAA);
-            OUT_BATCH_REGSEQ(R300_SC_SCREENDOOR, 1);
-            OUT_BATCH(0xffffff);
-            END_BATCH();
-        } else {
-            BEGIN_BATCH_NO_AUTOSTATE(3);
-            OUT_BATCH_REGSEQ(R300_SC_SCISSORS_TL, 2);
-            OUT_BATCH((R300_SCISSORS_OFFSET << R300_SCISSORS_X_SHIFT) |
-                    (R300_SCISSORS_OFFSET << R300_SCISSORS_Y_SHIFT));
-            OUT_BATCH(((rrb->base.Width + R300_SCISSORS_OFFSET - 1) << R300_SCISSORS_X_SHIFT) |
-                    ((rrb->base.Height + R300_SCISSORS_OFFSET - 1) << R300_SCISSORS_Y_SHIFT));
-            END_BATCH();
-            BEGIN_BATCH_NO_AUTOSTATE(16);
-            for (i = 0; i < 4; i++) {
-                OUT_BATCH_REGSEQ(R300_SC_CLIPRECT_TL_0 + (i * 8), 2);
-                OUT_BATCH((R300_SCISSORS_OFFSET << R300_CLIPRECT_X_SHIFT) | (R300_SCISSORS_OFFSET << R300_CLIPRECT_Y_SHIFT));
-                OUT_BATCH(((R300_SCISSORS_OFFSET + rrb->base.Width - 1) << R300_CLIPRECT_X_SHIFT) |
-                          ((R300_SCISSORS_OFFSET + rrb->base.Height - 1) << R300_CLIPRECT_Y_SHIFT));
-            }
-            OUT_BATCH_REGSEQ(R300_SC_CLIP_RULE, 1);
-            OUT_BATCH(0xAAAA);
-            OUT_BATCH_REGSEQ(R300_SC_SCREENDOOR, 1);
-            OUT_BATCH(0xffffff);
-            END_BATCH();
-        }
+        emit_scissor(r300, rrb->base.Width, rrb->base.Height);
     }
 }
 
@@ -440,7 +499,7 @@ static int check_variable(GLcontext *ctx, struct radeon_state_atom *atom)
 	return cnt ? cnt + 1 : 0;
 }
 
-int check_r500fp(GLcontext *ctx, struct radeon_state_atom *atom)
+static int check_r500fp(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	int cnt;
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
@@ -452,7 +511,7 @@ int check_r500fp(GLcontext *ctx, struct radeon_state_atom *atom)
 	return cnt ? (cnt * 6) + extra : 0;
 }
 
-int check_r500fp_const(GLcontext *ctx, struct radeon_state_atom *atom)
+static int check_r500fp_const(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	int cnt;
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
@@ -629,13 +688,13 @@ void r300InitCmdBuf(r300ContextPtr r300)
 		r300->hw.r500fp.cmd[R300_FPI_CMD_0] =
 			cmdr500fp(r300->radeon.radeonScreen, 0, 0, 0, 0);
 		if (r300->radeon.radeonScreen->kernel_mm)
-			r300->hw.r500fp.emit = emit_r500fp;
+			r300->hw.r500fp.emit = emit_r500fp_atom;
 
 		ALLOC_STATE(r500fp_const, r500fp_const, R500_FPP_CMDSIZE, 0);
 		r300->hw.r500fp_const.cmd[R300_FPI_CMD_0] =
 			cmdr500fp(r300->radeon.radeonScreen, 0, 0, 1, 0);
 		if (r300->radeon.radeonScreen->kernel_mm)
-			r300->hw.r500fp_const.emit = emit_r500fp;
+			r300->hw.r500fp_const.emit = emit_r500fp_atom;
 	} else {
 		ALLOC_STATE(fp, always, R300_FP_CMDSIZE, 0);
 		r300->hw.fp.cmd[R300_FP_CMD_0] = cmdpacket0(r300->radeon.radeonScreen, R300_US_CONFIG, 3);
@@ -679,7 +738,7 @@ void r300InitCmdBuf(r300ContextPtr r300)
 	ALLOC_STATE(rop, always, 2, 0);
 	r300->hw.rop.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_ROPCNTL, 1);
 	ALLOC_STATE(cb, cb_offset, R300_CB_CMDSIZE, 0);
-	r300->hw.cb.emit = &emit_cb_offset;
+	r300->hw.cb.emit = &emit_cb_offset_atom;
 	ALLOC_STATE(rb3d_dither_ctl, always, 10, 0);
 	r300->hw.rb3d_dither_ctl.cmd[0] = cmdpacket0(r300->radeon.radeonScreen, R300_RB3D_DITHER_CTL, 9);
 	ALLOC_STATE(rb3d_aaresolve_ctl, always, 2, 0);
@@ -743,20 +802,20 @@ void r300InitCmdBuf(r300ContextPtr r300)
 		r300->hw.vpi.cmd[0] =
 			cmdvpu(r300->radeon.radeonScreen, R300_PVS_CODE_START, 0);
 		if (r300->radeon.radeonScreen->kernel_mm)
-			r300->hw.vpi.emit = emit_vpu;
+			r300->hw.vpi.emit = emit_vpu_state;
 
 		if (is_r500) {
 			ALLOC_STATE(vpp, vpu, R300_VPP_CMDSIZE, 0);
 			r300->hw.vpp.cmd[0] =
 				cmdvpu(r300->radeon.radeonScreen, R500_PVS_CONST_START, 0);
 			if (r300->radeon.radeonScreen->kernel_mm)
-				r300->hw.vpp.emit = emit_vpu;
+				r300->hw.vpp.emit = emit_vpu_state;
 
 			ALLOC_STATE(vps, vpu, R300_VPS_CMDSIZE, 0);
 			r300->hw.vps.cmd[0] =
 				cmdvpu(r300->radeon.radeonScreen, R500_POINT_VPORT_SCALE_OFFSET, 1);
 			if (r300->radeon.radeonScreen->kernel_mm)
-				r300->hw.vps.emit = emit_vpu;
+				r300->hw.vps.emit = emit_vpu_state;
 
 			for (i = 0; i < 6; i++) {
 				ALLOC_STATE(vpucp[i], vpu, R300_VPUCP_CMDSIZE, 0);
@@ -764,20 +823,20 @@ void r300InitCmdBuf(r300ContextPtr r300)
 					cmdvpu(r300->radeon.radeonScreen,
 							R500_PVS_UCP_START + i, 1);
 				if (r300->radeon.radeonScreen->kernel_mm)
-					r300->hw.vpucp[i].emit = emit_vpu;
+					r300->hw.vpucp[i].emit = emit_vpu_state;
 			}
 		} else {
 			ALLOC_STATE(vpp, vpu, R300_VPP_CMDSIZE, 0);
 			r300->hw.vpp.cmd[0] =
 				cmdvpu(r300->radeon.radeonScreen, R300_PVS_CONST_START, 0);
 			if (r300->radeon.radeonScreen->kernel_mm)
-				r300->hw.vpp.emit = emit_vpu;
+				r300->hw.vpp.emit = emit_vpu_state;
 
 			ALLOC_STATE(vps, vpu, R300_VPS_CMDSIZE, 0);
 			r300->hw.vps.cmd[0] =
 				cmdvpu(r300->radeon.radeonScreen, R300_POINT_VPORT_SCALE_OFFSET, 1);
 			if (r300->radeon.radeonScreen->kernel_mm)
-				r300->hw.vps.emit = emit_vpu;
+				r300->hw.vps.emit = emit_vpu_state;
 
 			for (i = 0; i < 6; i++) {
 				ALLOC_STATE(vpucp[i], vpu, R300_VPUCP_CMDSIZE, 0);
@@ -785,7 +844,7 @@ void r300InitCmdBuf(r300ContextPtr r300)
 					cmdvpu(r300->radeon.radeonScreen,
 							R300_PVS_UCP_START + i, 1);
 				if (r300->radeon.radeonScreen->kernel_mm)
-					r300->hw.vpucp[i].emit = emit_vpu;
+					r300->hw.vpucp[i].emit = emit_vpu_state;
 			}
 		}
 	}
diff --git a/src/mesa/drivers/dri/r300/r300_cmdbuf.h b/src/mesa/drivers/dri/r300/r300_cmdbuf.h
index 1b703e518a..0e68da928e 100644
--- a/src/mesa/drivers/dri/r300/r300_cmdbuf.h
+++ b/src/mesa/drivers/dri/r300/r300_cmdbuf.h
@@ -44,14 +44,26 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define FIREAOS_BUFSZ          (3)
 #define SCISSORS_BUFSZ         (3)
 
-extern void r300InitCmdBuf(r300ContextPtr r300);
+void r300InitCmdBuf(r300ContextPtr r300);
 void r300_emit_scissor(GLcontext *ctx);
 
-void emit_vpu(GLcontext *ctx, struct radeon_state_atom * atom);
-int check_vpu(GLcontext *ctx, struct radeon_state_atom *atom);
+void r300_emit_vpu(struct r300_context *ctx,
+                   uint32_t *data,
+                   unsigned len,
+                   uint32_t addr);
 
-void emit_r500fp(GLcontext *ctx, struct radeon_state_atom * atom);
-int check_r500fp(GLcontext *ctx, struct radeon_state_atom *atom);
-int check_r500fp_const(GLcontext *ctx, struct radeon_state_atom *atom);
+void r500_emit_fp(struct r300_context *r300,
+                  uint32_t *data,
+                  unsigned len,
+                  uint32_t addr,
+                  unsigned type,
+                  unsigned clamp);
 
-#endif				/* __R300_CMDBUF_H__ */
+void r300_emit_cb_setup(struct r300_context *r300,
+                        struct radeon_bo *bo,
+                        uint32_t offset,
+                        GLuint format,
+                        unsigned cpp,
+                        unsigned pitch);
+
+#endif /* __R300_CMDBUF_H__ */
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 2c2b16aa98..1f6ccf6ddc 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -55,13 +55,14 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/t_vp_build.h"
 
 #include "drivers/common/driverfuncs.h"
+#include "drivers/common/meta.h"
 
 #include "r300_context.h"
 #include "radeon_context.h"
 #include "radeon_span.h"
+#include "r300_blit.h"
 #include "r300_cmdbuf.h"
 #include "r300_state.h"
-#include "r300_ioctl.h"
 #include "r300_tex.h"
 #include "r300_emit.h"
 #include "r300_swtcl.h"
@@ -90,10 +91,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_ATI_separate_stencil
 #define need_GL_NV_vertex_program
 
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
+void r300_init_texcopy_functions(struct dd_function_table *table);
 
-const struct dri_extension card_extensions[] = {
+static const struct dri_extension card_extensions[] = {
   /* *INDENT-OFF* */
   {"GL_ARB_depth_texture",		NULL},
   {"GL_ARB_fragment_program",		NULL},
@@ -145,7 +147,7 @@ const struct dri_extension card_extensions[] = {
 };
 
 
-const struct dri_extension mm_extensions[] = {
+static const struct dri_extension mm_extensions[] = {
   { "GL_EXT_framebuffer_blit",	GL_EXT_framebuffer_blit_functions },
   { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions },
   { NULL, NULL }
@@ -155,7 +157,7 @@ const struct dri_extension mm_extensions[] = {
  * The GL 2.0 functions are needed to make display lists work with
  * functions added by GL_ATI_separate_stencil.
  */
-const struct dri_extension gl_20_extension[] = {
+static const struct dri_extension gl_20_extension[] = {
   {"GL_VERSION_2_0",			GL_VERSION_2_0_functions },
 };
 
@@ -439,11 +441,11 @@ static void r300InitGLExtensions(GLcontext *ctx)
 	if (r300->options.stencil_two_side_disabled)
 		_mesa_disable_extension(ctx, "GL_EXT_stencil_two_side");
 
-	if (r300->options.s3tc_force_enabled) {
+	if (r300->options.s3tc_force_disabled) {
+		_mesa_disable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+	} else if (ctx->Mesa_DXTn || r300->options.s3tc_force_enabled) {
 		_mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 		_mesa_enable_extension(ctx, "GL_S3_s3tc");
-	} else if (r300->options.s3tc_force_disabled) {
-		_mesa_disable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 	}
 
 	if (!r300->radeon.radeonScreen->drmSupportsOcclusionQueries) {
@@ -451,13 +453,20 @@ static void r300InitGLExtensions(GLcontext *ctx)
 	}
 }
 
+static void r300InitIoctlFuncs(struct dd_function_table *functions)
+{
+	functions->Clear = _mesa_meta_Clear;
+	functions->Finish = radeonFinish;
+	functions->Flush = radeonFlush;
+}
+
 /* Create the device specific rendering context.
  */
 GLboolean r300CreateContext(const __GLcontextModes * glVisual,
-			    __DRIcontextPrivate * driContextPriv,
+			    __DRIcontext * driContextPriv,
 			    void *sharedContextPrivate)
 {
-	__DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+	__DRIscreen *sPriv = driContextPriv->driScreenPriv;
 	radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private);
 	struct dd_function_table functions;
 	r300ContextPtr r300;
@@ -484,6 +493,10 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	radeonInitQueryObjFunctions(&functions);
 	radeonInitBufferObjectFuncs(&functions);
 
+	if (r300->radeon.radeonScreen->kernel_mm) {
+		r300_init_texcopy_functions(&functions);
+	}
+
 	if (!radeonInitContext(&r300->radeon, &functions,
 			       glVisual, driContextPriv,
 			       sharedContextPrivate)) {
@@ -530,6 +543,7 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 		r300InitSwtcl(ctx);
 	}
 
+	r300_blit_init(r300);
 	radeon_fbo_init(&r300->radeon);
 	radeonInitSpanFuncs( ctx );
 	r300InitCmdBuf(r300);
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 518d5cdbf4..546cd8ddde 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -533,14 +533,19 @@ struct r300_context {
 
 	uint32_t fallback;
 
+	struct {
+		struct r300_vertex_program_code vp_code;
+		struct rX00_fragment_program_code fp_code;
+	} blit;
+
 	DECLARE_RENDERINPUTS(render_inputs_bitset);
 };
 
 #define R300_CONTEXT(ctx)		((r300ContextPtr)(ctx->DriverCtx))
 
-extern void r300DestroyContext(__DRIcontextPrivate * driContextPriv);
+extern void r300DestroyContext(__DRIcontext * driContextPriv);
 extern GLboolean r300CreateContext(const __GLcontextModes * glVisual,
-				   __DRIcontextPrivate * driContextPriv,
+				   __DRIcontext * driContextPriv,
 				   void *sharedContextPrivate);
 
 extern void r300InitShaderFuncs(struct dd_function_table *functions);
@@ -549,6 +554,8 @@ extern void r300InitShaderFunctions(r300ContextPtr r300);
 
 extern void r300InitDraw(GLcontext *ctx);
 
+extern void r300_init_texcopy_functions(struct dd_function_table *table);
+
 #define r300PackFloat32 radeonPackFloat32
 #define r300PackFloat24 radeonPackFloat24
 
diff --git a/src/mesa/drivers/dri/r300/r300_draw.c b/src/mesa/drivers/dri/r300/r300_draw.c
index e9968f9ffe..3dcd986e22 100644
--- a/src/mesa/drivers/dri/r300/r300_draw.c
+++ b/src/mesa/drivers/dri/r300/r300_draw.c
@@ -100,7 +100,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		GLubyte *in = (GLubyte *)src_ptr;
 
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo, &r300->ind_buf.bo_offset, size, 4);
-
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		out = (GLuint *)ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 
@@ -111,7 +111,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		if (i < mesa_ind_buf->count) {
 			*out++ = in[i];
 		}
-
+		radeon_bo_unmap(r300->ind_buf.bo);
 #if MESA_BIG_ENDIAN
 	} else { /* if (mesa_ind_buf->type == GL_UNSIGNED_SHORT) */
 		GLushort *in = (GLushort *)src_ptr;
@@ -120,6 +120,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo,
 				     &r300->ind_buf.bo_offset, size, 4);
 
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		out = (GLuint *)ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 
@@ -130,6 +131,7 @@ static void r300FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 		if (i < mesa_ind_buf->count) {
 			*out++ = in[i];
 		}
+		radeon_bo_unmap(r300->ind_buf.bo);
 #endif
 	}
 
@@ -173,10 +175,12 @@ static void r300SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 
 		radeonAllocDmaRegion(&r300->radeon, &r300->ind_buf.bo, &r300->ind_buf.bo_offset, size, 4);
 
+		radeon_bo_map(r300->ind_buf.bo, 1);
 		assert(r300->ind_buf.bo->ptr != NULL);
 		dst_ptr = ADD_POINTERS(r300->ind_buf.bo->ptr, r300->ind_buf.bo_offset);
 		_mesa_memcpy(dst_ptr, src_ptr, size);
 
+		radeon_bo_unmap(r300->ind_buf.bo);
 		r300->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT);
 		r300->ind_buf.count = mesa_ind_buf->count;
 
@@ -242,6 +246,7 @@ static void r300ConvertAttrib(GLcontext *ctx, int count, const struct gl_client_
 	}
 
 	radeonAllocDmaRegion(&r300->radeon, &attr->bo, &attr->bo_offset, sizeof(GLfloat) * input->Size * count, 32);
+	radeon_bo_map(attr->bo, 1);
 	dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
 
 	radeon_print(RADEON_FALLBACKS, RADEON_IMPORTANT,
@@ -280,6 +285,7 @@ static void r300ConvertAttrib(GLcontext *ctx, int count, const struct gl_client_
 			break;
 	}
 
+	radeon_bo_unmap(attr->bo);
 	if (mapped_named_bo) {
 		ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
 	}
@@ -294,6 +300,8 @@ static void r300AlignDataToDword(GLcontext *ctx, const struct gl_client_array *i
 
 	radeonAllocDmaRegion(&r300->radeon, &attr->bo, &attr->bo_offset, size, 32);
 
+	radeon_bo_map(attr->bo, 1);
+
 	if (!input->BufferObj->Pointer) {
 		ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
 		mapped_named_bo = GL_TRUE;
@@ -317,6 +325,7 @@ static void r300AlignDataToDword(GLcontext *ctx, const struct gl_client_array *i
 		ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
 	}
 
+	radeon_bo_unmap(attr->bo);
 	attr->stride = dst_stride;
 }
 
@@ -527,6 +536,7 @@ static void r300AllocDmaRegions(GLcontext *ctx, const struct gl_client_array *in
 				}
 
 				radeonAllocDmaRegion(&r300->radeon, &vbuf->attribs[index].bo, &vbuf->attribs[index].bo_offset, size, 32);
+				radeon_bo_map(vbuf->attribs[index].bo, 1);
 				assert(vbuf->attribs[index].bo->ptr != NULL);
 				dst = (uint32_t *)ADD_POINTERS(vbuf->attribs[index].bo->ptr, vbuf->attribs[index].bo_offset);
 				switch (vbuf->attribs[index].dwords) {
@@ -536,6 +546,7 @@ static void r300AllocDmaRegions(GLcontext *ctx, const struct gl_client_array *in
 					case 4: radeonEmitVec16(dst, input[i]->Ptr, input[i]->StrideB, local_count); break;
 					default: assert(0); break;
 				}
+				radeon_bo_unmap(vbuf->attribs[index].bo);
 
 			}
 		}
diff --git a/src/mesa/drivers/dri/r300/r300_emit.c b/src/mesa/drivers/dri/r300/r300_emit.c
index 07e6223087..15aeaf0514 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.c
+++ b/src/mesa/drivers/dri/r300/r300_emit.c
@@ -49,7 +49,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r300_context.h"
 #include "r300_state.h"
 #include "r300_emit.h"
-#include "r300_ioctl.h"
 #include "r300_render.h"
 #include "r300_swtcl.h"
 
@@ -118,7 +117,7 @@ GLuint r300VAPOutputCntl1(GLcontext * ctx, GLuint vp_writes)
 
 	if (first_free_texcoord > 8) {
 		fprintf(stderr, "\tout of free texcoords\n");
-		_mesa_exit(-1);
+		exit(-1);
 	}
 
 	return ret;
diff --git a/src/mesa/drivers/dri/r300/r300_emit.h b/src/mesa/drivers/dri/r300/r300_emit.h
index 8e57e354d1..a456d8867c 100644
--- a/src/mesa/drivers/dri/r300/r300_emit.h
+++ b/src/mesa/drivers/dri/r300/r300_emit.h
@@ -42,7 +42,6 @@
 #include "main/glheader.h"
 #include "r300_context.h"
 #include "r300_cmdbuf.h"
-#include "radeon_reg.h"
 
 static INLINE uint32_t cmdpacket0(struct radeon_screen *rscrn,
                                   int reg, int count)
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_common.c b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
index 267ee81a7a..2933d31136 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog_common.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_common.c
@@ -120,7 +120,7 @@ static void insert_WPOS_trailer(struct r300_fragment_program_compiler *compiler,
 		return;
 	}
 
-	rc_transform_fragment_wpos(&compiler->Base, FRAG_ATTRIB_WPOS, fp->wpos_attr);
+	rc_transform_fragment_wpos(&compiler->Base, FRAG_ATTRIB_WPOS, fp->wpos_attr, GL_FALSE);
 }
 
 /**
diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.c b/src/mesa/drivers/dri/r300/r300_ioctl.c
deleted file mode 100644
index 5cb04e2bb6..0000000000
--- a/src/mesa/drivers/dri/r300/r300_ioctl.c
+++ /dev/null
@@ -1,782 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.
-Copyright (C) 2004 Nicolai Haehnle.
-All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-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
- *
- * \author Keith Whitwell <keith@tungstengraphics.com>
- *
- * \author Nicolai Haehnle <prefect_@gmx.net>
- */
-
-#include <sched.h>
-#include <errno.h>
-
-#include "main/glheader.h"
-#include "main/imports.h"
-#include "main/macros.h"
-#include "main/context.h"
-#include "main/simple_list.h"
-#include "swrast/swrast.h"
-
-#include "radeon_common.h"
-#include "radeon_lock.h"
-#include "r300_context.h"
-#include "r300_ioctl.h"
-#include "r300_cmdbuf.h"
-#include "r300_state.h"
-#include "r300_vertprog.h"
-#include "radeon_reg.h"
-#include "r300_emit.h"
-#include "r300_context.h"
-
-#include "vblank.h"
-
-#define R200_3D_DRAW_IMMD_2      0xC0003500
-
-#define CLEARBUFFER_COLOR	0x1
-#define CLEARBUFFER_DEPTH	0x2
-#define CLEARBUFFER_STENCIL	0x4
-
-#if 1
-
-/**
- * Fragment program helper macros
- */
-
-/* Produce unshifted source selectors */
-#define FP_TMP(idx) (idx)
-#define FP_CONST(idx) ((idx) | (1 << 5))
-
-/* Produce source/dest selector dword */
-#define FP_SELC_MASK_NO		0
-#define FP_SELC_MASK_X		1
-#define FP_SELC_MASK_Y		2
-#define FP_SELC_MASK_XY		3
-#define FP_SELC_MASK_Z		4
-#define FP_SELC_MASK_XZ		5
-#define FP_SELC_MASK_YZ		6
-#define FP_SELC_MASK_XYZ	7
-
-#define FP_SELC(destidx,regmask,outmask,src0,src1,src2) \
-	(((destidx) << R300_ALU_DSTC_SHIFT) |		\
-	 (FP_SELC_MASK_##regmask << 23) |		\
-	 (FP_SELC_MASK_##outmask << 26) |		\
-	 ((src0) << R300_ALU_SRC0C_SHIFT) |		\
-	 ((src1) << R300_ALU_SRC1C_SHIFT) |		\
-	 ((src2) << R300_ALU_SRC2C_SHIFT))
-
-#define FP_SELA_MASK_NO		0
-#define FP_SELA_MASK_W		1
-
-#define FP_SELA(destidx,regmask,outmask,src0,src1,src2) \
-	(((destidx) << R300_ALU_DSTA_SHIFT) |		\
-	 (FP_SELA_MASK_##regmask << 23) |		\
-	 (FP_SELA_MASK_##outmask << 24) |		\
-	 ((src0) << R300_ALU_SRC0A_SHIFT) |		\
-	 ((src1) << R300_ALU_SRC1A_SHIFT) |		\
-	 ((src2) << R300_ALU_SRC2A_SHIFT))
-
-/* Produce unshifted argument selectors */
-#define FP_ARGC(source)	R300_ALU_ARGC_##source
-#define FP_ARGA(source) R300_ALU_ARGA_##source
-#define FP_ABS(arg) ((arg) | (1 << 6))
-#define FP_NEG(arg) ((arg) ^ (1 << 5))
-
-/* Produce instruction dword */
-#define FP_INSTRC(opcode,arg0,arg1,arg2) \
-	(R300_ALU_OUTC_##opcode | 		\
-	((arg0) << R300_ALU_ARG0C_SHIFT) |	\
-	((arg1) << R300_ALU_ARG1C_SHIFT) |	\
-	((arg2) << R300_ALU_ARG2C_SHIFT))
-
-#define FP_INSTRA(opcode,arg0,arg1,arg2) \
-	(R300_ALU_OUTA_##opcode | 		\
-	((arg0) << R300_ALU_ARG0A_SHIFT) |	\
-	((arg1) << R300_ALU_ARG1A_SHIFT) |	\
-	((arg2) << R300_ALU_ARG2A_SHIFT))
-
-#endif
-
-static void r300EmitClearState(GLcontext * ctx);
-
-static void r300ClearBuffer(r300ContextPtr r300, int flags,
-			    struct radeon_renderbuffer *rrb,
-			    struct radeon_renderbuffer *rrbd)
-{
-	BATCH_LOCALS(&r300->radeon);
-	GLcontext *ctx = r300->radeon.glCtx;
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&r300->radeon);
-	GLuint cbpitch = 0;
-	r300ContextPtr rmesa = r300;
-
-	if (RADEON_DEBUG & RADEON_IOCTL)
-		fprintf(stderr, "%s: buffer %p (%i,%i %ix%i)\n",
-			__FUNCTION__, rrb, dPriv->x, dPriv->y,
-			dPriv->w, dPriv->h);
-
-	if (rrb) {
-		cbpitch = (rrb->pitch / rrb->cpp);
-		if (rrb->cpp == 4)
-			cbpitch |= R300_COLOR_FORMAT_ARGB8888;
-		else
-			cbpitch |= R300_COLOR_FORMAT_RGB565;
-
-		if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE){
-			cbpitch |= R300_COLOR_TILE_ENABLE;
-        }
-	}
-
-	/* TODO in bufmgr */
-	cp_wait(&r300->radeon, R300_WAIT_3D | R300_WAIT_3D_CLEAN);
-	end_3d(&rmesa->radeon);
-
-	if (flags & CLEARBUFFER_COLOR) {
-		assert(rrb != 0);
-		BEGIN_BATCH_NO_AUTOSTATE(6);
-		OUT_BATCH_REGSEQ(R300_RB3D_COLOROFFSET0, 1);
-		OUT_BATCH_RELOC(0, rrb->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-		OUT_BATCH_REGVAL(R300_RB3D_COLORPITCH0, cbpitch);
-		END_BATCH();
-	}
-#if 1
-	if (flags & (CLEARBUFFER_DEPTH | CLEARBUFFER_STENCIL)) {
-		uint32_t zbpitch = (rrbd->pitch / rrbd->cpp);
-		if (rrbd->bo->flags & RADEON_BO_FLAGS_MACRO_TILE){
-			zbpitch |= R300_DEPTHMACROTILE_ENABLE;
-        }
-		if (rrbd->bo->flags & RADEON_BO_FLAGS_MICRO_TILE){
-            zbpitch |= R300_DEPTHMICROTILE_TILED;
-        }
-		BEGIN_BATCH_NO_AUTOSTATE(6);
-		OUT_BATCH_REGSEQ(R300_ZB_DEPTHOFFSET, 1);
-		OUT_BATCH_RELOC(0, rrbd->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-		OUT_BATCH_REGSEQ(R300_ZB_DEPTHPITCH, 1);
-		if (!r300->radeon.radeonScreen->kernel_mm)
-			OUT_BATCH(zbpitch);
-		else
-			OUT_BATCH_RELOC(zbpitch, rrbd->bo, zbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0);
-		END_BATCH();
-	}
-#endif
-	BEGIN_BATCH_NO_AUTOSTATE(6);
-	OUT_BATCH_REGSEQ(RB3D_COLOR_CHANNEL_MASK, 1);
-	if (flags & CLEARBUFFER_COLOR) {
-		OUT_BATCH((ctx->Color.ColorMask[BCOMP] ? RB3D_COLOR_CHANNEL_MASK_BLUE_MASK0 : 0) |
-			  (ctx->Color.ColorMask[GCOMP] ? RB3D_COLOR_CHANNEL_MASK_GREEN_MASK0 : 0) |
-			  (ctx->Color.ColorMask[RCOMP] ? RB3D_COLOR_CHANNEL_MASK_RED_MASK0 : 0) |
-			  (ctx->Color.ColorMask[ACOMP] ? RB3D_COLOR_CHANNEL_MASK_ALPHA_MASK0 : 0));
-	} else {
-		OUT_BATCH(0);
-	}
-
-
-	{
-		uint32_t t1, t2;
-
-		t1 = 0x0;
-		t2 = 0x0;
-
-		if (flags & CLEARBUFFER_DEPTH) {
-			t1 |= R300_Z_ENABLE | R300_Z_WRITE_ENABLE;
-			t2 |=
-			    (R300_ZS_ALWAYS << R300_Z_FUNC_SHIFT);
-		}
-
-		if (flags & CLEARBUFFER_STENCIL) {
-			t1 |= R300_STENCIL_ENABLE;
-			t2 |=
-			    (R300_ZS_ALWAYS <<
-			     R300_S_FRONT_FUNC_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_S_FRONT_SFAIL_OP_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_S_FRONT_ZPASS_OP_SHIFT) |
-			    (R300_ZS_REPLACE <<
-			     R300_S_FRONT_ZFAIL_OP_SHIFT);
-		}
-
-		OUT_BATCH_REGSEQ(R300_ZB_CNTL, 3);
-		OUT_BATCH(t1);
-		OUT_BATCH(t2);
-		OUT_BATCH(((ctx->Stencil.WriteMask[0] & R300_STENCILREF_MASK) <<
-                   R300_STENCILWRITEMASK_SHIFT) |
-			  (ctx->Stencil.Clear & R300_STENCILREF_MASK));
-		END_BATCH();
-	}
-
-	if (!rmesa->radeon.radeonScreen->kernel_mm) {
-		BEGIN_BATCH_NO_AUTOSTATE(9);
-		OUT_BATCH(cmdpacket3(r300->radeon.radeonScreen, R300_CMD_PACKET3_CLEAR));
-		OUT_BATCH_FLOAT32(dPriv->w / 2.0);
-		OUT_BATCH_FLOAT32(dPriv->h / 2.0);
-		OUT_BATCH_FLOAT32(ctx->Depth.Clear);
-		OUT_BATCH_FLOAT32(1.0);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[0]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[1]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[2]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[3]);
-		END_BATCH();
-	} else {
-		OUT_BATCH(CP_PACKET3(R200_3D_DRAW_IMMD_2, 8));
-		OUT_BATCH(R300_PRIM_TYPE_POINT | R300_PRIM_WALK_RING |
-			  (1 << R300_PRIM_NUM_VERTICES_SHIFT));
-		OUT_BATCH_FLOAT32(dPriv->w / 2.0);
-		OUT_BATCH_FLOAT32(dPriv->h / 2.0);
-		OUT_BATCH_FLOAT32(ctx->Depth.Clear);
-		OUT_BATCH_FLOAT32(1.0);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[0]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[1]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[2]);
-		OUT_BATCH_FLOAT32(ctx->Color.ClearColor[3]);
-	}
-
-	r300EmitCacheFlush(rmesa);
-	cp_wait(&r300->radeon, R300_WAIT_3D | R300_WAIT_3D_CLEAN);
-
-	R300_STATECHANGE(r300, cb);
-	R300_STATECHANGE(r300, cmk);
-	R300_STATECHANGE(r300, zs);
-}
-
-static void r300EmitClearState(GLcontext * ctx)
-{
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	BATCH_LOCALS(&r300->radeon);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&r300->radeon);
-	int i;
-	int has_tcl;
-	int is_r500 = 0;
-	GLuint vap_cntl;
-
-	has_tcl = r300->options.hw_tcl_enabled;
-
-	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
-		is_r500 = 1;
-
-	/* State atom dirty tracking is a little subtle here.
-	 *
-	 * On the one hand, we need to make sure base state is emitted
-	 * here if we start with an empty batch buffer, otherwise clear
-	 * works incorrectly with multiple processes. Therefore, the first
-	 * BEGIN_BATCH cannot be a BEGIN_BATCH_NO_AUTOSTATE.
-	 *
-	 * On the other hand, implicit state emission clears the state atom
-	 * dirty bits, so we have to call R300_STATECHANGE later than the
-	 * first BEGIN_BATCH.
-	 *
-	 * The final trickiness is that, because we change state, we need
-	 * to ensure that any stored swtcl primitives are flushed properly
-	 * before we start changing state. See the R300_NEWPRIM in r300Clear
-	 * for this.
-	 */
-	BEGIN_BATCH(31);
-	OUT_BATCH_REGSEQ(R300_VAP_PROG_STREAM_CNTL_0, 1);
-	if (!has_tcl)
-		OUT_BATCH(((((0 << R300_DST_VEC_LOC_SHIFT) | 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)));
-	else
-		OUT_BATCH(((((0 << R300_DST_VEC_LOC_SHIFT) | 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)));
-
-	OUT_BATCH_REGVAL(R300_FG_FOG_BLEND, 0);
-	OUT_BATCH_REGVAL(R300_VAP_PROG_STREAM_CNTL_EXT_0,
-	   ((((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_W << R300_SWIZZLE_SELECT_W_SHIFT) |
-	       ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y | R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT))
-	      << R300_SWIZZLE0_SHIFT) |
-	     (((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_W << R300_SWIZZLE_SELECT_W_SHIFT) |
-	       ((R300_WRITE_ENA_X | R300_WRITE_ENA_Y | R300_WRITE_ENA_Z | R300_WRITE_ENA_W) << R300_WRITE_ENA_SHIFT))
-	      << R300_SWIZZLE1_SHIFT)));
-
-	/* R300_VAP_INPUT_CNTL_0, R300_VAP_INPUT_CNTL_1 */
-	OUT_BATCH_REGSEQ(R300_VAP_VTX_STATE_CNTL, 2);
-	OUT_BATCH((R300_SEL_USER_COLOR_0 << R300_COLOR_0_ASSEMBLY_SHIFT));
-	OUT_BATCH(R300_INPUT_CNTL_POS | R300_INPUT_CNTL_COLOR | R300_INPUT_CNTL_TC0);
-
-	/* comes from fglrx startup of clear */
-	OUT_BATCH_REGSEQ(R300_SE_VTE_CNTL, 2);
-	OUT_BATCH(R300_VTX_W0_FMT | R300_VPORT_X_SCALE_ENA |
-		  R300_VPORT_X_OFFSET_ENA | R300_VPORT_Y_SCALE_ENA |
-		  R300_VPORT_Y_OFFSET_ENA | R300_VPORT_Z_SCALE_ENA |
-		  R300_VPORT_Z_OFFSET_ENA);
-	OUT_BATCH(0x8);
-
-	OUT_BATCH_REGVAL(R300_VAP_PSC_SGN_NORM_CNTL, 0xaaaaaaaa);
-
-	OUT_BATCH_REGSEQ(R300_VAP_OUTPUT_VTX_FMT_0, 2);
-	OUT_BATCH(R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT |
-		  R300_VAP_OUTPUT_VTX_FMT_0__COLOR_0_PRESENT);
-	OUT_BATCH(0); /* no textures */
-
-	OUT_BATCH_REGVAL(R300_TX_ENABLE, 0);
-
-	OUT_BATCH_REGSEQ(R300_SE_VPORT_XSCALE, 6);
-	OUT_BATCH_FLOAT32(1.0);
-	OUT_BATCH_FLOAT32(dPriv->x);
-	OUT_BATCH_FLOAT32(1.0);
-	OUT_BATCH_FLOAT32(dPriv->y);
-	OUT_BATCH_FLOAT32(1.0);
-	OUT_BATCH_FLOAT32(0.0);
-
-	OUT_BATCH_REGVAL(R300_FG_ALPHA_FUNC, 0);
-
-	OUT_BATCH_REGSEQ(R300_RB3D_CBLEND, 2);
-	OUT_BATCH(0x0);
-	OUT_BATCH(0x0);
-	END_BATCH();
-
-	R300_STATECHANGE(r300, vir[0]);
-	R300_STATECHANGE(r300, fogs);
-	R300_STATECHANGE(r300, vir[1]);
-	R300_STATECHANGE(r300, vic);
-	R300_STATECHANGE(r300, vte);
-	R300_STATECHANGE(r300, vof);
-	R300_STATECHANGE(r300, txe);
-	R300_STATECHANGE(r300, vpt);
-	R300_STATECHANGE(r300, at);
-	R300_STATECHANGE(r300, bld);
-	R300_STATECHANGE(r300, ps);
-
-	if (has_tcl) {
-		R300_STATECHANGE(r300, vap_clip_cntl);
-
-		BEGIN_BATCH_NO_AUTOSTATE(2);
-		OUT_BATCH_REGVAL(R300_VAP_CLIP_CNTL, R300_PS_UCP_MODE_CLIP_AS_TRIFAN | R300_CLIP_DISABLE);
-		END_BATCH();
-        }
-
-	BEGIN_BATCH_NO_AUTOSTATE(2);
-	OUT_BATCH_REGVAL(R300_GA_POINT_SIZE,
-		((dPriv->w * 6) << R300_POINTSIZE_X_SHIFT) |
-		((dPriv->h * 6) << R300_POINTSIZE_Y_SHIFT));
-	END_BATCH();
-
-	if (!is_r500) {
-		R300_STATECHANGE(r300, ri);
-		R300_STATECHANGE(r300, rc);
-		R300_STATECHANGE(r300, rr);
-
-		BEGIN_BATCH(14);
-		OUT_BATCH_REGSEQ(R300_RS_IP_0, 8);
-		for (i = 0; i < 8; ++i)
-			OUT_BATCH(R300_RS_SEL_T(1) | R300_RS_SEL_R(2) | R300_RS_SEL_Q(3));
-
-		OUT_BATCH_REGSEQ(R300_RS_COUNT, 2);
-		OUT_BATCH((1 << R300_IC_COUNT_SHIFT) | R300_HIRES_EN);
-		OUT_BATCH(0x0);
-
-		OUT_BATCH_REGVAL(R300_RS_INST_0, R300_RS_INST_COL_CN_WRITE);
-		END_BATCH();
-	} else {
-		R300_STATECHANGE(r300, ri);
-		R300_STATECHANGE(r300, rc);
-		R300_STATECHANGE(r300, rr);
-
-		BEGIN_BATCH(14);
-		OUT_BATCH_REGSEQ(R500_RS_IP_0, 8);
-		for (i = 0; i < 8; ++i) {
-			OUT_BATCH((R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_S_SHIFT) |
-				  (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_T_SHIFT) |
-				  (R500_RS_IP_PTR_K0 << R500_RS_IP_TEX_PTR_R_SHIFT) |
-				  (R500_RS_IP_PTR_K1 << R500_RS_IP_TEX_PTR_Q_SHIFT));
-		}
-
-		OUT_BATCH_REGSEQ(R300_RS_COUNT, 2);
-		OUT_BATCH((1 << R300_IC_COUNT_SHIFT) | R300_HIRES_EN);
-		OUT_BATCH(0x0);
-
-		OUT_BATCH_REGVAL(R500_RS_INST_0, R500_RS_INST_COL_CN_WRITE);
-		END_BATCH();
-	}
-
-	if (!is_r500) {
-		R300_STATECHANGE(r300, fp);
-		R300_STATECHANGE(r300, fpi[0]);
-		R300_STATECHANGE(r300, fpi[1]);
-		R300_STATECHANGE(r300, fpi[2]);
-		R300_STATECHANGE(r300, fpi[3]);
-
-		BEGIN_BATCH(17);
-		OUT_BATCH_REGSEQ(R300_US_CONFIG, 3);
-		OUT_BATCH(0x0);
-		OUT_BATCH(0x0);
-		OUT_BATCH(0x0);
-		OUT_BATCH_REGSEQ(R300_US_CODE_ADDR_0, 4);
-		OUT_BATCH(0x0);
-		OUT_BATCH(0x0);
-		OUT_BATCH(0x0);
-		OUT_BATCH(R300_RGBA_OUT);
-
-		OUT_BATCH_REGVAL(R300_US_ALU_RGB_INST_0,
-			FP_INSTRC(MAD, FP_ARGC(SRC0C_XYZ), FP_ARGC(ONE), FP_ARGC(ZERO)));
-		OUT_BATCH_REGVAL(R300_US_ALU_RGB_ADDR_0,
-			FP_SELC(0, NO, XYZ, FP_TMP(0), 0, 0));
-		OUT_BATCH_REGVAL(R300_US_ALU_ALPHA_INST_0,
-			FP_INSTRA(MAD, FP_ARGA(SRC0A), FP_ARGA(ONE), FP_ARGA(ZERO)));
-		OUT_BATCH_REGVAL(R300_US_ALU_ALPHA_ADDR_0,
-			FP_SELA(0, NO, W, FP_TMP(0), 0, 0));
-		END_BATCH();
-	} else {
-		struct radeon_state_atom r500fp;
-		uint32_t _cmd[10];
-
-		R300_STATECHANGE(r300, fp);
-		R300_STATECHANGE(r300, r500fp);
-
-		BEGIN_BATCH(7);
-		OUT_BATCH_REGSEQ(R500_US_CONFIG, 2);
-		OUT_BATCH(R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO);
-		OUT_BATCH(0x0);
-		OUT_BATCH_REGSEQ(R500_US_CODE_ADDR, 3);
-		OUT_BATCH(R500_US_CODE_START_ADDR(0) | R500_US_CODE_END_ADDR(1));
-		OUT_BATCH(R500_US_CODE_RANGE_ADDR(0) | R500_US_CODE_RANGE_SIZE(1));
-		OUT_BATCH(R500_US_CODE_OFFSET_ADDR(0));
-		END_BATCH();
-
-		r500fp.check = check_r500fp;
-		r500fp.cmd = _cmd;
-		r500fp.cmd[0] = cmdr500fp(r300->radeon.radeonScreen, 0, 1, 0, 0);
-		r500fp.cmd[1] = R500_INST_TYPE_OUT |
-			R500_INST_TEX_SEM_WAIT |
-			R500_INST_LAST |
-			R500_INST_RGB_OMASK_R |
-			R500_INST_RGB_OMASK_G |
-			R500_INST_RGB_OMASK_B |
-			R500_INST_ALPHA_OMASK |
-			R500_INST_RGB_CLAMP |
-			R500_INST_ALPHA_CLAMP;
-		r500fp.cmd[2] = R500_RGB_ADDR0(0) |
-			R500_RGB_ADDR1(0) |
-			R500_RGB_ADDR1_CONST |
-			R500_RGB_ADDR2(0) |
-			R500_RGB_ADDR2_CONST;
-		r500fp.cmd[3] = R500_ALPHA_ADDR0(0) |
-			R500_ALPHA_ADDR1(0) |
-			R500_ALPHA_ADDR1_CONST |
-			R500_ALPHA_ADDR2(0) |
-			R500_ALPHA_ADDR2_CONST;
-		r500fp.cmd[4] = 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;
-		r500fp.cmd[5] = R500_ALPHA_OP_CMP |
-			R500_ALPHA_SWIZ_A_A |
-			R500_ALPHA_SWIZ_B_A;
-		r500fp.cmd[6] = 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;
-
-		r500fp.cmd[7] = 0;
-		if (r300->radeon.radeonScreen->kernel_mm) {
-			emit_r500fp(ctx, &r500fp);
-		} else {
-			int dwords = r500fp.check(ctx,&r500fp);
-			BEGIN_BATCH_NO_AUTOSTATE(dwords);
-			OUT_BATCH_TABLE(r500fp.cmd, dwords);
-			END_BATCH();
-		}
-
-	}
-
-	BEGIN_BATCH(2);
-	OUT_BATCH_REGVAL(R300_VAP_PVS_STATE_FLUSH_REG, 0);
-	END_BATCH();
-
-	if (has_tcl) {
-		vap_cntl = ((10 << R300_PVS_NUM_SLOTS_SHIFT) |
-			(5 << R300_PVS_NUM_CNTLRS_SHIFT) |
-			(12 << R300_VF_MAX_VTX_NUM_SHIFT));
-		if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515)
-			vap_cntl |= R500_TCL_STATE_OPTIMIZATION;
-	} else {
-		vap_cntl = ((10 << R300_PVS_NUM_SLOTS_SHIFT) |
-			(5 << R300_PVS_NUM_CNTLRS_SHIFT) |
-			(5 << R300_VF_MAX_VTX_NUM_SHIFT));
-	}
-
-	if (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV515)
-		vap_cntl |= (2 << R300_PVS_NUM_FPUS_SHIFT);
-	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV530) ||
-		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV560) ||
-		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV570))
-		vap_cntl |= (5 << R300_PVS_NUM_FPUS_SHIFT);
-	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV410) ||
-		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R420))
-		vap_cntl |= (6 << R300_PVS_NUM_FPUS_SHIFT);
-	else if ((r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R520) ||
-		 (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_R580))
-		vap_cntl |= (8 << R300_PVS_NUM_FPUS_SHIFT);
-	else
-		vap_cntl |= (4 << R300_PVS_NUM_FPUS_SHIFT);
-
-	R300_STATECHANGE(r300, vap_cntl);
-
-	BEGIN_BATCH(2);
-	OUT_BATCH_REGVAL(R300_VAP_CNTL, vap_cntl);
-	END_BATCH();
-
-	if (has_tcl) {
-        struct radeon_state_atom vpu;
-        uint32_t _cmd[10];
-		R300_STATECHANGE(r300, pvs);
-		R300_STATECHANGE(r300, vap_flush);
-		R300_STATECHANGE(r300, vpi);
-
-		BEGIN_BATCH(4);
-		OUT_BATCH_REGSEQ(R300_VAP_PVS_CODE_CNTL_0, 3);
-		OUT_BATCH((0 << R300_PVS_FIRST_INST_SHIFT) |
-			  (0 << R300_PVS_XYZW_VALID_INST_SHIFT) |
-			  (1 << R300_PVS_LAST_INST_SHIFT));
-		OUT_BATCH((0 << R300_PVS_CONST_BASE_OFFSET_SHIFT) |
-			  (0 << R300_PVS_MAX_CONST_ADDR_SHIFT));
-		OUT_BATCH(1 << R300_PVS_LAST_VTX_SRC_INST_SHIFT);
-		END_BATCH();
-
-		vpu.check = check_vpu;
-		vpu.cmd = _cmd;
-		vpu.cmd[0] = cmdvpu(r300->radeon.radeonScreen, 0, 2);
-
-		vpu.cmd[1] = PVS_OP_DST_OPERAND(VE_ADD, GL_FALSE, GL_FALSE,
-                                         0, 0xf, PVS_DST_REG_OUT);
-		vpu.cmd[2] = PVS_SRC_OPERAND(0, PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y,
-                                      PVS_SRC_SELECT_Z, PVS_SRC_SELECT_W,
-                                      PVS_SRC_REG_INPUT, NEGATE_NONE);
-		vpu.cmd[3] = PVS_SRC_OPERAND(0, PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_REG_INPUT, NEGATE_NONE);
-		vpu.cmd[4] = 0x0;
-
-		vpu.cmd[5] = PVS_OP_DST_OPERAND(VE_ADD, GL_FALSE, GL_FALSE, 1, 0xf,
-                                         PVS_DST_REG_OUT);
-		vpu.cmd[6] = PVS_SRC_OPERAND(1, PVS_SRC_SELECT_X,
-                                      PVS_SRC_SELECT_Y, PVS_SRC_SELECT_Z,
-                                      PVS_SRC_SELECT_W, PVS_SRC_REG_INPUT,
-                                      NEGATE_NONE);
-		vpu.cmd[7] = PVS_SRC_OPERAND(1, PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_SELECT_FORCE_0,
-                                      PVS_SRC_REG_INPUT, NEGATE_NONE);
-		vpu.cmd[8] = 0x0;
-
-		if (r300->radeon.radeonScreen->kernel_mm) {
-			int dwords = r300->hw.vap_flush.check(ctx,&r300->hw.vap_flush);
-			BEGIN_BATCH_NO_AUTOSTATE(dwords);
-			OUT_BATCH_TABLE(r300->hw.vap_flush.cmd, dwords);
-			END_BATCH();
-			emit_vpu(ctx, &vpu);
-		} else {
-			int dwords = vpu.check(ctx,&vpu);
-			BEGIN_BATCH_NO_AUTOSTATE(dwords);
-			OUT_BATCH_TABLE(vpu.cmd, dwords);
-			END_BATCH();
-		}
-
-	}
-}
-
-static int r300KernelClear(GLcontext *ctx, GLuint flags)
-{
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&r300->radeon);
-	struct radeon_framebuffer *rfb = dPriv->driverPrivate;
-	struct radeon_renderbuffer *rrb;
-	struct radeon_renderbuffer *rrbd;
-	int bits = 0, ret;
-
-	/* Make sure it fits there. */
-	radeon_cs_space_reset_bos(r300->radeon.cmdbuf.cs);
-
-	if (flags & BUFFER_BIT_COLOR0) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_COLOR0);
-		radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
-						  rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM);
-	}
-
-	if (flags & BUFFER_BIT_FRONT_LEFT) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_FRONT_LEFT);
-		radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
-						  rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM);
-	}
-
-	if (flags & BUFFER_BIT_BACK_LEFT) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_BACK_LEFT);
-		radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
-						  rrb->bo, 0, RADEON_GEM_DOMAIN_VRAM);
-	}
-
-	rrbd = radeon_get_renderbuffer(&rfb->base, BUFFER_DEPTH);
-	if (rrbd) {
-		radeon_cs_space_add_persistent_bo(r300->radeon.cmdbuf.cs,
-						  rrbd->bo, 0, RADEON_GEM_DOMAIN_VRAM);
-	}
-
-	ret = radeon_cs_space_check(r300->radeon.cmdbuf.cs);
-	if (ret)
-	  return -1;
-
-	rcommonEnsureCmdBufSpace(&r300->radeon, 421 * 3, __FUNCTION__);
-	if (flags || bits)
-		r300EmitClearState(ctx);
-
-	rrbd = radeon_get_renderbuffer(&rfb->base, BUFFER_DEPTH);
-	if (rrbd && (flags & BUFFER_BIT_DEPTH))
-		bits |= CLEARBUFFER_DEPTH;
-
-	if (rrbd && (flags & BUFFER_BIT_STENCIL))
-		bits |= CLEARBUFFER_STENCIL;
-
-	if (flags & BUFFER_BIT_COLOR0) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_COLOR0);
-		r300ClearBuffer(r300, CLEARBUFFER_COLOR, rrb, NULL);
-		bits = 0;
-	}
-
-	if (flags & BUFFER_BIT_FRONT_LEFT) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_FRONT_LEFT);
-		r300ClearBuffer(r300, bits | CLEARBUFFER_COLOR, rrb, rrbd);
-		bits = 0;
-	}
-
-	if (flags & BUFFER_BIT_BACK_LEFT) {
-		rrb = radeon_get_renderbuffer(&rfb->base, BUFFER_BACK_LEFT);
-		r300ClearBuffer(r300, bits | CLEARBUFFER_COLOR, rrb, rrbd);
-		bits = 0;
-	}
-
-	if (bits)
-		r300ClearBuffer(r300, bits, NULL, rrbd);
-
-	COMMIT_BATCH();
-	return 0;
-}
-
-/**
- * Buffer clear
- */
-static void r300Clear(GLcontext * ctx, GLbitfield mask)
-{
-	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&r300->radeon);
-	const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
-	GLbitfield swrast_mask = 0, tri_mask = 0;
-	int i, ret;
-	struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-	if (RADEON_DEBUG & RADEON_IOCTL)
-		fprintf(stderr, "r300Clear\n");
-
-	if (!r300->radeon.radeonScreen->driScreen->dri2.enabled) {
-		LOCK_HARDWARE(&r300->radeon);
-		UNLOCK_HARDWARE(&r300->radeon);
-		if (dPriv->numClipRects == 0)
-			return;
-	}
-
-	/* Flush swtcl vertices if necessary, because we will change hardware
-	 * state during clear. See also the state-related comment in
-	 * r300EmitClearState.
-	 */
-	R300_NEWPRIM(r300);
-
-	if (colorMask == ~0)
-	  tri_mask |= (mask & BUFFER_BITS_COLOR);
-	else
-	  tri_mask |= (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT));
-
-
-	/* HW stencil */
-	if (mask & BUFFER_BIT_STENCIL) {
-		tri_mask |= BUFFER_BIT_STENCIL;
-	}
-
-	/* HW depth */
-	if (mask & BUFFER_BIT_DEPTH) {
-    	        tri_mask |= BUFFER_BIT_DEPTH;
-	}
-
-	/* If we're doing a tri pass for depth/stencil, include a likely color
-	 * buffer with it.
-	 */
-
-	for (i = 0; i < BUFFER_COUNT; i++) {
-	  GLuint bufBit = 1 << i;
-	  if ((tri_mask) & bufBit) {
-	    if (!fb->Attachment[i].Renderbuffer->ClassID) {
-	      tri_mask &= ~bufBit;
-	      swrast_mask |= bufBit;
-	    }
-	  }
-	}
-
-	/* SW fallback clearing */
-	swrast_mask = mask & ~tri_mask;
-
-	ret = 0;
-	if (tri_mask) {
-		if (r300->radeon.radeonScreen->kernel_mm)
-			radeonUserClear(ctx, tri_mask);
-		else {
-			/* if kernel clear fails due to size restraints fallback */
-			ret = r300KernelClear(ctx, tri_mask);
-			if (ret < 0)
-				swrast_mask |= tri_mask;
-		}
-	}
-
-	if (swrast_mask) {
-		if (RADEON_DEBUG & RADEON_FALLBACKS)
-			fprintf(stderr, "%s: swrast clear, mask: %x\n",
-				__FUNCTION__, swrast_mask);
-		_swrast_Clear(ctx, swrast_mask);
-	}
-}
-
-void r300InitIoctlFuncs(struct dd_function_table *functions)
-{
-	functions->Clear = r300Clear;
-	functions->Finish = radeonFinish;
-	functions->Flush = radeonFlush;
-}
diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.h b/src/mesa/drivers/dri/r300/r300_ioctl.h
deleted file mode 100644
index 3abfa71a6e..0000000000
--- a/src/mesa/drivers/dri/r300/r300_ioctl.h
+++ /dev/null
@@ -1,44 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-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>
- *   Nicolai Haehnle <prefect_@gmx.net>
- */
-
-#ifndef __R300_IOCTL_H__
-#define __R300_IOCTL_H__
-
-#include "r300_context.h"
-#include "radeon_drm.h"
-
-extern void r300InitIoctlFuncs(struct dd_function_table *functions);
-
-#endif				/* __R300_IOCTL_H__ */
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index 623da60333..ea684e7df1 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -1789,6 +1789,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_ALU_DSTC_OUTPUT_X           (1 << 26)
 #       define R300_ALU_DSTC_OUTPUT_Y           (1 << 27)
 #       define R300_ALU_DSTC_OUTPUT_Z           (1 << 28)
+#       define R300_RGB_TARGET(x)               ((x) << 29)
 
 #define R300_US_ALU_ALPHA_ADDR_0                 0x47C0
 #       define R300_ALU_SRC0A_SHIFT             0
@@ -1806,6 +1807,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #       define R300_ALU_DSTA_REG                (1 << 23)
 #       define R300_ALU_DSTA_OUTPUT             (1 << 24)
 #		define R300_ALU_DSTA_DEPTH              (1 << 27)
+#		define R300_ALPHA_TARGET(x)             ((x) << 25)
 
 #define R300_US_ALU_RGB_INST_0                   0x48C0
 #       define R300_ALU_ARGC_SRC0C_XYZ          0
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 3cd38753b8..02c94250a8 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -67,10 +67,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "vbo/vbo_split.h"
 #include "tnl/tnl.h"
 #include "tnl/t_vp_build.h"
-#include "radeon_reg.h"
-#include "radeon_macros.h"
 #include "r300_context.h"
-#include "r300_ioctl.h"
 #include "r300_state.h"
 #include "r300_reg.h"
 #include "r300_tex.h"
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index 3060f49aaf..c51285aad9 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -45,7 +45,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/framebuffer.h"
 #include "main/simple_list.h"
 #include "main/api_arrayelt.h"
-#include "main/texformat.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
@@ -56,7 +55,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/t_vp_build.h"
 
 #include "r300_context.h"
-#include "r300_ioctl.h"
 #include "r300_state.h"
 #include "r300_reg.h"
 #include "r300_emit.h"
@@ -999,7 +997,7 @@ static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
 static void r300UpdateWindow(GLcontext * ctx)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+	__DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
 	GLfloat xoffset = dPriv ? (GLfloat) dPriv->x : 0;
 	GLfloat yoffset = dPriv ? (GLfloat) dPriv->y + dPriv->h : 0;
 	const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1052,7 +1050,7 @@ static void r300DepthRange(GLcontext * ctx, GLclampd nearval, GLclampd farval)
 void r300UpdateViewportOffset(GLcontext * ctx)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+	__DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
 	GLfloat xoffset = (GLfloat) dPriv->x;
 	GLfloat yoffset = (GLfloat) dPriv->y + dPriv->h;
 	const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1313,7 +1311,7 @@ static void r300SetupTextures(GLcontext * ctx)
 		fprintf(stderr,
 			"Aiiee ! mtu=%d is greater than R300_MAX_TEXTURE_UNITS=%d\n",
 			mtu, R300_MAX_TEXTURE_UNITS);
-		_mesa_exit(-1);
+		exit(-1);
 	}
 
 	/* We cannot let disabled tmu offsets pass DRM */
@@ -1770,9 +1768,10 @@ static void r300ResetHwState(r300ContextPtr r300)
 	radeon_firevertices(&r300->radeon);
 
 	r300ColorMask(ctx,
-		      ctx->Color.ColorMask[RCOMP],
-		      ctx->Color.ColorMask[GCOMP],
-		      ctx->Color.ColorMask[BCOMP], ctx->Color.ColorMask[ACOMP]);
+		      ctx->Color.ColorMask[0][RCOMP],
+		      ctx->Color.ColorMask[0][GCOMP],
+		      ctx->Color.ColorMask[0][BCOMP],
+                      ctx->Color.ColorMask[0][ACOMP]);
 
 	r300Enable(ctx, GL_DEPTH_TEST, ctx->Depth.Test);
 	r300DepthMask(ctx, ctx->Depth.Mask);
@@ -2041,7 +2040,7 @@ static const GLfloat *get_fragmentprogram_constant(GLcontext *ctx, GLuint index,
 		}
 
 		case RC_STATE_R300_WINDOW_DIMENSION: {
-			__DRIdrawablePrivate * drawable = radeon_get_drawable(&rmesa->radeon);
+			__DRIdrawable * drawable = radeon_get_drawable(&rmesa->radeon);
 			buffer[0] = drawable->w * 0.5f;	/* width*0.5 */
 			buffer[1] = drawable->h * 0.5f;	/* height*0.5 */
 			buffer[2] = 0.5F;	/* for moving range [-1 1] -> [0 1] */
diff --git a/src/mesa/drivers/dri/r300/r300_swtcl.c b/src/mesa/drivers/dri/r300/r300_swtcl.c
index ee2c71e1a7..93983cee20 100644
--- a/src/mesa/drivers/dri/r300/r300_swtcl.c
+++ b/src/mesa/drivers/dri/r300/r300_swtcl.c
@@ -124,7 +124,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 	}
 
 	if (ctx->Light.Enabled && ctx->Light.Model.TwoSide) {
-		VB->AttribPtr[VERT_ATTRIB_GENERIC0] = VB->ColorPtr[1];
+		VB->AttribPtr[VERT_ATTRIB_GENERIC0] = VB->BackfaceColorPtr;
 		OutputsWritten |= 1 << VERT_RESULT_BFC0;
 #if MESA_LITTLE_ENDIAN
 		EMIT_ATTR( _TNL_ATTRIB_GENERIC0, EMIT_4UB_4F_RGBA );
@@ -134,7 +134,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		ADD_ATTR(VERT_ATTRIB_GENERIC0, R300_DATA_TYPE_BYTE, SWTCL_OVM_COLOR2, SWIZZLE_XYZW, MASK_XYZW, 1);
 #endif
 		if (fp_reads & FRAG_BIT_COL1) {
-			VB->AttribPtr[VERT_ATTRIB_GENERIC1] = VB->SecondaryColorPtr[1];
+			VB->AttribPtr[VERT_ATTRIB_GENERIC1] = VB->BackfaceSecondaryColorPtr;
 			GLuint swiz = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ONE);
 			OutputsWritten |= 1 << VERT_RESULT_BFC1;
 #if MESA_LITTLE_ENDIAN
@@ -159,7 +159,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		int tex_id = rmesa->selected_fp->wpos_attr - FRAG_ATTRIB_TEX0;
 
 		VB->AttribPtr[VERT_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
-		VB->TexCoordPtr[tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
+		VB->AttribPtr[_TNL_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_POS];
 		RENDERINPUTS_SET(tnl->render_inputs_bitset, _TNL_ATTRIB_TEX0 + tex_id);
 	}
 
@@ -167,7 +167,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		int tex_id = rmesa->selected_fp->fog_attr - FRAG_ATTRIB_TEX0;
 
 		VB->AttribPtr[VERT_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
-		VB->TexCoordPtr[tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
+		VB->AttribPtr[_TNL_ATTRIB_TEX0 + tex_id] = VB->AttribPtr[VERT_ATTRIB_FOG];
 		RENDERINPUTS_SET(tnl->render_inputs_bitset, _TNL_ATTRIB_TEX0 + tex_id);
 	}
 
@@ -180,7 +180,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 		GLuint swiz, format, hw_format;
 		for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 			if (fp_reads & FRAG_BIT_TEX(i)) {
-				switch (VB->TexCoordPtr[i]->size) {
+				switch (VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size) {
 					case 1:
 						format = EMIT_1F;
 						hw_format = R300_DATA_TYPE_FLOAT_1;
@@ -215,7 +215,7 @@ void r300ChooseSwtclVertexFormat(GLcontext *ctx, GLuint *_InputsRead,  GLuint *_
 
 	if (first_free_tex >= ctx->Const.MaxTextureUnits) {
 		fprintf(stderr, "\tout of free texcoords to write fog coordinate\n");
-		_mesa_exit(-1);
+		exit(-1);
 	}
 
 	R300_NEWPRIM(rmesa);
@@ -665,11 +665,11 @@ void r300_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
 	r300EmitCacheFlush(rmesa);
 
 	radeonEmitState(&rmesa->radeon);
-    r300_emit_scissor(ctx);
+	r300_emit_scissor(ctx);
 	r300EmitVertexAOS(rmesa,
-			rmesa->radeon.swtcl.vertex_size,
-			first_elem(&rmesa->radeon.dma.reserved)->bo,
-			current_offset);
+			  rmesa->radeon.swtcl.vertex_size,
+			  rmesa->radeon.swtcl.bo,
+			  current_offset);
 
 	r300EmitVbufPrim(rmesa,
 		   rmesa->radeon.swtcl.hw_primitive,
diff --git a/src/mesa/drivers/dri/r300/r300_tex.c b/src/mesa/drivers/dri/r300/r300_tex.c
index 433e5a87d4..963f648cb1 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.c
+++ b/src/mesa/drivers/dri/r300/r300_tex.c
@@ -40,7 +40,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/image.h"
 #include "main/mipmap.h"
 #include "main/simple_list.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
@@ -49,7 +48,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "r300_context.h"
 #include "r300_state.h"
-#include "r300_ioctl.h"
 #include "radeon_mipmap_tree.h"
 #include "r300_tex.h"
 
@@ -196,6 +194,7 @@ static void r300TexParameter(GLcontext * ctx, GLenum target,
 			     GLenum pname, const GLfloat * params)
 {
 	radeonTexObj* t = radeon_tex_obj(texObj);
+	GLenum texBaseFormat;
 
 	if (RADEON_DEBUG & (RADEON_STATE | RADEON_TEXTURE)) {
 		fprintf(stderr, "%s( %s )\n", __FUNCTION__,
@@ -216,30 +215,23 @@ static void r300TexParameter(GLcontext * ctx, GLenum target,
 		break;
 
 	case GL_TEXTURE_BORDER_COLOR:
-		r300SetTexBorderColor(t, texObj->BorderColor);
+		r300SetTexBorderColor(t, texObj->BorderColor.f);
 		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.  Since there's no LOD clamping,
-		 * we just have to rely on loading the right subset of mipmap levels
-		 * to simulate a clamped LOD.
-		 */
-		if (t->mt) {
-			radeon_miptree_unreference(t->mt);
-			t->mt = 0;
-			t->validated = GL_FALSE;
-		}
+		t->validated = GL_FALSE;
 		break;
 
 	case GL_DEPTH_TEXTURE_MODE:
 		if (!texObj->Image[0][texObj->BaseLevel])
 			return;
-		if (texObj->Image[0][texObj->BaseLevel]->TexFormat->BaseFormat
-		    == GL_DEPTH_COMPONENT) {
+		texBaseFormat = texObj->Image[0][texObj->BaseLevel]->_BaseFormat;
+
+		if (texBaseFormat == GL_DEPTH_COMPONENT ||
+			texBaseFormat == GL_DEPTH_STENCIL) {
 			r300SetDepthTexMode(texObj);
 			break;
 		} else {
@@ -268,7 +260,11 @@ static void r300DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 
 	if (rmesa) {
 		int i;
-		radeon_firevertices(&rmesa->radeon);
+		struct radeon_bo *bo;
+		bo = !t->mt ? t->bo : t->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->radeon.cmdbuf.cs)) {
+			radeon_firevertices(&rmesa->radeon);
+		}
 
 		for(i = 0; i < R300_MAX_TEXTURE_UNITS; ++i)
 			if (rmesa->hw.textures[i] == t)
@@ -280,10 +276,8 @@ static void r300DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 		t->bo = NULL;
 	}
 
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
+	radeon_miptree_unreference(&t->mt);
+
 	_mesa_delete_texture_object(ctx, texObj);
 }
 
@@ -313,7 +307,7 @@ static struct gl_texture_object *r300NewTextureObject(GLcontext * ctx,
 	/* Initialize hardware state */
 	r300UpdateTexWrap(t);
 	r300SetTexFilter(t, t->base.MinFilter, t->base.MagFilter, t->base.MaxAnisotropy);
-	r300SetTexBorderColor(t, t->base.BorderColor);
+	r300SetTexBorderColor(t, t->base.BorderColor.f);
 
 	return &t->base;
 }
diff --git a/src/mesa/drivers/dri/r300/r300_tex.h b/src/mesa/drivers/dri/r300/r300_tex.h
index 8a653ea2d1..6ede0fe25c 100644
--- a/src/mesa/drivers/dri/r300/r300_tex.h
+++ b/src/mesa/drivers/dri/r300/r300_tex.h
@@ -51,4 +51,6 @@ extern GLboolean r300ValidateBuffers(GLcontext * ctx);
 
 extern void r300InitTextureFuncs(struct dd_function_table *functions);
 
+int32_t r300TranslateTexFormat(gl_format mesaFormat);
+
 #endif				/* __r300_TEX_H__ */
diff --git a/src/mesa/drivers/dri/r300/r300_texcopy.c b/src/mesa/drivers/dri/r300/r300_texcopy.c
new file mode 100644
index 0000000000..ebc9c05b8a
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/r300_texcopy.c
@@ -0,0 +1,168 @@
+/*
+ * Copyright (C) 2009 Maciej Cencora <m.cencora@gmail.com>
+ *
+ * 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_common.h"
+#include "r300_context.h"
+
+#include "main/image.h"
+#include "main/teximage.h"
+#include "main/texstate.h"
+#include "drivers/common/meta.h"
+
+#include "radeon_mipmap_tree.h"
+#include "r300_blit.h"
+#include <main/debug.h>
+
+// TODO:
+// need to pass correct pitch for small dst textures!
+static GLboolean
+do_copy_texsubimage(GLcontext *ctx,
+                    GLenum target, GLint level,
+                    struct radeon_tex_obj *tobj,
+                    radeon_texture_image *timg,
+                    GLint dstx, GLint dsty,
+                    GLint x, GLint y,
+                    GLsizei width, GLsizei height)
+{
+    struct r300_context *r300 = R300_CONTEXT(ctx);
+    struct radeon_renderbuffer *rrb;
+
+    if (_mesa_get_format_bits(timg->base.TexFormat, GL_DEPTH_BITS) > 0) {
+        rrb = radeon_get_depthbuffer(&r300->radeon);
+    } else {
+        rrb = radeon_get_colorbuffer(&r300->radeon);
+    }
+
+    if (!timg->mt) {
+        radeon_validate_texture_miptree(ctx, &tobj->base);
+    }
+
+    assert(rrb && rrb->bo);
+    assert(timg->mt->bo);
+    assert(timg->base.Width >= dstx + width);
+    assert(timg->base.Height >= dsty + height);
+
+    intptr_t src_offset = rrb->draw_offset;
+    intptr_t dst_offset = radeon_miptree_image_offset(timg->mt, _mesa_tex_target_to_face(target), level);
+
+    if (src_offset % 32 || dst_offset % 32) {
+        return GL_FALSE;
+    }
+
+    if (0) {
+        fprintf(stderr, "%s: copying to face %d, level %d\n",
+                __FUNCTION__, _mesa_tex_target_to_face(target), level);
+        fprintf(stderr, "to: x %d, y %d, offset %d\n", dstx, dsty, (uint32_t) dst_offset);
+        fprintf(stderr, "from (%dx%d) width %d, height %d, offset %d, pitch %d\n",
+                x, y, rrb->base.Width, rrb->base.Height, (uint32_t) src_offset, rrb->pitch/rrb->cpp);
+        fprintf(stderr, "src size %d, dst size %d\n", rrb->bo->size, timg->mt->bo->size);
+
+    }
+
+    /* blit from src buffer to texture */
+    return r300_blit(r300, rrb->bo, src_offset, rrb->base.Format, rrb->pitch/rrb->cpp,
+                     rrb->base.Width, rrb->base.Height, x, y,
+                     timg->mt->bo, dst_offset, timg->base.TexFormat,
+                     timg->base.Width, timg->base.Width, timg->base.Height,
+                     dstx, dsty, width, height, 1);
+}
+
+static void
+r300CopyTexImage2D(GLcontext *ctx, GLenum target, GLint level,
+                   GLenum internalFormat,
+                   GLint x, GLint y, GLsizei width, GLsizei height,
+                   GLint border)
+{
+    struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
+    struct gl_texture_object *texObj =
+        _mesa_select_tex_object(ctx, texUnit, target);
+    struct gl_texture_image *texImage =
+        _mesa_select_tex_image(ctx, texObj, target, level);
+    int srcx, srcy, dstx, dsty;
+
+    if (border)
+        goto fail;
+
+    /* Setup or redefine the texture object, mipmap tree and texture
+     * image.  Don't populate yet.
+     */
+    ctx->Driver.TexImage2D(ctx, target, level, internalFormat,
+                           width, height, border,
+                           GL_RGBA, GL_UNSIGNED_BYTE, NULL,
+                           &ctx->DefaultPacking, texObj, texImage);
+
+    srcx = x;
+    srcy = y;
+    dstx = 0;
+    dsty = 0;
+    if (!_mesa_clip_copytexsubimage(ctx,
+                                    &dstx, &dsty,
+                                    &srcx, &srcy,
+                                    &width, &height)) {
+        return;
+    }
+
+    if (!do_copy_texsubimage(ctx, target, level,
+                             radeon_tex_obj(texObj), (radeon_texture_image *)texImage,
+                             0, 0, x, y, width, height)) {
+        goto fail;
+    }
+
+    return;
+
+fail:
+    _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
+                              width, height, border);
+}
+
+static void
+r300CopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
+                      GLint xoffset, GLint yoffset,
+                      GLint x, GLint y,
+                      GLsizei width, GLsizei height)
+{
+    struct gl_texture_unit *texUnit = _mesa_get_current_tex_unit(ctx);
+    struct gl_texture_object *texObj = _mesa_select_tex_object(ctx, texUnit, target);
+    struct gl_texture_image *texImage = _mesa_select_tex_image(ctx, texObj, target, level);
+
+    if (!do_copy_texsubimage(ctx, target, level,
+                             radeon_tex_obj(texObj), (radeon_texture_image *)texImage,
+                             xoffset, yoffset, x, y, width, height)) {
+
+       //DEBUG_FALLBACKS
+
+        _mesa_meta_CopyTexSubImage2D(ctx, target, level,
+                                     xoffset, yoffset, x, y, width, height);
+    }
+}
+
+
+void r300_init_texcopy_functions(struct dd_function_table *table)
+{
+    table->CopyTexImage2D = r300CopyTexImage2D;
+    table->CopyTexSubImage2D = r300CopyTexSubImage2D;
+}
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/r300_texstate.c b/src/mesa/drivers/dri/r300/r300_texstate.c
index f030451b28..78ff54574f 100644
--- a/src/mesa/drivers/dri/r300/r300_texstate.c
+++ b/src/mesa/drivers/dri/r300/r300_texstate.c
@@ -39,7 +39,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/enums.h"
@@ -47,19 +46,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "r300_context.h"
 #include "r300_state.h"
-#include "r300_ioctl.h"
 #include "radeon_mipmap_tree.h"
 #include "r300_tex.h"
 #include "r300_reg.h"
 
-#define VALID_FORMAT(f) ( ((f) <= MESA_FORMAT_RGBA_DXT5			\
-			   || ((f) >= MESA_FORMAT_RGBA_FLOAT32 &&	\
-			       (f) <= MESA_FORMAT_INTENSITY_FLOAT16))	\
-			  && tx_table[f].flag )
-
-#define _ASSIGN(entry, format)				\
-	[ MESA_FORMAT_ ## entry ] = { format, 0, 1}
-
 /*
  * Note that the _REV formats are the same as the non-REV formats.  This is
  * because the REV and non-REV formats are identical as a byte string, but
@@ -69,66 +59,119 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  * identically.  -- paulus
  */
 
-static const struct tx_table {
-	GLuint format, filter, flag;
-} tx_table[] = {
-	/* *INDENT-OFF* */
+int32_t r300TranslateTexFormat(gl_format mesaFormat)
+{
+	switch (mesaFormat)
+	{
 #ifdef MESA_LITTLE_ENDIAN
-	_ASSIGN(RGBA8888, R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8)),
-	_ASSIGN(RGBA8888_REV, R300_EASY_TX_FORMAT(Z, Y, X, W, W8Z8Y8X8)),
-	_ASSIGN(ARGB8888, R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8)),
-	_ASSIGN(ARGB8888_REV, R300_EASY_TX_FORMAT(W, Z, Y, X, W8Z8Y8X8)),
+		case MESA_FORMAT_RGBA8888:
+			return R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8);
+		case MESA_FORMAT_RGBA8888_REV:
+			return R300_EASY_TX_FORMAT(Z, Y, X, W, W8Z8Y8X8);
+		case MESA_FORMAT_ARGB8888:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
+		case MESA_FORMAT_ARGB8888_REV:
+			return R300_EASY_TX_FORMAT(W, Z, Y, X, W8Z8Y8X8);
 #else
-	_ASSIGN(RGBA8888, R300_EASY_TX_FORMAT(Z, Y, X, W, W8Z8Y8X8)),
-	_ASSIGN(RGBA8888_REV, R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8)),
-	_ASSIGN(ARGB8888, R300_EASY_TX_FORMAT(W, Z, Y, X, W8Z8Y8X8)),
-	_ASSIGN(ARGB8888_REV, R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8)),
+		case MESA_FORMAT_RGBA8888:
+			return R300_EASY_TX_FORMAT(Z, Y, X, W, W8Z8Y8X8);
+		case MESA_FORMAT_RGBA8888_REV:
+			return R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8);
+		case MESA_FORMAT_ARGB8888:
+			return R300_EASY_TX_FORMAT(W, Z, Y, X, W8Z8Y8X8);
+		case MESA_FORMAT_ARGB8888_REV:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
 #endif
-	_ASSIGN(RGB888, R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8)),
-	_ASSIGN(RGB565, R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5)),
-	_ASSIGN(RGB565_REV, R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5)),
-	_ASSIGN(ARGB4444, R300_EASY_TX_FORMAT(X, Y, Z, W, W4Z4Y4X4)),
-	_ASSIGN(ARGB4444_REV, R300_EASY_TX_FORMAT(X, Y, Z, W, W4Z4Y4X4)),
-	_ASSIGN(ARGB1555, R300_EASY_TX_FORMAT(X, Y, Z, W, W1Z5Y5X5)),
-	_ASSIGN(ARGB1555_REV, R300_EASY_TX_FORMAT(X, Y, Z, W, W1Z5Y5X5)),
-	_ASSIGN(AL88, R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8)),
-	_ASSIGN(AL88_REV, R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8)),
-	_ASSIGN(RGB332, R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z3Y3X2)),
-	_ASSIGN(A8, R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X8)),
-	_ASSIGN(L8, R300_EASY_TX_FORMAT(X, X, X, ONE, X8)),
-	_ASSIGN(I8, R300_EASY_TX_FORMAT(X, X, X, X, X8)),
-	_ASSIGN(CI8, R300_EASY_TX_FORMAT(X, X, X, X, X8)),
-	_ASSIGN(YCBCR, R300_EASY_TX_FORMAT(X, Y, Z, ONE, G8R8_G8B8) | R300_TX_FORMAT_YUV_MODE),
-	_ASSIGN(YCBCR_REV, R300_EASY_TX_FORMAT(X, Y, Z, ONE, G8R8_G8B8) | R300_TX_FORMAT_YUV_MODE),
-	_ASSIGN(RGB_DXT1, R300_EASY_TX_FORMAT(X, Y, Z, ONE, DXT1)),
-	_ASSIGN(RGBA_DXT1, R300_EASY_TX_FORMAT(X, Y, Z, W, DXT1)),
-	_ASSIGN(RGBA_DXT3, R300_EASY_TX_FORMAT(X, Y, Z, W, DXT3)),
-	_ASSIGN(RGBA_DXT5, R300_EASY_TX_FORMAT(Y, Z, W, X, DXT5)),
-	_ASSIGN(RGBA_FLOAT32, R300_EASY_TX_FORMAT(Z, Y, X, W, FL_R32G32B32A32)),
-	_ASSIGN(RGBA_FLOAT16, R300_EASY_TX_FORMAT(Z, Y, X, W, FL_R16G16B16A16)),
-	_ASSIGN(RGB_FLOAT32, 0xffffffff),
-	_ASSIGN(RGB_FLOAT16, 0xffffffff),
-	_ASSIGN(ALPHA_FLOAT32, R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, FL_I32)),
-	_ASSIGN(ALPHA_FLOAT16, R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, FL_I16)),
-	_ASSIGN(LUMINANCE_FLOAT32, R300_EASY_TX_FORMAT(X, X, X, ONE, FL_I32)),
-	_ASSIGN(LUMINANCE_FLOAT16, R300_EASY_TX_FORMAT(X, X, X, ONE, FL_I16)),
-	_ASSIGN(LUMINANCE_ALPHA_FLOAT32, R300_EASY_TX_FORMAT(X, X, X, Y, FL_I32A32)),
-	_ASSIGN(LUMINANCE_ALPHA_FLOAT16, R300_EASY_TX_FORMAT(X, X, X, Y, FL_I16A16)),
-	_ASSIGN(INTENSITY_FLOAT32, R300_EASY_TX_FORMAT(X, X, X, X, FL_I32)),
-	_ASSIGN(INTENSITY_FLOAT16, R300_EASY_TX_FORMAT(X, X, X, X, FL_I16)),
-	_ASSIGN(Z16, R300_EASY_TX_FORMAT(X, X, X, X, X16)),
-	_ASSIGN(Z24_S8, R300_EASY_TX_FORMAT(X, X, X, X, X24_Y8)),
-	_ASSIGN(S8_Z24, R300_EASY_TX_FORMAT(Y, Y, Y, Y, X24_Y8)),
-	_ASSIGN(Z32, R300_EASY_TX_FORMAT(X, X, X, X, X32)),
-	/* EXT_texture_sRGB */
-	_ASSIGN(SRGBA8, R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8) | R300_TX_FORMAT_GAMMA),
-	_ASSIGN(SLA8, R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8) | R300_TX_FORMAT_GAMMA),
-	_ASSIGN(SL8, R300_EASY_TX_FORMAT(X, X, X, ONE, X8) | R300_TX_FORMAT_GAMMA),
-	/* *INDENT-ON* */
+		case MESA_FORMAT_XRGB8888:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
+		case MESA_FORMAT_RGB888:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
+		case MESA_FORMAT_RGB565:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
+		case MESA_FORMAT_RGB565_REV:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
+		case MESA_FORMAT_ARGB4444:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W4Z4Y4X4);
+		case MESA_FORMAT_ARGB4444_REV:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W4Z4Y4X4);
+		case MESA_FORMAT_ARGB1555:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W1Z5Y5X5);
+		case MESA_FORMAT_ARGB1555_REV:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, W1Z5Y5X5);
+		case MESA_FORMAT_AL88:
+			return R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8);
+		case MESA_FORMAT_AL88_REV:
+			return R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8);
+		case MESA_FORMAT_RGB332:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z3Y3X2);
+		case MESA_FORMAT_A8:
+			return R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X8);
+		case MESA_FORMAT_L8:
+			return R300_EASY_TX_FORMAT(X, X, X, ONE, X8);
+		case MESA_FORMAT_I8:
+			return R300_EASY_TX_FORMAT(X, X, X, X, X8);
+		case MESA_FORMAT_CI8:
+			return R300_EASY_TX_FORMAT(X, X, X, X, X8);
+		case MESA_FORMAT_YCBCR:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, G8R8_G8B8) | R300_TX_FORMAT_YUV_MODE;
+		case MESA_FORMAT_YCBCR_REV:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, G8R8_G8B8) | R300_TX_FORMAT_YUV_MODE;
+		case MESA_FORMAT_RGB_DXT1:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, DXT1);
+		case MESA_FORMAT_RGBA_DXT1:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, DXT1);
+		case MESA_FORMAT_RGBA_DXT3:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, DXT3);
+		case MESA_FORMAT_RGBA_DXT5:
+			return R300_EASY_TX_FORMAT(Y, Z, W, X, DXT5);
+		case MESA_FORMAT_RGBA_FLOAT32:
+			return R300_EASY_TX_FORMAT(Z, Y, X, W, FL_R32G32B32A32);
+		case MESA_FORMAT_RGBA_FLOAT16:
+			return R300_EASY_TX_FORMAT(Z, Y, X, W, FL_R16G16B16A16);
+		case MESA_FORMAT_ALPHA_FLOAT32:
+			return R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, FL_I32);
+		case MESA_FORMAT_ALPHA_FLOAT16:
+			return R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, FL_I16);
+		case MESA_FORMAT_LUMINANCE_FLOAT32:
+			return R300_EASY_TX_FORMAT(X, X, X, ONE, FL_I32);
+		case MESA_FORMAT_LUMINANCE_FLOAT16:
+			return R300_EASY_TX_FORMAT(X, X, X, ONE, FL_I16);
+		case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32:
+			return R300_EASY_TX_FORMAT(X, X, X, Y, FL_I32A32);
+		case MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16:
+			return R300_EASY_TX_FORMAT(X, X, X, Y, FL_I16A16);
+		case MESA_FORMAT_INTENSITY_FLOAT32:
+			return R300_EASY_TX_FORMAT(X, X, X, X, FL_I32);
+		case MESA_FORMAT_INTENSITY_FLOAT16:
+			return R300_EASY_TX_FORMAT(X, X, X, X, FL_I16);
+		case MESA_FORMAT_Z16:
+			return R300_EASY_TX_FORMAT(X, X, X, X, X16);
+		case MESA_FORMAT_Z24_S8:
+			return R300_EASY_TX_FORMAT(X, X, X, X, X24_Y8);
+		case MESA_FORMAT_S8_Z24:
+			return R300_EASY_TX_FORMAT(Y, Y, Y, Y, X24_Y8);
+		case MESA_FORMAT_Z32:
+			return R300_EASY_TX_FORMAT(X, X, X, X, X32);
+		/* EXT_texture_sRGB */
+		case MESA_FORMAT_SRGBA8:
+			return R300_EASY_TX_FORMAT(Y, Z, W, X, W8Z8Y8X8) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SLA8:
+			return R300_EASY_TX_FORMAT(X, X, X, Y, Y8X8) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SL8:
+			return R300_EASY_TX_FORMAT(X, X, X, ONE, X8) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SRGB_DXT1:
+			return R300_EASY_TX_FORMAT(X, Y, Z, ONE, DXT1) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SRGBA_DXT1:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, DXT1) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SRGBA_DXT3:
+			return R300_EASY_TX_FORMAT(X, Y, Z, W, DXT3) | R300_TX_FORMAT_GAMMA;
+		case MESA_FORMAT_SRGBA_DXT5:
+			return R300_EASY_TX_FORMAT(Y, Z, W, X, DXT5) | R300_TX_FORMAT_GAMMA;
+		default:
+			return -1;
+	}
 };
 
-#undef _ASSIGN
-
 void r300SetDepthTexMode(struct gl_texture_object *tObj)
 {
 	static const GLuint formats[3][3] = {
@@ -138,9 +181,9 @@ void r300SetDepthTexMode(struct gl_texture_object *tObj)
 			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X16),
 		},
 		{
-			R300_EASY_TX_FORMAT(X, X, X, ONE, X24_Y8),
-			R300_EASY_TX_FORMAT(X, X, X, X, X24_Y8),
-			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, X, X24_Y8),
+			R300_EASY_TX_FORMAT(Y, Y, Y, ONE, X24_Y8),
+			R300_EASY_TX_FORMAT(Y, Y, Y, Y, X24_Y8),
+			R300_EASY_TX_FORMAT(ZERO, ZERO, ZERO, Y, X24_Y8),
 		},
 		{
 			R300_EASY_TX_FORMAT(X, X, X, ONE, X32),
@@ -156,11 +199,11 @@ void r300SetDepthTexMode(struct gl_texture_object *tObj)
 
 	t = radeon_tex_obj(tObj);
 
-	switch (tObj->Image[0][tObj->BaseLevel]->TexFormat->MesaFormat) {
+	switch (tObj->Image[0][tObj->BaseLevel]->TexFormat) {
 	case MESA_FORMAT_Z16:
 		format = formats[0];
 		break;
-	case MESA_FORMAT_Z24_S8:
+	case MESA_FORMAT_S8_Z24:
 		format = formats[1];
 		break;
 	case MESA_FORMAT_Z32:
@@ -203,32 +246,29 @@ void r300SetDepthTexMode(struct gl_texture_object *tObj)
 static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 {
 	const struct gl_texture_image *firstImage;
-	int firstlevel = t->mt ? t->mt->firstLevel : 0;
-	    
-	firstImage = t->base.Image[0][firstlevel];
+	firstImage = t->base.Image[0][t->minLod];
 
-	if (!t->image_override
-	    && VALID_FORMAT(firstImage->TexFormat->MesaFormat)) {
-		if (firstImage->TexFormat->BaseFormat == GL_DEPTH_COMPONENT) {
+	if (!t->image_override) {
+		if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
 			r300SetDepthTexMode(&t->base);
 		} else {
-			t->pp_txformat = tx_table[firstImage->TexFormat->MesaFormat].format;
+			int32_t txformat = r300TranslateTexFormat(firstImage->TexFormat);
+			if (txformat < 0) {
+				_mesa_problem(rmesa->radeon.glCtx, "%s: Invalid format %s",
+							  __FUNCTION__, _mesa_get_format_name(firstImage->TexFormat));
+				exit(1);
+			}
+			t->pp_txformat = (uint32_t) txformat;
 		}
-
-		t->pp_txfilter |= tx_table[firstImage->TexFormat->MesaFormat].filter;
-	} else if (!t->image_override) {
-		_mesa_problem(NULL, "unexpected texture format in %s",
-			      __FUNCTION__);
-		return;
 	}
 
 	if (t->image_override && t->bo)
 		return;
 
-	t->pp_txsize = (((firstImage->Width - 1) << R300_TX_WIDTHMASK_SHIFT)
-			| ((firstImage->Height - 1) << R300_TX_HEIGHTMASK_SHIFT)
-			| ((firstImage->DepthLog2) << R300_TX_DEPTHMASK_SHIFT)
-			| ((t->mt->lastLevel - t->mt->firstLevel) << R300_TX_MAX_MIP_LEVEL_SHIFT));
+	t->pp_txsize = (((R300_TX_WIDTHMASK_MASK & ((firstImage->Width - 1) << R300_TX_WIDTHMASK_SHIFT)))
+			| ((R300_TX_HEIGHTMASK_MASK & ((firstImage->Height - 1) << R300_TX_HEIGHTMASK_SHIFT)))
+			| ((R300_TX_DEPTHMASK_MASK & ((firstImage->DepthLog2) << R300_TX_DEPTHMASK_SHIFT)))
+			| ((R300_TX_MAX_MIP_LEVEL_MASK & ((t->maxLod - t->minLod) << R300_TX_MAX_MIP_LEVEL_SHIFT))));
 
 	t->tile_bits = 0;
 
@@ -239,7 +279,7 @@ static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 
 
 	if (t->base.Target == GL_TEXTURE_RECTANGLE_NV) {
-		unsigned int align = (64 / t->mt->bpp) - 1;
+		unsigned int align = (64 / _mesa_get_format_bytes(firstImage->TexFormat)) - 1;
 		t->pp_txsize |= R300_TX_SIZE_TXPITCH_EN;
 		if (!t->image_override)
 			t->pp_txpitch = ((firstImage->Width + align) & ~align) - 1;
@@ -248,8 +288,12 @@ static void setup_hardware_state(r300ContextPtr rmesa, radeonTexObj *t)
 	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
 	    if (firstImage->Width > 2048)
 		t->pp_txpitch |= R500_TXWIDTH_BIT11;
+            else
+		t->pp_txpitch &= ~R500_TXWIDTH_BIT11;
 	    if (firstImage->Height > 2048)
 		t->pp_txpitch |= R500_TXHEIGHT_BIT11;
+            else
+		t->pp_txpitch &= ~R500_TXHEIGHT_BIT11;
 	}
 }
 
@@ -355,18 +399,15 @@ void r300SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	switch (depth) {
 	case 32:
 		t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
-		t->pp_txfilter |= tx_table[2].filter;
 		pitch_val /= 4;
 		break;
 	case 24:
 	default:
 		t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
-		t->pp_txfilter |= tx_table[4].filter;
 		pitch_val /= 4;
 		break;
 	case 16:
 		t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
-		t->pp_txfilter |= tx_table[5].filter;
 		pitch_val /= 2;
 		break;
 	}
@@ -407,18 +448,7 @@ void r300SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
     	    return;
     	}
 
-	radeon_update_renderbuffers(pDRICtx, dPriv);
-	/* back & depth buffer are useless free them right away */
-	rb = (void*)rfb->base.Attachment[BUFFER_DEPTH].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-        rb->bo = NULL;
-	}
-	rb = (void*)rfb->base.Attachment[BUFFER_BACK_LEFT].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-		rb->bo = NULL;
-	}
+	radeon_update_renderbuffers(pDRICtx, dPriv, GL_TRUE);
 	rb = rfb->color_rb[0];
 	if (rb->bo == NULL) {
 		/* Failed to BO for the buffer */
@@ -434,20 +464,13 @@ void r300SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
-	texImage->TexFormat = radeonChooseTextureFormat(radeon->glCtx,
-							internalFormat,
-							type, format, 0);
 	rImage->bo = rb->bo;
 	radeon_bo_ref(rImage->bo);
 	t->bo = rb->bo;
@@ -463,32 +486,33 @@ void r300SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 			t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
 		else
 			t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, W, W8Z8Y8X8);
-		t->pp_txfilter |= tx_table[2].filter;
 		pitch_val /= 4;
 		break;
 	case 3:
 	default:
 		t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, ONE, W8Z8Y8X8);
-		t->pp_txfilter |= tx_table[4].filter;
 		pitch_val /= 4;
 		break;
 	case 2:
 		t->pp_txformat = R300_EASY_TX_FORMAT(X, Y, Z, ONE, Z5Y6X5);
-		t->pp_txfilter |= tx_table[5].filter;
 		pitch_val /= 2;
 		break;
 	}
 	pitch_val--;
-	t->pp_txsize = ((rb->base.Width - 1) << R300_TX_WIDTHMASK_SHIFT) |
-              ((rb->base.Height - 1) << R300_TX_HEIGHTMASK_SHIFT);
+	t->pp_txsize = (((R300_TX_WIDTHMASK_MASK & ((rb->base.Width - 1) << R300_TX_WIDTHMASK_SHIFT)))
+			| ((R300_TX_HEIGHTMASK_MASK & ((rb->base.Height - 1) << R300_TX_HEIGHTMASK_SHIFT))));
 	t->pp_txsize |= R300_TX_SIZE_TXPITCH_EN;
 	t->pp_txpitch |= pitch_val;
 
 	if (rmesa->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
 	    if (rb->base.Width > 2048)
 		t->pp_txpitch |= R500_TXWIDTH_BIT11;
+            else
+		t->pp_txpitch &= ~R500_TXWIDTH_BIT11;
 	    if (rb->base.Height > 2048)
 		t->pp_txpitch |= R500_TXHEIGHT_BIT11;
+            else
+		t->pp_txpitch &= ~R500_TXHEIGHT_BIT11;
 	}
 	t->validated = GL_TRUE;
 	_mesa_unlock_texture(radeon->glCtx, texObj);
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 43629d643b..aa98a049aa 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -62,12 +62,6 @@ static int r300VertexProgUpdateParams(GLcontext * ctx, struct r300_vertex_progra
 		}
 	}
 
-	if (vp->code.constants.Count * 4 > VSF_MAX_FRAGMENT_LENGTH) {
-		/* Should have checked this earlier... */
-		fprintf(stderr, "%s:Params exhausted\n", __FUNCTION__);
-		_mesa_exit(-1);
-	}
-
 	for(i = 0; i < vp->code.constants.Count; ++i) {
 		const float * src = 0;
 		const struct rc_constant * constant = &vp->code.constants.Constants[i];
@@ -281,6 +275,11 @@ static struct r300_vertex_program *build_program(GLcontext *ctx,
 	}
 
 	r3xx_compile_vertex_program(&compiler);
+
+	if (vp->code.constants.Count > ctx->Const.VertexProgram.MaxParameters) {
+		rc_error(&compiler.Base, "Program exceeds constant buffer size limit\n");
+	}
+
 	vp->error = compiler.Base.Error;
 
 	vp->Base->Base.InputsRead = vp->code.InputsRead;
@@ -334,7 +333,6 @@ struct r300_vertex_program * r300SelectAndTranslateVertexShader(GLcontext *ctx)
 #define bump_vpu_count(ptr, new_count)   do { \
 		drm_r300_cmd_header_t* _p=((drm_r300_cmd_header_t*)(ptr)); \
 		int _nc=(new_count)/4; \
-		assert(_nc < 256); \
 		if(_nc>_p->vpu.count)_p->vpu.count=_nc; \
 	} while(0)
 
