Merge remote branch 'origin/master' into radeon-rewrite

Conflicts: configure.ac src/mesa/drivers/dri/r200/r200_context.c src/mesa/drivers/dri/r300/r300_render.c
author: Dave Airlie <[email protected]> 2009-02-13 01:11:25 +1000
committer: Dave Airlie <[email protected]> 2009-02-13 01:11:25 +1000
commit: 53116910b10e3b8a05f42970eff311c21808699f (patch)
tree: 6831e50a56a02fed324f8cb119e05d837e00d7cf /src/mesa/drivers
parent: 7e104f9cde94279a902b408e8d1cf21779b393a8 (diff)
parent: 0ccbc3c905f0594a35d72887a1f115e148aaa596 (diff)
166 files changed, 6200 insertions, 4183 deletions
diff --git a/src/mesa/drivers/allegro/amesa.c b/src/mesa/drivers/allegro/amesa.c
index ade62518489..0744677d2bd 100644
--- a/src/mesa/drivers/allegro/amesa.c
+++ b/src/mesa/drivers/allegro/amesa.c
@@ -26,7 +26,7 @@
 #include "main/imports.h"
 #include "main/matrix.h"
 #include "main/mtypes.h"
-#include "GL/amesa.h"
+#include "amesa.h"
 
 
 struct amesa_visual
@@ -338,7 +338,7 @@ void AMesaDestroyBuffer(AMesaBuffer buffer)
 {
    if (buffer->Screen)     destroy_bitmap(buffer->Screen);
    if (buffer->Background) destroy_bitmap(buffer->Background);
-   _mesa_unreference_framebuffer(&buffer->GLBuffer);
+   _mesa_reference_framebuffer(&buffer->GLBuffer, NULL);
    free(buffer);
 }
 
diff --git a/src/mesa/drivers/allegro/amesa.h b/src/mesa/drivers/allegro/amesa.h
new file mode 100644
index 00000000000..852d34cf4f5
--- /dev/null
+++ b/src/mesa/drivers/allegro/amesa.h
@@ -0,0 +1,65 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  3.3
+ * 
+ * Copyright (C) 1999-2000  Brian Paul   All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/* Allegro (DJGPP) driver by Bernhard Tschirren ([email protected]) */
+
+
+#ifndef AMESA_H
+#define AMESA_H
+
+
+#define AMESA_MAJOR_VERSION 3
+#define AMESA_MINOR_VERSION 3
+
+
+typedef struct amesa_visual  *AMesaVisual;
+typedef struct amesa_buffer  *AMesaBuffer;
+typedef struct amesa_context *AMesaContext;
+
+
+extern AMesaVisual AMesaCreateVisual(GLboolean dbFlag, GLint depth,
+                                     GLint depthSize,
+                                     GLint stencilSize,
+                                     GLint accumSize);
+
+extern void AMesaDestroyVisual(AMesaVisual visual);
+
+extern AMesaBuffer AMesaCreateBuffer(AMesaVisual visual,
+                                     GLint width, GLint height);
+
+extern void AMesaDestroyBuffer(AMesaBuffer buffer);
+
+
+extern AMesaContext AMesaCreateContext(AMesaVisual visual,
+                                       AMesaContext sharelist);
+
+extern void AMesaDestroyContext(AMesaContext context);
+
+extern GLboolean AMesaMakeCurrent(AMesaContext context, AMesaBuffer buffer);
+
+extern void AMesaSwapBuffers(AMesaBuffer buffer);
+
+
+#endif /* AMESA_H */
diff --git a/src/mesa/drivers/dri/Makefile b/src/mesa/drivers/dri/Makefile
index eef68825bc9..32db0978613 100644
--- a/src/mesa/drivers/dri/Makefile
+++ b/src/mesa/drivers/dri/Makefile
@@ -6,7 +6,7 @@ include $(TOP)/configs/current
 
 
 
-default: $(TOP)/$(LIB_DIR) subdirs
+default: $(TOP)/$(LIB_DIR) subdirs dri.pc
 
 
 $(TOP)/$(LIB_DIR):
@@ -25,7 +25,8 @@ pcedit = sed \
 	-e 's,@INSTALL_LIB_DIR@,$(INSTALL_LIB_DIR),' \
 	-e 's,@INSTALL_INC_DIR@,$(INSTALL_INC_DIR),' \
 	-e 's,@VERSION@,$(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY),' \
-	-e 's,@DRI_DRIVER_DIR@,$(DRI_DRIVER_SEARCH_DIR),'
+	-e 's,@DRI_DRIVER_DIR@,$(DRI_DRIVER_SEARCH_DIR),' \
+	-e 's,@DRI_PC_REQ_PRIV@,$(DRI_PC_REQ_PRIV),'
 
 dri.pc: dri.pc.in
 	$(pcedit) $< > $@
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 2dc3664cc6a..2fa36bab3f9 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -2,13 +2,16 @@
 
 MESA_MODULES = $(TOP)/src/mesa/libmesa.a
 
-COMMON_SOURCES = \
+COMMON_GALLIUM_SOURCES = \
         ../common/utils.c \
-        ../common/texmem.c \
         ../common/vblank.c \
         ../common/dri_util.c \
-        ../common/xmlconfig.c \
-        ../common/drirenderbuffer.c 
+        ../common/xmlconfig.c
+
+COMMON_SOURCES = $(COMMON_GALLIUM_SOURCES) \
+        ../../common/driverfuncs.c \
+        ../common/texmem.c \
+        ../common/drirenderbuffer.c
 
 ifeq ($(WINDOW_SYSTEM),dri)
 WINOBJ=
@@ -59,9 +62,9 @@ SHARED_INCLUDES = \
 default: symlinks depend $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
 
 
-$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
+$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(PIPE_DRIVERS) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
 	$(MKLIB) -o $@ -noprefix -linker '$(CC)' -ldflags '$(LDFLAGS)' \
-		$(OBJECTS) $(MESA_MODULES) $(WINOBJ) $(DRI_LIB_DEPS)
+		$(OBJECTS) $(PIPE_DRIVERS) $(MESA_MODULES) $(WINOBJ) $(DRI_LIB_DEPS)
 
 
 $(TOP)/$(LIB_DIR)/$(LIBNAME): $(LIBNAME)
@@ -69,6 +72,7 @@ $(TOP)/$(LIB_DIR)/$(LIBNAME): $(LIBNAME)
 
 
 depend: $(C_SOURCES) $(ASM_SOURCES) $(SYMLINKS)
+	rm -f depend
 	touch depend
 	$(MKDEP) $(MKDEP_OPTIONS) $(DRIVER_DEFINES) $(INCLUDES) $(C_SOURCES) \
 		$(ASM_SOURCES)
diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c
index e9b1a80b913..ae790554055 100644
--- a/src/mesa/drivers/dri/common/dri_util.c
+++ b/src/mesa/drivers/dri/common/dri_util.c
@@ -314,13 +314,28 @@ static void driReportDamage(__DRIdrawable *pdp,
 static void driSwapBuffers(__DRIdrawable *dPriv)
 {
     __DRIscreen *psp = dPriv->driScreenPriv;
-
+    drm_clip_rect_t *rects;
+    int i;
+    
     if (!dPriv->numClipRects)
         return;
 
     psp->DriverAPI.SwapBuffers(dPriv);
 
-    driReportDamage(dPriv, dPriv->pClipRects, dPriv->numClipRects);
+    rects = _mesa_malloc(sizeof(*rects) * dPriv->numClipRects);
+
+    if (!rects)
+        return;
+
+    for (i = 0; i < dPriv->numClipRects; i++) {
+        rects[i].x1 = dPriv->pClipRects[i].x1 - dPriv->x;
+        rects[i].y1 = dPriv->pClipRects[i].y1 - dPriv->y;
+        rects[i].x2 = dPriv->pClipRects[i].x2 - dPriv->x;
+        rects[i].y2 = dPriv->pClipRects[i].y2 - dPriv->y;
+    }
+
+    driReportDamage(dPriv, rects, dPriv->numClipRects);
+    _mesa_free(rects);
 }
 
 static int driDrawableGetMSC( __DRIscreen *sPriv, __DRIdrawable *dPriv,
diff --git a/src/mesa/drivers/dri/common/dri_util.h b/src/mesa/drivers/dri/common/dri_util.h
index c0e1bea5e05..c95a5c8299a 100644
--- a/src/mesa/drivers/dri/common/dri_util.h
+++ b/src/mesa/drivers/dri/common/dri_util.h
@@ -1,25 +1,3 @@
-/* $XFree86: xc/lib/GL/dri/dri_util.h,v 1.1 2002/02/22 21:32:52 dawes Exp $ */
-/**
- * \file dri_util.h
- * DRI utility functions definitions.
- *
- * This module acts as glue between GLX and the actual hardware driver.  A DRI
- * driver doesn't really \e have to use any of this - it's optional.  But, some
- * useful stuff is done here that otherwise would have to be duplicated in most
- * drivers.
- * 
- * Basically, these utility functions take care of some of the dirty details of
- * screen initialization, context creation, context binding, DRM setup, etc.
- *
- * These functions are compiled into each DRI driver so libGL.so knows nothing
- * about them.
- *
- * \sa dri_util.c.
- * 
- * \author Kevin E. Martin <[email protected]>
- * \author Brian Paul <[email protected]>
- */
-
 /*
  * Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
  * All Rights Reserved.
@@ -45,6 +23,26 @@
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+/**
+ * \file dri_util.h
+ * DRI utility functions definitions.
+ *
+ * This module acts as glue between GLX and the actual hardware driver.  A DRI
+ * driver doesn't really \e have to use any of this - it's optional.  But, some
+ * useful stuff is done here that otherwise would have to be duplicated in most
+ * drivers.
+ * 
+ * Basically, these utility functions take care of some of the dirty details of
+ * screen initialization, context creation, context binding, DRM setup, etc.
+ *
+ * These functions are compiled into each DRI driver so libGL.so knows nothing
+ * about them.
+ *
+ * \sa dri_util.c.
+ * 
+ * \author Kevin E. Martin <[email protected]>
+ * \author Brian Paul <[email protected]>
+ */
 
 #ifndef _DRI_UTIL_H_
 #define _DRI_UTIL_H_
@@ -108,6 +106,28 @@ do {                                                                    \
     }                                                                   \
 } while (0)
 
+/**
+ * Same as above, but for two drawables simultaneously.
+ *
+ */
+
+#define DRI_VALIDATE_TWO_DRAWABLES_INFO(psp, pdp, prp)			\
+do {								\
+    while (*((pdp)->pStamp) != (pdp)->lastStamp ||			\
+	   *((prp)->pStamp) != (prp)->lastStamp) {			\
+        register unsigned int hwContext = (psp)->pSAREA->lock.lock &	\
+	    ~(DRM_LOCK_HELD | DRM_LOCK_CONT);				\
+	DRM_UNLOCK((psp)->fd, &(psp)->pSAREA->lock, hwContext);		\
+									\
+	DRM_SPINLOCK(&(psp)->pSAREA->drawable_lock, (psp)->drawLockID);	\
+	DRI_VALIDATE_DRAWABLE_INFO_ONCE(pdp);                           \
+	DRI_VALIDATE_DRAWABLE_INFO_ONCE(prp);				\
+	DRM_SPINUNLOCK(&(psp)->pSAREA->drawable_lock, (psp)->drawLockID); \
+									\
+	DRM_LIGHT_LOCK((psp)->fd, &(psp)->pSAREA->lock, hwContext);	\
+    }                                                                   \
+} while (0)
+
 
 /**
  * Driver callback functions.
diff --git a/src/mesa/drivers/dri/common/drirenderbuffer.c b/src/mesa/drivers/dri/common/drirenderbuffer.c
index b99bf2033b2..15af99136cd 100644
--- a/src/mesa/drivers/dri/common/drirenderbuffer.c
+++ b/src/mesa/drivers/dri/common/drirenderbuffer.c
@@ -209,6 +209,8 @@ driUpdateFramebufferSize(GLcontext *ctx, const __DRIdrawablePrivate *dPriv)
    struct gl_framebuffer *fb = (struct gl_framebuffer *) dPriv->driverPrivate;
    if (fb && (dPriv->w != fb->Width || dPriv->h != fb->Height)) {
       ctx->Driver.ResizeBuffers(ctx, fb, dPriv->w, dPriv->h);
+      /* if the driver needs the hw lock for ResizeBuffers, the drawable
+         might have changed again by now */
       assert(fb->Width == dPriv->w);
       assert(fb->Height == dPriv->h);
    }
diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h
index b977ebf0153..0d641f25f7c 100644
--- a/src/mesa/drivers/dri/common/extension_helper.h
+++ b/src/mesa/drivers/dri/common/extension_helper.h
@@ -119,9 +119,10 @@ static const char ActiveTextureARB_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_blit)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_blit)
 static const char BlitFramebufferEXT_names[] = 
     "iiiiiiiiii\0" /* Parameter signature */
+    "glBlitFramebuffer\0"
     "glBlitFramebufferEXT\0"
     "";
 #endif
@@ -433,17 +434,18 @@ static const char Color4ubVertex3fvSUN_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_texture_array)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_texture_array)
 static const char FramebufferTextureLayerEXT_names[] = 
     "iiiii\0" /* Parameter signature */
+    "glFramebufferTextureLayer\0"
     "glFramebufferTextureLayerEXT\0"
     "";
 #endif
 
 #if defined(need_GL_SGIX_list_priority)
-static const char GetListParameterivSGIX_names[] = 
+static const char GetListParameterfvSGIX_names[] = 
     "iip\0" /* Parameter signature */
-    "glGetListParameterivSGIX\0"
+    "glGetListParameterfvSGIX\0"
     "";
 #endif
 
@@ -777,6 +779,13 @@ static const char VertexAttribs4fvNV_names[] =
     "";
 #endif
 
+#if defined(need_GL_NV_vertex_array_range)
+static const char VertexArrayRangeNV_names[] = 
+    "ip\0" /* Parameter signature */
+    "glVertexArrayRangeNV\0"
+    "";
+#endif
+
 #if defined(need_GL_SGIX_fragment_lighting)
 static const char FragmentLightiSGIX_names[] = 
     "iii\0" /* Parameter signature */
@@ -946,9 +955,10 @@ static const char VertexAttribs3fvNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char GenerateMipmapEXT_names[] = 
     "i\0" /* Parameter signature */
+    "glGenerateMipmap\0"
     "glGenerateMipmapEXT\0"
     "";
 #endif
@@ -1077,13 +1087,6 @@ static const char AreTexturesResident_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
-static const char IsRenderbufferEXT_names[] = 
-    "i\0" /* Parameter signature */
-    "glIsRenderbufferEXT\0"
-    "";
-#endif
-
 #if defined(need_GL_VERSION_2_0) || defined(need_GL_ATI_separate_stencil)
 static const char StencilOpSeparate_names[] = 
     "iiii\0" /* Parameter signature */
@@ -1171,9 +1174,10 @@ static const char TexCoord2fNormal3fVertex3fSUN_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char RenderbufferStorageEXT_names[] = 
     "iiii\0" /* Parameter signature */
+    "glRenderbufferStorage\0"
     "glRenderbufferStorageEXT\0"
     "";
 #endif
@@ -1364,9 +1368,10 @@ static const char SpriteParameterfvSGIX_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char CheckFramebufferStatusEXT_names[] = 
     "i\0" /* Parameter signature */
+    "glCheckFramebufferStatus\0"
     "glCheckFramebufferStatusEXT\0"
     "";
 #endif
@@ -1554,13 +1559,6 @@ static const char FragmentLightivSGIX_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
-static const char DeleteRenderbuffersEXT_names[] = 
-    "ip\0" /* Parameter signature */
-    "glDeleteRenderbuffersEXT\0"
-    "";
-#endif
-
 #if defined(need_GL_EXT_pixel_transform)
 static const char PixelTransformParameterfvEXT_names[] = 
     "iip\0" /* Parameter signature */
@@ -1621,9 +1619,10 @@ static const char WindowPos2dMESA_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char FramebufferTexture3DEXT_names[] = 
     "iiiiii\0" /* Parameter signature */
+    "glFramebufferTexture3D\0"
     "glFramebufferTexture3DEXT\0"
     "";
 #endif
@@ -1673,10 +1672,11 @@ static const char IndexFuncEXT_names[] =
     "";
 #endif
 
-#if defined(need_GL_SGIX_list_priority)
-static const char GetListParameterfvSGIX_names[] = 
-    "iip\0" /* Parameter signature */
-    "glGetListParameterfvSGIX\0"
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
+static const char FramebufferTexture2DEXT_names[] = 
+    "iiiii\0" /* Parameter signature */
+    "glFramebufferTexture2D\0"
+    "glFramebufferTexture2DEXT\0"
     "";
 #endif
 
@@ -1763,13 +1763,6 @@ static const char DeleteFencesNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_SGIX_polynomial_ffd)
-static const char DeformationMap3dSGIX_names[] = 
-    "iddiiddiiddiip\0" /* Parameter signature */
-    "glDeformationMap3dSGIX\0"
-    "";
-#endif
-
 #if defined(need_GL_VERSION_2_0)
 static const char IsShader_names[] = 
     "i\0" /* Parameter signature */
@@ -1894,6 +1887,13 @@ static const char MatrixIndexuivARB_names[] =
     "";
 #endif
 
+#if defined(need_GL_ARB_framebuffer_object)
+static const char RenderbufferStorageMultisample_names[] = 
+    "iiiii\0" /* Parameter signature */
+    "glRenderbufferStorageMultisample\0"
+    "";
+#endif
+
 #if defined(need_GL_EXT_coordinate_frame)
 static const char Tangent3sEXT_names[] = 
     "iii\0" /* Parameter signature */
@@ -1944,9 +1944,10 @@ static const char VertexAttrib2dvNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char FramebufferRenderbufferEXT_names[] = 
     "iiii\0" /* Parameter signature */
+    "glFramebufferRenderbuffer\0"
     "glFramebufferRenderbufferEXT\0"
     "";
 #endif
@@ -1973,9 +1974,10 @@ static const char SetFenceNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char FramebufferTexture1DEXT_names[] = 
     "iiiii\0" /* Parameter signature */
+    "glFramebufferTexture1D\0"
     "glFramebufferTexture1DEXT\0"
     "";
 #endif
@@ -2105,9 +2107,10 @@ static const char Tangent3fvEXT_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char BindFramebufferEXT_names[] = 
     "ii\0" /* Parameter signature */
+    "glBindFramebuffer\0"
     "glBindFramebufferEXT\0"
     "";
 #endif
@@ -2488,9 +2491,10 @@ static const char MapParameterivNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char GenRenderbuffersEXT_names[] = 
     "ip\0" /* Parameter signature */
+    "glGenRenderbuffers\0"
     "glGenRenderbuffersEXT\0"
     "";
 #endif
@@ -2952,9 +2956,10 @@ static const char WindowPos2ivMESA_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char IsFramebufferEXT_names[] = 
     "i\0" /* Parameter signature */
+    "glIsFramebuffer\0"
     "glIsFramebufferEXT\0"
     "";
 #endif
@@ -3125,13 +3130,6 @@ static const char ReplacementCodeuiSUN_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
-static const char FramebufferTexture2DEXT_names[] = 
-    "iiiii\0" /* Parameter signature */
-    "glFramebufferTexture2DEXT\0"
-    "";
-#endif
-
 #if defined(need_GL_NV_vertex_program)
 static const char VertexAttribPointerNV_names[] = 
     "iiiip\0" /* Parameter signature */
@@ -3139,13 +3137,21 @@ static const char VertexAttribPointerNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char GetFramebufferAttachmentParameterivEXT_names[] = 
     "iiip\0" /* Parameter signature */
+    "glGetFramebufferAttachmentParameteriv\0"
     "glGetFramebufferAttachmentParameterivEXT\0"
     "";
 #endif
 
+#if defined(need_GL_EXT_pixel_transform)
+static const char PixelTransformParameterfEXT_names[] = 
+    "iif\0" /* Parameter signature */
+    "glPixelTransformParameterfEXT\0"
+    "";
+#endif
+
 #if defined(need_GL_VERSION_1_3)
 static const char MultiTexCoord4dvARB_names[] = 
     "ip\0" /* Parameter signature */
@@ -3264,9 +3270,10 @@ static const char VertexAttrib3fvARB_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char DeleteFramebuffersEXT_names[] = 
     "ip\0" /* Parameter signature */
+    "glDeleteFramebuffers\0"
     "glDeleteFramebuffersEXT\0"
     "";
 #endif
@@ -3360,9 +3367,10 @@ static const char GetPixelTexGenParameterfvSGIS_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char GenFramebuffersEXT_names[] = 
     "ip\0" /* Parameter signature */
+    "glGenFramebuffers\0"
     "glGenFramebuffersEXT\0"
     "";
 #endif
@@ -3374,10 +3382,10 @@ static const char GetProgramParameterdvNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_pixel_transform)
-static const char PixelTransformParameterfEXT_names[] = 
-    "iif\0" /* Parameter signature */
-    "glPixelTransformParameterfEXT\0"
+#if defined(need_GL_SGIX_instruments)
+static const char PollInstrumentsSGIX_names[] = 
+    "p\0" /* Parameter signature */
+    "glPollInstrumentsSGIX\0"
     "";
 #endif
 
@@ -3709,9 +3717,10 @@ static const char MultiDrawArraysEXT_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char BindRenderbufferEXT_names[] = 
     "ii\0" /* Parameter signature */
+    "glBindRenderbuffer\0"
     "glBindRenderbufferEXT\0"
     "";
 #endif
@@ -4530,6 +4539,13 @@ static const char Minmax_names[] =
     "";
 #endif
 
+#if defined(need_GL_SGIX_polynomial_ffd)
+static const char DeformationMap3dSGIX_names[] = 
+    "iddiiddiiddiip\0" /* Parameter signature */
+    "glDeformationMap3dSGIX\0"
+    "";
+#endif
+
 #if defined(need_GL_VERSION_1_4) || defined(need_GL_EXT_fog_coord)
 static const char FogCoorddvEXT_names[] = 
     "p\0" /* Parameter signature */
@@ -4605,10 +4621,11 @@ static const char WeightdvARB_names[] =
     "";
 #endif
 
-#if defined(need_GL_SGIX_instruments)
-static const char PollInstrumentsSGIX_names[] = 
-    "p\0" /* Parameter signature */
-    "glPollInstrumentsSGIX\0"
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
+static const char DeleteRenderbuffersEXT_names[] = 
+    "ip\0" /* Parameter signature */
+    "glDeleteRenderbuffers\0"
+    "glDeleteRenderbuffersEXT\0"
     "";
 #endif
 
@@ -4684,9 +4701,10 @@ static const char MultiTexCoord2fvARB_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_framebuffer_object)
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
 static const char GetRenderbufferParameterivEXT_names[] = 
     "iip\0" /* Parameter signature */
+    "glGetRenderbufferParameteriv\0"
     "glGetRenderbufferParameterivEXT\0"
     "";
 #endif
@@ -4742,6 +4760,13 @@ static const char Tangent3svEXT_names[] =
     "";
 #endif
 
+#if defined(need_GL_SGIX_list_priority)
+static const char GetListParameterivSGIX_names[] = 
+    "iip\0" /* Parameter signature */
+    "glGetListParameterivSGIX\0"
+    "";
+#endif
+
 #if defined(need_GL_VERSION_1_5) || defined(need_GL_ARB_vertex_buffer_object)
 static const char BindBufferARB_names[] = 
     "ii\0" /* Parameter signature */
@@ -4823,10 +4848,11 @@ static const char FragmentMaterialivSGIX_names[] =
     "";
 #endif
 
-#if defined(need_GL_NV_vertex_array_range)
-static const char VertexArrayRangeNV_names[] = 
-    "ip\0" /* Parameter signature */
-    "glVertexArrayRangeNV\0"
+#if defined(need_GL_ARB_framebuffer_object) || defined(need_GL_EXT_framebuffer_object)
+static const char IsRenderbufferEXT_names[] = 
+    "i\0" /* Parameter signature */
+    "glIsRenderbuffer\0"
+    "glIsRenderbufferEXT\0"
     "";
 #endif
 
@@ -4892,6 +4918,32 @@ static const struct dri_extension_function GL_ARB_draw_buffers_functions[] = {
 };
 #endif
 
+#if defined(need_GL_ARB_framebuffer_object)
+static const struct dri_extension_function GL_ARB_framebuffer_object_functions[] = {
+    { BlitFramebufferEXT_names, BlitFramebufferEXT_remap_index, -1 },
+    { FramebufferTextureLayerEXT_names, FramebufferTextureLayerEXT_remap_index, -1 },
+    { GenerateMipmapEXT_names, GenerateMipmapEXT_remap_index, -1 },
+    { RenderbufferStorageEXT_names, RenderbufferStorageEXT_remap_index, -1 },
+    { CheckFramebufferStatusEXT_names, CheckFramebufferStatusEXT_remap_index, -1 },
+    { FramebufferTexture3DEXT_names, FramebufferTexture3DEXT_remap_index, -1 },
+    { FramebufferTexture2DEXT_names, FramebufferTexture2DEXT_remap_index, -1 },
+    { RenderbufferStorageMultisample_names, RenderbufferStorageMultisample_remap_index, -1 },
+    { FramebufferRenderbufferEXT_names, FramebufferRenderbufferEXT_remap_index, -1 },
+    { FramebufferTexture1DEXT_names, FramebufferTexture1DEXT_remap_index, -1 },
+    { BindFramebufferEXT_names, BindFramebufferEXT_remap_index, -1 },
+    { GenRenderbuffersEXT_names, GenRenderbuffersEXT_remap_index, -1 },
+    { IsFramebufferEXT_names, IsFramebufferEXT_remap_index, -1 },
+    { GetFramebufferAttachmentParameterivEXT_names, GetFramebufferAttachmentParameterivEXT_remap_index, -1 },
+    { DeleteFramebuffersEXT_names, DeleteFramebuffersEXT_remap_index, -1 },
+    { GenFramebuffersEXT_names, GenFramebuffersEXT_remap_index, -1 },
+    { BindRenderbufferEXT_names, BindRenderbufferEXT_remap_index, -1 },
+    { DeleteRenderbuffersEXT_names, DeleteRenderbuffersEXT_remap_index, -1 },
+    { GetRenderbufferParameterivEXT_names, GetRenderbufferParameterivEXT_remap_index, -1 },
+    { IsRenderbufferEXT_names, IsRenderbufferEXT_remap_index, -1 },
+    { NULL, 0, 0 }
+};
+#endif
+
 #if defined(need_GL_ARB_matrix_palette)
 static const struct dri_extension_function GL_ARB_matrix_palette_functions[] = {
     { MatrixIndexusvARB_names, MatrixIndexusvARB_remap_index, -1 },
@@ -5319,22 +5371,22 @@ static const struct dri_extension_function GL_EXT_framebuffer_blit_functions[] =
 #if defined(need_GL_EXT_framebuffer_object)
 static const struct dri_extension_function GL_EXT_framebuffer_object_functions[] = {
     { GenerateMipmapEXT_names, GenerateMipmapEXT_remap_index, -1 },
-    { IsRenderbufferEXT_names, IsRenderbufferEXT_remap_index, -1 },
     { RenderbufferStorageEXT_names, RenderbufferStorageEXT_remap_index, -1 },
     { CheckFramebufferStatusEXT_names, CheckFramebufferStatusEXT_remap_index, -1 },
-    { DeleteRenderbuffersEXT_names, DeleteRenderbuffersEXT_remap_index, -1 },
     { FramebufferTexture3DEXT_names, FramebufferTexture3DEXT_remap_index, -1 },
+    { FramebufferTexture2DEXT_names, FramebufferTexture2DEXT_remap_index, -1 },
     { FramebufferRenderbufferEXT_names, FramebufferRenderbufferEXT_remap_index, -1 },
     { FramebufferTexture1DEXT_names, FramebufferTexture1DEXT_remap_index, -1 },
     { BindFramebufferEXT_names, BindFramebufferEXT_remap_index, -1 },
     { GenRenderbuffersEXT_names, GenRenderbuffersEXT_remap_index, -1 },
     { IsFramebufferEXT_names, IsFramebufferEXT_remap_index, -1 },
-    { FramebufferTexture2DEXT_names, FramebufferTexture2DEXT_remap_index, -1 },
     { GetFramebufferAttachmentParameterivEXT_names, GetFramebufferAttachmentParameterivEXT_remap_index, -1 },
     { DeleteFramebuffersEXT_names, DeleteFramebuffersEXT_remap_index, -1 },
     { GenFramebuffersEXT_names, GenFramebuffersEXT_remap_index, -1 },
     { BindRenderbufferEXT_names, BindRenderbufferEXT_remap_index, -1 },
+    { DeleteRenderbuffersEXT_names, DeleteRenderbuffersEXT_remap_index, -1 },
     { GetRenderbufferParameterivEXT_names, GetRenderbufferParameterivEXT_remap_index, -1 },
+    { IsRenderbufferEXT_names, IsRenderbufferEXT_remap_index, -1 },
     { NULL, 0, 0 }
 };
 #endif
@@ -5415,8 +5467,8 @@ static const struct dri_extension_function GL_EXT_paletted_texture_functions[] =
 #if defined(need_GL_EXT_pixel_transform)
 static const struct dri_extension_function GL_EXT_pixel_transform_functions[] = {
     { PixelTransformParameterfvEXT_names, PixelTransformParameterfvEXT_remap_index, -1 },
-    { PixelTransformParameteriEXT_names, PixelTransformParameteriEXT_remap_index, -1 },
     { PixelTransformParameterfEXT_names, PixelTransformParameterfEXT_remap_index, -1 },
+    { PixelTransformParameteriEXT_names, PixelTransformParameteriEXT_remap_index, -1 },
     { PixelTransformParameterivEXT_names, PixelTransformParameterivEXT_remap_index, -1 },
     { NULL, 0, 0 }
 };
@@ -5716,8 +5768,8 @@ static const struct dri_extension_function GL_NV_register_combiners2_functions[]
 
 #if defined(need_GL_NV_vertex_array_range)
 static const struct dri_extension_function GL_NV_vertex_array_range_functions[] = {
-    { FlushVertexArrayRangeNV_names, FlushVertexArrayRangeNV_remap_index, -1 },
     { VertexArrayRangeNV_names, VertexArrayRangeNV_remap_index, -1 },
+    { FlushVertexArrayRangeNV_names, FlushVertexArrayRangeNV_remap_index, -1 },
     { NULL, 0, 0 }
 };
 #endif
@@ -5946,11 +5998,11 @@ static const struct dri_extension_function GL_SGIX_instruments_functions[] = {
 #if defined(need_GL_SGIX_list_priority)
 static const struct dri_extension_function GL_SGIX_list_priority_functions[] = {
     { ListParameterfSGIX_names, ListParameterfSGIX_remap_index, -1 },
-    { GetListParameterivSGIX_names, GetListParameterivSGIX_remap_index, -1 },
     { GetListParameterfvSGIX_names, GetListParameterfvSGIX_remap_index, -1 },
     { ListParameteriSGIX_names, ListParameteriSGIX_remap_index, -1 },
     { ListParameterfvSGIX_names, ListParameterfvSGIX_remap_index, -1 },
     { ListParameterivSGIX_names, ListParameterivSGIX_remap_index, -1 },
+    { GetListParameterivSGIX_names, GetListParameterivSGIX_remap_index, -1 },
     { NULL, 0, 0 }
 };
 #endif
@@ -5965,9 +6017,9 @@ static const struct dri_extension_function GL_SGIX_pixel_texture_functions[] = {
 #if defined(need_GL_SGIX_polynomial_ffd)
 static const struct dri_extension_function GL_SGIX_polynomial_ffd_functions[] = {
     { LoadIdentityDeformationMapSGIX_names, LoadIdentityDeformationMapSGIX_remap_index, -1 },
-    { DeformationMap3dSGIX_names, DeformationMap3dSGIX_remap_index, -1 },
     { DeformSGIX_names, DeformSGIX_remap_index, -1 },
     { DeformationMap3fSGIX_names, DeformationMap3fSGIX_remap_index, -1 },
+    { DeformationMap3dSGIX_names, DeformationMap3dSGIX_remap_index, -1 },
     { NULL, 0, 0 }
 };
 #endif
diff --git a/src/mesa/drivers/dri/common/utils.c b/src/mesa/drivers/dri/common/utils.c
index 30c860b96c5..6b44ed9a673 100644
--- a/src/mesa/drivers/dri/common/utils.c
+++ b/src/mesa/drivers/dri/common/utils.c
@@ -179,13 +179,18 @@ driGetRendererString( char * buffer, const char * hardware_name,
 
 
 
+#define need_GL_ARB_draw_buffers
 #define need_GL_ARB_multisample
+#define need_GL_ARB_texture_compression
 #define need_GL_ARB_transpose_matrix
+#define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_window_pos
 #define need_GL_EXT_compiled_vertex_array
+#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_polygon_offset
 #define need_GL_EXT_texture_object
 #define need_GL_EXT_vertex_array
+#define need_GL_IBM_multimode_draw_arrays
 #define need_GL_MESA_window_pos
 
 /* These are needed in *all* drivers because Mesa internally implements
@@ -198,14 +203,19 @@ driGetRendererString( char * buffer, const char * hardware_name,
 #include "extension_helper.h"
 
 static const struct dri_extension all_mesa_extensions[] = {
+   { "GL_ARB_draw_buffers",          GL_ARB_draw_buffers_functions },
    { "GL_ARB_multisample",           GL_ARB_multisample_functions },
+   { "GL_ARB_texture_compression",   GL_ARB_texture_compression_functions },
    { "GL_ARB_transpose_matrix",      GL_ARB_transpose_matrix_functions },
+   { "GL_ARB_vertex_buffer_object",  GL_ARB_vertex_buffer_object_functions},
    { "GL_ARB_window_pos",            GL_ARB_window_pos_functions },
    { "GL_EXT_blend_func_separate",   GL_EXT_blend_func_separate_functions },
    { "GL_EXT_compiled_vertex_array", GL_EXT_compiled_vertex_array_functions },
+   { "GL_EXT_multi_draw_arrays",     GL_EXT_multi_draw_arrays_functions },
    { "GL_EXT_polygon_offset",        GL_EXT_polygon_offset_functions },
    { "GL_EXT_texture_object",        GL_EXT_texture_object_functions },
    { "GL_EXT_vertex_array",          GL_EXT_vertex_array_functions },
+   { "GL_IBM_multimode_draw_arrays", GL_IBM_multimode_draw_arrays_functions },
    { "GL_MESA_window_pos",           GL_MESA_window_pos_functions },
    { "GL_NV_vertex_program",         GL_NV_vertex_program_functions },
    { NULL,                           NULL }
@@ -310,8 +320,10 @@ void driInitSingleExtension( GLcontext * ctx,
 	     */
 	    offset = _glapi_add_dispatch( functions, parameter_signature );
 	    if (offset == -1) {
+#if 0 /* this causes noise with egl */
 		fprintf(stderr, "DISPATCH ERROR! _glapi_add_dispatch failed "
 			"to add %s!\n", functions[0]);
+#endif
 	    }
 	    else if (ext->functions[i].remap_index != -1) {
 		driDispatchRemapTable[ ext->functions[i].remap_index ] = 
@@ -504,6 +516,9 @@ GLboolean driClipRectToFramebuffer( const GLframebuffer *buffer,
  *                      \c GLX_SWAP_UNDEFINED_OML.  See the
  *                      GLX_OML_swap_method extension spec for more details.
  * \param num_db_modes  Number of entries in \c db_modes.
+ * \param msaa_samples  Array of msaa sample count. 0 represents a visual
+ *                      without a multisample buffer.
+ * \param num_msaa_modes Number of entries in \c msaa_samples.
  * \param visType       GLX visual type.  Usually either \c GLX_TRUE_COLOR or
  *                      \c GLX_DIRECT_COLOR.
  * 
@@ -523,7 +538,8 @@ __DRIconfig **
 driCreateConfigs(GLenum fb_format, GLenum fb_type,
 		 const uint8_t * depth_bits, const uint8_t * stencil_bits,
 		 unsigned num_depth_stencil_bits,
-		 const GLenum * db_modes, unsigned num_db_modes)
+		 const GLenum * db_modes, unsigned num_db_modes,
+	 	 const u_int8_t * msaa_samples, unsigned num_msaa_modes)
 {
    static const uint8_t bits_table[4][4] = {
      /* R  G  B  A */
@@ -583,9 +599,7 @@ driCreateConfigs(GLenum fb_format, GLenum fb_type,
    int index;
    __DRIconfig **configs, **c;
    __GLcontextModes *modes;
-   unsigned i;
-   unsigned j;
-   unsigned k;
+   unsigned i, j, k, h;
    unsigned num_modes;
    unsigned num_accum_bits = 2;
 
@@ -658,7 +672,7 @@ driCreateConfigs(GLenum fb_format, GLenum fb_type,
 	 break;
    }
 
-   num_modes = num_depth_stencil_bits * num_db_modes * num_accum_bits;
+   num_modes = num_depth_stencil_bits * num_db_modes * num_accum_bits * num_msaa_modes;
    configs = _mesa_calloc((num_modes + 1) * sizeof *configs);
    if (configs == NULL)
        return NULL;
@@ -666,66 +680,72 @@ driCreateConfigs(GLenum fb_format, GLenum fb_type,
     c = configs;
     for ( k = 0 ; k < num_depth_stencil_bits ; k++ ) {
 	for ( i = 0 ; i < num_db_modes ; i++ ) {
-	    for ( j = 0 ; j < num_accum_bits ; j++ ) {
-		*c = _mesa_malloc (sizeof **c);
-		modes = &(*c)->modes;
-		c++;
-
-		memset(modes, 0, sizeof *modes);
-		modes->redBits   = bits[0];
-		modes->greenBits = bits[1];
-		modes->blueBits  = bits[2];
-		modes->alphaBits = bits[3];
-		modes->redMask   = masks[0];
-		modes->greenMask = masks[1];
-		modes->blueMask  = masks[2];
-		modes->alphaMask = masks[3];
-		modes->rgbBits   = modes->redBits + modes->greenBits
-		    + modes->blueBits + modes->alphaBits;
-
-		modes->accumRedBits   = 16 * j;
-		modes->accumGreenBits = 16 * j;
-		modes->accumBlueBits  = 16 * j;
-		modes->accumAlphaBits = (masks[3] != 0) ? 16 * j : 0;
-		modes->visualRating = (j == 0) ? GLX_NONE : GLX_SLOW_CONFIG;
-
-		modes->stencilBits = stencil_bits[k];
-		modes->depthBits = depth_bits[k];
-
-		modes->transparentPixel = GLX_NONE;
-		modes->transparentRed = GLX_DONT_CARE;
-		modes->transparentGreen = GLX_DONT_CARE;
-		modes->transparentBlue = GLX_DONT_CARE;
-		modes->transparentAlpha = GLX_DONT_CARE;
-		modes->transparentIndex = GLX_DONT_CARE;
-		modes->visualType = GLX_DONT_CARE;
-		modes->renderType = GLX_RGBA_BIT;
-		modes->drawableType = GLX_WINDOW_BIT;
-		modes->rgbMode = GL_TRUE;
-
-		if ( db_modes[i] == GLX_NONE ) {
-		    modes->doubleBufferMode = GL_FALSE;
-		}
-		else {
-		    modes->doubleBufferMode = GL_TRUE;
-		    modes->swapMethod = db_modes[i];
-		}
-
-		modes->haveAccumBuffer = ((modes->accumRedBits +
+	    for ( h = 0 ; h < num_msaa_modes; h++ ) {
+	    	for ( j = 0 ; j < num_accum_bits ; j++ ) {
+		    *c = _mesa_malloc (sizeof **c);
+		    modes = &(*c)->modes;
+		    c++;
+
+		    memset(modes, 0, sizeof *modes);
+		    modes->redBits   = bits[0];
+		    modes->greenBits = bits[1];
+		    modes->blueBits  = bits[2];
+		    modes->alphaBits = bits[3];
+		    modes->redMask   = masks[0];
+		    modes->greenMask = masks[1];
+		    modes->blueMask  = masks[2];
+		    modes->alphaMask = masks[3];
+		    modes->rgbBits   = modes->redBits + modes->greenBits
+		    	+ modes->blueBits + modes->alphaBits;
+
+		    modes->accumRedBits   = 16 * j;
+		    modes->accumGreenBits = 16 * j;
+		    modes->accumBlueBits  = 16 * j;
+		    modes->accumAlphaBits = (masks[3] != 0) ? 16 * j : 0;
+		    modes->visualRating = (j == 0) ? GLX_NONE : GLX_SLOW_CONFIG;
+
+		    modes->stencilBits = stencil_bits[k];
+		    modes->depthBits = depth_bits[k];
+
+		    modes->transparentPixel = GLX_NONE;
+		    modes->transparentRed = GLX_DONT_CARE;
+		    modes->transparentGreen = GLX_DONT_CARE;
+		    modes->transparentBlue = GLX_DONT_CARE;
+		    modes->transparentAlpha = GLX_DONT_CARE;
+		    modes->transparentIndex = GLX_DONT_CARE;
+		    modes->visualType = GLX_DONT_CARE;
+		    modes->renderType = GLX_RGBA_BIT;
+		    modes->drawableType = GLX_WINDOW_BIT;
+		    modes->rgbMode = GL_TRUE;
+
+		    if ( db_modes[i] == GLX_NONE ) {
+		    	modes->doubleBufferMode = GL_FALSE;
+		    }
+		    else {
+		    	modes->doubleBufferMode = GL_TRUE;
+		    	modes->swapMethod = db_modes[i];
+		    }
+
+		    modes->samples = msaa_samples[h];
+		    modes->sampleBuffers = modes->samples ? 1 : 0;
+
+
+		    modes->haveAccumBuffer = ((modes->accumRedBits +
 					   modes->accumGreenBits +
 					   modes->accumBlueBits +
 					   modes->accumAlphaBits) > 0);
-		modes->haveDepthBuffer = (modes->depthBits > 0);
-		modes->haveStencilBuffer = (modes->stencilBits > 0);
-
-		modes->bindToTextureRgb = GL_TRUE;
-		modes->bindToTextureRgba = GL_TRUE;
-		modes->bindToMipmapTexture = GL_FALSE;
-		modes->bindToTextureTargets = modes->rgbMode ?
-		    __DRI_ATTRIB_TEXTURE_1D_BIT |
-		    __DRI_ATTRIB_TEXTURE_2D_BIT |
-		    __DRI_ATTRIB_TEXTURE_RECTANGLE_BIT :
-		    0;
+		    modes->haveDepthBuffer = (modes->depthBits > 0);
+		    modes->haveStencilBuffer = (modes->stencilBits > 0);
+
+		    modes->bindToTextureRgb = GL_TRUE;
+		    modes->bindToTextureRgba = GL_TRUE;
+		    modes->bindToMipmapTexture = GL_FALSE;
+		    modes->bindToTextureTargets = modes->rgbMode ?
+		    	__DRI_ATTRIB_TEXTURE_1D_BIT |
+		    	__DRI_ATTRIB_TEXTURE_2D_BIT |
+		    	__DRI_ATTRIB_TEXTURE_RECTANGLE_BIT :
+		    	0;
+		}
 	    }
 	}
     }
@@ -734,9 +754,10 @@ driCreateConfigs(GLenum fb_format, GLenum fb_type,
     return configs;
 }
 
-const __DRIconfig **driConcatConfigs(__DRIconfig **a, __DRIconfig **b)
+__DRIconfig **driConcatConfigs(__DRIconfig **a,
+			       __DRIconfig **b)
 {
-    const __DRIconfig **all;
+    __DRIconfig **all;
     int i, j, index;
 
     i = 0;
diff --git a/src/mesa/drivers/dri/common/utils.h b/src/mesa/drivers/dri/common/utils.h
index 0c974dbff3b..9e9e5bc224b 100644
--- a/src/mesa/drivers/dri/common/utils.h
+++ b/src/mesa/drivers/dri/common/utils.h
@@ -131,9 +131,11 @@ extern __DRIconfig **
 driCreateConfigs(GLenum fb_format, GLenum fb_type,
 		 const uint8_t * depth_bits, const uint8_t * stencil_bits,
 		 unsigned num_depth_stencil_bits,
-		 const GLenum * db_modes, unsigned num_db_modes);
+		 const GLenum * db_modes, unsigned num_db_modes,
+    		 const uint8_t * msaa_samples, unsigned num_msaa_modes);
 
-const __DRIconfig **driConcatConfigs(__DRIconfig **a, __DRIconfig **b);
+__DRIconfig **driConcatConfigs(__DRIconfig **a,
+			       __DRIconfig **b);
 
 int
 driGetConfigAttrib(const __DRIconfig *config,
diff --git a/src/mesa/drivers/dri/common/vblank.c b/src/mesa/drivers/dri/common/vblank.c
index d610253fe6f..12aeaa108f7 100644
--- a/src/mesa/drivers/dri/common/vblank.c
+++ b/src/mesa/drivers/dri/common/vblank.c
@@ -130,9 +130,8 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
 
 
    if ( divisor != 0 ) {
-      unsigned int target = (unsigned int)target_msc;
-      unsigned int next = target;
-      unsigned int r;
+      int64_t next = target_msc;
+      int64_t r;
       int dont_wait = (target_msc == 0);
 
       do {
@@ -154,9 +153,9 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
 
 	 *msc = vblank_to_msc(priv, vbl.reply.sequence);
 
-         dont_wait = 0;
-         if (target_msc != 0 && *msc == target)
+         if (!dont_wait && *msc == next)
             break;
+         dont_wait = 0;
 
          /* Assuming the wait-done test fails, the next refresh to wait for
           * will be one that satisfies (MSC % divisor) == remainder.  The
@@ -165,11 +164,12 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
           * If this refresh has already happened, we add divisor to obtain 
           * the next refresh after the current one that will satisfy it.
           */
-         r = (*msc % (unsigned int)divisor);
-         next = (*msc - r + (unsigned int)remainder);
-         if (next <= *msc) next += (unsigned int)divisor;
+         r = ((uint64_t)*msc % divisor);
+         next = (*msc - r + remainder);
+         if (next <= *msc)
+	    next += divisor;
 
-      } while ( r != (unsigned int)remainder );
+      } while (r != remainder);
    }
    else {
       /* If the \c divisor is zero, just wait until the MSC is greater
diff --git a/src/mesa/drivers/dri/dri.pc.in b/src/mesa/drivers/dri/dri.pc.in
index c47ee9c7e7d..695aa6cfd66 100644
--- a/src/mesa/drivers/dri/dri.pc.in
+++ b/src/mesa/drivers/dri/dri.pc.in
@@ -7,4 +7,5 @@ dridriverdir=@DRI_DRIVER_DIR@
 Name: dri
 Description: Direct Rendering Infrastructure
 Version: @VERSION@
+Requires.private: @DRI_PC_REQ_PRIV@
 Cflags: -I${includedir}
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.c b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
index 679f8561d24..00bdcec7a69 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.c
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
@@ -260,6 +260,8 @@ ffbCreateContext(const __GLcontextModes *mesaVis,
 	ctx->Const.MaxLineWidthAA = 1.0;
 	ctx->Const.LineWidthGranularity = 1.0;
 
+	ctx->Const.MaxDrawBuffers = 1;
+
 	/* Instead of having GCC emit these constants a zillion times
 	 * everywhere in the driver, put them here.
 	 */
@@ -626,6 +628,7 @@ ffbFillInModes( __DRIscreenPrivate *psp,
 
    uint8_t depth_bits_array[3];
    uint8_t stencil_bits_array[3];
+   uint8_t msaa_samples_array[1];
 
    depth_bits_array[0] = 0;
    depth_bits_array[1] = depth_bits;
@@ -639,6 +642,8 @@ ffbFillInModes( __DRIscreenPrivate *psp,
    stencil_bits_array[1] = 0;
    stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+   msaa_samples_array[0] = 0;
+
    depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
    back_buffer_factor  = (have_back_buffer) ? 3 : 1;
 
@@ -654,9 +659,10 @@ ffbFillInModes( __DRIscreenPrivate *psp,
    configs = driCreateConfigs(fb_format, fb_type,
 			      depth_bits_array, stencil_bits_array,
 			      depth_buffer_factor, back_buffer_modes,
-			      back_buffer_factor);
+			      back_buffer_factor,
+                               msaa_samples_array, 1);
    if (configs == NULL) {
-    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
               __LINE__);
       return NULL;
    }
diff --git a/src/mesa/drivers/dri/gamma/gamma_context.c b/src/mesa/drivers/dri/gamma/gamma_context.c
index c91bedce3a8..b0ac299daae 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.c
+++ b/src/mesa/drivers/dri/gamma/gamma_context.c
@@ -133,6 +133,8 @@ GLboolean gammaCreateContext( const __GLcontextModes *glVisual,
    ctx->Const.MaxPointSizeAA = 16.0; 
    ctx->Const.PointSizeGranularity = 0.25;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    gmesa->texHeap = mmInit( 0, gmesa->gammaScreen->textureSize );
 
    make_empty_list(&gmesa->TexObjList);
diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile
new file mode 100644
index 00000000000..ac7e1de9285
--- /dev/null
+++ b/src/mesa/drivers/dri/glcore/Makefile
@@ -0,0 +1,84 @@
+# src/mesa/drivers/dri/glcore/Makefile
+
+TOP = ../../../../..
+include $(TOP)/configs/current
+
+LIBNAME = glcore_dri.so
+
+DRIVER_SOURCES = glcore_driver.c \
+		 $(TOP)/src/mesa/drivers/common/driverfuncs.c \
+		 ../common/dri_util.c
+
+C_SOURCES = \
+	$(DRIVER_SOURCES) \
+	$(DRI_SOURCES) 
+
+
+# Include directories
+INCLUDE_DIRS = \
+	-I. \
+	-I../common \
+	-I../dri_client \
+	-I../dri_client/imports \
+	-Iserver \
+	-I$(TOP)/include \
+	-I$(DRM_SOURCE_PATH)/shared-core \
+	-I$(TOP)/src/mesa \
+	-I$(TOP)/src/mesa/main \
+	-I$(TOP)/src/mesa/glapi \
+	-I$(TOP)/src/mesa/math \
+	-I$(TOP)/src/mesa/transform \
+	-I$(TOP)/src/mesa/shader \
+	-I$(TOP)/src/mesa/swrast \
+	-I$(TOP)/src/mesa/swrast_setup
+
+# Core Mesa objects
+MESA_MODULES = $(TOP)/src/mesa/libmesa.a
+
+# Libraries that the driver shared lib depends on
+LIB_DEPS = -lm -lpthread -lc
+# LIB_DEPS = -lGL -lm -lpthread -lc
+
+
+ASM_SOURCES = 
+
+OBJECTS = $(C_SOURCES:.c=.o) \
+	  $(ASM_SOURCES:.S=.o) 
+
+
+##### RULES #####
+
+.c.o:
+	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES) $< -o $@
+
+.S.o:
+	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $(DEFINES)  $< -o $@
+
+
+##### TARGETS #####
+
+default: depend $(TOP)/$(LIB_DIR)/$(LIBNAME)
+
+
+$(TOP)/$(LIB_DIR)/$(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile
+	CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix -install $(TOP)/$(LIB_DIR) \
+		$(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES)
+
+
+depend: $(C_SOURCES) $(ASM_SOURCES)
+	rm -f depend
+	touch depend
+	$(MKDEP) $(MKDEP_OPTIONS) $(INCLUDE_DIRS) $(C_SOURCES) $(ASM_SOURCES) \
+		> /dev/null 
+
+
+# Emacs tags
+tags:
+	etags `find . -name \*.[ch]` `find ../include`
+
+
+clean:
+	-rm -f *.o server/*.o
+
+
+include depend
diff --git a/src/mesa/drivers/dri/i810/i810context.c b/src/mesa/drivers/dri/i810/i810context.c
index c281a4990e4..67856556865 100644
--- a/src/mesa/drivers/dri/i810/i810context.c
+++ b/src/mesa/drivers/dri/i810/i810context.c
@@ -63,11 +63,6 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "drirenderbuffer.h"
 #include "utils.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
-#include "extension_helper.h"
-
 #ifndef I810_DEBUG
 int I810_DEBUG = (0);
 #endif
@@ -123,14 +118,11 @@ static void i810BufferSize(GLframebuffer *buffer, GLuint *width, GLuint *height)
  */
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_env_add",            NULL },
     { "GL_ARB_texture_env_combine",        NULL },
     { "GL_ARB_texture_env_crossbar",       NULL },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_stencil_wrap",               NULL },
     { "GL_EXT_texture_edge_clamp",         NULL },
     { "GL_EXT_texture_env_combine",        NULL },
diff --git a/src/mesa/drivers/dri/i810/i810screen.c b/src/mesa/drivers/dri/i810/i810screen.c
index 48603f5d79f..9a5a39c1dbc 100644
--- a/src/mesa/drivers/dri/i810/i810screen.c
+++ b/src/mesa/drivers/dri/i810/i810screen.c
@@ -77,6 +77,7 @@ i810FillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[2];
     uint8_t stencil_bits_array[2];
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
@@ -88,13 +89,16 @@ i810FillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[0] = 0;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
     configs = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor,
-			       back_buffer_modes, back_buffer_factor);
+			       back_buffer_modes, back_buffer_factor,
+                               msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 5858e0ee9fd..954a7e2af1f 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -18,8 +18,9 @@ DRIVER_SOURCES = \
 	intel_regions.c \
 	intel_buffer_objects.c \
 	intel_batchbuffer.c \
+	intel_clear.c \
+	intel_extensions.c \
 	intel_mipmap_tree.c \
-	i915_tex_layout.c \
 	intel_tex_layout.c \
 	intel_tex_image.c \
 	intel_tex_subimage.c \
@@ -34,7 +35,9 @@ DRIVER_SOURCES = \
 	intel_pixel_read.c \
 	intel_buffers.c \
 	intel_blit.c \
+	intel_swapbuffers.c \
 	i915_tex.c \
+	i915_tex_layout.c \
 	i915_texstate.c \
 	i915_context.c \
 	i915_debug.c \
@@ -50,8 +53,7 @@ DRIVER_SOURCES = \
 	intel_span.c \
 	intel_state.c \
 	intel_tris.c \
-	intel_fbo.c \
-	intel_depthstencil.c
+	intel_fbo.c
 
 C_SOURCES = \
 	$(COMMON_SOURCES) \
diff --git a/src/mesa/drivers/dri/i915/i830_context.c b/src/mesa/drivers/dri/i915/i830_context.c
index 09b1ec922ff..fdce8af31f4 100644
--- a/src/mesa/drivers/dri/i915/i830_context.c
+++ b/src/mesa/drivers/dri/i915/i830_context.c
@@ -97,6 +97,8 @@ i830CreateContext(const __GLcontextModes * mesaVis,
    ctx->Const.MaxTextureRectSize = (1 << 11);
    ctx->Const.MaxTextureUnits = I830_TEX_UNITS;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
                       18 * sizeof(GLfloat));
 
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index 3b3ff2bceda..8fc8aa5f900 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -297,7 +297,7 @@ i830_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH(40, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(30, IGNORE_CLIPRECTS);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -491,8 +491,17 @@ i830_emit_state(struct intel_context *intel)
    }
 
    if (dirty & I830_UPLOAD_BUFFERS) {
+      GLuint count = 9; 
+
       DBG("I830_UPLOAD_BUFFERS:\n");
-      BEGIN_BATCH(I830_DEST_SETUP_SIZE + 2, IGNORE_CLIPRECTS);
+
+      if (state->depth_region)
+          count += 3;
+
+      if (intel->constant_cliprect)
+          count += 6;
+
+      BEGIN_BATCH(count, IGNORE_CLIPRECTS);
       OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR0]);
       OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR1]);
       OUT_RELOC(state->draw_region->buffer,
@@ -557,6 +566,8 @@ i830_emit_state(struct intel_context *intel)
          OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S4]);
          OUT_BATCH(state->Tex[i][I830_TEXREG_MCS]);
          OUT_BATCH(state->Tex[i][I830_TEXREG_CUBE]);
+
+         ADVANCE_BATCH();
       }
 
       if (dirty & I830_UPLOAD_TEXBLEND(i)) {
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index e0ddc7fd613..6e2d41e19ab 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -50,15 +50,6 @@
  * Mesa's Driver Functions
  ***************************************/
 
-static const struct dri_extension i915_extensions[] = {
-   {"GL_ARB_depth_texture", NULL},
-   {"GL_ARB_fragment_program", NULL},
-   {"GL_ARB_shadow", NULL},
-   {"GL_ARB_texture_non_power_of_two", NULL},
-   {"GL_EXT_shadow_funcs", NULL},
-   {NULL, NULL}
-};
-
 /* Override intel default.
  */
 static void
@@ -170,10 +161,8 @@ i915CreateContext(const __GLcontextModes * mesaVis,
    ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0; /* I don't think we have one */
 
    ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
-   ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
-
-   driInitExtensions(ctx, i915_extensions, GL_FALSE);
 
+   ctx->Const.MaxDrawBuffers = 1;
 
    _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
                       36 * sizeof(GLfloat));
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 8bd761ec6a1..f091d600c35 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -33,6 +33,7 @@
 #include "shader/prog_parameter.h"
 #include "shader/program.h"
 #include "shader/programopt.h"
+#include "shader/prog_print.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
@@ -1048,6 +1049,9 @@ i915ProgramStringNotify(GLcontext * ctx,
          _mesa_append_fog_code(ctx, &p->FragProg);
          p->FragProg.FogOption = GL_NONE;
       }
+
+      if (INTEL_DEBUG & DEBUG_STATE)
+	 _mesa_print_program(prog);
    }
 
    _tnl_program_string(ctx, target, prog);
@@ -1105,30 +1109,14 @@ i915ValidateFragmentProgram(struct i915_context *i915)
       EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4);
    }
 
-   if ((inputsRead & (FRAG_BIT_COL1 | FRAG_BIT_FOGC)) ||
-       i915->vertex_fog != I915_FOG_NONE) {
-
-      if (inputsRead & FRAG_BIT_COL1) {
-         intel->specoffset = offset / 4;
-         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3);
-      }
-      else
-         EMIT_PAD(3);
-
-      if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE)
-         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1);
-      else
-         EMIT_PAD(1);
+   if (inputsRead & FRAG_BIT_COL1) {
+       intel->specoffset = offset / 4;
+       EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_4UB_4F_BGRA, S4_VFMT_SPEC_FOG, 4);
    }
 
-   /* XXX this was disabled, but enabling this code helped fix the Glean
-    * tfragprog1 fog tests.
-    */
-#if 1
    if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE) {
       EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1F, S4_VFMT_FOG_PARAM, 4);
    }
-#endif
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
       if (inputsRead & FRAG_BIT_TEX(i)) {
diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c
index 9d04358e4f3..814fb59fd34 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -41,6 +41,7 @@
 #include "intel_fbo.h"
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
+#include "intel_buffers.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
@@ -301,6 +302,65 @@ i915DepthMask(GLcontext * ctx, GLboolean flag)
       i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_WRITE_ENABLE;
 }
 
+
+
+/**
+ * Update the viewport transformation matrix.  Depends on:
+ *  - viewport pos/size
+ *  - depthrange
+ *  - window pos/size or FBO size
+ */
+void
+intelCalcViewport(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   const GLfloat *v = ctx->Viewport._WindowMap.m;
+   const GLfloat depthScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
+   GLfloat *m = intel->ViewportMatrix.m;
+   GLfloat yScale, yBias;
+
+   if (ctx->DrawBuffer->Name) {
+      /* User created FBO */
+      /* y=0=bottom */
+      yScale = 1.0;
+      yBias = 0.0;
+   }
+   else {
+      /* window buffer, y=0=top */
+      yScale = -1.0;
+      yBias = (intel->driDrawable) ? intel->driDrawable->h : 0.0F;
+   }
+
+   m[MAT_SX] = v[MAT_SX];
+   m[MAT_TX] = v[MAT_TX];
+
+   m[MAT_SY] = v[MAT_SY] * yScale;
+   m[MAT_TY] = v[MAT_TY] * yScale + yBias;
+
+   m[MAT_SZ] = v[MAT_SZ] * depthScale;
+   m[MAT_TZ] = v[MAT_TZ] * depthScale;
+}
+
+
+/** Called from ctx->Driver.Viewport() */
+static void
+i915Viewport(GLcontext * ctx,
+              GLint x, GLint y, GLsizei width, GLsizei height)
+{
+   intelCalcViewport(ctx);
+
+   intel_viewport(ctx, x, y, width, height);
+}
+
+
+/** Called from ctx->Driver.DepthRange() */
+static void
+i915DepthRange(GLcontext * ctx, GLclampd nearval, GLclampd farval)
+{
+   intelCalcViewport(ctx);
+}
+
+
 /* =============================================================
  * Polygon stipple
  *
@@ -569,7 +629,7 @@ i915_update_fog(GLcontext * ctx)
    GLboolean enabled;
    GLboolean try_pixel_fog;
 
-   if (ctx->FragmentProgram._Active) {
+   if (ctx->FragmentProgram._Current) {
       /* Pull in static fog state from program */
       mode = ctx->FragmentProgram._Current->FogOption;
       enabled = (mode != GL_NONE);
@@ -964,6 +1024,8 @@ i915InitStateFunctions(struct dd_function_table *functions)
    functions->StencilFuncSeparate = i915StencilFuncSeparate;
    functions->StencilMaskSeparate = i915StencilMaskSeparate;
    functions->StencilOpSeparate = i915StencilOpSeparate;
+   functions->DepthRange = i915DepthRange;
+   functions->Viewport = i915Viewport;
 }
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index d1b0dcdf319..adbb52a3a3b 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -295,6 +295,13 @@ i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
            wt == GL_CLAMP_TO_BORDER || wr == GL_CLAMP_TO_BORDER))
          return GL_FALSE;
 
+      /* Only support TEXCOORDMODE_CLAMP_EDGE and TEXCOORDMODE_CUBE (not 
+       * used) when using cube map texture coordinates
+       */
+      if (tObj->Target == GL_TEXTURE_CUBE_MAP_ARB &&
+          (((ws != GL_CLAMP) && (ws != GL_CLAMP_TO_EDGE)) ||
+           ((wt != GL_CLAMP) && (wt != GL_CLAMP_TO_EDGE))))
+          return GL_FALSE;
 
       state[I915_TEXREG_SS3] = ss3;     /* SS3_NORMALIZED_COORDS */
 
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index e79c955d64d..3f6d282d342 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -173,7 +173,7 @@ i915_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH(200, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(20, IGNORE_CLIPRECTS);
 
    OUT_BATCH(_3DSTATE_AA_CMD |
              AA_LINE_ECAAR_WIDTH_ENABLE |
@@ -376,9 +376,18 @@ i915_emit_state(struct intel_context *intel)
    }
 
    if (dirty & I915_UPLOAD_BUFFERS) {
+      GLuint count = 9;
+
       if (INTEL_DEBUG & DEBUG_STATE)
          fprintf(stderr, "I915_UPLOAD_BUFFERS:\n");
-      BEGIN_BATCH(I915_DEST_SETUP_SIZE + 2, IGNORE_CLIPRECTS);
+
+      if (state->depth_region)
+          count += 3;
+
+      if (intel->constant_cliprect)
+          count += 6;
+
+      BEGIN_BATCH(count, IGNORE_CLIPRECTS);
       OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR0]);
       OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR1]);
       OUT_RELOC(state->draw_region->buffer,
diff --git a/src/mesa/drivers/dri/i915/intel_clear.c b/src/mesa/drivers/dri/i915/intel_clear.c
new file mode 120000
index 00000000000..9a2a742a0d9
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_clear.c
@@ -0,0 +1 @@
+../intel/intel_clear.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_depthstencil.c b/src/mesa/drivers/dri/i915/intel_depthstencil.c
deleted file mode 120000
index 4ac4ae690a3..00000000000
--- a/src/mesa/drivers/dri/i915/intel_depthstencil.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_depthstencil.c
-\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_extensions.c b/src/mesa/drivers/dri/i915/intel_extensions.c
new file mode 120000
index 00000000000..a2f3e8cd208
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_extensions.c
@@ -0,0 +1 @@
+../intel/intel_extensions.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index 467abe4d7cb..410052b3c2b 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -117,6 +117,26 @@ intelDmaPrimitive(struct intel_context *intel, GLenum prim)
    intel_set_prim(intel, hw_prim[prim]);
 }
 
+static inline GLuint intel_get_vb_max(struct intel_context *intel)
+{
+   GLuint ret;
+
+   if (intel->intelScreen->no_vbo)
+      ret = intel->batch->size - 1500;
+   else
+      ret = INTEL_VB_SIZE;
+   ret /= (intel->vertex_size * 4);
+   return ret;
+}
+
+static inline GLuint intel_get_current_max(struct intel_context *intel)
+{
+
+   if (intel->intelScreen->no_vbo)
+      return intel_get_vb_max(intel);
+   else
+      return (INTEL_VB_SIZE - intel->prim.current_offset) / (intel->vertex_size * 4);
+}
 
 #define LOCAL_VARS struct intel_context *intel = intel_context(ctx)
 #define INIT( prim ) 				\
@@ -126,9 +146,8 @@ do {						\
 
 #define FLUSH() INTEL_FIREVERTICES(intel)
 
-#define GET_SUBSEQUENT_VB_MAX_VERTS() (INTEL_VB_SIZE / (intel->vertex_size * 4))
-#define GET_CURRENT_VB_MAX_VERTS() \
-   ((INTEL_VB_SIZE - intel->prim.current_offset) / (intel->vertex_size * 4))
+#define GET_SUBSEQUENT_VB_MAX_VERTS() intel_get_vb_max(intel)
+#define GET_CURRENT_VB_MAX_VERTS() intel_get_current_max(intel)
 
 #define ALLOC_VERTS(nr) intel_get_prim_space(intel, nr)
 
diff --git a/src/mesa/drivers/dri/i915/intel_state.c b/src/mesa/drivers/dri/i915/intel_state.c
index 09aa62da361..519672fc359 100644..120000
--- a/src/mesa/drivers/dri/i915/intel_state.c
+++ b/src/mesa/drivers/dri/i915/intel_state.c
@@ -1,295 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/enums.h"
-#include "main/colormac.h"
-#include "main/dd.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_fbo.h"
-#include "intel_regions.h"
-#include "swrast/swrast.h"
-
-int 
-intel_translate_shadow_compare_func( GLenum func )
-{
-   switch(func) {
-   case GL_NEVER: 
-       return COMPAREFUNC_ALWAYS; 
-   case GL_LESS: 
-       return COMPAREFUNC_LEQUAL; 
-   case GL_LEQUAL: 
-       return COMPAREFUNC_LESS;
-   case GL_GREATER: 
-       return COMPAREFUNC_GEQUAL; 
-   case GL_GEQUAL: 
-      return COMPAREFUNC_GREATER; 
-   case GL_NOTEQUAL: 
-      return COMPAREFUNC_EQUAL; 
-   case GL_EQUAL: 
-      return COMPAREFUNC_NOTEQUAL; 
-   case GL_ALWAYS: 
-       return COMPAREFUNC_NEVER; 
-   }
-
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_NEVER; 
-}
-
-int
-intel_translate_compare_func(GLenum func)
-{
-   switch (func) {
-   case GL_NEVER:
-      return COMPAREFUNC_NEVER;
-   case GL_LESS:
-      return COMPAREFUNC_LESS;
-   case GL_LEQUAL:
-      return COMPAREFUNC_LEQUAL;
-   case GL_GREATER:
-      return COMPAREFUNC_GREATER;
-   case GL_GEQUAL:
-      return COMPAREFUNC_GEQUAL;
-   case GL_NOTEQUAL:
-      return COMPAREFUNC_NOTEQUAL;
-   case GL_EQUAL:
-      return COMPAREFUNC_EQUAL;
-   case GL_ALWAYS:
-      return COMPAREFUNC_ALWAYS;
-   }
-
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_ALWAYS;
-}
-
-int
-intel_translate_stencil_op(GLenum op)
-{
-   switch (op) {
-   case GL_KEEP:
-      return STENCILOP_KEEP;
-   case GL_ZERO:
-      return STENCILOP_ZERO;
-   case GL_REPLACE:
-      return STENCILOP_REPLACE;
-   case GL_INCR:
-      return STENCILOP_INCRSAT;
-   case GL_DECR:
-      return STENCILOP_DECRSAT;
-   case GL_INCR_WRAP:
-      return STENCILOP_INCR;
-   case GL_DECR_WRAP:
-      return STENCILOP_DECR;
-   case GL_INVERT:
-      return STENCILOP_INVERT;
-   default:
-      return STENCILOP_ZERO;
-   }
-}
-
-int
-intel_translate_blend_factor(GLenum factor)
-{
-   switch (factor) {
-   case GL_ZERO:
-      return BLENDFACT_ZERO;
-   case GL_SRC_ALPHA:
-      return BLENDFACT_SRC_ALPHA;
-   case GL_ONE:
-      return BLENDFACT_ONE;
-   case GL_SRC_COLOR:
-      return BLENDFACT_SRC_COLR;
-   case GL_ONE_MINUS_SRC_COLOR:
-      return BLENDFACT_INV_SRC_COLR;
-   case GL_DST_COLOR:
-      return BLENDFACT_DST_COLR;
-   case GL_ONE_MINUS_DST_COLOR:
-      return BLENDFACT_INV_DST_COLR;
-   case GL_ONE_MINUS_SRC_ALPHA:
-      return BLENDFACT_INV_SRC_ALPHA;
-   case GL_DST_ALPHA:
-      return BLENDFACT_DST_ALPHA;
-   case GL_ONE_MINUS_DST_ALPHA:
-      return BLENDFACT_INV_DST_ALPHA;
-   case GL_SRC_ALPHA_SATURATE:
-      return BLENDFACT_SRC_ALPHA_SATURATE;
-   case GL_CONSTANT_COLOR:
-      return BLENDFACT_CONST_COLOR;
-   case GL_ONE_MINUS_CONSTANT_COLOR:
-      return BLENDFACT_INV_CONST_COLOR;
-   case GL_CONSTANT_ALPHA:
-      return BLENDFACT_CONST_ALPHA;
-   case GL_ONE_MINUS_CONSTANT_ALPHA:
-      return BLENDFACT_INV_CONST_ALPHA;
-   }
-
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, factor);
-   return BLENDFACT_ZERO;
-}
-
-int
-intel_translate_logic_op(GLenum opcode)
-{
-   switch (opcode) {
-   case GL_CLEAR:
-      return LOGICOP_CLEAR;
-   case GL_AND:
-      return LOGICOP_AND;
-   case GL_AND_REVERSE:
-      return LOGICOP_AND_RVRSE;
-   case GL_COPY:
-      return LOGICOP_COPY;
-   case GL_COPY_INVERTED:
-      return LOGICOP_COPY_INV;
-   case GL_AND_INVERTED:
-      return LOGICOP_AND_INV;
-   case GL_NOOP:
-      return LOGICOP_NOOP;
-   case GL_XOR:
-      return LOGICOP_XOR;
-   case GL_OR:
-      return LOGICOP_OR;
-   case GL_OR_INVERTED:
-      return LOGICOP_OR_INV;
-   case GL_NOR:
-      return LOGICOP_NOR;
-   case GL_EQUIV:
-      return LOGICOP_EQUIV;
-   case GL_INVERT:
-      return LOGICOP_INV;
-   case GL_OR_REVERSE:
-      return LOGICOP_OR_RVRSE;
-   case GL_NAND:
-      return LOGICOP_NAND;
-   case GL_SET:
-      return LOGICOP_SET;
-   default:
-      return LOGICOP_SET;
-   }
-}
-
-
-static void
-intelClearColor(GLcontext * ctx, const GLfloat color[4])
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLubyte clear[4];
-
-   CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
-   CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
-
-   /* compute both 32 and 16-bit clear values */
-   intel->ClearColor8888 = INTEL_PACKCOLOR8888(clear[0], clear[1],
-                                               clear[2], clear[3]);
-   intel->ClearColor565 = INTEL_PACKCOLOR565(clear[0], clear[1], clear[2]);
-}
-
-
-/**
- * Update the viewport transformation matrix.  Depends on:
- *  - viewport pos/size
- *  - depthrange
- *  - window pos/size or FBO size
- */
-static void
-intelCalcViewport(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-   const GLfloat *v = ctx->Viewport._WindowMap.m;
-   const GLfloat depthScale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
-   GLfloat *m = intel->ViewportMatrix.m;
-   GLfloat yScale, yBias;
-
-   if (ctx->DrawBuffer->Name) {
-      /* User created FBO */
-      struct intel_renderbuffer *irb
-         = intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
-      if (irb && !irb->RenderToTexture) {
-         /* y=0=top */
-         yScale = -1.0;
-         yBias = irb->Base.Height;
-      }
-      else {
-         /* y=0=bottom */
-         yScale = 1.0;
-         yBias = 0.0;
-      }
-   }
-   else {
-      /* window buffer, y=0=top */
-      yScale = -1.0;
-      yBias = (intel->driDrawable) ? intel->driDrawable->h : 0.0F;
-   }
-
-   m[MAT_SX] = v[MAT_SX];
-   m[MAT_TX] = v[MAT_TX];
-
-   m[MAT_SY] = v[MAT_SY] * yScale;
-   m[MAT_TY] = v[MAT_TY] * yScale + yBias;
-
-   m[MAT_SZ] = v[MAT_SZ] * depthScale;
-   m[MAT_TZ] = v[MAT_TZ] * depthScale;
-}
-
-static void
-intelViewport(GLcontext * ctx,
-              GLint x, GLint y, GLsizei width, GLsizei height)
-{
-   intelCalcViewport(ctx);
-}
-
-static void
-intelDepthRange(GLcontext * ctx, GLclampd nearval, GLclampd farval)
-{
-   intelCalcViewport(ctx);
-}
-
-/* Fallback to swrast for select and feedback.
- */
-static void
-intelRenderMode(GLcontext * ctx, GLenum mode)
-{
-   struct intel_context *intel = intel_context(ctx);
-   FALLBACK(intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER));
-}
-
-
-void
-intelInitStateFuncs(struct dd_function_table *functions)
-{
-   functions->RenderMode = intelRenderMode;
-   functions->Viewport = intelViewport;
-   functions->DepthRange = intelDepthRange;
-   functions->ClearColor = intelClearColor;
-}
+../intel/intel_state.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_swapbuffers.c b/src/mesa/drivers/dri/i915/intel_swapbuffers.c
new file mode 120000
index 00000000000..148d5215aa6
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_swapbuffers.c
@@ -0,0 +1 @@
+../intel/intel_swapbuffers.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index 06210035af5..e80996580c7 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -61,9 +61,100 @@ static void intelRenderPrimitive(GLcontext * ctx, GLenum prim);
 static void intelRasterPrimitive(GLcontext * ctx, GLenum rprim,
                                  GLuint hwprim);
 
+static void
+intel_flush_inline_primitive(struct intel_context *intel)
+{
+   GLuint used = intel->batch->ptr - intel->prim.start_ptr;
+
+   assert(intel->prim.primitive != ~0);
+
+/*    _mesa_printf("/\n"); */
+
+   if (used < 8)
+      goto do_discard;
+
+   *(int *) intel->prim.start_ptr = (_3DPRIMITIVE |
+                                     intel->prim.primitive | (used / 4 - 2));
+
+   goto finished;
+
+ do_discard:
+   intel->batch->ptr -= used;
+
+ finished:
+   intel->prim.primitive = ~0;
+   intel->prim.start_ptr = 0;
+   intel->prim.flush = 0;
+}
+
+static void intel_start_inline(struct intel_context *intel, uint32_t prim)
+{
+   BATCH_LOCALS;
+   uint32_t batch_flags = LOOP_CLIPRECTS;
+
+   intel->vtbl.emit_state(intel);
+
+   intel->no_batch_wrap = GL_TRUE;
+
+   /*_mesa_printf("%s *", __progname);*/
+
+   /* Emit a slot which will be filled with the inline primitive
+    * command later.
+    */
+   BEGIN_BATCH(2, batch_flags);
+   OUT_BATCH(0);
+
+   assert((intel->batch->dirty_state & (1<<1)) == 0);
+
+   intel->prim.start_ptr = intel->batch->ptr;
+   intel->prim.primitive = prim;
+   intel->prim.flush = intel_flush_inline_primitive;
+
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+   intel->no_batch_wrap = GL_FALSE;
+/*    _mesa_printf(">"); */
+}
+
+static void intel_wrap_inline(struct intel_context *intel)
+{
+   GLuint prim = intel->prim.primitive;
+
+   intel_flush_inline_primitive(intel);
+   intel_batchbuffer_flush(intel->batch);
+   intel_start_inline(intel, prim);  /* ??? */
+}
+
+static GLuint *intel_extend_inline(struct intel_context *intel, GLuint dwords)
+{
+   GLuint sz = dwords * sizeof(GLuint);
+   GLuint *ptr;
+
+   assert(intel->prim.flush == intel_flush_inline_primitive);
+
+   if (intel_batchbuffer_space(intel->batch) < sz)
+      intel_wrap_inline(intel);
+
+/*    _mesa_printf("."); */
+
+   intel->vtbl.assert_not_dirty(intel);
+
+   ptr = (GLuint *) intel->batch->ptr;
+   intel->batch->ptr += sz;
+
+   return ptr;
+}
+
 /** Sets the primitive type for a primitive sequence, flushing as needed. */
 void intel_set_prim(struct intel_context *intel, uint32_t prim)
 {
+   /* if we have no VBOs */
+
+   if (intel->intelScreen->no_vbo) {
+      intel_start_inline(intel, prim);
+      return;
+   }
    if (prim != intel->prim.primitive) {
       INTEL_FIREVERTICES(intel);
       intel->prim.primitive = prim;
@@ -75,6 +166,10 @@ uint32_t *intel_get_prim_space(struct intel_context *intel, unsigned int count)
 {
    uint32_t *addr;
 
