Merge branch 'master' into glsl2

author: Ian Romanick <ian.d.romanick@intel.com> 2010-08-13 20:22:46 -0700
committer: Ian Romanick <ian.d.romanick@intel.com> 2010-08-13 20:22:46 -0700
commit: 443a7e4e9a360acbc3e662c098be436f180bf81d (patch)
tree: 71125fc394bdcbc17cd7ed74c592b9a7761541f8 /src/mesa
parent: 2f8ee757ab324d599fcb8287789eb5f1a7890d74 (diff)
parent: 27041d7cb3faeaed483538a228573466363ec1c7 (diff)
174 files changed, 1654 insertions, 1158 deletions
diff --git a/src/mesa/drivers/dri/common/dri_metaops.c b/src/mesa/drivers/dri/common/dri_metaops.c
index 86e59a8e51..a2f404b616 100644
--- a/src/mesa/drivers/dri/common/dri_metaops.c
+++ b/src/mesa/drivers/dri/common/dri_metaops.c
@@ -29,6 +29,7 @@
 #include "main/arbprogram.h"
 #include "main/arrayobj.h"
 #include "main/bufferobj.h"
+#include "main/context.h"
 #include "main/enable.h"
 #include "main/matrix.h"
 #include "main/texstate.h"
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 71ee753748..65fd658c04 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -56,7 +56,7 @@ C_SOURCES = \
 
 ASM_SOURCES = 
 
-DRIVER_DEFINES = -I../intel -I../intel/server -DI915 \
+DRIVER_DEFINES = -I../intel -DI915 \
 	$(shell pkg-config libdrm --atleast-version=2.3.1 \
 				&& echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP")
 
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 831981558d..e381a5c714 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -106,7 +106,7 @@ C_SOURCES = \
 
 ASM_SOURCES = 
 
-DRIVER_DEFINES = -I../intel -I../intel/server
+DRIVER_DEFINES = -I../intel
 
 INCLUDES += $(INTEL_CFLAGS)
 DRI_LIB_DEPS += $(INTEL_LIBS)
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index 40eece276b..af08446f2d 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -46,68 +46,68 @@ brw_add_validated_bo(struct brw_context *brw, drm_intel_bo *bo)
    }
 };
 
-const struct brw_tracked_state brw_blend_constant_color;
-const struct brw_tracked_state brw_cc_unit;
-const struct brw_tracked_state brw_check_fallback;
-const struct brw_tracked_state brw_clip_prog;
-const struct brw_tracked_state brw_clip_unit;
-const struct brw_tracked_state brw_vs_constants;
-const struct brw_tracked_state brw_wm_constants;
-const struct brw_tracked_state brw_constant_buffer;
-const struct brw_tracked_state brw_curbe_offsets;
-const struct brw_tracked_state brw_invarient_state;
-const struct brw_tracked_state brw_gs_prog;
-const struct brw_tracked_state brw_gs_unit;
-const struct brw_tracked_state brw_line_stipple;
-const struct brw_tracked_state brw_aa_line_parameters;
-const struct brw_tracked_state brw_pipelined_state_pointers;
-const struct brw_tracked_state brw_binding_table_pointers;
-const struct brw_tracked_state brw_depthbuffer;
-const struct brw_tracked_state brw_polygon_stipple_offset;
-const struct brw_tracked_state brw_polygon_stipple;
-const struct brw_tracked_state brw_program_parameters;
-const struct brw_tracked_state brw_recalculate_urb_fence;
-const struct brw_tracked_state brw_sf_prog;
-const struct brw_tracked_state brw_sf_unit;
-const struct brw_tracked_state brw_sf_vp;
-const struct brw_tracked_state brw_state_base_address;
-const struct brw_tracked_state brw_urb_fence;
-const struct brw_tracked_state brw_vertex_state;
-const struct brw_tracked_state brw_vs_surfaces;
-const struct brw_tracked_state brw_vs_prog;
-const struct brw_tracked_state brw_vs_unit;
-const struct brw_tracked_state brw_wm_input_sizes;
-const struct brw_tracked_state brw_wm_prog;
-const struct brw_tracked_state brw_wm_samplers;
-const struct brw_tracked_state brw_wm_constant_surface;
-const struct brw_tracked_state brw_wm_surfaces;
-const struct brw_tracked_state brw_wm_binding_table;
-const struct brw_tracked_state brw_wm_unit;
-
-const struct brw_tracked_state brw_psp_urb_cbs;
-
-const struct brw_tracked_state brw_pipe_control;
-
-const struct brw_tracked_state brw_drawing_rect;
-const struct brw_tracked_state brw_indices;
-const struct brw_tracked_state brw_vertices;
-const struct brw_tracked_state brw_index_buffer;
-const struct brw_tracked_state gen6_binding_table_pointers;
-const struct brw_tracked_state gen6_blend_state;
-const struct brw_tracked_state gen6_cc_state_pointers;
-const struct brw_tracked_state gen6_clip_state;
-const struct brw_tracked_state gen6_clip_vp;
-const struct brw_tracked_state gen6_color_calc_state;
-const struct brw_tracked_state gen6_depth_stencil_state;
-const struct brw_tracked_state gen6_gs_state;
-const struct brw_tracked_state gen6_sampler_state;
-const struct brw_tracked_state gen6_scissor_state;
-const struct brw_tracked_state gen6_sf_state;
-const struct brw_tracked_state gen6_sf_vp;
-const struct brw_tracked_state gen6_urb;
-const struct brw_tracked_state gen6_viewport_state;
-const struct brw_tracked_state gen6_vs_state;
-const struct brw_tracked_state gen6_wm_state;
+extern const struct brw_tracked_state brw_blend_constant_color;
+extern const struct brw_tracked_state brw_cc_unit;
+extern const struct brw_tracked_state brw_check_fallback;
+extern const struct brw_tracked_state brw_clip_prog;
+extern const struct brw_tracked_state brw_clip_unit;
+extern const struct brw_tracked_state brw_vs_constants;
+extern const struct brw_tracked_state brw_wm_constants;
+extern const struct brw_tracked_state brw_constant_buffer;
+extern const struct brw_tracked_state brw_curbe_offsets;
+extern const struct brw_tracked_state brw_invarient_state;
+extern const struct brw_tracked_state brw_gs_prog;
+extern const struct brw_tracked_state brw_gs_unit;
+extern const struct brw_tracked_state brw_line_stipple;
+extern const struct brw_tracked_state brw_aa_line_parameters;
+extern const struct brw_tracked_state brw_pipelined_state_pointers;
+extern const struct brw_tracked_state brw_binding_table_pointers;
+extern const struct brw_tracked_state brw_depthbuffer;
+extern const struct brw_tracked_state brw_polygon_stipple_offset;
+extern const struct brw_tracked_state brw_polygon_stipple;
+extern const struct brw_tracked_state brw_program_parameters;
+extern const struct brw_tracked_state brw_recalculate_urb_fence;
+extern const struct brw_tracked_state brw_sf_prog;
+extern const struct brw_tracked_state brw_sf_unit;
+extern const struct brw_tracked_state brw_sf_vp;
+extern const struct brw_tracked_state brw_state_base_address;
+extern const struct brw_tracked_state brw_urb_fence;
+extern const struct brw_tracked_state brw_vertex_state;
+extern const struct brw_tracked_state brw_vs_surfaces;
+extern const struct brw_tracked_state brw_vs_prog;
+extern const struct brw_tracked_state brw_vs_unit;
+extern const struct brw_tracked_state brw_wm_input_sizes;
+extern const struct brw_tracked_state brw_wm_prog;
+extern const struct brw_tracked_state brw_wm_samplers;
+extern const struct brw_tracked_state brw_wm_constant_surface;
+extern const struct brw_tracked_state brw_wm_surfaces;
+extern const struct brw_tracked_state brw_wm_binding_table;
+extern const struct brw_tracked_state brw_wm_unit;
+
+extern const struct brw_tracked_state brw_psp_urb_cbs;
+
+extern const struct brw_tracked_state brw_pipe_control;
+
+extern const struct brw_tracked_state brw_drawing_rect;
+extern const struct brw_tracked_state brw_indices;
+extern const struct brw_tracked_state brw_vertices;
+extern const struct brw_tracked_state brw_index_buffer;
+extern const struct brw_tracked_state gen6_binding_table_pointers;
+extern const struct brw_tracked_state gen6_blend_state;
+extern const struct brw_tracked_state gen6_cc_state_pointers;
+extern const struct brw_tracked_state gen6_clip_state;
+extern const struct brw_tracked_state gen6_clip_vp;
+extern const struct brw_tracked_state gen6_color_calc_state;
+extern const struct brw_tracked_state gen6_depth_stencil_state;
+extern const struct brw_tracked_state gen6_gs_state;
+extern const struct brw_tracked_state gen6_sampler_state;
+extern const struct brw_tracked_state gen6_scissor_state;
+extern const struct brw_tracked_state gen6_sf_state;
+extern const struct brw_tracked_state gen6_sf_vp;
+extern const struct brw_tracked_state gen6_urb;
+extern const struct brw_tracked_state gen6_viewport_state;
+extern const struct brw_tracked_state gen6_vs_state;
+extern const struct brw_tracked_state gen6_wm_state;
 
 /***********************************************************************
  * brw_state.c
diff --git a/src/mesa/drivers/dri/i965/brw_util.c b/src/mesa/drivers/dri/i965/brw_util.c
index 1db2a210d4..e878da3850 100644
--- a/src/mesa/drivers/dri/i965/brw_util.c
+++ b/src/mesa/drivers/dri/i965/brw_util.c
@@ -30,6 +30,8 @@
   */
          
 
+#include <assert.h>
+
 #include "main/mtypes.h"
 #include "program/prog_parameter.h"
 #include "brw_util.h"
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 5f2035d79c..e19f44035f 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -29,6 +29,7 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/extensions.h"
+#include "main/fbobject.h"
 #include "main/framebuffer.h"
 #include "main/imports.h"
 #include "main/points.h"
@@ -39,8 +40,6 @@
 #include "drivers/common/driverfuncs.h"
 #include "drivers/common/meta.h"
 
-#include "i830_dri.h"
-
 #include "intel_chipset.h"
 #include "intel_buffers.h"
 #include "intel_tex.h"
@@ -420,7 +419,7 @@ intel_prepare_render(struct intel_context *intel)
    __DRIdrawable *drawable;
 
    drawable = driContext->driDrawablePriv;
-   if (drawable->dri2.stamp != driContext->dri2.draw_stamp) {
+   if (drawable && drawable->dri2.stamp != driContext->dri2.draw_stamp) {
       if (drawable->lastStamp != drawable->dri2.stamp)
 	 intel_update_renderbuffers(driContext, drawable);
       intel_draw_buffer(&intel->ctx, intel->ctx.DrawBuffer);
@@ -428,7 +427,7 @@ intel_prepare_render(struct intel_context *intel)
    }
 
    drawable = driContext->driReadablePriv;
-   if (drawable->dri2.stamp != driContext->dri2.read_stamp) {
+   if (drawable && drawable->dri2.stamp != driContext->dri2.read_stamp) {
       if (drawable->lastStamp != drawable->dri2.stamp)
 	 intel_update_renderbuffers(driContext, drawable);
       driContext->dri2.read_stamp = drawable->dri2.stamp;
@@ -613,6 +612,7 @@ intelInitContext(struct intel_context *intel,
    __DRIscreen *sPriv = driContextPriv->driScreenPriv;
    struct intel_screen *intelScreen = sPriv->private;
    int bo_reuse_mode;
+   __GLcontextModes visual;
 
    /* we can't do anything without a connection to the device */
    if (intelScreen->bufmgr == NULL)
@@ -624,6 +624,11 @@ intelInitContext(struct intel_context *intel,
       functions->Viewport = intel_viewport;
    }
 
+   if (mesaVis == NULL) {
+      memset(&visual, 0, sizeof visual);
+      mesaVis = &visual;
+   }
+
    if (!_mesa_initialize_context_for_api(&intel->ctx, api, mesaVis, shareCtx,
 					 functions, (void *) intel)) {
       printf("%s: failed to init mesa context\n", __FUNCTION__);
@@ -890,14 +895,21 @@ intelMakeCurrent(__DRIcontext * driContextPriv,
    }
 
    if (driContextPriv) {
-      struct gl_framebuffer *fb = driDrawPriv->driverPrivate;
-      struct gl_framebuffer *readFb = driReadPriv->driverPrivate;
+      struct gl_framebuffer *fb, *readFb;
+      
+      if (driDrawPriv == NULL && driReadPriv == NULL) {
+	 fb = _mesa_get_incomplete_framebuffer();
+	 readFb = _mesa_get_incomplete_framebuffer();
+      } else {
+	 fb = driDrawPriv->driverPrivate;
+	 readFb = driReadPriv->driverPrivate;
+	 driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
+	 driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
+      }
 
-      driContextPriv->dri2.draw_stamp = driDrawPriv->dri2.stamp - 1;
-      driContextPriv->dri2.read_stamp = driReadPriv->dri2.stamp - 1;
       intel_prepare_render(intel);
       _mesa_make_current(&intel->ctx, fb, readFb);
-
+      
       /* We do this in intel_prepare_render() too, but intel->ctx.DrawBuffer
        * is NULL at that point.  We can't call _mesa_makecurrent()
        * first, since we need the buffer size for the initial
diff --git a/src/mesa/drivers/dri/intel/intel_extensions_es2.c b/src/mesa/drivers/dri/intel/intel_extensions_es2.c
index baf8e13001..de34bbb2ae 100644
--- a/src/mesa/drivers/dri/intel/intel_extensions_es2.c
+++ b/src/mesa/drivers/dri/intel/intel_extensions_es2.c
@@ -28,7 +28,6 @@
 #include "main/extensions.h"
 
 #include "intel_extensions.h"
-#include "utils.h"
 
 static const char *es2_extensions[] = {
    /* Used by mesa internally (cf all_mesa_extensions in ../common/utils.c) */
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index 0e2fe893fe..02c0ffce31 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -45,6 +45,7 @@
 #include "main/attrib.h"
 #include "main/enable.h"
 #include "main/viewport.h"
+#include "main/context.h"
 #include "swrast/swrast.h"
 
 #include "intel_screen.h"
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index fe4de18960..680d18ba29 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -155,6 +155,9 @@ intel_region_alloc_internal(struct intel_context *intel,
    }
 
    region = calloc(sizeof(*region), 1);
+   if (region == NULL)
+      return region;
+
    region->cpp = cpp;
    region->width = width;
    region->height = height;
@@ -189,6 +192,9 @@ intel_region_alloc(struct intel_context *intel,
 
    region = intel_region_alloc_internal(intel, cpp, width, height,
 					aligned_pitch / cpp, buffer);
+   if (region == NULL)
+      return region;
+
    region->tiling = tiling;
 
    return region;
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 224b506c05..6efb2ddc55 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -102,7 +102,7 @@ do_copy_texsubimage(struct intel_context *intel,
    GLcontext *ctx = &intel->ctx;
    const struct intel_region *src = get_teximage_source(intel, internalFormat);
 
-   if (!intelImage->mt || !src) {
+   if (!intelImage->mt || !src || !src->buffer) {
       if (INTEL_DEBUG & DEBUG_FALLBACKS)
 	 fprintf(stderr, "%s fail %p %p (0x%08x)\n",
 		 __FUNCTION__, intelImage->mt, src, internalFormat);
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
index 5f813c0efa..e03b203fb4 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_format.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -19,7 +19,6 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
                          GLenum format, GLenum type)
 {
    struct intel_context *intel = intel_context(ctx);
-   const GLboolean do32bpt = (intel->ctx.Visual.rgbBits >= 24);
 
 #if 0
    printf("%s intFmt=0x%x format=0x%x type=0x%x\n",
@@ -30,39 +29,28 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case 4:
    case GL_RGBA:
    case GL_COMPRESSED_RGBA:
-      if (format == GL_BGRA) {
-         if (type == GL_UNSIGNED_BYTE || type == GL_UNSIGNED_INT_8_8_8_8_REV) {
-            return MESA_FORMAT_ARGB8888;
-         }
-         else if (type == GL_UNSIGNED_SHORT_4_4_4_4_REV) {
-            return MESA_FORMAT_ARGB4444;
-         }
-         else if (type == GL_UNSIGNED_SHORT_1_5_5_5_REV) {
-            return MESA_FORMAT_ARGB1555;
-         }
-      }
-      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
+      if (type == GL_UNSIGNED_SHORT_4_4_4_4_REV)
+	 return MESA_FORMAT_ARGB4444;
+      else if (type == GL_UNSIGNED_SHORT_1_5_5_5_REV)
+	 return MESA_FORMAT_ARGB1555;
+      else
+	 return MESA_FORMAT_ARGB8888;
 
    case 3:
    case GL_RGB:
    case GL_COMPRESSED_RGB:
-      if (format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5) {
-         return MESA_FORMAT_RGB565;
-      }
-      if (do32bpt) {
-	 if (intel->has_xrgb_textures)
-	    return MESA_FORMAT_XRGB8888;
-	 else
-	    return MESA_FORMAT_ARGB8888;
-      } else {
+      if (type == GL_UNSIGNED_SHORT_5_6_5)
 	 return MESA_FORMAT_RGB565;
-      }
+      else if (intel->has_xrgb_textures)
+	 return MESA_FORMAT_XRGB8888;
+      else
+	 return MESA_FORMAT_ARGB8888;
 
    case GL_RGBA8:
    case GL_RGB10_A2:
    case GL_RGBA12:
    case GL_RGBA16:
-      return do32bpt ? MESA_FORMAT_ARGB8888 : MESA_FORMAT_ARGB4444;
+      return MESA_FORMAT_ARGB8888;
 
    case GL_RGBA4:
    case GL_RGBA2:
diff --git a/src/mesa/drivers/dri/intel/server/i830_dri.h b/src/mesa/drivers/dri/intel/server/i830_dri.h
deleted file mode 100644
index def049e7a6..0000000000
--- a/src/mesa/drivers/dri/intel/server/i830_dri.h
+++ /dev/null
@@ -1,62 +0,0 @@
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_dri.h,v 1.6 2003/09/28 20:15:59 alanh Exp $ */
-
-#ifndef _I830_DRI_H
-#define _I830_DRI_H
-
-#include "xf86drm.h"
-
-#define I830_MAX_DRAWABLES 256
-
-#define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 9
-#define I830_PATCHLEVEL 0
-
-#define I830_REG_SIZE 0x80000
-
-typedef struct _I830DRIRec {
-   drm_handle_t regs;
-   drmSize regsSize;
-
-   drmSize unused1; /* backbufferSize */
-   drm_handle_t unused2; /* backbuffer */
-
-   drmSize unused3; /* depthbufferSize */
-   drm_handle_t unused4; /* depthbuffer */
-
-   drmSize unused5; /* rotatedSize */
-   drm_handle_t unused6; /* rotatedbuffer */
-
-   drm_handle_t unused7; /* textures */
-   int unused8; /* textureSize */
-
-   drm_handle_t unused9; /* agp_buffers */
-   drmSize unused10; /* agp_buf_size */
-
-   int deviceID;
-   int width;
-   int height;
-   int mem;
-   int cpp;
-   int bitsPerPixel;
-
-   int unused11[8]; /* was front/back/depth/rotated offset/pitch */
-
-   int unused12; /* logTextureGranularity */
-   int unused13; /* textureOffset */
-
-   int irq;
-   int sarea_priv_offset;
-} I830DRIRec, *I830DRIPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830ConfigPrivRec, *I830ConfigPrivPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830DRIContextRec, *I830DRIContextPtr;
-
-
-#endif
diff --git a/src/mesa/drivers/dri/intel/server/intel.h b/src/mesa/drivers/dri/intel/server/intel.h
deleted file mode 100644
index 6ea72499c1..0000000000
--- a/src/mesa/drivers/dri/intel/server/intel.h
+++ /dev/null
@@ -1,331 +0,0 @@
-#ifndef _INTEL_H_
-#define _INTEL_H_
-
-#include "xf86drm.h"		/* drm_handle_t, etc */
-
-/* Intel */
-#ifndef PCI_CHIP_I810
-#define PCI_CHIP_I810              0x7121
-#define PCI_CHIP_I810_DC100        0x7123
-#define PCI_CHIP_I810_E            0x7125
-#define PCI_CHIP_I815              0x1132
-#define PCI_CHIP_I810_BRIDGE       0x7120
-#define PCI_CHIP_I810_DC100_BRIDGE 0x7122
-#define PCI_CHIP_I810_E_BRIDGE     0x7124
-#define PCI_CHIP_I815_BRIDGE       0x1130
-#endif
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-
-#ifndef PCI_CHIP_I855_GM
-#define PCI_CHIP_I855_GM	   0x3582
-#define PCI_CHIP_I855_GM_BRIDGE	   0x3580
-#endif
-
-#ifndef PCI_CHIP_I865_G
-#define PCI_CHIP_I865_G		   0x2572
-#define PCI_CHIP_I865_G_BRIDGE	   0x2570
-#endif
-
-#ifndef PCI_CHIP_I915_G
-#define PCI_CHIP_I915_G		   0x2582
-#define PCI_CHIP_I915_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I915_GM
-#define PCI_CHIP_I915_GM	   0x2592
-#define PCI_CHIP_I915_GM_BRIDGE	   0x2590
-#endif
-
-#ifndef PCI_CHIP_E7221_G
-#define PCI_CHIP_E7221_G	   0x258A
-/* Same as I915_G_BRIDGE */
-#define PCI_CHIP_E7221_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I945_G
-#define PCI_CHIP_I945_G        0x2772
-#define PCI_CHIP_I945_G_BRIDGE 0x2770
-#endif
-
-#ifndef PCI_CHIP_I945_GM
-#define PCI_CHIP_I945_GM        0x27A2
-#define PCI_CHIP_I945_GM_BRIDGE 0x27A0
-#endif
-
-#define IS_I810(pI810) (pI810->Chipset == PCI_CHIP_I810 ||	\
-			pI810->Chipset == PCI_CHIP_I810_DC100 || \
-			pI810->Chipset == PCI_CHIP_I810_E)
-#define IS_I815(pI810) (pI810->Chipset == PCI_CHIP_I815)
-#define IS_I830(pI810) (pI810->Chipset == PCI_CHIP_I830_M)
-#define IS_845G(pI810) (pI810->Chipset == PCI_CHIP_845_G)
-#define IS_I85X(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM)
-#define IS_I852(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I852_GM || pI810->variant == I852_GME))
-#define IS_I855(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I855_GM || pI810->variant == I855_GME))
-#define IS_I865G(pI810) (pI810->Chipset == PCI_CHIP_I865_G)
-
-#define IS_I915G(pI810) (pI810->Chipset == PCI_CHIP_I915_G || pI810->Chipset == PCI_CHIP_E7221_G)
-#define IS_I915GM(pI810) (pI810->Chipset == PCI_CHIP_I915_GM)
-#define IS_I945G(pI810) (pI810->Chipset == PCI_CHIP_I945_G)
-#define IS_I945GM(pI810) (pI810->Chipset == PCI_CHIP_I945_GM)
-#define IS_I9XX(pI810) (IS_I915G(pI810) || IS_I915GM(pI810) || IS_I945G(pI810) || IS_I945GM(pI810))
-
-#define IS_MOBILE(pI810) (IS_I830(pI810) || IS_I85X(pI810) || IS_I915GM(pI810) || IS_I945GM(pI810))
-
-#define I830_GMCH_CTRL		0x52
-
-#define I830_GMCH_MEM_MASK      0x1
-#define I830_GMCH_MEM_64M       0x1
-#define I830_GMCH_MEM_128M      0
-
-#define I830_GMCH_GMS_MASK			0x70
-#define I830_GMCH_GMS_DISABLED		0x00
-#define I830_GMCH_GMS_LOCAL			0x10
-#define I830_GMCH_GMS_STOLEN_512	0x20
-#define I830_GMCH_GMS_STOLEN_1024	0x30
-#define I830_GMCH_GMS_STOLEN_8192	0x40
-
-#define I855_GMCH_GMS_MASK			(0x7 << 4)
-#define I855_GMCH_GMS_DISABLED			0x00
-#define I855_GMCH_GMS_STOLEN_1M			(0x1 << 4)
-#define I855_GMCH_GMS_STOLEN_4M			(0x2 << 4)
-#define I855_GMCH_GMS_STOLEN_8M			(0x3 << 4)
-#define I855_GMCH_GMS_STOLEN_16M		(0x4 << 4)
-#define I855_GMCH_GMS_STOLEN_32M		(0x5 << 4)
-#define I915G_GMCH_GMS_STOLEN_48M		(0x6 << 4)
-#define I915G_GMCH_GMS_STOLEN_64M		(0x7 << 4)
-
-typedef unsigned char Bool;
-#define TRUE 1
-#define FALSE 0
-
-#define PIPE_NONE	0<<0
-#define PIPE_CRT	1<<0
-#define PIPE_TV		1<<1
-#define PIPE_DFP	1<<2
-#define PIPE_LFP	1<<3
-#define PIPE_CRT2	1<<4
-#define PIPE_TV2	1<<5
-#define PIPE_DFP2	1<<6
-#define PIPE_LFP2	1<<7
-
-typedef struct _I830MemPool *I830MemPoolPtr;
-typedef struct _I830MemRange *I830MemRangePtr;
-typedef struct _I830MemRange {
-   long Start;
-   long End;
-   long Size;
-   unsigned long Physical;
-   unsigned long Offset;		/* Offset of AGP-allocated portion */
-   unsigned long Alignment;
-   drm_handle_t Key;
-   unsigned long Pitch; // add pitch
-   I830MemPoolPtr Pool;
-} I830MemRange;
-
-typedef struct _I830MemPool {
-   I830MemRange Total;
-   I830MemRange Free;
-   I830MemRange Fixed;
-   I830MemRange Allocated;
-} I830MemPool;
-
-typedef struct {
-   int tail_mask;
-   I830MemRange mem;
-   unsigned char *virtual_start;
-   int head;
-   int tail;
-   int space;
-} I830RingBuffer;
-
-typedef struct _I830Rec {
-   unsigned char *MMIOBase;
-   unsigned char *FbBase;
-   int cpp;
-   uint32_t aper_size;
-   unsigned int bios_version;
-
-   /* These are set in PreInit and never changed. */
-   long FbMapSize;
-   long TotalVideoRam;
-   I830MemRange StolenMemory;		/* pre-allocated memory */
-   long BIOSMemorySize;			/* min stolen pool size */
-   int BIOSMemSizeLoc;
-
-   /* These change according to what has been allocated. */
-   long FreeMemory;
-   I830MemRange MemoryAperture;
-   I830MemPool StolenPool;
-   long allocatedMemory;
-
-   /* Regions allocated either from the above pools, or from agpgart. */
-   /* for single and dual head configurations */
-   I830MemRange FrontBuffer;
-   I830MemRange FrontBuffer2;
-   I830MemRange Scratch;
-   I830MemRange Scratch2;
-
-   I830RingBuffer *LpRing;
-
-   I830MemRange BackBuffer;
-   I830MemRange DepthBuffer;
-   I830MemRange TexMem;
-   int TexGranularity;
-   I830MemRange ContextMem;
-   int drmMinor;
-   Bool have3DWindows;
-
-   Bool NeedRingBufferLow;
-   Bool allowPageFlip;
-   Bool disableTiling;
-
-   int Chipset;
-   unsigned long LinearAddr;
-   unsigned long MMIOAddr;
-
-   drmSize           registerSize;     /**< \brief MMIO register map size */
-   drm_handle_t         registerHandle;   /**< \brief MMIO register map handle */
-  //   IOADDRESS ioBase;
-   int               irq;              /**< \brief IRQ number */
-   int GttBound;
-
-   drm_handle_t ring_map;
-   unsigned int Fence[8];
-
-} I830Rec;
-
-/*
- * 12288 is set as the maximum, chosen because it is enough for
- * 1920x1440@32bpp with a 2048 pixel line pitch with some to spare.
- */
-#define I830_MAXIMUM_VBIOS_MEM		12288
-#define I830_DEFAULT_VIDEOMEM_2D	(MB(32) / 1024)
-#define I830_DEFAULT_VIDEOMEM_3D	(MB(64) / 1024)
-
-/* Flags for memory allocation function */
-#define FROM_ANYWHERE			0x00000000
-#define FROM_POOL_ONLY			0x00000001
-#define FROM_NEW_ONLY			0x00000002
-#define FROM_MASK			0x0000000f
-
-#define ALLOCATE_AT_TOP			0x00000010
-#define ALLOCATE_AT_BOTTOM		0x00000020
-#define FORCE_GAPS			0x00000040
-
-#define NEED_PHYSICAL_ADDR		0x00000100
-#define ALIGN_BOTH_ENDS			0x00000200
-#define FORCE_LOW			0x00000400
-
-#define ALLOC_NO_TILING			0x00001000
-#define ALLOC_INITIAL			0x00002000
-
-#define ALLOCATE_DRY_RUN		0x80000000
-
-/* Chipset registers for VIDEO BIOS memory RW access */
-#define _855_DRAM_RW_CONTROL 0x58
-#define _845_DRAM_RW_CONTROL 0x90
-#define DRAM_WRITE    0x33330000
-
-#define KB(x) ((x) * 1024)
-#define MB(x) ((x) * KB(1024))
-
-#define GTT_PAGE_SIZE			KB(4)
-#define ROUND_TO(x, y)			(((x) + (y) - 1) / (y) * (y))
-#define ROUND_DOWN_TO(x, y)		((x) / (y) * (y))
-#define ROUND_TO_PAGE(x)		ROUND_TO((x), GTT_PAGE_SIZE)
-#define ROUND_TO_MB(x)			ROUND_TO((x), MB(1))
-#define PRIMARY_RINGBUFFER_SIZE		KB(128)
-
-
-/* Ring buffer registers, p277, overview p19
- */
-#define LP_RING     0x2030
-#define HP_RING     0x2040
-
-#define RING_TAIL      0x00
-#define TAIL_ADDR           0x000FFFF8
-#define I830_TAIL_MASK	    0x001FFFF8
-
-#define RING_HEAD      0x04
-#define HEAD_WRAP_COUNT     0xFFE00000
-#define HEAD_WRAP_ONE       0x00200000
-#define HEAD_ADDR           0x001FFFFC
-#define I830_HEAD_MASK      0x001FFFFC
-
-#define RING_START     0x08
-#define START_ADDR          0x03FFFFF8
-#define I830_RING_START_MASK	0xFFFFF000
-
-#define RING_LEN       0x0C
-#define RING_NR_PAGES       0x001FF000 
-#define I830_RING_NR_PAGES	0x001FF000
-#define RING_REPORT_MASK    0x00000006
-#define RING_REPORT_64K     0x00000002
-#define RING_REPORT_128K    0x00000004
-#define RING_NO_REPORT      0x00000000
-#define RING_VALID_MASK     0x00000001
-#define RING_VALID          0x00000001
-#define RING_INVALID        0x00000000
-
-
-/* Fence/Tiling ranges [0..7]
- */
-#define FENCE            0x2000
-#define FENCE_NR         8
-
-#define I915G_FENCE_START_MASK	0x0ff00000
-
-#define I830_FENCE_START_MASK	0x07f80000
-
-#define FENCE_START_MASK    0x03F80000
-#define FENCE_X_MAJOR       0x00000000
-#define FENCE_Y_MAJOR       0x00001000
-#define FENCE_SIZE_MASK     0x00000700
-#define FENCE_SIZE_512K     0x00000000
-#define FENCE_SIZE_1M       0x00000100
-#define FENCE_SIZE_2M       0x00000200
-#define FENCE_SIZE_4M       0x00000300
-#define FENCE_SIZE_8M       0x00000400
-#define FENCE_SIZE_16M      0x00000500
-#define FENCE_SIZE_32M      0x00000600
-#define FENCE_SIZE_64M	    0x00000700
-#define I915G_FENCE_SIZE_1M       0x00000000
-#define I915G_FENCE_SIZE_2M       0x00000100
-#define I915G_FENCE_SIZE_4M       0x00000200
-#define I915G_FENCE_SIZE_8M       0x00000300
-#define I915G_FENCE_SIZE_16M      0x00000400
-#define I915G_FENCE_SIZE_32M      0x00000500
-#define I915G_FENCE_SIZE_64M	0x00000600
-#define I915G_FENCE_SIZE_128M	0x00000700
-#define FENCE_PITCH_1       0x00000000
-#define FENCE_PITCH_2       0x00000010
-#define FENCE_PITCH_4       0x00000020
-#define FENCE_PITCH_8       0x00000030
-#define FENCE_PITCH_16      0x00000040
-#define FENCE_PITCH_32      0x00000050
-#define FENCE_PITCH_64	    0x00000060
-#define FENCE_VALID         0x00000001
-
-#include <mmio.h>
-
-#  define MMIO_IN8(base, offset) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset))
-#  define MMIO_IN32(base, offset) \
-	read_MMIO_LE32(base, offset)
-#  define MMIO_OUT8(base, offset, val) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset)) = (val)
-#  define MMIO_OUT32(base, offset, val) \
-	*(volatile unsigned int *)(void *)(((unsigned char*)(base)) + (offset)) = CPU_TO_LE32(val)
-
-
-				/* Memory mapped register access macros */
-#define INREG8(addr)        MMIO_IN8(MMIO, addr)
-#define INREG(addr)         MMIO_IN32(MMIO, addr)
-#define OUTREG8(addr, val)  MMIO_OUT8(MMIO, addr, val)
-#define OUTREG(addr, val)   MMIO_OUT32(MMIO, addr, val)
-
-#define DSPABASE		0x70184
-
-#endif
diff --git a/src/mesa/drivers/dri/mach64/mach64_ioctl.h b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
index 1ffda1932f..9145ee6e6c 100644
--- a/src/mesa/drivers/dri/mach64/mach64_ioctl.h
+++ b/src/mesa/drivers/dri/mach64/mach64_ioctl.h
@@ -32,6 +32,9 @@
 #ifndef __MACH64_IOCTL_H__
 #define __MACH64_IOCTL_H__
 