@@ -367,7 +365,7 @@ static void r300EmitVertexProgram(r300ContextPtr r300, int dest, struct r300_ver
 			break;
 		default:
 			fprintf(stderr, "%s:%s don't know how to handle dest %04x\n", __FILE__, __FUNCTION__, dest);
-			_mesa_exit(-1);
+			exit(-1);
 	}
 }
 
diff --git a/src/mesa/drivers/dri/r300/radeon_bo.c b/src/mesa/drivers/dri/r300/radeon_bo.c
new file mode 120000
index 0000000000..9448ffee54
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_bo.c
@@ -0,0 +1 @@
+../radeon/radeon_bo.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/radeon_bo_int_drm.h b/src/mesa/drivers/dri/r300/radeon_bo_int_drm.h
new file mode 120000
index 0000000000..029450928b
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_bo_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_bo_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/radeon_cs.c b/src/mesa/drivers/dri/r300/radeon_cs.c
new file mode 120000
index 0000000000..66b7ad1eb0
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_cs.c
@@ -0,0 +1 @@
+../radeon/radeon_cs.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r300/radeon_cs_int_drm.h b/src/mesa/drivers/dri/r300/radeon_cs_int_drm.h
new file mode 120000
index 0000000000..462f5245d0
--- /dev/null
+++ b/src/mesa/drivers/dri/r300/radeon_cs_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_cs_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r600/Makefile b/src/mesa/drivers/dri/r600/Makefile
index 7d5a7b1ab6..26f47b7268 100644
--- a/src/mesa/drivers/dri/r600/Makefile
+++ b/src/mesa/drivers/dri/r600/Makefile
@@ -14,7 +14,7 @@ EGL_SOURCES = server/radeon_egl.c
 endif
 
 ifeq ($(RADEON_LDFLAGS),)
-CS_SOURCES = radeon_cs_space_drm.c
+CS_SOURCES = radeon_cs_space_drm.c radeon_bo.c radeon_cs.c
 endif
 
 COMMON_SOURCES = \
@@ -76,4 +76,3 @@ DRI_LIB_DEPS += $(RADEON_LDFLAGS)
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.c b/src/mesa/drivers/dri/r600/r600_cmdbuf.c
index 3cfe03a45f..370bb04f93 100644
--- a/src/mesa/drivers/dri/r600/r600_cmdbuf.c
+++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.c
@@ -52,29 +52,49 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_mipmap_tree.h"
 #include "radeon_reg.h"
 
+#ifdef HAVE_LIBDRM_RADEON
+#include "radeon_cs_int.h"
+#else
+#include "radeon_cs_int_drm.h"
+#endif
 
+struct r600_cs_manager_legacy
+{
+    struct radeon_cs_manager    base;
+    struct radeon_context       *ctx;
+    /* hack for scratch stuff */
+    uint32_t                    pending_age;
+    uint32_t                    pending_count;
+};
+
+struct r600_cs_reloc_legacy {
+    struct radeon_cs_reloc  base;
+    uint32_t                cindices;
+    uint32_t                *indices;
+    uint32_t                *reloc_indices;
+};
 
-static struct radeon_cs * r600_cs_create(struct radeon_cs_manager *csm,
-                                   uint32_t ndw)
+static struct radeon_cs_int *r600_cs_create(struct radeon_cs_manager *csm,
+					    uint32_t ndw)
 {
-    struct radeon_cs *cs;
+    struct radeon_cs_int *csi;
 
-    cs = (struct radeon_cs*)calloc(1, sizeof(struct radeon_cs));
-    if (cs == NULL) {
+    csi = (struct radeon_cs_int*)calloc(1, sizeof(struct radeon_cs_int));
+    if (csi == NULL) {
         return NULL;
     }
-    cs->csm = csm;
-    cs->ndw = (ndw + 0x3FF) & (~0x3FF);
-    cs->packets = (uint32_t*)malloc(4*cs->ndw);
-    if (cs->packets == NULL) {
-        free(cs);
+    csi->csm = csm;
+    csi->ndw = (ndw + 0x3FF) & (~0x3FF);
+    csi->packets = (uint32_t*)malloc(4*csi->ndw);
+    if (csi->packets == NULL) {
+        free(csi);
         return NULL;
     }
-    cs->relocs_total_size = 0;
-    return cs;
+    csi->relocs_total_size = 0;
+    return csi;
 }
 
-static int r600_cs_write_reloc(struct radeon_cs *cs,
+static int r600_cs_write_reloc(struct radeon_cs_int *csi,
 			       struct radeon_bo *bo,
 			       uint32_t read_domain,
 			       uint32_t write_domain,
@@ -83,7 +103,7 @@ static int r600_cs_write_reloc(struct radeon_cs *cs,
     struct r600_cs_reloc_legacy *relocs;
     int i;
 
-    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
+    relocs = (struct r600_cs_reloc_legacy *)csi->relocs;
     /* check domains */
     if ((read_domain && write_domain) || (!read_domain && !write_domain)) {
         /* in one CS a bo can only be in read or write domain but not
@@ -98,7 +118,7 @@ static int r600_cs_write_reloc(struct radeon_cs *cs,
         return -EINVAL;
     }
     /* check if bo is already referenced */
-    for(i = 0; i < cs->crelocs; i++) {
+    for(i = 0; i < csi->crelocs; i++) {
         uint32_t *indices;
         uint32_t *reloc_indices;
 
@@ -129,109 +149,108 @@ static int r600_cs_write_reloc(struct radeon_cs *cs,
             }
             relocs[i].indices = indices;
             relocs[i].reloc_indices = reloc_indices;
-            relocs[i].indices[relocs[i].cindices - 1] = cs->cdw;
-            relocs[i].reloc_indices[relocs[i].cindices - 1] = cs->cdw;
-            cs->section_cdw += 2;
-	    cs->cdw += 2;
+            relocs[i].indices[relocs[i].cindices - 1] = csi->cdw;
+            relocs[i].reloc_indices[relocs[i].cindices - 1] = csi->cdw;
+            csi->section_cdw += 2;
+	    csi->cdw += 2;
 
             return 0;
         }
     }
     /* add bo to reloc */
     relocs = (struct r600_cs_reloc_legacy*)
-             realloc(cs->relocs,
-                     sizeof(struct r600_cs_reloc_legacy) * (cs->crelocs + 1));
+             realloc(csi->relocs,
+                     sizeof(struct r600_cs_reloc_legacy) * (csi->crelocs + 1));
     if (relocs == NULL) {
         return -ENOMEM;
     }
-    cs->relocs = relocs;
-    relocs[cs->crelocs].base.bo = bo;
-    relocs[cs->crelocs].base.read_domain = read_domain;
-    relocs[cs->crelocs].base.write_domain = write_domain;
-    relocs[cs->crelocs].base.flags = flags;
-    relocs[cs->crelocs].indices = (uint32_t*)malloc(4);
-    relocs[cs->crelocs].reloc_indices = (uint32_t*)malloc(4);
-    if ( (relocs[cs->crelocs].indices == NULL) || (relocs[cs->crelocs].reloc_indices == NULL) )
+    csi->relocs = relocs;
+    relocs[csi->crelocs].base.bo = bo;
+    relocs[csi->crelocs].base.read_domain = read_domain;
+    relocs[csi->crelocs].base.write_domain = write_domain;
+    relocs[csi->crelocs].base.flags = flags;
+    relocs[csi->crelocs].indices = (uint32_t*)malloc(4);
+    relocs[csi->crelocs].reloc_indices = (uint32_t*)malloc(4);
+    if ( (relocs[csi->crelocs].indices == NULL) || (relocs[csi->crelocs].reloc_indices == NULL) )
     {
         return -ENOMEM;
     }
 
-    relocs[cs->crelocs].indices[0] = cs->cdw;
-    relocs[cs->crelocs].reloc_indices[0] = cs->cdw;
-    cs->section_cdw += 2;
-    cs->cdw += 2;
-    relocs[cs->crelocs].cindices = 1;
-    cs->relocs_total_size += radeon_bo_legacy_relocs_size(bo);
-    cs->crelocs++;
+    relocs[csi->crelocs].indices[0] = csi->cdw;
+    relocs[csi->crelocs].reloc_indices[0] = csi->cdw;
+    csi->section_cdw += 2;
+    csi->cdw += 2;
+    relocs[csi->crelocs].cindices = 1;
+    csi->relocs_total_size += radeon_bo_legacy_relocs_size(bo);
+    csi->crelocs++;
 
     radeon_bo_ref(bo);
 
     return 0;
 }
 
-static int r600_cs_begin(struct radeon_cs *cs,
+static int r600_cs_begin(struct radeon_cs_int *csi,
                     uint32_t ndw,
                     const char *file,
                     const char *func,
                     int line)
 {
-    if (cs->section) {
+    if (csi->section_ndw) {
         fprintf(stderr, "CS already in a section(%s,%s,%d)\n",
-                cs->section_file, cs->section_func, cs->section_line);
+                csi->section_file, csi->section_func, csi->section_line);
         fprintf(stderr, "CS can't start section(%s,%s,%d)\n",
                 file, func, line);
         return -EPIPE;
     }
 
-    cs->section = 1;
-    cs->section_ndw = ndw;
-    cs->section_cdw = 0;
-    cs->section_file = file;
-    cs->section_func = func;
-    cs->section_line = line;
+    csi->section_ndw = ndw;
+    csi->section_cdw = 0;
+    csi->section_file = file;
+    csi->section_func = func;
+    csi->section_line = line;
 
-    if (cs->cdw + ndw > cs->ndw) {
+    if (csi->cdw + ndw > csi->ndw) {
         uint32_t tmp, *ptr;
 	int num = (ndw > 0x400) ? ndw : 0x400;
 
-        tmp = (cs->cdw + num + 0x3FF) & (~0x3FF);
-        ptr = (uint32_t*)realloc(cs->packets, 4 * tmp);
+        tmp = (csi->cdw + num + 0x3FF) & (~0x3FF);
+        ptr = (uint32_t*)realloc(csi->packets, 4 * tmp);
         if (ptr == NULL) {
             return -ENOMEM;
         }
-        cs->packets = ptr;
-        cs->ndw = tmp;
+        csi->packets = ptr;
+        csi->ndw = tmp;
     }
 
     return 0;
 }
 
-static int r600_cs_end(struct radeon_cs *cs,
+static int r600_cs_end(struct radeon_cs_int *csi,
                   const char *file,
                   const char *func,
                   int line)
 
 {
-    if (!cs->section) {
+    if (!csi->section_ndw) {
         fprintf(stderr, "CS no section to end at (%s,%s,%d)\n",
                 file, func, line);
         return -EPIPE;
     }
-    cs->section = 0;
 
-    if ( cs->section_ndw != cs->section_cdw ) {
+    if ( csi->section_ndw != csi->section_cdw ) {
         fprintf(stderr, "CS section size missmatch start at (%s,%s,%d) %d vs %d\n",
-                cs->section_file, cs->section_func, cs->section_line, cs->section_ndw, cs->section_cdw);
-        fprintf(stderr, "cs->section_ndw = %d, cs->cdw = %d, cs->section_cdw = %d \n",
-                cs->section_ndw, cs->cdw, cs->section_cdw);
+                csi->section_file, csi->section_func, csi->section_line, csi->section_ndw, csi->section_cdw);
+        fprintf(stderr, "csi->section_ndw = %d, csi->cdw = %d, csi->section_cdw = %d \n",
+                csi->section_ndw, csi->cdw, csi->section_cdw);
         fprintf(stderr, "CS section end at (%s,%s,%d)\n",
                 file, func, line);
         return -EPIPE;
     }
+    csi->section_ndw = 0;
 
-    if (cs->cdw > cs->ndw) {
+    if (csi->cdw > csi->ndw) {
 	    fprintf(stderr, "CS section overflow at (%s,%s,%d) cdw %d ndw %d\n",
-		    cs->section_file, cs->section_func, cs->section_line,cs->cdw,cs->ndw);
+		    csi->section_file, csi->section_func, csi->section_line,csi->cdw,csi->ndw);
 	    fprintf(stderr, "CS section end at (%s,%s,%d)\n",
 		    file, func, line);
 	    assert(0);
@@ -240,21 +259,21 @@ static int r600_cs_end(struct radeon_cs *cs,
     return 0;
 }
 
-static int r600_cs_process_relocs(struct radeon_cs *cs, 
+static int r600_cs_process_relocs(struct radeon_cs_int *csi, 
                                   uint32_t * reloc_chunk,
                                   uint32_t * length_dw_reloc_chunk) 
 {
-    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)csi->csm;
     struct r600_cs_reloc_legacy *relocs;
     int i, j, r;
 
     uint32_t offset_dw = 0;
 
-    csm = (struct r600_cs_manager_legacy*)cs->csm;
-    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
+    csm = (struct r600_cs_manager_legacy*)csi->csm;
+    relocs = (struct r600_cs_reloc_legacy *)csi->relocs;
 restart:
-    for (i = 0; i < cs->crelocs; i++) {
-            uint32_t soffset, eoffset, asicoffset;
+    for (i = 0; i < csi->crelocs; i++) {
+            uint32_t soffset, eoffset;
 
             r = radeon_bo_legacy_validate(relocs[i].base.bo,
 					  &soffset, &eoffset);
@@ -262,32 +281,20 @@ restart:
 		    goto restart;
             }
             if (r) {
-		    fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
+		    fprintf(stderr, "invalid bo(%p) [0x%08X, 0x%08X]\n",
 			    relocs[i].base.bo, soffset, eoffset);
 		    return r;
             }
-            asicoffset = soffset;
 
 	    for (j = 0; j < relocs[i].cindices; j++) {
-		    if (asicoffset >= eoffset) {
-			    /*                radeon_bo_debug(relocs[i].base.bo, 12); */
-			    fprintf(stderr, "validated %p [0x%08X, 0x%08X]\n",
-				    relocs[i].base.bo, soffset, eoffset);
-			    fprintf(stderr, "above end: %p 0x%08X 0x%08X\n",
-				    relocs[i].base.bo,
-				    cs->packets[relocs[i].indices[j]],
-				    eoffset);
-			    exit(0);
-			    return -EINVAL;
-		    }
 		    /* pkt3 nop header in ib chunk */
-		    cs->packets[relocs[i].reloc_indices[j]] = 0xC0001000;
+		    csi->packets[relocs[i].reloc_indices[j]] = 0xC0001000;
 		    /* reloc index in ib chunk */
-		    cs->packets[relocs[i].reloc_indices[j] + 1] = offset_dw;
+		    csi->packets[relocs[i].reloc_indices[j] + 1] = offset_dw;
 	    }
 
 	    /* asic offset in reloc chunk */ /* see alex drm r600_nomm_relocate */
-	    reloc_chunk[offset_dw] = asicoffset;
+	    reloc_chunk[offset_dw] = soffset;
 	    reloc_chunk[offset_dw + 3] = 0;
 
 	    offset_dw += 4;
@@ -298,14 +305,14 @@ restart:
     return 0;
 }
 
-static int r600_cs_set_age(struct radeon_cs *cs) /* -------------- */
+static int r600_cs_set_age(struct radeon_cs_int *csi) /* -------------- */
 {
-    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)csi->csm;
     struct r600_cs_reloc_legacy *relocs;
     int i;
 
-    relocs = (struct r600_cs_reloc_legacy *)cs->relocs;
-    for (i = 0; i < cs->crelocs; i++) {
+    relocs = (struct r600_cs_reloc_legacy *)csi->relocs;
+    for (i = 0; i < csi->crelocs; i++) {
         radeon_bo_legacy_pending(relocs[i].base.bo, csm->pending_age);
         radeon_bo_unref(relocs[i].base.bo);
     }
@@ -313,21 +320,21 @@ static int r600_cs_set_age(struct radeon_cs *cs) /* -------------- */
 }
 
 #if 0
-static void dump_cmdbuf(struct radeon_cs *cs)
+static void dump_cmdbuf(struct radeon_cs_int *csi)
 {
 	int i;
 	fprintf(stderr,"--start--\n");
-	for (i = 0; i < cs->cdw; i++){
-		fprintf(stderr,"0x%08x\n", cs->packets[i]);
+	for (i = 0; i < csi->cdw; i++){
+		fprintf(stderr,"0x%08x\n", csi->packets[i]);
 	}
 	fprintf(stderr,"--end--\n");
 
 }
 #endif
 
-static int r600_cs_emit(struct radeon_cs *cs)
+static int r600_cs_emit(struct radeon_cs_int *csi)
 {
-    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)cs->csm;
+    struct r600_cs_manager_legacy *csm = (struct r600_cs_manager_legacy*)csi->csm;
     struct drm_radeon_cs       cs_cmd;
     struct drm_radeon_cs_chunk cs_chunk[2];
     uint32_t length_dw_reloc_chunk;
@@ -341,9 +348,9 @@ static int r600_cs_emit(struct radeon_cs *cs)
 
     csm->pending_count = 1;
 
-    reloc_chunk = (uint32_t*)calloc(1, cs->crelocs * 4 * 4);
+    reloc_chunk = (uint32_t*)calloc(1, csi->crelocs * 4 * 4);
 
-    r = r600_cs_process_relocs(cs, reloc_chunk, &length_dw_reloc_chunk);
+    r = r600_cs_process_relocs(csi, reloc_chunk, &length_dw_reloc_chunk);
     if (r) {
 	free(reloc_chunk);
         return 0;
@@ -351,8 +358,8 @@ static int r600_cs_emit(struct radeon_cs *cs)
 
     /* raw ib chunk */
     cs_chunk[0].chunk_id   = RADEON_CHUNK_ID_IB;
-    cs_chunk[0].length_dw  = cs->cdw;
-    cs_chunk[0].chunk_data = (unsigned long)(cs->packets);
+    cs_chunk[0].length_dw  = csi->cdw;
+    cs_chunk[0].chunk_data = (unsigned long)(csi->packets);
 
     /* reloc chaunk */
     cs_chunk[1].chunk_id   = RADEON_CHUNK_ID_RELOCS;
@@ -370,7 +377,7 @@ static int r600_cs_emit(struct radeon_cs *cs)
 
     do 
     {
-        r = drmCommandWriteRead(cs->csm->fd, DRM_RADEON_CS, &cs_cmd, sizeof(cs_cmd));
+        r = drmCommandWriteRead(csi->csm->fd, DRM_RADEON_CS, &cs_cmd, sizeof(cs_cmd));
         retry++;
     } while (r == -EAGAIN && retry < 1000);
 
@@ -381,11 +388,11 @@ static int r600_cs_emit(struct radeon_cs *cs)
 
     csm->pending_age = cs_cmd.cs_id;
 
-    r600_cs_set_age(cs);
+    r600_cs_set_age(csi);
 
-    cs->csm->read_used = 0;
-    cs->csm->vram_write_used = 0;
-    cs->csm->gart_write_used = 0;
+    csi->csm->read_used = 0;
+    csi->csm->vram_write_used = 0;
+    csi->csm->gart_write_used = 0;
 
     free(reloc_chunk);
 
@@ -405,35 +412,34 @@ static void inline r600_cs_free_reloc(void *relocs_p, int crelocs)
     }
 }
 
-static int r600_cs_destroy(struct radeon_cs *cs)
+static int r600_cs_destroy(struct radeon_cs_int *csi)
 {
-    r600_cs_free_reloc(cs->relocs, cs->crelocs);
-    free(cs->relocs);
-    free(cs->packets);
-    free(cs);
+    r600_cs_free_reloc(csi->relocs, csi->crelocs);
+    free(csi->relocs);
+    free(csi->packets);
+    free(csi);
     return 0;
 }
 
-static int r600_cs_erase(struct radeon_cs *cs)
+static int r600_cs_erase(struct radeon_cs_int *csi)
 {
-    r600_cs_free_reloc(cs->relocs, cs->crelocs);
-    free(cs->relocs);
-    cs->relocs_total_size = 0;
-    cs->relocs = NULL;
-    cs->crelocs = 0;
-    cs->cdw = 0;
-    cs->section = 0;
+    r600_cs_free_reloc(csi->relocs, csi->crelocs);
+    free(csi->relocs);
+    csi->relocs_total_size = 0;
+    csi->relocs = NULL;
+    csi->crelocs = 0;
+    csi->cdw = 0;
     return 0;
 }
 
-static int r600_cs_need_flush(struct radeon_cs *cs)
+static int r600_cs_need_flush(struct radeon_cs_int *csi)
 {
     /* this function used to flush when the BO usage got to
      * a certain size, now the higher levels handle this better */
     return 0;
 }
 
-static void r600_cs_print(struct radeon_cs *cs, FILE *file)
+static void r600_cs_print(struct radeon_cs_int *csi, FILE *file)
 {
 }
 
diff --git a/src/mesa/drivers/dri/r600/r600_cmdbuf.h b/src/mesa/drivers/dri/r600/r600_cmdbuf.h
index eba43d37b6..dff0009699 100644
--- a/src/mesa/drivers/dri/r600/r600_cmdbuf.h
+++ b/src/mesa/drivers/dri/r600/r600_cmdbuf.h
@@ -118,22 +118,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R600_IT_SET_CTL_CONST                     0x00006F00
 #define R600_IT_SURFACE_BASE_UPDATE               0x00007300
 
-struct r600_cs_manager_legacy
-{
-    struct radeon_cs_manager    base;
-    struct radeon_context       *ctx;
-    /* hack for scratch stuff */
-    uint32_t                    pending_age;
-    uint32_t                    pending_count;
-};
-
-struct r600_cs_reloc_legacy {
-    struct radeon_cs_reloc  base;
-    uint32_t                cindices;
-    uint32_t                *indices;
-    uint32_t                *reloc_indices;
-};
-
 struct radeon_cs_manager * r600_radeon_cs_manager_legacy_ctor(struct radeon_context *ctx);
 
 /**
diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c
index 969144ba12..cb549497f5 100644
--- a/src/mesa/drivers/dri/r600/r600_context.c
+++ b/src/mesa/drivers/dri/r600/r600_context.c
@@ -64,6 +64,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r600_cmdbuf.h"
 #include "r600_emit.h"
 #include "radeon_bocs_wrapper.h"
+#include "radeon_queryobj.h"
 
 #include "r700_state.h"
 #include "r700_ioctl.h"
@@ -73,11 +74,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "utils.h"
 #include "xmlpool.h"		/* for symbolic values of enum-type options */
 
-/* hw_tcl_on derives from future_hw_tcl_on when its safe to change it. */
-int future_hw_tcl_on = 1;
-int hw_tcl_on = 1;
+#define R600_ENABLE_GLSL_TEST 1
 
 #define need_GL_VERSION_2_0
+#define need_GL_ARB_occlusion_query
 #define need_GL_ARB_point_parameters
 #define need_GL_ARB_vertex_program
 #define need_GL_EXT_blend_equation_separate
@@ -92,14 +92,15 @@ int hw_tcl_on = 1;
 #define need_GL_ATI_separate_stencil
 #define need_GL_NV_vertex_program
 
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
-extern const struct tnl_pipeline_stage *r700_pipeline[];
-
-const struct dri_extension card_extensions[] = {
+static const struct dri_extension card_extensions[] = {
   /* *INDENT-OFF* */
+  {"GL_ARB_depth_clamp",                NULL},
   {"GL_ARB_depth_texture",		NULL},
   {"GL_ARB_fragment_program",		NULL},
+  {"GL_ARB_fragment_program_shadow",	NULL},
+  {"GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions},
   {"GL_ARB_multitexture",		NULL},
   {"GL_ARB_point_parameters",		GL_ARB_point_parameters_functions},
   {"GL_ARB_shadow",			NULL},
@@ -111,6 +112,7 @@ const struct dri_extension card_extensions[] = {
   {"GL_ARB_texture_env_crossbar",	NULL},
   {"GL_ARB_texture_env_dot3",		NULL},
   {"GL_ARB_texture_mirrored_repeat",	NULL},
+  {"GL_ARB_texture_non_power_of_two",   NULL},
   {"GL_ARB_vertex_program",		GL_ARB_vertex_program_functions},
   {"GL_EXT_blend_equation_separate",	GL_EXT_blend_equation_separate_functions},
   {"GL_EXT_blend_func_separate",	GL_EXT_blend_func_separate_functions},
@@ -131,6 +133,7 @@ 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},
@@ -146,7 +149,7 @@ const struct dri_extension card_extensions[] = {
 };
 
 
-const struct dri_extension mm_extensions[] = {
+static const struct dri_extension mm_extensions[] = {
   { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions },
   { NULL, NULL }
 };
@@ -155,21 +158,29 @@ const struct dri_extension mm_extensions[] = {
  * The GL 2.0 functions are needed to make display lists work with
  * functions added by GL_ATI_separate_stencil.
  */
-const struct dri_extension gl_20_extension[] = {
+static const struct dri_extension gl_20_extension[] = {
+#ifdef R600_ENABLE_GLSL_TEST
+    {"GL_ARB_shading_language_100",			GL_VERSION_2_0_functions },
+#else
   {"GL_VERSION_2_0",			GL_VERSION_2_0_functions },
+#endif /* R600_ENABLE_GLSL_TEST */
+  {NULL, NULL}
 };
 
-
-static void r600RunPipeline(GLcontext * ctx)
-{
-    _mesa_lock_context_textures(ctx);
-
-    if (ctx->NewState)
-        _mesa_update_state_locked(ctx);
-    
-    _tnl_run_pipeline(ctx);
-    _mesa_unlock_context_textures(ctx);
-}
+static const struct tnl_pipeline_stage *r600_pipeline[] = {
+	/* Catch any t&l fallbacks
+	 */
+	&_tnl_vertex_transform_stage,
+	&_tnl_normal_transform_stage,
+	&_tnl_lighting_stage,
+	&_tnl_fog_coordinate_stage,
+	&_tnl_texgen_stage,
+	&_tnl_texture_transform_stage,
+	&_tnl_point_attenuation_stage,
+	&_tnl_vertex_program_stage,
+	&_tnl_render_stage,
+	0,
+};
 
 static void r600_get_lock(radeonContextPtr rmesa)
 {
@@ -180,7 +191,7 @@ static void r600_get_lock(radeonContextPtr rmesa)
 		if (!rmesa->radeonScreen->kernel_mm)
 			radeon_bo_legacy_texture_age(rmesa->radeonScreen->bom);
 	}
-}		  
+}
 
 static void r600_vtbl_emit_cs_header(struct radeon_cs *cs, radeonContextPtr rmesa)
 {
@@ -202,6 +213,24 @@ static void r600_fallback(GLcontext *ctx, GLuint bit, GLboolean mode)
 		context->radeon.Fallback &= ~bit;
 }
 
+static void r600_emit_query_finish(radeonContextPtr radeon)
+{
+	context_t *context = (context_t*) radeon;
+	BATCH_LOCALS(&context->radeon);
+
+	struct radeon_query_object *query = radeon->query.current;
+
+	BEGIN_BATCH_NO_AUTOSTATE(4 + 2);
+	R600_OUT_BATCH(CP_PACKET3(R600_IT_EVENT_WRITE, 2));
+	R600_OUT_BATCH(ZPASS_DONE);
+	R600_OUT_BATCH(query->curr_offset + 8); /* hw writes qwords */
+	R600_OUT_BATCH(0x00000000);
+	R600_OUT_BATCH_RELOC(VGT_EVENT_INITIATOR, query->bo, 0, 0, RADEON_GEM_DOMAIN_GTT, 0);
+	END_BATCH();
+	assert(query->curr_offset < RADEON_QUERY_PAGE_SIZE);
+	query->emitted_begin = GL_FALSE;
+}
+
 static void r600_init_vtbl(radeonContextPtr radeon)
 {
 	radeon->vtbl.get_lock = r600_get_lock;
@@ -210,71 +239,12 @@ static void r600_init_vtbl(radeonContextPtr radeon)
 	radeon->vtbl.swtcl_flush = NULL;
 	radeon->vtbl.pre_emit_atoms = r600_vtbl_pre_emit_atoms;
 	radeon->vtbl.fallback = r600_fallback;
+	radeon->vtbl.emit_query_finish = r600_emit_query_finish;
 }
 
-/* Create the device specific rendering context.
- */
-GLboolean r600CreateContext(const __GLcontextModes * glVisual,
-			    __DRIcontextPrivate * driContextPriv,
-			    void *sharedContextPrivate)
+static void r600InitConstValues(GLcontext *ctx, radeonScreenPtr screen)
 {
-	__DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-	radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private);
-	struct dd_function_table functions;
-	context_t *r600;
-	GLcontext *ctx;
-
-	assert(glVisual);
-	assert(driContextPriv);
-	assert(screen);
-
-	/* Allocate the R600 context */
-	r600 = (context_t*) CALLOC(sizeof(*r600));
-	if (!r600) {
-		radeon_error("Failed to allocate memory for context.\n");
-		return GL_FALSE;
-	}
-
-	if (!(screen->chip_flags & RADEON_CHIPSET_TCL))
-		hw_tcl_on = future_hw_tcl_on = 0;
-
-	r600_init_vtbl(&r600->radeon);
-	/* Parse configuration files.
-	 * Do this here so that initialMaxAnisotropy is set before we create
-	 * the default textures.
-	 */
-	driParseConfigFiles(&r600->radeon.optionCache, &screen->optionCache,
-			    screen->driScreen->myNum, "r600");
-
-	r600->radeon.initialMaxAnisotropy = driQueryOptionf(&r600->radeon.optionCache,
-						     "def_max_anisotropy");
-
-	/* Init default driver functions then plug in our R600-specific functions
-	 * (the texture functions are especially important)
-	 */
-	_mesa_init_driver_functions(&functions);
-
-	r700InitStateFuncs(&functions);
-	r600InitTextureFuncs(&functions);
-	r700InitShaderFuncs(&functions);
-	r700InitIoctlFuncs(&functions);
-    radeonInitBufferObjectFuncs(&functions);
-
-	if (!radeonInitContext(&r600->radeon, &functions,
-			       glVisual, driContextPriv,
-			       sharedContextPrivate)) {
-		radeon_error("Initializing context failed.\n");
-		FREE(r600);
-		return GL_FALSE;
-	}
-
-	/* Init r600 context data */
-	/* Set the maximum texture size small enough that we can guarentee that
-	 * all texture units can bind a maximal texture and have them both in
-	 * texturable memory at once.
-	 */
-
-	ctx = r600->radeon.glCtx;
+	context_t *r600 = R700_CONTEXT(ctx);
 
 	ctx->Const.MaxTextureImageUnits =
 	    driQueryOptioni(&r600->radeon.optionCache, "texture_image_units");
@@ -299,38 +269,7 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 	ctx->Const.MaxLineWidth   = 0xffff / 8.0;
 	ctx->Const.MaxLineWidthAA = 0xffff / 8.0;
 
-	/* Needs further modifications */
-#if 0
-	ctx->Const.MaxArrayLockSize =
-	    ( /*512 */ RADEON_BUFFER_SIZE * 16 * 1024) / (4 * 4);
-#endif
-
-	ctx->Const.MaxDrawBuffers = 1;
-
-	/* Initialize the software rasterizer and helper modules.
-	 */
-	_swrast_CreateContext(ctx);
-	_vbo_CreateContext(ctx);
-	_tnl_CreateContext(ctx);
-	_swsetup_CreateContext(ctx);
-	_swsetup_Wakeup(ctx);
-	_ae_create_context(ctx);
-
-	/* Install the customized pipeline:
-	 */
-	_tnl_destroy_pipeline(ctx);
-	_tnl_install_pipeline(ctx, r700_pipeline);
-
-	/* Try and keep materials and vertices separate:
-	 */
-/* 	_tnl_isolate_materials(ctx, GL_TRUE); */
-
-	/* Configure swrast and TNL to match hardware characteristics:
-	 */
-	_swrast_allow_pixel_fog(ctx, GL_FALSE);
-	_swrast_allow_vertex_fog(ctx, GL_TRUE);
-	_tnl_allow_pixel_fog(ctx, GL_FALSE);
-	_tnl_allow_vertex_fog(ctx, GL_TRUE);
+	ctx->Const.MaxDrawBuffers = 1; /* hw supports 8 */
 
 	/* 256 for reg-based consts, inline consts also supported */
 	ctx->Const.VertexProgram.MaxInstructions = 8192; /* in theory no limit */
@@ -354,15 +293,38 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 	ctx->Const.FragmentProgram.MaxNativeInstructions = 8192;
 	ctx->Const.FragmentProgram.MaxNativeTexIndirections = 8; /* ??? */
 	ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0;	/* and these are?? */
-	ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
-	ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
+}
 
-	radeon_init_debug();
+static void r600ParseOptions(context_t *r600, radeonScreenPtr screen)
+{
+	/* Parse configuration files.
+	 * Do this here so that initialMaxAnisotropy is set before we create
+	 * the default textures.
+	 */
+	driParseConfigFiles(&r600->radeon.optionCache, &screen->optionCache,
+			    screen->driScreen->myNum, "r600");
+
+	r600->radeon.initialMaxAnisotropy = driQueryOptionf(&r600->radeon.optionCache,
+							    "def_max_anisotropy");
+
+}
+
+static void r600InitGLExtensions(GLcontext *ctx)
+{
+	context_t *r600 = R700_CONTEXT(ctx);
 
 	driInitExtensions(ctx, card_extensions, GL_TRUE);
 	if (r600->radeon.radeonScreen->kernel_mm)
 	  driInitExtensions(ctx, mm_extensions, GL_FALSE);
 
+#ifdef R600_ENABLE_GLSL_TEST
+    driInitExtensions(ctx, gl_20_extension, GL_TRUE);
+    _mesa_enable_2_0_extensions(ctx);
+    
+    /* glsl compiler has problem if this is not GL_TRUE */
+    ctx->Shader.EmitCondCodes = GL_TRUE;
+#endif /* R600_ENABLE_GLSL_TEST */
+
 	if (driQueryOptionb
 	    (&r600->radeon.optionCache, "disable_stencil_two_side"))
 		_mesa_disable_extension(ctx, "GL_EXT_stencil_two_side");
@@ -377,21 +339,100 @@ GLboolean r600CreateContext(const __GLcontextModes * glVisual,
 		_mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 	}
 
-    r700InitDraw(ctx);
+	/* XXX: RV740 only seems to report results from half of its DBs */
+	if (r600->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV740)
+		_mesa_disable_extension(ctx, "GL_ARB_occlusion_query");
+}
 
-	radeon_fbo_init(&r600->radeon);
-   	radeonInitSpanFuncs( ctx );
+/* Create the device specific rendering context.
+ */
+GLboolean r600CreateContext(const __GLcontextModes * glVisual,
+			    __DRIcontext * driContextPriv,
+			    void *sharedContextPrivate)
+{
+	__DRIscreen *sPriv = driContextPriv->driScreenPriv;
+	radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private);
+	struct dd_function_table functions;
+	context_t *r600;
+	GLcontext *ctx;
 
-	r600InitCmdBuf(r600);
+	assert(glVisual);
+	assert(driContextPriv);
+	assert(screen);
 
-	r700InitState(r600->radeon.glCtx);
+	/* Allocate the R600 context */
+	r600 = (context_t*) CALLOC(sizeof(*r600));
+	if (!r600) {
+		radeon_error("Failed to allocate memory for context.\n");
+		return GL_FALSE;
+	}
+
+	r600ParseOptions(r600, screen);
+
+	r600->radeon.radeonScreen = screen;
+	r600_init_vtbl(&r600->radeon);
+
+	/* Init default driver functions then plug in our R600-specific functions
+	 * (the texture functions are especially important)
+	 */
+	_mesa_init_driver_functions(&functions);
 
-	TNL_CONTEXT(ctx)->Driver.RunPipeline = r600RunPipeline;
+	r700InitStateFuncs(&functions);
+	r600InitTextureFuncs(&functions);
+	r700InitShaderFuncs(&functions);
+	radeonInitQueryObjFunctions(&functions);
+	r700InitIoctlFuncs(&functions);
+	radeonInitBufferObjectFuncs(&functions);
 
-	if (driQueryOptionb(&r600->radeon.optionCache, "no_rast")) {
-		radeon_warning("disabling 3D acceleration\n");
+	if (!radeonInitContext(&r600->radeon, &functions,
+			       glVisual, driContextPriv,
+			       sharedContextPrivate)) {
+		radeon_error("Initializing context failed.\n");
+		FREE(r600);
+		return GL_FALSE;
 	}
 
+	ctx = r600->radeon.glCtx;
+
+	ctx->VertexProgram._MaintainTnlProgram = GL_TRUE;
+	ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
+
+	r600InitConstValues(ctx, screen);
+
+	_mesa_set_mvp_with_dp4( ctx, GL_TRUE );
+
+	/* Initialize the software rasterizer and helper modules.
+	 */
+	_swrast_CreateContext(ctx);
+	_vbo_CreateContext(ctx);
+	_tnl_CreateContext(ctx);
+	_swsetup_CreateContext(ctx);
+	_swsetup_Wakeup(ctx);
+
+	/* Install the customized pipeline:
+	 */
+	_tnl_destroy_pipeline(ctx);
+	_tnl_install_pipeline(ctx, r600_pipeline);
+	TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
+
+	/* Configure swrast and TNL to match hardware characteristics:
+	 */
+	_swrast_allow_pixel_fog(ctx, GL_FALSE);
+	_swrast_allow_vertex_fog(ctx, GL_TRUE);
+	_tnl_allow_pixel_fog(ctx, GL_FALSE);
+	_tnl_allow_vertex_fog(ctx, GL_TRUE);
+
+	radeon_init_debug();
+
+	r700InitDraw(ctx);
+
+	radeon_fbo_init(&r600->radeon);
+   	radeonInitSpanFuncs( ctx );
+	r600InitCmdBuf(r600);
+	r700InitState(r600->radeon.glCtx);
+
+	r600InitGLExtensions(ctx);
+
 	return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r600_context.h b/src/mesa/drivers/dri/r600/r600_context.h
index 7f68820fda..a1b4af715e 100644
--- a/src/mesa/drivers/dri/r600/r600_context.h
+++ b/src/mesa/drivers/dri/r600/r600_context.h
@@ -58,29 +58,6 @@ typedef struct r600_context context_t;
 
 #include "main/mm.h"
 
-/************ DMA BUFFERS **************/
-
-/* The blit width for texture uploads
- */
-#define R600_BLIT_WIDTH_BYTES 1024
-#define R600_MAX_TEXTURE_UNITS 8
-
-struct r600_texture_state {
-	int tc_count;		/* number of incoming texture coordinates from VAP */
-};
-
-/* Perhaps more if we store programs in vmem? */
-/* drm_r600_cmd_header_t->vpu->count is unsigned char */
-#define VSF_MAX_FRAGMENT_LENGTH (255*4)
-
-/* Can be tested with colormat currently. */
-#define VSF_MAX_FRAGMENT_TEMPS (14)
-
-#define STATE_R600_WINDOW_DIMENSION (STATE_INTERNAL_DRIVER+0)
-#define STATE_R600_TEXRECT_FACTOR (STATE_INTERNAL_DRIVER+1)
-
-extern int hw_tcl_on;
-
 #define COLOR_IS_RGBA
 #define TAG(x) r600##x
 #include "tnl_dd/t_dd_vertex.h"
@@ -131,6 +108,7 @@ typedef struct StreamDesc
 	GLint   size;   //number of data element
 	GLenum  type;  //data element type
 	GLsizei stride;
+	GLenum  format; // GL_RGBA,GLBGRA 
 
 	struct radeon_bo *bo;
 	GLint  bo_offset;
@@ -167,9 +145,6 @@ struct r600_context {
 
 	/* Vertex buffers
 	 */
-	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;
@@ -179,7 +154,7 @@ struct r600_context {
 #define GL_CONTEXT(context)     ((GLcontext *)(context->radeon.glCtx))
 
 extern GLboolean r600CreateContext(const __GLcontextModes * glVisual,
-				   __DRIcontextPrivate * driContextPriv,
+				   __DRIcontext * driContextPriv,
 				   void *sharedContextPrivate);
 
 #define R700_CONTEXT_STATES(context) ((R700_CHIP_CONTEXT *)(&context->hw))
@@ -203,7 +178,6 @@ extern GLboolean r700SyncSurf(context_t *context,
 			      uint32_t write_domain,
 			      uint32_t sync_type);
 
-extern void r700SetupStreams(GLcontext * ctx);
 extern void r700Start3D(context_t *context);
 extern void r600InitAtoms(context_t *context);
 extern void r700InitDraw(GLcontext *ctx);
@@ -213,7 +187,4 @@ extern void r700InitDraw(GLcontext *ctx);
 #define RADEON_D_PLAYBACK_RAW 2
 #define RADEON_D_T 3
 
-#define r600PackFloat32 radeonPackFloat32
-#define r600PackFloat24 radeonPackFloat24
-
 #endif				/* __R600_CONTEXT_H__ */
diff --git a/src/mesa/drivers/dri/r600/r600_reg_r6xx.h b/src/mesa/drivers/dri/r600/r600_reg_r6xx.h
index f7702c46de..74af7b4fed 100644
--- a/src/mesa/drivers/dri/r600/r600_reg_r6xx.h
+++ b/src/mesa/drivers/dri/r600/r600_reg_r6xx.h
@@ -415,11 +415,11 @@ enum {
 	ALPHA_TO_MASK_ENABLE                              = 1 << 0,
 	ALPHA_TO_MASK_OFFSET0_mask                        = 0x03 << 8,
 	ALPHA_TO_MASK_OFFSET0_shift                       = 8,
-	ALPHA_TO_MASK_OFFSET1_mask                        = 0x03 << 8,
+	ALPHA_TO_MASK_OFFSET1_mask                        = 0x03 << 10,
 	ALPHA_TO_MASK_OFFSET1_shift                       = 10,
-	ALPHA_TO_MASK_OFFSET2_mask                        = 0x03 << 8,
+	ALPHA_TO_MASK_OFFSET2_mask                        = 0x03 << 12,
 	ALPHA_TO_MASK_OFFSET2_shift                       = 12,
-	ALPHA_TO_MASK_OFFSET3_mask                        = 0x03 << 8,
+	ALPHA_TO_MASK_OFFSET3_mask                        = 0x03 << 14,
 	ALPHA_TO_MASK_OFFSET3_shift                       = 14,
 
 //  SQ_VTX_CONSTANT_WORD2_0                               = 0x00038008,
diff --git a/src/mesa/drivers/dri/r600/r600_reg_r7xx.h b/src/mesa/drivers/dri/r600/r600_reg_r7xx.h
index e5c01c861a..eb169bd885 100644
--- a/src/mesa/drivers/dri/r600/r600_reg_r7xx.h
+++ b/src/mesa/drivers/dri/r600/r600_reg_r7xx.h
@@ -143,6 +143,8 @@ enum {
 //  SQ_TEX_SAMPLER_MISC_0                                 = 0x0003d03c,
 	R7xx_TRUNCATE_COORD_bit                           = 1 << 9,
 	R7xx_DISABLE_CUBE_WRAP_bit                        = 1 << 10,
+//  DB_RENDER_CONTROL                                     = 0x00028d0c,
+	PERFECT_ZPASS_COUNTS_bit                          = 1 << 15,
 
 } ;
 
diff --git a/src/mesa/drivers/dri/r600/r600_tex.c b/src/mesa/drivers/dri/r600/r600_tex.c
index d105b90cd1..f745fe3e8a 100644
--- a/src/mesa/drivers/dri/r600/r600_tex.c
+++ b/src/mesa/drivers/dri/r600/r600_tex.c
@@ -40,7 +40,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/image.h"
 #include "main/mipmap.h"
 #include "main/simple_list.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
@@ -286,6 +285,7 @@ static void r600TexParameter(GLcontext * ctx, GLenum target,
 			     GLenum pname, const GLfloat * params)
 {
 	radeonTexObj* t = radeon_tex_obj(texObj);
+	GLenum baseFormat;
 
 	radeon_print(RADEON_STATE | RADEON_TEXTURE, RADEON_VERBOSE,
 			"%s( %s )\n", __FUNCTION__,
@@ -305,30 +305,22 @@ static void r600TexParameter(GLcontext * ctx, GLenum target,
 		break;
 
 	case GL_TEXTURE_BORDER_COLOR:
-		r600SetTexBorderColor(t, texObj->BorderColor);
+		r600SetTexBorderColor(t, texObj->BorderColor.f);
 		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.  Since there's no LOD clamping,
-		 * we just have to rely on loading the right subset of mipmap levels
-		 * to simulate a clamped LOD.
-		 */
-		if (t->mt) {
-			radeon_miptree_unreference(t->mt);
-			t->mt = 0;
-			t->validated = GL_FALSE;
-		}
+		t->validated = GL_FALSE;
 		break;
 
 	case GL_DEPTH_TEXTURE_MODE:
 		if (!texObj->Image[0][texObj->BaseLevel])
 			return;
-		if (texObj->Image[0][texObj->BaseLevel]->TexFormat->BaseFormat
-		    == GL_DEPTH_COMPONENT) {
+		baseFormat = texObj->Image[0][texObj->BaseLevel]->_BaseFormat;
+		if (baseFormat == GL_DEPTH_COMPONENT ||
+		    baseFormat == GL_DEPTH_STENCIL) {
 			r600SetDepthTexMode(texObj);
 			break;
 		} else {
@@ -368,10 +360,8 @@ static void r600DeleteTexture(GLcontext * ctx, struct gl_texture_object *texObj)
 		t->bo = NULL;
 	}
 
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
+	radeon_miptree_unreference(&t->mt);
+
 	_mesa_delete_texture_object(ctx, texObj);
 }
 
@@ -401,7 +391,7 @@ static struct gl_texture_object *r600NewTextureObject(GLcontext * ctx,
 	r600SetTexDefaultState(t);
 	r600UpdateTexWrap(t);
 	r600SetTexFilter(t, t->base.MinFilter, t->base.MagFilter, t->base.MaxAnisotropy);
-	r600SetTexBorderColor(t, t->base.BorderColor);
+	r600SetTexBorderColor(t, t->base.BorderColor.f);
 
 	return &t->base;
 }
diff --git a/src/mesa/drivers/dri/r600/r600_texstate.c b/src/mesa/drivers/dri/r600/r600_texstate.c
index 7d7e77d355..ae252c995b 100644
--- a/src/mesa/drivers/dri/r600/r600_texstate.c
+++ b/src/mesa/drivers/dri/r600/r600_texstate.c
@@ -39,7 +39,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 #include "main/enums.h"
@@ -78,7 +77,7 @@ void r600UpdateTextureState(GLcontext * ctx)
 	}
 }
 
-static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_format)
+static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, gl_format mesa_format)
 {
 	radeonTexObj *t = radeon_tex_obj(tObj);
 
@@ -87,9 +86,19 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 	CLEARfield(t->SQ_TEX_RESOURCE4, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_Z_mask);
 	CLEARfield(t->SQ_TEX_RESOURCE4, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
 
+	SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_UNSIGNED,
+		 FORMAT_COMP_X_shift, FORMAT_COMP_X_mask);
+	SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_UNSIGNED,
+		 FORMAT_COMP_Y_shift, FORMAT_COMP_Y_mask);
+	SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_UNSIGNED,
+		 FORMAT_COMP_Z_shift, FORMAT_COMP_Z_mask);
+	SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_UNSIGNED,
+		 FORMAT_COMP_W_shift, FORMAT_COMP_W_mask);
+
 	switch (mesa_format) /* This is mesa format. */
 	{
 	case MESA_FORMAT_RGBA8888:
+	case MESA_FORMAT_SIGNED_RGBA8888:
 		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);
 
@@ -101,8 +110,19 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 			 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);
+		if (mesa_format == MESA_FORMAT_SIGNED_RGBA8888) {
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_X_shift, FORMAT_COMP_X_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_Y_shift, FORMAT_COMP_Y_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_Z_shift, FORMAT_COMP_Z_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_W_shift, FORMAT_COMP_W_mask);
+		}
 		break;
 	case MESA_FORMAT_RGBA8888_REV:
+	case MESA_FORMAT_SIGNED_RGBA8888_REV:
 		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);
 
@@ -114,6 +134,16 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 			 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_W,
 			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
+		if (mesa_format == MESA_FORMAT_SIGNED_RGBA8888_REV) {
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_X_shift, FORMAT_COMP_X_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_Y_shift, FORMAT_COMP_Y_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_Z_shift, FORMAT_COMP_Z_mask);
+			SETfield(t->SQ_TEX_RESOURCE4, SQ_FORMAT_COMP_SIGNED,
+				 FORMAT_COMP_W_shift, FORMAT_COMP_W_mask);
+		}
 		break;
 	case MESA_FORMAT_ARGB8888:
 		SETfield(t->SQ_TEX_RESOURCE1, FMT_8_8_8_8,
@@ -480,13 +510,21 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 			 SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_shift, SQ_TEX_RESOURCE_WORD4_0__DST_SEL_W_mask);
 		break;
 	case MESA_FORMAT_Z16:
+	case MESA_FORMAT_X8_Z24:
+	case MESA_FORMAT_S8_Z24:
 	case MESA_FORMAT_Z24_S8:
 	case MESA_FORMAT_Z32:
+	case MESA_FORMAT_S8:
 		switch (mesa_format) {
 		case MESA_FORMAT_Z16:
 			SETfield(t->SQ_TEX_RESOURCE1, FMT_16,
 				 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
 			break;
+		case MESA_FORMAT_X8_Z24:
+		case MESA_FORMAT_S8_Z24:
+			SETfield(t->SQ_TEX_RESOURCE1, FMT_8_24,
+				 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
+			break;
 		case MESA_FORMAT_Z24_S8:
 			SETfield(t->SQ_TEX_RESOURCE1, FMT_24_8,
 				 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
@@ -495,6 +533,12 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 			SETfield(t->SQ_TEX_RESOURCE1, FMT_32,
 				 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
 			break;
+		case MESA_FORMAT_S8:
+			SETfield(t->SQ_TEX_RESOURCE1, FMT_8,
+				 SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_shift, SQ_TEX_RESOURCE_WORD1_0__DATA_FORMAT_mask);
+			break;
+		default:
+			break;
 		};
 		switch (tObj->DepthMode) {
 		case GL_LUMINANCE:  /* X, X, X, ONE */
@@ -582,6 +626,31 @@ static GLboolean r600GetTexFormat(struct gl_texture_object *tObj, GLuint mesa_fo
 	return GL_TRUE;
 }
 
+static GLuint r600_translate_shadow_func(GLenum func)
+{
+   switch (func) {
+   case GL_NEVER:
+      return SQ_TEX_DEPTH_COMPARE_NEVER;
+   case GL_LESS:
+      return SQ_TEX_DEPTH_COMPARE_LESS;
+   case GL_LEQUAL:
+      return SQ_TEX_DEPTH_COMPARE_LESSEQUAL;
+   case GL_GREATER:
+      return SQ_TEX_DEPTH_COMPARE_GREATER;
+   case GL_GEQUAL:
+      return SQ_TEX_DEPTH_COMPARE_GREATEREQUAL;
+   case GL_NOTEQUAL:
+      return SQ_TEX_DEPTH_COMPARE_NOTEQUAL;
+   case GL_EQUAL:
+      return SQ_TEX_DEPTH_COMPARE_EQUAL;
+   case GL_ALWAYS:
+      return SQ_TEX_DEPTH_COMPARE_ALWAYS;
+   default:
+      WARN_ONCE("Unknown shadow compare function! %d", func);
+      return 0;
+   }
+}
+
 void r600SetDepthTexMode(struct gl_texture_object *tObj)
 {
 	radeonTexObjPtr t;
@@ -591,7 +660,7 @@ void r600SetDepthTexMode(struct gl_texture_object *tObj)
 
 	t = radeon_tex_obj(tObj);
 
-	r600GetTexFormat(tObj, tObj->Image[0][tObj->BaseLevel]->TexFormat->MesaFormat);
+	r600GetTexFormat(tObj, tObj->Image[0][tObj->BaseLevel]->TexFormat);
 
 }
 
@@ -605,7 +674,6 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 {
 	radeonTexObj *t = radeon_tex_obj(texObj);
 	const struct gl_texture_image *firstImage;
-	int firstlevel = t->mt ? t->mt->firstLevel : 0;
 	GLuint uTexelPitch, row_align;
 
 	if (rmesa->radeon.radeonScreen->driScreen->dri2.enabled &&
@@ -613,10 +681,10 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	    t->bo)
 		return;
 
-	firstImage = t->base.Image[0][firstlevel];
+	firstImage = t->base.Image[0][t->minLod];
 
 	if (!t->image_override) {
-		if (!r600GetTexFormat(texObj, firstImage->TexFormat->MesaFormat)) {
+		if (!r600GetTexFormat(texObj, firstImage->TexFormat)) {
 			radeon_error("unexpected texture format in %s\n",
 				      __FUNCTION__);
 			return;
@@ -648,7 +716,8 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	}
 
 	row_align = rmesa->radeon.texture_row_align - 1;
-	uTexelPitch = ((firstImage->Width * t->mt->bpp + row_align) & ~row_align) / t->mt->bpp;
+	uTexelPitch = (_mesa_format_row_stride(firstImage->TexFormat, firstImage->Width) + row_align) & ~row_align;
+	uTexelPitch = uTexelPitch / _mesa_get_format_bytes(firstImage->TexFormat);
 	uTexelPitch = (uTexelPitch + R700_TEXEL_PITCH_ALIGNMENT_MASK)
 		& ~R700_TEXEL_PITCH_ALIGNMENT_MASK;
 
@@ -662,11 +731,22 @@ static void setup_hardware_state(context_t *rmesa, struct gl_texture_object *tex
 	SETfield(t->SQ_TEX_RESOURCE1, firstImage->Height - 1,
 		 TEX_HEIGHT_shift, TEX_HEIGHT_mask);
 
-	if ((t->mt->lastLevel - t->mt->firstLevel) > 0) {
-		t->SQ_TEX_RESOURCE3 = t->mt->levels[0].size / 256;
-		SETfield(t->SQ_TEX_RESOURCE4, t->mt->firstLevel, BASE_LEVEL_shift, BASE_LEVEL_mask);
-		SETfield(t->SQ_TEX_RESOURCE5, t->mt->lastLevel, LAST_LEVEL_shift, LAST_LEVEL_mask);
+	t->SQ_TEX_RESOURCE2 = get_base_teximage_offset(t) / 256;
+
+	if ((t->maxLod - t->minLod) > 0) {
+		t->SQ_TEX_RESOURCE3 = radeon_miptree_image_offset(t->mt, 0, t->minLod + 1) / 256;
+		SETfield(t->SQ_TEX_RESOURCE4, 0, BASE_LEVEL_shift, BASE_LEVEL_mask);
+		SETfield(t->SQ_TEX_RESOURCE5, t->maxLod - t->minLod, LAST_LEVEL_shift, LAST_LEVEL_mask);
 	}
+	if(texObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB)
+	{
+		SETfield(t->SQ_TEX_SAMPLER0, r600_translate_shadow_func(texObj->CompareFunc), DEPTH_COMPARE_FUNCTION_shift, DEPTH_COMPARE_FUNCTION_mask);
+	}
+	else
+	{
+		CLEARfield(t->SQ_TEX_SAMPLER0, DEPTH_COMPARE_FUNCTION_mask);
+	}
+
 }
 
 /**
@@ -764,7 +844,8 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	struct gl_texture_object *tObj =
 	    _mesa_lookup_texture(rmesa->radeon.glCtx, texname);
 	radeonTexObjPtr t = radeon_tex_obj(tObj);
-	uint32_t pitch_val, size;
+	const struct gl_texture_image *firstImage;
+	uint32_t pitch_val, size, row_align;
 
 	if (!tObj)
 		return;
@@ -774,7 +855,9 @@ void r600SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
 	if (!offset)
 		return;
 
-	size = pitch;//h * w * (depth / 8);
+	firstImage = t->base.Image[0][t->minLod];
+	row_align = rmesa->radeon.texture_row_align - 1;
+	size = ((_mesa_format_row_stride(firstImage->TexFormat, firstImage->Width) + row_align) & ~row_align) * firstImage->Height;
 	if (t->bo) {
 		radeon_bo_unref(t->bo);
 		t->bo = NULL;
@@ -870,18 +953,7 @@ void r600SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
     	    return;
     	}
 
-	radeon_update_renderbuffers(pDRICtx, dPriv);
-	/* back & depth buffer are useless free them right away */
-	rb = (void*)rfb->base.Attachment[BUFFER_DEPTH].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-        rb->bo = NULL;
-	}
-	rb = (void*)rfb->base.Attachment[BUFFER_BACK_LEFT].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-		rb->bo = NULL;
-	}
+	radeon_update_renderbuffers(pDRICtx, dPriv, GL_TRUE);
 	rb = rfb->color_rb[0];
 	if (rb->bo == NULL) {
 		/* Failed to BO for the buffer */
@@ -897,20 +969,14 @@ void r600SetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_fo
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
-	texImage->TexFormat = radeonChooseTextureFormat(radeon->glCtx,
-							internalFormat,
-							type, format, 0);
+
 	rImage->bo = rb->bo;
 	radeon_bo_ref(rImage->bo);
 	t->bo = rb->bo;
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c
index 903b6968be..0ff16b4ddd 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.c
+++ b/src/mesa/drivers/dri/r600/r700_assembler.c
@@ -32,12 +32,49 @@
 
 #include "main/mtypes.h"
 #include "main/imports.h"
+#include "shader/prog_parameter.h"
 
 #include "radeon_debug.h"
 #include "r600_context.h"
 
 #include "r700_assembler.h"
 
+#define USE_CF_FOR_CONTINUE_BREAK 1
+#define USE_CF_FOR_POP_AFTER      1
+
+struct prog_instruction noise1_insts[12] = { 
+    {OPCODE_BGNSUB , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_MOV , {{0, 0, 0, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 0, 2, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_MOV , {{8, 0, 0, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 0, 4, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_MOV , {{8, 0, 585, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 0, 8, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_SGT , {{0, 0, 585, 0, 0, 0}, {8, 0, 1170, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 1, 1, 0, 8, 1672, 0}, 1, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_IF , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 7, 0, 0}, 0, 0, 0, 1, 0, 0, 0, 15, 0, 0, 0}, 
+    {OPCODE_MOV , {{0, 0, 1755, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 0, 1, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_RET , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_ENDIF , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_MOV , {{0, 0, 1170, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {0, 0, 1, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_RET , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}, 
+    {OPCODE_ENDSUB , {{13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}, {13, 0, 1672, 0, 0, 0}}, {13, 0, 15, 0, 8, 1672, 0}, 0, 0, 0, 1, 0, 0, 0, -1, 0, 0, 0}
+};
+float noise1_const[2][4] = {
+    {0.300000f, 0.900000f, 0.500000f, 0.300000f}
+};
+
+COMPILED_SUB noise1_presub = {
+    &(noise1_insts[0]),
+    12, 
+    2, 
+    1, 
+    0, 
+    &(noise1_const[0]), 
+    SWIZZLE_X, 
+    SWIZZLE_X, 
+    SWIZZLE_X, 
+    SWIZZLE_X,
+    {0,0,0},
+    0 
+};
+
 BITS addrmode_PVSDST(PVSDST * pPVSDST)
 {
 	return pPVSDST->addrmode0 | ((BITS)pPVSDST->addrmode1 << 1);
@@ -327,21 +364,27 @@ GLuint GetSurfaceFormat(GLenum eType, GLuint nChannels, GLuint * pClient_size)
     return(format);
 }
 
-unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm)
+unsigned int r700GetNumOperands(GLuint opcode, GLuint nIsOp3) 
 {
-    if(pAsm->D.dst.op3)
+    if(nIsOp3 > 0)
     {
         return 3;
     }
 
-    switch (pAsm->D.dst.opcode)
+    switch (opcode)
     {
-    case SQ_OP2_INST_ADD:                          
+    case SQ_OP2_INST_ADD:
+    case SQ_OP2_INST_KILLE:
+    case SQ_OP2_INST_KILLGT:
+    case SQ_OP2_INST_KILLGE:
+    case SQ_OP2_INST_KILLNE:
     case SQ_OP2_INST_MUL: 
     case SQ_OP2_INST_MAX:
     case SQ_OP2_INST_MIN:
     //case SQ_OP2_INST_MAX_DX10:
     //case SQ_OP2_INST_MIN_DX10:
+    case SQ_OP2_INST_SETE: 
+    case SQ_OP2_INST_SETNE:
     case SQ_OP2_INST_SETGT:
     case SQ_OP2_INST_SETGE:
     case SQ_OP2_INST_PRED_SETE:
@@ -354,9 +397,10 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm)
         return 2;  
 
     case SQ_OP2_INST_MOV: 
+    case SQ_OP2_INST_MOVA_FLOOR:
     case SQ_OP2_INST_FRACT:
     case SQ_OP2_INST_FLOOR:
-    case SQ_OP2_INST_KILLGT:
+    case SQ_OP2_INST_TRUNC:
     case SQ_OP2_INST_EXP_IEEE:
     case SQ_OP2_INST_LOG_CLAMPED:
     case SQ_OP2_INST_LOG_IEEE:
@@ -368,7 +412,7 @@ unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm)
         return 1;
         
     default: radeon_error(
-		    "Need instruction operand number for %x.\n", pAsm->D.dst.opcode);
+		    "Need instruction operand number for %x.\n", opcode); 
     };
 
     return 3;
@@ -382,103 +426,128 @@ int Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700
     pAsm->pR700Shader = pShader;
     pAsm->currentShaderType = spt;
 
-	pAsm->cf_last_export_ptr   = NULL;
+    pAsm->cf_last_export_ptr   = NULL;
 
-	pAsm->cf_current_export_clause_ptr = NULL;
-	pAsm->cf_current_alu_clause_ptr    = NULL;
-	pAsm->cf_current_tex_clause_ptr    = NULL;
-	pAsm->cf_current_vtx_clause_ptr    = NULL;
-	pAsm->cf_current_cf_clause_ptr     = NULL;
+    pAsm->cf_current_export_clause_ptr = NULL;
+    pAsm->cf_current_alu_clause_ptr    = NULL;
+    pAsm->cf_current_tex_clause_ptr    = NULL;
+    pAsm->cf_current_vtx_clause_ptr    = NULL;
+    pAsm->cf_current_cf_clause_ptr     = NULL;
 
-	// No clause has been created yet
-	pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE;
+    // No clause has been created yet
+    pAsm->cf_current_clause_type = CF_EMPTY_CLAUSE;
 
-	pAsm->number_of_colorandz_exports = 0;
-	pAsm->number_of_exports           = 0;
-	pAsm->number_of_export_opcodes    = 0;
+    pAsm->number_of_colorandz_exports = 0;
+    pAsm->number_of_exports           = 0;
+    pAsm->number_of_export_opcodes    = 0;
 
+    pAsm->alu_x_opcode = 0;
 
-	pAsm->D.bits = 0;
-	pAsm->S[0].bits = 0;
-	pAsm->S[1].bits = 0;
-	pAsm->S[2].bits = 0;
+    pAsm->D2.bits = 0;
 
-	pAsm->uLastPosUpdate = 0; 
+    pAsm->D.bits = 0;
+    pAsm->S[0].bits = 0;
+    pAsm->S[1].bits = 0;
+    pAsm->S[2].bits = 0;
+
+    pAsm->uLastPosUpdate = 0; 
 	
-	*(BITS *) &pAsm->fp_stOutFmt0 = 0;
+    *(BITS *) &pAsm->fp_stOutFmt0 = 0;
 
-	pAsm->uIIns = 0;
-	pAsm->uOIns = 0;
-	pAsm->number_used_registers = 0;
-	pAsm->uUsedConsts = 256; 
+    pAsm->uIIns = 0;
+    pAsm->uOIns = 0;
+    pAsm->number_used_registers = 0;
+    pAsm->uUsedConsts = 256; 
 
 
-	// Fragment programs
-	pAsm->uBoolConsts = 0;
-	pAsm->uIntConsts = 0;
-	pAsm->uInsts = 0;
-	pAsm->uConsts = 0;
+    // Fragment programs
+    pAsm->uBoolConsts = 0;
+    pAsm->uIntConsts = 0;
+    pAsm->uInsts = 0;
+    pAsm->uConsts = 0;
 
-	pAsm->FCSP = 0;
-	pAsm->fc_stack[0].type = FC_NONE;
+    pAsm->FCSP = 0;
+    pAsm->fc_stack[0].type = FC_NONE;
 
-	pAsm->branch_depth     = 0;
-	pAsm->max_branch_depth = 0;
+    pAsm->aArgSubst[0] =
+    pAsm->aArgSubst[1] =
+    pAsm->aArgSubst[2] =
+    pAsm->aArgSubst[3] = (-1);
 
-	pAsm->aArgSubst[0] =
-	pAsm->aArgSubst[1] =
-	pAsm->aArgSubst[2] =
-	pAsm->aArgSubst[3] = (-1);
+    pAsm->uOutputs = 0;
 
-	pAsm->uOutputs = 0;
+    for (i=0; i<NUMBER_OF_OUTPUT_COLORS; i++) 
+    {
+        pAsm->color_export_register_number[i] = (-1);
+    }
 
-	for (i=0; i<NUMBER_OF_OUTPUT_COLORS; i++) 
-	{
-		pAsm->color_export_register_number[i] = (-1);
-	}
 
+    pAsm->depth_export_register_number = (-1);
+    pAsm->stencil_export_register_number = (-1);
+    pAsm->coverage_to_mask_export_register_number = (-1);
+    pAsm->mask_export_register_number = (-1);
 
-	pAsm->depth_export_register_number = (-1);
-	pAsm->stencil_export_register_number = (-1);
-	pAsm->coverage_to_mask_export_register_number = (-1);
-	pAsm->mask_export_register_number = (-1);
+    pAsm->starting_export_register_number = 0;
+    pAsm->starting_vfetch_register_number = 0;
+    pAsm->starting_temp_register_number   = 0;
+    pAsm->uFirstHelpReg = 0;
 
-	pAsm->starting_export_register_number = 0;
-	pAsm->starting_vfetch_register_number = 0;
-	pAsm->starting_temp_register_number   = 0;
-	pAsm->uFirstHelpReg = 0;
+    pAsm->input_position_is_used = GL_FALSE;
+    pAsm->input_normal_is_used   = GL_FALSE;
 
+    for (i=0; i<NUMBER_OF_INPUT_COLORS; i++) 
+    {
+        pAsm->input_color_is_used[ i ] = GL_FALSE;
+    }
 
-	pAsm->input_position_is_used = GL_FALSE;
-	pAsm->input_normal_is_used   = GL_FALSE;
+    for (i=0; i<NUMBER_OF_TEXTURE_UNITS; i++) 
+    {
+        pAsm->input_texture_unit_is_used[ i ] = GL_FALSE;
+    }
 
+    for (i=0; i<VERT_ATTRIB_MAX; i++) 
+    {
+        pAsm->vfetch_instruction_ptr_array[ i ] = NULL;
+    }
 
-	for (i=0; i<NUMBER_OF_INPUT_COLORS; i++) 
-	{
-		pAsm->input_color_is_used[ i ] = GL_FALSE;
-	}
+    pAsm->number_of_inputs = 0;
 
-	for (i=0; i<NUMBER_OF_TEXTURE_UNITS; i++) 
-	{
-		pAsm->input_texture_unit_is_used[ i ] = GL_FALSE;
-	}
+    pAsm->is_tex = GL_FALSE;
+    pAsm->need_tex_barrier = GL_FALSE;
 
-	for (i=0; i<VERT_ATTRIB_MAX; i++) 
-	{
-		pAsm->vfetch_instruction_ptr_array[ i ] = NULL;
-	}
+    pAsm->subs              = NULL;
+    pAsm->unSubArraySize    = 0;
+    pAsm->unSubArrayPointer = 0;
+    pAsm->callers              = NULL;
+    pAsm->unCallerArraySize    = 0;
+    pAsm->unCallerArrayPointer = 0;
+
+    pAsm->CALLSP = 0;
+    pAsm->CALLSTACK[0].FCSP_BeforeEntry = 0;
+    pAsm->CALLSTACK[0].plstCFInstructions_local
+          = &(pAsm->pR700Shader->lstCFInstructions);
+
+    pAsm->CALLSTACK[0].max = 0;
+    pAsm->CALLSTACK[0].current = 0;
+
+    SetActiveCFlist(pAsm->pR700Shader, pAsm->CALLSTACK[0].plstCFInstructions_local);
 
-	pAsm->number_of_inputs = 0;
+    pAsm->unCFflags = 0;
 
-	pAsm->is_tex = GL_FALSE;
-	pAsm->need_tex_barrier = GL_FALSE;
+    pAsm->presubs           = NULL;
+    pAsm->unPresubArraySize = 0;
+    pAsm->unNumPresub       = 0;
+    pAsm->unCurNumILInsts   = 0;
 
-	return 0;
+    pAsm->unVetTexBits      = 0;
+
+    return 0;
 }
 
 GLboolean IsTex(gl_inst_opcode Opcode)
 {
-    if( (OPCODE_TEX==Opcode) || (OPCODE_TXP==Opcode) || (OPCODE_TXB==Opcode) )
+    if( (OPCODE_TEX==Opcode) || (OPCODE_TXP==Opcode) || (OPCODE_TXB==Opcode) ||
+        (OPCODE_DDX==Opcode) || (OPCODE_DDY==Opcode) )
     {
         return GL_TRUE;
     }
@@ -591,6 +660,31 @@ int check_current_clause(r700_AssemblerBase* pAsm,
     return GL_TRUE;
 }
 
+GLboolean add_cf_instruction(r700_AssemblerBase* pAsm)
+{
+    if(GL_FALSE == check_current_clause(pAsm, CF_OTHER_CLAUSE))
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr = 
+      (R700ControlFlowGenericClause*) CALLOC_STRUCT(R700ControlFlowGenericClause);
+
+    if (pAsm->cf_current_cf_clause_ptr != NULL) 
+	{
+		Init_R700ControlFlowGenericClause(pAsm->cf_current_cf_clause_ptr);
+		AddCFInstruction( pAsm->pR700Shader, 
+                          (R700ControlFlowInstruction *)pAsm->cf_current_cf_clause_ptr );
+	}
+	else 
+	{
+        radeon_error("Could not allocate a new VFetch CF instruction.\n");
+		return GL_FALSE;
+	}
+
+    return GL_TRUE;
+}
+
 GLboolean add_vfetch_instruction(r700_AssemblerBase*     pAsm,
 								 R700VertexInstruction*  vertex_instruction_ptr)
 {
@@ -797,6 +891,7 @@ GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm,
                                        GLubyte             element,
                                        GLuint              _signed,
                                        GLboolean           normalize,
+                                       GLenum              format,
                                        VTX_FETCH_METHOD  * pFetchMethod)
 {
     GLuint client_size_inbyte;
@@ -845,10 +940,21 @@ GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm,
 	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;
+	if(format == GL_BGRA)
+	{
+		vfetch_instruction_ptr->m_Word1.f.dst_sel_x        = (size < 1) ? SQ_SEL_0 : SQ_SEL_Z;
+		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_X;
+		vfetch_instruction_ptr->m_Word1.f.dst_sel_w        = (size < 4) ? SQ_SEL_1 : SQ_SEL_W;
+	}
+	else
+	{
+		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;
@@ -986,7 +1092,8 @@ GLboolean checkop2(r700_AssemblerBase* pAsm)
 
     checkop_init(pAsm);
 
-    if( (pILInst->SrcReg[0].File == PROGRAM_CONSTANT)    ||
+    if( (pILInst->SrcReg[0].File == PROGRAM_UNIFORM)     || 
+        (pILInst->SrcReg[0].File == PROGRAM_CONSTANT)    ||
         (pILInst->SrcReg[0].File == PROGRAM_LOCAL_PARAM) ||
         (pILInst->SrcReg[0].File == PROGRAM_ENV_PARAM)   ||
         (pILInst->SrcReg[0].File == PROGRAM_STATE_VAR) )
@@ -997,7 +1104,8 @@ GLboolean checkop2(r700_AssemblerBase* pAsm)
     {
         bSrcConst[0] = GL_FALSE;
     }
-    if( (pILInst->SrcReg[1].File == PROGRAM_CONSTANT)    ||
+    if( (pILInst->SrcReg[1].File == PROGRAM_UNIFORM)     || 
+        (pILInst->SrcReg[1].File == PROGRAM_CONSTANT)    ||
         (pILInst->SrcReg[1].File == PROGRAM_LOCAL_PARAM) ||
         (pILInst->SrcReg[1].File == PROGRAM_ENV_PARAM)   ||
         (pILInst->SrcReg[1].File == PROGRAM_STATE_VAR) )
@@ -1030,7 +1138,8 @@ GLboolean checkop3(r700_AssemblerBase* pAsm)
 
     checkop_init(pAsm);
 
-    if( (pILInst->SrcReg[0].File == PROGRAM_CONSTANT)    ||
+    if( (pILInst->SrcReg[0].File == PROGRAM_UNIFORM)     || 
+        (pILInst->SrcReg[0].File == PROGRAM_CONSTANT)    ||
         (pILInst->SrcReg[0].File == PROGRAM_LOCAL_PARAM) ||
         (pILInst->SrcReg[0].File == PROGRAM_ENV_PARAM)   ||
         (pILInst->SrcReg[0].File == PROGRAM_STATE_VAR) )
@@ -1041,7 +1150,8 @@ GLboolean checkop3(r700_AssemblerBase* pAsm)
     {
         bSrcConst[0] = GL_FALSE;
     }
-    if( (pILInst->SrcReg[1].File == PROGRAM_CONSTANT)    ||
+    if( (pILInst->SrcReg[1].File == PROGRAM_UNIFORM)     || 
+        (pILInst->SrcReg[1].File == PROGRAM_CONSTANT)    ||
         (pILInst->SrcReg[1].File == PROGRAM_LOCAL_PARAM) ||
         (pILInst->SrcReg[1].File == PROGRAM_ENV_PARAM)   ||
         (pILInst->SrcReg[1].File == PROGRAM_STATE_VAR) )
@@ -1052,7 +1162,8 @@ GLboolean checkop3(r700_AssemblerBase* pAsm)
     {
         bSrcConst[1] = GL_FALSE;
     }
-    if( (pILInst->SrcReg[2].File == PROGRAM_CONSTANT)    ||
+    if( (pILInst->SrcReg[2].File == PROGRAM_UNIFORM)     || 
+        (pILInst->SrcReg[2].File == PROGRAM_CONSTANT)    ||
         (pILInst->SrcReg[2].File == PROGRAM_LOCAL_PARAM) ||
         (pILInst->SrcReg[2].File == PROGRAM_ENV_PARAM)   ||
         (pILInst->SrcReg[2].File == PROGRAM_STATE_VAR) )
@@ -1152,6 +1263,7 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm,
         case PROGRAM_LOCAL_PARAM:
         case PROGRAM_ENV_PARAM:
         case PROGRAM_STATE_VAR:
+        case PROGRAM_UNIFORM:
             if (1 == pILInst->SrcReg[src].RelAddr)
             {
                 setaddrmode_PVSSRC(&(pAsm->S[fld].src), ADDR_RELATIVE_A0);
@@ -1162,10 +1274,18 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm,
             }
 
             pAsm->S[fld].src.rtype = SRC_REG_CONSTANT;
-            pAsm->S[fld].src.reg   = pILInst->SrcReg[src].Index;
+            if(pILInst->SrcReg[src].Index < 0)
+            {
+                WARN_ONCE("Negative register offsets not supported yet!\n");
+                pAsm->S[fld].src.reg  = 0;
+            } 
+            else
+            {
+                pAsm->S[fld].src.reg = pILInst->SrcReg[src].Index;
+            }
             break;      
         case PROGRAM_INPUT:
-            setaddrmode_PVSSRC(&(pAsm->S[fld].src), ADDR_ABSOLUTE);
+            setaddrmode_PVSSRC(&(pAsm->S[fld].src), ADDR_ABSOLUTE); 
             pAsm->S[fld].src.rtype = SRC_REG_INPUT;
             switch (pAsm->currentShaderType)
             {
@@ -1178,7 +1298,7 @@ GLboolean assemble_src(r700_AssemblerBase *pAsm,
             }
             break;      
         default:
-            radeon_error("Invalid source argument type\n");
+            radeon_error("Invalid source argument type : %d \n", pILInst->SrcReg[src].File);
             return GL_FALSE;
         }
     } 
@@ -1234,6 +1354,15 @@ GLboolean assemble_dst(r700_AssemblerBase *pAsm)
     pAsm->D.dst.writez = (pILInst->DstReg.WriteMask >> 2) & 0x1;
     pAsm->D.dst.writew = (pILInst->DstReg.WriteMask >> 3) & 0x1;
   
+    if(pILInst->SaturateMode == SATURATE_ZERO_ONE)
+    {
+        pAsm->D2.dst2.SaturateMode = 1;
+    }
+    else
+    {
+        pAsm->D2.dst2.SaturateMode = 0;
+    }
+
     return GL_TRUE;
 }
 
@@ -1293,6 +1422,7 @@ GLboolean tex_src(r700_AssemblerBase *pAsm)
     else
     {
     switch (pILInst->SrcReg[0].File) {
+        case PROGRAM_UNIFORM: 
         case PROGRAM_CONSTANT:
         case PROGRAM_LOCAL_PARAM:
         case PROGRAM_ENV_PARAM:
@@ -1305,25 +1435,65 @@ GLboolean tex_src(r700_AssemblerBase *pAsm)
             pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
             break;
         case PROGRAM_INPUT:
-            switch (pILInst->SrcReg[0].Index)
+            if(SPT_VP == pAsm->currentShaderType)
             {
-                case FRAG_ATTRIB_COL0:
-                case FRAG_ATTRIB_COL1:
-                case FRAG_ATTRIB_TEX0:
-                case FRAG_ATTRIB_TEX1:
-                case FRAG_ATTRIB_TEX2:
-	        case FRAG_ATTRIB_TEX3:
-                case FRAG_ATTRIB_TEX4:
-                case FRAG_ATTRIB_TEX5:
-                case FRAG_ATTRIB_TEX6:
-                case FRAG_ATTRIB_TEX7:
-                    bValidTexCoord = GL_TRUE;
+                switch (pILInst->SrcReg[0].Index)
+                {
+                    case VERT_ATTRIB_TEX0:
+                    case VERT_ATTRIB_TEX1:
+                    case VERT_ATTRIB_TEX2:
+                    case VERT_ATTRIB_TEX3:
+                    case VERT_ATTRIB_TEX4:
+                    case VERT_ATTRIB_TEX5:
+                    case VERT_ATTRIB_TEX6:
+                    case VERT_ATTRIB_TEX7:
+                        bValidTexCoord = GL_TRUE;
+                        pAsm->S[0].src.reg   =
+                            pAsm->ucVP_AttributeMap[pILInst->SrcReg[0].Index];
+                        pAsm->S[0].src.rtype = SRC_REG_INPUT;
+                        break;
+                }
+            }
+            else
+            {
+                switch (pILInst->SrcReg[0].Index)
+                {
+                    case FRAG_ATTRIB_WPOS:
+                    case FRAG_ATTRIB_COL0:
+                    case FRAG_ATTRIB_COL1:
+                    case FRAG_ATTRIB_FOGC:
+                    case FRAG_ATTRIB_TEX0:
+                    case FRAG_ATTRIB_TEX1:
+                    case FRAG_ATTRIB_TEX2:
+                    case FRAG_ATTRIB_TEX3:
+                    case FRAG_ATTRIB_TEX4:
+                    case FRAG_ATTRIB_TEX5:
+                    case FRAG_ATTRIB_TEX6:
+                    case FRAG_ATTRIB_TEX7:
+                        bValidTexCoord = GL_TRUE;
+                        pAsm->S[0].src.reg   =
+                            pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index];
+                        pAsm->S[0].src.rtype = SRC_REG_INPUT;
+                        break;
+                    case FRAG_ATTRIB_FACE:
+                        fprintf(stderr, "FRAG_ATTRIB_FACE unsupported\n");
+                        break;
+                    case FRAG_ATTRIB_PNTC:
+                        fprintf(stderr, "FRAG_ATTRIB_PNTC unsupported\n");
+                        break;
+                }
+
+                if( (pILInst->SrcReg[0].Index >= FRAG_ATTRIB_VAR0) ||
+                    (pILInst->SrcReg[0].Index < FRAG_ATTRIB_MAX) )
+                {
+				    bValidTexCoord = GL_TRUE;
                     pAsm->S[0].src.reg   =
                         pAsm->uiFP_AttributeMap[pILInst->SrcReg[0].Index];
                     pAsm->S[0].src.rtype = SRC_REG_INPUT;
-                break;
+                }
             }
-        break;
+
+            break;
         }
     }
 
@@ -1368,8 +1538,17 @@ GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm, GLboolean normalize
     tex_instruction_ptr->m_Word0.f.tex_inst         = pAsm->D.dst.opcode;
     tex_instruction_ptr->m_Word0.f.bc_frac_mode     = 0x0;
     tex_instruction_ptr->m_Word0.f.fetch_whole_quad = 0x0;
+    tex_instruction_ptr->m_Word0.f.alt_const        = 0;
 
-    tex_instruction_ptr->m_Word0.f.resource_id      = texture_unit_source->reg;
+    if(SPT_VP == pAsm->currentShaderType)
+    {
+        tex_instruction_ptr->m_Word0.f.resource_id      = texture_unit_source->reg + VERT_ATTRIB_MAX;
+        pAsm->unVetTexBits |= 1 << texture_unit_source->reg;
+    }
+    else
+    {
+        tex_instruction_ptr->m_Word0.f.resource_id      = texture_unit_source->reg;
+    }
 
     tex_instruction_ptr->m_Word1.f.lod_bias     = 0x0;
     if (normalized) {
@@ -1388,7 +1567,6 @@ GLboolean assemble_tex_instruction(r700_AssemblerBase *pAsm, GLboolean normalize
     tex_instruction_ptr->m_Word2.f.offset_x   = 0x0;
     tex_instruction_ptr->m_Word2.f.offset_y   = 0x0;
     tex_instruction_ptr->m_Word2.f.offset_z   = 0x0;
-
     tex_instruction_ptr->m_Word2.f.sampler_id = texture_unit_source->reg;
 
     // dst
@@ -1505,6 +1683,10 @@ GLboolean assemble_alu_src(R700ALUInstruction*  alu_instruction_ptr,
         {
             src_sel = pSource->reg + CFILE_REGISTER_OFFSET;            
         }
+        else if (pSource->rtype == SRC_REC_LITERAL)
+        {
+            src_sel = SQ_ALU_SRC_LITERAL;            
+        }
         else
         {
             radeon_error("Source (%d) register type (%d) not one of TEMP, INPUT, or CONSTANT.\n",
@@ -1594,7 +1776,8 @@ GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
         return GL_FALSE;
     }
 
-    if ( pAsm->cf_current_alu_clause_ptr == NULL ||
+    if ( pAsm->alu_x_opcode != 0 ||
+         pAsm->cf_current_alu_clause_ptr == NULL ||
          ( (pAsm->cf_current_alu_clause_ptr != NULL) && 
            (pAsm->cf_current_alu_clause_ptr->m_Word1.f.count >= (GetCFMaxInstructions(pAsm->cf_current_alu_clause_ptr->m_ShaderInstType)-contiguous_slots_needed-1) )
          ) ) 
@@ -1624,9 +1807,17 @@ GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.kcache_addr0 = 0x0;
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.kcache_addr1 = 0x0;
 
-        //cf_current_alu_clause_ptr->m_Word1.f.count           = number_of_scalar_operations - 1;
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.count           = 0x0;
-        pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst         = SQ_CF_INST_ALU;
+
+        if(pAsm->alu_x_opcode != 0)
+        {
+            pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst = pAsm->alu_x_opcode;
+            pAsm->alu_x_opcode = 0;
+        }
+        else
+        {
+            pAsm->cf_current_alu_clause_ptr->m_Word1.f.cf_inst = SQ_CF_INST_ALU;
+        }
 
         pAsm->cf_current_alu_clause_ptr->m_Word1.f.whole_quad_mode = 0x0;
 
@@ -1634,7 +1825,7 @@ GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
     }
     else 
     {
-        pAsm->cf_current_alu_clause_ptr->m_Word1.f.count++;
+        pAsm->cf_current_alu_clause_ptr->m_Word1.f.count += (GetInstructionSize(alu_instruction_ptr->m_ShaderInstType) / 2);
     }
 
     // If this clause constains any instruction that is forward dependent on a TEX instruction, 
@@ -1911,7 +2102,7 @@ GLboolean check_scalar(r700_AssemblerBase* pAsm,
 
     GLuint swizzle_key;
 
-    GLuint number_of_operands = r700GetNumOperands(pAsm);
+    GLuint number_of_operands = r700GetNumOperands(pAsm->D.dst.opcode, pAsm->D.dst.op3);
 
     for (src=0; src<number_of_operands; src++) 
     {
@@ -2000,7 +2191,7 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
 
     GLuint swizzle_key;
 
-    GLuint number_of_operands = r700GetNumOperands(pAsm);
+    GLuint number_of_operands = r700GetNumOperands(pAsm->D.dst.opcode, pAsm->D.dst.op3);
 
     for (src=0; src<number_of_operands; src++) 
     {
@@ -2033,7 +2224,7 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
         if( is_gpr(sel) ) 
         {
             if( GL_FALSE == cycle_for_vector_bank_swizzle(bank_swizzle, src, &cycle) )
-            {
+            {             
                 return GL_FALSE;
             }
 
@@ -2045,7 +2236,7 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
             else 
             {
                 if( GL_FALSE == reserve_gpr(pAsm, sel, chan, cycle) )
-                {
+                {                    
                     return GL_FALSE;
                 }
             }
@@ -2057,7 +2248,7 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
             if( is_cfile(sel) ) 
             {        
                 if( GL_FALSE == reserve_cfile(pAsm, sel, chan) )
-                {
+                {                    
                     return GL_FALSE;
                 }
             }
@@ -2069,6 +2260,10 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
 
 GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
 {
+    R700ALUInstruction            * alu_instruction_ptr;
+    R700ALUInstructionHalfLiteral * alu_instruction_ptr_hl;
+    R700ALUInstructionFullLiteral * alu_instruction_ptr_fl;
+
     GLuint    number_of_scalar_operations;
     GLboolean is_single_scalar_operation;
     GLuint    scalar_channel_index;
@@ -2077,7 +2272,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
     int    current_source_index;
     GLuint contiguous_slots_needed;
 
-    GLuint    uNumSrc = r700GetNumOperands(pAsm);
+    GLuint    uNumSrc = r700GetNumOperands(pAsm->D.dst.opcode, pAsm->D.dst.op3);
     //GLuint    channel_swizzle, j;
     //GLuint    chan_counter[4] = {0, 0, 0, 0};
     //PVSSRC *  pSource[3];
@@ -2134,23 +2329,44 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
 
     contiguous_slots_needed = 0;
 
-    if(GL_TRUE == is_reduction_opcode(&(pAsm->D)) ) 
+    if(!is_single_scalar_operation) 
     {
         contiguous_slots_needed = 4;
     }
 
+    contiguous_slots_needed += pAsm->D2.dst2.literal_slots;
+
     initialize(pAsm);    
 
     for (scalar_channel_index=0;
             scalar_channel_index < number_of_scalar_operations; 
                 scalar_channel_index++) 
     {
-        R700ALUInstruction* alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
-        if (alu_instruction_ptr == NULL) 
-		{
-			return GL_FALSE;
-		}
-        Init_R700ALUInstruction(alu_instruction_ptr);
+        if(scalar_channel_index == (number_of_scalar_operations-1))
+        {
+            switch(pAsm->D2.dst2.literal_slots)
+            {
+            case 0:
+                alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
+                Init_R700ALUInstruction(alu_instruction_ptr);
+                break;
+            case 1:
+                alu_instruction_ptr_hl = (R700ALUInstructionHalfLiteral*) CALLOC_STRUCT(R700ALUInstructionHalfLiteral);
+                Init_R700ALUInstructionHalfLiteral(alu_instruction_ptr_hl, pAsm->C[0].f, pAsm->C[1].f);
+                alu_instruction_ptr = (R700ALUInstruction*)alu_instruction_ptr_hl;
+                break;
+            case 2:
+                alu_instruction_ptr_fl = (R700ALUInstructionFullLiteral*) CALLOC_STRUCT(R700ALUInstructionFullLiteral);
+                Init_R700ALUInstructionFullLiteral(alu_instruction_ptr_fl,pAsm->C[0].f, pAsm->C[1].f, pAsm->C[2].f, pAsm->C[3].f);
+                alu_instruction_ptr = (R700ALUInstruction*)alu_instruction_ptr_fl;
+            break;
+            };
+        }
+        else
+        {
+            alu_instruction_ptr = (R700ALUInstruction*) CALLOC_STRUCT(R700ALUInstruction);
+            Init_R700ALUInstruction(alu_instruction_ptr);
+        }
         
         //src 0
         current_source_index = 0;
@@ -2160,7 +2376,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
                                          current_source_index,
                                          pcurrent_source, 
                                          scalar_channel_index) )     
-        {
+        {            
             return GL_FALSE;
         }
    
@@ -2174,13 +2390,13 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
                                              current_source_index,
                                              pcurrent_source, 
                                              scalar_channel_index) ) 
-            {
+            {                
                 return GL_FALSE;
             }
         }
 
         //other bits
-        alu_instruction_ptr->m_Word0.f.index_mode = SQ_INDEX_LOOP;
+        alu_instruction_ptr->m_Word0.f.index_mode = pAsm->D2.dst2.index_mode;
 
         if(   (is_single_scalar_operation == GL_TRUE) 
            || (GL_TRUE == bSplitInst) )
@@ -2192,9 +2408,17 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
             alu_instruction_ptr->m_Word0.f.last = (scalar_channel_index == 3) ?  1 : 0;
         }
 
-        alu_instruction_ptr->m_Word0.f.pred_sel                = 0x0;
-        alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x0;  
-        alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x0;
+        alu_instruction_ptr->m_Word0.f.pred_sel = (pAsm->D.dst.pred_inv > 0) ? 1 : 0;
+        if(1 == pAsm->D.dst.predicated)
+        {
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x1;
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x1;
+        }
+        else
+        {
+            alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x0;
+            alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x0;
+        }
 
         // dst
         if( (pAsm->D.dst.rtype == DST_REG_TEMPORARY) || 
@@ -2203,7 +2427,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
             alu_instruction_ptr->m_Word1.f.dst_gpr  = pAsm->D.dst.reg;
         }
         else 
-        {
+        {            
             radeon_error("Only temp destination registers supported for ALU dest regs.\n");
             return GL_FALSE;
         }
@@ -2233,7 +2457,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
 
         alu_instruction_ptr->m_Word1.f.dst_chan = scalar_channel_index;
 
-        alu_instruction_ptr->m_Word1.f.clamp    = pAsm->pILInst[pAsm->uiCurInst].SaturateMode;
+        alu_instruction_ptr->m_Word1.f.clamp    = pAsm->D2.dst2.SaturateMode;
 
         if (pAsm->D.dst.op3) 
         {            
@@ -2260,8 +2484,8 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
             {
                 alu_instruction_ptr->m_Word1_OP2.f6.alu_inst           = pAsm->D.dst.opcode;
 
-                alu_instruction_ptr->m_Word1_OP2.f6.src0_abs           = 0x0;
-                alu_instruction_ptr->m_Word1_OP2.f6.src1_abs           = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f6.src0_abs           = pAsm->S[0].src.abs;
+                alu_instruction_ptr->m_Word1_OP2.f6.src1_abs           = pAsm->S[1].src.abs;
 
                 //alu_instruction_ptr->m_Word1_OP2.f6.update_execute_mask = 0x0;
                 //alu_instruction_ptr->m_Word1_OP2.f6.update_pred         = 0x0;
@@ -2289,8 +2513,8 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
             {
                 alu_instruction_ptr->m_Word1_OP2.f.alu_inst           = pAsm->D.dst.opcode;
 
-                alu_instruction_ptr->m_Word1_OP2.f.src0_abs           = 0x0;
-                alu_instruction_ptr->m_Word1_OP2.f.src1_abs           = 0x0;
+                alu_instruction_ptr->m_Word1_OP2.f.src0_abs           = pAsm->S[0].src.abs;
+                alu_instruction_ptr->m_Word1_OP2.f.src1_abs           = pAsm->S[1].src.abs;
 
                 //alu_instruction_ptr->m_Word1_OP2.f.update_execute_mask = 0x0;
                 //alu_instruction_ptr->m_Word1_OP2.f.update_pred         = 0x0;
@@ -2317,7 +2541,7 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
         }
 
         if(GL_FALSE == add_alu_instruction(pAsm, alu_instruction_ptr, contiguous_slots_needed) )
-        {
+        {            
             return GL_FALSE;
         }
 
@@ -2328,19 +2552,19 @@ GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm)
         if (is_single_scalar_operation) 
         {
             if(GL_FALSE == check_scalar(pAsm, alu_instruction_ptr) )
-            {
+            {                
                 return GL_FALSE;
             }
         }
         else 
         {
             if(GL_FALSE == check_vector(pAsm, alu_instruction_ptr) )
-            {
-                return 1;
+            {                
+                return GL_FALSE; 
             }
         }
 
-        contiguous_slots_needed = 0;
+        contiguous_slots_needed -= 1;
     }
 
     return GL_TRUE;
@@ -2391,11 +2615,14 @@ GLboolean next_ins(r700_AssemblerBase *pAsm)
     
     //reset for next inst.
     pAsm->D.bits    = 0;
+    pAsm->D2.bits   = 0;
     pAsm->S[0].bits = 0;
     pAsm->S[1].bits = 0;
     pAsm->S[2].bits = 0;
     pAsm->is_tex = GL_FALSE;
     pAsm->need_tex_barrier = GL_FALSE;
+    pAsm->D2.bits = 0;
+    pAsm->C[0].bits = pAsm->C[1].bits = pAsm->C[2].bits = pAsm->C[3].bits = 0;
     return GL_TRUE;
 }
 
