Merge branch 'master' of git+ssh://[email protected]/git/mesa/mesa

242 files changed, 16776 insertions, 34674 deletions

diff --git a/src/mesa/drivers/beos/Makefile b/src/mesa/drivers/beos/Makefile
index 1897d6aa3e8..9c7d6affc3b 100644
--- a/src/mesa/drivers/beos/Makefile
+++ b/src/mesa/drivers/beos/Makefile
@@ -179,7 +179,8 @@ $(TOP)/$(LIB_DIR):
 	mkdir $(TOP)/$(LIB_DIR)
 
 $(TOP)/$(LIB_DIR)/$(GL_LIB_NAME): $(OBJECTS) $(MESA_MODULES) $(GLU_MODULES)
-	@$(TOP)/bin/mklib -o $(GL_LIB) -install $(TOP)/$(LIB_DIR) -major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
+	@$(TOP)/bin/mklib -o $(GL_LIB) -ldflags '$(LDFLAGS)' -install $(TOP)/$(LIB_DIR) \
+		-major $(MESA_MAJOR) -minor $(MESA_MINOR) -patch $(MESA_TINY) \
 		$(MKLIB_OPTIONS) $(GL_LIB_DEPS) $(OBJECTS) $(MESA_MODULES) $(GLU_MODULES)
 
 # $(GLU_OBJECTS):
diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c
index 9b1c3f1e060..1e7c279875e 100644
--- a/src/mesa/drivers/common/driverfuncs.c
+++ b/src/mesa/drivers/common/driverfuncs.c
@@ -25,11 +25,10 @@
 
 #include "glheader.h"
 #include "imports.h"
+#include "arrayobj.h"
 #include "buffers.h"
 #include "context.h"
 #include "framebuffer.h"
-#include "program.h"
-#include "prog_execute.h"
 #include "queryobj.h"
 #include "renderbuffer.h"
 #include "texcompress.h"
@@ -44,9 +43,10 @@
 #include "fbobject.h"
 #include "texrender.h"
 #endif
-#include "shader_api.h"
-#include "arrayobj.h"
 
+#include "shader/program.h"
+#include "shader/prog_execute.h"
+#include "shader/shader_api.h"
 #include "driverfuncs.h"
 #include "tnl/tnl.h"
 #include "swrast/swrast.h"
diff --git a/src/mesa/drivers/dri/Makefile.template b/src/mesa/drivers/dri/Makefile.template
index 6f2314ee8cb..3abce004c92 100644
--- a/src/mesa/drivers/dri/Makefile.template
+++ b/src/mesa/drivers/dri/Makefile.template
@@ -13,7 +13,8 @@ COMMON_SOURCES = \
 
 COMMON_BM_SOURCES = \
 	../common/dri_bufmgr.c \
-	../common/dri_drmpool.c
+	../common/dri_bufmgr_ttm.c \
+	../common/dri_bufmgr_fake.c
 
 
 ifeq ($(WINDOW_SYSTEM),dri)
@@ -49,11 +50,6 @@ SHARED_INCLUDES = \
 	-I$(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$(TOP)/src/mesa/swrast_setup \
 	-I$(TOP)/src/egl/main \
 	-I$(TOP)/src/egl/drivers/dri \
 	$(LIBDRM_CFLAGS)
@@ -74,7 +70,7 @@ default: depend symlinks $(LIBNAME) $(TOP)/$(LIB_DIR)/$(LIBNAME)
 
 
 $(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(WINOBJ) Makefile $(TOP)/src/mesa/drivers/dri/Makefile.template
-	$(TOP)/bin/mklib -noprefix -o $@ \
+	$(TOP)/bin/mklib -ldflags '$(LDFLAGS)' -noprefix -o $@ \
 		$(OBJECTS) $(MESA_MODULES) $(WINOBJ) $(DRI_LIB_DEPS)
 
 
@@ -100,8 +96,8 @@ clean:
 
 
 install: $(LIBNAME)
-	$(INSTALL) -d $(DRI_DRIVER_INSTALL_DIR)
-	$(INSTALL) -m 755 $(LIBNAME) $(DRI_DRIVER_INSTALL_DIR)
+	$(INSTALL) -d $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR)
+	$(INSTALL) -m 755 $(LIBNAME) $(DESTDIR)$(DRI_DRIVER_INSTALL_DIR)
 
 
 include depend
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr.c b/src/mesa/drivers/dri/common/dri_bufmgr.c
index eaa4fb09c70..407409bf06d 100644
--- a/src/mesa/drivers/dri/common/dri_bufmgr.c
+++ b/src/mesa/drivers/dri/common/dri_bufmgr.c
@@ -1,499 +1,152 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
+/*
+ * Copyright © 2007 Intel Corporation
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, 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:
- * 
+ * 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 NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
  *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
- *          Keith Whitwell <keithw-at-tungstengraphics-dot-com>
  */
 
-#include <xf86drm.h>
-#include <stdlib.h>
-#include "glthread.h"
-#include "errno.h"
+#include "mtypes.h"
 #include "dri_bufmgr.h"
-#include "string.h"
-#include "imports.h"
-#include "dri_bufpool.h"
 
-_glthread_DECLARE_STATIC_MUTEX(bmMutex);
-
-/*
- * TODO: Introduce fence pools in the same way as 
- * buffer object pools.
+/** @file dri_bufmgr.c
+ *
+ * Convenience functions for buffer management methods.
  */
 
-
-
-typedef struct _DriFenceObject
-{
-   int fd;
-   _glthread_Mutex mutex;
-   int refCount;
-   const char *name;
-   drmFence fence;
-} DriFenceObject;
-
-typedef struct _DriBufferObject
-{
-   DriBufferPool *pool;
-   _glthread_Mutex mutex;
-   int refCount;
-   const char *name;
-   unsigned flags;
-   unsigned hint;
-   unsigned alignment;
-   void *private;
-} DriBufferObject;
-
-
-void
-bmError(int val, const char *file, const char *function, int line)
-{
-   _mesa_printf("Fatal video memory manager error \"%s\".\n"
-                "Check kernel logs or set the LIBGL_DEBUG\n"
-                "environment variable to \"verbose\" for more info.\n"
-                "Detected in file %s, line %d, function %s.\n",
-                strerror(-val), file, line, function);
-#ifndef NDEBUG
-   abort();
-#else
-   abort();
-#endif
-}
-
-DriFenceObject *
-driFenceBuffers(int fd, char *name, unsigned flags)
-{
-   DriFenceObject *fence = (DriFenceObject *) malloc(sizeof(*fence));
-   int ret;
-
-   if (!fence)
-      BM_CKFATAL(-EINVAL);
-
-   _glthread_LOCK_MUTEX(bmMutex);
-   fence->refCount = 1;
-   fence->name = name;
-   fence->fd = fd;
-   _glthread_INIT_MUTEX(fence->mutex);
-   ret = drmFenceBuffers(fd, flags, &fence->fence);
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   if (ret) {
-      free(fence);
-      BM_CKFATAL(ret);
-   }
-   return fence;
-}
-
-
-unsigned 
-driFenceType(DriFenceObject * fence)
-{
-    unsigned ret;
-
-    _glthread_LOCK_MUTEX(bmMutex);
-    ret = fence->fence.flags;
-    _glthread_UNLOCK_MUTEX(bmMutex);
-    
-    return ret;
-}
-
-
-DriFenceObject *
-driFenceReference(DriFenceObject * fence)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   ++fence->refCount;
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   return fence;
-}
-
-void
-driFenceUnReference(DriFenceObject * fence)
-{
-   if (!fence)
-      return;
-
-   _glthread_LOCK_MUTEX(bmMutex);
-   if (--fence->refCount == 0) {
-      drmFenceDestroy(fence->fd, &fence->fence);
-      free(fence);
-   }
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driFenceFinish(DriFenceObject * fence, unsigned type, int lazy)
-{
-   int ret;
-   unsigned flags = (lazy) ? DRM_FENCE_FLAG_WAIT_LAZY : 0;
-
-   _glthread_LOCK_MUTEX(fence->mutex);
-   ret = drmFenceWait(fence->fd, flags, &fence->fence, type);
-   _glthread_UNLOCK_MUTEX(fence->mutex);
-   BM_CKFATAL(ret);
-}
-
-int
-driFenceSignaled(DriFenceObject * fence, unsigned type)
-{
-   int signaled;
-   int ret;
-
-   if (fence == NULL)
-      return GL_TRUE;
-
-   _glthread_LOCK_MUTEX(fence->mutex);
-   ret = drmFenceSignaled(fence->fd, &fence->fence, type, &signaled);
-   _glthread_UNLOCK_MUTEX(fence->mutex);
-   BM_CKFATAL(ret);
-   return signaled;
-}
-
-
-extern drmBO *
-driBOKernel(struct _DriBufferObject *buf)
-{
-   drmBO *ret;
-
-   assert(buf->private != NULL);
-   ret = buf->pool->kernel(buf->pool, buf->private);
-   if (!ret)
-      BM_CKFATAL(-EINVAL);
-
-   return ret;
-}
-
-void
-driBOWaitIdle(struct _DriBufferObject *buf, int lazy)
+dri_bo *
+dri_bo_alloc(dri_bufmgr *bufmgr, const char *name, unsigned long size,
+	     unsigned int alignment, unsigned int location_mask)
 {
-   struct _DriBufferPool *pool;
-   void *priv;
+   assert((location_mask & ~(DRM_BO_FLAG_MEM_LOCAL | DRM_BO_FLAG_MEM_TT |
+			     DRM_BO_FLAG_MEM_VRAM | DRM_BO_FLAG_MEM_PRIV0 |
+			     DRM_BO_FLAG_MEM_PRIV1 | DRM_BO_FLAG_MEM_PRIV2 |
+			     DRM_BO_FLAG_MEM_PRIV3 |
+			     DRM_BO_FLAG_MEM_PRIV4)) == 0);
 
-   _glthread_LOCK_MUTEX(buf->mutex);
-   pool = buf->pool;
-   priv = buf->private;
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   
-   assert(priv != NULL);
-   BM_CKFATAL(buf->pool->waitIdle(pool, priv, lazy));
+   return bufmgr->bo_alloc(bufmgr, name, size, alignment, location_mask);
 }
 
-void *
-driBOMap(struct _DriBufferObject *buf, unsigned flags, unsigned hint)
+dri_bo *
+dri_bo_alloc_static(dri_bufmgr *bufmgr, const char *name, unsigned long offset,
+		    unsigned long size, void *virtual,
+		    unsigned int location_mask)
 {
-   void *virtual;
+   assert((location_mask & ~(DRM_BO_FLAG_MEM_LOCAL | DRM_BO_FLAG_MEM_TT |
+			     DRM_BO_FLAG_MEM_VRAM | DRM_BO_FLAG_MEM_PRIV0 |
+			     DRM_BO_FLAG_MEM_PRIV1 | DRM_BO_FLAG_MEM_PRIV2 |
+			     DRM_BO_FLAG_MEM_PRIV3 |
+			     DRM_BO_FLAG_MEM_PRIV4)) == 0);
 
-   assert(buf->private != NULL);
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   BM_CKFATAL(buf->pool->map(buf->pool, buf->private, flags, hint, &virtual));
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   return virtual;
+   return bufmgr->bo_alloc_static(bufmgr, name, offset, size, virtual,
+				  location_mask);
 }
 
 void
-driBOUnmap(struct _DriBufferObject *buf)
-{
-   assert(buf->private != NULL);
-
-   buf->pool->unmap(buf->pool, buf->private);
-}
-
-unsigned long
-driBOOffset(struct _DriBufferObject *buf)
-{
-   unsigned long ret;
-
-   assert(buf->private != NULL);
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   ret = buf->pool->offset(buf->pool, buf->private);
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   return ret;
-}
-
-unsigned
-driBOFlags(struct _DriBufferObject *buf)
-{
-   unsigned ret;
-
-   assert(buf->private != NULL);
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   ret = buf->pool->flags(buf->pool, buf->private);
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-   return ret;
-}
-
-struct _DriBufferObject *
-driBOReference(struct _DriBufferObject *buf)
+dri_bo_reference(dri_bo *bo)
 {
-   _glthread_LOCK_MUTEX(bmMutex);
-   if (++buf->refCount == 1) {
-      BM_CKFATAL(-EINVAL);
-   }
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   return buf;
+   bo->bufmgr->bo_reference(bo);
 }
 
 void
-driBOUnReference(struct _DriBufferObject *buf)
+dri_bo_unreference(dri_bo *bo)
 {
-   int tmp;
-
-   if (!buf)
+   if (bo == NULL)
       return;
 
-   _glthread_LOCK_MUTEX(bmMutex);
-   tmp = --buf->refCount;
-   _glthread_UNLOCK_MUTEX(bmMutex);
-   if (!tmp) {
-      buf->pool->destroy(buf->pool, buf->private);
-      free(buf);
-   }
+   bo->bufmgr->bo_unreference(bo);
 }
 
-void
-driBOData(struct _DriBufferObject *buf,
-          unsigned size, const void *data, unsigned flags)
-{
-   void *virtual;
-   int newBuffer;
-   struct _DriBufferPool *pool;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   pool = buf->pool;
-   if (!pool->create) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "driBOData called on invalid buffer\n");
-      BM_CKFATAL(-EINVAL);
-   }
-   newBuffer = !buf->private || (pool->size(pool, buf->private) < size) ||
-      pool->map(pool, buf->private, DRM_BO_FLAG_WRITE,
-                DRM_BO_HINT_DONT_BLOCK, &virtual);
-
-   if (newBuffer) {
-      if (buf->private)
-         pool->destroy(pool, buf->private);
-      if (!flags)
-         flags = buf->flags;
-      buf->private = pool->create(pool, size, flags, DRM_BO_HINT_DONT_FENCE, 
-				  buf->alignment);
-      if (!buf->private)
-         BM_CKFATAL(-ENOMEM);
-      BM_CKFATAL(pool->map(pool, buf->private,
-                           DRM_BO_FLAG_WRITE,
-                           DRM_BO_HINT_DONT_BLOCK, &virtual));
-   }
-
-   if (data != NULL)
-      memcpy(virtual, data, size);
-
-   BM_CKFATAL(pool->unmap(pool, buf->private));
-   _glthread_UNLOCK_MUTEX(buf->mutex);
-}
-
-void
-driBOSubData(struct _DriBufferObject *buf,
-             unsigned long offset, unsigned long size, const void *data)
+int
+dri_bo_map(dri_bo *buf, GLboolean write_enable)
 {
-   void *virtual;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (size && data) {
-      BM_CKFATAL(buf->pool->map(buf->pool, buf->private,
-                                DRM_BO_FLAG_WRITE, 0, &virtual));
-      memcpy((unsigned char *) virtual + offset, data, size);
-      BM_CKFATAL(buf->pool->unmap(buf->pool, buf->private));
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
+   return buf->bufmgr->bo_map(buf, write_enable);
 }
 
-void
-driBOGetSubData(struct _DriBufferObject *buf,
-                unsigned long offset, unsigned long size, void *data)
+int
+dri_bo_unmap(dri_bo *buf)
 {
-   void *virtual;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (size && data) {
-      BM_CKFATAL(buf->pool->map(buf->pool, buf->private,
-                                DRM_BO_FLAG_READ, 0, &virtual));
-      memcpy(data, (unsigned char *) virtual + offset, size);
-      BM_CKFATAL(buf->pool->unmap(buf->pool, buf->private));
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
+   return buf->bufmgr->bo_unmap(buf);
 }
 
-void
-driBOSetStatic(struct _DriBufferObject *buf,
-               unsigned long offset,
-               unsigned long size, void *virtual, unsigned flags)
+int
+dri_bo_validate(dri_bo *buf, unsigned int flags)
 {
-   _glthread_LOCK_MUTEX(buf->mutex);
-   if (buf->private != NULL) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "Invalid buffer for setStatic\n");
-      BM_CKFATAL(-EINVAL);
-   }
-   if (buf->pool->setstatic == NULL) {
-      _mesa_error(NULL, GL_INVALID_OPERATION,
-                  "Invalid buffer pool for setStatic\n");
-      BM_CKFATAL(-EINVAL);
-   }
-
-   if (!flags)
-      flags = buf->flags;
-
-   buf->private = buf->pool->setstatic(buf->pool, offset, size,
-                                       virtual, flags);
-   if (!buf->private) {
-      _mesa_error(NULL, GL_OUT_OF_MEMORY,
-                  "Invalid buffer pool for setStatic\n");
-      BM_CKFATAL(-ENOMEM);
-   }
-   _glthread_UNLOCK_MUTEX(buf->mutex);
+   return buf->bufmgr->bo_validate(buf, flags);
 }
 
-
-
-void
-driGenBuffers(struct _DriBufferPool *pool,
-              const char *name,
-              unsigned n,
-              struct _DriBufferObject *buffers[],
-              unsigned alignment, unsigned flags, unsigned hint)
+dri_fence *
+dri_fence_validated(dri_bufmgr *bufmgr, const char *name, GLboolean flushed)
 {
-   struct _DriBufferObject *buf;
-   int i;
-
-   flags = (flags) ? flags : DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_MEM_VRAM |
-      DRM_BO_FLAG_MEM_LOCAL | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE;
-
-
-   for (i = 0; i < n; ++i) {
-      buf = (struct _DriBufferObject *) calloc(1, sizeof(*buf));
-      if (!buf)
-         BM_CKFATAL(-ENOMEM);
-
-      _glthread_INIT_MUTEX(buf->mutex);
-      _glthread_LOCK_MUTEX(buf->mutex);
-      _glthread_LOCK_MUTEX(bmMutex);
-      buf->refCount = 1;
-      _glthread_UNLOCK_MUTEX(bmMutex);
-      buf->flags = flags;
-      buf->hint = hint;
-      buf->name = name;
-      buf->alignment = alignment;
-      buf->pool = pool;
-      _glthread_UNLOCK_MUTEX(buf->mutex);
-      buffers[i] = buf;
-   }
+   return bufmgr->fence_validated(bufmgr, name, flushed);
 }
 
 void
-driDeleteBuffers(unsigned n, struct _DriBufferObject *buffers[])
+dri_fence_wait(dri_fence *fence)
 {
-   int i;
-
-   for (i = 0; i < n; ++i) {
-      driBOUnReference(buffers[i]);
-   }
+   fence->bufmgr->fence_wait(fence);
 }
 
-
 void
-driInitBufMgr(int fd)
+dri_fence_reference(dri_fence *fence)
 {
-   ;
+   fence->bufmgr->fence_reference(fence);
 }
 
-
 void
-driBOCreateList(int target, drmBOList * list)
+dri_fence_unreference(dri_fence *fence)
 {
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOCreateList(target, list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
+   if (fence == NULL)
+      return;
 
-void
-driBOResetList(drmBOList * list)
-{
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOResetList(list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
+   fence->bufmgr->fence_unreference(fence);
 }
 
 void
-driBOAddListItem(drmBOList * list, struct _DriBufferObject *buf,
-                 unsigned flags, unsigned mask)
+dri_bo_subdata(dri_bo *bo, unsigned long offset,
+	       unsigned long size, const void *data)
 {
-   int newItem;
-
-   _glthread_LOCK_MUTEX(buf->mutex);
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmAddValidateItem(list, driBOKernel(buf),
-                                 flags, mask, &newItem));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-
-   /*
-    * Tell userspace pools to validate the buffer. This should be a 
-    * noop if the pool is already validated.
-    * FIXME: We should have a list for this as well.
-    */
-
-   if (buf->pool->validate) {
-      BM_CKFATAL(buf->pool->validate(buf->pool, buf->private));
-   }
+   if (size == 0 || data == NULL)
+      return;
 
-   _glthread_UNLOCK_MUTEX(buf->mutex);
+   dri_bo_map(bo, GL_TRUE);
+   memcpy((unsigned char *)bo->virtual + offset, data, size);
+   dri_bo_unmap(bo);
 }
 
 void
-driBOFence(struct _DriBufferObject *buf, struct _DriFenceObject *fence)
+dri_bo_get_subdata(dri_bo *bo, unsigned long offset,
+		   unsigned long size, void *data)
 {
-   _glthread_LOCK_MUTEX(buf->mutex);
-   BM_CKFATAL(buf->pool->fence(buf->pool, buf->private, fence));
-   _glthread_UNLOCK_MUTEX(buf->mutex);
+   if (size == 0 || data == NULL)
+      return;
 
+   dri_bo_map(bo, GL_FALSE);
+   memcpy(data, (unsigned char *)bo->virtual + offset, size);
+   dri_bo_unmap(bo);
 }
 
 void
-driBOValidateList(int fd, drmBOList * list)
+dri_bufmgr_destroy(dri_bufmgr *bufmgr)
 {
-   _glthread_LOCK_MUTEX(bmMutex);
-   BM_CKFATAL(drmBOValidateList(fd, list));
-   _glthread_UNLOCK_MUTEX(bmMutex);
-}
-
-void
-driPoolTakeDown(struct _DriBufferPool *pool)
-{
-   pool->takeDown(pool);
-
+   bufmgr->destroy(bufmgr);
 }
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr.h b/src/mesa/drivers/dri/common/dri_bufmgr.h
index 01f149ae4ed..3be342926f7 100644
--- a/src/mesa/drivers/dri/common/dri_bufmgr.h
+++ b/src/mesa/drivers/dri/common/dri_bufmgr.h
@@ -1,5 +1,6 @@
 /**************************************************************************
  * 
+ * Copyright � 2007 Intel Corporation
  * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
  * All Rights Reserved.
  * 
@@ -28,72 +29,170 @@
 /*
  * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
  *          Keith Whitwell <keithw-at-tungstengraphics-dot-com>
+ *	    Eric Anholt <[email protected]>
  */
 
 #ifndef _DRI_BUFMGR_H_
 #define _DRI_BUFMGR_H_
 #include <xf86drm.h>
 
+typedef struct _dri_bufmgr dri_bufmgr;
+typedef struct _dri_bo dri_bo;
+typedef struct _dri_fence dri_fence;
 
-struct _DriFenceObject;
-struct _DriBufferObject;
-struct _DriBufferPool;
+struct _dri_bo {
+   /** Size in bytes of the buffer object. */
+   unsigned long size;
+   /**
+    * Card virtual address (offset from the beginning of the aperture) for the
+    * object.  Only valid while validated.
+    */
+   unsigned long offset;
+   /**
+    * Virtual address for accessing the buffer data.  Only valid while mapped.
+    */
+   void *virtual;
+   /** Buffer manager context associated with this buffer object */
+   dri_bufmgr *bufmgr;
+};
 
-extern struct _DriFenceObject *driFenceBuffers(int fd, char *name,
-                                               unsigned flags);
+struct _dri_fence {
+   /**
+    * This is an ORed mask of DRM_BO_FLAG_READ, DRM_BO_FLAG_WRITE, and
+    * DRM_FLAG_EXE indicating the operations associated with this fence.
+    *
+    * It is constant for the life of the fence object.
+    */
+   unsigned int type;
+   /** Buffer manager context associated with this fence */
+   dri_bufmgr *bufmgr;
+};
 
-extern struct _DriFenceObject *driFenceReference(struct _DriFenceObject *fence);
+/**
+ * Context for a buffer manager instance.
+ *
+ * Contains public methods followed by private storage for the buffer manager.
+ */
+struct _dri_bufmgr {
+   /**
+    * Allocate a buffer object.
+    *
+    * Buffer objects are not necessarily initially mapped into CPU virtual
+    * address space or graphics device aperture.  They must be mapped using
+    * bo_map() to be used by the CPU, and validated for use using bo_validate()
+    * to be used from the graphics device.
+    */
+   dri_bo *(*bo_alloc)(dri_bufmgr *bufmgr_ctx, const char *name,
+		       unsigned long size, unsigned int alignment,
+		       unsigned int location_mask);
 
-extern void driFenceUnReference(struct _DriFenceObject *fence);
+   /**
+    * Allocates a buffer object for a static allocation.
+    *
+    * Static allocations are ones such as the front buffer that are offered by
+    * the X Server, which are never evicted and never moved.
+    */
+   dri_bo *(*bo_alloc_static)(dri_bufmgr *bufmgr_ctx, const char *name,
+			      unsigned long offset, unsigned long size,
+			      void *virtual, unsigned int location_mask);
 
-extern void
-driFenceFinish(struct _DriFenceObject *fence, unsigned type, int lazy);
+   /** Takes a reference on a buffer object */
+   void (*bo_reference)(dri_bo *bo);
 
-extern int driFenceSignaled(struct _DriFenceObject *fence, unsigned type);
-extern unsigned driFenceType(struct _DriFenceObject *fence);
+   /**
+    * Releases a reference on a buffer object, freeing the data if
+    * rerefences remain.
+    */
+   void (*bo_unreference)(dri_bo *bo);
 
-/*
- * Return a pointer to the libdrm buffer object this DriBufferObject
- * uses.
- */
+   /**
+    * Maps the buffer into userspace.
+    *
+    * This function will block waiting for any existing fence on the buffer to
+    * clear, first.  The resulting mapping is available at buf->virtual.
+\    */
+   int (*bo_map)(dri_bo *buf, GLboolean write_enable);
+
+   /** Reduces the refcount on the userspace mapping of the buffer object. */
+   int (*bo_unmap)(dri_bo *buf);
+
+   /**
+    * Makes the buffer accessible to the graphics chip.
+    *
+    * The resulting offset of the buffer within the graphics aperture is then
+    * available at buf->offset until the buffer is fenced.
+    *
+    * Flags should consist of the memory types that the buffer may be validated
+    * into and the read/write/exe flags appropriate to the use of the buffer.
+    */
+   int (*bo_validate)(dri_bo *buf, unsigned int flags);
+
+   /**
+    * Associates the current set of validated buffers with a fence.
+    *
+    * Once fenced, the buffer manager will allow the validated buffers to be
+    * evicted when the graphics device's execution has passed the fence
+    * command.
+    *
+    * The fence object will have flags for the sum of the read/write/exe flags
+    * of the validated buffers associated with it.
+    */
+   dri_fence * (*fence_validated)(dri_bufmgr *bufmgr, const char *name,
+				  GLboolean flushed);
+
+   /** Takes a reference on a fence object */
+   void (*fence_reference)(dri_fence *fence);
+
+   /**
+    * Releases a reference on a fence object, freeing the data if
+    * rerefences remain.
+    */
+   void (*fence_unreference)(dri_fence *fence);
+
+   /**
+    * Blocks until the given fence is signaled.
+    */
+   void (*fence_wait)(dri_fence *fence);
+
+   /**
+    * Tears down the buffer manager instance.
+    */
+   void (*destroy)(dri_bufmgr *bufmgr);
+};
+
+dri_bo *dri_bo_alloc(dri_bufmgr *bufmgr, const char *name, unsigned long size,
+		     unsigned int alignment, unsigned int location_mask);
+dri_bo *dri_bo_alloc_static(dri_bufmgr *bufmgr, const char *name,
+			    unsigned long offset, unsigned long size,
+			    void *virtual, unsigned int location_mask);
+void dri_bo_reference(dri_bo *bo);
+void dri_bo_unreference(dri_bo *bo);
+int dri_bo_map(dri_bo *buf, GLboolean write_enable);
+int dri_bo_unmap(dri_bo *buf);
+int dri_bo_validate(dri_bo *buf, unsigned int flags);
+dri_fence *dri_fence_validated(dri_bufmgr *bufmgr, const char *name,
+			       GLboolean flushed);
+void dri_fence_wait(dri_fence *fence);
+void dri_fence_reference(dri_fence *fence);
+void dri_fence_unreference(dri_fence *fence);
+
+void dri_bo_subdata(dri_bo *bo, unsigned long offset,
+		    unsigned long size, const void *data);
+void dri_bo_get_subdata(dri_bo *bo, unsigned long offset,
+			unsigned long size, void *data);
+
+dri_bufmgr *dri_bufmgr_ttm_init(int fd, unsigned int fence_type,
+				unsigned int fence_type_flush);
 
-extern drmBO *driBOKernel(struct _DriBufferObject *buf);
-extern void *driBOMap(struct _DriBufferObject *buf, unsigned flags,
-                      unsigned hint);
-extern void driBOUnmap(struct _DriBufferObject *buf);
-extern unsigned long driBOOffset(struct _DriBufferObject *buf);
-extern unsigned driBOFlags(struct _DriBufferObject *buf);
-extern struct _DriBufferObject *driBOReference(struct _DriBufferObject *buf);
-extern void driBOUnReference(struct _DriBufferObject *buf);
-extern void driBOData(struct _DriBufferObject *r_buf,
-                      unsigned size, const void *data, unsigned flags);
-extern void driBOSubData(struct _DriBufferObject *buf,
-                         unsigned long offset, unsigned long size,
-                         const void *data);
-extern void driBOGetSubData(struct _DriBufferObject *buf,
-                            unsigned long offset, unsigned long size,
-                            void *data);
-extern void driGenBuffers(struct _DriBufferPool *pool,
-                          const char *name,
-                          unsigned n,
-                          struct _DriBufferObject *buffers[],
-                          unsigned alignment, unsigned flags, unsigned hint);
-extern void driDeleteBuffers(unsigned n, struct _DriBufferObject *buffers[]);
-extern void driInitBufMgr(int fd);
-extern void driBOCreateList(int target, drmBOList * list);
-extern void driBOResetList(drmBOList * list);
-extern void driBOAddListItem(drmBOList * list, struct _DriBufferObject *buf,
-                             unsigned flags, unsigned mask);
-extern void driBOValidateList(int fd, drmBOList * list);
-
-extern void driBOFence(struct _DriBufferObject *buf,
-                       struct _DriFenceObject *fence);
-
-extern void driPoolTakeDown(struct _DriBufferPool *pool);
-extern void driBOSetStatic(struct _DriBufferObject *buf,
-                           unsigned long offset,
-                           unsigned long size, void *virtual, unsigned flags);
-extern void driBOWaitIdle(struct _DriBufferObject *buf, int lazy);
-extern void driPoolTakeDown(struct _DriBufferPool *pool);
+void dri_bufmgr_fake_contended_lock_take(dri_bufmgr *bufmgr);
+dri_bufmgr *dri_bufmgr_fake_init(unsigned long low_offset, void *low_virtual,
+				 unsigned long size,
+				 unsigned int (*fence_emit)(void *private),
+				 int (*fence_wait)(void *private,
+						   unsigned int cookie),
+				 void *driver_priv);
+void dri_bufmgr_destroy(dri_bufmgr *bufmgr);
+dri_bo *dri_ttm_bo_create_from_handle(dri_bufmgr *bufmgr, const char *name,
+				      unsigned int handle);
 
 #endif
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr_fake.c b/src/mesa/drivers/dri/common/dri_bufmgr_fake.c
new file mode 100644
index 00000000000..e0d23a36477
--- /dev/null
+++ b/src/mesa/drivers/dri/common/dri_bufmgr_fake.c
@@ -0,0 +1,882 @@
+/**************************************************************************
+ * 
+ * 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.
+ * 
+ **************************************************************************/
+
+/* Originally a fake version of the buffer manager so that we can
+ * prototype the changes in a driver fairly quickly, has been fleshed
+ * out to a fully functional interim solution.
+ *
+ * Basically wraps the old style memory management in the new
+ * programming interface, but is more expressive and avoids many of
+ * the bugs in the old texture manager.
+ */
+#include "mtypes.h"
+#include "dri_bufmgr.h"
+#include "drm.h"
+
+#include "simple_list.h"
+#include "mm.h"
+#include "imports.h"
+
+#if 0
+#define DBG(...) _mesa_printf(__VA_ARGS__)
+#else
+#define DBG(...)
+#endif
+
+/* Internal flags:
+ */
+#define BM_NO_BACKING_STORE			0x00000001
+#define BM_NO_FENCE_SUBDATA			0x00000002
+#define BM_PINNED				0x00000004
+
+/* Wrapper around mm.c's mem_block, which understands that you must
+ * wait for fences to expire before memory can be freed.  This is
+ * specific to our use of memcpy for uploads - an upload that was
+ * processed through the command queue wouldn't need to care about
+ * fences.
+ */
+struct block {
+   struct block *next, *prev;
+   struct mem_block *mem;	/* BM_MEM_AGP */
+
+   unsigned on_hardware:1;
+   unsigned fenced:1;
+
+   unsigned fence;		/* BM_MEM_AGP, Split to read_fence, write_fence */
+
+   dri_bo *bo;
+   void *virtual;
+};
+
+typedef struct _bufmgr_fake {
+   dri_bufmgr bufmgr;
+
+   _glthread_Mutex mutex;	/**< for thread safety */
+
+   unsigned long low_offset;
+   unsigned long size;
+   void *virtual;
+
+   struct mem_block *heap;
+   struct block lru;		/* only allocated, non-fence-pending blocks here */
+
+   unsigned buf_nr;		/* for generating ids */
+
+   struct block on_hardware;	/* after bmValidateBuffers */
+   struct block fenced;		/* after bmFenceBuffers (mi_flush, emit irq, write dword) */
+                                /* then to bufmgr->lru or free() */
+
+   unsigned int last_fence;
+
+   unsigned fail:1;
+   unsigned need_fence:1;
+   GLboolean thrashing;
+
+   /**
+    * Driver callback to emit a fence, returning the cookie.
+    *
+    * Currently, this also requires that a write flush be emitted before
+    * emitting the fence, but this should change.
+    */
+   unsigned int (*fence_emit)(void *private);
+   /** Driver callback to wait for a fence cookie to have passed. */
+   int (*fence_wait)(void *private, unsigned int fence_cookie);
+   /** Driver-supplied argument to driver callbacks */
+   void *driver_priv;
+} dri_bufmgr_fake;
+
+typedef struct _dri_bo_fake {
+   dri_bo bo;
+
+   unsigned id;			/* debug only */
+   const char *name;
+
+   unsigned dirty:1;
+   unsigned int refcount;
+   /* Flags may consist of any of the DRM_BO flags, plus
+    * DRM_BO_NO_BACKING_STORE and BM_NO_FENCE_SUBDATA, which are the first two
+    * driver private flags.
+    */
+   unsigned int flags;
+   unsigned int alignment;
+   GLboolean is_static;
+
+   struct block *block;
+   void *backing_store;
+   void (*invalidate_cb)(dri_bufmgr *bufmgr, void * );
+   void *invalidate_ptr;
+} dri_bo_fake;
+
+typedef struct _dri_fence_fake {
+   dri_fence fence;
+
+   const char *name;
+   unsigned int refcount;
+   unsigned int fence_cookie;
+   GLboolean flushed;
+} dri_fence_fake;
+
+static int clear_fenced(dri_bufmgr_fake *bufmgr_fake,
+			unsigned int fence_cookie);
+
+#define MAXFENCE 0x7fffffff
+
+static GLboolean FENCE_LTE( unsigned a, unsigned b )
+{
+   if (a == b)
+      return GL_TRUE;
+
+   if (a < b && b - a < (1<<24))
+      return GL_TRUE;
+
+   if (a > b && MAXFENCE - a + b < (1<<24))
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+static unsigned int
+_fence_emit_internal(dri_bufmgr_fake *bufmgr_fake)
+{
+   bufmgr_fake->last_fence = bufmgr_fake->fence_emit(bufmgr_fake->driver_priv);
+   return bufmgr_fake->last_fence;
+}
+
+static void
+_fence_wait_internal(dri_bufmgr_fake *bufmgr_fake, unsigned int cookie)
+{
+   int ret;
+
+   ret = bufmgr_fake->fence_wait(bufmgr_fake->driver_priv, cookie);
+   if (ret != 0) {
+      _mesa_printf("%s:%d: Error %d waiting for fence.\n",
+		   __FILE__, __LINE__);
+      abort();
+   }
+   clear_fenced(bufmgr_fake, cookie);
+}
+
+static GLboolean
+_fence_test(dri_bufmgr_fake *bufmgr_fake, unsigned fence)
+{
+   /* Slight problem with wrap-around:
+    */
+   return fence == 0 || FENCE_LTE(fence, bufmgr_fake->last_fence);
+}
+
+/**
+ * Allocate a memory manager block for the buffer.
+ */
+static GLboolean
+alloc_block(dri_bo *bo)
+{
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+   dri_bufmgr_fake *bufmgr_fake= (dri_bufmgr_fake *)bo->bufmgr;
+   struct block *block = (struct block *)calloc(sizeof *block, 1);
+   unsigned int align_log2 = ffs(bo_fake->alignment);
+   GLuint sz;
+
+   if (!block)
+      return GL_FALSE;
+
+   sz = (bo->size + bo_fake->alignment - 1) & ~(bo_fake->alignment - 1);
+
+   block->mem = mmAllocMem(bufmgr_fake->heap, sz, align_log2, 0);
+   if (!block->mem) {
+      free(block);
+      return GL_FALSE;
+   }
+
+   make_empty_list(block);
+
+   /* Insert at head or at tail???   
+    */
+   insert_at_tail(&bufmgr_fake->lru, block);
+
+   block->virtual = bufmgr_fake->virtual +
+      block->mem->ofs - bufmgr_fake->low_offset;
+   block->bo = bo;
+
+   bo_fake->block = block;
+
+   return GL_TRUE;
+}
+
+/* Release the card storage associated with buf:
+ */
+static void free_block(dri_bufmgr_fake *bufmgr_fake, struct block *block)
+{
+   DBG("free block %p\n", block);
+
+   if (!block)
+      return;
+
+   if (block->on_hardware) {
+      block->bo = NULL;
+   }
+   else if (block->fenced) {
+      block->bo = NULL;
+   }
+   else {
+      DBG("    - free immediately\n");
+      remove_from_list(block);
+
+      mmFreeMem(block->mem);
+      free(block);
+   }
+}
+
+static void
+alloc_backing_store(dri_bo *bo)
+{
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+   assert(!bo_fake->backing_store);
+   assert(!(bo_fake->flags & (BM_PINNED|BM_NO_BACKING_STORE)));
+
+   bo_fake->backing_store = ALIGN_MALLOC(bo->size, 64);
+}
+
+static void
+free_backing_store(dri_bo *bo)
+{
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+   assert(!(bo_fake->flags & (BM_PINNED|BM_NO_BACKING_STORE)));
+
+   if (bo_fake->backing_store) {
+      ALIGN_FREE(bo_fake->backing_store);
+      bo_fake->backing_store = NULL;
+   }
+}
+
+static void
+set_dirty(dri_bo *bo)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   if (bo_fake->flags & BM_NO_BACKING_STORE)
+      bo_fake->invalidate_cb(&bufmgr_fake->bufmgr, bo_fake->invalidate_ptr);
+
+   assert(!(bo_fake->flags & BM_PINNED));
+
+   DBG("set_dirty - buf %d\n", bo_fake->id);
+   bo_fake->dirty = 1;
+}
+
+static GLboolean
+evict_lru(dri_bufmgr_fake *bufmgr_fake, GLuint max_fence)
+{
+   struct block *block, *tmp;
+
+   DBG("%s\n", __FUNCTION__);
+
+   foreach_s(block, tmp, &bufmgr_fake->lru) {
+      dri_bo_fake *bo_fake = (dri_bo_fake *)block->bo;
+
+      if (bo_fake != NULL && (bo_fake->flags & BM_NO_FENCE_SUBDATA))
+	 continue;
+
+      if (block->fence && max_fence && !FENCE_LTE(block->fence, max_fence))
+	 return 0;
+
+      set_dirty(&bo_fake->bo);
+      bo_fake->block = NULL;
+
+      free_block(bufmgr_fake, block);
+      return GL_TRUE;
+   }
+
+   return GL_FALSE;
+}
+
+#define foreach_s_rev(ptr, t, list)   \
+        for(ptr=(list)->prev,t=(ptr)->prev; list != ptr; ptr=t, t=(t)->prev)
+
+static GLboolean
+evict_mru(dri_bufmgr_fake *bufmgr_fake)
+{
+   struct block *block, *tmp;
+
+   DBG("%s\n", __FUNCTION__);
+
+   foreach_s_rev(block, tmp, &bufmgr_fake->lru) {
+      dri_bo_fake *bo_fake = (dri_bo_fake *)block->bo;
+
+      if (bo_fake && (bo_fake->flags & BM_NO_FENCE_SUBDATA))
+	 continue;
+
+      set_dirty(&bo_fake->bo);
+      bo_fake->block = NULL;
+
+      free_block(bufmgr_fake, block);
+      return GL_TRUE;
+   }
+
+   return GL_FALSE;
+}
+
+/**
+ * Removes all objects from the fenced list older than the given fence.
+ */
+static int clear_fenced(dri_bufmgr_fake *bufmgr_fake,
+			unsigned int fence_cookie)
+{
+   struct block *block, *tmp;
+   int ret = 0;
+
+   foreach_s(block, tmp, &bufmgr_fake->fenced) {
+      assert(block->fenced);
+
+      if (_fence_test(bufmgr_fake, block->fence)) {
+
+	 block->fenced = 0;
+
+	 if (!block->bo) {
+	    DBG("delayed free: offset %x sz %x\n",
+		block->mem->ofs, block->mem->size);
+	    remove_from_list(block);
+	    mmFreeMem(block->mem);
+	    free(block);
+	 }
+	 else {
+	    DBG("return to lru: offset %x sz %x\n",
+		block->mem->ofs, block->mem->size);
+	    move_to_tail(&bufmgr_fake->lru, block);
+	 }
+
+	 ret = 1;
+      }
+      else {
+	 /* Blocks are ordered by fence, so if one fails, all from
+	  * here will fail also:
+	  */
+	 break;
+      }
+   }
+
+   DBG("%s: %d\n", __FUNCTION__, ret);
+   return ret;
+}
+
+static void fence_blocks(dri_bufmgr_fake *bufmgr_fake, unsigned fence)
+{
+   struct block *block, *tmp;
+
+   foreach_s (block, tmp, &bufmgr_fake->on_hardware) {
+      DBG("Fence block %p (sz 0x%x buf %p) with fence %d\n", block,
+	  block->mem->size, block->bo, fence);
+      block->fence = fence;
+
+      block->on_hardware = 0;
+      block->fenced = 1;
+
+      /* Move to tail of pending list here
+       */
+      move_to_tail(&bufmgr_fake->fenced, block);
+   }
+
+   assert(is_empty_list(&bufmgr_fake->on_hardware));
+}
+
+static GLboolean evict_and_alloc_block(dri_bo *bo)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   assert(bo_fake->block == NULL);
+
+   /* Search for already free memory:
+    */
+   if (alloc_block(bo))
+      return GL_TRUE;
+
+   /* If we're not thrashing, allow lru eviction to dig deeper into
+    * recently used textures.  We'll probably be thrashing soon:
+    */
+   if (!bufmgr_fake->thrashing) {
+      while (evict_lru(bufmgr_fake, 0))
+	 if (alloc_block(bo))
+	    return GL_TRUE;
+   }
+
+   /* Keep thrashing counter alive?
+    */
+   if (bufmgr_fake->thrashing)
+      bufmgr_fake->thrashing = 20;
+
+   /* Wait on any already pending fences - here we are waiting for any
+    * freed memory that has been submitted to hardware and fenced to
+    * become available:
+    */
+   while (!is_empty_list(&bufmgr_fake->fenced)) {
+      GLuint fence = bufmgr_fake->fenced.next->fence;
+      _fence_wait_internal(bufmgr_fake, fence);
+
+      if (alloc_block(bo))
+	 return GL_TRUE;
+   }
+
+   if (!is_empty_list(&bufmgr_fake->on_hardware)) {
+      while (!is_empty_list(&bufmgr_fake->fenced)) {
+	 GLuint fence = bufmgr_fake->fenced.next->fence;
+	 _fence_wait_internal(bufmgr_fake, fence);
+      }
+
+      if (!bufmgr_fake->thrashing) {
+	 DBG("thrashing\n");
+      }
+      bufmgr_fake->thrashing = 20;
+
+      if (alloc_block(bo))
+	 return GL_TRUE;
+   }
+
+   while (evict_mru(bufmgr_fake))
+      if (alloc_block(bo))
+	 return GL_TRUE;
+
+   DBG("%s 0x%x bytes failed\n", __FUNCTION__, bo->size);
+
+   assert(is_empty_list(&bufmgr_fake->on_hardware));
+   assert(is_empty_list(&bufmgr_fake->fenced));
+
+   return GL_FALSE;
+}
+
+/***********************************************************************
+ * Public functions
+ */
+
+/**
+ * Wait for hardware idle by emitting a fence and waiting for it.
+ */
+static void
+dri_bufmgr_fake_wait_idle(dri_bufmgr_fake *bufmgr_fake)
+{
+   unsigned int cookie;
+
+   cookie = bufmgr_fake->fence_emit(bufmgr_fake->driver_priv);
+   _fence_wait_internal(bufmgr_fake, cookie);
+}
+
+/* Specifically ignore texture memory sharing.
+ *  -- just evict everything
+ *  -- and wait for idle
+ */
+void
+dri_bufmgr_fake_contended_lock_take(dri_bufmgr *bufmgr)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bufmgr;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   {
+      struct block *block, *tmp;
+
+      bufmgr_fake->need_fence = 1;
+      bufmgr_fake->fail = 0;
+
+      /* Wait for hardware idle.  We don't know where acceleration has been
+       * happening, so we'll need to wait anyway before letting anything get
+       * put on the card again.
+       */
+      dri_bufmgr_fake_wait_idle(bufmgr_fake);
+
+      /* Check that we hadn't released the lock without having fenced the last
+       * set of buffers.
+       */
+      assert(is_empty_list(&bufmgr_fake->fenced));
+      assert(is_empty_list(&bufmgr_fake->on_hardware));
+
+      foreach_s(block, tmp, &bufmgr_fake->lru) {
+	 assert(_fence_test(bufmgr_fake, block->fence));
+	 set_dirty(block->bo);
+      }
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+static dri_bo *
+dri_fake_bo_alloc(dri_bufmgr *bufmgr, const char *name,
+		  unsigned long size, unsigned int alignment,
+		  unsigned int location_mask)
+{
+   dri_bufmgr_fake *bufmgr_fake;
+   dri_bo_fake *bo_fake;
+
+   bufmgr_fake = (dri_bufmgr_fake *)bufmgr;
+
+   bo_fake = calloc(1, sizeof(*bo_fake));
+   if (!bo_fake)
+      return NULL;
+
+   bo_fake->bo.size = size;
+   bo_fake->bo.offset = -1;
+   bo_fake->bo.virtual = NULL;
+   bo_fake->bo.bufmgr = bufmgr;
+   bo_fake->refcount = 1;
+
+   /* Alignment must be a power of two */
+   assert((alignment & (alignment - 1)) == 0);
+   if (alignment == 0)
+      alignment = 1;
+   bo_fake->alignment = alignment;
+   bo_fake->id = ++bufmgr_fake->buf_nr;
+   bo_fake->name = name;
+   bo_fake->flags = 0;
+   bo_fake->is_static = GL_FALSE;
+
+   DBG("drm_bo_alloc: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
+   return &bo_fake->bo;
+}
+
+static dri_bo *
+dri_fake_bo_alloc_static(dri_bufmgr *bufmgr, const char *name,
+			 unsigned long offset, unsigned long size,
+			 void *virtual, unsigned int location_mask)
+{
+   dri_bufmgr_fake *bufmgr_fake;
+   dri_bo_fake *bo_fake;
+
+   bufmgr_fake = (dri_bufmgr_fake *)bufmgr;
+
+   bo_fake = calloc(1, sizeof(*bo_fake));
+   if (!bo_fake)
+      return NULL;
+
+   bo_fake->bo.size = size;
+   bo_fake->bo.offset = offset;
+   bo_fake->bo.virtual = virtual;
+   bo_fake->bo.bufmgr = bufmgr;
+   bo_fake->refcount = 1;
+   bo_fake->id = ++bufmgr_fake->buf_nr;
+   bo_fake->name = name;
+   bo_fake->flags = BM_PINNED | DRM_BO_FLAG_NO_MOVE;
+   bo_fake->is_static = GL_TRUE;
+
+   DBG("drm_bo_alloc_static: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
+   return &bo_fake->bo;
+}
+
+static void
+dri_fake_bo_reference(dri_bo *bo)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   bo_fake->refcount++;
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+static void
+dri_fake_bo_unreference(dri_bo *bo)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   if (!bo)
+      return;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   if (--bo_fake->refcount == 0) {
+      /* No remaining references, so free it */
+      if (bo_fake->block)
+	 free_block(bufmgr_fake, bo_fake->block);
+      free_backing_store(bo);
+      _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+      free(bo);
+      return;
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+/**
+ * Map a buffer into bo->virtual, allocating either card memory space (If
+ * BM_NO_BACKING_STORE or BM_PINNED) or backing store, as necessary.
+ */
+static int
+dri_fake_bo_map(dri_bo *bo, GLboolean write_enable)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   /* Static buffers are always mapped. */
+   if (bo_fake->is_static)
+      return 0;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   {
+      DBG("drm_bo_map: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+	  bo_fake->bo.size / 1024);
+
+      if (bo->virtual != NULL) {
+	 _mesa_printf("%s: already mapped\n", __FUNCTION__);
+	 abort();
+      }
+      else if (bo_fake->flags & (BM_NO_BACKING_STORE|BM_PINNED)) {
+
+	 if (!bo_fake->block && !evict_and_alloc_block(bo)) {
+	    DBG("%s: alloc failed\n", __FUNCTION__);
+	    bufmgr_fake->fail = 1;
+	    _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+	    return 1;
+	 }
+	 else {
+	    assert(bo_fake->block);
+	    bo_fake->dirty = 0;
+
+	    if (!(bo_fake->flags & BM_NO_FENCE_SUBDATA))
+	       dri_bufmgr_fake_wait_idle(bufmgr_fake);
+
+	    bo->virtual = bo_fake->block->virtual;
+	 }
+      }
+      else {
+	 if (write_enable)
+	    set_dirty(bo);
+
+	 if (bo_fake->backing_store == 0)
+	    alloc_backing_store(bo);
+
+	 bo->virtual = bo_fake->backing_store;
+      }
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+   return 0;
+}
+
+static int
+dri_fake_bo_unmap(dri_bo *bo)
+{
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   /* Static buffers are always mapped. */
+   if (bo_fake->is_static)
+      return 0;
+
+   if (bo == NULL)
+      return 0;
+
+   DBG("drm_bo_unmap: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+
+   bo->virtual = NULL;
+
+   return 0;
+}
+
+static int
+dri_fake_bo_validate(dri_bo *bo, unsigned int flags)
+{
+   dri_bufmgr_fake *bufmgr_fake;
+   dri_bo_fake *bo_fake = (dri_bo_fake *)bo;
+
+   /* XXX: Sanity-check whether we've already validated this one under
+    * different flags.  See drmAddValidateItem().
+    */
+
+   DBG("drm_bo_validate: (buf %d: %s, %d kb)\n", bo_fake->id, bo_fake->name,
+       bo_fake->bo.size / 1024);
+   bufmgr_fake = (dri_bufmgr_fake *)bo->bufmgr;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   {
+      if (bo_fake->is_static) {
+	 /* Add it to the needs-fence list */
+	 bufmgr_fake->need_fence = 1;
+	 _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+	 return 0;
+      }
+
+      /* Allocate the card memory */
+      if (!bo_fake->block && !evict_and_alloc_block(bo)) {
+	 bufmgr_fake->fail = 1;
+	 _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+	 DBG("Failed to validate buf %d:%s\n", bo_fake->id, bo_fake->name);
+	 return -1;
+      }
+
+      assert(bo_fake->block);
+      assert(bo_fake->block->bo == &bo_fake->bo);
+
+      bo->offset = bo_fake->block->mem->ofs;
+
+      /* Upload the buffer contents if necessary */
+      if (bo_fake->dirty) {
+	 DBG("Upload dirty buf %d:%s, sz %d offset 0x%x\n", bo_fake->id,
+	     bo_fake->name, bo->size, bo_fake->block->mem->ofs);
+
+	 assert(!(bo_fake->flags &
+		  (BM_NO_BACKING_STORE|BM_PINNED)));
+
+	 /* Actually, should be able to just wait for a fence on the memory,
+	  * which we would be tracking when we free it.  Waiting for idle is
+	  * a sufficiently large hammer for now.
+	  */
+	 dri_bufmgr_fake_wait_idle(bufmgr_fake);
+
+	 memcpy(bo_fake->block->virtual, bo_fake->backing_store, bo->size);
+	 bo_fake->dirty = 0;
+      }
+
+      bo_fake->block->on_hardware = 1;
+      move_to_tail(&bufmgr_fake->on_hardware, bo_fake->block);
+
+      bufmgr_fake->need_fence = 1;
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+
+   return 0;
+}
+
+static dri_fence *
+dri_fake_fence_validated(dri_bufmgr *bufmgr, const char *name,
+			 GLboolean flushed)
+{
+   dri_fence_fake *fence_fake;
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bufmgr;
+   unsigned int cookie;
+
+   fence_fake = malloc(sizeof(*fence_fake));
+   if (!fence_fake)
+      return NULL;
+
+   fence_fake->refcount = 1;
+   fence_fake->name = name;
+   fence_fake->flushed = flushed;
+   fence_fake->fence.bufmgr = bufmgr;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   cookie = _fence_emit_internal(bufmgr_fake);
+   fence_fake->fence_cookie = cookie;
+   fence_blocks(bufmgr_fake, cookie);
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+
+   DBG("drm_fence_validated: 0x%08x cookie\n", fence_fake->fence_cookie);
+
+   return &fence_fake->fence;
+}
+
+static void
+dri_fake_fence_reference(dri_fence *fence)
+{
+   dri_fence_fake *fence_fake = (dri_fence_fake *)fence;
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)fence->bufmgr;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   ++fence_fake->refcount;
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+static void
+dri_fake_fence_unreference(dri_fence *fence)
+{
+   dri_fence_fake *fence_fake = (dri_fence_fake *)fence;
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)fence->bufmgr;
+
+   if (!fence)
+      return;
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   if (--fence_fake->refcount == 0) {
+      _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+      free(fence);
+      return;
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+static void
+dri_fake_fence_wait(dri_fence *fence)
+{
+   dri_fence_fake *fence_fake = (dri_fence_fake *)fence;
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)fence->bufmgr;
+
+   DBG("drm_fence_wait: 0x%08x cookie\n", fence_fake->fence_cookie);
+
+   _glthread_LOCK_MUTEX(bufmgr_fake->mutex);
+   _fence_wait_internal(bufmgr_fake, fence_fake->fence_cookie);
+   _glthread_UNLOCK_MUTEX(bufmgr_fake->mutex);
+}
+
+static void
+dri_fake_destroy(dri_bufmgr *bufmgr)
+{
+   dri_bufmgr_fake *bufmgr_fake = (dri_bufmgr_fake *)bufmgr;
+
+   _glthread_DESTROY_MUTEX(bufmgr_fake->mutex);
+   mmDestroy(bufmgr_fake->heap);
+   free(bufmgr);
+}
+
+dri_bufmgr *
+dri_bufmgr_fake_init(unsigned long low_offset, void *low_virtual,
+		     unsigned long size,
+		     unsigned int (*fence_emit)(void *private),
+		     int (*fence_wait)(void *private, unsigned int cookie),
+		     void *driver_priv)
+{
+   dri_bufmgr_fake *bufmgr_fake;
+
+   bufmgr_fake = calloc(1, sizeof(*bufmgr_fake));
+
+   /* Initialize allocator */
+   make_empty_list(&bufmgr_fake->fenced);
+   make_empty_list(&bufmgr_fake->on_hardware);
+   make_empty_list(&bufmgr_fake->lru);
+
+   bufmgr_fake->low_offset = low_offset;
+   bufmgr_fake->virtual = low_virtual;
+   bufmgr_fake->size = size;
+   bufmgr_fake->heap = mmInit(low_offset, size);
+
+   _glthread_INIT_MUTEX(bufmgr_fake->mutex);
+
+   /* Hook in methods */
+   bufmgr_fake->bufmgr.bo_alloc = dri_fake_bo_alloc;
+   bufmgr_fake->bufmgr.bo_alloc_static = dri_fake_bo_alloc_static;
+   bufmgr_fake->bufmgr.bo_reference = dri_fake_bo_reference;
+   bufmgr_fake->bufmgr.bo_unreference = dri_fake_bo_unreference;
+   bufmgr_fake->bufmgr.bo_map = dri_fake_bo_map;
+   bufmgr_fake->bufmgr.bo_unmap = dri_fake_bo_unmap;
+   bufmgr_fake->bufmgr.bo_validate = dri_fake_bo_validate;
+   bufmgr_fake->bufmgr.fence_validated = dri_fake_fence_validated;
+   bufmgr_fake->bufmgr.fence_wait = dri_fake_fence_wait;
+   bufmgr_fake->bufmgr.fence_reference = dri_fake_fence_reference;
+   bufmgr_fake->bufmgr.fence_unreference = dri_fake_fence_unreference;
+   bufmgr_fake->bufmgr.destroy = dri_fake_destroy;
+
+   bufmgr_fake->fence_emit = fence_emit;
+   bufmgr_fake->fence_wait = fence_wait;
+   bufmgr_fake->driver_priv = driver_priv;
+
+   return &bufmgr_fake->bufmgr;
+}
diff --git a/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c b/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c
new file mode 100644
index 00000000000..235398eb872
--- /dev/null
+++ b/src/mesa/drivers/dri/common/dri_bufmgr_ttm.c
@@ -0,0 +1,469 @@
+/**************************************************************************
+ * 
+ * Copyright � 2007 Intel Corporation
+ * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
+ * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
+ * USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * 
+ **************************************************************************/
+/*
+ * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
+ *          Keith Whitwell <keithw-at-tungstengraphics-dot-com>
+ *	    Eric Anholt <[email protected]>
+ */
+
+#include <xf86drm.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include "glthread.h"
+#include "errno.h"
+#include "mtypes.h"
+#include "dri_bufmgr.h"
+#include "string.h"
+#include "imports.h"
+
+#define BUFMGR_DEBUG 0
+
+typedef struct _dri_bufmgr_ttm {
+   dri_bufmgr bufmgr;
+
+   int fd;
+   _glthread_Mutex mutex;
+   unsigned int fence_type;
+   unsigned int fence_type_flush;
+} dri_bufmgr_ttm;
+
+typedef struct _dri_bo_ttm {
+   dri_bo bo;
+
+   int refcount;		/* Protected by bufmgr->mutex */
+   drmBO drm_bo;
+   const char *name;
+   /**
+    * Note whether we are the owner of the buffer, to determine if we must
+    * drmBODestroy or drmBOUnreference to unreference the buffer.
+    */
+   GLboolean owner;
+} dri_bo_ttm;
+
+typedef struct _dri_fence_ttm
+{
+   dri_fence fence;
+
+   int refcount;		/* Protected by bufmgr->mutex */
+   const char *name;
+   drmFence drm_fence;
+} dri_fence_ttm;
+
+#if 0
+int
+driFenceSignaled(DriFenceObject * fence, unsigned type)
+{
+   int signaled;
+   int ret;
+
+   if (fence == NULL)
+      return GL_TRUE;
+
+   _glthread_LOCK_MUTEX(fence->mutex);
+   ret = drmFenceSignaled(bufmgr_ttm->fd, &fence->fence, type, &signaled);
+   _glthread_UNLOCK_MUTEX(fence->mutex);
+   BM_CKFATAL(ret);
+   return signaled;
+}
+#endif
+
+static dri_bo *
+dri_ttm_alloc(dri_bufmgr *bufmgr, const char *name,
+	      unsigned long size, unsigned int alignment,
+	      unsigned int location_mask)
+{
+   dri_bufmgr_ttm *ttm_bufmgr;
+   dri_bo_ttm *ttm_buf;
+   unsigned int pageSize = getpagesize();
+   int ret;
+   unsigned int flags, hint;
+
+   ttm_bufmgr = (dri_bufmgr_ttm *)bufmgr;
+
+   ttm_buf = malloc(sizeof(*ttm_buf));
+   if (!ttm_buf)
+      return NULL;
+
+   /* The mask argument doesn't do anything for us that we want other than
+    * determine which pool (TTM or local) the buffer is allocated into, so just
+    * pass all of the allocation class flags.
+    */
+   flags = location_mask | DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE |
+      DRM_BO_FLAG_EXE;
+   /* No hints we want to use. */
+   hint = 0;
+
+   ret = drmBOCreate(ttm_bufmgr->fd, 0, size, alignment / pageSize,
+		     NULL, drm_bo_type_dc,
+                     flags, hint, &ttm_buf->drm_bo);
+   if (ret != 0) {
+      free(ttm_buf);
+      return NULL;
+   }
+   ttm_buf->bo.size = ttm_buf->drm_bo.size;
+   ttm_buf->bo.offset = ttm_buf->drm_bo.offset;
+   ttm_buf->bo.virtual = NULL;
+   ttm_buf->bo.bufmgr = bufmgr;
+   ttm_buf->name = name;
+   ttm_buf->refcount = 1;
+   ttm_buf->owner = GL_TRUE;
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "bo_create: %p (%s)\n", &ttm_buf->bo, ttm_buf->name);
+#endif
+
+   return &ttm_buf->bo;
+}
+
+/* Our TTM backend doesn't allow creation of static buffers, as that requires
+ * privelege for the non-fake case, and the lock in the fake case where we were
+ * working around the X Server not creating buffers and passing handles to us.
+ */
+static dri_bo *
+dri_ttm_alloc_static(dri_bufmgr *bufmgr, const char *name,
+		     unsigned long offset, unsigned long size, void *virtual,
+		     unsigned int location_mask)
+{
+   return NULL;
+}
+
+/** Returns a dri_bo wrapping the given buffer object handle.
+ *
+ * This can be used when one application needs to pass a buffer object
+ * to another.
+ */
+dri_bo *
+dri_ttm_bo_create_from_handle(dri_bufmgr *bufmgr, const char *name,
+			      unsigned int handle)
+{
+   dri_bufmgr_ttm *ttm_bufmgr;
+   dri_bo_ttm *ttm_buf;
+   int ret;
+
+   ttm_bufmgr = (dri_bufmgr_ttm *)bufmgr;
+
+   ttm_buf = malloc(sizeof(*ttm_buf));
+   if (!ttm_buf)
+      return NULL;
+
+   ret = drmBOReference(ttm_bufmgr->fd, handle, &ttm_buf->drm_bo);
+   if (ret != 0) {
+      free(ttm_buf);
+      return NULL;
+   }
+   ttm_buf->bo.size = ttm_buf->drm_bo.size;
+   ttm_buf->bo.offset = ttm_buf->drm_bo.offset;
+   ttm_buf->bo.virtual = NULL;
+   ttm_buf->bo.bufmgr = bufmgr;
+   ttm_buf->name = name;
+   ttm_buf->refcount = 1;
+   ttm_buf->owner = GL_FALSE;
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "bo_create_from_handle: %p (%s)\n", &ttm_buf->bo,
+	   ttm_buf->name);
+#endif
+
+   return &ttm_buf->bo;
+}
+
+static void
+dri_ttm_bo_reference(dri_bo *buf)
+{
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)buf->bufmgr;
+   dri_bo_ttm *ttm_buf = (dri_bo_ttm *)buf;
+
+   _glthread_LOCK_MUTEX(bufmgr_ttm->mutex);
+   ttm_buf->refcount++;
+   _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+}
+
+static void
+dri_ttm_bo_unreference(dri_bo *buf)
+{
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)buf->bufmgr;
+   dri_bo_ttm *ttm_buf = (dri_bo_ttm *)buf;
+
+   if (!buf)
+      return;
+
+   _glthread_LOCK_MUTEX(bufmgr_ttm->mutex);
+   if (--ttm_buf->refcount == 0) {
+      int ret;
+
+      /* XXX Having to use drmBODestroy as the opposite of drmBOCreate instead
+       * of simply unreferencing is madness, and leads to behaviors we may not
+       * want (making the buffer unsharable).
+       */
+      if (ttm_buf->owner)
+	 ret = drmBODestroy(bufmgr_ttm->fd, &ttm_buf->drm_bo);
+      else
+	 ret = drmBOUnReference(bufmgr_ttm->fd, &ttm_buf->drm_bo);
+      if (ret != 0) {
+	 fprintf(stderr, "drmBOUnReference failed (%s): %s\n", ttm_buf->name,
+		 strerror(-ret));
+      }
+#if BUFMGR_DEBUG
+      fprintf(stderr, "bo_unreference final: %p (%s)\n",
+	      &ttm_buf->bo, ttm_buf->name);
+#endif
+      _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+      free(buf);
+      return;
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+}
+
+static int
+dri_ttm_bo_map(dri_bo *buf, GLboolean write_enable)
+{
+   dri_bufmgr_ttm *bufmgr_ttm;
+   dri_bo_ttm *ttm_buf = (dri_bo_ttm *)buf;
+   unsigned int flags;
+
+   bufmgr_ttm = (dri_bufmgr_ttm *)buf->bufmgr;
+
+   flags = DRM_BO_FLAG_READ;
+   if (write_enable)
+       flags |= DRM_BO_FLAG_WRITE;
+
+   assert(buf->virtual == NULL);
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "bo_map: %p (%s)\n", &ttm_buf->bo, ttm_buf->name);
+#endif
+
+   return drmBOMap(bufmgr_ttm->fd, &ttm_buf->drm_bo, flags, 0, &buf->virtual);
+}
+
+static int
+dri_ttm_bo_unmap(dri_bo *buf)
+{
+   dri_bufmgr_ttm *bufmgr_ttm;
+   dri_bo_ttm *ttm_buf = (dri_bo_ttm *)buf;
+
+   if (buf == NULL)
+      return 0;
+
+   bufmgr_ttm = (dri_bufmgr_ttm *)buf->bufmgr;
+
+   assert(buf->virtual != NULL);
+
+   buf->virtual = NULL;
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "bo_unmap: %p (%s)\n", &ttm_buf->bo, ttm_buf->name);
+#endif
+
+   return drmBOUnmap(bufmgr_ttm->fd, &ttm_buf->drm_bo);
+}
+
+static int
+dri_ttm_validate(dri_bo *buf, unsigned int flags)
+{
+   dri_bufmgr_ttm *bufmgr_ttm;
+   dri_bo_ttm *ttm_buf = (dri_bo_ttm *)buf;
+   unsigned int mask;
+   int err;
+
+   /* XXX: Sanity-check whether we've already validated this one under
+    * different flags.  See drmAddValidateItem().
+    */
+
+   bufmgr_ttm = (dri_bufmgr_ttm *)buf->bufmgr;
+
+   /* Calculate the appropriate mask to pass to the DRM. There appears to be
+    * be a direct relationship to flags, so it's unnecessary to have it passed
+    * in as an argument.
+    */
+   mask = DRM_BO_MASK_MEM;
+   mask |= flags & (DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE | DRM_BO_FLAG_EXE);
+
+   err = drmBOValidate(bufmgr_ttm->fd, &ttm_buf->drm_bo, 0, flags, mask, 0);
+
+   if (err == 0) {
+      /* XXX: add to fence list for sanity checking */
+   } else {
+      fprintf(stderr, "failed to validate buffer (%s): %s\n",
+	      ttm_buf->name, strerror(-err));
+   }
+
+   buf->offset = ttm_buf->drm_bo.offset;
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "bo_validate: %p (%s)\n", &ttm_buf->bo, ttm_buf->name);
+#endif
+
+   return err;
+}
+
+static dri_fence *
+dri_ttm_fence_validated(dri_bufmgr *bufmgr, const char *name,
+			GLboolean flushed)
+{
+   dri_fence_ttm *fence_ttm = malloc(sizeof(*fence_ttm));
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)bufmgr;
+   int ret;
+   unsigned int type;
+
+   if (!fence_ttm)
+      return NULL;
+
+   if (flushed)
+      type = bufmgr_ttm->fence_type_flush;
+   else
+      type = bufmgr_ttm->fence_type;
+
+   fence_ttm->refcount = 1;
+   fence_ttm->name = name;
+   fence_ttm->fence.bufmgr = bufmgr;
+   ret = drmFenceBuffers(bufmgr_ttm->fd, type, 0, &fence_ttm->drm_fence);
+   if (ret) {
+      fprintf(stderr, "failed to fence (%s): %s\n", name, strerror(-ret));
+      free(fence_ttm);
+      return NULL;
+   }
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "fence_validated: %p (%s)\n", &fence_ttm->fence,
+	   fence_ttm->name);
+#endif
+
+   return &fence_ttm->fence;
+}
+
+static void
+dri_ttm_fence_reference(dri_fence *fence)
+{
+   dri_fence_ttm *fence_ttm = (dri_fence_ttm *)fence;
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)fence->bufmgr;
+
+   _glthread_LOCK_MUTEX(bufmgr_ttm->mutex);
+   ++fence_ttm->refcount;
+   _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+}
+
+static void
+dri_ttm_fence_unreference(dri_fence *fence)
+{
+   dri_fence_ttm *fence_ttm = (dri_fence_ttm *)fence;
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)fence->bufmgr;
+
+   if (!fence)
+      return;
+
+   _glthread_LOCK_MUTEX(bufmgr_ttm->mutex);
+   if (--fence_ttm->refcount == 0) {
+      int ret;
+
+      /* XXX Having to use drmFenceDestroy as the opposite of drmFenceBuffers
+       * instead of simply unreferencing is madness, and leads to behaviors we
+       * may not want (making the fence unsharable).  This behavior by the DRM
+       * ioctls should be fixed, and drmFenceDestroy eliminated.
+       */
+      ret = drmFenceDestroy(bufmgr_ttm->fd, &fence_ttm->drm_fence);
+      if (ret != 0) {
+	 fprintf(stderr, "drmFenceDestroy failed (%s): %s\n",
+		 fence_ttm->name, strerror(-ret));
+      }
+
+      _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+      free(fence);
+      return;
+   }
+   _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+}
+
+static void
+dri_ttm_fence_wait(dri_fence *fence)
+{
+   dri_fence_ttm *fence_ttm = (dri_fence_ttm *)fence;
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)fence->bufmgr;
+   int ret;
+
+   _glthread_LOCK_MUTEX(bufmgr_ttm->mutex);
+   ret = drmFenceWait(bufmgr_ttm->fd, 0, &fence_ttm->drm_fence, 0);
+   _glthread_UNLOCK_MUTEX(bufmgr_ttm->mutex);
+   if (ret != 0) {
+      _mesa_printf("%s:%d: Error %d waiting for fence %s.\n",
+		   __FILE__, __LINE__, ret, fence_ttm->name);
+      abort();
+   }
+
+#if BUFMGR_DEBUG
+   fprintf(stderr, "fence_wait: %p (%s)\n", &fence_ttm->fence,
+	   fence_ttm->name);
+#endif
+}
+
+static void
+dri_bufmgr_ttm_destroy(dri_bufmgr *bufmgr)
+{
+   dri_bufmgr_ttm *bufmgr_ttm = (dri_bufmgr_ttm *)bufmgr;
+
+   _glthread_DESTROY_MUTEX(bufmgr_ttm->mutex);
+   free(bufmgr);
+}
+
+/**
+ * Initializes the TTM buffer manager, which uses the kernel to allocate, map,
+ * and manage map buffer objections.
+ *
+ * \param fd File descriptor of the opened DRM device.
+ * \param fence_type Driver-specific fence type used for fences with no flush.
+ * \param fence_type_flush Driver-specific fence type used for fences with a
+ *	  flush.
+ */
+dri_bufmgr *
+dri_bufmgr_ttm_init(int fd, unsigned int fence_type,
+		    unsigned int fence_type_flush)
+{
+   dri_bufmgr_ttm *bufmgr_ttm;
+
+   bufmgr_ttm = malloc(sizeof(*bufmgr_ttm));
+   bufmgr_ttm->fd = fd;
+   bufmgr_ttm->fence_type = fence_type;
+   bufmgr_ttm->fence_type_flush = fence_type_flush;
+   _glthread_INIT_MUTEX(bufmgr_ttm->mutex);
+
+   bufmgr_ttm->bufmgr.bo_alloc = dri_ttm_alloc;
+   bufmgr_ttm->bufmgr.bo_alloc_static = dri_ttm_alloc_static;
+   bufmgr_ttm->bufmgr.bo_reference = dri_ttm_bo_reference;
+   bufmgr_ttm->bufmgr.bo_unreference = dri_ttm_bo_unreference;
+   bufmgr_ttm->bufmgr.bo_map = dri_ttm_bo_map;
+   bufmgr_ttm->bufmgr.bo_unmap = dri_ttm_bo_unmap;
+   bufmgr_ttm->bufmgr.bo_validate = dri_ttm_validate;
+   bufmgr_ttm->bufmgr.fence_validated = dri_ttm_fence_validated;
+   bufmgr_ttm->bufmgr.fence_reference = dri_ttm_fence_reference;
+   bufmgr_ttm->bufmgr.fence_unreference = dri_ttm_fence_unreference;
+   bufmgr_ttm->bufmgr.fence_wait = dri_ttm_fence_wait;
+   bufmgr_ttm->bufmgr.destroy = dri_bufmgr_ttm_destroy;
+
+   return &bufmgr_ttm->bufmgr;
+}
diff --git a/src/mesa/drivers/dri/common/dri_bufpool.h b/src/mesa/drivers/dri/common/dri_bufpool.h
deleted file mode 100644
index c6fb2c3ce01..00000000000
--- a/src/mesa/drivers/dri/common/dri_bufpool.h
+++ /dev/null
@@ -1,86 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
- */
-
-#ifndef _DRI_BUFPOOL_H_
-#define _DRI_BUFPOOL_H_
-
-#include <xf86drm.h>
-struct _DriFenceObject;
-
-typedef struct _DriBufferPool
-{
-   int fd;
-   int (*map) (struct _DriBufferPool * pool, void *private,
-               unsigned flags, int hint, void **virtual);
-   int (*unmap) (struct _DriBufferPool * pool, void *private);
-   int (*destroy) (struct _DriBufferPool * pool, void *private);
-   unsigned long (*offset) (struct _DriBufferPool * pool, void *private);
-   unsigned (*flags) (struct _DriBufferPool * pool, void *private);
-   unsigned long (*size) (struct _DriBufferPool * pool, void *private);
-   void *(*create) (struct _DriBufferPool * pool, unsigned long size,
-                    unsigned flags, unsigned hint, unsigned alignment);
-   int (*fence) (struct _DriBufferPool * pool, void *private,
-                 struct _DriFenceObject * fence);
-   drmBO *(*kernel) (struct _DriBufferPool * pool, void *private);
-   int (*validate) (struct _DriBufferPool * pool, void *private);
-   void *(*setstatic) (struct _DriBufferPool * pool, unsigned long offset,
-                       unsigned long size, void *virtual, unsigned flags);
-   int (*waitIdle) (struct _DriBufferPool *pool, void *private,
-		    int lazy);
-   void (*takeDown) (struct _DriBufferPool * pool);
-   void *data;
-} DriBufferPool;
-
-extern void bmError(int val, const char *file, const char *function,
-                    int line);
-#define BM_CKFATAL(val)					       \
-  do{							       \
-    int tstVal = (val);					       \
-    if (tstVal) 					       \
-      bmError(tstVal, __FILE__, __FUNCTION__, __LINE__);       \
-  } while(0);
-
-
-
-
-
-/*
- * Builtin pools.
- */
-
-/*
- * Kernel buffer objects. Size in multiples of page size. Page size aligned.
- */
-
-extern struct _DriBufferPool *driDRMPoolInit(int fd);
-extern struct _DriBufferPool *driDRMStaticPoolInit(int fd);
-
-#endif
diff --git a/src/mesa/drivers/dri/common/dri_drmpool.c b/src/mesa/drivers/dri/common/dri_drmpool.c
deleted file mode 100644
index 878a148b397..00000000000
--- a/src/mesa/drivers/dri/common/dri_drmpool.c
+++ /dev/null
@@ -1,227 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
- */
-
-#include <xf86drm.h>
-#include <stdlib.h>
-#include <unistd.h>
-#include "dri_bufpool.h"
-
-/*
- * Buffer pool implementation using DRM buffer objects as DRI buffer objects.
- */
-
-static void *
-pool_create(struct _DriBufferPool *pool,
-            unsigned long size, unsigned flags, unsigned hint,
-            unsigned alignment)
-{
-   drmBO *buf = (drmBO *) malloc(sizeof(*buf));
-   int ret;
-   unsigned pageSize = getpagesize();
-
-   if (!buf)
-      return NULL;
-
-   if ((alignment > pageSize) && (alignment % pageSize)) {
-      return NULL;
-   }
-
-   ret = drmBOCreate(pool->fd, 0, size, alignment / pageSize,
-		     NULL, drm_bo_type_dc,
-                     flags, hint, buf);
-   if (ret) {
-      free(buf);
-      return NULL;
-   }
-
-   return (void *) buf;
-}
-
-static int
-pool_destroy(struct _DriBufferPool *pool, void *private)
-{
-   int ret;
-   drmBO *buf = (drmBO *) private;
-   ret = drmBODestroy(pool->fd, buf);
-   free(buf);
-   return ret;
-}
-
-static int
-pool_map(struct _DriBufferPool *pool, void *private, unsigned flags,
-         int hint, void **virtual)
-{
-   drmBO *buf = (drmBO *) private;
-
-   return drmBOMap(pool->fd, buf, flags, hint, virtual);
-}
-
-static int
-pool_unmap(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return drmBOUnmap(pool->fd, buf);
-}
-
-static unsigned long
-pool_offset(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->offset;
-}
-
-static unsigned
-pool_flags(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->flags;
-}
-
-
-static unsigned long
-pool_size(struct _DriBufferPool *pool, void *private)
-{
-   drmBO *buf = (drmBO *) private;
-   return buf->size;
-}
-
-static int
-pool_fence(struct _DriBufferPool *pool, void *private,
-           struct _DriFenceObject *fence)
-{
-   /*
-    * Noop. The kernel handles all fencing.
-    */
-
-   return 0;
-}
-
-static drmBO *
-pool_kernel(struct _DriBufferPool *pool, void *private)
-{
-   return (drmBO *) private;
-}
-
-static int
-pool_waitIdle(struct _DriBufferPool *pool, void *private, int lazy)
-{
-   drmBO *buf = (drmBO *) private;
-   return drmBOWaitIdle(pool->fd, buf, (lazy) ? DRM_BO_HINT_WAIT_LAZY:0);
-}
-
-    
-static void
-pool_takedown(struct _DriBufferPool *pool)
-{
-   free(pool);
-}
-
-
-struct _DriBufferPool *
-driDRMPoolInit(int fd)
-{
-   struct _DriBufferPool *pool;
-
-   pool = (struct _DriBufferPool *) malloc(sizeof(*pool));
-
-   if (!pool)
-      return NULL;
-
-   pool->fd = fd;
-   pool->map = &pool_map;
-   pool->unmap = &pool_unmap;
-   pool->destroy = &pool_destroy;
-   pool->offset = &pool_offset;
-   pool->flags = &pool_flags;
-   pool->size = &pool_size;
-   pool->create = &pool_create;
-   pool->fence = &pool_fence;
-   pool->kernel = &pool_kernel;
-   pool->validate = NULL;
-   pool->setstatic = NULL;
-   pool->waitIdle = &pool_waitIdle;
-   pool->takeDown = &pool_takedown;
-   pool->data = NULL;
-   return pool;
-}
-
-
-static void *
-pool_setstatic(struct _DriBufferPool *pool, unsigned long offset,
-               unsigned long size, void *virtual, unsigned flags)
-{
-   drmBO *buf = (drmBO *) malloc(sizeof(*buf));
-   int ret;
-
-   if (!buf)
-      return NULL;
-
-   ret = drmBOCreate(pool->fd, offset, size, 0, NULL, drm_bo_type_fake,
-                     flags, DRM_BO_HINT_DONT_FENCE, buf);
-
-   if (ret) {
-      free(buf);
-      return NULL;
-   }
-
-   buf->virtual = virtual;
-
-   return (void *) buf;
-}
-
-
-struct _DriBufferPool *
-driDRMStaticPoolInit(int fd)
-{
-   struct _DriBufferPool *pool;
-
-   pool = (struct _DriBufferPool *) malloc(sizeof(*pool));
-
-   if (!pool)
-      return NULL;
-
-   pool->fd = fd;
-   pool->map = &pool_map;
-   pool->unmap = &pool_unmap;
-   pool->destroy = &pool_destroy;
-   pool->offset = &pool_offset;
-   pool->flags = &pool_flags;
-   pool->size = &pool_size;
-   pool->create = NULL;
-   pool->fence = &pool_fence;
-   pool->kernel = &pool_kernel;
-   pool->validate = NULL;
-   pool->setstatic = &pool_setstatic;
-   pool->waitIdle = &pool_waitIdle;
-   pool->takeDown = &pool_takedown;
-   pool->data = NULL;
-   return pool;
-}
diff --git a/src/mesa/drivers/dri/common/spantmp2.h b/src/mesa/drivers/dri/common/spantmp2.h
index 50f3cf55816..53f5f846a0b 100644
--- a/src/mesa/drivers/dri/common/spantmp2.h
+++ b/src/mesa/drivers/dri/common/spantmp2.h
@@ -114,7 +114,7 @@
     do {                                                                \
         GLuint p = *(volatile GLuint *) GET_PTR(_x, _y);                \
        __asm__ __volatile__( "bswap	%0; rorl $8, %0"                \
-				: "=r" (p) : "r" (p) );                 \
+				: "=r" (p) : "0" (p) );                 \
        ((GLuint *)rgba)[0] = p;                                         \
     } while (0)
 # elif defined( MESA_BIG_ENDIAN )
diff --git a/src/mesa/drivers/dri/glcore/Makefile b/src/mesa/drivers/dri/glcore/Makefile
index a9e96970fae..968190acfc0 100644
--- a/src/mesa/drivers/dri/glcore/Makefile
+++ b/src/mesa/drivers/dri/glcore/Makefile
@@ -61,7 +61,8 @@ 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) \
+	CC="$(CC)" CXX="$(CXX)" $(TOP)/bin/mklib -o $(LIBNAME) -noprefix \
+		-ldflags '$(LDFLAGS)' -install $(TOP)/$(LIB_DIR) \
 		$(OBJECTS) $(WINLIB) $(LIB_DEPS) $(WINOBJ) $(MESA_MODULES)
 
 
diff --git a/src/mesa/drivers/dri/i810/i810screen.c b/src/mesa/drivers/dri/i810/i810screen.c
index f64c10a9ae0..f8cf050d7e4 100644
--- a/src/mesa/drivers/dri/i810/i810screen.c
+++ b/src/mesa/drivers/dri/i810/i810screen.c
@@ -288,8 +288,8 @@ i810InitDriver(__DRIscreenPrivate *sPriv)
 	      i810Screen->depth.handle,
 	      i810Screen->depth.size,
 	      (drmAddress *)&i810Screen->depth.map) != 0) {
-      FREE(i810Screen);
       drmUnmap(i810Screen->back.map, i810Screen->back.size);
+      FREE(i810Screen);
       sPriv->private = NULL;
       __driUtilMessage("i810InitDriver: drmMap (2) failed");
       return GL_FALSE;
diff --git a/src/mesa/drivers/dri/i915/Makefile b/src/mesa/drivers/dri/i915/Makefile
index 805abf75e08..38e40902111 100644
--- a/src/mesa/drivers/dri/i915/Makefile
+++ b/src/mesa/drivers/dri/i915/Makefile
@@ -7,16 +7,6 @@ LIBNAME = i915_dri.so
 MINIGLX_SOURCES = server/intel_dri.c
 
 DRIVER_SOURCES = \
-	i915_context.c \
-	i915_debug.c \
-	i915_fragprog.c \
-	i915_metaops.c \
-	i915_program.c \
-	i915_state.c \
-	i915_tex.c \
-	i915_texprog.c \
-	i915_texstate.c \
-	i915_vtbl.c \
 	i830_context.c \
 	i830_metaops.c \
 	i830_state.c \
@@ -24,27 +14,59 @@ DRIVER_SOURCES = \
 	i830_tex.c \
 	i830_texstate.c \
 	i830_vtbl.c \
+	intel_render.c \
+	intel_regions.c \
+	intel_buffer_objects.c \
 	intel_batchbuffer.c \
+	intel_mipmap_tree.c \
+	i915_tex_layout.c \
+	intel_tex_layout.c \
+	intel_tex_image.c \
+	intel_tex_subimage.c \
+	intel_tex_copy.c \
+	intel_tex_validate.c \
+	intel_tex_format.c \
+	intel_tex.c \
+	intel_pixel.c \
+	intel_pixel_copy.c \
+	intel_pixel_read.c \
+	intel_pixel_draw.c \
+	intel_buffers.c \
+	intel_blit.c \
+	i915_tex.c \
+	i915_texstate.c \
+	i915_context.c \
+	i915_debug.c \
+	i915_debug_fp.c \
+	i915_fragprog.c \
+	i915_metaops.c \
+	i915_program.c \
+	i915_state.c \
+	i915_vtbl.c \
 	intel_context.c \
+	intel_decode.c \
 	intel_ioctl.c \
-	intel_pixel.c \
-	intel_render.c \
 	intel_rotate.c \
 	intel_screen.c \
 	intel_span.c \
 	intel_state.c \
-	intel_tex.c \
-	intel_texmem.c \
-	intel_tris.c 
+	intel_tris.c \
+	intel_fbo.c \
+	intel_depthstencil.c
 
 C_SOURCES = \
 	$(COMMON_SOURCES) \
+	$(COMMON_BM_SOURCES) \
 	$(DRIVER_SOURCES)
 
 ASM_SOURCES = 
 
-
+DRIVER_DEFINES = -I../intel $(shell pkg-config libdrm --atleast-version=2.3.1 \
+				&& echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP")
 
 include ../Makefile.template
 
+intel_decode.o: ../intel/intel_decode.c
+intel_tex_layout.o: ../intel/intel_tex_layout.c
+
 symlinks:
diff --git a/src/mesa/drivers/dri/i915/i830_context.c b/src/mesa/drivers/dri/i915/i830_context.c
index 7ca601e1b5c..2ff8621c42f 100644
--- a/src/mesa/drivers/dri/i915/i830_context.c
+++ b/src/mesa/drivers/dri/i915/i830_context.c
@@ -38,37 +38,38 @@
  * Mesa's Driver Functions
  ***************************************/
 
-static const struct dri_extension i830_extensions[] =
-{
-    { "GL_ARB_texture_env_crossbar",       NULL },
-    { NULL,                                NULL }
+static const struct dri_extension i830_extensions[] = {
+   {"GL_ARB_texture_env_crossbar", NULL},
+   {NULL, NULL}
 };
 
 
-static void i830InitDriverFunctions( struct dd_function_table *functions )
+static void
+i830InitDriverFunctions(struct dd_function_table *functions)
 {
-   intelInitDriverFunctions( functions );
-   i830InitStateFuncs( functions );
-   i830InitTextureFuncs( functions );
+   intelInitDriverFunctions(functions);
+   i830InitStateFuncs(functions);
+   i830InitTextureFuncs(functions);
 }
 
 
-GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate)
+GLboolean
+i830CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate)
 {
    struct dd_function_table functions;
-   i830ContextPtr i830 = (i830ContextPtr) CALLOC_STRUCT(i830_context);
-   intelContextPtr intel = &i830->intel;
+   struct i830_context *i830 = CALLOC_STRUCT(i830_context);
+   struct intel_context *intel = &i830->intel;
    GLcontext *ctx = &intel->ctx;
-   GLuint i;
-   if (!i830) return GL_FALSE;
+   if (!i830)
+      return GL_FALSE;
 
-   i830InitVtbl( i830 );
-   i830InitDriverFunctions( &functions );
+   i830InitVtbl(i830);
+   i830InitDriverFunctions(&functions);
 
-   if (!intelInitContext( intel, mesaVis, driContextPriv,
-			  sharedContextPrivate, &functions )) {
+   if (!intelInitContext(intel, mesaVis, driContextPriv,
+                         sharedContextPrivate, &functions)) {
       FREE(i830);
       return GL_FALSE;
    }
@@ -77,48 +78,27 @@ GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
    intel->ctx.Const.MaxTextureImageUnits = I830_TEX_UNITS;
    intel->ctx.Const.MaxTextureCoordUnits = I830_TEX_UNITS;
 
-   intel->nr_heaps = 1;
-   intel->texture_heaps[0] = 
-      driCreateTextureHeap( 0, intel,
-			    intel->intelScreen->tex.size,
-			    12,
-			    I830_NR_TEX_REGIONS,
-			    intel->sarea->texList,
-			    (unsigned *) & intel->sarea->texAge,
-			    & intel->swapped,
-			    sizeof( struct i830_texture_object ),
-			    (destroy_texture_object_t *)intelDestroyTexObj );
-
-   /* FIXME: driCalculateMaxTextureLevels assumes that mipmaps are tightly
-    * FIXME: packed, but they're not in Intel graphics hardware.
+   /* Advertise the full hardware capabilities.  The new memory
+    * manager should cope much better with overload situations:
     */
-   intel->ctx.Const.MaxTextureUnits = I830_TEX_UNITS;
-   i = driQueryOptioni( &intel->optionCache, "allow_large_textures");
-   driCalculateMaxTextureLevels( intel->texture_heaps,
-				 intel->nr_heaps,
-				 &intel->ctx.Const,
-				 4,
-				 11, /* max 2D texture size is 2048x2048 */
-				 8,  /* max 3D texture size is 256^3 */
-				 10, /* max CUBE texture size is 1024x1024 */
-				 11, /* max RECT. supported */
-				 12,
-				 GL_FALSE,
-				 i );
-
-   _tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-		       18 * sizeof(GLfloat) );
+   ctx->Const.MaxTextureLevels = 12;
+   ctx->Const.Max3DTextureLevels = 9;
+   ctx->Const.MaxCubeTextureLevels = 11;
+   ctx->Const.MaxTextureRectSize = (1 << 11);
+   ctx->Const.MaxTextureUnits = I830_TEX_UNITS;
 
-   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
+   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
+                      18 * sizeof(GLfloat));
 
-   driInitExtensions( ctx, i830_extensions, GL_FALSE );
+   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
 
-   i830InitState( i830 );
+   driInitExtensions(ctx, i830_extensions, GL_FALSE);
 
+   i830InitState(i830);
+   i830InitMetaFuncs(i830);
 
-   _tnl_allow_vertex_fog( ctx, 1 ); 
-   _tnl_allow_pixel_fog( ctx, 0 ); 
+   _tnl_allow_vertex_fog(ctx, 1);
+   _tnl_allow_pixel_fog(ctx, 0);
 
    return GL_TRUE;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_context.h b/src/mesa/drivers/dri/i915/i830_context.h
index bae777dd5a4..9397fa45b52 100644
--- a/src/mesa/drivers/dri/i915/i830_context.h
+++ b/src/mesa/drivers/dri/i915/i830_context.h
@@ -49,17 +49,15 @@
  */
 #define I830_DESTREG_CBUFADDR0 0
 #define I830_DESTREG_CBUFADDR1 1
-#define I830_DESTREG_CBUFADDR2 2
-#define I830_DESTREG_DBUFADDR0 3
-#define I830_DESTREG_DBUFADDR1 4
-#define I830_DESTREG_DBUFADDR2 5
-#define I830_DESTREG_DV0 6
-#define I830_DESTREG_DV1 7
-#define I830_DESTREG_SENABLE 8
-#define I830_DESTREG_SR0 9
-#define I830_DESTREG_SR1 10
-#define I830_DESTREG_SR2 11
-#define I830_DEST_SETUP_SIZE 12
+#define I830_DESTREG_DBUFADDR0 2
+#define I830_DESTREG_DBUFADDR1 3
+#define I830_DESTREG_DV0 4
+#define I830_DESTREG_DV1 5
+#define I830_DESTREG_SENABLE 6
+#define I830_DESTREG_SR0 7
+#define I830_DESTREG_SR1 8
+#define I830_DESTREG_SR2 9
+#define I830_DEST_SETUP_SIZE 10
 
 #define I830_CTXREG_STATE1		0
 #define I830_CTXREG_STATE2		1
@@ -73,7 +71,7 @@
 #define I830_CTXREG_AA			9
 #define I830_CTXREG_FOGCOLOR		10
 #define I830_CTXREG_BLENDCOLOR0		11
-#define I830_CTXREG_BLENDCOLOR1		12 
+#define I830_CTXREG_BLENDCOLOR1		12
 #define I830_CTXREG_VF			13
 #define I830_CTXREG_VF2			14
 #define I830_CTXREG_MCSB0		15
@@ -84,17 +82,16 @@
 #define I830_STPREG_ST1        1
 #define I830_STP_SETUP_SIZE    2
 
-#define I830_TEXREG_TM0LI      0 /* load immediate 2 texture map n */
-#define I830_TEXREG_TM0S0      1
-#define I830_TEXREG_TM0S1      2
-#define I830_TEXREG_TM0S2      3
-#define I830_TEXREG_TM0S3      4
-#define I830_TEXREG_TM0S4      5
-#define I830_TEXREG_MCS	       6	/* _3DSTATE_MAP_COORD_SETS */
-#define I830_TEXREG_CUBE       7	/* _3DSTATE_MAP_SUBE */
-#define I830_TEX_SETUP_SIZE    8
+#define I830_TEXREG_TM0LI      0        /* load immediate 2 texture map n */
+#define I830_TEXREG_TM0S1      1
+#define I830_TEXREG_TM0S2      2
+#define I830_TEXREG_TM0S3      3
+#define I830_TEXREG_TM0S4      4
+#define I830_TEXREG_MCS	       5        /* _3DSTATE_MAP_COORD_SETS */
+#define I830_TEXREG_CUBE       6        /* _3DSTATE_MAP_SUBE */
+#define I830_TEX_SETUP_SIZE    7
 
-#define I830_TEXBLEND_SIZE	12	/* (4 args + op) * 2 + COLOR_FACTOR */
+#define I830_TEXBLEND_SIZE	12      /* (4 args + op) * 2 + COLOR_FACTOR */
 
 struct i830_texture_object
 {
@@ -104,30 +101,39 @@ struct i830_texture_object
 
 #define I830_TEX_UNITS 4
 
-struct i830_hw_state {
+struct i830_hw_state
+{
    GLuint Ctx[I830_CTX_SETUP_SIZE];
    GLuint Buffer[I830_DEST_SETUP_SIZE];
    GLuint Stipple[I830_STP_SETUP_SIZE];
    GLuint Tex[I830_TEX_UNITS][I830_TEX_SETUP_SIZE];
    GLuint TexBlend[I830_TEX_UNITS][I830_TEXBLEND_SIZE];
    GLuint TexBlendWordsUsed[I830_TEX_UNITS];
-   GLuint emitted;		/* I810_UPLOAD_* */
+
+   struct intel_region *draw_region;
+   struct intel_region *depth_region;
+
+   /* Regions aren't actually that appropriate here as the memory may
+    * be from a PBO or FBO.  Will have to do this for draw and depth for
+    * FBO's...
+    */
+   dri_bo *tex_buffer[I830_TEX_UNITS];
+   GLuint tex_offset[I830_TEX_UNITS];
+
+   GLuint emitted;              /* I810_UPLOAD_* */
    GLuint active;
 };
 
-struct i830_context 
+struct i830_context
 {
    struct intel_context intel;
-   
-   DECLARE_RENDERINPUTS(last_index_bitset);
+
+   GLuint lodbias_tm0s3[MAX_TEXTURE_UNITS];
+     DECLARE_RENDERINPUTS(last_index_bitset);
 
    struct i830_hw_state meta, initial, state, *current;
 };
 
-typedef struct i830_context *i830ContextPtr;
-typedef struct i830_texture_object *i830TextureObjectPtr;
-
-#define I830_CONTEXT(ctx)	((i830ContextPtr)(ctx))
 
 
 
@@ -148,71 +154,60 @@ do {						\
 
 /* i830_vtbl.c
  */
-extern void 
-i830InitVtbl( i830ContextPtr i830 );
+extern void i830InitVtbl(struct i830_context *i830);
 
+extern void
+i830_state_draw_region(struct intel_context *intel,
+                       struct i830_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region);
 /* i830_context.c
  */
-extern GLboolean 
-i830CreateContext( const __GLcontextModes *mesaVis,
-		   __DRIcontextPrivate *driContextPriv,
-		   void *sharedContextPrivate);
+extern GLboolean
+i830CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate);
 
 /* i830_tex.c, i830_texstate.c
  */
-extern void 
-i830UpdateTextureState( intelContextPtr intel );
-
-extern void 
-i830InitTextureFuncs( struct dd_function_table *functions );
+extern void i830UpdateTextureState(struct intel_context *intel);
 
-extern intelTextureObjectPtr
-i830AllocTexObj( struct gl_texture_object *tObj );
+extern void i830InitTextureFuncs(struct dd_function_table *functions);
 
 /* i830_texblend.c
  */
-extern GLuint i830SetTexEnvCombine(i830ContextPtr i830,
-    const struct gl_tex_env_combine_state * combine, GLint blendUnit,
-     GLuint texel_op, GLuint *state, const GLfloat *factor );
+extern GLuint i830SetTexEnvCombine(struct i830_context *i830,
+                                   const struct gl_tex_env_combine_state
+                                   *combine, GLint blendUnit, GLuint texel_op,
+                                   GLuint * state, const GLfloat * factor);
 
-extern void 
-i830EmitTextureBlend( i830ContextPtr i830 );
+extern void i830EmitTextureBlend(struct i830_context *i830);
 
 
 /* i830_state.c
  */
-extern void 
-i830InitStateFuncs( struct dd_function_table *functions );
+extern void i830InitStateFuncs(struct dd_function_table *functions);
 
-extern void 
-i830EmitState( i830ContextPtr i830 );
+extern void i830EmitState(struct i830_context *i830);
 
-extern void 
-i830InitState( i830ContextPtr i830 );
+extern void i830InitState(struct i830_context *i830);
 
 /* i830_metaops.c
  */
-extern GLboolean
-i830TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels );
-
-extern GLboolean
-i830TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels );
-
-extern void 
-i830ClearWithTris( intelContextPtr intel, GLbitfield mask,
-		   GLboolean all, GLint cx, GLint cy, GLint cw, GLint ch);
+extern void i830InitMetaFuncs(struct i830_context *i830);
 
 extern void
-i830RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
+i830RotateWindow(struct intel_context *intel, __DRIdrawablePrivate * dPriv,
                  GLuint srcBuf);
 
-#endif
+/*======================================================================
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct i830_context *
+i830_context(GLcontext * ctx)
+{
+   return (struct i830_context *) ctx;
+}
 
+#endif
diff --git a/src/mesa/drivers/dri/i915/i830_metaops.c b/src/mesa/drivers/dri/i915/i830_metaops.c
index c1d7fe349c0..13e4ab3aaca 100644
--- a/src/mesa/drivers/dri/i915/i830_metaops.c
+++ b/src/mesa/drivers/dri/i915/i830_metaops.c
@@ -34,6 +34,7 @@
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
 #include "intel_ioctl.h"
+#include "intel_regions.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
@@ -41,34 +42,26 @@
 /* A large amount of state doesn't need to be uploaded.
  */
 #define ACTIVE (I830_UPLOAD_INVARIENT |         \
-		I830_UPLOAD_TEXBLEND(0) |	\
-		I830_UPLOAD_STIPPLE |		\
 		I830_UPLOAD_CTX |		\
 		I830_UPLOAD_BUFFERS |		\
-		I830_UPLOAD_TEX(0))		
+		I830_UPLOAD_STIPPLE |		\
+		I830_UPLOAD_TEXBLEND(0) |	\
+		I830_UPLOAD_TEX(0))
 
 
 #define SET_STATE( i830, STATE )		\
 do {						\
-   i830->current->emitted = 0;			\
+   i830->current->emitted &= ~ACTIVE;			\
    i830->current = &i830->STATE;		\
-   i830->current->emitted = 0;			\
+   i830->current->emitted &= ~ACTIVE;			\
 } while (0)
 
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void set_initial_state( i830ContextPtr i830 )
-{
-   memcpy(&i830->meta, &i830->initial, sizeof(i830->meta) );
-   i830->meta.active = ACTIVE;
-   i830->meta.emitted = 0;
-}
-
 
-static void set_no_depth_stencil_write( i830ContextPtr i830 )
+static void
+set_no_stencil_write(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
     */
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
@@ -76,6 +69,13 @@ static void set_no_depth_stencil_write( i830ContextPtr i830 )
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_STENCIL_WRITE;
 
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
+}
+
+static void
+set_no_depth_write(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
     */
@@ -87,35 +87,56 @@ static void set_no_depth_stencil_write( i830ContextPtr i830 )
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
-/* Set stencil unit to replace always with the reference value.
+/* Set depth unit to replace.
  */
-static void set_stencil_replace( i830ContextPtr i830,
-				 GLuint s_mask,
-				 GLuint s_clear)
+static void
+set_depth_replace(struct intel_context *intel)
 {
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
+    * ctx->Driver.DepthMask( ctx, GL_TRUE )
     */
    i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DEPTH_WRITE;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DEPTH_WRITE;
+
+   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
+    */
+   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
+   i830->meta.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
+                                          DEPTH_TEST_FUNC
+                                          (COMPAREFUNC_ALWAYS));
+
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
+}
+
+
+/* Set stencil unit to replace always with the reference value.
+ */
+static void
+set_stencil_replace(struct intel_context *intel,
+                    GLuint s_mask, GLuint s_clear)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
+    */
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
 
    /* ctx->Driver.StencilMask( ctx, s_mask )
     */
    i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					   STENCIL_WRITE_MASK((s_mask&0xff)));
+                                          STENCIL_WRITE_MASK((s_mask &
+                                                              0xff)));
 
    /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
     */
    i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |= 
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
       (ENABLE_STENCIL_PARMS |
        STENCIL_FAIL_OP(STENCILOP_REPLACE) |
        STENCIL_PASS_DEPTH_FAIL_OP(STENCILOP_REPLACE) |
@@ -125,14 +146,14 @@ static void set_stencil_replace( i830ContextPtr i830,
     */
    i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					   STENCIL_TEST_MASK(0xff));
+                                          STENCIL_TEST_MASK(0xff));
 
    i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-						ENABLE_STENCIL_TEST_FUNC_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |= 
+                                               ENABLE_STENCIL_TEST_FUNC_MASK);
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
       (ENABLE_STENCIL_REF_VALUE |
        ENABLE_STENCIL_TEST_FUNC |
-       STENCIL_REF_VALUE((s_clear&0xff)) |
+       STENCIL_REF_VALUE((s_clear & 0xff)) |
        STENCIL_TEST_FUNC(COMPAREFUNC_ALWAYS));
 
 
@@ -141,38 +162,43 @@ static void set_stencil_replace( i830ContextPtr i830,
 }
 
 
-static void set_color_mask( i830ContextPtr i830, GLboolean state )
+static void
+set_color_mask(struct intel_context *intel, GLboolean state)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
    const GLuint mask = ((1 << WRITEMASK_RED_SHIFT) |
-			(1 << WRITEMASK_GREEN_SHIFT) |
-			(1 << WRITEMASK_BLUE_SHIFT) |
-			(1 << WRITEMASK_ALPHA_SHIFT));
+                        (1 << WRITEMASK_GREEN_SHIFT) |
+                        (1 << WRITEMASK_BLUE_SHIFT) |
+                        (1 << WRITEMASK_ALPHA_SHIFT));
 
    i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~mask;
 
    if (state) {
-      i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= 
-	 (i830->state.Ctx[I830_CTXREG_ENABLES_2] & mask);
+      i830->meta.Ctx[I830_CTXREG_ENABLES_2] |=
+         (i830->state.Ctx[I830_CTXREG_ENABLES_2] & mask);
    }
-      
+
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
 /* Installs a one-stage passthrough texture blend pipeline.  Is there
  * more that can be done to turn off texturing?
  */
-static void set_no_texture( i830ContextPtr i830 )
+static void
+set_no_texture(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
    static const struct gl_tex_env_combine_state comb = {
       GL_NONE, GL_NONE,
-      { GL_TEXTURE, 0, 0, }, { GL_TEXTURE, 0, 0, },
-      { GL_SRC_COLOR, 0, 0 }, { GL_SRC_ALPHA, 0, 0 },
+      {GL_TEXTURE, 0, 0,}, {GL_TEXTURE, 0, 0,},
+      {GL_SRC_COLOR, 0, 0}, {GL_SRC_ALPHA, 0, 0},
       0, 0, 0, 0
    };
 
    i830->meta.TexBlendWordsUsed[0] =
-     i830SetTexEnvCombine( i830, & comb, 0, TEXBLENDARG_TEXEL0,
-			   i830->meta.TexBlend[0], NULL);
+      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
+                           i830->meta.TexBlend[0], NULL);
 
    i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
    i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
@@ -181,18 +207,22 @@ static void set_no_texture( i830ContextPtr i830 )
 /* Set up a single element blend stage for 'replace' texturing with no
  * funny ops.
  */
-static void enable_texture_blend_replace( i830ContextPtr i830 )
+static void
+set_texture_blend_replace(struct intel_context *intel)
 {
+   struct i830_context *i830 = i830_context(&intel->ctx);
    static const struct gl_tex_env_combine_state comb = {
       GL_REPLACE, GL_REPLACE,
-      { GL_TEXTURE, GL_TEXTURE, GL_TEXTURE }, { GL_TEXTURE, GL_TEXTURE, GL_TEXTURE, },
-      { GL_SRC_COLOR, GL_SRC_COLOR, GL_SRC_COLOR }, { GL_SRC_ALPHA, GL_SRC_ALPHA, GL_SRC_ALPHA },
+      {GL_TEXTURE, GL_TEXTURE, GL_TEXTURE,}, {GL_TEXTURE, GL_TEXTURE,
+                                              GL_TEXTURE,},
+      {GL_SRC_COLOR, GL_SRC_COLOR, GL_SRC_COLOR}, {GL_SRC_ALPHA, GL_SRC_ALPHA,
+                                                   GL_SRC_ALPHA},
       0, 0, 1, 1
    };
 
    i830->meta.TexBlendWordsUsed[0] =
-     i830SetTexEnvCombine( i830, & comb, 0, TEXBLENDARG_TEXEL0,
-			   i830->meta.TexBlend[0], NULL);
+      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
+                           i830->meta.TexBlend[0], NULL);
 
    i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
    i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
@@ -206,717 +236,222 @@ static void enable_texture_blend_replace( i830ContextPtr i830 )
 /* Set up an arbitary piece of memory as a rectangular texture
  * (including the front or back buffer).
  */
-static void set_tex_rect_source( i830ContextPtr i830,
-				 GLuint offset,
-				 GLuint width, 
-				 GLuint height,
-				 GLuint pitch, /* in bytes */
-				 GLuint textureFormat )
+static GLboolean
+set_tex_rect_source(struct intel_context *intel,
+                    dri_bo *buffer,
+                    GLuint offset,
+                    GLuint pitch, GLuint height, GLenum format, GLenum type)
 {
-   GLint numLevels = 1;
+   struct i830_context *i830 = i830_context(&intel->ctx);
    GLuint *setup = i830->meta.Tex[0];
+   GLint numLevels = 1;
+   GLuint textureFormat;
+   GLuint cpp;
 
-/*    fprintf(stderr, "%s: offset: %x w: %d h: %d pitch %d format %x\n", */
-/* 	   __FUNCTION__, offset, width, height, pitch, textureFormat ); */
+   /* A full implementation of this would do the upload through
+    * glTexImage2d, and get all the conversion operations at that
+    * point.  We are restricted, but still at least have access to the
+    * fragment program swizzle.
+    */
+   switch (format) {
+   case GL_BGRA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGBA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_BGR:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5_REV:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGB:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
 
-   setup[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 | 
-			       (LOAD_TEXTURE_MAP0 << 0) | 4);
-   setup[I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE | offset);
+   default:
+      return GL_FALSE;
+   }
+
+   i830->meta.tex_buffer[0] = buffer;
+   i830->meta.tex_offset[0] = offset;
+
+   setup[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
+                               (LOAD_TEXTURE_MAP0 << 0) | 4);
    setup[I830_TEXREG_TM0S1] = (((height - 1) << TM0S1_HEIGHT_SHIFT) |
-			       ((width - 1) << TM0S1_WIDTH_SHIFT) |
-			       textureFormat);
-   setup[I830_TEXREG_TM0S2] = ((((pitch / 4) - 1) << TM0S2_PITCH_SHIFT));   
-   setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAX_MIP_MASK;
-   setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_MIP_MASK;
-   setup[I830_TEXREG_TM0S3] |= ((numLevels - 1)*4) << TM0S3_MIN_MIP_SHIFT;
+                               ((pitch - 1) << TM0S1_WIDTH_SHIFT) |
+                               textureFormat);
+   setup[I830_TEXREG_TM0S2] =
+      (((((pitch * cpp) / 4) -
+         1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
+
+   setup[I830_TEXREG_TM0S3] =
+      ((((numLevels -
+          1) *
+         4) << TM0S3_MIN_MIP_SHIFT) | (FILTER_NEAREST <<
+                                       TM0S3_MIN_FILTER_SHIFT) |
+       (MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT) | (FILTER_NEAREST <<
+                                                     TM0S3_MAG_FILTER_SHIFT));
+
+   setup[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(0));
 
    setup[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-			     MAP_UNIT(0) |
-			     ENABLE_TEXCOORD_PARAMS |
-			     TEXCOORDS_ARE_IN_TEXELUNITS |
-			     TEXCOORDTYPE_CARTESIAN |
-			     ENABLE_ADDR_V_CNTL |
-			     TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
-			     ENABLE_ADDR_U_CNTL |
-			     TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
+                             MAP_UNIT(0) |
+                             ENABLE_TEXCOORD_PARAMS |
+                             TEXCOORDS_ARE_IN_TEXELUNITS |
+                             TEXCOORDTYPE_CARTESIAN |
+                             ENABLE_ADDR_V_CNTL |
+                             TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
+                             ENABLE_ADDR_U_CNTL |
+                             TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
 
    i830->meta.emitted &= ~I830_UPLOAD_TEX(0);
+   return GL_TRUE;
 }
 
 
-/* Select between front and back draw buffers.
- */
-static void set_draw_region( i830ContextPtr i830,
-			      const intelRegion *region )
-{
-   i830->meta.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->meta.Buffer[I830_DESTREG_CBUFADDR2] = region->offset;
-   i830->meta.emitted &= ~I830_UPLOAD_BUFFERS;
-}
-
-/* Setup an arbitary draw format, useful for targeting
- * texture or agp memory.
- */
-#if 0
-static void set_draw_format( i830ContextPtr i830,
-			     GLuint format,
-			     GLuint depth_format)
-{
-   i830->meta.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					  DSTORG_VERT_BIAS(0x8) | /* .5 */
-					  format |
-					  DEPTH_IS_Z |
-					  depth_format);
-}
-#endif
-
-
-static void set_vertex_format( i830ContextPtr i830 )
+static void
+set_vertex_format(struct intel_context *intel)
 {
-   i830->meta.Ctx[I830_CTXREG_VF] =  (_3DSTATE_VFT0_CMD |
-				      VFT0_TEX_COUNT(1) |
-				      VFT0_DIFFUSE |
-				      VFT0_SPEC |
-				      VFT0_XYZW);
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830->meta.Ctx[I830_CTXREG_VF] = (_3DSTATE_VFT0_CMD |
+                                     VFT0_TEX_COUNT(1) |
+                                     VFT0_DIFFUSE | VFT0_XYZ);
    i830->meta.Ctx[I830_CTXREG_VF2] = (_3DSTATE_VFT1_CMD |
-				      VFT1_TEX0_FMT(TEXCOORDFMT_2D) |
-				      VFT1_TEX1_FMT(TEXCOORDFMT_2D) | 
-				      VFT1_TEX2_FMT(TEXCOORDFMT_2D) |
-				      VFT1_TEX3_FMT(TEXCOORDFMT_2D));
+                                      VFT1_TEX0_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX1_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX2_FMT(TEXCOORDFMT_2D) |
+                                      VFT1_TEX3_FMT(TEXCOORDFMT_2D));
    i830->meta.emitted &= ~I830_UPLOAD_CTX;
 }
 
 
-static void draw_quad(i830ContextPtr i830, 
-		      GLfloat x0, GLfloat x1,
-		      GLfloat y0, GLfloat y1, 
-		      GLubyte red, GLubyte green,
-		      GLubyte blue, GLubyte alpha,
-		      GLfloat s0, GLfloat s1,
-		      GLfloat t0, GLfloat t1 )
-{
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i830->intel, 
-						PRIM3D_TRIFAN, 
-						4*vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i;
-
-   
-/*    fprintf(stderr, "%s: %f,%f-%f,%f 0x%x%x%x%x %f,%f-%f,%f\n", */
-/* 	   __FUNCTION__, */
-/* 	   x0,y0,x1,y1,red,green,blue,alpha,s0,t0,s1,t1); */
-
-
-   /* initial vertex, left bottom */
-   tmp.v.x = x0;
-   tmp.v.y = y0;
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0;
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-   tmp.v.u0 = s0;
-   tmp.v.v0 = t0;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right bottom */
-   vb += 8;
-   tmp.v.x = x1;
-   tmp.v.u0 = s1;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right top */
-   vb += 8;
-   tmp.v.y = y1;
-   tmp.v.v0 = t1;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* left top */
-   vb += 8;
-   tmp.v.x = x0;
-   tmp.v.u0 = s0;
-   for (i = 0 ; i < 8 ; i++)
-      vb[i] = tmp.ui[i];
-
-/*    fprintf(stderr, "%s: DV1: %x\n",  */
-/* 	   __FUNCTION__, i830->meta.Buffer[I830_DESTREG_DV1]); */
-}
-
-static void draw_poly(i830ContextPtr i830, 
-		      GLubyte red, GLubyte green, GLubyte blue, GLubyte alpha,
-                      GLuint numVerts,
-                      GLfloat verts[][2],
-                      GLfloat texcoords[][2])
+static void
+meta_import_pixel_state(struct intel_context *intel)
 {
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i830->intel, 
-						PRIM3D_TRIFAN, 
-						numVerts * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i, k;
-
-   /* initial constant vertex fields */
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-
-   for (k = 0; k < numVerts; k++) {
-      tmp.v.x = verts[k][0];
-      tmp.v.y = verts[k][1];
-      tmp.v.u0 = texcoords[k][0];
-      tmp.v.v0 = texcoords[k][1];
-
-      for (i = 0 ; i < vertex_size ; i++)
-         vb[i] = tmp.ui[i];
-
-      vb += vertex_size;
-   }
-}
-
-void 
-i830ClearWithTris(intelContextPtr intel, GLbitfield mask,
-		  GLboolean allFoo,
-		  GLint cxFoo, GLint cyFoo, GLint cwFoo, GLint chFoo)
-{
-   i830ContextPtr i830 = I830_CONTEXT( intel );
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   intelScreenPrivate *screen = intel->intelScreen;
-   int x0, y0, x1, y1;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-
-   INTEL_FIREVERTICES(intel);
-   SET_STATE( i830, meta );
-   set_initial_state( i830 );
-/*    set_no_texture( i830 ); */
-   set_vertex_format( i830 ); 
-
-   LOCK_HARDWARE(intel);
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   if(!all) {
-      x0 = cx;
-      y0 = cy;
-      x1 = x0 + cw;
-      y1 = y0 + ch;
-   } else {
-      x0 = 0;
-      y0 = 0;
-      x1 = x0 + dPriv->w;
-      y1 = y0 + dPriv->h;
-   }
-
-   /* Don't do any clipping to screen - these are window coordinates.
-    * The active cliprects will be applied as for any other geometry.
-    */
-
-   if(mask & BUFFER_BIT_FRONT_LEFT) {
-      set_no_depth_stencil_write( i830 );
-      set_color_mask( i830, GL_TRUE );
-      set_draw_region( i830, &screen->front );
-      draw_quad(i830, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if(mask & BUFFER_BIT_BACK_LEFT) {
-      set_no_depth_stencil_write( i830 );
-      set_color_mask( i830, GL_TRUE );
-      set_draw_region( i830, &screen->back );
-
-      draw_quad(i830, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if(mask & BUFFER_BIT_STENCIL) {
-      set_stencil_replace( i830, 
-			   intel->ctx.Stencil.WriteMask[0], 
-			   intel->ctx.Stencil.Clear);
-
-      set_color_mask( i830, GL_FALSE );
-      set_draw_region( i830, &screen->front );
-      draw_quad( i830, x0, x1, y0, y1, 0, 0, 0, 0, 0, 0, 0, 0 );
-   }
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   i830->meta.Ctx[I830_CTXREG_STATE1] = i830->state.Ctx[I830_CTXREG_STATE1];
+   i830->meta.Ctx[I830_CTXREG_STATE2] = i830->state.Ctx[I830_CTXREG_STATE2];
+   i830->meta.Ctx[I830_CTXREG_STATE3] = i830->state.Ctx[I830_CTXREG_STATE3];
+   i830->meta.Ctx[I830_CTXREG_STATE4] = i830->state.Ctx[I830_CTXREG_STATE4];
+   i830->meta.Ctx[I830_CTXREG_STATE5] = i830->state.Ctx[I830_CTXREG_STATE5];
+   i830->meta.Ctx[I830_CTXREG_IALPHAB] = i830->state.Ctx[I830_CTXREG_IALPHAB];
+   i830->meta.Ctx[I830_CTXREG_STENCILTST] =
+      i830->state.Ctx[I830_CTXREG_STENCILTST];
+   i830->meta.Ctx[I830_CTXREG_ENABLES_1] =
+      i830->state.Ctx[I830_CTXREG_ENABLES_1];
+   i830->meta.Ctx[I830_CTXREG_ENABLES_2] =
+      i830->state.Ctx[I830_CTXREG_ENABLES_2];
+   i830->meta.Ctx[I830_CTXREG_AA] = i830->state.Ctx[I830_CTXREG_AA];
+   i830->meta.Ctx[I830_CTXREG_FOGCOLOR] =
+      i830->state.Ctx[I830_CTXREG_FOGCOLOR];
+   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR0] =
+      i830->state.Ctx[I830_CTXREG_BLENDCOLOR0];
+   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR1] =
+      i830->state.Ctx[I830_CTXREG_BLENDCOLOR1];
+   i830->meta.Ctx[I830_CTXREG_MCSB0] = i830->state.Ctx[I830_CTXREG_MCSB0];
+   i830->meta.Ctx[I830_CTXREG_MCSB1] = i830->state.Ctx[I830_CTXREG_MCSB1];
+
+
+   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~CULLMODE_MASK;
+   i830->meta.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
+   i830->meta.emitted &= ~I830_UPLOAD_CTX;
 
-   UNLOCK_HARDWARE(intel);
 
-   INTEL_FIREVERTICES(intel);
-   SET_STATE( i830, state );
+   i830->meta.Buffer[I830_DESTREG_SENABLE] =
+      i830->state.Buffer[I830_DESTREG_SENABLE];
+   i830->meta.Buffer[I830_DESTREG_SR1] = i830->state.Buffer[I830_DESTREG_SR1];
+   i830->meta.Buffer[I830_DESTREG_SR2] = i830->state.Buffer[I830_DESTREG_SR2];
+   i830->meta.emitted &= ~I830_UPLOAD_BUFFERS;
 }
 
 
-#if 0
 
-GLboolean
-i830TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels )
+/* Select between front and back draw buffers.
+ */
+static void
+meta_draw_region(struct intel_context *intel,
+                 struct intel_region *color_region,
+                 struct intel_region *depth_region)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   intelScreenPrivate *screen = i830->intel.intelScreen;
-   GLint pitch = pack->RowLength ? pack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = i830->intel.driDrawable;
-   int textureFormat;
-   GLenum glTextureFormat;
-   int src_offset = i830->meta.Buffer[I830_DESTREG_CBUFADDR2];
-   int destOffset = intelAgpOffsetFromVirtual( &i830->intel, pixels);
-   int destFormat, depthFormat, destPitch;
-   drm_clip_rect_t tmp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-
-   if (	ctx->_ImageTransferState ||
-	pack->SwapBytes ||
-	pack->LsbFirst ||
-	!pack->Invert) {
-      fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   switch (screen->fbFormat) {
-   case DV_PF_565:
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      glTextureFormat = GL_RGB;
-      break;
-   case DV_PF_555:
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      glTextureFormat = GL_RGBA;
-      break;
-   case DV_PF_8888:
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      glTextureFormat = GL_RGBA;
-      break;
-   default:
-      fprintf(stderr, "%s: textureFormat failed %x\n", __FUNCTION__,
-	      screen->fbFormat);
-      return GL_FALSE;
-   }
-
-
-   switch (type) {
-   case GL_UNSIGNED_SHORT_5_6_5: 
-      if (format != GL_RGB) return GL_FALSE;
-      destFormat = COLR_BUF_RGB565; 
-      depthFormat = DEPTH_FRMT_16_FIXED;
-      destPitch = pitch * 2;
-      break;
-   case GL_UNSIGNED_INT_8_8_8_8_REV: 
-      if (format != GL_BGRA) return GL_FALSE;
-      destFormat = COLR_BUF_ARGB8888; 
-      depthFormat = DEPTH_FRMT_24_FIXED_8_OTHER;
-      destPitch = pitch * 4;
-      break;
-   default:
-      fprintf(stderr, "%s: destFormat failed %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(type));
-      return GL_FALSE;
-   }
-
-   destFormat |= (0x02<<24);
-
-/*    fprintf(stderr, "type: %s destFormat: %x\n", */
-/* 	   _mesa_lookup_enum_by_nr(type), */
-/* 	   destFormat); */
-
-   intelFlush( ctx );
-
-   SET_STATE( i830, meta );
-   set_initial_state( i830 );
-   set_no_depth_stencil_write( i830 );
-
-   LOCK_HARDWARE( intel );
-   {
-      intelWaitForIdle( intel ); /* required by GL */
-
-      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-#if 0
-      /* FIXME -- Just emit the correct state
-       */
-      if (i830SetParam(i830->driFd, I830_SETPARAM_CBUFFER_PITCH, 
-		      destPitch) != 0) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: setparam failed\n", __FUNCTION__);
-	 return GL_FALSE;
-      }
-#endif
-
-
-      y = dPriv->h - y - height;
-      x += dPriv->x;
-      y += dPriv->y;
-
-
-      /* Set the frontbuffer up as a large rectangular texture.
-       */
-      set_tex_rect_source( i830, 
-			   src_offset, 
-			   screen->width, 
-			   screen->height, 
-			   screen->front.pitch, 
-			   textureFormat ); 
-   
-   
-      enable_texture_blend_replace( i830 ); 
-
-
-      /* Set the 3d engine to draw into the agp memory
-       */
+   struct i830_context *i830 = i830_context(&intel->ctx);
 
-      set_draw_region( i830, destOffset ); 
-      set_draw_format( i830, destFormat, depthFormat );  
-
-
-      /* Draw a single quad, no cliprects:
-       */
-      i830->intel.numClipRects = 1;
-      i830->intel.pClipRects = &tmp;
-      i830->intel.pClipRects[0].x1 = 0;
-      i830->intel.pClipRects[0].y1 = 0;
-      i830->intel.pClipRects[0].x2 = width;
-      i830->intel.pClipRects[0].y2 = height;
-
-      draw_quad( i830, 
-		 0, width, 0, height, 
-		 0, 255, 0, 0, 
-		 x, x+width, y, y+height );
-
-      intelWindowMoved( intel );
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( ctx ); /* required by GL */
-
-   SET_STATE( i830, state );
-   return GL_TRUE;
+   i830_state_draw_region(intel, &i830->meta, color_region, depth_region);
 }
 
 
-GLboolean
-i830TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels )
+/* Operations where the 3D engine is decoupled temporarily from the
+ * current GL state and used for other purposes than simply rendering
+ * incoming triangles.
+ */
+static void
+install_meta_state(struct intel_context *intel)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLint pitch = unpack->RowLength ? unpack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   int textureFormat;
-   GLenum glTextureFormat;
-   int dst_offset = i830->meta.Buffer[I830_DESTREG_CBUFADDR2];
-   int src_offset = intelAgpOffsetFromVirtual( intel, pixels );
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Todo -- upload images that aren't in agp space, then texture
-    * from them.  
-    */
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   memcpy(&i830->meta, &i830->initial, sizeof(i830->meta));
 
-   if ( !intelIsAgpMemory( intel, pixels, pitch*height ) ) {
-      fprintf(stderr, "%s: intelIsAgpMemory failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Todo -- don't want to clobber all the drawing state like we do
-    * for readpixels -- most of this state can be handled just fine.
-    */
-   if (	ctx->_ImageTransferState ||
-	unpack->SwapBytes ||
-	unpack->LsbFirst ||
-	ctx->Color.AlphaEnabled || 
-	ctx->Depth.Test ||
-	ctx->Fog.Enabled ||
-	ctx->Scissor.Enabled ||
-	ctx->Stencil.Enabled ||
-	!ctx->Color.ColorMask[0] ||
-	!ctx->Color.ColorMask[1] ||
-	!ctx->Color.ColorMask[2] ||
-	!ctx->Color.ColorMask[3] ||
-	ctx->Color.ColorLogicOpEnabled ||
-	ctx->Texture._EnabledUnits ||
-	ctx->Depth.OcclusionTest) {
-      fprintf(stderr, "%s: other tests failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Todo -- remove these restrictions:
-    */
-   if (ctx->Pixel.ZoomX != 1.0F ||
-       ctx->Pixel.ZoomY != -1.0F)
-      return GL_FALSE;
-
-
-
-   switch (type) {
-   case GL_UNSIGNED_SHORT_1_5_5_5_REV:
-      if (format != GL_BGRA) return GL_FALSE;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      glTextureFormat = GL_RGBA;
-      break;
-   case GL_UNSIGNED_SHORT_5_6_5: 
-      if (format != GL_RGB) return GL_FALSE;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      glTextureFormat = GL_RGB;
-      break;
-   case GL_UNSIGNED_SHORT_8_8_MESA: 
-      if (format != GL_YCBCR_MESA) return GL_FALSE;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY 
-/*  		       | TM0S1_COLORSPACE_CONVERSION */
-	 );
-      glTextureFormat = GL_YCBCR_MESA;
-      break;
-   case GL_UNSIGNED_SHORT_8_8_REV_MESA: 
-      if (format != GL_YCBCR_MESA) return GL_FALSE;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL 
-/* 		       | TM0S1_COLORSPACE_CONVERSION */
-	 );
-      glTextureFormat = GL_YCBCR_MESA;
-      break;
-   case GL_UNSIGNED_INT_8_8_8_8_REV: 
-      if (format != GL_BGRA) return GL_FALSE;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      glTextureFormat = GL_RGBA;
-      break;
-   default:
-      fprintf(stderr, "%s: destFormat failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   intelFlush( ctx );
-
-   SET_STATE( i830, meta );
-
-   LOCK_HARDWARE( intel );
-   {
-      intelWaitForIdle( intel ); /* required by GL */
-
-      y -= height;			/* cope with pixel zoom */
-
-      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 SET_STATE(i830, state);
-	 fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-
-      y = dPriv->h - y - height;
-
-      set_initial_state( i830 );
-
-      /* Set the pixel image up as a rectangular texture.
-       */
-      set_tex_rect_source( i830, 
-			   src_offset, 
-			   width, 
-			   height, 
-			   pitch, /* XXXX!!!! -- /2 sometimes */
-			   textureFormat ); 
-   
-   
-      enable_texture_blend_replace( i830 ); 
-
-   
-      /* Draw to the current draw buffer:
-       */
-      set_draw_offset( i830, dst_offset );
-
-      /* Draw a quad, use regular cliprects
-       */
-/*       fprintf(stderr, "x: %d y: %d width %d height %d\n", x, y, width, height); */
+   i830->meta.active = ACTIVE;
+   i830->meta.emitted = 0;
 
-      draw_quad( i830, 
-		 x, x+width, y, y+height,
-		 0, 255, 0, 0, 
-		 0, width, 0, height );
+   SET_STATE(i830, meta);
+   set_vertex_format(intel);
+   set_no_texture(intel);
+}
 
-      intelWindowMoved( intel );
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( ctx ); /* required by GL */
-   
+static void
+leave_meta_state(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   intel_region_release(&i830->meta.draw_region);
+   intel_region_release(&i830->meta.depth_region);
+/*    intel_region_release(intel, &i830->meta.tex_region[0]); */
    SET_STATE(i830, state);
-
-   return GL_TRUE;
 }
 
-#endif
 
-/**
- * Copy the window contents named by dPriv to the rotated (or reflected)
- * color buffer.
- * srcBuf is BUFFER_BIT_FRONT_LEFT or BUFFER_BIT_BACK_LEFT to indicate the source.
- */
+
 void
-i830RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf)
+i830InitMetaFuncs(struct i830_context *i830)
 {
-   i830ContextPtr i830 = I830_CONTEXT( intel );
-   intelScreenPrivate *screen = intel->intelScreen;
-   const GLuint cpp = screen->cpp;
-   drm_clip_rect_t fullRect;
-   GLuint textureFormat, srcOffset, srcPitch;
-   const drm_clip_rect_t *clipRects;
-   int numClipRects;
-   int i;
-
-   int xOrig, yOrig;
-   int origNumClipRects;
-   drm_clip_rect_t *origRects;
-
-   /*
-    * set up hardware state
-    */
-   intelFlush( &intel->ctx );
-
-   SET_STATE( i830, meta ); 
-   set_initial_state( i830 ); 
-   set_no_texture( i830 ); 
-   set_vertex_format( i830 ); 
-   set_no_depth_stencil_write( i830 );
-   set_color_mask( i830, GL_FALSE );
-
-   LOCK_HARDWARE(intel);
-
-   /* save current drawing origin and cliprects (restored at end) */
-   xOrig = intel->drawX;
-   yOrig = intel->drawY;
-   origNumClipRects = intel->numClipRects;
-   origRects = intel->pClipRects;
-
-   if (!intel->numClipRects)
-      goto done;
-
-   /*
-    * set drawing origin, cliprects for full-screen access to rotated screen
-    */
-   fullRect.x1 = 0;
-   fullRect.y1 = 0;
-   fullRect.x2 = screen->rotatedWidth;
-   fullRect.y2 = screen->rotatedHeight;
-   intel->drawX = 0;
-   intel->drawY = 0;
-   intel->numClipRects = 1;
-   intel->pClipRects = &fullRect;
-
-   set_draw_region( i830, &screen->rotated );
-
-   if (cpp == 4)
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   else
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-
-   if (srcBuf == BUFFER_BIT_FRONT_LEFT) {
-      srcPitch = screen->front.pitch;   /* in bytes */
-      srcOffset = screen->front.offset; /* bytes */
-      clipRects = dPriv->pClipRects;
-      numClipRects = dPriv->numClipRects;
-   }
-   else {
-      srcPitch = screen->back.pitch;   /* in bytes */
-      srcOffset = screen->back.offset; /* bytes */
-      clipRects = dPriv->pBackClipRects;
-      numClipRects = dPriv->numBackClipRects;
-   }
-
-   /* set the whole screen up as a texture to avoid alignment issues */
-   set_tex_rect_source(i830,
-                       srcOffset,
-                       screen->width,
-                       screen->height,
-                       srcPitch,
-                       textureFormat);
-
-   enable_texture_blend_replace(i830);
-
-   /*
-    * loop over the source window's cliprects
-    */
-   for (i = 0; i < numClipRects; i++) {
-      int srcX0 = clipRects[i].x1;
-      int srcY0 = clipRects[i].y1;
-      int srcX1 = clipRects[i].x2;
-      int srcY1 = clipRects[i].y2;
-      GLfloat verts[4][2], tex[4][2];
-      int j;
-
-      /* build vertices for four corners of clip rect */
-      verts[0][0] = srcX0;  verts[0][1] = srcY0;
-      verts[1][0] = srcX1;  verts[1][1] = srcY0;
-      verts[2][0] = srcX1;  verts[2][1] = srcY1;
-      verts[3][0] = srcX0;  verts[3][1] = srcY1;
-
-      /* .. and texcoords */
-      tex[0][0] = srcX0;  tex[0][1] = srcY0;
-      tex[1][0] = srcX1;  tex[1][1] = srcY0;
-      tex[2][0] = srcX1;  tex[2][1] = srcY1;
-      tex[3][0] = srcX0;  tex[3][1] = srcY1;
-
-      /* transform coords to rotated screen coords */
-
-      for (j = 0; j < 4; j++) {
-         matrix23TransformCoordf(&screen->rotMatrix,
-                                 &verts[j][0], &verts[j][1]);
-      }
-
-      /* draw polygon to map source image to dest region */
-      draw_poly(i830, 255, 255, 255, 255, 4, verts, tex);
-
-   } /* cliprect loop */
-
-   intelFlushBatchLocked( intel, GL_FALSE, GL_FALSE, GL_FALSE );
-
- done:
-   /* restore original drawing origin and cliprects */
-   intel->drawX = xOrig;
-   intel->drawY = yOrig;
-   intel->numClipRects = origNumClipRects;
-   intel->pClipRects = origRects;
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i830, state );
+   i830->intel.vtbl.install_meta_state = install_meta_state;
+   i830->intel.vtbl.leave_meta_state = leave_meta_state;
+   i830->intel.vtbl.meta_no_depth_write = set_no_depth_write;
+   i830->intel.vtbl.meta_no_stencil_write = set_no_stencil_write;
+   i830->intel.vtbl.meta_stencil_replace = set_stencil_replace;
+   i830->intel.vtbl.meta_depth_replace = set_depth_replace;
+   i830->intel.vtbl.meta_color_mask = set_color_mask;
+   i830->intel.vtbl.meta_no_texture = set_no_texture;
+   i830->intel.vtbl.meta_texture_blend_replace = set_texture_blend_replace;
+   i830->intel.vtbl.meta_tex_rect_source = set_tex_rect_source;
+   i830->intel.vtbl.meta_draw_region = meta_draw_region;
+   i830->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_reg.h b/src/mesa/drivers/dri/i915/i830_reg.h
index 98cee2f214c..41280bca7ce 100644
--- a/src/mesa/drivers/dri/i915/i830_reg.h
+++ b/src/mesa/drivers/dri/i915/i830_reg.h
@@ -407,7 +407,7 @@
 #define LOGICOP_SET			0xf
 #define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
 #define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		((x)<<8)
+#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
 #define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
 #define ENABLE_STENCIL_WRITE_MASK	(1<<16)
 #define STENCIL_WRITE_MASK(x)		((x)&0xff)
@@ -554,8 +554,8 @@
 #define    MAPSURF_4BIT_INDEXED		   (7<<6)
 #define TM0S1_MT_FORMAT_MASK         (0x7 << 3)
 #define TM0S1_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3) /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_IDX_RGB565	           (0<<3) /* SURFACE_8BIT_INDEXED */
+#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
+#define    MT_8BIT_IDX_RGB565	           (0<<3)       /* SURFACE_8BIT_INDEXED */
 #define    MT_8BIT_IDX_ARGB1555	           (1<<3)
 #define    MT_8BIT_IDX_ARGB4444	           (2<<3)
 #define    MT_8BIT_IDX_AY88		   (3<<3)
@@ -563,9 +563,9 @@
 #define    MT_8BIT_IDX_BUMP_88DVDU 	   (5<<3)
 #define    MT_8BIT_IDX_BUMP_655LDVDU	   (6<<3)
 #define    MT_8BIT_IDX_ARGB8888	           (7<<3)
-#define    MT_8BIT_I8		           (0<<3) /* SURFACE_8BIT */
+#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
 #define    MT_8BIT_L8		           (1<<3)
-#define    MT_16BIT_RGB565 		   (0<<3) /* SURFACE_16BIT */
+#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
 #define    MT_16BIT_ARGB1555		   (1<<3)
 #define    MT_16BIT_ARGB4444		   (2<<3)
 #define    MT_16BIT_AY88		   (3<<3)
@@ -573,16 +573,17 @@
 #define    MT_16BIT_BUMP_88DVDU	           (5<<3)
 #define    MT_16BIT_BUMP_655LDVDU	   (6<<3)
 #define    MT_16BIT_DIB_RGB565_8888	   (7<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3) /* SURFACE_32BIT */
+#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
 #define    MT_32BIT_ABGR8888		   (1<<3)
+#define    MT_32BIT_XRGB8888		   (2<<3)       /* XXX: Guess from i915_reg.h */
 #define    MT_32BIT_BUMP_XLDVDU_8888	   (6<<3)
 #define    MT_32BIT_DIB_8888		   (7<<3)
-#define    MT_411_YUV411		   (0<<3) /* SURFACE_411 */
-#define    MT_422_YCRCB_SWAPY	           (0<<3) /* SURFACE_422 */
+#define    MT_411_YUV411		   (0<<3)       /* SURFACE_411 */
+#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
 #define    MT_422_YCRCB_NORMAL	           (1<<3)
 #define    MT_422_YCRCB_SWAPUV	           (2<<3)
 #define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3) /* SURFACE_COMPRESSED */
+#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
 #define    MT_COMPRESS_DXT2_3	           (1<<3)
 #define    MT_COMPRESS_DXT4_5	           (2<<3)
 #define    MT_COMPRESS_FXT1		   (3<<3)
diff --git a/src/mesa/drivers/dri/i915/i830_state.c b/src/mesa/drivers/dri/i915/i830_state.c
index f7980201f9a..3c149e69055 100644
--- a/src/mesa/drivers/dri/i915/i830_state.c
+++ b/src/mesa/drivers/dri/i915/i830_state.c
@@ -38,149 +38,151 @@
 
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
+#include "intel_fbo.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
 
+#define FILE_DEBUG_FLAG DEBUG_STATE
+
 static void
-i830StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func, GLint ref,
+i830StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
                         GLuint mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
 
    mask = mask & 0xff;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(func), ref, mask);
+   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(func), ref, mask);
 
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					STENCIL_TEST_MASK(mask));
+                                           STENCIL_TEST_MASK(mask));
    i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-					     ENABLE_STENCIL_TEST_FUNC_MASK);
+                                                ENABLE_STENCIL_TEST_FUNC_MASK);
    i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_REF_VALUE |
-					    ENABLE_STENCIL_TEST_FUNC |
-					    STENCIL_REF_VALUE(ref) |
-					    STENCIL_TEST_FUNC(test));
+                                               ENABLE_STENCIL_TEST_FUNC |
+                                               STENCIL_REF_VALUE(ref) |
+                                               STENCIL_TEST_FUNC(test));
 }
 
 static void
-i830StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+i830StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : mask 0x%x\n", __FUNCTION__, mask);
+   struct i830_context *i830 = i830_context(ctx);
 
+   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
+   
    mask = mask & 0xff;
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(mask));
+                                           STENCIL_WRITE_MASK(mask));
 }
 
 static void
-i830StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
+i830StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
                       GLenum zpass)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int fop, dfop, dpop;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(fail),
-	      _mesa_lookup_enum_by_nr(zfail),
-	      _mesa_lookup_enum_by_nr(zpass));
+   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(fail),
+       _mesa_lookup_enum_by_nr(zfail), 
+       _mesa_lookup_enum_by_nr(zpass));
 
-   fop = 0; dfop = 0; dpop = 0;
+   fop = 0;
+   dfop = 0;
+   dpop = 0;
 
-   switch(fail) {
-   case GL_KEEP: 
-      fop = STENCILOP_KEEP; 
+   switch (fail) {
+   case GL_KEEP:
+      fop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      fop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      fop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      fop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      fop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       fop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       fop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      fop = STENCILOP_INCR; 
+      fop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      fop = STENCILOP_DECR; 
+      fop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      fop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      fop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
-   switch(zfail) {
-   case GL_KEEP: 
-      dfop = STENCILOP_KEEP; 
+   switch (zfail) {
+   case GL_KEEP:
+      dfop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      dfop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      dfop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      dfop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      dfop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       dfop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       dfop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      dfop = STENCILOP_INCR; 
+      dfop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      dfop = STENCILOP_DECR; 
+      dfop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      dfop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      dfop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
-   switch(zpass) {
-   case GL_KEEP: 
-      dpop = STENCILOP_KEEP; 
+   switch (zpass) {
+   case GL_KEEP:
+      dpop = STENCILOP_KEEP;
       break;
-   case GL_ZERO: 
-      dpop = STENCILOP_ZERO; 
+   case GL_ZERO:
+      dpop = STENCILOP_ZERO;
       break;
-   case GL_REPLACE: 
-      dpop = STENCILOP_REPLACE; 
+   case GL_REPLACE:
+      dpop = STENCILOP_REPLACE;
       break;
-   case GL_INCR: 
+   case GL_INCR:
       dpop = STENCILOP_INCRSAT;
       break;
-   case GL_DECR: 
+   case GL_DECR:
       dpop = STENCILOP_DECRSAT;
       break;
    case GL_INCR_WRAP:
-      dpop = STENCILOP_INCR; 
+      dpop = STENCILOP_INCR;
       break;
    case GL_DECR_WRAP:
-      dpop = STENCILOP_DECR; 
+      dpop = STENCILOP_DECR;
       break;
-   case GL_INVERT: 
-      dpop = STENCILOP_INVERT; 
+   case GL_INVERT:
+      dpop = STENCILOP_INVERT;
       break;
-   default: 
+   default:
       break;
    }
 
@@ -188,27 +190,30 @@ i830StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
    i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_PARMS |
-					    STENCIL_FAIL_OP(fop) |
-					    STENCIL_PASS_DEPTH_FAIL_OP(dfop) |
-					    STENCIL_PASS_DEPTH_PASS_OP(dpop));
+                                               STENCIL_FAIL_OP(fop) |
+                                               STENCIL_PASS_DEPTH_FAIL_OP
+                                               (dfop) |
+                                               STENCIL_PASS_DEPTH_PASS_OP
+                                               (dpop));
 }
 
-static void i830AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+static void
+i830AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
    GLubyte refByte;
    GLuint refInt;
 
    UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
-   refInt = (GLuint)refByte;
+   refInt = (GLuint) refByte;
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE2] &= ~ALPHA_TEST_REF_MASK;
    i830->state.Ctx[I830_CTXREG_STATE2] |= (ENABLE_ALPHA_TEST_FUNC |
-					ENABLE_ALPHA_REF_VALUE |
-					ALPHA_TEST_FUNC(test) |
-					ALPHA_REF_VALUE(refInt));
+                                           ENABLE_ALPHA_REF_VALUE |
+                                           ALPHA_TEST_FUNC(test) |
+                                           ALPHA_REF_VALUE(refInt));
 }
 
 /**
@@ -221,45 +226,49 @@ static void i830AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
  * This function is substantially different from the old i830-specific driver.
  * I'm not sure which is correct.
  */
-static void i830EvalLogicOpBlendState(GLcontext *ctx)
+static void
+i830EvalLogicOpBlendState(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
    if (RGBA_LOGICOP_ENABLED(ctx)) {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-					      ENABLE_LOGIC_OP);
-   } else if (ctx->Color.BlendEnabled) {
+                                                 ENABLE_LOGIC_OP);
+   }
+   else if (ctx->Color.BlendEnabled) {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (ENABLE_COLOR_BLEND |
-					      DISABLE_LOGIC_OP);
-   } else {
+                                                 DISABLE_LOGIC_OP);
+   }
+   else {
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-					       ENABLE_LOGIC_OP_MASK);
+                                                  ENABLE_LOGIC_OP_MASK);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-					      DISABLE_LOGIC_OP);
+                                                 DISABLE_LOGIC_OP);
    }
 }
 
-static void i830BlendColor(GLcontext *ctx, const GLfloat color[4])
+static void
+i830BlendColor(GLcontext * ctx, const GLfloat color[4])
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    GLubyte r, g, b, a;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] = (a<<24) | (r<<16) | (g<<8) | b;
+   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] =
+      (a << 24) | (r << 16) | (g << 8) | b;
 }
 
 /**
@@ -268,9 +277,10 @@ static void i830BlendColor(GLcontext *ctx, const GLfloat color[4])
  * function because some blend equations (i.e., \c GL_MIN and \c GL_MAX)
  * change the interpretation of the blend function.
  */
-static void i830_set_blend_state( GLcontext * ctx )
+static void
+i830_set_blend_state(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int funcA;
    int funcRGB;
    int eqnA;
@@ -279,71 +289,72 @@ static void i830_set_blend_state( GLcontext * ctx )
    int s1;
 
 
-   funcRGB = SRC_BLND_FACT( intel_translate_blend_factor( ctx->Color.BlendSrcRGB ) )
-       | DST_BLND_FACT( intel_translate_blend_factor( ctx->Color.BlendDstRGB ) );
+   funcRGB =
+      SRC_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcRGB))
+      | DST_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstRGB));
 
-   switch(ctx->Color.BlendEquationRGB) {
+   switch (ctx->Color.BlendEquationRGB) {
    case GL_FUNC_ADD:
-      eqnRGB = BLENDFUNC_ADD; 
+      eqnRGB = BLENDFUNC_ADD;
       break;
    case GL_MIN:
       eqnRGB = BLENDFUNC_MIN;
       funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_MAX: 
+   case GL_MAX:
       eqnRGB = BLENDFUNC_MAX;
       funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_FUNC_SUBTRACT: 
-      eqnRGB = BLENDFUNC_SUB; 
+   case GL_FUNC_SUBTRACT:
+      eqnRGB = BLENDFUNC_SUB;
       break;
    case GL_FUNC_REVERSE_SUBTRACT:
-      eqnRGB = BLENDFUNC_RVRSE_SUB; 
+      eqnRGB = BLENDFUNC_RVRSE_SUB;
       break;
    default:
-      fprintf( stderr, "[%s:%u] Invalid RGB blend equation (0x%04x).\n",
-	       __FUNCTION__, __LINE__, ctx->Color.BlendEquationRGB );
+      fprintf(stderr, "[%s:%u] Invalid RGB blend equation (0x%04x).\n",
+              __FUNCTION__, __LINE__, ctx->Color.BlendEquationRGB);
       return;
    }
 
 
-   funcA = SRC_ABLEND_FACT( intel_translate_blend_factor( ctx->Color.BlendSrcA ) )
-       | DST_ABLEND_FACT( intel_translate_blend_factor( ctx->Color.BlendDstA ) );
+   funcA = SRC_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcA))
+      | DST_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstA));
 
-   switch(ctx->Color.BlendEquationA) {
+   switch (ctx->Color.BlendEquationA) {
    case GL_FUNC_ADD:
-      eqnA = BLENDFUNC_ADD; 
+      eqnA = BLENDFUNC_ADD;
       break;
-   case GL_MIN: 
+   case GL_MIN:
       eqnA = BLENDFUNC_MIN;
       funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_MAX: 
+   case GL_MAX:
       eqnA = BLENDFUNC_MAX;
       funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
       break;
-   case GL_FUNC_SUBTRACT: 
-      eqnA = BLENDFUNC_SUB; 
+   case GL_FUNC_SUBTRACT:
+      eqnA = BLENDFUNC_SUB;
       break;
    case GL_FUNC_REVERSE_SUBTRACT:
-      eqnA = BLENDFUNC_RVRSE_SUB; 
+      eqnA = BLENDFUNC_RVRSE_SUB;
       break;
    default:
-      fprintf( stderr, "[%s:%u] Invalid alpha blend equation (0x%04x).\n",
-	       __FUNCTION__, __LINE__, ctx->Color.BlendEquationA );
+      fprintf(stderr, "[%s:%u] Invalid alpha blend equation (0x%04x).\n",
+              __FUNCTION__, __LINE__, ctx->Color.BlendEquationA);
       return;
    }
 
    iab = eqnA | funcA
-       | _3DSTATE_INDPT_ALPHA_BLEND_CMD
-       | ENABLE_SRC_ABLEND_FACTOR | ENABLE_DST_ABLEND_FACTOR
-       | ENABLE_ALPHA_BLENDFUNC;
+      | _3DSTATE_INDPT_ALPHA_BLEND_CMD
+      | ENABLE_SRC_ABLEND_FACTOR | ENABLE_DST_ABLEND_FACTOR
+      | ENABLE_ALPHA_BLENDFUNC;
    s1 = eqnRGB | funcRGB
-       | _3DSTATE_MODES_1_CMD
-       | ENABLE_SRC_BLND_FACTOR | ENABLE_DST_BLND_FACTOR
-       | ENABLE_COLR_BLND_FUNC;
+      | _3DSTATE_MODES_1_CMD
+      | ENABLE_SRC_BLND_FACTOR | ENABLE_DST_BLND_FACTOR
+      | ENABLE_COLR_BLND_FUNC;
 
-   if ( (eqnA | funcA) != (eqnRGB | funcRGB) )
+   if ((eqnA | funcA) != (eqnRGB | funcRGB))
       iab |= ENABLE_INDPT_ALPHA_BLEND;
    else
       iab |= DISABLE_INDPT_ALPHA_BLEND;
@@ -363,70 +374,68 @@ static void i830_set_blend_state( GLcontext * ctx )
    i830EvalLogicOpBlendState(ctx);
 
    if (0) {
-      fprintf(stderr, "[%s:%u] STATE1: 0x%08x IALPHAB: 0x%08x blend is %sabled\n",
-	      __FUNCTION__, __LINE__,
-	      i830->state.Ctx[I830_CTXREG_STATE1],
-	      i830->state.Ctx[I830_CTXREG_IALPHAB],
-	      (ctx->Color.BlendEnabled) ? "en" : "dis");
+      fprintf(stderr,
+              "[%s:%u] STATE1: 0x%08x IALPHAB: 0x%08x blend is %sabled\n",
+              __FUNCTION__, __LINE__, i830->state.Ctx[I830_CTXREG_STATE1],
+              i830->state.Ctx[I830_CTXREG_IALPHAB],
+              (ctx->Color.BlendEnabled) ? "en" : "dis");
    }
 }
 
 
-static void i830BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB,
-				      GLenum modeA) 
+static void
+i830BlendEquationSeparate(GLcontext * ctx, GLenum modeRGB, GLenum modeA)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s -> %s, %s\n", __FUNCTION__,
-	     _mesa_lookup_enum_by_nr(modeRGB),
-	     _mesa_lookup_enum_by_nr(modeA));
+   DBG("%s -> %s, %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(modeRGB),
+       _mesa_lookup_enum_by_nr(modeA));
 
    (void) modeRGB;
    (void) modeA;
-   i830_set_blend_state( ctx );
+   i830_set_blend_state(ctx);
 }
 
 
-static void i830BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, 
-				  GLenum dfactorRGB, GLenum sfactorA,
-				  GLenum dfactorA )
+static void
+i830BlendFuncSeparate(GLcontext * ctx, GLenum sfactorRGB,
+                      GLenum dfactorRGB, GLenum sfactorA, GLenum dfactorA)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s -> RGB(%s, %s) A(%s, %s)\n", __FUNCTION__,
-	     _mesa_lookup_enum_by_nr(sfactorRGB),
-	     _mesa_lookup_enum_by_nr(dfactorRGB),
-	     _mesa_lookup_enum_by_nr(sfactorA),
-	     _mesa_lookup_enum_by_nr(dfactorA));
+   DBG("%s -> RGB(%s, %s) A(%s, %s)\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(sfactorRGB),
+       _mesa_lookup_enum_by_nr(dfactorRGB),
+       _mesa_lookup_enum_by_nr(sfactorA),
+       _mesa_lookup_enum_by_nr(dfactorA));
 
    (void) sfactorRGB;
    (void) dfactorRGB;
    (void) sfactorA;
    (void) dfactorA;
-   i830_set_blend_state( ctx );
+   i830_set_blend_state(ctx);
 }
 
 
 
-static void i830DepthFunc(GLcontext *ctx, GLenum func)
+static void
+i830DepthFunc(GLcontext * ctx, GLenum func)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    int test = intel_translate_compare_func(func);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
    i830->state.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
-				       DEPTH_TEST_FUNC(test));
+                                           DEPTH_TEST_FUNC(test));
 }
 
-static void i830DepthMask(GLcontext *ctx, GLboolean flag)
+static void
+i830DepthMask(GLcontext * ctx, GLboolean flag)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s flag (%d)\n", __FUNCTION__, flag);
+   struct i830_context *i830 = i830_context(ctx);
 
+   DBG("%s flag (%d)\n", __FUNCTION__, flag);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
    i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
@@ -443,14 +452,15 @@ static void i830DepthMask(GLcontext *ctx, GLboolean flag)
  * The i830 supports a 4x4 stipple natively, GL wants 32x32.
  * Fortunately stipple is usually a repeating pattern.
  */
-static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
+static void
+i830PolygonStipple(GLcontext * ctx, const GLubyte * mask)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    const GLubyte *m = mask;
    GLubyte p[4];
-   int i,j,k;
+   int i, j, k;
    int active = (ctx->Polygon.StippleFlag &&
-		 i830->intel.reduced_primitive == GL_TRIANGLES);
+                 i830->intel.reduced_primitive == GL_TRIANGLES);
    GLuint newMask;
 
    if (active) {
@@ -458,23 +468,26 @@ static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
       i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
    }
 
-   p[0] = mask[12] & 0xf; p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf; p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf; p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf; p[3] |= p[3] << 4;
-
-   for (k = 0 ; k < 8 ; k++)
-      for (j = 3 ; j >= 0; j--)
-	 for (i = 0 ; i < 4 ; i++, m++)
-	    if (*m != p[j]) {
-	       i830->intel.hw_stipple = 0;
-	       return;
-	    }
+   p[0] = mask[12] & 0xf;
+   p[0] |= p[0] << 4;
+   p[1] = mask[8] & 0xf;
+   p[1] |= p[1] << 4;
+   p[2] = mask[4] & 0xf;
+   p[2] |= p[2] << 4;
+   p[3] = mask[0] & 0xf;
+   p[3] |= p[3] << 4;
+
+   for (k = 0; k < 8; k++)
+      for (j = 3; j >= 0; j--)
+         for (i = 0; i < 4; i++, m++)
+            if (*m != p[j]) {
+               i830->intel.hw_stipple = 0;
+               return;
+            }
 
    newMask = (((p[0] & 0xf) << 0) |
-	      ((p[1] & 0xf) << 4) |
-	      ((p[2] & 0xf) << 8) |
-	      ((p[3] & 0xf) << 12));
+              ((p[1] & 0xf) << 4) |
+              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
 
 
    if (newMask == 0xffff || newMask == 0x0) {
@@ -495,49 +508,54 @@ static void i830PolygonStipple( GLcontext *ctx, const GLubyte *mask )
 /* =============================================================
  * Hardware clipping
  */
-static void i830Scissor(GLcontext *ctx, GLint x, GLint y, 
-			GLsizei w, GLsizei h)
+static void
+i830Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   intelScreenPrivate *screen = i830->intel.intelScreen;
+   struct i830_context *i830 = i830_context(ctx);
    int x1, y1, x2, y2;
 
-   if (!i830->intel.driDrawable)
+   if (!ctx->DrawBuffer)
       return;
 
-   x1 = x;
-   y1 = i830->intel.driDrawable->h - (y + h);
-   x2 = x + w - 1;
-   y2 = y1 + h - 1;
+   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "[%s] x(%d) y(%d) w(%d) h(%d)\n", __FUNCTION__,
-	      x, y, w, h);
-
-   if (x1 < 0) x1 = 0;
-   if (y1 < 0) y1 = 0;
-   if (x2 < 0) x2 = 0;
-   if (y2 < 0) y2 = 0;
-
-   if (x2 >= screen->width) x2 = screen->width-1;
-   if (y2 >= screen->height) y2 = screen->height-1;
-   if (x1 >= screen->width) x1 = screen->width-1;
-   if (y1 >= screen->height) y1 = screen->height-1;
+   if (ctx->DrawBuffer->Name == 0) {
+      x1 = x;
+      y1 = ctx->DrawBuffer->Height - (y + h);
+      x2 = x + w - 1;
+      y2 = y1 + h - 1;
+      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   else {
+      /* FBO - not inverted
+       */
+      x1 = x;
+      y1 = y;
+      x2 = x + w - 1;
+      y2 = y + h - 1;
+      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
 
+   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
+   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
+   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
+   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
+   
+   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
 
    I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
    i830->state.Buffer[I830_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
    i830->state.Buffer[I830_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
 }
 
-static void i830LogicOp(GLcontext *ctx, GLenum opcode)
+static void
+i830LogicOp(GLcontext * ctx, GLenum opcode)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   int tmp = intel_translate_logic_op( opcode );
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i830_context *i830 = i830_context(ctx);
+   int tmp = intel_translate_logic_op(opcode);
 
+   DBG("%s\n", __FUNCTION__);
+   
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE4] &= ~LOGICOP_MASK;
    i830->state.Ctx[I830_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
@@ -545,14 +563,14 @@ static void i830LogicOp(GLcontext *ctx, GLenum opcode)
 
 
 
-static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
+static void
+i830CullFaceFrontFace(GLcontext * ctx, GLenum unused)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    GLuint mode;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (!ctx->Polygon.CullFlag) {
       mode = CULLMODE_NONE;
    }
@@ -560,9 +578,9 @@ static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
       mode = CULLMODE_CW;
 
       if (ctx->Polygon.CullFaceMode == GL_FRONT)
-	 mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
+         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
       if (ctx->Polygon.FrontFace != GL_CCW)
-	 mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
+         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
    }
    else {
       mode = CULLMODE_BOTH;
@@ -573,18 +591,18 @@ static void i830CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    i830->state.Ctx[I830_CTXREG_STATE3] |= ENABLE_CULL_MODE | mode;
 }
 
-static void i830LineWidth( GLcontext *ctx, GLfloat widthf )
+static void
+i830LineWidth(GLcontext * ctx, GLfloat widthf)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
    int width;
    int state5;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   width = (int)(widthf * 2);
-   CLAMP_SELF(width, 1, 15);
+   DBG("%s\n", __FUNCTION__);
    
+   width = (int) (widthf * 2);
+   CLAMP_SELF(width, 1, 15);
+
    state5 = i830->state.Ctx[I830_CTXREG_STATE5] & ~FIXED_LINE_WIDTH_MASK;
    state5 |= (ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(width));
 
@@ -594,19 +612,19 @@ static void i830LineWidth( GLcontext *ctx, GLfloat widthf )
    }
 }
 
-static void i830PointSize(GLcontext *ctx, GLfloat size)
+static void
+i830PointSize(GLcontext * ctx, GLfloat size)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLint point_size = (int)size;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i830_context *i830 = i830_context(ctx);
+   GLint point_size = (int) size;
 
+   DBG("%s\n", __FUNCTION__);
+   
    CLAMP_SELF(point_size, 1, 256);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_STATE5] &= ~FIXED_POINT_WIDTH_MASK;
    i830->state.Ctx[I830_CTXREG_STATE5] |= (ENABLE_FIXED_POINT_WIDTH |
-				       FIXED_POINT_WIDTH(point_size));
+                                           FIXED_POINT_WIDTH(point_size));
 }
 
 
@@ -614,23 +632,21 @@ static void i830PointSize(GLcontext *ctx, GLfloat size)
  * Color masks
  */
 
-static void i830ColorMask(GLcontext *ctx,
-			  GLboolean r, GLboolean g,
-			  GLboolean b, GLboolean a)
+static void
+i830ColorMask(GLcontext * ctx,
+              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
    GLuint tmp = 0;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
+   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
 
    tmp = ((i830->state.Ctx[I830_CTXREG_ENABLES_2] & ~WRITEMASK_MASK) |
-	  ENABLE_COLOR_MASK |
-	  ENABLE_COLOR_WRITE |
-	  ((!r) << WRITEMASK_RED_SHIFT) |
-	  ((!g) << WRITEMASK_GREEN_SHIFT) |
-	  ((!b) << WRITEMASK_BLUE_SHIFT) |
-	  ((!a) << WRITEMASK_ALPHA_SHIFT));
+          ENABLE_COLOR_MASK |
+          ENABLE_COLOR_WRITE |
+          ((!r) << WRITEMASK_RED_SHIFT) |
+          ((!g) << WRITEMASK_GREEN_SHIFT) |
+          ((!b) << WRITEMASK_BLUE_SHIFT) | ((!a) << WRITEMASK_ALPHA_SHIFT));
 
    if (tmp != i830->state.Ctx[I830_CTXREG_ENABLES_2]) {
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
@@ -638,9 +654,10 @@ static void i830ColorMask(GLcontext *ctx,
    }
 }
 
-static void update_specular( GLcontext *ctx )
+static void
+update_specular(GLcontext * ctx)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
+   struct i830_context *i830 = i830_context(ctx);
 
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
    i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_SPEC_ADD_MASK;
@@ -651,22 +668,22 @@ static void update_specular( GLcontext *ctx )
       i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_SPEC_ADD;
 }
 
-static void i830LightModelfv(GLcontext *ctx, GLenum pname, 
-			     const GLfloat *param)
+static void
+i830LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular( ctx );
+      update_specular(ctx);
    }
 }
 
 /* In Mesa 3.5 we can reliably do native flatshading.
  */
-static void i830ShadeModel(GLcontext *ctx, GLenum mode)
+static void
+i830ShadeModel(GLcontext * ctx, GLenum mode)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
    I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
 
@@ -675,58 +692,62 @@ static void i830ShadeModel(GLcontext *ctx, GLenum mode)
    i830->state.Ctx[I830_CTXREG_STATE3] &= ~SHADE_MODE_MASK;
 
    if (mode == GL_FLAT) {
-     i830->state.Ctx[I830_CTXREG_STATE3] |= (ALPHA_SHADE_MODE(SHADE_MODE_FLAT) |
-					  FOG_SHADE_MODE(SHADE_MODE_FLAT) |
-					  SPEC_SHADE_MODE(SHADE_MODE_FLAT) |
-					  COLOR_SHADE_MODE(SHADE_MODE_FLAT));
-   } else {
-     i830->state.Ctx[I830_CTXREG_STATE3] |= (ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  COLOR_SHADE_MODE(SHADE_MODE_LINEAR));
+      i830->state.Ctx[I830_CTXREG_STATE3] |=
+         (ALPHA_SHADE_MODE(SHADE_MODE_FLAT) | FOG_SHADE_MODE(SHADE_MODE_FLAT)
+          | SPEC_SHADE_MODE(SHADE_MODE_FLAT) |
+          COLOR_SHADE_MODE(SHADE_MODE_FLAT));
+   }
+   else {
+      i830->state.Ctx[I830_CTXREG_STATE3] |=
+         (ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
+          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
+          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
+          COLOR_SHADE_MODE(SHADE_MODE_LINEAR));
    }
 }
 
 /* =============================================================
  * Fog
  */
-static void i830Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
+static void
+i830Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (pname == GL_FOG_COLOR) {      
-      GLuint color = (((GLubyte)(ctx->Fog.Color[0]*255.0F) << 16) |
-		      ((GLubyte)(ctx->Fog.Color[1]*255.0F) << 8) |
-		      ((GLubyte)(ctx->Fog.Color[2]*255.0F) << 0));
+   DBG("%s\n", __FUNCTION__);
+   
+   if (pname == GL_FOG_COLOR) {
+      GLuint color = (((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
+                      ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
+                      ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
 
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_FOGCOLOR] = (_3DSTATE_FOG_COLOR_CMD | color);
+      i830->state.Ctx[I830_CTXREG_FOGCOLOR] =
+         (_3DSTATE_FOG_COLOR_CMD | color);
    }
 }
 
 /* =============================================================
  */
 
-static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+static void
+i830Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 {
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   struct i830_context *i830 = i830_context(ctx);
 
-   switch(cap) {
+   switch (cap) {
    case GL_LIGHTING:
    case GL_COLOR_SUM:
-      update_specular( ctx );
+      update_specular(ctx);
       break;
 
    case GL_ALPHA_TEST:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_ALPHA_TEST_MASK;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_ALPHA_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_ALPHA_TEST;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_ALPHA_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_ALPHA_TEST;
 
       break;
 
@@ -740,17 +761,17 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       /* Logicop doesn't seem to work at 16bpp:
        */
       if (i830->intel.intelScreen->cpp == 2)
-	 FALLBACK( &i830->intel, I830_FALLBACK_LOGICOP, state );
+         FALLBACK(&i830->intel, I830_FALLBACK_LOGICOP, state);
       break;
- 
+
    case GL_DITHER:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DITHER;
 
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DITHER;
+         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DITHER;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DITHER;
+         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DITHER;
       break;
 
    case GL_DEPTH_TEST:
@@ -758,46 +779,44 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
 
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
 
       /* Also turn off depth writes when GL_DEPTH_TEST is disabled:
        */
-      i830DepthMask( ctx, ctx->Depth.Mask );
+      i830DepthMask(ctx, ctx->Depth.Mask);
       break;
 
    case GL_SCISSOR_TEST:
       I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
-      
+
       if (state)
-	 i830->state.Buffer[I830_DESTREG_SENABLE] = 
-	    (_3DSTATE_SCISSOR_ENABLE_CMD |
-	     ENABLE_SCISSOR_RECT);
+         i830->state.Buffer[I830_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
       else
-	 i830->state.Buffer[I830_DESTREG_SENABLE] = 
-	    (_3DSTATE_SCISSOR_ENABLE_CMD |
-	     DISABLE_SCISSOR_RECT);
+         i830->state.Buffer[I830_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
 
       break;
 
    case GL_LINE_SMOOTH:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      
+
       i830->state.Ctx[I830_CTXREG_AA] &= ~AA_LINE_ENABLE;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_ENABLE;
+         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_ENABLE;
       else
-	 i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_DISABLE;
+         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_DISABLE;
       break;
 
    case GL_FOG:
       I830_STATECHANGE(i830, I830_UPLOAD_CTX);
       i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_FOG_MASK;
       if (state)
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_FOG;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_FOG;
       else
-	 i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_FOG;
+         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_FOG;
       break;
 
    case GL_CULL_FACE:
@@ -808,20 +827,32 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       break;
 
    case GL_STENCIL_TEST:
-      if (i830->intel.hw_stencil) {
-	 I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-	 if (state) {
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-	 } else {
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_STENCIL_WRITE;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
-	    i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_STENCIL_WRITE;
-	 }
-      } else {
-	 FALLBACK( &i830->intel, I830_FALLBACK_STENCIL, state );
+      {
+         GLboolean hw_stencil = GL_FALSE;
+         if (ctx->DrawBuffer) {
+            struct intel_renderbuffer *irbStencil
+               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
+            hw_stencil = (irbStencil && irbStencil->region);
+         }
+         if (hw_stencil) {
+            I830_STATECHANGE(i830, I830_UPLOAD_CTX);
+
+            if (state) {
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
+            }
+            else {
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] &=
+                  ~ENABLE_STENCIL_WRITE;
+               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
+               i830->state.Ctx[I830_CTXREG_ENABLES_2] |=
+                  DISABLE_STENCIL_WRITE;
+            }
+         }
+         else {
+            FALLBACK(&i830->intel, I830_FALLBACK_STENCIL, state);
+         }
       }
       break;
 
@@ -830,13 +861,12 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
        * I'll do more testing later to find out exactly which hardware
        * supports it.  Disabled for now.
        */
-      if (i830->intel.hw_stipple && 
-	  i830->intel.reduced_primitive == GL_TRIANGLES)
-      {
-	 I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-	 i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
-	 if (state)
-	    i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
+      if (i830->intel.hw_stipple &&
+          i830->intel.reduced_primitive == GL_TRIANGLES) {
+         I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
+         i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
+         if (state)
+            i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
       }
       break;
 
@@ -846,7 +876,8 @@ static void i830Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 }
 
 
-static void i830_init_packets( i830ContextPtr i830 )
+static void
+i830_init_packets(struct i830_context *i830)
 {
    intelScreenPrivate *screen = i830->intel.intelScreen;
 
@@ -855,197 +886,192 @@ static void i830_init_packets( i830ContextPtr i830 )
 
    /* Set default blend state */
    i830->state.TexBlend[0][0] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-				  TEXPIPE_COLOR |
-				  ENABLE_TEXOUTPUT_WRT_SEL |
-				  TEXOP_OUTPUT_CURRENT |
-				  DISABLE_TEX_CNTRL_STAGE |
-				  TEXOP_SCALE_1X |
-				  TEXOP_MODIFY_PARMS |
-				  TEXOP_LAST_STAGE |
-				  TEXBLENDOP_ARG1);
+                                 TEXPIPE_COLOR |
+                                 ENABLE_TEXOUTPUT_WRT_SEL |
+                                 TEXOP_OUTPUT_CURRENT |
+                                 DISABLE_TEX_CNTRL_STAGE |
+                                 TEXOP_SCALE_1X |
+                                 TEXOP_MODIFY_PARMS |
+                                 TEXOP_LAST_STAGE | TEXBLENDOP_ARG1);
    i830->state.TexBlend[0][1] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-				  TEXPIPE_ALPHA |
-				  ENABLE_TEXOUTPUT_WRT_SEL |
-				  TEXOP_OUTPUT_CURRENT |
-				  TEXOP_SCALE_1X |
-				  TEXOP_MODIFY_PARMS |
-				  TEXBLENDOP_ARG1);
+                                 TEXPIPE_ALPHA |
+                                 ENABLE_TEXOUTPUT_WRT_SEL |
+                                 TEXOP_OUTPUT_CURRENT |
+                                 TEXOP_SCALE_1X |
+                                 TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    i830->state.TexBlend[0][2] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-				  TEXPIPE_COLOR |
-				  TEXBLEND_ARG1 |
-				  TEXBLENDARG_MODIFY_PARMS |
-				  TEXBLENDARG_DIFFUSE);
+                                 TEXPIPE_COLOR |
+                                 TEXBLEND_ARG1 |
+                                 TEXBLENDARG_MODIFY_PARMS |
+                                 TEXBLENDARG_DIFFUSE);
    i830->state.TexBlend[0][3] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-				  TEXPIPE_ALPHA |
-				  TEXBLEND_ARG1 |
-				  TEXBLENDARG_MODIFY_PARMS |
-				  TEXBLENDARG_DIFFUSE);
+                                 TEXPIPE_ALPHA |
+                                 TEXBLEND_ARG1 |
+                                 TEXBLENDARG_MODIFY_PARMS |
+                                 TEXBLENDARG_DIFFUSE);
 
    i830->state.TexBlendWordsUsed[0] = 4;
 
 
-   i830->state.Ctx[I830_CTXREG_VF] =  0;
+   i830->state.Ctx[I830_CTXREG_VF] = 0;
    i830->state.Ctx[I830_CTXREG_VF2] = 0;
 
    i830->state.Ctx[I830_CTXREG_AA] = (_3DSTATE_AA_CMD |
-				      AA_LINE_ECAAR_WIDTH_ENABLE |
-				      AA_LINE_ECAAR_WIDTH_1_0 |
-				      AA_LINE_REGION_WIDTH_ENABLE |
-				      AA_LINE_REGION_WIDTH_1_0 | 
-				      AA_LINE_DISABLE);
+                                      AA_LINE_ECAAR_WIDTH_ENABLE |
+                                      AA_LINE_ECAAR_WIDTH_1_0 |
+                                      AA_LINE_REGION_WIDTH_ENABLE |
+                                      AA_LINE_REGION_WIDTH_1_0 |
+                                      AA_LINE_DISABLE);
 
    i830->state.Ctx[I830_CTXREG_ENABLES_1] = (_3DSTATE_ENABLES_1_CMD |
-					     DISABLE_LOGIC_OP |
-					     DISABLE_STENCIL_TEST |
-					     DISABLE_DEPTH_BIAS |
-					     DISABLE_SPEC_ADD |
-					     DISABLE_FOG |
-					     DISABLE_ALPHA_TEST |
-					     DISABLE_COLOR_BLEND |
-					     DISABLE_DEPTH_TEST);
-
+                                             DISABLE_LOGIC_OP |
+                                             DISABLE_STENCIL_TEST |
+                                             DISABLE_DEPTH_BIAS |
+                                             DISABLE_SPEC_ADD |
+                                             DISABLE_FOG |
+                                             DISABLE_ALPHA_TEST |
+                                             DISABLE_COLOR_BLEND |
+                                             DISABLE_DEPTH_TEST);
+
+#if 000                         /* XXX all the stencil enable state is set in i830Enable(), right? */
    if (i830->intel.hw_stencil) {
       i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-						ENABLE_STENCIL_WRITE |
-						ENABLE_TEX_CACHE |
-						ENABLE_DITHER |
-						ENABLE_COLOR_MASK |
-						/* set no color comps disabled */
-						ENABLE_COLOR_WRITE |
-						ENABLE_DEPTH_WRITE);
-   } else {
+                                                ENABLE_STENCIL_WRITE |
+                                                ENABLE_TEX_CACHE |
+                                                ENABLE_DITHER |
+                                                ENABLE_COLOR_MASK |
+                                                /* set no color comps disabled */
+                                                ENABLE_COLOR_WRITE |
+                                                ENABLE_DEPTH_WRITE);
+   }
+   else
+#endif
+   {
       i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-						DISABLE_STENCIL_WRITE |
-						ENABLE_TEX_CACHE |
-						ENABLE_DITHER |
-						ENABLE_COLOR_MASK |
-						/* set no color comps disabled */
-						ENABLE_COLOR_WRITE |
-						ENABLE_DEPTH_WRITE);
+                                                DISABLE_STENCIL_WRITE |
+                                                ENABLE_TEX_CACHE |
+                                                ENABLE_DITHER |
+                                                ENABLE_COLOR_MASK |
+                                                /* set no color comps disabled */
+                                                ENABLE_COLOR_WRITE |
+                                                ENABLE_DEPTH_WRITE);
    }
 
    i830->state.Ctx[I830_CTXREG_STATE1] = (_3DSTATE_MODES_1_CMD |
-					  ENABLE_COLR_BLND_FUNC |
-					  BLENDFUNC_ADD |
-					  ENABLE_SRC_BLND_FACTOR |
-					  SRC_BLND_FACT(BLENDFACT_ONE) | 
-					  ENABLE_DST_BLND_FACTOR |
-					  DST_BLND_FACT(BLENDFACT_ZERO) );
+                                          ENABLE_COLR_BLND_FUNC |
+                                          BLENDFUNC_ADD |
+                                          ENABLE_SRC_BLND_FACTOR |
+                                          SRC_BLND_FACT(BLENDFACT_ONE) |
+                                          ENABLE_DST_BLND_FACTOR |
+                                          DST_BLND_FACT(BLENDFACT_ZERO));
 
    i830->state.Ctx[I830_CTXREG_STATE2] = (_3DSTATE_MODES_2_CMD |
-					  ENABLE_GLOBAL_DEPTH_BIAS | 
-					  GLOBAL_DEPTH_BIAS(0) |
-					  ENABLE_ALPHA_TEST_FUNC | 
-					  ALPHA_TEST_FUNC(COMPAREFUNC_ALWAYS) |
-					  ALPHA_REF_VALUE(0) );
+                                          ENABLE_GLOBAL_DEPTH_BIAS |
+                                          GLOBAL_DEPTH_BIAS(0) |
+                                          ENABLE_ALPHA_TEST_FUNC |
+                                          ALPHA_TEST_FUNC(COMPAREFUNC_ALWAYS)
+                                          | ALPHA_REF_VALUE(0));
 
    i830->state.Ctx[I830_CTXREG_STATE3] = (_3DSTATE_MODES_3_CMD |
-					  ENABLE_DEPTH_TEST_FUNC |
-					  DEPTH_TEST_FUNC(COMPAREFUNC_LESS) |
-					  ENABLE_ALPHA_SHADE_MODE |
-					  ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_FOG_SHADE_MODE |
-					  FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_SPEC_SHADE_MODE |
-					  SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_COLOR_SHADE_MODE |
-					  COLOR_SHADE_MODE(SHADE_MODE_LINEAR) |
-					  ENABLE_CULL_MODE |
-					  CULLMODE_NONE);
+                                          ENABLE_DEPTH_TEST_FUNC |
+                                          DEPTH_TEST_FUNC(COMPAREFUNC_LESS) |
+                                          ENABLE_ALPHA_SHADE_MODE |
+                                          ALPHA_SHADE_MODE(SHADE_MODE_LINEAR)
+                                          | ENABLE_FOG_SHADE_MODE |
+                                          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
+                                          ENABLE_SPEC_SHADE_MODE |
+                                          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
+                                          ENABLE_COLOR_SHADE_MODE |
+                                          COLOR_SHADE_MODE(SHADE_MODE_LINEAR)
+                                          | ENABLE_CULL_MODE | CULLMODE_NONE);
 
    i830->state.Ctx[I830_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-					  ENABLE_LOGIC_OP_FUNC |
-					  LOGIC_OP_FUNC(LOGICOP_COPY) |
-					  ENABLE_STENCIL_TEST_MASK |
-					  STENCIL_TEST_MASK(0xff) |
-					  ENABLE_STENCIL_WRITE_MASK |
-					  STENCIL_WRITE_MASK(0xff));
+                                          ENABLE_LOGIC_OP_FUNC |
+                                          LOGIC_OP_FUNC(LOGICOP_COPY) |
+                                          ENABLE_STENCIL_TEST_MASK |
+                                          STENCIL_TEST_MASK(0xff) |
+                                          ENABLE_STENCIL_WRITE_MASK |
+                                          STENCIL_WRITE_MASK(0xff));
 
    i830->state.Ctx[I830_CTXREG_STENCILTST] = (_3DSTATE_STENCIL_TEST_CMD |
-					      ENABLE_STENCIL_PARMS |
-					      STENCIL_FAIL_OP(STENCILOP_KEEP) |
-					      STENCIL_PASS_DEPTH_FAIL_OP(STENCILOP_KEEP) |
-					      STENCIL_PASS_DEPTH_PASS_OP(STENCILOP_KEEP) |
-					      ENABLE_STENCIL_TEST_FUNC |
-					      STENCIL_TEST_FUNC(COMPAREFUNC_ALWAYS) |
-					      ENABLE_STENCIL_REF_VALUE |
-					      STENCIL_REF_VALUE(0) );
-
-   i830->state.Ctx[I830_CTXREG_STATE5] = (_3DSTATE_MODES_5_CMD |
-					  FLUSH_TEXTURE_CACHE |
-					  ENABLE_SPRITE_POINT_TEX |
-					  SPRITE_POINT_TEX_OFF |
-					  ENABLE_FIXED_LINE_WIDTH |
-					  FIXED_LINE_WIDTH(0x2) | /* 1.0 */
-					  ENABLE_FIXED_POINT_WIDTH |
-					  FIXED_POINT_WIDTH(1) );
+                                              ENABLE_STENCIL_PARMS |
+                                              STENCIL_FAIL_OP(STENCILOP_KEEP)
+                                              |
+                                              STENCIL_PASS_DEPTH_FAIL_OP
+                                              (STENCILOP_KEEP) |
+                                              STENCIL_PASS_DEPTH_PASS_OP
+                                              (STENCILOP_KEEP) |
+                                              ENABLE_STENCIL_TEST_FUNC |
+                                              STENCIL_TEST_FUNC
+                                              (COMPAREFUNC_ALWAYS) |
+                                              ENABLE_STENCIL_REF_VALUE |
+                                              STENCIL_REF_VALUE(0));
+
+   i830->state.Ctx[I830_CTXREG_STATE5] = (_3DSTATE_MODES_5_CMD | FLUSH_TEXTURE_CACHE | ENABLE_SPRITE_POINT_TEX | SPRITE_POINT_TEX_OFF | ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(0x2) |       /* 1.0 */
+                                          ENABLE_FIXED_POINT_WIDTH |
+                                          FIXED_POINT_WIDTH(1));
 
    i830->state.Ctx[I830_CTXREG_IALPHAB] = (_3DSTATE_INDPT_ALPHA_BLEND_CMD |
-					   DISABLE_INDPT_ALPHA_BLEND |
-					   ENABLE_ALPHA_BLENDFUNC |
-					   ABLENDFUNC_ADD);
+                                           DISABLE_INDPT_ALPHA_BLEND |
+                                           ENABLE_ALPHA_BLENDFUNC |
+                                           ABLENDFUNC_ADD);
 
    i830->state.Ctx[I830_CTXREG_FOGCOLOR] = (_3DSTATE_FOG_COLOR_CMD |
-					    FOG_COLOR_RED(0) |
-					    FOG_COLOR_GREEN(0) |
-					    FOG_COLOR_BLUE(0));
+                                            FOG_COLOR_RED(0) |
+                                            FOG_COLOR_GREEN(0) |
+                                            FOG_COLOR_BLUE(0));
 
    i830->state.Ctx[I830_CTXREG_BLENDCOLOR0] = _3DSTATE_CONST_BLEND_COLOR_CMD;
    i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] = 0;
 
    i830->state.Ctx[I830_CTXREG_MCSB0] = _3DSTATE_MAP_COORD_SETBIND_CMD;
    i830->state.Ctx[I830_CTXREG_MCSB1] = (TEXBIND_SET3(TEXCOORDSRC_VTXSET_3) |
-					 TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |
-					 TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |
-					 TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));
-					 
+                                         TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |
+                                         TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |
+                                         TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));
+
 
    i830->state.Stipple[I830_STPREG_ST0] = _3DSTATE_STIPPLE;
 
    i830->state.Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] = 
-      (BUF_3D_ID_COLOR_BACK | 
-       BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
+   i830->state.Buffer[I830_DESTREG_CBUFADDR1] = (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(screen->front.pitch) |     /* pitch in bytes */
+                                                 BUF_3D_USE_FENCE);
 
 
    i830->state.Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] = 
-      (BUF_3D_ID_DEPTH |
-       BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_DBUFADDR2] = screen->depth.offset;
-
+   i830->state.Buffer[I830_DESTREG_DBUFADDR1] = (BUF_3D_ID_DEPTH | BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
+                                                 BUF_3D_USE_FENCE);
 
    i830->state.Buffer[I830_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
 
+#if 0
    switch (screen->fbFormat) {
-   case DV_PF_555:
    case DV_PF_565:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					      DSTORG_VERT_BIAS(0x8) | /* .5 */
-					      screen->fbFormat |
-					      DEPTH_IS_Z |
-					      DEPTH_FRMT_16_FIXED);
+      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |   /* .5 */
+                                              DSTORG_VERT_BIAS(0x8) |   /* .5 */
+                                              screen->fbFormat |
+                                              DEPTH_IS_Z |
+                                              DEPTH_FRMT_16_FIXED);
       break;
    case DV_PF_8888:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					      DSTORG_VERT_BIAS(0x8) | /* .5 */
-					      screen->fbFormat |
-					      DEPTH_IS_Z |
-					      DEPTH_FRMT_24_FIXED_8_OTHER);
+      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |   /* .5 */
+                                              DSTORG_VERT_BIAS(0x8) |   /* .5 */
+                                              screen->fbFormat |
+                                              DEPTH_IS_Z |
+                                              DEPTH_FRMT_24_FIXED_8_OTHER);
       break;
    }
-
+#endif
    i830->state.Buffer[I830_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-					       DISABLE_SCISSOR_RECT);
+                                               DISABLE_SCISSOR_RECT);
    i830->state.Buffer[I830_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
    i830->state.Buffer[I830_DESTREG_SR1] = 0;
    i830->state.Buffer[I830_DESTREG_SR2] = 0;
 }
 
 
-void i830InitStateFuncs( struct dd_function_table *functions )
+void
+i830InitStateFuncs(struct dd_function_table *functions)
 {
    functions->AlphaFunc = i830AlphaFunc;
    functions->BlendColor = i830BlendColor;
@@ -1070,20 +1096,21 @@ void i830InitStateFuncs( struct dd_function_table *functions )
    functions->StencilOpSeparate = i830StencilOpSeparate;
 }
 
-void i830InitState( i830ContextPtr i830 )
+void
+i830InitState(struct i830_context *i830)
 {
    GLcontext *ctx = &i830->intel.ctx;
 
-   i830_init_packets( i830 );
+   i830_init_packets(i830);
 
    _mesa_init_driver_state(ctx);
 
-   memcpy( &i830->initial, &i830->state, sizeof(i830->state) );
+   memcpy(&i830->initial, &i830->state, sizeof(i830->state));
 
    i830->current = &i830->state;
    i830->state.emitted = 0;
-   i830->state.active = (I830_UPLOAD_TEXBLEND(0) |
-			 I830_UPLOAD_STIPPLE |
-			 I830_UPLOAD_CTX |
-			 I830_UPLOAD_BUFFERS);
+   i830->state.active = (I830_UPLOAD_INVARIENT |
+                         I830_UPLOAD_TEXBLEND(0) |
+                         I830_UPLOAD_STIPPLE |
+                         I830_UPLOAD_CTX | I830_UPLOAD_BUFFERS);
 }
diff --git a/src/mesa/drivers/dri/i915/i830_tex.c b/src/mesa/drivers/dri/i915/i830_tex.c
index 3c4aedb35cb..fed464d1aac 100644
--- a/src/mesa/drivers/dri/i915/i830_tex.c
+++ b/src/mesa/drivers/dri/i915/i830_tex.c
@@ -45,261 +45,13 @@
 
 
 
-
-/**
- * Set the texture wrap modes.
- * 
- * The i830M (and related graphics cores) do not support GL_CLAMP.  The Intel
- * drivers for "other operating systems" implement GL_CLAMP as
- * GL_CLAMP_TO_EDGE, so the same is done here.
- * 
- * \param t Texture object whose wrap modes are to be set
- * \param swrap Wrap mode for the \a s texture coordinate
- * \param twrap Wrap mode for the \a t texture coordinate
- */
-static void i830SetTexWrapping(i830TextureObjectPtr tex,
-			       GLenum swrap, 
-			       GLenum twrap)
-{
-   tex->Setup[I830_TEXREG_MCS] &= ~(TEXCOORD_ADDR_U_MASK|TEXCOORD_ADDR_V_MASK);
-
-   switch( swrap ) {
-   case GL_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP);
-      break;
-   case GL_CLAMP:
-   case GL_CLAMP_TO_EDGE:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_CLAMP);
-      break;
-   case GL_CLAMP_TO_BORDER:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_CLAMP_BORDER);
-      break;
-   case GL_MIRRORED_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_MIRROR);
-      break;
-   default:
-      break;
-   }
-
-   switch( twrap ) {
-   case GL_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP);
-      break;
-   case GL_CLAMP:
-   case GL_CLAMP_TO_EDGE:
-      tex->Setup[I830_TEXREG_MCS] |= TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_CLAMP);
-      break;
-   case GL_CLAMP_TO_BORDER:
-      tex->Setup[I830_TEXREG_MCS] |= 
-			TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_CLAMP_BORDER);
-      break;
-   case GL_MIRRORED_REPEAT:
-      tex->Setup[I830_TEXREG_MCS] |=
-			TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_MIRROR);
-      break;
-   default:
-      break;
-   }
-}
-
-
-/**
- * Set the texture magnification and minification modes.
- * 
- * \param t Texture whose filter modes are to be set
- * \param minf Texture minification mode
- * \param magf Texture magnification mode
- * \param bias LOD bias for this texture unit.
- */
-
-static void i830SetTexFilter( i830TextureObjectPtr t, GLenum minf, GLenum magf,
-			      GLfloat maxanisotropy )
-{
-   int minFilt = 0, mipFilt = 0, magFilt = 0;
-
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if ( maxanisotropy > 1.0 ) {
-      minFilt = FILTER_ANISOTROPIC;
-      magFilt = FILTER_ANISOTROPIC;
-   }
-   else {
-      switch (minf) {
-      case GL_NEAREST:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_NONE;
-	 break;
-      case GL_LINEAR:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_NONE;
-	 break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_NEAREST;
-	 break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_NEAREST;
-	 break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-	 minFilt = FILTER_NEAREST;
-	 mipFilt = MIPFILTER_LINEAR;
-	 break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-	 minFilt = FILTER_LINEAR;
-	 mipFilt = MIPFILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }
-
-      switch (magf) {
-      case GL_NEAREST:
-	 magFilt = FILTER_NEAREST;
-	 break;
-      case GL_LINEAR:
-	 magFilt = FILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }  
-   }
-
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIP_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAG_FILTER_MASK;
-   t->Setup[I830_TEXREG_TM0S3] |= ((minFilt << TM0S3_MIN_FILTER_SHIFT) |
-				   (mipFilt << TM0S3_MIP_FILTER_SHIFT) |
-				   (magFilt << TM0S3_MAG_FILTER_SHIFT));
-}
-
-static void i830SetTexBorderColor(i830TextureObjectPtr t, GLubyte color[4])
-{
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-    t->Setup[I830_TEXREG_TM0S4] = 
-        INTEL_PACKCOLOR8888(color[0],color[1],color[2],color[3]);
-}
-
-
-/**
- * Allocate space for and load the mesa images into the texture memory block.
- * This will happen before drawing with a new texture, or drawing with a
- * texture after it was swapped out or teximaged again.
- */
-
-intelTextureObjectPtr i830AllocTexObj( struct gl_texture_object *texObj )
-{
-   i830TextureObjectPtr t = CALLOC_STRUCT( i830_texture_object );
-   if ( !t ) 
-      return NULL;
-
-   texObj->DriverData = t;
-   t->intel.base.tObj = texObj;
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-   make_empty_list( &t->intel.base );
-
-   t->Setup[I830_TEXREG_TM0LI] = 0; /* not used */
-   t->Setup[I830_TEXREG_TM0S0] = 0;
-   t->Setup[I830_TEXREG_TM0S1] = 0;
-   t->Setup[I830_TEXREG_TM0S2] = 0;
-   t->Setup[I830_TEXREG_TM0S3] = 0;
-   t->Setup[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-				MAP_UNIT(0) |
-				ENABLE_TEXCOORD_PARAMS |
-				TEXCOORDS_ARE_NORMAL |
-				TEXCOORDTYPE_CARTESIAN |
-				ENABLE_ADDR_V_CNTL |
-				TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
-				ENABLE_ADDR_U_CNTL |
-				TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
-
-   
-   i830SetTexWrapping( t, texObj->WrapS, texObj->WrapT );
-   i830SetTexFilter( t, texObj->MinFilter, texObj->MagFilter, 
-		     texObj->MaxAnisotropy );
-   i830SetTexBorderColor( t, texObj->_BorderChan );
-
-   return &t->intel;
-}
-
-
-static void i830TexParameter( GLcontext *ctx, GLenum target,
-			      struct gl_texture_object *tObj,
-			      GLenum pname, const GLfloat *params )
-{
-   i830TextureObjectPtr t = (i830TextureObjectPtr) tObj->DriverData;
-   if (!t)
-      return;
-
-   switch (pname) {
-   case GL_TEXTURE_MIN_FILTER:
-   case GL_TEXTURE_MAG_FILTER:
-   case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-      i830SetTexFilter( t, tObj->MinFilter, tObj->MagFilter,
-			tObj->MaxAnisotropy);
-      break;
-
-   case GL_TEXTURE_WRAP_S:
-   case GL_TEXTURE_WRAP_T:
-      i830SetTexWrapping( t, tObj->WrapS, tObj->WrapT );
-      break;
-  
-   case GL_TEXTURE_BORDER_COLOR:
-      i830SetTexBorderColor( t, tObj->_BorderChan );
-      break;
-
-   case GL_TEXTURE_BASE_LEVEL:
-   case GL_TEXTURE_MAX_LEVEL:
-   case GL_TEXTURE_MIN_LOD:
-   case GL_TEXTURE_MAX_LOD:
-      /* The i830 and its successors can do a lot of this without
-       * reloading the textures.  A project for someone?
-       */
-      intelFlush( ctx );
-      driSwapOutTextureObject( (driTextureObject *) t );
-      break;
-
-   default:
-      return;
-   }
-
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-}
-
-
-static void i830TexEnv( GLcontext *ctx, GLenum target, 
-			GLenum pname, const GLfloat *param )
+static void
+i830TexEnv(GLcontext * ctx, GLenum target,
+           GLenum pname, const GLfloat * param)
 {
-   i830ContextPtr i830 = I830_CONTEXT( ctx );
-   GLuint unit = ctx->Texture.CurrentUnit;
 
    switch (pname) {
-   case GL_TEXTURE_ENV_COLOR: 
-#if 0
-   {
-      GLubyte r, g, b, a;
-      GLuint col;
-      
-      UNCLAMPED_FLOAT_TO_UBYTE(r, param[RCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(g, param[GCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(b, param[BCOMP]);
-      UNCLAMPED_FLOAT_TO_UBYTE(a, param[ACOMP]);
-
-      col = ((a << 24) | (r << 16) | (g << 8) | b);
-
-      if (col != i830->state.TexEnv[unit][I830_TEXENVREG_COL1]) {
-	 I830_STATECHANGE(i830, I830_UPLOAD_TEXENV);
-	 i830->state.TexEnv[unit][I830_TEXENVREG_COL1] = col;
-      }
-
-      break;
-   }
-#endif
+   case GL_TEXTURE_ENV_COLOR:
    case GL_TEXTURE_ENV_MODE:
    case GL_COMBINE_RGB:
    case GL_COMBINE_ALPHA:
@@ -319,38 +71,30 @@ static void i830TexEnv( GLcontext *ctx, GLenum target,
    case GL_ALPHA_SCALE:
       break;
 
-   case GL_TEXTURE_LOD_BIAS: {
-      int b = (int) ((*param) * 16.0);
-      if (b > 63) b = 63;
-      if (b < -64) b = -64;
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_TM0S3] &= ~TM0S3_LOD_BIAS_MASK;
-      i830->state.Tex[unit][I830_TEXREG_TM0S3] |= 
-	 ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK);
-      break;
-   }
+   case GL_TEXTURE_LOD_BIAS:{
+         struct i830_context *i830 = i830_context(ctx);
+         GLuint unit = ctx->Texture.CurrentUnit;
+         int b = (int) ((*param) * 16.0);
+         if (b > 63)
+            b = 63;
+         if (b < -64)
+            b = -64;
+         I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
+         i830->lodbias_tm0s3[unit] =
+            ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK);
+         break;
+      }
 
    default:
       break;
    }
 }
 
-static void i830BindTexture( GLcontext *ctx, GLenum target,
-			    struct gl_texture_object *texObj )
-{
-   i830TextureObjectPtr tex;
-   
-   if (!texObj->DriverData)
-      i830AllocTexObj( texObj );
-   
-   tex = (i830TextureObjectPtr)texObj->DriverData;
-}
 
 
 
-void i830InitTextureFuncs( struct dd_function_table *functions )
+void
+i830InitTextureFuncs(struct dd_function_table *functions)
 {
-   functions->BindTexture 		= i830BindTexture;
-   functions->TexEnv                    = i830TexEnv;
-   functions->TexParameter              = i830TexParameter;
+   functions->TexEnv = i830TexEnv;
 }
diff --git a/src/mesa/drivers/dri/i915/i830_texblend.c b/src/mesa/drivers/dri/i915/i830_texblend.c
index 49e0347643c..58f220eb7ce 100644
--- a/src/mesa/drivers/dri/i915/i830_texblend.c
+++ b/src/mesa/drivers/dri/i915/i830_texblend.c
@@ -46,46 +46,42 @@
 /* ================================================================
  * Texture combine functions
  */
-static GLuint pass_through( GLuint *state, GLuint blendUnit )
+static GLuint
+pass_through(GLuint * state, GLuint blendUnit)
 {
    state[0] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-	       TEXPIPE_COLOR |
-	       ENABLE_TEXOUTPUT_WRT_SEL |
-	       TEXOP_OUTPUT_CURRENT |
-	       DISABLE_TEX_CNTRL_STAGE |
-	       TEXOP_SCALE_1X |
-	       TEXOP_MODIFY_PARMS |
-	       TEXBLENDOP_ARG1);
+               TEXPIPE_COLOR |
+               ENABLE_TEXOUTPUT_WRT_SEL |
+               TEXOP_OUTPUT_CURRENT |
+               DISABLE_TEX_CNTRL_STAGE |
+               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    state[1] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-	       TEXPIPE_ALPHA |
-	       ENABLE_TEXOUTPUT_WRT_SEL |
-	       TEXOP_OUTPUT_CURRENT |
-	       TEXOP_SCALE_1X |
-	       TEXOP_MODIFY_PARMS |
-	       TEXBLENDOP_ARG1);
+               TEXPIPE_ALPHA |
+               ENABLE_TEXOUTPUT_WRT_SEL |
+               TEXOP_OUTPUT_CURRENT |
+               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
    state[2] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-	       TEXPIPE_COLOR |
-	       TEXBLEND_ARG1 |
-	       TEXBLENDARG_MODIFY_PARMS |
-	       TEXBLENDARG_CURRENT);
+               TEXPIPE_COLOR |
+               TEXBLEND_ARG1 |
+               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
    state[3] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-	       TEXPIPE_ALPHA |
-	       TEXBLEND_ARG1 |
-	       TEXBLENDARG_MODIFY_PARMS |
-	       TEXBLENDARG_CURRENT);
+               TEXPIPE_ALPHA |
+               TEXBLEND_ARG1 |
+               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
 
    return 4;
 }
 
-static GLuint emit_factor( GLuint blendUnit, GLuint *state, GLuint count, 
-			   const GLfloat *factor )
+static GLuint
+emit_factor(GLuint blendUnit, GLuint * state, GLuint count,
+            const GLfloat * factor)
 {
    GLubyte r, g, b, a;
    GLuint col;
-      
+
    if (0)
       fprintf(stderr, "emit constant %d: %.2f %.2f %.2f %.2f\n",
-	  blendUnit, factor[0], factor[1], factor[2], factor[3]);
+              blendUnit, factor[0], factor[1], factor[2], factor[3]);
 
    UNCLAMPED_FLOAT_TO_UBYTE(r, factor[0]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, factor[1]);
@@ -94,21 +90,27 @@ static GLuint emit_factor( GLuint blendUnit, GLuint *state, GLuint count,
 
    col = ((a << 24) | (r << 16) | (g << 8) | b);
 
-   state[count++] = _3DSTATE_COLOR_FACTOR_N_CMD(blendUnit); 
+   state[count++] = _3DSTATE_COLOR_FACTOR_N_CMD(blendUnit);
    state[count++] = col;
 
    return count;
 }
 
 
-static __inline__ GLuint GetTexelOp(GLint unit)
+static INLINE GLuint
+GetTexelOp(GLint unit)
 {
-   switch(unit) {
-   case 0: return TEXBLENDARG_TEXEL0;
-   case 1: return TEXBLENDARG_TEXEL1;
-   case 2: return TEXBLENDARG_TEXEL2;
-   case 3: return TEXBLENDARG_TEXEL3;
-   default: return TEXBLENDARG_TEXEL0;
+   switch (unit) {
+   case 0:
+      return TEXBLENDARG_TEXEL0;
+   case 1:
+      return TEXBLENDARG_TEXEL1;
+   case 2:
+      return TEXBLENDARG_TEXEL2;
+   case 3:
+      return TEXBLENDARG_TEXEL3;
+   default:
+      return TEXBLENDARG_TEXEL0;
    }
 }
 
@@ -132,12 +134,10 @@ static __inline__ GLuint GetTexelOp(GLint unit)
  * partial support for the extension?
  */
 GLuint
-i830SetTexEnvCombine(i830ContextPtr i830,
-		     const struct gl_tex_env_combine_state * combine,
-		     GLint blendUnit,
-		     GLuint texel_op,
-		     GLuint *state,
-		     const GLfloat *factor )
+i830SetTexEnvCombine(struct i830_context * i830,
+                     const struct gl_tex_env_combine_state * combine,
+                     GLint blendUnit,
+                     GLuint texel_op, GLuint * state, const GLfloat * factor)
 {
    const GLuint numColorArgs = combine->_NumArgsRGB;
    const GLuint numAlphaArgs = combine->_NumArgsA;
@@ -162,7 +162,7 @@ i830SetTexEnvCombine(i830ContextPtr i830,
       TEXPIPE_ALPHA | TEXBLEND_ARG0 | TEXBLENDARG_MODIFY_PARMS,
    };
 
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
+   if (INTEL_DEBUG & DEBUG_TEXTURE)
       fprintf(stderr, "%s\n", __FUNCTION__);
 
 
@@ -188,23 +188,23 @@ i830SetTexEnvCombine(i830ContextPtr i830,
    }
 
 
-   switch(combine->ModeRGB) {
-   case GL_REPLACE: 
+   switch (combine->ModeRGB) {
+   case GL_REPLACE:
       blendop = TEXBLENDOP_ARG1;
       break;
-   case GL_MODULATE: 
+   case GL_MODULATE:
       blendop = TEXBLENDOP_MODULATE;
       break;
-   case GL_ADD: 
+   case GL_ADD:
       blendop = TEXBLENDOP_ADD;
       break;
    case GL_ADD_SIGNED:
-      blendop = TEXBLENDOP_ADDSIGNED; 
+      blendop = TEXBLENDOP_ADDSIGNED;
       break;
    case GL_INTERPOLATE:
-      blendop = TEXBLENDOP_BLEND; 
+      blendop = TEXBLENDOP_BLEND;
       break;
-   case GL_SUBTRACT: 
+   case GL_SUBTRACT:
       blendop = TEXBLENDOP_SUBTRACT;
       break;
    case GL_DOT3_RGB_EXT:
@@ -215,55 +215,54 @@ i830SetTexEnvCombine(i830ContextPtr i830,
    case GL_DOT3_RGBA:
       blendop = TEXBLENDOP_DOT3;
       break;
-   default: 
-      return pass_through( state, blendUnit );
+   default:
+      return pass_through(state, blendUnit);
    }
 
    blendop |= (rgb_shift << TEXOP_SCALE_SHIFT);
 
 
    /* Handle RGB args */
-   for(i = 0; i < 3; i++) {
-      switch(combine->SourceRGB[i]) {
-      case GL_TEXTURE: 
-	 args_RGB[i] = texel_op;
-	 break;
+   for (i = 0; i < 3; i++) {
+      switch (combine->SourceRGB[i]) {
+      case GL_TEXTURE:
+         args_RGB[i] = texel_op;
+         break;
       case GL_TEXTURE0:
       case GL_TEXTURE1:
       case GL_TEXTURE2:
       case GL_TEXTURE3:
-	 args_RGB[i] = GetTexelOp( combine->SourceRGB[i] - GL_TEXTURE0 );
-	 break;
+         args_RGB[i] = GetTexelOp(combine->SourceRGB[i] - GL_TEXTURE0);
+         break;
       case GL_CONSTANT:
-	 args_RGB[i] = TEXBLENDARG_FACTOR_N; 
-	 need_factor = 1;
-	 break;
+         args_RGB[i] = TEXBLENDARG_FACTOR_N;
+         need_factor = 1;
+         break;
       case GL_PRIMARY_COLOR:
-	 args_RGB[i] = TEXBLENDARG_DIFFUSE;
-	 break;
+         args_RGB[i] = TEXBLENDARG_DIFFUSE;
+         break;
       case GL_PREVIOUS:
-	 args_RGB[i] = TEXBLENDARG_CURRENT; 
-	 break;
-      default: 
-	 return pass_through( state, blendUnit );
+         args_RGB[i] = TEXBLENDARG_CURRENT;
+         break;
+      default:
+         return pass_through(state, blendUnit);
       }
 
-      switch(combine->OperandRGB[i]) {
-      case GL_SRC_COLOR: 
-	 args_RGB[i] |= 0;
-	 break;
-      case GL_ONE_MINUS_SRC_COLOR: 
-	 args_RGB[i] |= TEXBLENDARG_INV_ARG;
-	 break;
-      case GL_SRC_ALPHA: 
-	 args_RGB[i] |= TEXBLENDARG_REPLICATE_ALPHA;
-	 break;
-      case GL_ONE_MINUS_SRC_ALPHA: 
-	 args_RGB[i] |= (TEXBLENDARG_REPLICATE_ALPHA | 
-			 TEXBLENDARG_INV_ARG);
-	 break;
-      default: 
-	 return pass_through( state, blendUnit );
+      switch (combine->OperandRGB[i]) {
+      case GL_SRC_COLOR:
+         args_RGB[i] |= 0;
+         break;
+      case GL_ONE_MINUS_SRC_COLOR:
+         args_RGB[i] |= TEXBLENDARG_INV_ARG;
+         break;
+      case GL_SRC_ALPHA:
+         args_RGB[i] |= TEXBLENDARG_REPLICATE_ALPHA;
+         break;
+      case GL_ONE_MINUS_SRC_ALPHA:
+         args_RGB[i] |= (TEXBLENDARG_REPLICATE_ALPHA | TEXBLENDARG_INV_ARG);
+         break;
+      default:
+         return pass_through(state, blendUnit);
       }
    }
 
@@ -275,76 +274,76 @@ i830SetTexEnvCombine(i830ContextPtr i830,
     * Note - the global factor is set up with alpha == .5, so 
     * the alpha part of the DOT4 calculation should be zero.
     */
-   if ( combine->ModeRGB == GL_DOT3_RGBA_EXT || 
-	combine->ModeRGB == GL_DOT3_RGBA ) {
+   if (combine->ModeRGB == GL_DOT3_RGBA_EXT ||
+       combine->ModeRGB == GL_DOT3_RGBA) {
       ablendop = TEXBLENDOP_DOT4;
-      args_A[0] = TEXBLENDARG_FACTOR; /* the global factor */
+      args_A[0] = TEXBLENDARG_FACTOR;   /* the global factor */
       args_A[1] = TEXBLENDARG_FACTOR;
       args_A[2] = TEXBLENDARG_FACTOR;
    }
    else {
-      switch(combine->ModeA) {
-      case GL_REPLACE: 
-	 ablendop = TEXBLENDOP_ARG1;
-	 break;
-      case GL_MODULATE: 
-	 ablendop = TEXBLENDOP_MODULATE;
-	 break;
-      case GL_ADD: 
-	 ablendop = TEXBLENDOP_ADD;
-	 break;
+      switch (combine->ModeA) {
+      case GL_REPLACE:
+         ablendop = TEXBLENDOP_ARG1;
+         break;
+      case GL_MODULATE:
+         ablendop = TEXBLENDOP_MODULATE;
+         break;
+      case GL_ADD:
+         ablendop = TEXBLENDOP_ADD;
+         break;
       case GL_ADD_SIGNED:
-	 ablendop = TEXBLENDOP_ADDSIGNED; 
-	 break;
+         ablendop = TEXBLENDOP_ADDSIGNED;
+         break;
       case GL_INTERPOLATE:
-	 ablendop = TEXBLENDOP_BLEND; 
-	 break;
-      case GL_SUBTRACT: 
-	 ablendop = TEXBLENDOP_SUBTRACT;
-	 break;
+         ablendop = TEXBLENDOP_BLEND;
+         break;
+      case GL_SUBTRACT:
+         ablendop = TEXBLENDOP_SUBTRACT;
+         break;
       default:
-	 return pass_through( state, blendUnit );
+         return pass_through(state, blendUnit);
       }
 
 
       ablendop |= (alpha_shift << TEXOP_SCALE_SHIFT);
 
       /* Handle A args */
-      for(i = 0; i < 3; i++) {
-	 switch(combine->SourceA[i]) {
-	 case GL_TEXTURE: 
-	    args_A[i] = texel_op;
-	    break;
-	 case GL_TEXTURE0:
-	 case GL_TEXTURE1:
-	 case GL_TEXTURE2:
-	 case GL_TEXTURE3:
-	    args_A[i] = GetTexelOp( combine->SourceA[i] - GL_TEXTURE0 );
-	    break;
-	 case GL_CONSTANT:
-	    args_A[i] = TEXBLENDARG_FACTOR_N; 
-	    need_factor = 1;
-	    break;
-	 case GL_PRIMARY_COLOR:
-	    args_A[i] = TEXBLENDARG_DIFFUSE; 
-	    break;
-	 case GL_PREVIOUS:
-	    args_A[i] = TEXBLENDARG_CURRENT; 
-	    break;
-	 default: 
-	    return pass_through( state, blendUnit );
-	 }
-
-	 switch(combine->OperandA[i]) {
-	 case GL_SRC_ALPHA: 
-	    args_A[i] |= 0;
-	    break;
-	 case GL_ONE_MINUS_SRC_ALPHA: 
-	    args_A[i] |= TEXBLENDARG_INV_ARG;
-	    break;
-	 default: 
-	    return pass_through( state, blendUnit );
-	 }
+      for (i = 0; i < 3; i++) {
+         switch (combine->SourceA[i]) {
+         case GL_TEXTURE:
+            args_A[i] = texel_op;
+            break;
+         case GL_TEXTURE0:
+         case GL_TEXTURE1:
+         case GL_TEXTURE2:
+         case GL_TEXTURE3:
+            args_A[i] = GetTexelOp(combine->SourceA[i] - GL_TEXTURE0);
+            break;
+         case GL_CONSTANT:
+            args_A[i] = TEXBLENDARG_FACTOR_N;
+            need_factor = 1;
+            break;
+         case GL_PRIMARY_COLOR:
+            args_A[i] = TEXBLENDARG_DIFFUSE;
+            break;
+         case GL_PREVIOUS:
+            args_A[i] = TEXBLENDARG_CURRENT;
+            break;
+         default:
+            return pass_through(state, blendUnit);
+         }
+
+         switch (combine->OperandA[i]) {
+         case GL_SRC_ALPHA:
+            args_A[i] |= 0;
+            break;
+         case GL_ONE_MINUS_SRC_ALPHA:
+            args_A[i] |= TEXBLENDARG_INV_ARG;
+            break;
+         default:
+            return pass_through(state, blendUnit);
+         }
       }
    }
 
@@ -363,86 +362,86 @@ i830SetTexEnvCombine(i830ContextPtr i830,
 
    used = 0;
    state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-		    TEXPIPE_COLOR |
-		    ENABLE_TEXOUTPUT_WRT_SEL |
-		    TEXOP_OUTPUT_CURRENT |
-		    DISABLE_TEX_CNTRL_STAGE |
-		    TEXOP_MODIFY_PARMS |
-		    blendop);
+                    TEXPIPE_COLOR |
+                    ENABLE_TEXOUTPUT_WRT_SEL |
+                    TEXOP_OUTPUT_CURRENT |
+                    DISABLE_TEX_CNTRL_STAGE | TEXOP_MODIFY_PARMS | blendop);
    state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-		    TEXPIPE_ALPHA |
-		    ENABLE_TEXOUTPUT_WRT_SEL |
-		    TEXOP_OUTPUT_CURRENT |
-		    TEXOP_MODIFY_PARMS |
-		    ablendop);
+                    TEXPIPE_ALPHA |
+                    ENABLE_TEXOUTPUT_WRT_SEL |
+                    TEXOP_OUTPUT_CURRENT | TEXOP_MODIFY_PARMS | ablendop);
 
-   for ( i = 0 ; i < numColorArgs ; i++ ) {
+   for (i = 0; i < numColorArgs; i++) {
       state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-		       tex_blend_rgb[i] | args_RGB[i]);
+                       tex_blend_rgb[i] | args_RGB[i]);
    }
 
-   for ( i = 0 ; i < numAlphaArgs ; i++ ) {
+   for (i = 0; i < numAlphaArgs; i++) {
       state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-		       tex_blend_a[i] | args_A[i]);
+                       tex_blend_a[i] | args_A[i]);
    }
 
 
-   if (need_factor) 
-      return emit_factor( blendUnit, state, used, factor );
-   else 
+   if (need_factor)
+      return emit_factor(blendUnit, state, used, factor);
+   else
       return used;
 }
 
 
-static void emit_texblend( i830ContextPtr i830, GLuint unit, GLuint blendUnit,
-			   GLboolean last_stage )
+static void
+emit_texblend(struct i830_context *i830, GLuint unit, GLuint blendUnit,
+              GLboolean last_stage)
 {
    struct gl_texture_unit *texUnit = &i830->intel.ctx.Texture.Unit[unit];
    GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
 
 
-   if (0) fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
+   if (0)
+      fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
 
    /* Update i830->state.TexBlend
-    */ 
-   tmp_sz = i830SetTexEnvCombine(i830, texUnit->_CurrentCombine, blendUnit, 
-				 GetTexelOp(unit), tmp,
-				 texUnit->EnvColor );
+    */
+   tmp_sz = i830SetTexEnvCombine(i830, texUnit->_CurrentCombine, blendUnit,
+                                 GetTexelOp(unit), tmp, texUnit->EnvColor);
 
-   if (last_stage) 
+   if (last_stage)
       tmp[0] |= TEXOP_LAST_STAGE;
 
    if (tmp_sz != i830->state.TexBlendWordsUsed[blendUnit] ||
-       memcmp( tmp, i830->state.TexBlend[blendUnit], tmp_sz * sizeof(GLuint))) {
-      
-      I830_STATECHANGE( i830, I830_UPLOAD_TEXBLEND(blendUnit) );
-      memcpy( i830->state.TexBlend[blendUnit], tmp, tmp_sz * sizeof(GLuint));
+       memcmp(tmp, i830->state.TexBlend[blendUnit],
+              tmp_sz * sizeof(GLuint))) {
+
+      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(blendUnit));
+      memcpy(i830->state.TexBlend[blendUnit], tmp, tmp_sz * sizeof(GLuint));
       i830->state.TexBlendWordsUsed[blendUnit] = tmp_sz;
    }
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(blendUnit), GL_TRUE);
 }
 
-static void emit_passthrough( i830ContextPtr i830 )
+static void
+emit_passthrough(struct i830_context *i830)
 {
    GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
    GLuint unit = 0;
 
-   tmp_sz = pass_through( tmp, unit );
+   tmp_sz = pass_through(tmp, unit);
    tmp[0] |= TEXOP_LAST_STAGE;
 
    if (tmp_sz != i830->state.TexBlendWordsUsed[unit] ||
-       memcmp( tmp, i830->state.TexBlend[unit], tmp_sz * sizeof(GLuint))) {
-      
-      I830_STATECHANGE( i830, I830_UPLOAD_TEXBLEND(unit) );
-      memcpy( i830->state.TexBlend[unit], tmp, tmp_sz * sizeof(GLuint));
+       memcmp(tmp, i830->state.TexBlend[unit], tmp_sz * sizeof(GLuint))) {
+
+      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(unit));
+      memcpy(i830->state.TexBlend[unit], tmp, tmp_sz * sizeof(GLuint));
       i830->state.TexBlendWordsUsed[unit] = tmp_sz;
    }
 
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(unit), GL_TRUE);
 }
 
-void i830EmitTextureBlend( i830ContextPtr i830 )
+void
+i830EmitTextureBlend(struct i830_context *i830)
 {
    GLcontext *ctx = &i830->intel.ctx;
    GLuint unit, last_stage = 0, blendunit = 0;
@@ -450,16 +449,15 @@ void i830EmitTextureBlend( i830ContextPtr i830 )
    I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND_ALL, GL_FALSE);
 
    if (ctx->Texture._EnabledUnits) {
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) 
-	    last_stage = unit;
+      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
+         if (ctx->Texture.Unit[unit]._ReallyEnabled)
+            last_stage = unit;
 
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) 
-	    emit_texblend( i830, unit, blendunit++, last_stage == unit );
+      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
+         if (ctx->Texture.Unit[unit]._ReallyEnabled)
+            emit_texblend(i830, unit, blendunit++, last_stage == unit);
    }
    else {
-      emit_passthrough( i830 );
+      emit_passthrough(i830);
    }
 }
-
diff --git a/src/mesa/drivers/dri/i915/i830_texstate.c b/src/mesa/drivers/dri/i915/i830_texstate.c
index ba972dac8f1..7613b9d2a6f 100644
--- a/src/mesa/drivers/dri/i915/i830_texstate.c
+++ b/src/mesa/drivers/dri/i915/i830_texstate.c
@@ -25,459 +25,319 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
 #include "mtypes.h"
-#include "simple_list.h"
 #include "enums.h"
 #include "texformat.h"
-#include "texstore.h"
+#include "dri_bufmgr.h"
 
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_ioctl.h"
+#include "intel_mipmap_tree.h"
 #include "intel_tex.h"
 
 #include "i830_context.h"
 #include "i830_reg.h"
 
-static const GLint initial_offsets[6][2] = { {0,0},
-				       {0,2},
-				       {1,0},
-				       {1,2},
-				       {1,1},
-				       {1,3} };
-
-static const GLint step_offsets[6][2] = { {0,2},
-				    {0,2},
-				    {-1,2},
-				    {-1,2},
-				    {-1,1},
-				    {-1,1} };
 
-#define I830_TEX_UNIT_ENABLED(unit)		(1<<unit)
 
-static GLboolean i830SetTexImages( i830ContextPtr i830, 
-				  struct gl_texture_object *tObj )
+static GLuint
+translate_texture_format(GLuint mesa_format)
 {
-   GLuint total_height, pitch, i, textureFormat;
-   i830TextureObjectPtr t = (i830TextureObjectPtr) tObj->DriverData;
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   GLint firstLevel, lastLevel, numLevels;
-
-   switch( baseImage->TexFormat->MesaFormat ) {
+   switch (mesa_format) {
    case MESA_FORMAT_L8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_L8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_L8;
    case MESA_FORMAT_I8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8;
    case MESA_FORMAT_A8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8; /* Kludge -- check with conform, glean */
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8; /* Kludge! */
    case MESA_FORMAT_AL88:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_AY88;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_AY88;
    case MESA_FORMAT_RGB565:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_RGB565;
    case MESA_FORMAT_ARGB1555:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
    case MESA_FORMAT_ARGB4444:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB4444;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      t->intel.texelBytes = 4;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      break;
-
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
    case MESA_FORMAT_YCBCR_REV:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL | 
-		       TM0S1_COLORSPACE_CONVERSION);
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY | /* ??? */
-		       TM0S1_COLORSPACE_CONVERSION);
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
    case MESA_FORMAT_RGB_FXT1:
    case MESA_FORMAT_RGBA_FXT1:
-     t->intel.texelBytes = 2;
-     textureFormat = MAPSURF_COMPRESSED | MT_COMPRESS_FXT1;
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
    case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-     /* 
-      * DXTn pitches are Width/4 * blocksize in bytes 
-      * for DXT1: blocksize=8 so Width/4*8 = Width * 2 
-      * for DXT3/5: blocksize=16 so Width/4*16 = Width * 4
-      */
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-     break;
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
    case MESA_FORMAT_RGBA_DXT3:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-     break;
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
    case MESA_FORMAT_RGBA_DXT5:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
    default:
-      fprintf(stderr, "%s: bad image format\n", __FUNCTION__);
+      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
       abort();
+      return 0;
    }
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    * This depends on the base image size, GL_TEXTURE_MIN_LOD,
-    * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL.
-    * Yes, this looks overly complicated, but it's all needed.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
+}
 
 
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
 
 
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      pitch = dim * t->intel.texelBytes;
-      pitch *= 2;		/* double pitch for cube layouts */
-      pitch = (pitch + 3) & ~3;
-      
-      total_height = dim * 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    if (!t->intel.image[face][i].image) {
-	       fprintf(stderr, "no image %d %d\n", face, i);
-	       break;		/* can't happen */
-	    }
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    x += step_offsets[face][0] * d;
-	    y += step_offsets[face][1] * d;
-	 }
-      }
-      break;
-   }
+/* The i915 (and related graphics cores) do not support GL_CLAMP.  The
+ * Intel drivers for "other operating systems" implement GL_CLAMP as
+ * GL_CLAMP_TO_EDGE, so the same is done here.
+ */
+static GLuint
+translate_wrap_mode(GLenum wrap)
+{
+   switch (wrap) {
+   case GL_REPEAT:
+      return TEXCOORDMODE_WRAP;
+   case GL_CLAMP:
+   case GL_CLAMP_TO_EDGE:
+      return TEXCOORDMODE_CLAMP;        /* not really correct */
+   case GL_CLAMP_TO_BORDER:
+      return TEXCOORDMODE_CLAMP_BORDER;
+   case GL_MIRRORED_REPEAT:
+      return TEXCOORDMODE_MIRROR;
    default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      for ( total_height = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 if (t->intel.image[0][i].image->IsCompressed)
-	 {
-	   if (t->intel.image[0][i].image->Height > 4)
-	     total_height += t->intel.image[0][i].image->Height/4;
-	   else
-	     total_height += 1;
-	 }
-	 else
-	   total_height += MAX2(2, t->intel.image[0][i].image->Height);
-      }
-      break;
+      return TEXCOORDMODE_WRAP;
    }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = total_height*pitch;
-   t->intel.max_level = i-1;
-   t->Setup[I830_TEXREG_TM0S1] = 
-      (((tObj->Image[0][firstLevel]->Height - 1) << TM0S1_HEIGHT_SHIFT) |
-       ((tObj->Image[0][firstLevel]->Width - 1) << TM0S1_WIDTH_SHIFT) |
-       textureFormat);
-   t->Setup[I830_TEXREG_TM0S2] = 
-      (((pitch / 4) - 1) << TM0S2_PITCH_SHIFT) |
-      TM0S2_CUBE_FACE_ENA_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MAX_MIP_MASK;
-   t->Setup[I830_TEXREG_TM0S3] &= ~TM0S3_MIN_MIP_MASK;
-   t->Setup[I830_TEXREG_TM0S3] |= ((numLevels - 1)*4) << TM0S3_MIN_MIP_SHIFT;
-   t->intel.dirty = I830_UPLOAD_TEX_ALL;
-
-   return intelUploadTexImages( &i830->intel, &t->intel, 0 );
 }
 
 
-static void i830_import_tex_unit( i830ContextPtr i830, 
-			   i830TextureObjectPtr t,
-			   GLuint unit )
+/* Recalculate all state from scratch.  Perhaps not the most
+ * efficient, but this has gotten complex enough that we need
+ * something which is understandable and reliable.
+ */
+static GLboolean
+i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 {
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
-      fprintf(stderr, "%s unit(%d)\n", __FUNCTION__, unit);
-   
-   if (i830->intel.CurrentTexObj[unit]) 
-      i830->intel.CurrentTexObj[unit]->base.bound &= ~(1U << unit);
-
-   i830->intel.CurrentTexObj[unit] = (intelTextureObjectPtr)t;
-   t->intel.base.bound |= (1 << unit);
-
-   I830_STATECHANGE( i830, I830_UPLOAD_TEX(unit) );
-
-   i830->state.Tex[unit][I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 | 
-					       (LOAD_TEXTURE_MAP0 << unit) | 4);
-   i830->state.Tex[unit][I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE |
-					       t->intel.TextureOffset);
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S1] = t->Setup[I830_TEXREG_TM0S1];
-   i830->state.Tex[unit][I830_TEXREG_TM0S2] = t->Setup[I830_TEXREG_TM0S2];
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S3] &= TM0S3_LOD_BIAS_MASK;
-   i830->state.Tex[unit][I830_TEXREG_TM0S3] |= (t->Setup[I830_TEXREG_TM0S3] &
-						~TM0S3_LOD_BIAS_MASK);
-
-   i830->state.Tex[unit][I830_TEXREG_TM0S4] = t->Setup[I830_TEXREG_TM0S4];
-   i830->state.Tex[unit][I830_TEXREG_MCS] = (t->Setup[I830_TEXREG_MCS] & 
-					     ~MAP_UNIT_MASK);   
-   i830->state.Tex[unit][I830_TEXREG_CUBE] = t->Setup[I830_TEXREG_CUBE];
-   i830->state.Tex[unit][I830_TEXREG_MCS] |= MAP_UNIT(unit);
-
-   t->intel.dirty &= ~I830_UPLOAD_TEX(unit);
-}
-
+   GLcontext *ctx = &intel->ctx;
+   struct i830_context *i830 = i830_context(ctx);
+   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   struct gl_texture_image *firstImage;
+   GLuint *state = i830->state.Tex[unit], format, pitch;
 
+   memset(state, 0, sizeof(state));
 
-static GLboolean enable_tex_common( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i830TextureObjectPtr t = (i830TextureObjectPtr)tObj->DriverData;
+   /*We need to refcount these. */
 
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
+   if (i830->state.tex_buffer[unit] != NULL) {
+       dri_bo_unreference(i830->state.tex_buffer[unit]);
+       i830->state.tex_buffer[unit] = NULL;
+   }
 
-   /* Fallback if there's a texture border */
-   if ( tObj->Image[0][tObj->BaseLevel]->Border > 0 ) {
-      fprintf(stderr, "Texture border\n");
+   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
       return GL_FALSE;
-   }
 
-   /* Upload teximages (not pipelined)
+   /* Get first image here, since intelObj->firstLevel will get set in
+    * the intel_finalize_mipmap_tree() call above.
     */
-   if (t->intel.base.dirty_images[0]) {
-      if (!i830SetTexImages( i830, tObj )) {
-	 return GL_FALSE;
+   firstImage = tObj->Image[0][intelObj->firstLevel];
+
+   if (intelObj->imageOverride) {
+      i830->state.tex_buffer[unit] = NULL;
+      i830->state.tex_offset[unit] = intelObj->textureOffset;
+
+      switch (intelObj->depthOverride) {
+      case 32:
+	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
+	 break;
+      case 24:
+      default:
+	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
+	 break;
+      case 16:
+	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
+	 break;
       }
-   }
 
-   /* Update state if this is a different texture object to last
-    * time.
-    */
-   if (i830->intel.CurrentTexObj[unit] != &t->intel || 
-       (t->intel.dirty & I830_UPLOAD_TEX(unit))) {
-      i830_import_tex_unit( i830, t, unit);
-   }
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
+      pitch = intelObj->pitchOverride;
+   } else {
+      dri_bo_reference(intelObj->mt->region->buffer);
+      i830->state.tex_buffer[unit] = intelObj->mt->region->buffer;
+      i830->state.tex_offset[unit] = intel_miptree_image_offset(intelObj->mt,
+								0, intelObj->
+								firstLevel);
 
-   return GL_TRUE;
-}
-
-static GLboolean enable_tex_rect( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
-
-   mcs &= ~TEXCOORDS_ARE_NORMAL;
-   mcs |= TEXCOORDS_ARE_IN_TEXELUNITS;
-
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (0 != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = 0;
+      format = translate_texture_format(firstImage->TexFormat->MesaFormat);
+      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
    }
 
-   return GL_TRUE;
-}
+   state[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
+                               (LOAD_TEXTURE_MAP0 << unit) | 4);
 
+/*    state[I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE | */
+/* 			       t->intel.TextureOffset); */
 
-static GLboolean enable_tex_2d( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
 
-   mcs &= ~TEXCOORDS_ARE_IN_TEXELUNITS;
-   mcs |= TEXCOORDS_ARE_NORMAL;
+   state[I830_TEXREG_TM0S1] =
+      (((firstImage->Height - 1) << TM0S1_HEIGHT_SHIFT) |
+       ((firstImage->Width - 1) << TM0S1_WIDTH_SHIFT) | format);
+
+   state[I830_TEXREG_TM0S2] =
+      ((((pitch / 4) - 1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
 
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (0 != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = 0;
+   {
+      if (tObj->Target == GL_TEXTURE_CUBE_MAP)
+         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit) |
+                                    CUBE_NEGX_ENABLE |
+                                    CUBE_POSX_ENABLE |
+                                    CUBE_NEGY_ENABLE |
+                                    CUBE_POSY_ENABLE |
+                                    CUBE_NEGZ_ENABLE | CUBE_POSZ_ENABLE);
+      else
+         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit));
    }
 
-   return GL_TRUE;
-}
 
- 
-static GLboolean enable_tex_cube( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i830TextureObjectPtr t = (i830TextureObjectPtr)tObj->DriverData;
-   GLuint mcs = i830->state.Tex[unit][I830_TEXREG_MCS];
-   const GLuint cube = CUBE_NEGX_ENABLE | CUBE_POSX_ENABLE
-     | CUBE_NEGY_ENABLE | CUBE_POSY_ENABLE
-     | CUBE_NEGZ_ENABLE | CUBE_POSZ_ENABLE;
-   GLuint face;
-
-   mcs &= ~TEXCOORDS_ARE_IN_TEXELUNITS;
-   mcs |= TEXCOORDS_ARE_NORMAL;
-
-   if ((mcs != i830->state.Tex[unit][I830_TEXREG_MCS])
-       || (cube != i830->state.Tex[unit][I830_TEXREG_CUBE])) {
-      I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-      i830->state.Tex[unit][I830_TEXREG_MCS] = mcs;
-      i830->state.Tex[unit][I830_TEXREG_CUBE] = cube;
-   }
 
-   /* Upload teximages (not pipelined)
-    */
-   if ( t->intel.base.dirty_images[0] || t->intel.base.dirty_images[1] ||
-        t->intel.base.dirty_images[2] || t->intel.base.dirty_images[3] ||
-        t->intel.base.dirty_images[4] || t->intel.base.dirty_images[5] ) {
-      i830SetTexImages( i830, tObj );
-   }
 
-   /* upload (per face) */
-   for (face = 0; face < 6; face++) {
-      if (t->intel.base.dirty_images[face]) {
-	 if (!intelUploadTexImages( &i830->intel, &t->intel, face )) {
-	    return GL_FALSE;
-	 }
+   {
+      GLuint minFilt, mipFilt, magFilt;
+
+      switch (tObj->MinFilter) {
+      case GL_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_NEAREST_MIPMAP_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_LINEAR_MIPMAP_NEAREST:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_NEAREST_MIPMAP_LINEAR:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      case GL_LINEAR_MIPMAP_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      default:
+         return GL_FALSE;
       }
-   }
 
+      if (tObj->MaxAnisotropy > 1.0) {
+         minFilt = FILTER_ANISOTROPIC;
+         magFilt = FILTER_ANISOTROPIC;
+      }
+      else {
+         switch (tObj->MagFilter) {
+         case GL_NEAREST:
+            magFilt = FILTER_NEAREST;
+            break;
+         case GL_LINEAR:
+            magFilt = FILTER_LINEAR;
+            break;
+         default:
+            return GL_FALSE;
+         }
+      }
 
-   return GL_TRUE;
-}
+      state[I830_TEXREG_TM0S3] = i830->lodbias_tm0s3[unit];
 
+#if 0
+      /* YUV conversion:
+       */
+      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
+          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
+         state[I830_TEXREG_TM0S3] |= SS2_COLORSPACE_CONVERSION;
+#endif
+
+      state[I830_TEXREG_TM0S3] |= ((intelObj->lastLevel -
+                                    intelObj->firstLevel) *
+                                   4) << TM0S3_MIN_MIP_SHIFT;
+
+      state[I830_TEXREG_TM0S3] |= ((minFilt << TM0S3_MIN_FILTER_SHIFT) |
+                                   (mipFilt << TM0S3_MIP_FILTER_SHIFT) |
+                                   (magFilt << TM0S3_MAG_FILTER_SHIFT));
+   }
 
-static GLboolean disable_tex( GLcontext *ctx, GLuint unit )
-{
-   i830ContextPtr i830 = I830_CONTEXT(ctx);
+   {
+      GLenum ws = tObj->WrapS;
+      GLenum wt = tObj->WrapT;
 
-   /* This is happening too often.  I need to conditionally send diffuse
-    * state to the card.  Perhaps a diffuse dirty flag of some kind.
-    * Will need to change this logic if more than 2 texture units are
-    * used.  We need to only do this up to the last unit enabled, or unit
-    * one if nothing is enabled.
-    */
 
-   if ( i830->intel.CurrentTexObj[unit] != NULL ) {
-      /* The old texture is no longer bound to this texture unit.
-       * Mark it as such.
+      /* 3D textures not available on i830
        */
-
-      i830->intel.CurrentTexObj[unit]->base.bound &= ~(1U << 0);
-      i830->intel.CurrentTexObj[unit] = NULL;
+      if (tObj->Target == GL_TEXTURE_3D)
+         return GL_FALSE;
+
+      state[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
+                                MAP_UNIT(unit) |
+                                ENABLE_TEXCOORD_PARAMS |
+                                ss3 |
+                                ENABLE_ADDR_V_CNTL |
+                                TEXCOORD_ADDR_V_MODE(translate_wrap_mode(wt))
+                                | ENABLE_ADDR_U_CNTL |
+                                TEXCOORD_ADDR_U_MODE(translate_wrap_mode
+                                                     (ws)));
    }
 
-   return GL_TRUE;
-}
 
-static GLboolean i830UpdateTexUnit( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
+   state[I830_TEXREG_TM0S4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
+                                                  tObj->_BorderChan[1],
+                                                  tObj->_BorderChan[2],
+                                                  tObj->_BorderChan[3]);
 
-   if (texUnit->_ReallyEnabled &&
-       INTEL_CONTEXT(ctx)->intelScreen->tex.size < 2048 * 1024)
-      return GL_FALSE;
 
-   switch(texUnit->_ReallyEnabled) {
-   case TEXTURE_1D_BIT:
-   case TEXTURE_2D_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_2d( ctx, unit ));
-   case TEXTURE_RECT_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_rect( ctx, unit ));
-   case TEXTURE_CUBE_BIT:
-      return (enable_tex_common( ctx, unit ) &&
-	      enable_tex_cube( ctx, unit ));
-   case 0:
-      return disable_tex( ctx, unit );
-   default:
-      return GL_FALSE;
-   }
+   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
+   /* memcmp was already disabled, but definitely won't work as the
+    * region might now change and that wouldn't be detected:
+    */
+   I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
+   return GL_TRUE;
 }
 
 
-void i830UpdateTextureState( intelContextPtr intel )
-{
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   GLcontext *ctx = &intel->ctx;
-   GLboolean ok;
-
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
 
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX_ALL, GL_FALSE);
 
-   ok = (i830UpdateTexUnit( ctx, 0 ) &&
-	 i830UpdateTexUnit( ctx, 1 ) &&
-	 i830UpdateTexUnit( ctx, 2 ) &&
-	 i830UpdateTexUnit( ctx, 3 ));
+void
+i830UpdateTextureState(struct intel_context *intel)
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLboolean ok = GL_TRUE;
+   GLuint i;
+
+   for (i = 0; i < I830_TEX_UNITS && ok; i++) {
+      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
+      case TEXTURE_1D_BIT:
+      case TEXTURE_2D_BIT:
+      case TEXTURE_CUBE_BIT:
+         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_NORMAL);
+         break;
+      case TEXTURE_RECT_BIT:
+         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_IN_TEXELUNITS);
+         break;
+      case 0:{
+	 struct i830_context *i830 = i830_context(&intel->ctx);
+         if (i830->state.active & I830_UPLOAD_TEX(i)) 
+            I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(i), GL_FALSE);
+
+	 if (i830->state.tex_buffer[i] != NULL) {
+	    dri_bo_unreference(i830->state.tex_buffer[i]);
+	    i830->state.tex_buffer[i] = NULL;
+	 }
+         break;
+      }
+      case TEXTURE_3D_BIT:
+      default:
+         ok = GL_FALSE;
+         break;
+      }
+   }
 
-   FALLBACK( intel, I830_FALLBACK_TEXTURE, !ok );
+   FALLBACK(intel, I830_FALLBACK_TEXTURE, !ok);
 
    if (ok)
-      i830EmitTextureBlend( i830 );
+      i830EmitTextureBlend(i830);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i915/i830_vtbl.c b/src/mesa/drivers/dri/i915/i830_vtbl.c
index d40cf705a35..eecff2729fd 100644
--- a/src/mesa/drivers/dri/i915/i830_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i830_vtbl.c
@@ -28,14 +28,15 @@
 
 #include "i830_context.h"
 #include "i830_reg.h"
-
 #include "intel_batchbuffer.h"
-
+#include "intel_regions.h"
 #include "tnl/t_context.h"
 #include "tnl/t_vertex.h"
 
-static GLboolean i830_check_vertex_size( intelContextPtr intel,
-					 GLuint expected );
+#define FILE_DEBUG_FLAG DEBUG_STATE
+
+static GLboolean i830_check_vertex_size(struct intel_context *intel,
+                                        GLuint expected);
 
 #define SZ_TO_HW(sz)  ((sz-2)&0x3)
 #define EMIT_SZ(sz)   (EMIT_1F + (sz) - 1)
@@ -59,10 +60,16 @@ do {									\
 #define VRTX_TEX_SET_FMT(n, x)          ((x)<<((n)*2))
 #define TEXBIND_SET(n, x) 		((x)<<((n)*4))
 
-static void i830_render_start( intelContextPtr intel )
+static void
+i830_render_prevalidate(struct intel_context *intel)
+{
+}
+
+static void
+i830_render_start(struct intel_context *intel)
 {
    GLcontext *ctx = &intel->ctx;
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
    DECLARE_RENDERINPUTS(index_bitset);
@@ -70,7 +77,7 @@ static void i830_render_start( intelContextPtr intel )
    GLuint v2 = _3DSTATE_VFT1_CMD;
    GLuint mcsb1 = 0;
 
-   RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
+   RENDERINPUTS_COPY(index_bitset, tnl->render_inputs_bitset);
 
    /* Important:
     */
@@ -80,196 +87,215 @@ static void i830_render_start( intelContextPtr intel )
    /* EMIT_ATTR's must be in order as they tell t_vertex.c how to
     * build up a hardware vertex.
     */
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, VFT0_XYZW );
+   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, VFT0_XYZW);
       intel->coloroffset = 4;
    }
    else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, VFT0_XYZ );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, VFT0_XYZ);
       intel->coloroffset = 3;
    }
 
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POINTSIZE )) {
-      EMIT_ATTR( _TNL_ATTRIB_POINTSIZE, EMIT_1F, VFT0_POINT_WIDTH );
+   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_POINTSIZE)) {
+      EMIT_ATTR(_TNL_ATTRIB_POINTSIZE, EMIT_1F, VFT0_POINT_WIDTH);
    }
 
-   EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, VFT0_DIFFUSE );
-      
+   EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, VFT0_DIFFUSE);
+
    intel->specoffset = 0;
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 ) ||
-       RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG )) {
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 )) {
+   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1) ||
+       RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG)) {
+      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1)) {
          intel->specoffset = intel->coloroffset + 1;
-         EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, VFT0_SPEC );
+         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, VFT0_SPEC);
       }
       else
-         EMIT_PAD( 3 );
+         EMIT_PAD(3);
 
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG ))
-         EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1UB_1F, VFT0_SPEC );
+      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG))
+         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, VFT0_SPEC);
       else
-         EMIT_PAD( 1 );
+         EMIT_PAD(1);
    }
 
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
+   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
       int i, count = 0;
 
       for (i = 0; i < I830_TEX_UNITS; i++) {
-         if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
+         if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_TEX(i))) {
             GLuint sz = VB->TexCoordPtr[i]->size;
             GLuint emit;
-            GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] & 
+            GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] &
                           ~TEXCOORDTYPE_MASK);
 
-	    switch (sz) {
-	    case 1: 
-	    case 2: 
-	       emit = EMIT_2F; 
-	       sz = 2; 
-	       mcs |= TEXCOORDTYPE_CARTESIAN; 
-	       break;
-	    case 3:
-	       emit = EMIT_3F; 
-	       sz = 3;
-	       mcs |= TEXCOORDTYPE_VECTOR;
-	       break;
-	    case 4: 
-	       emit = EMIT_3F_XYW; 
-	       sz = 3;     
-	       mcs |= TEXCOORDTYPE_HOMOGENEOUS;
-	       break;
-	    default: 
-	       continue;
-	    };
-	      
-
-	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, emit, 0 );	       
-	    v2 |= VRTX_TEX_SET_FMT(count, SZ_TO_HW(sz));
-	    mcsb1 |= (count+8)<<(i*4);
-
-	    if (mcs != i830->state.Tex[i][I830_TEXREG_MCS]) {
-	       I830_STATECHANGE(i830, I830_UPLOAD_TEX(i));
-	       i830->state.Tex[i][I830_TEXREG_MCS] = mcs;
-	    }
-
-	    count++;
-	 }
+            switch (sz) {
+            case 1:
+            case 2:
+               emit = EMIT_2F;
+               sz = 2;
+               mcs |= TEXCOORDTYPE_CARTESIAN;
+               break;
+            case 3:
+               emit = EMIT_3F;
+               sz = 3;
+               mcs |= TEXCOORDTYPE_VECTOR;
+               break;
+            case 4:
+               emit = EMIT_3F_XYW;
+               sz = 3;
+               mcs |= TEXCOORDTYPE_HOMOGENEOUS;
+               break;
+            default:
+               continue;
+            };
+
+
+            EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, emit, 0);
+            v2 |= VRTX_TEX_SET_FMT(count, SZ_TO_HW(sz));
+            mcsb1 |= (count + 8) << (i * 4);
+
+            if (mcs != i830->state.Tex[i][I830_TEXREG_MCS]) {
+               I830_STATECHANGE(i830, I830_UPLOAD_TEX(i));
+               i830->state.Tex[i][I830_TEXREG_MCS] = mcs;
+            }
+
+            count++;
+         }
       }
 
       v0 |= VFT0_TEX_COUNT(count);
    }
-   
+
    /* Only need to change the vertex emit code if there has been a
     * statechange to a new hardware vertex format:
     */
    if (v0 != i830->state.Ctx[I830_CTXREG_VF] ||
        v2 != i830->state.Ctx[I830_CTXREG_VF2] ||
        mcsb1 != i830->state.Ctx[I830_CTXREG_MCSB1] ||
-       !RENDERINPUTS_EQUAL( index_bitset, i830->last_index_bitset )) {
-    
-      I830_STATECHANGE( i830, I830_UPLOAD_CTX );
+       !RENDERINPUTS_EQUAL(index_bitset, i830->last_index_bitset)) {
+      int k;
+
+      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
 
       /* Must do this *after* statechange, so as not to affect
        * buffered vertices reliant on the old state:
        */
-      intel->vertex_size = 
-	 _tnl_install_attrs( ctx, 
-			     intel->vertex_attrs, 
-			     intel->vertex_attr_count,
-			     intel->ViewportMatrix.m, 0 );
+      intel->vertex_size =
+         _tnl_install_attrs(ctx,
+                            intel->vertex_attrs,
+                            intel->vertex_attr_count,
+                            intel->ViewportMatrix.m, 0);
 
       intel->vertex_size >>= 2;
 
       i830->state.Ctx[I830_CTXREG_VF] = v0;
       i830->state.Ctx[I830_CTXREG_VF2] = v2;
       i830->state.Ctx[I830_CTXREG_MCSB1] = mcsb1;
-      RENDERINPUTS_COPY( i830->last_index_bitset, index_bitset );
+      RENDERINPUTS_COPY(i830->last_index_bitset, index_bitset);
 
-      assert(i830_check_vertex_size( intel, intel->vertex_size ));
+      k = i830_check_vertex_size(intel, intel->vertex_size);
+      assert(k);
    }
 }
 
-static void i830_reduced_primitive_state( intelContextPtr intel,
-					  GLenum rprim )
+static void
+i830_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
 {
-    i830ContextPtr i830 = I830_CONTEXT(intel);
-    GLuint st1 = i830->state.Stipple[I830_STPREG_ST1];
-
-    st1 &= ~ST1_ENABLE;
-
-    switch (rprim) {
-    case GL_TRIANGLES:
-       if (intel->ctx.Polygon.StippleFlag &&
-	   intel->hw_stipple)
-	  st1 |= ST1_ENABLE;
-       break;
-    case GL_LINES:
-    case GL_POINTS:
-    default:
-       break;
-    }
-
-    i830->intel.reduced_primitive = rprim;
-
-    if (st1 != i830->state.Stipple[I830_STPREG_ST1]) {
-       I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-       i830->state.Stipple[I830_STPREG_ST1] = st1;
-    }
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLuint st1 = i830->state.Stipple[I830_STPREG_ST1];
+
+   st1 &= ~ST1_ENABLE;
+
+   switch (rprim) {
+   case GL_TRIANGLES:
+      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
+         st1 |= ST1_ENABLE;
+      break;
+   case GL_LINES:
+   case GL_POINTS:
+   default:
+      break;
+   }
+
+   i830->intel.reduced_primitive = rprim;
+
+   if (st1 != i830->state.Stipple[I830_STPREG_ST1]) {
+      INTEL_FIREVERTICES(intel);
+
+      I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
+      i830->state.Stipple[I830_STPREG_ST1] = st1;
+   }
 }
 
 /* Pull apart the vertex format registers and figure out how large a
  * vertex is supposed to be. 
  */
-static GLboolean i830_check_vertex_size( intelContextPtr intel,
-					 GLuint expected )
+static GLboolean
+i830_check_vertex_size(struct intel_context *intel, GLuint expected)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(&intel->ctx);
    int vft0 = i830->current->Ctx[I830_CTXREG_VF];
    int vft1 = i830->current->Ctx[I830_CTXREG_VF2];
    int nrtex = (vft0 & VFT0_TEX_COUNT_MASK) >> VFT0_TEX_COUNT_SHIFT;
    int i, sz = 0;
 
    switch (vft0 & VFT0_XYZW_MASK) {
-   case VFT0_XY: sz = 2; break;
-   case VFT0_XYZ: sz = 3; break;
-   case VFT0_XYW: sz = 3; break;
-   case VFT0_XYZW: sz = 4; break;
-   default: 
+   case VFT0_XY:
+      sz = 2;
+      break;
+   case VFT0_XYZ:
+      sz = 3;
+      break;
+   case VFT0_XYW:
+      sz = 3;
+      break;
+   case VFT0_XYZW:
+      sz = 4;
+      break;
+   default:
       fprintf(stderr, "no xyzw specified\n");
       return 0;
    }
 
-   if (vft0 & VFT0_SPEC) sz++;
-   if (vft0 & VFT0_DIFFUSE) sz++;
-   if (vft0 & VFT0_DEPTH_OFFSET) sz++;
-   if (vft0 & VFT0_POINT_WIDTH) sz++;
-	
-   for (i = 0 ; i < nrtex ; i++) { 
+   if (vft0 & VFT0_SPEC)
+      sz++;
+   if (vft0 & VFT0_DIFFUSE)
+      sz++;
+   if (vft0 & VFT0_DEPTH_OFFSET)
+      sz++;
+   if (vft0 & VFT0_POINT_WIDTH)
+      sz++;
+
+   for (i = 0; i < nrtex; i++) {
       switch (vft1 & VFT1_TEX0_MASK) {
-      case TEXCOORDFMT_2D: sz += 2; break;
-      case TEXCOORDFMT_3D: sz += 3; break;
-      case TEXCOORDFMT_4D: sz += 4; break;
-      case TEXCOORDFMT_1D: sz += 1; break;
+      case TEXCOORDFMT_2D:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_3D:
+         sz += 3;
+         break;
+      case TEXCOORDFMT_4D:
+         sz += 4;
+         break;
+      case TEXCOORDFMT_1D:
+         sz += 1;
+         break;
       }
       vft1 >>= VFT1_TEX1_SHIFT;
    }
-	
-   if (sz != expected) 
+
+   if (sz != expected)
       fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-   
+
    return sz == expected;
 }
 