+#include <stdio.h>
+#include <stdlib.h>
+
 #include "mach64_dri.h"
 #include "mach64_reg.h"
 #include "mach64_lock.h"
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fbo.c b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
index 8be7edb150..bd1273beea 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
@@ -220,7 +220,7 @@ get_tex_format(struct gl_texture_image *ti)
 	case MESA_FORMAT_RGB565:
 		return GL_RGB5;
 	default:
-		assert(0);
+		return GL_NONE;
 	}
 }
 
@@ -231,7 +231,6 @@ nouveau_render_texture(GLcontext *ctx, struct gl_framebuffer *fb,
 	struct gl_renderbuffer *rb = att->Renderbuffer;
 	struct gl_texture_image *ti =
 		att->Texture->Image[att->CubeMapFace][att->TextureLevel];
-	int ret;
 
 	/* Allocate a renderbuffer object for the texture if we
 	 * haven't already done so. */
@@ -244,9 +243,7 @@ nouveau_render_texture(GLcontext *ctx, struct gl_framebuffer *fb,
 	}
 
 	/* Update the renderbuffer fields from the texture. */
-	ret = set_renderbuffer_format(rb, get_tex_format(ti));
-	assert(ret);
-
+	set_renderbuffer_format(rb, get_tex_format(ti));
 	rb->Width = ti->Width;
 	rb->Height = ti->Height;
 	nouveau_surface_ref(&to_nouveau_teximage(ti)->surface,
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_texture.c b/src/mesa/drivers/dri/nouveau/nouveau_texture.c
index dbf9a5cc61..442f4e899e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_texture.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_texture.c
@@ -38,6 +38,7 @@
 #include "main/mipmap.h"
 #include "main/texfetch.h"
 #include "main/teximage.h"
+#include "drivers/common/meta.h"
 
 static struct gl_texture_object *
 nouveau_texture_new(GLcontext *ctx, GLuint name, GLenum target)
@@ -182,10 +183,10 @@ teximage_fits(struct gl_texture_object *t, int level)
 	struct nouveau_surface *s = &to_nouveau_texture(t)->surfaces[level];
 	struct gl_texture_image *ti = t->Image[0][level];
 
-	return ti && (t->Target == GL_TEXTURE_RECTANGLE ||
-		      (s->bo && s->width == ti->Width &&
-		       s->height == ti->Height &&
-		       s->format == ti->TexFormat));
+	return ti && to_nouveau_teximage(ti)->surface.bo &&
+		(t->Target == GL_TEXTURE_RECTANGLE ||
+		 (s->bo && s->format == ti->TexFormat &&
+		  s->width == ti->Width && s->height == ti->Height));
 }
 
 static GLboolean
@@ -589,6 +590,53 @@ nouveau_texture_unmap(GLcontext *ctx, struct gl_texture_object *t)
 	}
 }
 
+static void
+store_mipmap(GLcontext *ctx, GLenum target, int first, int last,
+	     struct gl_texture_object *t)
+{
+	struct gl_pixelstore_attrib packing = {
+		.BufferObj = ctx->Shared->NullBufferObj,
+		.Alignment = 1
+	};
+	GLenum format = t->Image[0][first]->TexFormat;
+	unsigned base_format, type, comps;
+	int i;
+
+	base_format = _mesa_get_format_base_format(format);
+	_mesa_format_to_type_and_comps(format, &type, &comps);
+
+	for (i = first; i <= last; i++) {
+		struct gl_texture_image *ti = t->Image[0][i];
+		void *data = ti->Data;
+
+		nouveau_teximage(ctx, 3, target, i, ti->InternalFormat,
+				 ti->Width, ti->Height, ti->Depth,
+				 ti->Border, base_format, type, data,
+				 &packing, t, ti);
+
+		_mesa_free_texmemory(data);
+	}
+}
+
+static void
+nouveau_generate_mipmap(GLcontext *ctx, GLenum target,
+			struct gl_texture_object *t)
+{
+	if (_mesa_meta_check_generate_mipmap_fallback(ctx, target, t)) {
+		struct gl_texture_image *base = t->Image[0][t->BaseLevel];
+
+		nouveau_teximage_map(ctx, base);
+		_mesa_generate_mipmap(ctx, target, t);
+		nouveau_teximage_unmap(ctx, base);
+
+		store_mipmap(ctx, target, t->BaseLevel + 1,
+			     get_last_level(t), t);
+
+	} else {
+		_mesa_meta_GenerateMipmap(ctx, target, t);
+	}
+}
+
 void
 nouveau_texture_functions_init(struct dd_function_table *functions)
 {
@@ -607,4 +655,5 @@ nouveau_texture_functions_init(struct dd_function_table *functions)
 	functions->BindTexture = nouveau_bind_texture;
 	functions->MapTexture = nouveau_texture_map;
 	functions->UnmapTexture = nouveau_texture_unmap;
+	functions->GenerateMipmap = nouveau_generate_mipmap;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_fb.c b/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
index 21da4f7af1..95691cad04 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state_fb.c
@@ -72,7 +72,7 @@ nv20_emit_framebuffer(GLcontext *ctx, int emit)
 			fb->_ColorDrawBuffers[0])->surface;
 
 		rt_format |= get_rt_format(s->format);
-		zeta_pitch = rt_pitch = s->pitch;
+		rt_pitch = s->pitch;
 
 		nouveau_bo_markl(bctx, kelvin, NV20TCL_COLOR_OFFSET,
 				 s->bo, 0, bo_flags);
@@ -88,6 +88,9 @@ nv20_emit_framebuffer(GLcontext *ctx, int emit)
 
 		nouveau_bo_markl(bctx, kelvin, NV20TCL_ZETA_OFFSET,
 				 s->bo, 0, bo_flags);
+	} else {
+		rt_format |= get_rt_format(MESA_FORMAT_Z24_S8);
+		zeta_pitch = rt_pitch;
 	}
 
 	BEGIN_RING(chan, kelvin, NV20TCL_RT_FORMAT, 2);
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
index e46118e4fc..2d45513bb4 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state_tex.c
@@ -194,7 +194,8 @@ nv20_emit_tex_obj(GLcontext *ctx, int emit)
 		| nvgl_wrap_mode(t->WrapS) << 0;
 
 	tx_filter = nvgl_filter_mode(t->MagFilter) << 24
-		| nvgl_filter_mode(t->MinFilter) << 16;
+		| nvgl_filter_mode(t->MinFilter) << 16
+		| 2 << 12;
 
 	tx_enable = NV20TCL_TX_ENABLE_ENABLE
 		| log2i(t->MaxAnisotropy) << 4;
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index 262fe3cdde..dbf4ad477d 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -612,6 +612,8 @@ static void r200RasterPrimitive( GLcontext *ctx, GLuint hwprim )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
 
+   radeon_prepare_render(&rmesa->radeon);
+
    if (rmesa->radeon.swtcl.hw_primitive != hwprim) {
       /* need to disable perspective-correct texturing for point sprites */
       if ((hwprim & 0xf) == R200_VF_PRIM_POINT_SPRITES && ctx->Point.PointSprite) {
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index d43e14581e..4ae0f30491 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -264,6 +264,8 @@ void r200TclPrimitive( GLcontext *ctx,
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
    GLuint newprim = hw_prim | R200_VF_TCL_OUTPUT_VTX_ENABLE;
 
+   radeon_prepare_render(&rmesa->radeon);
+
    if (newprim != rmesa->tcl.hw_primitive ||
        !discrete_prim[hw_prim&0xf]) {
       /* need to disable perspective-correct texturing for point sprites */
diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
index a326ee4c4f..c6246a81a2 100644
--- a/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r3xx_fragprog.c
@@ -109,13 +109,13 @@ void r3xx_compile_fragment_program(struct r300_fragment_program_compiler* c)
 	debug_program_log(c, "before compilation");
 
 	if (c->Base.is_r500){
-		r500_transform_unroll_loops(&c->Base, &loop_state);	
-		debug_program_log(c, "after r500 transform loops");
+		rc_unroll_loops(&c->Base, R500_PFS_MAX_INST);
+		debug_program_log(c, "after unroll loops");
 	}
 	else{
-		rc_transform_unroll_loops(&c->Base, &loop_state);
+		rc_transform_loops(&c->Base, &loop_state);
 		debug_program_log(c, "after transform loops");
-		
+
 		rc_emulate_branches(&c->Base);
 		debug_program_log(c, "after emulate branches");
 	}
diff --git a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
index d347b4df9c..e940fedec2 100644
--- a/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r3xx_vertprog.c
@@ -633,7 +633,7 @@ static struct rc_swizzle_caps r300_vertprog_swizzle_caps = {
 void r3xx_compile_vertex_program(struct r300_vertex_program_compiler* compiler)
 {
 	struct emulate_loop_state loop_state;
-	
+
 	compiler->Base.SwizzleCaps = &r300_vertprog_swizzle_caps;
 
 	addArtificialOutputs(compiler);
@@ -643,13 +643,12 @@ void r3xx_compile_vertex_program(struct r300_vertex_program_compiler* compiler)
 	/* XXX Ideally this should be done only for r3xx, but since
 	 * we don't have branching support for r5xx, we use the emulation
 	 * on all chipsets. */
-	rc_transform_unroll_loops(&compiler->Base, &loop_state);
-	
-	debug_program_log(compiler, "after transform loops");
-	
+
 	if (compiler->Base.is_r500){
+		rc_transform_loops(&compiler->Base, &loop_state);
 		rc_emulate_loops(&loop_state, R500_VS_MAX_ALU);
 	} else {
+		rc_transform_loops(&compiler->Base, &loop_state);
 		rc_emulate_loops(&loop_state, R300_VS_MAX_ALU);
 	}
 	debug_program_log(compiler, "after emulate loops");
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
index e6b5522c5b..80a120497e 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.c
@@ -30,7 +30,6 @@
 #include <stdio.h>
 
 #include "../r300_reg.h"
-#include "radeon_emulate_loops.h"
 
 /**
  * Rewrite IF instructions to use the ALU result special register.
@@ -60,31 +59,6 @@ int r500_transform_IF(
 	return 1;
 }
 
-/**
- * Rewrite loops to make them easier to emit.  This is not a local
- * transformation, because it modifies and reorders an entire block of code.
- */
-void r500_transform_unroll_loops(struct radeon_compiler * c,
-						struct emulate_loop_state *s)
-{
-	int i;
-	
-	rc_transform_unroll_loops(c, s);
-	
-	for( i = s->LoopCount - 1; i >= 0; i-- ){
-		struct rc_instruction * inst_continue;
-		if(!s->Loops[i].EndLoop){
-			continue;
-		}
-		/* Insert a continue instruction at the end of the loop.  This
-		 * is required in order to emit loops correctly. */
-		inst_continue = rc_insert_new_instruction(c,
-						s->Loops[i].EndIf->Prev);
-		inst_continue->U.I.Opcode = RC_OPCODE_CONTINUE;
-	}
-
-}
-
 static int r500_swizzle_is_native(rc_opcode opcode, struct rc_src_register reg)
 {
 	unsigned int relevant;
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
index 0d005a794f..34173351f8 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog.h
@@ -49,6 +49,4 @@ extern int r500_transform_IF(
 	struct rc_instruction * inst,
 	void* data);
 
-void r500_transform_unroll_loops(struct radeon_compiler * c,
-						struct emulate_loop_state * s);
 #endif
diff --git a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
index 0bd8f0a239..c3f817ad4e 100644
--- a/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
+++ b/src/mesa/drivers/dri/r300/compiler/r500_fragprog_emit.c
@@ -64,7 +64,12 @@ struct branch_info {
 };
 
 struct loop_info {
-	int LoopStart;
+	int BgnLoop;
+
+	int BranchDepth;
+	int * Brks;
+	int BrkCount;
+	int BrkReserved;
 };
 
 struct emit_state {
@@ -368,6 +373,12 @@ static void emit_flowcontrol(struct emit_state * s, struct rc_instruction * inst
 
 	unsigned int newip = ++s->Code->inst_end;
 
+	/* Currently all loops use the same integer constant to intialize
+	 * the loop variables. */
+	if(!s->Code->int_constants[0]) {
+		s->Code->int_constants[0] = R500_FC_INT_CONST_KR(0xff);
+		s->Code->int_constant_count = 1;
+	}
 	s->Code->inst[newip].inst0 = R500_INST_TYPE_FC | R500_INST_ALU_WAIT;
 
 	switch(inst->U.I.Opcode){
@@ -378,32 +389,69 @@ static void emit_flowcontrol(struct emit_state * s, struct rc_instruction * inst
 			s->Loops, s->CurrentLoopDepth, s->LoopsReserved, 1);
 
 		loop = &s->Loops[s->CurrentLoopDepth++];
-		
-		/* We don't emit an instruction for BGNLOOP, so we need to
-		 * decrement the instruction counter, but first we need to
-		 * set LoopStart to the current value of inst_end, which
-		 * will end up being the first real instruction in the loop.*/
-		loop->LoopStart = s->Code->inst_end--;
+		memset(loop, 0, sizeof(struct loop_info));
+		loop->BranchDepth = s->CurrentBranchDepth;
+		loop->BgnLoop = newip;
+
+		s->Code->inst[newip].inst2 = R500_FC_OP_LOOP
+			| R500_FC_JUMP_FUNC(0x00)
+			| R500_FC_IGNORE_UNCOVERED
+			;
 		break;
-	
 	case RC_OPCODE_BRK:
-		/* Don't emit an instruction for BRK */
-		s->Code->inst_end--;
+		loop = &s->Loops[s->CurrentLoopDepth - 1];
+		memory_pool_array_reserve(&s->C->Pool, int, loop->Brks,
+					loop->BrkCount, loop->BrkReserved, 1);
+
+		loop->Brks[loop->BrkCount++] = newip;
+		s->Code->inst[newip].inst2 = R500_FC_OP_BREAKLOOP
+			| R500_FC_JUMP_FUNC(0xff)
+			| R500_FC_B_OP1_DECR
+			| R500_FC_B_POP_CNT(
+				s->CurrentBranchDepth - loop->BranchDepth)
+			| R500_FC_IGNORE_UNCOVERED
+			;
 		break;
 
 	case RC_OPCODE_CONTINUE:
 		loop = &s->Loops[s->CurrentLoopDepth - 1];
-		s->Code->inst[newip].inst2 = R500_FC_OP_JUMP |
-			R500_FC_JUMP_FUNC(0xff);
-		s->Code->inst[newip].inst3 = R500_FC_JUMP_ADDR(loop->LoopStart);
+		s->Code->inst[newip].inst2 = R500_FC_OP_JUMP
+			| R500_FC_JUMP_FUNC(0xff)
+			| R500_FC_B_OP1_DECR
+			| R500_FC_B_POP_CNT(
+				s->CurrentBranchDepth -	loop->BranchDepth)
+			;
+		s->Code->inst[newip].inst3 = R500_FC_JUMP_ADDR(loop->BgnLoop);
 		break;
 
 	case RC_OPCODE_ENDLOOP:
-		/* Don't emit an instruction for ENDLOOP */
-		s->Code->inst_end--;
+	{
+		unsigned int i;
+		loop = &s->Loops[s->CurrentLoopDepth - 1];
+		/* Emit ENDLOOP */
+		s->Code->inst[newip].inst2 = R500_FC_OP_ENDLOOP
+			| R500_FC_JUMP_FUNC(0xff)
+			| R500_FC_JUMP_ANY
+			| R500_FC_IGNORE_UNCOVERED
+			;
+		/* The constant integer at index 0 is used by all loops. */
+		s->Code->inst[newip].inst3 = R500_FC_INT_ADDR(0)
+			| R500_FC_JUMP_ADDR(loop->BgnLoop + 1)
+			;
+
+		/* Set jump address and int constant for BGNLOOP */
+		s->Code->inst[loop->BgnLoop].inst3 = R500_FC_INT_ADDR(0)
+			| R500_FC_JUMP_ADDR(newip)
+			;
+
+		/* Set jump address for the BRK instructions. */
+		while(loop->BrkCount--) {
+			s->Code->inst[loop->Brks[loop->BrkCount]].inst3 =
+						R500_FC_JUMP_ADDR(newip + 1);
+		}
 		s->CurrentLoopDepth--;
 		break;
-
+	}
 	case RC_OPCODE_IF:
 		if ( s->CurrentBranchDepth >= MAX_BRANCH_DEPTH_FULL) {
 			rc_error(s->C, "Branch depth exceeds hardware limit");
@@ -442,24 +490,16 @@ static void emit_flowcontrol(struct emit_state * s, struct rc_instruction * inst
 		}
 
 		branch = &s->Branches[s->CurrentBranchDepth - 1];
-		
-		if(inst->Prev->U.I.Opcode == RC_OPCODE_BRK){
-			branch->Endif = --s->Code->inst_end;
-			s->Code->inst[branch->Endif].inst2 |=
-				R500_FC_B_OP0_DECR;
-		}
-		else{
-			branch->Endif = newip;
-		
-			s->Code->inst[branch->Endif].inst2 = R500_FC_OP_JUMP
-				| R500_FC_A_OP_NONE /* no address stack */
-				| R500_FC_JUMP_ANY /* docs says set this, but I don't understand why */
-				| R500_FC_B_OP0_DECR /* decrement branch counter if stay */
-				| R500_FC_B_OP1_NONE /* no branch counter if stay */
-				| R500_FC_B_POP_CNT(1)
+		branch->Endif = newip;
+
+		s->Code->inst[branch->Endif].inst2 = R500_FC_OP_JUMP
+			| R500_FC_A_OP_NONE /* no address stack */
+			| R500_FC_JUMP_ANY /* docs says set this, but I don't understand why */
+			| R500_FC_B_OP0_DECR /* decrement branch counter if stay */
+			| R500_FC_B_OP1_NONE /* no branch counter if stay */
+			| R500_FC_B_POP_CNT(1)
 			;
-			s->Code->inst[branch->Endif].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
-		}
+		s->Code->inst[branch->Endif].inst3 = R500_FC_JUMP_ADDR(branch->Endif + 1);
 		s->Code->inst[branch->If].inst2 = R500_FC_OP_JUMP
 			| R500_FC_A_OP_NONE /* no address stack */
 			| R500_FC_JUMP_FUNC(0x0f) /* jump if ALU result is false */
@@ -544,11 +584,9 @@ void r500BuildFragmentProgramHwCode(struct r300_fragment_program_compiler *compi
 		code->inst[ip].inst0 = R500_INST_TYPE_OUT | R500_INST_TEX_SEM_WAIT;
 	}
 
-	/* Use FULL flow control mode if branches are nested deep enough.
-	 * We don not need to enable FULL flow control mode for loops, becasue
-	 * we aren't using the hardware loop instructions.
-	 */
-	if (s.MaxBranchDepth >= 4) {
+	/* Enable full flow control mode if we are using loops or have if
+	 * statements nested at least four deep. */
+	if (s.MaxBranchDepth >= 4 || s.LoopsReserved > 0) {
 		if (code->max_temp_idx < 1)
 			code->max_temp_idx = 1;
 
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_code.h b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
index d03689763b..e14a3520dd 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_code.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_code.h
@@ -221,6 +221,9 @@ struct r500_fragment_program_code {
 	int max_temp_idx;
 
 	uint32_t us_fc_ctrl;
+
+	uint32_t int_constants[32];
+	uint32_t int_constant_count;
 };
 
 struct rX00_fragment_program_code {
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c
index fbb4235c22..31566a937f 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_dataflow_deadcode.c
@@ -43,6 +43,12 @@ struct instruction_state {
 	unsigned char SrcReg[3];
 };
 
+struct loopinfo {
+	struct updatemask_state * Breaks;
+	unsigned int BreakCount;
+	unsigned int BreaksReserved;
+};
+
 struct branchinfo {
 	unsigned int HaveElse:1;
 
@@ -59,6 +65,10 @@ struct deadcode_state {
 	struct branchinfo * BranchStack;
 	unsigned int BranchStackSize;
 	unsigned int BranchStackReserved;
+
+	struct loopinfo * LoopStack;
+	unsigned int LoopStackSize;
+	unsigned int LoopStackReserved;
 };
 
 
@@ -78,6 +88,22 @@ static void or_updatemasks(
 	dst->Address = a->Address | b->Address;
 }
 
+static void push_break(struct deadcode_state *s)
+{
+	struct loopinfo * loop = &s->LoopStack[s->LoopStackSize - 1];
+	memory_pool_array_reserve(&s->C->Pool, struct updatemask_state,
+		loop->Breaks, loop->BreakCount, loop->BreaksReserved, 1);
+
+	memcpy(&loop->Breaks[loop->BreakCount++], &s->R, sizeof(s->R));
+}
+
+static void push_loop(struct deadcode_state * s)
+{
+	memory_pool_array_reserve(&s->C->Pool, struct loopinfo, s->LoopStack,
+			s->LoopStackSize, s->LoopStackReserved, 1);
+	memset(&s->LoopStack[s->LoopStackSize++], 0, sizeof(struct loopinfo));
+}
+
 static void push_branch(struct deadcode_state * s)
 {
 	memory_pool_array_reserve(&s->C->Pool, struct branchinfo, s->BranchStack,
@@ -233,11 +259,22 @@ void rc_dataflow_deadcode(struct radeon_compiler * c, rc_dataflow_mark_outputs_f
 					}
 				}
 			}
+			push_loop(&s);
 			break;
 		}
-		case RC_OPCODE_CONTINUE:
 		case RC_OPCODE_BRK:
+			push_break(&s);
+			break;
 		case RC_OPCODE_BGNLOOP:
+		{
+			unsigned int i;
+			struct loopinfo * loop = &s.LoopStack[s.LoopStackSize-1];
+			for(i = 0; i < loop->BreakCount; i++) {
+				or_updatemasks(&s.R, &s.R, &loop->Breaks[i]);
+			}
+			break;
+		}
+		case RC_OPCODE_CONTINUE:
 			break;
 		case RC_OPCODE_ENDIF:
 			push_branch(&s);
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.c b/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.c
index 131e9e7436..24c3ae57b6 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.c
@@ -39,7 +39,6 @@
 #define DBG(...) do { if (VERBOSE) fprintf(stderr, __VA_ARGS__); } while(0)
 
 struct const_value {
-	
 	struct radeon_compiler * C;
 	struct rc_src_register * Src;
 	float Value;
@@ -78,17 +77,17 @@ static int src_reg_is_immediate(struct rc_src_register * src,
 	c->Program.Constants.Constants[src->Index].Type==RC_CONSTANT_IMMEDIATE;
 }
 
-static unsigned int loop_calc_iterations(struct emulate_loop_state *s, 
-			struct loop_info * loop, unsigned int max_instructions)
+static unsigned int loop_max_possible_iterations(struct radeon_compiler *c,
+			struct loop_info * loop, unsigned int prog_inst_limit)
 {
-	unsigned int total_i = rc_recompute_ips(s->C);
+	unsigned int total_i = rc_recompute_ips(c);
 	unsigned int loop_i = (loop->EndLoop->IP - loop->BeginLoop->IP) - 1;
 	/* +1 because the program already has one iteration of the loop. */
-	return 1 + ((max_instructions - total_i) / (s->LoopCount * loop_i));
+	return 1 + ((prog_inst_limit - total_i) / loop_i);
 }
 
-static void loop_unroll(struct emulate_loop_state * s,
-			struct loop_info *loop, unsigned int iterations)
+static void unroll_loop(struct radeon_compiler * c, struct loop_info * loop,
+						unsigned int iterations)
 {
 	unsigned int i;
 	struct rc_instruction * ptr;
@@ -99,7 +98,7 @@ static void loop_unroll(struct emulate_loop_state * s,
 	rc_remove_instruction(loop->EndLoop);
 	for( i = 1; i < iterations; i++){
 		for(ptr = first; ptr != last->Next; ptr = ptr->Next){
-			struct rc_instruction *new = rc_alloc_instruction(s->C);
+			struct rc_instruction *new = rc_alloc_instruction(c);
 			memcpy(new, ptr, sizeof(struct rc_instruction));
 			rc_insert_instruction(append_to, new);
 			append_to = new;
@@ -115,7 +114,7 @@ static void update_const_value(void * data, struct rc_instruction * inst,
 	if(value->Src->File != file ||
 	   value->Src->Index != index ||
 	   !(1 << GET_SWZ(value->Src->Swizzle, 0) & mask)){
-	   	return;
+		return;
 	}
 	switch(inst->U.I.Opcode){
 	case RC_OPCODE_MOV:
@@ -140,7 +139,7 @@ static void get_incr_amount(void * data, struct rc_instruction * inst,
 	if(file != RC_FILE_TEMPORARY ||
 	   count_inst->Index != index ||
 	   (1 << GET_SWZ(count_inst->Swz,0) != mask)){
-	   	return;
+		return;
 	}
 	/* Find the index of the counter register. */
 	opcode = rc_get_opcode_info(inst->U.I.Opcode);
@@ -185,13 +184,16 @@ static void get_incr_amount(void * data, struct rc_instruction * inst,
 		count_inst->Unknown = 1;
 		return;
 	}
-	
 }
 
-static int transform_const_loop(struct emulate_loop_state * s,
-						struct loop_info * loop)
+/**
+ * If prog_inst_limit is -1, then all eligible loops will be unrolled regardless
+ * of how many iterations they have.
+ */
+static int try_unroll_loop(struct radeon_compiler * c, struct loop_info * loop,
+						unsigned int prog_inst_limit)
 {
-	int end_loops = 1;
+	int end_loops;
 	int iterations;
 	struct count_inst count_inst;
 	float limit_value;
@@ -201,12 +203,12 @@ static int transform_const_loop(struct emulate_loop_state * s,
 	struct rc_instruction * inst;
 
 	/* Find the counter and the upper limit */
-	
-	if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[0], s->C)){
+
+	if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[0], c)){
 		limit = &loop->Cond->U.I.SrcReg[0];
 		counter = &loop->Cond->U.I.SrcReg[1];
 	}
-	else if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[1], s->C)){
+	else if(src_reg_is_immediate(&loop->Cond->U.I.SrcReg[1], c)){
 		limit = &loop->Cond->U.I.SrcReg[1];
 		counter = &loop->Cond->U.I.SrcReg[0];
 	}
@@ -214,13 +216,13 @@ static int transform_const_loop(struct emulate_loop_state * s,
 		DBG("No constant limit.\n");
 		return 0;
 	}
-	
+
 	/* Find the initial value of the counter */
 	counter_value.Src = counter;
 	counter_value.Value = 0.0f;
 	counter_value.HasValue = 0;
-	counter_value.C = s->C;
-	for(inst = s->C->Program.Instructions.Next; inst != loop->BeginLoop;
+	counter_value.C = c;
+	for(inst = c->Program.Instructions.Next; inst != loop->BeginLoop;
 							inst = inst->Next){
 		rc_for_all_writes_mask(inst, update_const_value, &counter_value);
 	}
@@ -230,11 +232,12 @@ static int transform_const_loop(struct emulate_loop_state * s,
 	}
 	DBG("Initial counter value is %f\n", counter_value.Value);
 	/* Determine how the counter is modified each loop */
-	count_inst.C = s->C;
+	count_inst.C = c;
 	count_inst.Index = counter->Index;
 	count_inst.Swz = counter->Swizzle;
 	count_inst.Amount = 0.0f;
 	count_inst.Unknown = 0;
+	end_loops = 1;
 	for(inst = loop->BeginLoop->Next; end_loops > 0; inst = inst->Next){
 		switch(inst->U.I.Opcode){
 		/* XXX In the future we might want to try to unroll nested
@@ -246,6 +249,16 @@ static int transform_const_loop(struct emulate_loop_state * s,
 			loop->EndLoop = inst;
 			end_loops--;
 			break;
+		case RC_OPCODE_BRK:
+			/* Don't unroll loops if it has a BRK instruction
+			 * other one used when testing the main conditional
+			 * of the loop. */
+
+			/* Make sure we haven't entered a nested loops. */
+			if(inst != loop->Brk && end_loops == 1) {
+				return 0;
+			}
+			break;
 		/* XXX Check if the counter is modified within an if statement.
 		 */
 		case RC_OPCODE_IF:
@@ -266,17 +279,20 @@ static int transform_const_loop(struct emulate_loop_state * s,
 	/* Calculate the number of iterations of this loop.  Keeping this
 	 * simple, since we only support increment and decrement loops.
 	 */
-	limit_value = get_constant_value(s->C, limit, 0);
+	limit_value = get_constant_value(c, limit, 0);
 	DBG("Limit is %f.\n", limit_value);
+	/* The iteration calculations are opposite of what you would expect.
+	 * In a normal loop, if the condition is met, then loop continues, but
+	 * with our loops, if the condition is met, the is exited. */
 	switch(loop->Cond->U.I.Opcode){
-	case RC_OPCODE_SGT:
-	case RC_OPCODE_SLT:
+	case RC_OPCODE_SGE:
+	case RC_OPCODE_SLE:
 		iterations = (int) ceilf((limit_value - counter_value.Value) /
 							count_inst.Amount);
 		break;
 
-	case RC_OPCODE_SLE:
-	case RC_OPCODE_SGE:
+	case RC_OPCODE_SGT:
+	case RC_OPCODE_SLT:
 		iterations = (int) floorf((limit_value - counter_value.Value) /
 							count_inst.Amount) + 1;
 		break;
@@ -284,77 +300,85 @@ static int transform_const_loop(struct emulate_loop_state * s,
 		return 0;
 	}
 