@@ -2514,6 +2741,35 @@ GLboolean assemble_ADD(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
+GLboolean assemble_ARL(r700_AssemblerBase *pAsm)
+{ /* TODO: ar values dont' persist between clauses */
+    if( GL_FALSE == checkop1(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MOVA_FLOOR;
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg = 0;
+    pAsm->D.dst.writex = 0;
+    pAsm->D.dst.writey = 0;
+    pAsm->D.dst.writez = 0;
+    pAsm->D.dst.writew = 0;
+
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+        return GL_FALSE;
+    }
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    return GL_TRUE;
+}
+
 GLboolean assemble_BAD(char *opcode_str) 
 {
     radeon_error("Not yet implemented instruction (%s)\n", opcode_str);
@@ -2599,9 +2855,44 @@ GLboolean assemble_CMP(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
-GLboolean assemble_COS(r700_AssemblerBase *pAsm)
+GLboolean assemble_TRIG(r700_AssemblerBase *pAsm, BITS opcode)
 {
-    return assemble_math_function(pAsm, SQ_OP2_INST_COS);
+    int tmp;
+    checkop1(pAsm);
+
+    tmp = gethelpr(pAsm);
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MUL;
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+    pAsm->D.dst.writex = 1;
+
+    assemble_src(pAsm, 0, -1);
+
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+    pAsm->D2.dst2.literal_slots = 1;
+    pAsm->C[0].f = 1/(3.1415926535 * 2);
+    pAsm->C[1].f = 0.0F;
+    next_ins(pAsm);
+
+    pAsm->D.dst.opcode = opcode;
+    pAsm->D.dst.math = 1;
+
+    assemble_dst(pAsm);
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp;
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    next_ins(pAsm);
+
+    //TODO - replicate if more channels set in WriteMask
+    return GL_TRUE;
+
 }
  
 GLboolean assemble_DOT(r700_AssemblerBase *pAsm)
@@ -2635,7 +2926,7 @@ GLboolean assemble_DOT(r700_AssemblerBase *pAsm)
     }
     else if(pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_DPH) 
     {
-        onecomp_PVSSRC(&(pAsm->S[1].src), 3);
+        onecomp_PVSSRC(&(pAsm->S[0].src), 3);
     } 
 
     if ( GL_FALSE == next_ins(pAsm) ) 
@@ -2688,6 +2979,133 @@ GLboolean assemble_EX2(r700_AssemblerBase *pAsm)
 {
     return assemble_math_function(pAsm, SQ_OP2_INST_EXP_IEEE);
 }
+
+GLboolean assemble_EXP(r700_AssemblerBase *pAsm)
+{
+    BITS tmp;
+
+    checkop1(pAsm);
+
+    tmp = gethelpr(pAsm);
+
+    // FLOOR   tmp.x,    a.x
+    // EX2     dst.x     tmp.x
+
+    if (pAsm->pILInst->DstReg.WriteMask & 0x1) {
+        pAsm->D.dst.opcode = SQ_OP2_INST_FLOOR;
+
+        setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+        pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+        pAsm->D.dst.reg    = tmp;
+        pAsm->D.dst.writex = 1;
+
+        if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+        {
+            return GL_FALSE;
+        }
+
+        if( GL_FALSE == next_ins(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        pAsm->D.dst.opcode = SQ_OP2_INST_EXP_IEEE;
+        pAsm->D.dst.math = 1;
+
+        if( GL_FALSE == assemble_dst(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        pAsm->D.dst.writey = pAsm->D.dst.writez = pAsm->D.dst.writew = 0;
+
+        setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+        pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+        pAsm->S[0].src.reg   = tmp;
+
+        setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+        noneg_PVSSRC(&(pAsm->S[0].src));
+
+        if( GL_FALSE == next_ins(pAsm) )
+        {
+            return GL_FALSE;
+        }
+    }
+
+    // FRACT   dst.y     a.x
+
+    if ((pAsm->pILInst->DstReg.WriteMask >> 1) & 0x1) {
+        pAsm->D.dst.opcode = SQ_OP2_INST_FRACT;
+
+        if( GL_FALSE == assemble_dst(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+        {
+            return GL_FALSE;
+        }
+
+        pAsm->D.dst.writex = pAsm->D.dst.writez = pAsm->D.dst.writew = 0;
+
+        if( GL_FALSE == next_ins(pAsm) )
+        {
+            return GL_FALSE;
+        }
+    }
+
+    // EX2     dst.z,    a.x
+
+    if ((pAsm->pILInst->DstReg.WriteMask >> 2) & 0x1) {
+        pAsm->D.dst.opcode = SQ_OP2_INST_EXP_IEEE;
+        pAsm->D.dst.math = 1;
+
+        if( GL_FALSE == assemble_dst(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+        {
+            return GL_FALSE;
+        }
+
+        pAsm->D.dst.writex = pAsm->D.dst.writey = pAsm->D.dst.writew = 0;
+
+        if( GL_FALSE == next_ins(pAsm) )
+        {
+            return GL_FALSE;
+        }
+    }
+
+    // MOV     dst.w     1.0
+
+    if ((pAsm->pILInst->DstReg.WriteMask >> 3) & 0x1) {
+        pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+
+        if( GL_FALSE == assemble_dst(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        pAsm->D.dst.writex = pAsm->D.dst.writey = pAsm->D.dst.writez = 0;
+
+        setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+        pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+        pAsm->S[0].src.reg   = tmp;
+
+        setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_1);
+        noneg_PVSSRC(&(pAsm->S[0].src));
+
+        if( GL_FALSE == next_ins(pAsm) )
+        {
+            return GL_FALSE;
+        }
+    }
+
+    return GL_TRUE;
+}
  
 GLboolean assemble_FLR(r700_AssemblerBase *pAsm)
 {
@@ -2742,17 +3160,19 @@ GLboolean assemble_FRC(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
  
-GLboolean assemble_KIL(r700_AssemblerBase *pAsm)
-{
-    checkop1(pAsm);
+GLboolean assemble_KIL(r700_AssemblerBase *pAsm, GLuint opcode)
+{  
+    struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]);
 
-    pAsm->D.dst.opcode = SQ_OP2_INST_KILLGT;  
-  
-    if ( GL_FALSE == assemble_dst(pAsm) )
-    {
-        return GL_FALSE;
-    }
+    if(pILInst->Opcode == OPCODE_KIL)
+        checkop1(pAsm);
+
+    pAsm->D.dst.opcode = opcode;  
+    //pAsm->D.dst.math = 1;
 
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg   = 0;
     pAsm->D.dst.writex = 0;
     pAsm->D.dst.writey = 0;
     pAsm->D.dst.writez = 0;
@@ -2761,30 +3181,34 @@ GLboolean assemble_KIL(r700_AssemblerBase *pAsm)
     setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
     pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
     pAsm->S[0].src.reg = 0;
-
     setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_0);
     noneg_PVSSRC(&(pAsm->S[0].src));
 
-    pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
-
-    if(PROGRAM_TEMPORARY == pAsm->pILInst[pAsm->uiCurInst].DstReg.File)
+    if(pILInst->Opcode == OPCODE_KIL_NV)
     {
-        pAsm->S[1].src.reg = pAsm->pILInst[pAsm->uiCurInst].DstReg.Index + pAsm->starting_temp_register_number;
+        setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+        pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
+        pAsm->S[1].src.reg = 0;
+        setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_1);
+        neg_PVSSRC(&(pAsm->S[1].src));
     }
     else
-    {   //PROGRAM_OUTPUT
-        pAsm->S[1].src.reg = pAsm->uiFP_OutputMap[pAsm->pILInst[pAsm->uiCurInst].DstReg.Index];
+    {
+        if( GL_FALSE == assemble_src(pAsm, 0, 1) )
+        {
+            return GL_FALSE;
+        }
+
     }
-  
-    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
-    noswizzle_PVSSRC(&(pAsm->S[1].src));
-  
+
     if ( GL_FALSE == next_ins(pAsm) )
     {
         return GL_FALSE;
     }
 
+    /* Doc says KILL has to be last(end) ALU clause */
     pAsm->pR700Shader->killIsUsed = GL_TRUE;
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
     
     return GL_TRUE;
 }
@@ -2848,6 +3272,7 @@ GLboolean assemble_LRP(r700_AssemblerBase *pAsm)
     {
         return GL_FALSE;
     }
+
     if( GL_FALSE == assemble_src(pAsm, 2, -1) ) 
     {
         return GL_FALSE;
@@ -2878,6 +3303,217 @@ GLboolean assemble_LRP(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
+GLboolean assemble_LOG(r700_AssemblerBase *pAsm)
+{
+    BITS tmp1, tmp2, tmp3;
+
+    checkop1(pAsm);
+
+    tmp1 = gethelpr(pAsm);
+    tmp2 = gethelpr(pAsm);
+    tmp3 = gethelpr(pAsm);
+
+    // FIXME: The hardware can do fabs() directly on input
+    //        elements, but the compiler doesn't have the
+    //        capability to use that.
+
+    // MAX     tmp1.x,   a.x,    -a.x   (fabs(a.x))
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MAX;  
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp1;
+    pAsm->D.dst.writex = 1;
+
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+        return GL_FALSE;
+    }
+ 
+    pAsm->S[1].bits = pAsm->S[0].bits;
+    flipneg_PVSSRC(&(pAsm->S[1].src));
+
+    if ( GL_FALSE == next_ins(pAsm) ) 
+    {
+        return GL_FALSE;
+    }
+
+    // Entire algo:
+    //
+    // LG2     tmp2.x,   tmp1.x
+    // FLOOR   tmp3.x,   tmp2.x
+    // MOV     dst.x,    tmp3.x
+    // ADD     tmp3.x,   tmp2.x,    -tmp3.x
+    // EX2     dst.y,    tmp3.x
+    // MOV     dst.z,    tmp2.x
+    // MOV     dst.w,    1.0
+
+    // LG2     tmp2.x,   tmp1.x
+    // FLOOR   tmp3.x,   tmp2.x
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_LOG_IEEE;
+    pAsm->D.dst.math = 1;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp2;
+    pAsm->D.dst.writex = 1;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp1;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_FLOOR;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp3;
+    pAsm->D.dst.writex = 1;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp2;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    // MOV     dst.x,    tmp3.x
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.writey = pAsm->D.dst.writez = pAsm->D.dst.writew = 0;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp3;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    // ADD     tmp3.x,   tmp2.x,    -tmp3.x
+    // EX2     dst.y,    tmp3.x
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_ADD;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp3;
+    pAsm->D.dst.writex = 1;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp2;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    pAsm->S[1].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[1].src.reg   = tmp3;
+
+    setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+    neg_PVSSRC(&(pAsm->S[1].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_EXP_IEEE;
+    pAsm->D.dst.math = 1;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.writex = pAsm->D.dst.writez = pAsm->D.dst.writew = 0;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp3;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    // MOV     dst.z,    tmp2.x
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.writex = pAsm->D.dst.writey = pAsm->D.dst.writew = 0;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp2;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    // MOV     dst.w     1.0
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.writex = pAsm->D.dst.writey = pAsm->D.dst.writez = 0;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp1;
+
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_1);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    return GL_TRUE;
+}
+
 GLboolean assemble_MAD(struct r700_AssemblerBase *pAsm) 
 {
     int tmp, ii;
@@ -3035,6 +3671,7 @@ GLboolean assemble_LIT(r700_AssemblerBase *pAsm)
     pAsm->S[0].src.rtype = srcType;
     pAsm->S[0].src.reg   = srcReg;
     setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    swizzleagain_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X, SQ_SEL_X, SQ_SEL_X, SQ_SEL_X);
     pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
     pAsm->S[1].src.reg   = tmp;
     setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
@@ -3371,77 +4008,137 @@ GLboolean assemble_RSQ(r700_AssemblerBase *pAsm)
     return assemble_math_function(pAsm, SQ_OP2_INST_RECIPSQRT_IEEE);
 }
  
-GLboolean assemble_SIN(r700_AssemblerBase *pAsm) 
-{
-    return assemble_math_function(pAsm, SQ_OP2_INST_SIN);
-}
- 
 GLboolean assemble_SCS(r700_AssemblerBase *pAsm) 
 {
     BITS tmp;
 
-	checkop1(pAsm);
+    checkop1(pAsm);
 
-	tmp = gethelpr(pAsm);
+    tmp = gethelpr(pAsm);
+    /* tmp.x = src /2*PI */
+    pAsm->D.dst.opcode = SQ_OP2_INST_MUL;
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+    pAsm->D.dst.writex = 1;
 
-	// COS tmp.x,    a.x
-	pAsm->D.dst.opcode = SQ_OP2_INST_COS;
-	pAsm->D.dst.math = 1;
+    assemble_src(pAsm, 0, -1);
 
-	setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
-	pAsm->D.dst.rtype = DST_REG_TEMPORARY;
-	pAsm->D.dst.reg = tmp;
-	pAsm->D.dst.writex = 1;
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+    pAsm->D2.dst2.literal_slots = 1;
+    pAsm->C[0].f = 1/(3.1415926535 * 2);
+    pAsm->C[1].f = 0.0F;
 
-	if( GL_FALSE == assemble_src(pAsm, 0, -1) )
-	{
-		return GL_FALSE;
-	}
+    next_ins(pAsm);
 
-	if ( GL_FALSE == next_ins(pAsm) )
-	{
-		return GL_FALSE;
-	}
+    // COS dst.x,    a.x
+    pAsm->D.dst.opcode = SQ_OP2_INST_COS;
+    pAsm->D.dst.math = 1;
 
-	// SIN tmp.y,    a.x
-	pAsm->D.dst.opcode = SQ_OP2_INST_SIN;
-	pAsm->D.dst.math = 1;
+    assemble_dst(pAsm);
+    /* mask y */
+    pAsm->D.dst.writey = 0;
 
-	setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
-	pAsm->D.dst.rtype = DST_REG_TEMPORARY;
-	pAsm->D.dst.reg = tmp;
-	pAsm->D.dst.writey = 1;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp;
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
 
-	if( GL_FALSE == assemble_src(pAsm, 0, -1) )
-	{
-		return GL_FALSE;
-	}
+    if ( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
 
-	if( GL_FALSE == next_ins(pAsm) )
-	{
-		return GL_FALSE;
-	}
+    // SIN dst.y,    a.x
+    pAsm->D.dst.opcode = SQ_OP2_INST_SIN;
+    pAsm->D.dst.math = 1;
 
-	// MOV dst.mask,     tmp
-	pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+    assemble_dst(pAsm);
+    /* mask x */
+    pAsm->D.dst.writex = 0;
 
-	if( GL_FALSE == assemble_dst(pAsm) )
-	{
-		return GL_FALSE;
-	}
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = tmp;
+    setswizzle_PVSSRC(&(pAsm->S[0].src), SQ_SEL_X);
+    noneg_PVSSRC(&(pAsm->S[0].src));
 
-	setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
-	pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
-	pAsm->S[0].src.reg = tmp;
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
 
-	noswizzle_PVSSRC(&(pAsm->S[0].src));
-	pAsm->S[0].src.swizzlez = SQ_SEL_0;
-	pAsm->S[0].src.swizzlew = SQ_SEL_0;
+    return GL_TRUE;
+}
 
-	if ( GL_FALSE == next_ins(pAsm) )
-	{
-		return GL_FALSE;
-	}
+GLboolean assemble_LOGIC(r700_AssemblerBase *pAsm, BITS opcode) 
+{
+    if( GL_FALSE == checkop2(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = opcode;
+    //pAsm->D.dst.math   = 1;
+
+    if( GL_FALSE == assemble_dst(pAsm) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 0, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == assemble_src(pAsm, 1, -1) )
+    {
+	    return GL_FALSE;
+    }
+
+    if( GL_FALSE == next_ins(pAsm) ) 
+    {
+	    return GL_FALSE;
+    }
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_LOGIC_PRED(r700_AssemblerBase *pAsm, BITS opcode) 
+{
+    struct prog_instruction *pILInst = &(pAsm->pILInst[pAsm->uiCurInst]);
+
+    pAsm->D.dst.opcode = opcode;
+    pAsm->D.dst.math   = 1;
+    pAsm->D.dst.predicated = 1;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg = pAsm->uHelpReg;
+    pAsm->D.dst.writex = 1;
+    pAsm->D.dst.writey = pAsm->D.dst.writez = pAsm->D.dst.writew = 0;
+
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[0].src.reg = pAsm->last_cond_register + pAsm->starting_temp_register_number;
+    pAsm->S[0].src.swizzlex = pILInst->DstReg.CondSwizzle & 0x7;
+    noneg_PVSSRC(&(pAsm->S[0].src));
+
+    pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
+    pAsm->S[1].src.reg   = pAsm->uHelpReg;
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[1].src));
+    pAsm->S[1].src.swizzlex = SQ_SEL_0;
+    pAsm->S[1].src.swizzley = SQ_SEL_0;
+    pAsm->S[1].src.swizzlez = SQ_SEL_0;
+    pAsm->S[1].src.swizzlew = SQ_SEL_0;
+
+    if( GL_FALSE == next_ins(pAsm) ) 
+    {
+	    return GL_FALSE;
+    }
 
     return GL_TRUE;
 }
@@ -3524,6 +4221,7 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
     
     switch (pAsm->pILInst[pAsm->uiCurInst].SrcReg[0].File)
     {
+    case PROGRAM_UNIFORM: 
     case PROGRAM_CONSTANT:
     case PROGRAM_LOCAL_PARAM:
     case PROGRAM_ENV_PARAM:
@@ -3544,22 +4242,6 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
 	    need_barrier = GL_TRUE;
     }
 
-    switch (pAsm->pILInst[pAsm->uiCurInst].Opcode)
-    {
-        case OPCODE_TEX:
-            break;
-        case OPCODE_TXB:
-            radeon_error("do not support TXB yet\n");
-            return GL_FALSE;
-            break;
-        case OPCODE_TXP:
-            break;
-        default:
-            radeon_error("Internal error: bad texture op (not TEX)\n");
-            return GL_FALSE;
-            break;
-    }
-
     if (pAsm->pILInst[pAsm->uiCurInst].Opcode == OPCODE_TXP)
     {
         GLuint tmp = gethelpr(pAsm);
@@ -3637,24 +4319,6 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
             return GL_FALSE;
         }
  
-        /* tmp1.z = ABS(tmp1.z) dont have abs support in assembler currently
-         * have to do explicit instruction
-         */
-        pAsm->D.dst.opcode = SQ_OP2_INST_MAX;
-        setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
-        pAsm->D.dst.rtype = DST_REG_TEMPORARY;
-        pAsm->D.dst.reg   = tmp1;
-        pAsm->D.dst.writez = 1;
-
-        setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
-        pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
-        pAsm->S[0].src.reg = tmp1;
-	noswizzle_PVSSRC(&(pAsm->S[0].src));
-        pAsm->S[1].bits = pAsm->S[0].bits;
-        flipneg_PVSSRC(&(pAsm->S[1].src));
-        
-        next_ins(pAsm);
-
         /* tmp1.z = RCP_e(|tmp1.z|) */
         pAsm->D.dst.opcode = SQ_OP2_INST_RECIP_IEEE;
         pAsm->D.dst.math = 1;
@@ -3667,13 +4331,13 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
         pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
         pAsm->S[0].src.reg = tmp1;
         pAsm->S[0].src.swizzlex = SQ_SEL_Z;
+        pAsm->S[0].src.abs = 1;
 
         next_ins(pAsm);
 
         /* MULADD R0.x,  R0.x,  PS1,  (0x3FC00000, 1.5f).x
          * MULADD R0.y,  R0.y,  PS1,  (0x3FC00000, 1.5f).x
          * muladd has no writemask, have to use another temp 
-         * also no support for imm constants, so add 1 here
          */
         pAsm->D.dst.opcode = SQ_OP3_INST_MULADD;
         pAsm->D.dst.op3    = 1;
@@ -3690,30 +4354,12 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
         pAsm->S[1].src.reg   = tmp1;
         setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_Z);
         setaddrmode_PVSSRC(&(pAsm->S[2].src), ADDR_ABSOLUTE);
-        pAsm->S[2].src.rtype = SRC_REG_TEMPORARY;
+        /* immediate c 1.5 */
+        pAsm->D2.dst2.literal_slots = 1;
+        pAsm->C[0].f = 1.5F;
+        pAsm->S[2].src.rtype = SRC_REC_LITERAL;
         pAsm->S[2].src.reg   = tmp1;
-        setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_1);
-
-        next_ins(pAsm);
-
-        /* ADD the remaining .5 */
-        pAsm->D.dst.opcode = SQ_OP2_INST_ADD;
-        setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
-        pAsm->D.dst.rtype = DST_REG_TEMPORARY;
-        pAsm->D.dst.reg   = tmp2;
-        pAsm->D.dst.writex = 1;
-        pAsm->D.dst.writey = 1;
-        pAsm->D.dst.writez = 0;
-        pAsm->D.dst.writew = 0;
-
-        setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
-        pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
-        pAsm->S[0].src.reg   = tmp2;
-        noswizzle_PVSSRC(&(pAsm->S[0].src));
-        setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
-        pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
-        pAsm->S[1].src.reg   = 252; // SQ_ALU_SRC_0_5 
-        noswizzle_PVSSRC(&(pAsm->S[1].src));
+        setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_X);
 
         next_ins(pAsm);
 
@@ -3738,14 +4384,35 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
 
     }
 
-    pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE;
+    switch(pAsm->pILInst[pAsm->uiCurInst].Opcode)
+    {
+        case OPCODE_DDX:
+            /* will these need WQM(1) on CF inst ? */
+            pAsm->D.dst.opcode = SQ_TEX_INST_GET_GRADIENTS_H;
+            break;
+        case OPCODE_DDY:
+            pAsm->D.dst.opcode = SQ_TEX_INST_GET_GRADIENTS_V;
+            break;
+        case OPCODE_TXB:
+            pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE_L;
+            break;
+        default:
+            if(pAsm->pILInst[pAsm->uiCurInst].TexShadow == 1)
+                pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE_C;
+            else
+                pAsm->D.dst.opcode = SQ_TEX_INST_SAMPLE;
+    }
+
+    pAsm->is_tex = GL_TRUE;
+    if ( GL_TRUE == need_barrier )
+
     pAsm->is_tex = GL_TRUE;
     if ( GL_TRUE == need_barrier )
     {
         pAsm->need_tex_barrier = GL_TRUE;
     }
     // Set src1 to tex unit id
-    pAsm->S[1].src.reg   = pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit;
+    pAsm->S[1].src.reg   = pAsm->SamplerUnits[pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit];
     pAsm->S[1].src.rtype = SRC_REG_TEMPORARY;
 
     //No sw info from mesa compiler, so hard code here.
@@ -3779,11 +4446,46 @@ GLboolean assemble_TEX(r700_AssemblerBase *pAsm)
         pAsm->S[0].src.swizzlew = SQ_SEL_Y;
     }
  
+    if(pAsm->pILInst[pAsm->uiCurInst].TexShadow == 1)
+    {
+        /* compare value goes to w chan ? */
+        pAsm->S[0].src.swizzlew = SQ_SEL_Z;
+    }
+
     if ( GL_FALSE == next_ins(pAsm) )
         {
             return GL_FALSE;
         }
 
+    /* add ARB shadow ambient but clamp to 0..1 */
+    if(pAsm->pILInst[pAsm->uiCurInst].TexShadow == 1)
+    {
+	/* ADD_SAT dst,  dst,  ambient[texunit] */
+	pAsm->D.dst.opcode = SQ_OP2_INST_ADD;
+
+	if( GL_FALSE == assemble_dst(pAsm) )
+	{
+	    return GL_FALSE;
+	}
+	pAsm->D2.dst2.SaturateMode = 1;
+
+	pAsm->S[0].src.rtype = pAsm->D.dst.rtype;
+	pAsm->S[0].src.reg = pAsm->D.dst.reg;
+	noswizzle_PVSSRC(&(pAsm->S[0].src));
+	noneg_PVSSRC(&(pAsm->S[0].src));
+
+	pAsm->S[1].src.rtype = SRC_REG_CONSTANT;
+	pAsm->S[1].src.reg = pAsm->shadow_regs[pAsm->pILInst[pAsm->uiCurInst].TexSrcUnit];
+	noswizzle_PVSSRC(&(pAsm->S[1].src));
+	noneg_PVSSRC(&(pAsm->S[1].src));
+
+	if( GL_FALSE == next_ins(pAsm) )
+	{
+	    return GL_FALSE;
+	}
+
+    }
+
     return GL_TRUE;
 }
 
@@ -3902,27 +4604,909 @@ GLboolean assemble_EXPORT(r700_AssemblerBase *pAsm)
     return GL_TRUE;
 }
 
-GLboolean assemble_IF(r700_AssemblerBase *pAsm)
+static inline void decreaseCurrent(r700_AssemblerBase *pAsm, GLuint uReason)
 {
+    switch (uReason)
+    {
+    case FC_PUSH_VPM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current--;
+        break;
+    case FC_PUSH_WQM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 4;
+        break;
+    case FC_LOOP:
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 4;
+        break;
+    case FC_REP:
+        /* TODO : for 16 vp asic, should -= 2; */
+        pAsm->CALLSTACK[pAsm->CALLSP].current -= 1;
+        break;
+    };
+}
+
+static inline void checkStackDepth(r700_AssemblerBase *pAsm, GLuint uReason, GLboolean bCheckMaxOnly)
+{
+    if(GL_TRUE == bCheckMaxOnly)
+    {
+        switch (uReason)
+        {
+        case FC_PUSH_VPM:
+            if((pAsm->CALLSTACK[pAsm->CALLSP].current + 1)
+                    > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max =
+                    pAsm->CALLSTACK[pAsm->CALLSP].current + 1;
+            }
+            break;
+        case FC_PUSH_WQM:
+            if((pAsm->CALLSTACK[pAsm->CALLSP].current + 4)
+                    > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max =
+                    pAsm->CALLSTACK[pAsm->CALLSP].current + 4;
+            }
+            break;
+        }
+        return;
+    }
+
+    switch (uReason)
+    {
+    case FC_PUSH_VPM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current++;
+        break;
+    case FC_PUSH_WQM:
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 4;
+        break;
+    case FC_LOOP:
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 4;
+        break;
+    case FC_REP:
+        /* TODO : for 16 vp asic, should += 2; */
+        pAsm->CALLSTACK[pAsm->CALLSP].current += 1;
+        break;
+    };
+
+    if(pAsm->CALLSTACK[pAsm->CALLSP].current
+         > pAsm->CALLSTACK[pAsm->CALLSP].max)
+    {
+        pAsm->CALLSTACK[pAsm->CALLSP].max =
+            pAsm->CALLSTACK[pAsm->CALLSP].current;
+    }
+}
+
+GLboolean jumpToOffest(r700_AssemblerBase *pAsm, GLuint pops, GLint offset)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = pops;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_JUMP;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex + offset;
+
+    return GL_TRUE;
+}
+
+GLboolean pops(r700_AssemblerBase *pAsm, GLuint pops)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = pops;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_IF(r700_AssemblerBase *pAsm, GLboolean bHasElse)
+{
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+
+    assemble_LOGIC_PRED(pAsm, SQ_OP2_INST_PRED_SETNE);
+
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    if(GL_TRUE != bHasElse)
+    {
+        pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count = 1; 
+    }
+    else
+    {
+        pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count = 0;
+    }
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_JUMP;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->FCSP++;
+	pAsm->fc_stack[pAsm->FCSP].type  = FC_IF;
+    pAsm->fc_stack[pAsm->FCSP].mid   = NULL;
+    pAsm->fc_stack[pAsm->FCSP].midLen= 0;
+    pAsm->fc_stack[pAsm->FCSP].first = pAsm->cf_current_cf_clause_ptr;
+
+#ifndef USE_CF_FOR_POP_AFTER
+    if(GL_TRUE != bHasElse)
+    {
+        pAsm->alu_x_opcode = SQ_CF_INST_ALU_POP_AFTER;
+    }
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_FALSE); 
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_ELSE(r700_AssemblerBase *pAsm)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1; ///
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_ELSE;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[pAsm->FCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( (void *)pAsm->fc_stack[pAsm->FCSP].mid,
+                                                                                     0,
+                                                                                     sizeof(R700ControlFlowGenericClause *) );
+    pAsm->fc_stack[pAsm->FCSP].mid[0] = pAsm->cf_current_cf_clause_ptr;
+    //pAsm->fc_stack[pAsm->FCSP].unNumMid = 1;
+
+#ifndef USE_CF_FOR_POP_AFTER
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_POP_AFTER;
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode - 1; 
+
     return GL_TRUE;
 }
 
 GLboolean assemble_ENDIF(r700_AssemblerBase *pAsm)
 {
+#ifdef USE_CF_FOR_POP_AFTER
+    pops(pAsm, 1); 
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    if(NULL == pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        /* no else in between */
+        pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode;
+    }
+    else
+    {
+        pAsm->fc_stack[pAsm->FCSP].mid[0]->m_Word0.f.addr = pAsm->pR700Shader->plstCFInstructions_active->uNumOfNode;
+    }
+
+    if(NULL != pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        FREE(pAsm->fc_stack[pAsm->FCSP].mid);
+    }
+
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_IF)
+    {
+        radeon_error("if/endif in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+    
+    pAsm->FCSP--;
+
+    decreaseCurrent(pAsm, FC_PUSH_VPM);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_BGNLOOP(r700_AssemblerBase *pAsm)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_START_NO_AL;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->FCSP++;
+	pAsm->fc_stack[pAsm->FCSP].type  = FC_LOOP;
+    pAsm->fc_stack[pAsm->FCSP].mid   = NULL;
+    pAsm->fc_stack[pAsm->FCSP].unNumMid = 0;
+    pAsm->fc_stack[pAsm->FCSP].midLen   = 0;
+    pAsm->fc_stack[pAsm->FCSP].first    = pAsm->cf_current_cf_clause_ptr;
+
+    checkStackDepth(pAsm, FC_LOOP, GL_FALSE);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_BRK(r700_AssemblerBase *pAsm)
+{
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+
+    assemble_LOGIC_PRED(pAsm, SQ_OP2_INST_PRED_SETNE);
+    
+    unsigned int unFCSP;
+    for(unFCSP=pAsm->FCSP; unFCSP>0; unFCSP--)
+    {
+        if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+        {
+            break;
+        }
+    }
+    if(0 == FC_LOOP)
+    {
+        radeon_error("Break is not inside loop/endloop pair.\n");
+        return GL_FALSE;
+    }
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_BREAK;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+#endif //USE_CF_FOR_CONTINUE_BREAK
+    return GL_TRUE;
+}
+
+GLboolean assemble_CONT(r700_AssemblerBase *pAsm)
+{
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+
+    assemble_LOGIC_PRED(pAsm, SQ_OP2_INST_PRED_SETNE);
+
+    unsigned int unFCSP;
+    for(unFCSP=pAsm->FCSP; unFCSP>0; unFCSP--)
+    {
+        if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+        {
+            break;
+        }
+    }
+    if(0 == FC_LOOP)
+    {
+        radeon_error("Continue is not inside loop/endloop pair.\n");
+        return GL_FALSE;
+    }
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_CONTINUE;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_POP;
+ 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr             = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+#endif /* USE_CF_FOR_CONTINUE_BREAK */
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_ENDLOOP(r700_AssemblerBase *pAsm)
+{
+    GLuint i;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_END;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word0.f.addr   = pAsm->fc_stack[pAsm->FCSP].first->m_uIndex + 1;
+    pAsm->fc_stack[pAsm->FCSP].first->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex + 1;
+
+#ifdef USE_CF_FOR_CONTINUE_BREAK
+    for(i=0; i<pAsm->fc_stack[pAsm->FCSP].unNumMid; i++)
+    {
+        pAsm->fc_stack[pAsm->FCSP].mid[i]->m_Word0.f.addr = pAsm->cf_current_cf_clause_ptr->m_uIndex;
+    }
+    if(NULL != pAsm->fc_stack[pAsm->FCSP].mid)
+    {
+        FREE(pAsm->fc_stack[pAsm->FCSP].mid);
+    }
+#endif
+
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_LOOP)
+    {
+        radeon_error("loop/endloop in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+
+    GLuint unFCSP;
+    GLuint unIF = 0;
+    if((pAsm->unCFflags & HAS_CURRENT_LOOPRET) > 0)
+    {        
+        for(unFCSP=(pAsm->FCSP-1); unFCSP>pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry; unFCSP--)
+        {
+            if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+            {
+                breakLoopOnFlag(pAsm, unFCSP);
+                break;
+            }
+            else if(FC_IF == pAsm->fc_stack[unFCSP].type)
+            {
+                unIF++;
+            }
+        }
+        if(unFCSP <= pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry)
+        {            
+#ifdef USE_CF_FOR_POP_AFTER
+            returnOnFlag(pAsm, unIF); 
+#else
+            returnOnFlag(pAsm, 0);
+#endif /* USE_CF_FOR_POP_AFTER */
+            pAsm->unCFflags &= ~HAS_CURRENT_LOOPRET;
+        }
+    }
+
+    pAsm->FCSP--;
+
+    decreaseCurrent(pAsm, FC_LOOP);
+    
+    return GL_TRUE;
+}
+
+void add_return_inst(r700_AssemblerBase *pAsm)
+{
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+    //pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_RETURN;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+}
+
+GLboolean assemble_BGNSUB(r700_AssemblerBase *pAsm, GLint nILindex, GLuint uiIL_Shift)
+{
+    /* Put in sub */
+    if( (pAsm->unSubArrayPointer + 1) > pAsm->unSubArraySize )
+    {
+        pAsm->subs = (SUB_OFFSET*)_mesa_realloc( (void *)pAsm->subs,
+                                  sizeof(SUB_OFFSET) * pAsm->unSubArraySize,
+                                  sizeof(SUB_OFFSET) * (pAsm->unSubArraySize + 10) );
+        if(NULL == pAsm->subs)
+        {
+            return GL_FALSE;
+        }
+        pAsm->unSubArraySize += 10;
+    }
+
+    pAsm->subs[pAsm->unSubArrayPointer].subIL_Offset = nILindex + uiIL_Shift;
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.pHead=NULL;  
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.pTail=NULL;  
+    pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local.uNumOfNode=0;
+
+    pAsm->CALLSP++;
+    pAsm->CALLSTACK[pAsm->CALLSP].subDescIndex = pAsm->unSubArrayPointer;
+    pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry = pAsm->FCSP;
+    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local
+                   = &(pAsm->subs[pAsm->unSubArrayPointer].lstCFInstructions_local);
+    pAsm->CALLSTACK[pAsm->CALLSP].max = 0;
+    pAsm->CALLSTACK[pAsm->CALLSP].current = 0;
+    SetActiveCFlist(pAsm->pR700Shader, 
+                    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local);
+
+    pAsm->unSubArrayPointer++;
+
+    /* start sub */
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    pAsm->FCSP++;
+    pAsm->fc_stack[pAsm->FCSP].type  = FC_REP;
+
+    checkStackDepth(pAsm, FC_REP, GL_FALSE);
+
     return GL_TRUE;
 }
 
-GLboolean AssembleInstr(GLuint uiNumberInsts,
+GLboolean assemble_ENDSUB(r700_AssemblerBase *pAsm)
+{
+    if(pAsm->fc_stack[pAsm->FCSP].type != FC_REP)
+    {
+        radeon_error("BGNSUB/ENDSUB in shader code are not paired. \n");
+        return GL_FALSE;
+    }
+
+    /* copy max to sub structure */
+    pAsm->subs[pAsm->CALLSTACK[pAsm->CALLSP].subDescIndex].unStackDepthMax
+        = pAsm->CALLSTACK[pAsm->CALLSP].max;
+
+    decreaseCurrent(pAsm, FC_REP);
+
+    pAsm->CALLSP--;
+    SetActiveCFlist(pAsm->pR700Shader, 
+                    pAsm->CALLSTACK[pAsm->CALLSP].plstCFInstructions_local);
+    
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    pAsm->FCSP--;
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_RET(r700_AssemblerBase *pAsm)
+{
+    GLuint unIF = 0;
+
+    if(pAsm->CALLSP > 0)
+    {   /* in sub */
+        GLuint unFCSP;        
+        for(unFCSP=pAsm->FCSP; unFCSP>pAsm->CALLSTACK[pAsm->CALLSP].FCSP_BeforeEntry; unFCSP--)
+        {
+            if(FC_LOOP == pAsm->fc_stack[unFCSP].type)
+            {
+                setRetInLoopFlag(pAsm, SQ_SEL_1);
+                breakLoopOnFlag(pAsm, unFCSP);
+                pAsm->unCFflags |= LOOPRET_FLAGS;
+
+                return GL_TRUE;
+            }
+            else if(FC_IF == pAsm->fc_stack[unFCSP].type)
+            {
+                unIF++;
+            }
+        }
+    }
+
+#ifdef USE_CF_FOR_POP_AFTER    
+    if(unIF > 0)
+    {
+        pops(pAsm, unIF);
+    }
+#endif /* USE_CF_FOR_POP_AFTER */
+
+    add_return_inst(pAsm);
+
+    return GL_TRUE;
+}
+
+GLboolean assemble_CAL(r700_AssemblerBase *pAsm, 
+                       GLint nILindex,
+                       GLuint uiIL_Shift,
+                       GLuint uiNumberInsts,
+                       struct prog_instruction *pILInst,
+                       PRESUB_DESC * pPresubDesc)
+{
+    GLint uiIL_Offset;
+
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.call_count       = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_CALL;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    /* Put in caller */
+    if( (pAsm->unCallerArrayPointer + 1) > pAsm->unCallerArraySize )
+    {
+        pAsm->callers = (CALLER_POINTER*)_mesa_realloc( (void *)pAsm->callers, 
+                       sizeof(CALLER_POINTER) * pAsm->unCallerArraySize, 
+                       sizeof(CALLER_POINTER) * (pAsm->unCallerArraySize + 10) );
+        if(NULL == pAsm->callers)
+        {
+            return GL_FALSE;
+        }
+        pAsm->unCallerArraySize += 10;
+    }
+    
+    uiIL_Offset = nILindex + uiIL_Shift;
+    pAsm->callers[pAsm->unCallerArrayPointer].subIL_Offset = uiIL_Offset; 
+    pAsm->callers[pAsm->unCallerArrayPointer].cf_ptr       = pAsm->cf_current_cf_clause_ptr;
+    
+    pAsm->callers[pAsm->unCallerArrayPointer].finale_cf_ptr  = NULL; 
+    pAsm->callers[pAsm->unCallerArrayPointer].prelude_cf_ptr = NULL; 
+
+    pAsm->unCallerArrayPointer++;
+
+    int j;
+    GLuint max;
+    GLuint unSubID;
+    GLboolean bRet;
+    for(j=0; j<pAsm->unSubArrayPointer; j++)
+    {
+        if(uiIL_Offset == pAsm->subs[j].subIL_Offset)
+        {   /* compiled before */
+
+            max = pAsm->subs[j].unStackDepthMax 
+                + pAsm->CALLSTACK[pAsm->CALLSP].current;
+            if(max > pAsm->CALLSTACK[pAsm->CALLSP].max)
+            {
+                pAsm->CALLSTACK[pAsm->CALLSP].max = max;
+            }
+            
+            pAsm->callers[pAsm->unCallerArrayPointer - 1].subDescIndex = j; 
+            return GL_TRUE;
+        }
+    }
+
+    pAsm->callers[pAsm->unCallerArrayPointer - 1].subDescIndex = pAsm->unSubArrayPointer;
+    unSubID = pAsm->unSubArrayPointer;
+
+    bRet = AssembleInstr(nILindex, uiIL_Shift, uiNumberInsts, pILInst, pAsm);
+
+    if(GL_TRUE == bRet)
+    {
+        max = pAsm->subs[unSubID].unStackDepthMax 
+            + pAsm->CALLSTACK[pAsm->CALLSP].current;
+        if(max > pAsm->CALLSTACK[pAsm->CALLSP].max)
+        {
+            pAsm->CALLSTACK[pAsm->CALLSP].max = max;
+        }
+
+        pAsm->subs[unSubID].pPresubDesc = pPresubDesc;
+    }
+
+    return bRet;
+}
+
+GLboolean setRetInLoopFlag(r700_AssemblerBase *pAsm, GLuint flagValue)
+{
+    GLfloat fLiteral[2] = {0.1, 0.0};
+
+    pAsm->D.dst.opcode   = SQ_OP2_INST_MOV;
+    pAsm->D.dst.op3      = 0;
+    pAsm->D.dst.rtype    = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg      = pAsm->flag_reg_index;
+    pAsm->D.dst.writex   = 1;
+    pAsm->D.dst.writey   = 0;
+    pAsm->D.dst.writez   = 0;
+    pAsm->D.dst.writew   = 0;
+    pAsm->D2.dst2.literal_slots      = 1;
+    pAsm->D2.dst2.SaturateMode = SATURATE_OFF;
+    pAsm->D.dst.predicated     = 0;
+    /* in reloc where dislink flag init inst, only one slot alu inst is handled. */
+    pAsm->D.dst.math           = 1; /* TODO : not math really, but one channel op, more generic alu assembler needed */
+    pAsm->D2.dst2.index_mode = SQ_INDEX_LOOP; /* Check this ! */
+#if 0
+    pAsm->S[0].src.rtype = SRC_REC_LITERAL;
+    //pAsm->S[0].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = SQ_SEL_X;
+    pAsm->S[0].src.swizzley = SQ_SEL_Y;
+    pAsm->S[0].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[0].src.swizzlew = SQ_SEL_W;
+
+    if( GL_FALSE == next_ins_literal(pAsm, &(fLiteral[0])) )
+    {
+        return GL_FALSE;
+    }
+#else
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = flagValue;
+    pAsm->S[0].src.swizzley = flagValue;
+    pAsm->S[0].src.swizzlez = flagValue;
+    pAsm->S[0].src.swizzlew = flagValue;
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+#endif
+
+    return GL_TRUE;
+}
+
+GLboolean testFlag(r700_AssemblerBase *pAsm)
+{
+    GLfloat fLiteral[2] = {0.1, 0.0};
+
+    //Test flag
+    GLuint tmp = gethelpr(pAsm);
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU_PUSH_BEFORE;
+
+    pAsm->D.dst.opcode   = SQ_OP2_INST_PRED_SETE;
+    pAsm->D.dst.math     = 1;
+    pAsm->D.dst.rtype    = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg      = tmp;
+    pAsm->D.dst.writex   = 1;
+    pAsm->D.dst.writey   = 0;
+    pAsm->D.dst.writez   = 0;
+    pAsm->D.dst.writew   = 0;
+    pAsm->D2.dst2.literal_slots      = 1;
+    pAsm->D2.dst2.SaturateMode = SATURATE_OFF;
+    pAsm->D.dst.predicated     = 1;
+    pAsm->D2.dst2.index_mode = SQ_INDEX_LOOP; /* Check this ! */
+
+    pAsm->S[0].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[0].src.reg   = pAsm->flag_reg_index;
+    setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[0].src));
+    pAsm->S[0].src.swizzlex = SQ_SEL_X;
+    pAsm->S[0].src.swizzley = SQ_SEL_Y;
+    pAsm->S[0].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[0].src.swizzlew = SQ_SEL_W;
+#if 0
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    //pAsm->S[1].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[1].src));
+    pAsm->S[1].src.swizzlex = SQ_SEL_X;
+    pAsm->S[1].src.swizzley = SQ_SEL_Y;
+    pAsm->S[1].src.swizzlez = SQ_SEL_Z;
+    pAsm->S[1].src.swizzlew = SQ_SEL_W;
+
+    if( GL_FALSE == next_ins_literal(pAsm, &(fLiteral[0])) )
+    {
+        return GL_FALSE;
+    }
+#else
+    pAsm->S[1].src.rtype = DST_REG_TEMPORARY;
+    pAsm->S[1].src.reg   = 0;
+    setaddrmode_PVSSRC(&(pAsm->S[1].src), ADDR_ABSOLUTE);
+    noneg_PVSSRC(&(pAsm->S[1].src));
+    pAsm->S[1].src.swizzlex = SQ_SEL_1;
+    pAsm->S[1].src.swizzley = SQ_SEL_1;
+    pAsm->S[1].src.swizzlez = SQ_SEL_1;
+    pAsm->S[1].src.swizzlew = SQ_SEL_1;
+
+    if( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+#endif
+
+    checkStackDepth(pAsm, FC_PUSH_VPM, GL_TRUE);
+
+    return GL_TRUE;
+}
+
+GLboolean returnOnFlag(r700_AssemblerBase *pAsm, GLuint unIF)
+{
+    testFlag(pAsm);
+    jumpToOffest(pAsm, 1, 4);
+    setRetInLoopFlag(pAsm, SQ_SEL_0);
+    pops(pAsm, unIF + 1);
+    add_return_inst(pAsm);
+
+    return GL_TRUE;
+}
+
+GLboolean breakLoopOnFlag(r700_AssemblerBase *pAsm, GLuint unFCSP)
+{
+    testFlag(pAsm);
+ 
+    //break
+    if(GL_FALSE == add_cf_instruction(pAsm) )
+    {
+        return GL_FALSE;
+    }
+    
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.pop_count        = 1;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_const         = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cond             = SQ_CF_COND_ACTIVE;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.end_of_program   = 0x0;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.valid_pixel_mode = 0x0; 
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.cf_inst          = SQ_CF_INST_LOOP_BREAK;
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.whole_quad_mode  = 0x0;
+
+    pAsm->cf_current_cf_clause_ptr->m_Word1.f.barrier          = 0x1;
+
+    pAsm->fc_stack[unFCSP].mid = (R700ControlFlowGenericClause **)_mesa_realloc( 
+                                              (void *)pAsm->fc_stack[unFCSP].mid,
+                                              sizeof(R700ControlFlowGenericClause *) * pAsm->fc_stack[unFCSP].unNumMid,
+                                              sizeof(R700ControlFlowGenericClause *) * (pAsm->fc_stack[unFCSP].unNumMid + 1) );
+    pAsm->fc_stack[unFCSP].mid[pAsm->fc_stack[unFCSP].unNumMid] = pAsm->cf_current_cf_clause_ptr;
+    pAsm->fc_stack[unFCSP].unNumMid++;
+
+    pops(pAsm, 1);
+               
+    return GL_TRUE;
+}
+
+GLboolean AssembleInstr(GLuint uiFirstInst,
+                        GLuint uiIL_Shift,
+                        GLuint uiNumberInsts,
                         struct prog_instruction *pILInst, 
 						r700_AssemblerBase *pR700AsmCode)
 {
     GLuint i;
 
     pR700AsmCode->pILInst = pILInst;
-	for(i=0; i<uiNumberInsts; i++)
+	for(i=uiFirstInst; i<uiNumberInsts; i++)
     {
         pR700AsmCode->uiCurInst = i;
 
+#ifndef USE_CF_FOR_CONTINUE_BREAK
+        if(OPCODE_BRK == pILInst[i+1].Opcode)
+        {
+            switch(pILInst[i].Opcode)            
+            {
+            case OPCODE_SLE:
+                pILInst[i].Opcode = OPCODE_SGT;
+                break;
+            case OPCODE_SLT:
+                pILInst[i].Opcode = OPCODE_SGE;
+                break;
+            case OPCODE_SGE:
+                pILInst[i].Opcode = OPCODE_SLT;
+                break;
+            case OPCODE_SGT:
+                pILInst[i].Opcode = OPCODE_SLE;
+                break;
+            case OPCODE_SEQ:
+                pILInst[i].Opcode = OPCODE_SNE;
+                break;
+            case OPCODE_SNE:
+                pILInst[i].Opcode = OPCODE_SEQ;
+                break;
+            default:
+                break;
+            }
+        }
+#endif
+        if(pILInst[i].CondUpdate == 1)
+        {
+            /* remember dest register used for cond evaluation */
+            /* XXX also handle PROGRAM_OUTPUT registers here? */
+            pR700AsmCode->last_cond_register = pILInst[i].DstReg.Index; 
+        }
+
         switch (pILInst[i].Opcode)
         {
         case OPCODE_ABS: 
@@ -3936,8 +5520,7 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
             break;  
 
         case OPCODE_ARL: 
-            radeon_error("Not yet implemented instruction OPCODE_ARL \n");
-            //if ( GL_FALSE == assemble_BAD("ARL") ) 
+            if ( GL_FALSE == assemble_ARL(pR700AsmCode) ) 
                 return GL_FALSE;
             break;
         case OPCODE_ARR: 
@@ -3951,7 +5534,7 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;  
         case OPCODE_COS: 
-            if ( GL_FALSE == assemble_COS(pR700AsmCode) ) 
+            if ( GL_FALSE == assemble_TRIG(pR700AsmCode, SQ_OP2_INST_COS) ) 
                 return GL_FALSE;
             break;  
 
@@ -3972,16 +5555,16 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;  
         case OPCODE_EXP: 
-            radeon_error("Not yet implemented instruction OPCODE_EXP \n");
-            //if ( GL_FALSE == assemble_BAD("EXP") ) 
+            if ( GL_FALSE == assemble_EXP(pR700AsmCode) ) 
                 return GL_FALSE;
-            break; // approx of EX2
+            break;
 
         case OPCODE_FLR:     
             if ( GL_FALSE == assemble_FLR(pR700AsmCode) ) 
                 return GL_FALSE;
             break;  
-        //case OP_FLR_INT: 
+        //case OP_FLR_INT: ;
+
         //    if ( GL_FALSE == assemble_FLR_INT() ) 
         //        return GL_FALSE;
         //    break;  
@@ -3992,7 +5575,8 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
             break;  
 
         case OPCODE_KIL: 
-            if ( GL_FALSE == assemble_KIL(pR700AsmCode) ) 
+        case OPCODE_KIL_NV: 
+            if ( GL_FALSE == assemble_KIL(pR700AsmCode, SQ_OP2_INST_KILLGT) ) 
                 return GL_FALSE;
             break;
         case OPCODE_LG2: 
@@ -4008,10 +5592,9 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;  
         case OPCODE_LOG: 
-            radeon_error("Not yet implemented instruction OPCODE_LOG \n");
-            //if ( GL_FALSE == assemble_BAD("LOG") ) 
+            if ( GL_FALSE == assemble_LOG(pR700AsmCode) ) 
                 return GL_FALSE;
-            break; // approx of LG2
+            break;
 
         case OPCODE_MAD: 
             if ( GL_FALSE == assemble_MAD(pR700AsmCode) ) 
@@ -4033,6 +5616,26 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
         case OPCODE_MUL: 
             if ( GL_FALSE == assemble_MUL(pR700AsmCode) ) 
                 return GL_FALSE;
+            break;
+            
+        case OPCODE_NOISE1:
+            {                                               
+                callPreSub(pR700AsmCode, 
+                           GLSL_NOISE1,                         
+                           &noise1_presub,                                                  
+                           pILInst->DstReg.Index + pR700AsmCode->starting_temp_register_number, 
+                           1); 
+                radeon_error("noise1: not yet supported shader instruction\n");
+            };
+            break; 
+        case OPCODE_NOISE2: 
+            radeon_error("noise2: not yet supported shader instruction\n");
+            break; 
+        case OPCODE_NOISE3: 
+            radeon_error("noise3: not yet supported shader instruction\n");
+            break; 
+        case OPCODE_NOISE4: 
+            radeon_error("noise4: not yet supported shader instruction\n");
             break; 
 
         case OPCODE_POW: 
@@ -4048,22 +5651,78 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;  
         case OPCODE_SIN: 
-            if ( GL_FALSE == assemble_SIN(pR700AsmCode) ) 
+            if ( GL_FALSE == assemble_TRIG(pR700AsmCode, SQ_OP2_INST_SIN) ) 
                 return GL_FALSE;
             break;  
         case OPCODE_SCS: 
             if ( GL_FALSE == assemble_SCS(pR700AsmCode) ) 
                 return GL_FALSE;
-            break;  
+            break; 
+            
+        case OPCODE_SEQ:
+            if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETE) ) 
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_SGT: 
+            if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGT) ) 
+            {
+                return GL_FALSE;
+            }
+            break;
 
         case OPCODE_SGE: 
             if ( GL_FALSE == assemble_SGE(pR700AsmCode) ) 
+            { 
                 return GL_FALSE;
-            break; 
+            }
+            break;
+        
+        /* NO LT, LE, TODO : use GE => LE, GT => LT : reverse 2 src order would be simpliest. Or use SQ_CF_COND_FALSE for SQ_CF_COND_ACTIVE.*/
         case OPCODE_SLT: 
-            if ( GL_FALSE == assemble_SLT(pR700AsmCode) ) 
+            {
+                struct prog_src_register SrcRegSave[2];
+                SrcRegSave[0] = pILInst[i].SrcReg[0];
+                SrcRegSave[1] = pILInst[i].SrcReg[1];
+                pILInst[i].SrcReg[0] = SrcRegSave[1];
+                pILInst[i].SrcReg[1] = SrcRegSave[0];
+                if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGT) ) 
+                {
+                    pILInst[i].SrcReg[0] = SrcRegSave[0];
+                    pILInst[i].SrcReg[1] = SrcRegSave[1];
+                    return GL_FALSE;
+                }
+                pILInst[i].SrcReg[0] = SrcRegSave[0];
+                pILInst[i].SrcReg[1] = SrcRegSave[1];
+            }
+            break;
+
+        case OPCODE_SLE: 
+            {
+                struct prog_src_register SrcRegSave[2];
+                SrcRegSave[0] = pILInst[i].SrcReg[0];
+                SrcRegSave[1] = pILInst[i].SrcReg[1];
+                pILInst[i].SrcReg[0] = SrcRegSave[1];
+                pILInst[i].SrcReg[1] = SrcRegSave[0];
+                if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETGE) ) 
+                {
+                    pILInst[i].SrcReg[0] = SrcRegSave[0];
+                    pILInst[i].SrcReg[1] = SrcRegSave[1];
+                    return GL_FALSE;
+                }
+                pILInst[i].SrcReg[0] = SrcRegSave[0];
+                pILInst[i].SrcReg[1] = SrcRegSave[1];
+            }
+            break;
+
+        case OPCODE_SNE: 
+            if ( GL_FALSE == assemble_LOGIC(pR700AsmCode, SQ_OP2_INST_SETNE) ) 
+            {
                 return GL_FALSE;
-            break; 
+            }
+            break;
 
         //case OP_STP: 
         //    if ( GL_FALSE == assemble_STP(pR700AsmCode) ) 
@@ -4089,7 +5748,8 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 }
             }
             break;
-
+        case OPCODE_DDX:
+        case OPCODE_DDY:
         case OPCODE_TEX: 
         case OPCODE_TXB:  
         case OPCODE_TXP: 
@@ -4097,30 +5757,104 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
                 return GL_FALSE;
             break;
 
+        case OPCODE_TRUNC:
+            if ( GL_FALSE == assemble_math_function(pR700AsmCode, SQ_OP2_INST_TRUNC) )
+                return GL_FALSE;
+            break;
+
         case OPCODE_XPD: 
             if ( GL_FALSE == assemble_XPD(pR700AsmCode) ) 
                 return GL_FALSE;
             break;  
 
-        case OPCODE_IF   : 
-            if ( GL_FALSE == assemble_IF(pR700AsmCode) ) 
-                return GL_FALSE;
+        case OPCODE_IF:
+            {                
+                GLboolean bHasElse = GL_FALSE;
+
+                if(pILInst[pILInst[i].BranchTarget].Opcode == OPCODE_ELSE)
+                {
+                    bHasElse = GL_TRUE;
+                }
+
+                if ( GL_FALSE == assemble_IF(pR700AsmCode, bHasElse) ) 
+                {
+                    return GL_FALSE;
+                }
+            }
             break;
+
         case OPCODE_ELSE : 
-            radeon_error("Not yet implemented instruction OPCODE_ELSE \n");
-            //if ( GL_FALSE == assemble_BAD("ELSE") ) 
+            if ( GL_FALSE == assemble_ELSE(pR700AsmCode) ) 
                 return GL_FALSE;
             break;
+
         case OPCODE_ENDIF: 
             if ( GL_FALSE == assemble_ENDIF(pR700AsmCode) ) 
                 return GL_FALSE;
             break;
 
+        case OPCODE_BGNLOOP:
+            if( GL_FALSE == assemble_BGNLOOP(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_BRK:
+            if( GL_FALSE == assemble_BRK(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_CONT:
+            if( GL_FALSE == assemble_CONT(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_ENDLOOP:
+            if( GL_FALSE == assemble_ENDLOOP(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
+        case OPCODE_BGNSUB:
+            if( GL_FALSE == assemble_BGNSUB(pR700AsmCode, i, uiIL_Shift) )
+            {
+                return GL_FALSE;
+            }
+            break;
+        
+        case OPCODE_RET:
+            if( GL_FALSE == assemble_RET(pR700AsmCode) )
+            {
+                return GL_FALSE;
+            }
+            break;
+        
+        case OPCODE_CAL:
+            if( GL_FALSE == assemble_CAL(pR700AsmCode, 
+                                         pILInst[i].BranchTarget,
+                                         uiIL_Shift,
+                                         uiNumberInsts,
+                                         pILInst,
+                                         NULL) )
+            {
+                return GL_FALSE;
+            }
+            break;
+
         //case OPCODE_EXPORT: 
         //    if ( GL_FALSE == assemble_EXPORT() ) 
         //        return GL_FALSE;
         //    break;
 
+        case OPCODE_ENDSUB:
+            return assemble_ENDSUB(pR700AsmCode);
+
         case OPCODE_END: 
 			//pR700AsmCode->uiCurInst = i;
 			//This is to remaind that if in later exoort there is depth/stencil
@@ -4137,6 +5871,417 @@ GLboolean AssembleInstr(GLuint uiNumberInsts,
     return GL_TRUE;
 }
 
+GLboolean InitShaderProgram(r700_AssemblerBase * pAsm)
+{
+    setRetInLoopFlag(pAsm, SQ_SEL_0);
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+    return GL_TRUE;
+}
+
+GLboolean RelocProgram(r700_AssemblerBase * pAsm, struct gl_program * pILProg)
+{
+    GLuint i;
+    GLuint unCFoffset;
+    TypedShaderList * plstCFmain;
+    TypedShaderList * plstCFsub;
+
+    R700ShaderInstruction *        pInst;
+    R700ControlFlowGenericClause * pCFInst;
+
+    R700ControlFlowALUClause * pCF_ALU;
+    R700ALUInstruction       * pALU;
+    GLuint                     unConstOffset = 0;
+    GLuint                     unRegOffset;
+    GLuint                     unMinRegIndex;
+
+    plstCFmain = pAsm->CALLSTACK[0].plstCFInstructions_local;
+
+    /* remove flags init if they are not used */
+    if((pAsm->unCFflags & HAS_LOOPRET) == 0)
+    {
+        R700ControlFlowALUClause * pCF_ALU;
+        pInst = plstCFmain->pHead;
+        while(pInst)
+        {
+            if(SIT_CF_ALU == pInst->m_ShaderInstType)
+            {
+                pCF_ALU = (R700ControlFlowALUClause *)pInst;
+                if(0 == pCF_ALU->m_Word1.f.count)
+                {
+                    pCF_ALU->m_Word1.f.cf_inst = SQ_CF_INST_NOP;
+                }
+                else
+                {
+                    R700ALUInstruction * pALU = pCF_ALU->m_pLinkedALUInstruction;
+                    
+                    pALU->m_pLinkedALUClause = NULL;
+                    pALU = (R700ALUInstruction *)(pALU->pNextInst);
+                    pALU->m_pLinkedALUClause = pCF_ALU;
+                    pCF_ALU->m_pLinkedALUInstruction = pALU;
+
+                    pCF_ALU->m_Word1.f.count--;
+                }
+                break;
+            }
+            pInst = pInst->pNextInst;
+        };
+    }
+
+    if(pAsm->CALLSTACK[0].max > 0)
+    {
+        pAsm->pR700Shader->uStackSize = ((pAsm->CALLSTACK[0].max + 3)>>2) + 2;
+    }
+
+    if(0 == pAsm->unSubArrayPointer)
+    {
+        return GL_TRUE;
+    }
+
+    unCFoffset = plstCFmain->uNumOfNode;
+
+    if(NULL != pILProg->Parameters)
+    {        
+        unConstOffset = pILProg->Parameters->NumParameters;
+    }
+
+    /* Reloc subs */
+    for(i=0; i<pAsm->unSubArrayPointer; i++)
+    {
+        pAsm->subs[i].unCFoffset = unCFoffset;
+        plstCFsub = &(pAsm->subs[i].lstCFInstructions_local);
+
+        pInst = plstCFsub->pHead;
+
+        /* reloc instructions */
+        while(pInst)
+        {
+            if(SIT_CF_GENERIC == pInst->m_ShaderInstType)
+            {
+                pCFInst = (R700ControlFlowGenericClause *)pInst;
+
+                switch (pCFInst->m_Word1.f.cf_inst)
+                {
+                case SQ_CF_INST_POP:
+                case SQ_CF_INST_JUMP:
+                case SQ_CF_INST_ELSE:
+                case SQ_CF_INST_LOOP_END:
+                case SQ_CF_INST_LOOP_START:
+                case SQ_CF_INST_LOOP_START_NO_AL:
+                case SQ_CF_INST_LOOP_CONTINUE:
+                case SQ_CF_INST_LOOP_BREAK:
+                    pCFInst->m_Word0.f.addr += unCFoffset;
+                    break;
+                default:
+                    break;
+                }
+            }  
+            
+            pInst->m_uIndex += unCFoffset;
+
+            pInst = pInst->pNextInst;
+        };
+
+        if(NULL != pAsm->subs[i].pPresubDesc)
+        {
+            GLuint                     uNumSrc;            
+            
+            unMinRegIndex  = pAsm->subs[i].pPresubDesc->pCompiledSub->MinRegIndex;
+            unRegOffset    = pAsm->subs[i].pPresubDesc->maxStartReg;            
+            unConstOffset += pAsm->subs[i].pPresubDesc->unConstantsStart;
+
+            pInst = plstCFsub->pHead;
+            while(pInst)
+            {
+                if(SIT_CF_ALU == pInst->m_ShaderInstType)
+                {
+                    pCF_ALU = (R700ControlFlowALUClause *)pInst;
+
+                    pALU = pCF_ALU->m_pLinkedALUInstruction;
+                    for(int j=0; j<=pCF_ALU->m_Word1.f.count; j++)
+                    {
+                        pALU->m_Word1.f.dst_gpr = pALU->m_Word1.f.dst_gpr + unRegOffset - unMinRegIndex;
+
+                        if(pALU->m_Word0.f.src0_sel < SQ_ALU_SRC_GPR_SIZE)
+                        {   
+                            pALU->m_Word0.f.src0_sel = pALU->m_Word0.f.src0_sel + unRegOffset - unMinRegIndex;
+                        }
+                        else if(pALU->m_Word0.f.src0_sel >= SQ_ALU_SRC_CFILE_BASE)
+                        {   
+                            pALU->m_Word0.f.src0_sel += unConstOffset;
+                        }
+
+                        if( ((pALU->m_Word1.val >> SQ_ALU_WORD1_OP3_ALU_INST_SHIFT) & 0x0000001F) 
+                            >= SQ_OP3_INST_MUL_LIT )
+                        {   /* op3 : 3 srcs */
+                            if(pALU->m_Word1_OP3.f.src2_sel < SQ_ALU_SRC_GPR_SIZE)
+                            {   
+                                pALU->m_Word1_OP3.f.src2_sel = pALU->m_Word1_OP3.f.src2_sel + unRegOffset - unMinRegIndex;
+                            }
+                            else if(pALU->m_Word1_OP3.f.src2_sel >= SQ_ALU_SRC_CFILE_BASE)
+                            {   
+                                pALU->m_Word1_OP3.f.src2_sel += unConstOffset;
+                            }    
+                            if(pALU->m_Word0.f.src1_sel < SQ_ALU_SRC_GPR_SIZE)
+                            {   
+                                pALU->m_Word0.f.src1_sel = pALU->m_Word0.f.src1_sel + unRegOffset - unMinRegIndex;
+                            }
+                            else if(pALU->m_Word0.f.src1_sel >= SQ_ALU_SRC_CFILE_BASE)
+                            {   
+                                pALU->m_Word0.f.src1_sel += unConstOffset;
+                            }                                 
+                        }
+                        else
+                        {
+                            if(pAsm->bR6xx)
+                            {
+                                uNumSrc = r700GetNumOperands(pALU->m_Word1_OP2.f6.alu_inst, 0);
+                            }
+                            else
+                            {
+                                uNumSrc = r700GetNumOperands(pALU->m_Word1_OP2.f.alu_inst, 0);
+                            }
+                            if(2 == uNumSrc)
+                            {   /* 2 srcs */
+                                if(pALU->m_Word0.f.src1_sel < SQ_ALU_SRC_GPR_SIZE)
+                                {   
+                                    pALU->m_Word0.f.src1_sel = pALU->m_Word0.f.src1_sel + unRegOffset - unMinRegIndex;
+                                }
+                                else if(pALU->m_Word0.f.src1_sel >= SQ_ALU_SRC_CFILE_BASE)
+                                {   
+                                    pALU->m_Word0.f.src1_sel += unConstOffset;
+                                }                                  
+                            }                            
+                        }
+                        pALU = (R700ALUInstruction*)(pALU->pNextInst);
+                    }                    
+                }             
+                pInst = pInst->pNextInst;
+            };
+        }
+
+        /* Put sub into main */
+        plstCFmain->pTail->pNextInst = plstCFsub->pHead;
+        plstCFmain->pTail            = plstCFsub->pTail;
+        plstCFmain->uNumOfNode      += plstCFsub->uNumOfNode;
+
+        unCFoffset += plstCFsub->uNumOfNode;
+    }
+
+    /* reloc callers */
+    for(i=0; i<pAsm->unCallerArrayPointer; i++)
+    {
+        pAsm->callers[i].cf_ptr->m_Word0.f.addr
+            = pAsm->subs[pAsm->callers[i].subDescIndex].unCFoffset; 
+
+        if(NULL != pAsm->subs[pAsm->callers[i].subDescIndex].pPresubDesc)
+        {                 
+            unMinRegIndex = pAsm->subs[pAsm->callers[i].subDescIndex].pPresubDesc->pCompiledSub->MinRegIndex;
+            unRegOffset = pAsm->subs[pAsm->callers[i].subDescIndex].pPresubDesc->maxStartReg;
+
+            if(NULL != pAsm->callers[i].prelude_cf_ptr)
+            {                
+                pCF_ALU = (R700ControlFlowALUClause * )(pAsm->callers[i].prelude_cf_ptr);
+                pALU = pCF_ALU->m_pLinkedALUInstruction;
+                for(int j=0; j<=pCF_ALU->m_Word1.f.count; j++)
+                {
+                    pALU->m_Word1.f.dst_gpr = pALU->m_Word1.f.dst_gpr + unRegOffset - unMinRegIndex;
+                    pALU = (R700ALUInstruction*)(pALU->pNextInst);
+                }
+            }
+            if(NULL != pAsm->callers[i].finale_cf_ptr)
+            {
+                pCF_ALU = (R700ControlFlowALUClause * )(pAsm->callers[i].finale_cf_ptr);
+                pALU = pCF_ALU->m_pLinkedALUInstruction;
+                for(int j=0; j<=pCF_ALU->m_Word1.f.count; j++)
+                {
+                    pALU->m_Word0.f.src0_sel = pALU->m_Word0.f.src0_sel + unRegOffset - unMinRegIndex;
+                    pALU = (R700ALUInstruction*)(pALU->pNextInst);
+                }
+            }
+        }
+    }
+
+    return GL_TRUE;
+}
+
+GLboolean callPreSub(r700_AssemblerBase* pAsm, 
+                         LOADABLE_SCRIPT_SIGNITURE scriptSigniture,                          
+                         COMPILED_SUB * pCompiledSub,                                               
+                         GLshort uOutReg,
+                         GLshort uNumValidSrc)
+{
+    /* save assemble context */
+    GLuint starting_temp_register_number_save;
+    GLuint number_used_registers_save;
+    GLuint uFirstHelpReg_save;
+    GLuint uHelpReg_save;
+    GLuint uiCurInst_save;
+    struct prog_instruction *pILInst_save;
+    PRESUB_DESC * pPresubDesc;
+    GLboolean     bRet;
+    int i;
+
+    R700ControlFlowGenericClause* prelude_cf_ptr = NULL;
+
+    /* copy srcs to presub inputs */  
+    pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+    for(i=0; i<uNumValidSrc; i++)
+    {
+        pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+        setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+        pAsm->D.dst.rtype = DST_REG_TEMPORARY;
+        pAsm->D.dst.reg   = pCompiledSub->srcRegIndex[i];
+        pAsm->D.dst.writex = 1;
+        pAsm->D.dst.writey = 1;
+        pAsm->D.dst.writez = 1;
+        pAsm->D.dst.writew = 1;
+
+        if( GL_FALSE == assemble_src(pAsm, i, 0) )
+        {
+            return GL_FALSE;
+        }
+
+        next_ins(pAsm);
+    }
+    if(uNumValidSrc > 0)
+    {
+        prelude_cf_ptr     = pAsm->cf_current_alu_clause_ptr;
+        pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+    }
+
+    /* browse thro existing presubs. */
+    for(i=0; i<pAsm->unNumPresub; i++)
+    {
+        if(pAsm->presubs[i].sptSigniture == scriptSigniture)
+        {
+            break;
+        }
+    }
+
+    if(i == pAsm->unNumPresub)
+    {   /* not loaded yet */
+        /* save assemble context */
+        number_used_registers_save         = pAsm->number_used_registers;
+        uFirstHelpReg_save                 = pAsm->uFirstHelpReg;
+        uHelpReg_save                      = pAsm->uHelpReg;
+        starting_temp_register_number_save = pAsm->starting_temp_register_number;
+        pILInst_save                       = pAsm->pILInst;
+        uiCurInst_save                     = pAsm->uiCurInst;
+
+        /* alloc in presub */
+        if( (pAsm->unNumPresub + 1) > pAsm->unPresubArraySize )
+        {
+            pAsm->presubs = (PRESUB_DESC*)_mesa_realloc( (void *)pAsm->presubs,
+                                      sizeof(PRESUB_DESC) * pAsm->unPresubArraySize,
+                                      sizeof(PRESUB_DESC) * (pAsm->unPresubArraySize + 4) );
+            if(NULL == pAsm->presubs)
+            {
+                radeon_error("No memeory to allocate built in shader function description structures. \n");
+                return GL_FALSE;
+            }
+            pAsm->unPresubArraySize += 4;
+        }
+        
+        pPresubDesc = &(pAsm->presubs[i]);
+        pPresubDesc->sptSigniture = scriptSigniture;
+
+        /* constants offsets need to be final resolved at reloc. */
+        if(0 == pAsm->unNumPresub)
+        {
+            pPresubDesc->unConstantsStart = 0; 
+        }
+        else
+        {
+            pPresubDesc->unConstantsStart =  pAsm->presubs[i-1].unConstantsStart
+                                           + pAsm->presubs[i-1].pCompiledSub->NumParameters;
+        }
+
+        pPresubDesc->pCompiledSub = pCompiledSub;
+
+        pPresubDesc->subIL_Shift = pAsm->unCurNumILInsts;
+        pPresubDesc->maxStartReg  = uFirstHelpReg_save;
+        pAsm->unCurNumILInsts    += pCompiledSub->NumInstructions;
+
+        pAsm->unNumPresub++;
+
+        /* setup new assemble context */
+        pAsm->starting_temp_register_number = 0;
+        pAsm->number_used_registers = pCompiledSub->NumTemporaries;
+        pAsm->uFirstHelpReg         = pAsm->number_used_registers;
+        pAsm->uHelpReg              = pAsm->uFirstHelpReg;
+
+        bRet = assemble_CAL(pAsm, 
+                            0, 
+                            pPresubDesc->subIL_Shift, 
+                            pCompiledSub->NumInstructions,
+                            pCompiledSub->Instructions,
+                            pPresubDesc);
+
+        
+        pPresubDesc->number_used_registers = pAsm->number_used_registers;        
+
+        /* restore assemble context */
+        pAsm->number_used_registers         = number_used_registers_save; 
+        pAsm->uFirstHelpReg                 = uFirstHelpReg_save;
+        pAsm->uHelpReg                      = uHelpReg_save;
+        pAsm->starting_temp_register_number = starting_temp_register_number_save;
+        pAsm->pILInst                       = pILInst_save; 
+        pAsm->uiCurInst                     = uiCurInst_save;
+    }
+    else
+    {   /* was loaded */
+        pPresubDesc = &(pAsm->presubs[i]);  
+        
+        bRet = assemble_CAL(pAsm, 
+                            0, 
+                            pPresubDesc->subIL_Shift, 
+                            pCompiledSub->NumInstructions,
+                            pCompiledSub->Instructions,
+                            pPresubDesc);
+    }
+
+    if(GL_FALSE == bRet)
+    {
+        radeon_error("Shader presub assemble failed. \n");
+    }
+    else
+    {
+        /* copy presub output to real dst */ 
+        pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+        pAsm->D.dst.opcode = SQ_OP2_INST_MOV;
+
+        if( GL_FALSE == assemble_dst(pAsm) )
+        {
+            return GL_FALSE;
+        }
+
+        setaddrmode_PVSSRC(&(pAsm->S[0].src), ADDR_ABSOLUTE);
+        pAsm->S[0].src.rtype = SRC_REG_TEMPORARY;
+        pAsm->S[0].src.reg   = pCompiledSub->dstRegIndex;
+        pAsm->S[0].src.swizzlex = pCompiledSub->outputSwizzleX;
+        pAsm->S[0].src.swizzley = pCompiledSub->outputSwizzleY;
+        pAsm->S[0].src.swizzlez = pCompiledSub->outputSwizzleZ;
+        pAsm->S[0].src.swizzlew = pCompiledSub->outputSwizzleW;
+
+        next_ins(pAsm);        
+
+        pAsm->callers[pAsm->unCallerArrayPointer - 1].finale_cf_ptr  = pAsm->cf_current_alu_clause_ptr;
+        pAsm->callers[pAsm->unCallerArrayPointer - 1].prelude_cf_ptr = prelude_cf_ptr;
+        pAsm->alu_x_opcode = SQ_CF_INST_ALU;
+    }
+
+    if( (pPresubDesc->number_used_registers + pAsm->uFirstHelpReg) > pAsm->number_used_registers )
+    {
+        pAsm->number_used_registers = pPresubDesc->number_used_registers + pAsm->uFirstHelpReg;
+    }
+    if(pAsm->uFirstHelpReg > pPresubDesc->maxStartReg)
+    {
+        pPresubDesc->maxStartReg = pAsm->uFirstHelpReg;
+    }
+
+    return bRet;
+}
+
 GLboolean Process_Export(r700_AssemblerBase* pAsm,
                          GLuint type,
                          GLuint export_starting_index,
@@ -4282,6 +6427,7 @@ GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode,
                                    GLbitfield          OutputsWritten)  
 { 
     unsigned int unBit;
+    GLuint export_count = 0;
 
     if(pR700AsmCode->depth_export_register_number >= 0) 
     {
@@ -4303,6 +6449,7 @@ GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode,
         {
             return GL_FALSE;
         }
+        export_count++;
 	}
 	unBit = 1 << FRAG_RESULT_DEPTH;
 	if(OutputsWritten & unBit)
@@ -4316,8 +6463,15 @@ GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode,
         {
             return GL_FALSE;
         }
+        export_count++;
 	}
-
+    /* Need to export something, otherwise we'll hang
+     * results are undefined anyway */
+    if(export_count == 0)
+    {
+        Process_Export(pR700AsmCode, SQ_EXPORT_PIXEL, 0, 1, 0, GL_FALSE);
+    }
+    
     if(pR700AsmCode->cf_last_export_ptr != NULL) 
     {
         pR700AsmCode->cf_last_export_ptr->m_Word1.f.cf_inst        = SQ_CF_INST_EXPORT_DONE;
@@ -4422,6 +6576,25 @@ GLboolean Process_Vertex_Exports(r700_AssemblerBase *pR700AsmCode,
             export_starting_index++;
 		}
 	}
+    
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+        unBit = 1 << i;
+        if(OutputsWritten & unBit)
+		{
+            if( GL_FALSE == Process_Export(pR700AsmCode,
+                                          SQ_EXPORT_PARAM, 
+                                          export_starting_index, 
+                                          1, 
+                                          pR700AsmCode->ucVP_OutputMap[i],
+                                          GL_FALSE) )
+            {                
+                return GL_FALSE;
+            }
+
+            export_starting_index++;
+		}
+    }
 
     // At least one param should be exported
     if (export_count) 
@@ -4456,6 +6629,21 @@ GLboolean Clean_Up_Assembler(r700_AssemblerBase *pR700AsmCode)
 {
     FREE(pR700AsmCode->pucOutMask);
     FREE(pR700AsmCode->pInstDeps);
+
+    if(NULL != pR700AsmCode->subs)
+    {
+        FREE(pR700AsmCode->subs);
+    }
+    if(NULL != pR700AsmCode->callers)
+    {
+        FREE(pR700AsmCode->callers);
+    }
+
+    if(NULL != pR700AsmCode->presubs)
+    {
+        FREE(pR700AsmCode->presubs);
+    }
+
     return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.h b/src/mesa/drivers/dri/r600/r700_assembler.h
index 0d4283e4ba..56baf5b0d9 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.h
+++ b/src/mesa/drivers/dri/r600/r700_assembler.h
@@ -34,6 +34,45 @@
 #include "r700_shaderinst.h"
 #include "r700_shader.h"
 
+typedef enum LOADABLE_SCRIPT_SIGNITURE
+{
+    GLSL_NOISE1 = 0x10000001,
+    GLSL_NOISE2 = 0x10000002,
+    GLSL_NOISE3 = 0x10000003,
+    GLSL_NOISE4 = 0x10000004
+}LOADABLE_SCRIPT_SIGNITURE;
+
+typedef struct COMPILED_SUB
+{
+    struct  prog_instruction *Instructions;
+    GLuint  NumInstructions;
+    GLuint  NumTemporaries;
+    GLuint  NumParameters;
+    GLuint  MinRegIndex;
+    GLfloat (*ParameterValues)[4];
+    GLbyte  outputSwizzleX; 
+    GLbyte  outputSwizzleY;
+    GLbyte  outputSwizzleZ;
+    GLbyte  outputSwizzleW;
+    GLshort srcRegIndex[3];
+    GLushort dstRegIndex;
+}COMPILED_SUB;
+
+typedef struct PRESUB_DESCtag 
+{
+    LOADABLE_SCRIPT_SIGNITURE sptSigniture;
+    GLint  subIL_Shift;
+    struct prog_src_register InReg[3];
+    struct prog_dst_register OutReg;
+
+    GLushort maxStartReg;
+    GLushort number_used_registers;
+
+    GLuint   unConstantsStart;
+
+    COMPILED_SUB * pCompiledSub;
+} PRESUB_DESC;
+
 typedef enum SHADER_PIPE_TYPE 
 {
     SPT_VP = 0,
@@ -72,7 +111,8 @@ typedef enum SrcRegisterType
     SRC_REG_INPUT          = 1,
     SRC_REG_CONSTANT       = 2,
     SRC_REG_ALT_TEMPORARY  = 3,
-    NUMBER_OF_SRC_REG_TYPE = 4
+    SRC_REC_LITERAL        = 4, 
+    NUMBER_OF_SRC_REG_TYPE = 5
 } SrcRegisterType;
 
 typedef enum DstRegisterType 
@@ -111,16 +151,24 @@ typedef struct PVSDSTtag
 	BITS addrmode1:1; //32
 } PVSDST;
 
+typedef struct PVSINSTtag
+{
+    BITS literal_slots      :2; 
+    BITS SaturateMode :2; 
+    BITS index_mode   :3;
+} PVSINST;
+
 typedef struct PVSSRCtag 
 {
-	BITS rtype:4;            
+	BITS rtype:3;            
 	BITS addrmode0:1;        
-	BITS reg:10;      //15     (8)
+	BITS reg:10;      //14     (8)
 	BITS swizzlex:3;
 	BITS swizzley:3;
 	BITS swizzlez:3;
-	BITS swizzlew:3;  //27        
+	BITS swizzlew:3;  //26        
 
+	BITS abs:1;
 	BITS negx:1;
 	BITS negy:1;
 	BITS negz:1;
@@ -148,6 +196,7 @@ typedef union PVSDWORDtag
 {
 	BITS    bits;
 	PVSDST  dst;
+    PVSINST dst2;
 	PVSSRC  src;
 	PVSMATH math;
 	float   f;
@@ -251,6 +300,8 @@ enum
     FC_IF = 1,
     FC_LOOP = 2,
     FC_REP = 3,
+    FC_PUSH_VPM = 4,
+    FC_PUSH_WQM = 5,
 
     COND_NONE = 0,
     COND_BOOL = 1,
@@ -263,22 +314,56 @@ enum
 
 typedef struct FC_LEVEL 
 {
-	unsigned int           first; ///< first fc instruction on level (if, rep, loop)
-	unsigned int*          mid; ///< middle instructions - else or all breaks on this level
-	unsigned int           midLen;
-	unsigned int           type;
-	unsigned int           cond;
-	unsigned int           inv;
-	unsigned int           bpush; ///< 1 if first instruction does branch stack push
-			 int           id; ///< id of bool or int variable
+    R700ControlFlowGenericClause *  first;
+    R700ControlFlowGenericClause ** mid;
+    unsigned int unNumMid;
+    unsigned int midLen;
+    unsigned int type;
+    unsigned int cond;
+    unsigned int inv;
+    int id; ///< id of bool or int variable
 } FC_LEVEL;
 
 typedef struct VTX_FETCH_METHOD 
 {
-	GLboolean bEnableMini;
-	GLuint mega_fetch_remainder;
+    GLboolean bEnableMini;
+    GLuint mega_fetch_remainder;
 } VTX_FETCH_METHOD;
 
+typedef struct SUB_OFFSET
+{
+    GLint  subIL_Offset;
+    GLuint unCFoffset;
+    GLuint unStackDepthMax;
+    PRESUB_DESC *   pPresubDesc;
+    TypedShaderList lstCFInstructions_local;
+} SUB_OFFSET;
+
+typedef struct CALLER_POINTER
+{
+    GLint  subIL_Offset;
+    GLint  subDescIndex;
+    R700ControlFlowGenericClause* cf_ptr;
+
+    R700ControlFlowGenericClause* prelude_cf_ptr;
+    R700ControlFlowGenericClause* finale_cf_ptr;
+} CALLER_POINTER;
+
+#define SQ_MAX_CALL_DEPTH 0x00000020
+
+typedef struct CALL_LEVEL
+{
+    unsigned int      FCSP_BeforeEntry;
+    GLint             subDescIndex;
+    GLushort          current;
+    GLushort          max;
+    TypedShaderList * plstCFInstructions_local;
+} CALL_LEVEL;
+
+#define HAS_CURRENT_LOOPRET 0x1L
+#define HAS_LOOPRET         0x2L
+#define LOOPRET_FLAGS       HAS_LOOPRET | HAS_CURRENT_LOOPRET
+
 typedef struct r700_AssemblerBase 
 {
 	R700ControlFlowSXClause*      cf_last_export_ptr;
@@ -294,14 +379,19 @@ typedef struct r700_AssemblerBase
 	// No clause has been created yet
 	CF_CLAUSE_TYPE cf_current_clause_type;
 
+    BITS alu_x_opcode;
+
 	GLuint number_of_exports;
 	GLuint number_of_colorandz_exports;
 	GLuint number_of_export_opcodes;
 
 	PVSDWORD D;
+    PVSDWORD D2;
 	PVSDWORD S[3];
+        PVSDWORD C[4];
 
 	unsigned int uLastPosUpdate;
+	unsigned int last_cond_register;
 
 	OUT_FRAGMENT_FMT_0     fp_stOutFmt0;
 
@@ -310,6 +400,8 @@ typedef struct r700_AssemblerBase
 	unsigned int number_used_registers;
 	unsigned int uUsedConsts; 
 
+    unsigned int flag_reg_index;
+
 	// Fragment programs
 	unsigned int uiFP_AttributeMap[FRAG_ATTRIB_MAX];
 	unsigned int uiFP_OutputMap[FRAG_RESULT_MAX];
@@ -330,9 +422,6 @@ typedef struct r700_AssemblerBase
 	unsigned int FCSP;
 	FC_LEVEL fc_stack[32];
 
-	unsigned int branch_depth;
-	unsigned int max_branch_depth;
-
 	//-----------------------------------------------------------------------------------
 	// ArgSubst used in Assemble_Source() function
 	//-----------------------------------------------------------------------------------
@@ -373,11 +462,33 @@ typedef struct r700_AssemblerBase
     SHADER_PIPE_TYPE currentShaderType;
     struct prog_instruction * pILInst;
     GLuint             uiCurInst;
+    GLubyte SamplerUnits[MAX_SAMPLERS];
     GLboolean   bR6xx;
     /* helper to decide which type of instruction to assemble */
     GLboolean is_tex;
     /* we inserted helper intructions and need barrier on next TEX ins */ 
     GLboolean need_tex_barrier; 
+
+    SUB_OFFSET     * subs;
+    GLuint           unSubArraySize;
+    GLuint           unSubArrayPointer;
+    CALLER_POINTER * callers;
+    GLuint           unCallerArraySize;
+    GLuint           unCallerArrayPointer;
+    unsigned int     CALLSP;
+    CALL_LEVEL       CALLSTACK[SQ_MAX_CALL_DEPTH];
+
+    GLuint unCFflags;
+
+    PRESUB_DESC * presubs;
+    GLuint        unPresubArraySize;
+    GLuint        unNumPresub;
+    GLuint        unCurNumILInsts;
+
+    GLuint    unVetTexBits;
+
+    GLuint    shadow_regs[R700_MAX_TEXTURE_UNITS];
+
 } r700_AssemblerBase;
 
 //Internal use
@@ -399,7 +510,7 @@ BITS is_depth_component_exported(OUT_FRAGMENT_FMT_0* pFPOutFmt) ;
 GLboolean is_reduction_opcode(PVSDWORD * dest);
 GLuint GetSurfaceFormat(GLenum eType, GLuint nChannels, GLuint * pClient_size);
 
-unsigned int r700GetNumOperands(r700_AssemblerBase* pAsm);
+unsigned int r700GetNumOperands(GLuint opcode, GLuint nIsOp3);
 
 GLboolean IsTex(gl_inst_opcode Opcode);
 GLboolean IsAlu(gl_inst_opcode Opcode);
@@ -422,6 +533,7 @@ GLboolean assemble_vfetch_instruction2(r700_AssemblerBase* pAsm,
                                        GLubyte             element,
                                        GLuint              _signed,
                                        GLboolean           normalize,
+                                       GLenum              format,
                                        VTX_FETCH_METHOD  * pFetchMethod);
 GLboolean cleanup_vfetch_instructions(r700_AssemblerBase* pAsm);
 GLuint gethelpr(r700_AssemblerBase* pAsm);
@@ -446,6 +558,10 @@ GLboolean assemble_alu_src(R700ALUInstruction*  alu_instruction_ptr,
 GLboolean add_alu_instruction(r700_AssemblerBase* pAsm,
                               R700ALUInstruction* alu_instruction_ptr,
                               GLuint              contiguous_slots_needed);
+
+GLboolean add_cf_instruction(r700_AssemblerBase* pAsm);
+void add_return_inst(r700_AssemblerBase *pAsm);
+
 void get_src_properties(R700ALUInstruction*  alu_instruction_ptr,
                         int                  source_index,
                         BITS*                psrc_sel,
@@ -467,21 +583,31 @@ GLboolean check_vector(r700_AssemblerBase* pAsm,
                        R700ALUInstruction* alu_instruction_ptr);
 GLboolean assemble_alu_instruction(r700_AssemblerBase *pAsm);
 GLboolean next_ins(r700_AssemblerBase *pAsm);
+
+GLboolean pops(r700_AssemblerBase *pAsm, GLuint pops);
+GLboolean jumpToOffest(r700_AssemblerBase *pAsm, GLuint pops, GLint offset);
+GLboolean setRetInLoopFlag(r700_AssemblerBase *pAsm, GLuint flagValue);
+GLboolean testFlag(r700_AssemblerBase *pAsm);
+GLboolean breakLoopOnFlag(r700_AssemblerBase *pAsm, GLuint unFCSP);
+GLboolean returnOnFlag(r700_AssemblerBase *pAsm, GLuint unIF);
+
 GLboolean assemble_math_function(r700_AssemblerBase* pAsm, BITS opcode);
 GLboolean assemble_ABS(r700_AssemblerBase *pAsm);
 GLboolean assemble_ADD(r700_AssemblerBase *pAsm);
+GLboolean assemble_ARL(r700_AssemblerBase *pAsm);
 GLboolean assemble_BAD(char *opcode_str);
 GLboolean assemble_CMP(r700_AssemblerBase *pAsm);
-GLboolean assemble_COS(r700_AssemblerBase *pAsm);
 GLboolean assemble_DOT(r700_AssemblerBase *pAsm);
 GLboolean assemble_DST(r700_AssemblerBase *pAsm);
 GLboolean assemble_EX2(r700_AssemblerBase *pAsm);
+GLboolean assemble_EXP(r700_AssemblerBase *pAsm);
 GLboolean assemble_FLR(r700_AssemblerBase *pAsm);
 GLboolean assemble_FLR_INT(r700_AssemblerBase *pAsm);
 GLboolean assemble_FRC(r700_AssemblerBase *pAsm);
-GLboolean assemble_KIL(r700_AssemblerBase *pAsm);
+GLboolean assemble_KIL(r700_AssemblerBase *pAsm, GLuint opcode);
 GLboolean assemble_LG2(r700_AssemblerBase *pAsm);
 GLboolean assemble_LRP(r700_AssemblerBase *pAsm);
+GLboolean assemble_LOG(r700_AssemblerBase *pAsm);
 GLboolean assemble_MAD(r700_AssemblerBase *pAsm);
 GLboolean assemble_LIT(r700_AssemblerBase *pAsm);
 GLboolean assemble_MAX(r700_AssemblerBase *pAsm);
@@ -491,17 +617,37 @@ GLboolean assemble_MUL(r700_AssemblerBase *pAsm);
 GLboolean assemble_POW(r700_AssemblerBase *pAsm);
 GLboolean assemble_RCP(r700_AssemblerBase *pAsm);
 GLboolean assemble_RSQ(r700_AssemblerBase *pAsm);
-GLboolean assemble_SIN(r700_AssemblerBase *pAsm);
 GLboolean assemble_SCS(r700_AssemblerBase *pAsm);
 GLboolean assemble_SGE(r700_AssemblerBase *pAsm);
+
+GLboolean assemble_LOGIC(r700_AssemblerBase *pAsm, BITS opcode);
+GLboolean assemble_LOGIC_PRED(r700_AssemblerBase *pAsm, BITS opcode); 
+GLboolean assemble_TRIG(r700_AssemblerBase *pAsm, BITS opcode);
+
 GLboolean assemble_SLT(r700_AssemblerBase *pAsm);
 GLboolean assemble_STP(r700_AssemblerBase *pAsm);
 GLboolean assemble_TEX(r700_AssemblerBase *pAsm);
 GLboolean assemble_XPD(r700_AssemblerBase *pAsm);
 GLboolean assemble_EXPORT(r700_AssemblerBase *pAsm);
-GLboolean assemble_IF(r700_AssemblerBase *pAsm);
+GLboolean assemble_IF(r700_AssemblerBase *pAsm, GLboolean bHasElse);
+GLboolean assemble_ELSE(r700_AssemblerBase *pAsm);
 GLboolean assemble_ENDIF(r700_AssemblerBase *pAsm);
 
+GLboolean assemble_BGNLOOP(r700_AssemblerBase *pAsm);
+GLboolean assemble_BRK(r700_AssemblerBase *pAsm);
+GLboolean assemble_COND(r700_AssemblerBase *pAsm);
+GLboolean assemble_ENDLOOP(r700_AssemblerBase *pAsm);
+
+GLboolean assemble_BGNSUB(r700_AssemblerBase *pAsm, GLint nILindex, GLuint uiIL_Shift);
+GLboolean assemble_ENDSUB(r700_AssemblerBase *pAsm);
+GLboolean assemble_RET(r700_AssemblerBase *pAsm);
+GLboolean assemble_CAL(r700_AssemblerBase *pAsm, 
+                       GLint nILindex,
+                       GLuint uiIL_Offest,
+                       GLuint uiNumberInsts,
+                       struct prog_instruction *pILInst,
+                       PRESUB_DESC * pPresubDesc);
+
 GLboolean Process_Export(r700_AssemblerBase* pAsm,
                          GLuint type, 
                          GLuint export_starting_index,
@@ -511,14 +657,25 @@ GLboolean Process_Export(r700_AssemblerBase* pAsm,
 GLboolean Move_Depth_Exports_To_Correct_Channels(r700_AssemblerBase *pAsm, 
                                                  BITS depth_channel_select);
 
+GLboolean callPreSub(r700_AssemblerBase* pAsm, 
+                     LOADABLE_SCRIPT_SIGNITURE scriptSigniture,
+                     /* struct prog_instruction ** pILInstParent, */
+                     COMPILED_SUB * pCompiledSub,                                            
+                     GLshort uOutReg,
+                     GLshort uNumValidSrc);
 
 //Interface
-GLboolean AssembleInstr(GLuint uiNumberInsts,
+GLboolean AssembleInstr(GLuint uiFirstInst,
+                        GLuint uiIL_Shift,
+                        GLuint uiNumberInsts,
                         struct prog_instruction *pILInst, 
 						r700_AssemblerBase *pR700AsmCode);
 GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode, GLbitfield OutputsWritten);  
 GLboolean Process_Vertex_Exports(r700_AssemblerBase *pR700AsmCode, GLbitfield OutputsWritten);
 
+GLboolean RelocProgram(r700_AssemblerBase * pAsm, struct gl_program * pILProg);
+GLboolean InitShaderProgram(r700_AssemblerBase * pAsm);
+
 int       Init_r700_AssemblerBase(SHADER_PIPE_TYPE spt, r700_AssemblerBase* pAsm, R700_Shader* pShader);
 GLboolean Clean_Up_Assembler(r700_AssemblerBase *pR700AsmCode);
 