-static void i830_emit_invarient_state( intelContextPtr intel )
+static void
+i830_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH( 40 );
-
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(0));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(1));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(2));
-   OUT_BATCH(_3DSTATE_MAP_CUBE | MAP_UNIT(3));
+   BEGIN_BATCH(40, 0);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -282,37 +308,35 @@ static void i830_emit_invarient_state( intelContextPtr intel )
 
    OUT_BATCH(_3DSTATE_FOG_MODE_CMD);
    OUT_BATCH(FOGFUNC_ENABLE |
-	     FOG_LINEAR_CONST | 
-	     FOGSRC_INDEX_Z | 
-	     ENABLE_FOG_DENSITY);
+             FOG_LINEAR_CONST | FOGSRC_INDEX_Z | ENABLE_FOG_DENSITY);
    OUT_BATCH(0);
    OUT_BATCH(0);
 
 
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(0) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(0) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(0));
+             MAP_UNIT(0) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(0) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(0));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(1) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(1) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(1));
+             MAP_UNIT(1) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(1) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(1));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(2) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(2) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(2));
+             MAP_UNIT(2) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(2) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(2));
    OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-	     MAP_UNIT(3) |
-	     DISABLE_TEX_STREAM_BUMP |
-	     ENABLE_TEX_STREAM_COORD_SET |
-	     TEX_STREAM_COORD_SET(3) |
-	     ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(3));
+             MAP_UNIT(3) |
+             DISABLE_TEX_STREAM_BUMP |
+             ENABLE_TEX_STREAM_COORD_SET |
+             TEX_STREAM_COORD_SET(3) |
+             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(3));
 
    OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
    OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(0));
@@ -324,21 +348,13 @@ static void i830_emit_invarient_state( intelContextPtr intel )
    OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(3));
 
    OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-	     ENABLE_POINT_RASTER_RULE |
-	     OGL_POINT_RASTER_RULE |
-	     ENABLE_LINE_STRIP_PROVOKE_VRTX |
-	     ENABLE_TRI_FAN_PROVOKE_VRTX |
-	     ENABLE_TRI_STRIP_PROVOKE_VRTX |
-	     LINE_STRIP_PROVOKE_VRTX(1) |
-	     TRI_FAN_PROVOKE_VRTX(2) | 
-	     TRI_STRIP_PROVOKE_VRTX(2));
-
-   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | 
-	     DISABLE_SCISSOR_RECT);
-
-   OUT_BATCH(_3DSTATE_SCISSOR_RECT_0_CMD);
-   OUT_BATCH(0);
-   OUT_BATCH(0);
+             ENABLE_POINT_RASTER_RULE |
+             OGL_POINT_RASTER_RULE |
+             ENABLE_LINE_STRIP_PROVOKE_VRTX |
+             ENABLE_TRI_FAN_PROVOKE_VRTX |
+             ENABLE_TRI_STRIP_PROVOKE_VRTX |
+             LINE_STRIP_PROVOKE_VRTX(1) |
+             TRI_FAN_PROVOKE_VRTX(2) | TRI_STRIP_PROVOKE_VRTX(2));
 
    OUT_BATCH(_3DSTATE_VERTEX_TRANSFORM);
    OUT_BATCH(DISABLE_VIEWPORT_TRANSFORM | DISABLE_PERSPECTIVE_DIVIDE);
@@ -349,7 +365,7 @@ static void i830_emit_invarient_state( intelContextPtr intel )
 
 
    OUT_BATCH(_3DSTATE_COLOR_FACTOR_CMD);
-   OUT_BATCH(0x80808080);	/* .5 required in alpha for GL_DOT3_RGBA_EXT */
+   OUT_BATCH(0x80808080);       /* .5 required in alpha for GL_DOT3_RGBA_EXT */
 
    ADVANCE_BATCH();
 }
@@ -358,13 +374,16 @@ static void i830_emit_invarient_state( intelContextPtr intel )
 #define emit( intel, state, size )			\
 do {							\
    int k;						\
-   BEGIN_BATCH( size / sizeof(GLuint));			\
-   for (k = 0 ; k < size / sizeof(GLuint) ; k++)	\
+   BEGIN_BATCH(size / sizeof(GLuint), 0);		\
+   for (k = 0 ; k < size / sizeof(GLuint) ; k++) {	\
+      if (0) _mesa_printf("  0x%08x\n", state[k]);		\
       OUT_BATCH(state[k]);				\
+   }							\
    ADVANCE_BATCH();					\
-} while (0);
+} while (0)
 
-static GLuint get_state_size( struct i830_hw_state *state )
+static GLuint
+get_state_size(struct i830_hw_state *state)
 {
    GLuint dirty = state->active & ~state->emitted;
    GLuint sz = 0;
@@ -373,21 +392,21 @@ static GLuint get_state_size( struct i830_hw_state *state )
    if (dirty & I830_UPLOAD_INVARIENT)
       sz += 40 * sizeof(int);
 
-   if (dirty & I830_UPLOAD_CTX) 
+   if (dirty & I830_UPLOAD_CTX)
       sz += sizeof(state->Ctx);
 
-   if (dirty & I830_UPLOAD_BUFFERS) 
+   if (dirty & I830_UPLOAD_BUFFERS)
       sz += sizeof(state->Buffer);
 
-   if (dirty & I830_UPLOAD_STIPPLE) 
+   if (dirty & I830_UPLOAD_STIPPLE)
       sz += sizeof(state->Stipple);
 
    for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i)))  
-	 sz += sizeof(state->Tex[i]); 
+      if ((dirty & I830_UPLOAD_TEX(i)))
+         sz += sizeof(state->Tex[i]);
 
-      if (dirty & I830_UPLOAD_TEXBLEND(i)) 
-	 sz += state->TexBlendWordsUsed[i] * 4;
+      if (dirty & I830_UPLOAD_TEXBLEND(i))
+         sz += state->TexBlendWordsUsed[i] * 4;
    }
 
    return sz;
@@ -396,139 +415,260 @@ static GLuint get_state_size( struct i830_hw_state *state )
 
 /* Push the state into the sarea and/or texture memory.
  */
-static void i830_emit_state( intelContextPtr intel )
+static void
+i830_emit_state(struct intel_context *intel)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
+   struct i830_context *i830 = i830_context(&intel->ctx);
    struct i830_hw_state *state = i830->current;
    int i;
-   GLuint dirty = state->active & ~state->emitted;
-   GLuint counter = intel->batch.counter;
+   GLuint dirty;
    BATCH_LOCALS;
 
-   if (intel->batch.space < get_state_size(state)) {
-      intelFlushBatch(intel, GL_TRUE);
-      dirty = state->active & ~state->emitted;
-      counter = intel->batch.counter;
-   }
+   /* We don't hold the lock at this point, so want to make sure that
+    * there won't be a buffer wrap.  
+    *
+    * It might be better to talk about explicit places where
+    * scheduling is allowed, rather than assume that it is whenever a
+    * batchbuffer fills up.
+    */
+   intel_batchbuffer_require_space(intel->batch, get_state_size(state), 0);
+
+   /* Do this here as we may have flushed the batchbuffer above,
+    * causing more state to be dirty!
+    */
+   dirty = state->active & ~state->emitted;
 
    if (dirty & I830_UPLOAD_INVARIENT) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_INVARIENT:\n"); 
-      i830_emit_invarient_state( intel );
+      DBG("I830_UPLOAD_INVARIENT:\n");
+      i830_emit_invarient_state(intel);
    }
 
    if (dirty & I830_UPLOAD_CTX) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_CTX:\n"); 
-      emit( i830, state->Ctx, sizeof(state->Ctx) );
+      DBG("I830_UPLOAD_CTX:\n");
+      emit(i830, state->Ctx, sizeof(state->Ctx));
+
    }
 
    if (dirty & I830_UPLOAD_BUFFERS) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_BUFFERS:\n"); 
-      emit( i830, state->Buffer, sizeof(state->Buffer) );
-   }
+      DBG("I830_UPLOAD_BUFFERS:\n");
+      BEGIN_BATCH(I830_DEST_SETUP_SIZE + 2, 0);
+      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR1]);
+      OUT_RELOC(state->draw_region->buffer,
+                DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
+                state->draw_region->draw_offset);
+
+      if (state->depth_region) {
+         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR0]);
+         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR1]);
+         OUT_RELOC(state->depth_region->buffer,
+                   DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
+                   state->depth_region->draw_offset);
+      }
 
+      OUT_BATCH(state->Buffer[I830_DESTREG_DV0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_DV1]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SENABLE]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR0]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR1]);
+      OUT_BATCH(state->Buffer[I830_DESTREG_SR2]);
+      ADVANCE_BATCH();
+   }
+   
    if (dirty & I830_UPLOAD_STIPPLE) {
-      if (VERBOSE) fprintf(stderr, "I830_UPLOAD_STIPPLE:\n"); 
-      emit( i830, state->Stipple, sizeof(state->Stipple) );
+      DBG("I830_UPLOAD_STIPPLE:\n");
+      emit(i830, state->Stipple, sizeof(state->Stipple));
    }
 
    for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i))) { 
- 	 if (VERBOSE) fprintf(stderr, "I830_UPLOAD_TEX(%d):\n", i); 
-	 emit( i830, state->Tex[i], sizeof(state->Tex[i])); 
-      } 
+      if ((dirty & I830_UPLOAD_TEX(i))) {
+         DBG("I830_UPLOAD_TEX(%d):\n", i);
+
+         BEGIN_BATCH(I830_TEX_SETUP_SIZE + 1, 0);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0LI]);
+
+         if (state->tex_buffer[i]) {
+            OUT_RELOC(state->tex_buffer[i],
+                      DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
+                      state->tex_offset[i] | TM0S0_USE_FENCE);
+         }
+	 else if (state == &i830->meta) {
+	    assert(i == 0);
+	    OUT_BATCH(0);
+	 }
+	 else {
+	    OUT_BATCH(state->tex_offset[i]);
+	 }
+
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S1]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S2]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S3]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S4]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_MCS]);
+         OUT_BATCH(state->Tex[i][I830_TEXREG_CUBE]);
+      }
 
       if (dirty & I830_UPLOAD_TEXBLEND(i)) {
-	 if (VERBOSE) fprintf(stderr, "I830_UPLOAD_TEXBLEND(%d):\n", i); 
-	 emit( i830, state->TexBlend[i], 
-	       state->TexBlendWordsUsed[i] * 4 );
+         DBG("I830_UPLOAD_TEXBLEND(%d): %d words\n", i,
+             state->TexBlendWordsUsed[i]);
+         emit(i830, state->TexBlend[i], state->TexBlendWordsUsed[i] * 4);
       }
    }
 
    state->emitted |= dirty;
-   intel->batch.last_emit_state = counter;
-   assert(counter == intel->batch.counter);
 }
 
-static void i830_destroy_context( intelContextPtr intel )
+static void
+i830_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+
+   for (i = 0; i < I830_TEX_UNITS; i++) {
+      if (i830->state.tex_buffer[i] != NULL) {
+	 dri_bo_unreference(i830->state.tex_buffer[i]);
+	 i830->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
-static void
-i830_set_color_region(intelContextPtr intel, const intelRegion *region)
+
+void
+i830_state_draw_region(struct intel_context *intel,
+		       struct i830_hw_state *state,
+		       struct intel_region *color_region,
+		       struct intel_region *depth_region)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   I830_STATECHANGE( i830, I830_UPLOAD_BUFFERS );
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_CBUFADDR2] = region->offset;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   GLuint value;
+
+   ASSERT(state == &i830->state || state == &i830->meta);
+
+   if (state->draw_region != color_region) {
+      intel_region_release(&state->draw_region);
+      intel_region_reference(&state->draw_region, color_region);
+   }
+   if (state->depth_region != depth_region) {
+      intel_region_release(&state->depth_region);
+      intel_region_reference(&state->depth_region, depth_region);
+   }
+
+   /*
+    * Set stride/cpp values
+    */
+   if (color_region) {
+      state->Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I830_DESTREG_CBUFADDR1] =
+         (BUF_3D_ID_COLOR_BACK |
+          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   if (depth_region) {
+      state->Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I830_DESTREG_DBUFADDR1] =
+         (BUF_3D_ID_DEPTH |
+          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   /*
+    * Compute/set I830_DESTREG_DV1 value
+    */
+   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
+            DSTORG_VERT_BIAS(0x8) | DEPTH_IS_Z);    /* .5 */
+            
+   if (color_region && color_region->cpp == 4) {
+      value |= DV_PF_8888;
+   }
+   else {
+      value |= DV_PF_565;
+   }
+   if (depth_region && depth_region->cpp == 4) {
+      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
+   }
+   else {
+      value |= DEPTH_FRMT_16_FIXED;
+   }
+   state->Buffer[I830_DESTREG_DV1] = value;
+
+   I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
+
+
 }
 
 
 static void
-i830_set_z_region(intelContextPtr intel, const intelRegion *region)
+i830_set_draw_region(struct intel_context *intel,
+                     struct intel_region *color_region,
+                     struct intel_region *depth_region)
 {
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-   I830_STATECHANGE( i830, I830_UPLOAD_BUFFERS );
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_DBUFADDR2] = region->offset;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830_state_draw_region(intel, &i830->state, color_region, depth_region);
 }
 
-
+#if 0
 static void
 i830_update_color_z_regions(intelContextPtr intel,
-                            const intelRegion *colorRegion,
-                            const intelRegion *depthRegion)
+                            const intelRegion * colorRegion,
+                            const intelRegion * depthRegion)
 {
    i830ContextPtr i830 = I830_CONTEXT(intel);
 
    i830->state.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) | BUF_3D_USE_FENCE);
+      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) |
+       BUF_3D_USE_FENCE);
    i830->state.Buffer[I830_DESTREG_CBUFADDR2] = colorRegion->offset;
 
    i830->state.Buffer[I830_DESTREG_DBUFADDR1] =
       (BUF_3D_ID_DEPTH | BUF_3D_PITCH(depthRegion->pitch) | BUF_3D_USE_FENCE);
    i830->state.Buffer[I830_DESTREG_DBUFADDR2] = depthRegion->offset;
 }
+#endif
 
 
 /* This isn't really handled at the moment.
  */
-static void i830_lost_hardware( intelContextPtr intel )
+static void
+i830_lost_hardware(struct intel_context *intel)
 {
-   I830_CONTEXT(intel)->state.emitted = 0;
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   i830->state.emitted = 0;
 }
 
 
 
-static void i830_emit_flush( intelContextPtr intel )
+static GLuint
+i830_flush_cmd(void)
 {
-   BATCH_LOCALS;
-
-   BEGIN_BATCH(2);
-   OUT_BATCH( MI_FLUSH | FLUSH_MAP_CACHE ); 
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
+   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
 
+static void 
+i830_assert_not_dirty( struct intel_context *intel )
+{
+   struct i830_context *i830 = i830_context(&intel->ctx);
+   struct i830_hw_state *state = i830->current;
+   GLuint dirty = state->active & ~state->emitted;
+   assert(!dirty);
+}
 
 
-void i830InitVtbl( i830ContextPtr i830 )
+void
+i830InitVtbl(struct i830_context *i830)
 {
-   i830->intel.vtbl.alloc_tex_obj = i830AllocTexObj;
    i830->intel.vtbl.check_vertex_size = i830_check_vertex_size;
-   i830->intel.vtbl.clear_with_tris = i830ClearWithTris;
-   i830->intel.vtbl.rotate_window = i830RotateWindow;
    i830->intel.vtbl.destroy = i830_destroy_context;
    i830->intel.vtbl.emit_state = i830_emit_state;
    i830->intel.vtbl.lost_hardware = i830_lost_hardware;
    i830->intel.vtbl.reduced_primitive_state = i830_reduced_primitive_state;
-   i830->intel.vtbl.set_color_region = i830_set_color_region;
-   i830->intel.vtbl.set_z_region = i830_set_z_region;
-   i830->intel.vtbl.update_color_z_regions = i830_update_color_z_regions;
+   i830->intel.vtbl.set_draw_region = i830_set_draw_region;
    i830->intel.vtbl.update_texture_state = i830UpdateTextureState;
-   i830->intel.vtbl.emit_flush = i830_emit_flush;
+   i830->intel.vtbl.flush_cmd = i830_flush_cmd;
    i830->intel.vtbl.render_start = i830_render_start;
+   i830->intel.vtbl.render_prevalidate = i830_render_prevalidate;
+   i830->intel.vtbl.assert_not_dirty = i830_assert_not_dirty;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index 2bc1cae9c31..49e30141e43 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -36,84 +36,93 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "vbo/vbo.h"
-
 
 #include "utils.h"
 #include "i915_reg.h"
 
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
+
 /***************************************
  * 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_env_crossbar",       NULL },
-    { "GL_EXT_shadow_funcs",               NULL },
-    /* ARB extn won't work if not enabled */
-    { "GL_SGIX_depth_texture",             NULL },
-    { NULL,                                NULL }
+static const struct dri_extension i915_extensions[] = {
+   {"GL_ARB_depth_texture", NULL},
+   {"GL_ARB_fragment_program", NULL},
+   {"GL_ARB_shadow", NULL},
+   {"GL_ARB_texture_env_crossbar", NULL},
+   {"GL_ARB_texture_non_power_of_two", NULL},
+   {"GL_EXT_shadow_funcs", NULL},
+   /* ARB extn won't work if not enabled */
+   {"GL_SGIX_depth_texture", NULL},
+   {NULL, NULL}
 };
 
 /* Override intel default.
  */
-static void i915InvalidateState( GLcontext *ctx, GLuint new_state )
+static void
+i915InvalidateState(GLcontext * ctx, GLuint new_state)
 {
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   _tnl_invalidate_vertex_state( ctx, new_state );
-   INTEL_CONTEXT(ctx)->NewGLState |= new_state;
+   _swrast_InvalidateState(ctx, new_state);
+   _swsetup_InvalidateState(ctx, new_state);
+   _vbo_InvalidateState(ctx, new_state);
+   _tnl_InvalidateState(ctx, new_state);
+   _tnl_invalidate_vertex_state(ctx, new_state);
+   intel_context(ctx)->NewGLState |= new_state;
 
    /* Todo: gather state values under which tracked parameters become
     * invalidated, add callbacks for things like
     * ProgramLocalParameters, etc.
     */
    {
-      struct i915_fragment_program *p = 
-	 (struct i915_fragment_program *)ctx->FragmentProgram._Current;
+      struct i915_fragment_program *p =
+         (struct i915_fragment_program *) ctx->FragmentProgram._Current;
       if (p && p->nr_params)
-	 p->params_uptodate = 0;
+         p->params_uptodate = 0;
    }
 
-   if (new_state & (_NEW_FOG|_NEW_HINT|_NEW_PROGRAM))
+   if (new_state & (_NEW_FOG | _NEW_HINT | _NEW_PROGRAM))
       i915_update_fog(ctx);
 }
 
 
-static void i915InitDriverFunctions( struct dd_function_table *functions )
+static void
+i915InitDriverFunctions(struct dd_function_table *functions)
 {
-   intelInitDriverFunctions( functions );
-   i915InitStateFunctions( functions );
-   i915InitTextureFuncs( functions );
-   i915InitFragProgFuncs( functions );
+   intelInitDriverFunctions(functions);
+   i915InitStateFunctions(functions);
+   i915InitTextureFuncs(functions);
+   i915InitFragProgFuncs(functions);
    functions->UpdateState = i915InvalidateState;
 }
 
 
 
-GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate)
+GLboolean
+i915CreateContext(const __GLcontextModes * mesaVis,
+                  __DRIcontextPrivate * driContextPriv,
+                  void *sharedContextPrivate)
 {
    struct dd_function_table functions;
-   i915ContextPtr i915 = (i915ContextPtr) CALLOC_STRUCT(i915_context);
-   intelContextPtr intel = &i915->intel;
+   struct i915_context *i915 =
+      (struct i915_context *) CALLOC_STRUCT(i915_context);
+   struct intel_context *intel = &i915->intel;
    GLcontext *ctx = &intel->ctx;
-   GLuint i;
 
-   if (!i915) return GL_FALSE;
+   if (!i915)
+      return GL_FALSE;
+
+   if (0)
+      _mesa_printf("\ntexmem-0-3 branch\n\n");
 
-   i915InitVtbl( i915 );
+   i915InitVtbl(i915);
+   i915InitMetaFuncs(i915);
 
-   i915InitDriverFunctions( &functions );
+   i915InitDriverFunctions(&functions);
 
-   if (!intelInitContext( intel, mesaVis, driContextPriv,
-			  sharedContextPrivate, &functions )) {
+   if (!intelInitContext(intel, mesaVis, driContextPriv,
+                         sharedContextPrivate, &functions)) {
       FREE(i915);
       return GL_FALSE;
    }
@@ -122,65 +131,44 @@ GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
    ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS;
    ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS;
 
-   intel->nr_heaps = 1;
-   intel->texture_heaps[0] = 
-      driCreateTextureHeap( 0, intel,
-			    intel->intelScreen->tex.size,
-			    12,
-			    I830_NR_TEX_REGIONS,
-			    intel->sarea->texList,
-			    (unsigned *) & intel->sarea->texAge,
-			    & intel->swapped,
-			    sizeof( struct i915_texture_object ),
-			    (destroy_texture_object_t *)intelDestroyTexObj );
-
-   /* FIXME: driCalculateMaxTextureLevels assumes that mipmaps are
-    * tightly packed, but they're not in Intel graphics
-    * hardware.
+
+   /* Advertise the full hardware capabilities.  The new memory
+    * manager should cope much better with overload situations:
     */
+   ctx->Const.MaxTextureLevels = 12;
+   ctx->Const.Max3DTextureLevels = 9;
+   ctx->Const.MaxCubeTextureLevels = 12;
+   ctx->Const.MaxTextureRectSize = (1 << 11);
    ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
-   i = driQueryOptioni( &intel->optionCache, "allow_large_textures");
-   driCalculateMaxTextureLevels( intel->texture_heaps,
-				 intel->nr_heaps,
-				 &intel->ctx.Const,
-				 4,
-				 11, /* max 2D texture size is 2048x2048 */
-				 8,  /* 3D texture */
-				 11, /* cube texture. */
-				 11, /* rect texture */
-				 12,
-				 GL_FALSE,
-				 i );
 
    /* GL_ARB_fragment_program limits - don't think Mesa actually
     * validates programs against these, and in any case one ARB
     * instruction can translate to more than one HW instruction, so
     * we'll still have to check and fallback each time.
     */
-   
    ctx->Const.FragmentProgram.MaxNativeTemps = I915_MAX_TEMPORARY;
-   ctx->Const.FragmentProgram.MaxNativeAttribs = 11; /* 8 tex, 2 color, fog */
+   ctx->Const.FragmentProgram.MaxNativeAttribs = 11;    /* 8 tex, 2 color, fog */
    ctx->Const.FragmentProgram.MaxNativeParameters = I915_MAX_CONSTANT;
    ctx->Const.FragmentProgram.MaxNativeAluInstructions = I915_MAX_ALU_INSN;
    ctx->Const.FragmentProgram.MaxNativeTexInstructions = I915_MAX_TEX_INSN;
-   ctx->Const.FragmentProgram.MaxNativeInstructions = (I915_MAX_ALU_INSN + 
-						I915_MAX_TEX_INSN);
-   ctx->Const.FragmentProgram.MaxNativeTexIndirections = I915_MAX_TEX_INDIRECT;
+   ctx->Const.FragmentProgram.MaxNativeInstructions = (I915_MAX_ALU_INSN +
+                                                       I915_MAX_TEX_INSN);
+   ctx->Const.FragmentProgram.MaxNativeTexIndirections =
+      I915_MAX_TEX_INDIRECT;
    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);
 
-   driInitExtensions( ctx, i915_extensions, GL_FALSE );
 
-
-   _tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-		       36 * sizeof(GLfloat) );
+   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
+                      36 * sizeof(GLfloat));
 
    intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
 
-   i915InitState( i915 );
+   i915InitState(i915);
 
    return GL_TRUE;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i915_context.h b/src/mesa/drivers/dri/i915/i915_context.h
index ec1550126a6..1070de1a540 100644
--- a/src/mesa/drivers/dri/i915/i915_context.h
+++ b/src/mesa/drivers/dri/i915/i915_context.h
@@ -46,6 +46,7 @@
 #define I915_UPLOAD_CONSTANTS        0x10
 #define I915_UPLOAD_FOG              0x20
 #define I915_UPLOAD_INVARIENT        0x40
+#define I915_UPLOAD_DEFAULTS         0x80
 #define I915_UPLOAD_TEX(i)           (0x00010000<<(i))
 #define I915_UPLOAD_TEX_ALL          (0x00ff0000)
 #define I915_UPLOAD_TEX_0_SHIFT      16
@@ -55,10 +56,8 @@
  */
 #define I915_DESTREG_CBUFADDR0 0
 #define I915_DESTREG_CBUFADDR1 1
-#define I915_DESTREG_CBUFADDR2 2
 #define I915_DESTREG_DBUFADDR0 3
 #define I915_DESTREG_DBUFADDR1 4
-#define I915_DESTREG_DBUFADDR2 5
 #define I915_DESTREG_DV0 6
 #define I915_DESTREG_DV1 7
 #define I915_DESTREG_SENABLE 8
@@ -89,7 +88,6 @@
 #define I915_STPREG_ST1        1
 #define I915_STP_SETUP_SIZE    2
 
-#define I915_TEXREG_MS2        0
 #define I915_TEXREG_MS3        1
 #define I915_TEXREG_MS4        2
 #define I915_TEXREG_SS2        3
@@ -97,6 +95,15 @@
 #define I915_TEXREG_SS4        5
 #define I915_TEX_SETUP_SIZE    6
 
+#define I915_DEFREG_C0    0
+#define I915_DEFREG_C1    1
+#define I915_DEFREG_S0    2
+#define I915_DEFREG_S1    3
+#define I915_DEFREG_Z0    4
+#define I915_DEFREG_Z1    5
+#define I915_DEF_SETUP_SIZE    6
+
+
 #define I915_MAX_CONSTANT      32
 #define I915_CONSTANT_SIZE     (2+(4*I915_MAX_CONSTANT))
 
@@ -107,13 +114,14 @@
 /* Hardware version of a parsed fragment program.  "Derived" from the
  * mesa fragment_program struct.
  */
-struct i915_fragment_program {
+struct i915_fragment_program
+{
    struct gl_fragment_program FragProg;
 
    GLboolean translated;
    GLboolean params_uptodate;
    GLboolean on_hardware;
-   GLboolean error;		/* If program is malformed for any reason. */
+   GLboolean error;             /* If program is malformed for any reason. */
 
    GLuint nr_tex_indirect;
    GLuint nr_tex_insn;
@@ -135,22 +143,22 @@ struct i915_fragment_program {
    GLuint constant_flags[I915_MAX_CONSTANT];
    GLuint nr_constants;
 
-   GLuint *csr;			/* Cursor, points into program.
-				 */
+   GLuint *csr;                 /* Cursor, points into program.
+                                 */
 
-   GLuint *decl;		/* Cursor, points into declarations.
-				 */
-   
-   GLuint decl_s;		/* flags for which s regs need to be decl'd */
-   GLuint decl_t;		/* flags for which t regs need to be decl'd */
+   GLuint *decl;                /* Cursor, points into declarations.
+                                 */
 
-   GLuint temp_flag;		/* Tracks temporary regs which are in
-				 * use.
-				 */
+   GLuint decl_s;               /* flags for which s regs need to be decl'd */
+   GLuint decl_t;               /* flags for which t regs need to be decl'd */
 
-   GLuint utemp_flag;		/* Tracks TYPE_U temporary regs which are in
-				 * use.
-				 */
+   GLuint temp_flag;            /* Tracks temporary regs which are in
+                                 * use.
+                                 */
+
+   GLuint utemp_flag;           /* Tracks TYPE_U temporary regs which are in
+                                 * use.
+                                 */
 
 
 
@@ -159,28 +167,12 @@ struct i915_fragment_program {
    GLuint wpos_tex;
    GLboolean depth_written;
 
-   struct { 
-      GLuint reg;		/* Hardware constant idx */
-      const GLfloat *values; 	/* Pointer to tracked values */
+   struct
+   {
+      GLuint reg;               /* Hardware constant idx */
+      const GLfloat *values;    /* Pointer to tracked values */
    } param[I915_MAX_CONSTANT];
    GLuint nr_params;
-      
-
-
-
-   /* Helpers for i915_texprog.c:
-    */
-   GLuint src_texture;		/* Reg containing sampled texture color,
-				 * else UREG_BAD.
-				 */
-
-   GLuint src_previous;		/* Reg containing color from previous 
-				 * stage.  May need to be decl'd.
-				 */
-
-   GLuint last_tex_stage;	/* Number of last enabled texture unit */
-
-   struct vertex_buffer *VB;
 };
 
 
@@ -188,67 +180,68 @@ struct i915_fragment_program {
 
 
 
-struct i915_texture_object
-{
-   struct intel_texture_object intel;
-   GLenum lastTarget;
-   GLboolean refs_border_color;
-   GLuint Setup[I915_TEX_SETUP_SIZE];
-};
 
 #define I915_TEX_UNITS 8
 
 
-struct i915_hw_state {
+struct i915_hw_state
+{
    GLuint Ctx[I915_CTX_SETUP_SIZE];
    GLuint Buffer[I915_DEST_SETUP_SIZE];
    GLuint Stipple[I915_STP_SETUP_SIZE];
    GLuint Fog[I915_FOG_SETUP_SIZE];
+   GLuint Defaults[I915_DEF_SETUP_SIZE];
    GLuint Tex[I915_TEX_UNITS][I915_TEX_SETUP_SIZE];
    GLuint Constant[I915_CONSTANT_SIZE];
    GLuint ConstantSize;
    GLuint Program[I915_PROGRAM_SIZE];
    GLuint ProgramSize;
-   GLuint active;		/* I915_UPLOAD_* */
-   GLuint emitted;		/* I915_UPLOAD_* */
+
+   /* Region pointers for relocation: 
+    */
+   struct intel_region *draw_region;
+   struct intel_region *depth_region;
+/*    struct intel_region *tex_region[I915_TEX_UNITS]; */
+
+   /* Regions aren't actually that appropriate here as the memory may
+    * be from a PBO or FBO.  Will have to do this for draw and depth for
+    * FBO's...
+    */
+   dri_bo *tex_buffer[I915_TEX_UNITS];
+   GLuint tex_offset[I915_TEX_UNITS];
+
+
+   GLuint active;               /* I915_UPLOAD_* */
+   GLuint emitted;              /* I915_UPLOAD_* */
 };
 
 #define I915_FOG_PIXEL  2
 #define I915_FOG_VERTEX 1
 #define I915_FOG_NONE   0
 
-struct i915_context 
+struct i915_context
 {
    struct intel_context intel;
 
    GLuint last_ReallyEnabled;
    GLuint vertex_fog;
+   GLuint lodbias_ss2[MAX_TEXTURE_UNITS];
+
 
-   struct i915_fragment_program tex_program;
    struct i915_fragment_program *current_program;
 
    struct i915_hw_state meta, initial, state, *current;
 };
 
 
-typedef struct i915_context *i915ContextPtr;
-typedef struct i915_texture_object *i915TextureObjectPtr;
-
-#define I915_CONTEXT(ctx)	((i915ContextPtr)(ctx))
-
-
-
 #define I915_STATECHANGE(i915, flag)					\
 do {									\
-   if (0) fprintf(stderr, "I915_STATECHANGE %x in %s\n", flag, __FUNCTION__);	\
    INTEL_FIREVERTICES( &(i915)->intel );					\
    (i915)->state.emitted &= ~(flag);					\
 } while (0)
 
 #define I915_ACTIVESTATE(i915, flag, mode)			\
 do {								\
-   if (0) fprintf(stderr, "I915_ACTIVESTATE %x %d in %s\n",	\
-		  flag, mode, __FUNCTION__);			\
    INTEL_FIREVERTICES( &(i915)->intel );				\
    if (mode)							\
       (i915)->state.active |= (flag);				\
@@ -260,7 +253,13 @@ do {								\
 /*======================================================================
  * i915_vtbl.c
  */
-extern void i915InitVtbl( i915ContextPtr i915 );
+extern void i915InitVtbl(struct i915_context *i915);
+
+extern void
+i915_state_draw_region(struct intel_context *intel,
+                       struct i915_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region);
 
 
 
@@ -289,70 +288,58 @@ do {									\
 /*======================================================================
  * i915_context.c
  */
-extern GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-				    __DRIcontextPrivate *driContextPriv,
-				    void *sharedContextPrivate);
-
-
-/*======================================================================
- * i915_texprog.c
- */
-extern void i915ValidateTextureProgram( i915ContextPtr i915 );
+extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
 
 
 /*======================================================================
  * i915_debug.c
  */
-extern void i915_disassemble_program( const GLuint *program, GLuint sz );
-extern void i915_print_ureg( const char *msg, GLuint ureg );
+extern void i915_disassemble_program(const GLuint * program, GLuint sz);
+extern void i915_print_ureg(const char *msg, GLuint ureg);
 
 
 /*======================================================================
  * i915_state.c
  */
-extern void i915InitStateFunctions( struct dd_function_table *functions );
-extern void i915InitState( i915ContextPtr i915 );
-extern void i915_update_fog(GLcontext *ctxx);
+extern void i915InitStateFunctions(struct dd_function_table *functions);
+extern void i915InitState(struct i915_context *i915);
+extern void i915_update_fog(GLcontext * ctx);
 
 
 /*======================================================================
  * i915_tex.c
  */
-extern void i915UpdateTextureState( intelContextPtr intel );
-extern void i915InitTextureFuncs( struct dd_function_table *functions );
-extern intelTextureObjectPtr i915AllocTexObj( struct gl_texture_object *texObj );
+extern void i915UpdateTextureState(struct intel_context *intel);
+extern void i915InitTextureFuncs(struct dd_function_table *functions);
 
 /*======================================================================
  * i915_metaops.c
  */
-extern GLboolean
-i915TryTextureReadPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *pack,
-			  GLvoid *pixels );
-
-extern GLboolean
-i915TryTextureDrawPixels( GLcontext *ctx,
-			  GLint x, GLint y, GLsizei width, GLsizei height,
-			  GLenum format, GLenum type,
-			  const struct gl_pixelstore_attrib *unpack,
-			  const GLvoid *pixels );
+void i915InitMetaFuncs(struct i915_context *i915);
 
-extern void 
-i915ClearWithTris( intelContextPtr intel, GLbitfield mask,
-		   GLboolean all, GLint cx, GLint cy, GLint cw, GLint ch);
-
-
-extern void
-i915RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf);
 
 /*======================================================================
  * i915_fragprog.c
  */
-extern void i915ValidateFragmentProgram( i915ContextPtr i915 );
-extern void i915InitFragProgFuncs( struct dd_function_table *functions );
-	
-#endif
+extern void i915ValidateFragmentProgram(struct i915_context *i915);
+extern void i915InitFragProgFuncs(struct dd_function_table *functions);
+
+/*======================================================================
+ * Inline conversion functions.  These are better-typed than the
+ * macros used previously:
+ */
+static INLINE struct i915_context *
+i915_context(GLcontext * ctx)
+{
+   return (struct i915_context *) ctx;
+}
+
 
+
+#define I915_CONTEXT(ctx)	i915_context(ctx)
+
+
+
+#endif
diff --git a/src/mesa/drivers/dri/i915/i915_debug.c b/src/mesa/drivers/dri/i915/i915_debug.c
index 054b561605a..c0e1242a0ea 100644
--- a/src/mesa/drivers/dri/i915/i915_debug.c
+++ b/src/mesa/drivers/dri/i915/i915_debug.c
@@ -25,275 +25,819 @@
  * 
  **************************************************************************/
 
+#include "imports.h"
+
 #include "i915_reg.h"
 #include "i915_context.h"
-#include <stdio.h>
-
-
-static const char *opcodes[0x20] = {
-   "NOP",
-   "ADD",
-   "MOV",
-   "MUL",
-   "MAD",
-   "DP2ADD",
-   "DP3",
-   "DP4",
-   "FRC",
-   "RCP",
-   "RSQ",
-   "EXP",
-   "LOG",
-   "CMP",
-   "MIN",
-   "MAX",
-   "FLR",
-   "MOD",
-   "TRC",
-   "SGE",
-   "SLT",
-   "TEXLD",
-   "TEXLDP",
-   "TEXLDB",
-   "TEXKILL",
-   "DCL",
-   "0x1a",
-   "0x1b",
-   "0x1c",
-   "0x1d",
-   "0x1e",
-   "0x1f",
-};
-
-
-static const int args[0x20] = {
-   0,				/* 0 nop */
-   2,				/* 1 add */
-   1,				/* 2 mov */
-   2,				/* 3 m ul */
-   3, 				/* 4 mad */
-   3,				/* 5 dp2add */
-   2,				/* 6 dp3 */
-   2,				/* 7 dp4 */
-   1,				/* 8 frc */
-   1,				/* 9 rcp */
-   1,				/* a rsq */
-   1,				/* b exp */
-   1,				/* c log */
-   3,				/* d cmp */
-   2,				/* e min */
-   2,				/* f max */
-   1,				/* 10 flr */
-   1,				/* 11 mod */
-   1,				/* 12 trc */
-   2,				/* 13 sge */
-   2,				/* 14 slt */
-   1,
-   1,
-   1,
-   1,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-   0,
-};
-
-
-static const char *regname[0x8] = {
-   "R",
-   "T",
-   "CONST",
-   "S",
-   "OC",
-   "OD",
-   "U",
-   "UNKNOWN",
-};
-
-static void print_reg_type_nr( GLuint type, GLuint nr )
+#include "i915_debug.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
+
+static GLboolean debug( struct debug_stream *stream, const char *name, GLuint len )
 {
-   switch (type) {
-   case REG_TYPE_T:
-      switch (nr) {
-      case T_DIFFUSE: fprintf(stderr, "T_DIFFUSE"); return;
-      case T_SPECULAR: fprintf(stderr, "T_SPECULAR"); return;
-      case T_FOG_W: fprintf(stderr, "T_FOG_W"); return;
-      default: fprintf(stderr, "T_TEX%d", nr); return;
-      }
-   case REG_TYPE_OC:
-      if (nr == 0) {
-	 fprintf(stderr, "oC");
-	 return;
-      }
-      break;
-   case REG_TYPE_OD:
-      if (nr == 0) {
-	 fprintf(stderr, "oD");
-	 return;
-      }
-      break;
-   default:
-      break;
+   GLuint i;
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
    }
 
-   fprintf(stderr, "%s[%d]", regname[type], nr);
-}
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
 
-#define REG_SWIZZLE_MASK 0x7777
-#define REG_NEGATE_MASK 0x8888
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);   
+   PRINTF("\n");
 
-#define REG_SWIZZLE_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
-		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
 
 
-static void print_reg_neg_swizzle( GLuint reg )
+static const char *get_prim_name( GLuint val )
 {
-   int i;
-
-   if ((reg & REG_SWIZZLE_MASK) == REG_SWIZZLE_XYZW &&
-       (reg & REG_NEGATE_MASK) == 0)
-      return;
-
-   fprintf(stderr, ".");
-
-   for (i = 3 ; i >= 0; i--) {
-      if (reg & (1<<((i*4)+3))) 
-	 fprintf(stderr, "-");
-	 
-      switch ((reg>>(i*4)) & 0x7) {
-      case 0: fprintf(stderr, "x"); break;
-      case 1: fprintf(stderr, "y"); break;
-      case 2: fprintf(stderr, "z"); break;
-      case 3: fprintf(stderr, "w"); break;
-      case 4: fprintf(stderr, "0"); break;
-      case 5: fprintf(stderr, "1"); break;
-      default: fprintf(stderr, "?"); break;
-      }
+   switch (val & PRIM3D_MASK) {
+   case PRIM3D_TRILIST: return "TRILIST"; break;
+   case PRIM3D_TRISTRIP: return "TRISTRIP"; break;
+   case PRIM3D_TRISTRIP_RVRSE: return "TRISTRIP_RVRSE"; break;
+   case PRIM3D_TRIFAN: return "TRIFAN"; break;
+   case PRIM3D_POLY: return "POLY"; break;
+   case PRIM3D_LINELIST: return "LINELIST"; break;
+   case PRIM3D_LINESTRIP: return "LINESTRIP"; break;
+   case PRIM3D_RECTLIST: return "RECTLIST"; break;
+   case PRIM3D_POINTLIST: return "POINTLIST"; break;
+   case PRIM3D_DIB: return "DIB"; break;
+   case PRIM3D_CLEAR_RECT: return "CLEAR_RECT"; break;
+   case PRIM3D_ZONE_INIT: return "ZONE_INIT"; break;
+   default: return "????"; break;
    }
 }
 
-
-static void print_src_reg( GLuint dword )
+static GLboolean debug_prim( struct debug_stream *stream, const char *name, 
+			     GLboolean dump_floats,
+			     GLuint len )
 {
-   GLuint nr = (dword >> A2_SRC2_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A2_SRC2_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr( type, nr );
-   print_reg_neg_swizzle( dword );
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i;
+   
+
+
+   PRINTF("%s %s (%d dwords):\n", name, prim, len);
+   PRINTF("\t0x%08x\n",  ptr[0]);   
+   for (i = 1; i < len; i++) {
+      if (dump_floats)
+	 PRINTF("\t0x%08x // %f\n",  ptr[i], *(GLfloat *)&ptr[i]);   
+      else
+	 PRINTF("\t0x%08x\n",  ptr[i]);   
+   }
+
+      
+   PRINTF("\n");
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
 }
+   
+
+
 
-void i915_print_ureg( const char *msg, GLuint ureg )
+static GLboolean debug_program( struct debug_stream *stream, const char *name, GLuint len )
 {
-   fprintf(stderr, "%s: ", msg);
-   print_src_reg( ureg >> 8 );
-   fprintf(stderr, "\n");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+
+   if (len == 0) {
+      PRINTF("Error - zero length packet (0x%08x)\n", stream->ptr[0]);
+      assert(0);
+      return GL_FALSE;
+   }
+
+   if (stream->print_addresses)
+      PRINTF("%08x:  ", stream->offset);
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   i915_disassemble_program( ptr, len );
+
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
 }
 
-static void print_dest_reg( GLuint dword )
+
+static GLboolean debug_chain( struct debug_stream *stream, const char *name, GLuint len )
 {
-   GLuint nr = (dword >> A0_DEST_NR_SHIFT) & REG_NR_MASK;
-   GLuint type = (dword >> A0_DEST_TYPE_SHIFT) & REG_TYPE_MASK;
-   print_reg_type_nr( type, nr );
-   if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
-      return;
-   fprintf(stderr, ".");
-   if (dword & A0_DEST_CHANNEL_X) fprintf(stderr, "x");
-   if (dword & A0_DEST_CHANNEL_Y) fprintf(stderr, "y");
-   if (dword & A0_DEST_CHANNEL_Z) fprintf(stderr, "z");
-   if (dword & A0_DEST_CHANNEL_W) fprintf(stderr, "w");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint old_offset = stream->offset + len * sizeof(GLuint);
+   GLuint i;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+
+   stream->offset = ptr[1] & ~0x3;
+   
+   if (stream->offset < old_offset)
+      PRINTF("\n... skipping backwards from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+   else
+      PRINTF("\n... skipping from 0x%x --> 0x%x ...\n\n", 
+		   old_offset, stream->offset );
+
+
+   return GL_TRUE;
 }
 
 
-#define GET_SRC0_REG(r0, r1) ((r0<<14)|(r1>>A1_SRC0_CHANNEL_W_SHIFT))
-#define GET_SRC1_REG(r0, r1) ((r0<<8)|(r1>>A2_SRC1_CHANNEL_W_SHIFT))
-#define GET_SRC2_REG(r)      (r)
+static GLboolean debug_variable_length_prim( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   const char *prim = get_prim_name( ptr[0] );
+   GLuint i, len;
+
+   GLushort *idx = (GLushort *)(ptr+1);
+   for (i = 0; idx[i] != 0xffff; i++)
+      ;
+
+   len = 1+(i+2)/2;
 
+   PRINTF("3DPRIM, %s variable length %d indicies (%d dwords):\n", prim, i, len);
+   for (i = 0; i < len; i++)
+      PRINTF("\t0x%08x\n",  ptr[i]);
+   PRINTF("\n");
 
-static void print_arith_op( GLuint opcode, const GLuint *program )
+   stream->offset += len * sizeof(GLuint);
+   return GL_TRUE;
+}
+
+
+#define BITS( dw, hi, lo, ... )				\
+do {							\
+   unsigned himask = ~0UL >> (31 - (hi));		\
+   PRINTF("\t\t ");				\
+   PRINTF(__VA_ARGS__);			\
+   PRINTF(": 0x%x\n", ((dw) & himask) >> (lo));	\
+} while (0)
+
+#define MBZ( dw, hi, lo) do {							\
+   unsigned x = (dw) >> (lo);				\
+   unsigned lomask = (1 << (lo)) - 1;			\
+   unsigned himask;					\
+   himask = (1UL << (hi)) - 1;				\
+   assert ((x & himask & ~lomask) == 0);	\
+} while (0)
+
+#define FLAG( dw, bit, ... )			\
+do {							\
+   if (((dw) >> (bit)) & 1) {				\
+      PRINTF("\t\t ");				\
+      PRINTF(__VA_ARGS__);			\
+      PRINTF("\n");				\
+   }							\
+} while (0)
+
+static GLboolean debug_load_immediate( struct debug_stream *stream,
+				       const char *name,
+				       GLuint len )
 {
-   if (opcode != A0_NOP) {
-      print_dest_reg(program[0]);
-      if (program[0] & A0_DEST_SATURATE)
-	 fprintf(stderr, " = SATURATE ");
-      else
-	 fprintf(stderr, " = ");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 4) & 0xff;
+   GLuint j = 0;
+   
+   PRINTF("%s (%d dwords, flags: %x):\n", name, len, bits);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   if (bits & (1<<0)) {
+      PRINTF("\t  LIS0: 0x%08x\n", ptr[j]);
+      PRINTF("\t vb address: 0x%08x\n", (ptr[j] & ~0x3));
+      BITS(ptr[j], 0, 0, "vb invalidate disable");
+      j++;
    }
+   if (bits & (1<<1)) {
+      PRINTF("\t  LIS1: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 29, 24, "vb dword width");
+      BITS(ptr[j], 21, 16, "vb dword pitch");
+      BITS(ptr[j], 15, 0, "vb max index");
+      j++;
+   }
+   if (bits & (1<<2)) {
+      int i;
+      PRINTF("\t  LIS2: 0x%08x\n", ptr[j]);
+      for (i = 0; i < 8; i++) {
+	 unsigned tc = (ptr[j] >> (i * 4)) & 0xf;
+	 if (tc != 0xf)
+	    BITS(tc, 3, 0, "tex coord %d", i);
+      }
+      j++;
+   }
+   if (bits & (1<<3)) {
+      PRINTF("\t  LIS3: 0x%08x\n", ptr[j]);
+      j++;
+   }
+   if (bits & (1<<4)) {
+      PRINTF("\t  LIS4: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 23, "point width");
+      BITS(ptr[j], 22, 19, "line width");
+      FLAG(ptr[j], 18, "alpha flatshade");
+      FLAG(ptr[j], 17, "fog flatshade");
+      FLAG(ptr[j], 16, "spec flatshade");
+      FLAG(ptr[j], 15, "rgb flatshade");
+      BITS(ptr[j], 14, 13, "cull mode");
+      FLAG(ptr[j], 12, "vfmt: point width");
+      FLAG(ptr[j], 11, "vfmt: specular/fog");
+      FLAG(ptr[j], 10, "vfmt: rgba");
+      FLAG(ptr[j], 9, "vfmt: depth offset");
+      BITS(ptr[j], 8, 6, "vfmt: position (2==xyzw)");
+      FLAG(ptr[j], 5, "force dflt diffuse");
+      FLAG(ptr[j], 4, "force dflt specular");
+      FLAG(ptr[j], 3, "local depth offset enable");
+      FLAG(ptr[j], 2, "vfmt: fp32 fog coord");
+      FLAG(ptr[j], 1, "sprite point");
+      FLAG(ptr[j], 0, "antialiasing");
+      j++;
+   }
+   if (bits & (1<<5)) {
+      PRINTF("\t  LIS5: 0x%08x\n", ptr[j]);
+      BITS(ptr[j], 31, 28, "rgba write disables");
+      FLAG(ptr[j], 27,     "force dflt point width");
+      FLAG(ptr[j], 26,     "last pixel enable");
+      FLAG(ptr[j], 25,     "global z offset enable");
+      FLAG(ptr[j], 24,     "fog enable");
+      BITS(ptr[j], 23, 16, "stencil ref");
+      BITS(ptr[j], 15, 13, "stencil test");
+      BITS(ptr[j], 12, 10, "stencil fail op");
+      BITS(ptr[j], 9, 7,   "stencil pass z fail op");
+      BITS(ptr[j], 6, 4,   "stencil pass z pass op");
+      FLAG(ptr[j], 3,      "stencil write enable");
+      FLAG(ptr[j], 2,      "stencil test enable");
+      FLAG(ptr[j], 1,      "color dither enable");
+      FLAG(ptr[j], 0,      "logiop enable");
+      j++;
+   }
+   if (bits & (1<<6)) {
+      PRINTF("\t  LIS6: 0x%08x\n", ptr[j]);
+      FLAG(ptr[j], 31,      "alpha test enable");
+      BITS(ptr[j], 30, 28,  "alpha func");
+      BITS(ptr[j], 27, 20,  "alpha ref");
+      FLAG(ptr[j], 19,      "depth test enable");
+      BITS(ptr[j], 18, 16,  "depth func");
+      FLAG(ptr[j], 15,      "blend enable");
+      BITS(ptr[j], 14, 12,  "blend func");
+      BITS(ptr[j], 11, 8,   "blend src factor");
+      BITS(ptr[j], 7,  4,   "blend dst factor");
+      FLAG(ptr[j], 3,       "depth write enable");
+      FLAG(ptr[j], 2,       "color write enable");
+      BITS(ptr[j], 1,  0,   "provoking vertex"); 
+      j++;
+   }
+
+
+   PRINTF("\n");
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   assert(j == len);
 
-   print_src_reg(GET_SRC0_REG(program[0], program[1]));
-   if (args[opcode] == 1) {
-      fprintf(stderr, "\n");
-      return;
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 
+
+
+static GLboolean debug_load_indirect( struct debug_stream *stream,
+				      const char *name,
+				      GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint bits = (ptr[0] >> 8) & 0x3f;
+   GLuint i, j = 0;
+   
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   for (i = 0; i < 6; i++) {
+      if (bits & (1<<i)) {
+	 switch (1<<(8+i)) {
+	 case LI0_STATE_STATIC_INDIRECT:
+	    PRINTF("        STATIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_DYNAMIC_INDIRECT:
+	    PRINTF("       DYNAMIC: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    break;
+	 case LI0_STATE_SAMPLER:
+	    PRINTF("       SAMPLER: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_MAP:
+	    PRINTF("           MAP: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_PROGRAM:
+	    PRINTF("       PROGRAM: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 case LI0_STATE_CONSTANTS:
+	    PRINTF("     CONSTANTS: 0x%08x | %x\n", ptr[j]&~3, ptr[j]&3); j++;
+	    PRINTF("                0x%08x\n", ptr[j++]);
+	    break;
+	 default:
+	    assert(0);
+	    break;
+	 }
+      }
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC1_REG(program[1], program[2]));
-   if (args[opcode] == 2) { 
-      fprintf(stderr, "\n");
-      return;
+   if (bits == 0) {
+      PRINTF("\t  DUMMY: 0x%08x\n", ptr[j++]);
    }
 
-   fprintf(stderr, ", ");
-   print_src_reg(GET_SRC2_REG(program[2]));
-   fprintf(stderr, "\n");
-   return;
+   PRINTF("\n");
+
+
+   assert(j == len);
+
+   stream->offset += len * sizeof(GLuint);
+   
+   return GL_TRUE;
+}
+ 	
+static void BR13( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   FLAG(val, 30, "clipping enable");
+   BITS(val, 25, 24, "color depth (3==32bpp)");
+   BITS(val, 23, 16, "raster op");
+   BITS(val, 15, 0,  "dest pitch");
+}
+
+
+static void BR22( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "dest y1");
+   BITS(val, 15, 0,  "dest x1");
+}
+
+static void BR23( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "dest y2");
+   BITS(val, 15, 0,  "dest x2");
+}
+
+static void BR09( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- dest address\n",  val);
 }
 
+static void BR26( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 31, 16, "src y1");
+   BITS(val, 15, 0,  "src x1");
+}
 
-static void print_tex_op( GLuint opcode, const GLuint *program )
+static void BR11( struct debug_stream *stream,
+		  GLuint val )
 {
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, " = ");
+   PRINTF("\t0x%08x\n",  val);
+   BITS(val, 15, 0,  "src pitch");
+}
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+static void BR12( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- src address\n",  val);
+}
 
-   fprintf(stderr, "S[%d],", 
-	   program[0] & T0_SAMPLER_NR_MASK);
+static void BR16( struct debug_stream *stream,
+		  GLuint val )
+{
+   PRINTF("\t0x%08x -- color\n",  val);
+}
+   
+static GLboolean debug_copy_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   BR13(stream, ptr[j++]);
+   BR22(stream, ptr[j++]);
+   BR23(stream, ptr[j++]);
+   BR09(stream, ptr[j++]);
+   BR26(stream, ptr[j++]);
+   BR11(stream, ptr[j++]);
+   BR12(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_color_blit( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   BR13(stream, ptr[j++]);
+   BR22(stream, ptr[j++]);
+   BR23(stream, ptr[j++]);
+   BR09(stream, ptr[j++]);
+   BR16(stream, ptr[j++]);
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_modes4( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j]);
+   BITS(ptr[j], 21, 18, "logicop func");
+   FLAG(ptr[j], 17, "stencil test mask modify-enable");
+   FLAG(ptr[j], 16, "stencil write mask modify-enable");
+   BITS(ptr[j], 15, 8, "stencil test mask");
+   BITS(ptr[j], 7, 0,  "stencil write mask");
+   j++;
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean debug_map_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "map mask");
+      j++;
+   }
+
+   while (j < len) {
+      {
+	 PRINTF("\t  TMn.0: 0x%08x\n", ptr[j]);
+	 PRINTF("\t map address: 0x%08x\n", (ptr[j] & ~0x3));
+	 FLAG(ptr[j], 1, "vertical line stride");
+	 FLAG(ptr[j], 0, "vertical line stride offset");
+	 j++;
+      }
+
+      {
+	 PRINTF("\t  TMn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "height");
+	 BITS(ptr[j], 20, 10, "width");
+	 BITS(ptr[j], 9, 7, "surface format");
+	 BITS(ptr[j], 6, 3, "texel format");
+	 FLAG(ptr[j], 2, "use fence regs");
+	 FLAG(ptr[j], 1, "tiled surface");
+	 FLAG(ptr[j], 0, "tile walk ymajor");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TMn.2: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 21, "dword pitch");
+	 BITS(ptr[j], 20, 15, "cube face enables");
+	 BITS(ptr[j], 14, 9, "max lod");
+	 FLAG(ptr[j], 8,     "mip layout right");
+	 BITS(ptr[j], 7, 0, "depth");
+	 j++;
+      }
+   }
 
-   print_reg_type_nr( (program[1]>>T1_ADDRESS_REG_TYPE_SHIFT) & REG_TYPE_MASK,
-		      (program[1]>>T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK );
-   fprintf(stderr, "\n");
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
 }
 
-static void print_dcl_op( GLuint opcode, const GLuint *program )
+static GLboolean debug_sampler_state( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
 {
-   fprintf(stderr, "%s ", opcodes[opcode]);
-   print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, "\n");
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+   
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 15, 0,   "sampler mask");
+      j++;
+   }
+
+   while (j < len) {
+      {
+	 PRINTF("\t  TSn.0: 0x%08x\n", ptr[j]);
+	 FLAG(ptr[j], 31, "reverse gamma");
+	 FLAG(ptr[j], 30, "planar to packed");
+	 FLAG(ptr[j], 29, "yuv->rgb");
+	 BITS(ptr[j], 28, 27, "chromakey index");
+	 BITS(ptr[j], 26, 22, "base mip level");
+	 BITS(ptr[j], 21, 20, "mip mode filter");
+	 BITS(ptr[j], 19, 17, "mag mode filter");
+	 BITS(ptr[j], 16, 14, "min mode filter");
+	 BITS(ptr[j], 13, 5,  "lod bias (s4.4)");
+	 FLAG(ptr[j], 4,      "shadow enable");
+	 FLAG(ptr[j], 3,      "max-aniso-4");
+	 BITS(ptr[j], 2, 0,   "shadow func");
+	 j++;
+      }
+
+      {
+	 PRINTF("\t  TSn.1: 0x%08x\n", ptr[j]);
+	 BITS(ptr[j], 31, 24, "min lod");
+	 MBZ( ptr[j], 23, 18 );
+	 FLAG(ptr[j], 17,     "kill pixel enable");
+	 FLAG(ptr[j], 16,     "keyed tex filter mode");
+	 FLAG(ptr[j], 15,     "chromakey enable");
+	 BITS(ptr[j], 14, 12, "tcx wrap mode");
+	 BITS(ptr[j], 11, 9,  "tcy wrap mode");
+	 BITS(ptr[j], 8,  6,  "tcz wrap mode");
+	 FLAG(ptr[j], 5,      "normalized coords");
+	 BITS(ptr[j], 4,  1,  "map (surface) index");
+	 FLAG(ptr[j], 0,      "EAST deinterlacer enable");
+	 j++;
+      }
+      {
+	 PRINTF("\t  TSn.2: 0x%08x  (default color)\n", ptr[j]);
+	 j++;
+      }
+   }
+
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
 }
 
+static GLboolean debug_dest_vars( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      FLAG(ptr[j], 31,     "early classic ztest");
+      FLAG(ptr[j], 30,     "opengl tex default color");
+      FLAG(ptr[j], 29,     "bypass iz");
+      FLAG(ptr[j], 28,     "lod preclamp");
+      BITS(ptr[j], 27, 26, "dither pattern");
+      FLAG(ptr[j], 25,     "linear gamma blend");
+      FLAG(ptr[j], 24,     "debug dither");
+      BITS(ptr[j], 23, 20, "dstorg x");
+      BITS(ptr[j], 19, 16, "dstorg y");
+      MBZ (ptr[j], 15, 15 );
+      BITS(ptr[j], 14, 12, "422 write select");
+      BITS(ptr[j], 11, 8,  "cbuf format");
+      BITS(ptr[j], 3, 2,   "zbuf format");
+      FLAG(ptr[j], 1,      "vert line stride");
+      FLAG(ptr[j], 1,      "vert line stride offset");
+      j++;
+   }
+   
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
 
-void i915_disassemble_program( const GLuint *program, GLuint sz )
+static GLboolean debug_buf_info( struct debug_stream *stream,
+				  const char *name,
+				  GLuint len )
 {
-   GLuint size = program[0] & 0x1ff;
-   GLint i;
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   int j = 0;
+
+   PRINTF("%s (%d dwords):\n", name, len);
+   PRINTF("\t0x%08x\n",  ptr[j++]);
+
+   {
+      PRINTF("\t0x%08x\n",  ptr[j]);
+      BITS(ptr[j], 28, 28, "aux buffer id");
+      BITS(ptr[j], 27, 24, "buffer id (7=depth, 3=back)");
+      FLAG(ptr[j], 23,     "use fence regs");
+      FLAG(ptr[j], 22,     "tiled surface");
+      FLAG(ptr[j], 21,     "tile walk ymajor");
+      MBZ (ptr[j], 20, 14);
+      BITS(ptr[j], 13, 2,  "dword pitch");
+      MBZ (ptr[j], 2,  0);
+      j++;
+   }
    
-   fprintf(stderr, "BEGIN\n");
+   PRINTF("\t0x%08x -- buffer base address\n",  ptr[j++]);
 
-   if (size+2 != sz) {
-      fprintf(stderr, "%s: program size mismatch %d/%d\n", __FUNCTION__,
-	      size+2, sz);
-      exit(1);
+   stream->offset += len * sizeof(GLuint);
+   assert(j == len);
+   return GL_TRUE;
+}
+
+static GLboolean i915_debug_packet( struct debug_stream *stream )
+{
+   GLuint *ptr = (GLuint *)(stream->ptr + stream->offset);
+   GLuint cmd = *ptr;
+   
+   switch (((cmd >> 29) & 0x7)) {
+   case 0x0:
+      switch ((cmd >> 23) & 0x3f) {
+      case 0x0:
+	 return debug(stream, "MI_NOOP", 1);
+      case 0x3:
+	 return debug(stream, "MI_WAIT_FOR_EVENT", 1);
+      case 0x4:
+	 return debug(stream, "MI_FLUSH", 1);
+      case 0xA:
+	 debug(stream, "MI_BATCH_BUFFER_END", 1);
+	 return GL_FALSE;
+      case 0x22:
+	 return debug(stream, "MI_LOAD_REGISTER_IMM", 3);
+      case 0x31:
+	 return debug_chain(stream, "MI_BATCH_BUFFER_START", 2);
+      default:
+	 break;
+      }
+      break;
+   case 0x1:
+      break;
+   case 0x2:
+      switch ((cmd >> 22) & 0xff) {	 
+      case 0x50:
+	 return debug_color_blit(stream, "XY_COLOR_BLT", (cmd & 0xff) + 2);
+      case 0x53:
+	 return debug_copy_blit(stream, "XY_SRC_COPY_BLT", (cmd & 0xff) + 2);
+      default:
+	 return debug(stream, "blit command", (cmd & 0xff) + 2);
+      }
+      break;
+   case 0x3:
+      switch ((cmd >> 24) & 0x1f) {	 
+      case 0x6:
+	 return debug(stream, "3DSTATE_ANTI_ALIASING", 1);
+      case 0x7:
+	 return debug(stream, "3DSTATE_RASTERIZATION_RULES", 1);
+      case 0x8:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_OPS", 2);
+      case 0x9:
+	 return debug(stream, "3DSTATE_BACKFACE_STENCIL_MASKS", 1);
+      case 0xb:
+	 return debug(stream, "3DSTATE_INDEPENDENT_ALPHA_BLEND", 1);
+      case 0xc:
+	 return debug(stream, "3DSTATE_MODES5", 1);	 
+      case 0xd:
+	 return debug_modes4(stream, "3DSTATE_MODES4", 1);
+      case 0x15:
+	 return debug(stream, "3DSTATE_FOG_COLOR", 1);
+      case 0x16:
+	 return debug(stream, "3DSTATE_COORD_SET_BINDINGS", 1);
+      case 0x1c:
+	 /* 3DState16NP */
+	 switch((cmd >> 19) & 0x1f) {
+	 case 0x10:
+	    return debug(stream, "3DSTATE_SCISSOR_ENABLE", 1);
+	 case 0x11:
+	    return debug(stream, "3DSTATE_DEPTH_SUBRECTANGLE_DISABLE", 1);
+	 default:
+	    break;
+	 }
+	 break;
+      case 0x1d:
+	 /* 3DStateMW */
+	 switch ((cmd >> 16) & 0xff) {
+	 case 0x0:
+	    return debug_map_state(stream, "3DSTATE_MAP_STATE", (cmd & 0x1f) + 2);
+	 case 0x1:
+	    return debug_sampler_state(stream, "3DSTATE_SAMPLER_STATE", (cmd & 0x1f) + 2);
+	 case 0x4:
+	    return debug_load_immediate(stream, "3DSTATE_LOAD_STATE_IMMEDIATE", (cmd & 0xf) + 2);
+	 case 0x5:
+	    return debug_program(stream, "3DSTATE_PIXEL_SHADER_PROGRAM", (cmd & 0x1ff) + 2);
+	 case 0x6:
+	    return debug(stream, "3DSTATE_PIXEL_SHADER_CONSTANTS", (cmd & 0xff) + 2);
+	 case 0x7:
+	    return debug_load_indirect(stream, "3DSTATE_LOAD_INDIRECT", (cmd & 0xff) + 2);
+	 case 0x80:
+	    return debug(stream, "3DSTATE_DRAWING_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x81:
+	    return debug(stream, "3DSTATE_SCISSOR_RECTANGLE", (cmd & 0xffff) + 2);
+	 case 0x83:
+	    return debug(stream, "3DSTATE_SPAN_STIPPLE", (cmd & 0xffff) + 2);
+	 case 0x85:
+	    return debug_dest_vars(stream, "3DSTATE_DEST_BUFFER_VARS", (cmd & 0xffff) + 2);
+	 case 0x88:
+	    return debug(stream, "3DSTATE_CONSTANT_BLEND_COLOR", (cmd & 0xffff) + 2);
+	 case 0x89:
+	    return debug(stream, "3DSTATE_FOG_MODE", (cmd & 0xffff) + 2);
+	 case 0x8e:
+	    return debug_buf_info(stream, "3DSTATE_BUFFER_INFO", (cmd & 0xffff) + 2);
+	 case 0x97:
+	    return debug(stream, "3DSTATE_DEPTH_OFFSET_SCALE", (cmd & 0xffff) + 2);
+	 case 0x98:
+	    return debug(stream, "3DSTATE_DEFAULT_Z", (cmd & 0xffff) + 2);
+	 case 0x99:
+	    return debug(stream, "3DSTATE_DEFAULT_DIFFUSE", (cmd & 0xffff) + 2);
+	 case 0x9a:
+	    return debug(stream, "3DSTATE_DEFAULT_SPECULAR", (cmd & 0xffff) + 2);
+	 case 0x9c:
+	    return debug(stream, "3DSTATE_CLEAR_PARAMETERS", (cmd & 0xffff) + 2);
+	 default:
+	    assert(0);
+	    return 0;
+	 }
+	 break;
+      case 0x1e:
+	 if (cmd & (1 << 23))
+	    return debug(stream, "???", (cmd & 0xffff) + 1);
+	 else
+	    return debug(stream, "", 1);
+	 break;
+      case 0x1f:
+	 if ((cmd & (1 << 23)) == 0)	
+	    return debug_prim(stream, "3DPRIM (inline)", 1, (cmd & 0x1ffff) + 2);
+	 else if (cmd & (1 << 17)) 
+	 {
+	    if ((cmd & 0xffff) == 0)
+	       return debug_variable_length_prim(stream);
+	    else
+	       return debug_prim(stream, "3DPRIM (indexed)", 0, (((cmd & 0xffff) + 1) / 2) + 1);
+	 }
+	 else
+	    return debug_prim(stream, "3DPRIM  (indirect sequential)", 0, 2); 
+	 break;
+      default:
+	 return debug(stream, "", 0);
+      }
+   default:
+      assert(0);
+      return 0;
    }
 
-   program ++;
-   for (i = 1 ; i < sz ; i+=3, program+=3) {
-      GLuint opcode = program[0] & (0x1f<<24);
-
-      if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
-	 print_arith_op(opcode >> 24, program);
-      else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
-	 print_tex_op(opcode >> 24, program);
-      else if (opcode == D0_DCL)
-	 print_dcl_op(opcode >> 24, program);
-      else 
-	 fprintf(stderr, "Unknown opcode 0x%x\n", opcode);
+   assert(0);
+   return 0;
+}
+
+
+
+void
+i915_dump_batchbuffer( GLuint *start,
+		       GLuint *end )
+{
+   struct debug_stream stream;
+   GLuint bytes = (end - start) * 4;
+   GLboolean done = GL_FALSE;
+
+   PRINTF("\n\nBATCH: (%d)\n", bytes / 4);
+
+   stream.offset = 0;
+   stream.ptr = (char *)start;
+   stream.print_addresses = 0;
+
+   while (!done &&
+	  stream.offset < bytes &&
+	  stream.offset >= 0)
+   {
+      if (!i915_debug_packet( &stream ))
+	 break;
+
+      assert(stream.offset <= bytes &&
+	     stream.offset >= 0);
    }
 
-   fprintf(stderr, "END\n\n");
+   PRINTF("END-BATCH\n\n\n");
 }
+
+
diff --git a/src/mesa/drivers/dri/i915tex/intel_span.h b/src/mesa/drivers/dri/i915/i915_debug.h
index 5201f6d6c6e..0643a8c631c 100644
--- a/src/mesa/drivers/dri/i915tex/intel_span.h
+++ b/src/mesa/drivers/dri/i915/i915_debug.h
@@ -1,8 +1,8 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -25,14 +25,31 @@
  * 
  **************************************************************************/
 
-#ifndef _INTEL_SPAN_H
-#define _INTEL_SPAN_H
+/* Authors:  Keith Whitwell <[email protected]>
+ */
+
+#ifndef I915_DEBUG_H
+#define I915_DEBUG_H
+
+struct i915_context;
+
+struct debug_stream 
+{
+   unsigned offset;		/* current gtt offset */
+   char *ptr;		/* pointer to gtt offset zero */
+   char *end;		/* pointer to gtt offset zero */
+   unsigned print_addresses;
+};
+
+
+
+extern void i915_disassemble_program(const unsigned *program, unsigned sz);
+extern void i915_print_ureg(const char *msg, unsigned ureg);
 
-extern void intelInitSpanFuncs(GLcontext * ctx);
 
-extern void intelSpanRenderFinish(GLcontext * ctx);
-extern void intelSpanRenderStart(GLcontext * ctx);
+void
+i915_dump_batchbuffer( unsigned *start,
+		       unsigned *end );
 
-extern void intel_set_span_functions(struct gl_renderbuffer *rb);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915tex/i915_debug.c b/src/mesa/drivers/dri/i915/i915_debug_fp.c
index 974527e14cc..84347a01efe 100644
--- a/src/mesa/drivers/dri/i915tex/i915_debug.c
+++ b/src/mesa/drivers/dri/i915/i915_debug_fp.c
@@ -25,10 +25,16 @@
  * 
  **************************************************************************/
 
-#include "i915_reg.h"
-#include "i915_context.h"
 #include <stdio.h>
 
+#include "i915_reg.h"
+#include "i915_debug.h"
+#include "main/imports.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_print.h"
+
+#define PRINTF( ... ) _mesa_printf( __VA_ARGS__ )
 
 static const char *opcodes[0x20] = {
    "NOP",
@@ -120,27 +126,27 @@ print_reg_type_nr(GLuint type, GLuint nr)
    case REG_TYPE_T:
       switch (nr) {
       case T_DIFFUSE:
-         fprintf(stderr, "T_DIFFUSE");
+         PRINTF("T_DIFFUSE");
          return;
       case T_SPECULAR:
-         fprintf(stderr, "T_SPECULAR");
+         PRINTF("T_SPECULAR");
          return;
       case T_FOG_W:
-         fprintf(stderr, "T_FOG_W");
+         PRINTF("T_FOG_W");
          return;
       default:
-         fprintf(stderr, "T_TEX%d", nr);
+         PRINTF("T_TEX%d", nr);
          return;
       }
    case REG_TYPE_OC:
       if (nr == 0) {
-         fprintf(stderr, "oC");
+         PRINTF("oC");
          return;
       }
       break;
    case REG_TYPE_OD:
       if (nr == 0) {
-         fprintf(stderr, "oD");
+         PRINTF("oD");
          return;
       }
       break;
@@ -148,7 +154,7 @@ print_reg_type_nr(GLuint type, GLuint nr)
       break;
    }
 
-   fprintf(stderr, "%s[%d]", regname[type], nr);
+   PRINTF("%s[%d]", regname[type], nr);
 }
 
 #define REG_SWIZZLE_MASK 0x7777
@@ -169,33 +175,33 @@ print_reg_neg_swizzle(GLuint reg)
        (reg & REG_NEGATE_MASK) == 0)
       return;
 
-   fprintf(stderr, ".");
+   PRINTF(".");
 
    for (i = 3; i >= 0; i--) {
       if (reg & (1 << ((i * 4) + 3)))
-         fprintf(stderr, "-");
+         PRINTF("-");
 
       switch ((reg >> (i * 4)) & 0x7) {
       case 0:
-         fprintf(stderr, "x");
+         PRINTF("x");
          break;
       case 1:
-         fprintf(stderr, "y");
+         PRINTF("y");
          break;
       case 2:
-         fprintf(stderr, "z");
+         PRINTF("z");
          break;
       case 3:
-         fprintf(stderr, "w");
+         PRINTF("w");
          break;
       case 4:
-         fprintf(stderr, "0");
+         PRINTF("0");
          break;
       case 5:
-         fprintf(stderr, "1");
+         PRINTF("1");
          break;
       default:
-         fprintf(stderr, "?");
+         PRINTF("?");
          break;
       }
    }
@@ -211,13 +217,6 @@ print_src_reg(GLuint dword)
    print_reg_neg_swizzle(dword);
 }
 
-void
-i915_print_ureg(const char *msg, GLuint ureg)
-{
-   fprintf(stderr, "%s: ", msg);
-   print_src_reg(ureg >> 8);
-   fprintf(stderr, "\n");
-}
 
 static void
 print_dest_reg(GLuint dword)
@@ -227,15 +226,15 @@ print_dest_reg(GLuint dword)
    print_reg_type_nr(type, nr);
    if ((dword & A0_DEST_CHANNEL_ALL) == A0_DEST_CHANNEL_ALL)
       return;
-   fprintf(stderr, ".");
+   PRINTF(".");
    if (dword & A0_DEST_CHANNEL_X)
-      fprintf(stderr, "x");
+      PRINTF("x");
    if (dword & A0_DEST_CHANNEL_Y)
-      fprintf(stderr, "y");
+      PRINTF("y");
    if (dword & A0_DEST_CHANNEL_Z)
-      fprintf(stderr, "z");
+      PRINTF("z");
    if (dword & A0_DEST_CHANNEL_W)
-      fprintf(stderr, "w");
+      PRINTF("w");
 }
 
 
@@ -250,29 +249,29 @@ print_arith_op(GLuint opcode, const GLuint * program)
    if (opcode != A0_NOP) {
       print_dest_reg(program[0]);
       if (program[0] & A0_DEST_SATURATE)
-         fprintf(stderr, " = SATURATE ");
+         PRINTF(" = SATURATE ");
       else
-         fprintf(stderr, " = ");
+         PRINTF(" = ");
    }
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   PRINTF("%s ", opcodes[opcode]);
 
    print_src_reg(GET_SRC0_REG(program[0], program[1]));
    if (args[opcode] == 1) {
-      fprintf(stderr, "\n");
+      PRINTF("\n");
       return;
    }
 
-   fprintf(stderr, ", ");
+   PRINTF(", ");
    print_src_reg(GET_SRC1_REG(program[1], program[2]));
    if (args[opcode] == 2) {
-      fprintf(stderr, "\n");
+      PRINTF("\n");
       return;
    }
 
-   fprintf(stderr, ", ");
+   PRINTF(", ");
    print_src_reg(GET_SRC2_REG(program[2]));
-   fprintf(stderr, "\n");
+   PRINTF("\n");
    return;
 }
 
@@ -281,24 +280,24 @@ static void
 print_tex_op(GLuint opcode, const GLuint * program)
 {
    print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, " = ");
+   PRINTF(" = ");
 
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   PRINTF("%s ", opcodes[opcode]);
 
-   fprintf(stderr, "S[%d],", program[0] & T0_SAMPLER_NR_MASK);
+   PRINTF("S[%d],", program[0] & T0_SAMPLER_NR_MASK);
 
    print_reg_type_nr((program[1] >> T1_ADDRESS_REG_TYPE_SHIFT) &
                      REG_TYPE_MASK,
                      (program[1] >> T1_ADDRESS_REG_NR_SHIFT) & REG_NR_MASK);
-   fprintf(stderr, "\n");
+   PRINTF("\n");
 }
 
 static void
 print_dcl_op(GLuint opcode, const GLuint * program)
 {
-   fprintf(stderr, "%s ", opcodes[opcode]);
+   PRINTF("%s ", opcodes[opcode]);
    print_dest_reg(program[0] | A0_DEST_CHANNEL_ALL);
-   fprintf(stderr, "\n");
+   PRINTF("\n");
 }
 
 
@@ -308,18 +307,16 @@ i915_disassemble_program(const GLuint * program, GLuint sz)
    GLuint size = program[0] & 0x1ff;
    GLint i;
 
-   fprintf(stderr, "BEGIN\n");
+   PRINTF("\t\tBEGIN\n");
 
-   if (size + 2 != sz) {
-      fprintf(stderr, "%s: program size mismatch %d/%d\n", __FUNCTION__,
-              size + 2, sz);
-      exit(1);
-   }
+   assert(size + 2 == sz);
 
    program++;
    for (i = 1; i < sz; i += 3, program += 3) {
       GLuint opcode = program[0] & (0x1f << 24);
 
+      PRINTF("\t\t");
+
       if ((GLint) opcode >= A0_NOP && opcode <= A0_SLT)
          print_arith_op(opcode >> 24, program);
       else if (opcode >= T0_TEXLD && opcode <= T0_TEXKILL)
@@ -327,8 +324,10 @@ i915_disassemble_program(const GLuint * program, GLuint sz)
       else if (opcode == D0_DCL)
          print_dcl_op(opcode >> 24, program);
       else
-         fprintf(stderr, "Unknown opcode 0x%x\n", opcode);
+         PRINTF("Unknown opcode 0x%x\n", opcode);
    }
 
-   fprintf(stderr, "END\n\n");
+   PRINTF("\t\tEND\n\n");
 }
+
+
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 702b8788282..4c3f2236e5e 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -29,40 +29,44 @@
 #include "macros.h"
 #include "enums.h"
 
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/program.h"
+#include "shader/programopt.h"
+
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
+
 #include "intel_batchbuffer.h"
 
 #include "i915_reg.h"
 #include "i915_context.h"
 #include "i915_program.h"
 
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "program.h"
-#include "programopt.h"
-
 
 
 /* 1, -1/3!, 1/5!, -1/7! */
-static const GLfloat sin_constants[4] = {  1.0, 
-					   -1.0/(3*2*1),
-					   1.0/(5*4*3*2*1),
-					   -1.0/(7*6*5*4*3*2*1) };
+static const GLfloat sin_constants[4] = { 1.0,
+   -1.0 / (3 * 2 * 1),
+   1.0 / (5 * 4 * 3 * 2 * 1),
+   -1.0 / (7 * 6 * 5 * 4 * 3 * 2 * 1)
+};
 
 /* 1, -1/2!, 1/4!, -1/6! */
-static const GLfloat cos_constants[4] = {  1.0, 
-					   -1.0/(2*1),
-					   1.0/(4*3*2*1),
-					   -1.0/(6*5*4*3*2*1) };
+static const GLfloat cos_constants[4] = { 1.0,
+   -1.0 / (2 * 1),
+   1.0 / (4 * 3 * 2 * 1),
+   -1.0 / (6 * 5 * 4 * 3 * 2 * 1)
+};
 
 /**
  * Retrieve a ureg for the given source register.  Will emit
  * constants, apply swizzling and negation as needed.
  */
-static GLuint src_vector( struct i915_fragment_program *p,
-			  const struct prog_src_register *source,
-			  const struct gl_fragment_program *program )
+static GLuint
+src_vector(struct i915_fragment_program *p,
+           const struct prog_src_register *source,
+           const struct gl_fragment_program *program)
 {
    GLuint src;
 
@@ -70,136 +74,152 @@ static GLuint src_vector( struct i915_fragment_program *p,
 
       /* Registers:
        */
-      case PROGRAM_TEMPORARY:
-	 if (source->Index >= I915_MAX_TEMPORARY) {
-	    i915_program_error( p, "Exceeded max temporary reg" );
-	    return 0;
-	 }
-	 src = UREG( REG_TYPE_R, source->Index );
+   case PROGRAM_TEMPORARY:
+      if (source->Index >= I915_MAX_TEMPORARY) {
+         i915_program_error(p, "Exceeded max temporary reg");
+         return 0;
+      }
+      src = UREG(REG_TYPE_R, source->Index);
+      break;
+   case PROGRAM_INPUT:
+      switch (source->Index) {
+      case FRAG_ATTRIB_WPOS:
+         src = i915_emit_decl(p, REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL);
          break;
-      case PROGRAM_INPUT:
-	 switch (source->Index) {
-	 case FRAG_ATTRIB_WPOS:
-	    src = i915_emit_decl( p,  REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL ); 
-	    break;
-	 case FRAG_ATTRIB_COL0:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL ); 
-	    break;
-	 case FRAG_ATTRIB_COL1:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_SPECULAR, D0_CHANNEL_XYZ ); 
-	    src = swizzle( src, X, Y, Z, ONE );
-	    break;
-	 case FRAG_ATTRIB_FOGC:
-	    src = i915_emit_decl( p,  REG_TYPE_T, T_FOG_W, D0_CHANNEL_W ); 
-	    src = swizzle( src, W, W, W, W );
-	    break;
-	 case FRAG_ATTRIB_TEX0:
-	 case FRAG_ATTRIB_TEX1:
-	 case FRAG_ATTRIB_TEX2:
-	 case FRAG_ATTRIB_TEX3:
-	 case FRAG_ATTRIB_TEX4:
-	 case FRAG_ATTRIB_TEX5:
-	 case FRAG_ATTRIB_TEX6:
-	 case FRAG_ATTRIB_TEX7:
-	    src = i915_emit_decl( p,  REG_TYPE_T, 
-				 T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
-				 D0_CHANNEL_ALL ); 
-	    break;
-
-	 default:
-	    i915_program_error( p, "Bad source->Index" ); 
-	    return 0;
-	 }
+      case FRAG_ATTRIB_COL0:
+         src = i915_emit_decl(p, REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL);
+         break;
+      case FRAG_ATTRIB_COL1:
+         src = i915_emit_decl(p, REG_TYPE_T, T_SPECULAR, D0_CHANNEL_XYZ);
+         src = swizzle(src, X, Y, Z, ONE);
+         break;
+      case FRAG_ATTRIB_FOGC:
+         src = i915_emit_decl(p, REG_TYPE_T, T_FOG_W, D0_CHANNEL_W);
+         src = swizzle(src, W, W, W, W);
+         break;
+      case FRAG_ATTRIB_TEX0:
+      case FRAG_ATTRIB_TEX1:
+      case FRAG_ATTRIB_TEX2:
+      case FRAG_ATTRIB_TEX3:
+      case FRAG_ATTRIB_TEX4:
+      case FRAG_ATTRIB_TEX5:
+      case FRAG_ATTRIB_TEX6:
+      case FRAG_ATTRIB_TEX7:
+         src = i915_emit_decl(p, REG_TYPE_T,
+                              T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
+                              D0_CHANNEL_ALL);
          break;
-
-	 /* Various paramters and env values.  All emitted to
-	  * hardware as program constants.
-	  */
-      case PROGRAM_LOCAL_PARAM:
-         src = i915_emit_param4fv( 
-	    p, program->Base.LocalParams[source->Index]);
-	 break;
-
-      case PROGRAM_ENV_PARAM:
-         src = i915_emit_param4fv( 
-	    p, p->ctx->FragmentProgram.Parameters[source->Index]);
-	 break;
-
-      case PROGRAM_CONSTANT:
-      case PROGRAM_STATE_VAR:
-      case PROGRAM_NAMED_PARAM:
-         src = i915_emit_param4fv( 
-	    p, program->Base.Parameters->ParameterValues[source->Index] );
-	 break;
 
       default:
-	 i915_program_error( p, "Bad source->File" ); 
-	 return 0;
+         i915_program_error(p, "Bad source->Index");
+         return 0;
+      }
+      break;
+
+      /* Various paramters and env values.  All emitted to
+       * hardware as program constants.
+       */
+   case PROGRAM_LOCAL_PARAM:
+      src = i915_emit_param4fv(p, program->Base.LocalParams[source->Index]);
+      break;
+
+   case PROGRAM_ENV_PARAM:
+      src =
+         i915_emit_param4fv(p,
+                            p->ctx->FragmentProgram.Parameters[source->
+                                                               Index]);
+      break;
+
+   case PROGRAM_CONSTANT:
+   case PROGRAM_STATE_VAR:
+   case PROGRAM_NAMED_PARAM:
+      src =
+         i915_emit_param4fv(p,
+                            program->Base.Parameters->ParameterValues[source->
+                                                                      Index]);
+      break;
+
+   default:
+      i915_program_error(p, "Bad source->File");
+      return 0;
    }
 
-   src = swizzle(src, 
-		 GET_SWZ(source->Swizzle, 0),
-		 GET_SWZ(source->Swizzle, 1),
-		 GET_SWZ(source->Swizzle, 2),
-		 GET_SWZ(source->Swizzle, 3));
+   src = swizzle(src,
+                 GET_SWZ(source->Swizzle, 0),
+                 GET_SWZ(source->Swizzle, 1),
+                 GET_SWZ(source->Swizzle, 2), GET_SWZ(source->Swizzle, 3));
 
    if (source->NegateBase)
-      src = negate( src, 
-		    GET_BIT(source->NegateBase, 0),
-		    GET_BIT(source->NegateBase, 1),
-		    GET_BIT(source->NegateBase, 2),
-		    GET_BIT(source->NegateBase, 3));
+      src = negate(src,
+                   GET_BIT(source->NegateBase, 0),
+                   GET_BIT(source->NegateBase, 1),
+                   GET_BIT(source->NegateBase, 2),
+                   GET_BIT(source->NegateBase, 3));
 
    return src;
 }
 
 
-static GLuint get_result_vector( struct i915_fragment_program *p,
-				 const struct prog_instruction *inst )
+static GLuint
+get_result_vector(struct i915_fragment_program *p,
+                  const struct prog_instruction *inst)
 {
    switch (inst->DstReg.File) {
    case PROGRAM_OUTPUT:
       switch (inst->DstReg.Index) {
-      case FRAG_RESULT_COLR: 
-	 return UREG(REG_TYPE_OC, 0);
-      case FRAG_RESULT_DEPR: 
-	 p->depth_written = 1;
-	 return UREG(REG_TYPE_OD, 0);
-      default: 
-	 i915_program_error( p, "Bad inst->DstReg.Index" ); 
-	 return 0;
+      case FRAG_RESULT_COLR:
+         return UREG(REG_TYPE_OC, 0);
+      case FRAG_RESULT_DEPR:
+         p->depth_written = 1;
+         return UREG(REG_TYPE_OD, 0);
+      default:
+         i915_program_error(p, "Bad inst->DstReg.Index");
+         return 0;
       }
    case PROGRAM_TEMPORARY:
       return UREG(REG_TYPE_R, inst->DstReg.Index);
    default:
-      i915_program_error( p, "Bad inst->DstReg.File" ); 
+      i915_program_error(p, "Bad inst->DstReg.File");
       return 0;
    }
 }
-   
-static GLuint get_result_flags( const struct prog_instruction *inst )
+
+static GLuint
+get_result_flags(const struct prog_instruction *inst)
 {
    GLuint flags = 0;
 
-   if (inst->SaturateMode == SATURATE_ZERO_ONE) flags |= A0_DEST_SATURATE;
-   if (inst->DstReg.WriteMask & WRITEMASK_X) flags |= A0_DEST_CHANNEL_X;
-   if (inst->DstReg.WriteMask & WRITEMASK_Y) flags |= A0_DEST_CHANNEL_Y;
-   if (inst->DstReg.WriteMask & WRITEMASK_Z) flags |= A0_DEST_CHANNEL_Z;
-   if (inst->DstReg.WriteMask & WRITEMASK_W) flags |= A0_DEST_CHANNEL_W;
+   if (inst->SaturateMode == SATURATE_ZERO_ONE)
+      flags |= A0_DEST_SATURATE;
+   if (inst->DstReg.WriteMask & WRITEMASK_X)
+      flags |= A0_DEST_CHANNEL_X;
+   if (inst->DstReg.WriteMask & WRITEMASK_Y)
+      flags |= A0_DEST_CHANNEL_Y;
+   if (inst->DstReg.WriteMask & WRITEMASK_Z)
+      flags |= A0_DEST_CHANNEL_Z;
+   if (inst->DstReg.WriteMask & WRITEMASK_W)
+      flags |= A0_DEST_CHANNEL_W;
 
    return flags;
 }
 
-static GLuint translate_tex_src_target( struct i915_fragment_program *p,
-				     GLubyte bit )
+static GLuint
+translate_tex_src_target(struct i915_fragment_program *p, GLubyte bit)
 {
    switch (bit) {
-   case TEXTURE_1D_INDEX:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_2D_INDEX:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_RECT_INDEX: return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_3D_INDEX:   return D0_SAMPLE_TYPE_VOLUME;
-   case TEXTURE_CUBE_INDEX: return D0_SAMPLE_TYPE_CUBE;
-   default: i915_program_error(p, "TexSrcBit"); return 0;
+   case TEXTURE_1D_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_2D_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_RECT_INDEX:
+      return D0_SAMPLE_TYPE_2D;
+   case TEXTURE_3D_INDEX:
+      return D0_SAMPLE_TYPE_VOLUME;
+   case TEXTURE_CUBE_INDEX:
+      return D0_SAMPLE_TYPE_CUBE;
+   default:
+      i915_program_error(p, "TexSrcBit");
+      return 0;
    }
 }
 
@@ -246,9 +266,11 @@ do {									\
  * can lead to confusion -- hopefully we cope with it ok now.
  *
  */
-static void upload_program( struct i915_fragment_program *p )
+static void
+upload_program(struct i915_fragment_program *p)
 {
-   const struct gl_fragment_program *program = p->ctx->FragmentProgram._Current;
+   const struct gl_fragment_program *program =
+      p->ctx->FragmentProgram._Current;
    const struct prog_instruction *inst = program->Base.Instructions;
 
 /*    _mesa_debug_fp_inst(program->Base.NumInstructions, inst); */
@@ -258,12 +280,12 @@ static void upload_program( struct i915_fragment_program *p )
     * this being uploaded to hardware.
     */
    if (inst[0].Opcode == OPCODE_END) {
-      GLuint tmp = i915_get_utemp( p );
-      i915_emit_arith( p,
-		      A0_MOV,
-		      UREG(REG_TYPE_OC, 0), 
-		      A0_DEST_CHANNEL_ALL, 0,
-		      swizzle(tmp,ONE,ZERO,ONE,ONE), 0, 0);
+      GLuint tmp = i915_get_utemp(p);
+      i915_emit_arith(p,
+                      A0_MOV,
+                      UREG(REG_TYPE_OC, 0),
+                      A0_DEST_CHANNEL_ALL, 0,
+                      swizzle(tmp, ONE, ZERO, ONE, ONE), 0, 0);
       return;
    }
 
@@ -272,496 +294,465 @@ static void upload_program( struct i915_fragment_program *p )
       GLuint tmp = 0;
 
       switch (inst->Opcode) {
-      case OPCODE_ABS: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 i915_emit_arith( p, 
-			 A0_MAX,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, negate(src0, 1,1,1,1), 0);
-	 break;
-
-      case OPCODE_ADD: 
-	 EMIT_2ARG_ARITH( A0_ADD );
-	 break;
-
-      case OPCODE_CMP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 src2 = src_vector( p, &inst->SrcReg[2], program);
-	 i915_emit_arith( p, 
-			 A0_CMP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, src2, src1);	/* NOTE: order of src2, src1 */
-	 break;
+      case OPCODE_ABS:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         i915_emit_arith(p,
+                         A0_MAX,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         src0, negate(src0, 1, 1, 1, 1), 0);
+         break;
+
+      case OPCODE_ADD:
+         EMIT_2ARG_ARITH(A0_ADD);
+         break;
+
+      case OPCODE_CMP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         src2 = src_vector(p, &inst->SrcReg[2], program);
+         i915_emit_arith(p, A0_CMP, get_result_vector(p, inst), get_result_flags(inst), 0, src0, src2, src1);   /* NOTE: order of src2, src1 */
+         break;
 
       case OPCODE_COS:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
-			 0);
-
-	 i915_emit_arith( p, 
-			 A0_MOD,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 0, 0 );
-
-	 /* By choosing different taylor constants, could get rid of this mul:
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
-			 0);
-
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, 1
-	  * t0 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-	  * result = DP4 t0, cos_constants
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(tmp, X,X,ONE,ONE), 
-			 swizzle(tmp, X,ONE,ONE,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XYZ, 0,
-			 swizzle(tmp, X,Y,X,ONE), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XYZ, 0,
-			 swizzle(tmp, X,X,Z,ONE), 
-			 swizzle(tmp, Z,ONE,ONE,ONE), 0);
-	    
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(tmp, ONE,Z,Y,X),
-			 i915_emit_const4fv( p, cos_constants ), 0);
-
-	 break;
-
-      case OPCODE_DP3: 
-	 EMIT_2ARG_ARITH( A0_DP3 );
-	 break;
-
-      case OPCODE_DP4: 
-	 EMIT_2ARG_ARITH( A0_DP4 );
-	 break;
-
-      case OPCODE_DPH:  
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0, X,Y,Z,ONE), src1, 0);
-	 break;
-
-      case OPCODE_DST: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 /* result[0] = 1    * 1;
-	  * result[1] = a[1] * b[1];
-	  * result[2] = a[2] * 1;
-	  * result[3] = 1    * b[3];
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0, ONE, Y, Z,   ONE), 
-			 swizzle(src1, ONE, Y, ONE, W  ),
-			 0);
-	 break;
-
-      case OPCODE_EX2: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-
-	 i915_emit_arith( p, 
-			 A0_EXP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-
-      case OPCODE_FLR: 
-	 EMIT_1ARG_ARITH( A0_FLR );
-	 break;
-
-      case OPCODE_FRC: 
-	 EMIT_1ARG_ARITH( A0_FRC );
-	 break;
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         src0, i915_emit_const1f(p, 1.0 / (M_PI)), 0);
+
+         i915_emit_arith(p, A0_MOD, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
+
+         /* By choosing different taylor constants, could get rid of this mul:
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         tmp, i915_emit_const1f(p, (M_PI)), 0);
+
+         /* 
+          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
+          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, 1
+          * t0 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
+          * result = DP4 t0, cos_constants
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XY, 0,
+                         swizzle(tmp, X, X, ONE, ONE),
+                         swizzle(tmp, X, ONE, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XYZ, 0,
+                         swizzle(tmp, X, Y, X, ONE),
+                         swizzle(tmp, X, X, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XYZ, 0,
+                         swizzle(tmp, X, X, Z, ONE),
+                         swizzle(tmp, Z, ONE, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_DP4,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(tmp, ONE, Z, Y, X),
+                         i915_emit_const4fv(p, cos_constants), 0);
+
+         break;
+
+      case OPCODE_DP3:
+         EMIT_2ARG_ARITH(A0_DP3);
+         break;
+
+      case OPCODE_DP4:
+         EMIT_2ARG_ARITH(A0_DP4);
+         break;
+
+      case OPCODE_DPH:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+
+         i915_emit_arith(p,
+                         A0_DP4,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, Y, Z, ONE), src1, 0);
+         break;
+
+      case OPCODE_DST:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+
+         /* result[0] = 1    * 1;
+          * result[1] = a[1] * b[1];
+          * result[2] = a[2] * 1;
+          * result[3] = 1    * b[3];
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, ONE, Y, Z, ONE),
+                         swizzle(src1, ONE, Y, ONE, W), 0);
+         break;
+
+      case OPCODE_EX2:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_EXP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_FLR:
+         EMIT_1ARG_ARITH(A0_FLR);
+         break;
+
+      case OPCODE_FRC:
+         EMIT_1ARG_ARITH(A0_FRC);
+         break;
 
       case OPCODE_KIL:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 i915_emit_texld( p,
-			 tmp, A0_DEST_CHANNEL_ALL, /* use a dummy dest reg */
-			 0,
-			 src0,
-			 T0_TEXKILL );
-	 break;
-
-      case OPCODE_LG2: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-
-	 i915_emit_arith( p, 
-			 A0_LOG,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-
-      case OPCODE_LIT: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 /* tmp = max( a.xyzw, a.00zw )
-	  * XXX: Clamp tmp.w to -128..128
-	  * tmp.y = log(tmp.y)
-	  * tmp.y = tmp.w * tmp.y
-	  * tmp.y = exp(tmp.y)
-	  * result = cmp (a.11-x1, a.1x01, a.1xy1 )
-	  */
-	 i915_emit_arith( p, A0_MAX, tmp, A0_DEST_CHANNEL_ALL, 0, 
-			 src0, swizzle(src0, ZERO, ZERO, Z, W), 0 );
-
-	 i915_emit_arith( p, A0_LOG, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, Y, Y, Y, Y), 0, 0 );
-
-	 i915_emit_arith( p, A0_MUL, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, ZERO, Y, ZERO, ZERO), 
-			 swizzle(tmp, ZERO, W, ZERO, ZERO), 0 );
-
-	 i915_emit_arith( p, A0_EXP, tmp, A0_DEST_CHANNEL_Y, 0, 
-			 swizzle(tmp, Y, Y, Y, Y), 0, 0 );
-
-	 i915_emit_arith( p, A0_CMP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 negate(swizzle(tmp, ONE, ONE, X, ONE),0,0,1,0),
-			 swizzle(tmp, ONE, X, ZERO, ONE),
-			 swizzle(tmp, ONE, X, Y, ONE));
-		     
-	 break;
-
-      case OPCODE_LRP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 src2 = src_vector( p, &inst->SrcReg[2], program);
-	 flags = get_result_flags( inst );
-	 tmp = i915_get_utemp( p );
-
-	 /* b*a + c*(1-a)
-	  *
-	  * b*a + c - ca 
-	  *
-	  * tmp = b*a + c, 
-	  * result = (-c)*a + tmp 
-	  */
-	 i915_emit_arith( p, A0_MAD, tmp, 
-			 flags & A0_DEST_CHANNEL_ALL, 0,
-			 src1, src0, src2 );
-
-	 i915_emit_arith( p, A0_MAD, 
-			 get_result_vector( p, inst ), 
-			 flags, 0, 
-			 negate(src2, 1,1,1,1), src0, tmp );
-	 break;
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_texld(p, tmp, A0_DEST_CHANNEL_ALL,   /* use a dummy dest reg */
+                         0, src0, T0_TEXKILL);
+         break;
+
+      case OPCODE_LG2:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_LOG,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_LIT:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         /* tmp = max( a.xyzw, a.00zw )
+          * XXX: Clamp tmp.w to -128..128
+          * tmp.y = log(tmp.y)
+          * tmp.y = tmp.w * tmp.y
+          * tmp.y = exp(tmp.y)
+          * result = cmp (a.11-x1, a.1x01, a.1xy1 )
+          */
+         i915_emit_arith(p, A0_MAX, tmp, A0_DEST_CHANNEL_ALL, 0,
+                         src0, swizzle(src0, ZERO, ZERO, Z, W), 0);
+
+         i915_emit_arith(p, A0_LOG, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
+
+         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, ZERO, Y, ZERO, ZERO),
+                         swizzle(tmp, ZERO, W, ZERO, ZERO), 0);
+
+         i915_emit_arith(p, A0_EXP, tmp, A0_DEST_CHANNEL_Y, 0,
+                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
+
+         i915_emit_arith(p, A0_CMP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         negate(swizzle(tmp, ONE, ONE, X, ONE), 0, 0, 1, 0),
+                         swizzle(tmp, ONE, X, ZERO, ONE),
+                         swizzle(tmp, ONE, X, Y, ONE));
+
+         break;
+
+      case OPCODE_LRP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         src2 = src_vector(p, &inst->SrcReg[2], program);
+         flags = get_result_flags(inst);
+         tmp = i915_get_utemp(p);
+
+         /* b*a + c*(1-a)
+          *
+          * b*a + c - ca 
+          *
+          * tmp = b*a + c, 
+          * result = (-c)*a + tmp 
+          */
+         i915_emit_arith(p, A0_MAD, tmp,
+                         flags & A0_DEST_CHANNEL_ALL, 0, src1, src0, src2);
+
+         i915_emit_arith(p, A0_MAD,
+                         get_result_vector(p, inst),
+                         flags, 0, negate(src2, 1, 1, 1, 1), src0, tmp);
+         break;
 
       case OPCODE_MAD:
-	 EMIT_3ARG_ARITH( A0_MAD );
-	 break;
+         EMIT_3ARG_ARITH(A0_MAD);
+         break;
 
       case OPCODE_MAX:
-	 EMIT_2ARG_ARITH( A0_MAX );
-	 break;
-
-      case OPCODE_MIN: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 flags = get_result_flags( inst );
-
-	 i915_emit_arith( p, 
-			 A0_MAX,
-			 tmp, flags & A0_DEST_CHANNEL_ALL, 0,
-			 negate(src0,1,1,1,1), 
-			 negate(src1,1,1,1,1), 0);
-
-	 i915_emit_arith( p,
-			 A0_MOV,
-			 get_result_vector( p, inst ), 
-			 flags, 0,
-			 negate(tmp, 1,1,1,1), 0, 0);
-	 break;
-
-      case OPCODE_MOV: 
-	 EMIT_1ARG_ARITH( A0_MOV );
-	 break;
-
-      case OPCODE_MUL: 
-	 EMIT_2ARG_ARITH( A0_MUL );
-	 break;
-
-      case OPCODE_POW: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 flags = get_result_flags( inst );
-
-	 /* XXX: masking on intermediate values, here and elsewhere.
-	  */
-	 i915_emit_arith( p, 
-			 A0_LOG,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-
-	 i915_emit_arith( p,
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, src1, 0);
-
-
-	 i915_emit_arith( p,
-			 A0_EXP,
-			 get_result_vector( p, inst ), 
-			 flags, 0,
-			 swizzle(tmp,X,X,X,X), 0, 0);
-
-	 break;
-
-      case OPCODE_RCP: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-
-	 i915_emit_arith( p, 
-			 A0_RCP,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-
-      case OPCODE_RSQ: 
-
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-
-	 i915_emit_arith( p, 
-			 A0_RSQ,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,X,X,X,X), 0, 0);
-	 break;
-	 
+         EMIT_2ARG_ARITH(A0_MAX);
+         break;
+
+      case OPCODE_MIN:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+         flags = get_result_flags(inst);
+
+         i915_emit_arith(p,
+                         A0_MAX,
+                         tmp, flags & A0_DEST_CHANNEL_ALL, 0,
+                         negate(src0, 1, 1, 1, 1),
+                         negate(src1, 1, 1, 1, 1), 0);
+
+         i915_emit_arith(p,
+                         A0_MOV,
+                         get_result_vector(p, inst),
+                         flags, 0, negate(tmp, 1, 1, 1, 1), 0, 0);
+         break;
+
+      case OPCODE_MOV:
+         EMIT_1ARG_ARITH(A0_MOV);
+         break;
+
+      case OPCODE_MUL:
+         EMIT_2ARG_ARITH(A0_MUL);
+         break;
+
+      case OPCODE_POW:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+         flags = get_result_flags(inst);
+
+         /* XXX: masking on intermediate values, here and elsewhere.
+          */
+         i915_emit_arith(p,
+                         A0_LOG,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+
+         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_X, 0, tmp, src1, 0);
+
+
+         i915_emit_arith(p,
+                         A0_EXP,
+                         get_result_vector(p, inst),
+                         flags, 0, swizzle(tmp, X, X, X, X), 0, 0);
+
+         break;
+
+      case OPCODE_RCP:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_RCP,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
+      case OPCODE_RSQ:
+
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+
+         i915_emit_arith(p,
+                         A0_RSQ,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, X, X, X, X), 0, 0);
+         break;
+
       case OPCODE_SCS:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-	  * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-	  * scs.x = DP4 t1, sin_constants
-	  * t1 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-	  * scs.y = DP4 t1, cos_constants
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(src0, X,X,ONE,ONE), 
-			 swizzle(src0, X,ONE,ONE,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,X,Y), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
-
-	 if (inst->DstReg.WriteMask & WRITEMASK_Y) {
-	    GLuint tmp1;
-	    
-	    if (inst->DstReg.WriteMask & WRITEMASK_X)
-	       tmp1 = i915_get_utemp( p );
-	    else
-	       tmp1 = tmp;
-
-	    i915_emit_arith( p, 
-			    A0_MUL,
-			    tmp1, A0_DEST_CHANNEL_ALL, 0,
-			    swizzle(tmp, X,Y,Y,W), 
-			    swizzle(tmp, X,Z,ONE,ONE), 0);
-	    
-	    i915_emit_arith( p, 
-			    A0_DP4,
-			    get_result_vector( p, inst ), 
-			    A0_DEST_CHANNEL_Y, 0,
-			    swizzle(tmp1, W,Z,Y,X),
-			    i915_emit_const4fv( p, sin_constants ), 0);
-	 }
-
-	 if (inst->DstReg.WriteMask & WRITEMASK_X) {
-	    i915_emit_arith( p, 
-			    A0_MUL,
-			    tmp, A0_DEST_CHANNEL_XYZ, 0,
-			    swizzle(tmp, X,X,Z,ONE), 
-			    swizzle(tmp, Z,ONE,ONE,ONE), 0);
-	    
-	    i915_emit_arith( p, 
-			    A0_DP4,
-			    get_result_vector( p, inst ), 
-			    A0_DEST_CHANNEL_X, 0,
-			    swizzle(tmp, ONE,Z,Y,X), 
-			    i915_emit_const4fv( p, cos_constants ), 0);
-	 }
-	 break;
-
-      case OPCODE_SGE: 
-	 EMIT_2ARG_ARITH( A0_SGE );
-	 break;
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         /* 
+          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
+          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
+          * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
+          * scs.x = DP4 t1, sin_constants
+          * t1 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
+          * scs.y = DP4 t1, cos_constants
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XY, 0,
+                         swizzle(src0, X, X, ONE, ONE),
+                         swizzle(src0, X, ONE, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(tmp, X, Y, X, Y),
+                         swizzle(tmp, X, X, ONE, ONE), 0);
+
+         if (inst->DstReg.WriteMask & WRITEMASK_Y) {
+            GLuint tmp1;
+
+            if (inst->DstReg.WriteMask & WRITEMASK_X)
+               tmp1 = i915_get_utemp(p);
+            else
+               tmp1 = tmp;
+
+            i915_emit_arith(p,
+                            A0_MUL,
+                            tmp1, A0_DEST_CHANNEL_ALL, 0,
+                            swizzle(tmp, X, Y, Y, W),
+                            swizzle(tmp, X, Z, ONE, ONE), 0);
+
+            i915_emit_arith(p,
+                            A0_DP4,
+                            get_result_vector(p, inst),
+                            A0_DEST_CHANNEL_Y, 0,
+                            swizzle(tmp1, W, Z, Y, X),
+                            i915_emit_const4fv(p, sin_constants), 0);
+         }
+
+         if (inst->DstReg.WriteMask & WRITEMASK_X) {
+            i915_emit_arith(p,
+                            A0_MUL,
+                            tmp, A0_DEST_CHANNEL_XYZ, 0,
+                            swizzle(tmp, X, X, Z, ONE),
+                            swizzle(tmp, Z, ONE, ONE, ONE), 0);
+
+            i915_emit_arith(p,
+                            A0_DP4,
+                            get_result_vector(p, inst),
+                            A0_DEST_CHANNEL_X, 0,
+                            swizzle(tmp, ONE, Z, Y, X),
+                            i915_emit_const4fv(p, cos_constants), 0);
+         }
+         break;
+
+      case OPCODE_SGE:
+         EMIT_2ARG_ARITH(A0_SGE);
+         break;
 
       case OPCODE_SIN:
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 tmp = i915_get_utemp( p );
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 src0, 
-			 i915_emit_const1f(p, 1.0/(M_PI * 2)),
-			 0);
-
-	 i915_emit_arith( p, 
-			 A0_MOD,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 0, 0 );
-
-	 /* By choosing different taylor constants, could get rid of this mul:
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_X, 0,
-			 tmp, 
-			 i915_emit_const1f(p, (M_PI * 2)),
-			 0);
-
-	 /* 
-	  * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-	  * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-	  * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-	  * result = DP4 t1.wzyx, sin_constants
-	  */
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_XY, 0,
-			 swizzle(tmp, X,X,ONE,ONE), 
-			 swizzle(tmp, X,ONE,ONE,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,X,Y), 
-			 swizzle(tmp, X,X,ONE,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(tmp, X,Y,Y,W), 
-			 swizzle(tmp, X,Z,ONE,ONE), 0);
-	    
-	 i915_emit_arith( p, 
-			 A0_DP4,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(tmp, W, Z, Y, X ),
-			 i915_emit_const4fv( p, sin_constants ), 0);
-	 break;
-
-      case OPCODE_SLT: 
-	 EMIT_2ARG_ARITH( A0_SLT );
-	 break;
-
-      case OPCODE_SUB: 
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-
-	 i915_emit_arith( p, 
-			 A0_ADD,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 src0, negate(src1, 1,1,1,1), 0);
-	 break;
-
-      case OPCODE_SWZ: 
-	 EMIT_1ARG_ARITH( A0_MOV ); /* extended swizzle handled natively */
-	 break;
-
-      case OPCODE_TEX: 
-	 EMIT_TEX( T0_TEXLD );
-	 break;
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         src0, i915_emit_const1f(p, 1.0 / (M_PI)), 0);
+
+         i915_emit_arith(p, A0_MOD, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
+
+         /* By choosing different taylor constants, could get rid of this mul:
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_X, 0,
+                         tmp, i915_emit_const1f(p, (M_PI)), 0);
+
+         /* 
+          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
+          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
+          * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
+          * result = DP4 t1.wzyx, sin_constants
+          */
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_XY, 0,
+                         swizzle(tmp, X, X, ONE, ONE),
+                         swizzle(tmp, X, ONE, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(tmp, X, Y, X, Y),
+                         swizzle(tmp, X, X, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(tmp, X, Y, Y, W),
+                         swizzle(tmp, X, Z, ONE, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_DP4,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(tmp, W, Z, Y, X),
+                         i915_emit_const4fv(p, sin_constants), 0);
+         break;
+
+      case OPCODE_SLT:
+         EMIT_2ARG_ARITH(A0_SLT);
+         break;
+
+      case OPCODE_SUB:
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+
+         i915_emit_arith(p,
+                         A0_ADD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         src0, negate(src1, 1, 1, 1, 1), 0);
+         break;
+
+      case OPCODE_SWZ:
+         EMIT_1ARG_ARITH(A0_MOV);       /* extended swizzle handled natively */
+         break;
+
+      case OPCODE_TEX:
+         EMIT_TEX(T0_TEXLD);
+         break;
 
       case OPCODE_TXB:
-	 EMIT_TEX( T0_TEXLDB );
-	 break;
+         EMIT_TEX(T0_TEXLDB);
+         break;
 
       case OPCODE_TXP:
-	 EMIT_TEX( T0_TEXLDP );
-	 break;
+         EMIT_TEX(T0_TEXLDP);
+         break;
 
       case OPCODE_XPD:
-	 /* Cross product:
-	  *      result.x = src0.y * src1.z - src0.z * src1.y;
-	  *      result.y = src0.z * src1.x - src0.x * src1.z;
-	  *      result.z = src0.x * src1.y - src0.y * src1.x;
-	  *      result.w = undef;
-	  */
-	 src0 = src_vector( p, &inst->SrcReg[0], program);
-	 src1 = src_vector( p, &inst->SrcReg[1], program);
-	 tmp = i915_get_utemp( p );
-	 
-	 i915_emit_arith( p, 
-			 A0_MUL,
-			 tmp, A0_DEST_CHANNEL_ALL, 0,
-			 swizzle(src0,Z,X,Y,ONE), 
-			 swizzle(src1,Y,Z,X,ONE), 0);
-
-	 i915_emit_arith( p, 
-			 A0_MAD,
-			 get_result_vector( p, inst ), 
-			 get_result_flags( inst ), 0,
-			 swizzle(src0,Y,Z,X,ONE), 
-			 swizzle(src1,Z,X,Y,ONE), 
-			 negate(tmp,1,1,1,0));
-	 break;
+         /* Cross product:
+          *      result.x = src0.y * src1.z - src0.z * src1.y;
+          *      result.y = src0.z * src1.x - src0.x * src1.z;
+          *      result.z = src0.x * src1.y - src0.y * src1.x;
+          *      result.w = undef;
+          */
+         src0 = src_vector(p, &inst->SrcReg[0], program);
+         src1 = src_vector(p, &inst->SrcReg[1], program);
+         tmp = i915_get_utemp(p);
+
+         i915_emit_arith(p,
+                         A0_MUL,
+                         tmp, A0_DEST_CHANNEL_ALL, 0,
+                         swizzle(src0, Z, X, Y, ONE),
+                         swizzle(src1, Y, Z, X, ONE), 0);
+
+         i915_emit_arith(p,
+                         A0_MAD,
+                         get_result_vector(p, inst),
+                         get_result_flags(inst), 0,
+                         swizzle(src0, Y, Z, X, ONE),
+                         swizzle(src1, Z, X, Y, ONE),
+                         negate(tmp, 1, 1, 1, 0));
+         break;
 
       case OPCODE_END:
-	 return;
-	 
+         return;
+
       default:
-	 i915_program_error( p, "bad opcode" );
-	 return;
+         i915_program_error(p, "bad opcode");
+         return;
       }
 
       inst++;
-      i915_release_utemps( p ); 
+      i915_release_utemps(p);
    }
 }
 
@@ -769,21 +760,22 @@ static void upload_program( struct i915_fragment_program *p )
  * emit, just move the value into its correct position at the end of
  * the program:
  */
-static void fixup_depth_write( struct i915_fragment_program *p )
+static void
+fixup_depth_write(struct i915_fragment_program *p)
 {
    if (p->depth_written) {
       GLuint depth = UREG(REG_TYPE_OD, 0);
 
-      i915_emit_arith( p, 
-		      A0_MOV,
-		      depth, A0_DEST_CHANNEL_W, 0,
-		      swizzle(depth,X,Y,Z,Z), 
-		      0, 0);
+      i915_emit_arith(p,
+                      A0_MOV,
+                      depth, A0_DEST_CHANNEL_W, 0,
+                      swizzle(depth, X, Y, Z, Z), 0, 0);
    }
 }
 
 
-static void check_wpos( struct i915_fragment_program *p )
+static void
+check_wpos(struct i915_fragment_program *p)
 {
    GLuint inputs = p->FragProg.Base.InputsRead;
    GLint i;
@@ -791,12 +783,12 @@ static void check_wpos( struct i915_fragment_program *p )
    p->wpos_tex = -1;
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
-      if (inputs & FRAG_BIT_TEX(i)) 
-	 continue;
+      if (inputs & FRAG_BIT_TEX(i))
+         continue;
       else if (inputs & FRAG_BIT_WPOS) {
-	 p->wpos_tex = i;
-	 inputs &= ~FRAG_BIT_WPOS;
-      }   
+         p->wpos_tex = i;
+         inputs &= ~FRAG_BIT_WPOS;
+      }
    }
 
    if (inputs & FRAG_BIT_WPOS) {
@@ -805,53 +797,54 @@ static void check_wpos( struct i915_fragment_program *p )
 }
 
 
-static void translate_program( struct i915_fragment_program *p )
+static void
+translate_program(struct i915_fragment_program *p)
 {
-   i915ContextPtr i915 = I915_CONTEXT(p->ctx);
-   
-   i915_init_program( i915, p );
-   check_wpos( p ); 
-   upload_program( p );
-   fixup_depth_write( p );
-   i915_fini_program( p ); 
-   
+   struct i915_context *i915 = I915_CONTEXT(p->ctx);
+
+   i915_init_program(i915, p);
+   check_wpos(p);
+   upload_program(p);
+   fixup_depth_write(p);
+   i915_fini_program(p);
+
    p->translated = 1;
 }
 
 
-static void track_params( struct i915_fragment_program *p )
+static void
+track_params(struct i915_fragment_program *p)
 {
    GLint i;
 
    if (p->nr_params)
-      _mesa_load_state_parameters(p->ctx, p->FragProg.Base.Parameters); 
+      _mesa_load_state_parameters(p->ctx, p->FragProg.Base.Parameters);
 
    for (i = 0; i < p->nr_params; i++) {
       GLint reg = p->param[i].reg;
-      COPY_4V( p->constant[reg], p->param[i].values );
+      COPY_4V(p->constant[reg], p->param[i].values);
    }
-   
+
    p->params_uptodate = 1;
-   p->on_hardware = 0;		/* overkill */
+   p->on_hardware = 0;          /* overkill */
 }
 
 
-static void i915BindProgram( GLcontext *ctx,
-			    GLenum target, 
-			    struct gl_program *prog )
+static void
+i915BindProgram(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      i915ContextPtr i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_context *i915 = I915_CONTEXT(ctx);
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
+
+      if (i915->current_program == p)
+         return;
 
-      if (i915->current_program == p) 
-	 return;
-      
       if (i915->current_program) {
-	 i915->current_program->on_hardware = 0;
-	 i915->current_program->params_uptodate = 0;
+         i915->current_program->on_hardware = 0;
+         i915->current_program->params_uptodate = 0;
       }
-      
+
       i915->current_program = p;
 
       assert(p->on_hardware == 0);
@@ -860,71 +853,70 @@ static void i915BindProgram( GLcontext *ctx,
    }
 }
 
-static struct gl_program *i915NewProgram( GLcontext *ctx,
-				      GLenum target, 
-				      GLuint id )
+static struct gl_program *
+i915NewProgram(GLcontext * ctx, GLenum target, GLuint id)
 {
    switch (target) {
    case GL_VERTEX_PROGRAM_ARB:
-      return _mesa_init_vertex_program( ctx, CALLOC_STRUCT(gl_vertex_program),
-					target, id );
-
-   case GL_FRAGMENT_PROGRAM_ARB: {
-      struct i915_fragment_program *prog = CALLOC_STRUCT(i915_fragment_program);
-      if (prog) {
-	 i915_init_program( I915_CONTEXT(ctx), prog );
-
-	 return _mesa_init_fragment_program( ctx, &prog->FragProg,
-					     target, id );
+      return _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
+                                       target, id);
+
+   case GL_FRAGMENT_PROGRAM_ARB:{
+         struct i915_fragment_program *prog =
+            CALLOC_STRUCT(i915_fragment_program);
+         if (prog) {
+            i915_init_program(I915_CONTEXT(ctx), prog);
+
+            return _mesa_init_fragment_program(ctx, &prog->FragProg,
+                                               target, id);
+         }
+         else
+            return NULL;
       }
-      else
-	 return NULL;
-   }
 
    default:
       /* Just fallback:
        */
-      return _mesa_new_program( ctx, target, id );
+      return _mesa_new_program(ctx, target, id);
    }
 }
 
-static void i915DeleteProgram( GLcontext *ctx,
-			      struct gl_program *prog )
+static void
+i915DeleteProgram(GLcontext * ctx, struct gl_program *prog)
 {
    if (prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
-      i915ContextPtr i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
-      
-      if (i915->current_program == p) 
-	 i915->current_program = 0;
+      struct i915_context *i915 = I915_CONTEXT(ctx);
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
+
+      if (i915->current_program == p)
+         i915->current_program = 0;
    }
 
-   _mesa_delete_program( ctx, prog );
+   _mesa_delete_program(ctx, prog);
 }
 
 
-static GLboolean i915IsProgramNative( GLcontext *ctx,
-				     GLenum target, 
-				     struct gl_program *prog )
+static GLboolean
+i915IsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
 
       if (!p->translated)
-	 translate_program( p );
-      
+         translate_program(p);
+
       return !p->error;
    }
    else
       return GL_TRUE;
 }
 
-static void i915ProgramStringNotify( GLcontext *ctx,
-				    GLenum target,
-				    struct gl_program *prog )
+static void
+i915ProgramStringNotify(GLcontext * ctx,
+                        GLenum target, struct gl_program *prog)
 {
    if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *)prog;
+      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
       p->translated = 0;
 
       /* Hack: make sure fog is correctly enabled according to this
@@ -941,28 +933,28 @@ static void i915ProgramStringNotify( GLcontext *ctx,
 }
 
 
-void i915ValidateFragmentProgram( i915ContextPtr i915 )
+void
+i915ValidateFragmentProgram(struct i915_context *i915)
 {
    GLcontext *ctx = &i915->intel.ctx;
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
 
-   struct i915_fragment_program *p = 
-      (struct i915_fragment_program *)ctx->FragmentProgram._Current;
+   struct i915_fragment_program *p =
+      (struct i915_fragment_program *) ctx->FragmentProgram._Current;
 
    const GLuint inputsRead = p->FragProg.Base.InputsRead;
    GLuint s4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_VFMT_MASK;
    GLuint s2 = S2_TEXCOORD_NONE;
    int i, offset = 0;
 
-   if (i915->current_program != p) 
-   {
+   if (i915->current_program != p) {
       if (i915->current_program) {
-	 i915->current_program->on_hardware = 0;
-	 i915->current_program->params_uptodate = 0;
+         i915->current_program->on_hardware = 0;
+         i915->current_program->params_uptodate = 0;
       }
-      
+
       i915->current_program = p;
    }
 
@@ -971,8 +963,8 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
     */
    VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
 
-   if (!p->translated) 
-      translate_program( p );
+   if (!p->translated)
+      translate_program(p);
 
    intel->vertex_attr_count = 0;
    intel->wpos_offset = 0;
@@ -981,31 +973,31 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
    intel->specoffset = 0;
 
    if (inputsRead & FRAG_BITS_TEX_ANY) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16);
    }
    else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12 );
+      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12);
    }
 
    if (inputsRead & FRAG_BIT_COL0) {
       intel->coloroffset = offset / 4;
-      EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4 );
+      EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4);
    }
-   
-   if ((inputsRead & (FRAG_BIT_COL1|FRAG_BIT_FOGC)) || 
+
+   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 );
+         intel->specoffset = offset / 4;
+         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3);
       }
       else
-	 EMIT_PAD(3);
+         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 );
+      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 );
+         EMIT_PAD(1);
    }
 
    /* XXX this was disabled, but enabling this code helped fix the Glean
@@ -1013,63 +1005,66 @@ void i915ValidateFragmentProgram( i915ContextPtr i915 )
     */
 #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 );
+      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)) {
-	 int sz = VB->TexCoordPtr[i]->size;
-	    
-	 s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	 s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
+         int sz = VB->TexCoordPtr[i]->size;
+
+         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
+         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
 
-	 EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_SZ(sz), 0, sz * 4 );
+         EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, EMIT_SZ(sz), 0, sz * 4);
       }
       else if (i == p->wpos_tex) {
-	
-	 /* If WPOS is required, duplicate the XYZ position data in an
-	  * unused texture coordinate:
-	  */
-	 s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	 s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(3));
-
-	 intel->wpos_offset = offset;
-	 intel->wpos_size = 3 * sizeof(GLuint);
-
-	 EMIT_PAD( intel->wpos_size );
-      }   
+
+         /* If WPOS is required, duplicate the XYZ position data in an
+          * unused texture coordinate:
+          */
+         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
+         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(3));
+
+         intel->wpos_offset = offset;
+         intel->wpos_size = 3 * sizeof(GLuint);
+
+         EMIT_PAD(intel->wpos_size);
+      }
    }
 
    if (s2 != i915->state.Ctx[I915_CTXREG_LIS2] ||
        s4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-    
-      I915_STATECHANGE( i915, I915_UPLOAD_CTX );
+      int k;
+
+      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
       /* Must do this *after* statechange, so as not to affect
        * buffered vertices reliant on the old state:
        */
-      intel->vertex_size = _tnl_install_attrs( &intel->ctx, 
-					       intel->vertex_attrs, 
-					       intel->vertex_attr_count,
-					       intel->ViewportMatrix.m, 0 ); 
+      intel->vertex_size = _tnl_install_attrs(&intel->ctx,
+                                              intel->vertex_attrs,
+                                              intel->vertex_attr_count,
+                                              intel->ViewportMatrix.m, 0);
 
       intel->vertex_size >>= 2;
 
       i915->state.Ctx[I915_CTXREG_LIS2] = s2;
       i915->state.Ctx[I915_CTXREG_LIS4] = s4;
 
-      assert(intel->vtbl.check_vertex_size( intel, intel->vertex_size ));
+      k = intel->vtbl.check_vertex_size(intel, intel->vertex_size);
+      assert(k);
    }
 
-   if (!p->params_uptodate) 
-      track_params( p );
+   if (!p->params_uptodate)
+      track_params(p);
 
-   if (!p->on_hardware) 
-      i915_upload_program( i915, p );
+   if (!p->on_hardware)
+      i915_upload_program(i915, p);
 }
 
-void i915InitFragProgFuncs( struct dd_function_table *functions )
+void
+i915InitFragProgFuncs(struct dd_function_table *functions)
 {
    functions->BindProgram = i915BindProgram;
    functions->NewProgram = i915NewProgram;
diff --git a/src/mesa/drivers/dri/i915/i915_metaops.c b/src/mesa/drivers/dri/i915/i915_metaops.c
index 1be7ac4c485..a739bd65811 100644
--- a/src/mesa/drivers/dri/i915/i915_metaops.c
+++ b/src/mesa/drivers/dri/i915/i915_metaops.c
@@ -34,128 +34,170 @@
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
 #include "intel_ioctl.h"
+#include "intel_regions.h"
 #include "intel_rotate.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-/* A large amount of state doesn't need to be uploaded.
+/* We touch almost everything:
  */
-#define ACTIVE (I915_UPLOAD_INVARIENT |         \
-		I915_UPLOAD_PROGRAM | 		\
-		I915_UPLOAD_STIPPLE |		\
+#define ACTIVE (I915_UPLOAD_INVARIENT | 	\
 		I915_UPLOAD_CTX |		\
 		I915_UPLOAD_BUFFERS |		\
-		I915_UPLOAD_TEX(0))		
+		I915_UPLOAD_STIPPLE |		\
+                I915_UPLOAD_PROGRAM | 		\
+                I915_UPLOAD_FOG | 		\
+		I915_UPLOAD_TEX(0))
 
-#define SET_STATE( i915, STATE )			\
+#define SET_STATE( i915, STATE )		\
 do {						\
    i915->current->emitted &= ~ACTIVE;		\
-   i915->current = &i915->STATE;			\
+   i915->current = &i915->STATE;		\
    i915->current->emitted &= ~ACTIVE;		\
 } while (0)
 
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void set_initial_state( i915ContextPtr i915 )
-{
-   memcpy(&i915->meta, &i915->initial, sizeof(i915->meta) );
-   i915->meta.active = ACTIVE;
-   i915->meta.emitted = 0;
-}
 
-
-static void set_no_depth_stencil_write( i915ContextPtr i915 )
+static void
+meta_no_stencil_write(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
     */
-   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE | 
-				       S5_STENCIL_WRITE_ENABLE);
+   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
+                                         S5_STENCIL_WRITE_ENABLE);
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+static void
+meta_no_depth_write(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
    /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
     */
    i915->meta.Ctx[I915_CTXREG_LIS6] &= ~(S6_DEPTH_TEST_ENABLE |
-				       S6_DEPTH_WRITE_ENABLE);
+                                         S6_DEPTH_WRITE_ENABLE);
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
+static void
+meta_depth_replace(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
+   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_TRUE )
+    * ctx->Driver.DepthMask( ctx, GL_TRUE )
+    */
+   i915->meta.Ctx[I915_CTXREG_LIS6] |= (S6_DEPTH_TEST_ENABLE |
+                                        S6_DEPTH_WRITE_ENABLE);
+
+   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
+    */
+   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
+   i915->meta.Ctx[I915_CTXREG_LIS6] |=
+      COMPAREFUNC_ALWAYS << S6_DEPTH_TEST_FUNC_SHIFT;
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+
 /* Set stencil unit to replace always with the reference value.
  */
-static void set_stencil_replace( i915ContextPtr i915,
-				 GLuint s_mask,
-				 GLuint s_clear)
+static void
+meta_stencil_replace(struct intel_context *intel,
+                     GLuint s_mask, GLuint s_clear)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    GLuint op = STENCILOP_REPLACE;
    GLuint func = COMPAREFUNC_ALWAYS;
 
    /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
     */
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE | 
-				      S5_STENCIL_WRITE_ENABLE);
-
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~(S6_DEPTH_TEST_ENABLE |
-				       S6_DEPTH_WRITE_ENABLE);
-
+   i915->meta.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
+                                        S5_STENCIL_WRITE_ENABLE);
 
    /* ctx->Driver.StencilMask( ctx, s_mask )
     */
    i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
 
    i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(s_mask));
-
+                                          STENCIL_WRITE_MASK(s_mask));
 
    /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
     */
    i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-				       S5_STENCIL_PASS_Z_FAIL_MASK |
-				       S5_STENCIL_PASS_Z_PASS_MASK);
+                                         S5_STENCIL_PASS_Z_FAIL_MASK |
+                                         S5_STENCIL_PASS_Z_PASS_MASK);
 
    i915->meta.Ctx[I915_CTXREG_LIS5] |= ((op << S5_STENCIL_FAIL_SHIFT) |
-				      (op << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-				      (op << S5_STENCIL_PASS_Z_PASS_SHIFT));
+                                        (op << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
+                                        (op << S5_STENCIL_PASS_Z_PASS_SHIFT));
 
 
    /* ctx->Driver.StencilFunc( ctx, GL_ALWAYS, s_ref, ~0 )
     */
    i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					STENCIL_TEST_MASK(0xff));
+                                          STENCIL_TEST_MASK(0xff));
 
    i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-				       S5_STENCIL_TEST_FUNC_MASK);
-					
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((s_clear << S5_STENCIL_REF_SHIFT) |  
-				      (func << S5_STENCIL_TEST_FUNC_SHIFT)); 
+                                         S5_STENCIL_TEST_FUNC_MASK);
+
+   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((s_clear << S5_STENCIL_REF_SHIFT) |
+                                        (func << S5_STENCIL_TEST_FUNC_SHIFT));
 
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
 
-static void set_color_mask( i915ContextPtr i915, GLboolean state )
+static void
+meta_color_mask(struct intel_context *intel, GLboolean state)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    const GLuint mask = (S5_WRITEDISABLE_RED |
-			S5_WRITEDISABLE_GREEN |
-			S5_WRITEDISABLE_BLUE |
-			S5_WRITEDISABLE_ALPHA);
+                        S5_WRITEDISABLE_GREEN |
+                        S5_WRITEDISABLE_BLUE | S5_WRITEDISABLE_ALPHA);
 
    /* Copy colormask state from "regular" hw context.
     */
    if (state) {
       i915->meta.Ctx[I915_CTXREG_LIS5] &= ~mask;
-      i915->meta.Ctx[I915_CTXREG_LIS5] |= 
-	 (i915->state.Ctx[I915_CTXREG_LIS5] & mask);
+      i915->meta.Ctx[I915_CTXREG_LIS5] |=
+         (i915->state.Ctx[I915_CTXREG_LIS5] & mask);
    }
-   else 
+   else
       i915->meta.Ctx[I915_CTXREG_LIS5] |= mask;
-      
+
+   i915->meta.emitted &= ~I915_UPLOAD_CTX;
+}
+
+
+
+static void
+meta_import_pixel_state(struct intel_context *intel)
+{
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   memcpy(i915->meta.Fog, i915->state.Fog, I915_FOG_SETUP_SIZE * 4);
+
+   i915->meta.Ctx[I915_CTXREG_LIS5] = i915->state.Ctx[I915_CTXREG_LIS5];
+   i915->meta.Ctx[I915_CTXREG_LIS6] = i915->state.Ctx[I915_CTXREG_LIS6];
+   i915->meta.Ctx[I915_CTXREG_STATE4] = i915->state.Ctx[I915_CTXREG_STATE4];
+   i915->meta.Ctx[I915_CTXREG_BLENDCOLOR1] =
+      i915->state.Ctx[I915_CTXREG_BLENDCOLOR1];
+   i915->meta.Ctx[I915_CTXREG_IAB] = i915->state.Ctx[I915_CTXREG_IAB];
+
+   i915->meta.Buffer[I915_DESTREG_SENABLE] =
+      i915->state.Buffer[I915_DESTREG_SENABLE];
+   i915->meta.Buffer[I915_DESTREG_SR1] = i915->state.Buffer[I915_DESTREG_SR1];
+   i915->meta.Buffer[I915_DESTREG_SR2] = i915->state.Buffer[I915_DESTREG_SR2];
+
+   i915->meta.emitted &= ~I915_UPLOAD_FOG;
+   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
 }
 
@@ -212,69 +254,64 @@ static void set_color_mask( i915ContextPtr i915, GLboolean state )
 
 
 
-static void set_no_texture( i915ContextPtr i915 )
+static void
+meta_no_texture(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    static const GLuint prog[] = {
       _3DSTATE_PIXEL_SHADER_PROGRAM,
 
       /* Declare incoming diffuse color:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_T_DIFFUSE ) |
-       D0_CHANNEL_ALL),
+      (D0_DCL | D0_DECL_REG(REG_T_DIFFUSE) | D0_CHANNEL_ALL),
       D1_MBZ,
       D2_MBZ,
 
       /* output-color = mov(t_diffuse)
        */
       (A0_MOV |
-       A0_DEST_REG( REG_OC ) |
-       A0_DEST_CHANNEL_ALL |
-       A0_SRC0_REG( REG_T_DIFFUSE )),
+       A0_DEST_REG(REG_OC) |
+       A0_DEST_CHANNEL_ALL | A0_SRC0_REG(REG_T_DIFFUSE)),
       (A1_SRC0_XYZW),
       0,
    };
 
-   
-   memcpy( i915->meta.Program, prog, sizeof(prog) );
+
+   memcpy(i915->meta.Program, prog, sizeof(prog));
    i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
    i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
    i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
 }
 
-
-static void enable_texture_blend_replace( i915ContextPtr i915 )
+static void
+meta_texture_blend_replace(struct intel_context *intel)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
    static const GLuint prog[] = {
       _3DSTATE_PIXEL_SHADER_PROGRAM,
 
       /* Declare the sampler:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_S(0) ) |
-       D0_SAMPLE_TYPE_2D |
-       D0_CHANNEL_NONE),
+      (D0_DCL | D0_DECL_REG(REG_S(0)) | D0_SAMPLE_TYPE_2D | D0_CHANNEL_NONE),
       D1_MBZ,
       D2_MBZ,
 
       /* Declare the interpolated texture coordinate:
        */
-      (D0_DCL |
-       D0_DECL_REG( REG_T_TEX(0) ) |
-       D0_CHANNEL_ALL),
+      (D0_DCL | D0_DECL_REG(REG_T_TEX(0)) | D0_CHANNEL_ALL),
       D1_MBZ,
       D2_MBZ,
 
       /* output-color = texld(sample0, texcoord0) 
        */
-      (T0_TEXLD | 
-       T0_DEST_REG( REG_OC ) |
-       T0_SAMPLER( 0 )),
+      (T0_TEXLD | T0_DEST_REG(REG_OC) | T0_SAMPLER(0)),
       T1_ADDRESS_REG(REG_TYPE_T, 0),
       T2_MBZ
    };
 
-   memcpy( i915->meta.Program, prog, sizeof(prog) );
+   memcpy(i915->meta.Program, prog, sizeof(prog));
    i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
    i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
    i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
@@ -287,425 +324,186 @@ static void enable_texture_blend_replace( i915ContextPtr i915 )
 /* Set up an arbitary piece of memory as a rectangular texture
  * (including the front or back buffer).
  */
-static void set_tex_rect_source( i915ContextPtr i915,
-				 GLuint offset,
-				 GLuint width, 
-				 GLuint height,
-				 GLuint pitch, /* in bytes! */
-				 GLuint textureFormat )
+static GLboolean
+meta_tex_rect_source(struct intel_context *intel,
+                     dri_bo *buffer,
+                     GLuint offset,
+                     GLuint pitch, GLuint height, GLenum format, GLenum type)
 {
+   struct i915_context *i915 = i915_context(&intel->ctx);
    GLuint unit = 0;
    GLint numLevels = 1;
    GLuint *state = i915->meta.Tex[0];
+   GLuint textureFormat;
+   GLuint cpp;
 
-#if 0
-   printf("TexRect source offset 0x%x  pitch %d\n", offset, pitch);
-#endif
+   /* A full implementation of this would do the upload through
+    * glTexImage2d, and get all the conversion operations at that
+    * point.  We are restricted, but still at least have access to the
+    * fragment program swizzle.
+    */
+   switch (format) {
+   case GL_BGRA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGBA:
+      switch (type) {
+      case GL_UNSIGNED_INT_8_8_8_8_REV:
+      case GL_UNSIGNED_BYTE:
+         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
+         cpp = 4;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_BGR:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5_REV:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
+   case GL_RGB:
+      switch (type) {
+      case GL_UNSIGNED_SHORT_5_6_5:
+         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
+         cpp = 2;
+         break;
+      default:
+         return GL_FALSE;
+      }
+      break;
 
-/*    fprintf(stderr, "%s: offset: %x w: %d h: %d pitch %d format %x\n", */
-/* 	   __FUNCTION__, offset, width, height, pitch, textureFormat ); */
+   default:
+      return GL_FALSE;
+   }
+
+
+   if ((pitch * cpp) & 3) {
+      _mesa_printf("%s: texture is not dword pitch\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+/*    intel_region_release(&i915->meta.tex_region[0]); */
+/*    intel_region_reference(&i915->meta.tex_region[0], region); */
+   i915->meta.tex_buffer[0] = buffer;
+   i915->meta.tex_offset[0] = offset;
 
-   state[I915_TEXREG_MS2] = offset;
    state[I915_TEXREG_MS3] = (((height - 1) << MS3_HEIGHT_SHIFT) |
-			    ((width - 1) << MS3_WIDTH_SHIFT) |
-			    textureFormat |
-			    MS3_USE_FENCE_REGS);
+                             ((pitch - 1) << MS3_WIDTH_SHIFT) |
+                             textureFormat | MS3_USE_FENCE_REGS);
 
-   state[I915_TEXREG_MS4] = ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | 
-			    ((((numLevels-1) * 4)) << MS4_MAX_LOD_SHIFT));
+   state[I915_TEXREG_MS4] = (((((pitch * cpp) / 4) - 1) << MS4_PITCH_SHIFT) |
+                             MS4_CUBE_FACE_ENA_MASK |
+                             ((((numLevels - 1) * 4)) << MS4_MAX_LOD_SHIFT));
 
    state[I915_TEXREG_SS2] = ((FILTER_NEAREST << SS2_MIN_FILTER_SHIFT) |
-			    (MIPFILTER_NONE << SS2_MIP_FILTER_SHIFT) |
-			    (FILTER_NEAREST << SS2_MAG_FILTER_SHIFT));
+                             (MIPFILTER_NONE << SS2_MIP_FILTER_SHIFT) |
+                             (FILTER_NEAREST << SS2_MAG_FILTER_SHIFT));
+
    state[I915_TEXREG_SS3] = ((TEXCOORDMODE_WRAP << SS3_TCX_ADDR_MODE_SHIFT) |
-			    (TEXCOORDMODE_WRAP << SS3_TCY_ADDR_MODE_SHIFT) |
-			    (TEXCOORDMODE_WRAP << SS3_TCZ_ADDR_MODE_SHIFT) |
-			    (unit<<SS3_TEXTUREMAP_INDEX_SHIFT));
+                             (TEXCOORDMODE_WRAP << SS3_TCY_ADDR_MODE_SHIFT) |
+                             (TEXCOORDMODE_WRAP << SS3_TCZ_ADDR_MODE_SHIFT) |
+                             (unit << SS3_TEXTUREMAP_INDEX_SHIFT));
 
    state[I915_TEXREG_SS4] = 0;
 
    i915->meta.emitted &= ~I915_UPLOAD_TEX(0);
+   return GL_TRUE;
 }
 
 
-/* Select between front and back draw buffers.
+/**
+ * Set the color and depth drawing region for meta ops.
  */
-static void set_draw_region( i915ContextPtr i915, const intelRegion *region )
+static void
+meta_draw_region(struct intel_context *intel,
+                 struct intel_region *color_region,
+                 struct intel_region *depth_region)
 {
-#if 0
-   printf("Rotate into region: offset 0x%x  pitch %d\n",
-          region->offset, region->pitch);
-#endif
-   i915->meta.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->meta.Buffer[I915_DESTREG_CBUFADDR2] = region->offset;
-   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   i915_state_draw_region(intel, &i915->meta, color_region, depth_region);
 }
 
 
-#if 0
-/* Setup an arbitary draw format, useful for targeting texture or agp
- * memory.
- */
-static void set_draw_format( i915ContextPtr i915,
-			     GLuint format,
-			     GLuint depth_format)
+static void
+set_vertex_format(struct intel_context *intel)
 {
-   i915->meta.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					DSTORG_VERT_BIAS(0x8) | /* .5 */
-					format |
-					LOD_PRECLAMP_OGL |
-					TEX_DEFAULT_COLOR_OGL |
-					depth_format);
-
-   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
-/*    fprintf(stderr, "%s: DV1: %x\n",  */
-/* 	   __FUNCTION__, i915->meta.Buffer[I915_DESTREG_DV1]); */
-}
-#endif
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
-static void set_vertex_format( i915ContextPtr i915 )
-{
-   i915->meta.Ctx[I915_CTXREG_LIS2] = 
+   i915->meta.Ctx[I915_CTXREG_LIS2] =
       (S2_TEXCOORD_FMT(0, TEXCOORDFMT_2D) |
-       S2_TEXCOORD_FMT(1, TEXCOORDFMT_NOT_PRESENT) | 
+       S2_TEXCOORD_FMT(1, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(2, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(3, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(4, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(5, TEXCOORDFMT_NOT_PRESENT) | 
+       S2_TEXCOORD_FMT(5, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(6, TEXCOORDFMT_NOT_PRESENT) |
        S2_TEXCOORD_FMT(7, TEXCOORDFMT_NOT_PRESENT));
 
    i915->meta.Ctx[I915_CTXREG_LIS4] &= ~S4_VFMT_MASK;
 
-   i915->meta.Ctx[I915_CTXREG_LIS4] |= 
-      (S4_VFMT_COLOR |
-       S4_VFMT_SPEC_FOG |
-       S4_VFMT_XYZW);
+   i915->meta.Ctx[I915_CTXREG_LIS4] |= (S4_VFMT_COLOR | S4_VFMT_XYZ);
 
    i915->meta.emitted &= ~I915_UPLOAD_CTX;
-
 }
 
 
-static void draw_quad(i915ContextPtr i915, 
-		      GLfloat x0, GLfloat x1,
-		      GLfloat y0, GLfloat y1, 
-		      GLubyte red, GLubyte green,
-		      GLubyte blue, GLubyte alpha,
-		      GLfloat s0, GLfloat s1,
-		      GLfloat t0, GLfloat t1 )
-{
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i915->intel, 
-						PRIM3D_TRIFAN, 
-						4 * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i;
-
-   if (0)
-      fprintf(stderr, "%s: %f,%f-%f,%f 0x%x%x%x%x %f,%f-%f,%f\n",
-	      __FUNCTION__,
-	      x0,y0,x1,y1,red,green,blue,alpha,s0,t0,s1,t1);
-
-
-   /* initial vertex, left bottom */
-   tmp.v.x = x0;
-   tmp.v.y = y0;
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-   tmp.v.u0 = s0;
-   tmp.v.v0 = t0;
-
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right bottom */
-   vb += vertex_size;
-   tmp.v.x = x1;
-   tmp.v.u0 = s1;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* right top */
-   vb += vertex_size;
-   tmp.v.y = y1;
-   tmp.v.v0 = t1;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-
-   /* left top */
-   vb += vertex_size;
-   tmp.v.x = x0;
-   tmp.v.u0 = s0;
-   for (i = 0 ; i < vertex_size ; i++)
-      vb[i] = tmp.ui[i];
-}
-
 
-static void draw_poly(i915ContextPtr i915, 
-		      GLubyte red, GLubyte green, GLubyte blue, GLubyte alpha,
-                      GLuint numVerts,
-                      /*const*/ GLfloat verts[][2],
-                      /*const*/ GLfloat texcoords[][2])
+/* Operations where the 3D engine is decoupled temporarily from the
+ * current GL state and used for other purposes than simply rendering
+ * incoming triangles.
+ */
+static void
+install_meta_state(struct intel_context *intel)
 {
-   GLuint vertex_size = 8;
-   GLuint *vb = intelEmitInlinePrimitiveLocked( &i915->intel, 
-						PRIM3D_TRIFAN, 
-						numVerts * vertex_size,
-						vertex_size );
-   intelVertex tmp;
-   int i, k;
-
-   /* initial constant vertex fields */
-   tmp.v.z = 1.0;
-   tmp.v.w = 1.0; 
-   tmp.v.color.red = red;
-   tmp.v.color.green = green;
-   tmp.v.color.blue = blue;
-   tmp.v.color.alpha = alpha;
-   tmp.v.specular.red = 0;
-   tmp.v.specular.green = 0;
-   tmp.v.specular.blue = 0;
-   tmp.v.specular.alpha = 0;
-
-   for (k = 0; k < numVerts; k++) {
-      tmp.v.x = verts[k][0];
-      tmp.v.y = verts[k][1];
-      tmp.v.u0 = texcoords[k][0];
-      tmp.v.v0 = texcoords[k][1];
-
-      for (i = 0 ; i < vertex_size ; i++)
-         vb[i] = tmp.ui[i];
-
-      vb += vertex_size;
-   }
-}
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   memcpy(&i915->meta, &i915->initial, sizeof(i915->meta));
+   i915->meta.active = ACTIVE;
+   i915->meta.emitted = 0;
 
+   SET_STATE(i915, meta);
+   set_vertex_format(intel);
+   meta_no_texture(intel);
+}
 
-void 
-i915ClearWithTris(intelContextPtr intel, GLbitfield buffers,
-		  GLboolean allFoo,
-		  GLint cxFoo, GLint cyFoo, GLint cwFoo, GLint chFoo)
+static void
+leave_meta_state(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT( intel );
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   intelScreenPrivate *screen = intel->intelScreen;
-   int x0, y0, x1, y1;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-
-   SET_STATE( i915, meta ); 
-   set_initial_state( i915 ); 
-   set_no_texture( i915 ); 
-   set_vertex_format( i915 ); 
-
-   LOCK_HARDWARE(intel);
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   if (!all) {
-      x0 = cx;
-      y0 = cy;
-      x1 = x0 + cw;
-      y1 = y0 + ch;
-   } else {
-      x0 = 0;
-      y0 = 0;
-      x1 = x0 + dPriv->w;
-      y1 = y0 + dPriv->h;
-   }
-
-   /* Don't do any clipping to screen - these are window coordinates.
-    * The active cliprects will be applied as for any other geometry.
-    */
-
-   if (buffers & BUFFER_BIT_FRONT_LEFT) { 
-      set_no_depth_stencil_write( i915 );
-      set_color_mask( i915, GL_TRUE );
-      set_draw_region( i915, &screen->front );
-
-      draw_quad(i915, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green, 
- 		intel->clear_blue, intel->clear_alpha, 
-		0, 0, 0, 0);
-   }
-
-   if (buffers & BUFFER_BIT_BACK_LEFT) {
-      set_no_depth_stencil_write( i915 );
-      set_color_mask( i915, GL_TRUE );
-      set_draw_region( i915, &screen->back );
-
-      draw_quad(i915, x0, x1, y0, y1,
-		intel->clear_red, intel->clear_green,
-		intel->clear_blue, intel->clear_alpha,
-		0, 0, 0, 0);
-   }
-
-   if (buffers & BUFFER_BIT_STENCIL) {
-      set_stencil_replace( i915, 
-			   intel->ctx.Stencil.WriteMask[0], 
-			   intel->ctx.Stencil.Clear);
-      
-      set_color_mask( i915, GL_FALSE );
-      set_draw_region( i915, &screen->front ); /* could be either? */
-
-      draw_quad( i915, x0, x1, y0, y1, 0, 0, 0, 0, 0, 0, 0, 0 );
-   }
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i915, state );
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   intel_region_release(&i915->meta.draw_region);
+   intel_region_release(&i915->meta.depth_region);
+/*    intel_region_release(&i915->meta.tex_region[0]); */
+   SET_STATE(i915, state);
 }
 
 
-/**
- * Copy the window contents named by dPriv to the rotated (or reflected)
- * color buffer.
- * srcBuf is BUFFER_BIT_FRONT_LEFT or BUFFER_BIT_BACK_LEFT to indicate the source.
- */
+
 void
-i915RotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                 GLuint srcBuf)
+i915InitMetaFuncs(struct i915_context *i915)
 {
-   i915ContextPtr i915 = I915_CONTEXT( intel );
-   intelScreenPrivate *screen = intel->intelScreen;
-   const GLuint cpp = screen->cpp;
-   drm_clip_rect_t fullRect;
-   GLuint textureFormat, srcOffset, srcPitch;
-   const drm_clip_rect_t *clipRects;
-   int numClipRects;
-   int i;
-
-   int xOrig, yOrig;
-   int origNumClipRects;
-   drm_clip_rect_t *origRects;
-
-   /*
-    * set up hardware state
-    */
-   intelFlush( &intel->ctx );
-
-   SET_STATE( i915, meta ); 
-   set_initial_state( i915 ); 
-   set_no_texture( i915 ); 
-   set_vertex_format( i915 ); 
-   set_no_depth_stencil_write( i915 );
-   set_color_mask( i915, GL_TRUE );
-
-   LOCK_HARDWARE(intel);
-
-   /* save current drawing origin and cliprects (restored at end) */
-   xOrig = intel->drawX;
-   yOrig = intel->drawY;
-   origNumClipRects = intel->numClipRects;
-   origRects = intel->pClipRects;
-
-   if (!intel->numClipRects)
-      goto done;
-
-   /*
-    * set drawing origin, cliprects for full-screen access to rotated screen
-    */
-   fullRect.x1 = 0;
-   fullRect.y1 = 0;
-   fullRect.x2 = screen->rotatedWidth;
-   fullRect.y2 = screen->rotatedHeight;
-   intel->drawX = 0;
-   intel->drawY = 0;
-   intel->numClipRects = 1;
-   intel->pClipRects = &fullRect;
-
-   set_draw_region( i915, &screen->rotated );
-
-   if (cpp == 4)
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   else
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-
-   if (srcBuf == BUFFER_BIT_FRONT_LEFT) {
-      srcPitch = screen->front.pitch;   /* in bytes */
-      srcOffset = screen->front.offset; /* bytes */
-      clipRects = dPriv->pClipRects;
-      numClipRects = dPriv->numClipRects;
-   }
-   else {
-      srcPitch = screen->back.pitch;   /* in bytes */
-      srcOffset = screen->back.offset; /* bytes */
-      clipRects = dPriv->pBackClipRects;
-      numClipRects = dPriv->numBackClipRects;
-   }
-
-   /* set the whole screen up as a texture to avoid alignment issues */
-   set_tex_rect_source(i915,
-                       srcOffset,
-                       screen->width,
-		       screen->height,
-                       srcPitch,
-                       textureFormat);
-
-   enable_texture_blend_replace(i915);
-
-   /*
-    * loop over the source window's cliprects
-    */
-   for (i = 0; i < numClipRects; i++) {
-      int srcX0 = clipRects[i].x1;
-      int srcY0 = clipRects[i].y1;
-      int srcX1 = clipRects[i].x2;
-      int srcY1 = clipRects[i].y2;
-      GLfloat verts[4][2], tex[4][2];
-      int j;
-
-      /* build vertices for four corners of clip rect */
-      verts[0][0] = srcX0;  verts[0][1] = srcY0;
-      verts[1][0] = srcX1;  verts[1][1] = srcY0;
-      verts[2][0] = srcX1;  verts[2][1] = srcY1;
-      verts[3][0] = srcX0;  verts[3][1] = srcY1;
-
-      /* .. and texcoords */
-      tex[0][0] = srcX0;  tex[0][1] = srcY0;
-      tex[1][0] = srcX1;  tex[1][1] = srcY0;
-      tex[2][0] = srcX1;  tex[2][1] = srcY1;
-      tex[3][0] = srcX0;  tex[3][1] = srcY1;
-
-      /* transform coords to rotated screen coords */
-      for (j = 0; j < 4; j++) {
-         matrix23TransformCoordf(&screen->rotMatrix,
-                                 &verts[j][0], &verts[j][1]);
-      }
-
-      /* draw polygon to map source image to dest region */
-      draw_poly(i915, 255, 255, 255, 255, 4, verts, tex);
-
-   } /* cliprect loop */
-
-   intelFlushBatchLocked( intel, GL_FALSE, GL_FALSE, GL_FALSE );
-
- done:
-   /* restore original drawing origin and cliprects */
-   intel->drawX = xOrig;
-   intel->drawY = yOrig;
-   intel->numClipRects = origNumClipRects;
-   intel->pClipRects = origRects;
-
-   UNLOCK_HARDWARE(intel);
-
-   SET_STATE( i915, state );
+   i915->intel.vtbl.install_meta_state = install_meta_state;
+   i915->intel.vtbl.leave_meta_state = leave_meta_state;
+   i915->intel.vtbl.meta_no_depth_write = meta_no_depth_write;
+   i915->intel.vtbl.meta_no_stencil_write = meta_no_stencil_write;
+   i915->intel.vtbl.meta_stencil_replace = meta_stencil_replace;
+   i915->intel.vtbl.meta_depth_replace = meta_depth_replace;
+   i915->intel.vtbl.meta_color_mask = meta_color_mask;
+   i915->intel.vtbl.meta_no_texture = meta_no_texture;
+   i915->intel.vtbl.meta_texture_blend_replace = meta_texture_blend_replace;
+   i915->intel.vtbl.meta_tex_rect_source = meta_tex_rect_source;
+   i915->intel.vtbl.meta_draw_region = meta_draw_region;
+   i915->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
 }
-
diff --git a/src/mesa/drivers/dri/i915/i915_program.c b/src/mesa/drivers/dri/i915/i915_program.c
index 68491124449..c6c64340232 100644
--- a/src/mesa/drivers/dri/i915/i915_program.c
+++ b/src/mesa/drivers/dri/i915/i915_program.c
@@ -72,58 +72,62 @@
 
 #define I915_CONSTFLAG_PARAM 0x1f
 
-GLuint i915_get_temp( struct i915_fragment_program *p )
+GLuint
+i915_get_temp(struct i915_fragment_program *p)
 {
-   int bit = ffs( ~p->temp_flag );
+   int bit = ffs(~p->temp_flag);
    if (!bit) {
       fprintf(stderr, "%s: out of temporaries\n", __FILE__);
       exit(1);
    }
 
-   p->temp_flag |= 1<<(bit-1);
-   return UREG(REG_TYPE_R, (bit-1));
+   p->temp_flag |= 1 << (bit - 1);
+   return UREG(REG_TYPE_R, (bit - 1));
 }
 
 
-GLuint i915_get_utemp( struct i915_fragment_program *p )
+GLuint
+i915_get_utemp(struct i915_fragment_program * p)
 {
-   int bit = ffs( ~p->utemp_flag );
+   int bit = ffs(~p->utemp_flag);
    if (!bit) {
       fprintf(stderr, "%s: out of temporaries\n", __FILE__);
       exit(1);
    }
 
-   p->utemp_flag |= 1<<(bit-1);
-   return UREG(REG_TYPE_U, (bit-1));
+   p->utemp_flag |= 1 << (bit - 1);
+   return UREG(REG_TYPE_U, (bit - 1));
 }
 
-void i915_release_utemps( struct i915_fragment_program *p )
+void
+i915_release_utemps(struct i915_fragment_program *p)
 {
    p->utemp_flag = ~0x7;
 }
 
 
-GLuint i915_emit_decl( struct i915_fragment_program *p,
-		      GLuint type, GLuint nr, GLuint d0_flags )
+GLuint
+i915_emit_decl(struct i915_fragment_program *p,
+               GLuint type, GLuint nr, GLuint d0_flags)
 {
    GLuint reg = UREG(type, nr);
 
    if (type == REG_TYPE_T) {
-      if (p->decl_t & (1<<nr))
-	 return reg;
+      if (p->decl_t & (1 << nr))
+         return reg;
 
-      p->decl_t |= (1<<nr);
+      p->decl_t |= (1 << nr);
    }
    else if (type == REG_TYPE_S) {
-      if (p->decl_s & (1<<nr))
-	 return reg;
+      if (p->decl_s & (1 << nr))
+         return reg;
 
-      p->decl_s |= (1<<nr);
+      p->decl_s |= (1 << nr);
    }
-   else 
+   else
       return reg;
 
-   *(p->decl++) = (D0_DCL | D0_DEST( reg ) | d0_flags);
+   *(p->decl++) = (D0_DCL | D0_DEST(reg) | d0_flags);
    *(p->decl++) = D1_MBZ;
    *(p->decl++) = D2_MBZ;
 
@@ -131,24 +135,26 @@ GLuint i915_emit_decl( struct i915_fragment_program *p,
    return reg;
 }
 
-GLuint i915_emit_arith( struct i915_fragment_program *p,
-		       GLuint op,
-		       GLuint dest,
-		       GLuint mask,
-		       GLuint saturate,
-		       GLuint src0,
-		       GLuint src1,
-		       GLuint src2 )
+GLuint
+i915_emit_arith(struct i915_fragment_program * p,
+                GLuint op,
+                GLuint dest,
+                GLuint mask,
+                GLuint saturate, GLuint src0, GLuint src1, GLuint src2)
 {
    GLuint c[3];
    GLuint nr_const = 0;
 
    assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
-   assert(dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
+   dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest));
+   assert(dest);
 
-   if (GET_UREG_TYPE(src0) == REG_TYPE_CONST) c[nr_const++] = 0;
-   if (GET_UREG_TYPE(src1) == REG_TYPE_CONST) c[nr_const++] = 1;
-   if (GET_UREG_TYPE(src2) == REG_TYPE_CONST) c[nr_const++] = 2;
+   if (GET_UREG_TYPE(src0) == REG_TYPE_CONST)
+      c[nr_const++] = 0;
+   if (GET_UREG_TYPE(src1) == REG_TYPE_CONST)
+      c[nr_const++] = 1;
+   if (GET_UREG_TYPE(src2) == REG_TYPE_CONST)
+      c[nr_const++] = 2;
 
    /* Recursively call this function to MOV additional const values
     * into temporary registers.  Use utemp registers for this -
@@ -164,31 +170,25 @@ GLuint i915_emit_arith( struct i915_fragment_program *p,
       old_utemp_flag = p->utemp_flag;
 
       first = GET_UREG_NR(s[c[0]]);
-      for (i = 1 ; i < nr_const ; i++) {
-	 if (GET_UREG_NR(s[c[i]]) != first) {
-	    GLuint tmp = i915_get_utemp(p);
-
-	    i915_emit_arith( p, A0_MOV, tmp, A0_DEST_CHANNEL_ALL, 0,
-			    s[c[i]], 0, 0 );
-	    s[c[i]] = tmp;
-	 }
+      for (i = 1; i < nr_const; i++) {
+         if (GET_UREG_NR(s[c[i]]) != first) {
+            GLuint tmp = i915_get_utemp(p);
+
+            i915_emit_arith(p, A0_MOV, tmp, A0_DEST_CHANNEL_ALL, 0,
+                            s[c[i]], 0, 0);
+            s[c[i]] = tmp;
+         }
       }
 
       src0 = s[0];
       src1 = s[1];
       src2 = s[2];
-      p->utemp_flag = old_utemp_flag; /* restore */
+      p->utemp_flag = old_utemp_flag;   /* restore */
    }
 
-   *(p->csr++) = (op | 
-		  A0_DEST( dest ) |
-		  mask | 
-		  saturate |
-		  A0_SRC0( src0 ));
-   *(p->csr++) = (A1_SRC0( src0 ) |
-		  A1_SRC1( src1 ));
-   *(p->csr++) = (A2_SRC1( src1 ) |
-		  A2_SRC2( src2 ));
+   *(p->csr++) = (op | A0_DEST(dest) | mask | saturate | A0_SRC0(src0));
+   *(p->csr++) = (A1_SRC0(src0) | A1_SRC1(src1));
+   *(p->csr++) = (A2_SRC1(src1) | A2_SRC2(src2));
 
    p->nr_alu_insn++;
    return dest;
@@ -239,24 +239,28 @@ GLuint i915_emit_texld( struct i915_fragment_program *p,
 }
 
 
-GLuint i915_emit_const1f( struct i915_fragment_program *p, GLfloat c0 )
+GLuint
+i915_emit_const1f(struct i915_fragment_program * p, GLfloat c0)
 {
    GLint reg, idx;
 
-   if (c0 == 0.0) return swizzle(UREG(REG_TYPE_R, 0), ZERO, ZERO, ZERO, ZERO);
-   if (c0 == 1.0) return swizzle(UREG(REG_TYPE_R, 0), ONE,  ONE,  ONE,  ONE );
+   if (c0 == 0.0)
+      return swizzle(UREG(REG_TYPE_R, 0), ZERO, ZERO, ZERO, ZERO);
+   if (c0 == 1.0)
+      return swizzle(UREG(REG_TYPE_R, 0), ONE, ONE, ONE, ONE);
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-	 continue;
+         continue;
       for (idx = 0; idx < 4; idx++) {
-	 if (!(p->constant_flags[reg] & (1<<idx)) ||
-	     p->constant[reg][idx] == c0) {
-	    p->constant[reg][idx] = c0;
-	    p->constant_flags[reg] |= 1<<idx;
-	    if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	    return swizzle(UREG(REG_TYPE_CONST, reg),idx,ZERO,ZERO,ONE);
-	 }
+         if (!(p->constant_flags[reg] & (1 << idx)) ||
+             p->constant[reg][idx] == c0) {
+            p->constant[reg][idx] = c0;
+            p->constant_flags[reg] |= 1 << idx;
+            if (reg + 1 > p->nr_constants)
+               p->nr_constants = reg + 1;
+            return swizzle(UREG(REG_TYPE_CONST, reg), idx, ZERO, ZERO, ONE);
+         }
       }
    }
 
@@ -265,29 +269,35 @@ GLuint i915_emit_const1f( struct i915_fragment_program *p, GLfloat c0 )
    return 0;
 }
 
-GLuint i915_emit_const2f( struct i915_fragment_program *p, 
-			 GLfloat c0, GLfloat c1 )
+GLuint
+i915_emit_const2f(struct i915_fragment_program * p, GLfloat c0, GLfloat c1)
 {
    GLint reg, idx;
 
-   if (c0 == 0.0) return swizzle(i915_emit_const1f(p, c1), ZERO, X, Z, W);
-   if (c0 == 1.0) return swizzle(i915_emit_const1f(p, c1), ONE,  X, Z, W); 
+   if (c0 == 0.0)
+      return swizzle(i915_emit_const1f(p, c1), ZERO, X, Z, W);
+   if (c0 == 1.0)
+      return swizzle(i915_emit_const1f(p, c1), ONE, X, Z, W);
 
-   if (c1 == 0.0) return swizzle(i915_emit_const1f(p, c0), X, ZERO, Z, W);
-   if (c1 == 1.0) return swizzle(i915_emit_const1f(p, c0), X, ONE,  Z, W);
+   if (c1 == 0.0)
+      return swizzle(i915_emit_const1f(p, c0), X, ZERO, Z, W);
+   if (c1 == 1.0)
+      return swizzle(i915_emit_const1f(p, c0), X, ONE, Z, W);
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0xf ||
-	  p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-	 continue;
+          p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
+         continue;
       for (idx = 0; idx < 3; idx++) {
-	 if (!(p->constant_flags[reg] & (3<<idx))) {
-	    p->constant[reg][idx] = c0;
-	    p->constant[reg][idx+1] = c1;
-	    p->constant_flags[reg] |= 3<<idx;
-	    if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	    return swizzle(UREG(REG_TYPE_CONST, reg),idx,idx+1,ZERO,ONE);
-	 }
+         if (!(p->constant_flags[reg] & (3 << idx))) {
+            p->constant[reg][idx] = c0;
+            p->constant[reg][idx + 1] = c1;
+            p->constant_flags[reg] |= 3 << idx;
+            if (reg + 1 > p->nr_constants)
+               p->nr_constants = reg + 1;
+            return swizzle(UREG(REG_TYPE_CONST, reg), idx, idx + 1, ZERO,
+                           ONE);
+         }
       }
    }
 
@@ -298,27 +308,28 @@ GLuint i915_emit_const2f( struct i915_fragment_program *p,
 
 
 
-GLuint i915_emit_const4f( struct i915_fragment_program *p, 
-			 GLfloat c0, GLfloat c1, GLfloat c2, GLfloat c3 )
+GLuint
+i915_emit_const4f(struct i915_fragment_program * p,
+                  GLfloat c0, GLfloat c1, GLfloat c2, GLfloat c3)
 {
    GLint reg;
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0xf &&
-	  p->constant[reg][0] == c0 &&
-	  p->constant[reg][1] == c1 &&
-	  p->constant[reg][2] == c2 &&
-	  p->constant[reg][3] == c3) {
-	 return UREG(REG_TYPE_CONST, reg);
+          p->constant[reg][0] == c0 &&
+          p->constant[reg][1] == c1 &&
+          p->constant[reg][2] == c2 && p->constant[reg][3] == c3) {
+         return UREG(REG_TYPE_CONST, reg);
       }
       else if (p->constant_flags[reg] == 0) {
-	 p->constant[reg][0] = c0;
-	 p->constant[reg][1] = c1;
-	 p->constant[reg][2] = c2;
-	 p->constant[reg][3] = c3;
-	 p->constant_flags[reg] = 0xf;
-	 if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	 return UREG(REG_TYPE_CONST, reg);
+         p->constant[reg][0] = c0;
+         p->constant[reg][1] = c1;
+         p->constant[reg][2] = c2;
+         p->constant[reg][3] = c3;
+         p->constant_flags[reg] = 0xf;
+         if (reg + 1 > p->nr_constants)
+            p->nr_constants = reg + 1;
+         return UREG(REG_TYPE_CONST, reg);
       }
    }
 
@@ -328,34 +339,36 @@ GLuint i915_emit_const4f( struct i915_fragment_program *p,
 }
 
 
-GLuint i915_emit_const4fv( struct i915_fragment_program *p, const GLfloat *c )
+GLuint
+i915_emit_const4fv(struct i915_fragment_program * p, const GLfloat * c)
 {
-   return i915_emit_const4f( p, c[0], c[1], c[2], c[3] );
+   return i915_emit_const4f(p, c[0], c[1], c[2], c[3]);
 }
 
 
-GLuint i915_emit_param4fv( struct i915_fragment_program *p, 
-			  const GLfloat *values )
+GLuint
+i915_emit_param4fv(struct i915_fragment_program * p, const GLfloat * values)
 {
    GLint reg, i;
 
    for (i = 0; i < p->nr_params; i++) {
       if (p->param[i].values == values)
-	 return UREG(REG_TYPE_CONST, p->param[i].reg);
+         return UREG(REG_TYPE_CONST, p->param[i].reg);
    }
 
 
    for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
       if (p->constant_flags[reg] == 0) {
-	 p->constant_flags[reg] = I915_CONSTFLAG_PARAM;
-	 i = p->nr_params++;
+         p->constant_flags[reg] = I915_CONSTFLAG_PARAM;
+         i = p->nr_params++;
 
-	 p->param[i].values = values;
-	 p->param[i].reg = reg;
-	 p->params_uptodate = 0;
+         p->param[i].values = values;
+         p->param[i].reg = reg;
+         p->params_uptodate = 0;
 
-	 if (reg+1 > p->nr_constants) p->nr_constants = reg+1;
-	 return UREG(REG_TYPE_CONST, reg);
+         if (reg + 1 > p->nr_constants)
+            p->nr_constants = reg + 1;
+         return UREG(REG_TYPE_CONST, reg);
       }
    }
 
@@ -366,30 +379,31 @@ GLuint i915_emit_param4fv( struct i915_fragment_program *p,
 
 
 
-
-void i915_program_error( struct i915_fragment_program *p, const char *msg )
+void
+i915_program_error(struct i915_fragment_program *p, const char *msg)
 {
    _mesa_problem(NULL, "i915_program_error: %s", msg);
    p->error = 1;
 }
 
-void i915_init_program( i915ContextPtr i915, struct i915_fragment_program *p )
+
+void
+i915_init_program(struct i915_context *i915, struct i915_fragment_program *p)
 {
    GLcontext *ctx = &i915->intel.ctx;
-   TNLcontext *tnl = TNL_CONTEXT( ctx );
-   
+
    p->translated = 0;
    p->params_uptodate = 0;
    p->on_hardware = 0;
    p->error = 0;
 
-   p->nr_tex_indirect = 1;	/* correct? */
+   p->nr_tex_indirect = 1;      /* correct? */
    p->nr_tex_insn = 0;
    p->nr_alu_insn = 0;
    p->nr_decl_insn = 0;
 
-   p->ctx = ctx;  
-   memset( p->constant_flags, 0, sizeof(p->constant_flags) );
+   p->ctx = ctx;
+   memset(p->constant_flags, 0, sizeof(p->constant_flags));
 
    p->nr_constants = 0;
    p->csr = p->program;
@@ -402,21 +416,17 @@ void i915_init_program( i915ContextPtr i915, struct i915_fragment_program *p )
    p->depth_written = 0;
    p->nr_params = 0;
 
-   p->src_texture = UREG_BAD;
-   p->src_previous = UREG(REG_TYPE_T, T_DIFFUSE);
-   p->last_tex_stage = 0;
-   p->VB = &tnl->vb;
-
    *(p->decl++) = _3DSTATE_PIXEL_SHADER_PROGRAM;
 }
 
 
-void i915_fini_program( struct i915_fragment_program *p )
+void
+i915_fini_program(struct i915_fragment_program *p)
 {
    GLuint program_size = p->csr - p->program;
    GLuint decl_size = p->decl - p->declarations;
-   
-   if (p->nr_tex_indirect > I915_MAX_TEX_INDIRECT) 
+
+   if (p->nr_tex_indirect > I915_MAX_TEX_INDIRECT)
       i915_program_error(p, "Exceeded max nr indirect texture lookups");
 
    if (p->nr_tex_insn > I915_MAX_TEX_INSN)
@@ -446,22 +456,24 @@ void i915_fini_program( struct i915_fragment_program *p )
    p->declarations[0] |= program_size + decl_size - 2;
 }
 
-void i915_upload_program( i915ContextPtr i915, struct i915_fragment_program *p )
+void
+i915_upload_program(struct i915_context *i915,
+                    struct i915_fragment_program *p)
 {
    GLuint program_size = p->csr - p->program;
    GLuint decl_size = p->decl - p->declarations;
 
-   FALLBACK( &i915->intel, I915_FALLBACK_PROGRAM, p->error );
+   FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, p->error);
 
    /* Could just go straight to the batchbuffer from here:
     */
    if (i915->state.ProgramSize != (program_size + decl_size) ||
-       memcmp(i915->state.Program + decl_size, p->program, 
-	      program_size*sizeof(int)) != 0) {
-      I915_STATECHANGE( i915, I915_UPLOAD_PROGRAM );
-      memcpy(i915->state.Program, p->declarations, decl_size*sizeof(int));
+       memcmp(i915->state.Program + decl_size, p->program,
+              program_size * sizeof(int)) != 0) {
+      I915_STATECHANGE(i915, I915_UPLOAD_PROGRAM);
+      memcpy(i915->state.Program, p->declarations, decl_size * sizeof(int));
       memcpy(i915->state.Program + decl_size, p->program,
-	     program_size*sizeof(int));
+             program_size * sizeof(int));
       i915->state.ProgramSize = decl_size + program_size;
    }
 
@@ -470,30 +482,28 @@ void i915_upload_program( i915ContextPtr i915, struct i915_fragment_program *p )
     */
    if (p->nr_constants) {
       GLuint nr = p->nr_constants;
-      
-      I915_ACTIVESTATE( i915, I915_UPLOAD_CONSTANTS, 1 );
-      I915_STATECHANGE( i915, I915_UPLOAD_CONSTANTS );
+
+      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 1);
+      I915_STATECHANGE(i915, I915_UPLOAD_CONSTANTS);
 
       i915->state.Constant[0] = _3DSTATE_PIXEL_SHADER_CONSTANTS | ((nr) * 4);
-      i915->state.Constant[1] = (1<<(nr-1)) | ((1<<(nr-1))-1);
-      
-      memcpy(&i915->state.Constant[2], p->constant, 4*sizeof(int)*(nr));
+      i915->state.Constant[1] = (1 << (nr - 1)) | ((1 << (nr - 1)) - 1);
+
+      memcpy(&i915->state.Constant[2], p->constant, 4 * sizeof(int) * (nr));
       i915->state.ConstantSize = 2 + (nr) * 4;
 
       if (0) {
-	 GLuint i;
-	 for (i = 0; i < nr; i++) {
-	    fprintf(stderr, "const[%d]: %f %f %f %f\n", i, 
-		    p->constant[i][0],
-		    p->constant[i][1],
-		    p->constant[i][2],
-		    p->constant[i][3]);
-	 }
+         GLuint i;
+         for (i = 0; i < nr; i++) {
+            fprintf(stderr, "const[%d]: %f %f %f %f\n", i,
+                    p->constant[i][0],
+                    p->constant[i][1], p->constant[i][2], p->constant[i][3]);
+         }
       }
    }
    else {
-      I915_ACTIVESTATE( i915, I915_UPLOAD_CONSTANTS, 0 );
-   }  
+      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 0);
+   }
 
    p->on_hardware = 1;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_program.h b/src/mesa/drivers/dri/i915/i915_program.h
index 8891a177855..3c12b34f163 100644
--- a/src/mesa/drivers/dri/i915/i915_program.h
+++ b/src/mesa/drivers/dri/i915/i915_program.h
@@ -48,11 +48,11 @@
 #define UREG_CHANNEL_W_NEGATE_SHIFT   11
 #define UREG_CHANNEL_W_SHIFT          8
 #define UREG_CHANNEL_ZERO_NEGATE_MBZ  5
-#define UREG_CHANNEL_ZERO_SHIFT       4      
+#define UREG_CHANNEL_ZERO_SHIFT       4
 #define UREG_CHANNEL_ONE_NEGATE_MBZ   1
-#define UREG_CHANNEL_ONE_SHIFT        0      
+#define UREG_CHANNEL_ONE_SHIFT        0
 
-#define UREG_BAD          0xffffffff /* not a valid ureg */
+#define UREG_BAD          0xffffffff    /* not a valid ureg */
 
 #define X    SRC_X
 #define Y    SRC_Y
@@ -84,78 +84,75 @@
 
 /* One neat thing about the UREG representation:  
  */
-static __inline int swizzle( int reg, int x, int y, int z, int w )
+static INLINE int
+swizzle(int reg, int x, int y, int z, int w)
 {
    return ((reg & ~UREG_XYZW_CHANNEL_MASK) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, x ), 0 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, y ), 1 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, z ), 2 ) |
-	   CHANNEL_SRC( GET_CHANNEL_SRC( reg, w ), 3 ));
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, x), 0) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, y), 1) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, z), 2) |
+           CHANNEL_SRC(GET_CHANNEL_SRC(reg, w), 3));
 }
 
 /* Another neat thing about the UREG representation:  
  */
-static __inline int negate( int reg, int x, int y, int z, int w )
+static INLINE int
+negate(int reg, int x, int y, int z, int w)
 {
-   return reg ^ (((x&1)<<UREG_CHANNEL_X_NEGATE_SHIFT)|
-		 ((y&1)<<UREG_CHANNEL_Y_NEGATE_SHIFT)|
-		 ((z&1)<<UREG_CHANNEL_Z_NEGATE_SHIFT)|
-		 ((w&1)<<UREG_CHANNEL_W_NEGATE_SHIFT));
+   return reg ^ (((x & 1) << UREG_CHANNEL_X_NEGATE_SHIFT) |
+                 ((y & 1) << UREG_CHANNEL_Y_NEGATE_SHIFT) |
+                 ((z & 1) << UREG_CHANNEL_Z_NEGATE_SHIFT) |
+                 ((w & 1) << UREG_CHANNEL_W_NEGATE_SHIFT));
 }
 
 
-extern GLuint i915_get_temp( struct i915_fragment_program *p );
-extern GLuint i915_get_utemp( struct i915_fragment_program *p );
-extern void i915_release_utemps( struct i915_fragment_program *p );
+extern GLuint i915_get_temp(struct i915_fragment_program *p);
+extern GLuint i915_get_utemp(struct i915_fragment_program *p);
+extern void i915_release_utemps(struct i915_fragment_program *p);
 
 
-extern GLuint i915_emit_texld( struct i915_fragment_program *p,
-			      GLuint dest,
-			      GLuint destmask,
-			      GLuint sampler,
-			      GLuint coord,
-			      GLuint op );
+extern GLuint i915_emit_texld(struct i915_fragment_program *p,
+                              GLuint dest,
+                              GLuint destmask,
+                              GLuint sampler, GLuint coord, GLuint op);
 
-extern GLuint i915_emit_arith( struct i915_fragment_program *p,
-			      GLuint op,
-			      GLuint dest,
-			      GLuint mask,
-			      GLuint saturate,
-			      GLuint src0,
-			      GLuint src1,
-			      GLuint src2 );
+extern GLuint i915_emit_arith(struct i915_fragment_program *p,
+                              GLuint op,
+                              GLuint dest,
+                              GLuint mask,
+                              GLuint saturate,
+                              GLuint src0, GLuint src1, GLuint src2);
 
-extern GLuint i915_emit_decl( struct i915_fragment_program *p,
-			     GLuint type, GLuint nr, GLuint d0_flags );
+extern GLuint i915_emit_decl(struct i915_fragment_program *p,
+                             GLuint type, GLuint nr, GLuint d0_flags);
 
 
-extern GLuint i915_emit_const1f( struct i915_fragment_program *p, 
-				GLfloat c0 );
+extern GLuint i915_emit_const1f(struct i915_fragment_program *p, GLfloat c0);
 
-extern GLuint i915_emit_const2f( struct i915_fragment_program *p, 
-				GLfloat c0, GLfloat c1 );
+extern GLuint i915_emit_const2f(struct i915_fragment_program *p,
+                                GLfloat c0, GLfloat c1);
 
-extern GLuint i915_emit_const4fv( struct i915_fragment_program *p,
-				 const GLfloat *c );
+extern GLuint i915_emit_const4fv(struct i915_fragment_program *p,
+                                 const GLfloat * c);
 
-extern GLuint i915_emit_const4f( struct i915_fragment_program *p, 
-				GLfloat c0, GLfloat c1, 
-				GLfloat c2, GLfloat c3 );
+extern GLuint i915_emit_const4f(struct i915_fragment_program *p,
+                                GLfloat c0, GLfloat c1,
+                                GLfloat c2, GLfloat c3);
 
 
-extern GLuint i915_emit_param4fv( struct i915_fragment_program *p, 
-				 const GLfloat *values );
+extern GLuint i915_emit_param4fv(struct i915_fragment_program *p,
+                                 const GLfloat * values);
 
-extern void i915_program_error( struct i915_fragment_program *p,
-                                const char *msg );
+extern void i915_program_error(struct i915_fragment_program *p,
+                               const char *msg);
 
-extern void i915_init_program( i915ContextPtr i915,
-			      struct i915_fragment_program *p );
+extern void i915_init_program(struct i915_context *i915,
+                              struct i915_fragment_program *p);
 
-extern void i915_upload_program( i915ContextPtr i915, 
-				struct i915_fragment_program *p );
+extern void i915_upload_program(struct i915_context *i915,
+                                struct i915_fragment_program *p);
 
-extern void i915_fini_program( struct i915_fragment_program *p );
+extern void i915_fini_program(struct i915_fragment_program *p);
 
 
 
diff --git a/src/mesa/drivers/dri/i915/i915_reg.h b/src/mesa/drivers/dri/i915/i915_reg.h
index 694cd4c8c3c..a9fa56e8a60 100644
--- a/src/mesa/drivers/dri/i915/i915_reg.h
+++ b/src/mesa/drivers/dri/i915/i915_reg.h
@@ -112,6 +112,20 @@
 /* 3DSTATE_CHROMA_KEY */
 
 /* 3DSTATE_CLEAR_PARAMETERS, p150 */
+/* 
+ * Sets the color, depth and stencil clear values used by the
+ * CLEAR_RECT and ZONE_INIT primitive types, respectively.  These
+ * primitives set override most 3d state and only take a minimal x/y
+ * vertex.  The color/z/stencil information is supplied here and
+ * therefore cannot vary per vertex.
+ */
+#define _3DSTATE_CLEAR_PARAMETERS	(CMD_3D | (0x1d<<24) | (0x9c<<16) | 5)
+/* Dword 1 */
+#define CLEARPARAM_CLEAR_RECT		(1 << 16)
+#define CLEARPARAM_ZONE_INIT		(0 << 16)
+#define CLEARPARAM_WRITE_COLOR		(1 << 2)
+#define CLEARPARAM_WRITE_DEPTH		(1 << 1)
+#define CLEARPARAM_WRITE_STENCIL	(1 << 0)
 
 /* 3DSTATE_CONSTANT_BLEND_COLOR, p153 */
 #define _3DSTATE_CONST_BLEND_COLOR_CMD	(CMD_3D | (0x1d<<24) | (0x88<<16))
@@ -424,9 +438,22 @@
 
 #define S7_DEPTH_OFFSET_CONST_MASK     ~0
 
+
+/* Helper macros for blend factors
+ */
+#define DST_BLND_FACT(f) ((f)<<S6_CBUF_DST_BLEND_FACT_SHIFT)
+#define SRC_BLND_FACT(f) ((f)<<S6_CBUF_SRC_BLEND_FACT_SHIFT)
+#define DST_ABLND_FACT(f) ((f)<<IAB_DST_FACTOR_SHIFT)
+#define SRC_ABLND_FACT(f) ((f)<<IAB_SRC_FACTOR_SHIFT)
+
+
+
+
 /* 3DSTATE_MAP_DEINTERLACER_PARAMETERS */
-/* 3DSTATE_MAP_PALETTE_LOAD_32, p206 */
 
+/* 3DSTATE_MAP_PALETTE_LOAD_32, p206 */
+#define _3DSTATE_MAP_PALETTE_LOAD_32    (CMD_3D|(0x1d<<24)|(0x8f<<16))
+/* subsequent dwords up to length (max 16) are ARGB8888 color values */
 
 /* _3DSTATE_MODES_4, p218 */
 #define _3DSTATE_MODES_4_CMD		(CMD_3D|(0x0d<<24))
@@ -435,7 +462,7 @@
 #define LOGICOP_MASK			(0xf<<18)
 #define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
 #define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		((x)<<8)
+#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
 #define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
 #define ENABLE_STENCIL_WRITE_MASK	(1<<16)
 #define STENCIL_WRITE_MASK(x)		((x)&0xff)
@@ -458,7 +485,7 @@
 
 
 #define I915_MAX_TEX_INDIRECT 4
-#define I915_MAX_TEX_INSN     32     
+#define I915_MAX_TEX_INSN     32
 #define I915_MAX_ALU_INSN     64
 #define I915_MAX_DECL_INSN    27
 #define I915_MAX_TEMPORARY    16
@@ -470,33 +497,33 @@
  */
 #define _3DSTATE_PIXEL_SHADER_PROGRAM    (CMD_3D|(0x1d<<24)|(0x5<<16))
 
-#define REG_TYPE_R                 0 /* temporary regs, no need to
-				      * dcl, must be written before
-				      * read -- Preserved between
-				      * phases. 
-				      */
-#define REG_TYPE_T                 1 /* Interpolated values, must be
-				      * dcl'ed before use.
-				      *
-				      * 0..7: texture coord,
-				      * 8: diffuse spec,
-				      * 9: specular color,
-				      * 10: fog parameter in w.
-				      */
-#define REG_TYPE_CONST             2 /* Restriction: only one const
-				      * can be referenced per
-				      * instruction, though it may be
-				      * selected for multiple inputs.
-				      * Constants not initialized
-				      * default to zero.
-				      */
-#define REG_TYPE_S                 3 /* sampler */
-#define REG_TYPE_OC                4 /* output color (rgba) */
-#define REG_TYPE_OD                5 /* output depth (w), xyz are
-				      * temporaries.  If not written,
-				      * interpolated depth is used?
-				      */
-#define REG_TYPE_U                 6 /* unpreserved temporaries */
+#define REG_TYPE_R                 0    /* temporary regs, no need to
+                                         * dcl, must be written before
+                                         * read -- Preserved between
+                                         * phases. 
+                                         */
+#define REG_TYPE_T                 1    /* Interpolated values, must be
+                                         * dcl'ed before use.
+                                         *
+                                         * 0..7: texture coord,
+                                         * 8: diffuse spec,
+                                         * 9: specular color,
+                                         * 10: fog parameter in w.
+                                         */
+#define REG_TYPE_CONST             2    /* Restriction: only one const
+                                         * can be referenced per
+                                         * instruction, though it may be
+                                         * selected for multiple inputs.
+                                         * Constants not initialized
+                                         * default to zero.
+                                         */
+#define REG_TYPE_S                 3    /* sampler */
+#define REG_TYPE_OC                4    /* output color (rgba) */
+#define REG_TYPE_OD                5    /* output depth (w), xyz are
+                                         * temporaries.  If not written,
+                                         * interpolated depth is used?
+                                         */
+#define REG_TYPE_U                 6    /* unpreserved temporaries */
 #define REG_TYPE_MASK              0x7
 #define REG_NR_MASK                0xf
 
@@ -513,34 +540,34 @@
 #define T_TEX7     7
 #define T_DIFFUSE  8
 #define T_SPECULAR 9
-#define T_FOG_W    10		/* interpolated fog is in W coord */
+#define T_FOG_W    10           /* interpolated fog is in W coord */
 
 /* Arithmetic instructions */
 
 /* .replicate_swizzle == selection and replication of a particular
  * scalar channel, ie., .xxxx, .yyyy, .zzzz or .wwww 
  */
-#define A0_NOP    (0x0<<24)		/* no operation */
-#define A0_ADD    (0x1<<24)		/* dst = src0 + src1 */
-#define A0_MOV    (0x2<<24)		/* dst = src0 */
-#define A0_MUL    (0x3<<24)		/* dst = src0 * src1 */
-#define A0_MAD    (0x4<<24)		/* dst = src0 * src1 + src2 */
-#define A0_DP2ADD (0x5<<24)		/* dst.xyzw = src0.xy dot src1.xy + src2.replicate_swizzle */
-#define A0_DP3    (0x6<<24)		/* dst.xyzw = src0.xyz dot src1.xyz */
-#define A0_DP4    (0x7<<24)		/* dst.xyzw = src0.xyzw dot src1.xyzw */
-#define A0_FRC    (0x8<<24)		/* dst = src0 - floor(src0) */
-#define A0_RCP    (0x9<<24)		/* dst.xyzw = 1/(src0.replicate_swizzle) */
-#define A0_RSQ    (0xa<<24)		/* dst.xyzw = 1/(sqrt(abs(src0.replicate_swizzle))) */
-#define A0_EXP    (0xb<<24)		/* dst.xyzw = exp2(src0.replicate_swizzle) */
-#define A0_LOG    (0xc<<24)		/* dst.xyzw = log2(abs(src0.replicate_swizzle)) */
-#define A0_CMP    (0xd<<24)		/* dst = (src0 >= 0.0) ? src1 : src2 */
-#define A0_MIN    (0xe<<24)		/* dst = (src0 < src1) ? src0 : src1 */
-#define A0_MAX    (0xf<<24)		/* dst = (src0 >= src1) ? src0 : src1 */
-#define A0_FLR    (0x10<<24)		/* dst = floor(src0) */
-#define A0_MOD    (0x11<<24)		/* dst = src0 fmod 1.0 */
-#define A0_TRC    (0x12<<24)		/* dst = int(src0) */
-#define A0_SGE    (0x13<<24)		/* dst = src0 >= src1 ? 1.0 : 0.0 */
-#define A0_SLT    (0x14<<24)		/* dst = src0 < src1 ? 1.0 : 0.0 */
+#define A0_NOP    (0x0<<24)     /* no operation */
+#define A0_ADD    (0x1<<24)     /* dst = src0 + src1 */
+#define A0_MOV    (0x2<<24)     /* dst = src0 */
+#define A0_MUL    (0x3<<24)     /* dst = src0 * src1 */
+#define A0_MAD    (0x4<<24)     /* dst = src0 * src1 + src2 */
+#define A0_DP2ADD (0x5<<24)     /* dst.xyzw = src0.xy dot src1.xy + src2.replicate_swizzle */
+#define A0_DP3    (0x6<<24)     /* dst.xyzw = src0.xyz dot src1.xyz */
+#define A0_DP4    (0x7<<24)     /* dst.xyzw = src0.xyzw dot src1.xyzw */
+#define A0_FRC    (0x8<<24)     /* dst = src0 - floor(src0) */
+#define A0_RCP    (0x9<<24)     /* dst.xyzw = 1/(src0.replicate_swizzle) */
+#define A0_RSQ    (0xa<<24)     /* dst.xyzw = 1/(sqrt(abs(src0.replicate_swizzle))) */
+#define A0_EXP    (0xb<<24)     /* dst.xyzw = exp2(src0.replicate_swizzle) */
+#define A0_LOG    (0xc<<24)     /* dst.xyzw = log2(abs(src0.replicate_swizzle)) */
+#define A0_CMP    (0xd<<24)     /* dst = (src0 >= 0.0) ? src1 : src2 */
+#define A0_MIN    (0xe<<24)     /* dst = (src0 < src1) ? src0 : src1 */
+#define A0_MAX    (0xf<<24)     /* dst = (src0 >= src1) ? src0 : src1 */
+#define A0_FLR    (0x10<<24)    /* dst = floor(src0) */
+#define A0_MOD    (0x11<<24)    /* dst = src0 fmod 1.0 */
+#define A0_TRC    (0x12<<24)    /* dst = int(src0) */
+#define A0_SGE    (0x13<<24)    /* dst = src0 >= src1 ? 1.0 : 0.0 */
+#define A0_SLT    (0x14<<24)    /* dst = src0 < src1 ? 1.0 : 0.0 */
 #define A0_DEST_SATURATE                 (1<<22)
 #define A0_DEST_TYPE_SHIFT                19
 /* Allow: R, OC, OD, U */
@@ -599,23 +626,23 @@
 
 
 /* Texture instructions */
-#define T0_TEXLD     (0x15<<24)	/* Sample texture using predeclared
-				 * sampler and address, and output
-				 * filtered texel data to destination
-				 * register */
-#define T0_TEXLDP    (0x16<<24)	/* Same as texld but performs a
-				 * perspective divide of the texture
-				 * coordinate .xyz values by .w before
-				 * sampling. */
-#define T0_TEXLDB    (0x17<<24)	/* Same as texld but biases the
-				 * computed LOD by w.  Only S4.6 two's
-				 * comp is used.  This implies that a
-				 * float to fixed conversion is
-				 * done. */
-#define T0_TEXKILL   (0x18<<24)	/* Does not perform a sampling
-				 * operation.  Simply kills the pixel
-				 * if any channel of the address
-				 * register is < 0.0. */
+#define T0_TEXLD     (0x15<<24) /* Sample texture using predeclared
+                                 * sampler and address, and output
+                                 * filtered texel data to destination
+                                 * register */
+#define T0_TEXLDP    (0x16<<24) /* Same as texld but performs a
+                                 * perspective divide of the texture
+                                 * coordinate .xyz values by .w before
+                                 * sampling. */
+#define T0_TEXLDB    (0x17<<24) /* Same as texld but biases the
+                                 * computed LOD by w.  Only S4.6 two's
+                                 * comp is used.  This implies that a
+                                 * float to fixed conversion is
+                                 * done. */
+#define T0_TEXKILL   (0x18<<24) /* Does not perform a sampling
+                                 * operation.  Simply kills the pixel
+                                 * if any channel of the address
+                                 * register is < 0.0. */
 #define T0_DEST_TYPE_SHIFT                19
 /* Allow: R, OC, OD, U */
 /* Note: U (unpreserved) regs do not retain their values between
@@ -627,18 +654,18 @@
  */
 #define T0_DEST_NR_SHIFT                 14
 /* Allow R: 0..15, OC,OD: 0..0, U: 0..2 */
-#define T0_SAMPLER_NR_SHIFT              0 /* This field ignored for TEXKILL */
+#define T0_SAMPLER_NR_SHIFT              0      /* This field ignored for TEXKILL */
 #define T0_SAMPLER_NR_MASK               (0xf<<0)
 
-#define T1_ADDRESS_REG_TYPE_SHIFT        24 /* Reg to use as texture coord */
+#define T1_ADDRESS_REG_TYPE_SHIFT        24     /* Reg to use as texture coord */
 /* Allow R, T, OC, OD -- R, OC, OD are 'dependent' reads, new program phase */
 #define T1_ADDRESS_REG_NR_SHIFT          17
 #define T2_MBZ                           0
 
 /* Declaration instructions */
-#define D0_DCL       (0x19<<24)	/* Declare a t (interpolated attrib)
-				 * register or an s (sampler)
-				 * register. */
+#define D0_DCL       (0x19<<24) /* Declare a t (interpolated attrib)
+                                 * register or an s (sampler)
+                                 * register. */
 #define D0_SAMPLE_TYPE_SHIFT              22
 #define D0_SAMPLE_TYPE_2D                 (0x0<<22)
 #define D0_SAMPLE_TYPE_CUBE               (0x1<<22)
@@ -695,12 +722,12 @@
 #define    MAPSURF_4BIT_INDEXED		   (7<<7)
 #define MS3_MT_FORMAT_MASK         (0x7 << 3)
 #define MS3_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3) /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_I8		           (0<<3) /* SURFACE_8BIT */
+#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
+#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
 #define    MT_8BIT_L8		           (1<<3)
 #define    MT_8BIT_A8		           (4<<3)
 #define    MT_8BIT_MONO8	           (5<<3)
-#define    MT_16BIT_RGB565 		   (0<<3) /* SURFACE_16BIT */
+#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
 #define    MT_16BIT_ARGB1555		   (1<<3)
 #define    MT_16BIT_ARGB4444		   (2<<3)
 #define    MT_16BIT_AY88		   (3<<3)
@@ -709,7 +736,7 @@
 #define    MT_16BIT_I16	                   (7<<3)
 #define    MT_16BIT_L16	                   (8<<3)
 #define    MT_16BIT_A16	                   (9<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3) /* SURFACE_32BIT */
+#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
 #define    MT_32BIT_ABGR8888		   (1<<3)
 #define    MT_32BIT_XRGB8888		   (2<<3)
 #define    MT_32BIT_XBGR8888		   (3<<3)
@@ -725,11 +752,11 @@
 #define    MT_32BIT_xI824	           (0xD<<3)
 #define    MT_32BIT_xA824	           (0xE<<3)
 #define    MT_32BIT_xL824	           (0xF<<3)
-#define    MT_422_YCRCB_SWAPY	           (0<<3) /* SURFACE_422 */
+#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
 #define    MT_422_YCRCB_NORMAL	           (1<<3)
 #define    MT_422_YCRCB_SWAPUV	           (2<<3)
 #define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3) /* SURFACE_COMPRESSED */
+#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
 #define    MT_COMPRESS_DXT2_3	           (1<<3)
 #define    MT_COMPRESS_DXT4_5	           (2<<3)
 #define    MT_COMPRESS_FXT1		   (3<<3)
@@ -751,7 +778,7 @@
 #define MS4_MIP_LAYOUT_LEGACY           (0<<8)
 #define MS4_MIP_LAYOUT_BELOW_LPT        (0<<8)
 #define MS4_MIP_LAYOUT_RIGHT_LPT        (1<<8)
-#define MS4_VOLUME_DEPTH_SHIFT          0    
+#define MS4_VOLUME_DEPTH_SHIFT          0
 #define MS4_VOLUME_DEPTH_MASK           (0xff<<0)
 
 /* p244 */
@@ -779,7 +806,7 @@
 #define   FILTER_4X4_1    	3
 #define   FILTER_4X4_2    	4
 #define   FILTER_4X4_FLAT 	5
-#define   FILTER_6X5_MONO   	6 /* XXX - check */
+#define   FILTER_6X5_MONO   	6       /* XXX - check */
 #define SS2_MIN_FILTER_SHIFT          14
 #define SS2_MIN_FILTER_MASK           (0x7<<14)
 #define SS2_LOD_BIAS_SHIFT            5
@@ -826,10 +853,14 @@
 #define ST1_ENABLE               (1<<16)
 #define ST1_MASK                 (0xffff)
 
+#define _3DSTATE_DEFAULT_Z          ((0x3<<29)|(0x1d<<24)|(0x98<<16))
+#define _3DSTATE_DEFAULT_DIFFUSE    ((0x3<<29)|(0x1d<<24)|(0x99<<16))
+#define _3DSTATE_DEFAULT_SPECULAR   ((0x3<<29)|(0x1d<<24)|(0x9a<<16))
+
 
-#define MI_FLUSH           ((0<<29)|(4<<23))
-#define FLUSH_MAP_CACHE    (1<<0)
-#define FLUSH_RENDER_CACHE (1<<1)
+#define MI_FLUSH                   ((0<<29)|(4<<23))
+#define FLUSH_MAP_CACHE            (1<<0)
+#define INHIBIT_FLUSH_RENDER_CACHE (1<<2)
 
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c
index 1c4ec747558..e5d8d279936 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -38,97 +38,99 @@
 
 #include "drivers/common/driverfuncs.h"
 
+#include "intel_fbo.h"
 #include "intel_screen.h"
 #include "intel_batchbuffer.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-
+#define FILE_DEBUG_FLAG DEBUG_STATE
 
 static void
-i915StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func, GLint ref,
+i915StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
                         GLuint mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
 
    mask = mask & 0xff;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(func), ref, mask);
+   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(func), ref, mask);
 
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-					 STENCIL_TEST_MASK(mask));
+                                           STENCIL_TEST_MASK(mask));
 
    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-					S5_STENCIL_TEST_FUNC_MASK);
-					
-   i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) |  
- 				       (test << S5_STENCIL_TEST_FUNC_SHIFT)); 
+                                          S5_STENCIL_TEST_FUNC_MASK);
+
+   i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) |
+                                         (test <<
+                                          S5_STENCIL_TEST_FUNC_SHIFT));
 }
 
 static void
-i915StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+i915StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s : mask 0x%x\n", __FUNCTION__, mask);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
+   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
+   
    mask = mask & 0xff;
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-					STENCIL_WRITE_MASK(mask));
+                                           STENCIL_WRITE_MASK(mask));
 }
 
 
 static void
-i915StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail, GLenum zfail,
+i915StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
                       GLenum zpass)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int fop = intel_translate_stencil_op(fail); 
-   int dfop = intel_translate_stencil_op(zfail); 
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int fop = intel_translate_stencil_op(fail);
+   int dfop = intel_translate_stencil_op(zfail);
    int dpop = intel_translate_stencil_op(zpass);
 
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr(fail),
-	      _mesa_lookup_enum_by_nr(zfail),
-	      _mesa_lookup_enum_by_nr(zpass));
+   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
+       _mesa_lookup_enum_by_nr(fail),
+       _mesa_lookup_enum_by_nr(zfail), _mesa_lookup_enum_by_nr(zpass));
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-					S5_STENCIL_PASS_Z_FAIL_MASK |
-					S5_STENCIL_PASS_Z_PASS_MASK);
+                                          S5_STENCIL_PASS_Z_FAIL_MASK |
+                                          S5_STENCIL_PASS_Z_PASS_MASK);
 
    i915->state.Ctx[I915_CTXREG_LIS5] |= ((fop << S5_STENCIL_FAIL_SHIFT) |
-				       (dfop << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-				       (dpop << S5_STENCIL_PASS_Z_PASS_SHIFT));
+                                         (dfop <<
+                                          S5_STENCIL_PASS_Z_FAIL_SHIFT) |
+                                         (dpop <<
+                                          S5_STENCIL_PASS_Z_PASS_SHIFT));
 }
 
-static void i915AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+static void
+i915AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
    GLubyte refByte;
 
    UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_LIS6] &= ~(S6_ALPHA_TEST_FUNC_MASK |
-					S6_ALPHA_REF_MASK);
+                                          S6_ALPHA_REF_MASK);
    i915->state.Ctx[I915_CTXREG_LIS6] |= ((test << S6_ALPHA_TEST_FUNC_SHIFT) |
-				       (((GLuint)refByte) << S6_ALPHA_REF_SHIFT));
+                                         (((GLuint) refByte) <<
+                                          S6_ALPHA_REF_SHIFT));
 }
 
 /* This function makes sure that the proper enables are
@@ -137,41 +139,45 @@ static void i915AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
  * could change the LogicOp or Independant Alpha Blend without subsequent
  * calls to glEnable.
  */
-static void i915EvalLogicOpBlendState(GLcontext *ctx)
+static void
+i915EvalLogicOpBlendState(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (RGBA_LOGICOP_ENABLED(ctx)) {
       i915->state.Ctx[I915_CTXREG_LIS5] |= S5_LOGICOP_ENABLE;
       i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
-   } else {
+   }
+   else {
       i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_LOGICOP_ENABLE;
 
       if (ctx->Color.BlendEnabled) {
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_CBUF_BLEND_ENABLE;
-      } else {
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_CBUF_BLEND_ENABLE;
+      }
+      else {
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
       }
    }
 }
 
-static void i915BlendColor(GLcontext *ctx, const GLfloat color[4])
+static void
+i915BlendColor(GLcontext * ctx, const GLfloat color[4])
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLubyte r, g, b, a;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
    UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
 
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = (a<<24) | (r<<16) | (g<<8) | b;
+   i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] =
+      (a << 24) | (r << 16) | (g << 8) | b;
 }
 
 
@@ -182,31 +188,37 @@ static void i915BlendColor(GLcontext *ctx, const GLfloat color[4])
 
 
 
-static GLuint translate_blend_equation( GLenum mode )
+static GLuint
+translate_blend_equation(GLenum mode)
 {
    switch (mode) {
-   case GL_FUNC_ADD: return BLENDFUNC_ADD; 
-   case GL_MIN: return BLENDFUNC_MIN; 
-   case GL_MAX: return BLENDFUNC_MAX; 
-   case GL_FUNC_SUBTRACT: return BLENDFUNC_SUBTRACT; 
-   case GL_FUNC_REVERSE_SUBTRACT: return BLENDFUNC_REVERSE_SUBTRACT; 
-   default: return 0;
+   case GL_FUNC_ADD:
+      return BLENDFUNC_ADD;
+   case GL_MIN:
+      return BLENDFUNC_MIN;
+   case GL_MAX:
+      return BLENDFUNC_MAX;
+   case GL_FUNC_SUBTRACT:
+      return BLENDFUNC_SUBTRACT;
+   case GL_FUNC_REVERSE_SUBTRACT:
+      return BLENDFUNC_REVERSE_SUBTRACT;
+   default:
+      return 0;
    }
 }
 
-static void i915UpdateBlendState( GLcontext *ctx )
+static void
+i915UpdateBlendState(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   GLuint iab = (i915->state.Ctx[I915_CTXREG_IAB] & 
-		 ~(IAB_SRC_FACTOR_MASK |
-		   IAB_DST_FACTOR_MASK |
-		   (BLENDFUNC_MASK << IAB_FUNC_SHIFT) |
-		   IAB_ENABLE));
-
-   GLuint lis6 = (i915->state.Ctx[I915_CTXREG_LIS6] & 
-		  ~(S6_CBUF_SRC_BLEND_FACT_MASK |
-		    S6_CBUF_DST_BLEND_FACT_MASK |
-		    S6_CBUF_BLEND_FUNC_MASK));
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   GLuint iab = (i915->state.Ctx[I915_CTXREG_IAB] &
+                 ~(IAB_SRC_FACTOR_MASK |
+                   IAB_DST_FACTOR_MASK |
+                   (BLENDFUNC_MASK << IAB_FUNC_SHIFT) | IAB_ENABLE));
+
+   GLuint lis6 = (i915->state.Ctx[I915_CTXREG_LIS6] &
+                  ~(S6_CBUF_SRC_BLEND_FACT_MASK |
+                    S6_CBUF_DST_BLEND_FACT_MASK | S6_CBUF_BLEND_FUNC_MASK));
 
    GLuint eqRGB = ctx->Color.BlendEquationRGB;
    GLuint eqA = ctx->Color.BlendEquationA;
@@ -223,15 +235,15 @@ static void i915UpdateBlendState( GLcontext *ctx )
       srcA = dstA = GL_ONE;
    }
 
-   lis6 |= SRC_BLND_FACT(intel_translate_blend_factor(srcRGB)); 
-   lis6 |= DST_BLND_FACT(intel_translate_blend_factor(dstRGB)); 
-   lis6 |= translate_blend_equation( eqRGB ) << S6_CBUF_BLEND_FUNC_SHIFT;
+   lis6 |= SRC_BLND_FACT(intel_translate_blend_factor(srcRGB));
+   lis6 |= DST_BLND_FACT(intel_translate_blend_factor(dstRGB));
+   lis6 |= translate_blend_equation(eqRGB) << S6_CBUF_BLEND_FUNC_SHIFT;
 
-   iab |= SRC_ABLND_FACT(intel_translate_blend_factor(srcA)); 
-   iab |= DST_ABLND_FACT(intel_translate_blend_factor(dstA)); 
-   iab |= translate_blend_equation( eqA ) << IAB_FUNC_SHIFT;
+   iab |= SRC_ABLND_FACT(intel_translate_blend_factor(srcA));
+   iab |= DST_ABLND_FACT(intel_translate_blend_factor(dstA));
+   iab |= translate_blend_equation(eqA) << IAB_FUNC_SHIFT;
 
-   if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB) 
+   if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB)
       iab |= IAB_ENABLE;
 
    if (iab != i915->state.Ctx[I915_CTXREG_IAB] ||
@@ -246,41 +258,41 @@ static void i915UpdateBlendState( GLcontext *ctx )
 }
 
 
-static void i915BlendFuncSeparate(GLcontext *ctx, GLenum srcRGB, 
-				 GLenum dstRGB, GLenum srcA,
-				 GLenum dstA )
-{  
-   i915UpdateBlendState( ctx );
+static void
+i915BlendFuncSeparate(GLcontext * ctx, GLenum srcRGB,
+                      GLenum dstRGB, GLenum srcA, GLenum dstA)
+{
+   i915UpdateBlendState(ctx);
 }
 
 
-static void i915BlendEquationSeparate(GLcontext *ctx, GLenum eqRGB,
-				     GLenum eqA) 
+static void
+i915BlendEquationSeparate(GLcontext * ctx, GLenum eqRGB, GLenum eqA)
 {
-   i915UpdateBlendState( ctx );
+   i915UpdateBlendState(ctx);
 }
 
 
-static void i915DepthFunc(GLcontext *ctx, GLenum func)
+static void
+i915DepthFunc(GLcontext * ctx, GLenum func)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func( func );
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
+   int test = intel_translate_compare_func(func);
 
+   DBG("%s\n", __FUNCTION__);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
    i915->state.Ctx[I915_CTXREG_LIS6] |= test << S6_DEPTH_TEST_FUNC_SHIFT;
 }
 
-static void i915DepthMask(GLcontext *ctx, GLboolean flag)
+static void
+i915DepthMask(GLcontext * ctx, GLboolean flag)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s flag (%d)\n", __FUNCTION__, flag);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
+   DBG("%s flag (%d)\n", __FUNCTION__, flag);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (flag && ctx->Depth.Test)
@@ -295,14 +307,15 @@ static void i915DepthMask(GLcontext *ctx, GLboolean flag)
  * The i915 supports a 4x4 stipple natively, GL wants 32x32.
  * Fortunately stipple is usually a repeating pattern.
  */
-static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
+static void
+i915PolygonStipple(GLcontext * ctx, const GLubyte * mask)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    const GLubyte *m = mask;
    GLubyte p[4];
-   int i,j,k;
+   int i, j, k;
    int active = (ctx->Polygon.StippleFlag &&
-		 i915->intel.reduced_primitive == GL_TRIANGLES);
+                 i915->intel.reduced_primitive == GL_TRIANGLES);
    GLuint newMask;
 
    if (active) {
@@ -310,23 +323,26 @@ static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
       i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
    }
 
-   p[0] = mask[12] & 0xf; p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf; p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf; p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf; p[3] |= p[3] << 4;
-
-   for (k = 0 ; k < 8 ; k++)
-      for (j = 3 ; j >= 0; j--)
-	 for (i = 0 ; i < 4 ; i++, m++)
-	    if (*m != p[j]) {
-	       i915->intel.hw_stipple = 0;
-	       return;
-	    }
+   p[0] = mask[12] & 0xf;
+   p[0] |= p[0] << 4;
+   p[1] = mask[8] & 0xf;
+   p[1] |= p[1] << 4;
+   p[2] = mask[4] & 0xf;
+   p[2] |= p[2] << 4;
+   p[3] = mask[0] & 0xf;
+   p[3] |= p[3] << 4;
+
+   for (k = 0; k < 8; k++)
+      for (j = 3; j >= 0; j--)
+         for (i = 0; i < 4; i++, m++)
+            if (*m != p[j]) {
+               i915->intel.hw_stipple = 0;
+               return;
+            }
 
    newMask = (((p[0] & 0xf) << 0) |
-	      ((p[1] & 0xf) << 4) |
-	      ((p[2] & 0xf) << 8) |
-	      ((p[3] & 0xf) << 12));
+              ((p[1] & 0xf) << 4) |
+              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
 
 
    if (newMask == 0xffff || newMask == 0x0) {
@@ -347,49 +363,54 @@ static void i915PolygonStipple( GLcontext *ctx, const GLubyte *mask )
 /* =============================================================
  * Hardware clipping
  */
-static void i915Scissor(GLcontext *ctx, GLint x, GLint y, 
-			GLsizei w, GLsizei h)
+static void
+i915Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   intelScreenPrivate *screen = i915->intel.intelScreen;
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int x1, y1, x2, y2;
 
-   if (!i915->intel.driDrawable)
+   if (!ctx->DrawBuffer)
       return;
 
-   x1 = x;
-   y1 = i915->intel.driDrawable->h - (y + h);
-   x2 = x + w - 1;
-   y2 = y1 + h - 1;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "[%s] x(%d) y(%d) w(%d) h(%d)\n", __FUNCTION__,
-	      x, y, w, h);
-
-   if (x1 < 0) x1 = 0;
-   if (y1 < 0) y1 = 0;
-   if (x2 < 0) x2 = 0;
-   if (y2 < 0) y2 = 0;
-
-   if (x2 >= screen->width) x2 = screen->width-1;
-   if (y2 >= screen->height) y2 = screen->height-1;
-   if (x1 >= screen->width) x1 = screen->width-1;
-   if (y1 >= screen->height) y1 = screen->height-1;
+   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
 
+   if (ctx->DrawBuffer->Name == 0) {
+      x1 = x;
+      y1 = ctx->DrawBuffer->Height - (y + h);
+      x2 = x + w - 1;
+      y2 = y1 + h - 1;
+      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   else {
+      /* FBO - not inverted
+       */
+      x1 = x;
+      y1 = y;
+      x2 = x + w - 1;
+      y2 = y + h - 1;
+      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
+   }
+   
+   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
+   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
+   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
+   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
+   
+   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
 
    I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
    i915->state.Buffer[I915_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
    i915->state.Buffer[I915_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
 }
 
-static void i915LogicOp(GLcontext *ctx, GLenum opcode)
+static void
+i915LogicOp(GLcontext * ctx, GLenum opcode)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int tmp = intel_translate_logic_op(opcode);
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
    i915->state.Ctx[I915_CTXREG_STATE4] &= ~LOGICOP_MASK;
    i915->state.Ctx[I915_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
@@ -397,13 +418,14 @@ static void i915LogicOp(GLcontext *ctx, GLenum opcode)
 
 
 
-static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
+static void
+i915CullFaceFrontFace(GLcontext * ctx, GLenum unused)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLuint mode;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   DBG("%s %d\n", __FUNCTION__,
+       ctx->DrawBuffer ? ctx->DrawBuffer->Name : 0);
 
    if (!ctx->Polygon.CullFlag) {
       mode = S4_CULLMODE_NONE;
@@ -411,10 +433,12 @@ static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    else if (ctx->Polygon.CullFaceMode != GL_FRONT_AND_BACK) {
       mode = S4_CULLMODE_CW;
 
+      if (ctx->DrawBuffer && ctx->DrawBuffer->Name != 0)
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
       if (ctx->Polygon.CullFaceMode == GL_FRONT)
-	 mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
       if (ctx->Polygon.FrontFace != GL_CCW)
-	 mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
+         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
    }
    else {
       mode = S4_CULLMODE_BOTH;
@@ -425,16 +449,16 @@ static void i915CullFaceFrontFace(GLcontext *ctx, GLenum unused)
    i915->state.Ctx[I915_CTXREG_LIS4] |= mode;
 }
 
-static void i915LineWidth( GLcontext *ctx, GLfloat widthf )
+static void
+i915LineWidth(GLcontext * ctx, GLfloat widthf)
 {
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_LINE_WIDTH_MASK;
    int width;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   width = (int)(widthf * 2);
+   DBG("%s\n", __FUNCTION__);
+   
+   width = (int) (widthf * 2);
    CLAMP_SELF(width, 1, 0xf);
    lis4 |= width << S4_LINE_WIDTH_SHIFT;
 
@@ -444,15 +468,15 @@ static void i915LineWidth( GLcontext *ctx, GLfloat widthf )
    }
 }
 
-static void i915PointSize(GLcontext *ctx, GLfloat size)
+static void
+i915PointSize(GLcontext * ctx, GLfloat size)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_POINT_WIDTH_MASK;
-   GLint point_size = (int)size;
-
-   if (INTEL_DEBUG&DEBUG_DRI)
-     fprintf(stderr, "%s\n", __FUNCTION__);
+   GLint point_size = (int) size;
 
+   DBG("%s\n", __FUNCTION__);
+   
    CLAMP_SELF(point_size, 1, 255);
    lis4 |= point_size << S4_POINT_WIDTH_SHIFT;
 
@@ -467,20 +491,24 @@ static void i915PointSize(GLcontext *ctx, GLfloat size)
  * Color masks
  */
 
-static void i915ColorMask(GLcontext *ctx,
-			 GLboolean r, GLboolean g,
-			 GLboolean b, GLboolean a)
+static void
+i915ColorMask(GLcontext * ctx,
+              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
 {
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLuint tmp = i915->state.Ctx[I915_CTXREG_LIS5] & ~S5_WRITEDISABLE_MASK;
 
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
+   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b,
+       a);
 
-   if (!r) tmp |= S5_WRITEDISABLE_RED;
-   if (!g) tmp |= S5_WRITEDISABLE_GREEN;
-   if (!b) tmp |= S5_WRITEDISABLE_BLUE;
-   if (!a) tmp |= S5_WRITEDISABLE_ALPHA;
+   if (!r)
+      tmp |= S5_WRITEDISABLE_RED;
+   if (!g)
+      tmp |= S5_WRITEDISABLE_GREEN;
+   if (!b)
+      tmp |= S5_WRITEDISABLE_BLUE;
+   if (!a)
+      tmp |= S5_WRITEDISABLE_ALPHA;
 
    if (tmp != i915->state.Ctx[I915_CTXREG_LIS5]) {
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
@@ -488,54 +516,55 @@ static void i915ColorMask(GLcontext *ctx,
    }
 }
 
-static void update_specular( GLcontext *ctx )
+static void
+update_specular(GLcontext * ctx)
 {
    /* A hack to trigger the rebuild of the fragment program.
     */
-   INTEL_CONTEXT(ctx)->NewGLState |= _NEW_TEXTURE;
-   I915_CONTEXT(ctx)->tex_program.translated = 0; 
+   intel_context(ctx)->NewGLState |= _NEW_TEXTURE;
 }
 
-static void i915LightModelfv(GLcontext *ctx, GLenum pname, 
-			     const GLfloat *param)
+static void
+i915LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   if (INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
+   DBG("%s\n", __FUNCTION__);
+   
    if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular( ctx );
+      update_specular(ctx);
    }
 }
 
-static void i915ShadeModel(GLcontext *ctx, GLenum mode)
+static void
+i915ShadeModel(GLcontext * ctx, GLenum mode)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    I915_STATECHANGE(i915, I915_UPLOAD_CTX);
 
    if (mode == GL_SMOOTH) {
-     i915->state.Ctx[I915_CTXREG_LIS4] &= ~(S4_FLATSHADE_ALPHA | 
-					  S4_FLATSHADE_COLOR | 
-					  S4_FLATSHADE_SPECULAR);
-   } else {
-     i915->state.Ctx[I915_CTXREG_LIS4] |= (S4_FLATSHADE_ALPHA | 
-					 S4_FLATSHADE_COLOR | 
-					 S4_FLATSHADE_SPECULAR);
+      i915->state.Ctx[I915_CTXREG_LIS4] &= ~(S4_FLATSHADE_ALPHA |
+                                             S4_FLATSHADE_COLOR |
+                                             S4_FLATSHADE_SPECULAR);
+   }
+   else {
+      i915->state.Ctx[I915_CTXREG_LIS4] |= (S4_FLATSHADE_ALPHA |
+                                            S4_FLATSHADE_COLOR |
+                                            S4_FLATSHADE_SPECULAR);
    }
 }
 
 /* =============================================================
  * Fog
  */
-void i915_update_fog( GLcontext *ctx )
+void
+i915_update_fog(GLcontext * ctx)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
    GLenum mode;
    GLboolean enabled;
    GLboolean try_pixel_fog;
-   
+
    if (ctx->FragmentProgram._Active) {
       /* Pull in static fog state from program */
-      
       mode = ctx->FragmentProgram._Current->FogOption;
       enabled = (mode != GL_NONE);
       try_pixel_fog = 0;
@@ -546,7 +575,7 @@ void i915_update_fog( GLcontext *ctx )
 #if 0
       /* XXX - DISABLED -- Need ortho fallback */
       try_pixel_fog = (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT
-                       &&ctx->Hint.Fog == GL_NICEST);
+                       && ctx->Hint.Fog == GL_NICEST);
 #else
       try_pixel_fog = 0;
 #endif
@@ -559,48 +588,49 @@ void i915_update_fog( GLcontext *ctx )
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
       i915->vertex_fog = I915_FOG_PIXEL;
-	 
+
       switch (mode) {
       case GL_LINEAR:
-	 if (ctx->Fog.End <= ctx->Fog.Start) {
-	    /* XXX - this won't work with fragment programs.  Need to
-	     * either fallback or append fog instructions to end of
-	     * program in the case of linear fog.
-	     */
-	    i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
-	    i915->vertex_fog = I915_FOG_VERTEX;
-	 }
-	 else {
+         if (ctx->Fog.End <= ctx->Fog.Start) {
+            /* XXX - this won't work with fragment programs.  Need to
+             * either fallback or append fog instructions to end of
+             * program in the case of linear fog.
+             */
+            printf("vertex fog!\n");
+            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
+            i915->vertex_fog = I915_FOG_VERTEX;
+         }
+         else {
             GLfloat c2 = 1.0 / (ctx->Fog.End - ctx->Fog.Start);
             GLfloat c1 = ctx->Fog.End * c2;
 
-	    i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_C1_MASK;
-	    i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_LINEAR;
-	    i915->state.Fog[I915_FOGREG_MODE1] |= 
-	       ((GLuint)(c1 * FMC1_C1_ONE)) & FMC1_C1_MASK;
-
-	    if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-	       i915->state.Fog[I915_FOGREG_MODE2]
-                  = (GLuint)(c2 * FMC2_C2_ONE);
-	    }
-	    else {
-	       fi_type fi;
-	       fi.f = c2; 
-	       i915->state.Fog[I915_FOGREG_MODE2] = fi.i; 
-	    }
-	 }
-	 break;
+            i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_C1_MASK;
+            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_LINEAR;
+            i915->state.Fog[I915_FOGREG_MODE1] |=
+               ((GLuint) (c1 * FMC1_C1_ONE)) & FMC1_C1_MASK;
+
+            if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
+               i915->state.Fog[I915_FOGREG_MODE2]
+                  = (GLuint) (c2 * FMC2_C2_ONE);
+            }
+            else {
+               fi_type fi;
+               fi.f = c2;
+               i915->state.Fog[I915_FOGREG_MODE2] = fi.i;
+            }
+         }
+         break;
       case GL_EXP:
-	 i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP;
-	 break;
+         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP;
+         break;
       case GL_EXP2:
-	 i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP2;
-	 break;
+         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP2;
+         break;
       default:
-	 break;
+         break;
       }
    }
-   else /* if (i915->vertex_fog != I915_FOG_VERTEX) */ {      
+   else { /* if (i915->vertex_fog != I915_FOG_VERTEX) */
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
       i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
@@ -622,38 +652,38 @@ void i915_update_fog( GLcontext *ctx )
 }
 
 static void
-i915Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
+i915Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
    switch (pname) {
-   case GL_FOG_COORDINATE_SOURCE_EXT: 
+   case GL_FOG_COORDINATE_SOURCE_EXT:
    case GL_FOG_MODE:
    case GL_FOG_START:
-   case GL_FOG_END: 
+   case GL_FOG_END:
       break;
 
    case GL_FOG_DENSITY:
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
 
       if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-	 i915->state.Fog[I915_FOGREG_MODE3]
-            = (GLuint)(ctx->Fog.Density * FMC3_D_ONE);
+         i915->state.Fog[I915_FOGREG_MODE3] =
+            (GLuint) (ctx->Fog.Density * FMC3_D_ONE);
       }
       else {
-	 union { float f; int i; } fi;
-	 fi.f = ctx->Fog.Density; 
-	 i915->state.Fog[I915_FOGREG_MODE3] = fi.i; 
+         fi_type fi;
+         fi.f = ctx->Fog.Density;
+         i915->state.Fog[I915_FOGREG_MODE3] = fi.i;
       }
       break;
 
-   case GL_FOG_COLOR: 
+   case GL_FOG_COLOR:
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_COLOR] = 
-	 (_3DSTATE_FOG_COLOR_CMD | 
-	  ((GLubyte)(ctx->Fog.Color[0]*255.0F) << 16) |
-	  ((GLubyte)(ctx->Fog.Color[1]*255.0F) << 8) |
-	  ((GLubyte)(ctx->Fog.Color[2]*255.0F) << 0));
+      i915->state.Fog[I915_FOGREG_COLOR] =
+         (_3DSTATE_FOG_COLOR_CMD |
+          ((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
+          ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
+          ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
       break;
 
    default:
@@ -661,7 +691,8 @@ i915Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
    }
 }
 
-static void i915Hint(GLcontext *ctx, GLenum target, GLenum state)
+static void
+i915Hint(GLcontext * ctx, GLenum target, GLenum state)
 {
    switch (target) {
    case GL_FOG_HINT:
@@ -674,25 +705,26 @@ static void i915Hint(GLcontext *ctx, GLenum target, GLenum state)
 /* =============================================================
  */
 
-static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+static void
+i915Enable(GLcontext * ctx, GLenum cap, GLboolean state)
 {
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
-   switch(cap) {
+   switch (cap) {
    case GL_TEXTURE_2D:
       break;
 
    case GL_LIGHTING:
    case GL_COLOR_SUM:
-      update_specular( ctx );
+      update_specular(ctx);
       break;
 
    case GL_ALPHA_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_ALPHA_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_ALPHA_TEST_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_ALPHA_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_ALPHA_TEST_ENABLE;
       break;
 
    case GL_BLEND:
@@ -704,8 +736,8 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 
       /* Logicop doesn't seem to work at 16bpp:
        */
-      if (i915->intel.intelScreen->cpp == 2)
-	 FALLBACK( &i915->intel, I915_FALLBACK_LOGICOP, state );
+      if (i915->intel.intelScreen->cpp == 2)    /* XXX FBO fix */
+         FALLBACK(&i915->intel, I915_FALLBACK_LOGICOP, state);
       break;
 
    case GL_FRAGMENT_PROGRAM_ARB:
@@ -714,37 +746,37 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
    case GL_DITHER:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_COLOR_DITHER_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_COLOR_DITHER_ENABLE;
       break;
 
    case GL_DEPTH_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_TEST_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_ENABLE;
 
-      i915DepthMask( ctx, ctx->Depth.Mask );
+      i915DepthMask(ctx, ctx->Depth.Mask);
       break;
 
    case GL_SCISSOR_TEST:
       I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
       if (state)
-	 i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						   ENABLE_SCISSOR_RECT);
+         i915->state.Buffer[I915_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
       else
-	 i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						   DISABLE_SCISSOR_RECT);
+         i915->state.Buffer[I915_DESTREG_SENABLE] =
+            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
       break;
 
    case GL_LINE_SMOOTH:
       I915_STATECHANGE(i915, I915_UPLOAD_CTX);
       if (state)
-	 i915->state.Ctx[I915_CTXREG_LIS4] |= S4_LINE_ANTIALIAS_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS4] |= S4_LINE_ANTIALIAS_ENABLE;
       else
-	 i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_LINE_ANTIALIAS_ENABLE;
+         i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_LINE_ANTIALIAS_ENABLE;
       break;
 
    case GL_FOG:
@@ -755,16 +787,25 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
       break;
 
    case GL_STENCIL_TEST:
-      if (i915->intel.hw_stencil) {
-	 I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-	 if (state)
-	    i915->state.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
-						S5_STENCIL_WRITE_ENABLE);
-	 else
-	    i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE | 
-						 S5_STENCIL_WRITE_ENABLE);
-      } else {
-	 FALLBACK( &i915->intel, I915_FALLBACK_STENCIL, state );
+      {
+         GLboolean hw_stencil = GL_FALSE;
+         if (ctx->DrawBuffer) {
+            struct intel_renderbuffer *irbStencil
+               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
+            hw_stencil = (irbStencil && irbStencil->region);
+         }
+         if (hw_stencil) {
+            I915_STATECHANGE(i915, I915_UPLOAD_CTX);
+            if (state)
+               i915->state.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
+                                                     S5_STENCIL_WRITE_ENABLE);
+            else
+               i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
+                                                      S5_STENCIL_WRITE_ENABLE);
+         }
+         else {
+            FALLBACK(&i915->intel, I915_FALLBACK_STENCIL, state);
+         }
       }
       break;
 
@@ -773,23 +814,20 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
        * I'll do more testing later to find out exactly which hardware
        * supports it.  Disabled for now.
        */
-      if (i915->intel.hw_stipple && 
-	  i915->intel.reduced_primitive == GL_TRIANGLES)
-      {
-	 I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-	 if (state)
-	    i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
-	 else
-	    i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
+      if (i915->intel.hw_stipple &&
+          i915->intel.reduced_primitive == GL_TRIANGLES) {
+         I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
+         if (state)
+            i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
+         else
+            i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
       }
       break;
 
    case GL_POLYGON_SMOOTH:
-      FALLBACK( &i915->intel, I915_FALLBACK_POLYGON_SMOOTH, state );
       break;
 
    case GL_POINT_SMOOTH:
-      FALLBACK( &i915->intel, I915_FALLBACK_POINT_SMOOTH, state );
       break;
 
    default:
@@ -798,7 +836,8 @@ static void i915Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 }
 
 
-static void i915_init_packets( i915ContextPtr i915 )
+static void
+i915_init_packets(struct i915_context *i915)
 {
    intelScreenPrivate *screen = i915->intel.intelScreen;
 
@@ -811,39 +850,35 @@ static void i915_init_packets( i915ContextPtr i915 )
       /* Probably don't want to upload all this stuff every time one 
        * piece changes.
        */
-      i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 | 
-				       I1_LOAD_S(2) |
-				       I1_LOAD_S(4) |
-				       I1_LOAD_S(5) |
-				       I1_LOAD_S(6) | 
-				       (3));
+      i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
+                                         I1_LOAD_S(2) |
+                                         I1_LOAD_S(4) |
+                                         I1_LOAD_S(5) | I1_LOAD_S(6) | (3));
       i915->state.Ctx[I915_CTXREG_LIS2] = 0;
       i915->state.Ctx[I915_CTXREG_LIS4] = 0;
       i915->state.Ctx[I915_CTXREG_LIS5] = 0;
 
-      if (screen->cpp == 2)
-	 i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
+      if (screen->cpp == 2)     /* XXX FBO fix */
+         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
 
 
       i915->state.Ctx[I915_CTXREG_LIS6] = (S6_COLOR_WRITE_ENABLE |
-					 (2 << S6_TRISTRIP_PV_SHIFT));
+                                           (2 << S6_TRISTRIP_PV_SHIFT));
 
       i915->state.Ctx[I915_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-					   ENABLE_LOGIC_OP_FUNC |
-					   LOGIC_OP_FUNC(LOGICOP_COPY) |
-					   ENABLE_STENCIL_TEST_MASK |
-					   STENCIL_TEST_MASK(0xff) |
-					   ENABLE_STENCIL_WRITE_MASK |
-					   STENCIL_WRITE_MASK(0xff));
-
-
-      i915->state.Ctx[I915_CTXREG_IAB] = (_3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD |
-					IAB_MODIFY_ENABLE |
-					IAB_MODIFY_FUNC |
-					IAB_MODIFY_SRC_FACTOR |
-					IAB_MODIFY_DST_FACTOR);
-
-      i915->state.Ctx[I915_CTXREG_BLENDCOLOR0] = _3DSTATE_CONST_BLEND_COLOR_CMD;
+                                             ENABLE_LOGIC_OP_FUNC |
+                                             LOGIC_OP_FUNC(LOGICOP_COPY) |
+                                             ENABLE_STENCIL_TEST_MASK |
+                                             STENCIL_TEST_MASK(0xff) |
+                                             ENABLE_STENCIL_WRITE_MASK |
+                                             STENCIL_WRITE_MASK(0xff));
+
+      i915->state.Ctx[I915_CTXREG_IAB] =
+         (_3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD | IAB_MODIFY_ENABLE |
+          IAB_MODIFY_FUNC | IAB_MODIFY_SRC_FACTOR | IAB_MODIFY_DST_FACTOR);
+
+      i915->state.Ctx[I915_CTXREG_BLENDCOLOR0] =
+         _3DSTATE_CONST_BLEND_COLOR_CMD;
       i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = 0;
 
    }
@@ -858,11 +893,11 @@ static void i915_init_packets( i915ContextPtr i915 )
       I915_STATECHANGE(i915, I915_UPLOAD_FOG);
       i915->state.Fog[I915_FOGREG_MODE0] = _3DSTATE_FOG_MODE_CMD;
       i915->state.Fog[I915_FOGREG_MODE1] = (FMC1_FOGFUNC_MODIFY_ENABLE |
-					  FMC1_FOGFUNC_VERTEX |
-					  FMC1_FOGINDEX_MODIFY_ENABLE |
-					  FMC1_FOGINDEX_W |
-					  FMC1_C1_C2_MODIFY_ENABLE |
-					  FMC1_DENSITY_MODIFY_ENABLE);
+                                            FMC1_FOGFUNC_VERTEX |
+                                            FMC1_FOGINDEX_MODIFY_ENABLE |
+                                            FMC1_FOGINDEX_W |
+                                            FMC1_C1_C2_MODIFY_ENABLE |
+                                            FMC1_DENSITY_MODIFY_ENABLE);
       i915->state.Fog[I915_FOGREG_COLOR] = _3DSTATE_FOG_COLOR_CMD;
    }
 
@@ -871,66 +906,74 @@ static void i915_init_packets( i915ContextPtr i915 )
       I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
       /* color buffer offset/stride */
       i915->state.Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      i915->state.Buffer[I915_DESTREG_CBUFADDR1] = 
-	 (BUF_3D_ID_COLOR_BACK | 
-	  BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
-	  BUF_3D_USE_FENCE);
-      /*i915->state.Buffer[I915_DESTREG_CBUFADDR2] is the offset */
-
+      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
+      i915->state.Buffer[I915_DESTREG_CBUFADDR1] = (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
+                                                    BUF_3D_USE_FENCE);
 
-      /* depth/Z buffer offset/stride */
       i915->state.Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      i915->state.Buffer[I915_DESTREG_DBUFADDR1] = 
-	 (BUF_3D_ID_DEPTH |
-	  BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
-	  BUF_3D_USE_FENCE);
-      i915->state.Buffer[I915_DESTREG_DBUFADDR2] = screen->depth.offset;
-
+      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
+      i915->state.Buffer[I915_DESTREG_DBUFADDR1] = (BUF_3D_ID_DEPTH | BUF_3D_PITCH(screen->depth.pitch) |       /* pitch in bytes */
+                                                    BUF_3D_USE_FENCE);
 
       i915->state.Buffer[I915_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
 
-      /* color/depth pixel format */
+      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
+#if 0                           /* seems we don't need this */
       switch (screen->fbFormat) {
-      case DV_PF_555:
       case DV_PF_565:
-	 i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					       DSTORG_VERT_BIAS(0x8) | /* .5 */
-					       LOD_PRECLAMP_OGL |
-					       TEX_DEFAULT_COLOR_OGL |
-					       DITHER_FULL_ALWAYS |
-					       screen->fbFormat |
-					       DEPTH_FRMT_16_FIXED);
-	 break;
+         i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |        /* .5 */
+                                                 DSTORG_VERT_BIAS(0x8) |        /* .5 */
+                                                 LOD_PRECLAMP_OGL |
+                                                 TEX_DEFAULT_COLOR_OGL |
+                                                 DITHER_FULL_ALWAYS |
+                                                 screen->fbFormat |
+                                                 DEPTH_FRMT_16_FIXED);
+         break;
       case DV_PF_8888:
-	 i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) | /* .5 */
-					       DSTORG_VERT_BIAS(0x8) | /* .5 */
-					       LOD_PRECLAMP_OGL |
-					       TEX_DEFAULT_COLOR_OGL |
-					       screen->fbFormat |
-					       DEPTH_FRMT_24_FIXED_8_OTHER);
-	 break;
+         i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |        /* .5 */
+                                                 DSTORG_VERT_BIAS(0x8) |        /* .5 */
+                                                 LOD_PRECLAMP_OGL |
+                                                 TEX_DEFAULT_COLOR_OGL |
+                                                 screen->fbFormat |
+                                                 DEPTH_FRMT_24_FIXED_8_OTHER);
+         break;
       }
+#endif
+
 
       /* scissor */
-      i915->state.Buffer[I915_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-						DISABLE_SCISSOR_RECT);
+      i915->state.Buffer[I915_DESTREG_SENABLE] =
+         (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
       i915->state.Buffer[I915_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
       i915->state.Buffer[I915_DESTREG_SR1] = 0;
       i915->state.Buffer[I915_DESTREG_SR2] = 0;
    }
 
 
+#if 0
+   {
+      I915_STATECHANGE(i915, I915_UPLOAD_DEFAULTS);
+      i915->state.Default[I915_DEFREG_C0] = _3DSTATE_DEFAULT_DIFFUSE;
+      i915->state.Default[I915_DEFREG_C1] = 0;
+      i915->state.Default[I915_DEFREG_S0] = _3DSTATE_DEFAULT_SPECULAR;
+      i915->state.Default[I915_DEFREG_S1] = 0;
+      i915->state.Default[I915_DEFREG_Z0] = _3DSTATE_DEFAULT_Z;
+      i915->state.Default[I915_DEFREG_Z1] = 0;
+   }
+#endif
+
+
    /* These will be emitted every at the head of every buffer, unless
     * we get hardware contexts working.
     */
-   i915->state.active = (I915_UPLOAD_PROGRAM | 
-			 I915_UPLOAD_STIPPLE | 
-			 I915_UPLOAD_CTX | 
-			 I915_UPLOAD_BUFFERS | 
-			 I915_UPLOAD_INVARIENT);
+   i915->state.active = (I915_UPLOAD_PROGRAM |
+                         I915_UPLOAD_STIPPLE |
+                         I915_UPLOAD_CTX |
+                         I915_UPLOAD_BUFFERS | I915_UPLOAD_INVARIENT);
 }
 
-void i915InitStateFunctions( struct dd_function_table *functions )
+void
+i915InitStateFunctions(struct dd_function_table *functions)
 {
    functions->AlphaFunc = i915AlphaFunc;
    functions->BlendColor = i915BlendColor;
@@ -957,14 +1000,15 @@ void i915InitStateFunctions( struct dd_function_table *functions )
 }
 
 
-void i915InitState( i915ContextPtr i915 )
+void
+i915InitState(struct i915_context *i915)
 {
    GLcontext *ctx = &i915->intel.ctx;
 
-   i915_init_packets( i915 );
+   i915_init_packets(i915);
 
    _mesa_init_driver_state(ctx);
 
-   memcpy( &i915->initial, &i915->state, sizeof(i915->state) );
+   memcpy(&i915->initial, &i915->state, sizeof(i915->state));
    i915->current = &i915->state;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_tex.c b/src/mesa/drivers/dri/i915/i915_tex.c
index d9609d31933..59e148ca04e 100644
--- a/src/mesa/drivers/dri/i915/i915_tex.c
+++ b/src/mesa/drivers/dri/i915/i915_tex.c
@@ -45,109 +45,25 @@
 
 
 
-
-
-
-/**
- * Allocate space for and load the mesa images into the texture memory block.
- * This will happen before drawing with a new texture, or drawing with a
- * texture after it was swapped out or teximaged again.
- */
-
-intelTextureObjectPtr i915AllocTexObj( struct gl_texture_object *texObj )
+static void
+i915TexEnv(GLcontext * ctx, GLenum target,
+           GLenum pname, const GLfloat * param)
 {
-   i915TextureObjectPtr t = CALLOC_STRUCT( i915_texture_object );
-   if ( !t ) 
-      return NULL;
+   struct i915_context *i915 = I915_CONTEXT(ctx);
 
-   texObj->DriverData = t;
-   t->intel.base.tObj = texObj;
-   t->intel.dirty = I915_UPLOAD_TEX_ALL;
-   make_empty_list( &t->intel.base );
-   return &t->intel;
-}
-
-
-static void i915TexParameter( GLcontext *ctx, GLenum target,
-			     struct gl_texture_object *tObj,
-			     GLenum pname, const GLfloat *params )
-{
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
- 
    switch (pname) {
-   case GL_TEXTURE_MIN_FILTER:
-   case GL_TEXTURE_MAG_FILTER:
-   case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-   case GL_TEXTURE_WRAP_S:
-   case GL_TEXTURE_WRAP_T:
-   case GL_TEXTURE_WRAP_R:
-   case GL_TEXTURE_BORDER_COLOR:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   case GL_TEXTURE_COMPARE_MODE:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-   case GL_TEXTURE_COMPARE_FUNC:
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   case GL_TEXTURE_BASE_LEVEL:
-   case GL_TEXTURE_MAX_LEVEL:
-   case GL_TEXTURE_MIN_LOD:
-   case GL_TEXTURE_MAX_LOD:
-      /* The i915 and its successors can do a lot of this without
-       * reloading the textures.  A project for someone?
-       */
-      intelFlush( ctx );
-      driSwapOutTextureObject( (driTextureObject *) t );
-      t->intel.dirty = I915_UPLOAD_TEX_ALL;
-      break;
-
-   default:
-      return;
-   }
-}
-
-
-static void i915TexEnv( GLcontext *ctx, GLenum target, 
-			GLenum pname, const GLfloat *param )
-{
-   i915ContextPtr i915 = I915_CONTEXT( ctx );
-   GLuint unit = ctx->Texture.CurrentUnit;
-
-   switch (pname) {
-   case GL_TEXTURE_ENV_COLOR: 	/* Should be a tracked param */
-   case GL_TEXTURE_ENV_MODE:
-   case GL_COMBINE_RGB:
-   case GL_COMBINE_ALPHA:
-   case GL_SOURCE0_RGB:
-   case GL_SOURCE1_RGB:
-   case GL_SOURCE2_RGB:
-   case GL_SOURCE0_ALPHA:
-   case GL_SOURCE1_ALPHA:
-   case GL_SOURCE2_ALPHA:
-   case GL_OPERAND0_RGB:
-   case GL_OPERAND1_RGB:
-   case GL_OPERAND2_RGB:
-   case GL_OPERAND0_ALPHA:
-   case GL_OPERAND1_ALPHA:
-   case GL_OPERAND2_ALPHA:
-   case GL_RGB_SCALE:
-   case GL_ALPHA_SCALE:
-      i915->tex_program.translated = 0; 
-      break;
-
-   case GL_TEXTURE_LOD_BIAS: {
-      int b = (int) ((*param) * 16.0);
-      if (b > 255) b = 255;
-      if (b < -256) b = -256;
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS2] &= ~SS2_LOD_BIAS_MASK;
-      i915->state.Tex[unit][I915_TEXREG_SS2] |= 
-	 ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK);
-      break;
-   }
+   case GL_TEXTURE_LOD_BIAS:{
+         GLuint unit = ctx->Texture.CurrentUnit;
+         GLint b = (int) ((*param) * 16.0);
+         if (b > 255)
+            b = 255;
+         if (b < -256)
+            b = -256;
+         I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
+         i915->lodbias_ss2[unit] =
+            ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK);
+         break;
+      }
 
    default:
       break;
@@ -155,33 +71,8 @@ static void i915TexEnv( GLcontext *ctx, GLenum target,
 }
 