+	if (prog_inst_limit > 0
+		&& iterations > loop_max_possible_iterations(c, loop,
+							prog_inst_limit)) {
+		return 0;
+	}
+
 	DBG("Loop will have %d iterations.\n", iterations);
-	
+
 	/* Prepare loop for unrolling */
 	rc_remove_instruction(loop->Cond);
 	rc_remove_instruction(loop->If);
 	rc_remove_instruction(loop->Brk);
 	rc_remove_instruction(loop->EndIf);
-	
-	loop_unroll(s, loop, iterations);
+
+	unroll_loop(c, loop, iterations);
 	loop->EndLoop = NULL;
 	return 1;
 }
 
-/** 
- * This function prepares a loop to be unrolled by converting it into an if
- * statement.  Here is an outline of the conversion process:
- * BGNLOOP;                         	-> BGNLOOP;
- * <Additional conditional code>	-> <Additional conditional code>
- * SGE/SLT temp[0], temp[1], temp[2];	-> SLT/SGE temp[0], temp[1], temp[2];
- * IF temp[0];                      	-> IF temp[0];
- * BRK;                             	->
- * ENDIF;                           	-> <Loop Body>
- * <Loop Body>                      	-> ENDIF;
- * ENDLOOP;                         	-> ENDLOOP
- *
+/**
+ * @param c
+ * @param loop
  * @param inst A pointer to a BGNLOOP instruction.
- * @return If the loop can be unrolled, a pointer to the first instruction of
- * 		the unrolled loop.
- * 	   Otherwise, A pointer to the ENDLOOP instruction.
- * 	   Null if there is an error.
+ * @return 1 if all of the members of loop where set.
+ * @return 0 if there was an error and some members of loop are still NULL.
  */
-static struct rc_instruction * transform_loop(struct emulate_loop_state * s,
+static int build_loop_info(struct radeon_compiler * c, struct loop_info * loop,
 						struct rc_instruction * inst)
 {
-	struct loop_info *loop;
 	struct rc_instruction * ptr;
 
-	memory_pool_array_reserve(&s->C->Pool, struct loop_info,
-			s->Loops, s->LoopCount, s->LoopReserved, 1);
-
-	loop = &s->Loops[s->LoopCount++];
-	memset(loop, 0, sizeof(struct loop_info));
 	if(inst->U.I.Opcode != RC_OPCODE_BGNLOOP){
-		rc_error(s->C, "expected BGNLOOP\n", __FUNCTION__);
-		return NULL;
+		rc_error(c, "%s: expected BGNLOOP", __FUNCTION__);
+		return 0;
 	}
+
+	memset(loop, 0, sizeof(struct loop_info));
+
 	loop->BeginLoop = inst;
 
-	for(ptr = loop->BeginLoop->Next; !loop->EndLoop; ptr = ptr->Next){
+	for(ptr = loop->BeginLoop->Next; !loop->EndLoop; ptr = ptr->Next) {
+
+		if (ptr == &c->Program.Instructions) {
+			rc_error(c, "%s: BGNLOOP without an ENDLOOOP.\n",
+								__FUNCTION__);
+			return 0;
+		}
+
 		switch(ptr->U.I.Opcode){
 		case RC_OPCODE_BGNLOOP:
-			/* Nested loop */
-			ptr = transform_loop(s, ptr);
-			if(!ptr){
-				return NULL;
+		{
+			/* Nested loop, skip ahead to the end. */
+			unsigned int loop_depth = 1;
+			for(ptr = ptr->Next; ptr != &c->Program.Instructions;
+							ptr = ptr->Next){
+				if (ptr->U.I.Opcode == RC_OPCODE_BGNLOOP) {
+					loop_depth++;
+				} else if (ptr->U.I.Opcode == RC_OPCODE_ENDLOOP) {
+					if (!--loop_depth) {
+						break;
+					}
+				}
+			}
+			if (ptr == &c->Program.Instructions) {
+				rc_error(c, "%s: BGNLOOP without an ENDLOOOP\n",
+								__FUNCTION__);
+					return 0;
 			}
 			break;
+		}
 		case RC_OPCODE_BRK:
-			loop->Brk = ptr;
-			if(ptr->Next->U.I.Opcode != RC_OPCODE_ENDIF){
-				rc_error(s->C,
-					"%s: expected ENDIF\n",__FUNCTION__);
-				return NULL;
-			}
-			loop->EndIf = ptr->Next;
-			if(ptr->Prev->U.I.Opcode != RC_OPCODE_IF){
-				rc_error(s->C,
-					"%s: expected IF\n", __FUNCTION__);
-				return NULL;
+			if(ptr->Next->U.I.Opcode != RC_OPCODE_ENDIF
+					|| ptr->Prev->U.I.Opcode != RC_OPCODE_IF
+					|| loop->Brk){
+				continue;
 			}
+			loop->Brk = ptr;
 			loop->If = ptr->Prev;
+			loop->EndIf = ptr->Next;
 			switch(loop->If->Prev->U.I.Opcode){
 			case RC_OPCODE_SLT:
 			case RC_OPCODE_SGE:
@@ -364,18 +388,61 @@ static struct rc_instruction * transform_loop(struct emulate_loop_state * s,
 			case RC_OPCODE_SNE:
 				break;
 			default:
-				rc_error(s->C, "%s expected conditional\n",
+				rc_error(c, "%s: expected conditional",
 								__FUNCTION__);
-				return NULL;
+				return 0;
 			}
 			loop->Cond = loop->If->Prev;
-			ptr = loop->EndIf;
 			break;
+
 		case RC_OPCODE_ENDLOOP:
 			loop->EndLoop = ptr;
 			break;
 		}
 	}
+
+	if (loop->BeginLoop && loop->Brk && loop->If && loop->EndIf
+					&& loop->Cond && loop->EndLoop) {
+		return 1;
+	}
+	return 0;
+}
+
+/**
+ * This function prepares a loop to be unrolled by converting it into an if
+ * statement.  Here is an outline of the conversion process:
+ * BGNLOOP;                         	-> BGNLOOP;
+ * <Additional conditional code>	-> <Additional conditional code>
+ * SGE/SLT temp[0], temp[1], temp[2];	-> SLT/SGE temp[0], temp[1], temp[2];
+ * IF temp[0];                      	-> IF temp[0];
+ * BRK;                             	->
+ * ENDIF;                           	-> <Loop Body>
+ * <Loop Body>                      	-> ENDIF;
+ * ENDLOOP;                         	-> ENDLOOP
+ *
+ * @param inst A pointer to a BGNLOOP instruction.
+ * @return If the loop can be unrolled, a pointer to the first instruction of
+ * 		the unrolled loop.
+ * 	   Otherwise, A pointer to the ENDLOOP instruction.
+ * 	   Null if there is an error.
+ */
+static struct rc_instruction * transform_loop(struct emulate_loop_state * s,
+						struct rc_instruction * inst)
+{
+	struct loop_info * loop;
+
+	memory_pool_array_reserve(&s->C->Pool, struct loop_info,
+			s->Loops, s->LoopCount, s->LoopReserved, 1);
+
+	loop = &s->Loops[s->LoopCount++];
+
+	if (!build_loop_info(s->C, loop, inst))
+		return NULL;
+
+	if(try_unroll_loop(s->C, loop, -1)){
+		return loop->BeginLoop->Next;
+	}
+
 	/* Reverse the conditional instruction */
 	switch(loop->Cond->U.I.Opcode){
 	case RC_OPCODE_SGE:
@@ -400,24 +467,19 @@ static struct rc_instruction * transform_loop(struct emulate_loop_state * s,
 		rc_error(s->C, "loop->Cond is not a conditional.\n");
 		return NULL;
 	}
-	
-	/* Check if the number of loops is known at compile time. */
-	if(transform_const_loop(s, loop)){
-		return loop->BeginLoop->Next;
-	}
 
-	/* Prepare the loop to be unrolled */
+	/* Prepare the loop to be emulated */
 	rc_remove_instruction(loop->Brk);
 	rc_remove_instruction(loop->EndIf);
 	rc_insert_instruction(loop->EndLoop->Prev, loop->EndIf);
 	return loop->EndLoop;
 }
 
-void rc_transform_unroll_loops(struct radeon_compiler *c,
-					struct emulate_loop_state * s)
+void rc_transform_loops(struct radeon_compiler *c,
+						struct emulate_loop_state * s)
 {
 	struct rc_instruction * ptr;
-	
+
 	memset(s, 0, sizeof(struct emulate_loop_state));
 	s->C = c;
 	ptr = s->C->Program.Instructions.Next;
@@ -433,8 +495,23 @@ void rc_transform_unroll_loops(struct radeon_compiler *c,
 	}
 }
 
-void rc_emulate_loops(struct emulate_loop_state *s,
-						unsigned int max_instructions)
+void rc_unroll_loops(struct radeon_compiler *c, int prog_inst_limit)
+{
+	struct rc_instruction * inst;
+	struct loop_info loop;
+
+	for(inst = c->Program.Instructions.Next;
+			inst != &c->Program.Instructions; inst = inst->Next) {
+
+		if (inst->U.I.Opcode == RC_OPCODE_BGNLOOP) {
+			if (build_loop_info(c, &loop, inst)) {
+				try_unroll_loop(c, &loop, prog_inst_limit);
+			}
+		}
+	}
+}
+
+void rc_emulate_loops(struct emulate_loop_state *s, int prog_inst_limit)
 {
 	int i;
 	/* Iterate backwards of the list of loops so that loops that nested
@@ -444,8 +521,8 @@ void rc_emulate_loops(struct emulate_loop_state *s,
 		if(!s->Loops[i].EndLoop){
 			continue;
 		}
-		unsigned int iterations = loop_calc_iterations(s, &s->Loops[i],
-							max_instructions);
-		loop_unroll(s, &s->Loops[i], iterations);
+		unsigned int iterations = loop_max_possible_iterations(
+					s->C, &s->Loops[i], prog_inst_limit);
+		unroll_loop(s->C, &s->Loops[i], iterations);
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.h b/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.h
index 7748813c4e..86d91ef14b 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.h
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_emulate_loops.h
@@ -23,10 +23,11 @@ struct emulate_loop_state {
 	unsigned int LoopReserved;
 };
 
-void rc_transform_unroll_loops(struct radeon_compiler *c,
-					struct emulate_loop_state * s);
+void rc_transform_loops(struct radeon_compiler *c,
+						struct emulate_loop_state * s);
 
-void rc_emulate_loops(struct emulate_loop_state *s,
-					unsigned int max_instructions);
+void rc_unroll_loops(struct radeon_compiler * c, int prog_inst_limit);
+
+void rc_emulate_loops(struct emulate_loop_state * s, int prog_inst_limit);
 
 #endif /* RADEON_EMULATE_LOOPS_H */
diff --git a/src/mesa/drivers/dri/r300/compiler/radeon_optimize.c b/src/mesa/drivers/dri/r300/compiler/radeon_optimize.c
index eca0651536..7a3f35950a 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_optimize.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_optimize.c
@@ -164,7 +164,8 @@ static void peephole(struct radeon_compiler * c, struct rc_instruction * inst_mo
 	    inst = inst->Next) {
 		/* XXX In the future we might be able to make the optimizer
 		 * smart enough to handle loops. */
-		if(inst->U.I.Opcode == RC_OPCODE_BGNLOOP){
+		if(inst->U.I.Opcode == RC_OPCODE_BGNLOOP
+				|| inst->U.I.Opcode == RC_OPCODE_ENDLOOP){
 			return;
 		}
 		rc_for_all_reads_mask(inst, peephole_scan_read, &s);
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 3cc2897293..857aae5514 100644
--- a/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/compiler/radeon_program_alu.c
@@ -988,17 +988,22 @@ void radeonTransformKILP(struct radeon_compiler * c)
 	for (inst = c->Program.Instructions.Next;
 			inst != &c->Program.Instructions; inst = inst->Next) {
 
-		if (inst->U.I.Opcode != RC_OPCODE_KILP
-			|| inst->Prev->U.I.Opcode != RC_OPCODE_IF
-			|| inst->Next->U.I.Opcode != RC_OPCODE_ENDIF) {
+		if (inst->U.I.Opcode != RC_OPCODE_KILP)
 			continue;
-		}
+
 		inst->U.I.Opcode = RC_OPCODE_KIL;
-		inst->U.I.SrcReg[0] = negate(absolute(inst->Prev->U.I.SrcReg[0]));
 
-		/* Remove IF */
-		rc_remove_instruction(inst->Prev);
-		/* Remove ENDIF */
-		rc_remove_instruction(inst->Next);
+		if (inst->Prev->U.I.Opcode != RC_OPCODE_IF
+				|| inst->Next->U.I.Opcode != RC_OPCODE_ENDIF) {
+			inst->U.I.SrcReg[0] = negate(builtin_one);
+		} else {
+
+			inst->U.I.SrcReg[0] =
+				negate(absolute(inst->Prev->U.I.SrcReg[0]));
+			/* Remove IF */
+			rc_remove_instruction(inst->Prev);
+			/* Remove ENDIF */
+			rc_remove_instruction(inst->Next);
+		}
 	}
 }
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index bb8f91491f..cf89ab7ec3 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -327,6 +327,8 @@ void r300RunRenderPrimitive(GLcontext * ctx, int start, int end, int prim)
 	BATCH_LOCALS(&rmesa->radeon);
 	int type, num_verts;
 
+	radeon_prepare_render(&rmesa->radeon);
+
 	type = r300PrimitiveType(rmesa, prim);
 	num_verts = r300NumVerts(rmesa, end - start, prim);
 
diff --git a/src/mesa/drivers/dri/r600/r600_blit.c b/src/mesa/drivers/dri/r600/r600_blit.c
index 172f85eb26..619678214f 100644
--- a/src/mesa/drivers/dri/r600/r600_blit.c
+++ b/src/mesa/drivers/dri/r600/r600_blit.c
@@ -390,13 +390,20 @@ set_render_target(context_t *context, struct radeon_bo *bo, gl_format mesa_forma
 			 0, RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT, 0);
     END_BATCH();
 
-    BEGIN_BATCH_NO_AUTOSTATE(12);
+    BEGIN_BATCH_NO_AUTOSTATE(9);
     R600_OUT_BATCH_REGVAL(CB_COLOR0_SIZE + (4 * id), cb_color0_size);
     R600_OUT_BATCH_REGVAL(CB_COLOR0_VIEW + (4 * id), cb_color0_view);
-    R600_OUT_BATCH_REGVAL(CB_COLOR0_INFO + (4 * id), cb_color0_info);
     R600_OUT_BATCH_REGVAL(CB_COLOR0_MASK + (4 * id), 0);
     END_BATCH();
 
+    BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+    R600_OUT_BATCH_REGVAL(CB_COLOR0_INFO + (4 * id), cb_color0_info);
+    R600_OUT_BATCH_RELOC(0,
+			 bo,
+			 0,
+			 0, RADEON_GEM_DOMAIN_VRAM | RADEON_GEM_DOMAIN_GTT, 0);
+    END_BATCH();
+
     COMMIT_BATCH();
 
 }
@@ -1447,7 +1454,7 @@ set_default_state(context_t *context)
 	    SETbit(sq_dyn_gpr_cntl_ps_flush_req, VS_PC_LIMIT_ENABLE_bit);
     }
 
-    BEGIN_BATCH_NO_AUTOSTATE(117);
+    BEGIN_BATCH_NO_AUTOSTATE(114);
     R600_OUT_BATCH_REGSEQ(SQ_CONFIG, 6);
     R600_OUT_BATCH(sq_config);
     R600_OUT_BATCH(sq_gpr_resource_mgmt_1);
@@ -1477,7 +1484,6 @@ set_default_state(context_t *context)
                          (CLRCMP_SEL_SRC << CLRCMP_FCN_SEL_shift));
     R600_OUT_BATCH_REGVAL(SQ_VTX_BASE_VTX_LOC, 0);
     R600_OUT_BATCH_REGVAL(SQ_VTX_START_INST_LOC, 0);
-    R600_OUT_BATCH_REGVAL(DB_DEPTH_INFO, 0);
     R600_OUT_BATCH_REGVAL(DB_DEPTH_CONTROL, 0);
     R600_OUT_BATCH_REGVAL(CB_SHADER_MASK, (OUTPUT0_ENABLE_mask));
     R600_OUT_BATCH_REGVAL(CB_TARGET_MASK, (TARGET0_ENABLE_mask));
@@ -1607,7 +1613,7 @@ unsigned r600_blit(GLcontext *ctx,
     /* Flush is needed to make sure that source buffer has correct data */
     radeonFlush(ctx);
 
-    rcommonEnsureCmdBufSpace(&context->radeon, 304, __FUNCTION__);
+    rcommonEnsureCmdBufSpace(&context->radeon, 305, __FUNCTION__);
 
     /* load shaders */
     load_shaders(context->radeon.glCtx);
@@ -1616,7 +1622,7 @@ unsigned r600_blit(GLcontext *ctx,
         return GL_FALSE;
 
     /* set clear state */
-    /* 117 */
+    /* 114 */
     set_default_state(context);
 
     /* shaders */
@@ -1632,7 +1638,7 @@ unsigned r600_blit(GLcontext *ctx,
     set_tex_sampler(context);
 
     /* dst */
-    /* 27 */
+    /* 31 */
     set_render_target(context, dst_bo, dst_mesaformat,
 		      dst_pitch, dst_width, dst_height, dst_offset);
     /* scissors */
diff --git a/src/mesa/drivers/dri/r600/r700_assembler.c b/src/mesa/drivers/dri/r600/r700_assembler.c
index 99a33df4fc..9c954cbf70 100644
--- a/src/mesa/drivers/dri/r600/r700_assembler.c
+++ b/src/mesa/drivers/dri/r600/r700_assembler.c
@@ -275,7 +275,10 @@ GLuint GetSurfaceFormat(GLenum eType, GLuint nChannels, GLuint * pClient_size)
                 case 2:
                     format = FMT_8_8; break;
                 case 3:
-                    format = FMT_8_8_8; break;
+                    /* for some (small/unaligned) strides using 4 comps works
+                     * better, probably same as GL_SHORT below
+                     * test piglit/draw-vertices */
+                    format = FMT_8_8_8_8; break;
                 case 4:
                     format = FMT_8_8_8_8; break;
                 default:
@@ -2872,25 +2875,92 @@ GLboolean assemble_CMP(r700_AssemblerBase *pAsm)
 
 GLboolean assemble_TRIG(r700_AssemblerBase *pAsm, BITS opcode)
 {
+    /* 
+     * r600 - trunc to -PI..PI range
+     * r700 - normalize by dividing by 2PI
+     * see fdo bug 27901
+     */
+  
     int tmp;
     checkop1(pAsm);
 
     tmp = gethelpr(pAsm);
 
-    pAsm->D.dst.opcode = SQ_OP2_INST_MUL;
+    pAsm->D.dst.opcode = SQ_OP3_INST_MULADD;
+    pAsm->D.dst.op3    = 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;
 
     assemble_src(pAsm, 0, -1);
 
     pAsm->S[1].src.rtype = SRC_REC_LITERAL;
     setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+    
+    pAsm->S[2].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_Y);
+
     pAsm->D2.dst2.literal_slots = 1;
     pAsm->C[0].f = 1/(3.1415926535 * 2);
-    pAsm->C[1].f = 0.0F;
-    next_ins(pAsm);
+    pAsm->C[1].f = 0.5f;
+    
+    if ( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_FRACT;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+    pAsm->D.dst.writex = 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);
+
+    if(( GL_FALSE == next_ins(pAsm) ))
+    {
+        return GL_FALSE;
+    }
+    pAsm->D.dst.opcode = SQ_OP3_INST_MULADD;
+    pAsm->D.dst.op3    = 1;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+
+    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);
+
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+
+    pAsm->S[2].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_Y);
+
+    pAsm->D2.dst2.literal_slots = 1;
+
+    if (pAsm->bR6xx)
+    {
+       pAsm->C[0].f = 3.1415926535897f * 2.0f;
+       pAsm->C[1].f = -3.1415926535897f;
+    }
+    else 
+    {
+       pAsm->C[0].f = 1.0f;
+       pAsm->C[1].f = -0.5f;
+    }
+
+    if(( GL_FALSE == next_ins(pAsm) ))
+    {
+        return GL_FALSE;
+    }
 
     pAsm->D.dst.opcode = opcode;
     pAsm->D.dst.math = 1;
@@ -4030,22 +4100,79 @@ GLboolean assemble_SCS(r700_AssemblerBase *pAsm)
     checkop1(pAsm);
 
     tmp = gethelpr(pAsm);
-    /* tmp.x = src /2*PI */
-    pAsm->D.dst.opcode = SQ_OP2_INST_MUL;
+
+    pAsm->D.dst.opcode = SQ_OP3_INST_MULADD;
+    pAsm->D.dst.op3    = 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;
 
     assemble_src(pAsm, 0, -1);
 
     pAsm->S[1].src.rtype = SRC_REC_LITERAL;
     setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+
+    pAsm->S[2].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_Y);
+
     pAsm->D2.dst2.literal_slots = 1;
     pAsm->C[0].f = 1/(3.1415926535 * 2);
-    pAsm->C[1].f = 0.0F;
+    pAsm->C[1].f = 0.5F;
 
-    next_ins(pAsm);
+    if ( GL_FALSE == next_ins(pAsm) )
+    {
+        return GL_FALSE;
+    }
+
+    pAsm->D.dst.opcode = SQ_OP2_INST_FRACT;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+    pAsm->D.dst.writex = 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);
+
+    if(( GL_FALSE == next_ins(pAsm) ))
+    {
+        return GL_FALSE;
+    }
+    pAsm->D.dst.opcode = SQ_OP3_INST_MULADD;
+    pAsm->D.dst.op3    = 1;
+
+    setaddrmode_PVSDST(&(pAsm->D.dst), ADDR_ABSOLUTE);
+    pAsm->D.dst.rtype  = DST_REG_TEMPORARY;
+    pAsm->D.dst.reg    = tmp;
+
+    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);
+
+    pAsm->S[1].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[1].src), SQ_SEL_X);
+
+    pAsm->S[2].src.rtype = SRC_REC_LITERAL;
+    setswizzle_PVSSRC(&(pAsm->S[2].src), SQ_SEL_Y);
+
+    pAsm->D2.dst2.literal_slots = 1;
+
+    if(pAsm->bR6xx) {
+       pAsm->C[0].f = 3.1415926535897f * 2.0f;
+       pAsm->C[1].f = -3.1415926535897f;
+    } else {
+       pAsm->C[0].f = 1.0f;
+       pAsm->C[1].f = -0.5f;
+    }
+
+    if(( GL_FALSE == next_ins(pAsm) ))
+    {
+        return GL_FALSE;
+    }
 
     // COS dst.x,    a.x
     pAsm->D.dst.opcode = SQ_OP2_INST_COS;
@@ -6473,7 +6600,7 @@ GLboolean Process_Fragment_Exports(r700_AssemblerBase *pR700AsmCode,
      * results are undefined anyway */
     if(export_count == 0)
     {
-        Process_Export(pR700AsmCode, SQ_EXPORT_PIXEL, 0, 1, 0, GL_FALSE);
+        Process_Export(pR700AsmCode, SQ_EXPORT_PIXEL, 0, 1, pR700AsmCode->starting_export_register_number, GL_FALSE);
     }
     
     if(pR700AsmCode->cf_last_export_ptr != NULL) 
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index cefda3ac4b..1e955b93b2 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -617,18 +617,25 @@ static void r700SendDepthTargetState(GLcontext *ctx, struct radeon_state_atom *a
 
 	r700SetDepthTarget(context);
 
-        BEGIN_BATCH_NO_AUTOSTATE(8 + 2);
+        BEGIN_BATCH_NO_AUTOSTATE(7 + 2);
 	R600_OUT_BATCH_REGSEQ(DB_DEPTH_SIZE, 2);
 	R600_OUT_BATCH(r700->DB_DEPTH_SIZE.u32All);
 	R600_OUT_BATCH(r700->DB_DEPTH_VIEW.u32All);
-	R600_OUT_BATCH_REGSEQ(DB_DEPTH_BASE, 2);
+	R600_OUT_BATCH_REGSEQ(DB_DEPTH_BASE, 1);
 	R600_OUT_BATCH(r700->DB_DEPTH_BASE.u32All);
-	R600_OUT_BATCH(r700->DB_DEPTH_INFO.u32All);
 	R600_OUT_BATCH_RELOC(r700->DB_DEPTH_BASE.u32All,
 			     rrb->bo,
 			     r700->DB_DEPTH_BASE.u32All,
 			     0, RADEON_GEM_DOMAIN_VRAM, 0);
         END_BATCH();
+        BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+	R600_OUT_BATCH_REGSEQ(DB_DEPTH_INFO, 1);
+	R600_OUT_BATCH(r700->DB_DEPTH_INFO.u32All);
+	R600_OUT_BATCH_RELOC(r700->DB_DEPTH_INFO.u32All,
+			     rrb->bo,
+			     r700->DB_DEPTH_INFO.u32All,
+			     0, RADEON_GEM_DOMAIN_VRAM, 0);
+        END_BATCH();
 
 	if ((context->radeon.radeonScreen->chip_family > CHIP_FAMILY_R600) &&
 	    (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)) {
@@ -687,27 +694,35 @@ static void r700SendRenderTargetState(GLcontext *ctx, struct radeon_state_atom *
 	BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
 	R600_OUT_BATCH_REGSEQ(CB_COLOR0_TILE + (4 * id), 1);
 	R600_OUT_BATCH(r700->render_target[id].CB_COLOR0_TILE.u32All);
-	R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_BASE.u32All,
+	R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_TILE.u32All,
 			     rrb->bo,
-			     r700->render_target[id].CB_COLOR0_BASE.u32All,
+			     r700->render_target[id].CB_COLOR0_TILE.u32All,
 			     0, RADEON_GEM_DOMAIN_VRAM, 0);
 	END_BATCH();
 	BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
 	R600_OUT_BATCH_REGSEQ(CB_COLOR0_FRAG + (4 * id), 1);
 	R600_OUT_BATCH(r700->render_target[id].CB_COLOR0_FRAG.u32All);
-	R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_BASE.u32All,
+	R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_FRAG.u32All,
 			     rrb->bo,
-			     r700->render_target[id].CB_COLOR0_BASE.u32All,
+			     r700->render_target[id].CB_COLOR0_FRAG.u32All,
 			     0, RADEON_GEM_DOMAIN_VRAM, 0);
         END_BATCH();
 
-        BEGIN_BATCH_NO_AUTOSTATE(12);
+        BEGIN_BATCH_NO_AUTOSTATE(9);
 	R600_OUT_BATCH_REGVAL(CB_COLOR0_SIZE + (4 * id), r700->render_target[id].CB_COLOR0_SIZE.u32All);
 	R600_OUT_BATCH_REGVAL(CB_COLOR0_VIEW + (4 * id), r700->render_target[id].CB_COLOR0_VIEW.u32All);
-	R600_OUT_BATCH_REGVAL(CB_COLOR0_INFO + (4 * id), r700->render_target[id].CB_COLOR0_INFO.u32All);
 	R600_OUT_BATCH_REGVAL(CB_COLOR0_MASK + (4 * id), r700->render_target[id].CB_COLOR0_MASK.u32All);
         END_BATCH();
 
+	BEGIN_BATCH_NO_AUTOSTATE(3 + 2);
+	R600_OUT_BATCH_REGVAL(CB_COLOR0_INFO + (4 * id), r700->render_target[id].CB_COLOR0_INFO.u32All);
+	R600_OUT_BATCH_RELOC(r700->render_target[id].CB_COLOR0_INFO.u32All,
+			     rrb->bo,
+			     r700->render_target[id].CB_COLOR0_INFO.u32All,
+			     0, RADEON_GEM_DOMAIN_VRAM, 0);
+
+        END_BATCH();
+
 	COMMIT_BATCH();
 
 }
@@ -1567,7 +1582,7 @@ void r600InitAtoms(context_t *context)
 	ALLOC_STATE(sq, always, 34, r700SendSQConfig);
 	ALLOC_STATE(db, always, 17, r700SendDBState);
 	ALLOC_STATE(stencil, always, 4, r700SendStencilState);
-	ALLOC_STATE(db_target, always, 12, r700SendDepthTargetState);
+	ALLOC_STATE(db_target, always, 16, r700SendDepthTargetState);
 	ALLOC_STATE(sc, always, 15, r700SendSCState);
 	ALLOC_STATE(scissor, always, 22, r700SendScissorState);
 	ALLOC_STATE(aa, always, 12, r700SendAAState);
@@ -1578,7 +1593,7 @@ 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, 29, r700SendRenderTargetState);
+	ALLOC_STATE(cb_target, always, 31, r700SendRenderTargetState);
 	ALLOC_STATE(blnd, blnd, (6 + (R700_MAX_RENDER_TARGETS * 3)), r700SendCBBlendState);
 	ALLOC_STATE(blnd_clr, always, 6, r700SendCBBlendColorState);
 	ALLOC_STATE(sx, always, 9, r700SendSXState);
diff --git a/src/mesa/drivers/dri/r600/r700_clear.c b/src/mesa/drivers/dri/r600/r700_clear.c
index 09c48565b6..d1008f28b9 100644
--- a/src/mesa/drivers/dri/r600/r700_clear.c
+++ b/src/mesa/drivers/dri/r600/r700_clear.c
@@ -48,6 +48,7 @@ static GLboolean r700ClearFast(context_t *context, GLbitfield mask)
 void r700Clear(GLcontext * ctx, GLbitfield mask)
 {
     context_t *context = R700_CONTEXT(ctx);
+    radeonContextPtr radeon = &context->radeon;
     __DRIdrawable *dPriv = radeon_get_drawable(&context->radeon);
     const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask[0]);
     GLbitfield swrast_mask = 0, tri_mask = 0;
@@ -60,6 +61,8 @@ void r700Clear(GLcontext * ctx, GLbitfield mask)
         context->radeon.front_buffer_dirty = GL_TRUE;
     }
 
+    radeon_prepare_render(radeon);
+
     if( GL_TRUE == r700ClearFast(context, mask) )
     {
         return;
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index 1929b7cc12..ba55f38e05 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -625,11 +625,11 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
 
         stride = (input[i]->StrideB == 0) ? getTypeSize(input[i]->Type) * input[i]->Size : input[i]->StrideB;
 
-        if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT ||
+        if (input[i]->Type == GL_DOUBLE || input[i]->Type == GL_UNSIGNED_INT || input[i]->Type == GL_INT
 #if MESA_BIG_ENDIAN
-            getTypeSize(input[i]->Type) != 4 || 
+            || getTypeSize(input[i]->Type) != 4
 #endif
-            stride < 4) 
+            ) 
         {
             r700ConvertAttrib(ctx, count, input[i], &context->stream_desc[index]);
         } 
@@ -637,19 +637,10 @@ static void r700SetupStreams(GLcontext *ctx, const struct gl_client_array *input
         {
             if (input[i]->BufferObj->Name) 
             {
-                if (stride % 4 != 0) 
-                {
-                    assert(((intptr_t) input[i]->Ptr) % input[i]->StrideB == 0);
-                    r700AlignDataToDword(ctx, input[i], count, &context->stream_desc[index]);
-                    context->stream_desc[index].is_named_bo = GL_FALSE;
-                } 
-                else 
-                {
-                    context->stream_desc[index].stride = input[i]->StrideB;
-                    context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr;
-                    context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo;
-                    context->stream_desc[index].is_named_bo = GL_TRUE;
-                }
+                context->stream_desc[index].stride = input[i]->StrideB;
+                context->stream_desc[index].bo_offset = (intptr_t) input[i]->Ptr;
+                context->stream_desc[index].bo = get_radeon_buffer_object(input[i]->BufferObj)->bo;
+                context->stream_desc[index].is_named_bo = GL_TRUE;
             } 
             else 
             {
@@ -977,6 +968,10 @@ static void r700DrawPrims(GLcontext *ctx,
 {
 	GLboolean retval = GL_FALSE;
 
