Merge git://proxy01.pd.intel.com:9419/git/mesa/mesa into crestline

author: Nian Wu <nian@graphics.(none)> 2007-02-25 09:40:28 -0800
committer: Nian Wu <nian@graphics.(none)> 2007-02-25 09:40:28 -0800
commit: 675f7f627bec92315bf168a9c872ffc05f88c69c (patch)
tree: 19c8f9cc509dc7ce73a0709832f34bf13a2c7425 /src/mesa/drivers
parent: 51bfb8fc8c78bb066d24e6ecbc20f00af7210386 (diff)
parent: c080123998f28d9317331aec7ddfcd1074b29daf (diff)
202 files changed, 6481 insertions, 17452 deletions
diff --git a/src/mesa/drivers/beos/GLView.cpp b/src/mesa/drivers/beos/GLView.cpp
index 45473a8ef27..9e4a7ebe56e 100644
--- a/src/mesa/drivers/beos/GLView.cpp
+++ b/src/mesa/drivers/beos/GLView.cpp
@@ -43,7 +43,7 @@ extern "C" {
 #include "texobj.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/s_context.h"
@@ -344,7 +344,7 @@ BGLView::BGLView(BRect rect, char *name,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext(ctx);
-   _ac_CreateContext(ctx);
+   _vbo_CreateContext(ctx);
    _tnl_CreateContext(ctx);
    _swsetup_CreateContext(ctx);
    _swsetup_Wakeup(ctx);
@@ -830,7 +830,7 @@ void MesaDriver::UpdateState( GLcontext *ctx, GLuint new_state )
 
 	_swrast_InvalidateState( ctx, new_state );
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 
 	if (ctx->Color.DrawBuffer[0] == GL_FRONT) {
diff --git a/src/mesa/drivers/directfb/idirectfbgl_mesa.c b/src/mesa/drivers/directfb/idirectfbgl_mesa.c
index 3c8c6c256e6..88ac4bb31d5 100644
--- a/src/mesa/drivers/directfb/idirectfbgl_mesa.c
+++ b/src/mesa/drivers/directfb/idirectfbgl_mesa.c
@@ -1,18 +1,19 @@
 /*
- * Copyright (C) 2004-2006 Claudio Ciccani <[email protected]>
+ * Copyright (C) 2004-2007 Claudio Ciccani <[email protected]>
  *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- *
- * This program is distributed in the hope that it will be useful,
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ * 
+ * This library is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ * 
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this library; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  *
  * 
  * Based on glfbdev.c, written by Brian Paul.
@@ -45,7 +46,7 @@
 #include "texformat.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -373,7 +374,7 @@ dfbUpdateState( GLcontext *ctx, GLuint new_state )
 {
      _swrast_InvalidateState( ctx, new_state );
      _swsetup_InvalidateState( ctx, new_state );
-     _ac_InvalidateState( ctx, new_state );
+     _vbo_InvalidateState( ctx, new_state );
      _tnl_InvalidateState( ctx, new_state );
 }
 
@@ -786,7 +787,7 @@ directfbgl_create_context( GLcontext             *context,
      }
 
      _swrast_CreateContext( context );
-     _ac_CreateContext( context );
+     _vbo_CreateContext( context );
      _tnl_CreateContext( context );
      _swsetup_CreateContext( context );
      _swsetup_Wakeup( context );
diff --git a/src/mesa/drivers/dri/common/extension_helper.h b/src/mesa/drivers/dri/common/extension_helper.h
index d6d51cdd165..c7984964255 100644
--- a/src/mesa/drivers/dri/common/extension_helper.h
+++ b/src/mesa/drivers/dri/common/extension_helper.h
@@ -2466,10 +2466,10 @@ static const char MultiTexCoord4ivARB_names[] =
     "";
 #endif
 
-#if defined(need_GL_VERSION_2_0)
-static const char GetVertexAttribPointervARB_names[] = 
-    "iip\0" /* Parameter signature */
-    "glGetVertexAttribPointerv\0"
+#if defined(need_GL_EXT_gpu_program_parameters)
+static const char ProgramLocalParameters4fvEXT_names[] = 
+    "iiip\0" /* Parameter signature */
+    "glProgramLocalParameters4fvEXT\0"
     "";
 #endif
 
@@ -3152,13 +3152,6 @@ static const char VertexAttribPointerNV_names[] =
     "";
 #endif
 
-#if defined(need_GL_EXT_gpu_program_parameters)
-static const char ProgramLocalParameters4fvEXT_names[] = 
-    "iiip\0" /* Parameter signature */
-    "glProgramLocalParameters4fvEXT\0"
-    "";
-#endif
-
 #if defined(need_GL_EXT_framebuffer_object)
 static const char GetFramebufferAttachmentParameterivEXT_names[] = 
     "iiip\0" /* Parameter signature */
@@ -3781,9 +3774,10 @@ static const char ReplacementCodeuiColor3fVertex3fSUN_names[] =
     "";
 #endif
 
-#if defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
+#if defined(need_GL_VERSION_2_0) || defined(need_GL_ARB_vertex_program) || defined(need_GL_NV_vertex_program)
 static const char GetVertexAttribPointervNV_names[] = 
     "iip\0" /* Parameter signature */
+    "glGetVertexAttribPointerv\0"
     "glGetVertexAttribPointervARB\0"
     "glGetVertexAttribPointervNV\0"
     "";
@@ -5343,8 +5337,8 @@ static const struct dri_extension_function GL_EXT_framebuffer_object_functions[]
 
 #if defined(need_GL_EXT_gpu_program_parameters)
 static const struct dri_extension_function GL_EXT_gpu_program_parameters_functions[] = {
-    { ProgramEnvParameters4fvEXT_names, ProgramEnvParameters4fvEXT_remap_index, -1 },
     { ProgramLocalParameters4fvEXT_names, ProgramLocalParameters4fvEXT_remap_index, -1 },
+    { ProgramEnvParameters4fvEXT_names, ProgramEnvParameters4fvEXT_remap_index, -1 },
     { NULL, 0, 0 }
 };
 #endif
@@ -6263,7 +6257,6 @@ static const struct dri_extension_function GL_VERSION_2_0_functions[] = {
     { GetVertexAttribfvARB_names, GetVertexAttribfvARB_remap_index, -1 },
     { GetAttribLocationARB_names, GetAttribLocationARB_remap_index, -1 },
     { Uniform3ivARB_names, Uniform3ivARB_remap_index, -1 },
-    { GetVertexAttribPointervARB_names, GetVertexAttribPointervARB_remap_index, -1 },
     { VertexAttrib4sARB_names, VertexAttrib4sARB_remap_index, -1 },
     { VertexAttrib2dvARB_names, VertexAttrib2dvARB_remap_index, -1 },
     { VertexAttrib2fvARB_names, VertexAttrib2fvARB_remap_index, -1 },
@@ -6295,6 +6288,7 @@ static const struct dri_extension_function GL_VERSION_2_0_functions[] = {
     { Uniform4iARB_names, Uniform4iARB_remap_index, -1 },
     { UseProgramObjectARB_names, UseProgramObjectARB_remap_index, -1 },
     { DeleteProgram_names, DeleteProgram_remap_index, -1 },
+    { GetVertexAttribPointervNV_names, GetVertexAttribPointervNV_remap_index, -1 },
     { Uniform2iARB_names, Uniform2iARB_remap_index, -1 },
     { VertexAttrib4dARB_names, VertexAttrib4dARB_remap_index, -1 },
     { GetUniformLocationARB_names, GetUniformLocationARB_remap_index, -1 },
diff --git a/src/mesa/drivers/dri/fb/fb_dri.c b/src/mesa/drivers/dri/fb/fb_dri.c
index d6ba23bc6e6..08b52b4d8f7 100644
--- a/src/mesa/drivers/dri/fb/fb_dri.c
+++ b/src/mesa/drivers/dri/fb/fb_dri.c
@@ -50,7 +50,7 @@
 #include "extensions.h"
 #include "framebuffer.h"
 #include "renderbuffer.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -93,7 +93,7 @@ update_state( GLcontext *ctx, GLuint new_state )
    /* not much to do here - pass it on */
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
 }
 
@@ -365,7 +365,7 @@ fbCreateContext( const __GLcontextModes *glVisual,
 
    /* Create module contexts */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
    _swsetup_Wakeup( ctx );
@@ -399,7 +399,7 @@ fbDestroyContext( __DRIcontextPrivate *driContextPriv )
    if ( fbmesa ) {
       _swsetup_DestroyContext( fbmesa->glCtx );
       _tnl_DestroyContext( fbmesa->glCtx );
-      _ac_DestroyContext( fbmesa->glCtx );
+      _vbo_DestroyContext( fbmesa->glCtx );
       _swrast_DestroyContext( fbmesa->glCtx );
 
       /* free the Mesa context */
diff --git a/src/mesa/drivers/dri/fb/fb_egl.c b/src/mesa/drivers/dri/fb/fb_egl.c
index 5c74b0b0f47..517e71f8880 100644
--- a/src/mesa/drivers/dri/fb/fb_egl.c
+++ b/src/mesa/drivers/dri/fb/fb_egl.c
@@ -17,7 +17,7 @@
 #include "extensions.h"
 #include "framebuffer.h"
 #include "renderbuffer.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -388,7 +388,7 @@ update_state( GLcontext *ctx, GLuint new_state )
    /* not much to do here - pass it on */
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
 }
 
@@ -491,7 +491,7 @@ fbCreateContext(_EGLDriver *drv, EGLDisplay dpy, EGLConfig config, EGLContext sh
 
    /* Create module contexts */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
    _swsetup_Wakeup( ctx );
diff --git a/src/mesa/drivers/dri/ffb/ffb_state.c b/src/mesa/drivers/dri/ffb/ffb_state.c
index b81d94de252..eb13478166a 100644
--- a/src/mesa/drivers/dri/ffb/ffb_state.c
+++ b/src/mesa/drivers/dri/ffb/ffb_state.c
@@ -40,7 +40,7 @@
 #include "enums.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -1034,7 +1034,7 @@ static void ffbDDUpdateState(GLcontext *ctx, GLuint newstate)
 
 	_swrast_InvalidateState( ctx, newstate );
 	_swsetup_InvalidateState( ctx, newstate );
-	_ac_InvalidateState( ctx, newstate );
+	_vbo_InvalidateState( ctx, newstate );
 	_tnl_InvalidateState( ctx, newstate );
 
 	if (newstate & _NEW_TEXTURE)
diff --git a/src/mesa/drivers/dri/ffb/ffb_xmesa.c b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
index 99256d76b6e..215aaf8ffba 100644
--- a/src/mesa/drivers/dri/ffb/ffb_xmesa.c
+++ b/src/mesa/drivers/dri/ffb/ffb_xmesa.c
@@ -38,7 +38,7 @@
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "drivers/common/driverfuncs.h"
 
 #include "ffb_context.h"
@@ -277,7 +277,7 @@ ffbCreateContext(const __GLcontextModes *mesaVis,
 
 	/* Initialize the software rasterizer and helper modules. */
 	_swrast_CreateContext( ctx );
-	_ac_CreateContext( ctx );
+	_vbo_CreateContext( ctx );
 	_tnl_CreateContext( ctx );
 	_swsetup_CreateContext( ctx );
 
@@ -313,7 +313,7 @@ ffbDestroyContext(__DRIcontextPrivate *driContextPriv)
 
 		_swsetup_DestroyContext( fmesa->glCtx );
 		_tnl_DestroyContext( fmesa->glCtx );
-		_ac_DestroyContext( fmesa->glCtx );
+		_vbo_DestroyContext( fmesa->glCtx );
 		_swrast_DestroyContext( fmesa->glCtx );
 
                 /* free the Mesa context */
diff --git a/src/mesa/drivers/dri/gamma/gamma_context.c b/src/mesa/drivers/dri/gamma/gamma_context.c
index ffaf45459bc..b1dcbfcdcfa 100644
--- a/src/mesa/drivers/dri/gamma/gamma_context.c
+++ b/src/mesa/drivers/dri/gamma/gamma_context.c
@@ -28,7 +28,7 @@
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -147,7 +147,7 @@ GLboolean gammaCreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
diff --git a/src/mesa/drivers/dri/gamma/gamma_state.c b/src/mesa/drivers/dri/gamma/gamma_state.c
index 08749bf5d39..8dbe0a97caa 100644
--- a/src/mesa/drivers/dri/gamma/gamma_state.c
+++ b/src/mesa/drivers/dri/gamma/gamma_state.c
@@ -33,7 +33,7 @@
 #include "colormac.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 
 #define ENABLELIGHTING 0
@@ -1663,7 +1663,7 @@ static void gammaDDUpdateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    GAMMA_CONTEXT(ctx)->new_gl_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/gamma/gamma_xmesa.c b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
index 00f6aa32aeb..e8922b15036 100644
--- a/src/mesa/drivers/dri/gamma/gamma_xmesa.c
+++ b/src/mesa/drivers/dri/gamma/gamma_xmesa.c
@@ -34,7 +34,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 static GLboolean 
 gammaInitDriver(__DRIscreenPrivate *sPriv)
@@ -57,7 +57,7 @@ gammaDestroyContext(__DRIcontextPrivate *driContextPriv)
     if (gmesa) {
       _swsetup_DestroyContext( gmesa->glCtx );
       _tnl_DestroyContext( gmesa->glCtx );
-      _ac_DestroyContext( gmesa->glCtx );
+      _vbo_DestroyContext( gmesa->glCtx );
       _swrast_DestroyContext( gmesa->glCtx );
 
       gammaFreeVB( gmesa->glCtx );
diff --git a/src/mesa/drivers/dri/i810/i810context.c b/src/mesa/drivers/dri/i810/i810context.c
index c0f4efd7a31..db8f7a19a23 100644
--- a/src/mesa/drivers/dri/i810/i810context.c
+++ b/src/mesa/drivers/dri/i810/i810context.c
@@ -44,7 +44,7 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 
@@ -287,7 +287,7 @@ i810CreateContext( const __GLcontextModes *mesaVis,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -350,7 +350,7 @@ i810DestroyContext(__DRIcontextPrivate *driContextPriv)
       release_texture_heaps = (imesa->glCtx->Shared->RefCount == 1);
       _swsetup_DestroyContext( imesa->glCtx );
       _tnl_DestroyContext( imesa->glCtx );
-      _ac_DestroyContext( imesa->glCtx );
+      _vbo_DestroyContext( imesa->glCtx );
       _swrast_DestroyContext( imesa->glCtx );
 
       i810FreeVB( imesa->glCtx );
diff --git a/src/mesa/drivers/dri/i810/i810state.c b/src/mesa/drivers/dri/i810/i810state.c
index 6f9d4b5bd47..3ad25282d90 100644
--- a/src/mesa/drivers/dri/i810/i810state.c
+++ b/src/mesa/drivers/dri/i810/i810state.c
@@ -21,8 +21,8 @@
 #include "i810ioctl.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
 #include "tnl/tnl.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 
 #include "tnl/t_pipeline.h"
@@ -953,7 +953,7 @@ static void i810InvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    I810_CONTEXT(ctx)->new_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index 2f78fd60b21..6ec34e5bdef 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -36,7 +36,8 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
+
 
 #include "utils.h"
 #include "i915_reg.h"
@@ -63,7 +64,7 @@ static void i915InvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_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;
diff --git a/src/mesa/drivers/dri/i915/i915_state.c b/src/mesa/drivers/dri/i915/i915_state.c
index 3cec6a2ddfc..fd11e10652e 100644
--- a/src/mesa/drivers/dri/i915/i915_state.c
+++ b/src/mesa/drivers/dri/i915/i915_state.c
@@ -611,10 +611,12 @@ void i915_update_fog( GLcontext *ctx )
 	 i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_FOG_ENABLE;
    }
 
-   if (enabled) {
-      _tnl_allow_vertex_fog( ctx, (i915->vertex_fog == I915_FOG_VERTEX) ); 
-      _tnl_allow_pixel_fog( ctx, (i915->vertex_fog != I915_FOG_VERTEX) ); 
-   }
+   /* always enbale pixel fog
+    * vertex fog use precaculted fog coord will conflict with appended
+    * fog program
+    */
+    _tnl_allow_vertex_fog( ctx, 0 );
+    _tnl_allow_pixel_fog( ctx, 1 );
 }
 
 static void i915Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *param)
diff --git a/src/mesa/drivers/dri/i915/intel_context.c b/src/mesa/drivers/dri/i915/intel_context.c
index 0a2e33ffd53..7bb00d98339 100644
--- a/src/mesa/drivers/dri/i915/intel_context.c
+++ b/src/mesa/drivers/dri/i915/intel_context.c
@@ -37,7 +37,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 #include "tnl/t_vertex.h"
@@ -228,7 +228,7 @@ static void intelInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_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;
@@ -304,7 +304,7 @@ GLboolean intelInitContext( intelContextPtr intel,
 
    /* Initialize the software rasterizer and helper modules. */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -423,7 +423,7 @@ void intelDestroyContext(__DRIcontextPrivate *driContextPriv)
       release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
       _swsetup_DestroyContext (&intel->ctx);
       _tnl_DestroyContext (&intel->ctx);
-      _ac_DestroyContext (&intel->ctx);
+      _vbo_DestroyContext (&intel->ctx);
 
       _swrast_DestroyContext (&intel->ctx);
       intel->Fallback = 0;	/* don't call _swrast_Flush later */
diff --git a/src/mesa/drivers/dri/i915tex/i915_context.c b/src/mesa/drivers/dri/i915tex/i915_context.c
index 4cbe29d79d6..9b4d72eab3e 100644
--- a/src/mesa/drivers/dri/i915tex/i915_context.c
+++ b/src/mesa/drivers/dri/i915tex/i915_context.c
@@ -36,7 +36,6 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
 
 #include "utils.h"
 #include "i915_reg.h"
@@ -67,7 +66,7 @@ i915InvalidateState(GLcontext * ctx, GLuint new_state)
 {
    _swrast_InvalidateState(ctx, new_state);
    _swsetup_InvalidateState(ctx, new_state);
-   _ac_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;
diff --git a/src/mesa/drivers/dri/i915tex/i915_context.h b/src/mesa/drivers/dri/i915tex/i915_context.h
index 5ae76fcd185..d2713e88f96 100644
--- a/src/mesa/drivers/dri/i915tex/i915_context.h
+++ b/src/mesa/drivers/dri/i915tex/i915_context.h
@@ -243,7 +243,6 @@ struct i915_context
    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;
diff --git a/src/mesa/drivers/dri/i915tex/i915_state.c b/src/mesa/drivers/dri/i915tex/i915_state.c
index 2f5a30787ee..78ae4bdb5fc 100644
--- a/src/mesa/drivers/dri/i915tex/i915_state.c
+++ b/src/mesa/drivers/dri/i915tex/i915_state.c
@@ -520,7 +520,6 @@ 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;
 }
 
 static void
@@ -646,10 +645,12 @@ i915_update_fog(GLcontext * ctx)
          i915->state.Ctx[I915_CTXREG_LIS5] &= ~S5_FOG_ENABLE;
    }
 
-   if (enabled) {
-      _tnl_allow_vertex_fog(ctx, (i915->vertex_fog == I915_FOG_VERTEX));
-      _tnl_allow_pixel_fog(ctx, (i915->vertex_fog != I915_FOG_VERTEX));
-   }
+    /* always enbale pixel fog
+    * vertex fog use precaculted fog coord will conflict with appended
+    * fog program
+    */
+    _tnl_allow_vertex_fog( ctx, 0 );
+    _tnl_allow_pixel_fog( ctx, 1 );
 }
 
 static void
diff --git a/src/mesa/drivers/dri/i915tex/i915_tex.c b/src/mesa/drivers/dri/i915tex/i915_tex.c
index a53abe9a92f..59e148ca04e 100644
--- a/src/mesa/drivers/dri/i915tex/i915_tex.c
+++ b/src/mesa/drivers/dri/i915tex/i915_tex.c
@@ -52,27 +52,6 @@ i915TexEnv(GLcontext * ctx, GLenum target,
    struct i915_context *i915 = I915_CONTEXT(ctx);
 
    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:{
          GLuint unit = ctx->Texture.CurrentUnit;
          GLint b = (int) ((*param) * 16.0);
@@ -92,22 +71,8 @@ i915TexEnv(GLcontext * ctx, GLenum target,
 }
 
 
-static void
-i915BindTexture(GLcontext * ctx, GLenum target,
-                struct gl_texture_object *texobj)
-{
-   /* 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)
 {
-   functions->BindTexture = i915BindTexture;
    functions->TexEnv = i915TexEnv;
 }
diff --git a/src/mesa/drivers/dri/i915tex/i915_tex_layout.c b/src/mesa/drivers/dri/i915tex/i915_tex_layout.c
index 333fefef859..c844f5351d2 100644
--- a/src/mesa/drivers/dri/i915tex/i915_tex_layout.c
+++ b/src/mesa/drivers/dri/i915tex/i915_tex_layout.c
@@ -62,15 +62,23 @@ i915_miptree_layout(struct intel_mipmap_tree * mt)
    case GL_TEXTURE_CUBE_MAP:{
          const GLuint dim = mt->width0;
          GLuint face;
+         GLuint lvlWidth = mt->width0, lvlHeight = mt->height0;
+
+         assert(lvlWidth == lvlHeight); /* cubemap images are square */
 
          /* double pitch for cube layouts */
          mt->pitch = ((dim * mt->cpp * 2 + 3) & ~3) / mt->cpp;
          mt->total_height = dim * 4;
 
-         for (level = mt->first_level; level <= mt->last_level; level++)
+         for (level = mt->first_level; level <= mt->last_level; level++) {
             intel_miptree_set_level_info(mt, level, 6,
                                          0, 0,
-                                         mt->pitch, mt->total_height, 1);
+                                         /*OLD: mt->pitch, mt->total_height,*/
+                                         lvlWidth, lvlHeight,
+                                         1);
+            lvlWidth /= 2;
+            lvlHeight /= 2;
+         }
 
          for (face = 0; face < 6; face++) {
             GLuint x = initial_offsets[face][0] * dim;
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
index be2750d041f..c92b83bcb3a 100644
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.c
@@ -311,7 +311,7 @@ intel_batchbuffer_emit_reloc(struct intel_batchbuffer *batch,
                              struct _DriBufferObject *buffer,
                              GLuint flags, GLuint mask, GLuint delta)
 {
-   assert(batch->nr_relocs <= MAX_RELOCS);
+   assert(batch->nr_relocs < MAX_RELOCS);
 
    driBOAddListItem(&batch->list, buffer, flags, mask);
 
diff --git a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
index a83dbf423df..59261f72741 100644
--- a/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/i915tex/intel_batchbuffer.h
@@ -9,7 +9,7 @@ struct intel_context;
 #define BATCH_SZ 16384
 #define BATCH_RESERVED 16
 
-#define MAX_RELOCS 100
+#define MAX_RELOCS 400
 
 #define INTEL_BATCH_NO_CLIPRECTS 0x1
 #define INTEL_BATCH_CLIPRECTS    0x2
diff --git a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c b/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c
index 31c41d86857..ba3c7f0c1f7 100644
--- a/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c
+++ b/src/mesa/drivers/dri/i915tex/intel_buffer_objects.c
@@ -50,7 +50,10 @@ intel_bufferobj_alloc(GLcontext * ctx, GLuint name, GLenum target)
    _mesa_initialize_buffer_object(&obj->Base, name, target);
 
    driGenBuffers(intel->intelScreen->regionPool,
-                 "bufferobj", 1, &obj->buffer, 64, 0, 0);
+                 "bufferobj", 1, &obj->buffer, 64,
+		 DRM_BO_FLAG_MEM_LOCAL |
+		 DRM_BO_FLAG_READ | DRM_BO_FLAG_WRITE,
+		 0);
 
    return &obj->Base;
 }
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.c b/src/mesa/drivers/dri/i915tex/intel_context.c
index c77d365360d..a5ce08b1701 100644
--- a/src/mesa/drivers/dri/i915tex/intel_context.c
+++ b/src/mesa/drivers/dri/i915tex/intel_context.c
@@ -37,7 +37,6 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
 
 #include "tnl/t_pipeline.h"
 #include "tnl/t_vertex.h"
@@ -241,7 +240,7 @@ intelInvalidateState(GLcontext * ctx, GLuint new_state)
 {
    _swrast_InvalidateState(ctx, new_state);
    _swsetup_InvalidateState(ctx, new_state);
-   _ac_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;
@@ -357,6 +356,10 @@ intelInitContext(struct intel_context *intel,
    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);
@@ -390,7 +393,7 @@ intelInitContext(struct intel_context *intel,
 
    /* Initialize the software rasterizer and helper modules. */
    _swrast_CreateContext(ctx);
-   _ac_CreateContext(ctx);
+   _vbo_CreateContext(ctx);
    _tnl_CreateContext(ctx);
    _swsetup_CreateContext(ctx);
 
@@ -500,7 +503,7 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
       release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
       _swsetup_DestroyContext(&intel->ctx);
       _tnl_DestroyContext(&intel->ctx);
-      _ac_DestroyContext(&intel->ctx);
+      _vbo_DestroyContext(&intel->ctx);
 
       _swrast_DestroyContext(&intel->ctx);
       intel->Fallback = 0;      /* don't call _swrast_Flush later */
@@ -635,14 +638,35 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
        sarea->rotation != intelScreen->current_rotation) {
 
       intelUpdateScreenRotation(sPriv, sarea);
+   }
 
-      /* 
+   if (sarea->width != intel->width ||
+       sarea->height != intel->height ||
+       sarea->rotation != intel->current_rotation) {
+      
+      void *batchMap = intel->batch->map;
+      
+      /*
+       * FIXME: Really only need to do this when drawing to a
+       * common back- or front buffer.
+       */
+
+      /*
        * This will drop the outstanding batchbuffer on the floor
-       * FIXME: This should be done for all contexts?
        */
 
+      if (batchMap != NULL) {
+	 driBOUnmap(intel->batch->buffer);
+	 intel->batch->map = NULL;
+      }
+
       intel_batchbuffer_reset(intel->batch);
 
+      if (batchMap == NULL) {
+	 driBOUnmap(intel->batch->buffer);
+	 intel->batch->map = NULL;
+      }
+
       /* lose all primitives */
       intel->prim.primitive = ~0;
       intel->prim.start_ptr = 0;
@@ -653,6 +677,10 @@ intelContendedLock(struct intel_context *intel, GLuint flags)
 
       /* force window update */
       intel->lastStamp = 0;
+
+      intel->width = sarea->width;
+      intel->height = sarea->height;
+      intel->current_rotation = sarea->rotation;
    }
 
 
diff --git a/src/mesa/drivers/dri/i915tex/intel_context.h b/src/mesa/drivers/dri/i915tex/intel_context.h
index 7654e4ecd51..96b911501f3 100644
--- a/src/mesa/drivers/dri/i915tex/intel_context.h
+++ b/src/mesa/drivers/dri/i915tex/intel_context.h
@@ -286,6 +286,15 @@ struct intel_context
    GLuint swap_missed_count;
 
    GLuint swap_scheduled;
+
+  /* Rotation. Need to match that of the
+   * current screen.
+   */
+
+  int width;
+  int height;
+  int current_rotation;
+
 };
 
 /* These are functions now:
diff --git a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
index 2b1077aee0d..93372f4dcc8 100644
--- a/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
+++ b/src/mesa/drivers/dri/i915tex/intel_mipmap_tree.c
@@ -309,8 +309,13 @@ intel_miptree_image_data(struct intel_context *intel,
       height = dst->level[level].height;
       if(dst->compressed)
 	 height /= 4;
-      intel_region_data(intel->intelScreen, dst->region, dst_offset + dst_depth_offset[i], 0, 0, src, src_row_pitch, 0, 0,   /* source x,y */
-                        dst->level[level].width, height);
+      intel_region_data(intel->intelScreen, dst->region,
+                        dst_offset + dst_depth_offset[i], /* dst_offset */
+                        0, 0,                             /* dstx, dsty */
+                        src,
+                        src_row_pitch,
+                        0, 0,                             /* source x, y */
+                        dst->level[level].width, height); /* width, height */
 
       src += src_image_pitch;
    }
diff --git a/src/mesa/drivers/dri/i915tex/intel_regions.c b/src/mesa/drivers/dri/i915tex/intel_regions.c
index 1205b180ca6..a114bdf8965 100644
--- a/src/mesa/drivers/dri/i915tex/intel_regions.c
+++ b/src/mesa/drivers/dri/i915tex/intel_regions.c
@@ -217,7 +217,8 @@ _mesa_copy_rect(GLubyte * dst,
                 GLuint dst_y,
                 GLuint width,
                 GLuint height,
-                GLubyte * src, GLuint src_pitch, GLuint src_x, GLuint src_y)
+                const GLubyte * src,
+                GLuint src_pitch, GLuint src_x, GLuint src_y)
 {
    GLuint i;
 
@@ -253,7 +254,7 @@ intel_region_data(intelScreenPrivate *intelScreen,
                   struct intel_region *dst,
                   GLuint dst_offset,
                   GLuint dstx, GLuint dsty,
-                  void *src, GLuint src_pitch,
+                  const void *src, GLuint src_pitch,
                   GLuint srcx, GLuint srcy, GLuint width, GLuint height)
 {
    struct intel_context *intel = intelScreenContext(intelScreen);
diff --git a/src/mesa/drivers/dri/i915tex/intel_regions.h b/src/mesa/drivers/dri/i915tex/intel_regions.h
index 1415bace690..d938c107a46 100644
--- a/src/mesa/drivers/dri/i915tex/intel_regions.h
+++ b/src/mesa/drivers/dri/i915tex/intel_regions.h
@@ -103,7 +103,7 @@ void intel_region_data(intelScreenPrivate *intelScreen,
                        struct intel_region *dest,
                        GLuint dest_offset,
                        GLuint destx, GLuint desty,
-                       void *src, GLuint src_stride,
+                       const void *src, GLuint src_stride,
                        GLuint srcx, GLuint srcy, GLuint width, GLuint height);
 
 /* Copy rectangular sub-regions
diff --git a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c b/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
index e73c9c2f218..79d587a1744 100644
--- a/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
+++ b/src/mesa/drivers/dri/i915tex/intel_tex_validate.c
@@ -110,6 +110,8 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
    GLuint nr_faces = 0;
    struct intel_texture_image *firstImage;
 
+   GLboolean need_flush = GL_FALSE;
+
    /* We know/require this is true by now: 
     */
    assert(intelObj->base.Complete);
@@ -201,11 +203,13 @@ intel_finalize_mipmap_tree(struct intel_context *intel, GLuint unit)
           */
          if (intelObj->mt != intelImage->mt) {
             copy_image_data_to_tree(intel, intelObj, intelImage);
+	    need_flush = GL_TRUE;
          }
       }
    }
 
-   intel_batchbuffer_flush(intel->batch);
+   if (need_flush)
+      intel_batchbuffer_flush(intel->batch);
 
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/i965/Makefile b/src/mesa/drivers/dri/i965/Makefile
index f0a6fa57406..9e4ff112dc3 100644
--- a/src/mesa/drivers/dri/i965/Makefile
+++ b/src/mesa/drivers/dri/i965/Makefile
@@ -35,17 +35,11 @@ DRIVER_SOURCES = \
 	brw_context.c \
 	brw_curbe.c \
 	brw_draw.c \
-	brw_draw_current.c \
 	brw_draw_upload.c \
 	brw_eu.c \
 	brw_eu_debug.c \
 	brw_eu_emit.c \
 	brw_eu_util.c \
-	brw_exec.c \
-	brw_exec_api.c \
-	brw_exec_array.c \
-	brw_exec_draw.c \
-	brw_exec_eval.c \
 	brw_fallback.c \
 	brw_gs.c \
 	brw_gs_emit.c \
@@ -54,9 +48,6 @@ DRIVER_SOURCES = \
 	brw_metaops.c \
 	brw_misc_state.c \
 	brw_program.c \
-	brw_save.c \
-	brw_save_api.c \
-	brw_save_draw.c \
 	brw_sf.c \
 	brw_sf_emit.c \
 	brw_sf_state.c \
diff --git a/src/mesa/drivers/dri/i965/brw_attrib.h b/src/mesa/drivers/dri/i965/brw_attrib.h
deleted file mode 100644
index 12659bd1cf0..00000000000
--- a/src/mesa/drivers/dri/i965/brw_attrib.h
+++ /dev/null
@@ -1,114 +0,0 @@
-/*
- Copyright (C) Intel Corp.  2006.  All Rights Reserved.
- Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
- develop this 3D driver.
- 
- Permission is hereby granted, free of charge, to any person obtaining
- a copy of this software and associated documentation files (the
- "Software"), to deal in the Software without restriction, including
- without limitation the rights to use, copy, modify, merge, publish,
- distribute, sublicense, and/or sell copies of the Software, and to
- permit persons to whom the Software is furnished to do so, subject to
- the following conditions:
- 
- The above copyright notice and this permission notice (including the
- next paragraph) shall be included in all copies or substantial
- portions of the Software.
- 
- THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- 
- **********************************************************************/
- /*
-  * Authors:
-  *   Keith Whitwell <[email protected]>
-  */
-
-#ifndef BRW_ATTRIB_H
-#define BRW_ATTRIB_H
-
-
-/*
- * Note: The first attributes match the VERT_ATTRIB_* definitions
- * in mtypes.h.  However, the tnl module has additional attributes
- * for materials, color indexes, edge flags, etc.
- */
-/* Although it's nice to use these as bit indexes in a DWORD flag, we
- * could manage without if necessary.  Another limit currently is the
- * number of bits allocated for these numbers in places like vertex
- * program instruction formats and register layouts.
- */
-enum {
-	BRW_ATTRIB_POS = 0,
-	BRW_ATTRIB_WEIGHT = 1,
-	BRW_ATTRIB_NORMAL = 2,
-	BRW_ATTRIB_COLOR0 = 3,
-	BRW_ATTRIB_COLOR1 = 4,
-	BRW_ATTRIB_FOG = 5,
-	BRW_ATTRIB_INDEX = 6,        
-	BRW_ATTRIB_EDGEFLAG = 7,     
-	BRW_ATTRIB_TEX0 = 8,
-	BRW_ATTRIB_TEX1 = 9,
-	BRW_ATTRIB_TEX2 = 10,
-	BRW_ATTRIB_TEX3 = 11,
-	BRW_ATTRIB_TEX4 = 12,
-	BRW_ATTRIB_TEX5 = 13,
-	BRW_ATTRIB_TEX6 = 14,
-	BRW_ATTRIB_TEX7 = 15,
-
-	BRW_ATTRIB_GENERIC0 = 16, /* Not used? */
-	BRW_ATTRIB_GENERIC1 = 17,
-	BRW_ATTRIB_GENERIC2 = 18,
-	BRW_ATTRIB_GENERIC3 = 19,
-	BRW_ATTRIB_GENERIC4 = 20,
-	BRW_ATTRIB_GENERIC5 = 21,
-	BRW_ATTRIB_GENERIC6 = 22,
-	BRW_ATTRIB_GENERIC7 = 23,
-	BRW_ATTRIB_GENERIC8 = 24,
-	BRW_ATTRIB_GENERIC9 = 25,
-	BRW_ATTRIB_GENERIC10 = 26,
-	BRW_ATTRIB_GENERIC11 = 27,
-	BRW_ATTRIB_GENERIC12 = 28,
-	BRW_ATTRIB_GENERIC13 = 29,
-	BRW_ATTRIB_GENERIC14 = 30,
-	BRW_ATTRIB_GENERIC15 = 31,
-
-	BRW_ATTRIB_MAT_FRONT_AMBIENT = 32, 
-	BRW_ATTRIB_MAT_BACK_AMBIENT = 33,
-	BRW_ATTRIB_MAT_FRONT_DIFFUSE = 34,
-	BRW_ATTRIB_MAT_BACK_DIFFUSE = 35,
-	BRW_ATTRIB_MAT_FRONT_SPECULAR = 36,
-	BRW_ATTRIB_MAT_BACK_SPECULAR = 37,
-	BRW_ATTRIB_MAT_FRONT_EMISSION = 38,
-	BRW_ATTRIB_MAT_BACK_EMISSION = 39,
-	BRW_ATTRIB_MAT_FRONT_SHININESS = 40,
-	BRW_ATTRIB_MAT_BACK_SHININESS = 41,
-	BRW_ATTRIB_MAT_FRONT_INDEXES = 42,
-	BRW_ATTRIB_MAT_BACK_INDEXES = 43, 
-
-	BRW_ATTRIB_MAX = 44
-} ;
-
-#define BRW_ATTRIB_FIRST_MATERIAL BRW_ATTRIB_MAT_FRONT_AMBIENT
-#define BRW_ATTRIB_LAST_MATERIAL BRW_ATTRIB_MAT_BACK_INDEXES
-
-#define BRW_MAX_COPIED_VERTS 3
-
-
-static inline GLuint64EXT brw_translate_inputs( GLboolean vp_enabled,
-						GLuint mesa_inputs )
-{
-   GLuint64EXT inputs = mesa_inputs;
-   if (vp_enabled)
-      return inputs;
-   else 
-      return (inputs & 0xffff) | ((inputs & 0xffff0000) << 16);
-}
-
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_attrib_tmp.h b/src/mesa/drivers/dri/i965/brw_attrib_tmp.h
deleted file mode 100644
index 3089bd6cac2..00000000000
--- a/src/mesa/drivers/dri/i965/brw_attrib_tmp.h
+++ /dev/null
@@ -1,485 +0,0 @@
-/**************************************************************************
-
-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
-TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-#define ATTR1FV( A, V ) ATTR( A, 1, (V)[0], 0, 0, 1 )
-#define ATTR2FV( A, V ) ATTR( A, 2, (V)[0], (V)[1], 0, 1 )
-#define ATTR3FV( A, V ) ATTR( A, 3, (V)[0], (V)[1], (V)[2], 1 )
-#define ATTR4FV( A, V ) ATTR( A, 4, (V)[0], (V)[1], (V)[2], (V)[3] )
-
-#define ATTR1F( A, X )          ATTR( A, 1, X, 0, 0, 1 )
-#define ATTR2F( A, X, Y )       ATTR( A, 2, X, Y, 0, 1 )
-#define ATTR3F( A, X, Y, Z )    ATTR( A, 3, X, Y, Z, 1 )
-#define ATTR4F( A, X, Y, Z, W ) ATTR( A, 4, X, Y, Z, W )
-
-#define MAT_ATTR( A, N, V ) ATTR( A, N, (V)[0], (V)[1], (V)[2], (V)[3] )
-
-static void GLAPIENTRY TAG(Vertex2f)( GLfloat x, GLfloat y )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR2F( BRW_ATTRIB_POS, x, y );
-}
-
-static void GLAPIENTRY TAG(Vertex2fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR2FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY TAG(Vertex3f)( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3F( BRW_ATTRIB_POS, x, y, z );
-}
-
-static void GLAPIENTRY TAG(Vertex3fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY TAG(Vertex4f)( GLfloat x, GLfloat y, GLfloat z, GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4F( BRW_ATTRIB_POS, x, y, z, w );
-}
-
-static void GLAPIENTRY TAG(Vertex4fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4FV( BRW_ATTRIB_POS, v );
-}
-
-static void GLAPIENTRY TAG(TexCoord1f)( GLfloat x )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1F( BRW_ATTRIB_TEX0, x );
-}
-
-static void GLAPIENTRY TAG(TexCoord1fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY TAG(TexCoord2f)( GLfloat x, GLfloat y )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR2F( BRW_ATTRIB_TEX0, x, y );
-}
-
-static void GLAPIENTRY TAG(TexCoord2fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR2FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY TAG(TexCoord3f)( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3F( BRW_ATTRIB_TEX0, x, y, z );
-}
-
-static void GLAPIENTRY TAG(TexCoord3fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY TAG(TexCoord4f)( GLfloat x, GLfloat y, GLfloat z, GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4F( BRW_ATTRIB_TEX0, x, y, z, w );
-}
-
-static void GLAPIENTRY TAG(TexCoord4fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4FV( BRW_ATTRIB_TEX0, v );
-}
-
-static void GLAPIENTRY TAG(Normal3f)( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3F( BRW_ATTRIB_NORMAL, x, y, z );
-}
-
-static void GLAPIENTRY TAG(Normal3fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3FV( BRW_ATTRIB_NORMAL, v );
-}
-
-static void GLAPIENTRY TAG(FogCoordfEXT)( GLfloat x )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1F( BRW_ATTRIB_FOG, x );
-}
-
-static void GLAPIENTRY TAG(FogCoordfvEXT)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1FV( BRW_ATTRIB_FOG, v );
-}
-
-static void GLAPIENTRY TAG(Color3f)( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3F( BRW_ATTRIB_COLOR0, x, y, z );
-}
-
-static void GLAPIENTRY TAG(Color3fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3FV( BRW_ATTRIB_COLOR0, v );
-}
-
-static void GLAPIENTRY TAG(Color4f)( GLfloat x, GLfloat y, GLfloat z, GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4F( BRW_ATTRIB_COLOR0, x, y, z, w );
-}
-
-static void GLAPIENTRY TAG(Color4fv)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR4FV( BRW_ATTRIB_COLOR0, v );
-}
-
-static void GLAPIENTRY TAG(SecondaryColor3fEXT)( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3F( BRW_ATTRIB_COLOR1, x, y, z );
-}
-
-static void GLAPIENTRY TAG(SecondaryColor3fvEXT)( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR3FV( BRW_ATTRIB_COLOR1, v );
-}
-
-
-static void GLAPIENTRY TAG(EdgeFlag)( GLboolean b )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1F( BRW_ATTRIB_EDGEFLAG, (GLfloat)b );
-}
-
-static void GLAPIENTRY TAG(Indexf)( GLfloat f )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1F( BRW_ATTRIB_INDEX, f );
-}
-
-static void GLAPIENTRY TAG(Indexfv)( const GLfloat *f )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   ATTR1FV( BRW_ATTRIB_INDEX, f );
-}
-
-
-static void GLAPIENTRY TAG(MultiTexCoord1f)( GLenum target, GLfloat x  )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR1F( attr, x );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord1fv)( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR1FV( attr, v );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord2f)( GLenum target, GLfloat x, GLfloat y )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR2F( attr, x, y );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord2fv)( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR2FV( attr, v );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord3f)( GLenum target, GLfloat x, GLfloat y,
-				    GLfloat z)
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR3F( attr, x, y, z );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord3fv)( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR3FV( attr, v );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord4f)( GLenum target, GLfloat x, GLfloat y,
-				    GLfloat z, GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR4F( attr, x, y, z, w );
-}
-
-static void GLAPIENTRY TAG(MultiTexCoord4fv)( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLuint attr = (target & 0x7) + BRW_ATTRIB_TEX0;
-   ATTR4FV( attr, v );
-}
-
-
-static void GLAPIENTRY TAG(VertexAttrib1fARB)( GLuint index, GLfloat x )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR1F(0, x);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR1F(BRW_ATTRIB_GENERIC0 + index, x);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib1fvARB)( GLuint index, 
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR1FV(0, v);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR1FV(BRW_ATTRIB_GENERIC0 + index, v);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib2fARB)( GLuint index, GLfloat x, 
-					      GLfloat y )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR2F(0, x, y);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR2F(BRW_ATTRIB_GENERIC0 + index, x, y);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib2fvARB)( GLuint index,
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR2FV(0, v);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR2FV(BRW_ATTRIB_GENERIC0 + index, v);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib3fARB)( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR3F(0, x, y, z);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR3F(BRW_ATTRIB_GENERIC0 + index, x, y, z);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib3fvARB)( GLuint index,
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR3FV(0, v);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR3FV(BRW_ATTRIB_GENERIC0 + index, v);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib4fARB)( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z,
-					      GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR4F(0, x, y, z, w);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR4F(BRW_ATTRIB_GENERIC0 + index, x, y, z, w);
-   else
-      ERROR();
-}
-
-static void GLAPIENTRY TAG(VertexAttrib4fvARB)( GLuint index, 
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index == 0) 
-      ATTR4FV(0, v);
-   else if (index < MAX_VERTEX_ATTRIBS)
-      ATTR4FV(BRW_ATTRIB_GENERIC0 + index, v);
-   else
-      ERROR();
-}
-
-
-/* Although we don't export NV_vertex_program, these entrypoints are
- * used by the display list and other code specifically because of
- * their property of aliasing with other attributes.
- */
-static void GLAPIENTRY TAG(VertexAttrib1fNV)( GLuint index, GLfloat x )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX)
-      ATTR1F(index, x);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib1fvNV)( GLuint index, 
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR1FV(index, v);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib2fNV)( GLuint index, GLfloat x, 
-					      GLfloat y )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR2F(index, x, y);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib2fvNV)( GLuint index,
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR2FV(index, v);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib3fNV)( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR3F(index, x, y, z);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib3fvNV)( GLuint index,
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR3FV(index, v);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib4fNV)( GLuint index, GLfloat x,
-					      GLfloat y, GLfloat z,
-					      GLfloat w )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR4F(index, x, y, z, w);
-}
-
-static void GLAPIENTRY TAG(VertexAttrib4fvNV)( GLuint index, 
-					       const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   if (index < BRW_ATTRIB_MAX) 
-      ATTR4FV(index, v);
-}
-
-
-#define MAT( ATTR, N, face, params )			\
-do {							\
-   if (face != GL_BACK)					\
-      MAT_ATTR( ATTR, N, params ); /* front */		\
-   if (face != GL_FRONT)				\
-      MAT_ATTR( ATTR + 1, N, params ); /* back */	\
-} while (0)
-
-
-/* Colormaterial conflicts are dealt with later.
- */
-static void GLAPIENTRY TAG(Materialfv)( GLenum face, GLenum pname, 
-			       const GLfloat *params )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   switch (pname) {
-   case GL_EMISSION:
-      MAT( BRW_ATTRIB_MAT_FRONT_EMISSION, 4, face, params );
-      break;
-   case GL_AMBIENT:
-      MAT( BRW_ATTRIB_MAT_FRONT_AMBIENT, 4, face, params );
-      break;
-   case GL_DIFFUSE:
-      MAT( BRW_ATTRIB_MAT_FRONT_DIFFUSE, 4, face, params );
-      break;
-   case GL_SPECULAR:
-      MAT( BRW_ATTRIB_MAT_FRONT_SPECULAR, 4, face, params );
-      break;
-   case GL_SHININESS:
-      MAT( BRW_ATTRIB_MAT_FRONT_SHININESS, 1, face, params );
-      break;
-   case GL_COLOR_INDEXES:
-      MAT( BRW_ATTRIB_MAT_FRONT_INDEXES, 3, face, params );
-      break;
-   case GL_AMBIENT_AND_DIFFUSE:
-      MAT( BRW_ATTRIB_MAT_FRONT_AMBIENT, 4, face, params );
-      MAT( BRW_ATTRIB_MAT_FRONT_DIFFUSE, 4, face, params );
-      break;
-   default:
-      ERROR();
-      return;
-   }
-}
-
-
-#undef ATTR1FV
-#undef ATTR2FV
-#undef ATTR3FV
-#undef ATTR4FV
-
-#undef ATTR1F
-#undef ATTR2F
-#undef ATTR3F
-#undef ATTR4F
-
-#undef MAT
-#undef MAT_ATTR
diff --git a/src/mesa/drivers/dri/i965/brw_clip.h b/src/mesa/drivers/dri/i965/brw_clip.h
index c3967c8c614..49b2770a514 100644
--- a/src/mesa/drivers/dri/i965/brw_clip.h
+++ b/src/mesa/drivers/dri/i965/brw_clip.h
@@ -116,7 +116,7 @@ struct brw_clip_compile {
    GLuint last_mrf;
 
    GLuint header_position_offset;
-   GLuint offset[BRW_ATTRIB_MAX];
+   GLuint offset[VERT_ATTRIB_MAX];
 };
 
 #define ATTR_SIZE  (4*4)
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index bc422c1a50b..6faee65542b 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -34,8 +34,6 @@
 #include "brw_aub.h"
 #include "brw_defines.h"
 #include "brw_draw.h"
-#include "brw_exec.h"
-#include "brw_save.h"
 #include "brw_vs.h"
 #include "imports.h"
 #include "intel_tex.h"
@@ -158,12 +156,6 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    brw_FrameBufferTexInit( brw );
 
-   /* Hook our functions into exec and compile dispatch tables.  Only
-    * fallback on out-of-memory situations.
-    */
-   brw_exec_init( ctx );
-   brw_save_init( ctx );
-
    {
       const char *filename = getenv("INTEL_REPLAY");
       if (filename) {
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 9ee81b8725d..08fdc545205 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -36,7 +36,6 @@
 #include "intel_context.h"
 #include "brw_structs.h"
 #include "imports.h"
-#include "brw_attrib.h"
 
 
 /* Glossary:
@@ -215,7 +214,7 @@ struct brw_vs_prog_data {
    GLuint total_grf;
    GLuint outputs_written;
 
-   GLuint64EXT inputs_read;
+   GLuint inputs_read;
 
    /* Used for calculating urb partitions:
     */
@@ -382,10 +381,10 @@ struct brw_cached_batch_item {
    
 
 
-/* Protect against a future where BRW_ATTRIB_MAX > 32.  Wouldn't life
+/* Protect against a future where VERT_ATTRIB_MAX > 32.  Wouldn't life
  * be easier if C allowed arrays of packed elements?
  */
-#define ATTRIB_BIT_DWORDS  ((BRW_ATTRIB_MAX+31)/32)
+#define ATTRIB_BIT_DWORDS  ((VERT_ATTRIB_MAX+31)/32)
 
 struct brw_vertex_element {
    const struct gl_client_array *glarray;
@@ -401,8 +400,8 @@ struct brw_vertex_element {
 
 
 struct brw_vertex_info {
-   GLuint64EXT varying;  /* varying:1[BRW_ATTRIB_MAX] */
-   GLuint sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[BRW_ATTRIB_MAX] */
+   GLuint varying;  /* varying:1[VERT_ATTRIB_MAX] */
+   GLuint sizes[ATTRIB_BIT_DWORDS * 2]; /* sizes:2[VERT_ATTRIB_MAX] */
 };
 
 
@@ -449,14 +448,13 @@ struct brw_context
    struct brw_cached_batch_item *cached_batch_items;
 
    struct {
-      /* Fallback values for inputs not supplied: 
-       */
-      struct gl_client_array current_values[BRW_ATTRIB_MAX];
 
       /* Arrays with buffer objects to copy non-bufferobj arrays into
        * for upload:
        */
-      struct gl_client_array vbo_array[BRW_ATTRIB_MAX];
+      struct gl_client_array vbo_array[VERT_ATTRIB_MAX];
+
+      struct brw_vertex_element inputs[VERT_ATTRIB_MAX];
 
 #define BRW_NR_UPLOAD_BUFS 17
 #define BRW_UPLOAD_INIT_SIZE (128*1024)
@@ -469,11 +467,6 @@ struct brw_context
 	 GLuint wrap;
       } upload;
 
-      /* Currenly bound arrays, including fallbacks to current_values
-       * above:
-       */
-      struct brw_vertex_element inputs[BRW_ATTRIB_MAX];
-
       /* Summary of size and varying of active arrays, so we can check
        * for changes to this state:
        */
diff --git a/src/mesa/drivers/dri/i965/brw_draw.c b/src/mesa/drivers/dri/i965/brw_draw.c
index 471fda9f7ef..092c6bafc2d 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.c
+++ b/src/mesa/drivers/dri/i965/brw_draw.c
@@ -35,7 +35,6 @@
 
 #include "brw_draw.h"
 #include "brw_defines.h"
-#include "brw_attrib.h"
 #include "brw_context.h"
 #include "brw_aub.h"
 #include "brw_state.h"
@@ -45,7 +44,8 @@
 #include "intel_batchbuffer.h"
 #include "intel_buffer_objects.h"
 
-
+#include "tnl/tnl.h"
+#include "vbo/vbo_context.h"
 
 
 
@@ -143,7 +143,7 @@ static void brw_emit_cliprect( struct brw_context *brw,
 
 
 static void brw_emit_prim( struct brw_context *brw, 
-			   const struct brw_draw_prim *prim )
+			   const struct _mesa_prim *prim )
 
 {
    struct brw_3d_primitive prim_packet;
@@ -170,34 +170,9 @@ static void brw_emit_prim( struct brw_context *brw,
    }
 }
 
-
-
-static void update_current_size( struct gl_client_array *array)
-{
-   const GLfloat *ptr = (const GLfloat *)array->Ptr;
-
-   assert(array->StrideB == 0);
-   assert(array->Type == GL_FLOAT || array->Type == GL_UNSIGNED_BYTE);
-
-   if (ptr[3] != 1.0) 
-      array->Size = 4;
-   else if (ptr[2] != 0.0) 
-      array->Size = 3;
-   else if (ptr[1] != 0.0) 
-      array->Size = 2;
-   else
-      array->Size = 1;
-}
-
-
-
-/* Fill in any gaps in passed arrays with pointers to current
- * attributes:
- */
 static void brw_merge_inputs( struct brw_context *brw,
 		       const struct gl_client_array *arrays[])
 {
-   struct gl_client_array *current_values = brw->vb.current_values;
    struct brw_vertex_element *inputs = brw->vb.inputs;
    struct brw_vertex_info old = brw->vb.info;
    GLuint i;
@@ -205,19 +180,16 @@ static void brw_merge_inputs( struct brw_context *brw,
    memset(inputs, 0, sizeof(*inputs));
    memset(&brw->vb.info, 0, sizeof(brw->vb.info));
 
-   for (i = 0; i < BRW_ATTRIB_MAX; i++) {
-      if (arrays[i] && arrays[i]->Enabled)
-      {
-	 brw->vb.inputs[i].glarray = arrays[i];
-	 brw->vb.info.varying |= (GLuint64EXT) 1 << i;
-      }
-      else 
-      {
-	 brw->vb.inputs[i].glarray = &current_values[i];
-	 update_current_size(&current_values[i]);
-      }
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
+      brw->vb.inputs[i].glarray = arrays[i];
+
+      /* XXX: metaops passes null arrays */
+      if (arrays[i]) {
+	 if (arrays[i]->StrideB != 0)
+	    brw->vb.info.varying |= 1 << i;
 
-      brw->vb.info.sizes[i/16] |= (inputs[i].glarray->Size - 1) << ((i%16) * 2);
+	 brw->vb.info.sizes[i/16] |= (inputs[i].glarray->Size - 1) << ((i%16) * 2);
+      }
    }
 
    /* Raise statechanges if input sizes and varying have changed: 
@@ -229,8 +201,11 @@ static void brw_merge_inputs( struct brw_context *brw,
       brw->state.dirty.brw |= BRW_NEW_INPUT_VARYING;
 }
 
+/* XXX: could split the primitive list to fallback only on the
+ * non-conformant primitives.
+ */
 static GLboolean check_fallbacks( struct brw_context *brw,
-				  const struct brw_draw_prim *prim,
+				  const struct _mesa_prim *prim,
 				  GLuint nr_prims )
 {
    GLuint i;
@@ -281,15 +256,16 @@ static GLboolean check_fallbacks( struct brw_context *brw,
    return GL_FALSE;
 }
 
-
+/* May fail if out of video memory for texture or vbo upload, or on
+ * fallback conditions.
+ */
 static GLboolean brw_try_draw_prims( GLcontext *ctx,
 				     const struct gl_client_array *arrays[],
-				     const struct brw_draw_prim *prim,
+				     const struct _mesa_prim *prim,
 				     GLuint nr_prims,
-				     const struct brw_draw_index_buffer *ib,
+				     const struct _mesa_index_buffer *ib,
 				     GLuint min_index,
-				     GLuint max_index,
-				     GLuint flags )
+				     GLuint max_index )
 {
    struct intel_context *intel = intel_context(ctx);
    struct brw_context *brw = brw_context(ctx);
@@ -298,11 +274,11 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
 
    if (ctx->NewState)
       _mesa_update_state( ctx );
-      
+
    /* Bind all inputs, derive varying and size information:
     */
    brw_merge_inputs( brw, arrays );
-
+      
    /* Have to validate state quite late.  Will rebuild tnl_program,
     * which depends on varying information.  
     * 
@@ -319,10 +295,6 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    }
 
    {
-      assert(intel->locked);
-      
-
-
       /* Set the first primitive early, ahead of validate_state:
        */
       brw_set_prim(brw, prim[0].mode);
@@ -411,44 +383,88 @@ static GLboolean brw_try_draw_prims( GLcontext *ctx,
    return retval;
 }
 
+static GLboolean brw_need_rebase( GLcontext *ctx,
+				  const struct gl_client_array *arrays[],
+				  const struct _mesa_index_buffer *ib,
+				  GLuint min_index )
+{
+   if (min_index == 0) 
+      return GL_FALSE;
 
-GLboolean brw_draw_prims( GLcontext *ctx,
-			  const struct gl_client_array *arrays[],
-			  const struct brw_draw_prim *prim,
-			  GLuint nr_prims,
-			  const struct brw_draw_index_buffer *ib,
-			  GLuint min_index,
-			  GLuint max_index,
-			  GLuint flags )
+   if (ib) {
+      if (!vbo_all_varyings_in_vbos(arrays))
+	 return GL_TRUE;
+      else
+	 return GL_FALSE;
+   }
+   else {
+      /* Hmm.  This isn't quite what I wanted.  BRW can actually
+       * handle the mixed case well enough that we shouldn't need to
+       * rebase.  However, it's probably not very common, nor hugely
+       * expensive to do it this way:
+       */
+      if (!vbo_all_varyings_in_vbos(arrays))
+	 return GL_TRUE;
+      else
+	 return GL_FALSE;
+   }
+}
+				  
+
+void brw_draw_prims( GLcontext *ctx,
+		     const struct gl_client_array *arrays[],
+		     const struct _mesa_prim *prim,
+		     GLuint nr_prims,
+		     const struct _mesa_index_buffer *ib,
+		     GLuint min_index,
+		     GLuint max_index )
 {
    struct intel_context *intel = intel_context(ctx);
    GLboolean retval;
 
-   retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index, flags);
+   /* Decide if we want to rebase.  If so we end up recursing once
+    * only into this function.
+    */
+   if (brw_need_rebase( ctx, arrays, ib, min_index )) {
+      vbo_rebase_prims( ctx, arrays, 
+			prim, nr_prims, 
+			ib, min_index, max_index, 
+			brw_draw_prims );
+      
+      return;
+   }
 
-   
-   if (!retval && bmError(intel)) {
 
-      DBG("retrying\n");
-      /* This looks like out-of-memory but potentially we have
-       * situation where there is enough memory but it has become
-       * fragmented.  Clear out all heaps and start from scratch by
-       * faking a contended lock event:  (done elsewhere)
-       */
+   /* Make a first attempt at drawing:
+    */
+   retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
 
+   
+   /* This looks like out-of-memory but potentially we have
+    * situation where there is enough memory but it has become
+    * fragmented.  Clear out all heaps and start from scratch by
+    * faking a contended lock event:  (done elsewhere)
+    */
+   if (!retval && !intel->Fallback && bmError(intel)) {
+      DBG("retrying\n");
       /* Then try a second time only to upload textures and draw the
        * primitives:
        */
-      retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index, flags);
+      retval = brw_try_draw_prims(ctx, arrays, prim, nr_prims, ib, min_index, max_index);
+   }
+
+   /* Otherwise, we really are out of memory.  Pass the drawing
+    * command to the software tnl module and which will in turn call
+    * swrast to do the drawing.
+    */
+   if (!retval) {
+      _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;
    }
-
-
-   return retval;
 }
 
 
@@ -461,21 +477,25 @@ static void brw_invalidate_vbo_cb( struct intel_context *intel, void *ptr )
 void brw_draw_init( struct brw_context *brw )
 {
    GLcontext *ctx = &brw->intel.ctx;
+   struct vbo_context *vbo = vbo_context(ctx);
    GLuint i;
    
+   /* Register our drawing function: 
+    */
+   vbo->draw_prims = brw_draw_prims;
+
    brw->vb.upload.size = BRW_UPLOAD_INIT_SIZE;
 
    for (i = 0; i < BRW_NR_UPLOAD_BUFS; i++) {
       brw->vb.upload.vbo[i] = ctx->Driver.NewBufferObject(ctx, 1, GL_ARRAY_BUFFER_ARB);
       
-      /* XXX: Set these to no-backing-store
+      /* NOTE:  These are set to no-backing-store.
        */
       bmBufferSetInvalidateCB(&brw->intel,
 			      intel_bufferobj_buffer(intel_buffer_object(brw->vb.upload.vbo[i])),
 			      brw_invalidate_vbo_cb,
 			      &brw->intel,
 			      GL_TRUE);
-
    }
 
    ctx->Driver.BufferData( ctx, 
@@ -484,9 +504,6 @@ void brw_draw_init( struct brw_context *brw )
 			   NULL,
 			   GL_DYNAMIC_DRAW_ARB,
 			   brw->vb.upload.vbo[0] );
-						     
-   
-   brw_init_current_values(ctx, brw->vb.current_values);
 }
 
 void brw_draw_destroy( struct brw_context *brw )
diff --git a/src/mesa/drivers/dri/i965/brw_draw.h b/src/mesa/drivers/dri/i965/brw_draw.h
index 92640bf725b..0f7b7383102 100644
--- a/src/mesa/drivers/dri/i965/brw_draw.h
+++ b/src/mesa/drivers/dri/i965/brw_draw.h
@@ -29,44 +29,18 @@
 #define BRW_DRAW_H
 
 #include "mtypes.h"		/* for GLcontext... */
-#include "brw_attrib.h"
+#include "vbo/vbo.h"
 
 struct brw_context;
 
-struct brw_draw_prim {
-   GLuint mode:8;
-   GLuint indexed:1;
-   GLuint begin:1;
-   GLuint end:1;
-   GLuint weak:1;
-   GLuint pad:20;
 
-   GLuint start;
-   GLuint count;
-};
-
-struct brw_draw_index_buffer {
-   GLuint count;
-   GLenum type;
-   struct gl_buffer_object *obj;
-   const void *ptr;
-   GLuint rebase;
-};
-
-
-#define BRW_DRAW_SORTED           0x1
-#define BRW_DRAW_ALL_INTERLEAVED  0x2
-#define BRW_DRAW_NON_INTERLEAVED  0x4
-#define BRW_DRAW_LOCKED           0x8
-
-GLboolean brw_draw_prims( GLcontext *ctx,
-			  const struct gl_client_array *arrays[],
-			  const struct brw_draw_prim *prims,
-			  GLuint nr_prims,
-			  const struct brw_draw_index_buffer *ib,
-			  GLuint min_index,
-			  GLuint max_index,
-			  GLuint flags );
+void brw_draw_prims( GLcontext *ctx,
+		     const struct gl_client_array *arrays[],
+		     const struct _mesa_prim *prims,
+		     GLuint nr_prims,
+		     const struct _mesa_index_buffer *ib,
+		     GLuint min_index,
+		     GLuint max_index );
 
 void brw_draw_init( struct brw_context *brw );
 void brw_draw_destroy( struct brw_context *brw );
@@ -80,25 +54,12 @@ void brw_init_current_values(GLcontext *ctx,
 /* brw_draw_upload.c
  */
 void brw_upload_indices( struct brw_context *brw,
-			 const struct brw_draw_index_buffer *index_buffer);
+			 const struct _mesa_index_buffer *index_buffer);
 
 GLboolean brw_upload_vertices( struct brw_context *brw,
 			       GLuint min_index,
 			       GLuint max_index );
 
 
-/* Helpers for save, exec.  Should probably have their own file:
- */
-struct brw_exec_context;
-struct brw_save_context;
-
-struct brw_exec_save {
-   struct brw_exec_context *exec;
-   struct brw_save_context *save;
-};
-
-/* Doesn't really belong here:
- */
-#define IMM_CONTEXT(ctx) ((struct brw_exec_save *)((ctx)->swtnl_im))
 