+   if (intel->intelScreen->no_vbo) {
+      return intel_extend_inline(intel, count * intel->vertex_size);
+   }
+
    /* Check for space in the existing VB */
    if (intel->prim.vb_bo == NULL ||
        (intel->prim.current_offset +
@@ -118,8 +213,7 @@ void intel_flush_prim(struct intel_context *intel)
       return;
 
    /* Clear the current prims out of the context state so that a batch flush
-    * flush triggered by wait_flips or emit_state doesn't loop back to
-    * flush_prim again.
+    * flush triggered by emit_state doesn't loop back to flush_prim again.
     */
    vb_bo = intel->prim.vb_bo;
    dri_bo_reference(vb_bo);
@@ -131,8 +225,6 @@ void intel_flush_prim(struct intel_context *intel)
       intel->prim.start_offset = ALIGN(intel->prim.start_offset, 128);
    intel->prim.flush = NULL;
 
-   intel_wait_flips(intel);
-
    intel->vtbl.emit_state(intel);
 
    aper_array[0] = intel->batch->buf;
@@ -155,7 +247,7 @@ void intel_flush_prim(struct intel_context *intel)
 
 #if 0
    printf("emitting %d..%d=%d vertices size %d\n", offset,
-	  intel->prim.current_offset, intel->prim.count,
+	  intel->prim.current_offset, count,
 	  intel->vertex_size * 4);
 #endif
 
@@ -186,7 +278,7 @@ void intel_flush_prim(struct intel_context *intel)
       OUT_RELOC(vb_bo, I915_GEM_DOMAIN_VERTEX, 0,
 		offset | (intel->vertex_size << S0_VB_PITCH_SHIFT_830) |
 		S0_VB_ENABLE_830);
-      /* S1
+      /* S2
        * This is somewhat unfortunate -- VB width is tied up with
        * vertex format data that we've already uploaded through
        * _3DSTATE_VFT[01]_CMD.  We may want to replace emits of VFT state with
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 005460f3547..2934414d99a 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -9,9 +9,10 @@ DRIVER_SOURCES = \
 	intel_blit.c \
 	intel_buffer_objects.c \
 	intel_buffers.c \
+	intel_clear.c \
 	intel_context.c \
 	intel_decode.c \
-	intel_depthstencil.c \
+	intel_extensions.c \
 	intel_fbo.c \
 	intel_mipmap_tree.c \
 	intel_regions.c \
@@ -22,6 +23,7 @@ DRIVER_SOURCES = \
 	intel_pixel_copy.c \
 	intel_pixel_draw.c \
 	intel_state.c \
+	intel_swapbuffers.c \
 	intel_tex.c \
 	intel_tex_copy.c \
 	intel_tex_format.c \
@@ -49,7 +51,6 @@ DRIVER_SOURCES = \
 	brw_gs.c \
 	brw_gs_emit.c \
 	brw_gs_state.c \
-	brw_metaops.c \
 	brw_misc_state.c \
 	brw_program.c \
 	brw_queryobj.c \
@@ -68,7 +69,6 @@ DRIVER_SOURCES = \
 	brw_vs_constval.c \
 	brw_vs_emit.c \
 	brw_vs_state.c \
-	brw_vs_tnl.c \
 	brw_vtbl.c \
 	brw_wm.c \
 	brw_wm_debug.c \
diff --git a/src/mesa/drivers/dri/i965/brw_cc.c b/src/mesa/drivers/dri/i965/brw_cc.c
index fa8121e02d3..82370162f59 100644
--- a/src/mesa/drivers/dri/i965/brw_cc.c
+++ b/src/mesa/drivers/dri/i965/brw_cc.c
@@ -83,59 +83,60 @@ struct brw_cc_unit_key {
 static void
 cc_unit_populate_key(struct brw_context *brw, struct brw_cc_unit_key *key)
 {
-   struct gl_stencil_attrib *stencil = brw->attribs.Stencil;
+   GLcontext *ctx = &brw->intel.ctx;
+   const unsigned back = ctx->Stencil._BackFace;
 
    memset(key, 0, sizeof(*key));
 
-   key->stencil = stencil->Enabled;
-   key->stencil_two_side = stencil->_TestTwoSide;
+   key->stencil = ctx->Stencil.Enabled;
+   key->stencil_two_side = ctx->Stencil._TestTwoSide;
 
    if (key->stencil) {
-      key->stencil_func[0] = stencil->Function[0];
-      key->stencil_fail_op[0] = stencil->FailFunc[0];
-      key->stencil_pass_depth_fail_op[0] = stencil->ZFailFunc[0];
-      key->stencil_pass_depth_pass_op[0] = stencil->ZPassFunc[0];
-      key->stencil_ref[0] = stencil->Ref[0];
-      key->stencil_write_mask[0] = stencil->WriteMask[0];
-      key->stencil_test_mask[0] = stencil->ValueMask[0];
+      key->stencil_func[0] = ctx->Stencil.Function[0];
+      key->stencil_fail_op[0] = ctx->Stencil.FailFunc[0];
+      key->stencil_pass_depth_fail_op[0] = ctx->Stencil.ZFailFunc[0];
+      key->stencil_pass_depth_pass_op[0] = ctx->Stencil.ZPassFunc[0];
+      key->stencil_ref[0] = ctx->Stencil.Ref[0];
+      key->stencil_write_mask[0] = ctx->Stencil.WriteMask[0];
+      key->stencil_test_mask[0] = ctx->Stencil.ValueMask[0];
    }
    if (key->stencil_two_side) {
-      key->stencil_func[1] = stencil->Function[1];
-      key->stencil_fail_op[1] = stencil->FailFunc[1];
-      key->stencil_pass_depth_fail_op[1] = stencil->ZFailFunc[1];
-      key->stencil_pass_depth_pass_op[1] = stencil->ZPassFunc[1];
-      key->stencil_ref[1] = stencil->Ref[1];
-      key->stencil_write_mask[1] = stencil->WriteMask[1];
-      key->stencil_test_mask[1] = stencil->ValueMask[1];
+      key->stencil_func[1] = ctx->Stencil.Function[back];
+      key->stencil_fail_op[1] = ctx->Stencil.FailFunc[back];
+      key->stencil_pass_depth_fail_op[1] = ctx->Stencil.ZFailFunc[back];
+      key->stencil_pass_depth_pass_op[1] = ctx->Stencil.ZPassFunc[back];
+      key->stencil_ref[1] = ctx->Stencil.Ref[back];
+      key->stencil_write_mask[1] = ctx->Stencil.WriteMask[back];
+      key->stencil_test_mask[1] = ctx->Stencil.ValueMask[back];
    }
 
-   if (brw->attribs.Color->_LogicOpEnabled)
-      key->logic_op = brw->attribs.Color->LogicOp;
+   if (ctx->Color._LogicOpEnabled)
+      key->logic_op = ctx->Color.LogicOp;
    else
       key->logic_op = GL_COPY;
 
-   key->color_blend = brw->attribs.Color->BlendEnabled;
+   key->color_blend = ctx->Color.BlendEnabled;
    if (key->color_blend) {
-      key->blend_eq_rgb = brw->attribs.Color->BlendEquationRGB;
-      key->blend_eq_a = brw->attribs.Color->BlendEquationA;
-      key->blend_src_rgb = brw->attribs.Color->BlendSrcRGB;
-      key->blend_dst_rgb = brw->attribs.Color->BlendDstRGB;
-      key->blend_src_a = brw->attribs.Color->BlendSrcA;
-      key->blend_dst_a = brw->attribs.Color->BlendDstA;
+      key->blend_eq_rgb = ctx->Color.BlendEquationRGB;
+      key->blend_eq_a = ctx->Color.BlendEquationA;
+      key->blend_src_rgb = ctx->Color.BlendSrcRGB;
+      key->blend_dst_rgb = ctx->Color.BlendDstRGB;
+      key->blend_src_a = ctx->Color.BlendSrcA;
+      key->blend_dst_a = ctx->Color.BlendDstA;
    }
 
-   key->alpha_enabled = brw->attribs.Color->AlphaEnabled;
+   key->alpha_enabled = ctx->Color.AlphaEnabled;
    if (key->alpha_enabled) {
-      key->alpha_func = brw->attribs.Color->AlphaFunc;
-      key->alpha_ref = brw->attribs.Color->AlphaRef;
+      key->alpha_func = ctx->Color.AlphaFunc;
+      key->alpha_ref = ctx->Color.AlphaRef;
    }
 
-   key->dither = brw->attribs.Color->DitherFlag;
+   key->dither = ctx->Color.DitherFlag;
 
-   key->depth_test = brw->attribs.Depth->Test;
+   key->depth_test = ctx->Depth.Test;
    if (key->depth_test) {
-      key->depth_func = brw->attribs.Depth->Func;
-      key->depth_write = brw->attribs.Depth->Mask;
+      key->depth_func = ctx->Depth.Func;
+      key->depth_write = ctx->Depth.Mask;
    }
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_clip.c b/src/mesa/drivers/dri/i965/brw_clip.c
index 38d8b704d7e..5cffcebde43 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.c
+++ b/src/mesa/drivers/dri/i965/brw_clip.c
@@ -145,14 +145,14 @@ static void upload_clip_prog(struct brw_context *brw)
    /* CACHE_NEW_VS_PROG */
    key.attrs = brw->vs.prog_data->outputs_written;
    /* _NEW_LIGHT */
-   key.do_flat_shading = (brw->attribs.Light->ShadeModel == GL_FLAT);
+   key.do_flat_shading = (ctx->Light.ShadeModel == GL_FLAT);
    /* _NEW_TRANSFORM */
-   key.nr_userclip = brw_count_bits(brw->attribs.Transform->ClipPlanesEnabled);
+   key.nr_userclip = brw_count_bits(ctx->Transform.ClipPlanesEnabled);
    key.clip_mode = BRW_CLIPMODE_NORMAL;
 
    /* _NEW_POLYGON */
    if (key.primitive == GL_TRIANGLES) {
-      if (brw->attribs.Polygon->CullFaceMode == GL_FRONT_AND_BACK) 
+      if (ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) 
 	 key.clip_mode = BRW_CLIPMODE_REJECT_ALL;
       else {
 	 GLuint fill_front = CLIP_CULL;
@@ -160,44 +160,44 @@ static void upload_clip_prog(struct brw_context *brw)
 	 GLuint offset_front = 0;
 	 GLuint offset_back = 0;
 
-	 if (!brw->attribs.Polygon->CullFlag ||
-	     brw->attribs.Polygon->CullFaceMode != GL_FRONT) {
-	    switch (brw->attribs.Polygon->FrontMode) {
+	 if (!ctx->Polygon.CullFlag ||
+	     ctx->Polygon.CullFaceMode != GL_FRONT) {
+	    switch (ctx->Polygon.FrontMode) {
 	    case GL_FILL: 
 	       fill_front = CLIP_FILL; 
 	       offset_front = 0;
 	       break;
 	    case GL_LINE:
 	       fill_front = CLIP_LINE;
-	       offset_front = brw->attribs.Polygon->OffsetLine;
+	       offset_front = ctx->Polygon.OffsetLine;
 	       break;
 	    case GL_POINT:
 	       fill_front = CLIP_POINT;
-	       offset_front = brw->attribs.Polygon->OffsetPoint;
+	       offset_front = ctx->Polygon.OffsetPoint;
 	       break;
 	    }
 	 }
 
-	 if (!brw->attribs.Polygon->CullFlag ||
-	     brw->attribs.Polygon->CullFaceMode != GL_BACK) {
-	    switch (brw->attribs.Polygon->BackMode) {
+	 if (!ctx->Polygon.CullFlag ||
+	     ctx->Polygon.CullFaceMode != GL_BACK) {
+	    switch (ctx->Polygon.BackMode) {
 	    case GL_FILL: 
 	       fill_back = CLIP_FILL; 
 	       offset_back = 0;
 	       break;
 	    case GL_LINE:
 	       fill_back = CLIP_LINE;
-	       offset_back = brw->attribs.Polygon->OffsetLine;
+	       offset_back = ctx->Polygon.OffsetLine;
 	       break;
 	    case GL_POINT:
 	       fill_back = CLIP_POINT;
-	       offset_back = brw->attribs.Polygon->OffsetPoint;
+	       offset_back = ctx->Polygon.OffsetPoint;
 	       break;
 	    }
 	 }
 
-	 if (brw->attribs.Polygon->BackMode != GL_FILL ||
-	     brw->attribs.Polygon->FrontMode != GL_FILL) {
+	 if (ctx->Polygon.BackMode != GL_FILL ||
+	     ctx->Polygon.FrontMode != GL_FILL) {
 	    key.do_unfilled = 1;
 
 	    /* Most cases the fixed function units will handle.  Cases where
@@ -207,17 +207,17 @@ static void upload_clip_prog(struct brw_context *brw)
 
 	    if (offset_back || offset_front) {
 	       /* _NEW_POLYGON, _NEW_BUFFERS */
-	       key.offset_units = brw->attribs.Polygon->OffsetUnits * brw->intel.polygon_offset_scale;
-	       key.offset_factor = brw->attribs.Polygon->OffsetFactor * ctx->DrawBuffer->_MRD;
+	       key.offset_units = ctx->Polygon.OffsetUnits * brw->intel.polygon_offset_scale;
+	       key.offset_factor = ctx->Polygon.OffsetFactor * ctx->DrawBuffer->_MRD;
 	    }
 
-	    switch (brw->attribs.Polygon->FrontFace) {
+	    switch (ctx->Polygon.FrontFace) {
 	    case GL_CCW:
 	       key.fill_ccw = fill_front;
 	       key.fill_cw = fill_back;
 	       key.offset_ccw = offset_front;
 	       key.offset_cw = offset_back;
-	       if (brw->attribs.Light->Model.TwoSide &&
+	       if (ctx->Light.Model.TwoSide &&
 		   key.fill_cw != CLIP_CULL) 
 		  key.copy_bfc_cw = 1;
 	       break;
@@ -226,7 +226,7 @@ static void upload_clip_prog(struct brw_context *brw)
 	       key.fill_ccw = fill_back;
 	       key.offset_cw = offset_front;
 	       key.offset_ccw = offset_back;
-	       if (brw->attribs.Light->Model.TwoSide &&
+	       if (ctx->Light.Model.TwoSide &&
 		   key.fill_ccw != CLIP_CULL) 
 		  key.copy_bfc_ccw = 1;
 	       break;
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index e2bc08a6cb7..eaac6224f6e 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -32,6 +32,7 @@
 
 #include "main/imports.h"
 #include "main/api_noop.h"
+#include "main/macros.h"
 #include "main/vtxfmt.h"
 #include "main/simple_list.h"
 #include "shader/shader_api.h"
@@ -71,32 +72,10 @@ static void brwInitDriverFunctions( struct dd_function_table *functions )
    brwInitFragProgFuncs( functions );
    brwInitProgFuncs( functions );
    brw_init_queryobj_functions(functions);
-}
-
 
-static void brw_init_attribs( struct brw_context *brw )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-
-   brw->attribs.Color = &ctx->Color;
-   brw->attribs.Depth = &ctx->Depth;
-   brw->attribs.Fog = &ctx->Fog;
-   brw->attribs.Hint = &ctx->Hint;
-   brw->attribs.Light = &ctx->Light;
-   brw->attribs.Line = &ctx->Line;
-   brw->attribs.Point = &ctx->Point;
-   brw->attribs.Polygon = &ctx->Polygon;
-   brw->attribs.Scissor = &ctx->Scissor;
-   brw->attribs.Stencil = &ctx->Stencil;
-   brw->attribs.Texture = &ctx->Texture;
-   brw->attribs.Transform = &ctx->Transform;
-   brw->attribs.Viewport = &ctx->Viewport;
-   brw->attribs.VertexProgram = &ctx->VertexProgram;
-   brw->attribs.FragmentProgram = &ctx->FragmentProgram;
-   brw->attribs.PolygonStipple = &ctx->PolygonStipple[0];
+   functions->Viewport = intel_viewport;
 }
 
-
 GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 			    __DRIcontextPrivate *driContextPriv,
 			    void *sharedContextPrivate)
@@ -126,23 +105,24 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
 
-   ctx->Const.MaxTextureUnits = BRW_MAX_TEX_UNIT;
    ctx->Const.MaxTextureImageUnits = BRW_MAX_TEX_UNIT;
-   ctx->Const.MaxTextureCoordUnits = BRW_MAX_TEX_UNIT;
+   ctx->Const.MaxTextureCoordUnits = 8; /* Mesa limit */
+   ctx->Const.MaxTextureUnits = MIN2(ctx->Const.MaxTextureCoordUnits,
+                                     ctx->Const.MaxTextureImageUnits);
    ctx->Const.MaxVertexTextureImageUnits = 0; /* no vertex shader textures */
 
-   /* Advertise the full hardware capabilities.  The new memory
-    * manager should cope much better with overload situations:
+   /* Mesa limits textures to 4kx4k; it would be nice to fix that someday
     */
-   ctx->Const.MaxTextureLevels = 12;
+   ctx->Const.MaxTextureLevels = 13;
    ctx->Const.Max3DTextureLevels = 9;
    ctx->Const.MaxCubeTextureLevels = 12;
-   ctx->Const.MaxTextureRectSize = (1<<11);
+   ctx->Const.MaxTextureRectSize = (1<<12);
    
+   /* if conformance mode is set, swrast can handle any size AA point */
+   ctx->Const.MaxPointSizeAA = 255.0;
+
 /*    ctx->Const.MaxNativeVertexProgramTemps = 32; */
 
-   brw_init_attribs( brw );
-   brw_init_metaops( brw );
    brw_init_state( brw );
 
    brw->state.dirty.mesa = ~0;
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index e3904be977f..df90c2027f3 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -131,18 +131,17 @@ struct brw_context;
 #define BRW_NEW_WM_INPUT_DIMENSIONS     0x100
 #define BRW_NEW_INPUT_VARYING           0x200
 #define BRW_NEW_PSP                     0x800
-#define BRW_NEW_METAOPS                 0x1000
 #define BRW_NEW_FENCE                   0x2000
-#define BRW_NEW_LOCK                    0x4000
-#define BRW_NEW_INDICES			0x8000
-#define BRW_NEW_VERTICES		0x10000
+#define BRW_NEW_INDICES			0x4000
+#define BRW_NEW_VERTICES		0x8000
 /**
  * Used for any batch entry with a relocated pointer that will be used
  * by any 3D rendering.
  */
-#define BRW_NEW_BATCH			0x8000
+#define BRW_NEW_BATCH			0x10000
 /** brw->depth_region updated */
-#define BRW_NEW_DEPTH_BUFFER		0x10000
+#define BRW_NEW_DEPTH_BUFFER		0x20000
+#define BRW_NEW_NR_SURFACES		0x40000
 
 struct brw_state_flags {
    /** State update flags signalled by mesa internals */
@@ -158,7 +157,6 @@ struct brw_state_flags {
 struct brw_vertex_program {
    struct gl_vertex_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
@@ -166,7 +164,6 @@ struct brw_vertex_program {
 struct brw_fragment_program {
    struct gl_fragment_program program;
    GLuint id;
-   GLuint param_state;		/* flags indicating state tracked by params */
 };
 
 
@@ -240,7 +237,7 @@ struct brw_vs_ouput_sizes {
 };
 
 
-#define BRW_MAX_TEX_UNIT 8
+#define BRW_MAX_TEX_UNIT 16
 #define BRW_WM_MAX_SURF BRW_MAX_TEX_UNIT + MAX_DRAW_BUFFERS
 
 enum brw_cache_id {
@@ -305,26 +302,6 @@ struct brw_cache {
 };
 
 
-
-struct brw_state_pointers {
-   struct gl_colorbuffer_attrib	*Color;
-   struct gl_depthbuffer_attrib	*Depth;
-   struct gl_fog_attrib		*Fog;
-   struct gl_hint_attrib	*Hint;
-   struct gl_light_attrib	*Light;
-   struct gl_line_attrib	*Line;
-   struct gl_point_attrib	*Point;
-   struct gl_polygon_attrib	*Polygon;
-   GLuint                       *PolygonStipple;
-   struct gl_scissor_attrib	*Scissor;
-   struct gl_stencil_attrib	*Stencil;
-   struct gl_texture_attrib	*Texture;
-   struct gl_transform_attrib	*Transform;
-   struct gl_viewport_attrib	*Viewport;
-   struct gl_vertex_program_state *VertexProgram; 
-   struct gl_fragment_program_state *FragmentProgram;
-};
-
 /* Considered adding a member to this struct to document which flags
  * an update might raise so that ordering of the state atoms can be
  * checked or derived at runtime.  Dropped the idea in favor of having
@@ -459,7 +436,6 @@ struct brw_context
       int validated_bo_count;
    } state;
 
-   struct brw_state_pointers attribs;
    struct brw_cache cache;
    struct brw_cached_batch_item *cached_batch_items;
 
@@ -493,28 +469,6 @@ struct brw_context
       unsigned int offset;
    } ib;
 
-   struct {
-      /* Will be allocated on demand if needed.   
-       */
-      struct brw_state_pointers attribs;
-      struct gl_vertex_program *vp;
-      struct gl_fragment_program *fp, *fp_tex;
-
-      struct gl_buffer_object *vbo;
-
-      struct intel_region *saved_draw_region;
-      GLuint saved_nr_draw_regions;
-      struct intel_region *saved_depth_region;
-
-      GLuint restore_draw_buffers[MAX_DRAW_BUFFERS];
-      GLuint restore_num_draw_buffers;
-
-      struct gl_fragment_program *restore_fp;
-      
-      GLboolean active;
-   } metaops;
-
-
    /* Active vertex program: 
     */
    const struct gl_vertex_program *vertex_program;
@@ -705,13 +659,6 @@ void brw_FrameBufferTexInit( struct brw_context *brw,
 void brw_FrameBufferTexDestroy( struct brw_context *brw );
 void brw_validate_textures( struct brw_context *brw );
 
-/*======================================================================
- * brw_metaops.c
- */
-
-void brw_init_metaops( struct brw_context *brw );
-void brw_destroy_metaops( struct brw_context *brw );
-
 
 /*======================================================================
  * brw_program.c
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index c7bac7b0c52..4eaaa5f871b 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -48,6 +48,7 @@
  */
 static void calculate_curbe_offsets( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    /* CACHE_NEW_WM_PROG */
    GLuint nr_fp_regs = (brw->wm.prog_data->nr_params + 15) / 16;
    
@@ -58,8 +59,8 @@ static void calculate_curbe_offsets( struct brw_context *brw )
    GLuint total_regs;
 
    /* _NEW_TRANSFORM */
-   if (brw->attribs.Transform->ClipPlanesEnabled) {
-      GLuint nr_planes = 6 + brw_count_bits(brw->attribs.Transform->ClipPlanesEnabled);
+   if (ctx->Transform.ClipPlanesEnabled) {
+      GLuint nr_planes = 6 + brw_count_bits(ctx->Transform.ClipPlanesEnabled);
       nr_clip_regs = (nr_planes * 4 + 15) / 16;
    }
 
@@ -184,8 +185,8 @@ static void prepare_constant_buffer(struct brw_context *brw)
     * function will also be called whenever fp or vp changes.
     */
    brw->curbe.tracked_state.dirty.mesa = (_NEW_TRANSFORM|_NEW_PROJECTION);
-   brw->curbe.tracked_state.dirty.mesa |= vp->param_state;
-   brw->curbe.tracked_state.dirty.mesa |= fp->param_state;
+   brw->curbe.tracked_state.dirty.mesa |= vp->program.Base.Parameters->StateFlags;
+   brw->curbe.tracked_state.dirty.mesa |= fp->program.Base.Parameters->StateFlags;
 
    if (sz == 0) {
 
@@ -233,11 +234,11 @@ static void prepare_constant_buffer(struct brw_context *brw)
        */
       assert(MAX_CLIP_PLANES == 6);
       for (j = 0; j < MAX_CLIP_PLANES; j++) {
-	 if (brw->attribs.Transform->ClipPlanesEnabled & (1<<j)) {
-	    buf[offset + i * 4 + 0] = brw->attribs.Transform->_ClipUserPlane[j][0];
-	    buf[offset + i * 4 + 1] = brw->attribs.Transform->_ClipUserPlane[j][1];
-	    buf[offset + i * 4 + 2] = brw->attribs.Transform->_ClipUserPlane[j][2];
-	    buf[offset + i * 4 + 3] = brw->attribs.Transform->_ClipUserPlane[j][3];
+	 if (ctx->Transform.ClipPlanesEnabled & (1<<j)) {
+	    buf[offset + i * 4 + 0] = ctx->Transform._ClipUserPlane[j][0];
+	    buf[offset + i * 4 + 1] = ctx->Transform._ClipUserPlane[j][1];
+	    buf[offset + i * 4 + 2] = ctx->Transform._ClipUserPlane[j][2];
+	    buf[offset + i * 4 + 3] = ctx->Transform._ClipUserPlane[j][3];
 	    i++;
 	 }
       }
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index f893dd67423..99fd587e9fc 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -49,7 +49,7 @@
 
 #define FILE_DEBUG_FLAG DEBUG_BATCH
 
-static GLuint hw_prim[GL_POLYGON+1] = {
+static GLuint prim_to_hw_prim[GL_POLYGON+1] = {
    _3DPRIM_POINTLIST,
    _3DPRIM_LINELIST,
    _3DPRIM_LINELOOP,
@@ -84,15 +84,17 @@ static const GLenum reduced_prim[GL_POLYGON+1] = {
  */
 static GLuint brw_set_prim(struct brw_context *brw, GLenum prim)
 {
+   GLcontext *ctx = &brw->intel.ctx;
+
    if (INTEL_DEBUG & DEBUG_PRIMS)
       _mesa_printf("PRIM: %s\n", _mesa_lookup_enum_by_nr(prim));
    
    /* Slight optimization to avoid the GS program when not needed:
     */
    if (prim == GL_QUAD_STRIP &&
-       brw->attribs.Light->ShadeModel != GL_FLAT &&
-       brw->attribs.Polygon->FrontMode == GL_FILL &&
-       brw->attribs.Polygon->BackMode == GL_FILL)
+       ctx->Light.ShadeModel != GL_FLAT &&
+       ctx->Polygon.FrontMode == GL_FILL &&
+       ctx->Polygon.BackMode == GL_FILL)
       prim = GL_TRIANGLE_STRIP;
 
    if (prim != brw->primitive) {
@@ -103,12 +105,9 @@ static GLuint brw_set_prim(struct brw_context *brw, GLenum prim)
 	 brw->intel.reduced_primitive = reduced_prim[prim];
 	 brw->state.dirty.brw |= BRW_NEW_REDUCED_PRIMITIVE;
       }
-
-      brw_validate_state(brw);
-      brw_upload_state(brw);
    }
 
-   return hw_prim[prim];
+   return prim_to_hw_prim[prim];
 }
 
 
@@ -123,9 +122,9 @@ static GLuint trim(GLenum prim, GLuint length)
 }
 
 
-static void brw_emit_prim( struct brw_context *brw, 
-			   const struct _mesa_prim *prim )
-
+static void brw_emit_prim(struct brw_context *brw,
+			  const struct _mesa_prim *prim,
+			  uint32_t hw_prim)
 {
    struct brw_3d_primitive prim_packet;
 
@@ -136,7 +135,7 @@ static void brw_emit_prim( struct brw_context *brw,
    prim_packet.header.opcode = CMD_3D_PRIM;
    prim_packet.header.length = sizeof(prim_packet)/4 - 2;
    prim_packet.header.pad = 0;
-   prim_packet.header.topology = brw_set_prim(brw, prim->mode);
+   prim_packet.header.topology = hw_prim;
    prim_packet.header.indexed = prim->indexed;
 
    prim_packet.verts_per_instance = trim(prim->mode, prim->count);
@@ -169,14 +168,11 @@ static void brw_merge_inputs( struct brw_context *brw,
    for (i = 0; i < VERT_ATTRIB_MAX; i++) {
       brw->vb.inputs[i].glarray = arrays[i];
 
-      /* XXX: metaops passes null arrays */
-      if (arrays[i]) {
-	 if (arrays[i]->StrideB != 0)
-	    brw->vb.info.varying |= 1 << i;
+      if (arrays[i]->StrideB != 0)
+	 brw->vb.info.varying |= 1 << i;
 
 	 brw->vb.info.sizes[i/16] |= (brw->vb.inputs[i].glarray->Size - 1) <<
 	    ((i%16) * 2);
-      }
    }
 
    /* Raise statechanges if input sizes and varying have changed: 
@@ -195,12 +191,13 @@ static GLboolean check_fallbacks( struct brw_context *brw,
 				  const struct _mesa_prim *prim,
 				  GLuint nr_prims )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    GLuint i;
 
    if (!brw->intel.strict_conformance)
       return GL_FALSE;
 
-   if (brw->attribs.Polygon->SmoothFlag) {
+   if (ctx->Polygon.SmoothFlag) {
       for (i = 0; i < nr_prims; i++)
 	 if (reduced_prim[prim[i].mode] == GL_TRIANGLES) 
 	    return GL_TRUE;
@@ -209,7 +206,7 @@ static GLboolean check_fallbacks( struct brw_context *brw,
    /* BRW hardware will do AA lines, but they are non-conformant it
     * seems.  TBD whether we keep this fallback:
     */
-   if (brw->attribs.Line->SmoothFlag) {
+   if (ctx->Line.SmoothFlag) {
       for (i = 0; i < nr_prims; i++)
 	 if (reduced_prim[prim[i].mode] == GL_LINES) 
 	    return GL_TRUE;
@@ -218,7 +215,7 @@ static GLboolean check_fallbacks( struct brw_context *brw,
    /* Stipple -- these fallbacks could be resolved with a little
     * bit of work?
     */
-   if (brw->attribs.Line->StippleFlag) {
+   if (ctx->Line.StippleFlag) {
       for (i = 0; i < nr_prims; i++) {
 	 /* GS doesn't get enough information to know when to reset
 	  * the stipple counter?!?
@@ -227,14 +224,14 @@ static GLboolean check_fallbacks( struct brw_context *brw,
 	    return GL_TRUE;
 	    
 	 if (prim[i].mode == GL_POLYGON &&
-	     (brw->attribs.Polygon->FrontMode == GL_LINE ||
-	      brw->attribs.Polygon->BackMode == GL_LINE))
+	     (ctx->Polygon.FrontMode == GL_LINE ||
+	      ctx->Polygon.BackMode == GL_LINE))
 	    return GL_TRUE;
       }
    }
 
 
-   if (brw->attribs.Point->SmoothFlag) {
+   if (ctx->Point.SmoothFlag) {
       for (i = 0; i < nr_prims; i++)
 	 if (prim[i].mode == GL_POINTS) 
 	    return GL_TRUE;
@@ -258,11 +255,15 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    struct brw_context *brw = brw_context(ctx);
    GLboolean retval = GL_FALSE;
    GLboolean warn = GL_FALSE;
+   GLboolean first_time = GL_TRUE;
    GLuint i;
 
    if (ctx->NewState)
       _mesa_update_state( ctx );
 
+   if (check_fallbacks(brw, prim, nr_prims))
+      return GL_FALSE;
+
    brw_validate_textures( brw );
 
    /* Bind all inputs, derive varying and size information:
@@ -275,6 +276,7 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    brw->vb.min_index = min_index;
    brw->vb.max_index = max_index;
    brw->state.dirty.brw |= BRW_NEW_VERTICES;
+
    /* Have to validate state quite late.  Will rebuild tnl_program,
     * which depends on varying information.  
     * 
@@ -289,59 +291,58 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
       return GL_TRUE;
    }
 
-   /* Flush the batch if it's approaching full, so that we don't wrap while
-    * we've got validated state that needs to be in the same batch as the
-    * primitives.  This fraction is just a guess (minimal full state plus
-    * a primitive is around 512 bytes), and would be better if we had
-    * an upper bound of how much we might emit in a single
-    * brw_try_draw_prims().
-    */
-   intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
-				   LOOP_CLIPRECTS);
-   {
-      /* Set the first primitive early, ahead of validate_state:
+   for (i = 0; i < nr_prims; i++) {
+      uint32_t hw_prim;
+
+      /* Flush the batch if it's approaching full, so that we don't wrap while
+       * we've got validated state that needs to be in the same batch as the
+       * primitives.  This fraction is just a guess (minimal full state plus
+       * a primitive is around 512 bytes), and would be better if we had
+       * an upper bound of how much we might emit in a single
+       * brw_try_draw_prims().
        */
-      brw_set_prim(brw, prim[0].mode);
+      intel_batchbuffer_require_space(intel->batch, intel->batch->size / 4,
+				      LOOP_CLIPRECTS);
 
-      brw_validate_state( brw );
+      hw_prim = brw_set_prim(brw, prim[i].mode);
 
-      /* Various fallback checks:
-       */
-      if (brw->intel.Fallback) 
-	 goto out;
+      if (first_time || (brw->state.dirty.brw & BRW_NEW_PRIMITIVE)) {
+	 first_time = GL_FALSE;
 
-      if (check_fallbacks( brw, prim, nr_prims ))
-	 goto out;
+	 brw_validate_state(brw);
 
-      /* Check that we can fit our state in with our existing batchbuffer, or
-       * flush otherwise.
-       */
-      if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
-					  brw->state.validated_bo_count)) {
-	 static GLboolean warned;
-	 intel_batchbuffer_flush(intel->batch);
-
-	 /* Validate the state after we flushed the batch (which would have
-	  * changed the set of dirty state).  If we still fail to
-	  * check_aperture, warn of what's happening, but attempt to continue
-	  * on since it may succeed anyway, and the user would probably rather
-	  * see a failure and a warning than a fallback.
+	 /* Various fallback checks:  */
+	 if (brw->intel.Fallback)
+	    goto out;
+
+	 /* Check that we can fit our state in with our existing batchbuffer, or
+	  * flush otherwise.
 	  */
-	 brw_validate_state(brw);
-	 if (!warned &&
-	     dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+	 if (dri_bufmgr_check_aperture_space(brw->state.validated_bos,
 					     brw->state.validated_bo_count)) {
-	    warn = GL_TRUE;
-	    warned = GL_TRUE;
+	    static GLboolean warned;
+	    intel_batchbuffer_flush(intel->batch);
+
+	    /* Validate the state after we flushed the batch (which would have
+	     * changed the set of dirty state).  If we still fail to
+	     * check_aperture, warn of what's happening, but attempt to continue
+	     * on since it may succeed anyway, and the user would probably rather
+	     * see a failure and a warning than a fallback.
+	     */
+	    brw_validate_state(brw);
+	    if (!warned &&
+		dri_bufmgr_check_aperture_space(brw->state.validated_bos,
+						brw->state.validated_bo_count)) {
+	       warn = GL_TRUE;
+	       warned = GL_TRUE;
+	    }
 	 }
-      }
-
-      brw_upload_state(brw);
 
-      for (i = 0; i < nr_prims; i++) {
-	 brw_emit_prim(brw, &prim[i]);
+	 brw_upload_state(brw);
       }
 
+      brw_emit_prim(brw, &prim[i], hw_prim);
+
       retval = GL_TRUE;
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 73d6dea01ee..02998d59571 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -156,7 +156,13 @@ static GLuint byte_types_scale[5] = {
 };
 
 
-static GLuint get_surface_type( GLenum type, GLuint size, GLboolean normalized )
+/**
+ * Given vertex array type/size/format/normalized info, return
+ * the appopriate hardware surface type.
+ * Format will be GL_RGBA or possibly GL_BGRA for GLubyte[4] color arrays.
+ */
+static GLuint get_surface_type( GLenum type, GLuint size,
+                                GLenum format, GLboolean normalized )
 {
    if (INTEL_DEBUG & DEBUG_VERTS)
       _mesa_printf("type %s size %d normalized %d\n", 
@@ -171,11 +177,20 @@ static GLuint get_surface_type( GLenum type, GLuint size, GLboolean normalized )
       case GL_BYTE: return byte_types_norm[size];
       case GL_UNSIGNED_INT: return uint_types_norm[size];
       case GL_UNSIGNED_SHORT: return ushort_types_norm[size];
-      case GL_UNSIGNED_BYTE: return ubyte_types_norm[size];
+      case GL_UNSIGNED_BYTE:
+         if (format == GL_BGRA) {
+            /* See GL_EXT_vertex_array_bgra */
+            assert(size == 4);
+            return BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+         }
+         else {
+            return ubyte_types_norm[size];
+         }
       default: assert(0); return 0;
       }      
    }
    else {
+      assert(format == GL_RGBA); /* sanity check */
       switch (type) {
       case GL_DOUBLE: return double_types[size];
       case GL_FLOAT: return float_types[size];
@@ -281,7 +296,7 @@ copy_array_to_vbo_array( struct brw_context *brw,
    } else {
       void *data;
       char *dest;
-      const char *src = element->glarray->Ptr;
+      const unsigned char *src = element->glarray->Ptr;
       int i;
 
       data = _mesa_malloc(dst_stride * element->count);
@@ -484,6 +499,7 @@ static void brw_emit_vertices(struct brw_context *brw)
       struct brw_vertex_element *input = enabled[i];
       uint32_t format = get_surface_type(input->glarray->Type,
 					 input->glarray->Size,
+					 input->glarray->Format,
 					 input->glarray->Normalized);
       uint32_t comp0 = BRW_VE1_COMPONENT_STORE_SRC;
       uint32_t comp1 = BRW_VE1_COMPONENT_STORE_SRC;
@@ -549,7 +565,7 @@ static void brw_prepare_indices(struct brw_context *brw)
        */
       dri_bo_subdata(bo, offset, ib_size, index_buffer->ptr);
    } else {
-      offset = (GLuint)index_buffer->ptr;
+      offset = (GLuint) (unsigned long) index_buffer->ptr;
 
       /* If the index buffer isn't aligned to its element size, we have to
        * rebase it into a temporary.
diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h
index 49b422ee2ff..9e2b39af9bb 100644
--- a/src/mesa/drivers/dri/i965/brw_eu.h
+++ b/src/mesa/drivers/dri/i965/brw_eu.h
@@ -129,17 +129,28 @@ static INLINE int type_sz( GLuint type )
    }
 }
 
+/**
+ * Construct a brw_reg.
+ * \param file  one of the BRW_x_REGISTER_FILE values
+ * \param nr  register number/index
+ * \param subnr  register sub number
+ * \param type  one of BRW_REGISTER_TYPE_x
+ * \param vstride  one of BRW_VERTICAL_STRIDE_x
+ * \param width  one of BRW_WIDTH_x
+ * \param hstride  one of BRW_HORIZONTAL_STRIDE_x
+ * \param swizzle  one of BRW_SWIZZLE_x
+ * \param writemask  WRITEMASK_X/Y/Z/W bitfield
+ */
 static INLINE struct brw_reg brw_reg( GLuint file,
-					GLuint nr,
-					GLuint subnr,
-					GLuint type,
-					GLuint vstride,
-					GLuint width,
-					GLuint hstride,
-					GLuint swizzle,
-					GLuint writemask)
-{
-      
+                                      GLuint nr,
+                                      GLuint subnr,
+                                      GLuint type,
+                                      GLuint vstride,
+                                      GLuint width,
+                                      GLuint hstride,
+                                      GLuint swizzle,
+                                      GLuint writemask )
+{
    struct brw_reg reg;
    reg.type = type;
    reg.file = file;
@@ -166,6 +177,7 @@ static INLINE struct brw_reg brw_reg( GLuint file,
    return reg;
 }
 
+/** Construct float[16] register */
 static INLINE struct brw_reg brw_vec16_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -181,6 +193,7 @@ static INLINE struct brw_reg brw_vec16_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
+/** Construct float[8] register */
 static INLINE struct brw_reg brw_vec8_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -196,7 +209,7 @@ static INLINE struct brw_reg brw_vec8_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
-
+/** Construct float[4] register */
 static INLINE struct brw_reg brw_vec4_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -212,7 +225,7 @@ static INLINE struct brw_reg brw_vec4_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
-
+/** Construct float[2] register */
 static INLINE struct brw_reg brw_vec2_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -228,6 +241,7 @@ static INLINE struct brw_reg brw_vec2_reg( GLuint file,
 		  WRITEMASK_XY);
 }
 
+/** Construct float[1] register */
 static INLINE struct brw_reg brw_vec1_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -277,6 +291,7 @@ static INLINE struct brw_reg byte_offset( struct brw_reg reg,
 }
    
 
+/** Construct unsigned word[16] register */
 static INLINE struct brw_reg brw_uw16_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -284,6 +299,7 @@ static INLINE struct brw_reg brw_uw16_reg( GLuint file,
    return suboffset(retype(brw_vec16_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
+/** Construct unsigned word[8] register */
 static INLINE struct brw_reg brw_uw8_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
@@ -291,6 +307,7 @@ static INLINE struct brw_reg brw_uw8_reg( GLuint file,
    return suboffset(retype(brw_vec8_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
+/** Construct unsigned word[1] register */
 static INLINE struct brw_reg brw_uw1_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
@@ -311,6 +328,7 @@ static INLINE struct brw_reg brw_imm_reg( GLuint type )
 		   0);      
 }
 
+/** Construct float immediate register */
 static INLINE struct brw_reg brw_imm_f( GLfloat f )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_F);
@@ -318,6 +336,7 @@ static INLINE struct brw_reg brw_imm_f( GLfloat f )
    return imm;
 }
 
+/** Construct integer immediate register */
 static INLINE struct brw_reg brw_imm_d( GLint d )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_D);
@@ -325,6 +344,7 @@ static INLINE struct brw_reg brw_imm_d( GLint d )
    return imm;
 }
 
+/** Construct uint immediate register */
 static INLINE struct brw_reg brw_imm_ud( GLuint ud )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UD);
@@ -332,6 +352,7 @@ static INLINE struct brw_reg brw_imm_ud( GLuint ud )
    return imm;
 }
 
+/** Construct ushort immediate register */
 static INLINE struct brw_reg brw_imm_uw( GLushort uw )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UW);
@@ -339,6 +360,7 @@ static INLINE struct brw_reg brw_imm_uw( GLushort uw )
    return imm;
 }
 
+/** Construct short immediate register */
 static INLINE struct brw_reg brw_imm_w( GLshort w )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_W);
@@ -350,8 +372,7 @@ static INLINE struct brw_reg brw_imm_w( GLshort w )
  * numbers alias with _V and _VF below:
  */
 
-/* Vector of eight signed half-byte values: 
- */
+/** Construct vector of eight signed half-byte values */
 static INLINE struct brw_reg brw_imm_v( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_V);
@@ -362,8 +383,7 @@ static INLINE struct brw_reg brw_imm_v( GLuint v )
    return imm;
 }
 
-/* Vector of four 8-bit float values:
- */
+/** Construct vector of four 8-bit float values */
 static INLINE struct brw_reg brw_imm_vf( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_VF);
@@ -400,44 +420,43 @@ static INLINE struct brw_reg brw_address( struct brw_reg reg )
    return brw_imm_uw(reg.nr * REG_SIZE + reg.subnr);
 }
 
-
-static INLINE struct brw_reg brw_vec1_grf( GLuint nr,
-					       GLuint subnr )
+/** Construct float[1] general-purpose register */
+static INLINE struct brw_reg brw_vec1_grf( GLuint nr, GLuint subnr )
 {
    return brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_vec8_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[2] general-purpose register */
+static INLINE struct brw_reg brw_vec2_grf( GLuint nr, GLuint subnr )
 {
-   return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
+   return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_vec4_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[4] general-purpose register */
+static INLINE struct brw_reg brw_vec4_grf( GLuint nr, GLuint subnr )
 {
    return brw_vec4_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-
-static INLINE struct brw_reg brw_vec2_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[8] general-purpose register */
+static INLINE struct brw_reg brw_vec8_grf( GLuint nr, GLuint subnr )
 {
-   return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
+   return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_uw8_grf( GLuint nr,
-					    GLuint subnr )
+
+static INLINE struct brw_reg brw_uw8_grf( GLuint nr, GLuint subnr )
 {
    return brw_uw8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_uw16_grf( GLuint nr,
-					    GLuint subnr )
+static INLINE struct brw_reg brw_uw16_grf( GLuint nr, GLuint subnr )
 {
    return brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
+
+/** Construct null register (usually used for setting condition codes) */
 static INLINE struct brw_reg brw_null_reg( void )
 {
    return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
@@ -524,13 +543,13 @@ static INLINE struct brw_reg stride( struct brw_reg reg,
 				       GLuint width,
 				       GLuint hstride )
 {
-   
    reg.vstride = cvt(vstride);
    reg.width = cvt(width) - 1;
    reg.hstride = cvt(hstride);
    return reg;
 }
 
+
 static INLINE struct brw_reg vec16( struct brw_reg reg )
 {
    return stride(reg, 16,16,1);
@@ -556,6 +575,7 @@ static INLINE struct brw_reg vec1( struct brw_reg reg )
    return stride(reg, 0,1,0);
 }
 
+
 static INLINE struct brw_reg get_element( struct brw_reg reg, GLuint elt )
 {
    return vec1(suboffset(reg, elt));
@@ -687,7 +707,7 @@ static INLINE struct brw_indirect brw_indirect( GLuint addr_subnr, GLint offset
 
 static INLINE struct brw_instruction *current_insn( struct brw_compile *p)
 {
-	return &p->store[p->nr_insn];
+   return &p->store[p->nr_insn];
 }
 
 void brw_pop_insn_state( struct brw_compile *p );
@@ -733,6 +753,7 @@ ALU2(ADD)
 ALU2(MUL)
 ALU1(FRC)
 ALU1(RNDD)
+ALU1(RNDZ)
 ALU2(MAC)
 ALU2(MACH)
 ALU1(LZD)
diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c
index ce4cf46cfa6..4e099b5945c 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c
@@ -439,6 +439,7 @@ ALU2(ADD)
 ALU2(MUL)
 ALU1(FRC)
 ALU1(RNDD)
+ALU1(RNDZ)
 ALU2(MAC)
 ALU2(MACH)
 ALU1(LZD)
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.c b/src/mesa/drivers/dri/i965/brw_fallback.c
index 4ea660a51a3..e63098fdd4a 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.c
+++ b/src/mesa/drivers/dri/i965/brw_fallback.c
@@ -47,20 +47,12 @@ static GLboolean do_check_fallback(struct brw_context *brw)
    GLcontext *ctx = &brw->intel.ctx;
    GLuint i;
 
-   /* BRW_NEW_METAOPS
-    */
-   if (brw->metaops.active)
-      return GL_FALSE;
-
    if (brw->intel.no_rast) {
       DBG("FALLBACK: rasterization disabled\n");
       return GL_TRUE;
    }
 
    /* _NEW_RENDERMODE
-    *
-    * XXX: need to save/restore RenderMode in metaops state, or
-    * somehow move to a new attribs pointer:
     */
    if (ctx->RenderMode != GL_RENDER) {
       DBG("FALLBACK: render mode\n");
@@ -70,7 +62,7 @@ static GLboolean do_check_fallback(struct brw_context *brw)
    /* _NEW_TEXTURE:
     */
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
       if (texUnit->_ReallyEnabled) {
 	 struct intel_texture_object *intelObj = intel_texture_object(texUnit->_Current);
 	 struct gl_texture_image *texImage = intelObj->base.Image[0][intelObj->firstLevel];
@@ -83,7 +75,7 @@ static GLboolean do_check_fallback(struct brw_context *brw)
    
    /* _NEW_STENCIL 
     */
-   if (brw->attribs.Stencil->Enabled && 
+   if (ctx->Stencil.Enabled && 
        !brw->intel.hw_stencil) {
       DBG("FALLBACK: stencil\n");
       return GL_TRUE;
@@ -101,7 +93,7 @@ static void check_fallback(struct brw_context *brw)
 const struct brw_tracked_state brw_check_fallback = {
    .dirty = {
       .mesa = _NEW_BUFFERS | _NEW_RENDERMODE | _NEW_TEXTURE | _NEW_STENCIL,
-      .brw  = BRW_NEW_METAOPS,
+      .brw  = 0,
       .cache = 0
    },
    .prepare = check_fallback
diff --git a/src/mesa/drivers/dri/i965/brw_metaops.c b/src/mesa/drivers/dri/i965/brw_metaops.c
deleted file mode 100644
index 41bfa2e2567..00000000000
--- a/src/mesa/drivers/dri/i965/brw_metaops.c
+++ /dev/null
@@ -1,583 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <[email protected]>
-  *   frame buffer texture by Gary Wong <[email protected]>
-  */
- 
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/macros.h"
-
-#include "shader/arbprogparse.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-#include "intel_regions.h"
-#include "intel_buffers.h"
-
-#include "brw_context.h"
-#include "brw_defines.h"
-#include "brw_draw.h"
-#include "brw_fallback.h"
-
-#define INIT(brw, STRUCT, ATTRIB) 		\
-do {						\
-   brw->attribs.ATTRIB = &ctx->ATTRIB;		\
-} while (0)
-
-#define DUP(brw, STRUCT, ATTRIB) 		\
-do {						\
-   brw->metaops.attribs.ATTRIB = MALLOC_STRUCT(STRUCT);	\
-   memcpy(brw->metaops.attribs.ATTRIB, 			\
-	  brw->attribs.ATTRIB,			\
-	  sizeof(struct STRUCT));		\
-} while (0)
-
-
-#define INSTALL(brw, ATTRIB, STATE)		\
-do {						\
-   brw->attribs.ATTRIB = brw->metaops.attribs.ATTRIB;	\
-   brw->state.dirty.mesa |= STATE;		\
-} while (0)
-
-#define RESTORE(brw, ATTRIB, STATE)			\
-do {							\
-   brw->attribs.ATTRIB = &brw->intel.ctx.ATTRIB;	\
-   brw->state.dirty.mesa |= STATE;			\
-} while (0)
-
-static void init_attribs( struct brw_context *brw )
-{
-   DUP(brw, gl_colorbuffer_attrib, Color);
-   DUP(brw, gl_depthbuffer_attrib, Depth);
-   DUP(brw, gl_fog_attrib, Fog);
-   DUP(brw, gl_hint_attrib, Hint);
-   DUP(brw, gl_light_attrib, Light);
-   DUP(brw, gl_line_attrib, Line);
-   DUP(brw, gl_point_attrib, Point);
-   DUP(brw, gl_polygon_attrib, Polygon);
-   DUP(brw, gl_scissor_attrib, Scissor);
-   DUP(brw, gl_stencil_attrib, Stencil);
-   DUP(brw, gl_texture_attrib, Texture);
-   DUP(brw, gl_transform_attrib, Transform);
-   DUP(brw, gl_viewport_attrib, Viewport);
-   DUP(brw, gl_vertex_program_state, VertexProgram);
-   DUP(brw, gl_fragment_program_state, FragmentProgram);
-}
-
-static void install_attribs( struct brw_context *brw )
-{
-   INSTALL(brw, Color, _NEW_COLOR);
-   INSTALL(brw, Depth, _NEW_DEPTH);
-   INSTALL(brw, Fog, _NEW_FOG);
-   INSTALL(brw, Hint, _NEW_HINT);
-   INSTALL(brw, Light, _NEW_LIGHT);
-   INSTALL(brw, Line, _NEW_LINE);
-   INSTALL(brw, Point, _NEW_POINT);
-   INSTALL(brw, Polygon, _NEW_POLYGON);
-   INSTALL(brw, Scissor, _NEW_SCISSOR);
-   INSTALL(brw, Stencil, _NEW_STENCIL);
-   INSTALL(brw, Texture, _NEW_TEXTURE);
-   INSTALL(brw, Transform, _NEW_TRANSFORM);
-   INSTALL(brw, Viewport, _NEW_VIEWPORT);
-   INSTALL(brw, VertexProgram, _NEW_PROGRAM);
-   INSTALL(brw, FragmentProgram, _NEW_PROGRAM);
-}
-
-static void restore_attribs( struct brw_context *brw )
-{
-   RESTORE(brw, Color, _NEW_COLOR);
-   RESTORE(brw, Depth, _NEW_DEPTH);
-   RESTORE(brw, Fog, _NEW_FOG);
-   RESTORE(brw, Hint, _NEW_HINT);
-   RESTORE(brw, Light, _NEW_LIGHT);
-   RESTORE(brw, Line, _NEW_LINE);
-   RESTORE(brw, Point, _NEW_POINT);
-   RESTORE(brw, Polygon, _NEW_POLYGON);
-   RESTORE(brw, Scissor, _NEW_SCISSOR);
-   RESTORE(brw, Stencil, _NEW_STENCIL);
-   RESTORE(brw, Texture, _NEW_TEXTURE);
-   RESTORE(brw, Transform, _NEW_TRANSFORM);
-   RESTORE(brw, Viewport, _NEW_VIEWPORT);
-   RESTORE(brw, VertexProgram, _NEW_PROGRAM);
-   RESTORE(brw, FragmentProgram, _NEW_PROGRAM);
-}
-
-
-static const char *vp_prog =
-      "!!ARBvp1.0\n"
-      "MOV  result.color, vertex.color;\n"
-      "MOV  result.position, vertex.position;\n"
-      "END\n";
-
-static const char *fp_prog =
-      "!!ARBfp1.0\n"
-      "MOV result.color, fragment.color;\n"
-      "END\n";
-
-static const char *fp_tex_prog =
-      "!!ARBfp1.0\n"
-      "TEMP a;\n"
-      "ADD a, fragment.position, program.local[0];\n"
-      "MUL a, a, program.local[1];\n"
-      "TEX result.color, a, texture[0], 2D;\n"
-      "MOV result.depth.z, fragment.position;\n"
-      "END\n";
-
-/* Derived values of importance:
- *
- *   FragmentProgram->_Current
- *   VertexProgram->_Enabled
- *   brw->vertex_program
- *   DrawBuffer->_ColorDrawBufferIndexes[0]
- * 
- *
- * More if drawpixels-through-texture is added.  
- */
-static void init_metaops_state( struct brw_context *brw )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-
-   brw->metaops.vbo = ctx->Driver.NewBufferObject(ctx, 1, GL_ARRAY_BUFFER_ARB);
-
-   ctx->Driver.BufferData(ctx,
-			  GL_ARRAY_BUFFER_ARB,
-			  4096,
-			  NULL,
-			  GL_DYNAMIC_DRAW_ARB,
-			  brw->metaops.vbo);
-
-   brw->metaops.fp = (struct gl_fragment_program *)
-      ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 1 );
-
-   brw->metaops.fp_tex = (struct gl_fragment_program *)
-      ctx->Driver.NewProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, 1 );
-
-   brw->metaops.vp = (struct gl_vertex_program *)
-      ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 1 );
-
-   _mesa_parse_arb_fragment_program(ctx, GL_FRAGMENT_PROGRAM_ARB, 
-				    fp_prog, strlen(fp_prog),
-				    brw->metaops.fp);
-
-   _mesa_parse_arb_fragment_program(ctx, GL_FRAGMENT_PROGRAM_ARB, 
-				    fp_tex_prog, strlen(fp_tex_prog),
-				    brw->metaops.fp_tex);
-
-   _mesa_parse_arb_vertex_program(ctx, GL_VERTEX_PROGRAM_ARB, 
-				  vp_prog, strlen(vp_prog),
-				  brw->metaops.vp);
-
-   brw->metaops.attribs.VertexProgram->_Current = brw->metaops.vp;
-   brw->metaops.attribs.VertexProgram->_Enabled = GL_TRUE;
-
-   brw->metaops.attribs.FragmentProgram->_Current = brw->metaops.fp;
-}
-
-static void meta_flat_shade( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw->metaops.attribs.Light->ShadeModel = GL_FLAT;
-   brw->state.dirty.mesa |= _NEW_LIGHT;
-}
-
-
-static void meta_no_stencil_write( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw->metaops.attribs.Stencil->Enabled = GL_FALSE;
-   brw->metaops.attribs.Stencil->WriteMask[0] = GL_FALSE; 
-   brw->state.dirty.mesa |= _NEW_STENCIL;
-}
-
-static void meta_no_depth_write( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw->metaops.attribs.Depth->Test = GL_FALSE;
-   brw->metaops.attribs.Depth->Mask = GL_FALSE;
-   brw->state.dirty.mesa |= _NEW_DEPTH;
-}
-
-
-static void meta_depth_replace( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_TRUE )
-    * ctx->Driver.DepthMask( ctx, GL_TRUE )
-    */
-   brw->metaops.attribs.Depth->Test = GL_TRUE;
-   brw->metaops.attribs.Depth->Mask = GL_TRUE;
-   brw->state.dirty.mesa |= _NEW_DEPTH;
-
-   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
-    */
-   brw->metaops.attribs.Depth->Func = GL_ALWAYS;
-
-   brw->state.dirty.mesa |= _NEW_DEPTH;
-}
-
-
-static void meta_stencil_replace( struct intel_context *intel,
-				 GLuint s_mask,
-				 GLuint s_clear)
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw->metaops.attribs.Stencil->Enabled = GL_TRUE;
-   brw->metaops.attribs.Stencil->WriteMask[0] = s_mask;
-   brw->metaops.attribs.Stencil->ValueMask[0] = 0xff;
-   brw->metaops.attribs.Stencil->Ref[0] = s_clear;
-   brw->metaops.attribs.Stencil->Function[0] = GL_ALWAYS;
-   brw->metaops.attribs.Stencil->FailFunc[0] = GL_REPLACE;
-   brw->metaops.attribs.Stencil->ZPassFunc[0] = GL_REPLACE;
-   brw->metaops.attribs.Stencil->ZFailFunc[0] = GL_REPLACE;
-   brw->state.dirty.mesa |= _NEW_STENCIL;
-}
-
-
-static void meta_color_mask( struct intel_context *intel, GLboolean state )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   if (state)
-      COPY_4V(brw->metaops.attribs.Color->ColorMask, 
-	      brw->intel.ctx.Color.ColorMask); 
-   else
-      ASSIGN_4V(brw->metaops.attribs.Color->ColorMask, 0, 0, 0, 0);
-
-   brw->state.dirty.mesa |= _NEW_COLOR;
-}
-
-static void meta_no_texture( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   
-   brw->metaops.attribs.FragmentProgram->_Current = brw->metaops.fp;
-   
-   brw->metaops.attribs.Texture->CurrentUnit = 0;
-   brw->metaops.attribs.Texture->_EnabledUnits = 0;
-   brw->metaops.attribs.Texture->_EnabledCoordUnits = 0;
-   brw->metaops.attribs.Texture->Unit[ 0 ].Enabled = 0;
-   brw->metaops.attribs.Texture->Unit[ 0 ]._ReallyEnabled = 0;
-
-   brw->state.dirty.mesa |= _NEW_TEXTURE | _NEW_PROGRAM;
-}
-
-static void meta_texture_blend_replace(struct intel_context *intel)
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   brw->metaops.attribs.Texture->CurrentUnit = 0;
-   brw->metaops.attribs.Texture->_EnabledUnits = 1;
-   brw->metaops.attribs.Texture->_EnabledCoordUnits = 1;
-   brw->metaops.attribs.Texture->Unit[ 0 ].Enabled = TEXTURE_2D_BIT;
-   brw->metaops.attribs.Texture->Unit[ 0 ]._ReallyEnabled = TEXTURE_2D_BIT;
-   brw->metaops.attribs.Texture->Unit[ 0 ].Current2D =
-      intel->frame_buffer_texobj;
-   brw->metaops.attribs.Texture->Unit[ 0 ]._Current =
-      intel->frame_buffer_texobj;
-
-   brw->state.dirty.mesa |= _NEW_TEXTURE | _NEW_PROGRAM;
-}
-
-static void meta_import_pixel_state(struct intel_context *intel)
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   
-   RESTORE(brw, Color, _NEW_COLOR);
-   RESTORE(brw, Depth, _NEW_DEPTH);
-   RESTORE(brw, Fog, _NEW_FOG);
-   RESTORE(brw, Scissor, _NEW_SCISSOR);
-   RESTORE(brw, Stencil, _NEW_STENCIL);
-   RESTORE(brw, Texture, _NEW_TEXTURE);
-   RESTORE(brw, FragmentProgram, _NEW_PROGRAM);
-}
-
-static void meta_frame_buffer_texture( struct intel_context *intel,
-				       GLint xoff, GLint yoff )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   struct intel_region *region = intel_drawbuf_region( intel );
-   
-   INSTALL(brw, FragmentProgram, _NEW_PROGRAM);
-
-   brw->metaops.attribs.FragmentProgram->_Current = brw->metaops.fp_tex;
-   /* This is unfortunate, but seems to be necessary, since later on we
-      will end up calling _mesa_load_state_parameters to lookup the
-      local params (below), and that will want to look in ctx.FragmentProgram
-      instead of brw->attribs.FragmentProgram. */
-   intel->ctx.FragmentProgram.Current = brw->metaops.fp_tex;
-
-   brw->metaops.fp_tex->Base.LocalParams[ 0 ][ 0 ] = xoff;
-   brw->metaops.fp_tex->Base.LocalParams[ 0 ][ 1 ] = yoff;
-   brw->metaops.fp_tex->Base.LocalParams[ 0 ][ 2 ] = 0.0;
-   brw->metaops.fp_tex->Base.LocalParams[ 0 ][ 3 ] = 0.0;
-   brw->metaops.fp_tex->Base.LocalParams[ 1 ][ 0 ] =
-      1.0 / region->pitch;
-   brw->metaops.fp_tex->Base.LocalParams[ 1 ][ 1 ] =
-      -1.0 / region->height;
-   brw->metaops.fp_tex->Base.LocalParams[ 1 ][ 2 ] = 0.0;
-   brw->metaops.fp_tex->Base.LocalParams[ 1 ][ 3 ] = 1.0;
-   
-   brw->state.dirty.mesa |= _NEW_PROGRAM;
-}
-
-
-static void meta_draw_region( struct intel_context *intel,
-			     struct intel_region *draw_region,
-			     struct intel_region *depth_region )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-
-   if (!brw->metaops.saved_draw_region) {
-      brw->metaops.saved_draw_region = brw->state.draw_regions[0];
-      brw->metaops.saved_nr_draw_regions = brw->state.nr_draw_regions;
-      brw->metaops.saved_depth_region = brw->state.depth_region;
-   }
-
-   brw->state.draw_regions[0] = draw_region;
-   brw->state.nr_draw_regions = 1;
-   brw->state.depth_region = depth_region;
-
-   if (intel->frame_buffer_texobj != NULL)
-      brw_FrameBufferTexDestroy(brw);
-
-   if (draw_region)
-       brw_FrameBufferTexInit(brw, draw_region);
-
-   brw->state.dirty.mesa |= _NEW_BUFFERS;
-}
-
-
-static void meta_draw_quad(struct intel_context *intel, 
-			   GLfloat x0, GLfloat x1,
-			   GLfloat y0, GLfloat y1, 
-			   GLfloat z,
-			   GLuint color,
-			   GLfloat s0, GLfloat s1,
-			   GLfloat t0, GLfloat t1)
-{
-   GLcontext *ctx = &intel->ctx;
-   struct brw_context *brw = brw_context(&intel->ctx);
-   struct gl_client_array pos_array;
-   struct gl_client_array color_array;
-   struct gl_client_array *attribs[VERT_ATTRIB_MAX];
-   struct _mesa_prim prim[1];
-   GLfloat pos[4][3];
-
-   ctx->Driver.BufferData(ctx,
-			  GL_ARRAY_BUFFER_ARB,
-			  sizeof(pos) + sizeof(color),
-			  NULL,
-			  GL_DYNAMIC_DRAW_ARB,
-			  brw->metaops.vbo);
-
-   pos[0][0] = x0;
-   pos[0][1] = y0;
-   pos[0][2] = z;
-
-   pos[1][0] = x1;
-   pos[1][1] = y0;
-   pos[1][2] = z;
-
-   pos[2][0] = x1;
-   pos[2][1] = y1;
-   pos[2][2] = z;
-
-   pos[3][0] = x0;
-   pos[3][1] = y1;
-   pos[3][2] = z;
-
-   ctx->Driver.BufferSubData(ctx,
-			     GL_ARRAY_BUFFER_ARB,
-			     0,
-			     sizeof(pos),
-			     pos,
-			     brw->metaops.vbo);
-
-   /* Convert incoming ARGB to required RGBA */
-   /* Note this color is stored as GL_UNSIGNED_BYTE */
-   color = (color & 0xff00ff00) | (((color >> 16) | (color << 16)) & 0xff00ff);
-
-   ctx->Driver.BufferSubData(ctx,
-			     GL_ARRAY_BUFFER_ARB,
-			     sizeof(pos),
-			     sizeof(color),
-			     &color,
-			     brw->metaops.vbo);
-
-   /* Ignoring texture coords. 
-    */
-
-   memset(attribs, 0, VERT_ATTRIB_MAX * sizeof(*attribs));
-
-   attribs[VERT_ATTRIB_POS] = &pos_array;
-   attribs[VERT_ATTRIB_POS]->Ptr = 0;
-   attribs[VERT_ATTRIB_POS]->Type = GL_FLOAT;
-   attribs[VERT_ATTRIB_POS]->Enabled = 1;
-   attribs[VERT_ATTRIB_POS]->Size = 3;
-   attribs[VERT_ATTRIB_POS]->StrideB = 3 * sizeof(GLfloat);
-   attribs[VERT_ATTRIB_POS]->Stride = 3 * sizeof(GLfloat);
-   attribs[VERT_ATTRIB_POS]->_MaxElement = 4;
-   attribs[VERT_ATTRIB_POS]->Normalized = 0;
-   attribs[VERT_ATTRIB_POS]->BufferObj = brw->metaops.vbo;
-
-   attribs[VERT_ATTRIB_COLOR0] = &color_array;
-   attribs[VERT_ATTRIB_COLOR0]->Ptr = (const GLubyte *)sizeof(pos);
-   attribs[VERT_ATTRIB_COLOR0]->Type = GL_UNSIGNED_BYTE;
-   attribs[VERT_ATTRIB_COLOR0]->Enabled = 1;
-   attribs[VERT_ATTRIB_COLOR0]->Size = 4;
-   attribs[VERT_ATTRIB_COLOR0]->StrideB = 0;
-   attribs[VERT_ATTRIB_COLOR0]->Stride = 0;
-   attribs[VERT_ATTRIB_COLOR0]->_MaxElement = 1;
-   attribs[VERT_ATTRIB_COLOR0]->Normalized = 1;
-   attribs[VERT_ATTRIB_COLOR0]->BufferObj = brw->metaops.vbo;
-   
-   /* Just ignoring texture coordinates for now. 
-    */
-
-   memset(prim, 0, sizeof(*prim));
-
-   prim[0].mode = GL_TRIANGLE_FAN;
-   prim[0].begin = 1;
-   prim[0].end = 1;
-   prim[0].weak = 0;
-   prim[0].pad = 0;
-   prim[0].start = 0;
-   prim[0].count = 4;
-
-   brw_draw_prims(&brw->intel.ctx, 
-		  (const struct gl_client_array **)attribs,
-		  prim, 1,
-		  NULL,
-		  0,
-		  3 );
-}
-
-
-static void install_meta_state( struct intel_context *intel )
-{
-   GLcontext *ctx = &intel->ctx;
-   struct brw_context *brw = brw_context(ctx);
-   GLuint i;
-
-   if (!brw->metaops.vbo) {
-      init_metaops_state(brw);
-   }
-
-   install_attribs(brw);
-   
-   meta_no_texture(&brw->intel);
-   meta_flat_shade(&brw->intel);
-   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-      brw->metaops.restore_draw_buffers[i]
-         = ctx->DrawBuffer->_ColorDrawBufferIndexes[i];
-   }
-   brw->metaops.restore_num_draw_buffers = ctx->DrawBuffer->_NumColorDrawBuffers;
-
-   brw->metaops.restore_fp = ctx->FragmentProgram.Current;
-
-   /* This works without adjusting refcounts.  Fix later? 
-    */
-   brw->metaops.saved_draw_region = brw->state.draw_regions[0];
-   brw->metaops.saved_nr_draw_regions = brw->state.nr_draw_regions;
-   brw->metaops.saved_depth_region = brw->state.depth_region;
-   brw->metaops.active = 1;
-   
-   brw->state.dirty.brw |= BRW_NEW_METAOPS;
-}
-
-static void leave_meta_state( struct intel_context *intel )
-{
-   GLcontext *ctx = &intel->ctx;
-   struct brw_context *brw = brw_context(ctx);
-   GLuint i;
-
-   restore_attribs(brw);
-
-   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-      ctx->DrawBuffer->_ColorDrawBufferIndexes[i]
-         = brw->metaops.restore_draw_buffers[i];
-   }
-   ctx->DrawBuffer->_NumColorDrawBuffers = brw->metaops.restore_num_draw_buffers;
-
-   ctx->FragmentProgram.Current = brw->metaops.restore_fp;
-
-   brw->state.draw_regions[0] = brw->metaops.saved_draw_region;
-   brw->state.nr_draw_regions = brw->metaops.saved_nr_draw_regions;
-   brw->state.depth_region = brw->metaops.saved_depth_region;
-   brw->metaops.saved_draw_region = NULL;
-   brw->metaops.saved_depth_region = NULL;
-   brw->metaops.active = 0;
-
-   brw->state.dirty.mesa |= _NEW_BUFFERS;
-   brw->state.dirty.brw |= BRW_NEW_METAOPS;
-}
-
-
-
-void brw_init_metaops( struct brw_context *brw )
-{
-   init_attribs(brw);
-
-
-   brw->intel.vtbl.install_meta_state = install_meta_state;
-   brw->intel.vtbl.leave_meta_state = leave_meta_state;
-   brw->intel.vtbl.meta_no_depth_write = meta_no_depth_write;
-   brw->intel.vtbl.meta_no_stencil_write = meta_no_stencil_write;
-   brw->intel.vtbl.meta_stencil_replace = meta_stencil_replace;
-   brw->intel.vtbl.meta_depth_replace = meta_depth_replace;
-   brw->intel.vtbl.meta_color_mask = meta_color_mask;
-   brw->intel.vtbl.meta_no_texture = meta_no_texture;
-   brw->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
-   brw->intel.vtbl.meta_frame_buffer_texture = meta_frame_buffer_texture;
-   brw->intel.vtbl.meta_draw_region = meta_draw_region;
-   brw->intel.vtbl.meta_draw_quad = meta_draw_quad;
-   brw->intel.vtbl.meta_texture_blend_replace = meta_texture_blend_replace;
-/*    brw->intel.vtbl.meta_tex_rect_source = meta_tex_rect_source; */
-/*    brw->intel.vtbl.meta_draw_format = set_draw_format; */
-}
-
-void brw_destroy_metaops( struct brw_context *brw )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-
-   if (brw->metaops.vbo)
-      ctx->Driver.DeleteBuffer( ctx, brw->metaops.vbo );
-
-/*    ctx->Driver.DeleteProgram( ctx, brw->metaops.fp ); */
-/*    ctx->Driver.DeleteProgram( ctx, brw->metaops.fp_tex ); */
-/*    ctx->Driver.DeleteProgram( ctx, brw->metaops.vp ); */
-}
diff --git a/src/mesa/drivers/dri/i965/brw_misc_state.c b/src/mesa/drivers/dri/i965/brw_misc_state.c
index 627705fa9ba..9dcdad7b4ef 100644
--- a/src/mesa/drivers/dri/i965/brw_misc_state.c
+++ b/src/mesa/drivers/dri/i965/brw_misc_state.c
@@ -48,15 +48,16 @@
 
 static void upload_blend_constant_color(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_blend_constant_color bcc;
 
    memset(&bcc, 0, sizeof(bcc));      
    bcc.header.opcode = CMD_BLEND_CONSTANT_COLOR;
    bcc.header.length = sizeof(bcc)/4-2;
-   bcc.blend_constant_color[0] = brw->attribs.Color->BlendColor[0];
-   bcc.blend_constant_color[1] = brw->attribs.Color->BlendColor[1];
-   bcc.blend_constant_color[2] = brw->attribs.Color->BlendColor[2];
-   bcc.blend_constant_color[3] = brw->attribs.Color->BlendColor[3];
+   bcc.blend_constant_color[0] = ctx->Color.BlendColor[0];
+   bcc.blend_constant_color[1] = ctx->Color.BlendColor[1];
+   bcc.blend_constant_color[2] = ctx->Color.BlendColor[2];
+   bcc.blend_constant_color[3] = ctx->Color.BlendColor[3];
 
    BRW_CACHED_BATCH_STRUCT(brw, &bcc);
 }
@@ -154,10 +155,7 @@ static void upload_pipelined_state_pointers(struct brw_context *brw )
       OUT_RELOC(brw->gs.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 1);
    else
       OUT_BATCH(0);
-   if (!brw->metaops.active)
-      OUT_RELOC(brw->clip.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 1);
-   else
-      OUT_BATCH(0);
+   OUT_RELOC(brw->clip.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 1);
    OUT_RELOC(brw->sf.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
    OUT_RELOC(brw->wm.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
    OUT_RELOC(brw->cc.state_bo, I915_GEM_DOMAIN_INSTRUCTION, 0, 0);
@@ -186,7 +184,7 @@ static void upload_psp_urb_cbs(struct brw_context *brw )
 const struct brw_tracked_state brw_psp_urb_cbs = {
    .dirty = {
       .mesa = 0,
-      .brw = BRW_NEW_URB_FENCE | BRW_NEW_METAOPS | BRW_NEW_BATCH,
+      .brw = BRW_NEW_URB_FENCE | BRW_NEW_BATCH,
       .cache = (CACHE_NEW_VS_UNIT | 
 		CACHE_NEW_GS_UNIT | 
 		CACHE_NEW_GS_PROG | 
@@ -284,6 +282,7 @@ const struct brw_tracked_state brw_depthbuffer = {
 
 static void upload_polygon_stipple(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_polygon_stipple bps;
    GLuint i;
 
@@ -292,7 +291,7 @@ static void upload_polygon_stipple(struct brw_context *brw)
    bps.header.length = sizeof(bps)/4-2;
 
    for (i = 0; i < 32; i++)
-      bps.stipple[i] = brw->attribs.PolygonStipple[31 - i]; /* invert */
+      bps.stipple[i] = ctx->PolygonStipple[31 - i]; /* invert */
 
    BRW_CACHED_BATCH_STRUCT(brw, &bps);
 }
@@ -370,6 +369,7 @@ const struct brw_tracked_state brw_aa_line_parameters = {
 
 static void upload_line_stipple(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_line_stipple bls;
    GLfloat tmp;
    GLint tmpi;
@@ -378,10 +378,10 @@ static void upload_line_stipple(struct brw_context *brw)
    bls.header.opcode = CMD_LINE_STIPPLE_PATTERN;
    bls.header.length = sizeof(bls)/4 - 2;
 
-   bls.bits0.pattern = brw->attribs.Line->StipplePattern;
-   bls.bits1.repeat_count = brw->attribs.Line->StippleFactor;
+   bls.bits0.pattern = ctx->Line.StipplePattern;
+   bls.bits1.repeat_count = ctx->Line.StippleFactor;
 
-   tmp = 1.0 / (GLfloat) brw->attribs.Line->StippleFactor;
+   tmp = 1.0 / (GLfloat) ctx->Line.StippleFactor;
    tmpi = tmp * (1<<13);
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_program.c b/src/mesa/drivers/dri/i965/brw_program.c
index c38610b24ed..0c86911044b 100644
--- a/src/mesa/drivers/dri/i965/brw_program.c
+++ b/src/mesa/drivers/dri/i965/brw_program.c
@@ -111,13 +111,18 @@ static void brwProgramStringNotify( GLcontext *ctx,
 				    struct gl_program *prog )
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
+      struct gl_fragment_program *fprog = (struct gl_fragment_program *) prog;
       struct brw_context *brw = brw_context(ctx);
       struct brw_fragment_program *p = (struct brw_fragment_program *)prog;
       struct brw_fragment_program *fp = (struct brw_fragment_program *)brw->fragment_program;
+      if (fprog->FogOption) {
+         _mesa_append_fog_code(ctx, fprog);
+         fprog->FogOption = GL_NONE;
+      }
+
       if (p == fp)
 	 brw->state.dirty.brw |= BRW_NEW_FRAGMENT_PROGRAM;
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
    }
    else if (target == GL_VERTEX_PROGRAM_ARB) {
       struct brw_context *brw = brw_context(ctx);
@@ -129,7 +134,6 @@ static void brwProgramStringNotify( GLcontext *ctx,
 	 _mesa_insert_mvp_code(ctx, &p->program);
       }
       p->id = brw->program_id++;      
-      p->param_state = p->program.Base.Parameters->StateFlags;
 
       /* Also tell tnl about it:
        */
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index 9dce6cd8e6c..8c1711538af 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -46,6 +46,7 @@
 static void compile_sf_prog( struct brw_context *brw,
 			     struct brw_sf_prog_key *key )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_sf_compile c;
    const GLuint *program;
    GLuint program_size;
@@ -73,10 +74,12 @@ static void compile_sf_prog( struct brw_context *brw,
 	 c.attr_to_idx[i] = idx;
 	 c.idx_to_attr[idx] = i;
 	 if (i >= VERT_RESULT_TEX0 && i <= VERT_RESULT_TEX7) {
-		 c.point_attrs[i].CoordReplace = 
-			brw->attribs.Point->CoordReplace[i - VERT_RESULT_TEX0];
-	 } else
-		 c.point_attrs[i].CoordReplace = GL_FALSE;
+            c.point_attrs[i].CoordReplace = 
+               ctx->Point.CoordReplace[i - VERT_RESULT_TEX0];
+	 }
+         else {
+            c.point_attrs[i].CoordReplace = GL_FALSE;
+         }
 	 idx++;
       }
    
@@ -106,7 +109,6 @@ static void compile_sf_prog( struct brw_context *brw,
       assert(0);
       return;
    }
-	 
 
    /* get the program
     */
@@ -127,6 +129,7 @@ static void compile_sf_prog( struct brw_context *brw,
  */
 static void upload_sf_prog(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_sf_prog_key key;
 
    memset(&key, 0, sizeof(key));
@@ -157,15 +160,15 @@ static void upload_sf_prog(struct brw_context *brw)
       break;
    }
 