 
-static void i915BindTexture( GLcontext *ctx, GLenum target,
-			    struct gl_texture_object *texObj )
-{
-   i915TextureObjectPtr tex;
-   
-   if (!texObj->DriverData)
-      i915AllocTexObj( texObj );
-   
-   tex = (i915TextureObjectPtr)texObj->DriverData;
-
-   if (tex->lastTarget != texObj->Target) {
-      tex->intel.dirty = I915_UPLOAD_TEX_ALL;
-      tex->lastTarget = texObj->Target;
-   }
-
-   /* Need this if image format changes between bound textures.
-    * Could try and shortcircuit by checking for differences in
-    * state between incoming and outgoing textures:
-    */
-   I915_CONTEXT(ctx)->tex_program.translated = 0; 
-}
-
-
-
-void i915InitTextureFuncs( struct dd_function_table *functions )
+void
+i915InitTextureFuncs(struct dd_function_table *functions)
 {
-   functions->BindTexture = i915BindTexture;
    functions->TexEnv = i915TexEnv;
-   functions->TexParameter = i915TexParameter;
 }
diff --git a/src/mesa/drivers/dri/i915tex/i915_tex_layout.c b/src/mesa/drivers/dri/i915/i915_tex_layout.c
index 7b761a7b221..7b761a7b221 100644
--- a/src/mesa/drivers/dri/i915tex/i915_tex_layout.c
+++ b/src/mesa/drivers/dri/i915/i915_tex_layout.c
diff --git a/src/mesa/drivers/dri/i915/i915_texprog.c b/src/mesa/drivers/dri/i915/i915_texprog.c
deleted file mode 100644
index f6a8b0205a6..00000000000
--- a/src/mesa/drivers/dri/i915/i915_texprog.c
+++ /dev/null
@@ -1,676 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include <strings.h>
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "tnl/t_context.h"
-#include "intel_batchbuffer.h"
-
-#include "i915_reg.h"
-#include "i915_context.h"
-#include "i915_program.h"
-
-static GLuint translate_tex_src_bit( struct i915_fragment_program *p,
-				     GLubyte bit )
-{
-   switch (bit) {
-   case TEXTURE_1D_BIT:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_2D_BIT:   return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_RECT_BIT: return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_3D_BIT:   return D0_SAMPLE_TYPE_VOLUME;
-   case TEXTURE_CUBE_BIT: return D0_SAMPLE_TYPE_CUBE;
-   default: i915_program_error(p, "TexSrcBit"); return 0;
-   }
-}
-
-static GLuint get_source( struct i915_fragment_program *p, 
-			  GLenum src, GLuint unit )
-{
-   switch (src) {
-   case GL_TEXTURE: 
-      if (p->src_texture == UREG_BAD) {
-
-	 /* TODO: Use D0_CHANNEL_XY where possible.
-	  */
-	 GLuint dim = translate_tex_src_bit( p, p->ctx->Texture.Unit[unit]._ReallyEnabled);
-	 GLuint sampler = i915_emit_decl(p, REG_TYPE_S, unit, dim);
-	 GLuint texcoord = i915_emit_decl(p, REG_TYPE_T, unit, D0_CHANNEL_ALL);
-	 GLuint tmp = i915_get_temp( p );
-	 GLuint op = T0_TEXLD;
-
-	 if (p->VB->TexCoordPtr[unit]->size == 4)
-	    op = T0_TEXLDP;
-
-	 p->src_texture = i915_emit_texld( p, tmp, A0_DEST_CHANNEL_ALL, 
-					  sampler, texcoord, op );
-      }
-
-      return p->src_texture;
-
-      /* Crossbar: */
-   case GL_TEXTURE0:
-   case GL_TEXTURE1:
-   case GL_TEXTURE2:
-   case GL_TEXTURE3:
-   case GL_TEXTURE4:
-   case GL_TEXTURE5:
-   case GL_TEXTURE6:
-   case GL_TEXTURE7: {
-      return UREG_BAD;
-   }
-
-   case GL_CONSTANT:
-      return i915_emit_const4fv( p, p->ctx->Texture.Unit[unit].EnvColor );
-   case GL_PRIMARY_COLOR:
-      return i915_emit_decl(p, REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL);
-   case GL_PREVIOUS:
-   default: 
-      i915_emit_decl(p, 
-		GET_UREG_TYPE(p->src_previous),
-		GET_UREG_NR(p->src_previous), D0_CHANNEL_ALL); 
-      return p->src_previous;
-   }
-}
-			
-
-static GLuint emit_combine_source( struct i915_fragment_program *p, 
-				   GLuint mask,
-				   GLuint unit,
-				   GLenum source, 
-				   GLenum operand )
-{
-   GLuint arg, src;
-
-   src = get_source(p, source, unit);
-
-   switch (operand) {
-   case GL_ONE_MINUS_SRC_COLOR: 
-      /* Get unused tmp,
-       * Emit tmp = 1.0 + arg.-x-y-z-w
-       */
-      arg = i915_get_temp( p );
-      return i915_emit_arith( p, A0_ADD, arg, mask, 0,
-		  swizzle(src, ONE, ONE, ONE, ONE ),
-		  negate(src, 1,1,1,1), 0);
-
-   case GL_SRC_ALPHA: 
-      if (mask == A0_DEST_CHANNEL_W)
-	 return src;
-      else
-	 return swizzle( src, W, W, W, W );
-   case GL_ONE_MINUS_SRC_ALPHA: 
-      /* Get unused tmp,
-       * Emit tmp = 1.0 + arg.-w-w-w-w
-       */
-      arg = i915_get_temp( p );
-      return i915_emit_arith( p, A0_ADD, arg, mask, 0,
-			 swizzle(src, ONE, ONE, ONE, ONE ),
-			 negate( swizzle(src,W,W,W,W), 1,1,1,1), 0);
-   case GL_SRC_COLOR: 
-   default:
-      return src;
-   }
-}
-
-
-
-static int nr_args( GLenum mode )
-{
-   switch (mode) {
-   case GL_REPLACE: return 1; 
-   case GL_MODULATE: return 2;
-   case GL_ADD: return 2;
-   case GL_ADD_SIGNED: return 2;
-   case GL_INTERPOLATE:	return 3;
-   case GL_SUBTRACT: return 2;
-   case GL_DOT3_RGB_EXT: return 2;
-   case GL_DOT3_RGBA_EXT: return 2;
-   case GL_DOT3_RGB: return 2;
-   case GL_DOT3_RGBA: return 2;
-   default: return 0;
-   }
-}
-
-
-static GLboolean args_match( struct gl_texture_unit *texUnit )
-{
-   int i, nr = nr_args(texUnit->Combine.ModeRGB);
-
-   for (i = 0 ; i < nr ; i++) {
-      if (texUnit->Combine.SourceA[i] != texUnit->Combine.SourceRGB[i]) 
-	 return GL_FALSE;
-
-      switch(texUnit->Combine.OperandA[i]) {
-      case GL_SRC_ALPHA: 
-	 switch(texUnit->Combine.OperandRGB[i]) {
-	 case GL_SRC_COLOR: 
-	 case GL_SRC_ALPHA: 
-	    break;
-	 default:
-	    return GL_FALSE;
-	 }
-	 break;
-      case GL_ONE_MINUS_SRC_ALPHA: 
-	 switch(texUnit->Combine.OperandRGB[i]) {
-	 case GL_ONE_MINUS_SRC_COLOR: 
-	 case GL_ONE_MINUS_SRC_ALPHA: 
-	    break;
-	 default:
-	    return GL_FALSE;
-	 }
-	 break;
-      default: 
-	 return GL_FALSE;	/* impossible */
-      }
-   }
-
-   return GL_TRUE;
-}
-
-
-static GLuint emit_combine( struct i915_fragment_program *p,
-			    GLuint dest,
-			    GLuint mask,
-			    GLuint saturate,
-			    GLuint unit,
-			    GLenum mode,
-			    const GLenum *source,
-			    const GLenum *operand)
-{
-   int tmp, src[3], nr = nr_args(mode);
-   int i;
-
-   for (i = 0; i < nr; i++)
-      src[i] = emit_combine_source( p, mask, unit, source[i], operand[i] );
-
-   switch (mode) {
-   case GL_REPLACE: 
-      if (mask == A0_DEST_CHANNEL_ALL && !saturate)
-	 return src[0];
-      else
-	 return i915_emit_arith( p, A0_MOV, dest, mask, saturate, src[0], 0, 0 );
-   case GL_MODULATE: 
-      return i915_emit_arith( p, A0_MUL, dest, mask, saturate,
-			     src[0], src[1], 0 );
-   case GL_ADD: 
-      return i915_emit_arith( p, A0_ADD, dest, mask, saturate, 
-			     src[0], src[1], 0 );
-   case GL_ADD_SIGNED:
-      /* tmp = arg0 + arg1
-       * result = tmp + -.5
-       */
-      tmp = i915_emit_const1f(p, .5);
-      tmp = negate(swizzle(tmp,X,X,X,X),1,1,1,1);
-      i915_emit_arith( p, A0_ADD, dest, mask, 0, src[0], src[1], 0 );
-      i915_emit_arith( p, A0_ADD, dest, mask, saturate, dest, tmp, 0 );
-      return dest;
-   case GL_INTERPOLATE:		/* TWO INSTRUCTIONS */
-      /* Arg0 * (Arg2) + Arg1 * (1-Arg2)
-       *
-       * Arg0*Arg2 + Arg1 - Arg1Arg2 
-       *
-       * tmp = Arg0*Arg2 + Arg1, 
-       * result = (-Arg1)Arg2 + tmp 
-       */
-      tmp = i915_get_temp( p );
-      i915_emit_arith( p, A0_MAD, tmp, mask, 0, src[0], src[2], src[1] );
-      i915_emit_arith( p, A0_MAD, dest, mask, saturate, 
-		      negate(src[1], 1,1,1,1), src[2], tmp );
-      return dest;
-   case GL_SUBTRACT: 
-      /* negate src[1] */
-      return i915_emit_arith( p, A0_ADD, dest, mask, saturate, src[0],
-			 negate(src[1],1,1,1,1), 0 );
-
-   case GL_DOT3_RGBA:
-   case GL_DOT3_RGBA_EXT: 
-   case GL_DOT3_RGB_EXT:
-   case GL_DOT3_RGB: {
-      GLuint tmp0 = i915_get_temp( p );
-      GLuint tmp1 = i915_get_temp( p );
-      GLuint neg1 = negate(swizzle(i915_emit_const1f(p, 1),X,X,X,X), 1,1,1,1);
-      GLuint two = swizzle(i915_emit_const1f(p, 2),X,X,X,X);
-      i915_emit_arith( p, A0_MAD, tmp0, A0_DEST_CHANNEL_ALL, 0, 
-		      two, src[0], neg1);
-      if (src[0] == src[1])
-	 tmp1 = tmp0;
-      else
-	 i915_emit_arith( p, A0_MAD, tmp1, A0_DEST_CHANNEL_ALL, 0, 
-			 two, src[1], neg1);
-      i915_emit_arith( p, A0_DP3, dest, mask, saturate, tmp0, tmp1, 0);
-      return dest;
-   }
-
-   default: 
-      return src[0];
-   }
-}
-
-static GLuint get_dest( struct i915_fragment_program *p, int unit )
-{
-   if (p->ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)
-      return i915_get_temp( p );
-   else if (unit != p->last_tex_stage)
-      return i915_get_temp( p );
-   else
-      return UREG(REG_TYPE_OC, 0);
-}
-      
-
-
-static GLuint emit_texenv( struct i915_fragment_program *p, int unit )
-{
-   struct gl_texture_unit *texUnit = &p->ctx->Texture.Unit[unit];
-   GLenum envMode = texUnit->EnvMode;
-   struct gl_texture_object *tObj = texUnit->_Current;
-   GLenum format = tObj->Image[0][tObj->BaseLevel]->_BaseFormat;
-   GLuint saturate = unit < p->last_tex_stage ? A0_DEST_SATURATE : 0;
-
-   switch(envMode) {
-   case GL_BLEND: {
-      const int cf = get_source(p, GL_PREVIOUS, unit);
-      const int cc = get_source(p, GL_CONSTANT, unit);
-      const int cs = get_source(p, GL_TEXTURE, unit);
-      const int out = get_dest(p, unit);
-
-      if (format == GL_INTENSITY) {
-	 /* cv = cf(1 - cs) + cc.cs
-	  * cv = cf - cf.cs + cc.cs
-	  */
-	 /* u[2] = MAD( -cf * cs + cf )
-	  * cv   = MAD( cc * cs + u[2] )
-	  */
-	 
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0, 
-			 negate(cf,1,1,1,1), cs, cf );
-
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cc, cs, out );
-
-	 return out;
-      } else {
-	 /* cv = cf(1 - cs) + cc.cs
-	  * cv = cf - cf.cs + cc.cs
-	  * av =      af.as
-	  */
-	 /* u[2] = MAD( cf.-x-y-zw * cs.xyzw + cf.xyz0 )
-	  * oC   = MAD( cc.xyz0 * cs.xyz0 + u[2].xyzw )
-	  */
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0,
-			 negate(cf,1,1,1,0),  
-			 cs,
-			 swizzle(cf,X,Y,Z,ZERO) );
-
-
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cc,X,Y,Z,ZERO),  
-			 swizzle(cs,X,Y,Z,ZERO),
-			 out );
-
-	 return out;
-      }
-   }
-
-   case GL_DECAL: {
-      if (format == GL_RGB ||
-	  format == GL_RGBA) {
-	 int cf = get_source( p, GL_PREVIOUS, unit );
-	 int cs = get_source( p, GL_TEXTURE, unit );
-	 int out = get_dest(p, unit);
-	 
-	 /* cv = cf(1-as) + cs.as
-	  * cv = cf.(-as) + cf + cs.as
-	  * av = af
-	  */ 
-	 
-	 /* u[2] = mad( cf.xyzw * cs.-w-w-w1 + cf.xyz0 )
-	  * oc = mad( cs.xyz0 * cs.www0 + u[2].xyzw )
-	  */
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, 0,
-			 cf,  
-			 negate(swizzle(cs,W,W,W,ONE),1,1,1,0),
-			 swizzle(cf,X,Y,Z,ZERO) );
-	 
-	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cs,X,Y,Z,ZERO),  
-			 swizzle(cs,W,W,W,ZERO),
-			 out );
-	 return out;
-      }
-      else {
-	 return get_source( p, GL_PREVIOUS, unit );
-      }
-   }
-
-   case GL_REPLACE: {
-      const int cs = get_source( p, GL_TEXTURE, unit );	/* saturated */
-      switch (format) {
-      case GL_ALPHA: {
-	 const int cf = get_source( p, GL_PREVIOUS, unit ); /* saturated */
-	 i915_emit_arith( p, A0_MOV, cs, A0_DEST_CHANNEL_XYZ, 0, cf, 0, 0 );
-	 return cs;
-      }
-      case GL_RGB:
-      case GL_LUMINANCE: {
-	 const int cf = get_source( p, GL_PREVIOUS, unit ); /* saturated */
-	 i915_emit_arith( p, A0_MOV, cs, A0_DEST_CHANNEL_W, 0, cf, 0, 0 );
-	 return cs;
-      }
-      default:
-	 return cs;
-      }
-   }
-
-   case GL_MODULATE: {
-      const int cf = get_source( p, GL_PREVIOUS, unit );
-      const int cs = get_source( p, GL_TEXTURE, unit );
-      const int out = get_dest(p, unit);
-      switch (format) {
-      case GL_ALPHA: 
-	 i915_emit_arith( p, A0_MUL, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cs, ONE, ONE, ONE, W), cf, 0 );
-	 break;
-      default:
-	 i915_emit_arith( p, A0_MUL, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cs, cf, 0 );
-	 break;
-      }
-      return out;
-   }
-   case GL_ADD: {
-      int cf = get_source( p, GL_PREVIOUS, unit );
-      int cs = get_source( p, GL_TEXTURE, unit );
-      const int out = get_dest( p, unit );
-
-      if (format == GL_INTENSITY) {
-	 /* output-color.rgba = add( incoming, u[1] )
-	  */
-	 i915_emit_arith( p, A0_ADD, out, A0_DEST_CHANNEL_ALL, saturate, 
-			 cs, cf, 0 );
-	 return out;
-      }
-      else {
-	 /* cv.xyz = cf.xyz + cs.xyz
-	  * cv.w   = cf.w * cs.w
-	  *
-	  * cv.xyzw = MAD( cf.111w * cs.xyzw + cf.xyz0 )
-	  */
- 	 i915_emit_arith( p, A0_MAD, out, A0_DEST_CHANNEL_ALL, saturate,
-			 swizzle(cf,ONE,ONE,ONE,W), 
-			 cs,  
-			 swizzle(cf,X,Y,Z,ZERO) ); 
-	 return out;
-      }
-      break;
-   }
-   case GL_COMBINE: {
-      GLuint rgb_shift, alpha_shift, out, shift;
-      GLuint dest = get_dest(p, unit);
-
-      /* The EXT version of the DOT3 extension does not support the
-       * scale factor, but the ARB version (and the version in OpenGL
-       * 1.3) does.
-       */
-      switch (texUnit->Combine.ModeRGB) {
-      case GL_DOT3_RGB_EXT:
-	 alpha_shift = texUnit->Combine.ScaleShiftA;
-	 rgb_shift = 0;
-	 break;
-
-      case GL_DOT3_RGBA_EXT:
-	 alpha_shift = 0;
-	 rgb_shift = 0;
-	 break;
-
-      default:
-	 rgb_shift = texUnit->Combine.ScaleShiftRGB;
-	 alpha_shift = texUnit->Combine.ScaleShiftA;
-	 break;
-      }
-
-
-      /* Emit the RGB and A combine ops
-       */
-      if (texUnit->Combine.ModeRGB == texUnit->Combine.ModeA && 
-	  args_match( texUnit )) {
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_ALL, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-      }
-      else if (texUnit->Combine.ModeRGB == GL_DOT3_RGBA_EXT ||
-	       texUnit->Combine.ModeRGB == GL_DOT3_RGBA) {
-
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_ALL, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-      }
-      else {
-	 /* Need to do something to stop from re-emitting identical
-	  * argument calculations here:
-	  */
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_XYZ, saturate,
-			     unit,
-			     texUnit->Combine.ModeRGB,
-			     texUnit->Combine.SourceRGB,
-			     texUnit->Combine.OperandRGB );
-	 out = emit_combine( p, dest, A0_DEST_CHANNEL_W, saturate,
-			     unit,
-			     texUnit->Combine.ModeA,
-			     texUnit->Combine.SourceA,
-			     texUnit->Combine.OperandA );
-      }
-
-      /* Deal with the final shift:
-       */
-      if (alpha_shift || rgb_shift) {
-	 if (rgb_shift == alpha_shift) {
-	    shift = i915_emit_const1f(p, 1<<rgb_shift);
-	    shift = swizzle(shift,X,X,X,X);
-	 }
-	 else {
-	    shift = i915_emit_const2f(p, 1<<rgb_shift, 1<<alpha_shift);
-	    shift = swizzle(shift,X,X,X,Y);
-	 }
-	 return i915_emit_arith( p, A0_MUL, dest, A0_DEST_CHANNEL_ALL, 
-				saturate, out, shift, 0 );
-      }
-
-      return out;
-   }
-
-   default:
-      return get_source(p, GL_PREVIOUS, 0);
-   }
-}
-
-static void emit_program_fini( struct i915_fragment_program *p )
-{
-   int cf = get_source( p, GL_PREVIOUS, 0 );
-   int out = UREG( REG_TYPE_OC, 0 );
-
-   if (p->ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) {
-      /* Emit specular add.
-       */
-      GLuint s = i915_emit_decl(p, REG_TYPE_T, T_SPECULAR, D0_CHANNEL_ALL);
-      i915_emit_arith( p, A0_ADD, out, A0_DEST_CHANNEL_ALL, 0, cf, 
-		  swizzle(s, X,Y,Z,ZERO), 0 );
-   }
-   else if (cf != out) {
-      /* Will wind up in here if no texture enabled or a couple of
-       * other scenarios (GL_REPLACE for instance).
-       */
-      i915_emit_arith( p, A0_MOV, out, A0_DEST_CHANNEL_ALL, 0, cf, 0, 0 );
-   }
-}
-
-
-static void i915EmitTextureProgram( i915ContextPtr i915 )
-{
-   GLcontext *ctx = &i915->intel.ctx;
-   struct i915_fragment_program *p = &i915->tex_program;
-   GLuint unit;
-
-   if (0) fprintf(stderr, "%s\n", __FUNCTION__);
-
-   i915_init_program( i915, p );
-
-   if (ctx->Texture._EnabledUnits) {
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-	    p->last_tex_stage = unit;
-	 }
-
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++)
-	 if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-	    p->src_previous = emit_texenv( p, unit );
-	    p->src_texture = UREG_BAD;
-	    p->temp_flag = 0xffff000;
-	    p->temp_flag |= 1 << GET_UREG_NR(p->src_previous);
-	 }
-   }
-
-   emit_program_fini( p );
-
-   i915_fini_program( p );
-   i915_upload_program( i915, p );
-
-   p->translated = 1;
-}
-
-
-void i915ValidateTextureProgram( i915ContextPtr i915 )
-{
-   intelContextPtr intel = &i915->intel;
-   GLcontext *ctx = &intel->ctx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-   DECLARE_RENDERINPUTS(index_bitset);
-   int i, offset;
-   GLuint s4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_VFMT_MASK;
-   GLuint s2 = S2_TEXCOORD_NONE;
-
-   RENDERINPUTS_COPY( index_bitset, tnl->render_inputs_bitset );
-
-   /* Important:
-    */
-   VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
-   intel->vertex_attr_count = 0;
-   intel->coloroffset = 0;
-   intel->specoffset = 0;
-   offset = 0;
-
-   if (i915->current_program) {
-      i915->current_program->on_hardware = 0;
-      i915->current_program->params_uptodate = 0;
-   }
-
-   if (i915->vertex_fog == I915_FOG_PIXEL) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
-      RENDERINPUTS_CLEAR( index_bitset, _TNL_ATTRIB_FOG );
-   }
-   else if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16 );
-   }
-   else {
-      EMIT_ATTR( _TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12 );
-   }
-
-   /* How undefined is undefined? */
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_POINTSIZE )) {
-      EMIT_ATTR( _TNL_ATTRIB_POINTSIZE, EMIT_1F, S4_VFMT_POINT_WIDTH, 4 );
-   }
-      
-   intel->coloroffset = offset / 4;
-   EMIT_ATTR( _TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4 );
-            
-   if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 ) ||
-       RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG )) {
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_COLOR1 )) {
-	 intel->specoffset = offset / 4;
-	 EMIT_ATTR( _TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3 );
-      } else 
-	 EMIT_PAD( 3 );
-      
-      if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_FOG ))
-	 EMIT_ATTR( _TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1 );
-      else
-	 EMIT_PAD( 1 );
-   }
-
-   if (RENDERINPUTS_TEST_RANGE( index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX )) {
-      for (i = 0; i < 8; i++) {
-	 if (RENDERINPUTS_TEST( index_bitset, _TNL_ATTRIB_TEX(i) )) {
-	    int sz = VB->TexCoordPtr[i]->size;
-	    
-	    s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-	    s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
-
-	    EMIT_ATTR( _TNL_ATTRIB_TEX0+i, EMIT_SZ(sz), 0, sz * 4 );
-	 }
-      }
-   }
-
-   /* Only need to change the vertex emit code if there has been a
-    * statechange to a new hardware vertex format:
-    */
-   if (s2 != i915->state.Ctx[I915_CTXREG_LIS2] ||
-       s4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-    
-      I915_STATECHANGE( i915, I915_UPLOAD_CTX );
-
-      i915->tex_program.translated = 0;
-
-      /* Must do this *after* statechange, so as not to affect
-       * buffered vertices reliant on the old state:
-       */
-      intel->vertex_size = _tnl_install_attrs( ctx, 
-					       intel->vertex_attrs, 
-					       intel->vertex_attr_count,
-					       intel->ViewportMatrix.m, 0 ); 
-
-      intel->vertex_size >>= 2;
-
-      i915->state.Ctx[I915_CTXREG_LIS2] = s2;
-      i915->state.Ctx[I915_CTXREG_LIS4] = s4;
-
-      assert(intel->vtbl.check_vertex_size( intel, intel->vertex_size ));
-   }
-
-   if (!i915->tex_program.translated ||
-       i915->last_ReallyEnabled != ctx->Texture._EnabledUnits) {
-      i915EmitTextureProgram( i915 );      
-      i915->last_ReallyEnabled = ctx->Texture._EnabledUnits;
-   }
-}
diff --git a/src/mesa/drivers/dri/i915/i915_texstate.c b/src/mesa/drivers/dri/i915/i915_texstate.c
index a19d4b65840..09684e87b4e 100644
--- a/src/mesa/drivers/dri/i915/i915_texstate.c
+++ b/src/mesa/drivers/dri/i915/i915_texstate.c
@@ -25,902 +25,335 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
-#include "macros.h"
 #include "mtypes.h"
-#include "simple_list.h"
 #include "enums.h"
 #include "texformat.h"
-#include "texstore.h"
+#include "dri_bufmgr.h"
 
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_ioctl.h"
+#include "intel_mipmap_tree.h"
 #include "intel_tex.h"
 
 #include "i915_context.h"
 #include "i915_reg.h"
 
-static GLint initial_offsets[6][2] = { {0,0},
-				       {0,2},
-				       {1,0},
-				       {1,2},
-				       {1,1},
-				       {1,3} };
-
-
-static GLint step_offsets[6][2] = { {0,2},
-				    {0,2},
-				    {-1,2},
-				    {-1,2},
-				    {-1,1},
-				    {-1,1} };
-
-
-#define I915_TEX_UNIT_ENABLED(unit)		(1<<unit)
-
-static void i915LayoutTextureImages( i915ContextPtr i915,
-				     struct gl_texture_object *tObj )
-{
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   GLint firstLevel, lastLevel, numLevels;
-   GLint i, total_height, pitch;
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
-
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
-
-
-
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      pitch = dim * t->intel.texelBytes;
-      pitch *= 2;		/* double pitch for cube layouts */
-      pitch = (pitch + 3) & ~3;
-      
-      total_height = dim * 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    if (!t->intel.image[face][i].image) {
-	       fprintf(stderr, "no image %d %d\n", face, i);
-	       break;		/* can't happen */
-	    }
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    x += step_offsets[face][0] * d;
-	    y += step_offsets[face][1] * d;
-	 }
-      }
-      break;
-   }
-   case GL_TEXTURE_3D: {
-      GLuint virtual_height;
-      GLuint tmp_numLevels = numLevels;
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      /* Calculate the size of a single slice.  Hardware demands a
-       * minimum of 8 mipmaps, some of which might ultimately not be
-       * used:
-       */
-      if (tmp_numLevels < 9)
-	 tmp_numLevels = 9;
-
-      virtual_height = tObj->Image[0][firstLevel]->Height;
-
-      for ( total_height = i = 0 ; i < tmp_numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (t->intel.image[0][i].image) {
-	    t->intel.image[0][i].offset = total_height * pitch;
-	    t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 }
-
-	 total_height += MAX2(2, virtual_height);
-	 virtual_height >>= 1;
-      }
-
-      t->intel.depth_pitch = total_height * pitch;
-
-      /* Multiply slice size by texture depth for total size.  It's
-       * remarkable how wasteful of memory all the i8x0 texture
-       * layouts are.
-       */
-      total_height *= t->intel.image[0][0].image->Depth;
-      break;
-   }
-   default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-
-      for ( total_height = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 if (t->intel.image[0][i].image->IsCompressed) {
-	    total_height += (t->intel.image[0][i].image->Height + 3) / 4;
-	 }
-	 else
-	   total_height += MAX2(2, t->intel.image[0][i].image->Height);
-      }
-      break;
-   }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = total_height*pitch;
-   t->intel.max_level = numLevels-1;
-}
-
-
-static void i945LayoutTextureImages( i915ContextPtr i915,
-				    struct gl_texture_object *tObj )
-{
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   GLint firstLevel, lastLevel, numLevels;
-   GLint i, total_height, pitch, sz, max_offset = 0, offset;
-
-
-   /* Compute which mipmap levels we really want to send to the hardware.
-    */
-   driCalculateTextureFirstLastLevel( (driTextureObject *) t );
-
-   /* Figure out the amount of memory required to hold all the mipmap
-    * levels.  Choose the smallest pitch to accomodate the largest
-    * mipmap:
-    */
-   firstLevel = t->intel.base.firstLevel;
-   lastLevel = t->intel.base.lastLevel;
-   numLevels = lastLevel - firstLevel + 1;
-
-
-
-   /* All images must be loaded at this pitch.  Count the number of
-    * lines required:
-    */
-   switch (tObj->Target) {
-   case GL_TEXTURE_CUBE_MAP: {
-      const GLuint dim = tObj->Image[0][firstLevel]->Width;
-      GLuint face;
-
-      /* Depending on the size of the largest images, pitch can be
-       * determined either by the old-style packing of cubemap faces,
-       * or the final row of 4x4, 2x2 and 1x1 faces below this. 
-       */
-      if (dim > 32) {
-	 pitch = dim * t->intel.texelBytes;
-	 pitch *= 2;		/* double pitch for cube layouts */
-	 pitch = (pitch + 3) & ~3;
-      }
-      else {
-	 pitch = 14 * 8 * t->intel.texelBytes; /* determined by row of
-						* little maps at
-						* bottom */
-      }
-      
-      total_height = dim * 4 + 4;
-
-      for ( face = 0 ; face < 6 ; face++) {
-	 GLuint x = initial_offsets[face][0] * dim;
-	 GLuint y = initial_offsets[face][1] * dim;
-	 GLuint d = dim;
-	 
-	 if (dim == 4 && face >= 4) {
-	    y = total_height - 4;
-	    x = (face - 4) * 8;
-	 }
-	 else if (dim < 4) {
-	    y = total_height - 4;
-	    x = face * 8;
-	 }
-
-	 t->intel.base.dirty_images[face] = ~0;
-
-	 assert(tObj->Image[face][firstLevel]->Width == dim);
-	 assert(tObj->Image[face][firstLevel]->Height == dim);
-
-	 for (i = 0; i < numLevels; i++) {
-
-
-	    t->intel.image[face][i].image = tObj->Image[face][firstLevel + i];
-	    assert(t->intel.image[face][i].image);
-	 
-	    t->intel.image[face][i].offset = 
-	       y * pitch + x * t->intel.texelBytes;
-	    t->intel.image[face][i].internalFormat = baseImage->_BaseFormat;
-
-	    d >>= 1;
-	    
-	    switch (d) {
-	    case 4:
-	       switch (face) {
-	       case FACE_POS_X:
-	       case FACE_NEG_X:
-		  x += step_offsets[face][0] * d;
-		  y += step_offsets[face][1] * d;
-		  break;
-	       case FACE_POS_Y:
-	       case FACE_NEG_Y:
-		  y += 12;
-		  x -= 8;
-		  break;
-	       case FACE_POS_Z:
-	       case FACE_NEG_Z:
-		  y = total_height - 4;
-		  x = (face - 4) * 8;
-		  break;
-	       }
-
-	    case 2:
-	       y = total_height - 4;
-	       x = 16 + face * 8;
-	       break;
-
-	    case 1:
-	       x += 48;
-	       break;
-	       
-	    default:
-	       x += step_offsets[face][0] * d;
-	       y += step_offsets[face][1] * d;
-	       break;
-	    }
-	 }
-      }
-      max_offset = total_height * pitch;
-      break;
-   }
-   case GL_TEXTURE_3D: {
-      GLuint depth_packing = 0, depth_pack_pitch;
-      GLuint tmp_numLevels = numLevels;
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      depth_pack_pitch = pitch;
-      
-      t->intel.base.dirty_images[0] = ~0;
-
-
-      for ( total_height = i = 0 ; i < tmp_numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-
-	 
-	 t->intel.image[0][i].offset = total_height * pitch;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-	 
-
-
-	 total_height += MAX2(2, t->intel.image[0][i].image->Height) * 
-	    MAX2((t->intel.image[0][i].image->Depth >> depth_packing), 1);
-
-	 /* When alignment dominates, can't increase depth packing?
-	  * Or does pitch grow???  What are the alignment constraints,
-	  * anyway?
-	  */
-	 if (depth_pack_pitch > 4) {
-	    depth_packing++;
-	    depth_pack_pitch <<= 2;
-	 }
-      }
-
-      max_offset = total_height * pitch;
-      break;
-   }
-   default:
-      pitch = tObj->Image[0][firstLevel]->Width * t->intel.texelBytes;
-      pitch = (pitch + 3) & ~3;
-      t->intel.base.dirty_images[0] = ~0;
-      max_offset = 0;
-
-      for ( offset = i = 0 ; i < numLevels ; i++ ) {
-	 t->intel.image[0][i].image = tObj->Image[0][firstLevel + i];
-	 if (!t->intel.image[0][i].image) 
-	    break;
-	 
-	 t->intel.image[0][i].offset = offset;
-	 t->intel.image[0][i].internalFormat = baseImage->_BaseFormat;
-
-	 if (t->intel.image[0][i].image->IsCompressed)
-	    sz = MAX2(1, t->intel.image[0][i].image->Height/4) * pitch;
-	 else
-	    sz = MAX2(2, t->intel.image[0][i].image->Height) * pitch;
-	 
-	 /* Because the images are packed better, the final offset
-	  * might not be the maximal one:
-	  */
-	 max_offset = MAX2(max_offset, offset + sz);
-
-	 /* LPT change: step right after second mipmap.
-	  */
-	 if (i == 1) 
-	    offset += pitch / 2;
-	 else 
-	    offset += sz;
 
-      }
-      break;
-   }
-
-   t->intel.Pitch = pitch;
-   t->intel.base.totalSize = max_offset;
-   t->intel.max_level = numLevels-1;
-}
-
-
-
-
-static void i915SetTexImages( i915ContextPtr i915, 
-			     struct gl_texture_object *tObj )
+static GLuint
+translate_texture_format(GLuint mesa_format)
 {
-   GLuint textureFormat;
-   i915TextureObjectPtr t = (i915TextureObjectPtr) tObj->DriverData;
-   const struct gl_texture_image *baseImage = tObj->Image[0][tObj->BaseLevel];
-   GLint ss2 = 0;
-
-   switch( baseImage->TexFormat->MesaFormat ) {
+   switch (mesa_format) {
    case MESA_FORMAT_L8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_L8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_L8;
    case MESA_FORMAT_I8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_I8;
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_I8;
    case MESA_FORMAT_A8:
-      t->intel.texelBytes = 1;
-      textureFormat = MAPSURF_8BIT | MT_8BIT_A8; 
-      break;
-
+      return MAPSURF_8BIT | MT_8BIT_A8;
    case MESA_FORMAT_AL88:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_AY88;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_AY88;
    case MESA_FORMAT_RGB565:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_RGB565;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_RGB565;
    case MESA_FORMAT_ARGB1555:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB1555;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
    case MESA_FORMAT_ARGB4444:
-      t->intel.texelBytes = 2;
-      textureFormat = MAPSURF_16BIT | MT_16BIT_ARGB4444;
-      break;
-
+      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
    case MESA_FORMAT_ARGB8888:
-      t->intel.texelBytes = 4;
-      textureFormat = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-      break;
-
+      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
    case MESA_FORMAT_YCBCR_REV:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_NORMAL);
-      ss2 |= SS2_COLORSPACE_CONVERSION;
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
    case MESA_FORMAT_YCBCR:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_422 | MT_422_YCRCB_SWAPY);
-      ss2 |= SS2_COLORSPACE_CONVERSION;
-      break;
-
+      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
    case MESA_FORMAT_RGB_FXT1:
    case MESA_FORMAT_RGBA_FXT1:
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
    case MESA_FORMAT_Z16:
-      t->intel.texelBytes = 2;
-      textureFormat = (MAPSURF_16BIT | MT_16BIT_L16);
-      break;
-
+      return (MAPSURF_16BIT | MT_16BIT_L16);
    case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-     /* 
-      * DXTn pitches are Width/4 * blocksize in bytes 
-      * for DXT1: blocksize=8 so Width/4*8 = Width * 2 
-      * for DXT3/5: blocksize=16 so Width/4*16 = Width * 4
-      */
-     t->intel.texelBytes = 2;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
    case MESA_FORMAT_RGBA_DXT3:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-     break;
-
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
    case MESA_FORMAT_RGBA_DXT5:
-     t->intel.texelBytes = 4;
-     textureFormat = (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-     break;
-
-#if 0
+      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
    case MESA_FORMAT_Z24_S8:
-      t->intel.texelBytes = 4;
-      textureFormat = (MAPSURF_32BIT | MT_32BIT_xL824);
-      break;
-#endif
-
+      return (MAPSURF_32BIT | MT_32BIT_xL824);
    default:
-      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__,
-	      baseImage->TexFormat->MesaFormat);
+      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
       abort();
+      return 0;
    }
+}
 
-   switch (i915->intel.intelScreen->deviceID) {
-   case PCI_CHIP_I945_G:
-   case PCI_CHIP_I945_GM:
-   case PCI_CHIP_I945_GME:
-   case PCI_CHIP_G33_G:
-   case PCI_CHIP_Q33_G:
-   case PCI_CHIP_Q35_G:
-       i945LayoutTextureImages( i915, tObj );
-       break;
-   default:
-       i915LayoutTextureImages( i915, tObj );
-       break;
-   }
-
-   t->Setup[I915_TEXREG_MS3] = 
-      (((tObj->Image[0][t->intel.base.firstLevel]->Height - 1) << MS3_HEIGHT_SHIFT) |
-       ((tObj->Image[0][t->intel.base.firstLevel]->Width - 1) << MS3_WIDTH_SHIFT) |
-       textureFormat |
-       MS3_USE_FENCE_REGS);
-
-   t->Setup[I915_TEXREG_MS4] = 
-      ((((t->intel.Pitch / 4) - 1) << MS4_PITCH_SHIFT) | 
-       MS4_CUBE_FACE_ENA_MASK |
-       (((t->intel.max_level * 4)) << MS4_MAX_LOD_SHIFT) |
-       ((tObj->Image[0][t->intel.base.firstLevel]->Depth - 1) << MS4_VOLUME_DEPTH_SHIFT));
-
-   t->Setup[I915_TEXREG_SS2] &= ~(SS2_COLORSPACE_CONVERSION);
-   t->Setup[I915_TEXREG_SS2] |= ss2;
-
-   t->intel.dirty = I915_UPLOAD_TEX_ALL;
 
-}
 
 
 /* The i915 (and related graphics cores) do not support GL_CLAMP.  The
  * Intel drivers for "other operating systems" implement GL_CLAMP as
  * GL_CLAMP_TO_EDGE, so the same is done here.
  */
-static GLuint translate_wrap_mode( GLenum wrap )
+static GLuint
+translate_wrap_mode(GLenum wrap)
 {
-   switch( wrap ) {
-   case GL_REPEAT: return TEXCOORDMODE_WRAP;
-   case GL_CLAMP:  return TEXCOORDMODE_CLAMP_EDGE; /* not quite correct */
-   case GL_CLAMP_TO_EDGE: return TEXCOORDMODE_CLAMP_EDGE;
-   case GL_CLAMP_TO_BORDER: return TEXCOORDMODE_CLAMP_BORDER;
-   case GL_MIRRORED_REPEAT: return TEXCOORDMODE_MIRROR;
-   default: return TEXCOORDMODE_WRAP;
-   }
-}
-
-
-/**
- */
-static void i915ImportTexObjState( struct gl_texture_object *texObj )
-{   
-   i915TextureObjectPtr t = (i915TextureObjectPtr)texObj->DriverData;
-   int minFilt = 0, mipFilt = 0, magFilt = 0, shadow = 0;
-
-   if(INTEL_DEBUG&DEBUG_DRI)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   switch (texObj->MinFilter) {
-   case GL_NEAREST:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_NONE;
-      break;
-   case GL_LINEAR:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_NONE;
-      break;
-   case GL_NEAREST_MIPMAP_NEAREST:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_NEAREST;
-      break;
-   case GL_LINEAR_MIPMAP_NEAREST:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_NEAREST;
-      break;
-   case GL_NEAREST_MIPMAP_LINEAR:
-      minFilt = FILTER_NEAREST;
-      mipFilt = MIPFILTER_LINEAR;
-      break;
-   case GL_LINEAR_MIPMAP_LINEAR:
-      minFilt = FILTER_LINEAR;
-      mipFilt = MIPFILTER_LINEAR;
-      break;
+   switch (wrap) {
+   case GL_REPEAT:
+      return TEXCOORDMODE_WRAP;
+   case GL_CLAMP:
+      return TEXCOORDMODE_CLAMP_EDGE;   /* not quite correct */
+   case GL_CLAMP_TO_EDGE:
+      return TEXCOORDMODE_CLAMP_EDGE;
+   case GL_CLAMP_TO_BORDER:
+      return TEXCOORDMODE_CLAMP_BORDER;
+   case GL_MIRRORED_REPEAT:
+      return TEXCOORDMODE_MIRROR;
    default:
-      break;
-   }
-
-   if ( texObj->MaxAnisotropy > 1.0 ) {
-      minFilt = FILTER_ANISOTROPIC; 
-      magFilt = FILTER_ANISOTROPIC;
-   }
-   else {
-      switch (texObj->MagFilter) {
-      case GL_NEAREST:
-	 magFilt = FILTER_NEAREST;
-	 break;
-      case GL_LINEAR:
-	 magFilt = FILTER_LINEAR;
-	 break;
-      default:
-	 break;
-      }  
-   }
-
-   if (texObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB && 
-       texObj->Target != GL_TEXTURE_3D) {
-
-      shadow = SS2_SHADOW_ENABLE;
-      shadow |= intel_translate_compare_func( texObj->CompareFunc );
-      
-      minFilt = FILTER_4X4_FLAT;
-      magFilt = FILTER_4X4_FLAT;
-   }
-
-
-   t->Setup[I915_TEXREG_SS2] &= ~(SS2_MIN_FILTER_MASK |
-				 SS2_MIP_FILTER_MASK |
-				 SS2_MAG_FILTER_MASK |
-				 SS2_SHADOW_ENABLE |
-				 SS2_SHADOW_FUNC_MASK);
-   t->Setup[I915_TEXREG_SS2] |= ((minFilt << SS2_MIN_FILTER_SHIFT) |
-				(mipFilt << SS2_MIP_FILTER_SHIFT) |
-				(magFilt << SS2_MAG_FILTER_SHIFT) |
-				shadow);
-
-   {
-      GLuint ss3 = t->Setup[I915_TEXREG_SS3] & ~(SS3_TCX_ADDR_MODE_MASK |
-						SS3_TCY_ADDR_MODE_MASK |
-						SS3_TCZ_ADDR_MODE_MASK);
-      GLenum ws = texObj->WrapS;
-      GLenum wt = texObj->WrapT;
-      GLenum wr = texObj->WrapR;
-      
-      t->refs_border_color = 0;
-
-      if (texObj->Target == GL_TEXTURE_3D &&
-	  (texObj->MinFilter != GL_NEAREST ||
-	   texObj->MagFilter != GL_NEAREST)) {
-	 
-	 /* Try to mimic GL_CLAMP functionality a little better -
-	  * switch to CLAMP_TO_BORDER whenever a non-NEAREST filter is
-	  * in use.  Only do this for 3D textures at the moment --
-	  * doing it universally would fix the conform texbc.c
-	  * failure, though.
-	  */
-	 if (ws == GL_CLAMP) ws = GL_CLAMP_TO_BORDER;
-	 if (wt == GL_CLAMP) wt = GL_CLAMP_TO_BORDER;
-	 if (wr == GL_CLAMP) wr = GL_CLAMP_TO_BORDER;
-
-	 /* 3D textures don't seem to respect the border color.
-	  * Fallback if there's ever a danger that they might refer to
-	  * it.
-	  */
-	 if (ws == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-	 if (wt == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-	 if (wr == GL_CLAMP_TO_BORDER) t->refs_border_color = 1;
-      }
-
-      ss3 |= translate_wrap_mode(ws) << SS3_TCX_ADDR_MODE_SHIFT;
-      ss3 |= translate_wrap_mode(wt) << SS3_TCY_ADDR_MODE_SHIFT;
-      ss3 |= translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT;
-   
-      if (ss3 != t->Setup[I915_TEXREG_SS3]) {
-	 t->intel.dirty = I915_UPLOAD_TEX_ALL;
-	 t->Setup[I915_TEXREG_SS3] = ss3;
-      }
-   }
-
-   {   
-      const GLubyte *color = texObj->_BorderChan;
-
-      t->Setup[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(color[0],color[1],
-						     color[2],color[3]);
+      return TEXCOORDMODE_WRAP;
    }
 }
 
 
 
-static void i915_import_tex_unit( i915ContextPtr i915, 
-				 i915TextureObjectPtr t,
-				 GLuint unit )
+/* Recalculate all state from scratch.  Perhaps not the most
+ * efficient, but this has gotten complex enough that we need
+ * something which is understandable and reliable.
+ */
+static GLboolean
+i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
 {
-   GLuint state[I915_TEX_SETUP_SIZE];
+   GLcontext *ctx = &intel->ctx;
+   struct i915_context *i915 = i915_context(ctx);
+   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
+   struct intel_texture_object *intelObj = intel_texture_object(tObj);
+   struct gl_texture_image *firstImage;
+   GLuint *state = i915->state.Tex[unit], format, pitch;
 
-   if(INTEL_DEBUG&DEBUG_TEXTURE)
-      fprintf(stderr, "%s unit(%d)\n", __FUNCTION__, unit);
-   
-   if (i915->intel.CurrentTexObj[unit]) 
-      i915->intel.CurrentTexObj[unit]->base.bound &= ~(1U << unit);
+   memset(state, 0, sizeof(state));
 
-   i915->intel.CurrentTexObj[unit] = (intelTextureObjectPtr)t;
-   t->intel.base.bound |= (1 << unit);
+   /*We need to refcount these. */
 
-   if (t->intel.dirty & I915_UPLOAD_TEX(unit)) {
-      i915ImportTexObjState( t->intel.base.tObj );
-      t->intel.dirty &= ~I915_UPLOAD_TEX(unit);
+   if (i915->state.tex_buffer[unit] != NULL) {
+       dri_bo_unreference(i915->state.tex_buffer[unit]);
+       i915->state.tex_buffer[unit] = NULL;
    }
 
-   state[I915_TEXREG_MS2] = t->intel.TextureOffset;
-   state[I915_TEXREG_MS3] = t->Setup[I915_TEXREG_MS3];
-   state[I915_TEXREG_MS4] = t->Setup[I915_TEXREG_MS4];
-
-   state[I915_TEXREG_SS2] = (i915->state.Tex[unit][I915_TEXREG_SS2] &
-			    SS2_LOD_BIAS_MASK);
-   state[I915_TEXREG_SS2] |= (t->Setup[I915_TEXREG_SS2] & ~SS2_LOD_BIAS_MASK);
+   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
+      return GL_FALSE;
 
-   state[I915_TEXREG_SS3] = (i915->state.Tex[unit][I915_TEXREG_SS3] &
-			    SS3_NORMALIZED_COORDS);
-   state[I915_TEXREG_SS3] |= (t->Setup[I915_TEXREG_SS3] &
-			     ~(SS3_NORMALIZED_COORDS|
-			       SS3_TEXTUREMAP_INDEX_MASK));
+   /* Get first image here, since intelObj->firstLevel will get set in
+    * the intel_finalize_mipmap_tree() call above.
+    */
+   firstImage = tObj->Image[0][intelObj->firstLevel];
 
-   state[I915_TEXREG_SS3] |= (unit<<SS3_TEXTUREMAP_INDEX_SHIFT);
+   if (intelObj->imageOverride) {
+      i915->state.tex_buffer[unit] = NULL;
+      i915->state.tex_offset[unit] = intelObj->textureOffset;
 
-   state[I915_TEXREG_SS4] = t->Setup[I915_TEXREG_SS4];
+      switch (intelObj->depthOverride) {
+      case 32:
+	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
+	 break;
+      case 24:
+      default:
+	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
+	 break;
+      case 16:
+	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
+	 break;
+      }
 
+      pitch = intelObj->pitchOverride;
+   } else {
+      dri_bo_reference(intelObj->mt->region->buffer);
+      i915->state.tex_buffer[unit] = intelObj->mt->region->buffer;
+      i915->state.tex_offset[unit] =  intel_miptree_image_offset(intelObj->mt,
+								 0, intelObj->
+								 firstLevel);
 
-   if (memcmp(state, i915->state.Tex[unit], sizeof(state)) != 0) {
-      I915_STATECHANGE( i915, I915_UPLOAD_TEX(unit) );
-      memcpy(i915->state.Tex[unit], state, sizeof(state));
+      format = translate_texture_format(firstImage->TexFormat->MesaFormat);
+      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
    }
-}
-
 
+   state[I915_TEXREG_MS3] =
+      (((firstImage->Height - 1) << MS3_HEIGHT_SHIFT) |
+       ((firstImage->Width - 1) << MS3_WIDTH_SHIFT) | format |
+       MS3_USE_FENCE_REGS);
 
-static GLboolean enable_tex_common( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-
-   if (0) fprintf(stderr, "%s %d\n", __FUNCTION__, unit);
+   state[I915_TEXREG_MS4] =
+     ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | MS4_CUBE_FACE_ENA_MASK |
+       ((((intelObj->lastLevel - intelObj->firstLevel) * 4)) <<
+	MS4_MAX_LOD_SHIFT) | ((firstImage->Depth - 1) <<
+			      MS4_VOLUME_DEPTH_SHIFT));
 
-   if (!(i915->state.active & I915_UPLOAD_TEX(unit))) {
-      I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_TRUE);
-   }
 
-   /* Fallback if there's a texture border */
-   if ( tObj->Image[0][tObj->BaseLevel]->Border > 0 ) {
-      return GL_FALSE;
-   }
+   {
+      GLuint minFilt, mipFilt, magFilt;
 
+      switch (tObj->MinFilter) {
+      case GL_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NONE;
+         break;
+      case GL_NEAREST_MIPMAP_NEAREST:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_LINEAR_MIPMAP_NEAREST:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_NEAREST;
+         break;
+      case GL_NEAREST_MIPMAP_LINEAR:
+         minFilt = FILTER_NEAREST;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      case GL_LINEAR_MIPMAP_LINEAR:
+         minFilt = FILTER_LINEAR;
+         mipFilt = MIPFILTER_LINEAR;
+         break;
+      default:
+         return GL_FALSE;
+      }
 
-   /* Update state if this is a different texture object to last
-    * time.
-    */
-   if (i915->intel.CurrentTexObj[unit] != &t->intel || 
-       (t->intel.dirty & I915_UPLOAD_TEX(unit))) {
-      i915_import_tex_unit( i915, t, unit);
-      i915->tex_program.translated = 0;
-   }
+      if (tObj->MaxAnisotropy > 1.0) {
+         minFilt = FILTER_ANISOTROPIC;
+         magFilt = FILTER_ANISOTROPIC;
+      }
+      else {
+         switch (tObj->MagFilter) {
+         case GL_NEAREST:
+            magFilt = FILTER_NEAREST;
+            break;
+         case GL_LINEAR:
+            magFilt = FILTER_LINEAR;
+            break;
+         default:
+            return GL_FALSE;
+         }
+      }
 
-   return GL_TRUE;
-}
+      state[I915_TEXREG_SS2] = i915->lodbias_ss2[unit];
 
-static GLboolean enable_tex_rect( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
+      /* YUV conversion:
+       */
+      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
+          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
+         state[I915_TEXREG_SS2] |= SS2_COLORSPACE_CONVERSION;
 
-   ss3 &= ~SS3_NORMALIZED_COORDS;
+      /* Shadow:
+       */
+      if (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB &&
+          tObj->Target != GL_TEXTURE_3D) {
 
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
-   }
+         state[I915_TEXREG_SS2] |=
+            (SS2_SHADOW_ENABLE |
+             intel_translate_compare_func(tObj->CompareFunc));
 
-   /* Upload teximages (not pipelined)
-    */
-   if (t->intel.base.dirty_images[0]) {
-      i915SetTexImages( i915, tObj );
-      if (!intelUploadTexImages( &i915->intel, &t->intel, 0 )) {
-	 return GL_FALSE;
+         minFilt = FILTER_4X4_FLAT;
+         magFilt = FILTER_4X4_FLAT;
       }
-   }
-
-   return GL_TRUE;
-}
 
+      state[I915_TEXREG_SS2] |= ((minFilt << SS2_MIN_FILTER_SHIFT) |
+                                 (mipFilt << SS2_MIP_FILTER_SHIFT) |
+                                 (magFilt << SS2_MAG_FILTER_SHIFT));
+   }
 
-static GLboolean enable_tex_2d( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
+   {
+      GLenum ws = tObj->WrapS;
+      GLenum wt = tObj->WrapT;
+      GLenum wr = tObj->WrapR;
 
-   ss3 |= SS3_NORMALIZED_COORDS;
 
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
-   }
-
-   /* Upload teximages (not pipelined)
-    */
-   if (t->intel.base.dirty_images[0]) {
-      i915SetTexImages( i915, tObj );
-      if (!intelUploadTexImages( &i915->intel, &t->intel, 0 )) {
-	 return GL_FALSE;
-      }
-   }
+      /* 3D textures don't seem to respect the border color.
+       * Fallback if there's ever a danger that they might refer to
+       * it.  
+       * 
+       * Effectively this means fallback on 3D clamp or
+       * clamp_to_border.
+       */
+      if (tObj->Target == GL_TEXTURE_3D &&
+          (tObj->MinFilter != GL_NEAREST ||
+           tObj->MagFilter != GL_NEAREST) &&
+          (ws == GL_CLAMP ||
+           wt == GL_CLAMP ||
+           wr == GL_CLAMP ||
+           ws == GL_CLAMP_TO_BORDER ||
+           wt == GL_CLAMP_TO_BORDER || wr == GL_CLAMP_TO_BORDER))
+         return GL_FALSE;
 
-   return GL_TRUE;
-}
 
-static GLboolean enable_tex_cube( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-   struct gl_texture_object *tObj = texUnit->_Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
-   GLuint ss3 = i915->state.Tex[unit][I915_TEXREG_SS3];
-   GLuint face;
-
-   ss3 |= SS3_NORMALIZED_COORDS;
-
-   if (ss3 != i915->state.Tex[unit][I915_TEXREG_SS3]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-      i915->state.Tex[unit][I915_TEXREG_SS3] = ss3;
-   }
+      state[I915_TEXREG_SS3] = ss3;     /* SS3_NORMALIZED_COORDS */
 
-   /* Upload teximages (not pipelined)
-    */
-   if ( t->intel.base.dirty_images[0] || t->intel.base.dirty_images[1] ||
-        t->intel.base.dirty_images[2] || t->intel.base.dirty_images[3] ||
-        t->intel.base.dirty_images[4] || t->intel.base.dirty_images[5] ) {
-      i915SetTexImages( i915, tObj );
-   }
+      state[I915_TEXREG_SS3] |=
+         ((translate_wrap_mode(ws) << SS3_TCX_ADDR_MODE_SHIFT) |
+          (translate_wrap_mode(wt) << SS3_TCY_ADDR_MODE_SHIFT) |
+          (translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT));
 
-   /* upload (per face) */
-   for (face = 0; face < 6; face++) {
-      if (t->intel.base.dirty_images[face]) {
-	 if (!intelUploadTexImages( &i915->intel, &t->intel, face )) {
-	    return GL_FALSE;
-	 }
-      }
+      state[I915_TEXREG_SS3] |= (unit << SS3_TEXTUREMAP_INDEX_SHIFT);
    }
 
 
-   return GL_TRUE;
-}
+   state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
+                                                tObj->_BorderChan[1],
+                                                tObj->_BorderChan[2],
+                                                tObj->_BorderChan[3]);
 
-static GLboolean enable_tex_3d( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
-   i915TextureObjectPtr t = (i915TextureObjectPtr)tObj->DriverData;
 
-   /* 3D textures on I915 seem to get bogus border colors, hence this
-    * fallback:
+   I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_TRUE);
+   /* memcmp was already disabled, but definitely won't work as the
+    * region might now change and that wouldn't be detected:
     */
-   if (t->refs_border_color)
-      return GL_FALSE;
+   I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
 
-   return GL_TRUE;
-}
 
-
-
- 
-static GLboolean disable_tex( GLcontext *ctx, GLuint unit )
-{
-   i915ContextPtr i915 = I915_CONTEXT(ctx);
-
-   if (i915->state.active & I915_UPLOAD_TEX(unit)) {
-      I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_FALSE);
-   }
-
-   /* The old texture is no longer bound to this texture unit.
-    * Mark it as such.
-    */
-   if ( i915->intel.CurrentTexObj[unit] != NULL ) {
-      i915->intel.CurrentTexObj[unit]->base.bound &= ~(1U << 0);
-      i915->intel.CurrentTexObj[unit] = NULL;
-   }
+#if 0
+   DBG(TEXTURE, "state[I915_TEXREG_SS2] = 0x%x\n", state[I915_TEXREG_SS2]);
+   DBG(TEXTURE, "state[I915_TEXREG_SS3] = 0x%x\n", state[I915_TEXREG_SS3]);
+   DBG(TEXTURE, "state[I915_TEXREG_SS4] = 0x%x\n", state[I915_TEXREG_SS4]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS2] = 0x%x\n", state[I915_TEXREG_MS2]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS3] = 0x%x\n", state[I915_TEXREG_MS3]);
+   DBG(TEXTURE, "state[I915_TEXREG_MS4] = 0x%x\n", state[I915_TEXREG_MS4]);
+#endif
 
    return GL_TRUE;
 }
 
-static GLboolean i915UpdateTexUnit( GLcontext *ctx, GLuint unit )
-{
-   struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 
-   if (texUnit->_ReallyEnabled &&
-       INTEL_CONTEXT(ctx)->intelScreen->tex.size < 2048 * 1024)
-      return GL_FALSE;
 
-   switch (texUnit->_ReallyEnabled) {
-   case TEXTURE_1D_BIT:
-   case TEXTURE_2D_BIT:
-      return (enable_tex_2d( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_RECT_BIT:
-      return (enable_tex_rect( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_CUBE_BIT:
-      return (enable_tex_cube( ctx, unit ) &&
-	      enable_tex_common( ctx, unit ));
-   case TEXTURE_3D_BIT:
-       return (enable_tex_2d( ctx, unit ) && 
-	       enable_tex_common( ctx, unit ) &&
-	       enable_tex_3d( ctx, unit)); 
-   case 0:
-      return disable_tex( ctx, unit );
-   default:
-      return GL_FALSE;
-   }
-}
 
-
-void i915UpdateTextureState( intelContextPtr intel )
+void
+i915UpdateTextureState(struct intel_context *intel)
 {
-   GLcontext *ctx = &intel->ctx;
    GLboolean ok = GL_TRUE;
    GLuint i;
 
-   for (i = 0 ; i < I915_TEX_UNITS && ok ; i++) {
-      ok = i915UpdateTexUnit( ctx, i );
+   for (i = 0; i < I915_TEX_UNITS && ok; i++) {
+      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
+      case TEXTURE_1D_BIT:
+      case TEXTURE_2D_BIT:
+      case TEXTURE_CUBE_BIT:
+      case TEXTURE_3D_BIT:
+         ok = i915_update_tex_unit(intel, i, SS3_NORMALIZED_COORDS);
+         break;
+      case TEXTURE_RECT_BIT:
+         ok = i915_update_tex_unit(intel, i, 0);
+         break;
+      case 0:{
+            struct i915_context *i915 = i915_context(&intel->ctx);
+            if (i915->state.active & I915_UPLOAD_TEX(i))
+               I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(i), GL_FALSE);
+
+	    if (i915->state.tex_buffer[i] != NULL) {
+	       dri_bo_unreference(i915->state.tex_buffer[i]);
+	       i915->state.tex_buffer[i] = NULL;
+	    }
+
+            break;
+         }
+      default:
+         ok = GL_FALSE;
+         break;
+      }
    }
 
-   FALLBACK( intel, I915_FALLBACK_TEXTURE, !ok );
+   FALLBACK(intel, I915_FALLBACK_TEXTURE, !ok);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i915/i915_vtbl.c b/src/mesa/drivers/dri/i915/i915_vtbl.c
index cc8a605e505..35757e17eb9 100644
--- a/src/mesa/drivers/dri/i915/i915_vtbl.c
+++ b/src/mesa/drivers/dri/i915/i915_vtbl.c
@@ -37,115 +37,144 @@
 #include "tnl/t_vertex.h"
 
 #include "intel_batchbuffer.h"
+#include "intel_tex.h"
+#include "intel_regions.h"
 
 #include "i915_reg.h"
 #include "i915_context.h"
 
-static void i915_render_start( intelContextPtr intel )
+static void
+i915_render_prevalidate(struct intel_context *intel)
 {
-   GLcontext *ctx = &intel->ctx;
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
 
-   if (ctx->FragmentProgram._Active) 
-      i915ValidateFragmentProgram( i915 );
-   else {
-      assert(!ctx->FragmentProgram._MaintainTexEnvProgram);
-      i915ValidateTextureProgram( i915 );
-   }
+   i915ValidateFragmentProgram(i915);
+}
+
+static void
+i915_render_start(struct intel_context *intel)
+{
 }
 
 
-static void i915_reduced_primitive_state( intelContextPtr intel,
-					  GLenum rprim )
+static void
+i915_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
 {
-    i915ContextPtr i915 = I915_CONTEXT(intel);
-    GLuint st1 = i915->state.Stipple[I915_STPREG_ST1];
-
-    st1 &= ~ST1_ENABLE;
-
-    switch (rprim) {
-    case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
-    case GL_TRIANGLES:
-       if (intel->ctx.Polygon.StippleFlag &&
-	   intel->hw_stipple)
-	  st1 |= ST1_ENABLE;
-       break;
-    case GL_LINES:
-    case GL_POINTS:
-    default:
-       break;
-    }
-
-    i915->intel.reduced_primitive = rprim;
-
-    if (st1 != i915->state.Stipple[I915_STPREG_ST1]) {
-       I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-       i915->state.Stipple[I915_STPREG_ST1] = st1;
-    }
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   GLuint st1 = i915->state.Stipple[I915_STPREG_ST1];
+
+   st1 &= ~ST1_ENABLE;
+
+   switch (rprim) {
+   case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
+   case GL_TRIANGLES:
+      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
+         st1 |= ST1_ENABLE;
+      break;
+   case GL_LINES:
+   case GL_POINTS:
+   default:
+      break;
+   }
+
+   i915->intel.reduced_primitive = rprim;
+
+   if (st1 != i915->state.Stipple[I915_STPREG_ST1]) {
+      INTEL_FIREVERTICES(intel);
+
+      I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
+      i915->state.Stipple[I915_STPREG_ST1] = st1;
+   }
 }
 
 
 /* Pull apart the vertex format registers and figure out how large a
  * vertex is supposed to be. 
  */
-static GLboolean i915_check_vertex_size( intelContextPtr intel,
-					 GLuint expected )
+static GLboolean
+i915_check_vertex_size(struct intel_context *intel, GLuint expected)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
    int lis2 = i915->current->Ctx[I915_CTXREG_LIS2];
    int lis4 = i915->current->Ctx[I915_CTXREG_LIS4];
    int i, sz = 0;
 
    switch (lis4 & S4_VFMT_XYZW_MASK) {
-   case S4_VFMT_XY: sz = 2; break;
-   case S4_VFMT_XYZ: sz = 3; break;
-   case S4_VFMT_XYW: sz = 3; break;
-   case S4_VFMT_XYZW: sz = 4; break;
-   default: 
+   case S4_VFMT_XY:
+      sz = 2;
+      break;
+   case S4_VFMT_XYZ:
+      sz = 3;
+      break;
+   case S4_VFMT_XYW:
+      sz = 3;
+      break;
+   case S4_VFMT_XYZW:
+      sz = 4;
+      break;
+   default:
       fprintf(stderr, "no xyzw specified\n");
       return 0;
    }
 
-   if (lis4 & S4_VFMT_SPEC_FOG) sz++;
-   if (lis4 & S4_VFMT_COLOR) sz++;
-   if (lis4 & S4_VFMT_DEPTH_OFFSET) sz++;
-   if (lis4 & S4_VFMT_POINT_WIDTH) sz++;
-   if (lis4 & S4_VFMT_FOG_PARAM) sz++;
-	
-   for (i = 0 ; i < 8 ; i++) { 
+   if (lis4 & S4_VFMT_SPEC_FOG)
+      sz++;
+   if (lis4 & S4_VFMT_COLOR)
+      sz++;
+   if (lis4 & S4_VFMT_DEPTH_OFFSET)
+      sz++;
+   if (lis4 & S4_VFMT_POINT_WIDTH)
+      sz++;
+   if (lis4 & S4_VFMT_FOG_PARAM)
+      sz++;
+
+   for (i = 0; i < 8; i++) {
       switch (lis2 & S2_TEXCOORD_FMT0_MASK) {
-      case TEXCOORDFMT_2D: sz += 2; break;
-      case TEXCOORDFMT_3D: sz += 3; break;
-      case TEXCOORDFMT_4D: sz += 4; break;
-      case TEXCOORDFMT_1D: sz += 1; break;
-      case TEXCOORDFMT_2D_16: sz += 1; break;
-      case TEXCOORDFMT_4D_16: sz += 2; break;
-      case TEXCOORDFMT_NOT_PRESENT: break;
+      case TEXCOORDFMT_2D:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_3D:
+         sz += 3;
+         break;
+      case TEXCOORDFMT_4D:
+         sz += 4;
+         break;
+      case TEXCOORDFMT_1D:
+         sz += 1;
+         break;
+      case TEXCOORDFMT_2D_16:
+         sz += 1;
+         break;
+      case TEXCOORDFMT_4D_16:
+         sz += 2;
+         break;
+      case TEXCOORDFMT_NOT_PRESENT:
+         break;
       default:
-	 fprintf(stderr, "bad texcoord fmt %d\n", i);
-	 return GL_FALSE;
+         fprintf(stderr, "bad texcoord fmt %d\n", i);
+         return GL_FALSE;
       }
       lis2 >>= S2_TEXCOORD_FMT1_SHIFT;
    }
-	
-   if (sz != expected) 
+
+   if (sz != expected)
       fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-   
+
    return sz == expected;
 }
 
 
-static void i915_emit_invarient_state( intelContextPtr intel )
+static void
+i915_emit_invarient_state(struct intel_context *intel)
 {
    BATCH_LOCALS;
 
-   BEGIN_BATCH( 20 );
+   BEGIN_BATCH(200, 0);
 
    OUT_BATCH(_3DSTATE_AA_CMD |
-	     AA_LINE_ECAAR_WIDTH_ENABLE |
-	     AA_LINE_ECAAR_WIDTH_1_0 |
-	     AA_LINE_REGION_WIDTH_ENABLE |
-	     AA_LINE_REGION_WIDTH_1_0);
+             AA_LINE_ECAAR_WIDTH_ENABLE |
+             AA_LINE_ECAAR_WIDTH_1_0 |
+             AA_LINE_REGION_WIDTH_ENABLE | AA_LINE_REGION_WIDTH_1_0);
 
    OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
    OUT_BATCH(0);
@@ -158,35 +187,27 @@ static void i915_emit_invarient_state( intelContextPtr intel )
 
    /* Don't support texture crossbar yet */
    OUT_BATCH(_3DSTATE_COORD_SET_BINDINGS |
-	     CSB_TCB(0, 0) |
-	     CSB_TCB(1, 1) |
-	     CSB_TCB(2, 2) |
-	     CSB_TCB(3, 3) |
-	     CSB_TCB(4, 4) |
-	     CSB_TCB(5, 5) |
-	     CSB_TCB(6, 6) |
-	     CSB_TCB(7, 7));
+             CSB_TCB(0, 0) |
+             CSB_TCB(1, 1) |
+             CSB_TCB(2, 2) |
+             CSB_TCB(3, 3) |
+             CSB_TCB(4, 4) | CSB_TCB(5, 5) | CSB_TCB(6, 6) | CSB_TCB(7, 7));
 
    OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-	     ENABLE_POINT_RASTER_RULE |
-	     OGL_POINT_RASTER_RULE |
-	     ENABLE_LINE_STRIP_PROVOKE_VRTX |
-	     ENABLE_TRI_FAN_PROVOKE_VRTX |
-	     LINE_STRIP_PROVOKE_VRTX(1) |
-	     TRI_FAN_PROVOKE_VRTX(2) | 
-	     ENABLE_TEXKILL_3D_4D |
-	     TEXKILL_4D);
+             ENABLE_POINT_RASTER_RULE |
+             OGL_POINT_RASTER_RULE |
+             ENABLE_LINE_STRIP_PROVOKE_VRTX |
+             ENABLE_TRI_FAN_PROVOKE_VRTX |
+             LINE_STRIP_PROVOKE_VRTX(1) |
+             TRI_FAN_PROVOKE_VRTX(2) | ENABLE_TEXKILL_3D_4D | TEXKILL_4D);
 
    /* Need to initialize this to zero.
     */
-   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | 
-	     I1_LOAD_S(3) |
-	     (0));
+   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (0));
    OUT_BATCH(0);
- 
+
    /* XXX: Use this */
-   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | 
-	     DISABLE_SCISSOR_RECT);
+   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
 
    OUT_BATCH(_3DSTATE_SCISSOR_RECT_0_CMD);
    OUT_BATCH(0);
@@ -194,29 +215,23 @@ static void i915_emit_invarient_state( intelContextPtr intel )
 
    OUT_BATCH(_3DSTATE_DEPTH_SUBRECT_DISABLE);
 
-   OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0); /* disable indirect state */
+   OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0);       /* disable indirect state */
    OUT_BATCH(0);
 
 
    /* Don't support twosided stencil yet */
-   OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS |
-	     BFO_ENABLE_STENCIL_TWO_SIDE |
-	     0 );
-   
+   OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS | BFO_ENABLE_STENCIL_TWO_SIDE | 0);
+   OUT_BATCH(0);
+
    ADVANCE_BATCH();
 }
 
 
-#define emit( intel, state, size )			\
-do {							\
-   int k;						\
-   BEGIN_BATCH( (size) / sizeof(GLuint));		\
-   for (k = 0 ; k < (size) / sizeof(GLuint) ; k++)	\
-      OUT_BATCH((state)[k]);				\
-   ADVANCE_BATCH();					\
-} while (0);
+#define emit(intel, state, size )		     \
+   intel_batchbuffer_data(intel->batch, state, size, 0 )
 
-static GLuint get_dirty( struct i915_hw_state *state )
+static GLuint
+get_dirty(struct i915_hw_state *state)
 {
    GLuint dirty;
 
@@ -227,45 +242,45 @@ static GLuint get_dirty( struct i915_hw_state *state )
    if (dirty & I915_UPLOAD_TEX_ALL)
       state->emitted &= ~I915_UPLOAD_TEX_ALL;
    dirty = state->active & ~state->emitted;
-
    return dirty;
 }
 
 
-static GLuint get_state_size( struct i915_hw_state *state )
+static GLuint
+get_state_size(struct i915_hw_state *state)
 {
    GLuint dirty = get_dirty(state);
    GLuint i;
    GLuint sz = 0;
 
    if (dirty & I915_UPLOAD_INVARIENT)
-      sz += 20 * sizeof(int);
+      sz += 30 * 4;
 
    if (dirty & I915_UPLOAD_CTX)
       sz += sizeof(state->Ctx);
 
-   if (dirty & I915_UPLOAD_BUFFERS) 
+   if (dirty & I915_UPLOAD_BUFFERS)
       sz += sizeof(state->Buffer);
 
    if (dirty & I915_UPLOAD_STIPPLE)
       sz += sizeof(state->Stipple);
 
-   if (dirty & I915_UPLOAD_FOG) 
+   if (dirty & I915_UPLOAD_FOG)
       sz += sizeof(state->Fog);
 
    if (dirty & I915_UPLOAD_TEX_ALL) {
       int nr = 0;
-      for (i = 0; i < I915_TEX_UNITS; i++) 
-	 if (dirty & I915_UPLOAD_TEX(i)) 
-	    nr++;
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i))
+            nr++;
 
-      sz += (2+nr*3) * sizeof(GLuint) * 2;
+      sz += (2 + nr * 3) * sizeof(GLuint) * 2;
    }
 
-   if (dirty & I915_UPLOAD_CONSTANTS) 
+   if (dirty & I915_UPLOAD_CONSTANTS)
       sz += state->ConstantSize * sizeof(GLuint);
 
-   if (dirty & I915_UPLOAD_PROGRAM) 
+   if (dirty & I915_UPLOAD_PROGRAM)
       sz += state->ProgramSize * sizeof(GLuint);
 
    return sz;
@@ -274,47 +289,83 @@ static GLuint get_state_size( struct i915_hw_state *state )
 
 /* Push the state into the sarea and/or texture memory.
  */
-static void i915_emit_state( intelContextPtr intel )
+static void
+i915_emit_state(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
+   struct i915_context *i915 = i915_context(&intel->ctx);
    struct i915_hw_state *state = i915->current;
    int i;
-   GLuint dirty = get_dirty(state);
-   GLuint counter = intel->batch.counter;
+   GLuint dirty;
    BATCH_LOCALS;
 
-   if (intel->batch.space < get_state_size(state)) {
-      intelFlushBatch(intel, GL_TRUE);
-      dirty = get_dirty(state);
-      counter = intel->batch.counter;
-   }
+   /* We don't hold the lock at this point, so want to make sure that
+    * there won't be a buffer wrap.  
+    *
+    * It might be better to talk about explicit places where
+    * scheduling is allowed, rather than assume that it is whenever a
+    * batchbuffer fills up.
+    */
+   intel_batchbuffer_require_space(intel->batch, get_state_size(state), 0);
+
+   /* Do this here as we may have flushed the batchbuffer above,
+    * causing more state to be dirty!
+    */
+   dirty = get_dirty(state);
+   state->emitted |= dirty;
 
-   if (VERBOSE) 
+   if (INTEL_DEBUG & DEBUG_STATE)
       fprintf(stderr, "%s dirty: %x\n", __FUNCTION__, dirty);
 
    if (dirty & I915_UPLOAD_INVARIENT) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_INVARIENT:\n"); 
-      i915_emit_invarient_state( intel );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_INVARIENT:\n");
+      i915_emit_invarient_state(intel);
    }
 
    if (dirty & I915_UPLOAD_CTX) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_CTX:\n"); 
-      emit( i915, state->Ctx, sizeof(state->Ctx) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_CTX:\n");
+
+      emit(intel, state->Ctx, sizeof(state->Ctx));
    }
 
    if (dirty & I915_UPLOAD_BUFFERS) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_BUFFERS:\n"); 
-      emit( i915, state->Buffer, sizeof(state->Buffer) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_BUFFERS:\n");
+      BEGIN_BATCH(I915_DEST_SETUP_SIZE + 2, 0);
+      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR1]);
+      OUT_RELOC(state->draw_region->buffer,
+                DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
+                state->draw_region->draw_offset);
+
+      if (state->depth_region) {
+         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR0]);
+         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR1]);
+         OUT_RELOC(state->depth_region->buffer,
+                   DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
+                   state->depth_region->draw_offset);
+      }
+
+      OUT_BATCH(state->Buffer[I915_DESTREG_DV0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_DV1]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SENABLE]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR0]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR1]);
+      OUT_BATCH(state->Buffer[I915_DESTREG_SR2]);
+      ADVANCE_BATCH();
    }
 
    if (dirty & I915_UPLOAD_STIPPLE) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_STIPPLE:\n"); 
-      emit( i915, state->Stipple, sizeof(state->Stipple) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_STIPPLE:\n");
+      emit(intel, state->Stipple, sizeof(state->Stipple));
    }
 
    if (dirty & I915_UPLOAD_FOG) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_FOG:\n"); 
-      emit( i915, state->Fog, sizeof(state->Fog) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_FOG:\n");
+      emit(intel, state->Fog, sizeof(state->Fog));
    }
 
    /* Combine all the dirty texture state into a single command to
@@ -323,141 +374,198 @@ static void i915_emit_state( intelContextPtr intel )
    if (dirty & I915_UPLOAD_TEX_ALL) {
       int nr = 0;
 
-      for (i = 0; i < I915_TEX_UNITS; i++) 
-	 if (dirty & I915_UPLOAD_TEX(i)) 
-	    nr++;
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i))
+            nr++;
 
-      BEGIN_BATCH(2+nr*3);
-      OUT_BATCH(_3DSTATE_MAP_STATE | (3*nr));
+      BEGIN_BATCH(2 + nr * 3, 0);
+      OUT_BATCH(_3DSTATE_MAP_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0 ; i < I915_TEX_UNITS ; i++)
-	 if (dirty & I915_UPLOAD_TEX(i)) {
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS2]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS3]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_MS4]);
-	 }
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i)) {
+
+            if (state->tex_buffer[i]) {
+               OUT_RELOC(state->tex_buffer[i],
+                         DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
+                         state->tex_offset[i]);
+            }
+            else if (state == &i915->meta) {
+               assert(i == 0);
+               OUT_BATCH(0);
+            }
+            else {
+               OUT_BATCH(state->tex_offset[i]);
+            }
+
+            OUT_BATCH(state->Tex[i][I915_TEXREG_MS3]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_MS4]);
+         }
       ADVANCE_BATCH();
 
-      BEGIN_BATCH(2+nr*3);
-      OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3*nr));
+      BEGIN_BATCH(2 + nr * 3, 0);
+      OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3 * nr));
       OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0 ; i < I915_TEX_UNITS ; i++)
-	 if (dirty & I915_UPLOAD_TEX(i)) {
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS2]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS3]);
-	    OUT_BATCH(state->Tex[i][I915_TEXREG_SS4]);
-	 }
+      for (i = 0; i < I915_TEX_UNITS; i++)
+         if (dirty & I915_UPLOAD_TEX(i)) {
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS2]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS3]);
+            OUT_BATCH(state->Tex[i][I915_TEXREG_SS4]);
+         }
       ADVANCE_BATCH();
    }
 
    if (dirty & I915_UPLOAD_CONSTANTS) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_CONSTANTS:\n"); 
-      emit( i915, state->Constant, state->ConstantSize * sizeof(GLuint) );
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_CONSTANTS:\n");
+      emit(intel, state->Constant, state->ConstantSize * sizeof(GLuint));
    }
 
    if (dirty & I915_UPLOAD_PROGRAM) {
-      if (VERBOSE) fprintf(stderr, "I915_UPLOAD_PROGRAM:\n"); 
+      if (INTEL_DEBUG & DEBUG_STATE)
+         fprintf(stderr, "I915_UPLOAD_PROGRAM:\n");
+
+      assert((state->Program[0] & 0x1ff) + 2 == state->ProgramSize);
 
-      assert((state->Program[0] & 0x1ff)+2 == state->ProgramSize);
-      
-      emit( i915, state->Program, state->ProgramSize * sizeof(GLuint) );
-      if (VERBOSE)
-	 i915_disassemble_program( state->Program, state->ProgramSize );
+      emit(intel, state->Program, state->ProgramSize * sizeof(GLuint));
+      if (INTEL_DEBUG & DEBUG_STATE)
+         i915_disassemble_program(state->Program, state->ProgramSize);
    }
 
-   state->emitted |= dirty;
-   intel->batch.last_emit_state = counter;
-   assert(counter == intel->batch.counter);
+   assert(get_dirty(state) == 0);
 }
 
-static void i915_destroy_context( intelContextPtr intel )
+static void
+i915_destroy_context(struct intel_context *intel)
 {
+   GLuint i;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+
+   for (i = 0; i < I915_TEX_UNITS; i++) {
+      if (i915->state.tex_buffer[i] != NULL) {
+	 dri_bo_unreference(i915->state.tex_buffer[i]);
+	 i915->state.tex_buffer[i] = NULL;
+      }
+   }
+
    _tnl_free_vertices(&intel->ctx);
 }
 
 
 /**
- * Set the color buffer drawing region.
+ * Set the drawing regions for the color and depth/stencil buffers.
+ * This involves setting the pitch, cpp and buffer ID/location.
+ * Also set pixel format for color and Z rendering
+ * Used for setting both regular and meta state.
  */
-static void
-i915_set_color_region( intelContextPtr intel, const intelRegion *region)
+void
+i915_state_draw_region(struct intel_context *intel,
+                       struct i915_hw_state *state,
+                       struct intel_region *color_region,
+                       struct intel_region *depth_region)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
-   I915_STATECHANGE( i915, I915_UPLOAD_BUFFERS );
-   i915->state.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_CBUFADDR2] = region->offset;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   GLuint value;
+
+   ASSERT(state == &i915->state || state == &i915->meta);
+
+   if (state->draw_region != color_region) {
+      intel_region_release(&state->draw_region);
+      intel_region_reference(&state->draw_region, color_region);
+   }
+   if (state->depth_region != depth_region) {
+      intel_region_release(&state->depth_region);
+      intel_region_reference(&state->depth_region, depth_region);
+   }
+
+   /*
+    * Set stride/cpp values
+    */
+   if (color_region) {
+      state->Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I915_DESTREG_CBUFADDR1] =
+         (BUF_3D_ID_COLOR_BACK |
+          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   if (depth_region) {
+      state->Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
+      state->Buffer[I915_DESTREG_DBUFADDR1] =
+         (BUF_3D_ID_DEPTH |
+          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
+          BUF_3D_USE_FENCE);
+   }
+
+   /*
+    * Compute/set I915_DESTREG_DV1 value
+    */
+   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
+            DSTORG_VERT_BIAS(0x8) |     /* .5 */
+            LOD_PRECLAMP_OGL | TEX_DEFAULT_COLOR_OGL);
+   if (color_region && color_region->cpp == 4) {
+      value |= DV_PF_8888;
+   }
+   else {
+      value |= (DITHER_FULL_ALWAYS | DV_PF_565);
+   }
+   if (depth_region && depth_region->cpp == 4) {
+      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
+   }
+   else {
+      value |= DEPTH_FRMT_16_FIXED;
+   }
+   state->Buffer[I915_DESTREG_DV1] = value;
+
+   I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
 }
 
 
-/**
- * specify the z-buffer/stencil region
- */
 static void
-i915_set_z_region( intelContextPtr intel, const intelRegion *region)
+i915_set_draw_region(struct intel_context *intel,
+                     struct intel_region *color_region,
+                     struct intel_region *depth_region)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
-   I915_STATECHANGE( i915, I915_UPLOAD_BUFFERS );
-   i915->state.Buffer[I915_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH | BUF_3D_PITCH(region->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_DBUFADDR2] = region->offset;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   i915_state_draw_region(intel, &i915->state, color_region, depth_region);
 }
 
 
-/**
- * Set both the color and Z/stencil drawing regions.
- * Similar to two previous functions, but don't use I915_STATECHANGE()
- */
+
 static void
-i915_update_color_z_regions(intelContextPtr intel,
-                            const intelRegion *colorRegion,
-                            const intelRegion *depthRegion)
+i915_lost_hardware(struct intel_context *intel)
 {
-   i915ContextPtr i915 = I915_CONTEXT(intel);
-
-   i915->state.Buffer[I915_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) | BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_CBUFADDR2] = colorRegion->offset;
-
-   i915->state.Buffer[I915_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH |
-       BUF_3D_PITCH(depthRegion->pitch) |  /* pitch in bytes */
-       BUF_3D_USE_FENCE);
-   i915->state.Buffer[I915_DESTREG_DBUFADDR2] = depthRegion->offset;
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   i915->state.emitted = 0;
 }
 
-
-static void i915_lost_hardware( intelContextPtr intel )
+static GLuint
+i915_flush_cmd(void)
 {
-   I915_CONTEXT(intel)->state.emitted = 0;
+   return MI_FLUSH | FLUSH_MAP_CACHE;
 }
 
-static void i915_emit_flush( intelContextPtr intel )
+static void 
+i915_assert_not_dirty( struct intel_context *intel )
 {
-   BATCH_LOCALS;
-
-   BEGIN_BATCH(2);
-   OUT_BATCH( MI_FLUSH | FLUSH_MAP_CACHE | FLUSH_RENDER_CACHE ); 
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
+   struct i915_context *i915 = i915_context(&intel->ctx);
+   struct i915_hw_state *state = i915->current;
+   GLuint dirty = get_dirty(state);
+   assert(!dirty);
 }
 
 
-void i915InitVtbl( i915ContextPtr i915 )
+void
+i915InitVtbl(struct i915_context *i915)
 {
-   i915->intel.vtbl.alloc_tex_obj = i915AllocTexObj;
    i915->intel.vtbl.check_vertex_size = i915_check_vertex_size;
-   i915->intel.vtbl.clear_with_tris = i915ClearWithTris;
-   i915->intel.vtbl.rotate_window = i915RotateWindow;
    i915->intel.vtbl.destroy = i915_destroy_context;
    i915->intel.vtbl.emit_state = i915_emit_state;
    i915->intel.vtbl.lost_hardware = i915_lost_hardware;
    i915->intel.vtbl.reduced_primitive_state = i915_reduced_primitive_state;
    i915->intel.vtbl.render_start = i915_render_start;
-   i915->intel.vtbl.set_color_region = i915_set_color_region;
-   i915->intel.vtbl.set_z_region = i915_set_z_region;
-   i915->intel.vtbl.update_color_z_regions = i915_update_color_z_regions;
+   i915->intel.vtbl.render_prevalidate = i915_render_prevalidate;
+   i915->intel.vtbl.set_draw_region = i915_set_draw_region;
    i915->intel.vtbl.update_texture_state = i915UpdateTextureState;
-   i915->intel.vtbl.emit_flush = i915_emit_flush;
+   i915->intel.vtbl.flush_cmd = i915_flush_cmd;
+   i915->intel.vtbl.assert_not_dirty = i915_assert_not_dirty;
 }
-
diff --git a/src/mesa/drivers/dri/i915/intel_batchbuffer.c b/src/mesa/drivers/dri/i915/intel_batchbuffer.c
index 803b41b2567..045ff0a5b09 100644
--- a/src/mesa/drivers/dri/i915/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i915/intel_batchbuffer.c
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -25,805 +25,308 @@
  * 
  **************************************************************************/
 
-
-#include <stdio.h>
-#include <errno.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "enums.h"
-#include "vblank.h"
-
-#include "intel_reg.h"
 #include "intel_batchbuffer.h"
-#include "intel_context.h"
-
-
-
-
-/* ================================================================
- * Performance monitoring functions
+#include "intel_ioctl.h"
+#include "intel_decode.h"
+#include "i915_debug.h"
+
+/* Relocations in kernel space:
+ *    - pass dma buffer seperately
+ *    - memory manager knows how to patch
+ *    - pass list of dependent buffers
+ *    - pass relocation list
+ *
+ * Either:
+ *    - get back an offset for buffer to fire
+ *    - memory manager knows how to fire buffer
+ *
+ * Really want the buffer to be AGP and pinned.
+ *
  */
 
-static void intel_fill_box( intelContextPtr intel,
-			    GLshort x, GLshort y,
-			    GLshort w, GLshort h,
-			    GLubyte r, GLubyte g, GLubyte b )
-{
-   x += intel->drawX;
-   y += intel->drawY;
-
-   if (x >= 0 && y >= 0 &&
-       x+w < intel->intelScreen->width &&
-       y+h < intel->intelScreen->height)
-      intelEmitFillBlitLocked( intel, 
-			       intel->intelScreen->cpp,
-			       intel->intelScreen->back.pitch,
-			       intel->intelScreen->back.offset,
-			       x, y, w, h,
-			       INTEL_PACKCOLOR(intel->intelScreen->fbFormat,
-					       r,g,b,0xff));
-}
+/* Cliprect fence: The highest fence protecting a dma buffer
+ * containing explicit cliprect information.  Like the old drawable
+ * lock but irq-driven.  X server must wait for this fence to expire
+ * before changing cliprects [and then doing sw rendering?].  For
+ * other dma buffers, the scheduler will grab current cliprect info
+ * and mix into buffer.  X server must hold the lock while changing
+ * cliprects???  Make per-drawable.  Need cliprects in shared memory
+ * -- beats storing them with every cmd buffer in the queue.
+ *
+ * ==> X server must wait for this fence to expire before touching the
+ * framebuffer with new cliprects.
+ *
+ * ==> Cliprect-dependent buffers associated with a
+ * cliprect-timestamp.  All of the buffers associated with a timestamp
+ * must go to hardware before any buffer with a newer timestamp.
+ *
+ * ==> Dma should be queued per-drawable for correct X/GL
+ * synchronization.  Or can fences be used for this?
+ *
+ * Applies to: Blit operations, metaops, X server operations -- X
+ * server automatically waits on its own dma to complete before
+ * modifying cliprects ???
+ */
 
-static void intel_draw_performance_boxes( intelContextPtr intel )
+void
+intel_batchbuffer_reset(struct intel_batchbuffer *batch)
 {
-   /* Purple box for page flipping
-    */
-   if ( intel->perf_boxes & I830_BOX_FLIP ) 
-      intel_fill_box( intel, 4, 4, 8, 8, 255, 0, 255 );
-
-   /* Red box if we have to wait for idle at any point
-    */
-   if ( intel->perf_boxes & I830_BOX_WAIT ) 
-      intel_fill_box( intel, 16, 4, 8, 8, 255, 0, 0 );
-
-   /* Blue box: lost context?
-    */
-   if ( intel->perf_boxes & I830_BOX_LOST_CONTEXT ) 
-      intel_fill_box( intel, 28, 4, 8, 8, 0, 0, 255 );
-
-   /* Yellow box for texture swaps
-    */
-   if ( intel->perf_boxes & I830_BOX_TEXTURE_LOAD ) 
-      intel_fill_box( intel, 40, 4, 8, 8, 255, 255, 0 );
-
-   /* Green box if hardware never idles (as far as we can tell)
-    */
-   if ( !(intel->perf_boxes & I830_BOX_RING_EMPTY) ) 
-      intel_fill_box( intel, 64, 4, 8, 8, 0, 255, 0 );
+   struct intel_context *intel = batch->intel;
 
-
-   /* Draw bars indicating number of buffers allocated 
-    * (not a great measure, easily confused)
-    */
-#if 0
-   if (intel->dma_used) {
-      int bar = intel->dma_used / 10240;
-      if (bar > 100) bar = 100;
-      if (bar < 1) bar = 1;
-      intel_fill_box( intel, 4, 16, bar, 4, 196, 128, 128 );
-      intel->dma_used = 0;
+   if (batch->buf != NULL) {
+      dri_bo_unreference(batch->buf);
+      batch->buf = NULL;
    }
-#endif
 
-   intel->perf_boxes = 0;
+   batch->buf = dri_bo_alloc(intel->intelScreen->bufmgr, "batchbuffer",
+			     intel->intelScreen->maxBatchSize, 4096,
+			     DRM_BO_FLAG_MEM_TT);
+   dri_bo_map(batch->buf, GL_TRUE);
+   batch->map = batch->buf->virtual;
+   batch->size = intel->intelScreen->maxBatchSize;
+   batch->ptr = batch->map;
 }
 
-
-
-
-
-
-static int bad_prim_vertex_nr( int primitive, int nr )
+struct intel_batchbuffer *
+intel_batchbuffer_alloc(struct intel_context *intel)
 {
-   switch (primitive & PRIM3D_MASK) {
-   case PRIM3D_POINTLIST:
-      return nr < 1;
-   case PRIM3D_LINELIST:
-      return (nr & 1) || nr == 0;
-   case PRIM3D_LINESTRIP:
-      return nr < 2;
-   case PRIM3D_TRILIST:
-   case PRIM3D_RECTLIST:
-      return nr % 3 || nr == 0;
-   case PRIM3D_POLY:
-   case PRIM3D_TRIFAN:
-   case PRIM3D_TRISTRIP:
-   case PRIM3D_TRISTRIP_RVRSE:
-      return nr < 3;
-   default:
-      return 1;
-   }	
-}
+   struct intel_batchbuffer *batch = calloc(sizeof(*batch), 1);
 
-static void intel_flush_inline_primitive( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   GLuint used = intel->batch.ptr - intel->prim.start_ptr;
-   GLuint vertcount;
-
-   assert(intel->prim.primitive != ~0);
-
-   if (1) {
-      /* Check vertex size against the vertex we're specifying to
-       * hardware.  If it's wrong, ditch the primitive.
-       */ 
-      if (!intel->vtbl.check_vertex_size( intel, intel->vertex_size )) 
-	 goto do_discard;
-
-      vertcount = (used - 4)/ (intel->vertex_size * 4);
-
-      if (!vertcount)
-	 goto do_discard;
-      
-      if (vertcount * intel->vertex_size * 4 != used - 4) {
-	 fprintf(stderr, "vertex size confusion %d %d\n", used, 
-		 intel->vertex_size * vertcount * 4);
-	 goto do_discard;
-      }
+   batch->intel = intel;
+   batch->last_fence = NULL;
+   intel_batchbuffer_reset(batch);
 
-      if (bad_prim_vertex_nr( intel->prim.primitive, vertcount )) {
-	 fprintf(stderr, "bad_prim_vertex_nr %x %d\n", intel->prim.primitive,
-		 vertcount);
-	 goto do_discard;
-      }
-   }
-
-   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;
-   intel->batch.space += used;
-   assert(intel->batch.space >= 0);
-
- finished:
-   intel->prim.primitive = ~0;
-   intel->prim.start_ptr = 0;
-   intel->prim.flush = 0;
+   return batch;
 }
 
-
-/* Emit a primitive referencing vertices in a vertex buffer.
- */
-void intelStartInlinePrimitive( intelContextPtr intel, GLuint prim )
+void
+intel_batchbuffer_free(struct intel_batchbuffer *batch)
 {
-   BATCH_LOCALS;
-
-   if (0)
-      fprintf(stderr, "%s %x\n", __FUNCTION__, prim);
-
-
-   /* Finish any in-progress primitive:
-    */
-   INTEL_FIREVERTICES( intel );
-   
-   /* Emit outstanding state:
-    */
-   intel->vtbl.emit_state( intel );
-   
-   /* Make sure there is some space in this buffer:
-    */
-   if (intel->vertex_size * 10 * sizeof(GLuint) >= intel->batch.space) {
-      intelFlushBatch(intel, GL_TRUE); 
-      intel->vtbl.emit_state( intel );
+   if (batch->last_fence) {
+      dri_fence_wait(batch->last_fence);
+      dri_fence_unreference(batch->last_fence);
+      batch->last_fence = NULL;
    }
-
-#if 1
-   if (((unsigned long)intel->batch.ptr) & 0x4) {
-      BEGIN_BATCH(1);
-      OUT_BATCH(0);
-      ADVANCE_BATCH();
+   if (batch->map) {
+      dri_bo_unmap(batch->buf);
+      batch->map = NULL;
    }
-#endif
-
-   /* Emit a slot which will be filled with the inline primitive
-    * command later.
-    */
-   BEGIN_BATCH(2);
-   OUT_BATCH( 0 );
-
-   intel->prim.start_ptr = batch_ptr;
-   intel->prim.primitive = prim;
-   intel->prim.flush = intel_flush_inline_primitive;
-   intel->batch.contains_geometry = 1;
-
-   OUT_BATCH( 0 );
-   ADVANCE_BATCH();
+   dri_bo_unreference(batch->buf);
+   batch->buf = NULL;
+   free(batch);
 }
 
+static int
+relocation_sort(const void *a_in, const void *b_in) {
+   const struct buffer_reloc *a = a_in, *b = b_in;
 
-void intelRestartInlinePrimitive( intelContextPtr intel )
-{
-   GLuint prim = intel->prim.primitive;
-
-   intel_flush_inline_primitive( &intel->ctx );
-   if (1) intelFlushBatch(intel, GL_TRUE); /* GL_TRUE - is critical */
-   intelStartInlinePrimitive( intel, prim );
+   return (intptr_t)a->buf < (intptr_t)b->buf ? -1 : 1;
 }
 
 
-
-void intelWrapInlinePrimitive( intelContextPtr intel )
-{
-   GLuint prim = intel->prim.primitive;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-   intel_flush_inline_primitive( &intel->ctx );
-   intelFlushBatch(intel, GL_TRUE);
-   intelStartInlinePrimitive( intel, prim );
-}
-
-
-/* Emit a primitive with space for inline vertices.
+/* TODO: Push this whole function into bufmgr.
  */
-GLuint *intelEmitInlinePrimitiveLocked(intelContextPtr intel, 
-				       int primitive,
-				       int dwords,
-				       int vertex_size )
+static void
+do_flush_locked(struct intel_batchbuffer *batch,
+                GLuint used,
+                GLboolean ignore_cliprects, GLboolean allow_unlock)
 {
-   GLuint *tmp = 0;
-   BATCH_LOCALS;
-
-   if (0)
-      fprintf(stderr, "%s 0x%x %d\n", __FUNCTION__, primitive, dwords);
-
-   /* Emit outstanding state:
+   GLuint *ptr;
+   GLuint i;
+   struct intel_context *intel = batch->intel;
+   dri_fence *fo;
+   GLboolean performed_rendering = GL_FALSE;
+
+   assert(batch->buf->virtual != NULL);
+   ptr = batch->buf->virtual;
+
+   /* Sort our relocation list in terms of referenced buffer pointer.
+    * This lets us uniquely validate the buffers with the sum of all the flags,
+    * while avoiding O(n^2) on number of relocations.
     */
-   intel->vtbl.emit_state( intel );
-
-   if ((1+dwords)*4 >= intel->batch.space) {
-      intelFlushBatch(intel, GL_TRUE); 
-      intel->vtbl.emit_state( intel );
-   }
-
-
-   if (1) {
-      int used = dwords * 4;
-      int vertcount;
+   qsort(batch->reloc, batch->nr_relocs, sizeof(batch->reloc[0]),
+	 relocation_sort);
 
-      /* Check vertex size against the vertex we're specifying to
-       * hardware.  If it's wrong, ditch the primitive.
-       */ 
-      if (!intel->vtbl.check_vertex_size( intel, vertex_size )) 
-	 goto do_discard;
-
-      vertcount = dwords / vertex_size;
-      
-      if (dwords % vertex_size) {
-	 fprintf(stderr, "did not request a whole number of vertices\n");
-	 goto do_discard;
-      }
+   /* Perform the necessary validations of buffers, and enter the relocations
+    * in the batchbuffer.
+    */
+   for (i = 0; i < batch->nr_relocs; i++) {
+      struct buffer_reloc *r = &batch->reloc[i];
+
+      if (r->validate_flags & DRM_BO_FLAG_WRITE)
+	 performed_rendering = GL_TRUE;
+
+      /* If this is the first time we've seen this buffer in the relocation
+       * list, figure out our flags and validate it.
+       */
+      if (i == 0 || batch->reloc[i - 1].buf != r->buf) {
+	 uint32_t validate_flags;
+	 int j, ret;
+
+	 /* Accumulate the flags we need for validating this buffer. */
+	 validate_flags = r->validate_flags;
+	 for (j = i + 1; j < batch->nr_relocs; j++) {
+	    if (batch->reloc[j].buf != r->buf)
+	       break;
+	    validate_flags |= batch->reloc[j].validate_flags;
+	 }
 
-      if (bad_prim_vertex_nr( primitive, vertcount )) {
-	 fprintf(stderr, "bad_prim_vertex_nr %x %d\n", primitive, vertcount);
-	 goto do_discard;
+	 /* Validate.  If we fail, fence to clear the unfenced list and bail
+	  * out.
+	  */
+	 ret = dri_bo_validate(r->buf, validate_flags);
+	 if (ret != 0) {
+	    dri_bo_unmap(batch->buf);
+	    fo = dri_fence_validated(intel->intelScreen->bufmgr,
+				     "batchbuffer failure fence", GL_TRUE);
+	    dri_fence_unreference(fo);
+	    goto done;
+	 }
       }
-
-      if (used < 8)
-	 goto do_discard;
+      ptr[r->offset / 4] = r->buf->offset + r->delta;
+      dri_bo_unreference(r->buf);
    }
 
-   /* Emit 3D_PRIMITIVE commands:
-    */
-   BEGIN_BATCH(1 + dwords);
-   OUT_BATCH( _3DPRIMITIVE | 
-	      primitive |
-	      (dwords-1) );
-
-   tmp = (GLuint *)batch_ptr;
-   batch_ptr += dwords * 4;
-
-   ADVANCE_BATCH();
-
-   intel->batch.contains_geometry = 1;
-
- do_discard:
-   return tmp;
-}
+   dri_bo_unmap(batch->buf);
+   batch->map = NULL;
+   batch->ptr = NULL;
 
+   dri_bo_validate(batch->buf, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_EXE);
 
-static void intelWaitForFrameCompletion( intelContextPtr intel )
-{
-  drm_i915_sarea_t *sarea = (drm_i915_sarea_t *)intel->sarea;
+   batch->list_count = 0;
+   batch->nr_relocs = 0;
+   batch->flags = 0;
 
-   if (intel->do_irqs) {
-      if (intelGetLastFrame(intel) < sarea->last_dispatch) {
-	 if (!intel->irqsEmitted) {
-	    while (intelGetLastFrame (intel) < sarea->last_dispatch)
-	       ;
-	 }
-	 else {
-	    intelWaitIrq( intel, intel->alloc.irq_emitted );	
-	 }
-	 intel->irqsEmitted = 10;
-      }
+   /* Throw away non-effective packets.  Won't work once we have
+    * hardware contexts which would preserve statechanges beyond a
+    * single buffer.
+    */
 
-      if (intel->irqsEmitted) {
-	 LOCK_HARDWARE( intel ); 
-	 intelEmitIrqLocked( intel );
-	 intel->irqsEmitted--;
-	 UNLOCK_HARDWARE( intel ); 
-      }
-   } 
-   else {
-      while (intelGetLastFrame (intel) < sarea->last_dispatch) {
-	 if (intel->do_usleeps) 
-	    DO_USLEEP( 1 );
-      }
+   if (!(intel->numClipRects == 0 && !ignore_cliprects)) {
+      intel_batch_ioctl(batch->intel,
+                        batch->buf->offset,
+                        used, ignore_cliprects, allow_unlock);
    }
-}
 
-/*
- * Copy the back buffer to the front buffer. 
- */
-void intelCopyBuffer( const __DRIdrawablePrivate *dPriv,
-		      const drm_clip_rect_t	 *rect)
-{
-   intelContextPtr intel;
-   const intelScreenPrivate *intelScreen;
-   GLboolean   missed_target;
-   int64_t ust;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-
-   intelFlush( &intel->ctx );
-   
-   intelScreen = intel->intelScreen;
-
-   if (!rect && !intel->swap_scheduled && intelScreen->drmMinor >= 6 &&
-       !(intel->vblank_flags & VBLANK_FLAG_NO_IRQ) &&
-       intelScreen->current_rotation == 0) {
-      unsigned int interval = driGetVBlankInterval(dPriv, intel->vblank_flags);
-      unsigned int target;
-      drm_i915_vblank_swap_t swap;
-
-      swap.drawable = dPriv->hHWDrawable;
-      swap.seqtype = DRM_VBLANK_ABSOLUTE;
-      target = swap.sequence = intel->vbl_seq + interval;
-
-      if (intel->vblank_flags & VBLANK_FLAG_SYNC) {
-	 swap.seqtype |= DRM_VBLANK_NEXTONMISS;
-      } else if (interval == 0) {
-	 goto noschedule;
-      }
-
-      if ( intel->vblank_flags & VBLANK_FLAG_SECONDARY ) {
-	 swap.seqtype |= DRM_VBLANK_SECONDARY;
-      }
+   /* Associate a fence with the validated buffers, and note that we included
+    * a flush at the end.
+    */
+   fo = dri_fence_validated(intel->intelScreen->bufmgr,
+			    "Batch fence", GL_TRUE);
 
-      if (!drmCommandWriteRead(intel->driFd, DRM_I915_VBLANK_SWAP, &swap,
-                              sizeof(swap))) {
-        intel->swap_scheduled = 1;
-        intel->vbl_seq = swap.sequence;
-        swap.sequence -= target;
-        missed_target = swap.sequence > 0 && swap.sequence <= (1 << 23);
-      }
+   if (performed_rendering) {
+      dri_fence_unreference(batch->last_fence);
+      batch->last_fence = fo;
    } else {
-      intel->swap_scheduled = 0;
+     /* If we didn't validate any buffers for writing by the card, we don't
+      * need to track the fence for glFinish().
+      */
+      dri_fence_unreference(fo);
    }
-noschedule:
-
-   if (!intel->swap_scheduled) {
-      intelWaitForFrameCompletion( intel );
-      LOCK_HARDWARE( intel );
 
-      if (!rect)
-      {
-	 UNLOCK_HARDWARE( intel );
-	 driWaitForVBlank( dPriv, &intel->vbl_seq, intel->vblank_flags, & missed_target );
-	 LOCK_HARDWARE( intel );
-      }
-      {
-	 const intelScreenPrivate *intelScreen = intel->intelScreen;
-	 const __DRIdrawablePrivate *dPriv = intel->driDrawable;
-	 const int nbox = dPriv->numClipRects;
-	 const drm_clip_rect_t *pbox = dPriv->pClipRects;
-	 drm_clip_rect_t box;
-	 const int cpp = intelScreen->cpp;
-	 const int pitch = intelScreen->front.pitch; /* in bytes */
-	 int i;
-	 GLuint CMD, BR13;
-	 BATCH_LOCALS;
-
-	 switch(cpp) {
-	 case 2: 
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24);
-	    CMD = XY_SRC_COPY_BLT_CMD;
-	    break;
-	 case 4:
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24) | (1<<25);
-	    CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-		   XY_SRC_COPY_BLT_WRITE_RGB);
-	    break;
-	 default:
-	    BR13 = (pitch) | (0xCC << 16) | (1<<24);
-	    CMD = XY_SRC_COPY_BLT_CMD;
-	    break;
-	 }
-   
-	 if (0) 
-	    intel_draw_performance_boxes( intel );
-
-	 for (i = 0 ; i < nbox; i++, pbox++) 
-	 {
-	    if (pbox->x1 > pbox->x2 ||
-		pbox->y1 > pbox->y2 ||
-		pbox->x2 > intelScreen->width ||
-		pbox->y2 > intelScreen->height) {
-	       _mesa_warning(&intel->ctx, "Bad cliprect in intelCopyBuffer()");
-	       continue;
-	    }
-
-	    box = *pbox;
-
-	    if (rect)
-	    {
-	       if (rect->x1 > box.x1)
-		  box.x1 = rect->x1;
-	       if (rect->y1 > box.y1)
-		  box.y1 = rect->y1;
-	       if (rect->x2 < box.x2)
-		  box.x2 = rect->x2;
-	       if (rect->y2 < box.y2)
-		  box.y2 = rect->y2;
-
-	       if (box.x1 > box.x2 || box.y1 > box.y2)
-		  continue;
-	    }
-
-	    BEGIN_BATCH( 8);
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (box.y1 << 16) | box.x1 );
-	    OUT_BATCH( (box.y2 << 16) | box.x2 );
-
-	    if (intel->sarea->pf_current_page == 0) 
-	       OUT_BATCH( intelScreen->front.offset );
-	    else
-	       OUT_BATCH( intelScreen->back.offset );			
-
-	    OUT_BATCH( (box.y1 << 16) | box.x1 );
-	    OUT_BATCH( BR13 & 0xffff );
-
-	    if (intel->sarea->pf_current_page == 0) 
-	       OUT_BATCH( intelScreen->back.offset );			
-	    else
-	       OUT_BATCH( intelScreen->front.offset );
-
-	    ADVANCE_BATCH();
-	 }
+   if (intel->numClipRects == 0 && !ignore_cliprects) {
+      if (allow_unlock) {
+	 /* If we are not doing any actual user-visible rendering,
+	  * do a sched_yield to keep the app from pegging the cpu while
+	  * achieving nothing.
+	  */
+         UNLOCK_HARDWARE(intel);
+         sched_yield();
+         LOCK_HARDWARE(intel);
       }
-      intelFlushBatchLocked( intel, GL_TRUE, GL_TRUE, GL_TRUE );
-      UNLOCK_HARDWARE( intel );
+      intel->vtbl.lost_hardware(intel);
    }
 
-   if (!rect)
-   {
-       intel->swap_count++;
-       (*dri_interface->getUST)(&ust);
-       if (missed_target) {
-	   intel->swap_missed_count++;
-	   intel->swap_missed_ust = ust -  intel->swap_ust;
-       }
-   
-       intel->swap_ust = ust;
+done:
+   if (INTEL_DEBUG & DEBUG_BATCH) {
+      dri_bo_map(batch->buf, GL_FALSE);
+      intel_decode(ptr, used / 4, batch->buf->offset,
+		   intel->intelScreen->deviceID);
+      dri_bo_unmap(batch->buf);
    }
 }
 
 
-
-
-void intelEmitFillBlitLocked( intelContextPtr intel,
-			      GLuint cpp,
-			      GLshort dst_pitch,  /* in bytes */
-			      GLuint dst_offset,
-			      GLshort x, GLshort y, 
-			      GLshort w, GLshort h,
-			      GLuint color )
+void
+intel_batchbuffer_flush(struct intel_batchbuffer *batch)
 {
-   GLuint BR13, CMD;
-   BATCH_LOCALS;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = dst_pitch | (0xF0 << 16) | (1<<24);
-      CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = dst_pitch | (0xF0 << 16) | (1<<24) | (1<<25);
-      CMD = (XY_COLOR_BLT_CMD | XY_COLOR_BLT_WRITE_ALPHA |
-	     XY_COLOR_BLT_WRITE_RGB);
-      break;
-   default:
-      return;
-   }
-
-   BEGIN_BATCH( 6);
-   OUT_BATCH( CMD );
-   OUT_BATCH( BR13 );
-   OUT_BATCH( (y << 16) | x );
-   OUT_BATCH( ((y+h) << 16) | (x+w) );
-   OUT_BATCH( dst_offset );
-   OUT_BATCH( color );
-   ADVANCE_BATCH();
-}
+   struct intel_context *intel = batch->intel;
+   GLuint used = batch->ptr - batch->map;
+   GLboolean was_locked = intel->locked;
 
-
-/* Copy BitBlt
- */
-void intelEmitCopyBlitLocked( intelContextPtr intel,
-			      GLuint cpp,
-			      GLshort src_pitch,
-			      GLuint  src_offset,
-			      GLshort dst_pitch,
-			      GLuint  dst_offset,
-			      GLshort src_x, GLshort src_y,
-			      GLshort dst_x, GLshort dst_y,
-			      GLshort w, GLshort h )
-{
-   GLuint CMD, BR13;
-   int dst_y2 = dst_y + h;
-   int dst_x2 = dst_x + w;
-   BATCH_LOCALS;
-
-   src_pitch *= cpp;
-   dst_pitch *= cpp;
-
-   switch(cpp) {
-   case 1: 
-   case 2: 
-   case 3: 
-      BR13 = dst_pitch | (0xCC << 16) | (1<<24);
-      CMD = XY_SRC_COPY_BLT_CMD;
-      break;
-   case 4:
-      BR13 = dst_pitch | (0xCC << 16) | (1<<24) | (1<<25);
-      CMD = (XY_SRC_COPY_BLT_CMD | XY_SRC_COPY_BLT_WRITE_ALPHA |
-	     XY_SRC_COPY_BLT_WRITE_RGB);
-      break;
-   default:
+   if (used == 0)
       return;
-   }
-
-   if (dst_y2 < dst_y ||
-       dst_x2 < dst_x) {
-      return;
-   }
-
-   BEGIN_BATCH( 12);
-   OUT_BATCH( CMD );
-   OUT_BATCH( BR13 );
-   OUT_BATCH( (dst_y << 16) | dst_x );
-   OUT_BATCH( (dst_y2 << 16) | dst_x2 );
-   OUT_BATCH( dst_offset );	
-   OUT_BATCH( (src_y << 16) | src_x );
-   OUT_BATCH( src_pitch );
-   OUT_BATCH( src_offset ); 
-   ADVANCE_BATCH();
-}
-
-
 
-void intelClearWithBlit(GLcontext *ctx, GLbitfield buffers, GLboolean allFoo,
-                        GLint cx1Foo, GLint cy1Foo, GLint cwFoo, GLint chFoo)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   GLuint clear_depth, clear_color;
-   GLint cx, cy, cw, ch;
-   GLboolean all;
-   GLint pitch;
-   GLint cpp = intelScreen->cpp;
-   GLint i;
-   GLuint BR13, CMD, D_CMD;
-   BATCH_LOCALS;
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-
-   /* get clear bounds after locking */
-   cx = intel->ctx.DrawBuffer->_Xmin;
-   cy = intel->ctx.DrawBuffer->_Ymin;
-   cw = intel->ctx.DrawBuffer->_Xmax - cx;
-   ch = intel->ctx.DrawBuffer->_Ymax - cy;
-   all = (cw == intel->ctx.DrawBuffer->Width &&
-          ch == intel->ctx.DrawBuffer->Height);
-
-   pitch = intelScreen->front.pitch;
-
-   clear_color = intel->ClearColor;
-   clear_depth = 0;
-
-   if (buffers & BUFFER_BIT_DEPTH) {
-      clear_depth = (GLuint)(ctx->Depth.Clear * intel->ClearDepth);
-   }
-
-   if (buffers & BUFFER_BIT_STENCIL) {
-      clear_depth |= (ctx->Stencil.Clear & 0xff) << 24;
+   /* Add the MI_BATCH_BUFFER_END.  Always add an MI_FLUSH - this is a
+    * performance drain that we would like to avoid.
+    */
+   if (used & 4) {
+      ((int *) batch->ptr)[0] = intel->vtbl.flush_cmd();
+      ((int *) batch->ptr)[1] = 0;
+      ((int *) batch->ptr)[2] = MI_BATCH_BUFFER_END;
+      used += 12;
    }
-
-   switch(cpp) {
-   case 2: 
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24);
-      D_CMD = CMD = XY_COLOR_BLT_CMD;
-      break;
-   case 4:
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24) | (1<<25);
-      CMD = (XY_COLOR_BLT_CMD |
-	     XY_COLOR_BLT_WRITE_ALPHA | 
-	     XY_COLOR_BLT_WRITE_RGB);
-      D_CMD = XY_COLOR_BLT_CMD;
-      if (buffers & BUFFER_BIT_DEPTH) D_CMD |= XY_COLOR_BLT_WRITE_RGB;
-      if (buffers & BUFFER_BIT_STENCIL) D_CMD |= XY_COLOR_BLT_WRITE_ALPHA;
-      break;
-   default:
-      BR13 = (0xF0 << 16) | (pitch) | (1<<24);
-      D_CMD = CMD = XY_COLOR_BLT_CMD;
-      break;
+   else {
+      ((int *) batch->ptr)[0] = intel->vtbl.flush_cmd();
+      ((int *) batch->ptr)[1] = MI_BATCH_BUFFER_END;
+      used += 8;
    }
 
-   {
-      /* flip top to bottom */
-      cy = intel->driDrawable->h - cy - ch;
-      cx = cx + intel->drawX;
-      cy += intel->drawY;
-
-      /* adjust for page flipping */
-      if ( intel->sarea->pf_current_page == 1 ) {
-	 GLuint tmp = buffers;
-
-	 buffers &= ~(BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT);
-	 if ( tmp & BUFFER_BIT_FRONT_LEFT ) buffers |= BUFFER_BIT_BACK_LEFT;
-	 if ( tmp & BUFFER_BIT_BACK_LEFT )  buffers |= BUFFER_BIT_FRONT_LEFT;
-      }
-
-      for (i = 0 ; i < intel->numClipRects ; i++) 
-      { 	 
-	 drm_clip_rect_t *box = &intel->pClipRects[i];	 
-	 drm_clip_rect_t b;
-
-	 if (!all) {
-	    GLint x = box->x1;
-	    GLint y = box->y1;
-	    GLint w = box->x2 - x;
-	    GLint h = box->y2 - y;
-
-	    if (x < cx) w -= cx - x, x = cx; 
-	    if (y < cy) h -= cy - y, y = cy;
-	    if (x + w > cx + cw) w = cx + cw - x;
-	    if (y + h > cy + ch) h = cy + ch - y;
-	    if (w <= 0) continue;
-	    if (h <= 0) continue;
-
-	    b.x1 = x;
-	    b.y1 = y;
-	    b.x2 = x + w;
-	    b.y2 = y + h;      
-	 } else {
-	    b = *box;
-	 }
+   /* TODO: Just pass the relocation list and dma buffer up to the
+    * kernel.
+    */
+   if (!was_locked)
+      LOCK_HARDWARE(intel);
 
+   do_flush_locked(batch, used, !(batch->flags & INTEL_BATCH_CLIPRECTS),
+		   GL_FALSE);
 
-	 if (b.x1 > b.x2 ||
-	     b.y1 > b.y2 ||
-	     b.x2 > intelScreen->width ||
-	     b.y2 > intelScreen->height)
-	    continue;
-
-	 if ( buffers & BUFFER_BIT_FRONT_LEFT ) {	    
-	    BEGIN_BATCH( 6);	    
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->front.offset );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
-
-	 if ( buffers & BUFFER_BIT_BACK_LEFT ) {
-	    BEGIN_BATCH( 6); 
-	    OUT_BATCH( CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->back.offset );
-	    OUT_BATCH( clear_color );
-	    ADVANCE_BATCH();
-	 }
+   if (!was_locked)
+      UNLOCK_HARDWARE(intel);
 
-	 if ( buffers & (BUFFER_BIT_STENCIL | BUFFER_BIT_DEPTH) ) {
-	    BEGIN_BATCH( 6);
-	    OUT_BATCH( D_CMD );
-	    OUT_BATCH( BR13 );
-	    OUT_BATCH( (b.y1 << 16) | b.x1 );
-	    OUT_BATCH( (b.y2 << 16) | b.x2 );
-	    OUT_BATCH( intelScreen->depth.offset );
-	    OUT_BATCH( clear_depth );
-	    ADVANCE_BATCH();
-	 }      
-      }
-   }
-   intelFlushBatchLocked( intel, GL_TRUE, GL_FALSE, GL_TRUE );
-   UNLOCK_HARDWARE( intel );
+   /* Reset the buffer:
+    */
+   intel_batchbuffer_reset(batch);
 }
 
-
-
-
-void intelDestroyBatchBuffer( GLcontext *ctx )
+void
+intel_batchbuffer_finish(struct intel_batchbuffer *batch)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-
-   if (intel->alloc.offset) {
-      intelFreeAGP( intel, intel->alloc.ptr );
-      intel->alloc.ptr = NULL;
-      intel->alloc.offset = 0;
-   }
-   else if (intel->alloc.ptr) {
-      free(intel->alloc.ptr);
-      intel->alloc.ptr = NULL;
-   }
-
-   memset(&intel->batch, 0, sizeof(intel->batch));
+   intel_batchbuffer_flush(batch);
+   if (batch->last_fence != NULL)
+      dri_fence_wait(batch->last_fence);
 }
 
 
-void intelInitBatchBuffer( GLcontext *ctx )
+/*  This is the only way buffers get added to the validate list.
+ */
+GLboolean
+intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
+                             dri_bo *buffer,
+                             GLuint flags, GLuint delta)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct buffer_reloc *r = &batch->reloc[batch->nr_relocs++];
 
-   /* This path isn't really safe with rotate:
-    */
-   if (getenv("INTEL_BATCH") && intel->intelScreen->allow_batchbuffer) {      
-      switch (intel->intelScreen->deviceID) {
-      case PCI_CHIP_I865_G:
-	 /* HW bug?  Seems to crash if batchbuffer crosses 4k boundary.
-	  */
-	 intel->alloc.size = 8 * 1024; 
-	 break;
-      default:
-	 /* This is the smallest amount of memory the kernel deals with.
-	  * We'd ideally like to make this smaller.
-	  */
-	 intel->alloc.size = 1 << intel->intelScreen->logTextureGranularity;
-	 break;
-      }
+   assert(batch->nr_relocs <= MAX_RELOCS);
 
-      intel->alloc.ptr = intelAllocateAGP( intel, intel->alloc.size );
-      if (intel->alloc.ptr)
-	 intel->alloc.offset = 
-	    intelAgpOffsetFromVirtual( intel, intel->alloc.ptr );
-      else
-         intel->alloc.offset = 0; /* OK? */
-   }
+   dri_bo_reference(buffer);
+   r->buf = buffer;
+   r->offset = batch->ptr - batch->map;
+   r->delta = delta;
+   r->validate_flags = flags;
+
+   batch->ptr += 4;
+   return GL_TRUE;
+}
 
-   /* The default is now to use a local buffer and pass that to the
-    * kernel.  This is also a fallback if allocation fails on the
-    * above path:
-    */
-   if (!intel->alloc.ptr) {
-      intel->alloc.size = 8 * 1024;
-      intel->alloc.ptr = malloc( intel->alloc.size );
-      intel->alloc.offset = 0;
-   }
 
-   assert(intel->alloc.ptr);
+
+void
+intel_batchbuffer_data(struct intel_batchbuffer *batch,
+                       const void *data, GLuint bytes, GLuint flags)
+{
+   assert((bytes & 3) == 0);
+   intel_batchbuffer_require_space(batch, bytes, flags);
+   __memcpy(batch->ptr, data, bytes);
+   batch->ptr += bytes;
 }
diff --git a/src/mesa/drivers/dri/i915/intel_batchbuffer.h b/src/mesa/drivers/dri/i915/intel_batchbuffer.h
index 577d07137ff..850a91e1c91 100644
--- a/src/mesa/drivers/dri/i915/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/i915/intel_batchbuffer.h
@@ -1,126 +1,122 @@
-/**************************************************************************
- * 
- * 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_BATCHBUFFER_H
 #define INTEL_BATCHBUFFER_H
 
-#include "intel_context.h"
-#include "intel_ioctl.h"
+#include "mtypes.h"
+#include "dri_bufmgr.h"
 
+struct intel_context;
 
-#define BATCH_LOCALS	GLubyte *batch_ptr;
+#define BATCH_SZ 16384
+#define BATCH_RESERVED 16
 
-/* #define VERBOSE 0 */
-#ifndef VERBOSE
-extern int VERBOSE;
-#endif
+#define MAX_RELOCS 4096
 
+#define INTEL_BATCH_NO_CLIPRECTS 0x1
+#define INTEL_BATCH_CLIPRECTS    0x2
 
-#define BEGIN_BATCH(n)							\
-do {									\
-   if (VERBOSE) fprintf(stderr, 					\
-			"BEGIN_BATCH(%ld) in %s, %d dwords free\n",	\
-			((unsigned long)n), __FUNCTION__,		\
-			intel->batch.space/4);				\
-   if (intel->batch.space < (n)*4)					\
-      intelFlushBatch(intel, GL_TRUE);					\
-   if (intel->batch.space == intel->batch.size)	intel->batch.func = __FUNCTION__;			\
-   batch_ptr = intel->batch.ptr;					\
-} while (0)
+struct buffer_reloc
+{
+   dri_bo *buf;
+   GLuint offset;
+   GLuint delta;                /* not needed? */
+   GLuint validate_flags;
+};
 
-#define OUT_BATCH(n)					\
-do {							\
-   *(GLuint *)batch_ptr = (n);				\
-   if (VERBOSE) fprintf(stderr, " -- %08x at %s/%d\n", (n), __FILE__, __LINE__);	\
-   batch_ptr += 4;					\
-} while (0)
+struct intel_batchbuffer
+{
+   struct intel_context *intel;
+
+   dri_bo *buf;
+   dri_fence *last_fence;
+   GLuint flags;
+
+   drmBOList list;
+   GLuint list_count;
+   GLubyte *map;
+   GLubyte *ptr;
+
+   struct buffer_reloc reloc[MAX_RELOCS];
+   GLuint nr_relocs;
+   GLuint size;
+};
+
+struct intel_batchbuffer *intel_batchbuffer_alloc(struct intel_context
+                                                  *intel);
+
+void intel_batchbuffer_free(struct intel_batchbuffer *batch);
+
+
+void intel_batchbuffer_finish(struct intel_batchbuffer *batch);
+
+void intel_batchbuffer_flush(struct intel_batchbuffer *batch);
+
+void intel_batchbuffer_reset(struct intel_batchbuffer *batch);
 
-#define ADVANCE_BATCH()						\
-do {								\
-   if (VERBOSE) fprintf(stderr, "ADVANCE_BATCH()\n");		\
-   intel->batch.space -= (batch_ptr - intel->batch.ptr);	\
-   intel->batch.ptr = batch_ptr;				\
-   assert(intel->batch.space >= 0);				\
-} while(0)
-
-extern void intelInitBatchBuffer( GLcontext *ctx );
-extern void intelDestroyBatchBuffer( GLcontext *ctx );
-
-extern void intelStartInlinePrimitive( intelContextPtr intel, GLuint prim );
-extern void intelWrapInlinePrimitive( intelContextPtr intel );
-extern void intelRestartInlinePrimitive( intelContextPtr intel );
-extern GLuint *intelEmitInlinePrimitiveLocked(intelContextPtr intel, 
-					      int primitive, int dwords,
-					      int vertex_size);
-extern void intelCopyBuffer( const __DRIdrawablePrivate *dpriv,
-			     const drm_clip_rect_t	*rect);
-extern void intelClearWithBlit(GLcontext *ctx, GLbitfield mask, GLboolean all,
-			     GLint cx1, GLint cy1, GLint cw, GLint ch);
-
-extern void intelEmitCopyBlitLocked( intelContextPtr intel,
-				     GLuint cpp,
-				     GLshort src_pitch,
-				     GLuint  src_offset,
-				     GLshort dst_pitch,
-				     GLuint  dst_offset,
-				     GLshort srcx, GLshort srcy,
-				     GLshort dstx, GLshort dsty,
-				     GLshort w, GLshort h );
-
-extern void intelEmitFillBlitLocked( intelContextPtr intel,
-				     GLuint cpp,
-				     GLshort dst_pitch,
-				     GLuint dst_offset,
-				     GLshort x, GLshort y, 
-				     GLshort w, GLshort h,
-				     GLuint color );
-
-
-
-
-static __inline GLuint *intelExtendInlinePrimitive( intelContextPtr intel, 
-						GLuint dwords )
+
+/* Unlike bmBufferData, this currently requires the buffer be mapped.
+ * Consider it a convenience function wrapping multple
+ * intel_buffer_dword() calls.
+ */
+void intel_batchbuffer_data(struct intel_batchbuffer *batch,
+                            const void *data, GLuint bytes, GLuint flags);
+
+void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
+                                     GLuint bytes);
+
+GLboolean intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
+                                       dri_bo *buffer,
+                                       GLuint flags, GLuint offset);
+
+/* Inline functions - might actually be better off with these
+ * non-inlined.  Certainly better off switching all command packets to
+ * be passed as structs rather than dwords, but that's a little bit of
+ * work...
+ */
+static INLINE GLuint
+intel_batchbuffer_space(struct intel_batchbuffer *batch)
 {
-   GLuint sz = dwords * sizeof(GLuint);
-   GLuint *ptr;
+   return (batch->size - BATCH_RESERVED) - (batch->ptr - batch->map);
+}
 
-   if (intel->batch.space < sz) {
-      intelWrapInlinePrimitive( intel );
-/*       assert(intel->batch.space >= sz); */
-   }
 
-/*    assert(intel->prim.primitive != ~0); */
-   ptr = (GLuint *)intel->batch.ptr;
-   intel->batch.ptr += sz;
-   intel->batch.space -= sz;
+static INLINE void
+intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch, GLuint dword)
+{
+   assert(batch->map);
+   assert(intel_batchbuffer_space(batch) >= 4);
+   *(GLuint *) (batch->ptr) = dword;
+   batch->ptr += 4;
+}
 
-   return ptr;
+static INLINE void
+intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
+                                GLuint sz, GLuint flags)
+{
+   assert(sz < batch->size - 8);
+   if (intel_batchbuffer_space(batch) < sz ||
+       (batch->flags != 0 && flags != 0 && batch->flags != flags))
+      intel_batchbuffer_flush(batch);
+
+   batch->flags |= flags;
 }
 
+/* Here are the crusty old macros, to be removed:
+ */
+#define BATCH_LOCALS
+
+#define BEGIN_BATCH(n, flags) do {				\
+   assert(!intel->prim.flush);					\
+   intel_batchbuffer_require_space(intel->batch, (n)*4, flags);	\
+} while (0)
+
+#define OUT_BATCH(d)  intel_batchbuffer_emit_dword(intel->batch, d)
+
+#define OUT_RELOC(buf, flags, delta) do { 				\
+   assert((delta) >= 0);						\
+   intel_batchbuffer_emit_reloc(intel->batch, buf, flags, delta);	\
+} while (0)
+
+#define ADVANCE_BATCH() do { } while(0)
 
 
 #endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_blit.c b/src/mesa/drivers/dri/i915/intel_blit.c
index dbe4ba2ac5e..5d97f08434a 100644
--- a/src/mesa/drivers/dri/i915tex/intel_blit.c
+++ b/src/mesa/drivers/dri/i915/intel_blit.c
@@ -67,8 +67,8 @@ intelCopyBuffer(const __DRIdrawablePrivate * dPriv,
    intelScreen = intel->intelScreen;
 
    if (intel->last_swap_fence) {
-      driFenceFinish(intel->last_swap_fence, DRM_FENCE_TYPE_EXE, GL_TRUE);
-      driFenceUnReference(intel->last_swap_fence);
+      dri_fence_wait(intel->last_swap_fence);
+      dri_fence_unreference(intel->last_swap_fence);
       intel->last_swap_fence = NULL;
    }
    intel->last_swap_fence = intel->first_swap_fence;
@@ -140,19 +140,20 @@ intelCopyBuffer(const __DRIdrawablePrivate * dPriv,
 	 OUT_BATCH((pbox->y2 << 16) | pbox->x2);
 
 	 OUT_RELOC(frontRegion->buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-		   DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE, 0);
+		   0);
 	 OUT_BATCH((pbox->y1 << 16) | pbox->x1);
 	 OUT_BATCH(BR13 & 0xffff);
 	 OUT_RELOC(backRegion->buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-		   DRM_BO_MASK_MEM | DRM_BO_FLAG_READ, 0);
+		   0);
 
 	 ADVANCE_BATCH();
       }
 
       if (intel->first_swap_fence)
-	 driFenceUnReference(intel->first_swap_fence);
-      intel->first_swap_fence = intel_batchbuffer_flush(intel->batch);
-      driFenceReference(intel->first_swap_fence);
+	 dri_fence_unreference(intel->first_swap_fence);
+      intel_batchbuffer_flush(intel->batch);
+      intel->first_swap_fence = intel->batch->last_fence;
+      dri_fence_reference(intel->first_swap_fence);
    }
 
    UNLOCK_HARDWARE(intel);
@@ -165,7 +166,7 @@ void
 intelEmitFillBlit(struct intel_context *intel,
                   GLuint cpp,
                   GLshort dst_pitch,
-                  struct _DriBufferObject *dst_buffer,
+                  dri_bo *dst_buffer,
                   GLuint dst_offset,
                   GLshort x, GLshort y, GLshort w, GLshort h, GLuint color)
 {
@@ -199,8 +200,7 @@ intelEmitFillBlit(struct intel_context *intel,
    OUT_BATCH(BR13);
    OUT_BATCH((y << 16) | x);
    OUT_BATCH(((y + h) << 16) | (x + w));
-   OUT_RELOC(dst_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-             DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE, dst_offset);
+   OUT_RELOC(dst_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE, dst_offset);
    OUT_BATCH(color);
    ADVANCE_BATCH();
 }
@@ -236,10 +236,10 @@ void
 intelEmitCopyBlit(struct intel_context *intel,
                   GLuint cpp,
                   GLshort src_pitch,
-                  struct _DriBufferObject *src_buffer,
+                  dri_bo *src_buffer,
                   GLuint src_offset,
                   GLshort dst_pitch,
-                  struct _DriBufferObject *dst_buffer,
+                  dri_bo *dst_buffer,
                   GLuint dst_offset,
                   GLshort src_x, GLshort src_y,
                   GLshort dst_x, GLshort dst_y, 
@@ -297,12 +297,10 @@ intelEmitCopyBlit(struct intel_context *intel,
       OUT_BATCH(BR13);
       OUT_BATCH((dst_y << 16) | dst_x);
       OUT_BATCH((dst_y2 << 16) | dst_x2);
-      OUT_RELOC(dst_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE, dst_offset);
+      OUT_RELOC(dst_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE, dst_offset);
       OUT_BATCH((src_y << 16) | src_x);
       OUT_BATCH(((GLint) src_pitch & 0xffff));
-      OUT_RELOC(src_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_READ, src_offset);
+      OUT_RELOC(src_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ, src_offset);
       ADVANCE_BATCH();
    }
    else {
@@ -312,12 +310,10 @@ intelEmitCopyBlit(struct intel_context *intel,
       OUT_BATCH((0 << 16) | dst_x);
       OUT_BATCH((h << 16) | dst_x2);
       OUT_RELOC(dst_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
                 dst_offset + dst_y * dst_pitch);
       OUT_BATCH((0 << 16) | src_x);
       OUT_BATCH(((GLint) src_pitch & 0xffff));
       OUT_RELOC(src_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_READ,
                 src_offset + src_y * src_pitch);
       ADVANCE_BATCH();
    }
@@ -420,7 +416,7 @@ intelClearWithBlit(GLcontext * ctx, GLbitfield mask)
                /* OK, clear this renderbuffer */
                struct intel_region *irb_region =
 		  intel_get_rb_region(fb, buf);
-               struct _DriBufferObject *write_buffer =
+               dri_bo *write_buffer =
                   intel_region_buffer(intel->intelScreen, irb_region,
                                       all ? INTEL_WRITE_FULL :
                                       INTEL_WRITE_PART);
@@ -483,7 +479,6 @@ intelClearWithBlit(GLcontext * ctx, GLbitfield mask)
                OUT_BATCH((b.y1 << 16) | b.x1);
                OUT_BATCH((b.y2 << 16) | b.x2);
                OUT_RELOC(write_buffer, DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                         DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
                          irb_region->draw_offset);
                OUT_BATCH(clearVal);
                ADVANCE_BATCH();
diff --git a/src/mesa/drivers/dri/i915tex/intel_blit.h b/src/mesa/drivers/dri/i915/intel_blit.h
index e7bc280f58a..a66af863591 100644
--- a/src/mesa/drivers/dri/i915tex/intel_blit.h
+++ b/src/mesa/drivers/dri/i915/intel_blit.h
@@ -40,10 +40,10 @@ extern void intelClearWithBlit(GLcontext * ctx, GLbitfield mask);
 extern void intelEmitCopyBlit(struct intel_context *intel,
                               GLuint cpp,
                               GLshort src_pitch,
-                              struct _DriBufferObject *src_buffer,
+                              dri_bo *src_buffer,
                               GLuint src_offset,
                               GLshort dst_pitch,
-                              struct _DriBufferObject *dst_buffer,
+                              dri_bo *dst_buffer,
                               GLuint dst_offset,
                               GLshort srcx, GLshort srcy,
                               GLshort dstx, GLshort dsty,
@@ -53,7 +53,7 @@ extern void intelEmitCopyBlit(struct intel_context *intel,
 extern void intelEmitFillBlit(struct intel_context *intel,
                               GLuint cpp,
                               GLshort dst_pitch,
-                              struct _DriBufferObject *dst_buffer,
+                              dri_bo *dst_buffer,
                               GLuint dst_offset,
                               GLshort x, GLshort y,
                               GLshort w, GLshort h, GLuint color);
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c b/src/mesa/drivers/dri/i915/intel_buffer_objects.c
index 91c45ad95be..3c73c402d96 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/i915/intel_buffer_objects.c
@@ -35,6 +35,16 @@
 #include "intel_regions.h"
 #include "dri_bufmgr.h"
 
+/** Allocates a new dri_bo to store the data for the buffer object. */
+static void
+intel_bufferobj_alloc_buffer(struct intel_context *intel,
+			     struct intel_buffer_object *intel_obj)
+{
+   intel_obj->buffer = dri_bo_alloc(intel->intelScreen->bufmgr, "bufferobj",
+				    intel_obj->Base.Size, 64,
+				    DRM_BO_FLAG_MEM_TT);
+}
+
 /**
  * There is some duplication between mesa's bufferobjects and our
  * bufmgr buffers.  Both have an integer handle and a hashtable to
@@ -44,21 +54,15 @@
 static struct gl_buffer_object *
 intel_bufferobj_alloc(GLcontext * ctx, GLuint name, GLenum target)
 {
-   struct intel_context *intel = intel_context(ctx);
    struct intel_buffer_object *obj = CALLOC_STRUCT(intel_buffer_object);
 
    _mesa_initialize_buffer_object(&obj->Base, name, target);
 
-   driGenBuffers(intel->intelScreen->regionPool,
-                 "bufferobj", 1, &obj->buffer, 64,
-		 DRM_BO_FLAG_MEM_LOCAL |
-		 DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE,
-		 0);
+   obj->buffer = NULL;
 
    return &obj->Base;
 }
 
-
 /* Break the COW tie to the region.  The region gets to keep the data.
  */
 void
@@ -68,17 +72,9 @@ intel_bufferobj_release_region(struct intel_context *intel,
    assert(intel_obj->region->buffer == intel_obj->buffer);
    intel_obj->region->pbo = NULL;
    intel_obj->region = NULL;
-   driBOUnReference(intel_obj->buffer);
-   intel_obj->buffer = NULL;
 
-   /* This leads to a large number of buffer deletion/creation events.
-    * Currently the drm doesn't like that:
-    */
-   driGenBuffers(intel->intelScreen->regionPool,
-                 "buffer object", 1, &intel_obj->buffer, 64, 0, 0);
-   LOCK_HARDWARE(intel);
-   driBOData(intel_obj->buffer, intel_obj->Base.Size, NULL, 0);
-   UNLOCK_HARDWARE(intel);
+   dri_bo_unreference(intel_obj->buffer);
+   intel_obj->buffer = NULL;
 }
 
 /* Break the COW tie to the region.  Both the pbo and the region end
@@ -109,7 +105,7 @@ intel_bufferobj_free(GLcontext * ctx, struct gl_buffer_object *obj)
       intel_bufferobj_release_region(intel, intel_obj);
    }
    else if (intel_obj->buffer) {
-      driDeleteBuffers(1, &intel_obj->buffer);
+      dri_bo_unreference(intel_obj->buffer);
    }
 
    _mesa_free(intel_obj);
@@ -139,9 +135,15 @@ intel_bufferobj_data(GLcontext * ctx,
    if (intel_obj->region)
       intel_bufferobj_release_region(intel, intel_obj);
 
-   LOCK_HARDWARE(intel);
-   driBOData(intel_obj->buffer, size, data, 0);
-   UNLOCK_HARDWARE(intel);
+   if (intel_obj->buffer != NULL && intel_obj->buffer->size != size) {
+      dri_bo_unreference(intel_obj->buffer);
+      intel_obj->buffer = NULL;
+   }
+
+   intel_bufferobj_alloc_buffer(intel, intel_obj);
+
+   if (data != NULL)
+      dri_bo_subdata(intel_obj->buffer, 0, size, data);
 }
 
 
@@ -166,7 +168,7 @@ intel_bufferobj_subdata(GLcontext * ctx,
    if (intel_obj->region)
       intel_bufferobj_cow(intel, intel_obj);
 
-   driBOSubData(intel_obj->buffer, offset, size, data);
+   dri_bo_subdata(intel_obj->buffer, offset, size, data);
 }
 
 
@@ -183,7 +185,7 @@ intel_bufferobj_get_subdata(GLcontext * ctx,
    struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
 
    assert(intel_obj);
-   driBOGetSubData(intel_obj->buffer, offset, size, data);
+   dri_bo_get_subdata(intel_obj->buffer, offset, size, data);
 }
 
 
@@ -206,8 +208,13 @@ intel_bufferobj_map(GLcontext * ctx,
    if (intel_obj->region)
       intel_bufferobj_cow(intel, intel_obj);
 
-   obj->Pointer = driBOMap(intel_obj->buffer,
-                           DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 0);
+   if (intel_obj->buffer == NULL) {
+      obj->Pointer = NULL;
+      return NULL;
+   }
+
+   dri_bo_map(intel_obj->buffer, GL_TRUE);
+   obj->Pointer = intel_obj->buffer->virtual;
    return obj->Pointer;
 }
 
@@ -222,21 +229,25 @@ intel_bufferobj_unmap(GLcontext * ctx,
    struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
 
    assert(intel_obj);
-   assert(obj->Pointer);
-   driBOUnmap(intel_obj->buffer);
-   obj->Pointer = NULL;
+   if (intel_obj->buffer != NULL) {
+      assert(obj->Pointer);
+      dri_bo_unmap(intel_obj->buffer);
+      obj->Pointer = NULL;
+   }
    return GL_TRUE;
 }
 
-struct _DriBufferObject *
+dri_bo *
 intel_bufferobj_buffer(struct intel_context *intel,
                        struct intel_buffer_object *intel_obj, GLuint flag)
 {
    if (intel_obj->region) {
       if (flag == INTEL_WRITE_PART)
          intel_bufferobj_cow(intel, intel_obj);
-      else if (flag == INTEL_WRITE_FULL)
+      else if (flag == INTEL_WRITE_FULL) {
          intel_bufferobj_release_region(intel, intel_obj);
+	 intel_bufferobj_alloc_buffer(intel, intel_obj);
+      }
    }
 
    return intel_obj->buffer;
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.h b/src/mesa/drivers/dri/i915/intel_buffer_objects.h
index afe9b2f7cf2..db579a8ae40 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.h
+++ b/src/mesa/drivers/dri/i915/intel_buffer_objects.h
@@ -41,7 +41,7 @@ struct gl_buffer_object;
 struct intel_buffer_object
 {
    struct gl_buffer_object Base;
-   struct _DriBufferObject *buffer;     /* the low-level buffer manager's buffer handle */
+   dri_bo *buffer;     /* the low-level buffer manager's buffer handle */
 
    struct intel_region *region; /* Is there a zero-copy texture
                                    associated with this (pixel)
@@ -51,9 +51,9 @@ struct intel_buffer_object
 
 /* Get the bm buffer associated with a GL bufferobject:
  */
-struct _DriBufferObject *intel_bufferobj_buffer(struct intel_context *intel,
-                                                struct intel_buffer_object
-                                                *obj, GLuint flag);
+dri_bo *intel_bufferobj_buffer(struct intel_context *intel,
+			       struct intel_buffer_object
+			       *obj, GLuint flag);
 
 /* Hook the bufferobject implementation into mesa: 
  */
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffers.c b/src/mesa/drivers/dri/i915/intel_buffers.c
index 15d02f8e2cf..938bed6da7c 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffers.c
+++ b/src/mesa/drivers/dri/i915/intel_buffers.c
@@ -235,34 +235,34 @@ intelWindowMoved(struct intel_context *intel)
       drmI830Sarea *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 pipeA_rect = { .x1 = sarea->pipeA_x, .y1 = sarea->pipeA_y,
-				     .x2 = sarea->pipeA_x + sarea->pipeA_w,
-				     .y2 = sarea->pipeA_y + sarea->pipeA_h };
-      drm_clip_rect_t pipeB_rect = { .x1 = sarea->pipeB_x, .y1 = sarea->pipeB_y,
-				     .x2 = sarea->pipeB_x + sarea->pipeB_w,
-				     .y2 = sarea->pipeB_y + sarea->pipeB_h };
-      GLint areaA = driIntersectArea( drw_rect, pipeA_rect );
-      GLint areaB = driIntersectArea( drw_rect, pipeB_rect );
+      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 = intel_fb->vblank_flags;
       GLboolean pf_active;
-      GLint pf_pipes;
+      GLint pf_planes;
 
       /* Update page flipping info
        */
-      pf_pipes = 0;
+      pf_planes = 0;
 
       if (areaA > 0)
-	 pf_pipes |= 1;
+	 pf_planes |= 1;
 
       if (areaB > 0)
-	 pf_pipes |= 2;
+	 pf_planes |= 2;
 
       intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
-				   (intel_fb->pf_pipes & 0x2)) & 0x3;
+				   (intel_fb->pf_planes & 0x2)) & 0x3;
 
       intel_fb->pf_num_pages = intel->intelScreen->third.handle ? 3 : 2;
 
-      pf_active = pf_pipes && (pf_pipes & intel->sarea->pf_active) == pf_pipes;
+      pf_active = pf_planes && (pf_planes & intel->sarea->pf_active) == pf_planes;
 
       if (INTEL_DEBUG & DEBUG_LOCK)
 	 if (pf_active != intel_fb->pf_active)
@@ -270,8 +270,8 @@ intelWindowMoved(struct intel_context *intel)
 			 pf_active ? "" : "in");
 
       if (pf_active) {
-	 /* Sync pages between pipes if we're flipping on both at the same time */
-	 if (pf_pipes == 0x3 &&	pf_pipes != intel_fb->pf_pipes &&
+	 /* 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;
@@ -301,7 +301,7 @@ intelWindowMoved(struct intel_context *intel)
 	    drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
 	 }
 
-	 intel_fb->pf_pipes = pf_pipes;
+	 intel_fb->pf_planes = pf_planes;
       }
 
       intel_fb->pf_active = pf_active;
@@ -715,14 +715,14 @@ intel_wait_flips(struct intel_context *intel, GLuint batch_flags)
 			     BUFFER_BACK_LEFT);
 
    if (intel_fb->Base.Name == 0 && intel_rb->pf_pending == intel_fb->pf_seq) {
-      GLint pf_pipes = intel_fb->pf_pipes;
+      GLint pf_planes = intel_fb->pf_planes;
       BATCH_LOCALS;
 
       /* Wait for pending flips to take effect */
       BEGIN_BATCH(2, batch_flags);
-      OUT_BATCH(pf_pipes & 0x1 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_A_FLIP)
+      OUT_BATCH(pf_planes & 0x1 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_A_FLIP)
 		: 0);
-      OUT_BATCH(pf_pipes & 0x2 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_B_FLIP)
+      OUT_BATCH(pf_planes & 0x2 ? (MI_WAIT_FOR_EVENT | MI_WAIT_FOR_PLANE_B_FLIP)
 		: 0);
       ADVANCE_BATCH();
 
@@ -761,7 +761,7 @@ intelPageFlip(const __DRIdrawablePrivate * dPriv)
    if (dPriv->numClipRects && intel_fb->pf_active) {
       drm_i915_flip_t flip;
 
-      flip.pipes = intel_fb->pf_pipes;
+      flip.pipes = intel_fb->pf_planes;
 
       ret = drmCommandWrite(intel->driFd, DRM_I915_FLIP, &flip, sizeof(flip));
    }
@@ -776,7 +776,7 @@ intelPageFlip(const __DRIdrawablePrivate * dPriv)
    }
 
    intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
-				(intel_fb->pf_pipes & 0x2)) & 0x3;
+				(intel_fb->pf_planes & 0x2)) & 0x3;
 
    if (dPriv->numClipRects != 0) {
       intel_get_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT)->pf_pending =
@@ -860,7 +860,7 @@ intelScheduleSwap(const __DRIdrawablePrivate * dPriv, GLboolean *missed_target)
       swap.seqtype |= DRM_VBLANK_FLIP;
 
       intel_fb->pf_current_page = (((intel->sarea->pf_current_page >>
-				     (intel_fb->pf_pipes & 0x2)) & 0x3) + 1) %
+				     (intel_fb->pf_planes & 0x2)) & 0x3) + 1) %
 				  intel_fb->pf_num_pages;
    }
 
@@ -884,7 +884,7 @@ intelScheduleSwap(const __DRIdrawablePrivate * dPriv, GLboolean *missed_target)
    } else {
       if (swap.seqtype & DRM_VBLANK_FLIP) {
 	 intel_fb->pf_current_page = ((intel->sarea->pf_current_page >>
-					(intel_fb->pf_pipes & 0x2)) & 0x3) %
+					(intel_fb->pf_planes & 0x2)) & 0x3) %
 				     intel_fb->pf_num_pages;
       }
 
@@ -1124,6 +1124,15 @@ intel_draw_buffer(GLcontext * ctx, struct gl_framebuffer *fb)
       ctx->Driver.Enable(ctx, GL_STENCIL_TEST, ctx->Stencil.Enabled);
    }
 
+   /*
+    * Update depth test state
+    */
+   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);
+   }
 
    /**
     ** Release old regions, reference new regions
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffers.h b/src/mesa/drivers/dri/i915/intel_buffers.h
index 3b686cb5c18..3b686cb5c18 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffers.h
+++ b/src/mesa/drivers/dri/i915/intel_buffers.h
diff --git a/src/mesa/drivers/dri/i915/intel_context.c b/src/mesa/drivers/dri/i915/intel_context.c
index 11c23f24a1b..041a155fe73 100644
--- a/src/mesa/drivers/dri/i915/intel_context.c
+++ b/src/mesa/drivers/dri/i915/intel_context.c
@@ -38,7 +38,6 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 #include "tnl/t_vertex.h"
@@ -48,17 +47,24 @@
 #include "intel_screen.h"
 
 #include "i830_dri.h"
-#include "i830_common.h"
 
+#include "intel_buffers.h"
 #include "intel_tex.h"
 #include "intel_span.h"
 #include "intel_tris.h"
 #include "intel_ioctl.h"
 #include "intel_batchbuffer.h"
-
+#include "intel_blit.h"
+#include "intel_pixel.h"
+#include "intel_regions.h"
+#include "intel_buffer_objects.h"
+#include "intel_fbo.h"
+#include "intel_decode.h"
+
+#include "drirenderbuffer.h"
 #include "vblank.h"
 #include "utils.h"
-#include "xmlpool.h" /* for symbolic values of enum-type options */
+#include "xmlpool.h"            /* for symbolic values of enum-type options */
 #ifndef INTEL_DEBUG
 int INTEL_DEBUG = (0);
 #endif
@@ -75,67 +81,74 @@ int INTEL_DEBUG = (0);
 #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_secondary_color
 #define need_GL_NV_vertex_program
 #include "extension_helper.h"
 
-#ifndef VERBOSE
-int VERBOSE = 0;
-#endif
 
-#if DEBUG_LOCKING
-char *prevLockFile;
-int prevLockLine;
-#endif
+#define DRIVER_DATE                     "20061102"
 
-/***************************************
- * Mesa's Driver Functions
- ***************************************/
+_glthread_Mutex lockMutex;
+static GLboolean lockMutexInit = GL_FALSE;
 
-#define DRIVER_DATE "20061017"
 
-const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
+static const GLubyte *
+intelGetString(GLcontext * ctx, GLenum name)
 {
-   const char * chipset;
+   const char *chipset;
    static char buffer[128];
 
    switch (name) {
    case GL_VENDOR:
-      return (GLubyte *)"Tungsten Graphics, Inc";
+      return (GLubyte *) "Tungsten Graphics, Inc";
       break;
-      
+
    case GL_RENDERER:
-      switch (INTEL_CONTEXT(ctx)->intelScreen->deviceID) {
+      switch (intel_context(ctx)->intelScreen->deviceID) {
       case PCI_CHIP_845_G:
-	 chipset = "Intel(R) 845G"; break;
+         chipset = "Intel(R) 845G";
+         break;
       case PCI_CHIP_I830_M:
-	 chipset = "Intel(R) 830M"; break;
+         chipset = "Intel(R) 830M";
+         break;
       case PCI_CHIP_I855_GM:
-	 chipset = "Intel(R) 852GM/855GM"; break;
+         chipset = "Intel(R) 852GM/855GM";
+         break;
       case PCI_CHIP_I865_G:
-	 chipset = "Intel(R) 865G"; break;
+         chipset = "Intel(R) 865G";
+         break;
       case PCI_CHIP_I915_G:
-	 chipset = "Intel(R) 915G"; break;
+         chipset = "Intel(R) 915G";
+         break;
       case PCI_CHIP_I915_GM:
-	 chipset = "Intel(R) 915GM"; break;
+         chipset = "Intel(R) 915GM";
+         break;
       case PCI_CHIP_I945_G:
-	 chipset = "Intel(R) 945G"; break;
+         chipset = "Intel(R) 945G";
+         break;
       case PCI_CHIP_I945_GM:
-	 chipset = "Intel(R) 945GM"; break;
+         chipset = "Intel(R) 945GM";
+         break;
       case PCI_CHIP_I945_GME:
-	 chipset = "Intel(R) 945GME"; break;
+         chipset = "Intel(R) 945GME";
+         break;
       case PCI_CHIP_G33_G:
-	 chipset = "Intel(R) G33"; break;
+	 chipset = "Intel(R) G33";
+	 break;
       case PCI_CHIP_Q35_G:
-	 chipset = "Intel(R) Q35"; break;
+	 chipset = "Intel(R) Q35";
+	 break;
       case PCI_CHIP_Q33_G:
-	 chipset = "Intel(R) Q33"; break;
+	 chipset = "Intel(R) Q33";
+	 break;
       default:
-	 chipset = "Unknown Intel Chipset"; break;
+         chipset = "Unknown Intel Chipset";
+         break;
       }
 
-      (void) driGetRendererString( buffer, chipset, DRIVER_DATE, 0 );
+      (void) driGetRendererString(buffer, chipset, DRIVER_DATE, 0);
       return (GLubyte *) buffer;
 
    default:
@@ -146,51 +159,56 @@ const GLubyte *intelGetString( GLcontext *ctx, GLenum name )
 
 /**
  * Extension strings exported by the intel driver.
- *
- * \note
- * It appears that ARB_texture_env_crossbar has "disappeared" compared to the
- * old i830-specific driver.
  */
-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_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_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_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_vertex_program",              GL_NV_vertex_program_functions },
-    { "GL_NV_vertex_program1_1",           NULL },
-    { "GL_SGIS_generate_mipmap",           NULL },
-    { NULL,                                NULL }
+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_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_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},
+#if 1                           /* XXX FBO temporary? */
+   {"GL_EXT_packed_depth_stencil", NULL},
+#endif
+   {"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_vertex_program", GL_NV_vertex_program_functions},
+   {"GL_NV_vertex_program1_1", NULL},
+/*     { "GL_SGIS_generate_mipmap",           NULL }, */
+   {NULL, NULL}
+};
+
+const struct dri_extension ttm_extensions[] = {
+   {"GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions},
+   {"GL_ARB_pixel_buffer_object", NULL},
+   {NULL, NULL}
 };
 
 extern const struct tnl_pipeline_stage _intel_render_stage;
@@ -206,82 +224,144 @@ static const struct tnl_pipeline_stage *intel_pipeline[] = {
    &_tnl_point_attenuation_stage,
    &_tnl_vertex_program_stage,
 #if 1
-   &_intel_render_stage,     /* ADD: unclipped rastersetup-to-dma */
+   &_intel_render_stage,        /* ADD: unclipped rastersetup-to-dma */
 #endif
    &_tnl_render_stage,
    0,
 };
 
 
-static const struct dri_debug_control debug_control[] =
-{
-    { "fall",  DEBUG_FALLBACKS },
-    { "tex",   DEBUG_TEXTURE },
-    { "ioctl", DEBUG_IOCTL },
-    { "prim",  DEBUG_PRIMS },
-    { "vert",  DEBUG_VERTS },
-    { "state", DEBUG_STATE },
-    { "verb",  DEBUG_VERBOSE },
-    { "dri",   DEBUG_DRI },
-    { "dma",   DEBUG_DMA },
-    { "san",   DEBUG_SANITY },
-    { "sync",  DEBUG_SYNC },
-    { "sleep", DEBUG_SLEEP },
-    { "pix",   DEBUG_PIXEL },
-    { NULL,    0 }
+static const struct dri_debug_control debug_control[] = {
+   {"tex", DEBUG_TEXTURE},
+   {"state", DEBUG_STATE},
+   {"ioctl", DEBUG_IOCTL},
+   {"blit", DEBUG_BLIT},
+   {"mip", DEBUG_MIPTREE},
+   {"fall", DEBUG_FALLBACKS},
+   {"verb", DEBUG_VERBOSE},
+   {"bat", DEBUG_BATCH},
+   {"pix", DEBUG_PIXEL},
+   {"buf", DEBUG_BUFMGR},
+   {"reg", DEBUG_REGION},
+   {"fbo", DEBUG_FBO},
+   {"lock", DEBUG_LOCK},
+   {NULL, 0}
 };
 
 
-static void intelInvalidateState( GLcontext *ctx, GLuint new_state )
+static void
+intelInvalidateState(GLcontext * ctx, GLuint new_state)
 {
-   _swrast_InvalidateState( ctx, new_state );
-   _swsetup_InvalidateState( ctx, new_state );
-   _vbo_InvalidateState( ctx, new_state );
-   _tnl_InvalidateState( ctx, new_state );
-   _tnl_invalidate_vertex_state( ctx, new_state );
-   INTEL_CONTEXT(ctx)->NewGLState |= new_state;
+   _swrast_InvalidateState(ctx, new_state);
+   _swsetup_InvalidateState(ctx, new_state);
+   _vbo_InvalidateState(ctx, new_state);
+   _tnl_InvalidateState(ctx, new_state);
+   _tnl_invalidate_vertex_state(ctx, new_state);
+   intel_context(ctx)->NewGLState |= new_state;
 }
 
 
-void intelInitDriverFunctions( struct dd_function_table *functions )
+void
+intelFlush(GLcontext * ctx)
 {
-   _mesa_init_driver_functions( functions );
+   struct intel_context *intel = intel_context(ctx);
+
+   if (intel->Fallback)
+      _swrast_flush(ctx);
+
+   INTEL_FIREVERTICES(intel);
+
+   if (intel->batch->map != intel->batch->ptr)
+      intel_batchbuffer_flush(intel->batch);
+
+   /* XXX: Need to do an MI_FLUSH here.
+    */
+}
 
-   functions->Clear = intelClear;
-   functions->Flush = intelglFlush;
-   functions->Finish = intelFinish;
-   functions->GetString = intelGetString;
-   functions->UpdateState = intelInvalidateState;
 
-   intelInitTextureFuncs( functions );
-   intelInitPixelFuncs( functions );
-   intelInitStateFuncs( functions );
+/**
+ * Check if we need to rotate/warp the front color buffer to the
+ * rotated screen.  We generally need to do this when we get a glFlush
+ * or glFinish after drawing to the front color buffer.
+ */
+static void
+intelCheckFrontRotate(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] ==
+       BUFFER_BIT_FRONT_LEFT) {
+      intelScreenPrivate *screen = intel->intelScreen;
+      if (screen->current_rotation != 0) {
+         __DRIdrawablePrivate *dPriv = intel->driDrawable;
+         intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
+      }
+   }
 }
 
-static void intel_emit_invarient_state( GLcontext *ctx )
+
+/**
+ * Called via glFlush.
+ */
+static void
+intelglFlush(GLcontext * ctx)
 {
+   intelFlush(ctx);
+   intelCheckFrontRotate(ctx);
 }
 
+void
+intelFinish(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   intelFlush(ctx);
+   if (intel->batch->last_fence) {
+      dri_fence_wait(intel->batch->last_fence);
+      dri_fence_unreference(intel->batch->last_fence);
+      intel->batch->last_fence = NULL;
+   }
+   intelCheckFrontRotate(ctx);
+}
 
 
-GLboolean intelInitContext( intelContextPtr intel,
-			    const __GLcontextModes *mesaVis,
-			    __DRIcontextPrivate *driContextPriv,
-			    void *sharedContextPrivate,
-			    struct dd_function_table *functions )
+void
+intelInitDriverFunctions(struct dd_function_table *functions)
+{
+   _mesa_init_driver_functions(functions);
+
+   functions->Flush = intelglFlush;
+   functions->Finish = intelFinish;
+   functions->GetString = intelGetString;
+   functions->UpdateState = intelInvalidateState;
+   functions->CopyColorTable = _swrast_CopyColorTable;
+   functions->CopyColorSubTable = _swrast_CopyColorSubTable;
+   functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
+   functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
+
+   intelInitTextureFuncs(functions);
+   intelInitPixelFuncs(functions);
+   intelInitStateFuncs(functions);
+   intelInitBufferFuncs(functions);
+}
+
+
+GLboolean
+intelInitContext(struct intel_context *intel,
+                 const __GLcontextModes * mesaVis,
+                 __DRIcontextPrivate * driContextPriv,
+                 void *sharedContextPrivate,
+                 struct dd_function_table *functions)
 {
    GLcontext *ctx = &intel->ctx;
    GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
    __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
    drmI830Sarea *saPriv = (drmI830Sarea *)
-      (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
+      (((GLubyte *) sPriv->pSAREA) + intelScreen->sarea_priv_offset);
    int fthrottle_mode;
 
    if (!_mesa_initialize_context(&intel->ctx,
-				 mesaVis, shareCtx, 
-				 functions,
-				 (void*) intel))
+                                 mesaVis, shareCtx,
+                                 functions, (void *) intel))
       return GL_FALSE;
 
    driContextPriv->driverPrivate = intel;
@@ -289,15 +369,27 @@ GLboolean intelInitContext( intelContextPtr intel,
    intel->driScreen = sPriv;
    intel->sarea = saPriv;
 
+   intel->width = intelScreen->width;
+   intel->height = intelScreen->height;
+   intel->current_rotation = intelScreen->current_rotation;
 
-   (void) memset( intel->texture_heaps, 0, sizeof( intel->texture_heaps ) );
-   make_empty_list( & intel->swapped );
+   if (!lockMutexInit) {
+      lockMutexInit = GL_TRUE;
+      _glthread_INIT_MUTEX(lockMutex);
+   }
 
-   driParseConfigFiles (&intel->optionCache, &intelScreen->optionCache,
-			intel->driScreen->myNum, "i915");
+   driParseConfigFiles(&intel->optionCache, &intelScreen->optionCache,
+                       intel->driScreen->myNum, "i915");
 
    ctx->Const.MaxTextureMaxAnisotropy = 2.0;
 
+   /* This doesn't yet catch all non-conformant rendering, but it's a
+    * start.
+    */
+   if (getenv("INTEL_STRICT_CONFORMANCE")) {
+      intel->strict_conformance = 1;
+   }
+
    ctx->Const.MinLineWidth = 1.0;
    ctx->Const.MinLineWidthAA = 1.0;
    ctx->Const.MaxLineWidth = 3.0;
@@ -315,362 +407,256 @@ GLboolean intelInitContext( intelContextPtr intel,
     */
    _mesa_init_point(ctx);
 
+   ctx->Const.MaxColorAttachments = 4;  /* XXX FBO: review this */
+
    /* Initialize the software rasterizer and helper modules. */
-   _swrast_CreateContext( ctx );
-   _vbo_CreateContext( ctx );
-   _tnl_CreateContext( ctx );
-   _swsetup_CreateContext( ctx );
+   _swrast_CreateContext(ctx);
+   _vbo_CreateContext(ctx);
+   _tnl_CreateContext(ctx);
+   _swsetup_CreateContext(ctx);
 
    /* Install the customized pipeline: */
-   _tnl_destroy_pipeline( ctx );
-   _tnl_install_pipeline( ctx, intel_pipeline );
+   _tnl_destroy_pipeline(ctx);
+   _tnl_install_pipeline(ctx, intel_pipeline);
 
    /* Configure swrast to match hardware characteristics: */
-   _swrast_allow_pixel_fog( ctx, GL_FALSE );
-   _swrast_allow_vertex_fog( ctx, GL_TRUE );
+   _swrast_allow_pixel_fog(ctx, GL_FALSE);
+   _swrast_allow_vertex_fog(ctx, GL_TRUE);
 
    /* Dri stuff */
    intel->hHWContext = driContextPriv->hHWContext;
    intel->driFd = sPriv->fd;
-   intel->driHwLock = (drmLock *) &sPriv->pSAREA->lock;
+   intel->driHwLock = (drmLock *) & sPriv->pSAREA->lock;
 
-   intel->hw_stencil = mesaVis->stencilBits && mesaVis->depthBits == 24;
    intel->hw_stipple = 1;
 
-   switch(mesaVis->depthBits) {
-   case 0:			/* what to do in this case? */
+   /* XXX FBO: this doesn't seem to be used anywhere */
+   switch (mesaVis->depthBits) {
+   case 0:                     /* what to do in this case? */
    case 16:
-      intel->depth_scale = 1.0/0xffff;
-      intel->polygon_offset_scale = 1.0/0xffff;
-      intel->depth_clear_mask = ~0;
-      intel->ClearDepth = 0xffff;
+      intel->polygon_offset_scale = 1.0 / 0xffff;
       break;
    case 24:
-      intel->depth_scale = 1.0/0xffffff;
-      intel->polygon_offset_scale = 2.0/0xffffff; /* req'd to pass glean */
-      intel->depth_clear_mask = 0x00ffffff;
-      intel->stencil_clear_mask = 0xff000000;
-      intel->ClearDepth = 0x00ffffff;
+      intel->polygon_offset_scale = 2.0 / 0xffffff;     /* req'd to pass glean */
       break;
    default:
-      assert(0); 
+      assert(0);
       break;
    }
 
    /* Initialize swrast, tnl driver tables: */
-   intelInitSpanFuncs( ctx );
-   intelInitTriFuncs( ctx );
+   intelInitSpanFuncs(ctx);
+   intelInitTriFuncs(ctx);
 
 
    intel->RenderIndex = ~0;
 
    fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
-   intel->iw.irq_seq = -1;
    intel->irqsEmitted = 0;
 
    intel->do_irqs = (intel->intelScreen->irq_active &&
-		     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
+                     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
 
    intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
 
-   intel->vblank_flags = (intel->intelScreen->irq_active != 0)
-       ? driGetDefaultVBlankFlags(&intel->optionCache) : VBLANK_FLAG_NO_IRQ;
+   _math_matrix_ctr(&intel->ViewportMatrix);
+
+   /* Disable imaging extension until convolution is working in
+    * teximage paths:
+    */
+   driInitExtensions(ctx, card_extensions,
+/* 		      GL_TRUE, */
+                     GL_FALSE);
+
+   if (intelScreen->ttm)
+      driInitExtensions(ctx, ttm_extensions, GL_FALSE);
 
-   (*dri_interface->getUST)(&intel->swap_ust);
-   _math_matrix_ctr (&intel->ViewportMatrix);
 
-   driInitExtensions( ctx, card_extensions, GL_TRUE );
+   intel->batch = intel_batchbuffer_alloc(intel);
+   intel->last_swap_fence = NULL;
+   intel->first_swap_fence = NULL;
+
+   intel_bufferobj_init(intel);
+   intel_fbo_init(intel);
 
    if (intel->ctx.Mesa_DXTn) {
-     _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
-     _mesa_enable_extension( ctx, "GL_S3_s3tc" );
+      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
+      _mesa_enable_extension(ctx, "GL_S3_s3tc");
    }
-   else if (driQueryOptionb (&intel->optionCache, "force_s3tc_enable")) {
-     _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
+   else if (driQueryOptionb(&intel->optionCache, "force_s3tc_enable")) {
+      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
    }
 
-/*    driInitTextureObjects( ctx, & intel->swapped, */
-/* 			  DRI_TEXMGR_DO_TEXTURE_1D | */
-/* 			  DRI_TEXMGR_DO_TEXTURE_2D |  */
-/* 			  DRI_TEXMGR_DO_TEXTURE_RECT ); */
-
-
-   intelInitBatchBuffer(&intel->ctx);
-   intel->prim.flush = intel_emit_invarient_state;
    intel->prim.primitive = ~0;
 
 
 #if DO_DEBUG
-   INTEL_DEBUG  = driParseDebugString( getenv( "INTEL_DEBUG" ),
-				       debug_control );
-   INTEL_DEBUG |= driParseDebugString( getenv( "INTEL_DEBUG" ),
-				       debug_control );
-#endif
-
-#ifndef VERBOSE
-   if (getenv("INTEL_VERBOSE"))
-      VERBOSE=1;
+   INTEL_DEBUG = driParseDebugString(getenv("INTEL_DEBUG"), debug_control);
 #endif
 
-   if (getenv("INTEL_NO_RAST") || 
-       getenv("INTEL_NO_RAST")) {
+   if (getenv("INTEL_NO_RAST")) {
       fprintf(stderr, "disabling 3D rasterization\n");
-      FALLBACK(intel, INTEL_FALLBACK_USER, 1); 
+      FALLBACK(intel, INTEL_FALLBACK_USER, 1);
    }
 
    return GL_TRUE;
 }
 
-void intelDestroyContext(__DRIcontextPrivate *driContextPriv)
+void
+intelDestroyContext(__DRIcontextPrivate * driContextPriv)
 {
-   intelContextPtr intel = (intelContextPtr) driContextPriv->driverPrivate;
+   struct intel_context *intel =
+      (struct intel_context *) driContextPriv->driverPrivate;
 
-   assert(intel); /* should never be null */
+   assert(intel);               /* should never be null */
    if (intel) {
-      GLboolean   release_texture_heaps;
+      GLboolean release_texture_heaps;
 
-      INTEL_FIREVERTICES( intel );
+      INTEL_FIREVERTICES(intel);
 
-      intel->vtbl.destroy( intel );
+      intel->vtbl.destroy(intel);
 
       release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
-      _swsetup_DestroyContext (&intel->ctx);
-      _tnl_DestroyContext (&intel->ctx);
-      _vbo_DestroyContext (&intel->ctx);
+      _swsetup_DestroyContext(&intel->ctx);
+      _tnl_DestroyContext(&intel->ctx);
+      _vbo_DestroyContext(&intel->ctx);
 
-      _swrast_DestroyContext (&intel->ctx);
-      intel->Fallback = 0;	/* don't call _swrast_Flush later */
+      _swrast_DestroyContext(&intel->ctx);
+      intel->Fallback = 0;      /* don't call _swrast_Flush later */
 
-      intelDestroyBatchBuffer(&intel->ctx);
-      
+      intel_batchbuffer_free(intel->batch);
 
-      if ( release_texture_heaps ) {
+      if (intel->last_swap_fence) {
+	 dri_fence_wait(intel->last_swap_fence);
+	 dri_fence_unreference(intel->last_swap_fence);
+	 intel->last_swap_fence = NULL;
+      }
+      if (intel->first_swap_fence) {
+	 dri_fence_wait(intel->first_swap_fence);
+	 dri_fence_unreference(intel->first_swap_fence);
+	 intel->first_swap_fence = NULL;
+      }
+
+
+      if (release_texture_heaps) {
          /* This share group is about to go away, free our private
           * texture object data.
           */
-         int i;
-
-         for ( i = 0 ; i < intel->nr_heaps ; i++ ) {
-	    driDestroyTextureHeap( intel->texture_heaps[ i ] );
-	    intel->texture_heaps[ i ] = NULL;
-         }
-
-	 assert( is_empty_list( & intel->swapped ) );
+         if (INTEL_DEBUG & DEBUG_TEXTURE)
+            fprintf(stderr, "do something to free texture heaps\n");
       }
 
       /* free the Mesa context */
-      _mesa_destroy_context(&intel->ctx);
+      _mesa_free_context_data(&intel->ctx);
    }
 }
 
-void intelSetFrontClipRects( intelContextPtr intel )
+GLboolean
+intelUnbindContext(__DRIcontextPrivate * driContextPriv)
 {
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-
-   if (!dPriv) return;
-
-   intel->numClipRects = dPriv->numClipRects;
-   intel->pClipRects = dPriv->pClipRects;
-   intel->drawX = dPriv->x;
-   intel->drawY = dPriv->y;
+   return GL_TRUE;
 }
 
-
-void intelSetBackClipRects( intelContextPtr intel )
+GLboolean
+intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
+                 __DRIdrawablePrivate * driDrawPriv,
+                 __DRIdrawablePrivate * driReadPriv)
 {
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
 
-   if (!dPriv) return;
-
-   if (intel->sarea->pf_enabled == 0 && dPriv->numBackClipRects == 0) {
-      intel->numClipRects = dPriv->numClipRects;
-      intel->pClipRects = dPriv->pClipRects;
-      intel->drawX = dPriv->x;
-      intel->drawY = dPriv->y;
-   } else {
-      intel->numClipRects = dPriv->numBackClipRects;
-      intel->pClipRects = dPriv->pBackClipRects;
-      intel->drawX = dPriv->backX;
-      intel->drawY = dPriv->backY;
-      
-      if (dPriv->numBackClipRects == 1 &&
-	  dPriv->x == dPriv->backX &&
-	  dPriv->y == dPriv->backY) {
-      
-	 /* Repeat the calculation of the back cliprect dimensions here
-	  * as early versions of dri.a in the Xserver are incorrect.  Try
-	  * very hard not to restrict future versions of dri.a which
-	  * might eg. allocate truly private back buffers.
-	  */
-	 int x1, y1;
-	 int x2, y2;
-	 
-	 x1 = dPriv->x;
-	 y1 = dPriv->y;      
-	 x2 = dPriv->x + dPriv->w;
-	 y2 = dPriv->y + dPriv->h;
-	 
-	 if (x1 < 0) x1 = 0;
-	 if (y1 < 0) y1 = 0;
-	 if (x2 > intel->intelScreen->width) x2 = intel->intelScreen->width;
-	 if (y2 > intel->intelScreen->height) y2 = intel->intelScreen->height;
-
-	 if (x1 == dPriv->pBackClipRects[0].x1 &&
-	     y1 == dPriv->pBackClipRects[0].y1) {
-
-	    dPriv->pBackClipRects[0].x2 = x2;
-	    dPriv->pBackClipRects[0].y2 = y2;
-	 }
+   if (driContextPriv) {
+      struct intel_context *intel =
+         (struct intel_context *) driContextPriv->driverPrivate;
+      struct intel_framebuffer *intel_fb =
+	 (struct intel_framebuffer *) driDrawPriv->driverPrivate;
+      GLframebuffer *readFb = (GLframebuffer *) driReadPriv->driverPrivate;
+
+
+      /* XXX FBO temporary fix-ups! */
+      /* if the renderbuffers don't have regions, init them from the context */
+      {
+         struct intel_renderbuffer *irbDepth
+            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
+         struct intel_renderbuffer *irbStencil
+            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
+
+         if (intel_fb->color_rb[0] && !intel_fb->color_rb[0]->region) {
+            intel_region_reference(&intel_fb->color_rb[0]->region,
+				   intel->intelScreen->front_region);
+         }
+         if (intel_fb->color_rb[1] && !intel_fb->color_rb[1]->region) {
+            intel_region_reference(&intel_fb->color_rb[1]->region,
+				   intel->intelScreen->back_region);
+         }
+         if (intel_fb->color_rb[2] && !intel_fb->color_rb[2]->region) {
+            intel_region_reference(&intel_fb->color_rb[2]->region,
+				   intel->intelScreen->third_region);
+         }
+         if (irbDepth && !irbDepth->region) {
+            intel_region_reference(&irbDepth->region, intel->intelScreen->depth_region);
+         }
+         if (irbStencil && !irbStencil->region) {
+            intel_region_reference(&irbStencil->region, intel->intelScreen->depth_region);
+         }
       }
-   }
-}
 
+      /* set GLframebuffer size to match window, if needed */
+      driUpdateFramebufferSize(&intel->ctx, driDrawPriv);
 
-void intelWindowMoved( intelContextPtr intel )
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   GLframebuffer *drawFb = (GLframebuffer *) dPriv->driverPrivate;
-
-   if (!intel->ctx.DrawBuffer) {
-      intelSetFrontClipRects( intel );
-   }
-   else {
-      driUpdateFramebufferSize(&intel->ctx, dPriv);
-      switch (drawFb->_ColorDrawBufferMask[0]) {
-      case BUFFER_BIT_FRONT_LEFT:
-	 intelSetFrontClipRects( intel );
-	 break;
-      case BUFFER_BIT_BACK_LEFT:
-	 intelSetBackClipRects( intel );
-	 break;
-      default:
-	 /* glDrawBuffer(GL_NONE or GL_FRONT_AND_BACK): software fallback */
-	 intelSetFrontClipRects( intel );
+      if (driReadPriv != driDrawPriv) {
+	 driUpdateFramebufferSize(&intel->ctx, driReadPriv);
       }
-   }
 
-   if (drawFb->Width != dPriv->w || drawFb->Height != dPriv->h) {
-      /* update Mesa's notion of framebuffer/window size */
-      _mesa_resize_framebuffer(&intel->ctx, drawFb, dPriv->w, dPriv->h);
-      drawFb->Initialized = GL_TRUE; /* XXX remove someday */
-   }
+      _mesa_make_current(&intel->ctx, &intel_fb->Base, readFb);
 
-   /* Set state we know depends on drawable parameters:
-    */
-   {
-      GLcontext *ctx = &intel->ctx;
-
-      if (intel->intelScreen->driScrnPriv->ddxMinor >= 7) {
-	 drmI830Sarea *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 pipeA_rect = { .x1 = sarea->pipeA_x,
-					.x2 = sarea->pipeA_x + sarea->pipeA_w,
-					.y1 = sarea->pipeA_y,
-					.y2 = sarea->pipeA_y + sarea->pipeA_h };
-	 drm_clip_rect_t pipeB_rect = { .x1 = sarea->pipeB_x,
-					.x2 = sarea->pipeB_x + sarea->pipeB_w,
-					.y1 = sarea->pipeB_y,
-					.y2 = sarea->pipeB_y + sarea->pipeB_h };
-	 GLint areaA = driIntersectArea( drw_rect, pipeA_rect );
-	 GLint areaB = driIntersectArea( drw_rect, pipeB_rect );
-	 GLuint flags = intel->vblank_flags;
-
-	 if (areaB > areaA || (areaA == areaB && areaB > 0)) {
-	    flags = intel->vblank_flags | VBLANK_FLAG_SECONDARY;
-	 } else {
-	    flags = intel->vblank_flags & ~VBLANK_FLAG_SECONDARY;
+      /* The drawbuffer won't always be updated by _mesa_make_current: 
+       */
+      if (intel->ctx.DrawBuffer == &intel_fb->Base) {
+
+	 if (intel->driDrawable != driDrawPriv) {
+	    if (driDrawPriv->pdraw->swap_interval == (unsigned)-1) {
+	       int i;
+
+	       intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0)
+		  ? driGetDefaultVBlankFlags(&intel->optionCache)
+		 : VBLANK_FLAG_NO_IRQ;
+
+	       (*dri_interface->getUST) (&intel_fb->swap_ust);
+	       driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags,
+				     &intel_fb->vbl_seq);
+	       intel_fb->vbl_waited = intel_fb->vbl_seq;
+
+	       for (i = 0; i < (intel->intelScreen->third.handle ? 3 : 2); i++) {
+		  if (intel_fb->color_rb[i])
+		     intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_seq;
+	       }
+	    }
+	    intel->driDrawable = driDrawPriv;
+	    intelWindowMoved(intel);
 	 }
 
-	 if (flags != intel->vblank_flags) {
-	    intel->vblank_flags = flags;
-	    driGetCurrentVBlank(dPriv, intel->vblank_flags, &intel->vbl_seq);
-	 }
-      } else {
-	 intel->vblank_flags &= ~VBLANK_FLAG_SECONDARY;
+	 intel_draw_buffer(&intel->ctx, &intel_fb->Base);
       }
-
-      ctx->Driver.Scissor( ctx, ctx->Scissor.X, ctx->Scissor.Y,
-			   ctx->Scissor.Width, ctx->Scissor.Height );
-      
-      ctx->Driver.DepthRange( ctx, 
-			      ctx->Viewport.Near,
-			      ctx->Viewport.Far );
    }
-}
-
-GLboolean intelUnbindContext(__DRIcontextPrivate *driContextPriv)
-{
-   return GL_TRUE;
-}
-
-GLboolean intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-			  __DRIdrawablePrivate *driDrawPriv,
-			  __DRIdrawablePrivate *driReadPriv)
-{
-
-   if (driContextPriv) {
-      intelContextPtr intel = (intelContextPtr) driContextPriv->driverPrivate;
-
-      if ( intel->driDrawable != driDrawPriv ) {
-	 /* Shouldn't the readbuffer be stored also? */
-	 driDrawableInitVBlank( driDrawPriv, intel->vblank_flags,
-				&intel->vbl_seq );
-
-	 intel->driDrawable = driDrawPriv;
-	 intelWindowMoved( intel );
-      }
-
-      _mesa_make_current(&intel->ctx,
-			 (GLframebuffer *) driDrawPriv->driverPrivate,
-			 (GLframebuffer *) driReadPriv->driverPrivate);
-
-      intel->ctx.Driver.DrawBuffer( &intel->ctx, intel->ctx.Color.DrawBuffer[0] );
-   } else {
+   else {
       _mesa_make_current(NULL, NULL, NULL);
    }
 
    return GL_TRUE;
 }
 
-/**
- * Use the information in the sarea to update the screen parameters
- * related to screen rotation.
- */
 static void
-intelUpdateScreenRotation(intelContextPtr intel,
-                          __DRIscreenPrivate *sPriv,
-                          drmI830Sarea *sarea)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   intelRegion *colorBuf;
-
-   intelUnmapScreenRegions(intelScreen);
-
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   /* update the current hw offsets for the color and depth buffers */
-   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_BACK_LEFT)
-      colorBuf = &intelScreen->back;
-   else
-      colorBuf = &intelScreen->front;
-   intel->vtbl.update_color_z_regions(intel, colorBuf, &intelScreen->depth);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr, "ERROR Remapping screen regions!!!\n");
-   }
-}
-
-void intelGetLock( intelContextPtr intel, GLuint flags )
+intelContendedLock(struct intel_context *intel, GLuint flags)
 {
    __DRIdrawablePrivate *dPriv = intel->driDrawable;
    __DRIscreenPrivate *sPriv = intel->driScreen;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
-   drmI830Sarea * sarea = intel->sarea;
-   unsigned   i;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+   drmI830Sarea *sarea = intel->sarea;
 
    drmGetLock(intel->driFd, intel->hHWContext, flags);
 
+   if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - got contended lock\n", __progname);
+
    /* If the window moved, may need to set a new cliprect now.
     *
     * NOTE: This releases and regains the hw lock, so all state
@@ -679,98 +665,129 @@ void intelGetLock( intelContextPtr intel, GLuint flags )
    if (dPriv)
       DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
 
-   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
-      intelWindowMoved( intel );
-      intel->lastStamp = dPriv->lastStamp;
-   }
-
-   /* If we lost context, need to dump all registers to hardware.
-    * Note that we don't care about 2d contexts, even if they perform
-    * accelerated commands, so the DRI locking in the X server is even
-    * more broken than usual.
+   /* If the last consumer of the texture memory wasn't us, notify the fake
+    * bufmgr and record the new owner.  We should have the memory shared
+    * between contexts of a single fake bufmgr, but this will at least make
+    * things correct for now.
     */
+   if (!intel->intelScreen->ttm && sarea->texAge != intel->hHWContext) {
+      sarea->texAge = intel->hHWContext;
+      dri_bufmgr_fake_contended_lock_take(intel->intelScreen->bufmgr);
+      if (INTEL_DEBUG & DEBUG_BATCH)
+	 intel_decode_context_reset();
+   }
 
    if (sarea->width != intelScreen->width ||
        sarea->height != intelScreen->height ||
        sarea->rotation != intelScreen->current_rotation) {
-      intelUpdateScreenRotation(intel, sPriv, sarea);
 
-      /* This will drop the outstanding batchbuffer on the floor */
-      intel->batch.ptr -= (intel->batch.size - intel->batch.space);
-      intel->batch.space = intel->batch.size;
-      /* lose all primitives */
-      intel->prim.primitive = ~0;
-      intel->prim.start_ptr = 0;
-      intel->prim.flush = 0;
-      intel->vtbl.lost_hardware( intel ); 
+      intelUpdateScreenRotation(sPriv, sarea);
+   }
 
-      intel->lastStamp = 0; /* force window update */
+   if (sarea->width != intel->width ||
+       sarea->height != intel->height ||
+       sarea->rotation != intel->current_rotation) {
+      int numClipRects = intel->numClipRects;
 
-      /* Release batch buffer
+      /*
+       * FIXME: Really only need to do this when drawing to a
+       * common back- or front buffer.
        */
-      intelDestroyBatchBuffer(&intel->ctx);
-      intelInitBatchBuffer(&intel->ctx);
-      intel->prim.flush = intel_emit_invarient_state;
 
-      /* Still need to reset the global LRU?
+      /*
+       * This will essentially drop the outstanding batchbuffer on the floor.
        */
-      intel_driReinitTextureHeap( intel->texture_heaps[0], intel->intelScreen->tex.size );
+      intel->numClipRects = 0;
+
+      if (intel->Fallback)
+	 _swrast_flush(&intel->ctx);
+
+      INTEL_FIREVERTICES(intel);
+
+      if (intel->batch->map != intel->batch->ptr)
+	 intel_batchbuffer_flush(intel->batch);
+
+      intel->numClipRects = numClipRects;
+
+      /* force window update */
+      intel->lastStamp = 0;
+
+      intel->width = sarea->width;
+      intel->height = sarea->height;
+      intel->current_rotation = sarea->rotation;
    }
 
-   /* Shared texture managment - if another client has played with
-    * texture space, figure out which if any of our textures have been
-    * ejected, and update our global LRU.
+   /* Drawable changed?
     */
-   for ( i = 0 ; i < intel->nr_heaps ; i++ ) {
-      DRI_AGE_TEXTURES( intel->texture_heaps[ i ] );
+   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
+      intelWindowMoved(intel);
+      intel->lastStamp = dPriv->lastStamp;
    }
 }
 
 
-void intelSwapBuffers( __DRIdrawablePrivate *dPriv )
+
+/* Lock the hardware and validate our state.  
+ */
+void LOCK_HARDWARE( struct intel_context *intel )
 {
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      intelContextPtr intel;
-      GLcontext *ctx;
-      intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-      ctx = &intel->ctx;
-      if (ctx->Visual.doubleBufferMode) {
-         intelScreenPrivate *screen = intel->intelScreen;
-	 _mesa_notifySwapBuffers( ctx );  /* flush pending rendering comands */
-	 if ( 0 /*intel->doPageFlip*/ ) { /* doPageFlip is never set !!! */
-	    intelPageFlip( dPriv );
-	 } else {
-	     intelCopyBuffer( dPriv, NULL );
-	 }
-         if (screen->current_rotation != 0) {
-            intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
-         }
-      }
-   } 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__);
-   }
+    char __ret=0;
+    struct intel_framebuffer *intel_fb = NULL;
+    struct intel_renderbuffer *intel_rb = NULL;
+    _glthread_LOCK_MUTEX(lockMutex);
+    assert(!intel->locked);
+
+    if (intel->driDrawable) {
+       intel_fb = intel->driDrawable->driverPrivate;
+
+       if (intel_fb)
+	  intel_rb =
+	     intel_get_renderbuffer(&intel_fb->Base,
+				    intel_fb->Base._ColorDrawBufferMask[0] ==
+				    BUFFER_BIT_FRONT_LEFT ? BUFFER_FRONT_LEFT :
+				    BUFFER_BACK_LEFT);
+    }
+
+    if (intel_rb && intel_fb->vblank_flags &&
+	!(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) &&
+	(intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
+	drmVBlank vbl;
+
+	vbl.request.type = DRM_VBLANK_ABSOLUTE;
+
+	if ( intel_fb->vblank_flags & VBLANK_FLAG_SECONDARY ) {
+	    vbl.request.type |= DRM_VBLANK_SECONDARY;
+	}
+
+	vbl.request.sequence = intel_rb->vbl_pending;
+	drmWaitVBlank(intel->driFd, &vbl);
+	intel_fb->vbl_waited = vbl.reply.sequence;
+    }
+
+    DRM_CAS(intel->driHwLock, intel->hHWContext,
+        (DRM_LOCK_HELD|intel->hHWContext), __ret);
+
+    if (__ret)
+        intelContendedLock( intel, 0 );
+
+    if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - locked\n", __progname);
+
+    intel->locked = 1;
 }
 
-void intelCopySubBuffer( __DRIdrawablePrivate *dPriv,
-			 int x, int y, int w, int h )
+
+  /* Unlock the hardware using the global current context 
+   */
+void UNLOCK_HARDWARE( struct intel_context *intel )
 {
-   if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) {
-      intelContextPtr intel;
-      GLcontext *ctx;
-      intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-      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__);
-   }
-}
+   intel->locked = 0;
+
+   DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
+
+   _glthread_UNLOCK_MUTEX(lockMutex);
+
+   if (INTEL_DEBUG & DEBUG_LOCK)
+      _mesa_printf("%s - unlocked\n", __progname);
+} 
+
diff --git a/src/mesa/drivers/dri/i915/intel_context.h b/src/mesa/drivers/dri/i915/intel_context.h
index 3b50107d73f..1e9ccd5cdc0 100644
--- a/src/mesa/drivers/dri/i915/intel_context.h
+++ b/src/mesa/drivers/dri/i915/intel_context.h
@@ -34,7 +34,6 @@
 #include "drm.h"
 #include "mm.h"
 #include "texmem.h"
-#include "vblank.h"
 
 #include "intel_screen.h"
 #include "i915_drm.h"
@@ -49,156 +48,177 @@
 #define DV_PF_565  (2<<8)
 #define DV_PF_8888 (3<<8)
 
-#define INTEL_CONTEXT(ctx)	((intelContextPtr)(ctx))
+struct intel_region;
+struct intel_context;
 
-typedef struct intel_context intelContext;
-typedef struct intel_context *intelContextPtr;
-typedef struct intel_texture_object *intelTextureObjectPtr;
-
-typedef void (*intel_tri_func)(intelContextPtr, intelVertex *, intelVertex *,
-							  intelVertex *);
-typedef void (*intel_line_func)(intelContextPtr, intelVertex *, intelVertex *);
-typedef void (*intel_point_func)(intelContextPtr, intelVertex *);
+typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
+                                intelVertex *, intelVertex *);
+typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
+                                 intelVertex *);
+typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
 
 #define INTEL_FALLBACK_DRAW_BUFFER	 0x1
 #define INTEL_FALLBACK_READ_BUFFER	 0x2
-#define INTEL_FALLBACK_USER		 0x4
-#define INTEL_FALLBACK_NO_BATCHBUFFER	 0x8
-#define INTEL_FALLBACK_NO_TEXMEM	 0x10
+#define INTEL_FALLBACK_DEPTH_BUFFER      0x4
+#define INTEL_FALLBACK_STENCIL_BUFFER    0x8
+#define INTEL_FALLBACK_USER		 0x10
 #define INTEL_FALLBACK_RENDERMODE	 0x20
 
-extern void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode );
+extern void intelFallback(struct intel_context *intel, GLuint bit,
+                          GLboolean mode);
 #define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
 
 
-#define INTEL_TEX_MAXLEVELS 10
-
+#define INTEL_WRITE_PART  0x1
+#define INTEL_WRITE_FULL  0x2
+#define INTEL_READ        0x4
 
 struct intel_texture_object
 {
-   driTextureObject    base;	/* the parent class */
-
-   GLuint texelBytes;
-   GLuint age;
-   GLuint Pitch;
-   GLuint Height;
-   GLuint TextureOffset;
-   GLubyte *BufAddr;   
-
-   GLuint min_level;
-   GLuint max_level;
-   GLuint depth_pitch;
-
-   struct {
-      const struct gl_texture_image *image;
-      GLuint offset;       /* into BufAddr */
-      GLuint height;
-      GLuint internalFormat;
-   } image[6][INTEL_TEX_MAXLEVELS];
-
-   GLuint dirty;
-   GLuint firstLevel,lastLevel;
+   struct gl_texture_object base;       /* The "parent" object */
+
+   /* The mipmap tree must include at least these levels once
+    * validated:
+    */
+   GLuint firstLevel;
+   GLuint lastLevel;
+
+   /* Offset for firstLevel image:
+    */
+   GLuint textureOffset;
+
+   /* On validation any active images held in main memory or in other
+    * regions will be copied to this region and the old storage freed.
+    */
+   struct intel_mipmap_tree *mt;
+
+   GLboolean imageOverride;
+   GLint depthOverride;
+   GLuint pitchOverride;
+};
+
+
+
+struct intel_texture_image
+{
+   struct gl_texture_image base;
+
+   /* These aren't stored in gl_texture_image 
+    */
+   GLuint level;
+   GLuint face;
+
+   /* If intelImage->mt != NULL, image data is stored here.
+    * Else if intelImage->base.Data != NULL, image is stored there.
+    * Else there is no image data.
+    */
+   struct intel_mipmap_tree *mt;
 };
 
 
+#define INTEL_MAX_FIXUP 64
+
 struct intel_context
 {
-   GLcontext ctx;		/* the parent class */
+   GLcontext ctx;               /* the parent class */
+
+   struct
+   {
+      void (*destroy) (struct intel_context * intel);
+      void (*emit_state) (struct intel_context * intel);
+      void (*lost_hardware) (struct intel_context * intel);
+      void (*update_texture_state) (struct intel_context * intel);
+
+      void (*render_start) (struct intel_context * intel);
+      void (*render_prevalidate) (struct intel_context * intel);
+      void (*set_draw_region) (struct intel_context * intel,
+                               struct intel_region * draw_region,
+                               struct intel_region * depth_region);
+
+        GLuint(*flush_cmd) (void);
+
+      void (*reduced_primitive_state) (struct intel_context * intel,
+                                       GLenum rprim);
+
+        GLboolean(*check_vertex_size) (struct intel_context * intel,
+                                       GLuint expected);
+
+
+      /* Metaops: 
+       */
+      void (*install_meta_state) (struct intel_context * intel);
+      void (*leave_meta_state) (struct intel_context * intel);
+
+      void (*meta_draw_region) (struct intel_context * intel,
+                                struct intel_region * draw_region,
+                                struct intel_region * depth_region);
 
-   struct {
-      void (*destroy)( intelContextPtr intel ); 
-      void (*emit_state)( intelContextPtr intel );
-      void (*lost_hardware)( intelContextPtr intel );
-      void (*update_texture_state)( intelContextPtr intel );
+      void (*meta_color_mask) (struct intel_context * intel, GLboolean);
 
-      void (*render_start)( intelContextPtr intel );
-      void (*set_color_region)( intelContextPtr intel, const intelRegion *reg );
-      void (*set_z_region)( intelContextPtr intel, const intelRegion *reg );
-      void (*update_color_z_regions)(intelContextPtr intel,
-                                     const intelRegion *colorRegion,
-                                     const intelRegion *depthRegion);
-      void (*emit_flush)( intelContextPtr intel );
-      void (*reduced_primitive_state)( intelContextPtr intel, GLenum rprim );
+      void (*meta_stencil_replace) (struct intel_context * intel,
+                                    GLuint mask, GLuint clear);
 
-      GLboolean (*check_vertex_size)( intelContextPtr intel, GLuint expected );
+      void (*meta_depth_replace) (struct intel_context * intel);
 
-      void (*clear_with_tris)( intelContextPtr intel, GLbitfield mask,
-			       GLboolean all, 
-			       GLint cx, GLint cy, GLint cw, GLint ch);
+      void (*meta_texture_blend_replace) (struct intel_context * intel);
 
-      void (*rotate_window)( intelContextPtr intel,
-                             __DRIdrawablePrivate *dPriv, GLuint srcBuf);
+      void (*meta_no_stencil_write) (struct intel_context * intel);
+      void (*meta_no_depth_write) (struct intel_context * intel);
+      void (*meta_no_texture) (struct intel_context * intel);
 
-      intelTextureObjectPtr (*alloc_tex_obj)( struct gl_texture_object *tObj );
+      void (*meta_import_pixel_state) (struct intel_context * intel);
+
+      GLboolean(*meta_tex_rect_source) (struct intel_context * intel,
+					dri_bo * buffer,
+					GLuint offset,
+					GLuint pitch,
+					GLuint height,
+					GLenum format, GLenum type);
+      void (*rotate_window) (struct intel_context * intel,
+                             __DRIdrawablePrivate * dPriv, GLuint srcBuf);
+
+      void (*assert_not_dirty) (struct intel_context *intel);
 
    } vtbl;
 
-   GLint refcount;   
+   GLint refcount;
    GLuint Fallback;
    GLuint NewGLState;
-   
-   struct {
-      GLuint start_offset;
-      GLint size;
-      GLint space;
-      GLubyte *ptr;
-      GLuint counter;
-      GLuint last_emit_state;
-      GLboolean contains_geometry;
-      const char *func;
-      GLuint last_swap;
-   } batch;
-      
-   struct {
-      void *ptr;
-      GLint size;
-      GLuint offset;
-      GLuint active_buf;
-      GLuint irq_emitted;
-   } alloc;
-
-   struct {
+
+   dri_fence *last_swap_fence;
+   dri_fence *first_swap_fence;
+
+   struct intel_batchbuffer *batch;
+
+   struct
+   {
+      GLuint id;
       GLuint primitive;
-      GLubyte *start_ptr;      
-      void (*flush)( GLcontext * );
+      GLubyte *start_ptr;
+      void (*flush) (struct intel_context *);
    } prim;
 
    GLboolean locked;
+   char *prevLockFile;
+   int prevLockLine;
 
-   GLubyte clear_red;
-   GLubyte clear_green;
-   GLubyte clear_blue;
-   GLubyte clear_alpha;
-   GLuint ClearColor;
-   GLuint ClearDepth;
+   GLuint ClearColor565;
+   GLuint ClearColor8888;
 
+   /* Offsets of fields within the current vertex:
+    */
    GLuint coloroffset;
    GLuint specoffset;
-
-   /* Support for duplicating XYZW as WPOS parameter (crutch for I915).
-    */
    GLuint wpos_offset;
    GLuint wpos_size;
 
    struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
    GLuint vertex_attr_count;
 
-   GLfloat depth_scale;
-   GLfloat polygon_offset_scale; /* dependent on depth_scale, bpp */
-   GLuint depth_clear_mask;
-   GLuint stencil_clear_mask;
+   GLfloat polygon_offset_scale;        /* dependent on depth_scale, bpp */
 
-   GLboolean hw_stencil;
    GLboolean hw_stipple;
-   
-   /* Texture object bookkeeping
-    */
-   GLuint                nr_heaps;
-   driTexHeap          * texture_heaps[1];
-   driTextureObject      swapped;
-   GLuint                lastStamp;
-
-   struct intel_texture_object *CurrentTexObj[MAX_TEXTURE_UNITS];
+   GLboolean strict_conformance;
 
    /* State for intelvb.c and inteltris.c.
     */
@@ -207,8 +227,15 @@ struct intel_context
    GLenum render_primitive;
    GLenum reduced_primitive;
    GLuint vertex_size;
-   unsigned char *verts;	   /* points to tnl->clipspace.vertex_buf */
-
+   GLubyte *verts;              /* points to tnl->clipspace.vertex_buf */
+
+#if 0
+   struct intel_region *front_region;   /* XXX FBO: obsolete */
+   struct intel_region *rotated_region; /* XXX FBO: obsolete */
+   struct intel_region *back_region;    /* XXX FBO: obsolete */
+   struct intel_region *draw_region;    /* XXX FBO: rename to color_region */
+   struct intel_region *depth_region;   /**< currently bound depth/Z region */
+#endif
 
    /* Fallback rasterization functions 
     */
@@ -216,27 +243,18 @@ struct intel_context
    intel_line_func draw_line;
    intel_tri_func draw_tri;
 
-   /* Drawing buffer state
+   /* These refer to the current drawing buffer:
     */
-   intelRegion *drawRegion;  /* current drawing buffer */
-   intelRegion *readRegion;  /* current reading buffer */
-
-   int drawX;			/* origin of drawable in draw buffer */
-   int drawY;
-   GLuint numClipRects;		/* cliprects for that buffer */
+   int drawX, drawY;            /**< origin of drawing area within region */
+   GLuint numClipRects;         /**< cliprects for drawing */
    drm_clip_rect_t *pClipRects;
+   drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
 
-   int dirtyAge;
    int perf_boxes;
 
    GLuint do_usleeps;
    int do_irqs;
    GLuint irqsEmitted;
-   drm_i915_irq_wait_t iw;
-
-   GLboolean scissor;
-   drm_clip_rect_t draw_rect;
-   drm_clip_rect_t scissor_rect;
 
    drm_context_t hHWContext;
    drmLock *driHwLock;
@@ -244,118 +262,31 @@ struct intel_context
 
    __DRIdrawablePrivate *driDrawable;
    __DRIscreenPrivate *driScreen;
-   intelScreenPrivate *intelScreen; 
-   drmI830Sarea *sarea; 
+   intelScreenPrivate *intelScreen;
+   drmI830Sarea *sarea;
+
+   GLuint lastStamp;
 
    /**
     * Configuration cache
     */
    driOptionCache optionCache;
 
-   /* VBI
-    */
-   GLuint vbl_seq;
-   GLuint vblank_flags;
-
-   int64_t swap_ust;
-   int64_t swap_missed_ust;
-
-   GLuint swap_count;
-   GLuint swap_missed_count;
+  /* Rotation. Need to match that of the
+   * current screen.
+   */
 
-   GLuint swap_scheduled;
+  int width;
+  int height;
+  int current_rotation;
 };
 
-
-#define DEBUG_LOCKING	1
-
-#if DEBUG_LOCKING
-extern char *prevLockFile;
-extern int prevLockLine;
-
-#define DEBUG_LOCK()							\
-   do {									\
-      prevLockFile = (__FILE__);					\
-      prevLockLine = (__LINE__);					\
-   } while (0)
-
-#define DEBUG_RESET()							\
-   do {									\
-      prevLockFile = 0;							\
-      prevLockLine = 0;							\
-   } while (0)
-
-/* Slightly less broken way of detecting recursive locking in a
- * threaded environment.  The right way to do this would be to make
- * prevLockFile, prevLockLine thread-local.
- *
- * This technique instead checks to see if the same context is
- * requesting the lock twice -- this will not catch application
- * breakages where the same context is active in two different threads
- * at once, but it will catch driver breakages (recursive locking) in
- * threaded apps.
+/* These are functions now:
  */
-#define DEBUG_CHECK_LOCK()						\
-   do {									\
-      if ( *((volatile int *)intel->driHwLock) == 			\
-	   (DRM_LOCK_HELD | intel->hHWContext) ) {			\
-	 fprintf( stderr,						\
-		  "LOCK SET!\n\tPrevious %s:%d\n\tCurrent: %s:%d\n",	\
-		  prevLockFile, prevLockLine, __FILE__, __LINE__ );	\
-	 abort();							\
-      }									\
-   } while (0)
+void LOCK_HARDWARE( struct intel_context *intel );
+void UNLOCK_HARDWARE( struct intel_context *intel );
 
-#else
-
-#define DEBUG_LOCK()
-#define DEBUG_RESET()
-#define DEBUG_CHECK_LOCK()
-
-#endif
-
-
-
-
-/* Lock the hardware and validate our state.  
- */
-#define LOCK_HARDWARE( intel )				\
-do {							\
-    char __ret=0;					\
-    DEBUG_CHECK_LOCK();					\
-    assert(!(intel)->locked);				\
-    if ((intel)->swap_scheduled) {			\
-        drmVBlank vbl;					\
-        vbl.request.type = DRM_VBLANK_ABSOLUTE;		\
-        if ((intel)->vblank_flags &			\
-            VBLANK_FLAG_SECONDARY) {			\
-            vbl.request.type |= DRM_VBLANK_SECONDARY;	\
-        }						\
-        vbl.request.sequence = (intel)->vbl_seq;	\
-        drmWaitVBlank((intel)->driFd, &vbl);		\
-        (intel)->swap_scheduled = 0;			\
-    }							\
-    DRM_CAS((intel)->driHwLock, (intel)->hHWContext,	\
-        (DRM_LOCK_HELD|(intel)->hHWContext), __ret);	\
-    if (__ret)						\
-        intelGetLock( (intel), 0 );			\
-      DEBUG_LOCK();					\
-    (intel)->locked = 1;				\
-}while (0)
- 
-  
-  /* Unlock the hardware using the global current context 
-   */
-#define UNLOCK_HARDWARE(intel)						\
-do {									\
-   intel->locked = 0;							\
-   if (0) { 								\
-      intel->perf_boxes |= intel->sarea->perf_boxes;  			\
-      intel->sarea->perf_boxes = 0;					\
-   }									\
-   DRM_UNLOCK((intel)->driFd, (intel)->driHwLock, (intel)->hHWContext);	\
-   DEBUG_RESET();							\
-} while (0)
+extern char *__progname;
 
 
 #define SUBPIXEL_X 0.125
@@ -364,7 +295,7 @@ do {									\
 #define INTEL_FIREVERTICES(intel)		\
 do {						\
    if ((intel)->prim.flush)			\
-      (intel)->prim.flush(&(intel)->ctx);		\
+      (intel)->prim.flush(intel);		\
 } while (0)
 
 /* ================================================================
@@ -385,34 +316,26 @@ do {						\
   ((a<<24) | (r<<16) | (g<<8) | b)
 
 
-#define INTEL_PACKCOLOR(format, r,  g,  b, a)		\
-(format == DV_PF_555 ? INTEL_PACKCOLOR1555(r,g,b,a) :	\
- (format == DV_PF_565 ? INTEL_PACKCOLOR565(r,g,b) :	\
-  (format == DV_PF_8888 ? INTEL_PACKCOLOR8888(r,g,b,a) :	\
-   0)))
-
-
 
 /* ================================================================
  * From linux kernel i386 header files, copes with odd sizes better
  * than COPY_DWORDS would:
+ * XXX Put this in src/mesa/main/imports.h ???
  */
 #if defined(i386) || defined(__i386__)
-static __inline__ void * __memcpy(void * to, const void * from, size_t n)
+static INLINE void *
+__memcpy(void *to, const void *from, size_t n)
 {
    int d0, d1, d2;
-   __asm__ __volatile__(
-      "rep ; movsl\n\t"
-      "testb $2,%b4\n\t"
-      "je 1f\n\t"
-      "movsw\n"
-      "1:\ttestb $1,%b4\n\t"
-      "je 2f\n\t"
-      "movsb\n"
-      "2:"
-      : "=&c" (d0), "=&D" (d1), "=&S" (d2)
-      :"0" (n/4), "q" (n),"1" ((long) to),"2" ((long) from)
-      : "memory");
+   __asm__ __volatile__("rep ; movsl\n\t"
+                        "testb $2,%b4\n\t"
+                        "je 1f\n\t"
+                        "movsw\n"
+                        "1:\ttestb $1,%b4\n\t"
+                        "je 2f\n\t"
+                        "movsb\n" "2:":"=&c"(d0), "=&D"(d1), "=&S"(d2)
+                        :"0"(n / 4), "q"(n), "1"((long) to), "2"((long) from)
+                        :"memory");
    return (to);
 }
 #else
@@ -434,16 +357,18 @@ extern int INTEL_DEBUG;
 #define DEBUG_TEXTURE	0x1
 #define DEBUG_STATE	0x2
 #define DEBUG_IOCTL	0x4
-#define DEBUG_PRIMS	0x8
-#define DEBUG_VERTS	0x10
+#define DEBUG_BLIT	0x8
+#define DEBUG_MIPTREE   0x10
 #define DEBUG_FALLBACKS	0x20
 #define DEBUG_VERBOSE	0x40
-#define DEBUG_DRI       0x80
-#define DEBUG_DMA       0x100
-#define DEBUG_SANITY    0x200
-#define DEBUG_SYNC      0x400
-#define DEBUG_SLEEP     0x800
-#define DEBUG_PIXEL     0x1000
+#define DEBUG_BATCH     0x80
+#define DEBUG_PIXEL     0x100
+#define DEBUG_BUFMGR    0x200
+#define DEBUG_REGION    0x400
+#define DEBUG_FBO       0x800
+#define DEBUG_LOCK      0x1000
+
+#define DBG(...)  do { if (INTEL_DEBUG & FILE_DEBUG_FLAG) _mesa_printf(__VA_ARGS__); } while(0)
 
 
 #define PCI_CHIP_845_G			0x2562
@@ -464,26 +389,24 @@ extern int INTEL_DEBUG;
  * intel_context.c:
  */
 
-extern void intelInitDriverFunctions( struct dd_function_table *functions );
+extern GLboolean intelInitContext(struct intel_context *intel,
+                                  const __GLcontextModes * mesaVis,
+                                  __DRIcontextPrivate * driContextPriv,
+                                  void *sharedContextPrivate,
+                                  struct dd_function_table *functions);
 
-extern GLboolean intelInitContext( intelContextPtr intel, 
-				   const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate,
-				   struct dd_function_table *functions );
+extern void intelGetLock(struct intel_context *intel, GLuint flags);
 
-extern void intelGetLock(intelContextPtr intel, GLuint flags);
-extern void intelSetBackClipRects(intelContextPtr intel);
-extern void intelSetFrontClipRects(intelContextPtr intel);
-extern void intelWindowMoved( intelContextPtr intel );
+extern void intelFinish(GLcontext * ctx);
+extern void intelFlush(GLcontext * ctx);
 
-extern const GLubyte *intelGetString( GLcontext *ctx, GLenum name );
+extern void intelInitDriverFunctions(struct dd_function_table *functions);
 
 
 /* ================================================================
  * intel_state.c:
  */
-extern void intelInitStateFuncs( struct dd_function_table *functions );
+extern void intelInitStateFuncs(struct dd_function_table *functions);
 
 #define COMPAREFUNC_ALWAYS		0
 #define COMPAREFUNC_NEVER		0x1
@@ -537,27 +460,39 @@ extern void intelInitStateFuncs( struct dd_function_table *functions );
 #define BLENDFACT_INV_CONST_ALPHA	0x0f
 #define BLENDFACT_MASK          	0x0f
 
+#define MI_BATCH_BUFFER_END 	(0xA<<23)
 
-extern int intel_translate_compare_func( GLenum func );
-extern int intel_translate_stencil_op( GLenum op );
-extern int intel_translate_blend_factor( GLenum factor );
-extern int intel_translate_logic_op( GLenum opcode );
 
+extern int intel_translate_compare_func(GLenum func);
+extern int intel_translate_stencil_op(GLenum op);
+extern int intel_translate_blend_factor(GLenum factor);
+extern int intel_translate_logic_op(GLenum opcode);
 
-/* ================================================================
- * intel_ioctl.c:
+
+/*======================================================================
+ * Inline conversion functions.  
+ * These are better-typed than the macros used previously:
  */
-extern void intel_dump_batchbuffer( long offset,
-				    int *ptr,
-				    int count );
+static INLINE struct intel_context *
+intel_context(GLcontext * ctx)
+{
+   return (struct intel_context *) ctx;
+}
 
+static INLINE struct intel_texture_object *
+intel_texture_object(struct gl_texture_object *obj)
+{
+   return (struct intel_texture_object *) obj;
+}
 
-/* ================================================================
- * intel_pixel.c:
- */	
-extern void intelInitPixelFuncs( struct dd_function_table *functions );
+static INLINE struct intel_texture_image *
+intel_texture_image(struct gl_texture_image *img)
+{
+   return (struct intel_texture_image *) img;
+}
 
+extern struct intel_renderbuffer *intel_renderbuffer(struct gl_renderbuffer
+                                                     *rb);
 
 
 #endif
-
diff --git a/src/mesa/drivers/dri/i915/intel_decode.c b/src/mesa/drivers/dri/i915/intel_decode.c
new file mode 120000
index 00000000000..f671b6cbb13
--- /dev/null
+++ b/src/mesa/drivers/dri/i915/intel_decode.c
@@ -0,0 +1 @@
+../intel/intel_decode.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i915tex/intel_depthstencil.c b/src/mesa/drivers/dri/i915/intel_depthstencil.c
index d269a85a3c9..d269a85a3c9 100644
--- a/src/mesa/drivers/dri/i915tex/intel_depthstencil.c
+++ b/src/mesa/drivers/dri/i915/intel_depthstencil.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_depthstencil.h b/src/mesa/drivers/dri/i915/intel_depthstencil.h
index 2d3fc48b3a3..2d3fc48b3a3 100644
--- a/src/mesa/drivers/dri/i915tex/intel_depthstencil.h
+++ b/src/mesa/drivers/dri/i915/intel_depthstencil.h
diff --git a/src/mesa/drivers/dri/i915tex/intel_fbo.c b/src/mesa/drivers/dri/i915/intel_fbo.c
index 6f99f401c7c..6f99f401c7c 100644
--- a/src/mesa/drivers/dri/i915tex/intel_fbo.c
+++ b/src/mesa/drivers/dri/i915/intel_fbo.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_fbo.h b/src/mesa/drivers/dri/i915/intel_fbo.h
index 963f5e706f4..411d6342317 100644
--- a/src/mesa/drivers/dri/i915tex/intel_fbo.h
+++ b/src/mesa/drivers/dri/i915/intel_fbo.h
@@ -44,7 +44,7 @@ struct intel_framebuffer
    /* Drawable page flipping state */
    GLboolean pf_active;
    GLuint pf_seq;
-   GLint pf_pipes;
+   GLint pf_planes;
    GLint pf_current_page;
    GLint pf_num_pages;
 
diff --git a/src/mesa/drivers/dri/i915/intel_ioctl.c b/src/mesa/drivers/dri/i915/intel_ioctl.c
index ede3b6378fe..6e76737c783 100644
--- a/src/mesa/drivers/dri/i915/intel_ioctl.c
+++ b/src/mesa/drivers/dri/i915/intel_ioctl.c
@@ -38,622 +38,98 @@
 #include "intel_context.h"
 #include "intel_ioctl.h"
 #include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_regions.h"
 #include "drm.h"
 
-u_int32_t intelGetLastFrame (intelContextPtr intel) 
-{
-   int ret;
-   u_int32_t frame;
-   drm_i915_getparam_t gp;
-   
-   gp.param = I915_PARAM_LAST_DISPATCH;
-   gp.value = (int *)&frame;
-   ret = drmCommandWriteRead( intel->driFd, DRM_I915_GETPARAM,
-			      &gp, sizeof(gp) );
-   return frame;
-}
+#define FILE_DEBUG_FLAG DEBUG_IOCTL
 
-int intelEmitIrqLocked( intelContextPtr intel )
+int
+intelEmitIrqLocked(intelScreenPrivate *intelScreen)
 {
    drmI830IrqEmit ie;
    int ret, seq;
-      
-   assert(((*(int *)intel->driHwLock) & ~DRM_LOCK_CONT) == 
-	  (DRM_LOCK_HELD|intel->hHWContext));
 
    ie.irq_seq = &seq;
-	 
-   ret = drmCommandWriteRead( intel->driFd, DRM_I830_IRQ_EMIT, 
-			      &ie, sizeof(ie) );
-   if ( ret ) {
-      fprintf( stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret );
+
+   ret = drmCommandWriteRead(intelScreen->driScrnPriv->fd,
+			     DRM_I830_IRQ_EMIT, &ie, sizeof(ie));
+   if (ret) {
+      fprintf(stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret);
       exit(1);
    }
-   
-   if (0)
-      fprintf(stderr, "%s -->  %d\n", __FUNCTION__, seq );
+
+   DBG("%s -->  %d\n", __FUNCTION__, seq);
 
    return seq;
 }
 
-void intelWaitIrq( intelContextPtr intel, int seq )
+void
+intelWaitIrq(intelScreenPrivate *intelScreen, int seq)
 {
+   drm_i915_irq_wait_t iw;
    int ret;
-      
-   if (0)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, seq );
 
-   intel->iw.irq_seq = seq;
-	 
+   DBG("%s %d\n", __FUNCTION__, seq);
+
+   iw.irq_seq = seq;
+
    do {
-     ret = drmCommandWrite( intel->driFd, DRM_I830_IRQ_WAIT, &intel->iw, sizeof(intel->iw) );
+      ret = drmCommandWrite(intelScreen->driScrnPriv->fd,
+			    DRM_I830_IRQ_WAIT, &iw, sizeof(iw));
    } while (ret == -EAGAIN || ret == -EINTR);
 
-   if ( ret ) {
-      fprintf( stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret );
-      if (0)
-	 intel_dump_batchbuffer( intel->alloc.offset,
-				 intel->alloc.ptr,
-				 intel->alloc.size );
+   if (ret) {
+      fprintf(stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret);
       exit(1);
    }
 }
 
 
-
-static void age_intel( intelContextPtr intel, int age )
-{
-   GLuint i;
-
-   for (i = 0 ; i < MAX_TEXTURE_UNITS ; i++)
-      if (intel->CurrentTexObj[i]) 
-	 intel->CurrentTexObj[i]->age = age;
-}
-
-void intel_dump_batchbuffer( long offset,
-			     int *ptr,
-			     int count )
-{
-   int i;
-   fprintf(stderr, "\n\n\nSTART BATCH (%d dwords):\n", count);
-   for (i = 0; i < count/4; i += 4) 
-      fprintf(stderr, "\t0x%x: 0x%08x 0x%08x 0x%08x 0x%08x\n", 
-	      (unsigned int)offset + i*4, ptr[i], ptr[i+1], ptr[i+2], ptr[i+3]);
-   fprintf(stderr, "END BATCH\n\n\n");
-}
-
-void intelRefillBatchLocked( intelContextPtr intel, GLboolean allow_unlock )
-{
-   GLuint last_irq = intel->alloc.irq_emitted;
-   GLuint half = intel->alloc.size / 2;
-   GLuint buf = (intel->alloc.active_buf ^= 1);
-
-   intel->alloc.irq_emitted = intelEmitIrqLocked( intel );
-
-   if (last_irq) {
-      if (allow_unlock) UNLOCK_HARDWARE( intel ); 
-      intelWaitIrq( intel, last_irq );
-      if (allow_unlock) LOCK_HARDWARE( intel ); 
-   }
-
-   if (0)
-      fprintf(stderr, "%s: now using half %d\n", __FUNCTION__, buf);
-
-   intel->batch.start_offset = intel->alloc.offset + buf * half;
-   intel->batch.ptr = (unsigned char *)intel->alloc.ptr + buf * half;
-   intel->batch.size = half - 8;
-   intel->batch.space = half - 8;
-   assert(intel->batch.space >= 0);
-}
-
-#define MI_BATCH_BUFFER_END 	(0xA<<23)
-
-
-void intelFlushBatchLocked( intelContextPtr intel, 
-			    GLboolean ignore_cliprects,
-			    GLboolean refill,
-			    GLboolean allow_unlock)
+void
+intel_batch_ioctl(struct intel_context *intel,
+                  GLuint start_offset,
+                  GLuint used,
+                  GLboolean ignore_cliprects, GLboolean allow_unlock)
 {
    drmI830BatchBuffer batch;
 
    assert(intel->locked);
+   assert(used);
 
-   if (0)
-      fprintf(stderr, "%s used %d of %d offset %x..%x refill %d (started in %s)\n",
-	      __FUNCTION__, 
-	      (intel->batch.size - intel->batch.space), 
-	      intel->batch.size,
-	      intel->batch.start_offset,
-	      intel->batch.start_offset + 
-	      (intel->batch.size - intel->batch.space), 
-	      refill,
-	      intel->batch.func);
+   DBG("%s used %d offset %x..%x ignore_cliprects %d\n",
+       __FUNCTION__,
+       used, start_offset, start_offset + used, ignore_cliprects);
 
    /* Throw away non-effective packets.  Won't work once we have
     * hardware contexts which would preserve statechanges beyond a
     * single buffer.
     */
-   if (intel->numClipRects == 0 && !ignore_cliprects) {
-      
-      /* Without this yeild, an application with no cliprects can hog
-       * the hardware.  Without unlocking, the effect is much worse -
-       * effectively a lock-out of other contexts.
-       */
-      if (allow_unlock) {
-	 UNLOCK_HARDWARE( intel );
-	 sched_yield();
-	 LOCK_HARDWARE( intel );
-      }
-
-      /* Note that any state thought to have been emitted actually
-       * hasn't:
-       */
-      intel->batch.ptr -= (intel->batch.size - intel->batch.space);
-      intel->batch.space = intel->batch.size;
-      intel->vtbl.lost_hardware( intel ); 
-   }
-
-   if (intel->batch.space != intel->batch.size) {
-
-      if (intel->sarea->ctxOwner != intel->hHWContext) {
-	 intel->perf_boxes |= I830_BOX_LOST_CONTEXT;
-	 intel->sarea->ctxOwner = intel->hHWContext;
-      }
-
-      batch.start = intel->batch.start_offset;
-      batch.used = intel->batch.size - intel->batch.space;
-      batch.cliprects = intel->pClipRects;
-      batch.num_cliprects = ignore_cliprects ? 0 : intel->numClipRects;
-      batch.DR1 = 0;
-      batch.DR4 = ((((GLuint)intel->drawX) & 0xffff) | 
-		   (((GLuint)intel->drawY) << 16));
-      
-      if (intel->alloc.offset) {
-	 if ((batch.used & 0x4) == 0) {
-	    ((int *)intel->batch.ptr)[0] = 0;
-	    ((int *)intel->batch.ptr)[1] = MI_BATCH_BUFFER_END;
-	    batch.used += 0x8;
-	    intel->batch.ptr += 0x8;
-	 }
-	 else {
-	    ((int *)intel->batch.ptr)[0] = MI_BATCH_BUFFER_END;
-	    batch.used += 0x4;
-	    intel->batch.ptr += 0x4;
-	 }      
-      }
-
-      if (0)
- 	 intel_dump_batchbuffer( batch.start,
-				 (int *)(intel->batch.ptr - batch.used),
-				 batch.used );
-
-      intel->batch.start_offset += batch.used;
-      intel->batch.size -= batch.used;
-
-      if (intel->batch.size < 8) {
-	 refill = GL_TRUE;
-	 intel->batch.space = intel->batch.size = 0;
-      }
-      else {
-	 intel->batch.size -= 8;
-	 intel->batch.space = intel->batch.size;
-      }
-
-
-      assert(intel->batch.space >= 0);
-      assert(batch.start >= intel->alloc.offset);
-      assert(batch.start < intel->alloc.offset + intel->alloc.size);
-      assert(batch.start + batch.used > intel->alloc.offset);
-      assert(batch.start + batch.used <= 
-	     intel->alloc.offset + intel->alloc.size);
-
-
-      if (intel->alloc.offset) {
-	 if (drmCommandWrite (intel->driFd, DRM_I830_BATCHBUFFER, &batch, 
-			      sizeof(batch))) {
-	    fprintf(stderr, "DRM_I830_BATCHBUFFER: %d\n",  -errno);
-	    UNLOCK_HARDWARE(intel);
-	    exit(1);
-	 }
-      } else {
-	 drmI830CmdBuffer cmd;
-	 cmd.buf = (char *)intel->alloc.ptr + batch.start;
-	 cmd.sz = batch.used;
-	 cmd.DR1 = batch.DR1;
-	 cmd.DR4 = batch.DR4;
-	 cmd.num_cliprects = batch.num_cliprects;
-	 cmd.cliprects = batch.cliprects;
-	 
-	 if (drmCommandWrite (intel->driFd, DRM_I830_CMDBUFFER, &cmd, 
-			      sizeof(cmd))) {
-	    fprintf(stderr, "DRM_I830_CMDBUFFER: %d\n",  -errno);
-	    UNLOCK_HARDWARE(intel);
-	    exit(1);
-	 }
-      }	 
-
-      
-      age_intel(intel, intel->sarea->last_enqueue);
-
-      /* FIXME: use hardware contexts to avoid 'losing' hardware after
-       * each buffer flush.
-       */
-      if (intel->batch.contains_geometry) 
-	 assert(intel->batch.last_emit_state == intel->batch.counter);
-
-      intel->batch.counter++;
-      intel->batch.contains_geometry = 0;
-      intel->batch.func = 0;
-      intel->vtbl.lost_hardware( intel );
-   }
-
-   if (refill)
-      intelRefillBatchLocked( intel, allow_unlock );
-}
-
-void intelFlushBatch( intelContextPtr intel, GLboolean refill )
-{
-   if (intel->locked) {
-      intelFlushBatchLocked( intel, GL_FALSE, refill, GL_FALSE );
-   } 
-   else {
-      LOCK_HARDWARE(intel);
-      intelFlushBatchLocked( intel, GL_FALSE, refill, GL_TRUE );
-      UNLOCK_HARDWARE(intel);
-   }
-}
-
-
-void intelWaitForIdle( intelContextPtr intel )
-{   
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   intel->vtbl.emit_flush( intel );
-   intelFlushBatch( intel, GL_TRUE );
-
-   /* Use an irq to wait for dma idle -- Need to track lost contexts
-    * to shortcircuit consecutive calls to this function:
-    */
-   intelWaitIrq( intel, intel->alloc.irq_emitted );
-   intel->alloc.irq_emitted = 0;
-}
-
-
-/**
- * Check if we need to rotate/warp the front color buffer to the
- * rotated screen.  We generally need to do this when we get a glFlush
- * or glFinish after drawing to the front color buffer.
- */
-static void
-intelCheckFrontRotate(GLcontext *ctx)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
-      intelScreenPrivate *screen = intel->intelScreen;
-      if (screen->current_rotation != 0) {
-         __DRIdrawablePrivate *dPriv = intel->driDrawable;
-         intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
-      }
-   }
-}
-
-
-/**
- * NOT directly called via glFlush.
- */
-void intelFlush( GLcontext *ctx )
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-
-   if (intel->Fallback)
-      _swrast_flush( ctx );
-
-   INTEL_FIREVERTICES( intel );
-
-   if (intel->batch.size != intel->batch.space)
-      intelFlushBatch( intel, GL_FALSE );
-}
-
-
-/**
- * Called via glFlush.
- */
-void intelglFlush( GLcontext *ctx )
-{
-   intelFlush(ctx);
-   intelCheckFrontRotate(ctx);
-}
-
-
-void intelFinish( GLcontext *ctx  ) 
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   intelFlush( ctx );
-   intelWaitForIdle( intel );
-   intelCheckFrontRotate(ctx);
-}
-
-
-void intelClear(GLcontext *ctx, GLbitfield mask)
-{
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   const GLuint colorMask = *((GLuint *) &ctx->Color.ColorMask);
-   GLbitfield tri_mask = 0;
-   GLbitfield blit_mask = 0;
-   GLbitfield swrast_mask = 0;
-
-   if (0)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Take care of cliprects, which are handled differently for
-    * clears, etc.
-    */
-   intelFlush( &intel->ctx );
-
-   if (mask & BUFFER_BIT_FRONT_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_FRONT_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_FRONT_LEFT;
-      }
-   }
-
-   if (mask & BUFFER_BIT_BACK_LEFT) {
-      if (colorMask == ~0) {
-	 blit_mask |= BUFFER_BIT_BACK_LEFT;
-      } 
-      else {
-	 tri_mask |= BUFFER_BIT_BACK_LEFT;
-      }
-   }
-
-   if (mask & BUFFER_BIT_DEPTH) {
-      blit_mask |= BUFFER_BIT_DEPTH;
-   }
-
-   if (mask & BUFFER_BIT_STENCIL) {
-      if (!intel->hw_stencil) {
-	 swrast_mask |= BUFFER_BIT_STENCIL;
-      }
-      else if ((ctx->Stencil.WriteMask[0] & 0xff) != 0xff) {
-	 tri_mask |= BUFFER_BIT_STENCIL;
-      } 
-      else {
-	 blit_mask |= BUFFER_BIT_STENCIL;
-      }
-   }
-
-   swrast_mask |= (mask & BUFFER_BIT_ACCUM);
-
-   if (blit_mask) 
-      intelClearWithBlit( ctx, blit_mask, 0, 0, 0, 0, 0);
-
-   if (tri_mask) 
-      intel->vtbl.clear_with_tris( intel, tri_mask, 0, 0, 0, 0, 0);
-
-   if (swrast_mask)
-      _swrast_Clear( ctx, swrast_mask );
-}
 
 
-void
-intelRotateWindow(intelContextPtr intel, __DRIdrawablePrivate *dPriv,
-                  GLuint srcBuffer)
-{
-   if (intel->vtbl.rotate_window) {
-      intel->vtbl.rotate_window(intel, dPriv, srcBuffer);
-   }
-}
 
+   batch.start = start_offset;
+   batch.used = used;
+   batch.cliprects = intel->pClipRects;
+   batch.num_cliprects = ignore_cliprects ? 0 : intel->numClipRects;
+   batch.DR1 = 0;
+   batch.DR4 = ((((GLuint) intel->drawX) & 0xffff) |
+                (((GLuint) intel->drawY) << 16));
 
-void *intelAllocateAGP( intelContextPtr intel, GLsizei size )
-{
-   int region_offset;
-   drmI830MemAlloc alloc;
-   int ret;
-
-   if (0)
-      fprintf(stderr, "%s: %d bytes\n", __FUNCTION__, size);
-
-   alloc.region = I830_MEM_REGION_AGP;
-   alloc.alignment = 0;
-   alloc.size = size;
-   alloc.region_offset = &region_offset;
+   DBG("%s: 0x%x..0x%x DR4: %x cliprects: %d\n",
+       __FUNCTION__,
+       batch.start,
+       batch.start + batch.used * 4, batch.DR4, batch.num_cliprects);
 
-   LOCK_HARDWARE(intel);
-
-   /* Make sure the global heap is initialized
-    */
-   if (intel->texture_heaps[0])
-      driAgeTextures( intel->texture_heaps[0] );
-
-
-   ret = drmCommandWriteRead( intel->driFd,
-			      DRM_I830_ALLOC,
-			      &alloc, sizeof(alloc));
-   
-   if (ret) {
-      fprintf(stderr, "%s: DRM_I830_ALLOC ret %d\n", __FUNCTION__, ret);
+   if (drmCommandWrite(intel->driFd, DRM_I830_BATCHBUFFER, &batch,
+                       sizeof(batch))) {
+      fprintf(stderr, "DRM_I830_BATCHBUFFER: %d\n", -errno);
       UNLOCK_HARDWARE(intel);
-      return NULL;
-   }
-   
-   if (0)
-      fprintf(stderr, "%s: allocated %d bytes\n", __FUNCTION__, size);
-
-   /* Need to propogate this information (agp memory in use) to our
-    * local texture lru.  The kernel has already updated the global
-    * lru.  An alternative would have been to allocate memory the
-    * usual way and then notify the kernel to pin the allocation.
-    */
-   if (intel->texture_heaps[0])
-      driAgeTextures( intel->texture_heaps[0] );
-
-   UNLOCK_HARDWARE(intel);   
-
-   return (void *)((char *)intel->intelScreen->tex.map + region_offset);
-}
-
-void intelFreeAGP( intelContextPtr intel, void *pointer )
-{
-   int region_offset;
-   drmI830MemFree memfree;
-   int ret;
-
-   region_offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-
-   if (region_offset < 0 || 
-       region_offset > intel->intelScreen->tex.size) {
-      fprintf(stderr, "offset %d outside range 0..%d\n", region_offset,
-	      intel->intelScreen->tex.size);
-      return;
-   }
-
-   memfree.region = I830_MEM_REGION_AGP;
-   memfree.region_offset = region_offset;
-   
-   ret = drmCommandWrite( intel->driFd,
-			  DRM_I830_FREE,
-			  &memfree, sizeof(memfree));
-   
-   if (ret) 
-      fprintf(stderr, "%s: DRM_I830_FREE ret %d\n", __FUNCTION__, ret);
-}
-
-/* This version of AllocateMemoryMESA allocates only agp memory, and
- * only does so after the point at which the driver has been
- * initialized.
- *
- * Theoretically a valid context isn't required.  However, in this
- * implementation, it is, as I'm using the hardware lock to protect
- * the kernel data structures, and the current context to get the
- * device fd.
- */
-void *intelAllocateMemoryMESA(__DRInativeDisplay *dpy, int scrn,
-			      GLsizei size, GLfloat readfreq,
-			      GLfloat writefreq, GLfloat priority)
-{
-   GET_CURRENT_CONTEXT(ctx);
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s sz %d %f/%f/%f\n", __FUNCTION__, size, readfreq, 
-	      writefreq, priority);
-
-   if (getenv("INTEL_NO_ALLOC"))
-      return NULL;
-   
-   if (!ctx || INTEL_CONTEXT(ctx) == 0) 
-      return NULL;
-   
-   return intelAllocateAGP( INTEL_CONTEXT(ctx), size );
-}
-
-
-/* Called via glXFreeMemoryMESA() */
-void intelFreeMemoryMESA(__DRInativeDisplay *dpy, int scrn, GLvoid *pointer)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   if (INTEL_DEBUG & DEBUG_IOCTL) 
-      fprintf(stderr, "%s %p\n", __FUNCTION__, pointer);
-
-   if (!ctx || INTEL_CONTEXT(ctx) == 0) {
-      fprintf(stderr, "%s: no context\n", __FUNCTION__);
-      return;
-   }
-
-   intelFreeAGP( INTEL_CONTEXT(ctx), pointer );
-}
-
-/* Called via glXGetMemoryOffsetMESA() 
- *
- * Returns offset of pointer from the start of agp aperture.
- */
-GLuint intelGetMemoryOffsetMESA(__DRInativeDisplay *dpy, int scrn, 
-				const GLvoid *pointer)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   intelContextPtr intel;
-
-   if (!ctx || !(intel = INTEL_CONTEXT(ctx)) ) {
-      fprintf(stderr, "%s: no context\n", __FUNCTION__);
-      return ~0;
-   }
-
-   if (!intelIsAgpMemory( intel, pointer, 0 ))
-      return ~0;
-
-   return intelAgpOffsetFromVirtual( intel, pointer );
-}
-
-
-GLboolean intelIsAgpMemory( intelContextPtr intel, const GLvoid *pointer,
-			   GLint size )
-{
-   int offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-   int valid = (size >= 0 &&
-		offset >= 0 &&
-		offset + size < intel->intelScreen->tex.size);
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "intelIsAgpMemory( %p ) : %d\n", pointer, valid );
-   
-   return valid;
-}
-
-
-GLuint intelAgpOffsetFromVirtual( intelContextPtr intel, const GLvoid *pointer )
-{
-   int offset = (char *)pointer - (char *)intel->intelScreen->tex.map;
-
-   if (offset < 0 || offset > intel->intelScreen->tex.size)
-      return ~0;
-   else
-      return intel->intelScreen->tex.offset + offset;
-}
-
-
-
-
-
-/* Flip the front & back buffes
- */
-void intelPageFlip( const __DRIdrawablePrivate *dPriv )
-{
-#if 0
-   intelContextPtr intel;
-   int tmp, ret;
-
-   if (INTEL_DEBUG & DEBUG_IOCTL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(dPriv);
-   assert(dPriv->driContextPriv);
-   assert(dPriv->driContextPriv->driverPrivate);
-
-   intel = (intelContextPtr) dPriv->driContextPriv->driverPrivate;
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-
-   if (dPriv->pClipRects) {
-      *(drm_clip_rect_t *)intel->sarea->boxes = dPriv->pClipRects[0];
-      intel->sarea->nbox = 1;
-   }
-
-   ret = drmCommandNone(intel->driFd, DRM_I830_FLIP); 
-   if (ret) {
-      fprintf(stderr, "%s: %d\n", __FUNCTION__, ret);
-      UNLOCK_HARDWARE( intel );
       exit(1);
    }
 
-   tmp = intel->sarea->last_enqueue;
-   intelRefillBatchLocked( intel );
-   UNLOCK_HARDWARE( intel );
-
-
-   intelSetDrawBuffer( &intel->ctx, intel->ctx.Color.DriverDrawBuffer );
-#endif
+   /* FIXME: use hardware contexts to avoid 'losing' hardware after
+    * each buffer flush.
+    */
+   intel->vtbl.lost_hardware(intel);
 }
diff --git a/src/mesa/drivers/dri/i915/intel_ioctl.h b/src/mesa/drivers/dri/i915/intel_ioctl.h
index 6ea47e462e1..7a5b175ed1c 100644
--- a/src/mesa/drivers/dri/i915/intel_ioctl.h
+++ b/src/mesa/drivers/dri/i915/intel_ioctl.h
@@ -30,43 +30,11 @@
 
 #include "intel_context.h"
 
-extern void intelWaitAgeLocked( intelContextPtr intel, int age, GLboolean unlock );
+void intelWaitIrq(intelScreenPrivate *intelScreen, int seq);
+int intelEmitIrqLocked(intelScreenPrivate *intelScreen);
 
-extern void intelClear(GLcontext *ctx, GLbitfield mask);
-
-extern void intelPageFlip( const __DRIdrawablePrivate *dpriv );
-
-extern void intelRotateWindow(intelContextPtr intel,
-                              __DRIdrawablePrivate *dPriv, GLuint srcBuffer);
-
-extern void intelWaitForIdle( intelContextPtr intel );
-extern void intelFlushBatch( intelContextPtr intel, GLboolean refill );
-extern void intelFlushBatchLocked( intelContextPtr intel,
-				   GLboolean ignore_cliprects,
-				   GLboolean refill,
-				   GLboolean allow_unlock);
-extern void intelRefillBatchLocked( intelContextPtr intel, GLboolean allow_unlock );
-extern void intelFinish( GLcontext *ctx );
-extern void intelFlush( GLcontext *ctx );
-extern void intelglFlush( GLcontext *ctx );
-
-extern void *intelAllocateAGP( intelContextPtr intel, GLsizei size );
-extern void intelFreeAGP( intelContextPtr intel, void *pointer );
-
-extern void *intelAllocateMemoryMESA( __DRInativeDisplay *dpy, int scrn, 
-				      GLsizei size, GLfloat readfreq,
-				      GLfloat writefreq, GLfloat priority );
-
-extern void intelFreeMemoryMESA( __DRInativeDisplay *dpy, int scrn, 
-				 GLvoid *pointer );
-
-extern GLuint intelGetMemoryOffsetMESA( __DRInativeDisplay *dpy, int scrn, const GLvoid *pointer );
-extern GLboolean intelIsAgpMemory( intelContextPtr intel, const GLvoid *pointer,
-				  GLint size );
-
-extern GLuint intelAgpOffsetFromVirtual( intelContextPtr intel, const GLvoid *p );
-
-extern void intelWaitIrq( intelContextPtr intel, int seq );
-extern u_int32_t intelGetLastFrame (intelContextPtr intel);
-extern int intelEmitIrqLocked( intelContextPtr intel );
+void intel_batch_ioctl(struct intel_context *intel,
+                       GLuint start_offset,
+                       GLuint used,
+                       GLboolean ignore_cliprects, GLboolean allow_unlock);
 #endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c b/src/mesa/drivers/dri/i915/intel_mipmap_tree.c
index 843a78eb82b..74f6b2d8515 100644
--- a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i915/intel_mipmap_tree.c
@@ -97,9 +97,19 @@ intel_miptree_create(struct intel_context *intel,
       break;
    }
 
-   if (ok)
+   if (ok) {
+      if (!mt->compressed) {
+	 /* XXX: Align pitch to multiple of 64 bytes for now to allow
+	  * render-to-texture to work in all cases. This should probably be
+	  * replaced at some point by some scheme to only do this when really
+	  * necessary.
+	  */
+	 mt->pitch = ((mt->pitch * cpp + 63) & ~63) / cpp;
+      }
+
       mt->region = intel_region_alloc(intel->intelScreen,
                                       mt->cpp, mt->pitch, mt->total_height);
+   }
 
    if (!mt->region) {
       free(mt);
@@ -325,6 +335,7 @@ intel_miptree_image_data(struct intel_context *intel,
    }
 }
 
+extern GLuint intel_compressed_alignment(GLenum);
 /* Copy mipmap image between trees
  */
 void
@@ -342,8 +353,12 @@ intel_miptree_image_copy(struct intel_context *intel,
    const GLuint *src_depth_offset = intel_miptree_depth_offsets(src, level);
    GLuint i;
 
-   if (dst->compressed)
-      height /= 4;
+   if (dst->compressed) {
+       GLuint alignment = intel_compressed_alignment(dst->internal_format);
+       height = (height + 3) / 4;
+       width = ((width + alignment - 1) & ~(alignment - 1));
+   }
+
    for (i = 0; i < depth; i++) {
       intel_region_copy(intel->intelScreen,
                         dst->region, dst_offset + dst_depth_offset[i],
diff --git a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.h b/src/mesa/drivers/dri/i915/intel_mipmap_tree.h
index ecdb7be244f..ecdb7be244f 100644
--- a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.h
+++ b/src/mesa/drivers/dri/i915/intel_mipmap_tree.h
diff --git a/src/mesa/drivers/dri/i915/intel_pixel.c b/src/mesa/drivers/dri/i915/intel_pixel.c
index c3030d42b04..9018e3daef4 100644
--- a/src/mesa/drivers/dri/i915/intel_pixel.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel.c
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2006 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -12,7 +12,7 @@
  * the following conditions:
  * 
  * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
+ * next paragraph) shall be included in all copies or substantial portionsalloc
  * of the Software.
  * 
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
@@ -25,500 +25,96 @@
  * 
  **************************************************************************/
 
-#include "glheader.h"
 #include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
+#include "state.h"
 #include "swrast/swrast.h"
 
-#include "intel_screen.h"
 #include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-
-
-
-static GLboolean
-check_color( const GLcontext *ctx, GLenum type, GLenum format,
-	     const struct gl_pixelstore_attrib *packing,
-	     const void *pixels, GLint sz, GLint pitch )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLuint cpp = intel->intelScreen->cpp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (	(pitch & 63) ||
-	ctx->_ImageTransferState ||
-	packing->SwapBytes ||
-	packing->LsbFirst) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: failed 1\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   if ( type == GL_UNSIGNED_INT_8_8_8_8_REV && 
-	cpp == 4 && 
-	format == GL_BGRA ) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: passed 2\n", __FUNCTION__);
-      return GL_TRUE;
-   }
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s: failed\n", __FUNCTION__);
-
-   return GL_FALSE;
-}
-
-static GLboolean
-check_color_per_fragment_ops( const GLcontext *ctx )
-{
-   int result;
-   result = (!(     ctx->Color.AlphaEnabled || 
-		    ctx->Depth.Test ||
-		    ctx->Fog.Enabled ||
-		    ctx->Scissor.Enabled ||
-		    ctx->Stencil.Enabled ||
-		    !ctx->Color.ColorMask[0] ||
-		    !ctx->Color.ColorMask[1] ||
-		    !ctx->Color.ColorMask[2] ||
-		    !ctx->Color.ColorMask[3] ||
-		    ctx->Color.ColorLogicOpEnabled ||
-		    ctx->Texture._EnabledUnits
-           ) &&
-	   ctx->Current.RasterPosValid);
-   
-   return result;
-}
-
-
-/**
- * Clip the given rectangle against the buffer's bounds (including scissor).
- * \param size returns the 
- * \return GL_TRUE if any pixels remain, GL_FALSE if totally clipped.
- *
- * XXX Replace this with _mesa_clip_drawpixels() and _mesa_clip_readpixels()
- * from Mesa 6.4.  We shouldn't apply scissor for ReadPixels.
- */
-static GLboolean
-clip_pixelrect( const GLcontext *ctx,
-		const GLframebuffer *buffer,
-		GLint *x, GLint *y,
-		GLsizei *width, GLsizei *height)
-{
-   /* left clipping */
-   if (*x < buffer->_Xmin) {
-      *width -= (buffer->_Xmin - *x);
-      *x = buffer->_Xmin;
-   }
-
-   /* right clipping */
-   if (*x + *width > buffer->_Xmax)
-      *width -= (*x + *width - buffer->_Xmax - 1);
-
-   if (*width <= 0)
-      return GL_FALSE;
-
-   /* bottom clipping */
-   if (*y < buffer->_Ymin) {
-      *height -= (buffer->_Ymin - *y);
-      *y = buffer->_Ymin;
-   }
-
-   /* top clipping */
-   if (*y + *height > buffer->_Ymax)
-      *height -= (*y + *height - buffer->_Ymax - 1);
-
-   if (*height <= 0)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
+#include "intel_pixel.h"
+#include "intel_regions.h"
 
 
 /**
- * Compute intersection of a clipping rectangle and pixel rectangle,
- * returning results in x/y/w/hOut vars.
- * \return GL_TRUE if there's intersection, GL_FALSE if disjoint.
+ * Check if any fragment operations are in effect which might effect
+ * glDraw/CopyPixels.
  */
-static INLINE GLboolean
-intersect_region(const drm_clip_rect_t *box,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLint *xOut, GLint *yOut, GLint *wOut, GLint *hOut)
-{
-   GLint bx = box->x1;
-   GLint by = box->y1;
-   GLint bw = box->x2 - bx;
-   GLint bh = box->y2 - by;
-
-   if (bx < x) bw -= x - bx, bx = x;
-   if (by < y) bh -= y - by, by = y;
-   if (bx + bw > x + width) bw = x + width - bx;
-   if (by + bh > y + height) bh = y + height - by;
-
-   *xOut = bx;
-   *yOut = by;
-   *wOut = bw;
-   *hOut = bh;
-
-   if (bw <= 0) return GL_FALSE;
-   if (bh <= 0) return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-
-static GLboolean
-intelTryReadPixels( GLcontext *ctx,
-		  GLint x, GLint y, GLsizei width, GLsizei height,
-		  GLenum format, GLenum type,
-		  const struct gl_pixelstore_attrib *pack,
-		  GLvoid *pixels )
+GLboolean
+intel_check_blit_fragment_ops(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLint size = 0; /* not really used */
-   GLint pitch = pack->RowLength ? pack->RowLength : width;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   /* Only accelerate reading to agp buffers.
-    */
-   if ( !intelIsAgpMemory(intel, pixels, 
-			pitch * height * intel->intelScreen->cpp ) ) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: dest not agp\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   /* Need GL_PACK_INVERT_MESA to cope with upsidedown results from
-    * blitter:
-    */
-   if (!pack->Invert) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "%s: MESA_PACK_INVERT not set\n", __FUNCTION__);
-      return GL_FALSE;
-   }
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
 
-   if (!check_color(ctx, type, format, pack, pixels, size, pitch))
-      return GL_FALSE;
-
-   switch ( intel->intelScreen->cpp ) {
-   case 4:
-      break;
-   default:
-      return GL_FALSE;
-   }
-
-
-   /* Although the blits go on the command buffer, need to do this and
-    * fire with lock held to guarentee cliprects and drawing offset are
-    * correct.
-    *
-    * This is an unusual situation however, as the code which flushes
-    * a full command buffer expects to be called unlocked.  As a
-    * workaround, immediately flush the buffer on aquiring the lock.
+   /* XXX Note: Scissor could be done with the blitter:
     */
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-   {
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      int nbox = dPriv->numClipRects;
-      int src_offset = intel->readRegion->offset;
-      int src_pitch = intel->intelScreen->front.pitch;
-      int dst_offset = intelAgpOffsetFromVirtual( intel, pixels);
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      int i;
-
-      assert(dst_offset != ~0);  /* should have been caught above */
-
-      if (!clip_pixelrect(ctx, ctx->ReadBuffer, &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s totally clipped -- nothing to do\n",
-		    __FUNCTION__);
-	 return GL_TRUE;
-      }
-
-      /* convert to screen coords (y=0=top) */
-      y = dPriv->h - y - height;
-      x += dPriv->x;
-      y += dPriv->y;
-
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-	 fprintf(stderr, "readpixel blit src_pitch %d dst_pitch %d\n",
-		 src_pitch, pitch);
-
-      /* We don't really have to do window clipping for readpixels.
-       * The OpenGL spec says that pixels read from outside the
-       * visible window region (pixel ownership) have undefined value.
-       */
-      for (i = 0 ; i < nbox ; i++)
-      {
-         GLint bx, by, bw, bh;
-         if (intersect_region(box+i, x, y, width, height,
-                              &bx, &by, &bw, &bh)) {
-            intelEmitCopyBlitLocked( intel,
-                                     intel->intelScreen->cpp,
-                                     src_pitch, src_offset,
-                                     pitch, dst_offset,
-                                     bx, by,
-                                     bx - x, by - y,
-                                     bw, bh );
-         }
-      }
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( &intel->ctx );
-
-   return GL_TRUE;
-}
-
-static void
-intelReadPixels( GLcontext *ctx,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLenum format, GLenum type,
-		 const struct gl_pixelstore_attrib *pack,
-		 GLvoid *pixels )
-{
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (!intelTryReadPixels( ctx, x, y, width, height, format, type, pack, 
-                            pixels))
-      _swrast_ReadPixels( ctx, x, y, width, height, format, type, pack, 
-			  pixels);
+   return !(ctx->_ImageTransferState ||
+            ctx->Color.AlphaEnabled ||
+            ctx->Depth.Test ||
+            ctx->Fog.Enabled ||
+            ctx->Scissor.Enabled ||
+            ctx->Stencil.Enabled ||
+            !ctx->Color.ColorMask[0] ||
+            !ctx->Color.ColorMask[1] ||
+            !ctx->Color.ColorMask[2] ||
+            !ctx->Color.ColorMask[3] ||
+            ctx->Texture._EnabledUnits || 
+	    ctx->FragmentProgram._Enabled ||
+	    ctx->Color.BlendEnabled);
 }
 
 
-
-
-static void do_draw_pix( GLcontext *ctx,
-			 GLint x, GLint y, GLsizei width, GLsizei height,
-			 GLint pitch,
-			 const void *pixels,
-			 GLuint dest )
+GLboolean
+intel_check_meta_tex_fragment_ops(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   drm_clip_rect_t *box = dPriv->pClipRects;
-   int nbox = dPriv->numClipRects;
-   int i;
-   int src_offset = intelAgpOffsetFromVirtual( intel, pixels);
-   int src_pitch = pitch;
-
-   assert(src_offset != ~0);  /* should be caught earlier */
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   intelFlush( &intel->ctx );
-   LOCK_HARDWARE( intel );
-   if (ctx->DrawBuffer)
-   {
-      y -= height;			/* cope with pixel zoom */
-   
-      if (!clip_pixelrect(ctx, ctx->DrawBuffer,
-			  &x, &y, &width, &height)) {
-	 UNLOCK_HARDWARE( intel );
-	 return;
-      }
-
-      y = dPriv->h - y - height; 	/* convert from gl to hardware coords */
-      x += dPriv->x;
-      y += dPriv->y;
-
-      for (i = 0 ; i < nbox ; i++ )
-      {
-	 GLint bx, by, bw, bh;
-	 if (intersect_region(box + i, x, y, width, height,
-			      &bx, &by, &bw, &bh)) {
-            intelEmitCopyBlitLocked( intel,
-                                     intel->intelScreen->cpp,
-                                     src_pitch, src_offset,
-                                     intel->intelScreen->front.pitch,
-                                     intel->drawRegion->offset,
-                                     bx - x, by - y,
-                                     bx, by,
-                                     bw, bh );
-         }
-      }
-   }
-   UNLOCK_HARDWARE( intel );
-   intelFinish( &intel->ctx );
+   /* Some of _ImageTransferState (scale, bias) could be done with
+    * fragment programs on i915.
+    */
+   return !(ctx->_ImageTransferState || ctx->Fog.Enabled ||     /* not done yet */
+            ctx->Texture._EnabledUnits || ctx->FragmentProgram._Enabled);
 }
 