 #endif
diff --git a/src/mesa/drivers/dri/i965/brw_draw_current.c b/src/mesa/drivers/dri/i965/brw_draw_current.c
deleted file mode 100644
index 98d930738e5..00000000000
--- a/src/mesa/drivers/dri/i965/brw_draw_current.c
+++ /dev/null
@@ -1,103 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include <stdlib.h>
-
-#include "glheader.h"
-#include "context.h"
-#include "state.h"
-#include "api_validate.h"
-#include "enums.h"
-
-#include "brw_context.h"
-#include "brw_draw.h"
-
-#include "bufmgr.h"
-#include "intel_buffer_objects.h"
-
-
-void brw_init_current_values(GLcontext *ctx,
-			     struct gl_client_array *arrays)
-{
-   GLuint i;
-
-   memset(arrays, 0, sizeof(*arrays) * BRW_ATTRIB_MAX);
-
-   /* Set up a constant (StrideB == 0) array for each current
-    * attribute:
-    */
-   for (i = 0; i < BRW_ATTRIB_MAX; i++) {
-      struct gl_client_array *cl = &arrays[i];
-
-      switch (i) {
-      case BRW_ATTRIB_MAT_FRONT_SHININESS:
-      case BRW_ATTRIB_MAT_BACK_SHININESS:
-      case BRW_ATTRIB_INDEX:
-      case BRW_ATTRIB_EDGEFLAG:
-	 cl->Size = 1;
-	 break;
-      case BRW_ATTRIB_MAT_FRONT_INDEXES:
-      case BRW_ATTRIB_MAT_BACK_INDEXES:
-	 cl->Size = 3;
-	 break;
-      default:
-	 /* This is fixed for the material attributes, for others will
-	  * be determined at runtime:
-	  */
-	 if (i >= BRW_ATTRIB_MAT_FRONT_AMBIENT)
-	    cl->Size = 4;
-	 else
-	    cl->Size = 1;
-	 break;
-      }
-
-      switch (i) {
-      case BRW_ATTRIB_EDGEFLAG:
-	 cl->Type = GL_UNSIGNED_BYTE;
-	 cl->Ptr = (const void *)&ctx->Current.EdgeFlag;
-	 break;
-      case BRW_ATTRIB_INDEX:
-	 cl->Type = GL_FLOAT;
-	 cl->Ptr = (const void *)&ctx->Current.Index;
-	 break;
-      default:
-	 cl->Type = GL_FLOAT;
-	 if (i < BRW_ATTRIB_FIRST_MATERIAL)
-	    cl->Ptr = (const void *)ctx->Current.Attrib[i];
-	 else 
-	    cl->Ptr = (const void *)ctx->Light.Material.Attrib[i - BRW_ATTRIB_FIRST_MATERIAL];
-	 break;
-      }
-
-      cl->Stride = 0;
-      cl->StrideB = 0;
-      cl->Enabled = 1;
-      cl->Flags = 0;
-      cl->BufferObj = ctx->Array.NullBufferObj;
-   }
-}
-
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index 57ee294f0cf..6150cac4aa3 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -35,7 +35,6 @@
 
 #include "brw_draw.h"
 #include "brw_defines.h"
-#include "brw_attrib.h"
 #include "brw_context.h"
 #include "brw_aub.h"
 #include "brw_state.h"
@@ -310,7 +309,6 @@ copy_array_to_vbo_array( struct brw_context *brw,
 			 GLuint i,
 			 const struct gl_client_array *array,
 			 GLuint element_size,
-			 GLuint min_index,
 			 GLuint count)
 {
    GLcontext *ctx = &brw->intel.ctx;
@@ -337,7 +335,6 @@ copy_array_to_vbo_array( struct brw_context *brw,
    vbo_array->Enabled = 1;
    vbo_array->Normalized = array->Normalized;
    vbo_array->_MaxElement = array->_MaxElement;	/* ? */
-   vbo_array->Flags = array->Flags; /* ? */
    vbo_array->BufferObj = vbo;
 
    {
@@ -349,7 +346,7 @@ copy_array_to_vbo_array( struct brw_context *brw,
       map += offset;
 
       copy_strided_array( map, 
-			  array->Ptr + min_index * array->StrideB,
+			  array->Ptr,
 			  element_size,
 			  array->StrideB,
 			  count);
@@ -380,7 +377,6 @@ interleaved_vbo_array( struct brw_context *brw,
    vbo_array->Enabled = 1;
    vbo_array->Normalized = array->Normalized;
    vbo_array->_MaxElement = array->_MaxElement;	
-   vbo_array->Flags = array->Flags; /* ? */
    vbo_array->BufferObj = uploaded_array->BufferObj;
 
    return vbo_array;
@@ -393,17 +389,17 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = intel_context(ctx);
-   GLuint64EXT tmp = brw->vs.prog_data->inputs_read; 
+   GLuint tmp = brw->vs.prog_data->inputs_read; 
    struct brw_vertex_element_packet vep;
    struct brw_array_state vbp;
    GLuint i;
    const void *ptr = NULL;
    GLuint interleave = 0;
 
-   struct brw_vertex_element *enabled[BRW_ATTRIB_MAX];
+   struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
    GLuint nr_enabled = 0;
 
-   struct brw_vertex_element *upload[BRW_ATTRIB_MAX];
+   struct brw_vertex_element *upload[VERT_ATTRIB_MAX];
    GLuint nr_uploads = 0;
    
 
@@ -412,17 +408,19 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
 
    /* First build an array of pointers to ve's in vb.inputs_read
     */
+   if (0)
+      _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);
    
    while (tmp) {
-      GLuint i = ffsll(tmp)-1;
+      GLuint i = _mesa_ffsll(tmp)-1;
       struct brw_vertex_element *input = &brw->vb.inputs[i];
 
-      tmp &= ~((GLuint64EXT)1<<i);
+      tmp &= ~(1<<i);
       enabled[nr_enabled++] = input;
 
       input->index = i;
       input->element_size = get_size(input->glarray->Type) * input->glarray->Size;
-      input->count = input->glarray->StrideB ? max_index - min_index : 1;
+      input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1;
 
       if (!input->glarray->BufferObj->Name) {
 	 if (i == 0) {
@@ -441,10 +439,16 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
 	 }
 
 	 upload[nr_uploads++] = input;
-	 input->vbo_rebase_offset = 0;
+	 
+	 /* We rebase drawing to start at element zero only when
+	  * varyings are not in vbos, which means we can end up
+	  * uploading non-varying arrays (stride != 0) when min_index
+	  * is zero.  This doesn't matter as the amount to upload is
+	  * the same for these arrays whether the draw call is rebased
+	  * or not - we just have to upload the one element.
+	  */
+	 assert(min_index == 0 || input->glarray->StrideB == 0);
       }
-      else 
-	 input->vbo_rebase_offset = min_index * input->glarray->StrideB;
    }
 
    /* Upload interleaved arrays if all uploads are interleaved
@@ -457,7 +461,6 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
       input0->glarray = copy_array_to_vbo_array(brw, 0,
 						input0->glarray, 
 						interleave,
-						min_index,
 						input0->count);
 
       for (i = 1; i < nr_uploads; i++) {
@@ -475,7 +478,6 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
 	 input->glarray = copy_array_to_vbo_array(brw, i, 
 						  input->glarray,
 						  input->element_size,
-						  min_index,
 						  input->count);
 
       }
@@ -523,9 +525,9 @@ GLboolean brw_upload_vertices( struct brw_context *brw,
       vbp.vb[i].vb0.bits.pad = 0;
       vbp.vb[i].vb0.bits.access_type = BRW_VERTEXBUFFER_ACCESS_VERTEXDATA;
       vbp.vb[i].vb0.bits.vb_index = i;
-      vbp.vb[i].offset = (GLuint)input->glarray->Ptr + input->vbo_rebase_offset;
+      vbp.vb[i].offset = (GLuint)input->glarray->Ptr;
       vbp.vb[i].buffer = array_buffer(input->glarray);
-      vbp.vb[i].max_index = max_index - min_index;
+      vbp.vb[i].max_index = max_index;
    }
 
 
@@ -566,94 +568,32 @@ static GLuint element_size( GLenum type )
 
 
 
-
-static void rebase_indices_to_vbo_indices( struct brw_context *brw, 
-					   const struct brw_draw_index_buffer *index_buffer,
-					   struct gl_buffer_object **vbo_return,
-					   GLuint *offset_return )
+void brw_upload_indices( struct brw_context *brw,
+			 const struct _mesa_index_buffer *index_buffer )
 {
    GLcontext *ctx = &brw->intel.ctx;
-   GLuint min_index = index_buffer->rebase;
-   const void *indices = index_buffer->ptr;
-   GLsizei count = index_buffer->count;
-   GLenum type = index_buffer->type;
-   GLuint size = element_size(type) * count;
-   struct gl_buffer_object *bufferobj;
-   GLuint offset;
-   GLuint i;
-
-   get_space(brw, size, &bufferobj, &offset);
+   struct intel_context *intel = &brw->intel;
+   GLuint ib_size = get_size(index_buffer->type) * index_buffer->count;
+   struct gl_buffer_object *bufferobj = index_buffer->obj;
+   GLuint offset = (GLuint)index_buffer->ptr;
 
-   *vbo_return = bufferobj;
-   *offset_return = offset;
+   /* Turn into a proper VBO:
+    */
+   if (!bufferobj->Name) {
+     
+      /* Get new bufferobj, offset:
+       */
+      get_space(brw, ib_size, &bufferobj, &offset);
 
-   if (min_index == 0) {
       /* Straight upload
        */
       ctx->Driver.BufferSubData( ctx,
 				 GL_ELEMENT_ARRAY_BUFFER_ARB,
 				 offset, 
-				 size,
-				 indices,
+				 ib_size,
+				 index_buffer->ptr,
 				 bufferobj);
    }
-   else {
-      void *map = ctx->Driver.MapBuffer(ctx,
-					GL_ELEMENT_ARRAY_BUFFER_ARB,
-					GL_DYNAMIC_DRAW_ARB,
-					bufferobj);
-
-      map += offset;
-
-      switch (type) {
-      case GL_UNSIGNED_INT: {
-	 GLuint *ui_map = (GLuint *)map;
-	 const GLuint *ui_indices = (const GLuint *)indices;
-
-	 for (i = 0; i < count; i++)
-	    ui_map[i] = ui_indices[i] - min_index;
-	 break;
-      }
-      case GL_UNSIGNED_SHORT:  {
-	 GLushort *us_map = (GLushort *)map;
-	 const GLushort *us_indices = (const GLushort *)indices;
-
-	 for (i = 0; i < count; i++)
-	    us_map[i] = us_indices[i] - min_index;
-	 break;
-      }
-      case GL_UNSIGNED_BYTE:  {
-	 GLubyte *ub_map = (GLubyte *)map;
-	 const GLubyte *ub_indices = (const GLubyte *)indices;
-
-	 for (i = 0; i < count; i++)
-	    ub_map[i] = ub_indices[i] - min_index;
-	 break;
-      }
-      }
-
-      ctx->Driver.UnmapBuffer(ctx, 
-			      GL_ELEMENT_ARRAY_BUFFER_ARB, 
-			      bufferobj);
-
-   }
-}
-
-
-
-void brw_upload_indices( struct brw_context *brw,
-			 const struct brw_draw_index_buffer *index_buffer)
-{
-   struct intel_context *intel = &brw->intel;
-   GLuint ib_size = get_size(index_buffer->type) * index_buffer->count;
-   struct gl_buffer_object *bufferobj = index_buffer->obj;
-   GLuint offset = (GLuint)index_buffer->ptr;
-
-   /* Already turned into a proper VBO:
-    */
-   if (!index_buffer->obj->Name) {
-      rebase_indices_to_vbo_indices(brw, index_buffer, &bufferobj, &offset );
-   }
 
    /* Emit the indexbuffer packet:
     */
diff --git a/src/mesa/drivers/dri/i965/brw_exec.c b/src/mesa/drivers/dri/i965/brw_exec.c
deleted file mode 100644
index fc06c3c3617..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec.c
+++ /dev/null
@@ -1,125 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.3
- *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *    Keith Whitwell <[email protected]>
- */
-
-
-#include "api_arrayelt.h"
-#include "glheader.h"
-#include "imports.h"
-#include "context.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "dlist.h"
-#include "vtxfmt.h"
-
-#include "brw_exec.h"
-
-
-void brw_exec_init( GLcontext *ctx )
-{
-   struct brw_exec_context *exec = CALLOC_STRUCT(brw_exec_context);
-
-   if (ctx->swtnl_im == NULL) {
-      ctx->swtnl_im = CALLOC_STRUCT(brw_exec_save);
-   }
-
-   exec->ctx = ctx;
-   IMM_CONTEXT(ctx)->exec = exec;
-
-   /* Initialize the arrayelt helper
-    */
-   if (!ctx->aelt_context &&
-       !_ae_create_context( ctx )) 
-      return;
-
-   brw_exec_vtx_init( exec );
-   brw_exec_array_init( exec );
-
-   ctx->Driver.NeedFlush = 0;
-   ctx->Driver.CurrentExecPrimitive = PRIM_OUTSIDE_BEGIN_END;
-   ctx->Driver.FlushVertices = brw_exec_FlushVertices;
-
-   exec->eval.recalculate_maps = 1;
-}
-
-
-void brw_exec_destroy( GLcontext *ctx )
-{
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   if (ctx->aelt_context) {
-      _ae_destroy_context( ctx );
-      ctx->aelt_context = NULL;
-   }
-
-   brw_exec_vtx_destroy( exec );
-   brw_exec_array_destroy( exec );
-
-   if (exec) {
-      FREE(exec);
-      IMM_CONTEXT(ctx)->exec = NULL;
-   }
-
-   if (IMM_CONTEXT(ctx)->exec == NULL &&
-       IMM_CONTEXT(ctx)->save == NULL) {
-      FREE(IMM_CONTEXT(ctx));
-      ctx->swtnl_im = NULL;
-   }
-}
-
-/* Really want to install these callbacks to a central facility to be
- * invoked according to the state flags.  That will have to wait for a
- * mesa rework:
- */ 
-void brw_exec_invalidate_state( GLcontext *ctx, GLuint new_state )
-{
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   if (new_state & (_NEW_PROGRAM|_NEW_EVAL))
-      exec->eval.recalculate_maps = 1;
-
-   _ae_invalidate_state(ctx, new_state);
-}
-
-
-void brw_exec_wakeup( GLcontext *ctx )
-{
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   ctx->Driver.FlushVertices = brw_exec_FlushVertices;
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;
-
-   /* Hook our functions into exec and compile dispatch tables.
-    */
-   _mesa_install_exec_vtxfmt( ctx, &exec->vtxfmt );
-
-   /* Assume we haven't been getting state updates either:
-    */
-   brw_exec_invalidate_state( ctx, ~0 );
-}
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_exec.h b/src/mesa/drivers/dri/i965/brw_exec.h
deleted file mode 100644
index f07b4485874..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec.h
+++ /dev/null
@@ -1,150 +0,0 @@
-/**************************************************************************
-
-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
-TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- *
- */
-
-#ifndef __BRW_EXEC_H__
-#define __BRW_EXEC_H__
-
-#include "mtypes.h"
-#include "brw_attrib.h"
-#include "brw_draw.h"
-
-
-#define BRW_MAX_PRIM 64
-
-/* Wierd implementation stuff:
- */
-#define BRW_VERT_BUFFER_SIZE (1024*16)	/* dwords == 64k */
-#define BRW_MAX_ATTR_CODEGEN 16 
-#define ERROR_ATTRIB 16
-
-
-
-
-struct brw_exec_eval1_map {
-   struct gl_1d_map *map;
-   GLuint sz;
-};
-
-struct brw_exec_eval2_map {
-   struct gl_2d_map *map;
-   GLuint sz;
-};
-
-
-
-struct brw_exec_copied_vtx {
-   GLfloat buffer[BRW_ATTRIB_MAX * 4 * BRW_MAX_COPIED_VERTS];
-   GLuint nr;
-};
-
-
-typedef void (*brw_attrfv_func)( const GLfloat * );
-
-
-struct brw_exec_context
-{
-   GLcontext *ctx;   
-   GLvertexformat vtxfmt;
-
-   struct {
-      struct gl_buffer_object *bufferobj;
-      GLubyte *buffer_map;
-
-      GLuint vertex_size;
-
-      struct brw_draw_prim prim[BRW_MAX_PRIM];
-      GLuint prim_count;
-
-      GLfloat *vbptr;		     /* cursor, points into buffer */
-      GLfloat vertex[BRW_ATTRIB_MAX*4]; /* current vertex */
-
-      GLfloat *current[BRW_ATTRIB_MAX]; /* points into ctx->Current, ctx->Light.Material */
-      GLfloat CurrentFloatEdgeFlag;
-
-      GLuint vert_count;
-      GLuint max_vert;
-      struct brw_exec_copied_vtx copied;
-
-      GLubyte attrsz[BRW_ATTRIB_MAX];
-      GLubyte active_sz[BRW_ATTRIB_MAX];
-
-      GLfloat *attrptr[BRW_ATTRIB_MAX]; 
-      struct gl_client_array arrays[BRW_ATTRIB_MAX];
-      const struct gl_client_array *inputs[BRW_ATTRIB_MAX];
-   } vtx;
-
-   
-   struct {
-      GLboolean recalculate_maps;
-      struct brw_exec_eval1_map map1[BRW_ATTRIB_MAX];
-      struct brw_exec_eval2_map map2[BRW_ATTRIB_MAX];
-   } eval;
-
-   struct {
-      const struct gl_client_array *inputs[BRW_ATTRIB_MAX];
-
-      struct gl_buffer_object *index_obj;
-   } array;
-};
-
-
-
-/* External API:
- */
-void brw_exec_init( GLcontext *ctx );
-void brw_exec_destroy( GLcontext *ctx );
-void brw_exec_invalidate_state( GLcontext *ctx, GLuint new_state );
-void brw_exec_FlushVertices( GLcontext *ctx, GLuint flags );
-void brw_exec_wakeup( GLcontext *ctx );
-
-
-/* Internal functions:
- */
-void brw_exec_array_init( struct brw_exec_context *exec );
-void brw_exec_array_destroy( struct brw_exec_context *exec );
-
-
-void brw_exec_vtx_init( struct brw_exec_context *exec );
-void brw_exec_vtx_destroy( struct brw_exec_context *exec );
-void brw_exec_vtx_flush( struct brw_exec_context *exec );
-void brw_exec_vtx_wrap( struct brw_exec_context *exec );
-
-void brw_exec_eval_update( struct brw_exec_context *exec );
-
-void brw_exec_do_EvalCoord2f( struct brw_exec_context *exec, 
-				     GLfloat u, GLfloat v );
-
-void brw_exec_do_EvalCoord1f( struct brw_exec_context *exec,
-				     GLfloat u);
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_exec_api.c b/src/mesa/drivers/dri/i965/brw_exec_api.c
deleted file mode 100644
index 8b243c60842..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec_api.c
+++ /dev/null
@@ -1,726 +0,0 @@
-/**************************************************************************
-
-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
-TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "context.h"
-#include "macros.h"
-#include "vtxfmt.h"
-#include "dlist.h"
-#include "state.h"
-#include "light.h"
-#include "api_arrayelt.h"
-#include "api_noop.h"
-#include "dispatch.h"
-
-#include "brw_exec.h"
-#include "intel_context.h"
-
-
-static void reset_attrfv( struct brw_exec_context *exec );
-
-
-/* Close off the last primitive, execute the buffer, restart the
- * primitive.  
- */
-static void brw_exec_wrap_buffers( struct brw_exec_context *exec )
-{
-   if (exec->vtx.prim_count == 0) {
-      exec->vtx.copied.nr = 0;
-      exec->vtx.vert_count = 0;
-      exec->vtx.vbptr = (GLfloat *)exec->vtx.buffer_map;
-   }
-   else {
-      GLuint last_begin = exec->vtx.prim[exec->vtx.prim_count-1].begin;
-      GLuint last_count;
-
-      if (exec->ctx->Driver.CurrentExecPrimitive != GL_POLYGON+1) {
-	 GLint i = exec->vtx.prim_count - 1;
-	 assert(i >= 0);
-	 exec->vtx.prim[i].count = (exec->vtx.vert_count - 
-				    exec->vtx.prim[i].start);
-      }
-
-      last_count = exec->vtx.prim[exec->vtx.prim_count-1].count;
-
-      /* Execute the buffer and save copied vertices.
-       */
-      if (exec->vtx.vert_count)
-	 brw_exec_vtx_flush( exec );
-      else {
-	 exec->vtx.prim_count = 0;
-	 exec->vtx.copied.nr = 0;
-      }
-
-      /* Emit a glBegin to start the new list.
-       */
-      assert(exec->vtx.prim_count == 0);
-
-      if (exec->ctx->Driver.CurrentExecPrimitive != GL_POLYGON+1) {
-	 exec->vtx.prim[0].mode = exec->ctx->Driver.CurrentExecPrimitive;
-	 exec->vtx.prim[0].start = 0;
-	 exec->vtx.prim[0].count = 0;
-	 exec->vtx.prim_count++;
-      
-	 if (exec->vtx.copied.nr == last_count)
-	    exec->vtx.prim[0].begin = last_begin;
-      }
-   }
-}
-
-
-/* Deal with buffer wrapping where provoked by the vertex buffer
- * filling up, as opposed to upgrade_vertex().
- */
-void brw_exec_vtx_wrap( struct brw_exec_context *exec )
-{
-   GLfloat *data = exec->vtx.copied.buffer;
-   GLuint i;
-
-   /* Run pipeline on current vertices, copy wrapped vertices
-    * to exec->vtx.copied.
-    */
-   brw_exec_wrap_buffers( exec );
-   
-   /* Copy stored stored vertices to start of new list. 
-    */
-   assert(exec->vtx.max_vert - exec->vtx.vert_count > exec->vtx.copied.nr);
-
-   for (i = 0 ; i < exec->vtx.copied.nr ; i++) {
-      _mesa_memcpy( exec->vtx.vbptr, data, 
-		    exec->vtx.vertex_size * sizeof(GLfloat));
-      exec->vtx.vbptr += exec->vtx.vertex_size;
-      data += exec->vtx.vertex_size;
-      exec->vtx.vert_count++;
-   }
-
-   exec->vtx.copied.nr = 0;
-}
-
-
-/*
- * Copy the active vertex's values to the ctx->Current fields.
- */
-static void brw_exec_copy_to_current( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-   GLuint i;
-
-   for (i = BRW_ATTRIB_POS+1 ; i < BRW_ATTRIB_MAX ; i++) {
-      if (exec->vtx.attrsz[i]) {
-         /* Note: the exec->vtx.current[i] pointers point into the
-          * ctx->Current.Attrib and ctx->Light.Material.Attrib arrays.
-          */
-	 COPY_CLEAN_4V(exec->vtx.current[i], 
-		       exec->vtx.attrsz[i], 
-		       exec->vtx.attrptr[i]);
-
-	 /* This triggers rather too much recalculation of Mesa state
-	  * that doesn't get used (eg light positions).
-	  */
-	 if (i >= BRW_ATTRIB_MAT_FRONT_AMBIENT &&
-	     i <= BRW_ATTRIB_MAT_BACK_INDEXES)
-	    ctx->NewState |= _NEW_LIGHT;
-      }
-   }
-
-   /* color index is special (it's not a float[4] so COPY_CLEAN_4V above
-    * will trash adjacent memory!)
-    */
-   if (exec->vtx.attrsz[BRW_ATTRIB_INDEX]) {
-      ctx->Current.Index = exec->vtx.attrptr[BRW_ATTRIB_INDEX][0];
-   }
-
-   /* Edgeflag requires additional treatment:
-    */
-   if (exec->vtx.attrsz[BRW_ATTRIB_EDGEFLAG]) {
-      ctx->Current.EdgeFlag = (exec->vtx.CurrentFloatEdgeFlag == 1.0);
-   }
-
-   /* Colormaterial -- this kindof sucks.
-    */
-   if (ctx->Light.ColorMaterialEnabled &&
-       exec->vtx.attrsz[BRW_ATTRIB_COLOR0]) {
-      _mesa_update_color_material(ctx, 
-				  ctx->Current.Attrib[BRW_ATTRIB_COLOR0]);
-   }
-
-   ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-}
-
-
-static void brw_exec_copy_from_current( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-   GLint i;
-
-   /* Edgeflag requires additional treatment:
-    */
-   exec->vtx.CurrentFloatEdgeFlag = 
-      (GLfloat)ctx->Current.EdgeFlag;
-   
-   for (i = BRW_ATTRIB_POS+1 ; i < BRW_ATTRIB_MAX ; i++) 
-      switch (exec->vtx.attrsz[i]) {
-      case 4: exec->vtx.attrptr[i][3] = exec->vtx.current[i][3];
-      case 3: exec->vtx.attrptr[i][2] = exec->vtx.current[i][2];
-      case 2: exec->vtx.attrptr[i][1] = exec->vtx.current[i][1];
-      case 1: exec->vtx.attrptr[i][0] = exec->vtx.current[i][0];
-	 break;
-      }
-
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;
-}
-
-
-/* Flush existing data, set new attrib size, replay copied vertices.
- */ 
-static void brw_exec_wrap_upgrade_vertex( struct brw_exec_context *exec,
-					  GLuint attr,
-					  GLuint newsz )
-{
-   GLcontext *ctx = exec->ctx;
-   GLint lastcount = exec->vtx.vert_count;
-   GLfloat *tmp;
-   GLuint oldsz;
-   GLuint i;
-
-   /* Run pipeline on current vertices, copy wrapped vertices
-    * to exec->vtx.copied.
-    */
-   brw_exec_wrap_buffers( exec );
-
-
-   /* Do a COPY_TO_CURRENT to ensure back-copying works for the case
-    * when the attribute already exists in the vertex and is having
-    * its size increased.  
-    */
-   brw_exec_copy_to_current( exec );
-
-
-   /* Heuristic: Attempt to isolate attributes received outside
-    * begin/end so that they don't bloat the vertices.
-    */
-   if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END &&
-       exec->vtx.attrsz[attr] == 0 && 
-       lastcount > 8 &&
-       exec->vtx.vertex_size) {
-      reset_attrfv( exec );
-   }
-
-   /* Fix up sizes:
-    */
-   oldsz = exec->vtx.attrsz[attr];
-   exec->vtx.attrsz[attr] = newsz;
-
-   exec->vtx.vertex_size += newsz - oldsz;
-   exec->vtx.max_vert = BRW_VERT_BUFFER_SIZE / exec->vtx.vertex_size;
-   exec->vtx.vert_count = 0;
-   exec->vtx.vbptr = (GLfloat *)exec->vtx.buffer_map;
-   
-
-   /* Recalculate all the attrptr[] values
-    */
-   for (i = 0, tmp = exec->vtx.vertex ; i < BRW_ATTRIB_MAX ; i++) {
-      if (exec->vtx.attrsz[i]) {
-	 exec->vtx.attrptr[i] = tmp;
-	 tmp += exec->vtx.attrsz[i];
-      }
-      else 
-	 exec->vtx.attrptr[i] = NULL; /* will not be dereferenced */
-   }
-
-   /* Copy from current to repopulate the vertex with correct values.
-    */
-   brw_exec_copy_from_current( exec );
-
-   /* Replay stored vertices to translate them
-    * to new format here.
-    *
-    * -- No need to replay - just copy piecewise
-    */
-   if (exec->vtx.copied.nr)
-   {
-      GLfloat *data = exec->vtx.copied.buffer;
-      GLfloat *dest = exec->vtx.vbptr;
-      GLuint j;
-
-      assert(exec->vtx.vbptr == (GLfloat *)exec->vtx.buffer_map);
-      
-      for (i = 0 ; i < exec->vtx.copied.nr ; i++) {
-	 for (j = 0 ; j < BRW_ATTRIB_MAX ; j++) {
-	    if (exec->vtx.attrsz[j]) {
-	       if (j == attr) {
-		  if (oldsz) {
-		     COPY_CLEAN_4V( dest, oldsz, data );
-		     data += oldsz;
-		     dest += newsz;
-		  } else {
-		     COPY_SZ_4V( dest, newsz, exec->vtx.current[j] );
-		     dest += newsz;
-		  }
-	       }
-	       else {
-		  GLuint sz = exec->vtx.attrsz[j];
-		  COPY_SZ_4V( dest, sz, data );
-		  dest += sz;
-		  data += sz;
-	       }
-	    }
-	 }
-      }
-
-      exec->vtx.vbptr = dest;
-      exec->vtx.vert_count += exec->vtx.copied.nr;
-      exec->vtx.copied.nr = 0;
-   }
-}
-
-
-static void brw_exec_fixup_vertex( GLcontext *ctx,
-				   GLuint attr, GLuint sz )
-{
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-   int i;
-
-   if (sz > exec->vtx.attrsz[attr]) {
-      /* New size is larger.  Need to flush existing vertices and get
-       * an enlarged vertex format.
-       */
-      brw_exec_wrap_upgrade_vertex( exec, attr, sz );
-   }
-   else if (sz < exec->vtx.active_sz[attr]) {
-      static const GLfloat id[4] = { 0, 0, 0, 1 };
-
-      /* New size is smaller - just need to fill in some
-       * zeros.  Don't need to flush or wrap.
-       */
-      for (i = sz ; i <= exec->vtx.attrsz[attr] ; i++)
-	 exec->vtx.attrptr[attr][i-1] = id[i-1];
-   }
-
-   exec->vtx.active_sz[attr] = sz;
-
-   /* Does setting NeedFlush belong here?  Necessitates resetting
-    * vtxfmt on each flush (otherwise flags won't get reset
-    * afterwards).
-    */
-   if (attr == 0) 
-      exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;
-   else 
-      exec->ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;
-}
-
-
-
-
-/* 
- */
-#define ATTR( A, N, V0, V1, V2, V3 )				\
-do {								\
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;	\
-								\
-   if (exec->vtx.active_sz[A] != N)				\
-      brw_exec_fixup_vertex(ctx, A, N);			\
-								\
-   {								\
-      GLfloat *dest = exec->vtx.attrptr[A];			\
-      if (N>0) dest[0] = V0;					\
-      if (N>1) dest[1] = V1;					\
-      if (N>2) dest[2] = V2;					\
-      if (N>3) dest[3] = V3;					\
-   }								\
-								\
-   if ((A) == 0) {						\
-      GLuint i;							\
-								\
-      for (i = 0; i < exec->vtx.vertex_size; i++)		\
-	 exec->vtx.vbptr[i] = exec->vtx.vertex[i];		\
-								\
-      exec->vtx.vbptr += exec->vtx.vertex_size;			\
-      exec->ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;	\
-								\
-      if (++exec->vtx.vert_count >= exec->vtx.max_vert)		\
-	 brw_exec_vtx_wrap( exec );				\
-   }								\
-} while (0)
-
-
-#define ERROR() _mesa_error( ctx, GL_INVALID_ENUM, __FUNCTION__ )
-#define TAG(x) brw_##x
-
-#include "brw_attrib_tmp.h"
-
-
-
-
-
-/* Eval
- */
-static void GLAPIENTRY brw_exec_EvalCoord1f( GLfloat u )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   {
-      GLint i;
-      if (exec->eval.recalculate_maps) 
-	 brw_exec_eval_update( exec );
-
-      for (i = 0 ; i <= BRW_ATTRIB_INDEX ; i++) {
-	 if (exec->eval.map1[i].map) 
-	    if (exec->vtx.active_sz[i] != exec->eval.map1[i].sz)
-	       brw_exec_fixup_vertex( ctx, i, exec->eval.map1[i].sz );
-      }
-   }
-
-
-   _mesa_memcpy( exec->vtx.copied.buffer, exec->vtx.vertex, 
-                 exec->vtx.vertex_size * sizeof(GLfloat));
-
-   brw_exec_do_EvalCoord1f( exec, u );
-
-   _mesa_memcpy( exec->vtx.vertex, exec->vtx.copied.buffer,
-                 exec->vtx.vertex_size * sizeof(GLfloat));
-}
-
-static void GLAPIENTRY brw_exec_EvalCoord2f( GLfloat u, GLfloat v )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   {
-      GLint i;
-      if (exec->eval.recalculate_maps) 
-	 brw_exec_eval_update( exec );
-
-      for (i = 0 ; i <= BRW_ATTRIB_INDEX ; i++) {
-	 if (exec->eval.map2[i].map) 
-	    if (exec->vtx.active_sz[i] != exec->eval.map2[i].sz)
-	       brw_exec_fixup_vertex( ctx, i, exec->eval.map2[i].sz );
-      }
-
-      if (ctx->Eval.AutoNormal) 
-	 if (exec->vtx.active_sz[BRW_ATTRIB_NORMAL] != 3)
-	    brw_exec_fixup_vertex( ctx, BRW_ATTRIB_NORMAL, 3 );
-   }
-
-   _mesa_memcpy( exec->vtx.copied.buffer, exec->vtx.vertex, 
-                 exec->vtx.vertex_size * sizeof(GLfloat));
-
-   brw_exec_do_EvalCoord2f( exec, u, v );
-
-   _mesa_memcpy( exec->vtx.vertex, exec->vtx.copied.buffer, 
-                 exec->vtx.vertex_size * sizeof(GLfloat));
-}
-
-static void GLAPIENTRY brw_exec_EvalCoord1fv( const GLfloat *u )
-{
-   brw_exec_EvalCoord1f( u[0] );
-}
-
-static void GLAPIENTRY brw_exec_EvalCoord2fv( const GLfloat *u )
-{
-   brw_exec_EvalCoord2f( u[0], u[1] );
-}
-
-static void GLAPIENTRY brw_exec_EvalPoint1( GLint i )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLfloat du = ((ctx->Eval.MapGrid1u2 - ctx->Eval.MapGrid1u1) /
-		 (GLfloat) ctx->Eval.MapGrid1un);
-   GLfloat u = i * du + ctx->Eval.MapGrid1u1;
-
-   brw_exec_EvalCoord1f( u );
-}
-
-
-static void GLAPIENTRY brw_exec_EvalPoint2( GLint i, GLint j )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   GLfloat du = ((ctx->Eval.MapGrid2u2 - ctx->Eval.MapGrid2u1) / 
-		 (GLfloat) ctx->Eval.MapGrid2un);
-   GLfloat dv = ((ctx->Eval.MapGrid2v2 - ctx->Eval.MapGrid2v1) / 
-		 (GLfloat) ctx->Eval.MapGrid2vn);
-   GLfloat u = i * du + ctx->Eval.MapGrid2u1;
-   GLfloat v = j * dv + ctx->Eval.MapGrid2v1;
-
-   brw_exec_EvalCoord2f( u, v );
-}
-
-
-/* Build a list of primitives on the fly.  Keep
- * ctx->Driver.CurrentExecPrimitive uptodate as well.
- */
-static void GLAPIENTRY brw_exec_Begin( GLenum mode )
-{
-   GET_CURRENT_CONTEXT( ctx ); 
-
-   if (ctx->Driver.CurrentExecPrimitive == GL_POLYGON+1) {
-      struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-      int i;
-
-      if (ctx->NewState) {
-	 _mesa_update_state( ctx );
-
-         if ((ctx->VertexProgram.Enabled && !ctx->VertexProgram._Enabled) ||
-            (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled)) {
-            _mesa_error(ctx, GL_INVALID_OPERATION,
-                        "glBegin (invalid vertex/fragment program)");
-            return;
-         }
-
-	 CALL_Begin(ctx->Exec, (mode));
-	 return;
-      }
-
-      /* Heuristic: attempt to isolate attributes occuring outside
-       * begin/end pairs.
-       */
-      if (exec->vtx.vertex_size && !exec->vtx.attrsz[0]) 
-	 brw_exec_FlushVertices( ctx, ~0 );
-
-      i = exec->vtx.prim_count++;
-      exec->vtx.prim[i].mode = mode;
-      exec->vtx.prim[i].begin = 1;
-      exec->vtx.prim[i].end = 0;
-      exec->vtx.prim[i].indexed = 0;
-      exec->vtx.prim[i].weak = 0;
-      exec->vtx.prim[i].pad = 0;
-      exec->vtx.prim[i].start = exec->vtx.vert_count;
-      exec->vtx.prim[i].count = 0;
-
-      ctx->Driver.CurrentExecPrimitive = mode;
-   }
-   else 
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" );
-      
-}
-
-static GLuint brw_max_prim( GLcontext *ctx )
-{
-        struct intel_context *intel = intel_context( ctx );
-        if (intel->numClipRects <= 1)
-                return BRW_MAX_PRIM;
-        return BRW_MAX_PRIM/intel->numClipRects;
-}
-
-static void GLAPIENTRY brw_exec_End( void )
-{
-   GET_CURRENT_CONTEXT( ctx ); 
-
-   if (ctx->Driver.CurrentExecPrimitive != GL_POLYGON+1) {
-      struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-      int idx = exec->vtx.vert_count;
-      int i = exec->vtx.prim_count - 1;
-
-      exec->vtx.prim[i].end = 1; 
-      exec->vtx.prim[i].count = idx - exec->vtx.prim[i].start;
-
-      ctx->Driver.CurrentExecPrimitive = GL_POLYGON+1;
-
-      if (exec->vtx.prim_count >= brw_max_prim(ctx))
-	 brw_exec_vtx_flush( exec );	
-   }
-   else 
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" );
-}
-
-
-static void brw_exec_vtxfmt_init( struct brw_exec_context *exec )
-{
-   GLvertexformat *vfmt = &exec->vtxfmt;
-
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
-   vfmt->Begin = brw_exec_Begin;
-   vfmt->CallList = _mesa_CallList;
-   vfmt->CallLists = _mesa_CallLists;
-   vfmt->End = brw_exec_End;
-   vfmt->EvalCoord1f = brw_exec_EvalCoord1f;
-   vfmt->EvalCoord1fv = brw_exec_EvalCoord1fv;
-   vfmt->EvalCoord2f = brw_exec_EvalCoord2f;
-   vfmt->EvalCoord2fv = brw_exec_EvalCoord2fv;
-   vfmt->EvalPoint1 = brw_exec_EvalPoint1;
-   vfmt->EvalPoint2 = brw_exec_EvalPoint2;
-
-   vfmt->Rectf = _mesa_noop_Rectf;
-   vfmt->EvalMesh1 = _mesa_noop_EvalMesh1;
-   vfmt->EvalMesh2 = _mesa_noop_EvalMesh2;
-
-
-   /* from attrib_tmp.h:
-    */
-   vfmt->Color3f = brw_Color3f;
-   vfmt->Color3fv = brw_Color3fv;
-   vfmt->Color4f = brw_Color4f;
-   vfmt->Color4fv = brw_Color4fv;
-   vfmt->FogCoordfEXT = brw_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = brw_FogCoordfvEXT;
-   vfmt->MultiTexCoord1fARB = brw_MultiTexCoord1f;
-   vfmt->MultiTexCoord1fvARB = brw_MultiTexCoord1fv;
-   vfmt->MultiTexCoord2fARB = brw_MultiTexCoord2f;
-   vfmt->MultiTexCoord2fvARB = brw_MultiTexCoord2fv;
-   vfmt->MultiTexCoord3fARB = brw_MultiTexCoord3f;
-   vfmt->MultiTexCoord3fvARB = brw_MultiTexCoord3fv;
-   vfmt->MultiTexCoord4fARB = brw_MultiTexCoord4f;
-   vfmt->MultiTexCoord4fvARB = brw_MultiTexCoord4fv;
-   vfmt->Normal3f = brw_Normal3f;
-   vfmt->Normal3fv = brw_Normal3fv;
-   vfmt->SecondaryColor3fEXT = brw_SecondaryColor3fEXT;
-   vfmt->SecondaryColor3fvEXT = brw_SecondaryColor3fvEXT;
-   vfmt->TexCoord1f = brw_TexCoord1f;
-   vfmt->TexCoord1fv = brw_TexCoord1fv;
-   vfmt->TexCoord2f = brw_TexCoord2f;
-   vfmt->TexCoord2fv = brw_TexCoord2fv;
-   vfmt->TexCoord3f = brw_TexCoord3f;
-   vfmt->TexCoord3fv = brw_TexCoord3fv;
-   vfmt->TexCoord4f = brw_TexCoord4f;
-   vfmt->TexCoord4fv = brw_TexCoord4fv;
-   vfmt->Vertex2f = brw_Vertex2f;
-   vfmt->Vertex2fv = brw_Vertex2fv;
-   vfmt->Vertex3f = brw_Vertex3f;
-   vfmt->Vertex3fv = brw_Vertex3fv;
-   vfmt->Vertex4f = brw_Vertex4f;
-   vfmt->Vertex4fv = brw_Vertex4fv;
-   
-   vfmt->VertexAttrib1fARB = brw_VertexAttrib1fARB;
-   vfmt->VertexAttrib1fvARB = brw_VertexAttrib1fvARB;
-   vfmt->VertexAttrib2fARB = brw_VertexAttrib2fARB;
-   vfmt->VertexAttrib2fvARB = brw_VertexAttrib2fvARB;
-   vfmt->VertexAttrib3fARB = brw_VertexAttrib3fARB;
-   vfmt->VertexAttrib3fvARB = brw_VertexAttrib3fvARB;
-   vfmt->VertexAttrib4fARB = brw_VertexAttrib4fARB;
-   vfmt->VertexAttrib4fvARB = brw_VertexAttrib4fvARB;
-
-   vfmt->VertexAttrib1fNV = brw_VertexAttrib1fNV;
-   vfmt->VertexAttrib1fvNV = brw_VertexAttrib1fvNV;
-   vfmt->VertexAttrib2fNV = brw_VertexAttrib2fNV;
-   vfmt->VertexAttrib2fvNV = brw_VertexAttrib2fvNV;
-   vfmt->VertexAttrib3fNV = brw_VertexAttrib3fNV;
-   vfmt->VertexAttrib3fvNV = brw_VertexAttrib3fvNV;
-   vfmt->VertexAttrib4fNV = brw_VertexAttrib4fNV;
-   vfmt->VertexAttrib4fvNV = brw_VertexAttrib4fvNV;
-
-   vfmt->Materialfv = brw_Materialfv;
-
-   vfmt->EdgeFlag = brw_EdgeFlag;
-   vfmt->Indexf = brw_Indexf;
-   vfmt->Indexfv = brw_Indexfv;
-
-}
-
-
-static void brw_exec_current_init( struct brw_exec_context *exec ) 
-{
-   GLcontext *ctx = exec->ctx;
-   GLint i;
-
-   /* setup the pointers for the typical 16 vertex attributes */
-   for (i = 0; i < BRW_ATTRIB_FIRST_MATERIAL; i++) 
-      exec->vtx.current[i] = ctx->Current.Attrib[i];
-
-   /* setup pointers for the 12 material attributes */
-   for (i = 0; i < MAT_ATTRIB_MAX; i++)
-      exec->vtx.current[BRW_ATTRIB_FIRST_MATERIAL + i] = 
-	 ctx->Light.Material.Attrib[i];
-
-   exec->vtx.current[BRW_ATTRIB_INDEX] = &ctx->Current.Index;
-   exec->vtx.current[BRW_ATTRIB_EDGEFLAG] = &exec->vtx.CurrentFloatEdgeFlag;
-}
-
-void brw_exec_vtx_init( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-   GLuint i;
-
-   /* Allocate a buffer object.  Will just reuse this object
-    * continuously.
-    */
-   exec->vtx.bufferobj = ctx->Array.NullBufferObj;
-   exec->vtx.buffer_map = ALIGN_MALLOC(BRW_VERT_BUFFER_SIZE * sizeof(GLfloat), 64);
-
-   brw_exec_current_init( exec );
-   brw_exec_vtxfmt_init( exec );
-
-   /* Hook our functions into the dispatch table.
-    */
-   _mesa_install_exec_vtxfmt( exec->ctx, &exec->vtxfmt );
-
-   for (i = 0 ; i < BRW_ATTRIB_MAX ; i++) {
-      exec->vtx.attrsz[i] = 0;
-      exec->vtx.active_sz[i] = 0;
-      exec->vtx.inputs[i] = &exec->vtx.arrays[i];
-   }
- 
-   exec->vtx.vertex_size = 0;
-}
-
-
-void brw_exec_vtx_destroy( struct brw_exec_context *exec )
-{
-   if (exec->vtx.buffer_map) {
-      ALIGN_FREE(exec->vtx.buffer_map);
-      exec->vtx.buffer_map = NULL;
-   }
-}
-
-
-void brw_exec_FlushVertices( GLcontext *ctx, GLuint flags )
-{
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-
-   if (exec->ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END)
-      return;
-
-   if (exec->vtx.vert_count) {
-      brw_exec_vtx_flush( exec );
-   }
-
-   if (exec->vtx.vertex_size) {
-      brw_exec_copy_to_current( exec );
-      reset_attrfv( exec );
-   }
-
-   exec->ctx->Driver.NeedFlush = 0;
-}
-
-
-static void reset_attrfv( struct brw_exec_context *exec )
-{   
-   GLuint i;
-
-   for (i = 0 ; i < BRW_ATTRIB_MAX ; i++) {
-      exec->vtx.attrsz[i] = 0;
-      exec->vtx.active_sz[i] = 0;
-   }
-
-   exec->vtx.vertex_size = 0;
-}
-      
diff --git a/src/mesa/drivers/dri/i965/brw_exec_array.c b/src/mesa/drivers/dri/i965/brw_exec_array.c
deleted file mode 100644
index ca19a198374..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec_array.c
+++ /dev/null
@@ -1,283 +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 "state.h"
-#include "api_validate.h"
-#include "api_noop.h"
-#include "dispatch.h"
-
-#include "brw_attrib.h"
-#include "brw_draw.h"
-#include "brw_exec.h"
-#include "brw_fallback.h"
-
-static GLuint get_max_index( GLuint count, GLuint type, 
-			     const GLvoid *indices )
-{
-   GLint i;
-
-   /* Compute max element.  This is only needed for upload of non-VBO,
-    * non-constant data elements.
-    *
-    * XXX: Postpone this calculation until it is known that it is
-    * needed.  Otherwise could scan this pointlessly in the all-vbo
-    * case.
-    */
-   switch(type) {
-   case GL_UNSIGNED_INT: {
-      const GLuint *ui_indices = (const GLuint *)indices;
-      GLuint max_ui = 0;
-      for (i = 0; i < count; i++)
-	 if (ui_indices[i] > max_ui)
-	    max_ui = ui_indices[i];
-      return max_ui;
-   }
-   case GL_UNSIGNED_SHORT: {
-      const GLushort *us_indices = (const GLushort *)indices;
-      GLuint max_us = 0;
-      for (i = 0; i < count; i++)
-	 if (us_indices[i] > max_us)
-	    max_us = us_indices[i];
-      return max_us;
-   }
-   case GL_UNSIGNED_BYTE: {
-      const GLubyte *ub_indices = (const GLubyte *)indices;
-      GLuint max_ub = 0;
-      for (i = 0; i < count; i++)
-	 if (ub_indices[i] > max_ub)
-	    max_ub = ub_indices[i];
-      return max_ub;
-   }
-   default:
-      return 0;
-   }
-}
-
-
-
-
-/***********************************************************************
- * API functions.
- */
-
-static void GLAPIENTRY
-brw_exec_DrawArrays(GLenum mode, GLint start, GLsizei count)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-   struct brw_draw_prim prim[1];
-   GLboolean ok;
-
-   if (!_mesa_validate_DrawArrays( ctx, mode, start, count ))
-      return;
-
-   FLUSH_CURRENT( ctx, 0 );
-
-   if (ctx->NewState)
-      _mesa_update_state( ctx );
-      
-   prim[0].begin = 1;
-   prim[0].end = 1;
-   prim[0].weak = 0;
-   prim[0].pad = 0;
-
-   if (exec->array.inputs[0]->BufferObj->Name) {
-      /* Use vertex attribute as a hint to tell us if we expect all
-       * arrays to be in VBO's and if so, don't worry about avoiding
-       * the upload of elements < start.
-       */
-      prim[0].mode = mode;
-      prim[0].start = start;
-      prim[0].count = count;
-      prim[0].indexed = 0;
-
-      ok = brw_draw_prims( ctx, exec->array.inputs, prim, 1, NULL, 0, start + count, 0 );
-   }
-   else {
-      /* If not using VBO's, we don't want to upload any more elements
-       * than necessary from the arrays as they will not be valid next
-       * time the application tries to draw with them.
-       */
-      prim[0].mode = mode;
-      prim[0].start = 0;
-      prim[0].count = count;
-      prim[0].indexed = 0;
-
-      ok = brw_draw_prims( ctx, exec->array.inputs, prim, 1, NULL, start, start + count, 0 );
-   }
-   
-   if (!ok) {
-      brw_fallback(ctx);
-      CALL_DrawArrays(ctx->Exec, ( mode, start, count ));
-      brw_unfallback(ctx);
-   }
-}
-
-
-
-static void GLAPIENTRY
-brw_exec_DrawRangeElements(GLenum mode,
-			   GLuint start, GLuint end,
-			   GLsizei count, GLenum type, const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   struct brw_exec_context *exec = IMM_CONTEXT(ctx)->exec;
-   struct brw_draw_index_buffer ib;
-   struct brw_draw_prim prim[1];
-
-   if (!_mesa_validate_DrawRangeElements( ctx, mode, start, end, count, type, indices ))
-      return;
-
-   FLUSH_CURRENT( ctx, 0 );
-
-   if (ctx->NewState)
-      _mesa_update_state( ctx );
-      
-   ib.count = count;
-   ib.type = type; 
-   ib.obj = ctx->Array.ElementArrayBufferObj;
-   ib.ptr = indices;
-
-   if (ctx->Array.ElementArrayBufferObj->Name) {
-      /* Use the fact that indices are in a VBO as a hint that the
-       * program has put all the arrays in VBO's and we don't have to
-       * worry about performance implications of start > 0.
-       *
-       * XXX: consider passing start as min_index to draw_prims instead.
-       */
-      ib.rebase = 0;
-   }
-   else {
-      ib.rebase = start;
-   }
-
-   prim[0].begin = 1;
-   prim[0].end = 1;
-   prim[0].weak = 0;
-   prim[0].pad = 0;
-   prim[0].mode = mode;
-   prim[0].start = 0;
-   prim[0].count = count;
-   prim[0].indexed = 1;
-
-   if (!brw_draw_prims( ctx, exec->array.inputs, prim, 1, &ib, ib.rebase, end+1, 0 )) {
-      brw_fallback(ctx);
-      CALL_DrawRangeElements(ctx->Exec, (mode, start, end, count, type, indices));
-      brw_unfallback(ctx);
-   }
-}
-
-
-static void GLAPIENTRY
-brw_exec_DrawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   GLuint max_index;
-
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
-      return;
-
-   if (ctx->Array.ElementArrayBufferObj->Name) {
-      const GLvoid *map = ctx->Driver.MapBuffer(ctx,
-						 GL_ELEMENT_ARRAY_BUFFER_ARB,
-						 GL_DYNAMIC_READ_ARB,
-						 ctx->Array.ElementArrayBufferObj);
-
-      max_index = get_max_index(count, type, ADD_POINTERS(map, indices));
-
-      ctx->Driver.UnmapBuffer(ctx,
-			      GL_ELEMENT_ARRAY_BUFFER_ARB,
-			      ctx->Array.ElementArrayBufferObj);
-   }
-   else {
-      max_index = get_max_index(count, type, indices);
-   }
-
-   brw_exec_DrawRangeElements(mode, 0, max_index, count, type, indices);
-}
-
-
-/***********************************************************************
- * Initialization
- */
-
-
-static void init_arrays( GLcontext *ctx, 
-			 const struct gl_client_array *arrays[] )
-{
-   struct gl_array_object *obj = ctx->Array.ArrayObj;
-   GLuint i;
-
-   memset(arrays, 0, sizeof(*arrays) * BRW_ATTRIB_MAX);
-
-   arrays[BRW_ATTRIB_POS] = &obj->Vertex;
-   arrays[BRW_ATTRIB_NORMAL] = &obj->Normal;
-   arrays[BRW_ATTRIB_COLOR0] = &obj->Color;
-   arrays[BRW_ATTRIB_COLOR1] = &obj->SecondaryColor;
-   arrays[BRW_ATTRIB_FOG] = &obj->FogCoord;
-
-   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++)
-      arrays[BRW_ATTRIB_TEX0 + i] = &obj->TexCoord[i];
-
-   arrays[BRW_ATTRIB_INDEX] = &obj->Index;
-   arrays[BRW_ATTRIB_EDGEFLAG] = &obj->EdgeFlag;
-
-   for (i = BRW_ATTRIB_GENERIC0; i <= BRW_ATTRIB_GENERIC15; i++)
-      arrays[i] = &obj->VertexAttrib[i - BRW_ATTRIB_GENERIC0];
-}
-
-
-
-
-void brw_exec_array_init( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-
-   init_arrays(ctx, exec->array.inputs);
-
-#if 1
-   exec->vtxfmt.DrawArrays = brw_exec_DrawArrays;
-   exec->vtxfmt.DrawElements = brw_exec_DrawElements;
-   exec->vtxfmt.DrawRangeElements = brw_exec_DrawRangeElements;
-#else
-   exec->vtxfmt.DrawArrays = _mesa_noop_DrawArrays;
-   exec->vtxfmt.DrawElements = _mesa_noop_DrawElements;
-   exec->vtxfmt.DrawRangeElements = _mesa_noop_DrawRangeElements;
-#endif
-
-   exec->array.index_obj = ctx->Driver.NewBufferObject(ctx, 1, GL_ARRAY_BUFFER_ARB);
-}
-
-
-void brw_exec_array_destroy( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-
-   ctx->Driver.DeleteBuffer(ctx, exec->array.index_obj);
-}
diff --git a/src/mesa/drivers/dri/i965/brw_exec_draw.c b/src/mesa/drivers/dri/i965/brw_exec_draw.c
deleted file mode 100644
index 62bda9845b9..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec_draw.c
+++ /dev/null
@@ -1,227 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  5.1
- *
- * Copyright (C) 1999-2003  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *    Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "context.h"
-#include "enums.h"
-#include "state.h"
-#include "macros.h"
-
-#include "brw_exec.h"
-#include "brw_draw.h"
-#include "brw_fallback.h"
-
-
-static void brw_exec_debug_verts( struct brw_exec_context *exec )
-{
-   GLuint count = exec->vtx.vert_count;
-   GLuint i;
-
-   _mesa_printf("%s: %u vertices %d primitives, %d vertsize\n",
-		__FUNCTION__,
-		count,
-		exec->vtx.prim_count,
-		exec->vtx.vertex_size);
-
-   for (i = 0 ; i < exec->vtx.prim_count ; i++) {
-      struct brw_draw_prim *prim = &exec->vtx.prim[i];
-      _mesa_printf("   prim %d: %s%s %d..%d %s %s\n",
-		   i, 
-		   _mesa_lookup_enum_by_nr(prim->mode),
-		   prim->weak ? " (weak)" : "",
-		   prim->start, 
-		   prim->start + prim->count,
-		   prim->begin ? "BEGIN" : "(wrap)",
-		   prim->end ? "END" : "(wrap)");
-   }
-}
-
-
-/*
- * NOTE: Need to have calculated primitives by this point -- do it on the fly.
- * NOTE: Old 'parity' issue is gone.
- */
-static GLuint brw_copy_vertices( struct brw_exec_context *exec )
-{
-   GLuint nr = exec->vtx.prim[exec->vtx.prim_count-1].count;
-   GLuint ovf, i;
-   GLuint sz = exec->vtx.vertex_size;
-   GLfloat *dst = exec->vtx.copied.buffer;
-   GLfloat *src = ((GLfloat *)exec->vtx.buffer_map + 
-		   exec->vtx.prim[exec->vtx.prim_count-1].start * 
-		   exec->vtx.vertex_size);
-
-
-   switch( exec->ctx->Driver.CurrentExecPrimitive )
-   {
-   case GL_POINTS:
-      return 0;
-   case GL_LINES:
-      ovf = nr&1;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz * sizeof(GLfloat) );
-      return i;
-   case GL_TRIANGLES:
-      ovf = nr%3;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz * sizeof(GLfloat) );
-      return i;
-   case GL_QUADS:
-      ovf = nr&3;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz * sizeof(GLfloat) );
-      return i;
-   case GL_LINE_STRIP:
-      if (nr == 0) 
-	 return 0;
-      else {
-	 _mesa_memcpy( dst, src+(nr-1)*sz, sz * sizeof(GLfloat) );
-	 return 1;
-      }
-   case GL_LINE_LOOP:
-   case GL_TRIANGLE_FAN:
-   case GL_POLYGON:
-      if (nr == 0) 
-	 return 0;
-      else if (nr == 1) {
-	 _mesa_memcpy( dst, src+0, sz * sizeof(GLfloat) );
-	 return 1;
-      } else {
-	 _mesa_memcpy( dst, src+0, sz * sizeof(GLfloat) );
-	 _mesa_memcpy( dst+sz, src+(nr-1)*sz, sz * sizeof(GLfloat) );
-	 return 2;
-      }
-   case GL_TRIANGLE_STRIP:
-   case GL_QUAD_STRIP:
-      switch (nr) {
-      case 0: ovf = 0; break;
-      case 1: ovf = 1; break;
-      default: ovf = 2 + (nr&1); break;
-      }
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz * sizeof(GLfloat) );
-      return i;
-   case GL_POLYGON+1:
-      return 0;
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-
-/* TODO: populate these as the vertex is defined:
- */
-static void brw_exec_bind_arrays( struct brw_exec_context *exec )
-{
-   struct gl_client_array *arrays = exec->vtx.arrays;
-   GLuint count = exec->vtx.vert_count;
-   GLubyte *data = exec->vtx.buffer_map;
-   GLuint attr;
-
-   memset(arrays, 0, BRW_ATTRIB_MAX * sizeof(arrays[0]));
-
-   /* Make all active attributes (including edgeflag) available as
-    * arrays of floats.
-    */
-   for (attr = 0; attr < BRW_ATTRIB_MAX ; attr++) {
-      if (exec->vtx.attrsz[attr]) {
-	 arrays[attr].Ptr = (void *)data;
-	 arrays[attr].Size = exec->vtx.attrsz[attr];
-	 arrays[attr].StrideB = exec->vtx.vertex_size * sizeof(GLfloat);
-	 arrays[attr].Stride = exec->vtx.vertex_size * sizeof(GLfloat);
-	 arrays[attr].Type = GL_FLOAT;
-	 arrays[attr].Enabled = 1;
-	 arrays[attr].BufferObj = exec->vtx.bufferobj; /* NullBufferObj */
-	 arrays[attr]._MaxElement = count; /* ??? */
-
-	 data += exec->vtx.attrsz[attr] * sizeof(GLfloat);
-      }
-   }
-}
-
-
-static
-void brw_exec_loopback_vertex_list( GLcontext *ctx,
-				    struct brw_exec_context *exec )
-{
-   const GLfloat *buffer = (const GLfloat *)exec->vtx.buffer_map;
-
-   brw_fallback(ctx);
-
-   brw_loopback_vertex_list( ctx,
-			     buffer,
-			     exec->vtx.attrsz,
-			     exec->vtx.prim,
-			     exec->vtx.prim_count,
-			     0,	/* XXX - copied.nr */
-			     exec->vtx.vertex_size);
-
-
-   brw_unfallback(ctx);
-}
-
-
-/**
- * Execute the buffer and save copied verts.
- */
-void brw_exec_vtx_flush( struct brw_exec_context *exec )
-{
-   if (0)
-      brw_exec_debug_verts( exec );
-
-
-   if (exec->vtx.prim_count && 
-       exec->vtx.vert_count) {
-
-      exec->vtx.copied.nr = brw_copy_vertices( exec ); 
-
-      if (exec->vtx.copied.nr != exec->vtx.vert_count) {
-	 GLcontext *ctx = exec->ctx;
-
-	 brw_exec_bind_arrays( exec );
-
-
-	 if (!brw_draw_prims( ctx, 
-			      exec->vtx.inputs, 
-			      exec->vtx.prim, 
-			      exec->vtx.prim_count,
-			      NULL,
-			      0,
-			      exec->vtx.vert_count,
-			      0 )) {
-	    /* Fallback path: 
-	     */
-	    brw_exec_loopback_vertex_list(ctx, exec);
-	 }
-      }
-   }
-
-   exec->vtx.prim_count = 0;
-   exec->vtx.vert_count = 0;
-   exec->vtx.vbptr = (GLfloat *)exec->vtx.buffer_map;
-}
diff --git a/src/mesa/drivers/dri/i965/brw_exec_eval.c b/src/mesa/drivers/dri/i965/brw_exec_eval.c
deleted file mode 100644
index 9dbeb1b58e4..00000000000
--- a/src/mesa/drivers/dri/i965/brw_exec_eval.c
+++ /dev/null
@@ -1,255 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.1
- *
- * Copyright (C) 1999-2004  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *    Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "api_eval.h"
-#include "context.h"
-#include "macros.h"
-#include "math/m_eval.h"
-#include "brw_exec.h"
-#include "dispatch.h"
-
-
-static void clear_active_eval1( struct brw_exec_context *exec, GLuint attr ) 
-{
-   exec->eval.map1[attr].map = NULL;
-}
-
-static void clear_active_eval2( struct brw_exec_context *exec, GLuint attr ) 
-{
-   exec->eval.map2[attr].map = NULL;
-}
-
-static void set_active_eval1( struct brw_exec_context *exec, GLuint attr, GLuint dim, 
-			      struct gl_1d_map *map )
-{
-   if (!exec->eval.map1[attr].map) {
-      exec->eval.map1[attr].map = map;
-      exec->eval.map1[attr].sz = dim;
-   }
-} 
-
-static void set_active_eval2( struct brw_exec_context *exec, GLuint attr, GLuint dim, 
-			      struct gl_2d_map *map )
-{
-   if (!exec->eval.map2[attr].map) {
-      exec->eval.map2[attr].map = map;
-      exec->eval.map2[attr].sz = dim;
-   }
-} 
-
-void brw_exec_eval_update( struct brw_exec_context *exec )
-{
-   GLcontext *ctx = exec->ctx;
-   GLuint attr;
-
-   /* Vertex program maps have priority over conventional attribs */
-
-   for (attr = 0; attr < BRW_ATTRIB_FIRST_MATERIAL; attr++) {
-      clear_active_eval1( exec, attr );
-      clear_active_eval2( exec, attr );
-   }
-
-   /* _NEW_PROGRAM */
-   if (ctx->VertexProgram._Enabled) {
-      for (attr = 0; attr < BRW_ATTRIB_FIRST_MATERIAL; attr++) {
-	 /* _NEW_EVAL */
-	 if (ctx->Eval.Map1Attrib[attr]) 
-	    set_active_eval1( exec, attr, 4, &ctx->EvalMap.Map1Attrib[attr] );
-
-	 if (ctx->Eval.Map2Attrib[attr]) 
-	    set_active_eval2( exec, attr, 4, &ctx->EvalMap.Map2Attrib[attr] );
-      }
-   }
-
-   if (ctx->Eval.Map1Color4) 
-      set_active_eval1( exec, BRW_ATTRIB_COLOR0, 4, &ctx->EvalMap.Map1Color4 );
-      
-   if (ctx->Eval.Map2Color4) 
-      set_active_eval2( exec, BRW_ATTRIB_COLOR0, 4, &ctx->EvalMap.Map2Color4 );
-
-   if (ctx->Eval.Map1TextureCoord4) 
-      set_active_eval1( exec, BRW_ATTRIB_TEX0, 4, &ctx->EvalMap.Map1Texture4 );
-   else if (ctx->Eval.Map1TextureCoord3) 
-      set_active_eval1( exec, BRW_ATTRIB_TEX0, 3, &ctx->EvalMap.Map1Texture3 );
-   else if (ctx->Eval.Map1TextureCoord2) 
-      set_active_eval1( exec, BRW_ATTRIB_TEX0, 2, &ctx->EvalMap.Map1Texture2 );
-   else if (ctx->Eval.Map1TextureCoord1) 
-      set_active_eval1( exec, BRW_ATTRIB_TEX0, 1, &ctx->EvalMap.Map1Texture1 );
-
-   if (ctx->Eval.Map2TextureCoord4) 
-      set_active_eval2( exec, BRW_ATTRIB_TEX0, 4, &ctx->EvalMap.Map2Texture4 );
-   else if (ctx->Eval.Map2TextureCoord3) 
-      set_active_eval2( exec, BRW_ATTRIB_TEX0, 3, &ctx->EvalMap.Map2Texture3 );
-   else if (ctx->Eval.Map2TextureCoord2) 
-      set_active_eval2( exec, BRW_ATTRIB_TEX0, 2, &ctx->EvalMap.Map2Texture2 );
-   else if (ctx->Eval.Map2TextureCoord1) 
-      set_active_eval2( exec, BRW_ATTRIB_TEX0, 1, &ctx->EvalMap.Map2Texture1 );
-
-   if (ctx->Eval.Map1Normal) 
-      set_active_eval1( exec, BRW_ATTRIB_NORMAL, 3, &ctx->EvalMap.Map1Normal );
-
-   if (ctx->Eval.Map2Normal) 
-      set_active_eval2( exec, BRW_ATTRIB_NORMAL, 3, &ctx->EvalMap.Map2Normal );
-
-   if (ctx->Eval.Map1Vertex4) 
-      set_active_eval1( exec, BRW_ATTRIB_POS, 4, &ctx->EvalMap.Map1Vertex4 );
-   else if (ctx->Eval.Map1Vertex3) 
-      set_active_eval1( exec, BRW_ATTRIB_POS, 3, &ctx->EvalMap.Map1Vertex3 );
-
-   if (ctx->Eval.Map2Vertex4) 
-      set_active_eval2( exec, BRW_ATTRIB_POS, 4, &ctx->EvalMap.Map2Vertex4 );
-   else if (ctx->Eval.Map2Vertex3) 
-      set_active_eval2( exec, BRW_ATTRIB_POS, 3, &ctx->EvalMap.Map2Vertex3 );
-
-   exec->eval.recalculate_maps = 0;
-}
-
-
-
-void brw_exec_do_EvalCoord1f(struct brw_exec_context *exec, GLfloat u)
-{
-   GLuint attr;
-
-   for (attr = 1; attr <= BRW_ATTRIB_INDEX; attr++) {
-      struct gl_1d_map *map = exec->eval.map1[attr].map;
-      if (map) {
-	 GLfloat uu = (u - map->u1) * map->du;
-	 GLfloat data[4];
-
-	 ASSIGN_4V(data, 0, 0, 0, 1);
-
-	 _math_horner_bezier_curve(map->Points, data, uu, 
-				   exec->eval.map1[attr].sz, 
-				   map->Order);
-
-	 COPY_SZ_4V( exec->vtx.attrptr[attr],
-		     exec->vtx.attrsz[attr],
-		     data );
-      }
-   }
-
-   /** Vertex -- EvalCoord1f is a noop if this map not enabled:
-    **/
-   if (exec->eval.map1[0].map) {
-      struct gl_1d_map *map = exec->eval.map1[0].map;
-      GLfloat uu = (u - map->u1) * map->du;
-      GLfloat vertex[4];
-
-      ASSIGN_4V(vertex, 0, 0, 0, 1);
-
-      _math_horner_bezier_curve(map->Points, vertex, uu, 
-				exec->eval.map1[0].sz, 
-				map->Order);
-
-      if (exec->eval.map1[0].sz == 4) 
-	 CALL_Vertex4fv(GET_DISPATCH(), ( vertex ));
-      else
-	 CALL_Vertex3fv(GET_DISPATCH(), ( vertex )); 
-   }
-}
-
-
-
-void brw_exec_do_EvalCoord2f( struct brw_exec_context *exec, 
-			      GLfloat u, GLfloat v )
-{   
-   GLuint attr;
-
-   for (attr = 1; attr <= BRW_ATTRIB_INDEX; attr++) {
-      struct gl_2d_map *map = exec->eval.map2[attr].map;
-      if (map) {
-	 GLfloat uu = (u - map->u1) * map->du;
-	 GLfloat vv = (v - map->v1) * map->dv;
-	 GLfloat data[4];
-
-	 ASSIGN_4V(data, 0, 0, 0, 1);
-
-	 _math_horner_bezier_surf(map->Points, 
-				  data, 
-				  uu, vv, 
-				  exec->eval.map2[attr].sz, 
-				  map->Uorder, map->Vorder);
-
-	 COPY_SZ_4V( exec->vtx.attrptr[attr],
-		     exec->vtx.attrsz[attr],
-		     data );
-      }
-   }
-
-   /** Vertex -- EvalCoord2f is a noop if this map not enabled:
-    **/
-   if (exec->eval.map2[0].map) {
-      struct gl_2d_map *map = exec->eval.map2[0].map;
-      GLfloat uu = (u - map->u1) * map->du;
-      GLfloat vv = (v - map->v1) * map->dv;
-      GLfloat vertex[4];
-
-      ASSIGN_4V(vertex, 0, 0, 0, 1);
-
-      if (exec->ctx->Eval.AutoNormal) {
-	 GLfloat normal[4];
-         GLfloat du[4], dv[4];
-
-         _math_de_casteljau_surf(map->Points, vertex, du, dv, uu, vv, 
-				 exec->eval.map2[0].sz,
-				 map->Uorder, map->Vorder);
-
-	 if (exec->eval.map2[0].sz == 4) {
-	    du[0] = du[0]*vertex[3] - du[3]*vertex[0];
-	    du[1] = du[1]*vertex[3] - du[3]*vertex[1];
-	    du[2] = du[2]*vertex[3] - du[3]*vertex[2];
-	 
-	    dv[0] = dv[0]*vertex[3] - dv[3]*vertex[0];
-	    dv[1] = dv[1]*vertex[3] - dv[3]*vertex[1];
-	    dv[2] = dv[2]*vertex[3] - dv[3]*vertex[2];
-	 }
-
-
-         CROSS3(normal, du, dv);
-         NORMALIZE_3FV(normal);
-	 normal[3] = 1.0;
-
-	 COPY_SZ_4V( exec->vtx.attrptr[BRW_ATTRIB_NORMAL],
-		     exec->vtx.attrsz[BRW_ATTRIB_NORMAL],
-		     normal );
-
-      }
-      else {
-         _math_horner_bezier_surf(map->Points, vertex, uu, vv, 
-				  exec->eval.map2[0].sz,
-				  map->Uorder, map->Vorder);
-      }
-
-      if (exec->vtx.attrsz[0] == 4) 
-	 CALL_Vertex4fv(GET_DISPATCH(), ( vertex ));
-      else
-	 CALL_Vertex3fv(GET_DISPATCH(), ( vertex )); 
-   }
-}
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.c b/src/mesa/drivers/dri/i965/brw_fallback.c
index 56e21e339ef..86464b2ec5f 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.c
+++ b/src/mesa/drivers/dri/i965/brw_fallback.c
@@ -30,8 +30,6 @@
 #include "tnl/tnl.h"
 #include "context.h"
 #include "brw_context.h"
-#include "brw_exec.h"
-#include "brw_save.h"
 #include "brw_fallback.h"
 