+	context_t *context = R700_CONTEXT(ctx);
+	radeonContextPtr radeon = &context->radeon;
+	radeon_prepare_render(radeon);
+
 	/* This check should get folded into just the places that
 	 * min/max index are really needed.
 	 */
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index 137f3007ce..6a2a09eaf1 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -461,11 +461,11 @@ static void r700TranslateAttrib(GLcontext *ctx, GLuint unLoc, int count, const s
 	stride = (input->StrideB == 0) ? getTypeSize(input->Type) * input->Size 
                                    : input->StrideB;
 
-    if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT ||
+    if (input->Type == GL_DOUBLE || input->Type == GL_UNSIGNED_INT || input->Type == GL_INT
 #if MESA_BIG_ENDIAN
-        getTypeSize(input->Type) != 4 ||
+        || getTypeSize(input->Type) != 4
 #endif
-        stride < 4) 
+       ) 
     {
         pStreamDesc->type = GL_FLOAT;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_common.c b/src/mesa/drivers/dri/radeon/radeon_common.c
index 13f1f0611b..c1a660af3d 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common.c
@@ -708,7 +708,6 @@ void radeon_draw_buffer(GLcontext *ctx, struct gl_framebuffer *fb)
 		if (fb->_ColorDrawBufferIndexes[0] == BUFFER_FRONT_LEFT) {
 			rrbColor = radeon_renderbuffer(fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer);
 			radeon->front_cliprects = GL_TRUE;
-			radeon->front_buffer_dirty = GL_TRUE;
 		} else {
 			rrbColor = radeon_renderbuffer(fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer);
 			radeon->front_cliprects = GL_FALSE;
@@ -1132,17 +1131,13 @@ flush_front:
 		if (screen->dri2.loader && (screen->dri2.loader->base.version >= 2)
 			&& (screen->dri2.loader->flushFrontBuffer != NULL)) {
 			__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
-			 * enabled.  This is done so that the dirty bit can only be set in
-			 * glDrawBuffer.  Otherwise the dirty bit would have to be set at
-			 * each of N places that do rendering.  This has worse performances,
-			 * but it is much easier to get correct.
+			/* We set the dirty bit in radeon_prepare_render() if we're
+			 * front buffer rendering once we get there.
 			 */
-			if (!radeon->is_front_buffer_rendering) {
-				radeon->front_buffer_dirty = GL_FALSE;
-			}
+			radeon->front_buffer_dirty = GL_FALSE;
+
+			(*screen->dri2.loader->flushFrontBuffer)(drawable, drawable->loaderPrivate);
 		}
 	}
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.c b/src/mesa/drivers/dri/radeon/radeon_common_context.c
index 5a7d52c4d2..92663bf66d 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.c
@@ -493,6 +493,50 @@ radeon_bits_per_pixel(const struct radeon_renderbuffer *rb)
    return _mesa_get_format_bytes(rb->base.Format) * 8; 
 }
 
+/*
+ * Check if drawable has been invalidated by dri2InvalidateDrawable().
+ * Update renderbuffers if so. This prevents a client from accessing
+ * a backbuffer that has a swap pending but not yet completed.
+ *
+ * See intel_prepare_render for equivalent code in intel driver.
+ *
+ */
+void radeon_prepare_render(radeonContextPtr radeon)
+{
+    __DRIcontext *driContext = radeon->dri.context;
+    __DRIdrawable *drawable;
+    __DRIscreen *screen;
+
+    screen = driContext->driScreenPriv;
+    if (!screen->dri2.loader)
+        return;
+
+    drawable = driContext->driDrawablePriv;
+    if (drawable->dri2.stamp != driContext->dri2.draw_stamp) {
+	if (drawable->lastStamp != drawable->dri2.stamp)
+	    radeon_update_renderbuffers(driContext, drawable, GL_FALSE);
+
+	/* Intel driver does the equivalent of this, no clue if it is needed:
+	 * radeon_draw_buffer(radeon->glCtx, &(drawable->driverPrivate)->base);
+	 */
+	driContext->dri2.draw_stamp = drawable->dri2.stamp;
+    }
+
+    drawable = driContext->driReadablePriv;
+    if (drawable->dri2.stamp != driContext->dri2.read_stamp) {
+	if (drawable->lastStamp != drawable->dri2.stamp)
+	    radeon_update_renderbuffers(driContext, drawable, GL_FALSE);
+	driContext->dri2.read_stamp = drawable->dri2.stamp;
+    }
+
+    /* If we're currently rendering to the front buffer, the rendering
+     * that will happen next will probably dirty the front buffer.  So
+     * mark it as dirty here.
+     */
+    if (radeon->is_front_buffer_rendering)
+	radeon->front_buffer_dirty = GL_TRUE;
+}
+
 void
 radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
 			    GLboolean front_only)
@@ -514,6 +558,11 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
 	screen = context->driScreenPriv;
 	radeon = (radeonContextPtr) context->driverPrivate;
 
+	/* Set this up front, so that in case our buffers get invalidated
+	 * while we're getting new buffers, we don't clobber the stamp and
+	 * thus ignore the invalidate. */
+	drawable->lastStamp = drawable->dri2.stamp;
+
 	if (screen->dri2.loader
 	   && (screen->dri2.loader->base.version > 2)
 	   && (screen->dri2.loader->getBuffersWithFormat != NULL)) {
@@ -650,6 +699,13 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
 		rb->base.Height = drawable->h;
 		rb->has_surface = 0;
 
+		/* r6xx+ tiling */
+		rb->tile_config = radeon->radeonScreen->tile_config;
+		rb->group_bytes = radeon->radeonScreen->group_bytes;
+		rb->num_channels = radeon->radeonScreen->num_channels;
+		rb->num_banks = radeon->radeonScreen->num_banks;
+		rb->r7xx_bank_op = radeon->radeonScreen->r7xx_bank_op;
+
 		if (buffers[i].attachment == __DRI_BUFFER_STENCIL && depth_bo) {
 			if (RADEON_DEBUG & RADEON_DRI)
 				fprintf(stderr, "(reusing depth buffer as stencil)\n");
@@ -678,7 +734,7 @@ radeon_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable,
 				bo->flags |= RADEON_BO_FLAGS_MACRO_TILE;
 			if (tiling_flags & RADEON_TILING_MICRO)
 				bo->flags |= RADEON_BO_FLAGS_MICRO_TILE;
-			
+
 		}
 
 		if (buffers[i].attachment == __DRI_BUFFER_DEPTH) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_common_context.h b/src/mesa/drivers/dri/radeon/radeon_common_context.h
index 5156c5d0d0..f06e5fdf24 100644
--- a/src/mesa/drivers/dri/radeon/radeon_common_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_common_context.h
@@ -93,6 +93,13 @@ struct radeon_renderbuffer
 	GLuint pf_pending;  /**< sequence number of pending flip */
 	GLuint vbl_pending;   /**< vblank sequence number of pending flip */
 	__DRIdrawable *dPriv;
+
+	/* r6xx+ tiling */
+	GLuint tile_config;
+	GLint group_bytes;
+	GLint num_channels;
+	GLint num_banks;
+	GLint r7xx_bank_op;
 };
 
 struct radeon_framebuffer
@@ -614,5 +621,6 @@ GLboolean radeonMakeCurrent(__DRIcontext * driContextPriv,
 			    __DRIdrawable * driDrawPriv,
 			    __DRIdrawable * driReadPriv);
 extern void radeonDestroyContext(__DRIcontext * driContextPriv);
+void radeon_prepare_render(radeonContextPtr radeon);
 
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_pixel_read.c b/src/mesa/drivers/dri/radeon/radeon_pixel_read.c
index dadb8002c7..fb741173ca 100644
--- a/src/mesa/drivers/dri/radeon/radeon_pixel_read.c
+++ b/src/mesa/drivers/dri/radeon/radeon_pixel_read.c
@@ -179,6 +179,9 @@ radeonReadPixels(GLcontext * ctx,
                  GLenum format, GLenum type,
                  const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
 {
+    radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+    radeon_prepare_render(radeon);
+
     if (do_blit_readpixels(ctx, x, y, width, height, format, type, pack, pixels))
         return;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 82107cc6ae..fa97a19302 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -213,6 +213,10 @@ static const GLuint __driNConfigOptions = 17;
 
 static int getSwapInfo( __DRIdrawable *dPriv, __DRIswapInfo * sInfo );
 
+#ifndef RADEON_INFO_TILE_CONFIG
+#define RADEON_INFO_TILE_CONFIG 0x6
+#endif
+
 static int
 radeonGetParam(__DRIscreen *sPriv, int param, void *value)
 {
@@ -232,6 +236,9 @@ radeonGetParam(__DRIscreen *sPriv, int param, void *value)
       case RADEON_PARAM_NUM_Z_PIPES:
           info.request = RADEON_INFO_NUM_Z_PIPES;
           break;
+      case RADEON_INFO_TILE_CONFIG:
+	  info.request = RADEON_INFO_TILE_CONFIG;
+          break;
       default:
           return -EINVAL;
       }
@@ -376,6 +383,21 @@ static const __DRItexBufferExtension r600TexBufferExtension = {
 };
 #endif
 
+static void
+radeonDRI2Flush(__DRIdrawable *drawable)
+{
+    radeonContextPtr rmesa;
+
+    rmesa = (radeonContextPtr) drawable->driContextPriv->driverPrivate;
+    radeonFlush(rmesa->glCtx);
+}
+
+static const struct __DRI2flushExtensionRec radeonFlushExtension = {
+    { __DRI2_FLUSH, __DRI2_FLUSH_VERSION },
+    radeonDRI2Flush,
+    dri2InvalidateDrawable,
+};
+
 static int radeon_set_screen_flags(radeonScreenPtr screen, int device_id)
 {
    screen->device_id = device_id;
@@ -1305,6 +1327,56 @@ radeonCreateScreen2(__DRIscreen *sPriv)
    else
 	   screen->chip_flags |= RADEON_CLASS_R600;
 
+   /* r6xx+ tiling */
+   if (IS_R600_CLASS(screen) && (sPriv->drm_version.minor >= 6)) {
+	   ret = radeonGetParam(sPriv, RADEON_INFO_TILE_CONFIG, &temp);
+	   if (ret)
+		   fprintf(stderr, "failed to get tiling info\n");
+	   else {
+		   screen->tile_config = temp;
+		   screen->r7xx_bank_op = 0;
+		   switch((screen->tile_config & 0xe) >> 1) {
+		   case 0:
+			   screen->num_channels = 1;
+			   break;
+		   case 1:
+			   screen->num_channels = 2;
+			   break;
+		   case 2:
+			   screen->num_channels = 4;
+			   break;
+		   case 3:
+			   screen->num_channels = 8;
+			   break;
+		   default:
+			   fprintf(stderr, "bad channels\n");
+			   break;
+		   }
+		   switch((screen->tile_config & 0x30) >> 4) {
+		   case 0:
+			   screen->num_banks = 4;
+			   break;
+		   case 1:
+			   screen->num_banks = 8;
+			   break;
+		   default:
+			   fprintf(stderr, "bad banks\n");
+			   break;
+		   }
+		   switch((screen->tile_config & 0xc0) >> 6) {
+		   case 0:
+			   screen->group_bytes = 256;
+			   break;
+		   case 1:
+			   screen->group_bytes = 512;
+			   break;
+		   default:
+			   fprintf(stderr, "bad group_bytes\n");
+			   break;
+		   }
+	   }
+   }
+
    if (IS_R300_CLASS(screen)) {
        ret = radeonGetParam(sPriv, RADEON_PARAM_NUM_GB_PIPES, &temp);
        if (ret) {
@@ -1379,6 +1451,8 @@ radeonCreateScreen2(__DRIscreen *sPriv)
    screen->extensions[i++] = &r600TexBufferExtension.base;
 #endif
 
+   screen->extensions[i++] = &radeonFlushExtension.base;
+
    screen->extensions[i++] = NULL;
    sPriv->extensions = screen->extensions;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.h b/src/mesa/drivers/dri/radeon/radeon_screen.h
index 0d7e335fa3..2b33201a53 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.h
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.h
@@ -112,6 +112,13 @@ typedef struct radeon_screen {
    int kernel_mm;
    drm_radeon_sarea_t *sarea;	/* Private SAREA data */
    struct radeon_bo_manager *bom;
+
+   /* r6xx+ tiling */
+   GLuint tile_config;
+   GLint group_bytes;
+   GLint num_channels;
+   GLint num_banks;
+   GLint r7xx_bank_op;
 } radeonScreenRec, *radeonScreenPtr;
 
 #define IS_R100_CLASS(screen) \
diff --git a/src/mesa/drivers/dri/radeon/radeon_span.c b/src/mesa/drivers/dri/radeon/radeon_span.c
index 1adb609603..9dfe2dd243 100644
--- a/src/mesa/drivers/dri/radeon/radeon_span.c
+++ b/src/mesa/drivers/dri/radeon/radeon_span.c
@@ -111,7 +111,6 @@ static GLubyte *r200_depth_4byte(const struct radeon_renderbuffer * rrb,
  * two main types:
  * - 1D (akin to macro-linear/micro-tiled on older asics)
  * - 2D (akin to macro-tiled/micro-tiled on older asics)
- * only 1D tiling is implemented below
  */
 #if defined(RADEON_R600)
 static inline GLint r600_1d_tile_helper(const struct radeon_renderbuffer * rrb,
@@ -208,12 +207,190 @@ static inline GLint r600_1d_tile_helper(const struct radeon_renderbuffer * rrb,
     return offset;
 }
 
+static inline GLint r600_log2(GLint n)
+{
+	GLint log2 = 0;
+
+	while (n >>= 1)
+		++log2;
+	return log2;
+}
+
+static inline GLint r600_2d_tile_helper(const struct radeon_renderbuffer * rrb,
+					GLint x, GLint y, GLint is_depth, GLint is_stencil)
+{
+	GLint group_bytes = rrb->group_bytes;
+	GLint num_channels = rrb->num_channels;
+	GLint num_banks = rrb->num_banks;
+	GLint r7xx_bank_op = rrb->r7xx_bank_op;
+	/* */
+	GLint group_bits = r600_log2(group_bytes);
+	GLint channel_bits = r600_log2(num_channels);
+	GLint bank_bits = r600_log2(num_banks);
+	GLint element_bytes = rrb->cpp;
+	GLint num_samples = 1;
+	GLint tile_width = 8;
+	GLint tile_height = 8;
+	GLint tile_thickness = 1;
+	GLint macro_tile_width = num_banks;
+	GLint macro_tile_height = num_channels;
+	GLint pitch_elements = (rrb->pitch / element_bytes) / tile_width;
+	GLint height = rrb->base.Height / tile_height;
+	GLint z = 0;
+	GLint sample_number = 0;
+	/* */
+	GLint tile_bytes;
+	GLint macro_tile_bytes;
+	GLint macro_tiles_per_row;
+	GLint macro_tiles_per_slice;
+	GLint slice_offset;
+	GLint macro_tile_row_index;
+	GLint macro_tile_column_index;
+	GLint macro_tile_offset;
+	GLint pixel_number = 0;
+	GLint element_offset;
+	GLint bank = 0;
+	GLint channel = 0;
+	GLint total_offset;
+	GLint group_mask = (1 << group_bits) - 1;
+	GLint offset_low;
+	GLint offset_high;
+	GLint offset = 0;
+
+	switch (num_channels) {
+	case 2:
+	default:
+		// channel[0] = x[3] ^ y[3]
+		channel |= (((x >> 3) ^ (y >> 3)) & 1) << 0;
+		break;
+	case 4:
+		// channel[0] = x[4] ^ y[3]
+		channel |= (((x >> 4) ^ (y >> 3)) & 1) << 0;
+		// channel[1] = x[3] ^ y[4]
+		channel |= (((x >> 3) ^ (y >> 4)) & 1) << 1;
+		break;
+	case 8:
+		// channel[0] = x[5] ^ y[3]
+		channel |= (((x >> 5) ^ (y >> 3)) & 1) << 0;
+		// channel[0] = x[4] ^ x[5] ^ y[4]
+		channel |= (((x >> 4) ^ (x >> 5) ^ (y >> 4)) & 1) << 1;
+		// channel[0] = x[3] ^ y[5]
+		channel |= (((x >> 3) ^ (y >> 5)) & 1) << 2;
+		break;
+	}
+
+	switch (num_banks) {
+	case 4:
+		// bank[0] = x[3] ^ y[4 + log2(num_channels)]
+		bank |= (((x >> 3) ^ (y >> (4 + channel_bits))) & 1) << 0;
+		if (r7xx_bank_op)
+			// bank[1] = x[3] ^ y[4 + log2(num_channels)] ^ x[5]
+			bank |= (((x >> 4) ^ (y >> (3 + channel_bits)) ^ (x >> 5)) & 1) << 1;
+		else
+			// bank[1] = x[4] ^ y[3 + log2(num_channels)]
+			bank |= (((x >> 4) ^ (y >> (3 + channel_bits))) & 1) << 1;
+		break;
+	case 8:
+		// bank[0] = x[3] ^ y[5 + log2(num_channels)]
+		bank |= (((x >> 3) ^ (y >> (5 + channel_bits))) & 1) << 0;
+		// bank[1] = x[4] ^ y[4 + log2(num_channels)] ^ y[5 + log2(num_channels)]
+		bank |= (((x >> 4) ^ (y >> (4 + channel_bits)) ^ (y >> (5 + channel_bits))) & 1) << 1;
+		if (r7xx_bank_op)
+			// bank[2] = x[5] ^ y[3 + log2(num_channels)] ^ x[6]
+			bank |= (((x >> 5) ^ (y >> (3 + channel_bits)) ^ (x >> 6)) & 1) << 2;
+		else
+			// bank[2] = x[5] ^ y[3 + log2(num_channels)]
+			bank |= (((x >> 5) ^ (y >> (3 + channel_bits))) & 1) << 2;
+		break;
+	}
+
+	tile_bytes = tile_width * tile_height * tile_thickness * element_bytes * num_samples;
+	macro_tile_bytes = macro_tile_width * macro_tile_height * tile_bytes;
+	macro_tiles_per_row = pitch_elements / macro_tile_width;
+	macro_tiles_per_slice = macro_tiles_per_row * (height / macro_tile_height);
+	slice_offset = (z / tile_thickness) * macro_tiles_per_slice * macro_tile_bytes;
+	macro_tile_row_index = (y / tile_height) / macro_tile_height;
+	macro_tile_column_index = (x / tile_width) / macro_tile_width;
+	macro_tile_offset = ((macro_tile_row_index * macro_tiles_per_row) + macro_tile_column_index) * macro_tile_bytes;
+
+	if (is_depth) {
+		GLint pixel_offset = 0;
+
+		pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0]
+		pixel_number |= ((y >> 0) & 1) << 1; // pn[1] = y[0]
+		pixel_number |= ((x >> 1) & 1) << 2; // pn[2] = x[1]
+		pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1]
+		pixel_number |= ((x >> 2) & 1) << 4; // pn[4] = x[2]
+		pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2]
+		switch (element_bytes) {
+		case 2:
+			pixel_offset = pixel_number * element_bytes * num_samples;
+			break;
+		case 4:
+			/* stencil and depth data are stored separately within a tile.
+			 * stencil is stored in a contiguous tile before the depth tile.
+			 * stencil element is 1 byte, depth element is 3 bytes.
+			 * stencil tile is 64 bytes.
+			 */
+			if (is_stencil)
+				pixel_offset = pixel_number * 1 * num_samples;
+			else
+				pixel_offset = (pixel_number * 3 * num_samples) + 64;
+			break;
+		}
+		element_offset = pixel_offset + (sample_number * element_bytes);
+	} else {
+		GLint sample_offset;
+
+		switch (element_bytes) {
+		case 1:
+			pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0]
+			pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1]
+			pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2]
+			pixel_number |= ((y >> 1) & 1) << 3; // pn[3] = y[1]
+			pixel_number |= ((y >> 0) & 1) << 4; // pn[4] = y[0]
+			pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2]
+			break;
+		case 2:
+			pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0]
+			pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1]
+			pixel_number |= ((x >> 2) & 1) << 2; // pn[2] = x[2]
+			pixel_number |= ((y >> 0) & 1) << 3; // pn[3] = y[0]
+			pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1]
+			pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2]
+			break;
+		case 4:
+			pixel_number |= ((x >> 0) & 1) << 0; // pn[0] = x[0]
+			pixel_number |= ((x >> 1) & 1) << 1; // pn[1] = x[1]
+			pixel_number |= ((y >> 0) & 1) << 2; // pn[2] = y[0]
+			pixel_number |= ((x >> 2) & 1) << 3; // pn[3] = x[2]
+			pixel_number |= ((y >> 1) & 1) << 4; // pn[4] = y[1]
+			pixel_number |= ((y >> 2) & 1) << 5; // pn[5] = y[2]
+			break;
+		}
+		sample_offset = sample_number * (tile_bytes / num_samples);
+		element_offset = sample_offset + (pixel_number * element_bytes);
+	}
+	total_offset = (slice_offset + macro_tile_offset) >> (channel_bits + bank_bits);
+	total_offset += element_offset;
+
+	offset_low = total_offset & group_mask;
+	offset_high = (total_offset & ~group_mask) << (channel_bits + bank_bits);
+	offset = (bank << (group_bits + channel_bits)) + (channel << group_bits) + offset_low + offset_high;
+
+	return offset;
+}
+
 /* depth buffers */
 static GLubyte *r600_ptr_depth(const struct radeon_renderbuffer * rrb,
 			       GLint x, GLint y)
 {
     GLubyte *ptr = rrb->bo->ptr;
-    GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 0);
+    GLint offset;
+    if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
+	    offset = r600_2d_tile_helper(rrb, x, y, 1, 0);
+    else
+	    offset = r600_1d_tile_helper(rrb, x, y, 1, 0);
     return &ptr[offset];
 }
 
@@ -221,7 +398,11 @@ static GLubyte *r600_ptr_stencil(const struct radeon_renderbuffer * rrb,
 				 GLint x, GLint y)
 {
     GLubyte *ptr = rrb->bo->ptr;
-    GLint offset = r600_1d_tile_helper(rrb, x, y, 1, 1);
+    GLint offset;
+    if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
+	    offset = r600_2d_tile_helper(rrb, x, y, 1, 1);
+    else
+	    offset = r600_1d_tile_helper(rrb, x, y, 1, 1);
     return &ptr[offset];
 }
 
@@ -235,7 +416,10 @@ static GLubyte *r600_ptr_color(const struct radeon_renderbuffer * rrb,
     if (rrb->has_surface || !(rrb->bo->flags & mask)) {
         offset = x * rrb->cpp + y * rrb->pitch;
     } else {
-	    offset = r600_1d_tile_helper(rrb, x, y, 0, 0);
+	    if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
+		    offset = r600_2d_tile_helper(rrb, x, y, 0, 0);
+	    else
+		    offset = r600_1d_tile_helper(rrb, x, y, 0, 0);
     }
     return &ptr[offset];
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index f2fcb46688..67be466c3f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -408,6 +408,8 @@ static GLboolean radeon_run_render( GLcontext *ctx,
        !radeon_dma_validate_render( ctx, VB ))
       return GL_TRUE;		
 
+   radeon_prepare_render(&rmesa->radeon);
+
    tnl->Driver.Render.Start( ctx );
 
    for (i = 0 ; i < VB->PrimitiveCount ; i++)
diff --git a/src/mesa/drivers/dri/radeon/radeon_tcl.c b/src/mesa/drivers/dri/radeon/radeon_tcl.c
index ea796e1a45..5e1718f9df 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tcl.c
@@ -252,6 +252,8 @@ void radeonTclPrimitive( GLcontext *ctx,
    GLuint se_cntl;
    GLuint newprim = hw_prim | RADEON_CP_VC_CNTL_TCL_ENABLE;
 
+   radeon_prepare_render(&rmesa->radeon);
+
    if (newprim != rmesa->tcl.hw_primitive ||
        !discrete_prim[hw_prim&0xf]) {
       RADEON_NEWPRIM( rmesa );
diff --git a/src/mesa/drivers/dri/radeon/radeon_tex_copy.c b/src/mesa/drivers/dri/radeon/radeon_tex_copy.c
index 29fd31ac23..4cb0bb60c8 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tex_copy.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tex_copy.c
@@ -153,6 +153,9 @@ radeonCopyTexImage2D(GLcontext *ctx, GLenum target, GLint level,
         _mesa_select_tex_image(ctx, texObj, target, level);
     int srcx, srcy, dstx, dsty;
 
+    radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+    radeon_prepare_render(radeon);
+
     if (border)
         goto fail;
 
@@ -202,6 +205,9 @@ radeonCopyTexSubImage2D(GLcontext *ctx, GLenum target, GLint level,
     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);
 
+    radeonContextPtr radeon = RADEON_CONTEXT(ctx);
+    radeon_prepare_render(radeon);
+
     if (!do_copy_texsubimage(ctx, target, level,
                              radeon_tex_obj(texObj), (radeon_texture_image *)texImage,
                              xoffset, yoffset, x, y, width, height)) {
diff --git a/src/mesa/main/arbprogram.h b/src/mesa/main/arbprogram.h
index 787ffd62f4..e2e535e911 100644
--- a/src/mesa/main/arbprogram.h
+++ b/src/mesa/main/arbprogram.h
@@ -27,7 +27,6 @@
 #define ARBPROGRAM_H
 
 
-#include "compiler.h"
 #include "glheader.h"
 
 
diff --git a/src/mesa/main/arrayobj.h b/src/mesa/main/arrayobj.h
index 8999edc724..fdf7e2bca4 100644
--- a/src/mesa/main/arrayobj.h
+++ b/src/mesa/main/arrayobj.h
@@ -28,7 +28,7 @@
 #ifndef ARRAYOBJ_H
 #define ARRAYOBJ_H
 
-#include "context.h"
+#include "mtypes.h"
 
 /**
  * \file arrayobj.h
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index 002448fedb..753949be50 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -41,6 +41,7 @@
 #include "hint.h"
 #include "light.h"
 #include "lines.h"
+#include "macros.h"
 #include "matrix.h"
 #include "multisample.h"
 #include "points.h"
diff --git a/src/mesa/main/bufferobj.h b/src/mesa/main/bufferobj.h
index 912529cfdf..f234d06c6c 100644
--- a/src/mesa/main/bufferobj.h
+++ b/src/mesa/main/bufferobj.h
@@ -29,7 +29,7 @@
 #define BUFFEROBJ_H
 
 
-#include "context.h"
+#include "mtypes.h"
 
 
 /*
diff --git a/src/mesa/main/clear.c b/src/mesa/main/clear.c
index e76ab5527b..49d86b3b1f 100644
--- a/src/mesa/main/clear.c
+++ b/src/mesa/main/clear.c
@@ -35,6 +35,7 @@
 #include "context.h"
 #include "colormac.h"
 #include "enums.h"
+#include "macros.h"
 #include "state.h"
 
 
diff --git a/src/mesa/main/clear.h b/src/mesa/main/clear.h
index 4c78eeda48..6657370c4b 100644
--- a/src/mesa/main/clear.h
+++ b/src/mesa/main/clear.h
@@ -27,7 +27,7 @@
 #define CLEAR_H
 
 
-#include "main/mtypes.h"
+#include "glheader.h"
 
 
 extern void GLAPIENTRY
diff --git a/src/mesa/main/clip.h b/src/mesa/main/clip.h
index d53afb45bd..ac472d66e0 100644
--- a/src/mesa/main/clip.h
+++ b/src/mesa/main/clip.h
@@ -31,7 +31,7 @@
 #ifndef CLIP_H
 #define CLIP_H
 
-#include "mtypes.h"
+#include "glheader.h"
 
 extern void GLAPIENTRY 
 _mesa_ClipPlane( GLenum plane, const GLdouble *equation );
diff --git a/src/mesa/main/colormac.h b/src/mesa/main/colormac.h
index 905f4e2283..245fb658bb 100644
--- a/src/mesa/main/colormac.h
+++ b/src/mesa/main/colormac.h
@@ -33,9 +33,9 @@
 #define COLORMAC_H
 
 
-#include "imports.h"
 #include "config.h"
 #include "macros.h"
+#include "mtypes.h"
 
 
 /** \def BYTE_TO_CHAN
diff --git a/src/mesa/main/convolve.c b/src/mesa/main/convolve.c
index 15e8dffc23..f63bddc44d 100644
--- a/src/mesa/main/convolve.c
+++ b/src/mesa/main/convolve.c
@@ -37,6 +37,7 @@
 #include "convolve.h"
 #include "context.h"
 #include "image.h"
+#include "macros.h"
 #include "mtypes.h"
 #include "state.h"
 #include "main/dispatch.h"
diff --git a/src/mesa/main/debug.h b/src/mesa/main/debug.h
index 0449cb1798..b517cc8259 100644
--- a/src/mesa/main/debug.h
+++ b/src/mesa/main/debug.h
@@ -36,6 +36,9 @@
 #ifndef _DEBUG_H
 #define _DEBUG_H
 
+#include "glheader.h"
+#include "mtypes.h"
+
 #if _HAVE_FULL_GL
 
 extern void _mesa_print_tri_caps( const char *name, GLuint flags );
diff --git a/src/mesa/main/depthstencil.h b/src/mesa/main/depthstencil.h
index 3dde081f5a..afbac77f0e 100644
--- a/src/mesa/main/depthstencil.h
+++ b/src/mesa/main/depthstencil.h
@@ -26,6 +26,7 @@
 #ifndef DEPTHSTENCIL_H
 #define DEPTHSTENCIL_H
 
+#include "mtypes.h"
 
 extern struct gl_renderbuffer *
 _mesa_new_z24_renderbuffer_wrapper(GLcontext *ctx,
diff --git a/src/mesa/main/fbobject.c b/src/mesa/main/fbobject.c
index 8c86b392c7..9a84e5a79c 100644
--- a/src/mesa/main/fbobject.c
+++ b/src/mesa/main/fbobject.c
@@ -97,6 +97,11 @@ _mesa_init_fbobjects(GLcontext *ctx)
    DummyRenderbuffer.Delete = delete_dummy_renderbuffer;
 }
 
+struct gl_framebuffer *
+_mesa_get_incomplete_framebuffer(void)
+{
+   return &DummyFramebuffer;
+}
 
 /**
  * Helper routine for getting a gl_renderbuffer.
diff --git a/src/mesa/main/fbobject.h b/src/mesa/main/fbobject.h
index ff946033a4..9850ee9aa2 100644
--- a/src/mesa/main/fbobject.h
+++ b/src/mesa/main/fbobject.h
@@ -26,10 +26,14 @@
 #ifndef FBOBJECT_H
 #define FBOBJECT_H
 
+#include "mtypes.h"
 
 extern void
 _mesa_init_fbobjects(GLcontext *ctx);
 
+extern struct gl_framebuffer *
+_mesa_get_incomplete_framebuffer(void);
+
 extern struct gl_renderbuffer *
 _mesa_lookup_renderbuffer(GLcontext *ctx, GLuint id);
 
diff --git a/src/mesa/main/fog.c b/src/mesa/main/fog.c
index 269ff3f8b9..9f26c012d6 100644
--- a/src/mesa/main/fog.c
+++ b/src/mesa/main/fog.c
@@ -27,6 +27,7 @@
 #include "colormac.h"
 #include "context.h"
 #include "fog.h"
+#include "macros.h"
 #include "mtypes.h"
 
 
diff --git a/src/mesa/main/formats.c b/src/mesa/main/formats.c
index 49463fcc3c..90449cc04f 100644
--- a/src/mesa/main/formats.c
+++ b/src/mesa/main/formats.c
@@ -26,7 +26,6 @@
 