-
-
-static GLboolean
-intelTryDrawPixels( GLcontext *ctx,
-		  GLint x, GLint y, GLsizei width, GLsizei height,
-		  GLenum format, GLenum type,
-		  const struct gl_pixelstore_attrib *unpack,
-		  const GLvoid *pixels )
+/* The intel_region struct doesn't really do enough to capture the
+ * format of the pixels in the region.  For now this code assumes that
+ * the region is a display surface and hence is either ARGB8888 or
+ * RGB565.
+ * XXX FBO: If we'd pass in the intel_renderbuffer instead of region, we'd
+ * know the buffer's pixel format.
+ *
+ * \param format  as given to glDraw/ReadPixels
+ * \param type  as given to glDraw/ReadPixels
+ */
+GLboolean
+intel_check_blit_format(struct intel_region * region,
+                        GLenum format, GLenum type)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   GLint pitch = unpack->RowLength ? unpack->RowLength : width;
-   GLuint dest;
-   GLuint cpp = intel->intelScreen->cpp;
-   GLint size = width * pitch * cpp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   switch (format) {
-   case GL_RGB:
-   case GL_RGBA:
-   case GL_BGRA:
-      dest = intel->drawRegion->offset;
-
-      /* Planemask doesn't have full support in blits.
-       */
-      if (!ctx->Color.ColorMask[RCOMP] ||
-	  !ctx->Color.ColorMask[GCOMP] ||
-	  !ctx->Color.ColorMask[BCOMP] ||
-	  !ctx->Color.ColorMask[ACOMP]) {
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s: planemask\n", __FUNCTION__);
-	 return GL_FALSE;	
-      }
-
-      /* Can't do conversions on agp reads/draws. 
-       */
-      if ( !intelIsAgpMemory( intel, pixels, size ) ) {
-	 if (INTEL_DEBUG & DEBUG_PIXEL)
-	    fprintf(stderr, "%s: not agp memory\n", __FUNCTION__);
-	 return GL_FALSE;
-      }
-
-      if (!check_color(ctx, type, format, unpack, pixels, size, pitch)) {
-	 return GL_FALSE;
-      }
-      if (!check_color_per_fragment_ops(ctx)) {
-	 return GL_FALSE;
-      }
-
-      if (ctx->Pixel.ZoomX != 1.0F ||
-	  ctx->Pixel.ZoomY != -1.0F)
-	 return GL_FALSE;
-      break;
-
-   default:
-      return GL_FALSE;
+   if (region->cpp == 4 &&
+       (type == GL_UNSIGNED_INT_8_8_8_8_REV ||
+        type == GL_UNSIGNED_BYTE) && format == GL_BGRA) {
+      return GL_TRUE;
    }
 
-   if ( intelIsAgpMemory(intel, pixels, size) )
-   {
-      do_draw_pix( ctx, x, y, width, height, pitch, pixels, dest );
+   if (region->cpp == 2 &&
+       type == GL_UNSIGNED_SHORT_5_6_5_REV && format == GL_BGR) {
       return GL_TRUE;
    }
-   else if (0)
-   {
-      /* Pixels is in regular memory -- get dma buffers and perform
-       * upload through them.  No point doing this for regular uploads
-       * but once we remove some of the restrictions above (colormask,
-       * pixelformat conversion, zoom?, etc), this could be a win.
-       */
-   }
-   else
-      return GL_FALSE;
-
-   return GL_FALSE;
-}
 
-static void
-intelDrawPixels( GLcontext *ctx,
-		 GLint x, GLint y, GLsizei width, GLsizei height,
-		 GLenum format, GLenum type,
-		 const struct gl_pixelstore_attrib *unpack,
-		 const GLvoid *pixels )
-{
    if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (intelTryDrawPixels( ctx, x, y, width, height, format, type,
-                           unpack, pixels ))
-      return;
-
-   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;
-   }
-   else {
-      _swrast_DrawPixels( ctx, x, y, width, height, format, type,
-                          unpack, pixels );
-   }
-}
-
-
-
-
-/**
- * Implement glCopyPixels for the front color buffer (or back buffer Pixmap)
- * for the color buffer.  Don't support zooming, pixel transfer, etc.
- * We do support copying from one window to another, ala glXMakeCurrentRead.
- */
-static void
-intelCopyPixels( GLcontext *ctx,
-		 GLint srcx, GLint srcy, GLsizei width, GLsizei height,
-		 GLint destx, GLint desty, GLenum type )
-{
-#if 0
-   const XMesaContext xmesa = XMESA_CONTEXT(ctx);
-   const SWcontext *swrast = SWRAST_CONTEXT( ctx );
-   XMesaDisplay *dpy = xmesa->xm_visual->display;
-   const XMesaDrawable drawBuffer = xmesa->xm_draw_buffer->buffer;
-   const XMesaDrawable readBuffer = xmesa->xm_read_buffer->buffer;
-   const XMesaGC gc = xmesa->xm_draw_buffer->gc;
-
-   ASSERT(dpy);
-   ASSERT(gc);
+      fprintf(stderr, "%s: bad format for blit (cpp %d, type %s format %s)\n",
+              __FUNCTION__, region->cpp,
+              _mesa_lookup_enum_by_nr(type), _mesa_lookup_enum_by_nr(format));
 
-   if (drawBuffer &&  /* buffer != 0 means it's a Window or Pixmap */
-       readBuffer &&
-       type == GL_COLOR &&
-       (swrast->_RasterMask & ~CLIP_BIT) == 0 && /* no blend, z-test, etc */
-       ctx->_ImageTransferState == 0 &&  /* no color tables, scale/bias, etc */
-       ctx->Pixel.ZoomX == 1.0 &&        /* no zooming */
-       ctx->Pixel.ZoomY == 1.0) {
-      /* Note: we don't do any special clipping work here.  We could,
-       * but X will do it for us.
-       */
-      srcy = FLIP(xmesa->xm_read_buffer, srcy) - height + 1;
-      desty = FLIP(xmesa->xm_draw_buffer, desty) - height + 1;
-      XCopyArea(dpy, readBuffer, drawBuffer, gc,
-                srcx, srcy, width, height, destx, desty);
-   }
-#else
-   _swrast_CopyPixels(ctx, srcx, srcy, width, height, destx, desty, type );
-#endif
+   return GL_FALSE;
 }
 
 
-
-
-void intelInitPixelFuncs( struct dd_function_table *functions )
+void
+intelInitPixelFuncs(struct dd_function_table *functions)
 {
+   functions->Accum = _swrast_Accum;
+   functions->Bitmap = _swrast_Bitmap;
    functions->CopyPixels = intelCopyPixels;
-   if (!getenv("INTEL_NO_BLITS")) {
-      functions->ReadPixels = intelReadPixels;  
-      functions->DrawPixels = intelDrawPixels; 
-   }
+   functions->ReadPixels = intelReadPixels;
+   functions->DrawPixels = intelDrawPixels;
 }
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel.h b/src/mesa/drivers/dri/i915/intel_pixel.h
index a6fcf90ce03..a6fcf90ce03 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel.h
+++ b/src/mesa/drivers/dri/i915/intel_pixel.h
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
index 65bf338589e..65bf338589e 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel_bitmap.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel_copy.c b/src/mesa/drivers/dri/i915/intel_pixel_copy.c
index 9d478283e47..9d478283e47 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel_copy.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c b/src/mesa/drivers/dri/i915/intel_pixel_draw.c
index e4e57cb3a7e..afb586b4a33 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel_draw.c
@@ -217,7 +217,7 @@ do_blit_drawpixels(GLcontext * ctx,
    struct intel_buffer_object *src = intel_buffer_object(unpack->BufferObj);
    GLuint src_offset;
    GLuint rowLength;
-   struct _DriFenceObject *fence = NULL;
+   dri_fence *fence = NULL;
 
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s\n", __FUNCTION__);
@@ -298,8 +298,7 @@ do_blit_drawpixels(GLcontext * ctx,
       drm_clip_rect_t *box = dPriv->pClipRects;
       drm_clip_rect_t rect;
       drm_clip_rect_t dest_rect;
-      struct _DriBufferObject *src_buffer =
-         intel_bufferobj_buffer(intel, src, INTEL_READ);
+      dri_bo *src_buffer = intel_bufferobj_buffer(intel, src, INTEL_READ);
       int i;
 
       dest_rect.x1 = dPriv->x + x;
@@ -324,14 +323,15 @@ do_blit_drawpixels(GLcontext * ctx,
 			   ctx->Color.ColorLogicOpEnabled ?
 			   ctx->Color.LogicOp : GL_COPY);
       }
-      fence = intel_batchbuffer_flush(intel->batch);
-      driFenceReference(fence);
+      intel_batchbuffer_flush(intel->batch);
+      fence = intel->batch->last_fence;
+      dri_fence_reference(fence);
    }
    UNLOCK_HARDWARE(intel);
 
    if (fence) {
-      driFenceFinish(fence, DRM_FENCE_TYPE_EXE | DRM_I915_FENCE_TYPE_RW, GL_FALSE);
-      driFenceUnReference(fence);
+      dri_fence_wait(fence);
+      dri_fence_unreference(fence);
    }
 
    if (INTEL_DEBUG & DEBUG_PIXEL)
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel_read.c b/src/mesa/drivers/dri/i915/intel_pixel_read.c
index 24e49ae0663..a22844926c9 100644
--- a/src/mesa/drivers/dri/i915tex/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/i915/intel_pixel_read.c
@@ -173,7 +173,7 @@ do_blit_readpixels(GLcontext * ctx,
    struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj);
    GLuint dst_offset;
    GLuint rowLength;
-   struct _DriFenceObject *fence = NULL;
+   dri_fence *fence = NULL;
 
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s\n", __FUNCTION__);
@@ -241,9 +241,9 @@ do_blit_readpixels(GLcontext * ctx,
       GLboolean all = (width * height * src->cpp == dst->Base.Size &&
                        x == 0 && dst_offset == 0);
 
-      struct _DriBufferObject *dst_buffer =
-         intel_bufferobj_buffer(intel, dst, all ? INTEL_WRITE_FULL :
-                                INTEL_WRITE_PART);
+      dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
+						  all ? INTEL_WRITE_FULL :
+						  INTEL_WRITE_PART);
       __DRIdrawablePrivate *dPriv = intel->driDrawable;
       int nbox = dPriv->numClipRects;
       drm_clip_rect_t *box = dPriv->pClipRects;
@@ -275,16 +275,16 @@ do_blit_readpixels(GLcontext * ctx,
 			   GL_COPY);
       }
 
-      fence = intel_batchbuffer_flush(intel->batch);
-      driFenceReference(fence);
+      intel_batchbuffer_flush(intel->batch);
+      fence = intel->batch->last_fence;
+      dri_fence_reference(fence);
 
    }
    UNLOCK_HARDWARE(intel);
 
    if (fence) {
-      driFenceFinish(fence, DRM_FENCE_TYPE_EXE | DRM_I915_FENCE_TYPE_RW, 
-		     GL_FALSE);
-      driFenceUnReference(fence);
+      dri_fence_wait(fence);
+      dri_fence_unreference(fence);
    }
 
    if (INTEL_DEBUG & DEBUG_PIXEL)
diff --git a/src/mesa/drivers/dri/i915/intel_reg.h b/src/mesa/drivers/dri/i915/intel_reg.h
index 1ec153266c7..7828ba6ad39 100644
--- a/src/mesa/drivers/dri/i915/intel_reg.h
+++ b/src/mesa/drivers/dri/i915/intel_reg.h
@@ -81,4 +81,8 @@
 #define XY_SRC_COPY_BLT_WRITE_ALPHA     (1<<21)
 #define XY_SRC_COPY_BLT_WRITE_RGB       (1<<20)
 
+#define MI_WAIT_FOR_EVENT               ((0x3<<23))
+#define MI_WAIT_FOR_PLANE_B_FLIP        (1<<6)
+#define MI_WAIT_FOR_PLANE_A_FLIP        (1<<2)
+
 #endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_regions.c b/src/mesa/drivers/dri/i915/intel_regions.c
index 7d19bd07d36..4eac859a133 100644
--- a/src/mesa/drivers/dri/i915tex/intel_regions.c
+++ b/src/mesa/drivers/dri/i915/intel_regions.c
@@ -52,8 +52,17 @@ void
 intel_region_idle(intelScreenPrivate *intelScreen, struct intel_region *region)
 {
    DBG("%s\n", __FUNCTION__);
-   if (region && region->buffer)
-      driBOWaitIdle(region->buffer, GL_FALSE);
+   /* XXX: Using this function is likely bogus -- it ought to only have been
+    * used before a map, anyway, but leave this cheap implementation of it
+    * for now.
+    */
+   if (region && region->buffer) {
+      /* Mapping it for read will ensure that any acceleration to the region
+       * would have landed already.
+       */
+      dri_bo_map(region->buffer, GL_TRUE);
+      dri_bo_unmap(region->buffer);
+   }
 }
 
 /* XXX: Thread safety?
@@ -66,8 +75,8 @@ intel_region_map(intelScreenPrivate *intelScreen, struct intel_region *region)
       if (region->pbo)
          intel_region_cow(intelScreen, region);
 
-      region->map = driBOMap(region->buffer,
-                             DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 0);
+      dri_bo_map(region->buffer, GL_TRUE);
+      region->map = region->buffer->virtual;
    }
 
    return region->map;
@@ -78,19 +87,16 @@ intel_region_unmap(intelScreenPrivate *intelScreen, struct intel_region *region)
 {
    DBG("%s\n", __FUNCTION__);
    if (!--region->map_refcount) {
-      driBOUnmap(region->buffer);
+      dri_bo_unmap(region->buffer);
       region->map = NULL;
    }
 }
 
-#undef TEST_CACHED_TEXTURES
-
 struct intel_region *
 intel_region_alloc(intelScreenPrivate *intelScreen,
                    GLuint cpp, GLuint pitch, GLuint height)
 {
    struct intel_region *region = calloc(sizeof(*region), 1);
-   struct intel_context *intel = intelScreenContext(intelScreen);
 
    DBG("%s\n", __FUNCTION__);
 
@@ -99,18 +105,8 @@ intel_region_alloc(intelScreenPrivate *intelScreen,
    region->height = height;     /* needed? */
    region->refcount = 1;
 
-   driGenBuffers(intelScreen->regionPool,
-                 "region", 1, &region->buffer, 64,
-#ifdef TEST_CACHED_TEXTURES		 
-		 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_BIND_CACHED |
-		 DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 
-#else
-		 0,
-#endif
-		 0);
-   LOCK_HARDWARE(intel);
-   driBOData(region->buffer, pitch * cpp * height, NULL, 0);
-   UNLOCK_HARDWARE(intel);
+   region->buffer = dri_bo_alloc(intelScreen->bufmgr, "region",
+				 pitch * cpp * height, 64, DRM_BO_FLAG_MEM_TT);
    return region;
 }
 
@@ -141,7 +137,7 @@ intel_region_release(struct intel_region **region)
       if ((*region)->pbo)
 	 (*region)->pbo->region = NULL;
       (*region)->pbo = NULL;
-      driBOUnReference((*region)->buffer);
+      dri_bo_unreference((*region)->buffer);
       free(*region);
    }
    *region = NULL;
@@ -151,6 +147,7 @@ intel_region_release(struct intel_region **region)
 struct intel_region *
 intel_region_create_static(intelScreenPrivate *intelScreen,
                            GLuint mem_type,
+			   unsigned int bo_handle,
                            GLuint offset,
                            void *virtual,
                            GLuint cpp, GLuint pitch, GLuint height)
@@ -163,16 +160,18 @@ intel_region_create_static(intelScreenPrivate *intelScreen,
    region->height = height;     /* needed? */
    region->refcount = 1;
 
-   /*
-    * We use a "shared" buffer type to indicate buffers created and
-    * shared by others.
-    */
-
-   driGenBuffers(intelScreen->staticPool, "static region", 1,
-                 &region->buffer, 64,
-                 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_NO_MOVE |
-                 DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 0);
-   driBOSetStatic(region->buffer, offset, pitch * cpp * height, virtual, 0);
+   if (intelScreen->ttm) {
+      assert(bo_handle != -1);
+      region->buffer = dri_ttm_bo_create_from_handle(intelScreen->bufmgr,
+						     "static region",
+						     bo_handle);
+   } else {
+      region->buffer = dri_bo_alloc_static(intelScreen->bufmgr,
+					   "static region",
+					   offset, pitch * cpp * height,
+					   virtual,
+					   DRM_BO_FLAG_MEM_TT);
+   }
 
    return region;
 }
@@ -183,6 +182,7 @@ void
 intel_region_update_static(intelScreenPrivate *intelScreen,
 			   struct intel_region *region,
                            GLuint mem_type,
+			   unsigned int bo_handle,
                            GLuint offset,
                            void *virtual,
                            GLuint cpp, GLuint pitch, GLuint height)
@@ -198,13 +198,19 @@ intel_region_update_static(intelScreenPrivate *intelScreen,
     * shared by others.
     */
 
-   driDeleteBuffers(1, &region->buffer);
-   driGenBuffers(intelScreen->staticPool, "static region", 1,
-                 &region->buffer, 64,
-                 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_NO_MOVE |
-                 DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 0);
-   driBOSetStatic(region->buffer, offset, pitch * cpp * height, virtual, 0);
-
+   dri_bo_unreference(region->buffer);
+   if (intelScreen->ttm) {
+      assert(bo_handle != -1);
+      region->buffer = dri_ttm_bo_create_from_handle(intelScreen->bufmgr,
+						     "static region",
+						     bo_handle);
+   } else {
+      region->buffer = dri_bo_alloc_static(intelScreen->bufmgr,
+					   "static region",
+					   offset, pitch * cpp * height,
+					   virtual,
+					   DRM_BO_FLAG_MEM_TT);
+   }
 }
 
 
@@ -379,37 +385,33 @@ intel_region_attach_pbo(intelScreenPrivate *intelScreen,
    }
 
    if (region->buffer) {
-      driDeleteBuffers(1, &region->buffer);
+      dri_bo_unreference(region->buffer);
       region->buffer = NULL;
    }
 
    region->pbo = pbo;
    region->pbo->region = region;
-   region->buffer = driBOReference(pbo->buffer);
+   dri_bo_reference(pbo->buffer);
+   region->buffer = pbo->buffer;
 }
 
 
-/* Break the COW tie to the pbo.  The pbo gets to keep the data.
+/* Break the COW tie to the pbo and allocate a new buffer.
+ * The pbo gets to keep the data.
  */
 void
 intel_region_release_pbo(intelScreenPrivate *intelScreen,
                          struct intel_region *region)
 {
-   struct intel_context *intel = intelScreenContext(intelScreen);
-
    assert(region->buffer == region->pbo->buffer);
    region->pbo->region = NULL;
    region->pbo = NULL;
-   driBOUnReference(region->buffer);
+   dri_bo_unreference(region->buffer);
    region->buffer = NULL;
 
-   driGenBuffers(intelScreen->regionPool,
-                 "region", 1, &region->buffer, 64, 0, 0);
-   
-   LOCK_HARDWARE(intel);
-   driBOData(region->buffer,
-             region->cpp * region->pitch * region->height, NULL, 0);
-   UNLOCK_HARDWARE(intel);
+   region->buffer = dri_bo_alloc(intelScreen->bufmgr, "region",
+				 region->pitch * region->cpp * region->height,
+				 64, DRM_BO_FLAG_MEM_TT);
 }
 
 /* Break the COW tie to the pbo.  Both the pbo and the region end up
@@ -465,7 +467,7 @@ intel_region_cow(intelScreenPrivate *intelScreen, struct intel_region *region)
    }
 }
 
-struct _DriBufferObject *
+dri_bo *
 intel_region_buffer(intelScreenPrivate *intelScreen,
                     struct intel_region *region, GLuint flag)
 {
diff --git a/src/mesa/drivers/dri/i915tex/intel_regions.h b/src/mesa/drivers/dri/i915/intel_regions.h
index d938c107a46..42d7b177117 100644
--- a/src/mesa/drivers/dri/i915tex/intel_regions.h
+++ b/src/mesa/drivers/dri/i915/intel_regions.h
@@ -44,7 +44,7 @@ struct intel_buffer_object;
  */
 struct intel_region
 {
-   struct _DriBufferObject *buffer;   /**< buffer manager's buffer ID */
+   dri_bo *buffer;  /**< buffer manager's buffer */
    GLuint refcount; /**< Reference count for region */
    GLuint cpp;      /**< bytes per pixel */
    GLuint pitch;    /**< in pixels */
@@ -73,6 +73,7 @@ void intel_region_release(struct intel_region **ib);
 extern struct intel_region 
 *intel_region_create_static(intelScreenPrivate *intelScreen,
 			    GLuint mem_type,
+			    unsigned int bo_handle,
 			    GLuint offset,
 			    void *virtual,
 			    GLuint cpp,
@@ -81,6 +82,7 @@ extern void
 intel_region_update_static(intelScreenPrivate *intelScreen,
 			   struct intel_region *region,
 			   GLuint mem_type,
+			   unsigned int bo_handle,
 			   GLuint offset,
 			   void *virtual,
 			   GLuint cpp, GLuint pitch, GLuint height);
@@ -134,8 +136,8 @@ void intel_region_release_pbo(intelScreenPrivate *intelScreen,
 void intel_region_cow(intelScreenPrivate *intelScreen,
                       struct intel_region *region);
 
-struct _DriBufferObject *intel_region_buffer(intelScreenPrivate *intelScreen,
-                                             struct intel_region *region,
-                                             GLuint flag);
+dri_bo *intel_region_buffer(intelScreenPrivate *intelScreen,
+			    struct intel_region *region,
+			    GLuint flag);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/intel_render.c b/src/mesa/drivers/dri/i915/intel_render.c
index d9438ba0fd8..c8b6d308d96 100644
--- a/src/mesa/drivers/dri/i915/intel_render.c
+++ b/src/mesa/drivers/dri/i915/intel_render.c
@@ -51,14 +51,14 @@
  * dma buffers.  Use strip/fan hardware primitives where possible.
  * Try to simulate missing primitives with indexed vertices.
  */
-#define HAVE_POINTS      0  /* Has it, but can't use because subpixel has to
-			     * be adjusted for points on the INTEL/I845G
-			     */
+#define HAVE_POINTS      0      /* Has it, but can't use because subpixel has to
+                                 * be adjusted for points on the INTEL/I845G
+                                 */
 #define HAVE_LINES       1
 #define HAVE_LINE_STRIPS 1
 #define HAVE_TRIANGLES   1
 #define HAVE_TRI_STRIPS  1
-#define HAVE_TRI_STRIP_1 0  /* has it, template can't use it yet */
+#define HAVE_TRI_STRIP_1 0      /* has it, template can't use it yet */
 #define HAVE_TRI_FANS    1
 #define HAVE_POLYGONS    1
 #define HAVE_QUADS       0
@@ -66,7 +66,7 @@
 
 #define HAVE_ELTS        0
 
-static GLuint hw_prim[GL_POLYGON+1] = {
+static GLuint hw_prim[GL_POLYGON + 1] = {
    0,
    PRIM3D_LINELIST,
    PRIM3D_LINESTRIP,
@@ -79,7 +79,7 @@ static GLuint hw_prim[GL_POLYGON+1] = {
    PRIM3D_POLY
 };
 
-static const GLenum reduced_prim[GL_POLYGON+1] = {  
+static const GLenum reduced_prim[GL_POLYGON + 1] = {
    GL_POINTS,
    GL_LINES,
    GL_LINES,
@@ -92,58 +92,61 @@ static const GLenum reduced_prim[GL_POLYGON+1] = {
    GL_TRIANGLES
 };
 
-static const int scale_prim[GL_POLYGON+1] = {  
-   0,				/* fallback case */
+static const int scale_prim[GL_POLYGON + 1] = {
+   0,                           /* fallback case */
    1,
    2,
    2,
    1,
    3,
    3,
-   0,				/* fallback case */
-   0,				/* fallback case */
+   0,                           /* fallback case */
+   0,                           /* fallback case */
    3
 };
 
 
-static void intelDmaPrimitive( intelContextPtr intel, GLenum prim )
+static void
+intelDmaPrimitive(struct intel_context *intel, GLenum prim)
 {
-   if (0) fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
+   if (0)
+      fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
    INTEL_FIREVERTICES(intel);
-   intel->vtbl.reduced_primitive_state( intel, reduced_prim[prim] );
-   intelStartInlinePrimitive( intel, hw_prim[prim] );
+   intel->vtbl.reduced_primitive_state(intel, reduced_prim[prim]);
+   intelStartInlinePrimitive(intel, hw_prim[prim], INTEL_BATCH_CLIPRECTS);
 }
 
 
-#define LOCAL_VARS intelContextPtr intel = INTEL_CONTEXT(ctx)
+#define LOCAL_VARS struct intel_context *intel = intel_context(ctx)
 #define INIT( prim ) 				\
 do {						\
    intelDmaPrimitive( intel, prim );		\
 } while (0)
-#define FLUSH()  INTEL_FIREVERTICES( intel )
+
+#define FLUSH() INTEL_FIREVERTICES(intel)
 
 #define GET_SUBSEQUENT_VB_MAX_VERTS() \
-  (((intel->alloc.size / 2) - 1500) / (intel->vertex_size*4))
+  ((intel->batch->size - 1500) / (intel->vertex_size*4))
 #define GET_CURRENT_VB_MAX_VERTS() GET_SUBSEQUENT_VB_MAX_VERTS()
 
 #define ALLOC_VERTS( nr ) \
    intelExtendInlinePrimitive( intel, (nr) * intel->vertex_size )
-  
+
 #define EMIT_VERTS( ctx, j, nr, buf ) \
-  _tnl_emit_vertices_to_buffer(ctx, j, (j)+(nr), buf )  
+  _tnl_emit_vertices_to_buffer(ctx, j, (j)+(nr), buf )
 
 #define TAG(x) intel_##x
 #include "tnl_dd/t_dd_dmatmp.h"
-  
-  
+
+
 /**********************************************************************/
 /*                          Render pipeline stage                     */
 /**********************************************************************/
 
 /* Heuristic to choose between the two render paths:  
  */
-static GLboolean choose_render( intelContextPtr intel,
-				struct vertex_buffer *VB )
+static GLboolean
+choose_render(struct intel_context *intel, struct vertex_buffer *VB)
 {
    int vertsz = intel->vertex_size;
    int cost_render = 0;
@@ -153,20 +156,20 @@ static GLboolean choose_render( intelContextPtr intel,
    int nr_rverts = 0;
    int rprim = intel->reduced_primitive;
    int i = 0;
-   
-   for (i = 0 ; i < VB->PrimitiveCount ; i++) {
+
+   for (i = 0; i < VB->PrimitiveCount; i++) {
       GLuint prim = VB->Primitive[i].mode;
       GLuint length = VB->Primitive[i].count;
 
       if (!length)
-	 continue;
+         continue;
 
       nr_prims++;
       nr_rverts += length * scale_prim[prim & PRIM_MODE_MASK];
 
       if (reduced_prim[prim & PRIM_MODE_MASK] != rprim) {
-	 nr_rprims++;
-	 rprim = reduced_prim[prim & PRIM_MODE_MASK];
+         nr_rprims++;
+         rprim = reduced_prim[prim & PRIM_MODE_MASK];
       }
    }
 
@@ -177,64 +180,65 @@ static GLboolean choose_render( intelContextPtr intel,
 
    /* One point for every 1024 dwords (4k) of dma:
     */
-   cost_render += (vertsz * i) / 1024; 
-   cost_fallback += (vertsz * nr_rverts) / 1024; 
+   cost_render += (vertsz * i) / 1024;
+   cost_fallback += (vertsz * nr_rverts) / 1024;
 
    if (0)
       fprintf(stderr, "cost render: %d fallback: %d\n",
-	      cost_render, cost_fallback);
+              cost_render, cost_fallback);
 
-   if (cost_render > cost_fallback) 
+   if (cost_render > cost_fallback)
       return GL_FALSE;
 
    return GL_TRUE;
 }
 
 
-static GLboolean intel_run_render( GLcontext *ctx, 
-				 struct tnl_pipeline_stage *stage )
+static GLboolean
+intel_run_render(GLcontext * ctx, struct tnl_pipeline_stage *stage)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
    GLuint i;
 
+   intel->vtbl.render_prevalidate( intel );
+
    /* Don't handle clipping or indexed vertices.
     */
-   if (intel->RenderIndex != 0 || 
-       !intel_validate_render( ctx, VB ) || 
-       !choose_render( intel, VB )) {
+   if (intel->RenderIndex != 0 ||
+       !intel_validate_render(ctx, VB) || !choose_render(intel, VB)) {
       return GL_TRUE;
    }
 
    tnl->clipspace.new_inputs |= VERT_BIT_POS;
 
-   tnl->Driver.Render.Start( ctx );
-   
-   for (i = 0 ; i < VB->PrimitiveCount ; i++)
-   {
+   tnl->Driver.Render.Start(ctx);
+
+   for (i = 0; i < VB->PrimitiveCount; i++) {
       GLuint prim = VB->Primitive[i].mode;
       GLuint start = VB->Primitive[i].start;
       GLuint length = VB->Primitive[i].count;
 
       if (!length)
-	 continue;
+         continue;
 
-      intel_render_tab_verts[prim & PRIM_MODE_MASK]( ctx, start, start + length,
-						     prim );
+      intel_render_tab_verts[prim & PRIM_MODE_MASK] (ctx, start,
+                                                     start + length, prim);
    }
-      
-   tnl->Driver.Render.Finish( ctx );
 
-   return GL_FALSE;     /* finished the pipe */
+   tnl->Driver.Render.Finish(ctx);
+
+   INTEL_FIREVERTICES(intel);
+
+   return GL_FALSE;             /* finished the pipe */
 }
 
-const struct tnl_pipeline_stage _intel_render_stage =
-{
+const struct tnl_pipeline_stage _intel_render_stage = {
    "intel render",
    NULL,
    NULL,
    NULL,
    NULL,
-   intel_run_render	/* run */
+   intel_run_render             /* run */
 };
diff --git a/src/mesa/drivers/dri/i915/intel_rotate.c b/src/mesa/drivers/dri/i915/intel_rotate.c
index a77640ee54d..12d98c4ad2f 100644
--- a/src/mesa/drivers/dri/i915/intel_rotate.c
+++ b/src/mesa/drivers/dri/i915/intel_rotate.c
@@ -15,11 +15,14 @@
 
 void
 matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02,
-            int m10, int m11, int m12)
+            int m00, int m01, int m02, int m10, int m11, int m12)
 {
-   m->m00 = m00;   m->m01 = m01;   m->m02 = m02;
-   m->m10 = m10;   m->m11 = m11;   m->m12 = m12;
+   m->m00 = m00;
+   m->m01 = m01;
+   m->m02 = m02;
+   m->m10 = m10;
+   m->m11 = m11;
+   m->m12 = m12;
 }
 
 
@@ -66,9 +69,9 @@ matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist)
    *yDist = (y1 - y0) + (y2 - y0);
 
    if (*xDist < 0)
-       *xDist = -*xDist;
+      *xDist = -*xDist;
    if (*yDist < 0)
-       *yDist = -*yDist;
+      *yDist = -*yDist;
 }
 
 
@@ -76,7 +79,8 @@ matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist)
  * Transform the rect defined by (x, y, w, h) by m.
  */
 void
-matrix23TransformRect(const struct matrix23 *m, int *x, int *y, int *w, int *h)
+matrix23TransformRect(const struct matrix23 *m, int *x, int *y, int *w,
+                      int *h)
 {
    int x0 = *x, y0 = *y;
    int x1 = *x + *w, y1 = *y;
@@ -108,16 +112,16 @@ matrix23Rotate(struct matrix23 *m, int width, int height, int angle)
       matrix23Set(m, 1, 0, 0, 0, 1, 0);
       break;
    case 90:
-      matrix23Set(m, 0, 1, 0,  -1, 0, width);
+      matrix23Set(m, 0, 1, 0, -1, 0, width);
       break;
    case 180:
-      matrix23Set(m, -1, 0, width,  0, -1, height);
+      matrix23Set(m, -1, 0, width, 0, -1, height);
       break;
    case 270:
-      matrix23Set(m, 0, -1, height,  1, 0, 0);
+      matrix23Set(m, 0, -1, height, 1, 0, 0);
       break;
    default:
-      /*abort()*/;
+      /*abort() */ ;
    }
 }
 
@@ -129,16 +133,24 @@ void
 matrix23Flip(struct matrix23 *m, int width, int height, int xflip, int yflip)
 {
    if (xflip) {
-      m->m00 = -1;  m->m01 = 0;   m->m02 = width - 1;
+      m->m00 = -1;
+      m->m01 = 0;
+      m->m02 = width - 1;
    }
    else {
-      m->m00 = 1;   m->m01 = 0;   m->m02 = 0;
+      m->m00 = 1;
+      m->m01 = 0;
+      m->m02 = 0;
    }
    if (yflip) {
-      m->m10 = 0;   m->m11 = -1;  m->m12 = height - 1;
+      m->m10 = 0;
+      m->m11 = -1;
+      m->m12 = height - 1;
    }
    else {
-      m->m10 = 0;   m->m11 = 1;   m->m12 = 0;
+      m->m10 = 0;
+      m->m11 = 1;
+      m->m12 = 0;
    }
 }
 
@@ -169,14 +181,18 @@ main(int argc, char *argv[])
 {
    int width = 500, height = 400;
    int rot;
-   int fx = 0, fy = 0;  /* flip x and/or y ? */
+   int fx = 0, fy = 0;          /* flip x and/or y ? */
    int coords[4][2];
 
    /* four corner coords to test with */
-   coords[0][0] = 0;  coords[0][1] = 0;
-   coords[1][0] = width-1;  coords[1][1] = 0;
-   coords[2][0] = width-1;  coords[2][1] = height-1;
-   coords[3][0] = 0;  coords[3][1] = height-1;
+   coords[0][0] = 0;
+   coords[0][1] = 0;
+   coords[1][0] = width - 1;
+   coords[1][1] = 0;
+   coords[2][0] = width - 1;
+   coords[2][1] = height - 1;
+   coords[3][0] = 0;
+   coords[3][1] = height - 1;
 
 
    for (rot = 0; rot < 360; rot += 90) {
diff --git a/src/mesa/drivers/dri/i915/intel_rotate.h b/src/mesa/drivers/dri/i915/intel_rotate.h
index 0da45d20ce5..9c8802ca477 100644
--- a/src/mesa/drivers/dri/i915/intel_rotate.h
+++ b/src/mesa/drivers/dri/i915/intel_rotate.h
@@ -11,11 +11,9 @@ struct matrix23
 
 extern void
 matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02,
-            int m10, int m11, int m12);
+            int m00, int m01, int m02, int m10, int m11, int m12);
 
-extern void
-matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y);
+extern void matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y);
 
 extern void
 matrix23TransformCoordf(const struct matrix23 *m, float *x, float *y);
diff --git a/src/mesa/drivers/dri/i915/intel_screen.c b/src/mesa/drivers/dri/i915/intel_screen.c
index ca8610b4965..2721a900945 100644
--- a/src/mesa/drivers/dri/i915/intel_screen.c
+++ b/src/mesa/drivers/dri/i915/intel_screen.c
@@ -38,46 +38,48 @@
 
 #include "intel_screen.h"
 
+#include "intel_buffers.h"
 #include "intel_tex.h"
 #include "intel_span.h"
 #include "intel_tris.h"
 #include "intel_ioctl.h"
+#include "intel_fbo.h"
 
 #include "i830_dri.h"
+#include "dri_bufmgr.h"
+#include "intel_regions.h"
+#include "intel_batchbuffer.h"
 
 PUBLIC const char __driConfigOptions[] =
-DRI_CONF_BEGIN
-    DRI_CONF_SECTION_PERFORMANCE
-       DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS) 
-       DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
-    DRI_CONF_SECTION_END
-    DRI_CONF_SECTION_QUALITY
-       DRI_CONF_FORCE_S3TC_ENABLE(false)
-       DRI_CONF_ALLOW_LARGE_TEXTURES(1)
-      DRI_CONF_SECTION_END
-DRI_CONF_END;
-const GLuint __driNConfigOptions = 4;
+   DRI_CONF_BEGIN DRI_CONF_SECTION_PERFORMANCE
+   DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS)
+   DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
+   DRI_CONF_SECTION_END DRI_CONF_SECTION_QUALITY
+   DRI_CONF_FORCE_S3TC_ENABLE(false)
+   DRI_CONF_ALLOW_LARGE_TEXTURES(1)
+   DRI_CONF_SECTION_END DRI_CONF_END;
+     const GLuint __driNConfigOptions = 4;
 
 #ifdef USE_NEW_INTERFACE
-static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
-#endif /*USE_NEW_INTERFACE*/
+     static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
+#endif /*USE_NEW_INTERFACE */
 
-extern const struct dri_extension card_extensions[];
+     extern const struct dri_extension card_extensions[];
 
 /**
  * Map all the memory regions described by the screen.
  * \return GL_TRUE if success, GL_FALSE if error.
  */
 GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv)
+intelMapScreenRegions(__DRIscreenPrivate * sPriv)
 {
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
 
    if (intelScreen->front.handle) {
       if (drmMap(sPriv->fd,
                  intelScreen->front.handle,
                  intelScreen->front.size,
-                 (drmAddress *)&intelScreen->front.map) != 0) {
+                 (drmAddress *) & intelScreen->front.map) != 0) {
          _mesa_problem(NULL, "drmMap(frontbuffer) failed!");
          return GL_FALSE;
       }
@@ -86,42 +88,187 @@ intelMapScreenRegions(__DRIscreenPrivate *sPriv)
       _mesa_warning(NULL, "no front buffer handle in intelMapScreenRegions!");
    }
 
+   if (0)
+      _mesa_printf("Back 0x%08x ", intelScreen->back.handle);
    if (drmMap(sPriv->fd,
               intelScreen->back.handle,
               intelScreen->back.size,
-              (drmAddress *)&intelScreen->back.map) != 0) {
+              (drmAddress *) & intelScreen->back.map) != 0) {
       intelUnmapScreenRegions(intelScreen);
       return GL_FALSE;
    }
 
+   if (intelScreen->third.handle) {
+      if (0)
+	 _mesa_printf("Third 0x%08x ", intelScreen->third.handle);
+      if (drmMap(sPriv->fd,
+		 intelScreen->third.handle,
+		 intelScreen->third.size,
+		 (drmAddress *) & intelScreen->third.map) != 0) {
+	 intelUnmapScreenRegions(intelScreen);
+	 return GL_FALSE;
+      }
+   }
+
+   if (0)
+      _mesa_printf("Depth 0x%08x ", intelScreen->depth.handle);
    if (drmMap(sPriv->fd,
               intelScreen->depth.handle,
               intelScreen->depth.size,
-              (drmAddress *)&intelScreen->depth.map) != 0) {
+              (drmAddress *) & intelScreen->depth.map) != 0) {
       intelUnmapScreenRegions(intelScreen);
       return GL_FALSE;
    }
 
-   if (drmMap(sPriv->fd,
-              intelScreen->tex.handle,
-              intelScreen->tex.size,
-              (drmAddress *)&intelScreen->tex.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
+   if (0)
+      _mesa_printf("TEX 0x%08x ", intelScreen->tex.handle);
+   if (intelScreen->tex.size != 0) {
+      if (drmMap(sPriv->fd,
+		 intelScreen->tex.handle,
+		 intelScreen->tex.size,
+		 (drmAddress *) & intelScreen->tex.map) != 0) {
+	 intelUnmapScreenRegions(intelScreen);
+	 return GL_FALSE;
+      }
    }
 
    if (0)
-      printf("Mappings:  front: %p  back: %p  depth: %p  tex: %p\n",
-          intelScreen->front.map,
-          intelScreen->back.map,
-          intelScreen->depth.map,
-          intelScreen->tex.map);
+      printf("Mappings:  front: %p  back: %p  third: %p  depth: %p  tex: %p\n",
+             intelScreen->front.map,
+             intelScreen->back.map, intelScreen->third.map,
+             intelScreen->depth.map, intelScreen->tex.map);
    return GL_TRUE;
 }
 
+/** Driver-specific fence emit implementation for the fake memory manager. */
+static unsigned int
+intel_fence_emit(void *private)
+{
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *)private;
+   unsigned int fence;
+
+   /* XXX: Need to emit a flush, if we haven't already (at least with the
+    * current batchbuffer implementation, we have).
+    */
+
+   fence = intelEmitIrqLocked(intelScreen);
+
+   return fence;
+}
+
+/** Driver-specific fence wait implementation for the fake memory manager. */
+static int
+intel_fence_wait(void *private, unsigned int cookie)
+{
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *)private;
+
+   intelWaitIrq(intelScreen, cookie);
+
+   return 0;
+}
+
+static struct intel_region *
+intel_recreate_static(intelScreenPrivate *intelScreen,
+		      struct intel_region *region,
+		      intelRegion *region_desc,
+		      GLuint mem_type)
+{
+  if (region) {
+    intel_region_update_static(intelScreen, region, mem_type,
+			       region_desc->bo_handle, region_desc->offset,
+			       region_desc->map, intelScreen->cpp,
+			       region_desc->pitch / intelScreen->cpp,
+			       intelScreen->height);
+  } else {
+    region = intel_region_create_static(intelScreen, mem_type,
+					region_desc->bo_handle,
+					region_desc->offset,
+					region_desc->map, intelScreen->cpp,
+					region_desc->pitch / intelScreen->cpp,
+					intelScreen->height);
+  }
+
+  assert(region->buffer != NULL);
+
+  return region;
+}
+    
+
+/* Create intel_region structs to describe the static front,back,depth
+ * buffers created by the xserver. 
+ *
+ * Although FBO's mean we now no longer use these as render targets in
+ * all circumstances, they won't go away until the back and depth
+ * buffers become private, and the front and rotated buffers will
+ * remain even then.
+ *
+ * Note that these don't allocate video memory, just describe
+ * allocations alread made by the X server.
+ */
+static void
+intel_recreate_static_regions(intelScreenPrivate *intelScreen)
+{
+   intelScreen->front_region =
+      intel_recreate_static(intelScreen,
+			    intelScreen->front_region,
+			    &intelScreen->front,
+			    DRM_BO_FLAG_MEM_TT);
+
+   /* The rotated region is only used for old DDXes that didn't handle rotation
+\    * on their own.
+    */
+   if (intelScreen->driScrnPriv->ddxMinor < 8) {
+      intelScreen->rotated_region =
+	 intel_recreate_static(intelScreen,
+			       intelScreen->rotated_region,
+			       &intelScreen->rotated,
+			       DRM_BO_FLAG_MEM_TT);
+   }
+
+   intelScreen->back_region =
+      intel_recreate_static(intelScreen,
+			    intelScreen->back_region,
+			    &intelScreen->back,
+			    DRM_BO_FLAG_MEM_TT);
+
+   if (intelScreen->third.handle) {
+      intelScreen->third_region =
+	 intel_recreate_static(intelScreen,
+			       intelScreen->third_region,
+			       &intelScreen->third,
+			       DRM_BO_FLAG_MEM_TT);
+   }
+
+   /* Still assumes front.cpp == depth.cpp.  We can kill this when we move to
+    * private buffers.
+    */
+   intelScreen->depth_region =
+      intel_recreate_static(intelScreen,
+			    intelScreen->depth_region,
+			    &intelScreen->depth,
+			    DRM_BO_FLAG_MEM_TT);
+}
+
+/**
+ * Use the information in the sarea to update the screen parameters
+ * related to screen rotation. Needs to be called locked.
+ */
+void
+intelUpdateScreenRotation(__DRIscreenPrivate * sPriv, drmI830Sarea * sarea)
+{
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+
+   intelUnmapScreenRegions(intelScreen);
+   intelUpdateScreenFromSAREA(intelScreen, sarea);
+   if (!intelMapScreenRegions(sPriv)) {
+      fprintf(stderr, "ERROR Remapping screen regions!!!\n");
+   }
+   intel_recreate_static_regions(intelScreen);
+}
+
 
 void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen)
+intelUnmapScreenRegions(intelScreenPrivate * intelScreen)
 {
 #define REALLY_UNMAP 1
    if (intelScreen->front.map) {
@@ -138,6 +285,13 @@ intelUnmapScreenRegions(intelScreenPrivate *intelScreen)
 #endif
       intelScreen->back.map = NULL;
    }
+   if (intelScreen->third.map) {
+#if REALLY_UNMAP
+      if (drmUnmap(intelScreen->third.map, intelScreen->third.size) != 0)
+         printf("drmUnmap third failed!\n");
+#endif
+      intelScreen->third.map = NULL;
+   }
    if (intelScreen->depth.map) {
 #if REALLY_UNMAP
       drmUnmap(intelScreen->depth.map, intelScreen->depth.size);
@@ -154,9 +308,8 @@ intelUnmapScreenRegions(intelScreenPrivate *intelScreen)
 
 
 static void
-intelPrintDRIInfo(intelScreenPrivate *intelScreen,
-                  __DRIscreenPrivate *sPriv,
-                  I830DRIPtr gDRIPriv)
+intelPrintDRIInfo(intelScreenPrivate * intelScreen,
+                  __DRIscreenPrivate * sPriv, I830DRIPtr gDRIPriv)
 {
    fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
            intelScreen->front.size, intelScreen->front.offset,
@@ -177,9 +330,10 @@ intelPrintDRIInfo(intelScreenPrivate *intelScreen,
 
 
 static void
-intelPrintSAREA(const drmI830Sarea *sarea)
+intelPrintSAREA(const drmI830Sarea * sarea)
 {
-   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width, sarea->height);
+   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width,
+           sarea->height);
    fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
    fprintf(stderr,
            "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
@@ -193,8 +347,7 @@ intelPrintSAREA(const drmI830Sarea *sarea)
            sarea->depth_offset, sarea->depth_size,
            (unsigned) sarea->depth_handle);
    fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->tex_offset, sarea->tex_size,
-           (unsigned) sarea->tex_handle);
+           sarea->tex_offset, sarea->tex_size, (unsigned) sarea->tex_handle);
    fprintf(stderr, "SAREA: rotation: %d\n", sarea->rotation);
    fprintf(stderr,
            "SAREA: rotated offset: 0x%08x  size: 0x%x\n",
@@ -208,8 +361,8 @@ intelPrintSAREA(const drmI830Sarea *sarea)
  * information in the SAREA.  This function updates those parameters.
  */
 void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           drmI830Sarea *sarea)
+intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
+                           drmI830Sarea * sarea)
 {
    intelScreen->width = sarea->width;
    intelScreen->height = sarea->height;
@@ -223,12 +376,31 @@ intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
    intelScreen->back.pitch = sarea->pitch * intelScreen->cpp;
    intelScreen->back.handle = sarea->back_handle;
    intelScreen->back.size = sarea->back_size;
-			 
+
+   if (intelScreen->driScrnPriv->ddxMinor >= 8) {
+      intelScreen->third.offset = sarea->third_offset;
+      intelScreen->third.pitch = sarea->pitch * intelScreen->cpp;
+      intelScreen->third.handle = sarea->third_handle;
+      intelScreen->third.size = sarea->third_size;
+   }
+
    intelScreen->depth.offset = sarea->depth_offset;
    intelScreen->depth.pitch = sarea->pitch * intelScreen->cpp;
    intelScreen->depth.handle = sarea->depth_handle;
    intelScreen->depth.size = sarea->depth_size;
 
+   if (intelScreen->driScrnPriv->ddxMinor >= 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;
+   }
+
    intelScreen->tex.offset = sarea->tex_offset;
    intelScreen->logTextureGranularity = sarea->log_tex_granularity;
    intelScreen->tex.handle = sarea->tex_handle;
@@ -248,58 +420,75 @@ intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
 }
 
 
-static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
+static GLboolean
+intelInitDriver(__DRIscreenPrivate * sPriv)
 {
    intelScreenPrivate *intelScreen;
-   I830DRIPtr         gDRIPriv = (I830DRIPtr)sPriv->pDevPriv;
+   I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
    drmI830Sarea *sarea;
+
    PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-     (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->getProcAddress("glxEnableExtension"));
-   void * const psc = sPriv->psc->screenConfigs;
+      (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->
+                                      getProcAddress("glxEnableExtension"));
+   void *const psc = sPriv->psc->screenConfigs;
 
    if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
-      fprintf(stderr,"\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
+      fprintf(stderr,
+              "\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
       return GL_FALSE;
    }
 
    /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *)CALLOC(sizeof(intelScreenPrivate));
+   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
    if (!intelScreen) {
-      fprintf(stderr,"\nERROR!  Allocating private area failed\n");
+      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
       return GL_FALSE;
    }
    /* parse information in __driConfigOptions */
-   driParseOptionInfo (&intelScreen->optionCache,
-		       __driConfigOptions, __driNConfigOptions);
+   driParseOptionInfo(&intelScreen->optionCache,
+                      __driConfigOptions, __driNConfigOptions);
 
    intelScreen->driScrnPriv = sPriv;
-   sPriv->private = (void *)intelScreen;
+   sPriv->private = (void *) intelScreen;
    intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
    sarea = (drmI830Sarea *)
-         (((GLubyte *)sPriv->pSAREA)+intelScreen->sarea_priv_offset);
+      (((GLubyte *) sPriv->pSAREA) + intelScreen->sarea_priv_offset);
 
    intelScreen->deviceID = gDRIPriv->deviceID;
+   if (intelScreen->deviceID == PCI_CHIP_I865_G)
+      intelScreen->maxBatchSize = 4096;
+   else
+      intelScreen->maxBatchSize = BATCH_SZ;
+
    intelScreen->mem = gDRIPriv->mem;
    intelScreen->cpp = gDRIPriv->cpp;
 
    switch (gDRIPriv->bitsPerPixel) {
-   case 15: intelScreen->fbFormat = DV_PF_555; break;
-   case 16: intelScreen->fbFormat = DV_PF_565; break;
-   case 32: intelScreen->fbFormat = DV_PF_8888; break;
+   case 16:
+      intelScreen->fbFormat = DV_PF_565;
+      break;
+   case 32:
+      intelScreen->fbFormat = DV_PF_8888;
+      break;
+   default:
+      exit(1);
+      break;
    }
-			 
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
 
-   if (0)
-      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
+   intelUpdateScreenFromSAREA(intelScreen, sarea);
 
    if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr,"\nERROR!  mapping regions\n");
+      fprintf(stderr, "\nERROR!  mapping regions\n");
       _mesa_free(intelScreen);
       sPriv->private = NULL;
       return GL_FALSE;
    }
 
+   intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
+
+   if (0)
+      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
+
    intelScreen->drmMinor = sPriv->drmMinor;
 
    /* Determine if IRQs are active? */
@@ -310,11 +499,11 @@ static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
       gp.param = I830_PARAM_IRQ_ACTIVE;
       gp.value = &intelScreen->irq_active;
 
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
+      ret = drmCommandWriteRead(sPriv->fd, DRM_I830_GETPARAM,
+                                &gp, sizeof(gp));
       if (ret) {
-	 fprintf(stderr, "drmI830GetParam: %d\n", ret);
-	 return GL_FALSE;
+         fprintf(stderr, "drmI830GetParam: %d\n", ret);
+         return GL_FALSE;
       }
    }
 
@@ -326,128 +515,182 @@ static GLboolean intelInitDriver(__DRIscreenPrivate *sPriv)
       gp.param = I830_PARAM_ALLOW_BATCHBUFFER;
       gp.value = &intelScreen->allow_batchbuffer;
 
-      ret = drmCommandWriteRead( sPriv->fd, DRM_I830_GETPARAM,
-				 &gp, sizeof(gp));
+      ret = drmCommandWriteRead(sPriv->fd, DRM_I830_GETPARAM,
+                                &gp, sizeof(gp));
       if (ret) {
-	 fprintf(stderr, "drmI830GetParam: (%d) %d\n", gp.param, ret);
-	 return GL_FALSE;
+         fprintf(stderr, "drmI830GetParam: (%d) %d\n", gp.param, ret);
+         return GL_FALSE;
       }
    }
 
    if (glx_enable_extension != NULL) {
-      (*glx_enable_extension)( psc, "GLX_SGI_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_SGI_video_sync" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_control" );
-      (*glx_enable_extension)( psc, "GLX_MESA_swap_frame_usage" );
-      (*glx_enable_extension)( psc, "GLX_SGI_make_current_read" );
-      (*glx_enable_extension)( psc, "GLX_MESA_allocate_memory" );
-      (*glx_enable_extension)( psc, "GLX_MESA_copy_sub_buffer" );
-   }
-   
-   sPriv->psc->allocateMemory = (void *) intelAllocateMemoryMESA;
-   sPriv->psc->freeMemory     = (void *) intelFreeMemoryMESA;
-   sPriv->psc->memoryOffset   = (void *) intelGetMemoryOffsetMESA;
+      (*glx_enable_extension) (psc, "GLX_SGI_swap_control");
+      (*glx_enable_extension) (psc, "GLX_SGI_video_sync");
+      (*glx_enable_extension) (psc, "GLX_MESA_swap_control");
+      (*glx_enable_extension) (psc, "GLX_MESA_swap_frame_usage");
+      (*glx_enable_extension) (psc, "GLX_SGI_make_current_read");
+   }
+
+   /* If we've got a new enough DDX that's initializing TTM and giving us
+    * object handles for the shared buffers, use that.
+    */
+   intelScreen->ttm = GL_FALSE;
+   if (getenv("INTEL_NO_TTM") == NULL &&
+       intelScreen->driScrnPriv->ddxMinor >= 9 &&
+       intelScreen->front.bo_handle != -1) {
+      intelScreen->bufmgr = dri_bufmgr_ttm_init(sPriv->fd,
+						DRM_FENCE_TYPE_EXE,
+						DRM_FENCE_TYPE_EXE |
+						DRM_I915_FENCE_TYPE_RW);
+      if (intelScreen->bufmgr != NULL)
+	 intelScreen->ttm = GL_TRUE;
+   }
+   /* Otherwise, use the classic buffer manager. */
+   if (intelScreen->bufmgr == NULL) {
+      if (intelScreen->tex.size == 0) {
+	 fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
+		 __func__, __LINE__);
+	 return GL_FALSE;
+      }
+      fprintf(stderr, "[%s:%u] Failed to init TTM buffer manager, falling back"
+	      " to classic.\n", __func__, __LINE__);
+      intelScreen->bufmgr = dri_bufmgr_fake_init(intelScreen->tex.offset,
+						 intelScreen->tex.map,
+						 intelScreen->tex.size,
+						 intel_fence_emit,
+						 intel_fence_wait,
+						 intelScreen);
+   }
+
+   intel_recreate_static_regions(intelScreen);
 
    return GL_TRUE;
 }
-		
-		
-static void intelDestroyScreen(__DRIscreenPrivate *sPriv)
+
+
+static void
+intelDestroyScreen(__DRIscreenPrivate * sPriv)
 {
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
 
    intelUnmapScreenRegions(intelScreen);
 
-   driDestroyOptionInfo (&intelScreen->optionCache);
-
+   dri_bufmgr_destroy(intelScreen->bufmgr);
    FREE(intelScreen);
    sPriv->private = NULL;
 }
 
 
-static GLboolean intelCreateBuffer( __DRIscreenPrivate *driScrnPriv,
-				    __DRIdrawablePrivate *driDrawPriv,
-				    const __GLcontextModes *mesaVis,
-				    GLboolean isPixmap )
+/**
+ * This is called when we need to set up GL rendering to a new X window.
+ */
+static GLboolean
+intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
+                  __DRIdrawablePrivate * driDrawPriv,
+                  const __GLcontextModes * mesaVis, GLboolean isPixmap)
 {
    intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
 
    if (isPixmap) {
-      return GL_FALSE; /* not implemented */
-   } else {
-      GLboolean swStencil = (mesaVis->stencilBits > 0 && 
-			     mesaVis->depthBits != 24);
+      return GL_FALSE;          /* not implemented */
+   }
+   else {
+      GLboolean swStencil = (mesaVis->stencilBits > 0 &&
+                             mesaVis->depthBits != 24);
+      GLenum rgbFormat = (mesaVis->redBits == 5 ? GL_RGB5 : GL_RGBA8);
+
+      struct intel_framebuffer *intel_fb = CALLOC_STRUCT(intel_framebuffer);
+
+      if (!intel_fb)
+	 return GL_FALSE;
 
-      struct gl_framebuffer *fb = _mesa_create_framebuffer(mesaVis);
+      _mesa_initialize_framebuffer(&intel_fb->Base, mesaVis);
 
+      /* setup the hardware-based renderbuffers */
       {
-         driRenderbuffer *frontRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->front.map,
-                                 screen->cpp,
-                                 screen->front.offset, screen->front.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(frontRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &frontRb->Base);
+         intel_fb->color_rb[0]
+            = intel_create_renderbuffer(rgbFormat,
+                                        screen->width, screen->height,
+                                        screen->front.offset,
+                                        screen->front.pitch,
+                                        screen->cpp,
+                                        screen->front.map);
+         intel_set_span_functions(&intel_fb->color_rb[0]->Base);
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT,
+				&intel_fb->color_rb[0]->Base);
       }
 
       if (mesaVis->doubleBufferMode) {
-         driRenderbuffer *backRb
-            = driNewRenderbuffer(GL_RGBA,
-                                 screen->back.map,
-                                 screen->cpp,
-                                 screen->back.offset, screen->back.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(backRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &backRb->Base);
+         intel_fb->color_rb[1]
+            = intel_create_renderbuffer(rgbFormat,
+                                        screen->width, screen->height,
+                                        screen->back.offset,
+                                        screen->back.pitch,
+                                        screen->cpp,
+                                        screen->back.map);
+         intel_set_span_functions(&intel_fb->color_rb[1]->Base);
+         _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,
+					   screen->width, screen->height,
+					   screen->third.offset,
+					   screen->third.pitch,
+					   screen->cpp,
+					   screen->third.map);
+	    intel_set_span_functions(&intel_fb->color_rb[2]->Base);
+	    _mesa_reference_renderbuffer(&tmp_rb, &intel_fb->color_rb[2]->Base);
+	 }
       }
 
-      if (mesaVis->depthBits == 16) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT16,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
+      if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
+         /* combined depth/stencil buffer */
+         struct intel_renderbuffer *depthStencilRb
+            = intel_create_renderbuffer(GL_DEPTH24_STENCIL8_EXT,
+                                        screen->width, screen->height,
+                                        screen->depth.offset,
+                                        screen->depth.pitch,
+                                        screen->cpp,    /* 4! */
+                                        screen->depth.map);
+         intel_set_span_functions(&depthStencilRb->Base);
+         /* note: bind RB to two attachment points */
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
+				&depthStencilRb->Base);
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_STENCIL,
+				&depthStencilRb->Base);
       }
-      else if (mesaVis->depthBits == 24) {
-         driRenderbuffer *depthRb
-            = driNewRenderbuffer(GL_DEPTH_COMPONENT24,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(depthRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
+      else if (mesaVis->depthBits == 16) {
+         /* just 16-bit depth buffer, no hw stencil */
+         struct intel_renderbuffer *depthRb
+            = intel_create_renderbuffer(GL_DEPTH_COMPONENT16,
+                                        screen->width, screen->height,
+                                        screen->depth.offset,
+                                        screen->depth.pitch,
+                                        screen->cpp,    /* 2! */
+                                        screen->depth.map);
+         intel_set_span_functions(&depthRb->Base);
+         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH, &depthRb->Base);
       }
 
-      if (mesaVis->stencilBits > 0 && !swStencil) {
-         driRenderbuffer *stencilRb
-            = driNewRenderbuffer(GL_STENCIL_INDEX8_EXT,
-                                 screen->depth.map,
-                                 screen->cpp,
-                                 screen->depth.offset, screen->depth.pitch,
-                                 driDrawPriv);
-         intelSetSpanFunctions(stencilRb, mesaVis);
-         _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &stencilRb->Base);
-      }
-
-      _mesa_add_soft_renderbuffers(fb,
-                                   GL_FALSE, /* color */
-                                   GL_FALSE, /* depth */
-                                   swStencil,
-                                   mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* alpha */
-                                   GL_FALSE /* aux */);
-      driDrawPriv->driverPrivate = (void *) fb;
+      /* now add any/all software-based renderbuffers we may need */
+      _mesa_add_soft_renderbuffers(&intel_fb->Base,
+                                   GL_FALSE, /* never sw color */
+                                   GL_FALSE, /* never sw depth */
+                                   swStencil, mesaVis->accumRedBits > 0,
+                                   GL_FALSE, /* never sw alpha */
+                                   GL_FALSE  /* never sw aux */ );
+      driDrawPriv->driverPrivate = (void *) intel_fb;
 
-      return (driDrawPriv->driverPrivate != NULL);
+      return GL_TRUE;
    }
 }
 
-static void intelDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
+static void
+intelDestroyBuffer(__DRIdrawablePrivate * driDrawPriv)
 {
    _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
 }
@@ -457,24 +700,23 @@ static void intelDestroyBuffer(__DRIdrawablePrivate *driDrawPriv)
  * Get information about previous buffer swaps.
  */
 static int
-intelGetSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
+intelGetSwapInfo(__DRIdrawablePrivate * dPriv, __DRIswapInfo * sInfo)
 {
-   intelContextPtr intel;
+   struct intel_framebuffer *intel_fb;
 
-   if ( (dPriv == NULL) || (dPriv->driContextPriv == NULL)
-	|| (dPriv->driContextPriv->driverPrivate == NULL)
-	|| (sInfo == NULL) ) {
+   if ((dPriv == NULL) || (dPriv->driverPrivate == NULL)
+       || (sInfo == NULL)) {
       return -1;
    }
 
-   intel = dPriv->driContextPriv->driverPrivate;
-   sInfo->swap_count = intel->swap_count;
-   sInfo->swap_ust = intel->swap_ust;
-   sInfo->swap_missed_count = intel->swap_missed_count;
+   intel_fb = dPriv->driverPrivate;
+   sInfo->swap_count = intel_fb->swap_count;
+   sInfo->swap_ust = intel_fb->swap_ust;
+   sInfo->swap_missed_count = intel_fb->swap_missed_count;
 
    sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
-       ? driCalculateSwapUsage( dPriv, 0, intel->swap_missed_ust )
-       : 0.0;
+      ? driCalculateSwapUsage(dPriv, 0, intel_fb->swap_missed_ust)
+      : 0.0;
 
    return 0;
 }
@@ -484,31 +726,33 @@ intelGetSwapInfo( __DRIdrawablePrivate *dPriv, __DRIswapInfo * sInfo )
  * init-designated function to register chipids and createcontext
  * functions.
  */
-extern GLboolean i830CreateContext( const __GLcontextModes *mesaVis,
-				    __DRIcontextPrivate *driContextPriv,
-				    void *sharedContextPrivate);
+extern GLboolean i830CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
 
-extern GLboolean i915CreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate);
+extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
+                                   __DRIcontextPrivate * driContextPriv,
+                                   void *sharedContextPrivate);
 
 
 
 
-static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
-				   __DRIcontextPrivate *driContextPriv,
-				   void *sharedContextPrivate)
+static GLboolean
+intelCreateContext(const __GLcontextModes * mesaVis,
+                   __DRIcontextPrivate * driContextPriv,
+                   void *sharedContextPrivate)
 {
    __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *)sPriv->private;
+   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
 
    switch (intelScreen->deviceID) {
+      /* Don't deal with i830 until texture work complete:
+       */
    case PCI_CHIP_845_G:
    case PCI_CHIP_I830_M:
    case PCI_CHIP_I855_GM:
    case PCI_CHIP_I865_G:
-      return i830CreateContext( mesaVis, driContextPriv, 
-				sharedContextPrivate );
+      return i830CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
 
    case PCI_CHIP_I915_G:
    case PCI_CHIP_I915_GM:
@@ -518,9 +762,8 @@ static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
    case PCI_CHIP_G33_G:
    case PCI_CHIP_Q35_G:
    case PCI_CHIP_Q33_G:
-      return i915CreateContext( mesaVis, driContextPriv, 
-			       sharedContextPrivate );
- 
+      return i915CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
+
    default:
       fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
       return GL_FALSE;
@@ -529,30 +772,31 @@ static GLboolean intelCreateContext( const __GLcontextModes *mesaVis,
 
 
 static const struct __DriverAPIRec intelAPI = {
-   .InitDriver      = intelInitDriver,
-   .DestroyScreen   = intelDestroyScreen,
-   .CreateContext   = intelCreateContext,
-   .DestroyContext  = intelDestroyContext,
-   .CreateBuffer    = intelCreateBuffer,
-   .DestroyBuffer   = intelDestroyBuffer,
-   .SwapBuffers     = intelSwapBuffers,
-   .MakeCurrent     = intelMakeCurrent,
-   .UnbindContext   = intelUnbindContext,
-   .GetSwapInfo     = intelGetSwapInfo,
-   .GetMSC          = driGetMSC32,
-   .WaitForMSC      = driWaitForMSC32,
-   .WaitForSBC      = NULL,
-   .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = intelCopySubBuffer
+   .InitDriver = intelInitDriver,
+   .DestroyScreen = intelDestroyScreen,
+   .CreateContext = intelCreateContext,
+   .DestroyContext = intelDestroyContext,
+   .CreateBuffer = intelCreateBuffer,
+   .DestroyBuffer = intelDestroyBuffer,
+   .SwapBuffers = intelSwapBuffers,
+   .MakeCurrent = intelMakeCurrent,
+   .UnbindContext = intelUnbindContext,
+   .GetSwapInfo = intelGetSwapInfo,
+   .GetMSC = driGetMSC32,
+   .WaitForMSC = driWaitForMSC32,
+   .WaitForSBC = NULL,
+   .SwapBuffersMSC = NULL,
+   .CopySubBuffer = intelCopySubBuffer,
+   .setTexOffset = intelSetTexOffset,
 };
 
 
 static __GLcontextModes *
-intelFillInModes( unsigned pixel_bits, unsigned depth_bits,
-		 unsigned stencil_bits, GLboolean have_back_buffer )
+intelFillInModes(unsigned pixel_bits, unsigned depth_bits,
+                 unsigned stencil_bits, GLboolean have_back_buffer)
 {
-   __GLcontextModes * modes;
-   __GLcontextModes * m;
+   __GLcontextModes *modes;
+   __GLcontextModes *m;
    unsigned num_modes;
    unsigned depth_buffer_factor;
    unsigned back_buffer_factor;
@@ -580,46 +824,51 @@ intelFillInModes( unsigned pixel_bits, unsigned depth_bits,
     */
    stencil_bits_array[0] = 0;
    stencil_bits_array[1] = 0;
+   if (depth_bits == 24)
+      stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
+
    stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
 
    depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
-   back_buffer_factor  = (have_back_buffer) ? 3 : 1;
+   back_buffer_factor = (have_back_buffer) ? 3 : 1;
 
    num_modes = depth_buffer_factor * back_buffer_factor * 4;
 
-    if ( pixel_bits == 16 ) {
-        fb_format = GL_RGB;
-        fb_type = GL_UNSIGNED_SHORT_5_6_5;
-    }
-    else {
-        fb_format = GL_BGRA;
-        fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
-    }
+   if (pixel_bits == 16) {
+      fb_format = GL_RGB;
+      fb_type = GL_UNSIGNED_SHORT_5_6_5;
+   }
+   else {
+      fb_format = GL_BGRA;
+      fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
+   }
 
-   modes = (*dri_interface->createContextModes)( num_modes, sizeof( __GLcontextModes ) );
+   modes =
+      (*dri_interface->createContextModes) (num_modes,
+                                            sizeof(__GLcontextModes));
    m = modes;
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_TRUE_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
-   }
-   if ( ! driFillInModes( & m, fb_format, fb_type,
-			  depth_bits_array, stencil_bits_array, depth_buffer_factor,
-			  back_buffer_modes, back_buffer_factor,
-			  GLX_DIRECT_COLOR ) ) {
-	fprintf( stderr, "[%s:%u] Error creating FBConfig!\n",
-		 __func__, __LINE__ );
-	return NULL;
+   if (!driFillInModes(&m, fb_format, fb_type,
+                       depth_bits_array, stencil_bits_array,
+                       depth_buffer_factor, back_buffer_modes,
+                       back_buffer_factor, GLX_TRUE_COLOR)) {
+      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
+   }
+   if (!driFillInModes(&m, fb_format, fb_type,
+                       depth_bits_array, stencil_bits_array,
+                       depth_buffer_factor, back_buffer_modes,
+                       back_buffer_factor, GLX_DIRECT_COLOR)) {
+      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
+              __LINE__);
+      return NULL;
    }
 
    /* Mark the visual as slow if there are "fake" stencil bits.
     */
-   for ( m = modes ; m != NULL ; m = m->next ) {
-      if ( (m->stencilBits != 0) && (m->stencilBits != stencil_bits) ) {
-	 m->visualRating = GLX_SLOW_CONFIG;
+   for (m = modes; m != NULL; m = m->next) {
+      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
+         m->visualRating = GLX_SLOW_CONFIG;
       }
    }
 
@@ -637,43 +886,43 @@ intelFillInModes( unsigned pixel_bits, unsigned depth_bits,
  * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
  *         failure.
  */
-PUBLIC
-void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIscreen *psc,
-			     const __GLcontextModes * modes,
-			     const __DRIversion * ddx_version,
-			     const __DRIversion * dri_version,
-			     const __DRIversion * drm_version,
-			     const __DRIframebuffer * frame_buffer,
-			     drmAddress pSAREA, int fd, 
-			     int internal_api_version,
-			     const __DRIinterfaceMethods * interface,
-			     __GLcontextModes ** driver_modes )
-			     
+PUBLIC void *
+__driCreateNewScreen_20050727(__DRInativeDisplay * dpy, int scrn,
+                              __DRIscreen * psc,
+                              const __GLcontextModes * modes,
+                              const __DRIversion * ddx_version,
+                              const __DRIversion * dri_version,
+                              const __DRIversion * drm_version,
+                              const __DRIframebuffer * frame_buffer,
+                              drmAddress pSAREA, int fd,
+                              int internal_api_version,
+                              const __DRIinterfaceMethods * interface,
+                              __GLcontextModes ** driver_modes)
 {
    __DRIscreenPrivate *psp;
    static const __DRIversion ddx_expected = { 1, 5, 0 };
    static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 4, 0 };
+   static const __DRIversion drm_expected = { 1, 5, 0 };
 
    dri_interface = interface;
 
-   if ( ! driCheckDriDdxDrmVersions2( "i915",
-				      dri_version, & dri_expected,
-				      ddx_version, & ddx_expected,
-				      drm_version, & drm_expected ) ) {
+   if (!driCheckDriDdxDrmVersions2("i915",
+                                   dri_version, &dri_expected,
+                                   ddx_version, &ddx_expected,
+                                   drm_version, &drm_expected)) {
       return NULL;
    }
 
    psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-				  ddx_version, dri_version, drm_version,
-				  frame_buffer, pSAREA, fd,
-				  internal_api_version, &intelAPI);
-   if ( psp != NULL ) {
+                                  ddx_version, dri_version, drm_version,
+                                  frame_buffer, pSAREA, fd,
+                                  internal_api_version, &intelAPI);
+
+   if (psp != NULL) {
       I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-      *driver_modes = intelFillInModes( dri_priv->cpp * 8,
-					(dri_priv->cpp == 2) ? 16 : 24,
-					(dri_priv->cpp == 2) ? 0  : 8,
-					1 );
+      *driver_modes = intelFillInModes(dri_priv->cpp * 8,
+                                       (dri_priv->cpp == 2) ? 16 : 24,
+                                       (dri_priv->cpp == 2) ? 0 : 8, 1);
 
       /* Calling driInitExtensions here, with a NULL context pointer, does not actually
        * enable the extensions.  It just makes sure that all the dispatch offsets for all
@@ -683,8 +932,24 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc
        *
        * Hello chicken.  Hello egg.  How are you two today?
        */
-      driInitExtensions( NULL, card_extensions, GL_FALSE );
+      driInitExtensions(NULL, card_extensions, GL_FALSE);
    }
 
    return (void *) psp;
 }
+
+struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen)
+{
+  /*
+   * This should probably change to have the screen allocate a dummy
+   * context at screen creation. For now just use the current context.
+   */
+
+  GET_CURRENT_CONTEXT(ctx);
+  if (ctx == NULL) {
+     _mesa_problem(NULL, "No current context in intelScreenContext\n");
+     return NULL;
+  }
+  return intel_context(ctx);
+}
+
diff --git a/src/mesa/drivers/dri/i915/intel_screen.h b/src/mesa/drivers/dri/i915/intel_screen.h
index 24cfd9bf8b2..aa0ef2c5090 100644
--- a/src/mesa/drivers/dri/i915/intel_screen.h
+++ b/src/mesa/drivers/dri/i915/intel_screen.h
@@ -29,39 +29,51 @@
 #define _INTEL_INIT_H_
 
 #include <sys/time.h>
-#include "xmlconfig.h"
 #include "dri_util.h"
 #include "intel_rotate.h"
 #include "i830_common.h"
+#include "xmlconfig.h"
+#include "dri_bufmgr.h"
 
-
-/* This roughly corresponds to a gl_renderbuffer (Mesa 6.4) */
-typedef struct {
+/* XXX: change name or eliminate to avoid conflict with "struct
+ * intel_region"!!!
+ */
+typedef struct
+{
    drm_handle_t handle;
-   drmSize size;        /* region size in bytes */
-   char *map;           /* memory map */
-   int offset;          /* from start of video mem, in bytes */
-   int pitch;           /* row stride, in bytes */
+   drmSize size;                /* region size in bytes */
+   char *map;                   /* memory map */
+   int offset;                  /* from start of video mem, in bytes */
+   int pitch;                   /* row stride, in bytes */
+   unsigned int bo_handle;	/* buffer object id if available, or -1 */
 } intelRegion;
 
-typedef struct 
+typedef struct
 {
    intelRegion front;
    intelRegion back;
+   intelRegion third;
    intelRegion rotated;
    intelRegion depth;
    intelRegion tex;
-   
+
+   struct intel_region *front_region;
+   struct intel_region *back_region;
+   struct intel_region *third_region;
+   struct intel_region *depth_region;
+   struct intel_region *rotated_region;
+
    int deviceID;
    int width;
    int height;
-   int mem;         /* unused */
-   
-   int cpp;         /* for front and back buffers */
-   int fbFormat;
+   int mem;                     /* unused */
+
+   int cpp;                     /* for front and back buffers */
+/*    int bitsPerPixel;   */
+   int fbFormat;                /* XXX FBO: this is obsolete - remove after i830 updates */
 
    int logTextureGranularity;
-   
+
    __DRIscreenPrivate *driScrnPriv;
    unsigned int sarea_priv_offset;
 
@@ -72,41 +84,51 @@ typedef struct
 
    struct matrix23 rotMatrix;
 
-   int current_rotation;  /* 0, 90, 180 or 270 */
+   int current_rotation;        /* 0, 90, 180 or 270 */
    int rotatedWidth, rotatedHeight;
 
    /**
    * Configuration cache with default values for all contexts
    */
    driOptionCache optionCache;
+
+   dri_bufmgr *bufmgr;
+   unsigned int maxBatchSize;
+
+   /**
+    * This value indicates that the kernel memory manager is being used
+    * instead of the fake client-side memory manager.
+    */
+   GLboolean ttm;
 } intelScreenPrivate;
 
 
-extern GLboolean
-intelMapScreenRegions(__DRIscreenPrivate *sPriv);
 
-extern void
-intelUnmapScreenRegions(intelScreenPrivate *intelScreen);
+extern GLboolean intelMapScreenRegions(__DRIscreenPrivate * sPriv);
 
-extern void
-intelUpdateScreenFromSAREA(intelScreenPrivate *intelScreen,
-                           drmI830Sarea *sarea);
+extern void intelUnmapScreenRegions(intelScreenPrivate * intelScreen);
 
 extern void
-intelDestroyContext(__DRIcontextPrivate *driContextPriv);
+intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
+                           drmI830Sarea * sarea);
 
-extern GLboolean
-intelUnbindContext(__DRIcontextPrivate *driContextPriv);
+extern void intelDestroyContext(__DRIcontextPrivate * driContextPriv);
+
+extern GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv);
 
 extern GLboolean
-intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
-                 __DRIdrawablePrivate *driDrawPriv,
-                 __DRIdrawablePrivate *driReadPriv);
+intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
+                 __DRIdrawablePrivate * driDrawPriv,
+                 __DRIdrawablePrivate * driReadPriv);
+
+extern void intelSwapBuffers(__DRIdrawablePrivate * dPriv);
 
 extern void
-intelSwapBuffers(__DRIdrawablePrivate *dPriv);
+intelCopySubBuffer(__DRIdrawablePrivate * dPriv, int x, int y, int w, int h);
+
+extern struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen);
 
 extern void
-intelCopySubBuffer( __DRIdrawablePrivate *dPriv, int x, int y, int w, int h );
+intelUpdateScreenRotation(__DRIscreenPrivate * sPriv, drmI830Sarea * sarea);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/intel_span.c b/src/mesa/drivers/dri/i915/intel_span.c
index c3ffc4b2ac0..d1f8ef06beb 100644
--- a/src/mesa/drivers/dri/i915/intel_span.c
+++ b/src/mesa/drivers/dri/i915/intel_span.c
@@ -30,229 +30,380 @@
 #include "mtypes.h"
 #include "colormac.h"
 
+#include "intel_fbo.h"
 #include "intel_screen.h"
-
 #include "intel_span.h"
+#include "intel_regions.h"
 #include "intel_ioctl.h"
+#include "intel_tex.h"
+
 #include "swrast/swrast.h"
 
+/*
+  break intelWriteRGBASpan_ARGB8888
+*/
 
+#undef DBG
 #define DBG 0
 
-#define LOCAL_VARS						\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLushort p;							\
-   (void) buf; (void) p
-
-#define LOCAL_DEPTH_VARS					\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *) drb->Base.Data +			\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch
-
-#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS 
-
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR565(color[0],color[1],color[2])
-
-#define Y_FLIP(_y) (height - _y - 1)
+#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;	\
+   GLubyte *buf = (GLubyte *) irb->pfMap				\
+      + (intel->drawY * irb->pfPitch + intel->drawX) * irb->region->cpp;\
+   GLuint p;								\
+   assert(irb->pfMap);\
+   (void) p;
+
+/* XXX FBO: this is identical to the macro in spantmp2.h except we get
+ * the cliprect info from the context, not the driDrawable.
+ * Move this into spantmp2.h someday.
+ */
+#define HW_CLIPLOOP()							\
+   do {									\
+      int _nc = intel->numClipRects;					\
+      while ( _nc-- ) {							\
+	 int minx = intel->pClipRects[_nc].x1 - intel->drawX;		\
+	 int miny = intel->pClipRects[_nc].y1 - intel->drawY;		\
+	 int maxx = intel->pClipRects[_nc].x2 - intel->drawX;		\
+	 int maxy = intel->pClipRects[_nc].y2 - intel->drawY;
+
+
+
+
+#define Y_FLIP(_y) ((_y) * yScale + yBias)
 
 #define HW_LOCK()
 
 #define HW_UNLOCK()
 
-/* 16 bit, 565 rgb color spanline and pixel functions
+/* 16 bit, RGB565 color spanline and pixel functions
  */
-#define WRITE_RGBA( _x, _y, r, g, b, a )				\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = ( (((int)r & 0xf8) << 8) |	\
-		                             (((int)g & 0xfc) << 3) |	\
-		                             (((int)b & 0xf8) >> 3))
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch) = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (((p >> 11) & 0x1f) * 255) / 31;			\
-   rgba[1] = (((p >>  5) & 0x3f) * 255) / 63;			\
-   rgba[2] = (((p >>  0) & 0x1f) * 255) / 31;			\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_565
-#include "spantmp.h"
-
-/* 15 bit, 555 rgb color spanline and pixel functions
- */
-#define WRITE_RGBA( _x, _y, r, g, b, a )			\
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = (((r & 0xf8) << 7) |	\
-		                            ((g & 0xf8) << 3) |	\
-                         		    ((b & 0xf8) >> 3))
-
-#define WRITE_PIXEL( _x, _y, p )  \
-   *(GLushort *)(buf + _x*2 + _y*pitch)  = p
-
-#define READ_RGBA( rgba, _x, _y )				\
-do {								\
-   GLushort p = *(GLushort *)(buf + _x*2 + _y*pitch);		\
-   rgba[0] = (p >> 7) & 0xf8;					\
-   rgba[1] = (p >> 3) & 0xf8;					\
-   rgba[2] = (p << 3) & 0xf8;					\
-   rgba[3] = 255;						\
-} while(0)
-
-#define TAG(x) intel##x##_555
-#include "spantmp.h"
-
-/* 16 bit depthbuffer functions.
+#define SPANTMP_PIXEL_FMT GL_RGB
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
+
+#define TAG(x)    intel##x##_RGB565
+#define TAG2(x,y) intel##x##_RGB565##y
+#define GET_PTR(X,Y) (buf + ((Y) * irb->pfPitch + (X)) * 2)
+#include "spantmp2.h"
+
+/* 32 bit, ARGB8888 color spanline and pixel functions
  */
-#define WRITE_DEPTH( _x, _y, d ) \
-   *(GLushort *)(buf + (_x)*2 + (_y)*pitch)  = d;
+#define SPANTMP_PIXEL_FMT GL_BGRA
+#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
 
-#define READ_DEPTH( d, _x, _y )	\
-   d = *(GLushort *)(buf + (_x)*2 + (_y)*pitch);	 
+#define TAG(x)    intel##x##_ARGB8888
+#define TAG2(x,y) intel##x##_ARGB8888##y
+#define GET_PTR(X,Y) (buf + ((Y) * irb->pfPitch + (X)) * 4)
+#include "spantmp2.h"
 
 
-#define TAG(x) intel##x##_z16
-#include "depthtmp.h"
+#define LOCAL_DEPTH_VARS						\
+   struct intel_context *intel = intel_context(ctx);			\
+   struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
+   const GLuint pitch = irb->pfPitch/***XXX region->pitch*/; /* in pixels */ \
+   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
+   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1;	\
+   char *buf = (char *) irb->pfMap/*XXX use region->map*/ +             \
+      (intel->drawY * pitch + intel->drawX) * irb->region->cpp;
 
 
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);			\
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;		\
-   driRenderbuffer *drb = (driRenderbuffer *) rb;		\
-   GLuint pitch = drb->pitch;					\
-   GLuint height = dPriv->h;					\
-   char *buf = (char *)drb->Base.Data +				\
-			dPriv->x * drb->cpp +			\
-			dPriv->y * pitch;			\
-   GLuint p;							\
-   (void) buf; (void) p
-
-#undef INIT_MONO_PIXEL
-#define INIT_MONO_PIXEL(p,color)\
-	 p = INTEL_PACKCOLOR8888(color[0],color[1],color[2],color[3])
-
-/* 32 bit, 8888 argb color spanline and pixel functions
- */
-#define WRITE_RGBA(_x, _y, r, g, b, a)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = ((r << 16) |	\
-					  (g << 8)  |	\
-					  (b << 0)  |	\
-					  (a << 24) )
+#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS
 
-#define WRITE_PIXEL(_x, _y, p)			\
-    *(GLuint *)(buf + _x*4 + _y*pitch) = p
+/**
+ ** 16-bit depthbuffer functions.
+ **/
+#define WRITE_DEPTH( _x, _y, d ) \
+   ((GLushort *)buf)[(_x) + (_y) * pitch] = d;
 
+#define READ_DEPTH( d, _x, _y )	\
+   d = ((GLushort *)buf)[(_x) + (_y) * pitch];
 
-#define READ_RGBA(rgba, _x, _y)					\
-    do {							\
-	GLuint p = *(GLuint *)(buf + _x*4 + _y*pitch);		\
-	rgba[0] = (p >> 16) & 0xff;				\
-	rgba[1] = (p >> 8)  & 0xff;				\
-	rgba[2] = (p >> 0)  & 0xff;				\
-	rgba[3] = (p >> 24) & 0xff;				\
-    } while (0)
 
-#define TAG(x) intel##x##_8888
-#include "spantmp.h"
+#define TAG(x) intel##x##_z16
+#include "depthtmp.h"
 
 
-/* 24/8 bit interleaved depth/stencil functions
- */
-#define WRITE_DEPTH( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xff000000;					\
-   tmp |= (d) & 0xffffff;				\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
+/**
+ ** 24/8-bit interleaved depth/stencil functions
+ ** Note: we're actually reading back combined depth+stencil values.
+ ** The wrappers in main/depthstencil.c are used to extract the depth
+ ** and stencil values.
+ **/
+/* Change ZZZS -> SZZZ */
+#define WRITE_DEPTH( _x, _y, d ) {				\
+   GLuint tmp = ((d) >> 8) | ((d) << 24);			\
+   ((GLuint *)buf)[(_x) + (_y) * pitch] = tmp;			\
 }
 
-#define READ_DEPTH( d, _x, _y )		\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) & 0xffffff;
-
+/* Change SZZZ -> ZZZS */
+#define READ_DEPTH( d, _x, _y ) {				\
+   GLuint tmp = ((GLuint *)buf)[(_x) + (_y) * pitch];		\
+   d = (tmp << 8) | (tmp >> 24);				\
+}
 
 #define TAG(x) intel##x##_z24_s8
 #include "depthtmp.h"
 
-#define WRITE_STENCIL( _x, _y, d ) {			\
-   GLuint tmp = *(GLuint *)(buf + (_x)*4 + (_y)*pitch);	\
-   tmp &= 0xffffff;					\
-   tmp |= ((d)<<24);					\
-   *(GLuint *)(buf + (_x)*4 + (_y)*pitch) = tmp;		\
+
+/**
+ ** 8-bit stencil function (XXX FBO: This is obsolete)
+ **/
+#define WRITE_STENCIL( _x, _y, d ) {				\
+   GLuint tmp = ((GLuint *)buf)[(_x) + (_y) * pitch];		\
+   tmp &= 0xffffff;						\
+   tmp |= ((d) << 24);						\
+   ((GLuint *) buf)[(_x) + (_y) * pitch] = tmp;			\
 }
 
-#define READ_STENCIL( d, _x, _y )			\
-   d = *(GLuint *)(buf + (_x)*4 + (_y)*pitch) >> 24;
+#define READ_STENCIL( d, _x, _y )				\
+   d = ((GLuint *)buf)[(_x) + (_y) * pitch] >> 24;
 
 #define TAG(x) intel##x##_z24_s8
 #include "stenciltmp.h"
 
 
-/* Move locking out to get reasonable span performance.
+
+/**
+ * Map or unmap all the renderbuffers which we may need during
+ * software rendering.
+ * XXX in the future, we could probably convey extra information to
+ * reduce the number of mappings needed.  I.e. if doing a glReadPixels
+ * from the depth buffer, we really only need one mapping.
+ *
+ * XXX Rewrite this function someday.
+ * We can probably just loop over all the renderbuffer attachments,
+ * map/unmap all of them, and not worry about the _ColorDrawBuffers
+ * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields.
  */
-void intelSpanRenderStart( GLcontext *ctx )
+static void
+intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   GLcontext *ctx = &intel->ctx;
+   GLuint i, j;
+   struct intel_renderbuffer *irb;
+
+   /* color draw buffers */
+   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
+      for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers[i]; j++) {
+         struct gl_renderbuffer *rb =
+            ctx->DrawBuffer->_ColorDrawBuffers[i][j];
+         irb = intel_renderbuffer(rb);
+         if (irb) {
+            /* this is a user-created intel_renderbuffer */
+            if (irb->region) {
+               if (map)
+                  intel_region_map(intel->intelScreen, irb->region);
+               else
+                  intel_region_unmap(intel->intelScreen, irb->region);
+               irb->pfMap = irb->region->map;
+               irb->pfPitch = irb->region->pitch;
+            }
+         }
+      }
+   }
+
+   /* check for render to textures */
+   for (i = 0; i < BUFFER_COUNT; i++) {
+      struct gl_renderbuffer_attachment *att =
+         ctx->DrawBuffer->Attachment + i;
+      struct gl_texture_object *tex = att->Texture;
+      if (tex) {
+         /* render to texture */
+         ASSERT(att->Renderbuffer);
+         if (map) {
+            struct gl_texture_image *texImg;
+            texImg = tex->Image[att->CubeMapFace][att->TextureLevel];
+            intel_tex_map_images(intel, intel_texture_object(tex));
+         }
+         else {
+            intel_tex_unmap_images(intel, intel_texture_object(tex));
+         }
+      }
+   }
+
+   /* color read buffers */
+   irb = intel_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
+   if (irb && irb->region) {
+      if (map)
+         intel_region_map(intel->intelScreen, irb->region);
+      else
+         intel_region_unmap(intel->intelScreen, irb->region);
+      irb->pfMap = irb->region->map;
+      irb->pfPitch = irb->region->pitch;
+   }
 
-   intelFlush(&intel->ctx);
+   /* Account for front/back color page flipping.
+    * The span routines use the pfMap and pfPitch fields which will
+    * swap the front/back region map/pitch if we're page flipped.
+    * Do this after mapping, above, so the map field is valid.
+    */
+#if 0
+   if (map && ctx->DrawBuffer->Name == 0) {
+      struct intel_renderbuffer *irbFront
+         = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_FRONT_LEFT);
+      struct intel_renderbuffer *irbBack
+         = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_BACK_LEFT);
+      if (irbBack) {
+         /* double buffered */
+         if (intel->sarea->pf_current_page == 0) {
+            irbFront->pfMap = irbFront->region->map;
+            irbFront->pfPitch = irbFront->region->pitch;
+            irbBack->pfMap = irbBack->region->map;
+            irbBack->pfPitch = irbBack->region->pitch;
+         }
+         else {
+            irbFront->pfMap = irbBack->region->map;
+            irbFront->pfPitch = irbBack->region->pitch;
+            irbBack->pfMap = irbFront->region->map;
+            irbBack->pfPitch = irbFront->region->pitch;
+         }
+      }
+   }
+#endif
+
+   /* depth buffer (Note wrapper!) */
+   if (ctx->DrawBuffer->_DepthBuffer) {
+      irb = intel_renderbuffer(ctx->DrawBuffer->_DepthBuffer->Wrapped);
+      if (irb && irb->region && irb->Base.Name != 0) {
+         if (map) {
+            intel_region_map(intel->intelScreen, irb->region);
+            irb->pfMap = irb->region->map;
+            irb->pfPitch = irb->region->pitch;
+         }
+         else {
+            intel_region_unmap(intel->intelScreen, irb->region);
+            irb->pfMap = NULL;
+            irb->pfPitch = 0;
+         }
+      }
+   }
+
+   /* stencil buffer (Note wrapper!) */
+   if (ctx->DrawBuffer->_StencilBuffer) {
+      irb = intel_renderbuffer(ctx->DrawBuffer->_StencilBuffer->Wrapped);
+      if (irb && irb->region && irb->Base.Name != 0) {
+         if (map) {
+            intel_region_map(intel->intelScreen, irb->region);
+            irb->pfMap = irb->region->map;
+            irb->pfPitch = irb->region->pitch;
+         }
+         else {
+            intel_region_unmap(intel->intelScreen, irb->region);
+            irb->pfMap = NULL;
+            irb->pfPitch = 0;
+         }
+      }
+   }
+}
+
+
+
+/**
+ * Prepare for softare rendering.  Map current read/draw framebuffers'
+ * renderbuffes and all currently bound texture objects.
+ *
+ * Old note: Moved locking out to get reasonable span performance.
+ */
+void
+intelSpanRenderStart(GLcontext * ctx)
+{
+   struct intel_context *intel = intel_context(ctx);
+   GLuint i;
+
+   intelFinish(&intel->ctx);
    LOCK_HARDWARE(intel);
-   intelWaitForIdle(intel);
+
+#if 0
+   /* Just map the framebuffer and all textures.  Bufmgr code will
+    * take care of waiting on the necessary fences:
+    */
+   intel_region_map(intel->intelScreen, intel->front_region);
+   intel_region_map(intel->intelScreen, intel->back_region);
+   intel_region_map(intel->intelScreen, intel->intelScreen->depth_region);
+#endif
+
+   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; 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));
+      }
+   }
+
+   intel_map_unmap_buffers(intel, GL_TRUE);
 }
 
-void intelSpanRenderFinish( GLcontext *ctx )
+/**
+ * Called when done softare rendering.  Unmap the buffers we mapped in
+ * the above function.
+ */
+void
+intelSpanRenderFinish(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   _swrast_flush( ctx );
-   UNLOCK_HARDWARE( intel );
+   struct intel_context *intel = intel_context(ctx);
+   GLuint i;
+
+   _swrast_flush(ctx);
+
+   /* Now unmap the framebuffer:
+    */
+#if 0
+   intel_region_unmap(intel, intel->front_region);
+   intel_region_unmap(intel, intel->back_region);
+   intel_region_unmap(intel, intel->intelScreen->depth_region);
+#endif
+
+   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; 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));
+      }
+   }
+
+   intel_map_unmap_buffers(intel, GL_FALSE);
+
+   UNLOCK_HARDWARE(intel);
 }
 
-void intelInitSpanFuncs( GLcontext *ctx )
+
+void
+intelInitSpanFuncs(GLcontext * ctx)
 {
    struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
    swdd->SpanRenderStart = intelSpanRenderStart;
-   swdd->SpanRenderFinish = intelSpanRenderFinish; 
+   swdd->SpanRenderFinish = intelSpanRenderFinish;
 }
 
 
 /**
- * Plug in the Get/Put routines for the given driRenderbuffer.
+ * Plug in appropriate span read/write functions for the given renderbuffer.
+ * These are used for the software fallbacks.
  */
 void
-intelSetSpanFunctions(driRenderbuffer *drb, const GLvisual *vis)
+intel_set_span_functions(struct gl_renderbuffer *rb)
 {
-   if (drb->Base.InternalFormat == GL_RGBA) {
-      if (vis->redBits == 5 && vis->greenBits == 5 && vis->blueBits == 5) {
-         intelInitPointers_555(&drb->Base);
-      }
-      else if (vis->redBits == 5 && vis->greenBits == 6 && vis->blueBits == 5) {
-         intelInitPointers_565(&drb->Base);
-      }
-      else {
-         assert(vis->redBits == 8);
-         assert(vis->greenBits == 8);
-         assert(vis->blueBits == 8);
-         intelInitPointers_8888(&drb->Base);
-      }
+   if (rb->_ActualFormat == GL_RGB5) {
+      /* 565 RGB */
+      intelInitPointers_RGB565(rb);
+   }
+   else if (rb->_ActualFormat == GL_RGBA8) {
+      /* 8888 RGBA */
+      intelInitPointers_ARGB8888(rb);
+   }
+   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT16) {
+      intelInitDepthPointers_z16(rb);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT16) {
-      intelInitDepthPointers_z16(&drb->Base);
+   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24 ||        /* XXX FBO remove */
+            rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
+      intelInitDepthPointers_z24_s8(rb);
    }
-   else if (drb->Base.InternalFormat == GL_DEPTH_COMPONENT24) {
-      intelInitDepthPointers_z24_s8(&drb->Base);
+   else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) {       /* XXX FBO remove */
+      intelInitStencilPointers_z24_s8(rb);
    }
-   else if (drb->Base.InternalFormat == GL_STENCIL_INDEX8_EXT) {
-      intelInitStencilPointers_z24_s8(&drb->Base);
+   else {
+      _mesa_problem(NULL,
+                    "Unexpected _ActualFormat in intelSetSpanFunctions");
    }
 }
diff --git a/src/mesa/drivers/dri/i915/intel_span.h b/src/mesa/drivers/dri/i915/intel_span.h
index 2d4f8589d0f..5201f6d6c6e 100644
--- a/src/mesa/drivers/dri/i915/intel_span.h
+++ b/src/mesa/drivers/dri/i915/intel_span.h
@@ -28,14 +28,11 @@
 #ifndef _INTEL_SPAN_H
 #define _INTEL_SPAN_H
 
-#include "drirenderbuffer.h"
+extern void intelInitSpanFuncs(GLcontext * ctx);
 
-extern void intelInitSpanFuncs( GLcontext *ctx );
+extern void intelSpanRenderFinish(GLcontext * ctx);
+extern void intelSpanRenderStart(GLcontext * ctx);
 
-extern void intelSpanRenderFinish( GLcontext *ctx );
-extern void intelSpanRenderStart( GLcontext *ctx );
-
-extern void
-intelSetSpanFunctions(driRenderbuffer *rb, const GLvisual *vis);
+extern void intel_set_span_functions(struct gl_renderbuffer *rb);
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/intel_state.c b/src/mesa/drivers/dri/i915/intel_state.c
index e5988a5ed6c..271511037e9 100644
--- a/src/mesa/drivers/dri/i915/intel_state.c
+++ b/src/mesa/drivers/dri/i915/intel_state.c
@@ -30,252 +30,240 @@
 #include "context.h"
 #include "macros.h"
 #include "enums.h"
+#include "colormac.h"
 #include "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_compare_func( GLenum func )
+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; 
+   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; 
+   return COMPAREFUNC_ALWAYS;
 }
 
-int intel_translate_stencil_op( GLenum op )
+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: 
+   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: 
+   case GL_DECR:
       return STENCILOP_DECRSAT;
    case GL_INCR_WRAP:
-      return STENCILOP_INCR; 
+      return STENCILOP_INCR;
    case GL_DECR_WRAP:
-      return STENCILOP_DECR; 
-   case GL_INVERT: 
-      return STENCILOP_INVERT; 
-   default: 
+      return STENCILOP_DECR;
+   case GL_INVERT:
+      return STENCILOP_INVERT;
+   default:
       return STENCILOP_ZERO;
    }
 }
 
-int intel_translate_blend_factor( GLenum factor )
+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; 
+   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; 
+      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_INV_DST_ALPHA;
+   case GL_SRC_ALPHA_SATURATE:
       return BLENDFACT_SRC_ALPHA_SATURATE;
    case GL_CONSTANT_COLOR:
-      return BLENDFACT_CONST_COLOR; 
+      return BLENDFACT_CONST_COLOR;
    case GL_ONE_MINUS_CONSTANT_COLOR:
       return BLENDFACT_INV_CONST_COLOR;
    case GL_CONSTANT_ALPHA:
-      return BLENDFACT_CONST_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 )
+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:
+   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;
-   }
-}
-
-static void intelDrawBuffer(GLcontext *ctx, GLenum mode )
-{
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   int front = 0;
- 
-   if (!ctx->DrawBuffer)
-      return;
-
-   switch ( ctx->DrawBuffer->_ColorDrawBufferMask[0] ) {
-   case BUFFER_BIT_FRONT_LEFT:
-      front = 1;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
-   case BUFFER_BIT_BACK_LEFT:
-      front = 0;
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_FALSE );
-      break;
    default:
-      FALLBACK( intel, INTEL_FALLBACK_DRAW_BUFFER, GL_TRUE );
-      return;
-   }
-
-   if ( intel->sarea->pf_current_page == 1 ) 
-      front ^= 1;
-   
-   intelSetFrontClipRects( intel );
-
-   if (front) {
-      intel->drawRegion = &intel->intelScreen->front;
-      intel->readRegion = &intel->intelScreen->front;
-   } else {
-      intel->drawRegion = &intel->intelScreen->back;
-      intel->readRegion = &intel->intelScreen->back;
+      return LOGICOP_SET;
    }
-
-   intel->vtbl.set_color_region( intel, intel->drawRegion );
-}
-
-static void intelReadBuffer( GLcontext *ctx, GLenum mode )
-{
-   /* nothing, until we implement h/w glRead/CopyPixels or CopyTexImage */
 }
 
 
-static void intelClearColor(GLcontext *ctx, const GLfloat color[4])
+static void
+intelClearColor(GLcontext * ctx, const GLfloat color[4])
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   intelScreenPrivate *screen = intel->intelScreen;
+   struct intel_context *intel = intel_context(ctx);
+   GLubyte clear[4];
 
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_red, color[0]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_green, color[1]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_blue, color[2]);
-   CLAMPED_FLOAT_TO_UBYTE(intel->clear_alpha, color[3]);
+   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]);
 
-   intel->ClearColor = INTEL_PACKCOLOR(screen->fbFormat,
-				       intel->clear_red, 
-				       intel->clear_green, 
-				       intel->clear_blue, 
-				       intel->clear_alpha);
+   /* 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]);
 }
 
 
-static void intelCalcViewport( GLcontext *ctx )
+/**
+ * Update the viewport transformation matrix.  Depends on:
+ *  - viewport pos/size
+ *  - depthrange
+ *  - window pos/size or FBO size
+ */
+static void
+intelCalcViewport(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(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;
-   GLint h = 0;
+   GLfloat yScale, yBias;
 
-   if (intel->driDrawable) 
-      h = intel->driDrawable->h + SUBPIXEL_Y;
+   if (ctx->DrawBuffer->Name) {
+      /* User created FBO */
+      struct intel_renderbuffer *irb
+         = intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][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] + SUBPIXEL_X;
 
-   /* See also intel_translate_vertex.  SUBPIXEL adjustments can be done
-    * via state vars, too.
-    */
-   m[MAT_SX] =   v[MAT_SX];
-   m[MAT_TX] =   v[MAT_TX] + SUBPIXEL_X;
-   m[MAT_SY] = - v[MAT_SY];
-   m[MAT_TY] = - v[MAT_TY] + h;
-   m[MAT_SZ] =   v[MAT_SZ] * intel->depth_scale;
-   m[MAT_TZ] =   v[MAT_TZ] * intel->depth_scale;
+   m[MAT_SY] = v[MAT_SY] * yScale;
+   m[MAT_TY] = v[MAT_TY] * yScale + yBias + SUBPIXEL_Y;
+
+   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 )
+static void
+intelViewport(GLcontext * ctx,
+              GLint x, GLint y, GLsizei width, GLsizei height)
 {
-   intelCalcViewport( ctx );
+   intelCalcViewport(ctx);
 }
 
-static void intelDepthRange( GLcontext *ctx,
-			    GLclampd nearval, GLclampd farval )
+static void
+intelDepthRange(GLcontext * ctx, GLclampd nearval, GLclampd farval)
 {
-   intelCalcViewport( ctx );
+   intelCalcViewport(ctx);
 }
 
 /* Fallback to swrast for select and feedback.
  */
-static void intelRenderMode( GLcontext *ctx, GLenum mode )
+static void
+intelRenderMode(GLcontext * ctx, GLenum mode)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
-   FALLBACK( intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER) );
+   struct intel_context *intel = intel_context(ctx);
+   FALLBACK(intel, INTEL_FALLBACK_RENDERMODE, (mode != GL_RENDER));
 }
 
 
-void intelInitStateFuncs( struct dd_function_table *functions )
+void
+intelInitStateFuncs(struct dd_function_table *functions)
 {
-   functions->DrawBuffer = intelDrawBuffer;
-   functions->ReadBuffer = intelReadBuffer;
    functions->RenderMode = intelRenderMode;
    functions->Viewport = intelViewport;
    functions->DepthRange = intelDepthRange;
    functions->ClearColor = intelClearColor;
 }
-
diff --git a/src/mesa/drivers/dri/i915tex/intel_structs.h b/src/mesa/drivers/dri/i915/intel_structs.h
index 522e3bd92c2..522e3bd92c2 100644
--- a/src/mesa/drivers/dri/i915tex/intel_structs.h
+++ b/src/mesa/drivers/dri/i915/intel_structs.h
diff --git a/src/mesa/drivers/dri/i915/intel_tex.c b/src/mesa/drivers/dri/i915/intel_tex.c
index 5bd280652af..b08dee43bc1 100644
--- a/src/mesa/drivers/dri/i915/intel_tex.c
+++ b/src/mesa/drivers/dri/i915/intel_tex.c
@@ -1,877 +1,192 @@
-/**************************************************************************
- * 
- * 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 "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "teximage.h"
-#include "texmem.h"
 #include "texobj.h"
-#include "swrast/swrast.h"
-
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
 #include "intel_context.h"
+#include "intel_mipmap_tree.h"
 #include "intel_tex.h"
-#include "intel_ioctl.h"
-
 
+#define FILE_DEBUG_FLAG DEBUG_TEXTURE
 
 static GLboolean
-intelValidateClientStorage( intelContextPtr intel, GLenum target,
-			    GLint internalFormat,
-			    GLint srcWidth, GLint srcHeight, 
-			    GLenum format, GLenum type,  const void *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage)
-
+intelIsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj)
 {
-   GLcontext *ctx = &intel->ctx;
-   int texelBytes;
-
-   if (0)
-      fprintf(stderr, "intformat %s format %s type %s\n",
-	      _mesa_lookup_enum_by_nr( internalFormat ),
-	      _mesa_lookup_enum_by_nr( format ),
-	      _mesa_lookup_enum_by_nr( type ));
-
-   if (!ctx->Unpack.ClientStorage)
-      return 0;
-
-   if (ctx->_ImageTransferState ||
-       texImage->IsCompressed ||
-       texObj->GenerateMipmap)
-      return 0;
-
-
-   /* This list is incomplete
-    */
-   switch ( internalFormat ) {
-   case GL_RGBA:
-      if ( format == GL_BGRA && type == GL_UNSIGNED_INT_8_8_8_8_REV ) {
-	 texImage->TexFormat = &_mesa_texformat_argb8888;
-	 texelBytes = 4;
-      }
-      else
-	 return 0;
-      break;
-
-   case GL_RGB:
-      if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
-	 texImage->TexFormat = &_mesa_texformat_rgb565;
-	 texelBytes = 2;
-      }
-      else
-	 return 0;
-      break;
-
-   case GL_YCBCR_MESA:
-      if ( format == GL_YCBCR_MESA && 
-	   type == GL_UNSIGNED_SHORT_8_8_REV_APPLE ) {
-	 texImage->TexFormat = &_mesa_texformat_ycbcr_rev;
-	 texelBytes = 2;
-      }
-      else if ( format == GL_YCBCR_MESA && 
-		(type == GL_UNSIGNED_SHORT_8_8_APPLE || 
-		 type == GL_UNSIGNED_BYTE)) {
-	 texImage->TexFormat = &_mesa_texformat_ycbcr;
-	 texelBytes = 2;
-      }
-      else
-	 return 0;
-      break;
-      
-	 
-   default:
-      return 0;
-   }
+#if 0
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
 
-   /* Could deal with these packing issues, but currently don't:
-    */
-   if (packing->SkipPixels || 
-       packing->SkipRows || 
-       packing->SwapBytes ||
-       packing->LsbFirst) {
-      return 0;
-   }
-
-   {      
-      GLint srcRowStride = _mesa_image_row_stride(packing, srcWidth,
-						  format, type);
-
-      
-      if (0)
-	 fprintf(stderr, "%s: srcRowStride %d/%x\n", 
-		 __FUNCTION__, srcRowStride, srcRowStride);
-
-      /* Could check this later in upload, pitch restrictions could be
-       * relaxed, but would need to store the image pitch somewhere,
-       * as packing details might change before image is uploaded:
-       */
-      if (!intelIsAgpMemory( intel, pixels, srcHeight * srcRowStride ) ||
-	  (srcRowStride & 63))
-	 return 0;
-
-
-      /* Have validated that _mesa_transfer_teximage would be a straight
-       * memcpy at this point.  NOTE: future calls to TexSubImage will
-       * overwrite the client data.  This is explicitly mentioned in the
-       * extension spec.
-       */
-      texImage->Data = (void *)pixels;
-      texImage->IsClientData = GL_TRUE;
-      texImage->RowStride = srcRowStride / texelBytes;
-      return 1;
-   }
+   return
+      intelObj->mt &&
+      intelObj->mt->region &&
+      intel_is_region_resident(intel, intelObj->mt->region);
+#endif
+   return 1;
 }
 
- 
-
-static void intelTexImage1D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
 
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
 
-   texImage->IsClientData = GL_FALSE;
-
-   _mesa_store_teximage1d( ctx, target, level, internalFormat,
-			   width, border, format, type,
-			   pixels, packing, texObj, texImage );
-
-   t->dirty_images[0] |= (1 << level);
+static struct gl_texture_image *
+intelNewTextureImage(GLcontext * ctx)
+{
+   DBG("%s\n", __FUNCTION__);
+   (void) ctx;
+   return (struct gl_texture_image *) CALLOC_STRUCT(intel_texture_image);
 }
 
-static void intelTexSubImage1D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset,
-				GLsizei width,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
+
+static struct gl_texture_object *
+intelNewTextureObject(GLcontext * ctx, GLuint name, GLenum target)
 {
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
+   struct intel_texture_object *obj = CALLOC_STRUCT(intel_texture_object);
 
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
+   DBG("%s\n", __FUNCTION__);
+   _mesa_initialize_texture_object(&obj->base, name, target);
 
-   _mesa_store_texsubimage1d(ctx, target, level, xoffset, width, 
-			     format, type, pixels, packing, texObj,
-			     texImage);
+   return &obj->base;
 }
 
-
-/* Handles 2D, CUBE, RECT:
- */
-static void intelTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-			    GLint internalFormat,
-			    GLint width, GLint height, GLint border,
-			    GLenum format, GLenum type, const GLvoid *pixels,
-			    const struct gl_pixelstore_attrib *packing,
-			    struct gl_texture_object *texObj,
-			    struct gl_texture_image *texImage )
+static void 
+intelDeleteTextureObject(GLcontext *ctx,
+			 struct gl_texture_object *texObj)
 {
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_object *intelObj = intel_texture_object(texObj);
 
-   assert(t);
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   if (intelValidateClientStorage( INTEL_CONTEXT(ctx), target, 
-				   internalFormat, 
-				   width, height, 
-				   format, type, pixels, 
-				   packing, texObj, texImage)) {
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "%s: Using client storage\n", __FUNCTION__); 
-   }
-   else {
-      _mesa_store_teximage2d( ctx, target, level, internalFormat,
-			      width, height, border, format, type,
-			      pixels, packing, texObj, texImage );
+   if (intelObj->mt)
+      intel_miptree_release(intel, &intelObj->mt);
 
-      t->dirty_images[face] |= (1 << level);
-   }
+   _mesa_delete_texture_object(ctx, texObj);
 }
 
-static void intelTexSubImage2D( GLcontext *ctx, 
-			       GLenum target,
-			       GLint level,	
-			       GLint xoffset, GLint yoffset,
-			       GLsizei width, GLsizei height,
-			       GLenum format, GLenum type,
-			       const GLvoid *pixels,
-			       const struct gl_pixelstore_attrib *packing,
-			       struct gl_texture_object *texObj,
-			       struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
-   }
-
-   if (texImage->IsClientData &&
-       (char *)pixels == (char *)texImage->Data + 
-       ((xoffset + yoffset * texImage->RowStride) * 
-	texImage->TexFormat->TexelBytes)) {
 
-      /* Notification only - no upload required */
-   }
-   else {
-      assert( t ); /* this _should_ be true */
-      intelFlush( ctx );
-      driSwapOutTextureObject( t );
+static void
+intelFreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_texture_image *intelImage = intel_texture_image(texImage);
 
-      _mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width, 
-				height, format, type, pixels, packing, texObj,
-				texImage);
+   DBG("%s\n", __FUNCTION__);
 
-      t->dirty_images[face] |= (1 << level);
+   if (intelImage->mt) {
+      intel_miptree_release(intel, &intelImage->mt);
    }
-}
 
-static void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-                              GLint internalFormat,
-                              GLint width, GLint height, GLint border,
-                              GLsizei imageSize, const GLvoid *data,
-                              struct gl_texture_object *texObj,
-                              struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
+   if (texImage->Data) {
+      free(texImage->Data);
+      texImage->Data = NULL;
    }
-
-   assert(t);
-   intelFlush( ctx );
-   
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-     fprintf(stderr, "%s: Using normal storage\n", __FUNCTION__); 
-   
-   _mesa_store_compressed_teximage2d(ctx, target, level, internalFormat, width,
-				     height, border, imageSize, data, texObj, texImage);
-   
-   t->dirty_images[face] |= (1 << level);
 }
 
 
-static void intelCompressedTexSubImage2D( GLcontext *ctx, GLenum target, GLint level,
-                                 GLint xoffset, GLint yoffset,
-                                 GLsizei width, GLsizei height,
-                                 GLenum format,
-                                 GLsizei imageSize, const GLvoid *data,
-                                 struct gl_texture_object *texObj,
-                                 struct gl_texture_image *texImage )
+/* The system memcpy (at least on ubuntu 5.10) has problems copying
+ * to agp (writecombined) memory from a source which isn't 64-byte
+ * aligned - there is a 4x performance falloff.
+ *
+ * The x86 __memcpy is immune to this but is slightly slower
+ * (10%-ish) than the system memcpy.
+ *
+ * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
+ * isn't much faster than x86_memcpy for agp copies.
+ * 
+ * TODO: switch dynamically.
+ */
+static void *
+do_memcpy(void *dest, const void *src, size_t n)
 {
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-   GLuint face;
-
-
-   /* which cube face or ordinary 2D image */
-   switch (target) {
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_X:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_X:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Y:
-   case GL_TEXTURE_CUBE_MAP_POSITIVE_Z:
-   case GL_TEXTURE_CUBE_MAP_NEGATIVE_Z:
-      face = (GLuint) target - (GLuint) GL_TEXTURE_CUBE_MAP_POSITIVE_X;
-      ASSERT(face < 6);
-      break;
-   default:
-      face = 0;
+   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63)) {
+      return __memcpy(dest, src, n);
    }
-
-   assert( t ); /* this _should_ be true */
-   intelFlush( ctx );
-   driSwapOutTextureObject( t );
-   
-   _mesa_store_compressed_texsubimage2d(ctx, target, level, xoffset, yoffset, width,
-					height, format, imageSize, data, texObj, texImage);
-   
-   t->dirty_images[face] |= (1 << level);
+   else
+      return memcpy(dest, src, n);
 }
 
 
-static void intelTexImage3D( GLcontext *ctx, GLenum target, GLint level,
-                            GLint internalFormat,
-                            GLint width, GLint height, GLint depth,
-                            GLint border,
-                            GLenum format, GLenum type, const GLvoid *pixels,
-                            const struct gl_pixelstore_attrib *packing,
-                            struct gl_texture_object *texObj,
-                            struct gl_texture_image *texImage )
-{
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert(t);
-   driSwapOutTextureObject( t );
-   texImage->IsClientData = GL_FALSE;
-
-   _mesa_store_teximage3d(ctx, target, level, internalFormat,
-			  width, height, depth, border,
-			  format, type, pixels,
-			  &ctx->Unpack, texObj, texImage);
-   
-   t->dirty_images[0] |= (1 << level);
-}
-
+#if DO_DEBUG
 
-static void
-intelTexSubImage3D( GLcontext *ctx, GLenum target, GLint level,
-                   GLint xoffset, GLint yoffset, GLint zoffset,
-                   GLsizei width, GLsizei height, GLsizei depth,
-                   GLenum format, GLenum type,
-                   const GLvoid *pixels,
-                   const struct gl_pixelstore_attrib *packing,
-                   struct gl_texture_object *texObj,
-                   struct gl_texture_image *texImage )
+#ifndef __x86_64__
+static unsigned
+fastrdtsc(void)
 {
-   driTextureObject * t = (driTextureObject *) texObj->DriverData;
-
-   assert( t ); /* this _should_ be true */
-   driSwapOutTextureObject( t );
+   unsigned eax;
+   __asm__ volatile ("\t"
+                     "pushl  %%ebx\n\t"
+                     "cpuid\n\t" ".byte 0x0f, 0x31\n\t"
+                     "popl %%ebx\n":"=a" (eax)
+                     :"0"(0)
+                     :"ecx", "edx", "cc");
 
-   _mesa_store_texsubimage3d(ctx, target, level, xoffset, yoffset, zoffset,
-                             width, height, depth,
-                             format, type, pixels, packing, texObj, texImage);
-
-   t->dirty_images[0] |= (1 << level);
+   return eax;
 }
-
-
-
-
-static void intelDeleteTexture( GLcontext *ctx, struct gl_texture_object *tObj )
+#else
+static unsigned
+fastrdtsc(void)
 {
-   driTextureObject * t = (driTextureObject *) tObj->DriverData;
+   unsigned eax;
+   __asm__ volatile ("\t" "cpuid\n\t" ".byte 0x0f, 0x31\n\t":"=a" (eax)
+                     :"0"(0)
+                     :"ecx", "edx", "ebx", "cc");
 
-   if ( t != NULL ) {
-      intelFlush( ctx );
-      driDestroyTextureObject( t );
-   }
-   
-   /* Free mipmap images and the texture object itself */
-   _mesa_delete_texture_object(ctx, tObj);
+   return eax;
 }
+#endif
 
-
-static const struct gl_texture_format *
-intelChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
-			 GLenum format, GLenum type )
+static unsigned
+time_diff(unsigned t, unsigned t2)
 {
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
-   const GLboolean do32bpt = ( intel->intelScreen->cpp == 4 &&
-			       intel->intelScreen->tex.size > 4*1024*1024);
-
-   switch ( internalFormat ) {
-   case 4:
-   case GL_RGBA:
-   case GL_COMPRESSED_RGBA:
-      if ( format == GL_BGRA ) {
-	 if ( type == GL_UNSIGNED_INT_8_8_8_8_REV ) {
-	    return &_mesa_texformat_argb8888;
-	 }
-         else if ( type == GL_UNSIGNED_SHORT_4_4_4_4_REV ) {
-            return &_mesa_texformat_argb4444;
-	 }
-         else if ( type == GL_UNSIGNED_SHORT_1_5_5_5_REV ) {
-	    return &_mesa_texformat_argb1555;
-	 }
-      }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
-
-   case 3:
-   case GL_RGB:
-   case GL_COMPRESSED_RGB:
-      if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) {
-	 return &_mesa_texformat_rgb565;
-      }
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
-
-   case GL_RGBA8:
-   case GL_RGB10_A2:
-   case GL_RGBA12:
-   case GL_RGBA16:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_argb4444;
-
-   case GL_RGBA4:
-   case GL_RGBA2:
-      return &_mesa_texformat_argb4444;
-
-   case GL_RGB5_A1:
-      return &_mesa_texformat_argb1555;
-
-   case GL_RGB8:
-   case GL_RGB10:
-   case GL_RGB12:
-   case GL_RGB16:
-      return do32bpt ? &_mesa_texformat_argb8888 : &_mesa_texformat_rgb565;
-
-   case GL_RGB5:
-   case GL_RGB4:
-   case GL_R3_G3_B2:
-      return &_mesa_texformat_rgb565;
-
-   case GL_ALPHA:
-   case GL_ALPHA4:
-   case GL_ALPHA8:
-   case GL_ALPHA12:
-   case GL_ALPHA16:
-   case GL_COMPRESSED_ALPHA:
-      return &_mesa_texformat_a8;
-
-   case 1:
-   case GL_LUMINANCE:
-   case GL_LUMINANCE4:
-   case GL_LUMINANCE8:
-   case GL_LUMINANCE12:
-   case GL_LUMINANCE16:
-   case GL_COMPRESSED_LUMINANCE:
-      return &_mesa_texformat_l8;
-
-   case 2:
-   case GL_LUMINANCE_ALPHA:
-   case GL_LUMINANCE4_ALPHA4:
-   case GL_LUMINANCE6_ALPHA2:
-   case GL_LUMINANCE8_ALPHA8:
-   case GL_LUMINANCE12_ALPHA4:
-   case GL_LUMINANCE12_ALPHA12:
-   case GL_LUMINANCE16_ALPHA16:
-   case GL_COMPRESSED_LUMINANCE_ALPHA:
-      return &_mesa_texformat_al88;
-
-   case GL_INTENSITY:
-   case GL_INTENSITY4:
-   case GL_INTENSITY8:
-   case GL_INTENSITY12:
-   case GL_INTENSITY16:
-   case GL_COMPRESSED_INTENSITY:
-      return &_mesa_texformat_i8;
-
-   case GL_YCBCR_MESA:
-      if (type == GL_UNSIGNED_SHORT_8_8_MESA ||
-	  type == GL_UNSIGNED_BYTE)
-         return &_mesa_texformat_ycbcr;
-      else
-         return &_mesa_texformat_ycbcr_rev;
-
-   case GL_COMPRESSED_RGB_FXT1_3DFX:
-     return &_mesa_texformat_rgb_fxt1;
-   case GL_COMPRESSED_RGBA_FXT1_3DFX:
-     return &_mesa_texformat_rgba_fxt1;
-
-   case GL_RGB_S3TC:
-   case GL_RGB4_S3TC:
-   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgb_dxt1;
-
-   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgba_dxt1;
-
-   case GL_RGBA_S3TC:
-   case GL_RGBA4_S3TC:
-   case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-     return &_mesa_texformat_rgba_dxt3;
-
-   case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-      return &_mesa_texformat_rgba_dxt5;
-
-   case GL_DEPTH_COMPONENT:
-   case GL_DEPTH_COMPONENT16:
-   case GL_DEPTH_COMPONENT24:
-   case GL_DEPTH_COMPONENT32:
-      return &_mesa_texformat_z16;
-
-   default:
-      fprintf(stderr, "unexpected texture format %s in %s\n", 
-	      _mesa_lookup_enum_by_nr(internalFormat),
-	      __FUNCTION__);
-      return NULL;
-   }
-
-   return NULL; /* never get here */
+   return ((t < t2) ? t2 - t : 0xFFFFFFFFU - (t - t2 - 1));
 }
 
 
-
-void intelDestroyTexObj(intelContextPtr intel, intelTextureObjectPtr t)
+static void *
+timed_memcpy(void *dest, const void *src, size_t n)
 {
-   unsigned   i;
+   void *ret;
+   unsigned t1, t2;
+   double rate;
 
-   if ( intel == NULL ) 
-      return;
+   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63))
+      _mesa_printf("Warning - non-aligned texture copy!\n");
 
-   if ( t->age > intel->dirtyAge )
-      intel->dirtyAge = t->age;
+   t1 = fastrdtsc();
+   ret = do_memcpy(dest, src, n);
+   t2 = fastrdtsc();
 
-   for ( i = 0 ; i < MAX_TEXTURE_UNITS ; i++ ) {
-      if ( t == intel->CurrentTexObj[ i ] ) 
-	 intel->CurrentTexObj[ i ] = NULL;
-   }
+   rate = time_diff(t1, t2);
+   rate /= (double) n;
+   _mesa_printf("timed_memcpy: %u %u --> %f clocks/byte\n", t1, t2, rate);
+   return ret;
 }
+#endif /* DO_DEBUG */
 
 
-
-/* Upload an image from mesa's internal copy.  Image may be 1D, 2D or
- * 3D.  Cubemaps are expanded elsewhere.
- */
-static void intelUploadTexImage( intelContextPtr intel,
-				 intelTextureObjectPtr t,
-				 const struct gl_texture_image *image,
-				 const GLuint offset )
+void
+intelInitTextureFuncs(struct dd_function_table *functions)
 {
+   functions->ChooseTextureFormat = intelChooseTextureFormat;
+   functions->TexImage1D = intelTexImage1D;
+   functions->TexImage2D = intelTexImage2D;
+   functions->TexImage3D = intelTexImage3D;
+   functions->TexSubImage1D = intelTexSubImage1D;
+   functions->TexSubImage2D = intelTexSubImage2D;
+   functions->TexSubImage3D = intelTexSubImage3D;
+   functions->CopyTexImage1D = intelCopyTexImage1D;
+   functions->CopyTexImage2D = intelCopyTexImage2D;
+   functions->CopyTexSubImage1D = intelCopyTexSubImage1D;
+   functions->CopyTexSubImage2D = intelCopyTexSubImage2D;
+   functions->GetTexImage = intelGetTexImage;
 
-   if (!image || !image->Data) 
-      return;
-
-   if (image->Depth == 1 && image->IsClientData) {
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "Blit uploading\n");
-
-      /* Do it with a blit.
-       */
-      intelEmitCopyBlitLocked( intel,
-			       image->TexFormat->TexelBytes,
-			       image->RowStride, /* ? */
-			       intelGetMemoryOffsetMESA( NULL, 0, image->Data ),
-			       t->Pitch / image->TexFormat->TexelBytes,
-			       intelGetMemoryOffsetMESA( NULL, 0, t->BufAddr + offset ),
-			       0, 0,
-			       0, 0,
-			       image->Width,
-			       image->Height);
-   }
-   else if (image->IsCompressed) {
-      GLuint row_len = 0;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint j;
-
-      /* must always copy whole blocks (8/16 bytes) */
-      switch (image->InternalFormat) {
-	case GL_COMPRESSED_RGB_FXT1_3DFX:
-	case GL_COMPRESSED_RGBA_FXT1_3DFX:
-	case GL_RGB_S3TC:
-	case GL_RGB4_S3TC:
-	case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-	case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-	  row_len = (image->Width * 2 + 7) & ~7;
-	  break;
-	case GL_RGBA_S3TC:
-	case GL_RGBA4_S3TC:
-	case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
-	case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-	  row_len = (image->Width * 4 + 15) & ~15;
-	  break;
-	default:
-	  fprintf(stderr,"Internal Compressed format not supported %d\n", image->InternalFormat);
-	  break;
-      }
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len) {
-	 for (j = 0 ; j < (image->Height + 3)/4 ; j++, dst += (t->Pitch)) {
-	   __memcpy(dst, src, row_len );
-	   src += row_len;
-	 }
-      }
-   }
-   /* Time for another vtbl entry:
-    */
-   else if (intel->intelScreen->deviceID == PCI_CHIP_I945_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_I945_GM ||
-            intel->intelScreen->deviceID == PCI_CHIP_I945_GME ||
-            intel->intelScreen->deviceID == PCI_CHIP_G33_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_Q33_G ||
-            intel->intelScreen->deviceID == PCI_CHIP_Q35_G) {
-      GLuint row_len = image->Width * image->TexFormat->TexelBytes;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint d, j;
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len == t->Pitch) {
-	 memcpy( dst, src, row_len * image->Height * image->Depth );
-      }
-      else { 
-	 GLuint x = 0, y = 0;
-
-	 for (d = 0 ; d < image->Depth ; d++) {
-	    GLubyte *dst0 = dst + x + y * t->Pitch;
-
-	    for (j = 0 ; j < image->Height ; j++) {
-	       __memcpy(dst0, src, row_len );
-	       src += row_len;
-	       dst0 += t->Pitch;
-	    }
-
-	    x += MIN2(4, row_len); /* Guess: 4 byte minimum alignment */
-	    if (x > t->Pitch) {
-	       x = 0;
-	       y += image->Height;
-	    }
-	 }
-      }
-
-   }
-   else {
-      GLuint row_len = image->Width * image->TexFormat->TexelBytes;
-      GLubyte *dst = (GLubyte *)(t->BufAddr + offset);
-      GLubyte *src = (GLubyte *)image->Data;
-      GLuint d, j;
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, 
-		 "Upload image %dx%dx%d offset %xm row_len %x "
-		 "pitch %x depth_pitch %x\n",
-		 image->Width, image->Height, image->Depth, offset,
-		 row_len, t->Pitch, t->depth_pitch);
-
-      if (row_len == t->Pitch) {
-	 for (d = 0; d < image->Depth; d++) {
-	    memcpy( dst, src, t->Pitch * image->Height );
-	    dst += t->depth_pitch;
-	    src += row_len * image->Height;
-	 }
-      }
-      else { 
-	 for (d = 0 ; d < image->Depth ; d++) {
-	    for (j = 0 ; j < image->Height ; j++) {
-	       __memcpy(dst, src, row_len );
-	       src += row_len;
-	       dst += t->Pitch;
-	    }
-
-	    dst += t->depth_pitch - (t->Pitch * image->Height);
-	 }
-      }
-   }
-}
-
-
-
-int intelUploadTexImages( intelContextPtr intel, 
-			  intelTextureObjectPtr t,
-			  GLuint face)
-{
-   const int numLevels = t->base.lastLevel - t->base.firstLevel + 1;
-   const struct gl_texture_image *firstImage = t->image[face][t->base.firstLevel].image;
-   int pitch = firstImage->RowStride * firstImage->TexFormat->TexelBytes;
-
-   /* Can we texture out of the existing client data? */
-   if ( numLevels == 1 &&
-	firstImage->IsClientData &&
-	(pitch & 3) == 0) {
-
-      if (INTEL_DEBUG & DEBUG_TEXTURE)
-	 fprintf(stderr, "AGP texturing from client memory\n");
-
-      t->TextureOffset = intelAgpOffsetFromVirtual( intel, firstImage->Data );
-      t->BufAddr = 0;
-      t->dirty = ~0;
-      return GL_TRUE;
-   }
-   else {
-      if (INTEL_DEBUG & DEBUG_TEXTURE) 
-	 fprintf(stderr, "Uploading client data to agp\n");
-
-      INTEL_FIREVERTICES( intel );
-      LOCK_HARDWARE( intel );
-
-      if ( t->base.memBlock == NULL ) {
-	 int heap;
-
-	 heap = driAllocateTexture( intel->texture_heaps, intel->nr_heaps,
-				    (driTextureObject *) t );
-	 if ( heap == -1 ) {
-	    UNLOCK_HARDWARE( intel );
-	    return GL_FALSE;
-	 }
-
-	 /* Set the base offset of the texture image */
-	 t->BufAddr = (GLubyte *) (intel->intelScreen->tex.map + 
-				   t->base.memBlock->ofs);
-	 t->TextureOffset = intel->intelScreen->tex.offset + t->base.memBlock->ofs;
-	 t->dirty = ~0;
-      }
-
-
-      /* Let the world know we've used this memory recently.
-       */
-      driUpdateTextureLRU( (driTextureObject *) t );
-
-
-      /* Upload any images that are new */
-      if (t->base.dirty_images[face]) {
-	 int i;
-
- 	 intelWaitForIdle( intel );
-	    
-	 for (i = 0 ; i < numLevels ; i++) { 
-	    int level = i + t->base.firstLevel;
-
-	    if (t->base.dirty_images[face] & (1<<level)) {
-
-	       const struct gl_texture_image *image = t->image[face][i].image;
-	       GLuint offset = t->image[face][i].offset;
-
-     	       if (INTEL_DEBUG & DEBUG_TEXTURE)
-	          fprintf(stderr, "upload level %d, offset %x\n", 
-			  level, offset);
-
-	       intelUploadTexImage( intel, t, image, offset );
-	    }
-	 }
-	 t->base.dirty_images[face] = 0;
-	 intel->perf_boxes |= I830_BOX_TEXTURE_LOAD;
-      }
-      
-      UNLOCK_HARDWARE( intel );
-      return GL_TRUE;
-   }
-}
+   /* compressed texture functions */
+   functions->CompressedTexImage2D = intelCompressedTexImage2D;
+   functions->GetCompressedTexImage = intelGetCompressedTexImage;
 
-/**
- * Allocate a new texture object.
- * Called via ctx->Driver.NewTextureObject.
- * Note: this function will be called during context creation to
- * allocate the default texture objects.
- * Note: we could use containment here to 'derive' the driver-specific
- * texture object from the core mesa gl_texture_object.  Not done at this time.
- */
-static struct gl_texture_object *
-intelNewTextureObject( GLcontext *ctx, GLuint name, GLenum target )
-{
-   struct gl_texture_object *obj = _mesa_new_texture_object(ctx, name, target);
-   INTEL_CONTEXT(ctx)->vtbl.alloc_tex_obj( obj );
-   return obj;
-}
+   functions->NewTextureObject = intelNewTextureObject;
+   functions->NewTextureImage = intelNewTextureImage;
+   functions->DeleteTexture = intelDeleteTextureObject;
+   functions->FreeTexImageData = intelFreeTextureImageData;
+   functions->UpdateTexturePalette = 0;
+   functions->IsTextureResident = intelIsTextureResident;
 
-
-void intelInitTextureFuncs( struct dd_function_table *functions )
-{
-   functions->NewTextureObject          = intelNewTextureObject;
-   functions->ChooseTextureFormat       = intelChooseTextureFormat;
-   functions->TexImage1D                = intelTexImage1D;
-   functions->TexImage2D                = intelTexImage2D;
-   functions->TexImage3D                = intelTexImage3D;
-   functions->TexSubImage1D             = intelTexSubImage1D;
-   functions->TexSubImage2D             = intelTexSubImage2D;
-   functions->TexSubImage3D             = intelTexSubImage3D;
-   functions->CopyTexImage1D            = _swrast_copy_teximage1d;
-   functions->CopyTexImage2D            = _swrast_copy_teximage2d;
-   functions->CopyTexSubImage1D         = _swrast_copy_texsubimage1d;
-   functions->CopyTexSubImage2D         = _swrast_copy_texsubimage2d;
-   functions->CopyTexSubImage3D         = _swrast_copy_texsubimage3d;
-   functions->DeleteTexture             = intelDeleteTexture;
-   functions->UpdateTexturePalette      = NULL;
-   functions->IsTextureResident         = driIsTextureResident;
-   functions->TestProxyTexImage         = _mesa_test_proxy_teximage;
-   functions->DeleteTexture             = intelDeleteTexture;
-   functions->CompressedTexImage2D      = intelCompressedTexImage2D;
-   functions->CompressedTexSubImage2D   = intelCompressedTexSubImage2D;
+#if DO_DEBUG
+   if (INTEL_DEBUG & DEBUG_BUFMGR)
+      functions->TextureMemCpy = timed_memcpy;
+   else
+#endif
+      functions->TextureMemCpy = do_memcpy;
 }
diff --git a/src/mesa/drivers/dri/i915/intel_tex.h b/src/mesa/drivers/dri/i915/intel_tex.h
index 9b7e5502326..b77d7a1d8af 100644
--- a/src/mesa/drivers/dri/i915/intel_tex.h
+++ b/src/mesa/drivers/dri/i915/intel_tex.h
@@ -33,13 +33,119 @@
 #include "texmem.h"
 
 
-void intelInitTextureFuncs( struct dd_function_table *functions );
+void intelInitTextureFuncs(struct dd_function_table *functions);
 
-void intelDestroyTexObj( intelContextPtr intel, intelTextureObjectPtr t );
-int intelUploadTexImages( intelContextPtr intel, intelTextureObjectPtr t,
-			  GLuint face );
+const struct gl_texture_format *intelChooseTextureFormat(GLcontext * ctx,
+                                                         GLint internalFormat,
+                                                         GLenum format,
+                                                         GLenum type);
+
+
+void intelTexImage3D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint height, GLint depth,
+                     GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage3D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset, GLint yoffset, GLint zoffset,
+                        GLsizei width, GLsizei height, GLsizei depth,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelTexImage2D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint height, GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage2D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset, GLint yoffset,
+                        GLsizei width, GLsizei height,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelTexImage1D(GLcontext * ctx,
+                     GLenum target, GLint level,
+                     GLint internalFormat,
+                     GLint width, GLint border,
+                     GLenum format, GLenum type, const void *pixels,
+                     const struct gl_pixelstore_attrib *packing,
+                     struct gl_texture_object *texObj,
+                     struct gl_texture_image *texImage);
+
+void intelTexSubImage1D(GLcontext * ctx,
+                        GLenum target,
+                        GLint level,
+                        GLint xoffset,
+                        GLsizei width,
+                        GLenum format, GLenum type,
+                        const GLvoid * pixels,
+                        const struct gl_pixelstore_attrib *packing,
+                        struct gl_texture_object *texObj,
+                        struct gl_texture_image *texImage);
+
+void intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
+                         GLenum internalFormat,
+                         GLint x, GLint y, GLsizei width, GLint border);
+
+void intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
+                         GLenum internalFormat,
+                         GLint x, GLint y, GLsizei width, GLsizei height,
+                         GLint border);
+
+void intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
+                            GLint xoffset, GLint x, GLint y, GLsizei width);
+
+void intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
+                            GLint xoffset, GLint yoffset,
+                            GLint x, GLint y, GLsizei width, GLsizei height);
+
+void intelGetTexImage(GLcontext * ctx, GLenum target, GLint level,
+                      GLenum format, GLenum type, GLvoid * pixels,
+                      struct gl_texture_object *texObj,
+                      struct gl_texture_image *texImage);
+
+void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
+				GLint internalFormat,
+				GLint width, GLint height, GLint border,
+				GLsizei imageSize, const GLvoid *data,
+				struct gl_texture_object *texObj,
+				struct gl_texture_image *texImage );
+
+void intelGetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
+				GLvoid *pixels,
+				const struct gl_texture_object *texObj,
+				const struct gl_texture_image *texImage);
+
+void intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+		       unsigned long long offset, GLint depth, GLuint pitch);
+
+GLuint intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit);
+
+void intel_tex_map_images(struct intel_context *intel,
+                          struct intel_texture_object *intelObj);
+
+void intel_tex_unmap_images(struct intel_context *intel,
+                            struct intel_texture_object *intelObj);
+
+int intel_compressed_num_bytes(GLuint mesaFormat);
 
-GLboolean 
-intel_driReinitTextureHeap( driTexHeap *heap,
-			    unsigned size );
 #endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_copy.c b/src/mesa/drivers/dri/i915/intel_tex_copy.c
index b85a25642a2..b85a25642a2 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_copy.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_format.c b/src/mesa/drivers/dri/i915/intel_tex_format.c
index 6e058dff69f..6e058dff69f 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_format.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_format.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_image.c b/src/mesa/drivers/dri/i915/intel_tex_image.c
index f790b1e6f73..197cf35ebe6 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_image.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_image.c
@@ -221,11 +221,10 @@ try_pbo_upload(struct intel_context *intel,
    intelFlush(&intel->ctx);
    LOCK_HARDWARE(intel);
    {
-      struct _DriBufferObject *src_buffer =
-         intel_bufferobj_buffer(intel, pbo, INTEL_READ);
-      struct _DriBufferObject *dst_buffer =
-         intel_region_buffer(intel->intelScreen, intelImage->mt->region,
-                             INTEL_WRITE_FULL);
+      dri_bo *src_buffer = intel_bufferobj_buffer(intel, pbo, INTEL_READ);
+      dri_bo *dst_buffer = intel_region_buffer(intel->intelScreen,
+					       intelImage->mt->region,
+					       INTEL_WRITE_FULL);
 
 
       intelEmitCopyBlit(intel,
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_layout.c b/src/mesa/drivers/dri/i915/intel_tex_layout.c
index fe61b441945..fe61b441945 120000
--- a/src/mesa/drivers/dri/i915tex/intel_tex_layout.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_layout.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_subimage.c b/src/mesa/drivers/dri/i915/intel_tex_subimage.c
index 3935787806b..3935787806b 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_subimage.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_subimage.c
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c b/src/mesa/drivers/dri/i915/intel_tex_validate.c
index af18c26d55c..af18c26d55c 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i915/intel_tex_validate.c
diff --git a/src/mesa/drivers/dri/i915/intel_texmem.c b/src/mesa/drivers/dri/i915/intel_texmem.c
deleted file mode 100644
index 09beec95b8c..00000000000
--- a/src/mesa/drivers/dri/i915/intel_texmem.c
+++ /dev/null
@@ -1,72 +0,0 @@
-#include "texmem.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "macros.h"
-
-#include "intel_tex.h"
-
-static GLuint
-driLog2( GLuint n )
-{
-   GLuint log2;
-
-   for ( log2 = 1 ; n > 1 ; log2++ ) {
-      n >>= 1;
-   }
-
-   return log2;
-}
-
-static void calculate_heap_size( driTexHeap * heap, unsigned size, 
-				 unsigned nr_regions, unsigned alignmentShift )
-{
-   unsigned     l;
-
-   l = driLog2( (size - 1) / nr_regions );
-   if ( l < alignmentShift )
-   {
-      l = alignmentShift;
-   }
-
-   heap->logGranularity = l;
-   heap->size = size & ~((1L << l) - 1);
-}
-
-
-GLboolean 
-intel_driReinitTextureHeap( driTexHeap *heap,
-			    unsigned size )
-{
-   driTextureObject *t, *tmp;
-
-   /* Kick out everything:
-    */
-   foreach_s ( t, tmp, & heap->texture_objects ) {
-      if ( t->tObj != NULL ) {
-	 driSwapOutTextureObject( t );
-      }
-      else {
-	 driDestroyTextureObject( t );
-      }
-   }
-   
-   /* Destroy the memory manager:
-    */
-   mmDestroy( heap->memory_heap );
-      
-   /* Recreate the memory manager:
-    */
-   calculate_heap_size(heap, size, heap->nrRegions, heap->alignmentShift);
-   heap->memory_heap = mmInit( 0, heap->size );
-   if ( heap->memory_heap == NULL ) {
-      fprintf(stderr, "driReinitTextureHeap: couldn't recreate memory heap\n");
-      FREE( heap );
-      return GL_FALSE;
-   }
-
-   make_empty_list( & heap->texture_objects );
-
-   return GL_TRUE;
-}
-
-
diff --git a/src/mesa/drivers/dri/i915/intel_tris.c b/src/mesa/drivers/dri/i915/intel_tris.c
index b2787ee60ac..5fe3d4561fc 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.c
+++ b/src/mesa/drivers/dri/i915/intel_tris.c
@@ -29,6 +29,8 @@
 #include "context.h"
 #include "macros.h"
 #include "enums.h"
+#include "texobj.h"
+#include "state.h"
 #include "dd.h"
 
 #include "swrast/swrast.h"
@@ -38,19 +40,121 @@
 #include "tnl/t_vertex.h"
 
 #include "intel_screen.h"
+#include "intel_context.h"
 #include "intel_tris.h"
 #include "intel_batchbuffer.h"
+#include "intel_buffers.h"
 #include "intel_reg.h"
 #include "intel_span.h"
+#include "intel_tex.h"
 
-/* XXX we shouldn't include these headers in this file, but we need them
- * for fallbackStrings, below.
+static void intelRenderPrimitive(GLcontext * ctx, GLenum prim);
+static void intelRasterPrimitive(GLcontext * ctx, GLenum rprim,
+                                 GLuint hwprim);
+
+/*
  */
-#include "i830_context.h"
-#include "i915_context.h"
+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;
+}
+
+
+/* Emit a primitive referencing vertices in a vertex buffer.
+ */
+void
+intelStartInlinePrimitive(struct intel_context *intel,
+                          GLuint prim, GLuint batch_flags)
+{
+   BATCH_LOCALS;
+
+   intel->vtbl.emit_state(intel);
+
+   /* Need to make sure at the very least that we don't wrap
+    * batchbuffers in BEGIN_BATCH below, otherwise the primitive will
+    * be emitted to a batchbuffer missing the required full-state
+    * preamble.
+    */
+   if (intel_batchbuffer_space(intel->batch) < 100) {
+      intel_batchbuffer_flush(intel->batch);
+      intel->vtbl.emit_state(intel);
+   }
+
+/*    _mesa_printf("%s *", __progname); */
+
+   intel_wait_flips(intel, batch_flags);
+
+   /* Emit a slot which will be filled with the inline primitive
+    * command later.
+    */
+   BEGIN_BATCH(2, batch_flags);
+   OUT_BATCH(0);
+
+   intel->prim.start_ptr = intel->batch->ptr;
+   intel->prim.primitive = prim;
+   intel->prim.flush = intel_flush_inline_primitive;
+
+   OUT_BATCH(0);
+   ADVANCE_BATCH();
+
+/*    _mesa_printf(">"); */
+}
+
+
+void
+intelWrapInlinePrimitive(struct intel_context *intel)
+{
+   GLuint prim = intel->prim.primitive;
+   GLuint batchflags = intel->batch->flags;
+
+   intel_flush_inline_primitive(intel);
+   intel_batchbuffer_flush(intel->batch);
+   intelStartInlinePrimitive(intel, prim, batchflags);  /* ??? */
+}
+
+GLuint *
+intelExtendInlinePrimitive(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)
+      intelWrapInlinePrimitive(intel);
+
+/*    _mesa_printf("."); */
+
+   intel->vtbl.assert_not_dirty(intel);
+
+   ptr = (GLuint *) intel->batch->ptr;
+   intel->batch->ptr += sz;
+
+   return ptr;
+}
+
 
-static void intelRenderPrimitive( GLcontext *ctx, GLenum prim );
-static void intelRasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
 
 /***********************************************************************
  *                    Emit primitives as inline vertices               *
@@ -69,75 +173,81 @@ do {								\
 #else
 #define COPY_DWORDS( j, vb, vertsize, v )	\
 do {						\
-   if (0) fprintf(stderr, "\n");	\
    for ( j = 0 ; j < vertsize ; j++ ) {		\
-      if (0) fprintf(stderr, "   -- v(%d): %x/%f\n",j,	\
-	      ((GLuint *)v)[j],			\
-	      ((GLfloat *)v)[j]);		\
       vb[j] = ((GLuint *)v)[j];			\
    }						\
    vb += vertsize;				\
 } while (0)
 #endif
 
-static void __inline__ intel_draw_quad( intelContextPtr intel,
-					intelVertexPtr v0,
-					intelVertexPtr v1,
-					intelVertexPtr v2,
-					intelVertexPtr v3 )
+static void
+intel_draw_quad(struct intel_context *intel,
+                intelVertexPtr v0,
+                intelVertexPtr v1, intelVertexPtr v2, intelVertexPtr v3)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 6 * vertsize );
+   GLuint *vb = intelExtendInlinePrimitive(intel, 6 * vertsize);
    int j;
 
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v3 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v2 );
-   COPY_DWORDS( j, vb, vertsize, v3 );
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
+
+   /* If smooth shading, draw like a trifan which gives better
+    * rasterization.  Otherwise draw as two triangles with provoking
+    * vertex in third position as required for flat shading.
+    */
+   if (intel->ctx.Light.ShadeModel == GL_FLAT) {
+      COPY_DWORDS(j, vb, vertsize, v3);
+      COPY_DWORDS(j, vb, vertsize, v1);
+   }
+   else {
+      COPY_DWORDS(j, vb, vertsize, v2);
+      COPY_DWORDS(j, vb, vertsize, v0);
+   }
+
+   COPY_DWORDS(j, vb, vertsize, v2);
+   COPY_DWORDS(j, vb, vertsize, v3);
 }
 
-static void __inline__ intel_draw_triangle( intelContextPtr intel,
-					    intelVertexPtr v0,
-					    intelVertexPtr v1,
-					    intelVertexPtr v2 )
+static void
+intel_draw_triangle(struct intel_context *intel,
+                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 3 * vertsize );
+   GLuint *vb = intelExtendInlinePrimitive(intel, 3 * vertsize);
    int j;
-   
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
-   COPY_DWORDS( j, vb, vertsize, v2 );
+
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
+   COPY_DWORDS(j, vb, vertsize, v2);
 }
 
 
-static __inline__ void intel_draw_line( intelContextPtr intel,
-					intelVertexPtr v0,
-					intelVertexPtr v1 )
+static void
+intel_draw_line(struct intel_context *intel,
+                intelVertexPtr v0, intelVertexPtr v1)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, 2 * vertsize );
+   GLuint *vb = intelExtendInlinePrimitive(intel, 2 * vertsize);
    int j;
 
-   COPY_DWORDS( j, vb, vertsize, v0 );
-   COPY_DWORDS( j, vb, vertsize, v1 );
+   COPY_DWORDS(j, vb, vertsize, v0);
+   COPY_DWORDS(j, vb, vertsize, v1);
 }
 
 
-static __inline__ void intel_draw_point( intelContextPtr intel,
-					 intelVertexPtr v0 )
+static void
+intel_draw_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, vertsize );
+   GLuint *vb = intelExtendInlinePrimitive(intel, vertsize);
    int j;
 
    /* Adjust for sub pixel position -- still required for conform. */
-   *(float *)&vb[0] = v0->v.x - 0.125;
-   *(float *)&vb[1] = v0->v.y - 0.125;
-   for (j = 2 ; j < vertsize ; j++)
-     vb[j] = v0->ui[j];
+   *(float *) &vb[0] = v0->v.x - 0.125;
+   *(float *) &vb[1] = v0->v.y - 0.125;
+   for (j = 2; j < vertsize; j++)
+      vb[j] = v0->ui[j];
 }
 
 
@@ -146,13 +256,17 @@ static __inline__ void intel_draw_point( intelContextPtr intel,
  *                Fixup for ARB_point_parameters                       *
  ***********************************************************************/
 
-static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
+/* Currently not working - VERT_ATTRIB_POINTSIZE isn't correctly
+ * represented in the fragment program InputsRead field.
+ */
+static void
+intel_atten_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLcontext *ctx = &intel->ctx;
    GLfloat psz[4], col[4], restore_psz, restore_alpha;
 
-   _tnl_get_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz );
-   _tnl_get_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col );
+   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
 
    restore_psz = psz[0];
    restore_alpha = col[3];
@@ -170,19 +284,19 @@ static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
       psz[0] = 1.0;
 
    if (restore_psz != psz[0] || restore_alpha != col[3]) {
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col);
-   
-      intel_draw_point( intel, v0 );
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
+
+      intel_draw_point(intel, v0);
 
       psz[0] = restore_psz;
       col[3] = restore_alpha;
 
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr( ctx, v0, _TNL_ATTRIB_COLOR0, col);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
+      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
    }
    else
-      intel_draw_point( intel, v0 );
+      intel_draw_point(intel, v0);
 }
 
 
@@ -195,45 +309,44 @@ static void intel_atten_point( intelContextPtr intel, intelVertexPtr v0 )
 
 
 
-static void intel_wpos_triangle( intelContextPtr intel,
-				 intelVertexPtr v0,
-				 intelVertexPtr v1,
-				 intelVertexPtr v2 )
+static void
+intel_wpos_triangle(struct intel_context *intel,
+                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
-   
-   __memcpy( ((char *)v0) + offset, v0, size );
-   __memcpy( ((char *)v1) + offset, v1, size );
-   __memcpy( ((char *)v2) + offset, v2, size );
 
-   intel_draw_triangle( intel, v0, v1, v2 );
+   __memcpy(((char *) v0) + offset, v0, size);
+   __memcpy(((char *) v1) + offset, v1, size);
+   __memcpy(((char *) v2) + offset, v2, size);
+
+   intel_draw_triangle(intel, v0, v1, v2);
 }
 
 
-static void intel_wpos_line( intelContextPtr intel,
-			     intelVertexPtr v0,
-			     intelVertexPtr v1 )
+static void
+intel_wpos_line(struct intel_context *intel,
+                intelVertexPtr v0, intelVertexPtr v1)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
 
-   __memcpy( ((char *)v0) + offset, v0, size );
-   __memcpy( ((char *)v1) + offset, v1, size );
+   __memcpy(((char *) v0) + offset, v0, size);
+   __memcpy(((char *) v1) + offset, v1, size);
 
-   intel_draw_line( intel, v0, v1 );
+   intel_draw_line(intel, v0, v1);
 }
 
 
-static void intel_wpos_point( intelContextPtr intel,
-			      intelVertexPtr v0 )
+static void
+intel_wpos_point(struct intel_context *intel, intelVertexPtr v0)
 {
    GLuint offset = intel->wpos_offset;
    GLuint size = intel->wpos_size;
 
-   __memcpy( ((char *)v0) + offset, v0, size );
+   __memcpy(((char *) v0) + offset, v0, size);
 
-   intel_draw_point( intel, v0 );
+   intel_draw_point(intel, v0);
 }
 
 
@@ -290,11 +403,12 @@ do { 						\
 #define INTEL_MAX_TRIFUNC	0x10
 
 
-static struct {
-   tnl_points_func	        points;
-   tnl_line_func		line;
-   tnl_triangle_func	triangle;
-   tnl_quad_func		quad;
+static struct
+{
+   tnl_points_func points;
+   tnl_line_func line;
+   tnl_triangle_func triangle;
+   tnl_quad_func quad;
 } rast_tab[INTEL_MAX_TRIFUNC];
 
 
@@ -355,7 +469,7 @@ do {							\
 #define VERT_RESTORE_SPEC( idx ) if (specoffset) v[idx]->ui[specoffset] = spec[idx]
 
 #define LOCAL_VARS(n)							\
-   intelContextPtr intel = INTEL_CONTEXT(ctx);				\
+   struct intel_context *intel = intel_context(ctx);				\
    GLuint color[n], spec[n];						\
    GLuint coloroffset = intel->coloroffset;		\
    GLboolean specoffset = intel->specoffset;			\
@@ -366,7 +480,7 @@ do {							\
  *                Helpers for rendering unfilled primitives            *
  ***********************************************************************/
 
-static const GLuint hw_prim[GL_POLYGON+1] = {
+static const GLuint hw_prim[GL_POLYGON + 1] = {
    PRIM3D_POINTLIST,
    PRIM3D_LINELIST,
    PRIM3D_LINELIST,
@@ -456,7 +570,8 @@ static const GLuint hw_prim[GL_POLYGON+1] = {
 #include "tnl_dd/t_dd_tritmp.h"
 
 
-static void init_rast_tab( void )
+static void
+init_rast_tab(void)
 {
    init();
    init_offset();
@@ -487,10 +602,8 @@ static void init_rast_tab( void )
  * primitives.
  */
 static void
-intel_fallback_tri( intelContextPtr intel,
-		   intelVertex *v0,
-		   intelVertex *v1,
-		   intelVertex *v2 )
+intel_fallback_tri(struct intel_context *intel,
+                   intelVertex * v0, intelVertex * v1, intelVertex * v2)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[3];
@@ -498,19 +611,20 @@ intel_fallback_tri( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   _swsetup_Translate( ctx, v1, &v[1] );
-   _swsetup_Translate( ctx, v2, &v[2] );
-   intelSpanRenderStart( ctx );
-   _swrast_Triangle( ctx, &v[0], &v[1], &v[2] );
-   intelSpanRenderFinish( ctx );
+   INTEL_FIREVERTICES(intel);
+
+   _swsetup_Translate(ctx, v0, &v[0]);
+   _swsetup_Translate(ctx, v1, &v[1]);
+   _swsetup_Translate(ctx, v2, &v[2]);
+   intelSpanRenderStart(ctx);
+   _swrast_Triangle(ctx, &v[0], &v[1], &v[2]);
+   intelSpanRenderFinish(ctx);
 }
 
 
 static void
-intel_fallback_line( intelContextPtr intel,
-		    intelVertex *v0,
-		    intelVertex *v1 )
+intel_fallback_line(struct intel_context *intel,
+                    intelVertex * v0, intelVertex * v1)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[2];
@@ -518,17 +632,18 @@ intel_fallback_line( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   _swsetup_Translate( ctx, v1, &v[1] );
-   intelSpanRenderStart( ctx );
-   _swrast_Line( ctx, &v[0], &v[1] );
-   intelSpanRenderFinish( ctx );
-}
+   INTEL_FIREVERTICES(intel);
 
+   _swsetup_Translate(ctx, v0, &v[0]);
+   _swsetup_Translate(ctx, v1, &v[1]);
+   intelSpanRenderStart(ctx);
+   _swrast_Line(ctx, &v[0], &v[1]);
+   intelSpanRenderFinish(ctx);
+}
 
 static void
-intel_fallback_point( intelContextPtr intel,
-		     intelVertex *v0 )
+intel_fallback_point(struct intel_context *intel,
+		     intelVertex * v0)
 {
    GLcontext *ctx = &intel->ctx;
    SWvertex v[1];
@@ -536,12 +651,13 @@ intel_fallback_point( intelContextPtr intel,
    if (0)
       fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   _swsetup_Translate( ctx, v0, &v[0] );
-   intelSpanRenderStart( ctx );
-   _swrast_Point( ctx, &v[0] );
-   intelSpanRenderFinish( ctx );
-}
+   INTEL_FIREVERTICES(intel);
 
+   _swsetup_Translate(ctx, v0, &v[0]);
+   intelSpanRenderStart(ctx);
+   _swrast_Point(ctx, &v[0]);
+   intelSpanRenderFinish(ctx);
+}
 
 
 /**********************************************************************/
@@ -558,7 +674,7 @@ intel_fallback_point( intelContextPtr intel,
 #define INIT(x) intelRenderPrimitive( ctx, x )
 #undef LOCAL_VARS
 #define LOCAL_VARS						\
-    intelContextPtr intel = INTEL_CONTEXT(ctx);			\
+    struct intel_context *intel = intel_context(ctx);			\
     GLubyte *vertptr = (GLubyte *)intel->verts;			\
     const GLuint vertsize = intel->vertex_size;       	\
     const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;	\
@@ -581,10 +697,10 @@ intel_fallback_point( intelContextPtr intel,
 
 
 
-static void intelRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
-				   GLuint n )
+static void
+intelRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    GLuint prim = intel->render_primitive;
@@ -593,39 +709,40 @@ static void intelRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
     */
    {
       GLuint *tmp = VB->Elts;
-      VB->Elts = (GLuint *)elts;
-      tnl->Driver.Render.PrimTabElts[GL_POLYGON]( ctx, 0, n, 
-						  PRIM_BEGIN|PRIM_END );
+      VB->Elts = (GLuint *) elts;
+      tnl->Driver.Render.PrimTabElts[GL_POLYGON] (ctx, 0, n,
+                                                  PRIM_BEGIN | PRIM_END);
       VB->Elts = tmp;
    }
 
    /* Restore the render primitive
     */
    if (prim != GL_POLYGON)
-      tnl->Driver.Render.PrimitiveNotify( ctx, prim );
+      tnl->Driver.Render.PrimitiveNotify(ctx, prim);
 }
 
-static void intelRenderClippedLine( GLcontext *ctx, GLuint ii, GLuint jj )
+static void
+intelRenderClippedLine(GLcontext * ctx, GLuint ii, GLuint jj)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
 
-   tnl->Driver.Render.Line( ctx, ii, jj );
+   tnl->Driver.Render.Line(ctx, ii, jj);
 }
 
-static void intelFastRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
-				       GLuint n )
+static void
+intelFastRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
 {
-   intelContextPtr intel = INTEL_CONTEXT( ctx );
+   struct intel_context *intel = intel_context(ctx);
    const GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive( intel, (n-2) * 3 * vertsize );
-   GLubyte *vertptr = (GLubyte *)intel->verts;
-   const GLuint *start = (const GLuint *)V(elts[0]);
-   int i,j;
-
-   for (i = 2 ; i < n ; i++) {
-      COPY_DWORDS( j, vb, vertsize, V(elts[i-1]) );
-      COPY_DWORDS( j, vb, vertsize, V(elts[i]) );
-      COPY_DWORDS( j, vb, vertsize, start );
+   GLuint *vb = intelExtendInlinePrimitive(intel, (n - 2) * 3 * vertsize);
+   GLubyte *vertptr = (GLubyte *) intel->verts;
+   const GLuint *start = (const GLuint *) V(elts[0]);
+   int i, j;
+
+   for (i = 2; i < n; i++) {
+      COPY_DWORDS(j, vb, vertsize, V(elts[i - 1]));
+      COPY_DWORDS(j, vb, vertsize, V(elts[i]));
+      COPY_DWORDS(j, vb, vertsize, start);
    }
 }
 
@@ -636,68 +753,75 @@ static void intelFastRenderClippedPoly( GLcontext *ctx, const GLuint *elts,
 
 
 
-#define POINT_FALLBACK (0)
-#define LINE_FALLBACK (DD_LINE_STIPPLE)
-#define TRI_FALLBACK (0)
-#define ANY_FALLBACK_FLAGS (POINT_FALLBACK|LINE_FALLBACK|TRI_FALLBACK|\
-                            DD_TRI_STIPPLE|DD_POINT_ATTEN)
-#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE|DD_TRI_OFFSET|DD_TRI_UNFILLED)
+#define ANY_FALLBACK_FLAGS (DD_LINE_STIPPLE | DD_TRI_STIPPLE | DD_POINT_ATTEN | DD_POINT_SMOOTH | DD_TRI_SMOOTH)
+#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE | DD_TRI_OFFSET | DD_TRI_UNFILLED)
 
-void intelChooseRenderState(GLcontext *ctx)
+void
+intelChooseRenderState(GLcontext * ctx)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
    GLuint flags = ctx->_TriangleCaps;
    const struct gl_fragment_program *fprog = ctx->FragmentProgram._Current;
    GLboolean have_wpos = (fprog && (fprog->Base.InputsRead & FRAG_BIT_WPOS));
    GLuint index = 0;
 
    if (INTEL_DEBUG & DEBUG_STATE)
-     fprintf(stderr,"\n%s\n",__FUNCTION__);
+      fprintf(stderr, "\n%s\n", __FUNCTION__);
 
-   if ((flags & (ANY_FALLBACK_FLAGS|ANY_RASTER_FLAGS)) || have_wpos) {
+   if ((flags & (ANY_FALLBACK_FLAGS | ANY_RASTER_FLAGS)) || have_wpos) {
 
       if (flags & ANY_RASTER_FLAGS) {
-	 if (flags & DD_TRI_LIGHT_TWOSIDE)    index |= INTEL_TWOSIDE_BIT;
-	 if (flags & DD_TRI_OFFSET)	      index |= INTEL_OFFSET_BIT;
-	 if (flags & DD_TRI_UNFILLED)	      index |= INTEL_UNFILLED_BIT;
+         if (flags & DD_TRI_LIGHT_TWOSIDE)
+            index |= INTEL_TWOSIDE_BIT;
+         if (flags & DD_TRI_OFFSET)
+            index |= INTEL_OFFSET_BIT;
+         if (flags & DD_TRI_UNFILLED)
+            index |= INTEL_UNFILLED_BIT;
       }
 
       if (have_wpos) {
-	 intel->draw_point = intel_wpos_point;
-	 intel->draw_line = intel_wpos_line;
-	 intel->draw_tri = intel_wpos_triangle;
+         intel->draw_point = intel_wpos_point;
+         intel->draw_line = intel_wpos_line;
+         intel->draw_tri = intel_wpos_triangle;
 
-	 /* Make sure these get called:
-	  */
-	 index |= INTEL_FALLBACK_BIT;
+         /* Make sure these get called:
+          */
+         index |= INTEL_FALLBACK_BIT;
       }
       else {
-	 intel->draw_point = intel_draw_point;
-	 intel->draw_line = intel_draw_line;
-	 intel->draw_tri = intel_draw_triangle;
+         intel->draw_point = intel_draw_point;
+         intel->draw_line = intel_draw_line;
+         intel->draw_tri = intel_draw_triangle;
       }
 
       /* Hook in fallbacks for specific primitives.
        */
-      if (flags & ANY_FALLBACK_FLAGS)
-      {
-	 if (flags & POINT_FALLBACK)
-	    intel->draw_point = intel_fallback_point;
-
-	 if (flags & LINE_FALLBACK)
-	    intel->draw_line = intel_fallback_line;
-
-	 if (flags & TRI_FALLBACK)
-	    intel->draw_tri = intel_fallback_tri;
-
-	 if ((flags & DD_TRI_STIPPLE) && !intel->hw_stipple) 
-	    intel->draw_tri = intel_fallback_tri;
-
-	 if (flags & DD_POINT_ATTEN)
-	    intel->draw_point = intel_atten_point;
-
-	 index |= INTEL_FALLBACK_BIT;
+      if (flags & ANY_FALLBACK_FLAGS) {
+         if (flags & DD_LINE_STIPPLE)
+            intel->draw_line = intel_fallback_line;
+
+         if ((flags & DD_TRI_STIPPLE) && !intel->hw_stipple)
+            intel->draw_tri = intel_fallback_tri;
+
+         if (flags & DD_TRI_SMOOTH) {
+	    if (intel->strict_conformance)
+	       intel->draw_tri = intel_fallback_tri;
+	 }
+
+         if (flags & DD_POINT_ATTEN) {
+	    if (0)
+	       intel->draw_point = intel_atten_point;
+	    else
+	       intel->draw_point = intel_fallback_point;
+	 }
+
+	 if (flags & DD_POINT_SMOOTH) {
+	    if (intel->strict_conformance)
+	       intel->draw_point = intel_fallback_point;
+	 }
+
+         index |= INTEL_FALLBACK_BIT;
       }
    }
 
@@ -710,20 +834,21 @@ void intelChooseRenderState(GLcontext *ctx)
       tnl->Driver.Render.Quad = rast_tab[index].quad;
 
       if (index == 0) {
-	 tnl->Driver.Render.PrimTabVerts = intel_render_tab_verts;
-	 tnl->Driver.Render.PrimTabElts = intel_render_tab_elts;
-	 tnl->Driver.Render.ClippedLine = line; /* from tritmp.h */
-	 tnl->Driver.Render.ClippedPolygon = intelFastRenderClippedPoly;
-      } else {
-	 tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-	 tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
-	 tnl->Driver.Render.ClippedLine = intelRenderClippedLine;
-	 tnl->Driver.Render.ClippedPolygon = intelRenderClippedPoly;
+         tnl->Driver.Render.PrimTabVerts = intel_render_tab_verts;
+         tnl->Driver.Render.PrimTabElts = intel_render_tab_elts;
+         tnl->Driver.Render.ClippedLine = line; /* from tritmp.h */
+         tnl->Driver.Render.ClippedPolygon = intelFastRenderClippedPoly;
+      }
+      else {
+         tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
+         tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
+         tnl->Driver.Render.ClippedLine = intelRenderClippedLine;
+         tnl->Driver.Render.ClippedPolygon = intelRenderClippedPoly;
       }
    }
 }
 
-static const GLenum reduced_prim[GL_POLYGON+1] = {
+static const GLenum reduced_prim[GL_POLYGON + 1] = {
    GL_POINTS,
    GL_LINES,
    GL_LINES,
@@ -744,35 +869,52 @@ static const GLenum reduced_prim[GL_POLYGON+1] = {
 
 
 
-static void intelRunPipeline( GLcontext *ctx )
+static void
+intelRunPipeline(GLcontext * ctx)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
+
+   _mesa_lock_context_textures(ctx);
+   
+   if (ctx->NewState)
+      _mesa_update_state_locked(ctx);
 
    if (intel->NewGLState) {
       if (intel->NewGLState & _NEW_TEXTURE) {
-	 intel->vtbl.update_texture_state( intel ); 
+         intel->vtbl.update_texture_state(intel);
       }
 
       if (!intel->Fallback) {
-	 if (intel->NewGLState & _INTEL_NEW_RENDERSTATE)
-	    intelChooseRenderState( ctx );
+         if (intel->NewGLState & _INTEL_NEW_RENDERSTATE)
+            intelChooseRenderState(ctx);
       }
 
       intel->NewGLState = 0;
    }
 
-   _tnl_run_pipeline( ctx );
+   _tnl_run_pipeline(ctx);
+
+   _mesa_unlock_context_textures(ctx);
 }
 
-static void intelRenderStart( GLcontext *ctx )
+static void
+intelRenderStart(GLcontext * ctx)
 {
-   INTEL_CONTEXT(ctx)->vtbl.render_start( INTEL_CONTEXT(ctx) );
+   struct intel_context *intel = intel_context(ctx);
+
+   intel->vtbl.render_start(intel_context(ctx));
+   intel->vtbl.emit_state(intel);
 }
 
-static void intelRenderFinish( GLcontext *ctx )
+static void
+intelRenderFinish(GLcontext * ctx)
 {
-   if (INTEL_CONTEXT(ctx)->RenderIndex & INTEL_FALLBACK_BIT)
-      _swrast_flush( ctx );
+   struct intel_context *intel = intel_context(ctx);
+
+   if (intel->RenderIndex & INTEL_FALLBACK_BIT)
+      _swrast_flush(ctx);
+
+   INTEL_FIREVERTICES(intel);
 }
 
 
@@ -781,28 +923,33 @@ static void intelRenderFinish( GLcontext *ctx )
  /* System to flush dma and emit state changes based on the rasterized
   * primitive.
   */
-static void intelRasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim )
+static void
+intelRasterPrimitive(GLcontext * ctx, GLenum rprim, GLuint hwprim)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
 
    if (0)
-      fprintf(stderr, "%s %s %x\n", __FUNCTION__, 
-	      _mesa_lookup_enum_by_nr(rprim), hwprim);
+      fprintf(stderr, "%s %s %x\n", __FUNCTION__,
+              _mesa_lookup_enum_by_nr(rprim), hwprim);
+
+   intel->vtbl.reduced_primitive_state(intel, rprim);
 
-   intel->vtbl.reduced_primitive_state( intel, rprim );
-    
    /* Start a new primitive.  Arrange to have it flushed later on.
     */
-   if (hwprim != intel->prim.primitive) 
-      intelStartInlinePrimitive( intel, hwprim );
+   if (hwprim != intel->prim.primitive) {
+      INTEL_FIREVERTICES(intel);
+
+      intelStartInlinePrimitive(intel, hwprim, INTEL_BATCH_CLIPRECTS);
+   }
 }
 
 
-/* 
- */
-static void intelRenderPrimitive( GLcontext *ctx, GLenum prim )
+ /* 
+  */
+static void
+intelRenderPrimitive(GLcontext * ctx, GLenum prim)
 {
-   intelContextPtr intel = INTEL_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
 
    if (0)
       fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
@@ -817,63 +964,54 @@ static void intelRenderPrimitive( GLcontext *ctx, GLenum prim )
     * lower level functions in that case, potentially pingponging the
     * state:
     */
-   if (reduced_prim[prim] == GL_TRIANGLES && 
+   if (reduced_prim[prim] == GL_TRIANGLES &&
        (ctx->_TriangleCaps & DD_TRI_UNFILLED))
       return;
 
    /* Set some primitive-dependent state and Start? a new primitive.
     */
-   intelRasterPrimitive( ctx, reduced_prim[prim], hw_prim[prim] );
+   intelRasterPrimitive(ctx, reduced_prim[prim], hw_prim[prim]);
 }
 
 
-/**********************************************************************/
-/*           Transition to/from hardware rasterization.               */
-/**********************************************************************/
-
-static struct {
-   GLuint bit;
-   const char *str;
-} fallbackStrings[] = {
-   { INTEL_FALLBACK_DRAW_BUFFER, "Draw buffer" },
-   { INTEL_FALLBACK_READ_BUFFER, "Read buffer" },
-   { INTEL_FALLBACK_USER, "User" },
-   { INTEL_FALLBACK_NO_BATCHBUFFER, "No Batchbuffer" },
-   { INTEL_FALLBACK_NO_TEXMEM, "No Texmem" },
-   { INTEL_FALLBACK_RENDERMODE, "Rendermode" },
-
-   { I830_FALLBACK_TEXTURE, "i830 texture" },
-   { I830_FALLBACK_COLORMASK, "i830 colormask" },
-   { I830_FALLBACK_STENCIL, "i830 stencil" },
-   { I830_FALLBACK_STIPPLE, "i830 stipple" },
-   { I830_FALLBACK_LOGICOP, "i830 logicop" },
-
-   { I915_FALLBACK_TEXTURE, "i915 texture" },
-   { I915_FALLBACK_COLORMASK, "i915 colormask" },
-   { I915_FALLBACK_STENCIL, "i915 stencil" },
-   { I915_FALLBACK_STIPPLE, "i915 stipple" },
-   { I915_FALLBACK_PROGRAM, "i915 program" },
-   { I915_FALLBACK_LOGICOP, "i915 logicop" },
-   { I915_FALLBACK_POLYGON_SMOOTH, "i915 polygon smooth" },
-   { I915_FALLBACK_POINT_SMOOTH, "i915 point smooth" },
-
-   { 0, NULL }
+ /**********************************************************************/
+ /*           Transition to/from hardware rasterization.               */
+ /**********************************************************************/
+
+static char *fallbackStrings[] = {
+   [0] = "Draw buffer",
+   [1] = "Read buffer",
+   [2] = "Depth buffer",
+   [3] = "Stencil buffer",
+   [4] = "User disable",
+   [5] = "Render mode",
+
+   [12] = "Texture",
+   [13] = "Color mask",
+   [14] = "Stencil",
+   [15] = "Stipple",
+   [16] = "Program",
+   [17] = "Logic op",
+   [18] = "Smooth polygon",
+   [19] = "Smooth point",
 };
 
 
-static const char *
+static char *
 getFallbackString(GLuint bit)
 {
-   int i;
-   for (i = 0; fallbackStrings[i].bit; i++) {
-      if (fallbackStrings[i].bit == bit)
-         return fallbackStrings[i].str;
+   int i = 0;
+   while (bit > 1) {
+      i++;
+      bit >>= 1;
    }
-   return "unknown fallback bit";
+   return fallbackStrings[i];
 }
 
 
-void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
+
+void
+intelFallback(struct intel_context *intel, GLuint bit, GLboolean mode)
 {
    GLcontext *ctx = &intel->ctx;
    TNLcontext *tnl = TNL_CONTEXT(ctx);
@@ -883,20 +1021,19 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
       intel->Fallback |= bit;
       if (oldfallback == 0) {
          intelFlush(ctx);
-         if (INTEL_DEBUG & DEBUG_FALLBACKS) 
-            fprintf(stderr, "ENTER FALLBACK 0x%x: %s\n",
+         if (INTEL_DEBUG & DEBUG_FALLBACKS)
+            fprintf(stderr, "ENTER FALLBACK %x: %s\n",
                     bit, getFallbackString(bit));
-         _swsetup_Wakeup( ctx );
+         _swsetup_Wakeup(ctx);
          intel->RenderIndex = ~0;
       }
    }
    else {
       intel->Fallback &= ~bit;
       if (oldfallback == bit) {
-         _swrast_flush( ctx );
-         if (INTEL_DEBUG & DEBUG_FALLBACKS) 
-            fprintf(stderr, "LEAVE FALLBACK 0x%x: %s\n",
-                    bit, getFallbackString(bit));
+         _swrast_flush(ctx);
+         if (INTEL_DEBUG & DEBUG_FALLBACKS)
+            fprintf(stderr, "LEAVE FALLBACK %s\n", getFallbackString(bit));
          tnl->Driver.Render.Start = intelRenderStart;
          tnl->Driver.Render.PrimitiveNotify = intelRenderPrimitive;
          tnl->Driver.Render.Finish = intelRenderFinish;
@@ -904,18 +1041,87 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
          tnl->Driver.Render.CopyPV = _tnl_copy_pv;
          tnl->Driver.Render.Interp = _tnl_interp;
 
-         _tnl_invalidate_vertex_state( ctx, ~0 );
-         _tnl_invalidate_vertices( ctx, ~0 );
-         _tnl_install_attrs( ctx, 
-                             intel->vertex_attrs, 
-                             intel->vertex_attr_count,
-                             intel->ViewportMatrix.m, 0 ); 
+         _tnl_invalidate_vertex_state(ctx, ~0);
+         _tnl_invalidate_vertices(ctx, ~0);
+         _tnl_install_attrs(ctx,
+                            intel->vertex_attrs,
+                            intel->vertex_attr_count,
+                            intel->ViewportMatrix.m, 0);
 
          intel->NewGLState |= _INTEL_NEW_RENDERSTATE;
       }
    }
 }
 
+union fi
+{
+   GLfloat f;
+   GLint i;
+};
+
+
+/**********************************************************************/
+/*             Used only with the metaops callbacks.                  */
+/**********************************************************************/
+void
+intel_meta_draw_poly(struct intel_context *intel,
+                     GLuint n,
+                     GLfloat xy[][2],
+                     GLfloat z, GLuint color, GLfloat tex[][2])
+{
+   union fi *vb;
+   GLint i;
+
+   /* All 3d primitives should be emitted with INTEL_BATCH_CLIPRECTS,
+    * otherwise the drawing origin (DR4) might not be set correctly.
+    */
+   intelStartInlinePrimitive(intel, PRIM3D_TRIFAN, INTEL_BATCH_CLIPRECTS);
+   vb = (union fi *) intelExtendInlinePrimitive(intel, n * 6);
+
+   for (i = 0; i < n; i++) {
+      vb[0].f = xy[i][0];
+      vb[1].f = xy[i][1];
+      vb[2].f = z;
+      vb[3].i = color;
+      vb[4].f = tex[i][0];
+      vb[5].f = tex[i][1];
+      vb += 6;
+   }
+
+   INTEL_FIREVERTICES(intel);
+}
+
+void
+intel_meta_draw_quad(struct intel_context *intel,
+                     GLfloat x0, GLfloat x1,
+                     GLfloat y0, GLfloat y1,
+                     GLfloat z,
+                     GLuint color,
+                     GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1)
+{
+   GLfloat xy[4][2];
+   GLfloat tex[4][2];
+
+   xy[0][0] = x0;
+   xy[0][1] = y0;
+   xy[1][0] = x1;
+   xy[1][1] = y0;
+   xy[2][0] = x1;
+   xy[2][1] = y1;
+   xy[3][0] = x0;
+   xy[3][1] = y1;
+
+   tex[0][0] = s0;
+   tex[0][1] = t0;
+   tex[1][0] = s1;
+   tex[1][1] = t0;
+   tex[2][0] = s1;
+   tex[2][1] = t1;
+   tex[3][0] = s0;
+   tex[3][1] = t1;
+
+   intel_meta_draw_poly(intel, 4, xy, z, color, tex);
+}
 
 
 
@@ -924,7 +1130,8 @@ void intelFallback( intelContextPtr intel, GLuint bit, GLboolean mode )
 /**********************************************************************/
 
 
-void intelInitTriFuncs( GLcontext *ctx )
+void
+intelInitTriFuncs(GLcontext * ctx)
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    static int firsttime = 1;
diff --git a/src/mesa/drivers/dri/i915/intel_tris.h b/src/mesa/drivers/dri/i915/intel_tris.h
index d7e382fdb3e..b7bae8cd3bc 100644
--- a/src/mesa/drivers/dri/i915/intel_tris.h
+++ b/src/mesa/drivers/dri/i915/intel_tris.h
@@ -30,6 +30,8 @@
 
 #include "mtypes.h"
 
+
+
 #define _INTEL_NEW_RENDERSTATE (_DD_NEW_LINE_STIPPLE |		\
 			       _DD_NEW_TRI_UNFILLED |		\
 			       _DD_NEW_TRI_LIGHT_TWOSIDE |	\
@@ -38,9 +40,30 @@
 			       _NEW_PROGRAM |		\
 			       _NEW_POLYGONSTIPPLE)
 
-extern void intelInitTriFuncs( GLcontext *ctx );
+extern void intelInitTriFuncs(GLcontext * ctx);
+
+extern void intelChooseRenderState(GLcontext * ctx);
+
+extern void intelStartInlinePrimitive(struct intel_context *intel,
+                                      GLuint prim, GLuint flags);
+extern void intelWrapInlinePrimitive(struct intel_context *intel);
+
+GLuint *intelExtendInlinePrimitive(struct intel_context *intel,
+                                   GLuint dwords);
+
+
+void intel_meta_draw_quad(struct intel_context *intel,
+                          GLfloat x0, GLfloat x1,
+                          GLfloat y0, GLfloat y1,
+                          GLfloat z,
+                          GLuint color,
+                          GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1);
+
+void intel_meta_draw_poly(struct intel_context *intel,
+                          GLuint n,
+                          GLfloat xy[][2],
+                          GLfloat z, GLuint color, GLfloat tex[][2]);
+
 
-extern void intelPrintRenderState( const char *msg, GLuint state );
-extern void intelChooseRenderState( GLcontext *ctx );
 
 #endif
diff --git a/src/mesa/drivers/dri/i915/server/i830_common.h b/src/mesa/drivers/dri/i915/server/i830_common.h
index fb6ceaa52d4..f1fd3939ab9 100644
--- a/src/mesa/drivers/dri/i915/server/i830_common.h
+++ b/src/mesa/drivers/dri/i915/server/i830_common.h
@@ -52,6 +52,9 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define DRM_I830_INIT_HEAP                0x0a
 #define DRM_I830_CMDBUFFER                0x0b
 #define DRM_I830_DESTROY_HEAP             0x0c
+#define DRM_I830_SET_VBLANK_PIPE          0x0d
+#define DRM_I830_GET_VBLANK_PIPE          0x0e
+#define DRM_I830_MMIO		       	  0x10
 
 typedef struct {
    enum {
@@ -83,6 +86,7 @@ typedef struct {
         int last_enqueue;	/* last time a buffer was enqueued */
 	int last_dispatch;	/* age of the most recently dispatched buffer */
 	int ctxOwner;		/* last context to upload state */
+	/** Last context that used the buffer manager. */
 	int texAge;
         int pf_enabled;		/* is pageflipping allowed? */
         int pf_active;               
@@ -119,14 +123,29 @@ typedef struct {
 	unsigned int rotated_tiled;
 	unsigned int rotated2_tiled;
 
-	int pipeA_x;
-	int pipeA_y;
-	int pipeA_w;
-	int pipeA_h;
-	int pipeB_x;
-	int pipeB_y;
-	int pipeB_w;
-	int pipeB_h;
+	int planeA_x;
+	int planeA_y;
+	int planeA_w;
+	int planeA_h;
+	int planeB_x;
+	int planeB_y;
+	int planeB_w;
+	int planeB_h;
+
+	/* Triple buffering */
+	drm_handle_t third_handle;
+	int third_offset;
+	int third_size;
+	unsigned int third_tiled;
+
+	/* buffer object handles for the static buffers.  May change
+	 * over the lifetime of the client, though it doesn't in our current
+	 * implementation.
+	 */
+	unsigned int front_bo_handle;
+	unsigned int back_bo_handle;
+	unsigned int third_bo_handle;
+	unsigned int depth_bo_handle;
 } drmI830Sarea;
 
 /* Flags for perf_boxes
@@ -208,5 +227,30 @@ typedef struct {
 	int region;
 } drmI830MemDestroyHeap;
 
+#define DRM_I830_VBLANK_PIPE_A  1
+#define DRM_I830_VBLANK_PIPE_B  2
+
+typedef struct {
+        int pipe;
+} drmI830VBlankPipe;
+
+#define MMIO_READ  0
+#define MMIO_WRITE 1
+
+#define MMIO_REGS_IA_PRIMATIVES_COUNT           0
+#define MMIO_REGS_IA_VERTICES_COUNT             1
+#define MMIO_REGS_VS_INVOCATION_COUNT           2
+#define MMIO_REGS_GS_PRIMITIVES_COUNT           3
+#define MMIO_REGS_GS_INVOCATION_COUNT           4
+#define MMIO_REGS_CL_PRIMITIVES_COUNT           5
+#define MMIO_REGS_CL_INVOCATION_COUNT           6
+#define MMIO_REGS_PS_INVOCATION_COUNT           7
+#define MMIO_REGS_PS_DEPTH_COUNT                8
+
+typedef struct {
+        unsigned int read_write:1;
+        unsigned int reg:31;
+        void __user *data;
+} drmI830MMIO;
 
 #endif /* _I830_DRM_H_ */
diff --git a/src/mesa/drivers/dri/i915/server/i830_dri.h b/src/mesa/drivers/dri/i915/server/i830_dri.h
index 6c9a7090215..c2a3af8cbf7 100644
--- a/src/mesa/drivers/dri/i915/server/i830_dri.h
+++ b/src/mesa/drivers/dri/i915/server/i830_dri.h
@@ -9,8 +9,8 @@
 #define I830_MAX_DRAWABLES 256
 
 #define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 3
-#define I830_PATCHLEVEL 0
+#define I830_MINOR_VERSION 7
+#define I830_PATCHLEVEL 2
 
 #define I830_REG_SIZE 0x80000
 
@@ -18,20 +18,20 @@ typedef struct _I830DRIRec {
    drm_handle_t regs;
    drmSize regsSize;
 
-   drmSize backbufferSize;
-   drm_handle_t backbuffer;
+   drmSize unused1; /* backbufferSize */
+   drm_handle_t unused2; /* backbuffer */
 
-   drmSize depthbufferSize;
-   drm_handle_t depthbuffer;
+   drmSize unused3; /* depthbufferSize */
+   drm_handle_t unused4; /* depthbuffer */
 
-   drmSize rotatedSize;
-   drm_handle_t rotatedbuffer;
+   drmSize unused5; /* rotatedSize */
+   drm_handle_t unused6; /* rotatedbuffer */
 
-   drm_handle_t textures;
-   int textureSize;
+   drm_handle_t unused7; /* textures */
+   int unused8; /* textureSize */
 
-   drm_handle_t agp_buffers;
-   drmSize agp_buf_size;
+   drm_handle_t unused9; /* agp_buffers */
+   drmSize unused10; /* agp_buf_size */
 
    int deviceID;
    int width;
@@ -40,20 +40,10 @@ typedef struct _I830DRIRec {
    int cpp;
    int bitsPerPixel;
 
-   int fbOffset;
-   int fbStride;
-
-   int backOffset;
-   int backPitch;
-
-   int depthOffset;
-   int depthPitch;
-
-   int rotatedOffset;
-   int rotatedPitch;
-
-   int logTextureGranularity;
-   int textureOffset;
+   int unused11[8]; /* was front/back/depth/rotated offset/pitch */
+   
+   int unused12; /* logTextureGranularity */
+   int unused13; /* textureOffset */
 
    int irq;
    int sarea_priv_offset;
diff --git a/src/mesa/drivers/dri/i915/server/intel.h b/src/mesa/drivers/dri/i915/server/intel.h
index d7858a20c8d..6ea72499c1c 100644
--- a/src/mesa/drivers/dri/i915/server/intel.h
+++ b/src/mesa/drivers/dri/i915/server/intel.h
@@ -75,6 +75,9 @@
 
 #define I830_GMCH_CTRL		0x52
 
+#define I830_GMCH_MEM_MASK      0x1
+#define I830_GMCH_MEM_64M       0x1
+#define I830_GMCH_MEM_128M      0
 
 #define I830_GMCH_GMS_MASK			0x70
 #define I830_GMCH_GMS_DISABLED		0x00
@@ -141,7 +144,7 @@ typedef struct _I830Rec {
    unsigned char *MMIOBase;
    unsigned char *FbBase;
    int cpp;
-
+   uint32_t aper_size;
    unsigned int bios_version;
 
    /* These are set in PreInit and never changed. */
diff --git a/src/mesa/drivers/dri/i915/server/intel_dri.c b/src/mesa/drivers/dri/i915/server/intel_dri.c
index b6946b75d20..e49c4214ad4 100644
--- a/src/mesa/drivers/dri/i915/server/intel_dri.c
+++ b/src/mesa/drivers/dri/i915/server/intel_dri.c
@@ -292,15 +292,43 @@ static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
 
 static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
 {
-  struct pci_device host_bridge;
+  struct pci_device host_bridge, ig_dev;
   uint32_t gmch_ctrl;
   int memsize = 0;
   int range;
-
+  uint32_t aper_size;
+  uint32_t membase2 = 0;
+      
   memset(&host_bridge, 0, sizeof(host_bridge));
+  memset(&ig_dev, 0, sizeof(ig_dev));
+
+  ig_dev.dev = 2;
 
   pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
-  
+
+  if (IS_I830(pI830) || IS_845G(pI830)) {
+    if ((gmch_ctrl & I830_GMCH_MEM_MASK) == I830_GMCH_MEM_128M) {
+      aper_size = 0x80000000;
+    } else {
+      aper_size = 0x40000000;
+    }
+  } else {
+    if (IS_I9XX(pI830)) {
+      int ret;
+      ret = pci_device_cfg_read_u32(&ig_dev, &membase2, 0x18);
+      if (membase2 & 0x08000000)
+	aper_size = 0x8000000;
+      else
+	aper_size = 0x10000000;
+
+      fprintf(stderr,"aper size is %08X %08x %d\n", aper_size, membase2, ret);
+    } else
+      aper_size = 0x8000000;
+  }
+
+  pI830->aper_size = aper_size;
+
+
   /* We need to reduce the stolen size, by the GTT and the popup.
    * The GTT varying according the the FbMapSize and the popup is 4KB */
   range = (ctx->shared.fbSize / (1024*1024)) + 4;
@@ -577,7 +605,8 @@ I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
     fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
     return FALSE;
   }
-  
+
+#if 0  
   memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
   pI830->TexMem.Key = -1;
 
@@ -588,6 +617,7 @@ I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
     fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
     return FALSE;
   }
+#endif
 
   return TRUE;
 }
@@ -605,12 +635,29 @@ I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
     return FALSE;
   if (!BindAgpRange(ctx, &pI830->ContextMem))
     return FALSE;
+#if 0
   if (!BindAgpRange(ctx, &pI830->TexMem))
     return FALSE;
-
+#endif
   return TRUE;
 }
 
+static void SetupDRIMM(const DRIDriverContext *ctx, I830Rec *pI830)
+{
+  unsigned long aperEnd = ROUND_DOWN_TO(pI830->aper_size, GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
+  unsigned long aperStart = ROUND_TO(pI830->aper_size - KB(32768), GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
+
+  fprintf(stderr, "aper size is %08X\n", ctx->shared.fbSize);
+  if (drmMMInit(ctx->drmFD, aperStart, aperEnd - aperStart, DRM_BO_MEM_TT)) {
+      fprintf(stderr,
+	      "DRM MM Initialization Failed\n");
+  } else {
+    fprintf(stderr,
+	    "DRM MM Initialized at offset 0x%lx length %d page\n", aperStart, aperEnd-aperStart);
+  }
+
+}
+
 static Bool
 I830CleanupDma(const DRIDriverContext *ctx)
 {
@@ -810,6 +857,7 @@ I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sar
    fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
               sarea->depth_handle);
 
+#if 0
    if (drmAddMap(ctx->drmFD,
 		 (drm_handle_t)sarea->tex_offset,
 		 sarea->tex_size, DRM_AGP, 0,
@@ -820,7 +868,7 @@ I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sar
    }
    fprintf(stderr, "[drm] textures = 0x%08x\n",
 	      sarea->tex_handle);
-
+#endif
    return TRUE;
 }
 
@@ -848,29 +896,6 @@ I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sa
    }
 }
 
-static void
-I830InitTextureHeap(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   /* Start up the simple memory manager for agp space */
-   drmI830MemInitHeap drmHeap;
-   drmHeap.region = I830_MEM_REGION_AGP;
-   drmHeap.start  = 0;
-   drmHeap.size   = sarea->tex_size;
-      
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT_HEAP,
-			  &drmHeap, sizeof(drmHeap))) {
-      fprintf(stderr,
-		    "[drm] Failed to initialized agp heap manager\n");
-   } else {
-      fprintf(stderr,
-		    "[drm] Initialized kernel agp heap manager, %d\n",
-		    sarea->tex_size);
-
-      I830SetParam(ctx, I830_SETPARAM_TEX_LRU_LOG_GRANULARITY, 
-		      sarea->log_tex_granularity);
-   }
-}
-
 static Bool
 I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
 {
@@ -892,7 +917,7 @@ I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
    /* init to zero to be safe */
 
   I830DRIMapScreenRegions(ctx, pI830, sarea);
-  I830InitTextureHeap(ctx, pI830, sarea);
+  SetupDRIMM(ctx, pI830);
 
    if (ctx->pciDevice != PCI_CHIP_845_G &&
        ctx->pciDevice != PCI_CHIP_I830_M) {
@@ -1084,6 +1109,10 @@ I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
      return FALSE;
    }
 
+   pSAREAPriv->rotated_offset = -1;
+   pSAREAPriv->rotated_size = 0;
+   pSAREAPriv->rotated_pitch = ctx->shared.virtualWidth;
+
    pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
    pSAREAPriv->front_size = pI830->FrontBuffer.Size;
    pSAREAPriv->width = ctx->shared.virtualWidth;
@@ -1095,8 +1124,10 @@ I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
    pSAREAPriv->back_size = pI830->BackBuffer.Size;
    pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
    pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
+#if 0
    pSAREAPriv->tex_offset = pI830->TexMem.Start;
    pSAREAPriv->tex_size = pI830->TexMem.Size;
+#endif
    pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
 
    ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
@@ -1108,14 +1139,6 @@ I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
    pI830DRI->height = ctx->shared.virtualHeight;
    pI830DRI->mem = ctx->shared.fbSize;
    pI830DRI->cpp = ctx->cpp;
-   pI830DRI->backOffset = pI830->BackBuffer.Start;
-   pI830DRI->backPitch = pI830->BackBuffer.Pitch; 
-
-   pI830DRI->depthOffset = pI830->DepthBuffer.Start;
-   pI830DRI->depthPitch = pI830->DepthBuffer.Pitch; 
-
-   pI830DRI->fbOffset = pI830->FrontBuffer.Start;
-   pI830DRI->fbStride = pI830->FrontBuffer.Pitch;
 