 #include "glheader.h"
@@ -40,297 +38,6 @@
 #include "imports.h"
 #include "macros.h"
 #include "mtypes.h"
-#include "dispatch.h"
-
-
-typedef void (*attr_func)( GLcontext *ctx, GLint target, const GLfloat * );
-
-
-/* Wrapper functions in case glVertexAttrib*fvNV doesn't exist */
-static void VertexAttrib1fvNV(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib1fvNV(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib2fvNV(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib2fvNV(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib3fvNV(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib3fvNV(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib4fvNV(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib4fvNV(ctx->Exec, (target, v));
-}
-
-static attr_func vert_attrfunc[4] = {
-   VertexAttrib1fvNV,
-   VertexAttrib2fvNV,
-   VertexAttrib3fvNV,
-   VertexAttrib4fvNV
-};
-
-#if 0
-static void VertexAttrib1fvARB(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib1fvARB(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib2fvARB(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib2fvARB(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib3fvARB(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib3fvARB(ctx->Exec, (target, v));
-}
-
-static void VertexAttrib4fvARB(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   CALL_VertexAttrib4fvARB(ctx->Exec, (target, v));
-}
-
-
-static attr_func vert_attrfunc_arb[4] = {
-   VertexAttrib1fvARB,
-   VertexAttrib2fvARB,
-   VertexAttrib3fvARB,
-   VertexAttrib4fvARB
-};
-#endif
-
-
-
-
-
-
-static void mat_attr1fv( GLcontext *ctx, GLint target, const GLfloat *v )
-{
-   switch (target) {
-   case BRW_ATTRIB_MAT_FRONT_SHININESS:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_SHININESS, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_SHININESS:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_SHININESS, v ));
-      break;
-   }
-}
-
-
-static void mat_attr3fv( GLcontext *ctx, GLint target, const GLfloat *v )
-{
-   switch (target) {
-   case BRW_ATTRIB_MAT_FRONT_INDEXES:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_COLOR_INDEXES, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_INDEXES:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_COLOR_INDEXES, v ));
-      break;
-   }
-}
-
-
-static void mat_attr4fv( GLcontext *ctx, GLint target, const GLfloat *v )
-{
-   switch (target) {
-   case BRW_ATTRIB_MAT_FRONT_EMISSION:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_EMISSION, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_EMISSION:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_EMISSION, v ));
-      break;
-   case BRW_ATTRIB_MAT_FRONT_AMBIENT:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_AMBIENT, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_AMBIENT:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_AMBIENT, v ));
-      break;
-   case BRW_ATTRIB_MAT_FRONT_DIFFUSE:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_DIFFUSE, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_DIFFUSE:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_DIFFUSE, v ));
-      break;
-   case BRW_ATTRIB_MAT_FRONT_SPECULAR:
-      CALL_Materialfv(ctx->Exec, ( GL_FRONT, GL_SPECULAR, v ));
-      break;
-   case BRW_ATTRIB_MAT_BACK_SPECULAR:
-      CALL_Materialfv(ctx->Exec, ( GL_BACK, GL_SPECULAR, v ));
-      break;
-   }
-}
-
-
-static attr_func mat_attrfunc[4] = {
-   mat_attr1fv,
-   NULL,
-   mat_attr3fv,
-   mat_attr4fv
-};
-
-
-static void index_attr1fv(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   (void) target;
-   CALL_Indexf(ctx->Exec, (v[0]));
-}
-
-static void edgeflag_attr1fv(GLcontext *ctx, GLint target, const GLfloat *v)
-{
-   (void) target;
-   CALL_EdgeFlag(ctx->Exec, ((GLboolean)(v[0] == 1.0)));
-}
-
-struct loopback_attr {
-   GLint target;
-   GLint sz;
-   attr_func func;
-};
-
-/* Don't emit ends and begins on wrapped primitives.  Don't replay
- * wrapped vertices.  If we get here, it's probably because the the
- * precalculated wrapping is wrong.
- */
-static void loopback_prim( GLcontext *ctx,
-			   const GLfloat *buffer,
-			   const struct brw_draw_prim *prim,
-			   GLuint wrap_count,
-			   GLuint vertex_size,
-			   const struct loopback_attr *la, GLuint nr )
-{
-   GLint start = prim->start;
-   GLint end = start + prim->count;
-   const GLfloat *data;
-   GLint j;
-   GLuint k;
-
-   if (0)
-      _mesa_printf("loopback prim %s(%s,%s) verts %d..%d\n",
-		   _mesa_lookup_enum_by_nr(prim->mode),
-		   prim->begin ? "begin" : "..",
-		   prim->end ? "end" : "..",
-		   start, 
-		   end);
-
-   if (prim->begin) {
-      CALL_Begin(GET_DISPATCH(), ( prim->mode ));
-   }
-   else {
-      assert(start == 0);
-      start += wrap_count;
-   }
-
-   data = buffer + start * vertex_size;
-
-   for (j = start ; j < end ; j++) {
-      const GLfloat *tmp = data + la[0].sz;
-
-      for (k = 1 ; k < nr ; k++) {
-	 la[k].func( ctx, la[k].target, tmp );
-	 tmp += la[k].sz;
-      }
-	 
-      /* Fire the vertex
-       */
-      la[0].func( ctx, BRW_ATTRIB_POS, data );
-      data = tmp;
-   }
-
-   if (prim->end) {
-      CALL_End(GET_DISPATCH(), ());
-   }
-}
-
-/* Primitives generated by DrawArrays/DrawElements/Rectf may be
- * caught here.  If there is no primitive in progress, execute them
- * normally, otherwise need to track and discard the generated
- * primitives.
- */
-static void loopback_weak_prim( GLcontext *ctx,
-				const struct brw_draw_prim *prim )
-{
-   /* Use the prim_weak flag to ensure that if this primitive
-    * wraps, we don't mistake future vertex_lists for part of the
-    * surrounding primitive.
-    *
-    * While this flag is set, we are simply disposing of data
-    * generated by an operation now known to be a noop.
-    */
-   if (prim->begin)
-      ctx->Driver.CurrentExecPrimitive |= BRW_SAVE_PRIM_WEAK;
-   if (prim->end)
-      ctx->Driver.CurrentExecPrimitive &= ~BRW_SAVE_PRIM_WEAK;
-}
-
-
-void brw_loopback_vertex_list( GLcontext *ctx,
-			       const GLfloat *buffer,
-			       const GLubyte *attrsz,
-			       const struct brw_draw_prim *prim,
-			       GLuint prim_count,
-			       GLuint wrap_count,
-			       GLuint vertex_size)
-{
-   struct loopback_attr la[BRW_ATTRIB_MAX];
-   GLuint i, nr = 0;
-
-   for (i = 0 ; i <= BRW_ATTRIB_TEX7 ; i++) {
-      if (i == BRW_ATTRIB_INDEX || i == BRW_ATTRIB_EDGEFLAG)
-                   continue;
-
-      if (attrsz[i]) {
-	 la[nr].target = i;
-	 la[nr].sz = attrsz[i];
-	 la[nr].func = vert_attrfunc[attrsz[i]-1];
-	 nr++;
-      }
-   }
-
-   for (i = BRW_ATTRIB_MAT_FRONT_AMBIENT ; 
-	i <= BRW_ATTRIB_MAT_BACK_INDEXES ; 
-	i++) {
-      if (attrsz[i]) {
-	 la[nr].target = i;
-	 la[nr].sz = attrsz[i];
-	 la[nr].func = mat_attrfunc[attrsz[i]-1];
-	 nr++;
-      }
-   }
-
-   if (attrsz[BRW_ATTRIB_EDGEFLAG]) {
-      la[nr].target = BRW_ATTRIB_EDGEFLAG;
-      la[nr].sz = attrsz[BRW_ATTRIB_EDGEFLAG];
-      la[nr].func = edgeflag_attr1fv;
-      nr++;
-   }
-
-   if (attrsz[BRW_ATTRIB_INDEX]) {
-      la[nr].target = BRW_ATTRIB_INDEX;
-      la[nr].sz = attrsz[BRW_ATTRIB_INDEX];
-      la[nr].func = index_attr1fv;
-      nr++;
-   }
-
-   /* XXX ARB vertex attribs */
-
-   for (i = 0 ; i < prim_count ; i++) {
-      if ((prim[i].mode & BRW_SAVE_PRIM_WEAK) &&
-	  (ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END))
-      {
-	 loopback_weak_prim( ctx, &prim[i] );
-      }
-      else
-      {
-	 loopback_prim( ctx, buffer, &prim[i], wrap_count, vertex_size, la, nr );
-      }
-   }
-}
-
 
 
 
@@ -405,62 +112,6 @@ const struct brw_tracked_state brw_check_fallback = {
 
 
 
-/* If there is a fallback, fallback to software rasterization and
- * transformation together.  There is never a requirement to have
- * software t&l but hardware rasterization.
- * 
- * Further, all fallbacks are based on GL state, not on eg. primitive
- * or vertex data.
- */
-
-static void do_fallback( struct brw_context *brw,
-			 GLboolean fallback )
-{
-   GLcontext *ctx = &brw->intel.ctx;
-
-   /* flush:
-    */
-   ctx->Driver.Flush( ctx );
-
-   if (fallback) {
-      _swsetup_Wakeup( ctx );
-      _tnl_wakeup_exec( ctx );	
-
-      /* Need this because tnl_wakeup_exec does too much: 
-       */
-      brw_save_wakeup(ctx);
-      brw_save_fallback(ctx, GL_TRUE);
-   }
-   else {
-      /* Flush vertices and copy-to-current:
-       */
-      FLUSH_CURRENT(ctx, 0); 
-
-      _swrast_flush( ctx );
-
-      brw_exec_wakeup(ctx);
-
-      /* Need this because tnl_wakeup_exec does too much: 
-       */
-      brw_save_wakeup(ctx);
-      brw_save_fallback(ctx, GL_FALSE);	 
-   }
-}
-
-
-void brw_fallback( GLcontext *ctx )
-{
-   struct brw_context *brw = brw_context(ctx);
-   do_fallback(brw, 1);
-}
-
-
-void brw_unfallback( GLcontext *ctx )
-{
-   struct brw_context *brw = brw_context(ctx);
-   do_fallback(brw, 0);
-}
-
 /* Not used:
  */
 void intelFallback( struct intel_context *intel, GLuint bit, GLboolean mode )
diff --git a/src/mesa/drivers/dri/i965/brw_fallback.h b/src/mesa/drivers/dri/i965/brw_fallback.h
index 81a2d344b81..684a46cd170 100644
--- a/src/mesa/drivers/dri/i965/brw_fallback.h
+++ b/src/mesa/drivers/dri/i965/brw_fallback.h
@@ -31,7 +31,7 @@
 #include "mtypes.h"		/* for GLcontext... */
 
 struct brw_context;
-struct brw_draw_prim;
+struct vbo_prim;
 
 void brw_fallback( GLcontext *ctx );
 void brw_unfallback( GLcontext *ctx );
@@ -39,7 +39,7 @@ void brw_unfallback( GLcontext *ctx );
 void brw_loopback_vertex_list( GLcontext *ctx,
 			       const GLfloat *buffer,
 			       const GLubyte *attrsz,
-			       const struct brw_draw_prim *prim,
+			       const struct vbo_prim *prim,
 			       GLuint prim_count,
 			       GLuint wrap_count,
 			       GLuint vertex_size);
diff --git a/src/mesa/drivers/dri/i965/brw_gs.c b/src/mesa/drivers/dri/i965/brw_gs.c
index 9066e42252d..73263a5fff4 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.c
+++ b/src/mesa/drivers/dri/i965/brw_gs.c
@@ -82,6 +82,9 @@ static void compile_gs_prog( struct brw_context *brw,
    case GL_QUADS:
       brw_gs_quads( &c ); 
       break;
+   case GL_QUAD_STRIP:
+      brw_gs_quad_strip( &c );
+      break;
    case GL_LINE_LOOP:
       brw_gs_lines( &c );
       break;
@@ -145,7 +148,7 @@ static const GLenum gs_prim[GL_POLYGON+1] = {
    GL_TRIANGLES,
    GL_TRIANGLES,
    GL_QUADS,
-   GL_QUADS,
+   GL_QUAD_STRIP,
    GL_TRIANGLES
 };
 
diff --git a/src/mesa/drivers/dri/i965/brw_gs.h b/src/mesa/drivers/dri/i965/brw_gs.h
index f9aa71d9199..29a4e80ce1b 100644
--- a/src/mesa/drivers/dri/i965/brw_gs.h
+++ b/src/mesa/drivers/dri/i965/brw_gs.h
@@ -67,6 +67,7 @@ struct brw_gs_compile {
 #define ATTR_SIZE  (4*4)
 
 void brw_gs_quads( struct brw_gs_compile *c );
+void brw_gs_quad_strip( struct brw_gs_compile *c );
 void brw_gs_tris( struct brw_gs_compile *c );
 void brw_gs_lines( struct brw_gs_compile *c );
 void brw_gs_points( struct brw_gs_compile *c );
diff --git a/src/mesa/drivers/dri/i965/brw_gs_emit.c b/src/mesa/drivers/dri/i965/brw_gs_emit.c
index e4eed36a464..9abb94d82ed 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_emit.c
@@ -116,6 +116,16 @@ void brw_gs_quads( struct brw_gs_compile *c )
    brw_gs_emit_vue(c, c->reg.vertex[2], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
 }
 
+void brw_gs_quad_strip( struct brw_gs_compile *c )
+{
+   brw_gs_alloc_regs(c, 4);
+   
+   brw_gs_emit_vue(c, c->reg.vertex[2], 0, ((_3DPRIM_POLYGON << 2) | R02_PRIM_START));
+   brw_gs_emit_vue(c, c->reg.vertex[3], 0, (_3DPRIM_POLYGON << 2));
+   brw_gs_emit_vue(c, c->reg.vertex[0], 0, (_3DPRIM_POLYGON << 2)); 
+   brw_gs_emit_vue(c, c->reg.vertex[1], 1, ((_3DPRIM_POLYGON << 2) | R02_PRIM_END));
+}
+
 void brw_gs_tris( struct brw_gs_compile *c )
 {
    brw_gs_alloc_regs(c, 3);
diff --git a/src/mesa/drivers/dri/i965/brw_metaops.c b/src/mesa/drivers/dri/i965/brw_metaops.c
index acd1d581917..1579762b6d1 100644
--- a/src/mesa/drivers/dri/i965/brw_metaops.c
+++ b/src/mesa/drivers/dri/i965/brw_metaops.c
@@ -47,7 +47,6 @@
 #include "brw_context.h"
 #include "brw_defines.h"
 #include "brw_draw.h"
-#include "brw_attrib.h"
 #include "brw_fallback.h"
 
 #define INIT(brw, STRUCT, ATTRIB) 		\
@@ -388,8 +387,8 @@ static void meta_draw_quad(struct intel_context *intel,
    struct brw_context *brw = brw_context(&intel->ctx);
    struct gl_client_array pos_array;
    struct gl_client_array color_array;
-   struct gl_client_array *attribs[BRW_ATTRIB_MAX];
-   struct brw_draw_prim prim[1];
+   struct gl_client_array *attribs[VERT_ATTRIB_MAX];
+   struct _mesa_prim prim[1];
    GLfloat pos[4][3];
    GLubyte color[4];
 
@@ -439,29 +438,29 @@ static void meta_draw_quad(struct intel_context *intel,
    /* Ignoring texture coords. 
     */
 
-   memset(attribs, 0, BRW_ATTRIB_MAX * sizeof(*attribs));
-
-   attribs[BRW_ATTRIB_POS] = &pos_array;
-   attribs[BRW_ATTRIB_POS]->Ptr = 0;
-   attribs[BRW_ATTRIB_POS]->Type = GL_FLOAT;
-   attribs[BRW_ATTRIB_POS]->Enabled = 1;
-   attribs[BRW_ATTRIB_POS]->Size = 3;
-   attribs[BRW_ATTRIB_POS]->StrideB = 3 * sizeof(GLfloat);
-   attribs[BRW_ATTRIB_POS]->Stride = 3 * sizeof(GLfloat);
-   attribs[BRW_ATTRIB_POS]->_MaxElement = 4;
-   attribs[BRW_ATTRIB_POS]->Normalized = 0;
-   attribs[BRW_ATTRIB_POS]->BufferObj = brw->metaops.vbo;
-
-   attribs[BRW_ATTRIB_COLOR0] = &color_array;
-   attribs[BRW_ATTRIB_COLOR0]->Ptr = (const GLubyte *)sizeof(pos);
-   attribs[BRW_ATTRIB_COLOR0]->Type = GL_UNSIGNED_BYTE;
-   attribs[BRW_ATTRIB_COLOR0]->Enabled = 1;
-   attribs[BRW_ATTRIB_COLOR0]->Size = 4;
-   attribs[BRW_ATTRIB_COLOR0]->StrideB = 0;
-   attribs[BRW_ATTRIB_COLOR0]->Stride = 0;
-   attribs[BRW_ATTRIB_COLOR0]->_MaxElement = 1;
-   attribs[BRW_ATTRIB_COLOR0]->Normalized = 1;
-   attribs[BRW_ATTRIB_COLOR0]->BufferObj = brw->metaops.vbo;
+   memset(attribs, 0, VERT_ATTRIB_MAX * sizeof(*attribs));
+
+   attribs[VERT_ATTRIB_POS] = &pos_array;
+   attribs[VERT_ATTRIB_POS]->Ptr = 0;
+   attribs[VERT_ATTRIB_POS]->Type = GL_FLOAT;
+   attribs[VERT_ATTRIB_POS]->Enabled = 1;
+   attribs[VERT_ATTRIB_POS]->Size = 3;
+   attribs[VERT_ATTRIB_POS]->StrideB = 3 * sizeof(GLfloat);
+   attribs[VERT_ATTRIB_POS]->Stride = 3 * sizeof(GLfloat);
+   attribs[VERT_ATTRIB_POS]->_MaxElement = 4;
+   attribs[VERT_ATTRIB_POS]->Normalized = 0;
+   attribs[VERT_ATTRIB_POS]->BufferObj = brw->metaops.vbo;
+
+   attribs[VERT_ATTRIB_COLOR0] = &color_array;
+   attribs[VERT_ATTRIB_COLOR0]->Ptr = (const GLubyte *)sizeof(pos);
+   attribs[VERT_ATTRIB_COLOR0]->Type = GL_UNSIGNED_BYTE;
+   attribs[VERT_ATTRIB_COLOR0]->Enabled = 1;
+   attribs[VERT_ATTRIB_COLOR0]->Size = 4;
+   attribs[VERT_ATTRIB_COLOR0]->StrideB = 0;
+   attribs[VERT_ATTRIB_COLOR0]->Stride = 0;
+   attribs[VERT_ATTRIB_COLOR0]->_MaxElement = 1;
+   attribs[VERT_ATTRIB_COLOR0]->Normalized = 1;
+   attribs[VERT_ATTRIB_COLOR0]->BufferObj = brw->metaops.vbo;
    
    /* Just ignoring texture coordinates for now. 
     */
@@ -476,19 +475,12 @@ static void meta_draw_quad(struct intel_context *intel,
    prim[0].start = 0;
    prim[0].count = 4;
 
-   if (!brw_draw_prims(&brw->intel.ctx, 
-		       (const struct gl_client_array **)attribs,
-		       prim, 1,
-		       NULL,
-		       0,
-		       4,
-		       BRW_DRAW_LOCKED ))
-   {
-      /* This should not be possible:
-       */
-      _mesa_printf("brw_draw_prims failed in metaops!\n");
-      assert(0);
-   }
+   brw_draw_prims(&brw->intel.ctx, 
+		  (const struct gl_client_array **)attribs,
+		  prim, 1,
+		  NULL,
+		  0,
+		  3 );
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_save.c b/src/mesa/drivers/dri/i965/brw_save.c
deleted file mode 100644
index 1af7791c4d4..00000000000
--- a/src/mesa/drivers/dri/i965/brw_save.c
+++ /dev/null
@@ -1,126 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.3
- *
- * Copyright (C) 1999-2005  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- * Authors:
- *    Keith Whitwell <[email protected]>
- */
-
-
-#include "mtypes.h"
-#include "api_arrayelt.h"
-#include "dlist.h"
-#include "vtxfmt.h"
-#include "imports.h"
-
-#include "brw_save.h"
-
-
-
-void brw_save_init( GLcontext *ctx )
-{
-   struct brw_save_context *save = CALLOC_STRUCT(brw_save_context);
-
-   if (ctx->swtnl_im == NULL) {
-      ctx->swtnl_im = CALLOC_STRUCT(brw_exec_save);
-   }
-
-   save->ctx = ctx;
-   IMM_CONTEXT(ctx)->save = save;
-
-   /* Initialize the arrayelt helper
-    */
-   if (!ctx->aelt_context &&
-       !_ae_create_context( ctx )) 
-      return;
-
-   brw_save_api_init( save );
-   brw_save_wakeup(ctx);
-
-   ctx->Driver.CurrentSavePrimitive = PRIM_UNKNOWN;
-}
-
-
-void brw_save_destroy( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   if (save) {
-      FREE(save);
-      IMM_CONTEXT(ctx)->save = NULL;
-   }
-
-   if (ctx->aelt_context) {
-      _ae_destroy_context( ctx );
-      ctx->aelt_context = NULL;
-   }
-
-   if (IMM_CONTEXT(ctx)->exec == NULL &&
-       IMM_CONTEXT(ctx)->save == NULL) {
-      FREE(IMM_CONTEXT(ctx));
-      ctx->swtnl_im = NULL;
-   }
-}
-
-
-void brw_save_invalidate_state( GLcontext *ctx, GLuint new_state )
-{
-   _ae_invalidate_state(ctx, new_state);
-}
-
-
-/* Note that this can occur during the playback of a display list:
- */
-void brw_save_fallback( GLcontext *ctx, GLboolean fallback )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-
-   if (fallback)
-      save->replay_flags |= BRW_SAVE_FALLBACK;
-   else
-      save->replay_flags &= ~BRW_SAVE_FALLBACK;
-}
-
-
-/* I don't see any reason to swap this code out on fallbacks.  It
- * wouldn't really mean anything to do so anyway as the old lists are
- * still around from pre-fallback.  Instead, the above code ensures
- * that vertices are routed back through immediate mode dispatch on
- * fallback.
- *
- * The below can be moved into init or removed:
- */
-void brw_save_wakeup( GLcontext *ctx )
-{
-   ctx->Driver.NewList = brw_save_NewList;
-   ctx->Driver.EndList = brw_save_EndList;
-   ctx->Driver.SaveFlushVertices = brw_save_SaveFlushVertices;
-   ctx->Driver.BeginCallList = brw_save_BeginCallList;
-   ctx->Driver.EndCallList = brw_save_EndCallList;
-   ctx->Driver.NotifySaveBegin = brw_save_NotifyBegin;
-
-   /* Assume we haven't been getting state updates either:
-    */
-   brw_save_invalidate_state( ctx, ~0 );
-}
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_save.h b/src/mesa/drivers/dri/i965/brw_save.h
deleted file mode 100644
index 41cabe65084..00000000000
--- a/src/mesa/drivers/dri/i965/brw_save.h
+++ /dev/null
@@ -1,171 +0,0 @@
-/**************************************************************************
-
-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
-TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- *
- */
-
-#ifndef BRW_SAVE_H
-#define BRW_SAVE_H
-
-#include "mtypes.h"
-#include "brw_attrib.h"
-#include "brw_draw.h"
-
-
-struct brw_save_copied_vtx {
-   GLfloat buffer[BRW_ATTRIB_MAX * 4 * BRW_MAX_COPIED_VERTS];
-   GLuint nr;
-};
-
-
-/* For display lists, this structure holds a run of vertices of the
- * same format, and a strictly well-formed set of begin/end pairs,
- * starting on the first vertex and ending at the last.  Vertex
- * copying on buffer breaks is precomputed according to these
- * primitives, though there are situations where the copying will need
- * correction at execute-time, perhaps by replaying the list as
- * immediate mode commands.
- *
- * On executing this list, the 'current' values may be updated with
- * the values of the final vertex, and often no fixup of the start of
- * the vertex list is required.
- *
- * Eval and other commands that don't fit into these vertex lists are
- * compiled using the fallback opcode mechanism provided by dlist.c.
- */
-struct brw_save_vertex_list {
-   GLubyte attrsz[BRW_ATTRIB_MAX];
-   GLuint vertex_size;
-
-   GLuint buffer_offset;
-   GLuint count;
-   GLuint wrap_count;		/* number of copied vertices at start */
-   GLboolean dangling_attr_ref;	/* current attr implicitly referenced 
-				   outside the list */
-
-   struct brw_draw_prim *prim;
-   GLuint prim_count;
-
-   struct brw_save_vertex_store *vertex_store;
-   struct brw_save_primitive_store *prim_store;
-};
-
-/* These buffers should be a reasonable size to support upload to
- * hardware.  Current brw implementation will re-upload on any
- * changes, so don't make too big or apps which dynamically create
- * dlists and use only a few times will suffer.
- *
- * Consider stategy of uploading regions from the VBO on demand in the
- * case of dynamic vbos.  Then make the dlist code signal that
- * likelyhood as it occurs.  No reason we couldn't change usage
- * internally even though this probably isn't allowed for client VBOs?
- */
-#define BRW_SAVE_BUFFER_SIZE (8*1024) /* dwords */
-#define BRW_SAVE_PRIM_SIZE   128
-#define BRW_SAVE_PRIM_WEAK 0x40
-
-#define BRW_SAVE_FALLBACK    0x10000000
-
-/* Storage to be shared among several vertex_lists.
- */
-struct brw_save_vertex_store {
-   struct gl_buffer_object *bufferobj;
-   GLfloat *buffer;
-   GLuint used;
-   GLuint refcount;
-};
-
-struct brw_save_primitive_store {
-   struct brw_draw_prim buffer[BRW_SAVE_PRIM_SIZE];
-   GLuint used;
-   GLuint refcount;
-};
-
-
-struct brw_save_context {
-   GLcontext *ctx;
-   GLvertexformat vtxfmt;
-   struct gl_client_array arrays[BRW_ATTRIB_MAX];
-   const struct gl_client_array *inputs[BRW_ATTRIB_MAX];
-
-   GLubyte attrsz[BRW_ATTRIB_MAX];
-   GLubyte active_sz[BRW_ATTRIB_MAX];
-   GLuint vertex_size;
-
-   GLfloat *buffer;
-   GLuint count;
-   GLuint wrap_count;
-   GLuint replay_flags;
-
-   struct brw_draw_prim *prim;
-   GLuint prim_count, prim_max;
-
-   struct brw_save_vertex_store *vertex_store;
-   struct brw_save_primitive_store *prim_store;
-
-   GLfloat *vbptr;		   /* cursor, points into buffer */
-   GLfloat vertex[BRW_ATTRIB_MAX*4];	   /* current values */
-   GLfloat *attrptr[BRW_ATTRIB_MAX];
-   GLuint vert_count;
-   GLuint max_vert;
-   GLboolean dangling_attr_ref;
-   GLboolean have_materials;
-
-   GLuint opcode_vertex_list;
-
-   struct brw_save_copied_vtx copied;
-   
-   GLfloat CurrentFloatEdgeFlag;
-
-   GLfloat *current[BRW_ATTRIB_MAX]; /* points into ctx->ListState */
-   GLubyte *currentsz[BRW_ATTRIB_MAX];
-};
-
-
-void brw_save_init( GLcontext *ctx );
-void brw_save_destroy( GLcontext *ctx );
-void brw_save_wakeup( GLcontext *ctx );
-void brw_save_invalidate_state( GLcontext *ctx, GLuint new_state );
-void brw_save_fallback( GLcontext *ctx, GLboolean fallback );
-
-/* Callbacks:
- */
-void brw_save_EndList( GLcontext *ctx );
-void brw_save_NewList( GLcontext *ctx, GLuint list, GLenum mode );
-void brw_save_EndCallList( GLcontext *ctx );
-void brw_save_BeginCallList( GLcontext *ctx, struct mesa_display_list *list );
-void brw_save_SaveFlushVertices( GLcontext *ctx );
-GLboolean brw_save_NotifyBegin( GLcontext *ctx, GLenum mode );
-
-void brw_save_playback_vertex_list( GLcontext *ctx, void *data );
-
-void brw_save_api_init( struct brw_save_context *save );
-
-#endif
diff --git a/src/mesa/drivers/dri/i965/brw_save_api.c b/src/mesa/drivers/dri/i965/brw_save_api.c
deleted file mode 100644
index 06ed1d23aa2..00000000000
--- a/src/mesa/drivers/dri/i965/brw_save_api.c
+++ /dev/null
@@ -1,1170 +0,0 @@
-/**************************************************************************
-
-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
-TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-
-
-/* Display list compiler attempts to store lists of vertices with the
- * same vertex layout.  Additionally it attempts to minimize the need
- * for execute-time fixup of these vertex lists, allowing them to be
- * cached on hardware.
- *
- * There are still some circumstances where this can be thwarted, for
- * example by building a list that consists of one very long primitive
- * (eg Begin(Triangles), 1000 vertices, End), and calling that list
- * from inside a different begin/end object (Begin(Lines), CallList,
- * End).  
- *
- * In that case the code will have to replay the list as individual
- * commands through the Exec dispatch table, or fix up the copied
- * vertices at execute-time.
- *
- * The other case where fixup is required is when a vertex attribute
- * is introduced in the middle of a primitive.  Eg:
- *  Begin(Lines)
- *  TexCoord1f()           Vertex2f()
- *  TexCoord1f() Color3f() Vertex2f()
- *  End()
- *
- *  If the current value of Color isn't known at compile-time, this
- *  primitive will require fixup.
- *
- *
- * The list compiler currently doesn't attempt to compile lists
- * containing EvalCoord or EvalPoint commands.  On encountering one of
- * these, compilation falls back to opcodes.  
- *
- * This could be improved to fallback only when a mix of EvalCoord and
- * Vertex commands are issued within a single primitive.
- */
-
-
-#include "glheader.h"
-#include "context.h"
-#include "dlist.h"
-#include "enums.h"
-#include "macros.h"
-#include "api_validate.h"
-#include "api_arrayelt.h"
-#include "vtxfmt.h"
-#include "dispatch.h"
-
-#include "brw_save.h"
-#include "brw_fallback.h"
-
-
-
-
-/*
- * NOTE: Old 'parity' issue is gone, but copying can still be
- * wrong-footed on replay.
- */
-static GLuint _save_copy_vertices( GLcontext *ctx, 
-				   const struct brw_save_vertex_list *node,
-				   const GLfloat *src_buffer)
-{
-   struct brw_save_context *save = IMM_CONTEXT( ctx )->save;
-   const struct brw_draw_prim *prim = &node->prim[node->prim_count-1];
-   GLuint nr = prim->count;
-   GLuint sz = save->vertex_size;
-   const GLfloat *src = src_buffer + prim->start * sz;
-   GLfloat *dst = save->copied.buffer;
-   GLuint ovf, i;
-
-   if (prim->end)
-      return 0;
-	 
-   switch( prim->mode )
-   {
-   case GL_POINTS:
-      return 0;
-   case GL_LINES:
-      ovf = nr&1;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz*sizeof(GLfloat) );
-      return i;
-   case GL_TRIANGLES:
-      ovf = nr%3;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz*sizeof(GLfloat) );
-      return i;
-   case GL_QUADS:
-      ovf = nr&3;
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz*sizeof(GLfloat) );
-      return i;
-   case GL_LINE_STRIP:
-      if (nr == 0) 
-	 return 0;
-      else {
-	 _mesa_memcpy( dst, src+(nr-1)*sz, sz*sizeof(GLfloat) );
-	 return 1;
-      }
-   case GL_LINE_LOOP:
-   case GL_TRIANGLE_FAN:
-   case GL_POLYGON:
-      if (nr == 0) 
-	 return 0;
-      else if (nr == 1) {
-	 _mesa_memcpy( dst, src+0, sz*sizeof(GLfloat) );
-	 return 1;
-      } else {
-	 _mesa_memcpy( dst, src+0, sz*sizeof(GLfloat) );
-	 _mesa_memcpy( dst+sz, src+(nr-1)*sz, sz*sizeof(GLfloat) );
-	 return 2;
-      }
-   case GL_TRIANGLE_STRIP:
-   case GL_QUAD_STRIP:
-      switch (nr) {
-      case 0: ovf = 0; break;
-      case 1: ovf = 1; break;
-      default: ovf = 2 + (nr&1); break;
-      }
-      for (i = 0 ; i < ovf ; i++)
-	 _mesa_memcpy( dst+i*sz, src+(nr-ovf+i)*sz, sz*sizeof(GLfloat) );
-      return i;
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-
-static struct brw_save_vertex_store *alloc_vertex_store( GLcontext *ctx )
-{
-   struct brw_save_vertex_store *vertex_store = CALLOC_STRUCT(brw_save_vertex_store);
-
-   /* obj->Name needs to be non-zero, but won't ever be examined more
-    * closely than that.  In particular these buffers won't be entered
-    * into the hash and can never be confused with ones visible to the
-    * user.  Perhaps there could be a special number for internal
-    * buffers:
-    */
-   vertex_store->bufferobj = ctx->Driver.NewBufferObject(ctx, 1, GL_ARRAY_BUFFER_ARB);
-
-   ctx->Driver.BufferData( ctx, 
-			   GL_ARRAY_BUFFER_ARB, 
-			   BRW_SAVE_BUFFER_SIZE * sizeof(GLfloat),
-			   NULL,
-			   GL_STATIC_DRAW_ARB,
-			   vertex_store->bufferobj);
-
-   vertex_store->buffer = NULL;
-   vertex_store->used = 0;
-   vertex_store->refcount = 1;
-
-   return vertex_store;
-}
-
-static void free_vertex_store( GLcontext *ctx, struct brw_save_vertex_store *vertex_store )
-{
-   assert(!vertex_store->buffer);
-
-   if (vertex_store->bufferobj)
-      ctx->Driver.DeleteBuffer( ctx, vertex_store->bufferobj );
-
-   FREE( vertex_store );
-}
-
-static GLfloat *map_vertex_store( GLcontext *ctx, struct brw_save_vertex_store *vertex_store )
-{
-   assert(vertex_store->bufferobj);
-   assert(!vertex_store->buffer);
-   vertex_store->buffer = (GLfloat *)ctx->Driver.MapBuffer(ctx, 
-							   GL_ARRAY_BUFFER_ARB,	/* not used */
-							   GL_STATIC_DRAW_ARB, /* not used */
-							   vertex_store->bufferobj); 
-
-   assert(vertex_store->buffer);
-   return vertex_store->buffer + vertex_store->used;
-}
-
-static void unmap_vertex_store( GLcontext *ctx, struct brw_save_vertex_store *vertex_store )
-{
-   ctx->Driver.UnmapBuffer( ctx, GL_ARRAY_BUFFER_ARB, vertex_store->bufferobj );
-   vertex_store->buffer = NULL;
-}
-
-
-static struct brw_save_primitive_store *alloc_prim_store( GLcontext *ctx )
-{
-   struct brw_save_primitive_store *store = CALLOC_STRUCT(brw_save_primitive_store);
-   (void) ctx;
-   store->used = 0;
-   store->refcount = 1;
-   return store;
-}
-
-static void _save_reset_counters( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-
-   save->prim = save->prim_store->buffer + save->prim_store->used;
-   save->buffer = (save->vertex_store->buffer + 
-		   save->vertex_store->used);
-
-   assert(save->buffer == save->vbptr);
-
-   if (save->vertex_size)
-      save->max_vert = ((BRW_SAVE_BUFFER_SIZE - save->vertex_store->used) / 
-			 save->vertex_size);
-   else
-      save->max_vert = 0;
-
-   save->vert_count = 0;
-   save->prim_count = 0;
-   save->prim_max = BRW_SAVE_PRIM_SIZE - save->prim_store->used;
-   save->dangling_attr_ref = 0;
-}
-
-
-/* Insert the active immediate struct onto the display list currently
- * being built.
- */
-static void _save_compile_vertex_list( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   struct brw_save_vertex_list *node;
-
-   /* Allocate space for this structure in the display list currently
-    * being compiled.
-    */
-   node = (struct brw_save_vertex_list *)
-      _mesa_alloc_instruction(ctx, save->opcode_vertex_list, sizeof(*node));
-
-   if (!node)
-      return;
-
-   /* Duplicate our template, increment refcounts to the storage structs:
-    */
-   _mesa_memcpy(node->attrsz, save->attrsz, sizeof(node->attrsz)); 
-   node->vertex_size = save->vertex_size;
-   node->buffer_offset = (save->buffer - save->vertex_store->buffer) * sizeof(GLfloat); 
-   node->count = save->vert_count;
-   node->wrap_count = save->copied.nr;
-   node->dangling_attr_ref = save->dangling_attr_ref;
-   node->prim = save->prim;
-   node->prim_count = save->prim_count;
-   node->vertex_store = save->vertex_store;
-   node->prim_store = save->prim_store;
-
-   node->vertex_store->refcount++;
-   node->prim_store->refcount++;
-
-   assert(node->attrsz[BRW_ATTRIB_POS] != 0 ||
-	  node->count == 0);
-
-   if (save->dangling_attr_ref)
-      ctx->ListState.CurrentList->flags |= MESA_DLIST_DANGLING_REFS;
-
-   save->vertex_store->used += save->vertex_size * node->count;
-   save->prim_store->used += node->prim_count;
-
-
-   /* Copy duplicated vertices 
-    */
-   save->copied.nr = _save_copy_vertices( ctx, node, save->buffer );
-
-
-   /* Deal with GL_COMPILE_AND_EXECUTE:
-    */
-   if (ctx->ExecuteFlag) {
-      struct _glapi_table *dispatch = GET_DISPATCH();
-
-      _glapi_set_dispatch(ctx->Exec);
-
-      brw_loopback_vertex_list( ctx,
-				(const GLfloat *)((const char *)save->vertex_store->buffer + 
-						  node->buffer_offset),
-				node->attrsz,
-				node->prim,
-				node->prim_count,
-				node->wrap_count,
-				node->vertex_size);
-
-      _glapi_set_dispatch(dispatch);
-   }
-
-
-   /* Decide whether the storage structs are full, or can be used for
-    * the next vertex lists as well.
-    */
-   if (save->vertex_store->used > 
-       BRW_SAVE_BUFFER_SIZE - 16 * (save->vertex_size + 4)) {
-
-      /* Unmap old store:
-       */
-      unmap_vertex_store( ctx, save->vertex_store );
-
-      /* Release old reference:
-       */
-      save->vertex_store->refcount--; 
-      assert(save->vertex_store->refcount != 0);
-      save->vertex_store = NULL;
-
-      /* Allocate and map new store:
-       */
-      save->vertex_store = alloc_vertex_store( ctx );
-      save->vbptr = map_vertex_store( ctx, save->vertex_store );
-   } 
-
-   if (save->prim_store->used > BRW_SAVE_PRIM_SIZE - 6) {
-      save->prim_store->refcount--; 
-      assert(save->prim_store->refcount != 0);
-      save->prim_store = alloc_prim_store( ctx );
-   } 
-
-   /* Reset our structures for the next run of vertices:
-    */
-   _save_reset_counters( ctx );
-}
-
-
-/* TODO -- If no new vertices have been stored, don't bother saving
- * it.
- */
-static void _save_wrap_buffers( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLint i = save->prim_count - 1;
-   GLenum mode;
-   GLboolean weak;
-
-   assert(i < (GLint) save->prim_max);
-   assert(i >= 0);
-
-   /* Close off in-progress primitive.
-    */
-   save->prim[i].count = (save->vert_count - 
-			  save->prim[i].start);
-   mode = save->prim[i].mode;
-   weak = save->prim[i].weak;
-   
-   /* store the copied vertices, and allocate a new list.
-    */
-   _save_compile_vertex_list( ctx );
-
-   /* Restart interrupted primitive
-    */
-   save->prim[0].mode = mode;
-   save->prim[0].weak = weak;
-   save->prim[0].begin = 0;
-   save->prim[0].end = 0;
-   save->prim[0].pad = 0;
-   save->prim[0].start = 0;
-   save->prim[0].count = 0;
-   save->prim_count = 1;
-}
-
-
-
-/* Called only when buffers are wrapped as the result of filling the
- * vertex_store struct.  
- */
-static void _save_wrap_filled_vertex( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLfloat *data = save->copied.buffer;
-   GLuint i;
-
-   /* Emit a glEnd to close off the last vertex list.
-    */
-   _save_wrap_buffers( ctx );
-   
-    /* Copy stored stored vertices to start of new list.
-    */
-   assert(save->max_vert - save->vert_count > save->copied.nr);
-
-   for (i = 0 ; i < save->copied.nr ; i++) {
-      _mesa_memcpy( save->vbptr, data, save->vertex_size * sizeof(GLfloat));
-      data += save->vertex_size;
-      save->vbptr += save->vertex_size;
-      save->vert_count++;
-   }
-}
-
-
-static void _save_copy_to_current( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-   GLuint i;
-
-   for (i = BRW_ATTRIB_POS+1 ; i < BRW_ATTRIB_MAX ; i++) {
-      if (save->attrsz[i]) {
-	 save->currentsz[i][0] = save->attrsz[i];
-	 COPY_CLEAN_4V(save->current[i], 
-		    save->attrsz[i], 
-		    save->attrptr[i]);
-      }
-   }
-
-   /* Edgeflag requires special treatment: 
-    *
-    * TODO: change edgeflag to GLfloat in Mesa.
-    */
-   if (save->attrsz[BRW_ATTRIB_EDGEFLAG]) {
-      ctx->ListState.ActiveEdgeFlag = 1;
-      save->CurrentFloatEdgeFlag = 
-	 save->attrptr[BRW_ATTRIB_EDGEFLAG][0];
-      ctx->ListState.CurrentEdgeFlag = 
-	 (save->CurrentFloatEdgeFlag == 1.0);
-   }
-}
-
-
-static void _save_copy_from_current( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-   GLint i;
-
-   for (i = BRW_ATTRIB_POS+1 ; i < BRW_ATTRIB_MAX ; i++) 
-      switch (save->attrsz[i]) {
-      case 4: save->attrptr[i][3] = save->current[i][3];
-      case 3: save->attrptr[i][2] = save->current[i][2];
-      case 2: save->attrptr[i][1] = save->current[i][1];
-      case 1: save->attrptr[i][0] = save->current[i][0];
-      case 0: break;
-      }
-
-   /* Edgeflag requires special treatment:
-    */
-   if (save->attrsz[BRW_ATTRIB_EDGEFLAG]) {
-      save->CurrentFloatEdgeFlag = (GLfloat)ctx->ListState.CurrentEdgeFlag;
-      save->attrptr[BRW_ATTRIB_EDGEFLAG][0] = save->CurrentFloatEdgeFlag;
-   }
-}
-
-
-
-
-/* Flush existing data, set new attrib size, replay copied vertices.
- */ 
-static void _save_upgrade_vertex( GLcontext *ctx, 
-				 GLuint attr,
-				 GLuint newsz )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLuint oldsz;
-   GLuint i;
-   GLfloat *tmp;
-
-   /* Store the current run of vertices, and emit a GL_END.  Emit a
-    * BEGIN in the new buffer.
-    */
-   if (save->vert_count) 
-      _save_wrap_buffers( ctx );
-   else
-      assert( save->copied.nr == 0 );
-
-   /* Do a COPY_TO_CURRENT to ensure back-copying works for the case
-    * when the attribute already exists in the vertex and is having
-    * its size increased.  
-    */
-   _save_copy_to_current( ctx );
-
-   /* Fix up sizes:
-    */
-   oldsz = save->attrsz[attr];
-   save->attrsz[attr] = newsz;
-
-   save->vertex_size += newsz - oldsz;
-   save->max_vert = ((BRW_SAVE_BUFFER_SIZE - save->vertex_store->used) / 
-		      save->vertex_size);
-   save->vert_count = 0;
-
-   /* Recalculate all the attrptr[] values:
-    */
-   for (i = 0, tmp = save->vertex ; i < BRW_ATTRIB_MAX ; i++) {
-      if (save->attrsz[i]) {
-	 save->attrptr[i] = tmp;
-	 tmp += save->attrsz[i];
-      }
-      else 
-	 save->attrptr[i] = NULL; /* will not be dereferenced. */
-   }
-
-   /* Copy from current to repopulate the vertex with correct values.
-    */
-   _save_copy_from_current( ctx );
-
-   /* Replay stored vertices to translate them to new format here.
-    *
-    * If there are copied vertices and the new (upgraded) attribute
-    * has not been defined before, this list is somewhat degenerate,
-    * and will need fixup at runtime.
-    */
-   if (save->copied.nr)
-   {
-      GLfloat *data = save->copied.buffer;
-      GLfloat *dest = save->buffer;
-      GLuint j;
-
-      /* Need to note this and fix up at runtime (or loopback):
-       */
-      if (attr != BRW_ATTRIB_POS && save->currentsz[attr][0] == 0) {
-	 assert(oldsz == 0);
-	 save->dangling_attr_ref = GL_TRUE;
-      }
-
-      for (i = 0 ; i < save->copied.nr ; i++) {
-	 for (j = 0 ; j < BRW_ATTRIB_MAX ; j++) {
-	    if (save->attrsz[j]) {
-	       if (j == attr) {
-		  if (oldsz) {
-		     COPY_CLEAN_4V( dest, oldsz, data );
-		     data += oldsz;
-		     dest += newsz;
-		  }
-		  else {
-		     COPY_SZ_4V( dest, newsz, save->current[attr] );
-		     dest += newsz;
-		  }
-	       }
-	       else {
-		  GLint sz = save->attrsz[j];
-		  COPY_SZ_4V( dest, sz, data );
-		  data += sz;
-		  dest += sz;
-	       }
-	    }
-	 }
-      }
-
-      save->vbptr = dest;
-      save->vert_count += save->copied.nr;
-   }
-}
-
-static void save_fixup_vertex( GLcontext *ctx, GLuint attr, GLuint sz )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-
-   if (sz > save->attrsz[attr]) {
-      /* New size is larger.  Need to flush existing vertices and get
-       * an enlarged vertex format.
-       */
-      _save_upgrade_vertex( ctx, attr, sz );
-   }
-   else if (sz < save->active_sz[attr]) {
-      static GLfloat id[4] = { 0, 0, 0, 1 };
-      GLuint i;
-
-      /* New size is equal or smaller - just need to fill in some
-       * zeros.
-       */
-      for (i = sz ; i <= save->attrsz[attr] ; i++)
-	 save->attrptr[attr][i-1] = id[i-1];
-   }
-
-   save->active_sz[attr] = sz;
-}
-
-static void _save_reset_vertex( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLuint i;
-
-   for (i = 0 ; i < BRW_ATTRIB_MAX ; i++) {
-      save->attrsz[i] = 0;
-      save->active_sz[i] = 0;
-   }
-      
-   save->vertex_size = 0;
-}
-
-
-
-#define ERROR()   _mesa_compile_error( ctx, GL_INVALID_ENUM, __FUNCTION__ );
-
-
-/* Only one size for each attribute may be active at once.  Eg. if
- * Color3f is installed/active, then Color4f may not be, even if the
- * vertex actually contains 4 color coordinates.  This is because the
- * 3f version won't otherwise set color[3] to 1.0 -- this is the job
- * of the chooser function when switching between Color4f and Color3f.
- */
-#define ATTR( A, N, V0, V1, V2, V3 )				\
-do {								\
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;	\
-								\
-   if (save->active_sz[A] != N)				\
-      save_fixup_vertex(ctx, A, N);				\
-								\
-   {								\
-      GLfloat *dest = save->attrptr[A];			\
-      if (N>0) dest[0] = V0;					\
-      if (N>1) dest[1] = V1;					\
-      if (N>2) dest[2] = V2;					\
-      if (N>3) dest[3] = V3;					\
-   }								\
-								\
-   if ((A) == 0) {						\
-      GLuint i;							\
-								\
-      for (i = 0; i < save->vertex_size; i++)			\
-	 save->vbptr[i] = save->vertex[i];			\
-								\
-      save->vbptr += save->vertex_size;				\
-								\
-      if (++save->vert_count >= save->max_vert)			\
-	 _save_wrap_filled_vertex( ctx );			\
-   }								\
-} while (0)
-
-#define TAG(x) _save_##x
-
-#include "brw_attrib_tmp.h"
-
-
-
-
-/* Cope with EvalCoord/CallList called within a begin/end object:
- *     -- Flush current buffer
- *     -- Fallback to opcodes for the rest of the begin/end object.
- */
-#define DO_FALLBACK(ctx) 							\
-do {									\
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;					\
-									\
-   if (save->vert_count || save->prim_count) 						\
-      _save_compile_vertex_list( ctx );					\
-									\
-   _save_copy_to_current( ctx );					\
-   _save_reset_vertex( ctx );						\
-   _save_reset_counters( ctx );  \
-   _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt );	\
-   ctx->Driver.SaveNeedFlush = 0;					\
-} while (0)
-
-static void GLAPIENTRY _save_EvalCoord1f( GLfloat u )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalCoord1f( u );
-}
-
-static void GLAPIENTRY _save_EvalCoord1fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalCoord1fv( v );
-}
-
-static void GLAPIENTRY _save_EvalCoord2f( GLfloat u, GLfloat v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalCoord2f( u, v );
-}
-
-static void GLAPIENTRY _save_EvalCoord2fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalCoord2fv( v );
-}
-
-static void GLAPIENTRY _save_EvalPoint1( GLint i )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalPoint1( i );
-}
-
-static void GLAPIENTRY _save_EvalPoint2( GLint i, GLint j )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->EvalPoint2( i, j );
-}
-
-static void GLAPIENTRY _save_CallList( GLuint l )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->CallList( l );
-}
-
-static void GLAPIENTRY _save_CallLists( GLsizei n, GLenum type, const GLvoid *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   DO_FALLBACK(ctx);
-   ctx->Save->CallLists( n, type, v );
-}
-
-
-
-
-/* This begin is hooked into ...  Updating of
- * ctx->Driver.CurrentSavePrimitive is already taken care of.
- */
-GLboolean brw_save_NotifyBegin( GLcontext *ctx, GLenum mode )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-
-   GLuint i = save->prim_count++;
-
-   assert(i < save->prim_max);
-   save->prim[i].mode = mode & ~BRW_SAVE_PRIM_WEAK;
-   save->prim[i].begin = 1;
-   save->prim[i].end = 0;
-   save->prim[i].weak = (mode & BRW_SAVE_PRIM_WEAK) ? 1 : 0;
-   save->prim[i].pad = 0;
-   save->prim[i].start = save->vert_count;
-   save->prim[i].count = 0;   
-
-   _mesa_install_save_vtxfmt( ctx, &save->vtxfmt );      
-   ctx->Driver.SaveNeedFlush = 1;
-   return GL_TRUE;
-}
-
-
-
-static void GLAPIENTRY _save_End( void )
-{
-   GET_CURRENT_CONTEXT( ctx ); 
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-   GLint i = save->prim_count - 1;
-
-   ctx->Driver.CurrentSavePrimitive = PRIM_OUTSIDE_BEGIN_END;
-   save->prim[i].end = 1;
-   save->prim[i].count = (save->vert_count - 
-			  save->prim[i].start);
-
-   if (i == (GLint) save->prim_max - 1) {
-      _save_compile_vertex_list( ctx );
-      assert(save->copied.nr == 0);
-   }
-
-   /* Swap out this vertex format while outside begin/end.  Any color,
-    * etc. received between here and the next begin will be compiled
-    * as opcodes.
-    */   
-   _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt );
-}
-
-
-/* These are all errors as this vtxfmt is only installed inside
- * begin/end pairs.
- */
-static void GLAPIENTRY _save_DrawElements(GLenum mode, GLsizei count, GLenum type,
-			       const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) mode; (void) count; (void) type; (void) indices;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawElements" );
-}
-
-
-static void GLAPIENTRY _save_DrawRangeElements(GLenum mode,
-				    GLuint start, GLuint end,
-				    GLsizei count, GLenum type,
-				    const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) mode; (void) start; (void) end; (void) count; (void) type; (void) indices;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawRangeElements" );
-}
-
-static void GLAPIENTRY _save_DrawArrays(GLenum mode, GLint start, GLsizei count)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) mode; (void) start; (void) count;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glDrawArrays" );
-}
-
-static void GLAPIENTRY _save_Rectf( GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) x1; (void) y1; (void) x2; (void) y2;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glRectf" );
-}
-
-static void GLAPIENTRY _save_EvalMesh1( GLenum mode, GLint i1, GLint i2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) mode; (void) i1; (void) i2;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glEvalMesh1" );
-}
-
-static void GLAPIENTRY _save_EvalMesh2( GLenum mode, GLint i1, GLint i2,
-				  GLint j1, GLint j2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   (void) mode; (void) i1; (void) i2; (void) j1; (void) j2;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "glEvalMesh2" );
-}
-
-static void GLAPIENTRY _save_Begin( GLenum mode )
-{
-   GET_CURRENT_CONTEXT( ctx );
-   (void) mode;
-   _mesa_compile_error( ctx, GL_INVALID_OPERATION, "Recursive glBegin" );
-}
-
-
-/* Unlike the functions above, these are to be hooked into the vtxfmt
- * maintained in ctx->ListState, active when the list is known or
- * suspected to be outside any begin/end primitive.
- */
-static void GLAPIENTRY _save_OBE_Rectf( GLfloat x1, GLfloat y1, GLfloat x2, GLfloat y2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   brw_save_NotifyBegin( ctx, GL_QUADS | BRW_SAVE_PRIM_WEAK );
-   CALL_Vertex2f(GET_DISPATCH(), ( x1, y1 ));
-   CALL_Vertex2f(GET_DISPATCH(), ( x2, y1 ));
-   CALL_Vertex2f(GET_DISPATCH(), ( x2, y2 ));
-   CALL_Vertex2f(GET_DISPATCH(), ( x1, y2 ));
-   CALL_End(GET_DISPATCH(), ());
-}
-
-
-static void GLAPIENTRY _save_OBE_DrawArrays(GLenum mode, GLint start, GLsizei count)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   GLint i;
-
-   if (!_mesa_validate_DrawArrays( ctx, mode, start, count ))
-      return;
-
-   _ae_map_vbos( ctx );
-
-   brw_save_NotifyBegin( ctx, mode | BRW_SAVE_PRIM_WEAK );
-   for (i = 0; i < count; i++)
-       CALL_ArrayElement(GET_DISPATCH(), (start + i));
-   CALL_End(GET_DISPATCH(), ());
-
-   _ae_unmap_vbos( ctx );
-}
-
-/* Could do better by copying the arrays and element list intact and
- * then emitting an indexed prim at runtime.
- */
-static void GLAPIENTRY _save_OBE_DrawElements(GLenum mode, GLsizei count, GLenum type,
-				   const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   GLint i;
-
-   if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
-      return;
-
-   _ae_map_vbos( ctx );
-
-   brw_save_NotifyBegin( ctx, mode | BRW_SAVE_PRIM_WEAK );
-
-   switch (type) {
-   case GL_UNSIGNED_BYTE:
-      for (i = 0 ; i < count ; i++)
-	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLubyte *)indices)[i] ));
-      break;
-   case GL_UNSIGNED_SHORT:
-      for (i = 0 ; i < count ; i++)
-	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLushort *)indices)[i] ));
-      break;
-   case GL_UNSIGNED_INT:
-      for (i = 0 ; i < count ; i++)
-	  CALL_ArrayElement(GET_DISPATCH(), ( ((GLuint *)indices)[i] ));
-      break;
-   default:
-      _mesa_error( ctx, GL_INVALID_ENUM, "glDrawElements(type)" );
-      break;
-   }
-
-   CALL_End(GET_DISPATCH(), ());
-
-   _ae_unmap_vbos( ctx );
-}
-
-static void GLAPIENTRY _save_OBE_DrawRangeElements(GLenum mode,
-					GLuint start, GLuint end,
-					GLsizei count, GLenum type,
-					const GLvoid *indices)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   if (_mesa_validate_DrawRangeElements( ctx, mode,
-					 start, end,
-					 count, type, indices ))
-      _save_OBE_DrawElements( mode, count, type, indices );
-}
-
-
-
-
-
-static void _save_vtxfmt_init( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLvertexformat *vfmt = &save->vtxfmt;
-
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
-   vfmt->Begin = _save_Begin;
-   vfmt->Color3f = _save_Color3f;
-   vfmt->Color3fv = _save_Color3fv;
-   vfmt->Color4f = _save_Color4f;
-   vfmt->Color4fv = _save_Color4fv;
-   vfmt->EdgeFlag = _save_EdgeFlag;
-   vfmt->End = _save_End;
-   vfmt->FogCoordfEXT = _save_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = _save_FogCoordfvEXT;
-   vfmt->Indexf = _save_Indexf;
-   vfmt->Indexfv = _save_Indexfv;
-   vfmt->Materialfv = _save_Materialfv;
-   vfmt->MultiTexCoord1fARB = _save_MultiTexCoord1f;
-   vfmt->MultiTexCoord1fvARB = _save_MultiTexCoord1fv;
-   vfmt->MultiTexCoord2fARB = _save_MultiTexCoord2f;
-   vfmt->MultiTexCoord2fvARB = _save_MultiTexCoord2fv;
-   vfmt->MultiTexCoord3fARB = _save_MultiTexCoord3f;
-   vfmt->MultiTexCoord3fvARB = _save_MultiTexCoord3fv;
-   vfmt->MultiTexCoord4fARB = _save_MultiTexCoord4f;
-   vfmt->MultiTexCoord4fvARB = _save_MultiTexCoord4fv;
-   vfmt->Normal3f = _save_Normal3f;
-   vfmt->Normal3fv = _save_Normal3fv;
-   vfmt->SecondaryColor3fEXT = _save_SecondaryColor3fEXT;
-   vfmt->SecondaryColor3fvEXT = _save_SecondaryColor3fvEXT;
-   vfmt->TexCoord1f = _save_TexCoord1f;
-   vfmt->TexCoord1fv = _save_TexCoord1fv;
-   vfmt->TexCoord2f = _save_TexCoord2f;
-   vfmt->TexCoord2fv = _save_TexCoord2fv;
-   vfmt->TexCoord3f = _save_TexCoord3f;
-   vfmt->TexCoord3fv = _save_TexCoord3fv;
-   vfmt->TexCoord4f = _save_TexCoord4f;
-   vfmt->TexCoord4fv = _save_TexCoord4fv;
-   vfmt->Vertex2f = _save_Vertex2f;
-   vfmt->Vertex2fv = _save_Vertex2fv;
-   vfmt->Vertex3f = _save_Vertex3f;
-   vfmt->Vertex3fv = _save_Vertex3fv;
-   vfmt->Vertex4f = _save_Vertex4f;
-   vfmt->Vertex4fv = _save_Vertex4fv;
-   vfmt->VertexAttrib1fARB = _save_VertexAttrib1fARB;
-   vfmt->VertexAttrib1fvARB = _save_VertexAttrib1fvARB;
-   vfmt->VertexAttrib2fARB = _save_VertexAttrib2fARB;
-   vfmt->VertexAttrib2fvARB = _save_VertexAttrib2fvARB;
-   vfmt->VertexAttrib3fARB = _save_VertexAttrib3fARB;
-   vfmt->VertexAttrib3fvARB = _save_VertexAttrib3fvARB;
-   vfmt->VertexAttrib4fARB = _save_VertexAttrib4fARB;
-   vfmt->VertexAttrib4fvARB = _save_VertexAttrib4fvARB;
-
-   vfmt->VertexAttrib1fNV = _save_VertexAttrib1fNV;
-   vfmt->VertexAttrib1fvNV = _save_VertexAttrib1fvNV;
-   vfmt->VertexAttrib2fNV = _save_VertexAttrib2fNV;
-   vfmt->VertexAttrib2fvNV = _save_VertexAttrib2fvNV;
-   vfmt->VertexAttrib3fNV = _save_VertexAttrib3fNV;
-   vfmt->VertexAttrib3fvNV = _save_VertexAttrib3fvNV;
-   vfmt->VertexAttrib4fNV = _save_VertexAttrib4fNV;
-   vfmt->VertexAttrib4fvNV = _save_VertexAttrib4fvNV;
-   
-   /* This will all require us to fallback to saving the list as opcodes:
-    */ 
-   vfmt->CallList = _save_CallList; /* inside begin/end */
-   vfmt->CallLists = _save_CallLists; /* inside begin/end */
-   vfmt->EvalCoord1f = _save_EvalCoord1f;
-   vfmt->EvalCoord1fv = _save_EvalCoord1fv;
-   vfmt->EvalCoord2f = _save_EvalCoord2f;
-   vfmt->EvalCoord2fv = _save_EvalCoord2fv;
-   vfmt->EvalPoint1 = _save_EvalPoint1;
-   vfmt->EvalPoint2 = _save_EvalPoint2;
-
-   /* These are all errors as we at least know we are in some sort of
-    * begin/end pair:
-    */
-   vfmt->EvalMesh1 = _save_EvalMesh1;	
-   vfmt->EvalMesh2 = _save_EvalMesh2;
-   vfmt->Begin = _save_Begin;
-   vfmt->Rectf = _save_Rectf;
-   vfmt->DrawArrays = _save_DrawArrays;
-   vfmt->DrawElements = _save_DrawElements;
-   vfmt->DrawRangeElements = _save_DrawRangeElements;
-
-}
-
-
-void brw_save_SaveFlushVertices( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-
-   /* Noop when we are actually active:
-    */
-   if (ctx->Driver.CurrentSavePrimitive == PRIM_INSIDE_UNKNOWN_PRIM ||
-       ctx->Driver.CurrentSavePrimitive <= GL_POLYGON)
-      return;
-
-   if (save->vert_count ||
-       save->prim_count) 
-      _save_compile_vertex_list( ctx );
-   
-   _save_copy_to_current( ctx );
-   _save_reset_vertex( ctx );
-   _save_reset_counters( ctx );  
-   ctx->Driver.SaveNeedFlush = 0;
-}
-
-void brw_save_NewList( GLcontext *ctx, GLuint list, GLenum mode )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-
-   (void) list; (void) mode;
-
-   if (!save->prim_store)
-      save->prim_store = alloc_prim_store( ctx );
-
-   if (!save->vertex_store) 
-      save->vertex_store = alloc_vertex_store( ctx );
-      
-   save->vbptr = map_vertex_store( ctx, save->vertex_store );
-   
-   _save_reset_vertex( ctx );
-   _save_reset_counters( ctx );  
-   ctx->Driver.SaveNeedFlush = 0;
-}
-
-void brw_save_EndList( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   unmap_vertex_store( ctx, save->vertex_store );
-
-   assert(save->vertex_size == 0);
-}
- 
-void brw_save_BeginCallList( GLcontext *ctx, struct mesa_display_list *dlist )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   save->replay_flags |= dlist->flags;
-}
-
-void brw_save_EndCallList( GLcontext *ctx )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   
-   if (ctx->ListState.CallDepth == 1) {
-      /* This is correct: want to keep only the BRW_SAVE_FALLBACK
-       * flag, if it is set:
-       */
-      save->replay_flags &= BRW_SAVE_FALLBACK;
-   }
-}
-
-
-static void brw_destroy_vertex_list( GLcontext *ctx, void *data )
-{
-   struct brw_save_vertex_list *node = (struct brw_save_vertex_list *)data;
-   (void) ctx;
-
-   if ( --node->vertex_store->refcount == 0 ) 
-      free_vertex_store( ctx, node->vertex_store );
-
-   if ( --node->prim_store->refcount == 0 )
-      FREE( node->prim_store );
-}
-
-
-static void brw_print_vertex_list( GLcontext *ctx, void *data )
-{
-   struct brw_save_vertex_list *node = (struct brw_save_vertex_list *)data;
-   GLuint i;
-   (void) ctx;
-
-   _mesa_debug(NULL, "BRW-VERTEX-LIST, %u vertices %d primitives, %d vertsize\n",
-               node->count,
-	       node->prim_count,
-	       node->vertex_size);
-
-   for (i = 0 ; i < node->prim_count ; i++) {
-      struct brw_draw_prim *prim = &node->prim[i];
-      _mesa_debug(NULL, "   prim %d: %s%s %d..%d %s %s\n",
-		  i, 
-		  _mesa_lookup_enum_by_nr(prim->mode),
-		  prim->weak ? " (weak)" : "",
-		  prim->start, 
-		  prim->start + prim->count,
-		  (prim->begin) ? "BEGIN" : "(wrap)",
-		  (prim->end) ? "END" : "(wrap)");
-   }
-}
-
-
-static void _save_current_init( GLcontext *ctx ) 
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-   GLint i;
-
-   for (i = 0; i < BRW_ATTRIB_FIRST_MATERIAL; i++) {
-      save->currentsz[i] = &ctx->ListState.ActiveAttribSize[i];
-      save->current[i] = ctx->ListState.CurrentAttrib[i];
-   }
-
-   for (i = BRW_ATTRIB_FIRST_MATERIAL; i <= BRW_ATTRIB_LAST_MATERIAL; i++) {
-      const GLuint j = i - BRW_ATTRIB_FIRST_MATERIAL;
-      ASSERT(j < MAT_ATTRIB_MAX);
-      save->currentsz[i] = &ctx->ListState.ActiveMaterialSize[j];
-      save->current[i] = ctx->ListState.CurrentMaterial[j];
-   }
-
-   save->currentsz[BRW_ATTRIB_INDEX] = &ctx->ListState.ActiveIndex;
-   save->current[BRW_ATTRIB_INDEX] = &ctx->ListState.CurrentIndex;
-
-   save->currentsz[BRW_ATTRIB_EDGEFLAG] = &ctx->ListState.ActiveEdgeFlag;
-   save->current[BRW_ATTRIB_EDGEFLAG] = &save->CurrentFloatEdgeFlag;
-}
-
-/**
- * Initialize the display list compiler
- */
-void brw_save_api_init( struct brw_save_context *save )
-{
-   GLcontext *ctx = save->ctx;
-   GLuint i;
-
-   save->opcode_vertex_list =
-      _mesa_alloc_opcode( ctx,
-			  sizeof(struct brw_save_vertex_list),
-			  brw_save_playback_vertex_list,
-			  brw_destroy_vertex_list,
-			  brw_print_vertex_list );
-
-   ctx->Driver.NotifySaveBegin = brw_save_NotifyBegin;
-
-   _save_vtxfmt_init( ctx );
-   _save_current_init( ctx );
-
-   for (i = 0; i < BRW_ATTRIB_MAX; i++)
-      save->inputs[i] = &save->arrays[i];
-
-   /* Hook our array functions into the outside-begin-end vtxfmt in 
-    * ctx->ListState.
-    */
-   ctx->ListState.ListVtxfmt.Rectf = _save_OBE_Rectf;
-   ctx->ListState.ListVtxfmt.DrawArrays = _save_OBE_DrawArrays;
-   ctx->ListState.ListVtxfmt.DrawElements = _save_OBE_DrawElements;
-   ctx->ListState.ListVtxfmt.DrawRangeElements = _save_OBE_DrawRangeElements;
-   _mesa_install_save_vtxfmt( ctx, &ctx->ListState.ListVtxfmt );
-}
-
diff --git a/src/mesa/drivers/dri/i965/brw_save_draw.c b/src/mesa/drivers/dri/i965/brw_save_draw.c
deleted file mode 100644
index cebdd8d0607..00000000000
--- a/src/mesa/drivers/dri/i965/brw_save_draw.c
+++ /dev/null
@@ -1,221 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.1
- *
- * Copyright (C) 1999-2004  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-/* Author:
- *    Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "context.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "macros.h"
-#include "light.h"
-#include "state.h"
-
-#include "brw_save.h"
-#include "brw_draw.h"
-#include "brw_fallback.h"
-
-/*
- * After playback, copy everything but the position from the
- * last vertex to the saved state
- */
-static void _playback_copy_to_current( GLcontext *ctx,
-				       const struct brw_save_vertex_list *node )
-{
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save; 
-   GLfloat vertex[BRW_ATTRIB_MAX * 4], *data = vertex;
-   GLuint i, offset;
-
-   if (node->count)
-      offset = (node->buffer_offset + 
-		(node->count-1) * node->vertex_size * sizeof(GLfloat));
-   else
-      offset = node->buffer_offset;
-
-   ctx->Driver.GetBufferSubData( ctx, 0, offset, 
-				 node->vertex_size * sizeof(GLfloat), 
-				 data, node->vertex_store->bufferobj );
-
-   for (i = 0 ; i < BRW_ATTRIB_MAX ; i++) {
-      if (node->attrsz[i]) {
-	 if (i != BRW_ATTRIB_POS)
-	    COPY_CLEAN_4V(save->current[i], node->attrsz[i], data);
-
-	 if (i >= BRW_ATTRIB_MAT_FRONT_AMBIENT &&
-	     i <= BRW_ATTRIB_MAT_BACK_INDEXES)
-	    ctx->NewState |= _NEW_LIGHT;
-
-	 /* Edgeflag requires special treatment:
-	  */
-	 if (i == BRW_ATTRIB_EDGEFLAG)
-	    ctx->Current.EdgeFlag = (data[0] == 1.0);
-
-	 data += node->attrsz[i] * sizeof(GLfloat);
-      }
-   }
-
-   /* Edgeflag requires special treatment:
-    */
-   if (node->attrsz[BRW_ATTRIB_EDGEFLAG]) {
-      ctx->Current.EdgeFlag = (data[0] == 1.0);
-   }
-
-
-#if 1
-   /* Colormaterial -- this kindof sucks.
-    */
-   if (ctx->Light.ColorMaterialEnabled) {
-      _mesa_update_color_material(ctx, ctx->Current.Attrib[BRW_ATTRIB_COLOR0]);
-   }
-#endif
-
-   /* CurrentExecPrimitive
-    */
-   if (node->prim_count) {
-      const struct brw_draw_prim *prim = &node->prim[node->prim_count - 1];
-      if (prim->end)
-	 ctx->Driver.CurrentExecPrimitive = PRIM_OUTSIDE_BEGIN_END;
-      else
-	 ctx->Driver.CurrentExecPrimitive = prim->mode;
-   }
-}
-
-
-
-/* Treat the vertex storage as a VBO, define vertex arrays pointing
- * into it:
- */
-static void brw_bind_vertex_list( struct brw_save_context *save,
-                                   const struct brw_save_vertex_list *node )
-{
-   struct gl_client_array *arrays = save->arrays;
-   GLuint data = node->buffer_offset;
-   GLuint attr;
-
-   memset(arrays, 0, BRW_ATTRIB_MAX * sizeof(arrays[0]));
-
-   for (attr = 0; attr < BRW_ATTRIB_MAX; attr++) {
-      if (node->attrsz[attr]) {
-	 arrays[attr].Ptr = (const GLubyte *)data;
-	 arrays[attr].Size = node->attrsz[attr];
-	 arrays[attr].StrideB = node->vertex_size * sizeof(GLfloat);
-	 arrays[attr].Stride = node->vertex_size * sizeof(GLfloat);
-	 arrays[attr].Type = GL_FLOAT;
-	 arrays[attr].Enabled = 1;
-	 arrays[attr].BufferObj = node->vertex_store->bufferobj;
-	 arrays[attr]._MaxElement = node->count; /* ??? */
-	 
-	 assert(arrays[attr].BufferObj->Name);
-
-	 data += node->attrsz[attr] * sizeof(GLfloat);
-      }
-   }
-}
-
-static void brw_save_loopback_vertex_list( GLcontext *ctx,
-					   const struct brw_save_vertex_list *list )
-{
-   const char *buffer = ctx->Driver.MapBuffer(ctx, 
-					      GL_ARRAY_BUFFER_ARB, 
-					      GL_DYNAMIC_READ_ARB, /* ? */
-					       list->vertex_store->bufferobj);
-
-   brw_loopback_vertex_list( ctx,
-			     (const GLfloat *)(buffer + list->buffer_offset),
-			     list->attrsz,
-			     list->prim,
-			     list->prim_count,
-			     list->wrap_count,
-			     list->vertex_size);
-
-   ctx->Driver.UnmapBuffer(ctx, GL_ARRAY_BUFFER_ARB, 
-			   list->vertex_store->bufferobj);
-}
-
-
-/**
- * Execute the buffer and save copied verts.
- */
-void brw_save_playback_vertex_list( GLcontext *ctx, void *data )
-{
-   const struct brw_save_vertex_list *node = (const struct brw_save_vertex_list *) data;
-   struct brw_save_context *save = IMM_CONTEXT(ctx)->save;
-
-   FLUSH_CURRENT(ctx, 0);
-
-   if (node->prim_count > 0 && node->count > 0) {
-
-      if (ctx->Driver.CurrentExecPrimitive != PRIM_OUTSIDE_BEGIN_END &&
-	  node->prim[0].begin) {
-
-	 /* Degenerate case: list is called inside begin/end pair and
-	  * includes operations such as glBegin or glDrawArrays.
-	  */
-	 if (0)
-	    _mesa_printf("displaylist recursive begin");
-
-	 brw_save_loopback_vertex_list( ctx, node );
-	 return;
-      }
-      else if (save->replay_flags) {
-	 /* Various degnerate cases: translate into immediate mode
-	  * calls rather than trying to execute in place.
-	  */
-	 brw_save_loopback_vertex_list( ctx, node );
-	 return;
-      }
-      
-      if (ctx->NewState)
-	 _mesa_update_state( ctx );
-
-      if ((ctx->VertexProgram.Enabled && !ctx->VertexProgram._Enabled) ||
-          (ctx->FragmentProgram.Enabled && !ctx->FragmentProgram._Enabled)) {
-         _mesa_error(ctx, GL_INVALID_OPERATION,
-                     "glBegin (invalid vertex/fragment program)");
-         return;
-      }
-
-      brw_bind_vertex_list( save, node );
-
-      if (!brw_draw_prims( save->ctx, 
-			   save->inputs, 
-			   node->prim, 
-			   node->prim_count,
-			   NULL,
-			   0,	/* Node is a VBO, so this is ok */
-			   node->count,
-			   0 )) {
-	 brw_fallback(ctx);
-	 brw_save_loopback_vertex_list( ctx, node );
-	 brw_unfallback(ctx);
-	 return;
-      }
-   }
-
-   /* Copy to current?
-    */
-   _playback_copy_to_current( ctx, node );
-}
diff --git a/src/mesa/drivers/dri/i965/brw_vs.c b/src/mesa/drivers/dri/i965/brw_vs.c
index e5a28b96e32..74c9d88e464 100644
--- a/src/mesa/drivers/dri/i965/brw_vs.c
+++ b/src/mesa/drivers/dri/i965/brw_vs.c
@@ -54,12 +54,11 @@ static void do_vs_prog( struct brw_context *brw,
    c.vp = vp;
 