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index 3b7f6fffe0..3bc2d2ba02 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -45,6 +45,9 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	context_t         *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+
+    struct r700_vertex_program *vp = context->selected_vp;
+
 	struct radeon_bo *bo = NULL;
 	unsigned int i;
 	BATCH_LOCALS(&context->radeon);
@@ -52,13 +55,14 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
 	for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
-		if (ctx->Texture.Unit[i]._ReallyEnabled) {
+		if (ctx->Texture.Unit[i]._ReallyEnabled) {            
 			radeonTexObj *t = r700->textures[i];
 			if (t) {
-				if (!t->image_override)
+				if (!t->image_override) {
 					bo = t->mt->bo;
-				else
+				} else {
 					bo = t->bo;
+				}
 				if (bo) {
 
 					r700SyncSurf(context, bo,
@@ -67,7 +71,16 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 
 					BEGIN_BATCH_NO_AUTOSTATE(9 + 4);
 					R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_RESOURCE, 7));
-					R600_OUT_BATCH(i * 7);
+
+                    if( (1<<i) & vp->r700AsmCode.unVetTexBits )                    
+                    {   /* vs texture */                                     
+                        R600_OUT_BATCH((i + VERT_ATTRIB_MAX + SQ_FETCH_RESOURCE_VS_OFFSET) * FETCH_RESOURCE_STRIDE);
+                    }
+                    else
+                    {
+					    R600_OUT_BATCH(i * 7);
+                    }
+
 					R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE0);
 					R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE1);
 					R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE2);
@@ -77,7 +90,7 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 					R600_OUT_BATCH(r700->textures[i]->SQ_TEX_RESOURCE6);
 					R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE2,
 							     bo,
-							     0,
+							     r700->textures[i]->SQ_TEX_RESOURCE2,
 							     RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
 					R600_OUT_BATCH_RELOC(r700->textures[i]->SQ_TEX_RESOURCE3,
 							     bo,
@@ -91,21 +104,35 @@ static void r700SendTexState(GLcontext *ctx, struct radeon_state_atom *atom)
 	}
 }
 
+#define SAMPLER_STRIDE                 3
+
 static void r700SendTexSamplerState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	context_t         *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
 	unsigned int i;
+
+    struct r700_vertex_program *vp = context->selected_vp;
+
 	BATCH_LOCALS(&context->radeon);
 	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
 	for (i = 0; i < R700_TEXTURE_NUMBERUNITS; i++) {
-		if (ctx->Texture.Unit[i]._ReallyEnabled) {
+		if (ctx->Texture.Unit[i]._ReallyEnabled) {            
 			radeonTexObj *t = r700->textures[i];
 			if (t) {
 				BEGIN_BATCH_NO_AUTOSTATE(5);
 				R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_SAMPLER, 3));
-				R600_OUT_BATCH(i * 3);
+
+                if( (1<<i) & vp->r700AsmCode.unVetTexBits )                    
+                {   /* vs texture */
+                    R600_OUT_BATCH((i+SQ_TEX_SAMPLER_VS_OFFSET) * SAMPLER_STRIDE); //work 1
+                }
+                else
+                {
+				    R600_OUT_BATCH(i * 3);
+                }
+
 				R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER0);
 				R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER1);
 				R600_OUT_BATCH(r700->textures[i]->SQ_TEX_SAMPLER2);
@@ -141,77 +168,10 @@ static void r700SendTexBorderColorState(GLcontext *ctx, struct radeon_state_atom
 	}
 }
 
+extern int getTypeSize(GLenum type);
 static void r700SetupVTXConstants(GLcontext  * ctx,
-				  unsigned int nStreamID,
 				  void *       pAos,
-				  unsigned int size,      /* number of elements in vector */
-				  unsigned int stride,
-				  unsigned int count)     /* number of vectors in stream */
-{
-    context_t *context = R700_CONTEXT(ctx);
-    struct radeon_aos * paos = (struct radeon_aos *)pAos;
-    BATCH_LOCALS(&context->radeon);
-    radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
-
-    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);
-
-    uSQ_VTX_CONSTANT_WORD0_0 = paos->offset;
-    uSQ_VTX_CONSTANT_WORD1_0 = count * (size * 4) - 1;
-
-    SETfield(uSQ_VTX_CONSTANT_WORD2_0, 0, BASE_ADDRESS_HI_shift, BASE_ADDRESS_HI_mask); /* TODO */
-    SETfield(uSQ_VTX_CONSTANT_WORD2_0, stride, SQ_VTX_CONSTANT_WORD2_0__STRIDE_shift,
-	     SQ_VTX_CONSTANT_WORD2_0__STRIDE_mask);
-    SETfield(uSQ_VTX_CONSTANT_WORD2_0, GetSurfaceFormat(GL_FLOAT, 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 */
-    SETfield(uSQ_VTX_CONSTANT_WORD2_0, SQ_NUM_FORMAT_SCALED,
-	     SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_shift, SQ_VTX_CONSTANT_WORD2_0__NUM_FORMAT_ALL_mask);
-    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((nStreamID + 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();
-
-}
-
-extern int getTypeSize(GLenum type);
-static void r700SetupVTXConstants2(GLcontext  * ctx,
-				                   void *       pAos,
-                                   StreamDesc * pStreamDesc)     
+				  StreamDesc * pStreamDesc)
 {
     context_t *context = R700_CONTEXT(ctx);
     struct radeon_aos * paos = (struct radeon_aos *)pAos;
@@ -295,31 +255,6 @@ static void r700SetupVTXConstants2(GLcontext  * ctx,
 
 }
 
-void r700SetupStreams(GLcontext *ctx)
-{
-    context_t         *context = R700_CONTEXT(ctx);
-    struct r700_vertex_program *vp = context->selected_vp;
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
-    struct vertex_buffer *vb = &tnl->vb;
-    unsigned int i, j = 0;
-	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
-
-    R600_STATECHANGE(context, vtx);
-
-    for(i=0; i<VERT_ATTRIB_MAX; i++) {
-	    if(vp->mesa_program->Base.InputsRead & (1 << i)) {
-		    rcommon_emit_vector(ctx,
-					&context->radeon.tcl.aos[j],
-					vb->AttribPtr[i]->data,
-					vb->AttribPtr[i]->size,
-					vb->AttribPtr[i]->stride,
-					vb->Count);
-		    j++;
-	    }
-    }
-    context->radeon.tcl.aos_count = j;
-}
-
 static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
     context_t         *context = R700_CONTEXT(ctx);
@@ -343,25 +278,12 @@ 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)) 
-        {
-		    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++;
+	    if(vp->mesa_program->Base.InputsRead & (1 << i))
+	    {
+                r700SetupVTXConstants(ctx,
+				      (void*)(&context->radeon.tcl.aos[j]),
+				      &(context->stream_desc[j]));
+		j++;
 	    }
     }
 }
@@ -463,7 +385,6 @@ static void r700SendDepthTargetState(GLcontext *ctx, struct radeon_state_atom *a
 
 	rrb = radeon_get_depthbuffer(&context->radeon);
 	if (!rrb || !rrb->bo) {
-		fprintf(stderr, "no rrb\n");
 		return;
 	}
 
@@ -505,7 +426,6 @@ static void r700SendRenderTargetState(GLcontext *ctx, struct radeon_state_atom *
 
 	rrb = radeon_get_colorbuffer(&context->radeon);
 	if (!rrb || !rrb->bo) {
-		fprintf(stderr, "no rrb\n");
 		return;
 	}
 
@@ -549,68 +469,77 @@ static void r700SendRenderTargetState(GLcontext *ctx, struct radeon_state_atom *
 
 static void r700SendPSState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
-	context_t *context = R700_CONTEXT(ctx);
-	R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
-	struct radeon_bo * pbo;
-	BATCH_LOCALS(&context->radeon);
-	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+    context_t *context = R700_CONTEXT(ctx);
+    R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+    struct radeon_bo * pbo;
+    BATCH_LOCALS(&context->radeon);
+    radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
-	pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
+    pbo = (struct radeon_bo *)r700GetActiveFpShaderBo(GL_CONTEXT(context));
 
-	if (!pbo)
-		return;
+    if (!pbo)
+	    return;
 
-	r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+    r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
 
-        BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
-	R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
-	R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
-	R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
-			     pbo,
-			     r700->ps.SQ_PGM_START_PS.u32All,
-			     RADEON_GEM_DOMAIN_GTT, 0, 0);
-	END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+    R600_OUT_BATCH_REGSEQ(SQ_PGM_START_PS, 1);
+    R600_OUT_BATCH(r700->ps.SQ_PGM_START_PS.u32All);
+    R600_OUT_BATCH_RELOC(r700->ps.SQ_PGM_START_PS.u32All,
+		         pbo,
+		         r700->ps.SQ_PGM_START_PS.u32All,
+		         RADEON_GEM_DOMAIN_GTT, 0, 0);
+    END_BATCH();
 
-        BEGIN_BATCH_NO_AUTOSTATE(9);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
-        END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(9);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_PS, r700->ps.SQ_PGM_RESOURCES_PS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_EXPORTS_PS, r700->ps.SQ_PGM_EXPORTS_PS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_PS, r700->ps.SQ_PGM_CF_OFFSET_PS.u32All);
+    END_BATCH();
 
-	COMMIT_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    R600_OUT_BATCH_REGVAL(SQ_LOOP_CONST_0, 0x01000FFF);
+    END_BATCH();
+
+    COMMIT_BATCH();
 
 }
 
 static void r700SendVSState(GLcontext *ctx, struct radeon_state_atom *atom)
 {
-	context_t *context = R700_CONTEXT(ctx);
-	R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
-	struct radeon_bo * pbo;
-	BATCH_LOCALS(&context->radeon);
-	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+    context_t *context = R700_CONTEXT(ctx);
+    R700_CHIP_CONTEXT *r700 = R700_CONTEXT_STATES(context);
+    struct radeon_bo * pbo;
+    BATCH_LOCALS(&context->radeon);
+    radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
-	pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
+    pbo = (struct radeon_bo *)r700GetActiveVpShaderBo(GL_CONTEXT(context));
 
-	if (!pbo)
-		return;
+    if (!pbo)
+	    return;
 
-	r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
+    r700SyncSurf(context, pbo, RADEON_GEM_DOMAIN_GTT, 0, SH_ACTION_ENA_bit);
 
-        BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
-	R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
-	R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
-	R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
-			     pbo,
-			     r700->vs.SQ_PGM_START_VS.u32All,
-			     RADEON_GEM_DOMAIN_GTT, 0, 0);
-	END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+    R600_OUT_BATCH_REGSEQ(SQ_PGM_START_VS, 1);
+    R600_OUT_BATCH(r700->vs.SQ_PGM_START_VS.u32All);
+    R600_OUT_BATCH_RELOC(r700->vs.SQ_PGM_START_VS.u32All,
+		         pbo,
+		         r700->vs.SQ_PGM_START_VS.u32All,
+		         RADEON_GEM_DOMAIN_GTT, 0, 0);
+    END_BATCH();
 
-        BEGIN_BATCH_NO_AUTOSTATE(6);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
-	R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
-        END_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(6);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_RESOURCES_VS, r700->vs.SQ_PGM_RESOURCES_VS.u32All);
+    R600_OUT_BATCH_REGVAL(SQ_PGM_CF_OFFSET_VS, r700->vs.SQ_PGM_CF_OFFSET_VS.u32All);
+    END_BATCH();
 
-	COMMIT_BATCH();
+    BEGIN_BATCH_NO_AUTOSTATE(3);
+    R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + 32*4), 0x0100000F);
+    //R600_OUT_BATCH_REGVAL((SQ_LOOP_CONST_0 + (SQ_LOOP_CONST_vs<2)), 0x0100000F);
+    END_BATCH();
+
+    COMMIT_BATCH();
 }
 
 static void r700SendFSState(GLcontext *ctx, struct radeon_state_atom *atom)
@@ -891,8 +820,7 @@ static void r700SendDBState(GLcontext *ctx, struct radeon_state_atom *atom)
 	BATCH_LOCALS(&context->radeon);
 	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
 
-        BEGIN_BATCH_NO_AUTOSTATE(23);
-	R600_OUT_BATCH_REGVAL(DB_HTILE_DATA_BASE, r700->DB_HTILE_DATA_BASE.u32All);
+	BEGIN_BATCH_NO_AUTOSTATE(17);
 
 	R600_OUT_BATCH_REGSEQ(DB_STENCIL_CLEAR, 2);
 	R600_OUT_BATCH(r700->DB_STENCIL_CLEAR.u32All);
@@ -905,7 +833,6 @@ static void r700SendDBState(GLcontext *ctx, struct radeon_state_atom *atom)
 	R600_OUT_BATCH(r700->DB_RENDER_CONTROL.u32All);
 	R600_OUT_BATCH(r700->DB_RENDER_OVERRIDE.u32All);
 
-	R600_OUT_BATCH_REGVAL(DB_HTILE_SURFACE, r700->DB_HTILE_SURFACE.u32All);
 	R600_OUT_BATCH_REGVAL(DB_ALPHA_TO_MASK, r700->DB_ALPHA_TO_MASK.u32All);
 
 	END_BATCH();
@@ -1205,6 +1132,32 @@ static void r700SendVSConsts(GLcontext *ctx, struct radeon_state_atom *atom)
 	COMMIT_BATCH();
 }
 
+static void r700SendQueryBegin(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+	struct radeon_query_object *query = radeon->query.current;
+	BATCH_LOCALS(radeon);
+	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s\n", __func__);
+
+	/* clear the buffer */
+	radeon_bo_map(query->bo, GL_FALSE);
+	memset(query->bo->ptr, 0, 4 * 2 * sizeof(uint64_t)); /* 4 DBs, 2 qwords each */
+	radeon_bo_unmap(query->bo);
+
+	radeon_cs_space_check_with_bo(radeon->cmdbuf.cs,
+				      query->bo,
+				      0, RADEON_GEM_DOMAIN_GTT);
+
+	BEGIN_BATCH_NO_AUTOSTATE(4 + 2);
+	R600_OUT_BATCH(CP_PACKET3(R600_IT_EVENT_WRITE, 2));
+	R600_OUT_BATCH(ZPASS_DONE);
+	R600_OUT_BATCH(query->curr_offset); /* hw writes qwords */
+	R600_OUT_BATCH(0x00000000);
+	R600_OUT_BATCH_RELOC(VGT_EVENT_INITIATOR, query->bo, 0, 0, RADEON_GEM_DOMAIN_GTT, 0);
+	END_BATCH();
+	query->emitted_begin = GL_TRUE;
+}
+
 static int check_always(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	return atom->cmd_size;
@@ -1233,7 +1186,11 @@ static int check_blnd(GLcontext *ctx, struct radeon_state_atom *atom)
 		count += 3;
 
 	if (context->radeon.radeonScreen->chip_family > CHIP_FAMILY_R600) {
-		for (ui = 0; ui < R700_MAX_RENDER_TARGETS; ui++) {
+		/* targets are enabled in r700SetRenderTarget but state
+		   size is calculated before that. Until MRT's are done
+		   hardcode target0 as enabled. */
+		count += 3;
+		for (ui = 1; ui < R700_MAX_RENDER_TARGETS; ui++) {
                         if (r700->render_target[ui].enabled)
 				count += 3;
 		}
@@ -1313,6 +1270,20 @@ static int check_vs_consts(GLcontext *ctx, struct radeon_state_atom *atom)
 	return count;
 }
 
+static int check_queryobj(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+	struct radeon_query_object *query = radeon->query.current;
+	int count;
+
+	if (!query || query->emitted_begin)
+		count = 0;
+	else
+		count = atom->cmd_size;
+	radeon_print(RADEON_STATE, RADEON_TRACE, "%s %d\n", __func__, count);
+	return count;
+}
+
 #define ALLOC_STATE( ATOM, CHK, SZ, EMIT )				\
 do {									\
 	context->atoms.ATOM.cmd_size = (SZ);				\
@@ -1326,6 +1297,19 @@ do {									\
 	insert_at_tail(&context->radeon.hw.atomlist, &context->atoms.ATOM); \
 } while (0)
 
+static void r600_init_query_stateobj(radeonContextPtr radeon, int SZ)
+{
+	radeon->query.queryobj.cmd_size = (SZ);
+	radeon->query.queryobj.cmd = NULL;
+	radeon->query.queryobj.name = "queryobj";
+	radeon->query.queryobj.idx = 0;
+	radeon->query.queryobj.check = check_queryobj;
+	radeon->query.queryobj.dirty = GL_FALSE;
+	radeon->query.queryobj.emit = r700SendQueryBegin;
+	radeon->hw.max_state_size += (SZ);
+	insert_at_tail(&radeon->hw.atomlist, &radeon->query.queryobj);
+}
+
 void r600InitAtoms(context_t *context)
 {
 	radeon_print(RADEON_STATE, RADEON_NORMAL, "%s %p\n", __func__, context);
@@ -1336,7 +1320,7 @@ void r600InitAtoms(context_t *context)
 	context->radeon.hw.atomlist.name = "atom-list";
 
 	ALLOC_STATE(sq, always, 34, r700SendSQConfig);
-	ALLOC_STATE(db, always, 23, r700SendDBState);
+	ALLOC_STATE(db, always, 17, r700SendDBState);
 	ALLOC_STATE(stencil, always, 4, r700SendStencilState);
 	ALLOC_STATE(db_target, always, 12, r700SendDepthTargetState);
 	ALLOC_STATE(sc, always, 15, r700SendSCState);
@@ -1349,22 +1333,23 @@ void r600InitAtoms(context_t *context)
 	ALLOC_STATE(poly, always, 10, r700SendPolyState);
 	ALLOC_STATE(cb, cb, 18, r700SendCBState);
 	ALLOC_STATE(clrcmp, always, 6, r700SendCBCLRCMPState);
+	ALLOC_STATE(cb_target, always, 25, r700SendRenderTargetState);
 	ALLOC_STATE(blnd, blnd, (6 + (R700_MAX_RENDER_TARGETS * 3)), r700SendCBBlendState);
 	ALLOC_STATE(blnd_clr, always, 6, r700SendCBBlendColorState);
-	ALLOC_STATE(cb_target, always, 25, r700SendRenderTargetState);
 	ALLOC_STATE(sx, always, 9, r700SendSXState);
 	ALLOC_STATE(vgt, always, 41, r700SendVGTState);
 	ALLOC_STATE(spi, always, (59 + R700_MAX_SHADER_EXPORTS), r700SendSPIState);
 	ALLOC_STATE(vpt, always, 16, r700SendViewportState);
 	ALLOC_STATE(fs, always, 18, r700SendFSState);
-	ALLOC_STATE(vs, always, 18, r700SendVSState);
-	ALLOC_STATE(ps, always, 21, r700SendPSState);
+	ALLOC_STATE(vs, always, 21, r700SendVSState);
+	ALLOC_STATE(ps, always, 24, r700SendPSState);
 	ALLOC_STATE(vs_consts, vs_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendVSConsts);
 	ALLOC_STATE(ps_consts, ps_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendPSConsts);
 	ALLOC_STATE(vtx, vtx, (6 + (VERT_ATTRIB_MAX * 18)), r700SendVTXState);
 	ALLOC_STATE(tx, tx, (R700_TEXTURE_NUMBERUNITS * 20), r700SendTexState);
 	ALLOC_STATE(tx_smplr, tx, (R700_TEXTURE_NUMBERUNITS * 5), r700SendTexSamplerState);
 	ALLOC_STATE(tx_brdr_clr, tx, (R700_TEXTURE_NUMBERUNITS * 6), r700SendTexBorderColorState);
+	r600_init_query_stateobj(&context->radeon, 6 * 2);
 
 	context->radeon.hw.is_dirty = GL_TRUE;
 	context->radeon.hw.all_dirty = GL_TRUE;
diff --git a/src/mesa/drivers/dri/r600/r700_clear.c b/src/mesa/drivers/dri/r600/r700_clear.c
index c6546ab00c..98bfdd0937 100644
--- a/src/mesa/drivers/dri/r600/r700_clear.c
+++ b/src/mesa/drivers/dri/r600/r700_clear.c
@@ -49,14 +49,18 @@ static GLboolean r700ClearFast(context_t *context, GLbitfield mask)
 void r700Clear(GLcontext * ctx, GLbitfield mask)
 {
     context_t *context = R700_CONTEXT(ctx);
-    __DRIdrawablePrivate *dPriv = radeon_get_drawable(&context->radeon);
-    const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
+    __DRIdrawable *dPriv = radeon_get_drawable(&context->radeon);
+    const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask[0]);
     GLbitfield swrast_mask = 0, tri_mask = 0;
     int i;
     struct gl_framebuffer *fb = ctx->DrawBuffer;
 
     radeon_print(RADEON_RENDER, RADEON_VERBOSE, "%s %x\n", __func__, mask);
 
+    if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
+        context->radeon.front_buffer_dirty = GL_TRUE;
+    }
+
     if( GL_TRUE == r700ClearFast(context, mask) )
     {
         return;
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c
index 62a1ea1a22..84d51e6606 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.c
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.c
@@ -34,6 +34,7 @@
 #include "main/imports.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
+#include "shader/program.h"
 
 #include "r600_context.h"
 #include "r600_cmdbuf.h"
@@ -42,14 +43,68 @@
 
 #include "r700_debug.h"
 
+void insert_wpos_code(GLcontext *ctx, struct gl_fragment_program *fprog)
+{
+    static const gl_state_index winstate[STATE_LENGTH]
+         = { STATE_INTERNAL, STATE_FB_SIZE, 0, 0, 0};
+    struct prog_instruction *newInst, *inst;
+    GLint  win_size;  /* state reference */
+    GLuint wpos_temp; /* temp register */
+    int i, j;
+
+    /* PARAM win_size = STATE_FB_SIZE */
+    win_size = _mesa_add_state_reference(fprog->Base.Parameters, winstate);
+
+    wpos_temp = fprog->Base.NumTemporaries++;
+
+    /* scan program where WPOS is used and replace with wpos_temp */
+    inst = fprog->Base.Instructions;
+    for (i = 0; i < fprog->Base.NumInstructions; i++) {
+        for (j=0; j < 3; j++) {
+            if(inst->SrcReg[j].File == PROGRAM_INPUT && 
+               inst->SrcReg[j].Index == FRAG_ATTRIB_WPOS) {
+                inst->SrcReg[j].File = PROGRAM_TEMPORARY;
+                inst->SrcReg[j].Index = wpos_temp;
+            }
+        }
+        inst++;
+    }
+
+    _mesa_insert_instructions(&(fprog->Base), 0, 1);
+
+    newInst = fprog->Base.Instructions;
+    /* invert wpos.y
+     * wpos_temp.xyzw = wpos.x-yzw + winsize.0y00 */
+    newInst[0].Opcode = OPCODE_ADD;
+    newInst[0].DstReg.File = PROGRAM_TEMPORARY;
+    newInst[0].DstReg.Index = wpos_temp;
+    newInst[0].DstReg.WriteMask = WRITEMASK_XYZW;
+
+    newInst[0].SrcReg[0].File = PROGRAM_INPUT;
+    newInst[0].SrcReg[0].Index = FRAG_ATTRIB_WPOS;
+    newInst[0].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+    newInst[0].SrcReg[0].Negate = NEGATE_Y;
+
+    newInst[0].SrcReg[1].File = PROGRAM_STATE_VAR;
+    newInst[0].SrcReg[1].Index = win_size;
+    newInst[0].SrcReg[1].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ZERO, SWIZZLE_Y, SWIZZLE_ZERO, SWIZZLE_ZERO);
+
+}
+
 //TODO : Validate FP input with VP output.
 void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
-						  struct gl_fragment_program *mesa_fp)
+						  struct gl_fragment_program *mesa_fp,
+                          GLcontext *ctx) 
 {
 	unsigned int unBit;
     unsigned int i;
     GLuint       ui;
 
+    /* match fp inputs with vp exports. */
+    struct r700_vertex_program_cont *vpc =
+		       (struct r700_vertex_program_cont *)ctx->VertexProgram._Current;
+    GLbitfield OutputsWritten = vpc->mesa_program.Base.OutputsWritten;
+    
 	pAsm->number_used_registers = 0;
 
 //Input mapping : mesa_fp->Base.InputsRead set the flag, set in 
@@ -61,32 +116,99 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
 		pAsm->uiFP_AttributeMap[FRAG_ATTRIB_WPOS] = pAsm->number_used_registers++;
 	}
 
-	unBit = 1 << FRAG_ATTRIB_COL0;
-	if(mesa_fp->Base.InputsRead & unBit)
+    unBit = 1 << VERT_RESULT_COL0;
+	if(OutputsWritten & unBit)
 	{
 		pAsm->uiFP_AttributeMap[FRAG_ATTRIB_COL0] = pAsm->number_used_registers++;
 	}
 
-	unBit = 1 << FRAG_ATTRIB_COL1;
-	if(mesa_fp->Base.InputsRead & unBit)
+	unBit = 1 << VERT_RESULT_COL1;
+	if(OutputsWritten & unBit)
 	{
 		pAsm->uiFP_AttributeMap[FRAG_ATTRIB_COL1] = pAsm->number_used_registers++;
 	}
 
-        unBit = 1 << FRAG_ATTRIB_FOGC;
-        if(mesa_fp->Base.InputsRead & unBit)
-        {
-                pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FOGC] = pAsm->number_used_registers++;
-        }
+    unBit = 1 << VERT_RESULT_FOGC;
+    if(OutputsWritten & unBit)
+    {
+        pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FOGC] = pAsm->number_used_registers++;
+    }
 
 	for(i=0; i<8; i++)
 	{
-		unBit = 1 << (FRAG_ATTRIB_TEX0 + i);
-		if(mesa_fp->Base.InputsRead & unBit)
+		unBit = 1 << (VERT_RESULT_TEX0 + i);
+		if(OutputsWritten & unBit)
 		{
 			pAsm->uiFP_AttributeMap[FRAG_ATTRIB_TEX0 + i] = pAsm->number_used_registers++;
 		}
 	}
+ 
+/* order has been taken care of */ 
+#if 1
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+        unBit = 1 << i;
+        if(OutputsWritten & unBit)
+		{
+            pAsm->uiFP_AttributeMap[i-VERT_RESULT_VAR0+FRAG_ATTRIB_VAR0] = pAsm->number_used_registers++;
+        }
+    }
+#else
+    if( (mesa_fp->Base.InputsRead >> FRAG_ATTRIB_VAR0) > 0 )
+    {
+	    struct r700_vertex_program_cont *vpc =
+		       (struct r700_vertex_program_cont *)ctx->VertexProgram._Current;
+        struct gl_program_parameter_list * VsVarying = vpc->mesa_program.Base.Varying;
+        struct gl_program_parameter_list * PsVarying = mesa_fp->Base.Varying;
+        struct gl_program_parameter      * pVsParam;
+        struct gl_program_parameter      * pPsParam;
+        GLuint j, k;
+        GLuint unMaxVarying = 0;
+
+        for(i=0; i<VsVarying->NumParameters; i++)
+        {
+            pAsm->uiFP_AttributeMap[i + FRAG_ATTRIB_VAR0] = 0;
+        }
+
+        for(i=FRAG_ATTRIB_VAR0; i<FRAG_ATTRIB_MAX; i++)
+	    {
+            unBit = 1 << i;
+            if(mesa_fp->Base.InputsRead & unBit)
+		    {
+                j = i - FRAG_ATTRIB_VAR0;
+                pPsParam = PsVarying->Parameters + j;
+
+                for(k=0; k<VsVarying->NumParameters; k++)
+                {					
+                    pVsParam = VsVarying->Parameters + k;
+
+			        if( strcmp(pPsParam->Name, pVsParam->Name) == 0)
+                    {
+                        pAsm->uiFP_AttributeMap[i] = pAsm->number_used_registers + k;                  
+                        if(k > unMaxVarying)
+                        {
+                            unMaxVarying = k;
+                        }
+                        break;
+                    }
+                }
+		    }
+        }
+
+        pAsm->number_used_registers += unMaxVarying + 1;
+    }
+#endif
+    unBit = 1 << FRAG_ATTRIB_FACE;
+    if(mesa_fp->Base.InputsRead & unBit)
+    {
+        pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FACE] = pAsm->number_used_registers++;
+    }
+
+    unBit = 1 << FRAG_ATTRIB_PNTC;
+    if(mesa_fp->Base.InputsRead & unBit)
+    {
+        pAsm->uiFP_AttributeMap[FRAG_ATTRIB_PNTC] = pAsm->number_used_registers++;
+    }
 
 /* Map temporary registers (GPRs) */
     pAsm->starting_temp_register_number = pAsm->number_used_registers;
@@ -127,6 +249,8 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
         pAsm->pucOutMask[ui] = 0x0;
     }
 
+    pAsm->flag_reg_index = pAsm->number_used_registers++;
+
     pAsm->uFirstHelpReg = pAsm->number_used_registers;
 }
 
@@ -135,15 +259,19 @@ GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
 {
     GLuint i, j;
     GLint * puiTEMPwrites;
+    GLint * puiTEMPreads;
     struct prog_instruction * pILInst;
     InstDeps         *pInstDeps;
     struct prog_instruction * texcoord_DepInst;
     GLint              nDepInstID;
 
     puiTEMPwrites = (GLint*) MALLOC(sizeof(GLuint)*mesa_fp->Base.NumTemporaries);
+    puiTEMPreads = (GLint*) MALLOC(sizeof(GLuint)*mesa_fp->Base.NumTemporaries);
+
     for(i=0; i<mesa_fp->Base.NumTemporaries; i++)
     {
         puiTEMPwrites[i] = -1;
+        puiTEMPreads[i] = -1;
     }
 
     pInstDeps = (InstDeps*)MALLOC(sizeof(InstDeps)*mesa_fp->Base.NumInstructions);
@@ -167,6 +295,11 @@ GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
             {
                 //Set dep.
                 pInstDeps[i].nSrcDeps[j] = puiTEMPwrites[pILInst->SrcReg[j].Index];
+                //Set first read
+                if(puiTEMPreads[pILInst->SrcReg[j].Index] < 0 )
+                {
+                    puiTEMPreads[pILInst->SrcReg[j].Index] = i;
+                }
             }
             else
             {
@@ -177,8 +310,6 @@ GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
 
     fp->r700AsmCode.pInstDeps = pInstDeps;
 
-    FREE(puiTEMPwrites);
-
     //Find dep for tex inst    
     for(i=0; i<mesa_fp->Base.NumInstructions; i++)
     {
@@ -203,29 +334,84 @@ GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
                 {   //... other deps?
                 }
             }
+            // make sure that we dont overwrite src used earlier
+            nDepInstID = puiTEMPreads[pILInst->DstReg.Index];
+            if(nDepInstID < i)
+            {
+                pInstDeps[i].nDstDep = puiTEMPreads[pILInst->DstReg.Index];
+                texcoord_DepInst = &(mesa_fp->Base.Instructions[nDepInstID]);
+                if(GL_TRUE == IsAlu(texcoord_DepInst->Opcode) )
+                {
+                    pInstDeps[nDepInstID].nDstDep = i;
+                }
+ 
+            }
+
         }
 	}
 
+    FREE(puiTEMPwrites);
+    FREE(puiTEMPreads);
+
     return GL_TRUE;
 }
 
 GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
-							     struct gl_fragment_program   *mesa_fp)
+							     struct gl_fragment_program   *mesa_fp,
+                                 GLcontext *ctx) 
 {
 	GLuint    number_of_colors_exported;
 	GLboolean z_enabled = GL_FALSE;
-	GLuint    unBit;
+	GLuint    unBit, shadow_unit;
+	int i;
+	struct prog_instruction *inst;
+	gl_state_index shadow_ambient[STATE_LENGTH]
+	    = { STATE_INTERNAL, STATE_SHADOW_AMBIENT, 0, 0, 0};
 
     //Init_Program
 	Init_r700_AssemblerBase( SPT_FP, &(fp->r700AsmCode), &(fp->r700Shader) );
-	Map_Fragment_Program(&(fp->r700AsmCode), mesa_fp);
+
+    if(mesa_fp->Base.InputsRead & FRAG_BIT_WPOS)
+    {
+        insert_wpos_code(ctx, mesa_fp);
+    }
+
+    /* add/map  consts for ARB_shadow_ambient */
+    if(mesa_fp->Base.ShadowSamplers)
+    {
+        inst = mesa_fp->Base.Instructions;
+        for (i = 0; i < mesa_fp->Base.NumInstructions; i++)
+        {
+            if(inst->TexShadow == 1)
+            {
+                shadow_unit = inst->TexSrcUnit;
+                shadow_ambient[2] = shadow_unit;
+                fp->r700AsmCode.shadow_regs[shadow_unit] = 
+                    _mesa_add_state_reference(mesa_fp->Base.Parameters, shadow_ambient);
+            }
+            inst++;
+        }
+    }
+
+    Map_Fragment_Program(&(fp->r700AsmCode), mesa_fp, ctx); 
 
     if( GL_FALSE == Find_Instruction_Dependencies_fp(fp, mesa_fp) )
 	{
 		return GL_FALSE;
     }
+
+    InitShaderProgram(&(fp->r700AsmCode));
 	
-	if( GL_FALSE == AssembleInstr(mesa_fp->Base.NumInstructions,
+    for(i=0; i < MAX_SAMPLERS; i++)
+    {
+         fp->r700AsmCode.SamplerUnits[i] = fp->mesa_program.Base.SamplerUnits[i];
+    }
+
+    fp->r700AsmCode.unCurNumILInsts = mesa_fp->Base.NumInstructions;
+
+	if( GL_FALSE == AssembleInstr(0,
+                                  0,
+                                  mesa_fp->Base.NumInstructions,
                                   &(mesa_fp->Base.Instructions[0]), 
                                   &(fp->r700AsmCode)) )
 	{
@@ -237,6 +423,11 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
         return GL_FALSE;
     }
 
+    if( GL_FALSE == RelocProgram(&(fp->r700AsmCode), &(mesa_fp->Base)) )
+    {
+        return GL_FALSE;
+    }
+
     fp->r700Shader.nRegs = (fp->r700AsmCode.number_used_registers == 0) ? 0 
                          : (fp->r700AsmCode.number_used_registers - 1);
 
@@ -251,7 +442,15 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
 		number_of_colors_exported--;
 	}
 
-	fp->r700Shader.exportMode = number_of_colors_exported << 1 | z_enabled;
+	/* illegal to set this to 0 */
+	if(number_of_colors_exported || z_enabled)
+	{
+	    fp->r700Shader.exportMode = number_of_colors_exported << 1 | z_enabled;
+	}
+	else
+	{
+	    fp->r700Shader.exportMode = (1 << 1);
+	}
 
     fp->translated = GL_TRUE;
 
@@ -269,7 +468,7 @@ void r700SelectFragmentShader(GLcontext *ctx)
     }
 
     if (GL_FALSE == fp->translated)
-	    r700TranslateFragmentShader(fp, &(fp->mesa_program));
+	    r700TranslateFragmentShader(fp, &(fp->mesa_program), ctx); 
 }
 
 void * r700GetActiveFpShaderBo(GLcontext * ctx)
@@ -294,6 +493,7 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
     unsigned int unNumOfReg;
     unsigned int unBit;
     GLuint exportCount;
+    GLboolean point_sprite = GL_FALSE;
 
     if(GL_FALSE == fp->loaded)
     {
@@ -347,6 +547,50 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
         CLEARbit(r700->SPI_INPUT_Z.u32All, PROVIDE_Z_TO_SPI_bit);
     }
 
+    if (mesa_fp->Base.InputsRead & (1 << FRAG_ATTRIB_FACE))
+    {
+        ui += 1;
+        SETfield(r700->SPI_PS_IN_CONTROL_0.u32All, ui, NUM_INTERP_shift, NUM_INTERP_mask);
+        SETbit(r700->SPI_PS_IN_CONTROL_1.u32All, FRONT_FACE_ENA_bit);
+        SETbit(r700->SPI_PS_IN_CONTROL_1.u32All, FRONT_FACE_ALL_BITS_bit);
+        SETfield(r700->SPI_PS_IN_CONTROL_1.u32All, pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FACE], FRONT_FACE_ADDR_shift, FRONT_FACE_ADDR_mask);
+    }
+    else
+    {
+        CLEARbit(r700->SPI_PS_IN_CONTROL_1.u32All, FRONT_FACE_ENA_bit);
+    }
+
+    /* see if we need any point_sprite replacements */
+    for (i = VERT_RESULT_TEX0; i<= VERT_RESULT_TEX7; i++)
+    {
+        if(ctx->Point.CoordReplace[i - VERT_RESULT_TEX0] == GL_TRUE)
+            point_sprite = GL_TRUE;
+    }
+
+    if ((mesa_fp->Base.InputsRead & (1 << FRAG_ATTRIB_PNTC)) || point_sprite)
+    {
+        /* for FRAG_ATTRIB_PNTC we need to increase num_interp */
+        if(mesa_fp->Base.InputsRead & (1 << FRAG_ATTRIB_PNTC))
+        {
+            ui++;
+            SETfield(r700->SPI_PS_IN_CONTROL_0.u32All, ui, NUM_INTERP_shift, NUM_INTERP_mask);
+        }
+        SETbit(r700->SPI_INTERP_CONTROL_0.u32All, PNT_SPRITE_ENA_bit);
+        SETfield(r700->SPI_INTERP_CONTROL_0.u32All, SPI_PNT_SPRITE_SEL_S, PNT_SPRITE_OVRD_X_shift, PNT_SPRITE_OVRD_X_mask);
+        SETfield(r700->SPI_INTERP_CONTROL_0.u32All, SPI_PNT_SPRITE_SEL_T, PNT_SPRITE_OVRD_Y_shift, PNT_SPRITE_OVRD_Y_mask);
+        SETfield(r700->SPI_INTERP_CONTROL_0.u32All, SPI_PNT_SPRITE_SEL_0, PNT_SPRITE_OVRD_Z_shift, PNT_SPRITE_OVRD_Z_mask);
+        SETfield(r700->SPI_INTERP_CONTROL_0.u32All, SPI_PNT_SPRITE_SEL_1, PNT_SPRITE_OVRD_W_shift, PNT_SPRITE_OVRD_W_mask);
+        if(ctx->Point.SpriteOrigin == GL_LOWER_LEFT)
+            SETbit(r700->SPI_INTERP_CONTROL_0.u32All, PNT_SPRITE_TOP_1_bit);
+        else
+            CLEARbit(r700->SPI_INTERP_CONTROL_0.u32All, PNT_SPRITE_TOP_1_bit);
+    }
+    else
+    {
+        CLEARbit(r700->SPI_INTERP_CONTROL_0.u32All, PNT_SPRITE_ENA_bit);
+    }
+
+
     ui = (unNumOfReg < ui) ? ui : unNumOfReg;
 
     SETfield(r700->ps.SQ_PGM_RESOURCES_PS.u32All, ui, NUM_GPRS_shift, NUM_GPRS_mask);
@@ -362,27 +606,14 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
     SETfield(r700->ps.SQ_PGM_EXPORTS_PS.u32All, fp->r700Shader.exportMode,
              EXPORT_MODE_shift, EXPORT_MODE_mask);
 
-    R600_STATECHANGE(context, db);
-
-    if(fp->r700Shader.killIsUsed)
-    {
-	    SETbit(r700->DB_SHADER_CONTROL.u32All, KILL_ENABLE_bit);
-    }
-    else
-    {
-        CLEARbit(r700->DB_SHADER_CONTROL.u32All, KILL_ENABLE_bit);
-    }
-
-    if(fp->r700Shader.depthIsExported)
-    {
-	    SETbit(r700->DB_SHADER_CONTROL.u32All, Z_EXPORT_ENABLE_bit);
-    }
-    else
-    {
-        CLEARbit(r700->DB_SHADER_CONTROL.u32All, Z_EXPORT_ENABLE_bit);
-    }
-
     // emit ps input map
+    struct r700_vertex_program_cont *vpc =
+		       (struct r700_vertex_program_cont *)ctx->VertexProgram._Current;
+    GLbitfield OutputsWritten = vpc->mesa_program.Base.OutputsWritten;
+    
+    for(ui = 0; ui < R700_MAX_SHADER_EXPORTS; ui++)
+        r700->SPI_PS_INPUT_CNTL[ui].u32All = 0;
+
     unBit = 1 << FRAG_ATTRIB_WPOS;
     if(mesa_fp->Base.InputsRead & unBit)
     {
@@ -396,8 +627,8 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
                     CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
     }
 
-    unBit = 1 << FRAG_ATTRIB_COL0;
-    if(mesa_fp->Base.InputsRead & unBit)
+    unBit = 1 << VERT_RESULT_COL0;
+    if(OutputsWritten & unBit)
     {
 	    ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_COL0];
 	    SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
@@ -409,8 +640,8 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 		    CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
     }
 
-    unBit = 1 << FRAG_ATTRIB_COL1;
-    if(mesa_fp->Base.InputsRead & unBit)
+    unBit = 1 << VERT_RESULT_COL1;
+    if(OutputsWritten & unBit)
     {
 	    ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_COL1];
 	    SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
@@ -422,8 +653,8 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 		    CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
     }
 
-    unBit = 1 << FRAG_ATTRIB_FOGC;
-    if(mesa_fp->Base.InputsRead & unBit)
+    unBit = 1 << VERT_RESULT_FOGC;
+    if(OutputsWritten & unBit)
     {
             ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FOGC];
             SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
@@ -437,25 +668,79 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 
     for(i=0; i<8; i++)
     {
-	    unBit = 1 << (FRAG_ATTRIB_TEX0 + i);
-	    if(mesa_fp->Base.InputsRead & unBit)
+	    unBit = 1 << (VERT_RESULT_TEX0 + i);
+	    if(OutputsWritten & unBit)
 	    {
 		    ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_TEX0 + i];
 		    SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
 		    SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui,
 			     SEMANTIC_shift, SEMANTIC_mask);
 		    CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+		    /* ARB_point_sprite */
+		    if(ctx->Point.CoordReplace[i] == GL_TRUE)
+		    {
+			     SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, PT_SPRITE_TEX_bit);
+		    }
 	    }
     }
 
-    R600_STATECHANGE(context, cb);
+    unBit = 1 << FRAG_ATTRIB_FACE;
+    if(mesa_fp->Base.InputsRead & unBit)
+    {
+            ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_FACE];
+            SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
+            SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui,
+                     SEMANTIC_shift, SEMANTIC_mask);
+            if (r700->SPI_INTERP_CONTROL_0.u32All & FLAT_SHADE_ENA_bit)
+                    SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+            else
+                    CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+    }
+    unBit = 1 << FRAG_ATTRIB_PNTC;
+    if(mesa_fp->Base.InputsRead & unBit)
+    {
+            ui = pAsm->uiFP_AttributeMap[FRAG_ATTRIB_PNTC];
+            SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
+            SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui,
+                     SEMANTIC_shift, SEMANTIC_mask);
+            if (r700->SPI_INTERP_CONTROL_0.u32All & FLAT_SHADE_ENA_bit)
+                    SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+            else
+                    CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+            SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, PT_SPRITE_TEX_bit);
+    }
+
+
+
+
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+        unBit = 1 << i;
+        if(OutputsWritten & unBit)
+		{
+            ui = pAsm->uiFP_AttributeMap[i-VERT_RESULT_VAR0+FRAG_ATTRIB_VAR0];
+            SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, SEL_CENTROID_bit);
+            SETfield(r700->SPI_PS_INPUT_CNTL[ui].u32All, ui,
+		             SEMANTIC_shift, SEMANTIC_mask);
+            if (r700->SPI_INTERP_CONTROL_0.u32All & FLAT_SHADE_ENA_bit)
+		        SETbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+            else
+		        CLEARbit(r700->SPI_PS_INPUT_CNTL[ui].u32All, FLAT_SHADE_bit);
+        }
+    }
+
     exportCount = (r700->ps.SQ_PGM_EXPORTS_PS.u32All & EXPORT_MODE_mask) / (1 << EXPORT_MODE_shift);
-    r700->CB_SHADER_CONTROL.u32All = (1 << exportCount) - 1;
+    if (r700->CB_SHADER_CONTROL.u32All != ((1 << exportCount) - 1))
+    {
+	    R600_STATECHANGE(context, cb);
+	    r700->CB_SHADER_CONTROL.u32All = (1 << exportCount) - 1;
+    }
 
     /* sent out shader constants. */
     paramList = fp->mesa_program.Base.Parameters;
 
-    if(NULL != paramList) {
+    if(NULL != paramList) 
+    {
 	    _mesa_load_state_parameters(ctx, paramList);
 
 	    if (paramList->NumParameters > R700_MAX_DX9_CONSTS)
@@ -468,14 +753,33 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 	    unNumParamData = paramList->NumParameters;
 
 	    for(ui=0; ui<unNumParamData; ui++) {
-		    r700->ps.consts[ui][0].f32All = paramList->ParameterValues[ui][0];
-		    r700->ps.consts[ui][1].f32All = paramList->ParameterValues[ui][1];
-		    r700->ps.consts[ui][2].f32All = paramList->ParameterValues[ui][2];
-		    r700->ps.consts[ui][3].f32All = paramList->ParameterValues[ui][3];
+		        r700->ps.consts[ui][0].f32All = paramList->ParameterValues[ui][0];
+		        r700->ps.consts[ui][1].f32All = paramList->ParameterValues[ui][1];
+		        r700->ps.consts[ui][2].f32All = paramList->ParameterValues[ui][2];
+		        r700->ps.consts[ui][3].f32All = paramList->ParameterValues[ui][3];
 	    }
     } else
 	    r700->ps.num_consts = 0;
 
+    COMPILED_SUB * pCompiledSub;
+    GLuint uj;
+    GLuint unConstOffset = r700->ps.num_consts;
+    for(ui=0; ui<pAsm->unNumPresub; ui++)
+    {
+        pCompiledSub = pAsm->presubs[ui].pCompiledSub;
+
+        r700->ps.num_consts += pCompiledSub->NumParameters;
+
+        for(uj=0; uj<pCompiledSub->NumParameters; uj++)
+        {
+            r700->ps.consts[uj + unConstOffset][0].f32All = pCompiledSub->ParameterValues[uj][0];
+		    r700->ps.consts[uj + unConstOffset][1].f32All = pCompiledSub->ParameterValues[uj][1];
+		    r700->ps.consts[uj + unConstOffset][2].f32All = pCompiledSub->ParameterValues[uj][2];
+		    r700->ps.consts[uj + unConstOffset][3].f32All = pCompiledSub->ParameterValues[uj][3];
+        }
+        unConstOffset += pCompiledSub->NumParameters;
+    }
+
     return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.h b/src/mesa/drivers/dri/r600/r700_fragprog.h
index cbb108d212..39c59c9201 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.h
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.h
@@ -48,13 +48,17 @@ struct r700_fragment_program
 };
 
 /* Internal */
+void insert_wpos_code(GLcontext *ctx, struct gl_fragment_program *fprog);
+
 void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
-			  struct gl_fragment_program *mesa_fp);
+			  struct gl_fragment_program *mesa_fp,
+                          GLcontext *ctx); 
 GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
 					   struct gl_fragment_program   *mesa_fp);
 
 GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
-				      struct gl_fragment_program   *mesa_vp);
+				      struct gl_fragment_program   *mesa_vp,
+                                      GLcontext *ctx); 
 
 /* Interface */
 extern void r700SelectFragmentShader(GLcontext *ctx);
diff --git a/src/mesa/drivers/dri/r600/r700_oglprog.c b/src/mesa/drivers/dri/r600/r700_oglprog.c
index 5290ef31be..0d476fcd86 100644
--- a/src/mesa/drivers/dri/r600/r700_oglprog.c
+++ b/src/mesa/drivers/dri/r600/r700_oglprog.c
@@ -40,6 +40,24 @@
 #include "r700_vertprog.h"
 
 
+static void freeVertProgCache(GLcontext *ctx, struct r700_vertex_program_cont *cache)
+{
+	struct r700_vertex_program *tmp, *vp = cache->progs;
+
+	while (vp) {
+		tmp = vp->next;
+		/* Release DMA region */
+		r600DeleteShader(ctx, vp->shaderbo);
+		/* Clean up */
+		Clean_Up_Assembler(&(vp->r700AsmCode));
+		Clean_Up_Shader(&(vp->r700Shader));
+		
+		_mesa_reference_vertprog(ctx, &vp->mesa_program, NULL);
+		_mesa_free(vp);
+		vp = tmp;
+	}
+}
+
 static struct gl_program *r700NewProgram(GLcontext * ctx, 
                                          GLenum target,
 					                     GLuint id)
@@ -84,8 +102,7 @@ static struct gl_program *r700NewProgram(GLcontext * ctx,
 
 static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 {
-    struct r700_vertex_program_cont   * vpc;
-    struct r700_vertex_program *vp, *tmp;
+    struct r700_vertex_program_cont *vpc = (struct r700_vertex_program_cont *)prog;
     struct r700_fragment_program * fp;
 
 	radeon_print(RADEON_SHADER, RADEON_VERBOSE,
@@ -95,20 +112,7 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
     {
     case GL_VERTEX_STATE_PROGRAM_NV:
     case GL_VERTEX_PROGRAM_ARB:	    
-        vpc = (struct r700_vertex_program_cont*)prog;
-        vp = vpc->progs;
-	while (vp) {
-		tmp = vp->next;
-		/* Release DMA region */
-	 
-	        r600DeleteShader(ctx, vp->shaderbo);
-
-	        /* Clean up */
-	        Clean_Up_Assembler(&(vp->r700AsmCode));
-	        Clean_Up_Shader(&(vp->r700Shader));
-		_mesa_free(vp);
-		vp = tmp;
-	}
+	    freeVertProgCache(ctx, vpc);
 	    break;
     case GL_FRAGMENT_PROGRAM_NV:
     case GL_FRAGMENT_PROGRAM_ARB:
@@ -131,7 +135,24 @@ static void r700DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 static void
 r700ProgramStringNotify(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
-
+	struct r700_vertex_program_cont *vpc = (struct r700_vertex_program_cont *)prog;
+	struct r700_fragment_program * fp = (struct r700_fragment_program*)prog;
+
+	switch (target) {
+	case GL_VERTEX_PROGRAM_ARB:
+		freeVertProgCache(ctx, vpc);
+		vpc->progs = NULL;
+		break;
+	case GL_FRAGMENT_PROGRAM_ARB:
+		r600DeleteShader(ctx, fp->shaderbo);
+		Clean_Up_Assembler(&(fp->r700AsmCode));
+		Clean_Up_Shader(&(fp->r700Shader));
+		fp->translated = GL_FALSE;
+		fp->loaded     = GL_FALSE;
+		fp->shaderbo   = NULL;
+		break;
+	}
+		
 }
 
 static GLboolean r700IsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index 4f39d9f1bd..eab27cbd84 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -59,9 +59,7 @@
 
 void r700WaitForIdle(context_t *context);
 void r700WaitForIdleClean(context_t *context);
-GLboolean r700SendTextureState(context_t *context);
 static unsigned int r700PrimitiveType(int prim);
-void r600UpdateTextureState(GLcontext * ctx);
 GLboolean r700SyncSurf(context_t *context,
 		       struct radeon_bo *pbo,
 		       uint32_t read_domain,
@@ -253,24 +251,12 @@ 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 type, 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);
@@ -282,90 +268,153 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim
     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 */
-                     + 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);
 
-	// index type
     SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
 
-    if(GL_TRUE == bUseDrawIndex)
+    if(GL_TRUE != context->ind_buf.is_32bit)
     {
-        if(GL_TRUE != context->ind_buf.is_32bit)
-        {
             SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
-        }
     }
 
+    vgt_num_indices = num_indices;
+    SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
+    SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask);
+
+    total_emit =   3  /* VGT_PRIMITIVE_TYPE */
+	         + 2  /* VGT_INDEX_TYPE */
+	         + 2  /* NUM_INSTANCES */
+	         + 5 + 2; /* DRAW_INDEX */
+
+    BEGIN_BATCH_NO_AUTOSTATE(total_emit);
+    // prim
+    R600_OUT_BATCH_REGSEQ(VGT_PRIMITIVE_TYPE, 1);
+    R600_OUT_BATCH(vgt_primitive_type);
+    // index type
     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
+    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);
+    END_BATCH();
+    COMMIT_BATCH();
+}
+
+static void r700RunRenderPrimitiveImmediate(GLcontext * ctx, int start, int end, int prim)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    BATCH_LOCALS(&context->radeon);
+    int type, i;
+    uint32_t num_indices, total_emit = 0;
+    uint32_t vgt_draw_initiator = 0;
+    uint32_t vgt_index_type     = 0;
+    uint32_t vgt_primitive_type = 0;
+    uint32_t vgt_num_indices    = 0;
+
+    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;
+
+    SETfield(vgt_primitive_type, type,
+	     VGT_PRIMITIVE_TYPE__PRIM_TYPE_shift, VGT_PRIMITIVE_TYPE__PRIM_TYPE_mask);
+
+    if (num_indices > 0xffff)
+    {
+	    SETfield(vgt_index_type, DI_INDEX_SIZE_32_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
+    }
+    else
+    {
+            SETfield(vgt_index_type, DI_INDEX_SIZE_16_BIT, INDEX_TYPE_shift, INDEX_TYPE_mask);
+    }
+
     vgt_num_indices = num_indices;
+    SETfield(vgt_draw_initiator, DI_MAJOR_MODE_0, MAJOR_MODE_shift, MAJOR_MODE_mask);
 
-    if(GL_TRUE == bUseDrawIndex)
+    if (start == 0)
     {
-        SETfield(vgt_draw_initiator, DI_SRC_SEL_DMA, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
+	SETfield(vgt_draw_initiator, DI_SRC_SEL_AUTO_INDEX, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
     }
     else
     {
-        SETfield(vgt_draw_initiator, DI_SRC_SEL_IMMEDIATE, SOURCE_SELECT_shift, SOURCE_SELECT_mask);
+	if (num_indices > 0xffff)
+	{
+		total_emit += num_indices;
+	}
+	else
+	{
+		total_emit += (num_indices + 1) / 2;
+	}
+	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);
+    total_emit +=   3 /* VGT_PRIMITIVE_TYPE */
+	          + 2 /* VGT_INDEX_TYPE */
+	          + 2 /* NUM_INSTANCES */
+	          + 3; /* DRAW */
 
-    if(GL_TRUE == bUseDrawIndex)
+    BEGIN_BATCH_NO_AUTOSTATE(total_emit);
+    // prim
+    R600_OUT_BATCH_REGSEQ(VGT_PRIMITIVE_TYPE, 1);
+    R600_OUT_BATCH(vgt_primitive_type);
+    // index type
+    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
+    if(start == 0)
     {
-        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(CP_PACKET3(R600_IT_DRAW_INDEX_AUTO, 1));
         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
+	if (num_indices > 0xffff)
+        {
+	    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++)
 	    {
-                R600_OUT_BATCH(i);
+		R600_OUT_BATCH(i);
 	    }
-        }
+	}
+	else
+        {
+	    R600_OUT_BATCH(CP_PACKET3(R600_IT_DRAW_INDEX_IMMD, (((num_indices + 1) / 2) + 1)));
+	    R600_OUT_BATCH(vgt_num_indices);
+	    R600_OUT_BATCH(vgt_draw_initiator);
+	    for (i = start; i < (start + num_indices); i += 2)
+	    {
+		if ((i + 1) == (start + num_indices))
+		{
+		    R600_OUT_BATCH(i);
+		}
+		else
+		{
+		    R600_OUT_BATCH(((i + 1) << 16) | (i));
+		}
+	    }
+	}
     }
 
     END_BATCH();
@@ -375,173 +424,45 @@ static void r700RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim
 /* start 3d, idle, cb/db flush */
 #define PRE_EMIT_STATE_BUFSZ 10 + 5 + 14
 
-static GLuint r700PredictRenderSize(GLcontext* ctx, GLuint nr_prims)
+static GLuint r700PredictRenderSize(GLcontext* ctx,
+				    const struct _mesa_prim *prim,
+				    const struct _mesa_index_buffer *ib,
+				    GLuint nr_prims)
 {
     context_t *context = R700_CONTEXT(ctx);
-    struct r700_vertex_program *vp = context->selected_vp;
     GLboolean flushed;
     GLuint dwords, i;
     GLuint state_size;
-    /* pre calculate aos count so state prediction works */
-    context->radeon.tcl.aos_count = _mesa_bitcount(vp->mesa_program->Base.InputsRead);
 
     dwords = PRE_EMIT_STATE_BUFSZ;
-    if (nr_prims)
+    if (ib)
 	    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;
+	    for (i = 0; i < nr_prims; ++i)
+	    {
+		    if (prim[i].start == 0)
+			    dwords += 10;
+		    else if (prim[i].count > 0xffff)
+			    dwords += prim[i].count + 10;
+		    else
+			    dwords += ((prim[i].count + 1) / 2) + 10;
+	    }
     }
+
     state_size = radeonCountStateEmitSize(&context->radeon);
     flushed = rcommonEnsureCmdBufSpace(&context->radeon,
-            dwords + state_size, __FUNCTION__);
-
+				       dwords + state_size,
+				       __FUNCTION__);
     if (flushed)
-        dwords += radeonCountStateEmitSize(&context->radeon);
+	    dwords += radeonCountStateEmitSize(&context->radeon);
     else
-        dwords += state_size;
+	    dwords += state_size;
 
-    radeon_print(RADEON_RENDER, RADEON_VERBOSE,
-	"%s: total prediction size is %d.\n", __FUNCTION__, dwords);
+    radeon_print(RADEON_RENDER, RADEON_VERBOSE, "%s: total prediction size is %d.\n", __FUNCTION__, dwords);
     return dwords;
-}
-
-static GLboolean r700RunRender(GLcontext * ctx,
-			       struct tnl_pipeline_stage *stage)
-{
-    context_t *context = R700_CONTEXT(ctx);
-    radeonContextPtr radeon = &context->radeon;
-    unsigned int i, id = 0;
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
-    struct vertex_buffer *vb = &tnl->vb;
-    struct radeon_renderbuffer *rrb;
-
-    radeon_print(RADEON_RENDER, RADEON_NORMAL, "%s: cs begin at %d\n",
-                __func__, context->radeon.cmdbuf.cs->cdw);
-
-    /* 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, 0)
-        + context->radeon.cmdbuf.cs->cdw;
-    r700SetupStreams(ctx);
-
-    radeonEmitState(radeon);
-
-    radeon_debug_add_indent();
-    /* richard test code */
-    for (i = 0; i < vb->PrimitiveCount; i++) {
-        GLuint prim = _tnl_translate_prim(&vb->Primitive[i]);
-        GLuint start = vb->Primitive[i].start;
-        GLuint end = vb->Primitive[i].start + vb->Primitive[i].count;
-        r700RunRenderPrimitive(ctx, start, end, prim);
-    }
-    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);
-
-    radeonReleaseArrays(ctx, ~0);
-
-    radeon_print(RADEON_RENDER, RADEON_TRACE, "%s: cs end at %d\n",
-                __func__, context->radeon.cmdbuf.cs->cdw);
-
-    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_FALSE;
-}
-
-static GLboolean r700RunNonTCLRender(GLcontext * ctx,
-				     struct tnl_pipeline_stage *stage) /* -------------------- */
-{
-	GLboolean bRet = GL_TRUE;
-	
-	return bRet;
-}
-
-static GLboolean r700RunTCLRender(GLcontext * ctx,  /*----------------------*/
-				  struct tnl_pipeline_stage *stage)
-{
-	GLboolean bRet = GL_FALSE;
 
-    /* TODO : sw fallback */
-
-    /* Need shader bo's setup before bo check */
-    r700UpdateShaders(ctx);
-    /**
-
-    * Ensure all enabled and complete textures are uploaded along with any buffers being used.
-    */
-    if(!r600ValidateBuffers(ctx))
-    {
-        return GL_TRUE;
-    }
-
-    bRet = r700RunRender(ctx, stage);
-
-    return bRet;
-	//GL_FALSE will stop to do other pipe stage in _tnl_run_pipeline
-    //The render here DOES finish the whole pipe, so GL_FALSE should be returned for success.
 }
 
-const struct tnl_pipeline_stage _r700_render_stage = {
-	"r700 Hardware Rasterization",
-	NULL,
-	NULL,
-	NULL,
-	NULL,
-	r700RunNonTCLRender
-};
-
-const struct tnl_pipeline_stage _r700_tcl_stage = {
-	"r700 Hardware Transform, Clipping and Lighting",
-	NULL,
-	NULL,
-	NULL,
-	NULL,
-	r700RunTCLRender
-};
-
-const struct tnl_pipeline_stage *r700_pipeline[] = 
-{
-    &_r700_tcl_stage,
-    &_tnl_vertex_transform_stage,
-	&_tnl_normal_transform_stage,
-	&_tnl_lighting_stage,
-	&_tnl_fog_coordinate_stage,
-	&_tnl_texgen_stage,
-	&_tnl_texture_transform_stage,
-	&_tnl_vertex_program_stage,
-
-    &_r700_render_stage,
-    &_tnl_render_stage,
-    0,
-};
-
 #define CONVERT( TYPE, MACRO ) do {		\
 	GLuint i, j, sz;				\
 	sz = input->Size;				\
@@ -605,6 +526,9 @@ static void r700ConvertAttrib(GLcontext *ctx, int count,
 
     radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, 
                          sizeof(GLfloat) * input->Size * count, 32);
+
+    radeon_bo_map(attr->bo, 1);
+
     dst_ptr = (GLfloat *)ADD_POINTERS(attr->bo->ptr, attr->bo_offset);
 
     assert(src_ptr != NULL);
@@ -638,6 +562,8 @@ static void r700ConvertAttrib(GLcontext *ctx, int count,
             break;
     }
 
+    radeon_bo_unmap(attr->bo);
+
     if (mapped_named_bo) 
     {
         ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
@@ -656,6 +582,8 @@ static void r700AlignDataToDword(GLcontext *ctx,
 
     radeonAllocDmaRegion(&context->radeon, &attr->bo, &attr->bo_offset, size, 32);
 
+    radeon_bo_map(attr->bo, 1);
+
     if (!input->BufferObj->Pointer) 
     {
         ctx->Driver.MapBuffer(ctx, GL_ARRAY_BUFFER, GL_READ_ONLY_ARB, input->BufferObj);
@@ -675,6 +603,7 @@ static void r700AlignDataToDword(GLcontext *ctx,
         }
     }
 
+    radeon_bo_unmap(attr->bo);
     if (mapped_named_bo) 
     {
         ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER, input->BufferObj);
@@ -683,7 +612,7 @@ static void r700AlignDataToDword(GLcontext *ctx,
     attr->stride = dst_stride;
 }
 
-static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *input[], int count)
+static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input[], int count)
 {
 	context_t *context = R700_CONTEXT(ctx);
     GLuint stride;
@@ -743,32 +672,33 @@ static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *inpu
 
                 radeonAllocDmaRegion(&context->radeon, &context->stream_desc[index].bo, 
                                      &context->stream_desc[index].bo_offset, size, 32);
+
+                radeon_bo_map(context->stream_desc[index].bo, 1);
                 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; 
+                    radeonEmitVec4(dst, input[i]->Ptr, input[i]->StrideB, local_count);
                     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;
                 }
+		radeon_bo_unmap(context->stream_desc[index].bo);
             }
         }
 
@@ -786,7 +716,6 @@ static void r700SetupStreams2(GLcontext *ctx, const struct gl_client_array *inpu
         }
     }
 
-    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);    
@@ -841,6 +770,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
 
+	radeon_bo_map(context->ind_buf.bo, 1);
 	assert(context->ind_buf.bo->ptr != NULL);
 	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
 
@@ -854,6 +784,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
             *out++ = in[i];
         }
 
+	radeon_bo_unmap(context->ind_buf.bo);
 #if MESA_BIG_ENDIAN
     }
     else
@@ -864,6 +795,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
 
+	radeon_bo_map(context->ind_buf.bo, 1);
 	assert(context->ind_buf.bo->ptr != NULL);
 	out = (GLuint *)ADD_POINTERS(context->ind_buf.bo->ptr, context->ind_buf.bo_offset);
 
@@ -876,6 +808,7 @@ static void r700FixupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
         {
             *out++ = in[i];
         }
+	radeon_bo_unmap(context->ind_buf.bo);
 #endif
     }
 
@@ -921,11 +854,13 @@ static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 
 	radeonAllocDmaRegion(&context->radeon, &context->ind_buf.bo,
 			     &context->ind_buf.bo_offset, size, 4);
+	radeon_bo_map(context->ind_buf.bo, 1);
 	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);
 
+	radeon_bo_unmap(context->ind_buf.bo);
         context->ind_buf.is_32bit = (mesa_ind_buf->type == GL_UNSIGNED_INT);
         context->ind_buf.count = mesa_ind_buf->count;
 
@@ -941,12 +876,12 @@ static void r700SetupIndexBuffer(GLcontext *ctx, const struct _mesa_index_buffer
 }
 
 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 )
+				  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;
@@ -954,14 +889,12 @@ static GLboolean r700TryDrawPrims(GLcontext *ctx,
     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);
+    r700UpdateShaders(ctx);
     if (!r600ValidateBuffers(ctx))
 	    return GL_FALSE;
 
@@ -975,23 +908,29 @@ static GLboolean r700TryDrawPrims(GLcontext *ctx,
     r700SetScissor(context);
     r700SetupVertexProgram(ctx);
     r700SetupFragmentProgram(ctx);
-    r600UpdateTextureState(ctx);
+    r700UpdateShaderStates(ctx);
 
-    GLuint emit_end = r700PredictRenderSize(ctx, nr_prims)
+    GLuint emit_end = r700PredictRenderSize(ctx, prim, ib, nr_prims)
                     + context->radeon.cmdbuf.cs->cdw;
 
     r700SetupIndexBuffer(ctx, ib);
-    r700SetupStreams2(ctx, arrays, max_index + 1);
+    r700SetupStreams(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);
+	    if (context->ind_buf.bo)
+		    r700RunRenderPrimitive(ctx,
+					   prim[i].start,
+					   prim[i].start + prim[i].count,
+					   prim[i].mode);
+	    else
+		    r700RunRenderPrimitiveImmediate(ctx,
+						    prim[i].start,
+						    prim[i].start + prim[i].count,
+						    prim[i].mode);
     }
     radeon_debug_remove_indent();
 
@@ -1019,18 +958,18 @@ static GLboolean r700TryDrawPrims(GLcontext *ctx,
     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)
+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)
 {
-    GLboolean retval = GL_FALSE;
+	GLboolean retval = GL_FALSE;
 
-    /* This check should get folded into just the places that
+	/* This check should get folded into just the places that
 	 * min/max index are really needed.
 	 */
 	if (!index_bounds_valid) {
@@ -1038,7 +977,7 @@ static void r700DrawPrimsRe(GLcontext *ctx,
 	}
 
 	if (min_index) {
-		vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrimsRe );
+		vbo_rebase_prims( ctx, arrays, prim, nr_prims, ib, min_index, max_index, r700DrawPrims );
 		return;
 	}
 
@@ -1050,30 +989,6 @@ static void r700DrawPrimsRe(GLcontext *ctx,
 		_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);
diff --git a/src/mesa/drivers/dri/r600/r700_shader.c b/src/mesa/drivers/dri/r600/r700_shader.c
index 955ea4e4e1..2eed1acc2f 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.c
+++ b/src/mesa/drivers/dri/r600/r700_shader.c
@@ -159,13 +159,18 @@ void Init_R700_Shader(R700_Shader * pShader)
 	pShader->lstVTXInstructions.uNumOfNode=0;
 }
 
+void SetActiveCFlist(R700_Shader *pShader, TypedShaderList * plstCF)
+{
+    pShader->plstCFInstructions_active = plstCF;
+}
+
 void AddCFInstruction(R700_Shader *pShader, R700ControlFlowInstruction *pCFInst)
 {
     R700ControlFlowSXClause*  pSXClause; 
     R700ControlFlowSMXClause* pSMXClause;
 
-    pCFInst->m_uIndex = pShader->lstCFInstructions.uNumOfNode;
-    AddInstToList(&(pShader->lstCFInstructions), 
+    pCFInst->m_uIndex = pShader->plstCFInstructions_active->uNumOfNode;
+    AddInstToList(pShader->plstCFInstructions_active, 
                   (R700ShaderInstruction*)pCFInst);
     pShader->uShaderBinaryDWORDSize += GetInstructionSize(pCFInst->m_ShaderInstType);
 
diff --git a/src/mesa/drivers/dri/r600/r700_shader.h b/src/mesa/drivers/dri/r600/r700_shader.h
index 997cb05aaf..0599ffd901 100644
--- a/src/mesa/drivers/dri/r600/r700_shader.h
+++ b/src/mesa/drivers/dri/r600/r700_shader.h
@@ -109,6 +109,7 @@ typedef struct R700_Shader
     GLuint  uStackSize;
     GLuint  uMaxCallDepth;
 
+    TypedShaderList * plstCFInstructions_active;
 	TypedShaderList lstCFInstructions;
 	TypedShaderList lstALUInstructions;
 	TypedShaderList lstTEXInstructions;
@@ -128,16 +129,17 @@ typedef struct R700_Shader
 
 //Internal
 void AddInstToList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst);
+void TakeInstOutFromList(TypedShaderList * plstCFInstructions, R700ShaderInstruction * pInst);
 void ResolveLinks(R700_Shader *pShader);
 void Assemble(R700_Shader *pShader);
 
-
 //Interface
 void Init_R700_Shader(R700_Shader * pShader);
 void AddCFInstruction(R700_Shader *pShader, R700ControlFlowInstruction *pCFInst);
 void AddVTXInstruction(R700_Shader *pShader, R700VertexInstruction *pVTXInst);
 void AddTEXInstruction(R700_Shader *pShader, R700TextureInstruction *pTEXInst);
 void AddALUInstruction(R700_Shader *pShader, R700ALUInstruction *pALUInst);
+void SetActiveCFlist(R700_Shader *pShader, TypedShaderList * plstCF);
 
 void LoadProgram(R700_Shader *pShader);
 void UpdateShaderRegisters(R700_Shader *pShader);
diff --git a/src/mesa/drivers/dri/r600/r700_shaderinst.h b/src/mesa/drivers/dri/r600/r700_shaderinst.h
index 2829cca0a3..cdb9a570f7 100644
--- a/src/mesa/drivers/dri/r600/r700_shaderinst.h
+++ b/src/mesa/drivers/dri/r600/r700_shaderinst.h
@@ -42,6 +42,13 @@
 #define SQ_FETCH_RESOURCE_VS_OFFSET    0x000000a0
 #define SQ_FETCH_RESOURCE_VS_COUNT     0x000000b0
 
+//richard dec.10 glsl
+#define SQ_TEX_SAMPLER_PS_OFFSET       0x00000000
+#define SQ_TEX_SAMPLER_PS_COUNT        0x00000012
+#define SQ_TEX_SAMPLER_VS_OFFSET       0x00000012
+#define SQ_TEX_SAMPLER_VS_COUNT        0x00000012
+//-------------------
+
 #define SHADERINST_TYPEMASK_CF  0x10
 #define SHADERINST_TYPEMASK_ALU 0x20
 #define SHADERINST_TYPEMASK_TEX 0x40
diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c
index 7e8b48f91e..3c8cb579f9 100644
--- a/src/mesa/drivers/dri/r600/r700_state.c
+++ b/src/mesa/drivers/dri/r600/r700_state.c
@@ -46,7 +46,6 @@
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
 #include "vbo/vbo.h"
-#include "main/texformat.h"
 
 #include "r600_context.h"
 
@@ -55,18 +54,15 @@
 #include "r700_fragprog.h"
 #include "r700_vertprog.h"
 
-
+void r600UpdateTextureState(GLcontext * ctx);
 static void r700SetClipPlaneState(GLcontext * ctx, GLenum cap, GLboolean state);
 static void r700UpdatePolygonMode(GLcontext * ctx);
 static void r700SetPolygonOffsetState(GLcontext * ctx, GLboolean state);
 static void r700SetStencilState(GLcontext * ctx, GLboolean state);
 
-void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
+void r700UpdateShaders(GLcontext * ctx)
 {
     context_t *context = R700_CONTEXT(ctx);
-    GLvector4f dummy_attrib[_TNL_ATTRIB_MAX];
-    GLvector4f *temp_attrib[_TNL_ATTRIB_MAX];
-    int i;
 
     /* should only happenen once, just after context is created */
     /* TODO: shouldn't we fallback to sw here? */
@@ -77,40 +73,7 @@ void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
 
     r700SelectFragmentShader(ctx);
 
-    if (context->radeon.NewGLState) {
-	    for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {
-		    /* mat states from state var not array for sw */
-		    dummy_attrib[i].stride = 0;
-	            temp_attrib[i] = TNL_CONTEXT(ctx)->vb.AttribPtr[i];
-		    TNL_CONTEXT(ctx)->vb.AttribPtr[i] = &(dummy_attrib[i]);
-	    }
-
-	    _tnl_UpdateFixedFunctionProgram(ctx);
-
-	    for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {
-		    TNL_CONTEXT(ctx)->vb.AttribPtr[i] = temp_attrib[i];
-	    }
-    }
-
-    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);
+    r700SelectVertexShader(ctx);
     r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS);
     context->radeon.NewGLState = 0;
 }
@@ -122,7 +85,7 @@ void r700UpdateViewportOffset(GLcontext * ctx) //------------------
 {
 	context_t *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&context->radeon);
+	__DRIdrawable *dPriv = radeon_get_drawable(&context->radeon);
 	GLfloat xoffset = (GLfloat) dPriv->x;
 	GLfloat yoffset = (GLfloat) dPriv->y + dPriv->h;
 	const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -228,6 +191,67 @@ static void r700InvalidateState(GLcontext * ctx, GLuint new_state) //-----------
     context->radeon.NewGLState |= new_state;
 }
 
+static void r700SetDBRenderState(GLcontext * ctx)
+{
+	context_t *context = R700_CONTEXT(ctx);
+	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+	struct r700_fragment_program *fp = (struct r700_fragment_program *)
+		(ctx->FragmentProgram._Current);
+
+	R600_STATECHANGE(context, db);
+
+	SETbit(r700->DB_SHADER_CONTROL.u32All, DUAL_EXPORT_ENABLE_bit);
+	SETfield(r700->DB_SHADER_CONTROL.u32All, EARLY_Z_THEN_LATE_Z, Z_ORDER_shift, Z_ORDER_mask);
+	/* XXX need to enable htile for hiz/s */
+	SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIZ_ENABLE_shift, FORCE_HIZ_ENABLE_mask);
+	SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE0_shift, FORCE_HIS_ENABLE0_mask);
+	SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask);
+
+	if (context->radeon.query.current)
+	{
+		SETbit(r700->DB_RENDER_OVERRIDE.u32All, NOOP_CULL_DISABLE_bit);
+		if (context->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV770)
+		{
+			SETbit(r700->DB_RENDER_CONTROL.u32All, PERFECT_ZPASS_COUNTS_bit);
+		}
+	}
+	else
+	{
+		CLEARbit(r700->DB_RENDER_OVERRIDE.u32All, NOOP_CULL_DISABLE_bit);
+		if (context->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV770)
+		{
+			CLEARbit(r700->DB_RENDER_CONTROL.u32All, PERFECT_ZPASS_COUNTS_bit);
+		}
+	}
+
+	if (fp)
+	{
+		if (fp->r700Shader.killIsUsed)
+		{
+			SETbit(r700->DB_SHADER_CONTROL.u32All, KILL_ENABLE_bit);
+		}
+		else
+		{
+			CLEARbit(r700->DB_SHADER_CONTROL.u32All, KILL_ENABLE_bit);
+		}
+
+		if (fp->r700Shader.depthIsExported)
+		{
+			SETbit(r700->DB_SHADER_CONTROL.u32All, Z_EXPORT_ENABLE_bit);
+		}
+		else
+		{
+			CLEARbit(r700->DB_SHADER_CONTROL.u32All, Z_EXPORT_ENABLE_bit);
+		}
+	}
+}
+
+void r700UpdateShaderStates(GLcontext * ctx)
+{
+	r700SetDBRenderState(ctx);
+	r600UpdateTextureState(ctx);
+}
+
 static void r700SetDepthState(GLcontext * ctx)
 {
 	context_t *context = R700_CONTEXT(ctx);
@@ -681,6 +705,10 @@ static void r700UpdateCulling(GLcontext * ctx)
             CLEARbit(r700->PA_SU_SC_MODE_CNTL.u32All, FACE_bit); /* default: ccw */
             break;
     }
+
+    /* Winding is inverted when rendering to FBO */
+    if (ctx->DrawBuffer && ctx->DrawBuffer->Name)
+	    r700->PA_SU_SC_MODE_CNTL.u32All ^= FACE_bit;
 }
 
 static void r700UpdateLineStipple(GLcontext * ctx)
@@ -1043,7 +1071,7 @@ static void r700UpdateWindow(GLcontext * ctx, int id) //--------------------
 {
 	context_t *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(&context->radeon);
+	__DRIdrawable *dPriv = radeon_get_drawable(&context->radeon);
 	GLfloat xoffset = dPriv ? (GLfloat) dPriv->x : 0;
 	GLfloat yoffset = dPriv ? (GLfloat) dPriv->y + dPriv->h : 0;
 	const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1067,6 +1095,7 @@ static void r700UpdateWindow(GLcontext * ctx, int id) //--------------------
 	GLfloat tz = v[MAT_TZ] * depthScale;
 
 	R600_STATECHANGE(context, vpt);
+	R600_STATECHANGE(context, cl);
 
 	r700->viewport[id].PA_CL_VPORT_XSCALE.f32All  = sx;
 	r700->viewport[id].PA_CL_VPORT_XOFFSET.f32All = tx;
@@ -1077,6 +1106,18 @@ static void r700UpdateWindow(GLcontext * ctx, int id) //--------------------
 	r700->viewport[id].PA_CL_VPORT_ZSCALE.f32All  = sz;
 	r700->viewport[id].PA_CL_VPORT_ZOFFSET.f32All = tz;
 
+	if (ctx->Transform.DepthClamp) {
+		r700->viewport[id].PA_SC_VPORT_ZMIN_0.f32All = MIN2(ctx->Viewport.Near, ctx->Viewport.Far);
+		r700->viewport[id].PA_SC_VPORT_ZMAX_0.f32All = MAX2(ctx->Viewport.Near, ctx->Viewport.Far);
+		SETbit(r700->PA_CL_CLIP_CNTL.u32All, ZCLIP_NEAR_DISABLE_bit);
+		SETbit(r700->PA_CL_CLIP_CNTL.u32All, ZCLIP_FAR_DISABLE_bit);
+	} else {
+		r700->viewport[id].PA_SC_VPORT_ZMIN_0.f32All = 0.0;
+		r700->viewport[id].PA_SC_VPORT_ZMAX_0.f32All = 1.0;
+		CLEARbit(r700->PA_CL_CLIP_CNTL.u32All, ZCLIP_NEAR_DISABLE_bit);
+		CLEARbit(r700->PA_CL_CLIP_CNTL.u32All, ZCLIP_FAR_DISABLE_bit);
+	}
+
 	r700->viewport[id].enabled = GL_TRUE;
 
 	r700SetScissor(context);
@@ -1190,13 +1231,8 @@ static void r700UpdatePolygonMode(GLcontext * ctx)
 		/* Handle GL_CW (clock wise and GL_CCW (counter clock wise)
 		 * correctly by selecting the correct front and back face
 		 */
-		if (ctx->Polygon.FrontFace == GL_CCW) {
-			f = ctx->Polygon.FrontMode;
-			b = ctx->Polygon.BackMode;
-		} else {
-			f = ctx->Polygon.BackMode;
-			b = ctx->Polygon.FrontMode;
-		}
+		f = ctx->Polygon.FrontMode;
+		b = ctx->Polygon.BackMode;
 
 		/* Enable polygon mode */
 		SETfield(r700->PA_SU_SC_MODE_CNTL.u32All, X_DUAL_MODE, POLY_MODE_shift, POLY_MODE_mask);
@@ -1295,11 +1331,15 @@ void r700SetScissor(context_t *context) //---------------
 		return;
 	}
 	if (context->radeon.state.scissor.enabled) {
-		/* r600 has exclusive scissors */
 		x1 = context->radeon.state.scissor.rect.x1;
 		y1 = context->radeon.state.scissor.rect.y1;
-		x2 = context->radeon.state.scissor.rect.x2 + 1;
-		y2 = context->radeon.state.scissor.rect.y2 + 1;
+		x2 = context->radeon.state.scissor.rect.x2;
+		y2 = context->radeon.state.scissor.rect.y2;
+		/* r600 has exclusive BR scissors */
+		if (context->radeon.radeonScreen->kernel_mm) {
+			x2++;
+			y2++;
+		}
 	} else {
 		if (context->radeon.radeonScreen->driScreen->dri2.enabled) {
 			x1 = 0;
@@ -1378,8 +1418,6 @@ void r700SetScissor(context_t *context) //---------------
 	SETfield(r700->viewport[id].PA_SC_VPORT_SCISSOR_0_BR.u32All, y2,
 		 PA_SC_VPORT_SCISSOR_0_BR__BR_Y_shift, PA_SC_VPORT_SCISSOR_0_BR__BR_Y_mask);
 
-	r700->viewport[id].PA_SC_VPORT_ZMIN_0.u32All = 0;
-	r700->viewport[id].PA_SC_VPORT_ZMAX_0.u32All = 0x3F800000;
 	r700->viewport[id].enabled = GL_TRUE;
 }
 
@@ -1686,27 +1724,18 @@ void r700InitState(GLcontext * ctx) //-------------------
     r700InitSQConfig(ctx);
 
     r700ColorMask(ctx,
-		  ctx->Color.ColorMask[RCOMP],
-		  ctx->Color.ColorMask[GCOMP],
-		  ctx->Color.ColorMask[BCOMP],
-		  ctx->Color.ColorMask[ACOMP]);
+		  ctx->Color.ColorMask[0][RCOMP],
+		  ctx->Color.ColorMask[0][GCOMP],
+		  ctx->Color.ColorMask[0][BCOMP],
+		  ctx->Color.ColorMask[0][ACOMP]);
 
     r700Enable(ctx, GL_DEPTH_TEST, ctx->Depth.Test);
     r700DepthMask(ctx, ctx->Depth.Mask);
     r700DepthFunc(ctx, ctx->Depth.Func);
-    SETbit(r700->DB_SHADER_CONTROL.u32All, DUAL_EXPORT_ENABLE_bit);
-
     r700->DB_DEPTH_CLEAR.u32All     = 0x3F800000;
-
-    r700->DB_RENDER_CONTROL.u32All  = 0;
     SETbit(r700->DB_RENDER_CONTROL.u32All, STENCIL_COMPRESS_DISABLE_bit);
     SETbit(r700->DB_RENDER_CONTROL.u32All, DEPTH_COMPRESS_DISABLE_bit);
-    r700->DB_RENDER_OVERRIDE.u32All = 0;
-    if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
-	    SETbit(r700->DB_RENDER_OVERRIDE.u32All, FORCE_SHADER_Z_ORDER_bit);
-    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIZ_ENABLE_shift, FORCE_HIZ_ENABLE_mask);
-    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE0_shift, FORCE_HIS_ENABLE0_mask);
-    SETfield(r700->DB_RENDER_OVERRIDE.u32All, FORCE_DISABLE, FORCE_HIS_ENABLE1_shift, FORCE_HIS_ENABLE1_mask);
+    r700SetDBRenderState(ctx);
 
     r700->DB_ALPHA_TO_MASK.u32All = 0;
     SETfield(r700->DB_ALPHA_TO_MASK.u32All, 2, ALPHA_TO_MASK_OFFSET0_shift, ALPHA_TO_MASK_OFFSET0_mask);
diff --git a/src/mesa/drivers/dri/r600/r700_state.h b/src/mesa/drivers/dri/r600/r700_state.h
index 209189d8d7..60c6a7f23c 100644
--- a/src/mesa/drivers/dri/r600/r700_state.h
+++ b/src/mesa/drivers/dri/r600/r700_state.h
@@ -35,7 +35,7 @@
 
 extern void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state);
 extern void r700UpdateShaders (GLcontext * ctx);
-extern void r700UpdateShaders2(GLcontext * ctx);
+extern void r700UpdateShaderStates(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 d12c39c9f7..782f151f5a 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -111,6 +111,15 @@ unsigned int Map_Vertex_Output(r700_AssemblerBase       *pAsm,
 		}
 	}
 
+    for(i=VERT_RESULT_VAR0; i<VERT_RESULT_MAX; i++)
+	{
+		unBit = 1 << i;
+		if(mesa_vp->Base.OutputsWritten & unBit)
+		{
+			pAsm->ucVP_OutputMap[i] = unTotal++;
+		}
+	}
+
 	return (unTotal - unStart);
 }
 
@@ -179,7 +188,8 @@ GLboolean Process_Vertex_Program_Vfetch_Instructions2(
                                       context->stream_desc[i].size,
                                       context->stream_desc[i].element,
                                       context->stream_desc[i]._signed,
-                                      context->stream_desc[i].normalize,						            
+                                      context->stream_desc[i].normalize,
+                                      context->stream_desc[i].format,
                                      &vtxFetchMethod);
     }
 
@@ -203,22 +213,11 @@ void Map_Vertex_Program(GLcontext *ctx,
 	pAsm->number_used_registers += num_inputs;
 
 	// Create VFETCH instructions for inputs
-	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; 
-	    }
-    }
+        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);
@@ -228,7 +227,7 @@ void Map_Vertex_Program(GLcontext *ctx,
 	pAsm->number_used_registers += pAsm->number_of_exports;
 
     pAsm->pucOutMask = (unsigned char*) MALLOC(pAsm->number_of_exports);
-    
+
     for(ui=0; ui<pAsm->number_of_exports; ui++)
     {
         pAsm->pucOutMask[ui] = 0x0;
@@ -245,7 +244,9 @@ void Map_Vertex_Program(GLcontext *ctx,
     {   /* fix func t_vp uses NumTemporaries */
         pAsm->number_used_registers += mesa_vp->Base.NumTemporaries;
     }
-	
+
+    pAsm->flag_reg_index = pAsm->number_used_registers++;
+
     pAsm->uFirstHelpReg = pAsm->number_used_registers;
 }
 
@@ -300,18 +301,13 @@ GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp,
 }
 
 struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
-						struct gl_vertex_program *mesa_vp,
-                        GLint nVer)
+						      struct gl_vertex_program *mesa_vp)
 {
 	context_t *context = R700_CONTEXT(ctx);
 	struct r700_vertex_program *vp;
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *vb = &tnl->vb;
-	unsigned int unBit;
 	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)
@@ -319,29 +315,14 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
                 _mesa_insert_mvp_code(ctx, vp->mesa_program);
         }
 
-	if( 1 == nVer )
+	for(i=0; i<context->nNumActiveAos; i++)
 	{
-	    for(i=0; i<VERT_ATTRIB_MAX; i++)
-	    {
-		unBit = 1 << i;
-		if(vp->mesa_program->Base.InputsRead & unBit) /* ctx->Array.ArrayObj->xxxxxxx */
-		{
-			vp->aos_desc[i].size   = vb->AttribPtr[i]->size;
-			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;
-		}
-	    }
+		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;
+		vp->aos_desc[i].format = context->stream_desc[i].format;
 	}
-	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)
 	{
 		vp->r700AsmCode.bR6xx = 1;
@@ -354,20 +335,36 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 	if(GL_FALSE == Find_Instruction_Dependencies_vp(vp, vp->mesa_program))
 	{
 		return NULL;
+	}
+
+    InitShaderProgram(&(vp->r700AsmCode));
+
+    for(i=0; i < MAX_SAMPLERS; i++)
+    {
+        vp->r700AsmCode.SamplerUnits[i] = vp->mesa_program->Base.SamplerUnits[i];
     }
 
-	if(GL_FALSE == AssembleInstr(vp->mesa_program->Base.NumInstructions,
-                                 &(vp->mesa_program->Base.Instructions[0]), 
+    vp->r700AsmCode.unCurNumILInsts = vp->mesa_program->Base.NumInstructions;
+
+	if(GL_FALSE == AssembleInstr(0,
+                                 0,
+                                 vp->mesa_program->Base.NumInstructions,
+                                 &(vp->mesa_program->Base.Instructions[0]),
                                  &(vp->r700AsmCode)) )
 	{
 		return NULL;
-	} 
+	}
 
     if(GL_FALSE == Process_Vertex_Exports(&(vp->r700AsmCode), vp->mesa_program->Base.OutputsWritten) )
     {
         return NULL;
     }
 
+    if( GL_FALSE == RelocProgram(&(vp->r700AsmCode), &(vp->mesa_program->Base)) )
+    {
+        return GL_FALSE;
+    }
+
     vp->r700Shader.nRegs = (vp->r700AsmCode.number_used_registers == 0) ? 0 
                          : (vp->r700AsmCode.number_used_registers - 1);
 
@@ -378,14 +375,11 @@ struct r700_vertex_program* r700TranslateVertexShader(GLcontext *ctx,
 	return vp;
 }
 
-void r700SelectVertexShader(GLcontext *ctx, GLint nVersion)
+void r700SelectVertexShader(GLcontext *ctx)
 {
     context_t *context = R700_CONTEXT(ctx);
     struct r700_vertex_program_cont *vpc;
     struct r700_vertex_program *vp;
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
-    struct vertex_buffer *vb = &tnl->vb;
-    unsigned int unBit;
     unsigned int i;
     GLboolean match;
     GLbitfield InputsRead;
@@ -396,47 +390,28 @@ void r700SelectVertexShader(GLcontext *ctx, GLint nVersion)
     if (vpc->mesa_program.IsPositionInvariant)
     {
 	InputsRead |= VERT_BIT_POS;
-    } 
-    
+    }
+
     for (vp = vpc->progs; vp; vp = vp->next)
     {
-	if (vp->uiVersion != nVersion ) 
-	    continue;
-	match = GL_TRUE;	
-	if ( 1 == nVersion ) 
+	match = GL_TRUE;
+	for(i=0; i<context->nNumActiveAos; i++)
 	{
-	    for(i=0; i<VERT_ATTRIB_MAX; i++)
-	    {
-		unBit = 1 << i;
-		if(InputsRead & unBit)
+		if (vp->aos_desc[i].size != context->stream_desc[i].size ||
+		    vp->aos_desc[i].format != context->stream_desc[i].format)
 		{
-		    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) 
+	if (match)
 	{
 		context->selected_vp = vp;
 		return;
 	}
     }
 
-    vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program), nVersion);
+    vp = r700TranslateVertexShader(ctx, &(vpc->mesa_program));
     if(!vp)
     {
 	radeon_error("Failed to translate vertex shader. \n");
@@ -526,6 +501,7 @@ static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const s
 	pStreamDesc->size = input->Size;
 	pStreamDesc->dst_loc = context->nNumActiveAos;
 	pStreamDesc->element = unLoc;
+	pStreamDesc->format = input->Format;
 
 	switch (pStreamDesc->type) 
 	{ //GetSurfaceFormat
@@ -570,6 +546,11 @@ void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[],
     unsigned int unBit = mesa_vp->Base.InputsRead;
     context->nNumActiveAos = 0;
 
+    if (mesa_vp->IsPositionInvariant)
+    {
+        unBit |= VERT_BIT_POS;
+    }
+
     while(unBit) 
     {
         if(unBit & 1)
@@ -580,6 +561,7 @@ void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[],
         unBit >>= 1;
         ++unLoc;
     }
+    context->radeon.tcl.aos_count = context->nNumActiveAos;
 }
 
 void * r700GetActiveVpShaderBo(GLcontext * ctx)
@@ -661,6 +643,12 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
     paramList = vp->mesa_program->Base.Parameters;
 
     if(NULL != paramList) {
+        /* vp->mesa_program was cloned, not updated by glsl shader api. */
+        /* _mesa_reference_program has already checked glsl shProg is ok and set ctx->VertexProgem._Current */
+        /* so, use ctx->VertexProgem._Current */       
+        struct gl_program_parameter_list *paramListOrginal = 
+                         paramListOrginal = ctx->VertexProgram._Current->Base.Parameters;
+         
 	    _mesa_load_state_parameters(ctx, paramList);
 
 	    if (paramList->NumParameters > R700_MAX_DX9_CONSTS)
@@ -673,13 +661,42 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
 	    unNumParamData = paramList->NumParameters;
 
 	    for(ui=0; ui<unNumParamData; ui++) {
-		    r700->vs.consts[ui][0].f32All = paramList->ParameterValues[ui][0];
-		    r700->vs.consts[ui][1].f32All = paramList->ParameterValues[ui][1];
-		    r700->vs.consts[ui][2].f32All = paramList->ParameterValues[ui][2];
-		    r700->vs.consts[ui][3].f32All = paramList->ParameterValues[ui][3];
+            if(paramList->Parameters[ui].Type == PROGRAM_UNIFORM) 
+            {
+                r700->vs.consts[ui][0].f32All = paramListOrginal->ParameterValues[ui][0];
+		        r700->vs.consts[ui][1].f32All = paramListOrginal->ParameterValues[ui][1];
+		        r700->vs.consts[ui][2].f32All = paramListOrginal->ParameterValues[ui][2];
+		        r700->vs.consts[ui][3].f32All = paramListOrginal->ParameterValues[ui][3];
+            }
+            else
+            {
+		        r700->vs.consts[ui][0].f32All = paramList->ParameterValues[ui][0];
+		        r700->vs.consts[ui][1].f32All = paramList->ParameterValues[ui][1];
+		        r700->vs.consts[ui][2].f32All = paramList->ParameterValues[ui][2];
+		        r700->vs.consts[ui][3].f32All = paramList->ParameterValues[ui][3];
+            }
 	    }
     } else
 	    r700->vs.num_consts = 0;
 
+    COMPILED_SUB * pCompiledSub;
+    GLuint uj;
+    GLuint unConstOffset = r700->vs.num_consts;
+    for(ui=0; ui<vp->r700AsmCode.unNumPresub; ui++)
+    {
+        pCompiledSub = vp->r700AsmCode.presubs[ui].pCompiledSub;
+
+        r700->vs.num_consts += pCompiledSub->NumParameters;
+
+        for(uj=0; uj<pCompiledSub->NumParameters; uj++)
+        {
+            r700->vs.consts[uj + unConstOffset][0].f32All = pCompiledSub->ParameterValues[uj][0];
+		    r700->vs.consts[uj + unConstOffset][1].f32All = pCompiledSub->ParameterValues[uj][1];
+		    r700->vs.consts[uj + unConstOffset][2].f32All = pCompiledSub->ParameterValues[uj][2];
+		    r700->vs.consts[uj + unConstOffset][3].f32All = pCompiledSub->ParameterValues[uj][3];
+        }
+        unConstOffset += pCompiledSub->NumParameters;
+    }
+
     return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h
index f9a3e395ee..645c9ac84a 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.h
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.h
@@ -39,6 +39,7 @@ typedef struct ArrayDesc //TEMP
 	GLint size;   //number of data element
 	GLenum type;  //data element type
 	GLsizei stride;
+	GLenum format; //GL_RGBA or GL_BGRA
 } ArrayDesc;
 
 struct r700_vertex_program 
@@ -52,8 +53,7 @@ struct r700_vertex_program
 
     GLboolean translated;
     GLboolean loaded;
-    GLint     uiVersion;
-	
+
     void * shaderbo;
 
     ArrayDesc              aos_desc[VERT_ATTRIB_MAX];
@@ -87,11 +87,10 @@ 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,
-                    GLint nVer);
+						      struct gl_vertex_program   *mesa_vp);
 