-   key.do_point_sprite = brw->attribs.Point->PointSprite;
-   key.SpriteOrigin = brw->attribs.Point->SpriteOrigin;
+   key.do_point_sprite = ctx->Point.PointSprite;
+   key.SpriteOrigin = ctx->Point.SpriteOrigin;
    /* _NEW_LIGHT */
-   key.do_flat_shading = (brw->attribs.Light->ShadeModel == GL_FLAT);
-   key.do_twoside_color = (brw->attribs.Light->Enabled && brw->attribs.Light->Model.TwoSide);
+   key.do_flat_shading = (ctx->Light.ShadeModel == GL_FLAT);
+   key.do_twoside_color = (ctx->Light.Enabled && ctx->Light.Model.TwoSide);
 
    /* _NEW_POLYGON */
    if (key.do_twoside_color)
-      key.frontface_ccw = (brw->attribs.Polygon->FrontFace == GL_CCW);
+      key.frontface_ccw = (ctx->Polygon.FrontFace == GL_CCW);
 
    dri_bo_unreference(brw->sf.prog_bo);
    brw->sf.prog_bo = brw_search_cache(&brw->cache, BRW_SF_PROG,
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index 506126fcfb0..e96d5354b30 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -42,45 +42,30 @@ static void upload_sf_vp(struct brw_context *brw)
    GLcontext *ctx = &brw->intel.ctx;
    const GLfloat depth_scale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
    struct brw_sf_viewport sfv;
-   struct intel_renderbuffer *irb =
-      intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0]);
    GLfloat y_scale, y_bias;
+   const GLboolean render_to_fbo = (ctx->DrawBuffer->Name != 0);
 
    memset(&sfv, 0, sizeof(sfv));
 
-   if (ctx->DrawBuffer->Name) {
-      /* User-created FBO */
-      if (irb && !irb->RenderToTexture) {
-	 y_scale = -1.0;
-	 y_bias = ctx->DrawBuffer->Height;
-      } else {
-	 y_scale = 1.0;
-	 y_bias = 0;
-      }
-   } else {
+   if (render_to_fbo) {
+      y_scale = 1.0;
+      y_bias = 0;
+   }
+   else {
       y_scale = -1.0;
       y_bias = ctx->DrawBuffer->Height;
    }
 
-   /* _NEW_VIEWPORT, BRW_NEW_METAOPS */
-
-   if (!brw->metaops.active) {
-      const GLfloat *v = ctx->Viewport._WindowMap.m;
-
-      sfv.viewport.m00 = v[MAT_SX];
-      sfv.viewport.m11 = v[MAT_SY] * y_scale;
-      sfv.viewport.m22 = v[MAT_SZ] * depth_scale;
-      sfv.viewport.m30 = v[MAT_TX];
-      sfv.viewport.m31 = v[MAT_TY] * y_scale + y_bias;
-      sfv.viewport.m32 = v[MAT_TZ] * depth_scale;
-   } else {
-      sfv.viewport.m00 =   1;
-      sfv.viewport.m11 = - 1;
-      sfv.viewport.m22 =   1;
-      sfv.viewport.m30 =   0;
-      sfv.viewport.m31 =   ctx->DrawBuffer->Height;
-      sfv.viewport.m32 =   0;
-   }
+   /* _NEW_VIEWPORT */
+
+   const GLfloat *v = ctx->Viewport._WindowMap.m;
+
+   sfv.viewport.m00 = v[MAT_SX];
+   sfv.viewport.m11 = v[MAT_SY] * y_scale;
+   sfv.viewport.m22 = v[MAT_SZ] * depth_scale;
+   sfv.viewport.m30 = v[MAT_TX];
+   sfv.viewport.m31 = v[MAT_TY] * y_scale + y_bias;
+   sfv.viewport.m32 = v[MAT_TZ] * depth_scale;
 
    /* _NEW_SCISSOR */
 
@@ -91,10 +76,20 @@ static void upload_sf_vp(struct brw_context *brw)
     * Note that the hardware's coordinates are inclusive, while Mesa's min is
     * inclusive but max is exclusive.
     */
-   sfv.scissor.xmin = ctx->DrawBuffer->_Xmin;
-   sfv.scissor.xmax = ctx->DrawBuffer->_Xmax - 1;
-   sfv.scissor.ymin = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymax;
-   sfv.scissor.ymax = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymin - 1;
+   if (render_to_fbo) {
+      /* texmemory: Y=0=bottom */
+      sfv.scissor.xmin = ctx->DrawBuffer->_Xmin;
+      sfv.scissor.xmax = ctx->DrawBuffer->_Xmax - 1;
+      sfv.scissor.ymin = ctx->DrawBuffer->_Ymin;
+      sfv.scissor.ymax = ctx->DrawBuffer->_Ymax - 1;
+   }
+   else {
+      /* memory: Y=0=top */
+      sfv.scissor.xmin = ctx->DrawBuffer->_Xmin;
+      sfv.scissor.xmax = ctx->DrawBuffer->_Xmax - 1;
+      sfv.scissor.ymin = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymax;
+      sfv.scissor.ymax = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymin - 1;
+   }
 
    dri_bo_unreference(brw->sf.vp_bo);
    brw->sf.vp_bo = brw_cache_data( &brw->cache, BRW_SF_VP, &sfv, NULL, 0 );
@@ -104,7 +99,7 @@ const struct brw_tracked_state brw_sf_vp = {
    .dirty = {
       .mesa  = (_NEW_VIEWPORT | 
 		_NEW_SCISSOR),
-      .brw   = BRW_NEW_METAOPS,
+      .brw   = 0,
       .cache = 0
    },
    .prepare = upload_sf_vp
@@ -117,7 +112,11 @@ struct brw_sf_unit_key {
    unsigned int nr_urb_entries, urb_size, sfsize;
 
    GLenum front_face, cull_face;
-   GLboolean scissor, line_smooth, point_sprite, point_attenuated;
+   unsigned scissor:1;
+   unsigned line_smooth:1;
+   unsigned point_sprite:1;
+   unsigned point_attenuated:1;
+   unsigned render_to_fbo:1;
    float line_width;
    float point_size;
 };
@@ -125,6 +124,7 @@ struct brw_sf_unit_key {
 static void
 sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_SF_PROG */
@@ -136,20 +136,22 @@ sf_unit_populate_key(struct brw_context *brw, struct brw_sf_unit_key *key)
    key->urb_size = brw->urb.vsize;
    key->sfsize = brw->urb.sfsize;
 
-   key->scissor = brw->attribs.Scissor->Enabled;
-   key->front_face = brw->attribs.Polygon->FrontFace;
+   key->scissor = ctx->Scissor.Enabled;
+   key->front_face = ctx->Polygon.FrontFace;
 
-   if (brw->attribs.Polygon->CullFlag)
-      key->cull_face = brw->attribs.Polygon->CullFaceMode;
+   if (ctx->Polygon.CullFlag)
+      key->cull_face = ctx->Polygon.CullFaceMode;
    else
       key->cull_face = GL_NONE;
 
-   key->line_width = brw->attribs.Line->Width;
-   key->line_smooth = brw->attribs.Line->SmoothFlag;
+   key->line_width = ctx->Line.Width;
+   key->line_smooth = ctx->Line.SmoothFlag;
+
+   key->point_sprite = ctx->Point.PointSprite;
+   key->point_size = ctx->Point.Size;
+   key->point_attenuated = ctx->Point._Attenuated;
 
-   key->point_sprite = brw->attribs.Point->PointSprite;
-   key->point_size = brw->attribs.Point->Size;
-   key->point_attenuated = brw->attribs.Point->_Attenuated;
+   key->render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
 }
 
 static dri_bo *
@@ -196,6 +198,11 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
    else
       sf.sf5.front_winding = BRW_FRONTWINDING_CW;
 
+   /* The viewport is inverted for rendering to a FBO, and that inverts
+    * polygon front/back orientation.
+    */
+   sf.sf5.front_winding ^= key->render_to_fbo;
+
    switch (key->cull_face) {
    case GL_FRONT:
       sf.sf6.cull_mode = BRW_CULLMODE_FRONT;
@@ -229,7 +236,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
    /* XXX clamp max depends on AA vs. non-AA */
 
    sf.sf7.sprite_point = key->point_sprite;
-   sf.sf7.point_size = CLAMP(nearbyint(key->point_size), 1, 255) * (1<<3);
+   sf.sf7.point_size = CLAMP(rint(key->point_size), 1, 255) * (1<<3);
    sf.sf7.use_point_size_state = !key->point_attenuated;
    sf.sf7.aa_line_distance_mode = 0;
 
@@ -294,8 +301,7 @@ const struct brw_tracked_state brw_sf_unit = {
 		_NEW_LINE | 
 		_NEW_POINT | 
 		_NEW_SCISSOR),
-      .brw   = (BRW_NEW_URB_FENCE |
-		BRW_NEW_METAOPS),
+      .brw   = BRW_NEW_URB_FENCE,
       .cache = (CACHE_NEW_SF_VP |
 		CACHE_NEW_SF_PROG)
    },
diff --git a/src/mesa/drivers/dri/i965/brw_state.h b/src/mesa/drivers/dri/i965/brw_state.h
index bb22c03eeb6..df839c5b300 100644
--- a/src/mesa/drivers/dri/i965/brw_state.h
+++ b/src/mesa/drivers/dri/i965/brw_state.h
@@ -83,7 +83,6 @@ const struct brw_tracked_state brw_wm_unit;
 
 const struct brw_tracked_state brw_psp_urb_cbs;
 
-const struct brw_tracked_state brw_active_vertprog;
 const struct brw_tracked_state brw_pipe_control;
 
 const struct brw_tracked_state brw_clear_surface_cache;
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index 7a642bd2a8f..5de1450e612 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -181,11 +181,123 @@ brw_clear_validated_bos(struct brw_context *brw)
    brw->state.validated_bo_count = 0;
 }
 
+struct dirty_bit_map {
+   uint32_t bit;
+   char *name;
+   uint32_t count;
+};
+
+#define DEFINE_BIT(name) {name, #name, 0}
+
+static struct dirty_bit_map mesa_bits[] = {
+   DEFINE_BIT(_NEW_MODELVIEW),
+   DEFINE_BIT(_NEW_PROJECTION),
+   DEFINE_BIT(_NEW_TEXTURE_MATRIX),
+   DEFINE_BIT(_NEW_COLOR_MATRIX),
+   DEFINE_BIT(_NEW_ACCUM),
+   DEFINE_BIT(_NEW_COLOR),
+   DEFINE_BIT(_NEW_DEPTH),
+   DEFINE_BIT(_NEW_EVAL),
+   DEFINE_BIT(_NEW_FOG),
+   DEFINE_BIT(_NEW_HINT),
+   DEFINE_BIT(_NEW_LIGHT),
+   DEFINE_BIT(_NEW_LINE),
+   DEFINE_BIT(_NEW_PIXEL),
+   DEFINE_BIT(_NEW_POINT),
+   DEFINE_BIT(_NEW_POLYGON),
+   DEFINE_BIT(_NEW_POLYGONSTIPPLE),
+   DEFINE_BIT(_NEW_SCISSOR),
+   DEFINE_BIT(_NEW_STENCIL),
+   DEFINE_BIT(_NEW_TEXTURE),
+   DEFINE_BIT(_NEW_TRANSFORM),
+   DEFINE_BIT(_NEW_VIEWPORT),
+   DEFINE_BIT(_NEW_PACKUNPACK),
+   DEFINE_BIT(_NEW_ARRAY),
+   DEFINE_BIT(_NEW_RENDERMODE),
+   DEFINE_BIT(_NEW_BUFFERS),
+   DEFINE_BIT(_NEW_MULTISAMPLE),
+   DEFINE_BIT(_NEW_TRACK_MATRIX),
+   DEFINE_BIT(_NEW_PROGRAM),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map brw_bits[] = {
+   DEFINE_BIT(BRW_NEW_URB_FENCE),
+   DEFINE_BIT(BRW_NEW_FRAGMENT_PROGRAM),
+   DEFINE_BIT(BRW_NEW_VERTEX_PROGRAM),
+   DEFINE_BIT(BRW_NEW_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_CURBE_OFFSETS),
+   DEFINE_BIT(BRW_NEW_REDUCED_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_PRIMITIVE),
+   DEFINE_BIT(BRW_NEW_CONTEXT),
+   DEFINE_BIT(BRW_NEW_WM_INPUT_DIMENSIONS),
+   DEFINE_BIT(BRW_NEW_INPUT_VARYING),
+   DEFINE_BIT(BRW_NEW_PSP),
+   DEFINE_BIT(BRW_NEW_FENCE),
+   DEFINE_BIT(BRW_NEW_INDICES),
+   DEFINE_BIT(BRW_NEW_VERTICES),
+   DEFINE_BIT(BRW_NEW_BATCH),
+   DEFINE_BIT(BRW_NEW_DEPTH_BUFFER),
+   {0, 0, 0}
+};
+
+static struct dirty_bit_map cache_bits[] = {
+   DEFINE_BIT(CACHE_NEW_CC_VP),
+   DEFINE_BIT(CACHE_NEW_CC_UNIT),
+   DEFINE_BIT(CACHE_NEW_WM_PROG),
+   DEFINE_BIT(CACHE_NEW_SAMPLER_DEFAULT_COLOR),
+   DEFINE_BIT(CACHE_NEW_SAMPLER),
+   DEFINE_BIT(CACHE_NEW_WM_UNIT),
+   DEFINE_BIT(CACHE_NEW_SF_PROG),
+   DEFINE_BIT(CACHE_NEW_SF_VP),
+   DEFINE_BIT(CACHE_NEW_SF_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_UNIT),
+   DEFINE_BIT(CACHE_NEW_VS_PROG),
+   DEFINE_BIT(CACHE_NEW_GS_UNIT),
+   DEFINE_BIT(CACHE_NEW_GS_PROG),
+   DEFINE_BIT(CACHE_NEW_CLIP_VP),
+   DEFINE_BIT(CACHE_NEW_CLIP_UNIT),
+   DEFINE_BIT(CACHE_NEW_CLIP_PROG),
+   DEFINE_BIT(CACHE_NEW_SURFACE),
+   DEFINE_BIT(CACHE_NEW_SURF_BIND),
+   {0, 0, 0}
+};
+
+
+static void
+brw_update_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      if (bit_map[i].bit & bits)
+	 bit_map[i].count++;
+   }
+}
+
+static void
+brw_print_dirty_count(struct dirty_bit_map *bit_map, int32_t bits)
+{
+   int i;
+
+   for (i = 0; i < 32; i++) {
+      if (bit_map[i].bit == 0)
+	 return;
+
+      fprintf(stderr, "0x%08x: %12d (%s)\n",
+	      bit_map[i].bit, bit_map[i].count, bit_map[i].name);
+   }
+}
+
 /***********************************************************************
  * Emit all state:
  */
 void brw_validate_state( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = &brw->intel;
    struct brw_state_flags *state = &brw->state.dirty;
    GLuint i;
@@ -202,17 +314,13 @@ void brw_validate_state( struct brw_context *brw )
       state->brw |= ~0;
    }
 
-   /* texenv program needs to notify us somehow when this happens: 
-    * Some confusion about which state flag should represent this change.
-    */
-   if (brw->fragment_program != brw->attribs.FragmentProgram->_Current) {
-      brw->fragment_program = brw->attribs.FragmentProgram->_Current;
-      brw->state.dirty.mesa |= _NEW_PROGRAM;
+   if (brw->fragment_program != ctx->FragmentProgram._Current) {
+      brw->fragment_program = ctx->FragmentProgram._Current;
       brw->state.dirty.brw |= BRW_NEW_FRAGMENT_PROGRAM;
    }
 
-   if (brw->vertex_program != brw->attribs.VertexProgram->_Current) {
-      brw->vertex_program = brw->attribs.VertexProgram->_Current;
+   if (brw->vertex_program != ctx->VertexProgram._Current) {
+      brw->vertex_program = ctx->VertexProgram._Current;
       brw->state.dirty.brw |= BRW_NEW_VERTEX_PROGRAM;
    }
 
@@ -246,6 +354,7 @@ void brw_upload_state(struct brw_context *brw)
 {
    struct brw_state_flags *state = &brw->state.dirty;
    int i;
+   static int dirty_count = 0;
 
    brw_clear_validated_bos(brw);
 
@@ -301,6 +410,18 @@ void brw_upload_state(struct brw_context *brw)
       }
    }
 
+   if (INTEL_DEBUG & DEBUG_STATE) {
+      brw_update_dirty_count(mesa_bits, state->mesa);
+      brw_update_dirty_count(brw_bits, state->brw);
+      brw_update_dirty_count(cache_bits, state->cache);
+      if (dirty_count++ % 1000 == 0) {
+	 brw_print_dirty_count(mesa_bits, state->mesa);
+	 brw_print_dirty_count(brw_bits, state->brw);
+	 brw_print_dirty_count(cache_bits, state->cache);
+	 fprintf(stderr, "\n");
+      }
+   }
+
    if (!brw->intel.Fallback)
       memset(state, 0, sizeof(*state));
 }
diff --git a/src/mesa/drivers/dri/i965/brw_structs.h b/src/mesa/drivers/dri/i965/brw_structs.h
index 4e577d0f6a8..d97ff27f0a1 100644
--- a/src/mesa/drivers/dri/i965/brw_structs.h
+++ b/src/mesa/drivers/dri/i965/brw_structs.h
@@ -1031,10 +1031,10 @@ struct brw_surface_state
       GLuint writedisable_green:1; 
       GLuint writedisable_red:1; 
       GLuint writedisable_alpha:1; 
-      GLuint surface_format:9; 
+      GLuint surface_format:9;     /**< BRW_SURFACEFORMAT_x */
       GLuint data_return_format:1; 
       GLuint pad0:1;
-      GLuint surface_type:3; 
+      GLuint surface_type:3;       /**< BRW_SURFACE_1D/2D/3D/CUBE */
    } ss0;
    
    struct {
diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c
index 0bb6f176a0d..ef99e9c1aea 100644
--- a/src/mesa/drivers/dri/i965/brw_tex.c
+++ b/src/mesa/drivers/dri/i965/brw_tex.c
@@ -86,11 +86,12 @@ void brw_FrameBufferTexDestroy( struct brw_context *brw )
  */
 void brw_validate_textures( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = &brw->intel;
    int i;
 
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
 
       if (texUnit->_ReallyEnabled) {
 	 intel_finalize_mipmap_tree(intel, i);
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index 1db7ceebcfb..e3111c66800 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -85,6 +85,7 @@ static void do_vs_prog( struct brw_context *brw,
 
 static void brw_upload_vs_prog(struct brw_context *brw)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct brw_vs_prog_key key;
    struct brw_vertex_program *vp = 
       (struct brw_vertex_program *)brw->vertex_program;
@@ -97,14 +98,9 @@ static void brw_upload_vs_prog(struct brw_context *brw)
     * the inputs it asks for, whether they are varying or not.
     */
    key.program_string_id = vp->id;
-   key.nr_userclip = brw_count_bits(brw->attribs.Transform->ClipPlanesEnabled);
-   key.copy_edgeflag = (brw->attribs.Polygon->FrontMode != GL_FILL ||
-			brw->attribs.Polygon->BackMode != GL_FILL);
-
-   /* BRW_NEW_METAOPS
-    */
-   if (brw->metaops.active)
-      key.know_w_is_one = 1;
+   key.nr_userclip = brw_count_bits(ctx->Transform.ClipPlanesEnabled);
+   key.copy_edgeflag = (ctx->Polygon.FrontMode != GL_FILL ||
+			ctx->Polygon.BackMode != GL_FILL);
 
    /* Make an early check for the key.
     */
@@ -123,7 +119,7 @@ static void brw_upload_vs_prog(struct brw_context *brw)
 const struct brw_tracked_state brw_vs_prog = {
    .dirty = {
       .mesa  = _NEW_TRANSFORM | _NEW_POLYGON,
-      .brw   = BRW_NEW_VERTEX_PROGRAM | BRW_NEW_METAOPS,
+      .brw   = BRW_NEW_VERTEX_PROGRAM,
       .cache = 0
    },
    .prepare = brw_upload_vs_prog
diff --git a/src/mesa/drivers/dri/i965/brw_vs.h b/src/mesa/drivers/dri/i965/brw_vs.h
index 22388ec99d0..99d0e937226 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.h
+++ b/src/mesa/drivers/dri/i965/brw_vs.h
@@ -43,7 +43,6 @@ struct brw_vs_prog_key {
    GLuint program_string_id;
    GLuint nr_userclip:4;
    GLuint copy_edgeflag:1;
-   GLuint know_w_is_one:1;
    GLuint pad:26;
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_constval.c b/src/mesa/drivers/dri/i965/brw_vs_constval.c
index 6fbac02de63..9977677fd74 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_constval.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_constval.c
@@ -168,6 +168,7 @@ static GLuint get_input_size(struct brw_context *brw,
  */
 static void calc_wm_input_sizes( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    /* BRW_NEW_VERTEX_PROGRAM */
    struct brw_vertex_program *vp = 
       (struct brw_vertex_program *)brw->vertex_program;
@@ -179,7 +180,7 @@ static void calc_wm_input_sizes( struct brw_context *brw )
    memset(&t, 0, sizeof(t));
 
    /* _NEW_LIGHT */
-   if (brw->attribs.Light->Model.TwoSide)
+   if (ctx->Light.Model.TwoSide)
       t.twoside = 1;
 
    for (i = 0; i < VERT_ATTRIB_MAX; i++) 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 25b4ee85cb0..24b7dc30fe1 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -73,8 +73,6 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 
    c->prog_data.curb_read_length = reg - 1;
 
-
-
    /* Allocate input regs:  
     */
    c->nr_inputs = 0;
@@ -84,8 +82,7 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 	 c->regs[PROGRAM_INPUT][i] = brw_vec8_grf(reg, 0);
 	 reg++;
       }
-   }     
-
+   }
 
    /* Allocate outputs: TODO: could organize the non-position outputs
     * to go straight into message regs.
@@ -196,6 +193,7 @@ static void unalias1( struct brw_vs_compile *c,
       struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
       func(c, tmp, arg0);
       brw_MOV(p, dst, tmp);
+      release_tmp(c, tmp);
    }
    else {
       func(c, dst, arg0);
@@ -217,12 +215,38 @@ static void unalias2( struct brw_vs_compile *c,
       struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
       func(c, tmp, arg0, arg1);
       brw_MOV(p, dst, tmp);
+      release_tmp(c, tmp);
    }
    else {
       func(c, dst, arg0, arg1);
    }
 }
 
+static void unalias3( struct brw_vs_compile *c,
+		      struct brw_reg dst,
+		      struct brw_reg arg0,
+		      struct brw_reg arg1,
+		      struct brw_reg arg2,
+		      void (*func)( struct brw_vs_compile *,
+				    struct brw_reg,
+				    struct brw_reg,
+				    struct brw_reg,
+				    struct brw_reg ))
+{
+   if ((dst.file == arg0.file && dst.nr == arg0.nr) ||
+       (dst.file == arg1.file && dst.nr == arg1.nr) ||
+       (dst.file == arg2.file && dst.nr == arg2.nr)) {
+      struct brw_compile *p = &c->func;
+      struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
+      func(c, tmp, arg0, arg1, arg2);
+      brw_MOV(p, dst, tmp);
+      release_tmp(c, tmp);
+   }
+   else {
+      func(c, dst, arg0, arg1, arg2);
+   }
+}
+
 static void emit_sop( struct brw_compile *p,
                       struct brw_reg dst,
                       struct brw_reg arg0,
@@ -339,6 +363,7 @@ static void emit_math1( struct brw_vs_compile *c,
    }
 }
 
+
 static void emit_math2( struct brw_vs_compile *c, 
 			GLuint function,
 			struct brw_reg dst,
@@ -370,7 +395,6 @@ static void emit_math2( struct brw_vs_compile *c,
       release_tmp(c, tmp);
    }
 }
-		     
 
 
 static void emit_exp_noalias( struct brw_vs_compile *c,
@@ -420,7 +444,7 @@ static void emit_exp_noalias( struct brw_vs_compile *c,
 		 BRW_MATH_FUNCTION_EXP, 
 		 brw_writemask(dst, WRITEMASK_Z),
 		 brw_swizzle1(arg0, 0), 
-		 BRW_MATH_PRECISION_PARTIAL);
+		 BRW_MATH_PRECISION_FULL);
    }  
 
    if (dst.dw1.bits.writemask & WRITEMASK_W) {
@@ -521,8 +545,6 @@ static void emit_log_noalias( struct brw_vs_compile *c,
 }
 
 
-      
-
 /* Need to unalias - consider swizzles:   r0 = DST r0.xxxx r1
  */
 static void emit_dst_noalias( struct brw_vs_compile *c, 
@@ -544,6 +566,7 @@ static void emit_dst_noalias( struct brw_vs_compile *c,
       brw_MOV(p, brw_writemask(dst, WRITEMASK_W), arg1);
 }
 
+
 static void emit_xpd( struct brw_compile *p,
 		      struct brw_reg dst,
 		      struct brw_reg t,
@@ -554,7 +577,6 @@ static void emit_xpd( struct brw_compile *p,
 }
 
 
-
 static void emit_lit_noalias( struct brw_vs_compile *c, 
 			      struct brw_reg dst,
 			      struct brw_reg arg0 )
@@ -595,8 +617,42 @@ static void emit_lit_noalias( struct brw_vs_compile *c,
    brw_ENDIF(p, if_insn);
 }
 
+static void emit_lrp_noalias(struct brw_vs_compile *c,
+			     struct brw_reg dst,
+			     struct brw_reg arg0,
+			     struct brw_reg arg1,
+			     struct brw_reg arg2)
+{
+   struct brw_compile *p = &c->func;
+
+   brw_ADD(p, dst, negate(arg0), brw_imm_f(1.0));
+   brw_MUL(p, brw_null_reg(), dst, arg2);
+   brw_MAC(p, dst, arg0, arg1);
+}
+
+/** 3 or 4-component vector normalization */
+static void emit_nrm( struct brw_vs_compile *c, 
+                      struct brw_reg dst,
+                      struct brw_reg arg0,
+                      int num_comps)
+{
+   struct brw_compile *p = &c->func;
+   struct brw_reg tmp = get_tmp(c);
+
+   /* tmp = dot(arg0, arg0) */
+   if (num_comps == 3)
+      brw_DP3(p, tmp, arg0, arg0);
+   else
+      brw_DP4(p, tmp, arg0, arg0);
+
+   /* tmp = 1 / sqrt(tmp) */
+   emit_math1(c, BRW_MATH_FUNCTION_RSQ, tmp, tmp, BRW_MATH_PRECISION_FULL);
 
+   /* dst = arg0 * tmp */
+   brw_MUL(p, dst, arg0, tmp);
 
+   release_tmp(c, tmp);
+}
 
 
 /* TODO: relative addressing!
@@ -634,7 +690,6 @@ static struct brw_reg get_reg( struct brw_vs_compile *c,
 }
 
 
-
 static struct brw_reg deref( struct brw_vs_compile *c,
 			     struct brw_reg arg,
 			     GLint offset)
@@ -728,8 +783,6 @@ static struct brw_reg get_dst( struct brw_vs_compile *c,
 }
 
 
-
-
 static void emit_swz( struct brw_vs_compile *c, 
 		      struct brw_reg dst,
 		      struct prog_src_register src )
@@ -801,8 +854,8 @@ static void emit_swz( struct brw_vs_compile *c,
 }
 
 
-
-/* Post-vertex-program processing.  Send the results to the URB.
+/**
+ * Post-vertex-program processing.  Send the results to the URB.
  */
 static void emit_vertex_write( struct brw_vs_compile *c)
 {
@@ -817,23 +870,16 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	      get_reg(c, PROGRAM_INPUT, VERT_ATTRIB_EDGEFLAG));
    }
 
-
    /* Build ndc coords */
-   if (!c->key.know_w_is_one) {
-      ndc = get_tmp(c);
-      emit_math1(c, BRW_MATH_FUNCTION_INV, ndc, brw_swizzle1(pos, 3), BRW_MATH_PRECISION_FULL);
-      brw_MUL(p, brw_writemask(ndc, WRITEMASK_XYZ), pos, ndc);
-   }
-   else {
-      ndc = pos;
-   }
+   ndc = get_tmp(c);
+   emit_math1(c, BRW_MATH_FUNCTION_INV, ndc, brw_swizzle1(pos, 3), BRW_MATH_PRECISION_FULL);
+   brw_MUL(p, brw_writemask(ndc, WRITEMASK_XYZ), pos, ndc);
 
    /* Update the header for point size, user clipping flags, and -ve rhw
     * workaround.
     */
    if ((c->prog_data.outputs_written & (1<<VERT_RESULT_PSIZ)) ||
-       c->key.nr_userclip ||
-       (!BRW_IS_G4X(p->brw) && !c->key.know_w_is_one))
+       c->key.nr_userclip || !BRW_IS_G4X(p->brw))
    {
       struct brw_reg header1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
       GLuint i;
@@ -848,7 +894,6 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	 brw_AND(p, brw_writemask(header1, WRITEMASK_W), header1, brw_imm_ud(0x7ff<<8));
       }
 
-
       for (i = 0; i < c->key.nr_userclip; i++) {
 	 brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
 	 brw_DP4(p, brw_null_reg(), pos, c->userplane[i]);
@@ -856,7 +901,6 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
       }
 
-
       /* i965 clipping workaround: 
        * 1) Test for -ve rhw
        * 2) If set, 
@@ -866,7 +910,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
        * Later, clipping will detect ucp[6] and ensure the primitive is
        * clipped against all fixed planes.
        */
-      if (!BRW_IS_G4X(p->brw) && !c->key.know_w_is_one) {
+      if (!BRW_IS_G4X(p->brw)) {
 	 brw_CMP(p,
 		 vec8(brw_null_reg()),
 		 BRW_CONDITIONAL_L,
@@ -888,14 +932,12 @@ static void emit_vertex_write( struct brw_vs_compile *c)
       brw_MOV(p, retype(brw_message_reg(1), BRW_REGISTER_TYPE_UD), brw_imm_ud(0));
    }
 
-
    /* Emit the (interleaved) headers for the two vertices - an 8-reg
     * of zeros followed by two sets of NDC coordinates:
     */
    brw_set_access_mode(p, BRW_ALIGN_1);
    brw_MOV(p, offset(m0, 2), ndc);
    brw_MOV(p, offset(m0, 3), pos);
-   
 
    brw_urb_WRITE(p, 
 		 brw_null_reg(), /* dest */
@@ -909,9 +951,9 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 		 1, 		/* writes complete */
 		 0, 		/* urb destination offset */
 		 BRW_URB_SWIZZLE_INTERLEAVE);
-
 }
 
+
 static void 
 post_vs_emit( struct brw_vs_compile *c, struct brw_instruction *end_inst )
 {
@@ -958,7 +1000,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
    GLuint file;
 
    if (INTEL_DEBUG & DEBUG_VS) {
-      _mesa_printf("\n\n\nvs-emit:\n");
+      _mesa_printf("vs-emit:\n");
       _mesa_print_program(&c->vp->program.Base); 
       _mesa_printf("\n");
    }
@@ -1016,6 +1058,11 @@ void brw_vs_emit(struct brw_vs_compile *c )
       else
 	  dst = get_dst(c, inst->DstReg);
 
+      if (inst->SaturateMode != SATURATE_OFF) {
+	 _mesa_problem(NULL, "Unsupported saturate %d in vertex shader",
+                       inst->SaturateMode);
+      }
+
       switch (inst->Opcode) {
       case OPCODE_ABS:
 	 brw_MOV(p, dst, brw_abs(args[0]));
@@ -1035,6 +1082,12 @@ void brw_vs_emit(struct brw_vs_compile *c )
       case OPCODE_DPH:
 	 brw_DPH(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_NRM3:
+	 emit_nrm(c, dst, args[0], 3);
+	 break;
+      case OPCODE_NRM4:
+	 emit_nrm(c, dst, args[0], 4);
+	 break;
       case OPCODE_DST:
 	 unalias2(c, dst, args[0], args[1], emit_dst_noalias); 
 	 break;
@@ -1062,6 +1115,9 @@ void brw_vs_emit(struct brw_vs_compile *c )
       case OPCODE_LIT:
 	 unalias1(c, dst, args[0], emit_lit_noalias);
 	 break;
+      case OPCODE_LRP:
+	 unalias3(c, dst, args[0], args[1], args[2], emit_lrp_noalias);
+	 break;
       case OPCODE_MAD:
 	 brw_MOV(p, brw_acc_reg(), args[2]);
 	 brw_MAC(p, dst, args[0], args[1]);
@@ -1102,7 +1158,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 break;
       case OPCODE_SGT:
          emit_sgt(p, dst, args[0], args[1]);
-        break;
+         break;
       case OPCODE_SLT:
 	 emit_slt(p, dst, args[0], args[1]);
 	 break;
@@ -1118,6 +1174,10 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	  */
 	 emit_swz(c, dst, inst->SrcReg[0] );
 	 break;
+      case OPCODE_TRUNC:
+         /* round toward zero */
+	 brw_RNDZ(p, dst, args[0]);
+	 break;
       case OPCODE_XPD:
 	 emit_xpd(p, dst, args[0], args[1]);
 	 break;
@@ -1136,7 +1196,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
          brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
          brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
          brw_set_predicate_control_flag_value(p, 0xff);
-        break;
+         break;
       case OPCODE_CAL:
 	 brw_set_access_mode(p, BRW_ALIGN_1);
 	 brw_ADD(p, deref_1d(stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
@@ -1145,7 +1205,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 			 get_addr_reg(stack_index), brw_imm_d(4));
 	 inst->Data = &p->store[p->nr_insn];
 	 brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-        break;
+         break;
       case OPCODE_RET:
 	 brw_ADD(p, get_addr_reg(stack_index),
 			 get_addr_reg(stack_index), brw_imm_d(-4));
@@ -1154,17 +1214,17 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 brw_set_access_mode(p, BRW_ALIGN_16);
       case OPCODE_END:	
          brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-        break;
+         break;
       case OPCODE_PRINT:
       case OPCODE_BGNSUB:
       case OPCODE_ENDSUB:
+         /* no-op instructions */
 	 break;
       default:
-	 _mesa_printf("Unsupported opcode %i (%s) in vertex shader\n",
-		      inst->Opcode, inst->Opcode < MAX_OPCODE ?
+	 _mesa_problem(NULL, "Unsupported opcode %i (%s) in vertex shader",
+                       inst->Opcode, inst->Opcode < MAX_OPCODE ?
 				    _mesa_opcode_string(inst->Opcode) :
 				    "unknown");
-	 break;
       }
 
       if ((inst->DstReg.File == PROGRAM_OUTPUT)
diff --git a/src/mesa/drivers/dri/i965/brw_vs_state.c b/src/mesa/drivers/dri/i965/brw_vs_state.c
index 942581696d8..1a63766ea1f 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_state.c
@@ -49,6 +49,8 @@ struct brw_vs_unit_key {
 static void
 vs_unit_populate_key(struct brw_context *brw, struct brw_vs_unit_key *key)
 {
+   GLcontext *ctx = &brw->intel.ctx;
+
    memset(key, 0, sizeof(*key));
 
    /* CACHE_NEW_VS_PROG */
@@ -61,7 +63,7 @@ vs_unit_populate_key(struct brw_context *brw, struct brw_vs_unit_key *key)
    key->urb_size = brw->urb.vsize;
 
    /* BRW_NEW_CURBE_OFFSETS, _NEW_TRANSFORM */
-   if (brw->attribs.Transform->ClipPlanesEnabled) {
+   if (ctx->Transform.ClipPlanesEnabled) {
       /* Note that we read in the userclip planes as well, hence
        * clip_start:
        */
diff --git a/src/mesa/drivers/dri/i965/brw_vs_tnl.c b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
deleted file mode 100644
index eacc289f1f1..00000000000
--- a/src/mesa/drivers/dri/i965/brw_vs_tnl.c
+++ /dev/null
@@ -1,59 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.3
- *
- * Copyright (C) 2005  Tungsten Graphics   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * TUNGSTEN GRAPHICS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
- * WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/**
- * \file t_vp_build.c
- * Create a vertex program to execute the current fixed function T&L pipeline.
- * \author Keith Whitwell
- */
-
-
-#include "main/glheader.h"
-#include "main/macros.h"
-#include "main/enums.h"
-#include "brw_vs.h"
-#include "brw_state.h"
-
-
-static void prepare_active_vertprog( struct brw_context *brw )
-{
-   const struct gl_vertex_program *prev = brw->vertex_program;
-
-   brw->vertex_program = brw->attribs.VertexProgram->_Current;
-
-   if (brw->vertex_program != prev) 
-      brw->state.dirty.brw |= BRW_NEW_VERTEX_PROGRAM;
-}
-
-
-
-const struct brw_tracked_state brw_active_vertprog = {
-   .dirty = {
-      .mesa = _NEW_PROGRAM,
-      .brw = 0,
-      .cache = 0
-   },
-   .prepare = prepare_active_vertprog
-};
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 2d4c81274e6..b501a59ccd5 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -23,14 +23,12 @@
  OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
  WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <[email protected]>
-  */
-            
-
+**********************************************************************/
 
+/*
+ * Authors:
+ *   Keith Whitwell <[email protected]>
+ */
 
 #include "main/glheader.h"
 #include "main/mtypes.h"
@@ -44,12 +42,11 @@
 #include "brw_context.h"
 #include "brw_defines.h"
 #include "brw_state.h"
-
 #include "brw_draw.h"
 #include "brw_state.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
-#include <stdarg.h>
+
 
 static void
 dri_bo_release(dri_bo **bo)
@@ -58,21 +55,22 @@ dri_bo_release(dri_bo **bo)
    *bo = NULL;
 }
 
-/* called from intelDestroyContext()
+
+/**
+ * called from intelDestroyContext()
  */
 static void brw_destroy_context( struct intel_context *intel )
 {
    struct brw_context *brw = brw_context(&intel->ctx);
    int i;
 
-   brw_destroy_metaops(brw);
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
 
    brw_FrameBufferTexDestroy( brw );
 
    for (i = 0; i < brw->state.nr_draw_regions; i++)
-       intel_region_release(&brw->state.draw_regions[i]);
+      intel_region_release(&brw->state.draw_regions[i]);
    brw->state.nr_draw_regions = 0;
    intel_region_release(&brw->state.depth_region);
 
@@ -99,37 +97,46 @@ static void brw_destroy_context( struct intel_context *intel )
    dri_bo_release(&brw->cc.vp_bo);
 }
 
-/* called from intelDrawBuffer()
+
+/**
+ * called from intelDrawBuffer()
  */
 static void brw_set_draw_region( struct intel_context *intel, 
-				  struct intel_region *draw_regions[],
-				  struct intel_region *depth_region,
-				GLuint num_regions)
+                                 struct intel_region *draw_regions[],
+                                 struct intel_region *depth_region,
+                                 GLuint num_regions)
 {
    struct brw_context *brw = brw_context(&intel->ctx);
    int i;
+
+   /* release old color/depth regions */
    if (brw->state.depth_region != depth_region)
       brw->state.dirty.brw |= BRW_NEW_DEPTH_BUFFER;
    for (i = 0; i < brw->state.nr_draw_regions; i++)
        intel_region_release(&brw->state.draw_regions[i]);
    intel_region_release(&brw->state.depth_region);
+
+   /* reference new color/depth regions */
    for (i = 0; i < num_regions; i++)
        intel_region_reference(&brw->state.draw_regions[i], draw_regions[i]);
    intel_region_reference(&brw->state.depth_region, depth_region);
    brw->state.nr_draw_regions = num_regions;
 }
 
-/* called from intel_batchbuffer_flush and children before sending a
+
+/**
+ * called from intel_batchbuffer_flush and children before sending a
  * batchbuffer off.
  */
 static void brw_finish_batch(struct intel_context *intel)
 {
    struct brw_context *brw = brw_context(&intel->ctx);
-
    brw_emit_query_end(brw);
 }
 
-/* called from intelFlushBatchLocked
+
+/**
+ * called from intelFlushBatchLocked
  */
 static void brw_new_batch( struct intel_context *intel )
 {
@@ -160,24 +167,21 @@ static void brw_new_batch( struct intel_context *intel )
    }
 }
 
-static void brw_note_fence( struct intel_context *intel, 
-			    GLuint fence )
+
+static void brw_note_fence( struct intel_context *intel, GLuint fence )
 {
    brw_context(&intel->ctx)->state.dirty.brw |= BRW_NEW_FENCE;
 }
- 
+
+
 static void brw_note_unlock( struct intel_context *intel )
 {
    struct brw_context *brw = brw_context(&intel->ctx);
-
    brw_state_cache_check_size(brw);
-
-   brw_context(&intel->ctx)->state.dirty.brw |= BRW_NEW_LOCK;
 }
 
 
-void brw_do_flush( struct brw_context *brw, 
-		   GLuint flags )
+void brw_do_flush( struct brw_context *brw, GLuint flags )
 {
    struct brw_mi_flush flush;
    memset(&flush, 0, sizeof(flush));      
@@ -187,8 +191,7 @@ void brw_do_flush( struct brw_context *brw,
 }
 
 
-static void brw_emit_flush( struct intel_context *intel,
-			GLuint unused )
+static void brw_emit_flush( struct intel_context *intel, GLuint unused )
 {
    brw_do_flush(brw_context(&intel->ctx),
 		BRW_FLUSH_STATE_CACHE|BRW_FLUSH_READ_CACHE);
@@ -208,6 +211,7 @@ static GLuint brw_flush_cmd( void )
    return *(GLuint *)&flush;
 }
 
+
 static void brw_invalidate_state( struct intel_context *intel, GLuint new_state )
 {
    /* nothing */
@@ -217,14 +221,14 @@ static void brw_invalidate_state( struct intel_context *intel, GLuint new_state
 void brwInitVtbl( struct brw_context *brw )
 {
    brw->intel.vtbl.check_vertex_size = 0;
-   brw->intel.vtbl.emit_state = 0; 
-   brw->intel.vtbl.reduced_primitive_state = 0;	
+   brw->intel.vtbl.emit_state = 0;
+   brw->intel.vtbl.reduced_primitive_state = 0;
    brw->intel.vtbl.render_start = 0;
-   brw->intel.vtbl.update_texture_state = 0; 
+   brw->intel.vtbl.update_texture_state = 0;
 
-   brw->intel.vtbl.invalidate_state = brw_invalidate_state; 
-   brw->intel.vtbl.note_fence = brw_note_fence; 
-   brw->intel.vtbl.note_unlock = brw_note_unlock; 
+   brw->intel.vtbl.invalidate_state = brw_invalidate_state;
+   brw->intel.vtbl.note_fence = brw_note_fence;
+   brw->intel.vtbl.note_unlock = brw_note_unlock;
    brw->intel.vtbl.new_batch = brw_new_batch;
    brw->intel.vtbl.finish_batch = brw_finish_batch;
    brw->intel.vtbl.destroy = brw_destroy_context;
@@ -233,4 +237,3 @@ void brwInitVtbl( struct brw_context *brw )
    brw->intel.vtbl.emit_flush = brw_emit_flush;
    brw->intel.vtbl.debug_batch = brw_debug_batch;
 }
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 361312c2ca1..ea708a06815 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -36,63 +36,24 @@
 #include "brw_state.h"
 
 
+/** Return number of src args for given instruction */
 GLuint brw_wm_nr_args( GLuint opcode )
 {
    switch (opcode) {
-
    case WM_PIXELXY:
-   case OPCODE_ABS:
-   case OPCODE_FLR:
-   case OPCODE_FRC:
-   case OPCODE_SWZ:
-   case OPCODE_MOV:
-   case OPCODE_COS:
-   case OPCODE_EX2:
-   case OPCODE_LG2:
-   case OPCODE_RCP:
-   case OPCODE_RSQ:
-   case OPCODE_SIN:
-   case OPCODE_SCS:
-   case OPCODE_TEX:
-   case OPCODE_TXB:
-   case OPCODE_TXP:	
-   case OPCODE_KIL:
-   case OPCODE_LIT: 
-   case WM_CINTERP: 
-   case WM_WPOSXY: 
+   case WM_CINTERP:
+   case WM_WPOSXY:
       return 1;
-
-   case OPCODE_POW:
-   case OPCODE_SUB:
-   case OPCODE_SGE:
-   case OPCODE_SGT:
-   case OPCODE_SLE:
-   case OPCODE_SLT:
-   case OPCODE_SEQ:
-   case OPCODE_SNE:
-   case OPCODE_ADD:
-   case OPCODE_MAX:
-   case OPCODE_MIN:
-   case OPCODE_MUL:
-   case OPCODE_XPD:
-   case OPCODE_DP3:	
-   case OPCODE_DP4:
-   case OPCODE_DPH:
-   case OPCODE_DST:
-   case WM_LINTERP: 
+   case WM_LINTERP:
    case WM_DELTAXY:
    case WM_PIXELW:
       return 2;
-
    case WM_FB_WRITE:
-   case WM_PINTERP: 
-   case OPCODE_MAD:	
-   case OPCODE_CMP:
-   case OPCODE_LRP:
+   case WM_PINTERP:
       return 3;
-      
    default:
-      return 0;
+      assert(opcode < MAX_OPCODE);
+      return _mesa_num_inst_src_regs(opcode);
    }
 }
 
@@ -175,6 +136,9 @@ static void do_wm_prog( struct brw_context *brw,
 	*/
        brw_wm_emit(c);
    }
+   if (INTEL_DEBUG & DEBUG_WM)
+      fprintf(stderr, "\n");
+
    /* get the program
     */
    program = brw_get_program(&c->func, &program_size);
@@ -193,6 +157,7 @@ static void do_wm_prog( struct brw_context *brw,
 static void brw_wm_populate_key( struct brw_context *brw,
 				 struct brw_wm_prog_key *key )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    /* BRW_NEW_FRAGMENT_PROGRAM */
    struct brw_fragment_program *fp = 
       (struct brw_fragment_program *)brw->fragment_program;
@@ -206,57 +171,50 @@ static void brw_wm_populate_key( struct brw_context *brw,
     */
    /* _NEW_COLOR */
    if (fp->program.UsesKill ||
-       brw->attribs.Color->AlphaEnabled)
+       ctx->Color.AlphaEnabled)
       lookup |= IZ_PS_KILL_ALPHATEST_BIT;
 
    if (fp->program.Base.OutputsWritten & (1<<FRAG_RESULT_DEPR))
       lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
 
    /* _NEW_DEPTH */
-   if (brw->attribs.Depth->Test)
+   if (ctx->Depth.Test)
       lookup |= IZ_DEPTH_TEST_ENABLE_BIT;
 
-   if (brw->attribs.Depth->Test &&  
-       brw->attribs.Depth->Mask) /* ?? */
+   if (ctx->Depth.Test &&  
+       ctx->Depth.Mask) /* ?? */
       lookup |= IZ_DEPTH_WRITE_ENABLE_BIT;
 
    /* _NEW_STENCIL */
-   if (brw->attribs.Stencil->Enabled) {
+   if (ctx->Stencil.Enabled) {
       lookup |= IZ_STENCIL_TEST_ENABLE_BIT;
 
-      if (brw->attribs.Stencil->WriteMask[0] ||
-	  (brw->attribs.Stencil->_TestTwoSide &&
-	   brw->attribs.Stencil->WriteMask[1]))
+      if (ctx->Stencil.WriteMask[0] ||
+	  ctx->Stencil.WriteMask[ctx->Stencil._BackFace])
 	 lookup |= IZ_STENCIL_WRITE_ENABLE_BIT;
    }
 
-   /* XXX: when should this be disabled?
-    */
-   if (1)
-      lookup |= IZ_EARLY_DEPTH_TEST_BIT;
-
-   
    line_aa = AA_NEVER;
 
    /* _NEW_LINE, _NEW_POLYGON, BRW_NEW_REDUCED_PRIMITIVE */
-   if (brw->attribs.Line->SmoothFlag) {
+   if (ctx->Line.SmoothFlag) {
       if (brw->intel.reduced_primitive == GL_LINES) {
 	 line_aa = AA_ALWAYS;
       }
       else if (brw->intel.reduced_primitive == GL_TRIANGLES) {
-	 if (brw->attribs.Polygon->FrontMode == GL_LINE) {
+	 if (ctx->Polygon.FrontMode == GL_LINE) {
 	    line_aa = AA_SOMETIMES;
 
-	    if (brw->attribs.Polygon->BackMode == GL_LINE ||
-		(brw->attribs.Polygon->CullFlag &&
-		 brw->attribs.Polygon->CullFaceMode == GL_BACK))
+	    if (ctx->Polygon.BackMode == GL_LINE ||
+		(ctx->Polygon.CullFlag &&
+		 ctx->Polygon.CullFaceMode == GL_BACK))
 	       line_aa = AA_ALWAYS;
 	 }
-	 else if (brw->attribs.Polygon->BackMode == GL_LINE) {
+	 else if (ctx->Polygon.BackMode == GL_LINE) {
 	    line_aa = AA_SOMETIMES;
 
-	    if ((brw->attribs.Polygon->CullFlag &&
-		 brw->attribs.Polygon->CullFaceMode == GL_FRONT))
+	    if ((ctx->Polygon.CullFlag &&
+		 ctx->Polygon.CullFaceMode == GL_FRONT))
 	       line_aa = AA_ALWAYS;
 	 }
       }
@@ -271,20 +229,25 @@ static void brw_wm_populate_key( struct brw_context *brw,
    key->projtex_mask = brw->wm.input_size_masks[4-1] >> (FRAG_ATTRIB_TEX0 - FRAG_ATTRIB_WPOS); 
 
    /* _NEW_LIGHT */
-   key->flat_shade = (brw->attribs.Light->ShadeModel == GL_FLAT);
+   key->flat_shade = (ctx->Light.ShadeModel == GL_FLAT);
 
    /* _NEW_TEXTURE */
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      const struct gl_texture_unit *unit = &brw->attribs.Texture->Unit[i];
-      const struct gl_texture_object *t = unit->_Current;
+      const struct gl_texture_unit *unit = &ctx->Texture.Unit[i];
 
       if (unit->_ReallyEnabled) {
-	 if (t->Image[0][t->BaseLevel]->InternalFormat == GL_YCBCR_MESA) {
-	    key->yuvtex_mask |= 1<<i;
-	    if (t->Image[0][t->BaseLevel]->TexFormat->MesaFormat == 
-		    MESA_FORMAT_YCBCR)
-		key->yuvtex_swap_mask |= 1<< i;
+         const struct gl_texture_object *t = unit->_Current;
+         const struct gl_texture_image *img = t->Image[0][t->BaseLevel];
+	 if (img->InternalFormat == GL_YCBCR_MESA) {
+	    key->yuvtex_mask |= 1 << i;
+	    if (img->TexFormat->MesaFormat == MESA_FORMAT_YCBCR)
+		key->yuvtex_swap_mask |= 1 << i;
 	 }
+
+         key->tex_swizzles[i] = t->_Swizzle;
+      }
+      else {
+         key->tex_swizzles[i] = SWIZZLE_NOOP;
       }
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index 896390c17b8..0f46a25b1a1 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -49,8 +49,7 @@
 #define IZ_DEPTH_TEST_ENABLE_BIT    0x8
 #define IZ_STENCIL_WRITE_ENABLE_BIT 0x10
 #define IZ_STENCIL_TEST_ENABLE_BIT  0x20
-#define IZ_EARLY_DEPTH_TEST_BIT     0x40
-#define IZ_BIT_MAX                  0x80
+#define IZ_BIT_MAX                  0x40
 
 #define AA_NEVER     0
 #define AA_SOMETIMES 1
@@ -61,16 +60,17 @@ struct brw_wm_prog_key {
    GLuint aa_dest_stencil_reg:3;
    GLuint dest_depth_reg:3;
    GLuint nr_depth_regs:3;
-   GLuint projtex_mask:8;
-   GLuint shadowtex_mask:8;
    GLuint computes_depth:1;	/* could be derived from program string */
    GLuint source_depth_to_render_target:1;
    GLuint flat_shade:1;
    GLuint runtime_check_aads_emit:1;
    
-   GLuint yuvtex_mask:8;
-   GLuint yuvtex_swap_mask:8;	/* UV swaped */
-   GLuint pad1:16;
+   GLuint projtex_mask:16;
+   GLuint shadowtex_mask:16;
+   GLuint yuvtex_mask:16;
+   GLuint yuvtex_swap_mask:16;	/* UV swaped */
+
+   GLuint tex_swizzles[BRW_MAX_TEX_UNIT];
 
    GLuint program_string_id:32;
    GLuint origin_x, origin_y;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_debug.c b/src/mesa/drivers/dri/i965/brw_wm_debug.c
index f31d0974ecf..8f07f89ebc5 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_debug.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_debug.c
@@ -163,9 +163,9 @@ void brw_wm_print_program( struct brw_wm_compile *c,
 {
    GLuint insn;
 
-   _mesa_printf("\n\n\n%s:\n", stage);
+   _mesa_printf("%s:\n", stage);
    for (insn = 0; insn < c->nr_insns; insn++)
       brw_wm_print_insn(c, &c->instruction[insn]);
-   _mesa_printf("\n\n\n");
+   _mesa_printf("\n");
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index 58c78c4b2c5..b5050a3e40b 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -194,7 +194,7 @@ static void emit_linterp( struct brw_compile *p,
    interp[2] = brw_vec1_grf(nr+1, 0);
    interp[3] = brw_vec1_grf(nr+1, 4);
 
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
 	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
@@ -219,42 +219,40 @@ static void emit_pinterp( struct brw_compile *p,
    interp[2] = brw_vec1_grf(nr+1, 0);
    interp[3] = brw_vec1_grf(nr+1, 4);
 
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
 	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
       }
    }
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_MUL(p, dst[i], dst[i], w[3]);
       }
    }
 }
 
+
 static void emit_cinterp( struct brw_compile *p, 
 			 const struct brw_reg *dst,
 			 GLuint mask,
 			 const struct brw_reg *arg0 )
 {
-	struct brw_reg interp[4];
-	GLuint nr = arg0[0].nr;
-	GLuint i;
-
-	interp[0] = brw_vec1_grf(nr, 0);
-	interp[1] = brw_vec1_grf(nr, 4);
-	interp[2] = brw_vec1_grf(nr+1, 0);
-	interp[3] = brw_vec1_grf(nr+1, 4);
-
-	for(i = 0; i < 4; i++ ) {
-		if (mask & (1<<i)) {
-			brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
-		}
-	}
-}
-
+   struct brw_reg interp[4];
+   GLuint nr = arg0[0].nr;
+   GLuint i;
 
+   interp[0] = brw_vec1_grf(nr, 0);
+   interp[1] = brw_vec1_grf(nr, 4);
+   interp[2] = brw_vec1_grf(nr+1, 0);
+   interp[3] = brw_vec1_grf(nr+1, 4);
 
+   for (i = 0; i < 4; i++) {
+      if (mask & (1<<i)) {
+         brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
+      }
+   }
+}
 
 
 static void emit_alu1( struct brw_compile *p, 
@@ -280,6 +278,7 @@ static void emit_alu1( struct brw_compile *p,
       brw_set_saturate(p, 0);
 }
 
+
 static void emit_alu2( struct brw_compile *p, 
 		       struct brw_instruction *(*func)(struct brw_compile *, 
 						       struct brw_reg, 
@@ -351,6 +350,7 @@ static void emit_lrp( struct brw_compile *p,
       }
    }
 }
+
 static void emit_sop( struct brw_compile *p, 
 		      const struct brw_reg *dst,
 		      GLuint mask,
@@ -376,7 +376,7 @@ static void emit_slt( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
 }
 
 static void emit_sle( struct brw_compile *p, 
@@ -385,7 +385,7 @@ static void emit_sle( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
 }
 
 static void emit_sgt( struct brw_compile *p, 
@@ -394,7 +394,7 @@ static void emit_sgt( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
 }
 
 static void emit_sge( struct brw_compile *p, 
@@ -403,7 +403,7 @@ static void emit_sge( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
 }
 
 static void emit_seq( struct brw_compile *p, 
@@ -412,7 +412,7 @@ static void emit_seq( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
 }
 
 static void emit_sne( struct brw_compile *p, 
@@ -421,7 +421,7 @@ static void emit_sne( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
 }
 
 static void emit_cmp( struct brw_compile *p, 
@@ -505,7 +505,7 @@ static void emit_dp3( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -525,7 +525,7 @@ static void emit_dp4( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -546,7 +546,7 @@ static void emit_dph( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -592,7 +592,7 @@ static void emit_math1( struct brw_compile *p,
 			const struct brw_reg *arg0 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    //assert((mask & WRITEMASK_XYZW) == WRITEMASK_X ||
    //	  function == BRW_MATH_FUNCTION_SINCOS);
@@ -619,7 +619,7 @@ static void emit_math2( struct brw_compile *p,
 			const struct brw_reg *arg1)
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -760,7 +760,6 @@ static void emit_txb( struct brw_wm_compile *c,
    brw_MOV(p, brw_message_reg(8), arg[3]);
    msgLength = 9;
 
-
    brw_SAMPLE(p, 
 	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
 	      1,
@@ -772,7 +771,6 @@ static void emit_txb( struct brw_wm_compile *c,
 	      8,		/* responseLength */
 	      msgLength,
 	      0);	
-
 }
 
 
@@ -823,7 +821,6 @@ static void emit_kil( struct brw_wm_compile *c,
    struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
    GLuint i;
    
-
    /* XXX - usually won't need 4 compares!
     */
    for (i = 0; i < 4; i++) {
@@ -836,6 +833,7 @@ static void emit_kil( struct brw_wm_compile *c,
    }
 }
 
+
 static void fire_fb_write( struct brw_wm_compile *c,
 			   GLuint base_reg,
 			   GLuint nr,
@@ -869,6 +867,7 @@ static void fire_fb_write( struct brw_wm_compile *c,
 		eot);
 }
 
+
 static void emit_aa( struct brw_wm_compile *c,
 		     struct brw_reg *arg1,
 		     GLuint reg )
@@ -962,7 +961,6 @@ static void emit_fb_write( struct brw_wm_compile *c,
       nr += 2;
    }
 
-
    if (!c->key.runtime_check_aads_emit) {
       if (c->key.aa_dest_stencil_reg)
 	 emit_aa(c, arg1, 2);
@@ -996,8 +994,6 @@ static void emit_fb_write( struct brw_wm_compile *c,
 }
 
 
-
-
 /* Post-fragment-program processing.  Send the results to the
  * framebuffer.
  */
@@ -1022,6 +1018,7 @@ static void emit_spill( struct brw_wm_compile *c,
 		   slot);
 }
 
+
 static void emit_unspill( struct brw_wm_compile *c,
 			  struct brw_reg reg,
 			  GLuint slot )
@@ -1047,7 +1044,6 @@ static void emit_unspill( struct brw_wm_compile *c,
 }
 
 
-
 /**
  * Retrieve upto 4 GEN4 register pairs for the given wm reg:
  */
@@ -1073,6 +1069,7 @@ static void get_argument_regs( struct brw_wm_compile *c,
    }
 }
 
+
 static void spill_values( struct brw_wm_compile *c,
 			  struct brw_wm_value *values,
 			  GLuint nr )
@@ -1085,7 +1082,6 @@ static void spill_values( struct brw_wm_compile *c,
 }
 
 
-
 /* Emit the fragment program instructions here.
  */
 void brw_wm_emit( struct brw_wm_compile *c )
@@ -1176,7 +1172,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]);
 	 break;
 
-      case OPCODE_DP3:	/*  */
+      case OPCODE_DP3:
 	 emit_dp3(p, dst, dst_flags, args[0], args[1]);
 	 break;
 
@@ -1188,7 +1184,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_dph(p, dst, dst_flags, args[0], args[1]);
 	 break;
 
-      case OPCODE_LRP:	/*  */
+      case OPCODE_LRP:
 	 emit_lrp(p, dst, dst_flags, args[0], args[1], args[2]);
 	 break;
 
@@ -1315,8 +1311,3 @@ void brw_wm_emit( struct brw_wm_compile *c )
 		      inst->dst[i]->spill_slot);
    }
 }
-
-
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index d3e926c328c..ea3f3fc678f 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -111,6 +111,12 @@ static struct prog_src_register src_swizzle1( struct prog_src_register reg, int
    return src_swizzle(reg, x, x, x, x);
 }
 
+static struct prog_src_register src_swizzle4( struct prog_src_register reg, uint swizzle )
+{
+   reg.Swizzle = swizzle;
+   return reg;
+}
+
 
 /***********************************************************************
  * Dest regs
@@ -122,10 +128,11 @@ static struct prog_dst_register dst_reg(GLuint file, GLuint idx)
    reg.File = file;
    reg.Index = idx;
    reg.WriteMask = WRITEMASK_XYZW;
+   reg.RelAddr = 0;
    reg.CondMask = 0;
    reg.CondSwizzle = 0;
-   reg.pad = 0;
    reg.CondSrc = 0;
+   reg.pad = 0;
    return reg;
 }
 
@@ -426,10 +433,6 @@ static struct prog_src_register search_or_add_param5(struct brw_wm_compile *c,
 
    idx = _mesa_add_state_reference( paramList, tokens );
 
-   /* Recalculate state dependency: 
-    */
-   c->fp->param_state = paramList->StateFlags;
-
    return src_reg(PROGRAM_STATE_VAR, idx);
 }
 
@@ -557,12 +560,19 @@ static void precalc_lit( struct brw_wm_compile *c,
    }
 }
 
+
+/**
+ * Some TEX instructions require extra code, cube map coordinate
+ * normalization, or coordinate scaling for RECT textures, etc.
+ * This function emits those extra instructions and the TEX
+ * instruction itself.
+ */
 static void precalc_tex( struct brw_wm_compile *c,
 			 const struct prog_instruction *inst )
 {
    struct prog_src_register coord;
    struct prog_dst_register tmpcoord;
-   GLuint unit = c->fp->program.Base.SamplerUnits[inst->TexSrcUnit];
+   const GLuint unit = c->fp->program.Base.SamplerUnits[inst->TexSrcUnit];
 
    if (inst->TexSrcTarget == TEXTURE_CUBE_INDEX) {
        struct prog_instruction *out;
@@ -572,9 +582,11 @@ static void precalc_tex( struct brw_wm_compile *c,
        struct prog_src_register tmp1src = src_reg_from_dst(tmp1);
        struct prog_src_register src0 = inst->SrcReg[0];
 
+       /* find longest component of coord vector and normalize it */
        tmpcoord = get_temp(c);
        coord = src_reg_from_dst(tmpcoord);
 
+       /* tmpcoord = src0 (i.e.: coord = src0) */
        out = emit_op(c, OPCODE_MOV,
                      tmpcoord,
                      0, 0, 0,
@@ -584,6 +596,7 @@ static void precalc_tex( struct brw_wm_compile *c,
        out->SrcReg[0].NegateBase = 0;
        out->SrcReg[0].Abs = 1;
 
+       /* tmp0 = MAX(coord.X, coord.Y) */
        emit_op(c, OPCODE_MAX,
                tmp0,
                0, 0, 0,
@@ -591,6 +604,7 @@ static void precalc_tex( struct brw_wm_compile *c,
                src_swizzle1(coord, Y),
                src_undef());
 
+       /* tmp1 = MAX(tmp0, coord.Z) */
        emit_op(c, OPCODE_MAX,
                tmp1,
                0, 0, 0,
@@ -598,6 +612,7 @@ static void precalc_tex( struct brw_wm_compile *c,
                src_swizzle1(coord, Z),
                src_undef());
 
+       /* tmp0 = 1 / tmp1 */
        emit_op(c, OPCODE_RCP,
                tmp0,
                0, 0, 0,
@@ -605,6 +620,7 @@ static void precalc_tex( struct brw_wm_compile *c,
                src_undef(),
                src_undef());
 
+       /* tmpCoord = src0 * tmp0 */
        emit_op(c, OPCODE_MUL,
                tmpcoord,
                0, 0, 0,
@@ -614,7 +630,8 @@ static void precalc_tex( struct brw_wm_compile *c,
 
        release_temp(c, tmp0);
        release_temp(c, tmp1);
-   } else if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
+   }
+   else if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
       struct prog_src_register scale = 
 	 search_or_add_param5( c, 
 			       STATE_INTERNAL, 
@@ -645,19 +662,9 @@ static void precalc_tex( struct brw_wm_compile *c,
     * conversion requires allocating a temporary variable which we
     * don't have the facility to do that late in the compilation.
     */
-   if (!(c->key.yuvtex_mask & (1<<unit))) {
-      emit_op(c, 
-	      OPCODE_TEX,
-	      inst->DstReg,
-	      inst->SaturateMode,
-	      unit,
-	      inst->TexSrcTarget,
-	      coord,
-	      src_undef(),
-	      src_undef());
-   }
-   else {
-       GLboolean  swap_uv = c->key.yuvtex_swap_mask & (1<<unit);
+   if (c->key.yuvtex_mask & (1 << unit)) {
+      /* convert ycbcr to RGBA */
+      GLboolean  swap_uv = c->key.yuvtex_swap_mask & (1<<unit);
 
       /* 
 	 CONST C0 = { -.5, -.0625,  -.5, 1.164 }
@@ -737,6 +744,31 @@ static void precalc_tex( struct brw_wm_compile *c,
 
       release_temp(c, tmp);
    }
+   else {
+      /* ordinary RGBA tex instruction */
+      emit_op(c, 
+	      OPCODE_TEX,
+	      inst->DstReg,
+	      inst->SaturateMode,
+	      unit,
+	      inst->TexSrcTarget,
+	      coord,
+	      src_undef(),
+	      src_undef());
+   }
+
+   /* For GL_EXT_texture_swizzle: */
+   if (c->key.tex_swizzles[unit] != SWIZZLE_NOOP) {
+      /* swizzle the result of the TEX instruction */
+      struct prog_src_register tmpsrc = src_reg_from_dst(inst->DstReg);
+      emit_op(c, OPCODE_SWZ,
+              inst->DstReg,
+              SATURATE_OFF, /* saturate already done above */
+              0, 0,   /* tex unit, target N/A */
+              src_swizzle4(tmpsrc, c->key.tex_swizzles[unit]),
+              src_undef(),
+              src_undef());
+   }
 
    if ((inst->TexSrcTarget == TEXTURE_RECT_INDEX) ||
        (inst->TexSrcTarget == TEXTURE_CUBE_INDEX))
@@ -814,62 +846,11 @@ static void precalc_txp( struct brw_wm_compile *c,
 
 
 
-
-
-/***********************************************************************
- * Add instructions to perform fog blending
- */
-
-static void fog_blend( struct brw_wm_compile *c,
-			     struct prog_src_register fog_factor )
-{
-   struct prog_dst_register outcolor = dst_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
-   struct prog_src_register fogcolor = search_or_add_param5( c, STATE_FOG_COLOR, 0,0,0,0 );
-
-   /* color.xyz = LRP fog_factor.xxxx, output_color, fog_color */
-   
-   emit_op(c, 
-	   OPCODE_LRP,
-	   dst_mask(outcolor, WRITEMASK_XYZ),
-	   0, 0, 0,
-	   fog_factor,
-	   src_reg_from_dst(outcolor),
-	   fogcolor);
-}
-
-
-
-/* This one is simple - just take the interpolated fog coordinate and
- * use it as the fog blend factor.
- */
-static void fog_interpolated( struct brw_wm_compile *c )
-{
-   struct prog_src_register fogc = src_reg(PROGRAM_INPUT, FRAG_ATTRIB_FOGC);
-   
-   if (!(c->fp_interp_emitted & (1<<FRAG_ATTRIB_FOGC))) 
-      emit_interp(c, FRAG_ATTRIB_FOGC);
-
-   fog_blend( c, src_swizzle1(fogc, GET_SWZ(fogc.Swizzle,X)));
-}
-
-static void emit_fog( struct brw_wm_compile *c ) 
-{
-   if (!c->fp->program.FogOption)
-      return;
-
-   if (1) 
-      fog_interpolated( c );
-   else {
-      /* TODO: per-pixel fog */
-      assert(0);
-   }
-}
-
 static void emit_fb_write( struct brw_wm_compile *c )
 {
-   struct prog_src_register outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
    struct prog_src_register payload_r0_depth = src_reg(PROGRAM_PAYLOAD, PAYLOAD_DEPTH);
    struct prog_src_register outdepth = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DEPR);
+   struct prog_src_register outcolor;
    GLuint i;
 
    struct prog_instruction *inst, *last_inst;
@@ -893,7 +874,14 @@ static void emit_fb_write( struct brw_wm_compile *c )
 	   }
        }
        last_inst->Sampler |= 1; //eot
-   }else {
+   }
+   else {
+      /* if gl_FragData[0] is written, use it, else use gl_FragColor */
+      if (c->fp->program.Base.OutputsWritten & (1 << FRAG_RESULT_DATA0))
+         outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_DATA0);
+      else 
+         outcolor = src_reg(PROGRAM_OUTPUT, FRAG_RESULT_COLR);
+
        inst = emit_op(c, WM_FB_WRITE, dst_mask(dst_undef(),0),
 	       0, 0, 0, outcolor, payload_r0_depth, outdepth);
        inst->Sampler = 1|(0<<1);
@@ -960,7 +948,7 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
    GLuint insn;
 
    if (INTEL_DEBUG & DEBUG_WM) {
-      _mesa_printf("\n\n\npre-fp:\n");
+      _mesa_printf("pre-fp:\n");
       _mesa_print_program(&fp->program.Base); 
       _mesa_printf("\n");
    }
@@ -1055,7 +1043,6 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
          emit_ddy(c, inst);
 	break;
       case OPCODE_END:
-	 emit_fog(c);
 	 emit_fb_write(c);
 	 break;
       case OPCODE_PRINT:
@@ -1068,7 +1055,7 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
    }
 
    if (INTEL_DEBUG & DEBUG_WM) {
-	   _mesa_printf("\n\n\npass_fp:\n");
+	   _mesa_printf("pass_fp:\n");
 	   print_insns( c->prog_instructions, c->nr_fp_insns );
 	   _mesa_printf("\n");
    }
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index cb728190f5c..8fd776ac393 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -267,7 +267,7 @@ static void emit_trunc( struct brw_wm_compile *c,
 	    struct brw_reg src, dst;
 	    dst = get_dst_reg(c, inst, i, 1) ;
 	    src = get_src_reg(c, &inst->SrcReg[0], i, 1);
-	    brw_RNDD(p, dst, src);
+	    brw_RNDZ(p, dst, src);
 	}
     }
     brw_set_saturate(p, 0);
@@ -623,7 +623,7 @@ static void emit_dph(struct brw_wm_compile *c,
     brw_MAC(p, brw_null_reg(), src0[1], src1[1]);
     brw_MAC(p, dst, src0[2], src1[2]);
     brw_set_saturate(p, (inst->SaturateMode != SATURATE_OFF) ? 1 : 0);
-    brw_ADD(p, dst, src0[3], src1[3]);
+    brw_ADD(p, dst, dst, src1[3]);
     brw_set_saturate(p, 0);
 }
 
@@ -892,15 +892,19 @@ static void emit_lrp(struct brw_wm_compile *c,
     }
 }
 
+/**
+ * For GLSL shaders, this KIL will be unconditional.
+ * It may be contained inside an IF/ENDIF structure of course.
+ */
 static void emit_kil(struct brw_wm_compile *c)
 {
-	struct brw_compile *p = &c->func;
-	struct brw_reg depth = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
-	brw_push_insn_state(p);
-	brw_set_mask_control(p, BRW_MASK_DISABLE);
-	brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
-	brw_AND(p, depth, c->emit_mask_reg, depth);
-	brw_pop_insn_state(p);
+    struct brw_compile *p = &c->func;
+    struct brw_reg depth = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
+    brw_push_insn_state(p);
+    brw_set_mask_control(p, BRW_MASK_DISABLE);
+    brw_NOT(p, c->emit_mask_reg, brw_mask_reg(1)); //IMASK
+    brw_AND(p, depth, c->emit_mask_reg, depth);
+    brw_pop_insn_state(p);
 }
 
 static void emit_mad(struct brw_wm_compile *c,
@@ -1095,7 +1099,7 @@ static void noise1_sub( struct brw_wm_compile *c ) {
     /* Arrange the two end coordinates into scalars (itmp0/itmp1) to
        be hashed.  Also compute the remainder (offset within the unit
        length), interleaved to reduce register dependency penalties. */
-    brw_RNDD( p, itmp[ 0 ], param );
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param );
     brw_FRC( p, param, param );
     brw_ADD( p, itmp[ 1 ], itmp[ 0 ], brw_imm_ud( 1 ) );
     brw_MOV( p, itmp[ 3 ], brw_imm_ud( 0x79D9 ) ); /* constant used later */
@@ -1220,8 +1224,8 @@ static void noise2_sub( struct brw_wm_compile *c ) {
     /* Arrange the four corner coordinates into scalars (itmp0..itmp3) to
        be hashed.  Also compute the remainders (offsets within the unit
        square), interleaved to reduce register dependency penalties. */
-    brw_RNDD( p, itmp[ 0 ], param0 );
-    brw_RNDD( p, itmp[ 1 ], param1 );
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param0 );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param1 );
     brw_FRC( p, param0, param0 );
     brw_FRC( p, param1, param1 );
     brw_MOV( p, itmp[ 4 ], brw_imm_ud( 0xBA97 ) ); /* constant used later */
@@ -1400,21 +1404,19 @@ static void noise3_sub( struct brw_wm_compile *c ) {
     /* Arrange the eight corner coordinates into scalars (itmp0..itmp3) to
        be hashed.  Also compute the remainders (offsets within the unit
        cube), interleaved to reduce register dependency penalties. */
-    brw_RNDD( p, itmp[ 0 ], param0 );
-    brw_RNDD( p, itmp[ 1 ], param1 );
-    brw_RNDD( p, itmp[ 2 ], param2 );
-    brw_MOV( p, itmp[ 4 ], brw_imm_ud( 0xBC8F ) ); /* constant used later */
-    brw_MOV( p, itmp[ 5 ], brw_imm_ud( 0xD0BD ) ); /* constant used later */
-    brw_MOV( p, itmp[ 6 ], brw_imm_ud( 0x9B93 ) ); /* constant used later */
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param0 );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param1 );
+    brw_RNDD( p, retype( itmp[ 2 ], BRW_REGISTER_TYPE_D ), param2 );
     brw_FRC( p, param0, param0 );
     brw_FRC( p, param1, param1 );
     brw_FRC( p, param2, param2 );
     /* Since we now have only 16 bits of precision in the hash, we must
        be more careful about thorough mixing to maintain entropy as we
        squash the input vector into a small scalar. */
-    brw_MUL( p, brw_acc_reg(), itmp[ 4 ], itmp[ 0 ] );
-    brw_MAC( p, brw_acc_reg(), itmp[ 5 ], itmp[ 1 ] );
-    brw_MAC( p, itmp[ 0 ], itmp[ 6 ], itmp[ 2 ] );
+    brw_MUL( p, brw_null_reg(), low_words( itmp[ 0 ] ), brw_imm_uw( 0xBC8F ) );
+    brw_MAC( p, brw_null_reg(), low_words( itmp[ 1 ] ), brw_imm_uw( 0xD0BD ) );
+    brw_MAC( p, low_words( itmp[ 0 ] ), low_words( itmp[ 2 ] ),
+	     brw_imm_uw( 0x9B93 ) );
     brw_ADD( p, high_words( itmp[ 0 ] ), low_words( itmp[ 0 ] ),
 	     brw_imm_uw( 0xBC8F ) );
 
@@ -1668,6 +1670,430 @@ static void emit_noise3( struct brw_wm_compile *c,
     release_tmps( c, mark );
 }
     