    pI830DRI->bitsPerPixel = ctx->bpp;
    pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
diff --git a/src/mesa/drivers/dri/i915tex/Makefile b/src/mesa/drivers/dri/i915tex/Makefile
deleted file mode 100644
index b218929dce7..00000000000
--- a/src/mesa/drivers/dri/i915tex/Makefile
+++ /dev/null
@@ -1,70 +0,0 @@
-
-TOP = ../../../../..
-include $(TOP)/configs/current
-
-LIBNAME = i915tex_dri.so
-
-MINIGLX_SOURCES = server/intel_dri.c
-
-DRIVER_SOURCES = \
-	i830_context.c \
-	i830_metaops.c \
-	i830_state.c \
-	i830_texblend.c \
-	i830_tex.c \
-	i830_texstate.c \
-	i830_vtbl.c \
-	intel_render.c \
-	intel_regions.c \
-	intel_buffer_objects.c \
-	intel_batchbuffer.c \
-	intel_mipmap_tree.c \
-	i915_tex_layout.c \
-	intel_tex_layout.c \
-	intel_tex_image.c \
-	intel_tex_subimage.c \
-	intel_tex_copy.c \
-	intel_tex_validate.c \
-	intel_tex_format.c \
-	intel_tex.c \
-	intel_pixel.c \
-	intel_pixel_copy.c \
-	intel_pixel_read.c \
-	intel_pixel_draw.c \
-	intel_buffers.c \
-	intel_blit.c \
-	i915_tex.c \
-	i915_texstate.c \
-	i915_context.c \
-	i915_debug.c \
-	i915_fragprog.c \
-	i915_metaops.c \
-	i915_program.c \
-	i915_state.c \
-	i915_vtbl.c \
-	intel_context.c \
-	intel_ioctl.c \
-	intel_rotate.c \
-	intel_screen.c \
-	intel_span.c \
-	intel_state.c \
-	intel_tris.c \
-	intel_fbo.c \
-	intel_depthstencil.c \
-	intel_batchpool.c
-
-C_SOURCES = \
-	$(COMMON_SOURCES) \
-	$(COMMON_BM_SOURCES) \
-	$(DRIVER_SOURCES)
-
-ASM_SOURCES = 
-
-DRIVER_DEFINES = -I../intel $(shell pkg-config libdrm --atleast-version=2.3.1 \
-				&& echo "-DDRM_VBLANK_FLIP=DRM_VBLANK_FLIP")
-
-include ../Makefile.template
-
-intel_tex_layout.o: ../intel/intel_tex_layout.c
-
-symlinks:
diff --git a/src/mesa/drivers/dri/i915tex/i830_context.c b/src/mesa/drivers/dri/i915tex/i830_context.c
deleted file mode 100644
index 2ff8621c42f..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_context.c
+++ /dev/null
@@ -1,104 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "i830_context.h"
-#include "imports.h"
-#include "texmem.h"
-#include "intel_tex.h"
-#include "tnl/tnl.h"
-#include "tnl/t_vertex.h"
-#include "tnl/t_context.h"
-#include "utils.h"
-
-/***************************************
- * Mesa's Driver Functions
- ***************************************/
-
-static const struct dri_extension i830_extensions[] = {
-   {"GL_ARB_texture_env_crossbar", NULL},
-   {NULL, NULL}
-};
-
-
-static void
-i830InitDriverFunctions(struct dd_function_table *functions)
-{
-   intelInitDriverFunctions(functions);
-   i830InitStateFuncs(functions);
-   i830InitTextureFuncs(functions);
-}
-
-
-GLboolean
-i830CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
-                  void *sharedContextPrivate)
-{
-   struct dd_function_table functions;
-   struct i830_context *i830 = CALLOC_STRUCT(i830_context);
-   struct intel_context *intel = &i830->intel;
-   GLcontext *ctx = &intel->ctx;
-   if (!i830)
-      return GL_FALSE;
-
-   i830InitVtbl(i830);
-   i830InitDriverFunctions(&functions);
-
-   if (!intelInitContext(intel, mesaVis, driContextPriv,
-                         sharedContextPrivate, &functions)) {
-      FREE(i830);
-      return GL_FALSE;
-   }
-
-   intel->ctx.Const.MaxTextureUnits = I830_TEX_UNITS;
-   intel->ctx.Const.MaxTextureImageUnits = I830_TEX_UNITS;
-   intel->ctx.Const.MaxTextureCoordUnits = I830_TEX_UNITS;
-
-   /* Advertise the full hardware capabilities.  The new memory
-    * manager should cope much better with overload situations:
-    */
-   ctx->Const.MaxTextureLevels = 12;
-   ctx->Const.Max3DTextureLevels = 9;
-   ctx->Const.MaxCubeTextureLevels = 11;
-   ctx->Const.MaxTextureRectSize = (1 << 11);
-   ctx->Const.MaxTextureUnits = I830_TEX_UNITS;
-
-   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
-                      18 * sizeof(GLfloat));
-
-   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
-
-   driInitExtensions(ctx, i830_extensions, GL_FALSE);
-
-   i830InitState(i830);
-   i830InitMetaFuncs(i830);
-
-   _tnl_allow_vertex_fog(ctx, 1);
-   _tnl_allow_pixel_fog(ctx, 0);
-
-   return GL_TRUE;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_context.h b/src/mesa/drivers/dri/i915tex/i830_context.h
deleted file mode 100644
index 3d754103c0a..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_context.h
+++ /dev/null
@@ -1,213 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef I830CONTEXT_INC
-#define I830CONTEXT_INC
-
-#include "intel_context.h"
-
-#define I830_FALLBACK_TEXTURE		 0x1000
-#define I830_FALLBACK_COLORMASK		 0x2000
-#define I830_FALLBACK_STENCIL		 0x4000
-#define I830_FALLBACK_STIPPLE		 0x8000
-#define I830_FALLBACK_LOGICOP		 0x10000
-
-#define I830_UPLOAD_CTX              0x1
-#define I830_UPLOAD_BUFFERS          0x2
-#define I830_UPLOAD_STIPPLE          0x4
-#define I830_UPLOAD_INVARIENT        0x8
-#define I830_UPLOAD_TEX(i)           (0x10<<(i))
-#define I830_UPLOAD_TEXBLEND(i)      (0x100<<(i))
-#define I830_UPLOAD_TEX_ALL          (0x0f0)
-#define I830_UPLOAD_TEXBLEND_ALL     (0xf00)
-
-/* State structure offsets - these will probably disappear.
- */
-#define I830_DESTREG_CBUFADDR0 0
-#define I830_DESTREG_CBUFADDR1 1
-#define I830_DESTREG_DBUFADDR0 2
-#define I830_DESTREG_DBUFADDR1 3
-#define I830_DESTREG_DV0 4
-#define I830_DESTREG_DV1 5
-#define I830_DESTREG_SENABLE 6
-#define I830_DESTREG_SR0 7
-#define I830_DESTREG_SR1 8
-#define I830_DESTREG_SR2 9
-#define I830_DEST_SETUP_SIZE 10
-
-#define I830_CTXREG_STATE1		0
-#define I830_CTXREG_STATE2		1
-#define I830_CTXREG_STATE3		2
-#define I830_CTXREG_STATE4		3
-#define I830_CTXREG_STATE5		4
-#define I830_CTXREG_IALPHAB		5
-#define I830_CTXREG_STENCILTST		6
-#define I830_CTXREG_ENABLES_1		7
-#define I830_CTXREG_ENABLES_2		8
-#define I830_CTXREG_AA			9
-#define I830_CTXREG_FOGCOLOR		10
-#define I830_CTXREG_BLENDCOLOR0		11
-#define I830_CTXREG_BLENDCOLOR1		12
-#define I830_CTXREG_VF			13
-#define I830_CTXREG_VF2			14
-#define I830_CTXREG_MCSB0		15
-#define I830_CTXREG_MCSB1		16
-#define I830_CTX_SETUP_SIZE		17
-
-#define I830_STPREG_ST0        0
-#define I830_STPREG_ST1        1
-#define I830_STP_SETUP_SIZE    2
-
-#define I830_TEXREG_TM0LI      0        /* load immediate 2 texture map n */
-#define I830_TEXREG_TM0S1      1
-#define I830_TEXREG_TM0S2      2
-#define I830_TEXREG_TM0S3      3
-#define I830_TEXREG_TM0S4      4
-#define I830_TEXREG_MCS	       5        /* _3DSTATE_MAP_COORD_SETS */
-#define I830_TEXREG_CUBE       6        /* _3DSTATE_MAP_SUBE */
-#define I830_TEX_SETUP_SIZE    7
-
-#define I830_TEXBLEND_SIZE	12      /* (4 args + op) * 2 + COLOR_FACTOR */
-
-struct i830_texture_object
-{
-   struct intel_texture_object intel;
-   GLuint Setup[I830_TEX_SETUP_SIZE];
-};
-
-#define I830_TEX_UNITS 4
-
-struct i830_hw_state
-{
-   GLuint Ctx[I830_CTX_SETUP_SIZE];
-   GLuint Buffer[I830_DEST_SETUP_SIZE];
-   GLuint Stipple[I830_STP_SETUP_SIZE];
-   GLuint Tex[I830_TEX_UNITS][I830_TEX_SETUP_SIZE];
-   GLuint TexBlend[I830_TEX_UNITS][I830_TEXBLEND_SIZE];
-   GLuint TexBlendWordsUsed[I830_TEX_UNITS];
-
-   struct intel_region *draw_region;
-   struct intel_region *depth_region;
-
-   /* Regions aren't actually that appropriate here as the memory may
-    * be from a PBO or FBO.  Just use the buffer id.  Will have to do
-    * this for draw and depth for FBO's...
-    */
-   struct _DriBufferObject *tex_buffer[I830_TEX_UNITS];
-   GLuint tex_offset[I830_TEX_UNITS];
-
-   GLuint emitted;              /* I810_UPLOAD_* */
-   GLuint active;
-};
-
-struct i830_context
-{
-   struct intel_context intel;
-
-   GLuint lodbias_tm0s3[MAX_TEXTURE_UNITS];
-     DECLARE_RENDERINPUTS(last_index_bitset);
-
-   struct i830_hw_state meta, initial, state, *current;
-};
-
-
-
-
-#define I830_STATECHANGE(i830, flag)				\
-do {								\
-   INTEL_FIREVERTICES( &i830->intel );				\
-   i830->state.emitted &= ~flag;					\
-} while (0)
-
-#define I830_ACTIVESTATE(i830, flag, mode)	\
-do {						\
-   INTEL_FIREVERTICES( &i830->intel );		\
-   if (mode)					\
-      i830->state.active |= flag;		\
-   else						\
-      i830->state.active &= ~flag;		\
-} while (0)
-
-/* i830_vtbl.c
- */
-extern void i830InitVtbl(struct i830_context *i830);
-
-extern void
-i830_state_draw_region(struct intel_context *intel,
-                       struct i830_hw_state *state,
-                       struct intel_region *color_region,
-                       struct intel_region *depth_region);
-/* i830_context.c
- */
-extern GLboolean
-i830CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
-                  void *sharedContextPrivate);
-
-/* i830_tex.c, i830_texstate.c
- */
-extern void i830UpdateTextureState(struct intel_context *intel);
-
-extern void i830InitTextureFuncs(struct dd_function_table *functions);
-
-/* i830_texblend.c
- */
-extern GLuint i830SetTexEnvCombine(struct i830_context *i830,
-                                   const struct gl_tex_env_combine_state
-                                   *combine, GLint blendUnit, GLuint texel_op,
-                                   GLuint * state, const GLfloat * factor);
-
-extern void i830EmitTextureBlend(struct i830_context *i830);
-
-
-/* i830_state.c
- */
-extern void i830InitStateFuncs(struct dd_function_table *functions);
-
-extern void i830EmitState(struct i830_context *i830);
-
-extern void i830InitState(struct i830_context *i830);
-
-/* i830_metaops.c
- */
-extern void i830InitMetaFuncs(struct i830_context *i830);
-
-extern void
-i830RotateWindow(struct intel_context *intel, __DRIdrawablePrivate * dPriv,
-                 GLuint srcBuf);
-
-/*======================================================================
- * Inline conversion functions.  These are better-typed than the
- * macros used previously:
- */
-static INLINE struct i830_context *
-i830_context(GLcontext * ctx)
-{
-   return (struct i830_context *) ctx;
-}
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/i830_metaops.c b/src/mesa/drivers/dri/i915tex/i830_metaops.c
deleted file mode 100644
index f76646d89db..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_metaops.c
+++ /dev/null
@@ -1,457 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "utils.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-#include "intel_ioctl.h"
-#include "intel_regions.h"
-
-#include "i830_context.h"
-#include "i830_reg.h"
-
-/* A large amount of state doesn't need to be uploaded.
- */
-#define ACTIVE (I830_UPLOAD_INVARIENT |         \
-		I830_UPLOAD_CTX |		\
-		I830_UPLOAD_BUFFERS |		\
-		I830_UPLOAD_STIPPLE |		\
-		I830_UPLOAD_TEXBLEND(0) |	\
-		I830_UPLOAD_TEX(0))
-
-
-#define SET_STATE( i830, STATE )		\
-do {						\
-   i830->current->emitted &= ~ACTIVE;			\
-   i830->current = &i830->STATE;		\
-   i830->current->emitted &= ~ACTIVE;			\
-} while (0)
-
-
-static void
-set_no_stencil_write(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_STENCIL_WRITE;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_STENCIL_WRITE;
-
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-static void
-set_no_depth_write(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DEPTH_WRITE;
-
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-/* Set depth unit to replace.
- */
-static void
-set_depth_replace(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
-    * ctx->Driver.DepthMask( ctx, GL_TRUE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DEPTH_WRITE;
-
-   /* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
-    */
-   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
-   i830->meta.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
-                                          DEPTH_TEST_FUNC
-                                          (COMPAREFUNC_ALWAYS));
-
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-
-/* Set stencil unit to replace always with the reference value.
- */
-static void
-set_stencil_replace(struct intel_context *intel,
-                    GLuint s_mask, GLuint s_clear)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
-    */
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-
-   /* ctx->Driver.StencilMask( ctx, s_mask )
-    */
-   i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
-   i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-                                          STENCIL_WRITE_MASK((s_mask &
-                                                              0xff)));
-
-   /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
-    */
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
-      (ENABLE_STENCIL_PARMS |
-       STENCIL_FAIL_OP(STENCILOP_REPLACE) |
-       STENCIL_PASS_DEPTH_FAIL_OP(STENCILOP_REPLACE) |
-       STENCIL_PASS_DEPTH_PASS_OP(STENCILOP_REPLACE));
-
-   /* ctx->Driver.StencilFunc( ctx, GL_ALWAYS, s_clear, ~0 )
-    */
-   i830->meta.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
-   i830->meta.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-                                          STENCIL_TEST_MASK(0xff));
-
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-                                               ENABLE_STENCIL_TEST_FUNC_MASK);
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] |=
-      (ENABLE_STENCIL_REF_VALUE |
-       ENABLE_STENCIL_TEST_FUNC |
-       STENCIL_REF_VALUE((s_clear & 0xff)) |
-       STENCIL_TEST_FUNC(COMPAREFUNC_ALWAYS));
-
-
-
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-
-static void
-set_color_mask(struct intel_context *intel, GLboolean state)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   const GLuint mask = ((1 << WRITEMASK_RED_SHIFT) |
-                        (1 << WRITEMASK_GREEN_SHIFT) |
-                        (1 << WRITEMASK_BLUE_SHIFT) |
-                        (1 << WRITEMASK_ALPHA_SHIFT));
-
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] &= ~mask;
-
-   if (state) {
-      i830->meta.Ctx[I830_CTXREG_ENABLES_2] |=
-         (i830->state.Ctx[I830_CTXREG_ENABLES_2] & mask);
-   }
-
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-/* Installs a one-stage passthrough texture blend pipeline.  Is there
- * more that can be done to turn off texturing?
- */
-static void
-set_no_texture(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   static const struct gl_tex_env_combine_state comb = {
-      GL_NONE, GL_NONE,
-      {GL_TEXTURE, 0, 0,}, {GL_TEXTURE, 0, 0,},
-      {GL_SRC_COLOR, 0, 0}, {GL_SRC_ALPHA, 0, 0},
-      0, 0, 0, 0
-   };
-
-   i830->meta.TexBlendWordsUsed[0] =
-      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
-                           i830->meta.TexBlend[0], NULL);
-
-   i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
-   i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
-}
-
-/* Set up a single element blend stage for 'replace' texturing with no
- * funny ops.
- */
-static void
-set_texture_blend_replace(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   static const struct gl_tex_env_combine_state comb = {
-      GL_REPLACE, GL_REPLACE,
-      {GL_TEXTURE, GL_TEXTURE, GL_TEXTURE,}, {GL_TEXTURE, GL_TEXTURE,
-                                              GL_TEXTURE,},
-      {GL_SRC_COLOR, GL_SRC_COLOR, GL_SRC_COLOR}, {GL_SRC_ALPHA, GL_SRC_ALPHA,
-                                                   GL_SRC_ALPHA},
-      0, 0, 1, 1
-   };
-
-   i830->meta.TexBlendWordsUsed[0] =
-      i830SetTexEnvCombine(i830, &comb, 0, TEXBLENDARG_TEXEL0,
-                           i830->meta.TexBlend[0], NULL);
-
-   i830->meta.TexBlend[0][0] |= TEXOP_LAST_STAGE;
-   i830->meta.emitted &= ~I830_UPLOAD_TEXBLEND(0);
-
-/*    fprintf(stderr, "%s: TexBlendWordsUsed[0]: %d\n",  */
-/* 	   __FUNCTION__, i830->meta.TexBlendWordsUsed[0]); */
-}
-
-
-
-/* Set up an arbitary piece of memory as a rectangular texture
- * (including the front or back buffer).
- */
-static GLboolean
-set_tex_rect_source(struct intel_context *intel,
-                    struct _DriBufferObject *buffer,
-                    GLuint offset,
-                    GLuint pitch, GLuint height, GLenum format, GLenum type)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   GLuint *setup = i830->meta.Tex[0];
-   GLint numLevels = 1;
-   GLuint textureFormat;
-   GLuint cpp;
-
-   /* A full implementation of this would do the upload through
-    * glTexImage2d, and get all the conversion operations at that
-    * point.  We are restricted, but still at least have access to the
-    * fragment program swizzle.
-    */
-   switch (format) {
-   case GL_BGRA:
-      switch (type) {
-      case GL_UNSIGNED_INT_8_8_8_8_REV:
-      case GL_UNSIGNED_BYTE:
-         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
-         cpp = 4;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_RGBA:
-      switch (type) {
-      case GL_UNSIGNED_INT_8_8_8_8_REV:
-      case GL_UNSIGNED_BYTE:
-         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
-         cpp = 4;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_BGR:
-      switch (type) {
-      case GL_UNSIGNED_SHORT_5_6_5_REV:
-         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
-         cpp = 2;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_RGB:
-      switch (type) {
-      case GL_UNSIGNED_SHORT_5_6_5:
-         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
-         cpp = 2;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-
-   default:
-      return GL_FALSE;
-   }
-
-   i830->meta.tex_buffer[0] = buffer;
-   i830->meta.tex_offset[0] = offset;
-
-   setup[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
-                               (LOAD_TEXTURE_MAP0 << 0) | 4);
-   setup[I830_TEXREG_TM0S1] = (((height - 1) << TM0S1_HEIGHT_SHIFT) |
-                               ((pitch - 1) << TM0S1_WIDTH_SHIFT) |
-                               textureFormat);
-   setup[I830_TEXREG_TM0S2] =
-      (((((pitch * cpp) / 4) -
-         1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
-
-   setup[I830_TEXREG_TM0S3] =
-      ((((numLevels -
-          1) *
-         4) << TM0S3_MIN_MIP_SHIFT) | (FILTER_NEAREST <<
-                                       TM0S3_MIN_FILTER_SHIFT) |
-       (MIPFILTER_NONE << TM0S3_MIP_FILTER_SHIFT) | (FILTER_NEAREST <<
-                                                     TM0S3_MAG_FILTER_SHIFT));
-
-   setup[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(0));
-
-   setup[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-                             MAP_UNIT(0) |
-                             ENABLE_TEXCOORD_PARAMS |
-                             TEXCOORDS_ARE_IN_TEXELUNITS |
-                             TEXCOORDTYPE_CARTESIAN |
-                             ENABLE_ADDR_V_CNTL |
-                             TEXCOORD_ADDR_V_MODE(TEXCOORDMODE_WRAP) |
-                             ENABLE_ADDR_U_CNTL |
-                             TEXCOORD_ADDR_U_MODE(TEXCOORDMODE_WRAP));
-
-   i830->meta.emitted &= ~I830_UPLOAD_TEX(0);
-   return GL_TRUE;
-}
-
-
-static void
-set_vertex_format(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   i830->meta.Ctx[I830_CTXREG_VF] = (_3DSTATE_VFT0_CMD |
-                                     VFT0_TEX_COUNT(1) |
-                                     VFT0_DIFFUSE | VFT0_XYZ);
-   i830->meta.Ctx[I830_CTXREG_VF2] = (_3DSTATE_VFT1_CMD |
-                                      VFT1_TEX0_FMT(TEXCOORDFMT_2D) |
-                                      VFT1_TEX1_FMT(TEXCOORDFMT_2D) |
-                                      VFT1_TEX2_FMT(TEXCOORDFMT_2D) |
-                                      VFT1_TEX3_FMT(TEXCOORDFMT_2D));
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-}
-
-
-static void
-meta_import_pixel_state(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   i830->meta.Ctx[I830_CTXREG_STATE1] = i830->state.Ctx[I830_CTXREG_STATE1];
-   i830->meta.Ctx[I830_CTXREG_STATE2] = i830->state.Ctx[I830_CTXREG_STATE2];
-   i830->meta.Ctx[I830_CTXREG_STATE3] = i830->state.Ctx[I830_CTXREG_STATE3];
-   i830->meta.Ctx[I830_CTXREG_STATE4] = i830->state.Ctx[I830_CTXREG_STATE4];
-   i830->meta.Ctx[I830_CTXREG_STATE5] = i830->state.Ctx[I830_CTXREG_STATE5];
-   i830->meta.Ctx[I830_CTXREG_IALPHAB] = i830->state.Ctx[I830_CTXREG_IALPHAB];
-   i830->meta.Ctx[I830_CTXREG_STENCILTST] =
-      i830->state.Ctx[I830_CTXREG_STENCILTST];
-   i830->meta.Ctx[I830_CTXREG_ENABLES_1] =
-      i830->state.Ctx[I830_CTXREG_ENABLES_1];
-   i830->meta.Ctx[I830_CTXREG_ENABLES_2] =
-      i830->state.Ctx[I830_CTXREG_ENABLES_2];
-   i830->meta.Ctx[I830_CTXREG_AA] = i830->state.Ctx[I830_CTXREG_AA];
-   i830->meta.Ctx[I830_CTXREG_FOGCOLOR] =
-      i830->state.Ctx[I830_CTXREG_FOGCOLOR];
-   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR0] =
-      i830->state.Ctx[I830_CTXREG_BLENDCOLOR0];
-   i830->meta.Ctx[I830_CTXREG_BLENDCOLOR1] =
-      i830->state.Ctx[I830_CTXREG_BLENDCOLOR1];
-   i830->meta.Ctx[I830_CTXREG_MCSB0] = i830->state.Ctx[I830_CTXREG_MCSB0];
-   i830->meta.Ctx[I830_CTXREG_MCSB1] = i830->state.Ctx[I830_CTXREG_MCSB1];
-
-
-   i830->meta.Ctx[I830_CTXREG_STATE3] &= ~CULLMODE_MASK;
-   i830->meta.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
-   i830->meta.emitted &= ~I830_UPLOAD_CTX;
-
-
-   i830->meta.Buffer[I830_DESTREG_SENABLE] =
-      i830->state.Buffer[I830_DESTREG_SENABLE];
-   i830->meta.Buffer[I830_DESTREG_SR1] = i830->state.Buffer[I830_DESTREG_SR1];
-   i830->meta.Buffer[I830_DESTREG_SR2] = i830->state.Buffer[I830_DESTREG_SR2];
-   i830->meta.emitted &= ~I830_UPLOAD_BUFFERS;
-}
-
-
-
-/* Select between front and back draw buffers.
- */
-static void
-meta_draw_region(struct intel_context *intel,
-                 struct intel_region *color_region,
-                 struct intel_region *depth_region)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-
-   i830_state_draw_region(intel, &i830->meta, color_region, depth_region);
-}
-
-
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void
-install_meta_state(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   memcpy(&i830->meta, &i830->initial, sizeof(i830->meta));
-
-   i830->meta.active = ACTIVE;
-   i830->meta.emitted = 0;
-
-   SET_STATE(i830, meta);
-   set_vertex_format(intel);
-   set_no_texture(intel);
-}
-
-static void
-leave_meta_state(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   intel_region_release(&i830->meta.draw_region);
-   intel_region_release(&i830->meta.depth_region);
-/*    intel_region_release(intel, &i830->meta.tex_region[0]); */
-   SET_STATE(i830, state);
-}
-
-
-
-void
-i830InitMetaFuncs(struct i830_context *i830)
-{
-   i830->intel.vtbl.install_meta_state = install_meta_state;
-   i830->intel.vtbl.leave_meta_state = leave_meta_state;
-   i830->intel.vtbl.meta_no_depth_write = set_no_depth_write;
-   i830->intel.vtbl.meta_no_stencil_write = set_no_stencil_write;
-   i830->intel.vtbl.meta_stencil_replace = set_stencil_replace;
-   i830->intel.vtbl.meta_depth_replace = set_depth_replace;
-   i830->intel.vtbl.meta_color_mask = set_color_mask;
-   i830->intel.vtbl.meta_no_texture = set_no_texture;
-   i830->intel.vtbl.meta_texture_blend_replace = set_texture_blend_replace;
-   i830->intel.vtbl.meta_tex_rect_source = set_tex_rect_source;
-   i830->intel.vtbl.meta_draw_region = meta_draw_region;
-   i830->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_reg.h b/src/mesa/drivers/dri/i915tex/i830_reg.h
deleted file mode 100644
index 41280bca7ce..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_reg.h
+++ /dev/null
@@ -1,642 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef _I830_REG_H_
-#define _I830_REG_H_
-
-
-#include "intel_reg.h"
-
-#define I830_SET_FIELD( var, mask, value ) (var &= ~(mask), var |= value)
-
-#define _3DSTATE_AA_CMD			(CMD_3D | (0x06<<24))
-#define AA_LINE_ECAAR_WIDTH_ENABLE	(1<<16)
-#define AA_LINE_ECAAR_WIDTH_0_5 	0
-#define AA_LINE_ECAAR_WIDTH_1_0		(1<<14)
-#define AA_LINE_ECAAR_WIDTH_2_0 	(2<<14)
-#define AA_LINE_ECAAR_WIDTH_4_0 	(3<<14)
-#define AA_LINE_REGION_WIDTH_ENABLE	(1<<8)
-#define AA_LINE_REGION_WIDTH_0_5	0
-#define AA_LINE_REGION_WIDTH_1_0	(1<<6)
-#define AA_LINE_REGION_WIDTH_2_0	(2<<6)
-#define AA_LINE_REGION_WIDTH_4_0	(3<<6)
-#define AA_LINE_ENABLE			((1<<1) | 1)
-#define AA_LINE_DISABLE			(1<<1)
-
-#define _3DSTATE_BUF_INFO_CMD	(CMD_3D | (0x1d<<24) | (0x8e<<16) | 1)
-/* Dword 1 */
-#define BUF_3D_ID_COLOR_BACK	(0x3<<24)
-#define BUF_3D_ID_DEPTH 	(0x7<<24)
-#define BUF_3D_USE_FENCE	(1<<23)
-#define BUF_3D_TILED_SURFACE	(1<<22)
-#define BUF_3D_TILE_WALK_X	0
-#define BUF_3D_TILE_WALK_Y	(1<<21)
-#define BUF_3D_PITCH(x)         (((x)/4)<<2)
-/* Dword 2 */
-#define BUF_3D_ADDR(x)		((x) & ~0x3)
-
-
-#define _3DSTATE_COLOR_FACTOR_CMD	(CMD_3D | (0x1d<<24) | (0x1<<16))
-
-#define _3DSTATE_COLOR_FACTOR_N_CMD(stage)	(CMD_3D | (0x1d<<24) | \
-					         ((0x90+(stage))<<16))
-
-#define _3DSTATE_CONST_BLEND_COLOR_CMD	(CMD_3D | (0x1d<<24) | (0x88<<16))
-
-#define _3DSTATE_DFLT_DIFFUSE_CMD	(CMD_3D | (0x1d<<24) | (0x99<<16))
-
-#define _3DSTATE_DFLT_SPEC_CMD		(CMD_3D | (0x1d<<24) | (0x9a<<16))
-
-#define _3DSTATE_DFLT_Z_CMD		(CMD_3D | (0x1d<<24) | (0x98<<16))
-
-
-#define _3DSTATE_DST_BUF_VARS_CMD	(CMD_3D | (0x1d<<24) | (0x85<<16))
-/* Dword 1 */
-#define DSTORG_HORT_BIAS(x)		((x)<<20)
-#define DSTORG_VERT_BIAS(x)		((x)<<16)
-#define COLOR_4_2_2_CHNL_WRT_ALL	0
-#define COLOR_4_2_2_CHNL_WRT_Y		(1<<12)
-#define COLOR_4_2_2_CHNL_WRT_CR		(2<<12)
-#define COLOR_4_2_2_CHNL_WRT_CB		(3<<12)
-#define COLOR_4_2_2_CHNL_WRT_CRCB	(4<<12)
-#define COLR_BUF_8BIT			0
-#define COLR_BUF_RGB555 		(1<<8)
-#define COLR_BUF_RGB565 		(2<<8)
-#define COLR_BUF_ARGB8888		(3<<8)
-#define DEPTH_IS_Z			0
-#define DEPTH_IS_W			(1<<6)
-#define DEPTH_FRMT_16_FIXED		0
-#define DEPTH_FRMT_16_FLOAT		(1<<2)
-#define DEPTH_FRMT_24_FIXED_8_OTHER	(2<<2)
-#define DEPTH_FRMT_24_FLOAT_8_OTHER	(3<<2)
-#define VERT_LINE_STRIDE_1		(1<<1)
-#define VERT_LINE_STRIDE_0		0
-#define VERT_LINE_STRIDE_OFS_1		1
-#define VERT_LINE_STRIDE_OFS_0		0
-
-
-#define _3DSTATE_DRAW_RECT_CMD		(CMD_3D|(0x1d<<24)|(0x80<<16)|3)
-/* Dword 1 */
-#define DRAW_RECT_DIS_DEPTH_OFS 	(1<<30)
-#define DRAW_DITHER_OFS_X(x)		((x)<<26)
-#define DRAW_DITHER_OFS_Y(x)		((x)<<24)
-/* Dword 2 */
-#define DRAW_YMIN(x)			((x)<<16)
-#define DRAW_XMIN(x)			(x)
-/* Dword 3 */
-#define DRAW_YMAX(x)			((x)<<16)
-#define DRAW_XMAX(x)			(x)
-/* Dword 4 */
-#define DRAW_YORG(x)			((x)<<16)
-#define DRAW_XORG(x)			(x)
-
-
-#define _3DSTATE_ENABLES_1_CMD		(CMD_3D|(0x3<<24))
-#define ENABLE_LOGIC_OP_MASK		((1<<23)|(1<<22))
-#define ENABLE_LOGIC_OP 		((1<<23)|(1<<22))
-#define DISABLE_LOGIC_OP		(1<<23)
-#define ENABLE_STENCIL_TEST		((1<<21)|(1<<20))
-#define DISABLE_STENCIL_TEST		(1<<21)
-#define ENABLE_DEPTH_BIAS		((1<<11)|(1<<10))
-#define DISABLE_DEPTH_BIAS		(1<<11)
-#define ENABLE_SPEC_ADD_MASK		((1<<9)|(1<<8))
-#define ENABLE_SPEC_ADD 		((1<<9)|(1<<8))
-#define DISABLE_SPEC_ADD		(1<<9)
-#define ENABLE_DIS_FOG_MASK		((1<<7)|(1<<6))
-#define ENABLE_FOG			((1<<7)|(1<<6))
-#define DISABLE_FOG			(1<<7)
-#define ENABLE_DIS_ALPHA_TEST_MASK	((1<<5)|(1<<4))
-#define ENABLE_ALPHA_TEST		((1<<5)|(1<<4))
-#define DISABLE_ALPHA_TEST		(1<<5)
-#define ENABLE_DIS_CBLEND_MASK		((1<<3)|(1<<2))
-#define ENABLE_COLOR_BLEND		((1<<3)|(1<<2))
-#define DISABLE_COLOR_BLEND		(1<<3)
-#define ENABLE_DIS_DEPTH_TEST_MASK	((1<<1)|1)
-#define ENABLE_DEPTH_TEST		((1<<1)|1)
-#define DISABLE_DEPTH_TEST		(1<<1)
-
-/* _3DSTATE_ENABLES_2, p138 */
-#define _3DSTATE_ENABLES_2_CMD		(CMD_3D|(0x4<<24))
-#define ENABLE_STENCIL_WRITE		((1<<21)|(1<<20))
-#define DISABLE_STENCIL_WRITE		(1<<21)
-#define ENABLE_TEX_CACHE		((1<<17)|(1<<16))
-#define DISABLE_TEX_CACHE		(1<<17)
-#define ENABLE_DITHER			((1<<9)|(1<<8))
-#define DISABLE_DITHER			(1<<9)
-#define ENABLE_COLOR_MASK		(1<<10)
-#define WRITEMASK_ALPHA			(1<<7)
-#define WRITEMASK_ALPHA_SHIFT		7
-#define WRITEMASK_RED			(1<<6)
-#define WRITEMASK_RED_SHIFT		6
-#define WRITEMASK_GREEN 		(1<<5)
-#define WRITEMASK_GREEN_SHIFT		5
-#define WRITEMASK_BLUE			(1<<4)
-#define WRITEMASK_BLUE_SHIFT		4
-#define WRITEMASK_MASK			((1<<4)|(1<<5)|(1<<6)|(1<<7))
-#define ENABLE_COLOR_WRITE		((1<<3)|(1<<2))
-#define DISABLE_COLOR_WRITE		(1<<3)
-#define ENABLE_DIS_DEPTH_WRITE_MASK	0x3
-#define ENABLE_DEPTH_WRITE		((1<<1)|1)
-#define DISABLE_DEPTH_WRITE		(1<<1)
-
-/* _3DSTATE_FOG_COLOR, p139 */
-#define _3DSTATE_FOG_COLOR_CMD		(CMD_3D|(0x15<<24))
-#define FOG_COLOR_RED(x)		((x)<<16)
-#define FOG_COLOR_GREEN(x)		((x)<<8)
-#define FOG_COLOR_BLUE(x)		(x)
-
-/* _3DSTATE_FOG_MODE, p140 */
-#define _3DSTATE_FOG_MODE_CMD		(CMD_3D|(0x1d<<24)|(0x89<<16)|2)
-/* Dword 1 */
-#define FOGFUNC_ENABLE			(1<<31)
-#define FOGFUNC_VERTEX			0
-#define FOGFUNC_PIXEL_EXP		(1<<28)
-#define FOGFUNC_PIXEL_EXP2		(2<<28)
-#define FOGFUNC_PIXEL_LINEAR		(3<<28)
-#define FOGSRC_INDEX_Z			(1<<27)
-#define FOGSRC_INDEX_W			((1<<27)|(1<<25))
-#define FOG_LINEAR_CONST		(1<<24)
-#define FOG_CONST_1(x)			((x)<<4)
-#define ENABLE_FOG_DENSITY		(1<<23)
-/* Dword 2 */
-#define FOG_CONST_2(x)			(x)
-/* Dword 3 */
-#define FOG_DENSITY(x)			(x)
-
-/* _3DSTATE_INDEPENDENT_ALPHA_BLEND, p142 */
-#define _3DSTATE_INDPT_ALPHA_BLEND_CMD	(CMD_3D|(0x0b<<24))
-#define ENABLE_INDPT_ALPHA_BLEND	((1<<23)|(1<<22))
-#define DISABLE_INDPT_ALPHA_BLEND	(1<<23)
-#define ALPHA_BLENDFUNC_MASK		0x3f0000
-#define ENABLE_ALPHA_BLENDFUNC		(1<<21)
-#define ABLENDFUNC_ADD			0
-#define ABLENDFUNC_SUB			(1<<16)
-#define ABLENDFUNC_RVSE_SUB		(2<<16)
-#define ABLENDFUNC_MIN			(3<<16)
-#define ABLENDFUNC_MAX			(4<<16)
-#define SRC_DST_ABLEND_MASK		0xfff
-#define ENABLE_SRC_ABLEND_FACTOR	(1<<11)
-#define SRC_ABLEND_FACT(x)		((x)<<6)
-#define ENABLE_DST_ABLEND_FACTOR	(1<<5)
-#define DST_ABLEND_FACT(x)		(x)
-
-
-/* _3DSTATE_MAP_BLEND_ARG, p152 */
-#define _3DSTATE_MAP_BLEND_ARG_CMD(stage)	(CMD_3D|(0x0e<<24)|((stage)<<20))
-
-#define TEXPIPE_COLOR			0
-#define TEXPIPE_ALPHA			(1<<18)
-#define TEXPIPE_KILL			(2<<18)
-#define TEXBLEND_ARG0			0
-#define TEXBLEND_ARG1			(1<<15)
-#define TEXBLEND_ARG2			(2<<15)
-#define TEXBLEND_ARG3			(3<<15)
-#define TEXBLENDARG_MODIFY_PARMS	(1<<6)
-#define TEXBLENDARG_REPLICATE_ALPHA 	(1<<5)
-#define TEXBLENDARG_INV_ARG 		(1<<4)
-#define TEXBLENDARG_ONE 		0
-#define TEXBLENDARG_FACTOR		0x01
-#define TEXBLENDARG_ACCUM		0x02
-#define TEXBLENDARG_DIFFUSE		0x03
-#define TEXBLENDARG_SPEC		0x04
-#define TEXBLENDARG_CURRENT		0x05
-#define TEXBLENDARG_TEXEL0		0x06
-#define TEXBLENDARG_TEXEL1		0x07
-#define TEXBLENDARG_TEXEL2		0x08
-#define TEXBLENDARG_TEXEL3		0x09
-#define TEXBLENDARG_FACTOR_N		0x0e
-
-/* _3DSTATE_MAP_BLEND_OP, p155 */
-#define _3DSTATE_MAP_BLEND_OP_CMD(stage)	(CMD_3D|(0x0d<<24)|((stage)<<20))
-#if 0
-#   define TEXPIPE_COLOR		0
-#   define TEXPIPE_ALPHA		(1<<18)
-#   define TEXPIPE_KILL			(2<<18)
-#endif
-#define ENABLE_TEXOUTPUT_WRT_SEL	(1<<17)
-#define TEXOP_OUTPUT_CURRENT		0
-#define TEXOP_OUTPUT_ACCUM		(1<<15)
-#define ENABLE_TEX_CNTRL_STAGE		((1<<12)|(1<<11))
-#define DISABLE_TEX_CNTRL_STAGE		(1<<12)
-#define TEXOP_SCALE_SHIFT		9
-#define TEXOP_SCALE_1X			(0 << TEXOP_SCALE_SHIFT)
-#define TEXOP_SCALE_2X			(1 << TEXOP_SCALE_SHIFT)
-#define TEXOP_SCALE_4X			(2 << TEXOP_SCALE_SHIFT)
-#define TEXOP_MODIFY_PARMS		(1<<8)
-#define TEXOP_LAST_STAGE		(1<<7)
-#define TEXBLENDOP_KILLPIXEL		0x02
-#define TEXBLENDOP_ARG1 		0x01
-#define TEXBLENDOP_ARG2 		0x02
-#define TEXBLENDOP_MODULATE		0x03
-#define TEXBLENDOP_ADD			0x06
-#define TEXBLENDOP_ADDSIGNED		0x07
-#define TEXBLENDOP_BLEND		0x08
-#define TEXBLENDOP_BLEND_AND_ADD	0x09
-#define TEXBLENDOP_SUBTRACT		0x0a
-#define TEXBLENDOP_DOT3 		0x0b
-#define TEXBLENDOP_DOT4 		0x0c
-#define TEXBLENDOP_MODULATE_AND_ADD	0x0d
-#define TEXBLENDOP_MODULATE_2X_AND_ADD	0x0e
-#define TEXBLENDOP_MODULATE_4X_AND_ADD	0x0f
-
-/* _3DSTATE_MAP_BUMP_TABLE, p160 TODO */
-/* _3DSTATE_MAP_COLOR_CHROMA_KEY, p161 TODO */
-
-#define _3DSTATE_MAP_COORD_TRANSFORM	((3<<29)|(0x1d<<24)|(0x8c<<16))
-#define DISABLE_TEX_TRANSFORM		(1<<28)
-#define TEXTURE_SET(x)			(x<<29)
-
-#define _3DSTATE_VERTEX_TRANSFORM	((3<<29)|(0x1d<<24)|(0x8b<<16))
-#define DISABLE_VIEWPORT_TRANSFORM	(1<<31)
-#define DISABLE_PERSPECTIVE_DIVIDE	(1<<29)
-
-
-/* _3DSTATE_MAP_COORD_SET_BINDINGS, p162 */
-#define _3DSTATE_MAP_COORD_SETBIND_CMD	(CMD_3D|(0x1d<<24)|(0x02<<16))
-#define TEXBIND_MASK3			((1<<15)|(1<<14)|(1<<13)|(1<<12))
-#define TEXBIND_MASK2			((1<<11)|(1<<10)|(1<<9)|(1<<8))
-#define TEXBIND_MASK1			((1<<7)|(1<<6)|(1<<5)|(1<<4))
-#define TEXBIND_MASK0			((1<<3)|(1<<2)|(1<<1)|1)
-
-#define TEXBIND_SET3(x) 		((x)<<12)
-#define TEXBIND_SET2(x) 		((x)<<8)
-#define TEXBIND_SET1(x) 		((x)<<4)
-#define TEXBIND_SET0(x) 		(x)
-
-#define TEXCOORDSRC_KEEP		0
-#define TEXCOORDSRC_DEFAULT		0x01
-#define TEXCOORDSRC_VTXSET_0		0x08
-#define TEXCOORDSRC_VTXSET_1		0x09
-#define TEXCOORDSRC_VTXSET_2		0x0a
-#define TEXCOORDSRC_VTXSET_3		0x0b
-#define TEXCOORDSRC_VTXSET_4		0x0c
-#define TEXCOORDSRC_VTXSET_5		0x0d
-#define TEXCOORDSRC_VTXSET_6		0x0e
-#define TEXCOORDSRC_VTXSET_7		0x0f
-
-#define MAP_UNIT(unit)			((unit)<<16)
-#define MAP_UNIT_MASK			(0x7<<16)
-
-/* _3DSTATE_MAP_COORD_SETS, p164 */
-#define _3DSTATE_MAP_COORD_SET_CMD	(CMD_3D|(0x1c<<24)|(0x01<<19))
-#define ENABLE_TEXCOORD_PARAMS		(1<<15)
-#define TEXCOORDS_ARE_NORMAL		(1<<14)
-#define TEXCOORDS_ARE_IN_TEXELUNITS	0
-#define TEXCOORDTYPE_CARTESIAN		0
-#define TEXCOORDTYPE_HOMOGENEOUS	(1<<11)
-#define TEXCOORDTYPE_VECTOR		(2<<11)
-#define TEXCOORDTYPE_MASK	        (0x7<<11)
-#define ENABLE_ADDR_V_CNTL		(1<<7)
-#define ENABLE_ADDR_U_CNTL		(1<<3)
-#define TEXCOORD_ADDR_V_MODE(x) 	((x)<<4)
-#define TEXCOORD_ADDR_U_MODE(x) 	(x)
-#define TEXCOORDMODE_WRAP		0
-#define TEXCOORDMODE_MIRROR		1
-#define TEXCOORDMODE_CLAMP		2
-#define TEXCOORDMODE_WRAP_SHORTEST	3
-#define TEXCOORDMODE_CLAMP_BORDER	4
-#define TEXCOORD_ADDR_V_MASK		0x70
-#define TEXCOORD_ADDR_U_MASK		0x7
-
-/* _3DSTATE_MAP_CUBE, p168 TODO */
-#define _3DSTATE_MAP_CUBE		(CMD_3D|(0x1c<<24)|(0x0a<<19))
-#define CUBE_NEGX_ENABLE                (1<<5)
-#define CUBE_POSX_ENABLE                (1<<4)
-#define CUBE_NEGY_ENABLE                (1<<3)
-#define CUBE_POSY_ENABLE                (1<<2)
-#define CUBE_NEGZ_ENABLE                (1<<1)
-#define CUBE_POSZ_ENABLE                (1<<0)
-
-
-/* _3DSTATE_MODES_1, p190 */
-#define _3DSTATE_MODES_1_CMD		(CMD_3D|(0x08<<24))
-#define BLENDFUNC_MASK			0x3f0000
-#define ENABLE_COLR_BLND_FUNC		(1<<21)
-#define BLENDFUNC_ADD			0
-#define BLENDFUNC_SUB			(1<<16)
-#define BLENDFUNC_RVRSE_SUB		(2<<16)
-#define BLENDFUNC_MIN			(3<<16)
-#define BLENDFUNC_MAX			(4<<16)
-#define SRC_DST_BLND_MASK		0xfff
-#define ENABLE_SRC_BLND_FACTOR		(1<<11)
-#define ENABLE_DST_BLND_FACTOR		(1<<5)
-#define SRC_BLND_FACT(x)		((x)<<6)
-#define DST_BLND_FACT(x)		(x)
-
-
-/* _3DSTATE_MODES_2, p192 */
-#define _3DSTATE_MODES_2_CMD		(CMD_3D|(0x0f<<24))
-#define ENABLE_GLOBAL_DEPTH_BIAS	(1<<22)
-#define GLOBAL_DEPTH_BIAS(x)		((x)<<14)
-#define ENABLE_ALPHA_TEST_FUNC		(1<<13)
-#define ENABLE_ALPHA_REF_VALUE		(1<<8)
-#define ALPHA_TEST_FUNC(x)		((x)<<9)
-#define ALPHA_REF_VALUE(x)		(x)
-
-#define ALPHA_TEST_REF_MASK		0x3fff
-
-/* _3DSTATE_MODES_3, p193 */
-#define _3DSTATE_MODES_3_CMD		(CMD_3D|(0x02<<24))
-#define DEPTH_TEST_FUNC_MASK		0x1f0000
-#define ENABLE_DEPTH_TEST_FUNC		(1<<20)
-/* Uses COMPAREFUNC */
-#define DEPTH_TEST_FUNC(x)		((x)<<16)
-#define ENABLE_ALPHA_SHADE_MODE 	(1<<11)
-#define ENABLE_FOG_SHADE_MODE		(1<<9)
-#define ENABLE_SPEC_SHADE_MODE		(1<<7)
-#define ENABLE_COLOR_SHADE_MODE 	(1<<5)
-#define ALPHA_SHADE_MODE(x)		((x)<<10)
-#define FOG_SHADE_MODE(x)		((x)<<8)
-#define SPEC_SHADE_MODE(x)		((x)<<6)
-#define COLOR_SHADE_MODE(x)		((x)<<4)
-#define CULLMODE_MASK			0xf
-#define ENABLE_CULL_MODE		(1<<3)
-#define CULLMODE_BOTH			0
-#define CULLMODE_NONE			1
-#define CULLMODE_CW			2
-#define CULLMODE_CCW			3
-
-#define SHADE_MODE_LINEAR		0
-#define SHADE_MODE_FLAT 		0x1
-
-/* _3DSTATE_MODES_4, p195 */
-#define _3DSTATE_MODES_4_CMD		(CMD_3D|(0x16<<24))
-#define ENABLE_LOGIC_OP_FUNC		(1<<23)
-#define LOGIC_OP_FUNC(x)		((x)<<18)
-#define LOGICOP_MASK			((1<<18)|(1<<19)|(1<<20)|(1<<21))
-#define LOGICOP_CLEAR			0
-#define LOGICOP_NOR			0x1
-#define LOGICOP_AND_INV 		0x2
-#define LOGICOP_COPY_INV		0x3
-#define LOGICOP_AND_RVRSE		0x4
-#define LOGICOP_INV			0x5
-#define LOGICOP_XOR			0x6
-#define LOGICOP_NAND			0x7
-#define LOGICOP_AND			0x8
-#define LOGICOP_EQUIV			0x9
-#define LOGICOP_NOOP			0xa
-#define LOGICOP_OR_INV			0xb
-#define LOGICOP_COPY			0xc
-#define LOGICOP_OR_RVRSE		0xd
-#define LOGICOP_OR			0xe
-#define LOGICOP_SET			0xf
-#define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
-#define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
-#define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
-#define ENABLE_STENCIL_WRITE_MASK	(1<<16)
-#define STENCIL_WRITE_MASK(x)		((x)&0xff)
-
-/* _3DSTATE_MODES_5, p196 */
-#define _3DSTATE_MODES_5_CMD		(CMD_3D|(0x0c<<24))
-#define ENABLE_SPRITE_POINT_TEX 	(1<<23)
-#define SPRITE_POINT_TEX_ON		(1<<22)
-#define SPRITE_POINT_TEX_OFF		0
-#define FLUSH_RENDER_CACHE		(1<<18)
-#define FLUSH_TEXTURE_CACHE		(1<<16)
-#define FIXED_LINE_WIDTH_MASK		0xfc00
-#define ENABLE_FIXED_LINE_WIDTH 	(1<<15)
-#define FIXED_LINE_WIDTH(x)		((x)<<10)
-#define FIXED_POINT_WIDTH_MASK		0x3ff
-#define ENABLE_FIXED_POINT_WIDTH	(1<<9)
-#define FIXED_POINT_WIDTH(x)		(x)
-
-/* _3DSTATE_RASTERIZATION_RULES, p198 */
-#define _3DSTATE_RASTER_RULES_CMD	(CMD_3D|(0x07<<24))
-#define ENABLE_POINT_RASTER_RULE	(1<<15)
-#define OGL_POINT_RASTER_RULE		(1<<13)
-#define ENABLE_LINE_STRIP_PROVOKE_VRTX	(1<<8)
-#define ENABLE_TRI_FAN_PROVOKE_VRTX	(1<<5)
-#define ENABLE_TRI_STRIP_PROVOKE_VRTX	(1<<2)
-#define LINE_STRIP_PROVOKE_VRTX(x)	((x)<<6)
-#define TRI_FAN_PROVOKE_VRTX(x) 	((x)<<3)
-#define TRI_STRIP_PROVOKE_VRTX(x)	(x)
-
-/* _3DSTATE_SCISSOR_ENABLE, p200 */
-#define _3DSTATE_SCISSOR_ENABLE_CMD	(CMD_3D|(0x1c<<24)|(0x10<<19))
-#define ENABLE_SCISSOR_RECT		((1<<1) | 1)
-#define DISABLE_SCISSOR_RECT		(1<<1)
-
-/* _3DSTATE_SCISSOR_RECTANGLE_0, p201 */
-#define _3DSTATE_SCISSOR_RECT_0_CMD	(CMD_3D|(0x1d<<24)|(0x81<<16)|1)
-/* Dword 1 */
-#define SCISSOR_RECT_0_YMIN(x)		((x)<<16)
-#define SCISSOR_RECT_0_XMIN(x)		(x)
-/* Dword 2 */
-#define SCISSOR_RECT_0_YMAX(x)		((x)<<16)
-#define SCISSOR_RECT_0_XMAX(x)		(x)
-
-/* _3DSTATE_STENCIL_TEST, p202 */
-#define _3DSTATE_STENCIL_TEST_CMD	(CMD_3D|(0x09<<24))
-#define ENABLE_STENCIL_PARMS		(1<<23)
-#define STENCIL_OPS_MASK		(0xffc000)
-#define STENCIL_FAIL_OP(x)		((x)<<20)
-#define STENCIL_PASS_DEPTH_FAIL_OP(x)	((x)<<17)
-#define STENCIL_PASS_DEPTH_PASS_OP(x)	((x)<<14)
-
-#define ENABLE_STENCIL_TEST_FUNC_MASK	((1<<13)|(1<<12)|(1<<11)|(1<<10)|(1<<9))
-#define ENABLE_STENCIL_TEST_FUNC	(1<<13)
-/* Uses COMPAREFUNC */
-#define STENCIL_TEST_FUNC(x)		((x)<<9)
-#define STENCIL_REF_VALUE_MASK		((1<<8)|0xff)
-#define ENABLE_STENCIL_REF_VALUE	(1<<8)
-#define STENCIL_REF_VALUE(x)		(x)
-
-/* _3DSTATE_VERTEX_FORMAT, p204 */
-#define _3DSTATE_VFT0_CMD	(CMD_3D|(0x05<<24))
-#define VFT0_POINT_WIDTH	(1<<12)
-#define VFT0_TEX_COUNT_MASK    	(7<<8)
-#define VFT0_TEX_COUNT_SHIFT    8
-#define VFT0_TEX_COUNT(x) 	((x)<<8)
-#define VFT0_SPEC		(1<<7)
-#define VFT0_DIFFUSE		(1<<6)
-#define VFT0_DEPTH_OFFSET  	(1<<5)
-#define VFT0_XYZ		(1<<1)
-#define VFT0_XYZW		(2<<1)
-#define VFT0_XY			(3<<1)
-#define VFT0_XYW		(4<<1)
-#define VFT0_XYZW_MASK          (7<<1)
-
-/* _3DSTATE_VERTEX_FORMAT_2, p206 */
-#define _3DSTATE_VFT1_CMD	(CMD_3D|(0x0a<<24))
-#define VFT1_TEX7_FMT(x)	((x)<<14)
-#define VFT1_TEX6_FMT(x)	((x)<<12)
-#define VFT1_TEX5_FMT(x)	((x)<<10)
-#define VFT1_TEX4_FMT(x)	((x)<<8)
-#define VFT1_TEX3_FMT(x)	((x)<<6)
-#define VFT1_TEX2_FMT(x)	((x)<<4)
-#define VFT1_TEX1_FMT(x)	((x)<<2)
-#define VFT1_TEX0_FMT(x)	(x)
-#define VFT1_TEX0_MASK          3
-#define VFT1_TEX1_SHIFT         2
-#define TEXCOORDFMT_2D		0
-#define TEXCOORDFMT_3D		1
-#define TEXCOORDFMT_4D		2
-#define TEXCOORDFMT_1D		3
-
-/*New stuff picked up along the way */
-
-#define MLC_LOD_BIAS_MASK ((1<<7)-1)
-
-
-/* _3DSTATE_VERTEX_TRANSFORM, p207 */
-#define _3DSTATE_VERTEX_TRANS_CMD	(CMD_3D|(0x1d<<24)|(0x8b<<16)|0)
-#define _3DSTATE_VERTEX_TRANS_MTX_CMD	(CMD_3D|(0x1d<<24)|(0x8b<<16)|6)
-/* Dword 1 */
-#define ENABLE_VIEWPORT_TRANSFORM	((1<<31)|(1<<30))
-#define DISABLE_VIEWPORT_TRANSFORM	(1<<31)
-#define ENABLE_PERSP_DIVIDE		((1<<29)|(1<<28))
-#define DISABLE_PERSP_DIVIDE		(1<<29)
-#define VRTX_TRANS_LOAD_MATRICES	0x7421
-#define VRTX_TRANS_NO_LOAD_MATRICES	0x0000
-/* Dword 2 -> 7  are matrix elements */
-
-/* _3DSTATE_W_STATE, p209 */
-#define _3DSTATE_W_STATE_CMD		(CMD_3D|(0x1d<<24)|(0x8d<<16)|1)
-/* Dword 1 */
-#define MAGIC_W_STATE_DWORD1		0x00000008
-/* Dword 2 */
-#define WFAR_VALUE(x)			(x)
-
-
-/* Stipple command, carried over from the i810, apparently:
- */
-#define _3DSTATE_STIPPLE           ((0x3<<29)|(0x1d<<24)|(0x83<<16))
-#define ST1_ENABLE               (1<<16)
-#define ST1_MASK                 (0xffff)
-
-
-
-#define _3DSTATE_LOAD_STATE_IMMEDIATE_2      ((0x3<<29)|(0x1d<<24)|(0x03<<16))
-#define LOAD_TEXTURE_MAP0                   (1<<11)
-#define LOAD_GLOBAL_COLOR_FACTOR            (1<<6)
-
-#define TM0S0_ADDRESS_MASK              0xfffffffc
-#define TM0S0_USE_FENCE                 (1<<1)
-
-#define TM0S1_HEIGHT_SHIFT              21
-#define TM0S1_WIDTH_SHIFT               10
-#define TM0S1_PALETTE_SELECT            (1<<9)
-#define TM0S1_MAPSURF_FORMAT_MASK       (0x7 << 6)
-#define TM0S1_MAPSURF_FORMAT_SHIFT      6
-#define    MAPSURF_8BIT_INDEXED		   (0<<6)
-#define    MAPSURF_8BIT		 	   (1<<6)
-#define    MAPSURF_16BIT		   (2<<6)
-#define    MAPSURF_32BIT		   (3<<6)
-#define    MAPSURF_411			   (4<<6)
-#define    MAPSURF_422			   (5<<6)
-#define    MAPSURF_COMPRESSED		   (6<<6)
-#define    MAPSURF_4BIT_INDEXED		   (7<<6)
-#define TM0S1_MT_FORMAT_MASK         (0x7 << 3)
-#define TM0S1_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_IDX_RGB565	           (0<<3)       /* SURFACE_8BIT_INDEXED */
-#define    MT_8BIT_IDX_ARGB1555	           (1<<3)
-#define    MT_8BIT_IDX_ARGB4444	           (2<<3)
-#define    MT_8BIT_IDX_AY88		   (3<<3)
-#define    MT_8BIT_IDX_ABGR8888	           (4<<3)
-#define    MT_8BIT_IDX_BUMP_88DVDU 	   (5<<3)
-#define    MT_8BIT_IDX_BUMP_655LDVDU	   (6<<3)
-#define    MT_8BIT_IDX_ARGB8888	           (7<<3)
-#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
-#define    MT_8BIT_L8		           (1<<3)
-#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
-#define    MT_16BIT_ARGB1555		   (1<<3)
-#define    MT_16BIT_ARGB4444		   (2<<3)
-#define    MT_16BIT_AY88		   (3<<3)
-#define    MT_16BIT_DIB_ARGB1555_8888      (4<<3)
-#define    MT_16BIT_BUMP_88DVDU	           (5<<3)
-#define    MT_16BIT_BUMP_655LDVDU	   (6<<3)
-#define    MT_16BIT_DIB_RGB565_8888	   (7<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
-#define    MT_32BIT_ABGR8888		   (1<<3)
-#define    MT_32BIT_XRGB8888		   (2<<3)       /* XXX: Guess from i915_reg.h */
-#define    MT_32BIT_BUMP_XLDVDU_8888	   (6<<3)
-#define    MT_32BIT_DIB_8888		   (7<<3)
-#define    MT_411_YUV411		   (0<<3)       /* SURFACE_411 */
-#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
-#define    MT_422_YCRCB_NORMAL	           (1<<3)
-#define    MT_422_YCRCB_SWAPUV	           (2<<3)
-#define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
-#define    MT_COMPRESS_DXT2_3	           (1<<3)
-#define    MT_COMPRESS_DXT4_5	           (2<<3)
-#define    MT_COMPRESS_FXT1		   (3<<3)
-#define TM0S1_COLORSPACE_CONVERSION     (1 << 2)
-#define TM0S1_TILED_SURFACE             (1 << 1)
-#define TM0S1_TILE_WALK                 (1 << 0)
-
-#define TM0S2_PITCH_SHIFT               21
-#define TM0S2_CUBE_FACE_ENA_SHIFT       15
-#define TM0S2_CUBE_FACE_ENA_MASK        (1<<15)
-#define TM0S2_MAP_FORMAT                (1<<14)
-#define TM0S2_VERTICAL_LINE_STRIDE      (1<<13)
-#define TM0S2_VERITCAL_LINE_STRIDE_OFF  (1<<12)
-#define TM0S2_OUTPUT_CHAN_SHIFT         10
-#define TM0S2_OUTPUT_CHAN_MASK          (3<<10)
-
-#define TM0S3_MIP_FILTER_MASK           (0x3<<30)
-#define TM0S3_MIP_FILTER_SHIFT          30
-#define MIPFILTER_NONE		0
-#define MIPFILTER_NEAREST	1
-#define MIPFILTER_LINEAR	3
-#define TM0S3_MAG_FILTER_MASK           (0x3<<28)
-#define TM0S3_MAG_FILTER_SHIFT          28
-#define TM0S3_MIN_FILTER_MASK           (0x3<<26)
-#define TM0S3_MIN_FILTER_SHIFT          26
-#define FILTER_NEAREST		0
-#define FILTER_LINEAR		1
-#define FILTER_ANISOTROPIC	2
-
-#define TM0S3_LOD_BIAS_SHIFT		17
-#define TM0S3_LOD_BIAS_MASK		(0x1ff<<17)
-#define TM0S3_MAX_MIP_SHIFT		9
-#define TM0S3_MAX_MIP_MASK		(0xff<<9)
-#define TM0S3_MIN_MIP_SHIFT		3
-#define TM0S3_MIN_MIP_MASK		(0x3f<<3)
-#define TM0S3_KILL_PIXEL		(1<<2)
-#define TM0S3_KEYED_FILTER		(1<<1)
-#define TM0S3_CHROMA_KEY		(1<<0)
-
-
-/* _3DSTATE_MAP_TEXEL_STREAM, p188 */
-#define _3DSTATE_MAP_TEX_STREAM_CMD	(CMD_3D|(0x1c<<24)|(0x05<<19))
-#define DISABLE_TEX_STREAM_BUMP 	(1<<12)
-#define ENABLE_TEX_STREAM_BUMP		((1<<12)|(1<<11))
-#define TEX_MODIFY_UNIT_0		0
-#define TEX_MODIFY_UNIT_1		(1<<8)
-#define ENABLE_TEX_STREAM_COORD_SET	(1<<7)
-#define TEX_STREAM_COORD_SET(x) 	((x)<<4)
-#define ENABLE_TEX_STREAM_MAP_IDX	(1<<3)
-#define TEX_STREAM_MAP_IDX(x)		(x)
-
-
-#define MI_FLUSH           ((0<<29)|(4<<23))
-#define FLUSH_MAP_CACHE    (1<<0)
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/i830_state.c b/src/mesa/drivers/dri/i915tex/i830_state.c
deleted file mode 100644
index 3c149e69055..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_state.c
+++ /dev/null
@@ -1,1116 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
-
-#include "texmem.h"
-
-#include "drivers/common/driverfuncs.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-#include "intel_fbo.h"
-
-#include "i830_context.h"
-#include "i830_reg.h"
-
-#define FILE_DEBUG_FLAG DEBUG_STATE
-
-static void
-i830StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
-                        GLuint mask)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int test = intel_translate_compare_func(func);
-
-   mask = mask & 0xff;
-
-   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(func), ref, mask);
-
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-                                           STENCIL_TEST_MASK(mask));
-   i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_REF_VALUE_MASK |
-                                                ENABLE_STENCIL_TEST_FUNC_MASK);
-   i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_REF_VALUE |
-                                               ENABLE_STENCIL_TEST_FUNC |
-                                               STENCIL_REF_VALUE(ref) |
-                                               STENCIL_TEST_FUNC(test));
-}
-
-static void
-i830StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
-   
-   mask = mask & 0xff;
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-                                           STENCIL_WRITE_MASK(mask));
-}
-
-static void
-i830StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
-                      GLenum zpass)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int fop, dfop, dpop;
-
-   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(fail),
-       _mesa_lookup_enum_by_nr(zfail), 
-       _mesa_lookup_enum_by_nr(zpass));
-
-   fop = 0;
-   dfop = 0;
-   dpop = 0;
-
-   switch (fail) {
-   case GL_KEEP:
-      fop = STENCILOP_KEEP;
-      break;
-   case GL_ZERO:
-      fop = STENCILOP_ZERO;
-      break;
-   case GL_REPLACE:
-      fop = STENCILOP_REPLACE;
-      break;
-   case GL_INCR:
-      fop = STENCILOP_INCRSAT;
-      break;
-   case GL_DECR:
-      fop = STENCILOP_DECRSAT;
-      break;
-   case GL_INCR_WRAP:
-      fop = STENCILOP_INCR;
-      break;
-   case GL_DECR_WRAP:
-      fop = STENCILOP_DECR;
-      break;
-   case GL_INVERT:
-      fop = STENCILOP_INVERT;
-      break;
-   default:
-      break;
-   }
-   switch (zfail) {
-   case GL_KEEP:
-      dfop = STENCILOP_KEEP;
-      break;
-   case GL_ZERO:
-      dfop = STENCILOP_ZERO;
-      break;
-   case GL_REPLACE:
-      dfop = STENCILOP_REPLACE;
-      break;
-   case GL_INCR:
-      dfop = STENCILOP_INCRSAT;
-      break;
-   case GL_DECR:
-      dfop = STENCILOP_DECRSAT;
-      break;
-   case GL_INCR_WRAP:
-      dfop = STENCILOP_INCR;
-      break;
-   case GL_DECR_WRAP:
-      dfop = STENCILOP_DECR;
-      break;
-   case GL_INVERT:
-      dfop = STENCILOP_INVERT;
-      break;
-   default:
-      break;
-   }
-   switch (zpass) {
-   case GL_KEEP:
-      dpop = STENCILOP_KEEP;
-      break;
-   case GL_ZERO:
-      dpop = STENCILOP_ZERO;
-      break;
-   case GL_REPLACE:
-      dpop = STENCILOP_REPLACE;
-      break;
-   case GL_INCR:
-      dpop = STENCILOP_INCRSAT;
-      break;
-   case GL_DECR:
-      dpop = STENCILOP_DECRSAT;
-      break;
-   case GL_INCR_WRAP:
-      dpop = STENCILOP_INCR;
-      break;
-   case GL_DECR_WRAP:
-      dpop = STENCILOP_DECR;
-      break;
-   case GL_INVERT:
-      dpop = STENCILOP_INVERT;
-      break;
-   default:
-      break;
-   }
-
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STENCILTST] &= ~(STENCIL_OPS_MASK);
-   i830->state.Ctx[I830_CTXREG_STENCILTST] |= (ENABLE_STENCIL_PARMS |
-                                               STENCIL_FAIL_OP(fop) |
-                                               STENCIL_PASS_DEPTH_FAIL_OP
-                                               (dfop) |
-                                               STENCIL_PASS_DEPTH_PASS_OP
-                                               (dpop));
-}
-
-static void
-i830AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int test = intel_translate_compare_func(func);
-   GLubyte refByte;
-   GLuint refInt;
-
-   UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
-   refInt = (GLuint) refByte;
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE2] &= ~ALPHA_TEST_REF_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE2] |= (ENABLE_ALPHA_TEST_FUNC |
-                                           ENABLE_ALPHA_REF_VALUE |
-                                           ALPHA_TEST_FUNC(test) |
-                                           ALPHA_REF_VALUE(refInt));
-}
-
-/**
- * Makes sure that the proper enables are set for LogicOp, Independant Alpha
- * Blend, and Blending.  It needs to be called from numerous places where we
- * could change the LogicOp or Independant Alpha Blend without subsequent
- * calls to glEnable.
- * 
- * \todo
- * This function is substantially different from the old i830-specific driver.
- * I'm not sure which is correct.
- */
-static void
-i830EvalLogicOpBlendState(GLcontext * ctx)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-   if (RGBA_LOGICOP_ENABLED(ctx)) {
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-                                                  ENABLE_LOGIC_OP_MASK);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-                                                 ENABLE_LOGIC_OP);
-   }
-   else if (ctx->Color.BlendEnabled) {
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-                                                  ENABLE_LOGIC_OP_MASK);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (ENABLE_COLOR_BLEND |
-                                                 DISABLE_LOGIC_OP);
-   }
-   else {
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~(ENABLE_COLOR_BLEND |
-                                                  ENABLE_LOGIC_OP_MASK);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] |= (DISABLE_COLOR_BLEND |
-                                                 DISABLE_LOGIC_OP);
-   }
-}
-
-static void
-i830BlendColor(GLcontext * ctx, const GLfloat color[4])
-{
-   struct i830_context *i830 = i830_context(ctx);
-   GLubyte r, g, b, a;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] =
-      (a << 24) | (r << 16) | (g << 8) | b;
-}
-
-/**
- * Sets both the blend equation (called "function" in i830 docs) and the
- * blend function (called "factor" in i830 docs).  This is done in a single
- * function because some blend equations (i.e., \c GL_MIN and \c GL_MAX)
- * change the interpretation of the blend function.
- */
-static void
-i830_set_blend_state(GLcontext * ctx)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int funcA;
-   int funcRGB;
-   int eqnA;
-   int eqnRGB;
-   int iab;
-   int s1;
-
-
-   funcRGB =
-      SRC_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcRGB))
-      | DST_BLND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstRGB));
-
-   switch (ctx->Color.BlendEquationRGB) {
-   case GL_FUNC_ADD:
-      eqnRGB = BLENDFUNC_ADD;
-      break;
-   case GL_MIN:
-      eqnRGB = BLENDFUNC_MIN;
-      funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
-      break;
-   case GL_MAX:
-      eqnRGB = BLENDFUNC_MAX;
-      funcRGB = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
-      break;
-   case GL_FUNC_SUBTRACT:
-      eqnRGB = BLENDFUNC_SUB;
-      break;
-   case GL_FUNC_REVERSE_SUBTRACT:
-      eqnRGB = BLENDFUNC_RVRSE_SUB;
-      break;
-   default:
-      fprintf(stderr, "[%s:%u] Invalid RGB blend equation (0x%04x).\n",
-              __FUNCTION__, __LINE__, ctx->Color.BlendEquationRGB);
-      return;
-   }
-
-
-   funcA = SRC_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendSrcA))
-      | DST_ABLEND_FACT(intel_translate_blend_factor(ctx->Color.BlendDstA));
-
-   switch (ctx->Color.BlendEquationA) {
-   case GL_FUNC_ADD:
-      eqnA = BLENDFUNC_ADD;
-      break;
-   case GL_MIN:
-      eqnA = BLENDFUNC_MIN;
-      funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
-      break;
-   case GL_MAX:
-      eqnA = BLENDFUNC_MAX;
-      funcA = SRC_BLND_FACT(BLENDFACT_ONE) | DST_BLND_FACT(BLENDFACT_ONE);
-      break;
-   case GL_FUNC_SUBTRACT:
-      eqnA = BLENDFUNC_SUB;
-      break;
-   case GL_FUNC_REVERSE_SUBTRACT:
-      eqnA = BLENDFUNC_RVRSE_SUB;
-      break;
-   default:
-      fprintf(stderr, "[%s:%u] Invalid alpha blend equation (0x%04x).\n",
-              __FUNCTION__, __LINE__, ctx->Color.BlendEquationA);
-      return;
-   }
-
-   iab = eqnA | funcA
-      | _3DSTATE_INDPT_ALPHA_BLEND_CMD
-      | ENABLE_SRC_ABLEND_FACTOR | ENABLE_DST_ABLEND_FACTOR
-      | ENABLE_ALPHA_BLENDFUNC;
-   s1 = eqnRGB | funcRGB
-      | _3DSTATE_MODES_1_CMD
-      | ENABLE_SRC_BLND_FACTOR | ENABLE_DST_BLND_FACTOR
-      | ENABLE_COLR_BLND_FUNC;
-
-   if ((eqnA | funcA) != (eqnRGB | funcRGB))
-      iab |= ENABLE_INDPT_ALPHA_BLEND;
-   else
-      iab |= DISABLE_INDPT_ALPHA_BLEND;
-
-   if (iab != i830->state.Ctx[I830_CTXREG_IALPHAB] ||
-       s1 != i830->state.Ctx[I830_CTXREG_STATE1]) {
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_IALPHAB] = iab;
-      i830->state.Ctx[I830_CTXREG_STATE1] = s1;
-   }
-
-   /* This will catch a logicop blend equation.  It will also ensure
-    * independant alpha blend is really in the correct state (either enabled
-    * or disabled) if blending is already enabled.
-    */
-
-   i830EvalLogicOpBlendState(ctx);
-
-   if (0) {
-      fprintf(stderr,
-              "[%s:%u] STATE1: 0x%08x IALPHAB: 0x%08x blend is %sabled\n",
-              __FUNCTION__, __LINE__, i830->state.Ctx[I830_CTXREG_STATE1],
-              i830->state.Ctx[I830_CTXREG_IALPHAB],
-              (ctx->Color.BlendEnabled) ? "en" : "dis");
-   }
-}
-
-
-static void
-i830BlendEquationSeparate(GLcontext * ctx, GLenum modeRGB, GLenum modeA)
-{
-   DBG("%s -> %s, %s\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(modeRGB),
-       _mesa_lookup_enum_by_nr(modeA));
-
-   (void) modeRGB;
-   (void) modeA;
-   i830_set_blend_state(ctx);
-}
-
-
-static void
-i830BlendFuncSeparate(GLcontext * ctx, GLenum sfactorRGB,
-                      GLenum dfactorRGB, GLenum sfactorA, GLenum dfactorA)
-{
-   DBG("%s -> RGB(%s, %s) A(%s, %s)\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(sfactorRGB),
-       _mesa_lookup_enum_by_nr(dfactorRGB),
-       _mesa_lookup_enum_by_nr(sfactorA),
-       _mesa_lookup_enum_by_nr(dfactorA));
-
-   (void) sfactorRGB;
-   (void) dfactorRGB;
-   (void) sfactorA;
-   (void) dfactorA;
-   i830_set_blend_state(ctx);
-}
-
-
-
-static void
-i830DepthFunc(GLcontext * ctx, GLenum func)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int test = intel_translate_compare_func(func);
-
-   DBG("%s\n", __FUNCTION__);
-   
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE3] &= ~DEPTH_TEST_FUNC_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE3] |= (ENABLE_DEPTH_TEST_FUNC |
-                                           DEPTH_TEST_FUNC(test));
-}
-
-static void
-i830DepthMask(GLcontext * ctx, GLboolean flag)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   DBG("%s flag (%d)\n", __FUNCTION__, flag);
-   
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-   i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DIS_DEPTH_WRITE_MASK;
-
-   if (flag && ctx->Depth.Test)
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DEPTH_WRITE;
-   else
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DEPTH_WRITE;
-}
-
-/* =============================================================
- * Polygon stipple
- *
- * The i830 supports a 4x4 stipple natively, GL wants 32x32.
- * Fortunately stipple is usually a repeating pattern.
- */
-static void
-i830PolygonStipple(GLcontext * ctx, const GLubyte * mask)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   const GLubyte *m = mask;
-   GLubyte p[4];
-   int i, j, k;
-   int active = (ctx->Polygon.StippleFlag &&
-                 i830->intel.reduced_primitive == GL_TRIANGLES);
-   GLuint newMask;
-
-   if (active) {
-      I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-      i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
-   }
-
-   p[0] = mask[12] & 0xf;
-   p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf;
-   p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf;
-   p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf;
-   p[3] |= p[3] << 4;
-
-   for (k = 0; k < 8; k++)
-      for (j = 3; j >= 0; j--)
-         for (i = 0; i < 4; i++, m++)
-            if (*m != p[j]) {
-               i830->intel.hw_stipple = 0;
-               return;
-            }
-
-   newMask = (((p[0] & 0xf) << 0) |
-              ((p[1] & 0xf) << 4) |
-              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
-
-
-   if (newMask == 0xffff || newMask == 0x0) {
-      /* this is needed to make conform pass */
-      i830->intel.hw_stipple = 0;
-      return;
-   }
-
-   i830->state.Stipple[I830_STPREG_ST1] &= ~0xffff;
-   i830->state.Stipple[I830_STPREG_ST1] |= newMask;
-   i830->intel.hw_stipple = 1;
-
-   if (active)
-      i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
-}
-
-
-/* =============================================================
- * Hardware clipping
- */
-static void
-i830Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int x1, y1, x2, y2;
-
-   if (!ctx->DrawBuffer)
-      return;
-
-   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
-
-   if (ctx->DrawBuffer->Name == 0) {
-      x1 = x;
-      y1 = ctx->DrawBuffer->Height - (y + h);
-      x2 = x + w - 1;
-      y2 = y1 + h - 1;
-      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
-   }
-   else {
-      /* FBO - not inverted
-       */
-      x1 = x;
-      y1 = y;
-      x2 = x + w - 1;
-      y2 = y + h - 1;
-      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
-   }
-
-   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
-   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
-   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
-   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
-   
-   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
-
-   I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
-   i830->state.Buffer[I830_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
-   i830->state.Buffer[I830_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
-}
-
-static void
-i830LogicOp(GLcontext * ctx, GLenum opcode)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int tmp = intel_translate_logic_op(opcode);
-
-   DBG("%s\n", __FUNCTION__);
-   
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE4] &= ~LOGICOP_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
-}
-
-
-
-static void
-i830CullFaceFrontFace(GLcontext * ctx, GLenum unused)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   GLuint mode;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   if (!ctx->Polygon.CullFlag) {
-      mode = CULLMODE_NONE;
-   }
-   else if (ctx->Polygon.CullFaceMode != GL_FRONT_AND_BACK) {
-      mode = CULLMODE_CW;
-
-      if (ctx->Polygon.CullFaceMode == GL_FRONT)
-         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
-      if (ctx->Polygon.FrontFace != GL_CCW)
-         mode ^= (CULLMODE_CW ^ CULLMODE_CCW);
-   }
-   else {
-      mode = CULLMODE_BOTH;
-   }
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE3] &= ~CULLMODE_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE3] |= ENABLE_CULL_MODE | mode;
-}
-
-static void
-i830LineWidth(GLcontext * ctx, GLfloat widthf)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   int width;
-   int state5;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   width = (int) (widthf * 2);
-   CLAMP_SELF(width, 1, 15);
-
-   state5 = i830->state.Ctx[I830_CTXREG_STATE5] & ~FIXED_LINE_WIDTH_MASK;
-   state5 |= (ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(width));
-
-   if (state5 != i830->state.Ctx[I830_CTXREG_STATE5]) {
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_STATE5] = state5;
-   }
-}
-
-static void
-i830PointSize(GLcontext * ctx, GLfloat size)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   GLint point_size = (int) size;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   CLAMP_SELF(point_size, 1, 256);
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_STATE5] &= ~FIXED_POINT_WIDTH_MASK;
-   i830->state.Ctx[I830_CTXREG_STATE5] |= (ENABLE_FIXED_POINT_WIDTH |
-                                           FIXED_POINT_WIDTH(point_size));
-}
-
-
-/* =============================================================
- * Color masks
- */
-
-static void
-i830ColorMask(GLcontext * ctx,
-              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   GLuint tmp = 0;
-
-   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b, a);
-
-   tmp = ((i830->state.Ctx[I830_CTXREG_ENABLES_2] & ~WRITEMASK_MASK) |
-          ENABLE_COLOR_MASK |
-          ENABLE_COLOR_WRITE |
-          ((!r) << WRITEMASK_RED_SHIFT) |
-          ((!g) << WRITEMASK_GREEN_SHIFT) |
-          ((!b) << WRITEMASK_BLUE_SHIFT) | ((!a) << WRITEMASK_ALPHA_SHIFT));
-
-   if (tmp != i830->state.Ctx[I830_CTXREG_ENABLES_2]) {
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] = tmp;
-   }
-}
-
-static void
-update_specular(GLcontext * ctx)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-   i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_SPEC_ADD_MASK;
-
-   if (NEED_SECONDARY_COLOR(ctx))
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_SPEC_ADD;
-   else
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_SPEC_ADD;
-}
-
-static void
-i830LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
-{
-   DBG("%s\n", __FUNCTION__);
-   
-   if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular(ctx);
-   }
-}
-
-/* In Mesa 3.5 we can reliably do native flatshading.
- */
-static void
-i830ShadeModel(GLcontext * ctx, GLenum mode)
-{
-   struct i830_context *i830 = i830_context(ctx);
-   I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-
-#define SHADE_MODE_MASK ((1<<10)|(1<<8)|(1<<6)|(1<<4))
-
-   i830->state.Ctx[I830_CTXREG_STATE3] &= ~SHADE_MODE_MASK;
-
-   if (mode == GL_FLAT) {
-      i830->state.Ctx[I830_CTXREG_STATE3] |=
-         (ALPHA_SHADE_MODE(SHADE_MODE_FLAT) | FOG_SHADE_MODE(SHADE_MODE_FLAT)
-          | SPEC_SHADE_MODE(SHADE_MODE_FLAT) |
-          COLOR_SHADE_MODE(SHADE_MODE_FLAT));
-   }
-   else {
-      i830->state.Ctx[I830_CTXREG_STATE3] |=
-         (ALPHA_SHADE_MODE(SHADE_MODE_LINEAR) |
-          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-          COLOR_SHADE_MODE(SHADE_MODE_LINEAR));
-   }
-}
-
-/* =============================================================
- * Fog
- */
-static void
-i830Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   DBG("%s\n", __FUNCTION__);
-   
-   if (pname == GL_FOG_COLOR) {
-      GLuint color = (((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
-                      ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
-                      ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
-
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_FOGCOLOR] =
-         (_3DSTATE_FOG_COLOR_CMD | color);
-   }
-}
-
-/* =============================================================
- */
-
-static void
-i830Enable(GLcontext * ctx, GLenum cap, GLboolean state)
-{
-   struct i830_context *i830 = i830_context(ctx);
-
-   switch (cap) {
-   case GL_LIGHTING:
-   case GL_COLOR_SUM:
-      update_specular(ctx);
-      break;
-
-   case GL_ALPHA_TEST:
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_ALPHA_TEST_MASK;
-      if (state)
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_ALPHA_TEST;
-      else
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_ALPHA_TEST;
-
-      break;
-
-   case GL_BLEND:
-      i830EvalLogicOpBlendState(ctx);
-      break;
-
-   case GL_COLOR_LOGIC_OP:
-      i830EvalLogicOpBlendState(ctx);
-
-      /* Logicop doesn't seem to work at 16bpp:
-       */
-      if (i830->intel.intelScreen->cpp == 2)
-         FALLBACK(&i830->intel, I830_FALLBACK_LOGICOP, state);
-      break;
-
-   case GL_DITHER:
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] &= ~ENABLE_DITHER;
-
-      if (state)
-         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_DITHER;
-      else
-         i830->state.Ctx[I830_CTXREG_ENABLES_2] |= DISABLE_DITHER;
-      break;
-
-   case GL_DEPTH_TEST:
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_DEPTH_TEST_MASK;
-
-      if (state)
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_DEPTH_TEST;
-      else
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_DEPTH_TEST;
-
-      /* Also turn off depth writes when GL_DEPTH_TEST is disabled:
-       */
-      i830DepthMask(ctx, ctx->Depth.Mask);
-      break;
-
-   case GL_SCISSOR_TEST:
-      I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
-
-      if (state)
-         i830->state.Buffer[I830_DESTREG_SENABLE] =
-            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
-      else
-         i830->state.Buffer[I830_DESTREG_SENABLE] =
-            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
-
-      break;
-
-   case GL_LINE_SMOOTH:
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-      i830->state.Ctx[I830_CTXREG_AA] &= ~AA_LINE_ENABLE;
-      if (state)
-         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_ENABLE;
-      else
-         i830->state.Ctx[I830_CTXREG_AA] |= AA_LINE_DISABLE;
-      break;
-
-   case GL_FOG:
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-      i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_DIS_FOG_MASK;
-      if (state)
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_FOG;
-      else
-         i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_FOG;
-      break;
-
-   case GL_CULL_FACE:
-      i830CullFaceFrontFace(ctx, 0);
-      break;
-
-   case GL_TEXTURE_2D:
-      break;
-
-   case GL_STENCIL_TEST:
-      {
-         GLboolean hw_stencil = GL_FALSE;
-         if (ctx->DrawBuffer) {
-            struct intel_renderbuffer *irbStencil
-               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
-            hw_stencil = (irbStencil && irbStencil->region);
-         }
-         if (hw_stencil) {
-            I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-            if (state) {
-               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= ENABLE_STENCIL_TEST;
-               i830->state.Ctx[I830_CTXREG_ENABLES_2] |= ENABLE_STENCIL_WRITE;
-            }
-            else {
-               i830->state.Ctx[I830_CTXREG_ENABLES_1] &= ~ENABLE_STENCIL_TEST;
-               i830->state.Ctx[I830_CTXREG_ENABLES_2] &=
-                  ~ENABLE_STENCIL_WRITE;
-               i830->state.Ctx[I830_CTXREG_ENABLES_1] |= DISABLE_STENCIL_TEST;
-               i830->state.Ctx[I830_CTXREG_ENABLES_2] |=
-                  DISABLE_STENCIL_WRITE;
-            }
-         }
-         else {
-            FALLBACK(&i830->intel, I830_FALLBACK_STENCIL, state);
-         }
-      }
-      break;
-
-   case GL_POLYGON_STIPPLE:
-      /* The stipple command worked on my 855GM box, but not my 845G.
-       * I'll do more testing later to find out exactly which hardware
-       * supports it.  Disabled for now.
-       */
-      if (i830->intel.hw_stipple &&
-          i830->intel.reduced_primitive == GL_TRIANGLES) {
-         I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-         i830->state.Stipple[I830_STPREG_ST1] &= ~ST1_ENABLE;
-         if (state)
-            i830->state.Stipple[I830_STPREG_ST1] |= ST1_ENABLE;
-      }
-      break;
-
-   default:
-      ;
-   }
-}
-
-
-static void
-i830_init_packets(struct i830_context *i830)
-{
-   intelScreenPrivate *screen = i830->intel.intelScreen;
-
-   /* Zero all state */
-   memset(&i830->state, 0, sizeof(i830->state));
-
-   /* Set default blend state */
-   i830->state.TexBlend[0][0] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-                                 TEXPIPE_COLOR |
-                                 ENABLE_TEXOUTPUT_WRT_SEL |
-                                 TEXOP_OUTPUT_CURRENT |
-                                 DISABLE_TEX_CNTRL_STAGE |
-                                 TEXOP_SCALE_1X |
-                                 TEXOP_MODIFY_PARMS |
-                                 TEXOP_LAST_STAGE | TEXBLENDOP_ARG1);
-   i830->state.TexBlend[0][1] = (_3DSTATE_MAP_BLEND_OP_CMD(0) |
-                                 TEXPIPE_ALPHA |
-                                 ENABLE_TEXOUTPUT_WRT_SEL |
-                                 TEXOP_OUTPUT_CURRENT |
-                                 TEXOP_SCALE_1X |
-                                 TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
-   i830->state.TexBlend[0][2] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-                                 TEXPIPE_COLOR |
-                                 TEXBLEND_ARG1 |
-                                 TEXBLENDARG_MODIFY_PARMS |
-                                 TEXBLENDARG_DIFFUSE);
-   i830->state.TexBlend[0][3] = (_3DSTATE_MAP_BLEND_ARG_CMD(0) |
-                                 TEXPIPE_ALPHA |
-                                 TEXBLEND_ARG1 |
-                                 TEXBLENDARG_MODIFY_PARMS |
-                                 TEXBLENDARG_DIFFUSE);
-
-   i830->state.TexBlendWordsUsed[0] = 4;
-
-
-   i830->state.Ctx[I830_CTXREG_VF] = 0;
-   i830->state.Ctx[I830_CTXREG_VF2] = 0;
-
-   i830->state.Ctx[I830_CTXREG_AA] = (_3DSTATE_AA_CMD |
-                                      AA_LINE_ECAAR_WIDTH_ENABLE |
-                                      AA_LINE_ECAAR_WIDTH_1_0 |
-                                      AA_LINE_REGION_WIDTH_ENABLE |
-                                      AA_LINE_REGION_WIDTH_1_0 |
-                                      AA_LINE_DISABLE);
-
-   i830->state.Ctx[I830_CTXREG_ENABLES_1] = (_3DSTATE_ENABLES_1_CMD |
-                                             DISABLE_LOGIC_OP |
-                                             DISABLE_STENCIL_TEST |
-                                             DISABLE_DEPTH_BIAS |
-                                             DISABLE_SPEC_ADD |
-                                             DISABLE_FOG |
-                                             DISABLE_ALPHA_TEST |
-                                             DISABLE_COLOR_BLEND |
-                                             DISABLE_DEPTH_TEST);
-
-#if 000                         /* XXX all the stencil enable state is set in i830Enable(), right? */
-   if (i830->intel.hw_stencil) {
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-                                                ENABLE_STENCIL_WRITE |
-                                                ENABLE_TEX_CACHE |
-                                                ENABLE_DITHER |
-                                                ENABLE_COLOR_MASK |
-                                                /* set no color comps disabled */
-                                                ENABLE_COLOR_WRITE |
-                                                ENABLE_DEPTH_WRITE);
-   }
-   else
-#endif
-   {
-      i830->state.Ctx[I830_CTXREG_ENABLES_2] = (_3DSTATE_ENABLES_2_CMD |
-                                                DISABLE_STENCIL_WRITE |
-                                                ENABLE_TEX_CACHE |
-                                                ENABLE_DITHER |
-                                                ENABLE_COLOR_MASK |
-                                                /* set no color comps disabled */
-                                                ENABLE_COLOR_WRITE |
-                                                ENABLE_DEPTH_WRITE);
-   }
-
-   i830->state.Ctx[I830_CTXREG_STATE1] = (_3DSTATE_MODES_1_CMD |
-                                          ENABLE_COLR_BLND_FUNC |
-                                          BLENDFUNC_ADD |
-                                          ENABLE_SRC_BLND_FACTOR |
-                                          SRC_BLND_FACT(BLENDFACT_ONE) |
-                                          ENABLE_DST_BLND_FACTOR |
-                                          DST_BLND_FACT(BLENDFACT_ZERO));
-
-   i830->state.Ctx[I830_CTXREG_STATE2] = (_3DSTATE_MODES_2_CMD |
-                                          ENABLE_GLOBAL_DEPTH_BIAS |
-                                          GLOBAL_DEPTH_BIAS(0) |
-                                          ENABLE_ALPHA_TEST_FUNC |
-                                          ALPHA_TEST_FUNC(COMPAREFUNC_ALWAYS)
-                                          | ALPHA_REF_VALUE(0));
-
-   i830->state.Ctx[I830_CTXREG_STATE3] = (_3DSTATE_MODES_3_CMD |
-                                          ENABLE_DEPTH_TEST_FUNC |
-                                          DEPTH_TEST_FUNC(COMPAREFUNC_LESS) |
-                                          ENABLE_ALPHA_SHADE_MODE |
-                                          ALPHA_SHADE_MODE(SHADE_MODE_LINEAR)
-                                          | ENABLE_FOG_SHADE_MODE |
-                                          FOG_SHADE_MODE(SHADE_MODE_LINEAR) |
-                                          ENABLE_SPEC_SHADE_MODE |
-                                          SPEC_SHADE_MODE(SHADE_MODE_LINEAR) |
-                                          ENABLE_COLOR_SHADE_MODE |
-                                          COLOR_SHADE_MODE(SHADE_MODE_LINEAR)
-                                          | ENABLE_CULL_MODE | CULLMODE_NONE);
-
-   i830->state.Ctx[I830_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-                                          ENABLE_LOGIC_OP_FUNC |
-                                          LOGIC_OP_FUNC(LOGICOP_COPY) |
-                                          ENABLE_STENCIL_TEST_MASK |
-                                          STENCIL_TEST_MASK(0xff) |
-                                          ENABLE_STENCIL_WRITE_MASK |
-                                          STENCIL_WRITE_MASK(0xff));
-
-   i830->state.Ctx[I830_CTXREG_STENCILTST] = (_3DSTATE_STENCIL_TEST_CMD |
-                                              ENABLE_STENCIL_PARMS |
-                                              STENCIL_FAIL_OP(STENCILOP_KEEP)
-                                              |
-                                              STENCIL_PASS_DEPTH_FAIL_OP
-                                              (STENCILOP_KEEP) |
-                                              STENCIL_PASS_DEPTH_PASS_OP
-                                              (STENCILOP_KEEP) |
-                                              ENABLE_STENCIL_TEST_FUNC |
-                                              STENCIL_TEST_FUNC
-                                              (COMPAREFUNC_ALWAYS) |
-                                              ENABLE_STENCIL_REF_VALUE |
-                                              STENCIL_REF_VALUE(0));
-
-   i830->state.Ctx[I830_CTXREG_STATE5] = (_3DSTATE_MODES_5_CMD | FLUSH_TEXTURE_CACHE | ENABLE_SPRITE_POINT_TEX | SPRITE_POINT_TEX_OFF | ENABLE_FIXED_LINE_WIDTH | FIXED_LINE_WIDTH(0x2) |       /* 1.0 */
-                                          ENABLE_FIXED_POINT_WIDTH |
-                                          FIXED_POINT_WIDTH(1));
-
-   i830->state.Ctx[I830_CTXREG_IALPHAB] = (_3DSTATE_INDPT_ALPHA_BLEND_CMD |
-                                           DISABLE_INDPT_ALPHA_BLEND |
-                                           ENABLE_ALPHA_BLENDFUNC |
-                                           ABLENDFUNC_ADD);
-
-   i830->state.Ctx[I830_CTXREG_FOGCOLOR] = (_3DSTATE_FOG_COLOR_CMD |
-                                            FOG_COLOR_RED(0) |
-                                            FOG_COLOR_GREEN(0) |
-                                            FOG_COLOR_BLUE(0));
-
-   i830->state.Ctx[I830_CTXREG_BLENDCOLOR0] = _3DSTATE_CONST_BLEND_COLOR_CMD;
-   i830->state.Ctx[I830_CTXREG_BLENDCOLOR1] = 0;
-
-   i830->state.Ctx[I830_CTXREG_MCSB0] = _3DSTATE_MAP_COORD_SETBIND_CMD;
-   i830->state.Ctx[I830_CTXREG_MCSB1] = (TEXBIND_SET3(TEXCOORDSRC_VTXSET_3) |
-                                         TEXBIND_SET2(TEXCOORDSRC_VTXSET_2) |
-                                         TEXBIND_SET1(TEXCOORDSRC_VTXSET_1) |
-                                         TEXBIND_SET0(TEXCOORDSRC_VTXSET_0));
-
-
-   i830->state.Stipple[I830_STPREG_ST0] = _3DSTATE_STIPPLE;
-
-   i830->state.Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] = (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(screen->front.pitch) |     /* pitch in bytes */
-                                                 BUF_3D_USE_FENCE);
-
-
-   i830->state.Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] = (BUF_3D_ID_DEPTH | BUF_3D_PITCH(screen->depth.pitch) |  /* pitch in bytes */
-                                                 BUF_3D_USE_FENCE);
-
-   i830->state.Buffer[I830_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
-
-#if 0
-   switch (screen->fbFormat) {
-   case DV_PF_565:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |   /* .5 */
-                                              DSTORG_VERT_BIAS(0x8) |   /* .5 */
-                                              screen->fbFormat |
-                                              DEPTH_IS_Z |
-                                              DEPTH_FRMT_16_FIXED);
-      break;
-   case DV_PF_8888:
-      i830->state.Buffer[I830_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |   /* .5 */
-                                              DSTORG_VERT_BIAS(0x8) |   /* .5 */
-                                              screen->fbFormat |
-                                              DEPTH_IS_Z |
-                                              DEPTH_FRMT_24_FIXED_8_OTHER);
-      break;
-   }
-#endif
-   i830->state.Buffer[I830_DESTREG_SENABLE] = (_3DSTATE_SCISSOR_ENABLE_CMD |
-                                               DISABLE_SCISSOR_RECT);
-   i830->state.Buffer[I830_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
-   i830->state.Buffer[I830_DESTREG_SR1] = 0;
-   i830->state.Buffer[I830_DESTREG_SR2] = 0;
-}
-
-
-void
-i830InitStateFuncs(struct dd_function_table *functions)
-{
-   functions->AlphaFunc = i830AlphaFunc;
-   functions->BlendColor = i830BlendColor;
-   functions->BlendEquationSeparate = i830BlendEquationSeparate;
-   functions->BlendFuncSeparate = i830BlendFuncSeparate;
-   functions->ColorMask = i830ColorMask;
-   functions->CullFace = i830CullFaceFrontFace;
-   functions->DepthFunc = i830DepthFunc;
-   functions->DepthMask = i830DepthMask;
-   functions->Enable = i830Enable;
-   functions->Fogfv = i830Fogfv;
-   functions->FrontFace = i830CullFaceFrontFace;
-   functions->LightModelfv = i830LightModelfv;
-   functions->LineWidth = i830LineWidth;
-   functions->LogicOpcode = i830LogicOp;
-   functions->PointSize = i830PointSize;
-   functions->PolygonStipple = i830PolygonStipple;
-   functions->Scissor = i830Scissor;
-   functions->ShadeModel = i830ShadeModel;
-   functions->StencilFuncSeparate = i830StencilFuncSeparate;
-   functions->StencilMaskSeparate = i830StencilMaskSeparate;
-   functions->StencilOpSeparate = i830StencilOpSeparate;
-}
-
-void
-i830InitState(struct i830_context *i830)
-{
-   GLcontext *ctx = &i830->intel.ctx;
-
-   i830_init_packets(i830);
-
-   _mesa_init_driver_state(ctx);
-
-   memcpy(&i830->initial, &i830->state, sizeof(i830->state));
-
-   i830->current = &i830->state;
-   i830->state.emitted = 0;
-   i830->state.active = (I830_UPLOAD_INVARIENT |
-                         I830_UPLOAD_TEXBLEND(0) |
-                         I830_UPLOAD_STIPPLE |
-                         I830_UPLOAD_CTX | I830_UPLOAD_BUFFERS);
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_tex.c b/src/mesa/drivers/dri/i915tex/i830_tex.c
deleted file mode 100644
index fed464d1aac..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_tex.c
+++ /dev/null
@@ -1,100 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "texmem.h"
-#include "swrast/swrast.h"
-
-#include "mm.h"
-
-#include "intel_ioctl.h"
-
-#include "i830_context.h"
-#include "i830_reg.h"
-
-
-
-static void
-i830TexEnv(GLcontext * ctx, GLenum target,
-           GLenum pname, const GLfloat * param)
-{
-
-   switch (pname) {
-   case GL_TEXTURE_ENV_COLOR:
-   case GL_TEXTURE_ENV_MODE:
-   case GL_COMBINE_RGB:
-   case GL_COMBINE_ALPHA:
-   case GL_SOURCE0_RGB:
-   case GL_SOURCE1_RGB:
-   case GL_SOURCE2_RGB:
-   case GL_SOURCE0_ALPHA:
-   case GL_SOURCE1_ALPHA:
-   case GL_SOURCE2_ALPHA:
-   case GL_OPERAND0_RGB:
-   case GL_OPERAND1_RGB:
-   case GL_OPERAND2_RGB:
-   case GL_OPERAND0_ALPHA:
-   case GL_OPERAND1_ALPHA:
-   case GL_OPERAND2_ALPHA:
-   case GL_RGB_SCALE:
-   case GL_ALPHA_SCALE:
-      break;
-
-   case GL_TEXTURE_LOD_BIAS:{
-         struct i830_context *i830 = i830_context(ctx);
-         GLuint unit = ctx->Texture.CurrentUnit;
-         int b = (int) ((*param) * 16.0);
-         if (b > 63)
-            b = 63;
-         if (b < -64)
-            b = -64;
-         I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-         i830->lodbias_tm0s3[unit] =
-            ((b << TM0S3_LOD_BIAS_SHIFT) & TM0S3_LOD_BIAS_MASK);
-         break;
-      }
-
-   default:
-      break;
-   }
-}
-
-
-
-
-void
-i830InitTextureFuncs(struct dd_function_table *functions)
-{
-   functions->TexEnv = i830TexEnv;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_texblend.c b/src/mesa/drivers/dri/i915tex/i830_texblend.c
deleted file mode 100644
index 58f220eb7ce..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_texblend.c
+++ /dev/null
@@ -1,463 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "texformat.h"
-#include "texstore.h"
-
-#include "mm.h"
-
-#include "intel_screen.h"
-#include "intel_ioctl.h"
-#include "intel_tex.h"
-
-#include "i830_context.h"
-#include "i830_reg.h"
-
-
-/* ================================================================
- * Texture combine functions
- */
-static GLuint
-pass_through(GLuint * state, GLuint blendUnit)
-{
-   state[0] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-               TEXPIPE_COLOR |
-               ENABLE_TEXOUTPUT_WRT_SEL |
-               TEXOP_OUTPUT_CURRENT |
-               DISABLE_TEX_CNTRL_STAGE |
-               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
-   state[1] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-               TEXPIPE_ALPHA |
-               ENABLE_TEXOUTPUT_WRT_SEL |
-               TEXOP_OUTPUT_CURRENT |
-               TEXOP_SCALE_1X | TEXOP_MODIFY_PARMS | TEXBLENDOP_ARG1);
-   state[2] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-               TEXPIPE_COLOR |
-               TEXBLEND_ARG1 |
-               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
-   state[3] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-               TEXPIPE_ALPHA |
-               TEXBLEND_ARG1 |
-               TEXBLENDARG_MODIFY_PARMS | TEXBLENDARG_CURRENT);
-
-   return 4;
-}
-
-static GLuint
-emit_factor(GLuint blendUnit, GLuint * state, GLuint count,
-            const GLfloat * factor)
-{
-   GLubyte r, g, b, a;
-   GLuint col;
-
-   if (0)
-      fprintf(stderr, "emit constant %d: %.2f %.2f %.2f %.2f\n",
-              blendUnit, factor[0], factor[1], factor[2], factor[3]);
-
-   UNCLAMPED_FLOAT_TO_UBYTE(r, factor[0]);
-   UNCLAMPED_FLOAT_TO_UBYTE(g, factor[1]);
-   UNCLAMPED_FLOAT_TO_UBYTE(b, factor[2]);
-   UNCLAMPED_FLOAT_TO_UBYTE(a, factor[3]);
-
-   col = ((a << 24) | (r << 16) | (g << 8) | b);
-
-   state[count++] = _3DSTATE_COLOR_FACTOR_N_CMD(blendUnit);
-   state[count++] = col;
-
-   return count;
-}
-
-
-static INLINE GLuint
-GetTexelOp(GLint unit)
-{
-   switch (unit) {
-   case 0:
-      return TEXBLENDARG_TEXEL0;
-   case 1:
-      return TEXBLENDARG_TEXEL1;
-   case 2:
-      return TEXBLENDARG_TEXEL2;
-   case 3:
-      return TEXBLENDARG_TEXEL3;
-   default:
-      return TEXBLENDARG_TEXEL0;
-   }
-}
-
-
-/**
- * Calculate the hardware instuctions to setup the current texture enviromnemt
- * settings.  Since \c gl_texture_unit::_CurrentCombine is used, both
- * "classic" texture enviroments and GL_ARB_texture_env_combine type texture
- * environments are treated identically.
- *
- * \todo
- * This function should return \c GLboolean.  When \c GL_FALSE is returned,
- * it means that an environment is selected that the hardware cannot do.  This
- * is the way the Radeon and R200 drivers work.
- * 
- * \todo
- * Looking at i830_3d_regs.h, it seems the i830 can do part of
- * GL_ATI_texture_env_combine3.  It can handle using \c GL_ONE and
- * \c GL_ZERO as combine inputs (which the code already supports).  It can
- * also handle the \c GL_MODULATE_ADD_ATI mode.  Is it worth investigating
- * partial support for the extension?
- */
-GLuint
-i830SetTexEnvCombine(struct i830_context * i830,
-                     const struct gl_tex_env_combine_state * combine,
-                     GLint blendUnit,
-                     GLuint texel_op, GLuint * state, const GLfloat * factor)
-{
-   const GLuint numColorArgs = combine->_NumArgsRGB;
-   const GLuint numAlphaArgs = combine->_NumArgsA;
-
-   GLuint blendop;
-   GLuint ablendop;
-   GLuint args_RGB[3];
-   GLuint args_A[3];
-   GLuint rgb_shift;
-   GLuint alpha_shift;
-   GLboolean need_factor = 0;
-   int i;
-   unsigned used;
-   static const GLuint tex_blend_rgb[3] = {
-      TEXPIPE_COLOR | TEXBLEND_ARG1 | TEXBLENDARG_MODIFY_PARMS,
-      TEXPIPE_COLOR | TEXBLEND_ARG2 | TEXBLENDARG_MODIFY_PARMS,
-      TEXPIPE_COLOR | TEXBLEND_ARG0 | TEXBLENDARG_MODIFY_PARMS,
-   };
-   static const GLuint tex_blend_a[3] = {
-      TEXPIPE_ALPHA | TEXBLEND_ARG1 | TEXBLENDARG_MODIFY_PARMS,
-      TEXPIPE_ALPHA | TEXBLEND_ARG2 | TEXBLENDARG_MODIFY_PARMS,
-      TEXPIPE_ALPHA | TEXBLEND_ARG0 | TEXBLENDARG_MODIFY_PARMS,
-   };
-
-   if (INTEL_DEBUG & DEBUG_TEXTURE)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-
-   /* The EXT version of the DOT3 extension does not support the
-    * scale factor, but the ARB version (and the version in OpenGL
-    * 1.3) does.
-    */
-   switch (combine->ModeRGB) {
-   case GL_DOT3_RGB_EXT:
-      alpha_shift = combine->ScaleShiftA;
-      rgb_shift = 0;
-      break;
-
-   case GL_DOT3_RGBA_EXT:
-      alpha_shift = 0;
-      rgb_shift = 0;
-      break;
-
-   default:
-      rgb_shift = combine->ScaleShiftRGB;
-      alpha_shift = combine->ScaleShiftA;
-      break;
-   }
-
-
-   switch (combine->ModeRGB) {
-   case GL_REPLACE:
-      blendop = TEXBLENDOP_ARG1;
-      break;
-   case GL_MODULATE:
-      blendop = TEXBLENDOP_MODULATE;
-      break;
-   case GL_ADD:
-      blendop = TEXBLENDOP_ADD;
-      break;
-   case GL_ADD_SIGNED:
-      blendop = TEXBLENDOP_ADDSIGNED;
-      break;
-   case GL_INTERPOLATE:
-      blendop = TEXBLENDOP_BLEND;
-      break;
-   case GL_SUBTRACT:
-      blendop = TEXBLENDOP_SUBTRACT;
-      break;
-   case GL_DOT3_RGB_EXT:
-   case GL_DOT3_RGB:
-      blendop = TEXBLENDOP_DOT3;
-      break;
-   case GL_DOT3_RGBA_EXT:
-   case GL_DOT3_RGBA:
-      blendop = TEXBLENDOP_DOT3;
-      break;
-   default:
-      return pass_through(state, blendUnit);
-   }
-
-   blendop |= (rgb_shift << TEXOP_SCALE_SHIFT);
-
-
-   /* Handle RGB args */
-   for (i = 0; i < 3; i++) {
-      switch (combine->SourceRGB[i]) {
-      case GL_TEXTURE:
-         args_RGB[i] = texel_op;
-         break;
-      case GL_TEXTURE0:
-      case GL_TEXTURE1:
-      case GL_TEXTURE2:
-      case GL_TEXTURE3:
-         args_RGB[i] = GetTexelOp(combine->SourceRGB[i] - GL_TEXTURE0);
-         break;
-      case GL_CONSTANT:
-         args_RGB[i] = TEXBLENDARG_FACTOR_N;
-         need_factor = 1;
-         break;
-      case GL_PRIMARY_COLOR:
-         args_RGB[i] = TEXBLENDARG_DIFFUSE;
-         break;
-      case GL_PREVIOUS:
-         args_RGB[i] = TEXBLENDARG_CURRENT;
-         break;
-      default:
-         return pass_through(state, blendUnit);
-      }
-
-      switch (combine->OperandRGB[i]) {
-      case GL_SRC_COLOR:
-         args_RGB[i] |= 0;
-         break;
-      case GL_ONE_MINUS_SRC_COLOR:
-         args_RGB[i] |= TEXBLENDARG_INV_ARG;
-         break;
-      case GL_SRC_ALPHA:
-         args_RGB[i] |= TEXBLENDARG_REPLICATE_ALPHA;
-         break;
-      case GL_ONE_MINUS_SRC_ALPHA:
-         args_RGB[i] |= (TEXBLENDARG_REPLICATE_ALPHA | TEXBLENDARG_INV_ARG);
-         break;
-      default:
-         return pass_through(state, blendUnit);
-      }
-   }
-
-
-   /* Need to knobble the alpha calculations of TEXBLENDOP_DOT4 to
-    * match the spec.  Can't use DOT3 as it won't propogate values
-    * into alpha as required:
-    *
-    * Note - the global factor is set up with alpha == .5, so 
-    * the alpha part of the DOT4 calculation should be zero.
-    */
-   if (combine->ModeRGB == GL_DOT3_RGBA_EXT ||
-       combine->ModeRGB == GL_DOT3_RGBA) {
-      ablendop = TEXBLENDOP_DOT4;
-      args_A[0] = TEXBLENDARG_FACTOR;   /* the global factor */
-      args_A[1] = TEXBLENDARG_FACTOR;
-      args_A[2] = TEXBLENDARG_FACTOR;
-   }
-   else {
-      switch (combine->ModeA) {
-      case GL_REPLACE:
-         ablendop = TEXBLENDOP_ARG1;
-         break;
-      case GL_MODULATE:
-         ablendop = TEXBLENDOP_MODULATE;
-         break;
-      case GL_ADD:
-         ablendop = TEXBLENDOP_ADD;
-         break;
-      case GL_ADD_SIGNED:
-         ablendop = TEXBLENDOP_ADDSIGNED;
-         break;
-      case GL_INTERPOLATE:
-         ablendop = TEXBLENDOP_BLEND;
-         break;
-      case GL_SUBTRACT:
-         ablendop = TEXBLENDOP_SUBTRACT;
-         break;
-      default:
-         return pass_through(state, blendUnit);
-      }
-
-
-      ablendop |= (alpha_shift << TEXOP_SCALE_SHIFT);
-
-      /* Handle A args */
-      for (i = 0; i < 3; i++) {
-         switch (combine->SourceA[i]) {
-         case GL_TEXTURE:
-            args_A[i] = texel_op;
-            break;
-         case GL_TEXTURE0:
-         case GL_TEXTURE1:
-         case GL_TEXTURE2:
-         case GL_TEXTURE3:
-            args_A[i] = GetTexelOp(combine->SourceA[i] - GL_TEXTURE0);
-            break;
-         case GL_CONSTANT:
-            args_A[i] = TEXBLENDARG_FACTOR_N;
-            need_factor = 1;
-            break;
-         case GL_PRIMARY_COLOR:
-            args_A[i] = TEXBLENDARG_DIFFUSE;
-            break;
-         case GL_PREVIOUS:
-            args_A[i] = TEXBLENDARG_CURRENT;
-            break;
-         default:
-            return pass_through(state, blendUnit);
-         }
-
-         switch (combine->OperandA[i]) {
-         case GL_SRC_ALPHA:
-            args_A[i] |= 0;
-            break;
-         case GL_ONE_MINUS_SRC_ALPHA:
-            args_A[i] |= TEXBLENDARG_INV_ARG;
-            break;
-         default:
-            return pass_through(state, blendUnit);
-         }
-      }
-   }
-
-
-
-   /* Native Arg1 == Arg0 in GL_EXT_texture_env_combine spec */
-   /* Native Arg2 == Arg1 in GL_EXT_texture_env_combine spec */
-   /* Native Arg0 == Arg2 in GL_EXT_texture_env_combine spec */
-
-   /* When we render we need to figure out which is the last really enabled
-    * tex unit, and put last stage on it
-    */
-
-
-   /* Build color & alpha pipelines */
-
-   used = 0;
-   state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-                    TEXPIPE_COLOR |
-                    ENABLE_TEXOUTPUT_WRT_SEL |
-                    TEXOP_OUTPUT_CURRENT |
-                    DISABLE_TEX_CNTRL_STAGE | TEXOP_MODIFY_PARMS | blendop);
-   state[used++] = (_3DSTATE_MAP_BLEND_OP_CMD(blendUnit) |
-                    TEXPIPE_ALPHA |
-                    ENABLE_TEXOUTPUT_WRT_SEL |
-                    TEXOP_OUTPUT_CURRENT | TEXOP_MODIFY_PARMS | ablendop);
-
-   for (i = 0; i < numColorArgs; i++) {
-      state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-                       tex_blend_rgb[i] | args_RGB[i]);
-   }
-
-   for (i = 0; i < numAlphaArgs; i++) {
-      state[used++] = (_3DSTATE_MAP_BLEND_ARG_CMD(blendUnit) |
-                       tex_blend_a[i] | args_A[i]);
-   }
-
-
-   if (need_factor)
-      return emit_factor(blendUnit, state, used, factor);
-   else
-      return used;
-}
-
-
-static void
-emit_texblend(struct i830_context *i830, GLuint unit, GLuint blendUnit,
-              GLboolean last_stage)
-{
-   struct gl_texture_unit *texUnit = &i830->intel.ctx.Texture.Unit[unit];
-   GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
-
-
-   if (0)
-      fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
-
-   /* Update i830->state.TexBlend
-    */
-   tmp_sz = i830SetTexEnvCombine(i830, texUnit->_CurrentCombine, blendUnit,
-                                 GetTexelOp(unit), tmp, texUnit->EnvColor);
-
-   if (last_stage)
-      tmp[0] |= TEXOP_LAST_STAGE;
-
-   if (tmp_sz != i830->state.TexBlendWordsUsed[blendUnit] ||
-       memcmp(tmp, i830->state.TexBlend[blendUnit],
-              tmp_sz * sizeof(GLuint))) {
-
-      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(blendUnit));
-      memcpy(i830->state.TexBlend[blendUnit], tmp, tmp_sz * sizeof(GLuint));
-      i830->state.TexBlendWordsUsed[blendUnit] = tmp_sz;
-   }
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(blendUnit), GL_TRUE);
-}
-
-static void
-emit_passthrough(struct i830_context *i830)
-{
-   GLuint tmp[I830_TEXBLEND_SIZE], tmp_sz;
-   GLuint unit = 0;
-
-   tmp_sz = pass_through(tmp, unit);
-   tmp[0] |= TEXOP_LAST_STAGE;
-
-   if (tmp_sz != i830->state.TexBlendWordsUsed[unit] ||
-       memcmp(tmp, i830->state.TexBlend[unit], tmp_sz * sizeof(GLuint))) {
-
-      I830_STATECHANGE(i830, I830_UPLOAD_TEXBLEND(unit));
-      memcpy(i830->state.TexBlend[unit], tmp, tmp_sz * sizeof(GLuint));
-      i830->state.TexBlendWordsUsed[unit] = tmp_sz;
-   }
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND(unit), GL_TRUE);
-}
-
-void
-i830EmitTextureBlend(struct i830_context *i830)
-{
-   GLcontext *ctx = &i830->intel.ctx;
-   GLuint unit, last_stage = 0, blendunit = 0;
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEXBLEND_ALL, GL_FALSE);
-
-   if (ctx->Texture._EnabledUnits) {
-      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
-         if (ctx->Texture.Unit[unit]._ReallyEnabled)
-            last_stage = unit;
-
-      for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
-         if (ctx->Texture.Unit[unit]._ReallyEnabled)
-            emit_texblend(i830, unit, blendunit++, last_stage == unit);
-   }
-   else {
-      emit_passthrough(i830);
-   }
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_texstate.c b/src/mesa/drivers/dri/i915tex/i830_texstate.c
deleted file mode 100644
index 0d3f053226f..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_texstate.c
+++ /dev/null
@@ -1,343 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "mtypes.h"
-#include "enums.h"
-#include "texformat.h"
-#include "dri_bufmgr.h"
-
-#include "intel_mipmap_tree.h"
-#include "intel_tex.h"
-
-#include "i830_context.h"
-#include "i830_reg.h"
-
-
-
-static GLuint
-translate_texture_format(GLuint mesa_format)
-{
-   switch (mesa_format) {
-   case MESA_FORMAT_L8:
-      return MAPSURF_8BIT | MT_8BIT_L8;
-   case MESA_FORMAT_I8:
-      return MAPSURF_8BIT | MT_8BIT_I8;
-   case MESA_FORMAT_A8:
-      return MAPSURF_8BIT | MT_8BIT_I8; /* Kludge! */
-   case MESA_FORMAT_AL88:
-      return MAPSURF_16BIT | MT_16BIT_AY88;
-   case MESA_FORMAT_RGB565:
-      return MAPSURF_16BIT | MT_16BIT_RGB565;
-   case MESA_FORMAT_ARGB1555:
-      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
-   case MESA_FORMAT_ARGB4444:
-      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
-   case MESA_FORMAT_ARGB8888:
-      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   case MESA_FORMAT_YCBCR_REV:
-      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
-   case MESA_FORMAT_YCBCR:
-      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
-   case MESA_FORMAT_RGB_FXT1:
-   case MESA_FORMAT_RGBA_FXT1:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
-   case MESA_FORMAT_RGBA_DXT1:
-   case MESA_FORMAT_RGB_DXT1:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-   case MESA_FORMAT_RGBA_DXT3:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-   case MESA_FORMAT_RGBA_DXT5:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-   default:
-      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
-      abort();
-      return 0;
-   }
-}
-
-
-
-
-/* The i915 (and related graphics cores) do not support GL_CLAMP.  The
- * Intel drivers for "other operating systems" implement GL_CLAMP as
- * GL_CLAMP_TO_EDGE, so the same is done here.
- */
-static GLuint
-translate_wrap_mode(GLenum wrap)
-{
-   switch (wrap) {
-   case GL_REPEAT:
-      return TEXCOORDMODE_WRAP;
-   case GL_CLAMP:
-   case GL_CLAMP_TO_EDGE:
-      return TEXCOORDMODE_CLAMP;        /* not really correct */
-   case GL_CLAMP_TO_BORDER:
-      return TEXCOORDMODE_CLAMP_BORDER;
-   case GL_MIRRORED_REPEAT:
-      return TEXCOORDMODE_MIRROR;
-   default:
-      return TEXCOORDMODE_WRAP;
-   }
-}
-
-
-/* Recalculate all state from scratch.  Perhaps not the most
- * efficient, but this has gotten complex enough that we need
- * something which is understandable and reliable.
- */
-static GLboolean
-i830_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
-{
-   GLcontext *ctx = &intel->ctx;
-   struct i830_context *i830 = i830_context(ctx);
-   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
-   struct intel_texture_object *intelObj = intel_texture_object(tObj);
-   struct gl_texture_image *firstImage;
-   GLuint *state = i830->state.Tex[unit], format, pitch;
-
-   memset(state, 0, sizeof(state));
-
-   /*We need to refcount these. */
-
-   if (i830->state.tex_buffer[unit] != NULL) {
-       driBOUnReference(i830->state.tex_buffer[unit]);
-       i830->state.tex_buffer[unit] = NULL;
-   }
-
-   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
-      return GL_FALSE;
-
-   /* Get first image here, since intelObj->firstLevel will get set in
-    * the intel_finalize_mipmap_tree() call above.
-    */
-   firstImage = tObj->Image[0][intelObj->firstLevel];
-
-   if (intelObj->imageOverride) {
-      i830->state.tex_buffer[unit] = NULL;
-      i830->state.tex_offset[unit] = intelObj->textureOffset;
-
-      switch (intelObj->depthOverride) {
-      case 32:
-	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-	 break;
-      case 24:
-      default:
-	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
-	 break;
-      case 16:
-	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
-	 break;
-      }
-
-      pitch = intelObj->pitchOverride;
-   } else {
-      i830->state.tex_buffer[unit] = driBOReference(intelObj->mt->region->
-						    buffer);
-      i830->state.tex_offset[unit] = intel_miptree_image_offset(intelObj->mt,
-								0, intelObj->
-								firstLevel);
-
-      format = translate_texture_format(firstImage->TexFormat->MesaFormat);
-      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
-   }
-
-   state[I830_TEXREG_TM0LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_2 |
-                               (LOAD_TEXTURE_MAP0 << unit) | 4);
-
-/*    state[I830_TEXREG_TM0S0] = (TM0S0_USE_FENCE | */
-/* 			       t->intel.TextureOffset); */
-
-
-   state[I830_TEXREG_TM0S1] =
-      (((firstImage->Height - 1) << TM0S1_HEIGHT_SHIFT) |
-       ((firstImage->Width - 1) << TM0S1_WIDTH_SHIFT) | format);
-
-   state[I830_TEXREG_TM0S2] =
-      ((((pitch / 4) - 1) << TM0S2_PITCH_SHIFT) | TM0S2_CUBE_FACE_ENA_MASK);
-
-   {
-      if (tObj->Target == GL_TEXTURE_CUBE_MAP)
-         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit) |
-                                    CUBE_NEGX_ENABLE |
-                                    CUBE_POSX_ENABLE |
-                                    CUBE_NEGY_ENABLE |
-                                    CUBE_POSY_ENABLE |
-                                    CUBE_NEGZ_ENABLE | CUBE_POSZ_ENABLE);
-      else
-         state[I830_TEXREG_CUBE] = (_3DSTATE_MAP_CUBE | MAP_UNIT(unit));
-   }
-
-
-
-
-   {
-      GLuint minFilt, mipFilt, magFilt;
-
-      switch (tObj->MinFilter) {
-      case GL_NEAREST:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_NONE;
-         break;
-      case GL_LINEAR:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_NONE;
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_NEAREST;
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_NEAREST;
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_LINEAR;
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_LINEAR;
-         break;
-      default:
-         return GL_FALSE;
-      }
-
-      if (tObj->MaxAnisotropy > 1.0) {
-         minFilt = FILTER_ANISOTROPIC;
-         magFilt = FILTER_ANISOTROPIC;
-      }
-      else {
-         switch (tObj->MagFilter) {
-         case GL_NEAREST:
-            magFilt = FILTER_NEAREST;
-            break;
-         case GL_LINEAR:
-            magFilt = FILTER_LINEAR;
-            break;
-         default:
-            return GL_FALSE;
-         }
-      }
-
-      state[I830_TEXREG_TM0S3] = i830->lodbias_tm0s3[unit];
-
-#if 0
-      /* YUV conversion:
-       */
-      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
-          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
-         state[I830_TEXREG_TM0S3] |= SS2_COLORSPACE_CONVERSION;
-#endif
-
-      state[I830_TEXREG_TM0S3] |= ((intelObj->lastLevel -
-                                    intelObj->firstLevel) *
-                                   4) << TM0S3_MIN_MIP_SHIFT;
-
-      state[I830_TEXREG_TM0S3] |= ((minFilt << TM0S3_MIN_FILTER_SHIFT) |
-                                   (mipFilt << TM0S3_MIP_FILTER_SHIFT) |
-                                   (magFilt << TM0S3_MAG_FILTER_SHIFT));
-   }
-
-   {
-      GLenum ws = tObj->WrapS;
-      GLenum wt = tObj->WrapT;
-
-
-      /* 3D textures not available on i830
-       */
-      if (tObj->Target == GL_TEXTURE_3D)
-         return GL_FALSE;
-
-      state[I830_TEXREG_MCS] = (_3DSTATE_MAP_COORD_SET_CMD |
-                                MAP_UNIT(unit) |
-                                ENABLE_TEXCOORD_PARAMS |
-                                ss3 |
-                                ENABLE_ADDR_V_CNTL |
-                                TEXCOORD_ADDR_V_MODE(translate_wrap_mode(wt))
-                                | ENABLE_ADDR_U_CNTL |
-                                TEXCOORD_ADDR_U_MODE(translate_wrap_mode
-                                                     (ws)));
-   }
-
-
-   state[I830_TEXREG_TM0S4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
-                                                  tObj->_BorderChan[1],
-                                                  tObj->_BorderChan[2],
-                                                  tObj->_BorderChan[3]);
-
-
-   I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(unit), GL_TRUE);
-   /* memcmp was already disabled, but definitely won't work as the
-    * region might now change and that wouldn't be detected:
-    */
-   I830_STATECHANGE(i830, I830_UPLOAD_TEX(unit));
-   return GL_TRUE;
-}
-
-
-
-
-void
-i830UpdateTextureState(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   GLboolean ok = GL_TRUE;
-   GLuint i;
-
-   for (i = 0; i < I830_TEX_UNITS && ok; i++) {
-      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
-      case TEXTURE_1D_BIT:
-      case TEXTURE_2D_BIT:
-      case TEXTURE_CUBE_BIT:
-         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_NORMAL);
-         break;
-      case TEXTURE_RECT_BIT:
-         ok = i830_update_tex_unit(intel, i, TEXCOORDS_ARE_IN_TEXELUNITS);
-         break;
-      case 0:{
-	 struct i830_context *i830 = i830_context(&intel->ctx);
-         if (i830->state.active & I830_UPLOAD_TEX(i)) 
-            I830_ACTIVESTATE(i830, I830_UPLOAD_TEX(i), GL_FALSE);
-
-	 if (i830->state.tex_buffer[i] != NULL) {
-	    driBOUnReference(i830->state.tex_buffer[i]);
-	    i830->state.tex_buffer[i] = NULL;
-	 }
-         break;
-      }
-      case TEXTURE_3D_BIT:
-      default:
-         ok = GL_FALSE;
-         break;
-      }
-   }
-
-   FALLBACK(intel, I830_FALLBACK_TEXTURE, !ok);
-
-   if (ok)
-      i830EmitTextureBlend(i830);
-}
diff --git a/src/mesa/drivers/dri/i915tex/i830_vtbl.c b/src/mesa/drivers/dri/i915tex/i830_vtbl.c
deleted file mode 100644
index e432648ada0..00000000000
--- a/src/mesa/drivers/dri/i915tex/i830_vtbl.c
+++ /dev/null
@@ -1,661 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "i830_context.h"
-#include "i830_reg.h"
-#include "intel_batchbuffer.h"
-#include "intel_regions.h"
-#include "tnl/t_context.h"
-#include "tnl/t_vertex.h"
-
-#define FILE_DEBUG_FLAG DEBUG_STATE
-
-static GLboolean i830_check_vertex_size(struct intel_context *intel,
-                                        GLuint expected);
-
-#define SZ_TO_HW(sz)  ((sz-2)&0x3)
-#define EMIT_SZ(sz)   (EMIT_1F + (sz) - 1)
-#define EMIT_ATTR( ATTR, STYLE, V0 )					\
-do {									\
-   intel->vertex_attrs[intel->vertex_attr_count].attrib = (ATTR);	\
-   intel->vertex_attrs[intel->vertex_attr_count].format = (STYLE);	\
-   intel->vertex_attr_count++;						\
-   v0 |= V0;								\
-} while (0)
-
-#define EMIT_PAD( N )							\
-do {									\
-   intel->vertex_attrs[intel->vertex_attr_count].attrib = 0;		\
-   intel->vertex_attrs[intel->vertex_attr_count].format = EMIT_PAD;	\
-   intel->vertex_attrs[intel->vertex_attr_count].offset = (N);		\
-   intel->vertex_attr_count++;						\
-} while (0)
-
-
-#define VRTX_TEX_SET_FMT(n, x)          ((x)<<((n)*2))
-#define TEXBIND_SET(n, x) 		((x)<<((n)*4))
-
-static void
-i830_render_start(struct intel_context *intel)
-{
-   GLcontext *ctx = &intel->ctx;
-   struct i830_context *i830 = i830_context(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-   DECLARE_RENDERINPUTS(index_bitset);
-   GLuint v0 = _3DSTATE_VFT0_CMD;
-   GLuint v2 = _3DSTATE_VFT1_CMD;
-   GLuint mcsb1 = 0;
-
-   RENDERINPUTS_COPY(index_bitset, tnl->render_inputs_bitset);
-
-   /* Important:
-    */
-   VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
-   intel->vertex_attr_count = 0;
-
-   /* EMIT_ATTR's must be in order as they tell t_vertex.c how to
-    * build up a hardware vertex.
-    */
-   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
-      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, VFT0_XYZW);
-      intel->coloroffset = 4;
-   }
-   else {
-      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, VFT0_XYZ);
-      intel->coloroffset = 3;
-   }
-
-   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_POINTSIZE)) {
-      EMIT_ATTR(_TNL_ATTRIB_POINTSIZE, EMIT_1F, VFT0_POINT_WIDTH);
-   }
-
-   EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, VFT0_DIFFUSE);
-
-   intel->specoffset = 0;
-   if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1) ||
-       RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG)) {
-      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_COLOR1)) {
-         intel->specoffset = intel->coloroffset + 1;
-         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, VFT0_SPEC);
-      }
-      else
-         EMIT_PAD(3);
-
-      if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_FOG))
-         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, VFT0_SPEC);
-      else
-         EMIT_PAD(1);
-   }
-
-   if (RENDERINPUTS_TEST_RANGE(index_bitset, _TNL_FIRST_TEX, _TNL_LAST_TEX)) {
-      int i, count = 0;
-
-      for (i = 0; i < I830_TEX_UNITS; i++) {
-         if (RENDERINPUTS_TEST(index_bitset, _TNL_ATTRIB_TEX(i))) {
-            GLuint sz = VB->TexCoordPtr[i]->size;
-            GLuint emit;
-            GLuint mcs = (i830->state.Tex[i][I830_TEXREG_MCS] &
-                          ~TEXCOORDTYPE_MASK);
-
-            switch (sz) {
-            case 1:
-            case 2:
-               emit = EMIT_2F;
-               sz = 2;
-               mcs |= TEXCOORDTYPE_CARTESIAN;
-               break;
-            case 3:
-               emit = EMIT_3F;
-               sz = 3;
-               mcs |= TEXCOORDTYPE_VECTOR;
-               break;
-            case 4:
-               emit = EMIT_3F_XYW;
-               sz = 3;
-               mcs |= TEXCOORDTYPE_HOMOGENEOUS;
-               break;
-            default:
-               continue;
-            };
-
-
-            EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, emit, 0);
-            v2 |= VRTX_TEX_SET_FMT(count, SZ_TO_HW(sz));
-            mcsb1 |= (count + 8) << (i * 4);
-
-            if (mcs != i830->state.Tex[i][I830_TEXREG_MCS]) {
-               I830_STATECHANGE(i830, I830_UPLOAD_TEX(i));
-               i830->state.Tex[i][I830_TEXREG_MCS] = mcs;
-            }
-
-            count++;
-         }
-      }
-
-      v0 |= VFT0_TEX_COUNT(count);
-   }
-
-   /* Only need to change the vertex emit code if there has been a
-    * statechange to a new hardware vertex format:
-    */
-   if (v0 != i830->state.Ctx[I830_CTXREG_VF] ||
-       v2 != i830->state.Ctx[I830_CTXREG_VF2] ||
-       mcsb1 != i830->state.Ctx[I830_CTXREG_MCSB1] ||
-       !RENDERINPUTS_EQUAL(index_bitset, i830->last_index_bitset)) {
-      int k;
-
-      I830_STATECHANGE(i830, I830_UPLOAD_CTX);
-
-      /* Must do this *after* statechange, so as not to affect
-       * buffered vertices reliant on the old state:
-       */
-      intel->vertex_size =
-         _tnl_install_attrs(ctx,
-                            intel->vertex_attrs,
-                            intel->vertex_attr_count,
-                            intel->ViewportMatrix.m, 0);
-
-      intel->vertex_size >>= 2;
-
-      i830->state.Ctx[I830_CTXREG_VF] = v0;
-      i830->state.Ctx[I830_CTXREG_VF2] = v2;
-      i830->state.Ctx[I830_CTXREG_MCSB1] = mcsb1;
-      RENDERINPUTS_COPY(i830->last_index_bitset, index_bitset);
-
-      k = i830_check_vertex_size(intel, intel->vertex_size);
-      assert(k);
-   }
-}
-
-static void
-i830_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   GLuint st1 = i830->state.Stipple[I830_STPREG_ST1];
-
-   st1 &= ~ST1_ENABLE;
-
-   switch (rprim) {
-   case GL_TRIANGLES:
-      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
-         st1 |= ST1_ENABLE;
-      break;
-   case GL_LINES:
-   case GL_POINTS:
-   default:
-      break;
-   }
-
-   i830->intel.reduced_primitive = rprim;
-
-   if (st1 != i830->state.Stipple[I830_STPREG_ST1]) {
-      INTEL_FIREVERTICES(intel);
-
-      I830_STATECHANGE(i830, I830_UPLOAD_STIPPLE);
-      i830->state.Stipple[I830_STPREG_ST1] = st1;
-   }
-}
-
-/* Pull apart the vertex format registers and figure out how large a
- * vertex is supposed to be. 
- */
-static GLboolean
-i830_check_vertex_size(struct intel_context *intel, GLuint expected)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   int vft0 = i830->current->Ctx[I830_CTXREG_VF];
-   int vft1 = i830->current->Ctx[I830_CTXREG_VF2];
-   int nrtex = (vft0 & VFT0_TEX_COUNT_MASK) >> VFT0_TEX_COUNT_SHIFT;
-   int i, sz = 0;
-
-   switch (vft0 & VFT0_XYZW_MASK) {
-   case VFT0_XY:
-      sz = 2;
-      break;
-   case VFT0_XYZ:
-      sz = 3;
-      break;
-   case VFT0_XYW:
-      sz = 3;
-      break;
-   case VFT0_XYZW:
-      sz = 4;
-      break;
-   default:
-      fprintf(stderr, "no xyzw specified\n");
-      return 0;
-   }
-
-   if (vft0 & VFT0_SPEC)
-      sz++;
-   if (vft0 & VFT0_DIFFUSE)
-      sz++;
-   if (vft0 & VFT0_DEPTH_OFFSET)
-      sz++;
-   if (vft0 & VFT0_POINT_WIDTH)
-      sz++;
-
-   for (i = 0; i < nrtex; i++) {
-      switch (vft1 & VFT1_TEX0_MASK) {
-      case TEXCOORDFMT_2D:
-         sz += 2;
-         break;
-      case TEXCOORDFMT_3D:
-         sz += 3;
-         break;
-      case TEXCOORDFMT_4D:
-         sz += 4;
-         break;
-      case TEXCOORDFMT_1D:
-         sz += 1;
-         break;
-      }
-      vft1 >>= VFT1_TEX1_SHIFT;
-   }
-
-   if (sz != expected)
-      fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-
-   return sz == expected;
-}
-
-static void
-i830_emit_invarient_state(struct intel_context *intel)
-{
-   BATCH_LOCALS;
-
-   BEGIN_BATCH(40, 0);
-
-   OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_DFLT_SPEC_CMD);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_DFLT_Z_CMD);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_FOG_MODE_CMD);
-   OUT_BATCH(FOGFUNC_ENABLE |
-             FOG_LINEAR_CONST | FOGSRC_INDEX_Z | ENABLE_FOG_DENSITY);
-   OUT_BATCH(0);
-   OUT_BATCH(0);
-
-
-   OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-             MAP_UNIT(0) |
-             DISABLE_TEX_STREAM_BUMP |
-             ENABLE_TEX_STREAM_COORD_SET |
-             TEX_STREAM_COORD_SET(0) |
-             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(0));
-   OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-             MAP_UNIT(1) |
-             DISABLE_TEX_STREAM_BUMP |
-             ENABLE_TEX_STREAM_COORD_SET |
-             TEX_STREAM_COORD_SET(1) |
-             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(1));
-   OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-             MAP_UNIT(2) |
-             DISABLE_TEX_STREAM_BUMP |
-             ENABLE_TEX_STREAM_COORD_SET |
-             TEX_STREAM_COORD_SET(2) |
-             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(2));
-   OUT_BATCH(_3DSTATE_MAP_TEX_STREAM_CMD |
-             MAP_UNIT(3) |
-             DISABLE_TEX_STREAM_BUMP |
-             ENABLE_TEX_STREAM_COORD_SET |
-             TEX_STREAM_COORD_SET(3) |
-             ENABLE_TEX_STREAM_MAP_IDX | TEX_STREAM_MAP_IDX(3));
-
-   OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
-   OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(0));
-   OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
-   OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(1));
-   OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
-   OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(2));
-   OUT_BATCH(_3DSTATE_MAP_COORD_TRANSFORM);
-   OUT_BATCH(DISABLE_TEX_TRANSFORM | TEXTURE_SET(3));
-
-   OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-             ENABLE_POINT_RASTER_RULE |
-             OGL_POINT_RASTER_RULE |
-             ENABLE_LINE_STRIP_PROVOKE_VRTX |
-             ENABLE_TRI_FAN_PROVOKE_VRTX |
-             ENABLE_TRI_STRIP_PROVOKE_VRTX |
-             LINE_STRIP_PROVOKE_VRTX(1) |
-             TRI_FAN_PROVOKE_VRTX(2) | TRI_STRIP_PROVOKE_VRTX(2));
-
-   OUT_BATCH(_3DSTATE_VERTEX_TRANSFORM);
-   OUT_BATCH(DISABLE_VIEWPORT_TRANSFORM | DISABLE_PERSPECTIVE_DIVIDE);
-
-   OUT_BATCH(_3DSTATE_W_STATE_CMD);
-   OUT_BATCH(MAGIC_W_STATE_DWORD1);
-   OUT_BATCH(0x3f800000 /* 1.0 in IEEE float */ );
-
-
-   OUT_BATCH(_3DSTATE_COLOR_FACTOR_CMD);
-   OUT_BATCH(0x80808080);       /* .5 required in alpha for GL_DOT3_RGBA_EXT */
-
-   ADVANCE_BATCH();
-}
-
-
-#define emit( intel, state, size )			\
-do {							\
-   int k;						\
-   BEGIN_BATCH(size / sizeof(GLuint), 0);		\
-   for (k = 0 ; k < size / sizeof(GLuint) ; k++) {	\
-      if (0) _mesa_printf("  0x%08x\n", state[k]);		\
-      OUT_BATCH(state[k]);				\
-   }							\
-   ADVANCE_BATCH();					\
-} while (0)
-
-static GLuint
-get_state_size(struct i830_hw_state *state)
-{
-   GLuint dirty = state->active & ~state->emitted;
-   GLuint sz = 0;
-   GLuint i;
-
-   if (dirty & I830_UPLOAD_INVARIENT)
-      sz += 40 * sizeof(int);
-
-   if (dirty & I830_UPLOAD_CTX)
-      sz += sizeof(state->Ctx);
-
-   if (dirty & I830_UPLOAD_BUFFERS)
-      sz += sizeof(state->Buffer);
-
-   if (dirty & I830_UPLOAD_STIPPLE)
-      sz += sizeof(state->Stipple);
-
-   for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i)))
-         sz += sizeof(state->Tex[i]);
-
-      if (dirty & I830_UPLOAD_TEXBLEND(i))
-         sz += state->TexBlendWordsUsed[i] * 4;
-   }
-
-   return sz;
-}
-
-
-/* Push the state into the sarea and/or texture memory.
- */
-static void
-i830_emit_state(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   struct i830_hw_state *state = i830->current;
-   int i;
-   GLuint dirty;
-   BATCH_LOCALS;
-
-   /* We don't hold the lock at this point, so want to make sure that
-    * there won't be a buffer wrap.  
-    *
-    * It might be better to talk about explicit places where
-    * scheduling is allowed, rather than assume that it is whenever a
-    * batchbuffer fills up.
-    */
-   intel_batchbuffer_require_space(intel->batch, get_state_size(state), 0);
-
-   /* Do this here as we may have flushed the batchbuffer above,
-    * causing more state to be dirty!
-    */
-   dirty = state->active & ~state->emitted;
-
-   if (dirty & I830_UPLOAD_INVARIENT) {
-      DBG("I830_UPLOAD_INVARIENT:\n");
-      i830_emit_invarient_state(intel);
-   }
-
-   if (dirty & I830_UPLOAD_CTX) {
-      DBG("I830_UPLOAD_CTX:\n");
-      emit(i830, state->Ctx, sizeof(state->Ctx));
-
-   }
-
-   if (dirty & I830_UPLOAD_BUFFERS) {
-      DBG("I830_UPLOAD_BUFFERS:\n");
-      BEGIN_BATCH(I830_DEST_SETUP_SIZE + 2, 0);
-      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR0]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_CBUFADDR1]);
-      OUT_RELOC(state->draw_region->buffer,
-                DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
-                state->draw_region->draw_offset);
-
-      if (state->depth_region) {
-         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR0]);
-         OUT_BATCH(state->Buffer[I830_DESTREG_DBUFADDR1]);
-         OUT_RELOC(state->depth_region->buffer,
-                   DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                   DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
-                   state->depth_region->draw_offset);
-      }
-
-      OUT_BATCH(state->Buffer[I830_DESTREG_DV0]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_DV1]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_SENABLE]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_SR0]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_SR1]);
-      OUT_BATCH(state->Buffer[I830_DESTREG_SR2]);
-      ADVANCE_BATCH();
-   }
-   
-   if (dirty & I830_UPLOAD_STIPPLE) {
-      DBG("I830_UPLOAD_STIPPLE:\n");
-      emit(i830, state->Stipple, sizeof(state->Stipple));
-   }
-
-   for (i = 0; i < I830_TEX_UNITS; i++) {
-      if ((dirty & I830_UPLOAD_TEX(i))) {
-         DBG("I830_UPLOAD_TEX(%d):\n", i);
-
-         BEGIN_BATCH(I830_TEX_SETUP_SIZE + 1, 0);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0LI]);
-
-         if (state->tex_buffer[i]) {
-            OUT_RELOC(state->tex_buffer[i],
-                      DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-                      DRM_BO_MASK_MEM | DRM_BO_FLAG_READ,
-                      state->tex_offset[i] | TM0S0_USE_FENCE);
-         }
-	 else if (state == &i830->meta) {
-	    assert(i == 0);
-	    OUT_BATCH(0);
-	 }
-	 else {
-	    OUT_BATCH(state->tex_offset[i]);
-	 }
-
-         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S1]);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S2]);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S3]);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_TM0S4]);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_MCS]);
-         OUT_BATCH(state->Tex[i][I830_TEXREG_CUBE]);
-      }
-
-      if (dirty & I830_UPLOAD_TEXBLEND(i)) {
-         DBG("I830_UPLOAD_TEXBLEND(%d): %d words\n", i,
-             state->TexBlendWordsUsed[i]);
-         emit(i830, state->TexBlend[i], state->TexBlendWordsUsed[i] * 4);
-      }
-   }
-
-   state->emitted |= dirty;
-}
-
-static void
-i830_destroy_context(struct intel_context *intel)
-{
-   _tnl_free_vertices(&intel->ctx);
-}
-
-
-void
-i830_state_draw_region(struct intel_context *intel,
-		       struct i830_hw_state *state,
-		       struct intel_region *color_region,
-		       struct intel_region *depth_region)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   GLuint value;
-
-   ASSERT(state == &i830->state || state == &i830->meta);
-
-   if (state->draw_region != color_region) {
-      intel_region_release(&state->draw_region);
-      intel_region_reference(&state->draw_region, color_region);
-   }
-   if (state->depth_region != depth_region) {
-      intel_region_release(&state->depth_region);
-      intel_region_reference(&state->depth_region, depth_region);
-   }
-
-   /*
-    * Set stride/cpp values
-    */
-   if (color_region) {
-      state->Buffer[I830_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      state->Buffer[I830_DESTREG_CBUFADDR1] =
-         (BUF_3D_ID_COLOR_BACK |
-          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
-          BUF_3D_USE_FENCE);
-   }
-
-   if (depth_region) {
-      state->Buffer[I830_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      state->Buffer[I830_DESTREG_DBUFADDR1] =
-         (BUF_3D_ID_DEPTH |
-          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
-          BUF_3D_USE_FENCE);
-   }
-
-   /*
-    * Compute/set I830_DESTREG_DV1 value
-    */
-   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
-            DSTORG_VERT_BIAS(0x8) | DEPTH_IS_Z);    /* .5 */
-            
-   if (color_region && color_region->cpp == 4) {
-      value |= DV_PF_8888;
-   }
-   else {
-      value |= DV_PF_565;
-   }
-   if (depth_region && depth_region->cpp == 4) {
-      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
-   }
-   else {
-      value |= DEPTH_FRMT_16_FIXED;
-   }
-   state->Buffer[I830_DESTREG_DV1] = value;
-
-   I830_STATECHANGE(i830, I830_UPLOAD_BUFFERS);
-
-
-}
-
-
-static void
-i830_set_draw_region(struct intel_context *intel,
-                     struct intel_region *color_region,
-                     struct intel_region *depth_region)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   i830_state_draw_region(intel, &i830->state, color_region, depth_region);
-}
-
-#if 0
-static void
-i830_update_color_z_regions(intelContextPtr intel,
-                            const intelRegion * colorRegion,
-                            const intelRegion * depthRegion)
-{
-   i830ContextPtr i830 = I830_CONTEXT(intel);
-
-   i830->state.Buffer[I830_DESTREG_CBUFADDR1] =
-      (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(colorRegion->pitch) |
-       BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_CBUFADDR2] = colorRegion->offset;
-
-   i830->state.Buffer[I830_DESTREG_DBUFADDR1] =
-      (BUF_3D_ID_DEPTH | BUF_3D_PITCH(depthRegion->pitch) | BUF_3D_USE_FENCE);
-   i830->state.Buffer[I830_DESTREG_DBUFADDR2] = depthRegion->offset;
-}
-#endif
-
-
-/* This isn't really handled at the moment.
- */
-static void
-i830_lost_hardware(struct intel_context *intel)
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   i830->state.emitted = 0;
-}
-
-
-
-static GLuint
-i830_flush_cmd(void)
-{
-   return MI_FLUSH | FLUSH_MAP_CACHE;
-}
-
-
-static void 
-i830_assert_not_dirty( struct intel_context *intel )
-{
-   struct i830_context *i830 = i830_context(&intel->ctx);
-   struct i830_hw_state *state = i830->current;
-   GLuint dirty = state->active & ~state->emitted;
-   assert(!dirty);
-}
-
-
-void
-i830InitVtbl(struct i830_context *i830)
-{
-   i830->intel.vtbl.check_vertex_size = i830_check_vertex_size;
-   i830->intel.vtbl.destroy = i830_destroy_context;
-   i830->intel.vtbl.emit_state = i830_emit_state;
-   i830->intel.vtbl.lost_hardware = i830_lost_hardware;
-   i830->intel.vtbl.reduced_primitive_state = i830_reduced_primitive_state;
-   i830->intel.vtbl.set_draw_region = i830_set_draw_region;
-   i830->intel.vtbl.update_texture_state = i830UpdateTextureState;
-   i830->intel.vtbl.flush_cmd = i830_flush_cmd;
-   i830->intel.vtbl.render_start = i830_render_start;
-   i830->intel.vtbl.assert_not_dirty = i830_assert_not_dirty;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_context.c b/src/mesa/drivers/dri/i915tex/i915_context.c
deleted file mode 100644
index 49e30141e43..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_context.c
+++ /dev/null
@@ -1,174 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "i915_context.h"
-#include "imports.h"
-#include "intel_tex.h"
-#include "intel_tris.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vertex.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/tnl.h"
-
-#include "utils.h"
-#include "i915_reg.h"
-
-#include "intel_regions.h"
-#include "intel_batchbuffer.h"
-
-/***************************************
- * 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_env_crossbar", NULL},
-   {"GL_ARB_texture_non_power_of_two", NULL},
-   {"GL_EXT_shadow_funcs", NULL},
-   /* ARB extn won't work if not enabled */
-   {"GL_SGIX_depth_texture", NULL},
-   {NULL, NULL}
-};
-
-/* Override intel default.
- */
-static void
-i915InvalidateState(GLcontext * ctx, GLuint new_state)
-{
-   _swrast_InvalidateState(ctx, new_state);
-   _swsetup_InvalidateState(ctx, new_state);
-   _vbo_InvalidateState(ctx, new_state);
-   _tnl_InvalidateState(ctx, new_state);
-   _tnl_invalidate_vertex_state(ctx, new_state);
-   intel_context(ctx)->NewGLState |= new_state;
-
-   /* Todo: gather state values under which tracked parameters become
-    * invalidated, add callbacks for things like
-    * ProgramLocalParameters, etc.
-    */
-   {
-      struct i915_fragment_program *p =
-         (struct i915_fragment_program *) ctx->FragmentProgram._Current;
-      if (p && p->nr_params)
-         p->params_uptodate = 0;
-   }
-
-   if (new_state & (_NEW_FOG | _NEW_HINT | _NEW_PROGRAM))
-      i915_update_fog(ctx);
-}
-
-
-static void
-i915InitDriverFunctions(struct dd_function_table *functions)
-{
-   intelInitDriverFunctions(functions);
-   i915InitStateFunctions(functions);
-   i915InitTextureFuncs(functions);
-   i915InitFragProgFuncs(functions);
-   functions->UpdateState = i915InvalidateState;
-}
-
-
-
-GLboolean
-i915CreateContext(const __GLcontextModes * mesaVis,
-                  __DRIcontextPrivate * driContextPriv,
-                  void *sharedContextPrivate)
-{
-   struct dd_function_table functions;
-   struct i915_context *i915 =
-      (struct i915_context *) CALLOC_STRUCT(i915_context);
-   struct intel_context *intel = &i915->intel;
-   GLcontext *ctx = &intel->ctx;
-
-   if (!i915)
-      return GL_FALSE;
-
-   if (0)
-      _mesa_printf("\ntexmem-0-3 branch\n\n");
-
-   i915InitVtbl(i915);
-   i915InitMetaFuncs(i915);
-
-   i915InitDriverFunctions(&functions);
-
-   if (!intelInitContext(intel, mesaVis, driContextPriv,
-                         sharedContextPrivate, &functions)) {
-      FREE(i915);
-      return GL_FALSE;
-   }
-
-   ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
-   ctx->Const.MaxTextureImageUnits = I915_TEX_UNITS;
-   ctx->Const.MaxTextureCoordUnits = I915_TEX_UNITS;
-
-
-   /* Advertise the full hardware capabilities.  The new memory
-    * manager should cope much better with overload situations:
-    */
-   ctx->Const.MaxTextureLevels = 12;
-   ctx->Const.Max3DTextureLevels = 9;
-   ctx->Const.MaxCubeTextureLevels = 12;
-   ctx->Const.MaxTextureRectSize = (1 << 11);
-   ctx->Const.MaxTextureUnits = I915_TEX_UNITS;
-
-   /* GL_ARB_fragment_program limits - don't think Mesa actually
-    * validates programs against these, and in any case one ARB
-    * instruction can translate to more than one HW instruction, so
-    * we'll still have to check and fallback each time.
-    */
-   ctx->Const.FragmentProgram.MaxNativeTemps = I915_MAX_TEMPORARY;
-   ctx->Const.FragmentProgram.MaxNativeAttribs = 11;    /* 8 tex, 2 color, fog */
-   ctx->Const.FragmentProgram.MaxNativeParameters = I915_MAX_CONSTANT;
-   ctx->Const.FragmentProgram.MaxNativeAluInstructions = I915_MAX_ALU_INSN;
-   ctx->Const.FragmentProgram.MaxNativeTexInstructions = I915_MAX_TEX_INSN;
-   ctx->Const.FragmentProgram.MaxNativeInstructions = (I915_MAX_ALU_INSN +
-                                                       I915_MAX_TEX_INSN);
-   ctx->Const.FragmentProgram.MaxNativeTexIndirections =
-      I915_MAX_TEX_INDIRECT;
-   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);
-
-
-   _tnl_init_vertices(ctx, ctx->Const.MaxArrayLockSize + 12,
-                      36 * sizeof(GLfloat));
-
-   intel->verts = TNL_CONTEXT(ctx)->clipspace.vertex_buf;
-
-   i915InitState(i915);
-
-   return GL_TRUE;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_context.h b/src/mesa/drivers/dri/i915tex/i915_context.h
deleted file mode 100644
index d2713e88f96..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_context.h
+++ /dev/null
@@ -1,366 +0,0 @@
- /**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef I915CONTEXT_INC
-#define I915CONTEXT_INC
-
-#include "intel_context.h"
-
-#define I915_FALLBACK_TEXTURE		 0x1000
-#define I915_FALLBACK_COLORMASK		 0x2000
-#define I915_FALLBACK_STENCIL		 0x4000
-#define I915_FALLBACK_STIPPLE		 0x8000
-#define I915_FALLBACK_PROGRAM		 0x10000
-#define I915_FALLBACK_LOGICOP		 0x20000
-#define I915_FALLBACK_POLYGON_SMOOTH	 0x40000
-#define I915_FALLBACK_POINT_SMOOTH	 0x80000
-
-#define I915_UPLOAD_CTX              0x1
-#define I915_UPLOAD_BUFFERS          0x2
-#define I915_UPLOAD_STIPPLE          0x4
-#define I915_UPLOAD_PROGRAM          0x8
-#define I915_UPLOAD_CONSTANTS        0x10
-#define I915_UPLOAD_FOG              0x20
-#define I915_UPLOAD_INVARIENT        0x40
-#define I915_UPLOAD_DEFAULTS         0x80
-#define I915_UPLOAD_TEX(i)           (0x00010000<<(i))
-#define I915_UPLOAD_TEX_ALL          (0x00ff0000)
-#define I915_UPLOAD_TEX_0_SHIFT      16
-
-
-/* State structure offsets - these will probably disappear.
- */
-#define I915_DESTREG_CBUFADDR0 0
-#define I915_DESTREG_CBUFADDR1 1
-#define I915_DESTREG_DBUFADDR0 3
-#define I915_DESTREG_DBUFADDR1 4
-#define I915_DESTREG_DV0 6
-#define I915_DESTREG_DV1 7
-#define I915_DESTREG_SENABLE 8
-#define I915_DESTREG_SR0 9
-#define I915_DESTREG_SR1 10
-#define I915_DESTREG_SR2 11
-#define I915_DEST_SETUP_SIZE 12
-
-#define I915_CTXREG_STATE4		0
-#define I915_CTXREG_LI	        	1
-#define I915_CTXREG_LIS2		        2
-#define I915_CTXREG_LIS4	        	3
-#define I915_CTXREG_LIS5	        	4
-#define I915_CTXREG_LIS6	         	5
-#define I915_CTXREG_IAB   	 	6
-#define I915_CTXREG_BLENDCOLOR0		7
-#define I915_CTXREG_BLENDCOLOR1		8
-#define I915_CTX_SETUP_SIZE		9
-
-#define I915_FOGREG_COLOR		0
-#define I915_FOGREG_MODE0		1
-#define I915_FOGREG_MODE1		2
-#define I915_FOGREG_MODE2		3
-#define I915_FOGREG_MODE3		4
-#define I915_FOG_SETUP_SIZE		5
-
-#define I915_STPREG_ST0        0
-#define I915_STPREG_ST1        1
-#define I915_STP_SETUP_SIZE    2
-
-#define I915_TEXREG_MS3        1
-#define I915_TEXREG_MS4        2
-#define I915_TEXREG_SS2        3
-#define I915_TEXREG_SS3        4
-#define I915_TEXREG_SS4        5
-#define I915_TEX_SETUP_SIZE    6
-
-#define I915_DEFREG_C0    0
-#define I915_DEFREG_C1    1
-#define I915_DEFREG_S0    2
-#define I915_DEFREG_S1    3
-#define I915_DEFREG_Z0    4
-#define I915_DEFREG_Z1    5
-#define I915_DEF_SETUP_SIZE    6
-
-
-#define I915_MAX_CONSTANT      32
-#define I915_CONSTANT_SIZE     (2+(4*I915_MAX_CONSTANT))
-
-
-#define I915_PROGRAM_SIZE      192
-
-
-/* Hardware version of a parsed fragment program.  "Derived" from the
- * mesa fragment_program struct.
- */
-struct i915_fragment_program
-{
-   struct gl_fragment_program FragProg;
-
-   GLboolean translated;
-   GLboolean params_uptodate;
-   GLboolean on_hardware;
-   GLboolean error;             /* If program is malformed for any reason. */
-
-   GLuint nr_tex_indirect;
-   GLuint nr_tex_insn;
-   GLuint nr_alu_insn;
-   GLuint nr_decl_insn;
-
-
-
-
-   /* TODO: split between the stored representation of a program and
-    * the state used to build that representation.
-    */
-   GLcontext *ctx;
-
-   GLuint declarations[I915_PROGRAM_SIZE];
-   GLuint program[I915_PROGRAM_SIZE];
-
-   GLfloat constant[I915_MAX_CONSTANT][4];
-   GLuint constant_flags[I915_MAX_CONSTANT];
-   GLuint nr_constants;
-
-   GLuint *csr;                 /* Cursor, points into program.
-                                 */
-
-   GLuint *decl;                /* Cursor, points into declarations.
-                                 */
-
-   GLuint decl_s;               /* flags for which s regs need to be decl'd */
-   GLuint decl_t;               /* flags for which t regs need to be decl'd */
-
-   GLuint temp_flag;            /* Tracks temporary regs which are in
-                                 * use.
-                                 */
-
-   GLuint utemp_flag;           /* Tracks TYPE_U temporary regs which are in
-                                 * use.
-                                 */
-
-
-
-   /* Helpers for i915_fragprog.c:
-    */
-   GLuint wpos_tex;
-   GLboolean depth_written;
-
-   struct
-   {
-      GLuint reg;               /* Hardware constant idx */
-      const GLfloat *values;    /* Pointer to tracked values */
-   } param[I915_MAX_CONSTANT];
-   GLuint nr_params;
-
-
-   /* Helpers for i915_texprog.c:
-    */
-   GLuint src_texture;          /* Reg containing sampled texture color,
-                                 * else UREG_BAD.
-                                 */
-
-   GLuint src_previous;         /* Reg containing color from previous 
-                                 * stage.  May need to be decl'd.
-                                 */
-
-   GLuint last_tex_stage;       /* Number of last enabled texture unit */
-
-   struct vertex_buffer *VB;
-};
-
-
-
-
-
-
-
-#define I915_TEX_UNITS 8
-
-
-struct i915_hw_state
-{
-   GLuint Ctx[I915_CTX_SETUP_SIZE];
-   GLuint Buffer[I915_DEST_SETUP_SIZE];
-   GLuint Stipple[I915_STP_SETUP_SIZE];
-   GLuint Fog[I915_FOG_SETUP_SIZE];
-   GLuint Defaults[I915_DEF_SETUP_SIZE];
-   GLuint Tex[I915_TEX_UNITS][I915_TEX_SETUP_SIZE];
-   GLuint Constant[I915_CONSTANT_SIZE];
-   GLuint ConstantSize;
-   GLuint Program[I915_PROGRAM_SIZE];
-   GLuint ProgramSize;
-
-   /* Region pointers for relocation: 
-    */
-   struct intel_region *draw_region;
-   struct intel_region *depth_region;
-/*    struct intel_region *tex_region[I915_TEX_UNITS]; */
-
-   /* Regions aren't actually that appropriate here as the memory may
-    * be from a PBO or FBO.  Just use the buffer id.  Will have to do
-    * this for draw and depth for FBO's...
-    */
-   struct _DriBufferObject *tex_buffer[I915_TEX_UNITS];
-   GLuint tex_offset[I915_TEX_UNITS];
-
-
-   GLuint active;               /* I915_UPLOAD_* */
-   GLuint emitted;              /* I915_UPLOAD_* */
-};
-
-#define I915_FOG_PIXEL  2
-#define I915_FOG_VERTEX 1
-#define I915_FOG_NONE   0
-
-struct i915_context
-{
-   struct intel_context intel;
-
-   GLuint last_ReallyEnabled;
-   GLuint vertex_fog;
-   GLuint lodbias_ss2[MAX_TEXTURE_UNITS];
-
-
-   struct i915_fragment_program *current_program;
-
-   struct i915_hw_state meta, initial, state, *current;
-};
-
-
-#define I915_STATECHANGE(i915, flag)					\
-do {									\
-   INTEL_FIREVERTICES( &(i915)->intel );					\
-   (i915)->state.emitted &= ~(flag);					\
-} while (0)
-
-#define I915_ACTIVESTATE(i915, flag, mode)			\
-do {								\
-   INTEL_FIREVERTICES( &(i915)->intel );				\
-   if (mode)							\
-      (i915)->state.active |= (flag);				\
-   else								\
-      (i915)->state.active &= ~(flag);				\
-} while (0)
-
-
-/*======================================================================
- * i915_vtbl.c
- */
-extern void i915InitVtbl(struct i915_context *i915);
-
-extern void
-i915_state_draw_region(struct intel_context *intel,
-                       struct i915_hw_state *state,
-                       struct intel_region *color_region,
-                       struct intel_region *depth_region);
-
-
-
-#define SZ_TO_HW(sz)  ((sz-2)&0x3)
-#define EMIT_SZ(sz)   (EMIT_1F + (sz) - 1)
-#define EMIT_ATTR( ATTR, STYLE, S4, SZ )				\
-do {									\
-   intel->vertex_attrs[intel->vertex_attr_count].attrib = (ATTR);	\
-   intel->vertex_attrs[intel->vertex_attr_count].format = (STYLE);	\
-   s4 |= S4;								\
-   intel->vertex_attr_count++;						\
-   offset += (SZ);							\
-} while (0)
-
-#define EMIT_PAD( N )							\
-do {									\
-   intel->vertex_attrs[intel->vertex_attr_count].attrib = 0;		\
-   intel->vertex_attrs[intel->vertex_attr_count].format = EMIT_PAD;	\
-   intel->vertex_attrs[intel->vertex_attr_count].offset = (N);		\
-   intel->vertex_attr_count++;						\
-   offset += (N);							\
-} while (0)
-
-
-
-/*======================================================================
- * i915_context.c
- */
-extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
-                                   void *sharedContextPrivate);
-
-
-/*======================================================================
- * i915_texprog.c
- */
-extern void i915ValidateTextureProgram(struct i915_context *i915);
-
-
-/*======================================================================
- * i915_debug.c
- */
-extern void i915_disassemble_program(const GLuint * program, GLuint sz);
-extern void i915_print_ureg(const char *msg, GLuint ureg);
-
-
-/*======================================================================
- * i915_state.c
- */
-extern void i915InitStateFunctions(struct dd_function_table *functions);
-extern void i915InitState(struct i915_context *i915);
-extern void i915_update_fog(GLcontext * ctx);
-
-
-/*======================================================================
- * i915_tex.c
- */
-extern void i915UpdateTextureState(struct intel_context *intel);
-extern void i915InitTextureFuncs(struct dd_function_table *functions);
-
-/*======================================================================
- * i915_metaops.c
- */
-void i915InitMetaFuncs(struct i915_context *i915);
-
-
-/*======================================================================
- * i915_fragprog.c
- */
-extern void i915ValidateFragmentProgram(struct i915_context *i915);
-extern void i915InitFragProgFuncs(struct dd_function_table *functions);
-
-/*======================================================================
- * Inline conversion functions.  These are better-typed than the
- * macros used previously:
- */
-static INLINE struct i915_context *
-i915_context(GLcontext * ctx)
-{
-   return (struct i915_context *) ctx;
-}
-
-
-
-#define I915_CONTEXT(ctx)	i915_context(ctx)
-
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/i915_fragprog.c b/src/mesa/drivers/dri/i915tex/i915_fragprog.c
deleted file mode 100644
index a4b22a0c32c..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_fragprog.c
+++ /dev/null
@@ -1,1073 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-#include "intel_batchbuffer.h"
-
-#include "i915_reg.h"
-#include "i915_context.h"
-#include "i915_program.h"
-
-#include "prog_instruction.h"
-#include "prog_parameter.h"
-#include "program.h"
-#include "programopt.h"
-
-
-
-/* 1, -1/3!, 1/5!, -1/7! */
-static const GLfloat sin_constants[4] = { 1.0,
-   -1.0 / (3 * 2 * 1),
-   1.0 / (5 * 4 * 3 * 2 * 1),
-   -1.0 / (7 * 6 * 5 * 4 * 3 * 2 * 1)
-};
-
-/* 1, -1/2!, 1/4!, -1/6! */
-static const GLfloat cos_constants[4] = { 1.0,
-   -1.0 / (2 * 1),
-   1.0 / (4 * 3 * 2 * 1),
-   -1.0 / (6 * 5 * 4 * 3 * 2 * 1)
-};
-
-/**
- * Retrieve a ureg for the given source register.  Will emit
- * constants, apply swizzling and negation as needed.
- */
-static GLuint
-src_vector(struct i915_fragment_program *p,
-           const struct prog_src_register *source,
-           const struct gl_fragment_program *program)
-{
-   GLuint src;
-
-   switch (source->File) {
-
-      /* Registers:
-       */
-   case PROGRAM_TEMPORARY:
-      if (source->Index >= I915_MAX_TEMPORARY) {
-         i915_program_error(p, "Exceeded max temporary reg");
-         return 0;
-      }
-      src = UREG(REG_TYPE_R, source->Index);
-      break;
-   case PROGRAM_INPUT:
-      switch (source->Index) {
-      case FRAG_ATTRIB_WPOS:
-         src = i915_emit_decl(p, REG_TYPE_T, p->wpos_tex, D0_CHANNEL_ALL);
-         break;
-      case FRAG_ATTRIB_COL0:
-         src = i915_emit_decl(p, REG_TYPE_T, T_DIFFUSE, D0_CHANNEL_ALL);
-         break;
-      case FRAG_ATTRIB_COL1:
-         src = i915_emit_decl(p, REG_TYPE_T, T_SPECULAR, D0_CHANNEL_XYZ);
-         src = swizzle(src, X, Y, Z, ONE);
-         break;
-      case FRAG_ATTRIB_FOGC:
-         src = i915_emit_decl(p, REG_TYPE_T, T_FOG_W, D0_CHANNEL_W);
-         src = swizzle(src, W, W, W, W);
-         break;
-      case FRAG_ATTRIB_TEX0:
-      case FRAG_ATTRIB_TEX1:
-      case FRAG_ATTRIB_TEX2:
-      case FRAG_ATTRIB_TEX3:
-      case FRAG_ATTRIB_TEX4:
-      case FRAG_ATTRIB_TEX5:
-      case FRAG_ATTRIB_TEX6:
-      case FRAG_ATTRIB_TEX7:
-         src = i915_emit_decl(p, REG_TYPE_T,
-                              T_TEX0 + (source->Index - FRAG_ATTRIB_TEX0),
-                              D0_CHANNEL_ALL);
-         break;
-
-      default:
-         i915_program_error(p, "Bad source->Index");
-         return 0;
-      }
-      break;
-
-      /* Various paramters and env values.  All emitted to
-       * hardware as program constants.
-       */
-   case PROGRAM_LOCAL_PARAM:
-      src = i915_emit_param4fv(p, program->Base.LocalParams[source->Index]);
-      break;
-
-   case PROGRAM_ENV_PARAM:
-      src =
-         i915_emit_param4fv(p,
-                            p->ctx->FragmentProgram.Parameters[source->
-                                                               Index]);
-      break;
-
-   case PROGRAM_CONSTANT:
-   case PROGRAM_STATE_VAR:
-   case PROGRAM_NAMED_PARAM:
-      src =
-         i915_emit_param4fv(p,
-                            program->Base.Parameters->ParameterValues[source->
-                                                                      Index]);
-      break;
-
-   default:
-      i915_program_error(p, "Bad source->File");
-      return 0;
-   }
-
-   src = swizzle(src,
-                 GET_SWZ(source->Swizzle, 0),
-                 GET_SWZ(source->Swizzle, 1),
-                 GET_SWZ(source->Swizzle, 2), GET_SWZ(source->Swizzle, 3));
-
-   if (source->NegateBase)
-      src = negate(src,
-                   GET_BIT(source->NegateBase, 0),
-                   GET_BIT(source->NegateBase, 1),
-                   GET_BIT(source->NegateBase, 2),
-                   GET_BIT(source->NegateBase, 3));
-
-   return src;
-}
-
-
-static GLuint
-get_result_vector(struct i915_fragment_program *p,
-                  const struct prog_instruction *inst)
-{
-   switch (inst->DstReg.File) {
-   case PROGRAM_OUTPUT:
-      switch (inst->DstReg.Index) {
-      case FRAG_RESULT_COLR:
-         return UREG(REG_TYPE_OC, 0);
-      case FRAG_RESULT_DEPR:
-         p->depth_written = 1;
-         return UREG(REG_TYPE_OD, 0);
-      default:
-         i915_program_error(p, "Bad inst->DstReg.Index");
-         return 0;
-      }
-   case PROGRAM_TEMPORARY:
-      return UREG(REG_TYPE_R, inst->DstReg.Index);
-   default:
-      i915_program_error(p, "Bad inst->DstReg.File");
-      return 0;
-   }
-}
-
-static GLuint
-get_result_flags(const struct prog_instruction *inst)
-{
-   GLuint flags = 0;
-
-   if (inst->SaturateMode == SATURATE_ZERO_ONE)
-      flags |= A0_DEST_SATURATE;
-   if (inst->DstReg.WriteMask & WRITEMASK_X)
-      flags |= A0_DEST_CHANNEL_X;
-   if (inst->DstReg.WriteMask & WRITEMASK_Y)
-      flags |= A0_DEST_CHANNEL_Y;
-   if (inst->DstReg.WriteMask & WRITEMASK_Z)
-      flags |= A0_DEST_CHANNEL_Z;
-   if (inst->DstReg.WriteMask & WRITEMASK_W)
-      flags |= A0_DEST_CHANNEL_W;
-
-   return flags;
-}
-
-static GLuint
-translate_tex_src_target(struct i915_fragment_program *p, GLubyte bit)
-{
-   switch (bit) {
-   case TEXTURE_1D_INDEX:
-      return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_2D_INDEX:
-      return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_RECT_INDEX:
-      return D0_SAMPLE_TYPE_2D;
-   case TEXTURE_3D_INDEX:
-      return D0_SAMPLE_TYPE_VOLUME;
-   case TEXTURE_CUBE_INDEX:
-      return D0_SAMPLE_TYPE_CUBE;
-   default:
-      i915_program_error(p, "TexSrcBit");
-      return 0;
-   }
-}
-
-#define EMIT_TEX( OP )						\
-do {								\
-   GLuint dim = translate_tex_src_target( p, inst->TexSrcTarget );	\
-   GLuint sampler = i915_emit_decl(p, REG_TYPE_S,		\
-				  inst->TexSrcUnit, dim);	\
-   GLuint coord = src_vector( p, &inst->SrcReg[0], program);	\
-   /* Texel lookup */						\
-								\
-   i915_emit_texld( p,						\
-	       get_result_vector( p, inst ),			\
-	       get_result_flags( inst ),			\
-	       sampler,						\
-	       coord,						\
-	       OP);						\
-} while (0)
-
-#define EMIT_ARITH( OP, N )						\
-do {									\
-   i915_emit_arith( p,							\
-	       OP,							\
-	       get_result_vector( p, inst ), 				\
-	       get_result_flags( inst ), 0,			\
-	       (N<1)?0:src_vector( p, &inst->SrcReg[0], program),	\
-	       (N<2)?0:src_vector( p, &inst->SrcReg[1], program),	\
-	       (N<3)?0:src_vector( p, &inst->SrcReg[2], program));	\
-} while (0)
-
-#define EMIT_1ARG_ARITH( OP ) EMIT_ARITH( OP, 1 )
-#define EMIT_2ARG_ARITH( OP ) EMIT_ARITH( OP, 2 )
-#define EMIT_3ARG_ARITH( OP ) EMIT_ARITH( OP, 3 )
-
-
-/* Possible concerns:
- *
- * SIN, COS -- could use another taylor step?
- * LIT      -- results seem a little different to sw mesa
- * LOG      -- different to mesa on negative numbers, but this is conformant.
- * 
- * Parse failures -- Mesa doesn't currently give a good indication
- * internally whether a particular program string parsed or not.  This
- * can lead to confusion -- hopefully we cope with it ok now.
- *
- */
-static void
-upload_program(struct i915_fragment_program *p)
-{
-   const struct gl_fragment_program *program =
-      p->ctx->FragmentProgram._Current;
-   const struct prog_instruction *inst = program->Base.Instructions;
-
-/*    _mesa_debug_fp_inst(program->Base.NumInstructions, inst); */
-
-   /* Is this a parse-failed program?  Ensure a valid program is
-    * loaded, as the flagging of an error isn't sufficient to stop
-    * this being uploaded to hardware.
-    */
-   if (inst[0].Opcode == OPCODE_END) {
-      GLuint tmp = i915_get_utemp(p);
-      i915_emit_arith(p,
-                      A0_MOV,
-                      UREG(REG_TYPE_OC, 0),
-                      A0_DEST_CHANNEL_ALL, 0,
-                      swizzle(tmp, ONE, ZERO, ONE, ONE), 0, 0);
-      return;
-   }
-
-   while (1) {
-      GLuint src0, src1, src2, flags;
-      GLuint tmp = 0;
-
-      switch (inst->Opcode) {
-      case OPCODE_ABS:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         i915_emit_arith(p,
-                         A0_MAX,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         src0, negate(src0, 1, 1, 1, 1), 0);
-         break;
-
-      case OPCODE_ADD:
-         EMIT_2ARG_ARITH(A0_ADD);
-         break;
-
-      case OPCODE_CMP:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-         src2 = src_vector(p, &inst->SrcReg[2], program);
-         i915_emit_arith(p, A0_CMP, get_result_vector(p, inst), get_result_flags(inst), 0, src0, src2, src1);   /* NOTE: order of src2, src1 */
-         break;
-
-      case OPCODE_COS:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         tmp = i915_get_utemp(p);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_X, 0,
-                         src0, i915_emit_const1f(p, 1.0 / (M_PI * 2)), 0);
-
-         i915_emit_arith(p, A0_MOD, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
-
-         /* By choosing different taylor constants, could get rid of this mul:
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_X, 0,
-                         tmp, i915_emit_const1f(p, (M_PI * 2)), 0);
-
-         /* 
-          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, 1
-          * t0 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-          * result = DP4 t0, cos_constants
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_XY, 0,
-                         swizzle(tmp, X, X, ONE, ONE),
-                         swizzle(tmp, X, ONE, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_XYZ, 0,
-                         swizzle(tmp, X, Y, X, ONE),
-                         swizzle(tmp, X, X, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_XYZ, 0,
-                         swizzle(tmp, X, X, Z, ONE),
-                         swizzle(tmp, Z, ONE, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_DP4,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(tmp, ONE, Z, Y, X),
-                         i915_emit_const4fv(p, cos_constants), 0);
-
-         break;
-
-      case OPCODE_DP3:
-         EMIT_2ARG_ARITH(A0_DP3);
-         break;
-
-      case OPCODE_DP4:
-         EMIT_2ARG_ARITH(A0_DP4);
-         break;
-
-      case OPCODE_DPH:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-
-         i915_emit_arith(p,
-                         A0_DP4,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, X, Y, Z, ONE), src1, 0);
-         break;
-
-      case OPCODE_DST:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-
-         /* result[0] = 1    * 1;
-          * result[1] = a[1] * b[1];
-          * result[2] = a[2] * 1;
-          * result[3] = 1    * b[3];
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, ONE, Y, Z, ONE),
-                         swizzle(src1, ONE, Y, ONE, W), 0);
-         break;
-
-      case OPCODE_EX2:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-
-         i915_emit_arith(p,
-                         A0_EXP,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, X, X, X, X), 0, 0);
-         break;
-
-      case OPCODE_FLR:
-         EMIT_1ARG_ARITH(A0_FLR);
-         break;
-
-      case OPCODE_FRC:
-         EMIT_1ARG_ARITH(A0_FRC);
-         break;
-
-      case OPCODE_KIL:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         tmp = i915_get_utemp(p);
-
-         i915_emit_texld(p, tmp, A0_DEST_CHANNEL_ALL,   /* use a dummy dest reg */
-                         0, src0, T0_TEXKILL);
-         break;
-
-      case OPCODE_LG2:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-
-         i915_emit_arith(p,
-                         A0_LOG,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, X, X, X, X), 0, 0);
-         break;
-
-      case OPCODE_LIT:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         tmp = i915_get_utemp(p);
-
-         /* tmp = max( a.xyzw, a.00zw )
-          * XXX: Clamp tmp.w to -128..128
-          * tmp.y = log(tmp.y)
-          * tmp.y = tmp.w * tmp.y
-          * tmp.y = exp(tmp.y)
-          * result = cmp (a.11-x1, a.1x01, a.1xy1 )
-          */
-         i915_emit_arith(p, A0_MAX, tmp, A0_DEST_CHANNEL_ALL, 0,
-                         src0, swizzle(src0, ZERO, ZERO, Z, W), 0);
-
-         i915_emit_arith(p, A0_LOG, tmp, A0_DEST_CHANNEL_Y, 0,
-                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
-
-         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_Y, 0,
-                         swizzle(tmp, ZERO, Y, ZERO, ZERO),
-                         swizzle(tmp, ZERO, W, ZERO, ZERO), 0);
-
-         i915_emit_arith(p, A0_EXP, tmp, A0_DEST_CHANNEL_Y, 0,
-                         swizzle(tmp, Y, Y, Y, Y), 0, 0);
-
-         i915_emit_arith(p, A0_CMP,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         negate(swizzle(tmp, ONE, ONE, X, ONE), 0, 0, 1, 0),
-                         swizzle(tmp, ONE, X, ZERO, ONE),
-                         swizzle(tmp, ONE, X, Y, ONE));
-
-         break;
-
-      case OPCODE_LRP:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-         src2 = src_vector(p, &inst->SrcReg[2], program);
-         flags = get_result_flags(inst);
-         tmp = i915_get_utemp(p);
-
-         /* b*a + c*(1-a)
-          *
-          * b*a + c - ca 
-          *
-          * tmp = b*a + c, 
-          * result = (-c)*a + tmp 
-          */
-         i915_emit_arith(p, A0_MAD, tmp,
-                         flags & A0_DEST_CHANNEL_ALL, 0, src1, src0, src2);
-
-         i915_emit_arith(p, A0_MAD,
-                         get_result_vector(p, inst),
-                         flags, 0, negate(src2, 1, 1, 1, 1), src0, tmp);
-         break;
-
-      case OPCODE_MAD:
-         EMIT_3ARG_ARITH(A0_MAD);
-         break;
-
-      case OPCODE_MAX:
-         EMIT_2ARG_ARITH(A0_MAX);
-         break;
-
-      case OPCODE_MIN:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-         tmp = i915_get_utemp(p);
-         flags = get_result_flags(inst);
-
-         i915_emit_arith(p,
-                         A0_MAX,
-                         tmp, flags & A0_DEST_CHANNEL_ALL, 0,
-                         negate(src0, 1, 1, 1, 1),
-                         negate(src1, 1, 1, 1, 1), 0);
-
-         i915_emit_arith(p,
-                         A0_MOV,
-                         get_result_vector(p, inst),
-                         flags, 0, negate(tmp, 1, 1, 1, 1), 0, 0);
-         break;
-
-      case OPCODE_MOV:
-         EMIT_1ARG_ARITH(A0_MOV);
-         break;
-
-      case OPCODE_MUL:
-         EMIT_2ARG_ARITH(A0_MUL);
-         break;
-
-      case OPCODE_POW:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-         tmp = i915_get_utemp(p);
-         flags = get_result_flags(inst);
-
-         /* XXX: masking on intermediate values, here and elsewhere.
-          */
-         i915_emit_arith(p,
-                         A0_LOG,
-                         tmp, A0_DEST_CHANNEL_X, 0,
-                         swizzle(src0, X, X, X, X), 0, 0);
-
-         i915_emit_arith(p, A0_MUL, tmp, A0_DEST_CHANNEL_X, 0, tmp, src1, 0);
-
-
-         i915_emit_arith(p,
-                         A0_EXP,
-                         get_result_vector(p, inst),
-                         flags, 0, swizzle(tmp, X, X, X, X), 0, 0);
-
-         break;
-
-      case OPCODE_RCP:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-
-         i915_emit_arith(p,
-                         A0_RCP,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, X, X, X, X), 0, 0);
-         break;
-
-      case OPCODE_RSQ:
-
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-
-         i915_emit_arith(p,
-                         A0_RSQ,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, X, X, X, X), 0, 0);
-         break;
-
-      case OPCODE_SCS:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         tmp = i915_get_utemp(p);
-
-         /* 
-          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-          * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-          * scs.x = DP4 t1, sin_constants
-          * t1 = MUL t0.xxz1 t0.z111    ; x^6 x^4 x^2 1
-          * scs.y = DP4 t1, cos_constants
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_XY, 0,
-                         swizzle(src0, X, X, ONE, ONE),
-                         swizzle(src0, X, ONE, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_ALL, 0,
-                         swizzle(tmp, X, Y, X, Y),
-                         swizzle(tmp, X, X, ONE, ONE), 0);
-
-         if (inst->DstReg.WriteMask & WRITEMASK_Y) {
-            GLuint tmp1;
-
-            if (inst->DstReg.WriteMask & WRITEMASK_X)
-               tmp1 = i915_get_utemp(p);
-            else
-               tmp1 = tmp;
-
-            i915_emit_arith(p,
-                            A0_MUL,
-                            tmp1, A0_DEST_CHANNEL_ALL, 0,
-                            swizzle(tmp, X, Y, Y, W),
-                            swizzle(tmp, X, Z, ONE, ONE), 0);
-
-            i915_emit_arith(p,
-                            A0_DP4,
-                            get_result_vector(p, inst),
-                            A0_DEST_CHANNEL_Y, 0,
-                            swizzle(tmp1, W, Z, Y, X),
-                            i915_emit_const4fv(p, sin_constants), 0);
-         }
-
-         if (inst->DstReg.WriteMask & WRITEMASK_X) {
-            i915_emit_arith(p,
-                            A0_MUL,
-                            tmp, A0_DEST_CHANNEL_XYZ, 0,
-                            swizzle(tmp, X, X, Z, ONE),
-                            swizzle(tmp, Z, ONE, ONE, ONE), 0);
-
-            i915_emit_arith(p,
-                            A0_DP4,
-                            get_result_vector(p, inst),
-                            A0_DEST_CHANNEL_X, 0,
-                            swizzle(tmp, ONE, Z, Y, X),
-                            i915_emit_const4fv(p, cos_constants), 0);
-         }
-         break;
-
-      case OPCODE_SGE:
-         EMIT_2ARG_ARITH(A0_SGE);
-         break;
-
-      case OPCODE_SIN:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         tmp = i915_get_utemp(p);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_X, 0,
-                         src0, i915_emit_const1f(p, 1.0 / (M_PI * 2)), 0);
-
-         i915_emit_arith(p, A0_MOD, tmp, A0_DEST_CHANNEL_X, 0, tmp, 0, 0);
-
-         /* By choosing different taylor constants, could get rid of this mul:
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_X, 0,
-                         tmp, i915_emit_const1f(p, (M_PI * 2)), 0);
-
-         /* 
-          * t0.xy = MUL x.xx11, x.x1111  ; x^2, x, 1, 1
-          * t0 = MUL t0.xyxy t0.xx11 ; x^4, x^3, x^2, x
-          * t1 = MUL t0.xyyw t0.yz11    ; x^7 x^5 x^3 x
-          * result = DP4 t1.wzyx, sin_constants
-          */
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_XY, 0,
-                         swizzle(tmp, X, X, ONE, ONE),
-                         swizzle(tmp, X, ONE, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_ALL, 0,
-                         swizzle(tmp, X, Y, X, Y),
-                         swizzle(tmp, X, X, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_ALL, 0,
-                         swizzle(tmp, X, Y, Y, W),
-                         swizzle(tmp, X, Z, ONE, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_DP4,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(tmp, W, Z, Y, X),
-                         i915_emit_const4fv(p, sin_constants), 0);
-         break;
-
-      case OPCODE_SLT:
-         EMIT_2ARG_ARITH(A0_SLT);
-         break;
-
-      case OPCODE_SUB:
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-
-         i915_emit_arith(p,
-                         A0_ADD,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         src0, negate(src1, 1, 1, 1, 1), 0);
-         break;
-
-      case OPCODE_SWZ:
-         EMIT_1ARG_ARITH(A0_MOV);       /* extended swizzle handled natively */
-         break;
-
-      case OPCODE_TEX:
-         EMIT_TEX(T0_TEXLD);
-         break;
-
-      case OPCODE_TXB:
-         EMIT_TEX(T0_TEXLDB);
-         break;
-
-      case OPCODE_TXP:
-         EMIT_TEX(T0_TEXLDP);
-         break;
-
-      case OPCODE_XPD:
-         /* Cross product:
-          *      result.x = src0.y * src1.z - src0.z * src1.y;
-          *      result.y = src0.z * src1.x - src0.x * src1.z;
-          *      result.z = src0.x * src1.y - src0.y * src1.x;
-          *      result.w = undef;
-          */
-         src0 = src_vector(p, &inst->SrcReg[0], program);
-         src1 = src_vector(p, &inst->SrcReg[1], program);
-         tmp = i915_get_utemp(p);
-
-         i915_emit_arith(p,
-                         A0_MUL,
-                         tmp, A0_DEST_CHANNEL_ALL, 0,
-                         swizzle(src0, Z, X, Y, ONE),
-                         swizzle(src1, Y, Z, X, ONE), 0);
-
-         i915_emit_arith(p,
-                         A0_MAD,
-                         get_result_vector(p, inst),
-                         get_result_flags(inst), 0,
-                         swizzle(src0, Y, Z, X, ONE),
-                         swizzle(src1, Z, X, Y, ONE),
-                         negate(tmp, 1, 1, 1, 0));
-         break;
-
-      case OPCODE_END:
-         return;
-
-      default:
-         i915_program_error(p, "bad opcode");
-         return;
-      }
-
-      inst++;
-      i915_release_utemps(p);
-   }
-}
-
-/* Rather than trying to intercept and jiggle depth writes during
- * emit, just move the value into its correct position at the end of
- * the program:
- */
-static void
-fixup_depth_write(struct i915_fragment_program *p)
-{
-   if (p->depth_written) {
-      GLuint depth = UREG(REG_TYPE_OD, 0);
-
-      i915_emit_arith(p,
-                      A0_MOV,
-                      depth, A0_DEST_CHANNEL_W, 0,
-                      swizzle(depth, X, Y, Z, Z), 0, 0);
-   }
-}
-
-
-static void
-check_wpos(struct i915_fragment_program *p)
-{
-   GLuint inputs = p->FragProg.Base.InputsRead;
-   GLint i;
-
-   p->wpos_tex = -1;
-
-   for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
-      if (inputs & FRAG_BIT_TEX(i))
-         continue;
-      else if (inputs & FRAG_BIT_WPOS) {
-         p->wpos_tex = i;
-         inputs &= ~FRAG_BIT_WPOS;
-      }
-   }
-
-   if (inputs & FRAG_BIT_WPOS) {
-      i915_program_error(p, "No free texcoord for wpos value");
-   }
-}
-
-
-static void
-translate_program(struct i915_fragment_program *p)
-{
-   struct i915_context *i915 = I915_CONTEXT(p->ctx);
-
-   i915_init_program(i915, p);
-   check_wpos(p);
-   upload_program(p);
-   fixup_depth_write(p);
-   i915_fini_program(p);
-
-   p->translated = 1;
-}
-
-
-static void
-track_params(struct i915_fragment_program *p)
-{
-   GLint i;
-
-   if (p->nr_params)
-      _mesa_load_state_parameters(p->ctx, p->FragProg.Base.Parameters);
-
-   for (i = 0; i < p->nr_params; i++) {
-      GLint reg = p->param[i].reg;
-      COPY_4V(p->constant[reg], p->param[i].values);
-   }
-
-   p->params_uptodate = 1;
-   p->on_hardware = 0;          /* overkill */
-}
-
-
-static void
-i915BindProgram(GLcontext * ctx, GLenum target, struct gl_program *prog)
-{
-   if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_context *i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
-
-      if (i915->current_program == p)
-         return;
-
-      if (i915->current_program) {
-         i915->current_program->on_hardware = 0;
-         i915->current_program->params_uptodate = 0;
-      }
-
-      i915->current_program = p;
-
-      assert(p->on_hardware == 0);
-      assert(p->params_uptodate == 0);
-
-   }
-}
-
-static struct gl_program *
-i915NewProgram(GLcontext * ctx, GLenum target, GLuint id)
-{
-   switch (target) {
-   case GL_VERTEX_PROGRAM_ARB:
-      return _mesa_init_vertex_program(ctx, CALLOC_STRUCT(gl_vertex_program),
-                                       target, id);
-
-   case GL_FRAGMENT_PROGRAM_ARB:{
-         struct i915_fragment_program *prog =
-            CALLOC_STRUCT(i915_fragment_program);
-         if (prog) {
-            i915_init_program(I915_CONTEXT(ctx), prog);
-
-            return _mesa_init_fragment_program(ctx, &prog->FragProg,
-                                               target, id);
-         }
-         else
-            return NULL;
-      }
-
-   default:
-      /* Just fallback:
-       */
-      return _mesa_new_program(ctx, target, id);
-   }
-}
-
-static void
-i915DeleteProgram(GLcontext * ctx, struct gl_program *prog)
-{
-   if (prog->Target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_context *i915 = I915_CONTEXT(ctx);
-      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
-
-      if (i915->current_program == p)
-         i915->current_program = 0;
-   }
-
-   _mesa_delete_program(ctx, prog);
-}
-
-
-static GLboolean
-i915IsProgramNative(GLcontext * ctx, GLenum target, struct gl_program *prog)
-{
-   if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
-
-      if (!p->translated)
-         translate_program(p);
-
-      return !p->error;
-   }
-   else
-      return GL_TRUE;
-}
-
-static void
-i915ProgramStringNotify(GLcontext * ctx,
-                        GLenum target, struct gl_program *prog)
-{
-   if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      struct i915_fragment_program *p = (struct i915_fragment_program *) prog;
-      p->translated = 0;
-
-      /* Hack: make sure fog is correctly enabled according to this
-       * fragment program's fog options.
-       */
-      if (p->FragProg.FogOption) {
-         /* add extra instructions to do fog, then turn off FogOption field */
-         _mesa_append_fog_code(ctx, &p->FragProg);
-         p->FragProg.FogOption = GL_NONE;
-      }
-   }
-
-   _tnl_program_string(ctx, target, prog);
-}
-
-
-void
-i915ValidateFragmentProgram(struct i915_context *i915)
-{
-   GLcontext *ctx = &i915->intel.ctx;
-   struct intel_context *intel = intel_context(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-
-   struct i915_fragment_program *p =
-      (struct i915_fragment_program *) ctx->FragmentProgram._Current;
-
-   const GLuint inputsRead = p->FragProg.Base.InputsRead;
-   GLuint s4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_VFMT_MASK;
-   GLuint s2 = S2_TEXCOORD_NONE;
-   int i, offset = 0;
-
-   if (i915->current_program != p) {
-      if (i915->current_program) {
-         i915->current_program->on_hardware = 0;
-         i915->current_program->params_uptodate = 0;
-      }
-
-      i915->current_program = p;
-   }
-
-
-   /* Important:
-    */
-   VB->AttribPtr[VERT_ATTRIB_POS] = VB->NdcPtr;
-
-   if (!p->translated)
-      translate_program(p);
-
-   intel->vertex_attr_count = 0;
-   intel->wpos_offset = 0;
-   intel->wpos_size = 0;
-   intel->coloroffset = 0;
-   intel->specoffset = 0;
-
-   if (inputsRead & FRAG_BITS_TEX_ANY) {
-      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_4F_VIEWPORT, S4_VFMT_XYZW, 16);
-   }
-   else {
-      EMIT_ATTR(_TNL_ATTRIB_POS, EMIT_3F_VIEWPORT, S4_VFMT_XYZ, 12);
-   }
-
-   if (inputsRead & FRAG_BIT_COL0) {
-      intel->coloroffset = offset / 4;
-      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);
-   }
-
-   /* 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)) {
-         int sz = VB->TexCoordPtr[i]->size;
-
-         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(sz));
-
-         EMIT_ATTR(_TNL_ATTRIB_TEX0 + i, EMIT_SZ(sz), 0, sz * 4);
-      }
-      else if (i == p->wpos_tex) {
-
-         /* If WPOS is required, duplicate the XYZ position data in an
-          * unused texture coordinate:
-          */
-         s2 &= ~S2_TEXCOORD_FMT(i, S2_TEXCOORD_FMT0_MASK);
-         s2 |= S2_TEXCOORD_FMT(i, SZ_TO_HW(3));
-
-         intel->wpos_offset = offset;
-         intel->wpos_size = 3 * sizeof(GLuint);
-
-         EMIT_PAD(intel->wpos_size);
-      }
-   }
-
-   if (s2 != i915->state.Ctx[I915_CTXREG_LIS2] ||
-       s4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-      int k;
-
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-
-      /* Must do this *after* statechange, so as not to affect
-       * buffered vertices reliant on the old state:
-       */
-      intel->vertex_size = _tnl_install_attrs(&intel->ctx,
-                                              intel->vertex_attrs,
-                                              intel->vertex_attr_count,
-                                              intel->ViewportMatrix.m, 0);
-
-      intel->vertex_size >>= 2;
-
-      i915->state.Ctx[I915_CTXREG_LIS2] = s2;
-      i915->state.Ctx[I915_CTXREG_LIS4] = s4;
-
-      k = intel->vtbl.check_vertex_size(intel, intel->vertex_size);
-      assert(k);
-   }
-
-   if (!p->params_uptodate)
-      track_params(p);
-
-   if (!p->on_hardware)
-      i915_upload_program(i915, p);
-}
-
-void
-i915InitFragProgFuncs(struct dd_function_table *functions)
-{
-   functions->BindProgram = i915BindProgram;
-   functions->NewProgram = i915NewProgram;
-   functions->DeleteProgram = i915DeleteProgram;
-   functions->IsProgramNative = i915IsProgramNative;
-   functions->ProgramStringNotify = i915ProgramStringNotify;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_metaops.c b/src/mesa/drivers/dri/i915tex/i915_metaops.c
deleted file mode 100644
index 3ab62bc806b..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_metaops.c
+++ /dev/null
@@ -1,509 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "enums.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "utils.h"
-
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-#include "intel_ioctl.h"
-#include "intel_regions.h"
-#include "intel_rotate.h"
-
-#include "i915_context.h"
-#include "i915_reg.h"
-
-/* We touch almost everything:
- */
-#define ACTIVE (I915_UPLOAD_INVARIENT | 	\
-		I915_UPLOAD_CTX |		\
-		I915_UPLOAD_BUFFERS |		\
-		I915_UPLOAD_STIPPLE |		\
-                I915_UPLOAD_PROGRAM | 		\
-                I915_UPLOAD_FOG | 		\
-		I915_UPLOAD_TEX(0))
-
-#define SET_STATE( i915, STATE )		\
-do {						\
-   i915->current->emitted &= ~ACTIVE;		\
-   i915->current = &i915->STATE;		\
-   i915->current->emitted &= ~ACTIVE;		\
-} while (0)
-
-
-static void
-meta_no_stencil_write(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_FALSE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
-                                         S5_STENCIL_WRITE_ENABLE);
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-static void
-meta_no_depth_write(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_FALSE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~(S6_DEPTH_TEST_ENABLE |
-                                         S6_DEPTH_WRITE_ENABLE);
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-static void
-meta_depth_replace(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   /* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_TRUE )
-    * ctx->Driver.DepthMask( ctx, GL_TRUE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS6] |= (S6_DEPTH_TEST_ENABLE |
-                                        S6_DEPTH_WRITE_ENABLE);
-
-   /* ctx->Driver.DepthFunc( ctx, GL_REPLACE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
-   i915->meta.Ctx[I915_CTXREG_LIS6] |=
-      COMPAREFUNC_ALWAYS << S6_DEPTH_TEST_FUNC_SHIFT;
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-
-/* Set stencil unit to replace always with the reference value.
- */
-static void
-meta_stencil_replace(struct intel_context *intel,
-                     GLuint s_mask, GLuint s_clear)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   GLuint op = STENCILOP_REPLACE;
-   GLuint func = COMPAREFUNC_ALWAYS;
-
-   /* ctx->Driver.Enable( ctx, GL_STENCIL_TEST, GL_TRUE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
-                                        S5_STENCIL_WRITE_ENABLE);
-
-   /* ctx->Driver.StencilMask( ctx, s_mask )
-    */
-   i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
-
-   i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-                                          STENCIL_WRITE_MASK(s_mask));
-
-   /* ctx->Driver.StencilOp( ctx, GL_REPLACE, GL_REPLACE, GL_REPLACE )
-    */
-   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-                                         S5_STENCIL_PASS_Z_FAIL_MASK |
-                                         S5_STENCIL_PASS_Z_PASS_MASK);
-
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((op << S5_STENCIL_FAIL_SHIFT) |
-                                        (op << S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-                                        (op << S5_STENCIL_PASS_Z_PASS_SHIFT));
-
-
-   /* ctx->Driver.StencilFunc( ctx, GL_ALWAYS, s_ref, ~0 )
-    */
-   i915->meta.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
-   i915->meta.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-                                          STENCIL_TEST_MASK(0xff));
-
-   i915->meta.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-                                         S5_STENCIL_TEST_FUNC_MASK);
-
-   i915->meta.Ctx[I915_CTXREG_LIS5] |= ((s_clear << S5_STENCIL_REF_SHIFT) |
-                                        (func << S5_STENCIL_TEST_FUNC_SHIFT));
-
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-
-static void
-meta_color_mask(struct intel_context *intel, GLboolean state)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   const GLuint mask = (S5_WRITEDISABLE_RED |
-                        S5_WRITEDISABLE_GREEN |
-                        S5_WRITEDISABLE_BLUE | S5_WRITEDISABLE_ALPHA);
-
-   /* Copy colormask state from "regular" hw context.
-    */
-   if (state) {
-      i915->meta.Ctx[I915_CTXREG_LIS5] &= ~mask;
-      i915->meta.Ctx[I915_CTXREG_LIS5] |=
-         (i915->state.Ctx[I915_CTXREG_LIS5] & mask);
-   }
-   else
-      i915->meta.Ctx[I915_CTXREG_LIS5] |= mask;
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-
-
-static void
-meta_import_pixel_state(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   memcpy(i915->meta.Fog, i915->state.Fog, I915_FOG_SETUP_SIZE * 4);
-
-   i915->meta.Ctx[I915_CTXREG_LIS5] = i915->state.Ctx[I915_CTXREG_LIS5];
-   i915->meta.Ctx[I915_CTXREG_LIS6] = i915->state.Ctx[I915_CTXREG_LIS6];
-   i915->meta.Ctx[I915_CTXREG_STATE4] = i915->state.Ctx[I915_CTXREG_STATE4];
-   i915->meta.Ctx[I915_CTXREG_BLENDCOLOR1] =
-      i915->state.Ctx[I915_CTXREG_BLENDCOLOR1];
-   i915->meta.Ctx[I915_CTXREG_IAB] = i915->state.Ctx[I915_CTXREG_IAB];
-
-   i915->meta.Buffer[I915_DESTREG_SENABLE] =
-      i915->state.Buffer[I915_DESTREG_SENABLE];
-   i915->meta.Buffer[I915_DESTREG_SR1] = i915->state.Buffer[I915_DESTREG_SR1];
-   i915->meta.Buffer[I915_DESTREG_SR2] = i915->state.Buffer[I915_DESTREG_SR2];
-
-   i915->meta.emitted &= ~I915_UPLOAD_FOG;
-   i915->meta.emitted &= ~I915_UPLOAD_BUFFERS;
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-
-
-
-#define REG( type, nr ) (((type)<<5)|(nr))
-
-#define REG_R(x)       REG(REG_TYPE_R, x)
-#define REG_T(x)       REG(REG_TYPE_T, x)
-#define REG_CONST(x)   REG(REG_TYPE_CONST, x)
-#define REG_S(x)       REG(REG_TYPE_S, x)
-#define REG_OC         REG(REG_TYPE_OC, 0)
-#define REG_OD	       REG(REG_TYPE_OD, 0)
-#define REG_U(x)       REG(REG_TYPE_U, x)
-
-#define REG_T_DIFFUSE  REG(REG_TYPE_T, T_DIFFUSE)
-#define REG_T_SPECULAR REG(REG_TYPE_T, T_SPECULAR)
-#define REG_T_FOG_W    REG(REG_TYPE_T, T_FOG_W)
-#define REG_T_TEX(x)   REG(REG_TYPE_T, x)
-
-
-#define A0_DEST_REG( reg ) ( (reg) << A0_DEST_NR_SHIFT )
-#define A0_SRC0_REG( reg ) ( (reg) << A0_SRC0_NR_SHIFT )
-#define A1_SRC1_REG( reg ) ( (reg) << A1_SRC1_NR_SHIFT )
-#define A1_SRC2_REG( reg ) ( (reg) << A1_SRC2_NR_SHIFT )
-#define A2_SRC2_REG( reg ) ( (reg) << A2_SRC2_NR_SHIFT )
-#define D0_DECL_REG( reg ) ( (reg) << D0_NR_SHIFT )
-#define T0_DEST_REG( reg ) ( (reg) << T0_DEST_NR_SHIFT )
-
-#define T0_SAMPLER( unit )     ((unit)<<T0_SAMPLER_NR_SHIFT)
-
-#define T1_ADDRESS_REG( type, nr ) (((type)<<T1_ADDRESS_REG_TYPE_SHIFT)| \
-				    ((nr)<<T1_ADDRESS_REG_NR_SHIFT))
-
-
-#define A1_SRC0_XYZW ((SRC_X << A1_SRC0_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A1_SRC0_CHANNEL_Y_SHIFT) |	\
-		      (SRC_Z << A1_SRC0_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A1_SRC0_CHANNEL_W_SHIFT))
-
-#define A1_SRC1_XY   ((SRC_X << A1_SRC1_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A1_SRC1_CHANNEL_Y_SHIFT))
-
-#define A2_SRC1_ZW   ((SRC_Z << A2_SRC1_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A2_SRC1_CHANNEL_W_SHIFT))
-
-#define A2_SRC2_XYZW ((SRC_X << A2_SRC2_CHANNEL_X_SHIFT) |	\
-		      (SRC_Y << A2_SRC2_CHANNEL_Y_SHIFT) |	\
-		      (SRC_Z << A2_SRC2_CHANNEL_Z_SHIFT) |	\
-		      (SRC_W << A2_SRC2_CHANNEL_W_SHIFT))
-
-
-
-
-
-static void
-meta_no_texture(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   static const GLuint prog[] = {
-      _3DSTATE_PIXEL_SHADER_PROGRAM,
-
-      /* Declare incoming diffuse color:
-       */
-      (D0_DCL | D0_DECL_REG(REG_T_DIFFUSE) | D0_CHANNEL_ALL),
-      D1_MBZ,
-      D2_MBZ,
-
-      /* output-color = mov(t_diffuse)
-       */
-      (A0_MOV |
-       A0_DEST_REG(REG_OC) |
-       A0_DEST_CHANNEL_ALL | A0_SRC0_REG(REG_T_DIFFUSE)),
-      (A1_SRC0_XYZW),
-      0,
-   };
-
-
-   memcpy(i915->meta.Program, prog, sizeof(prog));
-   i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
-   i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
-   i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
-}
-
-static void
-meta_texture_blend_replace(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   static const GLuint prog[] = {
-      _3DSTATE_PIXEL_SHADER_PROGRAM,
-
-      /* Declare the sampler:
-       */
-      (D0_DCL | D0_DECL_REG(REG_S(0)) | D0_SAMPLE_TYPE_2D | D0_CHANNEL_NONE),
-      D1_MBZ,
-      D2_MBZ,
-
-      /* Declare the interpolated texture coordinate:
-       */
-      (D0_DCL | D0_DECL_REG(REG_T_TEX(0)) | D0_CHANNEL_ALL),
-      D1_MBZ,
-      D2_MBZ,
-
-      /* output-color = texld(sample0, texcoord0) 
-       */
-      (T0_TEXLD | T0_DEST_REG(REG_OC) | T0_SAMPLER(0)),
-      T1_ADDRESS_REG(REG_TYPE_T, 0),
-      T2_MBZ
-   };
-
-   memcpy(i915->meta.Program, prog, sizeof(prog));
-   i915->meta.ProgramSize = sizeof(prog) / sizeof(*prog);
-   i915->meta.Program[0] |= i915->meta.ProgramSize - 2;
-   i915->meta.emitted &= ~I915_UPLOAD_PROGRAM;
-}
-
-
-
-
-
-/* Set up an arbitary piece of memory as a rectangular texture
- * (including the front or back buffer).
- */
-static GLboolean
-meta_tex_rect_source(struct intel_context *intel,
-                     struct _DriBufferObject *buffer,
-                     GLuint offset,
-                     GLuint pitch, GLuint height, GLenum format, GLenum type)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   GLuint unit = 0;
-   GLint numLevels = 1;
-   GLuint *state = i915->meta.Tex[0];
-   GLuint textureFormat;
-   GLuint cpp;
-
-   /* A full implementation of this would do the upload through
-    * glTexImage2d, and get all the conversion operations at that
-    * point.  We are restricted, but still at least have access to the
-    * fragment program swizzle.
-    */
-   switch (format) {
-   case GL_BGRA:
-      switch (type) {
-      case GL_UNSIGNED_INT_8_8_8_8_REV:
-      case GL_UNSIGNED_BYTE:
-         textureFormat = (MAPSURF_32BIT | MT_32BIT_ARGB8888);
-         cpp = 4;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_RGBA:
-      switch (type) {
-      case GL_UNSIGNED_INT_8_8_8_8_REV:
-      case GL_UNSIGNED_BYTE:
-         textureFormat = (MAPSURF_32BIT | MT_32BIT_ABGR8888);
-         cpp = 4;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_BGR:
-      switch (type) {
-      case GL_UNSIGNED_SHORT_5_6_5_REV:
-         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
-         cpp = 2;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-   case GL_RGB:
-      switch (type) {
-      case GL_UNSIGNED_SHORT_5_6_5:
-         textureFormat = (MAPSURF_16BIT | MT_16BIT_RGB565);
-         cpp = 2;
-         break;
-      default:
-         return GL_FALSE;
-      }
-      break;
-
-   default:
-      return GL_FALSE;
-   }
-
-
-   if ((pitch * cpp) & 3) {
-      _mesa_printf("%s: texture is not dword pitch\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-/*    intel_region_release(&i915->meta.tex_region[0]); */
-/*    intel_region_reference(&i915->meta.tex_region[0], region); */
-   i915->meta.tex_buffer[0] = buffer;
-   i915->meta.tex_offset[0] = offset;
-
-   state[I915_TEXREG_MS3] = (((height - 1) << MS3_HEIGHT_SHIFT) |
-                             ((pitch - 1) << MS3_WIDTH_SHIFT) |
-                             textureFormat | MS3_USE_FENCE_REGS);
-
-   state[I915_TEXREG_MS4] = (((((pitch * cpp) / 4) - 1) << MS4_PITCH_SHIFT) |
-                             MS4_CUBE_FACE_ENA_MASK |
-                             ((((numLevels - 1) * 4)) << MS4_MAX_LOD_SHIFT));
-
-   state[I915_TEXREG_SS2] = ((FILTER_NEAREST << SS2_MIN_FILTER_SHIFT) |
-                             (MIPFILTER_NONE << SS2_MIP_FILTER_SHIFT) |
-                             (FILTER_NEAREST << SS2_MAG_FILTER_SHIFT));
-
-   state[I915_TEXREG_SS3] = ((TEXCOORDMODE_WRAP << SS3_TCX_ADDR_MODE_SHIFT) |
-                             (TEXCOORDMODE_WRAP << SS3_TCY_ADDR_MODE_SHIFT) |
-                             (TEXCOORDMODE_WRAP << SS3_TCZ_ADDR_MODE_SHIFT) |
-                             (unit << SS3_TEXTUREMAP_INDEX_SHIFT));
-
-   state[I915_TEXREG_SS4] = 0;
-
-   i915->meta.emitted &= ~I915_UPLOAD_TEX(0);
-   return GL_TRUE;
-}
-
-
-/**
- * Set the color and depth drawing region for meta ops.
- */
-static void
-meta_draw_region(struct intel_context *intel,
-                 struct intel_region *color_region,
-                 struct intel_region *depth_region)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   i915_state_draw_region(intel, &i915->meta, color_region, depth_region);
-}
-
-
-static void
-set_vertex_format(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   i915->meta.Ctx[I915_CTXREG_LIS2] =
-      (S2_TEXCOORD_FMT(0, TEXCOORDFMT_2D) |
-       S2_TEXCOORD_FMT(1, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(2, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(3, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(4, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(5, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(6, TEXCOORDFMT_NOT_PRESENT) |
-       S2_TEXCOORD_FMT(7, TEXCOORDFMT_NOT_PRESENT));
-
-   i915->meta.Ctx[I915_CTXREG_LIS4] &= ~S4_VFMT_MASK;
-
-   i915->meta.Ctx[I915_CTXREG_LIS4] |= (S4_VFMT_COLOR | S4_VFMT_XYZ);
-
-   i915->meta.emitted &= ~I915_UPLOAD_CTX;
-}
-
-
-
-/* Operations where the 3D engine is decoupled temporarily from the
- * current GL state and used for other purposes than simply rendering
- * incoming triangles.
- */
-static void
-install_meta_state(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   memcpy(&i915->meta, &i915->initial, sizeof(i915->meta));
-   i915->meta.active = ACTIVE;
-   i915->meta.emitted = 0;
-
-   SET_STATE(i915, meta);
-   set_vertex_format(intel);
-   meta_no_texture(intel);
-}
-
-static void
-leave_meta_state(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   intel_region_release(&i915->meta.draw_region);
-   intel_region_release(&i915->meta.depth_region);
-/*    intel_region_release(&i915->meta.tex_region[0]); */
-   SET_STATE(i915, state);
-}
-
-
-
-void
-i915InitMetaFuncs(struct i915_context *i915)
-{
-   i915->intel.vtbl.install_meta_state = install_meta_state;
-   i915->intel.vtbl.leave_meta_state = leave_meta_state;
-   i915->intel.vtbl.meta_no_depth_write = meta_no_depth_write;
-   i915->intel.vtbl.meta_no_stencil_write = meta_no_stencil_write;
-   i915->intel.vtbl.meta_stencil_replace = meta_stencil_replace;
-   i915->intel.vtbl.meta_depth_replace = meta_depth_replace;
-   i915->intel.vtbl.meta_color_mask = meta_color_mask;
-   i915->intel.vtbl.meta_no_texture = meta_no_texture;
-   i915->intel.vtbl.meta_texture_blend_replace = meta_texture_blend_replace;
-   i915->intel.vtbl.meta_tex_rect_source = meta_tex_rect_source;
-   i915->intel.vtbl.meta_draw_region = meta_draw_region;
-   i915->intel.vtbl.meta_import_pixel_state = meta_import_pixel_state;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_program.c b/src/mesa/drivers/dri/i915tex/i915_program.c
deleted file mode 100644
index ed45e4f8069..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_program.c
+++ /dev/null
@@ -1,515 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include <strings.h>
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "tnl/t_context.h"
-#include "intel_batchbuffer.h"
-
-#include "i915_reg.h"
-#include "i915_context.h"
-#include "i915_program.h"
-
-
-#define A0_DEST( reg ) (((reg)&UREG_TYPE_NR_MASK)>>UREG_A0_DEST_SHIFT_LEFT)
-#define D0_DEST( reg ) (((reg)&UREG_TYPE_NR_MASK)>>UREG_A0_DEST_SHIFT_LEFT)
-#define T0_DEST( reg ) (((reg)&UREG_TYPE_NR_MASK)>>UREG_A0_DEST_SHIFT_LEFT)
-#define A0_SRC0( reg ) (((reg)&UREG_MASK)>>UREG_A0_SRC0_SHIFT_LEFT)
-#define A1_SRC0( reg ) (((reg)&UREG_MASK)<<UREG_A1_SRC0_SHIFT_RIGHT)
-#define A1_SRC1( reg ) (((reg)&UREG_MASK)>>UREG_A1_SRC1_SHIFT_LEFT)
-#define A2_SRC1( reg ) (((reg)&UREG_MASK)<<UREG_A2_SRC1_SHIFT_RIGHT)
-#define A2_SRC2( reg ) (((reg)&UREG_MASK)>>UREG_A2_SRC2_SHIFT_LEFT)
-
-/* These are special, and don't have swizzle/negate bits.
- */
-#define T0_SAMPLER( reg )     (GET_UREG_NR(reg)<<T0_SAMPLER_NR_SHIFT)
-#define T1_ADDRESS_REG( reg ) ((GET_UREG_NR(reg)<<T1_ADDRESS_REG_NR_SHIFT) | \
-			       (GET_UREG_TYPE(reg)<<T1_ADDRESS_REG_TYPE_SHIFT))
-
-
-/* Macros for translating UREG's into the various register fields used
- * by the I915 programmable unit.
- */
-#define UREG_A0_DEST_SHIFT_LEFT  (UREG_TYPE_SHIFT - A0_DEST_TYPE_SHIFT)
-#define UREG_A0_SRC0_SHIFT_LEFT  (UREG_TYPE_SHIFT - A0_SRC0_TYPE_SHIFT)
-#define UREG_A1_SRC0_SHIFT_RIGHT (A1_SRC0_CHANNEL_W_SHIFT - UREG_CHANNEL_W_SHIFT)
-#define UREG_A1_SRC1_SHIFT_LEFT  (UREG_TYPE_SHIFT - A1_SRC1_TYPE_SHIFT)
-#define UREG_A2_SRC1_SHIFT_RIGHT (A2_SRC1_CHANNEL_W_SHIFT - UREG_CHANNEL_W_SHIFT)
-#define UREG_A2_SRC2_SHIFT_LEFT  (UREG_TYPE_SHIFT - A2_SRC2_TYPE_SHIFT)
-
-#define UREG_MASK         0xffffff00
-#define UREG_TYPE_NR_MASK ((REG_TYPE_MASK << UREG_TYPE_SHIFT) | \
-  			   (REG_NR_MASK << UREG_NR_SHIFT))
-
-
-#define I915_CONSTFLAG_PARAM 0x1f
-
-GLuint
-i915_get_temp(struct i915_fragment_program *p)
-{
-   int bit = ffs(~p->temp_flag);
-   if (!bit) {
-      fprintf(stderr, "%s: out of temporaries\n", __FILE__);
-      exit(1);
-   }
-
-   p->temp_flag |= 1 << (bit - 1);
-   return UREG(REG_TYPE_R, (bit - 1));
-}
-
-
-GLuint
-i915_get_utemp(struct i915_fragment_program * p)
-{
-   int bit = ffs(~p->utemp_flag);
-   if (!bit) {
-      fprintf(stderr, "%s: out of temporaries\n", __FILE__);
-      exit(1);
-   }
-
-   p->utemp_flag |= 1 << (bit - 1);
-   return UREG(REG_TYPE_U, (bit - 1));
-}
-
-void
-i915_release_utemps(struct i915_fragment_program *p)
-{
-   p->utemp_flag = ~0x7;
-}
-
-
-GLuint
-i915_emit_decl(struct i915_fragment_program *p,
-               GLuint type, GLuint nr, GLuint d0_flags)
-{
-   GLuint reg = UREG(type, nr);
-
-   if (type == REG_TYPE_T) {
-      if (p->decl_t & (1 << nr))
-         return reg;
-
-      p->decl_t |= (1 << nr);
-   }
-   else if (type == REG_TYPE_S) {
-      if (p->decl_s & (1 << nr))
-         return reg;
-
-      p->decl_s |= (1 << nr);
-   }
-   else
-      return reg;
-
-   *(p->decl++) = (D0_DCL | D0_DEST(reg) | d0_flags);
-   *(p->decl++) = D1_MBZ;
-   *(p->decl++) = D2_MBZ;
-
-   p->nr_decl_insn++;
-   return reg;
-}
-
-GLuint
-i915_emit_arith(struct i915_fragment_program * p,
-                GLuint op,
-                GLuint dest,
-                GLuint mask,
-                GLuint saturate, GLuint src0, GLuint src1, GLuint src2)
-{
-   GLuint c[3];
-   GLuint nr_const = 0;
-
-   assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
-   dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest));
-   assert(dest);
-
-   if (GET_UREG_TYPE(src0) == REG_TYPE_CONST)
-      c[nr_const++] = 0;
-   if (GET_UREG_TYPE(src1) == REG_TYPE_CONST)
-      c[nr_const++] = 1;
-   if (GET_UREG_TYPE(src2) == REG_TYPE_CONST)
-      c[nr_const++] = 2;
-
-   /* Recursively call this function to MOV additional const values
-    * into temporary registers.  Use utemp registers for this -
-    * currently shouldn't be possible to run out, but keep an eye on
-    * this.
-    */
-   if (nr_const > 1) {
-      GLuint s[3], first, i, old_utemp_flag;
-
-      s[0] = src0;
-      s[1] = src1;
-      s[2] = src2;
-      old_utemp_flag = p->utemp_flag;
-
-      first = GET_UREG_NR(s[c[0]]);
-      for (i = 1; i < nr_const; i++) {
-         if (GET_UREG_NR(s[c[i]]) != first) {
-            GLuint tmp = i915_get_utemp(p);
-
-            i915_emit_arith(p, A0_MOV, tmp, A0_DEST_CHANNEL_ALL, 0,
-                            s[c[i]], 0, 0);
-            s[c[i]] = tmp;
-         }
-      }
-
-      src0 = s[0];
-      src1 = s[1];
-      src2 = s[2];
-      p->utemp_flag = old_utemp_flag;   /* restore */
-   }
-
-   *(p->csr++) = (op | A0_DEST(dest) | mask | saturate | A0_SRC0(src0));
-   *(p->csr++) = (A1_SRC0(src0) | A1_SRC1(src1));
-   *(p->csr++) = (A2_SRC1(src1) | A2_SRC2(src2));
-
-   p->nr_alu_insn++;
-   return dest;
-}
-
-GLuint i915_emit_texld( struct i915_fragment_program *p,
-			GLuint dest,
-			GLuint destmask,
-			GLuint sampler,
-			GLuint coord,
-			GLuint op )
-{
-   if (coord != UREG(GET_UREG_TYPE(coord), GET_UREG_NR(coord))) {
-      /* No real way to work around this in the general case - need to
-       * allocate and declare a new temporary register (a utemp won't
-       * do).  Will fallback for now.
-       */
-      i915_program_error(p, "Can't (yet) swizzle TEX arguments");
-      return 0;
-   }
-
-   /* Don't worry about saturate as we only support  
-    */
-   if (destmask != A0_DEST_CHANNEL_ALL) {
-      GLuint tmp = i915_get_utemp(p);
-      i915_emit_texld( p, tmp, A0_DEST_CHANNEL_ALL, sampler, coord, op );
-      i915_emit_arith( p, A0_MOV, dest, destmask, 0, tmp, 0, 0 );
-      return dest;
-   }
-   else {
-      assert(GET_UREG_TYPE(dest) != REG_TYPE_CONST);
-      assert(dest = UREG(GET_UREG_TYPE(dest), GET_UREG_NR(dest)));
-
-      if (GET_UREG_TYPE(coord) != REG_TYPE_T) {
-	 p->nr_tex_indirect++;
-      }
-
-      *(p->csr++) = (op | 
-		     T0_DEST( dest ) |
-		     T0_SAMPLER( sampler ));
-
-      *(p->csr++) = T1_ADDRESS_REG( coord );
-      *(p->csr++) = T2_MBZ;
-
-      p->nr_tex_insn++;
-      return dest;
-   }
-}
-
-
-GLuint
-i915_emit_const1f(struct i915_fragment_program * p, GLfloat c0)
-{
-   GLint reg, idx;
-
-   if (c0 == 0.0)
-      return swizzle(UREG(REG_TYPE_R, 0), ZERO, ZERO, ZERO, ZERO);
-   if (c0 == 1.0)
-      return swizzle(UREG(REG_TYPE_R, 0), ONE, ONE, ONE, ONE);
-
-   for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
-      if (p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-         continue;
-      for (idx = 0; idx < 4; idx++) {
-         if (!(p->constant_flags[reg] & (1 << idx)) ||
-             p->constant[reg][idx] == c0) {
-            p->constant[reg][idx] = c0;
-            p->constant_flags[reg] |= 1 << idx;
-            if (reg + 1 > p->nr_constants)
-               p->nr_constants = reg + 1;
-            return swizzle(UREG(REG_TYPE_CONST, reg), idx, ZERO, ZERO, ONE);
-         }
-      }
-   }
-
-   fprintf(stderr, "%s: out of constants\n", __FUNCTION__);
-   p->error = 1;
-   return 0;
-}
-
-GLuint
-i915_emit_const2f(struct i915_fragment_program * p, GLfloat c0, GLfloat c1)
-{
-   GLint reg, idx;
-
-   if (c0 == 0.0)
-      return swizzle(i915_emit_const1f(p, c1), ZERO, X, Z, W);
-   if (c0 == 1.0)
-      return swizzle(i915_emit_const1f(p, c1), ONE, X, Z, W);
-
-   if (c1 == 0.0)
-      return swizzle(i915_emit_const1f(p, c0), X, ZERO, Z, W);
-   if (c1 == 1.0)
-      return swizzle(i915_emit_const1f(p, c0), X, ONE, Z, W);
-
-   for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
-      if (p->constant_flags[reg] == 0xf ||
-          p->constant_flags[reg] == I915_CONSTFLAG_PARAM)
-         continue;
-      for (idx = 0; idx < 3; idx++) {
-         if (!(p->constant_flags[reg] & (3 << idx))) {
-            p->constant[reg][idx] = c0;
-            p->constant[reg][idx + 1] = c1;
-            p->constant_flags[reg] |= 3 << idx;
-            if (reg + 1 > p->nr_constants)
-               p->nr_constants = reg + 1;
-            return swizzle(UREG(REG_TYPE_CONST, reg), idx, idx + 1, ZERO,
-                           ONE);
-         }
-      }
-   }
-
-   fprintf(stderr, "%s: out of constants\n", __FUNCTION__);
-   p->error = 1;
-   return 0;
-}
-
-
-
-GLuint
-i915_emit_const4f(struct i915_fragment_program * p,
-                  GLfloat c0, GLfloat c1, GLfloat c2, GLfloat c3)
-{
-   GLint reg;
-
-   for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
-      if (p->constant_flags[reg] == 0xf &&
-          p->constant[reg][0] == c0 &&
-          p->constant[reg][1] == c1 &&
-          p->constant[reg][2] == c2 && p->constant[reg][3] == c3) {
-         return UREG(REG_TYPE_CONST, reg);
-      }
-      else if (p->constant_flags[reg] == 0) {
-         p->constant[reg][0] = c0;
-         p->constant[reg][1] = c1;
-         p->constant[reg][2] = c2;
-         p->constant[reg][3] = c3;
-         p->constant_flags[reg] = 0xf;
-         if (reg + 1 > p->nr_constants)
-            p->nr_constants = reg + 1;
-         return UREG(REG_TYPE_CONST, reg);
-      }
-   }
-
-   fprintf(stderr, "%s: out of constants\n", __FUNCTION__);
-   p->error = 1;
-   return 0;
-}
-
-
-GLuint
-i915_emit_const4fv(struct i915_fragment_program * p, const GLfloat * c)
-{
-   return i915_emit_const4f(p, c[0], c[1], c[2], c[3]);
-}
-
-
-GLuint
-i915_emit_param4fv(struct i915_fragment_program * p, const GLfloat * values)
-{
-   GLint reg, i;
-
-   for (i = 0; i < p->nr_params; i++) {
-      if (p->param[i].values == values)
-         return UREG(REG_TYPE_CONST, p->param[i].reg);
-   }
-
-
-   for (reg = 0; reg < I915_MAX_CONSTANT; reg++) {
-      if (p->constant_flags[reg] == 0) {
-         p->constant_flags[reg] = I915_CONSTFLAG_PARAM;
-         i = p->nr_params++;
-
-         p->param[i].values = values;
-         p->param[i].reg = reg;
-         p->params_uptodate = 0;
-
-         if (reg + 1 > p->nr_constants)
-            p->nr_constants = reg + 1;
-         return UREG(REG_TYPE_CONST, reg);
-      }
-   }
-
-   fprintf(stderr, "%s: out of constants\n", __FUNCTION__);
-   p->error = 1;
-   return 0;
-}
-
-
-
-void
-i915_program_error(struct i915_fragment_program *p, const char *msg)
-{
-   _mesa_problem(NULL, "i915_program_error: %s", msg);
-   p->error = 1;
-}
-
-
-void
-i915_init_program(struct i915_context *i915, struct i915_fragment_program *p)
-{
-   GLcontext *ctx = &i915->intel.ctx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-   p->translated = 0;
-   p->params_uptodate = 0;
-   p->on_hardware = 0;
-   p->error = 0;
-
-   p->nr_tex_indirect = 1;      /* correct? */
-   p->nr_tex_insn = 0;
-   p->nr_alu_insn = 0;
-   p->nr_decl_insn = 0;
-
-   p->ctx = ctx;
-   memset(p->constant_flags, 0, sizeof(p->constant_flags));
-
-   p->nr_constants = 0;
-   p->csr = p->program;
-   p->decl = p->declarations;
-   p->decl_s = 0;
-   p->decl_t = 0;
-   p->temp_flag = 0xffff000;
-   p->utemp_flag = ~0x7;
-   p->wpos_tex = -1;
-   p->depth_written = 0;
-   p->nr_params = 0;
-
-   p->src_texture = UREG_BAD;
-   p->src_previous = UREG(REG_TYPE_T, T_DIFFUSE);
-   p->last_tex_stage = 0;
-   p->VB = &tnl->vb;
-
-   *(p->decl++) = _3DSTATE_PIXEL_SHADER_PROGRAM;
-}
-
-
-void
-i915_fini_program(struct i915_fragment_program *p)
-{
-   GLuint program_size = p->csr - p->program;
-   GLuint decl_size = p->decl - p->declarations;
-
-   if (p->nr_tex_indirect > I915_MAX_TEX_INDIRECT)
-      i915_program_error(p, "Exceeded max nr indirect texture lookups");
-
-   if (p->nr_tex_insn > I915_MAX_TEX_INSN)
-      i915_program_error(p, "Exceeded max TEX instructions");
-
-   if (p->nr_alu_insn > I915_MAX_ALU_INSN)
-      i915_program_error(p, "Exceeded max ALU instructions");
-
-   if (p->nr_decl_insn > I915_MAX_DECL_INSN)
-      i915_program_error(p, "Exceeded max DECL instructions");
-
-   if (p->error) {
-      p->FragProg.Base.NumNativeInstructions = 0;
-      p->FragProg.Base.NumNativeAluInstructions = 0;
-      p->FragProg.Base.NumNativeTexInstructions = 0;
-      p->FragProg.Base.NumNativeTexIndirections = 0;
-   }
-   else {
-      p->FragProg.Base.NumNativeInstructions = (p->nr_alu_insn +
-                                                p->nr_tex_insn +
-                                                p->nr_decl_insn);
-      p->FragProg.Base.NumNativeAluInstructions = p->nr_alu_insn;
-      p->FragProg.Base.NumNativeTexInstructions = p->nr_tex_insn;
-      p->FragProg.Base.NumNativeTexIndirections = p->nr_tex_indirect;
-   }
-
-   p->declarations[0] |= program_size + decl_size - 2;
-}
-
-void
-i915_upload_program(struct i915_context *i915,
-                    struct i915_fragment_program *p)
-{
-   GLuint program_size = p->csr - p->program;
-   GLuint decl_size = p->decl - p->declarations;
-
-   FALLBACK(&i915->intel, I915_FALLBACK_PROGRAM, p->error);
-
-   /* Could just go straight to the batchbuffer from here:
-    */
-   if (i915->state.ProgramSize != (program_size + decl_size) ||
-       memcmp(i915->state.Program + decl_size, p->program,
-              program_size * sizeof(int)) != 0) {
-      I915_STATECHANGE(i915, I915_UPLOAD_PROGRAM);
-      memcpy(i915->state.Program, p->declarations, decl_size * sizeof(int));
-      memcpy(i915->state.Program + decl_size, p->program,
-             program_size * sizeof(int));
-      i915->state.ProgramSize = decl_size + program_size;
-   }
-
-   /* Always seemed to get a failure if I used memcmp() to
-    * shortcircuit this state upload.  Needs further investigation?
-    */
-   if (p->nr_constants) {
-      GLuint nr = p->nr_constants;
-
-      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 1);
-      I915_STATECHANGE(i915, I915_UPLOAD_CONSTANTS);
-
-      i915->state.Constant[0] = _3DSTATE_PIXEL_SHADER_CONSTANTS | ((nr) * 4);
-      i915->state.Constant[1] = (1 << (nr - 1)) | ((1 << (nr - 1)) - 1);
-
-      memcpy(&i915->state.Constant[2], p->constant, 4 * sizeof(int) * (nr));
-      i915->state.ConstantSize = 2 + (nr) * 4;
-
-      if (0) {
-         GLuint i;
-         for (i = 0; i < nr; i++) {
-            fprintf(stderr, "const[%d]: %f %f %f %f\n", i,
-                    p->constant[i][0],
-                    p->constant[i][1], p->constant[i][2], p->constant[i][3]);
-         }
-      }
-   }
-   else {
-      I915_ACTIVESTATE(i915, I915_UPLOAD_CONSTANTS, 0);
-   }
-
-   p->on_hardware = 1;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_program.h b/src/mesa/drivers/dri/i915tex/i915_program.h
deleted file mode 100644
index 3c12b34f163..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_program.h
+++ /dev/null
@@ -1,160 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef I915_PROGRAM_H
-#define I915_PROGRAM_H
-
-#include "i915_context.h"
-#include "i915_reg.h"
-
-
-
-/* Having zero and one in here makes the definition of swizzle a lot
- * easier.
- */
-#define UREG_TYPE_SHIFT               29
-#define UREG_NR_SHIFT                 24
-#define UREG_CHANNEL_X_NEGATE_SHIFT   23
-#define UREG_CHANNEL_X_SHIFT          20
-#define UREG_CHANNEL_Y_NEGATE_SHIFT   19
-#define UREG_CHANNEL_Y_SHIFT          16
-#define UREG_CHANNEL_Z_NEGATE_SHIFT   15
-#define UREG_CHANNEL_Z_SHIFT          12
-#define UREG_CHANNEL_W_NEGATE_SHIFT   11
-#define UREG_CHANNEL_W_SHIFT          8
-#define UREG_CHANNEL_ZERO_NEGATE_MBZ  5
-#define UREG_CHANNEL_ZERO_SHIFT       4
-#define UREG_CHANNEL_ONE_NEGATE_MBZ   1
-#define UREG_CHANNEL_ONE_SHIFT        0
-
-#define UREG_BAD          0xffffffff    /* not a valid ureg */
-
-#define X    SRC_X
-#define Y    SRC_Y
-#define Z    SRC_Z
-#define W    SRC_W
-#define ZERO SRC_ZERO
-#define ONE  SRC_ONE
-
-/* Construct a ureg:
- */
-#define UREG( type, nr ) (((type)<< UREG_TYPE_SHIFT) |		\
-			  ((nr)  << UREG_NR_SHIFT) |		\
-			  (X     << UREG_CHANNEL_X_SHIFT) |	\
-			  (Y     << UREG_CHANNEL_Y_SHIFT) |	\
-			  (Z     << UREG_CHANNEL_Z_SHIFT) |	\
-			  (W     << UREG_CHANNEL_W_SHIFT) |	\
-			  (ZERO  << UREG_CHANNEL_ZERO_SHIFT) |	\
-			  (ONE   << UREG_CHANNEL_ONE_SHIFT))
-
-#define GET_CHANNEL_SRC( reg, channel ) ((reg<<(channel*4)) & (0xf<<20))
-#define CHANNEL_SRC( src, channel ) (src>>(channel*4))
-
-#define GET_UREG_TYPE(reg) (((reg)>>UREG_TYPE_SHIFT)&REG_TYPE_MASK)
-#define GET_UREG_NR(reg)   (((reg)>>UREG_NR_SHIFT)&REG_NR_MASK)
-
-
-
-#define UREG_XYZW_CHANNEL_MASK 0x00ffff00
-
-/* One neat thing about the UREG representation:  
- */
-static INLINE int
-swizzle(int reg, int x, int y, int z, int w)
-{
-   return ((reg & ~UREG_XYZW_CHANNEL_MASK) |
-           CHANNEL_SRC(GET_CHANNEL_SRC(reg, x), 0) |
-           CHANNEL_SRC(GET_CHANNEL_SRC(reg, y), 1) |
-           CHANNEL_SRC(GET_CHANNEL_SRC(reg, z), 2) |
-           CHANNEL_SRC(GET_CHANNEL_SRC(reg, w), 3));
-}
-
-/* Another neat thing about the UREG representation:  
- */
-static INLINE int
-negate(int reg, int x, int y, int z, int w)
-{
-   return reg ^ (((x & 1) << UREG_CHANNEL_X_NEGATE_SHIFT) |
-                 ((y & 1) << UREG_CHANNEL_Y_NEGATE_SHIFT) |
-                 ((z & 1) << UREG_CHANNEL_Z_NEGATE_SHIFT) |
-                 ((w & 1) << UREG_CHANNEL_W_NEGATE_SHIFT));
-}
-
-
-extern GLuint i915_get_temp(struct i915_fragment_program *p);
-extern GLuint i915_get_utemp(struct i915_fragment_program *p);
-extern void i915_release_utemps(struct i915_fragment_program *p);
-
-
-extern GLuint i915_emit_texld(struct i915_fragment_program *p,
-                              GLuint dest,
-                              GLuint destmask,
-                              GLuint sampler, GLuint coord, GLuint op);
-
-extern GLuint i915_emit_arith(struct i915_fragment_program *p,
-                              GLuint op,
-                              GLuint dest,
-                              GLuint mask,
-                              GLuint saturate,
-                              GLuint src0, GLuint src1, GLuint src2);
-
-extern GLuint i915_emit_decl(struct i915_fragment_program *p,
-                             GLuint type, GLuint nr, GLuint d0_flags);
-
-
-extern GLuint i915_emit_const1f(struct i915_fragment_program *p, GLfloat c0);
-
-extern GLuint i915_emit_const2f(struct i915_fragment_program *p,
-                                GLfloat c0, GLfloat c1);
-
-extern GLuint i915_emit_const4fv(struct i915_fragment_program *p,
-                                 const GLfloat * c);
-
-extern GLuint i915_emit_const4f(struct i915_fragment_program *p,
-                                GLfloat c0, GLfloat c1,
-                                GLfloat c2, GLfloat c3);
-
-
-extern GLuint i915_emit_param4fv(struct i915_fragment_program *p,
-                                 const GLfloat * values);
-
-extern void i915_program_error(struct i915_fragment_program *p,
-                               const char *msg);
-
-extern void i915_init_program(struct i915_context *i915,
-                              struct i915_fragment_program *p);
-
-extern void i915_upload_program(struct i915_context *i915,
-                                struct i915_fragment_program *p);
-
-extern void i915_fini_program(struct i915_fragment_program *p);
-
-
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/i915_reg.h b/src/mesa/drivers/dri/i915tex/i915_reg.h
deleted file mode 100644
index 34c68214056..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_reg.h
+++ /dev/null
@@ -1,841 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef _I915_REG_H_
-#define _I915_REG_H_
-
-
-#include "intel_reg.h"
-
-#define I915_SET_FIELD( var, mask, value ) (var &= ~(mask), var |= value)
-
-#define CMD_3D (0x3<<29)
-
-#define PRIM3D_INLINE		(CMD_3D | (0x1f<<24))
-#define PRIM3D_TRILIST		(0x0<<18)
-#define PRIM3D_TRISTRIP 	(0x1<<18)
-#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
-#define PRIM3D_TRIFAN		(0x3<<18)
-#define PRIM3D_POLY		(0x4<<18)
-#define PRIM3D_LINELIST 	(0x5<<18)
-#define PRIM3D_LINESTRIP	(0x6<<18)
-#define PRIM3D_RECTLIST 	(0x7<<18)
-#define PRIM3D_POINTLIST	(0x8<<18)
-#define PRIM3D_DIB		(0x9<<18)
-#define PRIM3D_CLEAR_RECT	(0xa<<18)
-#define PRIM3D_ZONE_INIT	(0xd<<18)
-#define PRIM3D_MASK		(0x1f<<18)
-
-/* p137 */
-#define _3DSTATE_AA_CMD			(CMD_3D | (0x06<<24))
-#define AA_LINE_ECAAR_WIDTH_ENABLE	(1<<16)
-#define AA_LINE_ECAAR_WIDTH_0_5 	0
-#define AA_LINE_ECAAR_WIDTH_1_0		(1<<14)
-#define AA_LINE_ECAAR_WIDTH_2_0 	(2<<14)
-#define AA_LINE_ECAAR_WIDTH_4_0 	(3<<14)
-#define AA_LINE_REGION_WIDTH_ENABLE	(1<<8)
-#define AA_LINE_REGION_WIDTH_0_5	0
-#define AA_LINE_REGION_WIDTH_1_0	(1<<6)
-#define AA_LINE_REGION_WIDTH_2_0	(2<<6)
-#define AA_LINE_REGION_WIDTH_4_0	(3<<6)
-
-/* 3DSTATE_BACKFACE_STENCIL_OPS, p138*/
-#define _3DSTATE_BACKFACE_STENCIL_OPS    (CMD_3D | (0x8<<24))
-#define BFO_ENABLE_STENCIL_REF          (1<<23)
-#define BFO_STENCIL_REF_SHIFT           15
-#define BFO_STENCIL_REF_MASK            (0xff<<15)
-#define BFO_ENABLE_STENCIL_FUNCS        (1<<14)
-#define BFO_STENCIL_TEST_SHIFT          11
-#define BFO_STENCIL_TEST_MASK           (0x7<<11)
-#define BFO_STENCIL_FAIL_SHIFT          8
-#define BFO_STENCIL_FAIL_MASK           (0x7<<8)
-#define BFO_STENCIL_PASS_Z_FAIL_SHIFT   5
-#define BFO_STENCIL_PASS_Z_FAIL_MASK    (0x7<<5)
-#define BFO_STENCIL_PASS_Z_PASS_SHIFT   2
-#define BFO_STENCIL_PASS_Z_PASS_MASK    (0x7<<2)
-#define BFO_ENABLE_STENCIL_TWO_SIDE     (1<<1)
-#define BFO_STENCIL_TWO_SIDE            (1<<0)
-
-
-/* 3DSTATE_BACKFACE_STENCIL_MASKS, p140 */
-#define _3DSTATE_BACKFACE_STENCIL_MASKS    (CMD_3D | (0x9<<24))
-#define BFM_ENABLE_STENCIL_TEST_MASK      (1<<17)
-#define BFM_ENABLE_STENCIL_WRITE_MASK     (1<<16)
-#define BFM_STENCIL_TEST_MASK_SHIFT       8
-#define BFM_STENCIL_TEST_MASK_MASK        (0xff<<8)
-#define BFM_STENCIL_WRITE_MASK_SHIFT      0
-#define BFM_STENCIL_WRITE_MASK_MASK       (0xff<<0)
-
-
-
-/* 3DSTATE_BIN_CONTROL p141 */
-
-/* p143 */
-#define _3DSTATE_BUF_INFO_CMD	(CMD_3D | (0x1d<<24) | (0x8e<<16) | 1)
-/* Dword 1 */
-#define BUF_3D_ID_COLOR_BACK	(0x3<<24)
-#define BUF_3D_ID_DEPTH 	(0x7<<24)
-#define BUF_3D_USE_FENCE	(1<<23)
-#define BUF_3D_TILED_SURFACE	(1<<22)
-#define BUF_3D_TILE_WALK_X	0
-#define BUF_3D_TILE_WALK_Y	(1<<21)
-#define BUF_3D_PITCH(x)         (((x)/4)<<2)
-/* Dword 2 */
-#define BUF_3D_ADDR(x)		((x) & ~0x3)
-
-
-/* 3DSTATE_CHROMA_KEY */
-
-/* 3DSTATE_CLEAR_PARAMETERS, p150 */
-
-/* 3DSTATE_CONSTANT_BLEND_COLOR, p153 */
-#define _3DSTATE_CONST_BLEND_COLOR_CMD	(CMD_3D | (0x1d<<24) | (0x88<<16))
-
-
-
-/* 3DSTATE_COORD_SET_BINDINGS, p154 */
-#define _3DSTATE_COORD_SET_BINDINGS      (CMD_3D | (0x16<<24))
-#define CSB_TCB(iunit, eunit)           ((eunit)<<(iunit*3))
-
-/* p156 */
-#define _3DSTATE_DFLT_DIFFUSE_CMD	(CMD_3D | (0x1d<<24) | (0x99<<16))
-
-/* p157 */
-#define _3DSTATE_DFLT_SPEC_CMD		(CMD_3D | (0x1d<<24) | (0x9a<<16))
-
-/* p158 */
-#define _3DSTATE_DFLT_Z_CMD		(CMD_3D | (0x1d<<24) | (0x98<<16))
-
-
-/* 3DSTATE_DEPTH_OFFSET_SCALE, p159 */
-#define _3DSTATE_DEPTH_OFFSET_SCALE       (CMD_3D | (0x1d<<24) | (0x97<<16))
-/* scale in dword 1 */
-
-
-/* 3DSTATE_DEPTH_SUBRECT_DISABLE, p160 */
-#define _3DSTATE_DEPTH_SUBRECT_DISABLE    (CMD_3D | (0x1c<<24) | (0x11<<19) | 0x2)
-
-/* p161 */
-#define _3DSTATE_DST_BUF_VARS_CMD	(CMD_3D | (0x1d<<24) | (0x85<<16))
-/* Dword 1 */
-#define TEX_DEFAULT_COLOR_OGL           (0<<30)
-#define TEX_DEFAULT_COLOR_D3D           (1<<30)
-#define ZR_EARLY_DEPTH                  (1<<29)
-#define LOD_PRECLAMP_OGL                (1<<28)
-#define LOD_PRECLAMP_D3D                (0<<28)
-#define DITHER_FULL_ALWAYS              (0<<26)
-#define DITHER_FULL_ON_FB_BLEND         (1<<26)
-#define DITHER_CLAMPED_ALWAYS           (2<<26)
-#define LINEAR_GAMMA_BLEND_32BPP        (1<<25)
-#define DEBUG_DISABLE_ENH_DITHER        (1<<24)
-#define DSTORG_HORT_BIAS(x)		((x)<<20)
-#define DSTORG_VERT_BIAS(x)		((x)<<16)
-#define COLOR_4_2_2_CHNL_WRT_ALL	0
-#define COLOR_4_2_2_CHNL_WRT_Y		(1<<12)
-#define COLOR_4_2_2_CHNL_WRT_CR		(2<<12)
-#define COLOR_4_2_2_CHNL_WRT_CB		(3<<12)
-#define COLOR_4_2_2_CHNL_WRT_CRCB	(4<<12)
-#define COLR_BUF_8BIT			0
-#define COLR_BUF_RGB555 		(1<<8)
-#define COLR_BUF_RGB565 		(2<<8)
-#define COLR_BUF_ARGB8888		(3<<8)
-#define DEPTH_FRMT_16_FIXED		0
-#define DEPTH_FRMT_16_FLOAT		(1<<2)
-#define DEPTH_FRMT_24_FIXED_8_OTHER	(2<<2)
-#define VERT_LINE_STRIDE_1		(1<<1)
-#define VERT_LINE_STRIDE_0		(0<<1)
-#define VERT_LINE_STRIDE_OFS_1		1
-#define VERT_LINE_STRIDE_OFS_0		0
-
-/* p166 */
-#define _3DSTATE_DRAW_RECT_CMD		(CMD_3D|(0x1d<<24)|(0x80<<16)|3)
-/* Dword 1 */
-#define DRAW_RECT_DIS_DEPTH_OFS 	(1<<30)
-#define DRAW_DITHER_OFS_X(x)		((x)<<26)
-#define DRAW_DITHER_OFS_Y(x)		((x)<<24)
-/* Dword 2 */
-#define DRAW_YMIN(x)			((x)<<16)
-#define DRAW_XMIN(x)			(x)
-/* Dword 3 */
-#define DRAW_YMAX(x)			((x)<<16)
-#define DRAW_XMAX(x)			(x)
-/* Dword 4 */
-#define DRAW_YORG(x)			((x)<<16)
-#define DRAW_XORG(x)			(x)
-
-
-/* 3DSTATE_FILTER_COEFFICIENTS_4X4, p170 */
-
-/* 3DSTATE_FILTER_COEFFICIENTS_6X5, p172 */
-
-
-/* _3DSTATE_FOG_COLOR, p173 */
-#define _3DSTATE_FOG_COLOR_CMD		(CMD_3D|(0x15<<24))
-#define FOG_COLOR_RED(x)		((x)<<16)
-#define FOG_COLOR_GREEN(x)		((x)<<8)
-#define FOG_COLOR_BLUE(x)		(x)
-
-/* _3DSTATE_FOG_MODE, p174 */
-#define _3DSTATE_FOG_MODE_CMD		(CMD_3D|(0x1d<<24)|(0x89<<16)|2)
-/* Dword 1 */
-#define FMC1_FOGFUNC_MODIFY_ENABLE	(1<<31)
-#define FMC1_FOGFUNC_VERTEX		(0<<28)
-#define FMC1_FOGFUNC_PIXEL_EXP		(1<<28)
-#define FMC1_FOGFUNC_PIXEL_EXP2		(2<<28)
-#define FMC1_FOGFUNC_PIXEL_LINEAR	(3<<28)
-#define FMC1_FOGFUNC_MASK		(3<<28)
-#define FMC1_FOGINDEX_MODIFY_ENABLE     (1<<27)
-#define FMC1_FOGINDEX_Z		        (0<<25)
-#define FMC1_FOGINDEX_W   		(1<<25)
-#define FMC1_C1_C2_MODIFY_ENABLE	(1<<24)
-#define FMC1_DENSITY_MODIFY_ENABLE	(1<<23)
-#define FMC1_C1_ONE      	        (1<<13)
-#define FMC1_C1_MASK		        (0xffff<<4)
-/* Dword 2 */
-#define FMC2_C2_ONE		        (1<<16)
-/* Dword 3 */
-#define FMC3_D_ONE      		(1<<16)
-
-
-
-/* _3DSTATE_INDEPENDENT_ALPHA_BLEND, p177 */
-#define _3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD	(CMD_3D|(0x0b<<24))
-#define IAB_MODIFY_ENABLE	        (1<<23)
-#define IAB_ENABLE       	        (1<<22)
-#define IAB_MODIFY_FUNC         	(1<<21)
-#define IAB_FUNC_SHIFT          	16
-#define IAB_MODIFY_SRC_FACTOR   	(1<<11)
-#define IAB_SRC_FACTOR_SHIFT		6
-#define IAB_SRC_FACTOR_MASK		(BLENDFACT_MASK<<6)
-#define IAB_MODIFY_DST_FACTOR	        (1<<5)
-#define IAB_DST_FACTOR_SHIFT		0
-#define IAB_DST_FACTOR_MASK		(BLENDFACT_MASK<<0)
-
-
-#define BLENDFUNC_ADD			0x0
-#define BLENDFUNC_SUBTRACT		0x1
-#define BLENDFUNC_REVERSE_SUBTRACT	0x2
-#define BLENDFUNC_MIN			0x3
-#define BLENDFUNC_MAX			0x4
-#define BLENDFUNC_MASK			0x7
-
-/* 3DSTATE_LOAD_INDIRECT, p180 */
-
-#define _3DSTATE_LOAD_INDIRECT	        (CMD_3D|(0x1d<<24)|(0x7<<16))
-#define LI0_STATE_STATIC_INDIRECT       (0x01<<8)
-#define LI0_STATE_DYNAMIC_INDIRECT      (0x02<<8)
-#define LI0_STATE_SAMPLER               (0x04<<8)
-#define LI0_STATE_MAP                   (0x08<<8)
-#define LI0_STATE_PROGRAM               (0x10<<8)
-#define LI0_STATE_CONSTANTS             (0x20<<8)
-
-#define SIS0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define SIS0_FORCE_LOAD                 (1<<1)
-#define SIS0_BUFFER_VALID               (1<<0)
-#define SIS1_BUFFER_LENGTH(x)           ((x)&0xff)
-
-#define DIS0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define DIS0_BUFFER_RESET               (1<<1)
-#define DIS0_BUFFER_VALID               (1<<0)
-
-#define SSB0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define SSB0_FORCE_LOAD                 (1<<1)
-#define SSB0_BUFFER_VALID               (1<<0)
-#define SSB1_BUFFER_LENGTH(x)           ((x)&0xff)
-
-#define MSB0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define MSB0_FORCE_LOAD                 (1<<1)
-#define MSB0_BUFFER_VALID               (1<<0)
-#define MSB1_BUFFER_LENGTH(x)           ((x)&0xff)
-
-#define PSP0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define PSP0_FORCE_LOAD                 (1<<1)
-#define PSP0_BUFFER_VALID               (1<<0)
-#define PSP1_BUFFER_LENGTH(x)           ((x)&0xff)
-
-#define PSC0_BUFFER_ADDRESS(x)          ((x)&~0x3)
-#define PSC0_FORCE_LOAD                 (1<<1)
-#define PSC0_BUFFER_VALID               (1<<0)
-#define PSC1_BUFFER_LENGTH(x)           ((x)&0xff)
-
-
-
-
-
-/* _3DSTATE_RASTERIZATION_RULES */
-#define _3DSTATE_RASTER_RULES_CMD	(CMD_3D|(0x07<<24))
-#define ENABLE_POINT_RASTER_RULE	(1<<15)
-#define OGL_POINT_RASTER_RULE		(1<<13)
-#define ENABLE_TEXKILL_3D_4D            (1<<10)
-#define TEXKILL_3D                      (0<<9)
-#define TEXKILL_4D                      (1<<9)
-#define ENABLE_LINE_STRIP_PROVOKE_VRTX	(1<<8)
-#define ENABLE_TRI_FAN_PROVOKE_VRTX	(1<<5)
-#define LINE_STRIP_PROVOKE_VRTX(x)	((x)<<6)
-#define TRI_FAN_PROVOKE_VRTX(x) 	((x)<<3)
-
-/* _3DSTATE_SCISSOR_ENABLE, p256 */
-#define _3DSTATE_SCISSOR_ENABLE_CMD	(CMD_3D|(0x1c<<24)|(0x10<<19))
-#define ENABLE_SCISSOR_RECT		((1<<1) | 1)
-#define DISABLE_SCISSOR_RECT		(1<<1)
-
-/* _3DSTATE_SCISSOR_RECTANGLE_0, p257 */
-#define _3DSTATE_SCISSOR_RECT_0_CMD	(CMD_3D|(0x1d<<24)|(0x81<<16)|1)
-/* Dword 1 */
-#define SCISSOR_RECT_0_YMIN(x)		((x)<<16)
-#define SCISSOR_RECT_0_XMIN(x)		(x)
-/* Dword 2 */
-#define SCISSOR_RECT_0_YMAX(x)		((x)<<16)
-#define SCISSOR_RECT_0_XMAX(x)		(x)
-
-/* p189 */
-#define _3DSTATE_LOAD_STATE_IMMEDIATE_1   ((0x3<<29)|(0x1d<<24)|(0x04<<16))
-#define I1_LOAD_S(n)                      (1<<(4+n))
-
-#define S0_VB_OFFSET_MASK              0xffffffc
-#define S0_AUTO_CACHE_INV_DISABLE      (1<<0)
-
-#define S1_VERTEX_WIDTH_SHIFT          24
-#define S1_VERTEX_WIDTH_MASK           (0x3f<<24)
-#define S1_VERTEX_PITCH_SHIFT          16
-#define S1_VERTEX_PITCH_MASK           (0x3f<<16)
-
-#define TEXCOORDFMT_2D                 0x0
-#define TEXCOORDFMT_3D                 0x1
-#define TEXCOORDFMT_4D                 0x2
-#define TEXCOORDFMT_1D                 0x3
-#define TEXCOORDFMT_2D_16              0x4
-#define TEXCOORDFMT_4D_16              0x5
-#define TEXCOORDFMT_NOT_PRESENT        0xf
-#define S2_TEXCOORD_FMT0_MASK            0xf
-#define S2_TEXCOORD_FMT1_SHIFT           4
-#define S2_TEXCOORD_FMT(unit, type)    ((type)<<(unit*4))
-#define S2_TEXCOORD_NONE               (~0)
-
-/* S3 not interesting */
-
-#define S4_POINT_WIDTH_SHIFT           23
-#define S4_POINT_WIDTH_MASK            (0x1ff<<23)
-#define S4_LINE_WIDTH_SHIFT            19
-#define S4_LINE_WIDTH_ONE              (0x2<<19)
-#define S4_LINE_WIDTH_MASK             (0xf<<19)
-#define S4_FLATSHADE_ALPHA             (1<<18)
-#define S4_FLATSHADE_FOG               (1<<17)
-#define S4_FLATSHADE_SPECULAR          (1<<16)
-#define S4_FLATSHADE_COLOR             (1<<15)
-#define S4_CULLMODE_BOTH	       (0<<13)
-#define S4_CULLMODE_NONE	       (1<<13)
-#define S4_CULLMODE_CW		       (2<<13)
-#define S4_CULLMODE_CCW		       (3<<13)
-#define S4_CULLMODE_MASK	       (3<<13)
-#define S4_VFMT_POINT_WIDTH            (1<<12)
-#define S4_VFMT_SPEC_FOG               (1<<11)
-#define S4_VFMT_COLOR                  (1<<10)
-#define S4_VFMT_DEPTH_OFFSET           (1<<9)
-#define S4_VFMT_XYZ     	       (1<<6)
-#define S4_VFMT_XYZW     	       (2<<6)
-#define S4_VFMT_XY     		       (3<<6)
-#define S4_VFMT_XYW     	       (4<<6)
-#define S4_VFMT_XYZW_MASK              (7<<6)
-#define S4_FORCE_DEFAULT_DIFFUSE       (1<<5)
-#define S4_FORCE_DEFAULT_SPECULAR      (1<<4)
-#define S4_LOCAL_DEPTH_OFFSET_ENABLE   (1<<3)
-#define S4_VFMT_FOG_PARAM              (1<<2)
-#define S4_SPRITE_POINT_ENABLE         (1<<1)
-#define S4_LINE_ANTIALIAS_ENABLE       (1<<0)
-
-#define S4_VFMT_MASK (S4_VFMT_POINT_WIDTH   | 	\
-		      S4_VFMT_SPEC_FOG      |	\
-		      S4_VFMT_COLOR         |	\
-		      S4_VFMT_DEPTH_OFFSET  |	\
-		      S4_VFMT_XYZW_MASK     |	\
-		      S4_VFMT_FOG_PARAM)
-
-
-#define S5_WRITEDISABLE_ALPHA          (1<<31)
-#define S5_WRITEDISABLE_RED            (1<<30)
-#define S5_WRITEDISABLE_GREEN          (1<<29)
-#define S5_WRITEDISABLE_BLUE           (1<<28)
-#define S5_WRITEDISABLE_MASK           (0xf<<28)
-#define S5_FORCE_DEFAULT_POINT_SIZE    (1<<27)
-#define S5_LAST_PIXEL_ENABLE           (1<<26)
-#define S5_GLOBAL_DEPTH_OFFSET_ENABLE  (1<<25)
-#define S5_FOG_ENABLE                  (1<<24)
-#define S5_STENCIL_REF_SHIFT           16
-#define S5_STENCIL_REF_MASK            (0xff<<16)
-#define S5_STENCIL_TEST_FUNC_SHIFT     13
-#define S5_STENCIL_TEST_FUNC_MASK      (0x7<<13)
-#define S5_STENCIL_FAIL_SHIFT          10
-#define S5_STENCIL_FAIL_MASK           (0x7<<10)
-#define S5_STENCIL_PASS_Z_FAIL_SHIFT   7
-#define S5_STENCIL_PASS_Z_FAIL_MASK    (0x7<<7)
-#define S5_STENCIL_PASS_Z_PASS_SHIFT   4
-#define S5_STENCIL_PASS_Z_PASS_MASK    (0x7<<4)
-#define S5_STENCIL_WRITE_ENABLE        (1<<3)
-#define S5_STENCIL_TEST_ENABLE         (1<<2)
-#define S5_COLOR_DITHER_ENABLE         (1<<1)
-#define S5_LOGICOP_ENABLE              (1<<0)
-
-
-#define S6_ALPHA_TEST_ENABLE           (1<<31)
-#define S6_ALPHA_TEST_FUNC_SHIFT       28
-#define S6_ALPHA_TEST_FUNC_MASK        (0x7<<28)
-#define S6_ALPHA_REF_SHIFT             20
-#define S6_ALPHA_REF_MASK              (0xff<<20)
-#define S6_DEPTH_TEST_ENABLE           (1<<19)
-#define S6_DEPTH_TEST_FUNC_SHIFT       16
-#define S6_DEPTH_TEST_FUNC_MASK        (0x7<<16)
-#define S6_CBUF_BLEND_ENABLE           (1<<15)
-#define S6_CBUF_BLEND_FUNC_SHIFT       12
-#define S6_CBUF_BLEND_FUNC_MASK        (0x7<<12)
-#define S6_CBUF_SRC_BLEND_FACT_SHIFT   8
-#define S6_CBUF_SRC_BLEND_FACT_MASK    (0xf<<8)
-#define S6_CBUF_DST_BLEND_FACT_SHIFT   4
-#define S6_CBUF_DST_BLEND_FACT_MASK    (0xf<<4)
-#define S6_DEPTH_WRITE_ENABLE          (1<<3)
-#define S6_COLOR_WRITE_ENABLE          (1<<2)
-#define S6_TRISTRIP_PV_SHIFT           0
-#define S6_TRISTRIP_PV_MASK            (0x3<<0)
-
-#define S7_DEPTH_OFFSET_CONST_MASK     ~0
-
-/* 3DSTATE_MAP_DEINTERLACER_PARAMETERS */
-
-/* 3DSTATE_MAP_PALETTE_LOAD_32, p206 */
-#define _3DSTATE_MAP_PALETTE_LOAD_32    (CMD_3D|(0x1d<<24)|(0x8f<<16))
-/* subsequent dwords up to length (max 16) are ARGB8888 color values */
-
-/* _3DSTATE_MODES_4, p218 */
-#define _3DSTATE_MODES_4_CMD		(CMD_3D|(0x0d<<24))
-#define ENABLE_LOGIC_OP_FUNC		(1<<23)
-#define LOGIC_OP_FUNC(x)		((x)<<18)
-#define LOGICOP_MASK			(0xf<<18)
-#define MODE4_ENABLE_STENCIL_TEST_MASK	((1<<17)|(0xff00))
-#define ENABLE_STENCIL_TEST_MASK	(1<<17)
-#define STENCIL_TEST_MASK(x)		(((x)&0xff)<<8)
-#define MODE4_ENABLE_STENCIL_WRITE_MASK	((1<<16)|(0x00ff))
-#define ENABLE_STENCIL_WRITE_MASK	(1<<16)
-#define STENCIL_WRITE_MASK(x)		((x)&0xff)
-
-/* _3DSTATE_MODES_5, p220 */
-#define _3DSTATE_MODES_5_CMD		(CMD_3D|(0x0c<<24))
-#define PIPELINE_FLUSH_RENDER_CACHE	(1<<18)
-#define PIPELINE_FLUSH_TEXTURE_CACHE	(1<<16)
-
-
-/* p221 */
-#define _3DSTATE_PIXEL_SHADER_CONSTANTS  (CMD_3D|(0x1d<<24)|(0x6<<16))
-#define PS1_REG(n)                      (1<<(n))
-#define PS2_CONST_X(n)                  (n)
-#define PS3_CONST_Y(n)                  (n)
-#define PS4_CONST_Z(n)                  (n)
-#define PS5_CONST_W(n)                  (n)
-
-/* p222 */
-
-
-#define I915_MAX_TEX_INDIRECT 4
-#define I915_MAX_TEX_INSN     32
-#define I915_MAX_ALU_INSN     64
-#define I915_MAX_DECL_INSN    27
-#define I915_MAX_TEMPORARY    16
-
-
-/* Each instruction is 3 dwords long, though most don't require all
- * this space.  Maximum of 123 instructions.  Smaller maxes per insn
- * type.
- */
-#define _3DSTATE_PIXEL_SHADER_PROGRAM    (CMD_3D|(0x1d<<24)|(0x5<<16))
-
-#define REG_TYPE_R                 0    /* temporary regs, no need to
-                                         * dcl, must be written before
-                                         * read -- Preserved between
-                                         * phases. 
-                                         */
-#define REG_TYPE_T                 1    /* Interpolated values, must be
-                                         * dcl'ed before use.
-                                         *
-                                         * 0..7: texture coord,
-                                         * 8: diffuse spec,
-                                         * 9: specular color,
-                                         * 10: fog parameter in w.
-                                         */
-#define REG_TYPE_CONST             2    /* Restriction: only one const
-                                         * can be referenced per
-                                         * instruction, though it may be
-                                         * selected for multiple inputs.
-                                         * Constants not initialized
-                                         * default to zero.
-                                         */
-#define REG_TYPE_S                 3    /* sampler */
-#define REG_TYPE_OC                4    /* output color (rgba) */
-#define REG_TYPE_OD                5    /* output depth (w), xyz are
-                                         * temporaries.  If not written,
-                                         * interpolated depth is used?
-                                         */
-#define REG_TYPE_U                 6    /* unpreserved temporaries */
-#define REG_TYPE_MASK              0x7
-#define REG_NR_MASK                0xf
-
-
-/* REG_TYPE_T:
- */
-#define T_TEX0     0
-#define T_TEX1     1
-#define T_TEX2     2
-#define T_TEX3     3
-#define T_TEX4     4
-#define T_TEX5     5
-#define T_TEX6     6
-#define T_TEX7     7
-#define T_DIFFUSE  8
-#define T_SPECULAR 9
-#define T_FOG_W    10           /* interpolated fog is in W coord */
-
-/* Arithmetic instructions */
-
-/* .replicate_swizzle == selection and replication of a particular
- * scalar channel, ie., .xxxx, .yyyy, .zzzz or .wwww 
- */
-#define A0_NOP    (0x0<<24)     /* no operation */
-#define A0_ADD    (0x1<<24)     /* dst = src0 + src1 */
-#define A0_MOV    (0x2<<24)     /* dst = src0 */
-#define A0_MUL    (0x3<<24)     /* dst = src0 * src1 */
-#define A0_MAD    (0x4<<24)     /* dst = src0 * src1 + src2 */
-#define A0_DP2ADD (0x5<<24)     /* dst.xyzw = src0.xy dot src1.xy + src2.replicate_swizzle */
-#define A0_DP3    (0x6<<24)     /* dst.xyzw = src0.xyz dot src1.xyz */
-#define A0_DP4    (0x7<<24)     /* dst.xyzw = src0.xyzw dot src1.xyzw */
-#define A0_FRC    (0x8<<24)     /* dst = src0 - floor(src0) */
-#define A0_RCP    (0x9<<24)     /* dst.xyzw = 1/(src0.replicate_swizzle) */
-#define A0_RSQ    (0xa<<24)     /* dst.xyzw = 1/(sqrt(abs(src0.replicate_swizzle))) */
-#define A0_EXP    (0xb<<24)     /* dst.xyzw = exp2(src0.replicate_swizzle) */
-#define A0_LOG    (0xc<<24)     /* dst.xyzw = log2(abs(src0.replicate_swizzle)) */
-#define A0_CMP    (0xd<<24)     /* dst = (src0 >= 0.0) ? src1 : src2 */
-#define A0_MIN    (0xe<<24)     /* dst = (src0 < src1) ? src0 : src1 */
-#define A0_MAX    (0xf<<24)     /* dst = (src0 >= src1) ? src0 : src1 */
-#define A0_FLR    (0x10<<24)    /* dst = floor(src0) */
-#define A0_MOD    (0x11<<24)    /* dst = src0 fmod 1.0 */
-#define A0_TRC    (0x12<<24)    /* dst = int(src0) */
-#define A0_SGE    (0x13<<24)    /* dst = src0 >= src1 ? 1.0 : 0.0 */
-#define A0_SLT    (0x14<<24)    /* dst = src0 < src1 ? 1.0 : 0.0 */
-#define A0_DEST_SATURATE                 (1<<22)
-#define A0_DEST_TYPE_SHIFT                19
-/* Allow: R, OC, OD, U */
-#define A0_DEST_NR_SHIFT                 14
-/* Allow R: 0..15, OC,OD: 0..0, U: 0..2 */
-#define A0_DEST_CHANNEL_X                (1<<10)
-#define A0_DEST_CHANNEL_Y                (2<<10)
-#define A0_DEST_CHANNEL_Z                (4<<10)
-#define A0_DEST_CHANNEL_W                (8<<10)
-#define A0_DEST_CHANNEL_ALL              (0xf<<10)
-#define A0_DEST_CHANNEL_SHIFT            10
-#define A0_SRC0_TYPE_SHIFT               7
-#define A0_SRC0_NR_SHIFT                 2
-
-#define A0_DEST_CHANNEL_XY              (A0_DEST_CHANNEL_X|A0_DEST_CHANNEL_Y)
-#define A0_DEST_CHANNEL_XYZ             (A0_DEST_CHANNEL_XY|A0_DEST_CHANNEL_Z)
-
-
-#define SRC_X        0
-#define SRC_Y        1
-#define SRC_Z        2
-#define SRC_W        3
-#define SRC_ZERO     4
-#define SRC_ONE      5
-
-#define A1_SRC0_CHANNEL_X_NEGATE         (1<<31)
-#define A1_SRC0_CHANNEL_X_SHIFT          28
-#define A1_SRC0_CHANNEL_Y_NEGATE         (1<<27)
-#define A1_SRC0_CHANNEL_Y_SHIFT          24
-#define A1_SRC0_CHANNEL_Z_NEGATE         (1<<23)
-#define A1_SRC0_CHANNEL_Z_SHIFT          20
-#define A1_SRC0_CHANNEL_W_NEGATE         (1<<19)
-#define A1_SRC0_CHANNEL_W_SHIFT          16
-#define A1_SRC1_TYPE_SHIFT               13
-#define A1_SRC1_NR_SHIFT                 8
-#define A1_SRC1_CHANNEL_X_NEGATE         (1<<7)
-#define A1_SRC1_CHANNEL_X_SHIFT          4
-#define A1_SRC1_CHANNEL_Y_NEGATE         (1<<3)
-#define A1_SRC1_CHANNEL_Y_SHIFT          0
-
-#define A2_SRC1_CHANNEL_Z_NEGATE         (1<<31)
-#define A2_SRC1_CHANNEL_Z_SHIFT          28
-#define A2_SRC1_CHANNEL_W_NEGATE         (1<<27)
-#define A2_SRC1_CHANNEL_W_SHIFT          24
-#define A2_SRC2_TYPE_SHIFT               21
-#define A2_SRC2_NR_SHIFT                 16
-#define A2_SRC2_CHANNEL_X_NEGATE         (1<<15)
-#define A2_SRC2_CHANNEL_X_SHIFT          12
-#define A2_SRC2_CHANNEL_Y_NEGATE         (1<<11)
-#define A2_SRC2_CHANNEL_Y_SHIFT          8
-#define A2_SRC2_CHANNEL_Z_NEGATE         (1<<7)
-#define A2_SRC2_CHANNEL_Z_SHIFT          4
-#define A2_SRC2_CHANNEL_W_NEGATE         (1<<3)
-#define A2_SRC2_CHANNEL_W_SHIFT          0
-
-
-
-/* Texture instructions */
-#define T0_TEXLD     (0x15<<24) /* Sample texture using predeclared
-                                 * sampler and address, and output
-                                 * filtered texel data to destination
-                                 * register */
-#define T0_TEXLDP    (0x16<<24) /* Same as texld but performs a
-                                 * perspective divide of the texture
-                                 * coordinate .xyz values by .w before
-                                 * sampling. */
-#define T0_TEXLDB    (0x17<<24) /* Same as texld but biases the
-                                 * computed LOD by w.  Only S4.6 two's
-                                 * comp is used.  This implies that a
-                                 * float to fixed conversion is
-                                 * done. */
-#define T0_TEXKILL   (0x18<<24) /* Does not perform a sampling
-                                 * operation.  Simply kills the pixel
-                                 * if any channel of the address
-                                 * register is < 0.0. */
-#define T0_DEST_TYPE_SHIFT                19
-/* Allow: R, OC, OD, U */
-/* Note: U (unpreserved) regs do not retain their values between
- * phases (cannot be used for feedback) 
- *
- * Note: oC and OD registers can only be used as the destination of a
- * texture instruction once per phase (this is an implementation
- * restriction). 
- */
-#define T0_DEST_NR_SHIFT                 14
-/* Allow R: 0..15, OC,OD: 0..0, U: 0..2 */
-#define T0_SAMPLER_NR_SHIFT              0      /* This field ignored for TEXKILL */
-#define T0_SAMPLER_NR_MASK               (0xf<<0)
-
-#define T1_ADDRESS_REG_TYPE_SHIFT        24     /* Reg to use as texture coord */
-/* Allow R, T, OC, OD -- R, OC, OD are 'dependent' reads, new program phase */
-#define T1_ADDRESS_REG_NR_SHIFT          17
-#define T2_MBZ                           0
-
-/* Declaration instructions */
-#define D0_DCL       (0x19<<24) /* Declare a t (interpolated attrib)
-                                 * register or an s (sampler)
-                                 * register. */
-#define D0_SAMPLE_TYPE_SHIFT              22
-#define D0_SAMPLE_TYPE_2D                 (0x0<<22)
-#define D0_SAMPLE_TYPE_CUBE               (0x1<<22)
-#define D0_SAMPLE_TYPE_VOLUME             (0x2<<22)
-#define D0_SAMPLE_TYPE_MASK               (0x3<<22)
-
-#define D0_TYPE_SHIFT                19
-/* Allow: T, S */
-#define D0_NR_SHIFT                  14
-/* Allow T: 0..10, S: 0..15 */
-#define D0_CHANNEL_X                (1<<10)
-#define D0_CHANNEL_Y                (2<<10)
-#define D0_CHANNEL_Z                (4<<10)
-#define D0_CHANNEL_W                (8<<10)
-#define D0_CHANNEL_ALL              (0xf<<10)
-#define D0_CHANNEL_NONE             (0<<10)
-
-#define D0_CHANNEL_XY               (D0_CHANNEL_X|D0_CHANNEL_Y)
-#define D0_CHANNEL_XYZ              (D0_CHANNEL_XY|D0_CHANNEL_Z)
-
-/* I915 Errata: Do not allow (xz), (xw), (xzw) combinations for diffuse
- * or specular declarations. 
- *
- * For T dcls, only allow: (x), (xy), (xyz), (w), (xyzw) 
- *
- * Must be zero for S (sampler) dcls
- */
-#define D1_MBZ                          0
-#define D2_MBZ                          0
-
-
-
-/* p207 */
-#define _3DSTATE_MAP_STATE               (CMD_3D|(0x1d<<24)|(0x0<<16))
-
-#define MS1_MAPMASK_SHIFT               0
-#define MS1_MAPMASK_MASK                (0x8fff<<0)
-
-#define MS2_UNTRUSTED_SURFACE           (1<<31)
-#define MS2_ADDRESS_MASK                0xfffffffc
-#define MS2_VERTICAL_LINE_STRIDE        (1<<1)
-#define MS2_VERTICAL_OFFSET             (1<<1)
-
-#define MS3_HEIGHT_SHIFT              21
-#define MS3_WIDTH_SHIFT               10
-#define MS3_PALETTE_SELECT            (1<<9)
-#define MS3_MAPSURF_FORMAT_SHIFT      7
-#define MS3_MAPSURF_FORMAT_MASK       (0x7<<7)
-#define    MAPSURF_8BIT		 	   (1<<7)
-#define    MAPSURF_16BIT		   (2<<7)
-#define    MAPSURF_32BIT		   (3<<7)
-#define    MAPSURF_422			   (5<<7)
-#define    MAPSURF_COMPRESSED		   (6<<7)
-#define    MAPSURF_4BIT_INDEXED		   (7<<7)
-#define MS3_MT_FORMAT_MASK         (0x7 << 3)
-#define MS3_MT_FORMAT_SHIFT        3
-#define    MT_4BIT_IDX_ARGB8888	           (7<<3)       /* SURFACE_4BIT_INDEXED */
-#define    MT_8BIT_I8		           (0<<3)       /* SURFACE_8BIT */
-#define    MT_8BIT_L8		           (1<<3)
-#define    MT_8BIT_A8		           (4<<3)
-#define    MT_8BIT_MONO8	           (5<<3)
-#define    MT_16BIT_RGB565 		   (0<<3)       /* SURFACE_16BIT */
-#define    MT_16BIT_ARGB1555		   (1<<3)
-#define    MT_16BIT_ARGB4444		   (2<<3)
-#define    MT_16BIT_AY88		   (3<<3)
-#define    MT_16BIT_88DVDU	           (5<<3)
-#define    MT_16BIT_BUMP_655LDVDU	   (6<<3)
-#define    MT_16BIT_I16	                   (7<<3)
-#define    MT_16BIT_L16	                   (8<<3)
-#define    MT_16BIT_A16	                   (9<<3)
-#define    MT_32BIT_ARGB8888		   (0<<3)       /* SURFACE_32BIT */
-#define    MT_32BIT_ABGR8888		   (1<<3)
-#define    MT_32BIT_XRGB8888		   (2<<3)
-#define    MT_32BIT_XBGR8888		   (3<<3)
-#define    MT_32BIT_QWVU8888		   (4<<3)
-#define    MT_32BIT_AXVU8888		   (5<<3)
-#define    MT_32BIT_LXVU8888	           (6<<3)
-#define    MT_32BIT_XLVU8888	           (7<<3)
-#define    MT_32BIT_ARGB2101010	           (8<<3)
-#define    MT_32BIT_ABGR2101010	           (9<<3)
-#define    MT_32BIT_AWVU2101010	           (0xA<<3)
-#define    MT_32BIT_GR1616	           (0xB<<3)
-#define    MT_32BIT_VU1616	           (0xC<<3)
-#define    MT_32BIT_xI824	           (0xD<<3)
-#define    MT_32BIT_xA824	           (0xE<<3)
-#define    MT_32BIT_xL824	           (0xF<<3)
-#define    MT_422_YCRCB_SWAPY	           (0<<3)       /* SURFACE_422 */
-#define    MT_422_YCRCB_NORMAL	           (1<<3)
-#define    MT_422_YCRCB_SWAPUV	           (2<<3)
-#define    MT_422_YCRCB_SWAPUVY	           (3<<3)
-#define    MT_COMPRESS_DXT1		   (0<<3)       /* SURFACE_COMPRESSED */
-#define    MT_COMPRESS_DXT2_3	           (1<<3)
-#define    MT_COMPRESS_DXT4_5	           (2<<3)
-#define    MT_COMPRESS_FXT1		   (3<<3)
-#define    MT_COMPRESS_DXT1_RGB		   (4<<3)
-#define MS3_USE_FENCE_REGS              (1<<2)
-#define MS3_TILED_SURFACE             (1<<1)
-#define MS3_TILE_WALK                 (1<<0)
-
-#define MS4_PITCH_SHIFT                 21
-#define MS4_CUBE_FACE_ENA_NEGX          (1<<20)
-#define MS4_CUBE_FACE_ENA_POSX          (1<<19)
-#define MS4_CUBE_FACE_ENA_NEGY          (1<<18)
-#define MS4_CUBE_FACE_ENA_POSY          (1<<17)
-#define MS4_CUBE_FACE_ENA_NEGZ          (1<<16)
-#define MS4_CUBE_FACE_ENA_POSZ          (1<<15)
-#define MS4_CUBE_FACE_ENA_MASK          (0x3f<<15)
-#define MS4_MAX_LOD_SHIFT		9
-#define MS4_MAX_LOD_MASK		(0x3f<<9)
-#define MS4_MIP_LAYOUT_LEGACY           (0<<8)
-#define MS4_MIP_LAYOUT_BELOW_LPT        (0<<8)
-#define MS4_MIP_LAYOUT_RIGHT_LPT        (1<<8)
-#define MS4_VOLUME_DEPTH_SHIFT          0
-#define MS4_VOLUME_DEPTH_MASK           (0xff<<0)
-
-/* p244 */
-#define _3DSTATE_SAMPLER_STATE         (CMD_3D|(0x1d<<24)|(0x1<<16))
-
-#define SS1_MAPMASK_SHIFT               0
-#define SS1_MAPMASK_MASK                (0x8fff<<0)
-
-#define SS2_REVERSE_GAMMA_ENABLE        (1<<31)
-#define SS2_PACKED_TO_PLANAR_ENABLE     (1<<30)
-#define SS2_COLORSPACE_CONVERSION       (1<<29)
-#define SS2_CHROMAKEY_SHIFT             27
-#define SS2_BASE_MIP_LEVEL_SHIFT        22
-#define SS2_BASE_MIP_LEVEL_MASK         (0x1f<<22)
-#define SS2_MIP_FILTER_SHIFT            20
-#define SS2_MIP_FILTER_MASK             (0x3<<20)
-#define   MIPFILTER_NONE       	0
-#define   MIPFILTER_NEAREST	1
-#define   MIPFILTER_LINEAR	3
-#define SS2_MAG_FILTER_SHIFT          17
-#define SS2_MAG_FILTER_MASK           (0x7<<17)
-#define   FILTER_NEAREST	0
-#define   FILTER_LINEAR		1
-#define   FILTER_ANISOTROPIC	2
-#define   FILTER_4X4_1    	3
-#define   FILTER_4X4_2    	4
-#define   FILTER_4X4_FLAT 	5
-#define   FILTER_6X5_MONO   	6       /* XXX - check */
-#define SS2_MIN_FILTER_SHIFT          14
-#define SS2_MIN_FILTER_MASK           (0x7<<14)
-#define SS2_LOD_BIAS_SHIFT            5
-#define SS2_LOD_BIAS_ONE              (0x10<<5)
-#define SS2_LOD_BIAS_MASK             (0x1ff<<5)
-/* Shadow requires:
- *  MT_X8{I,L,A}24 or MT_{I,L,A}16 texture format
- *  FILTER_4X4_x  MIN and MAG filters
- */
-#define SS2_SHADOW_ENABLE             (1<<4)
-#define SS2_MAX_ANISO_MASK            (1<<3)
-#define SS2_MAX_ANISO_2               (0<<3)
-#define SS2_MAX_ANISO_4               (1<<3)
-#define SS2_SHADOW_FUNC_SHIFT         0
-#define SS2_SHADOW_FUNC_MASK          (0x7<<0)
-/* SS2_SHADOW_FUNC values: see COMPAREFUNC_* */
-
-#define SS3_MIN_LOD_SHIFT            24
-#define SS3_MIN_LOD_ONE              (0x10<<24)
-#define SS3_MIN_LOD_MASK             (0xff<<24)
-#define SS3_KILL_PIXEL_ENABLE        (1<<17)
-#define SS3_TCX_ADDR_MODE_SHIFT      12
-#define SS3_TCX_ADDR_MODE_MASK       (0x7<<12)
-#define   TEXCOORDMODE_WRAP		0
-#define   TEXCOORDMODE_MIRROR		1
-#define   TEXCOORDMODE_CLAMP_EDGE	2
-#define   TEXCOORDMODE_CUBE       	3
-#define   TEXCOORDMODE_CLAMP_BORDER	4
-#define   TEXCOORDMODE_MIRROR_ONCE      5
-#define SS3_TCY_ADDR_MODE_SHIFT      9
-#define SS3_TCY_ADDR_MODE_MASK       (0x7<<9)
-#define SS3_TCZ_ADDR_MODE_SHIFT      6
-#define SS3_TCZ_ADDR_MODE_MASK       (0x7<<6)
-#define SS3_NORMALIZED_COORDS        (1<<5)
-#define SS3_TEXTUREMAP_INDEX_SHIFT   1
-#define SS3_TEXTUREMAP_INDEX_MASK    (0xf<<1)
-#define SS3_DEINTERLACER_ENABLE      (1<<0)
-
-#define SS4_BORDER_COLOR_MASK        (~0)
-
-/* 3DSTATE_SPAN_STIPPLE, p258
- */
-#define _3DSTATE_STIPPLE           ((0x3<<29)|(0x1d<<24)|(0x83<<16))
-#define ST1_ENABLE               (1<<16)
-#define ST1_MASK                 (0xffff)
-
-#define _3DSTATE_DEFAULT_Z          ((0x3<<29)|(0x1d<<24)|(0x98<<16))
-#define _3DSTATE_DEFAULT_DIFFUSE    ((0x3<<29)|(0x1d<<24)|(0x99<<16))
-#define _3DSTATE_DEFAULT_SPECULAR   ((0x3<<29)|(0x1d<<24)|(0x9a<<16))
-
-
-#define MI_FLUSH                   ((0<<29)|(4<<23))
-#define FLUSH_MAP_CACHE            (1<<0)
-#define INHIBIT_FLUSH_RENDER_CACHE (1<<2)
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/i915_state.c b/src/mesa/drivers/dri/i915tex/i915_state.c
deleted file mode 100644
index e5d8d279936..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_state.c
+++ /dev/null
@@ -1,1014 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "dd.h"
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-
-#include "texmem.h"
-
-#include "drivers/common/driverfuncs.h"
-
-#include "intel_fbo.h"
-#include "intel_screen.h"
-#include "intel_batchbuffer.h"
-
-#include "i915_context.h"
-#include "i915_reg.h"
-
-#define FILE_DEBUG_FLAG DEBUG_STATE
-
-static void
-i915StencilFuncSeparate(GLcontext * ctx, GLenum face, GLenum func, GLint ref,
-                        GLuint mask)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func(func);
-
-   mask = mask & 0xff;
-
-   DBG("%s : func: %s, ref : 0x%x, mask: 0x%x\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(func), ref, mask);
-
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_TEST_MASK;
-   i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_TEST_MASK |
-                                           STENCIL_TEST_MASK(mask));
-
-   i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_REF_MASK |
-                                          S5_STENCIL_TEST_FUNC_MASK);
-
-   i915->state.Ctx[I915_CTXREG_LIS5] |= ((ref << S5_STENCIL_REF_SHIFT) |
-                                         (test <<
-                                          S5_STENCIL_TEST_FUNC_SHIFT));
-}
-
-static void
-i915StencilMaskSeparate(GLcontext * ctx, GLenum face, GLuint mask)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   DBG("%s : mask 0x%x\n", __FUNCTION__, mask);
-   
-   mask = mask & 0xff;
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_STATE4] &= ~MODE4_ENABLE_STENCIL_WRITE_MASK;
-   i915->state.Ctx[I915_CTXREG_STATE4] |= (ENABLE_STENCIL_WRITE_MASK |
-                                           STENCIL_WRITE_MASK(mask));
-}
-
-
-static void
-i915StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail, GLenum zfail,
-                      GLenum zpass)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int fop = intel_translate_stencil_op(fail);
-   int dfop = intel_translate_stencil_op(zfail);
-   int dpop = intel_translate_stencil_op(zpass);
-
-
-   DBG("%s: fail : %s, zfail: %s, zpass : %s\n", __FUNCTION__,
-       _mesa_lookup_enum_by_nr(fail),
-       _mesa_lookup_enum_by_nr(zfail), _mesa_lookup_enum_by_nr(zpass));
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-
-   i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_FAIL_MASK |
-                                          S5_STENCIL_PASS_Z_FAIL_MASK |
-                                          S5_STENCIL_PASS_Z_PASS_MASK);
-
-   i915->state.Ctx[I915_CTXREG_LIS5] |= ((fop << S5_STENCIL_FAIL_SHIFT) |
-                                         (dfop <<
-                                          S5_STENCIL_PASS_Z_FAIL_SHIFT) |
-                                         (dpop <<
-                                          S5_STENCIL_PASS_Z_PASS_SHIFT));
-}
-
-static void
-i915AlphaFunc(GLcontext * ctx, GLenum func, GLfloat ref)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func(func);
-   GLubyte refByte;
-
-   UNCLAMPED_FLOAT_TO_UBYTE(refByte, ref);
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_LIS6] &= ~(S6_ALPHA_TEST_FUNC_MASK |
-                                          S6_ALPHA_REF_MASK);
-   i915->state.Ctx[I915_CTXREG_LIS6] |= ((test << S6_ALPHA_TEST_FUNC_SHIFT) |
-                                         (((GLuint) refByte) <<
-                                          S6_ALPHA_REF_SHIFT));
-}
-
-/* This function makes sure that the proper enables are
- * set for LogicOp, Independant Alpha Blend, and Blending.
- * It needs to be called from numerous places where we
- * could change the LogicOp or Independant Alpha Blend without subsequent
- * calls to glEnable.
- */
-static void
-i915EvalLogicOpBlendState(GLcontext * ctx)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-
-   if (RGBA_LOGICOP_ENABLED(ctx)) {
-      i915->state.Ctx[I915_CTXREG_LIS5] |= S5_LOGICOP_ENABLE;
-      i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
-   }
-   else {
-      i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_LOGICOP_ENABLE;
-
-      if (ctx->Color.BlendEnabled) {
-         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_CBUF_BLEND_ENABLE;
-      }
-      else {
-         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_CBUF_BLEND_ENABLE;
-      }
-   }
-}
-
-static void
-i915BlendColor(GLcontext * ctx, const GLfloat color[4])
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   GLubyte r, g, b, a;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   UNCLAMPED_FLOAT_TO_UBYTE(r, color[RCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(g, color[GCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(b, color[BCOMP]);
-   UNCLAMPED_FLOAT_TO_UBYTE(a, color[ACOMP]);
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] =
-      (a << 24) | (r << 16) | (g << 8) | b;
-}
-
-
-#define DST_BLND_FACT(f) ((f)<<S6_CBUF_DST_BLEND_FACT_SHIFT)
-#define SRC_BLND_FACT(f) ((f)<<S6_CBUF_SRC_BLEND_FACT_SHIFT)
-#define DST_ABLND_FACT(f) ((f)<<IAB_DST_FACTOR_SHIFT)
-#define SRC_ABLND_FACT(f) ((f)<<IAB_SRC_FACTOR_SHIFT)
-
-
-
-static GLuint
-translate_blend_equation(GLenum mode)
-{
-   switch (mode) {
-   case GL_FUNC_ADD:
-      return BLENDFUNC_ADD;
-   case GL_MIN:
-      return BLENDFUNC_MIN;
-   case GL_MAX:
-      return BLENDFUNC_MAX;
-   case GL_FUNC_SUBTRACT:
-      return BLENDFUNC_SUBTRACT;
-   case GL_FUNC_REVERSE_SUBTRACT:
-      return BLENDFUNC_REVERSE_SUBTRACT;
-   default:
-      return 0;
-   }
-}
-
-static void
-i915UpdateBlendState(GLcontext * ctx)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   GLuint iab = (i915->state.Ctx[I915_CTXREG_IAB] &
-                 ~(IAB_SRC_FACTOR_MASK |
-                   IAB_DST_FACTOR_MASK |
-                   (BLENDFUNC_MASK << IAB_FUNC_SHIFT) | IAB_ENABLE));
-
-   GLuint lis6 = (i915->state.Ctx[I915_CTXREG_LIS6] &
-                  ~(S6_CBUF_SRC_BLEND_FACT_MASK |
-                    S6_CBUF_DST_BLEND_FACT_MASK | S6_CBUF_BLEND_FUNC_MASK));
-
-   GLuint eqRGB = ctx->Color.BlendEquationRGB;
-   GLuint eqA = ctx->Color.BlendEquationA;
-   GLuint srcRGB = ctx->Color.BlendSrcRGB;
-   GLuint dstRGB = ctx->Color.BlendDstRGB;
-   GLuint srcA = ctx->Color.BlendSrcA;
-   GLuint dstA = ctx->Color.BlendDstA;
-
-   if (eqRGB == GL_MIN || eqRGB == GL_MAX) {
-      srcRGB = dstRGB = GL_ONE;
-   }
-
-   if (eqA == GL_MIN || eqA == GL_MAX) {
-      srcA = dstA = GL_ONE;
-   }
-
-   lis6 |= SRC_BLND_FACT(intel_translate_blend_factor(srcRGB));
-   lis6 |= DST_BLND_FACT(intel_translate_blend_factor(dstRGB));
-   lis6 |= translate_blend_equation(eqRGB) << S6_CBUF_BLEND_FUNC_SHIFT;
-
-   iab |= SRC_ABLND_FACT(intel_translate_blend_factor(srcA));
-   iab |= DST_ABLND_FACT(intel_translate_blend_factor(dstA));
-   iab |= translate_blend_equation(eqA) << IAB_FUNC_SHIFT;
-
-   if (srcA != srcRGB || dstA != dstRGB || eqA != eqRGB)
-      iab |= IAB_ENABLE;
-
-   if (iab != i915->state.Ctx[I915_CTXREG_IAB] ||
-       lis6 != i915->state.Ctx[I915_CTXREG_LIS6]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      i915->state.Ctx[I915_CTXREG_IAB] = iab;
-      i915->state.Ctx[I915_CTXREG_LIS6] = lis6;
-   }
-
-   /* This will catch a logicop blend equation */
-   i915EvalLogicOpBlendState(ctx);
-}
-
-
-static void
-i915BlendFuncSeparate(GLcontext * ctx, GLenum srcRGB,
-                      GLenum dstRGB, GLenum srcA, GLenum dstA)
-{
-   i915UpdateBlendState(ctx);
-}
-
-
-static void
-i915BlendEquationSeparate(GLcontext * ctx, GLenum eqRGB, GLenum eqA)
-{
-   i915UpdateBlendState(ctx);
-}
-
-
-static void
-i915DepthFunc(GLcontext * ctx, GLenum func)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int test = intel_translate_compare_func(func);
-
-   DBG("%s\n", __FUNCTION__);
-   
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_FUNC_MASK;
-   i915->state.Ctx[I915_CTXREG_LIS6] |= test << S6_DEPTH_TEST_FUNC_SHIFT;
-}
-
-static void
-i915DepthMask(GLcontext * ctx, GLboolean flag)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   DBG("%s flag (%d)\n", __FUNCTION__, flag);
-   
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-
-   if (flag && ctx->Depth.Test)
-      i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_WRITE_ENABLE;
-   else
-      i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_WRITE_ENABLE;
-}
-
-/* =============================================================
- * Polygon stipple
- *
- * The i915 supports a 4x4 stipple natively, GL wants 32x32.
- * Fortunately stipple is usually a repeating pattern.
- */
-static void
-i915PolygonStipple(GLcontext * ctx, const GLubyte * mask)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   const GLubyte *m = mask;
-   GLubyte p[4];
-   int i, j, k;
-   int active = (ctx->Polygon.StippleFlag &&
-                 i915->intel.reduced_primitive == GL_TRIANGLES);
-   GLuint newMask;
-
-   if (active) {
-      I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-      i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
-   }
-
-   p[0] = mask[12] & 0xf;
-   p[0] |= p[0] << 4;
-   p[1] = mask[8] & 0xf;
-   p[1] |= p[1] << 4;
-   p[2] = mask[4] & 0xf;
-   p[2] |= p[2] << 4;
-   p[3] = mask[0] & 0xf;
-   p[3] |= p[3] << 4;
-
-   for (k = 0; k < 8; k++)
-      for (j = 3; j >= 0; j--)
-         for (i = 0; i < 4; i++, m++)
-            if (*m != p[j]) {
-               i915->intel.hw_stipple = 0;
-               return;
-            }
-
-   newMask = (((p[0] & 0xf) << 0) |
-              ((p[1] & 0xf) << 4) |
-              ((p[2] & 0xf) << 8) | ((p[3] & 0xf) << 12));
-
-
-   if (newMask == 0xffff || newMask == 0x0) {
-      /* this is needed to make conform pass */
-      i915->intel.hw_stipple = 0;
-      return;
-   }
-
-   i915->state.Stipple[I915_STPREG_ST1] &= ~0xffff;
-   i915->state.Stipple[I915_STPREG_ST1] |= newMask;
-   i915->intel.hw_stipple = 1;
-
-   if (active)
-      i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
-}
-
-
-/* =============================================================
- * Hardware clipping
- */
-static void
-i915Scissor(GLcontext * ctx, GLint x, GLint y, GLsizei w, GLsizei h)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int x1, y1, x2, y2;
-
-   if (!ctx->DrawBuffer)
-      return;
-
-   DBG("%s %d,%d %dx%d\n", __FUNCTION__, x, y, w, h);
-
-   if (ctx->DrawBuffer->Name == 0) {
-      x1 = x;
-      y1 = ctx->DrawBuffer->Height - (y + h);
-      x2 = x + w - 1;
-      y2 = y1 + h - 1;
-      DBG("%s %d..%d,%d..%d (inverted)\n", __FUNCTION__, x1, x2, y1, y2);
-   }
-   else {
-      /* FBO - not inverted
-       */
-      x1 = x;
-      y1 = y;
-      x2 = x + w - 1;
-      y2 = y + h - 1;
-      DBG("%s %d..%d,%d..%d (not inverted)\n", __FUNCTION__, x1, x2, y1, y2);
-   }
-   
-   x1 = CLAMP(x1, 0, ctx->DrawBuffer->Width - 1);
-   y1 = CLAMP(y1, 0, ctx->DrawBuffer->Height - 1);
-   x2 = CLAMP(x2, 0, ctx->DrawBuffer->Width - 1);
-   y2 = CLAMP(y2, 0, ctx->DrawBuffer->Height - 1);
-   
-   DBG("%s %d..%d,%d..%d (clamped)\n", __FUNCTION__, x1, x2, y1, y2);
-
-   I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
-   i915->state.Buffer[I915_DESTREG_SR1] = (y1 << 16) | (x1 & 0xffff);
-   i915->state.Buffer[I915_DESTREG_SR2] = (y2 << 16) | (x2 & 0xffff);
-}
-
-static void
-i915LogicOp(GLcontext * ctx, GLenum opcode)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int tmp = intel_translate_logic_op(opcode);
-
-   DBG("%s\n", __FUNCTION__);
-   
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_STATE4] &= ~LOGICOP_MASK;
-   i915->state.Ctx[I915_CTXREG_STATE4] |= LOGIC_OP_FUNC(tmp);
-}
-
-
-
-static void
-i915CullFaceFrontFace(GLcontext * ctx, GLenum unused)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   GLuint mode;
-
-   DBG("%s %d\n", __FUNCTION__,
-       ctx->DrawBuffer ? ctx->DrawBuffer->Name : 0);
-
-   if (!ctx->Polygon.CullFlag) {
-      mode = S4_CULLMODE_NONE;
-   }
-   else if (ctx->Polygon.CullFaceMode != GL_FRONT_AND_BACK) {
-      mode = S4_CULLMODE_CW;
-
-      if (ctx->DrawBuffer && ctx->DrawBuffer->Name != 0)
-         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
-      if (ctx->Polygon.CullFaceMode == GL_FRONT)
-         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
-      if (ctx->Polygon.FrontFace != GL_CCW)
-         mode ^= (S4_CULLMODE_CW ^ S4_CULLMODE_CCW);
-   }
-   else {
-      mode = S4_CULLMODE_BOTH;
-   }
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_CULLMODE_MASK;
-   i915->state.Ctx[I915_CTXREG_LIS4] |= mode;
-}
-
-static void
-i915LineWidth(GLcontext * ctx, GLfloat widthf)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_LINE_WIDTH_MASK;
-   int width;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   width = (int) (widthf * 2);
-   CLAMP_SELF(width, 1, 0xf);
-   lis4 |= width << S4_LINE_WIDTH_SHIFT;
-
-   if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      i915->state.Ctx[I915_CTXREG_LIS4] = lis4;
-   }
-}
-
-static void
-i915PointSize(GLcontext * ctx, GLfloat size)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   int lis4 = i915->state.Ctx[I915_CTXREG_LIS4] & ~S4_POINT_WIDTH_MASK;
-   GLint point_size = (int) size;
-
-   DBG("%s\n", __FUNCTION__);
-   
-   CLAMP_SELF(point_size, 1, 255);
-   lis4 |= point_size << S4_POINT_WIDTH_SHIFT;
-
-   if (lis4 != i915->state.Ctx[I915_CTXREG_LIS4]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      i915->state.Ctx[I915_CTXREG_LIS4] = lis4;
-   }
-}
-
-
-/* =============================================================
- * Color masks
- */
-
-static void
-i915ColorMask(GLcontext * ctx,
-              GLboolean r, GLboolean g, GLboolean b, GLboolean a)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   GLuint tmp = i915->state.Ctx[I915_CTXREG_LIS5] & ~S5_WRITEDISABLE_MASK;
-
-   DBG("%s r(%d) g(%d) b(%d) a(%d)\n", __FUNCTION__, r, g, b,
-       a);
-
-   if (!r)
-      tmp |= S5_WRITEDISABLE_RED;
-   if (!g)
-      tmp |= S5_WRITEDISABLE_GREEN;
-   if (!b)
-      tmp |= S5_WRITEDISABLE_BLUE;
-   if (!a)
-      tmp |= S5_WRITEDISABLE_ALPHA;
-
-   if (tmp != i915->state.Ctx[I915_CTXREG_LIS5]) {
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      i915->state.Ctx[I915_CTXREG_LIS5] = tmp;
-   }
-}
-
-static void
-update_specular(GLcontext * ctx)
-{
-   /* A hack to trigger the rebuild of the fragment program.
-    */
-   intel_context(ctx)->NewGLState |= _NEW_TEXTURE;
-}
-
-static void
-i915LightModelfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
-{
-   DBG("%s\n", __FUNCTION__);
-   
-   if (pname == GL_LIGHT_MODEL_COLOR_CONTROL) {
-      update_specular(ctx);
-   }
-}
-
-static void
-i915ShadeModel(GLcontext * ctx, GLenum mode)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-
-   if (mode == GL_SMOOTH) {
-      i915->state.Ctx[I915_CTXREG_LIS4] &= ~(S4_FLATSHADE_ALPHA |
-                                             S4_FLATSHADE_COLOR |
-                                             S4_FLATSHADE_SPECULAR);
-   }
-   else {
-      i915->state.Ctx[I915_CTXREG_LIS4] |= (S4_FLATSHADE_ALPHA |
-                                            S4_FLATSHADE_COLOR |
-                                            S4_FLATSHADE_SPECULAR);
-   }
-}
-
-/* =============================================================
- * Fog
- */
-void
-i915_update_fog(GLcontext * ctx)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-   GLenum mode;
-   GLboolean enabled;
-   GLboolean try_pixel_fog;
-
-   if (ctx->FragmentProgram._Active) {
-      /* Pull in static fog state from program */
-      mode = ctx->FragmentProgram._Current->FogOption;
-      enabled = (mode != GL_NONE);
-      try_pixel_fog = 0;
-   }
-   else {
-      enabled = ctx->Fog.Enabled;
-      mode = ctx->Fog.Mode;
-#if 0
-      /* XXX - DISABLED -- Need ortho fallback */
-      try_pixel_fog = (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT
-                       && ctx->Hint.Fog == GL_NICEST);
-#else
-      try_pixel_fog = 0;
-#endif
-   }
-
-   if (!enabled) {
-      i915->vertex_fog = I915_FOG_NONE;
-   }
-   else if (try_pixel_fog) {
-      I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
-      i915->vertex_fog = I915_FOG_PIXEL;
-
-      switch (mode) {
-      case GL_LINEAR:
-         if (ctx->Fog.End <= ctx->Fog.Start) {
-            /* XXX - this won't work with fragment programs.  Need to
-             * either fallback or append fog instructions to end of
-             * program in the case of linear fog.
-             */
-            printf("vertex fog!\n");
-            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
-            i915->vertex_fog = I915_FOG_VERTEX;
-         }
-         else {
-            GLfloat c2 = 1.0 / (ctx->Fog.End - ctx->Fog.Start);
-            GLfloat c1 = ctx->Fog.End * c2;
-
-            i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_C1_MASK;
-            i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_LINEAR;
-            i915->state.Fog[I915_FOGREG_MODE1] |=
-               ((GLuint) (c1 * FMC1_C1_ONE)) & FMC1_C1_MASK;
-
-            if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-               i915->state.Fog[I915_FOGREG_MODE2]
-                  = (GLuint) (c2 * FMC2_C2_ONE);
-            }
-            else {
-               fi_type fi;
-               fi.f = c2;
-               i915->state.Fog[I915_FOGREG_MODE2] = fi.i;
-            }
-         }
-         break;
-      case GL_EXP:
-         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP;
-         break;
-      case GL_EXP2:
-         i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_PIXEL_EXP2;
-         break;
-      default:
-         break;
-      }
-   }
-   else { /* if (i915->vertex_fog != I915_FOG_VERTEX) */
-      I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_MODE1] &= ~FMC1_FOGFUNC_MASK;
-      i915->state.Fog[I915_FOGREG_MODE1] |= FMC1_FOGFUNC_VERTEX;
-      i915->vertex_fog = I915_FOG_VERTEX;
-   }
-
-   I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-   I915_ACTIVESTATE(i915, I915_UPLOAD_FOG, enabled);
-   if (enabled)
-      i915->state.Ctx[I915_CTXREG_LIS5] |= S5_FOG_ENABLE;
-   else
-      i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_FOG_ENABLE;
-
-   /* Always enable pixel fog.  Vertex fog using fog coord will conflict
-    * with fog code appended onto fragment program.
-    */
-    _tnl_allow_vertex_fog( ctx, 0 );
-    _tnl_allow_pixel_fog( ctx, 1 );
-}
-
-static void
-i915Fogfv(GLcontext * ctx, GLenum pname, const GLfloat * param)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   switch (pname) {
-   case GL_FOG_COORDINATE_SOURCE_EXT:
-   case GL_FOG_MODE:
-   case GL_FOG_START:
-   case GL_FOG_END:
-      break;
-
-   case GL_FOG_DENSITY:
-      I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-
-      if (i915->state.Fog[I915_FOGREG_MODE1] & FMC1_FOGINDEX_Z) {
-         i915->state.Fog[I915_FOGREG_MODE3] =
-            (GLuint) (ctx->Fog.Density * FMC3_D_ONE);
-      }
-      else {
-         fi_type fi;
-         fi.f = ctx->Fog.Density;
-         i915->state.Fog[I915_FOGREG_MODE3] = fi.i;
-      }
-      break;
-
-   case GL_FOG_COLOR:
-      I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_COLOR] =
-         (_3DSTATE_FOG_COLOR_CMD |
-          ((GLubyte) (ctx->Fog.Color[0] * 255.0F) << 16) |
-          ((GLubyte) (ctx->Fog.Color[1] * 255.0F) << 8) |
-          ((GLubyte) (ctx->Fog.Color[2] * 255.0F) << 0));
-      break;
-
-   default:
-      break;
-   }
-}
-
-static void
-i915Hint(GLcontext * ctx, GLenum target, GLenum state)
-{
-   switch (target) {