 #include "imports.h"
 #include "formats.h"
-#include "config.h"
 
 
 /**
diff --git a/src/mesa/main/formats.h b/src/mesa/main/formats.h
index aa14185628..ad176caaa0 100644
--- a/src/mesa/main/formats.h
+++ b/src/mesa/main/formats.h
@@ -33,7 +33,7 @@
 #define FORMATS_H
 
 
-#include "main/mtypes.h"
+#include <GL/gl.h>
 
 
 
diff --git a/src/mesa/main/framebuffer.c b/src/mesa/main/framebuffer.c
index 56558cfcc1..e0aac26f62 100644
--- a/src/mesa/main/framebuffer.c
+++ b/src/mesa/main/framebuffer.c
@@ -75,7 +75,6 @@ compute_depth_max(struct gl_framebuffer *fb)
    fb->_MRD = (GLfloat)1.0 / fb->_DepthMaxF;
 }
 
-
 /**
  * Create and initialize a gl_framebuffer object.
  * This is intended for creating _window_system_ framebuffers, not generic
diff --git a/src/mesa/main/framebuffer.h b/src/mesa/main/framebuffer.h
index 1b6e3b1f0c..2e9844282f 100644
--- a/src/mesa/main/framebuffer.h
+++ b/src/mesa/main/framebuffer.h
@@ -26,6 +26,7 @@
 #ifndef FRAMEBUFFER_H
 #define FRAMEBUFFER_H
 
+#include "mtypes.h"
 
 extern struct gl_framebuffer *
 _mesa_create_framebuffer(const GLvisual *visual);
diff --git a/src/mesa/main/get.h b/src/mesa/main/get.h
index 320492b4ce..99a004b71d 100644
--- a/src/mesa/main/get.h
+++ b/src/mesa/main/get.h
@@ -32,7 +32,7 @@
 #define GET_H
 
 
-#include "mtypes.h"
+#include "glheader.h"
 
 
 extern void GLAPIENTRY
diff --git a/src/mesa/main/histogram.c b/src/mesa/main/histogram.c
index 3a65bb1926..4e482bcd54 100644
--- a/src/mesa/main/histogram.c
+++ b/src/mesa/main/histogram.c
@@ -29,6 +29,7 @@
 #include "context.h"
 #include "image.h"
 #include "histogram.h"
+#include "macros.h"
 #include "main/dispatch.h"
 
 
diff --git a/src/mesa/main/image.c b/src/mesa/main/image.c
index 63c28342f2..86aa6d0d70 100644
--- a/src/mesa/main/image.c
+++ b/src/mesa/main/image.c
@@ -32,7 +32,6 @@
 
 #include "glheader.h"
 #include "colormac.h"
-#include "context.h"
 #include "enums.h"
 #include "image.h"
 #include "imports.h"
diff --git a/src/mesa/main/mm.c b/src/mesa/main/mm.c
index 3ef38e94be..25a0293703 100644
--- a/src/mesa/main/mm.c
+++ b/src/mesa/main/mm.c
@@ -22,6 +22,11 @@
  *
  */
 
+#include <assert.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include "compiler.h"
 #include "mm.h"
 
 
diff --git a/src/mesa/main/mm.h b/src/mesa/main/mm.h
index df340808ac..228721ca2a 100644
--- a/src/mesa/main/mm.h
+++ b/src/mesa/main/mm.h
@@ -32,9 +32,6 @@
 #define MM_H
 
 
-#include "imports.h"
-
-
 struct mem_block {
    struct mem_block *next, *prev;
    struct mem_block *next_free, *prev_free;
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index f17529c6c7..e89f55befe 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -36,7 +36,6 @@
 
 #include "main/glheader.h"
 #include "main/config.h"
-#include "main/compiler.h"
 #include "main/mfeatures.h"
 #include "glapi/glapi.h"
 #include "math/m_matrix.h"	/* GLmatrix */
@@ -1808,6 +1807,11 @@ struct gl_program
    /** Which texture target is being sampled (TEXTURE_1D/2D/3D/etc_INDEX) */
    gl_texture_index SamplerTargets[MAX_SAMPLERS];
 
+   /** Bitmask of which register files are read/written with indirect
+    * addressing.  Mask of (1 << PROGRAM_x) bits.
+    */
+   GLbitfield IndirectRegisterFiles;
+
    /** Logical counts */
    /*@{*/
    GLuint NumInstructions;
diff --git a/src/mesa/main/multisample.h b/src/mesa/main/multisample.h
index 4305900cc4..998488ef42 100644
--- a/src/mesa/main/multisample.h
+++ b/src/mesa/main/multisample.h
@@ -26,6 +26,7 @@
 #ifndef MULTISAMPLE_H
 #define MULTISAMPLE_H
 
+#include "mtypes.h"
 
 extern void GLAPIENTRY
 _mesa_SampleCoverageARB(GLclampf value, GLboolean invert);
diff --git a/src/mesa/main/nvprogram.h b/src/mesa/main/nvprogram.h
index 8ee59661bd..260a25ba9e 100644
--- a/src/mesa/main/nvprogram.h
+++ b/src/mesa/main/nvprogram.h
@@ -29,6 +29,8 @@
 #ifndef NVPROGRAM_H
 #define NVPROGRAM_H
 
+#include "glheader.h"
+#include "mtypes.h"
 
 extern void GLAPIENTRY
 _mesa_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params);
diff --git a/src/mesa/main/pixelstore.h b/src/mesa/main/pixelstore.h
index ee963f9ba3..47bff4276d 100644
--- a/src/mesa/main/pixelstore.h
+++ b/src/mesa/main/pixelstore.h
@@ -33,6 +33,7 @@
 
 
 #include "glheader.h"
+#include "mtypes.h"
 
 
 extern void GLAPIENTRY
diff --git a/src/mesa/main/remap.h b/src/mesa/main/remap.h
index 7afdee36f5..a2a55f615d 100644
--- a/src/mesa/main/remap.h
+++ b/src/mesa/main/remap.h
@@ -28,6 +28,7 @@
 #define REMAP_H
 
 
+#include "main/compiler.h"
 #include "main/mfeatures.h"
 
 struct gl_function_pool_remap {
diff --git a/src/mesa/main/renderbuffer.h b/src/mesa/main/renderbuffer.h
index 7c205e141c..bc92b26982 100644
--- a/src/mesa/main/renderbuffer.h
+++ b/src/mesa/main/renderbuffer.h
@@ -26,6 +26,11 @@
 #ifndef RENDERBUFFER_H
 #define RENDERBUFFER_H
 
+#include "glheader.h"
+#include "mtypes.h"
+
+struct gl_framebuffer;
+struct gl_renderbuffer;
 
 extern void
 _mesa_init_renderbuffer(struct gl_renderbuffer *rb, GLuint name);
diff --git a/src/mesa/main/restart.h b/src/mesa/main/restart.h
index 931cd70128..25f58f24c3 100644
--- a/src/mesa/main/restart.h
+++ b/src/mesa/main/restart.h
@@ -28,6 +28,7 @@
 #ifndef RESTART_H
 #define RESTART_H
 
+#include "glheader.h"
 
 extern void GLAPIENTRY
 _mesa_PrimitiveRestart(void);
diff --git a/src/mesa/main/shared.c b/src/mesa/main/shared.c
index f9d10f3bbe..cbe004518a 100644
--- a/src/mesa/main/shared.c
+++ b/src/mesa/main/shared.c
@@ -32,7 +32,6 @@
 #include "imports.h"
 #include "mtypes.h"
 #include "hash.h"
-#include "arrayobj.h"
 #if FEATURE_ATI_fragment_shader
 #include "atifragshader.h"
 #endif
diff --git a/src/mesa/main/shared.h b/src/mesa/main/shared.h
index ef164a1459..5166a0ce51 100644
--- a/src/mesa/main/shared.h
+++ b/src/mesa/main/shared.h
@@ -25,6 +25,7 @@
 #ifndef SHARED_H
 #define SHARED_H
 
+#include "mtypes.h"
 
 struct gl_shared_state *
 _mesa_alloc_shared_state(GLcontext *ctx);
diff --git a/src/mesa/main/syncobj.h b/src/mesa/main/syncobj.h
index f23fa281e2..c53511995b 100644
--- a/src/mesa/main/syncobj.h
+++ b/src/mesa/main/syncobj.h
@@ -31,7 +31,10 @@
 #ifndef SYNCOBJ_H
 #define SYNCOBJ_H
 
-#include "context.h"
+#include "glheader.h"
+#include "mtypes.h"
+
+struct dd_function_table;
 
 extern void
 _mesa_init_sync_object_functions(struct dd_function_table *driver);
diff --git a/src/mesa/main/texcompress.c b/src/mesa/main/texcompress.c
index f4b1119eb1..e911524cbc 100644
--- a/src/mesa/main/texcompress.c
+++ b/src/mesa/main/texcompress.c
@@ -33,7 +33,6 @@
 #include "glheader.h"
 #include "imports.h"
 #include "colormac.h"
-#include "context.h"
 #include "formats.h"
 #include "texcompress.h"
 
diff --git a/src/mesa/main/texcompress_fxt1.c b/src/mesa/main/texcompress_fxt1.c
index 04acf05e52..c8b45bd3a5 100644
--- a/src/mesa/main/texcompress_fxt1.c
+++ b/src/mesa/main/texcompress_fxt1.c
@@ -32,9 +32,9 @@
 #include "glheader.h"
 #include "imports.h"
 #include "colormac.h"
-#include "context.h"
 #include "convolve.h"
 #include "image.h"
+#include "macros.h"
 #include "mipmap.h"
 #include "texcompress.h"
 #include "texcompress_fxt1.h"
diff --git a/src/mesa/main/texcompress_fxt1.h b/src/mesa/main/texcompress_fxt1.h
index d63ca71e21..38048b26cc 100644
--- a/src/mesa/main/texcompress_fxt1.h
+++ b/src/mesa/main/texcompress_fxt1.h
@@ -25,9 +25,11 @@
 #ifndef TEXCOMPRESS_FXT1_H
 #define TEXCOMPRESS_FXT1_H
 
-#include "main/mtypes.h"
+#include "glheader.h"
 #include "texstore.h"
 
+struct gl_texture_image;
+
 #if FEATURE_texture_fxt1
 
 extern GLboolean
diff --git a/src/mesa/main/texcompress_s3tc.c b/src/mesa/main/texcompress_s3tc.c
index 85c394b051..c70792cab6 100644
--- a/src/mesa/main/texcompress_s3tc.c
+++ b/src/mesa/main/texcompress_s3tc.c
@@ -36,10 +36,10 @@
 #include "glheader.h"
 #include "imports.h"
 #include "colormac.h"
-#include "context.h"
 #include "convolve.h"
 #include "dlopen.h"
 #include "image.h"
+#include "macros.h"
 #include "texcompress.h"
 #include "texcompress_s3tc.h"
 #include "texstore.h"
diff --git a/src/mesa/main/texfetch.c b/src/mesa/main/texfetch.c
index fe002082cc..c03bc71cd7 100644
--- a/src/mesa/main/texfetch.c
+++ b/src/mesa/main/texfetch.c
@@ -34,7 +34,7 @@
 
 
 #include "colormac.h"
-#include "context.h"
+#include "macros.h"
 #include "texcompress.h"
 #include "texcompress_fxt1.h"
 #include "texcompress_s3tc.h"
diff --git a/src/mesa/main/texgen.h b/src/mesa/main/texgen.h
index 397d89e630..2224a93761 100644
--- a/src/mesa/main/texgen.h
+++ b/src/mesa/main/texgen.h
@@ -27,7 +27,10 @@
 #define TEXGEN_H
 
 
-#include "main/mtypes.h"
+#include "compiler.h"
+#include "glheader.h"
+
+struct _glapi_table;
 
 
 #if FEATURE_texgen
diff --git a/src/mesa/main/texgetimage.h b/src/mesa/main/texgetimage.h
index 088d27c7e1..866ab70494 100644
--- a/src/mesa/main/texgetimage.h
+++ b/src/mesa/main/texgetimage.h
@@ -27,6 +27,7 @@
 #ifndef TEXGETIMAGE_H
 #define TEXGETIMAGE_H
 
+#include "mtypes.h"
 
 extern void
 _mesa_get_teximage(GLcontext *ctx, GLenum target, GLint level,
diff --git a/src/mesa/main/texrender.c b/src/mesa/main/texrender.c
index d29af5a5b2..c68105b395 100644
--- a/src/mesa/main/texrender.c
+++ b/src/mesa/main/texrender.c
@@ -1,6 +1,7 @@
 
 #include "context.h"
 #include "colormac.h"
+#include "macros.h"
 #include "texfetch.h"
 #include "texrender.h"
 #include "renderbuffer.h"
diff --git a/src/mesa/main/texrender.h b/src/mesa/main/texrender.h
index 7c3fb0871b..1e87d594a2 100644
--- a/src/mesa/main/texrender.h
+++ b/src/mesa/main/texrender.h
@@ -1,6 +1,7 @@
 #ifndef TEXRENDER_H
 #define TEXRENDER_H
 
+#include "mtypes.h"
 
 extern void
 _mesa_render_texture(GLcontext *ctx,
diff --git a/src/mesa/main/texstate.h b/src/mesa/main/texstate.h
index 17ac68000c..912cb67798 100644
--- a/src/mesa/main/texstate.h
+++ b/src/mesa/main/texstate.h
@@ -32,6 +32,7 @@
 #define TEXSTATE_H
 
 
+#include "compiler.h"
 #include "mtypes.h"
 
 
diff --git a/src/mesa/main/texstore.c b/src/mesa/main/texstore.c
index 0f21395af3..2989fdb72e 100644
--- a/src/mesa/main/texstore.c
+++ b/src/mesa/main/texstore.c
@@ -55,7 +55,6 @@
 #include "glheader.h"
 #include "bufferobj.h"
 #include "colormac.h"
-#include "context.h"
 #include "convolve.h"
 #include "image.h"
 #include "macros.h"
diff --git a/src/mesa/main/uniforms.h b/src/mesa/main/uniforms.h
index 29f77cb35a..ef98fe16bb 100644
--- a/src/mesa/main/uniforms.h
+++ b/src/mesa/main/uniforms.h
@@ -25,6 +25,10 @@
 #ifndef UNIFORMS_H
 #define UNIFORMS_H
 
+#include "glheader.h"
+
+struct gl_program;
+struct _glapi_table;
 
 extern void GLAPIENTRY
 _mesa_Uniform1fARB(GLint, GLfloat);
diff --git a/src/mesa/main/viewport.h b/src/mesa/main/viewport.h
index f08fef2797..ec054a7c59 100644
--- a/src/mesa/main/viewport.h
+++ b/src/mesa/main/viewport.h
@@ -27,6 +27,8 @@
 #ifndef VIEWPORT_H
 #define VIEWPORT_H
 
+#include "glheader.h"
+#include "mtypes.h"
 
 extern void GLAPIENTRY
 _mesa_Viewport(GLint x, GLint y, GLsizei width, GLsizei height);
diff --git a/src/mesa/main/vtxfmt.h b/src/mesa/main/vtxfmt.h
index fb6c23abe9..aad38b87c3 100644
--- a/src/mesa/main/vtxfmt.h
+++ b/src/mesa/main/vtxfmt.h
@@ -33,6 +33,9 @@
 #ifndef _VTXFMT_H_
 #define _VTXFMT_H_
 
+#include "compiler.h"
+#include "mtypes.h"
+
 #if FEATURE_beginend
 
 extern void _mesa_init_exec_vtxfmt( GLcontext *ctx );
diff --git a/src/mesa/math/m_matrix.h b/src/mesa/math/m_matrix.h
index 3bc5de6cd4..a69afb8589 100644
--- a/src/mesa/math/m_matrix.h
+++ b/src/mesa/math/m_matrix.h
@@ -32,6 +32,8 @@
 #define _M_MATRIX_H
 
 
+#include "main/glheader.h"
+
 
 /**
  * \name Symbolic names to some of the entries in the matrix
diff --git a/src/mesa/math/m_translate.c b/src/mesa/math/m_translate.c
index b12b07957c..51daf7bfd3 100644
--- a/src/mesa/math/m_translate.c
+++ b/src/mesa/math/m_translate.c
@@ -29,8 +29,8 @@
 
 
 #include "main/glheader.h"
+#include "main/macros.h"
 #include "main/mtypes.h"		/* GLchan hack */
-#include "main/colormac.h"
 
 #include "m_translate.h"
 
diff --git a/src/mesa/math/m_translate.h b/src/mesa/math/m_translate.h
index c677682d50..5804103116 100644
--- a/src/mesa/math/m_translate.h
+++ b/src/mesa/math/m_translate.h
@@ -26,7 +26,8 @@
 #ifndef _M_TRANSLATE_H_
 #define _M_TRANSLATE_H_
 
-#include "main/config.h"
+#include "main/compiler.h"
+#include "main/glheader.h"
 #include "main/mtypes.h"		/* hack for GLchan */
 
 
diff --git a/src/mesa/math/m_xform.h b/src/mesa/math/m_xform.h
index 33421ad1c0..14ac956a7b 100644
--- a/src/mesa/math/m_xform.h
+++ b/src/mesa/math/m_xform.h
@@ -27,10 +27,10 @@
 #define _M_XFORM_H
 
 
+#include "main/compiler.h"
 #include "main/glheader.h"
-#include "main/config.h"
-#include "math/m_vector.h"
 #include "math/m_matrix.h"
+#include "math/m_vector.h"
 
 #ifdef USE_X86_ASM
 #define _XFORMAPI _ASMAPI
diff --git a/src/mesa/program/arbprogparse.c b/src/mesa/program/arbprogparse.c
index 6373529e4e..f834aaf568 100644
--- a/src/mesa/program/arbprogparse.c
+++ b/src/mesa/program/arbprogparse.c
@@ -109,6 +109,7 @@ _mesa_parse_arb_fragment_program(GLcontext* ctx, GLenum target,
    program->Base.NumNativeTexIndirections = prog.NumTexIndirections;
    program->Base.InputsRead      = prog.InputsRead;
    program->Base.OutputsWritten  = prog.OutputsWritten;
+   program->Base.IndirectRegisterFiles = prog.IndirectRegisterFiles;
    for (i = 0; i < MAX_TEXTURE_IMAGE_UNITS; i++) {
       program->Base.TexturesUsed[i] = prog.TexturesUsed[i];
       if (prog.TexturesUsed[i])
@@ -199,6 +200,7 @@ _mesa_parse_arb_vertex_program(GLcontext *ctx, GLenum target,
    program->Base.NumNativeAddressRegs = prog.NumNativeAddressRegs;
    program->Base.InputsRead     = prog.InputsRead;
    program->Base.OutputsWritten = prog.OutputsWritten;
+   program->Base.IndirectRegisterFiles = prog.IndirectRegisterFiles;
    program->IsPositionInvariant = (state.option.PositionInvariant)
       ? GL_TRUE : GL_FALSE;
 
diff --git a/src/mesa/program/nvvertparse.c b/src/mesa/program/nvvertparse.c
index e2afcfd4ce..1ac83d0e59 100644
--- a/src/mesa/program/nvvertparse.c
+++ b/src/mesa/program/nvvertparse.c
@@ -64,6 +64,7 @@ struct parse_state {
    GLbitfield inputsRead;
    GLbitfield outputsWritten;
    GLboolean anyProgRegsWritten;
+   GLboolean indirectRegisterFiles;
    GLuint numInst;                 /* number of instructions parsed */
 };
 
@@ -410,6 +411,7 @@ Parse_ParamReg(struct parse_state *parseState, struct prog_src_register *srcReg)
 
       srcReg->RelAddr = GL_TRUE;
       srcReg->File = PROGRAM_ENV_PARAM;
+      parseState->indirectRegisterFiles |= (1 << srcReg->File);
       /* Look for +/-N offset */
       if (!Peek_Token(parseState, token))
          RETURN_ERROR;
@@ -1308,6 +1310,7 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget,
    parseState.inputsRead = 0;
    parseState.outputsWritten = 0;
    parseState.anyProgRegsWritten = GL_FALSE;
+   parseState.indirectRegisterFiles = 0x0;
 
    /* Reset error state */
    _mesa_set_program_error(ctx, -1, NULL);
@@ -1408,6 +1411,8 @@ _mesa_parse_nv_vertex_program(GLcontext *ctx, GLenum dstTarget,
       program->Base.Parameters = _mesa_new_parameter_list ();
       program->Base.NumParameters = 0;
 
+      program->Base.IndirectRegisterFiles = parseState.indirectRegisterFiles;
+
       state_tokens[0] = STATE_VERTEX_PROGRAM;
       state_tokens[1] = STATE_ENV;
       /* Add refs to all of the potential params, in order.  If we want to not
diff --git a/src/mesa/program/prog_execute.c b/src/mesa/program/prog_execute.c
index b6da3449b2..1222a0f4c8 100644
--- a/src/mesa/program/prog_execute.c
+++ b/src/mesa/program/prog_execute.c
@@ -37,7 +37,7 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
+#include "main/macros.h"
 #include "prog_execute.h"
 #include "prog_instruction.h"
 #include "prog_parameter.h"
@@ -81,6 +81,22 @@ static const GLfloat ZeroVec[4] = { 0.0F, 0.0F, 0.0F, 0.0F };
 
 
 /**
+ * Return TRUE for +0 and other positive values, FALSE otherwise.
+ * Used for RCC opcode.
+ */
+static INLINE GLboolean
+positive(float x)
+{
+   fi_type fi;
+   fi.f = x;
+   if (fi.i & 0x80000000)
+      return GL_FALSE;
+   return GL_TRUE;
+}
+
+
+
+/**
  * Return a pointer to the 4-element float vector specified by the given
  * source register.
  */
@@ -1340,6 +1356,44 @@ _mesa_execute_program(GLcontext * ctx,
             store_vector4(inst, machine, result);
          }
          break;
+      case OPCODE_RCC:  /* clamped riciprocal */
+         {
+            const float largest = 1.884467e+19, smallest = 5.42101e-20;
+            GLfloat a[4], r, result[4];
+            fetch_vector1(&inst->SrcReg[0], machine, a);
+            if (DEBUG_PROG) {
+               if (a[0] == 0)
+                  printf("RCC(0)\n");
+               else if (IS_INF_OR_NAN(a[0]))
+                  printf("RCC(inf)\n");
+            }
+            if (a[0] == 1.0F) {
+               r = 1.0F;
+            }
+            else {
+               r = 1.0F / a[0];
+            }
+            if (positive(r)) {
+               if (r > largest) {
+                  r = largest;
+               }
+               else if (r < smallest) {
+                  r = smallest;
+               }
+            }
+            else {
+               if (r < -largest) {
+                  r = -largest;
+               }
+               else if (r > -smallest) {
+                  r = -smallest;
+               }
+            }
+            result[0] = result[1] = result[2] = result[3] = r;
+            store_vector4(inst, machine, result);
+         }
+         break;
+
       case OPCODE_RCP:
          {
             GLfloat a[4], result[4];
diff --git a/src/mesa/program/prog_instruction.h b/src/mesa/program/prog_instruction.h
index 02df208945..1d850c4d5d 100644
--- a/src/mesa/program/prog_instruction.h
+++ b/src/mesa/program/prog_instruction.h
@@ -149,20 +149,20 @@ typedef enum prog_opcode {
    OPCODE_ADD,       /*   X        X       X       X         X   */
    OPCODE_AND,       /*                                          */
    OPCODE_ARA,       /*                    2                     */
-   OPCODE_ARL,       /*   X                X                     */
+   OPCODE_ARL,       /*   X                X                 X   */
    OPCODE_ARL_NV,    /*                    2                     */
    OPCODE_ARR,       /*                    2                     */
    OPCODE_BGNLOOP,   /*                                     opt  */
    OPCODE_BGNSUB,    /*                                     opt  */
    OPCODE_BRA,       /*                    2                 X   */
    OPCODE_BRK,       /*                    2                opt  */
-   OPCODE_CAL,       /*                    2       2             */
-   OPCODE_CMP,       /*            X                             */
+   OPCODE_CAL,       /*                    2       2         X   */
+   OPCODE_CMP,       /*            X                         X   */
    OPCODE_CONT,      /*                                     opt  */
    OPCODE_COS,       /*            X       2       X         X   */
    OPCODE_DDX,       /*                            X         X   */
    OPCODE_DDY,       /*                            X         X   */
-   OPCODE_DP2,       /*                            2             */
+   OPCODE_DP2,       /*                            2         X   */
    OPCODE_DP2A,      /*                            2             */
    OPCODE_DP3,       /*   X        X       X       X         X   */
    OPCODE_DP4,       /*   X        X       X       X         X   */
@@ -185,7 +185,7 @@ typedef enum prog_opcode {
    OPCODE_LG2,       /*   X        X       2       X         X   */
    OPCODE_LIT,       /*   X        X       X       X             */
    OPCODE_LOG,       /*   X                X                 X   */
-   OPCODE_LRP,       /*            X               X             */
+   OPCODE_LRP,       /*            X               X         X   */
    OPCODE_MAD,       /*   X        X       X       X         X   */
    OPCODE_MAX,       /*   X        X       X       X         X   */
    OPCODE_MIN,       /*   X        X       X       X         X   */
@@ -196,8 +196,8 @@ typedef enum prog_opcode {
    OPCODE_NOISE3,    /*                                      X   */
    OPCODE_NOISE4,    /*                                      X   */
    OPCODE_NOT,       /*                                          */
-   OPCODE_NRM3,      /*                                          */
-   OPCODE_NRM4,      /*                                          */
+   OPCODE_NRM3,      /*                                      X   */
+   OPCODE_NRM4,      /*                                      X   */
    OPCODE_OR,        /*                                          */
    OPCODE_PK2H,      /*                            X             */
    OPCODE_PK2US,     /*                            X             */
@@ -209,7 +209,7 @@ typedef enum prog_opcode {
    OPCODE_PUSHA,     /*                    3                     */
    OPCODE_RCC,       /*                    1.1                   */
    OPCODE_RCP,       /*   X        X       X       X         X   */
-   OPCODE_RET,       /*                    2       2             */
+   OPCODE_RET,       /*                    2       2         X   */
    OPCODE_RFL,       /*            X               X             */
    OPCODE_RSQ,       /*   X        X       X       X         X   */
    OPCODE_SCS,       /*            X                             */
diff --git a/src/mesa/program/prog_parameter_layout.c b/src/mesa/program/prog_parameter_layout.c
index a888573832..d7dc97edbf 100644
--- a/src/mesa/program/prog_parameter_layout.c
+++ b/src/mesa/program/prog_parameter_layout.c
@@ -28,6 +28,7 @@
  * \author Ian Romanick <ian.d.romanick@intel.com>
  */
 
+#include "main/compiler.h"
 #include "main/mtypes.h"
 #include "prog_parameter.h"
 #include "prog_parameter_layout.h"
diff --git a/src/mesa/program/prog_print.c b/src/mesa/program/prog_print.c
index 95db9b0eb3..00aa6de963 100644
--- a/src/mesa/program/prog_print.c
+++ b/src/mesa/program/prog_print.c
@@ -924,6 +924,8 @@ _mesa_fprint_program_parameters(FILE *f,
    fprintf(f, "NumParameters=%d\n", prog->NumParameters);
    fprintf(f, "NumAttributes=%d\n", prog->NumAttributes);
    fprintf(f, "NumAddressRegs=%d\n", prog->NumAddressRegs);
+   fprintf(f, "IndirectRegisterFiles: 0x%x (0b%s)\n",
+           prog->IndirectRegisterFiles, binary(prog->IndirectRegisterFiles));
    fprintf(f, "SamplersUsed: 0x%x (0b%s)\n",
                  prog->SamplersUsed, binary(prog->SamplersUsed));
    fprintf(f, "Samplers=[ ");
diff --git a/src/mesa/program/program.c b/src/mesa/program/program.c
index cf46095ce8..d99584d63b 100644
--- a/src/mesa/program/program.c
+++ b/src/mesa/program/program.c
@@ -512,6 +512,7 @@ _mesa_clone_program(GLcontext *ctx, const struct gl_program *prog)
    if (prog->Attributes)
       clone->Attributes = _mesa_clone_parameter_list(prog->Attributes);
    memcpy(clone->LocalParams, prog->LocalParams, sizeof(clone->LocalParams));
+   clone->IndirectRegisterFiles = prog->IndirectRegisterFiles;
    clone->NumInstructions = prog->NumInstructions;
    clone->NumTemporaries = prog->NumTemporaries;
    clone->NumParameters = prog->NumParameters;
diff --git a/src/mesa/program/program_parse.tab.c b/src/mesa/program/program_parse.tab.c
index 6421d1f58a..31a609600b 100644
--- a/src/mesa/program/program_parse.tab.c
+++ b/src/mesa/program/program_parse.tab.c
@@ -798,29 +798,29 @@ static const yytype_uint16 yyrline[] =
      415,   459,   464,   474,   518,   524,   525,   526,   527,   528,
      529,   530,   531,   532,   533,   534,   535,   538,   550,   558,
      575,   582,   601,   612,   632,   657,   664,   697,   704,   719,
-     774,   817,   826,   847,   857,   861,   890,   909,   909,   911,
-     918,   930,   931,   932,   935,   949,   963,   983,   994,  1006,
-    1008,  1009,  1010,  1011,  1014,  1014,  1014,  1014,  1015,  1018,
-    1022,  1027,  1034,  1041,  1048,  1071,  1094,  1095,  1096,  1097,
-    1098,  1099,  1102,  1121,  1125,  1131,  1135,  1139,  1143,  1152,
-    1161,  1165,  1170,  1176,  1187,  1187,  1188,  1190,  1194,  1198,
-    1202,  1208,  1208,  1210,  1228,  1254,  1257,  1268,  1274,  1280,
-    1281,  1288,  1294,  1300,  1308,  1314,  1320,  1328,  1334,  1340,
-    1348,  1349,  1352,  1353,  1354,  1355,  1356,  1357,  1358,  1359,
-    1360,  1361,  1362,  1365,  1374,  1378,  1382,  1388,  1397,  1401,
-    1405,  1414,  1418,  1424,  1430,  1437,  1442,  1450,  1460,  1462,
-    1470,  1476,  1480,  1484,  1490,  1501,  1510,  1514,  1519,  1523,
-    1527,  1531,  1537,  1544,  1548,  1554,  1562,  1573,  1580,  1584,
-    1590,  1600,  1611,  1615,  1633,  1642,  1645,  1651,  1655,  1659,
-    1665,  1676,  1681,  1686,  1691,  1696,  1701,  1709,  1712,  1717,
-    1730,  1738,  1749,  1757,  1757,  1759,  1759,  1761,  1771,  1776,
-    1783,  1793,  1802,  1807,  1814,  1824,  1834,  1846,  1846,  1847,
-    1847,  1849,  1859,  1867,  1877,  1885,  1893,  1902,  1913,  1917,
-    1923,  1924,  1925,  1928,  1928,  1931,  1966,  1970,  1970,  1973,
-    1980,  1989,  2003,  2012,  2021,  2025,  2034,  2043,  2054,  2061,
-    2066,  2075,  2087,  2090,  2099,  2110,  2111,  2112,  2115,  2116,
-    2117,  2120,  2121,  2124,  2125,  2128,  2129,  2132,  2143,  2154,
-    2165,  2191,  2192
+     774,   817,   826,   848,   858,   862,   891,   910,   910,   912,
+     919,   931,   932,   933,   936,   950,   964,   984,   995,  1007,
+    1009,  1010,  1011,  1012,  1015,  1015,  1015,  1015,  1016,  1019,
+    1023,  1028,  1035,  1042,  1049,  1072,  1095,  1096,  1097,  1098,
+    1099,  1100,  1103,  1122,  1126,  1132,  1136,  1140,  1144,  1153,
+    1162,  1166,  1171,  1177,  1188,  1188,  1189,  1191,  1195,  1199,
+    1203,  1209,  1209,  1211,  1229,  1255,  1258,  1269,  1275,  1281,
+    1282,  1289,  1295,  1301,  1309,  1315,  1321,  1329,  1335,  1341,
+    1349,  1350,  1353,  1354,  1355,  1356,  1357,  1358,  1359,  1360,
+    1361,  1362,  1363,  1366,  1375,  1379,  1383,  1389,  1398,  1402,
+    1406,  1415,  1419,  1425,  1431,  1438,  1443,  1451,  1461,  1463,
+    1471,  1477,  1481,  1485,  1491,  1502,  1511,  1515,  1520,  1524,
+    1528,  1532,  1538,  1545,  1549,  1555,  1563,  1574,  1581,  1585,
+    1591,  1601,  1612,  1616,  1634,  1643,  1646,  1652,  1656,  1660,
+    1666,  1677,  1682,  1687,  1692,  1697,  1702,  1710,  1713,  1718,
+    1731,  1739,  1750,  1758,  1758,  1760,  1760,  1762,  1772,  1777,
+    1784,  1794,  1803,  1808,  1815,  1825,  1835,  1847,  1847,  1848,
+    1848,  1850,  1860,  1868,  1878,  1886,  1894,  1903,  1914,  1918,
+    1924,  1925,  1926,  1929,  1929,  1932,  1967,  1971,  1971,  1974,
+    1981,  1990,  2004,  2013,  2022,  2026,  2035,  2044,  2055,  2062,
+    2067,  2076,  2088,  2091,  2100,  2111,  2112,  2113,  2116,  2117,
+    2118,  2121,  2122,  2125,  2126,  2129,  2130,  2133,  2144,  2155,
+    2166,  2192,  2193
 };
 #endif
 