+/* For the four-dimensional case, the little micro-optimisation benefits
+   we obtain by unrolling all the loops aren't worth the massive bloat it
+   now causes.  Instead, we loop twice around performing a similar operation
+   to noise3, once for the w=0 cube and once for the w=1, with a bit more
+   code to glue it all together. */
+static void noise4_sub( struct brw_wm_compile *c ) {
+
+    struct brw_compile *p = &c->func;
+    struct brw_reg param[ 4 ],
+	x0y0, x0y1, x1y0, x1y1, /* gradients at four of the corners */
+	w0, /* noise for the w=0 cube */
+	floors[ 2 ], /* integer coordinates of base corner of hypercube */
+	interp[ 4 ], /* interpolation coefficients */
+	t, tmp[ 8 ], /* float temporaries */
+	itmp[ 8 ], /* unsigned integer temporaries (aliases of floats above) */
+	wtmp[ 8 ]; /* 16-way unsigned word temporaries (aliases of above) */
+    int i, j;
+    int mark = mark_tmps( c );
+    GLuint loop, origin;
+    
+    x0y0 = alloc_tmp( c );
+    x0y1 = alloc_tmp( c );
+    x1y0 = alloc_tmp( c );
+    x1y1 = alloc_tmp( c );
+    t = alloc_tmp( c );
+    w0 = alloc_tmp( c );    
+    floors[ 0 ] = retype( alloc_tmp( c ), BRW_REGISTER_TYPE_UD );
+    floors[ 1 ] = retype( alloc_tmp( c ), BRW_REGISTER_TYPE_UD );
+
+    for( i = 0; i < 4; i++ ) {
+	param[ i ] = lookup_tmp( c, mark - 5 + i );
+	interp[ i ] = alloc_tmp( c );
+    }
+    
+    for( i = 0; i < 8; i++ ) {
+	tmp[ i ] = alloc_tmp( c );
+	itmp[ i ] = retype( tmp[ i ], BRW_REGISTER_TYPE_UD );
+	wtmp[ i ] = brw_uw16_grf( tmp[ i ].nr, 0 );
+    }
+
+    brw_set_access_mode( p, BRW_ALIGN_1 );
+
+    /* We only want 16 bits of precision from the integral part of each
+       co-ordinate, but unfortunately the RNDD semantics would saturate
+       at 16 bits if we performed the operation directly to a 16-bit
+       destination.  Therefore, we round to 32-bit temporaries where
+       appropriate, and then store only the lower 16 bits. */
+    brw_RNDD( p, retype( floors[ 0 ], BRW_REGISTER_TYPE_D ), param[ 0 ] );
+    brw_RNDD( p, retype( itmp[ 0 ], BRW_REGISTER_TYPE_D ), param[ 1 ] );
+    brw_RNDD( p, retype( floors[ 1 ], BRW_REGISTER_TYPE_D ), param[ 2 ] );
+    brw_RNDD( p, retype( itmp[ 1 ], BRW_REGISTER_TYPE_D ), param[ 3 ] );
+    brw_MOV( p, high_words( floors[ 0 ] ), low_words( itmp[ 0 ] ) );
+    brw_MOV( p, high_words( floors[ 1 ] ), low_words( itmp[ 1 ] ) );
+
+    /* Modify the flag register here, because the side effect is useful
+       later (see below).  We know for certain that all flags will be
+       cleared, since the FRC instruction cannot possibly generate
+       negative results.  Even for exceptional inputs (infinities, denormals,
+       NaNs), the architecture guarantees that the L conditional is false. */
+    brw_set_conditionalmod( p, BRW_CONDITIONAL_L );
+    brw_FRC( p, param[ 0 ], param[ 0 ] );
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    for( i = 1; i < 4; i++ )	
+	brw_FRC( p, param[ i ], param[ i ] );
+    
+    /* Calculate the interpolation coefficients (6t^5 - 15t^4 + 10t^3) first
+       of all. */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, interp[ i ], param[ i ], brw_imm_f( 6.0 ) );
+    for( i = 0; i < 4; i++ )
+	brw_ADD( p, interp[ i ], interp[ i ], brw_imm_f( -15.0 ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, interp[ i ], interp[ i ], param[ i ] );
+    for( i = 0; i < 4; i++ )
+	brw_ADD( p, interp[ i ], interp[ i ], brw_imm_f( 10.0 ) );
+    for( j = 0; j < 3; j++ )
+	for( i = 0; i < 4; i++ )
+	    brw_MUL( p, interp[ i ], interp[ i ], param[ i ] );
+
+    /* Mark the current address, as it will be a jump destination.  The
+       following code will be executed twice: first, with the flag
+       register clear indicating the w=0 case, and second with flags
+       set for w=1. */
+    loop = p->nr_insn;
+    
+    /* Arrange the eight corner coordinates into scalars (itmp0..itmp3) to
+       be hashed.  Since we have only 16 bits of precision in the hash, we
+       must be careful about thorough mixing to maintain entropy as we
+       squash the input vector into a small scalar. */
+    brw_MUL( p, brw_null_reg(), low_words( floors[ 0 ] ),
+	     brw_imm_uw( 0xBC8F ) );
+    brw_MAC( p, brw_null_reg(), high_words( floors[ 0 ] ),
+	     brw_imm_uw( 0xD0BD ) );
+    brw_MAC( p, brw_null_reg(), low_words( floors[ 1 ] ),
+	     brw_imm_uw( 0x9B93 ) );
+    brw_MAC( p, low_words( itmp[ 0 ] ), high_words( floors[ 1 ] ),
+	     brw_imm_uw( 0xA359 ) );
+    brw_ADD( p, high_words( itmp[ 0 ] ), low_words( itmp[ 0 ] ),
+	     brw_imm_uw( 0xBC8F ) );
+
+    /* Temporarily disable the execution mask while we work with ExecSize=16
+       channels (the mask is set for ExecSize=8 and is probably incorrect).
+       Although this might cause execution of unwanted channels, the code
+       writes only to temporary registers and has no side effects, so
+       disabling the mask is harmless. */
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_ADD( p, wtmp[ 1 ], wtmp[ 0 ], brw_imm_uw( 0xD0BD ) );
+    brw_ADD( p, wtmp[ 2 ], wtmp[ 0 ], brw_imm_uw( 0x9B93 ) );
+    brw_ADD( p, wtmp[ 3 ], wtmp[ 1 ], brw_imm_uw( 0x9B93 ) );
+
+    /* We're now ready to perform the hashing.  The eight hashes are
+       interleaved for performance.  The hash function used is
+       designed to rapidly achieve avalanche and require only 16x16
+       bit multiplication, and 8-bit swizzles (which we get for
+       free). */
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0x28D9 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    for( i = 0; i < 4; i++ )
+	brw_MUL( p, wtmp[ i ], wtmp[ i ], brw_imm_uw( 0xC6D5 ) );
+    for( i = 0; i < 4; i++ )
+	brw_XOR( p, even_bytes( wtmp[ i ] ), even_bytes( wtmp[ i ] ),
+		 odd_bytes( wtmp[ i ] ) );
+    brw_pop_insn_state( p );
+
+    /* Now we want to initialise the four rear gradients based on the
+       hashes.  Format conversion from signed integer to float leaves
+       everything scaled too high by a factor of pow( 2, 15 ), but
+       we correct for that right at the end. */
+    /* x component */
+    brw_ADD( p, t, param[ 0 ], brw_imm_f( -1.0 ) );
+    brw_MOV( p, x0y0, low_words( tmp[ 0 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 1 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 0 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 1 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+    
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param[ 1 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param[ 0 ] );
+    brw_MUL( p, x0y1, x0y1, param[ 0 ] );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );    
+    /* prepare t for the w component (used below): w the first time through
+       the loop; w - 1 the second time) */
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_ADD( p, t, param[ 3 ], brw_imm_f( -1.0 ) );
+    p->current->header.predicate_inverse = 1;
+    brw_MOV( p, t, param[ 3 ] );
+    p->current->header.predicate_inverse = 0;
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 1 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 1 ] );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 0 ], wtmp[ 0 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 1 ], wtmp[ 1 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 2 ] );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], param[ 2 ] );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* w component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 1 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 0 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 1 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param[ 0 ], brw_imm_f( -1.0 ) );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* Here we interpolate in the y dimension... */
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) );
+    brw_MUL( p, x0y1, x0y1, interp[ 1 ] );
+    brw_MUL( p, x1y1, x1y1, interp[ 1 ] );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  Leave the result in tmp[ 0 ] (see below)... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, interp[ 0 ] );
+    brw_ADD( p, tmp[ 0 ], x0y0, x1y0 );
+
+    /* Now do the same thing for the front four gradients... */
+    /* x component */
+    brw_MOV( p, x0y0, low_words( tmp[ 2 ] ) );
+    brw_MOV( p, x0y1, low_words( tmp[ 3 ] ) );
+    brw_MOV( p, x1y0, high_words( tmp[ 2 ] ) );
+    brw_MOV( p, x1y1, high_words( tmp[ 3 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, x1y0, x1y0, t );
+    brw_MUL( p, x1y1, x1y1, t );
+    brw_ADD( p, t, param[ 1 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, x0y0, x0y0, param[ 0 ] );
+    brw_MUL( p, x0y1, x0y1, param[ 0 ] );
+
+    /* y component */
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    brw_ADD( p, t, param[ 2 ], brw_imm_f( -1.0 ) );
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], param[ 1 ] );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], param[ 1 ] );
+    
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    
+    /* z component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_SHL( p, wtmp[ 2 ], wtmp[ 2 ], brw_imm_uw( 4 ) );
+    brw_SHL( p, wtmp[ 3 ], wtmp[ 3 ], brw_imm_uw( 4 ) );
+    brw_pop_insn_state( p );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    /* prepare t for the w component (used below): w the first time through
+       the loop; w - 1 the second time) */
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_ADD( p, t, param[ 3 ], brw_imm_f( -1.0 ) );
+    p->current->header.predicate_inverse = 1;
+    brw_MOV( p, t, param[ 3 ] );
+    p->current->header.predicate_inverse = 0;
+    brw_set_predicate_control( p, BRW_PREDICATE_NONE );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* w component */
+    brw_MOV( p, tmp[ 4 ], low_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 5 ], low_words( tmp[ 3 ] ) );
+    brw_MOV( p, tmp[ 6 ], high_words( tmp[ 2 ] ) );
+    brw_MOV( p, tmp[ 7 ], high_words( tmp[ 3 ] ) );
+
+    brw_MUL( p, tmp[ 4 ], tmp[ 4 ], t );
+    brw_MUL( p, tmp[ 5 ], tmp[ 5 ], t );
+    brw_MUL( p, tmp[ 6 ], tmp[ 6 ], t );
+    brw_MUL( p, tmp[ 7 ], tmp[ 7 ], t );
+    
+    brw_ADD( p, x0y0, x0y0, tmp[ 4 ] );
+    brw_ADD( p, x0y1, x0y1, tmp[ 5 ] );
+    brw_ADD( p, x1y0, x1y0, tmp[ 6 ] );
+    brw_ADD( p, x1y1, x1y1, tmp[ 7 ] );
+
+    /* Interpolate in the y dimension: */
+    brw_ADD( p, x0y1, x0y1, negate( x0y0 ) );
+    brw_ADD( p, x1y1, x1y1, negate( x1y0 ) );
+    brw_MUL( p, x0y1, x0y1, interp[ 1 ] );
+    brw_MUL( p, x1y1, x1y1, interp[ 1 ] );
+    brw_ADD( p, x0y0, x0y0, x0y1 );
+    brw_ADD( p, x1y0, x1y0, x1y1 );
+
+    /* And now in x.  The rear face is in tmp[ 0 ] (see above), so this
+       time put the front face in tmp[ 1 ] and we're nearly there... */
+    brw_ADD( p, x1y0, x1y0, negate( x0y0 ) );
+    brw_MUL( p, x1y0, x1y0, interp[ 0 ] );
+    brw_ADD( p, tmp[ 1 ], x0y0, x1y0 );
+
+    /* Another interpolation, in the z dimension: */
+    brw_ADD( p, tmp[ 1 ], tmp[ 1 ], negate( tmp[ 0 ] ) );    
+    brw_MUL( p, tmp[ 1 ], tmp[ 1 ], interp[ 2 ] );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], tmp[ 1 ] );
+
+    /* Exit the loop if we've computed both cubes... */
+    origin = p->nr_insn;
+    brw_push_insn_state( p );
+    brw_set_predicate_control( p, BRW_PREDICATE_NORMAL );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_ADD( p, brw_ip_reg(), brw_ip_reg(), brw_imm_d( 0 ) );
+    brw_pop_insn_state( p );
+
+    /* Save the result for the w=0 case, and increment the w coordinate: */
+    brw_MOV( p, w0, tmp[ 0 ] );
+    brw_ADD( p, high_words( floors[ 1 ] ), high_words( floors[ 1 ] ),
+	     brw_imm_uw( 1 ) );
+
+    /* Loop around for the other cube.  Explicitly set the flag register
+       (unfortunately we must spend an extra instruction to do this: we
+       can't rely on a side effect of the previous MOV or ADD because
+       conditional modifiers which are normally true might be false in
+       exceptional circumstances, e.g. given a NaN input; the add to
+       brw_ip_reg() is not suitable because the IP is not an 8-vector). */
+    brw_push_insn_state( p );
+    brw_set_mask_control( p, BRW_MASK_DISABLE );
+    brw_MOV( p, brw_flag_reg(), brw_imm_uw( 0xFF ) );
+    brw_ADD( p, brw_ip_reg(), brw_ip_reg(),
+	     brw_imm_d( ( loop - p->nr_insn ) << 4 ) );
+    brw_pop_insn_state( p );
+
+    /* Patch the previous conditional branch now that we know the
+       destination address. */
+    brw_set_src1( p->store + origin,
+		  brw_imm_d( ( p->nr_insn - origin ) << 4 ) );
+
+    /* The very last interpolation. */
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], negate( w0 ) );    
+    brw_MUL( p, tmp[ 0 ], tmp[ 0 ], interp[ 3 ] );
+    brw_ADD( p, tmp[ 0 ], tmp[ 0 ], w0 );
+
+    /* scale by pow( 2, -15 ), as described above */
+    brw_MUL( p, param[ 0 ], tmp[ 0 ], brw_imm_f( 0.000030517578125 ) );
+
+    release_tmps( c, mark );
+}
+
+static void emit_noise4( struct brw_wm_compile *c,
+			 struct prog_instruction *inst )
+{
+    struct brw_compile *p = &c->func;
+    struct brw_reg src0, src1, src2, src3, param0, param1, param2, param3, dst;
+    GLuint mask = inst->DstReg.WriteMask;
+    int i;
+    int mark = mark_tmps( c );
+
+    assert( mark == 0 );
+    
+    src0 = get_src_reg( c, inst->SrcReg, 0, 1 );
+    src1 = get_src_reg( c, inst->SrcReg, 1, 1 );
+    src2 = get_src_reg( c, inst->SrcReg, 2, 1 );
+    src3 = get_src_reg( c, inst->SrcReg, 3, 1 );
+
+    param0 = alloc_tmp( c );
+    param1 = alloc_tmp( c );
+    param2 = alloc_tmp( c );
+    param3 = alloc_tmp( c );
+
+    brw_MOV( p, param0, src0 );
+    brw_MOV( p, param1, src1 );
+    brw_MOV( p, param2, src2 );
+    brw_MOV( p, param3, src3 );
+
+    invoke_subroutine( c, SUB_NOISE4, noise4_sub );
+    
+    /* Fill in the result: */
+    brw_set_saturate( p, inst->SaturateMode == SATURATE_ZERO_ONE );
+    for (i = 0 ; i < 4; i++) {
+	if (mask & (1<<i)) {
+	    dst = get_dst_reg(c, inst, i, 1);
+	    brw_MOV( p, dst, param0 );
+	}
+    }
+    if( inst->SaturateMode == SATURATE_ZERO_ONE )
+	brw_set_saturate( p, 0 );
+    
+    release_tmps( c, mark );
+}
+    
 static void emit_wpos_xy(struct brw_wm_compile *c,
                 struct prog_instruction *inst)
 {
@@ -1996,8 +2422,9 @@ static void brw_wm_emit_glsl(struct brw_context *brw, struct brw_wm_compile *c)
 	    case OPCODE_NOISE3:
 		emit_noise3(c, inst);
 		break;
-	    /* case OPCODE_NOISE4: */
-		/* not yet implemented */
+	    case OPCODE_NOISE4:
+		emit_noise4(c, inst);
+		break;
 	    case OPCODE_TEX:
 		emit_tex(c, inst);
 		break;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_iz.c b/src/mesa/drivers/dri/i965/brw_wm_iz.c
index 0bb5d5ba833..bd60ac9b315 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_iz.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_iz.c
@@ -52,70 +52,6 @@ const struct {
 {
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 1, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 0, 1, 0, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 1, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 1, 1, 0 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 0, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 1, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 0, 1, 0, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 1, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { C, 0, 1, 1, 1 }, 
- { P, 0, 0, 0, 0 }, 
- { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
  { P, 0, 0, 0, 0 }, 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
index 205a7160d39..590cd946ec3 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
@@ -51,6 +51,7 @@ static struct brw_wm_value *get_value( struct brw_wm_compile *c)
    return &c->vreg[c->nr_vreg++];
 }
 
+/** return pointer to a newly allocated instruction */
 static struct brw_wm_instruction *get_instruction( struct brw_wm_compile *c )
 {
    assert(c->nr_insns < BRW_WM_MAX_INSN);
@@ -60,6 +61,7 @@ static struct brw_wm_instruction *get_instruction( struct brw_wm_compile *c )
 /***********************************************************************
  */
 
+/** Init the "undef" register */
 static void pass0_init_undef( struct brw_wm_compile *c)
 {
    struct brw_wm_ref *ref = &c->undef_ref;
@@ -69,6 +71,7 @@ static void pass0_init_undef( struct brw_wm_compile *c)
    ref->prevuse = NULL;
 }
 
+/** Set a FP register to a value */
 static void pass0_set_fpreg_value( struct brw_wm_compile *c,
 				   GLuint file,
 				   GLuint idx,
@@ -83,6 +86,7 @@ static void pass0_set_fpreg_value( struct brw_wm_compile *c,
    c->pass0_fp_reg[file][idx][component] = ref;
 }
 
+/** Set a FP register to a ref */
 static void pass0_set_fpreg_ref( struct brw_wm_compile *c,
 				 GLuint file,
 				 GLuint idx,
@@ -115,12 +119,13 @@ static const struct brw_wm_ref *get_param_ref( struct brw_wm_compile *c,
       ref->value = &c->creg[i/16];
       ref->insn = 0;
       ref->prevuse = NULL;
-      
+
       return ref;
    }
 }
 
 
+/** Return a ref to a constant/literal value */
 static const struct brw_wm_ref *get_const_ref( struct brw_wm_compile *c,
 					       const GLfloat *constval )
 {
@@ -142,7 +147,7 @@ static const struct brw_wm_ref *get_const_ref( struct brw_wm_compile *c,
        */
       c->constref[i].constval = *constval;
       c->constref[i].ref = get_param_ref(c, constval);
-   
+
       return c->constref[i].ref;
    }
    else {
@@ -187,7 +192,7 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
 	 
 	 /* There's something really hokey about parameters parsed in
 	  * arb programs - they all end up in here, whether they be
-	  * state values, paramters or constants.  This duplicates the
+	  * state values, parameters or constants.  This duplicates the
 	  * structure above & also seems to subvert the limits set for
 	  * each type of constant/param.
 	  */ 
@@ -198,7 +203,7 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
 	     */
 	    ref = get_const_ref(c, &plist->ParameterValues[idx][component]);
 	    break;
-	    
+
 	 case PROGRAM_STATE_VAR:
 	 case PROGRAM_UNIFORM:
 	    /* These may change from run to run:
@@ -229,14 +234,13 @@ static const struct brw_wm_ref *pass0_get_reg( struct brw_wm_compile *c,
 
 
 
-
 /***********************************************************************
  * Straight translation to internal instruction format
  */
 
 static void pass0_set_dst( struct brw_wm_compile *c,
-			   struct brw_wm_instruction *out,		     
-			   const struct prog_instruction *inst,		     
+			   struct brw_wm_instruction *out,
+			   const struct prog_instruction *inst,
 			   GLuint writemask )
 {
    const struct prog_dst_register *dst = &inst->DstReg;
@@ -245,18 +249,17 @@ static void pass0_set_dst( struct brw_wm_compile *c,
    for (i = 0; i < 4; i++) {
       if (writemask & (1<<i)) {
 	 out->dst[i] = get_value(c);
-
 	 pass0_set_fpreg_value(c, dst->File, dst->Index, i, out->dst[i]);
       }
    }
-   
+
    out->writemask = writemask;
 }
 
 
 static void pass0_set_dst_scalar( struct brw_wm_compile *c,
-				  struct brw_wm_instruction *out,		     
-				  const struct prog_instruction *inst,		     
+				  struct brw_wm_instruction *out,
+                                  const struct prog_instruction *inst,
 				  GLuint writemask )
 {
    if (writemask) {
@@ -282,7 +285,6 @@ static void pass0_set_dst_scalar( struct brw_wm_compile *c,
 }
 
 
-
 static const struct brw_wm_ref *get_fp_src_reg_ref( struct brw_wm_compile *c,
 						    struct prog_src_register src,
 						    GLuint i )
@@ -292,14 +294,13 @@ static const struct brw_wm_ref *get_fp_src_reg_ref( struct brw_wm_compile *c,
    static const GLfloat const_zero = 0.0;
    static const GLfloat const_one = 1.0;
 
-	 
    if (component == SWIZZLE_ZERO) 
       src_ref = get_const_ref(c, &const_zero);
    else if (component == SWIZZLE_ONE) 
       src_ref = get_const_ref(c, &const_one);
    else 
       src_ref = pass0_get_reg(c, src.File, src.Index, component);
-	 
+
    return src_ref;
 }
 
@@ -311,19 +312,19 @@ static struct brw_wm_ref *get_new_ref( struct brw_wm_compile *c,
 {
    const struct brw_wm_ref *ref = get_fp_src_reg_ref(c, src, i);
    struct brw_wm_ref *newref = get_ref(c);
-      
+
    newref->value = ref->value;
    newref->hw_reg = ref->hw_reg;
 
-   if (insn) { 
+   if (insn) {
       newref->insn = insn - c->instruction;
       newref->prevuse = newref->value->lastuse;
       newref->value->lastuse = newref;
    }
 
-   if (src.NegateBase & (1<<i)) 
+   if (src.NegateBase & (1<<i))
       newref->hw_reg.negate ^= 1;
-	    
+
    if (src.Abs) {
       newref->hw_reg.negate = 0;
       newref->hw_reg.abs = 1;
@@ -333,7 +334,6 @@ static struct brw_wm_ref *get_new_ref( struct brw_wm_compile *c,
 }
 
 
-
 static struct brw_wm_instruction *translate_insn( struct brw_wm_compile *c,
 						  const struct prog_instruction *inst )
 {
@@ -379,14 +379,22 @@ static void pass0_precalc_mov( struct brw_wm_compile *c,
 {
    const struct prog_dst_register *dst = &inst->DstReg;
    GLuint writemask = inst->DstReg.WriteMask;
+   struct brw_wm_ref *refs[4];
    GLuint i;
 
    /* Get the effect of a MOV by manipulating our register table:
+    * First get all refs, then assign refs.  This ensures that "in-place"
+    * swizzles such as:
+    *   MOV t, t.xxyx
+    * are handled correctly.  Previously, these two steps were done in
+    * one loop and the above case was incorrectly handled.
     */
    for (i = 0; i < 4; i++) {
-      if (writemask & (1<<i)) {	    
-	 pass0_set_fpreg_ref( c, dst->File, dst->Index, i, 
-			      get_new_ref(c, inst->SrcReg[0], i, NULL));
+      refs[i] = get_new_ref(c, inst->SrcReg[0], i, NULL);
+   }
+   for (i = 0; i < 4; i++) {
+      if (writemask & (1 << i)) {	    
+         pass0_set_fpreg_ref( c, dst->File, dst->Index, i, refs[i]);
       }
    }
 }
@@ -418,6 +426,7 @@ static void pass0_init_payload( struct brw_wm_compile *c )
 			     &c->payload.input_interp[i] );      
 }
 
+
 /***********************************************************************
  * PASS 0
  *
@@ -440,7 +449,6 @@ void brw_wm_pass0( struct brw_wm_compile *c )
    for (insn = 0; insn < c->nr_fp_insns; insn++) {
       const struct prog_instruction *inst = &c->prog_instructions[insn];
 
-
       /* Optimize away moves, otherwise emit translated instruction:
        */      
       switch (inst->Opcode) {
@@ -453,8 +461,6 @@ void brw_wm_pass0( struct brw_wm_compile *c )
 	    translate_insn(c, inst);
 	 }
 	 break;
-	 
-
       default:
 	 translate_insn(c, inst);
 	 break;
@@ -465,4 +471,3 @@ void brw_wm_pass0( struct brw_wm_compile *c )
       brw_wm_print_program(c, "pass0");
    }
 }
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass1.c b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
index f6f3a38e9e0..6eaed8a665b 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass1.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass1.c
@@ -58,7 +58,8 @@ static void unlink_ref(struct brw_wm_ref *ref)
 
    if (ref == value->lastuse) {
       value->lastuse = ref->prevuse;
-   } else {
+   }
+   else {
       struct brw_wm_ref *i = value->lastuse;
       while (i->prevuse != ref) i = i->prevuse;
       i->prevuse = ref->prevuse;
@@ -75,8 +76,9 @@ static void track_arg(struct brw_wm_compile *c,
    for (i = 0; i < 4; i++) {
       struct brw_wm_ref *ref = inst->src[arg][i];
       if (ref) {
-	 if (readmask & (1<<i)) 
+	 if (readmask & (1<<i)) {
 	    ref->value->contributes_to_output = 1;
+         }
 	 else {
 	    unlink_ref(ref);
 	    inst->src[arg][i] = NULL;
@@ -88,15 +90,21 @@ static void track_arg(struct brw_wm_compile *c,
 static GLuint get_texcoord_mask( GLuint tex_idx )
 {
    switch (tex_idx) {
-   case TEXTURE_1D_INDEX: return WRITEMASK_X;
-   case TEXTURE_2D_INDEX: return WRITEMASK_XY;
-   case TEXTURE_3D_INDEX: return WRITEMASK_XYZ;
-   case TEXTURE_CUBE_INDEX: return WRITEMASK_XYZ;
-   case TEXTURE_RECT_INDEX: return WRITEMASK_XY;
+   case TEXTURE_1D_INDEX:
+      return WRITEMASK_X;
+   case TEXTURE_2D_INDEX:
+      return WRITEMASK_XY;
+   case TEXTURE_3D_INDEX:
+      return WRITEMASK_XYZ;
+   case TEXTURE_CUBE_INDEX:
+      return WRITEMASK_XYZ;
+   case TEXTURE_RECT_INDEX:
+      return WRITEMASK_XY;
    default: return 0;
    }
 }
 
+
 /* Step two: Basically this is dead code elimination.  
  *
  * Iterate backwards over instructions, noting which values
@@ -273,6 +281,3 @@ void brw_wm_pass1( struct brw_wm_compile *c )
       brw_wm_print_program(c, "pass1");
    }
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass2.c b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
index 6fca9ad220a..780edbc42e6 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass2.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass2.c
@@ -84,7 +84,7 @@ static void init_registers( struct brw_wm_compile *c )
    for (j = 0; j < c->nr_creg; j++) 
       prealloc_reg(c, &c->creg[j], i++);
 
-   for (j = 0; j < FRAG_ATTRIB_MAX; j++) 
+   for (j = 0; j < FRAG_ATTRIB_MAX; j++) {
       if (inputs & (1<<j)) {
 	 /* index for vs output and ps input are not the same 
 	    in shader varying */
@@ -96,6 +96,7 @@ static void init_registers( struct brw_wm_compile *c )
 	 nr_interp_regs++;
 	 prealloc_reg(c, &c->payload.input_interp[index], i++);
       }
+   }
 
    assert(nr_interp_regs >= 1);
 
@@ -120,7 +121,7 @@ static void update_register_usage(struct brw_wm_compile *c,
       /* Only search those which can change:
        */
       if (grf->nextuse < thisinsn) {
-	 struct brw_wm_ref *ref = grf->value->lastuse;
+	 const struct brw_wm_ref *ref = grf->value->lastuse;
 
 	 /* Has last use of value been passed?
 	  */
@@ -148,7 +149,7 @@ static void spill_value(struct brw_wm_compile *c,
    /* Allocate a spill slot.  Note that allocations start from 0x40 -
     * the first slot is reserved to mean "undef" in brw_wm_emit.c
     */
-   if (!value->spill_slot) {  
+   if (!value->spill_slot) {
       c->last_scratch += 0x40;	
       value->spill_slot = c->last_scratch;
    }
@@ -189,7 +190,7 @@ static GLuint search_contiguous_regs(struct brw_wm_compile *c,
 	 if (grf[i+j].nextuse < group_nextuse)
 	    group_nextuse = grf[i+j].nextuse;
       }
-	 
+
       if (group_nextuse > furthest) {
 	 furthest = group_nextuse;
 	 reg = i;
@@ -197,7 +198,7 @@ static GLuint search_contiguous_regs(struct brw_wm_compile *c,
    }
 
    assert(furthest != thisinsn);
-   
+
    /* Any non-empty regs will need to be spilled:
     */
    for (j = 0; j < nr; j++) 
@@ -243,7 +244,7 @@ static void alloc_contiguous_dest(struct brw_wm_compile *c,
 
 static void load_args(struct brw_wm_compile *c, 
 		      struct brw_wm_instruction *inst)
-{   
+{
    GLuint thisinsn = inst - c->instruction;
    GLuint i,j;
 
@@ -258,17 +259,17 @@ static void load_args(struct brw_wm_compile *c,
 		* register allocation and mark the ref as requiring a fill.
 		*/
 	       GLuint reg = search_contiguous_regs(c, 1, thisinsn);
-            
+
 	       c->pass2_grf[reg].value = ref->value;
 	       c->pass2_grf[reg].nextuse = thisinsn;
-	    
+
 	       ref->value->resident = &c->pass2_grf[reg];
 
 	       /* Note that a fill is required:
 		*/
 	       ref->unspill_reg = reg*2;
 	    }
-	    
+
 	    /* Adjust the hw_reg to point at the value's current location:
 	     */
 	    assert(ref->value == ref->value->resident->value);
@@ -294,7 +295,7 @@ void brw_wm_pass2( struct brw_wm_compile *c )
 
    for (insn = 0; insn < c->nr_insns; insn++) {
       struct brw_wm_instruction *inst = &c->instruction[insn];
-      
+
       /* Update registers' nextuse values:
        */
       update_register_usage(c, insn);
@@ -322,11 +323,11 @@ void brw_wm_pass2( struct brw_wm_compile *c )
 	 break;
       }
 
-      if (TEST_DST_SPILLS && inst->opcode != WM_PIXELXY)
+      if (TEST_DST_SPILLS && inst->opcode != WM_PIXELXY) {
 	 for (i = 0; i < 4; i++)	
 	    if (inst->dst[i])
 	       spill_value(c, inst->dst[i]);
-
+      }
    }
 
    if (INTEL_DEBUG & DEBUG_WM) {
@@ -339,6 +340,3 @@ void brw_wm_pass2( struct brw_wm_compile *c )
        brw_wm_print_program(c, "pass2/done");
    }
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
index f12ef47a7d7..b6dac0d6983 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
@@ -220,14 +220,15 @@ static void
 brw_wm_sampler_populate_key(struct brw_context *brw,
 			    struct wm_sampler_key *key)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    int unit;
 
    memset(key, 0, sizeof(*key));
 
    for (unit = 0; unit < BRW_MAX_TEX_UNIT; unit++) {
-      if (brw->attribs.Texture->Unit[unit]._ReallyEnabled) {
+      if (ctx->Texture.Unit[unit]._ReallyEnabled) {
 	 struct wm_sampler_entry *entry = &key->sampler[unit];
-	 struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[unit];
+	 struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 	 struct gl_texture_object *texObj = texUnit->_Current;
 	 struct intel_texture_object *intelObj = intel_texture_object(texObj);
 	 struct gl_texture_image *firstImage =
@@ -244,7 +245,7 @@ brw_wm_sampler_populate_key(struct brw_context *brw,
 	 entry->minfilter = texObj->MinFilter;
 	 entry->magfilter = texObj->MagFilter;
 	 entry->comparemode = texObj->CompareMode;
-    entry->comparefunc = texObj->CompareFunc;
+         entry->comparefunc = texObj->CompareFunc;
 
 	 dri_bo_unreference(brw->wm.sdc_bo[unit]);
 	 if (firstImage->_BaseFormat == GL_DEPTH_COMPONENT) {
@@ -274,6 +275,7 @@ brw_wm_sampler_populate_key(struct brw_context *brw,
  */
 static void upload_wm_samplers( struct brw_context *brw )
 {
+   GLcontext *ctx = &brw->intel.ctx;
    struct wm_sampler_key key;
    int i;
 
@@ -317,7 +319,7 @@ static void upload_wm_samplers( struct brw_context *brw )
 
       /* Emit SDC relocations */
       for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-	 if (!brw->attribs.Texture->Unit[i]._ReallyEnabled)
+	 if (!ctx->Texture.Unit[i]._ReallyEnabled)
 	    continue;
 
 	 dri_bo_emit_reloc(brw->wm.sampler_bo,
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index fd461618bc4..3c3b3473d63 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -60,6 +60,7 @@ struct brw_wm_unit_key {
 static void
 wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    const struct gl_fragment_program *fp = brw->fragment_program;
    struct intel_context *intel = &brw->intel;
 
@@ -88,14 +89,14 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
    /* BRW_NEW_CURBE_OFFSETS */
    key->curbe_offset = brw->curbe.wm_start;
 
-   /* CACHE_NEW_SURFACE */
+   /* BRW_NEW_NR_SURFACEs */
    key->nr_surfaces = brw->wm.nr_surfaces;
 
    /* CACHE_NEW_SAMPLER */
    key->sampler_count = brw->wm.sampler_count;
 
    /* _NEW_POLYGONSTIPPLE */
-   key->polygon_stipple = brw->attribs.Polygon->StippleFlag;
+   key->polygon_stipple = ctx->Polygon.StippleFlag;
 
    /* BRW_NEW_FRAGMENT_PROGRAM */
    key->uses_depth = (fp->Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) != 0;
@@ -105,19 +106,19 @@ wm_unit_populate_key(struct brw_context *brw, struct brw_wm_unit_key *key)
       (fp->Base.OutputsWritten & (1 << FRAG_RESULT_DEPR)) != 0;
 
    /* _NEW_COLOR */
-   key->uses_kill = fp->UsesKill || brw->attribs.Color->AlphaEnabled;
+   key->uses_kill = fp->UsesKill || ctx->Color.AlphaEnabled;
    key->is_glsl = brw_wm_is_glsl(fp);
 
    /* XXX: This needs a flag to indicate when it changes. */
    key->stats_wm = intel->stats_wm;
 
    /* _NEW_LINE */
-   key->line_stipple = brw->attribs.Line->StippleFlag;
+   key->line_stipple = ctx->Line.StippleFlag;
 
    /* _NEW_POLYGON */
-   key->offset_enable = brw->attribs.Polygon->OffsetFill;
-   key->offset_units = brw->attribs.Polygon->OffsetUnits;
-   key->offset_factor = brw->attribs.Polygon->OffsetFactor;
+   key->offset_enable = ctx->Polygon.OffsetFill;
+   key->offset_units = ctx->Polygon.OffsetUnits;
+   key->offset_factor = ctx->Polygon.OffsetFactor;
 }
 
 static dri_bo *
@@ -281,10 +282,9 @@ const struct brw_tracked_state brw_wm_unit = {
 
       .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
 	      BRW_NEW_CURBE_OFFSETS |
-	      BRW_NEW_LOCK),
+	      BRW_NEW_NR_SURFACES),
 
-      .cache = (CACHE_NEW_SURFACE | 
-		CACHE_NEW_WM_PROG | 
+      .cache = (CACHE_NEW_WM_PROG |
 		CACHE_NEW_SAMPLER)
    },
    .prepare = upload_wm_unit,
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index 3790b50c976..d70f9c646cd 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -192,21 +192,27 @@ brw_create_texture_surface( struct brw_context *brw,
    if (key->bo) 
       surf.ss0.surface_format = translate_tex_format(key->format, key->depthmode);
    else {
-     switch(key->depth) {
-     case 32: surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM; break;
-     default:
-     case 24: surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8X8_UNORM; break;
-     case 16: surf.ss0.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM; break;
-     }
+      switch (key->depth) {
+      case 32:
+         surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8A8_UNORM;
+         break;
+      default:
+      case 24:
+         surf.ss0.surface_format = BRW_SURFACEFORMAT_B8G8R8X8_UNORM;
+         break;
+      case 16:
+         surf.ss0.surface_format = BRW_SURFACEFORMAT_B5G6R5_UNORM;
+         break;
+      }
    }
 
    /* This is ok for all textures with channel width 8bit or less:
     */
 /*    surf.ss0.data_return_format = BRW_SURFACERETURNFORMAT_S1; */
    if (key->bo)
-     surf.ss1.base_addr = key->bo->offset; /* reloc */
+      surf.ss1.base_addr = key->bo->offset; /* reloc */
    else
-     surf.ss1.base_addr = key->offset;
+      surf.ss1.base_addr = key->offset;
 
    surf.ss2.mip_count = key->last_level - key->first_level;
    surf.ss2.width = key->width - 1;
@@ -247,7 +253,7 @@ static void
 brw_update_texture_surface( GLcontext *ctx, GLuint unit )
 {
    struct brw_context *brw = brw_context(ctx);
-   struct gl_texture_object *tObj = brw->attribs.Texture->Unit[unit]._Current;
+   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
    struct intel_texture_object *intelObj = intel_texture_object(tObj);
    struct gl_texture_image *firstImage = tObj->Image[0][intelObj->firstLevel];
    struct brw_wm_surface_key key;
@@ -295,6 +301,7 @@ static void
 brw_update_region_surface(struct brw_context *brw, struct intel_region *region,
 			  unsigned int unit, GLboolean cached)
 {
+   GLcontext *ctx = &brw->intel.ctx;
    dri_bo *region_bo = NULL;
    struct {
       unsigned int surface_type;
@@ -327,10 +334,10 @@ brw_update_region_surface(struct brw_context *brw, struct intel_region *region,
       key.height = 1;
       key.cpp = 4;
    }
-   memcpy(key.color_mask, brw->attribs.Color->ColorMask,
+   memcpy(key.color_mask, ctx->Color.ColorMask,
 	  sizeof(key.color_mask));
-   key.color_blend = (!brw->attribs.Color->_LogicOpEnabled &&
-		      brw->attribs.Color->BlendEnabled);
+   key.color_blend = (!ctx->Color._LogicOpEnabled &&
+		      ctx->Color.BlendEnabled);
 
    dri_bo_unreference(brw->wm.surf_bo[unit]);
    brw->wm.surf_bo[unit] = NULL;
@@ -438,6 +445,7 @@ static void prepare_wm_surfaces(struct brw_context *brw )
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = &brw->intel;
    GLuint i;
+   int old_nr_surfaces;
 
    if (brw->state.nr_draw_regions  > 1) {
       for (i = 0; i < brw->state.nr_draw_regions; i++) {
@@ -448,10 +456,11 @@ static void prepare_wm_surfaces(struct brw_context *brw )
       brw_update_region_surface(brw, brw->state.draw_regions[0], 0, GL_TRUE);
    }
 
+   old_nr_surfaces = brw->wm.nr_surfaces;
    brw->wm.nr_surfaces = MAX_DRAW_BUFFERS;
 
    for (i = 0; i < BRW_MAX_TEX_UNIT; i++) {
-      struct gl_texture_unit *texUnit = &brw->attribs.Texture->Unit[i];
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
 
       /* _NEW_TEXTURE, BRW_NEW_TEXDATA */
       if(texUnit->_ReallyEnabled) {
@@ -473,6 +482,9 @@ static void prepare_wm_surfaces(struct brw_context *brw )
 
    dri_bo_unreference(brw->wm.bind_bo);
    brw->wm.bind_bo = brw_wm_get_binding_table(brw);
+
+   if (brw->wm.nr_surfaces != old_nr_surfaces)
+      brw->state.dirty.brw |= BRW_NEW_NR_SURFACES;
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/intel_clear.c b/src/mesa/drivers/dri/i965/intel_clear.c
new file mode 120000
index 00000000000..9a2a742a0d9
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_clear.c
@@ -0,0 +1 @@
+../intel/intel_clear.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_depthstencil.c b/src/mesa/drivers/dri/i965/intel_depthstencil.c
deleted file mode 120000
index 4ac4ae690a3..00000000000
--- a/src/mesa/drivers/dri/i965/intel_depthstencil.c
+++ /dev/null
@@ -1 +0,0 @@
-../intel/intel_depthstencil.c
-\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_extensions.c b/src/mesa/drivers/dri/i965/intel_extensions.c
new file mode 120000
index 00000000000..a2f3e8cd208
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_extensions.c
@@ -0,0 +1 @@
+../intel/intel_extensions.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_state.c b/src/mesa/drivers/dri/i965/intel_state.c
index 67ef5f78c1a..519672fc359 100644..120000
--- a/src/mesa/drivers/dri/i965/intel_state.c
+++ b/src/mesa/drivers/dri/i965/intel_state.c
@@ -1,225 +1 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "main/enums.h"
-#include "main/colormac.h"
-#include "main/dd.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_regions.h"
-#include "swrast/swrast.h"
-
-int intel_translate_shadow_compare_func( GLenum func )
-{
-   switch(func) {
-   case GL_NEVER: 
-       return COMPAREFUNC_ALWAYS; 
-   case GL_LESS: 
-       return COMPAREFUNC_LEQUAL; 
-   case GL_LEQUAL: 
-       return COMPAREFUNC_LESS;
-   case GL_GREATER: 
-       return COMPAREFUNC_GEQUAL; 
-   case GL_GEQUAL: 
-      return COMPAREFUNC_GREATER; 
-   case GL_NOTEQUAL: 
-      return COMPAREFUNC_EQUAL; 
-   case GL_EQUAL: 
-      return COMPAREFUNC_NOTEQUAL; 
-   case GL_ALWAYS: 
-       return COMPAREFUNC_NEVER; 
-   }
-
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_NEVER; 
-}
-
-int intel_translate_compare_func( GLenum func )
-{
-   switch(func) {
-   case GL_NEVER: 
-      return COMPAREFUNC_NEVER; 
-   case GL_LESS: 
-      return COMPAREFUNC_LESS; 
-   case GL_LEQUAL: 
-      return COMPAREFUNC_LEQUAL; 
-   case GL_GREATER: 
-      return COMPAREFUNC_GREATER; 
-   case GL_GEQUAL: 
-      return COMPAREFUNC_GEQUAL; 
-   case GL_NOTEQUAL: 
-      return COMPAREFUNC_NOTEQUAL; 
-   case GL_EQUAL: 
-      return COMPAREFUNC_EQUAL; 
-   case GL_ALWAYS: 
-      return COMPAREFUNC_ALWAYS; 
-   }
-
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
-   return COMPAREFUNC_ALWAYS; 
-}
-
-int intel_translate_stencil_op( GLenum op )
-{
-   switch(op) {
-   case GL_KEEP: 
-      return STENCILOP_KEEP; 
-   case GL_ZERO: 
-      return STENCILOP_ZERO; 
-   case GL_REPLACE: 
-      return STENCILOP_REPLACE; 
-   case GL_INCR: 
-      return STENCILOP_INCRSAT;
-   case GL_DECR: 
-      return STENCILOP_DECRSAT;
-   case GL_INCR_WRAP:
-      return STENCILOP_INCR; 
-   case GL_DECR_WRAP:
-      return STENCILOP_DECR; 
-   case GL_INVERT: 
-      return STENCILOP_INVERT; 
-   default: 
-      return STENCILOP_ZERO;
-   }
-}
-
-int intel_translate_blend_factor( GLenum factor )
-{
-   switch(factor) {
-   case GL_ZERO: 
-      return BLENDFACT_ZERO; 
-   case GL_SRC_ALPHA: 
-      return BLENDFACT_SRC_ALPHA; 
-   case GL_ONE: 
-      return BLENDFACT_ONE; 
-   case GL_SRC_COLOR: 
-      return BLENDFACT_SRC_COLR; 
-   case GL_ONE_MINUS_SRC_COLOR: 
-      return BLENDFACT_INV_SRC_COLR; 
-   case GL_DST_COLOR: 
-      return BLENDFACT_DST_COLR; 
-   case GL_ONE_MINUS_DST_COLOR: 
-      return BLENDFACT_INV_DST_COLR; 
-   case GL_ONE_MINUS_SRC_ALPHA:
-      return BLENDFACT_INV_SRC_ALPHA; 
-   case GL_DST_ALPHA: 
-      return BLENDFACT_DST_ALPHA; 
-   case GL_ONE_MINUS_DST_ALPHA:
-      return BLENDFACT_INV_DST_ALPHA; 
-   case GL_SRC_ALPHA_SATURATE: 
-      return BLENDFACT_SRC_ALPHA_SATURATE;
-   case GL_CONSTANT_COLOR:
-      return BLENDFACT_CONST_COLOR; 
-   case GL_ONE_MINUS_CONSTANT_COLOR:
-      return BLENDFACT_INV_CONST_COLOR;
-   case GL_CONSTANT_ALPHA:
-      return BLENDFACT_CONST_ALPHA; 
-   case GL_ONE_MINUS_CONSTANT_ALPHA:
-      return BLENDFACT_INV_CONST_ALPHA;
-   }
-   
-   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, factor);
-   return BLENDFACT_ZERO;
-}
-
-int intel_translate_logic_op( GLenum opcode )
-{
-   switch(opcode) {
-   case GL_CLEAR: 
-      return LOGICOP_CLEAR; 
-   case GL_AND: 
-      return LOGICOP_AND; 
-   case GL_AND_REVERSE: 
-      return LOGICOP_AND_RVRSE; 
-   case GL_COPY: 
-      return LOGICOP_COPY; 
-   case GL_COPY_INVERTED: 
-      return LOGICOP_COPY_INV; 
-   case GL_AND_INVERTED: 
-      return LOGICOP_AND_INV; 
-   case GL_NOOP: 
-      return LOGICOP_NOOP; 
-   case GL_XOR: 
-      return LOGICOP_XOR; 
-   case GL_OR: 
-      return LOGICOP_OR; 
-   case GL_OR_INVERTED: 
-      return LOGICOP_OR_INV; 
-   case GL_NOR: 
-      return LOGICOP_NOR; 
-   case GL_EQUIV: 
-      return LOGICOP_EQUIV; 
-   case GL_INVERT: 
-      return LOGICOP_INV; 
-   case GL_OR_REVERSE: 
-      return LOGICOP_OR_RVRSE; 
-   case GL_NAND: 
-      return LOGICOP_NAND; 
-   case GL_SET: 
-      return LOGICOP_SET; 
-   default:
-      return LOGICOP_SET;
-   }
-}
-
-
-static void intelClearColor(GLcontext *ctx, const GLfloat color[4])
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   UNCLAMPED_FLOAT_TO_RGBA_CHAN(intel->clear_chan, color);
-
-   intel->ClearColor8888 = INTEL_PACKCOLOR8888(intel->clear_chan[0],
-					       intel->clear_chan[1],
-					       intel->clear_chan[2],
-					       intel->clear_chan[3]);
-   intel->ClearColor565 = INTEL_PACKCOLOR565(intel->clear_chan[0],
-					     intel->clear_chan[1],
-					     intel->clear_chan[2]);
-}
-
-
-
-/* Fallback to swrast for select and feedback.
- */
-static void intelRenderMode( GLcontext *ctx, GLenum mode )
-{
-   struct intel_context *intel = intel_context(ctx);
-   FALLBACK( intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER) );
-}
-
-
-void intelInitStateFuncs( struct dd_function_table *functions )
-{
-   functions->RenderMode = intelRenderMode;
-   functions->ClearColor = intelClearColor;
-}
+../intel/intel_state.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_swapbuffers.c b/src/mesa/drivers/dri/i965/intel_swapbuffers.c
new file mode 120000
index 00000000000..148d5215aa6
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_swapbuffers.c
@@ -0,0 +1 @@
+../intel/intel_swapbuffers.c
+\ No newline at end of file
diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
index 8129996979f..51579df09e7 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -55,6 +55,12 @@ struct intel_batchbuffer
 
    GLuint size;
 
+   /** Tracking of BEGIN_BATCH()/OUT_BATCH()/ADVANCE_BATCH() debugging */
+   struct {
+      GLuint total;
+      GLubyte *start_ptr;
+   } emit;
+
    GLuint dirty_state;
 };
 
@@ -143,9 +149,12 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
 
 #define BEGIN_BATCH(n, cliprect_mode) do {				\
    intel_batchbuffer_require_space(intel->batch, (n)*4, cliprect_mode); \
+   assert(intel->batch->emit.start_ptr == NULL);			\
+   intel->batch->emit.total = (n) * 4;					\
+   intel->batch->emit.start_ptr = intel->batch->ptr;			\
 } while (0)
 
-#define OUT_BATCH(d)  intel_batchbuffer_emit_dword(intel->batch, d)
+#define OUT_BATCH(d) intel_batchbuffer_emit_dword(intel->batch, d)
 
 #define OUT_RELOC(buf, read_domains, write_domain, delta) do {		\
    assert((delta) >= 0);						\
@@ -153,7 +162,16 @@ intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
 				read_domains, write_domain, delta);	\
 } while (0)
 
-#define ADVANCE_BATCH() do { } while(0)
+#define ADVANCE_BATCH() do {						\
+   unsigned int _n = intel->batch->ptr - intel->batch->emit.start_ptr;	\
+   assert(intel->batch->emit.start_ptr != NULL);			\
+   if (_n != intel->batch->emit.total) {				\
+      fprintf(stderr, "ADVANCE_BATCH: %d of %d dwords emitted\n",	\
+	      _n, intel->batch->emit.total);				\
+      abort();								\
+   }									\
+   intel->batch->emit.start_ptr = NULL;					\
+} while(0)
 
 
 static INLINE void
diff --git a/src/mesa/drivers/dri/intel/intel_blit.c b/src/mesa/drivers/dri/intel/intel_blit.c
index e1046f4a5de..208f90c0abb 100644
--- a/src/mesa/drivers/dri/intel/intel_blit.c
+++ b/src/mesa/drivers/dri/intel/intel_blit.c
@@ -40,6 +40,7 @@
 #include "intel_reg.h"
 #include "intel_regions.h"
 #include "intel_batchbuffer.h"
+#include "intel_chipset.h"
 
 #define FILE_DEBUG_FLAG DEBUG_BLIT
 
@@ -359,51 +360,24 @@ intelEmitCopyBlit(struct intel_context *intel,
       return;
    }
 
-   /* Initial y values don't seem to work with negative pitches.  If
-    * we adjust the offsets manually (below), it seems to work fine.
-    *
-    * On the other hand, if we always adjust, the hardware doesn't
-    * know which blit directions to use, so overlapping copypixels get
-    * the wrong result.
-    */
-   if (dst_pitch > 0 && src_pitch > 0) {
-      assert(dst_x < dst_x2);
-      assert(dst_y < dst_y2);
-
-      BEGIN_BATCH(8, NO_LOOP_CLIPRECTS);
-      OUT_BATCH(CMD);
-      OUT_BATCH(BR13 | dst_pitch);
-      OUT_BATCH((dst_y << 16) | dst_x);
-      OUT_BATCH((dst_y2 << 16) | dst_x2);
-      OUT_RELOC(dst_buffer,
-		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
-		dst_offset);
-      OUT_BATCH((src_y << 16) | src_x);
-      OUT_BATCH(src_pitch);
-      OUT_RELOC(src_buffer,
-		I915_GEM_DOMAIN_RENDER, 0,
-		src_offset);
-      ADVANCE_BATCH();
-   }
-   else {
-      assert(dst_x < dst_x2);
-      assert(h > 0);
-
-      BEGIN_BATCH(8, NO_LOOP_CLIPRECTS);
-      OUT_BATCH(CMD);
-      OUT_BATCH(BR13 | ((uint16_t)dst_pitch));
-      OUT_BATCH((0 << 16) | dst_x);
-      OUT_BATCH((h << 16) | dst_x2);
-      OUT_RELOC(dst_buffer,
-		I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
-		dst_offset + dst_y * dst_pitch);
-      OUT_BATCH((0 << 16) | src_x);
-      OUT_BATCH(src_pitch);
-      OUT_RELOC(src_buffer,
-		I915_GEM_DOMAIN_RENDER, 0,
-		src_offset + src_y * src_pitch);
-      ADVANCE_BATCH();
-   }
+   assert(dst_x < dst_x2);
+   assert(dst_y < dst_y2);
+
+   BEGIN_BATCH(8, NO_LOOP_CLIPRECTS);
+   OUT_BATCH(CMD);
+   OUT_BATCH(BR13 | (uint16_t)dst_pitch);
+   OUT_BATCH((dst_y << 16) | dst_x);
+   OUT_BATCH((dst_y2 << 16) | dst_x2);
+   OUT_RELOC(dst_buffer,
+	     I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
+	     dst_offset);
+   OUT_BATCH((src_y << 16) | src_x);
+   OUT_BATCH((uint16_t)src_pitch);
+   OUT_RELOC(src_buffer,
+	     I915_GEM_DOMAIN_RENDER, 0,
+	     src_offset);
+   ADVANCE_BATCH();
+
    intel_batchbuffer_emit_mi_flush(intel->batch);
 }
 
@@ -570,7 +544,6 @@ intelClearWithBlit(GLcontext *ctx, GLbitfield mask)
                   _mesa_debug(ctx, "hardware blit clear buf %d rb id %d\n",
                   buf, irb->Base.Name);
                 */
-	       intel_wait_flips(intel);
 
                assert(b.x1 < b.x2);
                assert(b.y1 < b.y2);
@@ -626,7 +599,7 @@ intelEmitImmediateColorExpandBlit(struct intel_context *intel,
    intel_batchbuffer_require_space( intel->batch,
 				    (8 * 4) +
 				    (3 * 4) +
-				    dwords,
+				    dwords * 4,
 				    REFERENCES_CLIPRECTS );
 
    opcode = XY_SETUP_BLT_CMD;
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.c b/src/mesa/drivers/dri/intel/intel_buffers.c
index f8f009c6a30..7f2144abd4b 100644
--- a/src/mesa/drivers/dri/intel/intel_buffers.c
+++ b/src/mesa/drivers/dri/intel/intel_buffers.c
@@ -25,41 +25,14 @@
  * 
  **************************************************************************/
 
-#include "intel_screen.h"
 #include "intel_context.h"
-#include "intel_blit.h"
 #include "intel_buffers.h"
-#include "intel_chipset.h"
-#include "intel_depthstencil.h"
 #include "intel_fbo.h"
 #include "intel_regions.h"
 #include "intel_batchbuffer.h"
-#include "intel_reg.h"
-#include "main/context.h"
 #include "main/framebuffer.h"
-#include "swrast/swrast.h"
-#include "utils.h"
 #include "drirenderbuffer.h"
-#include "vblank.h"
-#include "i915_drm.h"
 
-/* This block can be removed when libdrm >= 2.3.1 is required */
-
-#ifndef DRM_IOCTL_I915_FLIP
-
-#define DRM_VBLANK_FLIP 0x8000000
-
-typedef struct drm_i915_flip {
-   int pipes;
-} drm_i915_flip_t;
-
-#undef DRM_IOCTL_I915_FLIP
-#define DRM_IOCTL_I915_FLIP DRM_IOW(DRM_COMMAND_BASE + DRM_I915_FLIP, \
-				    drm_i915_flip_t)
-
-#endif
-
-#define FILE_DEBUG_FLAG DEBUG_BLIT
 
 /**
  * XXX move this into a new dri/common/cliprects.c file.
@@ -130,7 +103,6 @@ intel_get_cliprects(struct intel_context *intel,
 		    int *x_off, int *y_off)
 {
    __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
 
    if (intel->constant_cliprect) {
       /* FBO or DRI2 rendering, which can just use the fb's size. */
@@ -143,8 +115,7 @@ intel_get_cliprects(struct intel_context *intel,
       *num_cliprects = 1;
       *x_off = 0;
       *y_off = 0;
-   } else if (intel->front_cliprects ||
-       intel_fb->pf_active || dPriv->numBackClipRects == 0) {
+   } else if (intel->front_cliprects || dPriv->numBackClipRects == 0) {
       /* use the front clip rects */
       *cliprects = dPriv->pClipRects;
       *num_cliprects = dPriv->numClipRects;
@@ -160,641 +131,6 @@ intel_get_cliprects(struct intel_context *intel,
    }
 }
 
-static void
-intelUpdatePageFlipping(struct intel_context *intel,
-			GLint areaA, GLint areaB)
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-   GLboolean pf_active;
-   GLint pf_planes;
-
-   /* Update page flipping info */
-   pf_planes = 0;
-
-   if (areaA > 0)
-      pf_planes |= 1;
-
-   if (areaB > 0)
-      pf_planes |= 2;
-
-   intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
-				(intel_fb->pf_planes & 0x2)) & 0x3;
-
-   intel_fb->pf_num_pages = intel->intelScreen->third.handle ? 3 : 2;
-
-   pf_active = pf_planes && (pf_planes & intel->sarea->pf_active) == pf_planes;
-
-   if (INTEL_DEBUG & DEBUG_LOCK)
-      if (pf_active != intel_fb->pf_active)
-	 _mesa_printf("%s - Page flipping %sactive\n", __progname,
-		      pf_active ? "" : "in");
-
-   if (pf_active) {
-      /* Sync pages between planes if flipping on both at the same time */
-      if (pf_planes == 0x3 && pf_planes != intel_fb->pf_planes &&
-	  (intel->sarea->pf_current_page & 0x3) !=
-	  (((intel->sarea->pf_current_page) >> 2) & 0x3)) {
-	 drm_i915_flip_t flip;
-
-	 if (intel_fb->pf_current_page ==
-	     (intel->sarea->pf_current_page & 0x3)) {
-	    /* XXX: This is ugly, but emitting two flips 'in a row' can cause
-	     * lockups for unknown reasons.
-	     */
-	    intel->sarea->pf_current_page =
-	       intel->sarea->pf_current_page & 0x3;
-	    intel->sarea->pf_current_page |=
-	       ((intel_fb->pf_current_page + intel_fb->pf_num_pages - 1) %
-		intel_fb->pf_num_pages) << 2;
-
-	    flip.pipes = 0x2;
-	 } else {
-	    intel->sarea->pf_current_page =
-	       intel->sarea->pf_current_page & (0x3 << 2);
-	    intel->sarea->pf_current_page |=
-	       (intel_fb->pf_current_page + intel_fb->pf_num_pages - 1) %
-	       intel_fb->pf_num_pages;
-
-	    flip.pipes = 0x1;
-	 }
-
-	 drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
-      }
-
-      intel_fb->pf_planes = pf_planes;
-   }
-
-   intel_fb->pf_active = pf_active;
-   intel_flip_renderbuffers(intel_fb);
-   intel_draw_buffer(&intel->ctx, intel->ctx.DrawBuffer);
-}
-
-/**
- * This will be called whenever the currently bound window is moved/resized.
- * XXX: actually, it seems to NOT be called when the window is only moved (BP).
- */
-void
-intelWindowMoved(struct intel_context *intel)
-{
-   GLcontext *ctx = &intel->ctx;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-
-   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
-       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
-      volatile struct drm_i915_sarea *sarea = intel->sarea;
-      drm_clip_rect_t drw_rect = { .x1 = dPriv->x, .x2 = dPriv->x + dPriv->w,
-				   .y1 = dPriv->y, .y2 = dPriv->y + dPriv->h };
-      drm_clip_rect_t planeA_rect = { .x1 = sarea->planeA_x, .y1 = sarea->planeA_y,
-				     .x2 = sarea->planeA_x + sarea->planeA_w,
-				     .y2 = sarea->planeA_y + sarea->planeA_h };
-      drm_clip_rect_t planeB_rect = { .x1 = sarea->planeB_x, .y1 = sarea->planeB_y,
-				     .x2 = sarea->planeB_x + sarea->planeB_w,
-				     .y2 = sarea->planeB_y + sarea->planeB_h };
-      GLint areaA = driIntersectArea( drw_rect, planeA_rect );
-      GLint areaB = driIntersectArea( drw_rect, planeB_rect );
-      GLuint flags = dPriv->vblFlags;
-
-      intelUpdatePageFlipping(intel, areaA, areaB);
-
-      /* Update vblank info
-       */
-      if (areaB > areaA || (areaA == areaB && areaB > 0)) {
-	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
-      } else {
-	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
-      }
-
-      /* Check to see if we changed pipes */
-      if (flags != dPriv->vblFlags && dPriv->vblFlags &&
-	  !(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ)) {
-	 int64_t count;
-	 drmVBlank vbl;
-	 int i;
-
-	 /*
-	  * Deal with page flipping
-	  */
-	 vbl.request.type = DRM_VBLANK_ABSOLUTE;
-
-	 if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
-	    vbl.request.type |= DRM_VBLANK_SECONDARY;
-	 }
-
-	 for (i = 0; i < intel_fb->pf_num_pages; i++) {
-	    if (!intel_fb->color_rb[i] ||
-		(intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <=
-		(1<<23))
-	       continue;
-
-	    vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending;
-	    drmWaitVBlank(intel->driFd, &vbl);
-	 }
-
-	 /*
-	  * Update msc_base from old pipe
-	  */
-	 driDrawableGetMSC32(dPriv->driScreenPriv, dPriv, &count);
-	 dPriv->msc_base = count;
-	 /*
-	  * Then get new vblank_base and vblSeq values
-	  */
-	 dPriv->vblFlags = flags;
-	 driGetCurrentVBlank(dPriv);
-	 dPriv->vblank_base = dPriv->vblSeq;
-
-	 intel_fb->vbl_waited = dPriv->vblSeq;
-
-	 for (i = 0; i < intel_fb->pf_num_pages; i++) {
-	    if (intel_fb->color_rb[i])
-	       intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_waited;
-	 }
-      }
-   } else {
-      dPriv->vblFlags &= ~VBLANK_FLAG_SECONDARY;
-   }
-
-   /* Update Mesa's notion of window size */
-   driUpdateFramebufferSize(ctx, dPriv);
-   intel_fb->Base.Initialized = GL_TRUE; /* XXX remove someday */
-
-   /* Update hardware scissor */
-   if (ctx->Driver.Scissor != NULL) {
-      ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			  ctx->Scissor.Width, ctx->Scissor.Height);
-   }
-
-   /* Re-calculate viewport related state */
-   if (ctx->Driver.DepthRange != NULL)
-      ctx->Driver.DepthRange( ctx, ctx->Viewport.Near, ctx->Viewport.Far );
-}
-
-
-
-/* A true meta version of this would be very simple and additionally
- * machine independent.  Maybe we'll get there one day.
- */
-static void
-intelClearWithTris(struct intel_context *intel, GLbitfield mask)
-{
-   GLcontext *ctx = &intel->ctx;
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   GLuint buf;
-
-   intel->vtbl.install_meta_state(intel);
-
-   /* Back and stencil cliprects are the same.  Try and do both
-    * buffers at once:
-    */
-   if (mask & (BUFFER_BIT_BACK_LEFT | BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH)) {
-      struct intel_region *backRegion =
-	 intel_get_rb_region(fb, BUFFER_BACK_LEFT);
-      struct intel_region *depthRegion =
-	 intel_get_rb_region(fb, BUFFER_DEPTH);
-
-      intel->vtbl.meta_draw_region(intel, backRegion, depthRegion);
-
-      if (mask & BUFFER_BIT_BACK_LEFT)
-	 intel->vtbl.meta_color_mask(intel, GL_TRUE);
-      else
-	 intel->vtbl.meta_color_mask(intel, GL_FALSE);
-
-      if (mask & BUFFER_BIT_STENCIL)
-	 intel->vtbl.meta_stencil_replace(intel,
-					  intel->ctx.Stencil.WriteMask[0],
-					  intel->ctx.Stencil.Clear);
-      else
-	 intel->vtbl.meta_no_stencil_write(intel);
-
-      if (mask & BUFFER_BIT_DEPTH)
-	 intel->vtbl.meta_depth_replace(intel);
-      else
-	 intel->vtbl.meta_no_depth_write(intel);
-
-      intel->vtbl.meta_draw_quad(intel,
-				 fb->_Xmin,
-				 fb->_Xmax,
-				 fb->_Ymin,
-				 fb->_Ymax,
-				 intel->ctx.Depth.Clear,
-				 intel->ClearColor8888,
-				 0, 0, 0, 0);   /* texcoords */
-
-      mask &= ~(BUFFER_BIT_BACK_LEFT | BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH);
-   }
-
-   /* clear the remaining (color) renderbuffers */
-   for (buf = 0; buf < BUFFER_COUNT && mask; buf++) {
-      const GLuint bufBit = 1 << buf;
-      if (mask & bufBit) {
-	 struct intel_renderbuffer *irbColor =
-	    intel_renderbuffer(fb->Attachment[buf].Renderbuffer);
-
-	 ASSERT(irbColor);
-
-	 intel->vtbl.meta_no_depth_write(intel);
-	 intel->vtbl.meta_no_stencil_write(intel);
-	 intel->vtbl.meta_color_mask(intel, GL_TRUE);
-	 intel->vtbl.meta_draw_region(intel, irbColor->region, NULL);
-
-	 intel->vtbl.meta_draw_quad(intel,
-				    fb->_Xmin,
-				    fb->_Xmax,
-				    fb->_Ymin,
-				    fb->_Ymax,
-				    0, intel->ClearColor8888,
-				    0, 0, 0, 0);   /* texcoords */
-
-	 mask &= ~bufBit;
-      }
-   }
-
-   intel->vtbl.leave_meta_state(intel);
-}
-
-static const char *buffer_names[] = {
-   [BUFFER_FRONT_LEFT] = "front",
-   [BUFFER_BACK_LEFT] = "back",
-   [BUFFER_FRONT_RIGHT] = "front right",
-   [BUFFER_BACK_RIGHT] = "back right",
-   [BUFFER_AUX0] = "aux0",
-   [BUFFER_AUX1] = "aux1",
-   [BUFFER_AUX2] = "aux2",
-   [BUFFER_AUX3] = "aux3",
-   [BUFFER_DEPTH] = "depth",
-   [BUFFER_STENCIL] = "stencil",
-   [BUFFER_ACCUM] = "accum",
-   [BUFFER_COLOR0] = "color0",
-   [BUFFER_COLOR1] = "color1",
-   [BUFFER_COLOR2] = "color2",
-   [BUFFER_COLOR3] = "color3",
-   [BUFFER_COLOR4] = "color4",
-   [BUFFER_COLOR5] = "color5",
-   [BUFFER_COLOR6] = "color6",
-   [BUFFER_COLOR7] = "color7",
-};
-
-/**
- * Called by ctx->Driver.Clear.
- */
-static void
-intelClear(GLcontext *ctx, GLbitfield mask)
-{
-   struct intel_context *intel = intel_context(ctx);
-   const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
-   GLbitfield tri_mask = 0;
-   GLbitfield blit_mask = 0;
-   GLbitfield swrast_mask = 0;
-   struct gl_framebuffer *fb = ctx->DrawBuffer;
-   GLuint i;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* HW color buffers (front, back, aux, generic FBO, etc) */
-   if (colorMask == ~0) {
-      /* clear all R,G,B,A */
-      /* XXX FBO: need to check if colorbuffers are software RBOs! */
-      blit_mask |= (mask & BUFFER_BITS_COLOR);
-   }
-   else {
-      /* glColorMask in effect */
-      tri_mask |= (mask & BUFFER_BITS_COLOR);
-   }
-
-   /* HW stencil */
-   if (mask & BUFFER_BIT_STENCIL) {
-      const struct intel_region *stencilRegion
-         = intel_get_rb_region(fb, BUFFER_STENCIL);
-      if (stencilRegion) {
-         /* have hw stencil */
-         if (IS_965(intel->intelScreen->deviceID) ||
-	     (ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
-	    /* We have to use the 3D engine if we're clearing a partial mask
-	     * of the stencil buffer, or if we're on a 965 which has a tiled
-	     * depth/stencil buffer in a layout we can't blit to.
-	     */
-            tri_mask |= BUFFER_BIT_STENCIL;
-         }
-         else {
-            /* clearing all stencil bits, use blitting */
-            blit_mask |= BUFFER_BIT_STENCIL;
-         }
-      }
-   }
-
-   /* HW depth */
-   if (mask & BUFFER_BIT_DEPTH) {
-      /* clear depth with whatever method is used for stencil (see above) */
-      if (IS_965(intel->intelScreen->deviceID) ||
-	  tri_mask & BUFFER_BIT_STENCIL)
-         tri_mask |= BUFFER_BIT_DEPTH;
-      else
-         blit_mask |= BUFFER_BIT_DEPTH;
-   }
-
-   /* SW fallback clearing */
-   swrast_mask = mask & ~tri_mask & ~blit_mask;
-
-   for (i = 0; i < BUFFER_COUNT; i++) {
-      GLuint bufBit = 1 << i;
-      if ((blit_mask | tri_mask) & bufBit) {
-         if (!fb->Attachment[i].Renderbuffer->ClassID) {
-            blit_mask &= ~bufBit;
-            tri_mask &= ~bufBit;
-            swrast_mask |= bufBit;
-         }
-      }
-   }
-
-   if (blit_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("blit clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (blit_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
-      intelClearWithBlit(ctx, blit_mask);
-   }
-
-   if (tri_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("tri clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (tri_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
-      intelClearWithTris(intel, tri_mask);
-   }
-
-   if (swrast_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("swrast clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (swrast_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
-      _swrast_Clear(ctx, swrast_mask);
-   }
-}
-
-
-/* Emit wait for pending flips */
-void
-intel_wait_flips(struct intel_context *intel)
-{
-   struct intel_framebuffer *intel_fb =
-      (struct intel_framebuffer *) intel->ctx.DrawBuffer;
-   struct intel_renderbuffer *intel_rb =
-      intel_get_renderbuffer(&intel_fb->Base,
-			     intel_fb->Base._ColorDrawBufferIndexes[0] ==
-			     BUFFER_FRONT_LEFT ? BUFFER_FRONT_LEFT :
-			     BUFFER_BACK_LEFT);
-
-   if (intel->intelScreen->driScrnPriv->dri2.enabled)
-      return;
-
-   if (intel_fb->Base.Name == 0 && intel_rb &&
-       intel_rb->pf_pending == intel_fb->pf_seq) {
-      GLint pf_planes = intel_fb->pf_planes;
-      BATCH_LOCALS;
-
-      /* Wait for pending flips to take effect */
-      BEGIN_BATCH(2, NO_LOOP_CLIPRECTS);
-      OUT_BATCH(pf_planes & 0x1 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_A_FLIP)
-		: 0);
-      OUT_BATCH(pf_planes & 0x2 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_B_FLIP)
-		: 0);
-      ADVANCE_BATCH();
-
-      intel_rb->pf_pending--;
-   }
-}
-
-
-/* Flip the front & back buffers
- */
-static GLboolean
-intelPageFlip(const __DRIdrawablePrivate * dPriv)
-{
-   struct intel_context *intel;
-   int ret;
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-
-   if (intel->intelScreen->drmMinor < 9)
-      return GL_FALSE;
-
-   intelFlush(&intel->ctx);
-
-   ret = 0;
-
-   LOCK_HARDWARE(intel);
-
-   if (dPriv->numClipRects && intel_fb->pf_active) {
-      drm_i915_flip_t flip;
-
-      flip.pipes = intel_fb->pf_planes;
-
-      ret = drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
-   }
-
-   UNLOCK_HARDWARE(intel);
-
-   if (ret || !intel_fb->pf_active)
-      return GL_FALSE;
-
-   if (!dPriv->numClipRects) {
-      usleep(10000);	/* throttle invisible client 10ms */
-   }
-
-   intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
-				(intel_fb->pf_planes & 0x2)) & 0x3;
-
-   if (dPriv->numClipRects != 0) {
-      intel_get_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT)->pf_pending =
-      intel_get_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT)->pf_pending =
-	 ++intel_fb->pf_seq;
-   }
-
-   intel_flip_renderbuffers(intel_fb);
-   intel_draw_buffer(&intel->ctx, &intel_fb->Base);
-
-   return GL_TRUE;
-}
-
-static GLboolean
-intelScheduleSwap(__DRIdrawablePrivate * dPriv, GLboolean *missed_target)
-{
-   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-   unsigned int interval;
-   struct intel_context *intel =
-      intelScreenContext(dPriv->driScreenPriv->private);
-   const intelScreenPrivate *intelScreen = intel->intelScreen;
-   unsigned int target;
-   drm_i915_vblank_swap_t swap;
-   GLboolean ret;
-
-   if (!dPriv->vblFlags ||
-       (dPriv->vblFlags & VBLANK_FLAG_NO_IRQ) ||
-       intelScreen->drmMinor < (intel_fb->pf_active ? 9 : 6))
-      return GL_FALSE;
-
-   interval = driGetVBlankInterval(dPriv);
-
-   swap.seqtype = DRM_VBLANK_ABSOLUTE;
-
-   if (dPriv->vblFlags & VBLANK_FLAG_SYNC) {
-      swap.seqtype |= DRM_VBLANK_NEXTONMISS;
-   } else if (interval == 0)
-      return GL_FALSE;
-
-   swap.drawable = dPriv->hHWDrawable;
-   target = swap.sequence = dPriv->vblSeq + interval;
-
-   if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
-      swap.seqtype |= DRM_VBLANK_SECONDARY;
-   }
-
-   LOCK_HARDWARE(intel);
-
-   intel_batchbuffer_flush(intel->batch);
-
-   if ( intel_fb->pf_active ) {
-      swap.seqtype |= DRM_VBLANK_FLIP;
-
-      intel_fb->pf_current_page = (((intel->sarea->pf_current_page >>
-				     (intel_fb->pf_planes & 0x2)) & 0x3) + 1) %
-				  intel_fb->pf_num_pages;
-   }
-
-   if (!drmCommandWriteRead(intel->driFd, DRM_I915_VBLANK_SWAP, &swap,
-			    sizeof(swap))) {
-      dPriv->vblSeq = swap.sequence;
-      swap.sequence -= target;
-      *missed_target = swap.sequence > 0 && swap.sequence <= (1 << 23);
-
-      intel_get_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT)->vbl_pending =
-	 intel_get_renderbuffer(&intel_fb->Base,
-				BUFFER_FRONT_LEFT)->vbl_pending =
-	 dPriv->vblSeq;
-
-      if (swap.seqtype & DRM_VBLANK_FLIP) {
-	 intel_flip_renderbuffers(intel_fb);
-	 intel_draw_buffer(&intel->ctx, intel->ctx.DrawBuffer);
-      }
-
-      ret = GL_TRUE;
-   } else {
-      if (swap.seqtype & DRM_VBLANK_FLIP) {
-	 intel_fb->pf_current_page = ((intel->sarea->pf_current_page >>
-					(intel_fb->pf_planes & 0x2)) & 0x3) %
-				     intel_fb->pf_num_pages;
-      }
-
-      ret = GL_FALSE;
-   }
-
-   UNLOCK_HARDWARE(intel);
-
-   return ret;
-}
-  
-void
-intelSwapBuffers(__DRIdrawablePrivate * dPriv)
-{
-   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
-
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      GET_CURRENT_CONTEXT(ctx);
-      struct intel_context *intel;
-
-      if (ctx == NULL)
-	 return;
-
-      intel = intel_context(ctx);
-
-      if (ctx->Visual.doubleBufferMode) {
-	 GLboolean missed_target;
-	 struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
-	 int64_t ust;
-         
-	 _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
-
-         if (!intelScheduleSwap(dPriv, &missed_target)) {
-	    driWaitForVBlank(dPriv, &missed_target);
-
-	    /*
-	     * Update each buffer's vbl_pending so we don't get too out of
-	     * sync
-	     */
-	    intel_get_renderbuffer(&intel_fb->Base,
-				   BUFFER_BACK_LEFT)->vbl_pending = 
-		    intel_get_renderbuffer(&intel_fb->Base,
-					   BUFFER_FRONT_LEFT)->vbl_pending =
-		    dPriv->vblSeq;
-	    if (!intelPageFlip(dPriv)) {
-	       intelCopyBuffer(dPriv, NULL);
-	    }
-	 }
-
-	 intel_fb->swap_count++;
-	 (*psp->systemTime->getUST) (&ust);
-	 if (missed_target) {
-	    intel_fb->swap_missed_count++;
-	    intel_fb->swap_missed_ust = ust - intel_fb->swap_ust;
-	 }
-
-	 intel_fb->swap_ust = ust;
-      }
-      drmCommandNone(intel->driFd, DRM_I915_GEM_THROTTLE);
-
-   }
-   else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
-void
-intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h)
-{
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      struct intel_context *intel =
-         (struct intel_context *) dPriv->driContextPriv->driverPrivate;
-      GLcontext *ctx = &intel->ctx;
-
-      if (ctx->Visual.doubleBufferMode) {
-         drm_clip_rect_t rect;
-         rect.x1 = x + dPriv->x;
-         rect.y1 = (dPriv->h - y - h) + dPriv->y;
-         rect.x2 = rect.x1 + w;
-         rect.y2 = rect.y1 + h;
-         _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
-         intelCopyBuffer(dPriv, &rect);
-      }
-   }
-   else {
-      /* XXX this shouldn't be an error but we can't handle it for now */
-      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
-   }
-}
-
 
 /**
  * Update the hardware state for drawing into a window or framebuffer object.
@@ -818,7 +154,7 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       return;
    }
 
-   /* Do this here, note core Mesa, since this function is called from
+   /* Do this here, not core Mesa, since this function is called from
     * many places within the driver.
     */
    if (ctx->NewState & (_NEW_BUFFERS | _NEW_COLOR | _NEW_PIXEL)) {
@@ -836,9 +172,6 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       return;
    }
 
-   if (fb->Name)
-      intel_validate_paired_depth_stencil(ctx, fb);
-
    /*
     * How many color buffers are we drawing into?
     */
@@ -846,7 +179,8 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       /* writing to 0  */
       colorRegions[0] = NULL;
       intel->constant_cliprect = GL_TRUE;
-   } else if (fb->_NumColorDrawBuffers > 1) {
+   }
+   else if (fb->_NumColorDrawBuffers > 1) {
        int i;
        struct intel_renderbuffer *irb;
 
@@ -885,14 +219,6 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       }
    }
 
-   /* Update culling direction which changes depending on the
-    * orientation of the buffer:
-    */
-   if (ctx->Driver.FrontFace)
-      ctx->Driver.FrontFace(ctx, ctx->Polygon.FrontFace);
-   else
-      ctx->NewState |= _NEW_POLYGON;
-
    if (!colorRegions[0]) {
       FALLBACK(intel, INTEL_FALLBACK_DRAW_BUFFER, GL_TRUE);
    }
@@ -924,20 +250,13 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
    /***
     *** Stencil buffer
     *** This can only be hardware accelerated if we're using a
-    *** combined DEPTH_STENCIL buffer (for now anyway).
+    *** combined DEPTH_STENCIL buffer.
     ***/
    if (fb->_StencilBuffer && fb->_StencilBuffer->Wrapped) {
       irbStencil = intel_renderbuffer(fb->_StencilBuffer->Wrapped);
       if (irbStencil && irbStencil->region) {
          ASSERT(irbStencil->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
          FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
-         /* need to re-compute stencil hw state */
-	 if (ctx->Driver.Enable != NULL)
-	    ctx->Driver.Enable(ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled);
-	 else
-	    ctx->NewState |= _NEW_STENCIL;
-         if (!depthRegion)
-            depthRegion = irbStencil->region;
       }
       else {
          FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_TRUE);
@@ -946,37 +265,30 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
    else {
       /* XXX FBO: instead of FALSE, pass ctx->Stencil.Enabled ??? */
       FALLBACK(intel, INTEL_FALLBACK_STENCIL_BUFFER, GL_FALSE);
-      /* need to re-compute stencil hw state */
-      if (ctx->Driver.Enable != NULL)
-	 ctx->Driver.Enable(ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled);
-      else
-	 ctx->NewState |= _NEW_STENCIL;
    }
 