-   case GL_FOG_HINT:
-      break;
-   default:
-      break;
-   }
-}
-
-/* =============================================================
- */
-
-static void
-i915Enable(GLcontext * ctx, GLenum cap, GLboolean state)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   switch (cap) {
-   case GL_TEXTURE_2D:
-      break;
-
-   case GL_LIGHTING:
-   case GL_COLOR_SUM:
-      update_specular(ctx);
-      break;
-
-   case GL_ALPHA_TEST:
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      if (state)
-         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_ALPHA_TEST_ENABLE;
-      else
-         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_ALPHA_TEST_ENABLE;
-      break;
-
-   case GL_BLEND:
-      i915EvalLogicOpBlendState(ctx);
-      break;
-
-   case GL_COLOR_LOGIC_OP:
-      i915EvalLogicOpBlendState(ctx);
-
-      /* Logicop doesn't seem to work at 16bpp:
-       */
-      if (i915->intel.intelScreen->cpp == 2)    /* XXX FBO fix */
-         FALLBACK(&i915->intel, I915_FALLBACK_LOGICOP, state);
-      break;
-
-   case GL_FRAGMENT_PROGRAM_ARB:
-      break;
-
-   case GL_DITHER:
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      if (state)
-         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
-      else
-         i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_COLOR_DITHER_ENABLE;
-      break;
-
-   case GL_DEPTH_TEST:
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      if (state)
-         i915->state.Ctx[I915_CTXREG_LIS6] |= S6_DEPTH_TEST_ENABLE;
-      else
-         i915->state.Ctx[I915_CTXREG_LIS6] &= ~S6_DEPTH_TEST_ENABLE;
-
-      i915DepthMask(ctx, ctx->Depth.Mask);
-      break;
-
-   case GL_SCISSOR_TEST:
-      I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
-      if (state)
-         i915->state.Buffer[I915_DESTREG_SENABLE] =
-            (_3DSTATE_SCISSOR_ENABLE_CMD | ENABLE_SCISSOR_RECT);
-      else
-         i915->state.Buffer[I915_DESTREG_SENABLE] =
-            (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
-      break;
-
-   case GL_LINE_SMOOTH:
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      if (state)
-         i915->state.Ctx[I915_CTXREG_LIS4] |= S4_LINE_ANTIALIAS_ENABLE;
-      else
-         i915->state.Ctx[I915_CTXREG_LIS4] &= ~S4_LINE_ANTIALIAS_ENABLE;
-      break;
-
-   case GL_FOG:
-      break;
-
-   case GL_CULL_FACE:
-      i915CullFaceFrontFace(ctx, 0);
-      break;
-
-   case GL_STENCIL_TEST:
-      {
-         GLboolean hw_stencil = GL_FALSE;
-         if (ctx->DrawBuffer) {
-            struct intel_renderbuffer *irbStencil
-               = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_STENCIL);
-            hw_stencil = (irbStencil && irbStencil->region);
-         }
-         if (hw_stencil) {
-            I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-            if (state)
-               i915->state.Ctx[I915_CTXREG_LIS5] |= (S5_STENCIL_TEST_ENABLE |
-                                                     S5_STENCIL_WRITE_ENABLE);
-            else
-               i915->state.Ctx[I915_CTXREG_LIS5] &= ~(S5_STENCIL_TEST_ENABLE |
-                                                      S5_STENCIL_WRITE_ENABLE);
-         }
-         else {
-            FALLBACK(&i915->intel, I915_FALLBACK_STENCIL, state);
-         }
-      }
-      break;
-
-   case GL_POLYGON_STIPPLE:
-      /* The stipple command worked on my 855GM box, but not my 845G.
-       * I'll do more testing later to find out exactly which hardware
-       * supports it.  Disabled for now.
-       */
-      if (i915->intel.hw_stipple &&
-          i915->intel.reduced_primitive == GL_TRIANGLES) {
-         I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-         if (state)
-            i915->state.Stipple[I915_STPREG_ST1] |= ST1_ENABLE;
-         else
-            i915->state.Stipple[I915_STPREG_ST1] &= ~ST1_ENABLE;
-      }
-      break;
-
-   case GL_POLYGON_SMOOTH:
-      break;
-
-   case GL_POINT_SMOOTH:
-      break;
-
-   default:
-      ;
-   }
-}
-
-
-static void
-i915_init_packets(struct i915_context *i915)
-{
-   intelScreenPrivate *screen = i915->intel.intelScreen;
-
-   /* Zero all state */
-   memset(&i915->state, 0, sizeof(i915->state));
-
-
-   {
-      I915_STATECHANGE(i915, I915_UPLOAD_CTX);
-      /* Probably don't want to upload all this stuff every time one 
-       * piece changes.
-       */
-      i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
-                                         I1_LOAD_S(2) |
-                                         I1_LOAD_S(4) |
-                                         I1_LOAD_S(5) | I1_LOAD_S(6) | (3));
-      i915->state.Ctx[I915_CTXREG_LIS2] = 0;
-      i915->state.Ctx[I915_CTXREG_LIS4] = 0;
-      i915->state.Ctx[I915_CTXREG_LIS5] = 0;
-
-      if (screen->cpp == 2)     /* XXX FBO fix */
-         i915->state.Ctx[I915_CTXREG_LIS5] |= S5_COLOR_DITHER_ENABLE;
-
-
-      i915->state.Ctx[I915_CTXREG_LIS6] = (S6_COLOR_WRITE_ENABLE |
-                                           (2 << S6_TRISTRIP_PV_SHIFT));
-
-      i915->state.Ctx[I915_CTXREG_STATE4] = (_3DSTATE_MODES_4_CMD |
-                                             ENABLE_LOGIC_OP_FUNC |
-                                             LOGIC_OP_FUNC(LOGICOP_COPY) |
-                                             ENABLE_STENCIL_TEST_MASK |
-                                             STENCIL_TEST_MASK(0xff) |
-                                             ENABLE_STENCIL_WRITE_MASK |
-                                             STENCIL_WRITE_MASK(0xff));
-
-      i915->state.Ctx[I915_CTXREG_IAB] =
-         (_3DSTATE_INDEPENDENT_ALPHA_BLEND_CMD | IAB_MODIFY_ENABLE |
-          IAB_MODIFY_FUNC | IAB_MODIFY_SRC_FACTOR | IAB_MODIFY_DST_FACTOR);
-
-      i915->state.Ctx[I915_CTXREG_BLENDCOLOR0] =
-         _3DSTATE_CONST_BLEND_COLOR_CMD;
-      i915->state.Ctx[I915_CTXREG_BLENDCOLOR1] = 0;
-
-   }
-
-   {
-      I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-      i915->state.Stipple[I915_STPREG_ST0] = _3DSTATE_STIPPLE;
-   }
-
-
-   {
-      I915_STATECHANGE(i915, I915_UPLOAD_FOG);
-      i915->state.Fog[I915_FOGREG_MODE0] = _3DSTATE_FOG_MODE_CMD;
-      i915->state.Fog[I915_FOGREG_MODE1] = (FMC1_FOGFUNC_MODIFY_ENABLE |
-                                            FMC1_FOGFUNC_VERTEX |
-                                            FMC1_FOGINDEX_MODIFY_ENABLE |
-                                            FMC1_FOGINDEX_W |
-                                            FMC1_C1_C2_MODIFY_ENABLE |
-                                            FMC1_DENSITY_MODIFY_ENABLE);
-      i915->state.Fog[I915_FOGREG_COLOR] = _3DSTATE_FOG_COLOR_CMD;
-   }
-
-
-   {
-      I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
-      /* color buffer offset/stride */
-      i915->state.Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
-      i915->state.Buffer[I915_DESTREG_CBUFADDR1] = (BUF_3D_ID_COLOR_BACK | BUF_3D_PITCH(screen->front.pitch) |  /* pitch in bytes */
-                                                    BUF_3D_USE_FENCE);
-
-      i915->state.Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
-      i915->state.Buffer[I915_DESTREG_DBUFADDR1] = (BUF_3D_ID_DEPTH | BUF_3D_PITCH(screen->depth.pitch) |       /* pitch in bytes */
-                                                    BUF_3D_USE_FENCE);
-
-      i915->state.Buffer[I915_DESTREG_DV0] = _3DSTATE_DST_BUF_VARS_CMD;
-
-      /* XXX FBO: remove this?  Also get set in i915_set_draw_region() */
-#if 0                           /* seems we don't need this */
-      switch (screen->fbFormat) {
-      case DV_PF_565:
-         i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |        /* .5 */
-                                                 DSTORG_VERT_BIAS(0x8) |        /* .5 */
-                                                 LOD_PRECLAMP_OGL |
-                                                 TEX_DEFAULT_COLOR_OGL |
-                                                 DITHER_FULL_ALWAYS |
-                                                 screen->fbFormat |
-                                                 DEPTH_FRMT_16_FIXED);
-         break;
-      case DV_PF_8888:
-         i915->state.Buffer[I915_DESTREG_DV1] = (DSTORG_HORT_BIAS(0x8) |        /* .5 */
-                                                 DSTORG_VERT_BIAS(0x8) |        /* .5 */
-                                                 LOD_PRECLAMP_OGL |
-                                                 TEX_DEFAULT_COLOR_OGL |
-                                                 screen->fbFormat |
-                                                 DEPTH_FRMT_24_FIXED_8_OTHER);
-         break;
-      }
-#endif
-
-
-      /* scissor */
-      i915->state.Buffer[I915_DESTREG_SENABLE] =
-         (_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
-      i915->state.Buffer[I915_DESTREG_SR0] = _3DSTATE_SCISSOR_RECT_0_CMD;
-      i915->state.Buffer[I915_DESTREG_SR1] = 0;
-      i915->state.Buffer[I915_DESTREG_SR2] = 0;
-   }
-
-
-#if 0
-   {
-      I915_STATECHANGE(i915, I915_UPLOAD_DEFAULTS);
-      i915->state.Default[I915_DEFREG_C0] = _3DSTATE_DEFAULT_DIFFUSE;
-      i915->state.Default[I915_DEFREG_C1] = 0;
-      i915->state.Default[I915_DEFREG_S0] = _3DSTATE_DEFAULT_SPECULAR;
-      i915->state.Default[I915_DEFREG_S1] = 0;
-      i915->state.Default[I915_DEFREG_Z0] = _3DSTATE_DEFAULT_Z;
-      i915->state.Default[I915_DEFREG_Z1] = 0;
-   }
-#endif
-
-
-   /* These will be emitted every at the head of every buffer, unless
-    * we get hardware contexts working.
-    */
-   i915->state.active = (I915_UPLOAD_PROGRAM |
-                         I915_UPLOAD_STIPPLE |
-                         I915_UPLOAD_CTX |
-                         I915_UPLOAD_BUFFERS | I915_UPLOAD_INVARIENT);
-}
-
-void
-i915InitStateFunctions(struct dd_function_table *functions)
-{
-   functions->AlphaFunc = i915AlphaFunc;
-   functions->BlendColor = i915BlendColor;
-   functions->BlendEquationSeparate = i915BlendEquationSeparate;
-   functions->BlendFuncSeparate = i915BlendFuncSeparate;
-   functions->ColorMask = i915ColorMask;
-   functions->CullFace = i915CullFaceFrontFace;
-   functions->DepthFunc = i915DepthFunc;
-   functions->DepthMask = i915DepthMask;
-   functions->Enable = i915Enable;
-   functions->Fogfv = i915Fogfv;
-   functions->FrontFace = i915CullFaceFrontFace;
-   functions->Hint = i915Hint;
-   functions->LightModelfv = i915LightModelfv;
-   functions->LineWidth = i915LineWidth;
-   functions->LogicOpcode = i915LogicOp;
-   functions->PointSize = i915PointSize;
-   functions->PolygonStipple = i915PolygonStipple;
-   functions->Scissor = i915Scissor;
-   functions->ShadeModel = i915ShadeModel;
-   functions->StencilFuncSeparate = i915StencilFuncSeparate;
-   functions->StencilMaskSeparate = i915StencilMaskSeparate;
-   functions->StencilOpSeparate = i915StencilOpSeparate;
-}
-
-
-void
-i915InitState(struct i915_context *i915)
-{
-   GLcontext *ctx = &i915->intel.ctx;
-
-   i915_init_packets(i915);
-
-   _mesa_init_driver_state(ctx);
-
-   memcpy(&i915->initial, &i915->state, sizeof(i915->state));
-   i915->current = &i915->state;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_tex.c b/src/mesa/drivers/dri/i915tex/i915_tex.c
deleted file mode 100644
index 59e148ca04e..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_tex.c
+++ /dev/null
@@ -1,78 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "simple_list.h"
-#include "enums.h"
-#include "image.h"
-#include "texstore.h"
-#include "texformat.h"
-#include "texmem.h"
-#include "swrast/swrast.h"
-
-#include "mm.h"
-
-#include "intel_ioctl.h"
-
-#include "i915_context.h"
-#include "i915_reg.h"
-
-
-
-static void
-i915TexEnv(GLcontext * ctx, GLenum target,
-           GLenum pname, const GLfloat * param)
-{
-   struct i915_context *i915 = I915_CONTEXT(ctx);
-
-   switch (pname) {
-   case GL_TEXTURE_LOD_BIAS:{
-         GLuint unit = ctx->Texture.CurrentUnit;
-         GLint b = (int) ((*param) * 16.0);
-         if (b > 255)
-            b = 255;
-         if (b < -256)
-            b = -256;
-         I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-         i915->lodbias_ss2[unit] =
-            ((b << SS2_LOD_BIAS_SHIFT) & SS2_LOD_BIAS_MASK);
-         break;
-      }
-
-   default:
-      break;
-   }
-}
-
-
-void
-i915InitTextureFuncs(struct dd_function_table *functions)
-{
-   functions->TexEnv = i915TexEnv;
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_texstate.c b/src/mesa/drivers/dri/i915tex/i915_texstate.c
deleted file mode 100644
index 3d68187cf88..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_texstate.c
+++ /dev/null
@@ -1,359 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "mtypes.h"
-#include "enums.h"
-#include "texformat.h"
-#include "dri_bufmgr.h"
-
-#include "intel_mipmap_tree.h"
-#include "intel_tex.h"
-
-#include "i915_context.h"
-#include "i915_reg.h"
-
-
-static GLuint
-translate_texture_format(GLuint mesa_format)
-{
-   switch (mesa_format) {
-   case MESA_FORMAT_L8:
-      return MAPSURF_8BIT | MT_8BIT_L8;
-   case MESA_FORMAT_I8:
-      return MAPSURF_8BIT | MT_8BIT_I8;
-   case MESA_FORMAT_A8:
-      return MAPSURF_8BIT | MT_8BIT_A8;
-   case MESA_FORMAT_AL88:
-      return MAPSURF_16BIT | MT_16BIT_AY88;
-   case MESA_FORMAT_RGB565:
-      return MAPSURF_16BIT | MT_16BIT_RGB565;
-   case MESA_FORMAT_ARGB1555:
-      return MAPSURF_16BIT | MT_16BIT_ARGB1555;
-   case MESA_FORMAT_ARGB4444:
-      return MAPSURF_16BIT | MT_16BIT_ARGB4444;
-   case MESA_FORMAT_ARGB8888:
-      return MAPSURF_32BIT | MT_32BIT_ARGB8888;
-   case MESA_FORMAT_YCBCR_REV:
-      return (MAPSURF_422 | MT_422_YCRCB_NORMAL);
-   case MESA_FORMAT_YCBCR:
-      return (MAPSURF_422 | MT_422_YCRCB_SWAPY);
-   case MESA_FORMAT_RGB_FXT1:
-   case MESA_FORMAT_RGBA_FXT1:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_FXT1);
-   case MESA_FORMAT_Z16:
-      return (MAPSURF_16BIT | MT_16BIT_L16);
-   case MESA_FORMAT_RGBA_DXT1:
-   case MESA_FORMAT_RGB_DXT1:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT1);
-   case MESA_FORMAT_RGBA_DXT3:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT2_3);
-   case MESA_FORMAT_RGBA_DXT5:
-      return (MAPSURF_COMPRESSED | MT_COMPRESS_DXT4_5);
-   case MESA_FORMAT_Z24_S8:
-      return (MAPSURF_32BIT | MT_32BIT_xL824);
-   default:
-      fprintf(stderr, "%s: bad image format %x\n", __FUNCTION__, mesa_format);
-      abort();
-      return 0;
-   }
-}
-
-
-
-
-/* The i915 (and related graphics cores) do not support GL_CLAMP.  The
- * Intel drivers for "other operating systems" implement GL_CLAMP as
- * GL_CLAMP_TO_EDGE, so the same is done here.
- */
-static GLuint
-translate_wrap_mode(GLenum wrap)
-{
-   switch (wrap) {
-   case GL_REPEAT:
-      return TEXCOORDMODE_WRAP;
-   case GL_CLAMP:
-      return TEXCOORDMODE_CLAMP_EDGE;   /* not quite correct */
-   case GL_CLAMP_TO_EDGE:
-      return TEXCOORDMODE_CLAMP_EDGE;
-   case GL_CLAMP_TO_BORDER:
-      return TEXCOORDMODE_CLAMP_BORDER;
-   case GL_MIRRORED_REPEAT:
-      return TEXCOORDMODE_MIRROR;
-   default:
-      return TEXCOORDMODE_WRAP;
-   }
-}
-
-
-
-/* Recalculate all state from scratch.  Perhaps not the most
- * efficient, but this has gotten complex enough that we need
- * something which is understandable and reliable.
- */
-static GLboolean
-i915_update_tex_unit(struct intel_context *intel, GLuint unit, GLuint ss3)
-{
-   GLcontext *ctx = &intel->ctx;
-   struct i915_context *i915 = i915_context(ctx);
-   struct gl_texture_object *tObj = ctx->Texture.Unit[unit]._Current;
-   struct intel_texture_object *intelObj = intel_texture_object(tObj);
-   struct gl_texture_image *firstImage;
-   GLuint *state = i915->state.Tex[unit], format, pitch;
-
-   memset(state, 0, sizeof(state));
-
-   /*We need to refcount these. */
-
-   if (i915->state.tex_buffer[unit] != NULL) {
-       driBOUnReference(i915->state.tex_buffer[unit]);
-       i915->state.tex_buffer[unit] = NULL;
-   }
-
-   if (!intelObj->imageOverride && !intel_finalize_mipmap_tree(intel, unit))
-      return GL_FALSE;
-
-   /* Get first image here, since intelObj->firstLevel will get set in
-    * the intel_finalize_mipmap_tree() call above.
-    */
-   firstImage = tObj->Image[0][intelObj->firstLevel];
-
-   if (intelObj->imageOverride) {
-      i915->state.tex_buffer[unit] = NULL;
-      i915->state.tex_offset[unit] = intelObj->textureOffset;
-
-      switch (intelObj->depthOverride) {
-      case 32:
-	 format = MAPSURF_32BIT | MT_32BIT_ARGB8888;
-	 break;
-      case 24:
-      default:
-	 format = MAPSURF_32BIT | MT_32BIT_XRGB8888;
-	 break;
-      case 16:
-	 format = MAPSURF_16BIT | MT_16BIT_RGB565;
-	 break;
-      }
-
-      pitch = intelObj->pitchOverride;
-   } else {
-      i915->state.tex_buffer[unit] = driBOReference(intelObj->mt->region->
-						    buffer);
-      i915->state.tex_offset[unit] =  intel_miptree_image_offset(intelObj->mt,
-								 0, intelObj->
-								 firstLevel);
-
-      format = translate_texture_format(firstImage->TexFormat->MesaFormat);
-      pitch = intelObj->mt->pitch * intelObj->mt->cpp;
-   }
-
-   state[I915_TEXREG_MS3] =
-      (((firstImage->Height - 1) << MS3_HEIGHT_SHIFT) |
-       ((firstImage->Width - 1) << MS3_WIDTH_SHIFT) | format |
-       MS3_USE_FENCE_REGS);
-
-   state[I915_TEXREG_MS4] =
-     ((((pitch / 4) - 1) << MS4_PITCH_SHIFT) | MS4_CUBE_FACE_ENA_MASK |
-       ((((intelObj->lastLevel - intelObj->firstLevel) * 4)) <<
-	MS4_MAX_LOD_SHIFT) | ((firstImage->Depth - 1) <<
-			      MS4_VOLUME_DEPTH_SHIFT));
-
-
-   {
-      GLuint minFilt, mipFilt, magFilt;
-
-      switch (tObj->MinFilter) {
-      case GL_NEAREST:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_NONE;
-         break;
-      case GL_LINEAR:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_NONE;
-         break;
-      case GL_NEAREST_MIPMAP_NEAREST:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_NEAREST;
-         break;
-      case GL_LINEAR_MIPMAP_NEAREST:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_NEAREST;
-         break;
-      case GL_NEAREST_MIPMAP_LINEAR:
-         minFilt = FILTER_NEAREST;
-         mipFilt = MIPFILTER_LINEAR;
-         break;
-      case GL_LINEAR_MIPMAP_LINEAR:
-         minFilt = FILTER_LINEAR;
-         mipFilt = MIPFILTER_LINEAR;
-         break;
-      default:
-         return GL_FALSE;
-      }
-
-      if (tObj->MaxAnisotropy > 1.0) {
-         minFilt = FILTER_ANISOTROPIC;
-         magFilt = FILTER_ANISOTROPIC;
-      }
-      else {
-         switch (tObj->MagFilter) {
-         case GL_NEAREST:
-            magFilt = FILTER_NEAREST;
-            break;
-         case GL_LINEAR:
-            magFilt = FILTER_LINEAR;
-            break;
-         default:
-            return GL_FALSE;
-         }
-      }
-
-      state[I915_TEXREG_SS2] = i915->lodbias_ss2[unit];
-
-      /* YUV conversion:
-       */
-      if (firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR ||
-          firstImage->TexFormat->MesaFormat == MESA_FORMAT_YCBCR_REV)
-         state[I915_TEXREG_SS2] |= SS2_COLORSPACE_CONVERSION;
-
-      /* Shadow:
-       */
-      if (tObj->CompareMode == GL_COMPARE_R_TO_TEXTURE_ARB &&
-          tObj->Target != GL_TEXTURE_3D) {
-
-         state[I915_TEXREG_SS2] |=
-            (SS2_SHADOW_ENABLE |
-             intel_translate_compare_func(tObj->CompareFunc));
-
-         minFilt = FILTER_4X4_FLAT;
-         magFilt = FILTER_4X4_FLAT;
-      }
-
-      state[I915_TEXREG_SS2] |= ((minFilt << SS2_MIN_FILTER_SHIFT) |
-                                 (mipFilt << SS2_MIP_FILTER_SHIFT) |
-                                 (magFilt << SS2_MAG_FILTER_SHIFT));
-   }
-
-   {
-      GLenum ws = tObj->WrapS;
-      GLenum wt = tObj->WrapT;
-      GLenum wr = tObj->WrapR;
-
-
-      /* 3D textures don't seem to respect the border color.
-       * Fallback if there's ever a danger that they might refer to
-       * it.  
-       * 
-       * Effectively this means fallback on 3D clamp or
-       * clamp_to_border.
-       */
-      if (tObj->Target == GL_TEXTURE_3D &&
-          (tObj->MinFilter != GL_NEAREST ||
-           tObj->MagFilter != GL_NEAREST) &&
-          (ws == GL_CLAMP ||
-           wt == GL_CLAMP ||
-           wr == GL_CLAMP ||
-           ws == GL_CLAMP_TO_BORDER ||
-           wt == GL_CLAMP_TO_BORDER || wr == GL_CLAMP_TO_BORDER))
-         return GL_FALSE;
-
-
-      state[I915_TEXREG_SS3] = ss3;     /* SS3_NORMALIZED_COORDS */
-
-      state[I915_TEXREG_SS3] |=
-         ((translate_wrap_mode(ws) << SS3_TCX_ADDR_MODE_SHIFT) |
-          (translate_wrap_mode(wt) << SS3_TCY_ADDR_MODE_SHIFT) |
-          (translate_wrap_mode(wr) << SS3_TCZ_ADDR_MODE_SHIFT));
-
-      state[I915_TEXREG_SS3] |= (unit << SS3_TEXTUREMAP_INDEX_SHIFT);
-   }
-
-
-   state[I915_TEXREG_SS4] = INTEL_PACKCOLOR8888(tObj->_BorderChan[0],
-                                                tObj->_BorderChan[1],
-                                                tObj->_BorderChan[2],
-                                                tObj->_BorderChan[3]);
-
-
-   I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(unit), GL_TRUE);
-   /* memcmp was already disabled, but definitely won't work as the
-    * region might now change and that wouldn't be detected:
-    */
-   I915_STATECHANGE(i915, I915_UPLOAD_TEX(unit));
-
-
-#if 0
-   DBG(TEXTURE, "state[I915_TEXREG_SS2] = 0x%x\n", state[I915_TEXREG_SS2]);
-   DBG(TEXTURE, "state[I915_TEXREG_SS3] = 0x%x\n", state[I915_TEXREG_SS3]);
-   DBG(TEXTURE, "state[I915_TEXREG_SS4] = 0x%x\n", state[I915_TEXREG_SS4]);
-   DBG(TEXTURE, "state[I915_TEXREG_MS2] = 0x%x\n", state[I915_TEXREG_MS2]);
-   DBG(TEXTURE, "state[I915_TEXREG_MS3] = 0x%x\n", state[I915_TEXREG_MS3]);
-   DBG(TEXTURE, "state[I915_TEXREG_MS4] = 0x%x\n", state[I915_TEXREG_MS4]);
-#endif
-
-   return GL_TRUE;
-}
-
-
-
-
-void
-i915UpdateTextureState(struct intel_context *intel)
-{
-   GLboolean ok = GL_TRUE;
-   GLuint i;
-
-   for (i = 0; i < I915_TEX_UNITS && ok; i++) {
-      switch (intel->ctx.Texture.Unit[i]._ReallyEnabled) {
-      case TEXTURE_1D_BIT:
-      case TEXTURE_2D_BIT:
-      case TEXTURE_CUBE_BIT:
-      case TEXTURE_3D_BIT:
-         ok = i915_update_tex_unit(intel, i, SS3_NORMALIZED_COORDS);
-         break;
-      case TEXTURE_RECT_BIT:
-         ok = i915_update_tex_unit(intel, i, 0);
-         break;
-      case 0:{
-            struct i915_context *i915 = i915_context(&intel->ctx);
-            if (i915->state.active & I915_UPLOAD_TEX(i))
-               I915_ACTIVESTATE(i915, I915_UPLOAD_TEX(i), GL_FALSE);
-
-	    if (i915->state.tex_buffer[i] != NULL) {
-	       driBOUnReference(i915->state.tex_buffer[i]);
-	       i915->state.tex_buffer[i] = NULL;
-	    }
-
-            break;
-         }
-      default:
-         ok = GL_FALSE;
-         break;
-      }
-   }
-
-   FALLBACK(intel, I915_FALLBACK_TEXTURE, !ok);
-}
diff --git a/src/mesa/drivers/dri/i915tex/i915_vtbl.c b/src/mesa/drivers/dri/i915tex/i915_vtbl.c
deleted file mode 100644
index ad333b490b7..00000000000
--- a/src/mesa/drivers/dri/i915tex/i915_vtbl.c
+++ /dev/null
@@ -1,553 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-
-#include "glheader.h"
-#include "mtypes.h"
-#include "imports.h"
-#include "macros.h"
-#include "colormac.h"
-
-#include "tnl/t_context.h"
-#include "tnl/t_vertex.h"
-
-#include "intel_batchbuffer.h"
-#include "intel_tex.h"
-#include "intel_regions.h"
-
-#include "i915_reg.h"
-#include "i915_context.h"
-
-static void
-i915_render_start(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-
-   i915ValidateFragmentProgram(i915);
-}
-
-
-static void
-i915_reduced_primitive_state(struct intel_context *intel, GLenum rprim)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   GLuint st1 = i915->state.Stipple[I915_STPREG_ST1];
-
-   st1 &= ~ST1_ENABLE;
-
-   switch (rprim) {
-   case GL_QUADS: /* from RASTERIZE(GL_QUADS) in t_dd_tritemp.h */
-   case GL_TRIANGLES:
-      if (intel->ctx.Polygon.StippleFlag && intel->hw_stipple)
-         st1 |= ST1_ENABLE;
-      break;
-   case GL_LINES:
-   case GL_POINTS:
-   default:
-      break;
-   }
-
-   i915->intel.reduced_primitive = rprim;
-
-   if (st1 != i915->state.Stipple[I915_STPREG_ST1]) {
-      INTEL_FIREVERTICES(intel);
-
-      I915_STATECHANGE(i915, I915_UPLOAD_STIPPLE);
-      i915->state.Stipple[I915_STPREG_ST1] = st1;
-   }
-}
-
-
-/* Pull apart the vertex format registers and figure out how large a
- * vertex is supposed to be. 
- */
-static GLboolean
-i915_check_vertex_size(struct intel_context *intel, GLuint expected)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   int lis2 = i915->current->Ctx[I915_CTXREG_LIS2];
-   int lis4 = i915->current->Ctx[I915_CTXREG_LIS4];
-   int i, sz = 0;
-
-   switch (lis4 & S4_VFMT_XYZW_MASK) {
-   case S4_VFMT_XY:
-      sz = 2;
-      break;
-   case S4_VFMT_XYZ:
-      sz = 3;
-      break;
-   case S4_VFMT_XYW:
-      sz = 3;
-      break;
-   case S4_VFMT_XYZW:
-      sz = 4;
-      break;
-   default:
-      fprintf(stderr, "no xyzw specified\n");
-      return 0;
-   }
-
-   if (lis4 & S4_VFMT_SPEC_FOG)
-      sz++;
-   if (lis4 & S4_VFMT_COLOR)
-      sz++;
-   if (lis4 & S4_VFMT_DEPTH_OFFSET)
-      sz++;
-   if (lis4 & S4_VFMT_POINT_WIDTH)
-      sz++;
-   if (lis4 & S4_VFMT_FOG_PARAM)
-      sz++;
-
-   for (i = 0; i < 8; i++) {
-      switch (lis2 & S2_TEXCOORD_FMT0_MASK) {
-      case TEXCOORDFMT_2D:
-         sz += 2;
-         break;
-      case TEXCOORDFMT_3D:
-         sz += 3;
-         break;
-      case TEXCOORDFMT_4D:
-         sz += 4;
-         break;
-      case TEXCOORDFMT_1D:
-         sz += 1;
-         break;
-      case TEXCOORDFMT_2D_16:
-         sz += 1;
-         break;
-      case TEXCOORDFMT_4D_16:
-         sz += 2;
-         break;
-      case TEXCOORDFMT_NOT_PRESENT:
-         break;
-      default:
-         fprintf(stderr, "bad texcoord fmt %d\n", i);
-         return GL_FALSE;
-      }
-      lis2 >>= S2_TEXCOORD_FMT1_SHIFT;
-   }
-
-   if (sz != expected)
-      fprintf(stderr, "vertex size mismatch %d/%d\n", sz, expected);
-
-   return sz == expected;
-}
-
-
-static void
-i915_emit_invarient_state(struct intel_context *intel)
-{
-   BATCH_LOCALS;
-
-   BEGIN_BATCH(200, 0);
-
-   OUT_BATCH(_3DSTATE_AA_CMD |
-             AA_LINE_ECAAR_WIDTH_ENABLE |
-             AA_LINE_ECAAR_WIDTH_1_0 |
-             AA_LINE_REGION_WIDTH_ENABLE | AA_LINE_REGION_WIDTH_1_0);
-
-   OUT_BATCH(_3DSTATE_DFLT_DIFFUSE_CMD);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_DFLT_SPEC_CMD);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_DFLT_Z_CMD);
-   OUT_BATCH(0);
-
-   /* Don't support texture crossbar yet */
-   OUT_BATCH(_3DSTATE_COORD_SET_BINDINGS |
-             CSB_TCB(0, 0) |
-             CSB_TCB(1, 1) |
-             CSB_TCB(2, 2) |
-             CSB_TCB(3, 3) |
-             CSB_TCB(4, 4) | CSB_TCB(5, 5) | CSB_TCB(6, 6) | CSB_TCB(7, 7));
-
-   OUT_BATCH(_3DSTATE_RASTER_RULES_CMD |
-             ENABLE_POINT_RASTER_RULE |
-             OGL_POINT_RASTER_RULE |
-             ENABLE_LINE_STRIP_PROVOKE_VRTX |
-             ENABLE_TRI_FAN_PROVOKE_VRTX |
-             LINE_STRIP_PROVOKE_VRTX(1) |
-             TRI_FAN_PROVOKE_VRTX(2) | ENABLE_TEXKILL_3D_4D | TEXKILL_4D);
-
-   /* Need to initialize this to zero.
-    */
-   OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 | I1_LOAD_S(3) | (1));
-   OUT_BATCH(0);
-
-   /* XXX: Use this */
-   OUT_BATCH(_3DSTATE_SCISSOR_ENABLE_CMD | DISABLE_SCISSOR_RECT);
-
-   OUT_BATCH(_3DSTATE_SCISSOR_RECT_0_CMD);
-   OUT_BATCH(0);
-   OUT_BATCH(0);
-
-   OUT_BATCH(_3DSTATE_DEPTH_SUBRECT_DISABLE);
-
-   OUT_BATCH(_3DSTATE_LOAD_INDIRECT | 0);       /* disable indirect state */
-   OUT_BATCH(0);
-
-
-   /* Don't support twosided stencil yet */
-   OUT_BATCH(_3DSTATE_BACKFACE_STENCIL_OPS | BFO_ENABLE_STENCIL_TWO_SIDE | 0);
-
-   ADVANCE_BATCH();
-}
-
-
-#define emit(intel, state, size )		     \
-   intel_batchbuffer_data(intel->batch, state, size, 0 )
-
-static GLuint
-get_dirty(struct i915_hw_state *state)
-{
-   GLuint dirty;
-
-   /* Workaround the multitex hang - if one texture unit state is
-    * modified, emit all texture units.
-    */
-   dirty = state->active & ~state->emitted;
-   if (dirty & I915_UPLOAD_TEX_ALL)
-      state->emitted &= ~I915_UPLOAD_TEX_ALL;
-   dirty = state->active & ~state->emitted;
-   return dirty;
-}
-
-
-static GLuint
-get_state_size(struct i915_hw_state *state)
-{
-   GLuint dirty = get_dirty(state);
-   GLuint i;
-   GLuint sz = 0;
-
-   if (dirty & I915_UPLOAD_CTX)
-      sz += sizeof(state->Ctx);
-
-   if (dirty & I915_UPLOAD_BUFFERS)
-      sz += sizeof(state->Buffer);
-
-   if (dirty & I915_UPLOAD_STIPPLE)
-      sz += sizeof(state->Stipple);
-
-   if (dirty & I915_UPLOAD_FOG)
-      sz += sizeof(state->Fog);
-
-   if (dirty & I915_UPLOAD_TEX_ALL) {
-      int nr = 0;
-      for (i = 0; i < I915_TEX_UNITS; i++)
-         if (dirty & I915_UPLOAD_TEX(i))
-            nr++;
-
-      sz += (2 + nr * 3) * sizeof(GLuint) * 2;
-   }
-
-   if (dirty & I915_UPLOAD_CONSTANTS)
-      sz += state->ConstantSize * sizeof(GLuint);
-
-   if (dirty & I915_UPLOAD_PROGRAM)
-      sz += state->ProgramSize * sizeof(GLuint);
-
-   return sz;
-}
-
-
-/* Push the state into the sarea and/or texture memory.
- */
-static void
-i915_emit_state(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   struct i915_hw_state *state = i915->current;
-   int i;
-   GLuint dirty;
-   BATCH_LOCALS;
-
-   /* We don't hold the lock at this point, so want to make sure that
-    * there won't be a buffer wrap.  
-    *
-    * It might be better to talk about explicit places where
-    * scheduling is allowed, rather than assume that it is whenever a
-    * batchbuffer fills up.
-    */
-   intel_batchbuffer_require_space(intel->batch, get_state_size(state), 0);
-
-   /* Do this here as we may have flushed the batchbuffer above,
-    * causing more state to be dirty!
-    */
-   dirty = get_dirty(state);
-
-   if (INTEL_DEBUG & DEBUG_STATE)
-      fprintf(stderr, "%s dirty: %x\n", __FUNCTION__, dirty);
-
-   if (dirty & I915_UPLOAD_INVARIENT) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_INVARIENT:\n");
-      i915_emit_invarient_state(intel);
-   }
-
-   if (dirty & I915_UPLOAD_CTX) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_CTX:\n");
-
-      emit(intel, state->Ctx, sizeof(state->Ctx));
-   }
-
-   if (dirty & I915_UPLOAD_BUFFERS) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_BUFFERS:\n");
-      BEGIN_BATCH(I915_DEST_SETUP_SIZE + 2, 0);
-      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR0]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_CBUFADDR1]);
-      OUT_RELOC(state->draw_region->buffer,
-                DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
-                state->draw_region->draw_offset);
-
-      if (state->depth_region) {
-         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR0]);
-         OUT_BATCH(state->Buffer[I915_DESTREG_DBUFADDR1]);
-         OUT_RELOC(state->depth_region->buffer,
-                   DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_WRITE,
-                   DRM_BO_MASK_MEM | DRM_BO_FLAG_WRITE,
-                   state->depth_region->draw_offset);
-      }
-
-      OUT_BATCH(state->Buffer[I915_DESTREG_DV0]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_DV1]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_SENABLE]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_SR0]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_SR1]);
-      OUT_BATCH(state->Buffer[I915_DESTREG_SR2]);
-      ADVANCE_BATCH();
-   }
-
-   if (dirty & I915_UPLOAD_STIPPLE) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_STIPPLE:\n");
-      emit(intel, state->Stipple, sizeof(state->Stipple));
-   }
-
-   if (dirty & I915_UPLOAD_FOG) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_FOG:\n");
-      emit(intel, state->Fog, sizeof(state->Fog));
-   }
-
-   /* Combine all the dirty texture state into a single command to
-    * avoid lockups on I915 hardware. 
-    */
-   if (dirty & I915_UPLOAD_TEX_ALL) {
-      int nr = 0;
-
-      for (i = 0; i < I915_TEX_UNITS; i++)
-         if (dirty & I915_UPLOAD_TEX(i))
-            nr++;
-
-      BEGIN_BATCH(2 + nr * 3, 0);
-      OUT_BATCH(_3DSTATE_MAP_STATE | (3 * nr));
-      OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0; i < I915_TEX_UNITS; i++)
-         if (dirty & I915_UPLOAD_TEX(i)) {
-
-            if (state->tex_buffer[i]) {
-               OUT_RELOC(state->tex_buffer[i],
-                         DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
-                         DRM_BO_MASK_MEM | DRM_BO_FLAG_READ,
-                         state->tex_offset[i]);
-            }
-            else if (state == &i915->meta) {
-               assert(i == 0);
-               OUT_BATCH(0);
-            }
-            else {
-               OUT_BATCH(state->tex_offset[i]);
-            }
-
-            OUT_BATCH(state->Tex[i][I915_TEXREG_MS3]);
-            OUT_BATCH(state->Tex[i][I915_TEXREG_MS4]);
-         }
-      ADVANCE_BATCH();
-
-      BEGIN_BATCH(2 + nr * 3, 0);
-      OUT_BATCH(_3DSTATE_SAMPLER_STATE | (3 * nr));
-      OUT_BATCH((dirty & I915_UPLOAD_TEX_ALL) >> I915_UPLOAD_TEX_0_SHIFT);
-      for (i = 0; i < I915_TEX_UNITS; i++)
-         if (dirty & I915_UPLOAD_TEX(i)) {
-            OUT_BATCH(state->Tex[i][I915_TEXREG_SS2]);
-            OUT_BATCH(state->Tex[i][I915_TEXREG_SS3]);
-            OUT_BATCH(state->Tex[i][I915_TEXREG_SS4]);
-         }
-      ADVANCE_BATCH();
-   }
-
-   if (dirty & I915_UPLOAD_CONSTANTS) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_CONSTANTS:\n");
-      emit(intel, state->Constant, state->ConstantSize * sizeof(GLuint));
-   }
-
-   if (dirty & I915_UPLOAD_PROGRAM) {
-      if (INTEL_DEBUG & DEBUG_STATE)
-         fprintf(stderr, "I915_UPLOAD_PROGRAM:\n");
-
-      assert((state->Program[0] & 0x1ff) + 2 == state->ProgramSize);
-
-      emit(intel, state->Program, state->ProgramSize * sizeof(GLuint));
-      if (INTEL_DEBUG & DEBUG_STATE)
-         i915_disassemble_program(state->Program, state->ProgramSize);
-   }
-
-   state->emitted |= dirty;
-}
-
-static void
-i915_destroy_context(struct intel_context *intel)
-{
-   _tnl_free_vertices(&intel->ctx);
-}
-
-
-/**
- * Set the drawing regions for the color and depth/stencil buffers.
- * This involves setting the pitch, cpp and buffer ID/location.
- * Also set pixel format for color and Z rendering
- * Used for setting both regular and meta state.
- */
-void
-i915_state_draw_region(struct intel_context *intel,
-                       struct i915_hw_state *state,
-                       struct intel_region *color_region,
-                       struct intel_region *depth_region)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   GLuint value;
-
-   ASSERT(state == &i915->state || state == &i915->meta);
-
-   if (state->draw_region != color_region) {
-      intel_region_release(&state->draw_region);
-      intel_region_reference(&state->draw_region, color_region);
-   }
-   if (state->depth_region != depth_region) {
-      intel_region_release(&state->depth_region);
-      intel_region_reference(&state->depth_region, depth_region);
-   }
-
-   /*
-    * Set stride/cpp values
-    */
-   if (color_region) {
-      state->Buffer[I915_DESTREG_CBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      state->Buffer[I915_DESTREG_CBUFADDR1] =
-         (BUF_3D_ID_COLOR_BACK |
-          BUF_3D_PITCH(color_region->pitch * color_region->cpp) |
-          BUF_3D_USE_FENCE);
-   }
-
-   if (depth_region) {
-      state->Buffer[I915_DESTREG_DBUFADDR0] = _3DSTATE_BUF_INFO_CMD;
-      state->Buffer[I915_DESTREG_DBUFADDR1] =
-         (BUF_3D_ID_DEPTH |
-          BUF_3D_PITCH(depth_region->pitch * depth_region->cpp) |
-          BUF_3D_USE_FENCE);
-   }
-
-   /*
-    * Compute/set I915_DESTREG_DV1 value
-    */
-   value = (DSTORG_HORT_BIAS(0x8) |     /* .5 */
-            DSTORG_VERT_BIAS(0x8) |     /* .5 */
-            LOD_PRECLAMP_OGL | TEX_DEFAULT_COLOR_OGL);
-   if (color_region && color_region->cpp == 4) {
-      value |= DV_PF_8888;
-   }
-   else {
-      value |= (DITHER_FULL_ALWAYS | DV_PF_565);
-   }
-   if (depth_region && depth_region->cpp == 4) {
-      value |= DEPTH_FRMT_24_FIXED_8_OTHER;
-   }
-   else {
-      value |= DEPTH_FRMT_16_FIXED;
-   }
-   state->Buffer[I915_DESTREG_DV1] = value;
-
-   I915_STATECHANGE(i915, I915_UPLOAD_BUFFERS);
-}
-
-
-static void
-i915_set_draw_region(struct intel_context *intel,
-                     struct intel_region *color_region,
-                     struct intel_region *depth_region)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   i915_state_draw_region(intel, &i915->state, color_region, depth_region);
-}
-
-
-
-static void
-i915_lost_hardware(struct intel_context *intel)
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   i915->state.emitted = 0;
-}
-
-static GLuint
-i915_flush_cmd(void)
-{
-   return MI_FLUSH | FLUSH_MAP_CACHE;
-}
-
-static void 
-i915_assert_not_dirty( struct intel_context *intel )
-{
-   struct i915_context *i915 = i915_context(&intel->ctx);
-   struct i915_hw_state *state = i915->current;
-   GLuint dirty = get_dirty(state);
-   assert(!dirty);
-}
-
-
-void
-i915InitVtbl(struct i915_context *i915)
-{
-   i915->intel.vtbl.check_vertex_size = i915_check_vertex_size;
-   i915->intel.vtbl.destroy = i915_destroy_context;
-   i915->intel.vtbl.emit_state = i915_emit_state;
-   i915->intel.vtbl.lost_hardware = i915_lost_hardware;
-   i915->intel.vtbl.reduced_primitive_state = i915_reduced_primitive_state;
-   i915->intel.vtbl.render_start = i915_render_start;
-   i915->intel.vtbl.set_draw_region = i915_set_draw_region;
-   i915->intel.vtbl.update_texture_state = i915UpdateTextureState;
-   i915->intel.vtbl.flush_cmd = i915_flush_cmd;
-   i915->intel.vtbl.assert_not_dirty = i915_assert_not_dirty;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
deleted file mode 100644
index c92b83bcb3a..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
+++ /dev/null
@@ -1,340 +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 "intel_batchbuffer.h"
-#include "intel_ioctl.h"
-
-/* Relocations in kernel space:
- *    - pass dma buffer seperately
- *    - memory manager knows how to patch
- *    - pass list of dependent buffers
- *    - pass relocation list
- *
- * Either:
- *    - get back an offset for buffer to fire
- *    - memory manager knows how to fire buffer
- *
- * Really want the buffer to be AGP and pinned.
- *
- */
-
-/* Cliprect fence: The highest fence protecting a dma buffer
- * containing explicit cliprect information.  Like the old drawable
- * lock but irq-driven.  X server must wait for this fence to expire
- * before changing cliprects [and then doing sw rendering?].  For
- * other dma buffers, the scheduler will grab current cliprect info
- * and mix into buffer.  X server must hold the lock while changing
- * cliprects???  Make per-drawable.  Need cliprects in shared memory
- * -- beats storing them with every cmd buffer in the queue.
- *
- * ==> X server must wait for this fence to expire before touching the
- * framebuffer with new cliprects.
- *
- * ==> Cliprect-dependent buffers associated with a
- * cliprect-timestamp.  All of the buffers associated with a timestamp
- * must go to hardware before any buffer with a newer timestamp.
- *
- * ==> Dma should be queued per-drawable for correct X/GL
- * synchronization.  Or can fences be used for this?
- *
- * Applies to: Blit operations, metaops, X server operations -- X
- * server automatically waits on its own dma to complete before
- * modifying cliprects ???
- */
-
-static void
-intel_dump_batchbuffer(GLuint offset, GLuint * ptr, GLuint count)
-{
-   int i;
-   fprintf(stderr, "\n\n\nSTART BATCH (%d dwords):\n", count / 4);
-   for (i = 0; i < count / 4; i += 4)
-      fprintf(stderr, "0x%x:\t0x%08x 0x%08x 0x%08x 0x%08x\n",
-              offset + i * 4, ptr[i], ptr[i + 1], ptr[i + 2], ptr[i + 3]);
-   fprintf(stderr, "END BATCH\n\n\n");
-}
-
-void
-intel_batchbuffer_reset(struct intel_batchbuffer *batch)
-{
-
-   int i;
-
-   /*
-    * Get a new, free batchbuffer.
-    */
-
-   batch->size =  batch->intel->intelScreen->maxBatchSize;
-   driBOData(batch->buffer, batch->size, NULL, 0);
-
-   driBOResetList(&batch->list);
-
-   /*
-    * Unreference buffers previously on the relocation list.
-    */
-
-   for (i = 0; i < batch->nr_relocs; i++) {
-      struct buffer_reloc *r = &batch->reloc[i];
-      driBOUnReference(r->buf);
-   }
-
-   batch->list_count = 0;
-   batch->nr_relocs = 0;
-   batch->flags = 0;
-
-   /*
-    * We don't refcount the batchbuffer itself since we can't destroy it
-    * while it's on the list.
-    */
-
-
-   driBOAddListItem(&batch->list, batch->buffer,
-                    DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_EXE,
-                    DRM_BO_MASK_MEM | DRM_BO_FLAG_EXE);
-
-
-   batch->map = driBOMap(batch->buffer, DRM_BO_FLAG_WRITE, 0);
-   batch->ptr = batch->map;
-}
-
-/*======================================================================
- * Public functions
- */
-struct intel_batchbuffer *
-intel_batchbuffer_alloc(struct intel_context *intel)
-{
-   struct intel_batchbuffer *batch = calloc(sizeof(*batch), 1);
-
-   batch->intel = intel;
-
-   driGenBuffers(intel->intelScreen->batchPool, "batchbuffer", 1,
-                 &batch->buffer, 4096,
-                 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_EXE, 0);
-   batch->last_fence = NULL;
-   driBOCreateList(20, &batch->list);
-   intel_batchbuffer_reset(batch);
-   return batch;
-}
-
-void
-intel_batchbuffer_free(struct intel_batchbuffer *batch)
-{
-   if (batch->last_fence) {
-      driFenceFinish(batch->last_fence,
-      DRM_FENCE_TYPE_EXE | DRM_I915_FENCE_TYPE_RW, GL_FALSE);
-      driFenceUnReference(batch->last_fence);
-      batch->last_fence = NULL;
-   }
-   if (batch->map) {
-      driBOUnmap(batch->buffer);
-      batch->map = NULL;
-   }
-   driBOUnReference(batch->buffer);
-   batch->buffer = NULL;
-   free(batch);
-}
-
-/* TODO: Push this whole function into bufmgr.
- */
-static void
-do_flush_locked(struct intel_batchbuffer *batch,
-                GLuint used,
-                GLboolean ignore_cliprects, GLboolean allow_unlock)
-{
-   GLuint *ptr;
-   GLuint i;
-   struct intel_context *intel = batch->intel;
-   unsigned fenceFlags;
-   struct _DriFenceObject *fo;
-
-   driBOValidateList(batch->intel->driFd, &batch->list);
-
-   /* Apply the relocations.  This nasty map indicates to me that the
-    * whole task should be done internally by the memory manager, and
-    * that dma buffers probably need to be pinned within agp space.
-    */
-   ptr = (GLuint *) driBOMap(batch->buffer, DRM_BO_FLAG_WRITE,
-                             DRM_BO_HINT_ALLOW_UNFENCED_MAP);
-
-
-   for (i = 0; i < batch->nr_relocs; i++) {
-      struct buffer_reloc *r = &batch->reloc[i];
-
-      ptr[r->offset / 4] = driBOOffset(r->buf) + r->delta;
-   }
-
-   if (INTEL_DEBUG & DEBUG_BATCH)
-      intel_dump_batchbuffer(0, ptr, used);
-
-   driBOUnmap(batch->buffer);
-   batch->map = NULL;
-
-   /* Throw away non-effective packets.  Won't work once we have
-    * hardware contexts which would preserve statechanges beyond a
-    * single buffer.
-    */
-
-   if (!(intel->numClipRects == 0 && !ignore_cliprects)) {
-      intel_batch_ioctl(batch->intel,
-                        driBOOffset(batch->buffer),
-                        used, ignore_cliprects, allow_unlock);
-   }
-
-
-   /*
-    * Kernel fencing. The flags tells the kernel that we've 
-    * programmed an MI_FLUSH.
-    */
-   
-   fenceFlags = DRM_I915_FENCE_FLAG_FLUSHED;
-   fo = driFenceBuffers(batch->intel->driFd,
-			"Batch fence", fenceFlags);
-
-   /*
-    * User space fencing.
-    */
-
-   driBOFence(batch->buffer, fo);
-
-   if (driFenceType(fo) == DRM_FENCE_TYPE_EXE) {
-
-     /*
-      * Oops. We only validated a batch buffer. This means we
-      * didn't do any proper rendering. Discard this fence object.
-      */
-
-      driFenceUnReference(fo);
-   } else {
-      driFenceUnReference(batch->last_fence);
-      batch->last_fence = fo;
-      for (i = 0; i < batch->nr_relocs; i++) {
-	struct buffer_reloc *r = &batch->reloc[i];
-	driBOFence(r->buf, fo);
-      }
-   }
-
-   if (intel->numClipRects == 0 && !ignore_cliprects) {
-      if (allow_unlock) {
-         UNLOCK_HARDWARE(intel);
-         sched_yield();
-         LOCK_HARDWARE(intel);
-      }
-      intel->vtbl.lost_hardware(intel);
-   }
-}
-
-
-struct _DriFenceObject *
-intel_batchbuffer_flush(struct intel_batchbuffer *batch)
-{
-   struct intel_context *intel = batch->intel;
-   GLuint used = batch->ptr - batch->map;
-   GLboolean was_locked = intel->locked;
-
-   if (used == 0)
-      return batch->last_fence;
-
-   /* Add the MI_BATCH_BUFFER_END.  Always add an MI_FLUSH - this is a
-    * performance drain that we would like to avoid.
-    */
-   if (used & 4) {
-      ((int *) batch->ptr)[0] = intel->vtbl.flush_cmd();
-      ((int *) batch->ptr)[1] = 0;
-      ((int *) batch->ptr)[2] = MI_BATCH_BUFFER_END;
-      used += 12;
-   }
-   else {
-      ((int *) batch->ptr)[0] = intel->vtbl.flush_cmd();
-      ((int *) batch->ptr)[1] = MI_BATCH_BUFFER_END;
-      used += 8;
-   }
-
-   driBOUnmap(batch->buffer);
-   batch->ptr = NULL;
-   batch->map = NULL;
-
-   /* TODO: Just pass the relocation list and dma buffer up to the
-    * kernel.
-    */
-   if (!was_locked)
-      LOCK_HARDWARE(intel);
-
-   do_flush_locked(batch, used, !(batch->flags & INTEL_BATCH_CLIPRECTS),
-		   GL_FALSE);
-
-   if (!was_locked)
-      UNLOCK_HARDWARE(intel);
-
-   /* Reset the buffer:
-    */
-   intel_batchbuffer_reset(batch);
-   return batch->last_fence;
-}
-
-void
-intel_batchbuffer_finish(struct intel_batchbuffer *batch)
-{
-   struct _DriFenceObject *fence = intel_batchbuffer_flush(batch);
-   driFenceReference(fence);
-   driFenceFinish(fence, 3, GL_FALSE);
-   driFenceUnReference(fence);
-}
-
-
-/*  This is the only way buffers get added to the validate list.
- */
-GLboolean
-intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
-                             struct _DriBufferObject *buffer,
-                             GLuint flags, GLuint mask, GLuint delta)
-{
-   assert(batch->nr_relocs < MAX_RELOCS);
-
-   driBOAddListItem(&batch->list, buffer, flags, mask);
-
-   {
-      struct buffer_reloc *r = &batch->reloc[batch->nr_relocs++];
-      driBOReference(buffer);
-      r->buf = buffer;
-      r->offset = batch->ptr - batch->map;
-      r->delta = delta;
-   }
-
-   batch->ptr += 4;
-   return GL_TRUE;
-}
-
-
-
-void
-intel_batchbuffer_data(struct intel_batchbuffer *batch,
-                       const void *data, GLuint bytes, GLuint flags)
-{
-   assert((bytes & 3) == 0);
-   intel_batchbuffer_require_space(batch, bytes, flags);
-   __memcpy(batch->ptr, data, bytes);
-   batch->ptr += bytes;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
deleted file mode 100644
index 212f1301012..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
+++ /dev/null
@@ -1,124 +0,0 @@
-#ifndef INTEL_BATCHBUFFER_H
-#define INTEL_BATCHBUFFER_H
-
-#include "mtypes.h"
-#include "dri_bufmgr.h"
-
-struct intel_context;
-
-#define BATCH_SZ 16384
-#define BATCH_RESERVED 16
-
-#define MAX_RELOCS 4096
-
-#define INTEL_BATCH_NO_CLIPRECTS 0x1
-#define INTEL_BATCH_CLIPRECTS    0x2
-
-struct buffer_reloc
-{
-   struct _DriBufferObject *buf;
-   GLuint offset;
-   GLuint delta;                /* not needed? */
-};
-
-struct intel_batchbuffer
-{
-   struct bufmgr *bm;
-   struct intel_context *intel;
-
-   struct _DriBufferObject *buffer;
-   struct _DriFenceObject *last_fence;
-   GLuint flags;
-
-   drmBOList list;
-   GLuint list_count;
-   GLubyte *map;
-   GLubyte *ptr;
-
-   struct buffer_reloc reloc[MAX_RELOCS];
-   GLuint nr_relocs;
-   GLuint size;
-};
-
-struct intel_batchbuffer *intel_batchbuffer_alloc(struct intel_context
-                                                  *intel);
-
-void intel_batchbuffer_free(struct intel_batchbuffer *batch);
-
-
-void intel_batchbuffer_finish(struct intel_batchbuffer *batch);
-
-struct _DriFenceObject *intel_batchbuffer_flush(struct intel_batchbuffer
-                                                *batch);
-
-void intel_batchbuffer_reset(struct intel_batchbuffer *batch);
-
-
-/* Unlike bmBufferData, this currently requires the buffer be mapped.
- * Consider it a convenience function wrapping multple
- * intel_buffer_dword() calls.
- */
-void intel_batchbuffer_data(struct intel_batchbuffer *batch,
-                            const void *data, GLuint bytes, GLuint flags);
-
-void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
-                                     GLuint bytes);
-
-GLboolean intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
-                                       struct _DriBufferObject *buffer,
-                                       GLuint flags,
-                                       GLuint mask, GLuint offset);
-
-/* Inline functions - might actually be better off with these
- * non-inlined.  Certainly better off switching all command packets to
- * be passed as structs rather than dwords, but that's a little bit of
- * work...
- */
-static INLINE GLuint
-intel_batchbuffer_space(struct intel_batchbuffer *batch)
-{
-   return (batch->size - BATCH_RESERVED) - (batch->ptr - batch->map);
-}
-
-
-static INLINE void
-intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch, GLuint dword)
-{
-   assert(batch->map);
-   assert(intel_batchbuffer_space(batch) >= 4);
-   *(GLuint *) (batch->ptr) = dword;
-   batch->ptr += 4;
-}
-
-static INLINE void
-intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
-                                GLuint sz, GLuint flags)
-{
-   assert(sz < batch->size - 8);
-   if (intel_batchbuffer_space(batch) < sz ||
-       (batch->flags != 0 && flags != 0 && batch->flags != flags))
-      intel_batchbuffer_flush(batch);
-
-   batch->flags |= flags;
-}
-
-/* Here are the crusty old macros, to be removed:
- */
-#define BATCH_LOCALS
-
-#define BEGIN_BATCH(n, flags) do {				\
-   assert(!intel->prim.flush);					\
-   intel_batchbuffer_require_space(intel->batch, (n)*4, flags);	\
-} while (0)
-
-#define OUT_BATCH(d)  intel_batchbuffer_emit_dword(intel->batch, d)
-
-#define OUT_RELOC(buf,flags,mask,delta) do { 				\
-   assert((delta) >= 0);							\
-   intel_batchbuffer_emit_reloc(intel->batch, buf, flags, mask, delta);	\
-} while (0)
-
-#define ADVANCE_BATCH() do { } while(0)
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchpool.c b/src/mesa/drivers/dri/i915tex/intel_batchpool.c
deleted file mode 100644
index 2503b8a62a1..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_batchpool.c
+++ /dev/null
@@ -1,418 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2006 Tungsten Graphics, Inc., Bismarck, ND., USA
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
- * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
- * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR 
- * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE 
- * USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * 
- **************************************************************************/
-/*
- * Authors: Thomas Hellstr�m <thomas-at-tungstengraphics-dot-com>
- */
-
-#include <xf86drm.h>
-#include <stdlib.h>
-#include <errno.h>
-#include "imports.h"
-#include "glthread.h"
-#include "dri_bufpool.h"
-#include "dri_bufmgr.h"
-#include "intel_screen.h"
-
-typedef struct
-{
-   drmMMListHead head;
-   struct _BPool *parent;
-   struct _DriFenceObject *fence;
-   unsigned long start;
-   int unfenced;
-   int mapped;
-} BBuf;
-
-typedef struct _BPool
-{
-   _glthread_Mutex mutex;
-   unsigned long bufSize;
-   unsigned poolSize;
-   unsigned numFree;
-   unsigned numTot;
-   unsigned numDelayed;
-   unsigned checkDelayed;
-   drmMMListHead free;
-   drmMMListHead delayed;
-   drmMMListHead head;
-   drmBO kernelBO;
-   void *virtual;
-   BBuf *bufs;
-} BPool;
-
-
-static BPool *
-createBPool(int fd, unsigned long bufSize, unsigned numBufs, unsigned flags,
-            unsigned checkDelayed)
-{
-   BPool *p = (BPool *) malloc(sizeof(*p));
-   BBuf *buf;
-   int i;
-
-   if (!p)
-      return NULL;
-
-   p->bufs = (BBuf *) malloc(numBufs * sizeof(*p->bufs));
-   if (!p->bufs) {
-      free(p);
-      return NULL;
-   }
-
-   DRMINITLISTHEAD(&p->free);
-   DRMINITLISTHEAD(&p->head);
-   DRMINITLISTHEAD(&p->delayed);
-
-   p->numTot = numBufs;
-   p->numFree = numBufs;
-   p->bufSize = bufSize;
-   p->numDelayed = 0;
-   p->checkDelayed = checkDelayed;
-
-   _glthread_INIT_MUTEX(p->mutex);
-
-   if (drmBOCreate(fd, 0, numBufs * bufSize, 0, NULL, drm_bo_type_dc,
-                   flags, DRM_BO_HINT_DONT_FENCE, &p->kernelBO)) {
-      free(p->bufs);
-      free(p);
-      return NULL;
-   }
-   if (drmBOMap(fd, &p->kernelBO, DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE, 0,
-                &p->virtual)) {
-      drmBODestroy(fd, &p->kernelBO);
-      free(p->bufs);
-      free(p);
-      return NULL;
-   }
-
-   /*
-    * We unmap the buffer so that we can validate it later. Note that this is
-    * just a synchronizing operation. The buffer will have a virtual mapping
-    * until it is destroyed.
-    */
-
-   drmBOUnmap(fd, &p->kernelBO);
-
-   buf = p->bufs;
-   for (i = 0; i < numBufs; ++i) {
-      buf->parent = p;
-      buf->fence = NULL;
-      buf->start = i * bufSize;
-      buf->mapped = 0;
-      buf->unfenced = 0;
-      DRMLISTADDTAIL(&buf->head, &p->free);
-      buf++;
-   }
-
-   return p;
-}
-
-
-static void
-pool_checkFree(BPool * p, int wait)
-{
-   drmMMListHead *list, *prev;
-   BBuf *buf;
-   int signaled = 0;
-   int i;
-
-   list = p->delayed.next;
-
-   if (p->numDelayed > 3) {
-      for (i = 0; i < p->numDelayed; i += 3) {
-         list = list->next;
-      }
-   }
-
-   prev = list->prev;
-   for (; list != &p->delayed; list = prev, prev = list->prev) {
-
-      buf = DRMLISTENTRY(BBuf, list, head);
-
-      if (!signaled) {
-         if (wait) {
-            driFenceFinish(buf->fence, DRM_FENCE_TYPE_EXE, 1);
-            signaled = 1;
-         }
-         else {
-            signaled = driFenceSignaled(buf->fence, DRM_FENCE_TYPE_EXE);
-         }
-      }
-
-      if (!signaled)
-         break;
-
-      driFenceUnReference(buf->fence);
-      buf->fence = NULL;
-      DRMLISTDEL(list);
-      p->numDelayed--;
-      DRMLISTADD(list, &p->free);
-      p->numFree++;
-   }
-}
-
-static void *
-pool_create(struct _DriBufferPool *pool,
-            unsigned long size, unsigned flags, unsigned hint,
-            unsigned alignment)
-{
-   BPool *p = (BPool *) pool->data;
-
-   drmMMListHead *item;
-
-   if (alignment && (alignment != 4096))
-      return NULL;
-
-   _glthread_LOCK_MUTEX(p->mutex);
-
-   if (p->numFree == 0)
-      pool_checkFree(p, GL_TRUE);
-
-   if (p->numFree == 0) {
-      fprintf(stderr, "Out of fixed size buffer objects\n");
-      BM_CKFATAL(-ENOMEM);
-   }
-
-   item = p->free.next;
-
-   if (item == &p->free) {
-      fprintf(stderr, "Fixed size buffer pool corruption\n");
-   }
-
-   DRMLISTDEL(item);
-   --p->numFree;
-
-   _glthread_UNLOCK_MUTEX(p->mutex);
-   return (void *) DRMLISTENTRY(BBuf, item, head);
-}
-
-
-static int
-pool_destroy(struct _DriBufferPool *pool, void *private)
-{
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-
-   _glthread_LOCK_MUTEX(p->mutex);
-
-   if (buf->fence) {
-      DRMLISTADDTAIL(&buf->head, &p->delayed);
-      p->numDelayed++;
-   }
-   else {
-      buf->unfenced = 0;
-      DRMLISTADD(&buf->head, &p->free);
-      p->numFree++;
-   }
-
-   if ((p->numDelayed % p->checkDelayed) == 0)
-      pool_checkFree(p, 0);
-
-   _glthread_UNLOCK_MUTEX(p->mutex);
-   return 0;
-}
-
-
-static int
-pool_map(struct _DriBufferPool *pool, void *private, unsigned flags,
-         int hint, void **virtual)
-{
-
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-
-   _glthread_LOCK_MUTEX(p->mutex);
-
-   /*
-    * Currently Mesa doesn't have any condition variables to resolve this
-    * cleanly in a multithreading environment.
-    * We bail out instead.
-    */
-
-   if (buf->mapped) {
-      fprintf(stderr, "Trying to map already mapped buffer object\n");
-      BM_CKFATAL(-EINVAL);
-   }
-
-#if 0
-   if (buf->unfenced && !(hint & DRM_BO_HINT_ALLOW_UNFENCED_MAP)) {
-      fprintf(stderr, "Trying to map an unfenced buffer object 0x%08x"
-              " 0x%08x %d\n", hint, flags, buf->start);
-      BM_CKFATAL(-EINVAL);
-   }
-
-#endif
-
-   if (buf->fence) {
-      _glthread_UNLOCK_MUTEX(p->mutex);
-      return -EBUSY;
-   }
-
-   buf->mapped = GL_TRUE;
-   *virtual = (unsigned char *) p->virtual + buf->start;
-   _glthread_UNLOCK_MUTEX(p->mutex);
-   return 0;
-}
-
-static int
-pool_waitIdle(struct _DriBufferPool *pool, void *private, int lazy)
-{
-   BBuf *buf = (BBuf *) private;
-   driFenceFinish(buf->fence, 0, lazy);
-   return 0;
-}
-
-static int
-pool_unmap(struct _DriBufferPool *pool, void *private)
-{
-   BBuf *buf = (BBuf *) private;
-
-   buf->mapped = 0;
-   return 0;
-}
-
-static unsigned long
-pool_offset(struct _DriBufferPool *pool, void *private)
-{
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-
-   return p->kernelBO.offset + buf->start;
-}
-
-static unsigned
-pool_flags(struct _DriBufferPool *pool, void *private)
-{
-   BPool *p = (BPool *) pool->data;
-
-   return p->kernelBO.flags;
-}
-
-static unsigned long
-pool_size(struct _DriBufferPool *pool, void *private)
-{
-   BPool *p = (BPool *) pool->data;
-
-   return p->bufSize;
-}
-
-
-static int
-pool_fence(struct _DriBufferPool *pool, void *private,
-           struct _DriFenceObject *fence)
-{
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-
-   _glthread_LOCK_MUTEX(p->mutex);
-   if (buf->fence) {
-      driFenceUnReference(buf->fence);
-   }
-   buf->fence = fence;
-   buf->unfenced = 0;
-   driFenceReference(buf->fence);
-   _glthread_UNLOCK_MUTEX(p->mutex);
-
-   return 0;
-}
-
-static drmBO *
-pool_kernel(struct _DriBufferPool *pool, void *private)
-{
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-
-   return &p->kernelBO;
-}
-
-static int
-pool_validate(struct _DriBufferPool *pool, void *private)
-{
-   BBuf *buf = (BBuf *) private;
-   BPool *p = buf->parent;
-   _glthread_LOCK_MUTEX(p->mutex);
-   buf->unfenced = GL_TRUE;
-   _glthread_UNLOCK_MUTEX(p->mutex);
-   return 0;
-}
-
-static void
-pool_takedown(struct _DriBufferPool *pool)
-{
-   BPool *p = (BPool *) pool->data;
-
-   /*
-    * Wait on outstanding fences. 
-    */
-
-   _glthread_LOCK_MUTEX(p->mutex);
-   while ((p->numFree < p->numTot) && p->numDelayed) {
-      _glthread_UNLOCK_MUTEX(p->mutex);
-      sched_yield();
-      pool_checkFree(p, GL_TRUE);
-      _glthread_LOCK_MUTEX(p->mutex);
-   }
-
-   drmBODestroy(pool->fd, &p->kernelBO);
-   free(p->bufs);
-   _glthread_UNLOCK_MUTEX(p->mutex);
-   free(p);
-   free(pool);
-}
-
-
-struct _DriBufferPool *
-driBatchPoolInit(int fd, unsigned flags,
-                 unsigned long bufSize,
-                 unsigned numBufs, unsigned checkDelayed)
-{
-   struct _DriBufferPool *pool;
-
-   pool = (struct _DriBufferPool *) malloc(sizeof(*pool));
-   if (!pool)
-      return NULL;
-
-   pool->data = createBPool(fd, bufSize, numBufs, flags, checkDelayed);
-   if (!pool->data)
-      return NULL;
-
-   pool->fd = fd;
-   pool->map = &pool_map;
-   pool->unmap = &pool_unmap;
-   pool->destroy = &pool_destroy;
-   pool->offset = &pool_offset;
-   pool->flags = &pool_flags;
-   pool->size = &pool_size;
-   pool->create = &pool_create;
-   pool->fence = &pool_fence;
-   pool->kernel = &pool_kernel;
-   pool->validate = &pool_validate;
-   pool->waitIdle = &pool_waitIdle;
-   pool->setstatic = NULL;
-   pool->takeDown = &pool_takedown;
-   return pool;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.c b/src/mesa/drivers/dri/i915tex/intel_context.c
deleted file mode 100644
index 40ea7564126..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_context.c
+++ /dev/null
@@ -1,787 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "matrix.h"
-#include "simple_list.h"
-#include "extensions.h"
-#include "framebuffer.h"
-#include "imports.h"
-#include "points.h"
-
-#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_buffers.h"
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_tris.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_pixel.h"
-#include "intel_regions.h"
-#include "intel_buffer_objects.h"
-#include "intel_fbo.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_point_parameters
-#define need_GL_ARB_texture_compression
-#define need_GL_ARB_vertex_buffer_object
-#define need_GL_ARB_vertex_program
-#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_secondary_color
-#define need_GL_NV_vertex_program
-#include "extension_helper.h"
-
-
-#define DRIVER_DATE                     "20061102"
-
-_glthread_Mutex lockMutex;
-static GLboolean lockMutexInit = GL_FALSE;
-
-
-static const GLubyte *
-intelGetString(GLcontext * ctx, GLenum name)
-{
-   const char *chipset;
-   static char buffer[128];
-
-   switch (name) {
-   case GL_VENDOR:
-      return (GLubyte *) "Tungsten Graphics, Inc";
-      break;
-
-   case GL_RENDERER:
-      switch (intel_context(ctx)->intelScreen->deviceID) {
-      case PCI_CHIP_845_G:
-         chipset = "Intel(R) 845G";
-         break;
-      case PCI_CHIP_I830_M:
-         chipset = "Intel(R) 830M";
-         break;
-      case PCI_CHIP_I855_GM:
-         chipset = "Intel(R) 852GM/855GM";
-         break;
-      case PCI_CHIP_I865_G:
-         chipset = "Intel(R) 865G";
-         break;
-      case PCI_CHIP_I915_G:
-         chipset = "Intel(R) 915G";
-         break;
-      case PCI_CHIP_I915_GM:
-         chipset = "Intel(R) 915GM";
-         break;
-      case PCI_CHIP_I945_G:
-         chipset = "Intel(R) 945G";
-         break;
-      case PCI_CHIP_I945_GM:
-         chipset = "Intel(R) 945GM";
-         break;
-      case PCI_CHIP_I945_GME:
-         chipset = "Intel(R) 945GME";
-         break;
-      case PCI_CHIP_G33_G:
-	 chipset = "Intel(R) G33";
-	 break;
-      case PCI_CHIP_Q35_G:
-	 chipset = "Intel(R) Q35";
-	 break;
-      case PCI_CHIP_Q33_G:
-	 chipset = "Intel(R) Q33";
-	 break;
-      default:
-         chipset = "Unknown Intel Chipset";
-         break;
-      }
-
-      (void) driGetRendererString(buffer, chipset, DRIVER_DATE, 0);
-      return (GLubyte *) buffer;
-
-   default:
-      return NULL;
-   }
-}
-
-
-/**
- * Extension strings exported by the intel driver.
- *
- * \note
- * It appears that ARB_texture_env_crossbar has "disappeared" compared to the
- * old i830-specific driver.
- */
-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_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_pixel_buffer_object", 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_subtract", NULL},
-   {"GL_EXT_cull_vertex", GL_EXT_cull_vertex_functions},
-   {"GL_EXT_fog_coord", GL_EXT_fog_coord_functions},
-   {"GL_EXT_framebuffer_object", GL_EXT_framebuffer_object_functions},
-   {"GL_EXT_multi_draw_arrays", GL_EXT_multi_draw_arrays_functions},
-#if 1                           /* XXX FBO temporary? */
-   {"GL_EXT_packed_depth_stencil", NULL},
-#endif
-   {"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_vertex_program", GL_NV_vertex_program_functions},
-   {"GL_NV_vertex_program1_1", NULL},
-/*     { "GL_SGIS_generate_mipmap",           NULL }, */
-   {NULL, NULL}
-};
-
-extern const struct tnl_pipeline_stage _intel_render_stage;
-
-static const struct tnl_pipeline_stage *intel_pipeline[] = {
-   &_tnl_vertex_transform_stage,
-   &_tnl_vertex_cull_stage,
-   &_tnl_normal_transform_stage,
-   &_tnl_lighting_stage,
-   &_tnl_fog_coordinate_stage,
-   &_tnl_texgen_stage,
-   &_tnl_texture_transform_stage,
-   &_tnl_point_attenuation_stage,
-   &_tnl_vertex_program_stage,
-#if 1
-   &_intel_render_stage,        /* ADD: unclipped rastersetup-to-dma */
-#endif
-   &_tnl_render_stage,
-   0,
-};
-
-
-static const struct dri_debug_control debug_control[] = {
-   {"tex", DEBUG_TEXTURE},
-   {"state", DEBUG_STATE},
-   {"ioctl", DEBUG_IOCTL},
-   {"blit", DEBUG_BLIT},
-   {"mip", DEBUG_MIPTREE},
-   {"fall", DEBUG_FALLBACKS},
-   {"verb", DEBUG_VERBOSE},
-   {"bat", DEBUG_BATCH},
-   {"pix", DEBUG_PIXEL},
-   {"buf", DEBUG_BUFMGR},
-   {"reg", DEBUG_REGION},
-   {"fbo", DEBUG_FBO},
-   {"lock", DEBUG_LOCK},
-   {NULL, 0}
-};
-
-
-static void
-intelInvalidateState(GLcontext * ctx, GLuint new_state)
-{
-   _swrast_InvalidateState(ctx, new_state);
-   _swsetup_InvalidateState(ctx, new_state);
-   _vbo_InvalidateState(ctx, new_state);
-   _tnl_InvalidateState(ctx, new_state);
-   _tnl_invalidate_vertex_state(ctx, new_state);
-   intel_context(ctx)->NewGLState |= new_state;
-}
-
-
-void
-intelFlush(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   if (intel->Fallback)
-      _swrast_flush(ctx);
-
-   INTEL_FIREVERTICES(intel);
-
-   if (intel->batch->map != intel->batch->ptr)
-      intel_batchbuffer_flush(intel->batch);
-
-   /* XXX: Need to do an MI_FLUSH here.
-    */
-}
-
-
-/**
- * Check if we need to rotate/warp the front color buffer to the
- * rotated screen.  We generally need to do this when we get a glFlush
- * or glFinish after drawing to the front color buffer.
- */
-static void
-intelCheckFrontRotate(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-   if (intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] ==
-       BUFFER_BIT_FRONT_LEFT) {
-      intelScreenPrivate *screen = intel->intelScreen;
-      if (screen->current_rotation != 0) {
-         __DRIdrawablePrivate *dPriv = intel->driDrawable;
-         intelRotateWindow(intel, dPriv, BUFFER_BIT_FRONT_LEFT);
-      }
-   }
-}
-
-
-/**
- * Called via glFlush.
- */
-static void
-intelglFlush(GLcontext * ctx)
-{
-   intelFlush(ctx);
-   intelCheckFrontRotate(ctx);
-}
-
-void
-intelFinish(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-   intelFlush(ctx);
-   if (intel->batch->last_fence) {
-      driFenceFinish(intel->batch->last_fence,
-                     0, GL_FALSE);
-      driFenceUnReference(intel->batch->last_fence);
-      intel->batch->last_fence = NULL;
-   }
-   intelCheckFrontRotate(ctx);
-}
-
-
-void
-intelInitDriverFunctions(struct dd_function_table *functions)
-{
-   _mesa_init_driver_functions(functions);
-
-   functions->Flush = intelglFlush;
-   functions->Finish = intelFinish;
-   functions->GetString = intelGetString;
-   functions->UpdateState = intelInvalidateState;
-   functions->CopyColorTable = _swrast_CopyColorTable;
-   functions->CopyColorSubTable = _swrast_CopyColorSubTable;
-   functions->CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
-   functions->CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
-
-   intelInitTextureFuncs(functions);
-   intelInitPixelFuncs(functions);
-   intelInitStateFuncs(functions);
-   intelInitBufferFuncs(functions);
-}
-
-
-GLboolean
-intelInitContext(struct intel_context *intel,
-                 const __GLcontextModes * mesaVis,
-                 __DRIcontextPrivate * driContextPriv,
-                 void *sharedContextPrivate,
-                 struct dd_function_table *functions)
-{
-   GLcontext *ctx = &intel->ctx;
-   GLcontext *shareCtx = (GLcontext *) sharedContextPrivate;
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-   drmI830Sarea *saPriv = (drmI830Sarea *)
-      (((GLubyte *) sPriv->pSAREA) + intelScreen->sarea_priv_offset);
-   int fthrottle_mode;
-   GLboolean havePools;
-
-   DRM_LIGHT_LOCK(sPriv->fd, &sPriv->pSAREA->lock, driContextPriv->hHWContext);
-   havePools = intelCreatePools(intelScreen);
-   DRM_UNLOCK(sPriv->fd, &sPriv->pSAREA->lock, driContextPriv->hHWContext);
-
-   if (!havePools)
-      return GL_FALSE;
-     
-   if (!_mesa_initialize_context(&intel->ctx,
-                                 mesaVis, shareCtx,
-                                 functions, (void *) intel))
-      return GL_FALSE;
-
-   driContextPriv->driverPrivate = intel;
-   intel->intelScreen = intelScreen;
-   intel->driScreen = sPriv;
-   intel->sarea = saPriv;
-
-   intel->width = intelScreen->width;
-   intel->height = intelScreen->height;
-   intel->current_rotation = intelScreen->current_rotation;
-
-   if (!lockMutexInit) {
-      lockMutexInit = GL_TRUE;
-      _glthread_INIT_MUTEX(lockMutex);
-   }
-
-   driParseConfigFiles(&intel->optionCache, &intelScreen->optionCache,
-                       intel->driScreen->myNum, "i915");
-
-   ctx->Const.MaxTextureMaxAnisotropy = 2.0;
-
-   /* This doesn't yet catch all non-conformant rendering, but it's a
-    * start.
-    */
-   if (getenv("INTEL_STRICT_CONFORMANCE")) {
-      intel->strict_conformance = 1;
-   }
-
-   ctx->Const.MinLineWidth = 1.0;
-   ctx->Const.MinLineWidthAA = 1.0;
-   ctx->Const.MaxLineWidth = 3.0;
-   ctx->Const.MaxLineWidthAA = 3.0;
-   ctx->Const.LineWidthGranularity = 1.0;
-
-   ctx->Const.MinPointSize = 1.0;
-   ctx->Const.MinPointSizeAA = 1.0;
-   ctx->Const.MaxPointSize = 255.0;
-   ctx->Const.MaxPointSizeAA = 3.0;
-   ctx->Const.PointSizeGranularity = 1.0;
-
-   /* reinitialize the context point state.
-    * It depend on constants in __GLcontextRec::Const
-    */
-   _mesa_init_point(ctx);
-
-   ctx->Const.MaxColorAttachments = 4;  /* XXX FBO: review this */
-
-   /* Initialize the software rasterizer and helper modules. */
-   _swrast_CreateContext(ctx);
-   _vbo_CreateContext(ctx);
-   _tnl_CreateContext(ctx);
-   _swsetup_CreateContext(ctx);
-
-   /* Install the customized pipeline: */
-   _tnl_destroy_pipeline(ctx);
-   _tnl_install_pipeline(ctx, intel_pipeline);
-
-   /* Configure swrast to match hardware characteristics: */
-   _swrast_allow_pixel_fog(ctx, GL_FALSE);
-   _swrast_allow_vertex_fog(ctx, GL_TRUE);
-
-   /* Dri stuff */
-   intel->hHWContext = driContextPriv->hHWContext;
-   intel->driFd = sPriv->fd;
-   intel->driHwLock = (drmLock *) & sPriv->pSAREA->lock;
-
-   intel->hw_stipple = 1;
-
-   /* XXX FBO: this doesn't seem to be used anywhere */
-   switch (mesaVis->depthBits) {
-   case 0:                     /* what to do in this case? */
-   case 16:
-      intel->polygon_offset_scale = 1.0 / 0xffff;
-      break;
-   case 24:
-      intel->polygon_offset_scale = 2.0 / 0xffffff;     /* req'd to pass glean */
-      break;
-   default:
-      assert(0);
-      break;
-   }
-
-   /* Initialize swrast, tnl driver tables: */
-   intelInitSpanFuncs(ctx);
-   intelInitTriFuncs(ctx);
-
-
-   intel->RenderIndex = ~0;
-
-   fthrottle_mode = driQueryOptioni(&intel->optionCache, "fthrottle_mode");
-   intel->iw.irq_seq = -1;
-   intel->irqsEmitted = 0;
-
-   intel->do_irqs = (intel->intelScreen->irq_active &&
-                     fthrottle_mode == DRI_CONF_FTHROTTLE_IRQS);
-
-   intel->do_usleeps = (fthrottle_mode == DRI_CONF_FTHROTTLE_USLEEPS);
-
-   _math_matrix_ctr(&intel->ViewportMatrix);
-
-   /* Disable imaging extension until convolution is working in
-    * teximage paths:
-    */
-   driInitExtensions(ctx, card_extensions,
-/* 		      GL_TRUE, */
-                     GL_FALSE);
-
-
-   intel->batch = intel_batchbuffer_alloc(intel);
-   intel->last_swap_fence = NULL;
-   intel->first_swap_fence = NULL;
-
-   intel_bufferobj_init(intel);
-   intel_fbo_init(intel);
-
-   if (intel->ctx.Mesa_DXTn) {
-      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
-      _mesa_enable_extension(ctx, "GL_S3_s3tc");
-   }
-   else if (driQueryOptionb(&intel->optionCache, "force_s3tc_enable")) {
-      _mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
-   }
-
-   intel->prim.primitive = ~0;
-
-
-#if DO_DEBUG
-   INTEL_DEBUG = driParseDebugString(getenv("INTEL_DEBUG"), debug_control);
-#endif
-
-   if (getenv("INTEL_NO_RAST")) {
-      fprintf(stderr, "disabling 3D rasterization\n");
-      FALLBACK(intel, INTEL_FALLBACK_USER, 1);
-   }
-
-   return GL_TRUE;
-}
-
-void
-intelDestroyContext(__DRIcontextPrivate * driContextPriv)
-{
-   struct intel_context *intel =
-      (struct intel_context *) driContextPriv->driverPrivate;
-
-   assert(intel);               /* should never be null */
-   if (intel) {
-      GLboolean release_texture_heaps;
-
-      INTEL_FIREVERTICES(intel);
-
-      intel->vtbl.destroy(intel);
-
-      release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
-      _swsetup_DestroyContext(&intel->ctx);
-      _tnl_DestroyContext(&intel->ctx);
-      _vbo_DestroyContext(&intel->ctx);
-
-      _swrast_DestroyContext(&intel->ctx);
-      intel->Fallback = 0;      /* don't call _swrast_Flush later */
-
-      intel_batchbuffer_free(intel->batch);
-
-      if (intel->last_swap_fence) {
-	 driFenceFinish(intel->last_swap_fence, DRM_FENCE_TYPE_EXE, GL_TRUE);
-	 driFenceUnReference(intel->last_swap_fence);
-	 intel->last_swap_fence = NULL;
-      }
-      if (intel->first_swap_fence) {
-	 driFenceFinish(intel->first_swap_fence, DRM_FENCE_TYPE_EXE, GL_TRUE);
-	 driFenceUnReference(intel->first_swap_fence);
-	 intel->first_swap_fence = NULL;
-      }
-
-
-      if (release_texture_heaps) {
-         /* This share group is about to go away, free our private
-          * texture object data.
-          */
-         if (INTEL_DEBUG & DEBUG_TEXTURE)
-            fprintf(stderr, "do something to free texture heaps\n");
-      }
-
-      /* free the Mesa context */
-      _mesa_free_context_data(&intel->ctx);
-   }
-}
-
-GLboolean
-intelUnbindContext(__DRIcontextPrivate * driContextPriv)
-{
-   return GL_TRUE;
-}
-
-GLboolean
-intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
-                 __DRIdrawablePrivate * driDrawPriv,
-                 __DRIdrawablePrivate * driReadPriv)
-{
-
-   if (driContextPriv) {
-      struct intel_context *intel =
-         (struct intel_context *) driContextPriv->driverPrivate;
-      struct intel_framebuffer *intel_fb =
-	 (struct intel_framebuffer *) driDrawPriv->driverPrivate;
-      GLframebuffer *readFb = (GLframebuffer *) driReadPriv->driverPrivate;
-
-
-      /* XXX FBO temporary fix-ups! */
-      /* if the renderbuffers don't have regions, init them from the context */
-      {
-         struct intel_renderbuffer *irbDepth
-            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_DEPTH);
-         struct intel_renderbuffer *irbStencil
-            = intel_get_renderbuffer(&intel_fb->Base, BUFFER_STENCIL);
-
-         if (intel_fb->color_rb[0] && !intel_fb->color_rb[0]->region) {
-            intel_region_reference(&intel_fb->color_rb[0]->region,
-				   intel->intelScreen->front_region);
-         }
-         if (intel_fb->color_rb[1] && !intel_fb->color_rb[1]->region) {
-            intel_region_reference(&intel_fb->color_rb[1]->region,
-				   intel->intelScreen->back_region);
-         }
-         if (intel_fb->color_rb[2] && !intel_fb->color_rb[2]->region) {
-            intel_region_reference(&intel_fb->color_rb[2]->region,
-				   intel->intelScreen->third_region);
-         }
-         if (irbDepth && !irbDepth->region) {
-            intel_region_reference(&irbDepth->region, intel->intelScreen->depth_region);
-         }
-         if (irbStencil && !irbStencil->region) {
-            intel_region_reference(&irbStencil->region, intel->intelScreen->depth_region);
-         }
-      }
-
-      /* set GLframebuffer size to match window, if needed */
-      driUpdateFramebufferSize(&intel->ctx, driDrawPriv);
-
-      if (driReadPriv != driDrawPriv) {
-	 driUpdateFramebufferSize(&intel->ctx, driReadPriv);
-      }
-
-      _mesa_make_current(&intel->ctx, &intel_fb->Base, readFb);
-
-      /* The drawbuffer won't always be updated by _mesa_make_current: 
-       */
-      if (intel->ctx.DrawBuffer == &intel_fb->Base) {
-
-	 if (intel->driDrawable != driDrawPriv) {
-	    if (driDrawPriv->pdraw->swap_interval == (unsigned)-1) {
-	       int i;
-
-	       intel_fb->vblank_flags = (intel->intelScreen->irq_active != 0)
-		  ? driGetDefaultVBlankFlags(&intel->optionCache)
-		 : VBLANK_FLAG_NO_IRQ;
-
-	       (*dri_interface->getUST) (&intel_fb->swap_ust);
-	       driDrawableInitVBlank(driDrawPriv, intel_fb->vblank_flags,
-				     &intel_fb->vbl_seq);
-	       intel_fb->vbl_waited = intel_fb->vbl_seq;
-
-	       for (i = 0; i < (intel->intelScreen->third.handle ? 3 : 2); i++) {
-		  if (intel_fb->color_rb[i])
-		     intel_fb->color_rb[i]->vbl_pending = intel_fb->vbl_seq;
-	       }
-	    }
-	    intel->driDrawable = driDrawPriv;
-	    intelWindowMoved(intel);
-	 }
-
-	 intel_draw_buffer(&intel->ctx, &intel_fb->Base);
-      }
-   }
-   else {
-      _mesa_make_current(NULL, NULL, NULL);
-   }
-
-   return GL_TRUE;
-}
-
-static void
-intelContendedLock(struct intel_context *intel, GLuint flags)
-{
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   __DRIscreenPrivate *sPriv = intel->driScreen;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-   drmI830Sarea *sarea = intel->sarea;
-
-   drmGetLock(intel->driFd, intel->hHWContext, flags);
-
-   if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - got contended lock\n", __progname);
-
-   /* If the window moved, may need to set a new cliprect now.
-    *
-    * NOTE: This releases and regains the hw lock, so all state
-    * checking must be done *after* this call:
-    */
-   if (dPriv)
-      DRI_VALIDATE_DRAWABLE_INFO(sPriv, dPriv);
-
-   if (sarea->width != intelScreen->width ||
-       sarea->height != intelScreen->height ||
-       sarea->rotation != intelScreen->current_rotation) {
-
-      intelUpdateScreenRotation(sPriv, sarea);
-   }
-
-   if (sarea->width != intel->width ||
-       sarea->height != intel->height ||
-       sarea->rotation != intel->current_rotation) {
-      int numClipRects = intel->numClipRects;
-
-      /*
-       * FIXME: Really only need to do this when drawing to a
-       * common back- or front buffer.
-       */
-
-      /*
-       * This will essentially drop the outstanding batchbuffer on the floor.
-       */
-      intel->numClipRects = 0;
-
-      if (intel->Fallback)
-	 _swrast_flush(&intel->ctx);
-
-      INTEL_FIREVERTICES(intel);
-
-      if (intel->batch->map != intel->batch->ptr)
-	 intel_batchbuffer_flush(intel->batch);
-
-      intel->numClipRects = numClipRects;
-
-      /* force window update */
-      intel->lastStamp = 0;
-
-      intel->width = sarea->width;
-      intel->height = sarea->height;
-      intel->current_rotation = sarea->rotation;
-   }
-
-   /* Drawable changed?
-    */
-   if (dPriv && intel->lastStamp != dPriv->lastStamp) {
-      intelWindowMoved(intel);
-      intel->lastStamp = dPriv->lastStamp;
-   }
-}
-
-
-
-/* Lock the hardware and validate our state.  
- */
-void LOCK_HARDWARE( struct intel_context *intel )
-{
-    char __ret=0;
-    struct intel_framebuffer *intel_fb = NULL;
-    struct intel_renderbuffer *intel_rb = NULL;
-    _glthread_LOCK_MUTEX(lockMutex);
-    assert(!intel->locked);
-
-    if (intel->driDrawable) {
-       intel_fb = intel->driDrawable->driverPrivate;
-
-       if (intel_fb)
-	  intel_rb =
-	     intel_get_renderbuffer(&intel_fb->Base,
-				    intel_fb->Base._ColorDrawBufferMask[0] ==
-				    BUFFER_BIT_FRONT_LEFT ? BUFFER_FRONT_LEFT :
-				    BUFFER_BACK_LEFT);
-    }
-
-    if (intel_rb && intel_fb->vblank_flags &&
-	!(intel_fb->vblank_flags & VBLANK_FLAG_NO_IRQ) &&
-	(intel_fb->vbl_waited - intel_rb->vbl_pending) > (1<<23)) {
-	drmVBlank vbl;
-
-	vbl.request.type = DRM_VBLANK_ABSOLUTE;
-
-	if ( intel_fb->vblank_flags & VBLANK_FLAG_SECONDARY ) {
-	    vbl.request.type |= DRM_VBLANK_SECONDARY;
-	}
-
-	vbl.request.sequence = intel_rb->vbl_pending;
-	drmWaitVBlank(intel->driFd, &vbl);
-	intel_fb->vbl_waited = vbl.reply.sequence;
-    }
-
-    DRM_CAS(intel->driHwLock, intel->hHWContext,
-        (DRM_LOCK_HELD|intel->hHWContext), __ret);
-
-    if (__ret)
-        intelContendedLock( intel, 0 );
-
-    if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - locked\n", __progname);
-
-    intel->locked = 1;
-}
-
-
-  /* Unlock the hardware using the global current context 
-   */
-void UNLOCK_HARDWARE( struct intel_context *intel )
-{
-   intel->locked = 0;
-
-   DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
-
-   _glthread_UNLOCK_MUTEX(lockMutex);
-
-   if (INTEL_DEBUG & DEBUG_LOCK)
-      _mesa_printf("%s - unlocked\n", __progname);
-} 
-
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.h b/src/mesa/drivers/dri/i915tex/intel_context.h
deleted file mode 100644
index 9d060eb866f..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_context.h
+++ /dev/null
@@ -1,504 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTELCONTEXT_INC
-#define INTELCONTEXT_INC
-
-
-
-#include "mtypes.h"
-#include "drm.h"
-#include "mm.h"
-#include "texmem.h"
-
-#include "intel_screen.h"
-#include "i915_drm.h"
-#include "i830_common.h"
-#include "tnl/t_vertex.h"
-
-#define TAG(x) intel##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-#define DV_PF_555  (1<<8)
-#define DV_PF_565  (2<<8)
-#define DV_PF_8888 (3<<8)
-
-struct intel_region;
-struct intel_context;
-struct _DriBufferObject;
-
-typedef void (*intel_tri_func) (struct intel_context *, intelVertex *,
-                                intelVertex *, intelVertex *);
-typedef void (*intel_line_func) (struct intel_context *, intelVertex *,
-                                 intelVertex *);
-typedef void (*intel_point_func) (struct intel_context *, intelVertex *);
-
-#define INTEL_FALLBACK_DRAW_BUFFER	 0x1
-#define INTEL_FALLBACK_READ_BUFFER	 0x2
-#define INTEL_FALLBACK_DEPTH_BUFFER      0x4
-#define INTEL_FALLBACK_STENCIL_BUFFER    0x8
-#define INTEL_FALLBACK_USER		 0x10
-#define INTEL_FALLBACK_RENDERMODE	 0x20
-
-extern void intelFallback(struct intel_context *intel, GLuint bit,
-                          GLboolean mode);
-#define FALLBACK( intel, bit, mode ) intelFallback( intel, bit, mode )
-
-
-#define INTEL_WRITE_PART  0x1
-#define INTEL_WRITE_FULL  0x2
-#define INTEL_READ        0x4
-
-struct intel_texture_object
-{
-   struct gl_texture_object base;       /* The "parent" object */
-
-   /* The mipmap tree must include at least these levels once
-    * validated:
-    */
-   GLuint firstLevel;
-   GLuint lastLevel;
-
-   /* Offset for firstLevel image:
-    */
-   GLuint textureOffset;
-
-   /* On validation any active images held in main memory or in other
-    * regions will be copied to this region and the old storage freed.
-    */
-   struct intel_mipmap_tree *mt;
-
-   GLboolean imageOverride;
-   GLint depthOverride;
-   GLuint pitchOverride;
-};
-
-
-
-struct intel_texture_image
-{
-   struct gl_texture_image base;
-
-   /* These aren't stored in gl_texture_image 
-    */
-   GLuint level;
-   GLuint face;
-
-   /* If intelImage->mt != NULL, image data is stored here.
-    * Else if intelImage->base.Data != NULL, image is stored there.
-    * Else there is no image data.
-    */
-   struct intel_mipmap_tree *mt;
-};
-
-
-#define INTEL_MAX_FIXUP 64
-
-struct intel_context
-{
-   GLcontext ctx;               /* the parent class */
-
-   struct
-   {
-      void (*destroy) (struct intel_context * intel);
-      void (*emit_state) (struct intel_context * intel);
-      void (*lost_hardware) (struct intel_context * intel);
-      void (*update_texture_state) (struct intel_context * intel);
-
-      void (*render_start) (struct intel_context * intel);
-      void (*set_draw_region) (struct intel_context * intel,
-                               struct intel_region * draw_region,
-                               struct intel_region * depth_region);
-
-        GLuint(*flush_cmd) (void);
-
-      void (*reduced_primitive_state) (struct intel_context * intel,
-                                       GLenum rprim);
-
-        GLboolean(*check_vertex_size) (struct intel_context * intel,
-                                       GLuint expected);
-
-
-      /* Metaops: 
-       */
-      void (*install_meta_state) (struct intel_context * intel);
-      void (*leave_meta_state) (struct intel_context * intel);
-
-      void (*meta_draw_region) (struct intel_context * intel,
-                                struct intel_region * draw_region,
-                                struct intel_region * depth_region);
-
-      void (*meta_color_mask) (struct intel_context * intel, GLboolean);
-
-      void (*meta_stencil_replace) (struct intel_context * intel,
-                                    GLuint mask, GLuint clear);
-
-      void (*meta_depth_replace) (struct intel_context * intel);
-
-      void (*meta_texture_blend_replace) (struct intel_context * intel);
-
-      void (*meta_no_stencil_write) (struct intel_context * intel);
-      void (*meta_no_depth_write) (struct intel_context * intel);
-      void (*meta_no_texture) (struct intel_context * intel);
-
-      void (*meta_import_pixel_state) (struct intel_context * intel);
-
-        GLboolean(*meta_tex_rect_source) (struct intel_context * intel,
-                                          struct _DriBufferObject * buffer,
-                                          GLuint offset,
-                                          GLuint pitch,
-                                          GLuint height,
-                                          GLenum format, GLenum type);
-      void (*rotate_window) (struct intel_context * intel,
-                             __DRIdrawablePrivate * dPriv, GLuint srcBuf);
-
-      void (*assert_not_dirty) (struct intel_context *intel);
-
-   } vtbl;
-
-   GLint refcount;
-   GLuint Fallback;
-   GLuint NewGLState;
-
-   struct _DriFenceObject *last_swap_fence;
-   struct _DriFenceObject *first_swap_fence;
-
-   struct intel_batchbuffer *batch;
-
-   struct
-   {
-      GLuint id;
-      GLuint primitive;
-      GLubyte *start_ptr;
-      void (*flush) (struct intel_context *);
-   } prim;
-
-   GLboolean locked;
-   char *prevLockFile;
-   int prevLockLine;
-
-   GLuint ClearColor565;
-   GLuint ClearColor8888;
-
-   /* Offsets of fields within the current vertex:
-    */
-   GLuint coloroffset;
-   GLuint specoffset;
-   GLuint wpos_offset;
-   GLuint wpos_size;
-
-   struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
-   GLuint vertex_attr_count;
-
-   GLfloat polygon_offset_scale;        /* dependent on depth_scale, bpp */
-
-   GLboolean hw_stipple;
-   GLboolean strict_conformance;
-
-   /* AGP memory buffer manager:
-    */
-   struct bufmgr *bm;
-
-
-   /* State for intelvb.c and inteltris.c.
-    */
-   GLuint RenderIndex;
-   GLmatrix ViewportMatrix;
-   GLenum render_primitive;
-   GLenum reduced_primitive;
-   GLuint vertex_size;
-   GLubyte *verts;              /* points to tnl->clipspace.vertex_buf */
-
-#if 0
-   struct intel_region *front_region;   /* XXX FBO: obsolete */
-   struct intel_region *rotated_region; /* XXX FBO: obsolete */
-   struct intel_region *back_region;    /* XXX FBO: obsolete */
-   struct intel_region *draw_region;    /* XXX FBO: rename to color_region */
-   struct intel_region *depth_region;   /**< currently bound depth/Z region */
-#endif
-
-   /* Fallback rasterization functions 
-    */
-   intel_point_func draw_point;
-   intel_line_func draw_line;
-   intel_tri_func draw_tri;
-
-   /* These refer to the current drawing buffer:
-    */
-   int drawX, drawY;            /**< origin of drawing area within region */
-   GLuint numClipRects;         /**< cliprects for drawing */
-   drm_clip_rect_t *pClipRects;
-   drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
-
-   int perf_boxes;
-
-   GLuint do_usleeps;
-   int do_irqs;
-   GLuint irqsEmitted;
-   drm_i915_irq_wait_t iw;
-
-   drm_context_t hHWContext;
-   drmLock *driHwLock;
-   int driFd;
-
-   __DRIdrawablePrivate *driDrawable;
-   __DRIscreenPrivate *driScreen;
-   intelScreenPrivate *intelScreen;
-   drmI830Sarea *sarea;
-
-   GLuint lastStamp;
-
-   /**
-    * Configuration cache
-    */
-   driOptionCache optionCache;
-
-  /* Rotation. Need to match that of the
-   * current screen.
-   */
-
-  int width;
-  int height;
-  int current_rotation;
-};
-
-/* These are functions now:
- */
-void LOCK_HARDWARE( struct intel_context *intel );
-void UNLOCK_HARDWARE( struct intel_context *intel );
-
-extern char *__progname;
-
-
-#define SUBPIXEL_X 0.125
-#define SUBPIXEL_Y 0.125
-
-#define INTEL_FIREVERTICES(intel)		\
-do {						\
-   if ((intel)->prim.flush)			\
-      (intel)->prim.flush(intel);		\
-} while (0)
-
-/* ================================================================
- * Color packing:
- */
-
-#define INTEL_PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define INTEL_PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define INTEL_PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define INTEL_PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-
-/* ================================================================
- * From linux kernel i386 header files, copes with odd sizes better
- * than COPY_DWORDS would:
- * XXX Put this in src/mesa/main/imports.h ???
- */
-#if defined(i386) || defined(__i386__)
-static INLINE void *
-__memcpy(void *to, const void *from, size_t n)
-{
-   int d0, d1, d2;
-   __asm__ __volatile__("rep ; movsl\n\t"
-                        "testb $2,%b4\n\t"
-                        "je 1f\n\t"
-                        "movsw\n"
-                        "1:\ttestb $1,%b4\n\t"
-                        "je 2f\n\t"
-                        "movsb\n" "2:":"=&c"(d0), "=&D"(d1), "=&S"(d2)
-                        :"0"(n / 4), "q"(n), "1"((long) to), "2"((long) from)
-                        :"memory");
-   return (to);
-}
-#else
-#define __memcpy(a,b,c) memcpy(a,b,c)
-#endif
-
-
-
-/* ================================================================
- * Debugging:
- */
-#define DO_DEBUG		0
-#if DO_DEBUG
-extern int INTEL_DEBUG;
-#else
-#define INTEL_DEBUG		0
-#endif
-
-#define DEBUG_TEXTURE	0x1
-#define DEBUG_STATE	0x2
-#define DEBUG_IOCTL	0x4
-#define DEBUG_BLIT	0x8
-#define DEBUG_MIPTREE   0x10
-#define DEBUG_FALLBACKS	0x20
-#define DEBUG_VERBOSE	0x40
-#define DEBUG_BATCH     0x80
-#define DEBUG_PIXEL     0x100
-#define DEBUG_BUFMGR    0x200
-#define DEBUG_REGION    0x400
-#define DEBUG_FBO       0x800
-#define DEBUG_LOCK      0x1000
-
-#define DBG(...)  do { if (INTEL_DEBUG & FILE_DEBUG_FLAG) _mesa_printf(__VA_ARGS__); } while(0)
-
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-#define PCI_CHIP_I855_GM		0x3582
-#define PCI_CHIP_I865_G			0x2572
-#define PCI_CHIP_I915_G			0x2582
-#define PCI_CHIP_I915_GM		0x2592
-#define PCI_CHIP_I945_G			0x2772
-#define PCI_CHIP_I945_GM		0x27A2
-#define PCI_CHIP_I945_GME		0x27AE
-#define PCI_CHIP_G33_G			0x29C2
-#define PCI_CHIP_Q35_G			0x29B2
-#define PCI_CHIP_Q33_G			0x29D2
-
-
-/* ================================================================
- * intel_context.c:
- */
-
-extern GLboolean intelInitContext(struct intel_context *intel,
-                                  const __GLcontextModes * mesaVis,
-                                  __DRIcontextPrivate * driContextPriv,
-                                  void *sharedContextPrivate,
-                                  struct dd_function_table *functions);
-
-extern void intelGetLock(struct intel_context *intel, GLuint flags);
-
-extern void intelFinish(GLcontext * ctx);
-extern void intelFlush(GLcontext * ctx);
-
-extern void intelInitDriverFunctions(struct dd_function_table *functions);
-
-
-/* ================================================================
- * intel_state.c:
- */
-extern void intelInitStateFuncs(struct dd_function_table *functions);
-
-#define COMPAREFUNC_ALWAYS		0
-#define COMPAREFUNC_NEVER		0x1
-#define COMPAREFUNC_LESS		0x2
-#define COMPAREFUNC_EQUAL		0x3
-#define COMPAREFUNC_LEQUAL		0x4
-#define COMPAREFUNC_GREATER		0x5
-#define COMPAREFUNC_NOTEQUAL		0x6
-#define COMPAREFUNC_GEQUAL		0x7
-
-#define STENCILOP_KEEP			0
-#define STENCILOP_ZERO			0x1
-#define STENCILOP_REPLACE		0x2
-#define STENCILOP_INCRSAT		0x3
-#define STENCILOP_DECRSAT		0x4
-#define STENCILOP_INCR			0x5
-#define STENCILOP_DECR			0x6
-#define STENCILOP_INVERT		0x7
-
-#define LOGICOP_CLEAR			0
-#define LOGICOP_NOR			0x1
-#define LOGICOP_AND_INV 		0x2
-#define LOGICOP_COPY_INV		0x3
-#define LOGICOP_AND_RVRSE		0x4
-#define LOGICOP_INV			0x5
-#define LOGICOP_XOR			0x6
-#define LOGICOP_NAND			0x7
-#define LOGICOP_AND			0x8
-#define LOGICOP_EQUIV			0x9
-#define LOGICOP_NOOP			0xa
-#define LOGICOP_OR_INV			0xb
-#define LOGICOP_COPY			0xc
-#define LOGICOP_OR_RVRSE		0xd
-#define LOGICOP_OR			0xe
-#define LOGICOP_SET			0xf
-
-#define BLENDFACT_ZERO			0x01
-#define BLENDFACT_ONE			0x02
-#define BLENDFACT_SRC_COLR		0x03
-#define BLENDFACT_INV_SRC_COLR 		0x04
-#define BLENDFACT_SRC_ALPHA		0x05
-#define BLENDFACT_INV_SRC_ALPHA 	0x06
-#define BLENDFACT_DST_ALPHA		0x07
-#define BLENDFACT_INV_DST_ALPHA 	0x08
-#define BLENDFACT_DST_COLR		0x09
-#define BLENDFACT_INV_DST_COLR		0x0a
-#define BLENDFACT_SRC_ALPHA_SATURATE	0x0b
-#define BLENDFACT_CONST_COLOR		0x0c
-#define BLENDFACT_INV_CONST_COLOR	0x0d
-#define BLENDFACT_CONST_ALPHA		0x0e
-#define BLENDFACT_INV_CONST_ALPHA	0x0f
-#define BLENDFACT_MASK          	0x0f
-
-#define MI_BATCH_BUFFER_END 	(0xA<<23)
-
-
-extern int intel_translate_compare_func(GLenum func);
-extern int intel_translate_stencil_op(GLenum op);
-extern int intel_translate_blend_factor(GLenum factor);
-extern int intel_translate_logic_op(GLenum opcode);
-
-
-/*======================================================================
- * Inline conversion functions.  
- * These are better-typed than the macros used previously:
- */
-static INLINE struct intel_context *
-intel_context(GLcontext * ctx)
-{
-   return (struct intel_context *) ctx;
-}
-
-static INLINE struct intel_texture_object *
-intel_texture_object(struct gl_texture_object *obj)
-{
-   return (struct intel_texture_object *) obj;
-}
-
-static INLINE struct intel_texture_image *
-intel_texture_image(struct gl_texture_image *img)
-{
-   return (struct intel_texture_image *) img;
-}
-
-extern struct intel_renderbuffer *intel_renderbuffer(struct gl_renderbuffer
-                                                     *rb);
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_ioctl.c b/src/mesa/drivers/dri/i915tex/intel_ioctl.c
deleted file mode 100644
index 3250c6b3a9c..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_ioctl.c
+++ /dev/null
@@ -1,138 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include <stdio.h>
-#include <unistd.h>
-#include <errno.h>
-#include <sched.h>
-
-#include "mtypes.h"
-#include "context.h"
-#include "swrast/swrast.h"
-
-#include "intel_context.h"
-#include "intel_ioctl.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_regions.h"
-#include "drm.h"
-
-#define FILE_DEBUG_FLAG DEBUG_IOCTL
-
-int
-intelEmitIrqLocked(struct intel_context *intel)
-{
-   drmI830IrqEmit ie;
-   int ret, seq;
-
-   assert(((*(int *) intel->driHwLock) & ~DRM_LOCK_CONT) ==
-          (DRM_LOCK_HELD | intel->hHWContext));
-
-   ie.irq_seq = &seq;
-
-   ret = drmCommandWriteRead(intel->driFd, DRM_I830_IRQ_EMIT,
-                             &ie, sizeof(ie));
-   if (ret) {
-      fprintf(stderr, "%s: drmI830IrqEmit: %d\n", __FUNCTION__, ret);
-      exit(1);
-   }
-
-   DBG("%s -->  %d\n", __FUNCTION__, seq);
-
-   return seq;
-}
-
-void
-intelWaitIrq(struct intel_context *intel, int seq)
-{
-   int ret;
-
-   DBG("%s %d\n", __FUNCTION__, seq);
-
-   intel->iw.irq_seq = seq;
-
-   do {
-      ret =
-         drmCommandWrite(intel->driFd, DRM_I830_IRQ_WAIT, &intel->iw,
-                         sizeof(intel->iw));
-   } while (ret == -EAGAIN || ret == -EINTR);
-
-   if (ret) {
-      fprintf(stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret);
-      exit(1);
-   }
-}
-
-
-void
-intel_batch_ioctl(struct intel_context *intel,
-                  GLuint start_offset,
-                  GLuint used,
-                  GLboolean ignore_cliprects, GLboolean allow_unlock)
-{
-   drmI830BatchBuffer batch;
-
-   assert(intel->locked);
-   assert(used);
-
-   DBG("%s used %d offset %x..%x ignore_cliprects %d\n",
-       __FUNCTION__,
-       used, start_offset, start_offset + used, ignore_cliprects);
-
-   /* Throw away non-effective packets.  Won't work once we have
-    * hardware contexts which would preserve statechanges beyond a
-    * single buffer.
-    */
-
-
-
-   batch.start = start_offset;
-   batch.used = used;
-   batch.cliprects = intel->pClipRects;
-   batch.num_cliprects = ignore_cliprects ? 0 : intel->numClipRects;
-   batch.DR1 = 0;
-   batch.DR4 = ((((GLuint) intel->drawX) & 0xffff) |
-                (((GLuint) intel->drawY) << 16));
-
-   DBG("%s: 0x%x..0x%x DR4: %x cliprects: %d\n",
-       __FUNCTION__,
-       batch.start,
-       batch.start + batch.used * 4, batch.DR4, batch.num_cliprects);
-
-   if (drmCommandWrite(intel->driFd, DRM_I830_BATCHBUFFER, &batch,
-                       sizeof(batch))) {
-      fprintf(stderr, "DRM_I830_BATCHBUFFER: %d\n", -errno);
-      UNLOCK_HARDWARE(intel);
-      exit(1);
-   }
-
-   /* FIXME: use hardware contexts to avoid 'losing' hardware after
-    * each buffer flush.
-    */
-   intel->vtbl.lost_hardware(intel);
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_ioctl.h b/src/mesa/drivers/dri/i915tex/intel_ioctl.h
deleted file mode 100644
index e8d07de893e..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_ioctl.h
+++ /dev/null
@@ -1,40 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTEL_IOCTL_H
-#define INTEL_IOCTL_H
-
-#include "intel_context.h"
-
-void intelWaitIrq(struct intel_context *intel, int seq);
-int intelEmitIrqLocked(struct intel_context *intel);
-
-void intel_batch_ioctl(struct intel_context *intel,
-                       GLuint start_offset,
-                       GLuint used,
-                       GLboolean ignore_cliprects, GLboolean allow_unlock);
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_pixel.c b/src/mesa/drivers/dri/i915tex/intel_pixel.c
deleted file mode 100644
index 9018e3daef4..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_pixel.c
+++ /dev/null
@@ -1,120 +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 portionsalloc
- * 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 "enums.h"
-#include "state.h"
-#include "swrast/swrast.h"
-
-#include "intel_context.h"
-#include "intel_pixel.h"
-#include "intel_regions.h"
-
-
-/**
- * Check if any fragment operations are in effect which might effect
- * glDraw/CopyPixels.
- */
-GLboolean
-intel_check_blit_fragment_ops(GLcontext * ctx)
-{
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   /* XXX Note: Scissor could be done with the blitter:
-    */
-   return !(ctx->_ImageTransferState ||
-            ctx->Color.AlphaEnabled ||
-            ctx->Depth.Test ||
-            ctx->Fog.Enabled ||
-            ctx->Scissor.Enabled ||
-            ctx->Stencil.Enabled ||
-            !ctx->Color.ColorMask[0] ||
-            !ctx->Color.ColorMask[1] ||
-            !ctx->Color.ColorMask[2] ||
-            !ctx->Color.ColorMask[3] ||
-            ctx->Texture._EnabledUnits || 
-	    ctx->FragmentProgram._Enabled ||
-	    ctx->Color.BlendEnabled);
-}
-
-
-GLboolean
-intel_check_meta_tex_fragment_ops(GLcontext * ctx)
-{
-   if (ctx->NewState)
-      _mesa_update_state(ctx);
-
-   /* Some of _ImageTransferState (scale, bias) could be done with
-    * fragment programs on i915.
-    */
-   return !(ctx->_ImageTransferState || ctx->Fog.Enabled ||     /* not done yet */
-            ctx->Texture._EnabledUnits || ctx->FragmentProgram._Enabled);
-}
-
-/* The intel_region struct doesn't really do enough to capture the
- * format of the pixels in the region.  For now this code assumes that
- * the region is a display surface and hence is either ARGB8888 or
- * RGB565.
- * XXX FBO: If we'd pass in the intel_renderbuffer instead of region, we'd
- * know the buffer's pixel format.
- *
- * \param format  as given to glDraw/ReadPixels
- * \param type  as given to glDraw/ReadPixels
- */
-GLboolean
-intel_check_blit_format(struct intel_region * region,
-                        GLenum format, GLenum type)
-{
-   if (region->cpp == 4 &&
-       (type == GL_UNSIGNED_INT_8_8_8_8_REV ||
-        type == GL_UNSIGNED_BYTE) && format == GL_BGRA) {
-      return GL_TRUE;
-   }
-
-   if (region->cpp == 2 &&
-       type == GL_UNSIGNED_SHORT_5_6_5_REV && format == GL_BGR) {
-      return GL_TRUE;
-   }
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s: bad format for blit (cpp %d, type %s format %s)\n",
-              __FUNCTION__, region->cpp,
-              _mesa_lookup_enum_by_nr(type), _mesa_lookup_enum_by_nr(format));
-
-   return GL_FALSE;
-}
-
-
-void
-intelInitPixelFuncs(struct dd_function_table *functions)
-{
-   functions->Accum = _swrast_Accum;
-   functions->Bitmap = _swrast_Bitmap;
-   functions->CopyPixels = intelCopyPixels;
-   functions->ReadPixels = intelReadPixels;
-   functions->DrawPixels = intelDrawPixels;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_reg.h b/src/mesa/drivers/dri/i915tex/intel_reg.h
deleted file mode 100644
index 7828ba6ad39..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_reg.h
+++ /dev/null
@@ -1,88 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#ifndef _INTEL_REG_H_
-#define _INTEL_REG_H_
-
-
-
-#define CMD_3D (0x3<<29)
-
-
-#define _3DPRIMITIVE         ((0x3<<29)|(0x1f<<24))
-#define PRIM_INDIRECT            (1<<23)
-#define PRIM_INLINE              (0<<23)
-#define PRIM_INDIRECT_SEQUENTIAL (0<<17)
-#define PRIM_INDIRECT_ELTS       (1<<17)
-
-#define PRIM3D_TRILIST		(0x0<<18)
-#define PRIM3D_TRISTRIP 	(0x1<<18)
-#define PRIM3D_TRISTRIP_RVRSE	(0x2<<18)
-#define PRIM3D_TRIFAN		(0x3<<18)
-#define PRIM3D_POLY		(0x4<<18)
-#define PRIM3D_LINELIST 	(0x5<<18)
-#define PRIM3D_LINESTRIP	(0x6<<18)
-#define PRIM3D_RECTLIST 	(0x7<<18)
-#define PRIM3D_POINTLIST	(0x8<<18)
-#define PRIM3D_DIB		(0x9<<18)
-#define PRIM3D_MASK		(0x1f<<18)
-
-#define I915PACKCOLOR4444(r,g,b,a) \
-  ((((a) & 0xf0) << 8) | (((r) & 0xf0) << 4) | ((g) & 0xf0) | ((b) >> 4))
-
-#define I915PACKCOLOR1555(r,g,b,a) \
-  ((((r) & 0xf8) << 7) | (((g) & 0xf8) << 2) | (((b) & 0xf8) >> 3) | \
-    ((a) ? 0x8000 : 0))
-
-#define I915PACKCOLOR565(r,g,b) \
-  ((((r) & 0xf8) << 8) | (((g) & 0xfc) << 3) | (((b) & 0xf8) >> 3))
-
-#define I915PACKCOLOR8888(r,g,b,a) \
-  ((a<<24) | (r<<16) | (g<<8) | b)
-
-
-
-
-#define BR00_BITBLT_CLIENT   0x40000000
-#define BR00_OP_COLOR_BLT    0x10000000
-#define BR00_OP_SRC_COPY_BLT 0x10C00000
-#define BR13_SOLID_PATTERN   0x80000000
-
-#define XY_COLOR_BLT_CMD		((2<<29)|(0x50<<22)|0x4)
-#define XY_COLOR_BLT_WRITE_ALPHA	(1<<21)
-#define XY_COLOR_BLT_WRITE_RGB		(1<<20)
-
-#define XY_SRC_COPY_BLT_CMD             ((2<<29)|(0x53<<22)|6)
-#define XY_SRC_COPY_BLT_WRITE_ALPHA     (1<<21)
-#define XY_SRC_COPY_BLT_WRITE_RGB       (1<<20)
-
-#define MI_WAIT_FOR_EVENT               ((0x3<<23))
-#define MI_WAIT_FOR_PLANE_B_FLIP        (1<<6)
-#define MI_WAIT_FOR_PLANE_A_FLIP        (1<<2)
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_render.c b/src/mesa/drivers/dri/i915tex/intel_render.c
deleted file mode 100644
index f9fa55051ea..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_render.c
+++ /dev/null
@@ -1,242 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-/*
- * Render unclipped vertex buffers by emitting vertices directly to
- * dma buffers.  Use strip/fan hardware acceleration where possible.
- *
- */
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "enums.h"
-
-#include "tnl/t_context.h"
-#include "tnl/t_vertex.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_tris.h"
-#include "intel_batchbuffer.h"
-#include "intel_reg.h"
-
-/*
- * Render unclipped vertex buffers by emitting vertices directly to
- * dma buffers.  Use strip/fan hardware primitives where possible.
- * Try to simulate missing primitives with indexed vertices.
- */
-#define HAVE_POINTS      0      /* Has it, but can't use because subpixel has to
-                                 * be adjusted for points on the INTEL/I845G
-                                 */
-#define HAVE_LINES       1
-#define HAVE_LINE_STRIPS 1
-#define HAVE_TRIANGLES   1
-#define HAVE_TRI_STRIPS  1
-#define HAVE_TRI_STRIP_1 0      /* has it, template can't use it yet */
-#define HAVE_TRI_FANS    1
-#define HAVE_POLYGONS    1
-#define HAVE_QUADS       0
-#define HAVE_QUAD_STRIPS 0
-
-#define HAVE_ELTS        0
-
-static GLuint hw_prim[GL_POLYGON + 1] = {
-   0,
-   PRIM3D_LINELIST,
-   PRIM3D_LINESTRIP,
-   PRIM3D_LINESTRIP,
-   PRIM3D_TRILIST,
-   PRIM3D_TRISTRIP,
-   PRIM3D_TRIFAN,
-   0,
-   0,
-   PRIM3D_POLY
-};
-
-static const GLenum reduced_prim[GL_POLYGON + 1] = {
-   GL_POINTS,
-   GL_LINES,
-   GL_LINES,
-   GL_LINES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES
-};
-
-static const int scale_prim[GL_POLYGON + 1] = {
-   0,                           /* fallback case */
-   1,
-   2,
-   2,
-   1,
-   3,
-   3,
-   0,                           /* fallback case */
-   0,                           /* fallback case */
-   3
-};
-
-
-static void
-intelDmaPrimitive(struct intel_context *intel, GLenum prim)
-{
-   if (0)
-      fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
-   INTEL_FIREVERTICES(intel);
-   intel->vtbl.reduced_primitive_state(intel, reduced_prim[prim]);
-   intelStartInlinePrimitive(intel, hw_prim[prim], INTEL_BATCH_CLIPRECTS);
-}
-
-
-#define LOCAL_VARS struct intel_context *intel = intel_context(ctx)
-#define INIT( prim ) 				\
-do {						\
-   intelDmaPrimitive( intel, prim );		\
-} while (0)
-
-#define FLUSH() INTEL_FIREVERTICES(intel)
-
-#define GET_SUBSEQUENT_VB_MAX_VERTS() \
-  ((intel->batch->size - 1500) / (intel->vertex_size*4))
-#define GET_CURRENT_VB_MAX_VERTS() GET_SUBSEQUENT_VB_MAX_VERTS()
-
-#define ALLOC_VERTS( nr ) \
-   intelExtendInlinePrimitive( intel, (nr) * intel->vertex_size )
-
-#define EMIT_VERTS( ctx, j, nr, buf ) \
-  _tnl_emit_vertices_to_buffer(ctx, j, (j)+(nr), buf )
-
-#define TAG(x) intel_##x
-#include "tnl_dd/t_dd_dmatmp.h"
-
-
-/**********************************************************************/
-/*                          Render pipeline stage                     */
-/**********************************************************************/
-
-/* Heuristic to choose between the two render paths:  
- */
-static GLboolean
-choose_render(struct intel_context *intel, struct vertex_buffer *VB)
-{
-   int vertsz = intel->vertex_size;
-   int cost_render = 0;
-   int cost_fallback = 0;
-   int nr_prims = 0;
-   int nr_rprims = 0;
-   int nr_rverts = 0;
-   int rprim = intel->reduced_primitive;
-   int i = 0;
-
-   for (i = 0; i < VB->PrimitiveCount; i++) {
-      GLuint prim = VB->Primitive[i].mode;
-      GLuint length = VB->Primitive[i].count;
-
-      if (!length)
-         continue;
-
-      nr_prims++;
-      nr_rverts += length * scale_prim[prim & PRIM_MODE_MASK];
-
-      if (reduced_prim[prim & PRIM_MODE_MASK] != rprim) {
-         nr_rprims++;
-         rprim = reduced_prim[prim & PRIM_MODE_MASK];
-      }
-   }
-
-   /* One point for each generated primitive:
-    */
-   cost_render = nr_prims;
-   cost_fallback = nr_rprims;
-
-   /* One point for every 1024 dwords (4k) of dma:
-    */
-   cost_render += (vertsz * i) / 1024;
-   cost_fallback += (vertsz * nr_rverts) / 1024;
-
-   if (0)
-      fprintf(stderr, "cost render: %d fallback: %d\n",
-              cost_render, cost_fallback);
-
-   if (cost_render > cost_fallback)
-      return GL_FALSE;
-
-   return GL_TRUE;
-}
-
-
-static GLboolean
-intel_run_render(GLcontext * ctx, struct tnl_pipeline_stage *stage)
-{
-   struct intel_context *intel = intel_context(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-   GLuint i;
-
-   /* Don't handle clipping or indexed vertices.
-    */
-   if (intel->RenderIndex != 0 ||
-       !intel_validate_render(ctx, VB) || !choose_render(intel, VB)) {
-      return GL_TRUE;
-   }
-
-   tnl->clipspace.new_inputs |= VERT_BIT_POS;
-
-   tnl->Driver.Render.Start(ctx);
-
-   for (i = 0; i < VB->PrimitiveCount; i++) {
-      GLuint prim = VB->Primitive[i].mode;
-      GLuint start = VB->Primitive[i].start;
-      GLuint length = VB->Primitive[i].count;
-
-      if (!length)
-         continue;
-
-      intel_render_tab_verts[prim & PRIM_MODE_MASK] (ctx, start,
-                                                     start + length, prim);
-   }
-
-   tnl->Driver.Render.Finish(ctx);
-
-   INTEL_FIREVERTICES(intel);
-
-   return GL_FALSE;             /* finished the pipe */
-}
-
-const struct tnl_pipeline_stage _intel_render_stage = {
-   "intel render",
-   NULL,
-   NULL,
-   NULL,
-   NULL,
-   intel_run_render             /* run */
-};
diff --git a/src/mesa/drivers/dri/i915tex/intel_rotate.c b/src/mesa/drivers/dri/i915tex/intel_rotate.c
deleted file mode 100644
index 12d98c4ad2f..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_rotate.c
+++ /dev/null
@@ -1,237 +0,0 @@
-
-/**
- * Routines for simple 2D->2D transformations for rotated, flipped screens.
- *
- * XXX This code is not intel-specific.  Move it into a common/utility
- * someday.
- */
-
-#include "intel_rotate.h"
-
-#define MIN2(A, B)   ( ((A) < (B)) ? (A) : (B) )
-
-#define ABS(A)  ( ((A) < 0) ? -(A) : (A) )
-
-
-void
-matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02, int m10, int m11, int m12)
-{
-   m->m00 = m00;
-   m->m01 = m01;
-   m->m02 = m02;
-   m->m10 = m10;
-   m->m11 = m11;
-   m->m12 = m12;
-}
-
-
-/*
- * Transform (x,y) coordinate by the given matrix.
- */
-void
-matrix23TransformCoordf(const struct matrix23 *m, float *x, float *y)
-{
-   const float x0 = *x;
-   const float y0 = *y;
-
-   *x = m->m00 * x0 + m->m01 * y0 + m->m02;
-   *y = m->m10 * x0 + m->m11 * y0 + m->m12;
-}
-
-
-void
-matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y)
-{
-   const int x0 = *x;
-   const int y0 = *y;
-
-   *x = m->m00 * x0 + m->m01 * y0 + m->m02;
-   *y = m->m10 * x0 + m->m11 * y0 + m->m12;
-}
-
-
-/*
- * Transform a width and height by the given matrix.
- * XXX this could be optimized quite a bit.
- */
-void
-matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist)
-{
-   int x0 = 0, y0 = 0;
-   int x1 = *xDist, y1 = 0;
-   int x2 = 0, y2 = *yDist;
-   matrix23TransformCoordi(m, &x0, &y0);
-   matrix23TransformCoordi(m, &x1, &y1);
-   matrix23TransformCoordi(m, &x2, &y2);
-
-   *xDist = (x1 - x0) + (x2 - x0);
-   *yDist = (y1 - y0) + (y2 - y0);
-
-   if (*xDist < 0)
-      *xDist = -*xDist;
-   if (*yDist < 0)
-      *yDist = -*yDist;
-}
-
-
-/**
- * Transform the rect defined by (x, y, w, h) by m.
- */
-void
-matrix23TransformRect(const struct matrix23 *m, int *x, int *y, int *w,
-                      int *h)
-{
-   int x0 = *x, y0 = *y;
-   int x1 = *x + *w, y1 = *y;
-   int x2 = *x + *w, y2 = *y + *h;
-   int x3 = *x, y3 = *y + *h;
-   matrix23TransformCoordi(m, &x0, &y0);
-   matrix23TransformCoordi(m, &x1, &y1);
-   matrix23TransformCoordi(m, &x2, &y2);
-   matrix23TransformCoordi(m, &x3, &y3);
-   *w = ABS(x1 - x0) + ABS(x2 - x1);
-   /**w = ABS(*w);*/
-   *h = ABS(y1 - y0) + ABS(y2 - y1);
-   /**h = ABS(*h);*/
-   *x = MIN2(x0, x1);
-   *x = MIN2(*x, x2);
-   *y = MIN2(y0, y1);
-   *y = MIN2(*y, y2);
-}
-
-
-/*
- * Make rotation matrix for width X height screen.
- */
-void
-matrix23Rotate(struct matrix23 *m, int width, int height, int angle)
-{
-   switch (angle) {
-   case 0:
-      matrix23Set(m, 1, 0, 0, 0, 1, 0);
-      break;
-   case 90:
-      matrix23Set(m, 0, 1, 0, -1, 0, width);
-      break;
-   case 180:
-      matrix23Set(m, -1, 0, width, 0, -1, height);
-      break;
-   case 270:
-      matrix23Set(m, 0, -1, height, 1, 0, 0);
-      break;
-   default:
-      /*abort() */ ;
-   }
-}
-
-
-/*
- * Make flip/reflection matrix for width X height screen.
- */
-void
-matrix23Flip(struct matrix23 *m, int width, int height, int xflip, int yflip)
-{
-   if (xflip) {
-      m->m00 = -1;
-      m->m01 = 0;
-      m->m02 = width - 1;
-   }
-   else {
-      m->m00 = 1;
-      m->m01 = 0;
-      m->m02 = 0;
-   }
-   if (yflip) {
-      m->m10 = 0;
-      m->m11 = -1;
-      m->m12 = height - 1;
-   }
-   else {
-      m->m10 = 0;
-      m->m11 = 1;
-      m->m12 = 0;
-   }
-}
-
-
-/*
- * result = a * b
- */
-void
-matrix23Multiply(struct matrix23 *result,
-                 const struct matrix23 *a, const struct matrix23 *b)
-{
-   result->m00 = a->m00 * b->m00 + a->m01 * b->m10;
-   result->m01 = a->m00 * b->m01 + a->m01 * b->m11;
-   result->m02 = a->m00 * b->m02 + a->m01 * b->m12 + a->m02;
-
-   result->m10 = a->m10 * b->m00 + a->m11 * b->m10;
-   result->m11 = a->m10 * b->m01 + a->m11 * b->m11;
-   result->m12 = a->m10 * b->m02 + a->m11 * b->m12 + a->m12;
-}
-
-
-#if 000
-
-#include <stdio.h>
-
-int
-main(int argc, char *argv[])
-{
-   int width = 500, height = 400;
-   int rot;
-   int fx = 0, fy = 0;          /* flip x and/or y ? */
-   int coords[4][2];
-
-   /* four corner coords to test with */
-   coords[0][0] = 0;
-   coords[0][1] = 0;
-   coords[1][0] = width - 1;
-   coords[1][1] = 0;
-   coords[2][0] = width - 1;
-   coords[2][1] = height - 1;
-   coords[3][0] = 0;
-   coords[3][1] = height - 1;
-
-
-   for (rot = 0; rot < 360; rot += 90) {
-      struct matrix23 rotate, flip, m;
-      int i;
-
-      printf("Rot %d, xFlip %d, yFlip %d:\n", rot, fx, fy);
-
-      /* make transformation matrix 'm' */
-      matrix23Rotate(&rotate, width, height, rot);
-      matrix23Flip(&flip, width, height, fx, fy);
-      matrix23Multiply(&m, &rotate, &flip);
-
-      /* xform four coords */
-      for (i = 0; i < 4; i++) {
-         int x = coords[i][0];
-         int y = coords[i][1];
-         matrix23TransformCoordi(&m, &x, &y);
-         printf("  %d, %d  -> %d %d\n", coords[i][0], coords[i][1], x, y);
-      }
-
-      /* xform width, height */
-      {
-         int x = width;
-         int y = height;
-         matrix23TransformDistance(&m, &x, &y);
-         printf("  %d x %d -> %d x %d\n", width, height, x, y);
-      }
-
-      /* xform rect */
-      {
-         int x = 50, y = 10, w = 200, h = 100;
-         matrix23TransformRect(&m, &x, &y, &w, &h);
-         printf("  %d,%d %d x %d -> %d, %d %d x %d\n", 50, 10, 200, 100,
-                x, y, w, h);
-      }
-
-   }
-
-   return 0;
-}
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_rotate.h b/src/mesa/drivers/dri/i915tex/intel_rotate.h
deleted file mode 100644
index 9c8802ca477..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_rotate.h
+++ /dev/null
@@ -1,39 +0,0 @@
-#ifndef INTEL_ROTATE_H
-#define INTEL_ROTATE_H 1
-
-struct matrix23
-{
-   int m00, m01, m02;
-   int m10, m11, m12;
-};
-
-
-
-extern void
-matrix23Set(struct matrix23 *m,
-            int m00, int m01, int m02, int m10, int m11, int m12);
-
-extern void matrix23TransformCoordi(const struct matrix23 *m, int *x, int *y);
-
-extern void
-matrix23TransformCoordf(const struct matrix23 *m, float *x, float *y);
-
-extern void
-matrix23TransformDistance(const struct matrix23 *m, int *xDist, int *yDist);
-
-extern void
-matrix23TransformRect(const struct matrix23 *m,
-                      int *x, int *y, int *w, int *h);
-
-extern void
-matrix23Rotate(struct matrix23 *m, int width, int height, int angle);
-
-extern void
-matrix23Flip(struct matrix23 *m, int width, int height, int xflip, int yflip);
-
-extern void
-matrix23Multiply(struct matrix23 *result,
-                 const struct matrix23 *a, const struct matrix23 *b);
-
-
-#endif /* INTEL_ROTATE_H */
diff --git a/src/mesa/drivers/dri/i915tex/intel_screen.c b/src/mesa/drivers/dri/i915tex/intel_screen.c
deleted file mode 100644
index 2acdead63d5..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_screen.c
+++ /dev/null
@@ -1,949 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "context.h"
-#include "framebuffer.h"
-#include "matrix.h"
-#include "renderbuffer.h"
-#include "simple_list.h"
-#include "utils.h"
-#include "vblank.h"
-#include "xmlpool.h"
-
-
-#include "intel_screen.h"
-
-#include "intel_buffers.h"
-#include "intel_tex.h"
-#include "intel_span.h"
-#include "intel_tris.h"
-#include "intel_ioctl.h"
-#include "intel_fbo.h"
-
-#include "i830_dri.h"
-#include "dri_bufpool.h"
-#include "intel_regions.h"
-#include "intel_batchbuffer.h"
-
-PUBLIC const char __driConfigOptions[] =
-   DRI_CONF_BEGIN DRI_CONF_SECTION_PERFORMANCE
-   DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS)
-   DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
-   DRI_CONF_SECTION_END DRI_CONF_SECTION_QUALITY
-   DRI_CONF_FORCE_S3TC_ENABLE(false)
-   DRI_CONF_ALLOW_LARGE_TEXTURES(1)
-   DRI_CONF_SECTION_END DRI_CONF_END;
-     const GLuint __driNConfigOptions = 4;
-
-#ifdef USE_NEW_INTERFACE
-     static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
-#endif /*USE_NEW_INTERFACE */
-
-     extern const struct dri_extension card_extensions[];
-
-/**
- * Map all the memory regions described by the screen.
- * \return GL_TRUE if success, GL_FALSE if error.
- */
-GLboolean
-intelMapScreenRegions(__DRIscreenPrivate * sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-
-   if (intelScreen->front.handle) {
-      if (drmMap(sPriv->fd,
-                 intelScreen->front.handle,
-                 intelScreen->front.size,
-                 (drmAddress *) & intelScreen->front.map) != 0) {
-         _mesa_problem(NULL, "drmMap(frontbuffer) failed!");
-         return GL_FALSE;
-      }
-   }
-   else {
-      _mesa_warning(NULL, "no front buffer handle in intelMapScreenRegions!");
-   }
-
-   if (0)
-      _mesa_printf("Back 0x%08x ", intelScreen->back.handle);
-   if (drmMap(sPriv->fd,
-              intelScreen->back.handle,
-              intelScreen->back.size,
-              (drmAddress *) & intelScreen->back.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-   if (intelScreen->third.handle) {
-      if (0)
-	 _mesa_printf("Third 0x%08x ", intelScreen->third.handle);
-      if (drmMap(sPriv->fd,
-		 intelScreen->third.handle,
-		 intelScreen->third.size,
-		 (drmAddress *) & intelScreen->third.map) != 0) {
-	 intelUnmapScreenRegions(intelScreen);
-	 return GL_FALSE;
-      }
-   }
-
-   if (0)
-      _mesa_printf("Depth 0x%08x ", intelScreen->depth.handle);
-   if (drmMap(sPriv->fd,
-              intelScreen->depth.handle,
-              intelScreen->depth.size,
-              (drmAddress *) & intelScreen->depth.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-
-#if 0
-   _mesa_printf("TEX 0x%08x ", intelScreen->tex.handle);
-   if (drmMap(sPriv->fd,
-              intelScreen->tex.handle,
-              intelScreen->tex.size,
-              (drmAddress *) & intelScreen->tex.map) != 0) {
-      intelUnmapScreenRegions(intelScreen);
-      return GL_FALSE;
-   }
-#endif
-   if (0)
-      printf("Mappings:  front: %p  back: %p  third: %p  depth: %p  tex: %p\n",
-             intelScreen->front.map,
-             intelScreen->back.map, intelScreen->third.map,
-             intelScreen->depth.map, intelScreen->tex.map);
-   return GL_TRUE;
-}
-
-
-static struct intel_region *
-intel_recreate_static(intelScreenPrivate *intelScreen,
-		      struct intel_region *region,
-		      GLuint mem_type,
-		      GLuint offset,
-		      void *virtual,
-		      GLuint cpp, GLuint pitch, GLuint height)
-{
-  if (region) {
-    intel_region_update_static(intelScreen, region, mem_type, offset,
-			       virtual, cpp, pitch, height);
-  } else {
-    region = intel_region_create_static(intelScreen, mem_type, offset,
-					virtual, cpp, pitch, height);
-  }
-  return region;
-}
-    
-
-/* Create intel_region structs to describe the static front,back,depth
- * buffers created by the xserver. 
- *
- * Although FBO's mean we now no longer use these as render targets in
- * all circumstances, they won't go away until the back and depth
- * buffers become private, and the front and rotated buffers will
- * remain even then.
- *
- * Note that these don't allocate video memory, just describe
- * allocations alread made by the X server.
- */
-static void
-intel_recreate_static_regions(intelScreenPrivate *intelScreen)
-{
-   intelScreen->front_region =
-      intel_recreate_static(intelScreen,
-			    intelScreen->front_region,
-			    DRM_BO_FLAG_MEM_TT,
-			    intelScreen->front.offset,
-			    intelScreen->front.map,
-			    intelScreen->cpp,
-			    intelScreen->front.pitch / intelScreen->cpp,
-			    intelScreen->height);
-
-   intelScreen->rotated_region =
-      intel_recreate_static(intelScreen,
-			    intelScreen->rotated_region,
-			    DRM_BO_FLAG_MEM_TT,
-			    intelScreen->rotated.offset,
-			    intelScreen->rotated.map,
-			    intelScreen->cpp,
-			    intelScreen->rotated.pitch /
-			    intelScreen->cpp, intelScreen->height);
-
-
-   intelScreen->back_region =
-      intel_recreate_static(intelScreen,
-			    intelScreen->back_region,
-			    DRM_BO_FLAG_MEM_TT,
-			    intelScreen->back.offset,
-			    intelScreen->back.map,
-			    intelScreen->cpp,
-			    intelScreen->back.pitch / intelScreen->cpp,
-			    intelScreen->height);
-
-   if (intelScreen->third.handle) {
-      intelScreen->third_region =
-	 intel_recreate_static(intelScreen,
-			       intelScreen->third_region,
-			       DRM_BO_FLAG_MEM_TT,
-			       intelScreen->third.offset,
-			       intelScreen->third.map,
-			       intelScreen->cpp,
-			       intelScreen->third.pitch / intelScreen->cpp,
-			       intelScreen->height);
-   }
-
-   /* Still assuming front.cpp == depth.cpp
-    */
-   intelScreen->depth_region =
-      intel_recreate_static(intelScreen,
-			    intelScreen->depth_region,
-			    DRM_BO_FLAG_MEM_TT,
-			    intelScreen->depth.offset,
-			    intelScreen->depth.map,
-			    intelScreen->cpp,
-			    intelScreen->depth.pitch / intelScreen->cpp,
-			    intelScreen->height);
-}
-
-/**
- * Use the information in the sarea to update the screen parameters
- * related to screen rotation. Needs to be called locked.
- */
-void
-intelUpdateScreenRotation(__DRIscreenPrivate * sPriv, drmI830Sarea * sarea)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-
-   intelUnmapScreenRegions(intelScreen);
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr, "ERROR Remapping screen regions!!!\n");
-   }
-   intel_recreate_static_regions(intelScreen);
-}
-
-
-void
-intelUnmapScreenRegions(intelScreenPrivate * intelScreen)
-{
-#define REALLY_UNMAP 1
-   if (intelScreen->front.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->front.map, intelScreen->front.size) != 0)
-         printf("drmUnmap front failed!\n");
-#endif
-      intelScreen->front.map = NULL;
-   }
-   if (intelScreen->back.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->back.map, intelScreen->back.size) != 0)
-         printf("drmUnmap back failed!\n");
-#endif
-      intelScreen->back.map = NULL;
-   }
-   if (intelScreen->third.map) {
-#if REALLY_UNMAP
-      if (drmUnmap(intelScreen->third.map, intelScreen->third.size) != 0)
-         printf("drmUnmap third failed!\n");
-#endif
-      intelScreen->third.map = NULL;
-   }
-   if (intelScreen->depth.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->depth.map, intelScreen->depth.size);
-      intelScreen->depth.map = NULL;
-#endif
-   }
-   if (intelScreen->tex.map) {
-#if REALLY_UNMAP
-      drmUnmap(intelScreen->tex.map, intelScreen->tex.size);
-      intelScreen->tex.map = NULL;
-#endif
-   }
-}
-
-
-static void
-intelPrintDRIInfo(intelScreenPrivate * intelScreen,
-                  __DRIscreenPrivate * sPriv, I830DRIPtr gDRIPriv)
-{
-   fprintf(stderr, "*** Front size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->front.size, intelScreen->front.offset,
-           intelScreen->front.pitch);
-   fprintf(stderr, "*** Back size:    0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->back.size, intelScreen->back.offset,
-           intelScreen->back.pitch);
-   fprintf(stderr, "*** Depth size:   0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->depth.size, intelScreen->depth.offset,
-           intelScreen->depth.pitch);
-   fprintf(stderr, "*** Rotated size: 0x%x  offset: 0x%x  pitch: %d\n",
-           intelScreen->rotated.size, intelScreen->rotated.offset,
-           intelScreen->rotated.pitch);
-   fprintf(stderr, "*** Texture size: 0x%x  offset: 0x%x\n",
-           intelScreen->tex.size, intelScreen->tex.offset);
-   fprintf(stderr, "*** Memory : 0x%x\n", gDRIPriv->mem);
-}
-
-
-static void
-intelPrintSAREA(const drmI830Sarea * sarea)
-{
-   fprintf(stderr, "SAREA: sarea width %d  height %d\n", sarea->width,
-           sarea->height);
-   fprintf(stderr, "SAREA: pitch: %d\n", sarea->pitch);
-   fprintf(stderr,
-           "SAREA: front offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->front_offset, sarea->front_size,
-           (unsigned) sarea->front_handle);
-   fprintf(stderr,
-           "SAREA: back  offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->back_offset, sarea->back_size,
-           (unsigned) sarea->back_handle);
-   fprintf(stderr, "SAREA: depth offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->depth_offset, sarea->depth_size,
-           (unsigned) sarea->depth_handle);
-   fprintf(stderr, "SAREA: tex   offset: 0x%08x  size: 0x%x  handle: 0x%x\n",
-           sarea->tex_offset, sarea->tex_size, (unsigned) sarea->tex_handle);
-   fprintf(stderr, "SAREA: rotation: %d\n", sarea->rotation);
-   fprintf(stderr,
-           "SAREA: rotated offset: 0x%08x  size: 0x%x\n",
-           sarea->rotated_offset, sarea->rotated_size);
-   fprintf(stderr, "SAREA: rotated pitch: %d\n", sarea->rotated_pitch);
-}
-
-
-/**
- * A number of the screen parameters are obtained/computed from
- * information in the SAREA.  This function updates those parameters.
- */
-void
-intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
-                           drmI830Sarea * sarea)
-{
-   intelScreen->width = sarea->width;
-   intelScreen->height = sarea->height;
-
-   intelScreen->front.offset = sarea->front_offset;
-   intelScreen->front.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->front.handle = sarea->front_handle;
-   intelScreen->front.size = sarea->front_size;
-
-   intelScreen->back.offset = sarea->back_offset;
-   intelScreen->back.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->back.handle = sarea->back_handle;
-   intelScreen->back.size = sarea->back_size;
-
-   if (intelScreen->driScrnPriv->ddxMinor >= 8) {
-      intelScreen->third.offset = sarea->third_offset;
-      intelScreen->third.pitch = sarea->pitch * intelScreen->cpp;
-      intelScreen->third.handle = sarea->third_handle;
-      intelScreen->third.size = sarea->third_size;
-   }
-
-   intelScreen->depth.offset = sarea->depth_offset;
-   intelScreen->depth.pitch = sarea->pitch * intelScreen->cpp;
-   intelScreen->depth.handle = sarea->depth_handle;
-   intelScreen->depth.size = sarea->depth_size;
-
-   intelScreen->tex.offset = sarea->tex_offset;
-   intelScreen->logTextureGranularity = sarea->log_tex_granularity;
-   intelScreen->tex.handle = sarea->tex_handle;
-   intelScreen->tex.size = sarea->tex_size;
-
-   intelScreen->rotated.offset = sarea->rotated_offset;
-   intelScreen->rotated.pitch = sarea->rotated_pitch * intelScreen->cpp;
-   intelScreen->rotated.size = sarea->rotated_size;
-   intelScreen->current_rotation = sarea->rotation;
-   matrix23Rotate(&intelScreen->rotMatrix,
-                  sarea->width, sarea->height, sarea->rotation);
-   intelScreen->rotatedWidth = sarea->virtualX;
-   intelScreen->rotatedHeight = sarea->virtualY;
-
-   if (0)
-      intelPrintSAREA(sarea);
-}
-
-GLboolean
-intelCreatePools(intelScreenPrivate *intelScreen)
-{
-   unsigned batchPoolSize = 1024*1024;
-   __DRIscreenPrivate * sPriv = intelScreen->driScrnPriv;
-
-   if (intelScreen->havePools)
-      return GL_TRUE;
-
-   batchPoolSize /= intelScreen->maxBatchSize;
-   intelScreen->regionPool = driDRMPoolInit(sPriv->fd);
-
-   if (!intelScreen->regionPool)
-      return GL_FALSE;
-
-   intelScreen->staticPool = driDRMStaticPoolInit(sPriv->fd);
-
-   if (!intelScreen->staticPool)
-      return GL_FALSE;
-
-   intelScreen->texPool = intelScreen->regionPool;
-
-   intelScreen->batchPool = driBatchPoolInit(sPriv->fd,
-                                             DRM_BO_FLAG_EXE |
-                                             DRM_BO_FLAG_MEM_TT |
-                                             DRM_BO_FLAG_MEM_LOCAL,
-                                             intelScreen->maxBatchSize, 
-					     batchPoolSize, 5);
-   if (!intelScreen->batchPool) {
-      fprintf(stderr, "Failed to initialize batch pool - possible incorrect agpgart installed\n");
-      return GL_FALSE;
-   }
-   
-   intel_recreate_static_regions(intelScreen);
-   intelScreen->havePools = GL_TRUE;
-
-   return GL_TRUE;
-}
-
-
-static GLboolean
-intelInitDriver(__DRIscreenPrivate * sPriv)
-{
-   intelScreenPrivate *intelScreen;
-   I830DRIPtr gDRIPriv = (I830DRIPtr) sPriv->pDevPriv;
-   drmI830Sarea *sarea;
-
-   PFNGLXSCRENABLEEXTENSIONPROC glx_enable_extension =
-      (PFNGLXSCRENABLEEXTENSIONPROC) (*dri_interface->
-                                      getProcAddress("glxEnableExtension"));
-   void *const psc = sPriv->psc->screenConfigs;
-
-   if (sPriv->devPrivSize != sizeof(I830DRIRec)) {
-      fprintf(stderr,
-              "\nERROR!  sizeof(I830DRIRec) does not match passed size from device driver\n");
-      return GL_FALSE;
-   }
-
-   /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
-   if (!intelScreen) {
-      fprintf(stderr, "\nERROR!  Allocating private area failed\n");
-      return GL_FALSE;
-   }
-   /* parse information in __driConfigOptions */
-   driParseOptionInfo(&intelScreen->optionCache,
-                      __driConfigOptions, __driNConfigOptions);
-
-   intelScreen->driScrnPriv = sPriv;
-   sPriv->private = (void *) intelScreen;
-   intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
-   sarea = (drmI830Sarea *)
-      (((GLubyte *) sPriv->pSAREA) + intelScreen->sarea_priv_offset);
-
-   intelScreen->maxBatchSize = BATCH_SZ;
-   intelScreen->deviceID = gDRIPriv->deviceID;
-   if (intelScreen->deviceID == PCI_CHIP_I865_G)
-      intelScreen->maxBatchSize = 4096;
-
-   intelScreen->mem = gDRIPriv->mem;
-   intelScreen->cpp = gDRIPriv->cpp;
-
-   switch (gDRIPriv->bitsPerPixel) {
-   case 16:
-      intelScreen->fbFormat = DV_PF_565;
-      break;
-   case 32:
-      intelScreen->fbFormat = DV_PF_8888;
-      break;
-   default:
-      exit(1);
-      break;
-   }
-
-   intelUpdateScreenFromSAREA(intelScreen, sarea);
-
-   if (!intelMapScreenRegions(sPriv)) {
-      fprintf(stderr, "\nERROR!  mapping regions\n");
-      _mesa_free(intelScreen);
-      sPriv->private = NULL;
-      return GL_FALSE;
-   }
-
-#if 0
-
-   /*
-    * FIXME: Remove this code and its references.
-    */
-
-   intelScreen->tex.offset = gDRIPriv->textureOffset;
-   intelScreen->logTextureGranularity = gDRIPriv->logTextureGranularity;
-   intelScreen->tex.handle = gDRIPriv->textures;
-   intelScreen->tex.size = gDRIPriv->textureSize;
-
-#else
-   intelScreen->tex.offset = 0;
-   intelScreen->logTextureGranularity = 0;
-   intelScreen->tex.handle = 0;
-   intelScreen->tex.size = 0;
-#endif
-
-   intelScreen->sarea_priv_offset = gDRIPriv->sarea_priv_offset;
-
-   if (0)
-      intelPrintDRIInfo(intelScreen, sPriv, gDRIPriv);
-
-   intelScreen->drmMinor = sPriv->drmMinor;
-
-   /* Determine if IRQs are active? */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_IRQ_ACTIVE;
-      gp.value = &intelScreen->irq_active;
-
-      ret = drmCommandWriteRead(sPriv->fd, DRM_I830_GETPARAM,
-                                &gp, sizeof(gp));
-      if (ret) {
-         fprintf(stderr, "drmI830GetParam: %d\n", ret);
-         return GL_FALSE;
-      }
-   }
-
-   /* Determine if batchbuffers are allowed */
-   {
-      int ret;
-      drmI830GetParam gp;
-
-      gp.param = I830_PARAM_ALLOW_BATCHBUFFER;
-      gp.value = &intelScreen->allow_batchbuffer;
-
-      ret = drmCommandWriteRead(sPriv->fd, DRM_I830_GETPARAM,
-                                &gp, sizeof(gp));
-      if (ret) {
-         fprintf(stderr, "drmI830GetParam: (%d) %d\n", gp.param, ret);
-         return GL_FALSE;
-      }
-   }
-
-   if (glx_enable_extension != NULL) {
-      (*glx_enable_extension) (psc, "GLX_SGI_swap_control");
-      (*glx_enable_extension) (psc, "GLX_SGI_video_sync");
-      (*glx_enable_extension) (psc, "GLX_MESA_swap_control");
-      (*glx_enable_extension) (psc, "GLX_MESA_swap_frame_usage");
-      (*glx_enable_extension) (psc, "GLX_SGI_make_current_read");
-   }
-
-   return GL_TRUE;
-}
-
-
-static void
-intelDestroyScreen(__DRIscreenPrivate * sPriv)
-{
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-
-   intelUnmapScreenRegions(intelScreen);
-
-   if (intelScreen->havePools) {
-      driPoolTakeDown(intelScreen->regionPool);
-      driPoolTakeDown(intelScreen->staticPool);
-      driPoolTakeDown(intelScreen->batchPool);
-   }
-   FREE(intelScreen);
-   sPriv->private = NULL;
-}
-
-
-/**
- * This is called when we need to set up GL rendering to a new X window.
- */
-static GLboolean
-intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
-                  __DRIdrawablePrivate * driDrawPriv,
-                  const __GLcontextModes * mesaVis, GLboolean isPixmap)
-{
-   intelScreenPrivate *screen = (intelScreenPrivate *) driScrnPriv->private;
-
-   if (isPixmap) {
-      return GL_FALSE;          /* not implemented */
-   }
-   else {
-      GLboolean swStencil = (mesaVis->stencilBits > 0 &&
-                             mesaVis->depthBits != 24);
-      GLenum rgbFormat = (mesaVis->redBits == 5 ? GL_RGB5 : GL_RGBA8);
-
-      struct intel_framebuffer *intel_fb = CALLOC_STRUCT(intel_framebuffer);
-
-      if (!intel_fb)
-	 return GL_FALSE;
-
-      _mesa_initialize_framebuffer(&intel_fb->Base, mesaVis);
-
-      /* setup the hardware-based renderbuffers */
-      {
-         intel_fb->color_rb[0]
-            = intel_create_renderbuffer(rgbFormat,
-                                        screen->width, screen->height,
-                                        screen->front.offset,
-                                        screen->front.pitch,
-                                        screen->cpp,
-                                        screen->front.map);
-         intel_set_span_functions(&intel_fb->color_rb[0]->Base);
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_FRONT_LEFT,
-				&intel_fb->color_rb[0]->Base);
-      }
-
-      if (mesaVis->doubleBufferMode) {
-         intel_fb->color_rb[1]
-            = intel_create_renderbuffer(rgbFormat,
-                                        screen->width, screen->height,
-                                        screen->back.offset,
-                                        screen->back.pitch,
-                                        screen->cpp,
-                                        screen->back.map);
-         intel_set_span_functions(&intel_fb->color_rb[1]->Base);
-         _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,
-					   screen->width, screen->height,
-					   screen->third.offset,
-					   screen->third.pitch,
-					   screen->cpp,
-					   screen->third.map);
-	    intel_set_span_functions(&intel_fb->color_rb[2]->Base);
-	    _mesa_reference_renderbuffer(&tmp_rb, &intel_fb->color_rb[2]->Base);
-	 }
-      }
-
-      if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
-         /* combined depth/stencil buffer */
-         struct intel_renderbuffer *depthStencilRb
-            = intel_create_renderbuffer(GL_DEPTH24_STENCIL8_EXT,
-                                        screen->width, screen->height,
-                                        screen->depth.offset,
-                                        screen->depth.pitch,
-                                        screen->cpp,    /* 4! */
-                                        screen->depth.map);
-         intel_set_span_functions(&depthStencilRb->Base);
-         /* note: bind RB to two attachment points */
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH,
-				&depthStencilRb->Base);
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_STENCIL,
-				&depthStencilRb->Base);
-      }
-      else if (mesaVis->depthBits == 16) {
-         /* just 16-bit depth buffer, no hw stencil */
-         struct intel_renderbuffer *depthRb
-            = intel_create_renderbuffer(GL_DEPTH_COMPONENT16,
-                                        screen->width, screen->height,
-                                        screen->depth.offset,
-                                        screen->depth.pitch,
-                                        screen->cpp,    /* 2! */
-                                        screen->depth.map);
-         intel_set_span_functions(&depthRb->Base);
-         _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_DEPTH, &depthRb->Base);
-      }
-
-      /* now add any/all software-based renderbuffers we may need */
-      _mesa_add_soft_renderbuffers(&intel_fb->Base,
-                                   GL_FALSE, /* never sw color */
-                                   GL_FALSE, /* never sw depth */
-                                   swStencil, mesaVis->accumRedBits > 0,
-                                   GL_FALSE, /* never sw alpha */
-                                   GL_FALSE  /* never sw aux */ );
-      driDrawPriv->driverPrivate = (void *) intel_fb;
-
-      return GL_TRUE;
-   }
-}
-
-static void
-intelDestroyBuffer(__DRIdrawablePrivate * driDrawPriv)
-{
-   _mesa_unreference_framebuffer((GLframebuffer **)(&(driDrawPriv->driverPrivate)));
-}
-
-
-/**
- * Get information about previous buffer swaps.
- */
-static int
-intelGetSwapInfo(__DRIdrawablePrivate * dPriv, __DRIswapInfo * sInfo)
-{
-   struct intel_framebuffer *intel_fb;
-
-   if ((dPriv == NULL) || (dPriv->driverPrivate == NULL)
-       || (sInfo == NULL)) {
-      return -1;
-   }
-
-   intel_fb = dPriv->driverPrivate;
-   sInfo->swap_count = intel_fb->swap_count;
-   sInfo->swap_ust = intel_fb->swap_ust;
-   sInfo->swap_missed_count = intel_fb->swap_missed_count;
-
-   sInfo->swap_missed_usage = (sInfo->swap_missed_count != 0)
-      ? driCalculateSwapUsage(dPriv, 0, intel_fb->swap_missed_ust)
-      : 0.0;
-
-   return 0;
-}
-
-
-/* There are probably better ways to do this, such as an
- * init-designated function to register chipids and createcontext
- * functions.
- */
-extern GLboolean i830CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
-                                   void *sharedContextPrivate);
-
-extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
-                                   __DRIcontextPrivate * driContextPriv,
-                                   void *sharedContextPrivate);
-
-
-
-
-static GLboolean
-intelCreateContext(const __GLcontextModes * mesaVis,
-                   __DRIcontextPrivate * driContextPriv,
-                   void *sharedContextPrivate)
-{
-   __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
-
-   switch (intelScreen->deviceID) {
-      /* Don't deal with i830 until texture work complete:
-       */
-   case PCI_CHIP_845_G:
-   case PCI_CHIP_I830_M:
-   case PCI_CHIP_I855_GM:
-   case PCI_CHIP_I865_G:
-      return i830CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
-
-   case PCI_CHIP_I915_G:
-   case PCI_CHIP_I915_GM:
-   case PCI_CHIP_I945_G:
-   case PCI_CHIP_I945_GM:
-   case PCI_CHIP_I945_GME:
-   case PCI_CHIP_G33_G:
-   case PCI_CHIP_Q35_G:
-   case PCI_CHIP_Q33_G:
-      return i915CreateContext(mesaVis, driContextPriv, sharedContextPrivate);
-
-   default:
-      fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
-      return GL_FALSE;
-   }
-}
-
-
-static const struct __DriverAPIRec intelAPI = {
-   .InitDriver = intelInitDriver,
-   .DestroyScreen = intelDestroyScreen,
-   .CreateContext = intelCreateContext,
-   .DestroyContext = intelDestroyContext,
-   .CreateBuffer = intelCreateBuffer,
-   .DestroyBuffer = intelDestroyBuffer,
-   .SwapBuffers = intelSwapBuffers,
-   .MakeCurrent = intelMakeCurrent,
-   .UnbindContext = intelUnbindContext,
-   .GetSwapInfo = intelGetSwapInfo,
-   .GetMSC = driGetMSC32,
-   .WaitForMSC = driWaitForMSC32,
-   .WaitForSBC = NULL,
-   .SwapBuffersMSC = NULL,
-   .CopySubBuffer = intelCopySubBuffer,
-   .setTexOffset = intelSetTexOffset,
-};
-
-
-static __GLcontextModes *
-intelFillInModes(unsigned pixel_bits, unsigned depth_bits,
-                 unsigned stencil_bits, GLboolean have_back_buffer)
-{
-   __GLcontextModes *modes;
-   __GLcontextModes *m;
-   unsigned num_modes;
-   unsigned depth_buffer_factor;
-   unsigned back_buffer_factor;
-   GLenum fb_format;
-   GLenum fb_type;
-
-   /* 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
-   };
-
-   u_int8_t depth_bits_array[3];
-   u_int8_t stencil_bits_array[3];
-
-
-   depth_bits_array[0] = 0;
-   depth_bits_array[1] = depth_bits;
-   depth_bits_array[2] = depth_bits;
-
-   /* Just like with the accumulation buffer, always provide some modes
-    * with a stencil buffer.  It will be a sw fallback, but some apps won't
-    * care about that.
-    */
-   stencil_bits_array[0] = 0;
-   stencil_bits_array[1] = 0;
-   if (depth_bits == 24)
-      stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   stencil_bits_array[2] = (stencil_bits == 0) ? 8 : stencil_bits;
-
-   depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 3 : 1;
-   back_buffer_factor = (have_back_buffer) ? 3 : 1;
-
-   num_modes = depth_buffer_factor * back_buffer_factor * 4;
-
-   if (pixel_bits == 16) {
-      fb_format = GL_RGB;
-      fb_type = GL_UNSIGNED_SHORT_5_6_5;
-   }
-   else {
-      fb_format = GL_BGRA;
-      fb_type = GL_UNSIGNED_INT_8_8_8_8_REV;
-   }
-
-   modes =
-      (*dri_interface->createContextModes) (num_modes,
-                                            sizeof(__GLcontextModes));
-   m = modes;
-   if (!driFillInModes(&m, fb_format, fb_type,
-                       depth_bits_array, stencil_bits_array,
-                       depth_buffer_factor, back_buffer_modes,
-                       back_buffer_factor, GLX_TRUE_COLOR)) {
-      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
-              __LINE__);
-      return NULL;
-   }
-   if (!driFillInModes(&m, fb_format, fb_type,
-                       depth_bits_array, stencil_bits_array,
-                       depth_buffer_factor, back_buffer_modes,
-                       back_buffer_factor, GLX_DIRECT_COLOR)) {
-      fprintf(stderr, "[%s:%u] Error creating FBConfig!\n", __func__,
-              __LINE__);
-      return NULL;
-   }
-
-   /* Mark the visual as slow if there are "fake" stencil bits.
-    */
-   for (m = modes; m != NULL; m = m->next) {
-      if ((m->stencilBits != 0) && (m->stencilBits != stencil_bits)) {
-         m->visualRating = GLX_SLOW_CONFIG;
-      }
-   }
-
-   return modes;
-}
-
-
-/**
- * This is the bootstrap function for the driver.  libGL supplies all of the
- * requisite information about the system, and the driver initializes itself.
- * This routine also fills in the linked list pointed to by \c driver_modes
- * with the \c __GLcontextModes that the driver can support for windows or
- * pbuffers.
- * 
- * \return A pointer to a \c __DRIscreenPrivate on success, or \c NULL on 
- *         failure.
- */
-PUBLIC void *
-__driCreateNewScreen_20050727(__DRInativeDisplay * dpy, int scrn,
-                              __DRIscreen * psc,
-                              const __GLcontextModes * modes,
-                              const __DRIversion * ddx_version,
-                              const __DRIversion * dri_version,
-                              const __DRIversion * drm_version,
-                              const __DRIframebuffer * frame_buffer,
-                              drmAddress pSAREA, int fd,
-                              int internal_api_version,
-                              const __DRIinterfaceMethods * interface,
-                              __GLcontextModes ** driver_modes)
-{
-   __DRIscreenPrivate *psp;
-   static const __DRIversion ddx_expected = { 1, 5, 0 };
-   static const __DRIversion dri_expected = { 4, 0, 0 };
-   static const __DRIversion drm_expected = { 1, 7, 0 };
-
-   dri_interface = interface;
-
-   if (!driCheckDriDdxDrmVersions2("i915",
-                                   dri_version, &dri_expected,
-                                   ddx_version, &ddx_expected,
-                                   drm_version, &drm_expected)) {
-      return NULL;
-   }
-
-   psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
-                                  ddx_version, dri_version, drm_version,
-                                  frame_buffer, pSAREA, fd,
-                                  internal_api_version, &intelAPI);
-
-   if (psp != NULL) {
-      I830DRIPtr dri_priv = (I830DRIPtr) psp->pDevPriv;
-      *driver_modes = intelFillInModes(dri_priv->cpp * 8,
-                                       (dri_priv->cpp == 2) ? 16 : 24,
-                                       (dri_priv->cpp == 2) ? 0 : 8, 1);
-
-      /* Calling driInitExtensions here, with a NULL context pointer, does not actually
-       * enable the extensions.  It just makes sure that all the dispatch offsets for all
-       * the extensions that *might* be enables are known.  This is needed because the
-       * dispatch offsets need to be known when _mesa_context_create is called, but we can't
-       * enable the extensions until we have a context pointer.
-       *
-       * Hello chicken.  Hello egg.  How are you two today?
-       */
-      driInitExtensions(NULL, card_extensions, GL_FALSE);
-   }
-
-   return (void *) psp;
-}
-
-struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen)
-{
-  /*
-   * This should probably change to have the screen allocate a dummy
-   * context at screen creation. For now just use the current context.
-   */
-
-  GET_CURRENT_CONTEXT(ctx);
-  if (ctx == NULL) {
-     _mesa_problem(NULL, "No current context in intelScreenContext\n");
-     return NULL;
-  }
-  return intel_context(ctx);
-}
-
diff --git a/src/mesa/drivers/dri/i915tex/intel_screen.h b/src/mesa/drivers/dri/i915tex/intel_screen.h
deleted file mode 100644
index bac43aadddf..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_screen.h
+++ /dev/null
@@ -1,137 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef _INTEL_INIT_H_
-#define _INTEL_INIT_H_
-
-#include <sys/time.h>
-#include "dri_util.h"
-#include "intel_rotate.h"
-#include "i830_common.h"
-#include "xmlconfig.h"
-#include "dri_bufpool.h"
-
-/* XXX: change name or eliminate to avoid conflict with "struct
- * intel_region"!!!
- */
-typedef struct
-{
-   drm_handle_t handle;
-   drmSize size;                /* region size in bytes */
-   char *map;                   /* memory map */
-   int offset;                  /* from start of video mem, in bytes */
-   int pitch;                   /* row stride, in bytes */
-} intelRegion;
-
-typedef struct
-{
-   intelRegion front;
-   intelRegion back;
-   intelRegion third;
-   intelRegion rotated;
-   intelRegion depth;
-   intelRegion tex;
-
-   struct intel_region *front_region;
-   struct intel_region *back_region;
-   struct intel_region *third_region;
-   struct intel_region *depth_region;
-   struct intel_region *rotated_region;
-
-   int deviceID;
-   int width;
-   int height;
-   int mem;                     /* unused */
-
-   int cpp;                     /* for front and back buffers */
-/*    int bitsPerPixel;   */
-   int fbFormat;                /* XXX FBO: this is obsolete - remove after i830 updates */
-
-   int logTextureGranularity;
-
-   __DRIscreenPrivate *driScrnPriv;
-   unsigned int sarea_priv_offset;
-
-   int drmMinor;
-
-   int irq_active;
-   int allow_batchbuffer;
-
-   struct matrix23 rotMatrix;
-
-   int current_rotation;        /* 0, 90, 180 or 270 */
-   int rotatedWidth, rotatedHeight;
-
-   /**
-   * Configuration cache with default values for all contexts
-   */
-   driOptionCache optionCache;
-   struct _DriBufferPool *batchPool;
-   struct _DriBufferPool *texPool;
-   struct _DriBufferPool *regionPool;
-   struct _DriBufferPool *staticPool;
-   unsigned int maxBatchSize;
-   GLboolean havePools;
-} intelScreenPrivate;
-
-
-
-extern GLboolean intelMapScreenRegions(__DRIscreenPrivate * sPriv);
-
-extern void intelUnmapScreenRegions(intelScreenPrivate * intelScreen);
-
-extern void
-intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
-                           drmI830Sarea * sarea);
-
-extern void intelDestroyContext(__DRIcontextPrivate * driContextPriv);
-
-extern GLboolean intelUnbindContext(__DRIcontextPrivate * driContextPriv);
-
-extern GLboolean
-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 _DriBufferPool *driBatchPoolInit(int fd, unsigned flags,
-                                               unsigned long bufSize,
-                                               unsigned numBufs,
-                                               unsigned checkDelayed);
-
-extern struct intel_context *intelScreenContext(intelScreenPrivate *intelScreen);
-
-extern void
-intelUpdateScreenRotation(__DRIscreenPrivate * sPriv, drmI830Sarea * sarea);
-extern GLboolean
-intelCreatePools(intelScreenPrivate *intelScreen);
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_span.c b/src/mesa/drivers/dri/i915tex/intel_span.c
deleted file mode 100644
index ab0874e4fd6..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_span.c
+++ /dev/null
@@ -1,409 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
-
-#include "intel_fbo.h"
-#include "intel_screen.h"
-#include "intel_span.h"
-#include "intel_regions.h"
-#include "intel_ioctl.h"
-#include "intel_tex.h"
-
-#include "swrast/swrast.h"
-
-/*
-  break intelWriteRGBASpan_ARGB8888
-*/
-
-#undef DBG
-#define DBG 0
-
-#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;	\
-   GLubyte *buf = (GLubyte *) irb->pfMap				\
-      + (intel->drawY * irb->pfPitch + intel->drawX) * irb->region->cpp;\
-   GLuint p;								\
-   assert(irb->pfMap);\
-   (void) p;
-
-/* XXX FBO: this is identical to the macro in spantmp2.h except we get
- * the cliprect info from the context, not the driDrawable.
- * Move this into spantmp2.h someday.
- */
-#define HW_CLIPLOOP()							\
-   do {									\
-      int _nc = intel->numClipRects;					\
-      while ( _nc-- ) {							\
-	 int minx = intel->pClipRects[_nc].x1 - intel->drawX;		\
-	 int miny = intel->pClipRects[_nc].y1 - intel->drawY;		\
-	 int maxx = intel->pClipRects[_nc].x2 - intel->drawX;		\
-	 int maxy = intel->pClipRects[_nc].y2 - intel->drawY;
-
-
-
-
-#define Y_FLIP(_y) ((_y) * yScale + yBias)
-
-#define HW_LOCK()
-
-#define HW_UNLOCK()
-
-/* 16 bit, RGB565 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_RGB
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
-
-#define TAG(x)    intel##x##_RGB565
-#define TAG2(x,y) intel##x##_RGB565##y
-#define GET_PTR(X,Y) (buf + ((Y) * irb->pfPitch + (X)) * 2)
-#include "spantmp2.h"
-
-/* 32 bit, ARGB8888 color spanline and pixel functions
- */
-#define SPANTMP_PIXEL_FMT GL_BGRA
-#define SPANTMP_PIXEL_TYPE GL_UNSIGNED_INT_8_8_8_8_REV
-
-#define TAG(x)    intel##x##_ARGB8888
-#define TAG2(x,y) intel##x##_ARGB8888##y
-#define GET_PTR(X,Y) (buf + ((Y) * irb->pfPitch + (X)) * 4)
-#include "spantmp2.h"
-
-
-#define LOCAL_DEPTH_VARS						\
-   struct intel_context *intel = intel_context(ctx);			\
-   struct intel_renderbuffer *irb = intel_renderbuffer(rb);		\
-   const GLuint pitch = irb->pfPitch/***XXX region->pitch*/; /* in pixels */ \
-   const GLint yScale = irb->RenderToTexture ? 1 : -1;			\
-   const GLint yBias = irb->RenderToTexture ? 0 : irb->Base.Height - 1;	\
-   char *buf = (char *) irb->pfMap/*XXX use region->map*/ +             \
-      (intel->drawY * pitch + intel->drawX) * irb->region->cpp;
-
-
-#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS
-
-/**
- ** 16-bit depthbuffer functions.
- **/
-#define WRITE_DEPTH( _x, _y, d ) \
-   ((GLushort *)buf)[(_x) + (_y) * pitch] = d;
-
-#define READ_DEPTH( d, _x, _y )	\
-   d = ((GLushort *)buf)[(_x) + (_y) * pitch];
-
-
-#define TAG(x) intel##x##_z16
-#include "depthtmp.h"
-
-
-/**
- ** 24/8-bit interleaved depth/stencil functions
- ** Note: we're actually reading back combined depth+stencil values.
- ** The wrappers in main/depthstencil.c are used to extract the depth
- ** and stencil values.
- **/
-/* Change ZZZS -> SZZZ */
-#define WRITE_DEPTH( _x, _y, d ) {				\
-   GLuint tmp = ((d) >> 8) | ((d) << 24);			\
-   ((GLuint *)buf)[(_x) + (_y) * pitch] = tmp;			\
-}
-
-/* Change SZZZ -> ZZZS */
-#define READ_DEPTH( d, _x, _y ) {				\
-   GLuint tmp = ((GLuint *)buf)[(_x) + (_y) * pitch];		\
-   d = (tmp << 8) | (tmp >> 24);				\
-}
-
-#define TAG(x) intel##x##_z24_s8
-#include "depthtmp.h"
-
-
-/**
- ** 8-bit stencil function (XXX FBO: This is obsolete)
- **/
-#define WRITE_STENCIL( _x, _y, d ) {				\
-   GLuint tmp = ((GLuint *)buf)[(_x) + (_y) * pitch];		\
-   tmp &= 0xffffff;						\
-   tmp |= ((d) << 24);						\
-   ((GLuint *) buf)[(_x) + (_y) * pitch] = tmp;			\
-}
-
-#define READ_STENCIL( d, _x, _y )				\
-   d = ((GLuint *)buf)[(_x) + (_y) * pitch] >> 24;
-
-#define TAG(x) intel##x##_z24_s8
-#include "stenciltmp.h"
-
-
-
-/**
- * Map or unmap all the renderbuffers which we may need during
- * software rendering.
- * XXX in the future, we could probably convey extra information to
- * reduce the number of mappings needed.  I.e. if doing a glReadPixels
- * from the depth buffer, we really only need one mapping.
- *
- * XXX Rewrite this function someday.
- * We can probably just loop over all the renderbuffer attachments,
- * map/unmap all of them, and not worry about the _ColorDrawBuffers
- * _ColorReadBuffer, _DepthBuffer or _StencilBuffer fields.
- */
-static void
-intel_map_unmap_buffers(struct intel_context *intel, GLboolean map)
-{
-   GLcontext *ctx = &intel->ctx;
-   GLuint i, j;
-   struct intel_renderbuffer *irb;
-
-   /* color draw buffers */
-   for (i = 0; i < ctx->Const.MaxDrawBuffers; i++) {
-      for (j = 0; j < ctx->DrawBuffer->_NumColorDrawBuffers[i]; j++) {
-         struct gl_renderbuffer *rb =
-            ctx->DrawBuffer->_ColorDrawBuffers[i][j];
-         irb = intel_renderbuffer(rb);
-         if (irb) {
-            /* this is a user-created intel_renderbuffer */
-            if (irb->region) {
-               if (map)
-                  intel_region_map(intel->intelScreen, irb->region);
-               else
-                  intel_region_unmap(intel->intelScreen, irb->region);
-            }
-            irb->pfMap = irb->region->map;
-            irb->pfPitch = irb->region->pitch;
-         }
-      }
-   }
-
-   /* check for render to textures */
-   for (i = 0; i < BUFFER_COUNT; i++) {
-      struct gl_renderbuffer_attachment *att =
-         ctx->DrawBuffer->Attachment + i;
-      struct gl_texture_object *tex = att->Texture;
-      if (tex) {
-         /* render to texture */
-         ASSERT(att->Renderbuffer);
-         if (map) {
-            struct gl_texture_image *texImg;
-            texImg = tex->Image[att->CubeMapFace][att->TextureLevel];
-            intel_tex_map_images(intel, intel_texture_object(tex));
-         }
-         else {
-            intel_tex_unmap_images(intel, intel_texture_object(tex));
-         }
-      }
-   }
-
-   /* color read buffers */
-   irb = intel_renderbuffer(ctx->ReadBuffer->_ColorReadBuffer);
-   if (irb && irb->region) {
-      if (map)
-         intel_region_map(intel->intelScreen, irb->region);
-      else
-         intel_region_unmap(intel->intelScreen, irb->region);
-      irb->pfMap = irb->region->map;
-      irb->pfPitch = irb->region->pitch;
-   }
-
-   /* Account for front/back color page flipping.
-    * The span routines use the pfMap and pfPitch fields which will
-    * swap the front/back region map/pitch if we're page flipped.
-    * Do this after mapping, above, so the map field is valid.
-    */
-#if 0
-   if (map && ctx->DrawBuffer->Name == 0) {
-      struct intel_renderbuffer *irbFront
-         = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_FRONT_LEFT);
-      struct intel_renderbuffer *irbBack
-         = intel_get_renderbuffer(ctx->DrawBuffer, BUFFER_BACK_LEFT);
-      if (irbBack) {
-         /* double buffered */
-         if (intel->sarea->pf_current_page == 0) {
-            irbFront->pfMap = irbFront->region->map;
-            irbFront->pfPitch = irbFront->region->pitch;
-            irbBack->pfMap = irbBack->region->map;
-            irbBack->pfPitch = irbBack->region->pitch;
-         }
-         else {
-            irbFront->pfMap = irbBack->region->map;
-            irbFront->pfPitch = irbBack->region->pitch;
-            irbBack->pfMap = irbFront->region->map;
-            irbBack->pfPitch = irbFront->region->pitch;
-         }
-      }
-   }
-#endif
-
-   /* depth buffer (Note wrapper!) */
-   if (ctx->DrawBuffer->_DepthBuffer) {
-      irb = intel_renderbuffer(ctx->DrawBuffer->_DepthBuffer->Wrapped);
-      if (irb && irb->region && irb->Base.Name != 0) {
-         if (map) {
-            intel_region_map(intel->intelScreen, irb->region);
-            irb->pfMap = irb->region->map;
-            irb->pfPitch = irb->region->pitch;
-         }
-         else {
-            intel_region_unmap(intel->intelScreen, irb->region);
-            irb->pfMap = NULL;
-            irb->pfPitch = 0;
-         }
-      }
-   }
-
-   /* stencil buffer (Note wrapper!) */
-   if (ctx->DrawBuffer->_StencilBuffer) {
-      irb = intel_renderbuffer(ctx->DrawBuffer->_StencilBuffer->Wrapped);
-      if (irb && irb->region && irb->Base.Name != 0) {
-         if (map) {
-            intel_region_map(intel->intelScreen, irb->region);
-            irb->pfMap = irb->region->map;
-            irb->pfPitch = irb->region->pitch;
-         }
-         else {
-            intel_region_unmap(intel->intelScreen, irb->region);
-            irb->pfMap = NULL;
-            irb->pfPitch = 0;
-         }
-      }
-   }
-}
-
-
-
-/**
- * Prepare for softare rendering.  Map current read/draw framebuffers'
- * renderbuffes and all currently bound texture objects.
- *
- * Old note: Moved locking out to get reasonable span performance.
- */
-void
-intelSpanRenderStart(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLuint i;
-
-   intelFinish(&intel->ctx);
-   LOCK_HARDWARE(intel);
-
-#if 0
-   /* Just map the framebuffer and all textures.  Bufmgr code will
-    * take care of waiting on the necessary fences:
-    */
-   intel_region_map(intel->intelScreen, intel->front_region);
-   intel_region_map(intel->intelScreen, intel->back_region);
-   intel_region_map(intel->intelScreen, intel->intelScreen->depth_region);
-#endif
-
-   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; 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));
-      }
-   }
-
-   intel_map_unmap_buffers(intel, GL_TRUE);
-}
-
-/**
- * Called when done softare rendering.  Unmap the buffers we mapped in
- * the above function.
- */
-void
-intelSpanRenderFinish(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-   GLuint i;
-
-   _swrast_flush(ctx);
-
-   /* Now unmap the framebuffer:
-    */
-#if 0
-   intel_region_unmap(intel, intel->front_region);
-   intel_region_unmap(intel, intel->back_region);
-   intel_region_unmap(intel, intel->intelScreen->depth_region);
-#endif
-
-   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; 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));
-      }
-   }
-
-   intel_map_unmap_buffers(intel, GL_FALSE);
-
-   UNLOCK_HARDWARE(intel);
-}
-
-
-void
-intelInitSpanFuncs(GLcontext * ctx)
-{
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart = intelSpanRenderStart;
-   swdd->SpanRenderFinish = intelSpanRenderFinish;
-}
-
-
-/**
- * Plug in appropriate span read/write functions for the given renderbuffer.
- * These are used for the software fallbacks.
- */
-void
-intel_set_span_functions(struct gl_renderbuffer *rb)
-{
-   if (rb->_ActualFormat == GL_RGB5) {
-      /* 565 RGB */
-      intelInitPointers_RGB565(rb);
-   }
-   else if (rb->_ActualFormat == GL_RGBA8) {
-      /* 8888 RGBA */
-      intelInitPointers_ARGB8888(rb);
-   }
-   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT16) {
-      intelInitDepthPointers_z16(rb);
-   }
-   else if (rb->_ActualFormat == GL_DEPTH_COMPONENT24 ||        /* XXX FBO remove */
-            rb->_ActualFormat == GL_DEPTH24_STENCIL8_EXT) {
-      intelInitDepthPointers_z24_s8(rb);
-   }
-   else if (rb->_ActualFormat == GL_STENCIL_INDEX8_EXT) {       /* XXX FBO remove */
-      intelInitStencilPointers_z24_s8(rb);
-   }
-   else {
-      _mesa_problem(NULL,
-                    "Unexpected _ActualFormat in intelSetSpanFunctions");
-   }
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_state.c b/src/mesa/drivers/dri/i915tex/intel_state.c
deleted file mode 100644
index 271511037e9..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_state.c
+++ /dev/null
@@ -1,269 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "colormac.h"
-#include "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_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][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] + SUBPIXEL_X;
-
-   m[MAT_SY] = v[MAT_SY] * yScale;
-   m[MAT_TY] = v[MAT_TY] * yScale + yBias + SUBPIXEL_Y;
-
-   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;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex.c b/src/mesa/drivers/dri/i915tex/intel_tex.c
deleted file mode 100644
index b08dee43bc1..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_tex.c
+++ /dev/null
@@ -1,192 +0,0 @@
-#include "texobj.h"
-#include "intel_context.h"
-#include "intel_mipmap_tree.h"
-#include "intel_tex.h"
-
-#define FILE_DEBUG_FLAG DEBUG_TEXTURE
-
-static GLboolean
-intelIsTextureResident(GLcontext * ctx, struct gl_texture_object *texObj)
-{
-#if 0
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   return
-      intelObj->mt &&
-      intelObj->mt->region &&
-      intel_is_region_resident(intel, intelObj->mt->region);
-#endif
-   return 1;
-}
-
-
-
-static struct gl_texture_image *
-intelNewTextureImage(GLcontext * ctx)
-{
-   DBG("%s\n", __FUNCTION__);
-   (void) ctx;
-   return (struct gl_texture_image *) CALLOC_STRUCT(intel_texture_image);
-}
-
-
-static struct gl_texture_object *
-intelNewTextureObject(GLcontext * ctx, GLuint name, GLenum target)
-{
-   struct intel_texture_object *obj = CALLOC_STRUCT(intel_texture_object);
-
-   DBG("%s\n", __FUNCTION__);
-   _mesa_initialize_texture_object(&obj->base, name, target);
-
-   return &obj->base;
-}
-
-static void 
-intelDeleteTextureObject(GLcontext *ctx,
-			 struct gl_texture_object *texObj)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_object *intelObj = intel_texture_object(texObj);
-
-   if (intelObj->mt)
-      intel_miptree_release(intel, &intelObj->mt);
-
-   _mesa_delete_texture_object(ctx, texObj);
-}
-
-
-static void
-intelFreeTextureImageData(GLcontext * ctx, struct gl_texture_image *texImage)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_texture_image *intelImage = intel_texture_image(texImage);
-
-   DBG("%s\n", __FUNCTION__);
-
-   if (intelImage->mt) {
-      intel_miptree_release(intel, &intelImage->mt);
-   }
-
-   if (texImage->Data) {
-      free(texImage->Data);
-      texImage->Data = NULL;
-   }
-}
-
-
-/* The system memcpy (at least on ubuntu 5.10) has problems copying
- * to agp (writecombined) memory from a source which isn't 64-byte
- * aligned - there is a 4x performance falloff.
- *
- * The x86 __memcpy is immune to this but is slightly slower
- * (10%-ish) than the system memcpy.
- *
- * The sse_memcpy seems to have a slight cliff at 64/32 bytes, but
- * isn't much faster than x86_memcpy for agp copies.
- * 
- * TODO: switch dynamically.
- */
-static void *
-do_memcpy(void *dest, const void *src, size_t n)
-{
-   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63)) {
-      return __memcpy(dest, src, n);
-   }
-   else
-      return memcpy(dest, src, n);
-}
-
-
-#if DO_DEBUG
-
-#ifndef __x86_64__
-static unsigned
-fastrdtsc(void)
-{
-   unsigned eax;
-   __asm__ volatile ("\t"
-                     "pushl  %%ebx\n\t"
-                     "cpuid\n\t" ".byte 0x0f, 0x31\n\t"
-                     "popl %%ebx\n":"=a" (eax)
-                     :"0"(0)
-                     :"ecx", "edx", "cc");
-
-   return eax;
-}
-#else
-static unsigned
-fastrdtsc(void)
-{
-   unsigned eax;
-   __asm__ volatile ("\t" "cpuid\n\t" ".byte 0x0f, 0x31\n\t":"=a" (eax)
-                     :"0"(0)
-                     :"ecx", "edx", "ebx", "cc");
-
-   return eax;
-}
-#endif
-
-static unsigned
-time_diff(unsigned t, unsigned t2)
-{
-   return ((t < t2) ? t2 - t : 0xFFFFFFFFU - (t - t2 - 1));
-}
-
-
-static void *
-timed_memcpy(void *dest, const void *src, size_t n)
-{
-   void *ret;
-   unsigned t1, t2;
-   double rate;
-
-   if ((((unsigned) src) & 63) || (((unsigned) dest) & 63))
-      _mesa_printf("Warning - non-aligned texture copy!\n");
-
-   t1 = fastrdtsc();
-   ret = do_memcpy(dest, src, n);
-   t2 = fastrdtsc();
-
-   rate = time_diff(t1, t2);
-   rate /= (double) n;
-   _mesa_printf("timed_memcpy: %u %u --> %f clocks/byte\n", t1, t2, rate);
-   return ret;
-}
-#endif /* DO_DEBUG */
-
-
-void
-intelInitTextureFuncs(struct dd_function_table *functions)
-{
-   functions->ChooseTextureFormat = intelChooseTextureFormat;
-   functions->TexImage1D = intelTexImage1D;
-   functions->TexImage2D = intelTexImage2D;
-   functions->TexImage3D = intelTexImage3D;
-   functions->TexSubImage1D = intelTexSubImage1D;
-   functions->TexSubImage2D = intelTexSubImage2D;
-   functions->TexSubImage3D = intelTexSubImage3D;
-   functions->CopyTexImage1D = intelCopyTexImage1D;
-   functions->CopyTexImage2D = intelCopyTexImage2D;
-   functions->CopyTexSubImage1D = intelCopyTexSubImage1D;
-   functions->CopyTexSubImage2D = intelCopyTexSubImage2D;
-   functions->GetTexImage = intelGetTexImage;
-
-   /* compressed texture functions */
-   functions->CompressedTexImage2D = intelCompressedTexImage2D;
-   functions->GetCompressedTexImage = intelGetCompressedTexImage;
-
-   functions->NewTextureObject = intelNewTextureObject;
-   functions->NewTextureImage = intelNewTextureImage;
-   functions->DeleteTexture = intelDeleteTextureObject;
-   functions->FreeTexImageData = intelFreeTextureImageData;
-   functions->UpdateTexturePalette = 0;
-   functions->IsTextureResident = intelIsTextureResident;
-
-#if DO_DEBUG
-   if (INTEL_DEBUG & DEBUG_BUFMGR)
-      functions->TextureMemCpy = timed_memcpy;
-   else
-#endif
-      functions->TextureMemCpy = do_memcpy;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex.h b/src/mesa/drivers/dri/i915tex/intel_tex.h
deleted file mode 100644
index b77d7a1d8af..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_tex.h
+++ /dev/null
@@ -1,151 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTELTEX_INC
-#define INTELTEX_INC
-
-#include "mtypes.h"
-#include "intel_context.h"
-#include "texmem.h"
-
-
-void intelInitTextureFuncs(struct dd_function_table *functions);
-
-const struct gl_texture_format *intelChooseTextureFormat(GLcontext * ctx,
-                                                         GLint internalFormat,
-                                                         GLenum format,
-                                                         GLenum type);
-
-
-void intelTexImage3D(GLcontext * ctx,
-                     GLenum target, GLint level,
-                     GLint internalFormat,
-                     GLint width, GLint height, GLint depth,
-                     GLint border,
-                     GLenum format, GLenum type, const void *pixels,
-                     const struct gl_pixelstore_attrib *packing,
-                     struct gl_texture_object *texObj,
-                     struct gl_texture_image *texImage);
-
-void intelTexSubImage3D(GLcontext * ctx,
-                        GLenum target,
-                        GLint level,
-                        GLint xoffset, GLint yoffset, GLint zoffset,
-                        GLsizei width, GLsizei height, GLsizei depth,
-                        GLenum format, GLenum type,
-                        const GLvoid * pixels,
-                        const struct gl_pixelstore_attrib *packing,
-                        struct gl_texture_object *texObj,
-                        struct gl_texture_image *texImage);
-
-void intelTexImage2D(GLcontext * ctx,
-                     GLenum target, GLint level,
-                     GLint internalFormat,
-                     GLint width, GLint height, GLint border,
-                     GLenum format, GLenum type, const void *pixels,
-                     const struct gl_pixelstore_attrib *packing,
-                     struct gl_texture_object *texObj,
-                     struct gl_texture_image *texImage);
-
-void intelTexSubImage2D(GLcontext * ctx,
-                        GLenum target,
-                        GLint level,
-                        GLint xoffset, GLint yoffset,
-                        GLsizei width, GLsizei height,
-                        GLenum format, GLenum type,
-                        const GLvoid * pixels,
-                        const struct gl_pixelstore_attrib *packing,
-                        struct gl_texture_object *texObj,
-                        struct gl_texture_image *texImage);
-
-void intelTexImage1D(GLcontext * ctx,
-                     GLenum target, GLint level,
-                     GLint internalFormat,
-                     GLint width, GLint border,
-                     GLenum format, GLenum type, const void *pixels,
-                     const struct gl_pixelstore_attrib *packing,
-                     struct gl_texture_object *texObj,
-                     struct gl_texture_image *texImage);
-
-void intelTexSubImage1D(GLcontext * ctx,
-                        GLenum target,
-                        GLint level,
-                        GLint xoffset,
-                        GLsizei width,
-                        GLenum format, GLenum type,
-                        const GLvoid * pixels,
-                        const struct gl_pixelstore_attrib *packing,
-                        struct gl_texture_object *texObj,
-                        struct gl_texture_image *texImage);
-
-void intelCopyTexImage1D(GLcontext * ctx, GLenum target, GLint level,
-                         GLenum internalFormat,
-                         GLint x, GLint y, GLsizei width, GLint border);
-
-void intelCopyTexImage2D(GLcontext * ctx, GLenum target, GLint level,
-                         GLenum internalFormat,
-                         GLint x, GLint y, GLsizei width, GLsizei height,
-                         GLint border);
-
-void intelCopyTexSubImage1D(GLcontext * ctx, GLenum target, GLint level,
-                            GLint xoffset, GLint x, GLint y, GLsizei width);
-
-void intelCopyTexSubImage2D(GLcontext * ctx, GLenum target, GLint level,
-                            GLint xoffset, GLint yoffset,
-                            GLint x, GLint y, GLsizei width, GLsizei height);
-
-void intelGetTexImage(GLcontext * ctx, GLenum target, GLint level,
-                      GLenum format, GLenum type, GLvoid * pixels,
-                      struct gl_texture_object *texObj,
-                      struct gl_texture_image *texImage);
-
-void intelCompressedTexImage2D( GLcontext *ctx, GLenum target, GLint level,
-				GLint internalFormat,
-				GLint width, GLint height, GLint border,
-				GLsizei imageSize, const GLvoid *data,
-				struct gl_texture_object *texObj,
-				struct gl_texture_image *texImage );
-
-void intelGetCompressedTexImage(GLcontext *ctx, GLenum target, GLint level,
-				GLvoid *pixels,
-				const struct gl_texture_object *texObj,
-				const struct gl_texture_image *texImage);
-
-void intelSetTexOffset(__DRIcontext *pDRICtx, GLint texname,
-		       unsigned long long offset, GLint depth, GLuint pitch);
-
-GLuint intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit);
-
-void intel_tex_map_images(struct intel_context *intel,
-                          struct intel_texture_object *intelObj);
-
-void intel_tex_unmap_images(struct intel_context *intel,
-                            struct intel_texture_object *intelObj);
-
-int intel_compressed_num_bytes(GLuint mesaFormat);
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/intel_tris.c b/src/mesa/drivers/dri/i915tex/intel_tris.c
deleted file mode 100644
index 5fe3d4561fc..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_tris.c
+++ /dev/null
@@ -1,1152 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "enums.h"
-#include "texobj.h"
-#include "state.h"
-#include "dd.h"
-
-#include "swrast/swrast.h"
-#include "swrast_setup/swrast_setup.h"
-#include "tnl/t_context.h"
-#include "tnl/t_pipeline.h"
-#include "tnl/t_vertex.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_tris.h"
-#include "intel_batchbuffer.h"
-#include "intel_buffers.h"
-#include "intel_reg.h"
-#include "intel_span.h"
-#include "intel_tex.h"
-
-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;
-}
-
-
-/* Emit a primitive referencing vertices in a vertex buffer.
- */
-void
-intelStartInlinePrimitive(struct intel_context *intel,
-                          GLuint prim, GLuint batch_flags)
-{
-   BATCH_LOCALS;
-
-   intel->vtbl.emit_state(intel);
-
-   /* Need to make sure at the very least that we don't wrap
-    * batchbuffers in BEGIN_BATCH below, otherwise the primitive will
-    * be emitted to a batchbuffer missing the required full-state
-    * preamble.
-    */
-   if (intel_batchbuffer_space(intel->batch) < 100) {
-      intel_batchbuffer_flush(intel->batch);
-      intel->vtbl.emit_state(intel);
-   }
-
-/*    _mesa_printf("%s *", __progname); */
-
-   intel_wait_flips(intel, batch_flags);
-
-   /* Emit a slot which will be filled with the inline primitive
-    * command later.
-    */
-   BEGIN_BATCH(2, batch_flags);
-   OUT_BATCH(0);
-
-   intel->prim.start_ptr = intel->batch->ptr;
-   intel->prim.primitive = prim;
-   intel->prim.flush = intel_flush_inline_primitive;
-
-   OUT_BATCH(0);
-   ADVANCE_BATCH();
-
-/*    _mesa_printf(">"); */
-}
-
-
-void
-intelWrapInlinePrimitive(struct intel_context *intel)
-{
-   GLuint prim = intel->prim.primitive;
-   GLuint batchflags = intel->batch->flags;
-
-   intel_flush_inline_primitive(intel);
-   intel_batchbuffer_flush(intel->batch);
-   intelStartInlinePrimitive(intel, prim, batchflags);  /* ??? */
-}
-
-GLuint *
-intelExtendInlinePrimitive(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)
-      intelWrapInlinePrimitive(intel);
-
-/*    _mesa_printf("."); */
-
-   intel->vtbl.assert_not_dirty(intel);
-
-   ptr = (GLuint *) intel->batch->ptr;
-   intel->batch->ptr += sz;
-
-   return ptr;
-}
-
-
-
-/***********************************************************************
- *                    Emit primitives as inline vertices               *
- ***********************************************************************/
-
-#ifdef __i386__
-#define COPY_DWORDS( j, vb, vertsize, v )			\
-do {								\
-   int __tmp;							\
-   __asm__ __volatile__( "rep ; movsl"				\
-			 : "=%c" (j), "=D" (vb), "=S" (__tmp)	\
-			 : "0" (vertsize),			\
-			 "D" ((long)vb),			\
-			 "S" ((long)v) );			\
-} while (0)
-#else
-#define COPY_DWORDS( j, vb, vertsize, v )	\
-do {						\
-   for ( j = 0 ; j < vertsize ; j++ ) {		\
-      vb[j] = ((GLuint *)v)[j];			\
-   }						\
-   vb += vertsize;				\
-} while (0)
-#endif
-
-static void
-intel_draw_quad(struct intel_context *intel,
-                intelVertexPtr v0,
-                intelVertexPtr v1, intelVertexPtr v2, intelVertexPtr v3)
-{
-   GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive(intel, 6 * vertsize);
-   int j;
-
-   COPY_DWORDS(j, vb, vertsize, v0);
-   COPY_DWORDS(j, vb, vertsize, v1);
-
-   /* If smooth shading, draw like a trifan which gives better
-    * rasterization.  Otherwise draw as two triangles with provoking
-    * vertex in third position as required for flat shading.
-    */
-   if (intel->ctx.Light.ShadeModel == GL_FLAT) {
-      COPY_DWORDS(j, vb, vertsize, v3);
-      COPY_DWORDS(j, vb, vertsize, v1);
-   }
-   else {
-      COPY_DWORDS(j, vb, vertsize, v2);
-      COPY_DWORDS(j, vb, vertsize, v0);
-   }
-
-   COPY_DWORDS(j, vb, vertsize, v2);
-   COPY_DWORDS(j, vb, vertsize, v3);
-}
-
-static void
-intel_draw_triangle(struct intel_context *intel,
-                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
-{
-   GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive(intel, 3 * vertsize);
-   int j;
-
-   COPY_DWORDS(j, vb, vertsize, v0);
-   COPY_DWORDS(j, vb, vertsize, v1);
-   COPY_DWORDS(j, vb, vertsize, v2);
-}
-
-
-static void
-intel_draw_line(struct intel_context *intel,
-                intelVertexPtr v0, intelVertexPtr v1)
-{
-   GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive(intel, 2 * vertsize);
-   int j;
-
-   COPY_DWORDS(j, vb, vertsize, v0);
-   COPY_DWORDS(j, vb, vertsize, v1);
-}
-
-
-static void
-intel_draw_point(struct intel_context *intel, intelVertexPtr v0)
-{
-   GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive(intel, vertsize);
-   int j;
-
-   /* Adjust for sub pixel position -- still required for conform. */
-   *(float *) &vb[0] = v0->v.x - 0.125;
-   *(float *) &vb[1] = v0->v.y - 0.125;
-   for (j = 2; j < vertsize; j++)
-      vb[j] = v0->ui[j];
-}
-
-
-
-/***********************************************************************
- *                Fixup for ARB_point_parameters                       *
- ***********************************************************************/
-
-/* Currently not working - VERT_ATTRIB_POINTSIZE isn't correctly
- * represented in the fragment program InputsRead field.
- */
-static void
-intel_atten_point(struct intel_context *intel, intelVertexPtr v0)
-{
-   GLcontext *ctx = &intel->ctx;
-   GLfloat psz[4], col[4], restore_psz, restore_alpha;
-
-   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-   _tnl_get_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
-
-   restore_psz = psz[0];
-   restore_alpha = col[3];
-
-   if (psz[0] >= ctx->Point.Threshold) {
-      psz[0] = MIN2(psz[0], ctx->Point.MaxSize);
-   }
-   else {
-      GLfloat dsize = psz[0] / ctx->Point.Threshold;
-      psz[0] = MAX2(ctx->Point.Threshold, ctx->Point.MinSize);
-      col[3] *= dsize * dsize;
-   }
-
-   if (psz[0] < 1.0)
-      psz[0] = 1.0;
-
-   if (restore_psz != psz[0] || restore_alpha != col[3]) {
-      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
-
-      intel_draw_point(intel, v0);
-
-      psz[0] = restore_psz;
-      col[3] = restore_alpha;
-
-      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_POINTSIZE, psz);
-      _tnl_set_attr(ctx, v0, _TNL_ATTRIB_COLOR0, col);
-   }
-   else
-      intel_draw_point(intel, v0);
-}
-
-
-
-
-
-/***********************************************************************
- *                Fixup for I915 WPOS texture coordinate                *
- ***********************************************************************/
-
-
-
-static void
-intel_wpos_triangle(struct intel_context *intel,
-                    intelVertexPtr v0, intelVertexPtr v1, intelVertexPtr v2)
-{
-   GLuint offset = intel->wpos_offset;
-   GLuint size = intel->wpos_size;
-
-   __memcpy(((char *) v0) + offset, v0, size);
-   __memcpy(((char *) v1) + offset, v1, size);
-   __memcpy(((char *) v2) + offset, v2, size);
-
-   intel_draw_triangle(intel, v0, v1, v2);
-}
-
-
-static void
-intel_wpos_line(struct intel_context *intel,
-                intelVertexPtr v0, intelVertexPtr v1)
-{
-   GLuint offset = intel->wpos_offset;
-   GLuint size = intel->wpos_size;
-
-   __memcpy(((char *) v0) + offset, v0, size);
-   __memcpy(((char *) v1) + offset, v1, size);
-
-   intel_draw_line(intel, v0, v1);
-}
-
-
-static void
-intel_wpos_point(struct intel_context *intel, intelVertexPtr v0)
-{
-   GLuint offset = intel->wpos_offset;
-   GLuint size = intel->wpos_size;
-
-   __memcpy(((char *) v0) + offset, v0, size);
-
-   intel_draw_point(intel, v0);
-}
-
-
-
-
-
-
-/***********************************************************************
- *          Macros for t_dd_tritmp.h to draw basic primitives          *
- ***********************************************************************/
-
-#define TRI( a, b, c )				\
-do { 						\
-   if (DO_FALLBACK)				\
-      intel->draw_tri( intel, a, b, c );	\
-   else						\
-      intel_draw_triangle( intel, a, b, c );	\
-} while (0)
-
-#define QUAD( a, b, c, d )			\
-do { 						\
-   if (DO_FALLBACK) {				\
-      intel->draw_tri( intel, a, b, d );	\
-      intel->draw_tri( intel, b, c, d );	\
-   } else					\
-      intel_draw_quad( intel, a, b, c, d );	\
-} while (0)
-
-#define LINE( v0, v1 )				\
-do { 						\
-   if (DO_FALLBACK)				\
-      intel->draw_line( intel, v0, v1 );	\
-   else						\
-      intel_draw_line( intel, v0, v1 );		\
-} while (0)
-
-#define POINT( v0 )				\
-do { 						\
-   if (DO_FALLBACK)				\
-      intel->draw_point( intel, v0 );		\
-   else						\
-      intel_draw_point( intel, v0 );		\
-} while (0)
-
-
-/***********************************************************************
- *              Build render functions from dd templates               *
- ***********************************************************************/
-
-#define INTEL_OFFSET_BIT 	0x01
-#define INTEL_TWOSIDE_BIT	0x02
-#define INTEL_UNFILLED_BIT	0x04
-#define INTEL_FALLBACK_BIT	0x08
-#define INTEL_MAX_TRIFUNC	0x10
-
-
-static struct
-{
-   tnl_points_func points;
-   tnl_line_func line;
-   tnl_triangle_func triangle;
-   tnl_quad_func quad;
-} rast_tab[INTEL_MAX_TRIFUNC];
-
-
-#define DO_FALLBACK (IND & INTEL_FALLBACK_BIT)
-#define DO_OFFSET   (IND & INTEL_OFFSET_BIT)
-#define DO_UNFILLED (IND & INTEL_UNFILLED_BIT)
-#define DO_TWOSIDE  (IND & INTEL_TWOSIDE_BIT)
-#define DO_FLAT      0
-#define DO_TRI       1
-#define DO_QUAD      1
-#define DO_LINE      1
-#define DO_POINTS    1
-#define DO_FULL_QUAD 1
-
-#define HAVE_RGBA         1
-#define HAVE_SPEC         1
-#define HAVE_BACK_COLORS  0
-#define HAVE_HW_FLATSHADE 1
-#define VERTEX            intelVertex
-#define TAB               rast_tab
-
-/* Only used to pull back colors into vertices (ie, we know color is
- * floating point).
- */
-#define INTEL_COLOR( dst, src )				\
-do {							\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[0], (src)[2]);	\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[1], (src)[1]);	\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[2], (src)[0]);	\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[3], (src)[3]);	\
-} while (0)
-
-#define INTEL_SPEC( dst, src )				\
-do {							\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[0], (src)[2]);	\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[1], (src)[1]);	\
-   UNCLAMPED_FLOAT_TO_UBYTE((dst)[2], (src)[0]);	\
-} while (0)
-
-
-#define DEPTH_SCALE intel->polygon_offset_scale
-#define UNFILLED_TRI unfilled_tri
-#define UNFILLED_QUAD unfilled_quad
-#define VERT_X(_v) _v->v.x
-#define VERT_Y(_v) _v->v.y
-#define VERT_Z(_v) _v->v.z
-#define AREA_IS_CCW( a ) (a > 0)
-#define GET_VERTEX(e) (intel->verts + (e * intel->vertex_size * sizeof(GLuint)))
-
-#define VERT_SET_RGBA( v, c )    if (coloroffset) INTEL_COLOR( v->ub4[coloroffset], c )
-#define VERT_COPY_RGBA( v0, v1 ) if (coloroffset) v0->ui[coloroffset] = v1->ui[coloroffset]
-#define VERT_SAVE_RGBA( idx )    if (coloroffset) color[idx] = v[idx]->ui[coloroffset]
-#define VERT_RESTORE_RGBA( idx ) if (coloroffset) v[idx]->ui[coloroffset] = color[idx]
-
-#define VERT_SET_SPEC( v, c )    if (specoffset) INTEL_SPEC( v->ub4[specoffset], c )
-#define VERT_COPY_SPEC( v0, v1 ) if (specoffset) COPY_3V(v0->ub4[specoffset], v1->ub4[specoffset])
-#define VERT_SAVE_SPEC( idx )    if (specoffset) spec[idx] = v[idx]->ui[specoffset]
-#define VERT_RESTORE_SPEC( idx ) if (specoffset) v[idx]->ui[specoffset] = spec[idx]
-
-#define LOCAL_VARS(n)							\
-   struct intel_context *intel = intel_context(ctx);				\
-   GLuint color[n], spec[n];						\
-   GLuint coloroffset = intel->coloroffset;		\
-   GLboolean specoffset = intel->specoffset;			\
-   (void) color; (void) spec; (void) coloroffset; (void) specoffset;
-
-
-/***********************************************************************
- *                Helpers for rendering unfilled primitives            *
- ***********************************************************************/
-
-static const GLuint hw_prim[GL_POLYGON + 1] = {
-   PRIM3D_POINTLIST,
-   PRIM3D_LINELIST,
-   PRIM3D_LINELIST,
-   PRIM3D_LINELIST,
-   PRIM3D_TRILIST,
-   PRIM3D_TRILIST,
-   PRIM3D_TRILIST,
-   PRIM3D_TRILIST,
-   PRIM3D_TRILIST,
-   PRIM3D_TRILIST
-};
-
-#define RASTERIZE(x) intelRasterPrimitive( ctx, x, hw_prim[x] )
-#define RENDER_PRIMITIVE intel->render_primitive
-#define TAG(x) x
-#define IND INTEL_FALLBACK_BIT
-#include "tnl_dd/t_dd_unfilled.h"
-#undef IND
-
-/***********************************************************************
- *                      Generate GL render functions                   *
- ***********************************************************************/
-
-#define IND (0)
-#define TAG(x) x
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_OFFSET_BIT)
-#define TAG(x) x##_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT)
-#define TAG(x) x##_twoside
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_OFFSET_BIT)
-#define TAG(x) x##_twoside_offset
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_UNFILLED_BIT)
-#define TAG(x) x##_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_OFFSET_BIT|INTEL_UNFILLED_BIT)
-#define TAG(x) x##_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_UNFILLED_BIT)
-#define TAG(x) x##_twoside_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_OFFSET_BIT|INTEL_UNFILLED_BIT)
-#define TAG(x) x##_twoside_offset_unfilled
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_FALLBACK_BIT)
-#define TAG(x) x##_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_OFFSET_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_offset_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_twoside_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_OFFSET_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_twoside_offset_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_UNFILLED_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_OFFSET_BIT|INTEL_UNFILLED_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_offset_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_UNFILLED_BIT|INTEL_FALLBACK_BIT)
-#define TAG(x) x##_twoside_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-#define IND (INTEL_TWOSIDE_BIT|INTEL_OFFSET_BIT|INTEL_UNFILLED_BIT| \
-	     INTEL_FALLBACK_BIT)
-#define TAG(x) x##_twoside_offset_unfilled_fallback
-#include "tnl_dd/t_dd_tritmp.h"
-
-
-static void
-init_rast_tab(void)
-{
-   init();
-   init_offset();
-   init_twoside();
-   init_twoside_offset();
-   init_unfilled();
-   init_offset_unfilled();
-   init_twoside_unfilled();
-   init_twoside_offset_unfilled();
-   init_fallback();
-   init_offset_fallback();
-   init_twoside_fallback();
-   init_twoside_offset_fallback();
-   init_unfilled_fallback();
-   init_offset_unfilled_fallback();
-   init_twoside_unfilled_fallback();
-   init_twoside_offset_unfilled_fallback();
-}
-
-
-/***********************************************************************
- *                    Rasterization fallback helpers                   *
- ***********************************************************************/
-
-
-/* This code is hit only when a mix of accelerated and unaccelerated
- * primitives are being drawn, and only for the unaccelerated
- * primitives.
- */
-static void
-intel_fallback_tri(struct intel_context *intel,
-                   intelVertex * v0, intelVertex * v1, intelVertex * v2)
-{
-   GLcontext *ctx = &intel->ctx;
-   SWvertex v[3];
-
-   if (0)
-      fprintf(stderr, "\n%s\n", __FUNCTION__);
-
-   INTEL_FIREVERTICES(intel);
-
-   _swsetup_Translate(ctx, v0, &v[0]);
-   _swsetup_Translate(ctx, v1, &v[1]);
-   _swsetup_Translate(ctx, v2, &v[2]);
-   intelSpanRenderStart(ctx);
-   _swrast_Triangle(ctx, &v[0], &v[1], &v[2]);
-   intelSpanRenderFinish(ctx);
-}
-
-
-static void
-intel_fallback_line(struct intel_context *intel,
-                    intelVertex * v0, intelVertex * v1)
-{
-   GLcontext *ctx = &intel->ctx;
-   SWvertex v[2];
-
-   if (0)
-      fprintf(stderr, "\n%s\n", __FUNCTION__);
-
-   INTEL_FIREVERTICES(intel);
-
-   _swsetup_Translate(ctx, v0, &v[0]);
-   _swsetup_Translate(ctx, v1, &v[1]);
-   intelSpanRenderStart(ctx);
-   _swrast_Line(ctx, &v[0], &v[1]);
-   intelSpanRenderFinish(ctx);
-}
-
-static void
-intel_fallback_point(struct intel_context *intel,
-		     intelVertex * v0)
-{
-   GLcontext *ctx = &intel->ctx;
-   SWvertex v[1];
-
-   if (0)
-      fprintf(stderr, "\n%s\n", __FUNCTION__);
-
-   INTEL_FIREVERTICES(intel);
-
-   _swsetup_Translate(ctx, v0, &v[0]);
-   intelSpanRenderStart(ctx);
-   _swrast_Point(ctx, &v[0]);
-   intelSpanRenderFinish(ctx);
-}
-
-
-/**********************************************************************/
-/*               Render unclipped begin/end objects                   */
-/**********************************************************************/
-
-#define IND 0
-#define V(x) (intelVertex *)(vertptr + ((x)*vertsize*sizeof(GLuint)))
-#define RENDER_POINTS( start, count )	\
-   for ( ; start < count ; start++) POINT( V(ELT(start)) );
-#define RENDER_LINE( v0, v1 )         LINE( V(v0), V(v1) )
-#define RENDER_TRI(  v0, v1, v2 )     TRI(  V(v0), V(v1), V(v2) )
-#define RENDER_QUAD( v0, v1, v2, v3 ) QUAD( V(v0), V(v1), V(v2), V(v3) )
-#define INIT(x) intelRenderPrimitive( ctx, x )
-#undef LOCAL_VARS
-#define LOCAL_VARS						\
-    struct intel_context *intel = intel_context(ctx);			\
-    GLubyte *vertptr = (GLubyte *)intel->verts;			\
-    const GLuint vertsize = intel->vertex_size;       	\
-    const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;	\
-    (void) elt;
-#define RESET_STIPPLE
-#define RESET_OCCLUSION
-#define PRESERVE_VB_DEFS
-#define ELT(x) x
-#define TAG(x) intel_##x##_verts
-#include "tnl/t_vb_rendertmp.h"
-#undef ELT
-#undef TAG
-#define TAG(x) intel_##x##_elts
-#define ELT(x) elt[x]
-#include "tnl/t_vb_rendertmp.h"
-
-/**********************************************************************/
-/*                   Render clipped primitives                        */
-/**********************************************************************/
-
-
-
-static void
-intelRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
-{
-   struct intel_context *intel = intel_context(ctx);
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
-   GLuint prim = intel->render_primitive;
-
-   /* Render the new vertices as an unclipped polygon.
-    */
-   {
-      GLuint *tmp = VB->Elts;
-      VB->Elts = (GLuint *) elts;
-      tnl->Driver.Render.PrimTabElts[GL_POLYGON] (ctx, 0, n,
-                                                  PRIM_BEGIN | PRIM_END);
-      VB->Elts = tmp;
-   }
-
-   /* Restore the render primitive
-    */
-   if (prim != GL_POLYGON)
-      tnl->Driver.Render.PrimitiveNotify(ctx, prim);
-}
-
-static void
-intelRenderClippedLine(GLcontext * ctx, GLuint ii, GLuint jj)
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-
-   tnl->Driver.Render.Line(ctx, ii, jj);
-}
-
-static void
-intelFastRenderClippedPoly(GLcontext * ctx, const GLuint * elts, GLuint n)
-{
-   struct intel_context *intel = intel_context(ctx);
-   const GLuint vertsize = intel->vertex_size;
-   GLuint *vb = intelExtendInlinePrimitive(intel, (n - 2) * 3 * vertsize);
-   GLubyte *vertptr = (GLubyte *) intel->verts;
-   const GLuint *start = (const GLuint *) V(elts[0]);
-   int i, j;
-
-   for (i = 2; i < n; i++) {
-      COPY_DWORDS(j, vb, vertsize, V(elts[i - 1]));
-      COPY_DWORDS(j, vb, vertsize, V(elts[i]));
-      COPY_DWORDS(j, vb, vertsize, start);
-   }
-}
-
-/**********************************************************************/
-/*                    Choose render functions                         */
-/**********************************************************************/
-
-
-
-
-#define ANY_FALLBACK_FLAGS (DD_LINE_STIPPLE | DD_TRI_STIPPLE | DD_POINT_ATTEN | DD_POINT_SMOOTH | DD_TRI_SMOOTH)
-#define ANY_RASTER_FLAGS (DD_TRI_LIGHT_TWOSIDE | DD_TRI_OFFSET | DD_TRI_UNFILLED)
-
-void
-intelChooseRenderState(GLcontext * ctx)
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct intel_context *intel = intel_context(ctx);
-   GLuint flags = ctx->_TriangleCaps;
-   const struct gl_fragment_program *fprog = ctx->FragmentProgram._Current;
-   GLboolean have_wpos = (fprog && (fprog->Base.InputsRead & FRAG_BIT_WPOS));
-   GLuint index = 0;
-
-   if (INTEL_DEBUG & DEBUG_STATE)
-      fprintf(stderr, "\n%s\n", __FUNCTION__);
-
-   if ((flags & (ANY_FALLBACK_FLAGS | ANY_RASTER_FLAGS)) || have_wpos) {
-
-      if (flags & ANY_RASTER_FLAGS) {
-         if (flags & DD_TRI_LIGHT_TWOSIDE)
-            index |= INTEL_TWOSIDE_BIT;
-         if (flags & DD_TRI_OFFSET)
-            index |= INTEL_OFFSET_BIT;
-         if (flags & DD_TRI_UNFILLED)
-            index |= INTEL_UNFILLED_BIT;
-      }
-
-      if (have_wpos) {
-         intel->draw_point = intel_wpos_point;
-         intel->draw_line = intel_wpos_line;
-         intel->draw_tri = intel_wpos_triangle;
-
-         /* Make sure these get called:
-          */
-         index |= INTEL_FALLBACK_BIT;
-      }
-      else {
-         intel->draw_point = intel_draw_point;
-         intel->draw_line = intel_draw_line;
-         intel->draw_tri = intel_draw_triangle;
-      }
-
-      /* Hook in fallbacks for specific primitives.
-       */
-      if (flags & ANY_FALLBACK_FLAGS) {
-         if (flags & DD_LINE_STIPPLE)
-            intel->draw_line = intel_fallback_line;
-
-         if ((flags & DD_TRI_STIPPLE) && !intel->hw_stipple)
-            intel->draw_tri = intel_fallback_tri;
-
-         if (flags & DD_TRI_SMOOTH) {
-	    if (intel->strict_conformance)
-	       intel->draw_tri = intel_fallback_tri;
-	 }
-
-         if (flags & DD_POINT_ATTEN) {
-	    if (0)
-	       intel->draw_point = intel_atten_point;
-	    else
-	       intel->draw_point = intel_fallback_point;
-	 }
-
-	 if (flags & DD_POINT_SMOOTH) {
-	    if (intel->strict_conformance)
-	       intel->draw_point = intel_fallback_point;
-	 }
-
-         index |= INTEL_FALLBACK_BIT;
-      }
-   }
-
-   if (intel->RenderIndex != index) {
-      intel->RenderIndex = index;
-
-      tnl->Driver.Render.Points = rast_tab[index].points;
-      tnl->Driver.Render.Line = rast_tab[index].line;
-      tnl->Driver.Render.Triangle = rast_tab[index].triangle;
-      tnl->Driver.Render.Quad = rast_tab[index].quad;
-
-      if (index == 0) {
-         tnl->Driver.Render.PrimTabVerts = intel_render_tab_verts;
-         tnl->Driver.Render.PrimTabElts = intel_render_tab_elts;
-         tnl->Driver.Render.ClippedLine = line; /* from tritmp.h */
-         tnl->Driver.Render.ClippedPolygon = intelFastRenderClippedPoly;
-      }
-      else {
-         tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts;
-         tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts;
-         tnl->Driver.Render.ClippedLine = intelRenderClippedLine;
-         tnl->Driver.Render.ClippedPolygon = intelRenderClippedPoly;
-      }
-   }
-}
-
-static const GLenum reduced_prim[GL_POLYGON + 1] = {
-   GL_POINTS,
-   GL_LINES,
-   GL_LINES,
-   GL_LINES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES,
-   GL_TRIANGLES
-};
-
-
-/**********************************************************************/
-/*                 High level hooks for t_vb_render.c                 */
-/**********************************************************************/
-
-
-
-
-static void
-intelRunPipeline(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   _mesa_lock_context_textures(ctx);
-   
-   if (ctx->NewState)
-      _mesa_update_state_locked(ctx);
-
-   if (intel->NewGLState) {
-      if (intel->NewGLState & _NEW_TEXTURE) {
-         intel->vtbl.update_texture_state(intel);
-      }
-
-      if (!intel->Fallback) {
-         if (intel->NewGLState & _INTEL_NEW_RENDERSTATE)
-            intelChooseRenderState(ctx);
-      }
-
-      intel->NewGLState = 0;
-   }
-
-   _tnl_run_pipeline(ctx);
-
-   _mesa_unlock_context_textures(ctx);
-}
-
-static void
-intelRenderStart(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   intel->vtbl.render_start(intel_context(ctx));
-   intel->vtbl.emit_state(intel);
-}
-
-static void
-intelRenderFinish(GLcontext * ctx)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   if (intel->RenderIndex & INTEL_FALLBACK_BIT)
-      _swrast_flush(ctx);
-
-   INTEL_FIREVERTICES(intel);
-}
-
-
-
-
- /* System to flush dma and emit state changes based on the rasterized
-  * primitive.
-  */
-static void
-intelRasterPrimitive(GLcontext * ctx, GLenum rprim, GLuint hwprim)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   if (0)
-      fprintf(stderr, "%s %s %x\n", __FUNCTION__,
-              _mesa_lookup_enum_by_nr(rprim), hwprim);
-
-   intel->vtbl.reduced_primitive_state(intel, rprim);
-
-   /* Start a new primitive.  Arrange to have it flushed later on.
-    */
-   if (hwprim != intel->prim.primitive) {
-      INTEL_FIREVERTICES(intel);
-
-      intelStartInlinePrimitive(intel, hwprim, INTEL_BATCH_CLIPRECTS);
-   }
-}
-
-
- /* 
-  */
-static void
-intelRenderPrimitive(GLcontext * ctx, GLenum prim)
-{
-   struct intel_context *intel = intel_context(ctx);
-
-   if (0)
-      fprintf(stderr, "%s %s\n", __FUNCTION__, _mesa_lookup_enum_by_nr(prim));
-
-   /* Let some clipping routines know which primitive they're dealing
-    * with.
-    */
-   intel->render_primitive = prim;
-
-   /* Shortcircuit this when called from t_dd_rendertmp.h for unfilled
-    * triangles.  The rasterized primitive will always be reset by
-    * lower level functions in that case, potentially pingponging the
-    * state:
-    */
-   if (reduced_prim[prim] == GL_TRIANGLES &&
-       (ctx->_TriangleCaps & DD_TRI_UNFILLED))
-      return;
-
-   /* Set some primitive-dependent state and Start? a new primitive.
-    */
-   intelRasterPrimitive(ctx, reduced_prim[prim], hw_prim[prim]);
-}
-
-
- /**********************************************************************/
- /*           Transition to/from hardware rasterization.               */
- /**********************************************************************/
-
-static char *fallbackStrings[] = {
-   [0] = "Draw buffer",
-   [1] = "Read buffer",
-   [2] = "Depth buffer",
-   [3] = "Stencil buffer",
-   [4] = "User disable",
-   [5] = "Render mode",
-
-   [12] = "Texture",
-   [13] = "Color mask",
-   [14] = "Stencil",
-   [15] = "Stipple",
-   [16] = "Program",
-   [17] = "Logic op",
-   [18] = "Smooth polygon",
-   [19] = "Smooth point",
-};
-
-
-static char *
-getFallbackString(GLuint bit)
-{
-   int i = 0;
-   while (bit > 1) {
-      i++;
-      bit >>= 1;
-   }
-   return fallbackStrings[i];
-}
-
-
-
-void
-intelFallback(struct intel_context *intel, GLuint bit, GLboolean mode)
-{
-   GLcontext *ctx = &intel->ctx;
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   GLuint oldfallback = intel->Fallback;
-
-   if (mode) {
-      intel->Fallback |= bit;
-      if (oldfallback == 0) {
-         intelFlush(ctx);
-         if (INTEL_DEBUG & DEBUG_FALLBACKS)
-            fprintf(stderr, "ENTER FALLBACK %x: %s\n",
-                    bit, getFallbackString(bit));
-         _swsetup_Wakeup(ctx);
-         intel->RenderIndex = ~0;
-      }
-   }
-   else {
-      intel->Fallback &= ~bit;
-      if (oldfallback == bit) {
-         _swrast_flush(ctx);
-         if (INTEL_DEBUG & DEBUG_FALLBACKS)
-            fprintf(stderr, "LEAVE FALLBACK %s\n", getFallbackString(bit));
-         tnl->Driver.Render.Start = intelRenderStart;
-         tnl->Driver.Render.PrimitiveNotify = intelRenderPrimitive;
-         tnl->Driver.Render.Finish = intelRenderFinish;
-         tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
-         tnl->Driver.Render.CopyPV = _tnl_copy_pv;
-         tnl->Driver.Render.Interp = _tnl_interp;
-
-         _tnl_invalidate_vertex_state(ctx, ~0);
-         _tnl_invalidate_vertices(ctx, ~0);
-         _tnl_install_attrs(ctx,
-                            intel->vertex_attrs,
-                            intel->vertex_attr_count,
-                            intel->ViewportMatrix.m, 0);
-
-         intel->NewGLState |= _INTEL_NEW_RENDERSTATE;
-      }
-   }
-}
-
-union fi
-{
-   GLfloat f;
-   GLint i;
-};
-
-
-/**********************************************************************/
-/*             Used only with the metaops callbacks.                  */
-/**********************************************************************/
-void
-intel_meta_draw_poly(struct intel_context *intel,
-                     GLuint n,
-                     GLfloat xy[][2],
-                     GLfloat z, GLuint color, GLfloat tex[][2])
-{
-   union fi *vb;
-   GLint i;
-
-   /* All 3d primitives should be emitted with INTEL_BATCH_CLIPRECTS,
-    * otherwise the drawing origin (DR4) might not be set correctly.
-    */
-   intelStartInlinePrimitive(intel, PRIM3D_TRIFAN, INTEL_BATCH_CLIPRECTS);
-   vb = (union fi *) intelExtendInlinePrimitive(intel, n * 6);
-
-   for (i = 0; i < n; i++) {
-      vb[0].f = xy[i][0];
-      vb[1].f = xy[i][1];
-      vb[2].f = z;
-      vb[3].i = color;
-      vb[4].f = tex[i][0];
-      vb[5].f = tex[i][1];
-      vb += 6;
-   }
-
-   INTEL_FIREVERTICES(intel);
-}
-
-void
-intel_meta_draw_quad(struct intel_context *intel,
-                     GLfloat x0, GLfloat x1,
-                     GLfloat y0, GLfloat y1,
-                     GLfloat z,
-                     GLuint color,
-                     GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1)
-{
-   GLfloat xy[4][2];
-   GLfloat tex[4][2];
-
-   xy[0][0] = x0;
-   xy[0][1] = y0;
-   xy[1][0] = x1;
-   xy[1][1] = y0;
-   xy[2][0] = x1;
-   xy[2][1] = y1;
-   xy[3][0] = x0;
-   xy[3][1] = y1;
-
-   tex[0][0] = s0;
-   tex[0][1] = t0;
-   tex[1][0] = s1;
-   tex[1][1] = t0;
-   tex[2][0] = s1;
-   tex[2][1] = t1;
-   tex[3][0] = s0;
-   tex[3][1] = t1;
-
-   intel_meta_draw_poly(intel, 4, xy, z, color, tex);
-}
-
-
-
-/**********************************************************************/
-/*                            Initialization.                         */
-/**********************************************************************/
-
-
-void
-intelInitTriFuncs(GLcontext * ctx)
-{
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   static int firsttime = 1;
-
-   if (firsttime) {
-      init_rast_tab();
-      firsttime = 0;
-   }
-
-   tnl->Driver.RunPipeline = intelRunPipeline;
-   tnl->Driver.Render.Start = intelRenderStart;
-   tnl->Driver.Render.Finish = intelRenderFinish;
-   tnl->Driver.Render.PrimitiveNotify = intelRenderPrimitive;
-   tnl->Driver.Render.ResetLineStipple = _swrast_ResetLineStipple;
-   tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
-   tnl->Driver.Render.CopyPV = _tnl_copy_pv;
-   tnl->Driver.Render.Interp = _tnl_interp;
-}
diff --git a/src/mesa/drivers/dri/i915tex/intel_tris.h b/src/mesa/drivers/dri/i915tex/intel_tris.h
deleted file mode 100644
index b7bae8cd3bc..00000000000
--- a/src/mesa/drivers/dri/i915tex/intel_tris.h
+++ /dev/null
@@ -1,69 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#ifndef INTELTRIS_INC
-#define INTELTRIS_INC
-
-#include "mtypes.h"
-
-
-
-#define _INTEL_NEW_RENDERSTATE (_DD_NEW_LINE_STIPPLE |		\
-			       _DD_NEW_TRI_UNFILLED |		\
-			       _DD_NEW_TRI_LIGHT_TWOSIDE |	\
-			       _DD_NEW_TRI_OFFSET |		\
-			       _DD_NEW_TRI_STIPPLE |		\
-			       _NEW_PROGRAM |		\
-			       _NEW_POLYGONSTIPPLE)
-
-extern void intelInitTriFuncs(GLcontext * ctx);
-
-extern void intelChooseRenderState(GLcontext * ctx);
-
-extern void intelStartInlinePrimitive(struct intel_context *intel,
-                                      GLuint prim, GLuint flags);
-extern void intelWrapInlinePrimitive(struct intel_context *intel);
-
-GLuint *intelExtendInlinePrimitive(struct intel_context *intel,
-                                   GLuint dwords);
-
-
-void intel_meta_draw_quad(struct intel_context *intel,
-                          GLfloat x0, GLfloat x1,
-                          GLfloat y0, GLfloat y1,
-                          GLfloat z,
-                          GLuint color,
-                          GLfloat s0, GLfloat s1, GLfloat t0, GLfloat t1);
-
-void intel_meta_draw_poly(struct intel_context *intel,
-                          GLuint n,
-                          GLfloat xy[][2],
-                          GLfloat z, GLuint color, GLfloat tex[][2]);
-
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/server/i830_common.h b/src/mesa/drivers/dri/i915tex/server/i830_common.h
deleted file mode 100644
index d4d58886ce0..00000000000
--- a/src/mesa/drivers/dri/i915tex/server/i830_common.h
+++ /dev/null
@@ -1,226 +0,0 @@
-/**************************************************************************
-
-Copyright 2001 VA Linux Systems Inc., Fremont, California.
-Copyright 2002 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
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ATI, VA LINUX SYSTEMS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_common.h,v 1.1 2002/09/11 00:29:32 dawes Exp $ */
-
-#ifndef _I830_COMMON_H_
-#define _I830_COMMON_H_
-
-
-#define I830_NR_TEX_REGIONS 255	/* maximum due to use of chars for next/prev */
-#define I830_LOG_MIN_TEX_REGION_SIZE 14
-
-
-/* Driver specific DRM command indices
- * NOTE: these are not OS specific, but they are driver specific
- */
-#define DRM_I830_INIT                     0x00
-#define DRM_I830_FLUSH                    0x01
-#define DRM_I830_FLIP                     0x02
-#define DRM_I830_BATCHBUFFER              0x03
-#define DRM_I830_IRQ_EMIT                 0x04
-#define DRM_I830_IRQ_WAIT                 0x05
-#define DRM_I830_GETPARAM                 0x06
-#define DRM_I830_SETPARAM                 0x07
-#define DRM_I830_ALLOC                    0x08
-#define DRM_I830_FREE                     0x09
-#define DRM_I830_INIT_HEAP                0x0a
-#define DRM_I830_CMDBUFFER                0x0b
-#define DRM_I830_DESTROY_HEAP             0x0c
-#define DRM_I830_SET_VBLANK_PIPE          0x0d
-#define DRM_I830_GET_VBLANK_PIPE          0x0e
-
-typedef struct {
-   enum {
-      I830_INIT_DMA = 0x01,
-      I830_CLEANUP_DMA = 0x02,
-      I830_RESUME_DMA = 0x03
-   } func;
-   unsigned int mmio_offset;
-   int sarea_priv_offset;
-   unsigned int ring_start;
-   unsigned int ring_end;
-   unsigned int ring_size;
-   unsigned int front_offset;
-   unsigned int back_offset;
-   unsigned int depth_offset;
-   unsigned int w;
-   unsigned int h;
-   unsigned int pitch;
-   unsigned int pitch_bits;
-   unsigned int back_pitch;
-   unsigned int depth_pitch;
-   unsigned int cpp;
-   unsigned int chipset;
-} drmI830Init;
-
-typedef struct {
-	drmTextureRegion texList[I830_NR_TEX_REGIONS+1];
-        int last_upload;	/* last time texture was uploaded */
-        int last_enqueue;	/* last time a buffer was enqueued */
-	int last_dispatch;	/* age of the most recently dispatched buffer */
-	int ctxOwner;		/* last context to upload state */
-	int texAge;
-        int pf_enabled;		/* is pageflipping allowed? */
-        int pf_active;               
-        int pf_current_page;	/* which buffer is being displayed? */
-        int perf_boxes;	        /* performance boxes to be displayed */   
-	int width, height;      /* screen size in pixels */
-
-	drm_handle_t front_handle;
-	int front_offset;
-	int front_size;
-
-	drm_handle_t back_handle;
-	int back_offset;
-	int back_size;
-
-	drm_handle_t depth_handle;
-	int depth_offset;
-	int depth_size;
-
-	drm_handle_t tex_handle;
-	int tex_offset;
-	int tex_size;
-	int log_tex_granularity;
-	int pitch;
-	int rotation;           /* 0, 90, 180 or 270 */
-	int rotated_offset;
-	int rotated_size;
-	int rotated_pitch;
-	int virtualX, virtualY;
-
-	unsigned int front_tiled;
-	unsigned int back_tiled;
-	unsigned int depth_tiled;
-	unsigned int rotated_tiled;
-	unsigned int rotated2_tiled;
-
-	int pipeA_x;
-	int pipeA_y;
-	int pipeA_w;
-	int pipeA_h;
-	int pipeB_x;
-	int pipeB_y;
-	int pipeB_w;
-	int pipeB_h;
-
-	/* Triple buffering */
-	drm_handle_t third_handle;
-	int third_offset;
-	int third_size;
-	unsigned int third_tiled;
-} drmI830Sarea;
-
-/* Flags for perf_boxes
- */
-#define I830_BOX_RING_EMPTY    0x1 /* populated by kernel */
-#define I830_BOX_FLIP          0x2 /* populated by kernel */
-#define I830_BOX_WAIT          0x4 /* populated by kernel & client */
-#define I830_BOX_TEXTURE_LOAD  0x8 /* populated by kernel */
-#define I830_BOX_LOST_CONTEXT  0x10 /* populated by client */
-
-
-typedef struct {
-   	int start;		/* agp offset */
-	int used;		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830BatchBuffer;
-
-typedef struct {
-   	char *buf;		/* agp offset */
-	int sz; 		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830CmdBuffer;
- 
-typedef struct {
-	int *irq_seq;
-} drmI830IrqEmit;
-
-typedef struct {
-	int irq_seq;
-} drmI830IrqWait;
-
-typedef struct {
-	int param;
-	int *value;
-} drmI830GetParam;
-
-#define I830_PARAM_IRQ_ACTIVE     1
-#define I830_PARAM_ALLOW_BATCHBUFFER   2 
-
-typedef struct {
-	int param;
-	int value;
-} drmI830SetParam;
-
-#define I830_SETPARAM_USE_MI_BATCHBUFFER_START  1
-#define I830_SETPARAM_TEX_LRU_LOG_GRANULARITY   2
-#define I830_SETPARAM_ALLOW_BATCHBUFFER         3
-
-
-/* A memory manager for regions of shared memory:
- */
-#define I830_MEM_REGION_AGP 1
-
-typedef struct {
-	int region;
-	int alignment;
-	int size;
-	int *region_offset;	/* offset from start of fb or agp */
-} drmI830MemAlloc;
-
-typedef struct {
-	int region;
-	int region_offset;
-} drmI830MemFree;
-
-typedef struct {
-	int region;
-	int size;
-	int start;	
-} drmI830MemInitHeap;
-
-typedef struct {
-	int region;
-} drmI830MemDestroyHeap;
-
-#define DRM_I830_VBLANK_PIPE_A  1
-#define DRM_I830_VBLANK_PIPE_B  2
-
-typedef struct {
-        int pipe;
-} drmI830VBlankPipe;
-
-#endif /* _I830_DRM_H_ */
diff --git a/src/mesa/drivers/dri/i915tex/server/i830_dri.h b/src/mesa/drivers/dri/i915tex/server/i830_dri.h
deleted file mode 100644
index c2a3af8cbf7..00000000000
--- a/src/mesa/drivers/dri/i915tex/server/i830_dri.h
+++ /dev/null
@@ -1,63 +0,0 @@
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_dri.h,v 1.4 2002/10/30 12:52:18 alanh Exp $ */
-
-#ifndef _I830_DRI_H
-#define _I830_DRI_H
-
-#include "xf86drm.h"
-#include "i830_common.h"
-
-#define I830_MAX_DRAWABLES 256
-
-#define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 7
-#define I830_PATCHLEVEL 2
-
-#define I830_REG_SIZE 0x80000
-
-typedef struct _I830DRIRec {
-   drm_handle_t regs;
-   drmSize regsSize;
-
-   drmSize unused1; /* backbufferSize */
-   drm_handle_t unused2; /* backbuffer */
-
-   drmSize unused3; /* depthbufferSize */
-   drm_handle_t unused4; /* depthbuffer */
-
-   drmSize unused5; /* rotatedSize */
-   drm_handle_t unused6; /* rotatedbuffer */
-
-   drm_handle_t unused7; /* textures */
-   int unused8; /* textureSize */
-
-   drm_handle_t unused9; /* agp_buffers */
-   drmSize unused10; /* agp_buf_size */
-
-   int deviceID;
-   int width;
-   int height;
-   int mem;
-   int cpp;
-   int bitsPerPixel;
-
-   int unused11[8]; /* was front/back/depth/rotated offset/pitch */
-   
-   int unused12; /* logTextureGranularity */
-   int unused13; /* textureOffset */
-
-   int irq;
-   int sarea_priv_offset;
-} I830DRIRec, *I830DRIPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830ConfigPrivRec, *I830ConfigPrivPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830DRIContextRec, *I830DRIContextPtr;
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/server/intel.h b/src/mesa/drivers/dri/i915tex/server/intel.h
deleted file mode 100644
index 6ea72499c1c..00000000000
--- a/src/mesa/drivers/dri/i915tex/server/intel.h
+++ /dev/null
@@ -1,331 +0,0 @@
-#ifndef _INTEL_H_
-#define _INTEL_H_
-
-#include "xf86drm.h"		/* drm_handle_t, etc */
-
-/* Intel */
-#ifndef PCI_CHIP_I810
-#define PCI_CHIP_I810              0x7121
-#define PCI_CHIP_I810_DC100        0x7123
-#define PCI_CHIP_I810_E            0x7125
-#define PCI_CHIP_I815              0x1132
-#define PCI_CHIP_I810_BRIDGE       0x7120
-#define PCI_CHIP_I810_DC100_BRIDGE 0x7122
-#define PCI_CHIP_I810_E_BRIDGE     0x7124
-#define PCI_CHIP_I815_BRIDGE       0x1130
-#endif
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-
-#ifndef PCI_CHIP_I855_GM
-#define PCI_CHIP_I855_GM	   0x3582
-#define PCI_CHIP_I855_GM_BRIDGE	   0x3580
-#endif
-
-#ifndef PCI_CHIP_I865_G
-#define PCI_CHIP_I865_G		   0x2572
-#define PCI_CHIP_I865_G_BRIDGE	   0x2570
-#endif
-
-#ifndef PCI_CHIP_I915_G
-#define PCI_CHIP_I915_G		   0x2582
-#define PCI_CHIP_I915_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I915_GM
-#define PCI_CHIP_I915_GM	   0x2592
-#define PCI_CHIP_I915_GM_BRIDGE	   0x2590
-#endif
-
-#ifndef PCI_CHIP_E7221_G
-#define PCI_CHIP_E7221_G	   0x258A
-/* Same as I915_G_BRIDGE */
-#define PCI_CHIP_E7221_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I945_G
-#define PCI_CHIP_I945_G        0x2772
-#define PCI_CHIP_I945_G_BRIDGE 0x2770
-#endif
-
-#ifndef PCI_CHIP_I945_GM
-#define PCI_CHIP_I945_GM        0x27A2
-#define PCI_CHIP_I945_GM_BRIDGE 0x27A0
-#endif
-
-#define IS_I810(pI810) (pI810->Chipset == PCI_CHIP_I810 ||	\
-			pI810->Chipset == PCI_CHIP_I810_DC100 || \
-			pI810->Chipset == PCI_CHIP_I810_E)
-#define IS_I815(pI810) (pI810->Chipset == PCI_CHIP_I815)
-#define IS_I830(pI810) (pI810->Chipset == PCI_CHIP_I830_M)
-#define IS_845G(pI810) (pI810->Chipset == PCI_CHIP_845_G)
-#define IS_I85X(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM)
-#define IS_I852(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I852_GM || pI810->variant == I852_GME))
-#define IS_I855(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I855_GM || pI810->variant == I855_GME))
-#define IS_I865G(pI810) (pI810->Chipset == PCI_CHIP_I865_G)
-
-#define IS_I915G(pI810) (pI810->Chipset == PCI_CHIP_I915_G || pI810->Chipset == PCI_CHIP_E7221_G)
-#define IS_I915GM(pI810) (pI810->Chipset == PCI_CHIP_I915_GM)
-#define IS_I945G(pI810) (pI810->Chipset == PCI_CHIP_I945_G)
-#define IS_I945GM(pI810) (pI810->Chipset == PCI_CHIP_I945_GM)
-#define IS_I9XX(pI810) (IS_I915G(pI810) || IS_I915GM(pI810) || IS_I945G(pI810) || IS_I945GM(pI810))
-
-#define IS_MOBILE(pI810) (IS_I830(pI810) || IS_I85X(pI810) || IS_I915GM(pI810) || IS_I945GM(pI810))
-
-#define I830_GMCH_CTRL		0x52
-
-#define I830_GMCH_MEM_MASK      0x1
-#define I830_GMCH_MEM_64M       0x1
-#define I830_GMCH_MEM_128M      0
-
-#define I830_GMCH_GMS_MASK			0x70
-#define I830_GMCH_GMS_DISABLED		0x00
-#define I830_GMCH_GMS_LOCAL			0x10
-#define I830_GMCH_GMS_STOLEN_512	0x20
-#define I830_GMCH_GMS_STOLEN_1024	0x30
-#define I830_GMCH_GMS_STOLEN_8192	0x40
-
-#define I855_GMCH_GMS_MASK			(0x7 << 4)
-#define I855_GMCH_GMS_DISABLED			0x00
-#define I855_GMCH_GMS_STOLEN_1M			(0x1 << 4)
-#define I855_GMCH_GMS_STOLEN_4M			(0x2 << 4)
-#define I855_GMCH_GMS_STOLEN_8M			(0x3 << 4)
-#define I855_GMCH_GMS_STOLEN_16M		(0x4 << 4)
-#define I855_GMCH_GMS_STOLEN_32M		(0x5 << 4)
-#define I915G_GMCH_GMS_STOLEN_48M		(0x6 << 4)
-#define I915G_GMCH_GMS_STOLEN_64M		(0x7 << 4)
-
-typedef unsigned char Bool;
-#define TRUE 1
-#define FALSE 0
-
-#define PIPE_NONE	0<<0
-#define PIPE_CRT	1<<0
-#define PIPE_TV		1<<1
-#define PIPE_DFP	1<<2
-#define PIPE_LFP	1<<3
-#define PIPE_CRT2	1<<4
-#define PIPE_TV2	1<<5
-#define PIPE_DFP2	1<<6
-#define PIPE_LFP2	1<<7
-
-typedef struct _I830MemPool *I830MemPoolPtr;
-typedef struct _I830MemRange *I830MemRangePtr;
-typedef struct _I830MemRange {
-   long Start;
-   long End;
-   long Size;
-   unsigned long Physical;
-   unsigned long Offset;		/* Offset of AGP-allocated portion */
-   unsigned long Alignment;
-   drm_handle_t Key;
-   unsigned long Pitch; // add pitch
-   I830MemPoolPtr Pool;
-} I830MemRange;
-
-typedef struct _I830MemPool {
-   I830MemRange Total;
-   I830MemRange Free;
-   I830MemRange Fixed;
-   I830MemRange Allocated;
-} I830MemPool;
-
-typedef struct {
-   int tail_mask;
-   I830MemRange mem;
-   unsigned char *virtual_start;
-   int head;
-   int tail;
-   int space;
-} I830RingBuffer;
-
-typedef struct _I830Rec {
-   unsigned char *MMIOBase;
-   unsigned char *FbBase;
-   int cpp;
-   uint32_t aper_size;
-   unsigned int bios_version;
-
-   /* These are set in PreInit and never changed. */
-   long FbMapSize;
-   long TotalVideoRam;
-   I830MemRange StolenMemory;		/* pre-allocated memory */
-   long BIOSMemorySize;			/* min stolen pool size */
-   int BIOSMemSizeLoc;
-
-   /* These change according to what has been allocated. */
-   long FreeMemory;
-   I830MemRange MemoryAperture;
-   I830MemPool StolenPool;
-   long allocatedMemory;
-
-   /* Regions allocated either from the above pools, or from agpgart. */
-   /* for single and dual head configurations */
-   I830MemRange FrontBuffer;
-   I830MemRange FrontBuffer2;
-   I830MemRange Scratch;
-   I830MemRange Scratch2;
-
-   I830RingBuffer *LpRing;
-
-   I830MemRange BackBuffer;
-   I830MemRange DepthBuffer;
-   I830MemRange TexMem;
-   int TexGranularity;
-   I830MemRange ContextMem;
-   int drmMinor;
-   Bool have3DWindows;
-
-   Bool NeedRingBufferLow;
-   Bool allowPageFlip;
-   Bool disableTiling;
-
-   int Chipset;
-   unsigned long LinearAddr;
-   unsigned long MMIOAddr;
-
-   drmSize           registerSize;     /**< \brief MMIO register map size */
-   drm_handle_t         registerHandle;   /**< \brief MMIO register map handle */
-  //   IOADDRESS ioBase;
-   int               irq;              /**< \brief IRQ number */
-   int GttBound;
-
-   drm_handle_t ring_map;
-   unsigned int Fence[8];
-
-} I830Rec;
-
-/*
- * 12288 is set as the maximum, chosen because it is enough for
- * 1920x1440@32bpp with a 2048 pixel line pitch with some to spare.
- */
-#define I830_MAXIMUM_VBIOS_MEM		12288
-#define I830_DEFAULT_VIDEOMEM_2D	(MB(32) / 1024)
-#define I830_DEFAULT_VIDEOMEM_3D	(MB(64) / 1024)
-
-/* Flags for memory allocation function */
-#define FROM_ANYWHERE			0x00000000
-#define FROM_POOL_ONLY			0x00000001
-#define FROM_NEW_ONLY			0x00000002
-#define FROM_MASK			0x0000000f
-
-#define ALLOCATE_AT_TOP			0x00000010
-#define ALLOCATE_AT_BOTTOM		0x00000020
-#define FORCE_GAPS			0x00000040
-
-#define NEED_PHYSICAL_ADDR		0x00000100
-#define ALIGN_BOTH_ENDS			0x00000200
-#define FORCE_LOW			0x00000400
-
-#define ALLOC_NO_TILING			0x00001000
-#define ALLOC_INITIAL			0x00002000
-
-#define ALLOCATE_DRY_RUN		0x80000000
-
-/* Chipset registers for VIDEO BIOS memory RW access */
-#define _855_DRAM_RW_CONTROL 0x58
-#define _845_DRAM_RW_CONTROL 0x90
-#define DRAM_WRITE    0x33330000
-
-#define KB(x) ((x) * 1024)
-#define MB(x) ((x) * KB(1024))
-
-#define GTT_PAGE_SIZE			KB(4)
-#define ROUND_TO(x, y)			(((x) + (y) - 1) / (y) * (y))
-#define ROUND_DOWN_TO(x, y)		((x) / (y) * (y))
-#define ROUND_TO_PAGE(x)		ROUND_TO((x), GTT_PAGE_SIZE)
-#define ROUND_TO_MB(x)			ROUND_TO((x), MB(1))
-#define PRIMARY_RINGBUFFER_SIZE		KB(128)
-
-
-/* Ring buffer registers, p277, overview p19
- */
-#define LP_RING     0x2030
-#define HP_RING     0x2040
-
-#define RING_TAIL      0x00
-#define TAIL_ADDR           0x000FFFF8
-#define I830_TAIL_MASK	    0x001FFFF8
-
-#define RING_HEAD      0x04
-#define HEAD_WRAP_COUNT     0xFFE00000
-#define HEAD_WRAP_ONE       0x00200000
-#define HEAD_ADDR           0x001FFFFC
-#define I830_HEAD_MASK      0x001FFFFC
-
-#define RING_START     0x08
-#define START_ADDR          0x03FFFFF8
-#define I830_RING_START_MASK	0xFFFFF000
-
-#define RING_LEN       0x0C
-#define RING_NR_PAGES       0x001FF000 
-#define I830_RING_NR_PAGES	0x001FF000
-#define RING_REPORT_MASK    0x00000006
-#define RING_REPORT_64K     0x00000002
-#define RING_REPORT_128K    0x00000004
-#define RING_NO_REPORT      0x00000000
-#define RING_VALID_MASK     0x00000001
-#define RING_VALID          0x00000001
-#define RING_INVALID        0x00000000
-
-
-/* Fence/Tiling ranges [0..7]
- */
-#define FENCE            0x2000
-#define FENCE_NR         8
-
-#define I915G_FENCE_START_MASK	0x0ff00000
-
-#define I830_FENCE_START_MASK	0x07f80000
-
-#define FENCE_START_MASK    0x03F80000
-#define FENCE_X_MAJOR       0x00000000
-#define FENCE_Y_MAJOR       0x00001000
-#define FENCE_SIZE_MASK     0x00000700
-#define FENCE_SIZE_512K     0x00000000
-#define FENCE_SIZE_1M       0x00000100
-#define FENCE_SIZE_2M       0x00000200
-#define FENCE_SIZE_4M       0x00000300
-#define FENCE_SIZE_8M       0x00000400
-#define FENCE_SIZE_16M      0x00000500
-#define FENCE_SIZE_32M      0x00000600
-#define FENCE_SIZE_64M	    0x00000700
-#define I915G_FENCE_SIZE_1M       0x00000000
-#define I915G_FENCE_SIZE_2M       0x00000100
-#define I915G_FENCE_SIZE_4M       0x00000200
-#define I915G_FENCE_SIZE_8M       0x00000300
-#define I915G_FENCE_SIZE_16M      0x00000400
-#define I915G_FENCE_SIZE_32M      0x00000500
-#define I915G_FENCE_SIZE_64M	0x00000600
-#define I915G_FENCE_SIZE_128M	0x00000700
-#define FENCE_PITCH_1       0x00000000
-#define FENCE_PITCH_2       0x00000010
-#define FENCE_PITCH_4       0x00000020
-#define FENCE_PITCH_8       0x00000030
-#define FENCE_PITCH_16      0x00000040
-#define FENCE_PITCH_32      0x00000050
-#define FENCE_PITCH_64	    0x00000060
-#define FENCE_VALID         0x00000001
-
-#include <mmio.h>
-
-#  define MMIO_IN8(base, offset) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset))
-#  define MMIO_IN32(base, offset) \
-	read_MMIO_LE32(base, offset)
-#  define MMIO_OUT8(base, offset, val) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset)) = (val)
-#  define MMIO_OUT32(base, offset, val) \
-	*(volatile unsigned int *)(void *)(((unsigned char*)(base)) + (offset)) = CPU_TO_LE32(val)
-
-
-				/* Memory mapped register access macros */
-#define INREG8(addr)        MMIO_IN8(MMIO, addr)
-#define INREG(addr)         MMIO_IN32(MMIO, addr)
-#define OUTREG8(addr, val)  MMIO_OUT8(MMIO, addr, val)
-#define OUTREG(addr, val)   MMIO_OUT32(MMIO, addr, val)
-
-#define DSPABASE		0x70184
-
-#endif
diff --git a/src/mesa/drivers/dri/i915tex/server/intel_dri.c b/src/mesa/drivers/dri/i915tex/server/intel_dri.c
deleted file mode 100644
index e49c4214ad4..00000000000
--- a/src/mesa/drivers/dri/i915tex/server/intel_dri.c
+++ /dev/null
@@ -1,1306 +0,0 @@
-/**
- * \file server/intel_dri.c
- * \brief File to perform the device-specific initialization tasks typically
- * done in the X server.
- *
- * Here they are converted to run in the client (or perhaps a standalone
- * process), and to work with the frame buffer device rather than the X
- * server infrastructure.
- * 
- * Copyright (C) 2006 Dave Airlie ([email protected])
-
- 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 THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR
- ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-*/
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <errno.h>
-#include <unistd.h>
-
-#include "driver.h"
-#include "drm.h"
-
-#include "intel.h"
-#include "i830_dri.h"
-
-#include "memops.h"
-#include "pciaccess.h"
-
-static size_t drm_page_size;
-static int nextTile = 0;
-#define xf86DrvMsg(...) do {} while(0)
-
-static const int pitches[] = {
-  128 * 8,
-  128 * 16,
-  128 * 32,
-  128 * 64,
-  0
-};
-
-static Bool I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea);
-
-static unsigned long
-GetBestTileAlignment(unsigned long size)
-{
-   unsigned long i;
-
-   for (i = KB(512); i < size; i <<= 1)
-      ;
-
-   if (i > MB(64))
-      i = MB(64);
-
-   return i;
-}
-
-static void SetFenceRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-  for (i = 0; i < 8; i++) {
-    OUTREG(FENCE + i * 4, pI830->Fence[i]);
-    //    if (I810_DEBUG & DEBUG_VERBOSE_VGA)
-    fprintf(stderr,"Fence Register : %x\n", pI830->Fence[i]);
-  }
-}
-
-/* Tiled memory is good... really, really good...
- *
- * Need to make it less likely that we miss out on this - probably
- * need to move the frontbuffer away from the 'guarenteed' alignment
- * of the first memory segment, or perhaps allocate a discontigous
- * framebuffer to get more alignment 'sweet spots'.
- */
-static void
-SetFence(const DRIDriverContext *ctx, I830Rec *pI830,
-	 int nr, unsigned int start, unsigned int pitch,
-         unsigned int size)
-{
-   unsigned int val;
-   unsigned int fence_mask = 0;
-   unsigned int fence_pitch;
-
-   if (nr < 0 || nr > 7) {
-      fprintf(stderr,
-		 "SetFence: fence %d out of range\n",nr);
-      return;
-   }
-
-   pI830->Fence[nr] = 0;
-
-   if (IS_I9XX(pI830))
-   	fence_mask = ~I915G_FENCE_START_MASK;
-   else
-   	fence_mask = ~I830_FENCE_START_MASK;
-
-   if (start & fence_mask) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not %s aligned\n",
-		 nr, start, (IS_I9XX(pI830)) ? "1MB" : "512k");
-      return;
-   }
-
-   if (start % size) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not size (%dk) aligned\n",
-		 nr, start, size / 1024);
-      return;
-   }
-
-   if (pitch & 127) {
-      fprintf(stderr,
-		 "SetFence: %d: pitch (%d) not a multiple of 128 bytes\n",
-		 nr, pitch);
-      return;
-   }
-
-   val = (start | FENCE_X_MAJOR | FENCE_VALID);
-
-   if (IS_I9XX(pI830)) {
-   	switch (size) {
-	   case MB(1):
-      		val |= I915G_FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= I915G_FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= I915G_FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= I915G_FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= I915G_FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= I915G_FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= I915G_FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-    } else {
-   	switch (size) {
-	   case KB(512):
-      		val |= FENCE_SIZE_512K;
-      		break;
-	   case MB(1):
-      		val |= FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-   }
-
-   if (IS_I9XX(pI830))
-	fence_pitch = pitch / 512;
-   else
-	fence_pitch = pitch / 128;
-
-   switch (fence_pitch) {
-   case 1:
-      val |= FENCE_PITCH_1;
-      break;
-   case 2:
-      val |= FENCE_PITCH_2;
-      break;
-   case 4:
-      val |= FENCE_PITCH_4;
-      break;
-   case 8:
-      val |= FENCE_PITCH_8;
-      break;
-   case 16:
-      val |= FENCE_PITCH_16;
-      break;
-   case 32:
-      val |= FENCE_PITCH_32;
-      break;
-   case 64:
-      val |= FENCE_PITCH_64;
-      break;
-   default:
-      fprintf(stderr,
-		 "SetFence: %d: illegal pitch (%d)\n", nr, pitch);
-      return;
-   }
-
-   pI830->Fence[nr] = val;
-}
-
-static Bool
-MakeTiles(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *pMem)
-{
-   int pitch, ntiles, i;
-
-   pitch = pMem->Pitch * ctx->cpp;
-   /*
-    * Simply try to break the region up into at most four pieces of size
-    * equal to the alignment.
-    */
-   ntiles = ROUND_TO(pMem->Size, pMem->Alignment) / pMem->Alignment;
-   if (ntiles >= 4) {
-      return FALSE;
-   }
-
-   for (i = 0; i < ntiles; i++, nextTile++) {
-     SetFence(ctx, pI830, nextTile, pMem->Start + i * pMem->Alignment,
-	       pitch, pMem->Alignment);
-   }
-   return TRUE;
-}
-
-static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-
-  /* Clear out */
-  for (i = 0; i < 8; i++)
-    pI830->Fence[i] = 0;
-  
-  nextTile = 0;
-
-  if (pI830->BackBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->BackBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the back buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the back buffer.\n");
-      pI830->allowPageFlip = FALSE;
-    }
-  }
-  
-  if (pI830->DepthBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->DepthBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the depth buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the depth buffer.\n");
-    }
-  }
-
-  return;
-}
-
-static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  struct pci_device host_bridge, ig_dev;
-  uint32_t gmch_ctrl;
-  int memsize = 0;
-  int range;
-  uint32_t aper_size;
-  uint32_t membase2 = 0;
-      
-  memset(&host_bridge, 0, sizeof(host_bridge));
-  memset(&ig_dev, 0, sizeof(ig_dev));
-
-  ig_dev.dev = 2;
-
-  pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
-
-  if (IS_I830(pI830) || IS_845G(pI830)) {
-    if ((gmch_ctrl & I830_GMCH_MEM_MASK) == I830_GMCH_MEM_128M) {
-      aper_size = 0x80000000;
-    } else {
-      aper_size = 0x40000000;
-    }
-  } else {
-    if (IS_I9XX(pI830)) {
-      int ret;
-      ret = pci_device_cfg_read_u32(&ig_dev, &membase2, 0x18);
-      if (membase2 & 0x08000000)
-	aper_size = 0x8000000;
-      else
-	aper_size = 0x10000000;
-
-      fprintf(stderr,"aper size is %08X %08x %d\n", aper_size, membase2, ret);
-    } else
-      aper_size = 0x8000000;
-  }
-
-  pI830->aper_size = aper_size;
-
-
-  /* We need to reduce the stolen size, by the GTT and the popup.
-   * The GTT varying according the the FbMapSize and the popup is 4KB */
-  range = (ctx->shared.fbSize / (1024*1024)) + 4;
-
-   if (IS_I85X(pI830) || IS_I865G(pI830) || IS_I9XX(pI830)) {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I855_GMCH_GMS_STOLEN_1M:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_4M:
-	 memsize = MB(4) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_8M:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_16M:
-	 memsize = MB(16) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_32M:
-	 memsize = MB(32) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_48M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(48) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_64M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(64) - KB(range);
-	 break;
-      }
-   } else {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I830_GMCH_GMS_STOLEN_512:
-	 memsize = KB(512) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_1024:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_8192:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I830_GMCH_GMS_LOCAL:
-	 memsize = 0;
-	 xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
-		    "Local memory found, but won't be used.\n");
-	 break;
-      }
-   }
-   if (memsize > 0) {
-     fprintf(stderr,
-		 "detected %d kB stolen memory.\n", memsize / 1024);
-   } else {
-     fprintf(stderr,
-		 "no video memory detected.\n");
-   }
-   return memsize;
-}
-
-static int AgpInit(const DRIDriverContext *ctx, I830Rec *info)
-{
-  unsigned long mode = 0x4;
-
-  if (drmAgpAcquire(ctx->drmFD) < 0) {
-    fprintf(stderr, "[gart] AGP not available\n");
-    return 0;
-  }
-  
-  if (drmAgpEnable(ctx->drmFD, mode) < 0) {
-    fprintf(stderr, "[gart] AGP not enabled\n");
-    drmAgpRelease(ctx->drmFD);
-    return 0;
-  }
-  else
-    fprintf(stderr, "[gart] AGP enabled at %dx\n", ctx->agpmode);
-
-  return 1;
-}
-
-/*
- * Allocate memory from the given pool.  Grow the pool if needed and if
- * possible.
- */
-static unsigned long
-AllocFromPool(const DRIDriverContext *ctx, I830Rec *pI830, 
-	      I830MemRange *result, I830MemPool *pool,
-	      long size, unsigned long alignment, int flags)
-{
-   long needed, start, end;
-
-   if (!result || !pool || !size)
-      return 0;
-
-   /* Calculate how much space is needed. */
-   if (alignment <= GTT_PAGE_SIZE)
-      needed = size;
-   else {
-	 start = ROUND_TO(pool->Free.Start, alignment);
-	 end = ROUND_TO(start + size, alignment);
-	 needed = end - pool->Free.Start;
-   }
-   if (needed > pool->Free.Size) {
-     return 0;
-   }
-
-   result->Start = ROUND_TO(pool->Free.Start, alignment);
-   pool->Free.Start += needed;
-   result->End = pool->Free.Start;
-
-   pool->Free.Size = pool->Free.End - pool->Free.Start;
-   result->Size = result->End - result->Start;
-   result->Pool = pool;
-   result->Alignment = alignment;
-   return needed;
-}
-
-static unsigned long AllocFromAGP(const DRIDriverContext *ctx, I830Rec *pI830, long size, unsigned long alignment, I830MemRange  *result)
-{
-   unsigned long start, end;
-   unsigned long newApStart, newApEnd;
-   int ret;
-   if (!result || !size)
-      return 0;
-   
-   if (!alignment)
-     alignment = 4;
-
-   start = ROUND_TO(pI830->MemoryAperture.Start, alignment);
-   end = ROUND_TO(start + size, alignment);
-   newApStart = end;
-   newApEnd = pI830->MemoryAperture.End;
-
-   ret=drmAgpAlloc(ctx->drmFD, size, 0, &(result->Physical), (drm_handle_t *)&(result->Key));
-   
-   if (ret)
-   {
-     fprintf(stderr,"drmAgpAlloc failed %d\n", ret);
-     return 0;
-   }
-   pI830->allocatedMemory += size;
-   pI830->MemoryAperture.Start = newApStart;
-   pI830->MemoryAperture.End = newApEnd;
-   pI830->MemoryAperture.Size = newApEnd - newApStart;
-   //   pI830->FreeMemory -= size;
-   result->Start = start;
-   result->End = start + size;
-   result->Size = size;
-   result->Offset = start;
-   result->Alignment = alignment;
-   result->Pool = NULL;
-  
-   return size;
-}
-
-unsigned long
-I830AllocVidMem(const DRIDriverContext *ctx, I830Rec *pI830,
-                I830MemRange *result, I830MemPool *pool, long size,
-                unsigned long alignment, int flags)
-{
-   unsigned long ret;
-
-   if (!result)
-      return 0;
-
-   /* Make sure these are initialised. */
-   result->Size = 0;
-   result->Key = -1;
-
-   if (!size) {
-      return 0;
-   }
-
-   if (pool->Free.Size < size) {
-      ret = AllocFromAGP(ctx, pI830, size, alignment, result);
-   }
-   else {
-      ret = AllocFromPool(ctx, pI830, result, pool, size, alignment, flags);
-      if (ret == 0)
-         ret = AllocFromAGP(ctx, pI830, size, alignment, result);
-   }
-   return ret;
-}
-
-static Bool BindAgpRange(const DRIDriverContext *ctx, I830MemRange *mem)
-{
-  if (!mem)
-    return FALSE;
-  
-  if (mem->Key == -1)
-    return TRUE;
-
-  return !drmAgpBind(ctx->drmFD, mem->Key, mem->Offset);
-}
-
-/* simple memory allocation routines needed */
-/* put ring buffer in low memory */
-/* need to allocate front, back, depth buffers aligned correctly,
-   allocate ring buffer, 
-*/
-
-/* */
-static Bool
-I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned long size, ret;
-  unsigned long lines, lineSize, align;
-
-  /* allocate ring buffer */
-  memset(pI830->LpRing, 0, sizeof(I830RingBuffer));
-  pI830->LpRing->mem.Key = -1;
-
-  size = PRIMARY_RINGBUFFER_SIZE;
-  
-  ret = I830AllocVidMem(ctx, pI830, &pI830->LpRing->mem, &pI830->StolenPool, size, 0x1000, 0);
-  
-  if (ret != size)
-  {
-    fprintf(stderr,"unable to allocate ring buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  pI830->LpRing->tail_mask = pI830->LpRing->mem.Size - 1;
-
-  
-  /* allocate front buffer */
-  memset(&(pI830->FrontBuffer), 0, sizeof(pI830->FrontBuffer));
-  pI830->FrontBuffer.Key = -1;
-  pI830->FrontBuffer.Pitch = ctx->shared.virtualWidth;
-
-  align = KB(512);  
-
-  lineSize = ctx->shared.virtualWidth * ctx->cpp;
-  lines = (ctx->shared.virtualHeight + 15) / 16 * 16;
-  size = lineSize * lines;
-  size = ROUND_TO_PAGE(size);
-
-  align = GetBestTileAlignment(size);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->FrontBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate front buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->BackBuffer), 0, sizeof(pI830->BackBuffer));
-  pI830->BackBuffer.Key = -1;
-  pI830->BackBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->BackBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate back buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->DepthBuffer), 0, sizeof(pI830->DepthBuffer));
-  pI830->DepthBuffer.Key = -1;
-  pI830->DepthBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->DepthBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate depth buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->ContextMem), 0, sizeof(pI830->ContextMem));
-  pI830->ContextMem.Key = -1;
-  size = KB(32);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->ContextMem, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
-    return FALSE;
-  }
-
-#if 0  
-  memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
-  pI830->TexMem.Key = -1;
-
-  size = 32768 * 1024;
-  ret = AllocFromAGP(ctx, pI830, size, align, &pI830->TexMem);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
-    return FALSE;
-  }
-#endif
-
-  return TRUE;
-}
-
-static Bool
-I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  if (!BindAgpRange(ctx, &pI830->LpRing->mem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->FrontBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->BackBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->DepthBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->ContextMem))
-    return FALSE;
-#if 0
-  if (!BindAgpRange(ctx, &pI830->TexMem))
-    return FALSE;
-#endif
-  return TRUE;
-}
-
-static void SetupDRIMM(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned long aperEnd = ROUND_DOWN_TO(pI830->aper_size, GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
-  unsigned long aperStart = ROUND_TO(pI830->aper_size - KB(32768), GTT_PAGE_SIZE) / GTT_PAGE_SIZE;
-
-  fprintf(stderr, "aper size is %08X\n", ctx->shared.fbSize);
-  if (drmMMInit(ctx->drmFD, aperStart, aperEnd - aperStart, DRM_BO_MEM_TT)) {
-      fprintf(stderr,
-	      "DRM MM Initialization Failed\n");
-  } else {
-    fprintf(stderr,
-	    "DRM MM Initialized at offset 0x%lx length %d page\n", aperStart, aperEnd-aperStart);
-  }
-
-}
-
-static Bool
-I830CleanupDma(const DRIDriverContext *ctx)
-{
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_CLEANUP_DMA;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-     fprintf(stderr, "I830 Dma Cleanup Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830InitDma(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-   I830RingBuffer *ring = pI830->LpRing;
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_INIT_DMA;
-
-   info.ring_start = ring->mem.Start + pI830->LinearAddr;
-   info.ring_end = ring->mem.End + pI830->LinearAddr;
-   info.ring_size = ring->mem.Size;
-
-   info.mmio_offset = (unsigned int)ctx->MMIOStart;
-
-   info.sarea_priv_offset = sizeof(drm_sarea_t);
-
-   info.front_offset = pI830->FrontBuffer.Start;
-   info.back_offset = pI830->BackBuffer.Start;
-   info.depth_offset = pI830->DepthBuffer.Start;
-   info.w = ctx->shared.virtualWidth;
-   info.h = ctx->shared.virtualHeight;
-   info.pitch = ctx->shared.virtualWidth;
-   info.back_pitch = pI830->BackBuffer.Pitch;
-   info.depth_pitch = pI830->DepthBuffer.Pitch;
-   info.cpp = ctx->cpp;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-      fprintf(stderr,
-		 "I830 Dma Initialization Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static int I830CheckDRMVersion( const DRIDriverContext *ctx,
-				  I830Rec *pI830 )
-{
-   drmVersionPtr  version;
-
-   version = drmGetVersion(ctx->drmFD);
-
-   if (version) {
-     int req_minor, req_patch;
-
-     req_minor = 4;
-     req_patch = 0;	
-
-     if (version->version_major != 1 ||
-	 version->version_minor < req_minor ||
-	 (version->version_minor == req_minor && 
-	  version->version_patchlevel < req_patch)) {
-       /* Incompatible drm version */
-       fprintf(stderr,
-	       "[dri] I830DRIScreenInit failed because of a version "
-	       "mismatch.\n"
-	       "[dri] i915.o kernel module version is %d.%d.%d "
-	       "but version 1.%d.%d or newer is needed.\n"
-	       "[dri] Disabling DRI.\n",
-	       version->version_major,
-	       version->version_minor,
-	       version->version_patchlevel,
-	       req_minor,
-	       req_patch);
-       drmFreeVersion(version);
-       return 0;
-     }
-     
-     pI830->drmMinor = version->version_minor;
-     drmFreeVersion(version);
-   }
-   return 1;
-}
-
-static void
-I830SetRingRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned int itemp;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-   OUTREG(LP_RING + RING_LEN, 0);
-   OUTREG(LP_RING + RING_TAIL, 0);
-   OUTREG(LP_RING + RING_HEAD, 0);
-
-   if ((long)(pI830->LpRing->mem.Start & I830_RING_START_MASK) !=
-       pI830->LpRing->mem.Start) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer start (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Start, I830_RING_START_MASK);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = pI830->LpRing->mem.Start & I830_RING_START_MASK;
-   OUTREG(LP_RING + RING_START, itemp);
-
-   if (((pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES) !=
-       pI830->LpRing->mem.Size - 4096) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer size - 4096 (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Size - 4096,
-		 I830_RING_NR_PAGES);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = (pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES;
-   itemp |= (RING_NO_REPORT | RING_VALID);
-   OUTREG(LP_RING + RING_LEN, itemp);
-
-   pI830->LpRing->head = INREG(LP_RING + RING_HEAD) & I830_HEAD_MASK;
-   pI830->LpRing->tail = INREG(LP_RING + RING_TAIL);
-   pI830->LpRing->space = pI830->LpRing->head - (pI830->LpRing->tail + 8);
-   if (pI830->LpRing->space < 0)
-      pI830->LpRing->space += pI830->LpRing->mem.Size;
-
-   SetFenceRegs(ctx, pI830);
-   
-   /* RESET THE DISPLAY PIPE TO POINT TO THE FRONTBUFFER - hacky
-      hacky hacky */
-   OUTREG(DSPABASE, pI830->FrontBuffer.Start + pI830->LinearAddr);
-
-}
-
-static Bool
-I830SetParam(const DRIDriverContext *ctx, int param, int value)
-{
-   drmI830SetParam sp;
-
-   memset(&sp, 0, sizeof(sp));
-   sp.param = param;
-   sp.value = value;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_SETPARAM, &sp, sizeof(sp))) {
-      fprintf(stderr, "I830 SetParam Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   fprintf(stderr,
-              "[drm] Mapping front buffer\n");
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->front_offset + pI830->LinearAddr),
-                 sarea->front_size,
-                 DRM_FRAME_BUFFER,  /*DRM_AGP,*/
-                 0,
-                 &sarea->front_handle) < 0) {
-     fprintf(stderr,
-	     "[drm] drmAddMap(front_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   ctx->shared.hFrameBuffer = sarea->front_handle;
-   ctx->shared.fbSize = sarea->front_size;
-   fprintf(stderr, "[drm] Front Buffer = 0x%08x\n",
-	   sarea->front_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->back_offset),
-                 sarea->back_size, DRM_AGP, 0,
-                 &sarea->back_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(back_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Back Buffer = 0x%08x\n",
-              sarea->back_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)sarea->depth_offset,
-                 sarea->depth_size, DRM_AGP, 0,
-                 &sarea->depth_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(depth_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
-              sarea->depth_handle);
-
-#if 0
-   if (drmAddMap(ctx->drmFD,
-		 (drm_handle_t)sarea->tex_offset,
-		 sarea->tex_size, DRM_AGP, 0,
-		 &sarea->tex_handle) < 0) {
-      fprintf(stderr,
-		 "[drm] drmAddMap(tex_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] textures = 0x%08x\n",
-	      sarea->tex_handle);
-#endif
-   return TRUE;
-}
-
-
-static void
-I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-#if 1
-   if (sarea->front_handle) {
-      drmRmMap(ctx->drmFD, sarea->front_handle);
-      sarea->front_handle = 0;
-   }
-#endif
-   if (sarea->back_handle) {
-      drmRmMap(ctx->drmFD, sarea->back_handle);
-      sarea->back_handle = 0;
-   }
-   if (sarea->depth_handle) {
-      drmRmMap(ctx->drmFD, sarea->depth_handle);
-      sarea->depth_handle = 0;
-   }
-   if (sarea->tex_handle) {
-      drmRmMap(ctx->drmFD, sarea->tex_handle);
-      sarea->tex_handle = 0;
-   }
-}
-
-static Bool
-I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  if (drmAddMap(ctx->drmFD,
-		(drm_handle_t)pI830->LpRing->mem.Start,
-		pI830->LpRing->mem.Size, DRM_AGP, 0,
-		&pI830->ring_map) < 0) {
-    fprintf(stderr,
-	    "[drm] drmAddMap(ring_map) failed. Disabling DRI\n");
-    return FALSE;
-  }
-  fprintf(stderr, "[drm] ring buffer = 0x%08x\n",
-	  pI830->ring_map);
-
-  if (I830InitDma(ctx, pI830) == FALSE) {
-    return FALSE;
-  }
-  
-   /* init to zero to be safe */
-
-  I830DRIMapScreenRegions(ctx, pI830, sarea);
-  SetupDRIMM(ctx, pI830);
-
-   if (ctx->pciDevice != PCI_CHIP_845_G &&
-       ctx->pciDevice != PCI_CHIP_I830_M) {
-      I830SetParam(ctx, I830_SETPARAM_USE_MI_BATCHBUFFER_START, 1 );
-   }
-
-   /* Okay now initialize the dma engine */
-   {
-      pI830->irq = drmGetInterruptFromBusID(ctx->drmFD,
-					    ctx->pciBus,
-					    ctx->pciDevice,
-					    ctx->pciFunc);
-
-      if (drmCtlInstHandler(ctx->drmFD, pI830->irq)) {
-	 fprintf(stderr,
-		    "[drm] failure adding irq handler\n");
-	 pI830->irq = 0;
-	 return FALSE;
-      }
-      else
-	 fprintf(stderr,
-		    "[drm] dma control initialized, using IRQ %d\n",
-		    pI830->irq);
-   }
-
-   fprintf(stderr, "[dri] visual configs initialized\n");
-
-   return TRUE;
-}
-
-static Bool
-I830ClearScreen(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  /* need to drmMap front and back buffers and zero them */
-  drmAddress map_addr;
-  int ret;
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->front_handle,
-	       sarea->front_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map front buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->front_size);
-  drmUnmap(map_addr, sarea->front_size);
-
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->back_handle,
-	       sarea->back_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map back buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->back_size);
-  drmUnmap(map_addr, sarea->back_size);
-
-  return TRUE;
-}
-
-static Bool
-I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
-		  
-{
-   I830DRIPtr pI830DRI;
-   drmI830Sarea *pSAREAPriv;
-   int err;
-      
-   drm_page_size = getpagesize();   
-
-   pI830->registerSize = ctx->MMIOSize;
-   /* This is a hack for now.  We have to have more than a 4k page here
-    * because of the size of the state.  However, the state should be
-    * in a per-context mapping.  This will be added in the Mesa 3.5 port
-    * of the I830 driver.
-    */
-   ctx->shared.SAREASize = SAREA_MAX;
-
-   /* Note that drmOpen will try to load the kernel module, if needed. */
-   ctx->drmFD = drmOpen("i915", NULL );
-   if (ctx->drmFD < 0) {
-      fprintf(stderr, "[drm] drmOpen failed\n");
-      return 0;
-   }
-
-   if ((err = drmSetBusid(ctx->drmFD, ctx->pciBusID)) < 0) {
-      fprintf(stderr, "[drm] drmSetBusid failed (%d, %s), %s\n",
-	      ctx->drmFD, ctx->pciBusID, strerror(-err));
-      return 0;
-   }
-
-   if (drmAddMap( ctx->drmFD,
-		  0,
-		  ctx->shared.SAREASize,
-		  DRM_SHM,
-		  DRM_CONTAINS_LOCK,
-		  &ctx->shared.hSAREA) < 0)
-   {
-     fprintf(stderr, "[drm] drmAddMap failed\n");
-     return 0;
-   }
-
-   fprintf(stderr, "[drm] added %d byte SAREA at 0x%08x\n",
-	   ctx->shared.SAREASize, ctx->shared.hSAREA);
-   
-   if (drmMap( ctx->drmFD,
-	       ctx->shared.hSAREA,
-	       ctx->shared.SAREASize,
-	       (drmAddressPtr)(&ctx->pSAREA)) < 0)
-   {
-      fprintf(stderr, "[drm] drmMap failed\n");
-      return 0;
-   
-   }
-   
-   memset(ctx->pSAREA, 0, ctx->shared.SAREASize);
-   fprintf(stderr, "[drm] mapped SAREA 0x%08x to %p, size %d\n",
-	   ctx->shared.hSAREA, ctx->pSAREA, ctx->shared.SAREASize);
-   
-
-   if (drmAddMap(ctx->drmFD, 
-		 ctx->MMIOStart,
-		 ctx->MMIOSize,
-		 DRM_REGISTERS, 
-		 DRM_READ_ONLY, 
-		 &pI830->registerHandle) < 0) {
-      fprintf(stderr, "[drm] drmAddMap mmio failed\n");	
-      return 0;
-   }
-   fprintf(stderr,
-	   "[drm] register handle = 0x%08x\n", pI830->registerHandle);
-
-
-   if (!I830CheckDRMVersion(ctx, pI830)) {
-     return FALSE;
-   }
-
-   /* Create a 'server' context so we can grab the lock for
-    * initialization ioctls.
-    */
-   if ((err = drmCreateContext(ctx->drmFD, &ctx->serverContext)) != 0) {
-      fprintf(stderr, "%s: drmCreateContext failed %d\n", __FUNCTION__, err);
-      return 0;
-   }
-
-   DRM_LOCK(ctx->drmFD, ctx->pSAREA, ctx->serverContext, 0); 
-
-   /* Initialize the SAREA private data structure */
-   pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				 sizeof(drm_sarea_t));
-   memset(pSAREAPriv, 0, sizeof(*pSAREAPriv));
-
-   pI830->StolenMemory.Size = I830DetectMemory(ctx, pI830);
-   pI830->StolenMemory.Start = 0;
-   pI830->StolenMemory.End = pI830->StolenMemory.Size;
-
-   pI830->MemoryAperture.Start = pI830->StolenMemory.End;
-   pI830->MemoryAperture.End = KB(40000);
-   pI830->MemoryAperture.Size = pI830->MemoryAperture.End - pI830->MemoryAperture.Start;
-
-   pI830->StolenPool.Fixed = pI830->StolenMemory;
-   pI830->StolenPool.Total = pI830->StolenMemory;
-   pI830->StolenPool.Free = pI830->StolenPool.Total;
-   pI830->FreeMemory = pI830->StolenPool.Total.Size;
-
-   if (!AgpInit(ctx, pI830))
-     return FALSE;
-
-   if (I830AllocateMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   if (I830BindMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   pSAREAPriv->rotated_offset = -1;
-   pSAREAPriv->rotated_size = 0;
-   pSAREAPriv->rotated_pitch = ctx->shared.virtualWidth;
-
-   pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
-   pSAREAPriv->front_size = pI830->FrontBuffer.Size;
-   pSAREAPriv->width = ctx->shared.virtualWidth;
-   pSAREAPriv->height = ctx->shared.virtualHeight;
-   pSAREAPriv->pitch = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualX = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualY = ctx->shared.virtualHeight;
-   pSAREAPriv->back_offset = pI830->BackBuffer.Start;
-   pSAREAPriv->back_size = pI830->BackBuffer.Size;
-   pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
-   pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
-#if 0
-   pSAREAPriv->tex_offset = pI830->TexMem.Start;
-   pSAREAPriv->tex_size = pI830->TexMem.Size;
-#endif
-   pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
-
-   ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
-   ctx->driverClientMsgSize = sizeof(I830DRIRec);
-   pI830DRI = (I830DRIPtr)ctx->driverClientMsg;
-   pI830DRI->deviceID = pI830->Chipset;
-   pI830DRI->regsSize = I830_REG_SIZE;
-   pI830DRI->width = ctx->shared.virtualWidth;
-   pI830DRI->height = ctx->shared.virtualHeight;
-   pI830DRI->mem = ctx->shared.fbSize;
-   pI830DRI->cpp = ctx->cpp;
-
-   pI830DRI->bitsPerPixel = ctx->bpp;
-   pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
-   
-   err = I830DRIDoMappings(ctx, pI830, pSAREAPriv);
-   if (err == FALSE)
-       return FALSE;
-
-   I830SetupMemoryTiling(ctx, pI830);
-
-   /* Quick hack to clear the front & back buffers.  Could also use
-    * the clear ioctl to do this, but would need to setup hw state
-    * first.
-    */
-   I830ClearScreen(ctx, pI830, pSAREAPriv);
-
-   I830SetRingRegs(ctx, pI830);
-
-   return TRUE;
-}
-
-
-/**
- * \brief Validate the fbdev mode.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Saves some registers and returns 1.
- *
- * \sa radeonValidateMode().
- */
-static int i830ValidateMode( const DRIDriverContext *ctx )
-{
-  return 1;
-}
-
-/**
- * \brief Examine mode returned by fbdev.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Restores registers that fbdev has clobbered and returns 1.
- *
- * \sa i810ValidateMode().
- */
-static int i830PostValidateMode( const DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = ctx->driverPrivate;
-
-  I830SetRingRegs(ctx, pI830);
-  return 1;
-}
-
-
-/**
- * \brief Initialize the framebuffer device mode
- *
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Fills in \p info with some default values and some information from \p ctx
- * and then calls I810ScreenInit() for the screen initialization.
- * 
- * Before exiting clears the framebuffer memory accessing it directly.
- */
-static int i830InitFBDev( DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = calloc(1, sizeof(I830Rec));
-  int i;
-
-   {
-      int  dummy = ctx->shared.virtualWidth;
-
-      switch (ctx->bpp / 8) {
-      case 1: dummy = (ctx->shared.virtualWidth + 127) & ~127; break;
-      case 2: dummy = (ctx->shared.virtualWidth +  31) &  ~31; break;
-      case 3:
-      case 4: dummy = (ctx->shared.virtualWidth +  15) &  ~15; break;
-      }
-
-      ctx->shared.virtualWidth = dummy;
-      ctx->shared.Width = ctx->shared.virtualWidth;
-   }
-
-
-   for (i = 0; pitches[i] != 0; i++) {
-     if (pitches[i] >= ctx->shared.virtualWidth) {
-       ctx->shared.virtualWidth = pitches[i];
-       break;
-     }
-   }
-
-   ctx->driverPrivate = (void *)pI830;
-   
-   pI830->LpRing = calloc(1, sizeof(I830RingBuffer));
-   pI830->Chipset = ctx->chipset;
-   pI830->LinearAddr = ctx->FBStart;
-
-   if (!I830ScreenInit( ctx, pI830 ))
-      return 0;
-
-   
-   return 1;
-}
-
-
-/**
- * \brief The screen is being closed, so clean up any state and free any
- * resources used by the DRI.
- *
- * \param ctx display handle.
- *
- * Unmaps the SAREA, closes the DRM device file descriptor and frees the driver
- * private data.
- */
-static void i830HaltFBDev( DRIDriverContext *ctx )
-{
-  drmI830Sarea *pSAREAPriv;
-  I830Rec *pI830 = ctx->driverPrivate;
-
-   if (pI830->irq) {
-       drmCtlUninstHandler(ctx->drmFD);
-       pI830->irq = 0;   }
-
-   I830CleanupDma(ctx);
-
-  pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				sizeof(drm_sarea_t));
-
-  I830DRIUnmapScreenRegions(ctx, pI830, pSAREAPriv);
-  drmUnmap( ctx->pSAREA, ctx->shared.SAREASize );
-  drmClose(ctx->drmFD);
-  
-  if (ctx->driverPrivate) {
-    free(ctx->driverPrivate);
-    ctx->driverPrivate = 0;
-  }
-}
-
-
-extern void i810NotifyFocus( int );
-
-/**
- * \brief Exported driver interface for Mini GLX.
- *
- * \sa DRIDriverRec.
- */
-const struct DRIDriverRec __driDriver = {
-   i830ValidateMode,
-   i830PostValidateMode,
-   i830InitFBDev,
-   i830HaltFBDev,
-   NULL,//I830EngineShutdown,
-   NULL, //I830EngineRestore,  
-#ifndef _EMBEDDED
-   0,
-#else
-   i810NotifyFocus, 
-#endif
-};
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index 9e4ff112dc3..7e07bc9c1a5 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -11,6 +11,7 @@ DRIVER_SOURCES = \
 	intel_buffer_objects.c \
 	intel_buffers.c \
 	intel_context.c \
+	intel_decode.c \
 	intel_ioctl.c \
 	intel_mipmap_tree.c \
 	intel_regions.c \
@@ -22,8 +23,6 @@ DRIVER_SOURCES = \
 	intel_tex.c \
 	intel_tex_layout.c \
 	intel_tex_validate.c \
-	brw_aub.c \
-	brw_aub_playback.c \
 	brw_cc.c \
 	brw_clip.c \
 	brw_clip_line.c \
@@ -77,6 +76,12 @@ DRIVER_SOURCES = \
 	brw_wm_state.c \
 	brw_wm_surface_state.c 
 