@@ -2844,6 +2844,7 @@ yyreduce:
 	   (yyval.src_reg).Base.File = (yyvsp[(1) - (4)].sym)->param_binding_type;
 
 	   if ((yyvsp[(3) - (4)].src_reg).Base.RelAddr) {
+              state->prog->IndirectRegisterFiles |= (1 << (yyval.src_reg).Base.File);
 	      (yyvsp[(1) - (4)].sym)->param_accessed_indirectly = 1;
 
 	      (yyval.src_reg).Base.RelAddr = 1;
@@ -2858,7 +2859,7 @@ yyreduce:
   case 63:
 
 /* Line 1455 of yacc.c  */
-#line 848 "program_parse.y"
+#line 849 "program_parse.y"
     {
            gl_register_file file = ((yyvsp[(1) - (1)].temp_sym).name != NULL) 
 	      ? (yyvsp[(1) - (1)].temp_sym).param_binding_type
@@ -2871,7 +2872,7 @@ yyreduce:
   case 64:
 
 /* Line 1455 of yacc.c  */
-#line 858 "program_parse.y"
+#line 859 "program_parse.y"
     {
 	   set_dst_reg(& (yyval.dst_reg), PROGRAM_OUTPUT, (yyvsp[(1) - (1)].result));
 	;}
@@ -2880,7 +2881,7 @@ yyreduce:
   case 65:
 
 /* Line 1455 of yacc.c  */
-#line 862 "program_parse.y"
+#line 863 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2912,7 +2913,7 @@ yyreduce:
   case 66:
 
 /* Line 1455 of yacc.c  */
-#line 891 "program_parse.y"
+#line 892 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -2934,7 +2935,7 @@ yyreduce:
   case 69:
 
 /* Line 1455 of yacc.c  */
-#line 912 "program_parse.y"
+#line 913 "program_parse.y"
     {
 	   init_src_reg(& (yyval.src_reg));
 	   (yyval.src_reg).Base.Index = (yyvsp[(1) - (1)].integer);
@@ -2944,7 +2945,7 @@ yyreduce:
   case 70:
 
 /* Line 1455 of yacc.c  */
-#line 919 "program_parse.y"
+#line 920 "program_parse.y"
     {
 	   /* FINISHME: Add support for multiple address registers.
 	    */
@@ -2959,30 +2960,30 @@ yyreduce:
   case 71:
 
 /* Line 1455 of yacc.c  */
-#line 930 "program_parse.y"
+#line 931 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 72:
 
 /* Line 1455 of yacc.c  */
-#line 931 "program_parse.y"
+#line 932 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;}
     break;
 
   case 73:
 
 /* Line 1455 of yacc.c  */
-#line 932 "program_parse.y"
+#line 933 "program_parse.y"
     { (yyval.integer) = -(yyvsp[(2) - (2)].integer); ;}
     break;
 
   case 74:
 
 /* Line 1455 of yacc.c  */
-#line 936 "program_parse.y"
+#line 937 "program_parse.y"
     {
-	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 63)) {
+	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 4095)) {
               char s[100];
               _mesa_snprintf(s, sizeof(s),
                              "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer));
@@ -2997,9 +2998,9 @@ yyreduce:
   case 75:
 
 /* Line 1455 of yacc.c  */
-#line 950 "program_parse.y"
+#line 951 "program_parse.y"
     {
-	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 64)) {
+	   if (((yyvsp[(1) - (1)].integer) < 0) || ((yyvsp[(1) - (1)].integer) > 4096)) {
               char s[100];
               _mesa_snprintf(s, sizeof(s),
                              "relative address offset too large (%d)", (yyvsp[(1) - (1)].integer));
@@ -3014,7 +3015,7 @@ yyreduce:
   case 76:
 
 /* Line 1455 of yacc.c  */
-#line 964 "program_parse.y"
+#line 965 "program_parse.y"
     {
 	   struct asm_symbol *const s = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(1) - (1)].string));
@@ -3037,7 +3038,7 @@ yyreduce:
   case 77:
 
 /* Line 1455 of yacc.c  */
-#line 984 "program_parse.y"
+#line 985 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid address component selector");
@@ -3051,7 +3052,7 @@ yyreduce:
   case 78:
 
 /* Line 1455 of yacc.c  */
-#line 995 "program_parse.y"
+#line 996 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].swiz_mask).mask != WRITEMASK_X) {
 	      yyerror(& (yylsp[(1) - (1)]), state,
@@ -3066,21 +3067,21 @@ yyreduce:
   case 83:
 
 /* Line 1455 of yacc.c  */
-#line 1011 "program_parse.y"
+#line 1012 "program_parse.y"
     { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;}
     break;
 
   case 88:
 
 /* Line 1455 of yacc.c  */
-#line 1015 "program_parse.y"
+#line 1016 "program_parse.y"
     { (yyval.swiz_mask).swizzle = SWIZZLE_NOOP; (yyval.swiz_mask).mask = WRITEMASK_XYZW; ;}
     break;
 
   case 89:
 
 /* Line 1455 of yacc.c  */
-#line 1019 "program_parse.y"
+#line 1020 "program_parse.y"
     {
 	   (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg);
 	;}
@@ -3089,7 +3090,7 @@ yyreduce:
   case 90:
 
 /* Line 1455 of yacc.c  */
-#line 1023 "program_parse.y"
+#line 1024 "program_parse.y"
     {
 	   (yyval.dst_reg) = (yyvsp[(2) - (3)].dst_reg);
 	;}
@@ -3098,7 +3099,7 @@ yyreduce:
   case 91:
 
 /* Line 1455 of yacc.c  */
-#line 1027 "program_parse.y"
+#line 1028 "program_parse.y"
     {
 	   (yyval.dst_reg).CondMask = COND_TR;
 	   (yyval.dst_reg).CondSwizzle = SWIZZLE_NOOP;
@@ -3109,7 +3110,7 @@ yyreduce:
   case 92:
 
 /* Line 1455 of yacc.c  */
-#line 1035 "program_parse.y"
+#line 1036 "program_parse.y"
     {
 	   (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg);
 	   (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle;
@@ -3119,7 +3120,7 @@ yyreduce:
   case 93:
 
 /* Line 1455 of yacc.c  */
-#line 1042 "program_parse.y"
+#line 1043 "program_parse.y"
     {
 	   (yyval.dst_reg) = (yyvsp[(1) - (2)].dst_reg);
 	   (yyval.dst_reg).CondSwizzle = (yyvsp[(2) - (2)].swiz_mask).swizzle;
@@ -3129,7 +3130,7 @@ yyreduce:
   case 94:
 
 /* Line 1455 of yacc.c  */
-#line 1049 "program_parse.y"
+#line 1050 "program_parse.y"
     {
 	   const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string));
 	   if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) {
@@ -3155,7 +3156,7 @@ yyreduce:
   case 95:
 
 /* Line 1455 of yacc.c  */
-#line 1072 "program_parse.y"
+#line 1073 "program_parse.y"
     {
 	   const int cond = _mesa_parse_cc((yyvsp[(1) - (1)].string));
 	   if ((cond == 0) || ((yyvsp[(1) - (1)].string)[2] != '\0')) {
@@ -3181,7 +3182,7 @@ yyreduce:
   case 102:
 
 /* Line 1455 of yacc.c  */
-#line 1103 "program_parse.y"
+#line 1104 "program_parse.y"
     {
 	   struct asm_symbol *const s =
 	      declare_variable(state, (yyvsp[(2) - (4)].string), at_attrib, & (yylsp[(2) - (4)]));
@@ -3203,7 +3204,7 @@ yyreduce:
   case 103:
 
 /* Line 1455 of yacc.c  */
-#line 1122 "program_parse.y"
+#line 1123 "program_parse.y"
     {
 	   (yyval.attrib) = (yyvsp[(2) - (2)].attrib);
 	;}
@@ -3212,7 +3213,7 @@ yyreduce:
   case 104:
 
 /* Line 1455 of yacc.c  */
-#line 1126 "program_parse.y"
+#line 1127 "program_parse.y"
     {
 	   (yyval.attrib) = (yyvsp[(2) - (2)].attrib);
 	;}
@@ -3221,7 +3222,7 @@ yyreduce:
   case 105:
 
 /* Line 1455 of yacc.c  */
-#line 1132 "program_parse.y"
+#line 1133 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_POS;
 	;}
@@ -3230,7 +3231,7 @@ yyreduce:
   case 106:
 
 /* Line 1455 of yacc.c  */
-#line 1136 "program_parse.y"
+#line 1137 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_WEIGHT;
 	;}
@@ -3239,7 +3240,7 @@ yyreduce:
   case 107:
 
 /* Line 1455 of yacc.c  */
-#line 1140 "program_parse.y"
+#line 1141 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_NORMAL;
 	;}
@@ -3248,7 +3249,7 @@ yyreduce:
   case 108:
 
 /* Line 1455 of yacc.c  */
-#line 1144 "program_parse.y"
+#line 1145 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_secondary_color) {
 	      yyerror(& (yylsp[(2) - (2)]), state, "GL_EXT_secondary_color not supported");
@@ -3262,7 +3263,7 @@ yyreduce:
   case 109:
 
 /* Line 1455 of yacc.c  */
-#line 1153 "program_parse.y"
+#line 1154 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_fog_coord) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "GL_EXT_fog_coord not supported");
@@ -3276,7 +3277,7 @@ yyreduce:
   case 110:
 
 /* Line 1455 of yacc.c  */
-#line 1162 "program_parse.y"
+#line 1163 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer);
 	;}
@@ -3285,7 +3286,7 @@ yyreduce:
   case 111:
 
 /* Line 1455 of yacc.c  */
-#line 1166 "program_parse.y"
+#line 1167 "program_parse.y"
     {
 	   yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported");
 	   YYERROR;
@@ -3295,7 +3296,7 @@ yyreduce:
   case 112:
 
 /* Line 1455 of yacc.c  */
-#line 1171 "program_parse.y"
+#line 1172 "program_parse.y"
     {
 	   (yyval.attrib) = VERT_ATTRIB_GENERIC0 + (yyvsp[(3) - (4)].integer);
 	;}
@@ -3304,7 +3305,7 @@ yyreduce:
   case 113:
 
 /* Line 1455 of yacc.c  */
-#line 1177 "program_parse.y"
+#line 1178 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxAttribs) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid vertex attribute reference");
@@ -3318,7 +3319,7 @@ yyreduce:
   case 117:
 
 /* Line 1455 of yacc.c  */
-#line 1191 "program_parse.y"
+#line 1192 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_WPOS;
 	;}
@@ -3327,7 +3328,7 @@ yyreduce:
   case 118:
 
 /* Line 1455 of yacc.c  */
-#line 1195 "program_parse.y"
+#line 1196 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_COL0 + (yyvsp[(2) - (2)].integer);
 	;}
@@ -3336,7 +3337,7 @@ yyreduce:
   case 119:
 
 /* Line 1455 of yacc.c  */
-#line 1199 "program_parse.y"
+#line 1200 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_FOGC;
 	;}
@@ -3345,7 +3346,7 @@ yyreduce:
   case 120:
 
 /* Line 1455 of yacc.c  */
-#line 1203 "program_parse.y"
+#line 1204 "program_parse.y"
     {
 	   (yyval.attrib) = FRAG_ATTRIB_TEX0 + (yyvsp[(2) - (2)].integer);
 	;}
@@ -3354,7 +3355,7 @@ yyreduce:
   case 123:
 
 /* Line 1455 of yacc.c  */
-#line 1211 "program_parse.y"
+#line 1212 "program_parse.y"
     {
 	   struct asm_symbol *const s =
 	      declare_variable(state, (yyvsp[(2) - (3)].string), at_param, & (yylsp[(2) - (3)]));
@@ -3375,7 +3376,7 @@ yyreduce:
   case 124:
 
 /* Line 1455 of yacc.c  */
-#line 1229 "program_parse.y"
+#line 1230 "program_parse.y"
     {
 	   if (((yyvsp[(4) - (6)].integer) != 0) && ((unsigned) (yyvsp[(4) - (6)].integer) != (yyvsp[(6) - (6)].temp_sym).param_binding_length)) {
 	      free((yyvsp[(2) - (6)].string));
@@ -3403,7 +3404,7 @@ yyreduce:
   case 125:
 
 /* Line 1455 of yacc.c  */
-#line 1254 "program_parse.y"
+#line 1255 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
@@ -3412,7 +3413,7 @@ yyreduce:
   case 126:
 
 /* Line 1455 of yacc.c  */
-#line 1258 "program_parse.y"
+#line 1259 "program_parse.y"
     {
 	   if (((yyvsp[(1) - (1)].integer) < 1) || ((unsigned) (yyvsp[(1) - (1)].integer) > state->limits->MaxParameters)) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid parameter array size");
@@ -3426,7 +3427,7 @@ yyreduce:
   case 127:
 
 /* Line 1455 of yacc.c  */
-#line 1269 "program_parse.y"
+#line 1270 "program_parse.y"
     {
 	   (yyval.temp_sym) = (yyvsp[(2) - (2)].temp_sym);
 	;}
@@ -3435,7 +3436,7 @@ yyreduce:
   case 128:
 
 /* Line 1455 of yacc.c  */
-#line 1275 "program_parse.y"
+#line 1276 "program_parse.y"
     {
 	   (yyval.temp_sym) = (yyvsp[(3) - (4)].temp_sym);
 	;}
@@ -3444,7 +3445,7 @@ yyreduce:
   case 130:
 
 /* Line 1455 of yacc.c  */
-#line 1282 "program_parse.y"
+#line 1283 "program_parse.y"
     {
 	   (yyvsp[(1) - (3)].temp_sym).param_binding_length += (yyvsp[(3) - (3)].temp_sym).param_binding_length;
 	   (yyval.temp_sym) = (yyvsp[(1) - (3)].temp_sym);
@@ -3454,7 +3455,7 @@ yyreduce:
   case 131:
 
 /* Line 1455 of yacc.c  */
-#line 1289 "program_parse.y"
+#line 1290 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3465,7 +3466,7 @@ yyreduce:
   case 132:
 
 /* Line 1455 of yacc.c  */
-#line 1295 "program_parse.y"
+#line 1296 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3476,7 +3477,7 @@ yyreduce:
   case 133:
 
 /* Line 1455 of yacc.c  */
-#line 1301 "program_parse.y"
+#line 1302 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3487,7 +3488,7 @@ yyreduce:
   case 134:
 
 /* Line 1455 of yacc.c  */
-#line 1309 "program_parse.y"
+#line 1310 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3498,7 +3499,7 @@ yyreduce:
   case 135:
 
 /* Line 1455 of yacc.c  */
-#line 1315 "program_parse.y"
+#line 1316 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3509,7 +3510,7 @@ yyreduce:
   case 136:
 
 /* Line 1455 of yacc.c  */
-#line 1321 "program_parse.y"
+#line 1322 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3520,7 +3521,7 @@ yyreduce:
   case 137:
 
 /* Line 1455 of yacc.c  */
-#line 1329 "program_parse.y"
+#line 1330 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3531,7 +3532,7 @@ yyreduce:
   case 138:
 
 /* Line 1455 of yacc.c  */
-#line 1335 "program_parse.y"
+#line 1336 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3542,7 +3543,7 @@ yyreduce:
   case 139:
 
 /* Line 1455 of yacc.c  */
-#line 1341 "program_parse.y"
+#line 1342 "program_parse.y"
     {
 	   memset(& (yyval.temp_sym), 0, sizeof((yyval.temp_sym)));
 	   (yyval.temp_sym).param_binding_begin = ~0;
@@ -3553,98 +3554,98 @@ yyreduce:
   case 140:
 
 /* Line 1455 of yacc.c  */
-#line 1348 "program_parse.y"
+#line 1349 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(1) - (1)].state), sizeof((yyval.state))); ;}
     break;
 
   case 141:
 
 /* Line 1455 of yacc.c  */
-#line 1349 "program_parse.y"
+#line 1350 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 142:
 
 /* Line 1455 of yacc.c  */
-#line 1352 "program_parse.y"
+#line 1353 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 143:
 
 /* Line 1455 of yacc.c  */
-#line 1353 "program_parse.y"
+#line 1354 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 144:
 
 /* Line 1455 of yacc.c  */
-#line 1354 "program_parse.y"
+#line 1355 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 145:
 
 /* Line 1455 of yacc.c  */
-#line 1355 "program_parse.y"
+#line 1356 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 146:
 
 /* Line 1455 of yacc.c  */
-#line 1356 "program_parse.y"
+#line 1357 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 147:
 
 /* Line 1455 of yacc.c  */
-#line 1357 "program_parse.y"
+#line 1358 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 148:
 
 /* Line 1455 of yacc.c  */
-#line 1358 "program_parse.y"
+#line 1359 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 149:
 
 /* Line 1455 of yacc.c  */
-#line 1359 "program_parse.y"
+#line 1360 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 150:
 
 /* Line 1455 of yacc.c  */
-#line 1360 "program_parse.y"
+#line 1361 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 151:
 
 /* Line 1455 of yacc.c  */
-#line 1361 "program_parse.y"
+#line 1362 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 152:
 
 /* Line 1455 of yacc.c  */
-#line 1362 "program_parse.y"
+#line 1363 "program_parse.y"
     { memcpy((yyval.state), (yyvsp[(2) - (2)].state), sizeof((yyval.state))); ;}
     break;
 
   case 153:
 
 /* Line 1455 of yacc.c  */
-#line 1366 "program_parse.y"
+#line 1367 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_MATERIAL;
@@ -3656,7 +3657,7 @@ yyreduce:
   case 154:
 
 /* Line 1455 of yacc.c  */
-#line 1375 "program_parse.y"
+#line 1376 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
@@ -3665,7 +3666,7 @@ yyreduce:
   case 155:
 
 /* Line 1455 of yacc.c  */
-#line 1379 "program_parse.y"
+#line 1380 "program_parse.y"
     {
 	   (yyval.integer) = STATE_EMISSION;
 	;}
@@ -3674,7 +3675,7 @@ yyreduce:
   case 156:
 
 /* Line 1455 of yacc.c  */
-#line 1383 "program_parse.y"
+#line 1384 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SHININESS;
 	;}
@@ -3683,7 +3684,7 @@ yyreduce:
   case 157:
 
 /* Line 1455 of yacc.c  */
-#line 1389 "program_parse.y"
+#line 1390 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHT;
@@ -3695,7 +3696,7 @@ yyreduce:
   case 158:
 
 /* Line 1455 of yacc.c  */
-#line 1398 "program_parse.y"
+#line 1399 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
@@ -3704,7 +3705,7 @@ yyreduce:
   case 159:
 
 /* Line 1455 of yacc.c  */
-#line 1402 "program_parse.y"
+#line 1403 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POSITION;
 	;}
@@ -3713,7 +3714,7 @@ yyreduce:
   case 160:
 
 /* Line 1455 of yacc.c  */
-#line 1406 "program_parse.y"
+#line 1407 "program_parse.y"
     {
 	   if (!state->ctx->Extensions.EXT_point_parameters) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "GL_ARB_point_parameters not supported");
@@ -3727,7 +3728,7 @@ yyreduce:
   case 161:
 
 /* Line 1455 of yacc.c  */
-#line 1415 "program_parse.y"
+#line 1416 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(2) - (2)].integer);
 	;}
@@ -3736,7 +3737,7 @@ yyreduce:
   case 162:
 
 /* Line 1455 of yacc.c  */
-#line 1419 "program_parse.y"
+#line 1420 "program_parse.y"
     {
 	   (yyval.integer) = STATE_HALF_VECTOR;
 	;}
@@ -3745,7 +3746,7 @@ yyreduce:
   case 163:
 
 /* Line 1455 of yacc.c  */
-#line 1425 "program_parse.y"
+#line 1426 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SPOT_DIRECTION;
 	;}
@@ -3754,7 +3755,7 @@ yyreduce:
   case 164:
 
 /* Line 1455 of yacc.c  */
-#line 1431 "program_parse.y"
+#line 1432 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].state)[1];
@@ -3764,7 +3765,7 @@ yyreduce:
   case 165:
 
 /* Line 1455 of yacc.c  */
-#line 1438 "program_parse.y"
+#line 1439 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTMODEL_AMBIENT;
@@ -3774,7 +3775,7 @@ yyreduce:
   case 166:
 
 /* Line 1455 of yacc.c  */
-#line 1443 "program_parse.y"
+#line 1444 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTMODEL_SCENECOLOR;
@@ -3785,7 +3786,7 @@ yyreduce:
   case 167:
 
 /* Line 1455 of yacc.c  */
-#line 1451 "program_parse.y"
+#line 1452 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_LIGHTPROD;
@@ -3798,7 +3799,7 @@ yyreduce:
   case 169:
 
 /* Line 1455 of yacc.c  */
-#line 1463 "program_parse.y"
+#line 1464 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(3) - (3)].integer);
@@ -3809,7 +3810,7 @@ yyreduce:
   case 170:
 
 /* Line 1455 of yacc.c  */
-#line 1471 "program_parse.y"
+#line 1472 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXENV_COLOR;
 	;}
@@ -3818,7 +3819,7 @@ yyreduce:
   case 171:
 
 /* Line 1455 of yacc.c  */
-#line 1477 "program_parse.y"
+#line 1478 "program_parse.y"
     {
 	   (yyval.integer) = STATE_AMBIENT;
 	;}
@@ -3827,7 +3828,7 @@ yyreduce:
   case 172:
 
 /* Line 1455 of yacc.c  */
-#line 1481 "program_parse.y"
+#line 1482 "program_parse.y"
     {
 	   (yyval.integer) = STATE_DIFFUSE;
 	;}
@@ -3836,7 +3837,7 @@ yyreduce:
   case 173:
 
 /* Line 1455 of yacc.c  */
-#line 1485 "program_parse.y"
+#line 1486 "program_parse.y"
     {
 	   (yyval.integer) = STATE_SPECULAR;
 	;}
@@ -3845,7 +3846,7 @@ yyreduce:
   case 174:
 
 /* Line 1455 of yacc.c  */
-#line 1491 "program_parse.y"
+#line 1492 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxLights) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid light selector");
@@ -3859,7 +3860,7 @@ yyreduce:
   case 175:
 
 /* Line 1455 of yacc.c  */
-#line 1502 "program_parse.y"
+#line 1503 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_TEXGEN;
@@ -3871,7 +3872,7 @@ yyreduce:
   case 176:
 
 /* Line 1455 of yacc.c  */
-#line 1511 "program_parse.y"
+#line 1512 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_S;
 	;}
@@ -3880,7 +3881,7 @@ yyreduce:
   case 177:
 
 /* Line 1455 of yacc.c  */
-#line 1515 "program_parse.y"
+#line 1516 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_OBJECT_S;
 	;}
@@ -3889,7 +3890,7 @@ yyreduce:
   case 178:
 
 /* Line 1455 of yacc.c  */
-#line 1520 "program_parse.y"
+#line 1521 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_S - STATE_TEXGEN_EYE_S;
 	;}
@@ -3898,7 +3899,7 @@ yyreduce:
   case 179:
 
 /* Line 1455 of yacc.c  */
-#line 1524 "program_parse.y"
+#line 1525 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_T - STATE_TEXGEN_EYE_S;
 	;}
@@ -3907,7 +3908,7 @@ yyreduce:
   case 180:
 
 /* Line 1455 of yacc.c  */
-#line 1528 "program_parse.y"
+#line 1529 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_R - STATE_TEXGEN_EYE_S;
 	;}
@@ -3916,7 +3917,7 @@ yyreduce:
   case 181:
 
 /* Line 1455 of yacc.c  */
-#line 1532 "program_parse.y"
+#line 1533 "program_parse.y"
     {
 	   (yyval.integer) = STATE_TEXGEN_EYE_Q - STATE_TEXGEN_EYE_S;
 	;}
@@ -3925,7 +3926,7 @@ yyreduce:
   case 182:
 
 /* Line 1455 of yacc.c  */
-#line 1538 "program_parse.y"
+#line 1539 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].integer);
@@ -3935,7 +3936,7 @@ yyreduce:
   case 183:
 
 /* Line 1455 of yacc.c  */
-#line 1545 "program_parse.y"
+#line 1546 "program_parse.y"
     {
 	   (yyval.integer) = STATE_FOG_COLOR;
 	;}
@@ -3944,7 +3945,7 @@ yyreduce:
   case 184:
 
 /* Line 1455 of yacc.c  */
-#line 1549 "program_parse.y"
+#line 1550 "program_parse.y"
     {
 	   (yyval.integer) = STATE_FOG_PARAMS;
 	;}
@@ -3953,7 +3954,7 @@ yyreduce:
   case 185:
 
 /* Line 1455 of yacc.c  */
-#line 1555 "program_parse.y"
+#line 1556 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_CLIPPLANE;
@@ -3964,7 +3965,7 @@ yyreduce:
   case 186:
 
 /* Line 1455 of yacc.c  */
-#line 1563 "program_parse.y"
+#line 1564 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxClipPlanes) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid clip plane selector");
@@ -3978,7 +3979,7 @@ yyreduce:
   case 187:
 
 /* Line 1455 of yacc.c  */
-#line 1574 "program_parse.y"
+#line 1575 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = (yyvsp[(2) - (2)].integer);
@@ -3988,7 +3989,7 @@ yyreduce:
   case 188:
 
 /* Line 1455 of yacc.c  */
-#line 1581 "program_parse.y"
+#line 1582 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POINT_SIZE;
 	;}
@@ -3997,7 +3998,7 @@ yyreduce:
   case 189:
 
 /* Line 1455 of yacc.c  */
-#line 1585 "program_parse.y"
+#line 1586 "program_parse.y"
     {
 	   (yyval.integer) = STATE_POINT_ATTENUATION;
 	;}
@@ -4006,7 +4007,7 @@ yyreduce:
   case 190:
 
 /* Line 1455 of yacc.c  */
-#line 1591 "program_parse.y"
+#line 1592 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (5)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(1) - (5)].state)[1];
@@ -4019,7 +4020,7 @@ yyreduce:
   case 191:
 
 /* Line 1455 of yacc.c  */
-#line 1601 "program_parse.y"
+#line 1602 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (2)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(1) - (2)].state)[1];
@@ -4032,7 +4033,7 @@ yyreduce:
   case 192:
 
 /* Line 1455 of yacc.c  */
-#line 1611 "program_parse.y"
+#line 1612 "program_parse.y"
     {
 	   (yyval.state)[2] = 0;
 	   (yyval.state)[3] = 3;
@@ -4042,7 +4043,7 @@ yyreduce:
   case 193:
 
 /* Line 1455 of yacc.c  */
-#line 1616 "program_parse.y"
+#line 1617 "program_parse.y"
     {
 	   /* It seems logical that the matrix row range specifier would have
 	    * to specify a range or more than one row (i.e., $5 > $3).
@@ -4063,7 +4064,7 @@ yyreduce:
   case 194:
 
 /* Line 1455 of yacc.c  */
-#line 1634 "program_parse.y"
+#line 1635 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(2) - (3)].state)[0];
 	   (yyval.state)[1] = (yyvsp[(2) - (3)].state)[1];
@@ -4074,7 +4075,7 @@ yyreduce:
   case 195:
 
 /* Line 1455 of yacc.c  */
-#line 1642 "program_parse.y"
+#line 1643 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
@@ -4083,7 +4084,7 @@ yyreduce:
   case 196:
 
 /* Line 1455 of yacc.c  */
-#line 1646 "program_parse.y"
+#line 1647 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(1) - (1)].integer);
 	;}
@@ -4092,7 +4093,7 @@ yyreduce:
   case 197:
 
 /* Line 1455 of yacc.c  */
-#line 1652 "program_parse.y"
+#line 1653 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_INVERSE;
 	;}
@@ -4101,7 +4102,7 @@ yyreduce:
   case 198:
 
 /* Line 1455 of yacc.c  */
-#line 1656 "program_parse.y"
+#line 1657 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_TRANSPOSE;
 	;}
@@ -4110,7 +4111,7 @@ yyreduce:
   case 199:
 
 /* Line 1455 of yacc.c  */
-#line 1660 "program_parse.y"
+#line 1661 "program_parse.y"
     {
 	   (yyval.integer) = STATE_MATRIX_INVTRANS;
 	;}
@@ -4119,7 +4120,7 @@ yyreduce:
   case 200:
 
 /* Line 1455 of yacc.c  */
-#line 1666 "program_parse.y"
+#line 1667 "program_parse.y"
     {
 	   if ((yyvsp[(1) - (1)].integer) > 3) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid matrix row reference");
@@ -4133,7 +4134,7 @@ yyreduce:
   case 201:
 
 /* Line 1455 of yacc.c  */
-#line 1677 "program_parse.y"
+#line 1678 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_MODELVIEW_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].integer);
@@ -4143,7 +4144,7 @@ yyreduce:
   case 202:
 
 /* Line 1455 of yacc.c  */
-#line 1682 "program_parse.y"
+#line 1683 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_PROJECTION_MATRIX;
 	   (yyval.state)[1] = 0;
@@ -4153,7 +4154,7 @@ yyreduce:
   case 203:
 
 /* Line 1455 of yacc.c  */
-#line 1687 "program_parse.y"
+#line 1688 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_MVP_MATRIX;
 	   (yyval.state)[1] = 0;
@@ -4163,7 +4164,7 @@ yyreduce:
   case 204:
 
 /* Line 1455 of yacc.c  */
-#line 1692 "program_parse.y"
+#line 1693 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_TEXTURE_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(2) - (2)].integer);
@@ -4173,7 +4174,7 @@ yyreduce:
   case 205:
 
 /* Line 1455 of yacc.c  */
-#line 1697 "program_parse.y"
+#line 1698 "program_parse.y"
     {
 	   yyerror(& (yylsp[(1) - (4)]), state, "GL_ARB_matrix_palette not supported");
 	   YYERROR;
@@ -4183,7 +4184,7 @@ yyreduce:
   case 206:
 