 /* Interface */
-extern void r700SelectVertexShader(GLcontext *ctx, GLint nVersion);
+extern void r700SelectVertexShader(GLcontext *ctx);
 extern void r700SetVertexFormat(GLcontext *ctx, const struct gl_client_array *arrays[], int count);
 
 extern GLboolean r700SetupVertexProgram(GLcontext * ctx);
diff --git a/src/mesa/drivers/dri/r600/radeon_bo.c b/src/mesa/drivers/dri/r600/radeon_bo.c
new file mode 120000
index 0000000000..9448ffee54
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_bo.c
@@ -0,0 +1 @@
+../radeon/radeon_bo.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r600/radeon_bo_int_drm.h b/src/mesa/drivers/dri/r600/radeon_bo_int_drm.h
new file mode 120000
index 0000000000..029450928b
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_bo_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_bo_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r600/radeon_cs.c b/src/mesa/drivers/dri/r600/radeon_cs.c
new file mode 120000
index 0000000000..66b7ad1eb0
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_cs.c
@@ -0,0 +1 @@
+../radeon/radeon_cs.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/r600/radeon_cs_int_drm.h b/src/mesa/drivers/dri/r600/radeon_cs_int_drm.h
new file mode 120000
index 0000000000..462f5245d0
--- /dev/null
+++ b/src/mesa/drivers/dri/r600/radeon_cs_int_drm.h
@@ -0,0 +1 @@
+../radeon/radeon_cs_int_drm.h
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/radeon/Makefile b/src/mesa/drivers/dri/radeon/Makefile
index b1efc72872..2b2f2c4aa7 100644
--- a/src/mesa/drivers/dri/radeon/Makefile
+++ b/src/mesa/drivers/dri/radeon/Makefile
@@ -11,7 +11,7 @@ LIBNAME = radeon_dri.so
 MINIGLX_SOURCES = server/radeon_dri.c 
 
 ifeq ($(RADEON_LDFLAGS),)
-CS_SOURCES = radeon_cs_space_drm.c
+CS_SOURCES = radeon_cs_space_drm.c radeon_bo.c radeon_cs.c
 endif
 
 RADEON_COMMON_SOURCES = \
@@ -55,4 +55,3 @@ X86_SOURCES =
 
 include ../Makefile.template
 
-symlinks:
diff --git a/src/mesa/drivers/dri/radeon/radeon_bo.c b/src/mesa/drivers/dri/radeon/radeon_bo.c
new file mode 100644
index 0000000000..393d156cde
--- /dev/null
+++ b/src/mesa/drivers/dri/radeon/radeon_bo.c
@@ -0,0 +1,110 @@
+#include <radeon_bocs_wrapper.h>
+#include <radeon_bo_int_drm.h>
+
+void radeon_bo_debug(struct radeon_bo *bo,
+		     const char *op)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+
+    fprintf(stderr, "%s %p 0x%08X 0x%08X 0x%08X\n",
+            op, bo, bo->handle, boi->size, boi->cref);
+}
+
+struct radeon_bo *radeon_bo_open(struct radeon_bo_manager *bom,
+				 uint32_t handle,
+				 uint32_t size,
+				 uint32_t alignment,
+				 uint32_t domains,
+				 uint32_t flags)
+{
+    struct radeon_bo *bo;
+    bo = bom->funcs->bo_open(bom, handle, size, alignment, domains, flags);
+    return bo;
+}
+
+void radeon_bo_ref(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    boi->cref++;
+    boi->bom->funcs->bo_ref(boi);
+}
+
+struct radeon_bo *radeon_bo_unref(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    boi->cref--;
+    return boi->bom->funcs->bo_unref(boi);
+}
+
+int radeon_bo_map(struct radeon_bo *bo, int write)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->bom->funcs->bo_map(boi, write);
+}
+
+int radeon_bo_unmap(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->bom->funcs->bo_unmap(boi);
+}
+
+int radeon_bo_wait(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    if (!boi->bom->funcs->bo_wait)
+	return 0;
+    return boi->bom->funcs->bo_wait(boi);
+}
+
+int radeon_bo_is_busy(struct radeon_bo *bo,
+		      uint32_t *domain)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->bom->funcs->bo_is_busy(boi, domain);
+}
+
+int radeon_bo_set_tiling(struct radeon_bo *bo,
+			 uint32_t tiling_flags, uint32_t pitch)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->bom->funcs->bo_set_tiling(boi, tiling_flags, pitch);
+}
+
+int radeon_bo_get_tiling(struct radeon_bo *bo,
+			  uint32_t *tiling_flags, uint32_t *pitch)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->bom->funcs->bo_get_tiling(boi, tiling_flags, pitch);
+}
+
+int radeon_bo_is_static(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    if (boi->bom->funcs->bo_is_static)
+	return boi->bom->funcs->bo_is_static(boi);
+    return 0;
+}
+
+int radeon_bo_is_referenced_by_cs(struct radeon_bo *bo,
+				  struct radeon_cs *cs)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    return boi->cref > 1;
+}
+
+uint32_t radeon_bo_get_handle(struct radeon_bo *bo)
+{
+    return bo->handle;
+}
+
+uint32_t radeon_bo_get_src_domain(struct radeon_bo *bo)
+{
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    uint32_t src_domain;
+
+    src_domain = boi->space_accounted & 0xffff;
+    if (!src_domain)
+	src_domain = boi->space_accounted >> 16;
+
+    return src_domain;
+}
diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
index 7141371633..beb2369880 100644
--- a/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
+++ b/src/mesa/drivers/dri/radeon/radeon_bo_drm.h
@@ -32,188 +32,44 @@
 
 #include <stdio.h>
 #include <stdint.h>
-//#include "radeon_track.h"
 
 /* bo object */
 #define RADEON_BO_FLAGS_MACRO_TILE  1
 #define RADEON_BO_FLAGS_MICRO_TILE  2
 
 struct radeon_bo_manager;
+struct radeon_cs;
 
 struct radeon_bo {
-    uint32_t                    alignment;
+    void                        *ptr;
+    uint32_t                    flags;
     uint32_t                    handle;
     uint32_t                    size;
-    uint32_t                    domains;
-    uint32_t                    flags;
-    unsigned                    cref;
-#ifdef RADEON_BO_TRACK
-    struct radeon_track         *track;
-#endif
-    void                        *ptr;
-    struct radeon_bo_manager    *bom;
-    uint32_t                    space_accounted;
-};
-
-/* bo functions */
-struct radeon_bo_funcs {
-    struct radeon_bo *(*bo_open)(struct radeon_bo_manager *bom,
-                                 uint32_t handle,
-                                 uint32_t size,
-                                 uint32_t alignment,
-                                 uint32_t domains,
-                                 uint32_t flags);
-    void (*bo_ref)(struct radeon_bo *bo);
-    struct radeon_bo *(*bo_unref)(struct radeon_bo *bo);
-    int (*bo_map)(struct radeon_bo *bo, int write);
-    int (*bo_unmap)(struct radeon_bo *bo);
-    int (*bo_wait)(struct radeon_bo *bo);
-    int (*bo_is_static)(struct radeon_bo *bo);
-    int (*bo_set_tiling)(struct radeon_bo *bo, uint32_t tiling_flags,
-			  uint32_t pitch);
-    int (*bo_get_tiling)(struct radeon_bo *bo, uint32_t *tiling_flags,
-			  uint32_t *pitch);
-    int (*bo_is_busy)(struct radeon_bo *bo, uint32_t *domain);
 };
 
-struct radeon_bo_manager {
-    struct radeon_bo_funcs  *funcs;
-    int                     fd;
-
-#ifdef RADEON_BO_TRACK
-    struct radeon_tracker   tracker;
-#endif
-};
-    
-static inline void _radeon_bo_debug(struct radeon_bo *bo,
-                                    const char *op,
-                                    const char *file,
-                                    const char *func,
-                                    int line)
-{
-    fprintf(stderr, "%s %p 0x%08X 0x%08X 0x%08X [%s %s %d]\n",
-            op, bo, bo->handle, bo->size, bo->cref, file, func, line);
-}
-
-static inline struct radeon_bo *_radeon_bo_open(struct radeon_bo_manager *bom,
-                                                uint32_t handle,
-                                                uint32_t size,
-                                                uint32_t alignment,
-                                                uint32_t domains,
-                                                uint32_t flags,
-                                                const char *file,
-                                                const char *func,
-                                                int line)
-{
-    struct radeon_bo *bo;
-
-    bo = bom->funcs->bo_open(bom, handle, size, alignment, domains, flags);
-
-#ifdef RADEON_BO_TRACK
-    if (bo) {
-        bo->track = radeon_tracker_add_track(&bom->tracker, bo->handle);
-        radeon_track_add_event(bo->track, file, func, "open", line);
-    }
-#endif
-    return bo;
-}
-
-static inline void _radeon_bo_ref(struct radeon_bo *bo,
-                                  const char *file,
-                                  const char *func,
-                                  int line)
-{
-    bo->cref++;
-#ifdef RADEON_BO_TRACK
-    radeon_track_add_event(bo->track, file, func, "ref", line); 
-#endif
-    bo->bom->funcs->bo_ref(bo);
-}
-
-static inline struct radeon_bo *_radeon_bo_unref(struct radeon_bo *bo,
-                                                 const char *file,
-                                                 const char *func,
-                                                 int line)
-{
-    bo->cref--;
-#ifdef RADEON_BO_TRACK
-    radeon_track_add_event(bo->track, file, func, "unref", line);
-    if (bo->cref <= 0) {
-        radeon_tracker_remove_track(&bo->bom->tracker, bo->track);
-        bo->track = NULL;
-    }
-#endif
-    return bo->bom->funcs->bo_unref(bo);
-}
-
-static inline int _radeon_bo_map(struct radeon_bo *bo,
-                                 int write,
-                                 const char *file,
-                                 const char *func,
-                                 int line)
-{
-    return bo->bom->funcs->bo_map(bo, write);
-}
-
-static inline int _radeon_bo_unmap(struct radeon_bo *bo,
-                                   const char *file,
-                                   const char *func,
-                                   int line)
-{
-    return bo->bom->funcs->bo_unmap(bo);
-}
-
-static inline int _radeon_bo_wait(struct radeon_bo *bo,
-                                  const char *file,
-                                  const char *func,
-                                  int line)
-{
-    return bo->bom->funcs->bo_wait(bo);
-}
-
-static inline int _radeon_bo_is_busy(struct radeon_bo *bo,
-				     uint32_t *domain,
-                                     const char *file,
-                                     const char *func,
-                                     int line)
-{
-    return bo->bom->funcs->bo_is_busy(bo, domain);
-}
-
-static inline int radeon_bo_set_tiling(struct radeon_bo *bo,
-				       uint32_t tiling_flags, uint32_t pitch)
-{
-    return bo->bom->funcs->bo_set_tiling(bo, tiling_flags, pitch);
-}
-
-static inline int radeon_bo_get_tiling(struct radeon_bo *bo,
-				       uint32_t *tiling_flags, uint32_t *pitch)
-{
-    return bo->bom->funcs->bo_get_tiling(bo, tiling_flags, pitch);
-}
-
-static inline int radeon_bo_is_static(struct radeon_bo *bo)
-{
-	if (bo->bom->funcs->bo_is_static)
-		return bo->bom->funcs->bo_is_static(bo);
-	return 0;
-}
-
-#define radeon_bo_open(bom, h, s, a, d, f)\
-    _radeon_bo_open(bom, h, s, a, d, f, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_ref(bo)\
-    _radeon_bo_ref(bo, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_unref(bo)\
-    _radeon_bo_unref(bo, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_map(bo, w)\
-    _radeon_bo_map(bo, w, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_unmap(bo)\
-    _radeon_bo_unmap(bo, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_debug(bo, opcode)\
-    _radeon_bo_debug(bo, opcode, __FILE__, __FUNCTION__, __LINE__)
-#define radeon_bo_wait(bo) \
-    _radeon_bo_wait(bo, __FILE__, __func__, __LINE__)
-#define radeon_bo_is_busy(bo, domain) \
-    _radeon_bo_is_busy(bo, domain, __FILE__, __func__, __LINE__)
+struct radeon_bo_manager;
 
+void radeon_bo_debug(struct radeon_bo *bo,
+		     const char *op);
+
+struct radeon_bo *radeon_bo_open(struct radeon_bo_manager *bom,
+				  uint32_t handle,
+				  uint32_t size,
+				  uint32_t alignment,
+				  uint32_t domains,
+				  uint32_t flags);
+
+void radeon_bo_ref(struct radeon_bo *bo);
+struct radeon_bo *radeon_bo_unref(struct radeon_bo *bo);
+int radeon_bo_map(struct radeon_bo *bo, int write);
+int radeon_bo_unmap(struct radeon_bo *bo);
+int radeon_bo_wait(struct radeon_bo *bo);
+int radeon_bo_is_busy(struct radeon_bo *bo, uint32_t *domain);
+int radeon_bo_set_tiling(struct radeon_bo *bo, uint32_t tiling_flags, uint32_t pitch);
+int radeon_bo_get_tiling(struct radeon_bo *bo, uint32_t *tiling_flags, uint32_t *pitch);
+int radeon_bo_is_static(struct radeon_bo *bo);
+int radeon_bo_is_referenced_by_cs(struct radeon_bo *bo,
+				  struct radeon_cs *cs);
+uint32_t radeon_bo_get_handle(struct radeon_bo *bo);
+uint32_t radeon_bo_get_src_domain(struct radeon_bo *bo);
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_int_drm.h b/src/mesa/drivers/dri/radeon/radeon_bo_int_drm.h
new file mode 100644
index 0000000000..190c332475
--- /dev/null
+++ b/src/mesa/drivers/dri/radeon/radeon_bo_int_drm.h
@@ -0,0 +1,45 @@
+#ifndef RADEON_BO_INT
+#define RADEON_BO_INT
+
+struct radeon_bo_manager {
+    struct radeon_bo_funcs  *funcs;
+    int                     fd;
+};
+
+struct radeon_bo_int {
+    void                        *ptr;
+    uint32_t                    flags;
+    uint32_t                    handle;
+    uint32_t                    size;
+    /* private members */
+    uint32_t                    alignment;
+    uint32_t                    domains;
+    unsigned                    cref;
+    struct radeon_bo_manager    *bom;
+    uint32_t                    space_accounted;
+    uint32_t                    referenced_in_cs;
+};
+
+/* bo functions */
+struct radeon_bo_funcs {
+    struct radeon_bo *(*bo_open)(struct radeon_bo_manager *bom,
+                                 uint32_t handle,
+                                 uint32_t size,
+                                 uint32_t alignment,
+                                 uint32_t domains,
+                                 uint32_t flags);
+    void (*bo_ref)(struct radeon_bo_int *bo);
+    struct radeon_bo *(*bo_unref)(struct radeon_bo_int *bo);
+    int (*bo_map)(struct radeon_bo_int *bo, int write);
+    int (*bo_unmap)(struct radeon_bo_int *bo);
+    int (*bo_wait)(struct radeon_bo_int *bo);
+    int (*bo_is_static)(struct radeon_bo_int *bo);
+    int (*bo_set_tiling)(struct radeon_bo_int *bo, uint32_t tiling_flags,
+			  uint32_t pitch);
+    int (*bo_get_tiling)(struct radeon_bo_int *bo, uint32_t *tiling_flags,
+			  uint32_t *pitch);
+    int (*bo_is_busy)(struct radeon_bo_int *bo, uint32_t *domain);
+    int (*bo_is_referenced_by_cs)(struct radeon_bo_int *bo, struct radeon_cs *cs);
+};
+
+#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c
index 3e7547d2f9..cf12664bac 100644
--- a/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c
+++ b/src/mesa/drivers/dri/radeon/radeon_bo_legacy.c
@@ -50,6 +50,12 @@
 #include "radeon_bocs_wrapper.h"
 #include "radeon_macros.h"
 
+#ifdef HAVE_LIBDRM_RADEON
+#include "radeon_bo_int.h"
+#else
+#include "radeon_bo_int_drm.h"
+#endif
+
 /* no seriously texmem.c is this screwed up */
 struct bo_legacy_texture_object {
     driTextureObject    base;
@@ -57,7 +63,7 @@ struct bo_legacy_texture_object {
 };
 
 struct bo_legacy {
-    struct radeon_bo    base;
+    struct radeon_bo_int    base;
     int                 map_count;
     uint32_t            pending;
     int                 is_pending;
@@ -187,10 +193,10 @@ static void legacy_get_current_age(struct bo_manager_legacy *boml)
     }
 }
 
-static int legacy_is_pending(struct radeon_bo *bo)
+static int legacy_is_pending(struct radeon_bo_int *boi)
 {
-    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
-    struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
+    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)boi->bom;
+    struct bo_legacy *bo_legacy = (struct bo_legacy*)boi;
 
     if (bo_legacy->is_pending <= 0) {
         bo_legacy->is_pending = 0;
@@ -204,13 +210,13 @@ static int legacy_is_pending(struct radeon_bo *bo)
         if (bo_legacy->pnext) {
             bo_legacy->pnext->pprev = bo_legacy->pprev;
         }
-	assert(bo_legacy->is_pending <= bo->cref);
+	assert(bo_legacy->is_pending <= boi->cref);
         while (bo_legacy->is_pending--) {
-	    bo = radeon_bo_unref(bo);
-	    if (!bo)
+	    boi = (struct radeon_bo_int *)radeon_bo_unref((struct radeon_bo *)boi);
+	    if (!boi)
 	      break;
         }
-	if (bo)
+	if (boi)
 	  bo_legacy->is_pending = 0;
         boml->cpendings--;
         return 0;
@@ -218,7 +224,7 @@ static int legacy_is_pending(struct radeon_bo *bo)
     return 1;
 }
 
-static int legacy_wait_pending(struct radeon_bo *bo)
+static int legacy_wait_pending(struct radeon_bo_int *bo)
 {
     struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
@@ -323,7 +329,7 @@ static struct bo_legacy *bo_allocate(struct bo_manager_legacy *boml,
     return bo_legacy;
 }
 
-static int bo_dma_alloc(struct radeon_bo *bo)
+static int bo_dma_alloc(struct radeon_bo_int *bo)
 {
     struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
@@ -333,7 +339,7 @@ static int bo_dma_alloc(struct radeon_bo *bo)
     int r;
 
     /* align size on 4Kb */
-    size = (((4 * 1024) - 1) + bo->size) & ~((4 * 1024) - 1);
+    size = (((4 * 1024) - 1) + bo_legacy->base.size) & ~((4 * 1024) - 1);
     alloc.region = RADEON_MEM_REGION_GART;
     alloc.alignment = bo_legacy->base.alignment;
     alloc.size = size;
@@ -355,7 +361,7 @@ static int bo_dma_alloc(struct radeon_bo *bo)
     return 0;
 }
 
-static int bo_dma_free(struct radeon_bo *bo)
+static int bo_dma_free(struct radeon_bo_int *bo)
 {
     struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
@@ -428,7 +434,7 @@ static struct radeon_bo *bo_open(struct radeon_bo_manager *bom,
         bo_legacy = boml->bos.next;
         while (bo_legacy) {
             if (bo_legacy->base.handle == handle) {
-                radeon_bo_ref(&(bo_legacy->base));
+                radeon_bo_ref((struct radeon_bo *)&(bo_legacy->base));
                 return (struct radeon_bo*)bo_legacy;
             }
             bo_legacy = bo_legacy->next;
@@ -468,20 +474,20 @@ retry:
             return NULL;
         }
     }
-    radeon_bo_ref(&(bo_legacy->base));
+    radeon_bo_ref((struct radeon_bo *)&(bo_legacy->base));
 
     return (struct radeon_bo*)bo_legacy;
 }
 
-static void bo_ref(struct radeon_bo *bo)
+static void bo_ref(struct radeon_bo_int *bo)
 {
 }
 
-static struct radeon_bo *bo_unref(struct radeon_bo *bo)
+static struct radeon_bo *bo_unref(struct radeon_bo_int *boi)
 {
-    struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
+    struct bo_legacy *bo_legacy = (struct bo_legacy*)boi;
 
-    if (bo->cref <= 0) {
+    if (boi->cref <= 0) {
         bo_legacy->prev->next = bo_legacy->next;
         if (bo_legacy->next) {
             bo_legacy->next->prev = bo_legacy->prev;
@@ -491,10 +497,10 @@ static struct radeon_bo *bo_unref(struct radeon_bo *bo)
         }
         return NULL;
     }
-    return bo;
+    return (struct radeon_bo *)boi;
 }
 
-static int bo_map(struct radeon_bo *bo, int write)
+static int bo_map(struct radeon_bo_int *bo, int write)
 {
     struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
@@ -528,7 +534,7 @@ static int bo_map(struct radeon_bo *bo, int write)
     return 0;
 }
 
-static int bo_unmap(struct radeon_bo *bo)
+static int bo_unmap(struct radeon_bo_int *bo)
 {
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
 
@@ -542,7 +548,7 @@ static int bo_unmap(struct radeon_bo *bo)
     return 0;
 }
 
-static int bo_is_busy(struct radeon_bo *bo, uint32_t *domain)
+static int bo_is_busy(struct radeon_bo_int *bo, uint32_t *domain)
 {
     *domain = 0;
     if (bo->domains & RADEON_GEM_DOMAIN_GTT)
@@ -555,7 +561,7 @@ static int bo_is_busy(struct radeon_bo *bo, uint32_t *domain)
         return 0;
 }
 
-static int bo_is_static(struct radeon_bo *bo)
+static int bo_is_static(struct radeon_bo_int *bo)
 {
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
     return bo_legacy->static_bo;
@@ -574,7 +580,7 @@ static struct radeon_bo_funcs bo_legacy_funcs = {
     bo_is_busy
 };
 
-static int bo_vram_validate(struct radeon_bo *bo,
+static int bo_vram_validate(struct radeon_bo_int *bo,
                             uint32_t *soffset,
                             uint32_t *eoffset)
 {
@@ -700,25 +706,30 @@ int radeon_bo_legacy_validate(struct radeon_bo *bo,
                               uint32_t *soffset,
                               uint32_t *eoffset)
 {
-    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)boi->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
     int r;
     int retries = 0;
 
     if (bo_legacy->map_count) {
         fprintf(stderr, "bo(%p, %d) is mapped (%d) can't valide it.\n",
-                bo, bo->size, bo_legacy->map_count);
+                bo, boi->size, bo_legacy->map_count);
+        return -EINVAL;
+    }
+    if(boi->size == 0) {
+        fprintf(stderr, "bo(%p) has size 0.\n", bo);
         return -EINVAL;
     }
     if (bo_legacy->static_bo || bo_legacy->validated) {
         *soffset = bo_legacy->offset;
-        *eoffset = bo_legacy->offset + bo->size;
+        *eoffset = bo_legacy->offset + boi->size;
 
         return 0;
     }
-    if (!(bo->domains & RADEON_GEM_DOMAIN_GTT)) {
+    if (!(boi->domains & RADEON_GEM_DOMAIN_GTT)) {
 
-        r = bo_vram_validate(bo, soffset, eoffset);
+        r = bo_vram_validate(boi, soffset, eoffset);
         if (r) {
 	    legacy_track_pending(&boml->base, 0);
 	    legacy_kick_all_buffers(boml);
@@ -732,7 +743,7 @@ int radeon_bo_legacy_validate(struct radeon_bo *bo,
         }
     }
     *soffset = bo_legacy->offset;
-    *eoffset = bo_legacy->offset + bo->size;
+    *eoffset = bo_legacy->offset + boi->size;
     bo_legacy->validated = 1;
 
     return 0;
@@ -740,7 +751,8 @@ int radeon_bo_legacy_validate(struct radeon_bo *bo,
 
 void radeon_bo_legacy_pending(struct radeon_bo *bo, uint32_t pending)
 {
-    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)bo->bom;
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
+    struct bo_manager_legacy *boml = (struct bo_manager_legacy *)boi->bom;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
 
     bo_legacy->pending = pending;
@@ -795,7 +807,7 @@ static struct bo_legacy *radeon_legacy_bo_alloc_static(struct bo_manager_legacy
     if (bo->base.handle > bom->nhandle) {
         bom->nhandle = bo->base.handle + 1;
     }
-    radeon_bo_ref(&(bo->base));
+    radeon_bo_ref((struct radeon_bo *)&(bo->base));
     return bo;
 }
 
@@ -890,12 +902,13 @@ void radeon_bo_legacy_texture_age(struct radeon_bo_manager *bom)
 
 unsigned radeon_bo_legacy_relocs_size(struct radeon_bo *bo)
 {
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
     struct bo_legacy *bo_legacy = (struct bo_legacy*)bo;
 
-    if (bo_legacy->static_bo || (bo->domains & RADEON_GEM_DOMAIN_GTT)) {
+    if (bo_legacy->static_bo || (boi->domains & RADEON_GEM_DOMAIN_GTT)) {
         return 0;
     }
-    return bo->size;
+    return boi->size;
 }
 
 /*
@@ -920,7 +933,7 @@ struct radeon_bo *radeon_legacy_bo_alloc_fake(struct radeon_bo_manager *bom,
     if (bo->base.handle > boml->nhandle) {
         boml->nhandle = bo->base.handle + 1;
     }
-    radeon_bo_ref(&(bo->base));
-    return &(bo->base);
+    radeon_bo_ref((struct radeon_bo *)&(bo->base));
+    return (struct radeon_bo *)&(bo->base);
 }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
index 4520a7d7d4..6c2648b6bd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
+++ b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
@@ -18,8 +18,11 @@
 #define RADEON_TILING_MACRO 0x1
 #define RADEON_TILING_MICRO 0x2
 #define RADEON_TILING_SWAP 0x4
+
+#ifndef RADEON_TILING_SURFACE
 #define RADEON_TILING_SURFACE 0x8 /* this object requires a surface
 				   * when mapped - i.e. front buffer */
+#endif
 
 /* to be used to build locally in mesa with no libdrm bits */
 #include "../radeon/radeon_bo_drm.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
index 8fac5c6c51..99d3ec7005 100644
--- a/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
+++ b/src/mesa/drivers/dri/radeon/radeon_buffer_objects.c
@@ -136,8 +136,13 @@ radeonBufferSubData(GLcontext * ctx,
                     const GLvoid * data,
                     struct gl_buffer_object *obj)
 {
+    radeonContextPtr radeon = RADEON_CONTEXT(ctx);
     struct radeon_buffer_object *radeon_obj = get_radeon_buffer_object(obj);
 
+    if (radeon_bo_is_referenced_by_cs(radeon_obj->bo, radeon->cmdbuf.cs)) {
+        radeon_firevertices(radeon);
+    }
+
     radeon_bo_map(radeon_obj->bo, GL_TRUE);
 
     _mesa_memcpy(radeon_obj->bo->ptr + offset, data, size);
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c
index f8a4cdb495..e0b853bc97 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common.c
@@ -137,7 +137,7 @@ void radeon_get_cliprects(radeonContextPtr radeon,
 			  unsigned int *num_cliprects,
 			  int *x_off, int *y_off)
 {
-	__DRIdrawablePrivate *dPriv = radeon_get_drawable(radeon);
+	__DRIdrawable *dPriv = radeon_get_drawable(radeon);
 	struct radeon_framebuffer *rfb = dPriv->driverPrivate;
 
 	if (radeon->constant_cliprect) {
@@ -169,8 +169,8 @@ void radeon_get_cliprects(radeonContextPtr radeon,
  */
 void radeonSetCliprects(radeonContextPtr radeon)
 {
-	__DRIdrawablePrivate *const drawable = radeon_get_drawable(radeon);
-	__DRIdrawablePrivate *const readable = radeon_get_readable(radeon);
+	__DRIdrawable *const drawable = radeon_get_drawable(radeon);
+	__DRIdrawable *const readable = radeon_get_readable(radeon);
 	struct radeon_framebuffer *const draw_rfb = drawable->driverPrivate;
 	struct radeon_framebuffer *const read_rfb = readable->driverPrivate;
 	int x_off, y_off;
@@ -229,16 +229,15 @@ void radeonUpdateScissor( GLcontext *ctx )
 	}
 	if (!rmesa->radeonScreen->kernel_mm) {
 	   /* Fix scissors for dri 1 */
-
-	   __DRIdrawablePrivate *dPriv = radeon_get_drawable(rmesa);
+	   __DRIdrawable *dPriv = radeon_get_drawable(rmesa);
 	   x1 += dPriv->x;
-	   x2 += dPriv->x;
+	   x2 += dPriv->x + 1;
 	   min_x += dPriv->x;
-	   max_x += dPriv->x;
+	   max_x += dPriv->x + 1;
 	   y1 += dPriv->y;
-	   y2 += dPriv->y;
+	   y2 += dPriv->y + 1;
 	   min_y += dPriv->y;
-	   max_y += dPriv->y;
+	   max_y += dPriv->y + 1;
 	}
 
 	rmesa->state.scissor.rect.x1 = CLAMP(x1,  min_x, max_x);
@@ -263,29 +262,6 @@ void radeonScissor(GLcontext* ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 	}
 }
 
-void radeonPolygonStipplePreKMS( GLcontext *ctx, const GLubyte *mask )
-{
-   radeonContextPtr radeon = RADEON_CONTEXT(ctx);
-   GLuint i;
-   drm_radeon_stipple_t stipple;
-
-   /* Must flip pattern upside down.
-   */
-   for ( i = 0 ; i < 32 ; i++ ) {
-      stipple.mask[31 - i] = ((GLuint *) mask)[i];
-   }
-
-   /* TODO: push this into cmd mechanism
-   */
-   radeon_firevertices(radeon);
-   LOCK_HARDWARE( radeon );
-
-   drmCommandWrite( radeon->dri.fd, DRM_RADEON_STIPPLE,
-	 &stipple, sizeof(stipple) );
-   UNLOCK_HARDWARE( radeon );
-}
-
-
 /* ================================================================
  * SwapBuffers with client-side throttling
  */
@@ -452,7 +428,7 @@ static void radeon_flip_renderbuffers(struct radeon_framebuffer *rfb)
 
 /* Copy the back color buffer to the front color buffer.
  */
-void radeonCopyBuffer( __DRIdrawablePrivate *dPriv,
+void radeonCopyBuffer( __DRIdrawable *dPriv,
 		       const drm_clip_rect_t	  *rect)
 {
 	radeonContextPtr rmesa;
@@ -520,7 +496,7 @@ void radeonCopyBuffer( __DRIdrawablePrivate *dPriv,
 	UNLOCK_HARDWARE( rmesa );
 }
 
-static int radeonScheduleSwap(__DRIdrawablePrivate *dPriv, GLboolean *missed_target)
+static int radeonScheduleSwap(__DRIdrawable *dPriv, GLboolean *missed_target)
 {
 	radeonContextPtr rmesa;
 
@@ -543,11 +519,11 @@ static int radeonScheduleSwap(__DRIdrawablePrivate *dPriv, GLboolean *missed_tar
 	return 0;
 }
 
-static GLboolean radeonPageFlip( __DRIdrawablePrivate *dPriv )
+static GLboolean radeonPageFlip( __DRIdrawable *dPriv )
 {
 	radeonContextPtr radeon;
 	GLint ret;
-	__DRIscreenPrivate *psp;
+	__DRIscreen *psp;
 	struct radeon_renderbuffer *rrb;
 	struct radeon_framebuffer *rfb;
 
@@ -595,10 +571,10 @@ static GLboolean radeonPageFlip( __DRIdrawablePrivate *dPriv )
 /**
  * Swap front and back buffer.
  */
-void radeonSwapBuffers(__DRIdrawablePrivate * dPriv)
+void radeonSwapBuffers(__DRIdrawable * dPriv)
 {
 	int64_t ust;
-	__DRIscreenPrivate *psp;
+	__DRIscreen *psp;
 
 	if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
 		radeonContextPtr radeon;
@@ -639,7 +615,7 @@ void radeonSwapBuffers(__DRIdrawablePrivate * dPriv)
 	}
 }
 
-void radeonCopySubBuffer(__DRIdrawablePrivate * dPriv,
+void radeonCopySubBuffer(__DRIdrawable * dPriv,
 			 int x, int y, int w, int h )
 {
 	if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
@@ -665,6 +641,27 @@ void radeonCopySubBuffer(__DRIdrawablePrivate * dPriv,
 	}
 }
 
+/**
+ * Check if we're about to draw into the front color buffer.
+ * If so, set the intel->front_buffer_dirty field to true.
+ */
+void
+radeon_check_front_buffer_rendering(GLcontext *ctx)
+{
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+	const struct gl_framebuffer *fb = ctx->DrawBuffer;
+
+	if (fb->Name == 0) {
+		/* drawing to window system buffer */
+		if (fb->_NumColorDrawBuffers > 0) {
+			if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
+				radeon->front_buffer_dirty = GL_TRUE;
+			}
+		}
+	}
+}
+
+
 void radeon_draw_buffer(GLcontext *ctx, struct gl_framebuffer *fb)
 {
 	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
@@ -841,7 +838,7 @@ void radeonDrawBuffer( GLcontext *ctx, GLenum mode )
        */
 		if (!was_front_buffer_rendering && radeon->is_front_buffer_rendering) {
 			radeon_update_renderbuffers(radeon->dri.context,
-				radeon->dri.context->driDrawablePriv);
+				radeon->dri.context->driDrawablePriv, GL_FALSE);
       }
 	}
 
@@ -858,7 +855,7 @@ void radeonReadBuffer( GLcontext *ctx, GLenum mode )
 
 		if (!was_front_buffer_reading && rmesa->is_front_buffer_reading) {
 			radeon_update_renderbuffers(rmesa->dri.context,
-						    rmesa->dri.context->driReadablePriv);
+						    rmesa->dri.context->driReadablePriv, GL_FALSE);
 	 	}
 	}
 	/* nothing, until we implement h/w glRead/CopyPixels or CopyTexImage */
@@ -909,9 +906,9 @@ void radeon_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei width, GLsizei he
 		if (radeon->is_front_buffer_rendering) {
 			ctx->Driver.Flush(ctx);
 		}
-		radeon_update_renderbuffers(driContext, driContext->driDrawablePriv);
+		radeon_update_renderbuffers(driContext, driContext->driDrawablePriv, GL_FALSE);
 		if (driContext->driDrawablePriv != driContext->driReadablePriv)
-			radeon_update_renderbuffers(driContext, driContext->driReadablePriv);
+			radeon_update_renderbuffers(driContext, driContext->driReadablePriv, GL_FALSE);
 	}
 
 	old_viewport = ctx->Driver.Viewport;
@@ -1119,22 +1116,21 @@ void radeonFlush(GLcontext *ctx)
 	   then no point flushing anything at all.
 	*/
 	if (!radeon->dma.flush && !radeon->cmdbuf.cs->cdw && is_empty_list(&radeon->dma.reserved))
-		return;
+		goto flush_front;
 
 	if (radeon->dma.flush)
 		radeon->dma.flush( ctx );
 
-	radeonEmitState(radeon);
-
 	if (radeon->cmdbuf.cs->cdw)
 		rcommonFlushCmdBuf(radeon, __FUNCTION__);
 
+flush_front:
 	if ((ctx->DrawBuffer->Name == 0) && radeon->front_buffer_dirty) {
 		__DRIscreen *const screen = radeon->radeonScreen->driScreen;
 
 		if (screen->dri2.loader && (screen->dri2.loader->base.version >= 2)
 			&& (screen->dri2.loader->flushFrontBuffer != NULL)) {
-			__DRIdrawablePrivate * drawable = radeon_get_drawable(radeon);
+			__DRIdrawable * drawable = radeon_get_drawable(radeon);
 			(*screen->dri2.loader->flushFrontBuffer)(drawable, drawable->loaderPrivate);
 
 			/* Only clear the dirty bit if front-buffer rendering is no longer
@@ -1148,9 +1144,6 @@ void radeonFlush(GLcontext *ctx)
 			}
 		}
 	}
-
-	make_empty_list(&radeon->query.not_flushed_head);
-
 }
 
 /* Make sure all commands have been sent to the hardware and have
@@ -1237,7 +1230,7 @@ int rcommonFlushCmdBuf(radeonContextPtr rmesa, const char *caller)
 		fprintf(stderr, "drmRadeonCmdBuffer: %d. Kernel failed to "
 				"parse or rejected command stream. See dmesg "
 				"for more info.\n", ret);
-		_mesa_exit(ret);
+		exit(ret);
 	}
 
 	return ret;
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.h b/src/mesa/drivers/dri/radeon/radeon_common.h
index f3201911ac..f31f08edf3 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.h
+++ b/src/mesa/drivers/dri/radeon/radeon_common.h
@@ -10,14 +10,13 @@ void radeonRecalcScissorRects(radeonContextPtr radeon);
 void radeonSetCliprects(radeonContextPtr radeon);
 void radeonUpdateScissor( GLcontext *ctx );
 void radeonScissor(GLcontext* ctx, GLint x, GLint y, GLsizei w, GLsizei h);
-void radeonPolygonStipplePreKMS( GLcontext *ctx, const GLubyte *mask );
 
 void radeonWaitForIdleLocked(radeonContextPtr radeon);
 extern uint32_t radeonGetAge(radeonContextPtr radeon);
-void radeonCopyBuffer( __DRIdrawablePrivate *dPriv,
+void radeonCopyBuffer( __DRIdrawable *dPriv,
 		       const drm_clip_rect_t	  *rect);
-void radeonSwapBuffers(__DRIdrawablePrivate * dPriv);
-void radeonCopySubBuffer(__DRIdrawablePrivate * dPriv,
+void radeonSwapBuffers(__DRIdrawable * dPriv);
+void radeonCopySubBuffer(__DRIdrawable * dPriv,
 			 int x, int y, int w, int h );
 
 void radeonUpdatePageFlipping(radeonContextPtr rmesa);
@@ -43,7 +42,9 @@ void
 radeon_renderbuffer_set_bo(struct radeon_renderbuffer *rb,
 			   struct radeon_bo *bo);
 struct radeon_renderbuffer *
-radeon_create_renderbuffer(GLenum format, __DRIdrawablePrivate *driDrawPriv);
+radeon_create_renderbuffer(gl_format format, __DRIdrawable *driDrawPriv);
+
+void radeon_check_front_buffer_rendering(GLcontext *ctx);
 static inline struct radeon_renderbuffer *radeon_renderbuffer(struct gl_renderbuffer *rb)
 {
 	struct radeon_renderbuffer *rrb = (struct radeon_renderbuffer *)rb;
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c
index 6b9b1e3c5e..b9c29b937e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c
@@ -181,10 +181,10 @@ static void radeonInitDriverFuncs(struct dd_function_table *functions)
 GLboolean radeonInitContext(radeonContextPtr radeon,
 			    struct dd_function_table* functions,
 			    const __GLcontextModes * glVisual,
-			    __DRIcontextPrivate * driContextPriv,
+			    __DRIcontext * driContextPriv,
 			    void *sharedContextPrivate)
 {
-	__DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+	__DRIscreen *sPriv = driContextPriv->driScreenPriv;
 	radeonScreenPtr screen = (radeonScreenPtr) (sPriv->private);
 	GLcontext* ctx;
 	GLcontext* shareCtx;
@@ -262,10 +262,9 @@ GLboolean radeonInitContext(radeonContextPtr radeon,
 		else
 			radeon->texture_row_align = 32;
 		radeon->texture_rect_row_align = 64;
-		radeon->texture_compressed_row_align = 64;
+		radeon->texture_compressed_row_align = 32;
 	}
 
-	make_empty_list(&radeon->query.not_flushed_head);
 	radeon_init_dma(radeon);
 
 	return GL_TRUE;
@@ -292,7 +291,7 @@ static void radeon_destroy_atom_list(radeonContextPtr radeon)
  * Cleanup common context fields.
  * Called by r200DestroyContext/r300DestroyContext
  */
-void radeonDestroyContext(__DRIcontextPrivate *driContextPriv )
+void radeonDestroyContext(__DRIcontext *driContextPriv )
 {
 #ifdef RADEON_BO_TRACK
 	FILE *track;
@@ -356,7 +355,7 @@ void radeonDestroyContext(__DRIcontextPrivate *driContextPriv )
 
 /* Force the context `c' to be unbound from its buffer.
  */
-GLboolean radeonUnbindContext(__DRIcontextPrivate * driContextPriv)
+GLboolean radeonUnbindContext(__DRIcontext * driContextPriv)
 {
 	radeonContextPtr radeon = (radeonContextPtr) driContextPriv->driverPrivate;
 
@@ -496,23 +495,12 @@ radeon_make_renderbuffer_current(radeonContextPtr radeon,
 static unsigned
 radeon_bits_per_pixel(const struct radeon_renderbuffer *rb)
 {
-   switch (rb->base._ActualFormat) {
-   case GL_RGB5:
-   case GL_DEPTH_COMPONENT16:
-      return 16;
-   case GL_RGB8:
-   case GL_RGBA8:
-   case GL_DEPTH_COMPONENT24:
-   case GL_DEPTH24_STENCIL8_EXT:
-   case GL_STENCIL_INDEX8_EXT:
-      return 32;
-   default:
-      return 0;
-   }
+   return _mesa_get_format_bytes(rb->base.Format) * 8; 
 }
 
 void
-radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
+radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
+			    GLboolean front_only)
 {
 	unsigned int attachments[10];
 	__DRIbuffer *buffers = NULL;
@@ -538,7 +526,7 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 		struct radeon_renderbuffer *stencil_rb;
 
 		i = 0;
-		if ((radeon->is_front_buffer_rendering ||
+		if ((front_only || radeon->is_front_buffer_rendering ||
 		     radeon->is_front_buffer_reading ||
 		     !draw->color_rb[1])
 		    && draw->color_rb[0]) {
@@ -546,23 +534,25 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 			attachments[i++] = radeon_bits_per_pixel(draw->color_rb[0]);
 		}
 
-		if (draw->color_rb[1]) {
-			attachments[i++] = __DRI_BUFFER_BACK_LEFT;
-			attachments[i++] = radeon_bits_per_pixel(draw->color_rb[1]);
-		}
+		if (!front_only) {
+			if (draw->color_rb[1]) {
+				attachments[i++] = __DRI_BUFFER_BACK_LEFT;
+				attachments[i++] = radeon_bits_per_pixel(draw->color_rb[1]);
+			}
 
-		depth_rb = radeon_get_renderbuffer(&draw->base, BUFFER_DEPTH);
-		stencil_rb = radeon_get_renderbuffer(&draw->base, BUFFER_STENCIL);
-
-		if ((depth_rb != NULL) && (stencil_rb != NULL)) {
-			attachments[i++] = __DRI_BUFFER_DEPTH_STENCIL;
-			attachments[i++] = radeon_bits_per_pixel(depth_rb);
-		} else if (depth_rb != NULL) {
-			attachments[i++] = __DRI_BUFFER_DEPTH;
-			attachments[i++] = radeon_bits_per_pixel(depth_rb);
-		} else if (stencil_rb != NULL) {
-			attachments[i++] = __DRI_BUFFER_STENCIL;
-			attachments[i++] = radeon_bits_per_pixel(stencil_rb);
+			depth_rb = radeon_get_renderbuffer(&draw->base, BUFFER_DEPTH);
+			stencil_rb = radeon_get_renderbuffer(&draw->base, BUFFER_STENCIL);
+
+			if ((depth_rb != NULL) && (stencil_rb != NULL)) {
+				attachments[i++] = __DRI_BUFFER_DEPTH_STENCIL;
+				attachments[i++] = radeon_bits_per_pixel(depth_rb);
+			} else if (depth_rb != NULL) {
+				attachments[i++] = __DRI_BUFFER_DEPTH;
+				attachments[i++] = radeon_bits_per_pixel(depth_rb);
+			} else if (stencil_rb != NULL) {
+				attachments[i++] = __DRI_BUFFER_STENCIL;
+				attachments[i++] = radeon_bits_per_pixel(stencil_rb);
+			}
 		}
 
 		buffers = (*screen->dri2.loader->getBuffersWithFormat)(drawable,
@@ -575,12 +565,14 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 		i = 0;
 		if (draw->color_rb[0])
 			attachments[i++] = __DRI_BUFFER_FRONT_LEFT;
-		if (draw->color_rb[1])
-			attachments[i++] = __DRI_BUFFER_BACK_LEFT;
-		if (radeon_get_renderbuffer(&draw->base, BUFFER_DEPTH))
-			attachments[i++] = __DRI_BUFFER_DEPTH;
-		if (radeon_get_renderbuffer(&draw->base, BUFFER_STENCIL))
-			attachments[i++] = __DRI_BUFFER_STENCIL;
+		if (!front_only) {
+			if (draw->color_rb[1])
+				attachments[i++] = __DRI_BUFFER_BACK_LEFT;
+			if (radeon_get_renderbuffer(&draw->base, BUFFER_DEPTH))
+				attachments[i++] = __DRI_BUFFER_DEPTH;
+			if (radeon_get_renderbuffer(&draw->base, BUFFER_STENCIL))
+				attachments[i++] = __DRI_BUFFER_STENCIL;
+		}
 
 		buffers = (*screen->dri2.loader->getBuffers)(drawable,
 								 &drawable->w,
@@ -728,9 +720,9 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 /* Force the context `c' to be the current context and associate with it
  * buffer `b'.
  */
-GLboolean radeonMakeCurrent(__DRIcontextPrivate * driContextPriv,
-			    __DRIdrawablePrivate * driDrawPriv,
-			    __DRIdrawablePrivate * driReadPriv)
+GLboolean radeonMakeCurrent(__DRIcontext * driContextPriv,
+			    __DRIdrawable * driDrawPriv,
+			    __DRIdrawable * driReadPriv)
 {
 	radeonContextPtr radeon;
 	struct radeon_framebuffer *drfb;
@@ -748,9 +740,9 @@ GLboolean radeonMakeCurrent(__DRIcontextPrivate * driContextPriv,
 	readfb = driReadPriv->driverPrivate;
 
 	if (driContextPriv->driScreenPriv->dri2.enabled) {
-		radeon_update_renderbuffers(driContextPriv, driDrawPriv);
+		radeon_update_renderbuffers(driContextPriv, driDrawPriv, GL_FALSE);
 		if (driDrawPriv != driReadPriv)
-			radeon_update_renderbuffers(driContextPriv, driReadPriv);
+			radeon_update_renderbuffers(driContextPriv, driReadPriv, GL_FALSE);
 		_mesa_reference_renderbuffer(&radeon->state.color.rb,
 			&(radeon_get_renderbuffer(&drfb->base, BUFFER_BACK_LEFT)->base));
 		_mesa_reference_renderbuffer(&radeon->state.depth.rb,
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.h b/src/mesa/drivers/dri/radeon/radeon_common_context.h
index 0309345393..ab79d2dc0f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.h
@@ -92,7 +92,7 @@ struct radeon_renderbuffer
 
 	GLuint pf_pending;  /**< sequence number of pending flip */
 	GLuint vbl_pending;   /**< vblank sequence number of pending flip */
-	__DRIdrawablePrivate *dPriv;
+	__DRIdrawable *dPriv;
 };
 
 struct radeon_framebuffer
@@ -208,6 +208,10 @@ struct radeon_tex_obj {
 	 * and so on.
 	 */
 	GLboolean validated;
+	/* Minimum LOD to be used during rendering */
+	unsigned minLod;
+	/* Miximum LOD to be used during rendering */
+	unsigned maxLod;
 
 	GLuint override_offset;
 	GLboolean image_override; /* Image overridden by GLX_EXT_tfp */
@@ -324,6 +328,7 @@ struct radeon_swtcl_info {
 	GLuint vertex_attr_count;
 
 	GLuint emit_prediction;
+        struct radeon_bo *bo;
 };
 
 #define RADEON_MAX_AOS_ARRAYS		16
@@ -376,8 +381,8 @@ struct radeon_store {
 };
 
 struct radeon_dri_mirror {
-	__DRIcontextPrivate *context;	/* DRI context */
-	__DRIscreenPrivate *screen;	/* DRI screen */
+	__DRIcontext *context;	/* DRI context */
+	__DRIscreen *screen;	/* DRI screen */
 
 	drm_context_t hwContext;
 	drm_hw_lock_t *hwLock;
@@ -401,9 +406,6 @@ struct radeon_state {
 	struct radeon_depthbuffer_state depth;
 	struct radeon_scissor_state scissor;
 	struct radeon_stencilbuffer_state stencil;
-
-	struct radeon_cs_space_check bos[RADEON_MAX_BOS];
-	int validated_bo_count;
 };
 
 /**
@@ -502,7 +504,6 @@ struct radeon_context {
 
    struct {
 	struct radeon_query_object *current;
-	struct radeon_query_object not_flushed_head;
 	struct radeon_state_atom queryobj;
    } query;
 
@@ -522,12 +523,12 @@ struct radeon_context {
 
 #define RADEON_CONTEXT(glctx) ((radeonContextPtr)(ctx->DriverCtx))
 
-static inline __DRIdrawablePrivate* radeon_get_drawable(radeonContextPtr radeon)
+static inline __DRIdrawable* radeon_get_drawable(radeonContextPtr radeon)
 {
 	return radeon->dri.context->driDrawablePriv;
 }
 
-static inline __DRIdrawablePrivate* radeon_get_readable(radeonContextPtr radeon)
+static inline __DRIdrawable* radeon_get_readable(radeonContextPtr radeon)
 {
 	return radeon->dri.context->driReadablePriv;
 }
@@ -580,15 +581,16 @@ static INLINE uint32_t radeonPackFloat24(float f)
 GLboolean radeonInitContext(radeonContextPtr radeon,
 			    struct dd_function_table* functions,
 			    const __GLcontextModes * glVisual,
-			    __DRIcontextPrivate * driContextPriv,
+			    __DRIcontext * driContextPriv,
 			    void *sharedContextPrivate);
 
 void radeonCleanupContext(radeonContextPtr radeon);
-GLboolean radeonUnbindContext(__DRIcontextPrivate * driContextPriv);
-void radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable);
-GLboolean radeonMakeCurrent(__DRIcontextPrivate * driContextPriv,
-			    __DRIdrawablePrivate * driDrawPriv,
-			    __DRIdrawablePrivate * driReadPriv);
-extern void radeonDestroyContext(__DRIcontextPrivate * driContextPriv);
+GLboolean radeonUnbindContext(__DRIcontext * driContextPriv);
+void radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
+				 GLboolean front_only);
+GLboolean radeonMakeCurrent(__DRIcontext * driContextPriv,
+			    __DRIdrawable * driDrawPriv,
+			    __DRIdrawable * driReadPriv);
+extern void radeonDestroyContext(__DRIcontext * driContextPriv);
 
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index 8f4485aee7..3cd305b0a2 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -69,7 +69,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_framebuffer_object
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 #define DRIVER_DATE	"20061018"
 
@@ -79,7 +79,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 /* Extension strings exported by the R100 driver.
  */
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_occlusion_query",		   GL_ARB_occlusion_query_functions},
@@ -109,7 +109,7 @@ const struct dri_extension card_extensions[] =
     { NULL,                                NULL }
 };
 
-const struct dri_extension mm_extensions[] = {
+static const struct dri_extension mm_extensions[] = {
   { "GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions },
   { NULL, NULL }
 };
@@ -208,10 +208,10 @@ static void r100_init_vtbl(radeonContextPtr radeon)
  */
 GLboolean
 r100CreateContext( const __GLcontextModes *glVisual,
-                     __DRIcontextPrivate *driContextPriv,
+                     __DRIcontext *driContextPriv,
                      void *sharedContextPrivate)
 {
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    radeonScreenPtr screen = (radeonScreenPtr)(sPriv->private);
    struct dd_function_table functions;
    r100ContextPtr rmesa;
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.h b/src/mesa/drivers/dri/radeon/radeon_context.h
index 4e2c52c835..dfedc38bfd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -331,8 +331,12 @@ struct r100_hw_state {
 	struct radeon_state_atom stp;
 };
 
+struct radeon_stipple_state {
+	GLuint mask[32];
+};
 
 struct r100_state {
+	struct radeon_stipple_state stipple;
 	struct radeon_texture_state texture;
 };
 
@@ -447,7 +451,7 @@ struct r100_context {
 #define RADEON_OLD_PACKETS 1
 
 extern GLboolean r100CreateContext( const __GLcontextModes *glVisual,
-				    __DRIcontextPrivate *driContextPriv,
+				    __DRIcontext *driContextPriv,
 				    void *sharedContextPrivate);
   
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs.c b/src/mesa/drivers/dri/radeon/radeon_cs.c
new file mode 100644
index 0000000000..17e7433369
--- /dev/null
+++ b/src/mesa/drivers/dri/radeon/radeon_cs.c
@@ -0,0 +1,95 @@
+
+#include <stdio.h>
+#include <stdint.h>
+#include "drm.h"
+#include "radeon_drm.h"
+#include "radeon_bocs_wrapper.h"
+#include "radeon_cs_int_drm.h"
+
+struct radeon_cs *radeon_cs_create(struct radeon_cs_manager *csm,
+			    uint32_t ndw)
+{
+    struct radeon_cs_int *csi = csm->funcs->cs_create(csm, ndw);
+    return (struct radeon_cs *)csi;
+}
+
+int radeon_cs_write_reloc(struct radeon_cs *cs,
+			  struct radeon_bo *bo,
+			  uint32_t read_domain,
+			  uint32_t write_domain,
+			  uint32_t flags)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+
+    return csi->csm->funcs->cs_write_reloc(csi,
+					   bo,
+					   read_domain,
+					   write_domain,
+					   flags);
+}
+
+int radeon_cs_begin(struct radeon_cs *cs,
+		    uint32_t ndw,
+		    const char *file,
+		    const char *func,
+		    int line)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_begin(csi, ndw, file, func, line);
+}
+
+int radeon_cs_end(struct radeon_cs *cs,
+		  const char *file,
+		  const char *func,
+		  int line)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_end(csi, file, func, line);
+}
+
+int radeon_cs_emit(struct radeon_cs *cs)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_emit(csi);
+}
+
+int radeon_cs_destroy(struct radeon_cs *cs)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_destroy(csi);
+}
+
+int radeon_cs_erase(struct radeon_cs *cs)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_erase(csi);
+}
+
+int radeon_cs_need_flush(struct radeon_cs *cs)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return csi->csm->funcs->cs_need_flush(csi);
+}
+
+void radeon_cs_print(struct radeon_cs *cs, FILE *file)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    csi->csm->funcs->cs_print(csi, file);
+}
+
+void radeon_cs_set_limit(struct radeon_cs *cs, uint32_t domain, uint32_t limit)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    if (domain == RADEON_GEM_DOMAIN_VRAM)
+	csi->csm->vram_limit = limit;
+    else
+	csi->csm->gart_limit = limit;
+}
+
+void radeon_cs_space_set_flush(struct radeon_cs *cs, void (*fn)(void *), void *data)
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    csi->space_flush_fn = fn;
+    csi->space_flush_data = data;
+}
+
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs_drm.h b/src/mesa/drivers/dri/radeon/radeon_cs_drm.h
index ab4eca31a3..a3f1750c6e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_cs_drm.h
+++ b/src/mesa/drivers/dri/radeon/radeon_cs_drm.h
@@ -36,6 +36,7 @@
 #include <string.h>
 #include "drm.h"
 #include "radeon_drm.h"
+#include "radeon_bo_drm.h"
 
 struct radeon_cs_reloc {
     struct radeon_bo    *bo;
@@ -49,173 +50,41 @@ struct radeon_cs_reloc {
 #define RADEON_CS_SPACE_OP_TO_BIG 1
 #define RADEON_CS_SPACE_FLUSH 2
 
-struct radeon_cs_space_check {
-    struct radeon_bo *bo;
-    uint32_t read_domains;
-    uint32_t write_domain;
-    uint32_t new_accounted;
-};
-
-#define MAX_SPACE_BOS (32)
-
-struct radeon_cs_manager;
-
 struct radeon_cs {
-    struct radeon_cs_manager    *csm;
-    void                        *relocs;
-    uint32_t                    *packets;
-    unsigned                    crelocs;
-    unsigned                    relocs_total_size;
-    unsigned                    cdw;
-    unsigned                    ndw;
-    int                         section;
+    uint32_t *packets;
+    unsigned cdw;
+    unsigned ndw;
     unsigned                    section_ndw;
     unsigned                    section_cdw;
-    const char                  *section_file;
-    const char                  *section_func;
-    int                         section_line;
-    struct radeon_cs_space_check bos[MAX_SPACE_BOS];
-    int                         bo_count;
-    void                        (*space_flush_fn)(void *);
-    void                        *space_flush_data;
-};
-
-/* cs functions */
-struct radeon_cs_funcs {
-    struct radeon_cs *(*cs_create)(struct radeon_cs_manager *csm,
-                                   uint32_t ndw);
-    int (*cs_write_reloc)(struct radeon_cs *cs,
-                          struct radeon_bo *bo,
-                          uint32_t read_domain,
-                          uint32_t write_domain,
-                          uint32_t flags);
-    int (*cs_begin)(struct radeon_cs *cs,
-                    uint32_t ndw,
-                    const char *file,
-                    const char *func,
-                    int line);
-    int (*cs_end)(struct radeon_cs *cs,
-                  const char *file,
-                  const char *func,
-                  int line);
-    int (*cs_emit)(struct radeon_cs *cs);
-    int (*cs_destroy)(struct radeon_cs *cs);
-    int (*cs_erase)(struct radeon_cs *cs);
-    int (*cs_need_flush)(struct radeon_cs *cs);
-    void (*cs_print)(struct radeon_cs *cs, FILE *file);
-};
-
-struct radeon_cs_manager {
-    struct radeon_cs_funcs  *funcs;
-    int                     fd;
-    int32_t vram_limit, gart_limit;
-    int32_t vram_write_used, gart_write_used;
-    int32_t read_used;
 };
 
-static inline struct radeon_cs *radeon_cs_create(struct radeon_cs_manager *csm,
-                                                 uint32_t ndw)
-{
-    return csm->funcs->cs_create(csm, ndw);
-}
-
-static inline int radeon_cs_write_reloc(struct radeon_cs *cs,
-                                        struct radeon_bo *bo,
-                                        uint32_t read_domain,
-                                        uint32_t write_domain,
-                                        uint32_t flags)
-{
-    return cs->csm->funcs->cs_write_reloc(cs,
-                                          bo,
-                                          read_domain,
-                                          write_domain,
-                                          flags);
-}
-
-static inline int radeon_cs_begin(struct radeon_cs *cs,
-                                  uint32_t ndw,
-                                  const char *file,
-                                  const char *func,
-                                  int line)
-{
-    return cs->csm->funcs->cs_begin(cs, ndw, file, func, line);
-}
-
-static inline int radeon_cs_end(struct radeon_cs *cs,
-                                const char *file,
-                                const char *func,
-                                int line)
-{
-    return cs->csm->funcs->cs_end(cs, file, func, line);
-}
-
-static inline int radeon_cs_emit(struct radeon_cs *cs)
-{
-    return cs->csm->funcs->cs_emit(cs);
-}
-
-static inline int radeon_cs_destroy(struct radeon_cs *cs)
-{
-    return cs->csm->funcs->cs_destroy(cs);
-}
-
-static inline int radeon_cs_erase(struct radeon_cs *cs)
-{
-    return cs->csm->funcs->cs_erase(cs);
-}
-
-static inline int radeon_cs_need_flush(struct radeon_cs *cs)
-{
-    return cs->csm->funcs->cs_need_flush(cs);
-}
-
-static inline void radeon_cs_print(struct radeon_cs *cs, FILE *file)
-{
-    cs->csm->funcs->cs_print(cs, file);
-}
-
-static inline void radeon_cs_set_limit(struct radeon_cs *cs, uint32_t domain, uint32_t limit)
-{
-    
-    if (domain == RADEON_GEM_DOMAIN_VRAM)
-	cs->csm->vram_limit = limit;
-    else
-	cs->csm->gart_limit = limit;
-}
-
-static inline void radeon_cs_write_dword(struct radeon_cs *cs, uint32_t dword)
-{
-    cs->packets[cs->cdw++] = dword;
-    if (cs->section) {
-        cs->section_cdw++;
-    }
-}
-
-static inline void radeon_cs_write_qword(struct radeon_cs *cs, uint64_t qword)
-{
-
-    memcpy(cs->packets + cs->cdw, &qword, sizeof(qword));
-    cs->cdw+=2;
-    if (cs->section) {
-        cs->section_cdw+=2;
-    }
-}
-
-static inline void radeon_cs_write_table(struct radeon_cs *cs, void *data, uint32_t size)
-{
-    memcpy(cs->packets + cs->cdw, data, size * 4);
-    cs->cdw += size;
-    if (cs->section) {
-	    cs->section_cdw += size;
-    }
-}
+#define MAX_SPACE_BOS (32)
 
-static inline void radeon_cs_space_set_flush(struct radeon_cs *cs, void (*fn)(void *), void *data)
-{
-    cs->space_flush_fn = fn;
-    cs->space_flush_data = data;
-}
+struct radeon_cs_manager;
 
+extern struct radeon_cs *radeon_cs_create(struct radeon_cs_manager *csm,
+					  uint32_t ndw);
+
+extern int radeon_cs_begin(struct radeon_cs *cs,
+			   uint32_t ndw,
+			   const char *file,
+			   const char *func, int line);
+extern int radeon_cs_end(struct radeon_cs *cs,
+			 const char *file,
+			 const char *func,
+			 int line);
+extern int radeon_cs_emit(struct radeon_cs *cs);
+extern int radeon_cs_destroy(struct radeon_cs *cs);
+extern int radeon_cs_erase(struct radeon_cs *cs);
+extern int radeon_cs_need_flush(struct radeon_cs *cs);
+extern void radeon_cs_print(struct radeon_cs *cs, FILE *file);
+extern void radeon_cs_set_limit(struct radeon_cs *cs, uint32_t domain, uint32_t limit);
+extern void radeon_cs_space_set_flush(struct radeon_cs *cs, void (*fn)(void *), void *data);
+extern int radeon_cs_write_reloc(struct radeon_cs *cs,
+				 struct radeon_bo *bo,
+				 uint32_t read_domain,
+				 uint32_t write_domain,
+				 uint32_t flags);
 
 /*
  * add a persistent BO to the list
@@ -243,4 +112,30 @@ int radeon_cs_space_check_with_bo(struct radeon_cs *cs,
 				  uint32_t read_domains,
 				  uint32_t write_domain);
 
+static inline void radeon_cs_write_dword(struct radeon_cs *cs, uint32_t dword)
+{
+    cs->packets[cs->cdw++] = dword;
+    if (cs->section_ndw) {
+        cs->section_cdw++;
+    }
+}
+
+static inline void radeon_cs_write_qword(struct radeon_cs *cs, uint64_t qword)
+{
+    memcpy(cs->packets + cs->cdw, &qword, sizeof(uint64_t));
+    cs->cdw += 2;
+    if (cs->section_ndw) {
+        cs->section_cdw += 2;
+    }
+}
+
+static inline void radeon_cs_write_table(struct radeon_cs *cs,
+					 void *data, uint32_t size)
+{
+    memcpy(cs->packets + cs->cdw, data, size * 4);
+    cs->cdw += size;
+    if (cs->section_ndw) {
+	cs->section_cdw += size;
+    }
+}
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs_int_drm.h b/src/mesa/drivers/dri/radeon/radeon_cs_int_drm.h
new file mode 100644
index 0000000000..8ba76bf951
--- /dev/null
+++ b/src/mesa/drivers/dri/radeon/radeon_cs_int_drm.h
@@ -0,0 +1,66 @@
+
+#ifndef _RADEON_CS_INT_H_
+#define _RADEON_CS_INT_H_
+
+struct radeon_cs_space_check {
+    struct radeon_bo_int *bo;
+    uint32_t read_domains;
+    uint32_t write_domain;
+    uint32_t new_accounted;
+};
+
+struct radeon_cs_int {
+    /* keep first two in same place */
+    uint32_t                    *packets;    
+    unsigned                    cdw;
+    unsigned                    ndw;
+    unsigned                    section_ndw;
+    unsigned                    section_cdw;
+    /* private members */
+    struct radeon_cs_manager    *csm;
+    void                        *relocs;
+    unsigned                    crelocs;
+    unsigned                    relocs_total_size;
+    const char                  *section_file;
+    const char                  *section_func;
+    int                         section_line;
+    struct radeon_cs_space_check bos[MAX_SPACE_BOS];
+    int                         bo_count;
+    void                        (*space_flush_fn)(void *);
+    void                        *space_flush_data;
+};
+
+/* cs functions */
+struct radeon_cs_funcs {
+    struct radeon_cs_int *(*cs_create)(struct radeon_cs_manager *csm,
+                                   uint32_t ndw);
+    int (*cs_write_reloc)(struct radeon_cs_int *cs,
+                          struct radeon_bo *bo,
+                          uint32_t read_domain,
+                          uint32_t write_domain,
+                          uint32_t flags);
+    int (*cs_begin)(struct radeon_cs_int *cs,
+                    uint32_t ndw,
+		    const char *file,
+		    const char *func,
+		    int line);
+    int (*cs_end)(struct radeon_cs_int *cs,
+		  const char *file, const char *func,
+		  int line);
+
+
+    int (*cs_emit)(struct radeon_cs_int *cs);
+    int (*cs_destroy)(struct radeon_cs_int *cs);
+    int (*cs_erase)(struct radeon_cs_int *cs);
+    int (*cs_need_flush)(struct radeon_cs_int *cs);
+    void (*cs_print)(struct radeon_cs_int *cs, FILE *file);
+};
+
+struct radeon_cs_manager {
+    struct radeon_cs_funcs  *funcs;
+    int                     fd;
+    int32_t vram_limit, gart_limit;
+    int32_t vram_write_used, gart_write_used;
+    int32_t read_used;
+};
+#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c b/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
index f1addb299e..45b608a1b9 100644
--- a/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
+++ b/src/mesa/drivers/dri/radeon/radeon_cs_legacy.c
@@ -30,10 +30,18 @@
  *      Jérôme Glisse <glisse@freedesktop.org>
  */
 #include <errno.h>
+#include <unistd.h>
+#include <stdint.h>
+#include "drm.h"
+#include "radeon_drm.h"
 
 #include "radeon_bocs_wrapper.h"
 #include "radeon_common.h"
-
+#ifdef HAVE_LIBDRM_RADEON
+#include "radeon_cs_int.h"
+#else
+#include "radeon_cs_int_drm.h"
+#endif
 struct cs_manager_legacy {
     struct radeon_cs_manager    base;
     struct radeon_context       *ctx;
@@ -51,27 +59,27 @@ struct cs_reloc_legacy {
 };
 
 
-static struct radeon_cs *cs_create(struct radeon_cs_manager *csm,
-                                   uint32_t ndw)
+static struct radeon_cs_int *cs_create(struct radeon_cs_manager *csm,
+				       uint32_t ndw)
 {
-    struct radeon_cs *cs;
+    struct radeon_cs_int *csi;
 
-    cs = (struct radeon_cs*)calloc(1, sizeof(struct radeon_cs));
-    if (cs == NULL) {
+    csi = (struct radeon_cs_int*)calloc(1, sizeof(struct radeon_cs_int));
+    if (csi == NULL) {
         return NULL;
     }
-    cs->csm = csm;
-    cs->ndw = (ndw + 0x3FF) & (~0x3FF);
-    cs->packets = (uint32_t*)malloc(4*cs->ndw);
-    if (cs->packets == NULL) {
-        free(cs);
+    csi->csm = csm;
+    csi->ndw = (ndw + 0x3FF) & (~0x3FF);
+    csi->packets = (uint32_t*)malloc(4*csi->ndw);
+    if (csi->packets == NULL) {
+        free(csi);
         return NULL;
     }
-    cs->relocs_total_size = 0;
-    return cs;
+    csi->relocs_total_size = 0;
+    return csi;
 }
 
-static int cs_write_reloc(struct radeon_cs *cs,
+static int cs_write_reloc(struct radeon_cs_int *cs,
                           struct radeon_bo *bo,
                           uint32_t read_domain,
                           uint32_t write_domain,
@@ -150,20 +158,19 @@ static int cs_write_reloc(struct radeon_cs *cs,
     return 0;
 }
 
-static int cs_begin(struct radeon_cs *cs,
+static int cs_begin(struct radeon_cs_int *cs,
                     uint32_t ndw,
                     const char *file,
                     const char *func,
                     int line)
 {
-    if (cs->section) {
+    if (cs->section_ndw) {
         fprintf(stderr, "CS already in a section(%s,%s,%d)\n",
                 cs->section_file, cs->section_func, cs->section_line);
         fprintf(stderr, "CS can't start section(%s,%s,%d)\n",
                 file, func, line);
         return -EPIPE;
     }
-    cs->section = 1;
     cs->section_ndw = ndw;
     cs->section_cdw = 0;
     cs->section_file = file;
@@ -187,18 +194,17 @@ static int cs_begin(struct radeon_cs *cs,
     return 0;
 }
 
-static int cs_end(struct radeon_cs *cs,
+static int cs_end(struct radeon_cs_int *cs,
                   const char *file,
                   const char *func,
                   int line)
 
 {
-    if (!cs->section) {
+    if (!cs->section_ndw) {
         fprintf(stderr, "CS no section to end at (%s,%s,%d)\n",
                 file, func, line);
         return -EPIPE;
     }
-    cs->section = 0;
     if (cs->section_ndw != cs->section_cdw) {
         fprintf(stderr, "CS section size missmatch start at (%s,%s,%d) %d vs %d\n",
                 cs->section_file, cs->section_func, cs->section_line, cs->section_ndw, cs->section_cdw);
@@ -206,10 +212,12 @@ static int cs_end(struct radeon_cs *cs,
                 file, func, line);
         return -EPIPE;
     }
+    cs->section_ndw = 0;
+
     return 0;
 }
 
-static int cs_process_relocs(struct radeon_cs *cs)
+static int cs_process_relocs(struct radeon_cs_int *cs)
 {
     struct cs_manager_legacy *csm = (struct cs_manager_legacy*)cs->csm;
     struct cs_reloc_legacy *relocs;
@@ -254,7 +262,7 @@ restart:
     return 0;
 }
 
-static int cs_set_age(struct radeon_cs *cs)
+static int cs_set_age(struct radeon_cs_int *cs)
 {
     struct cs_manager_legacy *csm = (struct cs_manager_legacy*)cs->csm;
     struct cs_reloc_legacy *relocs;
@@ -268,7 +276,7 @@ static int cs_set_age(struct radeon_cs *cs)
     return 0;
 }
 
-static int cs_emit(struct radeon_cs *cs)
+static int cs_emit(struct radeon_cs_int *cs)
 {
     struct cs_manager_legacy *csm = (struct cs_manager_legacy*)cs->csm;
     drm_radeon_cmd_buffer_t cmd;
@@ -276,7 +284,7 @@ static int cs_emit(struct radeon_cs *cs)
     uint64_t ull;
     int r;
 
-    csm->ctx->vtbl.emit_cs_header(cs, csm->ctx);
+    csm->ctx->vtbl.emit_cs_header((struct radeon_cs *)cs, csm->ctx);
 
     /* append buffer age */
     if ( IS_R300_CLASS(csm->ctx->radeonScreen) )
@@ -289,9 +297,9 @@ static int cs_emit(struct radeon_cs *cs)
       age.scratch.reg = 2;
       age.scratch.n_bufs = 1;
       age.scratch.flags = 0;
-      radeon_cs_write_dword(cs, age.u);
-      radeon_cs_write_qword(cs, ull);
-      radeon_cs_write_dword(cs, 0);
+      radeon_cs_write_dword((struct radeon_cs *)cs, age.u);
+      radeon_cs_write_qword((struct radeon_cs *)cs, ull);
+      radeon_cs_write_dword((struct radeon_cs *)cs, 0);
     }
 
     r = cs_process_relocs(cs);
@@ -342,7 +350,7 @@ static void inline cs_free_reloc(void *relocs_p, int crelocs)
       free(relocs[i].indices);
 }
 
-static int cs_destroy(struct radeon_cs *cs)
+static int cs_destroy(struct radeon_cs_int *cs)
 {
     cs_free_reloc(cs->relocs, cs->crelocs);
     free(cs->relocs);
@@ -351,7 +359,7 @@ static int cs_destroy(struct radeon_cs *cs)
     return 0;
 }
 
-static int cs_erase(struct radeon_cs *cs)
+static int cs_erase(struct radeon_cs_int *cs)
 {
     cs_free_reloc(cs->relocs, cs->crelocs);
     free(cs->relocs);
@@ -359,18 +367,18 @@ static int cs_erase(struct radeon_cs *cs)
     cs->relocs = NULL;
     cs->crelocs = 0;
     cs->cdw = 0;
-    cs->section = 0;
+    cs->section_ndw = 0;
     return 0;
 }
 
-static int cs_need_flush(struct radeon_cs *cs)
+static int cs_need_flush(struct radeon_cs_int *cs)
 {
     /* this function used to flush when the BO usage got to
      * a certain size, now the higher levels handle this better */
     return 0;
 }
 
-static void cs_print(struct radeon_cs *cs, FILE *file)
+static void cs_print(struct radeon_cs_int *cs, FILE *file)
 {
 }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_cs_space_drm.c b/src/mesa/drivers/dri/radeon/radeon_cs_space_drm.c
index 89cbbb5a6b..e22b437d56 100644
--- a/src/mesa/drivers/dri/radeon/radeon_cs_space_drm.c
+++ b/src/mesa/drivers/dri/radeon/radeon_cs_space_drm.c
@@ -29,6 +29,8 @@
 #include <errno.h>
 #include <stdlib.h>
 #include "radeon_bocs_wrapper.h"
+#include "radeon_bo_int_drm.h"
+#include "radeon_cs_int_drm.h"
 
 struct rad_sizes {
     int32_t op_read;
@@ -39,7 +41,7 @@ struct rad_sizes {
 static inline int radeon_cs_setup_bo(struct radeon_cs_space_check *sc, struct rad_sizes *sizes)
 {
     uint32_t read_domains, write_domain;
-    struct radeon_bo *bo;
+    struct radeon_bo_int *bo;
 
     bo = sc->bo;
     sc->new_accounted = 0;
@@ -47,7 +49,7 @@ static inline int radeon_cs_setup_bo(struct radeon_cs_space_check *sc, struct ra
     write_domain = sc->write_domain;
 
     /* legacy needs a static check */
-    if (radeon_bo_is_static(bo)) {
+    if (radeon_bo_is_static((struct radeon_bo *)sc->bo)) {
 	bo->space_accounted = sc->new_accounted = (read_domains << 16) | write_domain;
 	return 0;
     }
@@ -100,11 +102,11 @@ static inline int radeon_cs_setup_bo(struct radeon_cs_space_check *sc, struct ra
     return 0;
 }
 
-static int radeon_cs_do_space_check(struct radeon_cs *cs, struct radeon_cs_space_check *new_tmp)
+static int radeon_cs_do_space_check(struct radeon_cs_int *cs, struct radeon_cs_space_check *new_tmp)
 {
     struct radeon_cs_manager *csm = cs->csm;
     int i;
-    struct radeon_bo *bo;
+    struct radeon_bo_int *bo;
     struct rad_sizes sizes;
     int ret;
 
@@ -158,25 +160,28 @@ static int radeon_cs_do_space_check(struct radeon_cs *cs, struct radeon_cs_space
 
 void radeon_cs_space_add_persistent_bo(struct radeon_cs *cs, struct radeon_bo *bo, uint32_t read_domains, uint32_t write_domain)
 {
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
     int i;
-    for (i = 0; i < cs->bo_count; i++) {
-	if (cs->bos[i].bo == bo &&
-	    cs->bos[i].read_domains == read_domains &&
-	    cs->bos[i].write_domain == write_domain)
+    for (i = 0; i < csi->bo_count; i++) {
+	if (csi->bos[i].bo == boi &&
+	    csi->bos[i].read_domains == read_domains &&
+	    csi->bos[i].write_domain == write_domain)
 	    return;
     }
     radeon_bo_ref(bo);
-    i = cs->bo_count;
-    cs->bos[i].bo = bo;
-    cs->bos[i].read_domains = read_domains;
-    cs->bos[i].write_domain = write_domain;
-    cs->bos[i].new_accounted = 0;
-    cs->bo_count++;
-
-    assert(cs->bo_count < MAX_SPACE_BOS);
+    i = csi->bo_count;
+    csi->bos[i].bo = boi;
+    csi->bos[i].read_domains = read_domains;
+    csi->bos[i].write_domain = write_domain;
+    csi->bos[i].new_accounted = 0;
+    csi->bo_count++;
+
+    assert(csi->bo_count < MAX_SPACE_BOS);
 }
 
-static int radeon_cs_check_space_internal(struct radeon_cs *cs, struct radeon_cs_space_check *tmp_bo)
+static int radeon_cs_check_space_internal(struct radeon_cs_int *cs,
+					  struct radeon_cs_space_check *tmp_bo)
 {
     int ret;
     int flushed = 0;
@@ -198,37 +203,42 @@ again:
 int radeon_cs_space_check_with_bo(struct radeon_cs *cs,
 				  struct radeon_bo *bo,
 				  uint32_t read_domains, uint32_t write_domain)
-{									
+{
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    struct radeon_bo_int *boi = (struct radeon_bo_int *)bo;
     struct radeon_cs_space_check temp_bo;
+    
     int ret = 0;
 
     if (bo) {
-	temp_bo.bo = bo;
+	temp_bo.bo = boi;
 	temp_bo.read_domains = read_domains;
 	temp_bo.write_domain = write_domain;
 	temp_bo.new_accounted = 0;
     }
 
-    ret = radeon_cs_check_space_internal(cs, bo ? &temp_bo : NULL);
+    ret = radeon_cs_check_space_internal(csi, bo ? &temp_bo : NULL);
     return ret;
 }
 
 int radeon_cs_space_check(struct radeon_cs *cs)
 {
-    return radeon_cs_check_space_internal(cs, NULL);
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
+    return radeon_cs_check_space_internal(csi, NULL);
 }
 
 void radeon_cs_space_reset_bos(struct radeon_cs *cs)
 {
+    struct radeon_cs_int *csi = (struct radeon_cs_int *)cs;
     int i;
-    for (i = 0; i < cs->bo_count; i++) {
-	radeon_bo_unref(cs->bos[i].bo);
-	cs->bos[i].bo = NULL;
-	cs->bos[i].read_domains = 0;
-	cs->bos[i].write_domain = 0;
-	cs->bos[i].new_accounted = 0;
+    for (i = 0; i < csi->bo_count; i++) {
+	radeon_bo_unref((struct radeon_bo *)csi->bos[i].bo);
+	csi->bos[i].bo = NULL;
+	csi->bos[i].read_domains = 0;
+	csi->bos[i].write_domain = 0;
+	csi->bos[i].new_accounted = 0;
     }
-    cs->bo_count = 0;
+    csi->bo_count = 0;
 }
 
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_dma.c b/src/mesa/drivers/dri/radeon/radeon_dma.c
index c6edbae9a1..d31e4e47dd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_dma.c
+++ b/src/mesa/drivers/dri/radeon/radeon_dma.c
@@ -151,6 +151,7 @@ void rcommon_emit_vector(GLcontext * ctx, struct radeon_aos *aos,
 	aos->components = size;
 	aos->count = count;
 
+	radeon_bo_map(aos->bo, 1);
 	out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
 	switch (size) {
 	case 1: radeonEmitVec4(out, data, stride, count); break;
@@ -161,6 +162,7 @@ void rcommon_emit_vector(GLcontext * ctx, struct radeon_aos *aos,
 		assert(0);
 		break;
 	}
+	radeon_bo_unmap(aos->bo);
 }
 
 void radeon_init_dma(radeonContextPtr rmesa)
@@ -183,10 +185,6 @@ void radeonRefillCurrentDmaRegion(radeonContextPtr rmesa, int size)
 			__FUNCTION__, size, rmesa->dma.minimum_size);
 
 
-	/* unmap old reserved bo */
-	if (!is_empty_list(&rmesa->dma.reserved))
-		radeon_bo_unmap(first_elem(&rmesa->dma.reserved)->bo);
-
 	if (is_empty_list(&rmesa->dma.free)
 	      || last_elem(&rmesa->dma.free)->bo->size < size) {
 		dma_bo = CALLOC_STRUCT(radeon_dma_bo);
@@ -223,8 +221,6 @@ again_alloc:
         /* Cmd buff have been flushed in radeon_revalidate_bos */
 		goto again_alloc;
 	}
-
-	radeon_bo_map(first_elem(&rmesa->dma.reserved)->bo, 1);
 }
 
 /* Allocates a region from rmesa->dma.current.  If there isn't enough
@@ -281,7 +277,6 @@ void radeonFreeDmaRegions(radeonContextPtr rmesa)
 
 	foreach_s(dma_bo, temp, &rmesa->dma.reserved) {
 		remove_from_list(dma_bo);
-		radeon_bo_unmap(dma_bo->bo);
 	        radeon_bo_unref(dma_bo->bo);
 		FREE(dma_bo);
 	}
@@ -306,10 +301,6 @@ static int radeon_bo_is_idle(struct radeon_bo* bo)
 		WARN_ONCE("Your libdrm or kernel doesn't have support for busy query.\n"
 			"This may cause small performance drop for you.\n");
 	}
-	/* Protect against bug in legacy bo handling that causes bos stay
-	 * referenced even after they should be freed */
-	if (bo->cref != 1)
-		return 0;
 	return ret != -EBUSY;
 }
 
@@ -346,9 +337,7 @@ void radeonReleaseDmaRegions(radeonContextPtr rmesa)
 	foreach_s(dma_bo, temp, &rmesa->dma.wait) {
 		if (dma_bo->expire_counter == time) {
 			WARN_ONCE("Leaking dma buffer object!\n");
-			/* force free of buffer so we don't realy start
-			 * leaking stuff now*/
-			while ((dma_bo->bo = radeon_bo_unref(dma_bo->bo))) {}
+			radeon_bo_unref(dma_bo->bo);
 			remove_from_list(dma_bo);
 			FREE(dma_bo);
 			continue;
@@ -367,9 +356,6 @@ void radeonReleaseDmaRegions(radeonContextPtr rmesa)
 		insert_at_tail(&rmesa->dma.free, dma_bo);
 	}
 
-	/* unmap the last dma region */
-	if (!is_empty_list(&rmesa->dma.reserved))
-		radeon_bo_unmap(first_elem(&rmesa->dma.reserved)->bo);
 	/* move reserved to wait list */
 	foreach_s(dma_bo, temp, &rmesa->dma.reserved) {
 		/* free objects that are too small to be used because of large request */
@@ -403,11 +389,12 @@ void rcommon_flush_last_swtcl_prim( GLcontext *ctx  )
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	struct radeon_dma *dma = &rmesa->dma;
 		
-
 	if (RADEON_DEBUG & RADEON_IOCTL)
 		fprintf(stderr, "%s\n", __FUNCTION__);
 	dma->flush = NULL;
 
+	radeon_bo_unmap(rmesa->swtcl.bo);
+
 	if (!is_empty_list(&dma->reserved)) {
 	    GLuint current_offset = dma->current_used;
 
@@ -422,6 +409,8 @@ void rcommon_flush_last_swtcl_prim( GLcontext *ctx  )
 	    }
 	    rmesa->swtcl.numverts = 0;
 	}
+	radeon_bo_unref(rmesa->swtcl.bo);
+	rmesa->swtcl.bo = NULL;
 }
 /* Alloc space in the current dma region.
  */
@@ -432,6 +421,7 @@ rcommonAllocDmaLowVerts( radeonContextPtr rmesa, int nverts, int vsize )
 	void *head;
 	if (RADEON_DEBUG & RADEON_IOCTL)
 		fprintf(stderr, "%s\n", __FUNCTION__);
+
 	if(is_empty_list(&rmesa->dma.reserved)
 	      ||rmesa->dma.current_vertexptr + bytes > first_elem(&rmesa->dma.reserved)->bo->size) {
 		if (rmesa->dma.flush) {
@@ -455,7 +445,13 @@ rcommonAllocDmaLowVerts( radeonContextPtr rmesa, int nverts, int vsize )
                 rmesa->swtcl.numverts * rmesa->swtcl.vertex_size * 4 ==
                 rmesa->dma.current_vertexptr );
 
-	head = (first_elem(&rmesa->dma.reserved)->bo->ptr + rmesa->dma.current_vertexptr);
+	if (!rmesa->swtcl.bo) {
+		rmesa->swtcl.bo = first_elem(&rmesa->dma.reserved)->bo;
+		radeon_bo_ref(rmesa->swtcl.bo);
+		radeon_bo_map(rmesa->swtcl.bo, 1);
+	}
+
+	head = (rmesa->swtcl.bo->ptr + rmesa->dma.current_vertexptr);
 	rmesa->dma.current_vertexptr += bytes;
 	rmesa->swtcl.numverts += nverts;
 	return head;
diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c
index 7ac53ec0ca..7b1f84a715 100644
--- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
+++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
@@ -33,7 +33,6 @@
 #include "main/framebuffer.h"
 #include "main/renderbuffer.h"
 #include "main/context.h"
-#include "main/texformat.h"
 #include "main/texrender.h"
 #include "drivers/common/meta.h"
 
@@ -91,11 +90,8 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_R3_G3_B2:
    case GL_RGB4:
    case GL_RGB5:
-      rb->_ActualFormat = GL_RGB5;
+      rb->Format = _dri_texformat_rgb565;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 5;
-      rb->GreenBits = 6;
-      rb->BlueBits = 5;
       cpp = 2;
       break;
    case GL_RGB:
@@ -103,12 +99,8 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      rb->_ActualFormat = GL_RGB8;
+      rb->Format = _dri_texformat_argb8888;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 8;
-      rb->GreenBits = 8;
-      rb->BlueBits = 8;
-      rb->AlphaBits = 0;
       cpp = 4;
       break;
    case GL_RGBA:
@@ -119,12 +111,8 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      rb->_ActualFormat = GL_RGBA8;
+      rb->Format = _dri_texformat_argb8888;
       rb->DataType = GL_UNSIGNED_BYTE;
-      rb->RedBits = 8;
-      rb->GreenBits = 8;
-      rb->BlueBits = 8;
-      rb->AlphaBits = 8;
       cpp = 4;
       break;
    case GL_STENCIL_INDEX:
@@ -133,39 +121,36 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    case GL_STENCIL_INDEX8_EXT:
    case GL_STENCIL_INDEX16_EXT:
       /* alloc a depth+stencil buffer */
-      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->Format = MESA_FORMAT_S8_Z24;
       rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->StencilBits = 8;
       cpp = 4;
       break;
    case GL_DEPTH_COMPONENT16:
-      rb->_ActualFormat = GL_DEPTH_COMPONENT16;
+      rb->Format = MESA_FORMAT_Z16;
       rb->DataType = GL_UNSIGNED_SHORT;
-      rb->DepthBits = 16;
       cpp = 2;
       break;
    case GL_DEPTH_COMPONENT:
    case GL_DEPTH_COMPONENT24:
    case GL_DEPTH_COMPONENT32:
-      rb->_ActualFormat = GL_DEPTH_COMPONENT24;
+      rb->Format = MESA_FORMAT_X8_Z24;
       rb->DataType = GL_UNSIGNED_INT;
-      rb->DepthBits = 24;
       cpp = 4;
       break;
    case GL_DEPTH_STENCIL_EXT:
    case GL_DEPTH24_STENCIL8_EXT:
-      rb->_ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+      rb->Format = MESA_FORMAT_S8_Z24;
       rb->DataType = GL_UNSIGNED_INT_24_8_EXT;
-      rb->DepthBits = 24;
-      rb->StencilBits = 8;
       cpp = 4;
       break;
    default:
       _mesa_problem(ctx,
-                    "Unexpected format in intel_alloc_renderbuffer_storage");
+                    "Unexpected format in radeon_alloc_renderbuffer_storage");
       return GL_FALSE;
    }
 
+  rb->_BaseFormat = _mesa_base_fbo_format(ctx, internalFormat);
+
   if (ctx->Driver.Flush)
 	  ctx->Driver.Flush(ctx); /* +r6/r7 */
 
@@ -181,8 +166,9 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
      uint32_t size;
      uint32_t pitch = ((cpp * width + 63) & ~63) / cpp;
 
-     fprintf(stderr,"Allocating %d x %d radeon RBO (pitch %d)\n", width,
-	  height, pitch);
+     if (RADEON_DEBUG & RADEON_MEMORY)
+	     fprintf(stderr,"Allocating %d x %d radeon RBO (pitch %d)\n", width,
+		     height, pitch);
 
      size = pitch * height * cpp;
      rrb->pitch = pitch * cpp;
@@ -213,7 +199,7 @@ radeon_alloc_window_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    ASSERT(rb->Name == 0);
    rb->Width = width;
    rb->Height = height;
-   rb->_ActualFormat = internalFormat;
+   rb->InternalFormat = internalFormat;
 
    return GL_TRUE;
 }
@@ -255,8 +241,13 @@ radeon_nop_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    return GL_FALSE;
 }
 
+
+/**
+ * Create a renderbuffer for a window's color, depth and/or stencil buffer.
+ * Not used for user-created renderbuffers.
+ */
 struct radeon_renderbuffer *
-radeon_create_renderbuffer(GLenum format, __DRIdrawablePrivate *driDrawPriv)
+radeon_create_renderbuffer(gl_format format, __DRIdrawable *driDrawPriv)
 {
     struct radeon_renderbuffer *rrb;
 
@@ -267,67 +258,64 @@ radeon_create_renderbuffer(GLenum format, __DRIdrawablePrivate *driDrawPriv)
     _mesa_init_renderbuffer(&rrb->base, 0);
     rrb->base.ClassID = RADEON_RB_CLASS;
 
-    /* XXX format junk */
+    rrb->base.Format = format;
+
     switch (format) {
-	case GL_RGB5:
-	    rrb->base._ActualFormat = GL_RGB5;
-	    rrb->base._BaseFormat = GL_RGBA;
-	    rrb->base.RedBits = 5;
-	    rrb->base.GreenBits = 6;
-	    rrb->base.BlueBits = 5;
+        case MESA_FORMAT_RGB565:
+	    assert(_mesa_little_endian());
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
+	    break;
+        case MESA_FORMAT_RGB565_REV:
+	    assert(!_mesa_little_endian());
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
+	    break;
+        case MESA_FORMAT_XRGB8888:
+	    assert(_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
 	    break;
-	case GL_RGB8:
-	    rrb->base._ActualFormat = GL_RGB8;
-	    rrb->base._BaseFormat = GL_RGB;
-	    rrb->base.RedBits = 8;
-	    rrb->base.GreenBits = 8;
-	    rrb->base.BlueBits = 8;
-	    rrb->base.AlphaBits = 0;
+        case MESA_FORMAT_XRGB8888_REV:
+	    assert(!_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGB;
 	    break;
-	case GL_RGBA8:
-	    rrb->base._ActualFormat = GL_RGBA8;
-	    rrb->base._BaseFormat = GL_RGBA;
-	    rrb->base.RedBits = 8;
-	    rrb->base.GreenBits = 8;
-	    rrb->base.BlueBits = 8;
-	    rrb->base.AlphaBits = 8;
+	case MESA_FORMAT_ARGB8888:
+	    assert(_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGBA;
 	    break;
-	case GL_STENCIL_INDEX8_EXT:
-	    rrb->base._ActualFormat = GL_STENCIL_INDEX8_EXT;
-	    rrb->base._BaseFormat = GL_STENCIL_INDEX;
-	    rrb->base.StencilBits = 8;
+	case MESA_FORMAT_ARGB8888_REV:
+	    assert(!_mesa_little_endian());
 	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_RGBA;
 	    break;
-	case GL_DEPTH_COMPONENT16:
-	    rrb->base._ActualFormat = GL_DEPTH_COMPONENT16;
-	    rrb->base._BaseFormat = GL_DEPTH_COMPONENT;
-	    rrb->base.DepthBits = 16;
+	case MESA_FORMAT_S8:
+	    rrb->base.DataType = GL_UNSIGNED_BYTE;
+            rrb->base._BaseFormat = GL_STENCIL_INDEX;
+	    break;
+	case MESA_FORMAT_Z16:
 	    rrb->base.DataType = GL_UNSIGNED_SHORT;
+            rrb->base._BaseFormat = GL_DEPTH_COMPONENT;
 	    break;
-	case GL_DEPTH_COMPONENT24:
-	    rrb->base._ActualFormat = GL_DEPTH_COMPONENT24;
-	    rrb->base._BaseFormat = GL_DEPTH_COMPONENT;
-	    rrb->base.DepthBits = 24;
+	case MESA_FORMAT_X8_Z24:
 	    rrb->base.DataType = GL_UNSIGNED_INT;
+            rrb->base._BaseFormat = GL_DEPTH_COMPONENT;
 	    break;
-	case GL_DEPTH24_STENCIL8_EXT:
-	    rrb->base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
-	    rrb->base._BaseFormat = GL_DEPTH_STENCIL_EXT;
-	    rrb->base.DepthBits = 24;
-	    rrb->base.StencilBits = 8;
+	case MESA_FORMAT_S8_Z24:
 	    rrb->base.DataType = GL_UNSIGNED_INT_24_8_EXT;
+            rrb->base._BaseFormat = GL_DEPTH_STENCIL;
 	    break;
 	default:
-	    fprintf(stderr, "%s: Unknown format 0x%04x\n", __FUNCTION__, format);
+	    fprintf(stderr, "%s: Unknown format %s\n",
+                    __FUNCTION__, _mesa_get_format_name(format));
 	    _mesa_delete_renderbuffer(&rrb->base);
 	    return NULL;
     }
 
     rrb->dPriv = driDrawPriv;
-    rrb->base.InternalFormat = format;
+    rrb->base.InternalFormat = _mesa_get_format_base_format(format);
 
     rrb->base.Delete = radeon_delete_renderbuffer;
     rrb->base.AllocStorage = radeon_alloc_window_storage;
@@ -382,51 +370,41 @@ radeon_framebuffer_renderbuffer(GLcontext * ctx,
 }
 