    /*
-    * Update depth test state
+    * Update depth and stencil test state
     */
    if (ctx->Driver.Enable) {
-      if (ctx->Depth.Test && fb->Visual.depthBits > 0) {
-	 ctx->Driver.Enable(ctx, GL_DEPTH_TEST, GL_TRUE);
-      } else {
-	 ctx->Driver.Enable(ctx, GL_DEPTH_TEST, GL_FALSE);
-      }
-   } else {
-      ctx->NewState |= _NEW_DEPTH;
+      ctx->Driver.Enable(ctx, GL_DEPTH_TEST,
+                         (ctx->Depth.Test && fb->Visual.depthBits > 0));
+      ctx->Driver.Enable(ctx, GL_STENCIL_TEST,
+                         (ctx->Stencil.Enabled && fb->Visual.stencilBits > 0));
+   }
+   else {
+      ctx->NewState |= (_NEW_DEPTH | _NEW_STENCIL);
    }
 
    intel->vtbl.set_draw_region(intel, colorRegions, depthRegion, 
-	fb->_NumColorDrawBuffers);
+                               fb->_NumColorDrawBuffers);
 
    /* update viewport since it depends on window size */
-   if (ctx->Driver.Viewport) {
-      ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
-			   ctx->Viewport.Width, ctx->Viewport.Height);
-   } else {
-      ctx->NewState |= _NEW_VIEWPORT;
-   }
-
+#ifdef I915
+   intelCalcViewport(ctx);
+#else
+   ctx->NewState |= _NEW_VIEWPORT;
+#endif
    /* Set state we know depends on drawable parameters:
     */
    if (ctx->Driver.Scissor)
@@ -988,6 +300,14 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       ctx->Driver.DepthRange(ctx,
 			     ctx->Viewport.Near,
 			     ctx->Viewport.Far);
+
+   /* Update culling direction which changes depending on the
+    * orientation of the buffer:
+    */
+   if (ctx->Driver.FrontFace)
+      ctx->Driver.FrontFace(ctx, ctx->Polygon.FrontFace);
+   else
+      ctx->NewState |= _NEW_POLYGON;
 }
 
 
@@ -1018,7 +338,6 @@ intelReadBuffer(GLcontext * ctx, GLenum mode)
 void
 intelInitBufferFuncs(struct dd_function_table *functions)
 {
-   functions->Clear = intelClear;
    functions->DrawBuffer = intelDrawBuffer;
    functions->ReadBuffer = intelReadBuffer;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.h b/src/mesa/drivers/dri/intel/intel_buffers.h
index e5afb37dd1a..6069d38e9eb 100644
--- a/src/mesa/drivers/dri/intel/intel_buffers.h
+++ b/src/mesa/drivers/dri/intel/intel_buffers.h
@@ -45,12 +45,6 @@ extern struct intel_region *intel_readbuf_region(struct intel_context *intel);
 
 extern struct intel_region *intel_drawbuf_region(struct intel_context *intel);
 
-extern void intel_wait_flips(struct intel_context *intel);
-
-extern void intelSwapBuffers(__DRIdrawablePrivate * dPriv);
-
-extern void intelWindowMoved(struct intel_context *intel);
-
 extern void intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb);
 
 extern void intelInitBufferFuncs(struct dd_function_table *functions);
@@ -59,5 +53,8 @@ void intel_get_cliprects(struct intel_context *intel,
 			 struct drm_clip_rect **cliprects,
 			 unsigned int *num_cliprects,
 			 int *x_off, int *y_off);
+#ifdef I915
+void intelCalcViewport(GLcontext * ctx);
+#endif
 
 #endif /* INTEL_BUFFERS_H */
diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c
new file mode 100644
index 00000000000..b2291363164
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_clear.c
@@ -0,0 +1,376 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2009 Intel Corporation.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/image.h"
+#include "main/mtypes.h"
+#include "main/attrib.h"
+#include "main/blend.h"
+#include "main/bufferobj.h"
+#include "main/buffers.h"
+#include "main/depth.h"
+#include "main/enable.h"
+#include "main/macros.h"
+#include "main/matrix.h"
+#include "main/texstate.h"
+#include "main/shaders.h"
+#include "main/stencil.h"
+#include "main/varray.h"
+#include "glapi/dispatch.h"
+#include "swrast/swrast.h"
+
+#include "intel_context.h"
+#include "intel_blit.h"
+#include "intel_chipset.h"
+#include "intel_clear.h"
+#include "intel_fbo.h"
+#include "intel_pixel.h"
+
+#define FILE_DEBUG_FLAG DEBUG_BLIT
+
+/**
+ * Perform glClear where mask contains only color, depth, and/or stencil.
+ *
+ * The implementation is based on calling into Mesa to set GL state and
+ * performing normal triangle rendering.  The intent of this path is to
+ * have as generic a path as possible, so that any driver could make use of
+ * it.
+ */
+void
+intel_clear_tris(GLcontext *ctx, GLbitfield mask)
+{
+   struct intel_context *intel = intel_context(ctx);
+   GLfloat vertices[4][3];
+   GLfloat color[4][4];
+   GLfloat dst_z;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   int i;
+   GLboolean saved_fp_enable = GL_FALSE, saved_vp_enable = GL_FALSE;
+   GLboolean saved_shader_program = 0;
+   unsigned int saved_active_texture;
+
+   assert((mask & ~(BUFFER_BIT_BACK_LEFT | BUFFER_BIT_FRONT_LEFT |
+		    BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL)) == 0);
+
+   _mesa_PushAttrib(GL_COLOR_BUFFER_BIT |
+		    GL_CURRENT_BIT |
+		    GL_DEPTH_BUFFER_BIT |
+		    GL_ENABLE_BIT |
+		    GL_STENCIL_BUFFER_BIT |
+		    GL_TRANSFORM_BIT |
+		    GL_CURRENT_BIT);
+   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
+   saved_active_texture = ctx->Texture.CurrentUnit;
+
+   /* Disable existing GL state we don't want to apply to a clear. */
+   _mesa_Disable(GL_ALPHA_TEST);
+   _mesa_Disable(GL_BLEND);
+   _mesa_Disable(GL_CULL_FACE);
+   _mesa_Disable(GL_FOG);
+   _mesa_Disable(GL_POLYGON_SMOOTH);
+   _mesa_Disable(GL_POLYGON_STIPPLE);
+   _mesa_Disable(GL_POLYGON_OFFSET_FILL);
+   _mesa_Disable(GL_LIGHTING);
+   _mesa_Disable(GL_CLIP_PLANE0);
+   _mesa_Disable(GL_CLIP_PLANE1);
+   _mesa_Disable(GL_CLIP_PLANE2);
+   _mesa_Disable(GL_CLIP_PLANE3);
+   _mesa_Disable(GL_CLIP_PLANE4);
+   _mesa_Disable(GL_CLIP_PLANE5);
+   if (ctx->Extensions.ARB_fragment_program && ctx->FragmentProgram.Enabled) {
+      saved_fp_enable = GL_TRUE;
+      _mesa_Disable(GL_FRAGMENT_PROGRAM_ARB);
+   }
+   if (ctx->Extensions.ARB_vertex_program && ctx->VertexProgram.Enabled) {
+      saved_vp_enable = GL_TRUE;
+      _mesa_Disable(GL_VERTEX_PROGRAM_ARB);
+   }
+   if (ctx->Extensions.ARB_shader_objects && ctx->Shader.CurrentProgram) {
+      saved_shader_program = ctx->Shader.CurrentProgram->Name;
+      _mesa_UseProgramObjectARB(0);
+   }
+
+   if (ctx->Texture._EnabledUnits != 0) {
+      int i;
+
+      for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
+	 _mesa_ActiveTextureARB(GL_TEXTURE0 + i);
+	 _mesa_Disable(GL_TEXTURE_1D);
+	 _mesa_Disable(GL_TEXTURE_2D);
+	 _mesa_Disable(GL_TEXTURE_3D);
+	 if (ctx->Extensions.ARB_texture_cube_map)
+	    _mesa_Disable(GL_TEXTURE_CUBE_MAP_ARB);
+	 if (ctx->Extensions.NV_texture_rectangle)
+	    _mesa_Disable(GL_TEXTURE_RECTANGLE_NV);
+	 if (ctx->Extensions.MESA_texture_array) {
+	    _mesa_Disable(GL_TEXTURE_1D_ARRAY_EXT);
+	    _mesa_Disable(GL_TEXTURE_2D_ARRAY_EXT);
+	 }
+      }
+   }
+
+   intel_meta_set_passthrough_transform(intel);
+
+   for (i = 0; i < 4; i++) {
+      color[i][0] = ctx->Color.ClearColor[0];
+      color[i][1] = ctx->Color.ClearColor[1];
+      color[i][2] = ctx->Color.ClearColor[2];
+      color[i][3] = ctx->Color.ClearColor[3];
+   }
+
+   /* convert clear Z from [0,1] to NDC coord in [-1,1] */
+   dst_z = -1.0 + 2.0 * ctx->Depth.Clear;
+
+   /* Prepare the vertices, which are the same regardless of which buffer we're
+    * drawing to.
+    */
+   vertices[0][0] = fb->_Xmin;
+   vertices[0][1] = fb->_Ymin;
+   vertices[0][2] = dst_z;
+   vertices[1][0] = fb->_Xmax;
+   vertices[1][1] = fb->_Ymin;
+   vertices[1][2] = dst_z;
+   vertices[2][0] = fb->_Xmax;
+   vertices[2][1] = fb->_Ymax;
+   vertices[2][2] = dst_z;
+   vertices[3][0] = fb->_Xmin;
+   vertices[3][1] = fb->_Ymax;
+   vertices[3][2] = dst_z;
+
+   _mesa_ColorPointer(4, GL_FLOAT, 4 * sizeof(GLfloat), &color);
+   _mesa_VertexPointer(3, GL_FLOAT, 3 * sizeof(GLfloat), &vertices);
+   _mesa_Enable(GL_COLOR_ARRAY);
+   _mesa_Enable(GL_VERTEX_ARRAY);
+
+   while (mask != 0) {
+      GLuint this_mask = 0;
+
+      if (mask & BUFFER_BIT_BACK_LEFT)
+	 this_mask = BUFFER_BIT_BACK_LEFT;
+      else if (mask & BUFFER_BIT_FRONT_LEFT)
+	 this_mask = BUFFER_BIT_FRONT_LEFT;
+
+      /* Clear depth/stencil in the same pass as color. */
+      this_mask |= (mask & (BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL));
+
+      /* Select the current color buffer and use the color write mask if
+       * we have one, otherwise don't write any color channels.
+       */
+      if (this_mask & BUFFER_BIT_FRONT_LEFT)
+	 _mesa_DrawBuffer(GL_FRONT_LEFT);
+      else if (this_mask & BUFFER_BIT_BACK_LEFT)
+	 _mesa_DrawBuffer(GL_BACK_LEFT);
+      else
+	 _mesa_ColorMask(GL_FALSE, GL_FALSE, GL_FALSE, GL_FALSE);
+
+      /* Control writing of the depth clear value to depth. */
+      if (this_mask & BUFFER_BIT_DEPTH) {
+	 _mesa_DepthFunc(GL_ALWAYS);
+	 _mesa_Enable(GL_DEPTH_TEST);
+      } else {
+	 _mesa_Disable(GL_DEPTH_TEST);
+	 _mesa_DepthMask(GL_FALSE);
+      }
+
+      /* Control writing of the stencil clear value to stencil. */
+      if (this_mask & BUFFER_BIT_STENCIL) {
+	 _mesa_Enable(GL_STENCIL_TEST);
+	 _mesa_StencilOp(GL_REPLACE, GL_REPLACE, GL_REPLACE);
+	 _mesa_StencilFuncSeparate(GL_FRONT, GL_ALWAYS, ctx->Stencil.Clear,
+				   ctx->Stencil.WriteMask[0]);
+      } else {
+	 _mesa_Disable(GL_STENCIL_TEST);
+      }
+
+      CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
+
+      mask &= ~this_mask;
+   }
+
+   intel_meta_restore_transform(intel);
+
+   _mesa_ActiveTextureARB(GL_TEXTURE0 + saved_active_texture);
+   if (saved_fp_enable)
+      _mesa_Enable(GL_FRAGMENT_PROGRAM_ARB);
+   if (saved_vp_enable)
+      _mesa_Enable(GL_VERTEX_PROGRAM_ARB);
+
+   if (saved_shader_program)
+      _mesa_UseProgramObjectARB(saved_shader_program);
+
+   _mesa_PopClientAttrib();
+   _mesa_PopAttrib();
+}
+
+static const char *buffer_names[] = {
+   [BUFFER_FRONT_LEFT] = "front",
+   [BUFFER_BACK_LEFT] = "back",
+   [BUFFER_FRONT_RIGHT] = "front right",
+   [BUFFER_BACK_RIGHT] = "back right",
+   [BUFFER_AUX0] = "aux0",
+   [BUFFER_AUX1] = "aux1",
+   [BUFFER_AUX2] = "aux2",
+   [BUFFER_AUX3] = "aux3",
+   [BUFFER_DEPTH] = "depth",
+   [BUFFER_STENCIL] = "stencil",
+   [BUFFER_ACCUM] = "accum",
+   [BUFFER_COLOR0] = "color0",
+   [BUFFER_COLOR1] = "color1",
+   [BUFFER_COLOR2] = "color2",
+   [BUFFER_COLOR3] = "color3",
+   [BUFFER_COLOR4] = "color4",
+   [BUFFER_COLOR5] = "color5",
+   [BUFFER_COLOR6] = "color6",
+   [BUFFER_COLOR7] = "color7",
+};
+
+/**
+ * Called by ctx->Driver.Clear.
+ */
+static void
+intelClear(GLcontext *ctx, GLbitfield mask)
+{
+   struct intel_context *intel = intel_context(ctx);
+   const GLuint colorMask = *((GLuint *) & ctx->Color.ColorMask);
+   GLbitfield tri_mask = 0;
+   GLbitfield blit_mask = 0;
+   GLbitfield swrast_mask = 0;
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   GLuint i;
+
+   if (0)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   /* HW color buffers (front, back, aux, generic FBO, etc) */
+   if (colorMask == ~0) {
+      /* clear all R,G,B,A */
+      /* XXX FBO: need to check if colorbuffers are software RBOs! */
+      blit_mask |= (mask & BUFFER_BITS_COLOR);
+   }
+   else {
+      /* glColorMask in effect */
+      tri_mask |= (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT));
+   }
+
+   /* HW stencil */
+   if (mask & BUFFER_BIT_STENCIL) {
+      const struct intel_region *stencilRegion
+         = intel_get_rb_region(fb, BUFFER_STENCIL);
+      if (stencilRegion) {
+         /* have hw stencil */
+         if (IS_965(intel->intelScreen->deviceID) ||
+	     (ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
+	    /* We have to use the 3D engine if we're clearing a partial mask
+	     * of the stencil buffer, or if we're on a 965 which has a tiled
+	     * depth/stencil buffer in a layout we can't blit to.
+	     */
+            tri_mask |= BUFFER_BIT_STENCIL;
+         }
+         else {
+            /* clearing all stencil bits, use blitting */
+            blit_mask |= BUFFER_BIT_STENCIL;
+         }
+      }
+   }
+
+   /* HW depth */
+   if (mask & BUFFER_BIT_DEPTH) {
+      /* clear depth with whatever method is used for stencil (see above) */
+      if (IS_965(intel->intelScreen->deviceID) ||
+	  tri_mask & BUFFER_BIT_STENCIL)
+         tri_mask |= BUFFER_BIT_DEPTH;
+      else
+         blit_mask |= BUFFER_BIT_DEPTH;
+   }
+
+   /* If we're doing a tri pass for depth/stencil, include a likely color
+    * buffer with it.
+    */
+   if (mask & (BUFFER_BIT_DEPTH | BUFFER_BIT_STENCIL)) {
+      tri_mask |= blit_mask & BUFFER_BIT_BACK_LEFT;
+      blit_mask &= ~BUFFER_BIT_BACK_LEFT;
+   }
+
+   /* SW fallback clearing */
+   swrast_mask = mask & ~tri_mask & ~blit_mask;
+
+   for (i = 0; i < BUFFER_COUNT; i++) {
+      GLuint bufBit = 1 << i;
+      if ((blit_mask | tri_mask) & bufBit) {
+         if (!fb->Attachment[i].Renderbuffer->ClassID) {
+            blit_mask &= ~bufBit;
+            tri_mask &= ~bufBit;
+            swrast_mask |= bufBit;
+         }
+      }
+   }
+
+   if (blit_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("blit clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (blit_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      intelClearWithBlit(ctx, blit_mask);
+   }
+
+   if (tri_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("tri clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (tri_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      intel_clear_tris(ctx, tri_mask);
+   }
+
+   if (swrast_mask) {
+      if (INTEL_DEBUG & DEBUG_BLIT) {
+	 DBG("swrast clear:");
+	 for (i = 0; i < BUFFER_COUNT; i++) {
+	    if (swrast_mask & (1 << i))
+	       DBG(" %s", buffer_names[i]);
+	 }
+	 DBG("\n");
+      }
+      _swrast_Clear(ctx, swrast_mask);
+   }
+}
+
+
+void
+intelInitClearFuncs(struct dd_function_table *functions)
+{
+   functions->Clear = intelClear;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_clear.h b/src/mesa/drivers/dri/intel/intel_clear.h
new file mode 100644
index 00000000000..7fd6b310a9d
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_clear.h
@@ -0,0 +1,38 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_CLEAR_H
+#define INTEL_CLEAR_H
+
+struct dd_function_table;
+
+extern void
+intelInitClearFuncs(struct dd_function_table *functions);
+
+
+#endif /* INTEL_CLEAR_H */
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 9ac18e69609..d7ccfa06058 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -28,8 +28,6 @@
 
 #include "main/glheader.h"
 #include "main/context.h"
-#include "main/matrix.h"
-#include "main/simple_list.h"
 #include "main/extensions.h"
 #include "main/framebuffer.h"
 #include "main/imports.h"
@@ -38,66 +36,40 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vertex.h"
-
 #include "drivers/common/driverfuncs.h"
 
-#include "intel_screen.h"
-
 #include "i830_dri.h"
 
 #include "intel_chipset.h"
 #include "intel_buffers.h"
 #include "intel_tex.h"
 #include "intel_batchbuffer.h"
-#include "intel_blit.h"
+#include "intel_clear.h"
+#include "intel_extensions.h"
 #include "intel_pixel.h"
 #include "intel_regions.h"
 #include "intel_buffer_objects.h"
 #include "intel_fbo.h"
 #include "intel_decode.h"
 #include "intel_bufmgr.h"
+#include "intel_screen.h"
+#include "intel_swapbuffers.h"
 
 #include "drirenderbuffer.h"
 #include "vblank.h"
 #include "utils.h"
 #include "xmlpool.h"            /* for symbolic values of enum-type options */
+
+
 #ifndef INTEL_DEBUG
 int INTEL_DEBUG = (0);
 #endif
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_occlusion_query
-#define need_GL_ARB_point_parameters
-#define need_GL_ARB_shader_objects
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
-#define need_GL_ARB_vertex_program
-#define need_GL_ARB_vertex_shader
-#define need_GL_ARB_window_pos
-#define need_GL_EXT_blend_color
-#define need_GL_EXT_blend_equation_separate
-#define need_GL_EXT_blend_func_separate
-#define need_GL_EXT_blend_minmax
-#define need_GL_EXT_cull_vertex
-#define need_GL_EXT_fog_coord
-#define need_GL_EXT_framebuffer_object
-#define need_GL_EXT_multi_draw_arrays
-#define need_GL_EXT_point_parameters
-#define need_GL_EXT_secondary_color
-#define need_GL_ATI_separate_stencil
-#define need_GL_NV_point_sprite
-#define need_GL_NV_vertex_program
-#define need_GL_VERSION_2_0
-#define need_GL_VERSION_2_1
-
-#include "extension_helper.h"
-
-#define DRIVER_DATE                     "20080716"
+
+#define DRIVER_DATE                     "20090114"
 #define DRIVER_DATE_GEM                 "GEM " DRIVER_DATE
 
+
 static const GLubyte *
 intelGetString(GLcontext * ctx, GLenum name)
 {
@@ -282,6 +254,9 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
 	   return;
        }
 
+       if (rb == NULL)
+	  continue;
+
        if (rb->region) {
 	  dri_bo_flink(rb->region->buffer, &name);
 	  if (name == buffers[i].name)
@@ -317,11 +292,13 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
    driUpdateFramebufferSize(&intel->ctx, drawable);
 }
 
-static void
+void
 intel_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
     struct intel_context *intel = intel_context(ctx);
     __DRIcontext *driContext = intel->driContext;
+    void (*old_viewport)(GLcontext *ctx, GLint x, GLint y,
+			 GLsizei w, GLsizei h);
 
     if (!driContext->driScreenPriv->dri2.enabled)
 	return;
@@ -330,118 +307,14 @@ intel_viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
     if (driContext->driDrawablePriv != driContext->driReadablePriv)
 	intel_update_renderbuffers(driContext, driContext->driReadablePriv);
 
+    old_viewport = ctx->Driver.Viewport;
     ctx->Driver.Viewport = NULL;
     intel->driDrawable = driContext->driDrawablePriv;
     intelWindowMoved(intel);
     intel_draw_buffer(ctx, intel->ctx.DrawBuffer);
-    ctx->Driver.Viewport = intel_viewport;
+    ctx->Driver.Viewport = old_viewport;
 }
 
-/**
- * Extension strings exported by the intel driver.
- *
- * Extensions supported by all chips supported by i830_dri, i915_dri, or
- * i965_dri.
- */
-static const struct dri_extension card_extensions[] = {
-   { "GL_ARB_multisample",                GL_ARB_multisample_functions },
-   { "GL_ARB_multitexture",               NULL },
-   { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
-   { "GL_ARB_texture_border_clamp",       NULL },
-   { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
-   { "GL_ARB_texture_cube_map",           NULL },
-   { "GL_ARB_texture_env_add",            NULL },
-   { "GL_ARB_texture_env_combine",        NULL },
-   { "GL_ARB_texture_env_crossbar",       NULL },
-   { "GL_ARB_texture_env_dot3",           NULL },
-   { "GL_ARB_texture_mirrored_repeat",    NULL },
-   { "GL_ARB_texture_rectangle",          NULL },
-   { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
-   { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
-   { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
-   { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
-   { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
-   { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
-   { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
-   { "GL_EXT_blend_logic_op",             NULL },
-   { "GL_EXT_blend_subtract",             NULL },
-   { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
-   { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
-   { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
-   { "GL_EXT_packed_depth_stencil",       NULL },
-   { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
-   { "GL_EXT_stencil_wrap",               NULL },
-   { "GL_EXT_texture_edge_clamp",         NULL },
-   { "GL_EXT_texture_env_combine",        NULL },
-   { "GL_EXT_texture_env_dot3",           NULL },
-   { "GL_EXT_texture_filter_anisotropic", NULL },
-   { "GL_EXT_texture_lod_bias",           NULL },
-   { "GL_3DFX_texture_compression_FXT1",  NULL },
-   { "GL_APPLE_client_storage",           NULL },
-   { "GL_MESA_pack_invert",               NULL },
-   { "GL_MESA_ycbcr_texture",             NULL },
-   { "GL_NV_blend_square",                NULL },
-   { "GL_NV_point_sprite",                GL_NV_point_sprite_functions },
-   { "GL_NV_vertex_program",              GL_NV_vertex_program_functions },
-   { "GL_NV_vertex_program1_1",           NULL },
-   { "GL_SGIS_generate_mipmap",           NULL },
-   { NULL, NULL }
-};
-
-static const struct dri_extension brw_extensions[] = {
-   { "GL_ARB_depth_texture",              NULL },
-   { "GL_ARB_draw_buffers",               NULL },
-   { "GL_ARB_fragment_program",           NULL },
-   { "GL_ARB_fragment_program_shadow",    NULL },
-   { "GL_ARB_fragment_shader",            NULL },
-   { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
-   { "GL_ARB_point_sprite", 		  NULL },
-   { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
-   { "GL_ARB_shading_language_100",       GL_VERSION_2_0_functions },
-#if 0
-   /* Support for GLSL 1.20 is currently broken in core Mesa.
-    */
-   { "GL_ARB_shading_language_120",       GL_VERSION_2_1_functions },
-#endif
-   { "GL_ARB_shadow",                     NULL },
-   { "GL_ARB_texture_non_power_of_two",   NULL },
-   { "GL_ARB_vertex_shader",              GL_ARB_vertex_shader_functions },
-   { "GL_EXT_shadow_funcs",               NULL },
-   { "GL_EXT_texture_sRGB",		  NULL },
-   { "GL_ATI_separate_stencil",           GL_ATI_separate_stencil_functions },
-   { NULL,                                NULL }
-};
-
-static const struct dri_extension arb_oq_extensions[] = {
-   { NULL, NULL }
-};
-
-static const struct dri_extension ttm_extensions[] = {
-   { "GL_ARB_pixel_buffer_object",        NULL },
-   { "GL_EXT_framebuffer_object",         GL_EXT_framebuffer_object_functions },
-   { NULL, NULL }
-};
-
-/**
- * Initializes potential list of extensions if ctx == NULL, or actually enables
- * extensions for a context.
- */
-void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
-{
-   struct intel_context *intel = ctx?intel_context(ctx):NULL;
-
-   /* Disable imaging extension until convolution is working in teximage paths.
-    */
-   enable_imaging = GL_FALSE;
-
-   driInitExtensions(ctx, card_extensions, enable_imaging);
-
-   if (intel == NULL || intel->ttm)
-      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
-
-   if (intel == NULL || IS_965(intel->intelScreen->deviceID))
-      driInitExtensions(ctx, brw_extensions, GL_FALSE);
-}
 
 static const struct dri_debug_control debug_control[] = {
    { "tex",   DEBUG_TEXTURE},
@@ -492,9 +365,8 @@ intelInvalidateState(GLcontext * ctx, GLuint new_state)
       intel->vtbl.invalidate_state( intel, new_state );
 }
 
-
-void
-intelFlush(GLcontext * ctx)
+static void
+intel_flush(GLcontext *ctx, GLboolean needs_mi_flush)
 {
    struct intel_context *intel = intel_context(ctx);
 
@@ -508,13 +380,26 @@ intelFlush(GLcontext * ctx)
     * lands onscreen in a timely manner, even if the X Server doesn't trigger
     * a flush for us.
     */
-   intel_batchbuffer_emit_mi_flush(intel->batch);
+   if (needs_mi_flush)
+      intel_batchbuffer_emit_mi_flush(intel->batch);
 
    if (intel->batch->map != intel->batch->ptr)
       intel_batchbuffer_flush(intel->batch);
 }
 
 void
+intelFlush(GLcontext * ctx)
+{
+   intel_flush(ctx, GL_FALSE);
+}
+
+static void
+intel_glFlush(GLcontext *ctx)
+{
+   intel_flush(ctx, GL_TRUE);
+}
+
+void
 intelFinish(GLcontext * ctx)
 {
    struct gl_framebuffer *fb = ctx->DrawBuffer;
@@ -540,11 +425,10 @@ intelInitDriverFunctions(struct dd_function_table *functions)
 {
    _mesa_init_driver_functions(functions);
 
-   functions->Flush = intelFlush;
+   functions->Flush = intel_glFlush;
    functions->Finish = intelFinish;
    functions->GetString = intelGetString;
    functions->UpdateState = intelInvalidateState;
-   functions->Viewport = intel_viewport;
 
    functions->CopyColorTable = _swrast_CopyColorTable;
    functions->CopyColorSubTable = _swrast_CopyColorSubTable;
@@ -553,6 +437,7 @@ intelInitDriverFunctions(struct dd_function_table *functions)
 
    intelInitTextureFuncs(functions);
    intelInitStateFuncs(functions);
+   intelInitClearFuncs(functions);
    intelInitBufferFuncs(functions);
    intelInitPixelFuncs(functions);
 }
@@ -773,7 +658,6 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
 
       intel_region_release(&intel->front_region);
       intel_region_release(&intel->back_region);
-      intel_region_release(&intel->third_region);
       intel_region_release(&intel->depth_region);
 
       driDestroyOptionCache(&intel->optionCache);
@@ -823,12 +707,7 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 	    intel_renderbuffer_set_region(intel_fb->color_rb[1],
 					  intel->back_region);
          }
-#if 0
-         if (intel_fb->color_rb[2]) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[2],
-					  intel->third_region);
-         }
-#endif
+
          if (irbDepth) {
 	    intel_renderbuffer_set_region(irbDepth, intel->depth_region);
          }
@@ -861,11 +740,16 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 		  ? driGetDefaultVBlankFlags(&intel->optionCache)
 		 : VBLANK_FLAG_NO_IRQ;
 
+	       /* Prevent error printf if one crtc is disabled, this will
+		* be properly calculated in intelWindowMoved() next.
+		*/
+		driDrawPriv->vblFlags = intelFixupVblank(intel, driDrawPriv);
+
 	       (*psp->systemTime->getUST) (&intel_fb->swap_ust);
 	       driDrawableInitVBlank(driDrawPriv);
 	       intel_fb->vbl_waited = driDrawPriv->vblSeq;
 
-	       for (i = 0; i < (intel->intelScreen->third.handle ? 3 : 2); i++) {
+	       for (i = 0; i < 2; i++) {
 		  if (intel_fb->color_rb[i])
 		     intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
 	       }
@@ -889,7 +773,7 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
 {
    __DRIdrawablePrivate *dPriv = intel->driDrawable;
    __DRIscreenPrivate *sPriv = intel->driScreen;
-   volatile struct drm_i915_sarea *sarea = intel->sarea;
+   volatile drm_i915_sarea_t *sarea = intel->sarea;
    int me = intel->hHWContext;
 
    drmGetLock(intel->driFd, intel->hHWContext, flags);
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index 3938af4c72c..18dc43c4a42 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -157,6 +157,20 @@ struct intel_context
       void (*debug_batch)(struct intel_context *intel);
    } vtbl;
 
+   struct {
+      struct gl_fragment_program *bitmap_fp;
+      struct gl_vertex_program *passthrough_vp;
+
+      struct gl_fragment_program *saved_fp;
+      GLboolean saved_fp_enable;
+      struct gl_vertex_program *saved_vp;
+      GLboolean saved_vp_enable;
+
+      GLint saved_vp_x, saved_vp_y;
+      GLsizei saved_vp_width, saved_vp_height;
+      GLenum saved_matrix_mode;
+   } meta;
+
    GLint refcount;
    GLuint Fallback;
    GLuint NewGLState;
@@ -166,7 +180,6 @@ struct intel_context
 
    struct intel_region *front_region;
    struct intel_region *back_region;
-   struct intel_region *third_region;
    struct intel_region *depth_region;
 
    /**
@@ -184,6 +197,7 @@ struct intel_context
       GLuint id;
       uint32_t primitive;	/**< Current hardware primitive type */
       void (*flush) (struct intel_context *);
+      GLubyte *start_ptr; /**< for i8xx */
       dri_bo *vb_bo;
       uint8_t *vb;
       unsigned int start_offset; /**< Byte offset of primitive sequence */
@@ -196,7 +210,6 @@ struct intel_context
    char *prevLockFile;
    int prevLockLine;
 
-   GLubyte clear_chan[4];
    GLuint ClearColor565;
    GLuint ClearColor8888;
 
@@ -424,7 +437,6 @@ extern void intelFinish(GLcontext * ctx);
 extern void intelFlush(GLcontext * ctx);
 
 extern void intelInitDriverFunctions(struct dd_function_table *functions);
-extern void intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging);
 
 
 /* ================================================================
@@ -495,6 +507,9 @@ extern int intel_translate_stencil_op(GLenum op);
 extern int intel_translate_blend_factor(GLenum factor);
 extern int intel_translate_logic_op(GLenum opcode);
 
+void intel_viewport(GLcontext * ctx, GLint x, GLint y,
+		    GLsizei width, GLsizei height);
+
 void intel_update_renderbuffers(__DRIcontext *context,
 				__DRIdrawable *drawable);
 
diff --git a/src/mesa/drivers/dri/intel/intel_decode.c b/src/mesa/drivers/dri/intel/intel_decode.c
index 9c105013c05..136221c37fc 100644
--- a/src/mesa/drivers/dri/intel/intel_decode.c
+++ b/src/mesa/drivers/dri/intel/intel_decode.c
@@ -37,6 +37,7 @@
 
 #include <stdio.h>
 #include <stdarg.h>
+#include <string.h>
 #include <inttypes.h>
 
 #include "intel_decode.h"
@@ -86,27 +87,28 @@ decode_mi(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 
     struct {
 	uint32_t opcode;
+	int len_mask;
 	int min_len;
 	int max_len;
 	char *name;
     } opcodes_mi[] = {
-	{ 0x08, 1, 1, "MI_ARB_ON_OFF" },
-	{ 0x0a, 1, 1, "MI_BATCH_BUFFER_END" },
-	{ 0x31, 2, 2, "MI_BATCH_BUFFER_START" },
-	{ 0x14, 3, 3, "MI_DISPLAY_BUFFER_INFO" },
-	{ 0x04, 1, 1, "MI_FLUSH" },
-	{ 0x22, 3, 3, "MI_LOAD_REGISTER_IMM" },
-	{ 0x13, 2, 2, "MI_LOAD_SCAN_LINES_EXCL" },
-	{ 0x12, 2, 2, "MI_LOAD_SCAN_LINES_INCL" },
-	{ 0x00, 1, 1, "MI_NOOP" },
-	{ 0x11, 2, 2, "MI_OVERLAY_FLIP" },
-	{ 0x07, 1, 1, "MI_REPORT_HEAD" },
-	{ 0x18, 2, 2, "MI_SET_CONTEXT" },
-	{ 0x20, 3, 4, "MI_STORE_DATA_IMM" },
-	{ 0x21, 3, 4, "MI_STORE_DATA_INDEX" },
-	{ 0x24, 3, 3, "MI_STORE_REGISTER_MEM" },
-	{ 0x02, 1, 1, "MI_USER_INTERRUPT" },
-	{ 0x03, 1, 1, "MI_WAIT_FOR_EVENT" },
+	{ 0x08, 0, 1, 1, "MI_ARB_ON_OFF" },
+	{ 0x0a, 0, 1, 1, "MI_BATCH_BUFFER_END" },
+	{ 0x31, 0x3f, 2, 2, "MI_BATCH_BUFFER_START" },
+	{ 0x14, 0x3f, 3, 3, "MI_DISPLAY_BUFFER_INFO" },
+	{ 0x04, 0, 1, 1, "MI_FLUSH" },
+	{ 0x22, 0, 3, 3, "MI_LOAD_REGISTER_IMM" },
+	{ 0x13, 0x3f, 2, 2, "MI_LOAD_SCAN_LINES_EXCL" },
+	{ 0x12, 0x3f, 2, 2, "MI_LOAD_SCAN_LINES_INCL" },
+	{ 0x00, 0, 1, 1, "MI_NOOP" },
+	{ 0x11, 0x3f, 2, 2, "MI_OVERLAY_FLIP" },
+	{ 0x07, 0, 1, 1, "MI_REPORT_HEAD" },
+	{ 0x18, 0x3f, 2, 2, "MI_SET_CONTEXT" },
+	{ 0x20, 0x3f, 3, 4, "MI_STORE_DATA_IMM" },
+	{ 0x21, 0x3f, 3, 4, "MI_STORE_DATA_INDEX" },
+	{ 0x24, 0x3f, 3, 3, "MI_STORE_REGISTER_MEM" },
+	{ 0x02, 0, 1, 1, "MI_USER_INTERRUPT" },
+	{ 0x03, 0, 1, 1, "MI_WAIT_FOR_EVENT" },
     };
 
 
@@ -117,12 +119,14 @@ decode_mi(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 
 	    instr_out(data, hw_offset, 0, "%s\n", opcodes_mi[opcode].name);
 	    if (opcodes_mi[opcode].max_len > 1) {
-		len = (data[0] & 0x000000ff) + 2;
+		len = (data[0] & opcodes_mi[opcode].len_mask) + 2;
 		if (len < opcodes_mi[opcode].min_len ||
 		    len > opcodes_mi[opcode].max_len)
 		{
-		    fprintf(out, "Bad length in %s\n",
-			    opcodes_mi[opcode].name);
+		    fprintf(out, "Bad length (%d) in %s, [%d, %d]\n",
+			    len, opcodes_mi[opcode].name,
+			    opcodes_mi[opcode].min_len,
+			    opcodes_mi[opcode].max_len);
 		}
 	    }
 
@@ -308,6 +312,15 @@ decode_3d_1c(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     case 0x10:
 	instr_out(data, hw_offset, 0, "3DSTATE_SCISSOR_ENABLE\n");
 	return 1;
+    case 0x01:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_COORD_SET_I830\n");
+	return 1;
+    case 0x0a:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_CUBE_I830\n");
+	return 1;
+    case 0x05:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_TEX_STREAM_I830\n");
+	return 1;
     }
 
     instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
@@ -315,34 +328,508 @@ decode_3d_1c(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     return 1;
 }
 
+/** Sets the string dstname to describe the destination of the PS instruction */
+static void
+i915_get_instruction_dst(uint32_t *data, int i, char *dstname, int do_mask)
+{
+    uint32_t a0 = data[i];
+    int dst_nr = (a0 >> 14) & 0xf;
+    char dstmask[8];
+    char *sat;
+
+    if (do_mask) {
+	if (((a0 >> 10) & 0xf) == 0xf) {
+	    dstmask[0] = 0;
+	} else {
+	    int dstmask_index = 0;
+
+	    dstmask[dstmask_index++] = '.';
+	    if (a0 & (1 << 10))
+		dstmask[dstmask_index++] = 'x';
+	    if (a0 & (1 << 11))
+		dstmask[dstmask_index++] = 'y';
+	    if (a0 & (1 << 12))
+		dstmask[dstmask_index++] = 'z';
+	    if (a0 & (1 << 13))
+		dstmask[dstmask_index++] = 'w';
+	    dstmask[dstmask_index++] = 0;
+	}
+
+	if (a0 & (1 << 22))
+	    sat = ".sat";
+	else
+	    sat = "";
+    } else {
+	dstmask[0] = 0;
+	sat = "";
+    }
+
+    switch ((a0 >> 19) & 0x7) {
+    case 0:
+	if (dst_nr > 15)
+	    fprintf(out, "bad destination reg R%d\n", dst_nr);
+	sprintf(dstname, "R%d%s%s", dst_nr, dstmask, sat);
+	break;
+    case 4:
+	if (dst_nr > 0)
+	    fprintf(out, "bad destination reg oC%d\n", dst_nr);
+	sprintf(dstname, "oC%s%s", dstmask, sat);
+	break;
+    case 5:
+	if (dst_nr > 0)
+	    fprintf(out, "bad destination reg oD%d\n", dst_nr);
+	sprintf(dstname, "oD%s%s",  dstmask, sat);
+	break;
+    case 6:
+	if (dst_nr > 2)
+	    fprintf(out, "bad destination reg U%d\n", dst_nr);
+	sprintf(dstname, "U%d%s%s", dst_nr, dstmask, sat);
+	break;
+    default:
+	sprintf(dstname, "RESERVED");
+	break;
+    }
+}
+
+static char *
+i915_get_channel_swizzle(uint32_t select)
+{
+    switch (select & 0x7) {
+    case 0:
+	return (select & 8) ? "-x" : "x";
+    case 1:
+	return (select & 8) ? "-y" : "y";
+    case 2:
+	return (select & 8) ? "-z" : "z";
+    case 3:
+	return (select & 8) ? "-w" : "w";
+    case 4:
+	return (select & 8) ? "-0" : "0";
+    case 5:
+	return (select & 8) ? "-1" : "1";
+    default:
+	return (select & 8) ? "-bad" : "bad";
+    }
+}
+
+static void
+i915_get_instruction_src_name(uint32_t src_type, uint32_t src_nr, char *name)
+{
+    switch (src_type) {
+    case 0:
+	sprintf(name, "R%d", src_nr);
+	if (src_nr > 15)
+	    fprintf(out, "bad src reg %s\n", name);
+	break;
+    case 1:
+	if (src_nr < 8)
+	    sprintf(name, "T%d", src_nr);
+	else if (src_nr == 8)
+	    sprintf(name, "DIFFUSE");
+	else if (src_nr == 9)
+	    sprintf(name, "SPECULAR");
+	else if (src_nr == 10)
+	    sprintf(name, "FOG");
+	else {
+	    fprintf(out, "bad src reg T%d\n", src_nr);
+	    sprintf(name, "RESERVED");
+	}
+	break;
+    case 2:
+	sprintf(name, "C%d", src_nr);
+	if (src_nr > 31)
+	    fprintf(out, "bad src reg %s\n", name);
+	break;
+    case 4:
+	sprintf(name, "oC");
+	if (src_nr > 0)
+	    fprintf(out, "bad src reg oC%d\n", src_nr);
+	break;
+    case 5:
+	sprintf(name, "oD");
+	if (src_nr > 0)
+	    fprintf(out, "bad src reg oD%d\n", src_nr);
+	break;
+    case 6:
+	sprintf(name, "U%d", src_nr);
+	if (src_nr > 2)
+	    fprintf(out, "bad src reg %s\n", name);
+	break;
+    default:
+	fprintf(out, "bad src reg type %d\n", src_type);
+	sprintf(name, "RESERVED");
+	break;
+    }
+}
+
+static void
+i915_get_instruction_src0(uint32_t *data, int i, char *srcname)
+{
+    uint32_t a0 = data[i];
+    uint32_t a1 = data[i + 1];
+    int src_nr = (a0 >> 2) & 0x1f;
+    char *swizzle_x = i915_get_channel_swizzle((a1 >> 28) & 0xf);
+    char *swizzle_y = i915_get_channel_swizzle((a1 >> 24) & 0xf);
+    char *swizzle_z = i915_get_channel_swizzle((a1 >> 20) & 0xf);
+    char *swizzle_w = i915_get_channel_swizzle((a1 >> 16) & 0xf);
+    char swizzle[100];
+
+    i915_get_instruction_src_name((a0 >> 7) & 0x7, src_nr, srcname);
+    sprintf(swizzle, ".%s%s%s%s", swizzle_x, swizzle_y, swizzle_z, swizzle_w);
+    if (strcmp(swizzle, ".xyzw") != 0)
+	strcat(srcname, swizzle);
+}
+
+static void
+i915_get_instruction_src1(uint32_t *data, int i, char *srcname)
+{
+    uint32_t a1 = data[i + 1];
+    uint32_t a2 = data[i + 2];
+    int src_nr = (a1 >> 8) & 0x1f;
+    char *swizzle_x = i915_get_channel_swizzle((a1 >> 4) & 0xf);
+    char *swizzle_y = i915_get_channel_swizzle((a1 >> 0) & 0xf);
+    char *swizzle_z = i915_get_channel_swizzle((a2 >> 28) & 0xf);
+    char *swizzle_w = i915_get_channel_swizzle((a2 >> 24) & 0xf);
+    char swizzle[100];
+
+    i915_get_instruction_src_name((a1 >> 13) & 0x7, src_nr, srcname);
+    sprintf(swizzle, ".%s%s%s%s", swizzle_x, swizzle_y, swizzle_z, swizzle_w);
+    if (strcmp(swizzle, ".xyzw") != 0)
+	strcat(srcname, swizzle);
+}
+
+static void
+i915_get_instruction_src2(uint32_t *data, int i, char *srcname)
+{
+    uint32_t a2 = data[i + 2];
+    int src_nr = (a2 >> 16) & 0x1f;
+    char *swizzle_x = i915_get_channel_swizzle((a2 >> 12) & 0xf);
+    char *swizzle_y = i915_get_channel_swizzle((a2 >> 8) & 0xf);
+    char *swizzle_z = i915_get_channel_swizzle((a2 >> 4) & 0xf);
+    char *swizzle_w = i915_get_channel_swizzle((a2 >> 0) & 0xf);
+    char swizzle[100];
+
+    i915_get_instruction_src_name((a2 >> 21) & 0x7, src_nr, srcname);
+    sprintf(swizzle, ".%s%s%s%s", swizzle_x, swizzle_y, swizzle_z, swizzle_w);
+    if (strcmp(swizzle, ".xyzw") != 0)
+	strcat(srcname, swizzle);
+}
+
+static void
+i915_get_instruction_addr(uint32_t src_type, uint32_t src_nr, char *name)
+{
+    switch (src_type) {
+    case 0:
+	sprintf(name, "R%d", src_nr);
+	if (src_nr > 15)
+	    fprintf(out, "bad src reg %s\n", name);
+	break;
+    case 1:
+	if (src_nr < 8)
+	    sprintf(name, "T%d", src_nr);
+	else if (src_nr == 8)
+	    sprintf(name, "DIFFUSE");
+	else if (src_nr == 9)
+	    sprintf(name, "SPECULAR");
+	else if (src_nr == 10)
+	    sprintf(name, "FOG");
+	else {
+	    fprintf(out, "bad src reg T%d\n", src_nr);
+	    sprintf(name, "RESERVED");
+	}
+	break;
+    case 4:
+	sprintf(name, "oC");
+	if (src_nr > 0)
+	    fprintf(out, "bad src reg oC%d\n", src_nr);
+	break;
+    case 5:
+	sprintf(name, "oD");
+	if (src_nr > 0)
+	    fprintf(out, "bad src reg oD%d\n", src_nr);
+	break;
+    default:
+	fprintf(out, "bad src reg type %d\n", src_type);
+	sprintf(name, "RESERVED");
+	break;
+    }
+}
+
+static void
+i915_decode_alu1(uint32_t *data, uint32_t hw_offset,
+		 int i, char *instr_prefix, char *op_name)
+{
+    char dst[100], src0[100];
+
+    i915_get_instruction_dst(data, i, dst, 1);
+    i915_get_instruction_src0(data, i, src0);
+
+    instr_out(data, hw_offset, i++, "%s: %s %s, %s\n", instr_prefix,
+	      op_name, dst, src0);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+}
+
+static void
+i915_decode_alu2(uint32_t *data, uint32_t hw_offset,
+		 int i, char *instr_prefix, char *op_name)
+{
+    char dst[100], src0[100], src1[100];
+
+    i915_get_instruction_dst(data, i, dst, 1);
+    i915_get_instruction_src0(data, i, src0);
+    i915_get_instruction_src1(data, i, src1);
+
+    instr_out(data, hw_offset, i++, "%s: %s %s, %s, %s\n", instr_prefix,
+	      op_name, dst, src0, src1);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+}
+
+static void
+i915_decode_alu3(uint32_t *data, uint32_t hw_offset,
+		 int i, char *instr_prefix, char *op_name)
+{
+    char dst[100], src0[100], src1[100], src2[100];
+
+    i915_get_instruction_dst(data, i, dst, 1);
+    i915_get_instruction_src0(data, i, src0);
+    i915_get_instruction_src1(data, i, src1);
+    i915_get_instruction_src2(data, i, src2);
+
+    instr_out(data, hw_offset, i++, "%s: %s %s, %s, %s, %s\n", instr_prefix,
+	      op_name, dst, src0, src1, src2);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+}
+
+static void
+i915_decode_tex(uint32_t *data, uint32_t hw_offset, int i, char *instr_prefix,
+		char *tex_name)
+{
+    uint32_t t0 = data[i];
+    uint32_t t1 = data[i + 1];
+    char dst_name[100];
+    char addr_name[100];
+    int sampler_nr;
+
+    i915_get_instruction_dst(data, i, dst_name, 0);
+    i915_get_instruction_addr((t1 >> 24) & 0x7,
+			      (t1 >> 17) & 0xf,
+			      addr_name);
+    sampler_nr = t0 & 0xf;
+
+    instr_out(data, hw_offset, i++, "%s: %s %s, S%d, %s\n", instr_prefix,
+	      tex_name, dst_name, sampler_nr, addr_name);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+    instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+}
+
+static void
+i915_decode_dcl(uint32_t *data, uint32_t hw_offset, int i, char *instr_prefix)
+{
+    uint32_t d0 = data[i];
+    char *sampletype;
+    int dcl_nr = (d0 >> 14) & 0xf;
+    char *dcl_x = d0 & (1 << 10) ? "x" : "";
+    char *dcl_y = d0 & (1 << 11) ? "y" : "";
+    char *dcl_z = d0 & (1 << 12) ? "z" : "";
+    char *dcl_w = d0 & (1 << 13) ? "w" : "";
+    char dcl_mask[10];
+
+    switch ((d0 >> 19) & 0x3) {
+    case 1:
+	sprintf(dcl_mask, ".%s%s%s%s", dcl_x, dcl_y, dcl_z, dcl_w);
+	if (strcmp(dcl_mask, ".") == 0)
+	    fprintf(out, "bad (empty) dcl mask\n");
+
+	if (dcl_nr > 10)
+	    fprintf(out, "bad T%d dcl register number\n", dcl_nr);
+	if (dcl_nr < 8) {
+	    if (strcmp(dcl_mask, ".x") != 0 &&
+		strcmp(dcl_mask, ".xy") != 0 &&
+		strcmp(dcl_mask, ".xz") != 0 &&
+		strcmp(dcl_mask, ".w") != 0 &&
+		strcmp(dcl_mask, ".xyzw") != 0) {
+		fprintf(out, "bad T%d.%s dcl mask\n", dcl_nr, dcl_mask);
+	    }
+	    instr_out(data, hw_offset, i++, "%s: DCL T%d%s\n", instr_prefix,
+		      dcl_nr, dcl_mask);
+	} else {
+	    if (strcmp(dcl_mask, ".xz") == 0)
+		fprintf(out, "errataed bad dcl mask %s\n", dcl_mask);
+	    else if (strcmp(dcl_mask, ".xw") == 0)
+		fprintf(out, "errataed bad dcl mask %s\n", dcl_mask);
+	    else if (strcmp(dcl_mask, ".xzw") == 0)
+		fprintf(out, "errataed bad dcl mask %s\n", dcl_mask);
+
+	    if (dcl_nr == 8) {
+		instr_out(data, hw_offset, i++, "%s: DCL DIFFUSE%s\n", instr_prefix,
+			  dcl_mask);
+	    } else if (dcl_nr == 9) {
+		instr_out(data, hw_offset, i++, "%s: DCL SPECULAR%s\n", instr_prefix,
+			  dcl_mask);
+	    } else if (dcl_nr == 10) {
+		instr_out(data, hw_offset, i++, "%s: DCL FOG%s\n", instr_prefix,
+			  dcl_mask);
+	    }
+	}
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	break;
+    case 3:
+	switch ((d0 >> 22) & 0x3) {
+	case 0:
+	    sampletype = "2D";
+	    break;
+	case 1:
+	    sampletype = "CUBE";
+	    break;
+	case 2:
+	    sampletype = "3D";
+	    break;
+	default:
+	    sampletype = "RESERVED";
+	    break;
+	}
+	if (dcl_nr > 15)
+	    fprintf(out, "bad S%d dcl register number\n", dcl_nr);
+	instr_out(data, hw_offset, i++, "%s: DCL S%d %s\n", instr_prefix,
+		  dcl_nr, sampletype);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	break;
+    default:
+	instr_out(data, hw_offset, i++, "%s: DCL RESERVED%d\n", instr_prefix, dcl_nr);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+    }
+}
+
+static void
+i915_decode_instruction(uint32_t *data, uint32_t hw_offset,
+			int i, char *instr_prefix)
+{
+    switch ((data[i] >> 24) & 0x1f) {
+    case 0x0:
+	instr_out(data, hw_offset, i++, "%s: NOP\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	break;
+    case 0x01:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "ADD");
+	break;
+    case 0x02:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "MOV");
+	break;
+    case 0x03:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "MUL");
+	break;
+    case 0x04:
+	i915_decode_alu3(data, hw_offset, i, instr_prefix, "MAD");
+	break;
+    case 0x05:
+	i915_decode_alu3(data, hw_offset, i, instr_prefix, "DP2ADD");
+	break;
+    case 0x06:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "DP3");
+	break;
+    case 0x07:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "DP4");
+	break;
+    case 0x08:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "FRC");
+	break;
+    case 0x09:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "RCP");
+	break;
+    case 0x0a:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "RSQ");
+	break;
+    case 0x0b:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "EXP");
+	break;
+    case 0x0c:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "LOG");
+	break;
+    case 0x0d:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "CMP");
+	break;
+    case 0x0e:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "MIN");
+	break;
+    case 0x0f:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "MAX");
+	break;
+    case 0x10:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "FLR");
+	break;
+    case 0x11:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "MOD");
+	break;
+    case 0x12:
+	i915_decode_alu1(data, hw_offset, i, instr_prefix, "TRC");
+	break;
+    case 0x13:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "SGE");
+	break;
+    case 0x14:
+	i915_decode_alu2(data, hw_offset, i, instr_prefix, "SLT");
+	break;
+    case 0x15:
+	i915_decode_tex(data, hw_offset, i, instr_prefix, "TEXLD");
+	break;
+    case 0x16:
+	i915_decode_tex(data, hw_offset, i, instr_prefix, "TEXLDP");
+	break;
+    case 0x17:
+	i915_decode_tex(data, hw_offset, i, instr_prefix, "TEXLDB");
+	break;
+    case 0x19:
+	i915_decode_dcl(data, hw_offset, i, instr_prefix);
+	break;
+    default:
+	instr_out(data, hw_offset, i++, "%s: unknown\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	instr_out(data, hw_offset, i++, "%s\n", instr_prefix);
+	break;
+    }
+}
+
 static int
-decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures, int i830)
 {
     unsigned int len, i, c, opcode, word, map, sampler, instr;
 
     struct {
 	uint32_t opcode;
+	int i830_only;
 	int min_len;
 	int max_len;
 	char *name;
     } opcodes_3d_1d[] = {
-	{ 0x8e, 3, 3, "3DSTATE_BUFFER_INFO" },
-	{ 0x86, 4, 4, "3DSTATE_CHROMA_KEY" },
-	{ 0x9c, 1, 1, "3DSTATE_CLEAR_PARAMETERS" },
-	{ 0x88, 2, 2, "3DSTATE_CONSTANT_BLEND_COLOR" },
-	{ 0x99, 2, 2, "3DSTATE_DEFAULT_DIFFUSE" },
-	{ 0x9a, 2, 2, "3DSTATE_DEFAULT_SPECULAR" },
-	{ 0x98, 2, 2, "3DSTATE_DEFAULT_Z" },
-	{ 0x97, 2, 2, "3DSTATE_DEPTH_OFFSET_SCALE" },
-	{ 0x85, 2, 2, "3DSTATE_DEST_BUFFER_VARIABLES" },
-	{ 0x80, 5, 5, "3DSTATE_DRAWING_RECTANGLE" },
-	{ 0x8e, 3, 3, "3DSTATE_BUFFER_INFO" },
-	{ 0x9d, 65, 65, "3DSTATE_FILTER_COEFFICIENTS_4X4" },
-	{ 0x9e, 4, 4, "3DSTATE_MONO_FILTER" },
-	{ 0x89, 4, 4, "3DSTATE_FOG_MODE" },
-	{ 0x8f, 2, 16, "3DSTATE_MAP_PALLETE_LOAD_32" },
-	{ 0x81, 3, 3, "3DSTATE_SCISSOR_RECTANGLE" },
-	{ 0x83, 2, 2, "3DSTATE_SPAN_STIPPLE" },
+	{ 0x8e, 0, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x86, 0, 4, 4, "3DSTATE_CHROMA_KEY" },
+	{ 0x9c, 0, 1, 1, "3DSTATE_CLEAR_PARAMETERS" },
+	{ 0x88, 0, 2, 2, "3DSTATE_CONSTANT_BLEND_COLOR" },
+	{ 0x99, 0, 2, 2, "3DSTATE_DEFAULT_DIFFUSE" },
+	{ 0x9a, 0, 2, 2, "3DSTATE_DEFAULT_SPECULAR" },
+	{ 0x98, 0, 2, 2, "3DSTATE_DEFAULT_Z" },
+	{ 0x97, 0, 2, 2, "3DSTATE_DEPTH_OFFSET_SCALE" },
+	{ 0x85, 0, 2, 2, "3DSTATE_DEST_BUFFER_VARIABLES" },
+	{ 0x80, 0, 5, 5, "3DSTATE_DRAWING_RECTANGLE" },
+	{ 0x8e, 0, 3, 3, "3DSTATE_BUFFER_INFO" },
+	{ 0x9d, 0, 65, 65, "3DSTATE_FILTER_COEFFICIENTS_4X4" },
+	{ 0x9e, 0, 4, 4, "3DSTATE_MONO_FILTER" },
+	{ 0x89, 0, 4, 4, "3DSTATE_FOG_MODE" },
+	{ 0x8f, 0, 2, 16, "3DSTATE_MAP_PALLETE_LOAD_32" },
+	{ 0x81, 0, 3, 3, "3DSTATE_SCISSOR_RECTANGLE" },
+	{ 0x83, 0, 2, 2, "3DSTATE_SPAN_STIPPLE" },
+	{ 0x8c, 1, 2, 2, "3DSTATE_MAP_COORD_TRANSFORM_I830" },
+	{ 0x8b, 1, 2, 2, "3DSTATE_MAP_VERTEX_TRANSFORM_I830" },
+	{ 0x8d, 1, 3, 3, "3DSTATE_W_STATE_I830" },
+	{ 0x01, 1, 2, 2, "3DSTATE_COLOR_FACTOR_I830" },
+	{ 0x02, 1, 2, 2, "3DSTATE_MAP_COORD_SETBIND_I830" },
     };
 
     switch ((data[0] & 0x00ff0000) >> 16) {
@@ -426,8 +913,9 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     case 0x00:
 	instr_out(data, hw_offset, 0, "3DSTATE_MAP_STATE\n");
 	len = (data[0] & 0x0000003f) + 2;
+	instr_out(data, hw_offset, 1, "mask\n");
 
-	i = 1;
+	i = 2;
 	for (map = 0; map <= 15; map++) {
 	    if (data[1] & (1 << map)) {
 		if (i + 3 >= count)
@@ -447,7 +935,7 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_CONSTANTS\n");
 	len = (data[0] & 0x000000ff) + 2;
 
-	i = 1;
+	i = 2;
 	for (c = 0; c <= 31; c++) {
 	    if (data[1] & (1 << c)) {
 		if (i + 4 >= count)
@@ -467,7 +955,7 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 	    }
 	}
 	if (len != i) {
-	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    fprintf(out, "Bad count in 3DSTATE_PIXEL_SHADER_CONSTANTS\n");
 	    (*failures)++;
 	}
 	return len;
@@ -480,17 +968,22 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 	}
 	i = 1;
 	for (instr = 0; instr < (len - 1) / 3; instr++) {
+	    char instr_prefix[10];
+
 	    if (i + 3 >= count)
-		BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
-	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
-	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
-	    instr_out(data, hw_offset, i++, "PS%03x\n", instr);
+		BUFFER_FAIL(count, len, "3DSTATE_PIXEL_SHADER_PROGRAM");
+	    sprintf(instr_prefix, "PS%03d", instr);
+	    i915_decode_instruction(data, hw_offset, i, instr_prefix);
+	    i += 3;
 	}
 	return len;
     case 0x01:
+	if (i830)
+	    break;
 	instr_out(data, hw_offset, 0, "3DSTATE_SAMPLER_STATE\n");
+	instr_out(data, hw_offset, 1, "mask\n");
 	len = (data[0] & 0x0000003f) + 2;
-	i = 1;
+	i = 2;
 	for (sampler = 0; sampler <= 15; sampler++) {
 	    if (data[1] & (1 << sampler)) {
 		if (i + 3 >= count)
@@ -513,6 +1006,9 @@ decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     for (opcode = 0; opcode < sizeof(opcodes_3d_1d) / sizeof(opcodes_3d_1d[0]);
 	 opcode++)
     {
+	if (opcodes_3d_1d[opcode].i830_only && !i830)
+	    continue;
+
 	if (((data[0] & 0x00ff0000) >> 16) == opcodes_3d_1d[opcode].opcode) {
 	    len = 1;
 
@@ -756,7 +1252,7 @@ decode_3d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     case 0x1f:
 	return decode_3d_primitive(data, count, hw_offset, failures);
     case 0x1d:
-	return decode_3d_1d(data, count, hw_offset, failures);
+	return decode_3d_1d(data, count, hw_offset, failures, 0);
     case 0x1c:
 	return decode_3d_1c(data, count, hw_offset, failures);
     }
@@ -816,10 +1312,71 @@ get_965_depthformat(unsigned int depthformat)
     }
 }
 
+static const char *
+get_965_element_component(uint32_t data, int component)
+{
+    uint32_t component_control = (data >> (16 + (3 - component) * 4)) & 0x7;
+
+    switch (component_control) {
+    case 0:
+	return "nostore";
+    case 1:
+	switch (component) {
+	case 0: return "X";
+	case 1: return "Y";
+	case 2: return "Z";
+	case 3: return "W";
+	default: return "fail";
+	}
+    case 2:
+	return "0.0";
+    case 3:
+	return "1.0";
+    case 4:
+	return "0x1";
+    case 5:
+	return "VID";
+    default:
+	return "fail";
+    }
+}
+
+static const char *
+get_965_prim_type(uint32_t data)
+{
+    uint32_t primtype = (data >> 10) & 0x1f;
+
+    switch (primtype) {
+    case 0x01: return "point list";
+    case 0x02: return "line list";
+    case 0x03: return "line strip";
+    case 0x04: return "tri list";
+    case 0x05: return "tri strip";
+    case 0x06: return "tri fan";
+    case 0x07: return "quad list";
+    case 0x08: return "quad strip";
+    case 0x09: return "line list adj";
+    case 0x0a: return "line strip adj";
+    case 0x0b: return "tri list adj";
+    case 0x0c: return "tri strip adj";
+    case 0x0d: return "tri strip reverse";
+    case 0x0e: return "polygon";
+    case 0x0f: return "rect list";
+    case 0x10: return "line loop";
+    case 0x11: return "point list bf";
+    case 0x12: return "line strip cont";
+    case 0x13: return "line strip bf";
+    case 0x14: return "line strip cont bf";
+    case 0x15: return "tri fan no stipple";
+    default: return "fail";
+    }
+}
+
 static int
 decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 {
     unsigned int opcode, len;
+    int i;
 
     struct {
 	uint32_t opcode;
@@ -840,8 +1397,7 @@ decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 	{ 0x780b, 1, 1, "3DSTATE_VF_STATISTICS" },
 	{ 0x7808, 5, 257, "3DSTATE_VERTEX_BUFFERS" },
 	{ 0x7809, 3, 256, "3DSTATE_VERTEX_ELEMENTS" },
-	/* 0x7808: 3DSTATE_VERTEX_BUFFERS */
-	/* 0x7809: 3DSTATE_VERTEX_ELEMENTS */
+	{ 0x780a, 3, 3, "3DSTATE_INDEX_BUFFER" },
 	{ 0x7900, 4, 4, "3DSTATE_DRAWING_RECTANGLE" },
 	{ 0x7901, 5, 5, "3DSTATE_CONSTANT_COLOR" },
 	{ 0x7905, 5, 7, "3DSTATE_DEPTH_BUFFER" },
@@ -927,6 +1483,64 @@ decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 
 	return len;
 
+    case 0x7808:
+	len = (data[0] & 0xff) + 2;
+	if ((len - 1) % 4 != 0)
+	    fprintf(out, "Bad count in 3DSTATE_VERTEX_BUFFERS\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_VERTEX_BUFFERS");
+	instr_out(data, hw_offset, 0, "3DSTATE_VERTEX_BUFFERS\n");
+
+	for (i = 1; i < len;) {
+	    instr_out(data, hw_offset, i, "buffer %d: %s, pitch %db\n",
+		      data[i] >> 27,
+		      data[i] & (1 << 26) ? "random" : "sequential",
+		      data[i] & 0x07ff);
+	    i++;
+	    instr_out(data, hw_offset, i++, "buffer address\n");
+	    instr_out(data, hw_offset, i++, "max index\n");
+	    instr_out(data, hw_offset, i++, "mbz\n");
+	}
+	return len;
+
+    case 0x7809:
+	len = (data[0] & 0xff) + 2;
+	if ((len + 1) % 2 != 0)
+	    fprintf(out, "Bad count in 3DSTATE_VERTEX_ELEMENTS\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_VERTEX_ELEMENTS");
+	instr_out(data, hw_offset, 0, "3DSTATE_VERTEX_ELEMENTS\n");
+
+	for (i = 1; i < len;) {
+	    instr_out(data, hw_offset, i, "buffer %d: %svalid, type 0x%04x, "
+		      "src offset 0x%04xd bytes\n",
+		      data[i] >> 27,
+		      data[i] & (1 << 26) ? "" : "in",
+		      (data[i] >> 16) & 0x1ff,
+		      data[i] & 0x07ff);
+	    i++;
+	    instr_out(data, hw_offset, i, "(%s, %s, %s, %s), "
+		      "dst offset 0x%02x bytes\n",
+		      get_965_element_component(data[i], 0),
+		      get_965_element_component(data[i], 1),
+		      get_965_element_component(data[i], 2),
+		      get_965_element_component(data[i], 3),
+		      (data[i] & 0xff) * 4);
+	    i++;
+	}
+	return len;
+
+    case 0x780a:
+	len = (data[0] & 0xff) + 2;
+	if (len != 3)
+	    fprintf(out, "Bad count in 3DSTATE_INDEX_BUFFER\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DSTATE_INDEX_BUFFER");
+	instr_out(data, hw_offset, 0, "3DSTATE_INDEX_BUFFER\n");
+	instr_out(data, hw_offset, 1, "beginning buffer address\n");
+	instr_out(data, hw_offset, 2, "ending buffer address\n");
+	return len;
+
     case 0x7900:
 	if (len != 4)
 	    fprintf(out, "Bad count in 3DSTATE_DRAWING_RECTANGLE\n");
@@ -948,9 +1562,9 @@ decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 	return len;
 
     case 0x7905:
-	if (len != 5)
+	if (len != 5 && len != 6)
 	    fprintf(out, "Bad count in 3DSTATE_DEPTH_BUFFER\n");
-	if (count < 5)
+	if (count < len)
 	    BUFFER_FAIL(count, len, "3DSTATE_DEPTH_BUFFER");
 
 	instr_out(data, hw_offset, 0,
@@ -965,8 +1579,28 @@ decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
 		  ((data[3] & 0x0007ffc0) >> 6) + 1,
 		  ((data[3] & 0xfff80000) >> 19) + 1);
 	instr_out(data, hw_offset, 4, "volume depth\n");
+	if (len == 6)
+	    instr_out(data, hw_offset, 5, "\n");
 
 	return len;
+
+    case 0x7b00:
+	len = (data[0] & 0xff) + 2;
+	if (len != 6)
+	    fprintf(out, "Bad count in 3DPRIMITIVE\n");
+	if (count < len)
+	    BUFFER_FAIL(count, len, "3DPRIMITIVE");
+
+	instr_out(data, hw_offset, 0,
+		  "3DPRIMITIVE: %s %s\n",
+		  get_965_prim_type(data[0]),
+		  (data[0] & (1 << 15)) ? "random" : "sequential");
+	instr_out(data, hw_offset, 1, "primitive count\n");
+	instr_out(data, hw_offset, 2, "start vertex\n");
+	instr_out(data, hw_offset, 3, "instance count\n");
+	instr_out(data, hw_offset, 4, "start instance\n");
+	instr_out(data, hw_offset, 5, "index bias\n");
+	return len;
     }
 
     for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
@@ -999,6 +1633,73 @@ decode_3d_965(uint32_t *data, int count, uint32_t hw_offset, int *failures)
     return 1;
 }
 