 /* Line 1455 of yacc.c  */
-#line 1702 "program_parse.y"
+#line 1703 "program_parse.y"
     {
 	   (yyval.state)[0] = STATE_PROGRAM_MATRIX;
 	   (yyval.state)[1] = (yyvsp[(3) - (4)].integer);
@@ -4193,7 +4194,7 @@ yyreduce:
   case 207:
 
 /* Line 1455 of yacc.c  */
-#line 1709 "program_parse.y"
+#line 1710 "program_parse.y"
     {
 	   (yyval.integer) = 0;
 	;}
@@ -4202,7 +4203,7 @@ yyreduce:
   case 208:
 
 /* Line 1455 of yacc.c  */
-#line 1713 "program_parse.y"
+#line 1714 "program_parse.y"
     {
 	   (yyval.integer) = (yyvsp[(2) - (3)].integer);
 	;}
@@ -4211,7 +4212,7 @@ yyreduce:
   case 209:
 
 /* Line 1455 of yacc.c  */
-#line 1718 "program_parse.y"
+#line 1719 "program_parse.y"
     {
 	   /* Since GL_ARB_vertex_blend isn't supported, only modelview matrix
 	    * zero is valid.
@@ -4228,7 +4229,7 @@ yyreduce:
   case 210:
 
 /* Line 1455 of yacc.c  */
-#line 1731 "program_parse.y"
+#line 1732 "program_parse.y"
     {
 	   /* Since GL_ARB_matrix_palette isn't supported, just let any value
 	    * through here.  The error will be generated later.
@@ -4240,7 +4241,7 @@ yyreduce:
   case 211:
 
 /* Line 1455 of yacc.c  */
-#line 1739 "program_parse.y"
+#line 1740 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxProgramMatrices) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid program matrix selector");
@@ -4254,7 +4255,7 @@ yyreduce:
   case 212:
 
 /* Line 1455 of yacc.c  */
-#line 1750 "program_parse.y"
+#line 1751 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = STATE_DEPTH_RANGE;
@@ -4264,7 +4265,7 @@ yyreduce:
   case 217:
 
 /* Line 1455 of yacc.c  */
-#line 1762 "program_parse.y"
+#line 1763 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4277,7 +4278,7 @@ yyreduce:
   case 218:
 
 /* Line 1455 of yacc.c  */
-#line 1772 "program_parse.y"
+#line 1773 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (1)].integer);
 	   (yyval.state)[1] = (yyvsp[(1) - (1)].integer);
@@ -4287,7 +4288,7 @@ yyreduce:
   case 219:
 
 /* Line 1455 of yacc.c  */
-#line 1777 "program_parse.y"
+#line 1778 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (3)].integer);
 	   (yyval.state)[1] = (yyvsp[(3) - (3)].integer);
@@ -4297,7 +4298,7 @@ yyreduce:
   case 220:
 
 /* Line 1455 of yacc.c  */
-#line 1784 "program_parse.y"
+#line 1785 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4310,7 +4311,7 @@ yyreduce:
   case 221:
 
 /* Line 1455 of yacc.c  */
-#line 1794 "program_parse.y"
+#line 1795 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4323,7 +4324,7 @@ yyreduce:
   case 222:
 
 /* Line 1455 of yacc.c  */
-#line 1803 "program_parse.y"
+#line 1804 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (1)].integer);
 	   (yyval.state)[1] = (yyvsp[(1) - (1)].integer);
@@ -4333,7 +4334,7 @@ yyreduce:
   case 223:
 
 /* Line 1455 of yacc.c  */
-#line 1808 "program_parse.y"
+#line 1809 "program_parse.y"
     {
 	   (yyval.state)[0] = (yyvsp[(1) - (3)].integer);
 	   (yyval.state)[1] = (yyvsp[(3) - (3)].integer);
@@ -4343,7 +4344,7 @@ yyreduce:
   case 224:
 
 /* Line 1455 of yacc.c  */
-#line 1815 "program_parse.y"
+#line 1816 "program_parse.y"
     {
 	   memset((yyval.state), 0, sizeof((yyval.state)));
 	   (yyval.state)[0] = state->state_param_enum;
@@ -4356,7 +4357,7 @@ yyreduce:
   case 225:
 
 /* Line 1455 of yacc.c  */
-#line 1825 "program_parse.y"
+#line 1826 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxEnvParams) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid environment parameter reference");
@@ -4369,7 +4370,7 @@ yyreduce:
   case 226:
 
 /* Line 1455 of yacc.c  */
-#line 1835 "program_parse.y"
+#line 1836 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->limits->MaxLocalParams) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid local parameter reference");
@@ -4382,7 +4383,7 @@ yyreduce:
   case 231:
 
 /* Line 1455 of yacc.c  */
-#line 1850 "program_parse.y"
+#line 1851 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(1) - (1)].real);
@@ -4395,7 +4396,7 @@ yyreduce:
   case 232:
 
 /* Line 1455 of yacc.c  */
-#line 1860 "program_parse.y"
+#line 1861 "program_parse.y"
     {
 	   (yyval.vector).count = 1;
 	   (yyval.vector).data[0] = (yyvsp[(1) - (1)].real);
@@ -4408,7 +4409,7 @@ yyreduce:
   case 233:
 
 /* Line 1455 of yacc.c  */
-#line 1868 "program_parse.y"
+#line 1869 "program_parse.y"
     {
 	   (yyval.vector).count = 1;
 	   (yyval.vector).data[0] = (float) (yyvsp[(1) - (1)].integer);
@@ -4421,7 +4422,7 @@ yyreduce:
   case 234:
 
 /* Line 1455 of yacc.c  */
-#line 1878 "program_parse.y"
+#line 1879 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (3)].real);
@@ -4434,7 +4435,7 @@ yyreduce:
   case 235:
 
 /* Line 1455 of yacc.c  */
-#line 1886 "program_parse.y"
+#line 1887 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (5)].real);
@@ -4447,7 +4448,7 @@ yyreduce:
   case 236:
 
 /* Line 1455 of yacc.c  */
-#line 1895 "program_parse.y"
+#line 1896 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (7)].real);
@@ -4460,7 +4461,7 @@ yyreduce:
   case 237:
 
 /* Line 1455 of yacc.c  */
-#line 1904 "program_parse.y"
+#line 1905 "program_parse.y"
     {
 	   (yyval.vector).count = 4;
 	   (yyval.vector).data[0] = (yyvsp[(2) - (9)].real);
@@ -4473,7 +4474,7 @@ yyreduce:
   case 238:
 
 /* Line 1455 of yacc.c  */
-#line 1914 "program_parse.y"
+#line 1915 "program_parse.y"
     {
 	   (yyval.real) = ((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].real) : (yyvsp[(2) - (2)].real);
 	;}
@@ -4482,7 +4483,7 @@ yyreduce:
   case 239:
 
 /* Line 1455 of yacc.c  */
-#line 1918 "program_parse.y"
+#line 1919 "program_parse.y"
     {
 	   (yyval.real) = (float)(((yyvsp[(1) - (2)].negate)) ? -(yyvsp[(2) - (2)].integer) : (yyvsp[(2) - (2)].integer));
 	;}
@@ -4491,35 +4492,35 @@ yyreduce:
   case 240:
 
 /* Line 1455 of yacc.c  */
-#line 1923 "program_parse.y"
+#line 1924 "program_parse.y"
     { (yyval.negate) = FALSE; ;}
     break;
 
   case 241:
 
 /* Line 1455 of yacc.c  */
-#line 1924 "program_parse.y"
+#line 1925 "program_parse.y"
     { (yyval.negate) = TRUE;  ;}
     break;
 
   case 242:
 
 /* Line 1455 of yacc.c  */
-#line 1925 "program_parse.y"
+#line 1926 "program_parse.y"
     { (yyval.negate) = FALSE; ;}
     break;
 
   case 243:
 
 /* Line 1455 of yacc.c  */
-#line 1928 "program_parse.y"
+#line 1929 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (2)].integer); ;}
     break;
 
   case 245:
 
 /* Line 1455 of yacc.c  */
-#line 1932 "program_parse.y"
+#line 1933 "program_parse.y"
     {
 	   /* NV_fragment_program_option defines the size qualifiers in a
 	    * fairly broken way.  "SHORT" or "LONG" can optionally be used
@@ -4558,7 +4559,7 @@ yyreduce:
   case 246:
 
 /* Line 1455 of yacc.c  */
-#line 1966 "program_parse.y"
+#line 1967 "program_parse.y"
     {
 	;}
     break;
@@ -4566,14 +4567,14 @@ yyreduce:
   case 247:
 
 /* Line 1455 of yacc.c  */
-#line 1970 "program_parse.y"
+#line 1971 "program_parse.y"
     { (yyval.integer) = (yyvsp[(1) - (1)].integer); ;}
     break;
 
   case 249:
 
 /* Line 1455 of yacc.c  */
-#line 1974 "program_parse.y"
+#line 1975 "program_parse.y"
     {
 	   if (!declare_variable(state, (yyvsp[(3) - (3)].string), (yyvsp[(0) - (3)].integer), & (yylsp[(3) - (3)]))) {
 	      free((yyvsp[(3) - (3)].string));
@@ -4585,7 +4586,7 @@ yyreduce:
   case 250:
 
 /* Line 1455 of yacc.c  */
-#line 1981 "program_parse.y"
+#line 1982 "program_parse.y"
     {
 	   if (!declare_variable(state, (yyvsp[(1) - (1)].string), (yyvsp[(0) - (1)].integer), & (yylsp[(1) - (1)]))) {
 	      free((yyvsp[(1) - (1)].string));
@@ -4597,7 +4598,7 @@ yyreduce:
   case 251:
 
 /* Line 1455 of yacc.c  */
-#line 1990 "program_parse.y"
+#line 1991 "program_parse.y"
     {
 	   struct asm_symbol *const s =
 	      declare_variable(state, (yyvsp[(3) - (5)].string), at_output, & (yylsp[(3) - (5)]));
@@ -4614,7 +4615,7 @@ yyreduce:
   case 252:
 
 /* Line 1455 of yacc.c  */
-#line 2004 "program_parse.y"
+#line 2005 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_HPOS;
@@ -4628,7 +4629,7 @@ yyreduce:
   case 253:
 
 /* Line 1455 of yacc.c  */
-#line 2013 "program_parse.y"
+#line 2014 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_FOGC;
@@ -4642,7 +4643,7 @@ yyreduce:
   case 254:
 
 /* Line 1455 of yacc.c  */
-#line 2022 "program_parse.y"
+#line 2023 "program_parse.y"
     {
 	   (yyval.result) = (yyvsp[(2) - (2)].result);
 	;}
@@ -4651,7 +4652,7 @@ yyreduce:
   case 255:
 
 /* Line 1455 of yacc.c  */
-#line 2026 "program_parse.y"
+#line 2027 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_PSIZ;
@@ -4665,7 +4666,7 @@ yyreduce:
   case 256:
 
 /* Line 1455 of yacc.c  */
-#line 2035 "program_parse.y"
+#line 2036 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.result) = VERT_RESULT_TEX0 + (yyvsp[(3) - (3)].integer);
@@ -4679,7 +4680,7 @@ yyreduce:
   case 257:
 
 /* Line 1455 of yacc.c  */
-#line 2044 "program_parse.y"
+#line 2045 "program_parse.y"
     {
 	   if (state->mode == ARB_fragment) {
 	      (yyval.result) = FRAG_RESULT_DEPTH;
@@ -4693,7 +4694,7 @@ yyreduce:
   case 258:
 
 /* Line 1455 of yacc.c  */
-#line 2055 "program_parse.y"
+#line 2056 "program_parse.y"
     {
 	   (yyval.result) = (yyvsp[(2) - (3)].integer) + (yyvsp[(3) - (3)].integer);
 	;}
@@ -4702,7 +4703,7 @@ yyreduce:
   case 259:
 
 /* Line 1455 of yacc.c  */
-#line 2061 "program_parse.y"
+#line 2062 "program_parse.y"
     {
 	   (yyval.integer) = (state->mode == ARB_vertex)
 	      ? VERT_RESULT_COL0
@@ -4713,7 +4714,7 @@ yyreduce:
   case 260:
 
 /* Line 1455 of yacc.c  */
-#line 2067 "program_parse.y"
+#line 2068 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = VERT_RESULT_COL0;
@@ -4727,7 +4728,7 @@ yyreduce:
   case 261:
 
 /* Line 1455 of yacc.c  */
-#line 2076 "program_parse.y"
+#line 2077 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = VERT_RESULT_BFC0;
@@ -4741,7 +4742,7 @@ yyreduce:
   case 262:
 
 /* Line 1455 of yacc.c  */
-#line 2087 "program_parse.y"
+#line 2088 "program_parse.y"
     {
 	   (yyval.integer) = 0; 
 	;}
@@ -4750,7 +4751,7 @@ yyreduce:
   case 263:
 
 /* Line 1455 of yacc.c  */
-#line 2091 "program_parse.y"
+#line 2092 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = 0;
@@ -4764,7 +4765,7 @@ yyreduce:
   case 264:
 
 /* Line 1455 of yacc.c  */
-#line 2100 "program_parse.y"
+#line 2101 "program_parse.y"
     {
 	   if (state->mode == ARB_vertex) {
 	      (yyval.integer) = 1;
@@ -4778,91 +4779,91 @@ yyreduce:
   case 265:
 
 /* Line 1455 of yacc.c  */
-#line 2110 "program_parse.y"
+#line 2111 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 266:
 
 /* Line 1455 of yacc.c  */
-#line 2111 "program_parse.y"
+#line 2112 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 267:
 
 /* Line 1455 of yacc.c  */
-#line 2112 "program_parse.y"
+#line 2113 "program_parse.y"
     { (yyval.integer) = 1; ;}
     break;
 
   case 268:
 
 /* Line 1455 of yacc.c  */
-#line 2115 "program_parse.y"
+#line 2116 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 269:
 
 /* Line 1455 of yacc.c  */
-#line 2116 "program_parse.y"
+#line 2117 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 270:
 
 /* Line 1455 of yacc.c  */
-#line 2117 "program_parse.y"
+#line 2118 "program_parse.y"
     { (yyval.integer) = 1; ;}
     break;
 
   case 271:
 
 /* Line 1455 of yacc.c  */
-#line 2120 "program_parse.y"
+#line 2121 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 272:
 
 /* Line 1455 of yacc.c  */
-#line 2121 "program_parse.y"
+#line 2122 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
   case 273:
 
 /* Line 1455 of yacc.c  */
-#line 2124 "program_parse.y"
+#line 2125 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 274:
 
 /* Line 1455 of yacc.c  */
-#line 2125 "program_parse.y"
+#line 2126 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
   case 275:
 
 /* Line 1455 of yacc.c  */
-#line 2128 "program_parse.y"
+#line 2129 "program_parse.y"
     { (yyval.integer) = 0; ;}
     break;
 
   case 276:
 
 /* Line 1455 of yacc.c  */
-#line 2129 "program_parse.y"
+#line 2130 "program_parse.y"
     { (yyval.integer) = (yyvsp[(2) - (3)].integer); ;}
     break;
 
   case 277:
 
 /* Line 1455 of yacc.c  */
-#line 2133 "program_parse.y"
+#line 2134 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureCoordUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture coordinate unit selector");
@@ -4876,7 +4877,7 @@ yyreduce:
   case 278:
 
 /* Line 1455 of yacc.c  */
-#line 2144 "program_parse.y"
+#line 2145 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureImageUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture image unit selector");
@@ -4890,7 +4891,7 @@ yyreduce:
   case 279:
 
 /* Line 1455 of yacc.c  */
-#line 2155 "program_parse.y"
+#line 2156 "program_parse.y"
     {
 	   if ((unsigned) (yyvsp[(1) - (1)].integer) >= state->MaxTextureUnits) {
 	      yyerror(& (yylsp[(1) - (1)]), state, "invalid texture unit selector");
@@ -4904,7 +4905,7 @@ yyreduce:
   case 280:
 
 /* Line 1455 of yacc.c  */
-#line 2166 "program_parse.y"
+#line 2167 "program_parse.y"
     {
 	   struct asm_symbol *exist = (struct asm_symbol *)
 	      _mesa_symbol_table_find_symbol(state->st, 0, (yyvsp[(2) - (4)].string));
@@ -4933,7 +4934,7 @@ yyreduce:
 
 
 /* Line 1455 of yacc.c  */
-#line 4937 "program_parse.tab.c"
+#line 4938 "program_parse.tab.c"
       default: break;
     }
   YY_SYMBOL_PRINT ("-> $$ =", yyr1[yyn], &yyval, &yyloc);
@@ -5152,7 +5153,7 @@ yyreturn:
 
 
 /* Line 1675 of yacc.c  */
-#line 2195 "program_parse.y"
+#line 2196 "program_parse.y"
 
 
 void
diff --git a/src/mesa/program/program_parse.y b/src/mesa/program/program_parse.y
index 861927c744..fb6ef85a9f 100644
--- a/src/mesa/program/program_parse.y
+++ b/src/mesa/program/program_parse.y
@@ -835,6 +835,7 @@ srcReg: USED_IDENTIFIER /* temporaryReg | progParamSingle */
 	   $$.Base.File = $1->param_binding_type;
 
 	   if ($3.Base.RelAddr) {
+              state->prog->IndirectRegisterFiles |= (1 << $$.Base.File);
 	      $1->param_accessed_indirectly = 1;
 
 	      $$.Base.RelAddr = 1;
@@ -934,7 +935,7 @@ addrRegRelOffset:              { $$ = 0; }
 
 addrRegPosOffset: INTEGER
 	{
-	   if (($1 < 0) || ($1 > 63)) {
+	   if (($1 < 0) || ($1 > 4095)) {
               char s[100];
               _mesa_snprintf(s, sizeof(s),
                              "relative address offset too large (%d)", $1);
@@ -948,7 +949,7 @@ addrRegPosOffset: INTEGER
 
 addrRegNegOffset: INTEGER
 	{
-	   if (($1 < 0) || ($1 > 64)) {
+	   if (($1 < 0) || ($1 > 4096)) {
               char s[100];
               _mesa_snprintf(s, sizeof(s),
                              "relative address offset too large (%d)", $1);
diff --git a/src/mesa/slang/library/slang_common_builtin.gc b/src/mesa/slang/library/slang_common_builtin.gc
index d75354deff..1f5ddbc1ee 100644
--- a/src/mesa/slang/library/slang_common_builtin.gc
+++ b/src/mesa/slang/library/slang_common_builtin.gc
@@ -411,7 +411,7 @@ float atan(const float y, const float x)
    if (abs(x) > 1.0e-4) {
       r = atan(y / x);
       if (x < 0.0) {
-         r = r + sign(y) * 3.141593;
+         r = r + 3.141593 - 6.283186 * float(y < 0.0);
       }
    }
    else {
diff --git a/src/mesa/slang/slang_link.c b/src/mesa/slang/slang_link.c
index 00c2c13cc6..94db6f918d 100644
--- a/src/mesa/slang/slang_link.c
+++ b/src/mesa/slang/slang_link.c
@@ -756,6 +756,8 @@ _slang_update_inputs_outputs(struct gl_program *prog)
    prog->InputsRead = 0x0;
    prog->OutputsWritten = 0x0;
 
+   prog->IndirectRegisterFiles = 0x0;
+
    for (i = 0; i < prog->NumInstructions; i++) {
       const struct prog_instruction *inst = prog->Instructions + i;
       const GLuint numSrc = _mesa_num_inst_src_regs(inst->Opcode);
@@ -774,6 +776,9 @@ _slang_update_inputs_outputs(struct gl_program *prog)
          else if (inst->SrcReg[j].File == PROGRAM_ADDRESS) {
             maxAddrReg = MAX2(maxAddrReg, (GLuint) (inst->SrcReg[j].Index + 1));
          }
+
+         if (inst->SrcReg[j].RelAddr)
+            prog->IndirectRegisterFiles |= (1 << inst->SrcReg[j].File);
       }
 
       if (inst->DstReg.File == PROGRAM_OUTPUT) {
@@ -784,6 +789,8 @@ _slang_update_inputs_outputs(struct gl_program *prog)
       else if (inst->DstReg.File == PROGRAM_ADDRESS) {
          maxAddrReg = MAX2(maxAddrReg, inst->DstReg.Index + 1);
       }
+      if (inst->DstReg.RelAddr)
+         prog->IndirectRegisterFiles |= (1 << inst->DstReg.File);
    }
    prog->NumAddressRegs = maxAddrReg;
 }
diff --git a/src/mesa/state_tracker/st_atom.h b/src/mesa/state_tracker/st_atom.h
index 1f0fef63df..c7a04951bf 100644
--- a/src/mesa/state_tracker/st_atom.h
+++ b/src/mesa/state_tracker/st_atom.h
@@ -34,6 +34,8 @@
 #ifndef ST_ATOM_H
 #define ST_ATOM_H
 
+#include "main/glheader.h"
+
 struct st_context;
 struct st_tracked_state;
 
diff --git a/src/mesa/state_tracker/st_atom_constbuf.h b/src/mesa/state_tracker/st_atom_constbuf.h
index f707534e2c..97b076629e 100644
--- a/src/mesa/state_tracker/st_atom_constbuf.h
+++ b/src/mesa/state_tracker/st_atom_constbuf.h
@@ -29,6 +29,9 @@
 #ifndef ST_ATOM_CONSTBUF_H
 #define ST_ATOM_CONSTBUF_H
 
+struct gl_program_parameter_list;
+struct st_context;
+
 
 void st_upload_constants( struct st_context *st,
                           struct gl_program_parameter_list *params,
diff --git a/src/mesa/state_tracker/st_atom_depth.c b/src/mesa/state_tracker/st_atom_depth.c
index 3c07afba9a..1616e945fe 100644
--- a/src/mesa/state_tracker/st_atom_depth.c
+++ b/src/mesa/state_tracker/st_atom_depth.c
@@ -33,6 +33,8 @@
   */
  
 
+#include <assert.h>
+
 #include "st_context.h"
 #include "st_atom.h"
 #include "pipe/p_context.h"
diff --git a/src/mesa/state_tracker/st_atom_pixeltransfer.c b/src/mesa/state_tracker/st_atom_pixeltransfer.c
index b88c74fa03..8a8d17599e 100644
--- a/src/mesa/state_tracker/st_atom_pixeltransfer.c
+++ b/src/mesa/state_tracker/st_atom_pixeltransfer.c
@@ -37,6 +37,7 @@
 #include "main/image.h"
 #include "main/macros.h"
 #include "program/program.h"
+#include "program/prog_cache.h"
 #include "program/prog_instruction.h"
 #include "program/prog_parameter.h"
 #include "program/prog_print.h"
diff --git a/src/mesa/state_tracker/st_atom_shader.h b/src/mesa/state_tracker/st_atom_shader.h
index 8403bc66c9..56d4c68f4f 100644
--- a/src/mesa/state_tracker/st_atom_shader.h
+++ b/src/mesa/state_tracker/st_atom_shader.h
@@ -30,6 +30,9 @@
 #define ST_ATOM_SHADER_H
 
 
+struct st_context;
+struct translated_vertex_program;
+
 extern void
 st_free_translated_vertex_programs(struct st_context *st,
                                    struct translated_vertex_program *xvp);
diff --git a/src/mesa/state_tracker/st_atom_stipple.c b/src/mesa/state_tracker/st_atom_stipple.c
index 31e124b329..ecdd9f06f6 100644
--- a/src/mesa/state_tracker/st_atom_stipple.c
+++ b/src/mesa/state_tracker/st_atom_stipple.c
@@ -33,6 +33,8 @@
   */
  
 
+#include <assert.h>
+
 #include "st_context.h"
 #include "st_atom.h"
 #include "pipe/p_context.h"
diff --git a/src/mesa/state_tracker/st_cache.h b/src/mesa/state_tracker/st_cache.h
index b81de316ec..6d5de7b13a 100644
--- a/src/mesa/state_tracker/st_cache.h
+++ b/src/mesa/state_tracker/st_cache.h
@@ -33,10 +33,11 @@
 #ifndef ST_CACHE_H
 #define ST_CACHE_H
 
-#include "cso_cache/cso_cache.h"
-
 struct pipe_blend_state;
+struct pipe_depth_stencil_alpha_state;
+struct pipe_rasterizer_state;
 struct pipe_sampler_state;
+struct pipe_shader_state;
 struct st_context;
 
 
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index ba600ccef6..0b8ecd27cb 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -46,6 +46,7 @@
 
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
 #include "util/u_inlines.h"
 #include "util/u_draw_quad.h"
 #include "util/u_simple_shaders.h"
diff --git a/src/mesa/state_tracker/st_cb_bitmap.h b/src/mesa/state_tracker/st_cb_bitmap.h
index 8af975b74f..d04b2b6779 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.h
+++ b/src/mesa/state_tracker/st_cb_bitmap.h
@@ -30,7 +30,10 @@
 #define ST_CB_BITMAP_H
 
 
-#include "main/mtypes.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
+struct st_context;
 
 #if FEATURE_drawpix
 
diff --git a/src/mesa/state_tracker/st_cb_blit.h b/src/mesa/state_tracker/st_cb_blit.h
index 7ab9a54df9..c230652cef 100644
--- a/src/mesa/state_tracker/st_cb_blit.h
+++ b/src/mesa/state_tracker/st_cb_blit.h
@@ -29,8 +29,10 @@
 #define ST_CB_BLIT_H
 
 
-#include "main/mtypes.h"
-#include "st_context.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
+struct st_context;
 
 
 extern void
diff --git a/src/mesa/state_tracker/st_cb_bufferobjects.h b/src/mesa/state_tracker/st_cb_bufferobjects.h
index a27daac2bf..1c991d2083 100644
--- a/src/mesa/state_tracker/st_cb_bufferobjects.h
+++ b/src/mesa/state_tracker/st_cb_bufferobjects.h
@@ -28,9 +28,12 @@
 #ifndef ST_CB_BUFFEROBJECTS_H
 #define ST_CB_BUFFEROBJECTS_H
 
-struct st_context;
-struct gl_buffer_object;
+#include "main/compiler.h"
+#include "main/mtypes.h"
+
+struct dd_function_table;
 struct pipe_resource;
+struct st_context;
 
 /**
  * State_tracker vertex/pixel buffer object, derived from Mesa's
diff --git a/src/mesa/state_tracker/st_cb_clear.c b/src/mesa/state_tracker/st_cb_clear.c
index ea2414c4a0..246ab2e957 100644
--- a/src/mesa/state_tracker/st_cb_clear.c
+++ b/src/mesa/state_tracker/st_cb_clear.c
@@ -45,6 +45,7 @@
 #include "st_program.h"
 
 #include "pipe/p_context.h"
+#include "pipe/p_shader_tokens.h"
 #include "pipe/p_state.h"
 #include "pipe/p_defines.h"
 #include "util/u_format.h"
diff --git a/src/mesa/state_tracker/st_cb_clear.h b/src/mesa/state_tracker/st_cb_clear.h
index bc035ac25c..b27c09d10e 100644
--- a/src/mesa/state_tracker/st_cb_clear.h
+++ b/src/mesa/state_tracker/st_cb_clear.h
@@ -30,6 +30,9 @@
 #define ST_CB_CLEAR_H
 
 
+struct dd_function_table;
+struct st_context;
+
 extern void
 st_init_clear(struct st_context *st);
 
diff --git a/src/mesa/state_tracker/st_cb_condrender.h b/src/mesa/state_tracker/st_cb_condrender.h
index 891f1cbcd8..79d0db8d08 100644
--- a/src/mesa/state_tracker/st_cb_condrender.h
+++ b/src/mesa/state_tracker/st_cb_condrender.h
@@ -29,6 +29,8 @@
 #define ST_CB_CONDRENDER_H
 
 
+struct dd_function_table;
+
 extern void st_init_cond_render_functions(struct dd_function_table *functions);
 
 
diff --git a/src/mesa/state_tracker/st_cb_drawpixels.h b/src/mesa/state_tracker/st_cb_drawpixels.h
index 7d5e901ccc..575f169e08 100644
--- a/src/mesa/state_tracker/st_cb_drawpixels.h
+++ b/src/mesa/state_tracker/st_cb_drawpixels.h
@@ -30,7 +30,10 @@
 #define ST_CB_DRAWPIXELS_H
 
 
-#include "main/mtypes.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
+struct st_context;
 
 #if FEATURE_drawpix
 
diff --git a/src/mesa/state_tracker/st_cb_drawtex.c b/src/mesa/state_tracker/st_cb_drawtex.c
index b191a7f890..c99a8d792e 100644
--- a/src/mesa/state_tracker/st_cb_drawtex.c
+++ b/src/mesa/state_tracker/st_cb_drawtex.c
@@ -14,7 +14,6 @@
 
 #include "main/imports.h"
 #include "main/image.h"
-#include "main/bufferobj.h"
 #include "main/macros.h"
 #include "program/program.h"
 #include "program/prog_print.h"
diff --git a/src/mesa/state_tracker/st_cb_drawtex.h b/src/mesa/state_tracker/st_cb_drawtex.h
index a3f54a349c..d21262f897 100644
--- a/src/mesa/state_tracker/st_cb_drawtex.h
+++ b/src/mesa/state_tracker/st_cb_drawtex.h
@@ -10,7 +10,10 @@
 #define ST_CB_DRAWTEX_H
 
 
-#include "main/mtypes.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
+struct st_context;
 
 #if FEATURE_OES_draw_texture
 
diff --git a/src/mesa/state_tracker/st_cb_eglimage.c b/src/mesa/state_tracker/st_cb_eglimage.c
index 4aaf91d5a1..037e576fab 100644
--- a/src/mesa/state_tracker/st_cb_eglimage.c
+++ b/src/mesa/state_tracker/st_cb_eglimage.c
@@ -33,6 +33,7 @@
 #include "util/u_format.h"
 #include "st_cb_eglimage.h"
 #include "st_cb_fbo.h"
+#include "st_context.h"
 #include "st_texture.h"
 #include "st_format.h"
 #include "st_manager.h"
diff --git a/src/mesa/state_tracker/st_cb_eglimage.h b/src/mesa/state_tracker/st_cb_eglimage.h
index d6953e99f6..b6e44d5aff 100644
--- a/src/mesa/state_tracker/st_cb_eglimage.h
+++ b/src/mesa/state_tracker/st_cb_eglimage.h
@@ -29,8 +29,9 @@
 #ifndef ST_CB_EGLIMAGE_H
 #define ST_CB_EGLIMAGE_H
 
-#include "main/mtypes.h"
-#include "main/dd.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
 
 #if FEATURE_OES_EGL_image
 
diff --git a/src/mesa/state_tracker/st_cb_fbo.h b/src/mesa/state_tracker/st_cb_fbo.h
index 43b6c1e75f..62a9bbcb25 100644
--- a/src/mesa/state_tracker/st_cb_fbo.h
+++ b/src/mesa/state_tracker/st_cb_fbo.h
@@ -29,6 +29,15 @@
 #ifndef ST_CB_FBO_H
 #define ST_CB_FBO_H
 
+#include "main/compiler.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+
+#include "pipe/p_compiler.h"
+#include "pipe/p_format.h"
+
+struct dd_function_table;
+struct pipe_context;
 
 /**
  * Derived renderbuffer class.  Just need to add a pointer to the
diff --git a/src/mesa/state_tracker/st_cb_feedback.h b/src/mesa/state_tracker/st_cb_feedback.h
index 706d84960f..f2342f5823 100644
--- a/src/mesa/state_tracker/st_cb_feedback.h
+++ b/src/mesa/state_tracker/st_cb_feedback.h
@@ -30,7 +30,9 @@
 #define ST_CB_FEEDBACK_H
 