 
+/* TODO: According to EXT_fbo spec internal format of texture image
+ * once set during glTexImage call, should be preserved when
+ * attaching image to renderbuffer. When HW doesn't support
+ * rendering to format of attached image, set framebuffer
+ * completeness accordingly in radeon_validate_framebuffer (issue #79).
+ */
 static GLboolean
 radeon_update_wrapper(GLcontext *ctx, struct radeon_renderbuffer *rrb, 
 		     struct gl_texture_image *texImage)
 {
 	int retry = 0;
+	gl_format texFormat;
+
 restart:
-	if (texImage->TexFormat == &_mesa_texformat_argb8888) {
-		rrb->cpp = 4;
-		rrb->base._ActualFormat = GL_RGBA8;
-		rrb->base._BaseFormat = GL_RGBA;
+	if (texImage->TexFormat == _dri_texformat_argb8888) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to RGBA8 texture OK\n");
 	}
-	else if (texImage->TexFormat == &_mesa_texformat_rgb565) {
-		rrb->cpp = 2;
-		rrb->base._ActualFormat = GL_RGB5;
-		rrb->base._BaseFormat = GL_RGB;
+	else if (texImage->TexFormat == _dri_texformat_rgb565) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to RGB5 texture OK\n");
 	}
-	else if (texImage->TexFormat == &_mesa_texformat_argb1555) {
-		rrb->cpp = 2;
-		rrb->base._ActualFormat = GL_RGB5_A1;
-		rrb->base._BaseFormat = GL_RGBA;
+	else if (texImage->TexFormat == _dri_texformat_argb1555) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
 		DBG("Render to ARGB1555 texture OK\n");
 	}
-	else if (texImage->TexFormat == &_mesa_texformat_argb4444) {
-		rrb->cpp = 2;
-		rrb->base._ActualFormat = GL_RGBA4;
-		rrb->base._BaseFormat = GL_RGBA;
+	else if (texImage->TexFormat == _dri_texformat_argb4444) {
 		rrb->base.DataType = GL_UNSIGNED_BYTE;
-		DBG("Render to ARGB1555 texture OK\n");
+		DBG("Render to ARGB4444 texture OK\n");
 	}
-	else if (texImage->TexFormat == &_mesa_texformat_z16) {
-		rrb->cpp = 2;
-		rrb->base._ActualFormat = GL_DEPTH_COMPONENT16;
-		rrb->base._BaseFormat = GL_DEPTH_COMPONENT;
+	else if (texImage->TexFormat == MESA_FORMAT_Z16) {
 		rrb->base.DataType = GL_UNSIGNED_SHORT;
 		DBG("Render to DEPTH16 texture OK\n");
 	}
-	else if (texImage->TexFormat == &_mesa_texformat_s8_z24) {
-		rrb->cpp = 4;
-		rrb->base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
-		rrb->base._BaseFormat = GL_DEPTH_STENCIL_EXT;
+	else if (texImage->TexFormat == MESA_FORMAT_S8_Z24) {
 		rrb->base.DataType = GL_UNSIGNED_INT_24_8_EXT;
 		DBG("Render to DEPTH_STENCIL texture OK\n");
 	}
@@ -434,27 +412,31 @@ restart:
 		/* try redoing the FBO */
 		if (retry == 1) {
 			DBG("Render to texture BAD FORMAT %d\n",
-			    texImage->TexFormat->MesaFormat);
+			    texImage->TexFormat);
 			return GL_FALSE;
 		}
+                /* XXX why is the tex format being set here?
+                 * I think this can be removed.
+                 */
 		texImage->TexFormat = radeonChooseTextureFormat(ctx, texImage->InternalFormat, 0,
-								texImage->TexFormat->DataType,
+								_mesa_get_format_datatype(texImage->TexFormat),
 								1);
 
 		retry++;
 		goto restart;
 	}
 	
+	texFormat = texImage->TexFormat;
+
+	rrb->base.Format = texFormat;
+
+        rrb->cpp = _mesa_get_format_bytes(texFormat);
 	rrb->pitch = texImage->Width * rrb->cpp;
-	rrb->base.InternalFormat = rrb->base._ActualFormat;
+	rrb->base.InternalFormat = texImage->InternalFormat;
+        rrb->base._BaseFormat = _mesa_base_fbo_format(ctx, rrb->base.InternalFormat);
+
 	rrb->base.Width = texImage->Width;
 	rrb->base.Height = texImage->Height;
-	rrb->base.RedBits = texImage->TexFormat->RedBits;
-	rrb->base.GreenBits = texImage->TexFormat->GreenBits;
-	rrb->base.BlueBits = texImage->TexFormat->BlueBits;
-	rrb->base.AlphaBits = texImage->TexFormat->AlphaBits;
-	rrb->base.DepthBits = texImage->TexFormat->DepthBits;
-	rrb->base.StencilBits = texImage->TexFormat->StencilBits;
 	
 	rrb->base.Delete = radeon_delete_renderbuffer;
 	rrb->base.AllocStorage = radeon_nop_alloc_storage;
@@ -555,8 +537,10 @@ radeon_render_texture(GLcontext * ctx,
       imageOffset += offsets[att->Zoffset];
    }
 
-   /* store that offset in the region */
+   /* store that offset in the region, along with the correct pitch for
+    * the image we are rendering to */
    rrb->draw_offset = imageOffset;
+   rrb->pitch = radeon_image->mt->levels[att->TextureLevel].rowstride;
 
    /* update drawing region, etc */
    radeon_draw_buffer(ctx, fb);
diff --git a/src/mesa/drivers/dri/radeon/radeon_ioctl.c b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
index a0106d00fa..a9d50c5d07 100644
--- a/src/mesa/drivers/dri/radeon/radeon_ioctl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
@@ -449,7 +449,7 @@ void radeonEmitAOS( r100ContextPtr rmesa,
 static void radeonKernelClear(GLcontext *ctx, GLuint flags)
 {
      r100ContextPtr rmesa = R100_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    drm_radeon_sarea_t *sarea = rmesa->radeon.sarea;
    uint32_t clear;
    GLint ret, i;
@@ -570,11 +570,15 @@ static void radeonKernelClear(GLcontext *ctx, GLuint flags)
 static void radeonClear( GLcontext *ctx, GLbitfield mask )
 {
    r100ContextPtr rmesa = R100_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLuint flags = 0;
    GLuint color_mask = 0;
    GLuint orig_mask = mask;
 
+   if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
+      rmesa->radeon.front_buffer_dirty = GL_TRUE;
+   }
+
    if ( RADEON_DEBUG & RADEON_IOCTL ) {
       fprintf( stderr, "radeonClear\n");
    }
diff --git a/src/mesa/drivers/dri/radeon/radeon_lock.c b/src/mesa/drivers/dri/radeon/radeon_lock.c
index 02de8e5fd1..9dee691938 100644
--- a/src/mesa/drivers/dri/radeon/radeon_lock.c
+++ b/src/mesa/drivers/dri/radeon/radeon_lock.c
@@ -58,11 +58,9 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 void radeonGetLock(radeonContextPtr rmesa, GLuint flags)
 {
-	__DRIdrawablePrivate *const drawable = radeon_get_drawable(rmesa);
-	__DRIdrawablePrivate *const readable = radeon_get_readable(rmesa);
-	__DRIscreenPrivate *sPriv = rmesa->dri.screen;
-
-	assert(drawable != NULL);
+	__DRIdrawable *const drawable = radeon_get_drawable(rmesa);
+	__DRIdrawable *const readable = radeon_get_readable(rmesa);
+	__DRIscreen *sPriv = rmesa->dri.screen;
 
 	drmGetLock(rmesa->dri.fd, rmesa->dri.hwContext, flags);
 
@@ -74,12 +72,13 @@ void radeonGetLock(radeonContextPtr rmesa, GLuint flags)
 	 * 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(sPriv, drawable);
-	if (drawable != readable) {
+	if (drawable)
+		DRI_VALIDATE_DRAWABLE_INFO(sPriv, drawable);
+	if (readable && drawable != readable) {
 		DRI_VALIDATE_DRAWABLE_INFO(sPriv, readable);
 	}
 
-	if (rmesa->lastStamp != drawable->lastStamp) {
+	if (drawable && (rmesa->lastStamp != drawable->lastStamp)) {
 		radeon_window_moved(rmesa);
 		rmesa->lastStamp = drawable->lastStamp;
 	}
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
index 08e1c5d00d..d810e6080e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
@@ -76,12 +76,14 @@ static void emit_vecfog(GLcontext *ctx, struct radeon_aos *aos,
 
    /* Emit the data
     */
+   radeon_bo_map(aos->bo, 1);
    out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
    for (i = 0; i < count; i++) {
       out[0] = radeonComputeFogBlendFactor( ctx, *(GLfloat *)data );
       out++;
       data += stride;
    }
+   radeon_bo_unmap(aos->bo);
 }
 
 static void emit_s0_vec(uint32_t *out, GLvoid *data, int stride, int count)
@@ -151,6 +153,7 @@ static void emit_tex_vector(GLcontext *ctx, struct radeon_aos *aos,
 
    /* Emit the data
     */
+   radeon_bo_map(aos->bo, 1);
    out = (uint32_t*)((char*)aos->bo->ptr + aos->offset);
    switch (size) {
    case 1:
@@ -170,6 +173,7 @@ static void emit_tex_vector(GLcontext *ctx, struct radeon_aos *aos,
       exit(1);
       break;
    }
+   radeon_bo_unmap(aos->bo);
 }
 
 
@@ -196,12 +200,12 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.obj.buf) 
 	rcommon_emit_vector( ctx, 
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->ObjPtr->data,
-			     VB->ObjPtr->size,
-			     VB->ObjPtr->stride,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+			     VB->AttribPtr[_TNL_ATTRIB_POS]->size,
+			     VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
 			     count);
 
-      switch( VB->ObjPtr->size ) {
+      switch( VB->AttribPtr[_TNL_ATTRIB_POS]->size ) {
       case 4: vfmt |= RADEON_CP_VC_FRMT_W0;
       case 3: vfmt |= RADEON_CP_VC_FRMT_Z;
       case 2: vfmt |= RADEON_CP_VC_FRMT_XY;
@@ -216,9 +220,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.norm.buf)
 	 rcommon_emit_vector( ctx, 
 			      &(rmesa->tcl.aos[nr]),
-			      (char *)VB->NormalPtr->data,
+			      (char *)VB->AttribPtr[_TNL_ATTRIB_NORMAL]->data,
 			      3,
-			      VB->NormalPtr->stride,
+			      VB->AttribPtr[_TNL_ATTRIB_NORMAL]->stride,
 			      count);
 
       vfmt |= RADEON_CP_VC_FRMT_N0;
@@ -227,9 +231,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
    if (inputs & VERT_BIT_COLOR0) {
       int emitsize;
-      if (VB->ColorPtr[0]->size == 4 &&
-	  (VB->ColorPtr[0]->stride != 0 ||
-	   VB->ColorPtr[0]->data[0][3] != 1.0)) {
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size == 4 &&
+	  (VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride != 0 ||
+	   VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data[0][3] != 1.0)) {
 	 vfmt |= RADEON_CP_VC_FRMT_FPCOLOR | RADEON_CP_VC_FRMT_FPALPHA;
 	 emitsize = 4;
       }
@@ -242,9 +246,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.rgba.buf)
 	rcommon_emit_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->ColorPtr[0]->data,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data,
 			     emitsize,
-			     VB->ColorPtr[0]->stride,
+			     VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride,
 			     count);
 
       nr++;
@@ -256,9 +260,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
 	rcommon_emit_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->SecondaryColorPtr[0]->data,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data,
 			     3,
-			     VB->SecondaryColorPtr[0]->stride,
+			     VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride,
 			     count);
       }
 
@@ -273,8 +277,8 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       if (!rmesa->tcl.fog.buf)
 	 emit_vecfog( ctx,
 		      &(rmesa->tcl.aos[nr]),
-		      (char *)VB->FogCoordPtr->data,
-		      VB->FogCoordPtr->stride,
+		      (char *)VB->AttribPtr[_TNL_ATTRIB_FOG]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_FOG]->stride,
 		      count);
 
       vfmt |= RADEON_CP_VC_FRMT_FPFOG;
@@ -290,24 +294,24 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 if (!rmesa->tcl.tex[unit].buf)
 	    emit_tex_vector( ctx,
 			     &(rmesa->tcl.aos[nr]),
-			     (char *)VB->TexCoordPtr[unit]->data,
-			     VB->TexCoordPtr[unit]->size,
-			     VB->TexCoordPtr[unit]->stride,
+			     (char *)VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->data,
+			     VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size,
+			     VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->stride,
 			     count );
 	 nr++;
 
 	 vfmt |= RADEON_ST_BIT(unit);
          /* assume we need the 3rd coord if texgen is active for r/q OR at least
 	    3 coords are submitted. This may not be 100% correct */
-         if (VB->TexCoordPtr[unit]->size >= 3) {
+         if (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) {
 	    vtx |= RADEON_Q_BIT(unit);
 	    vfmt |= RADEON_Q_BIT(unit);
 	 }
 	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
 	    vtx |= RADEON_Q_BIT(unit);
-	 else if ((VB->TexCoordPtr[unit]->size >= 3) &&
+	 else if ((VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) &&
 	          ((ctx->Texture.Unit[unit]._ReallyEnabled & (TEXTURE_CUBE_BIT)) == 0)) {
-	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    GLuint swaptexmatcol = (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size - 3);
 	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
 		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
 	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
index 515783135d..d764ccb982 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
@@ -56,18 +56,18 @@ static void TAG(emit)( GLcontext *ctx,
 
    radeon_print(RADEON_SWRENDER, RADEON_VERBOSE, "%s\n", __FUNCTION__);
 
-   coord = (GLuint (*)[4])VB->ObjPtr->data;
-   coord_stride = VB->ObjPtr->stride;
+   coord = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_POS]->data;
+   coord_stride = VB->AttribPtr[_TNL_ATTRIB_POS]->stride;
 
    if (DO_TEX2) {
-      if (VB->TexCoordPtr[2]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX2]) {
 	 const GLuint t2 = GET_TEXSOURCE(2);
-	 tc2 = (GLuint (*)[4])VB->TexCoordPtr[t2]->data;
-	 tc2_stride = VB->TexCoordPtr[t2]->stride;
-	 if (DO_PTEX && VB->TexCoordPtr[t2]->size < 3) {
+	 tc2 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->data;
+	 tc2_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->stride;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size < 3) {
 	    fill_tex |= (1<<2);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t2]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t2]->size < 4) {
 	    rqcoordsnoswap |= (1<<2);
 	 }
       } else {
@@ -77,14 +77,14 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_TEX1) {
-      if (VB->TexCoordPtr[1]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]) {
 	 const GLuint t1 = GET_TEXSOURCE(1);
-	 tc1 = (GLuint (*)[4])VB->TexCoordPtr[t1]->data;
-	 tc1_stride = VB->TexCoordPtr[t1]->stride;
-	 if (DO_PTEX && VB->TexCoordPtr[t1]->size < 3) {
+	 tc1 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->data;
+	 tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->stride;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size < 3) {
 	    fill_tex |= (1<<1);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t1]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t1]->size < 4) {
 	    rqcoordsnoswap |= (1<<1);
 	 }
       } else {
@@ -94,14 +94,14 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_TEX0) {
-      if (VB->TexCoordPtr[0]) {
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]) {
 	 const GLuint t0 = GET_TEXSOURCE(0);
-	 tc0_stride = VB->TexCoordPtr[t0]->stride;
-	 tc0 = (GLuint (*)[4])VB->TexCoordPtr[t0]->data;
-	 if (DO_PTEX && VB->TexCoordPtr[t0]->size < 3) {
+	 tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->stride;
+	 tc0 = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->data;
+	 if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size < 3) {
 	    fill_tex |= (1<<0);
 	 }
-	 else if (DO_PTEX && VB->TexCoordPtr[t0]->size < 4) {
+	 else if (DO_PTEX && VB->AttribPtr[_TNL_ATTRIB_TEX0 + t0]->size < 4) {
 	    rqcoordsnoswap |= (1<<0);
 	 }
       } else {
@@ -112,9 +112,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_NORM) {
-      if (VB->NormalPtr) {
-	 norm_stride = VB->NormalPtr->stride;
-	 norm = (GLuint (*)[4])VB->NormalPtr->data;
+      if (VB->AttribPtr[_TNL_ATTRIB_NORMAL]) {
+	 norm_stride = VB->AttribPtr[_TNL_ATTRIB_NORMAL]->stride;
+	 norm = (GLuint (*)[4])VB->AttribPtr[_TNL_ATTRIB_NORMAL]->data;
       } else {
 	 norm_stride = 0;
 	 norm = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_NORMAL];
@@ -122,9 +122,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_RGBA) {
-      if (VB->ColorPtr[0]) {
-	 col = VB->ColorPtr[0]->data;
-	 col_stride = VB->ColorPtr[0]->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR0]) {
+	 col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+	 col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
       } else {
 	 col = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
 	 col_stride = 0;
@@ -132,9 +132,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_SPEC_OR_FOG) {
-      if (VB->SecondaryColorPtr[0]) {
-	 spec = VB->SecondaryColorPtr[0]->data;
-	 spec_stride = VB->SecondaryColorPtr[0]->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_COLOR1]) {
+	 spec = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->data;
+	 spec_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR1]->stride;
       } else {
 	 spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
 	 spec_stride = 0;
@@ -142,9 +142,9 @@ static void TAG(emit)( GLcontext *ctx,
    }
 
    if (DO_SPEC_OR_FOG) {
-      if (VB->FogCoordPtr) {
-	 fog = VB->FogCoordPtr->data;
-	 fog_stride = VB->FogCoordPtr->stride;
+      if (VB->AttribPtr[_TNL_ATTRIB_FOG]) {
+	 fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+	 fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
       } else {
 	 fog = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_FOG];
 	 fog_stride = 0;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index 78ec119302..98f96ff2a7 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -326,7 +326,7 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 
    if (1) {
       req |= RADEON_CP_VC_FRMT_Z;
-      if (VB->ObjPtr->size == 4) {
+      if (VB->AttribPtr[_TNL_ATTRIB_POS]->size == 4) {
 	 req |= RADEON_CP_VC_FRMT_W0;
       }
    }
@@ -348,15 +348,15 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 req |= RADEON_ST_BIT(unit);
 	 /* assume we need the 3rd coord if texgen is active for r/q OR at least
 	    3 coords are submitted. This may not be 100% correct */
-	 if (VB->TexCoordPtr[unit]->size >= 3) {
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) {
 	    req |= RADEON_Q_BIT(unit);
 	    vtx |= RADEON_Q_BIT(unit);
 	 }
 	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
 	    vtx |= RADEON_Q_BIT(unit);
-	 else if ((VB->TexCoordPtr[unit]->size >= 3) &&
+	 else if ((VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size >= 3) &&
 	          ((ctx->Texture.Unit[unit]._ReallyEnabled & (TEXTURE_CUBE_BIT)) == 0)) {
-	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    GLuint swaptexmatcol = (VB->AttribPtr[_TNL_ATTRIB_TEX0 + unit]->size - 3);
 	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
 		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
 	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
@@ -390,19 +390,19 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
     * this, add more vertex code (for obj-2, obj-3) or preferably move
     * to maos.  
     */
-   if (VB->ObjPtr->size < 3 || 
-       (VB->ObjPtr->size == 3 && 
+   if (VB->AttribPtr[_TNL_ATTRIB_POS]->size < 3 ||
+       (VB->AttribPtr[_TNL_ATTRIB_POS]->size == 3 &&
 	(setup_tab[i].vertex_format & RADEON_CP_VC_FRMT_W0))) {
 
       _math_trans_4f( rmesa->tcl.ObjClean.data,
-		      VB->ObjPtr->data,
-		      VB->ObjPtr->stride,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->data,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->stride,
 		      GL_FLOAT,
-		      VB->ObjPtr->size,
+		      VB->AttribPtr[_TNL_ATTRIB_POS]->size,
 		      0,
 		      VB->Count );
 
-      switch (VB->ObjPtr->size) {
+      switch (VB->AttribPtr[_TNL_ATTRIB_POS]->size) {
       case 1:
 	    _mesa_vector4f_clean_elem(&rmesa->tcl.ObjClean, VB->Count, 1);
       case 2:
@@ -416,14 +416,14 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 	 break;
       }
 
-      VB->ObjPtr = &rmesa->tcl.ObjClean;
+      VB->AttribPtr[_TNL_ATTRIB_POS] = &rmesa->tcl.ObjClean;
    }
 
 
-
+   radeon_bo_map(rmesa->radeon.tcl.aos[0].bo, 1);
    setup_tab[i].emit( ctx, 0, VB->Count, 
 		      rmesa->radeon.tcl.aos[0].bo->ptr + rmesa->radeon.tcl.aos[0].offset);
-
+   radeon_bo_unmap(rmesa->radeon.tcl.aos[0].bo);
    //   rmesa->radeon.tcl.aos[0].size = setup_tab[i].vertex_size;
    rmesa->radeon.tcl.aos[0].stride = setup_tab[i].vertex_size;
    rmesa->tcl.vertex_format = setup_tab[i].vertex_format;
diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
index 38db305e2a..033f26db2a 100644
--- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
+++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.c
@@ -1,4 +1,5 @@
 /*
+ * Copyright (C) 2009 Maciej Cencora.
  * Copyright (C) 2008 Nicolai Haehnle.
  *
  * All Rights Reserved.
@@ -32,51 +33,52 @@
 
 #include "main/simple_list.h"
 #include "main/texcompress.h"
-#include "main/texformat.h"
-
-static GLuint radeon_compressed_texture_size(GLcontext *ctx,
-		GLsizei width, GLsizei height, GLsizei depth,
-		GLuint mesaFormat)
+#include "main/teximage.h"
+#include "main/texobj.h"
+#include "radeon_texture.h"
+
+static unsigned get_aligned_compressed_row_stride(
+		gl_format format,
+		unsigned width,
+		unsigned minStride)
 {
-	GLuint size = _mesa_compressed_texture_size(ctx, width, height, depth, mesaFormat);
-
-	if (mesaFormat == MESA_FORMAT_RGB_DXT1 ||
-	    mesaFormat == MESA_FORMAT_RGBA_DXT1) {
-		if (width + 3 < 8)	/* width one block */
-			size = size * 4;
-		else if (width + 3 < 16)
-			size = size * 2;
-	} else {
-		/* DXT3/5, 16 bytes per block */
-	  //		WARN_ONCE("DXT 3/5 suffers from multitexturing problems!\n");
-		if (width + 3 < 8)
-			size = size * 2;
+	const unsigned blockSize = _mesa_get_format_bytes(format);
+	unsigned blockWidth, blockHeight, numXBlocks;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
+	numXBlocks = (width + blockWidth - 1) / blockWidth;
+
+	while (numXBlocks * blockSize < minStride)
+	{
+		++numXBlocks;
 	}
 
-	return size;
+	return numXBlocks * blockSize;
 }
 
+static unsigned get_compressed_image_size(
+		gl_format format,
+		unsigned rowStride,
+		unsigned height)
+{
+	unsigned blockWidth, blockHeight;
+
+	_mesa_get_format_block_size(format, &blockWidth, &blockHeight);
+
+	return rowStride * ((height + blockHeight - 1) / blockHeight);
+}
 
-static int radeon_compressed_num_bytes(GLuint mesaFormat)
+static int find_next_power_of_two(GLuint value)
 {
-   int bytes = 0;
-   switch(mesaFormat) {
-     
-   case MESA_FORMAT_RGB_FXT1:
-   case MESA_FORMAT_RGBA_FXT1:
-   case MESA_FORMAT_RGB_DXT1:
-   case MESA_FORMAT_RGBA_DXT1:
-     bytes = 2;
-     break;
-     
-   case MESA_FORMAT_RGBA_DXT3:
-   case MESA_FORMAT_RGBA_DXT5:
-     bytes = 4;
-   default:
-     break;
-   }
-   
-   return bytes;
+	int i, tmp;
+
+	i = 0;
+	tmp = value - 1;
+	while (tmp) {
+		tmp >>= 1;
+		i++;
+	}
+	return (1 << i);
 }
 
 /**
@@ -91,28 +93,28 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree
 {
 	radeon_mipmap_level *lvl = &mt->levels[level];
 	uint32_t row_align;
+	GLuint height;
+
+	height = find_next_power_of_two(lvl->height);
 
 	/* Find image size in bytes */
-	if (mt->compressed) {
-		/* TODO: Is this correct? Need test cases for compressed textures! */
-		row_align = rmesa->texture_compressed_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = radeon_compressed_texture_size(mt->radeon->glCtx,
-							   lvl->width, lvl->height, lvl->depth, mt->compressed);
+	if (_mesa_is_format_compressed(mt->mesaFormat)) {
+		lvl->rowstride = get_aligned_compressed_row_stride(mt->mesaFormat, lvl->width, rmesa->texture_compressed_row_align);
+		lvl->size = get_compressed_image_size(mt->mesaFormat, lvl->rowstride, height);
 	} else if (mt->target == GL_TEXTURE_RECTANGLE_NV) {
 		row_align = rmesa->texture_rect_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = lvl->rowstride * lvl->height;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
+		lvl->size = lvl->rowstride * height;
 	} else if (mt->tilebits & RADEON_TXO_MICRO_TILE) {
 		/* tile pattern is 16 bytes x2. mipmaps stay 32 byte aligned,
 		 * though the actual offset may be different (if texture is less than
 		 * 32 bytes width) to the untiled case */
-		lvl->rowstride = (lvl->width * mt->bpp * 2 + 31) & ~31;
-		lvl->size = lvl->rowstride * ((lvl->height + 1) / 2) * lvl->depth;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) * 2 + 31) & ~31;
+		lvl->size = lvl->rowstride * ((height + 1) / 2) * lvl->depth;
 	} else {
 		row_align = rmesa->texture_row_align - 1;
-		lvl->rowstride = (lvl->width * mt->bpp + row_align) & ~row_align;
-		lvl->size = lvl->rowstride * lvl->height * lvl->depth;
+		lvl->rowstride = (_mesa_format_row_stride(mt->mesaFormat, lvl->width) + row_align) & ~row_align;
+		lvl->size = lvl->rowstride * height * lvl->depth;
 	}
 	assert(lvl->size > 0);
 
@@ -124,7 +126,7 @@ static void compute_tex_image_offset(radeonContextPtr rmesa, radeon_mipmap_tree
 	if (RADEON_DEBUG & RADEON_TEXTURE)
 	  fprintf(stderr,
 		  "level %d, face %d: rs:%d %dx%d at %d\n",
-		  level, face, lvl->rowstride, lvl->width, lvl->height, lvl->faces[face].offset);
+		  level, face, lvl->rowstride, lvl->width, height, lvl->faces[face].offset);
 }
 
 static GLuint minify(GLuint size, GLuint levels)
@@ -138,22 +140,19 @@ static GLuint minify(GLuint size, GLuint levels)
 
 static void calculate_miptree_layout_r100(radeonContextPtr rmesa, radeon_mipmap_tree *mt)
 {
-	GLuint curOffset;
-	GLuint numLevels;
-	GLuint i;
-	GLuint face;
+	GLuint curOffset, i, face, level;
 
-	numLevels = mt->lastLevel - mt->firstLevel + 1;
-	assert(numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
+	assert(mt->numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
 
 	curOffset = 0;
 	for(face = 0; face < mt->faces; face++) {
 
-		for(i = 0; i < numLevels; i++) {
-			mt->levels[i].width = minify(mt->width0, i);
-			mt->levels[i].height = minify(mt->height0, i);
-			mt->levels[i].depth = minify(mt->depth0, i);
-			compute_tex_image_offset(rmesa, mt, face, i, &curOffset);
+		for(i = 0, level = mt->baseLevel; i < mt->numLevels; i++, level++) {
+			mt->levels[level].valid = 1;
+			mt->levels[level].width = minify(mt->width0, i);
+			mt->levels[level].height = minify(mt->height0, i);
+			mt->levels[level].depth = minify(mt->depth0, i);
+			compute_tex_image_offset(rmesa, mt, face, level, &curOffset);
 		}
 	}
 
@@ -163,23 +162,21 @@ static void calculate_miptree_layout_r100(radeonContextPtr rmesa, radeon_mipmap_
 
 static void calculate_miptree_layout_r300(radeonContextPtr rmesa, radeon_mipmap_tree *mt)
 {
-	GLuint curOffset;
-	GLuint numLevels;
-	GLuint i;
+	GLuint curOffset, i, level;
 
-	numLevels = mt->lastLevel - mt->firstLevel + 1;
-	assert(numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
+	assert(mt->numLevels <= rmesa->glCtx->Const.MaxTextureLevels);
 
 	curOffset = 0;
-	for(i = 0; i < numLevels; i++) {
+	for(i = 0, level = mt->baseLevel; i < mt->numLevels; i++, level++) {
 		GLuint face;
 
-		mt->levels[i].width = minify(mt->width0, i);
-		mt->levels[i].height = minify(mt->height0, i);
-		mt->levels[i].depth = minify(mt->depth0, i);
+		mt->levels[level].valid = 1;
+		mt->levels[level].width = minify(mt->width0, i);
+		mt->levels[level].height = minify(mt->height0, i);
+		mt->levels[level].depth = minify(mt->depth0, i);
 
 		for(face = 0; face < mt->faces; face++)
-			compute_tex_image_offset(rmesa, mt, face, i, &curOffset);
+			compute_tex_image_offset(rmesa, mt, face, level, &curOffset);
 	}
 
 	/* Note the required size in memory */
@@ -189,27 +186,22 @@ static void calculate_miptree_layout_r300(radeonContextPtr rmesa, radeon_mipmap_
 /**
  * Create a new mipmap tree, calculate its layout and allocate memory.
  */
-radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *t,
-		GLenum target, GLenum internal_format, GLuint firstLevel, GLuint lastLevel,
-		GLuint width0, GLuint height0, GLuint depth0,
-		GLuint bpp, GLuint tilebits, GLuint compressed)
+static radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa,
+		GLenum target, gl_format mesaFormat, GLuint baseLevel, GLuint numLevels,
+		GLuint width0, GLuint height0, GLuint depth0, GLuint tilebits)
 {
 	radeon_mipmap_tree *mt = CALLOC_STRUCT(_radeon_mipmap_tree);
 
-	mt->radeon = rmesa;
-	mt->internal_format = internal_format;
+	mt->mesaFormat = mesaFormat;
 	mt->refcount = 1;
-	mt->t = t;
 	mt->target = target;
 	mt->faces = (target == GL_TEXTURE_CUBE_MAP) ? 6 : 1;
-	mt->firstLevel = firstLevel;
-	mt->lastLevel = lastLevel;
+	mt->baseLevel = baseLevel;
+	mt->numLevels = numLevels;
 	mt->width0 = width0;
 	mt->height0 = height0;
 	mt->depth0 = depth0;
-	mt->bpp = compressed ? radeon_compressed_num_bytes(compressed) : bpp;
 	mt->tilebits = tilebits;
-	mt->compressed = compressed;
 
 	if (rmesa->radeonScreen->chip_family >= CHIP_FAMILY_R300)
 		calculate_miptree_layout_r300(rmesa, mt);
@@ -224,53 +216,43 @@ radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *
 	return mt;
 }
 
-void radeon_miptree_reference(radeon_mipmap_tree *mt)
+void radeon_miptree_reference(radeon_mipmap_tree *mt, radeon_mipmap_tree **ptr)
 {
+	assert(!*ptr);
+
 	mt->refcount++;
 	assert(mt->refcount > 0);
+
+	*ptr = mt;
 }
 
-void radeon_miptree_unreference(radeon_mipmap_tree *mt)
+void radeon_miptree_unreference(radeon_mipmap_tree **ptr)
 {
+	radeon_mipmap_tree *mt = *ptr;
 	if (!mt)
 		return;
 
 	assert(mt->refcount > 0);
+
 	mt->refcount--;
 	if (!mt->refcount) {
 		radeon_bo_unref(mt->bo);
 		free(mt);
 	}
-}
 
+	*ptr = 0;
+}
 
 /**
- * Calculate first and last mip levels for the given texture object,
- * where the dimensions are taken from the given texture image at
- * the given level.
- *
- * Note: level is the OpenGL level number, which is not necessarily the same
- * as the first level that is actually present.
- *
- * The base level image of the given texture face must be non-null,
- * or this will fail.
+ * Calculate min and max LOD for the given texture object.
+ * @param[in] tObj texture object whose LOD values to calculate
+ * @param[out] pminLod minimal LOD
+ * @param[out] pmaxLod maximal LOD
  */
-static void calculate_first_last_level(struct gl_texture_object *tObj,
-				       GLuint *pfirstLevel, GLuint *plastLevel,
-				       GLuint face, GLuint level)
+static void calculate_min_max_lod(struct gl_texture_object *tObj,
+				       unsigned *pminLod, unsigned *pmaxLod)
 {
-	const struct gl_texture_image * const baseImage =
-		tObj->Image[face][level];
-
-	assert(baseImage);
-	
-	/* These must be signed values.  MinLod and MaxLod can be negative numbers,
-	* and having firstLevel and lastLevel as signed prevents the need for
-	* extra sign checks.
-	*/
-	int   firstLevel;
-	int   lastLevel;
-
+	int minLod, maxLod;
 	/* Yes, this looks overly complicated, but it's all needed.
 	*/
 	switch (tObj->Target) {
@@ -281,32 +263,30 @@ static void calculate_first_last_level(struct gl_texture_object *tObj,
 		if (tObj->MinFilter == GL_NEAREST || tObj->MinFilter == GL_LINEAR) {
 			/* GL_NEAREST and GL_LINEAR only care about GL_TEXTURE_BASE_LEVEL.
 			*/
-			firstLevel = lastLevel = tObj->BaseLevel;
+			minLod = maxLod = tObj->BaseLevel;
 		} else {
-			firstLevel = tObj->BaseLevel + (GLint)(tObj->MinLod + 0.5);
-			firstLevel = MAX2(firstLevel, tObj->BaseLevel);
-			firstLevel = MIN2(firstLevel, level + baseImage->MaxLog2);
-			lastLevel = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5);
-			lastLevel = MAX2(lastLevel, tObj->BaseLevel);
-			lastLevel = MIN2(lastLevel, level + baseImage->MaxLog2);
-			lastLevel = MIN2(lastLevel, tObj->MaxLevel);
-			lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */
+			minLod = tObj->BaseLevel + (GLint)(tObj->MinLod);
+			minLod = MAX2(minLod, tObj->BaseLevel);
+			minLod = MIN2(minLod, tObj->MaxLevel);
+			maxLod = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5);
+			maxLod = MIN2(maxLod, tObj->MaxLevel);
+			maxLod = MIN2(maxLod, tObj->Image[0][minLod]->MaxLog2 + minLod);
+			maxLod = MAX2(maxLod, minLod); /* need at least one level */
 		}
 		break;
 	case GL_TEXTURE_RECTANGLE_NV:
 	case GL_TEXTURE_4D_SGIS:
-		firstLevel = lastLevel = 0;
+		minLod = maxLod = 0;
 		break;
 	default:
 		return;
 	}
 
 	/* save these values */
-	*pfirstLevel = firstLevel;
-	*plastLevel = lastLevel;
+	*pminLod = minLod;
+	*pmaxLod = maxLod;
 }
 
-
 /**
  * Checks whether the given miptree can hold the given texture image at the
  * given face and level.
@@ -316,20 +296,15 @@ GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
 {
 	radeon_mipmap_level *lvl;
 
-	if (face >= mt->faces || level < mt->firstLevel || level > mt->lastLevel)
+	if (face >= mt->faces)
 		return GL_FALSE;
 
-	if (texImage->InternalFormat != mt->internal_format ||
-	    texImage->IsCompressed != mt->compressed)
+	if (texImage->TexFormat != mt->mesaFormat)
 		return GL_FALSE;
 
-	if (!texImage->IsCompressed &&
-	    !mt->compressed &&
-	    texImage->TexFormat->TexelBytes != mt->bpp)
-		return GL_FALSE;
-
-	lvl = &mt->levels[level - mt->firstLevel];
-	if (lvl->width != texImage->Width ||
+	lvl = &mt->levels[level];
+	if (!lvl->valid ||
+	    lvl->width != texImage->Width ||
 	    lvl->height != texImage->Height ||
 	    lvl->depth != texImage->Depth)
 		return GL_FALSE;
@@ -337,59 +312,72 @@ GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
 	return GL_TRUE;
 }
 
-
 /**
  * Checks whether the given miptree has the right format to store the given texture object.
  */
-GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj)
+static GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj)
 {
 	struct gl_texture_image *firstImage;
-	GLuint compressed;
-	GLuint numfaces = 1;
-	GLuint firstLevel, lastLevel;
-
-	calculate_first_last_level(texObj, &firstLevel, &lastLevel, 0, texObj->BaseLevel);
-	if (texObj->Target == GL_TEXTURE_CUBE_MAP)
-		numfaces = 6;
-
-	firstImage = texObj->Image[0][firstLevel];
-	compressed = firstImage->IsCompressed ? firstImage->TexFormat->MesaFormat : 0;
-
-	return (mt->firstLevel == firstLevel &&
-	        mt->lastLevel == lastLevel &&
-	        mt->width0 == firstImage->Width &&
-	        mt->height0 == firstImage->Height &&
-	        mt->depth0 == firstImage->Depth &&
-	        mt->compressed == compressed &&
-	        (!mt->compressed ? (mt->bpp == firstImage->TexFormat->TexelBytes) : 1));
-}
+	unsigned numLevels;
+	radeon_mipmap_level *mtBaseLevel;
+
+	if (texObj->BaseLevel < mt->baseLevel)
+		return GL_FALSE;
 
+	mtBaseLevel = &mt->levels[texObj->BaseLevel - mt->baseLevel];
+	firstImage = texObj->Image[0][texObj->BaseLevel];
+	numLevels = MIN2(texObj->MaxLevel - texObj->BaseLevel + 1, firstImage->MaxLog2 + 1);
+
+	if (RADEON_DEBUG & RADEON_TEXTURE) {
+		fprintf(stderr, "Checking if miptree %p matches texObj %p\n", mt, texObj);
+		fprintf(stderr, "target %d vs %d\n", mt->target, texObj->Target);
+		fprintf(stderr, "format %d vs %d\n", mt->mesaFormat, firstImage->TexFormat);
+		fprintf(stderr, "numLevels %d vs %d\n", mt->numLevels, numLevels);
+		fprintf(stderr, "width0 %d vs %d\n", mtBaseLevel->width, firstImage->Width);
+		fprintf(stderr, "height0 %d vs %d\n", mtBaseLevel->height, firstImage->Height);
+		fprintf(stderr, "depth0 %d vs %d\n", mtBaseLevel->depth, firstImage->Depth);
+		if (mt->target == texObj->Target &&
+	        mt->mesaFormat == firstImage->TexFormat &&
+	        mt->numLevels >= numLevels &&
+	        mtBaseLevel->width == firstImage->Width &&
+	        mtBaseLevel->height == firstImage->Height &&
+	        mtBaseLevel->depth == firstImage->Depth) {
+			fprintf(stderr, "MATCHED\n");
+		} else {
+			fprintf(stderr, "NOT MATCHED\n");
+		}
+	}
+
+	return (mt->target == texObj->Target &&
+	        mt->mesaFormat == firstImage->TexFormat &&
+	        mt->numLevels >= numLevels &&
+	        mtBaseLevel->width == firstImage->Width &&
+	        mtBaseLevel->height == firstImage->Height &&
+	        mtBaseLevel->depth == firstImage->Depth);
+}
 
 /**
- * Try to allocate a mipmap tree for the given texture that will fit the
- * given image in the given position.
+ * Try to allocate a mipmap tree for the given texture object.
+ * @param[in] rmesa radeon context
+ * @param[in] t radeon texture object
  */
-void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
-		radeon_texture_image *image, GLuint face, GLuint level)
+void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t)
 {
-	GLuint compressed = image->base.IsCompressed ? image->base.TexFormat->MesaFormat : 0;
-	GLuint numfaces = 1;
-	GLuint firstLevel, lastLevel;
+	struct gl_texture_object *texObj = &t->base;
+	struct gl_texture_image *texImg = texObj->Image[0][texObj->BaseLevel];
+	GLuint numLevels;
 
 	assert(!t->mt);
 
-	calculate_first_last_level(&t->base, &firstLevel, &lastLevel, face, level);
-	if (t->base.Target == GL_TEXTURE_CUBE_MAP)
-		numfaces = 6;
-
-	if (level != firstLevel || face >= numfaces)
+	if (!texImg)
 		return;
 
-	t->mt = radeon_miptree_create(rmesa, t, t->base.Target,
-		image->base.InternalFormat,
-		firstLevel, lastLevel,
-		image->base.Width, image->base.Height, image->base.Depth,
-		image->base.TexFormat->TexelBytes, t->tile_bits, compressed);
+	numLevels = MIN2(texObj->MaxLevel - texObj->BaseLevel + 1, texImg->MaxLog2 + 1);
+
+	t->mt = radeon_miptree_create(rmesa, t->base.Target,
+		texImg->TexFormat, texObj->BaseLevel,
+		numLevels, texImg->Width, texImg->Height,
+		texImg->Depth, t->tile_bits);
 }
 
 /* Although we use the image_offset[] array to store relative offsets
@@ -401,21 +389,234 @@ void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
 void
 radeon_miptree_depth_offsets(radeon_mipmap_tree *mt, GLuint level, GLuint *offsets)
 {
-     if (mt->target != GL_TEXTURE_3D || mt->faces == 1)
-        offsets[0] = 0;
-     else {
-	int i;
-	for (i = 0; i < 6; i++)
-		offsets[i] = mt->levels[level].faces[i].offset;
-     }
+	if (mt->target != GL_TEXTURE_3D || mt->faces == 1) {
+		offsets[0] = 0;
+	} else {
+		int i;
+		for (i = 0; i < 6; i++) {
+			offsets[i] = mt->levels[level].faces[i].offset;
+		}
+	}
 }
 
 GLuint
 radeon_miptree_image_offset(radeon_mipmap_tree *mt,
 			    GLuint face, GLuint level)
 {
-   if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
-      return (mt->levels[level].faces[face].offset);
-   else
-      return mt->levels[level].faces[0].offset;
+	if (mt->target == GL_TEXTURE_CUBE_MAP_ARB)
+		return (mt->levels[level].faces[face].offset);
+	else
+		return mt->levels[level].faces[0].offset;
+}
+
+/**
+ * Ensure that the given image is stored in the given miptree from now on.
+ */
+static void migrate_image_to_miptree(radeon_mipmap_tree *mt,
+									 radeon_texture_image *image,
+									 int face, int level)
+{
+	radeon_mipmap_level *dstlvl = &mt->levels[level];
+	unsigned char *dest;
+
+	assert(image->mt != mt);
+	assert(dstlvl->valid);
+	assert(dstlvl->width == image->base.Width);
+	assert(dstlvl->height == image->base.Height);
+	assert(dstlvl->depth == image->base.Depth);
+
+	radeon_bo_map(mt->bo, GL_TRUE);
+	dest = mt->bo->ptr + dstlvl->faces[face].offset;
+
+	if (image->mt) {
+		/* Format etc. should match, so we really just need a memcpy().
+		 * In fact, that memcpy() could be done by the hardware in many
+		 * cases, provided that we have a proper memory manager.
+		 */
+		assert(mt->mesaFormat == image->base.TexFormat);
+
+		radeon_mipmap_level *srclvl = &image->mt->levels[image->mtlevel];
+
+		/* TODO: bring back these assertions once the FBOs are fixed */
+#if 0
+		assert(image->mtlevel == level);
+		assert(srclvl->size == dstlvl->size);
+		assert(srclvl->rowstride == dstlvl->rowstride);
+#endif
+
+		radeon_bo_map(image->mt->bo, GL_FALSE);
+
+		memcpy(dest,
+			image->mt->bo->ptr + srclvl->faces[face].offset,
+			dstlvl->size);
+		radeon_bo_unmap(image->mt->bo);
+
+		radeon_miptree_unreference(&image->mt);
+	} else if (image->base.Data) {
+		/* This condition should be removed, it's here to workaround
+		 * a segfault when mapping textures during software fallbacks.
+		 */
+		const uint32_t srcrowstride = _mesa_format_row_stride(image->base.TexFormat, image->base.Width);
+		uint32_t rows = image->base.Height * image->base.Depth;
+
+		if (_mesa_is_format_compressed(image->base.TexFormat)) {
+			uint32_t blockWidth, blockHeight;
+			_mesa_get_format_block_size(image->base.TexFormat, &blockWidth, &blockHeight);
+			rows = (rows + blockHeight - 1) / blockHeight;
+		}
+
+		copy_rows(dest, dstlvl->rowstride, image->base.Data, srcrowstride,
+				  rows, srcrowstride);
+
+		_mesa_free_texmemory(image->base.Data);
+		image->base.Data = 0;
+	}
+
+	radeon_bo_unmap(mt->bo);
+
+	radeon_miptree_reference(mt, &image->mt);
+	image->mtface = face;
+	image->mtlevel = level;
+}
+
+/**
+ * Filter matching miptrees, and select one with the most of data.
+ * @param[in] texObj radeon texture object
+ * @param[in] firstLevel first texture level to check
+ * @param[in] lastLevel last texture level to check
+ */
+static radeon_mipmap_tree * get_biggest_matching_miptree(radeonTexObj *texObj,
+														 unsigned firstLevel,
+														 unsigned lastLevel)
+{
+	const unsigned numLevels = lastLevel - firstLevel + 1;
+	unsigned *mtSizes = calloc(numLevels, sizeof(unsigned));
+	radeon_mipmap_tree **mts = calloc(numLevels, sizeof(radeon_mipmap_tree *));
+	unsigned mtCount = 0;
+	unsigned maxMtIndex = 0;
+	radeon_mipmap_tree *tmp;
+
+	for (unsigned level = firstLevel; level <= lastLevel; ++level) {
+		radeon_texture_image *img = get_radeon_texture_image(texObj->base.Image[0][level]);
+		unsigned found = 0;
+		// TODO: why this hack??
+		if (!img)
+			break;
+
+		if (!img->mt)
+			continue;
+
+		for (int i = 0; i < mtCount; ++i) {
+			if (mts[i] == img->mt) {
+				found = 1;
+				mtSizes[i] += img->mt->levels[img->mtlevel].size;
+				break;
+			}
+		}
+
+		if (!found && radeon_miptree_matches_texture(img->mt, &texObj->base)) {
+			mtSizes[mtCount] = img->mt->levels[img->mtlevel].size;
+			mts[mtCount] = img->mt;
+			mtCount++;
+		}
+	}
+
+	if (mtCount == 0) {
+		return NULL;
+	}
+
+	for (int i = 1; i < mtCount; ++i) {
+		if (mtSizes[i] > mtSizes[maxMtIndex]) {
+			maxMtIndex = i;
+		}
+	}
+
+	tmp = mts[maxMtIndex];
+	free(mtSizes);
+	free(mts);
+
+	return tmp;
+}
+
+/**
+ * Validate texture mipmap tree.
+ * If individual images are stored in different mipmap trees
+ * use the mipmap tree that has the most of the correct data.
+ */
+int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *texObj)
+{
+	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
+	radeonTexObj *t = radeon_tex_obj(texObj);
+
+	if (t->validated || t->image_override) {
+		return GL_TRUE;
+	}
+
+	if (texObj->Image[0][texObj->BaseLevel]->Border > 0)
+		return GL_FALSE;
+
+	_mesa_test_texobj_completeness(rmesa->glCtx, texObj);
+	if (!texObj->_Complete) {
+		return GL_FALSE;
+	}
+
+	calculate_min_max_lod(&t->base, &t->minLod, &t->maxLod);
+
+	if (RADEON_DEBUG & RADEON_TEXTURE)
+		fprintf(stderr, "%s: Validating texture %p now, minLod = %d, maxLod = %d\n",
+				__FUNCTION__, texObj ,t->minLod, t->maxLod);
+
+	radeon_mipmap_tree *dst_miptree;
+	dst_miptree = get_biggest_matching_miptree(t, t->minLod, t->maxLod);
+
+	if (!dst_miptree) {
+		radeon_miptree_unreference(&t->mt);
+		radeon_try_alloc_miptree(rmesa, t);
+		dst_miptree = t->mt;
+		if (RADEON_DEBUG & RADEON_TEXTURE) {
+			fprintf(stderr, "%s: No matching miptree found, allocated new one %p\n", __FUNCTION__, t->mt);
+		}
+	} else if (RADEON_DEBUG & RADEON_TEXTURE) {
+		fprintf(stderr, "%s: Using miptree %p\n", __FUNCTION__, t->mt);
+	}
+
+	const unsigned faces = texObj->Target == GL_TEXTURE_CUBE_MAP ? 6 : 1;
+	unsigned face, level;
+	radeon_texture_image *img;
+	/* Validate only the levels that will actually be used during rendering */
+	for (face = 0; face < faces; ++face) {
+		for (level = t->minLod; level <= t->maxLod; ++level) {
+			img = get_radeon_texture_image(texObj->Image[face][level]);
+
+			if (RADEON_DEBUG & RADEON_TEXTURE) {
+				fprintf(stderr, "Checking image level %d, face %d, mt %p ... ", level, face, img->mt);
+			}
+			
+			if (img->mt != dst_miptree) {
+				if (RADEON_DEBUG & RADEON_TEXTURE) {
+					fprintf(stderr, "MIGRATING\n");
+				}
+				struct radeon_bo *src_bo = (img->mt) ? img->mt->bo : img->bo;
+				if (src_bo && radeon_bo_is_referenced_by_cs(src_bo, rmesa->cmdbuf.cs)) {
+					radeon_firevertices(rmesa);
+				}
+				migrate_image_to_miptree(dst_miptree, img, face, level);
+			} else if (RADEON_DEBUG & RADEON_TEXTURE) {
+				fprintf(stderr, "OK\n");
+			}
+		}
+	}
+
+	t->validated = GL_TRUE;
+
+	return GL_TRUE;
 }
+
+uint32_t get_base_teximage_offset(radeonTexObj *texObj)
+{
+	if (!texObj->mt) {
+		return 0;
+	} else {
+		return radeon_miptree_image_offset(texObj->mt, 0, texObj->minLod);
+	}
+}
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
index db28252da3..a10649b5ae 100644
--- a/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
+++ b/src/mesa/drivers/dri/radeon/radeon_mipmap_tree.h
@@ -44,6 +44,7 @@ struct _radeon_mipmap_level {
 	GLuint depth;
 	GLuint size; /** Size of each image, in bytes */
 	GLuint rowstride; /** in bytes */
+	GLuint valid;
 	radeon_mipmap_image faces[6];
 };
 
@@ -59,43 +60,35 @@ struct _radeon_mipmap_level {
  * changed.
  */
 struct _radeon_mipmap_tree {
-	radeonContextPtr radeon;
-	radeonTexObj *t;
 	struct radeon_bo *bo;
 	GLuint refcount;
 
 	GLuint totalsize; /** total size of the miptree, in bytes */
 
 	GLenum target; /** GL_TEXTURE_xxx */
-	GLenum internal_format;
+	GLenum mesaFormat; /** MESA_FORMAT_xxx */
 	GLuint faces; /** # of faces: 6 for cubemaps, 1 otherwise */
-	GLuint firstLevel; /** First mip level stored in this mipmap tree */
-	GLuint lastLevel; /** Last mip level stored in this mipmap tree */
+	GLuint baseLevel; /** gl_texture_object->baseLevel it was created for */
+	GLuint numLevels; /** Number of mip levels stored in this mipmap tree */
 
-	GLuint width0; /** Width of firstLevel image */
-	GLuint height0; /** Height of firstLevel image */
-	GLuint depth0; /** Depth of firstLevel image */
+	GLuint width0; /** Width of baseLevel image */
+	GLuint height0; /** Height of baseLevel image */
+	GLuint depth0; /** Depth of baseLevel image */
 
-	GLuint bpp; /** Bytes per texel */
 	GLuint tilebits; /** RADEON_TXO_xxx_TILE */
-	GLuint compressed; /** MESA_FORMAT_xxx indicating a compressed format, or 0 if uncompressed */
 
 	radeon_mipmap_level levels[RADEON_MIPTREE_MAX_TEXTURE_LEVELS];
 };
 
-radeon_mipmap_tree* radeon_miptree_create(radeonContextPtr rmesa, radeonTexObj *t,
-		GLenum target, GLenum internal_format, GLuint firstLevel, GLuint lastLevel,
-		GLuint width0, GLuint height0, GLuint depth0,
-		GLuint bpp, GLuint tilebits, GLuint compressed);
-void radeon_miptree_reference(radeon_mipmap_tree *mt);
-void radeon_miptree_unreference(radeon_mipmap_tree *mt);
+void radeon_miptree_reference(radeon_mipmap_tree *mt, radeon_mipmap_tree **ptr);
+void radeon_miptree_unreference(radeon_mipmap_tree **ptr);
 
 GLboolean radeon_miptree_matches_image(radeon_mipmap_tree *mt,
 		struct gl_texture_image *texImage, GLuint face, GLuint level);
-GLboolean radeon_miptree_matches_texture(radeon_mipmap_tree *mt, struct gl_texture_object *texObj);
-void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t,
-			      radeon_texture_image *texImage, GLuint face, GLuint level);
+void radeon_try_alloc_miptree(radeonContextPtr rmesa, radeonTexObj *t);
 GLuint radeon_miptree_image_offset(radeon_mipmap_tree *mt,
 				   GLuint face, GLuint level);
 void radeon_miptree_depth_offsets(radeon_mipmap_tree *mt, GLuint level, GLuint *offsets);
+
+uint32_t get_base_teximage_offset(radeonTexObj *texObj);
 #endif /* __RADEON_MIPMAP_TREE_H_ */
diff --git a/src/mesa/drivers/dri/radeon/radeon_queryobj.c b/src/mesa/drivers/dri/radeon/radeon_queryobj.c
index b79d864ba2..98117cdfc1 100644
--- a/src/mesa/drivers/dri/radeon/radeon_queryobj.c
+++ b/src/mesa/drivers/dri/radeon/radeon_queryobj.c
@@ -31,24 +31,11 @@
 #include "main/imports.h"
 #include "main/simple_list.h"
 
-static int radeonQueryIsFlushed(GLcontext *ctx, struct gl_query_object *q)
-{
-	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
-	struct radeon_query_object *tmp, *query = (struct radeon_query_object *)q;
-
-	foreach(tmp, &radeon->query.not_flushed_head) {
-		if (tmp == query) {
-			return 0;
-		}
-	}
-
-	return 1;
-}
-
 static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 {
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
 	struct radeon_query_object *query = (struct radeon_query_object *)q;
-	uint32_t *result;
+        uint32_t *result;
 	int i;
 
 	radeon_print(RADEON_STATE, RADEON_VERBOSE,
@@ -56,13 +43,35 @@ static void radeonQueryGetResult(GLcontext *ctx, struct gl_query_object *q)
 			__FUNCTION__, query->Base.Id, (int) query->Base.Result);
 
 	radeon_bo_map(query->bo, GL_FALSE);
-
-	result = query->bo->ptr;
+        result = query->bo->ptr;
 
 	query->Base.Result = 0;
-	for (i = 0; i < query->curr_offset/sizeof(uint32_t); ++i) {
-		query->Base.Result += result[i];
-		radeon_print(RADEON_STATE, RADEON_TRACE, "result[%d] = %d\n", i, result[i]);
+	if (IS_R600_CLASS(radeon->radeonScreen)) {
+		/* ZPASS EVENT writes alternating qwords
+		 * At query start we set the start offset to 0 and
+		 * hw writes zpass start counts to qwords 0, 2, 4, 6.
+		 * At query end we set the start offset to 8 and
+		 * hw writes zpass end counts to qwords 1, 3, 5, 7.
+		 * then we substract. MSB is the valid bit.
+		 */
+		for (i = 0; i < 16; i += 4) {
+			uint64_t start = (uint64_t)LE32_TO_CPU(result[i]) |
+					 (uint64_t)LE32_TO_CPU(result[i + 1]) << 32;
+			uint64_t end = (uint64_t)LE32_TO_CPU(result[i + 2]) |
+				       (uint64_t)LE32_TO_CPU(result[i + 3]) << 32;
+			if ((start & 0x8000000000000000) && (end & 0x8000000000000000)) {
+				uint64_t query_count = end - start;
+				query->Base.Result += query_count;
+
+			}
+			radeon_print(RADEON_STATE, RADEON_TRACE,
+				     "%d start: %lx, end: %lx %ld\n", i, start, end, end - start);
+		}
+	} else {
+		for (i = 0; i < query->curr_offset/sizeof(uint32_t); ++i) {
+			query->Base.Result += LE32_TO_CPU(result[i]);
+			radeon_print(RADEON_STATE, RADEON_TRACE, "result[%d] = %d\n", i, LE32_TO_CPU(result[i]));
+		}
 	}
 
 	radeon_bo_unmap(query->bo);
@@ -99,10 +108,11 @@ static void radeonDeleteQuery(GLcontext *ctx, struct gl_query_object *q)
 
 static void radeonWaitQuery(GLcontext *ctx, struct gl_query_object *q)
 {
+	radeonContextPtr radeon = RADEON_CONTEXT(ctx);
 	struct radeon_query_object *query = (struct radeon_query_object *)q;
 
 	/* If the cmdbuf with packets for this query hasn't been flushed yet, do it now */
-	if (!radeonQueryIsFlushed(ctx, q))
+	if (radeon_bo_is_referenced_by_cs(query->bo, radeon->cmdbuf.cs))
 		ctx->Driver.Flush(ctx);
 
 	radeon_print(RADEON_STATE, RADEON_VERBOSE, "%s: query id %d, bo %p, offset %d\n", __FUNCTION__, q->Id, query->bo, query->curr_offset);
@@ -134,8 +144,6 @@ static void radeonBeginQuery(GLcontext *ctx, struct gl_query_object *q)
 
 	radeon->query.queryobj.dirty = GL_TRUE;
 	radeon->hw.is_dirty = GL_TRUE;
-	insert_at_tail(&radeon->query.not_flushed_head, query);
-
 }
 
 void radeonEmitQueryEnd(GLcontext *ctx)
@@ -183,7 +191,7 @@ static void radeonCheckQuery(GLcontext *ctx, struct gl_query_object *q)
 		uint32_t domain;
 
 		/* Need to perform a flush, as per ARB_occlusion_query spec */
-		if (!radeonQueryIsFlushed(ctx, q)) {
+		if (radeon_bo_is_referenced_by_cs(query->bo, radeon->cmdbuf.cs)) {
 			ctx->Driver.Flush(ctx);
 		}
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 573eb6c9c1..3080a0fcd0 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -141,12 +141,6 @@ DRI_CONF_BEGIN
 DRI_CONF_END;
 static const GLuint __driNConfigOptions = 17;
 
-extern const struct dri_extension blend_extensions[];
-extern const struct dri_extension ARB_vp_extension[];
-extern const struct dri_extension NV_vp_extension[];
-extern const struct dri_extension ATI_fs_extension[];
-extern const struct dri_extension point_extensions[];
-
 #elif defined(RADEON_R300) || defined(RADEON_R600)
 
 #define DRI_CONF_FP_OPTIMIZATION_SPEED   0
@@ -218,17 +212,12 @@ DRI_CONF_BEGIN
 DRI_CONF_END;
 static const GLuint __driNConfigOptions = 17;
 
-extern const struct dri_extension gl_20_extension[];
-
 #endif
 
-extern const struct dri_extension card_extensions[];
-extern const struct dri_extension mm_extensions[];
-
-static int getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo );
+static int getSwapInfo( __DRIdrawable *dPriv, __DRIswapInfo * sInfo );
 
 static int
-radeonGetParam(__DRIscreenPrivate *sPriv, int param, void *value)
+radeonGetParam(__DRIscreen *sPriv, int param, void *value)
 {
   int ret;
   drm_radeon_getparam_t gp = { 0 };
@@ -260,7 +249,7 @@ radeonGetParam(__DRIscreenPrivate *sPriv, int param, void *value)
 }
 
 static const __DRIconfig **
-radeonFillInModes( __DRIscreenPrivate *psp,
+radeonFillInModes( __DRIscreen *psp,
 		   unsigned pixel_bits, unsigned depth_bits,
 		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -401,12 +390,14 @@ static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id)
    screen->device_id = device_id;
    screen->chip_flags = 0;
    switch ( device_id ) {
+   case PCI_CHIP_RN50_515E:
+   case PCI_CHIP_RN50_5969:
+	return -1;
+
    case PCI_CHIP_RADEON_LY:
    case PCI_CHIP_RADEON_LZ:
    case PCI_CHIP_RADEON_QY:
    case PCI_CHIP_RADEON_QZ:
-   case PCI_CHIP_RN50_515E:
-   case PCI_CHIP_RN50_5969:
       screen->chip_family = CHIP_FAMILY_RV100;
       break;
 
@@ -920,7 +911,7 @@ static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id)
 /* Create the device specific screen private data struct.
  */
 static radeonScreenPtr
-radeonCreateScreen( __DRIscreenPrivate *sPriv )
+radeonCreateScreen( __DRIscreen *sPriv )
 {
    radeonScreenPtr screen;
    RADEONDRIPtr dri_priv = (RADEONDRIPtr)sPriv->pDevPriv;
@@ -1259,7 +1250,7 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
 }
 
 static radeonScreenPtr
-radeonCreateScreen2(__DRIscreenPrivate *sPriv)
+radeonCreateScreen2(__DRIscreen *sPriv)
 {
    radeonScreenPtr screen;
    int i;
@@ -1410,7 +1401,7 @@ radeonCreateScreen2(__DRIscreenPrivate *sPriv)
 /* Destroy the device specific screen private data struct.
  */
 static void
-radeonDestroyScreen( __DRIscreenPrivate *sPriv )
+radeonDestroyScreen( __DRIscreen *sPriv )
 {
     radeonScreenPtr screen = (radeonScreenPtr)sPriv->private;
 
@@ -1444,7 +1435,7 @@ radeonDestroyScreen( __DRIscreenPrivate *sPriv )
 /* Initialize the driver specific screen private data.
  */
 static GLboolean
-radeonInitDriver( __DRIscreenPrivate *sPriv )
+radeonInitDriver( __DRIscreen *sPriv )
 {
     if (sPriv->dri2.enabled) {
         sPriv->private = (void *) radeonCreateScreen2( sPriv );
@@ -1468,8 +1459,8 @@ radeonInitDriver( __DRIscreenPrivate *sPriv )
  * pbuffers.
  */
 static GLboolean
-radeonCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                    __DRIdrawablePrivate *driDrawPriv,
+radeonCreateBuffer( __DRIscreen *driScrnPriv,
+                    __DRIdrawable *driDrawPriv,
                     const __GLcontextModes *mesaVis,
                     GLboolean isPixmap )
 {
@@ -1480,7 +1471,7 @@ radeonCreateBuffer( __DRIscreenPrivate *driScrnPriv,
     const GLboolean swAccum = mesaVis->accumRedBits > 0;
     const GLboolean swStencil = mesaVis->stencilBits > 0 &&
 	mesaVis->depthBits != 24;
-    GLenum rgbFormat;
+    gl_format rgbFormat;
     struct radeon_framebuffer *rfb;
 
     if (isPixmap)
@@ -1493,11 +1484,11 @@ radeonCreateBuffer( __DRIscreenPrivate *driScrnPriv,
     _mesa_initialize_framebuffer(&rfb->base, mesaVis);
 
     if (mesaVis->redBits == 5)
-        rgbFormat = GL_RGB5;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_RGB565 : MESA_FORMAT_RGB565_REV;
     else if (mesaVis->alphaBits == 0)
-        rgbFormat = GL_RGB8;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_XRGB8888 : MESA_FORMAT_XRGB8888_REV;
     else
-        rgbFormat = GL_RGBA8;
+        rgbFormat = _mesa_little_endian() ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB8888_REV;
 
     /* front color renderbuffer */
     rfb->color_rb[0] = radeon_create_renderbuffer(rgbFormat, driDrawPriv);
@@ -1513,19 +1504,22 @@ radeonCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
     if (mesaVis->depthBits == 24) {
       if (mesaVis->stencilBits == 8) {
-	struct radeon_renderbuffer *depthStencilRb = radeon_create_renderbuffer(GL_DEPTH24_STENCIL8_EXT, driDrawPriv);
+	struct radeon_renderbuffer *depthStencilRb =
+           radeon_create_renderbuffer(MESA_FORMAT_S8_Z24, driDrawPriv);
 	_mesa_add_renderbuffer(&rfb->base, BUFFER_DEPTH, &depthStencilRb->base);
 	_mesa_add_renderbuffer(&rfb->base, BUFFER_STENCIL, &depthStencilRb->base);
 	depthStencilRb->has_surface = screen->depthHasSurface;
       } else {
 	/* depth renderbuffer */
-	struct radeon_renderbuffer *depth = radeon_create_renderbuffer(GL_DEPTH_COMPONENT24, driDrawPriv);
+	struct radeon_renderbuffer *depth =
+           radeon_create_renderbuffer(MESA_FORMAT_X8_Z24, driDrawPriv);
 	_mesa_add_renderbuffer(&rfb->base, BUFFER_DEPTH, &depth->base);
 	depth->has_surface = screen->depthHasSurface;
       }
     } else if (mesaVis->depthBits == 16) {
-      /* just 16-bit depth buffer, no hw stencil */
-	struct radeon_renderbuffer *depth = radeon_create_renderbuffer(GL_DEPTH_COMPONENT16, driDrawPriv);
+        /* just 16-bit depth buffer, no hw stencil */
+	struct radeon_renderbuffer *depth =
+           radeon_create_renderbuffer(MESA_FORMAT_Z16, driDrawPriv);
 	_mesa_add_renderbuffer(&rfb->base, BUFFER_DEPTH, &depth->base);
 	depth->has_surface = screen->depthHasSurface;
     }
@@ -1565,7 +1559,7 @@ static void radeon_cleanup_renderbuffers(struct radeon_framebuffer *rfb)
 }
 
 void
-radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+radeonDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
     struct radeon_framebuffer *rfb;
     if (!driDrawPriv)
@@ -1587,7 +1581,7 @@ radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-radeonInitScreen(__DRIscreenPrivate *psp)
+radeonInitScreen(__DRIscreen *psp)
 {
 #if defined(RADEON_R100)
    static const char *driver_name = "Radeon";
@@ -1619,27 +1613,6 @@ radeonInitScreen(__DRIscreenPrivate *psp)
       return NULL;
    }
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create
-    * is called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-#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_R300) || defined(RADEON_R600))
-   driInitSingleExtension( NULL, gl_20_extension );
-#endif
-
    if (!radeonInitDriver(psp))
        return NULL;
 
@@ -1658,7 +1631,7 @@ radeonInitScreen(__DRIscreenPrivate *psp)
  * \return the __GLcontextModes supported by this driver
  */
 static const
-__DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp)
+__DRIconfig **radeonInitScreen2(__DRIscreen *psp)
 {
    GLenum fb_format[3];
    GLenum fb_type[3];
@@ -1672,28 +1645,6 @@ __DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp)
    int color;
    __DRIconfig **configs = NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create
-    * is called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-   driInitExtensions( NULL, mm_extensions, GL_FALSE );
-#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_R300) || defined(RADEON_R600))
-   driInitSingleExtension( NULL, gl_20_extension );
-#endif
-
    if (!radeonInitDriver(psp)) {
        return NULL;
     }
@@ -1747,7 +1698,7 @@ __DRIconfig **radeonInitScreen2(__DRIscreenPrivate *psp)
  * Get information about previous buffer swaps.
  */
 static int
-getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
+getSwapInfo( __DRIdrawable *dPriv, __DRIswapInfo * sInfo )
 {
     struct radeon_framebuffer *rfb;
 
@@ -1800,3 +1751,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .InitScreen2     = radeonInitScreen2,
 };
 
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    &driDRI2Extension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.h b/src/mesa/drivers/dri/radeon/radeon_screen.h
index 15744e8828..5e6d432e11 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.h
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.h
@@ -86,7 +86,7 @@ typedef struct radeon_screen {
 
    __volatile__ uint32_t *scratch;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
    unsigned int sarea_priv_offset;
    unsigned int gart_buffer_offset;	/* offset in card memory space */
    unsigned int gart_texture_offset;	/* offset in card memory space */
@@ -123,5 +123,5 @@ typedef struct radeon_screen {
 #define IS_R600_CLASS(screen) \
 	((screen->chip_flags & RADEON_CLASS_MASK) == RADEON_CLASS_R600)
 
-extern void radeonDestroyBuffer(__DRIdrawablePrivate *driDrawPriv);
+extern void radeonDestroyBuffer(__DRIdrawable *driDrawPriv);
 #endif /* __RADEON_SCREEN_H__ */
diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c
index 0c49c3713a..8db3d2b143 100644
--- a/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -41,6 +41,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
 #include "main/glheader.h"
+#include "main/texformat.h"
 #include "swrast/swrast.h"
 
 #include "radeon_common.h"
@@ -334,22 +335,6 @@ static GLubyte *radeon_ptr_2byte_8x2(const struct radeon_renderbuffer * rrb,
 
 #endif
 
-#ifndef RADEON_R300
-#ifndef RADEON_R600
-static uint32_t
-z24s8_to_s8z24(uint32_t val)
-{
-   return (val << 24) | (val >> 8);
-}
-
-static uint32_t
-s8z24_to_z24s8(uint32_t val)
-{
-   return (val >> 24) | (val << 8);
-}
-#endif
-#endif
-
 /*
  * Note that all information needed to access pixels in a renderbuffer
  * should be obtained through the gl_renderbuffer parameter, not per-context
@@ -416,6 +401,18 @@ s8z24_to_z24s8(uint32_t val)
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5_REV
+
+#define TAG(x)    radeon##x##_RGB565_REV
+#define TAG2(x,y) radeon##x##_RGB565_REV##y
+#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)
+#endif
+#include "spantmp2.h"
+
 /* 16 bit, ARGB1555 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -430,6 +427,18 @@ s8z24_to_z24s8(uint32_t val)
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_1_5_5_5
+
+#define TAG(x)    radeon##x##_ARGB1555_REV
+#define TAG2(x,y) radeon##x##_ARGB1555_REV##y
+#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)
+#endif
+#include "spantmp2.h"
+
 /* 16 bit, RGBA4 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -444,6 +453,18 @@ s8z24_to_z24s8(uint32_t val)
 #endif
 #include "spantmp2.h"
 
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_4_4_4_4
+
+#define TAG(x)    radeon##x##_ARGB4444_REV
+#define TAG2(x,y) radeon##x##_ARGB4444_REV##y
+#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)
+#endif
+#include "spantmp2.h"
+
 /* 32 bit, xRGB8888 color spanline and pixel functions
  */
 #define SPANTMP_PIXEL_FMT GL_BGRA
@@ -488,6 +509,42 @@ s8z24_to_z24s8(uint32_t val)
 #endif
 #include "spantmp2.h"
 
+/* 32 bit, BGRx8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8
+
+#define TAG(x)    radeon##x##_BGRx8888
+#define TAG2(x,y) radeon##x##_BGRx8888##y
+#if defined(RADEON_R600)
+#define GET_VALUE(_x, _y) ((*(GLuint*)(r600_ptr_color(rrb, _x + x_off, _y + y_off)) | 0x000000ff))
+#define PUT_VALUE(_x, _y, d) { \
+   GLuint *_ptr = (GLuint*)r600_ptr_color( rrb, _x + x_off, _y + y_off );		\
+   *_ptr = d;								\
+} while (0)
+#else
+#define GET_VALUE(_x, _y) ((*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)) | 0x000000ff))
+#define PUT_VALUE(_x, _y, d) { \
+   GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
+   *_ptr = d;								\
+} while (0)
+#endif
+#include "spantmp2.h"
+
+/* 32 bit, BGRA8888 color spanline and pixel functions
+ */
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8
+
+#define TAG(x)    radeon##x##_BGRA8888
+#define TAG2(x,y) radeon##x##_BGRA8888##y
+#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_4byte(rrb, (X) + x_off, (Y) + y_off)
+#endif
+#include "spantmp2.h"
+
 /* ================================================================
  * Depth buffer
  */
@@ -542,10 +599,10 @@ s8z24_to_z24s8(uint32_t val)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x000000ff;							\
    tmp |= ((d << 8) & 0xffffff00);					\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
@@ -560,26 +617,26 @@ do {									\
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xff000000;							\
    tmp |= ((d) & 0x00ffffff);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #else
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );	\
-   GLuint tmp = *_ptr;							\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xff000000;							\
    tmp |= ((d) & 0x00ffffff);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #endif
 
 #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; \
+    d = (LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))) & 0xffffff00) >> 8; \
   }while(0)
 #elif defined(RADEON_R600)
 #define READ_DEPTH( d, _x, _y )						\
@@ -589,11 +646,11 @@ do {									\
 #elif defined(RADEON_R200)
 #define READ_DEPTH( d, _x, _y )						\
   do {									\
-    d = *(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off)) & 0x00ffffff; \
+    d = LE32_TO_CPU(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))) & 0x00ffffff; \
   }while(0)
 #else
 #define READ_DEPTH( d, _x, _y )	\
-  d = *(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off)) & 0x00ffffff;
+  d = LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off))) & 0x00ffffff;
 #endif
 
 #define TAG(x) radeon##x##_z24
@@ -611,7 +668,7 @@ do {									\
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   *_ptr = d;								\
+   *_ptr = CPU_TO_LE32((((d) & 0xff000000) >> 24) | (((d) & 0x00ffffff) << 8));   \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_DEPTH( _x, _y, d )					\
@@ -619,53 +676,52 @@ do {									\
    GLuint *_ptr = (GLuint*)r600_ptr_depth( rrb, _x + x_off, _y + y_off );		\
    GLuint tmp = *_ptr;				\
    tmp &= 0xff000000;							\
-   tmp |= (((d) >> 8) & 0x00ffffff);					\
+   tmp |= ((d) & 0x00ffffff);					\
    *_ptr = tmp;					\
    _ptr = (GLuint*)r600_ptr_stencil(rrb, _x + x_off, _y + y_off);		\
    tmp = *_ptr;				\
    tmp &= 0xffffff00;							\
-   tmp |= (d) & 0xff;							\
+   tmp |= ((d) >> 24) & 0xff;						\
    *_ptr = tmp;					\
 } while (0)
 #elif defined(RADEON_R200)
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = z24s8_to_s8z24(d);					\
-   *_ptr = tmp;								\
+   *_ptr = CPU_TO_LE32(d);						\
 } while (0)
 #else
 #define WRITE_DEPTH( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );	\
-   GLuint tmp = z24s8_to_s8z24(d);					\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(d);						\
 } while (0)
 #endif
 
 #if defined(RADEON_R300)
 #define READ_DEPTH( d, _x, _y )						\
   do { \
-    d = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)));	\
+    GLuint tmp = (*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off)));	\
+    d = LE32_TO_CPU(((tmp & 0x000000ff) << 24) | ((tmp & 0xffffff00) >> 8));	\
   }while(0)
 #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;	\
+    d = (*(GLuint*)(r600_ptr_depth(rrb, _x + x_off, _y + y_off))) & 0x00ffffff; \
+    d |= ((*(GLuint*)(r600_ptr_stencil(rrb, _x + x_off, _y + y_off))) << 24) & 0xff000000; \
   }while(0)
 #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)));	\
+    d = LE32_TO_CPU(*(GLuint*)(r200_depth_4byte(rrb, _x + x_off, _y + y_off))); \
   }while(0)
 #else
 #define READ_DEPTH( d, _x, _y )	do {					\
-    d = s8z24_to_z24s8(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off,	_y + y_off ))); \
+    d = LE32_TO_CPU(*(GLuint*)(radeon_ptr_4byte(rrb, _x + x_off, _y + y_off))); \
   } while (0)
 #endif
 
-#define TAG(x) radeon##x##_z24_s8
+#define TAG(x) radeon##x##_s8_z24
 #include "depthtmp.h"
 
 /* ================================================================
@@ -678,10 +734,10 @@ do {									\
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0xffffff00;							\
    tmp |= (d) & 0xff;							\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #elif defined(RADEON_R600)
 #define WRITE_STENCIL( _x, _y, d )					\
@@ -696,19 +752,19 @@ do {									\
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x00ffffff;							\
    tmp |= (((d) & 0xff) << 24);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #else
 #define WRITE_STENCIL( _x, _y, d )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte(rrb, _x + x_off, _y + y_off);		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    tmp &= 0x00ffffff;							\
    tmp |= (((d) & 0xff) << 24);						\
-   *_ptr = tmp;					\
+   *_ptr = CPU_TO_LE32(tmp);                                            \
 } while (0)
 #endif
 
@@ -716,7 +772,7 @@ do {									\
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = tmp & 0x000000ff;						\
 } while (0)
 #elif defined(RADEON_R600)
@@ -730,19 +786,19 @@ do {									\
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)r200_depth_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = (tmp & 0xff000000) >> 24;					\
 } while (0)
 #else
 #define READ_STENCIL( d, _x, _y )					\
 do {									\
    GLuint *_ptr = (GLuint*)radeon_ptr_4byte( rrb, _x + x_off, _y + y_off );		\
-   GLuint tmp = *_ptr;				\
+   GLuint tmp = LE32_TO_CPU(*_ptr);                                     \
    d = (tmp & 0xff000000) >> 24;					\
 } while (0)
 #endif
 
-#define TAG(x) radeon##x##_z24_s8
+#define TAG(x) radeon##x##_s8_z24
 #include "stenciltmp.h"
 
 
@@ -755,8 +811,7 @@ static void map_unmap_rb(struct gl_renderbuffer *rb, int flag)
 		return;
 
 	if (flag) {
-		if (rrb->bo->bom->funcs->bo_wait)
-			radeon_bo_wait(rrb->bo);
+	        radeon_bo_wait(rrb->bo);
 		r = radeon_bo_map(rrb->bo, 1);
 		if (r) {
 			fprintf(stderr, "(%s) error(%d) mapping buffer.\n",
@@ -772,18 +827,21 @@ static void map_unmap_rb(struct gl_renderbuffer *rb, int flag)
 }
 
 static void
-radeon_map_unmap_buffers(GLcontext *ctx, GLboolean map)
+radeon_map_unmap_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb,
+			     GLboolean map)
 {
 	GLuint i, j;
 
 	/* color draw buffers */
 	for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers; j++)
-		map_unmap_rb(ctx->DrawBuffer->_ColorDrawBuffers[j], map);
+		map_unmap_rb(fb->_ColorDrawBuffers[j], map);
+
+	map_unmap_rb(fb->_ColorReadBuffer, map);
 
 	/* 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. Note that a mipmapped texture need not
@@ -799,15 +857,15 @@ radeon_map_unmap_buffers(GLcontext *ctx, GLboolean map)
 				radeon_teximage_unmap(image);
 		}
 	}
-
-	map_unmap_rb(ctx->ReadBuffer->_ColorReadBuffer, map);
-
+	
 	/* depth buffer (Note wrapper!) */
-	if (ctx->DrawBuffer->_DepthBuffer)
-		map_unmap_rb(ctx->DrawBuffer->_DepthBuffer->Wrapped, map);
+	if (fb->_DepthBuffer)
+		map_unmap_rb(fb->_DepthBuffer->Wrapped, map);
+
+	if (fb->_StencilBuffer)
+		map_unmap_rb(fb->_StencilBuffer->Wrapped, map);
 
-	if (ctx->DrawBuffer->_StencilBuffer)
-		map_unmap_rb(ctx->DrawBuffer->_StencilBuffer->Wrapped, map);
+	radeon_check_front_buffer_rendering(ctx);
 }
 
 static void radeonSpanRenderStart(GLcontext * ctx)
@@ -832,23 +890,30 @@ static void radeonSpanRenderStart(GLcontext * ctx)
 			ctx->Driver.MapTexture(ctx, ctx->Texture.Unit[i]._Current);
 	}
 
-	radeon_map_unmap_buffers(ctx, 1);
+	radeon_map_unmap_framebuffer(ctx, ctx->DrawBuffer, GL_TRUE);
+	if (ctx->ReadBuffer != ctx->DrawBuffer)
+		radeon_map_unmap_framebuffer(ctx, ctx->ReadBuffer, GL_TRUE);
 }
 
 static void radeonSpanRenderFinish(GLcontext * ctx)
 {
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	int i;
+
 	_swrast_flush(ctx);
-	if (!rmesa->radeonScreen->driScreen->dri2.enabled) {
-		UNLOCK_HARDWARE(rmesa);
-	}
+
 	for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
 		if (ctx->Texture.Unit[i]._ReallyEnabled)
 			ctx->Driver.UnmapTexture(ctx, ctx->Texture.Unit[i]._Current);
 	}
 
-	radeon_map_unmap_buffers(ctx, 0);
+	radeon_map_unmap_framebuffer(ctx, ctx->DrawBuffer, GL_FALSE);
+	if (ctx->ReadBuffer != ctx->DrawBuffer)
+		radeon_map_unmap_framebuffer(ctx, ctx->ReadBuffer, GL_FALSE);
+
+	if (!rmesa->radeonScreen->driScreen->dri2.enabled) {
+		UNLOCK_HARDWARE(rmesa);
+	}
 }
 
 void radeonInitSpanFuncs(GLcontext * ctx)
@@ -864,25 +929,35 @@ void radeonInitSpanFuncs(GLcontext * ctx)
  */
 static void radeonSetSpanFunctions(struct radeon_renderbuffer *rrb)
 {
-	if (rrb->base._ActualFormat == GL_RGB5) {
+	if (rrb->base.Format == MESA_FORMAT_RGB565) {
 		radeonInitPointers_RGB565(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_RGB8) {
+	} else if (rrb->base.Format == MESA_FORMAT_RGB565_REV) {
+		radeonInitPointers_RGB565_REV(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_XRGB8888) {
 		radeonInitPointers_xRGB8888(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_RGBA8) {
+        } else if (rrb->base.Format == MESA_FORMAT_XRGB8888_REV) {
+		radeonInitPointers_BGRx8888(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB8888) {
 		radeonInitPointers_ARGB8888(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_RGBA4) {
+        } else if (rrb->base.Format == MESA_FORMAT_ARGB8888_REV) {
+		radeonInitPointers_BGRA8888(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB4444) {
 		radeonInitPointers_ARGB4444(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_RGB5_A1) {
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB4444_REV) {
+		radeonInitPointers_ARGB4444_REV(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB1555) {
 		radeonInitPointers_ARGB1555(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_DEPTH_COMPONENT16) {
+	} else if (rrb->base.Format == MESA_FORMAT_ARGB1555_REV) {
+		radeonInitPointers_ARGB1555_REV(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_Z16) {
 		radeonInitDepthPointers_z16(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_DEPTH_COMPONENT24) {
+	} else if (rrb->base.Format == MESA_FORMAT_X8_Z24) {
 		radeonInitDepthPointers_z24(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
-		radeonInitDepthPointers_z24_s8(&rrb->base);
-	} else if (rrb->base._ActualFormat == GL_STENCIL_INDEX8_EXT) {
-		radeonInitStencilPointers_z24_s8(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_S8_Z24) {
+		radeonInitDepthPointers_s8_z24(&rrb->base);
+	} else if (rrb->base.Format == MESA_FORMAT_S8) {
+		radeonInitStencilPointers_s8_z24(&rrb->base);
 	} else {
-		fprintf(stderr, "radeonSetSpanFunctions: bad actual format: 0x%04X\n", rrb->base._ActualFormat);
+		fprintf(stderr, "radeonSetSpanFunctions: bad format: 0x%04X\n", rrb->base.Format);
 	}
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c
index 4d0d35ee0c..1c9ec36dae 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state.c
@@ -521,10 +521,10 @@ static void radeonColorMask( GLcontext *ctx,
      return;
 
    mask = radeonPackColor( rrb->cpp,
-			   ctx->Color.ColorMask[RCOMP],
-			   ctx->Color.ColorMask[GCOMP],
-			   ctx->Color.ColorMask[BCOMP],
-			   ctx->Color.ColorMask[ACOMP] );
+			   ctx->Color.ColorMask[0][RCOMP],
+			   ctx->Color.ColorMask[0][GCOMP],
+			   ctx->Color.ColorMask[0][BCOMP],
+			   ctx->Color.ColorMask[0][ACOMP] );
 
    if ( rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK] != mask ) {
       RADEON_STATECHANGE( rmesa, msk );
@@ -550,6 +550,31 @@ static void radeonPolygonOffset( GLcontext *ctx,
    rmesa->hw.zbs.cmd[ZBS_SE_ZBIAS_CONSTANT] = constant.ui32;
 }
 
+static void radeonPolygonStipplePreKMS( GLcontext *ctx, const GLubyte *mask )
+{
+   r100ContextPtr rmesa = R100_CONTEXT(ctx);
+   GLuint i;
+   drm_radeon_stipple_t stipple;
+
+   /* Must flip pattern upside down.
+    */
+   for ( i = 0 ; i < 32 ; i++ ) {
+      rmesa->state.stipple.mask[31 - i] = ((GLuint *) mask)[i];
+   }
+
+   /* TODO: push this into cmd mechanism
+    */
+   radeon_firevertices(&rmesa->radeon);
+   LOCK_HARDWARE( &rmesa->radeon );
+
+   /* FIXME: Use window x,y offsets into stipple RAM.
+    */
+   stipple.mask = rmesa->state.stipple.mask;
+   drmCommandWrite( rmesa->radeon.dri.fd, DRM_RADEON_STIPPLE,
+		    &stipple, sizeof(drm_radeon_stipple_t) );
+   UNLOCK_HARDWARE( &rmesa->radeon );
+}
+
 static void radeonPolygonMode( GLcontext *ctx, GLenum face, GLenum mode )
 {
    r100ContextPtr rmesa = R100_CONTEXT(ctx);
@@ -1375,7 +1400,7 @@ static void radeonClearStencil( GLcontext *ctx, GLint s )
 void radeonUpdateWindow( GLcontext *ctx )
 {
    r100ContextPtr rmesa = R100_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLfloat xoffset = dPriv ? (GLfloat) dPriv->x : 0;
    GLfloat yoffset = dPriv ? (GLfloat) dPriv->y + dPriv->h : 0;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -1430,7 +1455,7 @@ static void radeonDepthRange( GLcontext *ctx, GLclampd nearval,
 void radeonUpdateViewportOffset( GLcontext *ctx )
 {
    r100ContextPtr rmesa = R100_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = radeon_get_drawable(&rmesa->radeon);
+   __DRIdrawable *dPriv = radeon_get_drawable(&rmesa->radeon);
    GLfloat xoffset = (GLfloat)dPriv->x;
    GLfloat yoffset = (GLfloat)dPriv->y + dPriv->h;
    const GLfloat *v = ctx->Viewport._WindowMap.m;
diff --git a/src/mesa/drivers/dri/radeon/radeon_state_init.c b/src/mesa/drivers/dri/radeon/radeon_state_init.c
index f3ad0dd17a..dd82888254 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -440,16 +440,18 @@ static void ctx_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom)
    atom->cmd[CTX_RB3D_CNTL] &= ~(0xf << 10);
    if (rrb->cpp == 4)
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB8888;
-   else switch (rrb->base._ActualFormat) {
-   case GL_RGB5:
+   else switch (rrb->base.Format) {
+   case MESA_FORMAT_RGB565:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_RGB565;
 	break;
-   case GL_RGBA4:
+   case MESA_FORMAT_ARGB4444:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB4444;
 	break;
-   case GL_RGB5_A1:
+   case MESA_FORMAT_ARGB1555:
 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB1555;
 	break;
+   default:
+	_mesa_problem(ctx, "unexpected format in ctx_emit_cs()");
    }
 
    cbpitch = (rrb->pitch / rrb->cpp);
@@ -643,11 +645,11 @@ static void tex_emit_cs(GLcontext *ctx, struct radeon_state_atom *atom)
      OUT_BATCH(CP_PACKET0(RADEON_PP_TXOFFSET_0 + (24 * i), 0));
      if (t->mt && !t->image_override) {
         if ((ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_CUBE_BIT)) {
-            lvl = &t->mt->levels[0];
+            lvl = &t->mt->levels[t->minLod];
 	    OUT_BATCH_RELOC(lvl->faces[5].offset, t->mt->bo, lvl->faces[5].offset,
 			RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
         } else {
-           OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
+           OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, get_base_teximage_offset(t),
 		     RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
         }
       } else {
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index e61f59eaea..8bf1bfbc57 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -179,7 +179,7 @@ static void radeonSetVertexFormat( GLcontext *ctx )
 
       for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    GLuint sz = VB->TexCoordPtr[i]->size;
+	    GLuint sz = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size;
 
 	    switch (sz) {
 	    case 1:
@@ -309,7 +309,7 @@ void r100_swtcl_flush(GLcontext *ctx, uint32_t current_offset)
    radeonEmitState(&rmesa->radeon);
    radeonEmitVertexAOS( rmesa,
 			rmesa->radeon.swtcl.vertex_size,
-			first_elem(&rmesa->radeon.dma.reserved)->bo,
+			rmesa->radeon.swtcl.bo,
 			current_offset);
 
 		      
diff --git a/src/mesa/drivers/dri/radeon/radeon_tex.c b/src/mesa/drivers/dri/radeon/radeon_tex.c
index 99865fff27..14163f13af 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tex.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tex.c
@@ -38,7 +38,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/enums.h"
 #include "main/image.h"
 #include "main/simple_list.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
@@ -342,24 +341,14 @@ static void radeonTexParameter( GLcontext *ctx, GLenum target,
       break;
 
    case GL_TEXTURE_BORDER_COLOR:
-      radeonSetTexBorderColor( t, texObj->BorderColor );
+      radeonSetTexBorderColor( t, texObj->BorderColor.f );
       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.  Since there's no LOD clamping,
-       * we just have to rely on loading the right subset of mipmap levels
-       * to simulate a clamped LOD.
-       */
-      if (t->mt) {
-         radeon_miptree_unreference(t->mt);
-	 t->mt = 0;
-	 t->validated = GL_FALSE;
-      }
+      t->validated = GL_FALSE;
       break;
 
    default:
@@ -389,10 +378,8 @@ static void radeonDeleteTexture( GLcontext *ctx,
      }
    }
 
-   if (t->mt) {
-      radeon_miptree_unreference(t->mt);
-      t->mt = 0;
-   }
+   radeon_miptree_unreference(&t->mt);
+
    /* Free mipmap images and the texture object itself */
    _mesa_delete_texture_object(ctx, texObj);
 }
@@ -441,7 +428,7 @@ radeonNewTextureObject( GLcontext *ctx, GLuint name, GLenum target )
    radeonSetTexWrap( t, t->base.WrapS, t->base.WrapT );
    radeonSetTexMaxAnisotropy( t, t->base.MaxAnisotropy );
    radeonSetTexFilter( t, t->base.MinFilter, t->base.MagFilter );
-   radeonSetTexBorderColor( t, t->base.BorderColor );
+   radeonSetTexBorderColor( t, t->base.BorderColor.f );
    return &t->base;
 }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index ae41b90efe..84ddcfd4fd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -38,8 +38,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/colormac.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
+#include "main/texstate.h"
 #include "main/texobj.h"
 #include "main/enums.h"
 
@@ -81,8 +81,10 @@ struct tx_table {
    GLuint format, filter;
 };
 
+/* XXX verify this table against MESA_FORMAT_x values */
 static const struct tx_table tx_table[] =
 {
+   _INVALID(NONE), /* MESA_FORMAT_NONE */
    _ALPHA(RGBA8888),
    _ALPHA_REV(RGBA8888),
    _ALPHA(ARGB8888),
@@ -660,7 +662,7 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 	rmesa = pDRICtx->driverPrivate;
 
 	rfb = dPriv->driverPrivate;
-        texUnit = &radeon->glCtx->Texture.Unit[radeon->glCtx->Texture.CurrentUnit];
+        texUnit = _mesa_get_current_tex_unit(radeon->glCtx);
 	texObj = _mesa_select_tex_object(radeon->glCtx, texUnit, target);
         texImage = _mesa_get_tex_image(radeon->glCtx, texObj, target, 0);
 
@@ -670,24 +672,13 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
     	    return;
     	}
 
-	radeon_update_renderbuffers(pDRICtx, dPriv);
-	/* back & depth buffer are useless free them right away */
-	rb = (void*)rfb->base.Attachment[BUFFER_DEPTH].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-        rb->bo = NULL;
-	}
-	rb = (void*)rfb->base.Attachment[BUFFER_BACK_LEFT].Renderbuffer;
-	if (rb && rb->bo) {
-		radeon_bo_unref(rb->bo);
-		rb->bo = NULL;
-	}
+	radeon_update_renderbuffers(pDRICtx, dPriv, GL_TRUE);
 	rb = rfb->color_rb[0];
 	if (rb->bo == NULL) {
 		/* Failed to BO for the buffer */
 		return;
 	}
-	
+
 	_mesa_lock_texture(radeon->glCtx, texObj);
 	if (t->bo) {
 		radeon_bo_unref(t->bo);
@@ -697,20 +688,14 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 		radeon_bo_unref(rImage->bo);
 		rImage->bo = NULL;
 	}
-	if (t->mt) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = NULL;
-	}
-	if (rImage->mt) {
-		radeon_miptree_unreference(rImage->mt);
-		rImage->mt = NULL;
-	}
+
+	radeon_miptree_unreference(&t->mt);
+	radeon_miptree_unreference(&rImage->mt);
+
 	_mesa_init_teximage_fields(radeon->glCtx, target, texImage,
 				   rb->base.Width, rb->base.Height, 1, 0, rb->cpp);
 	texImage->RowStride = rb->pitch / rb->cpp;
-	texImage->TexFormat = radeonChooseTextureFormat(radeon->glCtx,
-							internalFormat,
-							type, format, 0);
+
 	rImage->bo = rb->bo;
 	radeon_bo_ref(rImage->bo);
 	t->bo = rb->bo;
@@ -718,8 +703,6 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 	t->tile_bits = 0;
 	t->image_override = GL_TRUE;
 	t->override_offset = 0;
-	t->pp_txpitch &= (1 << 13) -1;
-	pitch_val = rb->pitch;
 	switch (rb->cpp) {
 	case 4:
 		if (glx_texture_format == GLX_TEXTURE_FORMAT_RGB_EXT)
@@ -738,12 +721,17 @@ void radeonSetTexBuffer2(__DRIcontext *pDRICtx, GLint target, GLint glx_texture_
 		t->pp_txfilter |= tx_table[MESA_FORMAT_RGB565].filter;
 		break;
 	}
-        t->pp_txsize = ((rb->base.Width - 1) << RADEON_TEX_USIZE_SHIFT)
-		   | ((rb->base.Height - 1) << RADEON_TEX_VSIZE_SHIFT);
-        t->pp_txformat |= RADEON_TXFORMAT_NON_POWER2;
-	t->pp_txpitch = pitch_val;
-        t->pp_txpitch -= 32;
 
+	t->pp_txpitch &= (1 << 13) -1;
+	pitch_val = rb->pitch;
+
+        t->pp_txsize = ((rb->base.Width - 1) << RADEON_TEX_USIZE_SHIFT)
+		| ((rb->base.Height - 1) << RADEON_TEX_VSIZE_SHIFT);
+	if (target == GL_TEXTURE_RECTANGLE_NV) {
+		t->pp_txformat |= RADEON_TXFORMAT_NON_POWER2;
+		t->pp_txpitch = pitch_val;
+		t->pp_txpitch -= 32;
+	}
 	t->validated = GL_TRUE;
 	_mesa_unlock_texture(radeon->glCtx, texObj);
 	return;
@@ -1021,7 +1009,7 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 	return GL_TRUE;
    }
 