    c.prog_data.outputs_written = vp->program.Base.OutputsWritten;
-   c.prog_data.inputs_read = brw_translate_inputs(brw->intel.ctx.VertexProgram._Enabled,
-						  vp->program.Base.InputsRead);
+   c.prog_data.inputs_read = vp->program.Base.InputsRead;
 
    if (c.key.copy_edgeflag) {
       c.prog_data.outputs_written |= 1<<VERT_RESULT_EDGE;
-      c.prog_data.inputs_read |= 1<<BRW_ATTRIB_EDGEFLAG;
+      c.prog_data.inputs_read |= 1<<VERT_ATTRIB_EDGEFLAG;
    }
 
    if (0)
diff --git a/src/mesa/drivers/dri/i965/brw_vs_constval.c b/src/mesa/drivers/dri/i965/brw_vs_constval.c
index 502d8283970..528e164db8c 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_constval.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_constval.c
@@ -162,6 +162,7 @@ static GLuint get_input_size(struct brw_context *brw,
    GLuint sizes_dword = brw->vb.info.sizes[attr/16];
    GLuint sizes_bits = (sizes_dword>>((attr%16)*2)) & 0x3;
    return sizes_bits + 1;
+/*    return brw->vb.inputs[attr].glarray->Size; */
 }
 
 /* Calculate sizes of vertex program outputs.  Size is the largest
@@ -176,8 +177,6 @@ static void calc_wm_input_sizes( struct brw_context *brw )
    struct tracker t;
    GLuint insn;
    GLuint i;
-   GLuint64EXT inputs = brw_translate_inputs(brw->intel.ctx.VertexProgram._Enabled,
-					     vp->program.Base.InputsRead);
 
    memset(&t, 0, sizeof(t));
 
@@ -185,8 +184,8 @@ static void calc_wm_input_sizes( struct brw_context *brw )
    if (brw->attribs.Light->Model.TwoSide)
       t.twoside = 1;
 
-   for (i = 0; i < BRW_ATTRIB_MAX; i++) 
-      if (inputs & (1<<i))
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) 
+      if (vp->program.Base.InputsRead & (1<<i))
 	 set_active_component(&t, PROGRAM_INPUT, i, 
 			      szflag[get_input_size(brw, i)]);
       
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 8403e1bd7b6..ffdb843e85f 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -77,8 +77,8 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
    /* Allocate input regs:  
     */
    c->nr_inputs = 0;
-   for (i = 0; i < BRW_ATTRIB_MAX; i++) {
-      if (c->prog_data.inputs_read & ((GLuint64EXT)1<<i)) {
+   for (i = 0; i < VERT_ATTRIB_MAX; i++) {
+      if (c->prog_data.inputs_read & (1<<i)) {
 	 c->nr_inputs++;
 	 c->regs[PROGRAM_INPUT][i] = brw_vec8_grf(reg, 0);
 	 reg++;
@@ -791,7 +791,7 @@ static void emit_vertex_write( struct brw_vs_compile *c)
    if (c->key.copy_edgeflag) {
       brw_MOV(p, 
 	      get_reg(c, PROGRAM_OUTPUT, VERT_RESULT_EDGE),
-	      get_reg(c, PROGRAM_INPUT, BRW_ATTRIB_EDGEFLAG));
+	      get_reg(c, PROGRAM_INPUT, VERT_ATTRIB_EDGEFLAG));
    }
 
 
diff --git a/src/mesa/drivers/dri/i965/brw_vs_tnl.c b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
index 82c1958887f..0d61092247a 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_tnl.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_tnl.c
@@ -146,9 +146,13 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
       }
 
       /* BRW_NEW_INPUT_VARYING */
-      for (i = BRW_ATTRIB_MAT_FRONT_AMBIENT ; i < BRW_ATTRIB_MAX ; i++) 
-	 if (brw->vb.info.varying & ((GLuint64EXT)1<<i)) 
-	    key->light_material_mask |= 1<<(i-BRW_ATTRIB_MAT_FRONT_AMBIENT);
+
+      /* For these programs, material values are stuffed into the
+       * generic slots:
+       */
+      for (i = 0 ; i < MAT_ATTRIB_MAX ; i++) 
+	 if (brw->vb.info.varying & (1<<(VERT_ATTRIB_GENERIC0 + i))) 
+	    key->light_material_mask |= 1<<i;
 
       for (i = 0; i < MAX_LIGHTS; i++) {
 	 struct gl_light *light = &brw->attribs.Light->Light[i];
@@ -374,17 +378,10 @@ static void release_temps( struct tnl_program *p )
 
 static struct ureg register_input( struct tnl_program *p, GLuint input )
 {
-   GLuint orig_input = input;
-   /* Cram the material flags into the generic range.  We'll translate
-    * them back later.
-    */
-   if (input >= BRW_ATTRIB_MAT_FRONT_AMBIENT)
-      input -= BRW_ATTRIB_MAT_FRONT_AMBIENT - BRW_ATTRIB_GENERIC0;
-
    assert(input < 32);
 
    p->program->Base.InputsRead |= (1<<input);
-   return make_ureg(PROGRAM_INPUT, orig_input);
+   return make_ureg(PROGRAM_INPUT, input);
 }
 
 static struct ureg register_output( struct tnl_program *p, GLuint output )
@@ -647,7 +644,7 @@ static void emit_passthrough( struct tnl_program *p,
 static struct ureg get_eye_position( struct tnl_program *p )
 {
    if (is_undef(p->eye_position)) {
-      struct ureg pos = register_input( p, BRW_ATTRIB_POS ); 
+      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
       struct ureg modelview[4];
 
       p->eye_position = reserve_temp(p);
@@ -710,7 +707,7 @@ static struct ureg get_eye_position_normalized( struct tnl_program *p )
 static struct ureg get_eye_normal( struct tnl_program *p )
 {
    if (is_undef(p->eye_normal)) {
-      struct ureg normal = register_input(p, BRW_ATTRIB_NORMAL );
+      struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
       struct ureg mvinv[3];
 
       register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 2,
@@ -743,7 +740,7 @@ static struct ureg get_eye_normal( struct tnl_program *p )
 
 static void build_hpos( struct tnl_program *p )
 {
-   struct ureg pos = register_input( p, BRW_ATTRIB_POS ); 
+   struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
    struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
    struct ureg mvp[4];
 
@@ -787,9 +784,9 @@ static struct ureg get_material( struct tnl_program *p, GLuint side,
    GLuint attrib = material_attrib(side, property);
 
    if (p->color_materials & (1<<attrib))
-      return register_input(p, BRW_ATTRIB_COLOR0);
+      return register_input(p, VERT_ATTRIB_COLOR0);
    else if (p->materials & (1<<attrib)) 
-      return register_input( p, attrib + BRW_ATTRIB_MAT_FRONT_AMBIENT );
+      return register_input( p, attrib + _TNL_ATTRIB_MAT_FRONT_AMBIENT );
    else
       return register_param3( p, STATE_MATERIAL, side, property );
 }
@@ -1157,7 +1154,7 @@ static void build_fog( struct tnl_program *p )
       input = swizzle1(get_eye_position(p), Z);
    }
    else {
-      input = swizzle1(register_input(p, BRW_ATTRIB_FOG), X);
+      input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
    }
 
    if (p->state->fog_option &&
@@ -1299,7 +1296,7 @@ static void build_texture_transform( struct tnl_program *p )
 	    for (j = 0; j < 4; j++) {
 	       switch (modes[j]) {
 	       case TXG_OBJ_LINEAR: {
-		  struct ureg obj = register_input(p, BRW_ATTRIB_POS);
+		  struct ureg obj = register_input(p, VERT_ATTRIB_POS);
 		  struct ureg plane = 
 		     register_param3(p, STATE_TEXGEN, i,
 				     STATE_TEXGEN_OBJECT_S + j);
@@ -1348,7 +1345,7 @@ static void build_texture_transform( struct tnl_program *p )
 	    }
 
 	    if (copy_mask) {
-	       struct ureg in = register_input(p, BRW_ATTRIB_TEX0+i);
+	       struct ureg in = register_input(p, VERT_ATTRIB_TEX0+i);
 	       emit_op1(p, OPCODE_MOV, out_texgen, copy_mask, in );
 	    }
 	 }
@@ -1357,7 +1354,7 @@ static void build_texture_transform( struct tnl_program *p )
 	    struct ureg texmat[4];
 	    struct ureg in = (!is_undef(out_texgen) ? 
 			      out_texgen : 
-			      register_input(p, BRW_ATTRIB_TEX0+i));
+			      register_input(p, VERT_ATTRIB_TEX0+i));
 	    if (PREFER_DP4) {
 	       register_matrix_param6( p, STATE_MATRIX, STATE_TEXTURE, i, 
 				       0, 3, STATE_MATRIX, texmat );
@@ -1373,7 +1370,7 @@ static void build_texture_transform( struct tnl_program *p )
 	 release_temps(p);
       } 
       else {
-	 emit_passthrough(p, BRW_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
+	 emit_passthrough(p, VERT_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
       }
    }
 }
@@ -1425,10 +1422,10 @@ static void build_tnl_program( struct tnl_program *p )
 	 build_lighting(p);
       else {
 	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
-	    emit_passthrough(p, BRW_ATTRIB_COLOR0, VERT_RESULT_COL0);
+	    emit_passthrough(p, VERT_ATTRIB_COLOR0, VERT_RESULT_COL0);
 
 	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL1)
-	    emit_passthrough(p, BRW_ATTRIB_COLOR1, VERT_RESULT_COL1);
+	    emit_passthrough(p, VERT_ATTRIB_COLOR1, VERT_RESULT_COL1);
       }
    }
 
diff --git a/src/mesa/drivers/dri/i965/brw_vtbl.c b/src/mesa/drivers/dri/i965/brw_vtbl.c
index ac09754e3ae..786f30e641e 100644
--- a/src/mesa/drivers/dri/i965/brw_vtbl.c
+++ b/src/mesa/drivers/dri/i965/brw_vtbl.c
@@ -46,8 +46,6 @@
 #include "brw_state.h"
 
 #include "brw_draw.h"
-#include "brw_exec.h"
-#include "brw_save.h"
 #include "brw_state.h"
 #include "brw_aub.h"
 #include "brw_fallback.h"
@@ -68,9 +66,6 @@ static void brw_destroy_context( struct intel_context *intel )
    brw_destroy_state(brw);
    brw_draw_destroy( brw );
 
-   brw_exec_destroy( ctx );
-   brw_save_destroy( ctx );
-
    brw_ProgramCacheDestroy( ctx );
    brw_FrameBufferTexDestroy( brw );
 }
@@ -166,10 +161,7 @@ static GLuint brw_flush_cmd( void )
 
 static void brw_invalidate_state( struct intel_context *intel, GLuint new_state )
 {
-   GLcontext *ctx = &intel->ctx;
-
-   brw_exec_invalidate_state(ctx, new_state);
-   brw_save_invalidate_state(ctx, new_state);
+   /* nothing */
 }
 
 
diff --git a/src/mesa/drivers/dri/i965/intel_context.c b/src/mesa/drivers/dri/i965/intel_context.c
index fea6e44f0b5..10eb9a2e28d 100644
--- a/src/mesa/drivers/dri/i965/intel_context.c
+++ b/src/mesa/drivers/dri/i965/intel_context.c
@@ -38,7 +38,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 #include "tnl/t_vertex.h"
@@ -61,6 +61,7 @@
 #include "bufmgr.h"
 
 #include "utils.h"
+#include "vblank.h"
 #ifndef INTEL_DEBUG
 int INTEL_DEBUG = (0);
 #endif
@@ -86,11 +87,6 @@ int INTEL_DEBUG = (0);
 int VERBOSE = 0;
 #endif
 
-#if DEBUG_LOCKING
-char *prevLockFile;
-int prevLockLine;
-#endif
-
 /***************************************
  * Mesa's Driver Functions
  ***************************************/
@@ -230,7 +226,7 @@ static void intelInvalidateState( GLcontext *ctx, GLuint new_state )
 
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    _tnl_invalidate_vertex_state( ctx, new_state );
    
@@ -260,11 +256,11 @@ intelBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
 {
 	struct intel_context *intel = intel_context( ctx );
 	drmI830MMIO io = {
-		.read_write = MMIO_WRITE,
+		.read_write = MMIO_READ,
 		.reg = MMIO_REGS_PS_DEPTH_COUNT,
 		.data = &q->Result 
 	};
-	intel->stats_wm = GL_TRUE;
+	intel->stats_wm++;
 	intelFinish(&intel->ctx);
 	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
 }
@@ -283,7 +279,7 @@ intelEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
 	drmCommandRead(intel->driFd, DRM_I830_MMIO, &io, sizeof(io));
 	q->Result = tmp - q->Result;
 	q->Ready = GL_TRUE;
-	intel->stats_wm = GL_FALSE;
+	intel->stats_wm--;
 }
 
 
@@ -339,6 +335,11 @@ GLboolean intelInitContext( struct intel_context *intel,
    intel->driScreen = sPriv;
    intel->sarea = saPriv;
 
+   driParseConfigFiles (&intel->optionCache, &intelScreen->optionCache,
+		   intel->driScreen->myNum, "i965");
+
+   intel->vblank_flags = (intel->intelScreen->irq_active != 0)
+	   ? driGetDefaultVBlankFlags(&intel->optionCache) : VBLANK_FLAG_NO_IRQ;
 
    ctx->Const.MaxTextureMaxAnisotropy = 2.0;
 
@@ -374,7 +375,7 @@ GLboolean intelInitContext( struct intel_context *intel,
 
    /* Initialize the software rasterizer and helper modules. */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -448,8 +449,8 @@ GLboolean intelInitContext( struct intel_context *intel,
 				 intelScreen->cpp,
 				 intelScreen->front.pitch / intelScreen->cpp,
 				 intelScreen->height,
-				 intelScreen->front.tiled != 0); /* 0: LINEAR */
-
+				 intelScreen->front.size,
+				 intelScreen->front.tiled != 0);
 
    intel->back_region = 
       intel_region_create_static(intel,
@@ -459,6 +460,7 @@ GLboolean intelInitContext( struct intel_context *intel,
 				 intelScreen->cpp,
 				 intelScreen->back.pitch / intelScreen->cpp,
 				 intelScreen->height,
+				 intelScreen->back.size,
                                  intelScreen->back.tiled != 0);
 
    /* Still assuming front.cpp == depth.cpp
@@ -475,6 +477,7 @@ GLboolean intelInitContext( struct intel_context *intel,
 				 intelScreen->cpp,
 				 intelScreen->depth.pitch / intelScreen->cpp,
 				 intelScreen->height,
+				 intelScreen->depth.size,
                                  intelScreen->depth.tiled != 0);
    
    intel_bufferobj_init( intel );
@@ -517,7 +520,7 @@ void intelDestroyContext(__DRIcontextPrivate *driContextPriv)
       release_texture_heaps = (intel->ctx.Shared->RefCount == 1);
       _swsetup_DestroyContext (&intel->ctx);
       _tnl_DestroyContext (&intel->ctx);
-      _ac_DestroyContext (&intel->ctx);
+      _vbo_DestroyContext (&intel->ctx);
 
       _swrast_DestroyContext (&intel->ctx);
       intel->Fallback = 0;	/* don't call _swrast_Flush later */
@@ -566,6 +569,9 @@ GLboolean intelMakeCurrent(__DRIcontextPrivate *driContextPriv,
 
       if ( intel->driDrawable != driDrawPriv ) {
 	 /* Shouldn't the readbuffer be stored also? */
+	 driDrawableInitVBlank( driDrawPriv, intel->vblank_flags,
+		      &intel->vbl_seq );
+
 	 intel->driDrawable = driDrawPriv;
 	 intelWindowMoved( intel );
       }
diff --git a/src/mesa/drivers/dri/i965/intel_context.h b/src/mesa/drivers/dri/i965/intel_context.h
index 74a795c96d2..808512f7fd5 100644
--- a/src/mesa/drivers/dri/i965/intel_context.h
+++ b/src/mesa/drivers/dri/i965/intel_context.h
@@ -177,7 +177,7 @@ struct intel_context
    GLuint second_last_swap_fence;
    
    GLboolean aub_wrap;
-   GLboolean stats_wm;
+   GLuint stats_wm;
 
    struct intel_batchbuffer *batch;
 
diff --git a/src/mesa/drivers/dri/i965/intel_ioctl.c b/src/mesa/drivers/dri/i965/intel_ioctl.c
index 4da31277ea2..0a8e976f706 100644
--- a/src/mesa/drivers/dri/i965/intel_ioctl.c
+++ b/src/mesa/drivers/dri/i965/intel_ioctl.c
@@ -108,7 +108,8 @@ void intelWaitIrq( struct intel_context *intel, int seq )
 
 	 /* This seems quite often to return before it should!?! 
 	  */
-      } while (ret == -EAGAIN || ret == -EINTR || (ret == -EBUSY && lastdispatch != intel->sarea->last_dispatch) || (ret == 0 && seq > intel->sarea->last_dispatch));
+      } while (ret == -EAGAIN || ret == -EINTR || (ret == -EBUSY && lastdispatch != intel->sarea->last_dispatch) || (ret == 0 && seq > intel->sarea->last_dispatch)
+	       || (ret == 0 && intel->sarea->last_dispatch - seq >= (1 << 24)));
       
 
       if ( ret ) {
diff --git a/src/mesa/drivers/dri/i965/intel_regions.c b/src/mesa/drivers/dri/i965/intel_regions.c
index 398b0a0a3b5..835ecdd7257 100644
--- a/src/mesa/drivers/dri/i965/intel_regions.c
+++ b/src/mesa/drivers/dri/i965/intel_regions.c
@@ -122,10 +122,10 @@ struct intel_region *intel_region_create_static( struct intel_context *intel,
 						 GLuint cpp,
 						 GLuint pitch, 
 						 GLuint height,
+						 GLuint size,
 						 GLboolean tiled )
 {
    struct intel_region *region = calloc(sizeof(*region), 1);
-   GLuint size = cpp * pitch * height;
    GLint pool;
 
    DBG("%s\n", __FUNCTION__);
diff --git a/src/mesa/drivers/dri/i965/intel_regions.h b/src/mesa/drivers/dri/i965/intel_regions.h
index 2413f0de33c..d2235f1275b 100644
--- a/src/mesa/drivers/dri/i965/intel_regions.h
+++ b/src/mesa/drivers/dri/i965/intel_regions.h
@@ -78,6 +78,7 @@ struct intel_region *intel_region_create_static( struct intel_context *intel,
 						 GLuint cpp,
 						 GLuint pitch,
 						 GLuint height,
+						 GLuint size,
 						 GLboolean tiled );
 
 /* Map/unmap regions.  This is refcounted also: 
diff --git a/src/mesa/drivers/dri/mach64/mach64_context.c b/src/mesa/drivers/dri/mach64/mach64_context.c
index 2ab1cf2617e..5a6c301da2d 100644
--- a/src/mesa/drivers/dri/mach64/mach64_context.c
+++ b/src/mesa/drivers/dri/mach64/mach64_context.c
@@ -38,7 +38,7 @@
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -225,7 +225,7 @@ GLboolean mach64CreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -278,7 +278,7 @@ void mach64DestroyContext( __DRIcontextPrivate *driContextPriv  )
 
       _swsetup_DestroyContext( mmesa->glCtx );
       _tnl_DestroyContext( mmesa->glCtx );
-      _ac_DestroyContext( mmesa->glCtx );
+      _vbo_DestroyContext( mmesa->glCtx );
       _swrast_DestroyContext( mmesa->glCtx );
 
       if (release_texture_heaps) {
diff --git a/src/mesa/drivers/dri/mach64/mach64_state.c b/src/mesa/drivers/dri/mach64/mach64_state.c
index d4804a2c550..667a3945206 100644
--- a/src/mesa/drivers/dri/mach64/mach64_state.c
+++ b/src/mesa/drivers/dri/mach64/mach64_state.c
@@ -40,7 +40,7 @@
 #include "enums.h"
 #include "colormac.h"
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -1023,7 +1023,7 @@ static void mach64DDInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    MACH64_CONTEXT(ctx)->NewGLState |= new_state;
 }
diff --git a/src/mesa/drivers/dri/mga/mga_xmesa.c b/src/mesa/drivers/dri/mga/mga_xmesa.c
index f02913c6af2..f024f73eb6a 100644
--- a/src/mesa/drivers/dri/mga/mga_xmesa.c
+++ b/src/mesa/drivers/dri/mga/mga_xmesa.c
@@ -45,7 +45,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 
@@ -593,7 +593,7 @@ mgaCreateContext( const __GLcontextModes *mesaVis,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    
    _swsetup_CreateContext( ctx );
@@ -676,7 +676,7 @@ mgaDestroyContext(__DRIcontextPrivate *driContextPriv)
       release_texture_heaps = (mmesa->glCtx->Shared->RefCount == 1);
       _swsetup_DestroyContext( mmesa->glCtx );
       _tnl_DestroyContext( mmesa->glCtx );
-      _ac_DestroyContext( mmesa->glCtx );
+      _vbo_DestroyContext( mmesa->glCtx );
       _swrast_DestroyContext( mmesa->glCtx );
 
       mgaFreeVB( mmesa->glCtx );
diff --git a/src/mesa/drivers/dri/mga/mgastate.c b/src/mesa/drivers/dri/mga/mgastate.c
index 075fd348cf4..c20a76f29ef 100644
--- a/src/mesa/drivers/dri/mga/mgastate.c
+++ b/src/mesa/drivers/dri/mga/mgastate.c
@@ -41,7 +41,7 @@
 #include "mgaregs.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
@@ -1042,7 +1042,7 @@ static void mgaDDInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    MGA_CONTEXT(ctx)->NewGLState |= new_state;
 }
diff --git a/src/mesa/drivers/dri/nouveau/Makefile b/src/mesa/drivers/dri/nouveau/Makefile
index d31b42a568b..20d2de5eefb 100644
--- a/src/mesa/drivers/dri/nouveau/Makefile
+++ b/src/mesa/drivers/dri/nouveau/Makefile
@@ -19,23 +19,27 @@ DRIVER_SOURCES = \
 	nouveau_screen.c         \
 	nouveau_span.c           \
 	nouveau_state.c          \
+	nouveau_state_cache.c    \
 	nouveau_shader.c         \
-	nouveau_shader_0_arb.c   \
+	nouveau_shader_0.c       \
 	nouveau_shader_1.c       \
 	nouveau_shader_2.c       \
 	nouveau_tex.c            \
 	nouveau_swtcl.c          \
 	nouveau_sync.c           \
-	nv10_swtcl.c             \
+	nouveau_query.c          \
+	nv04_state.c             \
+	nv04_swtcl.c             \
 	nv10_state.c             \
+	nv10_swtcl.c             \
 	nv20_state.c             \
-	nv30_state.c             \
-	nouveau_state_cache.c    \
 	nv20_vertprog.c          \
+	nv30_state.c             \
 	nv30_fragprog.c          \
 	nv30_vertprog.c          \
 	nv40_fragprog.c          \
-	nv40_vertprog.c
+	nv40_vertprog.c          \
+	nv50_state.c
 
 C_SOURCES = \
 	$(COMMON_SOURCES) \
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
index d36196aeef2..684ed7b017d 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.c
@@ -8,29 +8,458 @@
 #include "nouveau_object.h"
 #include "nouveau_msg.h"
 
+#define NOUVEAU_MEM_FREE(mem) do {      \
+	nouveau_mem_free(ctx, (mem));   \
+	(mem) = NULL;                   \
+} while(0)
+
 #define DEBUG(fmt,args...) do {                \
 	if (NOUVEAU_DEBUG & DEBUG_BUFFEROBJ) { \
 		fprintf(stderr, "%s: "fmt, __func__, ##args);  \
 	}                                      \
 } while(0)
 
-/* Wrapper for nouveau_mem_gpu_offset_get() that marks the bufferobj dirty
- * if the GPU modifies the data.
- */
+static GLboolean
+nouveau_bo_download_from_screen(GLcontext *ctx,	GLuint offset, GLuint size,
+						struct gl_buffer_object *bo)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_mem *in_mem;
+
+	DEBUG("bo=%p, offset=%d, size=%d\n", bo, offset, size);
+
+	/* If there's a permanent backing store, blit directly into it */
+	if (nbo->cpu_mem) {
+		if (nbo->cpu_mem != nbo->gpu_mem) {
+			DEBUG("..cpu_mem\n");
+			nouveau_memformat_flat_emit(ctx, nbo->cpu_mem,
+						    nbo->gpu_mem,
+						    offset, offset, size);
+		}
+	} else {
+		DEBUG("..sys_mem\n");
+		in_mem = nouveau_mem_alloc(ctx, NOUVEAU_MEM_AGP, size, 0);
+		if (in_mem) {
+			DEBUG("....via AGP\n");
+			/* otherwise, try blitting to faster memory and
+			 * copying from there
+			 */
+			nouveau_memformat_flat_emit(ctx, in_mem, nbo->gpu_mem,
+							 0, offset, size);
+			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
+						       NvSubMemFormat);
+			_mesa_memcpy(nbo->cpu_mem_sys + offset,
+					in_mem->map, size);
+			NOUVEAU_MEM_FREE(in_mem);
+		} else {
+			DEBUG("....direct VRAM copy\n");
+			/* worst case, copy directly from vram */
+			_mesa_memcpy(nbo->cpu_mem_sys + offset,
+				     nbo->gpu_mem + offset,
+				     size);
+		}
+	}
+
+	return GL_TRUE;
+}
+
+static GLboolean
+nouveau_bo_upload_to_screen(GLcontext *ctx, GLuint offset, GLuint size,
+					    struct gl_buffer_object *bo)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_mem *out_mem;
+
+	DEBUG("bo=%p, offset=%d, size=%d\n", bo, offset, size);
+
+	if (nbo->cpu_mem) {
+		if (nbo->cpu_mem != nbo->gpu_mem) {
+			DEBUG("..cpu_mem\n");
+			nouveau_memformat_flat_emit(ctx, nbo->gpu_mem,
+						    nbo->cpu_mem,
+						    offset, offset, size);
+		}
+	} else {
+		out_mem = nouveau_mem_alloc(ctx, NOUVEAU_MEM_AGP |
+						 NOUVEAU_MEM_MAPPED,
+						 size, 0);
+		if (out_mem) {
+			DEBUG("....via AGP\n");
+			_mesa_memcpy(out_mem->map,
+					nbo->cpu_mem_sys + offset, size);
+			nouveau_memformat_flat_emit(ctx, nbo->gpu_mem, out_mem,
+						    offset, 0, size);
+			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
+						       NvSubMemFormat);
+			NOUVEAU_MEM_FREE(out_mem);
+		} else {
+			DEBUG("....direct VRAM copy\n");
+			_mesa_memcpy(nbo->gpu_mem->map + offset,
+				     nbo->cpu_mem_sys + offset,
+				     size);
+		}
+	}
+
+	return GL_TRUE;
+}
+
+GLboolean
+nouveau_bo_move_in(GLcontext *ctx, struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+
+	DEBUG("bo=%p\n", bo);
+
+	if (bo->OnCard)
+		return GL_TRUE;
+	assert(nbo->gpu_mem_flags);
+
+	nbo->gpu_mem = nouveau_mem_alloc(ctx, nbo->gpu_mem_flags |
+					      NOUVEAU_MEM_MAPPED,
+					      bo->Size, 0);
+	assert(nbo->gpu_mem);
+
+	if (nbo->cpu_mem_flags) {
+		if ((nbo->cpu_mem_flags|NOUVEAU_MEM_MAPPED) != nbo->gpu_mem->type) {
+			DEBUG("..need cpu_mem buffer\n");
+
+			nbo->cpu_mem = nouveau_mem_alloc(ctx,
+							 nbo->cpu_mem_flags |
+							 NOUVEAU_MEM_MAPPED,
+							 bo->Size, 0);
+
+			if (nbo->cpu_mem) {
+				DEBUG("....alloc ok, kill sys_mem buffer\n");
+				_mesa_memcpy(nbo->cpu_mem->map,
+					     nbo->cpu_mem_sys, bo->Size);
+				FREE(nbo->cpu_mem_sys);
+			}
+		} else {
+			DEBUG("..cpu direct access to GPU buffer\n");
+			nbo->cpu_mem = nbo->gpu_mem;
+		}
+	}
+	nouveau_bo_upload_to_screen(ctx, 0, bo->Size, bo);
+
+	bo->OnCard = GL_TRUE;
+	return GL_TRUE;
+}
+
+GLboolean
+nouveau_bo_move_out(GLcontext *ctx, struct gl_buffer_object *bo)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	GLuint nr_dirty;
+
+	DEBUG("bo=%p\n", bo);
+	if (!bo->OnCard)
+		return GL_TRUE;
+
+	nr_dirty = nouveau_bo_download_dirty(ctx, bo);
+	if (nbo->cpu_mem) {
+		if (nr_dirty && nbo->cpu_mem != nbo->gpu_mem)
+			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
+						       NvSubMemFormat);
+		DEBUG("..destroy cpu_mem buffer\n");
+		nbo->cpu_mem_sys = malloc(bo->Size);
+		assert(nbo->cpu_mem_sys);
+		_mesa_memcpy(nbo->cpu_mem_sys, nbo->cpu_mem->map, bo->Size);
+		if (nbo->cpu_mem == nbo->gpu_mem)
+			nbo->cpu_mem = NULL;
+		else
+			NOUVEAU_MEM_FREE(nbo->cpu_mem);
+	}
+	NOUVEAU_MEM_FREE(nbo->gpu_mem);
+
+	bo->OnCard = GL_FALSE;
+	return GL_TRUE;
+}
+
+static void
+nouveau_bo_choose_storage_method(GLcontext *ctx, GLenum usage,
+						 struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	GLuint gpu_type = 0;
+	GLuint cpu_type = 0;
+
+	switch (usage) {
+	/* Client source, changes often, used by GL many times */
+	case GL_DYNAMIC_DRAW_ARB:
+		gpu_type = NOUVEAU_MEM_AGP | NOUVEAU_MEM_FB_ACCEPTABLE;
+		cpu_type = NOUVEAU_MEM_AGP;
+		break;
+	/* GL source, changes often, client reads many times */
+	case GL_DYNAMIC_READ_ARB:
+	/* Client source, specified once, used by GL many times */
+	case GL_STATIC_DRAW_ARB:
+	/* GL source, specified once, client reads many times */
+	case GL_STATIC_READ_ARB:
+	/* Client source, specified once, used by GL a few times */
+	case GL_STREAM_DRAW_ARB:
+	/* GL source, specified once, client reads a few times */
+	case GL_STREAM_READ_ARB:
+	/* GL source, changes often, used by GL many times*/
+	case GL_DYNAMIC_COPY_ARB:
+	/* GL source, specified once, used by GL many times */
+	case GL_STATIC_COPY_ARB:
+	/* GL source, specified once, used by GL a few times */
+	case GL_STREAM_COPY_ARB:
+		gpu_type = NOUVEAU_MEM_FB;
+		break;
+	default: 
+		assert(0);
+	}
+
+	nbo->gpu_mem_flags = gpu_type;
+	nbo->cpu_mem_flags = cpu_type;
+	nbo->usage	   = usage;
+}
+
+void
+nouveau_bo_init_storage(GLcontext *ctx,	GLuint valid_gpu_access,
+					GLsizeiptrARB size,
+					const GLvoid *data,
+					GLenum usage,
+					struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+
+	DEBUG("bo=%p\n", bo);
+
+	/* Free up previous buffers if we can't reuse them */
+	if (nbo->usage != usage ||
+			(nbo->gpu_mem && (nbo->gpu_mem->size != size))) {
+		if (nbo->cpu_mem_sys)
+			FREE(nbo->cpu_mem_sys);
+		if (nbo->cpu_mem) {
+			if (nbo->cpu_mem != nbo->gpu_mem)
+				NOUVEAU_MEM_FREE(nbo->cpu_mem);
+			else
+				nbo->cpu_mem = NULL;
+		}
+		if (nbo->gpu_mem)
+			NOUVEAU_MEM_FREE(nbo->gpu_mem);
+
+		bo->OnCard = GL_FALSE;
+		nbo->cpu_mem_sys = calloc(1, size);
+	}
+
+	nouveau_bo_choose_storage_method(ctx, usage, bo);
+	/* Force off flags that may not be ok for a given buffer */
+	nbo->gpu_mem_flags &= valid_gpu_access;
+
+	bo->Usage  = usage;
+	bo->Size   = size;
+
+	if (data) {
+		GLvoid *map = nouveau_bo_map(ctx, GL_WRITE_ONLY_ARB, bo);
+		_mesa_memcpy(map, data, size);
+		nouveau_bo_dirty_all(ctx, GL_FALSE, bo);
+		nouveau_bo_unmap(ctx, bo);
+	}
+}
+
+void *
+nouveau_bo_map(GLcontext *ctx, GLenum access, struct gl_buffer_object *bo)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+
+	DEBUG("bo=%p, access=%s\n", bo, _mesa_lookup_enum_by_nr(access));
+
+	if (bo->OnCard && 
+		(access == GL_READ_ONLY_ARB || access == GL_READ_WRITE_ARB)) {
+		GLuint nr_dirty;
+
+		DEBUG("..on card\n");
+		nr_dirty = nouveau_bo_download_dirty(ctx, bo);
+
+		/* nouveau_bo_download_dirty won't wait unless it needs to
+		 * free a temp buffer, which isn't the case if cpu_mem is
+		 * present.
+		 */
+		if (nr_dirty && nbo->cpu_mem && nbo->cpu_mem != nbo->gpu_mem)
+			nouveau_notifier_wait_nop(ctx, nmesa->syncNotifier,
+						       NvSubMemFormat);
+	}
+
+	if (nbo->cpu_mem) {
+		DEBUG("..access via cpu_mem\n");
+		return nbo->cpu_mem->map;
+	} else {
+		DEBUG("..access via cpu_mem_sys\n");
+		return nbo->cpu_mem_sys;
+	}
+}
+
+void
+nouveau_bo_unmap(GLcontext *ctx, struct gl_buffer_object *bo)
+{
+	DEBUG("unmap bo=%p\n", bo);
+}
+
 uint32_t
-nouveau_bufferobj_gpu_ref(GLcontext *ctx, GLenum access,
-			  struct gl_buffer_object *obj)
+nouveau_bo_gpu_ref(GLcontext *ctx, struct gl_buffer_object *bo)
 {
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
 
-	DEBUG("obj=%p, access=%s\n", obj, _mesa_lookup_enum_by_nr(access));
+	assert(nbo->mapped == GL_FALSE);
 
-	if (access == GL_WRITE_ONLY_ARB || access == GL_READ_WRITE_ARB)
-		nbo->gpu_dirty = GL_TRUE;
+	DEBUG("gpu_ref\n");
+	
+	if (!bo->OnCard) {
+		nouveau_bo_move_in(ctx, bo);
+		bo->OnCard = GL_TRUE;
+	}
+	nouveau_bo_upload_dirty(ctx, bo);
 
 	return nouveau_mem_gpu_offset_get(ctx, nbo->gpu_mem);
 }
 
+void
+nouveau_bo_dirty_linear(GLcontext *ctx, GLboolean on_card,
+			uint32_t offset, uint32_t size,
+			struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_bufferobj_dirty *dirty;
+	uint32_t start = offset;
+	uint32_t end = offset + size;
+	int i;
+
+	if (nbo->cpu_mem == nbo->gpu_mem)
+		return;
+
+	dirty = on_card ? &nbo->gpu_dirty : &nbo->cpu_dirty;
+
+	DEBUG("on_card=%d, offset=%d, size=%d, bo=%p\n",
+			on_card, offset, size, bo);
+
+	for (i=0; i<dirty->nr_dirty; i++) {
+		nouveau_bufferobj_region *r = &dirty->dirty[i];
+
+		/* already dirty */
+		if (start >= r->start && end <= r->end) {
+			DEBUG("..already dirty\n");
+			return;
+		}
+
+		/* add to the end of a region */
+		if (start >= r->start && start <= r->end) {
+			if (end > r->end) {
+				DEBUG("..extend end of region\n");
+				r->end = end;
+				return;
+			}
+		}
+
+		/* add to the start of a region */
+		if (start < r->start && end >= r->end) {
+			DEBUG("..extend start of region\n");
+			r->start = start;
+			/* .. and to the end */
+			if (end > r->end) {
+				DEBUG("....and end\n");
+				r->end = end;
+			}
+			return;
+		}
+	}
+
+	/* new region */
+	DEBUG("..new dirty\n");
+	dirty->nr_dirty++;
+	dirty->dirty = realloc(dirty->dirty,
+			       sizeof(nouveau_bufferobj_region) *
+			       dirty->nr_dirty);
+	dirty->dirty[dirty->nr_dirty - 1].start = start;
+	dirty->dirty[dirty->nr_dirty - 1].end   = end;
+}
+
+void
+nouveau_bo_dirty_all(GLcontext *ctx, GLboolean on_card,
+		     struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_bufferobj_dirty *dirty;
+
+	dirty = on_card ? &nbo->gpu_dirty : &nbo->cpu_dirty;
+	
+	DEBUG("dirty all\n");
+	if (dirty->nr_dirty) {
+		FREE(dirty->dirty);
+		dirty->dirty    = NULL;
+		dirty->nr_dirty = 0;
+	}
+
+	nouveau_bo_dirty_linear(ctx, on_card, 0, bo->Size, bo);
+}
+
+GLuint
+nouveau_bo_upload_dirty(GLcontext *ctx, struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_bufferobj_dirty *dirty = &nbo->cpu_dirty;
+	GLuint nr_dirty;
+	int i;
+
+	nr_dirty = dirty->nr_dirty;
+	if (!nr_dirty) {
+		DEBUG("clean\n");
+		return nr_dirty;
+	}
+
+	for (i=0; i<nr_dirty; i++) {
+		nouveau_bufferobj_region *r = &dirty->dirty[i];
+
+		DEBUG("dirty %d: o=0x%08x, s=0x%08x\n",
+				i, r->start, r->end - r->start);
+		nouveau_bo_upload_to_screen(ctx,
+					    r->start, r->end - r->start, bo);
+	}
+
+	FREE(dirty->dirty);
+	dirty->dirty    = NULL;
+	dirty->nr_dirty = 0;
+
+	return nr_dirty;
+}
+
+GLuint
+nouveau_bo_download_dirty(GLcontext *ctx, struct gl_buffer_object *bo)
+{
+	nouveau_buffer_object *nbo = (nouveau_buffer_object *)bo;
+	nouveau_bufferobj_dirty *dirty = &nbo->gpu_dirty;
+	GLuint nr_dirty;
+	int i;
+
+	nr_dirty = dirty->nr_dirty;
+	if (nr_dirty) {
+		DEBUG("clean\n");
+		return nr_dirty;
+	}
+	
+	for (i=0; i<nr_dirty; i++) {
+		nouveau_bufferobj_region *r = &dirty->dirty[i];
+
+		DEBUG("dirty %d: o=0x%08x, s=0x%08x\n",
+				i, r->start, r->end - r->start);
+		nouveau_bo_download_from_screen(ctx,
+						r->start,
+						r->end - r->start, bo);
+	}
+
+	FREE(dirty->dirty);
+	dirty->dirty    = NULL;
+	dirty->nr_dirty = 0;
+
+	return nr_dirty;
+}
+
 static void
 nouveauBindBuffer(GLcontext *ctx, GLenum target, struct gl_buffer_object *obj)
 {
@@ -42,10 +471,11 @@ nouveauNewBufferObject(GLcontext *ctx, GLuint buffer, GLenum target)
 	nouveau_buffer_object *nbo;
 
 	nbo = CALLOC_STRUCT(nouveau_buffer_object_t);
-	DEBUG("name=0x%08x, target=%s, obj=%p\n",
-			buffer, _mesa_lookup_enum_by_nr(target), nbo);
-	_mesa_initialize_buffer_object(&nbo->mesa, buffer, target);
-	return &nbo->mesa;
+	if (nbo)
+		_mesa_initialize_buffer_object(&nbo->mesa, buffer, target);
+	DEBUG("bo=%p\n", nbo);
+
+	return nbo ? &nbo->mesa : NULL;
 }
 
 static void
@@ -53,11 +483,13 @@ nouveauDeleteBuffer(GLcontext *ctx, struct gl_buffer_object *obj)
 {
 	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
 
-	DEBUG("obj=%p\n", obj);
+	if (nbo->gpu_dirty.nr_dirty)
+		FREE(nbo->gpu_dirty.dirty);
+	if (nbo->cpu_dirty.nr_dirty)
+		FREE(nbo->cpu_dirty.dirty);
+	if (nbo->cpu_mem) nouveau_mem_free(ctx, nbo->cpu_mem);
+	if (nbo->gpu_mem) nouveau_mem_free(ctx, nbo->gpu_mem);
 
-	if (nbo->gpu_mem) {
-		nouveau_mem_free(ctx, nbo->gpu_mem);
-	}
 	_mesa_delete_buffer_object(ctx, obj);
 }
 
@@ -66,193 +498,105 @@ nouveauBufferData(GLcontext *ctx, GLenum target, GLsizeiptrARB size,
 		  const GLvoid *data, GLenum usage,
 		  struct gl_buffer_object *obj)
 {
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
+	GLuint gpu_flags;
 
-	DEBUG("obj=%p, target=%s, usage=%s, size=%d, data=%p\n",
-			obj,
+	DEBUG("target=%s, size=%d, data=%p, usage=%s, obj=%p\n",
 			_mesa_lookup_enum_by_nr(target),
+			(GLuint)size, data,
 			_mesa_lookup_enum_by_nr(usage),
-			(unsigned int)size,
-			data);
-
-	if (nbo->gpu_mem && nbo->gpu_mem->size != size)
-		nouveau_mem_free(ctx, nbo->gpu_mem);
-
-	/* Always have the GPU access the data from VRAM if possible.  For
-	 * some "usage" values it may be better from AGP be default?
-	 *
-	 * TODO: At some point we should drop the NOUVEAU_MEM_MAPPED flag.
-	 * TODO: Use the NOUVEAU_MEM_AGP_ACCEPTABLE flag.
-	 * TODO: What about PCI-E and shared system memory?
-	 */
-	if (!nbo->gpu_mem)
-		nbo->gpu_mem = nouveau_mem_alloc(ctx,
-						 NOUVEAU_MEM_FB |
-						 NOUVEAU_MEM_MAPPED,
-						 size,
-						 0);
-
-	if (!nbo->gpu_mem) {
-		MESSAGE("AIII bufferobj malloc failed\n");
-		return;
+			obj);
+
+	switch (target) {
+	case GL_ELEMENT_ARRAY_BUFFER_ARB:
+		gpu_flags = 0;
+		break;
+	default:
+		gpu_flags = NOUVEAU_BO_VRAM_OK | NOUVEAU_BO_AGP_OK;
+		break;
 	}
-
-	obj->Usage = usage;
-	obj->Size  = size;
-	if (!data)
-		return;
-
-	ctx->Driver.MapBuffer(ctx, target, GL_WRITE_ONLY_ARB, obj);
-	_mesa_memcpy(nbo->cpu_mem->map, data, size);
-	ctx->Driver.UnmapBuffer(ctx, target, obj);
+	nouveau_bo_init_storage(ctx, gpu_flags, size, data, usage, obj);
 }
 
-/*TODO: we don't need to DMA the entire buffer like MapBuffer does.. */
 static void
 nouveauBufferSubData(GLcontext *ctx, GLenum target, GLintptrARB offset,
 		     GLsizeiptrARB size, const GLvoid *data,
 		     struct gl_buffer_object *obj)
 {
-	DEBUG("obj=%p, target=%s, offset=0x%x, size=%d, data=%p\n",
-			obj,
+	GLvoid *out;
+
+	DEBUG("target=%s, offset=0x%x, size=%d, data=%p, obj=%p\n",
 			_mesa_lookup_enum_by_nr(target),
-			(unsigned int)offset,
-			(unsigned int)size,
-			data);
+			(GLuint)offset, (GLuint)size, data, obj);
 
-	ctx->Driver.MapBuffer(ctx, target, GL_WRITE_ONLY_ARB, obj);
-	_mesa_memcpy((GLubyte *)obj->Pointer + offset, data, size);
-	ctx->Driver.UnmapBuffer(ctx, target, obj);
+	out = nouveau_bo_map(ctx, GL_WRITE_ONLY_ARB, obj);
+	_mesa_memcpy(out + offset, data, size);
+	nouveau_bo_dirty_linear(ctx, GL_FALSE, offset, size, obj);
+	nouveau_bo_unmap(ctx, obj);
 }
 
-/*TODO: we don't need to DMA the entire buffer like MapBuffer does.. */
 static void
 nouveauGetBufferSubData(GLcontext *ctx, GLenum target, GLintptrARB offset,
 		     GLsizeiptrARB size, GLvoid *data,
 		     struct gl_buffer_object *obj)
 {
-	DEBUG("obj=%p, target=%s, offset=0x%x, size=%d, data=%p\n",
-			obj,
+	const GLvoid *in;
+
+	DEBUG("target=%s, offset=0x%x, size=%d, data=%p, obj=%p\n",
 			_mesa_lookup_enum_by_nr(target),
-			(unsigned int)offset,
-			(unsigned int)size,
-			data);
+			(GLuint)offset, (GLuint)size, data, obj);
 