+SYMLINKS = \
+	server/i830_dri.h \
+	server/i830_common.h \
+	server/intel_dri.c \
+	server/intel.h
+
 C_SOURCES = \
 	$(COMMON_SOURCES) \
 	$(MINIGLX_SOURCES) \
@@ -88,6 +93,13 @@ DRIVER_DEFINES = -I../intel
 
 include ../Makefile.template
 
+intel_decode.o: ../intel/intel_decode.c
 intel_tex_layout.o: ../intel/intel_tex_layout.c
 
-symlinks:
+server:
+	mkdir -p server
+
+$(SYMLINKS): server
+	@[ -e $@ ] || ln -sf ../../i915/$@ server/
+
+symlinks: $(SYMLINKS)
diff --git a/src/mesa/drivers/dri/i965/brw_aub.c b/src/mesa/drivers/dri/i965/brw_aub.c
deleted file mode 100644
index f851a5b7955..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub.c
+++ /dev/null
@@ -1,353 +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]>
-  */
-
-#include "brw_context.h"
-#include "brw_aub.h"
-#include "intel_regions.h"
-#include <stdio.h>
-
-extern char *__progname;
-
-
-/* Registers to control page table
- */
-#define PGETBL_CTL       0x2020
-#define PGETBL_ENABLED   0x1
-
-#define NR_GTT_ENTRIES  65536	/* 256 mb */
-
-#define FAIL										\
-do {											\
-   fprintf(stderr, "failed to write aub data at %s/%d\n", __FUNCTION__, __LINE__);	\
-   exit(1);										\
-} while (0)
-
-
-/* Emit the headers at the top of each aubfile.  Initialize the GTT.
- */
-static void init_aubfile( FILE *aub_file )
-{   
-   struct aub_file_header fh;
-   struct aub_block_header bh;
-   unsigned int data;
-
-   static int nr;
-   
-   nr++;
-
-   /* Emit the aub header:
-    */
-   memset(&fh, 0, sizeof(fh));
-
-   fh.instruction_type = AUB_FILE_HEADER;
-   fh.minor = 0x0;
-   fh.major = 0x7;
-   memcpy(fh.application, __progname, sizeof(fh.application));
-   fh.day = (nr>>24) & 0xff;
-   fh.month = 0x0;
-   fh.year = 0x0;
-   fh.timezone = 0x0;
-   fh.second = nr & 0xff;
-   fh.minute = (nr>>8) & 0xff;
-   fh.hour = (nr>>16) & 0xff;
-   fh.comment_length = 0x0;   
-
-   if (fwrite(&fh, sizeof(fh), 1, aub_file) < 0) 
-      FAIL;
-         
-   /* Setup the GTT starting at main memory address zero (!):
-    */
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_MMI0_WRITE32;
-   bh.type = 0x0;
-   bh.address_space = ADDR_GTT;	/* ??? */
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address = PGETBL_CTL;
-   bh.length = 0x4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   data = 0x0 | PGETBL_ENABLED;
-
-   if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-      FAIL;
-}
-
-
-static void init_aub_gtt( struct brw_context *brw,
-			  GLuint start_offset, 
-			  GLuint size )
-{
-   FILE *aub_file = brw->intel.aub_file;
-   struct aub_block_header bh;
-   unsigned int i;
-
-   assert(start_offset + size < NR_GTT_ENTRIES * 4096);
-
-
-   memset(&bh, 0, sizeof(bh));
-   
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = 0x0;
-   bh.address_space = ADDR_MAIN;
-   bh.general_state_type = 0x0;
-   bh.surface_state_type = 0x0;
-   bh.address =  start_offset / 4096 * 4;
-   bh.length = size / 4096 * 4;
-
-   if (fwrite(&bh, sizeof(bh), 1, aub_file) < 0) 
-      FAIL;
-
-   for (i = 0; i < size / 4096; i++) {
-      GLuint data = brw->next_free_page | 1;
-
-      brw->next_free_page += 4096;
-
-      if (fwrite(&data, sizeof(data), 1, aub_file) < 0) 
-	 FAIL;
-   }
-
-}
-
-static void write_block_header( FILE *aub_file,
-				struct aub_block_header *bh,
-				const GLuint *data,
-				GLuint sz )
-{
-   sz = (sz + 3) & ~3;
-
-   if (fwrite(bh, sizeof(*bh), 1, aub_file) < 0) 
-      FAIL;
-
-   if (fwrite(data, sz, 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-static void write_dump_bmp( FILE *aub_file,
-			    struct aub_dump_bmp *db )
-{
-   if (fwrite(db, sizeof(*db), 1, aub_file) < 0) 
-      FAIL;
-
-   fflush(aub_file);
-}
-
-
-
-static void brw_aub_gtt_data( struct intel_context *intel,
-			      GLuint offset,
-			      const void *data,
-			      GLuint sz,
-			      GLuint type,
-			      GLuint state_type )
-{
-   struct aub_block_header bh;
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_DATA_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-
-   if (type == DW_GENERAL_STATE) {
-      bh.general_state_type = state_type;
-      bh.surface_state_type = 0;
-   }
-   else {
-      bh.general_state_type = 0;
-      bh.surface_state_type = state_type;
-   }
-
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(intel->aub_file, &bh, data, sz);
-}
-
-
-
-static void brw_aub_gtt_cmds( struct intel_context *intel,
-			      GLuint offset,
-			      const void *data,
-			      GLuint sz )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   struct aub_block_header bh;   
-   GLuint type = CW_PRIMARY_RING_A;
-   
-
-   bh.instruction_type = AUB_BLOCK_HEADER;
-   bh.operation = BH_COMMAND_WRITE;
-   bh.type = type;
-   bh.address_space = ADDR_GTT;
-   bh.pad0 = 0;
-   bh.general_state_type = 0;
-   bh.surface_state_type = 0;
-   bh.pad1 = 0;
-   bh.address = offset;
-   bh.length = sz;
-
-   write_block_header(brw->intel.aub_file, &bh, data, sz);
-}
-
-static void brw_aub_dump_bmp( struct intel_context *intel,
-			      GLuint buffer )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);
-   intelScreenPrivate *intelScreen = brw->intel.intelScreen;
-   struct aub_dump_bmp db;
-   GLuint format;
-
-   if (intelScreen->cpp == 4)
-      format = 0x7;
-   else
-      format = 0x3;
-
-
-   if (buffer == 0) {
-      db.instruction_type = AUB_DUMP_BMP;
-      db.xmin = 0;
-      db.ymin = 0;
-      db.format = format;
-      db.bpp = intelScreen->cpp * 8;
-      db.pitch = intelScreen->front.pitch / intelScreen->cpp;
-      db.xsize = intelScreen->width;
-      db.ysize = intelScreen->height;
-      db.addr = intelScreen->front.offset;
-      db.unknown = 0x0;		/* 4: xmajor tiled, 0: not tiled */
-
-      write_dump_bmp(brw->intel.aub_file, &db);
-   }
-   else {
-      db.instruction_type = AUB_DUMP_BMP;
-      db.xmin = 0;
-      db.ymin = 0;
-      db.format = format;
-      db.bpp = intel->back_region->cpp * 8;
-      db.pitch = intel->back_region->pitch;
-      db.xsize = intel->back_region->pitch;
-      db.ysize = intel->back_region->height;
-      db.addr = intelScreen->back.offset;
-      db.unknown = intel->back_region->tiled ? 0x4 : 0x0;
-
-      write_dump_bmp(brw->intel.aub_file, &db);
-   }
-}
-
-/* Attempt to prevent monster aubfiles by closing and reopening when
- * the state pools wrap.
- */
-static void brw_aub_wrap( struct intel_context *intel )
-{
-   struct brw_context *brw = brw_context(&intel->ctx);   
-   if (intel->aub_file) {
-      brw_aub_destroy(brw);
-      brw_aub_init(brw);
-   }
-   brw->wrap = 1;		/* ??? */
-}
-
-
-int brw_aub_init( struct brw_context *brw )
-{
-   struct intel_context *intel = &brw->intel;
-   intelScreenPrivate *intelScreen = intel->intelScreen;
-   char filename[80];
-   int val;
-   static int i = 0;
-
-   i++;
-
-   if (_mesa_getenv("INTEL_REPLAY"))
-      return 0;
-
-   if (_mesa_getenv("INTEL_AUBFILE")) {
-      val = snprintf(filename, sizeof(filename), "%s%d.aub", _mesa_getenv("INTEL_AUBFILE"), i%4);
-      _mesa_printf("--> Aub file: %s\n", filename);
-      brw->intel.aub_file = fopen(filename, "w");
-   }
-   else if (_mesa_getenv("INTEL_AUB")) {
-      val = snprintf(filename, sizeof(filename), "%s.aub", __progname);
-      if (val < 0 || val > sizeof(filename)) 
-	 strcpy(filename, "default.aub");   
-   
-      _mesa_printf("--> Aub file: %s\n", filename);
-      brw->intel.aub_file = fopen(filename, "w");
-   }
-   else {
-      return 0;
-   }
-
-   if (!brw->intel.aub_file) {
-      _mesa_printf("couldn't open aubfile\n");
-      exit(1);
-   }
-
-   brw->intel.vtbl.aub_commands = brw_aub_gtt_cmds;
-   brw->intel.vtbl.aub_dump_bmp = brw_aub_dump_bmp;
-   brw->intel.vtbl.aub_gtt_data = brw_aub_gtt_data;
-   brw->intel.vtbl.aub_wrap = brw_aub_wrap;
-   
-   init_aubfile(brw->intel.aub_file);
-
-   /* The GTT is located starting address zero in main memory.  Pages
-    * to populate the gtt start after this point.
-    */
-   brw->next_free_page = (NR_GTT_ENTRIES * 4 + 4095) & ~4095;
-
-   /* More or less correspond with all the agp regions mapped by the
-    * driver:
-    */
-   init_aub_gtt(brw, 0, 4096*4); /* so new fulsim doesn't crash */
-   init_aub_gtt(brw, intelScreen->front.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->back.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->depth.offset, intelScreen->back.size);
-   init_aub_gtt(brw, intelScreen->tex.offset, intelScreen->tex.size);
-
-   return 0;
-}
-
-void brw_aub_destroy( struct brw_context *brw )
-{
-   if (brw->intel.aub_file) {
-      fclose(brw->intel.aub_file);
-      brw->intel.aub_file = NULL;
-   }
-}
diff --git a/src/mesa/drivers/dri/i965/brw_aub.h b/src/mesa/drivers/dri/i965/brw_aub.h
deleted file mode 100644
index 198e36dc3c0..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub.h
+++ /dev/null
@@ -1,172 +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]>
-  */
-
-#ifndef BRW_AUB_H
-#define BRW_AUB_H
-
-struct aub_file_header {
-   unsigned int instruction_type;
-   unsigned int pad0:16;
-   unsigned int minor:8;
-   unsigned int major:8;
-   unsigned char application[8*4];
-   unsigned int day:8;
-   unsigned int month:8;
-   unsigned int year:16;
-   unsigned int timezone:8;
-   unsigned int second:8;
-   unsigned int minute:8;
-   unsigned int hour:8;
-   unsigned int comment_length:16;   
-   unsigned int pad1:16;
-};
-
-struct aub_block_header {
-   unsigned int instruction_type;
-   unsigned int operation:8;
-   unsigned int type:8;
-   unsigned int address_space:8;
-   unsigned int pad0:8;
-   unsigned int general_state_type:8;
-   unsigned int surface_state_type:8;
-   unsigned int pad1:16;
-   unsigned int address;
-   unsigned int length;
-};
-
-struct aub_dump_bmp {
-   unsigned int instruction_type;
-   unsigned int xmin:16;
-   unsigned int ymin:16;
-   unsigned int pitch:16;
-   unsigned int bpp:8;
-   unsigned int format:8;
-   unsigned int xsize:16;
-   unsigned int ysize:16;
-   unsigned int addr;
-   unsigned int unknown;
-};
-
-enum bh_operation {
-   BH_COMMENT,
-   BH_DATA_WRITE,
-   BH_COMMAND_WRITE,
-   BH_MMI0_WRITE32,
-   BH_END_SCENE,
-   BH_CONFIG_MEMORY_MAP,
-   BH_MAX_OPERATION
-};
-
-enum command_write_type {
-   CW_HWB_RING = 1,
-   CW_PRIMARY_RING_A,
-   CW_PRIMARY_RING_B,		/* XXX - disagreement with listaub! */
-   CW_PRIMARY_RING_C,
-   CW_MAX_TYPE
-};
-
-enum data_write_type {
-   DW_NOTYPE,
-   DW_BATCH_BUFFER,
-   DW_BIN_BUFFER,
-   DW_BIN_POINTER_LIST,
-   DW_SLOW_STATE_BUFFER,
-   DW_VERTEX_BUFFER,
-   DW_2D_MAP,
-   DW_CUBE_MAP,
-   DW_INDIRECT_STATE_BUFFER,
-   DW_VOLUME_MAP,
-   DW_1D_MAP,
-   DW_CONSTANT_BUFFER,
-   DW_CONSTANT_URB_ENTRY,
-   DW_INDEX_BUFFER,
-   DW_GENERAL_STATE,
-   DW_SURFACE_STATE,
-   DW_MEDIA_OBJECT_INDIRECT_DATA,
-   DW_MAX_TYPE
-};
-
-enum data_write_general_state_type {
-   DWGS_NOTYPE,
-   DWGS_VERTEX_SHADER_STATE,
-   DWGS_GEOMETRY_SHADER_STATE ,
-   DWGS_CLIPPER_STATE,
-   DWGS_STRIPS_FANS_STATE,
-   DWGS_WINDOWER_IZ_STATE,
-   DWGS_COLOR_CALC_STATE,
-   DWGS_CLIPPER_VIEWPORT_STATE,	/* was 0x7 */
-   DWGS_STRIPS_FANS_VIEWPORT_STATE,
-   DWGS_COLOR_CALC_VIEWPORT_STATE, /* was 0x9 */
-   DWGS_SAMPLER_STATE,
-   DWGS_KERNEL_INSTRUCTIONS,
-   DWGS_SCRATCH_SPACE,
-   DWGS_SAMPLER_DEFAULT_COLOR,
-   DWGS_INTERFACE_DESCRIPTOR,
-   DWGS_VLD_STATE,
-   DWGS_VFE_STATE,
-   DWGS_MAX_TYPE
-};
-
-enum data_write_surface_state_type {
-   DWSS_NOTYPE,
-   DWSS_BINDING_TABLE_STATE,
-   DWSS_SURFACE_STATE,
-   DWSS_MAX_TYPE
-};
-
-enum memory_map_type {
-   MM_DEFAULT,
-   MM_DYNAMIC,
-   MM_MAX_TYPE
-};
-
-enum address_space {
-   ADDR_GTT,
-   ADDR_LOCAL,
-   ADDR_MAIN,
-   ADDR_MAX
-};
-
-
-#define AUB_FILE_HEADER 0xe085000b
-#define AUB_BLOCK_HEADER 0xe0c10003
-#define AUB_DUMP_BMP 0xe09e0004
-
-struct brw_context;
-struct intel_context;
-
-int brw_aub_init( struct brw_context *brw );
-void brw_aub_destroy( struct brw_context *brw );
-
-int brw_playback_aubfile(struct brw_context *brw,
-			 const char *filename);
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_aub_playback.c b/src/mesa/drivers/dri/i965/brw_aub_playback.c
deleted file mode 100644
index 2433d50c116..00000000000
--- a/src/mesa/drivers/dri/i965/brw_aub_playback.c
+++ /dev/null
@@ -1,443 +0,0 @@
-
-#include <stdio.h>
-#include <sys/mman.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <unistd.h>
-#include <fcntl.h>
-
-#include "brw_aub.h"
-#include "brw_defines.h"
-#include "brw_context.h"
-#include "intel_ioctl.h"
-#include "bufmgr.h"
-
-struct aub_state {
-   struct intel_context *intel;
-   const char *map;
-   unsigned int csr;
-   unsigned int sz;
-};
-
-
-static int gobble( struct aub_state *s, int size )
-{
-   if (s->csr + size > s->sz) {
-      _mesa_printf("EOF in %s\n", __FUNCTION__);
-      return 1;
-   }
-
-   s->csr += size;
-   return 0;
-}
-
-static void flush_and_fence( struct aub_state *s )
-{
-   struct intel_context *intel = s->intel;
-   GLuint buf[2];
-
-   buf[0] = intel->vtbl.flush_cmd();
-   buf[1] = 0;
-
-   intel_cmd_ioctl(intel, (char *)&buf, sizeof(buf));
-      
-   intelWaitIrq( intel, intelEmitIrqLocked( intel ));
-}
-
-static void flush_cmds( struct aub_state *s,
-			const void *data,
-			int len )
-{
-   DBG("%s %d\n", __FUNCTION__, len);
-
-   if (len & 0x4) {
-      unsigned int *tmp = malloc(len + 4);
-      DBG("padding to octword\n");
-      memcpy(tmp, data, len);
-      tmp[len/4] = MI_NOOP;
-      flush_cmds(s, tmp, len+4);
-      free(tmp);
-      return;
-   }
-
-   /* For ring data, just send off immediately via an ioctl.
-    * This differs slightly from how the stream was executed
-    * initially as this would have been a batchbuffer.
-    */
-   intel_cmd_ioctl(s->intel, (void *)data, len);
-
-   if (1)
-      flush_and_fence(s);
-}
-
-static const char *pstrings[] = {
-   "none",
-   "POINTLIST",
-   "LINELIST",
-   "LINESTRIP",
-   "TRILIST",
-   "TRISTRIP",
-   "TRIFAN",
-   "QUADLIST",
-   "QUADSTRIP",
-   "LINELIST_ADJ",
-   "LINESTRIP_ADJ",
-   "TRILIST_ADJ",
-   "TRISTRIP_ADJ",
-   "TRISTRIP_REVERSE",
-   "POLYGON",
-   "RECTLIST",
-   "LINELOOP",
-   "POINTLIST_BF",
-   "LINESTRIP_CONT",
-   "LINESTRIP_BF",
-   "LINESTRIP_CONT_BF",
-   "TRIFAN_NOSTIPPLE",
-};
-
-static void do_3d_prim( struct aub_state *s,
-			const void *data,
-			int len )
-{
-   struct brw_3d_primitive prim;
-   const struct brw_3d_primitive *orig = data;
-   int i;
-
-   assert(len == sizeof(prim));
-   memcpy(&prim, data, sizeof(prim));
-
-#define START 0
-#define BLOCK (12*28)
-
-   if (orig->verts_per_instance < BLOCK)
-      flush_cmds(s, &prim, sizeof(prim));
-   else {
-      for (i = START; i + BLOCK < orig->verts_per_instance; i += BLOCK/2) {
-	 prim.start_vert_location = i;
-	 prim.verts_per_instance = BLOCK;
-	 _mesa_printf("%sprim %d/%s verts %d..%d (of %d)\n", 
-		      prim.header.indexed ? "INDEXED " : "",
-		      prim.header.topology, pstrings[prim.header.topology%16],
-		      prim.start_vert_location, 
-		      prim.start_vert_location + prim.verts_per_instance,
-		      orig->verts_per_instance);
-	 flush_cmds(s, &prim, sizeof(prim));
-      }
-   }
-}
-
-
-
-static struct {
-   int cmd;
-   const char *name;
-   int has_length;
-} cmd_info[] = {
-   { 0, "NOOP", 0 },
-   { 0x5410, "XY_COLOR_BLT_RGB", 1 },
-   { 0x5430, "XY_COLOR_BLT_RGBA", 1 },
-   { 0x54d0, "XY_SRC_COPY_BLT_RGB", 1 },
-   { 0x54f0, "XY_SRC_COPY_BLT_RGBA", 1 },
-   { CMD_URB_FENCE, "URB_FENCE",  1 },
-   { CMD_CONST_BUFFER_STATE, "CONST_BUFFER_STATE",  1 },
-   { CMD_CONST_BUFFER, "CONST_BUFFER",  1 },
-   { CMD_STATE_BASE_ADDRESS, "STATE_BASE_ADDRESS",  1 },
-   { CMD_STATE_INSN_POINTER, "STATE_INSN_POINTER",  1 },
-   { CMD_PIPELINE_SELECT, "PIPELINE_SELECT", 0, },
-   { CMD_PIPELINED_STATE_POINTERS, "PIPELINED_STATE_POINTERS", 1 },
-   { CMD_BINDING_TABLE_PTRS, "BINDING_TABLE_PTRS", 1 },
-   { CMD_VERTEX_BUFFER, "VERTEX_BUFFER", 1 },
-   { CMD_VERTEX_ELEMENT, "VERTEX_ELEMENT", 1 },
-   { CMD_INDEX_BUFFER, "INDEX_BUFFER", 1 },
-   { CMD_VF_STATISTICS, "VF_STATISTICS", 0 },
-   { CMD_DRAW_RECT, "DRAW_RECT", 1 },
-   { CMD_BLEND_CONSTANT_COLOR, "BLEND_CONSTANT_COLOR", 1 },
-   { CMD_CHROMA_KEY, "CHROMA_KEY", 1 },
-   { CMD_DEPTH_BUFFER, "DEPTH_BUFFER", 1 },
-   { CMD_POLY_STIPPLE_OFFSET, "POLY_STIPPLE_OFFSET", 1 },
-   { CMD_POLY_STIPPLE_PATTERN, "POLY_STIPPLE_PATTERN", 1 },
-   { CMD_LINE_STIPPLE_PATTERN, "LINE_STIPPLE_PATTERN", 1 },
-   { CMD_GLOBAL_DEPTH_OFFSET_CLAMP, "GLOBAL_DEPTH_OFFSET_CLAMP", 1 },
-   { CMD_PIPE_CONTROL, "PIPE_CONTROL", 1 },
-   { CMD_MI_FLUSH, "MI_FLUSH", 0 },
-   { CMD_3D_PRIM, "3D_PRIM", 1 },
-};
-
-#define NR_CMDS (sizeof(cmd_info)/sizeof(cmd_info[0]))
-
-
-static int find_command( unsigned int cmd )
-{
-   int i;
-
-   for (i = 0; i < NR_CMDS; i++) 
-      if (cmd == cmd_info[i].cmd) 
-	 return i;
-
-   return -1;
-}
-
-
-
-static int parse_commands( struct aub_state *s,
-			   const unsigned int *data,
-			   int len )
-{
-   while (len) {
-      int cmd = data[0] >> 16;
-      int dwords;
-      int i;
-
-      i = find_command(cmd);
-
-      if (i < 0) {
-	 _mesa_printf("couldn't find info for cmd %x\n", cmd);
-	 return 1;
-      }
-
-      if (cmd_info[i].has_length)
-	 dwords = (data[0] & 0xff) + 2;
-      else
-	 dwords = 1;
-
-      _mesa_printf("%s (%d dwords) 0x%x\n", cmd_info[i].name, dwords, data[0]);
-
-      if (len < dwords * 4) {
-	 _mesa_printf("EOF in %s (%d bytes)\n", __FUNCTION__, len);
-	 return 1;
-      }
-
-
-      if (0 && cmd == CMD_3D_PRIM)
-	 do_3d_prim(s, data, dwords * 4);
-      else
-	 flush_cmds(s, data, dwords * 4);
-
-      data += dwords;
-      len -= dwords * 4;
-   }
-
-   return 0;
-}
-
-
-
-static void parse_data_write( struct aub_state *s,
-			     const struct aub_block_header *bh,
-			     void *dest,
-			     const unsigned int *data,
-			     int len )
-{
-   switch (bh->type) {
-   case DW_GENERAL_STATE:
-      switch (bh->general_state_type) {
-      case DWGS_VERTEX_SHADER_STATE: {
-	 struct brw_vs_unit_state vs;
-	 assert(len == sizeof(vs));
-
-	 _mesa_printf("DWGS_VERTEX_SHADER_STATE\n");
-	 memcpy(&vs, data, sizeof(vs));
-
-/* 	 vs.vs6.vert_cache_disable = 1;  */
-/*  	 vs.thread4.max_threads = 4;  */
-
-	 memcpy(dest, &vs, sizeof(vs));
-	 return;
-      }
-      case DWGS_CLIPPER_STATE: {
-	 struct brw_clip_unit_state clip;
-	 assert(len == sizeof(clip));
-
-	 _mesa_printf("DWGS_CLIPPER_STATE\n");
-	 memcpy(&clip, data, sizeof(clip));
-
-/* 	 clip.thread4.max_threads = 0; */
-/*   	 clip.clip5.clip_mode = BRW_CLIPMODE_REJECT_ALL;   */
-
-	 memcpy(dest, &clip, sizeof(clip));
-	 return;
-      }
-
-      case DWGS_NOTYPE:
-      case DWGS_GEOMETRY_SHADER_STATE:
-      case DWGS_STRIPS_FANS_STATE:
-	 break;
-
-      case DWGS_WINDOWER_IZ_STATE: {
-	    struct brw_wm_unit_state wm;
-	    assert(len == sizeof(wm));
-
-	    _mesa_printf("DWGS_WINDOWER_IZ_STATE\n");
-	    memcpy(&wm, data, sizeof(wm));
-
-/* 	    wm.wm5.max_threads = 10; */
-
-	    memcpy(dest, &wm, sizeof(wm));
-	    return;
-	 }
-
-      case DWGS_COLOR_CALC_STATE:
-      case DWGS_CLIPPER_VIEWPORT_STATE:
-      case DWGS_STRIPS_FANS_VIEWPORT_STATE:
-      case DWGS_COLOR_CALC_VIEWPORT_STATE:
-      case DWGS_SAMPLER_STATE:
-      case DWGS_KERNEL_INSTRUCTIONS:
-      case DWGS_SCRATCH_SPACE:
-      case DWGS_SAMPLER_DEFAULT_COLOR:
-      case DWGS_INTERFACE_DESCRIPTOR:
-      case DWGS_VLD_STATE:
-      case DWGS_VFE_STATE:
-      default:
-	 break;
-      }
-      break;
-   case DW_SURFACE_STATE:
-      break;
-   case DW_1D_MAP:
-   case DW_2D_MAP:
-   case DW_CUBE_MAP:
-   case DW_VOLUME_MAP:
-   case DW_CONSTANT_BUFFER:
-   case DW_CONSTANT_URB_ENTRY:
-   case DW_VERTEX_BUFFER:
-   case DW_INDEX_BUFFER:
-   default:
-      break;
-   }
-
-   memcpy(dest, data, len);
-}
-
-
-/* In order to work, the memory layout has to be the same as the X
- * server which created the aubfile.
- */
-static int parse_block_header( struct aub_state *s )
-{
-   struct aub_block_header *bh = (struct aub_block_header *)(s->map + s->csr);
-   void *data = (void *)(bh + 1);
-   unsigned int len = (bh->length + 3) & ~3;
-
-   _mesa_printf("block header at 0x%x\n", s->csr);
-
-   if (s->csr + len + sizeof(*bh) > s->sz) {
-      _mesa_printf("EOF in data in %s\n", __FUNCTION__);
-      return 1;
-   }
-
-   if (bh->address_space == ADDR_GTT) {
-
-      switch (bh->operation)
-      {
-      case BH_DATA_WRITE: {
-	 void *dest = bmFindVirtual( s->intel, bh->address, len );
-	 if (dest == NULL) {
-	    _mesa_printf("Couldn't find virtual address for offset %x\n", bh->address);
-	    return 1;
-	 }
-
-#if 1
-	 parse_data_write(s, bh, dest, data, len);
-#else
-	 memcpy(dest, data, len);
-#endif
-	 break;
-      }
-      case BH_COMMAND_WRITE:
-#if 0
-	 intel_cmd_ioctl(s->intel, (void *)data, len);
-#else
-	 if (parse_commands(s, data, len) != 0)
-	    _mesa_printf("parse_commands failed\n");
-#endif
-	 break;
-      default:
-	 break;
-      }
-   }
-
-   s->csr += sizeof(*bh) + len;
-   return 0;
-}
-
-
-#define AUB_FILE_HEADER 0xe085000b
-#define AUB_BLOCK_HEADER 0xe0c10003
-#define AUB_DUMP_BMP 0xe09e0004
-
-int brw_playback_aubfile(struct brw_context *brw,
-			 const char *filename)
-{
-   struct intel_context *intel = &brw->intel;
-   struct aub_state state;
-   struct stat sb;
-   int fd;
-   int retval = 0;
-
-   state.intel = intel;
-
-   fd = open(filename, O_RDONLY, 0);
-   if (fd < 0) {
-      _mesa_printf("couldn't open aubfile: %s\n", filename);
-      return 1;
-   }
-
-   if (fstat(fd, &sb) != 0) {
-      _mesa_printf("couldn't open %s\n", filename);
-      return 1;
-   }
-
-   state.csr = 0;
-   state.sz = sb.st_size;
-   state.map = mmap(NULL, sb.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
-   
-   if (state.map == NULL) {
-      _mesa_printf("couldn't mmap %s\n", filename);
-      return 1;
-   }
-
-   LOCK_HARDWARE(intel); 
-   {
-      /* Make sure we don't confuse anything that might happen to be
-       * going on with the hardware:
-       */
-/*       bmEvictAll(intel); */
-/*       intel->vtbl.lost_hardware(intel); */
-      
-
-      /* Replay the aubfile item by item: 
-       */
-      while (retval == 0 && 
-	     state.csr != state.sz) {
-	 unsigned int insn = *(unsigned int *)(state.map + state.csr);
-
-	 switch (insn) {
-	 case AUB_FILE_HEADER:
-	    retval = gobble(&state, sizeof(struct aub_file_header));
-	    break;
-	 
-	 case AUB_BLOCK_HEADER:   
-	    retval = parse_block_header(&state);
-	    break;
-	 
-	 case AUB_DUMP_BMP:
-	    retval = gobble(&state, sizeof(struct aub_dump_bmp));
-	    break;
-	 
-	 default:
-	    _mesa_printf("unknown instruction %x\n", insn);
-	    retval = 1;
-	    break;
-	 }
-      }
-   }
-   UNLOCK_HARDWARE(intel);
-   return retval;
-}
-
-
-
-
-
-
-		  
diff --git a/src/mesa/drivers/dri/i965/brw_clip.c b/src/mesa/drivers/dri/i965/brw_clip.c
index 3bec153075a..8287fd9edf3 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.c
+++ b/src/mesa/drivers/dri/i965/brw_clip.c
@@ -212,6 +212,10 @@ static void upload_clip_prog( struct brw_context *brw )
 	    }
 	 }
 
+    if (brw->attribs.Polygon->BackMode != GL_FILL ||
+        brw->attribs.Polygon->FrontMode != GL_FILL)
+        key.do_unfilled = 1;
+
 	 /* Most cases the fixed function units will handle.  Cases where
 	  * one or more polygon faces are unfilled will require help:
 	  */
diff --git a/src/mesa/drivers/dri/i965/brw_clip_line.c b/src/mesa/drivers/dri/i965/brw_clip_line.c
index 83182270eac..9ad00676d4b 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_line.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_line.c
@@ -130,6 +130,7 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
    struct brw_instruction *plane_loop;
    struct brw_instruction *plane_active;
    struct brw_instruction *is_negative;
+   struct brw_instruction *is_neg2;
    struct brw_instruction *not_culled;
    struct brw_reg v1_null_ud = retype(vec1(brw_null_reg()), BRW_REGISTER_TYPE_UD);
 
@@ -146,6 +147,13 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
    brw_clip_init_planes(c);
    brw_clip_init_clipmask(c);
 
+   /* -ve rhw workaround */
+   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+   brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2),
+	   brw_imm_ud(1<<20));
+   brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(0x3f));
+   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
    plane_loop = brw_DO(p, BRW_EXECUTE_1);
    {
       /* if (planemask & 1)
@@ -183,13 +191,20 @@ static void clip_and_emit_line( struct brw_clip_compile *c )
 	    /* Coming back in.  We know that both cannot be negative
 	     * because the line would have been culled in that case.
 	     */
-	    brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
-	    brw_math_invert(p, c->reg.t, c->reg.t);
-	    brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
 
-	    brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
-	    brw_MOV(p, c->reg.t0, c->reg.t);
-	    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	    /* If both are positive, do nothing */
+             brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_L, c->reg.dp0, brw_imm_f(0.0));
+             is_neg2 = brw_IF(p, BRW_EXECUTE_1);
+             {
+		brw_ADD(p, c->reg.t, c->reg.dp0, negate(c->reg.dp1));
+		brw_math_invert(p, c->reg.t, c->reg.t);
+		brw_MUL(p, c->reg.t, c->reg.t, c->reg.dp0);
+
+		brw_CMP(p, vec1(brw_null_reg()), BRW_CONDITIONAL_G, c->reg.t, c->reg.t0 );
+		brw_MOV(p, c->reg.t0, c->reg.t);
+		brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+	     }
+	     brw_ENDIF(p, is_neg2);
 	 }
 	 brw_ENDIF(p, is_negative);	 
       }
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index 1e6d6fa1762..ae46d7a86e0 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -55,7 +55,7 @@ static void upload_clip_unit( struct brw_context *brw )
    /* BRW_NEW_URB_FENCE */
    clip.thread4.nr_urb_entries = brw->urb.nr_clip_entries; 
    clip.thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
-   clip.thread4.max_threads = 0; /* Hmm, maybe the max is 1 or 2 threads */
+   clip.thread4.max_threads = 1; /* 2 threads */
 
    if (INTEL_DEBUG & DEBUG_STATS)
       clip.thread4.stats_enable = 1; 
diff --git a/src/mesa/drivers/dri/i965/brw_clip_tri.c b/src/mesa/drivers/dri/i965/brw_clip_tri.c
index f62b02cedfd..b7d30bb0c4d 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_tri.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_tri.c
@@ -42,6 +42,20 @@
 #include "brw_util.h"
 #include "brw_clip.h"
 
+static struct brw_reg get_tmp( struct brw_clip_compile *c )
+{
+   struct brw_reg tmp = brw_vec4_grf(c->last_tmp, 0);
+
+   if (++c->last_tmp > c->prog_data.total_grf)
+      c->prog_data.total_grf = c->last_tmp;
+
+   return tmp;
+}
+
+static void release_tmps( struct brw_clip_compile *c )
+{
+   c->last_tmp = c->first_tmp;
+}
 
 
 void brw_clip_tri_alloc_regs( struct brw_clip_compile *c, 
@@ -435,15 +449,102 @@ static void maybe_do_clip_tri( struct brw_clip_compile *c )
    brw_ENDIF(p, do_clip);
 }
 
-
+static void brw_clip_test( struct brw_clip_compile *c )
+{
+    struct brw_reg t = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t1 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t2 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+    struct brw_reg t3 = retype(get_tmp(c), BRW_REGISTER_TYPE_UD);
+
+    struct brw_reg v0 = get_tmp(c);
+    struct brw_reg v1 = get_tmp(c);
+    struct brw_reg v2 = get_tmp(c);
+
+    struct brw_indirect vt0 = brw_indirect(0, 0);
+    struct brw_indirect vt1 = brw_indirect(1, 0);
+    struct brw_indirect vt2 = brw_indirect(2, 0);
+
+    struct brw_compile *p = &c->func;
+
+    brw_MOV(p, get_addr_reg(vt0), brw_address(c->reg.vertex[0]));
+    brw_MOV(p, get_addr_reg(vt1), brw_address(c->reg.vertex[1]));
+    brw_MOV(p, get_addr_reg(vt2), brw_address(c->reg.vertex[2]));
+    brw_MOV(p, v0, deref_4f(vt0, c->offset[VERT_RESULT_HPOS]));
+    brw_MOV(p, v1, deref_4f(vt1, c->offset[VERT_RESULT_HPOS]));
+    brw_MOV(p, v2, deref_4f(vt2, c->offset[VERT_RESULT_HPOS]));
+
+    /* test nearz, xmin, ymin plane */
+    brw_CMP(p, t1, BRW_CONDITIONAL_LE, negate(v0), get_element(v0, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t2, BRW_CONDITIONAL_LE, negate(v1), get_element(v1, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t3, BRW_CONDITIONAL_LE, negate(v2), get_element(v2, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_XOR(p, t, t1, t2);
+    brw_XOR(p, t1, t2, t3);
+    brw_OR(p, t, t, t1);
+
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 0), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<5)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 1), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<3)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 2), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<1)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    /* test farz, xmax, ymax plane */
+    brw_CMP(p, t1, BRW_CONDITIONAL_L, v0, get_element(v0, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t2, BRW_CONDITIONAL_L, v1, get_element(v1, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, t3, BRW_CONDITIONAL_L, v2, get_element(v2, 3)); 
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    brw_XOR(p, t, t1, t2);
+    brw_XOR(p, t1, t2, t3);
+    brw_OR(p, t, t, t1);
+
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 0), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<4)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 1), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<2)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+    brw_CMP(p, brw_null_reg(), BRW_CONDITIONAL_NZ, 
+	    get_element(t, 2), brw_imm_ud(0));
+    brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud((1<<0)));
+    brw_set_predicate_control(p, BRW_PREDICATE_NONE);
+
+    release_tmps(c);
+}
 
 
 void brw_emit_tri_clip( struct brw_clip_compile *c )
 {
+   struct brw_instruction *neg_rhw;
+   struct brw_compile *p = &c->func;
    brw_clip_tri_alloc_regs(c, 3 + c->key.nr_userclip + 6);
    brw_clip_tri_init_vertices(c);
    brw_clip_init_clipmask(c);
 
+   /* if -ve rhw workaround bit is set, 
+      do cliptest */
+   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
+   brw_AND(p, brw_null_reg(), get_element_ud(c->reg.R0, 2), 
+	   brw_imm_ud(1<<20));
+   neg_rhw = brw_IF(p, BRW_EXECUTE_1); 
+   {
+       brw_clip_test(c);
+   }
+   brw_ENDIF(p, neg_rhw);
+
    /* Can't push into do_clip_tri because with polygon (or quad)
     * flatshading, need to apply the flatshade here because we don't
     * respect the PV when converting to trifan for emit:
@@ -462,6 +563,3 @@ void brw_emit_tri_clip( struct brw_clip_compile *c )
     */
    brw_clip_kill_thread(c);
 }
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_clip_util.c b/src/mesa/drivers/dri/i965/brw_clip_util.c
index 19bef19801a..c37bfeb1ce6 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_util.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_util.c
@@ -272,6 +272,7 @@ void brw_clip_kill_thread(struct brw_clip_compile *c)
 
 
 
+
 struct brw_reg brw_clip_plane0_address( struct brw_clip_compile *c )
 {
    return brw_address(c->reg.fixed_planes);
@@ -327,8 +328,7 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c )
    
    /* Shift so that lowest outcode bit is rightmost: 
     */
-   brw_MOV(p, c->reg.planemask, incoming);
-   brw_SHR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(26));
+   brw_SHR(p, c->reg.planemask, incoming, brw_imm_ud(26));
 
    if (c->key.nr_userclip) {
       struct brw_reg tmp = retype(vec1(get_tmp(c)), BRW_REGISTER_TYPE_UD);
@@ -342,13 +342,5 @@ void brw_clip_init_clipmask( struct brw_clip_compile *c )
       
       release_tmp(c, tmp);
    }
-
-   /* Test for -ve rhw workaround 
-    */
-   brw_set_conditionalmod(p, BRW_CONDITIONAL_NZ);
-   brw_AND(p, vec1(brw_null_reg()), incoming, brw_imm_ud(1<<20));
-   brw_OR(p, c->reg.planemask, c->reg.planemask, brw_imm_ud(0x3f));
-   brw_set_predicate_control(p, BRW_PREDICATE_NONE);
-
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index 397a9bd3f5c..0ccdd8a6610 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -31,7 +31,6 @@
 
 
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_defines.h"
 #include "brw_draw.h"
 #include "brw_vs.h"
@@ -135,8 +134,6 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    driInitExtensions( ctx, brw_extensions, GL_FALSE );
 
-   brw_aub_init( brw );
-
    brw_init_attribs( brw );
    brw_init_metaops( brw );
    brw_init_state( brw );
@@ -156,14 +153,6 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    brw_FrameBufferTexInit( brw );
 
-   {
-      const char *filename = getenv("INTEL_REPLAY");
-      if (filename) {
-	 brw_playback_aubfile(brw, filename);
-	 exit(0);
-      }
-   }
-
    return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 08fdc545205..aa797b72ce6 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -277,9 +277,6 @@ struct brw_cache {
    GLuint key_size;		/* for fixed-size keys */
    GLuint aux_size;
 
-   GLuint aub_type;
-   GLuint aub_sub_type;
-   
    GLuint last_addr;			/* offset of active item */
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index 3f0aaa1f86d..d3c88c1dca0 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -42,7 +42,6 @@
 #include "brw_defines.h"
 #include "brw_state.h"
 #include "brw_util.h"
-#include "brw_aub.h"
 
 
 /* Partition the CURBE between the various users of constant values:
@@ -90,7 +89,7 @@ static void calculate_curbe_offsets( struct brw_context *brw )
     */
    if (nr_fp_regs > brw->curbe.wm_size ||
        nr_vp_regs > brw->curbe.vs_size ||
-       nr_clip_regs > brw->curbe.clip_size ||
+       nr_clip_regs != brw->curbe.clip_size ||
        (total_regs < brw->curbe.total_size / 4 &&
 	brw->curbe.total_size > 16)) {
 
@@ -315,13 +314,11 @@ static void upload_constant_buffer(struct brw_context *brw)
 
       /* Copy data to the buffer:
        */
-      bmBufferSubDataAUB(&brw->intel,
-			 pool->buffer,
-			 brw->curbe.gs_offset, 
-			 bufsz, 
-			 buf,
-			 DW_CONSTANT_BUFFER,
-			 0);
+      bmBufferSubData(&brw->intel,
+		      pool->buffer,
+		      brw->curbe.gs_offset,
+		      bufsz,
+		      buf);
    }
 
    /* TODO: only emit the constant_buffer packet when necessary, ie:
diff --git a/src/mesa/drivers/dri/i965/brw_defines.h b/src/mesa/drivers/dri/i965/brw_defines.h
index e8f878a7018..9bb7d2f703d 100644
--- a/src/mesa/drivers/dri/i965/brw_defines.h
+++ b/src/mesa/drivers/dri/i965/brw_defines.h
@@ -240,6 +240,8 @@
 #define BRW_FRONTWINDING_CW      0
 #define BRW_FRONTWINDING_CCW     1
 
+#define BRW_SPRITE_POINT_ENABLE  16
+
 #define BRW_INDEX_BYTE     0
 #define BRW_INDEX_WORD     1
 #define BRW_INDEX_DWORD    2
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 0c64d7e756d..b23b3579888 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -36,7 +36,6 @@
 #include "brw_draw.h"
 #include "brw_defines.h"
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_state.h"
 #include "brw_fallback.h"
 
@@ -331,6 +330,7 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
       else {
 	 /* Otherwise, explicitly do the cliprects at this point:
 	  */
+          GLuint nprims = 0;
 	 for (j = 0; j < brw->intel.numClipRects; j++) {
 	    brw_emit_cliprect(brw, &brw->intel.pClipRects[j]);
 
@@ -338,6 +338,11 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
 	     */
 	    for (i = 0; i < nr_prims; i++) {
 	       brw_emit_prim(brw, &prim[i]);   
+
+          if (++nprims == VBO_MAX_PRIM) {
+              intel_batchbuffer_flush(brw->intel.batch);
+              nprims = 0;
+          }
 	    }
 	 }
       }
@@ -462,11 +467,6 @@ void brw_draw_prims( GLcontext *ctx,
        _swsetup_Wakeup(ctx);
       _tnl_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
    }
-
-   if (intel->aub_file && (INTEL_DEBUG & DEBUG_SYNC)) {
-      intelFinish( &intel->ctx );
-      intel->aub_wrap = 1;
-   }
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 6150cac4aa3..fc2e3035af1 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -36,7 +36,6 @@
 #include "brw_draw.h"
 #include "brw_defines.h"
 #include "brw_context.h"
-#include "brw_aub.h"
 #include "brw_state.h"
 #include "brw_fallback.h"
 
@@ -593,6 +592,28 @@ void brw_upload_indices( struct brw_context *brw,
 				 ib_size,
 				 index_buffer->ptr,
 				 bufferobj);
+   } else {
+       if (((1 << get_index_type(index_buffer->type)) - 1) & offset) {
+           struct gl_buffer_object *vbo;
+           GLuint voffset;
+           GLubyte *map = ctx->Driver.MapBuffer(ctx,
+                                                GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                                GL_DYNAMIC_DRAW_ARB,
+                                                bufferobj);
+           map += offset;
+           get_space(brw, ib_size, &vbo, &voffset);
+           
+           ctx->Driver.BufferSubData(ctx,
+                                     GL_ELEMENT_ARRAY_BUFFER_ARB,
+                                     voffset,
+                                     ib_size,
+                                     map,
+                                     vbo);
+           ctx->Driver.UnmapBuffer(ctx, GL_ELEMENT_ARRAY_BUFFER_ARB, bufferobj);
+
+           bufferobj = vbo;
+           offset = voffset;
+       }
    }
 
    /* Emit the indexbuffer packet:
diff --git a/src/mesa/drivers/dri/i965/brw_program.c b/src/mesa/drivers/dri/i965/brw_program.c
index 752fe49bcbf..8e8fea48e9f 100644
--- a/src/mesa/drivers/dri/i965/brw_program.c
+++ b/src/mesa/drivers/dri/i965/brw_program.c
@@ -29,15 +29,14 @@
   *   Keith Whitwell <[email protected]>
   */
   
+#include "main/imports.h"
+#include "main/enums.h"
 #include "shader/prog_parameter.h"
-#include "brw_context.h"
-#include "brw_aub.h"
-#include "brw_util.h"
-#include "program.h"
-#include "imports.h"
-#include "enums.h"
+#include "shader/program.h"
 #include "tnl/tnl.h"
 
+#include "brw_context.h"
+#include "brw_util.h"
 
 static void brwBindProgram( GLcontext *ctx,
 			    GLenum target, 
diff --git a/src/mesa/drivers/dri/i965/brw_sf.c b/src/mesa/drivers/dri/i965/brw_sf.c
index d5175399d6c..738ceb0552e 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.c
+++ b/src/mesa/drivers/dri/i965/brw_sf.c
@@ -43,7 +43,7 @@
 #include "brw_sf.h"
 #include "brw_state.h"
 
-#define DO_SETUP_BITS ((1<<FRAG_ATTRIB_MAX)-1)
+#define DO_SETUP_BITS ((1<<(FRAG_ATTRIB_MAX)) - 1)
 
 static void compile_sf_prog( struct brw_context *brw,
 			     struct brw_sf_prog_key *key )
@@ -74,6 +74,11 @@ static void compile_sf_prog( struct brw_context *brw,
       if (c.key.attrs & (1<<i)) {
 	 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;
 	 idx++;
       }
    
@@ -90,7 +95,10 @@ static void compile_sf_prog( struct brw_context *brw,
       break;
    case SF_POINTS:
       c.nr_verts = 1;
-      brw_emit_point_setup( &c );
+      if (key->do_point_sprite)
+	  brw_emit_point_sprite_setup( &c );
+      else
+	  brw_emit_point_setup( &c );
       break;
    case SF_UNFILLED_TRIS:
       c.nr_verts = 3;
@@ -162,7 +170,8 @@ static void upload_sf_prog( struct brw_context *brw )
       break;
    }
 
-
+   key.do_point_sprite = brw->attribs.Point->PointSprite;
+   key.SpriteOrigin = brw->attribs.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);
@@ -179,7 +188,7 @@ static void upload_sf_prog( struct brw_context *brw )
 
 const struct brw_tracked_state brw_sf_prog = {
    .dirty = {
-      .mesa  = (_NEW_LIGHT|_NEW_POLYGON),
+      .mesa  = (_NEW_LIGHT|_NEW_POLYGON|_NEW_POINT),
       .brw   = (BRW_NEW_REDUCED_PRIMITIVE),
       .cache = CACHE_NEW_VS_PROG
    },
diff --git a/src/mesa/drivers/dri/i965/brw_sf.h b/src/mesa/drivers/dri/i965/brw_sf.h
index fb72b84ba8a..e8946511dd4 100644
--- a/src/mesa/drivers/dri/i965/brw_sf.h
+++ b/src/mesa/drivers/dri/i965/brw_sf.h
@@ -34,9 +34,9 @@
 #define BRW_SF_H
 
 
+#include "shader/program.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "program.h"
 
 
 #define SF_POINTS    0
@@ -50,9 +50,14 @@ struct brw_sf_prog_key {
    GLuint do_flat_shading:1;
    GLuint attrs:16;
    GLuint frontface_ccw:1;
-   GLuint pad:11;
+   GLuint do_point_sprite:1;
+   GLuint pad:10;
+   GLenum SpriteOrigin;
 };
 
+struct brw_sf_point_tex {
+	GLboolean CoordReplace;	
+};
 
 struct brw_sf_compile {
    struct brw_compile func;
@@ -94,12 +99,14 @@ struct brw_sf_compile {
 
    GLubyte attr_to_idx[VERT_RESULT_MAX];   
    GLubyte idx_to_attr[VERT_RESULT_MAX];   
+   struct brw_sf_point_tex point_attrs[VERT_RESULT_MAX];
 };
 
  
 void brw_emit_tri_setup( struct brw_sf_compile *c );
 void brw_emit_line_setup( struct brw_sf_compile *c );
 void brw_emit_point_setup( struct brw_sf_compile *c );
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c );
 void brw_emit_anyprim_setup( struct brw_sf_compile *c );
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_sf_emit.c b/src/mesa/drivers/dri/i965/brw_sf_emit.c
index cbaf018c44a..5e86e428fa8 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_emit.c
@@ -497,6 +497,86 @@ void brw_emit_line_setup( struct brw_sf_compile *c )
    } 
 }
 
+void brw_emit_point_sprite_setup( struct brw_sf_compile *c )
+{
+   struct brw_compile *p = &c->func;
+   GLuint i;
+
+   c->nr_verts = 1;
+   alloc_regs(c);
+   copy_z_inv_w(c);
+   for (i = 0; i < c->nr_setup_regs; i++)
+   {
+      struct brw_sf_point_tex *tex = &c->point_attrs[c->idx_to_attr[2*i]];
+      struct brw_reg a0 = offset(c->vert[0], i);
+      GLushort pc, pc_persp, pc_linear;
+      GLboolean last = calculate_masks(c, i, &pc, &pc_persp, &pc_linear);
+            
+      if (pc_persp)
+      {				
+	  if (!tex->CoordReplace) {
+	      brw_set_predicate_control_flag_value(p, pc_persp);
+	      brw_MUL(p, a0, a0, c->inv_w[0]);
+	  }
+      }
+
+      if (tex->CoordReplace) {
+	  /* Caculate 1.0/PointWidth */
+	  brw_math(&c->func,
+		  c->tmp,
+		  BRW_MATH_FUNCTION_INV,
+		  BRW_MATH_SATURATE_NONE,
+		  0,
+		  c->dx0,
+		  BRW_MATH_DATA_SCALAR,
+		  BRW_MATH_PRECISION_FULL);
+
+	  if (c->key.SpriteOrigin == GL_UPPER_LEFT) {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, negate(c->inv_w[0]));
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  } else {
+	   	brw_MUL(p, c->m1Cx, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m1Cx, 1)), brw_imm_f(0.0));
+	  	brw_MUL(p, c->m2Cy, c->tmp, c->inv_w[0]);
+		brw_MOV(p, vec1(suboffset(c->m2Cy, 0)), brw_imm_f(0.0));
+	  }
+      } else {
+	  brw_MOV(p, c->m1Cx, brw_imm_ud(0));
+	  brw_MOV(p, c->m2Cy, brw_imm_ud(0));
+      }
+
+      {
+	 brw_set_predicate_control_flag_value(p, pc); 
+	 if (tex->CoordReplace) {
+	     if (c->key.SpriteOrigin == GL_UPPER_LEFT) {
+		 brw_MUL(p, c->m3C0, c->inv_w[0], brw_imm_f(1.0));
+		 brw_MOV(p, vec1(suboffset(c->m3C0, 0)), brw_imm_f(0.0));
+	     }
+	     else
+		 brw_MOV(p, c->m3C0, brw_imm_f(0.0));
+	 } else {
+	 	brw_MOV(p, c->m3C0, a0); /* constant value */
+	 }
+
+	 /* Copy m0..m3 to URB. 
+	  */
+	 brw_urb_WRITE(p, 
+		       brw_null_reg(),
+		       0,
+		       brw_vec8_grf(0, 0),
+		       0, 	/* allocate */
+		       1,	/* used */
+		       4, 	/* msg len */
+		       0,	/* response len */
+		       last, 	/* eot */
+		       last, 	/* writes complete */
+		       i*4,	/* urb destination offset */
+		       BRW_URB_SWIZZLE_TRANSPOSE);
+      }
+   }
+}
 
 /* Points setup - several simplifications as all attributes are
  * constant across the face of the point (point sprites excluded!)
@@ -561,6 +641,7 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
    struct brw_compile *p = &c->func;
    struct brw_reg ip = brw_ip_reg();
    struct brw_reg payload_prim = brw_uw1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0);
+   struct brw_reg payload_attr = get_element_ud(brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, 1, 0), 0); 
    struct brw_reg primmask;
    struct brw_instruction *jmp;
    struct brw_reg v1_null_ud = vec1(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD));
@@ -582,7 +663,9 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
 					       (1<<_3DPRIM_TRIFAN_NOSTIPPLE)));
    jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
    {
+      brw_push_insn_state(p); 
       brw_emit_tri_setup( c );
+      brw_pop_insn_state(p);
       /* note - thread killed in subroutine */
    }
    brw_land_fwd_jump(p, jmp);
@@ -596,11 +679,23 @@ void brw_emit_anyprim_setup( struct brw_sf_compile *c )
 					       (1<<_3DPRIM_LINESTRIP_CONT_BF)));
    jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
    {
+      brw_push_insn_state(p); 
       brw_emit_line_setup( c );
+      brw_pop_insn_state(p);
       /* note - thread killed in subroutine */
    }
    brw_land_fwd_jump(p, jmp); 
 
+   brw_set_conditionalmod(p, BRW_CONDITIONAL_Z);
+   brw_AND(p, v1_null_ud, payload_attr, brw_imm_ud(1<<BRW_SPRITE_POINT_ENABLE));
+   jmp = brw_JMPI(p, ip, ip, brw_imm_w(0));
+   {
+      brw_push_insn_state(p); 
+      brw_emit_point_sprite_setup( c );
+      brw_pop_insn_state(p);
+   }
+   brw_land_fwd_jump(p, jmp); 
+
    brw_emit_point_setup( c );
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index 9a6e5f5f192..236c6fd42a5 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -185,7 +185,9 @@ static void upload_sf_unit( struct brw_context *brw )
    /* _NEW_POINT */
    sf.sf6.point_rast_rule = 1;	/* opengl conventions */
    /* XXX clamp max depends on AA vs. non-AA */
-   sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 3.0) * (1<<3);
+
+   sf.sf7.sprite_point = brw->attribs.Point->PointSprite;
+   sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 255.0) * (1<<3);
    sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated;
       
    /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
diff --git a/src/mesa/drivers/dri/i965/brw_state_batch.c b/src/mesa/drivers/dri/i965/brw_state_batch.c
index 909b0acd121..b78b51328a2 100644
--- a/src/mesa/drivers/dri/i965/brw_state_batch.c
+++ b/src/mesa/drivers/dri/i965/brw_state_batch.c
@@ -32,7 +32,6 @@
 
 
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "intel_batchbuffer.h"
 #include "imports.h"
 
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index 71c6938f9a3..98d765ac0e0 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -31,7 +31,6 @@
       
 
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "intel_batchbuffer.h"
 #include "imports.h"
 
@@ -188,13 +187,11 @@ GLuint brw_upload_cache( struct brw_cache *cache,
 
    /* Copy data to the buffer:
     */
-   bmBufferSubDataAUB(&cache->brw->intel,
-		      cache->pool->buffer,
-		      offset, 
-		      data_size, 
-		      data,
-		      cache->aub_type,
-		      cache->aub_sub_type);
+   bmBufferSubData(&cache->brw->intel,
+		   cache->pool->buffer,
+		   offset,
+		   data_size,
+		   data);
    
 
    cache->brw->state.dirty.cache |= 1<<cache->id;
@@ -227,17 +224,17 @@ GLuint brw_cache_data(struct brw_cache *cache,
    return brw_cache_data_sz(cache, data, cache->key_size);
 }
 
-
-
-
+enum pool_type {
+   DW_SURFACE_STATE,
+   DW_GENERAL_STATE
+};
 
 static void brw_init_cache( struct brw_context *brw, 
 			    const char *name,
 			    GLuint id,
 			    GLuint key_size,
 			    GLuint aux_size,
-			    GLuint aub_type,
-			    GLuint aub_sub_type )
+			    enum pool_type pool_type)
 {
    struct brw_cache *cache = &brw->cache[id];
    cache->brw = brw;
@@ -254,9 +251,7 @@ static void brw_init_cache( struct brw_context *brw,
 
    cache->key_size = key_size;
    cache->aux_size = aux_size;
-   cache->aub_type = aub_type;
-   cache->aub_sub_type = aub_sub_type;
-   switch (aub_type) {
+   switch (pool_type) {
    case DW_GENERAL_STATE: cache->pool = &brw->pool[BRW_GS_POOL]; break;
    case DW_SURFACE_STATE: cache->pool = &brw->pool[BRW_SS_POOL]; break;
    default: assert(0); break;
@@ -271,136 +266,119 @@ void brw_init_caches( struct brw_context *brw )
 		  BRW_CC_VP,
 		  sizeof(struct brw_cc_viewport),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_COLOR_CALC_VIEWPORT_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "CC_UNIT",
 		  BRW_CC_UNIT,
 		  sizeof(struct brw_cc_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_COLOR_CALC_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "WM_PROG",
 		  BRW_WM_PROG,
 		  sizeof(struct brw_wm_prog_key),
 		  sizeof(struct brw_wm_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SAMPLER_DEFAULT_COLOR",
 		  BRW_SAMPLER_DEFAULT_COLOR,
 		  sizeof(struct brw_sampler_default_color),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_SAMPLER_DEFAULT_COLOR);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SAMPLER",
 		  BRW_SAMPLER,
 		  0,		/* variable key/data size */
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_SAMPLER_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "WM_UNIT",
 		  BRW_WM_UNIT,
 		  sizeof(struct brw_wm_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_WINDOWER_IZ_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SF_PROG",
 		  BRW_SF_PROG,
 		  sizeof(struct brw_sf_prog_key),
 		  sizeof(struct brw_sf_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SF_VP",
 		  BRW_SF_VP,
 		  sizeof(struct brw_sf_viewport),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_STRIPS_FANS_VIEWPORT_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SF_UNIT",
 		  BRW_SF_UNIT,
 		  sizeof(struct brw_sf_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_STRIPS_FANS_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "VS_UNIT",
 		  BRW_VS_UNIT,
 		  sizeof(struct brw_vs_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_VERTEX_SHADER_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "VS_PROG",
 		  BRW_VS_PROG,
 		  sizeof(struct brw_vs_prog_key),
 		  sizeof(struct brw_vs_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "CLIP_UNIT",
 		  BRW_CLIP_UNIT,
 		  sizeof(struct brw_clip_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_CLIPPER_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "CLIP_PROG",
 		  BRW_CLIP_PROG,
 		  sizeof(struct brw_clip_prog_key),
 		  sizeof(struct brw_clip_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "GS_UNIT",
 		  BRW_GS_UNIT,
 		  sizeof(struct brw_gs_unit_state),
 		  0,
-		  DW_GENERAL_STATE,
-		  DWGS_GEOMETRY_SHADER_STATE);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "GS_PROG",
 		  BRW_GS_PROG,
 		  sizeof(struct brw_gs_prog_key),
 		  sizeof(struct brw_gs_prog_data),
-		  DW_GENERAL_STATE,
-		  DWGS_KERNEL_INSTRUCTIONS);
+		  DW_GENERAL_STATE);
 
    brw_init_cache(brw,
 		  "SS_SURFACE",
 		  BRW_SS_SURFACE,
 		  sizeof(struct brw_surface_state),
 		  0,
-		  DW_SURFACE_STATE,
-		  DWSS_SURFACE_STATE);
+		  DW_SURFACE_STATE);
 
    brw_init_cache(brw,
 		  "SS_SURF_BIND",
 		  BRW_SS_SURF_BIND,
 		  sizeof(struct brw_surface_binding_table),
 		  0,
-		  DW_SURFACE_STATE,
-		  DWSS_BINDING_TABLE_STATE);
+		  DW_SURFACE_STATE);
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_state_pool.c b/src/mesa/drivers/dri/i965/brw_state_pool.c
index b9926f2a5d7..708ae857ab5 100644
--- a/src/mesa/drivers/dri/i965/brw_state_pool.c
+++ b/src/mesa/drivers/dri/i965/brw_state_pool.c
@@ -126,10 +126,7 @@ void brw_pool_check_wrap( struct brw_context *brw,
 			  struct brw_mem_pool *pool )
 {
    if (pool->offset > (pool->size * 3) / 4) {
-      if (brw->intel.aub_file)
-	 brw->intel.aub_wrap = 1;
-      else
-	 brw->state.dirty.brw |= BRW_NEW_CONTEXT;
+      brw->state.dirty.brw |= BRW_NEW_CONTEXT;
    }
 
 }
diff --git a/src/mesa/drivers/dri/i965/brw_tex.c b/src/mesa/drivers/dri/i965/brw_tex.c
index 9d4b9867d24..ad29316a42a 100644
--- a/src/mesa/drivers/dri/i965/brw_tex.c
+++ b/src/mesa/drivers/dri/i965/brw_tex.c
@@ -154,13 +154,19 @@ brwChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
 
    case GL_RGB_S3TC:
    case GL_RGB4_S3TC:
+   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
+       return &_mesa_texformat_rgb_dxt1;
+
+   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
+       return &_mesa_texformat_rgba_dxt1;
+
    case GL_RGBA_S3TC:
    case GL_RGBA4_S3TC:
    case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT:
+       return &_mesa_texformat_rgba_dxt3;
+
    case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT:
-   case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT:
-   case GL_COMPRESSED_RGB_S3TC_DXT1_EXT:
-     return &_mesa_texformat_rgb_dxt1; /* there is no rgba support? */
+       return &_mesa_texformat_rgba_dxt5;
 
    case GL_DEPTH_COMPONENT:
    case GL_DEPTH_COMPONENT16:
@@ -168,6 +174,25 @@ brwChooseTextureFormat( GLcontext *ctx, GLint internalFormat,
    case GL_DEPTH_COMPONENT32:
       return &_mesa_texformat_z16;
 
+   case GL_SRGB_EXT:
+   case GL_SRGB8_EXT:
+   case GL_SRGB_ALPHA_EXT:
+   case GL_SRGB8_ALPHA8_EXT:
+   case GL_SLUMINANCE_EXT:
+   case GL_SLUMINANCE8_EXT:
+   case GL_SLUMINANCE_ALPHA_EXT:
+   case GL_SLUMINANCE8_ALPHA8_EXT:
+   case GL_COMPRESSED_SRGB_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_EXT:
+   case GL_COMPRESSED_SLUMINANCE_EXT:
+   case GL_COMPRESSED_SLUMINANCE_ALPHA_EXT:
+	return &_mesa_texformat_srgba8;
+   case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT3_EXT:
+   case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT:
+     return &_mesa_texformat_srgb_dxt1;
+
    default:
       fprintf(stderr, "unexpected texture format %s in %s\n", 
 	      _mesa_lookup_enum_by_nr(internalFormat),
diff --git a/src/mesa/drivers/dri/i965/brw_tex_layout.c b/src/mesa/drivers/dri/i965/brw_tex_layout.c
index af1ad0f1ef1..2094a1c8ad7 100644
--- a/src/mesa/drivers/dri/i965/brw_tex_layout.c
+++ b/src/mesa/drivers/dri/i965/brw_tex_layout.c
@@ -37,6 +37,7 @@
 #include "intel_tex_layout.h"
 #include "macros.h"
 
+#define ALIGN(value, alignment)  ((value + alignment - 1) & ~(alignment - 1))
 
 GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 {
@@ -53,11 +54,20 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
       GLuint pack_x_pitch, pack_x_nr;
       GLuint pack_y_pitch;
       GLuint level;
+      GLuint align_h = 2;
+      GLuint align_w = 4;
 
-      mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
       mt->total_height = 0;
+      
+      if (mt->compressed) {
+          align_w = intel_compressed_alignment(mt->internal_format);
+          mt->pitch = ALIGN(width, align_w);
+          pack_y_pitch = (height + 3) / 4;
+      } else {
+          mt->pitch = ((mt->width0 * mt->cpp + 3) & ~3) / mt->cpp;
+          pack_y_pitch = ALIGN(mt->height0, align_h);
+      }
 
-      pack_y_pitch = MAX2(mt->height0, 2);
       pack_x_pitch = mt->pitch;
       pack_x_nr = 1;
 
@@ -83,20 +93,30 @@ GLboolean brw_miptree_layout( struct intel_mipmap_tree *mt )
 
 
 	 mt->total_height += y;
-
-	 if (pack_x_pitch > 4) {
-	    pack_x_pitch >>= 1;
-	    pack_x_nr <<= 1;
-	    assert(pack_x_pitch * pack_x_nr <= mt->pitch);
-	 }
-
-	 if (pack_y_pitch > 2) {
-	    pack_y_pitch >>= 1;
-	 }
-
 	 width  = minify(width);
 	 height = minify(height);
 	 depth  = minify(depth);
+
+    if (mt->compressed) {
+        pack_y_pitch = (height + 3) / 4;
+        
+        if (pack_x_pitch > ALIGN(width, align_w)) {
+            pack_x_pitch = ALIGN(width, align_w);
+            pack_x_nr <<= 1;
+        }
+    } else {
+        if (pack_x_pitch > 4) {
+            pack_x_pitch >>= 1;
+            pack_x_nr <<= 1;
+            assert(pack_x_pitch * pack_x_nr <= mt->pitch);
+        }
+
+        if (pack_y_pitch > 2) {
+            pack_y_pitch >>= 1;
+            pack_y_pitch = ALIGN(pack_y_pitch, align_h);
+        }
+    }
+
       }
       break;
    }
diff --git a/src/mesa/drivers/dri/i965/brw_vs.h b/src/mesa/drivers/dri/i965/brw_vs.h
index fdb5785d67d..8843f816f9c 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.h
+++ b/src/mesa/drivers/dri/i965/brw_vs.h
@@ -36,7 +36,7 @@
 
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "program.h"
+#include "shader/program.h"
 
 
 struct brw_vs_prog_key {
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 6eb11b19ad2..8733b470c25 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -30,8 +30,8 @@
   */
             
 
-#include "program.h"
-#include "macros.h"
+#include "main/macros.h"
+#include "shader/program.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_print.h"
 #include "brw_context.h"
@@ -201,7 +201,7 @@ static void unalias2( struct brw_vs_compile *c,
 				    struct brw_reg,
 				    struct brw_reg ))
 {
-   if ((dst.file == arg0.file && dst.nr == arg0.nr) &&
+   if ((dst.file == arg0.file && dst.nr == arg0.nr) ||
        (dst.file == arg1.file && dst.nr == arg1.nr)) {
       struct brw_compile *p = &c->func;
       struct brw_reg tmp = brw_writemask(get_tmp(c), dst.dw1.bits.writemask);
diff --git a/src/mesa/drivers/dri/i965/brw_vs_tnl.c b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
index b69be350a92..27210d1a374 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_tnl.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
@@ -1000,13 +1000,19 @@ static void build_lighting( struct tnl_program *p )
 					       STATE_POSITION); 
 	    struct ureg V = get_eye_position(p);
 	    struct ureg dist = get_temp(p);
+       struct ureg tmpPpli = get_temp(p);
 
 	    VPpli = get_temp(p); 
 	    half = get_temp(p);
+
+       /* In homogeneous object coordinates
+        */
+       emit_op1(p, OPCODE_RCP, dist, 0, swizzle1(Ppli, W));
+       emit_op2(p, OPCODE_MUL, tmpPpli, 0, Ppli, dist);
  
 	    /* Calulate VPpli vector
 	     */
-	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); 
+	    emit_op2(p, OPCODE_SUB, VPpli, 0, tmpPpli, V); 
 
 	    /* Normalize VPpli.  The dist value also used in
 	     * attenuation below.
@@ -1038,6 +1044,7 @@ static void build_lighting( struct tnl_program *p )
 	    emit_normalize_vec3(p, half, half);
 
 	    release_temp(p, dist);
+       release_temp(p, tmpPpli);
 	 }
 
 	 /* Calculate dot products:
@@ -1621,7 +1628,8 @@ const struct brw_tracked_state brw_tnl_vertprog = {
 	       _NEW_FOG | 
 	       _NEW_HINT | 
 	       _NEW_POINT | 
-	       _NEW_TEXTURE),
+	       _NEW_TEXTURE |
+          _NEW_TEXTURE_MATRIX),
       .brw = (BRW_NEW_FRAGMENT_PROGRAM | 
 	      BRW_NEW_INPUT_VARYING),
       .cache = 0
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index 786f30e641e..d8cb1688023 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -47,7 +47,6 @@
 
 #include "brw_draw.h"
 #include "brw_state.h"
-#include "brw_aub.h"
 #include "brw_fallback.h"
 #include "brw_vs.h"
 
@@ -60,8 +59,6 @@ static void brw_destroy_context( struct intel_context *intel )
    GLcontext *ctx = &intel->ctx;
    struct brw_context *brw = brw_context(&intel->ctx);
 
-   brw_aub_destroy(brw);
-
    brw_destroy_metaops(brw);
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index 1497dc79687..904c00bef87 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -285,7 +285,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
 
 
    /* BRW_NEW_WM_INPUT_DIMENSIONS */
-   key->projtex_mask = brw->wm.input_size_masks[4-1]; 
+   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);
diff --git a/src/mesa/drivers/dri/i965/brw_wm.h b/src/mesa/drivers/dri/i965/brw_wm.h
index f5fddfdb68a..6dcf4732a9f 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.h
+++ b/src/mesa/drivers/dri/i965/brw_wm.h
@@ -34,9 +34,9 @@
 #define BRW_WM_H
 
 
+#include "shader/prog_instruction.h"
 #include "brw_context.h"
 #include "brw_eu.h"
-#include "prog_instruction.h"
 
 /* A big lookup table is used to figure out which and how many
  * additional regs will inserted before the main payload in the WM
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index ff97d87dc45..67154c1b13a 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -525,7 +525,57 @@ static void precalc_tex( struct brw_wm_compile *c,
    struct prog_src_register coord;
    struct prog_dst_register tmpcoord;
 
-   if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
+   if (inst->TexSrcTarget == TEXTURE_CUBE_INDEX) {
+       struct prog_instruction *out;
+       struct prog_dst_register tmp0 = get_temp(c);
+       struct prog_src_register tmp0src = src_reg_from_dst(tmp0);
+       struct prog_dst_register tmp1 = get_temp(c);
+       struct prog_src_register tmp1src = src_reg_from_dst(tmp1);
+       struct prog_src_register src0 = inst->SrcReg[0];
+
+       tmpcoord = get_temp(c);
+       coord = src_reg_from_dst(tmpcoord);
+
+       out = emit_op(c, OPCODE_MOV,
+                     tmpcoord,
+                     0, 0, 0,
+                     src0,
+                     src_undef(),
+                     src_undef());
+       out->SrcReg[0].NegateBase = 0;
+       out->SrcReg[0].Abs = 1;
+
+       emit_op(c, OPCODE_MAX,
+               tmp0,
+               0, 0, 0,
+               src_swizzle1(coord, X),
+               src_swizzle1(coord, Y),
+               src_undef());
+
+       emit_op(c, OPCODE_MAX,
+               tmp1,
+               0, 0, 0,
+               tmp0src,
+               src_swizzle1(coord, Z),
+               src_undef());
+
+       emit_op(c, OPCODE_RCP,
+               tmp0,
+               0, 0, 0,
+               tmp1src,
+               src_undef(),
+               src_undef());
+
+       emit_op(c, OPCODE_MUL,
+               tmpcoord,
+               0, 0, 0,
+               src0,
+               tmp0src,
+               src_undef());
+
+       release_temp(c, tmp0);
+       release_temp(c, tmp1);
+   } else if (inst->TexSrcTarget == TEXTURE_RECT_INDEX) {
       struct prog_src_register scale = 
 	 search_or_add_param5( c, 
 			       STATE_INTERNAL, 
@@ -660,7 +710,7 @@ static GLboolean projtex( struct brw_wm_compile *c,
       return 0;  /* ut2004 gun rendering !?! */
    else if (src.File == PROGRAM_INPUT && 
 	    GET_SWZ(src.Swizzle, W) == W &&
-	    (c->key.projtex_mask & (1<<src.Index)) == 0)
+           (c->key.projtex_mask & (1<<(src.Index + FRAG_ATTRIB_WPOS - FRAG_ATTRIB_TEX0))) == 0)
       return 0;
    else
       return 1;
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 93d4cfc3a5f..3c0952acf0d 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_sampler_state.c
@@ -54,7 +54,7 @@ static GLuint translate_wrap_mode( GLenum wrap )
    case GL_REPEAT: 
       return BRW_TEXCOORDMODE_WRAP;
    case GL_CLAMP:  
-      return BRW_TEXCOORDMODE_CLAMP_BORDER; /* conform likes it this way */
+      return BRW_TEXCOORDMODE_CLAMP;
    case GL_CLAMP_TO_EDGE: 
       return BRW_TEXCOORDMODE_CLAMP; /* conform likes it this way */
    case GL_CLAMP_TO_BORDER: 
@@ -173,12 +173,12 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
        * message (sample_c).  So need to recompile WM program when
        * shadow comparison is enabled on each/any texture unit.
        */
-      sampler->ss0.shadow_function = intel_translate_compare_func(texObj->CompareFunc);
+      sampler->ss0.shadow_function = intel_translate_shadow_compare_func(texObj->CompareFunc);
    }
 
    /* Set LOD bias: 
     */
-   sampler->ss0.lod_bias = S_FIXED(texUnit->LodBias + texObj->LodBias, 6);
+   sampler->ss0.lod_bias = S_FIXED(CLAMP(texUnit->LodBias + texObj->LodBias, -16, 15), 6);
 
    sampler->ss0.lod_preclamp = 1; /* OpenGL mode */
    sampler->ss0.default_color_mode = 0; /* OpenGL/DX10 mode */
@@ -192,8 +192,8 @@ static void brw_update_sampler_state( struct gl_texture_unit *texUnit,
     */
    sampler->ss0.base_level = U_FIXED(0, 1);
 
-   sampler->ss1.max_lod = U_FIXED(MAX2(texObj->MaxLod, 0), 6);
-   sampler->ss1.min_lod = U_FIXED(MAX2(texObj->MinLod, 0), 6);
+   sampler->ss1.max_lod = U_FIXED(MIN2(MAX2(texObj->MaxLod, 0), 13), 6);
+   sampler->ss1.min_lod = U_FIXED(MIN2(MAX2(texObj->MinLod, 0), 13), 6);
    
    sampler->ss2.default_color_pointer = sdc_gs_offset >> 5;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index ff5cb31bdd1..5b4f2abd0e2 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -117,7 +117,7 @@ static void upload_wm_unit(struct brw_context *brw )
    wm.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
 
    /* CACHE_NEW_SAMPLER */
-   wm.wm4.sampler_count = brw->wm.sampler_count;
+   wm.wm4.sampler_count = (brw->wm.sampler_count + 1) / 4;
    wm.wm4.sampler_state_pointer = brw->wm.sampler_gs_offset >> 5;
 
    /* BRW_NEW_FRAGMENT_PROGRAM */
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 d24c618a668..0a45164a0fd 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -114,11 +114,24 @@ static GLuint translate_tex_format( GLuint mesa_format )
       return BRW_SURFACEFORMAT_FXT1;
 
    case MESA_FORMAT_Z16:
-      return BRW_SURFACEFORMAT_L16_UNORM;
+      return BRW_SURFACEFORMAT_I16_UNORM;
 
-   case MESA_FORMAT_RGBA_DXT1:
    case MESA_FORMAT_RGB_DXT1:
-      return BRW_SURFACEFORMAT_DXT1_RGB;
+       return BRW_SURFACEFORMAT_DXT1_RGB;
+
+   case MESA_FORMAT_RGBA_DXT1:
+       return BRW_SURFACEFORMAT_BC1_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT3:
+       return BRW_SURFACEFORMAT_BC2_UNORM;
+       
+   case MESA_FORMAT_RGBA_DXT5:
+       return BRW_SURFACEFORMAT_BC3_UNORM;
+
+   case MESA_FORMAT_SRGBA8:
+      return BRW_SURFACEFORMAT_R8G8B8A8_UNORM_SRGB;
+   case MESA_FORMAT_SRGB_DXT1:
+      return BRW_SURFACEFORMAT_BC1_UNORM_SRGB;
 
    default:
       assert(0);
diff --git a/src/mesa/drivers/dri/i965/bufmgr.h b/src/mesa/drivers/dri/i965/bufmgr.h
index e748c0d6d0b..b31c2e6d9bc 100644
--- a/src/mesa/drivers/dri/i965/bufmgr.h
+++ b/src/mesa/drivers/dri/i965/bufmgr.h
@@ -130,24 +130,6 @@ int bmBufferSubData(struct intel_context *,
 		     unsigned size, 
 		     const void *data );
 
-
-int bmBufferDataAUB(struct intel_context *, 
-		     struct buffer *buf, 
-		     unsigned size, 
-		     const void *data, 
-		     unsigned flags,
-		     unsigned aubtype,
-		     unsigned aubsubtype );
-
-int bmBufferSubDataAUB(struct intel_context *, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			const void *data,
-			unsigned aubtype,
-			unsigned aubsubtype );
-
-
 /* In this version, taking the offset will provoke an upload on
  * buffers not already resident in AGP:
  */
@@ -170,12 +152,6 @@ void *bmMapBuffer( struct intel_context *,
 void bmUnmapBuffer( struct intel_context *,
 		    struct buffer *buf );
 
-void bmUnmapBufferAUB( struct intel_context *,
-		       struct buffer *buf,
-		       unsigned aubtype,
-		       unsigned aubsubtype );
-
-
 /* Pertains to all buffers who's offset has been taken since the last
  * fence or release.
  */
diff --git a/src/mesa/drivers/dri/i965/bufmgr_fake.c b/src/mesa/drivers/dri/i965/bufmgr_fake.c
index 24ee11edd8c..a85121122fc 100644
--- a/src/mesa/drivers/dri/i965/bufmgr_fake.c
+++ b/src/mesa/drivers/dri/i965/bufmgr_fake.c
@@ -85,7 +85,6 @@ struct buffer {
 
    unsigned mapped:1;		
    unsigned dirty:1;		
-   unsigned aub_dirty:1;	
    unsigned alignment:13;
    unsigned flags:16;
 
@@ -906,86 +905,6 @@ int bmBufferSubData(struct intel_context *intel,
    return retval;
 }
 
-
-
-int bmBufferDataAUB(struct intel_context *intel, 
-		     struct buffer *buf, 
-		     unsigned size, 
-		     const void *data, 
-		     unsigned flags,
-		     unsigned aubtype,
-		     unsigned aubsubtype )
-{
-   int retval = bmBufferData(intel, buf, size, data, flags);
-   
-
-   /* This only works because in this version of the buffer manager we
-    * allocate all buffers statically in agp space and so can emit the
-    * uploads to the aub file with the correct offsets as they happen.
-    */
-   if (retval == 0 && data && intel->aub_file) {
-
-      if (buf->block && !buf->dirty) {
-	 intel->vtbl.aub_gtt_data(intel,
-				      buf->block->mem->ofs,
-				      buf->block->virtual,
-				      size,
-				      aubtype,
-				      aubsubtype);
-	 buf->aub_dirty = 0;
-      }
-   }
-   
-   return retval;
-}
-		       
-
-int bmBufferSubDataAUB(struct intel_context *intel, 
-			struct buffer *buf, 
-			unsigned offset, 
-			unsigned size, 
-			const void *data,
-			unsigned aubtype,
-			unsigned aubsubtype )
-{
-   int retval = bmBufferSubData(intel, buf, offset, size, data);
-   
-
-   /* This only works because in this version of the buffer manager we
-    * allocate all buffers statically in agp space and so can emit the
-    * uploads to the aub file with the correct offsets as they happen.
-    */
-   if (intel->aub_file) {
-      if (retval == 0 && buf->block && !buf->dirty)
-	 intel->vtbl.aub_gtt_data(intel,
-				      buf->block->mem->ofs + offset,
-				      ((const char *)buf->block->virtual) + offset,
-				      size,
-				      aubtype,
-				      aubsubtype);
-   }
-
-   return retval;
-}
-
-void bmUnmapBufferAUB( struct intel_context *intel, 
-		       struct buffer *buf,
-		       unsigned aubtype,
-		       unsigned aubsubtype )
-{
-   bmUnmapBuffer(intel, buf);
-
-   if (intel->aub_file) {
-      /* Hack - exclude the framebuffer mappings.  If you removed
-       * this, you'd get very big aubfiles, but you *would* be able to
-       * see fallback rendering.
-       */
-      if (buf->block  && !buf->dirty && buf->block->pool == &intel->bm->pool[0]) {
-	 buf->aub_dirty = 1;
-      }
-   }
-}
-
 unsigned bmBufferOffset(struct intel_context *intel, 
 			struct buffer *buf)
 {
@@ -1197,26 +1116,7 @@ int bmValidateBuffers( struct intel_context *intel )
 			 buf->backing_store, 
 			 buf->size);
 
-	       if (intel->aub_file) {
-		  intel->vtbl.aub_gtt_data(intel,
-					       buf->block->mem->ofs,
-					       buf->backing_store,
-					       buf->size,
-					       0,
-					       0);
-	       }
-
 	       buf->dirty = 0;
-	       buf->aub_dirty = 0;
-	    }
-	    else if (buf->aub_dirty) {
-	       intel->vtbl.aub_gtt_data(intel,
-					    buf->block->mem->ofs,
-					    buf->block->virtual,
-					    buf->size,
-					    0,
-					    0);
-	       buf->aub_dirty = 0;
 	    }
 
 	    block->referenced = 0;
diff --git a/src/mesa/drivers/dri/i965/intel_batchbuffer.c b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
index 64885ed9b4b..fb58c0e708c 100644
--- a/src/mesa/drivers/dri/i965/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i965/intel_batchbuffer.c
@@ -28,6 +28,7 @@
 #include "imports.h"
 #include "intel_batchbuffer.h"
 #include "intel_ioctl.h"
+#include "intel_decode.h"
 #include "bufmgr.h"
 
 
@@ -168,31 +169,22 @@ GLboolean intel_batchbuffer_flush( struct intel_batchbuffer *batch )
       goto out;
    }
 
+   if (INTEL_DEBUG & DEBUG_BATCH) {
+      char *map;
 
-   if (intel->aub_file) {
-      /* Send buffered commands to aubfile as a single packet. 
-       */
-      intel_batchbuffer_map(batch);
-      ((int *)batch->ptr)[-1] = intel->vtbl.flush_cmd();
-      intel->vtbl.aub_commands(intel,
-			       offset, /* Fulsim wierdness - don't adjust */
-			       batch->map + batch->offset,
-			       used);
-      ((int *)batch->ptr)[-1] = MI_BATCH_BUFFER_END;
-      intel_batchbuffer_unmap(batch);
+      map = bmMapBuffer(batch->intel, batch->buffer,
+			BM_MEM_AGP|BM_MEM_LOCAL|BM_CLIENT);
+      intel_decode((uint32_t *)(map + batch->offset), used / 4,
+		   offset + batch->offset, intel->intelScreen->deviceID);
+      bmUnmapBuffer(batch->intel, batch->buffer);
    }
 
-
    /* Fire the batch buffer, which was uploaded above:
     */
    intel_batch_ioctl(batch->intel, 
 		     offset + batch->offset,
 		     used);
 
-   if (intel->aub_file && 
-       intel->ctx.DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT)
-      intel->vtbl.aub_dump_bmp( intel, 0 );
-
    /* Reset the buffer:
     */
  out:
diff --git a/src/mesa/drivers/dri/i965/intel_buffer_objects.c b/src/mesa/drivers/dri/i965/intel_buffer_objects.c
index 015e433fd7a..3349284f5db 100644
--- a/src/mesa/drivers/dri/i965/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/i965/intel_buffer_objects.c
@@ -103,8 +103,7 @@ static void intel_bufferobj_data( GLcontext *ctx,
    obj->Size = size;
    obj->Usage = usage;
 
-   bmBufferDataAUB(intel, intel_obj->buffer, size, data, 0,
-		   0, 0);
+   bmBufferData(intel, intel_obj->buffer, size, data, 0);
 }
 
 
@@ -125,7 +124,7 @@ static void intel_bufferobj_subdata( GLcontext *ctx,
    struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
 
    assert(intel_obj);
-   bmBufferSubDataAUB(intel, intel_obj->buffer, offset, size, data, 0, 0);
+   bmBufferSubData(intel, intel_obj->buffer, offset, size, data);
 }
 
 
@@ -181,7 +180,7 @@ static GLboolean intel_bufferobj_unmap( GLcontext *ctx,
    assert(intel_obj);
    assert(intel_obj->buffer);
    assert(obj->Pointer);
-   bmUnmapBufferAUB(intel, intel_obj->buffer, 0, 0);
+   bmUnmapBuffer(intel, intel_obj->buffer);
    obj->Pointer = NULL;
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/i965/intel_buffers.c b/src/mesa/drivers/dri/i965/intel_buffers.c
index d155c039d77..6c8b0735026 100644
--- a/src/mesa/drivers/dri/i965/intel_buffers.c
+++ b/src/mesa/drivers/dri/i965/intel_buffers.c
@@ -450,12 +450,6 @@ void intelSwapBuffers( __DRIdrawablePrivate *dPriv )
 	 } else {
 	    intelCopyBuffer( dPriv, NULL );
 	 }
-	 if (intel->aub_file) {
-	    intelFlush(ctx);
-	    intel->vtbl.aub_dump_bmp( intel, 1 );
-
-	    intel->aub_wrap = 1;
-	 }
       }
    } else {
       /* XXX this shouldn't be an error but we can't handle it for now */
diff --git a/src/mesa/drivers/dri/i965/intel_context.c b/src/mesa/drivers/dri/i965/intel_context.c
index 4f51fefe0f6..2cf311c7135 100644
--- a/src/mesa/drivers/dri/i965/intel_context.c
+++ b/src/mesa/drivers/dri/i965/intel_context.c
@@ -46,6 +46,7 @@
 #include "drivers/common/driverfuncs.h"
 
 #include "intel_screen.h"
+#include "intel_chipset.h"
 
 #include "i830_dri.h"
 #include "i830_common.h"
@@ -66,6 +67,7 @@
 int INTEL_DEBUG = (0);
 #endif
 
+#define need_GL_NV_point_sprite
 #define need_GL_ARB_multisample
 #define need_GL_ARB_point_parameters
 #define need_GL_ARB_texture_compression
@@ -81,6 +83,7 @@ int INTEL_DEBUG = (0);
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_multi_draw_arrays
 #define need_GL_EXT_secondary_color
+#define need_GL_EXT_point_parameters
 #include "extension_helper.h"
 
 #ifndef VERBOSE
@@ -146,6 +149,7 @@ 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_NV_point_sprite",                GL_NV_point_sprite_functions },
     { "GL_ARB_texture_border_clamp",       NULL },
     { "GL_ARB_texture_compression",        GL_ARB_texture_compression_functions },
     { "GL_ARB_texture_cube_map",           NULL },
@@ -158,6 +162,8 @@ const struct dri_extension card_extensions[] =
     { "GL_NV_texture_rectangle",           NULL },
     { "GL_EXT_texture_rectangle",          NULL },
     { "GL_ARB_texture_rectangle",          NULL },
+    { "GL_ARB_point_sprite",               NULL},
+    { "GL_ARB_point_parameters",	   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 },
@@ -177,6 +183,7 @@ const struct dri_extension card_extensions[] =
     { "GL_EXT_texture_env_dot3",           NULL },
     { "GL_EXT_texture_filter_anisotropic", NULL },
     { "GL_EXT_texture_lod_bias",           NULL },
+    { "GL_EXT_texture_sRGB",               NULL },
     { "GL_3DFX_texture_compression_FXT1",  NULL },
     { "GL_APPLE_client_storage",           NULL },
     { "GL_MESA_pack_invert",               NULL },
@@ -219,6 +226,7 @@ static const struct dri_debug_control debug_control[] =
     { "thre",  DEBUG_SINGLE_THREAD },
     { "wm",    DEBUG_WM },
     { "vs",    DEBUG_VS },
+    { "bat",   DEBUG_BATCH },
     { NULL,    0 }
 };
 
@@ -265,7 +273,7 @@ intelBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
 	};
 	intel->stats_wm++;
 	intelFinish(&intel->ctx);
-	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
+	drmCommandWrite(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
 }
 
 static void
@@ -279,7 +287,7 @@ intelEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
 		.data = &tmp
 	};
 	intelFinish(&intel->ctx);
-	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
+	drmCommandWrite(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
 	q->Result = tmp - q->Result;
 	q->Ready = GL_TRUE;
 	intel->stats_wm--;
@@ -570,6 +578,10 @@ GLboolean intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
    if (driContextPriv) {
       struct intel_context *intel = (struct intel_context *) driContextPriv->driverPrivate;
 
+      if (intel->driReadDrawable != driReadPriv) {
+          intel->driReadDrawable = driReadPriv;
+      }
+
       if ( intel->driDrawable != driDrawPriv ) {
 	 /* Shouldn't the readbuffer be stored also? */
 	 driDrawableInitVBlank( driDrawPriv, intel->vblank_flags,
@@ -658,19 +670,12 @@ void LOCK_HARDWARE( struct intel_context *intel )
 
    intel->locked = 1;
 
-   if (intel->aub_wrap) {
-      bm_fake_NotifyContendedLockTake( intel ); 
-      intel->vtbl.lost_hardware( intel );
-      intel->vtbl.aub_wrap(intel);
-      intel->aub_wrap = 0;
-   }
-
    if (bmError(intel)) {
       bmEvictAll(intel);
       intel->vtbl.lost_hardware( intel );
    }
 
-   /* Make sure nothing has been emitted prior to getting the lock: 
+   /* Make sure nothing has been emitted prior to getting the lock:
     */
    assert(intel->batch->map == 0);
 
@@ -703,8 +708,6 @@ void UNLOCK_HARDWARE( struct intel_context *intel )
    intel->vtbl.note_unlock( intel );
    intel->locked = 0;
 
-
-
    DRM_UNLOCK(intel->driFd, intel->driHwLock, intel->hHWContext);
    _glthread_UNLOCK_MUTEX(lockMutex); 
 }
diff --git a/src/mesa/drivers/dri/i965/intel_context.h b/src/mesa/drivers/dri/i965/intel_context.h
index 406f8483dce..f63c2f613dc 100644
--- a/src/mesa/drivers/dri/i965/intel_context.h
+++ b/src/mesa/drivers/dri/i965/intel_context.h
@@ -109,20 +109,6 @@ struct intel_context
       void (*emit_flush)( struct intel_context *intel,
 			  GLuint unused );
 
-      void (*aub_commands)( struct intel_context *intel, 
-			    GLuint offset,
-			    const void *buf,
-			    GLuint sz );
-      void (*aub_dump_bmp)( struct intel_context *intel, GLuint buffer );
-      void (*aub_wrap)( struct intel_context *intel );
-      void (*aub_gtt_data)( struct intel_context *intel, 
-			    GLuint offset,
-			    const void *src,
-			    GLuint size,
-			    GLuint aubtype, 
-			    GLuint aubsubtype);
-
-
       void (*reduced_primitive_state)( struct intel_context *intel, GLenum rprim );
 
       GLboolean (*check_vertex_size)( struct intel_context *intel, GLuint expected );
@@ -176,7 +162,6 @@ struct intel_context
    GLuint last_swap_fence;
    GLuint second_last_swap_fence;
    
-   GLboolean aub_wrap;
    GLuint stats_wm;
 
    struct intel_batchbuffer *batch;
@@ -234,11 +219,10 @@ struct intel_context
    int driFd;
 
    __DRIdrawablePrivate *driDrawable;
+   __DRIdrawablePrivate *driReadDrawable;
    __DRIscreenPrivate *driScreen;
    intelScreenPrivate *intelScreen; 
    volatile drmI830Sarea *sarea; 
-   
-   FILE *aub_file;
 
    GLuint lastStamp;
 
@@ -371,22 +355,7 @@ extern int INTEL_DEBUG;
 #define DEBUG_WM        0x10000
 #define DEBUG_URB       0x20000
 #define DEBUG_VS        0x40000
-
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-#define PCI_CHIP_I855_GM		0x3582
-#define PCI_CHIP_I865_G			0x2572
-#define PCI_CHIP_I915_G			0x2582
-#define PCI_CHIP_I915_GM		0x2592
-#define PCI_CHIP_I945_G			0x2772
-#define PCI_CHIP_I965_G			0x29A2
-#define PCI_CHIP_I965_Q			0x2992
-#define PCI_CHIP_I965_G_1		0x2982
-#define PCI_CHIP_I946_GZ		0x2972
-#define PCI_CHIP_I965_GM                0x2A02
-#define PCI_CHIP_I965_GME               0x2A12
-
+#define DEBUG_BATCH	0x80000
 
 /* ================================================================
  * intel_context.c:
@@ -463,7 +432,7 @@ extern void intelInitStateFuncs( struct dd_function_table *functions );
 #define BLENDFACT_INV_CONST_ALPHA	0x0f
 #define BLENDFACT_MASK          	0x0f
 
-
+extern int intel_translate_shadow_compare_func( GLenum func );
 extern int intel_translate_compare_func( GLenum func );
 extern int intel_translate_stencil_op( GLenum op );
 extern int intel_translate_blend_factor( GLenum factor );
diff --git a/src/mesa/drivers/dri/i965/intel_decode.c b/src/mesa/drivers/dri/i965/intel_decode.c
new file mode 120000
index 00000000000..f671b6cbb13
--- /dev/null
+++ b/src/mesa/drivers/dri/i965/intel_decode.c
@@ -0,0 +1 @@
+../intel/intel_decode.c
\ No newline at end of file
diff --git a/src/mesa/drivers/dri/i965/intel_ioctl.c b/src/mesa/drivers/dri/i965/intel_ioctl.c
index 0a8e976f706..e7e736079f4 100644
--- a/src/mesa/drivers/dri/i965/intel_ioctl.c
+++ b/src/mesa/drivers/dri/i965/intel_ioctl.c
@@ -115,10 +115,6 @@ void intelWaitIrq( struct intel_context *intel, int seq )
       if ( ret ) {
 	 fprintf( stderr, "%s: drmI830IrqWait: %d\n", __FUNCTION__, ret );
 
-	 if (intel->aub_file) {
-	    intel->vtbl.aub_dump_bmp( intel, intel->ctx.Visual.doubleBufferMode ? 1 : 0 );
-	 }
-
 	 exit(1);
       }
    }
diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
index 8486086b274..0fb33e27f47 100644
--- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
@@ -75,7 +75,7 @@ struct intel_mipmap_tree *intel_miptree_create( struct intel_context *intel,
    mt->width0 = width0;
    mt->height0 = height0;
    mt->depth0 = depth0;
-   mt->cpp = compressed ? 2 : cpp;
+   mt->cpp = cpp;
    mt->compressed = compressed;
 
    switch (intel->intelScreen->deviceID) {
@@ -211,7 +211,7 @@ GLuint intel_miptree_image_offset(struct intel_mipmap_tree *mt,
 
 
 
-
+extern GLuint intel_compressed_alignment(GLenum);
 /* Upload data for a particular image.
  */
 GLboolean intel_miptree_image_data(struct intel_context *intel, 
@@ -226,6 +226,17 @@ GLboolean intel_miptree_image_data(struct intel_context *intel,
    GLuint dst_offset = intel_miptree_image_offset(dst, face, level);
    const GLuint *dst_depth_offset = intel_miptree_depth_offsets(dst, level);
    GLuint i;
+   GLuint width, height, alignment;
+
+   width = dst->level[level].width;
+   height = dst->level[level].height;
+
+   if (dst->compressed) {
+       alignment = intel_compressed_alignment(dst->internal_format);
+       src_row_pitch = ((src_row_pitch + alignment - 1) & ~(alignment - 1));
+       width = ((width + alignment - 1) & ~(alignment - 1));
+       height = (height + 3) / 4;
+   }
 
    DBG("%s\n", __FUNCTION__);
    for (i = 0; i < depth; i++) {
@@ -237,8 +248,8 @@ GLboolean intel_miptree_image_data(struct intel_context *intel,
 			     src,
 			     src_row_pitch,
 			     0, 0,	/* source x,y */
-			     dst->level[level].width,
-			     dst->level[level].height))
+			     width,
+			     height))
 	 return GL_FALSE;
       src += src_image_pitch;
    }
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
index 421fcc5e511..79c1fee9c07 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_bitmap.c
@@ -168,12 +168,15 @@ do_blit_bitmap( GLcontext *ctx,
 {
    struct intel_context *intel = intel_context(ctx);
    struct intel_region *dst = intel_drawbuf_region(intel);
-   
+   GLfloat tmpColor[4];
+
    union {
       GLuint ui;
       GLubyte ub[4];
    } color;
 
+   if (!dst)
+       return GL_FALSE;
 
    if (unpack->BufferObj->Name) {
       bitmap = map_pbo(ctx, width, height, unpack, bitmap);
@@ -181,10 +184,16 @@ do_blit_bitmap( GLcontext *ctx,
 	 return GL_TRUE;	/* even though this is an error, we're done */
    }
 
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[0], ctx->Current.RasterColor[2]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[1], ctx->Current.RasterColor[1]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[2], ctx->Current.RasterColor[0]);
-   UNCLAMPED_FLOAT_TO_CHAN(color.ub[3], ctx->Current.RasterColor[3]);
+   COPY_4V(tmpColor, ctx->Current.RasterColor);
+
+   if (NEED_SECONDARY_COLOR(ctx)) {
+       ADD_3V(tmpColor, tmpColor, ctx->Current.RasterSecondaryColor);
+   }
+
+   UNCLAMPED_FLOAT_TO_CHAN(color.ub[0], tmpColor[2]);
+   UNCLAMPED_FLOAT_TO_CHAN(color.ub[1], tmpColor[1]);
+   UNCLAMPED_FLOAT_TO_CHAN(color.ub[2], tmpColor[0]);
+   UNCLAMPED_FLOAT_TO_CHAN(color.ub[3], tmpColor[3]);
 
    /* Does zoom apply to bitmaps?
     */
@@ -226,10 +235,10 @@ do_blit_bitmap( GLcontext *ctx,
       dsty = dPriv->y + (dPriv->h - dsty - height);  
       dstx = dPriv->x + dstx;
 
-      dest_rect.x1 = dstx;
-      dest_rect.y1 = dsty;
-      dest_rect.x2 = dstx + width;
-      dest_rect.y2 = dsty + height;
+      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;
 
       for (i = 0; i < nbox; i++) {
          drm_clip_rect_t rect;
diff --git a/src/mesa/drivers/dri/i965/intel_pixel_copy.c b/src/mesa/drivers/dri/i965/intel_pixel_copy.c
index 58dc49505fe..3bdf2fb4790 100644
--- a/src/mesa/drivers/dri/i965/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/i965/intel_pixel_copy.c
@@ -231,6 +231,7 @@ do_blit_copypixels(GLcontext * ctx,
 
    if (intel->driDrawable->numClipRects) {
       __DRIdrawablePrivate *dPriv = intel->driDrawable;
+      __DRIdrawablePrivate *dReadPriv = intel->driReadDrawable;
       drm_clip_rect_t *box = dPriv->pClipRects;
       drm_clip_rect_t dest_rect;
       GLint nbox = dPriv->numClipRects;
@@ -262,8 +263,8 @@ do_blit_copypixels(GLcontext * ctx,
       srcy = dPriv->h - srcy - height;  
       dstx += dPriv->x;
       dsty += dPriv->y;
-      srcx += dPriv->x;
-      srcy += 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.
diff --git a/src/mesa/drivers/dri/i965/intel_regions.c b/src/mesa/drivers/dri/i965/intel_regions.c
index 835ecdd7257..b78eba898ff 100644
--- a/src/mesa/drivers/dri/i965/intel_regions.c
+++ b/src/mesa/drivers/dri/i965/intel_regions.c
@@ -64,7 +64,7 @@ void intel_region_unmap(struct intel_context *intel,
 {
    DBG("%s\n", __FUNCTION__);
    if (!--region->map_refcount) {
-      bmUnmapBufferAUB(intel, region->buffer, 0, 0);
+      bmUnmapBuffer(intel, region->buffer);
       region->map = NULL;
    }
 }
@@ -217,10 +217,10 @@ GLboolean intel_region_data(struct intel_context *intel,
        srcx == 0 &&
        srcy == 0) 
    {
-      return (bmBufferDataAUB(intel,
-			      dst->buffer,
-			      dst->cpp * width * dst->height,
-			      src, 0, 0, 0) == 0);
+      return (bmBufferData(intel,
+			   dst->buffer,
+			   dst->cpp * width * dst->height,
+			   src, 0) == 0);
    }
    else {
       GLubyte *map = intel_region_map(intel, dst);
diff --git a/src/mesa/drivers/dri/i965/intel_state.c b/src/mesa/drivers/dri/i965/intel_state.c
index 2e442db6198..2f5467a4e46 100644
--- a/src/mesa/drivers/dri/i965/intel_state.c
+++ b/src/mesa/drivers/dri/i965/intel_state.c
@@ -38,6 +38,31 @@
 #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) {
diff --git a/src/mesa/drivers/dri/i965/intel_tex_validate.c b/src/mesa/drivers/dri/i965/intel_tex_validate.c
index 44ee94614d5..8c05e7cdabc 100644
--- a/src/mesa/drivers/dri/i965/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i965/intel_tex_validate.c
@@ -122,6 +122,29 @@ static void intel_texture_invalidate_cb( struct intel_context *intel,
    intel_texture_invalidate( (struct intel_texture_object *) ptr );
 }
 
+#include "texformat.h"
+static GLuint intel_compressed_num_bytes(GLenum mesaFormat)
+{
+    GLuint bytes = 0;
+
+    switch (mesaFormat) {
+    case MESA_FORMAT_RGB_FXT1:
+    case MESA_FORMAT_RGBA_FXT1:
+    case MESA_FORMAT_RGB_DXT1:
+    case MESA_FORMAT_RGBA_DXT1:
+        bytes = 2;
+        break;
+
+    case MESA_FORMAT_RGBA_DXT3:
+    case MESA_FORMAT_RGBA_DXT5:
+        bytes = 4;
+    
+    default:
+        break;
+    }
+
+    return bytes;
+}
 
 /*  
  */
@@ -132,7 +155,8 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
    GLuint face, i;
    GLuint nr_faces = 0;
    struct gl_texture_image *firstImage;
-
+   GLuint cpp = 0;
+   
    if( tObj == intel->frame_buffer_texobj )
       return GL_FALSE;
    
@@ -165,6 +189,12 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 
 
 
+   if (firstImage->IsCompressed) {
+       cpp = intel_compressed_num_bytes(firstImage->TexFormat->MesaFormat);
+   } else {
+       cpp = firstImage->TexFormat->TexelBytes;
+   }
+       
    /* Check tree can hold all active levels.  Check tree matches
     * target, imageFormat, etc.
     */
@@ -176,7 +206,7 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 	intelObj->mt->width0 != firstImage->Width ||
 	intelObj->mt->height0 != firstImage->Height ||
 	intelObj->mt->depth0 != firstImage->Depth ||
-	intelObj->mt->cpp != firstImage->TexFormat->TexelBytes ||
+	intelObj->mt->cpp != cpp ||
 	intelObj->mt->compressed != firstImage->IsCompressed)) 
    {
       intel_miptree_destroy(intel, intelObj->mt);
@@ -199,7 +229,7 @@ GLuint intel_finalize_mipmap_tree( struct intel_context *intel,
 					  firstImage->Width,
 					  firstImage->Height,
 					  firstImage->Depth,
-					  firstImage->TexFormat->TexelBytes,
+					  cpp,
 					  firstImage->IsCompressed);
 
       /* Tell the buffer manager that we will manage the backing
diff --git a/src/mesa/drivers/dri/i965/server/i830_common.h b/src/mesa/drivers/dri/i965/server/i830_common.h
deleted file mode 100644
index f320378c2a5..00000000000
--- a/src/mesa/drivers/dri/i965/server/i830_common.h
+++ /dev/null
@@ -1,222 +0,0 @@
-/**************************************************************************
-
-Copyright 2001 VA Linux Systems Inc., Fremont, California.
-Copyright 2002 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
-on the rights to use, copy, modify, merge, publish, distribute, sub
-license, and/or sell copies of the Software, and to permit persons to whom
-the Software is furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice (including the next
-paragraph) shall be included in all copies or substantial portions of the
-Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
-ATI, VA LINUX SYSTEMS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM,
-DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
-OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
-USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_common.h,v 1.1 2002/09/11 00:29:32 dawes Exp $ */
-
-#ifndef _I830_COMMON_H_
-#define _I830_COMMON_H_
-
-
-#define I830_NR_TEX_REGIONS 255	/* maximum due to use of chars for next/prev */
-#define I830_LOG_MIN_TEX_REGION_SIZE 14
-
-
-/* Driver specific DRM command indices
- * NOTE: these are not OS specific, but they are driver specific
- */
-#define DRM_I830_INIT                     0x00
-#define DRM_I830_FLUSH                    0x01
-#define DRM_I830_FLIP                     0x02
-#define DRM_I830_BATCHBUFFER              0x03
-#define DRM_I830_IRQ_EMIT                 0x04
-#define DRM_I830_IRQ_WAIT                 0x05
-#define DRM_I830_GETPARAM                 0x06
-#define DRM_I830_SETPARAM                 0x07
-#define DRM_I830_ALLOC                    0x08
-#define DRM_I830_FREE                     0x09
-#define DRM_I830_INIT_HEAP                0x0a
-#define DRM_I830_CMDBUFFER                0x0b
-#define DRM_I830_DESTROY_HEAP             0x0c
-#define DRM_I830_MMIO		       	  0x10
-
-typedef struct {
-   enum {
-      I830_INIT_DMA = 0x01,
-      I830_CLEANUP_DMA = 0x02,
-      I830_RESUME_DMA = 0x03
-   } func;
-   unsigned int mmio_offset;
-   int sarea_priv_offset;
-   unsigned int ring_start;
-   unsigned int ring_end;
-   unsigned int ring_size;
-   unsigned int front_offset;
-   unsigned int back_offset;
-   unsigned int depth_offset;
-   unsigned int w;
-   unsigned int h;
-   unsigned int pitch;
-   unsigned int pitch_bits;
-   unsigned int back_pitch;
-   unsigned int depth_pitch;
-   unsigned int cpp;
-   unsigned int chipset;
-} drmI830Init;
-
-typedef struct {
-	drmTextureRegion texList[I830_NR_TEX_REGIONS+1];
-        int last_upload;	/* last time texture was uploaded */
-        int last_enqueue;	/* last time a buffer was enqueued */
-        volatile int last_dispatch;	/* age of the most recently dispatched buffer */
-	int ctxOwner;		/* last context to upload state */
-	int texAge;
-        int pf_enabled;		/* is pageflipping allowed? */
-        int pf_active;               
-        int pf_current_page;	/* which buffer is being displayed? */
-        int perf_boxes;	        /* performance boxes to be displayed */   
-	int width, height;      /* screen size in pixels */
-
-	drm_handle_t front_handle;
-	int front_offset;
-	int front_size;
-
-	drm_handle_t back_handle;
-	int back_offset;
-	int back_size;
-
-	drm_handle_t depth_handle;
-	int depth_offset;
-	int depth_size;
-
-	drm_handle_t tex_handle;
-	int tex_offset;
-	int tex_size;
-	int log_tex_granularity;
-	int pitch;
-	int rotation;           /* 0, 90, 180 or 270 */
-	int rotated_offset;
-	int rotated_size;
-	int rotated_pitch;
-	int virtualX, virtualY;
-
-	unsigned int front_tiled;
-        unsigned int back_tiled;
-        unsigned int depth_tiled;
-        unsigned int rotated_tiled;
-        unsigned int rotated2_tiled;
-} drmI830Sarea;
-
-/* Flags for perf_boxes
- */
-#define I830_BOX_RING_EMPTY    0x1 /* populated by kernel */
-#define I830_BOX_FLIP          0x2 /* populated by kernel */
-#define I830_BOX_WAIT          0x4 /* populated by kernel & client */
-#define I830_BOX_TEXTURE_LOAD  0x8 /* populated by kernel */
-#define I830_BOX_LOST_CONTEXT  0x10 /* populated by client */
-
-
-typedef struct {
-   	int start;		/* agp offset */
-	int used;		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830BatchBuffer;
-
-typedef struct {
-   	char *buf;		/* agp offset */
-	int sz; 		/* nr bytes in use */
-	int DR1;		/* hw flags for GFX_OP_DRAWRECT_INFO */
-        int DR4;		/* window origin for GFX_OP_DRAWRECT_INFO*/
-	int num_cliprects;	/* mulitpass with multiple cliprects? */
-        drm_clip_rect_t *cliprects; /* pointer to userspace cliprects */
-} drmI830CmdBuffer;
- 
-typedef struct {
-	int *irq_seq;
-} drmI830IrqEmit;
-
-typedef struct {
-	int irq_seq;
-} drmI830IrqWait;
-
-typedef struct {
-	int param;
-	int *value;
-} drmI830GetParam;
-
-#define I830_PARAM_IRQ_ACTIVE     1
-#define I830_PARAM_ALLOW_BATCHBUFFER   2 
-
-typedef struct {
-	int param;
-	int value;
-} drmI830SetParam;
-
-#define I830_SETPARAM_USE_MI_BATCHBUFFER_START  1
-#define I830_SETPARAM_TEX_LRU_LOG_GRANULARITY   2
-#define I830_SETPARAM_ALLOW_BATCHBUFFER         3
-
-
-/* A memory manager for regions of shared memory:
- */
-#define I830_MEM_REGION_AGP 1
-
-typedef struct {
-	int region;
-	int alignment;
-	int size;
-	int *region_offset;	/* offset from start of fb or agp */
-} drmI830MemAlloc;
-
-typedef struct {
-	int region;
-	int region_offset;
-} drmI830MemFree;
-
-typedef struct {
-	int region;
-	int size;
-	int start;	
-} drmI830MemInitHeap;
-
-typedef struct {
-	int region;
-} drmI830MemDestroyHeap;
-
-#define MMIO_READ  0
-#define MMIO_WRITE 1
-
-#define MMIO_REGS_IA_PRIMATIVES_COUNT           0
-#define MMIO_REGS_IA_VERTICES_COUNT             1
-#define MMIO_REGS_VS_INVOCATION_COUNT           2
-#define MMIO_REGS_GS_PRIMITIVES_COUNT           3
-#define MMIO_REGS_GS_INVOCATION_COUNT           4
-#define MMIO_REGS_CL_PRIMITIVES_COUNT           5
-#define MMIO_REGS_CL_INVOCATION_COUNT           6
-#define MMIO_REGS_PS_INVOCATION_COUNT           7
-#define MMIO_REGS_PS_DEPTH_COUNT                8
-
-typedef struct {
-        unsigned int read_write:1;
-        unsigned int reg:31;
-        void __user *data;
-} drmI830MMIO;
-
-#endif /* _I830_DRM_H_ */
diff --git a/src/mesa/drivers/dri/i965/server/i830_dri.h b/src/mesa/drivers/dri/i965/server/i830_dri.h
deleted file mode 100644
index 22951812ad3..00000000000
--- a/src/mesa/drivers/dri/i965/server/i830_dri.h
+++ /dev/null
@@ -1,63 +0,0 @@
-/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/i810/i830_dri.h,v 1.4 2002/10/30 12:52:18 alanh Exp $ */
-
-#ifndef _I830_DRI_H
-#define _I830_DRI_H
-
-#include "xf86drm.h"
-#include "i830_common.h"
-
-#define I830_MAX_DRAWABLES 256
-
-#define I830_MAJOR_VERSION 1
-#define I830_MINOR_VERSION 3
-#define I830_PATCHLEVEL 0
-
-#define I830_REG_SIZE 0x80000
-
-typedef struct _I830DRIRec {
-   drm_handle_t regs;
-   drmSize regsSize;
-
-   drmSize unused1; /* backbufferSize */
-   drm_handle_t unused2; /* backbuffer */
-
-   drmSize unused3; /* depthbufferSize */
-   drm_handle_t unused4; /* depthbuffer */
-
-   drmSize unused5; /* rotatedSize /*/
-   drm_handle_t unused6; /* rotatedbuffer */
-
-   drm_handle_t unused7; /* textures */
-   int unused8; /* textureSize */
-
-   drm_handle_t unused9; /* agp_buffers */
-   drmSize unused10; /* agp_buf_size */
-
-   int deviceID;
-   int width;
-   int height;
-   int mem;
-   int cpp;
-   int bitsPerPixel;
-
-   int unused11[8]; /* was front/back/depth/rotated offset/pitch */
-
-   int unused12; /* logTextureGranularity */
-   int unused13; /* textureOffset */
-
-   int irq;
-   int sarea_priv_offset;
-} I830DRIRec, *I830DRIPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830ConfigPrivRec, *I830ConfigPrivPtr;
-
-typedef struct {
-   /* Nothing here yet */
-   int dummy;
-} I830DRIContextRec, *I830DRIContextPtr;
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/server/intel.h b/src/mesa/drivers/dri/i965/server/intel.h
deleted file mode 100644
index d7858a20c8d..00000000000
--- a/src/mesa/drivers/dri/i965/server/intel.h
+++ /dev/null
@@ -1,328 +0,0 @@
-#ifndef _INTEL_H_
-#define _INTEL_H_
-
-#include "xf86drm.h"		/* drm_handle_t, etc */
-
-/* Intel */
-#ifndef PCI_CHIP_I810
-#define PCI_CHIP_I810              0x7121
-#define PCI_CHIP_I810_DC100        0x7123
-#define PCI_CHIP_I810_E            0x7125
-#define PCI_CHIP_I815              0x1132
-#define PCI_CHIP_I810_BRIDGE       0x7120
-#define PCI_CHIP_I810_DC100_BRIDGE 0x7122
-#define PCI_CHIP_I810_E_BRIDGE     0x7124
-#define PCI_CHIP_I815_BRIDGE       0x1130
-#endif
-
-#define PCI_CHIP_845_G			0x2562
-#define PCI_CHIP_I830_M			0x3577
-
-#ifndef PCI_CHIP_I855_GM
-#define PCI_CHIP_I855_GM	   0x3582
-#define PCI_CHIP_I855_GM_BRIDGE	   0x3580
-#endif
-
-#ifndef PCI_CHIP_I865_G
-#define PCI_CHIP_I865_G		   0x2572
-#define PCI_CHIP_I865_G_BRIDGE	   0x2570
-#endif
-
-#ifndef PCI_CHIP_I915_G
-#define PCI_CHIP_I915_G		   0x2582
-#define PCI_CHIP_I915_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I915_GM
-#define PCI_CHIP_I915_GM	   0x2592
-#define PCI_CHIP_I915_GM_BRIDGE	   0x2590
-#endif
-
-#ifndef PCI_CHIP_E7221_G
-#define PCI_CHIP_E7221_G	   0x258A
-/* Same as I915_G_BRIDGE */
-#define PCI_CHIP_E7221_G_BRIDGE	   0x2580
-#endif
-
-#ifndef PCI_CHIP_I945_G
-#define PCI_CHIP_I945_G        0x2772
-#define PCI_CHIP_I945_G_BRIDGE 0x2770
-#endif
-
-#ifndef PCI_CHIP_I945_GM
-#define PCI_CHIP_I945_GM        0x27A2
-#define PCI_CHIP_I945_GM_BRIDGE 0x27A0
-#endif
-
-#define IS_I810(pI810) (pI810->Chipset == PCI_CHIP_I810 ||	\
-			pI810->Chipset == PCI_CHIP_I810_DC100 || \
-			pI810->Chipset == PCI_CHIP_I810_E)
-#define IS_I815(pI810) (pI810->Chipset == PCI_CHIP_I815)
-#define IS_I830(pI810) (pI810->Chipset == PCI_CHIP_I830_M)
-#define IS_845G(pI810) (pI810->Chipset == PCI_CHIP_845_G)
-#define IS_I85X(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM)
-#define IS_I852(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I852_GM || pI810->variant == I852_GME))
-#define IS_I855(pI810)  (pI810->Chipset == PCI_CHIP_I855_GM && (pI810->variant == I855_GM || pI810->variant == I855_GME))
-#define IS_I865G(pI810) (pI810->Chipset == PCI_CHIP_I865_G)
-
-#define IS_I915G(pI810) (pI810->Chipset == PCI_CHIP_I915_G || pI810->Chipset == PCI_CHIP_E7221_G)
-#define IS_I915GM(pI810) (pI810->Chipset == PCI_CHIP_I915_GM)
-#define IS_I945G(pI810) (pI810->Chipset == PCI_CHIP_I945_G)
-#define IS_I945GM(pI810) (pI810->Chipset == PCI_CHIP_I945_GM)
-#define IS_I9XX(pI810) (IS_I915G(pI810) || IS_I915GM(pI810) || IS_I945G(pI810) || IS_I945GM(pI810))
-
-#define IS_MOBILE(pI810) (IS_I830(pI810) || IS_I85X(pI810) || IS_I915GM(pI810) || IS_I945GM(pI810))
-
-#define I830_GMCH_CTRL		0x52
-
-
-#define I830_GMCH_GMS_MASK			0x70
-#define I830_GMCH_GMS_DISABLED		0x00
-#define I830_GMCH_GMS_LOCAL			0x10
-#define I830_GMCH_GMS_STOLEN_512	0x20
-#define I830_GMCH_GMS_STOLEN_1024	0x30
-#define I830_GMCH_GMS_STOLEN_8192	0x40
-
-#define I855_GMCH_GMS_MASK			(0x7 << 4)
-#define I855_GMCH_GMS_DISABLED			0x00
-#define I855_GMCH_GMS_STOLEN_1M			(0x1 << 4)
-#define I855_GMCH_GMS_STOLEN_4M			(0x2 << 4)
-#define I855_GMCH_GMS_STOLEN_8M			(0x3 << 4)
-#define I855_GMCH_GMS_STOLEN_16M		(0x4 << 4)
-#define I855_GMCH_GMS_STOLEN_32M		(0x5 << 4)
-#define I915G_GMCH_GMS_STOLEN_48M		(0x6 << 4)
-#define I915G_GMCH_GMS_STOLEN_64M		(0x7 << 4)
-
-typedef unsigned char Bool;
-#define TRUE 1
-#define FALSE 0
-
-#define PIPE_NONE	0<<0
-#define PIPE_CRT	1<<0
-#define PIPE_TV		1<<1
-#define PIPE_DFP	1<<2
-#define PIPE_LFP	1<<3
-#define PIPE_CRT2	1<<4
-#define PIPE_TV2	1<<5
-#define PIPE_DFP2	1<<6
-#define PIPE_LFP2	1<<7
-
-typedef struct _I830MemPool *I830MemPoolPtr;
-typedef struct _I830MemRange *I830MemRangePtr;
-typedef struct _I830MemRange {
-   long Start;
-   long End;
-   long Size;
-   unsigned long Physical;
-   unsigned long Offset;		/* Offset of AGP-allocated portion */
-   unsigned long Alignment;
-   drm_handle_t Key;
-   unsigned long Pitch; // add pitch
-   I830MemPoolPtr Pool;
-} I830MemRange;
-
-typedef struct _I830MemPool {
-   I830MemRange Total;
-   I830MemRange Free;
-   I830MemRange Fixed;
-   I830MemRange Allocated;
-} I830MemPool;
-
-typedef struct {
-   int tail_mask;
-   I830MemRange mem;
-   unsigned char *virtual_start;
-   int head;
-   int tail;
-   int space;
-} I830RingBuffer;
-
-typedef struct _I830Rec {
-   unsigned char *MMIOBase;
-   unsigned char *FbBase;
-   int cpp;
-
-   unsigned int bios_version;
-
-   /* These are set in PreInit and never changed. */
-   long FbMapSize;
-   long TotalVideoRam;
-   I830MemRange StolenMemory;		/* pre-allocated memory */
-   long BIOSMemorySize;			/* min stolen pool size */
-   int BIOSMemSizeLoc;
-
-   /* These change according to what has been allocated. */
-   long FreeMemory;
-   I830MemRange MemoryAperture;
-   I830MemPool StolenPool;
-   long allocatedMemory;
-
-   /* Regions allocated either from the above pools, or from agpgart. */
-   /* for single and dual head configurations */
-   I830MemRange FrontBuffer;
-   I830MemRange FrontBuffer2;
-   I830MemRange Scratch;
-   I830MemRange Scratch2;
-
-   I830RingBuffer *LpRing;
-
-   I830MemRange BackBuffer;
-   I830MemRange DepthBuffer;
-   I830MemRange TexMem;
-   int TexGranularity;
-   I830MemRange ContextMem;
-   int drmMinor;
-   Bool have3DWindows;
-
-   Bool NeedRingBufferLow;
-   Bool allowPageFlip;
-   Bool disableTiling;
-
-   int Chipset;
-   unsigned long LinearAddr;
-   unsigned long MMIOAddr;
-
-   drmSize           registerSize;     /**< \brief MMIO register map size */
-   drm_handle_t         registerHandle;   /**< \brief MMIO register map handle */
-  //   IOADDRESS ioBase;
-   int               irq;              /**< \brief IRQ number */
-   int GttBound;
-
-   drm_handle_t ring_map;
-   unsigned int Fence[8];
-
-} I830Rec;
-
-/*
- * 12288 is set as the maximum, chosen because it is enough for
- * 1920x1440@32bpp with a 2048 pixel line pitch with some to spare.
- */
-#define I830_MAXIMUM_VBIOS_MEM		12288
-#define I830_DEFAULT_VIDEOMEM_2D	(MB(32) / 1024)
-#define I830_DEFAULT_VIDEOMEM_3D	(MB(64) / 1024)
-
-/* Flags for memory allocation function */
-#define FROM_ANYWHERE			0x00000000
-#define FROM_POOL_ONLY			0x00000001
-#define FROM_NEW_ONLY			0x00000002
-#define FROM_MASK			0x0000000f
-
-#define ALLOCATE_AT_TOP			0x00000010
-#define ALLOCATE_AT_BOTTOM		0x00000020
-#define FORCE_GAPS			0x00000040
-
-#define NEED_PHYSICAL_ADDR		0x00000100
-#define ALIGN_BOTH_ENDS			0x00000200
-#define FORCE_LOW			0x00000400
-
-#define ALLOC_NO_TILING			0x00001000
-#define ALLOC_INITIAL			0x00002000
-
-#define ALLOCATE_DRY_RUN		0x80000000
-
-/* Chipset registers for VIDEO BIOS memory RW access */
-#define _855_DRAM_RW_CONTROL 0x58
-#define _845_DRAM_RW_CONTROL 0x90
-#define DRAM_WRITE    0x33330000
-
-#define KB(x) ((x) * 1024)
-#define MB(x) ((x) * KB(1024))
-
-#define GTT_PAGE_SIZE			KB(4)
-#define ROUND_TO(x, y)			(((x) + (y) - 1) / (y) * (y))
-#define ROUND_DOWN_TO(x, y)		((x) / (y) * (y))
-#define ROUND_TO_PAGE(x)		ROUND_TO((x), GTT_PAGE_SIZE)
-#define ROUND_TO_MB(x)			ROUND_TO((x), MB(1))
-#define PRIMARY_RINGBUFFER_SIZE		KB(128)
-
-
-/* Ring buffer registers, p277, overview p19
- */
-#define LP_RING     0x2030
-#define HP_RING     0x2040
-
-#define RING_TAIL      0x00
-#define TAIL_ADDR           0x000FFFF8
-#define I830_TAIL_MASK	    0x001FFFF8
-
-#define RING_HEAD      0x04
-#define HEAD_WRAP_COUNT     0xFFE00000
-#define HEAD_WRAP_ONE       0x00200000
-#define HEAD_ADDR           0x001FFFFC
-#define I830_HEAD_MASK      0x001FFFFC
-
-#define RING_START     0x08
-#define START_ADDR          0x03FFFFF8
-#define I830_RING_START_MASK	0xFFFFF000
-
-#define RING_LEN       0x0C
-#define RING_NR_PAGES       0x001FF000 
-#define I830_RING_NR_PAGES	0x001FF000
-#define RING_REPORT_MASK    0x00000006
-#define RING_REPORT_64K     0x00000002
-#define RING_REPORT_128K    0x00000004
-#define RING_NO_REPORT      0x00000000
-#define RING_VALID_MASK     0x00000001
-#define RING_VALID          0x00000001
-#define RING_INVALID        0x00000000
-
-
-/* Fence/Tiling ranges [0..7]
- */
-#define FENCE            0x2000
-#define FENCE_NR         8
-
-#define I915G_FENCE_START_MASK	0x0ff00000
-
-#define I830_FENCE_START_MASK	0x07f80000
-
-#define FENCE_START_MASK    0x03F80000
-#define FENCE_X_MAJOR       0x00000000
-#define FENCE_Y_MAJOR       0x00001000
-#define FENCE_SIZE_MASK     0x00000700
-#define FENCE_SIZE_512K     0x00000000
-#define FENCE_SIZE_1M       0x00000100
-#define FENCE_SIZE_2M       0x00000200
-#define FENCE_SIZE_4M       0x00000300
-#define FENCE_SIZE_8M       0x00000400
-#define FENCE_SIZE_16M      0x00000500
-#define FENCE_SIZE_32M      0x00000600
-#define FENCE_SIZE_64M	    0x00000700
-#define I915G_FENCE_SIZE_1M       0x00000000
-#define I915G_FENCE_SIZE_2M       0x00000100
-#define I915G_FENCE_SIZE_4M       0x00000200
-#define I915G_FENCE_SIZE_8M       0x00000300
-#define I915G_FENCE_SIZE_16M      0x00000400
-#define I915G_FENCE_SIZE_32M      0x00000500
-#define I915G_FENCE_SIZE_64M	0x00000600
-#define I915G_FENCE_SIZE_128M	0x00000700
-#define FENCE_PITCH_1       0x00000000
-#define FENCE_PITCH_2       0x00000010
-#define FENCE_PITCH_4       0x00000020
-#define FENCE_PITCH_8       0x00000030
-#define FENCE_PITCH_16      0x00000040
-#define FENCE_PITCH_32      0x00000050
-#define FENCE_PITCH_64	    0x00000060
-#define FENCE_VALID         0x00000001
-
-#include <mmio.h>
-
-#  define MMIO_IN8(base, offset) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset))
-#  define MMIO_IN32(base, offset) \
-	read_MMIO_LE32(base, offset)
-#  define MMIO_OUT8(base, offset, val) \
-	*(volatile unsigned char *)(((unsigned char*)(base)) + (offset)) = (val)
-#  define MMIO_OUT32(base, offset, val) \
-	*(volatile unsigned int *)(void *)(((unsigned char*)(base)) + (offset)) = CPU_TO_LE32(val)
-
-
-				/* Memory mapped register access macros */
-#define INREG8(addr)        MMIO_IN8(MMIO, addr)
-#define INREG(addr)         MMIO_IN32(MMIO, addr)
-#define OUTREG8(addr, val)  MMIO_OUT8(MMIO, addr, val)
-#define OUTREG(addr, val)   MMIO_OUT32(MMIO, addr, val)
-
-#define DSPABASE		0x70184
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/server/intel_dri.c b/src/mesa/drivers/dri/i965/server/intel_dri.c
deleted file mode 100644
index 169fdbece30..00000000000
--- a/src/mesa/drivers/dri/i965/server/intel_dri.c
+++ /dev/null
@@ -1,1282 +0,0 @@
-/**
- * \file server/intel_dri.c
- * \brief File to perform the device-specific initialization tasks typically
- * done in the X server.
- *
- * Here they are converted to run in the client (or perhaps a standalone
- * process), and to work with the frame buffer device rather than the X
- * server infrastructure.
- * 
- * Copyright (C) 2006 Dave Airlie ([email protected])
-
- 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 THE COPYRIGHT HOLDERS AND/OR THEIR SUPPLIERS BE LIABLE FOR
- ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-*/
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <errno.h>
-#include <unistd.h>
-
-#include "driver.h"
-#include "drm.h"
-
-#include "intel.h"
-#include "i830_dri.h"
-
-#include "memops.h"
-#include "pciaccess.h"
-
-static size_t drm_page_size;
-static int nextTile = 0;
-#define xf86DrvMsg(...) do {} while(0)
-
-static const int pitches[] = {
-  128 * 8,
-  128 * 16,
-  128 * 32,
-  128 * 64,
-  0
-};
-
-static Bool I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea);
-
-static unsigned long
-GetBestTileAlignment(unsigned long size)
-{
-   unsigned long i;
-
-   for (i = KB(512); i < size; i <<= 1)
-      ;
-
-   if (i > MB(64))
-      i = MB(64);
-
-   return i;
-}
-
-static void SetFenceRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-  for (i = 0; i < 8; i++) {
-    OUTREG(FENCE + i * 4, pI830->Fence[i]);
-    //    if (I810_DEBUG & DEBUG_VERBOSE_VGA)
-    fprintf(stderr,"Fence Register : %x\n", pI830->Fence[i]);
-  }
-}
-
-/* Tiled memory is good... really, really good...
- *
- * Need to make it less likely that we miss out on this - probably
- * need to move the frontbuffer away from the 'guarenteed' alignment
- * of the first memory segment, or perhaps allocate a discontigous
- * framebuffer to get more alignment 'sweet spots'.
- */
-static void
-SetFence(const DRIDriverContext *ctx, I830Rec *pI830,
-	 int nr, unsigned int start, unsigned int pitch,
-         unsigned int size)
-{
-   unsigned int val;
-   unsigned int fence_mask = 0;
-   unsigned int fence_pitch;
-
-   if (nr < 0 || nr > 7) {
-      fprintf(stderr,
-		 "SetFence: fence %d out of range\n",nr);
-      return;
-   }
-
-   pI830->Fence[nr] = 0;
-
-   if (IS_I9XX(pI830))
-   	fence_mask = ~I915G_FENCE_START_MASK;
-   else
-   	fence_mask = ~I830_FENCE_START_MASK;
-
-   if (start & fence_mask) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not %s aligned\n",
-		 nr, start, (IS_I9XX(pI830)) ? "1MB" : "512k");
-      return;
-   }
-
-   if (start % size) {
-      fprintf(stderr,
-		 "SetFence: %d: start (0x%08x) is not size (%dk) aligned\n",
-		 nr, start, size / 1024);
-      return;
-   }
-
-   if (pitch & 127) {
-      fprintf(stderr,
-		 "SetFence: %d: pitch (%d) not a multiple of 128 bytes\n",
-		 nr, pitch);
-      return;
-   }
-
-   val = (start | FENCE_X_MAJOR | FENCE_VALID);
-
-   if (IS_I9XX(pI830)) {
-   	switch (size) {
-	   case MB(1):
-      		val |= I915G_FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= I915G_FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= I915G_FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= I915G_FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= I915G_FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= I915G_FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= I915G_FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-    } else {
-   	switch (size) {
-	   case KB(512):
-      		val |= FENCE_SIZE_512K;
-      		break;
-	   case MB(1):
-      		val |= FENCE_SIZE_1M;
-      		break;
-   	   case MB(2):
-      		val |= FENCE_SIZE_2M;
-      		break;
-   	   case MB(4):
-      		val |= FENCE_SIZE_4M;
-      		break;
-   	   case MB(8):
-      		val |= FENCE_SIZE_8M;
-      		break;
-   	   case MB(16):
-      		val |= FENCE_SIZE_16M;
-      		break;
-   	   case MB(32):
-      		val |= FENCE_SIZE_32M;
-      		break;
-   	   case MB(64):
-      		val |= FENCE_SIZE_64M;
-      		break;
-   	   default:
-      		fprintf(stderr,
-		 "SetFence: %d: illegal size (%d kByte)\n", nr, size / 1024);
-      		return;
-   	}
-   }
-
-   if (IS_I9XX(pI830))
-	fence_pitch = pitch / 512;
-   else
-	fence_pitch = pitch / 128;
-
-   switch (fence_pitch) {
-   case 1:
-      val |= FENCE_PITCH_1;
-      break;
-   case 2:
-      val |= FENCE_PITCH_2;
-      break;
-   case 4:
-      val |= FENCE_PITCH_4;
-      break;
-   case 8:
-      val |= FENCE_PITCH_8;
-      break;
-   case 16:
-      val |= FENCE_PITCH_16;
-      break;
-   case 32:
-      val |= FENCE_PITCH_32;
-      break;
-   case 64:
-      val |= FENCE_PITCH_64;
-      break;
-   default:
-      fprintf(stderr,
-		 "SetFence: %d: illegal pitch (%d)\n", nr, pitch);
-      return;
-   }
-
-   pI830->Fence[nr] = val;
-}
-
-static Bool
-MakeTiles(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *pMem)
-{
-   int pitch, ntiles, i;
-
-   pitch = pMem->Pitch * ctx->cpp;
-   /*
-    * Simply try to break the region up into at most four pieces of size
-    * equal to the alignment.
-    */
-   ntiles = ROUND_TO(pMem->Size, pMem->Alignment) / pMem->Alignment;
-   if (ntiles >= 4) {
-      return FALSE;
-   }
-
-   for (i = 0; i < ntiles; i++, nextTile++) {
-     SetFence(ctx, pI830, nextTile, pMem->Start + i * pMem->Alignment,
-	       pitch, pMem->Alignment);
-   }
-   return TRUE;
-}
-
-static void I830SetupMemoryTiling(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  int i;
-
-  /* Clear out */
-  for (i = 0; i < 8; i++)
-    pI830->Fence[i] = 0;
-  
-  nextTile = 0;
-
-  if (pI830->BackBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->BackBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the back buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the back buffer.\n");
-      pI830->allowPageFlip = FALSE;
-    }
-  }
-  
-  if (pI830->DepthBuffer.Alignment >= KB(512)) {
-    if (MakeTiles(ctx, pI830, &(pI830->DepthBuffer))) {
-      fprintf(stderr,
-		 "Activating tiled memory for the depth buffer.\n");
-    } else {
-      fprintf(stderr,
-		 "MakeTiles failed for the depth buffer.\n");
-    }
-  }
-
-  return;
-}
-
-static int I830DetectMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  struct pci_device host_bridge;
-  uint32_t gmch_ctrl;
-  int memsize = 0;
-  int range;
-
-  memset(&host_bridge, 0, sizeof(host_bridge));
-
-  pci_device_cfg_read_u32(&host_bridge, &gmch_ctrl, I830_GMCH_CTRL);
-  
-  /* We need to reduce the stolen size, by the GTT and the popup.
-   * The GTT varying according the the FbMapSize and the popup is 4KB */
-  range = (ctx->shared.fbSize / (1024*1024)) + 4;
-
-   if (IS_I85X(pI830) || IS_I865G(pI830) || IS_I9XX(pI830)) {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I855_GMCH_GMS_STOLEN_1M:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_4M:
-	 memsize = MB(4) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_8M:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_16M:
-	 memsize = MB(16) - KB(range);
-	 break;
-      case I855_GMCH_GMS_STOLEN_32M:
-	 memsize = MB(32) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_48M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(48) - KB(range);
-	 break;
-      case I915G_GMCH_GMS_STOLEN_64M:
-	 if (IS_I9XX(pI830))
-	    memsize = MB(64) - KB(range);
-	 break;
-      }
-   } else {
-      switch (gmch_ctrl & I830_GMCH_GMS_MASK) {
-      case I830_GMCH_GMS_STOLEN_512:
-	 memsize = KB(512) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_1024:
-	 memsize = MB(1) - KB(range);
-	 break;
-      case I830_GMCH_GMS_STOLEN_8192:
-	 memsize = MB(8) - KB(range);
-	 break;
-      case I830_GMCH_GMS_LOCAL:
-	 memsize = 0;
-	 xf86DrvMsg(pScrn->scrnIndex, X_WARNING,
-		    "Local memory found, but won't be used.\n");
-	 break;
-      }
-   }
-   if (memsize > 0) {
-     fprintf(stderr,
-		 "detected %d kB stolen memory.\n", memsize / 1024);
-   } else {
-     fprintf(stderr,
-		 "no video memory detected.\n");
-   }
-   return memsize;
-}
-
-static int AgpInit(const DRIDriverContext *ctx, I830Rec *info)
-{
-  unsigned long mode = 0x4;
-
-  if (drmAgpAcquire(ctx->drmFD) < 0) {
-    fprintf(stderr, "[gart] AGP not available\n");
-    return 0;
-  }
-  
-  if (drmAgpEnable(ctx->drmFD, mode) < 0) {
-    fprintf(stderr, "[gart] AGP not enabled\n");
-    drmAgpRelease(ctx->drmFD);
-    return 0;
-  }
-  else
-    fprintf(stderr, "[gart] AGP enabled at %dx\n", ctx->agpmode);
-
-  return 1;
-}
-
-/*
- * Allocate memory from the given pool.  Grow the pool if needed and if
- * possible.
- */
-static unsigned long
-AllocFromPool(const DRIDriverContext *ctx, I830Rec *pI830, 
-	      I830MemRange *result, I830MemPool *pool,
-	      long size, unsigned long alignment, int flags)
-{
-   long needed, start, end;
-
-   if (!result || !pool || !size)
-      return 0;
-
-   /* Calculate how much space is needed. */
-   if (alignment <= GTT_PAGE_SIZE)
-      needed = size;
-   else {
-	 start = ROUND_TO(pool->Free.Start, alignment);
-	 end = ROUND_TO(start + size, alignment);
-	 needed = end - pool->Free.Start;
-   }
-   if (needed > pool->Free.Size) {
-     return 0;
-   }
-
-   result->Start = ROUND_TO(pool->Free.Start, alignment);
-   pool->Free.Start += needed;
-   result->End = pool->Free.Start;
-
-   pool->Free.Size = pool->Free.End - pool->Free.Start;
-   result->Size = result->End - result->Start;
-   result->Pool = pool;
-   result->Alignment = alignment;
-   return needed;
-}
-
-static unsigned long AllocFromAGP(const DRIDriverContext *ctx, I830Rec *pI830, long size, unsigned long alignment, I830MemRange  *result)
-{
-   unsigned long start, end;
-   unsigned long newApStart, newApEnd;
-   int ret;
-   if (!result || !size)
-      return 0;
-   
-   if (!alignment)
-     alignment = 4;
-
-   start = ROUND_TO(pI830->MemoryAperture.Start, alignment);
-   end = ROUND_TO(start + size, alignment);
-   newApStart = end;
-   newApEnd = pI830->MemoryAperture.End;
-
-   ret=drmAgpAlloc(ctx->drmFD, size, 0, &(result->Physical), (drm_handle_t *)&(result->Key));
-   
-   if (ret)
-   {
-     fprintf(stderr,"drmAgpAlloc failed %d\n", ret);
-     return 0;
-   }
-   pI830->allocatedMemory += size;
-   pI830->MemoryAperture.Start = newApStart;
-   pI830->MemoryAperture.End = newApEnd;
-   pI830->MemoryAperture.Size = newApEnd - newApStart;
-   //   pI830->FreeMemory -= size;
-   result->Start = start;
-   result->End = start + size;
-   result->Size = size;
-   result->Offset = start;
-   result->Alignment = alignment;
-   result->Pool = NULL;
-  
-   return size;
-}
-
-unsigned long
-I830AllocVidMem(const DRIDriverContext *ctx, I830Rec *pI830, I830MemRange *result, I830MemPool *pool, long size, unsigned long alignment, int flags)
-{
-  int ret;
-
-  if (!result)
-    return 0;
-
-   /* Make sure these are initialised. */
-   result->Size = 0;
-   result->Key = -1;
-
-   if (!size) {
-      return 0;
-   }
-
-   if (pool->Free.Size < size)
-     return AllocFromAGP(ctx, pI830, size, alignment, result);
-   else
-   {
-     ret = AllocFromPool(ctx, pI830, result, pool, size, alignment, flags);
-
-     if (ret==0)
-       return AllocFromAGP(ctx, pI830, size, alignment, result);
-     return ret;
-   }
-}
-
-static Bool BindAgpRange(const DRIDriverContext *ctx, I830MemRange *mem)
-{
-  if (!mem)
-    return FALSE;
-  
-  if (mem->Key == -1)
-    return TRUE;
-
-  return !drmAgpBind(ctx->drmFD, mem->Key, mem->Offset);
-}
-
-/* simple memory allocation routines needed */
-/* put ring buffer in low memory */
-/* need to allocate front, back, depth buffers aligned correctly,
-   allocate ring buffer, 
-*/
-
-/* */
-static Bool
-I830AllocateMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned long size, ret;
-  unsigned long lines, lineSize, align;
-
-  /* allocate ring buffer */
-  memset(pI830->LpRing, 0, sizeof(I830RingBuffer));
-  pI830->LpRing->mem.Key = -1;
-
-  size = PRIMARY_RINGBUFFER_SIZE;
-  
-  ret = I830AllocVidMem(ctx, pI830, &pI830->LpRing->mem, &pI830->StolenPool, size, 0x1000, 0);
-  
-  if (ret != size)
-  {
-    fprintf(stderr,"unable to allocate ring buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  pI830->LpRing->tail_mask = pI830->LpRing->mem.Size - 1;
-
-  
-  /* allocate front buffer */
-  memset(&(pI830->FrontBuffer), 0, sizeof(pI830->FrontBuffer));
-  pI830->FrontBuffer.Key = -1;
-  pI830->FrontBuffer.Pitch = ctx->shared.virtualWidth;
-
-  align = KB(512);  
-
-  lineSize = ctx->shared.virtualWidth * ctx->cpp;
-  lines = (ctx->shared.virtualHeight + 15) / 16 * 16;
-  size = lineSize * lines;
-  size = ROUND_TO_PAGE(size);
-
-  align = GetBestTileAlignment(size);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->FrontBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate front buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->BackBuffer), 0, sizeof(pI830->BackBuffer));
-  pI830->BackBuffer.Key = -1;
-  pI830->BackBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->BackBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate back buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->DepthBuffer), 0, sizeof(pI830->DepthBuffer));
-  pI830->DepthBuffer.Key = -1;
-  pI830->DepthBuffer.Pitch = ctx->shared.virtualWidth;
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->DepthBuffer, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate depth buffer %ld\n", ret);
-    return FALSE;
-  }
-
-  memset(&(pI830->ContextMem), 0, sizeof(pI830->ContextMem));
-  pI830->ContextMem.Key = -1;
-  size = KB(32);
-
-  ret = I830AllocVidMem(ctx, pI830, &pI830->ContextMem, &pI830->StolenPool, size, align, 0);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate context buffer %ld\n", ret);
-    return FALSE;
-  }
-  
-  memset(&(pI830->TexMem), 0, sizeof(pI830->TexMem));
-  pI830->TexMem.Key = -1;
-
-  size = 32768 * 1024;
-  ret = AllocFromAGP(ctx, pI830, size, align, &pI830->TexMem);
-  if (ret < size)
-  {
-    fprintf(stderr,"unable to allocate texture memory %ld\n", ret);
-    return FALSE;
-  }
-
-  return TRUE;
-}
-
-static Bool
-I830BindMemory(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  if (!BindAgpRange(ctx, &pI830->LpRing->mem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->FrontBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->BackBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->DepthBuffer))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->ContextMem))
-    return FALSE;
-  if (!BindAgpRange(ctx, &pI830->TexMem))
-    return FALSE;
-
-  return TRUE;
-}
-
-static Bool
-I830CleanupDma(const DRIDriverContext *ctx)
-{
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_CLEANUP_DMA;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-     fprintf(stderr, "I830 Dma Cleanup Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830InitDma(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-   I830RingBuffer *ring = pI830->LpRing;
-   drmI830Init info;
-
-   memset(&info, 0, sizeof(drmI830Init));
-   info.func = I830_INIT_DMA;
-
-   info.ring_start = ring->mem.Start + pI830->LinearAddr;
-   info.ring_end = ring->mem.End + pI830->LinearAddr;
-   info.ring_size = ring->mem.Size;
-
-   info.mmio_offset = (unsigned int)ctx->MMIOStart;
-
-   info.sarea_priv_offset = sizeof(drm_sarea_t);
-
-   info.front_offset = pI830->FrontBuffer.Start;
-   info.back_offset = pI830->BackBuffer.Start;
-   info.depth_offset = pI830->DepthBuffer.Start;
-   info.w = ctx->shared.virtualWidth;
-   info.h = ctx->shared.virtualHeight;
-   info.pitch = ctx->shared.virtualWidth;
-   info.back_pitch = pI830->BackBuffer.Pitch;
-   info.depth_pitch = pI830->DepthBuffer.Pitch;
-   info.cpp = ctx->cpp;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT,
-		       &info, sizeof(drmI830Init))) {
-      fprintf(stderr,
-		 "I830 Dma Initialization Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static int I830CheckDRMVersion( const DRIDriverContext *ctx,
-				  I830Rec *pI830 )
-{
-   drmVersionPtr  version;
-
-   version = drmGetVersion(ctx->drmFD);
-
-   if (version) {
-     int req_minor, req_patch;
-
-     req_minor = 4;
-     req_patch = 0;	
-
-     if (version->version_major != 1 ||
-	 version->version_minor < req_minor ||
-	 (version->version_minor == req_minor && 
-	  version->version_patchlevel < req_patch)) {
-       /* Incompatible drm version */
-       fprintf(stderr,
-	       "[dri] I830DRIScreenInit failed because of a version "
-	       "mismatch.\n"
-	       "[dri] i915.o kernel module version is %d.%d.%d "
-	       "but version 1.%d.%d or newer is needed.\n"
-	       "[dri] Disabling DRI.\n",
-	       version->version_major,
-	       version->version_minor,
-	       version->version_patchlevel,
-	       req_minor,
-	       req_patch);
-       drmFreeVersion(version);
-       return 0;
-     }
-     
-     pI830->drmMinor = version->version_minor;
-     drmFreeVersion(version);
-   }
-   return 1;
-}
-
-static void
-I830SetRingRegs(const DRIDriverContext *ctx, I830Rec *pI830)
-{
-  unsigned int itemp;
-  unsigned char *MMIO = ctx->MMIOAddress;
-
-   OUTREG(LP_RING + RING_LEN, 0);
-   OUTREG(LP_RING + RING_TAIL, 0);
-   OUTREG(LP_RING + RING_HEAD, 0);
-
-   if ((long)(pI830->LpRing->mem.Start & I830_RING_START_MASK) !=
-       pI830->LpRing->mem.Start) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer start (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Start, I830_RING_START_MASK);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = pI830->LpRing->mem.Start & I830_RING_START_MASK;
-   OUTREG(LP_RING + RING_START, itemp);
-
-   if (((pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES) !=
-       pI830->LpRing->mem.Size - 4096) {
-      fprintf(stderr,
-		 "I830SetRingRegs: Ring buffer size - 4096 (%lx) violates its "
-		 "mask (%x)\n", pI830->LpRing->mem.Size - 4096,
-		 I830_RING_NR_PAGES);
-   }
-   /* Don't care about the old value.  Reserved bits must be zero anyway. */
-   itemp = (pI830->LpRing->mem.Size - 4096) & I830_RING_NR_PAGES;
-   itemp |= (RING_NO_REPORT | RING_VALID);
-   OUTREG(LP_RING + RING_LEN, itemp);
-
-   pI830->LpRing->head = INREG(LP_RING + RING_HEAD) & I830_HEAD_MASK;
-   pI830->LpRing->tail = INREG(LP_RING + RING_TAIL);
-   pI830->LpRing->space = pI830->LpRing->head - (pI830->LpRing->tail + 8);
-   if (pI830->LpRing->space < 0)
-      pI830->LpRing->space += pI830->LpRing->mem.Size;
-
-   SetFenceRegs(ctx, pI830);
-   
-   /* RESET THE DISPLAY PIPE TO POINT TO THE FRONTBUFFER - hacky
-      hacky hacky */
-   OUTREG(DSPABASE, pI830->FrontBuffer.Start + pI830->LinearAddr);
-
-}
-
-static Bool
-I830SetParam(const DRIDriverContext *ctx, int param, int value)
-{
-   drmI830SetParam sp;
-
-   memset(&sp, 0, sizeof(sp));
-   sp.param = param;
-   sp.value = value;
-
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_SETPARAM, &sp, sizeof(sp))) {
-      fprintf(stderr, "I830 SetParam Failed\n");
-      return FALSE;
-   }
-
-   return TRUE;
-}
-
-static Bool
-I830DRIMapScreenRegions(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   fprintf(stderr,
-              "[drm] Mapping front buffer\n");
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->front_offset + pI830->LinearAddr),
-                 sarea->front_size,
-                 DRM_FRAME_BUFFER,  /*DRM_AGP,*/
-                 0,
-                 &sarea->front_handle) < 0) {
-     fprintf(stderr,
-	     "[drm] drmAddMap(front_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   ctx->shared.hFrameBuffer = sarea->front_handle;
-   ctx->shared.fbSize = sarea->front_size;
-   fprintf(stderr, "[drm] Front Buffer = 0x%08x\n",
-	   sarea->front_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)(sarea->back_offset),
-                 sarea->back_size, DRM_AGP, 0,
-                 &sarea->back_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(back_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Back Buffer = 0x%08x\n",
-              sarea->back_handle);
-
-   if (drmAddMap(ctx->drmFD,
-                 (drm_handle_t)sarea->depth_offset,
-                 sarea->depth_size, DRM_AGP, 0,
-                 &sarea->depth_handle) < 0) {
-      fprintf(stderr,
-                 "[drm] drmAddMap(depth_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] Depth Buffer = 0x%08x\n",
-              sarea->depth_handle);
-
-   if (drmAddMap(ctx->drmFD,
-		 (drm_handle_t)sarea->tex_offset,
-		 sarea->tex_size, DRM_AGP, 0,
-		 &sarea->tex_handle) < 0) {
-      fprintf(stderr,
-		 "[drm] drmAddMap(tex_handle) failed. Disabling DRI\n");
-      return FALSE;
-   }
-   fprintf(stderr, "[drm] textures = 0x%08x\n",
-	      sarea->tex_handle);
-
-   return TRUE;
-}
-
-
-static void
-I830DRIUnmapScreenRegions(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-#if 1
-   if (sarea->front_handle) {
-      drmRmMap(ctx->drmFD, sarea->front_handle);
-      sarea->front_handle = 0;
-   }
-#endif
-   if (sarea->back_handle) {
-      drmRmMap(ctx->drmFD, sarea->back_handle);
-      sarea->back_handle = 0;
-   }
-   if (sarea->depth_handle) {
-      drmRmMap(ctx->drmFD, sarea->depth_handle);
-      sarea->depth_handle = 0;
-   }
-   if (sarea->tex_handle) {
-      drmRmMap(ctx->drmFD, sarea->tex_handle);
-      sarea->tex_handle = 0;
-   }
-}
-
-static void
-I830InitTextureHeap(const DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-   /* Start up the simple memory manager for agp space */
-   drmI830MemInitHeap drmHeap;
-   drmHeap.region = I830_MEM_REGION_AGP;
-   drmHeap.start  = 0;
-   drmHeap.size   = sarea->tex_size;
-      
-   if (drmCommandWrite(ctx->drmFD, DRM_I830_INIT_HEAP,
-			  &drmHeap, sizeof(drmHeap))) {
-      fprintf(stderr,
-		    "[drm] Failed to initialized agp heap manager\n");
-   } else {
-      fprintf(stderr,
-		    "[drm] Initialized kernel agp heap manager, %d\n",
-		    sarea->tex_size);
-
-      I830SetParam(ctx, I830_SETPARAM_TEX_LRU_LOG_GRANULARITY, 
-		      sarea->log_tex_granularity);
-   }
-}
-
-static Bool
-I830DRIDoMappings(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  if (drmAddMap(ctx->drmFD,
-		(drm_handle_t)pI830->LpRing->mem.Start,
-		pI830->LpRing->mem.Size, DRM_AGP, 0,
-		&pI830->ring_map) < 0) {
-    fprintf(stderr,
-	    "[drm] drmAddMap(ring_map) failed. Disabling DRI\n");
-    return FALSE;
-  }
-  fprintf(stderr, "[drm] ring buffer = 0x%08x\n",
-	  pI830->ring_map);
-
-  if (I830InitDma(ctx, pI830) == FALSE) {
-    return FALSE;
-  }
-  
-   /* init to zero to be safe */
-
-  I830DRIMapScreenRegions(ctx, pI830, sarea);
-  I830InitTextureHeap(ctx, pI830, sarea);
-
-   if (ctx->pciDevice != PCI_CHIP_845_G &&
-       ctx->pciDevice != PCI_CHIP_I830_M) {
-      I830SetParam(ctx, I830_SETPARAM_USE_MI_BATCHBUFFER_START, 1 );
-   }
-
-   /* Okay now initialize the dma engine */
-   {
-      pI830->irq = drmGetInterruptFromBusID(ctx->drmFD,
-					    ctx->pciBus,
-					    ctx->pciDevice,
-					    ctx->pciFunc);
-
-      if (drmCtlInstHandler(ctx->drmFD, pI830->irq)) {
-	 fprintf(stderr,
-		    "[drm] failure adding irq handler\n");
-	 pI830->irq = 0;
-	 return FALSE;
-      }
-      else
-	 fprintf(stderr,
-		    "[drm] dma control initialized, using IRQ %d\n",
-		    pI830->irq);
-   }
-
-   fprintf(stderr, "[dri] visual configs initialized\n");
-
-   return TRUE;
-}
-
-static Bool
-I830ClearScreen(DRIDriverContext *ctx, I830Rec *pI830, drmI830Sarea *sarea)
-{
-  /* need to drmMap front and back buffers and zero them */
-  drmAddress map_addr;
-  int ret;
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->front_handle,
-	       sarea->front_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map front buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->front_size);
-  drmUnmap(map_addr, sarea->front_size);
-
-
-  ret = drmMap(ctx->drmFD,
-	       sarea->back_handle,
-	       sarea->back_size,
-	       &map_addr);
-
-  if (ret)
-  {
-    fprintf(stderr, "Unable to map back buffer\n");
-    return FALSE;
-  }
-
-  drimemsetio((char *)map_addr,
-	      0,
-	      sarea->back_size);
-  drmUnmap(map_addr, sarea->back_size);
-
-  return TRUE;
-}
-
-static Bool
-I830ScreenInit(DRIDriverContext *ctx, I830Rec *pI830)
-		  
-{
-   I830DRIPtr pI830DRI;
-   drmI830Sarea *pSAREAPriv;
-   int err;
-      
-   drm_page_size = getpagesize();   
-
-   pI830->registerSize = ctx->MMIOSize;
-   /* This is a hack for now.  We have to have more than a 4k page here
-    * because of the size of the state.  However, the state should be
-    * in a per-context mapping.  This will be added in the Mesa 3.5 port
-    * of the I830 driver.
-    */
-   ctx->shared.SAREASize = SAREA_MAX;
-
-   /* Note that drmOpen will try to load the kernel module, if needed. */
-   ctx->drmFD = drmOpen("i915", NULL );
-   if (ctx->drmFD < 0) {
-      fprintf(stderr, "[drm] drmOpen failed\n");
-      return 0;
-   }
-
-   if ((err = drmSetBusid(ctx->drmFD, ctx->pciBusID)) < 0) {
-      fprintf(stderr, "[drm] drmSetBusid failed (%d, %s), %s\n",
-	      ctx->drmFD, ctx->pciBusID, strerror(-err));
-      return 0;
-   }
-
-   if (drmAddMap( ctx->drmFD,
-		  0,
-		  ctx->shared.SAREASize,
-		  DRM_SHM,
-		  DRM_CONTAINS_LOCK,
-		  &ctx->shared.hSAREA) < 0)
-   {
-     fprintf(stderr, "[drm] drmAddMap failed\n");
-     return 0;
-   }
-
-   fprintf(stderr, "[drm] added %d byte SAREA at 0x%08x\n",
-	   ctx->shared.SAREASize, ctx->shared.hSAREA);
-   
-   if (drmMap( ctx->drmFD,
-	       ctx->shared.hSAREA,
-	       ctx->shared.SAREASize,
-	       (drmAddressPtr)(&ctx->pSAREA)) < 0)
-   {
-      fprintf(stderr, "[drm] drmMap failed\n");
-      return 0;
-   
-   }
-   
-   memset(ctx->pSAREA, 0, ctx->shared.SAREASize);
-   fprintf(stderr, "[drm] mapped SAREA 0x%08x to %p, size %d\n",
-	   ctx->shared.hSAREA, ctx->pSAREA, ctx->shared.SAREASize);
-   
-
-   if (drmAddMap(ctx->drmFD, 
-		 ctx->MMIOStart,
-		 ctx->MMIOSize,
-		 DRM_REGISTERS, 
-		 DRM_READ_ONLY, 
-		 &pI830->registerHandle) < 0) {
-      fprintf(stderr, "[drm] drmAddMap mmio failed\n");	
-      return 0;
-   }
-   fprintf(stderr,
-	   "[drm] register handle = 0x%08x\n", pI830->registerHandle);
-
-
-   if (!I830CheckDRMVersion(ctx, pI830)) {
-     return FALSE;
-   }
-
-   /* Create a 'server' context so we can grab the lock for
-    * initialization ioctls.
-    */
-   if ((err = drmCreateContext(ctx->drmFD, &ctx->serverContext)) != 0) {
-      fprintf(stderr, "%s: drmCreateContext failed %d\n", __FUNCTION__, err);
-      return 0;
-   }
-
-   DRM_LOCK(ctx->drmFD, ctx->pSAREA, ctx->serverContext, 0); 
-
-   /* Initialize the SAREA private data structure */
-   pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				 sizeof(drm_sarea_t));
-   memset(pSAREAPriv, 0, sizeof(*pSAREAPriv));
-
-   pI830->StolenMemory.Size = I830DetectMemory(ctx, pI830);
-   pI830->StolenMemory.Start = 0;
-   pI830->StolenMemory.End = pI830->StolenMemory.Size;
-
-   pI830->MemoryAperture.Start = pI830->StolenMemory.End;
-   pI830->MemoryAperture.End = KB(40000);
-   pI830->MemoryAperture.Size = pI830->MemoryAperture.End - pI830->MemoryAperture.Start;
-
-   pI830->StolenPool.Fixed = pI830->StolenMemory;
-   pI830->StolenPool.Total = pI830->StolenMemory;
-   pI830->StolenPool.Free = pI830->StolenPool.Total;
-   pI830->FreeMemory = pI830->StolenPool.Total.Size;
-
-   if (!AgpInit(ctx, pI830))
-     return FALSE;
-
-   if (I830AllocateMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   if (I830BindMemory(ctx, pI830) == FALSE)
-   {
-     return FALSE;
-   }
-
-   pSAREAPriv->front_offset = pI830->FrontBuffer.Start;
-   pSAREAPriv->front_size = pI830->FrontBuffer.Size;
-   pSAREAPriv->width = ctx->shared.virtualWidth;
-   pSAREAPriv->height = ctx->shared.virtualHeight;
-   pSAREAPriv->pitch = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualX = ctx->shared.virtualWidth;
-   pSAREAPriv->virtualY = ctx->shared.virtualHeight;
-   pSAREAPriv->back_offset = pI830->BackBuffer.Start;
-   pSAREAPriv->back_size = pI830->BackBuffer.Size;
-   pSAREAPriv->depth_offset = pI830->DepthBuffer.Start;
-   pSAREAPriv->depth_size = pI830->DepthBuffer.Size;
-   pSAREAPriv->tex_offset = pI830->TexMem.Start;
-   pSAREAPriv->tex_size = pI830->TexMem.Size;
-   pSAREAPriv->log_tex_granularity = pI830->TexGranularity;
-
-   ctx->driverClientMsg = malloc(sizeof(I830DRIRec));
-   ctx->driverClientMsgSize = sizeof(I830DRIRec);
-   pI830DRI = (I830DRIPtr)ctx->driverClientMsg;
-   pI830DRI->deviceID = pI830->Chipset;
-   pI830DRI->regsSize = I830_REG_SIZE;
-   pI830DRI->width = ctx->shared.virtualWidth;
-   pI830DRI->height = ctx->shared.virtualHeight;
-   pI830DRI->mem = ctx->shared.fbSize;
-   pI830DRI->cpp = ctx->cpp;
-   pI830DRI->backOffset = pI830->BackBuffer.Start;
-   pI830DRI->backPitch = pI830->BackBuffer.Pitch; 
-
-   pI830DRI->depthOffset = pI830->DepthBuffer.Start;
-   pI830DRI->depthPitch = pI830->DepthBuffer.Pitch; 
-
-   pI830DRI->fbOffset = pI830->FrontBuffer.Start;
-   pI830DRI->fbStride = pI830->FrontBuffer.Pitch;
-
-   pI830DRI->bitsPerPixel = ctx->bpp;
-   pI830DRI->sarea_priv_offset = sizeof(drm_sarea_t);
-   
-   err = I830DRIDoMappings(ctx, pI830, pSAREAPriv);
-   if (err == FALSE)
-       return FALSE;
-
-   I830SetupMemoryTiling(ctx, pI830);
-
-   /* Quick hack to clear the front & back buffers.  Could also use
-    * the clear ioctl to do this, but would need to setup hw state
-    * first.
-    */
-   I830ClearScreen(ctx, pI830, pSAREAPriv);
-
-   I830SetRingRegs(ctx, pI830);
-
-   return TRUE;
-}
-
-
-/**
- * \brief Validate the fbdev mode.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Saves some registers and returns 1.
- *
- * \sa radeonValidateMode().
- */
-static int i830ValidateMode( const DRIDriverContext *ctx )
-{
-  return 1;
-}
-
-/**
- * \brief Examine mode returned by fbdev.
- * 
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Restores registers that fbdev has clobbered and returns 1.
- *
- * \sa i810ValidateMode().
- */
-static int i830PostValidateMode( const DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = ctx->driverPrivate;
-
-  I830SetRingRegs(ctx, pI830);
-  return 1;
-}
-
-
-/**
- * \brief Initialize the framebuffer device mode
- *
- * \param ctx display handle.
- *
- * \return one on success, or zero on failure.
- *
- * Fills in \p info with some default values and some information from \p ctx
- * and then calls I810ScreenInit() for the screen initialization.
- * 
- * Before exiting clears the framebuffer memory accessing it directly.
- */
-static int i830InitFBDev( DRIDriverContext *ctx )
-{
-  I830Rec *pI830 = calloc(1, sizeof(I830Rec));
-  int i;
-
-   {
-      int  dummy = ctx->shared.virtualWidth;
-
-      switch (ctx->bpp / 8) {
-      case 1: dummy = (ctx->shared.virtualWidth + 127) & ~127; break;
-      case 2: dummy = (ctx->shared.virtualWidth +  31) &  ~31; break;
-      case 3:
-      case 4: dummy = (ctx->shared.virtualWidth +  15) &  ~15; break;
-      }
-
-      ctx->shared.virtualWidth = dummy;
-      ctx->shared.Width = ctx->shared.virtualWidth;
-   }
-
-
-   for (i = 0; pitches[i] != 0; i++) {
-     if (pitches[i] >= ctx->shared.virtualWidth) {
-       ctx->shared.virtualWidth = pitches[i];
-       break;
-     }
-   }
-
-   ctx->driverPrivate = (void *)pI830;
-   
-   pI830->LpRing = calloc(1, sizeof(I830RingBuffer));
-   pI830->Chipset = ctx->chipset;
-   pI830->LinearAddr = ctx->FBStart;
-
-   if (!I830ScreenInit( ctx, pI830 ))
-      return 0;
-
-   
-   return 1;
-}
-
-
-/**
- * \brief The screen is being closed, so clean up any state and free any
- * resources used by the DRI.
- *
- * \param ctx display handle.
- *
- * Unmaps the SAREA, closes the DRM device file descriptor and frees the driver
- * private data.
- */
-static void i830HaltFBDev( DRIDriverContext *ctx )
-{
-  drmI830Sarea *pSAREAPriv;
-  I830Rec *pI830 = ctx->driverPrivate;
-
-   if (pI830->irq) {
-       drmCtlUninstHandler(ctx->drmFD);
-       pI830->irq = 0;   }
-
-   I830CleanupDma(ctx);
-
-  pSAREAPriv = (drmI830Sarea *)(((char*)ctx->pSAREA) + 
-				sizeof(drm_sarea_t));
-
-  I830DRIUnmapScreenRegions(ctx, pI830, pSAREAPriv);
-  drmUnmap( ctx->pSAREA, ctx->shared.SAREASize );
-  drmClose(ctx->drmFD);
-  
-  if (ctx->driverPrivate) {
-    free(ctx->driverPrivate);
-    ctx->driverPrivate = 0;
-  }
-}
-
-
-extern void i810NotifyFocus( int );
-
-/**
- * \brief Exported driver interface for Mini GLX.
- *
- * \sa DRIDriverRec.
- */
-const struct DRIDriverRec __driDriver = {
-   i830ValidateMode,
-   i830PostValidateMode,
-   i830InitFBDev,
-   i830HaltFBDev,
-   NULL,//I830EngineShutdown,
-   NULL, //I830EngineRestore,  
-#ifndef _EMBEDDED
-   0,
-#else
-   i810NotifyFocus, 
-#endif
-};
diff --git a/src/mesa/drivers/dri/intel/intel_chipset.h b/src/mesa/drivers/dri/intel/intel_chipset.h
new file mode 100644
index 00000000000..a18ca5be08a
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_chipset.h
@@ -0,0 +1,78 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
+ *
+ */
+
+#define PCI_CHIP_I810			0x7121
+#define PCI_CHIP_I810_DC100		0x7123
+#define PCI_CHIP_I810_E			0x7125
+#define PCI_CHIP_I815			0x1132
+
+#define PCI_CHIP_I830_M			0x3577
+#define PCI_CHIP_845_G			0x2562
+#define PCI_CHIP_I855_GM		0x3582
+#define PCI_CHIP_I865_G			0x2572
+
+#define PCI_CHIP_I915_G			0x2582
+#define PCI_CHIP_I915_GM		0x2592
+#define PCI_CHIP_I945_G			0x2772
+#define PCI_CHIP_I945_GM		0x27A2
+#define PCI_CHIP_I945_GME		0x27AE
+
+#define PCI_CHIP_Q35_G			0x29B2
+#define PCI_CHIP_G33_G			0x29C2
+#define PCI_CHIP_Q33_G			0x29D2
+
+#define PCI_CHIP_I965_G			0x29A2
+#define PCI_CHIP_I965_Q			0x2992
+#define PCI_CHIP_I965_G_1		0x2982
+#define PCI_CHIP_I946_GZ		0x2972
+#define PCI_CHIP_I965_GM                0x2A02
+#define PCI_CHIP_I965_GME               0x2A12
+
+#define IS_MOBILE(devid)	(devid == PCI_CHIP_I855_GM || \
+				 devid == PCI_CHIP_I915_GM || \
+				 devid == PCI_CHIP_I945_GM || \
+				 devid == PCI_CHIP_I945_GME || \
+				 devid == PCI_CHIP_I965_GM || \
+				 devid == PCI_CHIP_I965_GME)
+
+#define IS_965(devid)		(devid = PCI_CHIP_I965_G || \
+				 devid == PCI_CHIP_I965_Q || \
+				 devid == PCI_CHIP_I965_G_1 || \
+				 devid == PCI_CHIP_I965_GM || \
+				 devid == PCI_CHIP_I965_GME || \
+				 devid == PCI_CHIP_I946_GZ)
+
+#define IS_9XX(devid)		(devid == PCI_CHIP_I915G || \
+				 devid == PCI_CHIP_I915GM || \
+				 devid == PCI_CHIP_I945G || \
+				 devid == PCI_CHIP_I945GM || \
+				 devid == PCI_CHIP_I945GME || \
+				 devid == PCI_CHIP_G33_G || \
+				 devid == PCI_CHIP_Q35_G || \
+				 devid == PCI_CHIP_Q33_G || \
+				 IS_965(devid))
+
diff --git a/src/mesa/drivers/dri/intel/intel_decode.c b/src/mesa/drivers/dri/intel/intel_decode.c
new file mode 100644
index 00000000000..16881f297d3
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_decode.c
@@ -0,0 +1,900 @@
+/* -*- c-basic-offset: 4 -*- */
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
+ *
+ */
+
+/** @file intel_decode.c
+ * This file contains code to print out batchbuffer contents in a
+ * human-readable format.
+ *
+ * The current version only supports i915 packets, and only pretty-prints a
+ * subset of them.  The intention is for it to make just a best attempt to
+ * decode, but never crash in the process.
+ */
+
+#include <stdio.h>
+#include <stdarg.h>
+#include <inttypes.h>
+
+#include "intel_decode.h"
+#include "intel_chipset.h"
+
+#define BUFFER_FAIL(_count, _len, _name) do {			\
+    fprintf(out, "Buffer size too small in %s (%d < %d)\n",	\
+	    (_name), (_count), (_len));				\
+    (*failures)++;						\
+    return count;						\
+} while (0)
+
+static FILE *out;
+static uint32_t saved_s2 = 0, saved_s4 = 0;
+static char saved_s2_set = 0, saved_s4_set = 0;
+
+static float
+int_as_float(uint32_t intval)
+{
+    union intfloat {
+	uint32_t i;
+	float f;
+    } uval;
+
+    uval.i = intval;
+    return uval.f;
+}
+
+static void
+instr_out(uint32_t *data, uint32_t hw_offset, unsigned int index,
+	  char *fmt, ...)
+{
+    va_list va;
+
+    fprintf(out, "0x%08x: 0x%08x: ", hw_offset + index * 4, data[index]);
+    va_start(va, fmt);
+    vfprintf(out, fmt, va);
+    va_end(va);
+}
+
+
+static int
+decode_mi(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_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" },
+    };
+
+
+    for (opcode = 0; opcode < sizeof(opcodes_mi) / sizeof(opcodes_mi[0]);
+	 opcode++) {
+	if ((data[0] & 0x1e000000) >> 23 == opcodes_mi[opcode].opcode) {
+	    unsigned int len = 1, i;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_mi[opcode].name);
+	    if (opcodes_mi[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 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);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_mi[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "MI UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_2d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int opcode, len;
+    char *format = NULL;
+
+    struct {
+	uint32_t opcode;
+	int min_len;
+	int max_len;
+	char *name;
+    } opcodes_2d[] = {
+	{ 0x40, 5, 5, "COLOR_BLT" },
+	{ 0x43, 6, 6, "SRC_COPY_BLT" },
+	{ 0x01, 8, 8, "XY_SETUP_BLT" },
+	{ 0x11, 9, 9, "XY_SETUP_MONO_PATTERN_SL_BLT" },
+	{ 0x03, 3, 3, "XY_SETUP_CLIP_BLT" },
+	{ 0x24, 2, 2, "XY_PIXEL_BLT" },
+	{ 0x25, 3, 3, "XY_SCANLINES_BLT" },
+	{ 0x26, 4, 4, "Y_TEXT_BLT" },
+	{ 0x31, 5, 134, "XY_TEXT_IMMEDIATE_BLT" },
+	{ 0x50, 6, 6, "XY_COLOR_BLT" },
+	{ 0x51, 6, 6, "XY_PAT_BLT" },
+	{ 0x76, 8, 8, "XY_PAT_CHROMA_BLT" },
+	{ 0x72, 7, 135, "XY_PAT_BLT_IMMEDIATE" },
+	{ 0x77, 9, 137, "XY_PAT_CHROMA_BLT_IMMEDIATE" },
+	{ 0x52, 9, 9, "XY_MONO_PAT_BLT" },
+	{ 0x59, 7, 7, "XY_MONO_PAT_FIXED_BLT" },
+	{ 0x53, 8, 8, "XY_SRC_COPY_BLT" },
+	{ 0x54, 8, 8, "XY_MONO_SRC_COPY_BLT" },
+	{ 0x71, 9, 137, "XY_MONO_SRC_COPY_IMMEDIATE_BLT" },
+	{ 0x55, 9, 9, "XY_FULL_BLT" },
+	{ 0x55, 9, 137, "XY_FULL_IMMEDIATE_PATTERN_BLT" },
+	{ 0x56, 9, 9, "XY_FULL_MONO_SRC_BLT" },
+	{ 0x75, 10, 138, "XY_FULL_MONO_SRC_IMMEDIATE_PATTERN_BLT" },
+	{ 0x57, 12, 12, "XY_FULL_MONO_PATTERN_BLT" },
+	{ 0x58, 12, 12, "XY_FULL_MONO_PATTERN_MONO_SRC_BLT" },
+    };
+
+    switch ((data[0] & 0x1fc00000) >> 22) {
+    case 0x50:
+	instr_out(data, hw_offset, 0,
+		  "XY_COLOR_BLT (rgb %sabled, alpha %sabled)\n",
+		  (data[0] & (1 << 20)) ? "en" : "dis",
+		  (data[0] & (1 << 21)) ? "en" : "dis");
+
+	len = (data[0] & 0x000000ff) + 2;
+	if (len != 6)
+	    fprintf(out, "Bad count in XY_COLOR_BLT\n");
+	if (count < 6)
+	    BUFFER_FAIL(count, len, "XY_COLOR_BLT");
+
+	switch ((data[1] >> 24) & 0x3) {
+	case 0:
+	    format="8";
+	    break;
+	case 1:
+	    format="565";
+	    break;
+	case 2:
+	    format="1555";
+	    break;
+	case 3:
+	    format="8888";
+	    break;
+	}
+
+	instr_out(data, hw_offset, 1, "format %s, pitch %d, "
+		  "clipping %sabled\n", format,
+		  data[1] & 0xffff, data[1] & (1 << 30) ? "en" : "dis");
+	instr_out(data, hw_offset, 2, "(%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 3, "(%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 4, "offset 0x%08x\n", data[4]);
+	instr_out(data, hw_offset, 5, "color\n");
+	return len;
+    case 0x53:
+	instr_out(data, hw_offset, 0,
+		  "XY_SRC_COPY_BLT (rgb %sabled, alpha %sabled)\n",
+		  (data[0] & (1 << 20)) ? "en" : "dis",
+		  (data[0] & (1 << 21)) ? "en" : "dis");
+
+	len = (data[0] & 0x000000ff) + 2;
+	if (len != 8)
+	    fprintf(out, "Bad count in XY_SRC_COPY_BLT\n");
+	if (count < 8)
+	    BUFFER_FAIL(count, len, "XY_SRC_COPY_BLT");
+
+	switch ((data[1] >> 24) & 0x3) {
+	case 0:
+	    format="8";
+	    break;
+	case 1:
+	    format="565";
+	    break;
+	case 2:
+	    format="1555";
+	    break;
+	case 3:
+	    format="8888";
+	    break;
+	}
+
+	instr_out(data, hw_offset, 1, "format %s, dst pitch %d, "
+		  "clipping %sabled\n", format,
+		  data[1] & 0xffff, data[1] & (1 << 30) ? "en" : "dis");
+	instr_out(data, hw_offset, 2, "dst (%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 3, "dst (%d,%d)\n",
+		  data[2] & 0xffff, data[2] >> 16);
+	instr_out(data, hw_offset, 4, "dst offset 0x%08x\n", data[4]);
+	instr_out(data, hw_offset, 5, "src (%d,%d)\n",
+		  data[5] & 0xffff, data[5] >> 16);
+	instr_out(data, hw_offset, 6, "src pitch %d\n",
+		  data[6] & 0xffff);
+	instr_out(data, hw_offset, 7, "src offset 0x%08x\n", data[7]);
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_2d) / sizeof(opcodes_2d[0]);
+	 opcode++) {
+	if ((data[0] & 0x1fc00000) >> 22 == opcodes_2d[opcode].opcode) {
+	    unsigned int i;
+
+	    len = 1;
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_2d[opcode].name);
+	    if (opcodes_2d[opcode].max_len > 1) {
+		len = (data[0] & 0x000000ff) + 2;
+		if (len < opcodes_2d[opcode].min_len ||
+		    len > opcodes_2d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n", opcodes_2d[opcode].name);
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, opcodes_2d[opcode].name);
+		instr_out(data, hw_offset, i, "dword %d\n", i);
+	    }
+
+	    return len;
+	}
+    }
+
+    instr_out(data, hw_offset, 0, "2D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1c(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    switch ((data[0] & 0x00f80000) >> 19) {
+    case 0x11:
+	instr_out(data, hw_offset, 0, "3DSTATE_DEPTH_SUBRECTANGLE_DISALBE\n");
+	return 1;
+    case 0x10:
+	instr_out(data, hw_offset, 0, "3DSTATE_SCISSOR_ENABLE\n");
+	return 1;
+    }
+
+    instr_out(data, hw_offset, 0, "3D UNKNOWN\n");
+    (*failures)++;
+    return 1;
+}
+
+static int
+decode_3d_1d(uint32_t *data, int count, uint32_t hw_offset, int *failures)
+{
+    unsigned int len, i, c, opcode, word, map, sampler, instr;
+
+    struct {
+	uint32_t opcode;
+	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" },
+    };
+
+    switch ((data[0] & 0x00ff0000) >> 16) {
+    case 0x07:
+	/* This instruction is unusual.  A 0 length means just 1 DWORD instead of
+	 * 2.  The 0 length is specified in one place to be unsupported, but
+	 * stated to be required in another, and 0 length LOAD_INDIRECTs appear
+	 * to cause no harm at least.
+	 */
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_INDIRECT\n");
+	len = (data[0] & 0x000000ff) + 1;
+	i = 1;
+	if (data[0] & (0x01 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SIS.0\n");
+	    instr_out(data, hw_offset, i++, "SIS.1\n");
+	}
+	if (data[0] & (0x02 << 8)) {
+	    if (i + 1 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "DIS.0\n");
+	}
+	if (data[0] & (0x04 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "SSB.0\n");
+	    instr_out(data, hw_offset, i++, "SSB.1\n");
+	}
+	if (data[0] & (0x08 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "MSB.0\n");
+	    instr_out(data, hw_offset, i++, "MSB.1\n");
+	}
+	if (data[0] & (0x10 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSP.0\n");
+	    instr_out(data, hw_offset, i++, "PSP.1\n");
+	}
+	if (data[0] & (0x20 << 8)) {
+	    if (i + 2 >= count)
+		BUFFER_FAIL(count, len, "3DSTATE_LOAD_INDIRECT");
+	    instr_out(data, hw_offset, i++, "PSC.0\n");
+	    instr_out(data, hw_offset, i++, "PSC.1\n");
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x04:
+	instr_out(data, hw_offset, 0, "3DSTATE_LOAD_STATE_IMMEDIATE_1\n");
+	len = (data[0] & 0x0000000f) + 2;
+	i = 1;
+	for (word = 0; word <= 7; word++) {
+	    if (data[0] & (1 << (4 + word))) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_LOAD_STATE_IMMEDIATE_1");
+
+		/* save vertex state for decode */
+		if (word == 2) {
+		    saved_s2_set = 1;
+		    saved_s2 = data[i];
+		}
+		if (word == 4) {
+		    saved_s4_set = 1;
+		    saved_s4 = data[i];
+		}
+
+		instr_out(data, hw_offset, i++, "S%d\n", word);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_LOAD_INDIRECT\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x00:
+	instr_out(data, hw_offset, 0, "3DSTATE_MAP_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+
+	i = 1;
+	for (map = 0; map <= 15; map++) {
+	    if (data[1] & (1 << map)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_MAP_STATE");
+		instr_out(data, hw_offset, i++, "map %d MS2\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS3\n", map);
+		instr_out(data, hw_offset, i++, "map %d MS4\n", map);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	    return len;
+	}
+	return len;
+    case 0x06:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_CONSTANTS\n");
+	len = (data[0] & 0x000000ff) + 2;
+
+	i = 1;
+	for (c = 0; c <= 31; c++) {
+	    if (data[1] & (1 << c)) {
+		if (i + 4 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_PIXEL_SHADER_CONSTANTS");
+		instr_out(data, hw_offset, i, "C%d.X = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Y = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.Z = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+		instr_out(data, hw_offset, i, "C%d.W = %f\n",
+			  c, int_as_float(data[i]));
+		i++;
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_MAP_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    case 0x05:
+	instr_out(data, hw_offset, 0, "3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	len = (data[0] & 0x000000ff) + 2;
+	if ((len - 1) % 3 != 0 || len > 370) {
+	    fprintf(out, "Bad count in 3DSTATE_PIXEL_SHADER_PROGRAM\n");
+	    (*failures)++;
+	}
+	i = 1;
+	for (instr = 0; instr < (len - 1) / 3; instr++) {
+	    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);
+	}
+	return len;
+    case 0x01:
+	instr_out(data, hw_offset, 0, "3DSTATE_SAMPLER_STATE\n");
+	len = (data[0] & 0x0000003f) + 2;
+	i = 1;
+	for (sampler = 0; sampler <= 15; sampler++) {
+	    if (data[1] & (1 << sampler)) {
+		if (i + 3 >= count)
+		    BUFFER_FAIL(count, len, "3DSTATE_SAMPLER_STATE");
+		instr_out(data, hw_offset, i++, "sampler %d SS2\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS3\n",
+			  sampler);
+		instr_out(data, hw_offset, i++, "sampler %d SS4\n",
+			  sampler);
+	    }
+	}
+	if (len != i) {
+	    fprintf(out, "Bad count in 3DSTATE_SAMPLER_STATE\n");
+	    (*failures)++;
+	}
+	return len;
+    }
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d_1d) / sizeof(opcodes_3d_1d[0]);
+	 opcode++)
+    {
+	if (((data[0] & 0x00ff0000) >> 16) == opcodes_3d_1d[opcode].opcode) {
+	    len = 1;
+
+	    instr_out(data, hw_offset, 0, "%s\n", opcodes_3d_1d[opcode].name);
+	    if (opcodes_3d_1d[opcode].max_len > 1) {
+		len = (data[0] & 0x0000ffff) + 2;
+		if (len < opcodes_3d_1d[opcode].min_len ||
+		    len > opcodes_3d_1d[opcode].max_len)
+		{
+		    fprintf(out, "Bad count in %s\n",
+			    opcodes_3d_1d[opcode].name);
+		    (*failures)++;
+		}
+	    }
+
+	    for (i = 1; i < len; i++) {
+		if (i >= count)
+		    BUFFER_FAIL(count, len,  opcodes_3d_1d[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;
+}
+
+static int
+decode_3d_primitive(uint32_t *data, int count, uint32_t hw_offset,
+		    int *failures)
+{
+    char immediate = (data[0] & (1 << 23)) == 0;
+    unsigned int len, i;
+    char *primtype;
+
+    switch ((data[0] >> 18) & 0xf) {
+    case 0x0: primtype = "TRILIST"; break;
+    case 0x1: primtype = "TRISTRIP"; break;
+    case 0x2: primtype = "TRISTRIP_REVERSE"; break;
+    case 0x3: primtype = "TRIFAN"; break;
+    case 0x4: primtype = "POLYGON"; break;
+    case 0x5: primtype = "LINELIST"; break;
+    case 0x6: primtype = "LINESTRIP"; break;
+    case 0x7: primtype = "RECTLIST"; break;
+    case 0x8: primtype = "POINTLIST"; break;
+    case 0x9: primtype = "DIB"; break;
+    case 0xa: primtype = "CLEAR_RECT"; break;
+    default: primtype = "unknown"; break;
+    }
+
+    /* XXX: 3DPRIM_DIB not supported */
+    if (immediate) {
+	len = (data[0] & 0x0003ffff) + 2;
+	instr_out(data, hw_offset, 0, "3DPRIMITIVE inline %s\n", primtype);
+	if (count < len)
+	    BUFFER_FAIL(count, len,  "3DPRIMITIVE inline");
+	if (!saved_s2_set || !saved_s4_set) {
+	    fprintf(out, "unknown vertex format\n");
+	    for (i = 1; i < len; i++) {
+		instr_out(data, hw_offset, i,
+			  "           vertex data (%f float)\n",
+			  int_as_float(data[i]));
+	    }
+	} else {
+	    unsigned int vertex = 0;
+	    for (i = 1; i < len;) {
+		unsigned int tc;
+
+#define VERTEX_OUT(fmt, ...) do {					\
+    if (i < len)							\
+	instr_out(data, hw_offset, i, " V%d."fmt"\n", vertex, __VA_ARGS__); \
+    else								\
+	fprintf(out, " missing data in V%d\n", vertex);			\
+    i++;								\
+} while (0)
+
+		VERTEX_OUT("X = %f", int_as_float(data[i]));
+		VERTEX_OUT("Y = %f", int_as_float(data[i]));
+	        switch (saved_s4 >> 6 & 0x7) {
+		case 0x1:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    break;
+		case 0x2:
+		    VERTEX_OUT("Z = %f", int_as_float(data[i]));
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		case 0x3:
+		    break;
+		case 0x4:
+		    VERTEX_OUT("W = %f", int_as_float(data[i]));
+		    break;
+		default:
+		    fprintf(out, "bad S4 position mask\n");
+		}
+
+		if (saved_s4 & (1 << 10)) {
+		    VERTEX_OUT("color = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 11)) {
+		    VERTEX_OUT("spec = (A=0x%02x, R=0x%02x, G=0x%02x, "
+			       "B=0x%02x)",
+			       data[i] >> 24,
+			       (data[i] >> 16) & 0xff,
+			       (data[i] >> 8) & 0xff,
+			       data[i] & 0xff);
+		}
+		if (saved_s4 & (1 << 12))
+		    VERTEX_OUT("width = 0x%08x)", data[i]);
+
+		for (tc = 0; tc <= 7; tc++) {
+		    switch ((saved_s2 >> (tc * 4)) & 0xf) {
+		    case 0x0:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x1:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x2:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Y = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.Z = %f", tc, int_as_float(data[i]));
+			VERTEX_OUT("T%d.W = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x3:
+			VERTEX_OUT("T%d.X = %f", tc, int_as_float(data[i]));
+			break;
+		    case 0x4:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0x5:
+			VERTEX_OUT("T%d.XY = 0x%08x half-float", tc, data[i]);
+			VERTEX_OUT("T%d.ZW = 0x%08x half-float", tc, data[i]);
+			break;
+		    case 0xf:
+			break;
+		    default:
+			fprintf(out, "bad S2.T%d format\n", tc);
+		    }
+		}
+		vertex++;
+	    }
+	}
+    } else {
+	/* indirect vertices */
+	len = data[0] & 0x0000ffff; /* index count */
+	if (data[0] & (1 << 17)) {
+	    /* random vertex access */
+	    if (count < (len + 1) / 2 + 1) {
+		BUFFER_FAIL(count, (len + 1) / 2 + 1,
+			    "3DPRIMITIVE random indirect");
+	    }
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE random indirect %s (%d)\n", primtype, len);
+	    if (len == 0) {
+		/* vertex indices continue until 0xffff is found */
+		for (i = 1; i < count; i++) {
+		    if ((data[i] & 0xffff) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: (terminator)\n");
+			return i;
+		    } else if ((data[i] >> 16) == 0xffff) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, "
+				  "(terminator)\n",
+				  data[i] & 0xffff);
+			return i;
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+		fprintf(out,
+			"3DPRIMITIVE: no terminator found in index buffer\n");
+		(*failures)++;
+		return count;
+	    } else {
+		/* fixed size vertex index buffer */
+		for (i = 0; i < len; i += 2) {
+		    if (i * 2 == len - 1) {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x\n",
+				  data[i] & 0xffff);
+		    } else {
+			instr_out(data, hw_offset, i,
+				  "            indices: 0x%04x, 0x%04x\n",
+				  data[i] & 0xffff, data[i] >> 16);
+		    }
+		}
+	    }
+	    return (len + 1) / 2 + 1;
+	} else {
+	    /* sequential vertex access */
+	    if (count < 2)
+		BUFFER_FAIL(count, 2, "3DPRIMITIVE seq indirect");
+	    instr_out(data, hw_offset, 0,
+		      "3DPRIMITIVE sequential indirect %s, %d starting from "
+		      "%d\n", primtype, len, data[1] & 0xffff);
+	    instr_out(data, hw_offset, 1, "           start\n");
+	    return 2;
+	}
+    }
+
+    return len;
+}
+
+static int
+decode_3d(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[] = {
+	{ 0x06, 1, 1, "3DSTATE_ANTI_ALIASING" },
+	{ 0x08, 1, 1, "3DSTATE_BACKFACE_STENCIL_OPS" },
+	{ 0x09, 1, 1, "3DSTATE_BACKFACE_STENCIL_MASKS" },
+	{ 0x16, 1, 1, "3DSTATE_COORD_SET_BINDINGS" },
+	{ 0x15, 1, 1, "3DSTATE_FOG_COLOR" },
+	{ 0x0b, 1, 1, "3DSTATE_INDEPENDENT_ALPHA_BLEND" },
+	{ 0x0d, 1, 1, "3DSTATE_MODES_4" },
+	{ 0x0c, 1, 1, "3DSTATE_MODES_5" },
+	{ 0x07, 1, 1, "3DSTATE_RASTERIZATION_RULES" },
+    };
+
+    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);
+    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;
+}
+
+static int
+decode_3d_965(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[] = {
+	{ 0x6000, 3, 3, "URB_FENCE" },
+	{ 0x6001, 2, 2, "CS_URB_STATE" },
+	{ 0x6002, 2, 2, "CONSTANT_BUFFER" },
+	{ 0x6101, 6, 6, "STATE_BASE_ADDRESS" },
+	{ 0x6102, 2, 2 , "STATE_SIP" },
+	{ 0x6104, 1, 1, "3DSTATE_PIPELINE_SELECT" },
+	{ 0x7800, 7, 7, "3DSTATE_PIPELINED_POINTERS" },
+	{ 0x7801, 6, 6, "3DSTATE_BINDING_TABLE_POINTERS" },
+	{ 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 */
+	{ 0x7900, 4, 4, "3DSTATE_DRAWING_RECTANGLE" },
+	{ 0x7901, 5, 5, "3DSTATE_CONSTANT_COLOR" },
+	{ 0x7905, 5, 5, "3DSTATE_DEPTH_BUFFER" },
+	{ 0x7906, 2, 2, "3DSTATE_POLY_STIPPLE_OFFSET" },
+	{ 0x7907, 33, 33, "3DSTATE_POLY_STIPPLE_PATTERN" },
+	{ 0x7909, 2, 2, "3DSTATE_GLOBAL_DEPTH_OFFSET_CLAMP" },
+	{ 0x7908, 3, 3, "3DSTATE_LINE_STIPPLE" },
+	{ 0x7b00, 6, 6, "3DPRIMITIVE" },
+    };
+
+    for (opcode = 0; opcode < sizeof(opcodes_3d) / sizeof(opcodes_3d[0]);
+	 opcode++) {
+	if ((data[0] & 0xffff0000) >> 16 == 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.
+ *
+ * \param data batch buffer contents
+ * \param count number of DWORDs to decode in the batch buffer
+ * \param hw_offset hardware address for the buffer
+ */
+int
+intel_decode(uint32_t *data, int count, uint32_t hw_offset, uint32_t devid)
+{
+    int index = 0;
+    int failures = 0;
+
+    out = stdout;
+
+    while (index < count) {
+	switch ((data[index] & 0xe0000000) >> 29) {
+	case 0x0:
+	    index += decode_mi(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x2:
+	    index += decode_2d(data + index, count - index,
+			       hw_offset + index * 4, &failures);
+	    break;
+	case 0x3:
+	    if (IS_965(devid)) {
+		index += decode_3d_965(data + index, count - index,
+				       hw_offset + index * 4, &failures);
+	    } else {
+		index += decode_3d(data + index, count - index,
+				   hw_offset + index * 4, &failures);
+	    }
+	    break;
+	default:
+	    instr_out(data, hw_offset, index, "UNKNOWN\n");
+	    failures++;
+	    index++;
+	    break;
+	}
+	fflush(out);
+    }
+
+    return failures;
+}
+
+void intel_decode_context_reset(void)
+{
+    saved_s2_set = 0;
+    saved_s4_set = 1;
+}
+
diff --git a/src/mesa/drivers/dri/intel/intel_decode.h b/src/mesa/drivers/dri/intel/intel_decode.h
new file mode 100644
index 00000000000..c50644a46b5
--- /dev/null
+++ b/src/mesa/drivers/dri/intel/intel_decode.h
@@ -0,0 +1,29 @@
+/*
+ * Copyright © 2007 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Eric Anholt <[email protected]>
+ *
+ */
+
+int intel_decode(uint32_t *data, int count, uint32_t hw_offset, uint32_t devid);
+void intel_decode_context_reset(void);
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.c b/src/mesa/drivers/dri/intel/intel_tex_layout.c
index fcb5cc39068..fdecd3e1868 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.c
@@ -40,6 +40,23 @@ static int align(int value, int alignment)
    return (value + alignment - 1) & ~(alignment - 1);
 }
 
+GLuint intel_compressed_alignment(GLenum internalFormat)
+{
+    GLuint alignment = 4;
+
+    switch (internalFormat) {
+    case GL_COMPRESSED_RGB_FXT1_3DFX:
+    case GL_COMPRESSED_RGBA_FXT1_3DFX:
+        alignment = 8;
+        break;
+
+    default:
+        break;
+    }
+
+    return alignment;
+}
+
 void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 {
    GLint align_h = 2, align_w = 4;
@@ -51,17 +68,30 @@ void i945_miptree_layout_2d( struct intel_mipmap_tree *mt )
 
    mt->pitch = mt->width0;
 
+   if (mt->compressed) {
+       align_w = intel_compressed_alignment(mt->internal_format);
+       mt->pitch = align(mt->width0, align_w);
+   }
+
    /* May need to adjust pitch to accomodate the placement of
     * the 2nd mipmap.  This occurs when the alignment
     * constraints of mipmap placement push the right edge of the
     * 2nd mipmap out past the width of its parent.
     */
    if (mt->first_level != mt->last_level) {
-      GLuint mip1_width = align(minify(mt->width0), align_w)
-			+ minify(minify(mt->width0));
+       GLuint mip1_width;
+
+       if (mt->compressed) {
+           mip1_width = align(minify(mt->width0), align_w)
+               + align(minify(minify(mt->width0)), align_w);
+       } else {
+           mip1_width = align(minify(mt->width0), align_w)
+               + minify(minify(mt->width0));
+       }
 
-      if (mip1_width > mt->width0)
-	 mt->pitch = mip1_width;
+       if (mip1_width > mt->pitch) {
+           mt->pitch = mip1_width;
+       }
    }
 
    /* Pitch must be a whole number of dwords, even though we
diff --git a/src/mesa/drivers/dri/intel/intel_tex_layout.h b/src/mesa/drivers/dri/intel/intel_tex_layout.h
index 1e37f8f525f..99d41c3629e 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_layout.h
+++ b/src/mesa/drivers/dri/intel/intel_tex_layout.h
@@ -39,3 +39,4 @@ static GLuint minify( GLuint d )
 }
 
 extern void i945_miptree_layout_2d( struct intel_mipmap_tree *mt );
+extern GLuint intel_compressed_alignment(GLenum);
diff --git a/src/mesa/drivers/dri/nouveau/Makefile b/src/mesa/drivers/dri/nouveau/Makefile
index 20d2de5eefb..6ea4594f1e6 100644
--- a/src/mesa/drivers/dri/nouveau/Makefile
+++ b/src/mesa/drivers/dri/nouveau/Makefile
@@ -9,12 +9,13 @@ MINIGLX_SOURCES =
 
 DRIVER_SOURCES = \
 	nouveau_bufferobj.c      \
-	nouveau_buffers.c        \
 	nouveau_card.c           \
 	nouveau_context.c        \
 	nouveau_driver.c         \
+	nouveau_fbo.c            \
 	nouveau_fifo.c           \
 	nouveau_lock.c           \
+	nouveau_mem.c            \
 	nouveau_object.c         \
 	nouveau_screen.c         \
 	nouveau_span.c           \
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
index fc14060c049..be6455a01ed 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
@@ -2,11 +2,11 @@
 #include "enums.h"
 
 #include "nouveau_bufferobj.h"
-#include "nouveau_buffers.h"
 #include "nouveau_context.h"
 #include "nouveau_drm.h"
-#include "nouveau_object.h"
+#include "nouveau_mem.h"
 #include "nouveau_msg.h"
+#include "nouveau_object.h"
 
 #define NOUVEAU_MEM_FREE(mem) do {      \
 	nouveau_mem_free(ctx, (mem));   \
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
index 3439a35e7c8..cbc89a151da 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
@@ -2,7 +2,7 @@
 #define __NOUVEAU_BUFFEROBJ_H__
 
 #include "mtypes.h"
-#include "nouveau_buffers.h"
+#include "nouveau_mem.h"
 