-   firstImage = t->base.Image[0][t->mt->firstLevel];   
+   firstImage = t->base.Image[0][t->minLod];
 
    if (firstImage->Border > 0) {
       fprintf(stderr, "%s: border\n", __FUNCTION__);
@@ -1031,27 +1019,27 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
    log2Width  = firstImage->WidthLog2;
    log2Height = firstImage->HeightLog2;
    log2Depth  = firstImage->DepthLog2;
-   texelBytes = firstImage->TexFormat->TexelBytes;
+   texelBytes = _mesa_get_format_bytes(firstImage->TexFormat);
 
    if (!t->image_override) {
-      if (VALID_FORMAT(firstImage->TexFormat->MesaFormat)) {
+      if (VALID_FORMAT(firstImage->TexFormat)) {
 	const struct tx_table *table = tx_table;
 
 	 t->pp_txformat &= ~(RADEON_TXFORMAT_FORMAT_MASK |
 			     RADEON_TXFORMAT_ALPHA_IN_MAP);
 	 t->pp_txfilter &= ~RADEON_YUV_TO_RGB;	 
 	 
-	 t->pp_txformat |= table[ firstImage->TexFormat->MesaFormat ].format;
-	 t->pp_txfilter |= table[ firstImage->TexFormat->MesaFormat ].filter;
+	 t->pp_txformat |= table[ firstImage->TexFormat ].format;
+	 t->pp_txfilter |= table[ firstImage->TexFormat ].filter;
       } else {
 	 _mesa_problem(NULL, "unexpected texture format in %s",
 		       __FUNCTION__);
 	 return GL_FALSE;
       }
    }
-   
+
    t->pp_txfilter &= ~RADEON_MAX_MIP_LEVEL_MASK;
-   t->pp_txfilter |= (t->mt->lastLevel - t->mt->firstLevel) << RADEON_MAX_MIP_LEVEL_SHIFT;
+   t->pp_txfilter |= (t->maxLod - t->minLod) << RADEON_MAX_MIP_LEVEL_SHIFT;
 	
    t->pp_txformat &= ~(RADEON_TXFORMAT_WIDTH_MASK |
 		       RADEON_TXFORMAT_HEIGHT_MASK |
@@ -1060,9 +1048,9 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 		       RADEON_TXFORMAT_F5_HEIGHT_MASK);
    t->pp_txformat |= ((log2Width << RADEON_TXFORMAT_WIDTH_SHIFT) |
 		      (log2Height << RADEON_TXFORMAT_HEIGHT_SHIFT));
-   
+
    t->tile_bits = 0;
-   
+
    if (t->base.Target == GL_TEXTURE_CUBE_MAP) {
       ASSERT(log2Width == log2Height);
       t->pp_txformat |= ((log2Width << RADEON_TXFORMAT_F5_WIDTH_SHIFT) |
@@ -1083,7 +1071,7 @@ static GLboolean setup_hardware_state(r100ContextPtr rmesa, radeonTexObj *t, int
 		   | ((firstImage->Height - 1) << RADEON_TEX_VSIZE_SHIFT));
 
    if ( !t->image_override ) {
-      if (firstImage->IsCompressed)
+      if (_mesa_is_format_compressed(firstImage->TexFormat))
          t->pp_txpitch = (firstImage->Width + 63) & ~(63);
       else
          t->pp_txpitch = ((firstImage->Width * texelBytes) + 63) & ~(63);
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.c b/src/mesa/drivers/dri/radeon/radeon_texture.c
index 7b7392b217..03178116c1 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.c
@@ -1,4 +1,5 @@
 /*
+ * Copyright (C) 2009 Maciej Cencora.
  * Copyright (C) 2008 Nicolai Haehnle.
  * Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
  *
@@ -34,7 +35,6 @@
 #include "main/convolve.h"
 #include "main/mipmap.h"
 #include "main/texcompress.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
@@ -47,7 +47,7 @@
 #include "radeon_mipmap_tree.h"
 
 
-static void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
+void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
 	GLuint numrows, GLuint rowsize)
 {
 	assert(rowsize <= dststride);
@@ -82,8 +82,7 @@ void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage)
 	radeon_texture_image* image = get_radeon_texture_image(timage);
 
 	if (image->mt) {
-		radeon_miptree_unreference(image->mt);
-		image->mt = 0;
+		radeon_miptree_unreference(&image->mt);
 		assert(!image->base.Data);
 	} else {
 		_mesa_free_texture_image_data(ctx, timage);
@@ -109,7 +108,7 @@ static void teximage_set_map_data(radeon_texture_image *image)
 	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;
+	image->base.RowStride = lvl->rowstride / _mesa_get_format_bytes(image->base.TexFormat);
 }
 
 
@@ -144,7 +143,6 @@ static void map_override(GLcontext *ctx, radeonTexObj *t)
 	radeon_bo_map(t->bo, GL_FALSE);
 
 	img->base.Data = t->bo->ptr;
-	_mesa_set_fetch_functions(&img->base, 2);
 }
 
 static void unmap_override(GLcontext *ctx, radeonTexObj *t)
@@ -176,7 +174,7 @@ void radeonMapTexture(GLcontext *ctx, struct gl_texture_object *texObj)
 
 	radeon_bo_map(t->mt->bo, GL_FALSE);
 	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level)
+		for(level = t->minLod; level <= t->maxLod; ++level)
 			teximage_set_map_data(get_radeon_texture_image(texObj->Image[face][level]));
 	}
 }
@@ -193,7 +191,7 @@ void radeonUnmapTexture(GLcontext *ctx, struct gl_texture_object *texObj)
 	  return;
 
 	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level)
+		for(level = t->minLod; level <= t->maxLod; ++level)
 			texObj->Image[face][level]->Data = 0;
 	}
 	radeon_bo_unmap(t->mt->bo);
@@ -242,8 +240,7 @@ static void radeon_generate_mipmap(GLcontext *ctx, GLenum target,
 			image->mtlevel = i;
 			image->mtface = face;
 
-			radeon_miptree_unreference(image->mt);
-			image->mt = NULL;
+			radeon_miptree_unreference(&image->mt);
 		}
 	}
 	
@@ -261,9 +258,9 @@ void radeonGenerateMipmap(GLcontext* ctx, GLenum target, struct gl_texture_objec
 
 
 /* try to find a format which will only need a memcopy */
-static const struct gl_texture_format *radeonChoose8888TexFormat(radeonContextPtr rmesa,
-								 GLenum srcFormat,
-								 GLenum srcType, GLboolean fbo)
+static gl_format radeonChoose8888TexFormat(radeonContextPtr rmesa,
+					   GLenum srcFormat,
+					   GLenum srcType, GLboolean fbo)
 {
 	const GLuint ui = 1;
 	const GLubyte littleEndian = *((const GLubyte *)&ui);
@@ -276,37 +273,37 @@ static const struct gl_texture_format *radeonChoose8888TexFormat(radeonContextPt
 	    (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
 	    (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_INT_8_8_8_8_REV) ||
 	    (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE && littleEndian)) {
-		return &_mesa_texformat_rgba8888;
+		return MESA_FORMAT_RGBA8888;
 	} else if ((srcFormat == GL_RGBA && srcType == GL_UNSIGNED_INT_8_8_8_8_REV) ||
 		   (srcFormat == GL_RGBA && srcType == GL_UNSIGNED_BYTE && littleEndian) ||
 		   (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_INT_8_8_8_8) ||
 		   (srcFormat == GL_ABGR_EXT && srcType == GL_UNSIGNED_BYTE && !littleEndian)) {
-		return &_mesa_texformat_rgba8888_rev;
+		return MESA_FORMAT_RGBA8888_REV;
 	} else if (IS_R200_CLASS(rmesa->radeonScreen)) {
 		return _dri_texformat_argb8888;
 	} else if (srcFormat == GL_BGRA && ((srcType == GL_UNSIGNED_BYTE && !littleEndian) ||
 					    srcType == GL_UNSIGNED_INT_8_8_8_8)) {
-		return &_mesa_texformat_argb8888_rev;
+		return MESA_FORMAT_ARGB8888_REV;
 	} else if (srcFormat == GL_BGRA && ((srcType == GL_UNSIGNED_BYTE && littleEndian) ||
 					    srcType == GL_UNSIGNED_INT_8_8_8_8_REV)) {
-		return &_mesa_texformat_argb8888;
+		return MESA_FORMAT_ARGB8888;
 	} else
 		return _dri_texformat_argb8888;
 }
 
-const struct gl_texture_format *radeonChooseTextureFormat_mesa(GLcontext * ctx,
-							  GLint internalFormat,
-							  GLenum format,
-							  GLenum type)
+gl_format radeonChooseTextureFormat_mesa(GLcontext * ctx,
+					 GLint internalFormat,
+					 GLenum format,
+					 GLenum type)
 {
 	return radeonChooseTextureFormat(ctx, internalFormat, format,
 					 type, 0);
 }
 
-const struct gl_texture_format *radeonChooseTextureFormat(GLcontext * ctx,
-							  GLint internalFormat,
-							  GLenum format,
-							  GLenum type, GLboolean fbo)
+gl_format radeonChooseTextureFormat(GLcontext * ctx,
+				    GLint internalFormat,
+				    GLenum format,
+				    GLenum type, GLboolean fbo)
 {
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	const GLboolean do32bpt =
@@ -430,58 +427,72 @@ const struct gl_texture_format *radeonChooseTextureFormat(GLcontext * ctx,
 	case GL_YCBCR_MESA:
 		if (type == GL_UNSIGNED_SHORT_8_8_APPLE ||
 		    type == GL_UNSIGNED_BYTE)
-			return &_mesa_texformat_ycbcr;
+			return MESA_FORMAT_YCBCR;
 		else
-			return &_mesa_texformat_ycbcr_rev;
+			return MESA_FORMAT_YCBCR_REV;
 
 	case GL_RGB_S3TC:
 	case GL_RGB4_S3TC:
 	case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-		return &_mesa_texformat_rgb_dxt1;
+		return MESA_FORMAT_RGB_DXT1;
 
 	case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-		return &_mesa_texformat_rgba_dxt1;
+		return MESA_FORMAT_RGBA_DXT1;
 
 	case GL_RGBA_S3TC:
 	case GL_RGBA4_S3TC:
 	case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-		return &_mesa_texformat_rgba_dxt3;
+		return MESA_FORMAT_RGBA_DXT3;
 
 	case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-		return &_mesa_texformat_rgba_dxt5;
+		return MESA_FORMAT_RGBA_DXT5;
 
 	case GL_ALPHA16F_ARB:
-		return &_mesa_texformat_alpha_float16;
+		return MESA_FORMAT_ALPHA_FLOAT16;
 	case GL_ALPHA32F_ARB:
-		return &_mesa_texformat_alpha_float32;
+		return MESA_FORMAT_ALPHA_FLOAT32;
 	case GL_LUMINANCE16F_ARB:
-		return &_mesa_texformat_luminance_float16;
+		return MESA_FORMAT_LUMINANCE_FLOAT16;
 	case GL_LUMINANCE32F_ARB:
-		return &_mesa_texformat_luminance_float32;
+		return MESA_FORMAT_LUMINANCE_FLOAT32;
 	case GL_LUMINANCE_ALPHA16F_ARB:
-		return &_mesa_texformat_luminance_alpha_float16;
+		return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT16;
 	case GL_LUMINANCE_ALPHA32F_ARB:
-		return &_mesa_texformat_luminance_alpha_float32;
+		return MESA_FORMAT_LUMINANCE_ALPHA_FLOAT32;
 	case GL_INTENSITY16F_ARB:
-		return &_mesa_texformat_intensity_float16;
+		return MESA_FORMAT_INTENSITY_FLOAT16;
 	case GL_INTENSITY32F_ARB:
-		return &_mesa_texformat_intensity_float32;
+		return MESA_FORMAT_INTENSITY_FLOAT32;
 	case GL_RGB16F_ARB:
-		return &_mesa_texformat_rgba_float16;
+		return MESA_FORMAT_RGBA_FLOAT16;
 	case GL_RGB32F_ARB:
-		return &_mesa_texformat_rgba_float32;
+		return MESA_FORMAT_RGBA_FLOAT32;
 	case GL_RGBA16F_ARB:
-		return &_mesa_texformat_rgba_float16;
+		return MESA_FORMAT_RGBA_FLOAT16;
 	case GL_RGBA32F_ARB:
-		return &_mesa_texformat_rgba_float32;
+		return MESA_FORMAT_RGBA_FLOAT32;
 
+#ifdef RADEON_R300
 	case GL_DEPTH_COMPONENT:
 	case GL_DEPTH_COMPONENT16:
+		return MESA_FORMAT_Z16;
 	case GL_DEPTH_COMPONENT24:
 	case GL_DEPTH_COMPONENT32:
 	case GL_DEPTH_STENCIL_EXT:
 	case GL_DEPTH24_STENCIL8_EXT:
-		return &_mesa_texformat_s8_z24;
+		if (rmesa->radeonScreen->chip_family >= CHIP_FAMILY_RV515)
+			return MESA_FORMAT_S8_Z24;
+		else
+			return MESA_FORMAT_Z16;
+#else
+	case GL_DEPTH_COMPONENT:
+	case GL_DEPTH_COMPONENT16:
+	case GL_DEPTH_COMPONENT24:
+	case GL_DEPTH_COMPONENT32:
+	case GL_DEPTH_STENCIL_EXT:
+	case GL_DEPTH24_STENCIL8_EXT:
+		return MESA_FORMAT_S8_Z24;
+#endif
 
 	/* EXT_texture_sRGB */
 	case GL_SRGB:
@@ -490,26 +501,193 @@ const struct gl_texture_format *radeonChooseTextureFormat(GLcontext * ctx,
 	case GL_SRGB8_ALPHA8:
 	case GL_COMPRESSED_SRGB:
 	case GL_COMPRESSED_SRGB_ALPHA:
-		return &_mesa_texformat_srgba8;
+		return MESA_FORMAT_SRGBA8;
 
 	case GL_SLUMINANCE:
 	case GL_SLUMINANCE8:
 	case GL_COMPRESSED_SLUMINANCE:
-		return &_mesa_texformat_sl8;
+		return MESA_FORMAT_SL8;
 
 	case GL_SLUMINANCE_ALPHA:
 	case GL_SLUMINANCE8_ALPHA8:
 	case GL_COMPRESSED_SLUMINANCE_ALPHA:
-		return &_mesa_texformat_sla8;
+		return MESA_FORMAT_SLA8;
 
 	default:
 		_mesa_problem(ctx,
 			      "unexpected internalFormat 0x%x in %s",
 			      (int)internalFormat, __func__);
+		return MESA_FORMAT_NONE;
+	}
+
+	return MESA_FORMAT_NONE;		/* never get here */
+}
+
+/** Check if given image is valid within current texture object.
+ */
+static int image_matches_texture_obj(struct gl_texture_object *texObj,
+	struct gl_texture_image *texImage,
+	unsigned level)
+{
+	const struct gl_texture_image *baseImage = texObj->Image[0][texObj->BaseLevel];
+
+	if (!baseImage)
+		return 0;
+
+	if (level < texObj->BaseLevel || level > texObj->MaxLevel)
+		return 0;
+
+	const unsigned levelDiff = level - texObj->BaseLevel;
+	const unsigned refWidth = MAX2(baseImage->Width >> levelDiff, 1);
+	const unsigned refHeight = MAX2(baseImage->Height >> levelDiff, 1);
+	const unsigned refDepth = MAX2(baseImage->Depth >> levelDiff, 1);
+
+	return (texImage->Width == refWidth &&
+			texImage->Height == refHeight &&
+			texImage->Depth == refDepth);
+}
+
+static void teximage_assign_miptree(radeonContextPtr rmesa,
+	struct gl_texture_object *texObj,
+	struct gl_texture_image *texImage,
+	unsigned face,
+	unsigned level)
+{
+	radeonTexObj *t = radeon_tex_obj(texObj);
+	radeon_texture_image* image = get_radeon_texture_image(texImage);
+
+	/* Since miptree holds only images for levels <BaseLevel..MaxLevel>
+	 * don't allocate the miptree if the teximage won't fit.
+	 */
+	if (!image_matches_texture_obj(texObj, texImage, level))
+		return;
+
+	/* Try using current miptree, or create new if there isn't any */
+	if (!t->mt || !radeon_miptree_matches_image(t->mt, texImage, face, level)) {
+		radeon_miptree_unreference(&t->mt);
+		radeon_try_alloc_miptree(rmesa, t);
+		if (RADEON_DEBUG & RADEON_TEXTURE) {
+			fprintf(stderr, "%s: texObj %p, texImage %p, face %d, level %d, "
+				"texObj miptree doesn't match, allocated new miptree %p\n",
+				__FUNCTION__, texObj, texImage, face, level, t->mt);
+		}
+	}
+
+	/* Miptree alocation may have failed,
+	 * when there was no image for baselevel specified */
+	if (t->mt) {
+		image->mtface = face;
+		image->mtlevel = level;
+		radeon_miptree_reference(t->mt, &image->mt);
+	}
+}
+
+static GLuint * allocate_image_offsets(GLcontext *ctx,
+	unsigned alignedWidth,
+	unsigned height,
+	unsigned depth)
+{
+	int i;
+	GLuint *offsets;
+
+	offsets = _mesa_malloc(depth * sizeof(GLuint)) ;
+	if (!offsets) {
+		_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTex[Sub]Image");
 		return NULL;
 	}
 
-	return NULL;		/* never get here */
+	for (i = 0; i < depth; ++i) {
+		offsets[i] = alignedWidth * height * i;
+	}
+
+	return offsets;
+}
+
+/**
+ * Update a subregion of the given texture image.
+ */
+static void radeon_store_teximage(GLcontext* ctx, int dims,
+		GLint xoffset, GLint yoffset, GLint zoffset,
+		GLsizei width, GLsizei height, GLsizei depth,
+		GLsizei imageSize,
+		GLenum format, GLenum type,
+		const GLvoid * pixels,
+		const struct gl_pixelstore_attrib *packing,
+		struct gl_texture_object *texObj,
+		struct gl_texture_image *texImage,
+		int compressed)
+{
+	radeonTexObj *t = radeon_tex_obj(texObj);
+	radeon_texture_image* image = get_radeon_texture_image(texImage);
+
+	GLuint dstRowStride;
+	GLuint *dstImageOffsets;
+
+	if (image->mt) {
+		dstRowStride = image->mt->levels[image->mtlevel].rowstride;
+	} else if (t->bo) {
+		/* TFP case */
+		/* TODO */
+		assert(0);
+	} else {
+		dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
+	}
+
+	assert(dstRowStride);
+
+	if (dims == 3) {
+		unsigned alignedWidth = dstRowStride/_mesa_get_format_bytes(texImage->TexFormat);
+		dstImageOffsets = allocate_image_offsets(ctx, alignedWidth, texImage->Height, texImage->Depth);
+		if (!dstImageOffsets) {
+			return;
+		}
+	} else {
+		dstImageOffsets = texImage->ImageOffsets;
+	}
+
+	radeon_teximage_map(image, GL_TRUE);
+
+	if (compressed) {
+		uint32_t srcRowStride, bytesPerRow, rows, block_width, block_height;
+		GLubyte *img_start;
+
+		_mesa_get_format_block_size(texImage->TexFormat, &block_width, &block_height);
+
+		if (!image->mt) {
+			dstRowStride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
+			img_start = _mesa_compressed_image_address(xoffset, yoffset, 0,
+									texImage->TexFormat,
+									texImage->Width, texImage->Data);
+		}
+		else {
+			uint32_t offset;
+			offset = dstRowStride / _mesa_get_format_bytes(texImage->TexFormat) * yoffset / block_height + xoffset / block_width;
+			offset *= _mesa_get_format_bytes(texImage->TexFormat);
+			img_start = texImage->Data + offset;
+		}
+		srcRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
+		bytesPerRow = srcRowStride;
+		rows = (height + block_height - 1) / block_height;
+
+		copy_rows(img_start, dstRowStride, pixels, srcRowStride, rows, bytesPerRow);
+	}
+	else {
+		if (!_mesa_texstore(ctx, dims, texImage->_BaseFormat,
+					texImage->TexFormat, texImage->Data,
+					xoffset, yoffset, zoffset,
+					dstRowStride,
+					dstImageOffsets,
+					width, height, depth,
+					format, type, pixels, packing)) {
+			_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
+		}
+	}
+
+	if (dims == 3) {
+		_mesa_free(dstImageOffsets);
+	}
+
+	radeon_teximage_unmap(image);
 }
 
 /**
@@ -530,13 +708,22 @@ static void radeon_teximage(
 	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 	radeonTexObj* t = radeon_tex_obj(texObj);
 	radeon_texture_image* image = get_radeon_texture_image(texImage);
-	GLuint dstRowStride;
 	GLint postConvWidth = width;
 	GLint postConvHeight = height;
-	GLuint texelBytes;
 	GLuint face = radeon_face_for_target(target);
 
-	radeon_firevertices(rmesa);
+	{
+		struct radeon_bo *bo;
+		bo = !image->mt ? image->bo : image->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->cmdbuf.cs)) {
+			radeon_firevertices(rmesa);
+		}
+	}
+
+	if (RADEON_DEBUG & RADEON_TEXTURE) {
+		fprintf(stderr, "radeon_teximage%dd: texObj %p, texImage %p, face %d, level %d\n",
+				dims, texObj, texImage, face, level);
+	}
 
 	t->validated = GL_FALSE;
 
@@ -545,62 +732,35 @@ static void radeon_teximage(
 						  &postConvHeight);
 	}
 
-	/* Choose and fill in the texture format for this image */
-	texImage->TexFormat = radeonChooseTextureFormat(ctx, internalFormat, format, type, 0);
-	_mesa_set_fetch_functions(texImage, dims);
-
-	if (texImage->TexFormat->TexelBytes == 0) {
-		texelBytes = 0;
-		texImage->IsCompressed = GL_TRUE;
-		texImage->CompressedSize =
-			ctx->Driver.CompressedTextureSize(ctx, texImage->Width,
-					   texImage->Height, texImage->Depth,
-					   texImage->TexFormat->MesaFormat);
-	} else {
-		texImage->IsCompressed = GL_FALSE;
-		texImage->CompressedSize = 0;
-
-		texelBytes = texImage->TexFormat->TexelBytes;
+	if (!_mesa_is_format_compressed(texImage->TexFormat)) {
+		GLuint texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 		/* Minimum pitch of 32 bytes */
 		if (postConvWidth * texelBytes < 32) {
-		  postConvWidth = 32 / texelBytes;
-		  texImage->RowStride = postConvWidth;
+			postConvWidth = 32 / texelBytes;
+			texImage->RowStride = postConvWidth;
 		}
-		if (!image->mt) {      
+		if (!image->mt) {
 			assert(texImage->RowStride == postConvWidth);
 		}
 	}
 
-	/* Allocate memory for image */
-	radeonFreeTexImageData(ctx, texImage); /* Mesa core only clears texImage->Data but not image->mt */
-
-	if (t->mt &&
-	    t->mt->firstLevel == level &&
-	    t->mt->lastLevel == level &&
-	    t->mt->target != GL_TEXTURE_CUBE_MAP_ARB &&
-	    !radeon_miptree_matches_image(t->mt, texImage, face, level)) {
-	  radeon_miptree_unreference(t->mt);
-	  t->mt = NULL;
-	}
-
-	if (!t->mt)
-		radeon_try_alloc_miptree(rmesa, t, image, face, level);
-	if (t->mt && radeon_miptree_matches_image(t->mt, texImage, face, level)) {
-		radeon_mipmap_level *lvl;
-		image->mt = t->mt;
-		image->mtlevel = level - t->mt->firstLevel;
-		image->mtface = face;
-		radeon_miptree_reference(t->mt);
-		lvl = &image->mt->levels[image->mtlevel];
-		dstRowStride = lvl->rowstride;
-	} else {
-		int size;
-		if (texImage->IsCompressed) {
-			size = texImage->CompressedSize;
-		} else {
-			size = texImage->Width * texImage->Height * texImage->Depth * texImage->TexFormat->TexelBytes;
+	/* Mesa core only clears texImage->Data but not image->mt */
+	radeonFreeTexImageData(ctx, texImage);
+
+	if (!t->bo) {
+		teximage_assign_miptree(rmesa, texObj, texImage, face, level);
+		if (!image->mt) {
+			int size = _mesa_format_image_size(texImage->TexFormat,
+								texImage->Width,
+								texImage->Height,
+								texImage->Depth);
+			texImage->Data = _mesa_alloc_texmemory(size);
+			if (RADEON_DEBUG & RADEON_TEXTURE) {
+				fprintf(stderr, "radeon_teximage%dd: texObj %p, texImage %p, "
+					" no miptree assigned, using local memory %p\n",
+					dims, texObj, texImage, texImage->Data);
+			}
 		}
-		texImage->Data = _mesa_alloc_texmemory(size);
 	}
 
 	/* Upload texture image; note that the spec allows pixels to be NULL */
@@ -614,64 +774,16 @@ static void radeon_teximage(
 	}
 
 	if (pixels) {
-		radeon_teximage_map(image, GL_TRUE);
-		if (compressed) {
-			if (image->mt) {
-				uint32_t srcRowStride, bytesPerRow, rows;
-				srcRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
-				bytesPerRow = srcRowStride;
-				rows = (height + 3) / 4;
-				copy_rows(texImage->Data, image->mt->levels[level].rowstride,
-					  pixels, srcRowStride, rows, bytesPerRow);
-			} else {
-				memcpy(texImage->Data, pixels, imageSize);
-			}
-		} else {
-			GLuint dstRowStride;
-			GLuint *dstImageOffsets;
-
-			if (image->mt) {
-				radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel];
-				dstRowStride = lvl->rowstride;
-			} else {
-				dstRowStride = texImage->Width * texImage->TexFormat->TexelBytes;
-			}
-
-			if (dims == 3) {
-				int i;
-
-				dstImageOffsets = _mesa_malloc(depth * sizeof(GLuint)) ;
-				if (!dstImageOffsets)
-					_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
-
-				for (i = 0; i < depth; ++i) {
-					dstImageOffsets[i] = dstRowStride/texImage->TexFormat->TexelBytes * height * i;
-				}
-			} else {
-				dstImageOffsets = texImage->ImageOffsets;
-			}
-
-			if (!texImage->TexFormat->StoreImage(ctx, dims,
-						texImage->_BaseFormat,
-						texImage->TexFormat,
-						texImage->Data, 0, 0, 0, /* dstX/Y/Zoffset */
-						dstRowStride,
-						dstImageOffsets,
-						width, height, depth,
-						format, type, pixels, packing))
-				_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
-
-			if (dims == 3)
-				_mesa_free(dstImageOffsets);
-		}
+		radeon_store_teximage(ctx, dims,
+			0, 0, 0,
+			width, height, depth,
+			imageSize, format, type,
+			pixels, packing,
+			texObj, texImage,
+			compressed);
 	}
 
 	_mesa_unmap_teximage_pbo(ctx, packing);
-
-	if (pixels)
-	  radeon_teximage_unmap(image);
-
-
 }
 
 void radeonTexImage1D(GLcontext * ctx, GLenum target, GLint level,
@@ -724,7 +836,7 @@ void radeonTexImage3D(GLcontext * ctx, GLenum target, GLint level,
 }
 
 /**
- * Update a subregion of the given texture image.
+ * All glTexSubImage calls go through this function.
  */
 static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int level,
 		GLint xoffset, GLint yoffset, GLint zoffset,
@@ -741,64 +853,39 @@ static void radeon_texsubimage(GLcontext* ctx, int dims, GLenum target, int leve
 	radeonTexObj* t = radeon_tex_obj(texObj);
 	radeon_texture_image* image = get_radeon_texture_image(texImage);
 
-	radeon_firevertices(rmesa);
+	{
+		struct radeon_bo *bo;
+		bo = !image->mt ? image->bo : image->mt->bo;
+		if (bo && radeon_bo_is_referenced_by_cs(bo, rmesa->cmdbuf.cs)) {
+			radeon_firevertices(rmesa);
+		}
+	}
+
+	if (RADEON_DEBUG & RADEON_TEXTURE) {
+		fprintf(stderr, "radeon_texsubimage%dd: texObj %p, texImage %p, face %d, level %d\n",
+				dims, texObj, texImage, radeon_face_for_target(target), level);
+	}
 
 	t->validated = GL_FALSE;
 	if (compressed) {
 		pixels = _mesa_validate_pbo_compressed_teximage(
-			ctx, imageSize, pixels, packing, "glCompressedTexImage");
+			ctx, imageSize, pixels, packing, "glCompressedTexSubImage");
 	} else {
 		pixels = _mesa_validate_pbo_teximage(ctx, dims,
-			width, height, depth, format, type, pixels, packing, "glTexSubImage1D");
+			width, height, depth, format, type, pixels, packing, "glTexSubImage");
 	}
 
 	if (pixels) {
-		GLint dstRowStride;
-		radeon_teximage_map(image, GL_TRUE);
-
-		if (image->mt) {
-			radeon_mipmap_level *lvl = &image->mt->levels[image->mtlevel];
-			dstRowStride = lvl->rowstride;
-		} else {
-			dstRowStride = texImage->RowStride * texImage->TexFormat->TexelBytes;
-		}
-
-		if (compressed) {
-			uint32_t srcRowStride, bytesPerRow, rows;
-			GLubyte *img_start;
-			if (!image->mt) {
-				dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, texImage->Width);
-				img_start = _mesa_compressed_image_address(xoffset, yoffset, 0,
-									   texImage->TexFormat->MesaFormat,
-									   texImage->Width, texImage->Data);
-			}
-			else {
-				uint32_t blocks_x = dstRowStride / (image->mt->bpp * 4);
-				img_start = texImage->Data + image->mt->bpp * 4 * (blocks_x * (yoffset / 4) + xoffset / 4);
-			}
-			srcRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
-			bytesPerRow = srcRowStride;
-			rows = (height + 3) / 4;
-
-			copy_rows(img_start, dstRowStride,  pixels, srcRowStride, rows,  bytesPerRow);
-			
-		} else {
-			if (!texImage->TexFormat->StoreImage(ctx, dims, texImage->_BaseFormat,
-							     texImage->TexFormat, texImage->Data,
-							     xoffset, yoffset, zoffset,
-							     dstRowStride,
-							     texImage->ImageOffsets,
-							     width, height, depth,
-							     format, type, pixels, packing))
-				_mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage");
-		}
+		radeon_store_teximage(ctx, dims,
+			xoffset, yoffset, zoffset,
+			width, height, depth,
+			imageSize, format, type,
+			pixels, packing,
+			texObj, texImage,
+			compressed);
 	}
 
-	radeon_teximage_unmap(image);
-
 	_mesa_unmap_teximage_pbo(ctx, packing);
-
-
 }
 
 void radeonTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
@@ -854,143 +941,6 @@ void radeonTexSubImage3D(GLcontext * ctx, GLenum target, GLint level,
 		format, type, pixels, packing, texObj, texImage, 0);
 }
 
-
-
-/**
- * Ensure that the given image is stored in the given miptree from now on.
- */
-static void migrate_image_to_miptree(radeon_mipmap_tree *mt, radeon_texture_image *image, int face, int level)
-{
-	radeon_mipmap_level *dstlvl = &mt->levels[level - mt->firstLevel];
-	unsigned char *dest;
-
-	assert(image->mt != mt);
-	assert(dstlvl->width == image->base.Width);
-	assert(dstlvl->height == image->base.Height);
-	assert(dstlvl->depth == image->base.Depth);
-
-
-	radeon_bo_map(mt->bo, GL_TRUE);
-	dest = mt->bo->ptr + dstlvl->faces[face].offset;
-
-	if (image->mt) {
-		/* Format etc. should match, so we really just need a memcpy().
-		 * In fact, that memcpy() could be done by the hardware in many
-		 * cases, provided that we have a proper memory manager.
-		 */
-		radeon_mipmap_level *srclvl = &image->mt->levels[image->mtlevel-image->mt->firstLevel];
-
-		assert(srclvl->size == dstlvl->size);
-		assert(srclvl->rowstride == dstlvl->rowstride);
-
-		radeon_bo_map(image->mt->bo, GL_FALSE);
-
-		memcpy(dest,
-			image->mt->bo->ptr + srclvl->faces[face].offset,
-			dstlvl->size);
-		radeon_bo_unmap(image->mt->bo);
-
-		radeon_miptree_unreference(image->mt);
-	} else {
-		uint32_t srcrowstride;
-		uint32_t height;
-		/* need to confirm this value is correct */
-		if (mt->compressed) {
-			height = (image->base.Height + 3) / 4;
-			srcrowstride = _mesa_compressed_row_stride(image->base.TexFormat->MesaFormat, image->base.Width);
-		} else {
-			height = image->base.Height * image->base.Depth;
-			srcrowstride = image->base.Width * image->base.TexFormat->TexelBytes;
-		}
-
-//		if (mt->tilebits)
-//			WARN_ONCE("%s: tiling not supported yet", __FUNCTION__);
-
-		copy_rows(dest, dstlvl->rowstride, image->base.Data, srcrowstride,
-			  height, srcrowstride);
-
-		_mesa_free_texmemory(image->base.Data);
-		image->base.Data = 0;
-	}
-
-	radeon_bo_unmap(mt->bo);
-
-	image->mt = mt;
-	image->mtface = face;
-	image->mtlevel = level;
-	radeon_miptree_reference(image->mt);
-}
-
-int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *texObj)
-{
-	radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-	radeonTexObj *t = radeon_tex_obj(texObj);
-	radeon_texture_image *baseimage = get_radeon_texture_image(texObj->Image[0][texObj->BaseLevel]);
-	int face, level;
-
-	if (t->validated || t->image_override)
-		return GL_TRUE;
-
-	if (RADEON_DEBUG & RADEON_TEXTURE)
-		fprintf(stderr, "%s: Validating texture %p now\n", __FUNCTION__, texObj);
-
-	if (baseimage->base.Border > 0)
-		return GL_FALSE;
-
-	/* Ensure a matching miptree exists.
-	 *
-	 * Differing mipmap trees can result when the app uses TexImage to
-	 * change texture dimensions.
-	 *
-	 * Prefer to use base image's miptree if it
-	 * exists, since that most likely contains more valid data (remember
-	 * that the base level is usually significantly larger than the rest
-	 * of the miptree, so cubemaps are the only possible exception).
-	 */
-	if (baseimage->mt &&
-	    baseimage->mt != t->mt &&
-	    radeon_miptree_matches_texture(baseimage->mt, &t->base)) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = baseimage->mt;
-		radeon_miptree_reference(t->mt);
-	} else if (t->mt && !radeon_miptree_matches_texture(t->mt, &t->base)) {
-		radeon_miptree_unreference(t->mt);
-		t->mt = 0;
-	}
-
-	if (!t->mt) {
-		if (RADEON_DEBUG & RADEON_TEXTURE)
-			fprintf(stderr, " Allocate new miptree\n");
-		radeon_try_alloc_miptree(rmesa, t, baseimage, 0, texObj->BaseLevel);
-		if (!t->mt) {
-			_mesa_problem(ctx, "radeon_validate_texture failed to alloc miptree");
-			return GL_FALSE;
-		}
-	}
-
-	/* Ensure all images are stored in the single main miptree */
-	for(face = 0; face < t->mt->faces; ++face) {
-		for(level = t->mt->firstLevel; level <= t->mt->lastLevel; ++level) {
-			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 || (!image->mt && !image->base.Data)) {
-				if (RADEON_DEBUG & RADEON_TEXTURE)
-					fprintf(stderr, "OK\n");
-
-				continue;
-			}
-
-			if (RADEON_DEBUG & RADEON_TEXTURE)
-				fprintf(stderr, "migrating\n");
-			migrate_image_to_miptree(t->mt, image, face, level);
-		}
-	}
-
-	return GL_TRUE;
-}
-
-
 /**
  * Need to map texture image into memory before copying image data,
  * then unmap it.
diff --git a/src/mesa/drivers/dri/radeon/radeon_texture.h b/src/mesa/drivers/dri/radeon/radeon_texture.h
index 888a55ba91..906daf12d0 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texture.h
+++ b/src/mesa/drivers/dri/radeon/radeon_texture.h
@@ -30,6 +30,11 @@
 
 #ifndef RADEON_TEXTURE_H
 #define RADEON_TEXTURE_H
+
+#include "main/formats.h"
+
+void copy_rows(void* dst, GLuint dststride, const void* src, GLuint srcstride,
+	GLuint numrows, GLuint rowsize);
 struct gl_texture_image *radeonNewTextureImage(GLcontext *ctx);
 void radeonFreeTexImageData(GLcontext *ctx, struct gl_texture_image *timage);
 
@@ -40,14 +45,16 @@ void radeonUnmapTexture(GLcontext *ctx, struct gl_texture_object *texObj);
 void radeonGenerateMipmap(GLcontext* ctx, GLenum target, struct gl_texture_object *texObj);
 int radeon_validate_texture_miptree(GLcontext * ctx, struct gl_texture_object *texObj);
 GLuint radeon_face_for_target(GLenum target);
-const struct gl_texture_format *radeonChooseTextureFormat_mesa(GLcontext * ctx,
-							  GLint internalFormat,
-							  GLenum format,
-							  GLenum type);
-const struct gl_texture_format *radeonChooseTextureFormat(GLcontext * ctx,
-							  GLint internalFormat,
-							  GLenum format,
-							  GLenum type, GLboolean fbo);
+
+gl_format radeonChooseTextureFormat_mesa(GLcontext * ctx,
+                                         GLint internalFormat,
+                                         GLenum format,
+                                         GLenum type);
+
+gl_format radeonChooseTextureFormat(GLcontext * ctx,
+                                    GLint internalFormat,
+                                    GLenum format,
+                                    GLenum type, GLboolean fbo);
 
 void radeonTexImage1D(GLcontext * ctx, GLenum target, GLint level,
 		      GLint internalFormat,
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 ec1182f34f..0000000000
--- a/src/mesa/drivers/dri/s3v/s3v_tex.c
+++ /dev/null
@@ -1,548 +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.TexImage2D = s3vTexImage2D;
-	ctx->Driver.TexSubImage2D = s3vTexSubImage2D;
-	ctx->Driver.BindTexture = s3vBindTexture;
-	ctx->Driver.DeleteTexture = s3vDeleteTexture;
-	ctx->Driver.TexParameter = s3vTexParameter;
-	ctx->Driver.UpdateTexturePalette = 0;
-	ctx->Driver.IsTextureResident = s3vIsTextureResident;
-
-	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/savage/savage_init.h b/src/mesa/drivers/dri/savage/savage_init.h
index abb8440fc4..bfd3077d70 100644
--- a/src/mesa/drivers/dri/savage/savage_init.h
+++ b/src/mesa/drivers/dri/savage/savage_init.h
@@ -66,7 +66,7 @@ typedef struct {
    unsigned int logTextureGranularity[SAVAGE_NR_TEX_HEAPS];
    drmAddress texVirtual[SAVAGE_NR_TEX_HEAPS];
   
-   __DRIscreenPrivate *driScrnPriv;
+   __DRIscreen *driScrnPriv;
 
    savageRegion aperture;
    savageRegion agpTextures;
diff --git a/src/mesa/drivers/dri/savage/savage_xmesa.c b/src/mesa/drivers/dri/savage/savage_xmesa.c
index 931ceff0a8..8e879ca41c 100644
--- a/src/mesa/drivers/dri/savage/savage_xmesa.c
+++ b/src/mesa/drivers/dri/savage/savage_xmesa.c
@@ -59,7 +59,7 @@
 #include "texmem.h"
 
 #define need_GL_EXT_secondary_color
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 #include "xmlpool.h"
 
@@ -168,7 +168,7 @@ PUBLIC const __DRIextension *savageScreenExtensions[] = {
 };
 
 static GLboolean
-savageInitDriver(__DRIscreenPrivate *sPriv)
+savageInitDriver(__DRIscreen *sPriv)
 {
   savageScreenPrivate *savageScreen;
   SAVAGEDRIPtr         gDRIPriv = (SAVAGEDRIPtr)sPriv->pDevPriv;
@@ -272,7 +272,7 @@ savageInitDriver(__DRIscreenPrivate *sPriv)
 /* Accessed by dlsym from dri_mesa_init.c
  */
 static void
-savageDestroyScreen(__DRIscreenPrivate *sPriv)
+savageDestroyScreen(__DRIscreen *sPriv)
 {
    savageScreenPrivate *savageScreen = (savageScreenPrivate *)sPriv->private;
 
@@ -288,12 +288,12 @@ savageDestroyScreen(__DRIscreenPrivate *sPriv)
 
 static GLboolean
 savageCreateContext( const __GLcontextModes *mesaVis,
-		     __DRIcontextPrivate *driContextPriv,
+		     __DRIcontext *driContextPriv,
 		     void *sharedContextPrivate )
 {
    GLcontext *ctx, *shareCtx;
    savageContextPtr imesa;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    struct dd_function_table functions;
    savageScreenPrivate *savageScreen = (savageScreenPrivate *)sPriv->private;
    drm_savage_sarea_t *saPriv=(drm_savage_sarea_t *)(((char*)sPriv->pSAREA)+
@@ -436,7 +436,7 @@ savageCreateContext( const __GLcontextModes *mesaVis,
    if (ctx->Const.MaxTextureLevels <= 6) { /*spec requires at least 64x64*/
        __driUtilMessage("Not enough texture memory. "
 			"Falling back to indirect rendering.");
-       Xfree(imesa);
+       _mesa_free(imesa);
        return GL_FALSE;
    }
 
@@ -546,7 +546,7 @@ savageCreateContext( const __GLcontextModes *mesaVis,
 }
 
 static void
-savageDestroyContext(__DRIcontextPrivate *driContextPriv)
+savageDestroyContext(__DRIcontext *driContextPriv)
 {
    savageContextPtr imesa = (savageContextPtr) driContextPriv->driverPrivate;
    GLuint i;
@@ -574,14 +574,14 @@ savageDestroyContext(__DRIcontextPrivate *driContextPriv)
       _mesa_destroy_context(imesa->glCtx);
 
       /* no longer use vertex_dma_buf*/
-      Xfree(imesa);
+      _mesa_free(imesa);
    }
 }
 
 
 static GLboolean
-savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-		    __DRIdrawablePrivate *driDrawPriv,
+savageCreateBuffer( __DRIscreen *driScrnPriv,
+		    __DRIdrawable *driDrawPriv,
 		    const __GLcontextModes *mesaVis,
 		    GLboolean isPixmap)
 {
@@ -602,7 +602,7 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  (GLubyte *) screen->aperture.map
                                  + 0x01000000 * TARGET_FRONT,
                                  screen->cpp,
@@ -615,7 +615,7 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888,
                                  (GLubyte *) screen->aperture.map
                                  + 0x01000000 * TARGET_BACK,
                                  screen->cpp,
@@ -628,7 +628,7 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
+            = driNewRenderbuffer(MESA_FORMAT_Z16,
                                  (GLubyte *) screen->aperture.map
                                  + 0x01000000 * TARGET_DEPTH,
                                  screen->zpp,
@@ -639,7 +639,7 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 24) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
+            = driNewRenderbuffer(MESA_FORMAT_S8_Z24,
                                  (GLubyte *) screen->aperture.map
                                  + 0x01000000 * TARGET_DEPTH,
                                  screen->zpp,
@@ -651,7 +651,7 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0 && !swStencil) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
+            = driNewRenderbuffer(MESA_FORMAT_S8,
                                  (GLubyte *) screen->aperture.map
                                  + 0x01000000 * TARGET_DEPTH,
                                  screen->zpp,
@@ -675,13 +675,13 @@ savageCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 }
 
 static void
-savageDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+savageDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
 #if 0
-void XMesaSwapBuffers(__DRIdrawablePrivate *driDrawPriv)
+void XMesaSwapBuffers(__DRIdrawable *driDrawPriv)
 {
    /* XXX should do swap according to the buffer, not the context! */
    savageContextPtr imesa = savageCtx; 
@@ -694,7 +694,7 @@ void XMesaSwapBuffers(__DRIdrawablePrivate *driDrawPriv)
 
 void savageXMesaSetClipRects(savageContextPtr imesa)
 {
-   __DRIdrawablePrivate *dPriv = imesa->driDrawable;
+   __DRIdrawable *dPriv = imesa->driDrawable;
 
    if ((dPriv->numBackClipRects == 0)
        || (imesa->glCtx->DrawBuffer->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT)) {
@@ -715,8 +715,8 @@ void savageXMesaSetClipRects(savageContextPtr imesa)
 
 static void savageXMesaWindowMoved( savageContextPtr imesa ) 
 {
-   __DRIdrawablePrivate *const drawable = imesa->driDrawable;
-   __DRIdrawablePrivate *const readable = imesa->driReadable;
+   __DRIdrawable *const drawable = imesa->driDrawable;
+   __DRIdrawable *const readable = imesa->driReadable;
 
    if (0)
       fprintf(stderr, "savageXMesaWindowMoved\n\n");
@@ -731,7 +731,7 @@ static void savageXMesaWindowMoved( savageContextPtr imesa )
 
 
 static GLboolean
-savageUnbindContext(__DRIcontextPrivate *driContextPriv)
+savageUnbindContext(__DRIcontext *driContextPriv)
 {
    savageContextPtr savage = (savageContextPtr) driContextPriv->driverPrivate;
    if (savage)
@@ -742,7 +742,7 @@ savageUnbindContext(__DRIcontextPrivate *driContextPriv)
 
 #if 0
 static GLboolean
-savageOpenFullScreen(__DRIcontextPrivate *driContextPriv)
+savageOpenFullScreen(__DRIcontext *driContextPriv)
 {
     
   
@@ -761,7 +761,7 @@ savageOpenFullScreen(__DRIcontextPrivate *driContextPriv)
 }
 
 static GLboolean
-savageCloseFullScreen(__DRIcontextPrivate *driContextPriv)
+savageCloseFullScreen(__DRIcontext *driContextPriv)
 {
     
     if (driContextPriv) {
@@ -777,9 +777,9 @@ savageCloseFullScreen(__DRIcontextPrivate *driContextPriv)
 #endif
 
 static GLboolean
-savageMakeCurrent(__DRIcontextPrivate *driContextPriv,
-		  __DRIdrawablePrivate *driDrawPriv,
-		  __DRIdrawablePrivate *driReadPriv)
+savageMakeCurrent(__DRIcontext *driContextPriv,
+		  __DRIdrawable *driDrawPriv,
+		  __DRIdrawable *driReadPriv)
 {
    if (driContextPriv) {
       savageContextPtr imesa
@@ -816,9 +816,9 @@ savageMakeCurrent(__DRIcontextPrivate *driContextPriv,
 
 void savageGetLock( savageContextPtr imesa, GLuint flags ) 
 {
-   __DRIdrawablePrivate *const drawable = imesa->driDrawable;
-   __DRIdrawablePrivate *const readable = imesa->driReadable;
-   __DRIscreenPrivate *sPriv = imesa->driScreen;
+   __DRIdrawable *const drawable = imesa->driDrawable;
+   __DRIdrawable *const readable = imesa->driReadable;
+   __DRIscreen *sPriv = imesa->driScreen;
    drm_savage_sarea_t *sarea = imesa->sarea;
    int me = imesa->hHWContext;
    int stamp = drawable->lastStamp; 
@@ -883,7 +883,7 @@ void savageGetLock( savageContextPtr imesa, GLuint flags )
 }
 
 static const  __DRIconfig **
-savageFillInModes( __DRIscreenPrivate *psp,
+savageFillInModes( __DRIscreen *psp,
 		   unsigned pixel_bits, unsigned depth_bits,
 		   unsigned stencil_bits, GLboolean have_back_buffer )
 {
@@ -967,7 +967,7 @@ savageFillInModes( __DRIscreenPrivate *psp,
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-savageInitScreen(__DRIscreenPrivate *psp)
+savageInitScreen(__DRIscreen *psp)
 {
    static const __DRIversion ddx_expected = { 2, 0, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
@@ -980,18 +980,6 @@ savageInitScreen(__DRIscreenPrivate *psp)
 				      &psp->drm_version, & drm_expected ) )
       return NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-
    if (!savageInitDriver(psp))
        return NULL;
 
@@ -1013,3 +1001,10 @@ const struct __DriverAPIRec driDriverAPI = {
    savageMakeCurrent,
    savageUnbindContext
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/savage/savagecontext.h b/src/mesa/drivers/dri/savage/savagecontext.h
index 53a37db1cb..ba1e6e1e1a 100644
--- a/src/mesa/drivers/dri/savage/savagecontext.h
+++ b/src/mesa/drivers/dri/savage/savagecontext.h
@@ -271,10 +271,10 @@ struct savage_context_t {
     drm_hw_lock_t *driHwLock;
     GLuint driFd;
 
-    __DRIdrawablePrivate *driDrawable;
-    __DRIdrawablePrivate *driReadable;
+    __DRIdrawable *driDrawable;
+    __DRIdrawable *driReadable;
 
-    __DRIscreenPrivate *driScreen;
+    __DRIscreen *driScreen;
     savageScreenPrivate *savageScreen; 
     drm_savage_sarea_t *sarea;
 
diff --git a/src/mesa/drivers/dri/savage/savageioctl.c b/src/mesa/drivers/dri/savage/savageioctl.c
index 948ed18419..d0b64e801a 100644
--- a/src/mesa/drivers/dri/savage/savageioctl.c
+++ b/src/mesa/drivers/dri/savage/savageioctl.c
@@ -337,6 +337,8 @@ static void savageDDClear( GLcontext *ctx, GLbitfield mask )
    GLint ch = ctx->DrawBuffer->_Ymax - cy;
 
    /* XXX FIX ME: the cx,cy,cw,ch vars are currently ignored! */
+   (void) ch;
+   (void) cw;
 
    if (SAVAGE_DEBUG & DEBUG_VERBOSE_MSG)
        fprintf (stderr, "%s\n", __FUNCTION__);
@@ -358,15 +360,15 @@ static void savageDDClear( GLcontext *ctx, GLbitfield mask )
    depthMask = 0;
    switch (imesa->savageScreen->cpp) {
    case 2:
-       colorMask = PACK_COLOR_565(ctx->Color.ColorMask[0],
-				  ctx->Color.ColorMask[1],
-				  ctx->Color.ColorMask[2]);
+       colorMask = PACK_COLOR_565(ctx->Color.ColorMask[0][0],
+				  ctx->Color.ColorMask[0][1],
+				  ctx->Color.ColorMask[0][2]);
        break;
    case 4:
-       colorMask = PACK_COLOR_8888(ctx->Color.ColorMask[3],
-				   ctx->Color.ColorMask[2],
-				   ctx->Color.ColorMask[1],
-				   ctx->Color.ColorMask[0]);
+       colorMask = PACK_COLOR_8888(ctx->Color.ColorMask[0][3],
+				   ctx->Color.ColorMask[0][2],
+				   ctx->Color.ColorMask[0][1],
+				   ctx->Color.ColorMask[0][0]);
        break;
    }
 
@@ -431,7 +433,7 @@ static void savageDDClear( GLcontext *ctx, GLbitfield mask )
 /*
  * Copy the back buffer to the front buffer. 
  */
-void savageSwapBuffers( __DRIdrawablePrivate *dPriv )
+void savageSwapBuffers( __DRIdrawable *dPriv )
 {
    savageContextPtr imesa;
 
@@ -535,7 +537,7 @@ void savageFlushVertices( savageContextPtr imesa )
 
 void savageFlushCmdBufLocked( savageContextPtr imesa, GLboolean discard )
 {
-    __DRIdrawablePrivate *dPriv = imesa->driDrawable;
+    __DRIdrawable *dPriv = imesa->driDrawable;
 
     if (!imesa->dmaVtxBuf.total)
 	discard = GL_FALSE;
diff --git a/src/mesa/drivers/dri/savage/savageioctl.h b/src/mesa/drivers/dri/savage/savageioctl.h
index 639605cc51..e7e80816c1 100644
--- a/src/mesa/drivers/dri/savage/savageioctl.h
+++ b/src/mesa/drivers/dri/savage/savageioctl.h
@@ -39,7 +39,7 @@ void savageFlushCmdBuf( savageContextPtr imesa, GLboolean discard );
 
 void savageDDInitIoctlFuncs( GLcontext *ctx );
 
-void savageSwapBuffers( __DRIdrawablePrivate *dPriv );
+void savageSwapBuffers( __DRIdrawable *dPriv );
 
 #define WAIT_IDLE_EMPTY(imesa) do { \
     if (SAVAGE_DEBUG & DEBUG_VERBOSE_MSG) \
diff --git a/src/mesa/drivers/dri/savage/savagerender.c b/src/mesa/drivers/dri/savage/savagerender.c
index 32c74f9467..8221edf387 100644
--- a/src/mesa/drivers/dri/savage/savagerender.c
+++ b/src/mesa/drivers/dri/savage/savagerender.c
@@ -252,13 +252,13 @@ static GLboolean run_texnorm_stage( GLcontext *ctx,
          const GLboolean normalizeS = (texObj->WrapS == GL_REPEAT);
          const GLboolean normalizeT = (reallyEnabled & TEXTURE_2D_BIT) &&
             (texObj->WrapT == GL_REPEAT);
-         const GLfloat *in = (GLfloat *)VB->TexCoordPtr[i]->data;
-         const GLint instride = VB->TexCoordPtr[i]->stride;
+         const GLfloat *in = (GLfloat *)VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->data;
+         const GLint instride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->stride;
          GLfloat (*out)[4] = store->texcoord[i].data;
          GLint j;
 
          if (!ctx->Texture.Unit[i]._ReallyEnabled ||
-             VB->TexCoordPtr[i]->size == 4)
+             VB->AttribPtr[_TNL_ATTRIB_TEX0 + i]->size == 4)
             /* Never try to normalize homogenous tex coords! */
             continue;
 
@@ -297,7 +297,7 @@ static GLboolean run_texnorm_stage( GLcontext *ctx,
          }
 
          if (normalizeS || normalizeT)
-            VB->AttribPtr[VERT_ATTRIB_TEX0+i] = VB->TexCoordPtr[i] = &store->texcoord[i];
+            VB->AttribPtr[_TNL_ATTRIB_TEX0 + i] = &store->texcoord[i];
       }
    }
 
diff --git a/src/mesa/drivers/dri/savage/savagespan.c b/src/mesa/drivers/dri/savage/savagespan.c
index 9615e34013..792e166d9c 100644
--- a/src/mesa/drivers/dri/savage/savagespan.c
+++ b/src/mesa/drivers/dri/savage/savagespan.c
@@ -34,7 +34,7 @@
 
 #define LOCAL_VARS						\
    driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   __DRIdrawablePrivate *const dPriv = drb->dPriv;		\
+   __DRIdrawable *const dPriv = drb->dPriv;		\
    GLuint cpp   = drb->cpp;					\
    GLuint pitch = drb->pitch;					\
    GLuint height = dPriv->h;					\
@@ -44,7 +44,7 @@
 
 #define LOCAL_DEPTH_VARS					\
    driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   __DRIdrawablePrivate *const dPriv = drb->dPriv;		\
+   __DRIdrawable *const dPriv = drb->dPriv;		\
    GLuint zpp   = drb->cpp;					\
    GLuint pitch = drb->pitch;					\
    GLuint height = dPriv->h;					\
@@ -255,15 +255,13 @@ void
 savageSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis,
                        GLboolean float_depth)
 {
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         savageInitPointers_565(&drb->Base);
-      }
-      else {
-         savageInitPointers_8888(&drb->Base);
-      }
+   if (drb->Base.Format == MESA_FORMAT_RGB565) {
+      savageInitPointers_565(&drb->Base);
+   }
+   else if (drb->Base.Format == MESA_FORMAT_ARGB8888) {
+      savageInitPointers_8888(&drb->Base);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (drb->Base.Format == MESA_FORMAT_Z16) {
       if (float_depth) {
          savageInitDepthPointers_z16f(&drb->Base);
       }
@@ -271,7 +269,7 @@ savageSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis,
          savageInitDepthPointers_z16(&drb->Base);
       }
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
+   else if (drb->Base.Format == MESA_FORMAT_S8_Z24) {
       if (float_depth) {
          savageInitDepthPointers_s8_z24f(&drb->Base);
       }
@@ -279,7 +277,7 @@ savageSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis,
          savageInitDepthPointers_s8_z24(&drb->Base);
       }
    }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
+   else if (drb->Base.Format == MESA_FORMAT_S8) {
       savageInitStencilPointers_s8_z24(&drb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/savage/savagetex.c b/src/mesa/drivers/dri/savage/savagetex.c
index fe239e1b05..97598f599e 100644
--- a/src/mesa/drivers/dri/savage/savagetex.c
+++ b/src/mesa/drivers/dri/savage/savagetex.c
@@ -23,29 +23,24 @@
  */
 
 
-#include <stdlib.h>
-#include <stdio.h>
-
-#include <GL/gl.h>
-
+#include "main/context.h"
 #include "main/mm.h"
-#include "savagecontext.h"
-#include "savagetex.h"
-#include "savagetris.h"
-#include "savageioctl.h"
-#include "main/simple_list.h"
-#include "main/enums.h"
-#include "savage_bci.h"
-
 #include "main/macros.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/texobj.h"
 #include "main/convolve.h"
 #include "main/colormac.h"
+#include "main/simple_list.h"
+#include "main/enums.h"
 
 #include "swrast/swrast.h"
 
+#include "savagecontext.h"
+#include "savagetex.h"
+#include "savagetris.h"
+#include "savageioctl.h"
+#include "savage_bci.h"
+
 #include "xmlpool.h"
 
 #define TILE_INDEX_DXT1 0
@@ -512,7 +507,7 @@ savageAllocTexObj( struct gl_texture_object *texObj )
 
       savageSetTexWrapping(t,texObj->WrapS,texObj->WrapT);
       savageSetTexFilter(t,texObj->MinFilter,texObj->MagFilter);
-      savageSetTexBorderColor(t,texObj->BorderColor);
+      savageSetTexBorderColor(t,texObj->BorderColor.f);
    }
 
    return t;
@@ -527,6 +522,11 @@ savageAllocTexObj( struct gl_texture_object *texObj )
  * components to white. This way we get the correct result.
  */
 
+#if 0
+/* Using MESA_FORMAT_RGBA8888 to store alpha-only textures should
+ * work but is space inefficient.
+ */
+
 static GLboolean
 _savage_texstore_a1114444(TEXSTORE_PARAMS);
 
@@ -590,10 +590,11 @@ _savage_texstore_a1114444(TEXSTORE_PARAMS)
 	return GL_FALSE;
     _mesa_adjust_image_for_convolution(ctx, dims, &srcWidth, &srcHeight);
     for (img = 0; img < srcDepth; img++) {
+        GLuint texelBytes = _mesa_get_format_bytes(dstFormat);
         GLubyte *dstRow = (GLubyte *) dstAddr
-           + dstImageOffsets[dstZoffset + img] * dstFormat->TexelBytes
+           + dstImageOffsets[dstZoffset + img] * texelBytes
            + dstYoffset * dstRowStride
-           + dstXoffset * dstFormat->TexelBytes;
+           + dstXoffset * texelBytes;
 	for (row = 0; row < srcHeight; row++) {
             GLushort *dstUI = (GLushort *) dstRow;
 	    for (col = 0; col < srcWidth; col++) {
@@ -629,10 +630,11 @@ _savage_texstore_a1118888(TEXSTORE_PARAMS)
 	return GL_FALSE;
     _mesa_adjust_image_for_convolution(ctx, dims, &srcWidth, &srcHeight);
     for (img = 0; img < srcDepth; img++) {
+        GLuint texelBytes = _mesa_get_format_bytes(dstFormat);
         GLubyte *dstRow = (GLubyte *) dstAddr
-           + dstImageOffsets[dstZoffset + img] * dstFormat->TexelBytes
+           + dstImageOffsets[dstZoffset + img] * texelBytes
            + dstYoffset * dstRowStride
-           + dstXoffset * dstFormat->TexelBytes;
+           + dstXoffset * texelBytes;
 	for (row = 0; row < srcHeight; row++) {
             GLuint *dstUI = (GLuint *) dstRow;
 	    for (col = 0; col < srcWidth; col++) {
@@ -647,10 +649,11 @@ _savage_texstore_a1118888(TEXSTORE_PARAMS)
 
     return GL_TRUE;
 }
+#endif
 
 
 /* Called by the _mesa_store_teximage[123]d() functions. */
-static const struct gl_texture_format *
+static gl_format
 savageChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 			   GLenum format, GLenum type )
 {
@@ -669,15 +672,15 @@ savageChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_INT_10_10_10_2:
       case GL_UNSIGNED_INT_2_10_10_10_REV:
-	 return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555;
+	 return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
       }
 
    case 3:
@@ -686,129 +689,152 @@ savageChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_5_6_5:
       case GL_UNSIGNED_SHORT_5_6_5_REV:
-	 return &_mesa_texformat_rgb565;
+	 return MESA_FORMAT_RGB565;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
       }
 
    case GL_RGBA8:
    case GL_RGBA12:
    case GL_RGBA16:
       return !force16bpt ?
-	  &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+	  MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case GL_RGB10_A2:
       return !force16bpt ?
-	  &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555;
+	  MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555;
 
    case GL_RGBA4:
    case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
 
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+      return !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
 
    case GL_RGB5:
    case GL_RGB4:
    case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_ALPHA:
    case GL_COMPRESSED_ALPHA:
-      return isSavage4 ? &_mesa_texformat_a8 : (
+#if 0
+      return isSavage4 ? MESA_FORMAT_a8 : (
 	 do32bpt ? &_savage_texformat_a1118888 : &_savage_texformat_a1114444);
+#else
+      if (isSavage4)
+         return MESA_FORMAT_A8;
+      else if (do32bpt)
+         return MESA_FORMAT_ARGB8888;
+      else
+         return MESA_FORMAT_ARGB4444;
+#endif
    case GL_ALPHA4:
-      return isSavage4 ? &_mesa_texformat_a8 : &_savage_texformat_a1114444;
+#if 0
+      return isSavage4 ? MESA_FORMAT_a8 : &_savage_texformat_a1114444;
+#else
+      if (isSavage4)
+         return MESA_FORMAT_A8;
+      else
+         return MESA_FORMAT_ARGB4444;
+#endif
    case GL_ALPHA8:
    case GL_ALPHA12:
    case GL_ALPHA16:
-      return isSavage4 ? &_mesa_texformat_a8 : (
+#if 0
+      return isSavage4 ? MESA_FORMAT_a8 : (
 	 !force16bpt ? &_savage_texformat_a1118888 : &_savage_texformat_a1114444);
-
+#else
+      if (isSavage4)
+         return MESA_FORMAT_A8;
+      else if (force16bpt)
+         return MESA_FORMAT_ARGB4444;
+      else
+         return MESA_FORMAT_ARGB8888;
+#endif
    case 1:
    case GL_LUMINANCE:
    case GL_COMPRESSED_LUMINANCE:
       /* no alpha, but use argb1555 in 16bit case to get pure grey values */
-      return isSavage4 ? &_mesa_texformat_l8 : (
-	 do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555);
+      return isSavage4 ? MESA_FORMAT_L8 : (
+	 do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555);
    case GL_LUMINANCE4:
-      return isSavage4 ? &_mesa_texformat_l8 : &_mesa_texformat_argb1555;
+      return isSavage4 ? MESA_FORMAT_L8 : MESA_FORMAT_ARGB1555;
    case GL_LUMINANCE8:
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
-      return isSavage4 ? &_mesa_texformat_l8 : (
-	 !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555);
+      return isSavage4 ? MESA_FORMAT_L8 : (
+	 !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555);
 
    case 2:
    case GL_LUMINANCE_ALPHA:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
       /* Savage4 has a al44 texture format. But it's not supported by Mesa. */
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
    case GL_LUMINANCE4_ALPHA4:
    case GL_LUMINANCE6_ALPHA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
    case GL_LUMINANCE8_ALPHA8:
    case GL_LUMINANCE12_ALPHA4:
    case GL_LUMINANCE12_ALPHA12:
    case GL_LUMINANCE16_ALPHA16:
-      return !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 #if 0
    /* TFT_I8 produces garbage on ProSavageDDR and subsequent texture
     * disable keeps rendering garbage. Disabled for now. */
    case GL_INTENSITY:
    case GL_COMPRESSED_INTENSITY:
-      return isSavage4 ? &_mesa_texformat_i8 : (
-	 do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444);
+      return isSavage4 ? MESA_FORMAT_i8 : (
+	 do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444);
    case GL_INTENSITY4:
-      return isSavage4 ? &_mesa_texformat_i8 : &_mesa_texformat_argb4444;
+      return isSavage4 ? MESA_FORMAT_i8 : MESA_FORMAT_ARGB4444;
    case GL_INTENSITY8:
    case GL_INTENSITY12:
    case GL_INTENSITY16:
-      return isSavage4 ? &_mesa_texformat_i8 : (
-	 !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444);
+      return isSavage4 ? MESA_FORMAT_i8 : (
+	 !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444);
 #else
    case GL_INTENSITY:
    case GL_COMPRESSED_INTENSITY:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
    case GL_INTENSITY4:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
    case GL_INTENSITY8:
    case GL_INTENSITY12:
    case GL_INTENSITY16:
-      return !force16bpt ? &_mesa_texformat_argb8888 :
-	  &_mesa_texformat_argb4444;
+      return !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 #endif
 
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
    case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgb_dxt1;
+      return MESA_FORMAT_RGB_DXT1;
    case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgba_dxt1;
+      return MESA_FORMAT_RGBA_DXT1;
 
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-      return &_mesa_texformat_rgba_dxt3;
+      return MESA_FORMAT_RGBA_DXT3;
 
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
       if (!isSavage4)
 	 /* Not the best choice but Savage3D/MX/IX don't support DXT3 or DXT5. */
-	 return &_mesa_texformat_rgba_dxt1;
+	 return MESA_FORMAT_RGBA_DXT1;
       /* fall through */
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
+      return MESA_FORMAT_RGBA_DXT5;
 
 /*
    case GL_COLOR_INDEX:
@@ -822,7 +848,7 @@ savageChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 */
    default:
       _mesa_problem(ctx, "unexpected texture format in %s", __FUNCTION__);
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 }
 
@@ -837,7 +863,7 @@ static void savageSetTexImages( savageContextPtr imesa,
    assert(t);
    assert(image);
 
-   switch (image->TexFormat->MesaFormat) {
+   switch (image->TexFormat) {
    case MESA_FORMAT_ARGB8888:
       textureFormat = TFT_ARGB8888;
       t->texelBytes = tileIndex = 4;
@@ -2018,7 +2044,7 @@ static void savageTexParameter( GLcontext *ctx, GLenum target,
       break;
   
    case GL_TEXTURE_BORDER_COLOR:
-      savageSetTexBorderColor(t,tObj->BorderColor);
+      savageSetTexBorderColor(t,tObj->BorderColor.f);
       break;
 
    default:
@@ -2083,6 +2109,7 @@ void savageDDInitTextureFuncs( struct dd_function_table *functions )
 
    /* Texel fetching with our custom texture formats works just like
     * the standard argb formats. */
+#if 0
    _savage_texformat_a1114444.FetchTexel1D = _mesa_texformat_argb4444.FetchTexel1D;
    _savage_texformat_a1114444.FetchTexel2D = _mesa_texformat_argb4444.FetchTexel2D;
    _savage_texformat_a1114444.FetchTexel3D = _mesa_texformat_argb4444.FetchTexel3D;
@@ -2096,4 +2123,5 @@ void savageDDInitTextureFuncs( struct dd_function_table *functions )
    _savage_texformat_a1118888.FetchTexel1Df= _mesa_texformat_argb8888.FetchTexel1Df;
    _savage_texformat_a1118888.FetchTexel2Df= _mesa_texformat_argb8888.FetchTexel2Df;
    _savage_texformat_a1118888.FetchTexel3Df= _mesa_texformat_argb8888.FetchTexel3Df;
+#endif
 }
diff --git a/src/mesa/drivers/dri/savage/savagetris.c b/src/mesa/drivers/dri/savage/savagetris.c
index c04763b40e..9a92541ef7 100644
--- a/src/mesa/drivers/dri/savage/savagetris.c
+++ b/src/mesa/drivers/dri/savage/savagetris.c
@@ -435,7 +435,8 @@ do {								\
 
 #define LOCAL_VARS(n)						\
    savageContextPtr imesa = SAVAGE_CONTEXT(ctx);		\
-   GLuint color[n], spec[n];					\
+   GLuint color[n] = { 0 };					\
+   GLuint spec[n] = { 0 };					\
    GLuint coloroffset =						\
       ((imesa->skip & SAVAGE_SKIP_W) ? 3 : 4);			\
    GLboolean specoffset =					\
@@ -879,13 +880,13 @@ static GLboolean savageCheckPTexHack( GLcontext *ctx )
 
    RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->TexCoordPtr[0]->size == 4) {
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4) {
       if (!RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_ATTRIB_TEX1, _TNL_LAST_TEX ))
 	 return GL_TRUE; /* apply ptex hack */
       else
 	 FALLBACK(ctx, SAVAGE_FALLBACK_PROJ_TEXTURE, GL_TRUE);
    }
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->TexCoordPtr[1]->size == 4)
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
       FALLBACK(ctx, SAVAGE_FALLBACK_PROJ_TEXTURE, GL_TRUE);
 
    return GL_FALSE; /* don't apply ptex hack */
@@ -976,13 +977,13 @@ static INLINE GLuint savageChooseVertexFormat_s3d( GLcontext *ctx )
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
       if (imesa->ptexHack)
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_3F_XYW, SAVAGE_EMIT_STQ0, SAVAGE_SKIP_ST0);
-      else if (VB->TexCoordPtr[0]->size == 4)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[0]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size >= 2)
 	 /* The chromium menu emits some 3D tex coords even though no
 	  * 3D texture is enabled. Ignore the 3rd coordinate. */
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_2F, SAVAGE_EMIT_ST0, SAVAGE_SKIP_ST0 );
-      else if (VB->TexCoordPtr[0]->size == 1) {
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 1) {
 	 EMIT_ATTR( _TNL_ATTRIB_TEX0, EMIT_1F, SAVAGE_EMIT_S0, SAVAGE_SKIP_S0 );
 	 EMIT_PAD( 4 );
       } else
@@ -1025,9 +1026,9 @@ static INLINE GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
       if (imesa->ptexHack)
 	 NEED_ATTR( SAVAGE_EMIT_STQ0, SAVAGE_SKIP_ST0);
-      else if (VB->TexCoordPtr[0]->size == 4)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[0]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size >= 2)
 	 /* The chromium menu emits some 3D tex coords even though no
 	  * 3D texture is enabled. Ignore the 3rd coordinate. */
 	 NEED_ATTR( SAVAGE_EMIT_ST0, SAVAGE_SKIP_ST0 );
@@ -1035,10 +1036,10 @@ static INLINE GLuint savageChooseVertexFormat_s4( GLcontext *ctx )
 	 NEED_ATTR( SAVAGE_EMIT_S0, SAVAGE_SKIP_S0 );
    }
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 )) {
-      if (VB->TexCoordPtr[1]->size == 4)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
 	 /* projective textures are not supported by the hardware */
 	 assert (0); /* should be caught by savageCheckPTexHack */
-      else if (VB->TexCoordPtr[1]->size >= 2)
+      else if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size >= 2)
 	 NEED_ATTR( SAVAGE_EMIT_ST1, SAVAGE_SKIP_ST1 );
       else
 	 NEED_ATTR( SAVAGE_EMIT_S1, SAVAGE_SKIP_S1 );
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/sis/sis_clear.c b/src/mesa/drivers/dri/sis/sis_clear.c
index 323383da62..d358ef62dc 100644
--- a/src/mesa/drivers/dri/sis/sis_clear.c
+++ b/src/mesa/drivers/dri/sis/sis_clear.c
@@ -393,7 +393,6 @@ sis_clear_z_stencil_buffer( GLcontext * ctx, GLbitfield mask,
 			    GLint x, GLint y, GLint width, GLint height )
 {
    sisContextPtr smesa = SIS_CONTEXT(ctx);
-   int cmd;
 
    mWait3DCmdQueue (8);
    MMIO(REG_SRC_PITCH, (smesa->zFormat == SiS_ZFORMAT_Z16) ?
diff --git a/src/mesa/drivers/dri/sis/sis_context.c b/src/mesa/drivers/dri/sis/sis_context.c
index a070fe3d79..0944f4d8b4 100644
--- a/src/mesa/drivers/dri/sis/sis_context.c
+++ b/src/mesa/drivers/dri/sis/sis_context.c
@@ -59,7 +59,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 #ifndef SIS_DEBUG
 int SIS_DEBUG = 0;
@@ -69,7 +69,7 @@ int GlobalCurrentHwcx = -1;
 int GlobalHwcxCountBase = 1;
 int GlobalCmdQueueLen = 0;
 
-struct dri_extension card_extensions[] =
+static struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_border_clamp",       NULL },
@@ -83,13 +83,15 @@ struct dri_extension card_extensions[] =
     { NULL,                                NULL }
 };
 
-struct dri_extension card_extensions_6326[] =
+#if 0
+static struct dri_extension card_extensions_6326[] =
 {
     /*{ "GL_ARB_texture_border_clamp",       NULL },*/
     /*{ "GL_ARB_texture_mirrored_repeat",    NULL },*/
     /*{ "GL_MESA_ycbcr_texture",             NULL },*/
     { NULL,                                NULL }
 };
+#endif
 
 static const struct dri_debug_control debug_control[] =
 {
@@ -160,11 +162,11 @@ void sisReAllocateBuffers(GLcontext *ctx, GLframebuffer *drawbuffer,
 
 GLboolean
 sisCreateContext( const __GLcontextModes *glVisual,
-		  __DRIcontextPrivate *driContextPriv,
+		  __DRIcontext *driContextPriv,
                   void *sharedContextPrivate )
 {
    GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    sisContextPtr smesa;
    sisScreenPtr sisScreen;
    int i;
@@ -337,7 +339,7 @@ sisCreateContext( const __GLcontextModes *glVisual,
 }
 
 void
-sisDestroyContext ( __DRIcontextPrivate *driContextPriv )
+sisDestroyContext ( __DRIcontext *driContextPriv )
 {
    sisContextPtr smesa = (sisContextPtr)driContextPriv->driverPrivate;
 
@@ -365,9 +367,9 @@ sisDestroyContext ( __DRIcontextPrivate *driContextPriv )
 }
 
 GLboolean
-sisMakeCurrent( __DRIcontextPrivate *driContextPriv,
-                __DRIdrawablePrivate *driDrawPriv,
-                __DRIdrawablePrivate *driReadPriv )
+sisMakeCurrent( __DRIcontext *driContextPriv,
+                __DRIdrawable *driDrawPriv,
+                __DRIdrawable *driReadPriv )
 {
    if ( driContextPriv ) {
       GET_CURRENT_CONTEXT(ctx);
@@ -396,7 +398,7 @@ sisMakeCurrent( __DRIcontextPrivate *driContextPriv,
 }
 
 GLboolean
-sisUnbindContext( __DRIcontextPrivate *driContextPriv )
+sisUnbindContext( __DRIcontext *driContextPriv )
 {
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/sis/sis_context.h b/src/mesa/drivers/dri/sis/sis_context.h
index bc53cb5efa..4179ee081a 100644
--- a/src/mesa/drivers/dri/sis/sis_context.h
+++ b/src/mesa/drivers/dri/sis/sis_context.h
@@ -359,9 +359,9 @@ struct sis_context
 
   /* Mirrors of some DRI state
    */
-  __DRIcontextPrivate	*driContext;	/* DRI context */
-  __DRIscreenPrivate	*driScreen;	/* DRI screen */
-  __DRIdrawablePrivate	*driDrawable;	/* DRI drawable bound to this ctx */
+  __DRIcontext	*driContext;	/* DRI context */
+  __DRIscreen	*driScreen;	/* DRI screen */
+  __DRIdrawable	*driDrawable;	/* DRI drawable bound to this ctx */
 
   unsigned int lastStamp;	        /* mirror driDrawable->lastStamp */
 
@@ -439,18 +439,18 @@ enum _sis_verbose {
 };
 
 extern GLboolean sisCreateContext( const __GLcontextModes *glVisual,
-				   __DRIcontextPrivate *driContextPriv,
+				   __DRIcontext *driContextPriv,
                                    void *sharedContextPrivate );
-extern void sisDestroyContext( __DRIcontextPrivate * );
+extern void sisDestroyContext( __DRIcontext * );
 
 void sisReAllocateBuffers(GLcontext *ctx, GLframebuffer *drawbuffer,
                           GLuint width, GLuint height);
 
-extern GLboolean sisMakeCurrent( __DRIcontextPrivate *driContextPriv,
-                                  __DRIdrawablePrivate *driDrawPriv,
-                                  __DRIdrawablePrivate *driReadPriv );
+extern GLboolean sisMakeCurrent( __DRIcontext *driContextPriv,
+                                  __DRIdrawable *driDrawPriv,
+                                  __DRIdrawable *driReadPriv );
 
-extern GLboolean sisUnbindContext( __DRIcontextPrivate *driContextPriv );
+extern GLboolean sisUnbindContext( __DRIcontext *driContextPriv );
 
 void WaitEngIdle (sisContextPtr smesa);
 void Wait2DEngIdle (sisContextPtr smesa);
diff --git a/src/mesa/drivers/dri/sis/sis_dd.c b/src/mesa/drivers/dri/sis/sis_dd.c
index bddc4a9285..217d77557f 100644
--- a/src/mesa/drivers/dri/sis/sis_dd.c
+++ b/src/mesa/drivers/dri/sis/sis_dd.c
@@ -41,6 +41,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "sis_tris.h"
 
 #include "swrast/swrast.h"
+#include "main/formats.h"
 #include "main/framebuffer.h"
 #include "main/renderbuffer.h"
 
@@ -142,25 +143,25 @@ sisInitRenderbuffer(struct gl_renderbuffer *rb, GLenum format)
 
    if (format == GL_RGBA) {
       /* Color */
-      rb->_BaseFormat = GL_RGBA;
+      rb->Format = MESA_FORMAT_ARGB8888;
       rb->DataType = GL_UNSIGNED_BYTE;
    }
    else if (format == GL_DEPTH_COMPONENT16) {
       /* Depth */
-      rb->_BaseFormat = GL_DEPTH_COMPONENT;
       /* we always Get/Put 32-bit Z values */
+      rb->Format = MESA_FORMAT_Z16;
       rb->DataType = GL_UNSIGNED_INT;
    }
    else if (format == GL_DEPTH_COMPONENT24) {
       /* Depth */
-      rb->_BaseFormat = GL_DEPTH_COMPONENT;
       /* we always Get/Put 32-bit Z values */
+      rb->Format = MESA_FORMAT_Z32;
       rb->DataType = GL_UNSIGNED_INT;
    }
    else {
       /* Stencil */
       ASSERT(format == GL_STENCIL_INDEX8_EXT);
-      rb->_BaseFormat = GL_STENCIL_INDEX;
+      rb->Format = MESA_FORMAT_S8;
       rb->DataType = GL_UNSIGNED_BYTE;
    }
 
diff --git a/src/mesa/drivers/dri/sis/sis_lock.c b/src/mesa/drivers/dri/sis/sis_lock.c
index 806110cad4..b8ff4e31e2 100644
--- a/src/mesa/drivers/dri/sis/sis_lock.c
+++ b/src/mesa/drivers/dri/sis/sis_lock.c
@@ -46,8 +46,8 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 void
 sisGetLock( sisContextPtr smesa, GLuint flags )
 {
-   __DRIdrawablePrivate *dPriv = smesa->driDrawable;
-   __DRIscreenPrivate *sPriv = smesa->driScreen;
+   __DRIdrawable *dPriv = smesa->driDrawable;
+   __DRIscreen *sPriv = smesa->driScreen;
    SISSAREAPrivPtr sarea = smesa->sarea;
 
    drmGetLock( smesa->driFd, smesa->hHWContext, flags );
diff --git a/src/mesa/drivers/dri/sis/sis_screen.c b/src/mesa/drivers/dri/sis/sis_screen.c
index b5f04ae28d..d38b93ec9b 100644
--- a/src/mesa/drivers/dri/sis/sis_screen.c
+++ b/src/mesa/drivers/dri/sis/sis_screen.c
@@ -65,7 +65,7 @@ static const GLuint __driNConfigOptions = 3;
 extern const struct dri_extension card_extensions[];
 
 static const __DRIconfig **
-sisFillInModes(__DRIscreenPrivate *psp, int bpp)
+sisFillInModes(__DRIscreen *psp, int bpp)
 {
    __DRIconfig **configs;
    unsigned depth_buffer_factor;
@@ -117,7 +117,7 @@ sisFillInModes(__DRIscreenPrivate *psp, int bpp)
 /* Create the device specific screen private data struct.
  */
 static sisScreenPtr
-sisCreateScreen( __DRIscreenPrivate *sPriv )
+sisCreateScreen( __DRIscreen *sPriv )
 {
    sisScreenPtr sisScreen;
    SISDRIPtr sisDRIPriv = (SISDRIPtr)sPriv->pDevPriv;
@@ -172,7 +172,7 @@ sisCreateScreen( __DRIscreenPrivate *sPriv )
 /* Destroy the device specific screen private data struct.
  */
 static void
-sisDestroyScreen( __DRIscreenPrivate *sPriv )
+sisDestroyScreen( __DRIscreen *sPriv )
 {
    sisScreenPtr sisScreen = (sisScreenPtr)sPriv->private;
 
@@ -192,8 +192,8 @@ sisDestroyScreen( __DRIscreenPrivate *sPriv )
  * data.
  */
 static GLboolean
-sisCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
+sisCreateBuffer( __DRIscreen *driScrnPriv,
+                 __DRIdrawable *driDrawPriv,
                  const __GLcontextModes *mesaVis,
                  GLboolean isPixmap )
 {
@@ -219,12 +219,12 @@ sisCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-sisDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+sisDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
-static void sisCopyBuffer( __DRIdrawablePrivate *dPriv )
+static void sisCopyBuffer( __DRIdrawable *dPriv )
 {
    sisContextPtr smesa = (sisContextPtr)dPriv->driContextPriv->driverPrivate;
    int i;
@@ -259,7 +259,7 @@ static void sisCopyBuffer( __DRIdrawablePrivate *dPriv )
 
 /* Copy the back color buffer to the front color buffer */
 static void
-sisSwapBuffers(__DRIdrawablePrivate *dPriv)
+sisSwapBuffers(__DRIdrawable *dPriv)
 {
    if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
          sisContextPtr smesa = (sisContextPtr) dPriv->driContextPriv->driverPrivate;
@@ -284,7 +284,7 @@ sisSwapBuffers(__DRIdrawablePrivate *dPriv)
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-sisInitScreen(__DRIscreenPrivate *psp)
+sisInitScreen(__DRIscreen *psp)
 {
    static const __DRIversion ddx_expected = {0, 8, 0};
    static const __DRIversion dri_expected = {4, 0, 0};
@@ -298,18 +298,6 @@ sisInitScreen(__DRIscreenPrivate *psp)
 				   &psp->drm_version, &drm_expected))
       return NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-
    psp->private = sisCreateScreen(psp);
 
    if (!psp->private) {
@@ -337,3 +325,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .SwapBuffersMSC  = NULL
 
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/sis/sis_screen.h b/src/mesa/drivers/dri/sis/sis_screen.h
index 07c29cfa09..8009fecc31 100644
--- a/src/mesa/drivers/dri/sis/sis_screen.h
+++ b/src/mesa/drivers/dri/sis/sis_screen.h
@@ -50,7 +50,7 @@ typedef struct {
    int cpp;
    unsigned int screenX, screenY;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
    unsigned int sarea_priv_offset;
 
    /* Configuration cache with default values for all contexts */
diff --git a/src/mesa/drivers/dri/sis/sis_span.c b/src/mesa/drivers/dri/sis/sis_span.c
index 9e9a509755..008b00160e 100644
--- a/src/mesa/drivers/dri/sis/sis_span.c
+++ b/src/mesa/drivers/dri/sis/sis_span.c
@@ -42,7 +42,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define LOCAL_VARS							\
    sisContextPtr smesa = SIS_CONTEXT(ctx);				\
-   __DRIdrawablePrivate *dPriv = smesa->driDrawable;			\
+   __DRIdrawable *dPriv = smesa->driDrawable;			\
    struct sis_renderbuffer *srb = (struct sis_renderbuffer *) rb;	\
    GLuint pitch = srb->pitch;						\
    char *buf = srb->map;						\
@@ -52,7 +52,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define LOCAL_DEPTH_VARS						\
    sisContextPtr smesa = SIS_CONTEXT(ctx);				\
-   __DRIdrawablePrivate *dPriv = smesa->driDrawable;			\
+   __DRIdrawable *dPriv = smesa->driDrawable;			\
    struct sis_renderbuffer *srb = (struct sis_renderbuffer *) rb;	\
    char *buf = srb->map;
 
@@ -176,24 +176,22 @@ sisDDInitSpanFuncs( GLcontext *ctx )
 void
 sisSetSpanFunctions(struct sis_renderbuffer *srb, const GLvisual *vis)
 {
-   if (srb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         sisInitPointers_RGB565( &srb->Base );
-      }
-      else {
-         sisInitPointers_ARGB8888( &srb->Base );
-      }
+   if (srb->Base.Format == MESA_FORMAT_RGB565) {
+      sisInitPointers_RGB565( &srb->Base );
    }
-   else if (srb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (srb->Base.Format == MESA_FORMAT_ARGB8888) {
+      sisInitPointers_ARGB8888( &srb->Base );
+   }
+   else if (srb->Base.Format == MESA_FORMAT_Z16) {
       sisInitDepthPointers_z16(&srb->Base);
    }
-   else if (srb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
+   else if (srb->Base.Format == MESA_FORMAT_S8_Z24) {
       sisInitDepthPointers_z24_s8(&srb->Base);
    }
-   else if (srb->Base.InternalFormat == GL_DEPTH_COMPONENT32) {
+   else if (srb->Base.Format == MESA_FORMAT_Z32) {
       sisInitDepthPointers_z32(&srb->Base);
    }
-   else if (srb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
+   else if (srb->Base.Format == MESA_FORMAT_S8) {
       sisInitStencilPointers_z24_s8(&srb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/sis/sis_tex.c b/src/mesa/drivers/dri/sis/sis_tex.c
index 28ced6cfd5..951c470dad 100644
--- a/src/mesa/drivers/dri/sis/sis_tex.c
+++ b/src/mesa/drivers/dri/sis/sis_tex.c
@@ -28,17 +28,15 @@ CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *    Eric Anholt <anholt@FreeBSD.org>
  */
 
-#include "sis_context.h"
-#include "sis_alloc.h"
-#include "sis_tex.h"
-
 #include "swrast/swrast.h"
 #include "main/imports.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 #include "main/texobj.h"
 
+#include "sis_context.h"
+#include "sis_alloc.h"
+#include "sis_tex.h"
 #include "xmlpool.h"
 
 #define ALIGN(value, align) (GLubyte *)((long)(value + align - 1) & ~(align - 1))
@@ -65,7 +63,7 @@ sisAllocTexImage( sisContextPtr smesa, sisTexObjPtr t, int level,
 
    if (t->format == 0) {
       t->format = image->_BaseFormat;
-      switch (image->TexFormat->MesaFormat)
+      switch (image->TexFormat)
       {
       case MESA_FORMAT_ARGB8888:
          t->hwformat = TEXEL_ARGB_8888_32;
@@ -101,13 +99,12 @@ sisAllocTexImage( sisContextPtr smesa, sisTexObjPtr t, int level,
          t->hwformat = TEXEL_VUY422;
          break;
       default:
-         sis_fatal_error("Bad texture format 0x%x.\n",
-			 image->TexFormat->MesaFormat);
+         sis_fatal_error("Bad texture format 0x%x.\n", image->TexFormat);
       }
    }
    assert(t->format == image->_BaseFormat);
 
-   texel_size = image->TexFormat->TexelBytes;
+   texel_size = _mesa_get_format_bytes(image->TexFormat);
    size = image->Width * image->Height * texel_size + TEXTURE_HW_PLUS;
 
    addr = sisAllocFB( smesa, size, &t->image[level].handle );
@@ -230,7 +227,7 @@ static GLboolean sisIsTextureResident( GLcontext * ctx,
   return (texObj->DriverData != NULL);
 }
 
-static const struct gl_texture_format *
+static gl_format
 sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 			  GLenum format, GLenum type )
 {
@@ -248,15 +245,15 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_INT_10_10_10_2:
       case GL_UNSIGNED_INT_2_10_10_10_REV:
-	 return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555;
+	 return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
       }
 
    case 3:
@@ -265,46 +262,46 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
       switch ( type ) {
       case GL_UNSIGNED_SHORT_4_4_4_4:
       case GL_UNSIGNED_SHORT_4_4_4_4_REV:
-	 return &_mesa_texformat_argb4444;
+	 return MESA_FORMAT_ARGB4444;
       case GL_UNSIGNED_SHORT_5_5_5_1:
       case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-	 return &_mesa_texformat_argb1555;
+	 return MESA_FORMAT_ARGB1555;
       case GL_UNSIGNED_SHORT_5_6_5:
       case GL_UNSIGNED_SHORT_5_6_5_REV:
-	 return &_mesa_texformat_rgb565;
+	 return MESA_FORMAT_RGB565;
       default:
-         return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+         return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
       }
 
    case GL_RGBA8:
    case GL_RGBA12:
    case GL_RGBA16:
       return !force16bpt ?
-	  &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+	  MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case GL_RGB10_A2:
       return !force16bpt ?
-	  &_mesa_texformat_argb8888 : &_mesa_texformat_argb1555;
+	  MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB1555;
 
    case GL_RGBA4:
    case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
 
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return !force16bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+      return !force16bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
 
    case GL_RGB5:
    case GL_RGB4:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb332;
+      return MESA_FORMAT_RGB332;
 
    case GL_ALPHA:
    case GL_ALPHA4:		/* FIXME: This could use its own texstore */
@@ -312,7 +309,7 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_ALPHA12:
    case GL_ALPHA16:
    case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
+      return MESA_FORMAT_A8;
 
    case 1:
    case GL_LUMINANCE:
@@ -321,7 +318,7 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
+      return MESA_FORMAT_L8;
 
    case 2:
    case GL_LUMINANCE_ALPHA:
@@ -332,7 +329,7 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12_ALPHA12:
    case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
+      return MESA_FORMAT_AL88;
 
    case GL_INTENSITY:
    case GL_INTENSITY4:
@@ -340,19 +337,19 @@ sisChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY12:
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
+      return MESA_FORMAT_I8;
 
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_APPLE ||
           type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    default:
       _mesa_problem(ctx, "unexpected format in sisDDChooseTextureFormat: %d",
          internalFormat);
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 }
 
@@ -425,7 +422,7 @@ static void sisTexSubImage1D( GLcontext *ctx,
 
    /* Upload the texture */
    WaitEngIdle(smesa);
-   texelBytes = texImage->TexFormat->TexelBytes;
+   texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 
    copySize = width * texelBytes;
    src = (char *)texImage->Data + xoffset * texelBytes;
@@ -513,7 +510,7 @@ static void sisTexSubImage2D( GLcontext *ctx,
 
    /* Upload the texture */
    WaitEngIdle(smesa);
-   texelBytes = texImage->TexFormat->TexelBytes;
+   texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 
    copySize = width * texelBytes;
    src = (char *)texImage->Data + (xoffset + yoffset * texImage->Width) *
diff --git a/src/mesa/drivers/dri/sis/sis_texstate.c b/src/mesa/drivers/dri/sis/sis_texstate.c
index 46417ce414..4c22a10cf7 100644
--- a/src/mesa/drivers/dri/sis/sis_texstate.c
+++ b/src/mesa/drivers/dri/sis/sis_texstate.c
@@ -36,7 +36,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "main/colormac.h"
 #include "main/context.h"
 #include "main/macros.h"
-#include "main/texformat.h"
 
 #include "sis_context.h"
 #include "sis_state.h"
@@ -458,10 +457,10 @@ sis_set_texobj_parm( GLcontext *ctx, struct gl_texture_object *texObj,
 
    {
       GLubyte c[4];
-      CLAMPED_FLOAT_TO_UBYTE(c[0], texObj->BorderColor[0]);
-      CLAMPED_FLOAT_TO_UBYTE(c[1], texObj->BorderColor[1]);
-      CLAMPED_FLOAT_TO_UBYTE(c[2], texObj->BorderColor[2]);
-      CLAMPED_FLOAT_TO_UBYTE(c[3], texObj->BorderColor[3]);
+      CLAMPED_FLOAT_TO_UBYTE(c[0], texObj->BorderColor.f[0]);
+      CLAMPED_FLOAT_TO_UBYTE(c[1], texObj->BorderColor.f[1]);
+      CLAMPED_FLOAT_TO_UBYTE(c[2], texObj->BorderColor.f[2]);
+      CLAMPED_FLOAT_TO_UBYTE(c[3], texObj->BorderColor.f[3]);
 
       current->texture[hw_unit].hwTextureBorderColor = 
          PACK_COLOR_8888(c[3], c[0], c[1], c[2]);
diff --git a/src/mesa/drivers/dri/sis/sis_tris.c b/src/mesa/drivers/dri/sis/sis_tris.c
index 76d12d07b3..4690274c3c 100644
--- a/src/mesa/drivers/dri/sis/sis_tris.c
+++ b/src/mesa/drivers/dri/sis/sis_tris.c
@@ -430,7 +430,8 @@ do {								\
 
 #define LOCAL_VARS(n)						\
    sisContextPtr smesa = SIS_CONTEXT(ctx);			\
-   GLuint color[n], spec[n];					\
+   GLuint color[n] = { 0 };					\
+   GLuint spec[n] = { 0 };					\
    GLuint coloroffset = smesa->coloroffset;			\
    GLuint specoffset = smesa->specoffset;			\
    (void) color; (void) spec; (void) coloroffset; (void) specoffset;
@@ -903,14 +904,14 @@ static void sisRenderStart( GLcontext *ctx )
 
    /* projective textures are not supported by the hardware */
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 )) {
-      if (VB->TexCoordPtr[0]->size > 2)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size > 2)
 	 tex_fallback = GL_TRUE;
       EMIT_ATTR(_TNL_ATTRIB_TEX0, EMIT_2F);
       AGPParseSet |= SiS_PS_HAS_UV0;
    }
    /* Will only hit tex1 on SiS300 */
    if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 )) {
-      if (VB->TexCoordPtr[1]->size > 2)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size > 2)
 	 tex_fallback = GL_TRUE;
       EMIT_ATTR(_TNL_ATTRIB_TEX1, EMIT_2F);
       AGPParseSet |= SiS_PS_HAS_UV1;
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 f4947daa06..8340861aff 100644
--- a/src/mesa/drivers/dri/swrast/swrast.c
+++ b/src/mesa/drivers/dri/swrast/swrast.c
@@ -33,6 +33,7 @@
 