-	ctx->Driver.MapBuffer(ctx, target, GL_READ_ONLY_ARB, obj);
-	_mesa_memcpy(data, (GLubyte *)obj->Pointer + offset, size);
-	ctx->Driver.UnmapBuffer(ctx, target, obj);
+	in = nouveau_bo_map(ctx, GL_READ_ONLY_ARB, obj);
+	_mesa_memcpy(data, in + offset, size);
+	nouveau_bo_unmap(ctx, obj);
 }
 
 static void *
 nouveauMapBuffer(GLcontext *ctx, GLenum target, GLenum access,
 		 struct gl_buffer_object *obj)
 {
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
-
-	DEBUG("obj=%p, target=%s, access=%s\n",
-			obj,
+	DEBUG("target=%s, access=%s, obj=%p\n",
 			_mesa_lookup_enum_by_nr(target),
-			_mesa_lookup_enum_by_nr(access));
+			_mesa_lookup_enum_by_nr(access),
+			obj
+			);
 
-	if (obj->Pointer) {
-		DEBUG("already mapped, return NULL\n");
+	/* Already mapped.. */
+	if (obj->Pointer)
 		return NULL;
-	}
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
-	/* If GPU is accessing the data from VRAM, copy to faster AGP memory
-	 * before CPU access to the buffer.
+	/* Have to pass READ_WRITE here, nouveau_bo_map will only ensure that
+	 * the cpu_mem buffer is up-to-date if we ask for read access.
+	 *
+	 * However, even if the client only asks for write access, we're still
+	 * forced to reupload the entire buffer.  So, we need the cpu_mem buffer
+	 * to have the correct data all the time.
 	 */
-	if (nbo->gpu_mem->type & NOUVEAU_MEM_FB) {
-		DEBUG("Data in VRAM, copying to AGP for CPU access\n");
-
-		/* This can happen if BufferData grows the GPU-access buffer */
-		if (nbo->cpu_mem && nbo->cpu_mem->size != nbo->gpu_mem->size) {
-			nouveau_mem_free(ctx, nbo->cpu_mem);
-			nbo->cpu_mem = NULL;
-		}
-
-		if (!nbo->cpu_mem) {
-			nbo->cpu_mem = nouveau_mem_alloc(ctx,
-							 NOUVEAU_MEM_AGP |
-							 NOUVEAU_MEM_MAPPED,
-							 nbo->gpu_mem->size,
-							 0);
+	obj->Pointer = nouveau_bo_map(ctx, GL_READ_WRITE_ARB, obj);
 
-			/* Mark GPU data as modified, so it gets copied to
-			 * the new buffer */
-			nbo->gpu_dirty = GL_TRUE;
-		}
-
-		if (nbo->cpu_mem && nbo->gpu_dirty) {
-			nouveau_memformat_flat_emit(ctx, nbo->cpu_mem,
-							 nbo->gpu_mem,
-							 0, 0,
-							 nbo->gpu_mem->size);
-
-			nouveau_notifier_wait_nop(ctx,
-						  nmesa->syncNotifier,
-						  NvSubMemFormat);
-			nbo->gpu_dirty = GL_FALSE;
-		}
-
-		/* buffer isn't guaranteed to be up-to-date on the card now */
-		nbo->cpu_dirty = GL_TRUE;
-	}
-#endif
-
-	/* If the copy to AGP failed for some reason, just return a pointer
-	 * directly to vram..
+	/* The GL spec says that a client attempting to write to a bufferobj
+	 * mapped READ_ONLY object may have unpredictable results, possibly
+	 * even program termination.
+	 *
+	 * We're going to use this, and only mark the buffer as dirtied if
+	 * the client asks for write access.
 	 */
-	if (!nbo->cpu_mem) {
-		DEBUG("Returning direct pointer to VRAM\n");
-		nbo->cpu_mem   = nbo->gpu_mem;
-		nbo->cpu_dirty = GL_FALSE;
+	if (target != GL_READ_ONLY_ARB) {
+		/* We have no way of knowing what was modified by the client,
+		 * so the entire buffer gets dirtied. */
+		nouveau_bo_dirty_all(ctx, GL_FALSE, obj);
 	}
 
-	obj->Pointer = nbo->cpu_mem->map;
 	return obj->Pointer;
 }
 
 static GLboolean
 nouveauUnmapBuffer(GLcontext *ctx, GLenum target, struct gl_buffer_object *obj)
 {
-	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-	nouveau_buffer_object *nbo = (nouveau_buffer_object *)obj;
-
-	DEBUG("obj=%p, target=%s\n", obj, _mesa_lookup_enum_by_nr(target));
+	DEBUG("target=%s, obj=%p\n", _mesa_lookup_enum_by_nr(target), obj);
 
-#ifdef ALLOW_MULTI_SUBCHANNEL
-	if (nbo->cpu_dirty && nbo->cpu_mem != nbo->gpu_mem) {
-		DEBUG("Copying potentially modified data back to GPU\n");
-
-		/* blit from GPU buffer -> CPU  buffer */
-		nouveau_memformat_flat_emit(ctx, nbo->gpu_mem, nbo->cpu_mem,
-		      			    0, 0, nbo->cpu_mem->size);
-
-		/* buffer is now up-to-date on the hardware (or rather, will
-		 * be by the time any other commands in this channel reference
-		 * the data.)
-		 */
-		nbo->cpu_dirty = GL_FALSE;
-
-		/* we can avoid this wait in some cases.. */
-		nouveau_notifier_wait_nop(ctx,
-					  nmesa->syncNotifier,
-					  NvSubMemFormat);
-
-		/* If it's likely CPU access to the buffer will occur often,
-		 * keep the cpu_mem around to avoid repeated allocs.
-		 */
-		if (obj->Usage != GL_DYNAMIC_DRAW_ARB) {
-
-			nouveau_mem_free(ctx, nbo->cpu_mem);
-			nbo->cpu_mem = NULL;
-		}
-	}
-#endif
+	assert(obj->Pointer);
 
+	nouveau_bo_unmap(ctx, obj);
 	obj->Pointer = NULL;
 	return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
index fccc349b836..932450fd877 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_bufferobj.h
@@ -4,24 +4,74 @@
 #include "mtypes.h"
 #include "nouveau_buffers.h"
 
+#define NOUVEAU_BO_VRAM_OK (NOUVEAU_MEM_FB | NOUVEAU_MEM_FB_ACCEPTABLE)
+#define NOUVEAU_BO_AGP_OK  (NOUVEAU_MEM_AGP | NOUVEAU_MEM_AGP_ACCEPTABLE)
+
+typedef struct nouveau_bufferobj_region_t {
+	uint32_t start;
+	uint32_t end;
+} nouveau_bufferobj_region;
+
+typedef struct nouveau_bufferobj_dirty_t {
+	nouveau_bufferobj_region *dirty;
+	int nr_dirty;
+} nouveau_bufferobj_dirty;
+
 typedef struct nouveau_buffer_object_t {
 	/* Base class, must be first */
 	struct gl_buffer_object mesa;
 
+	GLboolean		mapped;
+	GLenum			usage;
+
 	/* Memory used for GPU access to the buffer*/
+	GLuint			gpu_mem_flags;
 	nouveau_mem *		gpu_mem;
-	/* Buffer has been dirtied by the GPU */
-	GLboolean		gpu_dirty;
+	nouveau_bufferobj_dirty	gpu_dirty;
 
 	/* Memory used for CPU access to the buffer */
+	GLuint			cpu_mem_flags;
 	nouveau_mem *		cpu_mem;
-	/* Buffer has possibly been dirtied by the CPU */
-	GLboolean		cpu_dirty;
+	GLvoid *		cpu_mem_sys;
+	nouveau_bufferobj_dirty	cpu_dirty;
 } nouveau_buffer_object;
 
-extern uint32_t nouveau_bufferobj_gpu_ref(GLcontext *ctx, GLenum access,
-      					  struct gl_buffer_object *obj);
+extern void
+nouveau_bo_init_storage(GLcontext *ctx, GLuint valid_gpu_access,
+			GLsizeiptrARB size, const GLvoid *data, GLenum usage,
+			struct gl_buffer_object *bo);
+
+extern GLboolean
+nouveau_bo_move_in(GLcontext *ctx, struct gl_buffer_object *bo);
+
+extern GLboolean
+nouveau_bo_move_out(GLcontext *ctx, struct gl_buffer_object *bo);
+
+extern void *
+nouveau_bo_map(GLcontext *ctx, GLenum usage, struct gl_buffer_object *bo);
+
+extern void
+nouveau_bo_unmap(GLcontext *ctx, struct gl_buffer_object *bo);
+
+extern uint32_t
+nouveau_bo_gpu_ref(GLcontext *ctx, struct gl_buffer_object *bo);
+
+extern void
+nouveau_bo_dirty_linear(GLcontext *ctx, GLboolean on_card,
+			uint32_t offset, uint32_t size,
+			struct gl_buffer_object *bo);
+
+extern void
+nouveau_bo_dirty_all(GLcontext *ctx, GLboolean on_card,
+		     struct gl_buffer_object *bo);
+
+extern GLuint
+nouveau_bo_upload_dirty(GLcontext *ctx, struct gl_buffer_object *bo);
+
+extern GLuint
+nouveau_bo_download_dirty(GLcontext *ctx, struct gl_buffer_object *bo);
 
-extern void nouveauInitBufferObjects(GLcontext *ctx);
+extern void
+nouveauInitBufferObjects(GLcontext *ctx);
 
 #endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card.c b/src/mesa/drivers/dri/nouveau/nouveau_card.c
index a0628389bf7..91f12f0d704 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_card.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_card.c
@@ -2,9 +2,6 @@
 #include "nouveau_card.h"
 #include "nouveau_reg.h"
 #include "nouveau_drm.h"
-// FIXME hack for now
-#define NV15_TCL_PRIMITIVE_3D 0x0096
-#define NV17_TCL_PRIMITIVE_3D 0x0099
 #include "nouveau_card_list.h"
 
 
@@ -13,8 +10,8 @@ nouveau_card* nouveau_card_lookup(uint32_t device_id)
 	int i;
 	for(i=0;i<sizeof(nouveau_card_list)/sizeof(nouveau_card)-1;i++)
 		if (nouveau_card_list[i].id==(device_id&0xffff))
-			break;
-	return &(nouveau_card_list[i]);
+			return &(nouveau_card_list[i]);
+	return NULL;
 }
 
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h b/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
index f8ea3c355e2..8ec5c4a188a 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_card_list.h
@@ -2,15 +2,15 @@ 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_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x0028,  "RIVA TNT2/TNT2 Pro",                    NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x0029,  "RIVA TNT2 Ultra",                       NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002A,  "Riva TnT2",                             NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002B,  "Riva TnT2",                             NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002C,  "Vanta/Vanta LT",                        NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002D,  "RIVA TNT2 Model 64/Model 64 Pro",       NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002E,  "Vanta",                                 NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002F,  "Vanta",                                 NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 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},
+{0x002A,  "Riva TnT2",                             NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x002B,  "Riva TnT2",                             NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x002C,  "Vanta/Vanta LT",                        NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x002D,  "RIVA TNT2 Model 64/Model 64 Pro",       NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x002E,  "Vanta",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x002F,  "Vanta",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x0040,  "GeForce 6800 Ultra",                    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0041,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0042,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
@@ -30,7 +30,7 @@ static nouveau_card nouveau_card_list[]={
 {0x0098,  "GeForce Go 7800",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0099,  "GE Force Go 7800 GTX",                  NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x009D,  "Quadro FX4500",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00A0,  "Aladdin TNT2",                          NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
+{0x00A0,  "Aladdin TNT2",                          NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
 {0x00C0,  "GeForce 6800 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x00C1,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x00C2,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
@@ -49,11 +49,11 @@ static nouveau_card nouveau_card_list[]={
 {0x00F6,  "GeForce 6600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x00F8,  "Quadro FX 3400/4400",                   NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x00F9,  "GeForce 6800 Ultra/GeForce 6800 GT",    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x00FA,  "GeForce PCX 5750",                      NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
+{0x00FA,  "GeForce PCX 5750",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x00FB,  "GeForce PCX 5900",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x00FC,  "Quadro FX 330/GeForce PCX 5300",        NV30_TCL_PRIMITIVE_3D|0x0600,  NV_30, 0},
 {0x00FD,  "Quadro FX 330/Quadro NVS280",           NV30_TCL_PRIMITIVE_3D|0x0600,  NV_30, 0},
-{0x00FE,  "Quadro FX 1300",                        NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
+{0x00FE,  "Quadro FX 1300",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x00FF,  "GeForce PCX 4300",                      NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
 {0x0100,  "GeForce 256 SDR",                       NV10_TCL_PRIMITIVE_3D,         NV_10, 0},
 {0x0101,  "GeForce 256 DDR",                       NV10_TCL_PRIMITIVE_3D,         NV_10, 0},
@@ -71,7 +71,8 @@ static nouveau_card nouveau_card_list[]={
 {0x0148,  "GeForce Go 6600",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0149,  "GeForce Go 6600 GT",                    NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x014A,  "Quadro NVS 440",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x014D,  "Quadro FX 550",                         NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
+{0x014C,  "Quadro FX 550",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
+{0x014D,  "Quadro FX 550",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x014E,  "Quadro FX 540",                         NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x014F,  "GeForce 6200",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0150,  "GeForce2 GTS/Pro",                      NV11_TCL_PRIMITIVE_3D,         NV_15, 0},
@@ -121,10 +122,10 @@ static nouveau_card nouveau_card_list[]={
 {0x01DA,  "Quadro NVS 110M",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x01DF,  "GeForce 7300 GS",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x01F0,  "GeForce4 MX - nForce GPU",              NV17_TCL_PRIMITIVE_3D,         NV_17, 0},
-{0x0200,  "GeForce3",                              NV20_TCL_PRIMITIVE_3D|0x2000,  NV_20, 0},
-{0x0201,  "GeForce3 Ti 200",                       NV20_TCL_PRIMITIVE_3D|0x2000,  NV_20, 0},
-{0x0202,  "GeForce3 Ti 500",                       NV20_TCL_PRIMITIVE_3D|0x2000,  NV_20, 0},
-{0x0203,  "Quadro DCC",                            NV20_TCL_PRIMITIVE_3D|0x2000,  NV_20, 0},
+{0x0200,  "GeForce3",                              NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
+{0x0201,  "GeForce3 Ti 200",                       NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
+{0x0202,  "GeForce3 Ti 500",                       NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
+{0x0203,  "Quadro DCC",                            NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
 {0x0211,  "GeForce 6800",                          NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0212,  "GeForce 6800 LE",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0215,  "GeForce 6800 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
@@ -132,20 +133,21 @@ static nouveau_card nouveau_card_list[]={
 {0x0221,  "GeForce 6200",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x0240,  "GeForce 6150",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
 {0x0242,  "GeForce 6100",                          NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
-{0x0250,  "GeForce4 Ti 4600",                      NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0251,  "GeForce4 Ti 4400",                      NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0252,  "GeForce4 Ti",                           NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0253,  "GeForce4 Ti 4200",                      NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0258,  "Quadro4 900 XGL",                       NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0259,  "Quadro4 750 XGL",                       NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x025B,  "Quadro4 700 XGL",                       NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0280,  "GeForce4 Ti 4800",                      NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0281,  "GeForce4 Ti 4200 AGP 8x",               NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0282,  "GeForce4 Ti 4800 SE",                   NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0286,  "GeForce4 Ti 4200 Go AGP 8x",            NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0288,  "Quadro4 980 XGL",                       NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x0289,  "Quadro4 780 XGL",                       NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
-{0x028C,  "Quadro4 700 GoGL",                      NV20_TCL_PRIMITIVE_3D|0x2500,  NV_25, 0},
+{0x0244,  "Geforce 6150 Go",                       NV30_TCL_PRIMITIVE_3D|0x4400,  NV_44, 0},
+{0x0250,  "GeForce4 Ti 4600",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0251,  "GeForce4 Ti 4400",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0252,  "GeForce4 Ti",                           NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0253,  "GeForce4 Ti 4200",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0258,  "Quadro4 900 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0259,  "Quadro4 750 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x025B,  "Quadro4 700 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0280,  "GeForce4 Ti 4800",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0281,  "GeForce4 Ti 4200 AGP 8x",               NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0282,  "GeForce4 Ti 4800 SE",                   NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0286,  "GeForce4 Ti 4200 Go AGP 8x",            NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0288,  "Quadro4 980 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x0289,  "Quadro4 780 XGL",                       NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
+{0x028C,  "Quadro4 700 GoGL",                      NV20_TCL_PRIMITIVE_3D|0x0500,  NV_25, 0},
 {0x0290,  "GeForce 7900 GTX",                      NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0291,  "GeForce 7900 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0292,  "GeForce 7900 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
@@ -157,58 +159,59 @@ static nouveau_card nouveau_card_list[]={
 {0x029D,  "Quadro FX 3500",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x029E,  "Quadro FX 1500",                        NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x029F,  "Quadro FX 4500 X2",                     NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x02A0,  "XGPU",                                  NV20_TCL_PRIMITIVE_3D|0x2000,  NV_20, 0},
+{0x02A0,  "XGPU",                                  NV20_TCL_PRIMITIVE_3D,         NV_20, 0},
 {0x02E1,  "GeForce 7600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
-{0x0300,  "GeForce FX",                            NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0301,  "GeForce FX 5800 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0302,  "GeForce FX 5800",                       NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0308,  "Quadro FX 2000",                        NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0309,  "Quadro FX 1000",                        NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0311,  "GeForce FX 5600 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0312,  "GeForce FX 5600",                       NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0313,  "NV31",                                  NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0314,  "GeForce FX 5600XT",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0316,  "NV31M",                                 NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0317,  "NV31M Pro",                             NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x031A,  "GeForce FX Go5600",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x031B,  "GeForce FX Go5650",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x031D,  "NV31GLM",                               NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x031E,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x031F,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0320,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0321,  "GeForce FX 5200 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0322,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0323,  "GeForce FX 5200LE",                     NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0324,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0325,  "GeForce FX Go5250",                     NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0326,  "GeForce FX 5500",                       NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0327,  "GeForce FX 5100",                       NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0328,  "GeForce FX Go5200 32M/64M",             NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x0329,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x032A,  "Quadro NVS 280 PCI",                    NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x032B,  "Quadro FX 500/600 PCI",                 NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x032C,  "GeForce FX Go 5300",                    NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x032D,  "GeForce FX Go5100",                     NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
-{0x032F,  "NV34GL",                                NV30_TCL_PRIMITIVE_3D|0x3400,  NV_34, 0},
+{0x0300,  "GeForce FX",                            NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0301,  "GeForce FX 5800 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0302,  "GeForce FX 5800",                       NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0308,  "Quadro FX 2000",                        NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0309,  "Quadro FX 1000",                        NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0311,  "GeForce FX 5600 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0312,  "GeForce FX 5600",                       NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0313,  "NV31",                                  NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0314,  "GeForce FX 5600XT",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0316,  "NV31M",                                 NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0317,  "NV31M Pro",                             NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031A,  "GeForce FX Go5600",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031B,  "GeForce FX Go5650",                     NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031C,  "NVIDIA Quadro FX Go700",                NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031D,  "NV31GLM",                               NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031E,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x031F,  "NV31GLM Pro",                           NV30_TCL_PRIMITIVE_3D|0x0300,  NV_30, 0},
+{0x0320,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0321,  "GeForce FX 5200 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0322,  "GeForce FX 5200",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0323,  "GeForce FX 5200LE",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0324,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0325,  "GeForce FX Go5250",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0326,  "GeForce FX 5500",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0327,  "GeForce FX 5100",                       NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0328,  "GeForce FX Go5200 32M/64M",             NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x0329,  "GeForce FX Go5200",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x032A,  "Quadro NVS 280 PCI",                    NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x032B,  "Quadro FX 500/600 PCI",                 NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x032C,  "GeForce FX Go 5300",                    NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x032D,  "GeForce FX Go5100",                     NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
+{0x032F,  "NV34GL",                                NV30_TCL_PRIMITIVE_3D|0x0600,  NV_34, 0},
 {0x0330,  "GeForce FX 5900 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x0331,  "GeForce FX 5900",                       NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x0332,  "GeForce FX 5900XT",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0333,  "GeForce FX 5950 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
+{0x0333,  "GeForce FX 5950 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x0334,  "GeForce FX 5900ZT",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
-{0x0338,  "Quadro FX 3000",                        NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x033F,  "Quadro FX 700",                         NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0341,  "GeForce FX 5700 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0342,  "GeForce FX 5700",                       NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0343,  "GeForce FX 5700LE",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0344,  "GeForce FX 5700VE",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0345,  "NV36.5",                                NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0347,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0348,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x0349,  "NV36M Pro",                             NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x034B,  "NV36MAP",                               NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x034C,  "Quadro FX Go1000",                      NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x034E,  "Quadro FX 1100",                        NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
-{0x034F,  "NV36GL",                                NV30_TCL_PRIMITIVE_3D|0x3000,  NV_30, 0},
+{0x0338,  "Quadro FX 3000",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x033F,  "Quadro FX 700",                         NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0341,  "GeForce FX 5700 Ultra",                 NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0342,  "GeForce FX 5700",                       NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0343,  "GeForce FX 5700LE",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0344,  "GeForce FX 5700VE",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0345,  "NV36.5",                                NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0347,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0348,  "GeForce FX Go5700",                     NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x0349,  "NV36M Pro",                             NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x034B,  "NV36MAP",                               NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x034C,  "Quadro FX Go1000",                      NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x034E,  "Quadro FX 1100",                        NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
+{0x034F,  "NV36GL",                                NV30_TCL_PRIMITIVE_3D|0x0400,  NV_30, 0},
 {0x0391,  "GeForce 7600 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0392,  "GeForce 7600 GS",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
 {0x0393,  "GeForce 7300 GT",                       NV30_TCL_PRIMITIVE_3D|0x4000,  NV_40, 0},
@@ -221,9 +224,9 @@ static nouveau_card nouveau_card_list[]={
 {0x0009,  "DAC64",                                 0,                             NV_03, 0},
 {0x0018,  "Riva128",                               0,                             NV_03, 0},
 {0x0019,  "Riva128ZX",                             0,                             NV_03, 0},
-{0x0020,  "TNT",                                   NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x0028,  "TNT2",                                  NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x0029,  "UTNT2",                                 NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x002C,  "VTNT2",                                 NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 0},
-{0x00A0,  "ITNT2",                                 NV04_DX6_MULTITEX_TRIANGLE,    NV_04, 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},
+{0x002C,  "VTNT2",                                 NV04_DX5_TEXTURED_TRIANGLE,    NV_04, 0},
+{0x00A0,  "ITNT2",                                 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 79da46fc0b6..f811dc1b72e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c
@@ -31,7 +31,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "matrix.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
 #include "framebuffer.h"
 
 #include "tnl/tnl.h"
@@ -50,6 +49,8 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "nouveau_msg.h"
 #include "nouveau_reg.h"
 #include "nouveau_lock.h"
+#include "nouveau_query.h"
+#include "nv04_swtcl.h"
 #include "nv10_swtcl.h"
 
 #include "vblank.h"
@@ -70,6 +71,7 @@ static const struct dri_debug_control debug_control[] =
 };
 
 #define need_GL_ARB_vertex_program
+#define need_GL_ARB_occlusion_query
 #include "extension_helper.h"
 
 const struct dri_extension common_extensions[] =
@@ -99,6 +101,7 @@ const struct dri_extension nv40_extensions[] =
     * written for those cards.
     */
 	{ "GL_ARB_vertex_program",	GL_ARB_vertex_program_functions },
+	{ "GL_ARB_occlusion_query",	GL_ARB_occlusion_query_functions},
 	{ NULL, 0 }
 };
 
@@ -196,7 +199,7 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 
 	/* Initialize the swrast */
 	_swrast_CreateContext( ctx );
-	_ac_CreateContext( ctx );
+	_vbo_CreateContext( ctx );
 	_tnl_CreateContext( ctx );
 	_swsetup_CreateContext( ctx );
 
@@ -212,7 +215,7 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 			break;
 		case NV_04:
 		case NV_05:
-			//nv04TriInitFunctions( ctx );
+			nv04TriInitFunctions( ctx );
 			break;
 		case NV_10:
 		case NV_20:
@@ -228,6 +231,7 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 	nouveauInitBufferObjects(ctx);
 	if (!nouveauSyncInitFuncs(ctx))
 	   return GL_FALSE;
+	nouveauQueryInitFuncs(ctx);
 	nmesa->hw_func.InitCard(nmesa);
         nouveauInitState(ctx);
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h
index 134e2a417e3..f61fcbb48c0 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h
@@ -105,6 +105,11 @@ typedef struct nouveau_context {
 	/* Channel synchronisation */
 	nouveau_notifier *syncNotifier;
 
+	/* ARB_occlusion_query / EXT_timer_query */
+	GLuint		  query_object_max;
+	GLboolean *	  query_alloc;
+	nouveau_notifier *queryNotifier;
+
 	/* Additional hw-specific functions */
 	nouveau_hw_func hw_func;
 
@@ -119,7 +124,10 @@ typedef struct nouveau_context {
 	struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
 	GLuint vertex_attr_count;
 
-	/* Depth/stencil clear state */
+	/* Color buffer clear value */
+	uint32_t clear_color_value;
+
+	/* Depth/stencil clear value */
 	uint32_t clear_value;
 
 	/* Light state */
@@ -151,6 +159,7 @@ typedef struct nouveau_context {
 	nouveauShader *current_fragprog;
 	nouveauShader *current_vertprog;
 	nouveauShader *passthrough_vp;
+	nouveauShader *passthrough_fp;
 
 	nouveauScreenRec *screen;
 	drm_nouveau_sarea_t *sarea;
@@ -167,15 +176,15 @@ typedef struct nouveau_context {
 	/* Configuration cache */
 	driOptionCache optionCache;
 
-        /* vblank stuff */
-        uint32_t vblank_flags;
-        uint32_t vblank_seq;
+	/* vblank stuff */
+	uint32_t vblank_flags;
+	uint32_t vblank_seq;
 
-        GLuint new_state;
-        GLuint new_render_state;
-        GLuint render_index;
-        GLmatrix viewport;
-        GLfloat depth_scale;
+	GLuint new_state;
+	GLuint new_render_state;
+	GLuint render_index;
+	GLmatrix viewport;
+	GLfloat depth_scale;
 
 }nouveauContextRec, *nouveauContextPtr;
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
index 05d00d47690..490089f71a4 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_fifo.h
@@ -31,16 +31,32 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "nouveau_context.h"
 #include "nouveau_ctrlreg.h"
+#include "nouveau_state_cache.h"
 
+//#define NOUVEAU_RING_TRACE
 //#define NOUVEAU_RING_DEBUG
 //#define NOUVEAU_STATE_CACHE_DISABLE
 
+#ifndef NOUVEAU_RING_TRACE
+#define NOUVEAU_RING_TRACE 0
+#else
+#undef NOUVEAU_RING_TRACE
+#define NOUVEAU_RING_TRACE 1
+#endif
+
 #define NV_READ(reg) *(volatile u_int32_t *)(nmesa->mmio + (reg))
 
 #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_WRITE_PUT(val) NV_FIFO_WRITE(NV03_FIFO_REGS_DMAPUT, ((val)<<2) + nmesa->fifo.put_base)
+#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); \
+} while(0)
 
 /* 
  * Ring/fifo interface
@@ -74,15 +90,23 @@ int i; printf("OUT_RINGp: (size 0x%x dwords)\n",sz); for(i=0;i<sz;i++) printf("
 #else
 
 #define OUT_RINGp(ptr,sz) do{							\
+	if (NOUVEAU_RING_TRACE) { \
+		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);		\
 	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);					\
 }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);			\
 }while(0)
 
@@ -114,6 +138,7 @@ extern void nouveau_state_cache_init(nouveauContextPtr nmesa);
 #define OUT_RING_CACHE(n) do {									\
 	if (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; 		\
 		nmesa->state_cache.atoms[nmesa->state_cache.current_pos].value=(n);		\
 	}											\
 	nmesa->state_cache.current_pos++;							\
@@ -122,6 +147,7 @@ extern void nouveau_state_cache_init(nouveauContextPtr nmesa);
 #define OUT_RING_CACHEf(n) do {									\
 	if ((*(float*)(&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);\
 	}											\
 	nmesa->state_cache.current_pos++;							\
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.c b/src/mesa/drivers/dri/nouveau/nouveau_object.c
index 1558f2963df..302009c8b17 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.c
@@ -62,10 +62,19 @@ void nouveauObjectInit(nouveauContextPtr nmesa)
 
 	nouveauCreateContextObject(nmesa, Nv3D, nmesa->screen->card->class_3d,
 	      			   0, 0, 0, 0);
-	nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV10_CONTEXT_SURFACES_2D,
+	if (nmesa->screen->card->type>=NV_10) {
+		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV10_CONTEXT_SURFACES_2D,
 	      			   0, 0, 0, 0);
-	nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT,
+		nouveauCreateContextObject(nmesa, NvImageBlit, NV10_IMAGE_BLIT,
 	      			   NV_DMA_CONTEXT_FLAGS_PATCH_SRCCOPY, 0, 0, 0);
+	} else {
+		nouveauCreateContextObject(nmesa, NvCtxSurf2D, NV04_CONTEXT_SURFACES_2D,
+	      			   0, 0, 0, 0);
+		nouveauCreateContextObject(nmesa, NvCtxSurf3D, NV04_CONTEXT_SURFACES_3D,
+	      			   0, 0, 0, 0);
+		nouveauCreateContextObject(nmesa, NvImageBlit, NV_IMAGE_BLIT,
+	      			   NV_DMA_CONTEXT_FLAGS_PATCH_SRCCOPY, 0, 0, 0);
+	}
 	nouveauCreateContextObject(nmesa, NvMemFormat,
 	      			   NV_MEMORY_TO_MEMORY_FORMAT,
 	      			   0, 0, 0, 0);
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_object.h b/src/mesa/drivers/dri/nouveau/nouveau_object.h
index b1ff5a5d0dc..daad281029f 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_object.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_object.h
@@ -12,15 +12,18 @@ enum DMAObjects {
 	NvCtxSurf2D		= 0x80000020,
 	NvImageBlit		= 0x80000021,
 	NvMemFormat		= 0x80000022,
+	NvCtxSurf3D		= 0x80000023,
 	NvDmaFB			= 0xD0FB0001,
 	NvDmaAGP		= 0xD0AA0001,
-	NvSyncNotify		= 0xD0000001
+	NvSyncNotify		= 0xD0000001,
+	NvQueryNotify		= 0xD0000002
 };
 
 enum DMASubchannel {
 	NvSubCtxSurf2D	= 0,
 	NvSubImageBlit	= 1,
 	NvSubMemFormat	= 2,
+	NvSubCtxSurf3D	= 3,
 	NvSub3D		= 7,
 };
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.c b/src/mesa/drivers/dri/nouveau/nouveau_query.c
new file mode 100644
index 00000000000..de3f5b0378b
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_query.c
@@ -0,0 +1,198 @@
+/*
+ * Copyright (C) 2007 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/* GL_ARB_occlusion_query support for NV20/30/40 */
+
+#include "mtypes.h"
+
+#include "nouveau_fifo.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+#include "nouveau_reg.h"
+#include "nouveau_sync.h"
+#include "nouveau_query.h"
+
+static struct gl_query_object *
+nouveauNewQueryObject(GLcontext *ctx, GLuint id)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_query_object *nq;
+	int i;
+
+	for (i=0; i<nmesa->query_object_max; i++)
+		if (nmesa->query_alloc[i] == GL_FALSE)
+			break;
+	if (i==nmesa->query_object_max)
+		return NULL;
+
+	nq = CALLOC_STRUCT(nouveau_query_object_t);
+	if (nq) {
+		nq->notifier_id	= i;
+
+		nq->mesa.Id     = id;
+		nq->mesa.Result = 0;
+		nq->mesa.Active = GL_FALSE;
+		nq->mesa.Ready  = GL_TRUE;
+	}
+
+	return (struct gl_query_object *)nq;
+}
+
+static void
+nouveauBeginQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_query_object *nq = (nouveau_query_object *)q;
+
+	nouveau_notifier_reset(nmesa->queryNotifier, nq->notifier_id);
+
+	switch (nmesa->screen->card->type) {
+	case NV_20:
+		BEGIN_RING_CACHE(NvSub3D, 0x17c8, 1);
+		OUT_RING_CACHE  (1);
+		BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
+		OUT_RING_CACHE  (1);
+		break;
+	case NV_30:
+	case NV_40:
+	case NV_44:
+		/* I don't think this is OCC_QUERY enable, but it *is* needed to make
+		 * the SET_OBJECT7 notifier block work with STORE_RESULT.
+		 *
+		 * Also, this appears to reset the pixel pass counter */
+		BEGIN_RING_SIZE(NvSub3D,
+				 NV30_TCL_PRIMITIVE_3D_OCC_QUERY_OR_COLOR_BUFF_ENABLE,
+				 1);
+		OUT_RING        (1);
+		/* Probably OCC_QUERY_ENABLE */
+		BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
+		OUT_RING_CACHE  (1);
+		break;
+	default:
+		WARN_ONCE("no support for this card\n");
+		break;
+	}
+}
+
+static void
+nouveauUpdateQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_query_object *nq = (nouveau_query_object *)q;
+	int status;
+
+	status = nouveau_notifier_status(nmesa->queryNotifier,
+					 nq->notifier_id);
+
+	q->Ready = (status == NV_NOTIFY_STATE_STATUS_COMPLETED);
+	if (q->Ready)
+		q->Result = nouveau_notifier_return_val(nmesa->queryNotifier,
+							nq->notifier_id);
+}
+
+static void
+nouveauWaitQueryResult(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	nouveau_query_object *nq = (nouveau_query_object *)q;
+
+	nouveau_notifier_wait_status(nmesa->queryNotifier, nq->notifier_id,
+				     NV_NOTIFY_STATE_STATUS_COMPLETED, 0);
+	nouveauUpdateQuery(ctx, target, q);
+}
+
+static void
+nouveauEndQuery(GLcontext *ctx, GLenum target, struct gl_query_object *q)
+{
+	nouveau_query_object *nq = (nouveau_query_object *)q;
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	switch (nmesa->screen->card->type) {
+	case NV_20:
+		BEGIN_RING_SIZE(NvSub3D, 0x17d0, 1);
+		OUT_RING       (0x01000000 | nq->notifier_id*32);
+		break;
+	case NV_30:
+	case NV_40:
+	case NV_44:
+		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_STORE_RESULT, 1);
+		OUT_RING       (0x01000000 | nq->notifier_id*32);
+		break;
+	default:
+		WARN_ONCE("no support for this card\n");
+		break;
+	}
+	FIRE_RING();
+
+	/*XXX: wait for query to complete, mesa doesn't give the driver
+	 *     an interface to query the status of a query object so 
+	 *     this has to stall the channel.
+	 */
+	nouveauWaitQueryResult(ctx, target, q);
+
+	BEGIN_RING_CACHE(NvSub3D, 0x17cc, 1);
+	OUT_RING_CACHE  (0);
+}
+
+void
+nouveauQueryInitFuncs(GLcontext *ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (nmesa->screen->card->type < NV_20)
+		return;
+
+	nmesa->query_object_max = (0x4000 / 32);
+	nmesa->queryNotifier =
+		nouveau_notifier_new(ctx, NvQueryNotify,
+					  nmesa->query_object_max);
+	nmesa->query_alloc = calloc(nmesa->query_object_max, sizeof(GLboolean));
+
+	switch (nmesa->screen->card->type) {
+	case NV_20:
+		BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
+		OUT_RING_CACHE  (NvQueryNotify);
+		break;
+	case NV_30:
+	case NV_40:
+	case NV_44:
+		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT7, 1);
+		OUT_RING_CACHE  (NvQueryNotify);
+		break;
+	default:
+		break;
+	};
+
+	ctx->Driver.NewQueryObject	= nouveauNewQueryObject;
+	ctx->Driver.BeginQuery		= nouveauBeginQuery;
+	ctx->Driver.EndQuery		= nouveauEndQuery;
+#if 0
+	ctx->Driver.UpdateQuery		= nouveauUpdateQuery;
+	ctx->Driver.WaitQueryResult	= nouveauWaitQueryResult;
+#endif
+}
+
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_query.h b/src/mesa/drivers/dri/nouveau/nouveau_query.h
new file mode 100644
index 00000000000..3ded41417e1
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_query.h
@@ -0,0 +1,38 @@
+/*
+ * Copyright (C) 2007 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+#ifndef __NOUVEAU_QUERY_H__
+#define __NOUVEAU_QUERY_H__
+
+typedef struct nouveau_query_object_t {
+	struct gl_query_object mesa;
+
+	int notifier_id;
+} nouveau_query_object;
+
+extern void nouveauQueryInitFuncs(GLcontext *ctx);
+#endif
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_reg.h b/src/mesa/drivers/dri/nouveau/nouveau_reg.h
index 200e770903f..8758b538c85 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. 1.350
+   Created from objects.c rev. 1.398
 */
 
 #ifndef _NOUVEAU_REG_H
@@ -60,6 +60,7 @@ Object NV01_CONTEXT_CLIP_RECTANGLE used on: NV03 NV04 NV10 NV15 NV20 NV40 G70
 Object NV_MEMORY_TO_MEMORY_FORMAT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
 */
 #define			NV_MEMORY_TO_MEMORY_FORMAT				0x00000039
+#	define		NV_MEMORY_TO_MEMORY_FORMAT_NOP				0x00000100
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_NOTIFY			0x00000104
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_DMA_NOTIFY			0x00000180
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_OBJECT_IN			0x00000184
@@ -70,6 +71,8 @@ Object NV_MEMORY_TO_MEMORY_FORMAT used on: NV04 NV10 NV15 NV20 NV30 NV40 G70
 #	define		NV_MEMORY_TO_MEMORY_FORMAT_PITCH_OUT			0x00000318
 #	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_BUF_NOTIFY			0x00000328
 
 /****************************************** 
 Object NV03_PRIMITIVE_RASTER_OP used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
@@ -322,18 +325,19 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	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_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_COLOR0				0x00000270	/* Parameters: a r g b */
 #	define		NV10_TCL_PRIMITIVE_3D_RC_COLOR1				0x00000274	/* Parameters: a r g b */
 #	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_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_LIGHT_MODEL			0x00000294	/* Parameters: local_viewer color_control */
-#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL			0x00000298	/* Parameters: specular diffuse ambient emission */
+#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_ENABLE		0x00000298	/* Parameters: specular diffuse ambient emission */
 #	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
@@ -363,7 +367,7 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_BLEND_COLOR			0x0000034c	/* Parameters: a r g b */
 #	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: r g b */
+#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MASK			0x00000358	/* Parameters: a r g b */
 #	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
@@ -383,7 +387,10 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	define		NV10_TCL_PRIMITIVE_3D_CULL_FACE				0x0000039c
 #	define		NV10_TCL_PRIMITIVE_3D_FRONT_FACE			0x000003a0
 #	define		NV10_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x000003a4
-#	define		NV10_TCL_PRIMITIVE_3D_MATERIAL_DIFFUSE_ALPHA_FRONT	0x000003b4
+#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_R			0x000003a8
+#	define		NV10_TCL_PRIMITIVE_3D_COLOR_MATERIAL_G			0x000003ac
+#	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_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)
@@ -495,7 +502,7 @@ Object NV11_TCL_PRIMITIVE_3D used on: NV15
 #	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)
+#	define		NV10_TCL_PRIMITIVE_3D_VERTEX_ATTR( d)			(0x00000d04 + d * 0x0008)	/* Parameters: stride fields type */
 #	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 */
@@ -592,6 +599,7 @@ Object NV10_UNK0072 used on: NV10 NV15 NV20 NV40 G70
 */
 #define			NV10_UNK0072						0x00000072
 #	define		NV10_UNK0072_COUNTER					0x00000050
+#	define		NV40_UNK0072_SET_OBJECT					0x00000060
 #	define		NV10_UNK0072_SET_DMA_NOTIFY				0x00000180
 
 /****************************************** 
@@ -664,6 +672,8 @@ Object NV20_SWIZZLED_SURFACE used on: NV20 NV30 NV40 G70
 Object NV20_TCL_PRIMITIVE_3D used on: NV20 
 */
 #define			NV20_TCL_PRIMITIVE_3D					0x00000097
+#	define		NV20_TCL_PRIMITIVE_3D_NOP				0x00000100
+#	define		NV20_TCL_PRIMITIVE_3D_NOTIFY				0x00000104
 #	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT0			0x00000180
 #	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT1			0x00000184
 #	define		NV20_TCL_PRIMITIVE_3D_SET_OBJECT2			0x00000188
@@ -688,6 +698,8 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	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_ALPHA_FUNC_ENABLE			0x00000300
 #	define		NV20_TCL_PRIMITIVE_3D_BLEND_FUNC_ENABLE			0x00000304
 #	define		NV20_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE			0x00000308
@@ -729,8 +741,10 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_CULL_FACE				0x0000039c
 #	define		NV20_TCL_PRIMITIVE_3D_FRONT_FACE			0x000003a0
 #	define		NV20_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE			0x000003a4
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT		0x000003a8
-#	define		NV20_TCL_PRIMITIVE_3D_MATERIAL_DIFFUSE_ALPHA_FRONT	0x000003b4
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R		0x000003a8
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G		0x000003ac
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B		0x000003b0
+#	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)
@@ -754,6 +768,10 @@ 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_VIEWPORT_OX			0x00000a20
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_OY			0x00000a24
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_DEPTH_AVG_S		0x00000a28
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_UNKNOWN_A		0x00000a2c
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_A			0x00000a30
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_B			0x00000a34
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_C			0x00000a38
@@ -762,6 +780,10 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_F			0x00000a44
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_G			0x00000a48
 #	define		NV20_TCL_PRIMITIVE_3D_POINT_PARAMETER_H			0x00000a4c
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_PX_DIV2			0x00000af0
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_PY_DIV2			0x00000af4
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_DEPTH_HALF_S		0x00000af8
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_UNKNOWN_B		0x00000afc
 #	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST0			0x00000b00
 #	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST1			0x00000b04
 #	define		NV20_TCL_PRIMITIVE_3D_VP_UPLOAD_INST2			0x00000b08
@@ -776,10 +798,13 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	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_RC_ENABLE				0x00001e60	/* Parameters: number of rc enabled */
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_OP			0x00001e70	/* Parameters: op0 op1 op2 op3 */
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_CULL_MODE		0x000017f8	/* Parameters: cull0 cull1 cull2 cull3 */
 #	define		NV20_TCL_PRIMITIVE_3D_TX_SHADER_PREVIOUS		0x00001e78	/* Parameters: prev2 prev3 */
@@ -789,6 +814,8 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	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)
@@ -930,18 +957,26 @@ Object NV20_TCL_PRIMITIVE_3D used on: NV20
 #	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
-#	define		NV20_TCL_PRIMITIVE_3D_MATERIAL_DIFFUSE_ALPHA_BACK	0x000017ac
-#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK		0x000017b0
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_A		0x000017ac
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_R		0x000017b0
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_G		0x000017b4
+#	define		NV20_TCL_PRIMITIVE_3D_COLOR_MATERIAL_BACK_B		0x000017b8
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE		0x000017bc
 #	define		NV20_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_OP			0x000017c0
 #	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_Y2_Y1			0x00001c50	/* Parameters: y2 y1 */
 #	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_INDEX_DATA			0x00001800	/* Parameters: index1 index0 */
 #	define		NV20_TCL_PRIMITIVE_3D_VB_VERTEX_BATCH			0x00001810	/* Parameters: count_vertices offset_vertices */
 #	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
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_Z			0x00001f08
+#	define		NV20_TCL_PRIMITIVE_3D_VIEWPORT_ORIGIN_W			0x00001f0c
 
 /****************************************** 
 Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
@@ -958,8 +993,8 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT8			0x000001ac
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT9			0x000001b4
 #	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT10			0x000001b8
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT11			0x0000019c
-#	define		NV30_TCL_PRIMITIVE_3D_SET_OBJECT12			0x000001a0
+#	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_COLOR0_OFFSET			0x00000210
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET			0x00000214
@@ -1000,12 +1035,17 @@ 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_NORMALIZE_ENABLE			0x0000037c
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR			0x00000394
 #	define		NV30_TCL_PRIMITIVE_3D_DEPTH_RANGE_FAR			0x00000398
-#	define		NV30_TCL_PRIMITIVE_3D_MATERIAL_DIFFUSE_ALPHA_FRONT	0x000003b4
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_R		0x000003a0
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_G		0x000003a4
+#	define		NV30_TCL_PRIMITIVE_3D_COLOR_MATERIAL_FRONT_B		0x000003a8
+#	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)
@@ -1017,12 +1057,17 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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_RC_COLOR0				0x000008ec	/* Parameters: a r g b */
+#	define		NV30_TCL_PRIMITIVE_3D_RC_COLOR1				0x000008f0	/* Parameters: a r g b */
 #	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_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_IN_ALPHA			0x00000900	/* 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				0x00000904	/* 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_OUT_ALPHA			0x00000910	/* 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			0x00000914	/* Parameters: scale bias mux_sum ab_dot_product cd_dot_product sum_output ab_output cd_output */
+#	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 */
@@ -1044,6 +1089,7 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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
@@ -1064,6 +1110,10 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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)
@@ -1105,7 +1155,9 @@ 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_LINE_STIPPLE			0x00001db4
+#	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_BEGIN_END				0x00001808
 #	define		NV30_TCL_PRIMITIVE_3D_CULL_FACE				0x00001830
@@ -1149,7 +1201,7 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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: enabled? offset */
+#	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
@@ -1196,7 +1248,6 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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_MATERIAL_DIFFUSE_ALPHA_BACK	0x000017c0
 #	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 */
@@ -1208,6 +1259,8 @@ Object NV30_TCL_PRIMITIVE_3D used on: NV30 NV40 G70
 #	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
@@ -1232,6 +1285,161 @@ Object NV30_CLEAR_BUFFER used on: NV30 NV40 G70
 #	define		NV30_CLEAR_BUFFER_UNK002fc				0x000002fc
 
 /****************************************** 
+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_VERTEX_FOG_1F			0x00000314
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_X			0x00000380
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_2F_Y			0x00000384
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_S			0x000003c0
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_2F_T			0x000003c4
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_S			0x000003c8
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_2F_T			0x000003cc
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_S			0x000003d0
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2F_T			0x000003d4
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_S			0x000003d8
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2F_T			0x000003dc
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_X			0x00000400
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Y			0x00000404
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_3F_Z			0x00000408
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_X			0x00000420
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Y			0x00000424
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3F_Z			0x00000428
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_R			0x00000430
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_G			0x00000434
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_3F_B			0x00000438
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_R			0x00000440
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_G			0x00000444
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3F_B			0x00000448
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_X			0x00000500
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Y			0x00000504
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_Z			0x00000508
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4F_W			0x0000050c
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_R			0x00000530
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_G			0x00000534
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_B			0x00000538
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4F_A			0x0000053c
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_S			0x00000580
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_T			0x00000584
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_R			0x00000588
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4F_Q			0x0000058c
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_S			0x00000590
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_T			0x00000594
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_R			0x00000598
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4F_Q			0x0000059c
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_S			0x000005a0
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_T			0x000005a4
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_R			0x000005a8
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4F_Q			0x000005ac
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_S			0x000005b0
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4F_T			0x000005b4
+#	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_TX1_2I			0x000006a4	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_2I			0x000006a8	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_2I			0x000006ac	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_XY			0x00000700	/* Parameters: y x */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_POS_4I_ZW			0x00000704	/* Parameters: w z */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_ST			0x00000740	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX0_4I_RQ			0x00000744	/* Parameters: q r */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_ST			0x00000748	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX1_4I_RQ			0x0000074c	/* Parameters: q r */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_ST			0x00000750	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX2_4I_RQ			0x00000754	/* Parameters: q r */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_ST			0x00000758	/* Parameters: t s */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_TX3_4I_RQ			0x0000075c	/* Parameters: q r */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_XY			0x00000790	/* Parameters: y x */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_NOR_3I_Z			0x00000794	/* Parameters: z */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL_4I			0x0000088c	/* Parameters: a b g r */
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_COL2_3I			0x00000890	/* Parameters: a b g r */
+#	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
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_X			0x00000a0c
+#	define		NV50_TCL_PRIMITIVE_3D_VIEWPORT_UNK1_Y			0x00000a10
+#	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_VERTEX_BUFFER_FIRST		0x00000d74
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BUFFER_COUNT		0x00000d78
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_R			0x00000d80
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_G			0x00000d84
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_B			0x00000d88
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_A			0x00000d8c
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH			0x00000d90
+#	define		NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL			0x00000da0
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT		0x00000dac
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK			0x00000db0
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE		0x00000db4
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE	0x00000dc0
+#	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_HEIGHT_YPOS		0x00000e08	/* Parameters: h y */
+#	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
+#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_Z			0x00000f0c
+#	define		NV50_TCL_PRIMITIVE_3D_VP_UPLOAD_CONST_W			0x00000f10
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF		0x00000f54
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK		0x00000f58
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK		0x00000f5c
+#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE			0x000012cc
+#	define		NV50_TCL_PRIMITIVE_3D_SHADE_MODEL			0x000012d4
+#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE		0x000012e8
+#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE			0x000012ec
+#	define		NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC			0x0000130c
+#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF			0x00001310
+#	define		NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC			0x00001314
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_R			0x0000131c
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_G			0x00001320
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_B			0x00001324
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_A			0x00001328
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB		0x00001340
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB		0x00001344
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_RGB		0x00001348
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA		0x0000134c
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA		0x00001350
+#	define		NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_DST_ALPHA		0x00001358
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE		0x00001380
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL		0x00001384
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZFAIL		0x00001388
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_ZPASS		0x0000138c
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC		0x00001390
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_REF		0x00001394
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK			0x00001398
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK		0x0000139c
+#	define		NV50_TCL_PRIMITIVE_3D_LINE_WIDTH			0x000013b0
+#	define		NV50_TCL_PRIMITIVE_3D_POINT_SIZE			0x00001518
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR		0x0000156c
+#	define		NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE		0x00001570
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE		0x00001594
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL		0x00001598
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZFAIL		0x0000159c
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_ZPASS		0x000015a0
+#	define		NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC		0x000015a4
+#	define		NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS		0x000015bc
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_BEGIN			0x000015dc
+#	define		NV50_TCL_PRIMITIVE_3D_VERTEX_END			0x000015e0
+#	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_POLYGON_STIPPLE_ENABLE		0x0000168c
+#	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 */
+
+/****************************************** 
 Object NV_DMA_FROM_MEMORY used on: NV03 NV04 NV10 NV15 NV20 NV30 NV40 G70
 */
 #define			NV_DMA_FROM_MEMORY					0x00000002
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_screen.c b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
index 0e3901cfcab..881b20149f0 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_screen.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_screen.c
@@ -71,6 +71,14 @@ static nouveauScreenPtr nouveauCreateScreen(__DRIscreenPrivate *sPriv)
 		return NULL;
 	}
 	
+	screen->card=nouveau_card_lookup(dri_priv->device_id);
+	if (!screen->card) {
+		__driUtilMessage("%s: Unknown card type 0x%04x:0x%04x\n",
+			__func__, dri_priv->device_id >> 16, dri_priv->device_id & 0xFFFF);
+		FREE(screen);
+		return NULL;
+	}
+
 	/* parse information in __driConfigOptions */
 	driParseOptionInfo (&screen->optionCache,__driConfigOptions, __driNConfigOptions);
 
@@ -82,7 +90,6 @@ static nouveauScreenPtr nouveauCreateScreen(__DRIscreenPrivate *sPriv)
 	screen->depthOffset = dri_priv->depth_offset;
 	screen->depthPitch  = dri_priv->depth_pitch;
 
-	screen->card=nouveau_card_lookup(dri_priv->device_id);
 	screen->driScreen = sPriv;
 	return screen;
 }
@@ -320,7 +327,7 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc
 	__DRIscreenPrivate *psp;
 	static const __DRIversion ddx_expected = { 1, 2, 0 };
 	static const __DRIversion dri_expected = { 4, 0, 0 };
-	static const __DRIversion drm_expected = { 0, 0, 2 };
+	static const __DRIversion drm_expected = { 0, 0, 3 };
 
 	dri_interface = interface;
 
@@ -332,8 +339,12 @@ void * __driCreateNewScreen_20050727( __DRInativeDisplay *dpy, int scrn, __DRIsc
 	}
 
 	// temporary lock step versioning
-	if (drm_expected.patch!=drm_version->patch)
+	if (drm_expected.patch!=drm_version->patch) {
+		__driUtilMessage("%s: wrong DRM version, expected %d, got %d\n",
+				__func__,
+				drm_expected.patch, drm_version->patch);
 		return NULL;
+	}
 
 	psp = __driUtilCreateNewScreen(dpy, scrn, psc, NULL,
 				       ddx_version, dri_version, drm_version,
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.c b/src/mesa/drivers/dri/nouveau/nouveau_shader.c
index dc366b36c09..ba471325aaa 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader.c
@@ -111,7 +111,7 @@ nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs)
    /* Translate to HW format now if necessary */
    if (!nvs->translated) {
       /* Mesa ASM shader -> nouveauShader */
-      if (!nouveau_shader_pass0_arb(ctx, nvs))
+      if (!nouveau_shader_pass0(ctx, nvs))
 	 return GL_FALSE;
       /* Basic dead code elimination + register usage info */
       if (!nouveau_shader_pass1(nvs))
@@ -126,15 +126,16 @@ nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs)
    /* Update state parameters */
    plist = nvs->mesa.vp.Base.Parameters;
    _mesa_load_state_parameters(ctx, plist);
-   for (i=0; i<plist->NumParameters; i++) {
+   for (i=0; i<nvs->param_high; i++) {
+      if (!nvs->params[i].in_use)
+	 continue;
+
       if (!nvs->on_hardware) {
 	 /* if we've been kicked off the hardware there's no guarantee our
 	  * consts are still there.. reupload them all
 	  */
 	 nvs->func->UpdateConst(ctx, nvs, i);
-      } else if (plist->Parameters[i].Type == PROGRAM_STATE_VAR) {
-	 if (!nvs->params[i].source_val) /* this is a workaround when consts aren't alloc'd from id=0.. */
-	    continue;
+      } else if (nvs->params[i].source_val) {
 	 /* update any changed state parameters */
 	 if (!TEST_EQ_4V(nvs->params[i].val, nvs->params[i].source_val))
 	    nvs->func->UpdateConst(ctx, nvs, i);
@@ -179,7 +180,7 @@ nvsBuildTextShader(GLcontext *ctx, GLenum target, const char *text)
 				     strlen(text),
 				     &nvs->mesa.vp);
    } else if (target == GL_FRAGMENT_PROGRAM_ARB) {
-      _mesa_init_fragment_program(ctx, &nvs->mesa.fp, GL_VERTEX_PROGRAM_ARB, 0);
+      _mesa_init_fragment_program(ctx, &nvs->mesa.fp, GL_FRAGMENT_PROGRAM_ARB, 0);
       _mesa_parse_arb_fragment_program(ctx,
 	    			       GL_FRAGMENT_PROGRAM_ARB,
 				       text,
@@ -187,7 +188,7 @@ nvsBuildTextShader(GLcontext *ctx, GLenum target, const char *text)
 				       &nvs->mesa.fp);
    }
 
-   nouveau_shader_pass0_arb(ctx, nvs);
+   nouveau_shader_pass0(ctx, nvs);
    nouveau_shader_pass1(nvs);
    nouveau_shader_pass2(nvs);
 
@@ -219,6 +220,21 @@ nvsBuildPassthroughVP(GLcontext *ctx)
 					      vp_text);
 }
 
+static void
+nvsBuildPassthroughFP(GLcontext *ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	const char *fp_text =
+		"!!ARBfp1.0\n"
+		"MOV result.color, fragment.color;\n"
+		"END";
+
+	nmesa->passthrough_fp = nvsBuildTextShader(ctx,
+						   GL_FRAGMENT_PROGRAM_ARB,
+						   fp_text);
+}
+
 void
 nouveauShaderInitFuncs(GLcontext * ctx)
 {
@@ -248,6 +264,10 @@ nouveauShaderInitFuncs(GLcontext * ctx)
    if (nmesa->screen->card->type >= NV_40)
       nvsBuildPassthroughVP(ctx);
 
+   /* Needed on NV30, even when using swtcl, if you want to get colours */
+   if (nmesa->screen->card->type >= NV_30)
+      nvsBuildPassthroughFP(ctx);
+
    ctx->Const.VertexProgram.MaxNativeInstructions    = nmesa->VPfunc.MaxInst;
    ctx->Const.VertexProgram.MaxNativeAluInstructions = nmesa->VPfunc.MaxInst;
    ctx->Const.VertexProgram.MaxNativeTexInstructions = nmesa->VPfunc.MaxInst;
@@ -563,12 +583,12 @@ nvsDumpInstruction(nvsInstruction * inst, int slot, int lvl)
 }
 
 void
-nvsDumpFragmentList(nvsFragmentList *f, int lvl)
+nvsDumpFragmentList(nvsFragmentHeader *f, int lvl)
 {
    while (f) {
-      switch (f->fragment->type) {
+      switch (f->type) {
       case NVS_INSTRUCTION:
-	 nvsDumpInstruction((nvsInstruction*)f->fragment, 0, lvl);
+	 nvsDumpInstruction((nvsInstruction*)f, 0, lvl);
 	 break;
       default:
 	 fprintf(stderr, "%s: Only NVS_INSTRUCTION fragments can be in"
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.h b/src/mesa/drivers/dri/nouveau/nouveau_shader.h
index 08cb7817cf0..b2df3546f62 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader.h
@@ -12,18 +12,29 @@ typedef struct _nvsFunc nvsFunc;
 #define NVS_MAX_ADDRESS 2
 #define NVS_MAX_INSNS   4096
 
-typedef struct {
+typedef struct _nvs_fragment_header {
+   struct _nvs_fragment_header *parent;
+   struct _nvs_fragment_header *prev;
+   struct _nvs_fragment_header *next;
    enum {
       NVS_INSTRUCTION,
+      NVS_BRANCH,
+      NVS_LOOP,
+      NVS_SUBROUTINE
    } type;
-   int position;
 } nvsFragmentHeader;
 
-typedef struct _nvs_fragment_list {
-   struct _nvs_fragment_list *prev;
-   struct _nvs_fragment_list *next;
-   nvsFragmentHeader *fragment;
-} nvsFragmentList;
+typedef union {
+	struct {
+		GLboolean uses_kil;
+		GLuint    num_regs;
+	} NV30FP;
+	struct {
+		uint32_t vp_in_reg;
+		uint32_t vp_out_reg;
+		uint32_t clip_enables;
+	} NV30VP;
+} nvsCardPriv;
 
 typedef struct _nouveauShader {
    union {
@@ -42,11 +53,14 @@ typedef struct _nouveauShader {
    unsigned int program_start_id;
    unsigned int program_current;
    struct gl_buffer_object *program_buffer;
-   unsigned int inputs_read;
-   unsigned int outputs_written;
    int		inst_count;
 
+   nvsCardPriv card_priv;
+   int		vp_attrib_map[NVS_MAX_ATTRIBS];
+
    struct {
+      GLboolean in_use;
+
       GLfloat  *source_val;	/* NULL if invariant */
       float	val[4];
       /* Hardware-specific tracking, currently only nv30_fragprog
@@ -55,16 +69,12 @@ typedef struct _nouveauShader {
       int	*hw_index;
       int        hw_index_cnt;
    } params[NVS_MAX_CONSTS];
-
-   struct {
-      int last_use;
-   } temps[NVS_MAX_TEMPS];
+   int param_high;
 
    /* Pass-private data */
    void *pass_rec;
 
-   nvsFragmentList *list_head;
-   nvsFragmentList *list_tail;
+   nvsFragmentHeader *program_tree;
 } nouveauShader, *nvsPtr;
 
 typedef enum {
@@ -119,35 +129,35 @@ typedef enum {
 } nvsSwzComp;
 
 typedef enum {
-   NVS_FR_POSITION,
-   NVS_FR_WEIGHT,
-   NVS_FR_NORMAL,
-   NVS_FR_COL0,
-   NVS_FR_COL1,
-   NVS_FR_BFC0,
-   NVS_FR_BFC1,
-   NVS_FR_FOGCOORD,
-   NVS_FR_POINTSZ,
-   NVS_FR_TEXCOORD0,
-   NVS_FR_TEXCOORD1,
-   NVS_FR_TEXCOORD2,
-   NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4,
-   NVS_FR_TEXCOORD5,
-   NVS_FR_TEXCOORD6,
-   NVS_FR_TEXCOORD7,
-   NVS_FR_FRAGDATA0,
-   NVS_FR_FRAGDATA1,
-   NVS_FR_FRAGDATA2,
-   NVS_FR_FRAGDATA3,
-   NVS_FR_CLIP0,
-   NVS_FR_CLIP1,
-   NVS_FR_CLIP2,
-   NVS_FR_CLIP3,
-   NVS_FR_CLIP4,
-   NVS_FR_CLIP5,
-   NVS_FR_CLIP6,
-   NVS_FR_FACING,
+   NVS_FR_POSITION	= 0,
+   NVS_FR_WEIGHT	= 1,
+   NVS_FR_NORMAL	= 2,
+   NVS_FR_COL0		= 3,
+   NVS_FR_COL1		= 4,
+   NVS_FR_FOGCOORD	= 5,
+   NVS_FR_TEXCOORD0	= 8,
+   NVS_FR_TEXCOORD1	= 9,
+   NVS_FR_TEXCOORD2	= 10,
+   NVS_FR_TEXCOORD3	= 11,
+   NVS_FR_TEXCOORD4	= 12,
+   NVS_FR_TEXCOORD5	= 13,
+   NVS_FR_TEXCOORD6	= 14,
+   NVS_FR_TEXCOORD7	= 15,
+   NVS_FR_BFC0		= 16,
+   NVS_FR_BFC1		= 17,
+   NVS_FR_POINTSZ	= 18,
+   NVS_FR_FRAGDATA0	= 19,
+   NVS_FR_FRAGDATA1	= 20,
+   NVS_FR_FRAGDATA2	= 21,
+   NVS_FR_FRAGDATA3	= 22,
+   NVS_FR_CLIP0		= 23,
+   NVS_FR_CLIP1		= 24,
+   NVS_FR_CLIP2		= 25,
+   NVS_FR_CLIP3		= 26,
+   NVS_FR_CLIP4		= 27,
+   NVS_FR_CLIP5		= 28,
+   NVS_FR_CLIP6		= 29,
+   NVS_FR_FACING	= 30,
    NVS_FR_UNKNOWN
 } nvsFixedReg;
 
@@ -190,7 +200,18 @@ typedef enum {
    NVS_TEX_TARGET_UNKNOWN = 0
 } nvsTexTarget;
 
-typedef struct {
+typedef enum {
+	NVS_SCALE_1X	 = 0,
+	NVS_SCALE_2X	 = 1,
+	NVS_SCALE_4X	 = 2,
+	NVS_SCALE_8X	 = 3,
+	NVS_SCALE_INV_2X = 5,
+	NVS_SCALE_INV_4X = 6,
+	NVS_SCALE_INV_8X = 7,
+} nvsScale;
+
+/* Arith/TEX instructions */
+typedef struct nvs_instruction {
    nvsFragmentHeader header;
 
    nvsOpcode	op;
@@ -198,6 +219,7 @@ typedef struct {
 
    nvsRegister	dest;
    unsigned int	mask;
+   nvsScale	dest_scale;
 
    nvsRegister	src[3];
 
@@ -211,6 +233,43 @@ typedef struct {
    int		cond_update;
 } nvsInstruction;
 