+static int
+decode_3d_i830(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_3d[] = {
+	{ 0x02, 1, 1, "3DSTATE_MODES_3" },
+	{ 0x03, 1, 1, "3DSTATE_ENABLES_1"},
+	{ 0x04, 1, 1, "3DSTATE_ENABLES_2"},
+	{ 0x05, 1, 1, "3DSTATE_VFT0"},
+	{ 0x06, 1, 1, "3DSTATE_AA"},
+	{ 0x07, 1, 1, "3DSTATE_RASTERIZATION_RULES" },
+	{ 0x08, 1, 1, "3DSTATE_MODES_1" },
+	{ 0x09, 1, 1, "3DSTATE_STENCIL_TEST" },
+	{ 0x0a, 1, 1, "3DSTATE_VFT1"},
+	{ 0x0b, 1, 1, "3DSTATE_INDPT_ALPHA_BLEND" },
+	{ 0x0c, 1, 1, "3DSTATE_MODES_5" },
+	{ 0x0d, 1, 1, "3DSTATE_MAP_BLEND_OP" },
+	{ 0x0e, 1, 1, "3DSTATE_MAP_BLEND_ARG" },
+	{ 0x0f, 1, 1, "3DSTATE_MODES_2" },
+	{ 0x15, 1, 1, "3DSTATE_FOG_COLOR" },
+	{ 0x16, 1, 1, "3DSTATE_MODES_4" },
+    };
+
+    switch ((data[0] & 0x1f000000) >> 24) {
+    case 0x1f:
+	return decode_3d_primitive(data, count, hw_offset, failures);
+    case 0x1d:
+	return decode_3d_1d(data, count, hw_offset, failures, 1);
+    case 0x1c:
+	return decode_3d_1c(data, count, hw_offset, failures);
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1f000000) >> 24 == opcodes_3d[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d[opcode].name);
+	    if (opcodes_3d[opcode].max_len > 1) {
+		len = (data[0] & 0xff) + 2;
+		if (len < opcodes_3d[opcode].min_len ||
+		    len > opcodes_3d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_3d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_3d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
 /**
  * Decodes an i830-i915 batch buffer, writing the output to stdout.
  *
@@ -1028,9 +1729,12 @@ intel_decode(uint32_t *data, int count, uint32_t hw_offset, uint32_t devid)
 	    if (IS_965(devid)) {
 		index += decode_3d_965(data + index, count - index,
 				       hw_offset + index * 4, &failures);
-	    } else {
+	    } else if (IS_9XX(devid)) {
 		index += decode_3d(data + index, count - index,
 				   hw_offset + index * 4, &failures);
+	    } else {
+		index += decode_3d_i830(data + index, count - index,
+					hw_offset + index * 4, &failures);
 	    }
 	    break;
 	default:
diff --git a/src/mesa/drivers/dri/intel/intel_depthstencil.c b/src/mesa/drivers/dri/intel/intel_depthstencil.c
deleted file mode 100644
index c2b4d7728b4..00000000000
--- a/src/mesa/drivers/dri/intel/intel_depthstencil.c
+++ /dev/null
@@ -1,252 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "main/glheader.h"
-#include "main/imports.h"
-#include "main/context.h"
-#include "main/depthstencil.h"
-#include "main/fbobject.h"
-#include "main/framebuffer.h"
-#include "main/hash.h"
-#include "main/mtypes.h"
-#include "main/renderbuffer.h"
-
-#include "intel_context.h"
-#include "intel_fbo.h"
-#include "intel_depthstencil.h"
-#include "intel_regions.h"
-#include "intel_span.h"
-
-/**
- * The GL_EXT_framebuffer_object allows the user to create their own
- * framebuffer objects consisting of color renderbuffers (0 or more),
- * depth renderbuffers (0 or 1) and stencil renderbuffers (0 or 1).
- *
- * The spec considers depth and stencil renderbuffers to be totally independent
- * buffers.  In reality, most graphics hardware today uses a combined
- * depth+stencil buffer (one 32-bit pixel = 24 bits of Z + 8 bits of stencil).
- *
- * This causes difficulty because the user may create some number of depth
- * renderbuffers and some number of stencil renderbuffers and bind them
- * together in framebuffers in any combination.
- *
- * This code manages all that.
- *
- * 1. Depth renderbuffers are always allocated in hardware as 32bpp
- *    GL_DEPTH24_STENCIL8 buffers.
- *
- * 2. Stencil renderbuffers are initially allocated in software as 8bpp
- *    GL_STENCIL_INDEX8 buffers.
- *
- * 3. Depth and Stencil renderbuffers use the PairedStencil and PairedDepth
- *    fields (respectively) to indicate if the buffer's currently paired
- *    with another stencil or depth buffer (respectively).
- *
- * 4. When a depth and stencil buffer are initially both attached to the
- *    current framebuffer, we merge the stencil buffer values into the
- *    depth buffer (really a depth+stencil buffer).  The then hardware uses
- *    the combined buffer.
- *
- * 5. Whenever a depth or stencil buffer is reallocated (with
- *    glRenderbufferStorage) we undo the pairing and copy the stencil values
- *    from the combined depth/stencil buffer back to the stencil-only buffer.
- *
- * 6. We also undo the pairing when we find a change in buffer bindings.
- *
- * 7. If a framebuffer is only using a depth renderbuffer (no stencil), we
- *    just use the combined depth/stencil buffer and ignore the stencil values.
- *
- * 8. If a framebuffer is only using a stencil renderbuffer (no depth) we have
- *    to promote the 8bpp software stencil buffer to a 32bpp hardware
- *    depth+stencil buffer.
- *
- */
-
-/**
- * Undo the pairing/interleaving between depth and stencil buffers.
- * irb should be a depth/stencil or stencil renderbuffer.
- */
-void
-intel_unpair_depth_stencil(GLcontext *ctx, struct intel_renderbuffer *irb)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct gl_renderbuffer *rb = &irb->Base;
-
-   if (irb->PairedStencil) {
-      /* irb is a depth/stencil buffer */
-      struct gl_renderbuffer *stencilRb;
-      struct intel_renderbuffer *stencilIrb;
-
-      ASSERT(rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-
-      stencilRb = _mesa_lookup_renderbuffer(ctx, irb->PairedStencil);
-      stencilIrb = intel_renderbuffer(stencilRb);
-      if (stencilIrb) {
-         /* need to extract stencil values from the depth buffer */
-	 ASSERT(stencilIrb->PairedDepth == rb->Name);
-	 intel_renderbuffer_map(intel, rb);
-	 intel_renderbuffer_map(intel, stencilRb);
-	 _mesa_extract_stencil(ctx, rb, stencilRb);
-	 intel_renderbuffer_unmap(intel, stencilRb);
-	 intel_renderbuffer_unmap(intel, rb);
-         stencilIrb->PairedDepth = 0;
-      }
-      irb->PairedStencil = 0;
-   }
-   else if (irb->PairedDepth) {
-      /* irb is a stencil buffer */
-      struct gl_renderbuffer *depthRb;
-      struct intel_renderbuffer *depthIrb;
-
-      ASSERT(rb->_ActualFormat == GL_STENCIL_INDEX8_EXT ||
-             rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-
-      depthRb = _mesa_lookup_renderbuffer(ctx, irb->PairedDepth);
-      depthIrb = intel_renderbuffer(depthRb);
-      if (depthIrb) {
-         /* need to extract stencil values from the depth buffer */
-	 ASSERT(depthIrb->PairedStencil == rb->Name);
-	 intel_renderbuffer_map(intel, rb);
-	 intel_renderbuffer_map(intel, depthRb);
-	 _mesa_extract_stencil(ctx, depthRb, rb);
-	 intel_renderbuffer_unmap(intel, depthRb);
-	 intel_renderbuffer_unmap(intel, rb);
-         depthIrb->PairedStencil = 0;
-      }
-      irb->PairedDepth = 0;
-   }
-   else {
-      _mesa_problem(ctx, "Problem in undo_depth_stencil_pairing");
-   }
-
-   ASSERT(irb->PairedStencil == 0);
-   ASSERT(irb->PairedDepth == 0);
-}
-
-
-/**
- * Examine the depth and stencil renderbuffers which are attached to the
- * framebuffer.  If both depth and stencil are attached, make sure that the
- * renderbuffers are 'paired' (combined).  If only depth or only stencil is
- * attached, undo any previous pairing.
- *
- * Must be called if NewState & _NEW_BUFFER (when renderbuffer attachments
- * change, for example).
- */
-void
-intel_validate_paired_depth_stencil(GLcontext * ctx,
-                                    struct gl_framebuffer *fb)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_renderbuffer *depthRb, *stencilRb;
-
-   depthRb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
-   stencilRb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
-
-   if (depthRb && stencilRb) {
-      if (depthRb == stencilRb) {
-         /* Using a user-created combined depth/stencil buffer.
-          * Nothing to do.
-          */
-         ASSERT(depthRb->Base._BaseFormat == GL_DEPTH_STENCIL_EXT);
-         ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-      }
-      else {
-         /* Separate depth/stencil buffers, need to interleave now */
-         ASSERT(depthRb->Base._BaseFormat == GL_DEPTH_COMPONENT);
-         ASSERT(stencilRb->Base._BaseFormat == GL_STENCIL_INDEX);
-         /* may need to interleave depth/stencil now */
-         if (depthRb->PairedStencil == stencilRb->Base.Name) {
-            /* OK, the depth and stencil buffers are already interleaved */
-            ASSERT(stencilRb->PairedDepth == depthRb->Base.Name);
-         }
-         else {
-            /* need to setup new pairing/interleaving */
-            if (depthRb->PairedStencil) {
-               intel_unpair_depth_stencil(ctx, depthRb);
-            }
-            if (stencilRb->PairedDepth) {
-               intel_unpair_depth_stencil(ctx, stencilRb);
-            }
-
-            ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-            ASSERT(stencilRb->Base._ActualFormat == GL_STENCIL_INDEX8_EXT ||
-                   stencilRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-
-            /* establish new pairing: interleave stencil into depth buffer */
-	    intel_renderbuffer_map(intel, &depthRb->Base);
-	    intel_renderbuffer_map(intel, &stencilRb->Base);
-            _mesa_insert_stencil(ctx, &depthRb->Base, &stencilRb->Base);
-	    intel_renderbuffer_unmap(intel, &stencilRb->Base);
-	    intel_renderbuffer_unmap(intel, &depthRb->Base);
-            depthRb->PairedStencil = stencilRb->Base.Name;
-            stencilRb->PairedDepth = depthRb->Base.Name;
-         }
-
-      }
-   }
-   else if (depthRb) {
-      /* Depth buffer but no stencil buffer.
-       * We'll use a GL_DEPTH24_STENCIL8 buffer and ignore the stencil bits.
-       */
-      /* can't assert this until storage is allocated:
-         ASSERT(depthRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-       */
-      /* intel_undo any previous pairing */
-      if (depthRb->PairedStencil) {
-         intel_unpair_depth_stencil(ctx, depthRb);
-      }
-   }
-   else if (stencilRb) {
-      /* Stencil buffer but no depth buffer.
-       * Since h/w doesn't typically support just 8bpp stencil w/out Z,
-       * we'll use a GL_DEPTH24_STENCIL8 buffer and ignore the depth bits.
-       */
-      /* undo any previous pairing */
-      if (stencilRb->PairedDepth) {
-         intel_unpair_depth_stencil(ctx, stencilRb);
-      }
-      if (stencilRb->Base._ActualFormat == GL_STENCIL_INDEX8_EXT) {
-         /* promote buffer to GL_DEPTH24_STENCIL8 for hw rendering */
-         _mesa_promote_stencil(ctx, &stencilRb->Base);
-         ASSERT(stencilRb->Base._ActualFormat == GL_DEPTH24_STENCIL8_EXT);
-      }
-   }
-
-   /* Finally, update the fb->_DepthBuffer and fb->_StencilBuffer fields */
-   _mesa_update_depth_buffer(ctx, fb, BUFFER_DEPTH);
-   if (depthRb && depthRb->PairedStencil)
-      _mesa_update_stencil_buffer(ctx, fb, BUFFER_DEPTH);
-   else
-      _mesa_update_stencil_buffer(ctx, fb, BUFFER_STENCIL);
-
-
-   /* The hardware should use fb->Attachment[BUFFER_DEPTH].Renderbuffer
-    * first, if present, then fb->Attachment[BUFFER_STENCIL].Renderbuffer
-    * if present.
-    */
-}
diff --git a/src/mesa/drivers/dri/intel/intel_depthstencil.h b/src/mesa/drivers/dri/intel/intel_depthstencil.h
deleted file mode 100644
index 740eb0d9895..00000000000
--- a/src/mesa/drivers/dri/intel/intel_depthstencil.h
+++ /dev/null
@@ -1,15 +0,0 @@
-
-#ifndef INTEL_DEPTH_STENCIL_H
-#define INTEL_DEPTH_STENCIL_H
-
-#include "intel_fbo.h"
-
-extern void
-intel_unpair_depth_stencil(GLcontext * ctx, struct intel_renderbuffer *irb);
-
-extern void
-intel_validate_paired_depth_stencil(GLcontext * ctx,
-                                    struct gl_framebuffer *fb);
-
-
-#endif /* INTEL_DEPTH_STENCIL_H */
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.c b/src/mesa/drivers/dri/intel/intel_extensions.c
new file mode 100644
index 00000000000..28223ca1415
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_extensions.c
@@ -0,0 +1,189 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "intel_chipset.h"
+#include "intel_context.h"
+#include "intel_extensions.h"
+
+
+#define need_GL_ARB_framebuffer_object
+#define need_GL_ARB_occlusion_query
+#define need_GL_ARB_point_parameters
+#define need_GL_ARB_shader_objects
+#define need_GL_ARB_vertex_program
+#define need_GL_ARB_vertex_shader
+#define need_GL_ARB_window_pos
+#define need_GL_EXT_blend_color
+#define need_GL_EXT_blend_equation_separate
+#define need_GL_EXT_blend_func_separate
+#define need_GL_EXT_blend_minmax
+#define need_GL_EXT_cull_vertex
+#define need_GL_EXT_fog_coord
+#define need_GL_EXT_framebuffer_object
+#define need_GL_EXT_framebuffer_blit
+#define need_GL_EXT_point_parameters
+#define need_GL_EXT_secondary_color
+#define need_GL_EXT_stencil_two_side
+#define need_GL_ATI_separate_stencil
+#define need_GL_NV_point_sprite
+#define need_GL_NV_vertex_program
+#define need_GL_VERSION_2_0
+#define need_GL_VERSION_2_1
+
+#include "extension_helper.h"
+
+
+/**
+ * Extension strings exported by the intel driver.
+ *
+ * Extensions supported by all chips supported by i830_dri, i915_dri, or
+ * i965_dri.
+ */
+static const struct dri_extension card_extensions[] = {
+   { "GL_ARB_multitexture",               NULL },
+   { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
+   { "GL_ARB_texture_border_clamp",       NULL },
+   { "GL_ARB_texture_cube_map",           NULL },
+   { "GL_ARB_texture_env_add",            NULL },
+   { "GL_ARB_texture_env_combine",        NULL },
+   { "GL_ARB_texture_env_crossbar",       NULL },
+   { "GL_ARB_texture_env_dot3",           NULL },
+   { "GL_ARB_texture_mirrored_repeat",    NULL },
+   { "GL_ARB_texture_rectangle",          NULL },
+   { "GL_ARB_vertex_program",             GL_ARB_vertex_program_functions },
+   { "GL_ARB_window_pos",                 GL_ARB_window_pos_functions },
+   { "GL_EXT_blend_color",                GL_EXT_blend_color_functions },
+   { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
+   { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
+   { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
+   { "GL_EXT_blend_logic_op",             NULL },
+   { "GL_EXT_blend_subtract",             NULL },
+   { "GL_EXT_cull_vertex",                GL_EXT_cull_vertex_functions },
+   { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
+   { "GL_EXT_packed_depth_stencil",       NULL },
+   { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
+   { "GL_EXT_stencil_wrap",               NULL },
+   { "GL_EXT_texture_edge_clamp",         NULL },
+   { "GL_EXT_texture_env_combine",        NULL },
+   { "GL_EXT_texture_env_dot3",           NULL },
+   { "GL_EXT_texture_filter_anisotropic", NULL },
+   { "GL_EXT_texture_lod_bias",           NULL },
+   { "GL_3DFX_texture_compression_FXT1",  NULL },
+   { "GL_APPLE_client_storage",           NULL },
+   { "GL_MESA_pack_invert",               NULL },
+   { "GL_MESA_ycbcr_texture",             NULL },
+   { "GL_NV_blend_square",                NULL },
+   { "GL_NV_point_sprite",                GL_NV_point_sprite_functions },
+   { "GL_NV_vertex_program",              GL_NV_vertex_program_functions },
+   { "GL_NV_vertex_program1_1",           NULL },
+   { "GL_SGIS_generate_mipmap",           NULL },
+   { NULL, NULL }
+};
+
+
+/** i915 / i945-only extensions */
+static const struct dri_extension i915_extensions[] = {
+   { "GL_ARB_depth_texture",              NULL },
+   { "GL_ARB_fragment_program",           NULL },
+   { "GL_ARB_shadow",                     NULL },
+   { "GL_ARB_texture_non_power_of_two",   NULL },
+   { "GL_ATI_texture_env_combine3",       NULL },
+   { "GL_EXT_shadow_funcs",               NULL },
+   { "GL_NV_texture_env_combine4",        NULL },
+   { NULL,                                NULL }
+};
+
+
+/** i965-only extensions */
+static const struct dri_extension brw_extensions[] = {
+   { "GL_ARB_depth_texture",              NULL },
+   { "GL_ARB_fragment_program",           NULL },
+   { "GL_ARB_fragment_program_shadow",    NULL },
+   { "GL_ARB_fragment_shader",            NULL },
+   { "GL_ARB_framebuffer_object",         GL_ARB_framebuffer_object_functions},
+   { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
+   { "GL_ARB_point_sprite", 		  NULL },
+   { "GL_ARB_shader_objects",             GL_ARB_shader_objects_functions },
+   { "GL_ARB_shading_language_100",       GL_VERSION_2_0_functions },
+#if 0
+   /* Support for GLSL 1.20 is currently broken in core Mesa.
+    */
+   { "GL_ARB_shading_language_120",       GL_VERSION_2_1_functions },
+#endif
+   { "GL_ARB_shadow",                     NULL },
+   { "GL_ARB_texture_non_power_of_two",   NULL },
+   { "GL_ARB_vertex_shader",              GL_ARB_vertex_shader_functions },
+   { "GL_EXT_shadow_funcs",               NULL },
+   { "GL_EXT_stencil_two_side",           GL_EXT_stencil_two_side_functions },
+   { "GL_EXT_texture_sRGB",		  NULL },
+   { "GL_EXT_texture_swizzle",		  NULL },
+   { "GL_EXT_vertex_array_bgra",	  NULL },
+   { "GL_ATI_separate_stencil",           GL_ATI_separate_stencil_functions },
+   { "GL_ATI_texture_env_combine3",       NULL },
+   { "GL_NV_texture_env_combine4",        NULL },
+   { NULL,                                NULL }
+};
+
+
+static const struct dri_extension arb_oq_extensions[] = {
+   { NULL, NULL }
+};
+
+
+static const struct dri_extension ttm_extensions[] = {
+   { "GL_ARB_pixel_buffer_object",      NULL },
+   { "GL_EXT_framebuffer_blit",         GL_EXT_framebuffer_blit_functions },
+   { "GL_EXT_framebuffer_object",       GL_EXT_framebuffer_object_functions },
+   { NULL, NULL }
+};
+
+
+/**
+ * Initializes potential list of extensions if ctx == NULL, or actually enables
+ * extensions for a context.
+ */
+void
+intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging)
+{
+   struct intel_context *intel = ctx?intel_context(ctx):NULL;
+
+   /* Disable imaging extension until convolution is working in teximage paths.
+    */
+   enable_imaging = GL_FALSE;
+
+   driInitExtensions(ctx, card_extensions, enable_imaging);
+
+   if (intel == NULL || intel->ttm)
+      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
+
+   if (intel == NULL || IS_965(intel->intelScreen->deviceID))
+      driInitExtensions(ctx, brw_extensions, GL_FALSE);
+
+   if (intel == NULL || IS_915(intel->intelScreen->deviceID)
+       || IS_945(intel->intelScreen->deviceID))
+      driInitExtensions(ctx, i915_extensions, GL_FALSE);
+}
diff --git a/src/mesa/drivers/dri/intel/intel_extensions.h b/src/mesa/drivers/dri/intel/intel_extensions.h
new file mode 100644
index 00000000000..97147ecdb0f
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_extensions.h
@@ -0,0 +1,36 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_EXTENSIONS_H
+#define INTEL_EXTENSIONS_H
+
+
+extern void
+intelInitExtensions(GLcontext *ctx, GLboolean enable_imaging);
+
+
+#endif
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.c b/src/mesa/drivers/dri/intel/intel_fbo.c
index fce5e36b9d1..54d6044ad3b 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.c
+++ b/src/mesa/drivers/dri/intel/intel_fbo.c
@@ -27,6 +27,7 @@
 
 
 #include "main/imports.h"
+#include "main/macros.h"
 #include "main/mtypes.h"
 #include "main/fbobject.h"
 #include "main/framebuffer.h"
@@ -37,95 +38,13 @@
 
 #include "intel_context.h"
 #include "intel_buffers.h"
-#include "intel_depthstencil.h"
 #include "intel_fbo.h"
 #include "intel_mipmap_tree.h"
 #include "intel_regions.h"
-#include "intel_span.h"
 
 
 #define FILE_DEBUG_FLAG DEBUG_FBO
 
-#define INTEL_RB_CLASS 0x12345678
-
-
-/* XXX FBO: move this to intel_context.h (inlined) */
-/**
- * Return a gl_renderbuffer ptr casted to intel_renderbuffer.
- * NULL will be returned if the rb isn't really an intel_renderbuffer.
- * This is determiend by checking the ClassID.
- */
-struct intel_renderbuffer *
-intel_renderbuffer(struct gl_renderbuffer *rb)
-{
-   struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
-   if (irb && irb->Base.ClassID == INTEL_RB_CLASS) {
-      /*_mesa_warning(NULL, "Returning non-intel Rb\n");*/
-      return irb;
-   }
-   else
-      return NULL;
-}
-
-
-struct intel_renderbuffer *
-intel_get_renderbuffer(struct gl_framebuffer *fb, int attIndex)
-{
-   if (attIndex >= 0)
-      return intel_renderbuffer(fb->Attachment[attIndex].Renderbuffer);
-   else
-      return NULL;
-}
-
-
-void
-intel_flip_renderbuffers(struct intel_framebuffer *intel_fb)
-{
-   int current_page = intel_fb->pf_current_page;
-   int next_page = (current_page + 1) % intel_fb->pf_num_pages;
-   struct gl_renderbuffer *tmp_rb;
-
-   /* Exchange renderbuffers if necessary but make sure their reference counts
-    * are preserved.
-    */
-   if (intel_fb->color_rb[current_page] &&
-       intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer !=
-       &intel_fb->color_rb[current_page]->Base) {
-      tmp_rb = NULL;
-      _mesa_reference_renderbuffer(&tmp_rb,
-	 intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer);
-      tmp_rb = &intel_fb->color_rb[current_page]->Base;
-      _mesa_reference_renderbuffer(
-	 &intel_fb->Base.Attachment[BUFFER_FRONT_LEFT].Renderbuffer, tmp_rb);
-      _mesa_reference_renderbuffer(&tmp_rb, NULL);
-   }
-
-   if (intel_fb->color_rb[next_page] &&
-       intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer !=
-       &intel_fb->color_rb[next_page]->Base) {
-      tmp_rb = NULL;
-      _mesa_reference_renderbuffer(&tmp_rb,
-	 intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer);
-      tmp_rb = &intel_fb->color_rb[next_page]->Base;
-      _mesa_reference_renderbuffer(
-	 &intel_fb->Base.Attachment[BUFFER_BACK_LEFT].Renderbuffer, tmp_rb);
-      _mesa_reference_renderbuffer(&tmp_rb, NULL);
-   }
-}
-
-
-struct intel_region *
-intel_get_rb_region(struct gl_framebuffer *fb, GLuint attIndex)
-{
-   struct intel_renderbuffer *irb = intel_get_renderbuffer(fb, attIndex);
-
-   if (irb)
-      return irb->region;
-   else
-      return NULL;
-}
-
-
 
 /**
  * Create a new framebuffer object.
@@ -140,6 +59,7 @@ intel_new_framebuffer(GLcontext * ctx, GLuint name)
 }
 
 
+/** Called by gl_renderbuffer::Delete() */
 static void
 intel_delete_renderbuffer(struct gl_renderbuffer *rb)
 {
@@ -149,10 +69,6 @@ intel_delete_renderbuffer(struct gl_renderbuffer *rb)
 
    ASSERT(irb);
 
-   if (irb->PairedStencil || irb->PairedDepth) {
-      intel_unpair_depth_stencil(ctx, irb);
-   }
-
    if (irb->span_cache != NULL)
       _mesa_free(irb->span_cache);
 
@@ -164,7 +80,6 @@ intel_delete_renderbuffer(struct gl_renderbuffer *rb)
 }
 
 
-
 /**
  * Return a pointer to a specific pixel in a renderbuffer.
  */
@@ -179,7 +94,6 @@ intel_get_pointer(GLcontext * ctx, struct gl_renderbuffer *rb,
 }
 
 
-
 /**
  * Called via glRenderbufferStorageEXT() to set the format and allocate
  * storage for a user-created renderbuffer.
@@ -310,7 +224,6 @@ intel_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
 }
 
 
-
 /**
  * Called for each hardware renderbuffer when a _window_ is resized.
  * Just update fields.
@@ -328,6 +241,7 @@ intel_alloc_window_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
    return GL_TRUE;
 }
 
+
 static void
 intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
 		     GLuint width, GLuint height)
@@ -344,7 +258,7 @@ intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
    }
 
    /* Make sure all window system renderbuffers are up to date */
-   for (i = 0; i < 3; i++) {
+   for (i = 0; i < 2; i++) {
       struct gl_renderbuffer *rb = &intel_fb->color_rb[i]->Base;
 
       /* only resize if size is changing */
@@ -354,6 +268,8 @@ intel_resize_buffers(GLcontext *ctx, struct gl_framebuffer *fb,
    }
 }
 
+
+/** Dummy function for gl_renderbuffer::AllocStorage() */
 static GLboolean
 intel_nop_alloc_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
                         GLenum internalFormat, GLuint width, GLuint height)
@@ -373,10 +289,9 @@ intel_renderbuffer_set_region(struct intel_renderbuffer *rb,
    rb->region = NULL;
    intel_region_reference(&rb->region, region);
    intel_region_release(&old);
-
-   rb->pfPitch = region->pitch;
 }
 
+
 /**
  * Create a new intel_renderbuffer which corresponds to an on-screen window,
  * not a user-created renderbuffer.
@@ -496,9 +411,6 @@ intel_bind_framebuffer(GLcontext * ctx, GLenum target,
 {
    if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
       intel_draw_buffer(ctx, fb);
-      /* Integer depth range depends on depth buffer bits */
-      if (ctx->Driver.DepthRange != NULL)
-	 ctx->Driver.DepthRange(ctx, ctx->Viewport.Near, ctx->Viewport.Far);
    }
    else {
       /* don't need to do anything if target == GL_READ_FRAMEBUFFER_EXT */
@@ -522,6 +434,7 @@ intel_framebuffer_renderbuffer(GLcontext * ctx,
    intel_draw_buffer(ctx, fb);
 }
 
+
 static GLboolean
 intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb, 
 		     struct gl_texture_image *texImage)
@@ -529,20 +442,25 @@ intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb,
    if (texImage->TexFormat == &_mesa_texformat_argb8888) {
       irb->Base._ActualFormat = GL_RGBA8;
       irb->Base._BaseFormat = GL_RGBA;
+      irb->Base.DataType = GL_UNSIGNED_BYTE;
       DBG("Render to RGBA8 texture OK\n");
    }
    else if (texImage->TexFormat == &_mesa_texformat_rgb565) {
       irb->Base._ActualFormat = GL_RGB5;
       irb->Base._BaseFormat = GL_RGB;
+      irb->Base.DataType = GL_UNSIGNED_SHORT;
       DBG("Render to RGB5 texture OK\n");
    }
    else if (texImage->TexFormat == &_mesa_texformat_z16) {
       irb->Base._ActualFormat = GL_DEPTH_COMPONENT16;
       irb->Base._BaseFormat = GL_DEPTH_COMPONENT;
+      irb->Base.DataType = GL_UNSIGNED_SHORT;
       DBG("Render to DEPTH16 texture OK\n");
-   } else if (texImage->TexFormat == &_mesa_texformat_s8_z24) {
+   }
+   else if (texImage->TexFormat == &_mesa_texformat_s8_z24) {
       irb->Base._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
       irb->Base._BaseFormat = GL_DEPTH_STENCIL_EXT;
+      irb->Base.DataType = GL_UNSIGNED_INT_24_8_EXT;
       DBG("Render to DEPTH_STENCIL texture OK\n");
    }
    else {
@@ -554,7 +472,6 @@ intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb,
    irb->Base.InternalFormat = irb->Base._ActualFormat;
    irb->Base.Width = texImage->Width;
    irb->Base.Height = texImage->Height;
-   irb->Base.DataType = GL_UNSIGNED_BYTE;       /* FBO XXX fix */
    irb->Base.RedBits = texImage->TexFormat->RedBits;
    irb->Base.GreenBits = texImage->TexFormat->GreenBits;
    irb->Base.BlueBits = texImage->TexFormat->BlueBits;
@@ -564,11 +481,10 @@ intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb,
    irb->Base.Delete = intel_delete_renderbuffer;
    irb->Base.AllocStorage = intel_nop_alloc_storage;
 
-   irb->RenderToTexture = GL_TRUE;
-
    return GL_TRUE;
 }
 
+
 /**
  * When glFramebufferTexture[123]D is called this function sets up the
  * gl_renderbuffer wrapper around the texture image.
@@ -577,7 +493,7 @@ intel_update_wrapper(GLcontext *ctx, struct intel_renderbuffer *irb,
 static struct intel_renderbuffer *
 intel_wrap_texture(GLcontext * ctx, struct gl_texture_image *texImage)
 {
-   const GLuint name = ~0;      /* not significant, but distinct for debugging */
+   const GLuint name = ~0;   /* not significant, but distinct for debugging */
    struct intel_renderbuffer *irb;
 
    /* make an intel_renderbuffer to wrap the texture image */
@@ -620,7 +536,15 @@ intel_render_texture(GLcontext * ctx,
 
    ASSERT(newImage);
 
-   if (!irb) {
+   if (newImage->Border != 0) {
+      /* Fallback on drawing to a texture with a border, which won't have a
+       * miptree.
+       */
+      _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
+      _mesa_render_texture(ctx, fb, att);
+      return;
+   }
+   else if (!irb) {
       irb = intel_wrap_texture(ctx, newImage);
       if (irb) {
          /* bind the wrapper to the attachment point */
@@ -631,7 +555,9 @@ intel_render_texture(GLcontext * ctx,
          _mesa_render_texture(ctx, fb, att);
          return;
       }
-   } if (!intel_update_wrapper(ctx, irb, newImage)) {
+   }
+
+   if (!intel_update_wrapper(ctx, irb, newImage)) {
        _mesa_reference_renderbuffer(&att->Renderbuffer, NULL);
        _mesa_render_texture(ctx, fb, att);
        return;
@@ -693,6 +619,94 @@ intel_finish_render_texture(GLcontext * ctx,
 
 
 /**
+ * Do additional "completeness" testing of a framebuffer object.
+ */
+static void
+intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
+{
+   const struct intel_renderbuffer *depthRb =
+      intel_get_renderbuffer(fb, BUFFER_DEPTH);
+   const struct intel_renderbuffer *stencilRb =
+      intel_get_renderbuffer(fb, BUFFER_STENCIL);
+
+   if (stencilRb && stencilRb != depthRb) {
+      /* we only support combined depth/stencil buffers, not separate
+       * stencil buffers.
+       */
+      fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+   }
+}
+
+
+/**
+ * Called from glBlitFramebuffer().
+ * For now, we're doing an approximation with glCopyPixels().
+ * XXX we need to bypass all the per-fragment operations, except scissor.
+ */
+static void
+intel_blit_framebuffer(GLcontext *ctx,
+                       GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
+                       GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
+                       GLbitfield mask, GLenum filter)
+{
+   const GLfloat xZoomSave = ctx->Pixel.ZoomX;
+   const GLfloat yZoomSave = ctx->Pixel.ZoomY;
+   GLsizei width, height;
+   GLfloat xFlip = 1.0F, yFlip = 1.0F;
+
+   if (srcX1 < srcX0) {
+      GLint tmp = srcX1;
+      srcX1 = srcX0;
+      srcX0 = tmp;
+      xFlip = -1.0F;
+   }
+
+   if (srcY1 < srcY0) {
+      GLint tmp = srcY1;
+      srcY1 = srcY0;
+      srcY0 = tmp;
+      yFlip = -1.0F;
+   }
+
+   width = srcX1 - srcX0;
+   height = srcY1 - srcY0;
+
+   ctx->Pixel.ZoomX = xFlip * (dstX1 - dstX0) / (srcX1 - srcY0);
+   ctx->Pixel.ZoomY = yFlip * (dstY1 - dstY0) / (srcY1 - srcY0);
+
+   if (ctx->Pixel.ZoomX < 0.0F) {
+      dstX0 = MAX2(dstX0, dstX1);
+   }
+   else {
+      dstX0 = MIN2(dstX0, dstX1);
+   }
+
+   if (ctx->Pixel.ZoomY < 0.0F) {
+      dstY0 = MAX2(dstY0, dstY1);
+   }
+   else {
+      dstY0 = MIN2(dstY0, dstY1);
+   }
+
+   if (mask & GL_COLOR_BUFFER_BIT) {
+      ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height,
+                             dstX0, dstY0, GL_COLOR);
+   }
+   if (mask & GL_DEPTH_BUFFER_BIT) {
+      ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height,
+                             dstX0, dstY0, GL_DEPTH);
+   }
+   if (mask & GL_STENCIL_BUFFER_BIT) {
+      ctx->Driver.CopyPixels(ctx, srcX0, srcY0, width, height,
+                             dstX0, dstY0, GL_STENCIL);
+   }
+      
+   ctx->Pixel.ZoomX = xZoomSave;
+   ctx->Pixel.ZoomY = yZoomSave;
+}
+
+
+/**
  * Do one-time context initializations related to GL_EXT_framebuffer_object.
  * Hook in device driver functions.
  */
@@ -706,4 +720,6 @@ intel_fbo_init(struct intel_context *intel)
    intel->ctx.Driver.RenderTexture = intel_render_texture;
    intel->ctx.Driver.FinishRenderTexture = intel_finish_render_texture;
    intel->ctx.Driver.ResizeBuffers = intel_resize_buffers;
+   intel->ctx.Driver.ValidateFramebuffer = intel_validate_framebuffer;
+   intel->ctx.Driver.BlitFramebuffer = intel_blit_framebuffer;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_fbo.h b/src/mesa/drivers/dri/intel/intel_fbo.h
index 9d15582d78c..7226ee026f6 100644
--- a/src/mesa/drivers/dri/intel/intel_fbo.h
+++ b/src/mesa/drivers/dri/intel/intel_fbo.h
@@ -39,14 +39,7 @@ struct intel_framebuffer
 {
    struct gl_framebuffer Base;
 
-   struct intel_renderbuffer *color_rb[3];
-
-   /* Drawable page flipping state */
-   GLboolean pf_active;
-   GLuint pf_seq;
-   GLint pf_planes;
-   GLint pf_current_page;
-   GLint pf_num_pages;
+   struct intel_renderbuffer *color_rb[2];
 
    /* VBI
     */
@@ -62,21 +55,11 @@ struct intel_framebuffer
 
 /**
  * Intel renderbuffer, derived from gl_renderbuffer.
- * Note: The PairedDepth and PairedStencil fields use renderbuffer IDs,
- * not pointers because in some circumstances a deleted renderbuffer could
- * result in a dangling pointer here.
  */
 struct intel_renderbuffer
 {
    struct gl_renderbuffer Base;
    struct intel_region *region;
-   GLuint pfPitch;              /* possibly paged flipped pitch */
-   GLboolean RenderToTexture;   /* RTT? */
-
-   GLuint PairedDepth;   /**< only used if this is a depth renderbuffer */
-   GLuint PairedStencil; /**< only used if this is a stencil renderbuffer */
-
-   GLuint pf_pending;  /**< sequence number of pending flip */
 
    GLuint vbl_pending;   /**< vblank sequence number of pending flip */
 
@@ -84,32 +67,71 @@ struct intel_renderbuffer
    unsigned long span_cache_offset;
 };
 
-extern struct intel_renderbuffer *intel_renderbuffer(struct gl_renderbuffer
-                                                     *rb);
+
+/**
+ * gl_renderbuffer is a base class which we subclass.  The Class field
+ * is used for simple run-time type checking.
+ */
+#define INTEL_RB_CLASS 0x12345678
+
+
+/**
+ * Return a gl_renderbuffer ptr casted to intel_renderbuffer.
+ * NULL will be returned if the rb isn't really an intel_renderbuffer.
+ * This is determined by checking the ClassID.
+ */
+static INLINE struct intel_renderbuffer *
+intel_renderbuffer(struct gl_renderbuffer *rb)
+{
+   struct intel_renderbuffer *irb = (struct intel_renderbuffer *) rb;
+   if (irb && irb->Base.ClassID == INTEL_RB_CLASS) {
+      /*_mesa_warning(NULL, "Returning non-intel Rb\n");*/
+      return irb;
+   }
+   else
+      return NULL;
+}
+
+
+/**
+ * Return a framebuffer's renderbuffer, named by a BUFFER_x index.
+ */
+static INLINE struct intel_renderbuffer *
+intel_get_renderbuffer(struct gl_framebuffer *fb, int attIndex)
+{
+   if (attIndex >= 0)
+      return intel_renderbuffer(fb->Attachment[attIndex].Renderbuffer);
+   else
+      return NULL;
+}
+
 
 extern void
 intel_renderbuffer_set_region(struct intel_renderbuffer *irb,
 			      struct intel_region *region);
 
+
 extern struct intel_renderbuffer *
 intel_create_renderbuffer(GLenum intFormat);
 
-extern void intel_fbo_init(struct intel_context *intel);
 
-
-/* XXX make inline or macro */
-extern struct intel_renderbuffer *intel_get_renderbuffer(struct gl_framebuffer
-                                                         *fb,
-                                                         int attIndex);
-
-extern void intel_flip_renderbuffers(struct intel_framebuffer *intel_fb);
+extern void
+intel_fbo_init(struct intel_context *intel);
 
 
-/* XXX make inline or macro */
-extern struct intel_region *intel_get_rb_region(struct gl_framebuffer *fb,
-                                                GLuint attIndex);
+extern void
+intel_flip_renderbuffers(struct intel_framebuffer *intel_fb);
 
 
+static INLINE struct intel_region *
+intel_get_rb_region(struct gl_framebuffer *fb, GLuint attIndex)
+{
+   struct intel_renderbuffer *irb = intel_get_renderbuffer(fb, attIndex);
+   if (irb)
+      return irb->region;
+   else
+      return NULL;
+}
 
 
 #endif /* INTEL_FBO_H */
diff --git a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
index b96ba72853e..bf1c3f03f0e 100644
--- a/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/intel/intel_mipmap_tree.c
@@ -111,9 +111,9 @@ intel_miptree_create(struct intel_context *intel,
 				      first_level, last_level, width0,
 				      height0, depth0, cpp, compress_byte);
    /*
-    * pitch == 0 indicates the null texture
+    * pitch == 0 || height == 0  indicates the null texture
     */
-   if (!mt || !mt->pitch)
+   if (!mt || !mt->pitch || !mt->total_height)
       return NULL;
 
    mt->region = intel_region_alloc(intel,
@@ -163,7 +163,7 @@ intel_miptree_create_for_region(struct intel_context *intel,
    mt->pitch = region->pitch;
 #endif
 
-   mt->region = region;
+   intel_region_reference(&mt->region, region);
 
    return mt;
  }
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
index b267ffd890c..5e32288844c 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -27,7 +27,12 @@
 
 #include "main/enums.h"
 #include "main/state.h"
+#include "main/context.h"
+#include "main/enable.h"
+#include "main/matrix.h"
 #include "swrast/swrast.h"
+#include "shader/arbprogram.h"
+#include "shader/program.h"
 
 #include "intel_context.h"
 #include "intel_pixel.h"
@@ -112,12 +117,6 @@ intel_check_blit_fragment_ops(GLcontext * ctx, GLboolean src_alpha_is_one)
       return GL_FALSE;
    }
 
-   if (ctx->Scissor.Enabled) {
-      /* XXX Note: Scissor could be done with the blitter */
-      DBG("fallback due to image scissor\n");
-      return GL_FALSE;
-   }
-
    if (ctx->RenderMode != GL_RENDER) {
       DBG("fallback due to render mode\n");
       return GL_FALSE;
@@ -173,6 +172,162 @@ intel_check_blit_format(struct intel_region * region,
    return GL_FALSE;
 }
 
+void
+intel_meta_set_passthrough_transform(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   intel->meta.saved_vp_x = ctx->Viewport.X;
+   intel->meta.saved_vp_y = ctx->Viewport.Y;
+   intel->meta.saved_vp_width = ctx->Viewport.Width;
+   intel->meta.saved_vp_height = ctx->Viewport.Height;
+   intel->meta.saved_matrix_mode = ctx->Transform.MatrixMode;
+
+   _mesa_Viewport(0, 0, ctx->DrawBuffer->Width, ctx->DrawBuffer->Height);
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
+
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+}
+
+void
+intel_meta_restore_transform(struct intel_context *intel)
+{
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PopMatrix();
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PopMatrix();
+
+   _mesa_MatrixMode(intel->meta.saved_matrix_mode);
+
+   _mesa_Viewport(intel->meta.saved_vp_x, intel->meta.saved_vp_y,
+		  intel->meta.saved_vp_width, intel->meta.saved_vp_height);
+}
+
+/**
+ * Set up a vertex program to pass through the position and first texcoord
+ * for pixel path.
+ */
+void
+intel_meta_set_passthrough_vertex_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+   static const char *vp =
+      "!!ARBvp1.0\n"
+      "TEMP vertexClip;\n"
+      "DP4 vertexClip.x, state.matrix.mvp.row[0], vertex.position;\n"
+      "DP4 vertexClip.y, state.matrix.mvp.row[1], vertex.position;\n"
+      "DP4 vertexClip.z, state.matrix.mvp.row[2], vertex.position;\n"
+      "DP4 vertexClip.w, state.matrix.mvp.row[3], vertex.position;\n"
+      "MOV result.position, vertexClip;\n"
+      "MOV result.texcoord[0], vertex.texcoord[0];\n"
+      "MOV result.color, vertex.color;\n"
+      "END\n";
+
+   assert(intel->meta.saved_vp == NULL);
+
+   _mesa_reference_vertprog(ctx, &intel->meta.saved_vp,
+			    ctx->VertexProgram.Current);
+   if (intel->meta.passthrough_vp == NULL) {
+      GLuint prog_name;
+      _mesa_GenPrograms(1, &prog_name);
+      _mesa_BindProgram(GL_VERTEX_PROGRAM_ARB, prog_name);
+      _mesa_ProgramStringARB(GL_VERTEX_PROGRAM_ARB,
+			     GL_PROGRAM_FORMAT_ASCII_ARB,
+			     strlen(vp), (const GLubyte *)vp);
+      _mesa_reference_vertprog(ctx, &intel->meta.passthrough_vp,
+			       ctx->VertexProgram.Current);
+      _mesa_DeletePrograms(1, &prog_name);
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current,
+			    intel->meta.passthrough_vp);
+   ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
+			   &intel->meta.passthrough_vp->Base);
+
+   intel->meta.saved_vp_enable = ctx->VertexProgram.Enabled;
+   _mesa_Enable(GL_VERTEX_PROGRAM_ARB);
+}
+
+/**
+ * Restores the previous vertex program after
+ * intel_meta_set_passthrough_vertex_program()
+ */
+void
+intel_meta_restore_vertex_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current,
+			    intel->meta.saved_vp);
+   _mesa_reference_vertprog(ctx, &intel->meta.saved_vp, NULL);
+   ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
+			   &ctx->VertexProgram.Current->Base);
+
+   if (!intel->meta.saved_vp_enable)
+      _mesa_Disable(GL_VERTEX_PROGRAM_ARB);
+}
+
+/**
+ * Binds the given program string to GL_FRAGMENT_PROGRAM_ARB, caching the
+ * program object.
+ */
+void
+intel_meta_set_fragment_program(struct intel_context *intel,
+				struct gl_fragment_program **prog,
+				const char *prog_string)
+{
+   GLcontext *ctx = &intel->ctx;
+   assert(intel->meta.saved_fp == NULL);
+
+   _mesa_reference_fragprog(ctx, &intel->meta.saved_fp,
+			    ctx->FragmentProgram.Current);
+   if (*prog == NULL) {
+      GLuint prog_name;
+      _mesa_GenPrograms(1, &prog_name);
+      _mesa_BindProgram(GL_FRAGMENT_PROGRAM_ARB, prog_name);
+      _mesa_ProgramStringARB(GL_FRAGMENT_PROGRAM_ARB,
+			     GL_PROGRAM_FORMAT_ASCII_ARB,
+			     strlen(prog_string), (const GLubyte *)prog_string);
+      _mesa_reference_fragprog(ctx, prog, ctx->FragmentProgram.Current);
+      /* Note that DeletePrograms unbinds the program on us */
+      _mesa_DeletePrograms(1, &prog_name);
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, *prog);
+   ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, &((*prog)->Base));
+
+   intel->meta.saved_fp_enable = ctx->FragmentProgram.Enabled;
+   _mesa_Enable(GL_FRAGMENT_PROGRAM_ARB);
+}
+
+/**
+ * Restores the previous fragment program after
+ * intel_meta_set_fragment_program()
+ */
+void
+intel_meta_restore_fragment_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current,
+			    intel->meta.saved_fp);
+   _mesa_reference_fragprog(ctx, &intel->meta.saved_fp, NULL);
+   ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB,
+			   &ctx->FragmentProgram.Current->Base);
+
+   if (!intel->meta.saved_fp_enable)
+      _mesa_Disable(GL_FRAGMENT_PROGRAM_ARB);
+}
 
 void
 intelInitPixelFuncs(struct dd_function_table *functions)
@@ -187,3 +342,13 @@ intelInitPixelFuncs(struct dd_function_table *functions)
 #endif
    }
 }
+
+void
+intel_free_pixel_state(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   _mesa_reference_vertprog(ctx, &intel->meta.passthrough_vp, NULL);
+   _mesa_reference_fragprog(ctx, &intel->meta.bitmap_fp, NULL);
+}
+
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.h b/src/mesa/drivers/dri/intel/intel_pixel.h
index 6fa6effe835..cb41fa182cb 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.h
+++ b/src/mesa/drivers/dri/intel/intel_pixel.h
@@ -31,6 +31,15 @@
 #include "main/mtypes.h"
 
 void intelInitPixelFuncs(struct dd_function_table *functions);
+void intel_meta_set_passthrough_transform(struct intel_context *intel);
+void intel_meta_restore_transform(struct intel_context *intel);
+void intel_meta_set_passthrough_vertex_program(struct intel_context *intel);
+void intel_meta_restore_vertex_program(struct intel_context *intel);
+void intel_meta_set_fragment_program(struct intel_context *intel,
+				     struct gl_fragment_program **prog,
+				     const char *prog_string);
+void intel_meta_restore_fragment_program(struct intel_context *intel);
+void intel_free_pixel_state(struct intel_context *intel);
 
 GLboolean intel_check_blit_fragment_ops(GLcontext * ctx,
 					GLboolean src_alpha_is_one);
@@ -67,4 +76,6 @@ void intelBitmap(GLcontext * ctx,
 		 const struct gl_pixelstore_attrib *unpack,
 		 const GLubyte * pixels);
 
+void intel_clear_tris(GLcontext *ctx, GLbitfield mask);
+
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index 0565197ea08..1db7f5594e9 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -32,6 +32,18 @@
 #include "main/mtypes.h"
 #include "main/macros.h"
 #include "main/bufferobj.h"
+#include "main/pixelstore.h"
+#include "main/state.h"
+#include "main/teximage.h"
+#include "main/texenv.h"
+#include "main/texobj.h"
+#include "main/texstate.h"
+#include "main/texparam.h"
+#include "main/varray.h"
+#include "main/attrib.h"
+#include "main/enable.h"
+#include "shader/arbprogram.h"
+#include "glapi/dispatch.h"
 #include "swrast/swrast.h"
 
 #include "intel_screen.h"
@@ -86,6 +98,11 @@ static GLboolean test_bit( const GLubyte *src,
    return (src[bit/8] & (1<<(bit % 8))) ? 1 : 0;
 }
 
+static GLboolean test_msb_bit(const GLubyte *src, GLuint bit)
+{
+   return (src[bit/8] & (1<<(7 - (bit % 8)))) ? 1 : 0;
+}
+
 static void set_bit( GLubyte *dest,
 			  GLuint bit )
 {
@@ -149,8 +166,18 @@ static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
    return count;
 }
 
-
-
+/**
+ * Returns the low Y value of the vertical range given, flipped according to
+ * whether the framebuffer is or not.
+ */
+static inline int
+y_flip(struct gl_framebuffer *fb, int y, int height)
+{
+   if (fb->Name != 0)
+      return y;
+   else
+      return fb->Height - y - height;
+}
 
 /*
  * Render a bitmap.
@@ -164,9 +191,26 @@ do_blit_bitmap( GLcontext *ctx,
 {
    struct intel_context *intel = intel_context(ctx);
    struct intel_region *dst = intel_drawbuf_region(intel);
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
    GLfloat tmpColor[4];
    GLubyte ubcolor[4];
    GLuint color8888, color565;
+   unsigned int num_cliprects;
+   drm_clip_rect_t *cliprects;
+   int x_off, y_off;
+   GLsizei bitmap_width = width;
+   GLsizei bitmap_height = height;
+
+   /* Update draw buffer bounds */
+   _mesa_update_state(ctx);
+
+   if (ctx->Depth.Test) {
+      /* The blit path produces incorrect results when depth testing is on.
+       * It seems the blit Z coord is always 1.0 (the far plane) so fragments
+       * will likely be obscured by other, closer geometry.
+       */
+      return GL_FALSE;
+   }
 
    if (!dst)
        return GL_FALSE;
@@ -196,59 +240,38 @@ do_blit_bitmap( GLcontext *ctx,
 
    LOCK_HARDWARE(intel);
 
-   if (intel->driDrawable->numClipRects) {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      drm_clip_rect_t dest_rect;
-      GLint nbox = dPriv->numClipRects;
-      GLint srcx = 0, srcy = 0;
-      GLint orig_screen_x1, orig_screen_y2;
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   if (num_cliprects != 0) {
       GLuint i;
+      GLint orig_dstx = dstx;
+      GLint orig_dsty = dsty;
 
-
-      orig_screen_x1 = dPriv->x + dstx;
-      orig_screen_y2 = dPriv->y + (dPriv->h - dsty);
-
-      /* Do scissoring in GL coordinates:
-       */
-      if (ctx->Scissor.Enabled)
-      {
-	 GLint x = ctx->Scissor.X;
-	 GLint y = ctx->Scissor.Y;
-	 GLuint w = ctx->Scissor.Width;
-	 GLuint h = ctx->Scissor.Height;
-
-         if (!_mesa_clip_to_region(x, y, x+w-1, y+h-1, &dstx, &dsty, &width, &height))
+      /* Clip to buffer bounds and scissor. */
+      if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
+				fb->_Xmax, fb->_Ymax,
+				&dstx, &dsty, &width, &height))
             goto out;
-      }
-
-      /* Convert from GL to hardware coordinates:
-       */
-      dsty = dPriv->y + (dPriv->h - dsty - height);  
-      dstx = dPriv->x + dstx;
 
-      dest_rect.x1 = dstx < 0 ? 0 : dstx;
-      dest_rect.y1 = dsty < 0 ? 0 : dsty;
-      dest_rect.x2 = dstx + width < 0 ? 0 : dstx + width;
-      dest_rect.y2 = dsty + height < 0 ? 0 : dsty + height;
+      dstx = x_off + dstx;
+      dsty = y_off + y_flip(fb, dsty, height);
 
-      for (i = 0; i < nbox; i++) {
-         drm_clip_rect_t rect;
-	 int box_w, box_h;
+      for (i = 0; i < num_cliprects; i++) {
+	 int box_x, box_y, box_w, box_h;
 	 GLint px, py;
 	 GLuint stipple[32];  
 
-         if (!intel_intersect_cliprects(&rect, &dest_rect, &box[i]))
-            continue;
-
-	 /* Now go back to GL coordinates to figure out what subset of
-	  * the bitmap we are uploading for this cliprect:
-	  */
-	 box_w = rect.x2 - rect.x1;
-	 box_h = rect.y2 - rect.y1;
-	 srcx = rect.x1 - orig_screen_x1;
-	 srcy = orig_screen_y2 - rect.y2;
+	 box_x = dstx;
+	 box_y = dsty;
+	 box_w = width;
+	 box_h = height;
 
+	 /* Clip to drawable cliprect */
+         if (!_mesa_clip_to_region(cliprects[i].x1,
+				   cliprects[i].y1,
+				   cliprects[i].x2,
+				   cliprects[i].y2,
+				   &box_x, &box_y, &box_w, &box_h))
+	    continue;
 
 #define DY 32
 #define DX 32
@@ -269,13 +292,19 @@ do_blit_bitmap( GLcontext *ctx,
 
 	       /* May need to adjust this when padding has been introduced in
 		* sz above:
+		*
+		* Have to translate destination coordinates back into source
+		* coordinates.
 		*/
-	       if (get_bitmap_rect(width, height, unpack, 
+	       if (get_bitmap_rect(bitmap_width, bitmap_height, unpack,
 				   bitmap,
-				   srcx + px, srcy + py, w, h,
+				   -orig_dstx + (box_x + px - x_off),
+				   -orig_dsty + y_flip(fb,
+						       box_y + py - y_off, h),
+				   w, h,
 				   (GLubyte *)stipple,
 				   8,
-				   GL_TRUE) == 0)
+				   fb->Name == 0 ? GL_TRUE : GL_FALSE) == 0)
 		  continue;
 
 	       /* 
@@ -289,8 +318,8 @@ do_blit_bitmap( GLcontext *ctx,
 						  dst->buffer,
 						  0,
 						  dst->tiling,
-						  rect.x1 + px,
-						  rect.y2 - (py + h),
+						  box_x + px,
+						  box_y + py,
 						  w, h,
 						  logic_op);
 	    } 
@@ -300,6 +329,8 @@ do_blit_bitmap( GLcontext *ctx,
 out:
    UNLOCK_HARDWARE(intel);
 
+   if (INTEL_DEBUG & DEBUG_SYNC)
+      intel_batchbuffer_flush(intel->batch);
 
    if (unpack->BufferObj->Name) {
       /* done with PBO so unmap it now */
@@ -310,9 +341,179 @@ out:
    return GL_TRUE;
 }
 
+static GLboolean
+intel_texture_bitmap(GLcontext * ctx,
+		     GLint dst_x, GLint dst_y,
+		     GLsizei width, GLsizei height,
+		     const struct gl_pixelstore_attrib *unpack,
+		     const GLubyte *bitmap)
+{
+   struct intel_context *intel = intel_context(ctx);
+   static const char *fp =
+      "!!ARBfp1.0\n"
+      "TEMP val;\n"
+      "PARAM color=program.local[0];\n"
+      "TEX val, fragment.texcoord[0], texture[0], 2D;\n"
+      "ADD val, val.wwww, {-.5, -.5, -.5, -.5};\n"
+      "KIL val;\n"
+      "MOV result.color, color;\n"
+      "END\n";
+   GLuint texname;
+   GLfloat vertices[4][4];
+   GLfloat texcoords[4][2];
+   GLint old_active_texture;
+   GLubyte *unpacked_bitmap;
+   GLubyte *a8_bitmap;
+   int x, y;
+   GLfloat dst_z;
+
+   /* We need a fragment program for the KIL effect */
+   if (!ctx->Extensions.ARB_fragment_program ||
+       !ctx->Extensions.ARB_vertex_program) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr,
+		 "glBitmap fallback: No fragment/vertex program support\n");
+      return GL_FALSE;
+   }
+
+   /* We're going to mess with texturing with no regard to existing texture
+    * state, so if there is some set up we have to bail.
+    */
+   if (ctx->Texture._EnabledUnits != 0) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: texturing enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Can't do textured DrawPixels with a fragment program, unless we were
+    * to generate a new program that sampled our texture and put the results
+    * in the fragment color before the user's program started.
+    */
+   if (ctx->FragmentProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: fragment program enabled\n");
+      return GL_FALSE;
+   }
 
+   if (ctx->VertexProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: vertex program enabled\n");
+      return GL_FALSE;
+   }
 
+   /* Check that we can load in a texture this big. */
+   if (width > (1 << (ctx->Const.MaxTextureLevels - 1)) ||
+       height > (1 << (ctx->Const.MaxTextureLevels - 1))) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: bitmap too large (%dx%d)\n",
+		 width, height);
+      return GL_FALSE;
+   }
 
+   /* Convert the A1 bitmap to an A8 format suitable for glTexImage */
+   if (unpack->BufferObj->Name) {
+      bitmap = map_pbo(ctx, width, height, unpack, bitmap);
+      if (bitmap == NULL)
+	 return GL_TRUE;	/* even though this is an error, we're done */
+   }
+   unpacked_bitmap = _mesa_unpack_bitmap(width, height, bitmap,
+					 unpack);
+   a8_bitmap = _mesa_calloc(width * height);
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+	 if (test_msb_bit(unpacked_bitmap, ALIGN(width, 8) * y + x))
+	    a8_bitmap[y * width + x] = 0xff;
+      }
+   }
+   _mesa_free(unpacked_bitmap);
+   if (unpack->BufferObj->Name) {
+      /* done with PBO so unmap it now */
+      ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
+                              unpack->BufferObj);
+   }
+
+   /* Save GL state before we start setting up our drawing */
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_CURRENT_BIT |
+		    GL_VIEWPORT_BIT);
+   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT |
+			  GL_CLIENT_PIXEL_STORE_BIT);
+   old_active_texture = ctx->Texture.CurrentUnit;
+
+   _mesa_Disable(GL_POLYGON_STIPPLE);
+
+   /* Upload our bitmap data to an alpha texture */
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
+   _mesa_Enable(GL_TEXTURE_2D);
+   _mesa_GenTextures(1, &texname);
+   _mesa_BindTexture(GL_TEXTURE_2D, texname);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+
+   _mesa_PixelStorei(GL_UNPACK_SWAP_BYTES, GL_FALSE);
+   _mesa_PixelStorei(GL_UNPACK_LSB_FIRST, GL_FALSE);
+   _mesa_PixelStorei(GL_UNPACK_ROW_LENGTH, 0);
+   _mesa_PixelStorei(GL_UNPACK_SKIP_PIXELS, 0);
+   _mesa_PixelStorei(GL_UNPACK_SKIP_ROWS, 0);
+   _mesa_PixelStorei(GL_UNPACK_ALIGNMENT, 1);
+   _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_ALPHA, width, height, 0,
+		    GL_ALPHA, GL_UNSIGNED_BYTE, a8_bitmap);
+   _mesa_free(a8_bitmap);
+
+   intel_meta_set_fragment_program(intel, &intel->meta.bitmap_fp, fp);
+   _mesa_ProgramLocalParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 0,
+				     ctx->Current.RasterColor);
+   intel_meta_set_passthrough_vertex_program(intel);
+   intel_meta_set_passthrough_transform(intel);
+
+   /* convert rasterpos Z from [0,1] to NDC coord in [-1,1] */
+   dst_z = -1.0 + 2.0 * ctx->Current.RasterPos[2];
+
+   vertices[0][0] = dst_x;
+   vertices[0][1] = dst_y;
+   vertices[0][2] = dst_z;
+   vertices[0][3] = 1.0;
+   vertices[1][0] = dst_x + width;
+   vertices[1][1] = dst_y;
+   vertices[1][2] = dst_z;
+   vertices[1][3] = 1.0;
+   vertices[2][0] = dst_x + width;
+   vertices[2][1] = dst_y + height;
+   vertices[2][2] = dst_z;
+   vertices[2][3] = 1.0;
+   vertices[3][0] = dst_x;
+   vertices[3][1] = dst_y + height;
+   vertices[3][2] = dst_z;
+   vertices[3][3] = 1.0;
+
+   texcoords[0][0] = 0.0;
+   texcoords[0][1] = 0.0;
+   texcoords[1][0] = 1.0;
+   texcoords[1][1] = 0.0;
+   texcoords[2][0] = 1.0;
+   texcoords[2][1] = 1.0;
+   texcoords[3][0] = 0.0;
+   texcoords[3][1] = 1.0;
+
+   _mesa_VertexPointer(4, GL_FLOAT, 4 * sizeof(GLfloat), &vertices);
+   _mesa_ClientActiveTextureARB(GL_TEXTURE0);
+   _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
+   _mesa_Enable(GL_VERTEX_ARRAY);
+   _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
+   CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
+
+   intel_meta_restore_transform(intel);
+   intel_meta_restore_fragment_program(intel);
+   intel_meta_restore_vertex_program(intel);
+
+   _mesa_PopClientAttrib();
+   _mesa_Disable(GL_TEXTURE_2D); /* asserted that it was disabled at entry */
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
+   _mesa_PopAttrib();
+
+   _mesa_DeleteTextures(1, &texname);
+
+   return GL_TRUE;
+}
 
 /* There are a large number of possible ways to implement bitmap on
  * this hardware, most of them have some sort of drawback.  Here are a
@@ -345,6 +546,10 @@ intelBitmap(GLcontext * ctx,
                           unpack, pixels))
       return;
 
+   if (intel_texture_bitmap(ctx, x, y, width, height,
+			    unpack, pixels))
+      return;
+
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
 
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
index 1b3cb5adcb7..7c7aa6097c8 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
@@ -119,6 +119,12 @@ do_texture_copypixels(GLcontext * ctx,
    if (!src || !dst || type != GL_COLOR)
       return GL_FALSE;
 
+   if (ctx->_ImageTransferState) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
    /* Can't handle overlapping regions.  Don't have sufficient control
     * over rasterization to pull it off in-place.  Punt on these for
     * now.
@@ -260,6 +266,14 @@ do_blit_copypixels(GLcontext * ctx,
    struct intel_context *intel = intel_context(ctx);
    struct intel_region *dst = intel_drawbuf_region(intel);
    struct intel_region *src = copypix_src_region(intel, type);
+   struct gl_framebuffer *fb = ctx->DrawBuffer;
+   struct gl_framebuffer *read_fb = ctx->ReadBuffer;
+   unsigned int num_cliprects;
+   drm_clip_rect_t *cliprects;
+   int x_off, y_off;
+
+   /* Update draw buffer bounds */
+   _mesa_update_state(ctx);
 
    /* Copypixels can be more than a straight copy.  Ensure all the
     * extra operations are disabled:
@@ -277,71 +291,74 @@ do_blit_copypixels(GLcontext * ctx,
 
    LOCK_HARDWARE(intel);
 
-   if (intel->driDrawable->numClipRects) {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      __DRIdrawablePrivate *dReadPriv = intel->driReadDrawable;
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      GLint nbox = dPriv->numClipRects;
-      GLint delta_x = 0;
-      GLint delta_y = 0;
+   intel_get_cliprects(intel, &cliprects, &num_cliprects, &x_off, &y_off);
+   if (num_cliprects != 0) {
+      GLint delta_x;
+      GLint delta_y;
+      GLint orig_dstx;
+      GLint orig_dsty;
+      GLint orig_srcx;
+      GLint orig_srcy;
       GLuint i;
 
-      /* Do scissoring in GL coordinates:
-       */
-      if (ctx->Scissor.Enabled)
-      {
-	 GLint x = ctx->Scissor.X;
-	 GLint y = ctx->Scissor.Y;
-	 GLuint w = ctx->Scissor.Width;
-	 GLuint h = ctx->Scissor.Height;
-	 GLint dx = dstx - srcx;
-         GLint dy = dsty - srcy;
-
-         if (!_mesa_clip_to_region(x, y, x+w-1, y+h-1, &dstx, &dsty, &width, &height))
-            goto out;
-	 
-         srcx = dstx - dx;
-         srcy = dsty - dy;
-      }
+      /* XXX: We fail to handle different inversion between read and draw framebuffer. */
+
+      /* Clip to destination buffer. */
+      orig_dstx = dstx;
+      orig_dsty = dsty;
+      if (!_mesa_clip_to_region(fb->_Xmin, fb->_Ymin,
+				fb->_Xmax, fb->_Ymax,
+				&dstx, &dsty, &width, &height))
+	 goto out;
+      /* Adjust src coords for our post-clipped destination origin */
+      srcx += dstx - orig_dstx;
+      srcy += dsty - orig_dsty;
+
+      /* Clip to source buffer. */
+      orig_srcx = srcx;
+      orig_srcy = srcy;
+      if (!_mesa_clip_to_region(0, 0,
+				read_fb->Width, read_fb->Height,
+				&srcx, &srcy, &width, &height))
+	 goto out;
+      /* Adjust dst coords for our post-clipped source origin */
+      dstx += srcx - orig_srcx;
+      dsty += srcy - orig_srcy;
 
       /* Convert from GL to hardware coordinates:
        */
-      dsty = dPriv->h - dsty - height;  
-      srcy = dPriv->h - srcy - height;  
-      dstx += dPriv->x;
-      dsty += dPriv->y;
-      srcx += dReadPriv->x;
-      srcy += dReadPriv->y;
-
-      /* Clip against the source region.  This is the only source
-       * clipping we do.  Dst is clipped with cliprects below.
-       */
-      {
-         delta_x = srcx - dstx;
-         delta_y = srcy - dsty;
-
-         if (!_mesa_clip_to_region(0, 0, src->pitch, src->height,
-                                   &srcx, &srcy, &width, &height))
-            goto out;
+      if (fb->Name == 0) {
+	 /* copypixels to a system framebuffer */
+	 dstx = x_off + dstx;
+	 dsty = y_off + (fb->Height - dsty - height);
+      } else {
+	 /* copypixels to a user framebuffer object */
+	 dstx = x_off + dstx;
+	 dsty = y_off + dsty;
+      }
 
-         dstx = srcx - delta_x;
-         dsty = srcy - delta_y;
+      /* Flip source Y if it's a system framebuffer. */
+      if (read_fb->Name == 0) {
+	 srcx = intel->driReadDrawable->x + srcx;
+	 srcy = intel->driReadDrawable->y + (fb->Height - srcy - height);
       }
 
+      delta_x = srcx - dstx;
+      delta_y = srcy - dsty;
       /* Could do slightly more clipping: Eg, take the intersection of
-       * the existing set of cliprects and those cliprects translated
-       * by delta_x, delta_y:
-       * 
+       * the destination cliprects and the read drawable cliprects
+       *
        * This code will not overwrite other windows, but will
        * introduce garbage when copying from obscured window regions.
        */
-      for (i = 0; i < nbox; i++) {
+      for (i = 0; i < num_cliprects; i++) {
 	 GLint clip_x = dstx;
 	 GLint clip_y = dsty;
 	 GLint clip_w = width;
 	 GLint clip_h = height;
 
-         if (!_mesa_clip_to_region(box[i].x1, box[i].y1, box[i].x2, box[i].y2,
+         if (!_mesa_clip_to_region(cliprects[i].x1, cliprects[i].y1,
+				   cliprects[i].x2, cliprects[i].y2,
 				   &clip_x, &clip_y, &clip_w, &clip_h))
             continue;
 
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
index 8ebbc95a1d0..7be7ea82b35 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
@@ -36,7 +36,6 @@
 #include "main/texobj.h"
 #include "main/texstate.h"
 #include "main/texparam.h"
-#include "main/matrix.h"
 #include "main/varray.h"
 #include "main/attrib.h"
 #include "main/enable.h"
@@ -68,9 +67,13 @@ intel_texture_drawpixels(GLcontext * ctx,
 			 const struct gl_pixelstore_attrib *unpack,
 			 const GLvoid *pixels)
 {
+   struct intel_context *intel = intel_context(ctx);
    GLuint texname;
    GLfloat vertices[4][4];
    GLfloat texcoords[4][2];
+   GLfloat z;
+   GLint old_active_texture;
+   GLenum internalFormat;
 
    /* We're going to mess with texturing with no regard to existing texture
     * state, so if there is some set up we have to bail.
@@ -91,7 +94,7 @@ intel_texture_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   /* We don't have a way to generate fragments with stencil values which *
+   /* We don't have a way to generate fragments with stencil values which
     * will set the resulting stencil value.
     */
    if (format == GL_STENCIL_INDEX)
@@ -117,13 +120,14 @@ intel_texture_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT |
 		    GL_CURRENT_BIT);
    _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
 
    /* XXX: pixel store stuff */
    _mesa_Disable(GL_POLYGON_STIPPLE);
 
+   old_active_texture = ctx->Texture.CurrentUnit;
    _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
    _mesa_Enable(GL_TEXTURE_2D);
    _mesa_GenTextures(1, &texname);
@@ -131,21 +135,17 @@ intel_texture_drawpixels(GLcontext * ctx,
    _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
    _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
    _mesa_TexEnvf(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, GL_REPLACE);
-   /*
-   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_COMBINE_RGB, GL_REPLACE);
-   _mesa_TexEnvf(GL_TEXTURE_ENV, GL_COMBINE_ALPHA, GL_REPLACE);
-   */
-   _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, format,
+   if (type == GL_ALPHA)
+      internalFormat = GL_ALPHA;
+   else
+      internalFormat = GL_RGBA;
+   _mesa_TexImage2D(GL_TEXTURE_2D, 0, internalFormat, width, height, 0, format,
 		    type, pixels);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
-   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
+   intel_meta_set_passthrough_transform(intel);
 
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
+   /* convert rasterpos Z from [0,1] to NDC coord in [-1,1] */
+   z = -1.0 + 2.0 * ctx->Current.RasterPos[2];
 
    /* Create the vertex buffer based on the current raster pos.  The x and y
     * we're handed are ctx->Current.RasterPos[0,1] rounded to integers.
@@ -154,19 +154,19 @@ intel_texture_drawpixels(GLcontext * ctx,
     */
    vertices[0][0] = x;
    vertices[0][1] = y;
-   vertices[0][2] = ctx->Current.RasterPos[2];
+   vertices[0][2] = z;
    vertices[0][3] = 1.0;
    vertices[1][0] = x + width * ctx->Pixel.ZoomX;
    vertices[1][1] = y;
-   vertices[1][2] = ctx->Current.RasterPos[2];
+   vertices[1][2] = z;
    vertices[1][3] = 1.0;
    vertices[2][0] = x + width * ctx->Pixel.ZoomX;
    vertices[2][1] = y + height * ctx->Pixel.ZoomY;
-   vertices[2][2] = ctx->Current.RasterPos[2];
+   vertices[2][2] = z;
    vertices[2][3] = 1.0;
    vertices[3][0] = x;
    vertices[3][1] = y + height * ctx->Pixel.ZoomY;
-   vertices[3][2] = ctx->Current.RasterPos[2];
+   vertices[3][2] = z;
    vertices[3][3] = 1.0;
 
    texcoords[0][0] = 0.0;
@@ -179,15 +179,15 @@ intel_texture_drawpixels(GLcontext * ctx,
    texcoords[3][1] = 1.0;
 
    _mesa_VertexPointer(4, GL_FLOAT, 4 * sizeof(GLfloat), &vertices);
+   _mesa_ClientActiveTextureARB(GL_TEXTURE0);
    _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
    _mesa_Enable(GL_VERTEX_ARRAY);
    _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
    CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PopMatrix();
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PopMatrix();
+   intel_meta_restore_transform(intel);
+
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
    _mesa_PopClientAttrib();
    _mesa_PopAttrib();
 
@@ -205,6 +205,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
 			 const struct gl_pixelstore_attrib *unpack,
 			 const GLvoid *pixels)
 {
+   struct intel_context *intel = intel_context(ctx);
    GLuint texname, rb_name, fb_name, old_fb_name;
    GLfloat vertices[4][2];
    GLfloat texcoords[4][2];
@@ -214,6 +215,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
    struct gl_pixelstore_attrib old_unpack;
    GLstencil *stencil_pixels;
    int row;
+   GLint old_active_texture;
 
    if (format != GL_STENCIL_INDEX)
       return GL_FALSE;
@@ -230,6 +232,10 @@ intel_stencil_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
+   /* We don't support stencil testing/ops here */
+   if (ctx->Stencil.Enabled)
+      return GL_FALSE;
+
    /* We use FBOs for our wrapping of the depthbuffer into a color
     * destination.
     */
@@ -267,10 +273,11 @@ intel_stencil_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT |
 		    GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
    _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
    old_fb_name = ctx->DrawBuffer->Name;
+   old_active_texture = ctx->Texture.CurrentUnit;
 
    _mesa_Disable(GL_POLYGON_STIPPLE);
    _mesa_Disable(GL_DEPTH_TEST);
@@ -335,14 +342,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
    ctx->Unpack = old_unpack;
    _mesa_free(stencil_pixels);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
-   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
-
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
+   intel_meta_set_passthrough_transform(intel);
 
    vertices[0][0] = x;
    vertices[0][1] = y;
@@ -363,17 +363,17 @@ intel_stencil_drawpixels(GLcontext * ctx,
    texcoords[3][1] = 1.0;
 
    _mesa_VertexPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &vertices);
+   _mesa_ClientActiveTextureARB(GL_TEXTURE0);
    _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
    _mesa_Enable(GL_VERTEX_ARRAY);
    _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
    CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
 
+   intel_meta_restore_transform(intel);
+
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
    _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, old_fb_name);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PopMatrix();
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PopMatrix();
    _mesa_PopClientAttrib();
    _mesa_PopAttrib();
 
@@ -404,27 +404,6 @@ intelDrawPixels(GLcontext * ctx,
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
 
-   if (ctx->FragmentProgram._Current == ctx->FragmentProgram._TexEnvProgram) {
-      /*
-       * We don't want the i915 texenv program to be applied to DrawPixels.
-       * This is really just a performance optimization (mesa will other-
-       * wise happily run the fragment program on each pixel in the image).
-       */
-      struct gl_fragment_program *fpSave = ctx->FragmentProgram._Current;
-   /* can't just set current frag prog to 0 here as on buffer resize
-      we'll get new state checks which will segfault. Remains a hack. */
-      ctx->FragmentProgram._Current = NULL;
-      ctx->FragmentProgram._UseTexEnvProgram = GL_FALSE;
-      ctx->FragmentProgram._Active = GL_FALSE;
-      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-                          unpack, pixels );
-      ctx->FragmentProgram._Current = fpSave;
-      ctx->FragmentProgram._UseTexEnvProgram = GL_TRUE;
-      ctx->FragmentProgram._Active = GL_TRUE;
-      _swrast_InvalidateState(ctx, _NEW_PROGRAM);
-   }
-   else {
-      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-                          unpack, pixels );
-   }
+   _swrast_DrawPixels(ctx, x, y, width, height, format, type,
+		      unpack, pixels);
 }
diff --git a/src/mesa/drivers/dri/intel/intel_reg.h b/src/mesa/drivers/dri/intel/intel_reg.h
index 81a7386e429..57ac8f0cc14 100644
--- a/src/mesa/drivers/dri/intel/intel_reg.h
+++ b/src/mesa/drivers/dri/intel/intel_reg.h
@@ -71,16 +71,16 @@
 /** @{
  * 915 definitions
  */
-#define S0_VB_OFFSET_MASK		0xffffffc
+#define S0_VB_OFFSET_MASK		0xffffffc0
 #define S0_AUTO_CACHE_INV_DISABLE	(1<<0)
 /** @} */
 
 /** @{
  * 830 definitions
  */
-#define S0_VB_OFFSET_MASK_830		0xffffff8
+#define S0_VB_OFFSET_MASK_830		0xffffff80
 #define S0_VB_PITCH_SHIFT_830		1
-#define S0_VB_ENABLE_830		0
+#define S0_VB_ENABLE_830		(1<<0)
 /** @} */
 
 #define S1_VERTEX_WIDTH_SHIFT          24
@@ -100,8 +100,8 @@
 #define S2_TEXCOORD_FMT(unit, type)    ((type)<<(unit*4))
 #define S2_TEXCOORD_NONE               (~0)
 #define S2_TEX_COUNT_SHIFT_830		12
-#define S2_VERTEX_0_WIDTH_SHIFT_830	0
-#define S2_VERTEX_1_WIDTH_SHIFT_830	6
+#define S2_VERTEX_1_WIDTH_SHIFT_830	0
+#define S2_VERTEX_0_WIDTH_SHIFT_830	6
 /* S3 not interesting */
 
 #define S4_POINT_WIDTH_SHIFT           23
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index 8dbcc3050ee..ec85c4131a1 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -152,7 +152,7 @@ void
 intel_region_reference(struct intel_region **dst, struct intel_region *src)
 {
    if (src)
-      DBG("%s %d\n", __FUNCTION__, src->refcount);
+      DBG("%s %p %d\n", __FUNCTION__, src, src->refcount);
 
    assert(*dst == NULL);
    if (src) {
@@ -169,7 +169,7 @@ intel_region_release(struct intel_region **region_handle)
    if (region == NULL)
       return;
 
-   DBG("%s %d\n", __FUNCTION__, region->refcount - 1);
+   DBG("%s %p %d\n", __FUNCTION__, region, region->refcount - 1);
 
    ASSERT(region->refcount > 0);
    region->refcount--;
@@ -550,15 +550,6 @@ intel_recreate_static_regions(struct intel_context *intel)
 			    intel->back_region,
 			    &intelScreen->back);
 
-#ifdef I915
-   if (intelScreen->third.handle) {
-      intel->third_region =
-	 intel_recreate_static(intel, "third",
-			       intel->third_region,
-			       &intelScreen->third);
-   }
-#endif /* I915 */
-
    /* Still assumes front.cpp == depth.cpp.  We can kill this when we move to
     * private buffers.
     */
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index cf09fad7451..a52271158c6 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -28,27 +28,27 @@
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/framebuffer.h"
-#include "main/matrix.h"
 #include "main/renderbuffer.h"
-#include "main/simple_list.h"
+
 #include "utils.h"
 #include "vblank.h"
 #include "xmlpool.h"
 
-
-#include "intel_screen.h"
-
+#include "intel_batchbuffer.h"
 #include "intel_buffers.h"
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_fbo.h"
+#include "intel_bufmgr.h"
 #include "intel_chipset.h"
+#include "intel_extensions.h"
+#include "intel_fbo.h"
+#include "intel_regions.h"
+#include "intel_swapbuffers.h"
+#include "intel_screen.h"
+#include "intel_span.h"
+#include "intel_tex.h"
 
 #include "i915_drm.h"
 #include "i830_dri.h"
-#include "intel_regions.h"
-#include "intel_batchbuffer.h"
-#include "intel_bufmgr.h"
+
 
 PUBLIC const char __driConfigOptions[] =
    DRI_CONF_BEGIN
@@ -134,7 +134,7 @@ intelPrintDRIInfo(intelScreenPrivate * intelScreen,
 
 
 static void
-intelPrintSAREA(const struct drm_i915_sarea * sarea)
+intelPrintSAREA(const drm_i915_sarea_t * sarea)
 {
    fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width,
            sarea->height);
@@ -159,9 +159,9 @@ intelPrintSAREA(const struct drm_i915_sarea * sarea)
  * A number of the screen parameters are obtained/computed from
  * information in the SAREA.  This function updates those parameters.
  */
-void
+static void
 intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
-                           struct drm_i915_sarea * sarea)
+                           drm_i915_sarea_t * sarea)
 {
    intelScreen->width = sarea->width;
    intelScreen->height = sarea->height;
@@ -177,13 +177,6 @@ intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
    intelScreen->back.size = sarea->back_size;
    intelScreen->back.tiled = sarea->back_tiled;
 
-   if (intelScreen->driScrnPriv->ddx_version.minor >= 8) {
-      intelScreen->third.offset = sarea->third_offset;
-      intelScreen->third.handle = sarea->third_handle;
-      intelScreen->third.size = sarea->third_size;
-      intelScreen->third.tiled = sarea->third_tiled;
-   }
-
    intelScreen->depth.offset = sarea->depth_offset;
    intelScreen->depth.handle = sarea->depth_handle;
    intelScreen->depth.size = sarea->depth_size;
@@ -192,12 +185,10 @@ intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
    if (intelScreen->driScrnPriv->ddx_version.minor >= 9) {
       intelScreen->front.bo_handle = sarea->front_bo_handle;
       intelScreen->back.bo_handle = sarea->back_bo_handle;
-      intelScreen->third.bo_handle = sarea->third_bo_handle;
       intelScreen->depth.bo_handle = sarea->depth_bo_handle;
    } else {
       intelScreen->front.bo_handle = -1;
       intelScreen->back.bo_handle = -1;
-      intelScreen->third.bo_handle = -1;
       intelScreen->depth.bo_handle = -1;
    }
 
@@ -253,7 +244,7 @@ static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
 {
    intelScreenPrivate *intelScreen;
    I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
-   struct drm_i915_sarea *sarea;
+   drm_i915_sarea_t *sarea;
 
    if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
       fprintf(stderr,
@@ -273,7 +264,7 @@ static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
 
    intelScreen->driScrnPriv = sPriv;
    sPriv->private = (void *) intelScreen;
-   sarea = (struct drm_i915_sarea *)
+   sarea = (drm_i915_sarea_t *)
       (((GLubyte *) sPriv->pSAREA) + gDRIPriv->sarea_priv_offset);
    intelScreen->sarea = sarea;
 
@@ -325,8 +316,6 @@ intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
                   __DRIdrawablePrivate * driDrawPriv,
                   const __GLcontextModes * mesaVis, GLboolean isPixmap)
 {
-   intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
-
    if (isPixmap) {
       return GL_FALSE;          /* not implemented */
    }
@@ -353,12 +342,6 @@ intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
          _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT,
 				&intel_fb->color_rb[1]->Base);
 
-	 if (screen->third.handle) {
-	    struct gl_renderbuffer *tmp_rb = NULL;
-
-	    intel_fb->color_rb[2] = intel_create_renderbuffer(rgbFormat);
-	    _mesa_reference_renderbuffer(&tmp_rb, &intel_fb->color_rb[2]->Base);
-	 }
       }
 
       if (mesaVis->depthBits == 24) {
@@ -461,6 +444,7 @@ intelCreateContext(const __GLcontextModes * mesaVis,
 				  sharedContextPrivate);
       }
    } else {
+      intelScreen->no_vbo = GL_TRUE;
       return i830CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
    }
 #else
@@ -481,8 +465,6 @@ intelFillInModes(__DRIscreenPrivate *psp,
    __GLcontextModes *m;
    unsigned depth_buffer_factor;
    unsigned back_buffer_factor;
-   GLenum fb_format;
-   GLenum fb_type;
    int i;
 
    /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
@@ -494,6 +476,7 @@ intelFillInModes(__DRIscreenPrivate *psp,
 
    uint8_t depth_bits_array[3];
    uint8_t stencil_bits_array[3];
+   uint8_t msaa_samples_array[1];
 
    depth_bits_array[0] = 0;
    depth_bits_array[1] = depth_bits;
@@ -510,22 +493,39 @@ intelFillInModes(__DRIscreenPrivate *psp,
 
    stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+   msaa_samples_array[0] = 0;
+
    depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
    back_buffer_factor = (have_back_buffer) ? 3 : 1;
 
    if (pixel_bits == 16) {
-      fb_format = GL_RGB;
-      fb_type = GL_UNSIGNED_SHORT_5_6_5;
+      configs = driCreateConfigs(GL_RGB, GL_UNSIGNED_SHORT_5_6_5,
+				 depth_bits_array, stencil_bits_array,
+				 depth_buffer_factor, back_buffer_modes,
+				 back_buffer_factor,
+				 msaa_samples_array, 1);
    }
    else {
-      fb_format = GL_BGRA;
-      fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+      __DRIconfig **configs_a8r8g8b8;
+      __DRIconfig **configs_x8r8g8b8;
+
+      configs_a8r8g8b8 = driCreateConfigs(GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV,
+					  depth_bits_array,
+					  stencil_bits_array,
+					  depth_buffer_factor,
+					  back_buffer_modes,
+					  back_buffer_factor,
+					  msaa_samples_array, 1);
+      configs_x8r8g8b8 = driCreateConfigs(GL_BGR, GL_UNSIGNED_INT_8_8_8_8_REV,
+					  depth_bits_array,
+					  stencil_bits_array,
+					  depth_buffer_factor,
+					  back_buffer_modes,
+					  back_buffer_factor,
+					  msaa_samples_array, 1);
+      configs = driConcatConfigs(configs_a8r8g8b8, configs_x8r8g8b8);
    }
 
-   configs = driCreateConfigs(fb_format, fb_type,
-			      depth_bits_array, stencil_bits_array,
-			      depth_buffer_factor, back_buffer_modes,
-			      back_buffer_factor);
    if (configs == NULL) {
     fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
               __LINE__);
@@ -687,6 +687,17 @@ static const
 __DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
 {
    intelScreenPrivate *intelScreen;
+   GLenum fb_format[3];
+   GLenum fb_type[3];
+   /* GLX_SWAP_COPY_OML is only supported because the Intel driver doesn't
+    * support pageflipping at all.
+    */
+   static const GLenum back_buffer_modes[] = {
+      GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
+   };
+   uint8_t depth_bits[4], stencil_bits[4], msaa_samples_array[1];
+   int color;
+   __DRIconfig **configs = NULL;
 
    /* Calling driInitExtensions here, with a NULL context pointer,
     * does not actually enable the extensions.  It just makes sure
@@ -726,8 +737,50 @@ __DRIconfig **intelInitScreen2(__DRIscreenPrivate *psp)
    intelScreen->irq_active = 1;
    psp->extensions = intelScreenExtensions;
 
-   return driConcatConfigs(intelFillInModes(psp, 16, 16, 0, 1),
-			   intelFillInModes(psp, 32, 24, 8, 1));
+   depth_bits[0] = 0;
+   stencil_bits[0] = 0;
+   depth_bits[1] = 16;
+   stencil_bits[1] = 0;
+   depth_bits[2] = 24;
+   stencil_bits[2] = 0;
+   depth_bits[3] = 24;
+   stencil_bits[3] = 8;
+
+   msaa_samples_array[0] = 0;
+
+   fb_format[0] = GL_RGB;
+   fb_type[0] = GL_UNSIGNED_SHORT_5_6_5;
+
+   fb_format[1] = GL_BGR;
+   fb_type[1] = GL_UNSIGNED_INT_8_8_8_8_REV;
+
+   fb_format[2] = GL_BGRA;
+   fb_type[2] = GL_UNSIGNED_INT_8_8_8_8_REV;
+
+   for (color = 0; color < ARRAY_SIZE(fb_format); color++) {
+      __DRIconfig **new_configs;
+
+      new_configs = driCreateConfigs(fb_format[color], fb_type[color],
+				     depth_bits,
+				     stencil_bits,
+				     ARRAY_SIZE(depth_bits),
+				     back_buffer_modes,
+				     ARRAY_SIZE(back_buffer_modes),
+				     msaa_samples_array,
+				     ARRAY_SIZE(msaa_samples_array));
+      if (configs == NULL)
+	 configs = new_configs;
+      else
+	 configs = driConcatConfigs(configs, new_configs);
+   }
+
+   if (configs == NULL) {
+      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
+   }
+
+   return (const __DRIconfig **)configs;
 }
 
 const struct __DriverAPIRec driDriverAPI = {
diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index fc913da5bfe..e1036de4db8 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -56,7 +56,6 @@ typedef struct
 {
    intelRegion front;
    intelRegion back;
-   intelRegion third;
    intelRegion depth;
    intelRegion tex;
 