 
-#include "main/mtypes.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
 
 #if FEATURE_feedback
 
diff --git a/src/mesa/state_tracker/st_cb_flush.h b/src/mesa/state_tracker/st_cb_flush.h
index 7fca0176a3..7672b4cf1d 100644
--- a/src/mesa/state_tracker/st_cb_flush.h
+++ b/src/mesa/state_tracker/st_cb_flush.h
@@ -30,6 +30,12 @@
 #define ST_CB_FLUSH_H
 
 
+#include "pipe/p_compiler.h"
+
+struct dd_function_table;
+struct pipe_fence_handle;
+struct st_context;
+
 extern void
 st_init_flush_functions(struct dd_function_table *functions);
 
diff --git a/src/mesa/state_tracker/st_cb_program.h b/src/mesa/state_tracker/st_cb_program.h
index 0de96f2fd2..0fd179ef3d 100644
--- a/src/mesa/state_tracker/st_cb_program.h
+++ b/src/mesa/state_tracker/st_cb_program.h
@@ -29,6 +29,10 @@
 #define ST_CB_PROGRAM_H
 
 
+#include "main/mtypes.h"
+
+struct dd_function_table;
+
 extern void
 st_init_program_functions(struct dd_function_table *functions);
 
diff --git a/src/mesa/state_tracker/st_cb_rasterpos.h b/src/mesa/state_tracker/st_cb_rasterpos.h
index d2ed7297f1..2dc109bb18 100644
--- a/src/mesa/state_tracker/st_cb_rasterpos.h
+++ b/src/mesa/state_tracker/st_cb_rasterpos.h
@@ -29,7 +29,9 @@
 #define ST_CB_RASTERPOS_H
 
 
-#include "main/mtypes.h"
+#include "main/compiler.h"
+
+struct dd_function_table;
 
 #if FEATURE_rastpos
 
diff --git a/src/mesa/state_tracker/st_cb_readpixels.h b/src/mesa/state_tracker/st_cb_readpixels.h
index c90ef02906..9e1f7b4925 100644
--- a/src/mesa/state_tracker/st_cb_readpixels.h
+++ b/src/mesa/state_tracker/st_cb_readpixels.h
@@ -29,6 +29,10 @@
 #ifndef ST_CB_READPIXELS_H
 #define ST_CB_READPIXELS_H
 
+#include "main/mtypes.h"
+
+struct dd_function_table;
+
 extern struct st_renderbuffer *
 st_get_color_read_renderbuffer(GLcontext *ctx);
 
diff --git a/src/mesa/state_tracker/st_cb_strings.h b/src/mesa/state_tracker/st_cb_strings.h
index 3b765aaa59..92d5d2d9ba 100644
--- a/src/mesa/state_tracker/st_cb_strings.h
+++ b/src/mesa/state_tracker/st_cb_strings.h
@@ -30,6 +30,8 @@
 #define ST_CB_STRINGS_H
 
 
+struct dd_function_table;
+
 extern void
 st_init_string_functions(struct dd_function_table *functions);
 
diff --git a/src/mesa/state_tracker/st_cb_texture.h b/src/mesa/state_tracker/st_cb_texture.h
index 1cd9fc3a50..6942478e81 100644
--- a/src/mesa/state_tracker/st_cb_texture.h
+++ b/src/mesa/state_tracker/st_cb_texture.h
@@ -30,6 +30,13 @@
 #define ST_CB_TEXTURE_H
 
 
+#include "main/glheader.h"
+#include "main/mtypes.h"
+
+struct dd_function_table;
+struct pipe_context;
+struct st_context;
+
 extern GLboolean
 st_finalize_texture(GLcontext *ctx,
 		    struct pipe_context *pipe, 
diff --git a/src/mesa/state_tracker/st_cb_viewport.h b/src/mesa/state_tracker/st_cb_viewport.h
index db7dd6eab8..bcfd7cb68a 100644
--- a/src/mesa/state_tracker/st_cb_viewport.h
+++ b/src/mesa/state_tracker/st_cb_viewport.h
@@ -25,5 +25,12 @@
  *
  **************************************************************************/
 
+#ifndef ST_CB_VIEWPORT_H
+#define ST_CB_VIEWPORT_H
+
+struct dd_function_table;
+
 extern void
 st_init_viewport_functions(struct dd_function_table *functions);
+
+#endif /* ST_CB_VIEW_PORT_H */
diff --git a/src/mesa/state_tracker/st_cb_xformfb.h b/src/mesa/state_tracker/st_cb_xformfb.h
index 50efcb9293..574cf481e1 100644
--- a/src/mesa/state_tracker/st_cb_xformfb.h
+++ b/src/mesa/state_tracker/st_cb_xformfb.h
@@ -29,6 +29,10 @@
 #define ST_CB_XFORMFB_H
 
 
+#include "main/compiler.h"
+
+struct dd_function_table;
+
 #if FEATURE_EXT_transform_feedback
 
 extern void
diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c
index 7eb5f32611..3eb4708688 100644
--- a/src/mesa/state_tracker/st_context.c
+++ b/src/mesa/state_tracker/st_context.c
@@ -28,6 +28,7 @@
 #include "main/imports.h"
 #include "main/context.h"
 #include "main/shaderobj.h"
+#include "program/prog_cache.h"
 #include "vbo/vbo.h"
 #include "glapi/glapi.h"
 #include "st_context.h"
@@ -62,6 +63,9 @@
 #include "cso_cache/cso_context.h"
 
 
+DEBUG_GET_ONCE_BOOL_OPTION(mesa_mvp_dp4, "MESA_MVP_DP4", FALSE);
+
+
 /**
  * Called via ctx->Driver.UpdateState()
  */
@@ -169,7 +173,7 @@ struct st_context *st_create_context(gl_api api, struct pipe_context *pipe,
    /* XXX: need a capability bit in gallium to query if the pipe
     * driver prefers DP4 or MUL/MAD for vertex transformation.
     */
-   if (debug_get_bool_option("MESA_MVP_DP4", FALSE))
+   if (debug_get_option_mesa_mvp_dp4())
       _mesa_set_mvp_with_dp4( ctx, GL_TRUE );
 
    return st_create_context_priv(ctx, pipe);
diff --git a/src/mesa/state_tracker/st_context.h b/src/mesa/state_tracker/st_context.h
index a147a02117..60c25fb8f0 100644
--- a/src/mesa/state_tracker/st_context.h
+++ b/src/mesa/state_tracker/st_context.h
@@ -1,3 +1,4 @@
+//struct dd_function_table;
 /**************************************************************************
  * 
  * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
@@ -29,21 +30,17 @@
 #define ST_CONTEXT_H
 
 #include "main/mtypes.h"
-#include "program/prog_cache.h"
 #include "pipe/p_state.h"
 #include "state_tracker/st_api.h"
 
-
-struct st_context;
-struct st_texture_object;
-struct st_fragment_program;
+struct bitmap_cache;
+struct blit_state;
+struct dd_function_table;
 struct draw_context;
 struct draw_stage;
-struct cso_cache;
-struct cso_blend;
 struct gen_mipmap_state;
-struct blit_state;
-struct bitmap_cache;
+struct st_context;
+struct st_fragment_program;
 
 
 #define ST_NEW_MESA                    0x1 /* Mesa state has changed */
diff --git a/src/mesa/state_tracker/st_debug.c b/src/mesa/state_tracker/st_debug.c
index ebf6ec6e7e..1260fe04b1 100644
--- a/src/mesa/state_tracker/st_debug.c
+++ b/src/mesa/state_tracker/st_debug.c
@@ -55,6 +55,8 @@ static const struct debug_named_value st_debug_flags[] = {
    { "query",    DEBUG_QUERY, NULL },
    DEBUG_NAMED_VALUE_END
 };
+
+DEBUG_GET_ONCE_FLAGS_OPTION(st_debug, "ST_DEBUG", st_debug_flags, 0);
 #endif
 
 
@@ -62,7 +64,7 @@ void
 st_debug_init(void)
 {
 #ifdef DEBUG
-   ST_DEBUG = debug_get_flags_option("ST_DEBUG", st_debug_flags, 0 );
+   ST_DEBUG = debug_get_option_st_debug();
 #endif
 }
 
diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c
index 5821da4889..5b05489270 100644
--- a/src/mesa/state_tracker/st_draw.c
+++ b/src/mesa/state_tracker/st_draw.c
@@ -58,6 +58,7 @@
 #include "util/u_inlines.h"
 #include "util/u_format.h"
 #include "util/u_prim.h"
+#include "util/u_draw_quad.h"
 #include "draw/draw_context.h"
 #include "cso_cache/cso_context.h"
 
@@ -494,6 +495,49 @@ setup_non_interleaved_attribs(GLcontext *ctx,
 }
 
 
+static void
+setup_index_buffer(GLcontext *ctx,
+                   const struct _mesa_index_buffer *ib,
+                   struct pipe_index_buffer *ibuffer)
+{
+   struct st_context *st = st_context(ctx);
+   struct pipe_context *pipe = st->pipe;
+
+   memset(ibuffer, 0, sizeof(*ibuffer));
+   if (ib) {
+      struct gl_buffer_object *bufobj = ib->obj;
+
+      switch (ib->type) {
+      case GL_UNSIGNED_INT:
+         ibuffer->index_size = 4;
+         break;
+      case GL_UNSIGNED_SHORT:
+         ibuffer->index_size = 2;
+         break;
+      case GL_UNSIGNED_BYTE:
+         ibuffer->index_size = 1;
+         break;
+      default:
+         assert(0);
+	 return;
+      }
+
+      /* get/create the index buffer object */
+      if (bufobj && bufobj->Name) {
+         /* elements/indexes are in a real VBO */
+         struct st_buffer_object *stobj = st_buffer_object(bufobj);
+         pipe_resource_reference(&ibuffer->buffer, stobj->buffer);
+         ibuffer->offset = pointer_to_offset(ib->ptr);
+      }
+      else {
+         /* element/indicies are in user space memory */
+         ibuffer->buffer =
+            pipe_user_buffer_create(pipe->screen, (void *) ib->ptr,
+                                    ib->count * ibuffer->index_size,
+                                    PIPE_BIND_INDEX_BUFFER);
+      }
+   }
+}
 
 /**
  * Prior to drawing, check that any uniforms referenced by the
@@ -568,8 +612,11 @@ st_draw_vbo(GLcontext *ctx,
    GLuint attr;
    struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS];
    unsigned num_vbuffers, num_velements;
+   struct pipe_index_buffer ibuffer;
    GLboolean userSpace = GL_FALSE;
    GLboolean vertDataEdgeFlags;
+   struct pipe_draw_info info;
+   unsigned i;
 
    /* Mesa core state should have been validated already */
    assert(ctx->NewState == 0x0);
@@ -647,113 +694,35 @@ st_draw_vbo(GLcontext *ctx,
    if (num_vbuffers == 0 || num_velements == 0)
       return;
 
-   /* do actual drawing */
-   if (ib) {
-      /* indexed primitive */
-      struct gl_buffer_object *bufobj = ib->obj;
-      struct pipe_resource *indexBuf = NULL;
-      unsigned indexSize, indexOffset, i;
+   setup_index_buffer(ctx, ib, &ibuffer);
+   pipe->set_index_buffer(pipe, &ibuffer);
 
-      switch (ib->type) {
-      case GL_UNSIGNED_INT:
-         indexSize = 4;
-         break;
-      case GL_UNSIGNED_SHORT:
-         indexSize = 2;
-         break;
-      case GL_UNSIGNED_BYTE:
-         indexSize = 1;
-         break;
-      default:
-         assert(0);
-	 return;
-      }
-
-      /* get/create the index buffer object */
-      if (bufobj && bufobj->Name) {
-         /* elements/indexes are in a real VBO */
-         struct st_buffer_object *stobj = st_buffer_object(bufobj);
-         pipe_resource_reference(&indexBuf, stobj->buffer);
-         indexOffset = pointer_to_offset(ib->ptr) / indexSize;
-      }
-      else {
-         /* element/indicies are in user space memory */
-         indexBuf = pipe_user_buffer_create(pipe->screen, (void *) ib->ptr,
-                                            ib->count * indexSize,
-					    PIPE_BIND_INDEX_BUFFER);
-         indexOffset = 0;
+   util_draw_init_info(&info);
+   if (ib) {
+      info.indexed = TRUE;
+      if (min_index != ~0 && max_index != ~0) {
+         info.min_index = min_index;
+         info.max_index = max_index;
       }
+   }
 
-      /* draw */
-      if (pipe->draw_range_elements && min_index != ~0 && max_index != ~0) {
-         /* XXX: exercise temporary path to pass min/max directly
-          * through to driver & draw module.  These interfaces still
-          * need a bit of work...
-          */
-         for (i = 0; i < nr_prims; i++) {
-            unsigned vcount = prims[i].count;
-            unsigned prim = translate_prim(ctx, prims[i].mode);
-
-            if (u_trim_pipe_prim(prims[i].mode, &vcount)) {
-               pipe->draw_range_elements(pipe, indexBuf, indexSize,
-                                         prims[i].basevertex,
-                                         min_index, max_index, prim,
-                                         prims[i].start + indexOffset, vcount);
-            }
-         }
-      }
-      else {
-         for (i = 0; i < nr_prims; i++) {
-            unsigned vcount = prims[i].count;
-            unsigned prim = translate_prim(ctx, prims[i].mode);
-            
-            if (u_trim_pipe_prim(prims[i].mode, &vcount)) {
-               if (prims[i].num_instances == 1) {
-                  pipe->draw_elements(pipe, indexBuf,
-                                      indexSize,
-                                      prims[i].basevertex,
-                                      prim,
-                                      prims[i].start + indexOffset,
-                                      vcount);
-               }
-               else {
-                  pipe->draw_elements_instanced(pipe, indexBuf,
-                                                indexSize,
-                                                prims[i].basevertex,
-                                                prim,
-                                                prims[i].start + indexOffset,
-                                                vcount,
-                                                0, /* startInstance */
-                                                prims[i].num_instances);
-               }
-            }
-         }
+   /* do actual drawing */
+   for (i = 0; i < nr_prims; i++) {
+      info.mode = translate_prim( ctx, prims[i].mode );
+      info.start = prims[i].start;
+      info.count = prims[i].count;
+      info.instance_count = prims[i].num_instances;
+      info.index_bias = prims[i].basevertex;
+      if (!ib) {
+         info.min_index = info.start;
+         info.max_index = info.start + info.count - 1;
       }
 
-      pipe_resource_reference(&indexBuf, NULL);
+      if (u_trim_pipe_prim(info.mode, &info.count))
+         pipe->draw_vbo(pipe, &info);
    }
-   else {
-      /* non-indexed */
-      GLuint i;
-
-      for (i = 0; i < nr_prims; i++) {
-         unsigned vcount = prims[i].count;
-         unsigned prim = translate_prim(ctx, prims[i].mode);
 
-         if (u_trim_pipe_prim(prims[i].mode, &vcount)) {
-            if (prims[i].num_instances == 1) {
-               pipe->draw_arrays(pipe, prim, prims[i].start, vcount);
-            }
-            else {
-               pipe->draw_arrays_instanced(pipe, prim,
-                                           prims[i].start,
-                                           vcount,
-                                           0, /* startInstance */
-                                           prims[i].num_instances);
-            }
-         }
-      }
-   }
+   pipe_resource_reference(&ibuffer.buffer, NULL);
 
    /* unreference buffers (frees wrapped user-space buffer objects) */
    for (attr = 0; attr < num_vbuffers; attr++) {
diff --git a/src/mesa/state_tracker/st_draw.h b/src/mesa/state_tracker/st_draw.h
index 3e0face656..f36184487a 100644
--- a/src/mesa/state_tracker/st_draw.h
+++ b/src/mesa/state_tracker/st_draw.h
@@ -34,8 +34,13 @@
 #ifndef ST_DRAW_H
 #define ST_DRAW_H
 
-struct _mesa_prim;
+#include "main/compiler.h"
+#include "main/glheader.h"
+#include "main/mtypes.h"
+
 struct _mesa_index_buffer;
+struct _mesa_prim;
+struct st_context;
 
 void st_init_draw( struct st_context *st );
 
diff --git a/src/mesa/state_tracker/st_extensions.h b/src/mesa/state_tracker/st_extensions.h
index 2994f16dd3..aa9b2b2b91 100644
--- a/src/mesa/state_tracker/st_extensions.h
+++ b/src/mesa/state_tracker/st_extensions.h
@@ -30,6 +30,8 @@
 #define ST_EXTENSIONS_H
 
 
+struct st_context;
+
 extern void st_init_limits(struct st_context *st);
 
 extern void st_init_extensions(struct st_context *st);
diff --git a/src/mesa/state_tracker/st_format.h b/src/mesa/state_tracker/st_format.h
index 29768f296d..841c58cadc 100644
--- a/src/mesa/state_tracker/st_format.h
+++ b/src/mesa/state_tracker/st_format.h
@@ -31,7 +31,12 @@
 #define ST_FORMAT_H
 
 #include "main/formats.h"
+#include "main/mtypes.h"
 
+#include "pipe/p_defines.h"
+#include "pipe/p_format.h"
+
+struct pipe_screen;
 
 extern GLenum
 st_format_datatype(enum pipe_format format);
diff --git a/src/mesa/state_tracker/st_gen_mipmap.h b/src/mesa/state_tracker/st_gen_mipmap.h
index 00fbae9302..016bf3f4bb 100644
--- a/src/mesa/state_tracker/st_gen_mipmap.h
+++ b/src/mesa/state_tracker/st_gen_mipmap.h
@@ -30,6 +30,10 @@
 #define ST_GEN_MIPMAP_H
 
 
+#include "main/mtypes.h"
+
+struct st_context;
+
 extern void
 st_init_generate_mipmap(struct st_context *st);
 
diff --git a/src/mesa/state_tracker/st_gl_api.h b/src/mesa/state_tracker/st_gl_api.h
index fe1aec207e..57c6d9f24d 100644
--- a/src/mesa/state_tracker/st_gl_api.h
+++ b/src/mesa/state_tracker/st_gl_api.h
@@ -2,8 +2,6 @@
 #ifndef ST_GL_API_H
 #define ST_GL_API_H
 
-#include "state_tracker/st_api.h"
-
 struct st_api *st_gl_api_create(void);
 struct st_api *st_gl_api_create_es1(void);
 struct st_api *st_gl_api_create_es2(void);
diff --git a/src/mesa/state_tracker/st_manager.h b/src/mesa/state_tracker/st_manager.h
index cd2887b1e0..48a9d4d99a 100644
--- a/src/mesa/state_tracker/st_manager.h
+++ b/src/mesa/state_tracker/st_manager.h
@@ -29,8 +29,11 @@
 #ifndef ST_MANAGER_H
 #define ST_MANAGER_H
 
-#include "state_tracker/st_api.h"
-#include "st_context.h"
+#include "main/mtypes.h"
+
+#include "pipe/p_compiler.h"
+
+struct st_context;
 
 struct pipe_surface *
 st_manager_get_egl_image_surface(struct st_context *st,
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index 97186f8dad..a19dcc9253 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -44,6 +44,15 @@
 #include "util/u_math.h"
 #include "util/u_memory.h"
 
+
+#define PROGRAM_ANY_CONST ((1 << PROGRAM_LOCAL_PARAM) |  \
+                           (1 << PROGRAM_ENV_PARAM) |    \
+                           (1 << PROGRAM_STATE_VAR) |    \
+                           (1 << PROGRAM_NAMED_PARAM) |  \
+                           (1 << PROGRAM_CONSTANT) |     \
+                           (1 << PROGRAM_UNIFORM))
+
+
 struct label {
    unsigned branch_target;
    unsigned token;
@@ -205,7 +214,7 @@ src_register( struct st_translate *t,
       return ureg_src_undef();
 
    case PROGRAM_TEMPORARY:
-      ASSERT(index >= 0);
+      assert(index >= 0);
       if (ureg_dst_is_undef(t->temps[index]))
          t->temps[index] = ureg_DECL_temporary( t->ureg );
       assert(index < Elements(t->temps));
@@ -215,7 +224,7 @@ src_register( struct st_translate *t,
    case PROGRAM_ENV_PARAM:
    case PROGRAM_LOCAL_PARAM:
    case PROGRAM_UNIFORM:
-      ASSERT(index >= 0);
+      assert(index >= 0);
       return t->constants[index];
    case PROGRAM_STATE_VAR:
    case PROGRAM_CONSTANT:       /* ie, immediate */
@@ -1059,6 +1068,16 @@ st_translate_mesa_program(
       t->address[0] = ureg_DECL_address( ureg );
    }
 
+   if (program->IndirectRegisterFiles & (1 << PROGRAM_TEMPORARY)) {
+      /* If temps are accessed with indirect addressing, declare temporaries
+       * in sequential order.  Else, we declare them on demand elsewhere.
+       */
+      for (i = 0; i < program->NumTemporaries; i++) {
+         /* XXX use TGSI_FILE_TEMPORARY_ARRAY when it's supported by ureg */
+         t->temps[i] = ureg_DECL_temporary( t->ureg );
+      }
+   }
+
    /* Emit constants and immediates.  Mesa uses a single index space
     * for these, so we put all the translated regs in t->constants.
     */
@@ -1069,7 +1088,7 @@ st_translate_mesa_program(
          ret = PIPE_ERROR_OUT_OF_MEMORY;
          goto out;
       }
-      
+
       for (i = 0; i < program->Parameters->NumParameters; i++) {
          switch (program->Parameters->Parameters[i].Type) {
          case PROGRAM_ENV_PARAM:
@@ -1080,13 +1099,14 @@ st_translate_mesa_program(
             t->constants[i] = ureg_DECL_constant( ureg, i );
             break;
 
-            /* Emit immediates only when there is no address register
-             * in use.  FIXME: Be smarter and recognize param arrays:
+            /* Emit immediates only when there's no indirect addressing of
+             * the const buffer.
+             * FIXME: Be smarter and recognize param arrays:
              * indirect addressing is only valid within the referenced
              * array.
              */
          case PROGRAM_CONSTANT:
-            if (program->NumAddressRegs > 0) 
+            if (program->IndirectRegisterFiles & PROGRAM_ANY_CONST)
                t->constants[i] = ureg_DECL_constant( ureg, i );
             else
                t->constants[i] = 
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h
index e3c5bd1d94..ca076ce362 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.h
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h
@@ -30,8 +30,10 @@
 #define ST_MESA_TO_TGSI_H
 
 #include "main/mtypes.h"
-#include "tgsi/tgsi_ureg.h"
 
+#include "pipe/p_compiler.h"
+
+struct ureg_program;
 
 #if defined __cplusplus
 extern "C" {
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index 6f3ecdbce1..91528c227b 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -41,6 +41,7 @@
 #include "pipe/p_shader_tokens.h"
 #include "draw/draw_context.h"
 #include "tgsi/tgsi_dump.h"
+#include "tgsi/tgsi_ureg.h"
 
 #include "st_debug.h"
 #include "st_context.h"
diff --git a/src/mesa/state_tracker/st_program.h b/src/mesa/state_tracker/st_program.h
index d779d5a6dd..3805b9a725 100644
--- a/src/mesa/state_tracker/st_program.h
+++ b/src/mesa/state_tracker/st_program.h
@@ -36,11 +36,8 @@
 
 #include "main/mtypes.h"
 #include "program/program.h"
-#include "pipe/p_shader_tokens.h"
-
-
-struct cso_fragment_shader;
-struct cso_vertex_shader;
+#include "pipe/p_state.h"
+#include "st_context.h"
 
 
 /**
diff --git a/src/mesa/state_tracker/st_texture.c b/src/mesa/state_tracker/st_texture.c
index dbdf1ea1ad..add6e949df 100644
--- a/src/mesa/state_tracker/st_texture.c
+++ b/src/mesa/state_tracker/st_texture.c
@@ -25,14 +25,14 @@
  * 
  **************************************************************************/
 
+#include <stdio.h>
+
 #include "st_context.h"
 #include "st_format.h"
 #include "st_texture.h"
 #include "st_cb_fbo.h"
 #include "main/enums.h"
 
-#undef Elements  /* fix re-defined macro warning */
-
 #include "pipe/p_state.h"
 #include "pipe/p_context.h"
 #include "pipe/p_defines.h"
diff --git a/src/mesa/swrast/s_aaline.h b/src/mesa/swrast/s_aaline.h
index f1d708ec80..922eb230e5 100644
--- a/src/mesa/swrast/s_aaline.h
+++ b/src/mesa/swrast/s_aaline.h
@@ -28,7 +28,7 @@
 #define S_AALINE_H
 
 
-#include "swrast.h"
+#include "main/mtypes.h"
 
 
 extern void
diff --git a/src/mesa/swrast/s_aatriangle.h b/src/mesa/swrast/s_aatriangle.h
index 4b57fa73a2..9aed41a191 100644
--- a/src/mesa/swrast/s_aatriangle.h
+++ b/src/mesa/swrast/s_aatriangle.h
@@ -28,7 +28,7 @@
 #define S_AATRIANGLE_H
 
 
-#include "swrast.h"
+#include "main/mtypes.h"
 
 
 extern void
diff --git a/src/mesa/swrast/s_atifragshader.c b/src/mesa/swrast/s_atifragshader.c
index fa280e72e4..336415d57a 100644
--- a/src/mesa/swrast/s_atifragshader.c
+++ b/src/mesa/swrast/s_atifragshader.c
@@ -21,7 +21,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/atifragshader.h"
 #include "swrast/s_atifragshader.h"
diff --git a/src/mesa/swrast/s_context.c b/src/mesa/swrast/s_context.c
index 6d2d17c61d..d8d8a80b7d 100644
--- a/src/mesa/swrast/s_context.c
+++ b/src/mesa/swrast/s_context.c
@@ -28,7 +28,6 @@
 
 #include "main/imports.h"
 #include "main/bufferobj.h"
-#include "main/context.h"
 #include "main/colormac.h"
 #include "main/mtypes.h"
 #include "main/teximage.h"
diff --git a/src/mesa/swrast/s_feedback.c b/src/mesa/swrast/s_feedback.c
index 373b1416e2..6ac8ac73b0 100644
--- a/src/mesa/swrast/s_feedback.c
+++ b/src/mesa/swrast/s_feedback.c
@@ -24,7 +24,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/feedback.h"
 #include "main/macros.h"
 
diff --git a/src/mesa/swrast/s_fog.c b/src/mesa/swrast/s_fog.c
index 3fc8439213..689500a613 100644
--- a/src/mesa/swrast/s_fog.c
+++ b/src/mesa/swrast/s_fog.c
@@ -25,7 +25,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 
 #include "s_context.h"
diff --git a/src/mesa/swrast/s_fragprog.c b/src/mesa/swrast/s_fragprog.c
index 413f136cd5..a2c2a10c3d 100644
--- a/src/mesa/swrast/s_fragprog.c
+++ b/src/mesa/swrast/s_fragprog.c
@@ -24,7 +24,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "program/prog_instruction.h"
 
 #include "s_fragprog.h"
diff --git a/src/mesa/swrast/s_points.c b/src/mesa/swrast/s_points.c
index 1663ece829..12431662c4 100644
--- a/src/mesa/swrast/s_points.c
+++ b/src/mesa/swrast/s_points.c
@@ -25,7 +25,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "s_context.h"
 #include "s_feedback.h"
diff --git a/src/mesa/swrast/s_readpix.c b/src/mesa/swrast/s_readpix.c
index 6ad9aceec7..553fd9a76d 100644
--- a/src/mesa/swrast/s_readpix.c
+++ b/src/mesa/swrast/s_readpix.c
@@ -27,7 +27,6 @@
 #include "main/bufferobj.h"
 #include "main/colormac.h"
 #include "main/convolve.h"
-#include "main/context.h"
 #include "main/feedback.h"
 #include "main/formats.h"
 #include "main/image.h"
diff --git a/src/mesa/swrast/s_span.c b/src/mesa/swrast/s_span.c
index 687c8eb0bf..3b3929a41b 100644
--- a/src/mesa/swrast/s_span.c
+++ b/src/mesa/swrast/s_span.c
@@ -33,7 +33,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/image.h"
diff --git a/src/mesa/tnl/t_rasterpos.c b/src/mesa/tnl/t_rasterpos.c
index 3596d162b2..d82d5b5073 100644
--- a/src/mesa/tnl/t_rasterpos.c
+++ b/src/mesa/tnl/t_rasterpos.c
@@ -25,7 +25,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/feedback.h"
 #include "main/light.h"
 #include "main/macros.h"
diff --git a/src/mesa/tnl/t_vb_cull.c b/src/mesa/tnl/t_vb_cull.c
index 712901acf3..22df716673 100644
--- a/src/mesa/tnl/t_vb_cull.c
+++ b/src/mesa/tnl/t_vb_cull.c
@@ -28,7 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/tnl/t_vb_fog.c b/src/mesa/tnl/t_vb_fog.c
index 4a0e6ad4f9..9faae24ec6 100644
--- a/src/mesa/tnl/t_vb_fog.c
+++ b/src/mesa/tnl/t_vb_fog.c
@@ -28,7 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/tnl/t_vb_normals.c b/src/mesa/tnl/t_vb_normals.c
index 61ac409573..c2aa655674 100644
--- a/src/mesa/tnl/t_vb_normals.c
+++ b/src/mesa/tnl/t_vb_normals.c
@@ -28,7 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/tnl/t_vb_program.c b/src/mesa/tnl/t_vb_program.c
index 614c67d05e..3973df9a67 100644
--- a/src/mesa/tnl/t_vb_program.c
+++ b/src/mesa/tnl/t_vb_program.c
@@ -33,7 +33,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "program/prog_instruction.h"
diff --git a/src/mesa/tnl/t_vb_texgen.c b/src/mesa/tnl/t_vb_texgen.c
index 9ef13bc96d..950e0f54e9 100644
--- a/src/mesa/tnl/t_vb_texgen.c
+++ b/src/mesa/tnl/t_vb_texgen.c
@@ -37,7 +37,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/tnl/t_vb_texmat.c b/src/mesa/tnl/t_vb_texmat.c
index 83688290e5..985d137e5c 100644
--- a/src/mesa/tnl/t_vb_texmat.c
+++ b/src/mesa/tnl/t_vb_texmat.c
@@ -28,7 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/tnl/t_vb_vertex.c b/src/mesa/tnl/t_vb_vertex.c
index a275342563..453479227b 100644
--- a/src/mesa/tnl/t_vb_vertex.c
+++ b/src/mesa/tnl/t_vb_vertex.c
@@ -28,7 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/colormac.h"
-#include "main/context.h"
 #include "main/macros.h"
 #include "main/imports.h"
 #include "main/mtypes.h"
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index 045af46da8..84ae1b87f9 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -27,7 +27,7 @@
 
 #include "main/glheader.h"
 #include "main/bufferobj.h"
-#include "main/context.h"
+#include "main/compiler.h"
 #include "main/enums.h"
 #include "main/state.h"
 
diff --git a/src/mesa/vf/vf_generic.c b/src/mesa/vf/vf_generic.c
index 0af8893c30..95a317e99d 100644
--- a/src/mesa/vf/vf_generic.c
+++ b/src/mesa/vf/vf_generic.c
@@ -29,6 +29,7 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/colormac.h"
+#include "main/macros.h"
 #include "main/simple_list.h"
 
 #include "vf/vf.h"
author	Ian Romanick <ian.d.romanick@intel.com>	2010-08-13 20:22:46 -0700
committer	Ian Romanick <ian.d.romanick@intel.com>	2010-08-13 20:22:46 -0700
commit	443a7e4e9a360acbc3e662c098be436f180bf81d (patch)
tree	71125fc394bdcbc17cd7ed74c592b9a7761541f8 /src/mesa
parent	2f8ee757ab324d599fcb8287789eb5f1a7890d74 (diff)
parent	27041d7cb3faeaed483538a228573466363ec1c7 (diff)