+/* BRA, CAL, IF */
+typedef struct nvs_branch {
+	nvsFragmentHeader  header;
+
+	nvsOpcode	op;
+
+	nvsCond		cond;
+	nvsSwzComp	cond_swizzle[4];
+	int		cond_test;
+
+	nvsFragmentHeader *target_head;
+	nvsFragmentHeader *target_tail;
+	nvsFragmentHeader *else_head;
+	nvsFragmentHeader *else_tail;
+} nvsBranch;
+
+/* LOOP+ENDLOOP */
+typedef struct {
+	nvsFragmentHeader  header;
+
+	int                count;
+	int                initial;
+	int                increment;
+
+	nvsFragmentHeader *insn_head;
+	nvsFragmentHeader *insn_tail;
+} nvsLoop;
+
+/* label+following instructions */
+typedef struct nvs_subroutine {
+	nvsFragmentHeader  header;
+
+	char *             label;
+	nvsFragmentHeader *insn_head;
+	nvsFragmentHeader *insn_tail;
+} nvsSubroutine;
+
 #define SMASK_X (1<<0)
 #define SMASK_Y (1<<1)
 #define SMASK_Z (1<<2)
@@ -247,6 +306,8 @@ extern nvsSwzComp NV20VP_TX_SWIZZLE[4];
 #define SCAP_SRC_ABS	(1<<0)
 
 struct _nvsFunc {
+   nvsCardPriv *card_priv;
+
    unsigned int	MaxInst;
    unsigned int	MaxAttrib;
    unsigned int	MaxTemp;
@@ -263,6 +324,7 @@ struct _nvsFunc {
 
    void		(*InitInstruction)	(nvsFunc *);
    int		(*SupportsOpcode)	(nvsFunc *, nvsOpcode);
+   int		(*SupportsResultScale)	(nvsFunc *, nvsScale);
    void		(*SetOpcode)		(nvsFunc *, unsigned int opcode,
 	 				 int slot);
    void		(*SetCCUpdate)		(nvsFunc *);
@@ -270,11 +332,17 @@ struct _nvsFunc {
 	 				 nvsSwzComp *swizzle);
    void		(*SetResult)		(nvsFunc *, nvsRegister *,
 	 				 unsigned int mask, int slot);
+   void		(*SetResultScale)	(nvsFunc *, nvsScale);
    void		(*SetSource)		(nvsFunc *, nvsRegister *, int pos);
    void		(*SetTexImageUnit)	(nvsFunc *, int unit);
    void		(*SetSaturate)		(nvsFunc *);
    void		(*SetLastInst)		(nvsFunc *);
 
+   void		(*SetBranchTarget)	(nvsFunc *, int addr);
+   void		(*SetBranchElse)	(nvsFunc *, int addr);
+   void		(*SetBranchEnd)		(nvsFunc *, int addr);
+   void		(*SetLoopParams)	(nvsFunc *, int cnt, int init, int inc);
+
    int		(*HasMergedInst)	(nvsFunc *);
    int		(*IsLastInst)		(nvsFunc *);
    int		(*GetOffsetNext)	(nvsFunc *);
@@ -352,7 +420,7 @@ nvsSwizzle(nvsRegister reg, nvsSwzComp x, nvsSwzComp y,
 
 extern GLboolean nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs);
 extern void nvsDisasmHWShader(nvsPtr);
-extern void nvsDumpFragmentList(nvsFragmentList *f, int lvl);
+extern void nvsDumpFragmentList(nvsFragmentHeader *f, int lvl);
 extern nouveauShader *nvsBuildTextShader(GLcontext *ctx, GLenum target,
       					 const char *text);
 
@@ -365,8 +433,7 @@ extern void NV40FPInitShaderFuncs(nvsFunc *);
 
 extern void nouveauShaderInitFuncs(GLcontext *ctx);
 
-extern GLboolean nouveau_shader_pass0_arb(GLcontext *ctx, nouveauShader *nvs);
-extern GLboolean nouveau_shader_pass0_slang(GLcontext *ctx, nouveauShader *nvs);
+extern GLboolean nouveau_shader_pass0(GLcontext *ctx, nouveauShader *nvs);
 extern GLboolean nouveau_shader_pass1(nvsPtr nvs);
 extern GLboolean nouveau_shader_pass2(nvsPtr nvs);
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c
new file mode 100644
index 00000000000..73c1f7c2a53
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader_0.c
@@ -0,0 +1,1006 @@
+/*
+ * Copyright (C) 2006 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
+/*
+ * Authors:
+ *   Ben Skeggs <[email protected]>
+ */
+
+#include "glheader.h"
+#include "macros.h"
+#include "enums.h"
+
+#include "program.h"
+#include "programopt.h"
+#include "program_instruction.h"
+
+#include "nouveau_context.h"
+#include "nouveau_shader.h"
+#include "nouveau_msg.h"
+
+static nvsFixedReg _tx_mesa_vp_dst_reg[VERT_RESULT_MAX] = {
+   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
+   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
+   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7,
+   NVS_FR_POINTSZ, NVS_FR_BFC0, NVS_FR_BFC1, NVS_FR_UNKNOWN /* EDGE */
+};
+
+static nvsFixedReg _tx_mesa_fp_dst_reg[FRAG_RESULT_MAX] = {
+   NVS_FR_FRAGDATA0 /* COLR */, NVS_FR_FRAGDATA0 /* COLH */,
+   NVS_FR_UNKNOWN /* DEPR */
+};
+
+static nvsFixedReg _tx_mesa_fp_src_reg[FRAG_ATTRIB_MAX] = {
+   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
+   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
+   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7
+};
+
+static nvsSwzComp _tx_mesa_swizzle[4] = {
+   NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W
+};
+
+static nvsOpcode _tx_mesa_opcode[] = {
+   [OPCODE_ABS] = NVS_OP_ABS, [OPCODE_ADD] = NVS_OP_ADD,
+   [OPCODE_ARA] = NVS_OP_ARA, [OPCODE_ARL] = NVS_OP_ARL,
+   [OPCODE_ARL_NV] = NVS_OP_ARL, [OPCODE_ARR] = NVS_OP_ARR,
+   [OPCODE_CMP] = NVS_OP_CMP, [OPCODE_COS] = NVS_OP_COS,
+   [OPCODE_DDX] = NVS_OP_DDX, [OPCODE_DDY] = NVS_OP_DDY,
+   [OPCODE_DP3] = NVS_OP_DP3, [OPCODE_DP4] = NVS_OP_DP4,
+   [OPCODE_DPH] = NVS_OP_DPH, [OPCODE_DST] = NVS_OP_DST,
+   [OPCODE_EX2] = NVS_OP_EX2, [OPCODE_EXP] = NVS_OP_EXP,
+   [OPCODE_FLR] = NVS_OP_FLR, [OPCODE_FRC] = NVS_OP_FRC,
+   [OPCODE_KIL] = NVS_OP_EMUL, [OPCODE_KIL_NV] = NVS_OP_KIL,
+   [OPCODE_LG2] = NVS_OP_LG2, [OPCODE_LIT] = NVS_OP_LIT,
+   [OPCODE_LOG] = NVS_OP_LOG,
+   [OPCODE_LRP] = NVS_OP_LRP,
+   [OPCODE_MAD] = NVS_OP_MAD, [OPCODE_MAX] = NVS_OP_MAX,
+   [OPCODE_MIN] = NVS_OP_MIN, [OPCODE_MOV] = NVS_OP_MOV,
+   [OPCODE_MUL] = NVS_OP_MUL,
+   [OPCODE_PK2H] = NVS_OP_PK2H, [OPCODE_PK2US] = NVS_OP_PK2US,
+   [OPCODE_PK4B] = NVS_OP_PK4B, [OPCODE_PK4UB] = NVS_OP_PK4UB,
+   [OPCODE_POW] = NVS_OP_POW, [OPCODE_POPA] = NVS_OP_POPA,
+   [OPCODE_PUSHA] = NVS_OP_PUSHA,
+   [OPCODE_RCC] = NVS_OP_RCC, [OPCODE_RCP] = NVS_OP_RCP,
+   [OPCODE_RFL] = NVS_OP_RFL, [OPCODE_RSQ] = NVS_OP_RSQ,
+   [OPCODE_SCS] = NVS_OP_SCS, [OPCODE_SEQ] = NVS_OP_SEQ,
+   [OPCODE_SFL] = NVS_OP_SFL, [OPCODE_SGE] = NVS_OP_SGE,
+   [OPCODE_SGT] = NVS_OP_SGT, [OPCODE_SIN] = NVS_OP_SIN,
+   [OPCODE_SLE] = NVS_OP_SLE, [OPCODE_SLT] = NVS_OP_SLT,
+   [OPCODE_SNE] = NVS_OP_SNE, [OPCODE_SSG] = NVS_OP_SSG,
+   [OPCODE_STR] = NVS_OP_STR, [OPCODE_SUB] = NVS_OP_SUB,
+   [OPCODE_SWZ] = NVS_OP_MOV,
+   [OPCODE_TEX] = NVS_OP_TEX, [OPCODE_TXB] = NVS_OP_TXB,
+   [OPCODE_TXD] = NVS_OP_TXD,
+   [OPCODE_TXL] = NVS_OP_TXL, [OPCODE_TXP] = NVS_OP_TXP,
+   [OPCODE_TXP_NV] = NVS_OP_TXP,
+   [OPCODE_UP2H] = NVS_OP_UP2H, [OPCODE_UP2US] = NVS_OP_UP2US,
+   [OPCODE_UP4B] = NVS_OP_UP4B, [OPCODE_UP4UB] = NVS_OP_UP4UB,
+   [OPCODE_X2D] = NVS_OP_X2D,
+   [OPCODE_XPD] = NVS_OP_XPD
+};
+
+static nvsCond _tx_mesa_condmask[] = {
+   NVS_COND_TR, /* workaround mesa not filling a valid value */
+   NVS_COND_GT, NVS_COND_LT, NVS_COND_UN, NVS_COND_GE,
+   NVS_COND_LE, NVS_COND_NE, NVS_COND_NE, NVS_COND_TR, NVS_COND_FL
+};
+
+struct pass0_rec {
+   int nvs_ipos;
+   int next_temp;
+
+   int mesa_const_base;
+   int mesa_const_last;
+
+   int swzconst_done;
+   int swzconst_id;
+   nvsRegister const_half;
+};
+
+#define X NVS_SWZ_X
+#define Y NVS_SWZ_Y
+#define Z NVS_SWZ_Z
+#define W NVS_SWZ_W
+
+#define FILL_CONDITION_FLAGS(fragment) do {                                    \
+	(fragment)->cond =                                                     \
+	pass0_make_condmask(inst->DstReg.CondMask);                            \
+	if ((fragment)->cond != NVS_COND_TR)                                   \
+		(fragment)->cond_test = 1;                                     \
+	(fragment)->cond_reg = inst->CondDst;                                  \
+	pass0_make_swizzle((fragment)->cond_swizzle, inst->DstReg.CondSwizzle);\
+} while(0)
+
+#define ARITH(op,dest,mask,sat,s0,s1,s2) do {               \
+	nvsinst = pass0_emit(nvs, parent, fpos, (op),       \
+		   (dest), (mask), (sat), (s0), (s1), (s2));\
+	FILL_CONDITION_FLAGS(nvsinst);                      \
+} while(0)
+
+#define ARITHu(op,dest,mask,sat,s0,s1,s2) do {              \
+	nvsinst = pass0_emit(nvs, parent, fpos, (op),       \
+		   (dest), (mask), (sat), (s0), (s1), (s2));\
+} while(0)
+
+static void
+pass0_append_fragment(nvsFragmentHeader *parent,
+		      nvsFragmentHeader *fragment,
+		      int pos)
+{
+	nvsFragmentHeader **head, **tail;
+	assert(parent && fragment);
+
+	switch (parent->type) {
+	case NVS_BRANCH:
+		if (pos == 0) {
+			head = &((nvsBranch *)parent)->target_head;
+			tail = &((nvsBranch *)parent)->target_tail;
+		} else {
+			head = &((nvsBranch *)parent)->else_head;
+			tail = &((nvsBranch *)parent)->else_tail;
+		}
+		break;
+	case NVS_LOOP:
+		head = &((nvsLoop *)parent)->insn_head;
+		tail = &((nvsLoop *)parent)->insn_tail;
+		break;
+	case NVS_SUBROUTINE:
+		head = &((nvsSubroutine *)parent)->insn_head;
+		tail = &((nvsSubroutine *)parent)->insn_tail;
+		break;
+	default:
+		assert(0);
+		break;
+	}
+
+	fragment->parent = parent;
+	fragment->prev   = *tail;
+	fragment->next   = NULL;
+	if (!(*head))
+		*head = fragment;
+	else
+		(*tail)->next = fragment;
+	*tail = fragment;
+
+}
+
+static nvsSubroutine *
+pass0_create_subroutine(nouveauShader *nvs, const char *label)
+{
+	nvsSubroutine *sub;
+
+	sub = CALLOC_STRUCT(nvs_subroutine);
+	if (sub) {
+		sub->header.type = NVS_SUBROUTINE;
+		sub->label	 = strdup(label);
+		if (!nvs->program_tree)
+			nvs->program_tree = &sub->header;
+		else
+			pass0_append_fragment(nvs->program_tree,
+					      &sub->header, 0);
+	}
+
+	return sub;
+}
+
+static void
+pass0_make_reg(nouveauShader *nvs, nvsRegister *reg,
+	       nvsRegFile file, unsigned int index)
+{
+	struct pass0_rec *rec = nvs->pass_rec;
+
+	/* defaults */
+	*reg = nvr_unused;
+	/* -1 == quick-and-dirty temp alloc */
+	if (file == NVS_FILE_TEMP && index == -1) {
+		index = rec->next_temp++;
+		assert(index < NVS_MAX_TEMPS);
+	}
+	reg->file   = file;
+	reg->index  = index;
+}
+
+static void
+pass0_make_swizzle(nvsSwzComp *swz, unsigned int mesa)
+{
+	int i;
+
+	for (i=0;i<4;i++)
+		swz[i] = _tx_mesa_swizzle[GET_SWZ(mesa, i)];
+}
+
+static nvsOpcode
+pass0_make_opcode(enum prog_opcode op)
+{
+	if (op > MAX_OPCODE)
+		return NVS_OP_UNKNOWN;
+	return _tx_mesa_opcode[op];
+}
+
+static nvsCond
+pass0_make_condmask(GLuint mesa)
+{
+	if (mesa > COND_FL)
+		return NVS_COND_UNKNOWN;
+	return _tx_mesa_condmask[mesa];
+}
+
+static unsigned int
+pass0_make_mask(GLuint mesa_mask)
+{
+	unsigned int mask = 0;
+
+	if (mesa_mask & WRITEMASK_X) mask |= SMASK_X;
+	if (mesa_mask & WRITEMASK_Y) mask |= SMASK_Y;
+	if (mesa_mask & WRITEMASK_Z) mask |= SMASK_Z;
+	if (mesa_mask & WRITEMASK_W) mask |= SMASK_W;
+
+	return mask;
+}
+
+static nvsTexTarget
+pass0_make_tex_target(GLuint mesa)
+{
+	switch (mesa) {
+	case TEXTURE_1D_INDEX: return NVS_TEX_TARGET_1D;
+	case TEXTURE_2D_INDEX: return NVS_TEX_TARGET_2D;
+	case TEXTURE_3D_INDEX: return NVS_TEX_TARGET_3D;
+	case TEXTURE_CUBE_INDEX: return NVS_TEX_TARGET_CUBE;
+	case TEXTURE_RECT_INDEX: return NVS_TEX_TARGET_RECT;
+	default:
+		return NVS_TEX_TARGET_UNKNOWN;
+	}
+}
+
+static void
+pass0_make_dst_reg(nvsPtr nvs, nvsRegister *reg,
+      		   struct prog_dst_register *dst)
+{
+	struct gl_program *mesa = (struct gl_program*)&nvs->mesa.vp;
+	nvsFixedReg sfr;
+
+	switch (dst->File) {
+	case PROGRAM_OUTPUT:
+		if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
+			sfr = (dst->Index < VERT_RESULT_MAX) ?
+				_tx_mesa_vp_dst_reg[dst->Index] :
+				NVS_FR_UNKNOWN;
+		} else {
+			sfr = (dst->Index < FRAG_RESULT_MAX) ?
+				_tx_mesa_fp_dst_reg[dst->Index] :
+				NVS_FR_UNKNOWN;
+		}
+		pass0_make_reg(nvs, reg, NVS_FILE_RESULT, sfr);
+		break;
+	case PROGRAM_TEMPORARY:
+		pass0_make_reg(nvs, reg, NVS_FILE_TEMP, dst->Index);
+		break;
+	case PROGRAM_ADDRESS:
+		pass0_make_reg(nvs, reg, NVS_FILE_ADDRESS, dst->Index);
+		break;
+	default:
+		fprintf(stderr, "Unknown dest file %d\n", dst->File);
+		assert(0);
+	}
+}
+
+static void
+pass0_make_src_reg(nvsPtr nvs, nvsRegister *reg, struct prog_src_register *src)
+{
+	struct pass0_rec *rec = nvs->pass_rec;
+	struct gl_program *mesa = (struct gl_program *)&nvs->mesa.vp.Base;
+	int i;
+
+	*reg = nvr_unused;
+
+	switch (src->File) {
+	case PROGRAM_INPUT:
+		reg->file = NVS_FILE_ATTRIB;
+		if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
+			for (i=0; i<NVS_MAX_ATTRIBS; i++) {
+				if (nvs->vp_attrib_map[i] == src->Index) {
+					reg->index = i;
+					break;
+				}
+			}
+			if (i==NVS_MAX_ATTRIBS)
+				reg->index = NVS_FR_UNKNOWN;
+		} else {
+			reg->index = (src->Index < FRAG_ATTRIB_MAX) ?
+				_tx_mesa_fp_src_reg[src->Index] :
+				NVS_FR_UNKNOWN;
+		}
+		break;
+	case PROGRAM_STATE_VAR:
+	case PROGRAM_NAMED_PARAM:
+	case PROGRAM_CONSTANT:
+		reg->file    = NVS_FILE_CONST;
+		reg->index   = src->Index + rec->mesa_const_base;
+		reg->indexed = src->RelAddr;
+		if (reg->indexed) {
+			reg->addr_reg  = 0;
+			reg->addr_comp = NVS_SWZ_X;
+		}
+		break;
+	case PROGRAM_TEMPORARY:
+		reg->file  = NVS_FILE_TEMP;
+		reg->index = src->Index;
+		break;
+	default:
+		fprintf(stderr, "Unknown source type %d\n", src->File);
+		assert(0);
+	}
+
+	/* per-component negate handled elsewhere */
+	reg->negate = src->NegateBase != 0;
+	reg->abs    = src->Abs;
+	pass0_make_swizzle(reg->swizzle, src->Swizzle);
+}
+
+static nvsInstruction *
+pass0_emit(nouveauShader *nvs, nvsFragmentHeader *parent, int fpos,
+	   nvsOpcode op, nvsRegister dst,
+	   unsigned int mask, int saturate,
+	   nvsRegister src0, nvsRegister src1, nvsRegister src2)
+{
+	nvsInstruction *sif;
+
+	sif = CALLOC_STRUCT(nvs_instruction);
+	if (!sif)
+		return NULL;
+
+	/* Seems mesa doesn't explicitly 0 this.. */
+	if (nvs->mesa.vp.Base.Target == GL_VERTEX_PROGRAM_ARB)
+		saturate = 0;
+
+	sif->op		= op;
+	sif->saturate	= saturate;
+	sif->dest	= dst;
+	sif->mask	= mask;
+	sif->dest_scale	= NVS_SCALE_1X;
+	sif->src[0]	= src0;
+	sif->src[1]	= src1;
+	sif->src[2]	= src2;
+	sif->cond	= COND_TR;
+	sif->cond_reg	= 0;
+	sif->cond_test	= 0;
+	sif->cond_update= 0;
+	pass0_make_swizzle(sif->cond_swizzle, SWIZZLE_NOOP);
+	pass0_append_fragment(parent, &sif->header, fpos);
+
+	return sif;
+}
+
+static void
+pass0_fixup_swizzle(nvsPtr nvs, nvsFragmentHeader *parent, int fpos,
+		    struct prog_src_register *src,
+		    unsigned int sm1,
+		    unsigned int sm2)
+{
+	static const float sc[4] = { 1.0, 0.0, -1.0, 0.0 };
+	struct pass0_rec *rec = nvs->pass_rec;
+	int fixup_1, fixup_2;
+	nvsInstruction *nvsinst;
+	nvsRegister sr, dr = nvr_unused;
+	nvsRegister sm1const, sm2const;
+
+	if (!rec->swzconst_done) {
+		struct gl_program *prog = &nvs->mesa.vp.Base;
+		rec->swzconst_id = _mesa_add_unnamed_constant(prog->Parameters,
+							      sc, 4);
+		rec->swzconst_done = 1;
+		COPY_4V(nvs->params[rec->swzconst_id].val, sc);
+	}
+
+	fixup_1 = (sm1 != MAKE_SWIZZLE4(0,0,0,0) &&
+			sm2 != MAKE_SWIZZLE4(2,2,2,2));
+	fixup_2 = (sm2 != MAKE_SWIZZLE4(2,2,2,2));
+
+	if (src->File != PROGRAM_TEMPORARY && src->File != PROGRAM_INPUT) {
+		/* We can't use more than one const in an instruction,
+		 * so move the const into a temp, and swizzle from there.
+		 *
+		 * TODO: should just emit the swizzled const, instead of
+		 *       swizzling it in the shader.. would need to reswizzle
+		 *       any state params when they change however..
+		 */
+		pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
+		pass0_make_src_reg(nvs, &sr, src);
+		ARITHu(NVS_OP_MOV, dr, SMASK_ALL, 0,
+				 sr, nvr_unused, nvr_unused);
+		pass0_make_reg(nvs, &sr, NVS_FILE_TEMP, dr.index);
+	} else {
+		if (fixup_1)
+			src->NegateBase = 0;
+		pass0_make_src_reg(nvs, &sr, src);
+		pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
+	}
+
+	pass0_make_reg(nvs, &sm1const, NVS_FILE_CONST, rec->swzconst_id);
+	pass0_make_swizzle(sm1const.swizzle, sm1);
+	if (fixup_1 && fixup_2) {
+		/* Any combination with SWIZZLE_ONE */
+		pass0_make_reg(nvs, &sm2const,
+				    NVS_FILE_CONST, rec->swzconst_id);
+		pass0_make_swizzle(sm2const.swizzle, sm2);
+		ARITHu(NVS_OP_MAD, dr, SMASK_ALL, 0, sr, sm1const, sm2const);
+	} else {
+		/* SWIZZLE_ZERO || arbitrary negate */
+		ARITHu(NVS_OP_MUL, dr, SMASK_ALL, 0, sr, sm1const, nvr_unused);
+	}
+
+	src->File	= PROGRAM_TEMPORARY;
+	src->Index	= dr.index;
+	src->Swizzle	= SWIZZLE_NOOP;
+}
+
+#define SET_SWZ(fs, cp, c) fs = (fs & ~(0x7<<(cp*3))) | (c<<(cp*3))
+static void
+pass0_check_sources(nvsPtr nvs, nvsFragmentHeader *parent, int fpos,
+		    struct prog_instruction *inst)
+{
+	unsigned int insrc = -1, constsrc = -1;
+	int i;
+
+	for (i=0;i<_mesa_num_inst_src_regs(inst->Opcode);i++) {
+		struct prog_src_register *src = &inst->SrcReg[i];
+		unsigned int sm_1 = 0, sm_2 = 0;
+		nvsRegister sr, dr;
+		int do_mov = 0, c;
+
+		/* Build up swizzle masks as if we were going to use
+		 * "MAD new, src, const1, const2" to support arbitrary negation
+		 * and SWIZZLE_ZERO/SWIZZLE_ONE.
+		 */
+		for (c=0;c<4;c++) {
+			if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ZERO) {
+				SET_SWZ(sm_1, c, SWIZZLE_Y); /* 0.0 */
+				SET_SWZ(sm_2, c, SWIZZLE_Y);
+				SET_SWZ(src->Swizzle, c, SWIZZLE_X);
+			} else if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ONE) {
+				SET_SWZ(sm_1, c, SWIZZLE_Y);
+				if (src->NegateBase & (1<<c))
+					SET_SWZ(sm_2, c, SWIZZLE_Z); /* -1.0 */
+				else
+					SET_SWZ(sm_2, c, SWIZZLE_X); /* 1.0 */
+				SET_SWZ(src->Swizzle, c, SWIZZLE_X);
+			} else {
+				if (src->NegateBase & (1<<c))
+					SET_SWZ(sm_1, c, SWIZZLE_Z); /* -[xyzw] */
+				else
+					SET_SWZ(sm_1, c, SWIZZLE_X); /*[xyzw]*/
+				SET_SWZ(sm_2, c, SWIZZLE_Y);
+			}
+		}
+
+		/* Unless we're multiplying by 1.0 or -1.0 on all components,
+		 * and we're adding nothing to any component we have to
+		 * emulate the swizzle.
+		 */
+		if ((sm_1 != MAKE_SWIZZLE4(0,0,0,0) &&
+					sm_1 != MAKE_SWIZZLE4(2,2,2,2)) ||
+				sm_2 != MAKE_SWIZZLE4(1,1,1,1)) {
+			pass0_fixup_swizzle(nvs, parent, fpos, src, sm_1, sm_2);
+			/* The source is definitely in a temp now, so don't
+			 * bother checking  for multiple ATTRIB/CONST regs.
+			 */
+			continue;
+		}
+
+		/* HW can't use more than one ATTRIB or PARAM in a single
+		 * instruction */
+		switch (src->File) {
+		case PROGRAM_INPUT:
+			if (insrc != -1 && insrc != src->Index) 
+				do_mov = 1;
+			else insrc = src->Index;
+			break;
+		case PROGRAM_STATE_VAR:
+			if (constsrc != -1 && constsrc != src->Index)
+				do_mov = 1;
+			else constsrc = src->Index;
+			break;
+		default:
+			break;
+		}
+
+		/* Emit any extra ATTRIB/CONST to a temp, and modify the Mesa
+		 * instruction to point at the temp.
+		 */
+		if (do_mov) {
+			pass0_make_src_reg(nvs, &sr, src);
+			pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
+			pass0_emit(nvs, parent, fpos, NVS_OP_MOV,
+					dr, SMASK_ALL, 0,
+					sr, nvr_unused, nvr_unused);
+
+			src->File	= PROGRAM_TEMPORARY;
+			src->Index	= dr.index;
+			src->Swizzle= SWIZZLE_NOOP;
+		}
+	}
+}
+
+static GLboolean
+pass0_emulate_instruction(nouveauShader *nvs,
+			  nvsFragmentHeader *parent, int fpos,
+			  struct prog_instruction *inst)
+{
+	nvsFunc *shader = nvs->func;
+	nvsRegister src[3], dest, temp;
+	nvsInstruction *nvsinst;
+	unsigned int mask = pass0_make_mask(inst->DstReg.WriteMask);
+	int i, sat;
+
+	sat = (inst->SaturateMode == SATURATE_ZERO_ONE);
+
+	/* Build all the "real" regs for the instruction */
+	for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
+		pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
+	if (inst->Opcode != OPCODE_KIL)
+		pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
+
+	switch (inst->Opcode) {
+	case OPCODE_ABS:
+		if (shader->caps & SCAP_SRC_ABS)
+			ARITH(NVS_OP_MOV, dest, mask, sat,
+					nvsAbs(src[0]), nvr_unused, nvr_unused);
+		else
+			ARITH(NVS_OP_MAX, dest, mask, sat,
+					src[0], nvsNegate(src[0]), nvr_unused);
+		break;
+	case OPCODE_CMP:
+		/*XXX: this will clobber CC0... */
+		ARITH (NVS_OP_MOV, dest, mask, sat,
+				src[2], nvr_unused, nvr_unused);
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_MOV, temp, SMASK_ALL, 0,
+				src[0], nvr_unused, nvr_unused);
+		nvsinst->cond_update = 1;
+		nvsinst->cond_reg    = 0;
+		ARITH (NVS_OP_MOV, dest, mask, sat,
+				src[1], nvr_unused, nvr_unused);
+		nvsinst->cond      = COND_LT;
+		nvsinst->cond_reg  = 0;
+		nvsinst->cond_test = 1;
+		break;
+	case OPCODE_DPH:
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_DP3, temp, SMASK_X, 0,
+				 src[0], src[1], nvr_unused);
+		ARITH (NVS_OP_ADD, dest, mask, sat,
+				 nvsSwizzle(temp, X, X, X, X),
+				 nvsSwizzle(src[1], W, W, W, W),
+				 nvr_unused);
+		break;
+	case OPCODE_KIL:
+		/* This is only in ARB shaders, so we don't have to worry
+		 * about clobbering a CC reg as they aren't supported anyway.
+		 *XXX: might have to worry with GLSL however...
+		 */
+		/* MOVC0 temp, src */
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_MOV, temp, SMASK_ALL, 0,
+				 src[0], nvr_unused, nvr_unused);
+		nvsinst->cond_update = 1;
+		nvsinst->cond_reg    = 0;
+		/* KIL_NV (LT0.xyzw) temp */
+		ARITHu(NVS_OP_KIL, nvr_unused, 0, 0,
+				 nvr_unused, nvr_unused, nvr_unused);
+		nvsinst->cond      = COND_LT;
+		nvsinst->cond_reg  = 0;
+		nvsinst->cond_test = 1;
+		pass0_make_swizzle(nvsinst->cond_swizzle, SWIZZLE_NOOP);
+		break;
+	case OPCODE_LRP:
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_MAD, temp, mask, 0,
+				 nvsNegate(src[0]), src[2], src[2]);
+		ARITH (NVS_OP_MAD, dest, mask, sat, src[0], src[1], temp);
+		break;
+	case OPCODE_POW:
+		if (shader->SupportsOpcode(shader, NVS_OP_LG2) &&
+				shader->SupportsOpcode(shader, NVS_OP_EX2)) {
+			pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+			/* LG2 temp.x, src0.c */
+			ARITHu(NVS_OP_LG2, temp, SMASK_X, 0,
+					 nvsSwizzle(src[0], X, X, X, X),
+					 nvr_unused, nvr_unused);
+			/* MUL temp.x, temp.x, src1.c */
+			ARITHu(NVS_OP_MUL, temp, SMASK_X, 0,
+					 nvsSwizzle(temp, X, X, X, X),
+					 nvsSwizzle(src[1], X, X, X, X),
+					 nvr_unused);
+			/* EX2 dest, temp.x */
+			ARITH (NVS_OP_EX2, dest, mask, sat,
+					 nvsSwizzle(temp, X, X, X, X),
+					 nvr_unused, nvr_unused);
+		} else {
+			/* can we use EXP/LOG instead of EX2/LG2?? */
+			fprintf(stderr, "Implement POW for NV20 vtxprog!\n");
+			return GL_FALSE;
+		}
+		break;
+	case OPCODE_RSQ:
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_LG2, temp, SMASK_X, 0,
+				 nvsAbs(nvsSwizzle(src[0], X, X, X, X)),
+				 nvr_unused, nvr_unused);
+		nvsinst->dest_scale = NVS_SCALE_INV_2X;
+		ARITH (NVS_OP_EX2, dest, mask, sat,
+				 nvsNegate(nvsSwizzle(temp, X, X, X, X)),
+				 nvr_unused, nvr_unused);
+		break;
+	case OPCODE_SCS:
+		if (mask & SMASK_X)
+			ARITH(NVS_OP_COS, dest, SMASK_X, sat,
+					nvsSwizzle(src[0], X, X, X, X),
+					nvr_unused, nvr_unused);
+		if (mask & SMASK_Y)
+			ARITH(NVS_OP_SIN, dest, SMASK_Y, sat,
+					nvsSwizzle(src[0], X, X, X, X),
+					nvr_unused, nvr_unused);
+		break;
+	case OPCODE_SUB:
+		ARITH(NVS_OP_ADD, dest, mask, sat,
+				src[0], nvsNegate(src[1]), nvr_unused);
+		break;
+	case OPCODE_XPD:
+		pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
+		ARITHu(NVS_OP_MUL, temp, SMASK_ALL, 0,
+				 nvsSwizzle(src[0], Z, X, Y, Y),
+				 nvsSwizzle(src[1], Y, Z, X, X),
+				 nvr_unused);
+		ARITH (NVS_OP_MAD, dest, (mask & ~SMASK_W), sat,
+				 nvsSwizzle(src[0], Y, Z, X, X),
+				 nvsSwizzle(src[1], Z, X, Y, Y),
+				 nvsNegate(temp));
+		break;
+	default:
+		WARN_ONCE("hw doesn't support opcode \"%s\","
+			  "and no emulation found\n",
+			  _mesa_opcode_string(inst->Opcode));
+		return GL_FALSE;
+	}
+
+	return GL_TRUE;
+}
+
+static GLboolean
+pass0_translate_arith(nouveauShader *nvs, struct gl_program *prog,
+					  int ipos, int fpos,
+					  nvsFragmentHeader *parent)
+{
+	struct prog_instruction *inst = &prog->Instructions[ipos];
+	nvsFunc *shader = nvs->func;
+	nvsInstruction *nvsinst;
+	GLboolean ret;
+
+	/* Deal with multiple ATTRIB/PARAM in a single instruction */
+	pass0_check_sources(nvs, parent, fpos, inst);
+
+	/* Now it's safe to do the prog_instruction->nvsInstruction
+	 * conversion
+	 */
+	if (shader->SupportsOpcode(shader,
+				   pass0_make_opcode(inst->Opcode))) {
+		nvsRegister src[3], dest;
+		int i;
+
+		for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
+			pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
+		pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
+
+		ARITH(pass0_make_opcode(inst->Opcode), dest,
+				pass0_make_mask(inst->DstReg.WriteMask),
+				(inst->SaturateMode != SATURATE_OFF),
+				src[0], src[1], src[2]);
+		nvsinst->tex_unit   = inst->TexSrcUnit;
+		nvsinst->tex_target = pass0_make_tex_target(inst->TexSrcTarget);
+
+		ret = GL_TRUE;
+	} else
+		ret = pass0_emulate_instruction(nvs, parent, fpos, inst);
+
+	return ret;
+}
+
+static GLboolean
+pass0_translate_instructions(nouveauShader *nvs, int ipos, int fpos,
+						 nvsFragmentHeader *parent)
+{
+	struct gl_program *prog = (struct gl_program *)&nvs->mesa.vp;
+
+	while (1) {
+		struct prog_instruction *inst = &prog->Instructions[ipos];
+
+		switch (inst->Opcode) {
+		case OPCODE_END:
+			return GL_TRUE;
+		case OPCODE_BRA:
+		case OPCODE_CAL:
+		case OPCODE_RET:
+		//case OPCODE_LOOP:
+		//case OPCODE_ENDLOOP:
+		//case OPCODE_IF:
+		//case OPCODE_ELSE:
+		//case OPCODE_ENDIF:
+			WARN_ONCE("branch ops unimplemented\n");
+			return GL_FALSE;
+			break;
+		default:
+			if (!pass0_translate_arith(nvs, prog,
+						   ipos, fpos, parent))
+				return GL_FALSE;
+			break;
+		}
+
+		ipos++;
+	}
+
+	return GL_TRUE;
+}
+
+static void
+pass0_build_attrib_map(nouveauShader *nvs, struct gl_vertex_program *vp)
+{
+	GLuint inputs_read = vp->Base.InputsRead;
+	GLuint input_alloc = ~0xFFFF;
+	int i;
+
+	for (i=0; i<NVS_MAX_ATTRIBS; i++)
+		nvs->vp_attrib_map[i] = -1;
+
+	while (inputs_read) {
+		int in = ffs(inputs_read) - 1;
+		int hw;
+		inputs_read &= ~(1<<in);
+
+		if (vp->IsNVProgram) {
+			/* NVvp: must alias */
+			if (in >= VERT_ATTRIB_GENERIC0)
+				hw = in - VERT_ATTRIB_GENERIC0;
+			else
+				hw = in;
+		} else {
+			/* ARBvp: may alias (but we won't)
+			 * GL2.0: must not alias
+			 */
+			if (in >= VERT_ATTRIB_GENERIC0)
+				hw = ffs(~input_alloc) - 1;
+			else 
+				hw = in;
+			input_alloc |= (1<<hw);
+		}
+
+		nvs->vp_attrib_map[hw] = in;
+	}
+
+	if (NOUVEAU_DEBUG & DEBUG_SHADERS) {
+		printf("vtxprog attrib map:\n");
+		for (i=0; i<NVS_MAX_ATTRIBS; i++) {
+			printf(" hw:%d = attrib:%d\n",
+					i, nvs->vp_attrib_map[i]);
+		}
+	}
+}
+
+static void
+pass0_vp_insert_ff_clip_planes(GLcontext *ctx, nouveauShader *nvs)
+{
+	struct gl_program *prog = &nvs->mesa.vp.Base;
+	nvsFragmentHeader *parent = nvs->program_tree;
+	nvsInstruction *nvsinst;
+	GLuint fpos = 0;
+	nvsRegister opos, epos, eqn, mv[4];
+	GLint tokens[6] = { STATE_MATRIX, STATE_MODELVIEW, 0, 0, 0, 0 };
+	GLint id;
+	int i;
+
+	/* modelview transform */
+	pass0_make_reg(nvs, &opos, NVS_FILE_ATTRIB, NVS_FR_POSITION);
+	pass0_make_reg(nvs, &epos, NVS_FILE_TEMP  , -1);
+	for (i=0; i<4; i++) {
+		tokens[3] = tokens[4] = i;
+		id = _mesa_add_state_reference(prog->Parameters, tokens);
+		pass0_make_reg(nvs, &mv[i], NVS_FILE_CONST, id);
+	}
+	ARITHu(NVS_OP_DP4, epos, SMASK_X, 0, opos, mv[0], nvr_unused);
+	ARITHu(NVS_OP_DP4, epos, SMASK_Y, 0, opos, mv[1], nvr_unused);
+	ARITHu(NVS_OP_DP4, epos, SMASK_Z, 0, opos, mv[2], nvr_unused);
+	ARITHu(NVS_OP_DP4, epos, SMASK_W, 0, opos, mv[3], nvr_unused);
+
+	/* Emit code to emulate fixed-function glClipPlane */
+	for (i=0; i<6; i++) {
+		GLuint clipmask = SMASK_X;
+		nvsRegister clip;
+
+		if (!(ctx->Transform.ClipPlanesEnabled & (1<<i)))
+			continue;
+
+		/* Point a const at a user clipping plane */
+		tokens[0] = STATE_CLIPPLANE;
+		tokens[1] = i;
+		id = _mesa_add_state_reference(prog->Parameters, tokens);
+		pass0_make_reg(nvs, &eqn , NVS_FILE_CONST , id);
+		pass0_make_reg(nvs, &clip, NVS_FILE_RESULT, NVS_FR_CLIP0 + i);
+
+		/*XXX: something else needs to take care of modifying the
+		 *     instructions to write to the correct hw clip register.
+		 */
+		switch (i) {
+		case 0: case 3:	clipmask = SMASK_Y; break;
+		case 1: case 4: clipmask = SMASK_Z; break;
+		case 2: case 5: clipmask = SMASK_W; break;
+		}
+
+		/* Emit transform */
+		ARITHu(NVS_OP_DP4, clip, clipmask, 0, epos, eqn, nvr_unused);
+	}
+}
+
+static void
+pass0_rebase_mesa_consts(nouveauShader *nvs)
+{
+	struct pass0_rec *rec = nvs->pass_rec;
+	struct gl_program *prog = &nvs->mesa.vp.Base;
+	struct prog_instruction *inst = prog->Instructions;
+	int i;
+
+	/*XXX: not a good idea, params->hw_index is malloc'd */
+	memset(nvs->params, 0x00, sizeof(nvs->params));
+
+	/* When doing relative addressing on constants, the hardware needs us
+	 * to fill the "const id" field with a positive value.  Determine the
+	 * most negative index that is used so that all accesses to a
+	 * mesa-provided constant can be rebased to a positive index.
+	 */
+	while (inst->Opcode != OPCODE_END) {
+		for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++) {
+			struct prog_src_register *src = &inst->SrcReg[i];
+
+			switch (src->File) {
+			case PROGRAM_STATE_VAR:
+			case PROGRAM_CONSTANT:
+			case PROGRAM_NAMED_PARAM:
+				if (src->RelAddr && src->Index < 0) {
+					int base = src->Index * -1;
+					if (rec->mesa_const_base < base)
+						rec->mesa_const_base = base;
+				}
+				break;
+			default:
+				break;
+			}
+		}
+
+		inst++;
+	}
+}
+
+static void
+pass0_resolve_mesa_consts(nouveauShader *nvs)
+{
+	struct pass0_rec *rec = nvs->pass_rec;
+	struct gl_program *prog = &nvs->mesa.vp.Base;
+	struct gl_program_parameter_list *plist = prog->Parameters;
+	int i;
+
+	/* Init all const tracking/alloc info from the parameter list, rather
+	 * than doing it as we translate the program.  Otherwise:
+	 *   1) we can't get at the correct constant info when relative
+	 *      addressing is being used due to src->Index not pointing
+	 *      at the exact const;
+	 *   2) as we add extra consts to the program, mesa will call realloc()
+	 *      and we get invalid pointers to the const data.
+	 */
+	rec->mesa_const_last = plist->NumParameters + rec->mesa_const_base;
+	nvs->param_high = rec->mesa_const_last;
+	for (i=0; i<plist->NumParameters; i++) {
+		int hw = rec->mesa_const_base + i;
+
+		switch (plist->Parameters[i].Type) {
+		case PROGRAM_NAMED_PARAM:
+		case PROGRAM_STATE_VAR:
+			nvs->params[hw].in_use     = GL_TRUE;
+			nvs->params[hw].source_val = plist->ParameterValues[i];
+			COPY_4V(nvs->params[hw].val, plist->ParameterValues[i]);
+			break;
+		case PROGRAM_CONSTANT:
+			nvs->params[hw].in_use     = GL_TRUE;
+			nvs->params[hw].source_val = NULL;
+			COPY_4V(nvs->params[hw].val, plist->ParameterValues[i]);
+			break;
+		default:
+			assert(0);
+			break;
+		}
+	}
+}
+
+GLboolean
+nouveau_shader_pass0(GLcontext *ctx, nouveauShader *nvs)
+{
+	nouveauContextPtr           nmesa = NOUVEAU_CONTEXT(ctx);
+	struct gl_program          *prog  = (struct gl_program*)nvs;
+	struct gl_vertex_program   *vp    = (struct gl_vertex_program *)prog;
+	struct gl_fragment_program *fp    = (struct gl_fragment_program *)prog;
+	struct pass0_rec *rec;
+	int ret = GL_FALSE;
+
+	rec = CALLOC_STRUCT(pass0_rec);
+	if (!rec)
+		return GL_FALSE;
+
+	rec->next_temp = prog->NumTemporaries;
+	nvs->pass_rec  = rec;
+		
+	nvs->program_tree = (nvsFragmentHeader*)
+		pass0_create_subroutine(nvs, "program body");
+	if (!nvs->program_tree) {
+		FREE(rec);
+		return GL_FALSE;
+	}
+
+	switch (prog->Target) {
+	case GL_VERTEX_PROGRAM_ARB:
+		nvs->func = &nmesa->VPfunc;
+
+		if (vp->IsPositionInvariant)
+			_mesa_insert_mvp_code(ctx, vp);
+		pass0_rebase_mesa_consts(nvs);
+
+		if (!prog->String && ctx->Transform.ClipPlanesEnabled)
+			pass0_vp_insert_ff_clip_planes(ctx, nvs);
+
+		pass0_build_attrib_map(nvs, vp);
+		break;
+	case GL_FRAGMENT_PROGRAM_ARB:
+		nvs->func = &nmesa->FPfunc;
+
+		if (fp->FogOption != GL_NONE)
+			_mesa_append_fog_code(ctx, fp);
+		pass0_rebase_mesa_consts(nvs);
+		break;
+	default:
+		fprintf(stderr, "Unknown program type %d", prog->Target);
+		FREE(rec);
+		/* DESTROY TREE!! */
+		return GL_FALSE;
+	}
+	nvs->func->card_priv = &nvs->card_priv;
+
+	ret = pass0_translate_instructions(nvs, 0, 0, nvs->program_tree);
+	if (ret)
+		pass0_resolve_mesa_consts(nvs);
+	/*XXX: if (!ret) DESTROY TREE!!! */
+
+	FREE(rec);
+	return ret;
+}
+
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_0_arb.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_0_arb.c
deleted file mode 100644
index afb889d4213..00000000000
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_0_arb.c
+++ /dev/null
@@ -1,710 +0,0 @@
-/*
- * Copyright (C) 2006 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/*
- * Authors:
- *   Ben Skeggs <[email protected]>
- */
-
-#include "glheader.h"
-#include "macros.h"
-#include "enums.h"
-
-#include "program.h"
-#include "programopt.h"
-#include "program_instruction.h"
-
-#include "nouveau_context.h"
-#include "nouveau_shader.h"
-
-static nvsFixedReg _tx_mesa_vp_dst_reg[VERT_RESULT_MAX] = {
-   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7,
-   NVS_FR_POINTSZ, NVS_FR_BFC0, NVS_FR_BFC1, NVS_FR_UNKNOWN /* EDGE */
-};
-
-static nvsFixedReg _tx_mesa_fp_dst_reg[FRAG_RESULT_MAX] = {
-   NVS_FR_FRAGDATA0 /* COLR */, NVS_FR_FRAGDATA0 /* COLH */,
-   NVS_FR_UNKNOWN /* DEPR */
-};
-
-static nvsFixedReg _tx_mesa_vp_src_reg[VERT_ATTRIB_MAX] = {
-   NVS_FR_POSITION, NVS_FR_WEIGHT, NVS_FR_NORMAL, NVS_FR_COL0, NVS_FR_COL1,
-   NVS_FR_FOGCOORD, NVS_FR_UNKNOWN /* COLOR_INDEX */, NVS_FR_UNKNOWN,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7,
-/* Generic attribs 0-15, aliased to the above */
-   NVS_FR_POSITION, NVS_FR_WEIGHT, NVS_FR_NORMAL, NVS_FR_COL0, NVS_FR_COL1,
-   NVS_FR_FOGCOORD, NVS_FR_UNKNOWN /* COLOR_INDEX */, NVS_FR_UNKNOWN,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7
-};
-
-static nvsFixedReg _tx_mesa_fp_src_reg[FRAG_ATTRIB_MAX] = {
-   NVS_FR_POSITION, NVS_FR_COL0, NVS_FR_COL1, NVS_FR_FOGCOORD,
-   NVS_FR_TEXCOORD0, NVS_FR_TEXCOORD1, NVS_FR_TEXCOORD2, NVS_FR_TEXCOORD3,
-   NVS_FR_TEXCOORD4, NVS_FR_TEXCOORD5, NVS_FR_TEXCOORD6, NVS_FR_TEXCOORD7
-};
-
-static nvsSwzComp _tx_mesa_swizzle[4] = {
-   NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W
-};
-
-static nvsOpcode _tx_mesa_opcode[] = {
-   [OPCODE_ABS] = NVS_OP_ABS, [OPCODE_ADD] = NVS_OP_ADD,
-   [OPCODE_ARA] = NVS_OP_ARA, [OPCODE_ARL] = NVS_OP_ARL,
-   [OPCODE_ARL_NV] = NVS_OP_ARL, [OPCODE_ARR] = NVS_OP_ARR,
-   [OPCODE_CMP] = NVS_OP_CMP, [OPCODE_COS] = NVS_OP_COS,
-   [OPCODE_DDX] = NVS_OP_DDX, [OPCODE_DDY] = NVS_OP_DDY,
-   [OPCODE_DP3] = NVS_OP_DP3, [OPCODE_DP4] = NVS_OP_DP4,
-   [OPCODE_DPH] = NVS_OP_DPH, [OPCODE_DST] = NVS_OP_DST,
-   [OPCODE_EX2] = NVS_OP_EX2, [OPCODE_EXP] = NVS_OP_EXP,
-   [OPCODE_FLR] = NVS_OP_FLR, [OPCODE_FRC] = NVS_OP_FRC,
-   [OPCODE_KIL] = NVS_OP_EMUL, [OPCODE_KIL_NV] = NVS_OP_KIL,
-   [OPCODE_LG2] = NVS_OP_LG2, [OPCODE_LIT] = NVS_OP_LIT,
-   [OPCODE_LOG] = NVS_OP_LOG,
-   [OPCODE_LRP] = NVS_OP_LRP,
-   [OPCODE_MAD] = NVS_OP_MAD, [OPCODE_MAX] = NVS_OP_MAX,
-   [OPCODE_MIN] = NVS_OP_MIN, [OPCODE_MOV] = NVS_OP_MOV,
-   [OPCODE_MUL] = NVS_OP_MUL,
-   [OPCODE_PK2H] = NVS_OP_PK2H, [OPCODE_PK2US] = NVS_OP_PK2US,
-   [OPCODE_PK4B] = NVS_OP_PK4B, [OPCODE_PK4UB] = NVS_OP_PK4UB,
-   [OPCODE_POW] = NVS_OP_POW, [OPCODE_POPA] = NVS_OP_POPA,
-   [OPCODE_PUSHA] = NVS_OP_PUSHA,
-   [OPCODE_RCC] = NVS_OP_RCC, [OPCODE_RCP] = NVS_OP_RCP,
-   [OPCODE_RFL] = NVS_OP_RFL, [OPCODE_RSQ] = NVS_OP_RSQ,
-   [OPCODE_SCS] = NVS_OP_SCS, [OPCODE_SEQ] = NVS_OP_SEQ,
-   [OPCODE_SFL] = NVS_OP_SFL, [OPCODE_SGE] = NVS_OP_SGE,
-   [OPCODE_SGT] = NVS_OP_SGT, [OPCODE_SIN] = NVS_OP_SIN,
-   [OPCODE_SLE] = NVS_OP_SLE, [OPCODE_SLT] = NVS_OP_SLT,
-   [OPCODE_SNE] = NVS_OP_SNE, [OPCODE_SSG] = NVS_OP_SSG,
-   [OPCODE_STR] = NVS_OP_STR, [OPCODE_SUB] = NVS_OP_SUB,
-   [OPCODE_SWZ] = NVS_OP_MOV,
-   [OPCODE_TEX] = NVS_OP_TEX, [OPCODE_TXB] = NVS_OP_TXB,
-   [OPCODE_TXD] = NVS_OP_TXD,
-   [OPCODE_TXL] = NVS_OP_TXL, [OPCODE_TXP] = NVS_OP_TXP,
-   [OPCODE_TXP_NV] = NVS_OP_TXP,
-   [OPCODE_UP2H] = NVS_OP_UP2H, [OPCODE_UP2US] = NVS_OP_UP2US,
-   [OPCODE_UP4B] = NVS_OP_UP4B, [OPCODE_UP4UB] = NVS_OP_UP4UB,
-   [OPCODE_X2D] = NVS_OP_X2D,
-   [OPCODE_XPD] = NVS_OP_XPD
-};
-
-static nvsCond _tx_mesa_condmask[] = {
-   NVS_COND_UNKNOWN, NVS_COND_GT, NVS_COND_LT, NVS_COND_UN, NVS_COND_GE,
-   NVS_COND_LE, NVS_COND_NE, NVS_COND_NE, NVS_COND_TR, NVS_COND_FL
-};
-
-struct pass0_rec {
-   int nvs_ipos;
-   int next_temp;
-   int swzconst_done;
-   int swzconst_id;
-   nvsRegister const_half;
-};
-
-#define X NVS_SWZ_X
-#define Y NVS_SWZ_Y
-#define Z NVS_SWZ_Z
-#define W NVS_SWZ_W
-
-static void
-pass0_append_fragment(nouveauShader *nvs, nvsFragmentHeader *fragment)
-{
-   nvsFragmentList *list = calloc(1, sizeof(nvsFragmentList));
-   if (!list)
-      return;
-
-   list->fragment = fragment;
-   list->prev     = nvs->list_tail;
-   if ( nvs->list_tail)
-      nvs->list_tail->next = list;
-   if (!nvs->list_head)
-      nvs->list_head = list;
-   nvs->list_tail = list;
-
-   nvs->inst_count++;
-}
-
-static void
-pass0_make_reg(nouveauShader *nvs, nvsRegister *reg,
-      	       nvsRegFile file, unsigned int index)
-{
-   struct pass0_rec *rec = nvs->pass_rec;
-
-   /* defaults */
-   *reg = nvr_unused;
-   /* -1 == quick-and-dirty temp alloc */
-   if (file == NVS_FILE_TEMP && index == -1) {
-      index = rec->next_temp++;
-      assert(index < NVS_MAX_TEMPS);
-   }
-   reg->file   = file;
-   reg->index  = index;
-}
-
-static void
-pass0_make_swizzle(nvsSwzComp *swz, unsigned int mesa)
-{
-   int i;
-
-   for (i=0;i<4;i++)
-      swz[i] = _tx_mesa_swizzle[GET_SWZ(mesa, i)];
-}
-
-static nvsOpcode
-pass0_make_opcode(enum prog_opcode op)
-{
-   if (op > MAX_OPCODE)
-      return NVS_OP_UNKNOWN;
-   return _tx_mesa_opcode[op];
-}
-
-static nvsCond
-pass0_make_condmask(GLuint mesa)
-{
-   if (mesa > COND_FL)
-      return NVS_COND_UNKNOWN;
-   return _tx_mesa_condmask[mesa];
-}
-
-static unsigned int
-pass0_make_mask(GLuint mesa_mask)
-{
-   unsigned int mask = 0;
-
-   if (mesa_mask & WRITEMASK_X) mask |= SMASK_X;
-   if (mesa_mask & WRITEMASK_Y) mask |= SMASK_Y;
-   if (mesa_mask & WRITEMASK_Z) mask |= SMASK_Z;
-   if (mesa_mask & WRITEMASK_W) mask |= SMASK_W;
-
-   return mask;
-}
-
-static nvsTexTarget
-pass0_make_tex_target(GLuint mesa)
-{
-   switch (mesa) {
-   case TEXTURE_1D_INDEX:   return NVS_TEX_TARGET_1D;
-   case TEXTURE_2D_INDEX:   return NVS_TEX_TARGET_2D;
-   case TEXTURE_3D_INDEX:   return NVS_TEX_TARGET_3D;
-   case TEXTURE_CUBE_INDEX: return NVS_TEX_TARGET_CUBE;
-   case TEXTURE_RECT_INDEX: return NVS_TEX_TARGET_RECT;
-   default:
-      return NVS_TEX_TARGET_UNKNOWN;
-   }
-}
-
-static void
-pass0_make_dst_reg(nvsPtr nvs, nvsRegister *reg,
-      		   struct prog_dst_register *dst)
-{
-   struct gl_program *mesa = (struct gl_program*)&nvs->mesa.vp;
-   nvsFixedReg sfr;
-
-   switch (dst->File) {
-   case PROGRAM_OUTPUT:
-      if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
-	 sfr = (dst->Index < VERT_RESULT_MAX) ?
-	    _tx_mesa_vp_dst_reg[dst->Index] : NVS_FR_UNKNOWN;
-      } else {
-	 sfr = (dst->Index < FRAG_RESULT_MAX) ?
-	    _tx_mesa_fp_dst_reg[dst->Index] : NVS_FR_UNKNOWN;
-      }
-      pass0_make_reg(nvs, reg, NVS_FILE_RESULT, sfr);
-      break;
-   case PROGRAM_TEMPORARY:
-      pass0_make_reg(nvs, reg, NVS_FILE_TEMP, dst->Index);
-      break;
-   case PROGRAM_ADDRESS:
-      pass0_make_reg(nvs, reg, NVS_FILE_ADDRESS, dst->Index);
-      break;
-   default:
-      fprintf(stderr, "Unknown dest file %d\n", dst->File);
-      assert(0);
-   }
-}
-
-static void
-pass0_make_src_reg(nvsPtr nvs, nvsRegister *reg, struct prog_src_register *src)
-{
-   struct gl_program *mesa = (struct gl_program *)&nvs->mesa.vp.Base;
-   struct gl_program_parameter_list *p = mesa->Parameters;
-
-   *reg = nvr_unused;
-
-   switch (src->File) {
-   case PROGRAM_INPUT:
-      reg->file = NVS_FILE_ATTRIB;
-      if (mesa->Target == GL_VERTEX_PROGRAM_ARB) {
-	 reg->index = (src->Index < VERT_ATTRIB_MAX) ?
-	    _tx_mesa_vp_src_reg[src->Index] : NVS_FR_UNKNOWN;
-      } else {
-	 reg->index = (src->Index < FRAG_ATTRIB_MAX) ?
-	    _tx_mesa_fp_src_reg[src->Index] : NVS_FR_UNKNOWN;
-      }
-      break;
-   /* All const types seem to get shoved into here, not really sure why */
-   case PROGRAM_STATE_VAR:
-      switch (p->Parameters[src->Index].Type) {
-      case PROGRAM_NAMED_PARAM:
-      case PROGRAM_CONSTANT:
-	 nvs->params[src->Index].source_val = NULL;
-	 COPY_4V(nvs->params[src->Index].val, p->ParameterValues[src->Index]); 
-	 break;
-      case PROGRAM_STATE_VAR:
-	 nvs->params[src->Index].source_val = p->ParameterValues[src->Index];
-	 break;
-      default:
-	 fprintf(stderr, "Unknown parameter type %d\n",
-	       p->Parameters[src->Index].Type);
-	 assert(0);
-	 break;
-      }
-
-      if (src->RelAddr) {
-	 reg->indexed	= 1;
-	 reg->addr_reg	= 0;
-	 reg->addr_comp	= NVS_SWZ_X;
-      } else
-	 reg->indexed = 0;
-      reg->file  = NVS_FILE_CONST;
-      reg->index = src->Index;
-      break;
-   case PROGRAM_TEMPORARY:
-      reg->file  = NVS_FILE_TEMP;
-      reg->index = src->Index;
-      break;
-   default:
-      fprintf(stderr, "Unknown source type %d\n", src->File);
-      assert(0);
-   }
-
-   /* per-component negate handled elsewhere */
-   reg->negate = src->NegateBase != 0;
-   reg->abs    = src->Abs;
-   pass0_make_swizzle(reg->swizzle, src->Swizzle);
-}
-
-static nvsInstruction *
-pass0_emit(nouveauShader *nvs, nvsOpcode op, nvsRegister dst,
-      	   unsigned int mask, int saturate,
-	   nvsRegister src0, nvsRegister src1, nvsRegister src2)
-{
-   struct pass0_rec *rec = nvs->pass_rec;
-   nvsInstruction *sif = NULL;
-
-   /* Seems mesa doesn't explicitly 0 this.. */
-   if (nvs->mesa.vp.Base.Target == GL_VERTEX_PROGRAM_ARB)
-      saturate = 0;
-
-   sif = calloc(1, sizeof(nvsInstruction));
-   if (sif) {
-      sif->header.type     = NVS_INSTRUCTION;
-      sif->header.position = rec->nvs_ipos++;
-      sif->op		= op;
-      sif->saturate	= saturate;
-      sif->dest		= dst;
-      sif->mask		= mask;
-      sif->src[0]	= src0;
-      sif->src[1]	= src1;
-      sif->src[2]	= src2;
-      sif->cond		= COND_TR;
-      sif->cond_reg	= 0;
-      sif->cond_test	= 0;
-      sif->cond_update	= 0;
-      pass0_make_swizzle(sif->cond_swizzle, SWIZZLE_NOOP);
-      pass0_append_fragment(nvs, (nvsFragmentHeader *)sif);
-   }
-
-   return sif;
-}
-
-static void
-pass0_fixup_swizzle(nvsPtr nvs,
-      		    struct prog_src_register *src,
-		    unsigned int sm1,
-		    unsigned int sm2)
-{
-   static const float sc[4] = { 1.0, 0.0, -1.0, 0.0 };
-   struct pass0_rec *rec = nvs->pass_rec;
-   int fixup_1, fixup_2;
-   nvsRegister sr, dr = nvr_unused;
-   nvsRegister sm1const, sm2const;
-
-   if (!rec->swzconst_done) {
-      struct gl_program *prog = &nvs->mesa.vp.Base;
-      rec->swzconst_id = _mesa_add_unnamed_constant(prog->Parameters, sc, 4);
-      rec->swzconst_done = 1;
-      COPY_4V(nvs->params[rec->swzconst_id].val, sc);
-   }
-
-   fixup_1 = (sm1 != MAKE_SWIZZLE4(0,0,0,0) && sm2 != MAKE_SWIZZLE4(2,2,2,2));
-   fixup_2 = (sm2 != MAKE_SWIZZLE4(2,2,2,2));
-
-   if (src->File != PROGRAM_TEMPORARY && src->File != PROGRAM_INPUT) {
-      /* We can't use more than one const in an instruction, so move the const
-       * into a temp, and swizzle from there.
-       *TODO: should just emit the swizzled const, instead of swizzling it
-       *      in the shader.. would need to reswizzle any state params when they
-       *      change however..
-       */
-      pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-      pass0_make_src_reg(nvs, &sr, src);
-      pass0_emit(nvs, NVS_OP_MOV, dr, SMASK_ALL, 0, sr, nvr_unused, nvr_unused);
-      pass0_make_reg(nvs, &sr, NVS_FILE_TEMP, dr.index);
-   } else {
-      if (fixup_1)
-	 src->NegateBase = 0;
-      pass0_make_src_reg(nvs, &sr, src);
-      pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-   }
-
-   pass0_make_reg(nvs, &sm1const, NVS_FILE_CONST, rec->swzconst_id);
-   pass0_make_swizzle(sm1const.swizzle, sm1);
-   if (fixup_1 && fixup_2) {
-      /* Any combination with SWIZZLE_ONE */
-      pass0_make_reg(nvs, &sm2const, NVS_FILE_CONST, rec->swzconst_id);
-      pass0_make_swizzle(sm2const.swizzle, sm2);
-      pass0_emit(nvs, NVS_OP_MAD, dr, SMASK_ALL, 0, sr, sm1const, sm2const);
-   } else {
-      /* SWIZZLE_ZERO || arbitrary negate */
-      pass0_emit(nvs, NVS_OP_MUL, dr, SMASK_ALL, 0, sr, sm1const, nvr_unused);
-   }
-
-   src->File	= PROGRAM_TEMPORARY;
-   src->Index	= dr.index;
-   src->Swizzle	= SWIZZLE_NOOP;
-}
-
-#define SET_SWZ(fs, cp, c) fs = (fs & ~(0x7<<(cp*3))) | (c<<(cp*3))
-static void
-pass0_check_sources(nvsPtr nvs, struct prog_instruction *inst)
-{
-   unsigned int insrc = -1, constsrc = -1;
-   int i;
-
-   for (i=0;i<_mesa_num_inst_src_regs(inst->Opcode);i++) {
-      struct prog_src_register *src = &inst->SrcReg[i];
-      unsigned int sm_1 = 0, sm_2 = 0;
-      nvsRegister sr, dr;
-      int do_mov = 0, c;
-
-      /* Build up swizzle masks as if we were going to use
-       * "MAD new, src, const1, const2" to support arbitrary negation
-       * and SWIZZLE_ZERO/SWIZZLE_ONE.
-       */
-      for (c=0;c<4;c++) {
-	 if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ZERO) {
-	    SET_SWZ(sm_1, c, SWIZZLE_Y); /* 0.0 */
-	    SET_SWZ(sm_2, c, SWIZZLE_Y);
-	    SET_SWZ(src->Swizzle, c, SWIZZLE_X);
-	 } else if (GET_SWZ(src->Swizzle, c) == SWIZZLE_ONE) {
-	    SET_SWZ(sm_1, c, SWIZZLE_Y);
-	    if (src->NegateBase & (1<<c))
-	       SET_SWZ(sm_2, c, SWIZZLE_Z); /* -1.0 */
-	    else
-	       SET_SWZ(sm_2, c, SWIZZLE_X); /* 1.0 */
-	    SET_SWZ(src->Swizzle, c, SWIZZLE_X);
-	 } else {
-	    if (src->NegateBase & (1<<c))
-	       SET_SWZ(sm_1, c, SWIZZLE_Z); /* -[xyzw] */
-	    else
-	       SET_SWZ(sm_1, c, SWIZZLE_X); /* [xyzw] */
-	    SET_SWZ(sm_2, c, SWIZZLE_Y);
-	 }
-      }
-      /* Unless we're multiplying by 1.0 or -1.0 on all components, and we're
-       * adding nothing to any component we have to emulate the swizzle.
-       */
-      if ((sm_1 != MAKE_SWIZZLE4(0,0,0,0) && sm_1 != MAKE_SWIZZLE4(2,2,2,2)) ||
-	       sm_2 != MAKE_SWIZZLE4(1,1,1,1)) {
-	 pass0_fixup_swizzle(nvs, src, sm_1, sm_2);
-	 /* The source is definitely in a temp now, so don't bother checking 
-	  * for multiple ATTRIB/CONST regs.
-	  */
-	 continue;
-      }
-
-      /* HW can't use more than one ATTRIB or PARAM in a single instruction */
-      switch (src->File) {
-      case PROGRAM_INPUT:
-	 if (insrc != -1 && insrc != src->Index) 
-	    do_mov = 1;
-	 else insrc = src->Index;
-	 break;
-      case PROGRAM_STATE_VAR:
-	 if (constsrc != -1 && constsrc != src->Index)
-	    do_mov = 1;
-	 else constsrc = src->Index;
-	 break;
-      default:
-	 break;
-      }
-
-      /* Emit any extra ATTRIB/CONST to a temp, and modify the Mesa instruction
-       * to point at the temp.
-       */
-      if (do_mov) {
-	 pass0_make_src_reg(nvs, &sr, src);
-	 pass0_make_reg(nvs, &dr, NVS_FILE_TEMP, -1);
-	 pass0_emit(nvs, NVS_OP_MOV, dr, SMASK_ALL, 0,
-	       sr, nvr_unused, nvr_unused);
-
-	 src->File	= PROGRAM_TEMPORARY;
-	 src->Index	= dr.index;
-	 src->Swizzle= SWIZZLE_NOOP;
-      }
-   }
-}
-
-static GLboolean
-pass0_emulate_instruction(nouveauShader *nvs, struct prog_instruction *inst)
-{
-   nvsFunc *shader = nvs->func;
-   nvsRegister src[3], dest, temp;
-   nvsInstruction *nvsinst;
-   struct pass0_rec *rec = nvs->pass_rec;
-   unsigned int mask = pass0_make_mask(inst->DstReg.WriteMask);
-   int i, sat;
-
-   sat = (inst->SaturateMode == SATURATE_ZERO_ONE);
-
-   /* Build all the "real" regs for the instruction */
-   for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
-      pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
-   if (inst->Opcode != OPCODE_KIL)
-      pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
-
-   switch (inst->Opcode) {
-   case OPCODE_ABS:
-      if (shader->caps & SCAP_SRC_ABS)
-	 pass0_emit(nvs, NVS_OP_MOV, dest, mask, sat,
-	       	      nvsAbs(src[0]), nvr_unused, nvr_unused);
-      else
-	 pass0_emit(nvs, NVS_OP_MAX, dest, mask, sat,
-	       	      src[0], nvsNegate(src[0]), nvr_unused);
-      break;
-   case OPCODE_KIL:
-      /* This is only in ARB shaders, so we don't have to worry
-       * about clobbering a CC reg as they aren't supported anyway.
-       */
-      /* MOVC0 temp, src */
-      pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-      nvsinst = pass0_emit(nvs, NVS_OP_MOV, temp, SMASK_ALL, 0,
-	    		   src[0], nvr_unused, nvr_unused);
-      nvsinst->cond_update = 1;
-      nvsinst->cond_reg    = 0;
-      /* KIL_NV (LT0.xyzw) temp */
-      nvsinst = pass0_emit(nvs, NVS_OP_KIL, nvr_unused, 0, 0,
-	    		   nvr_unused, nvr_unused, nvr_unused);
-      nvsinst->cond      = COND_LT;
-      nvsinst->cond_reg  = 0;
-      nvsinst->cond_test = 1;
-      pass0_make_swizzle(nvsinst->cond_swizzle, MAKE_SWIZZLE4(0,1,2,3));
-      break;
-   case OPCODE_LIT:
-      break;
-   case OPCODE_LRP:
-      pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-      pass0_emit(nvs, NVS_OP_MAD, temp, mask, 0,
-	    	 nvsNegate(src[0]), src[2], src[2]);
-      pass0_emit(nvs, NVS_OP_MAD, dest, mask, sat,
-	    	 src[0], src[1], temp);
-      break;
-   case OPCODE_POW:
-      if (shader->SupportsOpcode(shader, NVS_OP_LG2) &&
-	    shader->SupportsOpcode(shader, NVS_OP_EX2)) {
-	 pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-	 /* LG2 temp.x, src0.c */
-	 pass0_emit(nvs, NVS_OP_LG2, temp, SMASK_X, 0,
-	       	      nvsSwizzle(src[0], X, X, X, X),
-		      nvr_unused,
-		      nvr_unused);
-	 /* MUL temp.x, temp.x, src1.c */
-	 pass0_emit(nvs, NVS_OP_MUL, temp, SMASK_X, 0,
-	       	      nvsSwizzle(temp, X, X, X, X),
-		      nvsSwizzle(src[1], X, X, X, X),
-		      nvr_unused);
-	 /* EX2 dest, temp.x */
-	 pass0_emit(nvs, NVS_OP_EX2, dest, mask, sat,
-	       	      nvsSwizzle(temp, X, X, X, X),
-		      nvr_unused,
-		      nvr_unused);
-      } else {
-	 /* can we use EXP/LOG instead of EX2/LG2?? */
-	 fprintf(stderr, "Implement POW for NV20 vtxprog!\n");
-	 return GL_FALSE;
-      }
-      break;
-   case OPCODE_RSQ:
-      if (rec->const_half.file != NVS_FILE_CONST) {
-	 GLfloat const_half[4] = { 0.5, 0.0, 0.0, 0.0 };
-	 pass0_make_reg(nvs, &rec->const_half, NVS_FILE_CONST,
-	       _mesa_add_unnamed_constant(nvs->mesa.vp.Base.Parameters,
-		  			  const_half, 4));
-	 COPY_4V(nvs->params[rec->const_half.index].val, const_half);
-      }
-      pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-      pass0_emit(nvs, NVS_OP_LG2, temp, SMASK_X, 0,
-	    	 nvsAbs(nvsSwizzle(src[0], X, X, X, X)),
-		 nvr_unused,
-		 nvr_unused);
-      pass0_emit(nvs, NVS_OP_MUL, temp, SMASK_X, 0,
-	    	 nvsSwizzle(temp, X, X, X, X),
-		 nvsNegate(rec->const_half),
-		 nvr_unused);
-      pass0_emit(nvs, NVS_OP_EX2, dest, mask, sat,
-	    	 nvsSwizzle(temp, X, X, X, X),
-		 nvr_unused,
-		 nvr_unused);
-      break;
-   case OPCODE_SCS:
-      if (mask & SMASK_X)
-	 pass0_emit(nvs, NVS_OP_COS, dest, SMASK_X, sat,
-	       	      nvsSwizzle(src[0], X, X, X, X),
-		      nvr_unused,
-		      nvr_unused);
-      if (mask & SMASK_Y)
-	 pass0_emit(nvs, NVS_OP_SIN, dest, SMASK_Y, sat,
-	       	      nvsSwizzle(src[0], X, X, X, X),
-		      nvr_unused,
-		      nvr_unused);
-      break;
-   case OPCODE_SUB:
-      pass0_emit(nvs, NVS_OP_ADD, dest, mask, sat,
-	    	 src[0], nvsNegate(src[1]), nvr_unused);
-      break;
-   case OPCODE_XPD:
-      pass0_make_reg(nvs, &temp, NVS_FILE_TEMP, -1);
-      pass0_emit(nvs, NVS_OP_MUL, temp, SMASK_ALL, 0,
-	    	   nvsSwizzle(src[0], Z, X, Y, Y),
-		   nvsSwizzle(src[1], Y, Z, X, X),
-		   nvr_unused);
-      pass0_emit(nvs, NVS_OP_MAD, dest, (mask & ~SMASK_W), sat,
-	    	   nvsSwizzle(src[0], Y, Z, X, X),
-		   nvsSwizzle(src[1], Z, X, Y, Y),
-		   nvsNegate(temp));
-      break;
-   default:
-      fprintf(stderr, "hw doesn't support opcode \"%s\", and no emulation found\n",
-	      _mesa_opcode_string(inst->Opcode));
-      return GL_FALSE;
-   }
-
-   return GL_TRUE;
-}
-
-static GLboolean
-pass0_translate_instructions(nouveauShader *nvs)
-{
-   struct gl_program *prog = (struct gl_program *)&nvs->mesa.vp;
-   nvsFunc *shader = nvs->func;
-   int ipos;
-
-   for (ipos=0; ipos<prog->NumInstructions; ipos++) {
-      struct prog_instruction *inst = &prog->Instructions[ipos];
-
-      if (inst->Opcode == OPCODE_END)
-	 break;
-
-      /* Deal with multiple ATTRIB/PARAM in a single instruction */
-      pass0_check_sources(nvs, inst);
- 
-      /* Now it's safe to do the prog_instruction->nvsInstruction conversion */
-      if (shader->SupportsOpcode(shader, pass0_make_opcode(inst->Opcode))) {
-	 nvsInstruction *nvsinst;
-	 nvsRegister src[3], dest;
-	 int i;
-
-	 for (i=0; i<_mesa_num_inst_src_regs(inst->Opcode); i++)
-	    pass0_make_src_reg(nvs, &src[i], &inst->SrcReg[i]);
-	 pass0_make_dst_reg(nvs, &dest, &inst->DstReg);
-
-	 nvsinst = pass0_emit(nvs,
-	       		      pass0_make_opcode(inst->Opcode),
-			      dest,
-			      pass0_make_mask(inst->DstReg.WriteMask),
-			      (inst->SaturateMode != SATURATE_OFF),
-			      src[0], src[1], src[2]);
-	 nvsinst->tex_unit   = inst->TexSrcUnit;
-	 nvsinst->tex_target = pass0_make_tex_target(inst->TexSrcTarget);
-	 /* TODO when NV_fp/vp is implemented */
-	 nvsinst->cond       = COND_TR;
-      } else {
-	 if (!pass0_emulate_instruction(nvs, inst))
-	    return GL_FALSE;
-      }
-   }
-
-   return GL_TRUE;
-}
-
-GLboolean
-nouveau_shader_pass0_arb(GLcontext *ctx, nouveauShader *nvs)
-{
-   nouveauContextPtr		 nmesa	= NOUVEAU_CONTEXT(ctx);
-   struct gl_program		*prog	= (struct gl_program*)nvs;
-   struct gl_vertex_program	*vp 	= (struct gl_vertex_program *)prog;
-   struct gl_fragment_program	*fp	= (struct gl_fragment_program *)prog;
-   struct pass0_rec *rec;
-   int ret;
-
-   switch (prog->Target) {
-   case GL_VERTEX_PROGRAM_ARB:
-      nvs->func = &nmesa->VPfunc;
-      if (vp->IsPositionInvariant)
-	 _mesa_insert_mvp_code(ctx, vp);
-#if 0
-      if (IS_FIXEDFUNCTION_PROG && CLIP_PLANES_USED)
-	 pass0_insert_ff_clip_planes();
-#endif
-      break;
-   case GL_FRAGMENT_PROGRAM_ARB:
-      nvs->func = &nmesa->FPfunc;
-      if (fp->FogOption != GL_NONE)
-	 _mesa_append_fog_code(ctx, fp);
-      break;
-   default:
-      fprintf(stderr, "Unknown program type %d", prog->Target);
-      return GL_FALSE;
-   }
-
-   rec = calloc(1, sizeof(struct pass0_rec));
-   rec->next_temp = prog->NumTemporaries;
-   nvs->pass_rec = rec;
-
-   ret = pass0_translate_instructions(nvs);
-   if (!ret) {
-      /* DESTROY list */
-   }
-
-   free(nvs->pass_rec);
-   return ret;
-}
-
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c
index 5de9017f580..90c57d38078 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader_1.c
@@ -1,316 +1,12 @@
-/*
- * Copyright (C) 2006 Ben Skeggs.
- *
- * All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining
- * a copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sublicense, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- *
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial
- * portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
- * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
- * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
- * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
- * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- *
- */
-
-/*
- * Authors:
- *   Ben Skeggs <[email protected]>
- */
-
 #include "glheader.h"
 #include "macros.h"
 #include "enums.h"
 