@@ -69,7 +68,7 @@ typedef struct
 
    __DRIscreenPrivate *driScrnPriv;
 
-   volatile struct drm_i915_sarea *sarea;
+   volatile drm_i915_sarea_t *sarea;
 
    int drmMinor;
 
@@ -77,6 +76,7 @@ typedef struct
 
    GLboolean no_hw;
 
+   GLboolean no_vbo;
    int ttm;
    dri_bufmgr *bufmgr;
 
@@ -92,10 +92,6 @@ extern GLboolean intelMapScreenRegions(__DRIscreenPrivate * sPriv);
 
 extern void intelUnmapScreenRegions(intelScreenPrivate * intelScreen);
 
-extern void
-intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
-                           struct drm_i915_sarea * sarea);
-
 extern void intelDestroyContext(__DRIcontextPrivate * driContextPriv);
 
 extern GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv);
@@ -105,11 +101,6 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
                  __DRIdrawablePrivate * driDrawPriv,
                  __DRIdrawablePrivate * driReadPriv);
 
-extern void intelSwapBuffers(__DRIdrawablePrivate * dPriv);
-
-extern void
-intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h);
-
 extern struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen);
 
 #endif
diff --git a/src/mesa/drivers/dri/intel/intel_span.c b/src/mesa/drivers/dri/intel/intel_span.c
index 8f4e681ffea..bdd2fd9e857 100644
--- a/src/mesa/drivers/dri/intel/intel_span.c
+++ b/src/mesa/drivers/dri/intel/intel_span.c
@@ -150,7 +150,7 @@ static uint32_t x_tile_swizzle(struct intel_renderbuffer *irb,
 	int	x_tile_number, y_tile_number;
 	int	tile_off, tile_base;
 	
-	tile_stride = (irb->pfPitch * irb->region->cpp) << 3;
+	tile_stride = (irb->region->pitch * irb->region->cpp) << 3;
 
 	xbyte = x * irb->region->cpp;
 
@@ -190,7 +190,7 @@ static uint32_t x_tile_swizzle(struct intel_renderbuffer *irb,
 	printf("(%d,%d) -> %d + %d = %d (pitch = %d, tstride = %d)\n",
 	       x, y, tile_off, tile_base,
 	       tile_off + tile_base,
-	       irb->pfPitch, tile_stride);
+	       irb->region->pitch, tile_stride);
 #endif
 
 	return tile_base + tile_off;
@@ -205,7 +205,7 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 	int	x_tile_number, y_tile_number;
 	int	tile_off, tile_base;
 	
-	tile_stride = (irb->pfPitch * irb->region->cpp) << 5;
+	tile_stride = (irb->region->pitch * irb->region->cpp) << 5;
 
 	xbyte = x * irb->region->cpp;
 
@@ -255,8 +255,8 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 #define LOCAL_VARS							\
    struct intel_context *intel = intel_context(ctx);			\
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
-   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
-   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1;	\
+   const GLint yScale = ctx->DrawBuffer->Name ? 1 : -1;			\
+   const GLint yBias = ctx->DrawBuffer->Name ? 0 : irb->Base.Height - 1;\
    unsigned int num_cliprects;						\
    struct drm_clip_rect *cliprects;					\
    int x_off, y_off;							\
@@ -392,8 +392,8 @@ static uint32_t y_tile_swizzle(struct intel_renderbuffer *irb,
 #define LOCAL_DEPTH_VARS						\
    struct intel_context *intel = intel_context(ctx);			\
    struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
-   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
-   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1; \
+   const GLint yScale = ctx->DrawBuffer->Name ? 1 : -1;			\
+   const GLint yBias = ctx->DrawBuffer->Name ? 0 : irb->Base.Height - 1;\
    unsigned int num_cliprects;						\
    struct drm_clip_rect *cliprects;					\
    int x_off, y_off;							\
@@ -528,8 +528,6 @@ intel_renderbuffer_map(struct intel_context *intel, struct gl_renderbuffer *rb)
    if (irb == NULL || irb->region == NULL)
       return;
 
-   irb->pfPitch = irb->region->pitch;
-
    intel_set_span_functions(intel, rb);
 }
 
@@ -543,7 +541,6 @@ intel_renderbuffer_unmap(struct intel_context *intel,
       return;
 
    clear_span_cache(irb);
-   irb->pfPitch = 0;
 
    rb->GetRow = NULL;
    rb->PutRow = NULL;
@@ -633,7 +630,7 @@ intelSpanRenderStart(GLcontext * ctx)
    intelFlush(&intel->ctx);
    LOCK_HARDWARE(intel);
 
-   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
+   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
       if (ctx->Texture.Unit[i]._ReallyEnabled) {
          struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
          intel_tex_map_images(intel, intel_texture_object(texObj));
@@ -655,7 +652,7 @@ intelSpanRenderFinish(GLcontext * ctx)
 
    _swrast_flush(ctx);
 
-   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
+   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
       if (ctx->Texture.Unit[i]._ReallyEnabled) {
          struct gl_texture_object *texObj = ctx->Texture.Unit[i]._Current;
          intel_tex_unmap_images(intel, intel_texture_object(texObj));
diff --git a/src/mesa/drivers/dri/intel/intel_state.c b/src/mesa/drivers/dri/intel/intel_state.c
new file mode 100644
index 00000000000..4ee742377d6
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_state.c
@@ -0,0 +1,233 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/colormac.h"
+#include "main/dd.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_regions.h"
+#include "swrast/swrast.h"
+
+int
+intel_translate_shadow_compare_func(GLenum func)
+{
+   switch (func) {
+   case GL_NEVER: 
+       return COMPAREFUNC_ALWAYS;
+   case GL_LESS: 
+       return COMPAREFUNC_LEQUAL;
+   case GL_LEQUAL: 
+       return COMPAREFUNC_LESS;
+   case GL_GREATER: 
+       return COMPAREFUNC_GEQUAL;
+   case GL_GEQUAL: 
+      return COMPAREFUNC_GREATER;
+   case GL_NOTEQUAL: 
+      return COMPAREFUNC_EQUAL;
+   case GL_EQUAL: 
+      return COMPAREFUNC_NOTEQUAL;
+   case GL_ALWAYS: 
+       return COMPAREFUNC_NEVER;
+   }
+
+   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
+   return COMPAREFUNC_NEVER;
+}
+
+int
+intel_translate_compare_func(GLenum func)
+{
+   switch (func) {
+   case GL_NEVER:
+      return COMPAREFUNC_NEVER;
+   case GL_LESS:
+      return COMPAREFUNC_LESS;
+   case GL_LEQUAL:
+      return COMPAREFUNC_LEQUAL;
+   case GL_GREATER:
+      return COMPAREFUNC_GREATER;
+   case GL_GEQUAL:
+      return COMPAREFUNC_GEQUAL;
+   case GL_NOTEQUAL:
+      return COMPAREFUNC_NOTEQUAL;
+   case GL_EQUAL:
+      return COMPAREFUNC_EQUAL;
+   case GL_ALWAYS:
+      return COMPAREFUNC_ALWAYS;
+   }
+
+   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, func);
+   return COMPAREFUNC_ALWAYS;
+}
+
+int
+intel_translate_stencil_op(GLenum op)
+{
+   switch (op) {
+   case GL_KEEP:
+      return STENCILOP_KEEP;
+   case GL_ZERO:
+      return STENCILOP_ZERO;
+   case GL_REPLACE:
+      return STENCILOP_REPLACE;
+   case GL_INCR:
+      return STENCILOP_INCRSAT;
+   case GL_DECR:
+      return STENCILOP_DECRSAT;
+   case GL_INCR_WRAP:
+      return STENCILOP_INCR;
+   case GL_DECR_WRAP:
+      return STENCILOP_DECR;
+   case GL_INVERT:
+      return STENCILOP_INVERT;
+   default:
+      return STENCILOP_ZERO;
+   }
+}
+
+int
+intel_translate_blend_factor(GLenum factor)
+{
+   switch (factor) {
+   case GL_ZERO:
+      return BLENDFACT_ZERO;
+   case GL_SRC_ALPHA:
+      return BLENDFACT_SRC_ALPHA;
+   case GL_ONE:
+      return BLENDFACT_ONE;
+   case GL_SRC_COLOR:
+      return BLENDFACT_SRC_COLR;
+   case GL_ONE_MINUS_SRC_COLOR:
+      return BLENDFACT_INV_SRC_COLR;
+   case GL_DST_COLOR:
+      return BLENDFACT_DST_COLR;
+   case GL_ONE_MINUS_DST_COLOR:
+      return BLENDFACT_INV_DST_COLR;
+   case GL_ONE_MINUS_SRC_ALPHA:
+      return BLENDFACT_INV_SRC_ALPHA;
+   case GL_DST_ALPHA:
+      return BLENDFACT_DST_ALPHA;
+   case GL_ONE_MINUS_DST_ALPHA:
+      return BLENDFACT_INV_DST_ALPHA;
+   case GL_SRC_ALPHA_SATURATE:
+      return BLENDFACT_SRC_ALPHA_SATURATE;
+   case GL_CONSTANT_COLOR:
+      return BLENDFACT_CONST_COLOR;
+   case GL_ONE_MINUS_CONSTANT_COLOR:
+      return BLENDFACT_INV_CONST_COLOR;
+   case GL_CONSTANT_ALPHA:
+      return BLENDFACT_CONST_ALPHA;
+   case GL_ONE_MINUS_CONSTANT_ALPHA:
+      return BLENDFACT_INV_CONST_ALPHA;
+   }
+
+   fprintf(stderr, "Unknown value in %s: %x\n", __FUNCTION__, factor);
+   return BLENDFACT_ZERO;
+}
+
+int
+intel_translate_logic_op(GLenum opcode)
+{
+   switch (opcode) {
+   case GL_CLEAR:
+      return LOGICOP_CLEAR;
+   case GL_AND:
+      return LOGICOP_AND;
+   case GL_AND_REVERSE:
+      return LOGICOP_AND_RVRSE;
+   case GL_COPY:
+      return LOGICOP_COPY;
+   case GL_COPY_INVERTED:
+      return LOGICOP_COPY_INV;
+   case GL_AND_INVERTED:
+      return LOGICOP_AND_INV;
+   case GL_NOOP:
+      return LOGICOP_NOOP;
+   case GL_XOR:
+      return LOGICOP_XOR;
+   case GL_OR:
+      return LOGICOP_OR;
+   case GL_OR_INVERTED:
+      return LOGICOP_OR_INV;
+   case GL_NOR:
+      return LOGICOP_NOR;
+   case GL_EQUIV:
+      return LOGICOP_EQUIV;
+   case GL_INVERT:
+      return LOGICOP_INV;
+   case GL_OR_REVERSE:
+      return LOGICOP_OR_RVRSE;
+   case GL_NAND:
+      return LOGICOP_NAND;
+   case GL_SET:
+      return LOGICOP_SET;
+   default:
+      return LOGICOP_SET;
+   }
+}
+
+
+static void
+intelClearColor(GLcontext *ctx, const GLfloat color[4])
+{
+   struct intel_context *intel = intel_context(ctx);
+   GLubyte clear[4];
+
+   CLAMPED_FLOAT_TO_UBYTE(clear[0], color[0]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[1], color[1]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[2], color[2]);
+   CLAMPED_FLOAT_TO_UBYTE(clear[3], color[3]);
+
+   /* compute both 32 and 16-bit clear values */
+   intel->ClearColor8888 = INTEL_PACKCOLOR8888(clear[0], clear[1],
+                                               clear[2], clear[3]);
+   intel->ClearColor565 = INTEL_PACKCOLOR565(clear[0], clear[1], clear[2]);
+}
+
+
+/* Fallback to swrast for select and feedback.
+ */
+static void
+intelRenderMode(GLcontext *ctx, GLenum mode)
+{
+   struct intel_context *intel = intel_context(ctx);
+   FALLBACK(intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER));
+}
+
+
+void
+intelInitStateFuncs(struct dd_function_table *functions)
+{
+   functions->RenderMode = intelRenderMode;
+   functions->ClearColor = intelClearColor;
+}
diff --git a/src/mesa/drivers/dri/intel/intel_swapbuffers.c b/src/mesa/drivers/dri/intel/intel_swapbuffers.c
new file mode 100644
index 00000000000..7d035b9f6e4
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_swapbuffers.c
@@ -0,0 +1,248 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_swapbuffers.h"
+#include "intel_fbo.h"
+#include "intel_batchbuffer.h"
+#include "drirenderbuffer.h"
+#include "vblank.h"
+#include "i915_drm.h"
+
+
+
+/*
+ * Correct a drawablePrivate's set of vblank flags WRT the current context.
+ * When considering multiple crtcs.
+ */
+GLuint
+intelFixupVblank(struct intel_context *intel, __DRIdrawablePrivate *dPriv)
+{
+   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
+       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
+      volatile drm_i915_sarea_t *sarea = intel->sarea;
+      drm_clip_rect_t drw_rect = { .x1 = dPriv->x, .x2 = dPriv->x + dPriv->w,
+				   .y1 = dPriv->y, .y2 = dPriv->y + dPriv->h };
+      drm_clip_rect_t planeA_rect = { .x1 = sarea->planeA_x, .y1 = sarea->planeA_y,
+				     .x2 = sarea->planeA_x + sarea->planeA_w,
+				     .y2 = sarea->planeA_y + sarea->planeA_h };
+      drm_clip_rect_t planeB_rect = { .x1 = sarea->planeB_x, .y1 = sarea->planeB_y,
+				     .x2 = sarea->planeB_x + sarea->planeB_w,
+				     .y2 = sarea->planeB_y + sarea->planeB_h };
+      GLint areaA = driIntersectArea( drw_rect, planeA_rect );
+      GLint areaB = driIntersectArea( drw_rect, planeB_rect );
+      GLuint flags = dPriv->vblFlags;
+
+      /* Update vblank info
+       */
+      if (areaB > areaA || (areaA == areaB && areaB > 0)) {
+	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
+      } else {
+	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
+      }
+
+      /* Do the stupid test: Is one of them actually disabled?
+       */
+      if (sarea->planeA_w == 0 || sarea->planeA_h == 0) {
+	 flags = dPriv->vblFlags | VBLANK_FLAG_SECONDARY;
+      } else if (sarea->planeB_w == 0 || sarea->planeB_h == 0) {
+	 flags = dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
+      }
+
+      return flags;
+   } else {
+      return dPriv->vblFlags & ~VBLANK_FLAG_SECONDARY;
+   }
+}
+
+
+/**
+ * Called from driSwapBuffers()
+ */
+void
+intelSwapBuffers(__DRIdrawablePrivate * dPriv)
+{
+   __DRIscreenPrivate *psp = dPriv->driScreenPriv;
+
+   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
+      GET_CURRENT_CONTEXT(ctx);
+      struct intel_context *intel;
+
+      if (ctx == NULL)
+	 return;
+
+      intel = intel_context(ctx);
+
+      if (ctx->Visual.doubleBufferMode) {
+	 GLboolean missed_target;
+	 struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+	 int64_t ust;
+         
+	 _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
+
+	/*
+	 * The old swapping ioctl was incredibly racy, just wait for vblank
+	 * and do the swap ourselves.
+	 */
+	 driWaitForVBlank(dPriv, &missed_target);
+
+	 /*
+	  * Update each buffer's vbl_pending so we don't get too out of
+	  * sync
+	  */
+	 intel_get_renderbuffer(&intel_fb->Base,
+		   		BUFFER_BACK_LEFT)->vbl_pending = dPriv->vblSeq;
+         intel_get_renderbuffer(&intel_fb->Base,
+		   		BUFFER_FRONT_LEFT)->vbl_pending = dPriv->vblSeq;
+
+	 intelCopyBuffer(dPriv, NULL);
+
+	 intel_fb->swap_count++;
+	 (*psp->systemTime->getUST) (&ust);
+	 if (missed_target) {
+	    intel_fb->swap_missed_count++;
+	    intel_fb->swap_missed_ust = ust - intel_fb->swap_ust;
+	 }
+
+	 intel_fb->swap_ust = ust;
+      }
+      drmCommandNone(intel->driFd, DRM_I915_GEM_THROTTLE);
+   }
+   else {
+      /* XXX this shouldn't be an error but we can't handle it for now */
+      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
+   }
+}
+
+
+/**
+ * Called from driCopySubBuffer()
+ */
+void
+intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h)
+{
+   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
+      struct intel_context *intel =
+         (struct intel_context *) dPriv->driContextPriv->driverPrivate;
+      GLcontext *ctx = &intel->ctx;
+
+      if (ctx->Visual.doubleBufferMode) {
+         drm_clip_rect_t rect;
+         rect.x1 = x + dPriv->x;
+         rect.y1 = (dPriv->h - y - h) + dPriv->y;
+         rect.x2 = rect.x1 + w;
+         rect.y2 = rect.y1 + h;
+         _mesa_notifySwapBuffers(ctx);  /* flush pending rendering comands */
+         intelCopyBuffer(dPriv, &rect);
+      }
+   }
+   else {
+      /* XXX this shouldn't be an error but we can't handle it for now */
+      fprintf(stderr, "%s: drawable has no context!\n", __FUNCTION__);
+   }
+}
+
+
+/**
+ * This will be called whenever the currently bound window is moved/resized.
+ * XXX: actually, it seems to NOT be called when the window is only moved (BP).
+ */
+void
+intelWindowMoved(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   struct intel_framebuffer *intel_fb = dPriv->driverPrivate;
+
+   if (!intel->intelScreen->driScrnPriv->dri2.enabled &&
+       intel->intelScreen->driScrnPriv->ddx_version.minor >= 7) {
+      GLuint flags = intelFixupVblank(intel, dPriv);
+
+      /* Check to see if we changed pipes */
+      if (flags != dPriv->vblFlags && dPriv->vblFlags &&
+	  !(dPriv->vblFlags & VBLANK_FLAG_NO_IRQ)) {
+	 int64_t count;
+	 drmVBlank vbl;
+	 int i;
+
+	 /*
+	  * Deal with page flipping
+	  */
+	 vbl.request.type = DRM_VBLANK_ABSOLUTE;
+
+	 if ( dPriv->vblFlags & VBLANK_FLAG_SECONDARY ) {
+	    vbl.request.type |= DRM_VBLANK_SECONDARY;
+	 }
+
+	 for (i = 0; i < 2; i++) {
+	    if (!intel_fb->color_rb[i] ||
+		(intel_fb->vbl_waited - intel_fb->color_rb[i]->vbl_pending) <=
+		(1<<23))
+	       continue;
+
+	    vbl.request.sequence = intel_fb->color_rb[i]->vbl_pending;
+	    drmWaitVBlank(intel->driFd, &vbl);
+	 }
+
+	 /*
+	  * Update msc_base from old pipe
+	  */
+	 driDrawableGetMSC32(dPriv->driScreenPriv, dPriv, &count);
+	 dPriv->msc_base = count;
+	 /*
+	  * Then get new vblank_base and vblSeq values
+	  */
+	 dPriv->vblFlags = flags;
+	 driGetCurrentVBlank(dPriv);
+	 dPriv->vblank_base = dPriv->vblSeq;
+
+	 intel_fb->vbl_waited = dPriv->vblSeq;
+
+	 for (i = 0; i < 2; i++) {
+	    if (intel_fb->color_rb[i])
+	       intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_waited;
+	 }
+      }
+   } else {
+      dPriv->vblFlags &= ~VBLANK_FLAG_SECONDARY;
+   }
+
+   /* Update Mesa's notion of window size */
+   driUpdateFramebufferSize(ctx, dPriv);
+   intel_fb->Base.Initialized = GL_TRUE; /* XXX remove someday */
+
+   /* Update hardware scissor */
+   if (ctx->Driver.Scissor != NULL) {
+      ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			  ctx->Scissor.Width, ctx->Scissor.Height);
+   }
+
+   /* Re-calculate viewport related state */
+   if (ctx->Driver.DepthRange != NULL)
+      ctx->Driver.DepthRange( ctx, ctx->Viewport.Near, ctx->Viewport.Far );
+}
diff --git a/src/mesa/drivers/dri/intel/intel_swapbuffers.h b/src/mesa/drivers/dri/intel/intel_swapbuffers.h
new file mode 100644
index 00000000000..75bb6242ffc
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_swapbuffers.h
@@ -0,0 +1,52 @@
+
+/**************************************************************************
+ * 
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+#ifndef INTEL_SWAPBUFFERS_H
+#define INTEL_SWAPBUFFERS_H
+
+#include "dri_util.h"
+#include "drm.h"
+
+struct intel_context;
+struct intel_framebuffer;
+
+
+extern void
+intelSwapBuffers(__DRIdrawablePrivate * dPriv);
+
+extern void
+intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h);
+
+extern GLuint
+intelFixupVblank(struct intel_context *intel, __DRIdrawablePrivate *dPriv);
+
+extern void
+intelWindowMoved(struct intel_context *intel);
+
+
+#endif /* INTEL_SWAPBUFFERS_H */
diff --git a/src/mesa/drivers/dri/intel/intel_tex.c b/src/mesa/drivers/dri/intel/intel_tex.c
index 23455a4c745..ae0994b183a 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.c
+++ b/src/mesa/drivers/dri/intel/intel_tex.c
@@ -93,7 +93,7 @@ intelFreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage)
 static void *
 do_memcpy(void *dest, const void *src, size_t n)
 {
-   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63)) {
+   if ((((unsigned long) src) & 63) || (((unsigned long) dest) & 63)) {
       return __memcpy(dest, src, n);
    }
    else
@@ -222,22 +222,16 @@ intelInitTextureFuncs(struct dd_function_table *functions)
    functions->TexSubImage1D = intelTexSubImage1D;
    functions->TexSubImage2D = intelTexSubImage2D;
    functions->TexSubImage3D = intelTexSubImage3D;
-#ifdef I915
    functions->CopyTexImage1D = intelCopyTexImage1D;
    functions->CopyTexImage2D = intelCopyTexImage2D;
    functions->CopyTexSubImage1D = intelCopyTexSubImage1D;
    functions->CopyTexSubImage2D = intelCopyTexSubImage2D;
-#else
-   functions->CopyTexImage1D = _swrast_copy_teximage1d;
-   functions->CopyTexImage2D = _swrast_copy_teximage2d;
-   functions->CopyTexSubImage1D = _swrast_copy_texsubimage1d;
-   functions->CopyTexSubImage2D = _swrast_copy_texsubimage2d;
-#endif
    functions->GetTexImage = intelGetTexImage;
    functions->GenerateMipmap = intelGenerateMipmap;
 
    /* compressed texture functions */
    functions->CompressedTexImage2D = intelCompressedTexImage2D;
+   functions->CompressedTexSubImage2D = intelCompressedTexSubImage2D;
    functions->GetCompressedTexImage = intelGetCompressedTexImage;
 
    functions->NewTextureObject = intelNewTextureObject;
diff --git a/src/mesa/drivers/dri/intel/intel_tex.h b/src/mesa/drivers/dri/intel/intel_tex.h
index 6219c1c953b..742ccc043aa 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.h
+++ b/src/mesa/drivers/dri/intel/intel_tex.h
@@ -130,6 +130,16 @@ void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
 				struct gl_texture_object *texObj,
 				struct gl_texture_image *texImage );
 
+void intelCompressedTexSubImage2D(GLcontext * ctx,
+				  GLenum target,
+				  GLint level,
+				  GLint xoffset, GLint yoffset,
+				  GLsizei width, GLsizei height,
+				  GLenum format, GLsizei imageSize,
+				  const GLvoid * pixels,
+				  struct gl_texture_object *texObj,
+				  struct gl_texture_image *texImage);
+
 void intelGetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
 				GLvoid *pixels,
 				struct gl_texture_object *texObj,
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index f4cb4a781c6..08437aa0e2b 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -98,7 +98,9 @@ do_copy_texsubimage(struct intel_context *intel,
       get_teximage_source(intel, internalFormat);
 
    if (!intelImage->mt || !src) {
-      DBG("%s fail %p %p\n", __FUNCTION__, intelImage->mt, src);
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "%s fail %p %p\n",
+		 __FUNCTION__, intelImage->mt, src);
       return GL_FALSE;
    }
 
@@ -110,56 +112,53 @@ do_copy_texsubimage(struct intel_context *intel,
                                                        intelImage->level);
       const GLint orig_x = x;
       const GLint orig_y = y;
-      const struct gl_framebuffer *fb = ctx->DrawBuffer;
-
-      if (_mesa_clip_to_region(fb->_Xmin, fb->_Ymin, fb->_Xmax, fb->_Ymax,
-                               &x, &y, &width, &height)) {
-         /* Update dst for clipped src.  Need to also clip the source rect.
-          */
-         dstx += x - orig_x;
-         dsty += y - orig_y;
-
-         if (ctx->ReadBuffer->Name == 0) {
-            /* reading from a window, adjust x, y */
-            __DRIdrawablePrivate *dPriv = intel->driDrawable;
-            GLuint window_y;
-            /* window_y = position of window on screen if y=0=bottom */
-            window_y = intel->intelScreen->height - (dPriv->y + dPriv->h);
-            y = window_y + y;
-            x += dPriv->x;
-         }
-         else {
-            /* reading from a FBO */
-            /* invert Y */
-            y = ctx->ReadBuffer->Height - y - 1;
-         }
-
-
-         /* A bit of fiddling to get the blitter to work with -ve
-          * pitches.  But we get a nice inverted blit this way, so it's
-          * worth it:
-          */
-         intelEmitCopyBlit(intel,
-                           intelImage->mt->cpp,
-                           -src->pitch,
-                           src->buffer,
-                           src->height * src->pitch * src->cpp,
-			   src->tiling,
-                           intelImage->mt->pitch,
-                           intelImage->mt->region->buffer,
-                           image_offset,
-			   intelImage->mt->region->tiling,
-                           x, y + height, dstx, dsty, width, height,
-			   GL_COPY); /* ? */
+      GLshort src_pitch;
+
+      /* Update dst for clipped src.  Need to also clip the source rect. */
+      dstx += x - orig_x;
+      dsty += y - orig_y;
+
+      /* image_offset may be non-page-aligned, but that's illegal for tiling. */
+      assert(intelImage->mt->region->tiling == I915_TILING_NONE);
+
+      if (ctx->ReadBuffer->Name == 0) {
+	 /* reading from a window, adjust x, y */
+	 __DRIdrawablePrivate *dPriv = intel->driDrawable;
+	 y = dPriv->y + (dPriv->h - (y + height));
+	 x += dPriv->x;
+
+	 /* Invert the data coming from the source rectangle due to GL
+	  * and hardware disagreeing on where y=0 is.
+	  *
+	  * It appears that our offsets and pitches get mangled
+	  * appropriately by the hardware, and we don't need to adjust them
+	  * on our own.
+	  */
+	 src_pitch = -src->pitch;
+      } else {
+	 /* reading from a FBO, y is already oriented the way we like */
+	 src_pitch = src->pitch;
       }
-   }
 
+      intelEmitCopyBlit(intel,
+			intelImage->mt->cpp,
+			src_pitch,
+			src->buffer,
+			0,
+			src->tiling,
+			intelImage->mt->pitch,
+			intelImage->mt->region->buffer,
+			image_offset,
+			intelImage->mt->region->tiling,
+			x, y, dstx, dsty, width, height,
+			GL_COPY);
+   }
 
    UNLOCK_HARDWARE(intel);
 
    /* GL_SGIS_generate_mipmap */
    if (intelImage->level == texObj->BaseLevel && texObj->GenerateMipmap) {
-      intel_generate_mipmap(ctx, target, texObj);
+      ctx->Driver.GenerateMipmap(ctx, target, texObj);
    }
 
    return GL_TRUE;
@@ -180,6 +179,7 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
       _mesa_select_tex_image(ctx, texObj, target, level);
+   int srcx, srcy, dstx, dsty, height;
 
    if (border)
       goto fail;
@@ -191,10 +191,20 @@ intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
                           width, border,
                           GL_RGBA, CHAN_TYPE, NULL,
                           &ctx->DefaultPacking, texObj, texImage);
+   srcx = x;
+   srcy = y;
+   dstx = 0;
+   dsty = 0;
+   height = 1;
+   if (!_mesa_clip_copytexsubimage(ctx,
+				   &dstx, &dsty,
+				   &srcx, &srcy,
+				   &width, &height))
+      return;
 
    if (!do_copy_texsubimage(intel_context(ctx), target,
                             intel_texture_image(texImage),
-                            internalFormat, 0, 0, x, y, width, 1))
+                            internalFormat, 0, 0, x, y, width, height))
       goto fail;
 
    return;
@@ -216,10 +226,21 @@ intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
       _mesa_select_tex_object(ctx, texUnit, target);
    struct gl_texture_image *texImage =
       _mesa_select_tex_image(ctx, texObj, target, level);
+   int srcx, srcy, dstx, dsty;
 
    if (border)
       goto fail;
 
+   srcx = x;
+   srcy = y;
+   dstx = 0;
+   dsty = 0;
+   if (!_mesa_clip_copytexsubimage(ctx,
+				   &dstx, &dsty,
+				   &srcx, &srcy,
+				   &width, &height))
+      return;
+
    /* Setup or redefine the texture object, mipmap tree and texture
     * image.  Don't populate yet.  
     */
diff --git a/src/mesa/drivers/dri/intel/intel_tex_format.c b/src/mesa/drivers/dri/intel/intel_tex_format.c
index 2be060dd3e3..2715a540d02 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_format.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_format.c
@@ -16,7 +16,7 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
                          GLenum format, GLenum type)
 {
    struct intel_context *intel = intel_context(ctx);
-   const GLboolean do32bpt = (intel->ctx.Visual.rgbBits == 32);
+   const GLboolean do32bpt = (intel->ctx.Visual.rgbBits >= 24);
 
    switch (internalFormat) {
    case 4:
@@ -134,8 +134,14 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_DEPTH_COMPONENT16:
    case GL_DEPTH_COMPONENT24:
    case GL_DEPTH_COMPONENT32:
+#if 0
       return &_mesa_texformat_z16;
-
+#else
+      /* fall-through.
+       * 16bpp depth texture can't be paired with a stencil buffer so
+       * always used combined depth/stencil format.
+       */
+#endif
    case GL_DEPTH_STENCIL_EXT:
    case GL_DEPTH24_STENCIL8_EXT:
       return &_mesa_texformat_s8_z24;
@@ -158,7 +164,7 @@ intelChooseTextureFormat(GLcontext * ctx, GLint internalFormat,
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
    case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
-     return &_mesa_texformat_srgb_dxt1;
+      return &_mesa_texformat_srgb_dxt1;
 #endif
 
    default:
diff --git a/src/mesa/drivers/dri/intel/intel_tex_image.c b/src/mesa/drivers/dri/intel/intel_tex_image.c
index 2ac7dceb0fc..866022d0ce2 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_image.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_image.c
@@ -209,7 +209,8 @@ try_pbo_upload(struct intel_context *intel,
       return GL_FALSE;
    }
 
-   src_offset = (GLuint) pixels;
+   /* note: potential 64-bit ptr to 32-bit int cast */
+   src_offset = (GLuint) (unsigned long) pixels;
 
    if (unpack->RowLength > 0)
       src_stride = unpack->RowLength;
@@ -264,7 +265,8 @@ try_pbo_zcopy(struct intel_context *intel,
       return GL_FALSE;
    }
 
-   src_offset = (GLuint) pixels;
+   /* note: potential 64-bit ptr to 32-bit int cast */
+   src_offset = (GLuint) (unsigned long) pixels;
 
    if (unpack->RowLength > 0)
       src_stride = unpack->RowLength;
diff --git a/src/mesa/drivers/dri/intel/intel_tex_subimage.c b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
index b7523618868..f86de568976 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_subimage.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_subimage.c
@@ -184,3 +184,18 @@ intelTexSubImage1D(GLcontext * ctx,
                     format, type, pixels, packing, texObj, texImage);
 
 }
+
+void
+intelCompressedTexSubImage2D(GLcontext * ctx,
+			     GLenum target,
+			     GLint level,
+			     GLint xoffset, GLint yoffset,
+			     GLsizei width, GLsizei height,
+			     GLenum format, GLsizei imageSize,
+			     const GLvoid * pixels,
+			     struct gl_texture_object *texObj,
+			     struct gl_texture_image *texImage)
+{
+   fprintf(stderr, "stubbed CompressedTexSubImage2D: %dx%d@%dx%d\n",
+	   width, height, xoffset, yoffset);
+}
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.c b/src/mesa/drivers/dri/mach64/mach64_context.c
index 99abd209b69..9c7f513c6f5 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.c
+++ b/src/mesa/drivers/dri/mach64/mach64_context.c
@@ -57,10 +57,6 @@
 #include "utils.h"
 #include "vblank.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_vertex_buffer_object
-#include "extension_helper.h"
-
 #ifndef MACH64_DEBUG
 int MACH64_DEBUG = (0);
 #endif
@@ -82,9 +78,7 @@ static const struct dri_debug_control debug_control[] =
 
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_texture_edge_clamp",         NULL },
     { "GL_MESA_ycbcr_texture",             NULL },
     { "GL_SGIS_generate_mipmap",           NULL },
@@ -196,6 +190,7 @@ GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
    ctx->Const.MaxTextureUnits = 2;
    ctx->Const.MaxTextureImageUnits = 2;
    ctx->Const.MaxTextureCoordUnits = 2;
+   ctx->Const.MaxDrawBuffers = 1;
 
    heap = mach64Screen->IsPCI ? MACH64_CARD_HEAP : MACH64_AGP_HEAP;
 
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.h b/src/mesa/drivers/dri/mach64/mach64_context.h
index 55e0618ff80..854751626d0 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.h
+++ b/src/mesa/drivers/dri/mach64/mach64_context.h
@@ -294,7 +294,13 @@ extern GLboolean mach64UnbindContext( __DRIcontextPrivate *driContextPriv );
 #define LE32_OUT( x, y )	do { *(GLuint *)(x) = (y); } while (0)
 #define LE32_OUT_FLOAT( x, y )	do { *(GLfloat *)(x) = (y); } while (0)
 #else
+#ifndef __OpenBSD__
 #include <byteswap.h>
+#else
+#include <machine/endian.h>
+#define bswap_32 bswap32
+#endif
+
 #define LE32_IN( x )		bswap_32( *(GLuint *)(x) )
 #define LE32_IN_FLOAT( x )						\
 ({									\
diff --git a/src/mesa/drivers/dri/mach64/mach64_screen.c b/src/mesa/drivers/dri/mach64/mach64_screen.c
index 6bfb4c32b15..43e59594dd5 100644
--- a/src/mesa/drivers/dri/mach64/mach64_screen.c
+++ b/src/mesa/drivers/dri/mach64/mach64_screen.c
@@ -93,6 +93,7 @@ mach64FillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[2];
     uint8_t stencil_bits_array[2];
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
@@ -104,6 +105,8 @@ mach64FillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[0] = 0;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
@@ -119,7 +122,8 @@ mach64FillInModes( __DRIscreenPrivate *psp,
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor, back_buffer_modes,
-			       back_buffer_factor);
+			       back_buffer_factor,
+                               msaa_samples_array, 1);
     if (configs == NULL) {
        fprintf(stderr, "[%s:%u] Error creating FBConfig!\n",
 	       __func__, __LINE__);
diff --git a/src/mesa/drivers/dri/mach64/mach64_tex.c b/src/mesa/drivers/dri/mach64/mach64_tex.c
index 1f9d3c57ebe..9fe267eafd5 100644
--- a/src/mesa/drivers/dri/mach64/mach64_tex.c
+++ b/src/mesa/drivers/dri/mach64/mach64_tex.c
@@ -152,6 +152,7 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_ALPHA8:
    case GL_ALPHA12:
    case GL_ALPHA16:
+   case GL_COMPRESSED_ALPHA:
    case 2:
    case GL_LUMINANCE_ALPHA:
    case GL_LUMINANCE4_ALPHA4:
@@ -160,9 +161,11 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE12_ALPHA4:
    case GL_LUMINANCE12_ALPHA12:
    case GL_LUMINANCE16_ALPHA16:
+   case GL_COMPRESSED_LUMINANCE_ALPHA:
    case 4:
    case GL_RGBA:
    case GL_RGBA2:
+   case GL_COMPRESSED_RGBA:
       if (mmesa->mach64Screen->cpp == 4)
          return &_mesa_texformat_argb8888;
       else
@@ -193,6 +196,7 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_RGB10:
    case GL_RGB12:
    case GL_RGB16:
+   case GL_COMPRESSED_RGB:
       if (mmesa->mach64Screen->cpp == 4)
          return &_mesa_texformat_argb8888;
       else
@@ -204,6 +208,7 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_LUMINANCE8:
    case GL_LUMINANCE12:
    case GL_LUMINANCE16:
+   case GL_COMPRESSED_LUMINANCE:
       if (mmesa->mach64Screen->cpp == 4)
          return &_mesa_texformat_argb8888; /* inefficient but accurate */
       else
@@ -214,6 +219,7 @@ mach64ChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_INTENSITY8:
    case GL_INTENSITY12:
    case GL_INTENSITY16:
+   case GL_COMPRESSED_INTENSITY:
       if (mmesa->mach64Screen->cpp == 4)
          return &_mesa_texformat_argb8888; /* inefficient but accurate */
       else
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.c b/src/mesa/drivers/dri/mga/mga_xmesa.c
index 86da3a2cacf..926534d6dff 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.c
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.c
@@ -69,13 +69,9 @@
 
 #include "GL/internal/dri_interface.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_gpu_program_parameters
-#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
 #if 0
 #define need_GL_EXT_paletted_texture
@@ -133,6 +129,7 @@ mgaFillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[3];
     uint8_t stencil_bits_array[3];
+    uint8_t msaa_samples_array[1];
 
 
     depth_bits_array[0] = 0;
@@ -147,6 +144,8 @@ mgaFillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[1] = 0;
     stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
@@ -162,7 +161,8 @@ mgaFillInModes( __DRIscreenPrivate *psp,
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor,
-			       back_buffer_modes, back_buffer_factor);
+			       back_buffer_modes, back_buffer_factor,
+                               msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
@@ -385,13 +385,9 @@ static const struct dri_extension g400_extensions[] =
 
 static const struct dri_extension card_extensions[] =
 {
-   { "GL_ARB_multisample",            GL_ARB_multisample_functions },
-   { "GL_ARB_texture_compression",    GL_ARB_texture_compression_functions },
    { "GL_ARB_texture_rectangle",      NULL },
-   { "GL_ARB_vertex_buffer_object",   GL_ARB_vertex_buffer_object_functions },
    { "GL_EXT_blend_logic_op",         NULL },
    { "GL_EXT_fog_coord",              GL_EXT_fog_coord_functions },
-   { "GL_EXT_multi_draw_arrays",      GL_EXT_multi_draw_arrays_functions },
    /* paletted_textures currently doesn't work, but we could fix them later */
 #if defined( need_GL_EXT_paletted_texture )
    { "GL_EXT_shared_texture_palette", NULL },
@@ -539,6 +535,8 @@ mgaCreateContext( const __GLcontextModes *mesaVis,
    ctx->Const.MaxLineWidthAA = 10.0;
    ctx->Const.LineWidthGranularity = 1.0;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    mmesa->texture_depth = driQueryOptioni (&mmesa->optionCache,
 					   "texture_depth");
    if (mmesa->texture_depth == DRI_CONF_TEXTURE_DEPTH_FB)
diff --git a/src/mesa/drivers/dri/r128/r128_context.c b/src/mesa/drivers/dri/r128/r128_context.c
index 535a98cc01f..f511a67bada 100644
--- a/src/mesa/drivers/dri/r128/r128_context.c
+++ b/src/mesa/drivers/dri/r128/r128_context.c
@@ -65,9 +65,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 int R128_DEBUG = 0;
 #endif
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
@@ -75,12 +72,9 @@ int R128_DEBUG = 0;
 
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_env_add",            NULL },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_blend_subtract",             GL_EXT_blend_minmax_functions },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
     { "GL_EXT_texture_edge_clamp",         NULL },
@@ -229,6 +223,8 @@ GLboolean r128CreateContext( const __GLcontextModes *glVisual,
    ctx->Const.MaxLineWidthAA = 1.0;
    ctx->Const.LineWidthGranularity = 1.0;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
 #if ENABLE_PERF_BOXES
    rmesa->boxes = driQueryOptionb(&rmesa->optionCache, "performance_boxes");
 #endif
diff --git a/src/mesa/drivers/dri/r128/r128_screen.c b/src/mesa/drivers/dri/r128/r128_screen.c
index cb3a147dbac..7cda4ca5d38 100644
--- a/src/mesa/drivers/dri/r128/r128_screen.c
+++ b/src/mesa/drivers/dri/r128/r128_screen.c
@@ -422,7 +422,7 @@ r128FillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[2];
     uint8_t stencil_bits_array[2];
-
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
@@ -434,6 +434,8 @@ r128FillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[0] = 0;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
@@ -446,26 +448,27 @@ r128FillInModes( __DRIscreenPrivate *psp,
         fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
     }
 
-   configs = driCreateConfigs(fb_format, fb_type,
-			      depth_bits_array, stencil_bits_array,
-			      depth_buffer_factor, back_buffer_modes,
-			      back_buffer_factor);
-   if (configs == NULL) {
-    fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
-              __LINE__);
-      return NULL;
-   }
+    configs = driCreateConfigs(fb_format, fb_type,
+                               depth_bits_array, stencil_bits_array,
+                               depth_buffer_factor, back_buffer_modes,
+                               back_buffer_factor,
+                               msaa_samples_array, 1);
+    if (configs == NULL) {
+        fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+                __LINE__);
+        return NULL;
+    }
 
-   /* Mark the visual as slow if there are "fake" stencil bits.
-    */
-   for (i = 0; configs[i]; i++) {
-      m = &configs[i]->modes;
-      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
-         m->visualRating = GLX_SLOW_CONFIG;
-      }
-   }
+    /* Mark the visual as slow if there are "fake" stencil bits.
+     */
+    for (i = 0; configs[i]; i++) {
+        m = &configs[i]->modes;
+        if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+            m->visualRating = GLX_SLOW_CONFIG;
+        }
+    }
 
-   return (const __DRIconfig **) configs;
+    return (const __DRIconfig **) configs;
 }
 
 
diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c
index 1f5a9f2427b..431bf8235c3 100644
--- a/src/mesa/drivers/dri/r200/r200_context.c
+++ b/src/mesa/drivers/dri/r200/r200_context.c
@@ -63,14 +63,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "radeon_span.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
 #define need_GL_ATI_fragment_shader
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
-#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
 #define need_GL_EXT_blend_equation_separate
 #define need_GL_EXT_blend_func_separate
@@ -118,20 +114,16 @@ static const GLubyte *r200GetString( GLcontext *ctx, GLenum name )
  */
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_border_clamp",       NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_env_add",            NULL },
     { "GL_ARB_texture_env_combine",        NULL },
     { "GL_ARB_texture_env_dot3",           NULL },
     { "GL_ARB_texture_env_crossbar",       NULL },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_blend_minmax",               GL_EXT_blend_minmax_functions },
     { "GL_EXT_blend_subtract",             NULL },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
-    { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
     { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
     { "GL_EXT_stencil_wrap",               NULL },
     { "GL_EXT_texture_edge_clamp",         NULL },
@@ -426,6 +418,8 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
    ctx->Const.VertexProgram.MaxNativeParameters = R200_VSF_MAX_PARAM;
    ctx->Const.VertexProgram.MaxNativeAddressRegs = 1;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 8bad6467f2d..6d68309d3f0 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -76,19 +76,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 int future_hw_tcl_on = 1;
 int hw_tcl_on = 1;
 
-#define need_GL_EXT_stencil_two_side
-#define need_GL_ARB_multisample
+#define need_GL_VERSION_2_0
 #define need_GL_ARB_point_parameters
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_ARB_vertex_program
-#define need_GL_EXT_blend_minmax
-//#define need_GL_EXT_fog_coord
-#define need_GL_EXT_multi_draw_arrays
-#define need_GL_EXT_secondary_color
 #define need_GL_EXT_blend_equation_separate
 #define need_GL_EXT_blend_func_separate
+#define need_GL_EXT_blend_minmax
+//#define need_GL_EXT_fog_coord
 #define need_GL_EXT_gpu_program_parameters
+#define need_GL_EXT_secondary_color
+#define need_GL_EXT_stencil_two_side
+#define need_GL_ATI_separate_stencil
 #define need_GL_NV_vertex_program
 #include "extension_helper.h"
 
@@ -96,27 +94,23 @@ const struct dri_extension card_extensions[] = {
   /* *INDENT-OFF* */
   {"GL_ARB_depth_texture",		NULL},
   {"GL_ARB_fragment_program",		NULL},
-  {"GL_ARB_multisample",		GL_ARB_multisample_functions},
   {"GL_ARB_multitexture",		NULL},
   {"GL_ARB_point_parameters",		GL_ARB_point_parameters_functions},
   {"GL_ARB_shadow",			NULL},
   {"GL_ARB_shadow_ambient",		NULL},
   {"GL_ARB_texture_border_clamp",	NULL},
-  {"GL_ARB_texture_compression",	GL_ARB_texture_compression_functions},
   {"GL_ARB_texture_cube_map",		NULL},
   {"GL_ARB_texture_env_add",		NULL},
   {"GL_ARB_texture_env_combine",	NULL},
   {"GL_ARB_texture_env_crossbar",	NULL},
   {"GL_ARB_texture_env_dot3",		NULL},
   {"GL_ARB_texture_mirrored_repeat",	NULL},
-  {"GL_ARB_vertex_buffer_object",	GL_ARB_vertex_buffer_object_functions},
   {"GL_ARB_vertex_program",		GL_ARB_vertex_program_functions},
   {"GL_EXT_blend_equation_separate",	GL_EXT_blend_equation_separate_functions},
   {"GL_EXT_blend_func_separate",	GL_EXT_blend_func_separate_functions},
   {"GL_EXT_blend_minmax",		GL_EXT_blend_minmax_functions},
   {"GL_EXT_blend_subtract",		NULL},
 //  {"GL_EXT_fog_coord",			GL_EXT_fog_coord_functions },
-  {"GL_EXT_multi_draw_arrays",		GL_EXT_multi_draw_arrays_functions},
   {"GL_EXT_gpu_program_parameters",     GL_EXT_gpu_program_parameters_functions},
   {"GL_EXT_secondary_color", 		GL_EXT_secondary_color_functions},
   {"GL_EXT_shadow_funcs",		NULL},
@@ -129,6 +123,7 @@ const struct dri_extension card_extensions[] = {
   {"GL_EXT_texture_lod_bias",		NULL},
   {"GL_EXT_texture_mirror_clamp",	NULL},
   {"GL_EXT_texture_rectangle",		NULL},
+  {"GL_ATI_separate_stencil",		GL_ATI_separate_stencil_functions},
   {"GL_ATI_texture_env_combine3",	NULL},
   {"GL_ATI_texture_mirror_once",	NULL},
   {"GL_MESA_pack_invert",		NULL},
@@ -141,6 +136,16 @@ const struct dri_extension card_extensions[] = {
   /* *INDENT-ON* */
 };
 
+
+/**
+ * The GL 2.0 functions are needed to make display lists work with
+ * functions added by GL_ATI_separate_stencil.
+ */
+const struct dri_extension gl_20_extension[] = {
+  {"GL_VERSION_2_0",			GL_VERSION_2_0_functions },
+};
+
+
 extern struct tnl_pipeline_stage _r300_render_stage;
 extern const struct tnl_pipeline_stage _r300_tcl_stage;
 
@@ -346,6 +351,8 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	    ( /*512 */ RADEON_BUFFER_SIZE * 16 * 1024) / (4 * 4);
 #endif
 
+	ctx->Const.MaxDrawBuffers = 1;
+
 	/* Initialize the software rasterizer and helper modules.
 	 */
 	_swrast_CreateContext(ctx);
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index 8b3fe431ab3..5f344be1163 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -64,7 +64,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_SE_VPORT_ZSCALE                0x1DA8
 #define R300_SE_VPORT_ZOFFSET               0x1DAC
 
-
+#define R300_VAP_PORT_IDX0		    0x2040
 /*
  * Vertex Array Processing (VAP) Control
  */
@@ -656,7 +656,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_GB_FOG_SELECT_C3A           (3 << 0)
 #	define R300_GB_FOG_SELECT_1_1_W         (4 << 0)
 #	define R300_GB_FOG_SELECT_Z		(5 << 0)
-#	define R300_GB_DEPTH_SELECT_Z		(0 << 3
+#	define R300_GB_DEPTH_SELECT_Z		(0 << 3)
 #	define R300_GB_DEPTH_SELECT_1_1_W	(1 << 3)
 #	define R300_GB_W_SELECT_1_W		(0 << 4)
 #	define R300_GB_W_SELECT_1		(1 << 4)
@@ -730,8 +730,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R500_RS_IP_TEX_PTR_Q_SHIFT 			18
 #define R500_RS_IP_COL_PTR_SHIFT 			24
 #define R500_RS_IP_COL_FMT_SHIFT 			27
-#	define R500_RS_COL_PTR(x)		        (x << 24)
-#       define R500_RS_COL_FMT(x)                       (x << 27)
+#	define R500_RS_COL_PTR(x)		        ((x) << 24)
+#       define R500_RS_COL_FMT(x)                       ((x) << 27)
 /* gap */
 #define R500_RS_IP_OFFSET_DIS 				(0 << 31)
 #define R500_RS_IP_OFFSET_EN 				(1 << 31)
@@ -1172,9 +1172,9 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_RS_IP_3				        0x431C
 #       define R300_RS_INTERP_SRC_SHIFT          2 /* TODO: check for removal */
 #       define R300_RS_INTERP_SRC_MASK           (7 << 2) /* TODO: check for removal */
-#	define R300_RS_TEX_PTR(x)		        (x << 0)
-#	define R300_RS_COL_PTR(x)		        (x << 6)
-#	define R300_RS_COL_FMT(x)		        (x << 9)
+#	define R300_RS_TEX_PTR(x)		        ((x) << 0)
+#	define R300_RS_COL_PTR(x)		        ((x) << 6)
+#	define R300_RS_COL_FMT(x)		        ((x) << 9)
 #	define R300_RS_COL_FMT_RGBA		        0
 #	define R300_RS_COL_FMT_RGB0		        1
 #	define R300_RS_COL_FMT_RGB1		        2
@@ -1184,10 +1184,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_RS_COL_FMT_111A		        8
 #	define R300_RS_COL_FMT_1110		        9
 #	define R300_RS_COL_FMT_1111		        10
-#	define R300_RS_SEL_S(x)		                (x << 13)
-#	define R300_RS_SEL_T(x)		                (x << 16)
-#	define R300_RS_SEL_R(x)		                (x << 19)
-#	define R300_RS_SEL_Q(x)		                (x << 22)
+#	define R300_RS_SEL_S(x)		                ((x) << 13)
+#	define R300_RS_SEL_T(x)		                ((x) << 16)
+#	define R300_RS_SEL_R(x)		                ((x) << 19)
+#	define R300_RS_SEL_Q(x)		                ((x) << 22)
 #	define R300_RS_SEL_C0		                0
 #	define R300_RS_SEL_C1		                1
 #	define R300_RS_SEL_C2		                2
@@ -2712,7 +2712,7 @@ enum {
 #   define R500_ALPHA_OP_COS				13
 #   define R500_ALPHA_OP_MDH				14
 #   define R500_ALPHA_OP_MDV				15
-#   define R500_ALPHA_ADDRD(x)				(x << 4)
+#   define R500_ALPHA_ADDRD(x)				((x) << 4)
 #   define R500_ALPHA_ADDRD_REL				(1 << 11)
 #  define R500_ALPHA_SEL_A_SHIFT			12
 #   define R500_ALPHA_SEL_A_SRC0			(0 << 12)
@@ -2756,16 +2756,16 @@ enum {
 #   define R500_ALPHA_OMOD_DIV_4			(5 << 26)
 #   define R500_ALPHA_OMOD_DIV_8			(6 << 26)
 #   define R500_ALPHA_OMOD_DISABLE			(7 << 26)
-#   define R500_ALPHA_TARGET(x)				(x << 29)
+#   define R500_ALPHA_TARGET(x)				((x) << 29)
 #   define R500_ALPHA_W_OMASK				(1 << 31)
 #define R500_US_ALU_ALPHA_ADDR_0			0x9800
-#   define R500_ALPHA_ADDR0(x)				(x << 0)
+#   define R500_ALPHA_ADDR0(x)				((x) << 0)
 #   define R500_ALPHA_ADDR0_CONST			(1 << 8)
 #   define R500_ALPHA_ADDR0_REL				(1 << 9)
-#   define R500_ALPHA_ADDR1(x)				(x << 10)
+#   define R500_ALPHA_ADDR1(x)				((x) << 10)
 #   define R500_ALPHA_ADDR1_CONST			(1 << 18)
 #   define R500_ALPHA_ADDR1_REL				(1 << 19)
-#   define R500_ALPHA_ADDR2(x)				(x << 20)
+#   define R500_ALPHA_ADDR2(x)				((x) << 20)
 #   define R500_ALPHA_ADDR2_CONST			(1 << 28)
 #   define R500_ALPHA_ADDR2_REL				(1 << 29)
 #   define R500_ALPHA_SRCP_OP_1_MINUS_2A0		(0 << 30)
@@ -2786,7 +2786,7 @@ enum {
 #   define R500_ALU_RGBA_OP_SOP				(10 << 0)
 #   define R500_ALU_RGBA_OP_MDH				(11 << 0)
 #   define R500_ALU_RGBA_OP_MDV				(12 << 0)
-#   define R500_ALU_RGBA_ADDRD(x)			(x << 4)
+#   define R500_ALU_RGBA_ADDRD(x)			((x) << 4)
 #   define R500_ALU_RGBA_ADDRD_REL			(1 << 11)
 #  define R500_ALU_RGBA_SEL_C_SHIFT			12
 #   define R500_ALU_RGBA_SEL_C_SRC0			(0 << 12)
@@ -2913,16 +2913,16 @@ enum {
 #   define R500_ALU_RGB_OMOD_DIV_4			(5 << 26)
 #   define R500_ALU_RGB_OMOD_DIV_8			(6 << 26)
 #   define R500_ALU_RGB_OMOD_DISABLE			(7 << 26)
-#   define R500_ALU_RGB_TARGET(x)			(x << 29)
+#   define R500_ALU_RGB_TARGET(x)			((x) << 29)
 #   define R500_ALU_RGB_WMASK				(1 << 31)
 #define R500_US_ALU_RGB_ADDR_0				0x9000
-#   define R500_RGB_ADDR0(x)				(x << 0)
+#   define R500_RGB_ADDR0(x)				((x) << 0)
 #   define R500_RGB_ADDR0_CONST				(1 << 8)
 #   define R500_RGB_ADDR0_REL				(1 << 9)
-#   define R500_RGB_ADDR1(x)				(x << 10)
+#   define R500_RGB_ADDR1(x)				((x) << 10)
 #   define R500_RGB_ADDR1_CONST				(1 << 18)
 #   define R500_RGB_ADDR1_REL				(1 << 19)
-#   define R500_RGB_ADDR2(x)				(x << 20)
+#   define R500_RGB_ADDR2(x)				((x) << 20)
 #   define R500_RGB_ADDR2_CONST				(1 << 28)
 #   define R500_RGB_ADDR2_REL				(1 << 29)
 #   define R500_RGB_SRCP_OP_1_MINUS_2RGB0		(0 << 30)
@@ -2977,19 +2977,19 @@ enum {
 
 /* note that these are 8 bit lengths, despite the offsets, at least for R500 */
 #define R500_US_CODE_ADDR				0x4630
-#   define R500_US_CODE_START_ADDR(x)			(x << 0)
-#   define R500_US_CODE_END_ADDR(x)			(x << 16)
+#   define R500_US_CODE_START_ADDR(x)			((x) << 0)
+#   define R500_US_CODE_END_ADDR(x)			((x) << 16)
 #define R500_US_CODE_OFFSET				0x4638
-#   define R500_US_CODE_OFFSET_ADDR(x)			(x << 0)
+#   define R500_US_CODE_OFFSET_ADDR(x)			((x) << 0)
 #define R500_US_CODE_RANGE				0x4634
-#   define R500_US_CODE_RANGE_ADDR(x)			(x << 0)
-#   define R500_US_CODE_RANGE_SIZE(x)			(x << 16)
+#   define R500_US_CODE_RANGE_ADDR(x)			((x) << 0)
+#   define R500_US_CODE_RANGE_SIZE(x)			((x) << 16)
 #define R500_US_CONFIG					0x4600
 #   define R500_ZERO_TIMES_ANYTHING_EQUALS_ZERO		(1 << 1)
 #define R500_US_FC_ADDR_0				0xa000
-#   define R500_FC_BOOL_ADDR(x)				(x << 0)
-#   define R500_FC_INT_ADDR(x)				(x << 8)
-#   define R500_FC_JUMP_ADDR(x)				(x << 16)
+#   define R500_FC_BOOL_ADDR(x)				((x) << 0)
+#   define R500_FC_INT_ADDR(x)				((x) << 8)
+#   define R500_FC_JUMP_ADDR(x)				((x) << 16)
 #   define R500_FC_JUMP_GLOBAL				(1 << 31)
 #define R500_US_FC_BOOL_CONST				0x4620
 #   define R500_FC_KBOOL(x)				(x)
@@ -3010,8 +3010,8 @@ enum {
 #   define R500_FC_A_OP_NONE				(0 << 6)
 #   define R500_FC_A_OP_POP				(1 << 6)
 #   define R500_FC_A_OP_PUSH				(2 << 6)
-#   define R500_FC_JUMP_FUNC(x)				(x << 8)
-#   define R500_FC_B_POP_CNT(x)				(x << 16)
+#   define R500_FC_JUMP_FUNC(x)				((x) << 8)
+#   define R500_FC_B_POP_CNT(x)				((x) << 16)
 #   define R500_FC_B_OP0_NONE				(0 << 24)
 #   define R500_FC_B_OP0_DECR				(1 << 24)
 #   define R500_FC_B_OP0_INCR				(2 << 24)
@@ -3020,14 +3020,14 @@ enum {
 #   define R500_FC_B_OP1_INCR				(2 << 26)
 #   define R500_FC_IGNORE_UNCOVERED			(1 << 28)
 #define R500_US_FC_INT_CONST_0				0x4c00
-#   define R500_FC_INT_CONST_KR(x)			(x << 0)
-#   define R500_FC_INT_CONST_KG(x)			(x << 8)
-#   define R500_FC_INT_CONST_KB(x)			(x << 16)
+#   define R500_FC_INT_CONST_KR(x)			((x) << 0)
+#   define R500_FC_INT_CONST_KG(x)			((x) << 8)
+#   define R500_FC_INT_CONST_KB(x)			((x) << 16)
 /* _0 through _15 */
 #define R500_US_FORMAT0_0				0x4640
-#   define R500_FORMAT_TXWIDTH(x)			(x << 0)
-#   define R500_FORMAT_TXHEIGHT(x)			(x << 11)
-#   define R500_FORMAT_TXDEPTH(x)			(x << 22)
+#   define R500_FORMAT_TXWIDTH(x)			((x) << 0)
+#   define R500_FORMAT_TXHEIGHT(x)			((x) << 11)
+#   define R500_FORMAT_TXDEPTH(x)			((x) << 22)
 /* _0 through _3 */
 #define R500_US_OUT_FMT_0				0x46a4
 #   define R500_OUT_FMT_C4_8				(0 << 0)
@@ -3068,12 +3068,12 @@ enum {
 #   define R500_C3_SEL_R				(1 << 14)
 #   define R500_C3_SEL_G				(2 << 14)
 #   define R500_C3_SEL_B				(3 << 14)
-#   define R500_OUT_SIGN(x)				(x << 16)
+#   define R500_OUT_SIGN(x)				((x) << 16)
 #   define R500_ROUND_ADJ				(1 << 20)
 #define R500_US_PIXSIZE					0x4604
 #   define R500_PIX_SIZE(x)				(x)
 #define R500_US_TEX_ADDR_0				0x9800
-#   define R500_TEX_SRC_ADDR(x)				(x << 0)
+#   define R500_TEX_SRC_ADDR(x)				((x) << 0)
 #   define R500_TEX_SRC_ADDR_REL			(1 << 7)
 #   define R500_TEX_SRC_S_SWIZ_R			(0 << 8)
 #   define R500_TEX_SRC_S_SWIZ_G			(1 << 8)
@@ -3091,7 +3091,7 @@ enum {
 #   define R500_TEX_SRC_Q_SWIZ_G			(1 << 14)
 #   define R500_TEX_SRC_Q_SWIZ_B			(2 << 14)
 #   define R500_TEX_SRC_Q_SWIZ_A			(3 << 14)
-#   define R500_TEX_DST_ADDR(x)				(x << 16)
+#   define R500_TEX_DST_ADDR(x)				((x) << 16)
 #   define R500_TEX_DST_ADDR_REL			(1 << 23)
 #   define R500_TEX_DST_R_SWIZ_R			(0 << 24)
 #   define R500_TEX_DST_R_SWIZ_G			(1 << 24)
@@ -3110,7 +3110,7 @@ enum {
 #   define R500_TEX_DST_A_SWIZ_B			(2 << 30)
 #   define R500_TEX_DST_A_SWIZ_A			(3 << 30)
 #define R500_US_TEX_ADDR_DXDY_0				0xa000
-#   define R500_DX_ADDR(x)				(x << 0)
+#   define R500_DX_ADDR(x)				((x) << 0)
 #   define R500_DX_ADDR_REL				(1 << 7)
 #   define R500_DX_S_SWIZ_R				(0 << 8)
 #   define R500_DX_S_SWIZ_G				(1 << 8)
@@ -3128,7 +3128,7 @@ enum {
 #   define R500_DX_Q_SWIZ_G				(1 << 14)
 #   define R500_DX_Q_SWIZ_B				(2 << 14)
 #   define R500_DX_Q_SWIZ_A				(3 << 14)
-#   define R500_DY_ADDR(x)				(x << 16)
+#   define R500_DY_ADDR(x)				((x) << 16)
 #   define R500_DY_ADDR_REL				(1 << 17)
 #   define R500_DY_S_SWIZ_R				(0 << 24)
 #   define R500_DY_S_SWIZ_G				(1 << 24)
@@ -3147,7 +3147,7 @@ enum {
 #   define R500_DY_Q_SWIZ_B				(2 << 30)
 #   define R500_DY_Q_SWIZ_A				(3 << 30)
 #define R500_US_TEX_INST_0				0x9000
-#   define R500_TEX_ID(x)				(x << 16)
+#   define R500_TEX_ID(x)				((x) << 16)
 #   define R500_TEX_INST_NOP				(0 << 22)
 #   define R500_TEX_INST_LD				(1 << 22)
 #   define R500_TEX_INST_TEXKILL			(2 << 22)
@@ -3208,9 +3208,9 @@ enum {
 #define R300_PACKET3_3D_LOAD_VBPNTR         0x00002F00
 
 #define R300_PACKET3_INDX_BUFFER            0x00003300
-#    define R300_EB_UNK1_SHIFT                      24
-#    define R300_EB_UNK1                    (0x80<<24)
-#    define R300_EB_UNK2                        0x0810
+#    define R300_INDX_BUFFER_DST_SHIFT          0
+#    define R300_INDX_BUFFER_SKIP_SHIFT         16
+#    define R300_INDX_BUFFER_ONE_REG_WR		(1<<31)
 
 /* Same as R300_PACKET3_3D_DRAW_VBUF but without VAP_VTX_FMT */
 #define R300_PACKET3_3D_DRAW_VBUF_2         0x00003400
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 8055952a768..f46477f0f67 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -197,7 +197,8 @@ static void r300FireEB(r300ContextPtr rmesa, int vertex_count, int type)
 		
 		if (!rmesa->radeon.radeonScreen->kernel_mm) {
 			OUT_BATCH_PACKET3(R300_PACKET3_INDX_BUFFER, 2);
-			OUT_BATCH(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
+			OUT_BATCH(R300_INDX_BUFFER_ONE_REG_WR | (0 << R300_INDX_BUFFER_SKIP_SHIFT) |
+	    			 (R300_VAP_PORT_IDX0 >> 2));
 			OUT_BATCH_RELOC(rmesa->state.elt_dma_offset,
 					rmesa->state.elt_dma_bo,
 					rmesa->state.elt_dma_offset,
@@ -205,7 +206,8 @@ static void r300FireEB(r300ContextPtr rmesa, int vertex_count, int type)
 			OUT_BATCH(vertex_count);
 		} else {
 			OUT_BATCH_PACKET3(R300_PACKET3_INDX_BUFFER, 2);
-			OUT_BATCH(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
+			OUT_BATCH(R300_INDX_BUFFER_ONE_REG_WR | (0 << R300_INDX_BUFFER_SKIP_SHIFT) |
+	    			 (R300_VAP_PORT_IDX0 >> 2));
 			OUT_BATCH(rmesa->state.elt_dma_offset);
 			OUT_BATCH(vertex_count);
 			radeon_cs_write_reloc(rmesa->radeon.cmdbuf.cs,
@@ -424,6 +426,8 @@ static GLboolean r300RunRender(GLcontext * ctx,
 static int r300Fallback(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	const unsigned back = ctx->Stencil._BackFace;
+
 	/* Do we need to use new-style shaders?
 	 * Also is there a better way to do this? */
 	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
@@ -448,12 +452,14 @@ static int r300Fallback(GLcontext * ctx)
 
 	FALLBACK_IF(ctx->RenderMode != GL_RENDER);
 
-	FALLBACK_IF(ctx->Stencil._TestTwoSide
-		    && (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[1]
-			|| ctx->Stencil.ValueMask[0] !=
-			ctx->Stencil.ValueMask[1]
-			|| ctx->Stencil.WriteMask[0] !=
-			ctx->Stencil.WriteMask[1]));
+	/* If GL_EXT_stencil_two_side is disabled, this fallback check can
+	 * be removed.
+	 */
+	FALLBACK_IF(ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back]
+		    || ctx->Stencil.ValueMask[0] !=
+		    ctx->Stencil.ValueMask[back]
+		    || ctx->Stencil.WriteMask[0] !=
+		    ctx->Stencil.WriteMask[back]);
 
 	if (ctx->Extensions.NV_point_sprite || ctx->Extensions.ARB_point_sprite)
 		FALLBACK_IF(ctx->Point.PointSprite);
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index fe58ac00a8a..93ef06fb268 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -781,6 +781,7 @@ static void r300Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 			    R300_FG_FOG_BLEND_FN_EXP2;
 			fogScale.f = 0.3 * ctx->Fog.Density;
 			fogStart.f = 0.0;
+                        break;
 		default:
 			return;
 		}
@@ -972,15 +973,9 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	GLuint refmask =
-	    (((ctx->Stencil.
-	       Ref[0] & 0xff) << R300_STENCILREF_SHIFT) | ((ctx->
-							    Stencil.
-							    ValueMask
-							    [0] &
-							    0xff)
-							   <<
-							   R300_STENCILMASK_SHIFT));
-
+	    ((ctx->Stencil.Ref[0] & 0xff) << R300_STENCILREF_SHIFT)
+	     | ((ctx->Stencil.ValueMask[0] & 0xff) << R300_STENCILMASK_SHIFT);
+	const unsigned back = ctx->Stencil._BackFace;
 	GLuint flag;
 
 	R300_STATECHANGE(rmesa, zs);
@@ -998,8 +993,7 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 	    (flag << R300_S_FRONT_FUNC_SHIFT);
 
-	if (ctx->Stencil._TestTwoSide)
-		flag = translate_func(ctx->Stencil.Function[1]);
+	flag = translate_func(ctx->Stencil.Function[back]);
 
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 	    (flag << R300_S_BACK_FUNC_SHIFT);
@@ -1024,6 +1018,7 @@ static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
 				  GLenum fail, GLenum zfail, GLenum zpass)
 {
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
+	const unsigned back = ctx->Stencil._BackFace;
 
 	R300_STATECHANGE(rmesa, zs);
 	/* It is easier to mask what's left.. */
@@ -1040,23 +1035,13 @@ static void r300StencilOpSeparate(GLcontext * ctx, GLenum face,
 	    | (translate_stencil_op(ctx->Stencil.ZPassFunc[0]) <<
 	       R300_S_FRONT_ZPASS_OP_SHIFT);
 
-	if (ctx->Stencil._TestTwoSide) {
-		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
-		    (translate_stencil_op(ctx->Stencil.FailFunc[1]) <<
-		     R300_S_BACK_SFAIL_OP_SHIFT)
-		    | (translate_stencil_op(ctx->Stencil.ZFailFunc[1]) <<
-		       R300_S_BACK_ZFAIL_OP_SHIFT)
-		    | (translate_stencil_op(ctx->Stencil.ZPassFunc[1]) <<
-		       R300_S_BACK_ZPASS_OP_SHIFT);
-	} else {
-		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
-		    (translate_stencil_op(ctx->Stencil.FailFunc[0]) <<
-		     R300_S_BACK_SFAIL_OP_SHIFT)
-		    | (translate_stencil_op(ctx->Stencil.ZFailFunc[0]) <<
-		       R300_S_BACK_ZFAIL_OP_SHIFT)
-		    | (translate_stencil_op(ctx->Stencil.ZPassFunc[0]) <<
-		       R300_S_BACK_ZPASS_OP_SHIFT);
-	}
+	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
+	    (translate_stencil_op(ctx->Stencil.FailFunc[back]) <<
+	     R300_S_BACK_SFAIL_OP_SHIFT)
+	    | (translate_stencil_op(ctx->Stencil.ZFailFunc[back]) <<
+	       R300_S_BACK_ZFAIL_OP_SHIFT)
+	    | (translate_stencil_op(ctx->Stencil.ZPassFunc[back]) <<
+	       R300_S_BACK_ZPASS_OP_SHIFT);
 }
 
 /* =============================================================
@@ -1647,6 +1632,13 @@ static void r300SetupRSUnit(GLcontext * ctx)
 		rs_col_count += count;
 	}
 
+	if (InputsRead & FRAG_BIT_FOGC) {
+		/* XXX FIX THIS
+		 * Just turn off the bit for now.
+		 * Need to do something similar to the color/texcoord inputs.
+		 */
+		InputsRead &= ~FRAG_BIT_FOGC;
+	}
 
 	for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
 		int swiz;
diff --git a/src/mesa/drivers/dri/r300/radeon_program_pair.c b/src/mesa/drivers/dri/r300/radeon_program_pair.c
index a6da9f56b5d..8a945d8537c 100644
--- a/src/mesa/drivers/dri/r300/radeon_program_pair.c
+++ b/src/mesa/drivers/dri/r300/radeon_program_pair.c
@@ -473,6 +473,11 @@ static void allocate_input_registers(struct pair_state *s)
 		alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_COL1, hwindex++);
 	InputsRead &= ~FRAG_BIT_COL1;
 
+	/* Fog coordinate */
+	if (InputsRead & FRAG_BIT_FOGC)
+		alloc_hw_reg(s, PROGRAM_INPUT, FRAG_ATTRIB_FOGC, hwindex++);
+	InputsRead &= ~FRAG_BIT_FOGC;
+
 	/* Anything else */
 	if (InputsRead)
 		error("Don't know how to handle inputs 0x%x\n", InputsRead);
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index 0296cffc20a..7db62be7afd 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -63,9 +63,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_tcl.h"
 #include "radeon_maos.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
@@ -81,16 +78,13 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_border_clamp",       NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_env_add",            NULL },
     { "GL_ARB_texture_env_combine",        NULL },
     { "GL_ARB_texture_env_crossbar",       NULL },
     { "GL_ARB_texture_env_dot3",           NULL },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_blend_logic_op",             NULL },
     { "GL_EXT_blend_subtract",             GL_EXT_blend_minmax_functions },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
@@ -344,6 +338,8 @@ radeonCreateContext( const __GLcontextModes *glVisual,
 
    rmesa->boxes = 0;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 8b06fb4d3e9..21b94d2e527 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -212,6 +212,8 @@ DRI_CONF_BEGIN
 DRI_CONF_END;
 static const GLuint __driNConfigOptions = 17;
 
+extern const struct dri_extension gl_20_extension[];
+
 #ifndef RADEON_DEBUG
 
 static const struct dri_debug_control debug_control[] = {
@@ -279,7 +281,7 @@ radeonFillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[2];
     uint8_t stencil_bits_array[2];
-
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
@@ -291,6 +293,8 @@ radeonFillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[0] = 0;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
@@ -306,7 +310,8 @@ radeonFillInModes( __DRIscreenPrivate *psp,
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor,
-			       back_buffer_modes, back_buffer_factor);
+			       back_buffer_modes, back_buffer_factor,
+			       msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
@@ -1368,6 +1373,7 @@ static void radeonDestroyContext(__DRIcontextPrivate * driContextPriv)
 
 #endif
 
+
 /**
  * This is the driver specific part of the createNewScreen entry point.
  *
@@ -1420,6 +1426,8 @@ radeonInitScreen(__DRIscreenPrivate *psp)
    driInitSingleExtension( NULL, NV_vp_extension );
    driInitSingleExtension( NULL, ATI_fs_extension );
    driInitExtensions( NULL, point_extensions, GL_FALSE );
+#elif defined(RADEON_COMMON_FOR_R300)
+   driInitSingleExtension( NULL, gl_20_extension );
 #endif
 
    if (!radeonInitDriver(psp))
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.c b/src/mesa/drivers/dri/s3v/s3v_context.c
index 14502f95aec..0a3bf7258d1 100644
--- a/src/mesa/drivers/dri/s3v/s3v_context.c
+++ b/src/mesa/drivers/dri/s3v/s3v_context.c
@@ -108,6 +108,8 @@ GLboolean s3vCreateContext(const __GLcontextModes *glVisual,
 	ctx->Const.MaxLineWidthAA = 1.0;
 	ctx->Const.LineWidthGranularity = 1.0;
 
+	ctx->Const.MaxDrawBuffers = 1;
+
 	vmesa->texHeap = mmInit( 0, vmesa->s3vScreen->textureSize );
 	DEBUG(("vmesa->s3vScreen->textureSize = 0x%x\n",
 		vmesa->s3vScreen->textureSize));
diff --git a/src/mesa/drivers/dri/savage/savage_xmesa.c b/src/mesa/drivers/dri/savage/savage_xmesa.c
index a344aab71bb..326d595352f 100644
--- a/src/mesa/drivers/dri/savage/savage_xmesa.c
+++ b/src/mesa/drivers/dri/savage/savage_xmesa.c
@@ -59,9 +59,6 @@
 #include "drirenderbuffer.h"
 #include "texmem.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_EXT_secondary_color
 #include "extension_helper.h"
 
@@ -133,10 +130,7 @@ struct timeval tv_s1,tv_f1;
 
 static const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_stencil_wrap",               NULL },
     { "GL_EXT_texture_lod_bias",           NULL },
     { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
@@ -350,7 +344,9 @@ savageCreateContext( const __GLcontextModes *mesaVis,
    ctx->Const.MaxLineWidthAA = 3.0;
    ctx->Const.LineWidthGranularity = 1.0;
 #endif
-   
+
+   ctx->Const.MaxDrawBuffers = 1;
+
    /* Dri stuff
     */
    imesa->hHWContext = driContextPriv->hHWContext;
@@ -914,7 +910,7 @@ savageFillInModes( __DRIscreenPrivate *psp,
 
     uint8_t depth_bits_array[2];
     uint8_t stencil_bits_array[2];
-
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = depth_bits;
     depth_bits_array[1] = depth_bits;
@@ -926,6 +922,8 @@ savageFillInModes( __DRIscreenPrivate *psp,
     stencil_bits_array[0] = 0;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
@@ -941,7 +939,8 @@ savageFillInModes( __DRIscreenPrivate *psp,
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor,
-			       back_buffer_modes, back_buffer_factor);
+			       back_buffer_modes, back_buffer_factor,
+                               msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
 		 __func__, __LINE__ );
diff --git a/src/mesa/drivers/dri/sis/sis_alloc.c b/src/mesa/drivers/dri/sis/sis_alloc.c
index 4ca40528034..ce34e44da22 100644
--- a/src/mesa/drivers/dri/sis/sis_alloc.c
+++ b/src/mesa/drivers/dri/sis/sis_alloc.c
@@ -137,7 +137,7 @@ sisAllocZStencilBuffer( sisContextPtr smesa )
 {
    int cpp = ( smesa->glCtx->Visual.depthBits +
                smesa->glCtx->Visual.stencilBits ) / 8;
-   unsigned char *addr;
+   char *addr;
 
    smesa->depth.bpp = cpp * 8;
    smesa->depth.pitch = ALIGNMENT(smesa->driDrawable->w * cpp, 4);
@@ -150,7 +150,7 @@ sisAllocZStencilBuffer( sisContextPtr smesa )
    addr = (char *)ALIGNMENT((unsigned long)addr, Z_BUFFER_HW_ALIGNMENT);
 
    smesa->depth.map = addr;
-   smesa->depth.offset = addr - smesa->FbBase;
+   smesa->depth.offset = addr - (char *)smesa->FbBase;
 