 #include "main/context.h"
 #include "main/extensions.h"
+#include "main/formats.h"
 #include "main/framebuffer.h"
 #include "main/imports.h"
 #include "main/renderbuffer.h"
@@ -49,78 +50,6 @@
 #include "swrast_priv.h"
 
 
-#define need_GL_VERSION_1_3
-#define need_GL_VERSION_1_4
-#define need_GL_VERSION_1_5
-#define need_GL_VERSION_2_0
-#define need_GL_VERSION_2_1
-
-/* sw extensions for imaging */
-#define need_GL_EXT_blend_color
-#define need_GL_EXT_blend_minmax
-#define need_GL_EXT_convolution
-#define need_GL_EXT_histogram
-#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
-#define need_GL_ARB_sync
-#define need_GL_APPLE_vertex_array_object
-#define need_GL_ATI_fragment_shader
-#define need_GL_ATI_separate_stencil
-#define need_GL_EXT_depth_bounds_test
-#define need_GL_EXT_framebuffer_object
-#define need_GL_EXT_framebuffer_blit
-#define need_GL_EXT_gpu_program_parameters
-#define need_GL_EXT_paletted_texture
-#define need_GL_EXT_stencil_two_side
-#define need_GL_MESA_resize_buffers
-#define need_GL_NV_vertex_program
-#define need_GL_NV_fragment_program
-
-#include "extension_helper.h"
-
-const struct dri_extension card_extensions[] =
-{
-    { "GL_VERSION_1_3",			GL_VERSION_1_3_functions },
-    { "GL_VERSION_1_4",			GL_VERSION_1_4_functions },
-    { "GL_VERSION_1_5",			GL_VERSION_1_5_functions },
-    { "GL_VERSION_2_0",			GL_VERSION_2_0_functions },
-    { "GL_VERSION_2_1",			GL_VERSION_2_1_functions },
-
-    { "GL_EXT_blend_color",		GL_EXT_blend_color_functions },
-    { "GL_EXT_blend_minmax",		GL_EXT_blend_minmax_functions },
-    { "GL_EXT_convolution",		GL_EXT_convolution_functions },
-    { "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 },
-    { "GL_ARB_sync",			GL_ARB_sync_functions },
-    { "GL_APPLE_vertex_array_object",	GL_APPLE_vertex_array_object_functions },
-    { "GL_ATI_fragment_shader",		GL_ATI_fragment_shader_functions },
-    { "GL_ATI_separate_stencil",	GL_ATI_separate_stencil_functions },
-    { "GL_EXT_depth_bounds_test",	GL_EXT_depth_bounds_test_functions },
-    { "GL_EXT_framebuffer_object",	GL_EXT_framebuffer_object_functions },
-    { "GL_EXT_framebuffer_blit",	GL_EXT_framebuffer_blit_functions },
-    { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
-    { "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 }
-};
-
-
 /**
  * Screen and config-related functions
  */
@@ -244,7 +173,7 @@ driCreateNewScreen(int scrn, const __DRIextension **extensions,
     *driver_configs = (const __DRIconfig **)
        driConcatConfigs(configs24, configs32);
 
-    driInitExtensions( NULL, card_extensions, GL_FALSE );
+    driInitExtensions( NULL, NULL, GL_FALSE );
 
     return psp;
 }
@@ -378,50 +307,38 @@ swrast_new_renderbuffer(const GLvisual *visual, GLboolean front)
 
     switch (pixel_format) {
     case PF_A8R8G8B8:
+	xrb->Base.Format = MESA_FORMAT_ARGB8888;
 	xrb->Base.InternalFormat = GL_RGBA;
 	xrb->Base._BaseFormat = GL_RGBA;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
-	xrb->Base.RedBits   = 8 * sizeof(GLubyte);
-	xrb->Base.GreenBits = 8 * sizeof(GLubyte);
-	xrb->Base.BlueBits  = 8 * sizeof(GLubyte);
-	xrb->Base.AlphaBits = 8 * sizeof(GLubyte);
 	xrb->bpp = 32;
 	break;
     case PF_X8R8G8B8:
+	xrb->Base.Format = MESA_FORMAT_ARGB8888; /* XXX */
 	xrb->Base.InternalFormat = GL_RGB;
 	xrb->Base._BaseFormat = GL_RGB;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
-	xrb->Base.RedBits   = 8 * sizeof(GLubyte);
-	xrb->Base.GreenBits = 8 * sizeof(GLubyte);
-	xrb->Base.BlueBits  = 8 * sizeof(GLubyte);
-	xrb->Base.AlphaBits = 0;
 	xrb->bpp = 32;
 	break;
     case PF_R5G6B5:
+	xrb->Base.Format = MESA_FORMAT_RGB565;
 	xrb->Base.InternalFormat = GL_RGB;
 	xrb->Base._BaseFormat = GL_RGB;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
-	xrb->Base.RedBits   = 5 * sizeof(GLubyte);
-	xrb->Base.GreenBits = 6 * sizeof(GLubyte);
-	xrb->Base.BlueBits  = 5 * sizeof(GLubyte);
-	xrb->Base.AlphaBits = 0;
 	xrb->bpp = 16;
 	break;
     case PF_R3G3B2:
+	xrb->Base.Format = MESA_FORMAT_RGB332;
 	xrb->Base.InternalFormat = GL_RGB;
 	xrb->Base._BaseFormat = GL_RGB;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
-	xrb->Base.RedBits   = 3 * sizeof(GLubyte);
-	xrb->Base.GreenBits = 3 * sizeof(GLubyte);
-	xrb->Base.BlueBits  = 2 * sizeof(GLubyte);
-	xrb->Base.AlphaBits = 0;
 	xrb->bpp = 8;
 	break;
     case PF_CI8:
+	xrb->Base.Format = MESA_FORMAT_CI8;
 	xrb->Base.InternalFormat = GL_COLOR_INDEX8_EXT;
 	xrb->Base._BaseFormat = GL_COLOR_INDEX;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
-	xrb->Base.IndexBits = 8 * sizeof(GLubyte);
 	xrb->bpp = 8;
 	break;
     default:
diff --git a/src/mesa/drivers/dri/swrast/swrast_span.c b/src/mesa/drivers/dri/swrast/swrast_span.c
index 2d3c25dcbe..f8e503463f 100644
--- a/src/mesa/drivers/dri/swrast/swrast_span.c
+++ b/src/mesa/drivers/dri/swrast/swrast_span.c
@@ -63,56 +63,42 @@ static const GLubyte kernel[16] = {
 
 /* 32-bit BGRA */
 #define STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE) \
-   DST[3] = VALUE[ACOMP]; \
-   DST[2] = VALUE[RCOMP]; \
-   DST[1] = VALUE[GCOMP]; \
-   DST[0] = VALUE[BCOMP]
+   *DST = VALUE[ACOMP] << 24 | VALUE[RCOMP] << 16 | VALUE[GCOMP] << 8 | VALUE[BCOMP]
 #define STORE_PIXEL_RGB_A8R8G8B8(DST, X, Y, VALUE) \
-   DST[3] = 0xff; \
-   DST[2] = VALUE[RCOMP]; \
-   DST[1] = VALUE[GCOMP]; \
-   DST[0] = VALUE[BCOMP]
+   *DST = 0xff << 24 | VALUE[RCOMP] << 16 | VALUE[GCOMP] << 8 | VALUE[BCOMP]
 #define FETCH_PIXEL_A8R8G8B8(DST, SRC) \
-   DST[ACOMP] = SRC[3]; \
-   DST[RCOMP] = SRC[2]; \
-   DST[GCOMP] = SRC[1]; \
-   DST[BCOMP] = SRC[0]
+   DST[ACOMP] = *SRC >> 24;            \
+   DST[RCOMP] = (*SRC >> 16) & 0xff;   \
+   DST[GCOMP] = (*SRC >> 8) & 0xff;    \
+   DST[BCOMP] = *SRC & 0xff
 
 
 /* 32-bit BGRX */
 #define STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE) \
-   DST[3] = 0xff; \
-   DST[2] = VALUE[RCOMP]; \
-   DST[1] = VALUE[GCOMP]; \
-   DST[0] = VALUE[BCOMP]
+   *DST = 0xff << 24 | VALUE[RCOMP] << 16 | VALUE[GCOMP] << 8 | VALUE[BCOMP]
 #define STORE_PIXEL_RGB_X8R8G8B8(DST, X, Y, VALUE) \
-   DST[3] = 0xff; \
-   DST[2] = VALUE[RCOMP]; \
-   DST[1] = VALUE[GCOMP]; \
-   DST[0] = VALUE[BCOMP]
+   *DST = 0xff << 24 | VALUE[RCOMP] << 16 | VALUE[GCOMP] << 8 | VALUE[BCOMP]
 #define FETCH_PIXEL_X8R8G8B8(DST, SRC) \
-   DST[ACOMP] = 0xff; \
-   DST[RCOMP] = SRC[2]; \
-   DST[GCOMP] = SRC[1]; \
-   DST[BCOMP] = SRC[0]
+   DST[ACOMP] = 0xff;                  \
+   DST[RCOMP] = (*SRC >> 16) & 0xff;   \
+   DST[GCOMP] = (*SRC >> 8) & 0xff;    \
+   DST[BCOMP] = *SRC & 0xff
 
 
 /* 16-bit BGR */
 #define STORE_PIXEL_R5G6B5(DST, X, Y, VALUE) \
    do { \
    int d = DITHER_COMP(X, Y) >> 6; \
-   GLushort *p = (GLushort *)DST; \
-   *p = ( ((DITHER_CLAMP((VALUE[RCOMP]) + d) & 0xf8) << 8) | \
-	  ((DITHER_CLAMP((VALUE[GCOMP]) + d) & 0xfc) << 3) | \
-	  ((DITHER_CLAMP((VALUE[BCOMP]) + d) & 0xf8) >> 3) ); \
+   *DST = ( ((DITHER_CLAMP((VALUE[RCOMP]) + d) & 0xf8) << 8) | \
+            ((DITHER_CLAMP((VALUE[GCOMP]) + d) & 0xfc) << 3) | \
+            ((DITHER_CLAMP((VALUE[BCOMP]) + d) & 0xf8) >> 3) ); \
    } while(0)
 #define FETCH_PIXEL_R5G6B5(DST, SRC) \
    do { \
-   GLushort p = *(GLushort *)SRC; \
    DST[ACOMP] = 0xff; \
-   DST[RCOMP] = ((p >> 8) & 0xf8) * 255 / 0xf8; \
-   DST[GCOMP] = ((p >> 3) & 0xfc) * 255 / 0xfc; \
-   DST[BCOMP] = ((p << 3) & 0xf8) * 255 / 0xf8; \
+   DST[RCOMP] = ((*SRC >> 8) & 0xf8) * 255 / 0xf8; \
+   DST[GCOMP] = ((*SRC >> 3) & 0xfc) * 255 / 0xfc; \
+   DST[BCOMP] = ((*SRC << 3) & 0xf8) * 255 / 0xf8; \
    } while(0)
 
 
@@ -145,8 +131,8 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 4;
-#define INC_PIXEL_PTR(P) P += 4
+   GLuint *P = (GLuint *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch / 4 + (X)
+#define INC_PIXEL_PTR(P) P++
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE)
 #define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
@@ -163,8 +149,8 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 4;
-#define INC_PIXEL_PTR(P) P += 4
+   GLuint *P = (GLuint *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch / 4 + (X);
+#define INC_PIXEL_PTR(P) P++
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE)
 #define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
@@ -181,8 +167,8 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 2;
-#define INC_PIXEL_PTR(P) P += 2
+   GLushort *P = (GLushort *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch / 2 + (X);
+#define INC_PIXEL_PTR(P) P++
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_R5G6B5(DST, X, Y, VALUE)
 #define FETCH_PIXEL(DST, SRC) \
@@ -234,8 +220,8 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)row;
-#define INC_PIXEL_PTR(P) P += 4
+   GLuint *P = (GLuint *)row;
+#define INC_PIXEL_PTR(P) P++
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_A8R8G8B8(DST, X, Y, VALUE)
 #define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
@@ -252,8 +238,8 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)row;
-#define INC_PIXEL_PTR(P) P += 4
+   GLuint *P = (GLuint *)row;
+#define INC_PIXEL_PTR(P) P++
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE)
 #define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
@@ -270,7 +256,7 @@ static const GLubyte kernel[16] = {
 #define SPAN_VARS \
    struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
 #define INIT_PIXEL_PTR(P, X, Y) \
-   GLubyte *P = (GLubyte *)row;
+   GLushort *P = (GLushort *)row;
 #define INC_PIXEL_PTR(P) P += 2
 #define STORE_PIXEL(DST, X, Y, VALUE) \
    STORE_PIXEL_R5G6B5(DST, X, Y, VALUE)
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_context.c b/src/mesa/drivers/dri/tdfx/tdfx_context.c
index 68b5027561..edb1875f76 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.c
@@ -68,13 +68,13 @@
 #define need_GL_EXT_paletted_texture
 /* #define need_GL_EXT_secondary_color */
 /* #define need_GL_NV_vertex_program */
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 
 /**
  * Common extension strings exported by all cards
  */
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
@@ -107,7 +107,7 @@ const struct dri_extension card_extensions[] =
 /**
  * Extension strings exported only by Naplam (e.g., Voodoo4 & Voodoo5) cards.
  */
-const struct dri_extension napalm_extensions[] =
+static const struct dri_extension napalm_extensions[] =
 {
     { "GL_ARB_texture_env_combine",        NULL },
     { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
@@ -165,12 +165,12 @@ static const struct dri_debug_control debug_control[] =
 };
 
 GLboolean tdfxCreateContext( const __GLcontextModes *mesaVis,
-			     __DRIcontextPrivate *driContextPriv,
+			     __DRIcontext *driContextPriv,
                              void *sharedContextPrivate )
 {
    tdfxContextPtr fxMesa;
    GLcontext *ctx, *shareCtx;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+   __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    tdfxScreenPrivate *fxScreen = (tdfxScreenPrivate *) sPriv->private;
    TDFXSAREAPriv *saPriv = (TDFXSAREAPriv *) ((char *) sPriv->pSAREA +
 					      sizeof(drm_sarea_t));
@@ -441,7 +441,7 @@ static GLboolean tdfxInitVertexFormats( tdfxContextPtr fxMesa )
  * Initialize the state in an tdfxContextPtr struct.
  */
 static GLboolean
-tdfxInitContext( __DRIdrawablePrivate *driDrawPriv, tdfxContextPtr fxMesa )
+tdfxInitContext( __DRIdrawable *driDrawPriv, tdfxContextPtr fxMesa )
 {
    /* KW: Would be nice to make one of these a member of the other.
     */
@@ -563,7 +563,7 @@ tdfxInitContext( __DRIdrawablePrivate *driDrawPriv, tdfxContextPtr fxMesa )
 
 
 void
-tdfxDestroyContext( __DRIcontextPrivate *driContextPriv )
+tdfxDestroyContext( __DRIcontext *driContextPriv )
 {
    tdfxContextPtr fxMesa = (tdfxContextPtr) driContextPriv->driverPrivate;
 
@@ -607,7 +607,7 @@ tdfxDestroyContext( __DRIcontextPrivate *driContextPriv )
 
 
 GLboolean
-tdfxUnbindContext( __DRIcontextPrivate *driContextPriv )
+tdfxUnbindContext( __DRIcontext *driContextPriv )
 {
    GET_CURRENT_CONTEXT(ctx);
    tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
@@ -626,9 +626,9 @@ tdfxUnbindContext( __DRIcontextPrivate *driContextPriv )
 
 
 GLboolean
-tdfxMakeCurrent( __DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv )
+tdfxMakeCurrent( __DRIcontext *driContextPriv,
+                 __DRIdrawable *driDrawPriv,
+                 __DRIdrawable *driReadPriv )
 {
    if ( TDFX_DEBUG & DEBUG_VERBOSE_DRI ) {
       fprintf( stderr, "%s( %p )\n", __FUNCTION__, (void *)driContextPriv );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.h b/src/mesa/drivers/dri/tdfx/tdfx_context.h
index 3bcb545119..6e25cac301 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.h
@@ -892,18 +892,18 @@ struct tdfx_context {
    char rendererString[100];
 
    /* stuff added for DRI */
-   __DRIscreenPrivate *driScreen;
-   __DRIcontextPrivate *driContext;
+   __DRIscreen *driScreen;
+   __DRIcontext *driContext;
 
    /**
     * DRI drawable bound to this context for drawing.
     */
-   __DRIdrawablePrivate	*driDrawable;
+   __DRIdrawable	*driDrawable;
 
    /**
     * DRI drawable bound to this context for reading.
     */
-   __DRIdrawablePrivate	*driReadable;
+   __DRIdrawable	*driReadable;
 
    drm_context_t hHWContext;
    drm_hw_lock_t *driHwLock;
@@ -938,19 +938,19 @@ struct tdfx_context {
 
 extern GLboolean
 tdfxCreateContext( const __GLcontextModes *mesaVis,
-                   __DRIcontextPrivate *driContextPriv,
+                   __DRIcontext *driContextPriv,
                    void *sharedContextPrivate );
 
 extern void
-tdfxDestroyContext( __DRIcontextPrivate *driContextPriv );
+tdfxDestroyContext( __DRIcontext *driContextPriv );
 
 extern GLboolean
-tdfxUnbindContext( __DRIcontextPrivate *driContextPriv );
+tdfxUnbindContext( __DRIcontext *driContextPriv );
 
 extern GLboolean
-tdfxMakeCurrent( __DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv );
+tdfxMakeCurrent( __DRIcontext *driContextPriv,
+                 __DRIdrawable *driDrawPriv,
+                 __DRIdrawable *driReadPriv );
 
 extern GLboolean
 tdfxInitGlide( tdfxContextPtr tmesa );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_lock.c b/src/mesa/drivers/dri/tdfx/tdfx_lock.c
index 17cdc51ee1..4f84240104 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_lock.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_lock.c
@@ -45,10 +45,10 @@
 
 void tdfxGetLock( tdfxContextPtr fxMesa )
 {
-    __DRIcontextPrivate *cPriv = fxMesa->driContext;
-    __DRIdrawablePrivate *const drawable = cPriv->driDrawablePriv;
-    __DRIdrawablePrivate *const readable = cPriv->driReadablePriv;
-    __DRIscreenPrivate *sPriv = drawable->driScreenPriv;
+    __DRIcontext *cPriv = fxMesa->driContext;
+    __DRIdrawable *const drawable = cPriv->driDrawablePriv;
+    __DRIdrawable *const readable = cPriv->driReadablePriv;
+    __DRIscreen *sPriv = drawable->driScreenPriv;
     TDFXSAREAPriv *saPriv = (TDFXSAREAPriv *) (((char *) sPriv->pSAREA) +
 					fxMesa->fxScreen->sarea_priv_offset);
     unsigned int stamp = drawable->lastStamp;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_pixels.c b/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
index 18729d5ae0..65f0464f8a 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_pixels.c
@@ -495,7 +495,7 @@ tdfx_readpixels_R5G6B5(GLcontext * ctx, GLint x, GLint y,
    {
       tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
       GrLfbInfo_t info;
-      __DRIdrawablePrivate *const readable = fxMesa->driReadable;
+      __DRIdrawable *const readable = fxMesa->driReadable;
       const GLint winX = readable->x;
       const GLint winY = readable->y + readable->h - 1;
       const GLint scrX = winX + x;
@@ -553,7 +553,7 @@ tdfx_readpixels_R8G8B8A8(GLcontext * ctx, GLint x, GLint y,
    {
       tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
       GrLfbInfo_t info;
-      __DRIdrawablePrivate *const readable = fxMesa->driReadable;
+      __DRIdrawable *const readable = fxMesa->driReadable;
       const GLint winX = readable->x;
       const GLint winY = readable->y + readable->h - 1;
       const GLint scrX = winX + x;
@@ -611,10 +611,10 @@ tdfx_drawpixels_R8G8B8A8(GLcontext * ctx, GLint x, GLint y,
        ctx->Fog.Enabled ||
        ctx->Scissor.Enabled ||
        ctx->Stencil._Enabled ||
-       !ctx->Color.ColorMask[0] ||
-       !ctx->Color.ColorMask[1] ||
-       !ctx->Color.ColorMask[2] ||
-       !ctx->Color.ColorMask[3] ||
+       !ctx->Color.ColorMask[0][0] ||
+       !ctx->Color.ColorMask[0][1] ||
+       !ctx->Color.ColorMask[0][2] ||
+       !ctx->Color.ColorMask[0][3] ||
        ctx->Color.ColorLogicOpEnabled ||
        ctx->Texture._EnabledUnits ||
        fxMesa->Fallback)       
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_render.c b/src/mesa/drivers/dri/tdfx/tdfx_render.c
index 2cd8e12d95..979bcd4514 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_render.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_render.c
@@ -76,8 +76,8 @@ static void tdfxClear( GLcontext *ctx, GLbitfield mask )
 
    if (fxMesa->glCtx->Visual.redBits != 8) {
       /* can only do color masking if running in 24/32bpp on Napalm */
-      if (ctx->Color.ColorMask[RCOMP] != ctx->Color.ColorMask[GCOMP] ||
-          ctx->Color.ColorMask[GCOMP] != ctx->Color.ColorMask[BCOMP]) {
+      if (ctx->Color.ColorMask[0][RCOMP] != ctx->Color.ColorMask[0][GCOMP] ||
+          ctx->Color.ColorMask[0][GCOMP] != ctx->Color.ColorMask[0][BCOMP]) {
          softwareMask |= (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT));
          mask &= ~(BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT);
       }
@@ -556,7 +556,7 @@ static void uploadTextureImages( tdfxContextPtr fxMesa )
  */
 void tdfxUploadClipping( tdfxContextPtr fxMesa )
 {
-   __DRIdrawablePrivate *dPriv = fxMesa->driDrawable;
+   __DRIdrawable *dPriv = fxMesa->driDrawable;
 
    assert(dPriv);
 
@@ -721,7 +721,7 @@ void tdfxEmitHwStateLocked( tdfxContextPtr fxMesa )
 	 fxMesa->Glide.grColorMask( fxMesa->Color.ColorMask[RCOMP] ||
                                     fxMesa->Color.ColorMask[GCOMP] ||
                                     fxMesa->Color.ColorMask[BCOMP],
-                                    /*fxMesa->Color.ColorMask[ACOMP]*/GL_FALSE/*[dBorca] no-no*/ );
+                                    /*fxMesa->Color.ColorMask[0][ACOMP]*/GL_FALSE/*[dBorca] no-no*/ );
       }
       fxMesa->dirty &= ~TDFX_UPLOAD_COLOR_MASK;
    }
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_screen.c b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
index 58bd48b294..4422b5dec4 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_screen.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
@@ -69,11 +69,8 @@ static const __DRIextension *tdfxExtensions[] = {
 
 static const GLuint __driNConfigOptions = 1;
 
-extern const struct dri_extension card_extensions[];
-extern const struct dri_extension napalm_extensions[];
-
 static GLboolean
-tdfxCreateScreen( __DRIscreenPrivate *sPriv )
+tdfxCreateScreen( __DRIscreen *sPriv )
 {
    tdfxScreenPrivate *fxScreen;
    TDFXDRIPtr fxDRIPriv = (TDFXDRIPtr) sPriv->pDevPriv;
@@ -124,7 +121,7 @@ tdfxCreateScreen( __DRIscreenPrivate *sPriv )
 
 
 static void
-tdfxDestroyScreen( __DRIscreenPrivate *sPriv )
+tdfxDestroyScreen( __DRIscreen *sPriv )
 {
    tdfxScreenPrivate *fxScreen = (tdfxScreenPrivate *) sPriv->private;
 
@@ -142,7 +139,7 @@ tdfxDestroyScreen( __DRIscreenPrivate *sPriv )
 
 
 static GLboolean
-tdfxInitDriver( __DRIscreenPrivate *sPriv )
+tdfxInitDriver( __DRIscreen *sPriv )
 {
    if ( TDFX_DEBUG & DEBUG_VERBOSE_DRI ) {
       fprintf( stderr, "%s( %p )\n", __FUNCTION__, (void *)sPriv );
@@ -158,8 +155,8 @@ tdfxInitDriver( __DRIscreenPrivate *sPriv )
 
 
 static GLboolean
-tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-                  __DRIdrawablePrivate *driDrawPriv,
+tdfxCreateBuffer( __DRIscreen *driScrnPriv,
+                  __DRIdrawable *driDrawPriv,
                   const __GLcontextModes *mesaVis,
                   GLboolean isPixmap )
 {
@@ -173,7 +170,7 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL, screen->cpp,
                                  screen->fbOffset, screen->width, driDrawPriv);
          tdfxSetSpanFunctions(frontRb, mesaVis);
          _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
@@ -181,7 +178,7 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA, NULL, screen->cpp,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL, screen->cpp,
                                  screen->backOffset, screen->width,
                                  driDrawPriv);
          tdfxSetSpanFunctions(backRb, mesaVis);
@@ -191,7 +188,7 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16, NULL, screen->cpp,
+            = driNewRenderbuffer(MESA_FORMAT_Z16, NULL, screen->cpp,
                                  screen->depthOffset, screen->width,
                                  driDrawPriv);
          tdfxSetSpanFunctions(depthRb, mesaVis);
@@ -199,7 +196,7 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 24) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24, NULL, screen->cpp,
+            = driNewRenderbuffer(MESA_FORMAT_Z24_S8, NULL, screen->cpp,
                                  screen->depthOffset, screen->width,
                                  driDrawPriv);
          tdfxSetSpanFunctions(depthRb, mesaVis);
@@ -208,7 +205,7 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT, NULL, screen->cpp,
+            = driNewRenderbuffer(MESA_FORMAT_S8, NULL, screen->cpp,
                                  screen->depthOffset, screen->width,
                                  driDrawPriv);
          tdfxSetSpanFunctions(stencilRb, mesaVis);
@@ -230,14 +227,14 @@ tdfxCreateBuffer( __DRIscreenPrivate *driScrnPriv,
 
 
 static void
-tdfxDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+tdfxDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
 
 static void
-tdfxSwapBuffers( __DRIdrawablePrivate *driDrawPriv )
+tdfxSwapBuffers( __DRIdrawable *driDrawPriv )
 
 {
    GET_CURRENT_CONTEXT(ctx);
@@ -256,7 +253,7 @@ tdfxSwapBuffers( __DRIdrawablePrivate *driDrawPriv )
     * we have to do a glFinish (per the GLX spec).
     */
    if ( ctx ) {
-      __DRIdrawablePrivate *curDrawPriv;
+      __DRIdrawable *curDrawPriv;
       fxMesa = TDFX_CONTEXT(ctx);
       curDrawPriv = fxMesa->driContext->driDrawablePriv;
 
@@ -344,7 +341,7 @@ tdfxSwapBuffers( __DRIdrawablePrivate *driDrawPriv )
 }
 
 static const __DRIconfig **
-tdfxFillInModes(__DRIscreenPrivate *psp,
+tdfxFillInModes(__DRIscreen *psp,
 		unsigned pixel_bits,
 		unsigned depth_bits,
 		unsigned stencil_bits,
@@ -418,19 +415,6 @@ tdfxInitScreen(__DRIscreen *psp)
 				      &psp->drm_version, & drm_expected ) )
       return NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-   driInitExtensions( NULL, napalm_extensions, GL_FALSE );
-
    if (!tdfxInitDriver(psp))
       return NULL;
       
@@ -456,3 +440,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_screen.h b/src/mesa/drivers/dri/tdfx/tdfx_screen.h
index 5a68898b36..6aa42e8667 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_screen.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_screen.h
@@ -61,7 +61,7 @@ typedef struct {
    int textureOffset;
    int textureSize;
 
-   __DRIscreenPrivate *driScrnPriv;
+   __DRIscreen *driScrnPriv;
    unsigned int sarea_priv_offset;
 
    /* Configuration cache with default values for all contexts */
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_span.c b/src/mesa/drivers/dri/tdfx/tdfx_span.c
index 6b38fa5a01..a17bcd952a 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_span.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_span.c
@@ -47,7 +47,7 @@
 
 #define LOCAL_VARS							\
    driRenderbuffer *drb = (driRenderbuffer *) rb;			\
-   __DRIdrawablePrivate *const dPriv = drb->dPriv;			\
+   __DRIdrawable *const dPriv = drb->dPriv;			\
    GLuint pitch = drb->backBuffer ? info.strideInBytes			\
      : (drb->pitch * drb->cpp);						\
    const GLuint bottom = dPriv->h - 1;					\
@@ -104,7 +104,7 @@
 
 #define HW_READ_CLIPLOOP()						\
       do {								\
-         const __DRIdrawablePrivate *dPriv = fxMesa->driDrawable;	\
+         const __DRIdrawable *dPriv = fxMesa->driDrawable;	\
          drm_clip_rect_t *rect = dPriv->pClipRects;			\
          int _nc = dPriv->numClipRects;					\
          while (_nc--) {						\
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_state.c b/src/mesa/drivers/dri/tdfx/tdfx_state.c
index 591df8a905..cdb61a0ce0 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_state.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_state.c
@@ -40,7 +40,6 @@
 
 #include "main/mtypes.h"
 #include "main/colormac.h"
-#include "main/texformat.h"
 #include "main/texstore.h"
 #include "main/teximage.h"
 
@@ -622,7 +621,7 @@ static int intersect_rect( drm_clip_rect_t *out,
 void tdfxUpdateClipping( GLcontext *ctx )
 {
    tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = fxMesa->driDrawable;
+   __DRIdrawable *dPriv = fxMesa->driDrawable;
 
    if ( TDFX_DEBUG & DEBUG_VERBOSE_API ) {
       fprintf( stderr, "%s()\n", __FUNCTION__ );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
index f6a48b3ae1..e31ae97b02 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
@@ -42,7 +42,6 @@
 #include "main/image.h"
 #include "main/mipmap.h"
 #include "main/texcompress.h"
-#include "main/texformat.h"
 #include "main/teximage.h"
 #include "main/texstore.h"
 #include "main/texobj.h"
@@ -72,13 +71,13 @@ _mesa_halve2x2_teximage2d ( GLcontext *ctx,
    GLubyte *_d = NULL;
    GLenum _t = 0;
 
-   if (texImage->TexFormat->MesaFormat == MESA_FORMAT_RGB565) {
+   if (texImage->TexFormat == MESA_FORMAT_RGB565) {
       _t = GL_UNSIGNED_SHORT_5_6_5_REV;
       bpt = bytesPerPixel;
-   } else if (texImage->TexFormat->MesaFormat == MESA_FORMAT_ARGB4444) {
+   } else if (texImage->TexFormat == MESA_FORMAT_ARGB4444) {
       _t = GL_UNSIGNED_SHORT_4_4_4_4_REV;
       bpt = bytesPerPixel;
-   } else if (texImage->TexFormat->MesaFormat == MESA_FORMAT_ARGB1555) {
+   } else if (texImage->TexFormat == MESA_FORMAT_ARGB1555) {
       _t = GL_UNSIGNED_SHORT_1_5_5_5_REV;
       bpt = bytesPerPixel;
    }
@@ -93,13 +92,13 @@ _mesa_halve2x2_teximage2d ( GLcontext *ctx,
       }
       _s = src = MALLOC(srcRowStride * srcHeight);
       _d = dst = MALLOC(dstWidth * bytesPerPixel * dstHeight);
-      _mesa_texstore_rgba8888(ctx, 2, GL_RGBA,
-                              &_mesa_texformat_rgba8888_rev, src,
-                              0, 0, 0, /* dstX/Y/Zoffset */
-                              srcRowStride, /* dstRowStride */
-                              &dstImageOffsets,
-                              srcWidth, srcHeight, 1,
-                              texImage->_BaseFormat, _t, srcImage, &ctx->DefaultPacking);
+      _mesa_texstore(ctx, 2, GL_RGBA,
+                     MESA_FORMAT_RGBA8888_REV, src,
+                     0, 0, 0, /* dstX/Y/Zoffset */
+                     srcRowStride, /* dstRowStride */
+                     &dstImageOffsets,
+                     srcWidth, srcHeight, 1,
+                     texImage->_BaseFormat, _t, srcImage, &ctx->DefaultPacking);
    }
 
    if (srcHeight == 1) {
@@ -140,13 +139,13 @@ _mesa_halve2x2_teximage2d ( GLcontext *ctx,
    if (bpt) {
       src = _s;
       dst = _d;
-      texImage->TexFormat->StoreImage(ctx, 2, texImage->_BaseFormat,
-                                      texImage->TexFormat, dstImage,
-                                      0, 0, 0, /* dstX/Y/Zoffset */
-                                      dstWidth * bpt,
-                                      &dstImageOffsets,
-                                      dstWidth, dstHeight, 1,
-                                      GL_BGRA, CHAN_TYPE, dst, &ctx->DefaultPacking);
+      _mesa_texstore(ctx, 2, texImage->_BaseFormat,
+                     texImage->TexFormat, dstImage,
+                     0, 0, 0, /* dstX/Y/Zoffset */
+                     dstWidth * bpt,
+                     &dstImageOffsets,
+                     dstWidth, dstHeight, 1,
+                     GL_BGRA, CHAN_TYPE, dst, &ctx->DefaultPacking);
       FREE(dst);
       FREE(src);
    }
@@ -190,7 +189,7 @@ tdfxGenerateMipmap(GLcontext *ctx, GLenum target,
    const tdfxMipMapLevel *mml;
 
    texImage = _mesa_get_tex_image(ctx, texObj, target, level);
-   assert(!texImage->IsCompressed);
+   texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 
    mml = TDFX_TEXIMAGE_DATA(texImage);
 
@@ -760,7 +759,7 @@ fxTexusError(const char *string, FxBool fatal)
 #endif
 
 
-static const struct gl_texture_format *
+static gl_format
 tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
                            GLenum srcFormat, GLenum srcType )
 {
@@ -774,7 +773,7 @@ tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_ALPHA12:
    case GL_ALPHA16:
    case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
+      return MESA_FORMAT_A8;
    case 1:
    case GL_LUMINANCE:
    case GL_LUMINANCE4:
@@ -782,7 +781,7 @@ tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
+      return MESA_FORMAT_L8;
    case 2:
    case GL_LUMINANCE_ALPHA:
    case GL_LUMINANCE4_ALPHA4:
@@ -792,48 +791,47 @@ tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12_ALPHA12:
    case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
+      return MESA_FORMAT_AL88;
    case GL_INTENSITY:
    case GL_INTENSITY4:
    case GL_INTENSITY8:
    case GL_INTENSITY12:
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
+      return MESA_FORMAT_I8;
    case GL_R3_G3_B2:
    case GL_RGB4:
    case GL_RGB5:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
    case GL_COMPRESSED_RGB:
       /* intentional fall-through */
    case 3:
    case GL_RGB:
      if ( srcFormat == GL_RGB && srcType == GL_UNSIGNED_SHORT_5_6_5 ) {
-       return &_mesa_texformat_rgb565;
+       return MESA_FORMAT_RGB565;
      }
      /* intentional fall through */
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return (allow32bpt) ? &_mesa_texformat_argb8888
-                          : &_mesa_texformat_rgb565;
+      return (allow32bpt) ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
    case GL_RGBA2:
    case GL_RGBA4:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
    case GL_COMPRESSED_RGBA:
       /* intentional fall-through */
    case 4:
    case GL_RGBA:
      if ( srcFormat == GL_BGRA ) {
        if ( srcType == GL_UNSIGNED_INT_8_8_8_8_REV ) {
-         return &_mesa_texformat_argb8888;
+         return MESA_FORMAT_ARGB8888;
        }
        else if ( srcType == GL_UNSIGNED_SHORT_4_4_4_4_REV ) {
-         return &_mesa_texformat_argb4444;
+         return MESA_FORMAT_ARGB4444;
        }
        else if ( srcType == GL_UNSIGNED_SHORT_1_5_5_5_REV ) {
-         return &_mesa_texformat_argb1555;
+         return MESA_FORMAT_ARGB1555;
        }
      }
      /* intentional fall through */
@@ -841,10 +839,9 @@ tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      return allow32bpt ? &_mesa_texformat_argb8888
-                        : &_mesa_texformat_argb4444;
+      return allow32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
    case GL_COLOR_INDEX:
    case GL_COLOR_INDEX1_EXT:
    case GL_COLOR_INDEX2_EXT:
@@ -852,29 +849,29 @@ tdfxChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_COLOR_INDEX8_EXT:
    case GL_COLOR_INDEX12_EXT:
    case GL_COLOR_INDEX16_EXT:
-      return &_mesa_texformat_ci8;
+      return MESA_FORMAT_CI8;
    /* GL_EXT_texture_compression_s3tc */
    /* GL_S3_s3tc */
    case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
-      return &_mesa_texformat_rgb_dxt1;
+      return MESA_FORMAT_RGB_DXT1;
    case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgba_dxt1;
+      return MESA_FORMAT_RGBA_DXT1;
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
-      return &_mesa_texformat_rgba_dxt3;
+      return MESA_FORMAT_RGBA_DXT3;
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
+      return MESA_FORMAT_RGBA_DXT5;
    /* GL_3DFX_texture_compression_FXT1 */
    case GL_COMPRESSED_RGB_FXT1_3DFX:
-      return &_mesa_texformat_rgb_fxt1;
+      return MESA_FORMAT_RGB_FXT1;
    case GL_COMPRESSED_RGBA_FXT1_3DFX:
-      return &_mesa_texformat_rgba_fxt1;
+      return MESA_FORMAT_RGBA_FXT1;
    default:
       _mesa_problem(ctx, "unexpected format in tdfxChooseTextureFormat");
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 }
 
@@ -1126,7 +1123,9 @@ fetch_rgb_dxt1(const struct gl_texture_image *texImage,
     i = i * mml->wScale;
     j = j * mml->hScale;
 
+    /* XXX Get fetch func from _mesa_get_texel_fetch_func()
     _mesa_texformat_rgb_dxt1.FetchTexel2D(texImage, i, j, k, rgba);
+    */
 }
 
 
@@ -1139,7 +1138,9 @@ fetch_rgba_dxt1(const struct gl_texture_image *texImage,
     i = i * mml->wScale;
     j = j * mml->hScale;
 
+    /* XXX Get fetch func from _mesa_get_texel_fetch_func()
     _mesa_texformat_rgba_dxt1.FetchTexel2D(texImage, i, j, k, rgba);
+    */
 }
 
 
@@ -1152,7 +1153,9 @@ fetch_rgba_dxt3(const struct gl_texture_image *texImage,
     i = i * mml->wScale;
     j = j * mml->hScale;
 
+    /* XXX Get fetch func from _mesa_get_texel_fetch_func()
     _mesa_texformat_rgba_dxt3.FetchTexel2D(texImage, i, j, k, rgba);
+    */
 }
 
 
@@ -1165,7 +1168,9 @@ fetch_rgba_dxt5(const struct gl_texture_image *texImage,
     i = i * mml->wScale;
     j = j * mml->hScale;
 
+    /* XXX Get fetch func from _mesa_get_texel_fetch_func()
     _mesa_texformat_rgba_dxt5.FetchTexel2D(texImage, i, j, k, rgba);
+    */
 }
 
 
@@ -1226,20 +1231,21 @@ adjust2DRatio (GLcontext *ctx,
    GLvoid *tempImage;
    GLuint dstImageOffsets = 0;
 
-   if (!texImage->IsCompressed) {
+   if (!_mesa_is_format_compressed(texImage->TexFormat)) {
       GLubyte *destAddr;
+
       tempImage = MALLOC(width * height * texelBytes);
       if (!tempImage) {
          return GL_FALSE;
       }
 
-      texImage->TexFormat->StoreImage(ctx, 2, texImage->_BaseFormat,
-                                      texImage->TexFormat, tempImage,
-                                      0, 0, 0, /* dstX/Y/Zoffset */
-                                      width * texelBytes, /* dstRowStride */
-                                      &dstImageOffsets,
-                                      width, height, 1,
-                                      format, type, pixels, packing);
+      _mesa_texstore(ctx, 2, texImage->_BaseFormat,
+                     texImage->TexFormat, tempImage,
+                     0, 0, 0, /* dstX/Y/Zoffset */
+                     width * texelBytes, /* dstRowStride */
+                     &dstImageOffsets,
+                     width, height, 1,
+                     format, type, pixels, packing);
 
       /* now rescale */
       /* compute address of dest subimage within the overal tex image */
@@ -1256,6 +1262,7 @@ adjust2DRatio (GLcontext *ctx,
    } else {
       const GLint rawBytes = 4;
       GLvoid *rawImage = MALLOC(width * height * rawBytes);
+
       if (!rawImage) {
          return GL_FALSE;
       }
@@ -1265,26 +1272,26 @@ adjust2DRatio (GLcontext *ctx,
          return GL_FALSE;
       }
       /* unpack image, apply transfer ops and store in rawImage */
-      _mesa_texstore_rgba8888(ctx, 2, GL_RGBA,
-                              &_mesa_texformat_rgba8888_rev, rawImage,
-                              0, 0, 0, /* dstX/Y/Zoffset */
-                              width * rawBytes, /* dstRowStride */
-                              &dstImageOffsets,
-                              width, height, 1,
-                              format, type, pixels, packing);
+      _mesa_texstore(ctx, 2, GL_RGBA,
+                     MESA_FORMAT_RGBA8888_REV, rawImage,
+                     0, 0, 0, /* dstX/Y/Zoffset */
+                     width * rawBytes, /* dstRowStride */
+                     &dstImageOffsets,
+                     width, height, 1,
+                     format, type, pixels, packing);
       _mesa_rescale_teximage2d(rawBytes,
                                width,
                                newWidth * rawBytes, /* dst stride */
                                width, height, /* src */
                                newWidth, newHeight, /* dst */
                                rawImage /*src*/, tempImage /*dst*/ );
-      texImage->TexFormat->StoreImage(ctx, 2, texImage->_BaseFormat,
-                                      texImage->TexFormat, texImage->Data,
-                                      xoffset * mml->wScale, yoffset * mml->hScale, 0, /* dstX/Y/Zoffset */
-                                      dstRowStride,
-                                      &dstImageOffsets,
-                                      newWidth, newHeight, 1,
-                                      GL_RGBA, CHAN_TYPE, tempImage, &ctx->DefaultPacking);
+      _mesa_texstore(ctx, 2, texImage->_BaseFormat,
+                     texImage->TexFormat, texImage->Data,
+                     xoffset * mml->wScale, yoffset * mml->hScale, 0, /* dstX/Y/Zoffset */
+                     dstRowStride,
+                     &dstImageOffsets,
+                     newWidth, newHeight, 1,
+                     GL_RGBA, CHAN_TYPE, tempImage, &ctx->DefaultPacking);
       FREE(rawImage);
    }
 
@@ -1357,7 +1364,7 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level,
      * be correct, since it would mess with "compressedSize".
      * Ditto for GL_RGBA[4]_S3TC, which is always mapped to DXT3.
      */
-    if (texImage->IsCompressed) {
+    if (_mesa_is_format_compressed(texImage->TexFormat)) {
       switch (internalFormat) {
       case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
       case GL_RGB_S3TC:
@@ -1384,30 +1391,22 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level,
        }
        if (texNapalm) {
           texImage->InternalFormat = internalFormat = texNapalm;
-          texImage->IsCompressed = GL_TRUE;
+          _mesa_is_format_compressed(texImage->TexFormat) = GL_TRUE;
        }
     }
 #endif
 
-    /* choose the texture format */
-    assert(ctx->Driver.ChooseTextureFormat);
-    texImage->TexFormat = (*ctx->Driver.ChooseTextureFormat)(ctx,
-                                     internalFormat, format, type);
-    assert(texImage->TexFormat);
-    mesaFormat = texImage->TexFormat->MesaFormat;
+    mesaFormat = texImage->TexFormat;
     mml->glideFormat = fxGlideFormat(mesaFormat);
     ti->info.format = mml->glideFormat;
     texImage->FetchTexelc = fxFetchFunction(mesaFormat);
-    texelBytes = texImage->TexFormat->TexelBytes;
-
-    if (texImage->IsCompressed) {
-       texImage->CompressedSize = _mesa_compressed_texture_size(ctx,
-                                                        	mml->width,
-                                                        	mml->height,
-                                                        	1,
-                                                        	mesaFormat);
-       dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, mml->width);
-       texImage->Data = _mesa_alloc_texmemory(texImage->CompressedSize);
+    texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
+
+    if (_mesa_is_format_compressed(texImage->TexFormat)) {
+       GLuint compressedSize = _mesa_format_image_size(mesaFormat, mml->width,
+                                                       mml->height, 1);
+       dstRowStride = _mesa_format_row_stride(texImage->TexFormat, mml->width);
+       texImage->Data = _mesa_alloc_texmemory(compressedSize);
     } else {
        dstRowStride = mml->width * texelBytes;
        texImage->Data = _mesa_alloc_texmemory(mml->width * mml->height * texelBytes);
@@ -1437,13 +1436,13 @@ tdfxTexImage2D(GLcontext *ctx, GLenum target, GLint level,
        else {
           /* no rescaling needed */
           /* unpack image, apply transfer ops and store in texImage->Data */
-          texImage->TexFormat->StoreImage(ctx, 2, texImage->_BaseFormat,
-                                          texImage->TexFormat, texImage->Data,
-                                          0, 0, 0, /* dstX/Y/Zoffset */
-                                          dstRowStride,
-                                          texImage->ImageOffsets,
-                                          width, height, 1,
-                                          format, type, pixels, packing);
+          _mesa_texstore(ctx, 2, texImage->_BaseFormat,
+                         texImage->TexFormat, texImage->Data,
+                         0, 0, 0, /* dstX/Y/Zoffset */
+                         dstRowStride,
+                         texImage->ImageOffsets,
+                         width, height, 1,
+                         format, type, pixels, packing);
        }
     }
 
@@ -1482,9 +1481,9 @@ tdfxTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
     assert(texImage->Data);	/* must have an existing texture image! */
     assert(texImage->_BaseFormat);
 
-    texelBytes = texImage->TexFormat->TexelBytes;
-    if (texImage->IsCompressed) {
-       dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, mml->width);
+    texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
+    if (_mesa_is_format_compressed(texImage->TexFormat)) {
+       dstRowStride = _mesa_format_row_stride(texImage->TexFormat, mml->width);
     } else {
        dstRowStride = mml->width * texelBytes;
     }
@@ -1507,13 +1506,13 @@ tdfxTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
     }
     else {
         /* no rescaling needed */
-        texImage->TexFormat->StoreImage(ctx, 2, texImage->_BaseFormat,
-                                        texImage->TexFormat, texImage->Data,
-                                        xoffset, yoffset, 0,
-                                        dstRowStride,
-                                        texImage->ImageOffsets,
-                                        width, height, 1,
-                                        format, type, pixels, packing);
+       _mesa_texstore(ctx, 2, texImage->_BaseFormat,
+                      texImage->TexFormat, texImage->Data,
+                      xoffset, yoffset, 0,
+                      dstRowStride,
+                      texImage->ImageOffsets,
+                      width, height, 1,
+                      format, type, pixels, packing);
     }
 
     ti->reloadImages = GL_TRUE; /* signal the image needs to be reloaded */
@@ -1572,7 +1571,8 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
     tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
     tdfxTexInfo *ti;
     tdfxMipMapLevel *mml;
-    GLuint mesaFormat;
+    gl_format mesaFormat;
+    GLuint compressedSize = 0;
 
     if (TDFX_DEBUG & DEBUG_VERBOSE_DRI) {
         fprintf(stderr, "tdfxCompressedTexImage2D: id=%d int 0x%x  %dx%d\n",
@@ -1585,8 +1585,6 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
        return;
     }
 
-    assert(texImage->IsCompressed);
-
     ti = TDFX_TEXTURE_DATA(texObj);
     if (!ti) {
         texObj->DriverData = fxAllocTexObjData(fxMesa);
@@ -1615,28 +1613,19 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
     mml->height = height * mml->hScale;
 
 
-    /* choose the texture format */
-    assert(ctx->Driver.ChooseTextureFormat);
-    texImage->TexFormat = (*ctx->Driver.ChooseTextureFormat)(ctx,
-                                           internalFormat, -1/*format*/, -1/*type*/);
-    assert(texImage->TexFormat);
-
     /* Determine the appropriate Glide texel format,
      * given the user's internal texture format hint.
      */
-    mesaFormat = texImage->TexFormat->MesaFormat;
+    mesaFormat = texImage->TexFormat;
     mml->glideFormat = fxGlideFormat(mesaFormat);
     ti->info.format = mml->glideFormat;
     texImage->FetchTexelc = fxFetchFunction(mesaFormat);
 
     /* allocate new storage for texture image, if needed */
     if (!texImage->Data) {
-       texImage->CompressedSize = _mesa_compressed_texture_size(ctx,
-                                                                mml->width,
-                                                                mml->height,
-                                                                1,
-                                                                mesaFormat);
-       texImage->Data = _mesa_alloc_texmemory(texImage->CompressedSize);
+       compressedSize = _mesa_format_image_size(mesaFormat, mml->width,
+                                                mml->height, 1);
+       texImage->Data = _mesa_alloc_texmemory(compressedSize);
        if (!texImage->Data) {
           _mesa_error(ctx, GL_OUT_OF_MEMORY, "glCompressedTexImage2D");
           return;
@@ -1659,10 +1648,10 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
         *    we replicate the data over the padded area.
         * For now, we take 2) + 3) but texelfetchers will be wrong!
         */
-       const GLuint mesaFormat = texImage->TexFormat->MesaFormat;
-       GLuint srcRowStride = _mesa_compressed_row_stride(mesaFormat, width);
+       const GLuint mesaFormat = texImage->TexFormat;
+       GLuint srcRowStride = _mesa_format_row_stride(mesaFormat, width);
  
-       GLuint destRowStride = _mesa_compressed_row_stride(mesaFormat,
+       GLuint destRowStride = _mesa_format_row_stride(mesaFormat,
                                                    mml->width);
  
        _mesa_upscale_teximage2d(srcRowStride, (height+3) / 4,
@@ -1671,7 +1660,7 @@ tdfxCompressedTexImage2D (GLcontext *ctx, GLenum target,
                                 texImage->Data);
        ti->padded = GL_TRUE;
     } else {
-       MEMCPY(texImage->Data, data, texImage->CompressedSize);
+       MEMCPY(texImage->Data, data, compressedSize);
     }
 
     RevalidateTexture(ctx, texObj);
@@ -1696,7 +1685,7 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target,
     GLint destRowStride, srcRowStride;
     GLint i, rows;
     GLubyte *dest;
-    const GLuint mesaFormat = texImage->TexFormat->MesaFormat;
+    const GLuint mesaFormat = texImage->TexFormat;
 
     if (TDFX_DEBUG & DEBUG_VERBOSE_DRI) {
         fprintf(stderr, "tdfxCompressedTexSubImage2D: id=%d\n", texObj->Name);
@@ -1707,9 +1696,9 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target,
     mml = TDFX_TEXIMAGE_DATA(texImage);
     assert(mml);
 
-    srcRowStride = _mesa_compressed_row_stride(mesaFormat, width);
+    srcRowStride = _mesa_format_row_stride(mesaFormat, width);
 
-    destRowStride = _mesa_compressed_row_stride(mesaFormat, mml->width);
+    destRowStride = _mesa_format_row_stride(mesaFormat, mml->width);
     dest = _mesa_compressed_image_address(xoffset, yoffset, 0,
                                           mesaFormat,
                                           mml->width,
@@ -1727,9 +1716,9 @@ tdfxCompressedTexSubImage2D( GLcontext *ctx, GLenum target,
      * see fxDDCompressedTexImage2D for caveats
      */
     if (mml->wScale != 1 || mml->hScale != 1) {
-       srcRowStride = _mesa_compressed_row_stride(mesaFormat, texImage->Width);
+       srcRowStride = _mesa_format_row_stride(mesaFormat, texImage->Width);
  
-       destRowStride = _mesa_compressed_row_stride(mesaFormat, mml->width);
+       destRowStride = _mesa_format_row_stride(mesaFormat, mml->width);
        _mesa_upscale_teximage2d(srcRowStride, texImage->Height / 4,
                                 destRowStride, mml->height / 4,
                                 1, texImage->Data, destRowStride,
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_texstate.c b/src/mesa/drivers/dri/tdfx/tdfx_texstate.c
index bbd2c8cfee..3f737878ed 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_texstate.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_texstate.c
@@ -1314,7 +1314,7 @@ SetupDoubleTexEnvVoodoo3(GLcontext *ctx, int tmu0,
       fxMesa->TexCombine[0].InvertRGB = FXFALSE;
       fxMesa->TexCombine[0].InvertAlpha = FXFALSE;
 
-      if ((baseFormat0 == GL_RGB) && (baseFormat0 == GL_LUMINANCE)) {
+      if ((baseFormat0 == GL_RGB) || (baseFormat0 == GL_LUMINANCE)) {
          fxMesa->AlphaCombine.Function = GR_COMBINE_FUNCTION_LOCAL;
          fxMesa->AlphaCombine.Factor = GR_COMBINE_FACTOR_NONE;
          fxMesa->AlphaCombine.Local = locala;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vb.c b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
index 4928802232..c200ba3255 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vb.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vb.c
@@ -69,11 +69,11 @@ static void interp_extras( GLcontext *ctx,
 
    /*fprintf(stderr, "%s\n", __FUNCTION__);*/
 
-   if (VB->ColorPtr[1]) {
+   if (VB->BackfaceColorPtr) {
       INTERP_4F( t,
-		    GET_COLOR(VB->ColorPtr[1], dst),
-		    GET_COLOR(VB->ColorPtr[1], out),
-		    GET_COLOR(VB->ColorPtr[1], in) );
+		 GET_COLOR(VB->BackfaceColorPtr, dst),
+		 GET_COLOR(VB->BackfaceColorPtr, out),
+		 GET_COLOR(VB->BackfaceColorPtr, in) );
    }
 
    if (VB->EdgeFlag) {
@@ -88,9 +88,9 @@ static void copy_pv_extras( GLcontext *ctx, GLuint dst, GLuint src )
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
-   if (VB->ColorPtr[1]) {
-	 COPY_4FV( GET_COLOR(VB->ColorPtr[1], dst), 
-		     GET_COLOR(VB->ColorPtr[1], src) );
+   if (VB->BackfaceColorPtr) {
+      COPY_4FV( GET_COLOR(VB->BackfaceColorPtr, dst),
+		GET_COLOR(VB->BackfaceColorPtr, src) );
    }
 
    setup_tab[TDFX_CONTEXT(ctx)->SetupIndex].copy_pv(ctx, dst, src);
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h b/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
index 9b780761f4..19baf7d0d2 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
+++ b/src/mesa/drivers/dri/tdfx/tdfx_vbtmp.h
@@ -58,32 +58,32 @@ static void TAG(emit)( GLcontext *ctx,
 /*     fprintf(stderr, "%s\n", __FUNCTION__); */
 
    if (IND & TDFX_TEX0_BIT) {
-      tc0_stride = VB->TexCoordPtr[tmu0_source]->stride;
-      tc0 = VB->TexCoordPtr[tmu0_source]->data;
+      tc0_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->stride;
+      tc0 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->data;
       u0scale = fxMesa->sScale0;
       v0scale = fxMesa->tScale0;
       if (IND & TDFX_PTEX_BIT)
-	 tc0_size = VB->TexCoordPtr[tmu0_source]->size;
+	 tc0_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu0_source]->size;
    }
 
    if (IND & TDFX_TEX1_BIT) {
-      tc1 = VB->TexCoordPtr[tmu1_source]->data;
-      tc1_stride = VB->TexCoordPtr[tmu1_source]->stride;
+      tc1 = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->data;
+      tc1_stride = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->stride;
       u1scale = fxMesa->sScale1;
       v1scale = fxMesa->tScale1;
       if (IND & TDFX_PTEX_BIT)
-	 tc1_size = VB->TexCoordPtr[tmu1_source]->size;
+	 tc1_size = VB->AttribPtr[_TNL_ATTRIB_TEX0 + tmu1_source]->size;
    }
    
    if (IND & TDFX_RGBA_BIT) {
-      col = VB->ColorPtr[0]->data;
-      col_stride = VB->ColorPtr[0]->stride;
-      col_size = VB->ColorPtr[0]->size;
+      col = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->data;
+      col_stride = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->stride;
+      col_size = VB->AttribPtr[_TNL_ATTRIB_COLOR0]->size;
    }
    
    if (IND & TDFX_FOGC_BIT) {
-      fog = VB->FogCoordPtr->data;
-      fog_stride = VB->FogCoordPtr->stride;
+      fog = VB->AttribPtr[_TNL_ATTRIB_FOG]->data;
+      fog_stride = VB->AttribPtr[_TNL_ATTRIB_FOG]->stride;
    }
 
    {
@@ -168,14 +168,14 @@ static GLboolean TAG(check_tex_sizes)( GLcontext *ctx )
       struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
 
       if (IND & TDFX_TEX1_BIT) {
-	 if (VB->TexCoordPtr[0] == 0)
-	    VB->TexCoordPtr[0] = VB->TexCoordPtr[1];
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX0] == 0)
+	    VB->AttribPtr[_TNL_ATTRIB_TEX0] = VB->AttribPtr[_TNL_ATTRIB_TEX1];
 	 
-	 if (VB->TexCoordPtr[1]->size == 4)
+	 if (VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
 	    return GL_FALSE;
       }
 
-      if (VB->TexCoordPtr[0]->size == 4)
+      if (VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4)
 	 return GL_FALSE;
    }
 
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_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index 6eb19ac079..d17a160271 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -32,6 +32,7 @@
 
 #include "main/glheader.h"
 #include "main/context.h"
+#include "main/formats.h"
 #include "main/matrix.h"
 #include "main/state.h"
 #include "main/simple_list.h"
@@ -65,7 +66,7 @@
 #define need_GL_ARB_point_parameters
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
-#include "extension_helper.h"
+#include "main/remap_helper.h"
 
 #define DRIVER_DATE	"20060710"
 
@@ -147,7 +148,7 @@ viaRenderbufferStorage(GLcontext *ctx, struct gl_renderbuffer *rb,
 
 static void
 viaInitRenderbuffer(struct via_renderbuffer *vrb, GLenum format,
-		    __DRIdrawablePrivate *dPriv)
+		    __DRIdrawable *dPriv)
 {
    const GLuint name = 0;
    struct gl_renderbuffer *rb = & vrb->Base;
@@ -163,24 +164,28 @@ viaInitRenderbuffer(struct via_renderbuffer *vrb, GLenum format,
    if (format == GL_RGBA) {
       /* Color */
       rb->_BaseFormat = GL_RGBA;
+      rb->Format = MESA_FORMAT_ARGB8888;
       rb->DataType = GL_UNSIGNED_BYTE;
    }
    else if (format == GL_DEPTH_COMPONENT16) {
       /* Depth */
       rb->_BaseFormat = GL_DEPTH_COMPONENT;
       /* we always Get/Put 32-bit Z values */
+      rb->Format = MESA_FORMAT_Z16;
       rb->DataType = GL_UNSIGNED_INT;
    }
    else if (format == GL_DEPTH_COMPONENT24) {
       /* Depth */
       rb->_BaseFormat = GL_DEPTH_COMPONENT;
       /* we always Get/Put 32-bit Z values */
+      rb->Format = MESA_FORMAT_Z32;
       rb->DataType = GL_UNSIGNED_INT;
    }
    else {
       /* Stencil */
       ASSERT(format == GL_STENCIL_INDEX8_EXT);
       rb->_BaseFormat = GL_STENCIL_INDEX;
+      rb->Format = MESA_FORMAT_S8;
       rb->DataType = GL_UNSIGNED_BYTE;
    }
 
@@ -202,7 +207,7 @@ viaInitRenderbuffer(struct via_renderbuffer *vrb, GLenum format,
 static GLboolean
 calculate_buffer_parameters(struct via_context *vmesa,
 			    struct gl_framebuffer *fb,
-			    __DRIdrawablePrivate *dPriv)
+			    __DRIdrawable *dPriv)
 {
    const unsigned shift = vmesa->viaScreen->bitsPerPixel / 16;
    const unsigned extra = 32;
@@ -362,7 +367,7 @@ void viaReAllocateBuffers(GLcontext *ctx, GLframebuffer *drawbuffer,
 
 /* Extension strings exported by the Unichrome driver.
  */
-const struct dri_extension card_extensions[] =
+static const struct dri_extension card_extensions[] =
 {
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
@@ -455,12 +460,12 @@ FreeBuffer(struct via_context *vmesa)
 
 GLboolean
 viaCreateContext(const __GLcontextModes *visual,
-                 __DRIcontextPrivate *driContextPriv,
+                 __DRIcontext *driContextPriv,
                  void *sharedContextPrivate)
 {
     GLcontext *ctx, *shareCtx;
     struct via_context *vmesa;
-    __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
+    __DRIscreen *sPriv = driContextPriv->driScreenPriv;
     viaScreenPrivate *viaScreen = (viaScreenPrivate *)sPriv->private;
     drm_via_sarea_t *saPriv = (drm_via_sarea_t *)
         (((GLubyte *)sPriv->pSAREA) + viaScreen->sareaPrivOffset);
@@ -674,7 +679,7 @@ viaCreateContext(const __GLcontextModes *visual,
 }
 
 void
-viaDestroyContext(__DRIcontextPrivate *driContextPriv)
+viaDestroyContext(__DRIcontext *driContextPriv)
 {
     GET_CURRENT_CONTEXT(ctx);
     struct via_context *vmesa =
@@ -724,8 +729,8 @@ viaDestroyContext(__DRIcontextPrivate *driContextPriv)
 
 void viaXMesaWindowMoved(struct via_context *vmesa)
 {
-   __DRIdrawablePrivate *const drawable = vmesa->driDrawable;
-   __DRIdrawablePrivate *const readable = vmesa->driReadable;
+   __DRIdrawable *const drawable = vmesa->driDrawable;
+   __DRIdrawable *const readable = vmesa->driReadable;
    struct via_renderbuffer * draw_buffer;
    struct via_renderbuffer * read_buffer;
    GLuint bytePerPixel = vmesa->viaScreen->bitsPerPixel >> 3;
@@ -808,15 +813,15 @@ void viaXMesaWindowMoved(struct via_context *vmesa)
 }
 
 GLboolean
-viaUnbindContext(__DRIcontextPrivate *driContextPriv)
+viaUnbindContext(__DRIcontext *driContextPriv)
 {
     return GL_TRUE;
 }
 
 GLboolean
-viaMakeCurrent(__DRIcontextPrivate *driContextPriv,
-               __DRIdrawablePrivate *driDrawPriv,
-               __DRIdrawablePrivate *driReadPriv)
+viaMakeCurrent(__DRIcontext *driContextPriv,
+               __DRIdrawable *driDrawPriv,
+               __DRIdrawable *driReadPriv)
 {
     if (VIA_DEBUG & DEBUG_DRI) {
 	fprintf(stderr, "driContextPriv = %016lx\n", (unsigned long)driContextPriv);
@@ -892,8 +897,8 @@ viaMakeCurrent(__DRIcontextPrivate *driContextPriv,
 
 void viaGetLock(struct via_context *vmesa, GLuint flags)
 {
-    __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-    __DRIscreenPrivate *sPriv = vmesa->driScreen;
+    __DRIdrawable *dPriv = vmesa->driDrawable;
+    __DRIscreen *sPriv = vmesa->driScreen;
 
     drmGetLock(vmesa->driFd, vmesa->hHWContext, flags);
 
@@ -923,9 +928,9 @@ void viaGetLock(struct via_context *vmesa, GLuint flags)
 
 
 void
-viaSwapBuffers(__DRIdrawablePrivate *drawablePrivate)
+viaSwapBuffers(__DRIdrawable *drawablePrivate)
 {
-    __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *)drawablePrivate;
+    __DRIdrawable *dPriv = (__DRIdrawable *)drawablePrivate;
 
     if (dPriv && 
 	dPriv->driContextPriv && 
diff --git a/src/mesa/drivers/dri/unichrome/via_context.h b/src/mesa/drivers/dri/unichrome/via_context.h
index 4cc9e475c2..4e1ab3a6ca 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.h
+++ b/src/mesa/drivers/dri/unichrome/via_context.h
@@ -105,7 +105,7 @@ struct via_renderbuffer {
    int drawW;                  
    int drawH;    
 
-   __DRIdrawablePrivate *dPriv;
+   __DRIdrawable *dPriv;
 };
 
 
@@ -294,14 +294,14 @@ struct via_context {
    /**
     * DRI drawable bound to this context for drawing.
     */
-   __DRIdrawablePrivate	*driDrawable;
+   __DRIdrawable	*driDrawable;
 
    /**
     * DRI drawable bound to this context for reading.
     */
-   __DRIdrawablePrivate	*driReadable;
+   __DRIdrawable	*driReadable;
 
-   __DRIscreenPrivate *driScreen;
+   __DRIscreen *driScreen;
    viaScreenPrivate *viaScreen;
    drm_via_sarea_t *sarea;
    volatile GLuint* regMMIOBase;
diff --git a/src/mesa/drivers/dri/unichrome/via_ioctl.c b/src/mesa/drivers/dri/unichrome/via_ioctl.c
index 6746f552ae..8d4edfa305 100644
--- a/src/mesa/drivers/dri/unichrome/via_ioctl.c
+++ b/src/mesa/drivers/dri/unichrome/via_ioctl.c
@@ -205,7 +205,7 @@ static void viaFillBuffer(struct via_context *vmesa,
 static void viaClear(GLcontext *ctx, GLbitfield mask)
 {
    struct via_context *vmesa = VIA_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
+   __DRIdrawable *dPriv = vmesa->driDrawable;
    struct via_renderbuffer *const vrb = 
      (struct via_renderbuffer *) dPriv->driverPrivate;
    int flag = 0;
@@ -507,12 +507,12 @@ void viaWaitIdleLocked( struct via_context *vmesa, GLboolean light )
  * except that WAIT_IDLE() will spin the CPU polling, while this is
  * IRQ driven.
  */
-static void viaWaitIdleVBlank(  __DRIdrawablePrivate *dPriv, 
+static void viaWaitIdleVBlank(  __DRIdrawable *dPriv, 
 			       struct via_context *vmesa,
 			       GLuint value )
 {
    GLboolean missed_target;
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+   __DRIscreen *psp = dPriv->driScreenPriv;
 
    VIA_FLUSH_DMA(vmesa); 
 
@@ -591,11 +591,11 @@ void viaResetPageFlippingLocked(struct via_context *vmesa)
 /*
  * Copy the back buffer to the front buffer. 
  */
-void viaCopyBuffer(__DRIdrawablePrivate *dPriv)
+void viaCopyBuffer(__DRIdrawable *dPriv)
 {
    struct via_context *vmesa = 
       (struct via_context *)dPriv->driContextPriv->driverPrivate;
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+   __DRIscreen *psp = dPriv->driScreenPriv;
 
    if (VIA_DEBUG & DEBUG_IOCTL)
       fprintf(stderr, 
@@ -635,12 +635,12 @@ void viaCopyBuffer(__DRIdrawablePrivate *dPriv)
 }
 
 
-void viaPageFlip(__DRIdrawablePrivate *dPriv)
+void viaPageFlip(__DRIdrawable *dPriv)
 {
     struct via_context *vmesa = 
        (struct via_context *)dPriv->driContextPriv->driverPrivate;
     struct via_renderbuffer buffer_tmp;
-    __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+    __DRIscreen *psp = dPriv->driScreenPriv;
 
     VIA_FLUSH_DMA(vmesa);
    if (dPriv->vblFlags == VBLANK_FLAG_SYNC &&
@@ -885,9 +885,6 @@ void viaFlushDmaLocked(struct via_context *vmesa, GLuint flags)
    }
    else if (vmesa->numClipRects) {
       drm_clip_rect_t *pbox = vmesa->pClipRects;
-      __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
-      struct via_renderbuffer *const vrb = 
-	(struct via_renderbuffer *) dPriv->driverPrivate;
 
       for (i = 0; i < vmesa->numClipRects; i++) {
 	 drm_clip_rect_t b;
diff --git a/src/mesa/drivers/dri/unichrome/via_ioctl.h b/src/mesa/drivers/dri/unichrome/via_ioctl.h
index 14a833a97d..c6b32cf085 100644
--- a/src/mesa/drivers/dri/unichrome/via_ioctl.h
+++ b/src/mesa/drivers/dri/unichrome/via_ioctl.h
@@ -33,8 +33,8 @@ void viaFlushDma(struct via_context *vmesa);
 void viaFlushDmaLocked(struct via_context *vmesa, GLuint flags);
 
 void viaInitIoctlFuncs(GLcontext *ctx);
-void viaCopyBuffer(__DRIdrawablePrivate *dpriv);
-void viaPageFlip(__DRIdrawablePrivate *dpriv);
+void viaCopyBuffer(__DRIdrawable *dpriv);
+void viaPageFlip(__DRIdrawable *dpriv);
 void viaCheckDma(struct via_context *vmesa, GLuint bytes);
 void viaResetPageFlippingLocked(struct via_context *vmesa);
 void viaWaitIdle(struct via_context *vmesa, GLboolean light);
diff --git a/src/mesa/drivers/dri/unichrome/via_screen.c b/src/mesa/drivers/dri/unichrome/via_screen.c
index 3dbb570571..2cfb98317d 100644
--- a/src/mesa/drivers/dri/unichrome/via_screen.c
+++ b/src/mesa/drivers/dri/unichrome/via_screen.c
@@ -62,8 +62,6 @@ DRI_CONF_BEGIN
 DRI_CONF_END;
 static const GLuint __driNConfigOptions = 3;
 
-extern const struct dri_extension card_extensions[];
-
 static drmBufMapPtr via_create_empty_buffers(void)
 {
     drmBufMapPtr retval;
@@ -92,7 +90,7 @@ static void via_free_empty_buffers( drmBufMapPtr bufs )
 
 
 static GLboolean
-viaInitDriver(__DRIscreenPrivate *sPriv)
+viaInitDriver(__DRIscreen *sPriv)
 {
     viaScreenPrivate *viaScreen;
     VIADRIPtr gDRIPriv = (VIADRIPtr)sPriv->pDevPriv;
@@ -186,7 +184,7 @@ viaInitDriver(__DRIscreenPrivate *sPriv)
 }
 
 static void
-viaDestroyScreen(__DRIscreenPrivate *sPriv)
+viaDestroyScreen(__DRIscreen *sPriv)
 {
     viaScreenPrivate *viaScreen = (viaScreenPrivate *)sPriv->private;
     VIADRIPtr gDRIPriv = (VIADRIPtr)sPriv->pDevPriv;
@@ -205,12 +203,14 @@ viaDestroyScreen(__DRIscreenPrivate *sPriv)
 
 
 static GLboolean
-viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
-                __DRIdrawablePrivate *driDrawPriv,
+viaCreateBuffer(__DRIscreen *driScrnPriv,
+                __DRIdrawable *driDrawPriv,
                 const __GLcontextModes *mesaVis,
                 GLboolean isPixmap)
 {
+#if 0
     viaScreenPrivate *screen = (viaScreenPrivate *) driScrnPriv->private;
+#endif
 
     GLboolean swStencil = (mesaVis->stencilBits > 0 && 
 			   mesaVis->depthBits != 24);
@@ -248,7 +248,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
       /* XXX check/fix the offset/pitch parameters! */
       {
          driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(frontRb, mesaVis);
@@ -257,7 +257,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->doubleBufferMode) {
          driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_ARGB8888, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(backRb, mesaVis);
@@ -266,7 +266,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->depthBits == 16) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_Z16, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(depthRb, mesaVis);
@@ -274,7 +274,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 24) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_Z24_S8, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(depthRb, mesaVis);
@@ -282,7 +282,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
       }
       else if (mesaVis->depthBits == 32) {
          driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT32, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_Z32, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(depthRb, mesaVis);
@@ -291,7 +291,7 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
       if (mesaVis->stencilBits > 0 && !swStencil) {
          driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT, NULL,
+            = driNewRenderbuffer(MESA_FORMAT_S8, NULL,
                                  screen->bytesPerPixel,
                                  0, screen->width, driDrawPriv);
          viaSetSpanFunctions(stencilRb, mesaVis);
@@ -314,13 +314,13 @@ viaCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 
 
 static void
-viaDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+viaDestroyBuffer(__DRIdrawable *driDrawPriv)
 {
    _mesa_reference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)), NULL);
 }
 
 static const __DRIconfig **
-viaFillInModes( __DRIscreenPrivate *psp,
+viaFillInModes( __DRIscreen *psp,
 		unsigned pixel_bits, GLboolean have_back_buffer )
 {
     __DRIconfig **configs;
@@ -377,7 +377,7 @@ viaFillInModes( __DRIscreenPrivate *psp,
  * \return the __GLcontextModes supported by this driver
  */
 static const __DRIconfig **
-viaInitScreen(__DRIscreenPrivate *psp)
+viaInitScreen(__DRIscreen *psp)
 {
    static const __DRIversion ddx_expected = { VIA_DRIDDX_VERSION_MAJOR,
                                               VIA_DRIDDX_VERSION_MINOR,
@@ -393,18 +393,6 @@ viaInitScreen(__DRIscreenPrivate *psp)
 				      &psp->drm_version, & drm_expected) )
       return NULL;
 
-   /* Calling driInitExtensions here, with a NULL context pointer,
-    * does not actually enable the extensions.  It just makes sure
-    * that all the dispatch offsets for all the extensions that
-    * *might* be enables are known.  This is needed because the
-    * dispatch offsets need to be known when _mesa_context_create is
-    * called, but we can't enable the extensions until we have a
-    * context pointer.
-    *
-    * Hello chicken.  Hello egg.  How are you two today?
-    */
-   driInitExtensions( NULL, card_extensions, GL_FALSE );
-
    if (!viaInitDriver(psp))
        return NULL;
 
@@ -417,7 +405,7 @@ viaInitScreen(__DRIscreenPrivate *psp)
  * Get information about previous buffer swaps.
  */
 static int
-getSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
+getSwapInfo( __DRIdrawable *dPriv, __DRIswapInfo * sInfo )
 {
    struct via_context *vmesa;
 
@@ -455,3 +443,10 @@ const struct __DriverAPIRec driDriverAPI = {
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL
 };
+
+/* This is the table of extensions that the loader will dlsym() for. */
+PUBLIC const __DRIextension *__driDriverExtensions[] = {
+    &driCoreExtension.base,
+    &driLegacyExtension.base,
+    NULL
+};
diff --git a/src/mesa/drivers/dri/unichrome/via_screen.h b/src/mesa/drivers/dri/unichrome/via_screen.h
index c3ef722ff0..aa662e01c0 100644
--- a/src/mesa/drivers/dri/unichrome/via_screen.h
+++ b/src/mesa/drivers/dri/unichrome/via_screen.h
@@ -61,7 +61,7 @@ typedef struct {
     drmAddress agpLinearStart;
     GLuint agpBase;
 
-    __DRIscreenPrivate *driScrnPriv;
+    __DRIscreen *driScrnPriv;
     drmBufMapPtr bufs;
     unsigned int sareaPrivOffset;
     /*=* John Sheng [2003.12.9] Tuxracer & VQ *=*/
@@ -77,21 +77,21 @@ typedef struct {
 
 extern GLboolean
 viaCreateContext(const __GLcontextModes *mesaVis,
-                 __DRIcontextPrivate *driContextPriv,
+                 __DRIcontext *driContextPriv,
                  void *sharedContextPrivate);
 
 extern void
-viaDestroyContext(__DRIcontextPrivate *driContextPriv);
+viaDestroyContext(__DRIcontext *driContextPriv);
 
 extern GLboolean
-viaUnbindContext(__DRIcontextPrivate *driContextPriv);
+viaUnbindContext(__DRIcontext *driContextPriv);
 
 extern GLboolean
-viaMakeCurrent(__DRIcontextPrivate *driContextPriv,
-               __DRIdrawablePrivate *driDrawPriv,
-               __DRIdrawablePrivate *driReadPriv);
+viaMakeCurrent(__DRIcontext *driContextPriv,
+               __DRIdrawable *driDrawPriv,
+               __DRIdrawable *driReadPriv);
 
 extern void
-viaSwapBuffers(__DRIdrawablePrivate *drawablePrivate);
+viaSwapBuffers(__DRIdrawable *drawablePrivate);
 
 #endif
diff --git a/src/mesa/drivers/dri/unichrome/via_span.c b/src/mesa/drivers/dri/unichrome/via_span.c
index b908f0fb23..fa3cbf7a79 100644
--- a/src/mesa/drivers/dri/unichrome/via_span.c
+++ b/src/mesa/drivers/dri/unichrome/via_span.c
@@ -23,6 +23,7 @@
  */
 
 #include "main/glheader.h"
+#include "main/formats.h"
 #include "main/macros.h"
 #include "main/mtypes.h"
 #include "main/colormac.h"
@@ -42,7 +43,7 @@
 #undef LOCAL_VARS
 #define LOCAL_VARS                                                   	\
     struct via_renderbuffer *vrb = (struct via_renderbuffer *) rb;   	\
-    __DRIdrawablePrivate *dPriv = vrb->dPriv;                           \
+    __DRIdrawable *dPriv = vrb->dPriv;                           \
     GLuint pitch = vrb->pitch;                                          \
     GLuint height = dPriv->h;                                        	\
     GLint p = 0;							\
@@ -79,7 +80,7 @@
  */
 #define LOCAL_DEPTH_VARS                                            \
     struct via_renderbuffer *vrb = (struct via_renderbuffer *) rb;  \
-    __DRIdrawablePrivate *dPriv = vrb->dPriv;                       \
+    __DRIdrawable *dPriv = vrb->dPriv;                       \
     GLuint depth_pitch = vrb->pitch;                                \
     GLuint height = dPriv->h;                                       \
     char *buf = (char *)(vrb->map)
@@ -177,24 +178,22 @@ void viaInitSpanFuncs(GLcontext *ctx)
 void
 viaSetSpanFunctions(struct via_renderbuffer *vrb, const GLvisual *vis)
 {
-   if (vrb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         viaInitPointers_565(&vrb->Base);
-      }
-      else {
-         viaInitPointers_8888(&vrb->Base);
-      }
+   if (vrb->Base.Format == MESA_FORMAT_RGB565) {
+      viaInitPointers_565(&vrb->Base);
    }
-   else if (vrb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
+   else if (vrb->Base.Format == MESA_FORMAT_ARGB8888) {
+      viaInitPointers_8888(&vrb->Base);
+   }
+   else if (vrb->Base.Format == MESA_FORMAT_Z16) {
       viaInitDepthPointers_z16(&vrb->Base);
    }
-   else if (vrb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
+   else if (vrb->Base.Format == MESA_FORMAT_Z24_S8) {
       viaInitDepthPointers_z24_s8(&vrb->Base);
    }
-   else if (vrb->Base.InternalFormat == GL_DEPTH_COMPONENT32) {
+   else if (vrb->Base.Format == MESA_FORMAT_Z32) {
       viaInitDepthPointers_z32(&vrb->Base);
    }
-   else if (vrb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
+   else if (vrb->Base.Format == MESA_FORMAT_S8) {
       viaInitStencilPointers_z24_s8(&vrb->Base);
    }
 }
diff --git a/src/mesa/drivers/dri/unichrome/via_state.c b/src/mesa/drivers/dri/unichrome/via_state.c
index 840e4e42da..e6e5526d34 100644
--- a/src/mesa/drivers/dri/unichrome/via_state.c
+++ b/src/mesa/drivers/dri/unichrome/via_state.c
@@ -476,7 +476,7 @@ void viaEmitState(struct via_context *vmesa)
     */
    if (ctx->Polygon.StippleFlag) {
       GLuint *stipple = &ctx->PolygonStipple[0];
-      __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
+      __DRIdrawable *dPriv = vmesa->driDrawable;
       struct via_renderbuffer *const vrb = 
 	(struct via_renderbuffer *) dPriv->driverPrivate;
       GLint i;
@@ -722,7 +722,7 @@ static void viaColorMask(GLcontext *ctx,
 void viaCalcViewport(GLcontext *ctx)
 {
     struct via_context *vmesa = VIA_CONTEXT(ctx);
-    __DRIdrawablePrivate *dPriv = vmesa->driDrawable;
+    __DRIdrawable *dPriv = vmesa->driDrawable;
     struct via_renderbuffer *const vrb = 
       (struct via_renderbuffer *) dPriv->driverPrivate;
     const GLfloat *v = ctx->Viewport._WindowMap.m;
@@ -891,10 +891,10 @@ static GLboolean viaChooseTextureState(GLcontext *ctx)
             if (texObj->Image[0][texObj->BaseLevel]->Border > 0) {
 	       vmesa->regHTXnTB[0] |= (HC_HTXnTB_TBC_S | HC_HTXnTB_TBC_T);
 	       vmesa->regHTXnTBC[0] = 
-		  PACK_COLOR_888(FLOAT_TO_UBYTE(texObj->BorderColor[0]),
-				 FLOAT_TO_UBYTE(texObj->BorderColor[1]),
-				 FLOAT_TO_UBYTE(texObj->BorderColor[2]));
-	       vmesa->regHTXnTRAH[0] = FLOAT_TO_UBYTE(texObj->BorderColor[3]);
+		  PACK_COLOR_888(FLOAT_TO_UBYTE(texObj->BorderColor.f[0]),
+				 FLOAT_TO_UBYTE(texObj->BorderColor.f[1]),
+				 FLOAT_TO_UBYTE(texObj->BorderColor.f[2]));
+	       vmesa->regHTXnTRAH[0] = FLOAT_TO_UBYTE(texObj->BorderColor.f[3]);
             }
 
 	    if (texUnit0->LodBias != 0.0f) {
@@ -924,10 +924,10 @@ static GLboolean viaChooseTextureState(GLcontext *ctx)
             if (texObj->Image[0][texObj->BaseLevel]->Border > 0) {
 	       vmesa->regHTXnTB[1] |= (HC_HTXnTB_TBC_S | HC_HTXnTB_TBC_T);
 	       vmesa->regHTXnTBC[1] = 
-		  PACK_COLOR_888(FLOAT_TO_UBYTE(texObj->BorderColor[0]),
-				 FLOAT_TO_UBYTE(texObj->BorderColor[1]),
-				 FLOAT_TO_UBYTE(texObj->BorderColor[2]));
-	       vmesa->regHTXnTRAH[1] = FLOAT_TO_UBYTE(texObj->BorderColor[3]);
+		  PACK_COLOR_888(FLOAT_TO_UBYTE(texObj->BorderColor.f[0]),
+				 FLOAT_TO_UBYTE(texObj->BorderColor.f[1]),
+				 FLOAT_TO_UBYTE(texObj->BorderColor.f[2]));
+	       vmesa->regHTXnTRAH[1] = FLOAT_TO_UBYTE(texObj->BorderColor.f[3]);
             }
 
 
@@ -1238,12 +1238,12 @@ static void viaChooseColorState(GLcontext *ctx)
     else
         vmesa->regHROP = HC_HROP_P;
 
-    vmesa->regHFBBMSKL = PACK_COLOR_888(ctx->Color.ColorMask[0],
-					ctx->Color.ColorMask[1],
-					ctx->Color.ColorMask[2]);
-    vmesa->regHROP |= ctx->Color.ColorMask[3];
+    vmesa->regHFBBMSKL = PACK_COLOR_888(ctx->Color.ColorMask[0][0],
+					ctx->Color.ColorMask[0][1],
+					ctx->Color.ColorMask[0][2]);
+    vmesa->regHROP |= ctx->Color.ColorMask[0][3];
 
-    if (ctx->Color.ColorMask[3])
+    if (ctx->Color.ColorMask[0][3])
         vmesa->regEnable |= HC_HenAW_MASK;
     else
         vmesa->regEnable &= ~HC_HenAW_MASK;
diff --git a/src/mesa/drivers/dri/unichrome/via_tex.c b/src/mesa/drivers/dri/unichrome/via_tex.c
index 54073e7691..24924d2613 100644
--- a/src/mesa/drivers/dri/unichrome/via_tex.c
+++ b/src/mesa/drivers/dri/unichrome/via_tex.c
@@ -30,17 +30,17 @@
 #include "main/macros.h"
 #include "main/mtypes.h"
 #include "main/enums.h"
+#include "main/formats.h"
 #include "main/colortab.h"
 #include "main/convolve.h"
 #include "main/context.h"
 #include "main/mipmap.h"
+#include "main/mm.h"
 #include "main/simple_list.h"
 #include "main/texcompress.h"
-#include "main/texformat.h"
 #include "main/texobj.h"
 #include "main/texstore.h"
 
-#include "main/mm.h"
 #include "via_context.h"
 #include "via_fb.h"
 #include "via_tex.h"
@@ -48,7 +48,7 @@
 #include "via_ioctl.h"
 #include "via_3d_reg.h"
 
-static const struct gl_texture_format *
+static gl_format
 viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
 		    GLenum format, GLenum type )
 {
@@ -65,56 +65,56 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
       if ( format == GL_BGRA ) {
 	 if ( type == GL_UNSIGNED_INT_8_8_8_8_REV ||
 	      type == GL_UNSIGNED_BYTE ) {
-	    return &_mesa_texformat_argb8888;
+	    return MESA_FORMAT_ARGB8888;
 	 }
          else if ( type == GL_UNSIGNED_SHORT_4_4_4_4_REV ) {
-            return &_mesa_texformat_argb4444;
+            return MESA_FORMAT_ARGB4444;
 	 }
          else if ( type == GL_UNSIGNED_SHORT_1_5_5_5_REV ) {
-	    return &_mesa_texformat_argb1555;
+	    return MESA_FORMAT_ARGB1555;
 	 }
       }
       else if ( type == GL_UNSIGNED_BYTE ||
 		type == GL_UNSIGNED_INT_8_8_8_8_REV ||
 		type == GL_UNSIGNED_INT_8_8_8_8 ) {
-	 return &_mesa_texformat_argb8888;
+	 return MESA_FORMAT_ARGB8888;
       }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
 
    case 3:
    case GL_RGB:
    case GL_COMPRESSED_RGB:
       if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
-	 return &_mesa_texformat_rgb565;
+	 return MESA_FORMAT_RGB565;
       }
       else if ( type == GL_UNSIGNED_BYTE ) {
-	 return &_mesa_texformat_argb8888;
+	 return MESA_FORMAT_ARGB8888;
       }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
+      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_RGB565;
 
    case GL_RGBA8:
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      return &_mesa_texformat_argb8888;
+      return MESA_FORMAT_ARGB8888;
 
    case GL_RGBA4:
    case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
+      return MESA_FORMAT_ARGB4444;
 
    case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
+      return MESA_FORMAT_ARGB1555;
 
    case GL_RGB8:
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
-      return &_mesa_texformat_argb8888;
+      return MESA_FORMAT_ARGB8888;
 
    case GL_RGB5:
    case GL_RGB4:
    case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
+      return MESA_FORMAT_RGB565;
 
    case GL_ALPHA:
    case GL_ALPHA4:
@@ -122,7 +122,7 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
    case GL_ALPHA12:
    case GL_ALPHA16:
    case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
+      return MESA_FORMAT_A8;
 
    case 1:
    case GL_LUMINANCE:
@@ -131,7 +131,7 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
    case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
+      return MESA_FORMAT_L8;
 
    case 2:
    case GL_LUMINANCE_ALPHA:
@@ -142,7 +142,7 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12_ALPHA12:
    case GL_LUMINANCE16_ALPHA16:
    case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
+      return MESA_FORMAT_AL88;
 
    case GL_INTENSITY:
    case GL_INTENSITY4:
@@ -150,35 +150,35 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY12:
    case GL_INTENSITY16:
    case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
+      return MESA_FORMAT_I8;
 
    case GL_YCBCR_MESA:
       if (type == GL_UNSIGNED_SHORT_8_8_MESA ||
 	  type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
+         return MESA_FORMAT_YCBCR;
       else
-         return &_mesa_texformat_ycbcr_rev;
+         return MESA_FORMAT_YCBCR_REV;
 
    case GL_COMPRESSED_RGB_FXT1_3DFX:
-      return &_mesa_texformat_rgb_fxt1;
+      return MESA_FORMAT_RGB_FXT1;
    case GL_COMPRESSED_RGBA_FXT1_3DFX:
-      return &_mesa_texformat_rgba_fxt1;
+      return MESA_FORMAT_RGBA_FXT1;
 
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
    case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgb_dxt1;
+      return MESA_FORMAT_RGB_DXT1;
 
    case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-      return &_mesa_texformat_rgba_dxt1;
+      return MESA_FORMAT_RGBA_DXT1;
 
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-      return &_mesa_texformat_rgba_dxt3;
+      return MESA_FORMAT_RGBA_DXT3;
 
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
+      return MESA_FORMAT_RGBA_DXT5;
 
    case GL_COLOR_INDEX:	
    case GL_COLOR_INDEX1_EXT:	
@@ -187,16 +187,16 @@ viaChooseTexFormat( GLcontext *ctx, GLint internalFormat,
    case GL_COLOR_INDEX8_EXT:	
    case GL_COLOR_INDEX12_EXT:	    
    case GL_COLOR_INDEX16_EXT:
-      return &_mesa_texformat_ci8;    
+      return MESA_FORMAT_CI8;
 
    default:
       fprintf(stderr, "unexpected texture format %s in %s\n", 
 	      _mesa_lookup_enum_by_nr(internalFormat),
 	      __FUNCTION__);
-      return NULL;
+      return MESA_FORMAT_NONE;
    }
 
-   return NULL; /* never get here */
+   return MESA_FORMAT_NONE; /* never get here */
 }
 
 static int logbase2(int n)
@@ -457,7 +457,7 @@ static GLboolean viaSetTexImages(GLcontext *ctx,
    GLuint widthExp = 0;
    GLuint heightExp = 0;    
 
-   switch (baseImage->image.TexFormat->MesaFormat) {
+   switch (baseImage->image.TexFormat) {
    case MESA_FORMAT_ARGB8888:
       texFormat = HC_HTXnFM_ARGB8888;
       break;
@@ -689,24 +689,7 @@ static void viaTexImage(GLcontext *ctx,
 
    assert(texImage->TexFormat);
 
-   if (dims == 1) {
-      texImage->FetchTexelc = texImage->TexFormat->FetchTexel1D;
-      texImage->FetchTexelf = texImage->TexFormat->FetchTexel1Df;
-   }
-   else {
-      texImage->FetchTexelc = texImage->TexFormat->FetchTexel2D;
-      texImage->FetchTexelf = texImage->TexFormat->FetchTexel2Df;
-   }
-   texelBytes = texImage->TexFormat->TexelBytes;
-
-   if (texelBytes == 0) {
-      /* compressed format */
-      texImage->IsCompressed = GL_TRUE;
-      texImage->CompressedSize =
-         ctx->Driver.CompressedTextureSize(ctx, texImage->Width,
-                                           texImage->Height, texImage->Depth,
-                                           texImage->TexFormat->MesaFormat);
-   }
+   texelBytes = _mesa_get_format_bytes(texImage->TexFormat);
 
    /* Minimum pitch of 32 bytes */
    if (postConvWidth * texelBytes < 32) {
@@ -718,8 +701,11 @@ static void viaTexImage(GLcontext *ctx,
    viaImage->pitchLog2 = logbase2(postConvWidth * texelBytes);
 
    /* allocate memory */
-   if (texImage->IsCompressed)
-      sizeInBytes = texImage->CompressedSize;
+   if (_mesa_is_format_compressed(texImage->TexFormat))
+      sizeInBytes = _mesa_format_image_size(texImage->TexFormat,
+                                            texImage->Width,
+                                            texImage->Height,
+                                            texImage->Depth);
    else
       sizeInBytes = postConvWidth * postConvHeight * texelBytes;
 
@@ -797,22 +783,22 @@ static void viaTexImage(GLcontext *ctx,
    else {
       GLint dstRowStride;
       GLboolean success;
-      if (texImage->IsCompressed) {
-         dstRowStride = _mesa_compressed_row_stride(texImage->TexFormat->MesaFormat, width);
+
+      if (_mesa_is_format_compressed(texImage->TexFormat)) {
+         dstRowStride = _mesa_format_row_stride(texImage->TexFormat, width);
       }
       else {
-         dstRowStride = postConvWidth * texImage->TexFormat->TexelBytes;
+         dstRowStride = postConvWidth * _mesa_get_format_bytes(texImage->TexFormat);
       }
-      ASSERT(texImage->TexFormat->StoreImage);
-      success = texImage->TexFormat->StoreImage(ctx, dims,
-                                                texImage->_BaseFormat,
-                                                texImage->TexFormat,
-                                                texImage->Data,
-                                                0, 0, 0,  /* dstX/Y/Zoffset */
-                                                dstRowStride,
-                                                texImage->ImageOffsets,
-                                                width, height, 1,
-                                                format, type, pixels, packing);
+      success = _mesa_texstore(ctx, dims,
+                               texImage->_BaseFormat,
+                               texImage->TexFormat,
+                               texImage->Data,
+                               0, 0, 0,  /* dstX/Y/Zoffset */
+                               dstRowStride,
+                               texImage->ImageOffsets,
+                               width, height, 1,
+                               format, type, pixels, packing);
       if (!success) {
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage");
       }
diff --git a/src/mesa/drivers/dri/unichrome/via_tris.c b/src/mesa/drivers/dri/unichrome/via_tris.c
index 79e67620c9..01359d51ea 100644
--- a/src/mesa/drivers/dri/unichrome/via_tris.c
+++ b/src/mesa/drivers/dri/unichrome/via_tris.c
@@ -330,7 +330,8 @@ do {							\
 
 #define LOCAL_VARS(n)                                                   \
     struct via_context *vmesa = VIA_CONTEXT(ctx);                             \
-    GLuint color[n], spec[n];                                           \
+    GLuint color[n] = { 0 };                                          \
+    GLuint spec[n] = { 0 };                                           \
     GLuint coloroffset = vmesa->coloroffset;              \
     GLuint specoffset = vmesa->specoffset;                       \
     (void)color; (void)spec; (void)coloroffset; (void)specoffset;
@@ -832,13 +833,13 @@ static GLboolean viaCheckPTexHack( GLcontext *ctx )
 
    RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->TexCoordPtr[0]->size == 4) {
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX0 ) && VB->AttribPtr[_TNL_ATTRIB_TEX0]->size == 4) {
       if (!RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_ATTRIB_TEX1, _TNL_LAST_TEX ))
 	 ptexHack = GL_TRUE; 
       else
 	 fallback = GL_TRUE;
    }
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->TexCoordPtr[1]->size == 4)
+   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX1 ) && VB->AttribPtr[_TNL_ATTRIB_TEX1]->size == 4)
       fallback = GL_TRUE;
 
    FALLBACK(VIA_CONTEXT(ctx), VIA_FALLBACK_PROJ_TEXTURE, fallback);