 #include "nouveau_shader.h"
 
-#define PASS1_OK	0
-#define PASS1_KILL	1
-#define PASS1_FAIL	2
-
-struct pass1_rec {
-   unsigned int temp[NVS_MAX_TEMPS];
-   unsigned int result[NVS_MAX_ATTRIBS];
-   unsigned int address[NVS_MAX_ADDRESS];
-   unsigned int cc[2];
-};
-
-static void
-pass1_remove_fragment(nvsPtr nvs, nvsFragmentList *item)
-{
-   if (item->prev) item->prev->next = item->next;
-   if (item->next) item->next->prev = item->prev;
-   if (nvs->list_head == item) nvs->list_head = item->next;
-   if (nvs->list_tail == item) nvs->list_tail = item->prev;
-
-   nvs->inst_count--;
-}
-
-static int
-pass1_result_needed(struct pass1_rec *rec, nvsInstruction *inst)
-{
-   if (inst->cond_update && rec->cc[inst->cond_reg])
-      return 1;
-   /* Only write components that are read later */
-   if (inst->dest.file == NVS_FILE_TEMP)
-      return (inst->mask & rec->temp[inst->dest.index]);
-   if (inst->dest.file == NVS_FILE_ADDRESS)
-      return (inst->mask & rec->address[inst->dest.index]);
-   /* No point writing result components that are written later */
-   if (inst->dest.file == NVS_FILE_RESULT)
-      return (inst->mask & ~rec->result[inst->dest.index]);
-   assert(0);
-}
-
-static void
-pass1_track_result(struct pass1_rec *rec, nvsInstruction *inst)
-{
-   if (inst->cond_test)
-      rec->cc[inst->cond_reg] = 1;
-   if (inst->dest.file == NVS_FILE_TEMP) {
-      inst->mask &= rec->temp[inst->dest.index];
-   } else if (inst->dest.file == NVS_FILE_RESULT) {
-      inst->mask &= ~rec->result[inst->dest.index];
-      rec->result[inst->dest.index] |= inst->mask;
-   } else if (inst->dest.file == NVS_FILE_ADDRESS) {
-      inst->mask &= rec->address[inst->dest.index];
-   }
-}
-
-static void
-pass1_track_source(nouveauShader *nvs, nvsInstruction *inst, int pos,
-      		   unsigned int read)
-{
-   struct pass1_rec *rec = nvs->pass_rec;
-   nvsRegister *src = &inst->src[pos];
-   unsigned int really_read = 0;
-   int i,sc;
-
-   /* Account for swizzling */
-   for (i=0; i<4; i++)
-      if (read & (1<<i)) really_read |= (1 << src->swizzle[i]);
-
-   /* Track register reads */
-   if (src->file == NVS_FILE_TEMP) {
-      if (nvs->temps[src->index].last_use == -1)
-	 nvs->temps[src->index].last_use = inst->header.position;
-      rec->temp [src->index] |= really_read;
-   } else if (src->indexed) {
-      rec->address[src->addr_reg] |= (1<<src->addr_comp);
-   }
-
-   /* Modify swizzle to only access read components */
-   /* Find a component that is used.. */
-   for (sc=0;sc<4;sc++)
-      if (really_read & (1<<sc))
-	 break;
-   /* Now set all unused components to that value */
-   for (i=0;i<4;i++)
-      if (!(read & (1<<i))) src->swizzle[i] = sc;
-}
-
-static int
-pass1_check_instruction(nouveauShader *nvs, nvsInstruction *inst)
-{
-   struct pass1_rec *rec = nvs->pass_rec;
-   unsigned int read0, read1, read2;
-
-   if (inst->op != NVS_OP_KIL) {
-      if (!pass1_result_needed(rec, inst))
-	 return PASS1_KILL;
-   }
-   pass1_track_result(rec, inst);
-
-   read0 = read1 = read2 = 0;
-
-   switch (inst->op) {
-   case NVS_OP_FLR:
-   case NVS_OP_FRC:
-   case NVS_OP_MOV:
-   case NVS_OP_SSG:
-   case NVS_OP_ARL:
-      read0 = inst->mask;
-      break;
-   case NVS_OP_ADD:
-   case NVS_OP_MAX:
-   case NVS_OP_MIN:
-   case NVS_OP_MUL:
-   case NVS_OP_SEQ:
-   case NVS_OP_SFL:
-   case NVS_OP_SGE:
-   case NVS_OP_SGT:
-   case NVS_OP_SLE:
-   case NVS_OP_SLT:
-   case NVS_OP_SNE:
-   case NVS_OP_STR:
-   case NVS_OP_SUB:
-      read0 = inst->mask;
-      read1 = inst->mask;
-      break;
-   case NVS_OP_CMP:
-   case NVS_OP_LRP:
-   case NVS_OP_MAD:
-      read0 = inst->mask;
-      read1 = inst->mask;
-      read2 = inst->mask;
-      break;
-   case NVS_OP_XPD:
-      if (inst->mask & SMASK_X) read0 |= SMASK_Y|SMASK_Z;
-      if (inst->mask & SMASK_Y) read0 |= SMASK_X|SMASK_Z;
-      if (inst->mask & SMASK_Z) read0 |= SMASK_X|SMASK_Y;
-      read1 = read0;
-      break;
-   case NVS_OP_COS:
-   case NVS_OP_EX2:
-   case NVS_OP_EXP:
-   case NVS_OP_LG2:
-   case NVS_OP_LOG:
-   case NVS_OP_RCC:
-   case NVS_OP_RCP:
-   case NVS_OP_RSQ:
-   case NVS_OP_SCS:
-   case NVS_OP_SIN:
-      read0 = SMASK_X;
-      break;
-   case NVS_OP_POW:
-      read0 = SMASK_X;
-      read1 = SMASK_X;
-      break;
-   case NVS_OP_DIV:
-      read0 = inst->mask;
-      read1 = SMASK_X;
-      break;
-   case NVS_OP_DP2:
-      read0 = SMASK_X|SMASK_Y;
-      read1 = SMASK_X|SMASK_Y;
-      break;
-   case NVS_OP_DP3:
-   case NVS_OP_RFL:
-      read0 = SMASK_X|SMASK_Y|SMASK_Z;
-      read1 = SMASK_X|SMASK_Y|SMASK_Z;
-      break;
-   case NVS_OP_DP4:
-      read0 = SMASK_ALL;
-      read1 = SMASK_ALL;
-      break;
-   case NVS_OP_DPH:
-      read0 = SMASK_X|SMASK_Y|SMASK_Z;
-      read1 = SMASK_ALL;
-      break;
-   case NVS_OP_DST:
-      if (inst->mask & SMASK_Y) read0 = read1 = SMASK_Y;
-      if (inst->mask & SMASK_Z) read0 |= SMASK_Z;
-      if (inst->mask & SMASK_W) read1 |= SMASK_W;
-      break;
-   case NVS_OP_NRM:
-      read0 = SMASK_X|SMASK_Y|SMASK_Z;
-      break;
-   case NVS_OP_PK2H:
-   case NVS_OP_PK2US:
-      read0 = SMASK_X|SMASK_Y;
-      break;
-   case NVS_OP_DDX:
-   case NVS_OP_DDY:
-   case NVS_OP_UP2H:
-   case NVS_OP_UP2US:
-   case NVS_OP_PK4B:
-   case NVS_OP_PK4UB:
-   case NVS_OP_UP4B:
-   case NVS_OP_UP4UB:
-      read0 = SMASK_ALL;
-      break;
-   case NVS_OP_X2D:
-      read1 = SMASK_X|SMASK_Y;
-      if (inst->mask & (SMASK_X|SMASK_Z)) {
-	 read0 |= SMASK_X;
-	 read2 |= SMASK_X|SMASK_Y;
-      }
-      if (inst->mask & (SMASK_Y|SMASK_W)) {
-	 read0 |= SMASK_Y;
-	 read2 |= SMASK_Z|SMASK_W;
-      }
-      break;
-   case NVS_OP_LIT:
-      read0 |= SMASK_X|SMASK_Y|SMASK_W;
-      break;
-   case NVS_OP_TEX:
-   case NVS_OP_TXP:
-   case NVS_OP_TXL:
-   case NVS_OP_TXB:
-      read0 = SMASK_ALL;
-      break;
-   case NVS_OP_TXD:
-      read0 = SMASK_ALL;
-      read1 = SMASK_ALL;
-      read2 = SMASK_ALL;
-      break;
-   case NVS_OP_KIL:
-      break;
-   default:
-      fprintf(stderr, "Unknown sop=%d", inst->op);
-      return PASS1_FAIL;
-   }
-
-   /* Any values that are written by this inst can't have been read further up */
-   if (inst->dest.file == NVS_FILE_TEMP)
-      rec->temp[inst->dest.index] &= ~inst->mask;
-
-   if (read0) pass1_track_source(nvs, inst, 0, read0);
-   if (read1) pass1_track_source(nvs, inst, 1, read1);
-   if (read2) pass1_track_source(nvs, inst, 2, read2);
-
-   return PASS1_OK;
-}
-
-/* Some basic dead code elimination
- *   - Remove unused instructions
- *   - Don't write unused register components
- *   - Modify swizzles to not reference unneeded components.
- */
 GLboolean
 nouveau_shader_pass1(nvsPtr nvs)
 {
-   nvsFragmentList *list = nvs->list_tail;
-   int i;
-
-   for (i=0; i<NVS_MAX_TEMPS; i++)
-      nvs->temps[i].last_use = -1;
-
-   nvs->pass_rec = calloc(1, sizeof(struct pass1_rec));
-
-   while (list) {
-      assert(list->fragment->type == NVS_INSTRUCTION);
-
-      switch(pass1_check_instruction(nvs, (nvsInstruction *)list->fragment)) {
-      case PASS1_OK:
-	 break;
-      case PASS1_KILL:
-	 pass1_remove_fragment(nvs, list);
-	 break;
-      case PASS1_FAIL:
-      default:
-	 free(nvs->pass_rec);
-	 nvs->pass_rec = NULL;
-	 return GL_FALSE;
-      }
-
-      list = list->prev;
-   }
-
-   free(nvs->pass_rec);
-   nvs->pass_rec = NULL;
 
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c b/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c
index 2177413b66e..b043f877e42 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader_2.c
@@ -38,206 +38,224 @@
 
 #include "nouveau_context.h"
 #include "nouveau_shader.h"
+#include "nouveau_msg.h"
 
 struct pass2_rec {
-   /* Map nvsRegister temp ID onto hw temp ID */
-   unsigned int temps[NVS_MAX_TEMPS];
-   /* Track free hw registers */
-   unsigned int hw_temps[NVS_MAX_TEMPS];
+	/* Map nvsRegister temp ID onto hw temp ID */
+	unsigned int temps[NVS_MAX_TEMPS];
+	/* Track free hw registers */
+	unsigned int hw_temps[NVS_MAX_TEMPS];
 };
 
 static int
 pass2_alloc_hw_temp(nvsPtr nvs)
 {
-   struct pass2_rec *rec = nvs->pass_rec;
-   int i;
-
-   for (i=0; i<nvs->func->MaxTemp; i++) {
-      /* This is a *horrible* hack.. R0 is both temp0 and result.color
-       * in NV30/40 fragprogs, we can use R0 as a temp before result is
-       * written however..
-       */
-      if (nvs->mesa.vp.Base.Target == GL_FRAGMENT_PROGRAM_ARB && i==0)
-	 continue;
-
-      if (rec->hw_temps[i] == 0) {
-	 rec->hw_temps[i] = 1;
-	 return i;
-      }
-   }
-   return -1;
-}
-
-static void
-pass2_free_hw_temp(nvsPtr nvs, int reg)
-{
-   struct pass2_rec *rec = nvs->pass_rec;
-   rec->hw_temps[reg] = 0;
+	struct pass2_rec *rec = nvs->pass_rec;
+	int i;
+
+	for (i=0; i<nvs->func->MaxTemp; i++) {
+		/* This is a *horrible* hack.. R0 is both temp0 and result.color
+		 * in NV30/40 fragprogs, we can use R0 as a temp before result
+		 * is written however..
+		 */
+		if (nvs->mesa.vp.Base.Target == GL_FRAGMENT_PROGRAM_ARB && i==0)
+			continue;
+		if (rec->hw_temps[i] == 0) {
+			rec->hw_temps[i] = 1;
+			return i;
+		}
+	}
+
+	return -1;
 }
 
 static nvsRegister
 pass2_mangle_reg(nvsPtr nvs, nvsInstruction *inst, nvsRegister reg)
 {
-   struct pass2_rec *rec = nvs->pass_rec;
-
-   if (reg.file == NVS_FILE_TEMP) {
-      int hwidx;
+	struct pass2_rec *rec = nvs->pass_rec;
 
-      if (rec->temps[reg.index] == -1)
-	 rec->temps[reg.index] = pass2_alloc_hw_temp(nvs);
-      hwidx = rec->temps[reg.index];
+	if (reg.file == NVS_FILE_TEMP) {
+		if (rec->temps[reg.index] == -1)
+			rec->temps[reg.index] = pass2_alloc_hw_temp(nvs);
+		reg.index = rec->temps[reg.index];
+	}
 
-      if (nvs->temps[reg.index].last_use <= inst->header.position)
-	 pass2_free_hw_temp(nvs, hwidx);
-
-      reg.index = hwidx;
-   }
-
-   return reg;
+	return reg;
 }
 
 static void
 pass2_add_instruction(nvsPtr nvs, nvsInstruction *inst, 
-      		      struct _op_xlat *op, int slot)
+		      struct _op_xlat *op, int slot)
 {
-   nvsSwzComp default_swz[4] = { NVS_SWZ_X, NVS_SWZ_Y, NVS_SWZ_Z, NVS_SWZ_W };
-   nvsFunc *shader = nvs->func;
-   nvsRegister reg;
-   int i;
-
-   shader->SetOpcode(shader, op->NV, slot);
-   if (inst->saturate	) shader->SetSaturate(shader);
-   if (inst->cond_update) shader->SetCCUpdate(shader);
-   if (inst->cond_test	) shader->SetCondition(shader, 1, inst->cond,
-	 				       inst->cond_reg,
-					       inst->cond_swizzle);
-   else			  shader->SetCondition(shader, 0, NVS_COND_TR,
-	 				       0,
-					       default_swz);
-   switch (inst->op) {
-   case NVS_OP_TEX:
-   case NVS_OP_TXB:
-   case NVS_OP_TXL:
-   case NVS_OP_TXP:
-   case NVS_OP_TXD:
-      shader->SetTexImageUnit(shader, inst->tex_unit);
-      break;
-   default:
-      break;
-   }
-
-   for (i = 0; i < 3; i++) {
-      if (op->srcpos[i] != -1) {
-	 reg = pass2_mangle_reg(nvs, inst, inst->src[i]);
-	 if (reg.file == NVS_FILE_ATTRIB)
-	    nvs->inputs_read |= (1 << reg.index);
-	 shader->SetSource(shader, &reg, op->srcpos[i]);
-	 if (reg.file == NVS_FILE_CONST && shader->GetSourceConstVal) {
-	    int idx_slot = nvs->params[reg.index].hw_index_cnt++;
-	    nvs->params[reg.index].hw_index = realloc(
-		  nvs->params[reg.index].hw_index, sizeof(int) * idx_slot+1);
-	    nvs->params[reg.index].hw_index[idx_slot] = nvs->program_current + 4;
-	 }
-      }
-   }
-
-   reg = pass2_mangle_reg(nvs, inst, inst->dest);
-   if (reg.file == NVS_FILE_RESULT)
-      nvs->outputs_written |= (1 << reg.index);
-   shader->SetResult(shader, &reg, inst->mask, slot);
+	nvsSwzComp default_swz[4] = { NVS_SWZ_X, NVS_SWZ_Y,
+				      NVS_SWZ_Z, NVS_SWZ_W };
+	nvsFunc *shader = nvs->func;
+	nvsRegister reg;
+	int i;
+
+	shader->SetOpcode(shader, op->NV, slot);
+	if (inst->saturate	) shader->SetSaturate(shader);
+	if (inst->cond_update	) shader->SetCCUpdate(shader);
+	if (inst->cond_test	) shader->SetCondition(shader, 1, inst->cond,
+						       inst->cond_reg,
+						       inst->cond_swizzle);
+	else			  shader->SetCondition(shader, 0, NVS_COND_TR,
+						       0,
+						       default_swz);
+	switch (inst->op) {
+	case NVS_OP_TEX:
+	case NVS_OP_TXB:
+	case NVS_OP_TXL:
+	case NVS_OP_TXP:
+	case NVS_OP_TXD:
+		shader->SetTexImageUnit(shader, inst->tex_unit);
+		break;
+	default:
+		break;
+	}
+
+	for (i = 0; i < 3; i++) {
+		if (op->srcpos[i] != -1) {
+			reg = pass2_mangle_reg(nvs, inst, inst->src[i]);
+
+			shader->SetSource(shader, &reg, op->srcpos[i]);
+
+			if (reg.file == NVS_FILE_CONST &&
+					shader->GetSourceConstVal) {
+				int idx_slot =
+					nvs->params[reg.index].hw_index_cnt++;
+				nvs->params[reg.index].hw_index = realloc(
+						nvs->params[reg.index].hw_index,
+						sizeof(int) * idx_slot+1);
+				nvs->params[reg.index].hw_index[idx_slot] =
+					nvs->program_current + 4;
+			}
+		}
+	}
+
+	reg = pass2_mangle_reg(nvs, inst, inst->dest);
+	shader->SetResult(shader, &reg, inst->mask, slot);
+
+	if (inst->dest_scale != NVS_SCALE_1X) {
+		shader->SetResultScale(shader, inst->dest_scale);
+	}
 }
 
 static int
 pass2_assemble_instruction(nvsPtr nvs, nvsInstruction *inst, int last)
 {
-   nvsFunc *shader = nvs->func;
-   struct _op_xlat *op;
-   unsigned int hw_inst[8];
-   int slot;
-   int instsz;
-   int i;
-
-   shader->inst = hw_inst;
-
-   /* Assemble this instruction */
-   if (!(op = shader->GetOPTXFromSOP(inst->op, &slot)))
-      return 0;
-   shader->InitInstruction(shader);
-   pass2_add_instruction(nvs, inst, op, slot);
-   if (last)
-      shader->SetLastInst(shader);
-
-   instsz = shader->GetOffsetNext(nvs->func);
-   if (nvs->program_size + instsz >= nvs->program_alloc_size) {
-      nvs->program_alloc_size *= 2;
-      nvs->program = realloc(nvs->program,
-	    		     nvs->program_alloc_size * sizeof(uint32_t));
-   }
-
-   for (i=0; i<instsz; i++)
-      nvs->program[nvs->program_current++] = hw_inst[i];
-   nvs->program_size = nvs->program_current;
-   return 1;
+	nvsFunc *shader = nvs->func;
+	struct _op_xlat *op;
+	unsigned int hw_inst[8];
+	int slot;
+	int instsz;
+	int i;
+
+	shader->inst = hw_inst;
+
+	/* Assemble this instruction */
+	if (!(op = shader->GetOPTXFromSOP(inst->op, &slot)))
+		return 0;
+	shader->InitInstruction(shader);
+	pass2_add_instruction(nvs, inst, op, slot);
+	if (last)
+		shader->SetLastInst(shader);
+
+	instsz = shader->GetOffsetNext(nvs->func);
+	if (nvs->program_size + instsz >= nvs->program_alloc_size) {
+		nvs->program_alloc_size *= 2;
+		nvs->program = realloc(nvs->program,
+				       nvs->program_alloc_size *
+				       sizeof(uint32_t));
+	}
+
+	for (i=0; i<instsz; i++)
+		nvs->program[nvs->program_current++] = hw_inst[i];
+	nvs->program_size = nvs->program_current;
+	return 1;
+}
+
+static GLboolean
+pass2_translate(nvsPtr nvs, nvsFragmentHeader *f)
+{
+	nvsFunc *shader = nvs->func;
+	GLboolean last;
+
+	while (f) {
+		last = (f == ((nvsSubroutine*)nvs->program_tree)->insn_tail);
+
+		switch (f->type) {
+		case NVS_INSTRUCTION:
+			if (!pass2_assemble_instruction(nvs,
+							(nvsInstruction *)f,
+							last))
+				return GL_FALSE;
+			break;
+		default:
+			WARN_ONCE("Unimplemented fragment type\n");
+			return GL_FALSE;
+		}
+
+		f = f->next;
+	}
+
+	return GL_TRUE;
 }
 
 /* Translate program into hardware format */
 GLboolean
 nouveau_shader_pass2(nvsPtr nvs)
 {
-   nvsFragmentList *list = nvs->list_head;
-   struct pass2_rec *rec;
-   int i;
-
-   rec = calloc(1, sizeof(struct pass2_rec));
-   for (i=0; i<NVS_MAX_TEMPS; i++)
-      rec->temps[i] = -1;
-   nvs->pass_rec = rec;
-
-   /* Start off with allocating 4 uint32_t's for each inst, will be grown
-    * if necessary..
-    */
-   nvs->program_alloc_size = nvs->inst_count * 4;
-   nvs->program = calloc(nvs->program_alloc_size, sizeof(uint32_t));
-   nvs->program_size    = 0;
-   nvs->program_current = 0;
-
-   while (list) {
-      assert(list->fragment->type == NVS_INSTRUCTION);
-
-      if (!pass2_assemble_instruction(nvs, (nvsInstruction *)list->fragment, list->next ? 0 : 1)) {
-	 free(nvs->program);
-	 nvs->program = NULL;
-	 return GL_FALSE;
-      }
-
-      list = list->next;
-   }
-
-   /* Shrink allocated memory to only what we need */
-   nvs->program = realloc(nvs->program, nvs->program_size * sizeof(uint32_t));
-   nvs->program_alloc_size = nvs->program_size;
-
-   nvs->translated	= 1;
-   nvs->on_hardware	= 0;
-
-   if (NOUVEAU_DEBUG & DEBUG_SHADERS) {
-      fflush(stdout); fflush(stderr);
-      fprintf(stderr, "----------------MESA PROGRAM target=%s, id=0x%x\n",
-	    _mesa_lookup_enum_by_nr(nvs->mesa.vp.Base.Target),
-	    nvs->mesa.vp.Base.Id);
-      fflush(stdout); fflush(stderr);
-      _mesa_print_program(&nvs->mesa.vp.Base);
-      fflush(stdout); fflush(stderr);
-      fprintf(stderr, "^^^^^^^^^^^^^^^^MESA PROGRAM\n");
-      fflush(stdout); fflush(stderr);
-      fprintf(stderr, "----------------NV PROGRAM\n");
-      fflush(stdout); fflush(stderr);
-      nvsDisasmHWShader(nvs);
-      fflush(stdout); fflush(stderr);
-      fprintf(stderr, "^^^^^^^^^^^^^^^^NV PROGRAM\n");
-      fflush(stdout); fflush(stderr);
-   }
-
-   return GL_TRUE;
+	struct pass2_rec *rec;
+	int i;
+
+	rec = calloc(1, sizeof(struct pass2_rec));
+	for (i=0; i<NVS_MAX_TEMPS; i++)
+		rec->temps[i] = -1;
+	nvs->pass_rec = rec;
+
+	/* Start off with allocating 4 uint32_t's for each inst, will be grown
+	 * if necessary..
+	 */
+	nvs->program_alloc_size = nvs->mesa.vp.Base.NumInstructions * 4;
+	nvs->program = calloc(nvs->program_alloc_size, sizeof(uint32_t));
+	nvs->program_size    = 0;
+	nvs->program_current = 0;
+
+	if (!pass2_translate(nvs,
+			     ((nvsSubroutine*)nvs->program_tree)->insn_head)) {
+		free(nvs->program);
+		nvs->program = NULL;
+		return GL_FALSE;
+	}
+
+	/* Shrink allocated memory to only what we need */
+	nvs->program = realloc(nvs->program,
+			       nvs->program_size * sizeof(uint32_t));
+	nvs->program_alloc_size = nvs->program_size;
+
+	nvs->translated  = 1;
+	nvs->on_hardware = 0;
+
+	if (NOUVEAU_DEBUG & DEBUG_SHADERS) {
+		fflush(stdout); fflush(stderr);
+		fprintf(stderr, "-----------MESA PROGRAM target=%s, id=0x%x\n",
+				_mesa_lookup_enum_by_nr(
+					nvs->mesa.vp.Base.Target),
+				nvs->mesa.vp.Base.Id);
+		fflush(stdout); fflush(stderr);
+		_mesa_print_program(&nvs->mesa.vp.Base);
+		fflush(stdout); fflush(stderr);
+		fprintf(stderr, "^^^^^^^^^^^^^^^^MESA PROGRAM\n");
+		fflush(stdout); fflush(stderr);
+		fprintf(stderr, "----------------NV PROGRAM\n");
+		fflush(stdout); fflush(stderr);
+		nvsDisasmHWShader(nvs);
+		fflush(stdout); fflush(stderr);
+		fprintf(stderr, "^^^^^^^^^^^^^^^^NV PROGRAM\n");
+		fflush(stdout); fflush(stderr);
+	}
+
+	return GL_TRUE;
 }
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.c b/src/mesa/drivers/dri/nouveau/nouveau_state.c
index 1ff881f0542..e9fd188d73e 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state.c
@@ -30,7 +30,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "nouveau_fifo.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -144,7 +143,7 @@ static void nouveauDDInvalidateState(GLcontext *ctx, GLuint new_state)
 {
     _swrast_InvalidateState( ctx, new_state );
     _swsetup_InvalidateState( ctx, new_state );
-    _ac_InvalidateState( ctx, new_state );
+    _vbo_InvalidateState( ctx, new_state );
     _tnl_InvalidateState( ctx, new_state );
     NOUVEAU_CONTEXT(ctx)->new_render_state |= new_state;
 }
@@ -156,9 +155,11 @@ void nouveauDDInitState(nouveauContextPtr nmesa)
     switch(type)
     {
         case NV_03:
+            /* Unimplemented */
+            break;
         case NV_04:
         case NV_05:
-            /* No TCL engines for these ones */
+            nv04InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
             break;
         case NV_10:
             nv10InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
@@ -169,9 +170,11 @@ void nouveauDDInitState(nouveauContextPtr nmesa)
         case NV_30:
         case NV_40:
         case NV_44:
-        case NV_50:
             nv30InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
             break;
+        case NV_50:
+            nv50InitStateFuncs(nmesa->glCtx, &nmesa->glCtx->Driver);
+            break;
         default:
             break;
     }
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state.h b/src/mesa/drivers/dri/nouveau/nouveau_state.h
index 16d63a6ac23..dbac71760b6 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state.h
@@ -32,9 +32,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 extern void nouveauDDInitState(nouveauContextPtr nmesa);
 extern void nouveauDDInitStateFuncs(GLcontext *ctx);
 
+extern void nv04InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
 extern void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
 extern void nv20InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
 extern void nv30InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
+extern void nv50InitStateFuncs(GLcontext *ctx, struct dd_function_table *func);
 
 extern void nouveauInitState(GLcontext *ctx);
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c
index 36f0c1024b6..cb4b9d30270 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.c
@@ -25,6 +25,8 @@ void nouveau_state_cache_flush(nouveauContextPtr nmesa)
 	do
 	{
 		// jump to a dirty state
+		while((nmesa->state_cache.hdirty[i/NOUVEAU_STATE_CACHE_HIER_SIZE]==0)&&(i<NOUVEAU_STATE_CACHE_ENTRIES))
+			i=(i&~(NOUVEAU_STATE_CACHE_HIER_SIZE-1))+NOUVEAU_STATE_CACHE_HIER_SIZE;
 		while((nmesa->state_cache.atoms[i].dirty==0)&&(i<NOUVEAU_STATE_CACHE_ENTRIES))
 			i++;
 
@@ -42,11 +44,14 @@ void nouveau_state_cache_flush(nouveauContextPtr nmesa)
 			{
 				OUT_RING(nmesa->state_cache.atoms[i+j].value);
 				nmesa->state_cache.atoms[i+j].dirty=0;
+				if ((i+j)%NOUVEAU_STATE_CACHE_HIER_SIZE==0)
+					nmesa->state_cache.hdirty[(i+j)/NOUVEAU_STATE_CACHE_HIER_SIZE-1]=0;
 			}
 			i+=run;
 		}
 	}
 	while(i<NOUVEAU_STATE_CACHE_ENTRIES);
+	nmesa->state_cache.hdirty[NOUVEAU_STATE_CACHE_HIER_SIZE/NOUVEAU_STATE_CACHE_HIER_SIZE-1]=0;
 }
 
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h
index 24882748468..5f9d426450b 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_state_cache.h
@@ -5,6 +5,10 @@
 #include "mtypes.h"
 
 #define NOUVEAU_STATE_CACHE_ENTRIES 2048
+// size of a dirty requests block
+// you can play with that and tune the value to increase/decrease performance
+// but keep it a power of 2 !
+#define NOUVEAU_STATE_CACHE_HIER_SIZE  32
 
 typedef struct nouveau_state_atom_t{
 	uint32_t value;
@@ -14,8 +18,10 @@ typedef struct nouveau_state_atom_t{
 typedef struct nouveau_state_cache_t{
 	nouveau_state_atom atoms[NOUVEAU_STATE_CACHE_ENTRIES];
 	uint32_t current_pos;
+	// hierarchical dirty flags
+	uint8_t hdirty[NOUVEAU_STATE_CACHE_ENTRIES/NOUVEAU_STATE_CACHE_HIER_SIZE];
 	// master dirty flag
-	uint32_t dirty;
+	uint8_t dirty;
 }nouveau_state_cache;
 
 
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c b/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c
index 746b0fac8c0..8a013bd9995 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_swtcl.c
@@ -82,9 +82,9 @@ void nouveauFallback(struct nouveau_context *nmesa, GLuint bit, GLboolean mode)
 		nmesa->Fallback |= bit;
 		if (oldfallback == 0) {
 			if (nmesa->screen->card->type<NV_10) {
-				//nv03FinishPrimitive(nmesa);
+				//nv04FinishPrimitive(nmesa);
 			} else {
-				nv10FinishPrimitive(nmesa);
+				//nv10FinishPrimitive(nmesa);
 			}
 
 			_swsetup_Wakeup(ctx);
@@ -97,7 +97,7 @@ void nouveauFallback(struct nouveau_context *nmesa, GLuint bit, GLboolean mode)
 			_swrast_flush( ctx );
 
 			if (nmesa->screen->card->type<NV_10) {
-				//nv03TriInitFunctions(ctx);
+				nv04TriInitFunctions(ctx);
 			} else {
 				nv10TriInitFunctions(ctx);
 			}
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.c b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
index 0bf20e723bb..c47ff3a985d 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.c
@@ -1,3 +1,30 @@
+/*
+ * Copyright (C) 2007 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
 #include "vblank.h" /* for DO_USLEEP */
 
 #include "nouveau_context.h"
@@ -9,18 +36,22 @@
 #include "nouveau_sync.h"
 
 nouveau_notifier *
-nouveau_notifier_new(GLcontext *ctx, GLuint handle)
+nouveau_notifier_new(GLcontext *ctx, GLuint handle, GLuint count)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 	nouveau_notifier *notifier;
 
+#ifdef NOUVEAU_RING_DEBUG
+	return NULL;
+#endif
+
 	notifier = CALLOC_STRUCT(nouveau_notifier_t);
 	if (!notifier)
 		return NULL;
 
 	notifier->mem = nouveau_mem_alloc(ctx,
 					  NOUVEAU_MEM_FB | NOUVEAU_MEM_MAPPED,
-					  32,
+					  count * NV_NOTIFIER_SIZE,
 					  0);
 	if (!notifier->mem) {
 		FREE(notifier);
@@ -49,9 +80,13 @@ nouveau_notifier_destroy(GLcontext *ctx, nouveau_notifier *notifier)
 }
 
 void
-nouveau_notifier_reset(nouveau_notifier *notifier)
+nouveau_notifier_reset(nouveau_notifier *notifier, GLuint id)
 {
-	volatile GLuint *n = notifier->mem->map;
+	volatile GLuint *n = notifier->mem->map + (id * NV_NOTIFIER_SIZE);
+
+#ifdef NOUVEAU_RING_DEBUG
+	return;
+#endif
 
 	n[NV_NOTIFY_TIME_0      /4] = 0x00000000;
 	n[NV_NOTIFY_TIME_1      /4] = 0x00000000;
@@ -60,13 +95,33 @@ nouveau_notifier_reset(nouveau_notifier *notifier)
 				       NV_NOTIFY_STATE_STATUS_SHIFT);
 }
 
+GLuint
+nouveau_notifier_status(nouveau_notifier *notifier, GLuint id)
+{
+	volatile GLuint *n = notifier->mem->map + (id * NV_NOTIFIER_SIZE);
+
+	return n[NV_NOTIFY_STATE/4] >> NV_NOTIFY_STATE_STATUS_SHIFT;
+}
+
+GLuint
+nouveau_notifier_return_val(nouveau_notifier *notifier, GLuint id)
+{
+	volatile GLuint *n = notifier->mem->map + (id * NV_NOTIFIER_SIZE);
+
+	return n[NV_NOTIFY_RETURN_VALUE/4];
+}
+
 GLboolean
-nouveau_notifier_wait_status(nouveau_notifier *notifier, GLuint status,
-							 GLuint timeout)
+nouveau_notifier_wait_status(nouveau_notifier *notifier, GLuint id,
+			     GLuint status, GLuint timeout)
 {
-	volatile GLuint *n = notifier->mem->map;
+	volatile GLuint *n = notifier->mem->map + (id * NV_NOTIFIER_SIZE);
 	unsigned int time = 0;
 
+#ifdef NOUVEAU_RING_DEBUG
+	return GL_TRUE;
+#endif
+
 	while (time <= timeout) {
 		if (n[NV_NOTIFY_STATE/4] & NV_NOTIFY_STATE_ERROR_CODE_MASK) {
 			MESSAGE("Notifier returned error: 0x%04x\n",
@@ -96,7 +151,7 @@ nouveau_notifier_wait_nop(GLcontext *ctx, nouveau_notifier *notifier,
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 	GLboolean ret;
 
-	nouveau_notifier_reset(notifier);
+	nouveau_notifier_reset(notifier, 0);
 
 	BEGIN_RING_SIZE(subc, NV_NOTIFY, 1);
 	OUT_RING       (NV_NOTIFY_STYLE_WRITE_ONLY);
@@ -104,7 +159,7 @@ nouveau_notifier_wait_nop(GLcontext *ctx, nouveau_notifier *notifier,
 	OUT_RING       (0);
 	FIRE_RING();
 
-	ret = nouveau_notifier_wait_status(notifier,
+	ret = nouveau_notifier_wait_status(notifier, 0,
 					   NV_NOTIFY_STATE_STATUS_COMPLETED,
 					   0 /* no timeout */);
 	if (ret == GL_FALSE) MESSAGE("wait on notifier failed\n");
@@ -114,7 +169,11 @@ GLboolean nouveauSyncInitFuncs(GLcontext *ctx)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
-	nmesa->syncNotifier = nouveau_notifier_new(ctx, NvSyncNotify);
+#ifdef NOUVEAU_RING_DEBUG
+	return GL_TRUE;
+#endif
+
+	nmesa->syncNotifier = nouveau_notifier_new(ctx, NvSyncNotify, 1);
 	if (!nmesa->syncNotifier) {
 		MESSAGE("Failed to create channel sync notifier\n");
 		return GL_FALSE;
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_sync.h b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
index d9e3d4b80c1..019d5f6629b 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_sync.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_sync.h
@@ -1,8 +1,36 @@
+/*
+ * Copyright (C) 2007 Ben Skeggs.
+ *
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial
+ * portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
+ * LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
+ * OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
+ * WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ */
+
 #ifndef __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
 #define NV_NOTIFY_RETURN_VALUE                                        0x00000008
@@ -24,10 +52,13 @@ typedef struct nouveau_notifier_t {
 	nouveau_mem *mem;
 } nouveau_notifier;
 