    /* stencil buffer is same as depth buffer */
    smesa->stencil.size = smesa->depth.size;
@@ -173,7 +173,7 @@ void
 sisAllocBackbuffer( sisContextPtr smesa )
 {
    int cpp = smesa->bytesPerPixel;
-   unsigned char *addr;
+   char *addr;
 
    smesa->back.bpp = smesa->bytesPerPixel * 8;
    smesa->back.pitch = ALIGNMENT(smesa->driDrawable->w * cpp, 4);
@@ -186,7 +186,7 @@ sisAllocBackbuffer( sisContextPtr smesa )
    addr = (char *)ALIGNMENT((unsigned long)addr, DRAW_BUFFER_HW_ALIGNMENT);
 
    smesa->back.map = addr;
-   smesa->back.offset = addr - smesa->FbBase;
+   smesa->back.offset = addr - (char *)smesa->FbBase;
 }
 
 void
diff --git a/src/mesa/drivers/dri/sis/sis_context.c b/src/mesa/drivers/dri/sis/sis_context.c
index 00d17da3bad..a070fe3d79f 100644
--- a/src/mesa/drivers/dri/sis/sis_context.c
+++ b/src/mesa/drivers/dri/sis/sis_context.c
@@ -57,9 +57,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 
-#define need_GL_ARB_multisample
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
 #include "extension_helper.h"
@@ -74,12 +71,9 @@ int GlobalCmdQueueLen = 0;
 
 struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_texture_border_clamp",       NULL },
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     /*{ "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },*/
     { "GL_EXT_texture_lod_bias",           NULL },
     { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
@@ -91,9 +85,7 @@ struct dri_extension card_extensions[] =
 
 struct dri_extension card_extensions_6326[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     /*{ "GL_ARB_texture_border_clamp",       NULL },*/
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     /*{ "GL_ARB_texture_mirrored_repeat",    NULL },*/
     /*{ "GL_MESA_ycbcr_texture",             NULL },*/
     { NULL,                                NULL }
diff --git a/src/mesa/drivers/dri/sis/sis_screen.c b/src/mesa/drivers/dri/sis/sis_screen.c
index b1a5d15236f..9eb27fef087 100644
--- a/src/mesa/drivers/dri/sis/sis_screen.c
+++ b/src/mesa/drivers/dri/sis/sis_screen.c
@@ -77,6 +77,7 @@ sisFillInModes(__DRIscreenPrivate *psp, int bpp)
    };
    uint8_t depth_bits_array[4];
    uint8_t stencil_bits_array[4];
+   uint8_t msaa_samples_array[1];
 
    depth_bits_array[0] = 0;
    stencil_bits_array[0] = 0;
@@ -87,6 +88,8 @@ sisFillInModes(__DRIscreenPrivate *psp, int bpp)
    depth_bits_array[3] = 32;
    stencil_bits_array[3] = 0;
 
+   msaa_samples_array[0] = 0;
+
    depth_buffer_factor = 4;
    back_buffer_factor = 2;
 
@@ -100,7 +103,8 @@ sisFillInModes(__DRIscreenPrivate *psp, int bpp)
 
    configs = driCreateConfigs(fb_format, fb_type, depth_bits_array,
 			      stencil_bits_array, depth_buffer_factor,
-			      back_buffer_modes, back_buffer_factor);
+			      back_buffer_modes, back_buffer_factor,
+                              msaa_samples_array, 1);
    if (configs == NULL) {
       fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__, __LINE__);
       return NULL;
diff --git a/src/mesa/drivers/dri/sis/sis_tris.c b/src/mesa/drivers/dri/sis/sis_tris.c
index 095941aea2e..76d12d07b3c 100644
--- a/src/mesa/drivers/dri/sis/sis_tris.c
+++ b/src/mesa/drivers/dri/sis/sis_tris.c
@@ -994,7 +994,7 @@ sisFlushPrimsLocked(sisContextPtr smesa)
 	 MMIO(REG_3D_PrimitiveSet, smesa->dwPrimitiveSet);
       }
       while (smesa->vb_last < smesa->vb_cur) {
-	 sis_emit_func(smesa, smesa->vb_last);
+	 sis_emit_func(smesa, (char *)smesa->vb_last);
 	 smesa->vb_last += incr;
       }
       mWait3DCmdQueue(1);
diff --git a/src/mesa/drivers/dri/swrast/swrast.c b/src/mesa/drivers/dri/swrast/swrast.c
index 49f1b8b9ee7..2e7f11327e6 100644
--- a/src/mesa/drivers/dri/swrast/swrast.c
+++ b/src/mesa/drivers/dri/swrast/swrast.c
@@ -66,12 +66,13 @@
 #define need_GL_ARB_vertex_program
 #define need_GL_APPLE_vertex_array_object
 #define need_GL_ATI_fragment_shader
+#define need_GL_ATI_separate_stencil
 #define need_GL_EXT_depth_bounds_test
 #define need_GL_EXT_framebuffer_object
 #define need_GL_EXT_framebuffer_blit
 #define need_GL_EXT_gpu_program_parameters
 #define need_GL_EXT_paletted_texture
-#define need_GL_IBM_multimode_draw_arrays
+#define need_GL_EXT_stencil_two_side
 #define need_GL_MESA_resize_buffers
 #define need_GL_NV_vertex_program
 #define need_GL_NV_fragment_program
@@ -96,12 +97,13 @@ const struct dri_extension card_extensions[] =
     { "GL_ARB_vertex_program",		GL_ARB_vertex_program_functions },
     { "GL_APPLE_vertex_array_object",	GL_APPLE_vertex_array_object_functions },
     { "GL_ATI_fragment_shader",		GL_ATI_fragment_shader_functions },
+    { "GL_ATI_separate_stencil",	GL_ATI_separate_stencil_functions },
     { "GL_EXT_depth_bounds_test",	GL_EXT_depth_bounds_test_functions },
     { "GL_EXT_framebuffer_object",	GL_EXT_framebuffer_object_functions },
     { "GL_EXT_framebuffer_blit",	GL_EXT_framebuffer_blit_functions },
     { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
     { "GL_EXT_paletted_texture",	GL_EXT_paletted_texture_functions },
-    { "GL_IBM_multimode_draw_arrays",	GL_IBM_multimode_draw_arrays_functions },
+    { "GL_EXT_stencil_two_side",	GL_EXT_stencil_two_side_functions },
     { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
     { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
     { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
@@ -145,6 +147,7 @@ swrastFillInModes(__DRIscreen *psp,
 
     uint8_t depth_bits_array[4];
     uint8_t stencil_bits_array[4];
+    uint8_t msaa_samples_array[1];
 
     depth_bits_array[0] = 0;
     depth_bits_array[1] = 0;
@@ -159,26 +162,38 @@ swrastFillInModes(__DRIscreen *psp,
     stencil_bits_array[2] = 0;
     stencil_bits_array[3] = (stencil_bits == 0) ? 8 : stencil_bits;
 
+    msaa_samples_array[0] = 0;
+
     depth_buffer_factor = 4;
     back_buffer_factor = 2;
 
-    if (pixel_bits == 8) {
+    switch (pixel_bits) {
+    case 8:
 	fb_format = GL_RGB;
 	fb_type = GL_UNSIGNED_BYTE_2_3_3_REV;
-    }
-    else if (pixel_bits == 16) {
+	break;
+    case 16:
 	fb_format = GL_RGB;
 	fb_type = GL_UNSIGNED_SHORT_5_6_5;
-    }
-    else {
+	break;
+    case 24:
+	fb_format = GL_BGR;
+	fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+	break;
+    case 32:
 	fb_format = GL_BGRA;
 	fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+	break;
+    default:
+	fprintf(stderr, "[%s:%u] bad depth %d\n", __func__, __LINE__,
+		pixel_bits);
+	return NULL;
     }
 
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor, back_buffer_modes,
-			       back_buffer_factor);
+			       back_buffer_factor, msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
 		__LINE__);
@@ -194,7 +209,7 @@ driCreateNewScreen(int scrn, const __DRIextension **extensions,
 {
     static const __DRIextension *emptyExtensionList[] = { NULL };
     __DRIscreen *psp;
-    __DRIconfig **configs8, **configs16, **configs32;
+    __DRIconfig **configs8, **configs16, **configs24, **configs32;
 
     (void) data;
 
@@ -211,11 +226,13 @@ driCreateNewScreen(int scrn, const __DRIextension **extensions,
 
     configs8  = swrastFillInModes(psp,  8,  8, 0, 1);
     configs16 = swrastFillInModes(psp, 16, 16, 0, 1);
+    configs24 = swrastFillInModes(psp, 24, 24, 8, 1);
     configs32 = swrastFillInModes(psp, 32, 24, 8, 1);
 
-    configs16 = (__DRIconfig **)driConcatConfigs(configs8, configs16);
-
-    *driver_configs = driConcatConfigs(configs16, configs32);
+    configs16 = driConcatConfigs(configs8, configs16);
+    configs24 = driConcatConfigs(configs16, configs24);
+    *driver_configs = (const __DRIconfig **)
+       driConcatConfigs(configs24, configs32);
 
     driInitExtensions( NULL, card_extensions, GL_FALSE );
 
@@ -247,19 +264,24 @@ static GLuint
 choose_pixel_format(const GLvisual *v)
 {
     if (v->rgbMode) {
-	int bpp = v->rgbBits;
+	int depth = v->rgbBits;
 
-	if (bpp == 32
+	if (depth == 32
 	    && v->redMask   == 0xff0000
 	    && v->greenMask == 0x00ff00
 	    && v->blueMask  == 0x0000ff)
 	    return PF_A8R8G8B8;
-	else if (bpp == 16
+	else if (depth == 24
+	    && v->redMask   == 0xff0000
+	    && v->greenMask == 0x00ff00
+	    && v->blueMask  == 0x0000ff)
+	    return PF_X8R8G8B8;
+	else if (depth == 16
 	    && v->redMask   == 0xf800
 	    && v->greenMask == 0x07e0
 	    && v->blueMask  == 0x001f)
 	    return PF_R5G6B5;
-	else if (bpp == 8
+	else if (depth == 8
 	    && v->redMask   == 0x07
 	    && v->greenMask == 0x38
 	    && v->blueMask  == 0xc0)
@@ -288,7 +310,6 @@ swrast_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
 			   GLenum internalFormat, GLuint width, GLuint height)
 {
     struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
-    int bpp;
     unsigned mask = PITCH_ALIGN_BITS - 1;
 
     TRACE;
@@ -297,23 +318,8 @@ swrast_alloc_front_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
     rb->Width = width;
     rb->Height = height;
 
-    switch (internalFormat) {
-    case GL_RGB:
-	bpp = rb->RedBits + rb->GreenBits + rb->BlueBits;
-	break;
-    case GL_RGBA:
-	bpp = rb->RedBits + rb->GreenBits + rb->BlueBits + rb->AlphaBits;
-	break;
-    case GL_COLOR_INDEX8_EXT:
-	bpp = rb->IndexBits;
-	break;
-    default:
-	_mesa_problem( NULL, "unexpected format in %s", __FUNCTION__ );
-	return GL_FALSE;
-    }
-
     /* always pad to PITCH_ALIGN_BITS */
-    xrb->pitch = ((width * bpp + mask) & ~mask) / 8;
+    xrb->pitch = ((width * xrb->bpp + mask) & ~mask) / 8;
 
     return GL_TRUE;
 }
@@ -369,6 +375,17 @@ swrast_new_renderbuffer(const GLvisual *visual, GLboolean front)
 	xrb->Base.GreenBits = 8 * sizeof(GLubyte);
 	xrb->Base.BlueBits  = 8 * sizeof(GLubyte);
 	xrb->Base.AlphaBits = 8 * sizeof(GLubyte);
+	xrb->bpp = 32;
+	break;
+    case PF_X8R8G8B8:
+	xrb->Base.InternalFormat = GL_RGB;
+	xrb->Base._BaseFormat = GL_RGB;
+	xrb->Base.DataType = GL_UNSIGNED_BYTE;
+	xrb->Base.RedBits   = 8 * sizeof(GLubyte);
+	xrb->Base.GreenBits = 8 * sizeof(GLubyte);
+	xrb->Base.BlueBits  = 8 * sizeof(GLubyte);
+	xrb->Base.AlphaBits = 0;
+	xrb->bpp = 32;
 	break;
     case PF_R5G6B5:
 	xrb->Base.InternalFormat = GL_RGB;
@@ -378,6 +395,7 @@ swrast_new_renderbuffer(const GLvisual *visual, GLboolean front)
 	xrb->Base.GreenBits = 6 * sizeof(GLubyte);
 	xrb->Base.BlueBits  = 5 * sizeof(GLubyte);
 	xrb->Base.AlphaBits = 0;
+	xrb->bpp = 16;
 	break;
     case PF_R3G3B2:
 	xrb->Base.InternalFormat = GL_RGB;
@@ -387,12 +405,14 @@ swrast_new_renderbuffer(const GLvisual *visual, GLboolean front)
 	xrb->Base.GreenBits = 3 * sizeof(GLubyte);
 	xrb->Base.BlueBits  = 2 * sizeof(GLubyte);
 	xrb->Base.AlphaBits = 0;
+	xrb->bpp = 8;
 	break;
     case PF_CI8:
 	xrb->Base.InternalFormat = GL_COLOR_INDEX8_EXT;
 	xrb->Base._BaseFormat = GL_COLOR_INDEX;
 	xrb->Base.DataType = GL_UNSIGNED_BYTE;
 	xrb->Base.IndexBits = 8 * sizeof(GLubyte);
+	xrb->bpp = 8;
 	break;
     default:
 	return NULL;
diff --git a/src/mesa/drivers/dri/swrast/swrast_priv.h b/src/mesa/drivers/dri/swrast/swrast_priv.h
index a707ffc40a8..1a5fb31d5a8 100644
--- a/src/mesa/drivers/dri/swrast/swrast_priv.h
+++ b/src/mesa/drivers/dri/swrast/swrast_priv.h
@@ -90,6 +90,8 @@ struct swrast_renderbuffer {
 
     /* renderbuffer pitch (in bytes) */
     GLuint pitch;
+   /* bits per pixel of storage */
+    GLuint bpp;
 };
 
 static INLINE __DRIcontext *
@@ -115,10 +117,10 @@ swrast_renderbuffer(struct gl_renderbuffer *rb)
  * Pixel formats we support
  */
 #define PF_CI8        1		/**< Color Index mode */
-#define PF_A8R8G8B8   2		/**< 32-bit TrueColor:  8-A, 8-R, 8-G, 8-B bits */
-#define PF_R5G6B5     3		/**< 16-bit TrueColor:  5-R, 6-G, 5-B bits */
-#define PF_R3G3B2     4		/**<  8-bit TrueColor:  3-R, 3-G, 2-B bits */
-
+#define PF_A8R8G8B8   2		/**< 32bpp TrueColor:  8-A, 8-R, 8-G, 8-B bits */
+#define PF_R5G6B5     3		/**< 16bpp TrueColor:  5-R, 6-G, 5-B bits */
+#define PF_R3G3B2     4		/**<  8bpp TrueColor:  3-R, 3-G, 2-B bits */
+#define PF_X8R8G8B8   5		/**< 32bpp TrueColor:  8-R, 8-G, 8-B bits */
 
 /**
  * Renderbuffer pitch alignment (in bits).
diff --git a/src/mesa/drivers/dri/swrast/swrast_span.c b/src/mesa/drivers/dri/swrast/swrast_span.c
index 5e990368b2e..2d3c25dcbef 100644
--- a/src/mesa/drivers/dri/swrast/swrast_span.c
+++ b/src/mesa/drivers/dri/swrast/swrast_span.c
@@ -79,6 +79,24 @@ static const GLubyte kernel[16] = {
    DST[BCOMP] = SRC[0]
 
 
+/* 32-bit BGRX */
+#define STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE) \
+   DST[3] = 0xff; \
+   DST[2] = VALUE[RCOMP]; \
+   DST[1] = VALUE[GCOMP]; \
+   DST[0] = VALUE[BCOMP]
+#define STORE_PIXEL_RGB_X8R8G8B8(DST, X, Y, VALUE) \
+   DST[3] = 0xff; \
+   DST[2] = VALUE[RCOMP]; \
+   DST[1] = VALUE[GCOMP]; \
+   DST[0] = VALUE[BCOMP]
+#define FETCH_PIXEL_X8R8G8B8(DST, SRC) \
+   DST[ACOMP] = 0xff; \
+   DST[RCOMP] = SRC[2]; \
+   DST[GCOMP] = SRC[1]; \
+   DST[BCOMP] = SRC[0]
+
+
 /* 16-bit BGR */
 #define STORE_PIXEL_R5G6B5(DST, X, Y, VALUE) \
    do { \
@@ -139,6 +157,24 @@ static const GLubyte kernel[16] = {
 #include "swrast/s_spantemp.h"
 
 
+/* 32-bit BGRX */
+#define NAME(FUNC) FUNC##_X8R8G8B8
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)xrb->Base.Data + YFLIP(xrb, Y) * xrb->pitch + (X) * 4;
+#define INC_PIXEL_PTR(P) P += 4
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE)
+#define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
+   STORE_PIXEL_RGB_X8R8G8B8(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_X8R8G8B8(DST, SRC)
+
+#include "swrast/s_spantemp.h"
+
+
 /* 16-bit BGR */
 #define NAME(FUNC) FUNC##_R5G6B5
 #define RB_TYPE GLubyte
@@ -210,6 +246,24 @@ static const GLubyte kernel[16] = {
 #include "swrast_spantemp.h"
 
 
+/* 32-bit BGRX */
+#define NAME(FUNC) FUNC##_X8R8G8B8_front
+#define RB_TYPE GLubyte
+#define SPAN_VARS \
+   struct swrast_renderbuffer *xrb = swrast_renderbuffer(rb);
+#define INIT_PIXEL_PTR(P, X, Y) \
+   GLubyte *P = (GLubyte *)row;
+#define INC_PIXEL_PTR(P) P += 4
+#define STORE_PIXEL(DST, X, Y, VALUE) \
+   STORE_PIXEL_X8R8G8B8(DST, X, Y, VALUE)
+#define STORE_PIXEL_RGB(DST, X, Y, VALUE) \
+   STORE_PIXEL_RGB_X8R8G8B8(DST, X, Y, VALUE)
+#define FETCH_PIXEL(DST, SRC) \
+   FETCH_PIXEL_X8R8G8B8(DST, SRC)
+
+#include "swrast_spantemp.h"
+
+
 /* 16-bit BGR */
 #define NAME(FUNC) FUNC##_R5G6B5_front
 #define RB_TYPE GLubyte
@@ -279,6 +333,15 @@ swrast_set_span_funcs_back(struct swrast_renderbuffer *xrb,
 	xrb->Base.PutValues = put_values_A8R8G8B8;
 	xrb->Base.PutMonoValues = put_mono_values_A8R8G8B8;
 	break;
+    case PF_X8R8G8B8:
+	xrb->Base.GetRow = get_row_X8R8G8B8;
+	xrb->Base.GetValues = get_values_X8R8G8B8;
+	xrb->Base.PutRow = put_row_X8R8G8B8;
+	xrb->Base.PutRowRGB = put_row_rgb_X8R8G8B8;
+	xrb->Base.PutMonoRow = put_mono_row_X8R8G8B8;
+	xrb->Base.PutValues = put_values_X8R8G8B8;
+	xrb->Base.PutMonoValues = put_mono_values_X8R8G8B8;
+	break;
     case PF_R5G6B5:
 	xrb->Base.GetRow = get_row_R5G6B5;
 	xrb->Base.GetValues = get_values_R5G6B5;
@@ -334,6 +397,15 @@ swrast_set_span_funcs_front(struct swrast_renderbuffer *xrb,
 	xrb->Base.PutValues = put_values_A8R8G8B8_front;
 	xrb->Base.PutMonoValues = put_mono_values_A8R8G8B8_front;
 	break;
+    case PF_X8R8G8B8:
+	xrb->Base.GetRow = get_row_X8R8G8B8_front;
+	xrb->Base.GetValues = get_values_X8R8G8B8_front;
+	xrb->Base.PutRow = put_row_X8R8G8B8_front;
+	xrb->Base.PutRowRGB = put_row_rgb_X8R8G8B8_front;
+	xrb->Base.PutMonoRow = put_mono_row_X8R8G8B8_front;
+	xrb->Base.PutValues = put_values_X8R8G8B8_front;
+	xrb->Base.PutMonoValues = put_mono_values_X8R8G8B8_front;
+	break;
     case PF_R5G6B5:
 	xrb->Base.GetRow = get_row_R5G6B5_front;
 	xrb->Base.GetValues = get_values_R5G6B5_front;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.c b/src/mesa/drivers/dri/tdfx/tdfx_context.c
index ef688d103d3..20046fcb3af 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.c
@@ -58,20 +58,15 @@
 
 #include "utils.h"
 
-#define need_GL_ARB_multisample
 /* #define need_GL_ARB_point_parameters */
 #define need_GL_ARB_occlusion_query
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
 /* #define need_GL_ARB_vertex_program */
 #define need_GL_EXT_blend_equation_separate
 #define need_GL_EXT_blend_func_separate
 #define need_GL_EXT_blend_minmax
 #define need_GL_EXT_fog_coord
-#define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_paletted_texture
 /* #define need_GL_EXT_secondary_color */
-#define need_GL_IBM_multimode_draw_arrays
 /* #define need_GL_MESA_program_debug */
 /* #define need_GL_NV_vertex_program */
 #include "extension_helper.h"
@@ -82,20 +77,16 @@
  */
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_occlusion_query",            GL_ARB_occlusion_query_functions },
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
 
     { "GL_EXT_blend_func_separate",        GL_EXT_blend_func_separate_functions },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
-    { "GL_EXT_multi_draw_arrays",          GL_EXT_multi_draw_arrays_functions },
     { "GL_EXT_paletted_texture",           GL_EXT_paletted_texture_functions },
     { "GL_EXT_shared_texture_palette",     NULL },
     { "GL_EXT_stencil_wrap",               NULL },
     { "GL_EXT_texture_env_add",            NULL },
     { "GL_EXT_texture_lod_bias",           NULL },
-    { "GL_IBM_multimode_draw_arrays",      GL_IBM_multimode_draw_arrays_functions },
 
 #ifdef need_GL_ARB_point_parameters
     { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
@@ -122,7 +113,6 @@ const struct dri_extension card_extensions[] =
  */
 const struct dri_extension napalm_extensions[] =
 {
-    { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_env_combine",        NULL },
     { "GL_EXT_blend_equation_separate",    GL_EXT_blend_equation_separate_functions },
     { "GL_EXT_blend_subtract",             GL_EXT_blend_minmax_functions },
@@ -319,6 +309,8 @@ GLboolean tdfxCreateContext( const __GLcontextModes *mesaVis,
    ctx->Const.MaxLineWidthAA = 1.0;
    ctx->Const.LineWidthGranularity = 1.0;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_dd.c b/src/mesa/drivers/dri/tdfx/tdfx_dd.c
index 2cef0795153..8472df607ae 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_dd.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_dd.c
@@ -110,13 +110,13 @@ static const GLubyte *tdfxDDGetString( GLcontext *ctx, GLenum name )
 
 
 static void
-tdfxBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+tdfxBeginQuery(GLcontext *ctx, struct gl_query_object *q)
 {
    tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
 
    (void) q;
 
-   if (target == GL_SAMPLES_PASSED_ARB) {
+   if (q->Target == GL_SAMPLES_PASSED_ARB) {
       LOCK_HARDWARE(fxMesa);
       fxMesa->Glide.grFinish();
       fxMesa->Glide.grReset(GR_STATS_PIXELS);
@@ -126,14 +126,14 @@ tdfxBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
 
 
 static void
-tdfxEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+tdfxEndQuery(GLcontext *ctx, struct gl_query_object *q)
 {
    tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx);
    FxI32 total_pixels;
    FxI32 z_fail_pixels;
 
 
-   if (target == GL_SAMPLES_PASSED_ARB) {
+   if (q->Target == GL_SAMPLES_PASSED_ARB) {
       LOCK_HARDWARE(fxMesa);
       fxMesa->Glide.grFinish();
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_screen.c b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
index cd22b849511..5f2f5cfff51 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_screen.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_screen.c
@@ -361,6 +361,7 @@ tdfxFillInModes(__DRIscreenPrivate *psp,
 	static const GLenum db_modes[2] = { GLX_NONE, GLX_SWAP_UNDEFINED_OML };
 	uint8_t depth_bits_array[4];
 	uint8_t stencil_bits_array[4];
+        uint8_t msaa_samples_array[1];
 	if(deep) {
 		depth_bits_array[0] = 0;
 		depth_bits_array[1] = 24;
@@ -377,13 +378,17 @@ tdfxFillInModes(__DRIscreenPrivate *psp,
 		stencil_bits_array[3] = 8;
 	}
 
-	return driCreateConfigs(
-		deep ? GL_RGBA : GL_RGB,
-		deep ? GL_UNSIGNED_INT_8_8_8_8 : GL_UNSIGNED_SHORT_5_6_5,
-		depth_bits_array,
-		stencil_bits_array,
-		deep ? 2 : 4,
-		db_modes, 2);
+	msaa_samples_array[0] = 0;
+
+	return (const __DRIconfig **)
+	   driCreateConfigs(deep ? GL_RGBA : GL_RGB,
+			    deep ? GL_UNSIGNED_INT_8_8_8_8 :
+				   GL_UNSIGNED_SHORT_5_6_5,
+			    depth_bits_array,
+			    stencil_bits_array,
+			    deep ? 2 : 4,
+			    db_modes, 2,
+			    msaa_samples_array, 1);
 }
 
 /**
diff --git a/src/mesa/drivers/dri/trident/trident_context.c b/src/mesa/drivers/dri/trident/trident_context.c
index e134cfcf8e9..aefd8a243f6 100644
--- a/src/mesa/drivers/dri/trident/trident_context.c
+++ b/src/mesa/drivers/dri/trident/trident_context.c
@@ -128,6 +128,8 @@ tridentCreateContext( const __GLcontextModes *glVisual,
    ctx->Const.MaxPointSizeAA = 16.0; 
    ctx->Const.PointSizeGranularity = 0.25;
 
+   ctx->Const.MaxDrawBuffers = 1;
+
 #if 0
    tmesa->texHeap = mmInit( 0, tmesa->tridentScreen->textureSize );
 
diff --git a/src/mesa/drivers/dri/trident/trident_vb.c b/src/mesa/drivers/dri/trident/trident_vb.c
index b231f5ef15e..055a9145953 100644
--- a/src/mesa/drivers/dri/trident/trident_vb.c
+++ b/src/mesa/drivers/dri/trident/trident_vb.c
@@ -402,7 +402,7 @@ void tridentInitVB( GLcontext *ctx )
    tridentContextPtr tmesa = TRIDENT_CONTEXT(ctx);
    GLuint size = TNL_CONTEXT(ctx)->vb.Size;
 
-   tmesa->verts = (char *)ALIGN_MALLOC( size * 16 * 4, 32 );
+   tmesa->verts = (GLubyte *)ALIGN_MALLOC( size * 16 * 4, 32 );
 
    {
       static int firsttime = 1;
diff --git a/src/mesa/drivers/dri/unichrome/via_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index f5bdb65eb08..6eb19ac0790 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -62,9 +62,7 @@
 #include "main/macros.h"
 #include "drirenderbuffer.h"
 
-#define need_GL_ARB_multisample
 #define need_GL_ARB_point_parameters
-#define need_GL_ARB_vertex_buffer_object
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_secondary_color
 #include "extension_helper.h"
@@ -366,14 +364,12 @@ void viaReAllocateBuffers(GLcontext *ctx, GLframebuffer *drawbuffer,
  */
 const struct dri_extension card_extensions[] =
 {
-    { "GL_ARB_multisample",                GL_ARB_multisample_functions },
     { "GL_ARB_multitexture",               NULL },
     { "GL_ARB_point_parameters",           GL_ARB_point_parameters_functions },
     { "GL_ARB_texture_env_add",            NULL },
     { "GL_ARB_texture_env_combine",        NULL },
 /*    { "GL_ARB_texture_env_dot3",           NULL }, */
     { "GL_ARB_texture_mirrored_repeat",    NULL },
-    { "GL_ARB_vertex_buffer_object",       GL_ARB_vertex_buffer_object_functions },
     { "GL_EXT_fog_coord",                  GL_EXT_fog_coord_functions },
     { "GL_EXT_secondary_color",            GL_EXT_secondary_color_functions },
     { "GL_EXT_stencil_wrap",               NULL },
@@ -577,6 +573,8 @@ viaCreateContext(const __GLcontextModes *visual,
     ctx->Const.MaxPointSizeAA = 1.0;
     ctx->Const.PointSizeGranularity = 1.0;
 
+    ctx->Const.MaxDrawBuffers = 1;
+
     ctx->Driver.GetString = viaGetString;
 
     ctx->DriverCtx = (void *)vmesa;
diff --git a/src/mesa/drivers/dri/unichrome/via_screen.c b/src/mesa/drivers/dri/unichrome/via_screen.c
index 988f9935ac1..e9f566161c2 100644
--- a/src/mesa/drivers/dri/unichrome/via_screen.c
+++ b/src/mesa/drivers/dri/unichrome/via_screen.c
@@ -342,6 +342,7 @@ viaFillInModes( __DRIscreenPrivate *psp,
      */
     static const uint8_t depth_bits_array[4]   = { 0, 16, 24, 32 };
     static const uint8_t stencil_bits_array[4] = { 0,  0,  8,  0 };
+    uint8_t msaa_samples_array[1] = { 0 };
     const unsigned depth_buffer_factor = 3;
 
     if ( pixel_bits == 16 ) {
@@ -356,7 +357,8 @@ viaFillInModes( __DRIscreenPrivate *psp,
     configs = driCreateConfigs(fb_format, fb_type,
 			       depth_bits_array, stencil_bits_array,
 			       depth_buffer_factor, back_buffer_modes,
-			       back_buffer_factor);
+			       back_buffer_factor,
+                               msaa_samples_array, 1);
     if (configs == NULL) {
 	fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
 		__LINE__);
diff --git a/src/mesa/drivers/fbdev/glfbdev.c b/src/mesa/drivers/fbdev/glfbdev.c
index 3c874ba57ab..531558dc4d6 100644
--- a/src/mesa/drivers/fbdev/glfbdev.c
+++ b/src/mesa/drivers/fbdev/glfbdev.c
@@ -685,7 +685,7 @@ glFBDevDestroyBuffer( GLFBDevBufferPtr buffer )
       }
       {
          struct gl_framebuffer *fb = &buffer->glframebuffer;
-         _mesa_unreference_framebuffer(&fb);
+         _mesa_reference_framebuffer(&fb, NULL);
       }
    }
 }
diff --git a/src/mesa/drivers/glide/fxapi.c b/src/mesa/drivers/glide/fxapi.c
index b7708fd636a..238f4915995 100644
--- a/src/mesa/drivers/glide/fxapi.c
+++ b/src/mesa/drivers/glide/fxapi.c
@@ -728,7 +728,7 @@ errorhandler:
        FREE(fxMesa->fogTable);
     }
     if (fxMesa->glBuffer) {
-       _mesa_unreference_framebuffer(&fxMesa->glBuffer);
+       _mesa_reference_framebuffer(&fxMesa->glBuffer, NULL);
     }
     if (fxMesa->glVis) {
        _mesa_destroy_visual(fxMesa->glVis);
@@ -828,7 +828,7 @@ fxMesaDestroyContext(fxMesaContext fxMesa)
    fxDDDestroyFxMesaContext(fxMesa); /* must be before _mesa_destroy_context */
    _mesa_destroy_visual(fxMesa->glVis);
    _mesa_destroy_context(fxMesa->glCtx);
-   _mesa_unreference_framebuffer(&fxMesa->glBuffer);
+   _mesa_reference_framebuffer(&fxMesa->glBuffer, NULL);
    fxTMClose(fxMesa); /* must be after _mesa_destroy_context */
 
    FREE(fxMesa);
diff --git a/src/mesa/drivers/glide/fxdd.c b/src/mesa/drivers/glide/fxdd.c
index 213ef2382fe..338cd37382a 100644
--- a/src/mesa/drivers/glide/fxdd.c
+++ b/src/mesa/drivers/glide/fxdd.c
@@ -1773,6 +1773,8 @@ fxDDInitFxMesaContext(fxMesaContext fxMesa)
    ctx->Const.MaxTextureImageUnits = fxMesa->haveTwoTMUs ? 2 : 1;
    ctx->Const.MaxTextureUnits = MAX2(ctx->Const.MaxTextureImageUnits, ctx->Const.MaxTextureCoordUnits);
 
+   ctx->Const.MaxDrawBuffers = 1;
+
    fxMesa->new_state = _NEW_ALL;
    if (!fxMesa->haveHwStencil) {
       /* don't touch stencil if there is none */
@@ -1862,7 +1864,6 @@ fxDDInitExtensions(GLcontext * ctx)
    }
 
    if (fxMesa->type >= GR_SSTTYPE_Voodoo4) {
-      _mesa_enable_extension(ctx, "GL_ARB_texture_compression");
       _mesa_enable_extension(ctx, "GL_3DFX_texture_compression_FXT1");
       _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
       _mesa_enable_extension(ctx, "GL_S3_s3tc");
@@ -1894,7 +1895,6 @@ fxDDInitExtensions(GLcontext * ctx)
        * fxSetupSingleTMU_NoLock/fxSetupDoubleTMU_NoLock:
        *    grTexDownloadTable(GR_TEXTABLE_NCC0, &(ti->palette));
        */
-      /*_mesa_enable_extension(ctx, "GL_ARB_texture_compression");*/
       _mesa_enable_extension(ctx, "GL_SGIS_generate_mipmap");
    }
 
@@ -1917,9 +1917,6 @@ fxDDInitExtensions(GLcontext * ctx)
    }
 
    /* core-level extensions */
-   _mesa_enable_extension(ctx, "GL_EXT_multi_draw_arrays");
-   _mesa_enable_extension(ctx, "GL_IBM_multimode_draw_arrays");
-   _mesa_enable_extension(ctx, "GL_ARB_vertex_buffer_object");
    /* dangerous */
    if (getenv("MESA_FX_ALLOW_VP")) {
       _mesa_enable_extension(ctx, "GL_ARB_vertex_program");
diff --git a/src/mesa/drivers/glslcompiler/Makefile b/src/mesa/drivers/glslcompiler/Makefile
index dc4abd44d49..ffe4e0a2343 100644
--- a/src/mesa/drivers/glslcompiler/Makefile
+++ b/src/mesa/drivers/glslcompiler/Makefile
@@ -33,7 +33,7 @@ default: $(PROGRAM)
 
 
 glslcompiler: $(OBJECTS)
-	$(CC) $(OBJECTS) -lm -lpthread -o $@
+	$(CC) $(OBJECTS) $(GL_LIB_DEPS) -o $@
 
 
 glslcompiler.o: glslcompiler.c
diff --git a/src/mesa/drivers/glslcompiler/glslcompiler.c b/src/mesa/drivers/glslcompiler/glslcompiler.c
index 34cce977c86..918ec418563 100644
--- a/src/mesa/drivers/glslcompiler/glslcompiler.c
+++ b/src/mesa/drivers/glslcompiler/glslcompiler.c
@@ -227,7 +227,7 @@ PrintShaderInstructions(GLuint shader, FILE *f)
    GET_CURRENT_CONTEXT(ctx);
    struct gl_shader *sh = _mesa_lookup_shader(ctx, shader);
    struct gl_program *prog = sh->Program;
-   _mesa_print_program_opt(prog, Options.Mode, Options.LineNumbers);
+   _mesa_fprint_program_opt(stdout, prog, Options.Mode, Options.LineNumbers);
 }
 
 
diff --git a/src/mesa/drivers/osmesa/Makefile b/src/mesa/drivers/osmesa/Makefile
index d6a18b1d755..3b3984200aa 100644
--- a/src/mesa/drivers/osmesa/Makefile
+++ b/src/mesa/drivers/osmesa/Makefile
@@ -30,8 +30,7 @@ CORE_MESA = $(TOP)/src/mesa/libmesa.a $(TOP)/src/mesa/libglapi.a
 	$(CC) -c $(INCLUDE_DIRS) $(CFLAGS) $< -o $@
 
 
-default:
-# $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME)
+default: $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME)
 	@ if [ "${DRIVER_DIRS}" = "osmesa" ] ; then \
 		$(MAKE) osmesa16 ; \
 	else \
@@ -42,9 +41,9 @@ default:
 
 
 # The normal libOSMesa is used in conjuction with libGL
-osmesa8: $(TOP)/lib/$(OSMESA_LIB_NAME)
+osmesa8: $(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME)
 
-$(TOP)/lib/$(OSMESA_LIB_NAME): $(OBJECTS)
+$(TOP)/$(LIB_DIR)/$(OSMESA_LIB_NAME): $(OBJECTS)
 	$(MKLIB) -o $(OSMESA_LIB) -linker '$(CC)' -ldflags '$(LDFLAGS)' \
 		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
 		-install $(TOP)/$(LIB_DIR) $(MKLIB_OPTIONS) \
diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c
index f2367bbbb70..904659e3450 100644
--- a/src/mesa/drivers/osmesa/osmesa.c
+++ b/src/mesa/drivers/osmesa/osmesa.c
@@ -1310,7 +1310,7 @@ OSMesaDestroyContext( OSMesaContext osmesa )
       _swrast_DestroyContext( &osmesa->mesa );
 
       _mesa_destroy_visual( osmesa->gl_visual );
-      _mesa_unreference_framebuffer( &osmesa->gl_buffer );
+      _mesa_reference_framebuffer( &osmesa->gl_buffer, NULL );
 
       _mesa_free_context_data( &osmesa->mesa );
       _mesa_free( osmesa );
diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def
index 3f2d644e86e..b386e34aadf 100644
--- a/src/mesa/drivers/windows/gdi/mesa.def
+++ b/src/mesa/drivers/windows/gdi/mesa.def
@@ -867,6 +867,7 @@ EXPORTS
 	_glapi_get_proc_address
 	_mesa_add_soft_renderbuffers
 	_mesa_add_renderbuffer
+	_mesa_begin_query
 	_mesa_buffer_data
 	_mesa_buffer_get_subdata
 	_mesa_buffer_map
@@ -881,6 +882,7 @@ EXPORTS
 	_mesa_delete_array_object
 	_mesa_delete_buffer_object
 	_mesa_delete_program
+	_mesa_delete_query
 	_mesa_delete_texture_object
 	_mesa_destroy_framebuffer
 	_mesa_destroy_visual
@@ -890,6 +892,7 @@ EXPORTS
 	_mesa_enable_2_0_extensions
 	_mesa_enable_2_1_extensions
 	_mesa_enable_sw_extensions
+	_mesa_end_query
 	_mesa_error
 	_mesa_finish_render_texture
 	_mesa_framebuffer_renderbuffer
@@ -941,6 +944,7 @@ EXPORTS
 	_mesa_update_framebuffer_visual
 	_mesa_use_program
 	_mesa_Viewport
+	_mesa_wait_query
 	_swrast_Accum
 	_swrast_Bitmap
 	_swrast_BlitFramebuffer
@@ -973,3 +977,4 @@ EXPORTS
 	_tnl_InvalidateState
 	_tnl_run_pipeline
 	_tnl_program_string
+	_tnl_RasterPos
+\ No newline at end of file
diff --git a/src/mesa/drivers/windows/gdi/wgl.c b/src/mesa/drivers/windows/gdi/wgl.c
index 9f0bb9122a9..8d8087067f5 100644
--- a/src/mesa/drivers/windows/gdi/wgl.c
+++ b/src/mesa/drivers/windows/gdi/wgl.c
@@ -601,8 +601,9 @@ WINGDIAPI BOOL GLAPIENTRY wglCopyContext(HGLRC hglrcSrc,
 WINGDIAPI HGLRC GLAPIENTRY wglCreateLayerContext(HDC hdc,
 						 int iLayerPlane)
 {
-    (void) hdc; (void) iLayerPlane;
     SetLastError(0);
+    if (iLayerPlane == 0)
+        return wglCreateContext( hdc );
     return(NULL);
 }
 
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 9dafb747231..80746950c71 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -56,11 +56,13 @@ wmesa_free_framebuffer(HDC hdc)
 	prev = pwfb;
     }
     if (pwfb) {
+        struct gl_framebuffer *fb;
 	if (pwfb == FirstFramebuffer)
 	    FirstFramebuffer = pwfb->next;
 	else
 	    prev->next = pwfb->next;
-	free(pwfb);
+        fb = &pwfb->Base;
+        _mesa_unreference_framebuffer(&fb); 
     }
 }
 
@@ -1503,6 +1505,9 @@ WMesaContext WMesaCreateContext(HDC hDC,
     ctx = &c->gl_ctx;
     _mesa_initialize_context(ctx, visual, NULL, &functions, (void *)c);
 
+    /* visual no longer needed - it was copied by _mesa_initialize_context() */
+    _mesa_destroy_visual(visual);
+
     _mesa_enable_sw_extensions(ctx);
     _mesa_enable_1_3_extensions(ctx);
     _mesa_enable_1_4_extensions(ctx);
diff --git a/src/mesa/drivers/windows/gldirect/dglcontext.c b/src/mesa/drivers/windows/gldirect/dglcontext.c
index 4ad7a76e67f..e9c23d1ccb0 100644
--- a/src/mesa/drivers/windows/gldirect/dglcontext.c
+++ b/src/mesa/drivers/windows/gldirect/dglcontext.c
@@ -1482,6 +1482,7 @@ SkipPrimaryCreate:
 #else
 	lpCtx->glCtx->Const.MaxTextureSize = 1024;
 #endif
+	lpCtx->glCtx->Const.MaxDrawBuffers = 1;
 
 	// Setup the Display Driver pointers
 	dglSetupDDPointers(lpCtx->glCtx);
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_wgl_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_wgl_dx7.c
index 0f8fe33eb15..fa44a952a09 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_wgl_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_wgl_dx7.c
@@ -1422,6 +1422,8 @@ BOOL gldInitialiseMesa_DX(
 		lpCtx->glCtx->Const.MaxTextureUnits = 1;
 	}
 
+	lpCtx->glCtx->Const.MaxDrawBuffers = 1;
+
 	// max texture size
 //	MaxTextureSize = min(gld->d3dCaps8.MaxTextureHeight, gld->d3dCaps8.MaxTextureWidth);
 	MaxTextureSize = min(gld->d3dCaps.dwMaxTextureHeight, gld->d3dCaps.dwMaxTextureWidth);
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_wgl_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_wgl_dx8.c
index 690f68b68f1..011d810e975 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_wgl_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_wgl_dx8.c
@@ -1204,6 +1204,7 @@ BOOL gldInitialiseMesa_DX(
 		MaxTextureSize >>= 1;
 	}
 	lpCtx->glCtx->Const.MaxTextureLevels = (TextureLevels) ? TextureLevels : 8;
+	lpCtx->glCtx->Const.MaxDrawBuffers = 1;
 
 	IDirect3DDevice8_SetRenderState(gld->pDev, D3DRS_LIGHTING, FALSE);
 	IDirect3DDevice8_SetRenderState(gld->pDev, D3DRS_CULLMODE, D3DCULL_NONE);
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_wgl_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_wgl_dx9.c
index dc465c54185..a03b865bb44 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_wgl_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_wgl_dx9.c
@@ -1206,6 +1206,7 @@ BOOL gldInitialiseMesa_DX(
 		MaxTextureSize >>= 1;
 	}
 	lpCtx->glCtx->Const.MaxTextureLevels = (TextureLevels) ? TextureLevels : 8;
+	lpCtx->glCtx->Const.MaxDrawBuffers = 1;
 
 	IDirect3DDevice9_SetRenderState(gld->pDev, D3DRS_LIGHTING, FALSE);
 	IDirect3DDevice9_SetRenderState(gld->pDev, D3DRS_CULLMODE, D3DCULL_NONE);
diff --git a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
index b590dc795a0..342a742867a 100644
--- a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
+++ b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
@@ -1625,6 +1625,7 @@ BOOL gldInitialiseMesa_MesaSW(
 
 	// Added this to force max texture diminsion to 256. KeithH
 	ctx->Const.MaxTextureLevels = 8;
+	ctx->Const.MaxDrawBuffers = 1;
 
 	_mesa_enable_sw_extensions(ctx);
 	_mesa_enable_imaging_extensions(ctx);
diff --git a/src/mesa/drivers/x11/fakeglx.c b/src/mesa/drivers/x11/fakeglx.c
index 827d39f9957..3b004a3ee25 100644
--- a/src/mesa/drivers/x11/fakeglx.c
+++ b/src/mesa/drivers/x11/fakeglx.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.5
  *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -42,7 +43,6 @@
 
 #include "glxheader.h"
 #include "glxapi.h"
-#include "GL/xmesa.h"
 #include "main/context.h"
 #include "main/config.h"
 #include "main/macros.h"
@@ -1393,6 +1393,25 @@ Fake_glXChooseVisual( Display *dpy, int screen, int *list )
 }
 
 
+/**
+ * Init basic fields of a new fake_glx_context.
+ */
+static void
+init_glx_context(struct fake_glx_context *glxCtx, Display *dpy)
+{
+   /* Always return True.  See if anyone's confused... */
+   GLboolean direct = GL_TRUE;
+
+   glxCtx->xmesaContext->direct = direct;
+   glxCtx->glxContext.isDirect = direct;
+   glxCtx->glxContext.currentDpy = dpy;
+   glxCtx->glxContext.xid = (XID) glxCtx;  /* self pointer */
+
+   assert((void *) glxCtx == (void *) &(glxCtx->glxContext));
+}
+
+
+
 static GLXContext
 Fake_glXCreateContext( Display *dpy, XVisualInfo *visinfo,
                        GLXContext share_list, Bool direct )
@@ -1431,12 +1450,7 @@ Fake_glXCreateContext( Display *dpy, XVisualInfo *visinfo,
       return NULL;
    }
 
-   glxCtx->xmesaContext->direct = GL_FALSE;
-   glxCtx->glxContext.isDirect = GL_FALSE;
-   glxCtx->glxContext.currentDpy = dpy;
-   glxCtx->glxContext.xid = (XID) glxCtx;  /* self pointer */
-
-   assert((void *) glxCtx == (void *) &(glxCtx->glxContext));
+   init_glx_context(glxCtx, dpy);
 
    return (GLXContext) glxCtx;
 }
@@ -2442,12 +2456,7 @@ Fake_glXCreateNewContext( Display *dpy, GLXFBConfig config,
       return NULL;
    }
 
-   glxCtx->xmesaContext->direct = GL_FALSE;
-   glxCtx->glxContext.isDirect = GL_FALSE;
-   glxCtx->glxContext.currentDpy = dpy;
-   glxCtx->glxContext.xid = (XID) glxCtx;  /* self pointer */
-
-   assert((void *) glxCtx == (void *) &(glxCtx->glxContext));
+   init_glx_context(glxCtx, dpy);
 
    return (GLXContext) glxCtx;
 }
@@ -2665,12 +2674,7 @@ Fake_glXCreateContextWithConfigSGIX(Display *dpy, GLXFBConfigSGIX config, int re
       return NULL;
    }
 
-   glxCtx->xmesaContext->direct = GL_FALSE;
-   glxCtx->glxContext.isDirect = GL_FALSE;
-   glxCtx->glxContext.currentDpy = dpy;
-   glxCtx->glxContext.xid = (XID) glxCtx;  /* self pointer */
-
-   assert((void *) glxCtx == (void *) &(glxCtx->glxContext));
+   init_glx_context(glxCtx, dpy);
 
    return (GLXContext) glxCtx;
 }
diff --git a/src/mesa/drivers/x11/fxmesa.h b/src/mesa/drivers/x11/fxmesa.h
new file mode 100644
index 00000000000..f8e9661f9cc
--- /dev/null
+++ b/src/mesa/drivers/x11/fxmesa.h
@@ -0,0 +1,103 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  4.0
+ * Copyright (C) 1995-2001  Brian Paul
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Library General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Library General Public License for more details.
+ *
+ * You should have received a copy of the GNU Library General Public
+ * License along with this library; if not, write to the Free
+ * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+ */
+
+
+/*
+ * FXMesa - 3Dfx Glide driver for Mesa.  Contributed by David Bucciarelli
+ *
+ * NOTE: This version requires Glide3 (http://sourceforge.net/projects/glide)
+ */
+
+
+#ifndef FXMESA_H
+#define FXMESA_H
+
+
+#include <glide.h>
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+
+#define FXMESA_MAJOR_VERSION 6
+#define FXMESA_MINOR_VERSION 3
+
+
+/*
+ * Values for attribList parameter to fxMesaCreateContext():
+ */
+#define FXMESA_NONE		0	/* to terminate attribList */
+#define FXMESA_DOUBLEBUFFER	10
+#define FXMESA_ALPHA_SIZE	11      /* followed by an integer */
+#define FXMESA_DEPTH_SIZE	12      /* followed by an integer */
+#define FXMESA_STENCIL_SIZE	13      /* followed by an integer */
+#define FXMESA_ACCUM_SIZE	14      /* followed by an integer */
+#define FXMESA_COLORDEPTH	20      /* followed by an integer */
+#define FXMESA_SHARE_CONTEXT 990099	/* keep in sync with xmesa1.c! */
+
+
+
+typedef struct tfxMesaContext *fxMesaContext;
+
+
+#if defined (__BEOS__)
+#pragma export on
+#endif
+
+
+GLAPI fxMesaContext GLAPIENTRY fxMesaCreateContext(GLuint win, GrScreenResolution_t,
+						  GrScreenRefresh_t,
+						  const GLint attribList[]);
+
+GLAPI fxMesaContext GLAPIENTRY fxMesaCreateBestContext(GLuint win,
+						      GLint width, GLint height,
+						      const GLint attribList[]);
+GLAPI void GLAPIENTRY fxMesaDestroyContext(fxMesaContext ctx);
+
+GLAPI GLint GLAPIENTRY fxMesaSelectCurrentBoard(int n);
+
+GLAPI void GLAPIENTRY fxMesaMakeCurrent(fxMesaContext ctx);
+
+GLAPI fxMesaContext GLAPIENTRY fxMesaGetCurrentContext(void);
+
+GLAPI void GLAPIENTRY fxMesaSwapBuffers(void);
+
+GLAPI void GLAPIENTRY fxMesaSetNearFar(GLfloat nearVal, GLfloat farVal);
+
+GLAPI void GLAPIENTRY fxMesaUpdateScreenSize(fxMesaContext ctx);
+
+GLAPI void GLAPIENTRY fxCloseHardware(void);
+
+GLAPI void GLAPIENTRY fxGetScreenGeometry (GLint *w, GLint *h);
+
+
+#if defined (__BEOS__)
+#pragma export off
+#endif
+
+
+#ifdef __cplusplus
+}
+#endif
+
+
+#endif
diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c
index c9009bad031..2a8c3a4fd36 100644
--- a/src/mesa/drivers/x11/xm_api.c
+++ b/src/mesa/drivers/x11/xm_api.c
@@ -63,7 +63,6 @@
 #endif
 
 #include "glxheader.h"
-#include "GL/xmesa.h"
 #include "xmesaP.h"
 #include "main/context.h"
 #include "main/extensions.h"
@@ -493,7 +492,7 @@ xmesa_free_buffer(XMesaBuffer buffer)
          b->frontxrb->drawable = 0;
 
          /* Unreference.  If count = zero we'll really delete the buffer */
-         _mesa_unreference_framebuffer(&fb);
+         _mesa_reference_framebuffer(&fb, NULL);
 
          return;
       }
@@ -1326,7 +1325,6 @@ xmesa_convert_from_x_visual_type( int visualType )
 #define need_GL_EXT_framebuffer_blit
 #define need_GL_EXT_gpu_program_parameters
 #define need_GL_EXT_paletted_texture
-#define need_GL_IBM_multimode_draw_arrays
 #define need_GL_MESA_resize_buffers
 #define need_GL_NV_vertex_program
 #define need_GL_NV_fragment_program
@@ -1356,7 +1354,6 @@ const struct dri_extension card_extensions[] =
    { "GL_EXT_framebuffer_blit",		GL_EXT_framebuffer_blit_functions },
    { "GL_EXT_gpu_program_parameters",	GL_EXT_gpu_program_parameters_functions },
    { "GL_EXT_paletted_texture",		GL_EXT_paletted_texture_functions },
-   { "GL_IBM_multimode_draw_arrays",	GL_IBM_multimode_draw_arrays_functions },
    { "GL_MESA_resize_buffers",		GL_MESA_resize_buffers_functions },
    { "GL_NV_vertex_program",		GL_NV_vertex_program_functions },
    { "GL_NV_fragment_program",		GL_NV_fragment_program_functions },
diff --git a/src/mesa/drivers/x11/xm_buffer.c b/src/mesa/drivers/x11/xm_buffer.c
index f104d44d051..7ad67bc34da 100644
--- a/src/mesa/drivers/x11/xm_buffer.c
+++ b/src/mesa/drivers/x11/xm_buffer.c
@@ -30,7 +30,6 @@
 
 
 #include "glxheader.h"
-#include "GL/xmesa.h"
 #include "xmesaP.h"
 #include "main/imports.h"
 #include "main/framebuffer.h"
diff --git a/src/mesa/drivers/x11/xmesa.h b/src/mesa/drivers/x11/xmesa.h
new file mode 100644
index 00000000000..98139af8336
--- /dev/null
+++ b/src/mesa/drivers/x11/xmesa.h
@@ -0,0 +1,424 @@
+/*
+ * Mesa 3-D graphics library
+ * Version:  7.1
+ * 
+ * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ * 
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/*
+ * Mesa/X11 interface.  This header file serves as the documentation for
+ * the Mesa/X11 interface functions.
+ *
+ * Note: this interface isn't intended for user programs.  It's primarily
+ * just for implementing the pseudo-GLX interface.
+ */
+
+
+/* Sample Usage:
+
+In addition to the usual X calls to select a visual, create a colormap
+and create a window, you must do the following to use the X/Mesa interface:
+
+1. Call XMesaCreateVisual() to make an XMesaVisual from an XVisualInfo.
+
+2. Call XMesaCreateContext() to create an X/Mesa rendering context, given
+   the XMesaVisual.
+
+3. Call XMesaCreateWindowBuffer() to create an XMesaBuffer from an X window
+   and XMesaVisual.
+
+4. Call XMesaMakeCurrent() to bind the XMesaBuffer to an XMesaContext and
+   to make the context the current one.
+
+5. Make gl* calls to render your graphics.
+
+6. Use XMesaSwapBuffers() when double buffering to swap front/back buffers.
+
+7. Before the X window is destroyed, call XMesaDestroyBuffer().
+
+8. Before exiting, call XMesaDestroyVisual and XMesaDestroyContext.
+
+*/
+
+
+
+
+#ifndef XMESA_H
+#define XMESA_H
+
+#ifdef __VMS
+#include <GL/vms_x_fix.h>
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifdef XFree86Server
+#include "xmesa_xf86.h"
+#else
+#include <X11/Xlib.h>
+#include <X11/Xutil.h>
+#include "xmesa_x.h"
+#endif
+#include "GL/gl.h"
+
+#ifdef AMIWIN
+#include <pragmas/xlib_pragmas.h>
+extern struct Library *XLibBase;
+#endif
+
+
+#define XMESA_MAJOR_VERSION 6
+#define XMESA_MINOR_VERSION 3
+
+
+
+/*
+ * Values passed to XMesaGetString:
+ */
+#define XMESA_VERSION 1
+#define XMESA_EXTENSIONS 2
+
+
+/*
+ * Values passed to XMesaSetFXmode:
+ */
+#define XMESA_FX_WINDOW       1
+#define XMESA_FX_FULLSCREEN   2
+
+
+
+typedef struct xmesa_context *XMesaContext;
+
+typedef struct xmesa_visual *XMesaVisual;
+
+typedef struct xmesa_buffer *XMesaBuffer;
+
+
+
+/*
+ * Create a new X/Mesa visual.
+ * Input:  display - X11 display
+ *         visinfo - an XVisualInfo pointer
+ *         rgb_flag - GL_TRUE = RGB mode,
+ *                    GL_FALSE = color index mode
+ *         alpha_flag - alpha buffer requested?
+ *         db_flag - GL_TRUE = double-buffered,
+ *                   GL_FALSE = single buffered
+ *         stereo_flag - stereo visual?
+ *         ximage_flag - GL_TRUE = use an XImage for back buffer,
+ *                       GL_FALSE = use an off-screen pixmap for back buffer
+ *         depth_size - requested bits/depth values, or zero
+ *         stencil_size - requested bits/stencil values, or zero
+ *         accum_red_size - requested bits/red accum values, or zero
+ *         accum_green_size - requested bits/green accum values, or zero
+ *         accum_blue_size - requested bits/blue accum values, or zero
+ *         accum_alpha_size - requested bits/alpha accum values, or zero
+ *         num_samples - number of samples/pixel if multisampling, or zero
+ *         level - visual level, usually 0
+ *         visualCaveat - ala the GLX extension, usually GLX_NONE_EXT
+ * Return;  a new XMesaVisual or 0 if error.
+ */
+extern XMesaVisual XMesaCreateVisual( XMesaDisplay *display,
+                                      XMesaVisualInfo visinfo,
+                                      GLboolean rgb_flag,
+                                      GLboolean alpha_flag,
+                                      GLboolean db_flag,
+                                      GLboolean stereo_flag,
+                                      GLboolean ximage_flag,
+                                      GLint depth_size,
+                                      GLint stencil_size,
+                                      GLint accum_red_size,
+                                      GLint accum_green_size,
+                                      GLint accum_blue_size,
+                                      GLint accum_alpha_size,
+                                      GLint num_samples,
+                                      GLint level,
+                                      GLint visualCaveat );
+
+/*
+ * Destroy an XMesaVisual, but not the associated XVisualInfo.
+ */
+extern void XMesaDestroyVisual( XMesaVisual v );
+
+
+
+/*
+ * Create a new XMesaContext for rendering into an X11 window.
+ *
+ * Input:  visual - an XMesaVisual
+ *         share_list - another XMesaContext with which to share display
+ *                      lists or NULL if no sharing is wanted.
+ * Return:  an XMesaContext or NULL if error.
+ */
+extern XMesaContext XMesaCreateContext( XMesaVisual v,
+					XMesaContext share_list );
+
+
+/*
+ * Destroy a rendering context as returned by XMesaCreateContext()
+ */
+extern void XMesaDestroyContext( XMesaContext c );
+
+
+#ifdef XFree86Server
+/*
+ * These are the extra routines required for integration with XFree86.
+ * None of these routines should be user visible. -KEM
+ */
+extern GLboolean XMesaForceCurrent( XMesaContext c );
+
+extern GLboolean XMesaLoseCurrent( XMesaContext c );
+
+extern GLboolean XMesaCopyContext( XMesaContext src,
+				   XMesaContext dst,
+				   GLuint mask );
+#endif /* XFree86Server */
+
+
+/*
+ * Create an XMesaBuffer from an X window.
+ */
+extern XMesaBuffer XMesaCreateWindowBuffer( XMesaVisual v, XMesaWindow w );
+
+
+/*
+ * Create an XMesaBuffer from an X pixmap.
+ */
+extern XMesaBuffer XMesaCreatePixmapBuffer( XMesaVisual v,
+					    XMesaPixmap p,
+					    XMesaColormap cmap );
+
+
+/*
+ * Destroy an XMesaBuffer, but not the corresponding window or pixmap.
+ */
+extern void XMesaDestroyBuffer( XMesaBuffer b );
+
+
+/*
+ * Return the XMesaBuffer handle which corresponds to an X drawable, if any.
+ *
+ * New in Mesa 2.3.
+ */
+extern XMesaBuffer XMesaFindBuffer( XMesaDisplay *dpy,
+				    XMesaDrawable d );
+
+
+
+/*
+ * Bind a buffer to a context and make the context the current one.
+ */
+extern GLboolean XMesaMakeCurrent( XMesaContext c,
+				   XMesaBuffer b );
+
+
+/*
+ * Bind two buffers (read and draw) to a context and make the
+ * context the current one.
+ * New in Mesa 3.3
+ */
+extern GLboolean XMesaMakeCurrent2( XMesaContext c,
+                                    XMesaBuffer drawBuffer,
+                                    XMesaBuffer readBuffer );
+
+
+/*
+ * Unbind the current context from its buffer.
+ */
+extern GLboolean XMesaUnbindContext( XMesaContext c );
+
+
+/*
+ * Return a handle to the current context.
+ */
+extern XMesaContext XMesaGetCurrentContext( void );
+
+
+/*
+ * Return handle to the current (draw) buffer.
+ */
+extern XMesaBuffer XMesaGetCurrentBuffer( void );
+
+
+/*
+ * Return handle to the current read buffer.
+ * New in Mesa 3.3
+ */
+extern XMesaBuffer XMesaGetCurrentReadBuffer( void );
+
+
+/*
+ * Swap the front and back buffers for the given buffer.  No action is
+ * taken if the buffer is not double buffered.
+ */
+extern void XMesaSwapBuffers( XMesaBuffer b );
+
+
+/*
+ * Copy a sub-region of the back buffer to the front buffer.
+ *
+ * New in Mesa 2.6
+ */
+extern void XMesaCopySubBuffer( XMesaBuffer b,
+				int x,
+				int y,
+				int width,
+				int height );
+
+
+/*
+ * Return a pointer to the the Pixmap or XImage being used as the back
+ * color buffer of an XMesaBuffer.  This function is a way to get "under
+ * the hood" of X/Mesa so one can manipulate the back buffer directly.
+ * Input:  b - the XMesaBuffer
+ * Output:  pixmap - pointer to back buffer's Pixmap, or 0
+ *          ximage - pointer to back buffer's XImage, or NULL
+ * Return:  GL_TRUE = context is double buffered
+ *          GL_FALSE = context is single buffered
+ */
+extern GLboolean XMesaGetBackBuffer( XMesaBuffer b,
+				     XMesaPixmap *pixmap,
+				     XMesaImage **ximage );
+
+
+
+/*
+ * Return the depth buffer associated with an XMesaBuffer.
+ * Input:  b - the XMesa buffer handle
+ * Output:  width, height - size of buffer in pixels
+ *          bytesPerValue - bytes per depth value (2 or 4)
+ *          buffer - pointer to depth buffer values
+ * Return:  GL_TRUE or GL_FALSE to indicate success or failure.
+ *
+ * New in Mesa 2.4.
+ */
+extern GLboolean XMesaGetDepthBuffer( XMesaBuffer b,
+				      GLint *width,
+				      GLint *height,
+				      GLint *bytesPerValue,
+				      void **buffer );
+
+
+
+/*
+ * Flush/sync a context
+ */
+extern void XMesaFlush( XMesaContext c );
+
+
+
+/*
+ * Get an X/Mesa-specific string.
+ * Input:  name - either XMESA_VERSION or XMESA_EXTENSIONS
+ */
+extern const char *XMesaGetString( XMesaContext c, int name );
+
+
+
+/*
+ * Scan for XMesaBuffers whose window/pixmap has been destroyed, then free
+ * any memory used by that buffer.
+ *
+ * New in Mesa 2.3.
+ */
+extern void XMesaGarbageCollect( void );
+
+
+
+/*
+ * Return a dithered pixel value.
+ * Input:  c - XMesaContext
+ *         x, y - window coordinate
+ *         red, green, blue, alpha - color components in [0,1]
+ * Return:  pixel value
+ *
+ * New in Mesa 2.3.
+ */
+extern unsigned long XMesaDitherColor( XMesaContext xmesa,
+				       GLint x,
+				       GLint y,
+				       GLfloat red,
+				       GLfloat green,
+				       GLfloat blue,
+				       GLfloat alpha );
+
+
+
+/*
+ * 3Dfx Glide driver only!
+ * Set 3Dfx/Glide full-screen or window rendering mode.
+ * Input:  mode - either XMESA_FX_WINDOW (window rendering mode) or
+ *                XMESA_FX_FULLSCREEN (full-screen rendering mode)
+ * Return:  GL_TRUE if success
+ *          GL_FALSE if invalid mode or if not using 3Dfx driver
+ *
+ * New in Mesa 2.6.
+ */
+extern GLboolean XMesaSetFXmode( GLint mode );
+
+
+
+/*
+ * Reallocate the back/depth/stencil/accum/etc/ buffers associated with
+ * buffer <b> if its size has changed.
+ *
+ * New in Mesa 4.0.2
+ */
+extern void XMesaResizeBuffers( XMesaBuffer b );
+
+
+
+/*
+ * Create a pbuffer.
+ * New in Mesa 4.1
+ */
+extern XMesaBuffer XMesaCreatePBuffer(XMesaVisual v, XMesaColormap cmap,
+                                      unsigned int width, unsigned int height);
+
+
+
+/*
+ * Texture from Pixmap
+ * New in Mesa 7.1
+ */
+extern void
+XMesaBindTexImage(XMesaDisplay *dpy, XMesaBuffer drawable, int buffer,
+                  const int *attrib_list);
+
+extern void
+XMesaReleaseTexImage(XMesaDisplay *dpy, XMesaBuffer drawable, int buffer);
+
+
+extern XMesaBuffer
+XMesaCreatePixmapTextureBuffer(XMesaVisual v, XMesaPixmap p,
+                               XMesaColormap cmap,
+                               int format, int target, int mipmap);
+
+
+
+#ifdef __cplusplus
+}
+#endif
+
+
+#endif
diff --git a/src/mesa/drivers/x11/xmesaP.h b/src/mesa/drivers/x11/xmesaP.h
index 98867ac7106..65e747d7b9d 100644
--- a/src/mesa/drivers/x11/xmesaP.h
+++ b/src/mesa/drivers/x11/xmesaP.h
@@ -27,10 +27,10 @@
 #define XMESAP_H
 
 
-#include "GL/xmesa.h"
+#include "xmesa.h"
 #include "main/mtypes.h"
 #if defined(FX)
-#include "GL/fxmesa.h"
+#include "fxmesa.h"
 #include "xm_glide.h"
 #endif
 #ifdef XFree86Server
diff --git a/src/mesa/drivers/x11/xmesa_x.h b/src/mesa/drivers/x11/xmesa_x.h
new file mode 100644
index 00000000000..865bab4313b
--- /dev/null
+++ b/src/mesa/drivers/x11/xmesa_x.h
@@ -0,0 +1,86 @@
+
+/**************************************************************************
+
+Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sub license, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
+ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/*
+ * Authors:
+ *   Kevin E. Martin <[email protected]>
+ *
+ * When we're building the XMesa driver for stand-alone Mesa we
+ * include this file when building the xm_*.c files.
+ * We need to define some types and macros differently when building
+ * in the Xserver vs. stand-alone Mesa.
+ */
+
+#ifndef _XMESA_X_H_
+#define _XMESA_X_H_
+
+typedef Display      XMesaDisplay;
+typedef Pixmap       XMesaPixmap;
+typedef Colormap     XMesaColormap;
+typedef Drawable     XMesaDrawable;
+typedef Window       XMesaWindow;
+typedef GC           XMesaGC;
+typedef XVisualInfo *XMesaVisualInfo;
+typedef XImage       XMesaImage;
+typedef XPoint       XMesaPoint;
+typedef XColor       XMesaColor;
+
+#define XMesaDestroyImage      XDestroyImage
+
+#define XMesaPutPixel          XPutPixel
+#define XMesaGetPixel          XGetPixel
+
+#define XMesaSetForeground     XSetForeground
+#define XMesaSetBackground     XSetBackground
+#define XMesaSetPlaneMask      XSetPlaneMask
+#define XMesaSetFunction       XSetFunction
+#define XMesaSetFillStyle      XSetFillStyle
+#define XMesaSetTile           XSetTile
+
+#define XMesaDrawPoint         XDrawPoint
+#define XMesaDrawPoints        XDrawPoints
+#define XMesaDrawLine          XDrawLine
+#define XMesaFillRectangle     XFillRectangle
+#define XMesaGetImage          XGetImage
+#define XMesaPutImage          XPutImage
+#define XMesaCopyArea          XCopyArea
+
+#define XMesaCreatePixmap      XCreatePixmap
+#define XMesaFreePixmap        XFreePixmap
+#define XMesaFreeGC            XFreeGC
+
+#define GET_COLORMAP_SIZE(__v)  __v->visinfo->colormap_size
+#define GET_REDMASK(__v)        __v->mesa_visual.redMask
+#define GET_GREENMASK(__v)      __v->mesa_visual.greenMask
+#define GET_BLUEMASK(__v)       __v->mesa_visual.blueMask
+#define GET_VISUAL_DEPTH(__v)   __v->visinfo->depth
+#define GET_BLACK_PIXEL(__v)    BlackPixel(__v->display, __v->mesa_visual.screen)
+#define CHECK_BYTE_ORDER(__v)   host_byte_order()==ImageByteOrder(__v->display)
+#define CHECK_FOR_HPCR(__v)     XInternAtom(__v->display, "_HP_RGB_SMOOTH_MAP_LIST", True)
+
+#endif
diff --git a/src/mesa/drivers/x11/xmesa_xf86.h b/src/mesa/drivers/x11/xmesa_xf86.h
new file mode 100644
index 00000000000..10f93c3ab62
--- /dev/null
+++ b/src/mesa/drivers/x11/xmesa_xf86.h
@@ -0,0 +1,198 @@
+
+/**************************************************************************
+
+Copyright 1998-1999 Precision Insight, Inc., Cedar Park, Texas.
+All Rights Reserved.
+
+Permission is hereby granted, free of charge, to any person obtaining a
+copy of this software and associated documentation files (the
+"Software"), to deal in the Software without restriction, including
+without limitation the rights to use, copy, modify, merge, publish,
+distribute, sub license, and/or sell copies of the Software, and to
+permit persons to whom the Software is furnished to do so, subject to
+the following conditions:
+
+The above copyright notice and this permission notice (including the
+next paragraph) shall be included in all copies or substantial portions
+of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+IN NO EVENT SHALL PRECISION INSIGHT AND/OR ITS SUPPLIERS BE LIABLE FOR
+ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+**************************************************************************/
+
+/*
+ * Authors:
+ *   Kevin E. Martin <[email protected]>
+ *
+ * When we're building the XMesa driver for use in the X server (as the
+ * indirect render) we include this file when building the xm_*.c files.
+ * We need to define some types and macros differently when building
+ * in the Xserver vs. stand-alone Mesa.
+ */
+
+#ifndef _XMESA_XF86_H_
+#define _XMESA_XF86_H_
+
+#include "GL/glxtokens.h"
+#include "scrnintstr.h"
+#include "pixmapstr.h"
+#include "gcstruct.h"
+#include "servermd.h"
+
+
+typedef struct _XMesaImageRec {
+    int width, height;
+    char *data;
+    int bytes_per_line; /* Padded to 32 bits */
+    int bits_per_pixel;
+} XMesaImage;
+
+typedef ScreenRec   XMesaDisplay;
+typedef PixmapPtr   XMesaPixmap;
+typedef ColormapPtr XMesaColormap;
+typedef DrawablePtr XMesaDrawable;
+typedef WindowPtr   XMesaWindow;
+typedef GCPtr       XMesaGC;
+typedef VisualPtr   XMesaVisualInfo;
+typedef DDXPointRec XMesaPoint;
+typedef xColorItem  XMesaColor;
+
+#define XMesaSetGeneric(__d,__gc,__val,__mask) \
+do { \
+    CARD32 __v[1]; \
+    (void) __d; \
+    __v[0] = __val; \
+    dixChangeGC(NullClient, __gc, __mask, __v, NULL); \
+} while (0)
+
+#define XMesaSetGenericPtr(__d,__gc,__pval,__mask) \
+do { \
+    ChangeGCVal __v[1]; \
+    (void) __d; \
+    __v[0].ptr = __pval; \
+    dixChangeGC(NullClient, __gc, __mask, NULL, __v); \
+} while (0)
+
+#define XMesaSetForeground(d,gc,v) XMesaSetGeneric(d,gc,v,GCForeground)
+#define XMesaSetBackground(d,gc,v) XMesaSetGeneric(d,gc,v,GCBackground)
+#define XMesaSetPlaneMask(d,gc,v)  XMesaSetGeneric(d,gc,v,GCPlaneMask)
+#define XMesaSetFunction(d,gc,v)   XMesaSetGeneric(d,gc,v,GCFunction)
+#define XMesaSetFillStyle(d,gc,v)  XMesaSetGeneric(d,gc,v,GCFillStyle)
+
+#define XMesaSetTile(d,gc,v)       XMesaSetGenericPtr(d,gc,v,GCTile)
+
+#define XMesaDrawPoint(__d,__b,__gc,__x,__y) \
+do { \
+    XMesaPoint __p[1]; \
+    (void) __d; \
+    __p[0].x = __x; \
+    __p[0].y = __y; \
+    ValidateGC(__b, __gc); \
+    (*gc->ops->PolyPoint)(__b, __gc, CoordModeOrigin, 1, __p); \
+} while (0)
+
+#define XMesaDrawPoints(__d,__b,__gc,__p,__n,__m) \
+do { \
+    (void) __d; \
+    ValidateGC(__b, __gc); \
+    (*gc->ops->PolyPoint)(__b, __gc, __m, __n, __p); \
+} while (0)
+
+#define XMesaDrawLine(__d, __b, __gc, __x0, __y0, __x1, __y1) \
+do { \
+    XMesaPoint __p[2]; \
+    (void) __d; \
+    __p[0].x = __x0; \
+    __p[0].y = __y0; \
+    __p[1].x = __x1; \
+    __p[1].y = __y1; \
+    ValidateGC(__b, __gc); \
+    (*gc->ops->PolyLines)(__b, __gc, CoordModeOrigin, 2, __p); \
+} while (0)
+
+#define XMesaFillRectangle(__d,__b,__gc,__x,__y,__w,__h) \
+do { \
+    xRectangle __r[1]; \
+    (void) __d; \
+    ValidateGC((DrawablePtr)__b, __gc); \
+    __r[0].x = __x; \
+    __r[0].y = __y; \
+    __r[0].width = __w; \
+    __r[0].height = __h; \
+    (*__gc->ops->PolyFillRect)((DrawablePtr)__b, __gc, 1, __r); \
+} while (0)
+
+static _X_INLINE XMesaImage *XMesaGetImage(XMesaDisplay *dpy, PixmapPtr p, int x,
+					int y, unsigned int width,
+					unsigned int height,
+					unsigned long plane_mask, int format)
+{
+    XMesaImage *img = Xcalloc(sizeof(*img));
+
+    img->width = p->drawable.width;
+    img->height = p->drawable.height;
+    img->bits_per_pixel = p->drawable.bitsPerPixel;
+    img->bytes_per_line = PixmapBytePad(width, p->drawable.depth);
+    img->data = malloc(height * img->bytes_per_line);
+
+    /* Assumes: Images are always in ZPixmap format */
+    (*p->drawable.pScreen->GetImage)(&p->drawable, x, y, width, height,
+				     plane_mask, ZPixmap, img->data);
+
+    return img;
+}
+
+#define XMesaPutImage(__d,__b,__gc,__i,__sx,__sy,__x,__y,__w,__h) \
+do { \
+    /* Assumes: Images are always in ZPixmap format */ \
+    (void) __d; \
+    ASSERT(!__sx && !__sy); /* The SubImage case */     \
+    ValidateGC(__b, __gc); \
+    (*__gc->ops->PutImage)(__b, __gc, ((XMesaDrawable)(__b))->depth, \
+			   __x, __y, __w, __h, 0, ZPixmap, \
+			   ((XMesaImage *)(__i))->data); \
+} while (0)
+
+#define XMesaCopyArea(__d,__sb,__db,__gc,__sx,__sy,__w,__h,__x,__y) \
+do { \
+    (void) __d; \
+    ValidateGC(__db, __gc); \
+    (*__gc->ops->CopyArea)((DrawablePtr)__sb, __db, __gc, \
+			   __sx, __sy, __w, __h, __x, __y); \
+} while (0)
+
+
+/* CreatePixmap returns a PixmapPtr; so, it cannot be inside braces */
+#ifdef CREATE_PIXMAP_USAGE_SCRATCH
+#define XMesaCreatePixmap(__d,__b,__w,__h,__depth) \
+  (*__d->CreatePixmap)(__d, __w, __h, __depth, 0)
+#else
+#define XMesaCreatePixmap(__d,__b,__w,__h,__depth) \
+    (*__d->CreatePixmap)(__d, __w, __h, __depth)
+#endif
+
+#define XMesaFreePixmap(__d,__b) \
+    (*__d->DestroyPixmap)(__b)
+
+#define XMesaFreeGC(__d,__gc) \
+do { \
+    (void) __d; \
+    FreeScratchGC(__gc); \
+} while (0)
+
+#define GET_COLORMAP_SIZE(__v)  __v->ColormapEntries
+#define GET_REDMASK(__v)        __v->mesa_visual.redMask
+#define GET_GREENMASK(__v)      __v->mesa_visual.greenMask
+#define GET_BLUEMASK(__v)       __v->mesa_visual.blueMask
+#define GET_VISUAL_DEPTH(__v)   __v->nplanes
+#define GET_BLACK_PIXEL(__v)    __v->display->blackPixel
+#define CHECK_BYTE_ORDER(__v)   GL_TRUE
+#define CHECK_FOR_HPCR(__v)     GL_FALSE
+
+#endif
author	Dave Airlie <[email protected]>	2009-02-13 01:11:25 +1000
committer	Dave Airlie <[email protected]>	2009-02-13 01:11:25 +1000
commit	53116910b10e3b8a05f42970eff311c21808699f (patch)
tree	6831e50a56a02fed324f8cb119e05d837e00d7cf /src/mesa/drivers
parent	7e104f9cde94279a902b408e8d1cf21779b393a8 (diff)
parent	0ccbc3c905f0594a35d72887a1f115e148aaa596 (diff)