 #define NOUVEAU_BO_VRAM_OK (NOUVEAU_MEM_FB | NOUVEAU_MEM_FB_ACCEPTABLE)
 #define NOUVEAU_BO_GART_OK  (NOUVEAU_MEM_AGP | NOUVEAU_MEM_AGP_ACCEPTABLE)
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c b/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
deleted file mode 100644
index d498f616c90..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.c
+++ /dev/null
@@ -1,436 +0,0 @@
-#include "utils.h"
-#include "framebuffer.h"
-#include "renderbuffer.h"
-#include "fbobject.h"
-
-#include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
-#include "nouveau_fifo.h"
-#include "nouveau_reg.h"
-#include "nouveau_msg.h"
-
-#define MAX_MEMFMT_LENGTH 32768
-
-/* Unstrided blit using NV_MEMORY_TO_MEMORY_FORMAT */
-GLboolean
-nouveau_memformat_flat_emit(GLcontext *ctx,
-      			    nouveau_mem *dst, nouveau_mem *src,
-			    GLuint dst_offset, GLuint src_offset,
-			    GLuint size)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   uint32_t src_handle, dst_handle;
-   GLuint count;
-
-   if (src_offset + size > src->size) {
-      MESSAGE("src out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-   if (dst_offset + size > dst->size) {
-      MESSAGE("dst out of nouveau_mem bounds\n");
-      return GL_FALSE;
-   }
-
-   src_handle = (src->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   dst_handle = (dst->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
-   src_offset += nouveau_mem_gpu_offset_get(ctx, src);
-   dst_offset += nouveau_mem_gpu_offset_get(ctx, dst);
-
-   BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN, 2);
-   OUT_RING       (src_handle);
-   OUT_RING       (dst_handle);
-
-   count = (size / MAX_MEMFMT_LENGTH) + ((size % MAX_MEMFMT_LENGTH) ? 1 : 0);
-
-   while (count--) {
-      GLuint length = (size > MAX_MEMFMT_LENGTH) ? MAX_MEMFMT_LENGTH : size;
-
-      BEGIN_RING_SIZE(NvSubMemFormat, NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
-      OUT_RING       (src_offset);
-      OUT_RING       (dst_offset);
-      OUT_RING       (0); /* pitch in */
-      OUT_RING       (0); /* pitch out */
-      OUT_RING       (length); /* line length */
-      OUT_RING       (1); /* number of lines */
-      OUT_RING       ((1 << 8) /* dst_inc */ | (1 << 0) /* src_inc */);
-      OUT_RING       (0); /* buffer notify? */
-      FIRE_RING();
-
-      src_offset += length;
-      dst_offset += length;
-      size       -= length;
-   }
-
-   return GL_TRUE;
-}
-
-void
-nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   struct drm_nouveau_mem_free memf;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (mem->map)
-      drmUnmap(mem->map, mem->size);
-   memf.flags  = mem->type;
-   memf.offset = mem->offset;
-   drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf, sizeof(memf));
-   FREE(mem);
-}
-
-nouveau_mem *
-nouveau_mem_alloc(GLcontext *ctx, int type, GLuint size, GLuint align)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   struct drm_nouveau_mem_alloc mema;
-   nouveau_mem *mem;
-   int ret;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: requested: type=0x%x, size=0x%x, align=0x%x\n",
-	    __func__, type, (GLuint)size, align);
-   }
-
-   mem = CALLOC(sizeof(nouveau_mem));
-   if (!mem)
-      return NULL;
-
-   mema.flags     = type;
-   mema.size      = mem->size = size;
-   mema.alignment = align;
-   mem->map       = NULL;
-   ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_MEM_ALLOC,
-	 		     &mema, sizeof(mema));
-   if (ret) {
-      FREE(mem);
-      return NULL;
-   }
-   mem->offset = mema.offset;
-   mem->type   = mema.flags;
-
-   if (NOUVEAU_DEBUG & DEBUG_MEM)  {
-      fprintf(stderr, "%s: actual: type=0x%x, offset=0x%x, size=0x%x\n",
-	    __func__, mem->type, (GLuint)mem->offset, (GLuint)mem->size);
-   }
-
-   if (type & NOUVEAU_MEM_MAPPED)
-      ret = drmMap(nmesa->driFd, mema.map_handle, mem->size, &mem->map);
-   if (ret) {
-      mem->map = NULL;
-      nouveau_mem_free(ctx, mem);
-      mem = NULL;
-   }
-
-   return mem;
-}
-
-uint32_t
-nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-
-   return mem->offset;
-}
-
-static GLboolean
-nouveau_renderbuffer_pixelformat(nouveau_renderbuffer *nrb,
-      				 GLenum internalFormat)
-{
-      nrb->mesa.InternalFormat	= internalFormat;
-
-   /*TODO: We probably want to extend this a bit, and maybe make
-    *      card-specific? 
-    */
-      switch (internalFormat) {
-      case GL_RGBA:
-      case GL_RGBA8:
-	 nrb->mesa._BaseFormat	= GL_RGBA;
-	 nrb->mesa._ActualFormat= GL_RGBA8;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 8;
-	 nrb->mesa.GreenBits	= 8;
-	 nrb->mesa.BlueBits	= 8;
-	 nrb->mesa.AlphaBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_RGB:
-      case GL_RGB5:
-	 nrb->mesa._BaseFormat	= GL_RGB;
-	 nrb->mesa._ActualFormat= GL_RGB5;
-	 nrb->mesa.DataType	= GL_UNSIGNED_BYTE;
-	 nrb->mesa.RedBits	= 5;
-	 nrb->mesa.GreenBits	= 6;
-	 nrb->mesa.BlueBits	= 5;
-	 nrb->mesa.AlphaBits	= 0;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT16:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH_COMPONENT16;
-	 nrb->mesa.DataType	= GL_UNSIGNED_SHORT;
-	 nrb->mesa.DepthBits	= 16;
-	 nrb->cpp		= 2;
-	 break;
-      case GL_DEPTH_COMPONENT24:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_COMPONENT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_STENCIL_INDEX8_EXT:
-	 nrb->mesa._BaseFormat	= GL_STENCIL_INDEX;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.StencilBits	= 8;
-	 nrb->cpp		= 4;
-	 break;
-      case GL_DEPTH24_STENCIL8_EXT:
-	 nrb->mesa._BaseFormat	= GL_DEPTH_STENCIL_EXT;
-	 nrb->mesa._ActualFormat= GL_DEPTH24_STENCIL8_EXT;
-	 nrb->mesa.DataType	= GL_UNSIGNED_INT_24_8_EXT;
-	 nrb->mesa.DepthBits	= 24;
-	 nrb->mesa.StencilBits  = 8;
-	 nrb->cpp		= 4;
-	 break;
-      default:
-	 return GL_FALSE;
-	 break;
-      }
-
-      return GL_TRUE;
-}
-
-static GLboolean
-nouveau_renderbuffer_storage(GLcontext *ctx, struct gl_renderbuffer *rb,
-      			     GLenum internalFormat,
-			     GLuint width,
-			     GLuint height)
-{
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
-      fprintf(stderr, "%s: unknown internalFormat\n", __func__);
-      return GL_FALSE;
-   }
-
-   /* If this buffer isn't statically alloc'd, we may need to ask the
-    * drm for more memory */
-   if (!nrb->dPriv && (rb->Width != width || rb->Height != height)) {
-      GLuint pitch;
-
-      /* align pitches to 64 bytes */
-      pitch = ((width * nrb->cpp) + 63) & ~63;
-
-      if (nrb->mem)
-	 nouveau_mem_free(ctx, nrb->mem);
-      nrb->mem = nouveau_mem_alloc(ctx,
-	    			   NOUVEAU_MEM_FB | NOUVEAU_MEM_MAPPED,
-				   pitch*height,
-				   0);
-      if (!nrb->mem)
-	 return GL_FALSE;
-
-      /* update nouveau_renderbuffer info */
-      nrb->offset = nouveau_mem_gpu_offset_get(ctx, nrb->mem);
-      nrb->pitch  = pitch;
-   }
-
-   rb->Width = width;
-   rb->Height = height;
-   rb->InternalFormat = internalFormat;
-   return GL_TRUE;
-}
-
-static void
-nouveau_renderbuffer_delete(struct gl_renderbuffer *rb)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer*)rb;
-
-   if (nrb->mem)
-      nouveau_mem_free(ctx, nrb->mem);
-   FREE(nrb);
-}
-
-nouveau_renderbuffer *
-nouveau_renderbuffer_new(GLenum internalFormat, GLvoid *map,
-      			 GLuint offset,  GLuint pitch,
-			 __DRIdrawablePrivate *dPriv)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, 0);
-
-      nouveau_renderbuffer_pixelformat(nrb, internalFormat);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-
-      nrb->dPriv  = dPriv;
-      nrb->offset = offset;
-      nrb->pitch  = pitch;
-      nrb->map    = map;
-   }
-
-   return nrb;
-}
-
-static void
-nouveau_cliprects_drawable_set(nouveauContextPtr nmesa,
-      			       nouveau_renderbuffer *nrb)
-{
-   __DRIdrawablePrivate *dPriv = nrb->dPriv;
-
-   nmesa->numClipRects	= dPriv->numClipRects;
-   nmesa->pClipRects	= dPriv->pClipRects;
-   nmesa->drawX		= dPriv->x;
-   nmesa->drawY		= dPriv->y;
-   nmesa->drawW		= dPriv->w;
-   nmesa->drawH		= dPriv->h;
-}
-
-static void
-nouveau_cliprects_renderbuffer_set(nouveauContextPtr nmesa,
-      				   nouveau_renderbuffer *nrb)
-{
-   nmesa->numClipRects	= 1;
-   nmesa->pClipRects	= &nmesa->osClipRect;
-   nmesa->osClipRect.x1	= 0;
-   nmesa->osClipRect.y1	= 0;
-   nmesa->osClipRect.x2	= nrb->mesa.Width;
-   nmesa->osClipRect.y2	= nrb->mesa.Height;
-   nmesa->drawX		= 0;
-   nmesa->drawY		= 0;
-   nmesa->drawW		= nrb->mesa.Width;
-   nmesa->drawH		= nrb->mesa.Height;
-}
-
-void
-nouveau_window_moved(GLcontext *ctx)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *nrb;
-
-   nrb = (nouveau_renderbuffer *)ctx->DrawBuffer->_ColorDrawBuffers[0][0];
-   if (!nrb)
-      return;
-
-   if (!nrb->dPriv)
-      nouveau_cliprects_renderbuffer_set(nmesa, nrb);
-   else
-      nouveau_cliprects_drawable_set(nmesa, nrb);
-
-   /* Viewport depends on window size/position, nouveauCalcViewport
-    * will take care of calling the hw-specific WindowMoved
-    */
-   ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
-	 		     ctx->Viewport.Width, ctx->Viewport.Height);
-   /* Scissor depends on window position */
-   ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
-	 		    ctx->Scissor.Width, ctx->Scissor.Height);
-}
-
-GLboolean
-nouveau_build_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
-{
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   nouveau_renderbuffer *color[MAX_DRAW_BUFFERS];
-   nouveau_renderbuffer *depth;
-
-   _mesa_update_framebuffer(ctx);
-   _mesa_update_draw_buffer_bounds(ctx);
-
-   color[0] = (nouveau_renderbuffer *)fb->_ColorDrawBuffers[0][0];
-   if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped)
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer->Wrapped;
-   else
-      depth = (nouveau_renderbuffer *)fb->_DepthBuffer;
-
-   if (!nmesa->hw_func.BindBuffers(nmesa, 1, color, depth))
-      return GL_FALSE;
-   nouveau_window_moved(ctx);
-
-   return GL_TRUE;
-}
-
-static void
-nouveauDrawBuffer(GLcontext *ctx, GLenum buffer)
-{
-   nouveau_build_framebuffer(ctx, ctx->DrawBuffer);
-}
-
-static struct gl_framebuffer *
-nouveauNewFramebuffer(GLcontext *ctx, GLuint name)
-{
-   return _mesa_new_framebuffer(ctx, name);
-}
-
-static struct gl_renderbuffer *
-nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
-{
-   nouveau_renderbuffer *nrb;
-
-   nrb = CALLOC_STRUCT(nouveau_renderbuffer_t);
-   if (nrb) {
-      _mesa_init_renderbuffer(&nrb->mesa, name);
-
-      nrb->mesa.AllocStorage	= nouveau_renderbuffer_storage;
-      nrb->mesa.Delete		= nouveau_renderbuffer_delete;
-   }
-   return &nrb->mesa;
-}
-
-static void
-nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
-                       struct gl_framebuffer *fb, struct gl_framebuffer *fbread)
-{
-   if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
-      nouveau_build_framebuffer(ctx, fb);
-   }
-}
-
-static void
-nouveauFramebufferRenderbuffer(GLcontext *ctx,
-      			       struct gl_framebuffer *fb,
-      			       GLenum attachment,
-			       struct gl_renderbuffer *rb)
-{
-   _mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
-   nouveau_build_framebuffer(ctx, fb);
-}
-
-static void
-nouveauRenderTexture(GLcontext *ctx,
-      		     struct gl_framebuffer *fb,
-      		     struct gl_renderbuffer_attachment *att)
-{
-}
-
-static void
-nouveauFinishRenderTexture(GLcontext *ctx,
-      			   struct gl_renderbuffer_attachment *att)
-{
-}
-
-void
-nouveauInitBufferFuncs(struct dd_function_table *func)
-{
-   func->DrawBuffer		 = nouveauDrawBuffer;
-
-   func->NewFramebuffer		 = nouveauNewFramebuffer;
-   func->NewRenderbuffer	 = nouveauNewRenderbuffer;
-   func->BindFramebuffer	 = nouveauBindFramebuffer;
-   func->FramebufferRenderbuffer = nouveauFramebufferRenderbuffer;
-   func->RenderTexture		 = nouveauRenderTexture;
-   func->FinishRenderTexture	 = nouveauFinishRenderTexture;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h b/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
deleted file mode 100644
index d86455184c2..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_buffers.h
+++ /dev/null
@@ -1,48 +0,0 @@
-#ifndef __NOUVEAU_BUFFERS_H__
-#define __NOUVEAU_BUFFERS_H__
-
-#include <stdint.h>
-#include "mtypes.h"
-#include "utils.h"
-#include "renderbuffer.h"
-
-typedef struct nouveau_mem_t {
-   int type;
-   uint64_t offset;
-   uint64_t size;
-   void*    map;
-} nouveau_mem;
-
-extern nouveau_mem *nouveau_mem_alloc(GLcontext *ctx, int type,
-      				      GLuint size, GLuint align);
-extern void nouveau_mem_free(GLcontext *ctx, nouveau_mem *mem);
-extern uint32_t nouveau_mem_gpu_offset_get(GLcontext *ctx, nouveau_mem *mem);
-
-extern GLboolean nouveau_memformat_flat_emit(GLcontext *ctx,
-      					     nouveau_mem *dst,
-					     nouveau_mem *src,
-					     GLuint dst_offset,
-					     GLuint src_offset,
-					     GLuint size);
-
-typedef struct nouveau_renderbuffer_t {
-   struct gl_renderbuffer mesa; /* must be first! */
-   __DRIdrawablePrivate  *dPriv;
-
-   nouveau_mem *mem;
-   void *	map;
-
-   int		cpp;
-   uint32_t	offset;
-   uint32_t	pitch;
-} nouveau_renderbuffer;
-
-extern nouveau_renderbuffer *nouveau_renderbuffer_new(GLenum internalFormat,
-      GLvoid *map, GLuint offset, GLuint pitch, __DRIdrawablePrivate *dPriv);
-extern void nouveau_window_moved(GLcontext *ctx);
-extern GLboolean nouveau_build_framebuffer(GLcontext *, struct gl_framebuffer *);
-extern nouveau_renderbuffer *nouveau_current_draw_buffer(GLcontext *ctx);
-
-extern void nouveauInitBufferFuncs(struct dd_function_table *func);
-
-#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h b/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
index 8ec5c4a188a..b1a0e7bfd12 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
@@ -1,7 +1,4 @@
 static nouveau_card nouveau_card_list[]={
-{0x0008,  "EDGE 3D",                               0,                             NV_03, 0},
-{0x0009,  "EDGE 3D",                               0,                             NV_03, 0},
-{0x0010,  "Mutara V08",                            0,                             NV_03, 0},
 {0x0020,  "RIVA TNT",                              NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x0028,  "RIVA TNT2/TNT2 Pro",                    NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x0029,  "RIVA TNT2 Ultra",                       NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
@@ -116,6 +113,7 @@ static nouveau_card nouveau_card_list[]={
 {0x0193,  "GeForce 8800 GTS",                      NV30_TCL_PRIMITIVE_3D|0x5000,  NV_50, 0},
 {0x01A0,  "GeForce2 MX Integrated Graphics",       NV11_TCL_PRIMITIVE_3D,         NV_11, 0},
 {0x01D1,  "GeForce 7300 LE",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
+{0x01D3,  "GeForce 7300 SE",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x01D6,  "GeForce Go 7200",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x01D7,  "Quadro NVS 110M / GeForce Go 7300",     NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x01D8,  "GeForce Go 7400",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
@@ -220,10 +218,6 @@ static nouveau_card nouveau_card_list[]={
 {0x03D1,  "GeForce 6100 nForce 405",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x03D2,  "GeForce 6100 nForce 400",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x03D5,  "GeForce 6100 nForce 420",               NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0008,  "NV1",                                   0,                             NV_03, 0},
-{0x0009,  "DAC64",                                 0,                             NV_03, 0},
-{0x0018,  "Riva128",                               0,                             NV_03, 0},
-{0x0019,  "Riva128ZX",                             0,                             NV_03, 0},
 {0x0020,  "TNT",                                   NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x0028,  "TNT2",                                  NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x0029,  "UTNT2",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c
index 44392c0267c..f36483a3d49 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c
@@ -216,9 +216,6 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 	nouveauDDInitState( nmesa );
 	switch(nmesa->screen->card->type)
 	{
-		case NV_03:
-			//nv03TriInitFunctions( ctx );
-			break;
 		case NV_04:
 		case NV_05:
 			nv04TriInitFunctions( ctx );
@@ -315,21 +312,24 @@ GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv )
 	return GL_TRUE;
 }
 
-static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
-				 __DRIdrawablePrivate *dPriv)
+void
+nouveauDoSwapBuffers(nouveauContextPtr nmesa, __DRIdrawablePrivate *dPriv)
 {
 	struct gl_framebuffer *fb;
-	nouveau_renderbuffer *src, *dst;
+	nouveauScreenPtr screen = dPriv->driScreenPriv->private;
+	nouveau_renderbuffer_t *src;
 	drm_clip_rect_t *box;
 	int nbox, i;
 
 	fb = (struct gl_framebuffer *)dPriv->driverPrivate;
-	dst = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
-	src = (nouveau_renderbuffer*)
-		fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
+	if (fb->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT) {
+		src = (nouveau_renderbuffer_t *)
+			fb->Attachment[BUFFER_FRONT_LEFT].Renderbuffer;
+	} else {
+		src = (nouveau_renderbuffer_t *)
+			fb->Attachment[BUFFER_BACK_LEFT].Renderbuffer;
+	}
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	LOCK_HARDWARE(nmesa);
 	nbox = dPriv->numClipRects;
 	box  = dPriv->pClipRects;
@@ -341,13 +341,14 @@ static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
 			OUT_RING       (6); /* X8R8G8B8 */
 		else
 			OUT_RING       (4); /* R5G6B5 */
-		OUT_RING       ((dst->pitch << 16) | src->pitch);
-		OUT_RING       (src->offset);
-		OUT_RING       (dst->offset);
+		OUT_RING(((screen->frontPitch * screen->fbFormat) << 16) |
+			  src->pitch);
+		OUT_RING(src->offset);
+		OUT_RING(screen->frontOffset);
 	}
 
 	for (i=0; i<nbox; i++, box++) {
-		BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_POINT, 3);
+		BEGIN_RING_SIZE(NvSubImageBlit, NV_IMAGE_BLIT_POINT_IN, 3);
 		OUT_RING       (((box->y1 - dPriv->y) << 16) |
 				(box->x1 - dPriv->x));
 		OUT_RING       ((box->y1 << 16) | box->x1);
@@ -357,7 +358,6 @@ static void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
 	FIRE_RING();
 
 	UNLOCK_HARDWARE(nmesa);
-#endif
 }
 
 void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv)
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h
index 9a0be2cb2a5..77fe13a9cdc 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h
@@ -36,25 +36,24 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "mtypes.h"
 #include "tnl/t_vertex.h"
 
+#include "nouveau_fbo.h"
 #include "nouveau_screen.h"
-#include "nouveau_state_cache.h"
-#include "nouveau_buffers.h"
 #include "nouveau_shader.h"
+#include "nouveau_state_cache.h"
 #include "nouveau_sync.h"
 
 #include "xmlconfig.h"
 
-typedef struct nouveau_fifo_t{
-	int channel;
-	u_int32_t* buffer;
-	u_int32_t* mmio;
-	u_int32_t put_base;
-	u_int32_t current;
-	u_int32_t put;
-	u_int32_t free;
-	u_int32_t max;
-}
-nouveau_fifo;
+typedef struct nouveau_fifo {
+	struct drm_nouveau_channel_alloc drm;
+	uint32_t *pushbuf;
+	uint32_t *mmio;
+	uint32_t *notifier_block;
+	uint32_t  current;
+	uint32_t  put;
+	uint32_t  free;
+	uint32_t  max;
+} nouveau_fifo_t;
 
 #define TAG(x) nouveau##x
 #include "tnl_dd/t_dd_vertex.h"
@@ -83,10 +82,16 @@ typedef struct nouveau_hw_func_t {
 	GLboolean (*InitCard)(struct nouveau_context *);
 	/* Update buffer offset/pitch/format */
 	GLboolean (*BindBuffers)(struct nouveau_context *, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth);
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth);
 	/* Update anything that depends on the window position/size */
 	void      (*WindowMoved)(struct nouveau_context *);
+
+	/* Update projection matrix */
+	void	(*UpdateProjectionMatrix)(GLcontext *);
+
+	/* Update modelview matrix (used for lighting and vertex weight) */
+	void	(*UpdateModelviewMatrix)(GLcontext *);
 } nouveau_hw_func;
 
 typedef struct nouveau_context {
@@ -94,13 +99,7 @@ typedef struct nouveau_context {
 	GLcontext *glCtx;
 
 	/* The per-context fifo */
-	nouveau_fifo fifo;
-
-	/* The read-only regs */
-	volatile unsigned char* mmio;
-
-	/* The per-channel notifier block */
-	volatile void *notifier_block;
+	nouveau_fifo_t fifo;
 
 	/* Physical addresses of AGP/VRAM apertures */
 	uint64_t vram_phys;
@@ -109,12 +108,12 @@ typedef struct nouveau_context {
 	uint64_t gart_size;
 
 	/* Channel synchronisation */
-	struct drm_nouveau_notifier_alloc *syncNotifier;
+	struct drm_nouveau_notifierobj_alloc *syncNotifier;
 
 	/* ARB_occlusion_query / EXT_timer_query */
 	GLuint		  query_object_max;
 	GLboolean *	  query_alloc;
-	struct drm_nouveau_notifier_alloc *queryNotifier;
+	struct drm_nouveau_notifierobj_alloc *queryNotifier;
 
 	/* Additional hw-specific functions */
 	nouveau_hw_func hw_func;
@@ -130,6 +129,10 @@ typedef struct nouveau_context {
 	struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
 	GLuint vertex_attr_count;
 
+	/* Color and depth renderbuffers */
+	nouveau_renderbuffer_t *color_buffer;
+	nouveau_renderbuffer_t *depth_buffer;
+
 	/* Color buffer clear value */
 	uint32_t clear_color_value;
 
@@ -223,6 +226,9 @@ extern GLboolean nouveauMakeCurrent( __DRIcontextPrivate *driContextPriv,
 
 extern GLboolean nouveauUnbindContext( __DRIcontextPrivate *driContextPriv );
 
+extern void nouveauDoSwapBuffers(nouveauContextPtr nmesa,
+				 __DRIdrawablePrivate *dPriv);
+
 extern void nouveauSwapBuffers(__DRIdrawablePrivate *dPriv);
 
 extern void nouveauCopySubBuffer(__DRIdrawablePrivate *dPriv,
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_driver.c b/src/mesa/drivers/dri/nouveau/nouveau_driver.c
index ddc9535624b..4851c668356 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_driver.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_driver.c
@@ -117,6 +117,9 @@ static const GLubyte *nouveauGetString( GLcontext *ctx, GLenum name )
 static void nouveauFlush( GLcontext *ctx )
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (ctx->DrawBuffer->_ColorDrawBufferMask[0] == BUFFER_BIT_FRONT_LEFT)
+		nouveauDoSwapBuffers(nmesa, nmesa->driDrawable);
 	FIRE_RING();
 }
 
@@ -124,6 +127,7 @@ static void nouveauFlush( GLcontext *ctx )
 static void nouveauFinish( GLcontext *ctx )
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
 	nouveauFlush( ctx );
 	nouveauWaitForIdle( nmesa );
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fbo.c b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
new file mode 100644
index 00000000000..cc3da8b9bde
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fbo.c
@@ -0,0 +1,290 @@
+#include "utils.h"
+#include "framebuffer.h"
+#include "renderbuffer.h"
+#include "fbobject.h"
+
+#include "nouveau_context.h"
+#include "nouveau_fbo.h"
+#include "nouveau_fifo.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
+
+static GLboolean
+nouveau_renderbuffer_pixelformat(nouveau_renderbuffer_t * nrb,
+				 GLenum internalFormat)
+{
+	nrb->mesa.InternalFormat = internalFormat;
+
+	/*TODO: We probably want to extend this a bit, and maybe make
+	 *      card-specific? 
+	 */
+	switch (internalFormat) {
+	case GL_RGBA:
+	case GL_RGBA8:
+		nrb->mesa._BaseFormat = GL_RGBA;
+		nrb->mesa._ActualFormat = GL_RGBA8;
+		nrb->mesa.DataType = GL_UNSIGNED_BYTE;
+		nrb->mesa.RedBits = 8;
+		nrb->mesa.GreenBits = 8;
+		nrb->mesa.BlueBits = 8;
+		nrb->mesa.AlphaBits = 8;
+		nrb->cpp = 4;
+		break;
+	case GL_RGB:
+	case GL_RGB5:
+		nrb->mesa._BaseFormat = GL_RGB;
+		nrb->mesa._ActualFormat = GL_RGB5;
+		nrb->mesa.DataType = GL_UNSIGNED_BYTE;
+		nrb->mesa.RedBits = 5;
+		nrb->mesa.GreenBits = 6;
+		nrb->mesa.BlueBits = 5;
+		nrb->mesa.AlphaBits = 0;
+		nrb->cpp = 2;
+		break;
+	case GL_DEPTH_COMPONENT16:
+		nrb->mesa._BaseFormat = GL_DEPTH_COMPONENT;
+		nrb->mesa._ActualFormat = GL_DEPTH_COMPONENT16;
+		nrb->mesa.DataType = GL_UNSIGNED_SHORT;
+		nrb->mesa.DepthBits = 16;
+		nrb->cpp = 2;
+		break;
+	case GL_DEPTH_COMPONENT24:
+		nrb->mesa._BaseFormat = GL_DEPTH_COMPONENT;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.DepthBits = 24;
+		nrb->cpp = 4;
+		break;
+	case GL_STENCIL_INDEX8_EXT:
+		nrb->mesa._BaseFormat = GL_STENCIL_INDEX;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.StencilBits = 8;
+		nrb->cpp = 4;
+		break;
+	case GL_DEPTH24_STENCIL8_EXT:
+		nrb->mesa._BaseFormat = GL_DEPTH_STENCIL_EXT;
+		nrb->mesa._ActualFormat = GL_DEPTH24_STENCIL8_EXT;
+		nrb->mesa.DataType = GL_UNSIGNED_INT_24_8_EXT;
+		nrb->mesa.DepthBits = 24;
+		nrb->mesa.StencilBits = 8;
+		nrb->cpp = 4;
+		break;
+	default:
+		return GL_FALSE;
+		break;
+	}
+
+	return GL_TRUE;
+}
+
+static GLboolean
+nouveau_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
+			     GLenum internalFormat,
+			     GLuint width, GLuint height)
+{
+	nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *) rb;
+
+	if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
+		fprintf(stderr, "%s: unknown internalFormat\n", __func__);
+		return GL_FALSE;
+	}
+
+	/* If this buffer isn't statically alloc'd, we may need to ask the
+	 * drm for more memory */
+	if (rb->Width != width || rb->Height != height) {
+		GLuint pitch;
+
+		/* align pitches to 64 bytes */
+		pitch = ((width * nrb->cpp) + 63) & ~63;
+
+		if (nrb->mem)
+			nouveau_mem_free(ctx, nrb->mem);
+		nrb->mem = nouveau_mem_alloc(ctx,
+					     NOUVEAU_MEM_FB |
+					     NOUVEAU_MEM_MAPPED,
+					     pitch * height, 0);
+		if (!nrb->mem)
+			return GL_FALSE;
+
+		/* update nouveau_renderbuffer info */
+		nrb->offset = nouveau_mem_gpu_offset_get(ctx, nrb->mem);
+		nrb->pitch = pitch;
+	}
+
+	rb->Width = width;
+	rb->Height = height;
+	rb->InternalFormat = internalFormat;
+
+	nouveauSpanSetFunctions(nrb);
+
+	return GL_TRUE;
+}
+
+static void
+nouveau_renderbuffer_delete(struct gl_renderbuffer *rb)
+{
+	GET_CURRENT_CONTEXT(ctx);
+	nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *) rb;
+
+	if (nrb->mem)
+		nouveau_mem_free(ctx, nrb->mem);
+	FREE(nrb);
+}
+
+nouveau_renderbuffer_t *
+nouveau_renderbuffer_new(GLenum internalFormat)
+{
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = CALLOC_STRUCT(nouveau_renderbuffer);
+	if (!nrb)
+		return NULL;
+
+	_mesa_init_renderbuffer(&nrb->mesa, 0);
+
+	if (!nouveau_renderbuffer_pixelformat(nrb, internalFormat)) {
+		fprintf(stderr, "%s: unknown internalFormat\n", __func__);
+		return GL_FALSE;
+	}
+
+	nrb->mesa.AllocStorage = nouveau_renderbuffer_storage;
+	nrb->mesa.Delete = nouveau_renderbuffer_delete;
+
+	return nrb;
+}
+
+static void
+nouveau_cliprects_renderbuffer_set(nouveauContextPtr nmesa,
+				   nouveau_renderbuffer_t * nrb)
+{
+	nmesa->numClipRects = 1;
+	nmesa->pClipRects = &nmesa->osClipRect;
+	nmesa->osClipRect.x1 = 0;
+	nmesa->osClipRect.y1 = 0;
+	nmesa->osClipRect.x2 = nrb->mesa.Width;
+	nmesa->osClipRect.y2 = nrb->mesa.Height;
+	nmesa->drawX = 0;
+	nmesa->drawY = 0;
+	nmesa->drawW = nrb->mesa.Width;
+	nmesa->drawH = nrb->mesa.Height;
+}
+
+void
+nouveau_window_moved(GLcontext * ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = (nouveau_renderbuffer_t *)
+		ctx->DrawBuffer->_ColorDrawBuffers[0][0];
+	if (!nrb)
+		return;
+
+	nouveau_cliprects_renderbuffer_set(nmesa, nrb);
+
+	/* Viewport depends on window size/position, nouveauCalcViewport
+	 * will take care of calling the hw-specific WindowMoved
+	 */
+	ctx->Driver.Viewport(ctx, ctx->Viewport.X, ctx->Viewport.Y,
+			     ctx->Viewport.Width, ctx->Viewport.Height);
+	/* Scissor depends on window position */
+	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			    ctx->Scissor.Width, ctx->Scissor.Height);
+}
+
+GLboolean
+nouveau_build_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_renderbuffer_t *color[MAX_DRAW_BUFFERS];
+	nouveau_renderbuffer_t *depth;
+
+	_mesa_update_framebuffer(ctx);
+	_mesa_update_draw_buffer_bounds(ctx);
+
+	color[0] = (nouveau_renderbuffer_t *) fb->_ColorDrawBuffers[0][0];
+	if (fb->_DepthBuffer && fb->_DepthBuffer->Wrapped)
+		depth = (nouveau_renderbuffer_t *) fb->_DepthBuffer->Wrapped;
+	else
+		depth = (nouveau_renderbuffer_t *) fb->_DepthBuffer;
+
+	if (!nmesa->hw_func.BindBuffers(nmesa, 1, color, depth))
+		return GL_FALSE;
+	nouveau_window_moved(ctx);
+
+	return GL_TRUE;
+}
+
+static void
+nouveauDrawBuffer(GLcontext *ctx, GLenum buffer)
+{
+	nouveau_build_framebuffer(ctx, ctx->DrawBuffer);
+}
+
+static struct gl_framebuffer *
+nouveauNewFramebuffer(GLcontext *ctx, GLuint name)
+{
+	return _mesa_new_framebuffer(ctx, name);
+}
+
+static struct gl_renderbuffer *
+nouveauNewRenderbuffer(GLcontext *ctx, GLuint name)
+{
+	nouveau_renderbuffer_t *nrb;
+
+	nrb = CALLOC_STRUCT(nouveau_renderbuffer);
+	if (!nrb)
+		return NULL;
+
+	_mesa_init_renderbuffer(&nrb->mesa, name);
+
+	nrb->mesa.AllocStorage = nouveau_renderbuffer_storage;
+	nrb->mesa.Delete = nouveau_renderbuffer_delete;
+	return &nrb->mesa;
+}
+
+static void
+nouveauBindFramebuffer(GLcontext *ctx, GLenum target,
+		       struct gl_framebuffer *fb,
+		       struct gl_framebuffer *fbread)
+{
+	if (target == GL_FRAMEBUFFER_EXT || target == GL_DRAW_FRAMEBUFFER_EXT) {
+		nouveau_build_framebuffer(ctx, fb);
+	}
+}
+
+static void
+nouveauFramebufferRenderbuffer(GLcontext *ctx, struct gl_framebuffer *fb,
+			       GLenum attachment, struct gl_renderbuffer *rb)
+{
+	_mesa_framebuffer_renderbuffer(ctx, fb, attachment, rb);
+	nouveau_build_framebuffer(ctx, fb);
+}
+
+static void
+nouveauRenderTexture(GLcontext * ctx, struct gl_framebuffer *fb,
+		     struct gl_renderbuffer_attachment *att)
+{
+}
+
+static void
+nouveauFinishRenderTexture(GLcontext * ctx,
+			   struct gl_renderbuffer_attachment *att)
+{
+}
+
+void
+nouveauInitBufferFuncs(struct dd_function_table *func)
+{
+	func->DrawBuffer = nouveauDrawBuffer;
+
+	func->NewFramebuffer = nouveauNewFramebuffer;
+	func->NewRenderbuffer = nouveauNewRenderbuffer;
+	func->BindFramebuffer = nouveauBindFramebuffer;
+	func->FramebufferRenderbuffer = nouveauFramebufferRenderbuffer;
+	func->RenderTexture = nouveauRenderTexture;
+	func->FinishRenderTexture = nouveauFinishRenderTexture;
+}
+
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fbo.h b/src/mesa/drivers/dri/nouveau/nouveau_fbo.h
new file mode 100644
index 00000000000..787af4e9e4b
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fbo.h
@@ -0,0 +1,30 @@
+#ifndef __NOUVEAU_BUFFERS_H__
+#define __NOUVEAU_BUFFERS_H__
+
+#include <stdint.h>
+#include "mtypes.h"
+#include "utils.h"
+#include "renderbuffer.h"
+
+#include "nouveau_mem.h"
+
+typedef struct nouveau_renderbuffer {
+	struct gl_renderbuffer mesa;	/* must be first! */
+
+	nouveau_mem *mem;
+	void *map;
+
+	int cpp;
+	uint32_t offset;
+	uint32_t pitch;
+} nouveau_renderbuffer_t;
+
+extern nouveau_renderbuffer_t *nouveau_renderbuffer_new(GLenum internalFormat);
+extern void nouveau_window_moved(GLcontext *);
+extern GLboolean nouveau_build_framebuffer(GLcontext *,
+					   struct gl_framebuffer *);
+extern nouveau_renderbuffer_t *nouveau_current_draw_buffer(GLcontext *);
+
+extern void nouveauInitBufferFuncs(struct dd_function_table *);
+
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
index 7b5e96b4c26..5dc94e0520a 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.c
@@ -57,7 +57,8 @@ void WAIT_RING(nouveauContextPtr nmesa,u_int32_t size)
 		if(nmesa->fifo.put >= fifo_get) {
 			nmesa->fifo.free = nmesa->fifo.max - nmesa->fifo.current;
 			if(nmesa->fifo.free < size+1) {
-				OUT_RING(NV03_FIFO_CMD_JUMP | nmesa->fifo.put_base);
+				OUT_RING(NV03_FIFO_CMD_JUMP |
+					 nmesa->fifo.drm.put_base);
 				if(fifo_get <= RING_SKIPS) {
 					if(nmesa->fifo.put <= RING_SKIPS) /* corner case - will be idle */
 						NV_FIFO_WRITE_PUT(RING_SKIPS + 1);
@@ -98,54 +99,54 @@ void nouveauWaitForIdle(nouveauContextPtr nmesa)
 // here we call the fifo initialization ioctl and fill in stuff accordingly
 GLboolean nouveauFifoInit(nouveauContextPtr nmesa)
 {
-	struct drm_nouveau_fifo_alloc fifo_init;
 	int i, ret;
 
 #ifdef NOUVEAU_RING_DEBUG
 	return GL_TRUE;
 #endif
 
-	fifo_init.fb_ctxdma_handle = NvDmaFB;
-	fifo_init.tt_ctxdma_handle = NvDmaTT;
-	ret=drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_FIFO_ALLOC, &fifo_init, sizeof(fifo_init));
+	nmesa->fifo.drm.fb_ctxdma_handle = NvDmaFB;
+	nmesa->fifo.drm.tt_ctxdma_handle = NvDmaTT;
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_CHANNEL_ALLOC,
+				  &nmesa->fifo.drm, sizeof(nmesa->fifo.drm));
 	if (ret) {
-		FATAL("Fifo initialization ioctl failed (returned %d)\n",ret);
+		FATAL("Fifo initialization ioctl failed (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.cmdbuf, fifo_init.cmdbuf_size, &nmesa->fifo.buffer);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.cmdbuf,
+		     nmesa->fifo.drm.cmdbuf_size, &nmesa->fifo.pushbuf);
 	if (ret) {
-		FATAL("Unable to map the fifo (returned %d)\n",ret);
+		FATAL("Unable to map the fifo (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.ctrl, fifo_init.ctrl_size, &nmesa->fifo.mmio);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.ctrl,
+		     nmesa->fifo.drm.ctrl_size, &nmesa->fifo.mmio);
 	if (ret) {
-		FATAL("Unable to map the control regs (returned %d)\n",ret);
+		FATAL("Unable to map the control regs (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
-	ret = drmMap(nmesa->driFd, fifo_init.notifier,
-				   fifo_init.notifier_size,
-				   &nmesa->notifier_block);
+	ret = drmMap(nmesa->driFd, nmesa->fifo.drm.notifier,
+				   nmesa->fifo.drm.notifier_size,
+				   &nmesa->fifo.notifier_block);
 	if (ret) {
-		FATAL("Unable to map the notifier block (returned %d)\n",ret);
+		FATAL("Unable to map the notifier block (returned %d)\n", ret);
 		return GL_FALSE;
 	}
 
 	/* Setup our initial FIFO tracking params */
-	nmesa->fifo.channel  = fifo_init.channel;
-	nmesa->fifo.put_base = fifo_init.put_base;
 	nmesa->fifo.current  = 0;
 	nmesa->fifo.put      = 0;
-	nmesa->fifo.max      = (fifo_init.cmdbuf_size >> 2) - 1;
+	nmesa->fifo.max      = (nmesa->fifo.drm.cmdbuf_size >> 2) - 1;
 	nmesa->fifo.free     = nmesa->fifo.max - nmesa->fifo.current;
 
 	for (i=0; i<RING_SKIPS; i++)
 	   OUT_RING(0);
 	nmesa->fifo.free -= RING_SKIPS;
 
-	MESSAGE("Fifo init ok. Using context %d\n", fifo_init.channel);
+	MESSAGE("Fifo init ok. Using context %d\n", nmesa->fifo.drm.channel);
 	return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
index 23325dcea53..3ca1560e3ac 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
@@ -48,14 +48,14 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define NV_FIFO_READ(reg) *(volatile u_int32_t *)(nmesa->fifo.mmio + (reg/4))
 #define NV_FIFO_WRITE(reg,value) *(volatile u_int32_t *)(nmesa->fifo.mmio + (reg/4)) = value;
-#define NV_FIFO_READ_GET() ((NV_FIFO_READ(NV03_FIFO_REGS_DMAGET) - nmesa->fifo.put_base) >> 2)
+#define NV_FIFO_READ_GET() ((NV_FIFO_READ(NV03_FIFO_REGS_DMAGET) - nmesa->fifo.drm.put_base) >> 2)
 #define NV_FIFO_WRITE_PUT(val) do { \
 	if (NOUVEAU_RING_TRACE) {\
 		printf("FIRE_RING : 0x%08x\n", nmesa->fifo.current << 2); \
 		fflush(stdout); \
 		sleep(1); \
 	} \
-	NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.put_base); \
+	NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.drm.put_base); \
 } while(0)
 
 /* 
@@ -110,20 +110,20 @@ nouveau_fifo_remaining-=sz;							\
 		uint32_t* p=(uint32_t*)(ptr);							\
 		int i; printf("OUT_RINGp: (size 0x%x dwords) (%s)\n",sz, __func__); for(i=0;i<sz;i++) printf(" [0x%08x] 0x%08x   %f\n", (nmesa->fifo.current+i) << 2, *(p+i), *((float*)(p+i))); 	\
 	} \
-	memcpy(nmesa->fifo.buffer+nmesa->fifo.current,ptr,(sz)*4);		\
+	memcpy(nmesa->fifo.pushbuf+nmesa->fifo.current,ptr,(sz)*4);		\
 	nmesa->fifo.current+=(sz);						\
 }while(0)
 
 #define OUT_RING(n) do {							\
 if (NOUVEAU_RING_TRACE) \
     printf("OUT_RINGn: [0x%08x] 0x%08x (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-nmesa->fifo.buffer[nmesa->fifo.current++]=(n);					\
+nmesa->fifo.pushbuf[nmesa->fifo.current++]=(n);					\
 }while(0)
 
 #define OUT_RINGf(n) do {							\
 if (NOUVEAU_RING_TRACE) \
     printf("OUT_RINGf: [0x%08x] %.04f (%s)\n", nmesa->fifo.current << 2, n, __func__);        \
-*((float*)(nmesa->fifo.buffer+nmesa->fifo.current++))=(n);			\
+*((float*)(nmesa->fifo.pushbuf+nmesa->fifo.current++))=(n);			\
 }while(0)
 
 #define BEGIN_RING_SIZE(subchannel,tag,size) do {					\
@@ -145,10 +145,14 @@ extern void nouveau_state_cache_init(nouveauContextPtr nmesa);
 #define OUT_RING_CACHE(n) OUT_RING((n))
 #define OUT_RING_CACHEf(n) OUT_RINGf((n))
 #define OUT_RING_CACHEp(ptr, sz) OUT_RINGp((ptr), (sz))
+#define OUT_RING_CACHE_FORCE(n) OUT_RING((n))
+#define OUT_RING_CACHE_FORCEf(n) OUT_RINGf((n))
+#define OUT_RING_CACHE_FORCEp(ptr, sz) OUT_RINGp((ptr), (sz))
 #else
 #define BEGIN_RING_CACHE(subchannel,tag,size) do {					\
 	nmesa->state_cache.dirty=1;	 						\
 	nmesa->state_cache.current_pos=((tag)/4);					\
+	assert(nmesa->state_cache.current_pos + size <= NOUVEAU_STATE_CACHE_ENTRIES); \
 }while(0)
 
 #define OUT_RING_CACHE(n) do {									\
@@ -161,17 +165,40 @@ extern void nouveau_state_cache_init(nouveauContextPtr nmesa);
 }while(0)
 
 #define OUT_RING_CACHEf(n) do {									\
-	if ((*(float*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))!=(n)){	\
+	if ((*(GLfloat*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))!=(n)){	\
 		nmesa->state_cache.atoms[nmesa->state_cache.current_pos].dirty=1;	 	\
 		nmesa->state_cache.hdirty[nmesa->state_cache.current_pos/NOUVEAU_STATE_CACHE_HIER_SIZE]=1; 		\
-		(*(float*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))=(n);\
+		(*(GLfloat*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))=(n);\
 	}											\
 	nmesa->state_cache.current_pos++;							\
 }while(0)
 
 #define OUT_RING_CACHEp(ptr,sz) do {							\
-uint32_t* p=(uint32_t*)(ptr);								\
-int i; for(i=0;i<sz;i++) OUT_RING_CACHE(*(p+i)); 					\
+	GLuint* p=(GLuint*)(ptr);									\
+	int i;														\
+	for(i=0;i<sz;i++)											\
+		OUT_RING_CACHE(*(p+i)); 								\
+}while(0)
+
+#define OUT_RING_CACHE_FORCE(n) do {									\
+	nmesa->state_cache.atoms[nmesa->state_cache.current_pos].dirty=1; 		\
+	nmesa->state_cache.hdirty[nmesa->state_cache.current_pos/NOUVEAU_STATE_CACHE_HIER_SIZE]=1; 		\
+	nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value=(n);		\
+	nmesa->state_cache.current_pos++;							\
+}while(0)
+
+#define OUT_RING_CACHE_FORCEf(n) do {									\
+	nmesa->state_cache.atoms[nmesa->state_cache.current_pos].dirty=1;	 	\
+	nmesa->state_cache.hdirty[nmesa->state_cache.current_pos/NOUVEAU_STATE_CACHE_HIER_SIZE]=1; 		\
+	(*(GLfloat*)(&nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value))=(n);\
+	nmesa->state_cache.current_pos++;							\
+}while(0)
+
+#define OUT_RING_CACHE_FORCEp(ptr,sz) do {						\
+	GLuint* p=(GLuint*)(ptr);									\
+	int i;														\
+	for(i=0;i<sz;i++)											\
+		OUT_RING_CACHE_FORCE(*(p+i)); 							\
 }while(0)
 #endif
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_mem.c b/src/mesa/drivers/dri/nouveau/nouveau_mem.c
new file mode 100644
index 00000000000..35c53268b06
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_mem.c
@@ -0,0 +1,144 @@
+#include "mtypes.h"
+
+#include "nouveau_context.h"
+#include "nouveau_fifo.h"
+#include "nouveau_mem.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
+
+#define MAX_MEMFMT_LENGTH 32768
+
+/* Unstrided blit using NV_MEMORY_TO_MEMORY_FORMAT */
+GLboolean
+nouveau_memformat_flat_emit(GLcontext * ctx,
+			    nouveau_mem * dst, nouveau_mem * src,
+			    GLuint dst_offset, GLuint src_offset,
+			    GLuint size)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	uint32_t src_handle, dst_handle;
+	GLuint count;
+
+	if (src_offset + size > src->size) {
+		MESSAGE("src out of nouveau_mem bounds\n");
+		return GL_FALSE;
+	}
+	if (dst_offset + size > dst->size) {
+		MESSAGE("dst out of nouveau_mem bounds\n");
+		return GL_FALSE;
+	}
+
+	src_handle = (src->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
+	dst_handle = (dst->type & NOUVEAU_MEM_FB) ? NvDmaFB : NvDmaTT;
+	src_offset += nouveau_mem_gpu_offset_get(ctx, src);
+	dst_offset += nouveau_mem_gpu_offset_get(ctx, dst);
+
+	BEGIN_RING_SIZE(NvSubMemFormat,
+			NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN, 2);
+	OUT_RING(src_handle);
+	OUT_RING(dst_handle);
+
+	count = (size / MAX_MEMFMT_LENGTH) + 
+		((size % MAX_MEMFMT_LENGTH) ? 1 : 0);
+
+	while (count--) {
+		GLuint length =
+		    (size > MAX_MEMFMT_LENGTH) ? MAX_MEMFMT_LENGTH : size;
+
+		BEGIN_RING_SIZE(NvSubMemFormat,
+				NV_MEMORY_TO_MEMORY_FORMAT_OFFSET_IN, 8);
+		OUT_RING(src_offset);
+		OUT_RING(dst_offset);
+		OUT_RING(0);	/* pitch in */
+		OUT_RING(0);	/* pitch out */
+		OUT_RING(length);	/* line length */
+		OUT_RING(1);	/* number of lines */
+		OUT_RING((1 << 8) /* dst_inc */ |(1 << 0) /* src_inc */ );
+		OUT_RING(0);	/* buffer notify? */
+		FIRE_RING();
+
+		src_offset += length;
+		dst_offset += length;
+		size -= length;
+	}
+
+	return GL_TRUE;
+}
+
+void nouveau_mem_free(GLcontext * ctx, nouveau_mem * mem)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	struct drm_nouveau_mem_free memf;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr, "%s: type=0x%x, offset=0x%x, size=0x%x\n",
+			__func__, mem->type, (GLuint) mem->offset,
+			(GLuint) mem->size);
+	}
+
+	if (mem->map)
+		drmUnmap(mem->map, mem->size);
+	memf.flags = mem->type;
+	memf.offset = mem->offset;
+	drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_MEM_FREE, &memf,
+			sizeof(memf));
+	FREE(mem);
+}
+
+nouveau_mem *nouveau_mem_alloc(GLcontext *ctx, uint32_t flags, GLuint size,
+			       GLuint align)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	struct drm_nouveau_mem_alloc mema;
+	nouveau_mem *mem;
+	int ret;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr,
+			"%s: requested: flags=0x%x, size=0x%x, align=0x%x\n",
+			__func__, flags, (GLuint) size, align);
+	}
+
+	mem = CALLOC(sizeof(nouveau_mem));
+	if (!mem)
+		return NULL;
+
+	mema.flags = flags;
+	mema.size = mem->size = size;
+	mema.alignment = align;
+	mem->map = NULL;
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_MEM_ALLOC,
+				  &mema, sizeof(mema));
+	if (ret) {
+		FREE(mem);
+		return NULL;
+	}
+	mem->offset = mema.offset;
+	mem->type = mema.flags;
+
+	if (NOUVEAU_DEBUG & DEBUG_MEM) {
+		fprintf(stderr,
+			"%s: actual: type=0x%x, offset=0x%x, size=0x%x\n",
+			__func__, mem->type, (GLuint) mem->offset,
+			(GLuint) mem->size);
+	}
+
+	if (flags & NOUVEAU_MEM_MAPPED)
+		ret = drmMap(nmesa->driFd, mema.map_handle, mem->size,
+			     &mem->map);
+	if (ret) {
+		mem->map = NULL;
+		nouveau_mem_free(ctx, mem);
+		mem = NULL;
+	}
+
+	return mem;
+}
+
+uint32_t nouveau_mem_gpu_offset_get(GLcontext * ctx, nouveau_mem * mem)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	return mem->offset;
+}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_mem.h b/src/mesa/drivers/dri/nouveau/nouveau_mem.h
new file mode 100644
index 00000000000..6db73f4708e
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_mem.h
@@ -0,0 +1,23 @@
+#ifndef __NOUVEAU_MEM_H__
+#define __NOUVEAU_MEM_H__
+
+typedef struct nouveau_mem_t {
+	int type;
+	uint64_t offset;
+	uint64_t size;
+	void *map;
+} nouveau_mem;
+
+extern nouveau_mem *nouveau_mem_alloc(GLcontext *, uint32_t flags,
+				      GLuint size, GLuint align);
+extern void nouveau_mem_free(GLcontext *, nouveau_mem *);
+extern uint32_t nouveau_mem_gpu_offset_get(GLcontext *, nouveau_mem *);
+
+extern GLboolean nouveau_memformat_flat_emit(GLcontext *,
+					     nouveau_mem *dst,
+					     nouveau_mem *src,
+					     GLuint dst_offset,
+					     GLuint src_offset,
+					     GLuint size);
+
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.c b/src/mesa/drivers/dri/nouveau/nouveau_object.c
index a143488e8d5..8f33093f1a5 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.c
@@ -10,7 +10,7 @@ GLboolean nouveauCreateContextObject(nouveauContextPtr nmesa,
 	struct drm_nouveau_grobj_alloc cto;
 	int ret;
 
-	cto.channel = nmesa->fifo.channel;
+	cto.channel = nmesa->fifo.drm.channel;
 	cto.handle  = handle;
 	cto.class   = class;
 	ret = drmCommandWrite(nmesa->driFd, DRM_NOUVEAU_GROBJ_ALLOC,
@@ -39,26 +39,57 @@ void nouveauObjectInit(nouveauContextPtr nmesa)
 		nouveauCreateContextObject(nmesa, NvCtxSurf3D, NV04_CONTEXT_SURFACES_3D);
 	}
 	if (nmesa->screen->card->type>=NV_11) {
-		nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT);
+		nouveauCreateContextObject(nmesa, NvImageBlit, NV11_IMAGE_BLIT);
 	} else {
 		nouveauCreateContextObject(nmesa, NvImageBlit, NV_IMAGE_BLIT);
 	}
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<NV_20)) {
+		nouveauCreateContextObject(nmesa, NvGdiRectText, NV04_GDI_RECTANGLE_TEXT);
+		nouveauCreateContextObject(nmesa, NvRasterOp, NV03_PRIMITIVE_RASTER_OP);
+		nouveauCreateContextObject(nmesa, NvPattern, NV04_IMAGE_PATTERN);
+	}
 	nouveauCreateContextObject(nmesa, NvMemFormat, NV_MEMORY_TO_MEMORY_FORMAT);
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf2D, NvCtxSurf2D);
 	BEGIN_RING_SIZE(NvSubCtxSurf2D, NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY0, 2);
 	OUT_RING(NvDmaFB);
 	OUT_RING(NvDmaFB);
 
 	nouveauObjectOnSubchannel(nmesa, NvSubImageBlit, NvImageBlit);
-	BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_CONTEXT_SURFACES_2D, 1);
+	BEGIN_RING_SIZE(NvSubImageBlit, NV_IMAGE_BLIT_SET_SURFACES_2D, 1);
 	OUT_RING(NvCtxSurf2D);
-	BEGIN_RING_SIZE(NvSubImageBlit, NV10_IMAGE_BLIT_SET_OPERATION, 1);
+	BEGIN_RING_SIZE(NvSubImageBlit, NV_IMAGE_BLIT_OPERATION, 1);
 	OUT_RING(3); /* SRCCOPY */
 
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<NV_20)) {
+		nouveauObjectOnSubchannel(nmesa, NvSubPattern, NvPattern);
+
+		BEGIN_RING_SIZE(NvSubPattern, NV04_IMAGE_PATTERN_COLOR_FORMAT, 4);
+		OUT_RING(1);	/* A16R5G6B5 */
+		OUT_RING(1);	/* little endian */
+		OUT_RING(0);	/* 8x8 */
+		OUT_RING(1);	/* monochrome */
+
+		nouveauObjectOnSubchannel(nmesa, NvSubRasterOp, NvRasterOp);
+
+		BEGIN_RING_SIZE(NvSubRasterOp, NV03_PRIMITIVE_RASTER_OP_DMA_NOTIFY, 1);
+		OUT_RING(NvSyncNotify);
+
+		nouveauObjectOnSubchannel(nmesa, NvSubGdiRectText, NvGdiRectText);
+
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_SET_DMA_NOTIFY, 1);
+		OUT_RING(NvSyncNotify);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_PATTERN, 2);
+		OUT_RING(NvPattern);
+		OUT_RING(NvRasterOp);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_SURFACE, 1);
+		OUT_RING(NvCtxSurf2D);
+		BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_FORMAT, 2);
+		OUT_RING(1);	/* X1R5G5B5 */
+		OUT_RING(1);	/* little endian */
+	}
+
 	nouveauObjectOnSubchannel(nmesa, NvSubMemFormat, NvMemFormat);
-#endif
 
 	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.h b/src/mesa/drivers/dri/nouveau/nouveau_object.h
index 8c72d014daa..10859de1190 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.h
@@ -3,8 +3,6 @@
 
 #include "nouveau_context.h"
 
-#define ALLOW_MULTI_SUBCHANNEL
-
 void nouveauObjectInit(nouveauContextPtr nmesa);
 
 enum DMAObjects {
@@ -13,6 +11,9 @@ enum DMAObjects {
 	NvImageBlit		= 0x80000021,
 	NvMemFormat		= 0x80000022,
 	NvCtxSurf3D		= 0x80000023,
+	NvGdiRectText		= 0x80000024,
+	NvPattern		= 0x80000025,
+	NvRasterOp		= 0x80000026,
 	NvDmaFB			= 0xD0FB0001,
 	NvDmaTT			= 0xD0AA0001,
 	NvSyncNotify		= 0xD0000001,
@@ -24,6 +25,9 @@ enum DMASubchannel {
 	NvSubImageBlit	= 1,
 	NvSubMemFormat	= 2,
 	NvSubCtxSurf3D	= 3,
+	NvSubGdiRectText= 4,
+	NvSubPattern	= 5,
+	NvSubRasterOp	= 6,
 	NvSub3D		= 7,
 };
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.c b/src/mesa/drivers/dri/nouveau/nouveau_query.c
index 01541400690..e5c1750a8ec 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_query.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_query.c
@@ -167,7 +167,8 @@ nouveauQueryInitFuncs(GLcontext *ctx)
 	if (nmesa->screen->card->type < NV_20)
 		return;
 
-	nmesa->query_object_max = (0x4000 / 32);
+	/* Reserve half the notifier block for use as query objects */
+	nmesa->query_object_max = (nmesa->fifo.drm.notifier_size / 2) / 32;
 	nmesa->queryNotifier =
 		nouveau_notifier_new(ctx, NvQueryNotify,
 					  nmesa->query_object_max);
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_reg.h b/src/mesa/drivers/dri/nouveau/nouveau_reg.h
index 8758b538c85..6b90bab0766 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_reg.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_reg.h
@@ -43,7 +43,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 **************************************************************************
 
-   Created from objects.c rev. 1.398
+   Created from objects.c rev. 1.440
 */
 
 #ifndef _NOUVEAU_REG_H
@@ -53,8 +53,17 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 Object NV01_CONTEXT_CLIP_RECTANGLE used on: NV03 NV04 NV10 NV15 NV20 NV40 G70
 */
 #define			NV01_CONTEXT_CLIP_RECTANGLE				0x00000019
+#	define		NV01_CONTEXT_CLIP_RECTANGLE_NOP				0x00000100
+#	define		NV01_CONTEXT_CLIP_RECTANGLE_NOTIFY			0x00000104
+#	define		NV01_CONTEXT_CLIP_RECTANGLE_DMA_NOTIFY			0x00000180
 #	define		NV01_CONTEXT_CLIP_RECTANGLE_SET_POINT			0x00000300	/* Parameters: x y */
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_POINT_X_MASK	0x0000ffff
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_POINT_Y_MASK	0xffff0000
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_POINT_Y_SHIFT	16
 #	define		NV01_CONTEXT_CLIP_RECTANGLE_SET_SIZE			0x00000304	/* Parameters: width height */
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV01_CONTEXT_CLIP_RECTANGLE_SET_SIZE_HEIGHT_SHIFT	16
 
 /****************************************** 
 Object NV_MEMORY_TO_MEMORY_FORMAT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
@@ -72,31 +81,69 @@ Object NV_MEMORY_TO_MEMORY_FORMAT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_LINE_LENGTH_IN		0x0000031c
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_LINE_COUNT			0x00000320
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_FORMAT			0x00000324	/* Parameters: src_inc dst_inc */
+#		 define		NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_SRC_INC_MASK	0x00000007
+#		 define		NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_DST_INC_MASK	0x00000700
+#		 define		NV_MEMORY_TO_MEMORY_FORMAT_FORMAT_DST_INC_SHIFT	8
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_BUF_NOTIFY			0x00000328
 
 /****************************************** 
 Object NV03_PRIMITIVE_RASTER_OP used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
 */
 #define			NV03_PRIMITIVE_RASTER_OP				0x00000043
-#	define		NV03_PRIMITIVE_RASTER_OP_NOTIFY				0x00000100
+#	define		NV03_PRIMITIVE_RASTER_OP_NOP				0x00000100
+#	define		NV03_PRIMITIVE_RASTER_OP_NOTIFY				0x00000104
 #	define		NV03_PRIMITIVE_RASTER_OP_DMA_NOTIFY			0x00000180
 #	define		NV03_PRIMITIVE_RASTER_OP_LOGIC_OP			0x00000300	/* Parameters: logic_op */
+#		 define		NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_MASK	0x000000f0
+#		 define		NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_SHIFT	4
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_CLEAR	0x0000
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_NOR	0x0001
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_AND_INVERTED	0x0002
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_COPY_INVERTED	0x0003
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_AND_REVERSE	0x0004
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_INVERT	0x0005
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_XOR	0x0006
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_NAND	0x0007
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_AND	0x0008
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_EQUIV	0x0009
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_NOOP	0x000a
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_OR_INVERTED	0x000b
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_COPY	0x000c
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_OR_REVERSE	0x000d
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_OR	0x000e
+#		 define NV03_PRIMITIVE_RASTER_OP_LOGIC_OP_LOGIC_OP_SET	0x000f
 
 /****************************************** 
 Object NV04_GDI_RECTANGLE_TEXT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
 */
 #define			NV04_GDI_RECTANGLE_TEXT					0x0000004a
+#	define		NV04_GDI_RECTANGLE_TEXT_NOP				0x00000100
+#	define		NV04_GDI_RECTANGLE_TEXT_NOTIFY				0x00000104
 #	define		NV04_GDI_RECTANGLE_TEXT_SET_DMA_NOTIFY			0x00000180
+#	define		NV04_GDI_RECTANGLE_TEXT_SET_DMA_FONTS			0x00000184
 #	define		NV04_GDI_RECTANGLE_TEXT_PATTERN				0x00000188
 #	define		NV04_GDI_RECTANGLE_TEXT_ROP5				0x0000018c
 #	define		NV04_GDI_RECTANGLE_TEXT_SURFACE				0x00000198
 #	define		NV04_GDI_RECTANGLE_TEXT_OPERATION			0x000002fc
 #	define		NV04_GDI_RECTANGLE_TEXT_FORMAT				0x00000300
+#	define		NV04_GDI_RECTANGLE_TEXT_MONO_FORMAT			0x00000304
 #	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL			0x000005f4	/* Parameters: left top */
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL_LEFT_MASK	0x0000ffff
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL_TOP_MASK	0xffff0000
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL_TOP_SHIFT	16
 #	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_BR			0x000005f8	/* Parameters: right bottom */
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_BR_RIGHT_MASK	0x0000ffff
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_BR_BOTTOM_MASK	0xffff0000
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_BR_BOTTOM_SHIFT	16
 #	define		NV04_GDI_RECTANGLE_TEXT_FILL_VALUE			0x000005fc
 #	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_TL			0x00000600	/* Parameters: left top */
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_TL_LEFT_MASK	0x0000ffff
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_TL_TOP_MASK	0xffff0000
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_TL_TOP_SHIFT	16
 #	define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_BR			0x00000604	/* Parameters: right bottom */
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_BR_RIGHT_MASK	0x0000ffff
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_BR_BOTTOM_MASK	0xffff0000
+#		 define		NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL2_BR_BOTTOM_SHIFT	16
 
 /****************************************** 
 Object NV04_SWIZZLED_SURFACE used on: NV04 NV10 NV15 
@@ -104,7 +151,12 @@ Object NV04_SWIZZLED_SURFACE used on: NV04 NV10 NV15
 #define			NV04_SWIZZLED_SURFACE					0x00000052
 #	define		NV04_SWIZZLED_SURFACE_DMA_NOTIFY			0x00000180
 #	define		NV04_SWIZZLED_SURFACE_DMA_IMAGE				0x00000184
-#	define		NV04_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2(height) log2(width) color */
+#	define		NV04_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2_height log2_width color */
+#		 define		NV04_SWIZZLED_SURFACE_FORMAT_LOG2_HEIGHT_MASK	0xff000000
+#		 define		NV04_SWIZZLED_SURFACE_FORMAT_LOG2_HEIGHT_SHIFT	24
+#		 define		NV04_SWIZZLED_SURFACE_FORMAT_LOG2_WIDTH_MASK	0x00ff0000
+#		 define		NV04_SWIZZLED_SURFACE_FORMAT_LOG2_WIDTH_SHIFT	16
+#		 define		NV04_SWIZZLED_SURFACE_FORMAT_COLOR_MASK	0x0000ffff
 #	define		NV04_SWIZZLED_SURFACE_OFFSET				0x00000304
 
 /****************************************** 
@@ -115,10 +167,31 @@ Object NV04_CONTEXT_SURFACES_3D used on: NV04
 #	define		NV04_CONTEXT_SURFACES_3D_DMA_COLOR			0x00000184
 #	define		NV04_CONTEXT_SURFACES_3D_DMA_ZETA			0x00000188
 #	define		NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL		0x000002f8	/* Parameters: x width */
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_X_MASK	0x0000ffff
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_WIDTH_MASK	0xffff0000
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL_WIDTH_SHIFT	16
 #	define		NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL			0x000002fc	/* Parameters: y height */
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_Y_MASK	0x0000ffff
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_HEIGHT_MASK	0xffff0000
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_VERTICAL_HEIGHT_SHIFT	16
 #	define		NV04_CONTEXT_SURFACES_3D_FORMAT				0x00000300	/* Parameters: color type width height */
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_COLOR_MASK	0x000000ff
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_MASK	0x0000ff00
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_SHIFT	8
+#		 define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_pitch	0x0001
+#		 define NV04_CONTEXT_SURFACES_3D_FORMAT_TYPE_swizzle	0x0002
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_WIDTH_MASK	0x00ff0000
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_WIDTH_SHIFT	16
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_HEIGHT_MASK	0xff000000
+#		 define		NV04_CONTEXT_SURFACES_3D_FORMAT_HEIGHT_SHIFT	24
 #	define		NV04_CONTEXT_SURFACES_3D_CLIP_SIZE			0x00000304	/* Parameters: width height */
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_HEIGHT_MASK	0xffff8000
+#		 define		NV04_CONTEXT_SURFACES_3D_CLIP_SIZE_HEIGHT_SHIFT	15
 #	define		NV04_CONTEXT_SURFACES_3D_PITCH				0x00000308	/* Parameters: color zeta */
+#		 define		NV04_CONTEXT_SURFACES_3D_PITCH_COLOR_MASK	0x0000ffff
+#		 define		NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_MASK	0xffff0000
+#		 define		NV04_CONTEXT_SURFACES_3D_PITCH_ZETA_SHIFT	16
 #	define		NV04_CONTEXT_SURFACES_3D_OFFSET_COLOR			0x0000030c
 #	define		NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA			0x00000310
 
@@ -134,20 +207,137 @@ Object NV04_DX5_TEXTURED_TRIANGLE used on: NV04
 #	define		NV04_DX5_TEXTURED_TRIANGLE_SURFACE			0x0000018c
 #	define		NV04_DX5_TEXTURED_TRIANGLE_COLOR_KEY			0x00000300
 #	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_OFFSET		0x00000304
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT		0x00000308	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT		0x00000308	/* Parameters: color mipmaps log_u log_v wrap_s wrap_t */
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_MASK	0x00000f00
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_SHIFT	8
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_Y8	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_A1R5G5B5	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_X1R5G5B5	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_A4R4G4B4	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_R5G6B5	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_A8R8G8B8	0x0006
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_COLOR_X8R8G8B8	0x0007
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_MIPMAPS_MASK	0x0000f000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_MIPMAPS_SHIFT	12
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_LOG_U_MASK	0x000f0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_LOG_U_SHIFT	16
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_LOG_V_MASK	0x00f00000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_LOG_V_SHIFT	20
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_S_MASK	0x07000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_S_SHIFT	24
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_S_REPEAT	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_S_MIRRORED	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_S_CLAMP	0x0003
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_T_MASK	0x70000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_T_SHIFT	28
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_T_REPEAT	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_T_MIRRORED	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FORMAT_WRAP_T_CLAMP	0x0003
 #	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER		0x0000030c	/* Parameters: magfilter minfilter lodbias */
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_MASK	0x70000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_SHIFT	28
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_NEAREST	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_LINEAR	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MAGFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_MASK	0x07000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_SHIFT	24
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_NEAREST	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_LINEAR	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_MINFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_LODBIAS_MASK	0x00ff0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_FILTER_LODBIAS_SHIFT	16
 #	define		NV04_DX5_TEXTURED_TRIANGLE_BLEND			0x00000310	/* Parameters: texture benable dst src */
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_TEXTURE_MASK	0x0000000f
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_BENABLE_MASK	0x00100000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_BENABLE	(1 << 20)
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_BENABLE_TRUE	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_BENABLE_FALSE	0x0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_MASK	0xf0000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_SHIFT	28
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_0	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_1	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_src_col	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_inv_src_col	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_src_a	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_inv_src_a	0x0006
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_dst_col	0x0009
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_DST_inv_dst_col	0x000a
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_MASK	0x0f000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_SHIFT	24
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_0	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_1	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_src_col	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_inv_src_col	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_src_a	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_inv_src_a	0x0006
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_dst_col	0x0009
+#		 define NV04_DX5_TEXTURED_TRIANGLE_BLEND_SRC_inv_dst_col	0x000a
 #	define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL			0x00000314	/* Parameters: alpharef alphafunc alphaenable zenable zwrite zfunc cullmode */
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAREF_MASK	0x000000ff
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_MASK	0x00000f00
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_SHIFT	8
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_NEVER	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_LESS	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_EQUAL	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_LEQUAL	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_GREATER	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_NOTEQUAL	0x0006
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_GEQUAL	0x0007
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAFUNC_ALWAYS	0x0008
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAENABLE_MASK	0x00001000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAENABLE	(1 << 12)
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAENABLE_TRUE	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ALPHAENABLE_FALSE	0x0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZENABLE_MASK	0x00004000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZENABLE	(1 << 14)
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZENABLE_TRUE	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZENABLE_FALSE	0x0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZWRITE_MASK	0x01000000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZWRITE	(1 << 24)
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZWRITE_TRUE	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZWRITE_FALSE	0x0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_MASK	0x000f0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_SHIFT	16
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_NEVER	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_LESS	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_EQUAL	0x0003
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_LEQUAL	0x0004
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_GREATER	0x0005
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_NOTEQUAL	0x0006
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_GEQUAL	0x0007
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_ZFUNC_ALWAYS	0x0008
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULLMODE_MASK	0x00300000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULLMODE_SHIFT	20
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULLMODE_NONE	0x0001
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULLMODE_CW	0x0002
+#		 define NV04_DX5_TEXTURED_TRIANGLE_CONTROL_CULLMODE_CCW	0x0003
 #	define		NV04_DX5_TEXTURED_TRIANGLE_FOG_COLOR			0x00000318
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_INV_W( d)			(0x0000040c + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_COLOR( d)			(0x00000410 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_SPECULAR( d)			(0x00000414 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_S( d)		(0x00000418 + d * 0x0020)
-#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_T( d)		(0x0000041c + d * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_INV_W( d)			(0x0000040c + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_COLOR( d)			(0x00000410 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_SPECULAR( d)			(0x00000414 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_S( d)		(0x00000418 + (d) * 0x0020)
+#	define		NV04_DX5_TEXTURED_TRIANGLE_TEXTURE_T( d)		(0x0000041c + (d) * 0x0020)
 #	define		NV04_DX5_TEXTURED_TRIANGLE_DRAW				0x00000600	/* Parameters: v0 v1 v2 v3 v4 v5 */
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V0_MASK	0x0000000f
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V1_MASK	0x000000f0
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V1_SHIFT	4
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V2_MASK	0x00000f00
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V2_SHIFT	8
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V3_MASK	0x0000f000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V3_SHIFT	12
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V4_MASK	0x000f0000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V4_SHIFT	16
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V5_MASK	0x00f00000
+#		 define		NV04_DX5_TEXTURED_TRIANGLE_DRAW_V5_SHIFT	20
 
 /****************************************** 
 Object NV04_DX6_MULTITEX_TRIANGLE used on: NV04 NV10 NV15 
@@ -161,31 +351,252 @@ Object NV04_DX6_MULTITEX_TRIANGLE used on: NV04 NV10 NV15
 #	define		NV04_DX6_MULTITEX_TRIANGLE_SURFACE			0x0000018c
 #	define		NV04_DX6_MULTITEX_TRIANGLE_OFFSET0			0x00000308
 #	define		NV04_DX6_MULTITEX_TRIANGLE_OFFSET1			0x0000030c
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0			0x00000310	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
-#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1			0x00000314	/* Parameters: color mipmaps log(u) log(v) wrap_s wrap_t */
+#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0			0x00000310	/* Parameters: color mipmaps log_u log_v wrap_s wrap_t */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_MASK	0x00000f00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_SHIFT	8
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_Y8	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_A1R5G5B5	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_X1R5G5B5	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_A4R4G4B4	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_R5G6B5	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_A8R8G8B8	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_COLOR_X8R8G8B8	0x0007
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_MIPMAPS_MASK	0x0000f000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_MIPMAPS_SHIFT	12
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_LOG_U_MASK	0x000f0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_LOG_U_SHIFT	16
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_LOG_V_MASK	0x00f00000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_LOG_V_SHIFT	20
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_S_MASK	0x07000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_S_SHIFT	24
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_S_REPEAT	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_S_MIRRORED	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_S_CLAMP	0x0003
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_T_MASK	0x70000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_T_SHIFT	28
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_T_REPEAT	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_T_MIRRORED	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT0_WRAP_T_CLAMP	0x0003
+#	define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1			0x00000314	/* Parameters: color mipmaps log_u log_v wrap_s wrap_t */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_MASK	0x00000f00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_SHIFT	8
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_Y8	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_A1R5G5B5	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_X1R5G5B5	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_A4R4G4B4	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_R5G6B5	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_A8R8G8B8	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_COLOR_X8R8G8B8	0x0007
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_MIPMAPS_MASK	0x0000f000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_MIPMAPS_SHIFT	12
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_LOG_U_MASK	0x000f0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_LOG_U_SHIFT	16
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_LOG_V_MASK	0x00f00000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_LOG_V_SHIFT	20
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_S_MASK	0x07000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_S_SHIFT	24
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_S_REPEAT	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_S_MIRRORED	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_S_CLAMP	0x0003
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_T_MASK	0x70000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_T_SHIFT	28
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_T_REPEAT	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_T_MIRRORED	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FORMAT1_WRAP_T_CLAMP	0x0003
 #	define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0			0x00000318	/* Parameters: magfilter minfilter lodbias */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_MASK	0x70000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_SHIFT	28
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_NEAREST	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_LINEAR	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MAGFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_MASK	0x07000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_SHIFT	24
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_NEAREST	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_LINEAR	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER0_MINFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_LODBIAS_MASK	0x00ff0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER0_LODBIAS_SHIFT	16
 #	define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1			0x0000031c	/* Parameters: magfilter minfilter lodbias */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_MASK	0x70000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_SHIFT	28
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_NEAREST	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_LINEAR	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MAGFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_MASK	0x07000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_SHIFT	24
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_NEAREST	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_LINEAR	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_FILTER1_MINFILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_LODBIAS_MASK	0x00ff0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_FILTER1_LODBIAS_SHIFT	16
 #	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_ALPHA		0x00000320
 #	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_0_COLOR		0x00000324
 #	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_ALPHA		0x0000032c
 #	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_1_COLOR		0x00000330
 #	define		NV04_DX6_MULTITEX_TRIANGLE_COMBINE_FACTOR		0x00000334
 #	define		NV04_DX6_MULTITEX_TRIANGLE_BLEND			0x00000338	/* Parameters: benable dst src */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_BENABLE_MASK	0x00100000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_BENABLE	(1 << 20)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_BENABLE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_BENABLE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_MASK	0xf0000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_SHIFT	28
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_0	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_1	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_src_col	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_inv_src_col	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_src_a	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_inv_src_a	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_dst_col	0x0009
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_DST_inv_dst_col	0x000a
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_MASK	0x0f000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_SHIFT	24
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_0	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_1	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_src_col	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_inv_src_col	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_src_a	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_inv_src_a	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_dst_col	0x0009
+#		 define NV04_DX6_MULTITEX_TRIANGLE_BLEND_SRC_inv_dst_col	0x000a
 #	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0			0x0000033c	/* Parameters: red_write green_write blue_write alpha_write alpha_write stencil_write alpharef alphafunc alphaenable zenable zwrite zfunc cullmode */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_MASK	0x08000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE	(1 << 27)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_RED_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_MASK	0x10000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE	(1 << 28)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_GREEN_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_MASK	0x20000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE	(1 << 29)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_BLUE_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_MASK	0x04000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE	(1 << 26)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_MASK	0x04000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE	(1 << 26)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHA_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_MASK	0x02000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE	(1 << 25)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_STENCIL_WRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAREF_MASK	0x000000ff
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_MASK	0x00000f00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_SHIFT	8
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_NEVER	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_LESS	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_EQUAL	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_LEQUAL	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_GREATER	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_NOTEQUAL	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_GEQUAL	0x0007
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAFUNC_ALWAYS	0x0008
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAENABLE_MASK	0x00001000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAENABLE	(1 << 12)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAENABLE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ALPHAENABLE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZENABLE_MASK	0x00004000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZENABLE	(1 << 14)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZENABLE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZENABLE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZWRITE_MASK	0x01000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZWRITE	(1 << 24)
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZWRITE_TRUE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZWRITE_FALSE	0x0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_MASK	0x000f0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_SHIFT	16
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_NEVER	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_LESS	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_EQUAL	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_LEQUAL	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_GREATER	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_NOTEQUAL	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_GEQUAL	0x0007
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_ZFUNC_ALWAYS	0x0008
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULLMODE_MASK	0x00300000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULLMODE_SHIFT	20
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULLMODE_NONE	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULLMODE_CW	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL0_CULLMODE_CCW	0x0003
 #	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1			0x00000340	/* Parameters: stencil_enable stencil_mask_write stencil_mask_read stencilref stencilfunc */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_ENABLE_MASK	0x00000001
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_MASK	0xff000000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_WRITE_SHIFT	24
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_MASK	0x00ff0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCIL_MASK_READ_SHIFT	16
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILREF_MASK	0x0000ff00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILREF_SHIFT	8
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_MASK	0x000000f0
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_SHIFT	4
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_NEVER	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_LESS	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_EQUAL	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_LEQUAL	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_GREATER	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_NOTEQUAL	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_GEQUAL	0x0007
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL1_STENCILFUNC_ALWAYS	0x0008
 #	define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2			0x00000344	/* Parameters: stencil_fail stencil_zfail stencil_zpass */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_FAIL_MASK	0x0000000f
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_MASK	0x000000f0
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_SHIFT	4
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_KEEP	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_ZERO	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_REPLACE	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_INCR	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_DECR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_INVERT	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_INCR_WRAP	0x0007
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZFAIL_DECR_WRAP	0x0008
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_MASK	0x00000f00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_SHIFT	8
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_KEEP	0x0001
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_ZERO	0x0002
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_REPLACE	0x0003
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_INCR	0x0004
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_DECR	0x0005
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_INVERT	0x0006
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_INCR_WRAP	0x0007
+#		 define NV04_DX6_MULTITEX_TRIANGLE_CONTROL2_STENCIL_ZPASS_DECR_WRAP	0x0008
 #	define		NV04_DX6_MULTITEX_TRIANGLE_FOG_COLOR			0x00000348
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_INV_W( d)			(0x0000040c + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_COLOR( d)			(0x00000410 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_SPECULAR( d)			(0x00000414 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_S( d)		(0x00000418 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_T( d)		(0x0000041c + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_S( d)		(0x00000420 + d * 0x0028)
-#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_T( d)		(0x00000424 + d * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SX( d)		(0x00000400 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SY( d)		(0x00000404 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TLVERTEX_SZ( d)		(0x00000408 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_INV_W( d)			(0x0000040c + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_COLOR( d)			(0x00000410 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_SPECULAR( d)			(0x00000414 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_S( d)		(0x00000418 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE0_T( d)		(0x0000041c + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_S( d)		(0x00000420 + (d) * 0x0028)
+#	define		NV04_DX6_MULTITEX_TRIANGLE_TEXTURE1_T( d)		(0x00000424 + (d) * 0x0028)
 #	define		NV04_DX6_MULTITEX_TRIANGLE_DRAW				0x00000540	/* Parameters: v0 v1 v2 v3 v4 v5 */
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V0_MASK	0x0000000f
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V1_MASK	0x000000f0
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V1_SHIFT	4
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V2_MASK	0x00000f00
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V2_SHIFT	8
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V3_MASK	0x0000f000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V3_SHIFT	12
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V4_MASK	0x000f0000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V4_SHIFT	16
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V5_MASK	0x00f00000
+#		 define		NV04_DX6_MULTITEX_TRIANGLE_DRAW_V5_SHIFT	20
 
 /****************************************** 
 Object NV04_COLOR_KEY used on: NV04 NV10 NV15 NV20 NV40 
@@ -207,7 +618,13 @@ Object NV04_SOLID_LINE used on: NV04
 #	define		NV04_SOLID_LINE_COLOR_FORMAT				0x00000300
 #	define		NV04_SOLID_LINE_COLOR_VALUE				0x00000304
 #	define		NV04_SOLID_LINE_START					0x00000400	/* Parameters: x y */
+#		 define		NV04_SOLID_LINE_START_X_MASK	0x0000ffff
+#		 define		NV04_SOLID_LINE_START_Y_MASK	0xffff0000
+#		 define		NV04_SOLID_LINE_START_Y_SHIFT	16
 #	define		NV04_SOLID_LINE_END					0x00000400	/* Parameters: x y */
+#		 define		NV04_SOLID_LINE_END_X_MASK	0x0000ffff
+#		 define		NV04_SOLID_LINE_END_Y_MASK	0xffff0000
+#		 define		NV04_SOLID_LINE_END_Y_SHIFT	16
 
 /****************************************** 
 Object NV04_UNK005E used on: NV04 
@@ -230,21 +647,64 @@ Object NV05_SCALED_IMAGE_FROM_MEMORY used on: NV04
 Object NV04_SCALED_IMAGE_FROM_MEMORY used on: NV04 
 */
 #define			NV04_SCALED_IMAGE_FROM_MEMORY				0x00000077
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_NOP			0x00000100
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_NOTIFY			0x00000104
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_DMA_NOTIFY		0x00000180
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_DMA_IMAGE			0x00000184
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_PATTERN			0x00000188
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_ROP			0x0000018c
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_BETA1			0x00000190
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_BETA4			0x00000194
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_SURFACE			0x00000198
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_COLOR_FORMAT		0x00000300
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_OPERATION			0x00000304
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POS			0x00000308	/* Parameters: x y */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_X_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_Y_MASK	0xffff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_Y_SHIFT	16
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE			0x0000030c	/* Parameters: width height */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_HEIGHT_SHIFT	16
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POS			0x00000310	/* Parameters: x y */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POS_X_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POS_Y_MASK	0xffff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_POS_Y_SHIFT	16
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE			0x00000314	/* Parameters: width height */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX			0x00000318	/* Parameters: int frac*0x100000 */
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY			0x0000031c	/* Parameters: int frac*0x100000 */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_HEIGHT_SHIFT	16
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX			0x00000318	/* Parameters: int frac_mul_0x100000 */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX_INT_MASK	0xfff00000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX_INT_SHIFT	20
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DU_DX_FRAC_MUL_0X100000_MASK	0x000fffff
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY			0x0000031c	/* Parameters: int frac_mul_0x100000 */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY_INT_MASK	0xfff00000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY_INT_SHIFT	20
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_DV_DY_FRAC_MUL_0X100000_MASK	0x000fffff
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_SIZE			0x00000400	/* Parameters: width height */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_SIZE_HEIGHT_SHIFT	16
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT			0x00000404	/* Parameters: pitch origin filter */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK	0x0000ffff
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_MASK	0x00ff0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_SHIFT	16
+#		 define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CENTER	0x0001
+#		 define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_ORIGIN_CORNER	0x0002
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_MASK	0xff000000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_SHIFT	24
+#		 define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_POINT_SAMPLE	0x0001
+#		 define NV04_SCALED_IMAGE_FROM_MEMORY_FORMAT_FILTER_BILINEAR	0x0002
 #	define		NV04_SCALED_IMAGE_FROM_MEMORY_OFFSET			0x00000408
-#	define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac*0x10 v_int v_frac*0x10 */
+#	define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac_mul_0x10 v_int v_frac_mul_0x10 */
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_U_INT_MASK	0xfff00000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_U_INT_SHIFT	20
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_U_FRAC_MUL_0X10_MASK	0x000f0000
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_U_FRAC_MUL_0X10_SHIFT	16
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_V_INT_MASK	0x0000fff0
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_V_INT_SHIFT	4
+#		 define		NV04_SCALED_IMAGE_FROM_MEMORY_POINT_V_FRAC_MUL_0X10_MASK	0x0000000f
 
 /****************************************** 
 Object NV_IMAGE_FROM_CPU used on: NV04 
@@ -266,43 +726,63 @@ Object NV05_IMAGE_FROM_CPU used on: NV04
 #	define		NV05_IMAGE_FROM_CPU_CLIP_RECTANGLE			0x00000188
 #	define		NV05_IMAGE_FROM_CPU_PATTERN				0x0000018c
 #	define		NV05_IMAGE_FROM_CPU_ROP					0x00000190
+#	define		NV05_IMAGE_FROM_CPU_BETA1				0x00000194
+#	define		NV05_IMAGE_FROM_CPU_BETA4				0x00000198
 #	define		NV05_IMAGE_FROM_CPU_SURFACE				0x0000019c
 #	define		NV05_IMAGE_FROM_CPU_OPERATION				0x000002fc
 #	define		NV05_IMAGE_FROM_CPU_FORMAT				0x00000300
 #	define		NV05_IMAGE_FROM_CPU_POINT				0x00000304	/* Parameters: x y */
+#		 define		NV05_IMAGE_FROM_CPU_POINT_X_MASK	0x0000ffff
+#		 define		NV05_IMAGE_FROM_CPU_POINT_Y_MASK	0xffff0000
+#		 define		NV05_IMAGE_FROM_CPU_POINT_Y_SHIFT	16
 #	define		NV05_IMAGE_FROM_CPU_SIZE_OUT				0x00000308	/* Parameters: x y */
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_OUT_X_MASK	0x0000ffff
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_OUT_Y_MASK	0xffff0000
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_OUT_Y_SHIFT	16
 #	define		NV05_IMAGE_FROM_CPU_SIZE_IN				0x0000030c	/* Parameters: x y */
-#	define		NV05_IMAGE_FROM_CPU_COLOR( d)				(0x00000400 + d * 0x0004)
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_IN_X_MASK	0x0000ffff
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_IN_Y_MASK	0xffff0000
+#		 define		NV05_IMAGE_FROM_CPU_SIZE_IN_Y_SHIFT	16
+#	define		NV05_IMAGE_FROM_CPU_COLOR( d)				(0x00000400 + (d) * 0x0004)
 
 /****************************************** 
-Object NV_IMAGE_BLIT used on: NV04 NV10 NV15 NV20 NV40 
+Object NV_IMAGE_BLIT used on: NV04 NV10 NV15 NV20 
 */
 #define			NV_IMAGE_BLIT						0x0000005f
-#	define		NV_IMAGE_BLIT_DMA_NOTIFY				0x00000180
-#	define		NV_IMAGE_BLIT_COLOR_KEY					0x00000184
-#	define		NV_IMAGE_BLIT_CLIP_RECTANGLE				0x00000188
-#	define		NV_IMAGE_BLIT_PATTERN					0x0000018c
-#	define		NV_IMAGE_BLIT_ROP5					0x00000190
-#	define		NV_IMAGE_BLIT_SURFACE					0x0000019c
+#	define		NV_IMAGE_BLIT_SET_DMA_NOTIFY				0x00000180
+#	define		NV_IMAGE_BLIT_SET_COLOR_KEY				0x00000184
+#	define		NV_IMAGE_BLIT_SET_CLIP_RECTANGLE			0x00000188
+#	define		NV_IMAGE_BLIT_SET_PATTERN				0x0000018c
+#	define		NV_IMAGE_BLIT_SET_ROP5					0x00000190
+#	define		NV_IMAGE_BLIT_SET_SURFACES_2D				0x0000019c
 #	define		NV_IMAGE_BLIT_OPERATION					0x000002fc
 #	define		NV_IMAGE_BLIT_POINT_IN					0x00000300	/* Parameters: x y */
+#		 define		NV_IMAGE_BLIT_POINT_IN_X_MASK	0x0000ffff
+#		 define		NV_IMAGE_BLIT_POINT_IN_Y_MASK	0xffff0000
+#		 define		NV_IMAGE_BLIT_POINT_IN_Y_SHIFT	16
 #	define		NV_IMAGE_BLIT_POINT_OUT					0x00000304	/* Parameters: x y */
+#		 define		NV_IMAGE_BLIT_POINT_OUT_X_MASK	0x0000ffff
+#		 define		NV_IMAGE_BLIT_POINT_OUT_Y_MASK	0xffff0000
+#		 define		NV_IMAGE_BLIT_POINT_OUT_Y_SHIFT	16
 #	define		NV_IMAGE_BLIT_SIZE					0x00000308	/* Parameters: width height */
+#		 define		NV_IMAGE_BLIT_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV_IMAGE_BLIT_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV_IMAGE_BLIT_SIZE_HEIGHT_SHIFT	16
 
 /****************************************** 
-Object NV10_TCL_PRIMITIVE_3D used on: NV10 
+Object NV11_IMAGE_BLIT used on: NV15 NV20 
 */
-#define			NV10_TCL_PRIMITIVE_3D					0x00000056
+#define			NV11_IMAGE_BLIT						0x0000009f
 
 /****************************************** 
-Object NV17_TCL_PRIMITIVE_3D used on: NV15 
+Object NV30_IMAGE_BLIT used on: NV30 NV40 G70
 */
-#define			NV17_TCL_PRIMITIVE_3D					0x00000099
+#define			NV30_IMAGE_BLIT						0x0000009f
 
 /****************************************** 
-Object NV11_TCL_PRIMITIVE_3D used on: NV15 
+Object NV10_TCL_PRIMITIVE_3D used on: NV10 
 */
-#define			NV11_TCL_PRIMITIVE_3D					0x00000096
+#define			NV10_TCL_PRIMITIVE_3D					0x00000056
 #	define		NV10_TCL_PRIMITIVE_3D_NOP				0x00000100
 #	define		NV10_TCL_PRIMITIVE_3D_NOTIFY				0x00000104
 #	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_NOTIFY			0x00000180
@@ -311,40 +791,687 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_SET_DISPLAY_LIST			0x0000018c
 #	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY2		0x00000194
 #	define		NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY3		0x00000198
-#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY4		0x000001ac
-#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY5		0x000001b0
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ			0x00000200	/* Parameters: width x */
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_WIDTH_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_WIDTH_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT			0x00000204	/* Parameters: height y */
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_HEIGHT_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_HEIGHT_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT			0x00000208	/* Parameters: type color */
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_pitch	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_swizzle	0x0002
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_FORMAT_COLOR_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color buffer pitch */
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_DEPTH_STENCIL_BUFFER_PITCH_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_DEPTH_STENCIL_BUFFER_PITCH_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_BUFFER_PITCH_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_OFFSET			0x00000210
 #	define		NV10_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
-#	define		NV10_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00000218 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00000220 + d * 0x0004)	/* Parameters: wrap_t wrap_s log2(height) log2(width) lod npot format cube_map */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00000228 + d * 0x0004)	/* Parameters: enable anisotropy */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00000230 + d * 0x0004)	/* Parameters: pitch */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00000240 + d * 0x0004)	/* Parameters: width height */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00000248 + d * 0x0004)	/* Parameters: mag_filter min_filter */
-#	define		NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00000250 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x000003e0 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x00000540 + y * 0x0010 + x * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000268 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000278 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000280 + d * 0x0004)	/* Parameters: rc1_tx_units_enabled rc1_rc_enabled scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#	define		NV10_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00000218 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00000220 + (d) * 0x0004)	/* Parameters: wrap_t wrap_s log2_height log2_width lod npot format cube_map */
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MASK	0xf0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_SHIFT	28
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_REPEAT	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_MIRRORED_REPEAT	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_EDGE	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP_TO_BORDER	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_T_CLAMP	0x0005
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_REPEAT	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_MIRRORED_REPEAT	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_EDGE	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP_TO_BORDER	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_WRAP_S_CLAMP	0x0005
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_HEIGHT_MASK	0x00f00000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_HEIGHT_SHIFT	20
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_WIDTH_MASK	0x000f0000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_WIDTH_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOD_MASK	0x0000f000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_LOD_SHIFT	12
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT_MASK	0x00000800
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT	(1 << 11)
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_NPOT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK	0x00000780
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT	7
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8	0x000b
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5	0x000f
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT	0x0010
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT	0x0012
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT	0x0013
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8	0x001a
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT_2	0x001b
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT	0x001d
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT	0x0020
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_MASK	0x00000004
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP	(1 << 2)
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_FALSE	0x0000
+#	define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00000228 + (d) * 0x0004)	/* Parameters: enable anisotropy */
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_MASK	0x40000000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE	(1 << 30)
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK	0x00000030
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT	4
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_1	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_2	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_4	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_8	0x0003
+#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00000230 + (d) * 0x0004)	/* Parameters: pitch */
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT	16
+#	define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00000240 + (d) * 0x0004)	/* Parameters: width height */
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_WIDTH_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_WIDTH_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_HEIGHT_MASK	0x0000ffff
+#	define		NV10_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00000248 + (d) * 0x0004)	/* Parameters: mag_filter min_filter */
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_MASK	0xf0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_SHIFT	28
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#	define		NV10_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00000250 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x000003e0 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x00000540 + (y) * 0x0010 + (x) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + (d) * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SHIFT	29
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SHIFT	21
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_MASK	0x000f0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SHIFT	13
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SHIFT	5
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_INPUT_MASK	0x0000000f
+#	define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000268 + (d) * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SHIFT	29
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SHIFT	21
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_MASK	0x000f0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SHIFT	13
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SHIFT	5
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_INPUT_MASK	0x0000000f
+#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000278 + (d) * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK	0x00030000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_MASK	0x00008000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS	(1 << 15)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_MASK	0x00004000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM	(1 << 14)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT	4
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK	0x0000000f
+#	define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000280 + (d) * 0x0004)	/* Parameters: rc1_tx_units_enabled rc1_rc_enabled scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_TX_UNITS_ENABLED_MASK	0x30000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_TX_UNITS_ENABLED_SHIFT	28
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_RC_ENABLED_MASK	0x08000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_RC_ENABLED	(1 << 27)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_RC_ENABLED_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_RC1_RC_ENABLED_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK	0x00030000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_MASK	0x00008000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS	(1 << 15)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_MASK	0x00004000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM	(1 << 14)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT	4
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0				0x00000270	/* Parameters: a r g b */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_R_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_R_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR0_B_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1				0x00000274	/* Parameters: a r g b */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_R_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_R_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1_B_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0				0x00000288	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SHIFT	29
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SHIFT	21
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_MASK	0x000f0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SHIFT	13
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SHIFT	5
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_INPUT_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1				0x0000028c	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_MASK	0xe0000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SHIFT	29
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE	(1 << 28)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_MASK	0x0f000000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_MASK	0x00e00000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SHIFT	21
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE	(1 << 20)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_MASK	0x000f0000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_MASK	0x0000e000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SHIFT	13
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_RGB	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_MASK	0x00000f00
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_ZERO	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_FOG	0x0003
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_NV	0x000c
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE1_NV	0x000d
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_MASK	0x00000080
+#		 define		NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP	(1 << 7)
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_FALSE	0x0000
 #	define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL			0x00000294	/* Parameters: local_viewer color_control */
+#		 define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER_MASK	0x00010000
+#		 define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER	(1 << 16)
+#		 define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_LOCAL_VIEWER_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL_MASK	0x00000002
+#		 define		NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL	1 // Nothing to shift
+#		 define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL_SINGLE_COLOR	0x0000
+#		 define NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL_COLOR_CONTROL_SEPARATE_SPECULAR_COLOR	0x0001
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE		0x00000298	/* Parameters: specular diffuse ambient emission */
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR_MASK	0x00000008
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR	(1 << 3)
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_SPECULAR_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE_MASK	0x00000004
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE	(1 << 2)
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_DIFFUSE_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT_MASK	0x00000002
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT	1 // Nothing to shift
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_AMBIENT_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE_EMISSION_MASK	0x00000001
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_MODE				0x0000029c
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000002a0
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_ENABLE			0x000002a4
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR				0x000002a8	/* Parameters: a b g r */
-#	define		NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE			0x000002bc
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + d * 0x0004)	/* Parameters: x2 x1 */
-#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + d * 0x0004)	/* Parameters: y2 y1 */
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK	0x000000ff
+#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE		0x000002b4
+#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + (d) * 0x0004)	/* Parameters: enable clip at x2 x2 enable clip at x1 x1 */
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X2_MASK	0x08000000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X2	(1 << 27)
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X2_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X2_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_MASK	0x07ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X1_MASK	0x00000800
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X1	(1 << 11)
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X1_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_ENABLE_CLIP_AT_X1_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X1_MASK	0x000007ff
+#	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + (d) * 0x0004)	/* Parameters: enable clip at y2 y2 enable clip at y1 y1 */
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y2_MASK	0x08000000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y2	(1 << 27)
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y2_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y2_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_MASK	0x07ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y1_MASK	0x00000800
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y1	(1 << 11)
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y1_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_ENABLE_CLIP_AT_Y1_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y1_MASK	0x000007ff
 #	define		NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000300
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000304
 #	define		NV10_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00000308
@@ -365,9 +1492,29 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000344
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000348
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000034c	/* Parameters: a r g b */
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000350
 #	define		NV10_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x00000354
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000358	/* Parameters: a r g b */
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT	24
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_A_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT	16
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_R_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT	8
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_COLOR_MASK_G_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x0000035c
 #	define		NV10_TCL_PRIMITIVE_3D_STENCIL_MASK			0x00000360
 #	define		NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC			0x00000364
@@ -392,19 +1539,58 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_B			0x000003b0
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_A			0x000003b4
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL			0x000003b8	/* Parameters: color_control */
+#		 define		NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL_COLOR_CONTROL_MASK	0x00000001
 #	define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x000003bc	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE( d)		(0x000003c0 + d * 0x0004)
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_MASK	0x00004000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7	(1 << 14)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_MASK	0x00001000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6	(1 << 12)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_MASK	0x00000400
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5	(1 << 10)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_MASK	0x00000100
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4	(1 << 8)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_MASK	0x00000040
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3	(1 << 6)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_MASK	0x00000010
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2	(1 << 4)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_MASK	0x00000004
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1	(1 << 2)
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0_MASK	0x00000001
+#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE( d)		(0x000003c0 + (d) * 0x0004)
 #	define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE		0x000003e8	/* Parameters: projection modelview0 modelview1 */
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION_MASK	0x00000004
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION	(1 << 2)
+#		 define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_PROJECTION_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0_MASK	0x00000002
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0	1 // Nothing to shift
+#		 define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0_TRUE	0x0001
+#		 define NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW0_FALSE	0x0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE_MODELVIEW1_MASK	0x00000001
 #	define		NV10_TCL_PRIMITIVE_3D_POINT_SIZE			0x000003ec
-#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX( d)		(0x00000400 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX( d)		(0x00000440 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX( d)	(0x00000480 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX( d)	(0x000004c0 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000500 + d * 0x0004)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000600 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000604 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000608 + d * 0x0010)
-#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000060c + d * 0x0010)
+#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX( d)		(0x00000400 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_MODELVIEW1_MATRIX( d)		(0x00000440 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW0_MATRIX( d)	(0x00000480 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW1_MATRIX( d)	(0x000004c0 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000500 + (d) * 0x0004)
+#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000600 + (d) * 0x0010)
+#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000604 + (d) * 0x0010)
+#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000608 + (d) * 0x0010)
+#	define		NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000060c + (d) * 0x0010)
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x00000680
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x00000684
 #	define		NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x00000688
@@ -429,34 +1615,34 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x0000070c
 #	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00000710
 #	define		NV10_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00000714
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00000800 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00000804 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00000808 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000080c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00000810 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00000814 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00000818 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000081c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00000820 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00000828 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000082c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00000830 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00000834 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00000838 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000083c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00000840 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00000844 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00000848 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000084c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00000850 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00000854 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00000858 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000085c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00000860 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00000864 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00000868 + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000086c + d * 0x0080)
-#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00000870 + d * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00000800 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00000804 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00000808 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000080c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00000810 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00000814 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00000818 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000081c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00000820 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00000828 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000082c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00000830 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00000834 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00000838 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000083c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00000840 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00000844 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00000848 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000084c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00000850 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00000854 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00000858 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000085c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00000860 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00000864 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00000868 + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000086c + (d) * 0x0080)
+#	define		NV10_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00000870 + (d) * 0x0080)
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00000c00
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00000c04
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00000c08
@@ -468,7 +1654,11 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00000c34
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00000c38
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000c40	/* Parameters: y x */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000c44	/* Parameters: z */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00000c50
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00000c54
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00000c58
@@ -477,67 +1667,164 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00000c64
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00000c68
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x00000c6c	/* Parameters: a b g r */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00000c80
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00000c84
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00000c88
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00000c8c	/* Parameters: a b g r */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK	0x00ff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK	0x000000ff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x00000c90
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x00000c94
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00000c98	/* Parameters: t s */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x00000ca0
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x00000ca4
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x00000ca8
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x00000cac
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x00000cb0	/* Parameters: t s */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x00000cb4	/* Parameters: q r */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x00000cb8
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x00000cbc
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x00000cc0	/* Parameters: t s */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x00000cc8
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x00000ccc
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x00000cd0
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x00000cd4
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x00000cd8	/* Parameters: t s */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x00000cdc	/* Parameters: q r */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00000ce0
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_WGH_1F			0x00000ce4
 #	define		NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE			0x00000cec
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00000d04 + d * 0x0008)	/* Parameters: stride fields type */
+#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00000d04 + (d) * 0x0008)	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE		0x00000cf0
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_POS		0x00000d00
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS		0x00000d04	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_POS_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL		0x00000d08
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL		0x00000d0c	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_COL2		0x00000d10
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2		0x00000d14	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_COL2_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX0		0x00000d18
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0		0x00000d1c	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX0_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_TX1		0x00000d20
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1		0x00000d24	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_TX1_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_NOR		0x00000d28
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR		0x00000d2c	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_NOR_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_WGH		0x00000d30
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH		0x00000d34	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_WGH_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_OFFSET_FOG		0x00000d38
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG		0x00000d3c	/* Parameters: stride fields type */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_MASK	0x0000ff00
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_STRIDE_SHIFT	8
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_MASK	0x000000f0
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_FIELDS_SHIFT	4
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_FORMAT_FOG_TYPE_MASK	0x0000000f
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x00000d40
 #	define		NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x00000d44
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x00000d5c	/* Parameters: pitch */
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000d60
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE		0x00000d68
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_CLEAR_ENABLE		0x00000d6c
 #	define		NV10_TCL_PRIMITIVE_3D_BEGIN_END				0x00000dfc
 #	define		NV10_TCL_PRIMITIVE_3D_INDEX_DATA			0x00000e00	/* Parameters: index1 index0 */
+#		 define		NV10_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX1_MASK	0xffff0000
+#		 define		NV10_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX1_SHIFT	16
+#		 define		NV10_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX0_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_BEGIN_END		0x000013fc
-#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS		0x00001400	/* Parameters: count-1 first */
+#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS		0x00001400	/* Parameters: count_minus_1 first */
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_COUNT_MINUS_1_MASK	0xff000000
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_COUNT_MINUS_1_SHIFT	24
+#		 define		NV10_TCL_PRIMITIVE_3D_VERTEX_BUFFER_DRAW_ARRAYS_FIRST_MASK	0x0000ffff
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X			0x00001638
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Y			0x0000163c
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Z			0x00001640
 #	define		NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_W			0x00001644
-#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE			0x00001658
 #	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_DATA			0x00001800
 
 /****************************************** 
+Object NV11_TCL_PRIMITIVE_3D used on: NV15 
+*/
+#define			NV11_TCL_PRIMITIVE_3D					0x00000096
+
+/****************************************** 
+Object NV17_TCL_PRIMITIVE_3D used on: NV15 
+*/
+#define			NV17_TCL_PRIMITIVE_3D					0x00000099
+#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY4		0x000001ac
+#	define		NV17_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY5		0x000001b0
+#	define		NV17_TCL_PRIMITIVE_3D_COLOR_MASK_ENABLE			0x000002bc
+#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x00000d5c	/* Parameters: pitch */
+#		 define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH_PITCH_MASK	0x0000ffff
+#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000d60
+#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_FILL_VALUE		0x00000d68
+#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_CLEAR_ENABLE		0x00000d6c
+#	define		NV17_TCL_PRIMITIVE_3D_LMA_DEPTH_ENABLE			0x00001658
+
+/****************************************** 
 Object NV10_IMAGE_FROM_CPU used on: NV10 NV15 NV20 NV30 NV40 G70
 */
 #define			NV10_IMAGE_FROM_CPU					0x0000008a
@@ -549,8 +1836,17 @@ Object NV10_IMAGE_FROM_CPU used on: NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV10_IMAGE_FROM_CPU_OPERATION				0x000002fc
 #	define		NV10_IMAGE_FROM_CPU_FORMAT				0x00000300
 #	define		NV10_IMAGE_FROM_CPU_POINT				0x00000304	/* Parameters: x y */
+#		 define		NV10_IMAGE_FROM_CPU_POINT_X_MASK	0x0000ffff
+#		 define		NV10_IMAGE_FROM_CPU_POINT_Y_MASK	0xffff0000
+#		 define		NV10_IMAGE_FROM_CPU_POINT_Y_SHIFT	16
 #	define		NV10_IMAGE_FROM_CPU_SIZE_OUT				0x00000308	/* Parameters: width height */
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_OUT_WIDTH_MASK	0x0000ffff
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_OUT_HEIGHT_MASK	0xffff0000
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_OUT_HEIGHT_SHIFT	16
 #	define		NV10_IMAGE_FROM_CPU_SIZE_IN				0x0000030c	/* Parameters: width height */
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_IN_WIDTH_MASK	0x0000ffff
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_IN_HEIGHT_MASK	0xffff0000
+#		 define		NV10_IMAGE_FROM_CPU_SIZE_IN_HEIGHT_SHIFT	16
 #	define		NV10_IMAGE_FROM_CPU_HLINE				0x00000400
 
 /****************************************** 
@@ -561,26 +1857,22 @@ Object NV10_PRIMITIVE_2D used on: NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV10_PRIMITIVE_2D_SET_SURFACE				0x00000184
 #	define		NV10_PRIMITIVE_2D_SET_FORMAT				0x00000300
 #	define		NV10_PRIMITIVE_2D_SET_POINT				0x00000304	/* Parameters: x y */
+#		 define		NV10_PRIMITIVE_2D_SET_POINT_X_MASK	0x0000ffff
+#		 define		NV10_PRIMITIVE_2D_SET_POINT_Y_MASK	0xffff0000
+#		 define		NV10_PRIMITIVE_2D_SET_POINT_Y_SHIFT	16
 #	define		NV10_PRIMITIVE_2D_SET_SIZE				0x00000308	/* Parameters: width height */
+#		 define		NV10_PRIMITIVE_2D_SET_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV10_PRIMITIVE_2D_SET_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV10_PRIMITIVE_2D_SET_SIZE_HEIGHT_SHIFT	16
 #	define		NV10_PRIMITIVE_2D_SET_CLIP_HORIZ			0x0000030c	/* Parameters: width x */
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_HORIZ_WIDTH_MASK	0xffff0000
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_HORIZ_WIDTH_SHIFT	16
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_HORIZ_X_MASK	0x0000ffff
 #	define		NV10_PRIMITIVE_2D_SET_CLIP_VERT				0x00000310	/* Parameters: height y */
-#	define		NV10_PRIMITIVE_2D_SET_DATA( d)				(0x00000400 + d * 0x0004)
-
-/****************************************** 
-Object NV10_IMAGE_BLIT used on: NV10 NV15 NV20 NV30 NV40 G70
-*/
-#define			NV10_IMAGE_BLIT						0x0000009f
-#	define		NV10_IMAGE_BLIT_NOP					0x00000100
-#	define		NV10_IMAGE_BLIT_NOTIFY					0x00000104
-#	define		NV10_IMAGE_BLIT_SET_DMA_NOTIFY				0x00000180
-#	define		NV10_IMAGE_BLIT_SET_CONTEXT_CLIP_RECTANGLE		0x00000188
-#	define		NV10_IMAGE_BLIT_SET_IMAGE_PATTERN			0x0000018c
-#	define		NV10_IMAGE_BLIT_SET_RASTER_OP				0x00000190
-#	define		NV10_IMAGE_BLIT_SET_CONTEXT_SURFACES_2D			0x0000019c
-#	define		NV10_IMAGE_BLIT_SET_OPERATION				0x000002fc
-#	define		NV10_IMAGE_BLIT_SET_POINT				0x00000300	/* Parameters: x y */
-#	define		NV10_IMAGE_BLIT_SET_PITCH				0x00000304	/* Parameters: skip */
-#	define		NV10_IMAGE_BLIT_SET_SIZE				0x00000308	/* Parameters: width height */
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_VERT_HEIGHT_MASK	0xffff0000
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_VERT_HEIGHT_SHIFT	16
+#		 define		NV10_PRIMITIVE_2D_SET_CLIP_VERT_Y_MASK	0x0000ffff
+#	define		NV10_PRIMITIVE_2D_SET_DATA( d)				(0x00000400 + (d) * 0x0004)
 
 /****************************************** 
 Object NV10_VIDEO_DISPLAY used on: NV10 NV15 NV20 NV30 NV40 G70
@@ -592,6 +1884,9 @@ Object NV10_VIDEO_DISPLAY used on: NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV10_VIDEO_DISPLAY_SET_DMA_IN_MEMORY1			0x00000188
 #	define		NV10_VIDEO_DISPLAY_SET_OBJECT3				0x0000019c
 #	define		NV10_VIDEO_DISPLAY_SIZE					0x000002f8	/* Parameters: height width */
+#		 define		NV10_VIDEO_DISPLAY_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV10_VIDEO_DISPLAY_SIZE_HEIGHT_SHIFT	16
+#		 define		NV10_VIDEO_DISPLAY_SIZE_WIDTH_MASK	0x0000ffff
 #	define		NV10_VIDEO_DISPLAY_OFFSET				0x00000300
 
 /****************************************** 
@@ -613,13 +1908,36 @@ Object NV10_SCALED_IMAGE_FROM_MEMORY used on: NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_SET_SURFACE		0x00000198
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_OPERATION			0x00000304
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_POS			0x00000308	/* Parameters: x y */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_X_MASK	0x0000ffff
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_Y_MASK	0xffff0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_POS_Y_SHIFT	16
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE			0x0000030c	/* Parameters: width height */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_CLIP_SIZE_HEIGHT_SHIFT	16
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_POS			0x00000310	/* Parameters: x y */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_POS_X_MASK	0x0000ffff
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_POS_Y_MASK	0xffff0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_POS_Y_SHIFT	16
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE			0x00000314	/* Parameters: width height */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_OUT_SIZE_HEIGHT_SHIFT	16
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_SIZE			0x00000400	/* Parameters: width height */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_SIZE_HEIGHT_SHIFT	16
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_FORMAT			0x00000404	/* Parameters: pitch */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_FORMAT_PITCH_MASK	0x0000ffff
 #	define		NV10_SCALED_IMAGE_FROM_MEMORY_OFFSET			0x00000408
-#	define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac*0x10 v_int v_frac*0x10 */
+#	define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT			0x0000040c	/* Parameters: u_int u_frac_mul_0x10 v_int v_frac_mul_0x10 */
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_U_INT_MASK	0xfff00000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_U_INT_SHIFT	20
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_U_FRAC_MUL_0X10_MASK	0x000f0000
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_U_FRAC_MUL_0X10_SHIFT	16
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_V_INT_MASK	0x0000fff0
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_V_INT_SHIFT	4
+#		 define		NV10_SCALED_IMAGE_FROM_MEMORY_POINT_V_FRAC_MUL_0X10_MASK	0x0000000f
 
 /****************************************** 
 Object NV10_CONTEXT_SURFACES_2D used on: NV10 NV15 NV20 NV30 NV40 G70
@@ -629,7 +1947,19 @@ Object NV10_CONTEXT_SURFACES_2D used on: NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY0		0x00000184
 #	define		NV10_CONTEXT_SURFACES_2D_SET_DMA_IN_MEMORY1		0x00000188
 #	define		NV10_CONTEXT_SURFACES_2D_FORMAT				0x00000300	/* Parameters: color type width height */
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_COLOR_MASK	0x000000ff
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_TYPE_MASK	0x0000ff80
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_TYPE_SHIFT	7
+#		 define NV10_CONTEXT_SURFACES_2D_FORMAT_TYPE_pitch	0x0001
+#		 define NV10_CONTEXT_SURFACES_2D_FORMAT_TYPE_swizzle	0x0002
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_WIDTH_MASK	0x00ff0000
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_WIDTH_SHIFT	16
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_HEIGHT_MASK	0xff000000
+#		 define		NV10_CONTEXT_SURFACES_2D_FORMAT_HEIGHT_SHIFT	24
 #	define		NV10_CONTEXT_SURFACES_2D_PITCH				0x00000304	/* Parameters: src dst */
+#		 define		NV10_CONTEXT_SURFACES_2D_PITCH_SRC_MASK	0x0000ffff
+#		 define		NV10_CONTEXT_SURFACES_2D_PITCH_DST_MASK	0xffff0000
+#		 define		NV10_CONTEXT_SURFACES_2D_PITCH_DST_SHIFT	16
 #	define		NV10_CONTEXT_SURFACES_2D_OFFSET_SRC			0x00000308
 #	define		NV10_CONTEXT_SURFACES_2D_OFFSET_DST			0x0000030c
 
@@ -643,6 +1973,9 @@ Object NV04_CONTEXT_SURFACES_2D used on: NV04 NV10 NV15
 #	define		NV04_CONTEXT_SURFACES_2D_SET_DMA_IMAGE_DST		0x00000188
 #	define		NV04_CONTEXT_SURFACES_2D_FORMAT				0x00000300
 #	define		NV04_CONTEXT_SURFACES_2D_PITCH				0x00000304	/* Parameters: src dst */
+#		 define		NV04_CONTEXT_SURFACES_2D_PITCH_SRC_MASK	0x0000ffff
+#		 define		NV04_CONTEXT_SURFACES_2D_PITCH_DST_MASK	0xffff0000
+#		 define		NV04_CONTEXT_SURFACES_2D_PITCH_DST_SHIFT	16
 #	define		NV04_CONTEXT_SURFACES_2D_OFFSET_SRC			0x00000308
 #	define		NV04_CONTEXT_SURFACES_2D_OFFSET_DST			0x0000030c
 
@@ -665,7 +1998,12 @@ Object NV20_SWIZZLED_SURFACE used on: NV20 NV30 NV40 G70
 #define			NV20_SWIZZLED_SURFACE					0x0000009e
 #	define		NV20_SWIZZLED_SURFACE_SET_OBJECT0			0x00000180
 #	define		NV20_SWIZZLED_SURFACE_SET_OBJECT1			0x00000184
-#	define		NV20_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2(height) log2(width) color */
+#	define		NV20_SWIZZLED_SURFACE_FORMAT				0x00000300	/* Parameters: log2_height log2_width color */
+#		 define		NV20_SWIZZLED_SURFACE_FORMAT_LOG2_HEIGHT_MASK	0xff000000
+#		 define		NV20_SWIZZLED_SURFACE_FORMAT_LOG2_HEIGHT_SHIFT	24
+#		 define		NV20_SWIZZLED_SURFACE_FORMAT_LOG2_WIDTH_MASK	0x00ff0000
+#		 define		NV20_SWIZZLED_SURFACE_FORMAT_LOG2_WIDTH_SHIFT	16
+#		 define		NV20_SWIZZLED_SURFACE_FORMAT_COLOR_MASK	0x0000ffff
 #	define		NV20_SWIZZLED_SURFACE_OFFSET				0x00000304
 
 /****************************************** 
@@ -686,20 +2024,79 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT9			0x000001ac
 #	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT10			0x000001b0
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ			0x00000200	/* Parameters: width x */
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_WIDTH_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_WIDTH_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ_X_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT			0x00000204	/* Parameters: height y */
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT_HEIGHT_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT_HEIGHT_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_VERT_Y_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT			0x00000208	/* Parameters: type color */
-#	define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color buffer pitch */
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_pitch	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT_TYPE_swizzle	0x0002
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_FORMAT_COLOR_MASK	0x000000ff
+#	define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH			0x0000020c	/* Parameters: zs_pitch color_pitch */
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZS_PITCH_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH_ZS_PITCH_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_BUFFER_PITCH_COLOR_PITCH_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_OFFSET			0x00000210
 #	define		NV20_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
 #	define		NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x0000022c	/* Parameters: pitch */
+#		 define		NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH_PITCH_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000230
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_CONTROL			0x00000294
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL		0x00000298	/* Parameters: back_specular back_ambient back_diffuse back_emission front_specular front_ambient front_diffuse front_emission */
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_SPECULAR_MASK	0x00004000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_SPECULAR	(1 << 14)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_SPECULAR_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_SPECULAR_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_AMBIENT_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_AMBIENT	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_AMBIENT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_AMBIENT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_DIFFUSE_MASK	0x00000400
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_DIFFUSE	(1 << 10)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_DIFFUSE_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_DIFFUSE_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_EMISSION_MASK	0x00000100
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_EMISSION	(1 << 8)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_EMISSION_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_BACK_EMISSION_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_SPECULAR_MASK	0x00000040
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_SPECULAR	(1 << 6)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_SPECULAR_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_SPECULAR_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_AMBIENT_MASK	0x00000010
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_AMBIENT	(1 << 4)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_AMBIENT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_AMBIENT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_DIFFUSE_MASK	0x00000004
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_DIFFUSE	(1 << 2)
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_DIFFUSE_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_DIFFUSE_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_CONTROL_FRONT_EMISSION_MASK	0x00000001
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_MODE				0x0000029c
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000002a0
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_ENABLE			0x000002a4
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + d * 0x0004)	/* Parameters: x2 x1 */
-#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + d * 0x0004)	/* Parameters: y2 y1 */
+#	define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR				0x000002a8	/* Parameters: a b g r */
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK	0x000000ff
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE		0x000002b4
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x000002c0 + (d) * 0x0004)	/* Parameters: x2 x1 */
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X1_MASK	0x0000ffff
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x000002e0 + (d) * 0x0004)	/* Parameters: y2 y1 */
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y1_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000300
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000304
 #	define		NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00000308
@@ -719,9 +2116,29 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000344
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000348
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000034c	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000350
 #	define		NV20_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x00000354
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000358	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_A_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_A_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_R_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_R_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_G_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_COLOR_MASK_G_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x0000035c
 #	define		NV20_TCL_PRIMITIVE_3D_STENCIL_MASK			0x00000360
 #	define		NV20_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC			0x00000364
@@ -747,17 +2164,49 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A		0x000003b4
 #	define		NV20_TCL_PRIMITIVE_3D_SEPARATE_SPECULAR_ENABLE		0x000003b8
 #	define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x000003bc	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x000003c0 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000420 + d * 0x0004)
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_MASK	0x00002000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7	(1 << 13)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_MASK	0x00008000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6	(1 << 15)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_MASK	0x00000200
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5	(1 << 9)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_MASK	0x00000800
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4	(1 << 11)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_MASK	0x00000020
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3	(1 << 5)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_MASK	0x00000080
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2	(1 << 7)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_MASK	0x00000002
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1	1 // Nothing to shift
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0_MASK	0x00000008
+#		 define		NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0	(1 << 3)
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0_FALSE	0x0000
+#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x000003c0 + (d) * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000420 + (d) * 0x0004)
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_SIZE			0x0000043c
-#	define		NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + d * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + y * 0x0010 + x * 0x0004)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000840 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000844 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000848 + d * 0x0010)
-#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000084c + d * 0x0010)
+#	define		NV20_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + (d) * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + (d) * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + (d) * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + (y) * 0x0010 + (x) * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000840 + (d) * 0x0010)
+#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000844 + (d) * 0x0010)
+#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000848 + (d) * 0x0010)
+#	define		NV20_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x0000084c + (d) * 0x0010)
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x000009c0
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x000009c4
 #	define		NV20_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x000009c8
@@ -768,6 +2217,16 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_E	0x000009f0
 #	define		NV20_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_F	0x000009f4
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE			0x00000a1c	/* Parameters: coord_replace r_mode enable */
+#		 define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_MASK	0x00000800
+#		 define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE	(1 << 11)
+#		 define NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_MASK	0x00000006
+#		 define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_SHIFT	1
+#		 define NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_R	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_S	0x0002
+#		 define		NV20_TCL_PRIMITIVE_3D_POINT_SPRITE_ENABLE_MASK	0x00000001
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_OX			0x00000a20
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_OY			0x00000a24
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_DEPTH_AVG_S		0x00000a28
@@ -796,59 +2255,754 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x00000a10
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x00000a14
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x00000a18
-#	define		NV20_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00001b00 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00001b04 + d * 0x0040)	/* Parameters: log2(height) log2(width) lod format cube_map */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_WRAP(d)			(0x00001b08 + d * 0x0040)	/* Parameters: wrap_s wrap_t wrap_r */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00001b0c + d * 0x0040)	/* Parameters: enable anisotropy */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00001b10 + d * 0x0040)	/* Parameters: pitch */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00001b14 + d * 0x0040)	/* Parameters: mag_filter min_filter */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00001b1c + d * 0x0040)	/* Parameters: width height */
-#	define		NV20_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00001b20 + d * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_TX_OFFSET(d)			(0x00001b00 + (d) * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT(d)			(0x00001b04 + (d) * 0x0040)	/* Parameters: log2_height log2_width lod format cube_map */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_HEIGHT_MASK	0x0f000000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_HEIGHT_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_WIDTH_MASK	0x00f00000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOG2_WIDTH_SHIFT	20
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOD_MASK	0x000f0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_LOD_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_INDEX8	0x000b
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT1	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT3	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_DXT5	0x000f
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R5G5B5A1_RECT	0x0010
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R8G8B8A8_RECT	0x0012
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8_RECT	0x0013
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8	0x001a
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_A8_RECT_2	0x001b
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_R4G4B4A4_RECT	0x001d
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_FORMAT_L8A8_RECT	0x0020
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_MASK	0x00000004
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP	(1 << 2)
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FORMAT_CUBE_MAP_FALSE	0x0000
+#	define		NV20_TCL_PRIMITIVE_3D_TX_WRAP(d)			(0x00001b08 + (d) * 0x0040)	/* Parameters: wrap_s wrap_t wrap_r */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_S_MASK	0x000000ff
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_REPEAT	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_MIRRORED_REPEAT	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_CLAMP_TO_EDGE	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_CLAMP_TO_BORDER	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_T_CLAMP	0x0005
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_REPEAT	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_MIRRORED_REPEAT	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_CLAMP_TO_EDGE	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_CLAMP_TO_BORDER	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_WRAP_WRAP_R_CLAMP	0x0005
+#	define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE(d)			(0x00001b0c + (d) * 0x0040)	/* Parameters: enable anisotropy */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_MASK	0x40000000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE	(1 << 30)
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ENABLE_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_MASK	0x00000030
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_SHIFT	4
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_1	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_2	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_4	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_ENABLE_ANISOTROPY_8	0x0003
+#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_PITCH(d)			(0x00001b10 + (d) * 0x0040)	/* Parameters: pitch */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_PITCH_PITCH_SHIFT	16
+#	define		NV20_TCL_PRIMITIVE_3D_TX_FILTER(d)			(0x00001b14 + (d) * 0x0040)	/* Parameters: mag_filter min_filter */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MAG_FILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_FILTER_MIN_FILTER_LINEAR_MIPMAP_LINEAR	0x0006
+#	define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE(d)			(0x00001b1c + (d) * 0x0040)	/* Parameters: width height */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_WIDTH_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_WIDTH_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_NPOT_SIZE_HEIGHT_MASK	0x0000ffff
+#	define		NV20_TCL_PRIMITIVE_3D_TX_PALETTE_OFFSET(d)		(0x00001b20 + (d) * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR(d)		(0x00001b24 + (d) * 0x0040)	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_A_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_R_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_G_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_B_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_RC_ENABLE				0x00001e60	/* Parameters: number of rc enabled */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_ENABLE_NUMBER_OF_RC_ENABLED_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP			0x00001e70	/* Parameters: op0 op1 op2 op3 */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP0_MASK	0x0000001f
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_MASK	0x000003e0
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_SHIFT	5
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_TEXTURE_2D	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_PASS_THROUGH	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_CULL_FRAGMENT	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_DOT_PRODUCT_TEXTURE_2D	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_DEPENDANT_AR	0x000f
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_DEPENDANT_GB	0x0010
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP1_DOT_PRODUCT	0x0011
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_MASK	0x00007c00
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_SHIFT	10
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_TEXTURE_2D	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_PASS_THROUGH	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_CULL_FRAGMENT	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_DOT_PRODUCT_TEXTURE_2D	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_DEPENDANT_AR	0x000f
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_DEPENDANT_GB	0x0010
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP2_DOT_PRODUCT	0x0011
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_MASK	0x000f8000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_SHIFT	15
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_TEXTURE_2D	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_PASS_THROUGH	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_CULL_FRAGMENT	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_DOT_PRODUCT_TEXTURE_2D	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_DEPENDANT_AR	0x000f
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_DEPENDANT_GB	0x0010
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP_OP3_DOT_PRODUCT	0x0011
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE		0x000017f8	/* Parameters: cull0 cull1 cull2 cull3 */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL0_MASK	0x0000000f
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_SHIFT	4
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_gggg	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_lggg	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_glgg	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_llgg	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_gglg	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_lglg	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_gllg	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_lllg	0x0007
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_gggl	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_lggl	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_glgl	0x000a
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_llgl	0x000b
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_ggll	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_lgll	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_glll	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL1_llll	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_gggg	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_lggg	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_glgg	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_llgg	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_gglg	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_lglg	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_gllg	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_lllg	0x0007
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_gggl	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_lggl	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_glgl	0x000a
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_llgl	0x000b
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_ggll	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_lgll	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_glll	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL2_llll	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_MASK	0x0000f000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_SHIFT	12
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_gggg	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_lggg	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_glgg	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_llgg	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_gglg	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_lglg	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_gllg	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_lllg	0x0007
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_gggl	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_lggl	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_glgl	0x000a
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_llgl	0x000b
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_ggll	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_lgll	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_glll	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE_CULL3_llll	0x000f
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS		0x00001e78	/* Parameters: prev2 prev3 */
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS_PREV2_MASK	0x00030000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS_PREV2_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS_PREV3_MASK	0x00300000
+#		 define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS_PREV3_SHIFT	20
 #	define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0				0x00001e20	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_R_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR0_B_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1				0x00001e24	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_R_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_COLOR1_B_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0				0x00000288	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SHIFT	29
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_MASK	0x0f000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SHIFT	21
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_MASK	0x000f0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SHIFT	13
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SHIFT	5
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_INPUT_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1				0x0000028c	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000ac0 + d * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000a60 + d * 0x0004)	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x00000a80 + d * 0x0004)	/* Parameters: a r g b */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000aa0 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00001e40 + d * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000105c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001060 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001064 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_AMBIENT(d)	(0x00000c00 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_DIFFUSE(d)	(0x00000c0c + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_SPECULAR(d)	(0x00000c18 + d * 0x0040)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001068 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000106c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001070 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001040 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001044 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001048 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000104c + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001050 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001054 + d * 0x0080)
-#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001058 + d * 0x0080)
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_MASK	0xe0000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SHIFT	29
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE	(1 << 28)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_MASK	0x0f000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_MASK	0x00e00000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SHIFT	21
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE	(1 << 20)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_MASK	0x000f0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_MASK	0x0000e000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SHIFT	13
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_MASK	0x00000080
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP	(1 << 7)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_FALSE	0x0000
+#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000260 + (d) * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SHIFT	29
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_MASK	0x0f000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SHIFT	21
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_MASK	0x000f0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SHIFT	13
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SHIFT	5
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_INPUT_MASK	0x0000000f
+#	define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000ac0 + (d) * 0x0004)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SHIFT	29
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_MASK	0x0f000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SHIFT	24
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SHIFT	21
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_MASK	0x000f0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SHIFT	13
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SHIFT	5
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_INPUT_MASK	0x0000000f
+#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000a60 + (d) * 0x0004)	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_R_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_B_MASK	0x000000ff
+#	define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x00000a80 + (d) * 0x0004)	/* Parameters: a r g b */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_R_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_R_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_B_MASK	0x000000ff
+#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000aa0 + (d) * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK	0x00030000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_MASK	0x00008000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS	(1 << 15)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_MASK	0x00004000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM	(1 << 14)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT	4
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK	0x0000000f
+#	define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00001e40 + (d) * 0x0004)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK	0x00030000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT	16
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_MASK	0x00008000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS	(1 << 15)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_MASK	0x00004000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM	(1 << 14)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT	8
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT	4
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO	0x0000
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG	0x0003
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV20_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK	0x0000000f
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000105c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001060 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001064 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_AMBIENT(d)	(0x00000c00 + (d) * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_DIFFUSE(d)	(0x00000c0c + (d) * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_BACK_SIDE_PRODUCT_SPECULAR(d)	(0x00000c18 + (d) * 0x0040)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001068 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000106c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001070 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001040 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001044 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001048 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000104c + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001050 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001054 + (d) * 0x0080)
+#	define		NV20_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001058 + (d) * 0x0080)
 #	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_A		0x00001e28
 #	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_B		0x00001e2c
 #	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_C		0x00001e30
@@ -856,7 +3010,7 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_E		0x00001e38
 #	define		NV20_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_F		0x00001e3c
 #	define		NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000147c
-#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(d)	(0x00001480 + d * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(d)	(0x00001480 + (d) * 0x0004)
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00001500
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00001504
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00001508
@@ -865,12 +3019,22 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z			0x00001520
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W			0x00001524
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY			0x00001528	/* Parameters: y x */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW			0x0000152c	/* Parameters: w z */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X			0x00001530
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00001534
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00001538
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00001540	/* Parameters: y x */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00001544	/* Parameters: z */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00001550
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00001554
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00001558
@@ -879,81 +3043,291 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00001564
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00001568
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000156c	/* Parameters: a b g r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00001580
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00001584
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00001588
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x0000158c	/* Parameters: a b g r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK	0x00ff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK	0x000000ff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x00001590
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x00001594
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00001598	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x000015a0
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x000015a4
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x000015a8
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x000015ac
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x000015b0	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x000015b4	/* Parameters: q r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000015b8
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000015bc
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x000015c0	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x000015c8
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x000015cc
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x000015d0
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x000015d4
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x000015d8	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x000015dc	/* Parameters: q r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000015e0
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000015e4
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x000015e8	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S			0x000015f0
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T			0x000015f4
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R			0x000015f8
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q			0x000015fc
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x00001600	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x00001604	/* Parameters: q r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x00001608
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x0000160c
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x00001610	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S			0x00001620
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T			0x00001624
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R			0x00001628
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q			0x0000162c
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x00001630	/* Parameters: t s */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x00001634	/* Parameters: q r */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00001698
 #	define		NV20_TCL_PRIMITIVE_3D_EDGE_FLAG				0x000016bc
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS		0x00001720	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH		0x00001724	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR		0x00001728	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL		0x0000172c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2		0x00001730	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG		0x00001734	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6			0x00001738	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7			0x0000173c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0		0x00001740	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1		0x00001744	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2		0x00001748	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3		0x0000174c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4		0x00001750	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5		0x00001754	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6		0x00001758	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7		0x0000175c	/* Parameters: enabled? offset */
-#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00001760 + d * 0x0004)
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS		0x00001720	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR0_POS_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH		0x00001724	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR1_WGH_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR		0x00001728	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR2_NOR_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL		0x0000172c	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR3_COL_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2		0x00001730	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR4_COL2_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG		0x00001734	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR5_FOG_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6			0x00001738	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR6_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7			0x0000173c	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR7_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0		0x00001740	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1		0x00001744	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR9_TX1_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2		0x00001748	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR10_TX2_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3		0x0000174c	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR11_TX3_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4		0x00001750	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR12_TX4_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5		0x00001754	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR13_TX5_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6		0x00001758	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR14_TX6_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7		0x0000175c	/* Parameters: enabled offset */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7_ENABLED_MASK	0x80000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7_ENABLED	(1 << 31)
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7_ENABLED_TRUE	0x0001
+#		 define NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7_ENABLED_FALSE	0x0000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR15_TX7_OFFSET_MASK	0x1fffffff
+#	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00001760 + (d) * 0x0004)
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS			0x00001760	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH			0x00001764	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR			0x00001768	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL			0x0000176c	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2			0x00001770	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG			0x00001774	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6			0x00001778	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR6_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7			0x0000177c	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR7_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0			0x00001780	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1			0x00001784	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2			0x00001788	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3			0x0000178c	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4			0x00001790	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5			0x00001794	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6			0x00001798	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7			0x0000179c	/* Parameters: stride fields type */
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7_STRIDE_MASK	0x0000ff00
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7_STRIDE_SHIFT	8
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7_FIELDS_MASK	0x000000f0
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7_FIELDS_SHIFT	4
+#		 define		NV20_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7_TYPE_MASK	0x0000000f
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000017a0
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000017a4
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000017a8
@@ -966,12 +3340,35 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_LIGHT_MODEL_TWO_SIDE_ENABLE	0x000017c4
 #	define		NV20_TCL_PRIMITIVE_3D_BEGIN_END				0x000017fc
 #	define		NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1			0x00001c30	/* Parameters: x2 x1 */
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1_X2_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1_X2_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1_X1_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1			0x00001c50	/* Parameters: y2 y1 */
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1_Y2_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1_Y2_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_SCISSOR_Y2_Y1_Y1_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH			0x00001d8c
 #	define		NV20_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB			0x00001d90
 #	define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS		0x00001d94	/* Parameters: clear color a clear color b clear color g clear color r clear depth clear stencil */
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_A_MASK	0x00000080
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_A	(1 << 7)
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_B_MASK	0x00000040
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_B	(1 << 6)
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_G_MASK	0x00000020
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_G	(1 << 5)
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_R_MASK	0x00000010
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_COLOR_R	(1 << 4)
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_DEPTH_MASK	0x00000002
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_DEPTH	1 // Nothing to shift
+#		 define		NV20_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS_CLEAR_STENCIL_MASK	0x00000001
 #	define		NV20_TCL_PRIMITIVE_3D_INDEX_DATA			0x00001800	/* Parameters: index1 index0 */
+#		 define		NV20_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX1_MASK	0xffff0000
+#		 define		NV20_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX1_SHIFT	16
+#		 define		NV20_TCL_PRIMITIVE_3D_INDEX_DATA_INDEX0_MASK	0x0000ffff
 #	define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001810	/* Parameters: count_vertices offset_vertices */
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_COUNT_VERTICES_MASK	0xff000000
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_COUNT_VERTICES_SHIFT	24
+#		 define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_OFFSET_VERTICES_MASK	0x00ffffff
 #	define		NV20_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001818
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X			0x00001f00
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Y			0x00001f04
@@ -979,7 +3376,7 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_W			0x00001f0c
 
 /****************************************** 
-Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
+Object NV30_TCL_PRIMITIVE_3D used on: NV30 
 */
 #define			NV30_TCL_PRIMITIVE_3D					0x00000097
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT0			0x00000180
@@ -988,25 +3385,61 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT3			0x0000018c
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT4			0x00000194
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT5			0x00000198
+#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC0_OBJECT		0x0000019c
+#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC1_OBJECT		0x000001a0
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT6			0x000001a4
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT7			0x000001a8
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT8			0x000001ac
+#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT8B			0x000001b0
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT9			0x000001b4
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT10			0x000001b8
-#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC0_OBJECT		0x0000019c
-#	define		NV30_TCL_PRIMITIVE_3D_SET_VB_SRC1_OBJECT		0x000001a0
-#	define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH			0x0000020c	/* Parameters: depth/stencil buffer pitch color0 buffer pitch */
+#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0	0x00000200	/* Parameters: width x_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0_WIDTH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0_WIDTH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0_X_OFFSET_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1	0x00000204	/* Parameters: height y_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1_HEIGHT_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1_HEIGHT_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1_Y_OFFSET_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH			0x0000020c	/* Parameters: zs_pitch color0_pitch */
+#		 define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH_ZS_PITCH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH_ZS_PITCH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_BUFFER0_PITCH_COLOR0_PITCH_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_COLOR0_OFFSET			0x00000210
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
 #	define		NV30_TCL_PRIMITIVE_3D_COLOR1_OFFSET			0x00000218
 #	define		NV30_TCL_PRIMITIVE_3D_BUFFER1_PITCH			0x0000021c	/* Parameters: color1 buffer pitch */
+#		 define		NV30_TCL_PRIMITIVE_3D_BUFFER1_PITCH_COLOR1_BUFFER_PITCH_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS			0x00000220	/* Parameters: BUF0 BUF1 BUF2 BUF3 */
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF0_MASK	0x00000001
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF1_MASK	0x00000002
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF1	1 // Nothing to shift
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF2_MASK	0x00000004
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF2	(1 << 2)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF2_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF2_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF3_MASK	0x00000008
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF3	(1 << 3)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF3_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_BUFFERS_BUF3_FALSE	0x0000
 #	define		NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH		0x0000022c	/* Parameters: pitch */
+#		 define		NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH_PITCH_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_OFFSET		0x00000230
-#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000240 + d * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX_ENABLE(d)		(0x00000240 + (d) * 0x0004)
 #	define		NV30_TCL_PRIMITIVE_3D_BUFFER2_PITCH			0x00000280
 #	define		NV30_TCL_PRIMITIVE_3D_BUFFER3_PITCH			0x00000284
 #	define		NV30_TCL_PRIMITIVE_3D_BUFFER2_OFFSET			0x00000288
 #	define		NV30_TCL_PRIMITIVE_3D_BUFFER3_OFFSET			0x0000028c
+#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0	0x000002c0	/* Parameters: width x_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0_WIDTH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0_WIDTH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0_X_OFFSET_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1	0x000002c4	/* Parameters: height y_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1_HEIGHT_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1_HEIGHT_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1_Y_OFFSET_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_DITHER_ENABLE			0x00000300
 #	define		NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x00000304
 #	define		NV30_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x00000308
@@ -1015,8 +3448,28 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC			0x00000314
 #	define		NV30_TCL_PRIMITIVE_3D_BLEND_FUNC_DST			0x00000318
 #	define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000031c	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_BLEND_COLOR_B_MASK	0x000000ff
 #	define		NV30_TCL_PRIMITIVE_3D_BLEND_EQUATION			0x00000320
 #	define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000324	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_A_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_A_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_R_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_R_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_R_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_G_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_COLOR_MASK_G_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK	0x000000ff
 #	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE		0x00000328
 #	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK			0x0000032c
 #	define		NV30_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC		0x00000330
@@ -1035,8 +3488,16 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS		0x00000364
 #	define		NV30_TCL_PRIMITIVE_3D_SHADE_MODEL			0x00000368
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_ENABLE			0x0000036c
-#	define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR				0x00000370
-#	define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123		0x00000370	/* Parameters: buffer3 b buffer3 g buffer3 r buffer3 a buffer2 b buffer2 g buffer2 r buffer2 a buffer1 b buffer1 g buffer1 r buffer1 a */
+#	define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR				0x00000370	/* Parameters: a b g r */
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_B_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_B_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_FOG_COLOR_R_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x00000374
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x00000378
 #	define		NV30_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x0000037c
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000394
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000398
@@ -1046,50 +3507,563 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_A		0x000003b4
 #	define		NV30_TCL_PRIMITIVE_3D_LINE_WIDTH_SMOOTH			0x000003b8
 #	define		NV30_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x000003bc
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x00000400 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + y * 0x0010 + x * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM			0x000008e4
+#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(d)		(0x00000400 + (d) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_MODELVIEW_MATRIX( d)		(0x00000480 + (d) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_INVERSE_MODELVIEW_MATRIX( d)	(0x00000580 + (d) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_PROJECTION_MATRIX( d)		(0x00000680 + (d) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_TX_MATRIX(x,y)			(0x000006c0 + (y) * 0x0010 + (x) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS		0x000008c0	/* Parameters: width x_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_WIDTH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_WIDTH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_X_OFFSET_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS		0x000008c4	/* Parameters: height y_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_HEIGHT_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_HEIGHT_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_Y_OFFSET_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_COORD_DIST			0x000008c8
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_MODE				0x000008cc
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT		0x000008d0
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR		0x000008d4
 #	define		NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC		0x000008d8
+#	define		NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM			0x000008e4
 #	define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0				0x000008ec	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR0_B_MASK	0x000000ff
 #	define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1				0x000008f0	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1_B_MASK	0x000000ff
 #	define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0				0x000008f4	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SHIFT	29
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_MASK	0x0f000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SHIFT	21
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SHIFT	13
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SHIFT	5
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL0_VARD_INPUT_MASK	0x0000000f
 #	define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1				0x000008f8	/* Parameters: vare_mapping vare_component_usage vare_input varf_mapping varf_component_usage varf_input varg_mapping varg_component_usage varg_input color_sum_clamp */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_MASK	0xe0000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SHIFT	29
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE	(1 << 28)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_MASK	0x0f000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARE_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_MASK	0x00e00000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SHIFT	21
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE	(1 << 20)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARF_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_MASK	0x0000e000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SHIFT	13
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_VARG_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_MASK	0x00000080
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP	(1 << 7)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_FINAL1_COLOR_SUM_CLAMP_FALSE	0x0000
 #	define		NV30_TCL_PRIMITIVE_3D_RC_ENABLE				0x000008fc	/* Parameters: number of rc enabled */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000900 + d * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000904 + d * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000908 + d * 0x0020)	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x0000090c + d * 0x0020)	/* Parameters: a r g b */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000910 + d * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000914 + d * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0	0x00000200	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM1	0x00000204	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS0	0x000002c0	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_OFS1	0x000002c4	/* Parameters: height y_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_ENABLE_NUMBER_OF_RC_ENABLED_MASK	0x0000000f
+#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA(d)			(0x00000900 + (d) * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SHIFT	29
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_MASK	0x0f000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SHIFT	21
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SHIFT	13
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SHIFT	5
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_BLUE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_ALPHA_VARD_INPUT_MASK	0x0000000f
+#	define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB(d)			(0x00000904 + (d) * 0x0020)	/* Parameters: vara_mapping vara_component_usage vara_input varb_mapping varb_component_usage varb_input varc_mapping varc_component_usage varc_input vard_mapping vard_component_usage vard_input */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_MASK	0xe0000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SHIFT	29
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_MASK	0x10000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE	(1 << 28)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_MASK	0x0f000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARA_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_MASK	0x00e00000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SHIFT	21
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_MASK	0x00100000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE	(1 << 20)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARB_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_MASK	0x0000e000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SHIFT	13
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARC_INPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_MASK	0x000000e0
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SHIFT	5
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_IDENTITY_NV	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_UNSIGNED_INVERT_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NORMAL_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_EXPAND_NEGATE_NV	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NORMAL_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_HALF_BIAS_NEGATE_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_IDENTITY_NV	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_MAPPING_SIGNED_NEGATE_NV	0x0007
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_MASK	0x00000010
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE	(1 << 4)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_RGB	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_COMPONENT_USAGE_ALPHA	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_IN_RGB_VARD_INPUT_MASK	0x0000000f
+#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0(d)		(0x00000908 + (d) * 0x0020)	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR0_B_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1(d)		(0x0000090c + (d) * 0x0020)	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_CONSTANT_COLOR1_B_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA(d)			(0x00000910 + (d) * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_MASK	0x00030000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_NONE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_MASK	0x00008000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS	(1 << 15)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_NONE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_MASK	0x00004000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM	(1 << 14)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_MUX_SUM_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SHIFT	4
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_ALPHA_CD_OUTPUT_MASK	0x0000000f
+#	define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB(d)			(0x00000914 + (d) * 0x0020)	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_MASK	0x00030000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_NONE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_TWO_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_FOUR_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SCALE_SCALE_BY_ONE_HALF_NV	0x0003
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_MASK	0x00008000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS	(1 << 15)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_NONE	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_BIAS_BIAS_BY_NEGATIVE_ONE_HALF_NV	0x0001
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_MASK	0x00004000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM	(1 << 14)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_MUX_SUM_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_MASK	0x00002000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT	(1 << 13)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_DOT_PRODUCT_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_MASK	0x00000f00
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_SUM_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_MASK	0x000000f0
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SHIFT	4
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR0_NV	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_CONSTANT_COLOR1_NV	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_FOG	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_PRIMARY_COLOR_NV	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SECONDARY_COLOR_NV	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE1_ARB	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_TEXTURE0_ARB	0x0009
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_NV	0x000c
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE1_NV	0x000d
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_SPARE0_PLUS_SECONDARY_COLOR_NV	0x000e
+#		 define NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_AB_OUTPUT_E_TIMES_F_NV	0x000f
+#		 define		NV30_TCL_PRIMITIVE_3D_RC_OUT_RGB_CD_OUTPUT_MASK	0x0000000f
 #	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0			0x00000a00	/* Parameters: width x_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_WIDTH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_WIDTH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_X_OFFSET_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1			0x00000a04	/* Parameters: height y_offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_HEIGHT_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_HEIGHT_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_Y_OFFSET_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x00000a10
 #	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x00000a14
 #	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_FRONT_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x00000a18
-#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS		0x000008c0	/* Parameters: width x_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS		0x000008c4	/* Parameters: height y_offset */
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE			0x00001ee8	/* Parameters: coord_replace r_mode enable */
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_SIZE			0x00001ee0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x00001ec0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x00001ec4
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00001ec8
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_D			0x00001ecc
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_E			0x00001ed0
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x00001ed4
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00001ed8
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00001edc
-#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE		0x00001ee4
 #	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_OX			0x00000a20
 #	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_OY			0x00000a24
 #	define		NV30_TCL_PRIMITIVE_3D_VIEWPORT_XFRM_NPF_DIV2		0x00000a28
@@ -1106,48 +4080,48 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x00000a74
 #	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x00000a78
 #	define		NV30_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x00000a7c
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000a90	/* Parameters: y x */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000a94	/* Parameters: z */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK	0x0000ffff
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0			0x00000b80
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST1			0x00000b84
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST2			0x00000b88
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_INST3			0x00000b8c
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R		0x000017b0
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G		0x000017b4
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B		0x000017b8
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A		0x000017c0
-#	define		NV30_TCL_PRIMITIVE_3D_OCC_QUERY_OR_COLOR_BUFF_ENABLE	0x000017c8
-#	define		NV30_TCL_PRIMITIVE_3D_STORE_RESULT			0x00001800
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000e00 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000e04 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000e08 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x00000e0c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + d * 0x0080)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001228 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000122c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001230 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001200 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001204 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001208 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000120c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001210 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001214 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001218 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000121c + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001220 + d * 0x0040)
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001224 + d * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_A(d)			(0x00000e00 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_B(d)			(0x00000e04 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_C(d)			(0x00000e08 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_D(d)			(0x00000e0c + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_R(d)	(0x00001000 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_G(d)	(0x00001004 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_AMBIENT_B(d)	(0x00001008 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_R(d)	(0x0000100c + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_G(d)	(0x00001010 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_DIFFUSE_B(d)	(0x00001014 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_R(d)	(0x00001018 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_G(d)	(0x0000101c + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_FRONT_SIDE_PRODUCT_SPECULAR_B(d)	(0x00001020 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_X(d)		(0x00001028 + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Y(d)		(0x0000102c + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_HALF_VECTOR_Z(d)		(0x00001030 + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_X(d)		(0x00001034 + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Y(d)		(0x00001038 + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_DIRECTION_Z(d)		(0x0000103c + (d) * 0x0080)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_A(d)		(0x00001200 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_EXPONENT(d)		(0x00001204 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_B(d)		(0x00001208 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_X(d)		(0x0000120c + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Y(d)		(0x00001210 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_DIR_Z(d)		(0x00001214 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_SPOT_CUTOFF_C(d)		(0x00001218 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_X(d)		(0x0000121c + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Y(d)		(0x00001220 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_POSITION_Z(d)		(0x00001224 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_CONSTANT_ATTENUATION(d)	(0x00001228 + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_LINEAR_ATTENUATION(d)	(0x0000122c + (d) * 0x0040)
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_QUADRATIC_ATTENUATION(d)	(0x00001230 + (d) * 0x0040)
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_A	0x00001400
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_B	0x00001404
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_C	0x00001408
@@ -1155,27 +4129,303 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_E	0x00001410
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_MATERIAL_SHININESS_F	0x00001414
 #	define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS			0x00001420	/* Parameters: light 7 light 6 light 5 light 4 light 3 light 2 light 1 light 0 */
-#	define		NV30_TCL_PRIMITIVE_3D_UNK1D6C_OFFSET			0x00001d6c
-#	define		NV30_TCL_PRIMITIVE_3D_UNK1D70_VALUE			0x00001d70
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE		0x00001db4
-#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN		0x00001db8	/* Parameters: factor pattern */
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_MASK	0x00008000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7	(1 << 15)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_7_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_MASK	0x00002000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6	(1 << 13)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_6_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_MASK	0x00000800
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5	(1 << 11)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_5_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_MASK	0x00000200
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4	(1 << 9)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_4_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_MASK	0x00000080
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3	(1 << 7)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_3_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_MASK	0x00000020
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2	(1 << 5)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_2_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_MASK	0x00000008
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1	(1 << 3)
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_ENABLED_LIGHTS_LIGHT_0_MASK	0x00000001
+#	define		NV30_TCL_PRIMITIVE_3D_SET_CLIPPING_PLANES		0x00001478
+#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000147c
+#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001480 + (d) * 0x0004)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3F_X( d)			(0x00001500 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3F_Y( d)			(0x00001504 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3F_Z( d)			(0x00001508 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3F_W( d)			(0x0000150c + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VB_POINTER( d)			(0x00001680 + (d) * 0x0004)	/* Parameters: source offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_POINTER_SOURCE_MASK	0x80000000
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_POINTER_SOURCE	(1 << 31)
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_POINTER_OFFSET_MASK	0x1fffffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0			0x00001718
+#	define		NV30_TCL_PRIMITIVE_3D_VTXFMT( d)			(0x00001740 + (d) * 0x0004)	/* Parameters: stride ncomp type */
+#		 define		NV30_TCL_PRIMITIVE_3D_VTXFMT_STRIDE_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_VTXFMT_STRIDE_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_VTXFMT_NCOMP_MASK	0x000000f0
+#		 define		NV30_TCL_PRIMITIVE_3D_VTXFMT_NCOMP_SHIFT	4
+#		 define		NV30_TCL_PRIMITIVE_3D_VTXFMT_TYPE_MASK	0x0000000f
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000017a0
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000017a4
+#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000017a8
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R		0x000017b0
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G		0x000017b4
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B		0x000017b8
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A		0x000017c0
+#	define		NV30_TCL_PRIMITIVE_3D_OCC_QUERY_OR_COLOR_BUFF_ENABLE	0x000017c8
+#	define		NV30_TCL_PRIMITIVE_3D_STORE_RESULT			0x00001800
 #	define		NV30_TCL_PRIMITIVE_3D_BEGIN_END				0x00001808
+#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16			0x0000180c	/* Parameters: 1 0 */
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16_1_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16_1_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16_0_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U32			0x00001810
+#	define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001814	/* Parameters: count_vertices offset_vertices */
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_COUNT_VERTICES_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_COUNT_VERTICES_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH_OFFSET_VERTICES_MASK	0x00ffffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001818
+#	define		NV30_TCL_PRIMITIVE_3D_SET_DISPLAY_LIST_MEM_OFFSET	0x0000181c
+#	define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST		0x00001824	/* Parameters: length start offset */
+#		 define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST_LENGTH_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST_LENGTH_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST_START_OFFSET_MASK	0x00ffffff
+#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x00001828
+#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x0000182c
 #	define		NV30_TCL_PRIMITIVE_3D_CULL_FACE				0x00001830
 #	define		NV30_TCL_PRIMITIVE_3D_FRONT_FACE			0x00001834
 #	define		NV30_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00001838
 #	define		NV30_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x0000183c
+#	define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT( d)			(0x00001840 + (d) * 0x0004)	/* Parameters: depth NPOT pitch */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT_DEPTH_MASK	0xfff00000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT_DEPTH_SHIFT	20
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT_NPOT_PITCH_MASK	0x000fffff
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2F_X( d)			(0x00001880 + (d) * 0x0008)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2F_Y( d)			(0x00001884 + (d) * 0x0008)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2I( d)			(0x00001900 + (d) * 0x0004)	/* Parameters: x y */
+#		 define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2I_X_MASK	0x0000ffff
+#		 define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2I_Y_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_2I_Y_SHIFT	16
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000194c	/* Parameters: a b g r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00001950	/* Parameters: a b g r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x000019c0	/* Parameters: t s */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x000019c4	/* Parameters: q r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x000019c8	/* Parameters: t s */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x000019cc	/* Parameters: q r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x000019d0	/* Parameters: t s */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x000019d4	/* Parameters: q r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x000019d8	/* Parameters: t s */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x000019dc	/* Parameters: q r */
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_TX_ADDRESS_UNIT( d)		(0x00001a00 + (d) * 0x0020)
+#	define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT( d)		(0x00001a04 + (d) * 0x0020)	/* Parameters: mipmap type format ncomp cubic */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_MIPMAP_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_MIPMAP_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_TYPE_MASK	0x00006000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_TYPE_SHIFT	13
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_TYPE_POT	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_TYPE_NPOT	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_TYPE_RECT	0x0003
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_MASK	0x00001f00
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_L8	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_A1R5G5B5	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_A4R4G4B4	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_R5G6B5	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_A8R8G8B8	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_DXT1	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_DXT3	0x0007
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_DXT5	0x0008
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_L16	0x0014
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_L16A16	0x0015
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_L8A8	0x0018
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_SL8A8	0x0019
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_RGBA_F16	0x001a
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_RGBA_F32	0x001b
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_L_F32	0x001c
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_FORMAT_LA_F16	0x001f
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_NCOMP_MASK	0x000000f0
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_NCOMP_SHIFT	4
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_CUBIC_MASK	0x00000004
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_CUBIC	(1 << 2)
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_CUBIC_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT_CUBIC_FALSE	0x0000
+#	define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT( d)			(0x00001a08 + (d) * 0x0020)	/* Parameters: wrap_s wrap_t wrap_r */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_S_MASK	0x000000ff
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_REPEAT	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_MIRRORED_REPEAT	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_CLAMP_TO_EDGE	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_CLAMP_TO_BORDER	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_T_CLAMP	0x0005
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_REPEAT	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_MIRRORED_REPEAT	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_CLAMP_TO_EDGE	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_CLAMP_TO_BORDER	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT_WRAP_R_CLAMP	0x0005
+#	define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT( d)		(0x00001a0c + (d) * 0x0020)	/* Parameters: nv40_enable nv30_enable anisotropy */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV40_ENABLE_MASK	0x80000000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV40_ENABLE	(1 << 31)
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV40_ENABLE_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV40_ENABLE_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV30_ENABLE_MASK	0x40000000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV30_ENABLE	(1 << 30)
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV30_ENABLE_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_NV30_ENABLE_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_MASK	0x00000070
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_SHIFT	4
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_1	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_2	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_4	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_NV40_6_NV30_8	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_8	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_10	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_12	0x0006
+#		 define NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT_ANISOTROPY_16	0x0007
+#	define		NV30_TCL_PRIMITIVE_3D_TX_SWIZZLE_UNIT( d)		(0x00001a10 + (d) * 0x0020)
+#	define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT( d)		(0x00001a14 + (d) * 0x0020)	/* Parameters: filter_min filter_mag */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_MASK	0x000f0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_NEAREST	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_LINEAR	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MIN_LINEAR_MIPMAP_LINEAR	0x0006
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_MASK	0x0f000000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_NEAREST	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_LINEAR	0x0002
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_NEAREST_MIPMAP_NEAREST	0x0003
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_LINEAR_MIPMAP_NEAREST	0x0004
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_NEAREST_MIPMAP_LINEAR	0x0005
+#		 define NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT_FILTER_MAG_LINEAR_MIPMAP_LINEAR	0x0006
+#	define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT( d)		(0x00001a18 + (d) * 0x0020)	/* Parameters: width height */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT_WIDTH_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT_WIDTH_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT_HEIGHT_MASK	0x0000ffff
+#	define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT( d)		(0x00001a1c + (d) * 0x0020)	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_A_SHIFT	24
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_A_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_A_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_R_SHIFT	16
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_R_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_R_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_G_SHIFT	8
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_G_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_G_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_TX_BORDER_COLOR_UNIT_B_MASK	0x000000ff
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4F_X( d)			(0x00001c00 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4F_Y( d)			(0x00001c04 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4F_Z( d)			(0x00001c08 + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4F_W( d)			(0x00001c0c + (d) * 0x0010)
+#	define		NV30_TCL_PRIMITIVE_3D_FP_CONTROL			0x00001d60	/* Parameters: uses_kil */
+#		 define		NV30_TCL_PRIMITIVE_3D_FP_CONTROL_USES_KIL_MASK	0x00000080
+#		 define		NV30_TCL_PRIMITIVE_3D_FP_CONTROL_USES_KIL	(1 << 7)
+#		 define NV30_TCL_PRIMITIVE_3D_FP_CONTROL_USES_KIL_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_FP_CONTROL_USES_KIL_FALSE	0x0000
+#	define		NV30_TCL_PRIMITIVE_3D_UNK1D6C_OFFSET			0x00001d6c
+#	define		NV30_TCL_PRIMITIVE_3D_UNK1D70_VALUE			0x00001d70
 #	define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_DEPTH			0x00001d8c
 #	define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB			0x00001d90	/* Parameters: a r g b */
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_A_MASK	0xff000000
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_A_SHIFT	24
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_R_MASK	0x00ff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_R_SHIFT	16
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_G_MASK	0x0000ff00
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_G_SHIFT	8
+#		 define		NV30_TCL_PRIMITIVE_3D_CLEAR_VALUE_ARGB_B_MASK	0x000000ff
 #	define		NV30_TCL_PRIMITIVE_3D_CLEAR_WHICH_BUFFERS		0x00001d94
+#	define		NV30_TCL_PRIMITIVE_3D_DO_VERTICES			0x00001dac
+#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE		0x00001db4
+#	define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN		0x00001db8	/* Parameters: factor pattern */
+#		 define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK	0x0000ffff
+#		 define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK	0xffff0000
+#		 define		NV30_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT	16
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_A		0x00001e20
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_B		0x00001e24
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_C		0x00001e28
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_D		0x00001e2c
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_E		0x00001e30
 #	define		NV30_TCL_PRIMITIVE_3D_BACK_MATERIAL_SHININESS_F		0x00001e34
-#	define		NV30_TCL_PRIMITIVE_3D_DO_VERTICES			0x00001dac
+#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00001e54
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_FROM_ID			0x00001e9c
 #	define		NV30_TCL_PRIMITIVE_3D_VP_PROGRAM_START_ID		0x00001ea0
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x00001ec0
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x00001ec4
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00001ec8
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_D			0x00001ecc
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_E			0x00001ed0
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x00001ed4
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00001ed8
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00001edc
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_SIZE			0x00001ee0
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE		0x00001ee4
+#	define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE			0x00001ee8	/* Parameters: coord_replace r_mode enable */
+#		 define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_MASK	0x00000800
+#		 define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE	(1 << 11)
+#		 define NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_COORD_REPLACE_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_MASK	0x00000006
+#		 define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_SHIFT	1
+#		 define NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_ZERO	0x0000
+#		 define NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_R	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_R_MODE_GL_S	0x0002
+#		 define		NV30_TCL_PRIMITIVE_3D_POINT_SPRITE_ENABLE_MASK	0x00000001
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID		0x00001efc
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_X		0x00001f00
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P0_Y		0x00001f04
@@ -1193,86 +4443,192 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_Y		0x00001f34
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_Z		0x00001f38
 #	define		NV30_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_P3_W		0x00001f3c
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3X(d)			(0x00001500 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3Y(d)			(0x00001504 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3Z(d)			(0x00001508 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_3W(d)			(0x0000150c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4X(d)			(0x00001c00 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4Y(d)			(0x00001c04 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4Z(d)			(0x00001c08 + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VTX_ATTR_4W(d)			(0x00001c0c + d * 0x0010)
-#	define		NV30_TCL_PRIMITIVE_3D_VB_POINTER_ATTR(d)		(0x00001680 + d * 0x0004)	/* Parameters: source: offset */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000a90	/* Parameters: y x */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000a94	/* Parameters: z */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x000018c0
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x000018c4
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000018c8
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000018cc
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000018d0
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000018d4
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x000018d8
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x000018dc
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x00001920	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x00001924	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x00001928	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x0000192c	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000194c	/* Parameters: a b g r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00001950	/* Parameters: a b g r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x000019c0	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x000019c4	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x000019c8	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x000019cc	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x000019d0	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x000019d4	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x000019d8	/* Parameters: t s */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x000019dc	/* Parameters: q r */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00001e54
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_UNK_0			0x00001718
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00001740 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS			0x00001740	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR1_WGH			0x00001744	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR2_NOR			0x00001748	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR3_COL			0x0000174c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR4_COL2			0x00001750	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR5_FOG			0x00001754	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR6			0x00001758	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR7			0x0000175c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR8_TX0			0x00001760	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR9_TX1			0x00001764	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR10_TX2			0x00001768	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR11_TX3			0x0000176c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR12_TX4			0x00001770	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR13_TX5			0x00001774	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR14_TX6			0x00001778	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR15_TX7			0x0000177c	/* Parameters: stride fields type */
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_R	0x000017a0
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_G	0x000017a4
-#	define		NV30_TCL_PRIMITIVE_3D_LIGHT_MODEL_BACK_SIDE_PRODUCT_AMBIENT_PLUS_EMISSION_B	0x000017a8
-#	define		NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM			0x000008e4
-#	define		NV30_TCL_PRIMITIVE_3D_TX_ADDRESS_UNIT(d)		(0x00001a00 + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_FORMAT_UNIT(d)			(0x00001a04 + d * 0x0020)	/* Parameters: mipmap type format ncomp cubic */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_WRAP_UNIT(d)			(0x00001a08 + d * 0x0020)	/* Parameters: wrap_s wrap_t wrap_r */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_ENABLE_UNIT(d)			(0x00001a0c + d * 0x0020)	/* Parameters: nv40_enable nv30_enable anisotropy */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_SWIZZLE_UNIT(d)		(0x00001a10 + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_FILTER_UNIT(d)			(0x00001a14 + d * 0x0020)	/* Parameters: filter_min filter_mag */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_XY_DIM_UNIT(d)			(0x00001a18 + d * 0x0020)	/* Parameters: width height */
-#	define		NV30_TCL_PRIMITIVE_3D_TX_UNK07_UNIT(d)			(0x00001a1c + d * 0x0020)
-#	define		NV30_TCL_PRIMITIVE_3D_TX_DEPTH_UNIT(d)			(0x00001840 + d * 0x0004)	/* Parameters: depth NPOT pitch */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001814	/* Parameters: count_vertices offset_vertices */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U16			0x0000180c	/* Parameters: 1: 0: */
-#	define		NV30_TCL_PRIMITIVE_3D_VB_ELEMENT_U32			0x00001810
-#	define		NV30_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001818
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x00000374
-#	define		NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x00000378
-#	define		NV30_TCL_PRIMITIVE_3D_SET_DISPLAY_LIST_MEM_OFFSET	0x0000181c
-#	define		NV30_TCL_PRIMITIVE_3D_EXECUTE_DISPLAY_LIST		0x00001824	/* Parameters: length start offset */
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x00001828
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x0000182c
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000147c
-#	define		NV30_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001480 + d * 0x0004)
-#	define		NV30_TCL_PRIMITIVE_3D_SET_CLIPPING_PLANES		0x00001478
-#	define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG				0x00001ff0	/* Parameters: vertex pos weight normal primary color secondary color fogcoord texture coords 0 texture ccords 1 texture coords 2 texture coords 3 texture coords 4 texture coords 5 texture coords 6 texture coords 7 */
-#	define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG			0x00001ff4	/* Parameters: primary color secondary color backface primary color backface secondary color fogcoord pointsize clip plane 0 clip plane 1 clip plane 2 clip plane 3 clip plane 4 clip plane 5 texture coords 0 texture coords 1 texture coords 2 texture coords 3 texture coords 4 texture coords 5 texture coords 6 texture coords 7 */
+#	define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG				0x00001ff0	/* Parameters: POS WEIGHT NORMAL COL0 COL1 FOGC TEX0 TEX1 TEX2 TEX3 TEX4 TEX5 TEX6 TEX7 */
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_POS_MASK	0x00000001
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_WEIGHT_MASK	0x00000002
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_WEIGHT	1 // Nothing to shift
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_WEIGHT_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_WEIGHT_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_NORMAL_MASK	0x00000004
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_NORMAL	(1 << 2)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_NORMAL_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_NORMAL_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL0_MASK	0x00000008
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL0	(1 << 3)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL0_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL0_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL1_MASK	0x00000010
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL1	(1 << 4)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_COL1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_FOGC_MASK	0x00000020
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_FOGC	(1 << 5)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_FOGC_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_FOGC_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX0_MASK	0x00000100
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX0	(1 << 8)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX0_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX0_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX1_MASK	0x00000200
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX1	(1 << 9)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX2_MASK	0x00000400
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX2	(1 << 10)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX2_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX2_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX3_MASK	0x00000800
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX3	(1 << 11)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX3_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX3_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX4_MASK	0x00001000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX4	(1 << 12)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX4_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX4_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX5_MASK	0x00002000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX5	(1 << 13)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX5_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX5_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX6_MASK	0x00004000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX6	(1 << 14)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX6_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX6_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX7_MASK	0x00008000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX7	(1 << 15)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX7_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_IN_REG_TEX7_FALSE	0x0000
+#	define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG			0x00001ff4	/* Parameters: COL0 COL1 BFC0 BFC1 FOGC PSZ CLP0 CLP1 CLP2 CLP3 CLP4 CLP5 TEX0 TEX1 TEX2 TEX3 TEX4 TEX5 TEX6 TEX7 */
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL0_MASK	0x00000001
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1_MASK	0x00000002
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1	1 // Nothing to shift
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0_MASK	0x00000004
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0	(1 << 2)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1_MASK	0x00000008
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1	(1 << 3)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC_MASK	0x00000010
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC	(1 << 4)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ_MASK	0x00000020
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ	(1 << 5)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0_MASK	0x00000040
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0	(1 << 6)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1_MASK	0x00000080
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1	(1 << 7)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2_MASK	0x00000100
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2	(1 << 8)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3_MASK	0x00000200
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3	(1 << 9)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4_MASK	0x00000400
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4	(1 << 10)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5_MASK	0x00000800
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5	(1 << 11)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0_MASK	0x00004000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0	(1 << 14)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX1_MASK	0x00008000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX1	(1 << 15)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX1_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX1_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX2_MASK	0x00010000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX2	(1 << 16)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX2_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX2_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX3_MASK	0x00020000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX3	(1 << 17)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX3_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX3_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX4_MASK	0x00040000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX4	(1 << 18)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX4_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX4_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX5_MASK	0x00080000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX5	(1 << 19)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX5_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX5_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX6_MASK	0x00100000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX6	(1 << 20)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX6_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX6_FALSE	0x0000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX7_MASK	0x00200000
+#		 define		NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX7	(1 << 21)
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX7_TRUE	0x0001
+#		 define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX7_FALSE	0x0000
+
+/****************************************** 
+Object NV40_TCL_PRIMITIVE_3D used on: NV40 G70
+*/
+#define			NV40_TCL_PRIMITIVE_3D					0x00000097
+#	define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123		0x00000370	/* Parameters: buffer3 b buffer3 g buffer3 r buffer3 a buffer2 b buffer2 g buffer2 r buffer2 a buffer1 b buffer1 g buffer1 r buffer1 a */
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_B_MASK	0x00008000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_B	(1 << 15)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_B_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_B_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_G_MASK	0x00004000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_G	(1 << 14)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_G_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_G_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_R_MASK	0x00002000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_R	(1 << 13)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_R_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_R_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_A_MASK	0x00001000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_A	(1 << 12)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_A_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER3_A_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_B_MASK	0x00000800
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_B	(1 << 11)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_B_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_B_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_G_MASK	0x00000400
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_G	(1 << 10)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_G_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_G_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_R_MASK	0x00000200
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_R	(1 << 9)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_R_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_R_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_A_MASK	0x00000100
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_A	(1 << 8)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_A_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER2_A_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_B_MASK	0x00000080
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_B	(1 << 7)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_B_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_B_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_G_MASK	0x00000040
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_G	(1 << 6)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_G_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_G_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_R_MASK	0x00000020
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_R	(1 << 5)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_R_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_R_FALSE	0x0000
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_A_MASK	0x00000010
+#		 define		NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_A	(1 << 4)
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_A_TRUE	0x0001
+#		 define NV40_TCL_PRIMITIVE_3D_COLOR_MASK_BUFFER123_BUFFER1_A_FALSE	0x0000
 
 /****************************************** 
 Object NV30_CLEAR_BUFFER used on: NV30 NV40 G70
@@ -1288,8 +4644,8 @@ Object NV30_CLEAR_BUFFER used on: NV30 NV40 G70
 Object NV50_TCL_PRIMITIVE_3D used on: 
 */
 #define			NV50_TCL_PRIMITIVE_3D					0x00000097
-#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_0( d)			(0x00000180 + d * 0x0004)
-#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_1( d)			(0x000001c0 + d * 0x0004)
+#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_0( d)			(0x00000180 + (d) * 0x0004)
+#	define		NV50_TCL_PRIMITIVE_3D_SET_OBJECT_1( d)			(0x000001c0 + (d) * 0x0004)
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_FOG_1F			0x00000314
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_X			0x00000380
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_Y			0x00000384
@@ -1338,23 +4694,83 @@ Object NV50_TCL_PRIMITIVE_3D used on:
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_R			0x000005b8
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_Q			0x000005bc
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I			0x000006a0	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2I_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I			0x000006a4	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2I_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x000006a8	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x000006ac	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY			0x00000700	/* Parameters: y x */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_Y_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY_X_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW			0x00000704	/* Parameters: w z */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_W_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW_Z_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x00000740	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x00000744	/* Parameters: q r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_Q_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ_R_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x00000748	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x0000074c	/* Parameters: q r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_Q_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ_R_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x00000750	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x00000754	/* Parameters: q r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_Q_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ_R_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x00000758	/* Parameters: t s */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_T_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST_S_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x0000075c	/* Parameters: q r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_Q_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ_R_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000790	/* Parameters: y x */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_Y_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY_X_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000794	/* Parameters: z */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z_Z_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000088c	/* Parameters: a b g r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_MASK	0xff000000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_A_SHIFT	24
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_MASK	0x00ff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_B_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_MASK	0x0000ff00
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_G_SHIFT	8
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I_R_MASK	0x000000ff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00000890	/* Parameters: a b g r */
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_MASK	0xff000000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_A_SHIFT	24
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_MASK	0x00ff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_B_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_MASK	0x0000ff00
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_G_SHIFT	8
+#		 define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I_R_MASK	0x000000ff
 #	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_X			0x00000a00
 #	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_Y			0x00000a04
 #	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK0_Z			0x00000a08
@@ -1363,8 +4779,14 @@ Object NV50_TCL_PRIMITIVE_3D used on:
 #	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_Z			0x00000a14
 #	define		NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000c08
 #	define		NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000c0c
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x00000d00 + d * 0x0008)	/* Parameters: x2 x1 */
-#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x00000d04 + d * 0x0008)	/* Parameters: y2 y1 */
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(d)		(0x00000d00 + (d) * 0x0008)	/* Parameters: x2 x1 */
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X2_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ_X1_MASK	0x0000ffff
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(d)		(0x00000d04 + (d) * 0x0008)	/* Parameters: y2 y1 */
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y2_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT_Y1_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_FIRST		0x00000d74
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_COUNT		0x00000d78
 #	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_R			0x00000d80
@@ -1380,7 +4802,13 @@ Object NV50_TCL_PRIMITIVE_3D used on:
 #	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE	0x00000dc4
 #	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE	0x00000dc8
 #	define		NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS		0x00000e04	/* Parameters: w x */
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_W_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_W_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS_X_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS		0x00000e08	/* Parameters: h y */
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_H_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_H_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_SCISSOR_HEIGHT_YPOS_Y_MASK	0x0000ffff
 #	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_ID		0x00000f00
 #	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_X			0x00000f04
 #	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Y			0x00000f08
@@ -1429,15 +4857,44 @@ Object NV50_TCL_PRIMITIVE_3D used on:
 #	define		NV50_TCL_PRIMITIVE_3D_VERTEX_DATA			0x00001640
 #	define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE		0x0000166c
 #	define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN		0x00001680	/* Parameters: pattern factor */
+#		 define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_MASK	0x00ffff00
+#		 define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_PATTERN_SHIFT	8
+#		 define		NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN_FACTOR_MASK	0x000000ff
 #	define		NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE		0x0000168c
-#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001700 + d * 0x0004)
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN( d)	(0x00001700 + (d) * 0x0004)
 #	define		NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00001918
 #	define		NV50_TCL_PRIMITIVE_3D_FRONT_FACE			0x0000191c
 #	define		NV50_TCL_PRIMITIVE_3D_CULL_FACE				0x00001920
 #	define		NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE			0x000019c4
 #	define		NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP			0x000019c8
 #	define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS			0x000019d0	/* Parameters: color stencil depth */
-#	define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK( d)			(0x00001a00 + d * 0x0004)	/* Parameters: a b g r */
+#		 define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS_COLOR_MASK	0x0000003c
+#		 define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS_COLOR_SHIFT	2
+#		 define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS_STENCIL_MASK	0x00000002
+#		 define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS_STENCIL	1 // Nothing to shift
+#		 define		NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS_DEPTH_MASK	0x00000001
+#	define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK( d)			(0x00001a00 + (d) * 0x0004)	/* Parameters: a b g r */
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_A_MASK	0x0000f000
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_A_SHIFT	12
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_A_TRUE	0x0001
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_A_FALSE	0x0000
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_B_MASK	0x00000f00
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_B_SHIFT	8
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_B_TRUE	0x0001
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_B_FALSE	0x0000
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_G_MASK	0x000000f0
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_G_SHIFT	4
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_G_TRUE	0x0001
+#		 define NV50_TCL_PRIMITIVE_3D_COLOR_MASK_G_FALSE	0x0000
+#		 define		NV50_TCL_PRIMITIVE_3D_COLOR_MASK_R_MASK	0x0000000f
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0			0x00000c00	/* Parameters: width x_offset */
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_WIDTH_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_WIDTH_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_0_X_OFFSET_MASK	0x0000ffff
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1			0x00000c04	/* Parameters: height y_offset */
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_HEIGHT_MASK	0xffff0000
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_HEIGHT_SHIFT	16
+#		 define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_DIMS_1_Y_OFFSET_MASK	0x0000ffff
 
 /****************************************** 
 Object NV_DMA_FROM_MEMORY used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
@@ -1468,13 +4925,49 @@ Object NvType0046 used on: NV04
 #	define		NvType0046_PITCH1					0x00000304
 #	define		NvType0046_PITCH2					0x0000030c
 #	define		NvType0046_SIZE						0x00000340	/* Parameters: width height */
+#		 define		NvType0046_SIZE_WIDTH_MASK	0x0000ffff
+#		 define		NvType0046_SIZE_HEIGHT_MASK	0xffff0000
+#		 define		NvType0046_SIZE_HEIGHT_SHIFT	16
 #	define		NvType0046_WIDTH					0x00000344	/* Parameters: visible_width blank_width */
+#		 define		NvType0046_WIDTH_VISIBLE_WIDTH_MASK	0x0000ffff
+#		 define		NvType0046_WIDTH_BLANK_WIDTH_MASK	0xffff0000
+#		 define		NvType0046_WIDTH_BLANK_WIDTH_SHIFT	16
 #	define		NvType0046_HSYNC					0x00000348	/* Parameters: hsync_start hsync_len */
+#		 define		NvType0046_HSYNC_HSYNC_START_MASK	0x0000ffff
+#		 define		NvType0046_HSYNC_HSYNC_LEN_MASK	0xffff0000
+#		 define		NvType0046_HSYNC_HSYNC_LEN_SHIFT	16
 #	define		NvType0046_HEIGHT					0x0000034c	/* Parameters: visible_height blank_height */
+#		 define		NvType0046_HEIGHT_VISIBLE_HEIGHT_MASK	0x0000ffff
+#		 define		NvType0046_HEIGHT_BLANK_HEIGHT_MASK	0xffff0000
+#		 define		NvType0046_HEIGHT_BLANK_HEIGHT_SHIFT	16
 #	define		NvType0046_VSYNC					0x00000350	/* Parameters: vsync_start vsync_len */
+#		 define		NvType0046_VSYNC_VSYNC_START_MASK	0x0000ffff
+#		 define		NvType0046_VSYNC_VSYNC_LEN_MASK	0xffff0000
+#		 define		NvType0046_VSYNC_VSYNC_LEN_SHIFT	16
 #	define		NvType0046_FULL_SIZE					0x00000354	/* Parameters: full_width full_height */
+#		 define		NvType0046_FULL_SIZE_FULL_WIDTH_MASK	0x0000ffff
+#		 define		NvType0046_FULL_SIZE_FULL_HEIGHT_MASK	0xffff0000
+#		 define		NvType0046_FULL_SIZE_FULL_HEIGHT_SHIFT	16
 #	define		NvType0046_PIXEL_CLK					0x00000358
-#	define		NvType0046_FLAGS					0x0000035c	/* Parameters: doublescan -hsync -vsync depth */
+#	define		NvType0046_FLAGS					0x0000035c	/* Parameters: doublescan neg_hsync neg_vsync depth */
+#		 define		NvType0046_FLAGS_DOUBLESCAN_MASK	0x00000002
+#		 define		NvType0046_FLAGS_DOUBLESCAN	1 // Nothing to shift
+#		 define NvType0046_FLAGS_DOUBLESCAN_TRUE	0x0001
+#		 define NvType0046_FLAGS_DOUBLESCAN_FALSE	0x0000
+#		 define		NvType0046_FLAGS_NEG_HSYNC_MASK	0x00000008
+#		 define		NvType0046_FLAGS_NEG_HSYNC	(1 << 3)
+#		 define NvType0046_FLAGS_NEG_HSYNC_TRUE	0x0001
+#		 define NvType0046_FLAGS_NEG_HSYNC_FALSE	0x0000
+#		 define		NvType0046_FLAGS_NEG_VSYNC_MASK	0x00000010
+#		 define		NvType0046_FLAGS_NEG_VSYNC	(1 << 4)
+#		 define NvType0046_FLAGS_NEG_VSYNC_TRUE	0x0001
+#		 define NvType0046_FLAGS_NEG_VSYNC_FALSE	0x0000
+#		 define		NvType0046_FLAGS_DEPTH_MASK	0x00030000
+#		 define		NvType0046_FLAGS_DEPTH_SHIFT	16
+#		 define NvType0046_FLAGS_DEPTH_8 bpp	0x0000
+#		 define NvType0046_FLAGS_DEPTH_16 bpp	0x0001
+#		 define NvType0046_FLAGS_DEPTH_15 bpp	0x0002
+#		 define NvType0046_FLAGS_DEPTH_24 bpp	0x0003
 
 /****************************************** 
 Object NvType0047 used on: NV04 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.c b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
index 69b0691bb7b..065aa81746c 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_screen.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
@@ -132,10 +132,11 @@ nouveauCreateBuffer(__DRIscreenPrivate *driScrnPriv,
                     GLboolean isPixmap)
 {
 	nouveauScreenPtr screen = (nouveauScreenPtr) driScrnPriv->private;
-	nouveau_renderbuffer  *nrb;
+	nouveau_renderbuffer_t *nrb;
 	struct gl_framebuffer *fb;
 	const GLboolean swAccum = mesaVis->accumRedBits > 0;
-	const GLboolean swStencil = mesaVis->stencilBits > 0 && mesaVis->depthBits != 24;
+	const GLboolean swStencil = (mesaVis->stencilBits > 0 &&
+				     mesaVis->depthBits != 24);
 	GLenum color_format = screen->fbFormat == 4 ? GL_RGBA8 : GL_RGB5;
 
 	if (isPixmap)
@@ -146,44 +147,26 @@ nouveauCreateBuffer(__DRIscreenPrivate *driScrnPriv,
 		return GL_FALSE;
 
 	/* Front buffer */
-	nrb = nouveau_renderbuffer_new(color_format,
-				       driScrnPriv->pFB + screen->frontOffset,
-				       screen->frontOffset,
-				       screen->frontPitch * screen->fbFormat,
-				       driDrawPriv);
-	nouveauSpanSetFunctions(nrb, mesaVis);
+	nrb = nouveau_renderbuffer_new(color_format);
 	_mesa_add_renderbuffer(fb, BUFFER_FRONT_LEFT, &nrb->mesa);
 
-	if (0 /* unified buffers if we choose to support them.. */) {
-	} else {
-		if (mesaVis->doubleBufferMode) {
-			nrb = nouveau_renderbuffer_new(color_format, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &nrb->mesa);
-		}
+	if (mesaVis->doubleBufferMode) {
+		nrb = nouveau_renderbuffer_new(color_format);
+		_mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &nrb->mesa);
+	}
 
-		if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH24_STENCIL8_EXT, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-			_mesa_add_renderbuffer(fb, BUFFER_STENCIL, &nrb->mesa);
-		} else if (mesaVis->depthBits == 24) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT24, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		} else if (mesaVis->depthBits == 16) {
-			nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT16, NULL,
-						       0, 0,
-						       NULL);
-			nouveauSpanSetFunctions(nrb, mesaVis);
-			_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
-		}
+	if (mesaVis->depthBits == 24 && mesaVis->stencilBits == 8) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH24_STENCIL8_EXT);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
+		_mesa_add_renderbuffer(fb, BUFFER_STENCIL, &nrb->mesa);
+	} else
+	if (mesaVis->depthBits == 24) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT24);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
+	} else
+	if (mesaVis->depthBits == 16) {
+		nrb = nouveau_renderbuffer_new(GL_DEPTH_COMPONENT16);
+		_mesa_add_renderbuffer(fb, BUFFER_DEPTH, &nrb->mesa);
 	}
 
 	_mesa_add_soft_renderbuffers(fb,
@@ -328,7 +311,7 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc
 	static const __DRIversion ddx_expected = { 1, 2, 0 };
 	static const __DRIversion dri_expected = { 4, 0, 0 };
 	static const __DRIversion drm_expected = { 0, 0, NOUVEAU_DRM_HEADER_PATCHLEVEL };
-#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 9
+#if NOUVEAU_DRM_HEADER_PATCHLEVEL != 10
 #error nouveau_drm.h version doesn't match expected version
 #endif
 	dri_interface = interface;
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.c b/src/mesa/drivers/dri/nouveau/nouveau_span.c
index 6e3f9fadf4e..d62830ff53e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_span.c
@@ -37,8 +37,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define HAVE_HW_STENCIL_SPANS	0
 #define HAVE_HW_STENCIL_PIXELS	0
 
-static char *fake_span[1280*1024*4];
-
 #define HW_CLIPLOOP()							\
    do {									\
       int _nc = nmesa->numClipRects;					\
@@ -50,11 +48,10 @@ static char *fake_span[1280*1024*4];
 
 #define LOCAL_VARS							\
    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);			\
-   nouveau_renderbuffer *nrb = (nouveau_renderbuffer *)rb;		\
+   nouveau_renderbuffer_t *nrb = (nouveau_renderbuffer_t *)rb;		\
    GLuint height = nrb->mesa.Height;					\
    GLubyte *map = (GLubyte *)(nrb->map ? nrb->map : nrb->mem->map) +    \
 	 (nmesa->drawY * nrb->pitch) + (nmesa->drawX * nrb->cpp);       \
-   map = fake_span; \
    GLuint p;								\
    (void) p;
 
@@ -70,7 +67,7 @@ static char *fake_span[1280*1024*4];
  * Color buffers
  */
 
-/* RGB565 */ 
+/* RGB565 */
 #define SPANTMP_PIXEL_FMT GL_RGB
 #define SPANTMP_PIXEL_TYPE GL_UNSIGNED_SHORT_5_6_5
 
@@ -89,40 +86,38 @@ static char *fake_span[1280*1024*4];
 #define GET_PTR(X,Y) (map + (Y)*nrb->pitch + (X)*nrb->cpp)
 #include "spantmp2.h"
 
-static void
-nouveauSpanRenderStart( GLcontext *ctx )
+static void nouveauSpanRenderStart(GLcontext * ctx)
 {
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   FIRE_RING();
-   LOCK_HARDWARE(nmesa);
-   nouveauWaitForIdleLocked( nmesa );
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	FIRE_RING();
+	LOCK_HARDWARE(nmesa);
+	nouveauWaitForIdleLocked(nmesa);
 }
 
-static void
-nouveauSpanRenderFinish( GLcontext *ctx )
+static void nouveauSpanRenderFinish(GLcontext * ctx)
 {
-   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-   _swrast_flush( ctx );
-   nouveauWaitForIdleLocked( nmesa );
-   UNLOCK_HARDWARE( nmesa );
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	_swrast_flush(ctx);
+	nouveauWaitForIdleLocked(nmesa);
+	UNLOCK_HARDWARE(nmesa);
 }
 
-void nouveauSpanInitFunctions( GLcontext *ctx )
+void nouveauSpanInitFunctions(GLcontext * ctx)
 {
-   struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx);
-   swdd->SpanRenderStart	= nouveauSpanRenderStart;
-   swdd->SpanRenderFinish	= nouveauSpanRenderFinish;
+	struct swrast_device_driver *swdd =
+	    _swrast_GetDeviceDriverReference(ctx);
+	swdd->SpanRenderStart = nouveauSpanRenderStart;
+	swdd->SpanRenderFinish = nouveauSpanRenderFinish;
 }
 
 
 /**
  * Plug in the Get/Put routines for the given driRenderbuffer.
  */
-void
-nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis)
+void nouveauSpanSetFunctions(nouveau_renderbuffer_t * nrb)
 {
-   if (nrb->mesa._ActualFormat == GL_RGBA8)
-      nouveauInitPointers_ARGB8888(&nrb->mesa);
-   else // if (nrb->mesa._ActualFormat == GL_RGB5)
-      nouveauInitPointers_RGB565(&nrb->mesa);
+	if (nrb->mesa._ActualFormat == GL_RGBA8)
+		nouveauInitPointers_ARGB8888(&nrb->mesa);
+	else			// if (nrb->mesa._ActualFormat == GL_RGB5)
+		nouveauInitPointers_RGB565(&nrb->mesa);
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_span.h b/src/mesa/drivers/dri/nouveau/nouveau_span.h
index bc39ecd17b5..d3f31c9cb23 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_span.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_span.h
@@ -30,10 +30,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define __NOUVEAU_SPAN_H__
 
 #include "drirenderbuffer.h"
-#include "nouveau_buffers.h"
+#include "nouveau_fbo.h"
 
-extern void nouveauSpanInitFunctions( GLcontext *ctx );
-extern void nouveauSpanSetFunctions(nouveau_renderbuffer *nrb, const GLvisual *vis);
+extern void nouveauSpanInitFunctions(GLcontext *ctx);
+extern void nouveauSpanSetFunctions(nouveau_renderbuffer_t *nrb);
 
 #endif /* __NOUVEAU_SPAN_H__ */
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c
index f618dcfc99b..e2f9fb869a1 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c
@@ -100,6 +100,14 @@ static void nouveauDepthRange(GLcontext *ctx, GLclampd near, GLclampd far)
     nouveauCalcViewport(ctx);
 }
 
+static void nouveauUpdateProjectionMatrix(GLcontext *ctx)
+{
+}
+
+static void nouveauUpdateModelviewMatrix(GLcontext *ctx)
+{
+}
+
 static void nouveauDDUpdateHWState(GLcontext *ctx)
 {
     nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
@@ -141,6 +149,15 @@ static void nouveauDDUpdateHWState(GLcontext *ctx)
 
 static void nouveauDDInvalidateState(GLcontext *ctx, GLuint new_state)
 {
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if ( new_state & _NEW_PROJECTION ) {
+		nmesa->hw_func.UpdateProjectionMatrix(ctx);
+	}
+        if ( new_state & _NEW_MODELVIEW ) {
+		nmesa->hw_func.UpdateModelviewMatrix(ctx);
+	}
+
     _swrast_InvalidateState( ctx, new_state );
     _swsetup_InvalidateState( ctx, new_state );
     _vbo_InvalidateState( ctx, new_state );
@@ -154,9 +171,6 @@ void nouveauDDInitState(nouveauContextPtr nmesa)
     uint32_t type = nmesa->screen->card->type;
     switch(type)
     {
-        case NV_03:
-            /* Unimplemented */
-            break;
         case NV_04:
         case NV_05:
             nv04InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
@@ -186,6 +200,8 @@ void nouveauDDInitState(nouveauContextPtr nmesa)
 /* Initialize the driver's state functions */
 void nouveauDDInitStateFuncs(GLcontext *ctx)
 {
+   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
    ctx->Driver.UpdateState		= nouveauDDInvalidateState;
 
    ctx->Driver.ClearIndex		= NULL;
@@ -236,6 +252,10 @@ void nouveauDDInitStateFuncs(GLcontext *ctx)
    ctx->Driver.CopyColorSubTable = _swrast_CopyColorSubTable;
    ctx->Driver.CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D;
    ctx->Driver.CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D;
+
+   /* Matrix updates */
+   nmesa->hw_func.UpdateProjectionMatrix = nouveauUpdateProjectionMatrix;
+   nmesa->hw_func.UpdateModelviewMatrix = nouveauUpdateModelviewMatrix;
 }
 
 #define STATE_INIT(a) if (ctx->Driver.a) ctx->Driver.a
@@ -276,6 +296,7 @@ void nouveauInitState(GLcontext *ctx)
     STATE_INIT(CullFace)( ctx, ctx->Polygon.CullFaceMode );
     STATE_INIT(DepthFunc)( ctx, ctx->Depth.Func );
     STATE_INIT(DepthMask)( ctx, ctx->Depth.Mask );
+    STATE_INIT(DepthRange)( ctx, ctx->Viewport.Near, ctx->Viewport.Far );
 
     STATE_INIT(Enable)( ctx, GL_ALPHA_TEST, ctx->Color.AlphaEnabled );
     STATE_INIT(Enable)( ctx, GL_BLEND, ctx->Color.BlendEnabled );
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.c b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
index 8abc847e1e2..2ca038f4f8e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
@@ -28,36 +28,36 @@
 #include "vblank.h" /* for DO_USLEEP */
 
 #include "nouveau_context.h"
-#include "nouveau_buffers.h"
-#include "nouveau_object.h"
 #include "nouveau_fifo.h"
-#include "nouveau_reg.h"
+#include "nouveau_mem.h"
 #include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
 #include "nouveau_sync.h"
 
 #define NOTIFIER(__v) \
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); \
-	volatile uint32_t *__v = (void*)nmesa->notifier_block + notifier->offset
+	volatile uint32_t *__v = (void*)nmesa->fifo.notifier_block + \
+				 notifier->offset
 
-struct drm_nouveau_notifier_alloc *
+struct drm_nouveau_notifierobj_alloc *
 nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	struct drm_nouveau_notifier_alloc *notifier;
+	struct drm_nouveau_notifierobj_alloc *notifier;
 	int ret;
 
 #ifdef NOUVEAU_RING_DEBUG
 	return NULL;
 #endif
-
-	notifier = CALLOC_STRUCT(drm_nouveau_notifier_alloc);
+	notifier = CALLOC_STRUCT(drm_nouveau_notifierobj_alloc);
 	if (!notifier)
 		return NULL;
 
-	notifier->channel = nmesa->fifo.channel;
+	notifier->channel = nmesa->fifo.drm.channel;
 	notifier->handle  = handle;
 	notifier->count   = count;
-	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_NOTIFIER_ALLOC,
+	ret = drmCommandWriteRead(nmesa->driFd, DRM_NOUVEAU_NOTIFIEROBJ_ALLOC,
 				  notifier, sizeof(*notifier));
 	if (ret) {
 		MESSAGE("Failed to create notifier 0x%08x: %d\n", handle, ret);
@@ -70,7 +70,7 @@ nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
 
 void
 nouveau_notifier_destroy(GLcontext *ctx,
-			 struct drm_nouveau_notifier_alloc *notifier)
+			 struct drm_nouveau_notifierobj_alloc *notifier)
 {
 	/*XXX: free notifier object.. */
 	FREE(notifier);
@@ -78,7 +78,7 @@ nouveau_notifier_destroy(GLcontext *ctx,
 
 void
 nouveau_notifier_reset(GLcontext *ctx,
-		       struct drm_nouveau_notifier_alloc *notifier,
+		       struct drm_nouveau_notifierobj_alloc *notifier,
 		       GLuint id)
 {
 	NOTIFIER(n);
@@ -96,7 +96,7 @@ nouveau_notifier_reset(GLcontext *ctx,
 
 GLuint
 nouveau_notifier_status(GLcontext *ctx,
-			struct drm_nouveau_notifier_alloc *notifier,
+			struct drm_nouveau_notifierobj_alloc *notifier,
 			GLuint id)
 {
 	NOTIFIER(n);
@@ -106,7 +106,7 @@ nouveau_notifier_status(GLcontext *ctx,
 
 GLuint
 nouveau_notifier_return_val(GLcontext *ctx,
-			    struct drm_nouveau_notifier_alloc *notifier,
+			    struct drm_nouveau_notifierobj_alloc *notifier,
 			    GLuint id)
 {
 	NOTIFIER(n);
@@ -116,7 +116,7 @@ nouveau_notifier_return_val(GLcontext *ctx,
 
 GLboolean
 nouveau_notifier_wait_status(GLcontext *ctx,
-			     struct drm_nouveau_notifier_alloc *notifier,
+			     struct drm_nouveau_notifierobj_alloc *notifier,
 			     GLuint id, GLuint status, GLuint timeout)
 {
 	NOTIFIER(n);
@@ -150,7 +150,7 @@ nouveau_notifier_wait_status(GLcontext *ctx,
 
 void
 nouveau_notifier_wait_nop(GLcontext *ctx,
-			  struct drm_nouveau_notifier_alloc *notifier,
+			  struct drm_nouveau_notifierobj_alloc *notifier,
 			  GLuint subc)
 {
 	NOTIFIER(n);
@@ -189,11 +189,9 @@ GLboolean nouveauSyncInitFuncs(GLcontext *ctx)
 	 */
 	BEGIN_RING_CACHE(NvSub3D, 0x180, 1);
 	OUT_RING_CACHE  (NvSyncNotify);
-#ifdef ALLOW_MULTI_SUBCHANNEL
 	BEGIN_RING_SIZE(NvSubMemFormat,
 	      		NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY, 1);
 	OUT_RING       (NvSyncNotify);
-#endif
 
 	return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.h b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
index b76af172762..fc37efbe6b2 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
@@ -28,8 +28,6 @@
 #ifndef __NOUVEAU_SYNC_H__
 #define __NOUVEAU_SYNC_H__
 
-#include "nouveau_buffers.h"
-
 #define NV_NOTIFIER_SIZE                                                      32
 #define NV_NOTIFY_TIME_0                                              0x00000000
 #define NV_NOTIFY_TIME_1                                              0x00000004
@@ -47,24 +45,24 @@
 #define NV_NOTIFY                                                     0x00000104
 #define NV_NOTIFY_STYLE_WRITE_ONLY                                             0
 
-extern struct drm_nouveau_notifier_alloc *
+extern struct drm_nouveau_notifierobj_alloc *
 nouveau_notifier_new(GLcontext *, GLuint handle, GLuint count);
 extern void 
-nouveau_notifier_destroy(GLcontext *, struct drm_nouveau_notifier_alloc *);
+nouveau_notifier_destroy(GLcontext *, struct drm_nouveau_notifierobj_alloc *);
 extern void
-nouveau_notifier_reset(GLcontext *, struct drm_nouveau_notifier_alloc *,
+nouveau_notifier_reset(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 		       GLuint id);
 extern GLuint
-nouveau_notifier_status(GLcontext *, struct drm_nouveau_notifier_alloc *,
+nouveau_notifier_status(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 			GLuint id);
 extern GLuint
-nouveau_notifier_return_val(GLcontext *, struct drm_nouveau_notifier_alloc *,
+nouveau_notifier_return_val(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 			    GLuint id);
 extern GLboolean
-nouveau_notifier_wait_status(GLcontext *, struct drm_nouveau_notifier_alloc *,
+nouveau_notifier_wait_status(GLcontext *, struct drm_nouveau_notifierobj_alloc *,
 			     GLuint id, GLuint status, GLuint timeout);
 extern void
-nouveau_notifier_wait_nop(GLcontext *ctx, struct drm_nouveau_notifier_alloc *,
+nouveau_notifier_wait_nop(GLcontext *ctx, struct drm_nouveau_notifierobj_alloc *,
 			  GLuint subc);
 
 extern GLboolean nouveauSyncInitFuncs(GLcontext *ctx);
diff --git a/src/mesa/drivers/dri/nouveau/nv04_state.c b/src/mesa/drivers/dri/nouveau/nv04_state.c
index 25df3d2a624..d3031aa5b19 100644
--- a/src/mesa/drivers/dri/nouveau/nv04_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv04_state.c
@@ -451,8 +451,8 @@ static GLboolean nv04InitCard(nouveauContextPtr nmesa)
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv04BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
+		nouveau_renderbuffer_t **color,
+		nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	uint32_t depth_pitch=(depth?depth->pitch:0+15)&~15+16;
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state.c b/src/mesa/drivers/dri/nouveau/nv10_state.c
index 47c4b14ba6b..8cbe72020fe 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state.c
@@ -37,29 +37,28 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 static void nv10ViewportScale(nouveauContextPtr nmesa)
 {
 	GLcontext *ctx = nmesa->glCtx;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-
+	GLfloat w = ((GLfloat) ctx->Viewport.Width) * 0.5;
+	GLfloat h = ((GLfloat) ctx->Viewport.Height) * 0.5;
 	GLfloat max_depth = (ctx->Viewport.Near + ctx->Viewport.Far) * 0.5;
-/*	if (ctx->DrawBuffer) {
-		switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
-			case 16:
-				max_depth *= 32767.0;
-				break;
-			case 24:
-				max_depth *= 16777215.0;
-				break;
+
+	if (ctx->DrawBuffer) {
+		if (ctx->DrawBuffer->_DepthBuffer) {
+			switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
+				case 16:
+					max_depth *= 32767.0;
+					break;
+				case 24:
+					max_depth *= 16777215.0;
+					break;
+			}
 		}
-	} else {*/
-		/* Default to 24 bits range */	
-		max_depth *= 16777215.0;
-/*	}*/
+	}
 
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X, 4);
-	OUT_RING_CACHEf ((((GLfloat) w) * 0.5) - 2048.0);
-	OUT_RING_CACHEf ((((GLfloat) h) * 0.5) - 2048.0);
-	OUT_RING_CACHEf (max_depth);
-	OUT_RING_CACHEf (0.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X, 4);
+	OUT_RINGf (w - 2048.0);
+	OUT_RINGf (h - 2048.0);
+	OUT_RINGf (max_depth);
+	OUT_RINGf (0.0);
 }
 
 static void nv10AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
@@ -111,9 +110,58 @@ static void nv10BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfac
 	OUT_RING_CACHE(dfactorRGB);
 }
 
+static void nv10ClearBuffer(GLcontext *ctx, nouveau_renderbuffer_t *buffer, int fill, int mask)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	int dimensions;
+
+	if (!buffer) {
+		return;
+	}
+
+	/* Surface that we will work on */
+	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf2D, NvCtxSurf2D);
+
+	BEGIN_RING_SIZE(NvSubCtxSurf2D, NV10_CONTEXT_SURFACES_2D_FORMAT, 4);
+	OUT_RING(0x0b); /* Y32 color format */
+	OUT_RING((buffer->pitch<<16)|buffer->pitch);
+	OUT_RING(buffer->offset);
+	OUT_RING(buffer->offset);
+
+	/* Now clear a rectangle */
+	dimensions = ((buffer->mesa.Height)<<16) | (buffer->mesa.Width);
+
+	nouveauObjectOnSubchannel(nmesa, NvSubGdiRectText, NvGdiRectText);
+
+	BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_OPERATION, 1);
+	OUT_RING(3);	/* SRCCOPY */
+
+	BEGIN_RING_SIZE(NvSubGdiRectText, NV04_GDI_RECTANGLE_TEXT_BLOCK_LEVEL1_TL, 5);
+	OUT_RING(0);	/* top left */
+	OUT_RING(dimensions);	/* bottom right */
+	OUT_RING(fill);
+	OUT_RING(0);	/* top left */
+	OUT_RING(dimensions);	/* bottom right */
+}
+
 static void nv10Clear(GLcontext *ctx, GLbitfield mask)
 {
-	/* TODO */
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (mask & (BUFFER_BIT_FRONT_LEFT|BUFFER_BIT_BACK_LEFT)) {
+		nv10ClearBuffer(ctx, nmesa->color_buffer,
+			nmesa->clear_color_value, 0xffffffff);
+	}
+	/* FIXME: check depth bits */
+	if (mask & (BUFFER_BIT_DEPTH)) {
+		nv10ClearBuffer(ctx, nmesa->depth_buffer,
+			nmesa->clear_value, 0xffffff00);
+	}
+	/* FIXME: check about stencil? */
+	if (mask & (BUFFER_BIT_STENCIL)) {
+		nv10ClearBuffer(ctx, nmesa->depth_buffer,
+			nmesa->clear_value, 0x000000ff);
+	}
 }
 
 static void nv10ClearColor(GLcontext *ctx, const GLfloat color[4])
@@ -152,7 +200,7 @@ static void nv10ClearStencil(GLcontext *ctx, GLint s)
 static void nv10ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane), 4);
+	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane - GL_CLIP_PLANE0), 4);
 	OUT_RING_CACHEf(equation[0]);
 	OUT_RING_CACHEf(equation[1]);
 	OUT_RING_CACHEf(equation[2]);
@@ -205,8 +253,12 @@ static void nv10DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
 	GLfloat depth_scale = 16777216.0;
-	if (ctx->DrawBuffer->_DepthBuffer->DepthBits == 16) {
-		depth_scale = 32768.0;
+	if (ctx->DrawBuffer) {
+		if (ctx->DrawBuffer->_DepthBuffer) {
+			if (ctx->DrawBuffer->_DepthBuffer->DepthBits == 16) {
+				depth_scale = 32768.0;
+			}
+		}
 	}
 
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
@@ -245,8 +297,10 @@ static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 			OUT_RING_CACHE(state);
 			break;
 		case GL_COLOR_LOGIC_OP:
-			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
-			OUT_RING_CACHE(state);
+			if (nmesa->screen->card->type >= NV_11) {
+				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
+				OUT_RING_CACHE(state);
+			}
 			break;
 //		case GL_COLOR_MATERIAL:
 //		case GL_COLOR_SUM_EXT:
@@ -271,6 +325,11 @@ static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 			break;
 //		case GL_HISTOGRAM:
 //		case GL_INDEX_LOGIC_OP:
+#if 0
+			/* light is broken, the hardware seem to only allow to use light 
+			 * in order : ie GL_LIGHT0 & GL_LIGHT2 is invalid
+			 * In this case the blob remap GL_LIGHT2 to hw light 1
+			 */
 		case GL_LIGHT0:
 		case GL_LIGHT1:
 		case GL_LIGHT2:
@@ -281,7 +340,11 @@ static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 		case GL_LIGHT7:
 			{
 			uint32_t mask=1<<(2*(cap-GL_LIGHT0));
-			nmesa->enabled_lights=((nmesa->enabled_lights&mask)|(mask*state));
+			if (state)
+				nmesa->enabled_lights |= mask;
+			else
+				nmesa->enabled_lights &= ~mask;
+
 			if (nmesa->lighting_enabled)
 			{
 				BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ENABLED_LIGHTS, 1);
@@ -297,6 +360,7 @@ static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 			else
 				OUT_RING_CACHE(0x0);
 			break;
+#endif
 		case GL_LINE_SMOOTH:
 			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE, 1);
 			OUT_RING_CACHE(state);
@@ -366,11 +430,21 @@ static void nv10Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 static void nv10Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
 {
     nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+    GLubyte cf[4];
     switch(pname)
     {
         case GL_FOG_MODE:
-            //BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 1);
-            //OUT_RING_CACHE (params);
+            BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 1);
+            OUT_RING_CACHE (ctx->Fog.Mode); /* can we extract it from params ??? */
+            break;
+        case GL_FOG_COLOR:
+            CLAMPED_FLOAT_TO_UBYTE(cf[0], params[0]);
+            CLAMPED_FLOAT_TO_UBYTE(cf[1], params[1]);
+            CLAMPED_FLOAT_TO_UBYTE(cf[2], params[2]);
+            CLAMPED_FLOAT_TO_UBYTE(cf[3], params[3]);
+
+            BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_COLOR, 1);
+            OUT_RING_CACHE(PACK_COLOR_8888(cf[3], cf[1], cf[2], cf[0]));
             break;
             /* TODO: unsure about the rest.*/
         default:
@@ -522,6 +596,11 @@ static void nv10LineWidth(GLcontext *ctx, GLfloat width)
 static void nv10LogicOpcode(GLcontext *ctx, GLenum opcode)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (nmesa->screen->card->type < NV_11) {
+		return;
+	}
+
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP, 1);
 	OUT_RING_CACHE(opcode);
 }
@@ -574,9 +653,52 @@ void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
 /** Set rasterization mode */
 void (*RenderMode)(GLcontext *ctx, GLenum mode );
 
+/* Translate GL coords to window coords, clamping w/h to the
+ * dimensions of the window.
+ */
+static void nv10WindowCoords(nouveauContextPtr nmesa,
+			     GLuint x, GLuint y, GLuint w, GLuint h,
+			     GLuint *wX, GLuint *wY, GLuint *wW, GLuint *wH)
+{
+	if ((x+w) > nmesa->drawW)
+		w = nmesa->drawW - x;
+	(*wX) = x + nmesa->drawX;
+	(*wW) = w;
+
+	if ((y+h) > nmesa->drawH)
+		h = nmesa->drawH - y;
+	(*wY) = (nmesa->drawH - y) - h + nmesa->drawY;
+	(*wH) = h;
+}
+
 /** Define the scissor box */
 static void nv10Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLuint wX, wY, wW, wH;
+
+	/* There's no scissor enable bit, so adjust the scissor to cover the
+	 * maximum draw buffer bounds
+	 */
+	if (!ctx->Scissor.Enabled) {
+	   wX = nmesa->drawX;
+	   wY = nmesa->drawY;
+	   wW = nmesa->drawW;
+	   wH = nmesa->drawH;
+	} else {
+	   nv10WindowCoords(nmesa, x, y, w, h, &wX, &wY, &wW, &wH);
+	}
+
+	if (!wW || !wH) {
+		return;
+	}
+
+	BEGIN_RING_SIZE(NvSub3D,
+	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
+        OUT_RING(((wW+wX-1) << 16) | wX | 0x08000800);
+	BEGIN_RING_SIZE(NvSub3D,
+	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
+        OUT_RING(((wH+wY-1) << 16) | wY | 0x08000800);
 }
 
 /** Select flat or smooth shading */
@@ -643,55 +765,115 @@ static void nv10TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
         OUT_RING_CACHEp(mat->m, 16);
 }
 
+static void nv10UpdateProjectionMatrix(GLcontext *ctx)
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLfloat w = ((GLfloat) ctx->Viewport.Width) * 0.5;
+	GLfloat h = ((GLfloat) ctx->Viewport.Height) * 0.5;
+	GLfloat max_depth = (ctx->Viewport.Near + ctx->Viewport.Far) * 0.5;
+	GLfloat projection[16];
+	int i;
+
+	if (ctx->DrawBuffer) {
+		if (ctx->DrawBuffer->_DepthBuffer) {
+			switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
+				case 16:
+					max_depth *= 32767.0;
+					break;
+				case 24:
+					max_depth *= 16777215.0;
+					break;
+			}
+		}
+	}
+
+	/* Transpose and rescale for viewport */
+	for (i=0; i<4; i++) {
+		projection[i] = w * ctx->_ModelProjectMatrix.m[i*4];
+	}
+	for (i=0; i<4; i++) {
+		projection[i+4] = -h * ctx->_ModelProjectMatrix.m[i*4+1];
+	}
+	for (i=0; i<4; i++) {
+		projection[i+8] = max_depth * ctx->_ModelProjectMatrix.m[i*4+2];
+	}
+	for (i=0; i<4; i++) {
+		projection[i+12] = ctx->_ModelProjectMatrix.m[i*4+3];
+	}
+
+	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(0), 16);
+	OUT_RING_CACHE_FORCEp(projection, 16);
+}
+
+static void nv10UpdateModelviewMatrix(GLcontext *ctx)
+{
+	/* TODO	update modelview if lighting or vertex weight enabled
+		update inverse modelview if lighting enabled
+		or update projection if lighting and vertex weight disabled
+	*/
+
+	nv10UpdateProjectionMatrix(ctx);
+}
+
 /* Update anything that depends on the window position/size */
 static void nv10WindowMoved(nouveauContextPtr nmesa)
 {
 	GLcontext *ctx = nmesa->glCtx;
 	GLfloat *v = nmesa->viewport.m;
-	GLuint w = ctx->Viewport.Width;
-	GLuint h = ctx->Viewport.Height;
-	GLuint x = ctx->Viewport.X + nmesa->drawX;
-	GLuint y = ctx->Viewport.Y + nmesa->drawY;
-	int i;
+	GLuint wX, wY, wW, wH;
 
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
+	nv10WindowCoords(nmesa, ctx->Viewport.X, ctx->Viewport.Y, 
+				ctx->Viewport.Width, ctx->Viewport.Height,
+				&wX, &wY, &wW, &wH);
 
-	/* something to do with clears, possibly doesn't belong here */
-	BEGIN_RING_SIZE(NvSub3D, 0x02b4, 1);
-	OUT_RING(0);
-
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
-        OUT_RING_CACHE(((w+x-1) << 16) | x | 0x08000800);
-	BEGIN_RING_CACHE(NvSub3D,
-	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
-        OUT_RING_CACHE(((h+y-1) << 16) | y | 0x08000800);
-	for (i=1; i<8; i++) {
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 1);
-        	OUT_RING_CACHE(0);
-		BEGIN_RING_CACHE(NvSub3D,
-		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(i), 1);
-	        OUT_RING_CACHE(0);
-	}
+        BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
+        OUT_RING((wW << 16) | wX);
+        OUT_RING((wH << 16) | wY);
 
 	nv10ViewportScale(nmesa);
+
+	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+	      		    ctx->Scissor.Width, ctx->Scissor.Height);
 }
 
 /* Initialise any card-specific non-GL related state */
 static GLboolean nv10InitCard(nouveauContextPtr nmesa)
 {
+	int i;
+	GLfloat projection[16];
+
 	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
 
 	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY0, 2);
 	OUT_RING(NvDmaFB);	/* 184 dma_in_memory0 */
-	OUT_RING(NvDmaFB);	/* 188 dma_in_memory1 */
+	OUT_RING(NvDmaTT);	/* 188 dma_in_memory1 */
 	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_SET_DMA_IN_MEMORY2, 2);
 	OUT_RING(NvDmaFB);	/* 194 dma_in_memory2 */
 	OUT_RING(NvDmaFB);	/* 198 dma_in_memory3 */
 
+	/* 0x0 viewport size */
+        BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
+        OUT_RING(0);
+        OUT_RING(0);
+
+	/* Clipping regions */
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_MODE, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D,
+	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
+        OUT_RING(0x07ff0800);
+	BEGIN_RING_SIZE(NvSub3D,
+	      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
+        OUT_RING(0x07ff0800);
+	for (i=1; i<8; i++) {
+		BEGIN_RING_SIZE(NvSub3D,
+		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 1);
+        	OUT_RING(0);
+		BEGIN_RING_SIZE(NvSub3D,
+		      NV10_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(i), 1);
+	        OUT_RING(0);
+	}
+
 	BEGIN_RING_SIZE(NvSub3D, 0x0290, 1);
 	OUT_RING(0x00100001);
 	BEGIN_RING_SIZE(NvSub3D, 0x03f4, 1);
@@ -702,19 +884,126 @@ static GLboolean nv10InitCard(nouveauContextPtr nmesa)
         	OUT_RING(0);
 	        OUT_RING(1);
 	        OUT_RING(2);
+
+	        BEGIN_RING_SIZE(NvSubImageBlit, 0x120, 3);
+        	OUT_RING(0);
+	        OUT_RING(1);
+	        OUT_RING(2);
 	}
 
+	/* Set state for stuff not initialized in nouveau_state.c */
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_TX_ENABLE(0), 2);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_RC_IN_ALPHA(0), 12);
+	OUT_RING  (0x30141010);
+	OUT_RING  (0);
+	OUT_RING  (0x20040000);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0);
+	OUT_RING  (0x00000c00);
+	OUT_RING  (0x18000000);
+	OUT_RING  (0x300e0300);
+	OUT_RING  (0x0c091c80);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_WEIGHT_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LIGHT_MODEL, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_CONTROL, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
+	OUT_RING  (8);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POINT_PARAMETERS_ENABLE, 1);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_LINE_WIDTH, 1);
+	OUT_RING  (8);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(0), 8);
+	for (i=0;i<8;i++) {
+		OUT_RING  (0);
+	}
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT, 3);
+	OUT_RINGf (-1.50);
+	OUT_RINGf (-0.09);
+	OUT_RINGf ( 0.00);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 2);
+	OUT_RING  (0x802);
+	OUT_RING  (2);
+
+	/* Projection and modelview matrix */
+	memset(projection, 0, sizeof(projection));
+	projection[0*4+0] = 1.0;
+	projection[1*4+1] = 1.0;
+	projection[2*4+2] = 1.0;
+	projection[3*4+3] = 1.0;
+
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEW_MATRIX_ENABLE, 1);
+	OUT_RING  (6);	/* enable projection and modelview0 matrix */
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_PROJECTION_MATRIX(0), 16);
+	for (i=0; i<16; i++) {
+		OUT_RINGf (projection[i]);
+	}
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_MODELVIEW0_MATRIX(0), 16);
+	for (i=0; i<16; i++) {
+		OUT_RINGf (projection[i]);
+	}
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
+	OUT_RINGf  (0.0);
+	OUT_RINGf  (1.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_SCALE_X, 4);
+	OUT_RINGf  (1.0);
+	OUT_RINGf  (1.0);
+	OUT_RINGf  (1.0);
+	OUT_RINGf  (1.0);
+
+	/* Set per-vertex component */
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R, 4);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R, 3);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	OUT_RING  (0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X, 3);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S, 4);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S, 4);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (0.0);
+	OUT_RINGf (1.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_FOG_1F, 1);
+	OUT_RINGf (0.0);
+	BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_EDGEFLAG_ENABLE, 1);
+	OUT_RING  (1);
+
 	return GL_TRUE;
 }
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	GLuint pitch, format, depth_pitch;
 
+	/* Store buffer pointers in context */
+	nmesa->color_buffer = color[0];
+	nmesa->depth_buffer = depth;
+
 	w = color[0]->mesa.Width;
 	h = color[0]->mesa.Height;
 	x = nmesa->drawX;
@@ -723,26 +1012,19 @@ static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
 	if (num_color != 1)
 		return GL_FALSE;
 
-        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 6);
-        OUT_RING_CACHE((w << 16) | x);
-        OUT_RING_CACHE((h << 16) | y);
+        BEGIN_RING_SIZE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 6);
+        OUT_RING((w << 16) | x);
+        OUT_RING((h << 16) | y);
 	depth_pitch = (depth ? depth->pitch : color[0]->pitch);
 	pitch = (depth_pitch<<16) | color[0]->pitch;
 	format = 0x108;
 	if (color[0]->mesa._ActualFormat != GL_RGBA8) {
 		format = 0x103; /* R5G6B5 color buffer */
 	}
-	OUT_RING_CACHE(format);
-	OUT_RING_CACHE(pitch);
-	OUT_RING_CACHE(color[0]->offset);
-	OUT_RING_CACHE(depth ? depth->offset : color[0]->offset);
-
-	/* Always set to bottom left of buffer */
-	/*BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_X, 4);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf ((GLfloat) h);
-	OUT_RING_CACHEf (0.0);
-	OUT_RING_CACHEf (0.0);*/
+	OUT_RING(format);
+	OUT_RING(pitch);
+	OUT_RING(color[0]->offset);
+	OUT_RING(depth ? depth->offset : color[0]->offset);
 
 	return GL_TRUE;
 }
@@ -794,4 +1076,6 @@ void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 	nmesa->hw_func.InitCard		= nv10InitCard;
 	nmesa->hw_func.BindBuffers	= nv10BindBuffers;
 	nmesa->hw_func.WindowMoved	= nv10WindowMoved;
+	nmesa->hw_func.UpdateProjectionMatrix = nv10UpdateProjectionMatrix;
+	nmesa->hw_func.UpdateModelviewMatrix = nv10UpdateModelviewMatrix;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
index 611469b6e41..42b3d666fb9 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
@@ -54,9 +54,53 @@ static void nv10RasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
 static void nv10RenderPrimitive( GLcontext *ctx, GLenum prim );
 static void nv10ResetLineStipple( GLcontext *ctx );
 
+static const int default_attr_size[8]={3,3,3,4,3,1,4,4};
 
+/* Mesa requires us to put pos attribute as the first attribute of the
+ * vertex, but on NV10 it is the last attribute.
+ * To fix that we put the pos attribute first, and we swap the pos
+ * attribute before sending it to the card.
+ * Speed cost of the swap seems negligeable
+ */
+#if 0
+/* old stuff where pos attribute isn't put first for mesa.
+ * Usefull for speed comparaison
+ */
+#define INV_VERT(i) i
+#define OUT_RING_VERTp(nmesa, ptr,sz, vertex_size) OUT_RINGp(ptr,sz)
+#define OUT_RING_VERT(nmesa, ptr, vertex_size) OUT_RINGp(ptr,vertex_size)
+#else
+
+#define INV_VERT(i) (i==0?7:i-1)
+
+#define OUT_RING_VERT_RAW(ptr,vertex_size) do{						\
+	/* if the vertex size is not null, we have at least pos attribute */ \
+	OUT_RINGp((GLfloat *)(ptr) + default_attr_size[_TNL_ATTRIB_POS], (vertex_size) - default_attr_size[_TNL_ATTRIB_POS]); \
+	OUT_RINGp((GLfloat *)(ptr), default_attr_size[_TNL_ATTRIB_POS]); \
+}while(0)
+
+#define OUT_RING_VERT(nmesa,ptr,vertex_size) do{ \
+	if (nmesa->screen->card->type>=NV_20) \
+		OUT_RINGp(ptr, vertex_size); \
+	else \
+		OUT_RING_VERT_RAW(ptr, vertex_size); \
+}while(0)
 
-static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t primitive,uint32_t size)
+
+#define OUT_RING_VERTp(nmesa, ptr,sz, vertex_size) do{						\
+	int nb_vert; \
+	if (nmesa->screen->card->type>=NV_20) \
+		OUT_RINGp(ptr, sz); \
+	else \
+		for (nb_vert = 0; nb_vert < (sz)/(vertex_size); nb_vert++) { \
+			OUT_RING_VERT_RAW((GLfloat*)(ptr)+nb_vert*(vertex_size), vertex_size); \
+		} \
+}while(0)
+
+#endif
+
+
+static inline void nv10StartPrimitive(struct nouveau_context* nmesa,GLuint primitive,GLuint size)
 {
 	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
 		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
@@ -74,7 +118,7 @@ static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t pri
 		BEGIN_RING_SIZE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_DATA|NONINC_METHOD,size);
 }
 
-inline void nv10FinishPrimitive(struct nouveau_context *nmesa)
+void nv10FinishPrimitive(struct nouveau_context *nmesa)
 {
 	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17))
 		BEGIN_RING_SIZE(NvSub3D,NV10_TCL_PRIMITIVE_3D_BEGIN_END,1);
@@ -103,13 +147,13 @@ static inline void nv10ExtendPrimitive(struct nouveau_context* nmesa, int size)
 static inline void nv10_render_generic_primitive_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags,GLuint prim)
 {
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
 	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(count-start)/4;
+	GLuint size_dword = vertsize*(count-start);
 
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,prim+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(start*vertsize)),size_dword);
+	OUT_RING_VERTp(nmesa, (nouveauVertex*)(vertptr+(start*vertsize)),size_dword, vertsize);
 	nv10FinishPrimitive(nmesa);
 }
 
@@ -189,16 +233,16 @@ static void (*nv10_render_tab_verts[GL_POLYGON+2])(GLcontext *,
 static inline void nv10_render_generic_primitive_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags,GLuint prim)
 {
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
 	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(count-start)/4;
+	GLuint size_dword = vertsize*(count-start);
 	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
 	GLuint j;
 
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,prim+1,size_dword);
 	for (j=start; j<count; j++ ) {
-		OUT_RINGp((nouveauVertex*)(vertptr+(elt[j]*vertsize)),vertsize/4);
+		OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(elt[j]*vertsize)),vertsize);
 	}
 	nv10FinishPrimitive(nmesa);
 }
@@ -288,77 +332,80 @@ do {									\
    nmesa->vertex_attr_count++;						\
 } while (0)
 
-static void nv10_render_clipped_line(GLcontext *ctx,GLuint ii,GLuint jj)
+static inline void nv10_render_point(GLcontext *ctx, GLfloat *vertptr)
 {
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLuint vertsize = nmesa->vertex_size;
+	GLuint size_dword = vertsize;
 
-}
-
-static void nv10_render_clipped_poly(GLcontext *ctx,const GLuint *elts,GLuint n)
-{
-	TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *VB = &tnl->vb;
-	GLuint *tmp = VB->Elts;
-	VB->Elts = (GLuint *)elts;
-	nv10_render_generic_primitive_elts( ctx, 0, n, PRIM_BEGIN|PRIM_END,GL_POLYGON );
-	VB->Elts = tmp;
+	nv10ExtendPrimitive(nmesa, size_dword);
+	nv10StartPrimitive(nmesa,GL_POINTS+1,size_dword);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr),vertsize);
+	nv10FinishPrimitive(nmesa);
 }
 
 static inline void nv10_render_points(GLcontext *ctx,GLuint first,GLuint last)
 {
-	WARN_ONCE("Unimplemented\n");
+	struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	GLuint i;
+
+	if (VB->Elts) {
+		for (i = first; i < last; i++)
+			if (VB->ClipMask[VB->Elts[i]] == 0)
+				nv10_render_point(ctx, vertptr + (VB->Elts[i]*vertsize));
+	}
+	else {
+		for (i = first; i < last; i++)
+			if (VB->ClipMask[i] == 0)
+				nv10_render_point(ctx, vertptr + (i*vertsize));
+	}
 }
 
 static inline void nv10_render_line(GLcontext *ctx,GLuint v1,GLuint v2)
 {
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
 	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(2)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
+	GLuint size_dword = vertsize*2;
 
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,GL_LINES+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
 	nv10FinishPrimitive(nmesa);
 }
 
 static inline void nv10_render_triangle(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3)
 {
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
 	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(3)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
+	GLuint size_dword = vertsize*3;
 
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,GL_TRIANGLES+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
 	nv10FinishPrimitive(nmesa);
 }
 
 static inline void nv10_render_quad(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3,GLuint v4)
 {
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
-	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLfloat *vertptr = (GLfloat *)nmesa->verts;
 	GLuint vertsize = nmesa->vertex_size;
-	GLuint size_dword = vertsize*(4)/4;
-
-	/* OUT_RINGp wants size in DWORDS */
-	vertsize >>= 2;
+	GLuint size_dword = vertsize*4;
 
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,GL_QUADS+1,size_dword);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
-	OUT_RINGp((nouveauVertex*)(vertptr+(v4*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v1*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v2*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v3*vertsize)),vertsize);
+	OUT_RING_VERT(nmesa, (nouveauVertex*)(vertptr+(v4*vertsize)),vertsize);
 	nv10FinishPrimitive(nmesa);
 }
 
@@ -371,8 +418,8 @@ static void nv10ChooseRenderState(GLcontext *ctx)
 
 	tnl->Driver.Render.PrimTabVerts = nv10_render_tab_verts;
 	tnl->Driver.Render.PrimTabElts = nv10_render_tab_elts;
-	tnl->Driver.Render.ClippedLine = nv10_render_clipped_line;
-	tnl->Driver.Render.ClippedPolygon = nv10_render_clipped_poly;
+	tnl->Driver.Render.ClippedLine = _tnl_RenderClippedLine;
+	tnl->Driver.Render.ClippedPolygon = _tnl_RenderClippedPolygon;
 	tnl->Driver.Render.Points = nv10_render_points;
 	tnl->Driver.Render.Line = nv10_render_line;
 	tnl->Driver.Render.Triangle = nv10_render_triangle;
@@ -388,7 +435,12 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	DECLARE_RENDERINPUTS(index);
 	struct vertex_buffer *VB = &tnl->vb;
 	int attr_size[16];
-	int default_attr_size[8]={3,3,3,4,3,1,4,4};
+	const int nv10_vtx_attribs[8]={
+		_TNL_ATTRIB_FOG, _TNL_ATTRIB_WEIGHT,
+		_TNL_ATTRIB_NORMAL, _TNL_ATTRIB_TEX1,
+		_TNL_ATTRIB_TEX0, _TNL_ATTRIB_COLOR1,
+		_TNL_ATTRIB_COLOR0, _TNL_ATTRIB_POS
+	};
 	int i;
 	int slots=0;
 	int total_size=0;
@@ -417,31 +469,55 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	/*
 	 * Tell t_vertex about the vertex format
 	 */
-	for(i=0;i<16;i++)
-	{
-		if (RENDERINPUTS_TEST(index, i))
+	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) {
+		for(i=0;i<8;i++) {
+			int j = nv10_vtx_attribs[INV_VERT(i)];
+			if (RENDERINPUTS_TEST(index, j)) {
+				switch(attr_size[j])
+				{
+					case 1:
+						EMIT_ATTR(j,EMIT_1F);
+						break;
+					case 2:
+						EMIT_ATTR(j,EMIT_2F);
+						break;
+					case 3:
+						EMIT_ATTR(j,EMIT_3F);
+						break;
+					case 4:
+						EMIT_ATTR(j,EMIT_4F);
+						break;
+				}
+				total_size+=attr_size[j];
+			}
+		}
+	} else {
+		for(i=0;i<16;i++)
 		{
-			slots=i+1;
-			switch(attr_size[i])
+			if (RENDERINPUTS_TEST(index, i))
 			{
-				case 1:
-					EMIT_ATTR(i,EMIT_1F);
-					break;
-				case 2:
-					EMIT_ATTR(i,EMIT_2F);
-					break;
-				case 3:
-					EMIT_ATTR(i,EMIT_3F);
-					break;
-				case 4:
-					EMIT_ATTR(i,EMIT_4F);
-					break;
+				slots=i+1;
+				switch(attr_size[i])
+				{
+					case 1:
+						EMIT_ATTR(i,EMIT_1F);
+						break;
+					case 2:
+						EMIT_ATTR(i,EMIT_2F);
+						break;
+					case 3:
+						EMIT_ATTR(i,EMIT_3F);
+						break;
+					case 4:
+						EMIT_ATTR(i,EMIT_4F);
+						break;
+				}
+				if (i==_TNL_ATTRIB_COLOR0)
+					nmesa->color_offset=total_size;
+				if (i==_TNL_ATTRIB_COLOR1)
+					nmesa->specular_offset=total_size;
+				total_size+=attr_size[i];
 			}
-			if (i==_TNL_ATTRIB_COLOR0)
-				nmesa->color_offset=total_size;
-			if (i==_TNL_ATTRIB_COLOR1)
-				nmesa->specular_offset=total_size;
-			total_size+=attr_size[i];
 		}
 	}
 
@@ -449,33 +525,32 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 			nmesa->vertex_attrs, 
 			nmesa->vertex_attr_count,
 			NULL, 0 );
-	assert(nmesa->vertex_size==total_size*4);
+	/* OUT_RINGp wants size in DWORDS */
+	nmesa->vertex_size = nmesa->vertex_size / 4;
+	assert(nmesa->vertex_size==total_size);
 
 	/* 
 	 * Tell the hardware about the vertex format
 	 */
 	if ((nmesa->screen->card->type>=NV_10) && (nmesa->screen->card->type<=NV_17)) {
-		int size;
+		int total_stride = 0;
 
 #define NV_VERTEX_ATTRIBUTE_TYPE_FLOAT 2
 
-#define NV10_SET_VERTEX_ATTRIB(i,j) \
-	do {	\
-		size = attr_size[j] << 4;	\
-		size |= (attr_size[j]*4) << 8;	\
-		size |= NV_VERTEX_ATTRIBUTE_TYPE_FLOAT;	\
-		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR(i),1);	\
-		OUT_RING_CACHE(size);	\
-	} while (0)
-
-		NV10_SET_VERTEX_ATTRIB(0, _TNL_ATTRIB_POS);
-		NV10_SET_VERTEX_ATTRIB(1, _TNL_ATTRIB_COLOR0);
-		NV10_SET_VERTEX_ATTRIB(2, _TNL_ATTRIB_COLOR1);
-		NV10_SET_VERTEX_ATTRIB(3, _TNL_ATTRIB_TEX0);
-		NV10_SET_VERTEX_ATTRIB(4, _TNL_ATTRIB_TEX1);
-		NV10_SET_VERTEX_ATTRIB(5, _TNL_ATTRIB_NORMAL);
-		NV10_SET_VERTEX_ATTRIB(6, _TNL_ATTRIB_WEIGHT);
-		NV10_SET_VERTEX_ATTRIB(7, _TNL_ATTRIB_FOG);
+		for(i=0;i<8;i++) {
+			int j = nv10_vtx_attribs[i];
+			int size;
+			int stride = attr_size[j] << 2;
+			if (j==_TNL_ATTRIB_POS) {
+				stride += total_stride;
+			}
+			size = attr_size[j] << 4;
+			size |= stride << 8;
+			size |= NV_VERTEX_ATTRIBUTE_TYPE_FLOAT;
+			BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR((7-i)),1);
+			OUT_RING_CACHE(size);
+			total_stride += stride;
+		}
 
 		BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VERTEX_ARRAY_VALIDATE,1);
 		OUT_RING_CACHE(0);
@@ -489,7 +564,7 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	} else {
 		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DO_VERTICES, 1);
 		OUT_RING(0);
-		BEGIN_RING_CACHE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS,slots);
+		BEGIN_RING_CACHE(NvSub3D,NV20_TCL_PRIMITIVE_3D_VB_POINTER_ATTR8_TX0,slots);
 		for(i=0;i<slots;i++)
 		{
 			int size=attr_size[i];
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state.c b/src/mesa/drivers/dri/nouveau/nv20_state.c
index ccf2f6148b4..6b583980a49 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state.c
@@ -728,8 +728,8 @@ static GLboolean nv20InitCard(nouveauContextPtr nmesa)
 
 /* Update buffer offset/pitch/format */
 static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color,
-				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+				 nouveau_renderbuffer_t **color,
+				 nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 	GLuint pitch, format, depth_pitch;
diff --git a/src/mesa/drivers/dri/nouveau/nv30_state.c b/src/mesa/drivers/dri/nouveau/nv30_state.c
index 9b010954b33..cd3ee986880 100644
--- a/src/mesa/drivers/dri/nouveau/nv30_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv30_state.c
@@ -905,9 +905,9 @@ static GLboolean nv40InitCard(nouveauContextPtr nmesa)
 	return GL_TRUE;
 }
 
-static GLboolean nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
-		nouveau_renderbuffer **color,
-		nouveau_renderbuffer *depth)
+static GLboolean
+nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
+		nouveau_renderbuffer_t **color, nouveau_renderbuffer_t *depth)
 {
 	GLuint x, y, w, h;
 
diff --git a/src/mesa/drivers/dri/nouveau/nv50_state.c b/src/mesa/drivers/dri/nouveau/nv50_state.c
index 818e268615c..a9236f093c3 100644
--- a/src/mesa/drivers/dri/nouveau/nv50_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv50_state.c
@@ -584,9 +584,9 @@ static GLboolean nv50InitCard(nouveauContextPtr nmesa)
 	return GL_FALSE;
 }
 
-static GLboolean nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
-      				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
+static GLboolean
+nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
+		nouveau_renderbuffer_t **color, nouveau_renderbuffer_t *depth)
 {
 	return GL_FALSE;
 }
diff --git a/src/mesa/drivers/dri/r200/r200_context.h b/src/mesa/drivers/dri/r200/r200_context.h
index bec09e8ef6a..c80180bdbcc 100644
--- a/src/mesa/drivers/dri/r200/r200_context.h
+++ b/src/mesa/drivers/dri/r200/r200_context.h
@@ -179,6 +179,7 @@ struct r200_tex_obj {
 
    drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
 					/* Six, for the cube faces */
+   GLboolean image_override;		/* Image overridden by GLX_EXT_tfp */
 
    GLuint pp_txfilter;		        /* hardware register values */
    GLuint pp_txformat;
diff --git a/src/mesa/drivers/dri/r200/r200_fragshader.c b/src/mesa/drivers/dri/r200/r200_fragshader.c
index 5dd3adaef69..d514b28219a 100644
--- a/src/mesa/drivers/dri/r200/r200_fragshader.c
+++ b/src/mesa/drivers/dri/r200/r200_fragshader.c
@@ -24,13 +24,13 @@
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *
  **************************************************************************/
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
 
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
 #include "tnl/t_context.h"
-#include "atifragshader.h"
-#include "program.h"
+#include "shader/atifragshader.h"
+#include "shader/program.h"
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_tex.h"
diff --git a/src/mesa/drivers/dri/r200/r200_tex.h b/src/mesa/drivers/dri/r200/r200_tex.h
index e6c0e00eb07..10ff8e8a660 100644
--- a/src/mesa/drivers/dri/r200/r200_tex.h
+++ b/src/mesa/drivers/dri/r200/r200_tex.h
@@ -35,6 +35,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #ifndef __R200_TEX_H__
 #define __R200_TEX_H__
 
+extern void r200SetTexOffset(__DRIcontext *pDRICtx, GLint texname,
+			     unsigned long long offset, GLint depth,
+			     GLuint pitch);
+
 extern void r200UpdateTextureState( GLcontext *ctx );
 
 extern int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t, GLuint face );
diff --git a/src/mesa/drivers/dri/r200/r200_texmem.c b/src/mesa/drivers/dri/r200/r200_texmem.c
index d926313d576..183c4ca424b 100644
--- a/src/mesa/drivers/dri/r200/r200_texmem.c
+++ b/src/mesa/drivers/dri/r200/r200_texmem.c
@@ -181,7 +181,8 @@ static void r200UploadRectSubImage( r200ContextPtr rmesa,
       /* In this case, could also use GART texturing.  This is
        * currently disabled, but has been tested & works.
        */
-      t->pp_txoffset = r200GartOffsetFromVirtual( rmesa, texImage->Data );
+      if ( !t->image_override )
+         t->pp_txoffset = r200GartOffsetFromVirtual( rmesa, texImage->Data );
       t->pp_txpitch = texImage->RowStride * texFormat->TexelBytes - 32;
 
       if (R200_DEBUG & DEBUG_TEXTURE)
@@ -467,7 +468,7 @@ int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t, GLuint face )
 	       t->base.firstLevel, t->base.lastLevel );
    }
 
-   if ( !t || t->base.totalSize == 0 )
+   if ( !t || t->base.totalSize == 0 || t->image_override )
       return 0;
 
    if (R200_DEBUG & DEBUG_SYNC) {
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index ae02ec4b638..93c0fb7a65f 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -37,6 +37,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "context.h"
 #include "macros.h"
 #include "texformat.h"
+#include "texobj.h"
 #include "enums.h"
 
 #include "r200_context.h"
@@ -72,10 +73,11 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define VALID_FORMAT(f) ( ((f) <= MESA_FORMAT_RGBA_DXT5) \
 			     && (tx_table_le[f].format != 0xffffffff) )
 
-static const struct {
+struct tx_table {
    GLuint format, filter;
-}
-tx_table_be[] =
+};
+
+static const struct tx_table tx_table_be[] =
 {
    [ MESA_FORMAT_RGBA8888 ] = { R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP, 0 },
    _ALPHA_REV(RGBA8888),
@@ -104,16 +106,13 @@ tx_table_be[] =
    _ALPHA(RGBA_DXT5),
 };
 
-static const struct {
-   GLuint format, filter;
-}
-tx_table_le[] =
+static const struct tx_table tx_table_le[] =
 {
    _ALPHA(RGBA8888),
    [ MESA_FORMAT_RGBA8888_REV ] = { R200_TXFORMAT_ABGR8888 | R200_TXFORMAT_ALPHA_IN_MAP, 0 },
    _ALPHA(ARGB8888),
    _ALPHA_REV(ARGB8888),
-   _INVALID(RGB888),
+   [ MESA_FORMAT_RGB888 ] = { R200_TXFORMAT_ARGB8888, 0 },
    _COLOR(RGB565),
    _COLOR_REV(RGB565),
    _ALPHA(ARGB4444),
@@ -160,30 +159,23 @@ static void r200SetTexImages( r200ContextPtr rmesa,
    GLint i, texelBytes;
    GLint numLevels;
    GLint log2Width, log2Height, log2Depth;
-   const GLuint ui = 1;
-   const GLubyte littleEndian = *((const GLubyte *) &ui);
 
    /* Set the hardware texture format
     */
+   if ( !t->image_override ) {
+      if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
+         t->pp_txformat &= ~(R200_TXFORMAT_FORMAT_MASK |
+                             R200_TXFORMAT_ALPHA_IN_MAP);
+         t->pp_txfilter &= ~R200_YUV_TO_RGB;
 
-   t->pp_txformat &= ~(R200_TXFORMAT_FORMAT_MASK |
-		       R200_TXFORMAT_ALPHA_IN_MAP);
-   t->pp_txfilter &= ~R200_YUV_TO_RGB;
-
-   if ( VALID_FORMAT( baseImage->TexFormat->MesaFormat ) ) {
-      if (littleEndian) {
 	 t->pp_txformat |= tx_table_le[ baseImage->TexFormat->MesaFormat ].format;
 	 t->pp_txfilter |= tx_table_le[ baseImage->TexFormat->MesaFormat ].filter;
       }
       else {
-	 t->pp_txformat |= tx_table_be[ baseImage->TexFormat->MesaFormat ].format;
-	 t->pp_txfilter |= tx_table_be[ baseImage->TexFormat->MesaFormat ].filter;
+         _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
+         return;
       }
    }
-   else {
-      _mesa_problem(NULL, "unexpected texture format in %s", __FUNCTION__);
-      return;
-   }
 
    texelBytes = baseImage->TexFormat->TexelBytes;
 
@@ -380,11 +372,13 @@ static void r200SetTexImages( r200ContextPtr rmesa,
     * requires 64-byte aligned pitches, and we may/may not need the
     * blitter.   NPOT only!
     */
-   if (baseImage->IsCompressed)
-      t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
-   else
-      t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
-   t->pp_txpitch -= 32;
+   if ( !t->image_override ) {
+      if (baseImage->IsCompressed)
+         t->pp_txpitch = (tObj->Image[0][t->base.firstLevel]->Width + 63) & ~(63);
+      else
+         t->pp_txpitch = ((tObj->Image[0][t->base.firstLevel]->Width * texelBytes) + 63) & ~(63);
+      t->pp_txpitch -= 32;
+   }
 
    t->dirty_state = TEX_ALL;
 
@@ -979,6 +973,46 @@ static GLboolean r200UpdateTextureEnv( GLcontext *ctx, int unit, int slot, GLuin
    return GL_TRUE;
 }
 
+void r200SetTexOffset(__DRIcontext * pDRICtx, GLint texname,
+		      unsigned long long offset, GLint depth, GLuint pitch)
+{
+	r200ContextPtr rmesa =
+	    (r200ContextPtr) ((__DRIcontextPrivate *) pDRICtx->private)->
+	    driverPrivate;
+	struct gl_texture_object *tObj =
+	    _mesa_lookup_texture(rmesa->glCtx, texname);
+	r200TexObjPtr t;
+
+	if (!tObj)
+		return;
+
+	t = (r200TexObjPtr) tObj->DriverData;
+
+	t->image_override = GL_TRUE;
+
+	if (!offset)
+		return;
+
+	t->pp_txoffset = offset;
+	t->pp_txpitch = pitch - 32;
+
+	switch (depth) {
+	case 32:
+		t->pp_txformat = tx_table_le[2].format;
+		t->pp_txfilter |= tx_table_le[2].filter;
+		break;
+	case 24:
+	default:
+		t->pp_txformat = tx_table_le[4].format;
+		t->pp_txfilter |= tx_table_le[4].filter;
+		break;
+	case 16:
+		t->pp_txformat = tx_table_le[5].format;
+		t->pp_txfilter |= tx_table_le[5].filter;
+		break;
+	}
+}
+
 #define REF_COLOR 1
 #define REF_ALPHA 2
 
@@ -1560,7 +1594,7 @@ static GLboolean enable_tex_2d( GLcontext *ctx, int unit )
       R200_FIREVERTICES( rmesa );
       r200SetTexImages( rmesa, tObj );
       r200UploadTexImages( rmesa, (r200TexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock ) 
+      if ( !t->base.memBlock && !t->image_override ) 
 	 return GL_FALSE;
    }
 
@@ -1668,7 +1702,9 @@ static GLboolean enable_tex_rect( GLcontext *ctx, int unit )
       R200_FIREVERTICES( rmesa );
       r200SetTexImages( rmesa, tObj );
       r200UploadTexImages( rmesa, (r200TexObjPtr) tObj->DriverData, 0 );
-      if ( !t->base.memBlock && !rmesa->prefer_gart_client_texturing ) 
+      if ( !t->base.memBlock &&
+           !t->image_override &&
+           !rmesa->prefer_gart_client_texturing ) 
 	 return GL_FALSE;
    }
 
diff --git a/src/mesa/drivers/dri/r200/r200_vertprog.c b/src/mesa/drivers/dri/r200/r200_vertprog.c
index 6089d617c6b..604b9c6caec 100644
--- a/src/mesa/drivers/dri/r200/r200_vertprog.c
+++ b/src/mesa/drivers/dri/r200/r200_vertprog.c
@@ -30,10 +30,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Aapo Tahkola <[email protected]>
  *   Roland Scheidegger <[email protected]>
  */
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "program.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/program.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
diff --git a/src/mesa/drivers/dri/r300/r300_shader.c b/src/mesa/drivers/dri/r300/r300_shader.c
index 5f5ac7c4c71..77abf86a8ee 100644
--- a/src/mesa/drivers/dri/r300/r300_shader.c
+++ b/src/mesa/drivers/dri/r300/r300_shader.c
@@ -1,8 +1,7 @@
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
 
-#include "program.h"
+#include "main/glheader.h"
+
+#include "shader/program.h"
 #include "tnl/tnl.h"
 #include "r300_context.h"
 #include "r300_fragprog.h"
diff --git a/src/mesa/drivers/dri/r300/r300_swtcl.c b/src/mesa/drivers/dri/r300/r300_swtcl.c
index c949f33bf33..a732bdb5598 100644
--- a/src/mesa/drivers/dri/r300/r300_swtcl.c
+++ b/src/mesa/drivers/dri/r300/r300_swtcl.c
@@ -40,6 +40,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "enums.h"
 #include "image.h"
 #include "imports.h"
+#include "light.h"
 #include "macros.h"
 
 #include "swrast/s_context.h"
diff --git a/src/mesa/drivers/dri/r300/r300_texmem.c b/src/mesa/drivers/dri/r300/r300_texmem.c
index 38f0da8b7c0..723601ac4a6 100644
--- a/src/mesa/drivers/dri/r300/r300_texmem.c
+++ b/src/mesa/drivers/dri/r300/r300_texmem.c
@@ -505,7 +505,7 @@ int r300UploadTexImages(r300ContextPtr rmesa, r300TexObjPtr t, GLuint face)
 			t->base.lastLevel);
 	}
 
-	if (!t || t->base.totalSize == 0)
+	if (t->base.totalSize == 0)
 		return 0;
 
 	if (RADEON_DEBUG & DEBUG_SYNC) {
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 7d4e8c95112..4dd3fd6a673 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -35,10 +35,10 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
  * For a description of the vertex program instruction set see r300_reg.h.
  */
 
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-#include "program.h"
+#include "main/glheader.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "shader/program.h"
 #include "shader/prog_instruction.h"
 #include "shader/prog_parameter.h"
 #include "shader/prog_statevars.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index 9451ec4aa5b..b302275c714 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -35,14 +35,15 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  *   Keith Whitwell <[email protected]>
  */
 
-#include "glheader.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "simple_list.h"
-#include "imports.h"
-#include "matrix.h"
-#include "extensions.h"
-#include "framebuffer.h"
+#include "main/glheader.h"
+#include "main/api_arrayelt.h"
+#include "main/context.h"
+#include "main/simple_list.h"
+#include "main/imports.h"
+#include "main/matrix.h"
+#include "main/extensions.h"
+#include "main/framebuffer.h"
+#include "main/state.h"
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index aa7fb633dd1..682cf3a5ee5 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -53,6 +53,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_context.h"
 #include "r200_ioctl.h"
 #include "r200_span.h"
+#include "r200_tex.h"
 #elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
 #include "r300_context.h"
 #include "r300_fragprog.h"
@@ -973,7 +974,8 @@ static const struct __DriverAPIRec r200API = {
    .WaitForMSC      = driWaitForMSC32,
    .WaitForSBC      = NULL,
    .SwapBuffersMSC  = NULL,
-   .CopySubBuffer   = r200CopySubBuffer
+   .CopySubBuffer   = r200CopySubBuffer,
+   .setTexOffset    = r200SetTexOffset
 };
 #endif
 
diff --git a/src/mesa/drivers/dri/sis/sis_tex.c b/src/mesa/drivers/dri/sis/sis_tex.c
index be87f16e298..8fd7b26ba55 100644
--- a/src/mesa/drivers/dri/sis/sis_tex.c
+++ b/src/mesa/drivers/dri/sis/sis_tex.c
@@ -130,6 +130,8 @@ sisAllocTexImage( sisContextPtr smesa, sisTexObjPtr t, int level,
 static void
 sisFreeTexImage( sisContextPtr smesa, sisTexObjPtr t, int level )
 {
+   assert(level >= 0);
+   assert(level < SIS_MAX_TEXTURE_LEVELS);
    if (t->image[level].Data == NULL)
       return;
 
@@ -213,7 +215,7 @@ sisDeleteTexture( GLcontext * ctx, struct gl_texture_object *texObj )
        */
       return;
    }
-   for (i = 0; i < MAX_TEXTURE_LEVELS; i++) {
+   for (i = 0; i < SIS_MAX_TEXTURE_LEVELS; i++) {
       sisFreeTexImage( smesa, t, i );
    }
 
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_tex.c b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
index 89865d96379..64946385b6d 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_tex.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_tex.c
@@ -1822,7 +1822,7 @@ tdfxTestProxyTexImage(GLcontext *ctx, GLenum target,
             tdfxTexInfo *ti;
             int memNeeded;
 
-            tObj = ctx->Texture.Proxy2D;
+            tObj = ctx->Texture.ProxyTex[TEXTURE_2D_INDEX];
             if (!tObj->DriverData)
                 tObj->DriverData = fxAllocTexObjData(fxMesa);
             ti = TDFX_TEXTURE_DATA(tObj);
diff --git a/src/mesa/drivers/dri/unichrome/via_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index 4d25d328e33..7c738777209 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -733,14 +733,15 @@ void viaXMesaWindowMoved(struct via_context *vmesa)
 {
    __DRIdrawablePrivate *const drawable = vmesa->driDrawable;
    __DRIdrawablePrivate *const readable = vmesa->driReadable;
-   struct via_renderbuffer *const draw_buffer = 
-     (struct via_renderbuffer *) drawable->driverPrivate;
-   struct via_renderbuffer *const read_buffer =
-     (struct via_renderbuffer *) readable->driverPrivate;
+   struct via_renderbuffer * draw_buffer;
+   struct via_renderbuffer * read_buffer;
    GLuint bytePerPixel = vmesa->viaScreen->bitsPerPixel >> 3;
 
    if (!drawable)
       return;
+
+   draw_buffer =  (struct via_renderbuffer *) drawable->driverPrivate;
+   read_buffer =  (struct via_renderbuffer *) readable->driverPrivate;
    
    switch (vmesa->glCtx->DrawBuffer->_ColorDrawBufferMask[0]) {
    case BUFFER_BIT_BACK_LEFT: 
diff --git a/src/mesa/drivers/windows/gdi/wgl.c b/src/mesa/drivers/windows/gdi/wgl.c
index dad3dc11604..17eed1dc329 100644
--- a/src/mesa/drivers/windows/gdi/wgl.c
+++ b/src/mesa/drivers/windows/gdi/wgl.c
@@ -601,8 +601,8 @@ WINGDIAPI HGLRC GLAPIENTRY wglCreateLayerContext(HDC hdc,
 WINGDIAPI BOOL GLAPIENTRY wglShareLists(HGLRC hglrc1,
 					HGLRC hglrc2)
 {
-    (void) hglrc1; (void) hglrc2;
-    return(TRUE);
+	WMesaShareLists(hglrc1, hglrc2);
+	return(TRUE);
 }
 
 
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 5b67439f0f2..ea3f1780d3c 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -1401,6 +1401,11 @@ void WMesaSwapBuffers( HDC hdc )
     }
 }
 
+void WMesaShareLists(WMesaContext ctx_to_share, WMesaContext ctx)
+{
+	_mesa_share_state(&ctx->gl_ctx, &ctx_to_share->gl_ctx);	
+}
+
 /* This is hopefully a temporary hack to define some needed dispatch
  * table entries.  Hopefully, I'll find a better solution.  The
  * dispatch table generation scripts ought to be making these dummy