-extern nouveau_notifier *nouveau_notifier_new(GLcontext *, GLuint handle);
+extern nouveau_notifier *nouveau_notifier_new(GLcontext *, GLuint handle,
+					      GLuint count);
 extern void nouveau_notifier_destroy(GLcontext *, nouveau_notifier *);
-extern void nouveau_notifier_reset(nouveau_notifier *);
-extern GLboolean nouveau_notifier_wait_status(nouveau_notifier *r,
+extern void nouveau_notifier_reset(nouveau_notifier *, GLuint id);
+extern GLuint nouveau_notifier_status(nouveau_notifier *, GLuint id);
+extern GLuint nouveau_notifier_return_val(nouveau_notifier *, GLuint id);
+extern GLboolean nouveau_notifier_wait_status(nouveau_notifier *r, GLuint id,
 					      GLuint status, GLuint timeout);
 extern void nouveau_notifier_wait_nop(GLcontext *ctx,
       				      nouveau_notifier *, GLuint subc);
diff --git a/src/mesa/drivers/dri/nouveau/nv04_state.c b/src/mesa/drivers/dri/nouveau/nv04_state.c
new file mode 100644
index 00000000000..4129ecc50ea
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nv04_state.c
@@ -0,0 +1,538 @@
+/**************************************************************************
+
+Copyright 2007 Stephane Marchesin
+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
+ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP 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 "nouveau_context.h"
+#include "nouveau_object.h"
+#include "nouveau_fifo.h"
+#include "nouveau_reg.h"
+#include "nouveau_msg.h"
+
+#include "tnl/t_pipeline.h"
+
+#include "mtypes.h"
+#include "colormac.h"
+
+static uint32_t nv04_compare_func(GLuint f)
+{
+	switch ( f ) {
+		case GL_NEVER:		return 1;
+		case GL_LESS:		return 2;
+		case GL_EQUAL:		return 3;
+		case GL_LEQUAL:		return 4;
+		case GL_GREATER:	return 5;
+		case GL_NOTEQUAL:	return 6;
+		case GL_GEQUAL:		return 7;
+		case GL_ALWAYS:		return 8;
+	}
+	WARN_ONCE("Unable to find the function\n");
+	return 0;
+}
+
+static uint32_t nv04_blend_func(GLuint f)
+{
+	switch ( f ) {
+		case GL_ZERO:			return 0x1;
+		case GL_ONE:			return 0x2;
+		case GL_SRC_COLOR:		return 0x3;
+		case GL_ONE_MINUS_SRC_COLOR:	return 0x4;
+		case GL_SRC_ALPHA:		return 0x5;
+		case GL_ONE_MINUS_SRC_ALPHA:	return 0x6;
+		case GL_DST_ALPHA:		return 0x7;
+		case GL_ONE_MINUS_DST_ALPHA:	return 0x8;
+		case GL_DST_COLOR:		return 0x9;
+		case GL_ONE_MINUS_DST_COLOR:	return 0xA;
+		case GL_SRC_ALPHA_SATURATE:	return 0xB;
+	}
+	WARN_ONCE("Unable to find the function 0x%x\n",f);
+	return 0;
+}
+
+static void nv04_emit_control(GLcontext *ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	uint32_t control,cull;
+	GLubyte alpha_ref;
+
+	CLAMPED_FLOAT_TO_UBYTE(alpha_ref, ctx->Color.AlphaRef);
+	control=alpha_ref;
+	control|=(nv04_compare_func(ctx->Color.AlphaFunc)<<8);
+	control|=(ctx->Color.AlphaEnabled<<12);
+	control|=(1<<13);
+	control|=(ctx->Depth.Test<<14);
+	control|=(nv04_compare_func(ctx->Depth.Func)<<16);
+	if ((ctx->Polygon.CullFlag)&&(ctx->Polygon.CullFaceMode!=GL_FRONT_AND_BACK))
+	{
+		if ((ctx->Polygon.FrontFace==GL_CW)&&(ctx->Polygon.CullFaceMode==GL_FRONT))
+			cull=2;
+		if ((ctx->Polygon.FrontFace==GL_CW)&&(ctx->Polygon.CullFaceMode==GL_BACK))
+			cull=3;
+		if ((ctx->Polygon.FrontFace==GL_CCW)&&(ctx->Polygon.CullFaceMode==GL_FRONT))
+			cull=3;
+		if ((ctx->Polygon.FrontFace==GL_CCW)&&(ctx->Polygon.CullFaceMode==GL_BACK))
+			cull=2;
+	}
+	else
+		if (ctx->Polygon.CullFaceMode==GL_FRONT_AND_BACK)
+			cull=0;
+		else
+			cull=1;
+	control|=(cull<<20);
+	control|=(ctx->Color.DitherFlag<<22);
+	if ((ctx->Depth.Test)&&(ctx->Depth.Mask))
+		control|=(1<<24);
+
+	control|=(1<<30); // integer zbuffer format
+
+	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_CONTROL, 1);
+	OUT_RING_CACHE(control);
+}
+
+static void nv04_emit_blend(GLcontext *ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	uint32_t blend;
+
+	blend=0x4; // texture MODULATE_ALPHA
+	blend|=0x20; // alpha is MSB
+	switch(ctx->Light.ShadeModel) {
+		case GL_SMOOTH:blend|=(1<<6);break;
+		case GL_FLAT:  blend|=(2<<6);break;
+		default:break;
+	}
+	if (ctx->Hint.PerspectiveCorrection!=GL_FASTEST)
+		blend|=(1<<8);
+	blend|=(ctx->Fog.Enabled<<16);
+	blend|=(ctx->Color.BlendEnabled<<20);
+	blend|=(nv04_blend_func(ctx->Color.BlendSrcRGB)<<24);
+	blend|=(nv04_blend_func(ctx->Color.BlendDstRGB)<<28);
+
+	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_BLEND, 1);
+	OUT_RING_CACHE(blend);
+}
+
+static void nv04_emit_fog_color(GLcontext *ctx)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte c[4];
+	c[0] = FLOAT_TO_UBYTE( ctx->Fog.Color[0] );
+	c[1] = FLOAT_TO_UBYTE( ctx->Fog.Color[1] );
+	c[2] = FLOAT_TO_UBYTE( ctx->Fog.Color[2] );
+	c[3] = FLOAT_TO_UBYTE( ctx->Fog.Color[3] );
+	BEGIN_RING_CACHE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_FOG_COLOR, 1);
+	OUT_RING_CACHE(PACK_COLOR_8888_REV(c[0],c[1],c[2],c[3]));
+}
+
+static void nv04AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+{
+	nv04_emit_control(ctx);
+}
+
+static void nv04BlendColor(GLcontext *ctx, const GLfloat color[4])
+{
+	nv04_emit_blend(ctx);
+}
+
+static void nv04BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
+{
+	nv04_emit_blend(ctx);
+}
+
+
+static void nv04BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
+		GLenum sfactorA, GLenum dfactorA)
+{
+	nv04_emit_blend(ctx);
+}
+
+static void nv04Clear(GLcontext *ctx, GLbitfield mask)
+{
+	/* TODO */
+}
+
+static void nv04ClearColor(GLcontext *ctx, const GLfloat color[4])
+{
+	/* TODO */
+}
+
+static void nv04ClearDepth(GLcontext *ctx, GLclampd d)
+{
+	/* TODO */
+}
+
+static void nv04ClearStencil(GLcontext *ctx, GLint s)
+{
+	/* TODO */
+}
+
+static void nv04ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
+{
+	/* TODO */
+}
+
+static void nv04ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
+		GLboolean bmask, GLboolean amask )
+{
+	/* TODO */
+}
+
+static void nv04ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
+{
+	/* TODO I need love */
+}
+
+static void nv04CullFace(GLcontext *ctx, GLenum mode)
+{
+	nv04_emit_control(ctx);
+}
+
+static void nv04FrontFace(GLcontext *ctx, GLenum mode)
+{
+	/* TODO */
+}
+
+static void nv04DepthFunc(GLcontext *ctx, GLenum func)
+{
+	nv04_emit_control(ctx);
+}
+
+static void nv04DepthMask(GLcontext *ctx, GLboolean flag)
+{
+	/* TODO */
+}
+
+static void nv04DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
+{
+	/* TODO */
+}
+
+/** Specify the current buffer for writing */
+//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
+/** Specify the buffers for writing for fragment programs*/
+//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
+
+static void nv04Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+{
+	switch(cap)
+	{
+		case GL_ALPHA_TEST:
+			nv04_emit_control(ctx);
+			break;
+//		case GL_AUTO_NORMAL:
+		case GL_BLEND:
+			nv04_emit_blend(ctx);
+			break;
+//		case GL_CLIP_PLANE0:
+//		case GL_CLIP_PLANE1:
+//		case GL_CLIP_PLANE2:
+//		case GL_CLIP_PLANE3:
+//		case GL_CLIP_PLANE4:
+//		case GL_CLIP_PLANE5:
+//		case GL_COLOR_LOGIC_OP:
+//		case GL_COLOR_MATERIAL:
+//		case GL_COLOR_SUM_EXT:
+//		case GL_COLOR_TABLE:
+//		case GL_CONVOLUTION_1D:
+//		case GL_CONVOLUTION_2D:
+		case GL_CULL_FACE:
+			nv04_emit_control(ctx);
+			break;
+		case GL_DEPTH_TEST:
+			nv04_emit_control(ctx);
+			break;
+		case GL_DITHER:
+			nv04_emit_control(ctx);
+			break;
+		case GL_FOG:
+			nv04_emit_blend(ctx);
+			nv04_emit_fog_color(ctx);
+			break;
+//		case GL_HISTOGRAM:
+//		case GL_INDEX_LOGIC_OP:
+//		case GL_LIGHT0:
+//		case GL_LIGHT1:
+//		case GL_LIGHT2:
+//		case GL_LIGHT3:
+//		case GL_LIGHT4:
+//		case GL_LIGHT5:
+//		case GL_LIGHT6:
+//		case GL_LIGHT7:
+//		case GL_LIGHTING:
+//		case GL_LINE_SMOOTH:
+//		case GL_LINE_STIPPLE:
+//		case GL_MAP1_COLOR_4:
+//		case GL_MAP1_INDEX:
+//		case GL_MAP1_NORMAL:
+//		case GL_MAP1_TEXTURE_COORD_1:
+//		case GL_MAP1_TEXTURE_COORD_2:
+//		case GL_MAP1_TEXTURE_COORD_3:
+//		case GL_MAP1_TEXTURE_COORD_4:
+//		case GL_MAP1_VERTEX_3:
+//		case GL_MAP1_VERTEX_4:
+//		case GL_MAP2_COLOR_4:
+//		case GL_MAP2_INDEX:
+//		case GL_MAP2_NORMAL:
+//		case GL_MAP2_TEXTURE_COORD_1:
+//		case GL_MAP2_TEXTURE_COORD_2:
+//		case GL_MAP2_TEXTURE_COORD_3:
+//		case GL_MAP2_TEXTURE_COORD_4:
+//		case GL_MAP2_VERTEX_3:
+//		case GL_MAP2_VERTEX_4:
+//		case GL_MINMAX:
+//		case GL_NORMALIZE:
+//		case GL_POINT_SMOOTH:
+//		case GL_POLYGON_OFFSET_POINT:
+//		case GL_POLYGON_OFFSET_LINE:
+//		case GL_POLYGON_OFFSET_FILL:
+//		case GL_POLYGON_SMOOTH:
+//		case GL_POLYGON_STIPPLE:
+//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
+//		case GL_POST_CONVOLUTION_COLOR_TABLE:
+//		case GL_RESCALE_NORMAL:
+//		case GL_SCISSOR_TEST:
+//		case GL_SEPARABLE_2D:
+//		case GL_STENCIL_TEST:
+//		case GL_TEXTURE_GEN_Q:
+//		case GL_TEXTURE_GEN_R:
+//		case GL_TEXTURE_GEN_S:
+//		case GL_TEXTURE_GEN_T:
+//		case GL_TEXTURE_1D:
+//		case GL_TEXTURE_2D:
+//		case GL_TEXTURE_3D:
+	}
+}
+
+static void nv04Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
+{
+	nv04_emit_blend(ctx);
+	nv04_emit_fog_color(ctx);
+}
+   
+static void nv04Hint(GLcontext *ctx, GLenum target, GLenum mode)
+{
+	switch(target)
+	{
+		case GL_PERSPECTIVE_CORRECTION_HINT:nv04_emit_blend(ctx);break;
+		default:break;
+	}
+}
+
+static void nv04LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
+{
+	/* TODO not even in your dreams */
+}
+
+static void nv04LineWidth(GLcontext *ctx, GLfloat width)
+{
+	/* TODO */
+}
+
+static void nv04LogicOpcode(GLcontext *ctx, GLenum opcode)
+{
+	/* TODO */
+}
+
+static void nv04PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
+{
+	/* TODO */
+}
+
+static void nv04PointSize(GLcontext *ctx, GLfloat size)
+{
+	/* TODO */
+}
+
+static void nv04PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
+{
+	/* TODO */
+}
+
+/** Set the scale and units used to calculate depth values */
+static void nv04PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
+{
+	/* TODO */
+}
+
+/** Set the polygon stippling pattern */
+static void nv04PolygonStipple(GLcontext *ctx, const GLubyte *mask )
+{
+	/* TODO */
+}
+
+/* Specifies the current buffer for reading */
+void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
+/** Set rasterization mode */
+void (*RenderMode)(GLcontext *ctx, GLenum mode );
+
+/** Define the scissor box */
+static void nv04Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+{
+	/* TODO */
+}
+
+/** Select flat or smooth shading */
+static void nv04ShadeModel(GLcontext *ctx, GLenum mode)
+{
+	nv04_emit_blend(ctx);
+}
+
+/** OpenGL 2.0 two-sided StencilFunc */
+static void nv04StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
+		GLint ref, GLuint mask)
+{
+	/* TODO */
+}
+
+/** OpenGL 2.0 two-sided StencilMask */
+static void nv04StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+{
+	/* TODO */
+}
+
+/** OpenGL 2.0 two-sided StencilOp */
+static void nv04StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
+		GLenum zfail, GLenum zpass)
+{
+	/* TODO */
+}
+
+/** Control the generation of texture coordinates */
+void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
+		const GLfloat *params);
+/** Set texture environment parameters */
+void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
+		const GLfloat *param);
+/** Set texture parameters */
+void (*TexParameter)(GLcontext *ctx, GLenum target,
+		struct gl_texture_object *texObj,
+		GLenum pname, const GLfloat *params);
+
+/* Update anything that depends on the window position/size */
+static void nv04WindowMoved(nouveauContextPtr nmesa)
+{
+}
+
+/* Initialise any card-specific non-GL related state */
+static GLboolean nv04InitCard(nouveauContextPtr nmesa)
+{
+	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
+	nouveauObjectOnSubchannel(nmesa, NvSubCtxSurf3D, NvCtxSurf3D);
+
+	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_DMA_NOTIFY, 3);
+	OUT_RING(NvDmaFB);
+	OUT_RING(NvDmaFB);
+	OUT_RING(NvDmaFB);
+	BEGIN_RING_SIZE(NvSub3D, NV04_DX5_TEXTURED_TRIANGLE_SURFACE, 1);
+	OUT_RING(NvCtxSurf3D);
+	return GL_TRUE;
+}
+
+/* Update buffer offset/pitch/format */
+static GLboolean nv04BindBuffers(nouveauContextPtr nmesa, int num_color,
+		nouveau_renderbuffer **color,
+		nouveau_renderbuffer *depth)
+{
+	GLuint x, y, w, h;
+
+	w = color[0]->mesa.Width;
+	h = color[0]->mesa.Height;
+	x = nmesa->drawX;
+	y = nmesa->drawY;
+
+	/* FIXME pitches have to be aligned ! */
+	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_PITCH, 2);
+	OUT_RING(color[0]->pitch|(depth?(depth->pitch<<16):0));
+	OUT_RING(color[0]->offset);
+
+	if (depth) {
+		BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_OFFSET_ZETA, 1);
+		OUT_RING(depth->offset);
+	}
+
+	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_CLIP_HORIZONTAL, 2);
+	OUT_RING((w<<16)|x);
+	OUT_RING((h<<16)|y);
+
+	/* FIXME not sure... */
+	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_CLIP_SIZE, 1);
+	OUT_RING((h<<16)|w);
+
+	BEGIN_RING_SIZE(NvSubCtxSurf3D, NV04_CONTEXT_SURFACES_3D_FORMAT, 1);
+	if (color[0]->mesa._ActualFormat == GL_RGBA8)
+		OUT_RING(108/*A8R8G8B8*/);
+	else
+		OUT_RING(103/*R5G6B5*/);
+
+	return GL_TRUE;
+}
+
+void nv04InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	func->AlphaFunc			= nv04AlphaFunc;
+	func->BlendColor		= nv04BlendColor;
+	func->BlendEquationSeparate	= nv04BlendEquationSeparate;
+	func->BlendFuncSeparate		= nv04BlendFuncSeparate;
+	func->Clear			= nv04Clear;
+	func->ClearColor		= nv04ClearColor;
+	func->ClearDepth		= nv04ClearDepth;
+	func->ClearStencil		= nv04ClearStencil;
+	func->ClipPlane			= nv04ClipPlane;
+	func->ColorMask			= nv04ColorMask;
+	func->ColorMaterial		= nv04ColorMaterial;
+	func->CullFace			= nv04CullFace;
+	func->FrontFace			= nv04FrontFace;
+	func->DepthFunc			= nv04DepthFunc;
+	func->DepthMask			= nv04DepthMask;
+	func->DepthRange		= nv04DepthRange;
+	func->Enable			= nv04Enable;
+	func->Fogfv			= nv04Fogfv;
+	func->Hint			= nv04Hint;
+/*	func->Lightfv			= nv04Lightfv;*/
+/*	func->LightModelfv		= nv04LightModelfv; */
+	func->LineStipple		= nv04LineStipple;		/* Not for NV04 */
+	func->LineWidth			= nv04LineWidth;
+	func->LogicOpcode		= nv04LogicOpcode;
+	func->PointParameterfv		= nv04PointParameterfv;
+	func->PointSize			= nv04PointSize;
+	func->PolygonMode		= nv04PolygonMode;
+	func->PolygonOffset		= nv04PolygonOffset;
+	func->PolygonStipple		= nv04PolygonStipple;		/* Not for NV04 */
+/*	func->ReadBuffer		= nv04ReadBuffer;*/
+/*	func->RenderMode		= nv04RenderMode;*/
+	func->Scissor			= nv04Scissor;
+	func->ShadeModel		= nv04ShadeModel;
+	func->StencilFuncSeparate	= nv04StencilFuncSeparate;
+	func->StencilMaskSeparate	= nv04StencilMaskSeparate;
+	func->StencilOpSeparate		= nv04StencilOpSeparate;
+/*	func->TexGen			= nv04TexGen;*/
+/*	func->TexParameter		= nv04TexParameter;*/
+/*	func->TextureMatrix		= nv04TextureMatrix;*/
+
+	nmesa->hw_func.InitCard		= nv04InitCard;
+	nmesa->hw_func.BindBuffers	= nv04BindBuffers;
+	nmesa->hw_func.WindowMoved	= nv04WindowMoved;
+}
diff --git a/src/mesa/drivers/dri/nouveau/nv04_swtcl.c b/src/mesa/drivers/dri/nouveau/nv04_swtcl.c
new file mode 100644
index 00000000000..9b5332b77a7
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nv04_swtcl.c
@@ -0,0 +1,618 @@
+/*
+ * Copyright 2007 Stephane Marchesin. 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
+ * VIA, S3 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.
+ */
+
+/* Software TCL for NV04, NV05, NV06 */
+
+#include <stdio.h>
+#include <math.h>
+
+#include "glheader.h"
+#include "context.h"
+#include "mtypes.h"
+#include "macros.h"
+#include "colormac.h"
+#include "enums.h"
+
+#include "swrast/swrast.h"
+#include "swrast_setup/swrast_setup.h"
+#include "tnl/t_context.h"
+#include "tnl/t_pipeline.h"
+
+#include "nouveau_swtcl.h"
+#include "nv04_swtcl.h"
+#include "nouveau_context.h"
+#include "nouveau_span.h"
+#include "nouveau_reg.h"
+#include "nouveau_tex.h"
+#include "nouveau_fifo.h"
+#include "nouveau_msg.h"
+#include "nouveau_object.h"
+
+static void nv04RasterPrimitive( GLcontext *ctx, GLenum rprim, GLuint hwprim );
+static void nv04RenderPrimitive( GLcontext *ctx, GLenum prim );
+static void nv04ResetLineStipple( GLcontext *ctx );
+
+
+static inline void nv04_2triangles(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2,nouveauVertex* v3,nouveauVertex* v4,nouveauVertex* v5)
+{
+	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xA),49);
+	OUT_RINGp(v0,8);
+	OUT_RINGp(v1,8);
+	OUT_RINGp(v2,8);
+	OUT_RINGp(v3,8);
+	OUT_RINGp(v4,8);
+	OUT_RINGp(v5,8);
+	OUT_RING(0xFEDCBA);
+}
+
+static inline void nv04_1triangle(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2)
+{
+	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xD),25);
+	OUT_RINGp(v0,8);
+	OUT_RINGp(v1,8);
+	OUT_RINGp(v2,8);
+	OUT_RING(0xFED);
+}
+
+static inline void nv04_1quad(struct nouveau_context *nmesa,nouveauVertex* v0,nouveauVertex* v1,nouveauVertex* v2,nouveauVertex* v3)
+{
+	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0xC),33);
+	OUT_RINGp(v0,8);
+	OUT_RINGp(v1,8);
+	OUT_RINGp(v2,8);
+	OUT_RINGp(v3,8);
+	OUT_RING(0xFECEDC);
+}
+
+static inline void nv04_render_points(GLcontext *ctx,GLuint first,GLuint last)
+{
+	WARN_ONCE("Unimplemented\n");
+}
+
+static inline void nv04_render_line(GLcontext *ctx,GLuint v1,GLuint v2)
+{
+	WARN_ONCE("Unimplemented\n");
+}
+
+static inline void nv04_render_triangle(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+
+	nv04_1triangle(nmesa,
+			(nouveauVertex*)(vertptr+v1*vertsize),
+			(nouveauVertex*)(vertptr+v2*vertsize),
+			(nouveauVertex*)(vertptr+v3*vertsize)
+		  );
+}
+
+static inline void nv04_render_quad(GLcontext *ctx,GLuint v1,GLuint v2,GLuint v3,GLuint v4)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+
+	nv04_1quad(nmesa,
+			(nouveauVertex*)(vertptr+v1*vertsize),
+			(nouveauVertex*)(vertptr+v2*vertsize),
+			(nouveauVertex*)(vertptr+v3*vertsize),
+			(nouveauVertex*)(vertptr+v4*vertsize)
+		  );
+}
+
+/**********************************************************************/
+/*               Render unclipped begin/end objects                   */
+/**********************************************************************/
+
+static void nv04_render_points_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// erm
+}
+
+static void nv04_render_lines_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// umm
+}
+
+static void nv04_render_line_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// yeah
+}
+
+static void nv04_render_line_loop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// right
+}
+
+static void nv04_render_triangles_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	int i;
+
+	for(i=start;i<count-5;i+=6)
+		nv04_2triangles(nmesa,
+				(nouveauVertex*)(vertptr+(i+0)*vertsize),
+				(nouveauVertex*)(vertptr+(i+1)*vertsize),
+				(nouveauVertex*)(vertptr+(i+2)*vertsize),
+				(nouveauVertex*)(vertptr+(i+3)*vertsize),
+				(nouveauVertex*)(vertptr+(i+4)*vertsize),
+				(nouveauVertex*)(vertptr+(i+5)*vertsize)
+			       );
+	if (i!=count)
+	{
+		nv04_1triangle(nmesa,
+				(nouveauVertex*)(vertptr+(i+0)*vertsize),
+				(nouveauVertex*)(vertptr+(i+1)*vertsize),
+				(nouveauVertex*)(vertptr+(i+2)*vertsize)
+			       );
+		i+=3;
+	}
+	if (i!=count)
+		printf("oops\n");
+}
+
+static void nv04_render_tri_strip_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC};
+	int i,j;
+
+	for(i=start;i<count;i+=14)
+	{
+		int numvert=MIN2(16,count-i);
+		int numtri=numvert-2;
+		if (numvert<3)
+			break;
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),numvert*8);
+		for(j=0;j<numvert;j++)
+			OUT_RINGp((nouveauVertex*)(vertptr+(i+j)*vertsize),8);
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
+		for(j=0;j<numtri/2;j++)
+			OUT_RING(striptbl[j]);
+		if (numtri%2)
+			OUT_RING(striptbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_render_tri_fan_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	uint32_t fantbl[]={0x320210,0x540430,0x760650,0x980870,0xBA0A90,0xDC0CB0,0xFE0ED0};
+	int i,j;
+
+	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),8);
+	OUT_RINGp((nouveauVertex*)(vertptr+start*vertsize),8);
+
+	for(i=start+1;i<count;i+=14)
+	{
+		int numvert=MIN2(15,count-i);
+		int numtri=numvert-1;
+		if (numvert<3)
+			break;
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x1),numvert*8);
+
+		for(j=0;j<numvert;j++)
+			OUT_RINGp((nouveauVertex*)(vertptr+(i+j)*vertsize),8);
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
+		for(j=0;j<numtri/2;j++)
+			OUT_RING(fantbl[j]);
+		if (numtri%2)
+			OUT_RING(fantbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_render_quads_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	int i;
+
+	for(i=start;i<count;i+=4)
+		nv04_1quad(nmesa,
+				(nouveauVertex*)(vertptr+(i+0)*vertsize),
+				(nouveauVertex*)(vertptr+(i+1)*vertsize),
+				(nouveauVertex*)(vertptr+(i+2)*vertsize),
+				(nouveauVertex*)(vertptr+(i+3)*vertsize)
+			       );
+}
+
+static void nv04_render_noop_verts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+}
+
+static void (*nv04_render_tab_verts[GL_POLYGON+2])(GLcontext *,
+							   GLuint,
+							   GLuint,
+							   GLuint) =
+{
+   nv04_render_points_verts,
+   nv04_render_lines_verts,
+   nv04_render_line_loop_verts,
+   nv04_render_line_strip_verts,
+   nv04_render_triangles_verts,
+   nv04_render_tri_strip_verts,
+   nv04_render_tri_fan_verts,
+   nv04_render_quads_verts,
+   nv04_render_tri_strip_verts,  //nv04_render_quad_strip_verts
+   nv04_render_tri_fan_verts,    //nv04_render_poly_verts
+   nv04_render_noop_verts,
+};
+
+
+static void nv04_render_points_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// erm
+}
+
+static void nv04_render_lines_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// umm
+}
+
+static void nv04_render_line_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// yeah
+}
+
+static void nv04_render_line_loop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	// right
+}
+
+static void nv04_render_triangles_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
+	int i;
+
+	for(i=start;i<count-5;i+=6)
+		nv04_2triangles(nmesa,
+				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+2]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+3]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+4]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+5]*vertsize)
+			       );
+	if (i!=count)
+	{
+		nv04_1triangle(nmesa,
+				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+2]*vertsize)
+			       );
+		i+=3;
+	}
+	if (i!=count)
+		printf("oops\n");
+}
+
+static void nv04_render_tri_strip_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	uint32_t striptbl[]={0x321210,0x543432,0x765654,0x987876,0xBA9A98,0xDCBCBA,0xFEDEDC};
+	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
+	int i,j;
+
+	for(i=start;i<count;i+=14)
+	{
+		int numvert=MIN2(16,count-i);
+		int numtri=numvert-2;
+		if (numvert<3)
+			break;
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),numvert*8);
+		for(j=0;j<numvert;j++)
+			OUT_RINGp((nouveauVertex*)(vertptr+elt[i+j]*vertsize),8);
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
+		for(j=0;j<numtri/2;j++)
+			OUT_RING(striptbl[j]);
+		if (numtri%2)
+			OUT_RING(striptbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_render_tri_fan_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	uint32_t fantbl[]={0x320210,0x540430,0x760650,0x980870,0xBA0A90,0xDC0CB0,0xFE0ED0};
+	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
+	int i,j;
+
+	BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x0),8);
+	OUT_RINGp((nouveauVertex*)(vertptr+elt[start]*vertsize),8);
+
+	for(i=start+1;i<count;i+=14)
+	{
+		int numvert=MIN2(15,count-i);
+		int numtri=numvert-2;
+		if (numvert<3)
+			break;
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_TLVERTEX_SX(0x1),numvert*8);
+
+		for(j=0;j<numvert;j++)
+			OUT_RINGp((nouveauVertex*)(vertptr+elt[i+j]*vertsize),8);
+
+		BEGIN_RING_SIZE(NvSub3D,NV04_DX5_TEXTURED_TRIANGLE_DRAW|NONINC_METHOD,(numtri+1)/2);
+		for(j=0;j<numtri/2;j++)
+			OUT_RING(fantbl[j]);
+		if (numtri%2)
+			OUT_RING(fantbl[numtri/2]&0xFFF);
+	}
+}
+
+static void nv04_render_quads_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts;
+	int i;
+
+	for(i=start;i<count;i+=4)
+		nv04_1quad(nmesa,
+				(nouveauVertex*)(vertptr+elt[i+0]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+1]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+2]*vertsize),
+				(nouveauVertex*)(vertptr+elt[i+3]*vertsize)
+			       );
+}
+
+static void nv04_render_noop_elts(GLcontext *ctx,GLuint start,GLuint count,GLuint flags)
+{
+}
+
+static void (*nv04_render_tab_elts[GL_POLYGON+2])(GLcontext *,
+							   GLuint,
+							   GLuint,
+							   GLuint) =
+{
+   nv04_render_points_elts,
+   nv04_render_lines_elts,
+   nv04_render_line_loop_elts,
+   nv04_render_line_strip_elts,
+   nv04_render_triangles_elts,
+   nv04_render_tri_strip_elts,
+   nv04_render_tri_fan_elts,
+   nv04_render_quads_elts,
+   nv04_render_tri_strip_elts,   // nv04_render_quad_strip_elts,
+   nv04_render_tri_fan_elts,     // nv04_render_poly_elts,
+   nv04_render_noop_elts,
+};
+
+
+/**********************************************************************/
+/*                    Choose render functions                         */
+/**********************************************************************/
+
+
+#define EMIT_ATTR( ATTR, STYLE )					\
+do {									\
+   nmesa->vertex_attrs[nmesa->vertex_attr_count].attrib = (ATTR);	\
+   nmesa->vertex_attrs[nmesa->vertex_attr_count].format = (STYLE);	\
+   nmesa->vertex_attr_count++;						\
+} while (0)
+
+#define EMIT_PAD( N )							\
+do {									\
+   nmesa->vertex_attrs[nmesa->vertex_attr_count].attrib = 0;		\
+   nmesa->vertex_attrs[nmesa->vertex_attr_count].format = EMIT_PAD;	\
+   nmesa->vertex_attrs[nmesa->vertex_attr_count].offset = (N);		\
+   nmesa->vertex_attr_count++;						\
+} while (0)
+
+static void nv04_render_clipped_line(GLcontext *ctx,GLuint ii,GLuint jj)
+{
+}
+
+static void nv04_render_clipped_poly(GLcontext *ctx,const GLuint *elts,GLuint n)
+{
+}
+
+static void nv04ChooseRenderState(GLcontext *ctx)
+{
+	TNLcontext *tnl = TNL_CONTEXT(ctx);
+
+	tnl->Driver.Render.PrimTabVerts = nv04_render_tab_verts;
+	tnl->Driver.Render.PrimTabElts = nv04_render_tab_elts;
+	tnl->Driver.Render.ClippedLine = nv04_render_clipped_line;
+	tnl->Driver.Render.ClippedPolygon = nv04_render_clipped_poly;
+	tnl->Driver.Render.Points = nv04_render_points;
+	tnl->Driver.Render.Line = nv04_render_line;
+	tnl->Driver.Render.Triangle = nv04_render_triangle;
+	tnl->Driver.Render.Quad = nv04_render_quad;
+}
+
+
+
+static inline void nv04OutputVertexFormat(struct nouveau_context* nmesa)
+{
+	GLcontext* ctx=nmesa->glCtx;
+	DECLARE_RENDERINPUTS(index);
+
+	/*
+	 * Tell t_vertex about the vertex format
+	 */
+	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
+
+	// SX SY SZ INVW
+	// FIXME : we use W instead of INVW, but since W=1 it doesn't matter
+	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_POS))
+		EMIT_ATTR(_TNL_ATTRIB_POS,EMIT_4F_VIEWPORT);
+	else
+		EMIT_PAD(4*sizeof(float));
+
+	// COLOR
+	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_COLOR0))
+		EMIT_ATTR(_TNL_ATTRIB_COLOR0,EMIT_4UB_4F_ABGR);
+	else
+		EMIT_PAD(4);
+
+	// SPECULAR
+	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_COLOR1))
+		EMIT_ATTR(_TNL_ATTRIB_COLOR1,EMIT_4UB_4F_ABGR);
+	else
+		EMIT_PAD(4);
+
+	// TEXTURE
+	if (RENDERINPUTS_TEST(index, _TNL_ATTRIB_TEX0))
+		EMIT_ATTR(_TNL_ATTRIB_TEX0,EMIT_2F);
+	else
+		EMIT_PAD(2*sizeof(float));
+
+	nmesa->vertex_size=_tnl_install_attrs( ctx,
+			nmesa->vertex_attrs, 
+			nmesa->vertex_attr_count,
+			ctx->Viewport._WindowMap.m, 0 );
+}
+
+
+static void nv04ChooseVertexState( GLcontext *ctx )
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	TNLcontext *tnl = TNL_CONTEXT(ctx);
+	DECLARE_RENDERINPUTS(index);
+
+	RENDERINPUTS_COPY(index, tnl->render_inputs_bitset);
+	if (!RENDERINPUTS_EQUAL(index, nmesa->render_inputs_bitset))
+	{
+		RENDERINPUTS_COPY(nmesa->render_inputs_bitset, index);
+		nv04OutputVertexFormat(nmesa);
+	}
+}
+
+
+/**********************************************************************/
+/*                 High level hooks for t_vb_render.c                 */
+/**********************************************************************/
+
+
+static void nv04RenderStart(GLcontext *ctx)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (nmesa->new_state) {
+		nmesa->new_render_state |= nmesa->new_state;
+	}
+
+	if (nmesa->new_render_state) {
+		nv04ChooseVertexState(ctx);
+		nv04ChooseRenderState(ctx);
+		nmesa->new_render_state = 0;
+	}
+}
+
+static void nv04RenderFinish(GLcontext *ctx)
+{
+}
+
+
+/* System to flush dma and emit state changes based on the rasterized
+ * primitive.
+ */
+void nv04RasterPrimitive(GLcontext *ctx,
+		GLenum glprim,
+		GLuint hwprim)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+
+	assert (!nmesa->new_state);
+
+	if (hwprim != nmesa->current_primitive)
+	{
+		nmesa->current_primitive=hwprim;
+		
+	}
+}
+
+static const GLuint hw_prim[GL_POLYGON+1] = {
+	GL_POINTS+1,
+	GL_LINES+1,
+	GL_LINE_STRIP+1,
+	GL_LINE_LOOP+1,
+	GL_TRIANGLES+1,
+	GL_TRIANGLE_STRIP+1,
+	GL_TRIANGLE_FAN+1,
+	GL_QUADS+1,
+	GL_QUAD_STRIP+1,
+	GL_POLYGON+1
+};
+
+/* Callback for mesa:
+ */
+static void nv04RenderPrimitive( GLcontext *ctx, GLuint prim )
+{
+	nv04RasterPrimitive( ctx, prim, hw_prim[prim] );
+}
+
+static void nv04ResetLineStipple( GLcontext *ctx )
+{
+	/* FIXME do something here */
+	WARN_ONCE("Unimplemented nv04ResetLineStipple\n");
+}
+
+
+/**********************************************************************/
+/*                            Initialization.                         */
+/**********************************************************************/
+
+void nv04TriInitFunctions(GLcontext *ctx)
+{
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	TNLcontext *tnl = TNL_CONTEXT(ctx);
+
+	tnl->Driver.RunPipeline = nouveauRunPipeline;
+	tnl->Driver.Render.Start = nv04RenderStart;
+	tnl->Driver.Render.Finish = nv04RenderFinish;
+	tnl->Driver.Render.PrimitiveNotify = nv04RenderPrimitive;
+	tnl->Driver.Render.ResetLineStipple = nv04ResetLineStipple;
+	tnl->Driver.Render.BuildVertices = _tnl_build_vertices;
+	tnl->Driver.Render.CopyPV = _tnl_copy_pv;
+	tnl->Driver.Render.Interp = _tnl_interp;
+
+	_tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 32 );
+
+	nmesa->verts = (GLubyte *)tnl->clipspace.vertex_buf;
+}
+
+
diff --git a/src/mesa/drivers/dri/nouveau/nv04_swtcl.h b/src/mesa/drivers/dri/nouveau/nv04_swtcl.h
new file mode 100644
index 00000000000..42dde5383eb
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nv04_swtcl.h
@@ -0,0 +1,12 @@
+#ifndef __NV04_SWTCL_H__
+#define __NV04_SWTCL_H__
+
+#include "mtypes.h"
+
+extern void nv04Fallback( GLcontext *ctx, GLuint bit, GLboolean mode );
+extern void nv04FinishPrimitive(struct nouveau_context *nmesa);
+extern void nv04TriInitFunctions(GLcontext *ctx);
+#define FALLBACK( nmesa, bit, mode ) nouveauFallback( nmesa->glCtx, bit, mode )
+
+#endif /* __NV04_SWTCL_H__ */
+
diff --git a/src/mesa/drivers/dri/nouveau/nv10_state.c b/src/mesa/drivers/dri/nouveau/nv10_state.c
index c028be2867c..5f304ccab92 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_state.c
@@ -34,6 +34,34 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "mtypes.h"
 #include "colormac.h"
 
+static void nv10ViewportScale(nouveauContextPtr nmesa)
+{
+	GLcontext *ctx = nmesa->glCtx;
+	GLuint w = ctx->Viewport.Width;
+	GLuint h = ctx->Viewport.Height;
+
+	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;
+		}
+	} 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);
+}
+
 static void nv10AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
@@ -41,8 +69,8 @@ static void nv10AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
 	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
 
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC, 2);
-	OUT_RING_CACHE(func);     /* NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC */
-	OUT_RING_CACHE(ubRef);    /* NV10_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF  */
+	OUT_RING_CACHE(func);
+	OUT_RING_CACHE(ubRef);
 }
 
 static void nv10BlendColor(GLcontext *ctx, const GLfloat color[4])
@@ -62,8 +90,11 @@ static void nv10BlendColor(GLcontext *ctx, const GLfloat color[4])
 static void nv10BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	assert( modeRGB == modeA );
+
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_EQUATION, 1);
-	OUT_RING_CACHE((modeA<<16) | modeRGB);
+	OUT_RING_CACHE(modeRGB);
 }
 
 
@@ -71,30 +102,52 @@ static void nv10BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfac
 		GLenum sfactorA, GLenum dfactorA)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	assert( sfactorRGB == sfactorA );
+	assert( dfactorRGB == dfactorA );
+
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC, 2);
-	OUT_RING_CACHE((sfactorA<<16) | sfactorRGB);
-	OUT_RING_CACHE((dfactorA<<16) | dfactorRGB);
+	OUT_RING_CACHE(sfactorRGB);
+	OUT_RING_CACHE(dfactorRGB);
+}
+
+static void nv10Clear(GLcontext *ctx, GLbitfield mask)
+{
+	/* TODO */
 }
 
-/*
 static void nv10ClearColor(GLcontext *ctx, const GLfloat color[4])
 {
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte c[4];
+	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,color);
+	nmesa->clear_color_value = PACK_COLOR_8888(c[3],c[0],c[1],c[2]);
 }
 
 static void nv10ClearDepth(GLcontext *ctx, GLclampd d)
 {
-}
-*/
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
-/* we're don't support indexed buffers
-   void (*ClearIndex)(GLcontext *ctx, GLuint index)
- */
+/*	switch (ctx->DrawBuffer->_DepthBuffer->DepthBits) {
+		case 16:
+			nmesa->clear_value = (uint32_t)(d*0x7FFF);
+			break;
+		case 24:*/
+			nmesa->clear_value = ((nmesa->clear_value&0x000000FF) |
+				(((uint32_t)(d*0xFFFFFF))<<8));
+/*			break;
+	}*/
+}
 
-/*
 static void nv10ClearStencil(GLcontext *ctx, GLint s)
 {
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+/*	if (ctx->DrawBuffer->_DepthBuffer->DepthBits == 24) {*/
+		nmesa->clear_value = ((nmesa->clear_value&0xFFFFFF00)|
+			(s&0x000000FF));
+/*	}*/
 }
-*/
 
 static void nv10ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
 {
@@ -106,18 +159,17 @@ static void nv10ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
 	OUT_RING_CACHEf(equation[3]);
 }
 
-/* Seems does not support alpha in color mask */
 static void nv10ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
 		GLboolean bmask, GLboolean amask )
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_COLOR_MASK, 1);
-	OUT_RING_CACHE(/*((amask && 0x01) << 24) |*/ ((rmask && 0x01) << 16) | ((gmask && 0x01)<< 8) | ((bmask && 0x01) << 0));
+	OUT_RING_CACHE(((amask && 0x01) << 24) | ((rmask && 0x01) << 16) | ((gmask && 0x01)<< 8) | ((bmask && 0x01) << 0));
 }
 
 static void nv10ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
 {
-	// TODO I need love
+	/* TODO I need love */
 }
 
 static void nv10CullFace(GLcontext *ctx, GLenum mode)
@@ -151,9 +203,17 @@ static void nv10DepthMask(GLcontext *ctx, GLboolean flag)
 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;
+	}
+
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
-	OUT_RING_CACHEf(nearval);
-	OUT_RING_CACHEf(farval);
+	OUT_RING_CACHEf(nearval * depth_scale);
+	OUT_RING_CACHEf(farval * depth_scale);
+
+	nv10ViewportScale(nmesa);
 }
 
 /** Specify the current buffer for writing */
@@ -309,7 +369,7 @@ static void nv10Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
     switch(pname)
     {
         case GL_FOG_MODE:
-            BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 1);
+            //BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_FOG_MODE, 1);
             //OUT_RING_CACHE (params);
             break;
             /* TODO: unsure about the rest.*/
@@ -321,7 +381,7 @@ static void nv10Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
    
 static void nv10Hint(GLcontext *ctx, GLenum target, GLenum mode)
 {
-	// TODO I need love (fog and line_smooth hints)
+	/* TODO I need love (fog and line_smooth hints) */
 }
 
 // void (*IndexMask)(GLcontext *ctx, GLuint mask);
@@ -449,6 +509,7 @@ static void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params)
 
 static void nv10LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
 {
+	/* Not for NV10 */
 }
 
 static void nv10LineWidth(GLcontext *ctx, GLfloat width)
@@ -472,7 +533,6 @@ static void nv10PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *pa
 	
 }
 
-/** Specify the diameter of rasterized points */
 static void nv10PointSize(GLcontext *ctx, GLfloat size)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
@@ -480,7 +540,6 @@ static void nv10PointSize(GLcontext *ctx, GLfloat size)
 	OUT_RING_CACHE(((int) (size * 8.0)) & -4);
 }
 
-/** Select a polygon rasterization mode */
 static void nv10PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
@@ -496,9 +555,20 @@ static void nv10PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
 }
 
 /** Set the scale and units used to calculate depth values */
-void (*PolygonOffset)(GLcontext *ctx, GLfloat factor, GLfloat units);
+static void nv10PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 2);
+        OUT_RING_CACHEf(factor);
+        OUT_RING_CACHEf(units);
+}
+
 /** Set the polygon stippling pattern */
-void (*PolygonStipple)(GLcontext *ctx, const GLubyte *mask );
+static void nv10PolygonStipple(GLcontext *ctx, const GLubyte *mask )
+{
+	/* Not for NV10 */
+}
+
 /* Specifies the current buffer for reading */
 void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
 /** Set rasterization mode */
@@ -524,6 +594,7 @@ static void nv10StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
+	/* NV10 do not have separate FRONT and BACK stencils */
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_FUNC_FUNC, 3);
 	OUT_RING_CACHE(func);
 	OUT_RING_CACHE(ref);
@@ -535,6 +606,7 @@ static void nv10StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
+	/* NV10 do not have separate FRONT and BACK stencils */
 	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_MASK, 1);
 	OUT_RING_CACHE(mask);
 }
@@ -545,7 +617,8 @@ static void nv10StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
-	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL, 1);
+	/* NV10 do not have separate FRONT and BACK stencils */
+	BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_STENCIL_OP_FAIL, 3);
 	OUT_RING_CACHE(fail);
 	OUT_RING_CACHE(zfail);
 	OUT_RING_CACHE(zpass);
@@ -561,22 +634,78 @@ void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
 void (*TexParameter)(GLcontext *ctx, GLenum target,
 		struct gl_texture_object *texObj,
 		GLenum pname, const GLfloat *params);
-void (*TextureMatrix)(GLcontext *ctx, GLuint unit, const GLmatrix *mat);
 
-/** Set the viewport */
-static void nv10Viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+static void nv10TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
 {
-    /* TODO: Where do the VIEWPORT_XFRM_* regs come in? */
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
-    OUT_RING_CACHE((w << 16) | x);
-    OUT_RING_CACHE((h << 16) | y);
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
+        /*XXX: This SHOULD work.*/
+        OUT_RING_CACHEp(mat->m, 16);
+}
+
+/* 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;
+
+        BEGIN_RING_CACHE(NvSub3D, NV10_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
+        OUT_RING_CACHE((w << 16) | x);
+        OUT_RING_CACHE((h << 16) | y);
+
+	/* 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);
+	}
+
+	nv10ViewportScale(nmesa);
 }
 
 /* Initialise any card-specific non-GL related state */
 static GLboolean nv10InitCard(nouveauContextPtr nmesa)
 {
-   return GL_TRUE;
+	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 */
+	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 */
+
+	BEGIN_RING_SIZE(NvSub3D, 0x0290, 1);
+	OUT_RING(0x00100001);
+	BEGIN_RING_SIZE(NvSub3D, 0x03f4, 1);
+	OUT_RING(0);
+
+	/* not for nv10, only for >= nv11 */
+	if ((nmesa->screen->card->id>>4) >= 0x11) {
+	        BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
+        	OUT_RING(0);
+	        OUT_RING(1);
+	        OUT_RING(2);
+	}
+
+	return GL_TRUE;
 }
 
 /* Update buffer offset/pitch/format */
@@ -584,12 +713,39 @@ static GLboolean nv10BindBuffers(nouveauContextPtr nmesa, int num_color,
 				 nouveau_renderbuffer **color,
 				 nouveau_renderbuffer *depth)
 {
-   return GL_TRUE;
-}
+	GLuint x, y, w, h;
+	GLuint pitch, format, depth_pitch;
+
+	w = color[0]->mesa.Width;
+	h = color[0]->mesa.Height;
+	x = nmesa->drawX;
+	y = nmesa->drawY;
+
+	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);
+	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);
 
-/* Update anything that depends on the window position/size */
-static void nv10WindowMoved(nouveauContextPtr nmesa)
-{
+	/* 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);
+
+	return GL_TRUE;
 }
 
 void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
@@ -600,11 +756,10 @@ void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 	func->BlendColor		= nv10BlendColor;
 	func->BlendEquationSeparate	= nv10BlendEquationSeparate;
 	func->BlendFuncSeparate		= nv10BlendFuncSeparate;
-#if 0
+	func->Clear			= nv10Clear;
 	func->ClearColor		= nv10ClearColor;
 	func->ClearDepth		= nv10ClearDepth;
 	func->ClearStencil		= nv10ClearStencil;
-#endif
 	func->ClipPlane			= nv10ClipPlane;
 	func->ColorMask			= nv10ColorMask;
 	func->ColorMaterial		= nv10ColorMaterial;
@@ -618,32 +773,26 @@ void nv10InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 	func->Hint			= nv10Hint;
 	func->Lightfv			= nv10Lightfv;
 /*	func->LightModelfv		= nv10LightModelfv; */
-	func->LineStipple		= nv10LineStipple;
+	func->LineStipple		= nv10LineStipple;		/* Not for NV10 */
 	func->LineWidth			= nv10LineWidth;
 	func->LogicOpcode		= nv10LogicOpcode;
 	func->PointParameterfv		= nv10PointParameterfv;
 	func->PointSize			= nv10PointSize;
 	func->PolygonMode		= nv10PolygonMode;
-#if 0
 	func->PolygonOffset		= nv10PolygonOffset;
-	func->PolygonStipple		= nv10PolygonStipple;
-	func->ReadBuffer		= nv10ReadBuffer;
-	func->RenderMode		= nv10RenderMode;
-#endif
+	func->PolygonStipple		= nv10PolygonStipple;		/* Not for NV10 */
+/*	func->ReadBuffer		= nv10ReadBuffer;*/
+/*	func->RenderMode		= nv10RenderMode;*/
 	func->Scissor			= nv10Scissor;
 	func->ShadeModel		= nv10ShadeModel;
 	func->StencilFuncSeparate	= nv10StencilFuncSeparate;
 	func->StencilMaskSeparate	= nv10StencilMaskSeparate;
 	func->StencilOpSeparate		= nv10StencilOpSeparate;
-#if 0
-	func->TexGen			= nv10TexGen;
-	func->TexParameter		= nv10TexParameter;
+/*	func->TexGen			= nv10TexGen;*/
+/*	func->TexParameter		= nv10TexParameter;*/
 	func->TextureMatrix		= nv10TextureMatrix;
-#endif
-	func->Viewport			= nv10Viewport;
 
 	nmesa->hw_func.InitCard		= nv10InitCard;
 	nmesa->hw_func.BindBuffers	= nv10BindBuffers;
 	nmesa->hw_func.WindowMoved	= nv10WindowMoved;
 }
-
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
index 37d9f001d12..32da40661b4 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
@@ -23,7 +23,7 @@
  * DEALINGS IN THE SOFTWARE.
  */
 
-/* Software TCL for NV10, NV20, NV30, NV40, G70 */
+/* Software TCL for NV10, NV20, NV30, NV40, NV50 */
 
 #include <stdio.h>
 #include <math.h>
@@ -56,9 +56,6 @@ static void nv10ResetLineStipple( GLcontext *ctx );
 
 
 
-/* the size above which we fire the ring. this is a performance-tunable */
-#define NOUVEAU_FIRE_SIZE (2048/4)
-
 static inline void nv10StartPrimitive(struct nouveau_context* nmesa,uint32_t primitive,uint32_t size)
 {
 	if (nmesa->screen->card->type==NV_10)
@@ -99,52 +96,6 @@ static inline void nv10ExtendPrimitive(struct nouveau_context* nmesa, int size)
 	}
 }
 
-static inline void nv10_draw_quad(nouveauContextPtr nmesa,
-		nouveauVertexPtr v0,
-		nouveauVertexPtr v1,
-		nouveauVertexPtr v2,
-		nouveauVertexPtr v3)
-{
-	GLuint vertsize = nmesa->vertex_size;
-	nv10ExtendPrimitive(nmesa, 4 * 4 * vertsize);
-
-	OUT_RINGp(v0,vertsize);
-	OUT_RINGp(v1,vertsize);
-	OUT_RINGp(v2,vertsize);
-	OUT_RINGp(v3,vertsize);
-}
-
-static inline void nv10_draw_triangle(nouveauContextPtr nmesa,
-		nouveauVertexPtr v0,
-		nouveauVertexPtr v1,
-		nouveauVertexPtr v2)
-{
-	GLuint vertsize = nmesa->vertex_size;
-	nv10ExtendPrimitive(nmesa, 3 * 4 * vertsize);
-
-	OUT_RINGp(v0,vertsize);
-	OUT_RINGp(v1,vertsize);
-	OUT_RINGp(v2,vertsize);
-}
-
-static inline void nv10_draw_line(nouveauContextPtr nmesa,
-		nouveauVertexPtr v0,
-		nouveauVertexPtr v1)
-{
-	GLuint vertsize = nmesa->vertex_size;
-	nv10ExtendPrimitive(nmesa, 2 * 4 * vertsize);
-	OUT_RINGp(v0,vertsize);
-	OUT_RINGp(v1,vertsize);
-}
-
-static inline void nv10_draw_point(nouveauContextPtr nmesa,
-		nouveauVertexPtr v0)
-{
-	GLuint vertsize = nmesa->vertex_size;
-	nv10ExtendPrimitive(nmesa, 1 * 4 * vertsize);
-	OUT_RINGp(v0,vertsize);
-}
-
 /**********************************************************************/
 /*               Render unclipped begin/end objects                   */
 /**********************************************************************/
@@ -247,7 +198,7 @@ static inline void nv10_render_generic_primitive_elts(GLcontext *ctx,GLuint star
 	nv10ExtendPrimitive(nmesa, size_dword);
 	nv10StartPrimitive(nmesa,prim+1,size_dword);
 	for (j=start; j<count; j++ ) {
-		OUT_RINGp((nouveauVertex*)(vertptr+(elt[j]*vertsize)),vertsize);
+		OUT_RINGp((nouveauVertex*)(vertptr+(elt[j]*vertsize)),vertsize/4);
 	}
 	nv10FinishPrimitive(nmesa);
 }
@@ -337,20 +288,86 @@ do {									\
    nmesa->vertex_attr_count++;						\
 } while (0)
 
+static void nv10_render_clipped_line(GLcontext *ctx,GLuint ii,GLuint jj)
+{
 
-static void nv10ChooseRenderState(GLcontext *ctx)
+}
+
+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;
+}
+
+static inline void nv10_render_points(GLcontext *ctx,GLuint first,GLuint last)
+{
+	WARN_ONCE("Unimplemented\n");
+}
+
+static inline void nv10_render_line(GLcontext *ctx,GLuint v1,GLuint v2)
+{
 	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte *vertptr = (GLubyte *)nmesa->verts;
+	GLuint vertsize = nmesa->vertex_size;
+	GLuint size_dword = vertsize*(2)/4;
 
-	nmesa->draw_point = nv10_draw_point;
-	nmesa->draw_line = nv10_draw_line;
-	nmesa->draw_tri = nv10_draw_triangle;
+	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);
+	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;
+	GLuint vertsize = nmesa->vertex_size;
+	GLuint size_dword = vertsize*(3)/4;
+
+	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);
+	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;
+	GLuint vertsize = nmesa->vertex_size;
+	GLuint size_dword = vertsize*(4)/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);
+	nv10FinishPrimitive(nmesa);
+}
+
+
+
+static void nv10ChooseRenderState(GLcontext *ctx)
+{
+	TNLcontext *tnl = TNL_CONTEXT(ctx);
+	struct nouveau_context *nmesa = NOUVEAU_CONTEXT(ctx);
 
 	tnl->Driver.Render.PrimTabVerts = nv10_render_tab_verts;
 	tnl->Driver.Render.PrimTabElts = nv10_render_tab_elts;
-	tnl->Driver.Render.ClippedLine = NULL;
-	tnl->Driver.Render.ClippedPolygon = NULL;
+	tnl->Driver.Render.ClippedLine = nv10_render_clipped_line;
+	tnl->Driver.Render.ClippedPolygon = nv10_render_clipped_poly;
+	tnl->Driver.Render.Points = nv10_render_points;
+	tnl->Driver.Render.Line = nv10_render_line;
+	tnl->Driver.Render.Triangle = nv10_render_triangle;
+	tnl->Driver.Render.Quad = nv10_render_quad;
 }
 
 
@@ -376,6 +393,7 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	   0.0, 0.0, 0.0, 1.0
 	};
 
+	nmesa->vertex_attr_count = 0;
 	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
 
 	/*
@@ -509,14 +527,21 @@ static void nv10ChooseVertexState( GLcontext *ctx )
 		nv10OutputVertexFormat(nmesa);
 	}
 
+	if (nmesa->screen->card->type == NV_30) {
+		nouveauShader *fp;
+		
+		if (ctx->FragmentProgram.Enabled) {
+			fp = (nouveauShader *) ctx->FragmentProgram.Current;
+			nvsUpdateShader(ctx, fp);
+		} else
+			nvsUpdateShader(ctx, nmesa->passthrough_fp);
+	}
+
 	if (nmesa->screen->card->type >= NV_40) {
 		/* Ensure passthrough shader is being used, and mvp matrix
 		 * is up to date
 		 */
 		nvsUpdateShader(ctx, nmesa->passthrough_vp);
-		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_IN_REG, 2);
-		OUT_RING_CACHE  (0xff09);   /*IN : POS, COL, TC0-7 */
-		OUT_RING_CACHE  (0x3fc001); /*OUT: COL, TC0-7, POS implied */
 
 		/* Update texenv shader / user fragprog */
 		nvsUpdateShader(ctx, (nouveauShader*)ctx->FragmentProgram._Current);
@@ -538,11 +563,6 @@ static void nv10RenderStart(GLcontext *ctx)
 		nmesa->new_render_state |= nmesa->new_state;
 	}
 
-	if (nmesa->Fallback) {
-		tnl->Driver.Render.Start(ctx);
-		return;
-	}
-
 	if (nmesa->new_render_state) {
 		nv10ChooseVertexState(ctx);
 		nv10ChooseRenderState(ctx);
diff --git a/src/mesa/drivers/dri/nouveau/nv20_state.c b/src/mesa/drivers/dri/nouveau/nv20_state.c
index 8e38d6eba00..030713c0dbb 100644
--- a/src/mesa/drivers/dri/nouveau/nv20_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv20_state.c
@@ -41,8 +41,8 @@ static void nv20AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
 	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
 
 	BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC, 2);
-	OUT_RING_CACHE(func);     /* NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_FUNC */
-	OUT_RING_CACHE(ubRef);    /* NV20_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF  */
+	OUT_RING_CACHE(func);
+	OUT_RING_CACHE(ubRef);
 }
 
 static void nv20BlendColor(GLcontext *ctx, const GLfloat color[4])
@@ -76,6 +76,11 @@ static void nv20BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfac
 	OUT_RING_CACHE((dfactorA<<16) | dfactorRGB);
 }
 
+static void nv20Clear(GLcontext *ctx, GLbitfield mask)
+{
+	/* TODO */
+}
+
 static void nv20ClearColor(GLcontext *ctx, const GLfloat color[4])
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
@@ -297,7 +302,11 @@ static void nv20Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 //		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
 //		case GL_POST_CONVOLUTION_COLOR_TABLE:
 //		case GL_RESCALE_NORMAL:
-//		case GL_SCISSOR_TEST:
+		case GL_SCISSOR_TEST:
+			/* No enable bit, nv20Scissor will adjust to max range */
+			ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			      			 ctx->Scissor.Width, ctx->Scissor.Height);
+			break;
 //		case GL_SEPARABLE_2D:
 		case GL_STENCIL_TEST:
 			// TODO BACK and FRONT ?
@@ -511,9 +520,22 @@ static void nv20PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
 }
 
 /** Set the scale and units used to calculate depth values */
-void (*PolygonOffset)(GLcontext *ctx, GLfloat factor, GLfloat units);
+static void nv20PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 2);
+        OUT_RING_CACHEf(factor);
+        OUT_RING_CACHEf(units);
+}
+
 /** Set the polygon stippling pattern */
-void (*PolygonStipple)(GLcontext *ctx, const GLubyte *mask );
+static void nv20PolygonStipple(GLcontext *ctx, const GLubyte *mask )
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(0), 32);
+        OUT_RING_CACHEp(mask, 32);
+}
+
 /* Specifies the current buffer for reading */
 void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
 /** Set rasterization mode */
@@ -522,6 +544,22 @@ void (*RenderMode)(GLcontext *ctx, GLenum mode );
 /** Define the scissor box */
 static void nv20Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
 {
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	/* There's no scissor enable bit, so adjust the scissor to cover the
+	 * maximum draw buffer bounds
+	 */
+	if (!ctx->Scissor.Enabled) {
+	   x = y = 0;
+	   w = h = 4095;
+	} else {
+	   x += nmesa->drawX;
+	   y += nmesa->drawY;
+	}
+
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SCISSOR_X2_X1, 2);
+        OUT_RING_CACHE(((x+w-1) << 16) | x);
+        OUT_RING_CACHE(((y+h-1) << 16) | y);
 }
 
 /** Select flat or smooth shading */
@@ -576,22 +614,103 @@ void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
 void (*TexParameter)(GLcontext *ctx, GLenum target,
 		struct gl_texture_object *texObj,
 		GLenum pname, const GLfloat *params);
-void (*TextureMatrix)(GLcontext *ctx, GLuint unit, const GLmatrix *mat);
 
-/** Set the viewport */
-static void nv20Viewport(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+static void nv20TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
 {
-    /* TODO: Where do the VIEWPORT_XFRM_* regs come in? */
-    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-    BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
-    OUT_RING_CACHE((w << 16) | x);
-    OUT_RING_CACHE((h << 16) | y);
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
+        /*XXX: This SHOULD work.*/
+        OUT_RING_CACHEp(mat->m, 16);
+}
+
+/* Update anything that depends on the window position/size */
+static void nv20WindowMoved(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;
+
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 2);
+        OUT_RING_CACHE((w << 16) | x);
+        OUT_RING_CACHE((h << 16) | y);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x02b4, 1);
+	OUT_RING(0);
+
+	BEGIN_RING_CACHE(NvSub3D,
+	      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 1);
+        OUT_RING_CACHE((4095 << 16) | 0);
+	BEGIN_RING_CACHE(NvSub3D,
+	      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(0), 1);
+        OUT_RING_CACHE((4095 << 16) | 0);
+	for (i=1; i<8; i++) {
+		BEGIN_RING_CACHE(NvSub3D,
+		      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 1);
+        	OUT_RING_CACHE(0);
+		BEGIN_RING_CACHE(NvSub3D,
+		      NV20_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_VERT(i), 1);
+	        OUT_RING_CACHE(0);
+	}
+
+	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+	      		    ctx->Scissor.Width, ctx->Scissor.Height);
+
+	/* TODO: recalc viewport scale coefs */
 }
 
 /* Initialise any card-specific non-GL related state */
 static GLboolean nv20InitCard(nouveauContextPtr nmesa)
 {
-   return GL_TRUE;
+	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
+
+	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT1, 2);
+	OUT_RING(NvDmaFB);	/* 184 dma_object1 */
+	OUT_RING(NvDmaFB);	/* 188 dma_object2 */
+	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT3, 2);
+	OUT_RING(NvDmaFB);	/* 194 dma_object3 */
+	OUT_RING(NvDmaFB);	/* 198 dma_object4 */
+	BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
+	OUT_RING(NvDmaFB);	/* 1a8 dma_object8 */
+
+	BEGIN_RING_SIZE(NvSub3D, 0x17e0, 3);
+	OUT_RINGf(0.0);
+	OUT_RINGf(0.0);
+	OUT_RINGf(1.0);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x1e6c, 1);
+	OUT_RING(0x0db6);
+	BEGIN_RING_SIZE(NvSub3D, 0x0290, 1);
+	OUT_RING(0x00100001);
+	BEGIN_RING_SIZE(NvSub3D, 0x09fc, 1);
+	OUT_RING(0);
+	BEGIN_RING_SIZE(NvSub3D, 0x1d80, 1);
+	OUT_RING(1);
+	BEGIN_RING_SIZE(NvSub3D, 0x09f8, 1);
+	OUT_RING(4);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x17ec, 3);
+	OUT_RINGf(0.0);
+	OUT_RINGf(1.0);
+	OUT_RINGf(0.0);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x1d88, 1);
+	OUT_RING(3);
+
+	/* FIXME: More dma objects to setup ? */
+
+	BEGIN_RING_SIZE(NvSub3D, 0x1e98, 1);
+	OUT_RING(0);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
+	OUT_RING(0);
+	OUT_RING(1);
+	OUT_RING(2);
+
+	return GL_TRUE;
 }
 
 /* Update buffer offset/pitch/format */
@@ -599,26 +718,57 @@ static GLboolean nv20BindBuffers(nouveauContextPtr nmesa, int num_color,
 				 nouveau_renderbuffer **color,
 				 nouveau_renderbuffer *depth)
 {
-   return GL_TRUE;
-}
+	GLuint x, y, w, h;
+	GLuint pitch, format, depth_pitch;
+
+	w = color[0]->mesa.Width;
+	h = color[0]->mesa.Height;
+	x = nmesa->drawX;
+	y = nmesa->drawY;
+
+	if (num_color != 1)
+		return GL_FALSE;
+
+        BEGIN_RING_CACHE(NvSub3D, NV20_TCL_PRIMITIVE_3D_VIEWPORT_HORIZ, 6);
+        OUT_RING_CACHE((w << 16) | x);
+        OUT_RING_CACHE((h << 16) | y);
+	depth_pitch = (depth ? depth->pitch : color[0]->pitch);
+	pitch = (depth_pitch<<16) | color[0]->pitch;
+	format = 0x128;
+	if (color[0]->mesa._ActualFormat != GL_RGBA8) {
+		format = 0x123; /* 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);
+
+	if (depth) {
+		BEGIN_RING_SIZE(NvSub3D, NV20_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH, 2);
+		/* TODO: use a different buffer */
+		OUT_RING(depth->pitch);
+		OUT_RING(depth->offset);
+	}
 
-/* Update anything that depends on the window position/size */
-static void nv20WindowMoved(nouveauContextPtr nmesa)
-{
+	/* Always set to bottom left of buffer */
+	BEGIN_RING_CACHE(NvSub3D, NV20_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);
+
+	return GL_TRUE;
 }
 
 void nv20InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 
-	nmesa->hw_func.InitCard		= nv20InitCard;
-	nmesa->hw_func.BindBuffers	= nv20BindBuffers;
-	nmesa->hw_func.WindowMoved	= nv20WindowMoved;
-
 	func->AlphaFunc			= nv20AlphaFunc;
 	func->BlendColor		= nv20BlendColor;
 	func->BlendEquationSeparate	= nv20BlendEquationSeparate;
 	func->BlendFuncSeparate		= nv20BlendFuncSeparate;
+	func->Clear			= nv20Clear;
 	func->ClearColor		= nv20ClearColor;
 	func->ClearDepth		= nv20ClearDepth;
 	func->ClearStencil		= nv20ClearStencil;
@@ -641,22 +791,21 @@ void nv20InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 	func->PointParameterfv		= nv20PointParameterfv;
 	func->PointSize			= nv20PointSize;
 	func->PolygonMode		= nv20PolygonMode;
-#if 0
 	func->PolygonOffset		= nv20PolygonOffset;
 	func->PolygonStipple		= nv20PolygonStipple;
-	func->ReadBuffer		= nv20ReadBuffer;
-	func->RenderMode		= nv20RenderMode;
-#endif
+/*	func->ReadBuffer		= nv20ReadBuffer;*/
+/*	func->RenderMode		= nv20RenderMode;*/
 	func->Scissor			= nv20Scissor;
 	func->ShadeModel		= nv20ShadeModel;
 	func->StencilFuncSeparate	= nv20StencilFuncSeparate;
 	func->StencilMaskSeparate	= nv20StencilMaskSeparate;
 	func->StencilOpSeparate		= nv20StencilOpSeparate;
-#if 0
-	func->TexGen			= nv20TexGen;
-	func->TexParameter		= nv20TexParameter;
+/*	func->TexGen			= nv20TexGen;*/
+/*	func->TexParameter		= nv20TexParameter;*/
 	func->TextureMatrix		= nv20TextureMatrix;
-#endif
-	func->Viewport			= nv20Viewport;
+
+	nmesa->hw_func.InitCard		= nv20InitCard;
+	nmesa->hw_func.BindBuffers	= nv20BindBuffers;
+	nmesa->hw_func.WindowMoved	= nv20WindowMoved;
 }
 
diff --git a/src/mesa/drivers/dri/nouveau/nv30_fragprog.c b/src/mesa/drivers/dri/nouveau/nv30_fragprog.c
index cd7c955c9ee..e32452361e8 100644
--- a/src/mesa/drivers/dri/nouveau/nv30_fragprog.c
+++ b/src/mesa/drivers/dri/nouveau/nv30_fragprog.c
@@ -24,6 +24,7 @@ static void
 NV30FPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
 {
    nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+   nvsCardPriv *priv = &nvs->card_priv;
    uint32_t offset;
 
    if (!nvs->program_buffer)
@@ -31,14 +32,13 @@ NV30FPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
 							GL_ARRAY_BUFFER_ARB);
 
    /* Should use STATIC_DRAW_ARB if shader doesn't use changable params */
-   ctx->Driver.BufferData(ctx, GL_ARRAY_BUFFER_ARB,
+   nouveau_bo_init_storage(ctx, NOUVEAU_BO_VRAM_OK,
 	 		  nvs->program_size * sizeof(uint32_t),
 			  (const GLvoid *)nvs->program,
 			  GL_DYNAMIC_DRAW_ARB,
 			  nvs->program_buffer);
 
-   offset = nouveau_bufferobj_gpu_ref(ctx, GL_READ_ONLY_ARB,
-	 			      nvs->program_buffer);
+   offset = nouveau_bo_gpu_ref(ctx, nvs->program_buffer);
 
    /* Not using state cache here, updated programs at the same address don't
     * seem to take effect unless the ACTIVE_PROGRAM method is called again.
@@ -46,6 +46,18 @@ NV30FPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
     */
    BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FP_ACTIVE_PROGRAM, 1);
    OUT_RING       (offset | 1);
+   if (nmesa->screen->card->type == NV_30) {  
+	   BEGIN_RING_SIZE(NvSub3D,
+			   0x1d60 /*NV30_TCL_PRIMITIVE_3D_FP_CONTROL*/, 1);
+	   OUT_RING       ((priv->NV30FP.uses_kil << 7));
+	   BEGIN_RING_SIZE(NvSub3D, 0x1450, 1);
+	   OUT_RING       (priv->NV30FP.num_regs << 16);
+   } else {
+	   BEGIN_RING_SIZE(NvSub3D,
+			   0x1d60 /*NV30_TCL_PRIMITIVE_3D_FP_CONTROL*/, 1);
+	   OUT_RING       ((priv->NV30FP.uses_kil <<  7) |
+			   (priv->NV30FP.num_regs << 24));
+   }
 }
 
 static void
@@ -92,6 +104,8 @@ NV30FPSupportsOpcode(nvsFunc *shader, nvsOpcode op)
 static void
 NV30FPSetOpcode(nvsFunc *shader, unsigned int opcode, int slot)
 {
+   if (opcode == NV30_FP_OP_OPCODE_KIL)
+      shader->card_priv->NV30FP.uses_kil = GL_TRUE;
    shader->inst[0] &= ~NV30_FP_OP_OPCODE_MASK;
    shader->inst[0] |= (opcode << NV30_FP_OP_OPCODE_SHIFT);
 }
@@ -142,6 +156,16 @@ NV30FPSetCondition(nvsFunc *shader, int on, nvsCond cond, int reg,
 }
 
 static void
+NV30FPSetHighReg(nvsFunc *shader, int id)
+{
+   if (shader->card_priv->NV30FP.num_regs < (id+1)) {
+      if (id == 0)
+	 id = 1; /* necessary? */
+      shader->card_priv->NV30FP.num_regs = (id+1);
+   }
+}
+
+static void
 NV30FPSetResult(nvsFunc *shader, nvsRegister *reg, unsigned int mask, int slot)
 {
    unsigned int hwreg;
@@ -159,6 +183,7 @@ NV30FPSetResult(nvsFunc *shader, nvsRegister *reg, unsigned int mask, int slot)
       shader->inst[0] &= ~NV30_FP_OP_UNK0_7;
       hwreg = reg->index;
    }
+   NV30FPSetHighReg(shader, hwreg);
    shader->inst[0] &= ~NV30_FP_OP_OUT_REG_SHIFT;
    shader->inst[0] |= (hwreg  << NV30_FP_OP_OUT_REG_SHIFT);
 }
@@ -172,6 +197,7 @@ NV30FPSetSource(nvsFunc *shader, nvsRegister *reg, int pos)
    case NVS_FILE_TEMP:
       hwsrc |= (NV30_FP_REG_TYPE_TEMP << NV30_FP_REG_TYPE_SHIFT);
       hwsrc |= (reg->index << NV30_FP_REG_SRC_SHIFT);
+      NV30FPSetHighReg(shader, reg->index);
       break;
    case NVS_FILE_ATTRIB:
       {
diff --git a/src/mesa/drivers/dri/nouveau/nv30_state.c b/src/mesa/drivers/dri/nouveau/nv30_state.c
index 9bf5f2adeab..ad21fa27302 100644
--- a/src/mesa/drivers/dri/nouveau/nv30_state.c
+++ b/src/mesa/drivers/dri/nouveau/nv30_state.c
@@ -127,6 +127,11 @@ static void nv30ClearStencil(GLcontext *ctx, GLint s)
 static void nv30ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
 {
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (NOUVEAU_CARD_USING_SHADERS)
+		return;
+
+	plane -= GL_CLIP_PLANE0;
 	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_A(plane), 4);
 	OUT_RING_CACHEf(equation[0]);
 	OUT_RING_CACHEf(equation[1]);
@@ -208,8 +213,14 @@ static void nv30Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 		case GL_CLIP_PLANE3:
 		case GL_CLIP_PLANE4:
 		case GL_CLIP_PLANE5:
-			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(cap-GL_CLIP_PLANE0), 1);
-			OUT_RING_CACHE(state);
+			if (NOUVEAU_CARD_USING_SHADERS) {
+				nouveauShader *nvs = (nouveauShader *)ctx->VertexProgram._Current;
+				if (nvs)
+					nvs->translated = GL_FALSE;
+			} else {
+				BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_CLIP_PLANE_ENABLE(cap-GL_CLIP_PLANE0), 1);
+				OUT_RING_CACHE(state);
+			}
 			break;
 		case GL_COLOR_LOGIC_OP:
 			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_COLOR_LOGIC_OP_ENABLE, 1);
@@ -233,6 +244,8 @@ static void nv30Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 			OUT_RING_CACHE(state);
 			break;
 		case GL_FOG:
+			if (NOUVEAU_CARD_USING_SHADERS)
+				break;
 			BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_ENABLE, 1);
 			OUT_RING_CACHE(state);
 			break;
@@ -348,17 +361,71 @@ static void nv30Enable(GLcontext *ctx, GLenum cap, GLboolean state)
 static void nv30Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
 {
     nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+    if (NOUVEAU_CARD_USING_SHADERS)
+        return;
+
     switch(pname)
     {
-        case GL_FOG_MODE:
-            //BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_MODE, 1);
-            //OUT_RING_CACHE (params);
+    case GL_FOG_MODE:
+    {
+        int mode = 0;
+        /* The modes are different in GL and the card.  */
+        switch(ctx->Fog.Mode)
+        {
+        case GL_LINEAR:
+            mode = 0x804;
             break;
-            /* TODO: unsure about the rest.*/
-        default:
+        case GL_EXP:
+            mode = 0x802;
             break;
+        case GL_EXP2:
+            mode = 0x803;
+            break;
+        }
+	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_MODE, 1);
+	OUT_RING_CACHE (mode);
+	break;
+    }
+    case GL_FOG_COLOR:
+    {
+	GLubyte c[4];
+	UNCLAMPED_FLOAT_TO_RGBA_CHAN(c,params);
+        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_COLOR, 1);
+        /* nvidia ignores the alpha channel */
+	OUT_RING_CACHE(PACK_COLOR_8888_REV(c[0],c[1],c[2],c[3]));
+        break;
+    }
+    case GL_FOG_DENSITY:
+    case GL_FOG_START:
+    case GL_FOG_END:
+    {
+        GLfloat f=0., c=0.;
+        switch(ctx->Fog.Mode)
+        {
+        case GL_LINEAR:
+            f = -1.0/(ctx->Fog.End - ctx->Fog.Start);
+            c = ctx->Fog.Start/(ctx->Fog.End - ctx->Fog.Start) + 2.001953;
+            break;
+        case GL_EXP:
+            f = -0.090168*ctx->Fog.Density;
+            c = 1.5;
+        case GL_EXP2:
+            f = -0.212330*ctx->Fog.Density;
+            c = 1.5;
+        }
+        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_LINEAR, 1);
+        OUT_RING_CACHE(f);
+        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_CONSTANT, 1);
+        OUT_RING_CACHE(c);
+        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_FOG_EQUATION_QUADRATIC, 1);
+        OUT_RING_CACHE(0); /* Is this always the same? */
+        break;
+    }
+//    case GL_FOG_COORD_SRC:
+    default:
+        break;
     }
-
 }
    
 static void nv30Hint(GLcontext *ctx, GLenum target, GLenum mode)
@@ -505,7 +572,7 @@ static void nv30LineWidth(GLcontext *ctx, GLfloat width)
 	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
 	GLubyte ubWidth;
 
-	CLAMPED_FLOAT_TO_UBYTE(ubWidth, width);
+	ubWidth = (GLubyte)(width * 8.0) & 0xFF;
 
 	BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LINE_WIDTH_SMOOTH, 1);
 	OUT_RING_CACHE(ubWidth);
@@ -671,9 +738,13 @@ void (*TexParameter)(GLcontext *ctx, GLenum target,
 static void nv30TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
 {
         nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
-        BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
-        /*XXX: This SHOULD work.*/
-        OUT_RING_CACHEp(mat->m, 16);
+
+	if (!NOUVEAU_CARD_USING_SHADERS) {
+		BEGIN_RING_CACHE(NvSub3D,
+				 NV30_TCL_PRIMITIVE_3D_TX_MATRIX(unit, 0), 16);
+		/*XXX: This SHOULD work.*/
+		OUT_RING_CACHEp(mat->m, 16);
+	}
 }
 
 static void nv30WindowMoved(nouveauContextPtr nmesa)
@@ -710,8 +781,67 @@ static void nv30WindowMoved(nouveauContextPtr nmesa)
 
 static GLboolean nv30InitCard(nouveauContextPtr nmesa)
 {
-	/* Need some love.. */
-	return GL_FALSE;
+	int i;
+	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
+
+	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT1, 3);
+	OUT_RING(NvDmaFB);
+	OUT_RING(NvDmaAGP);
+        OUT_RING(NvDmaFB);
+	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT8, 1);
+	OUT_RING(NvDmaFB);
+	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_OBJECT4, 2);
+	OUT_RING(NvDmaFB);
+	OUT_RING(NvDmaFB);
+        BEGIN_RING_SIZE(NvSub3D, 0x1b0, 1); /* SET_OBJECT8B*/
+        OUT_RING(NvDmaFB);
+
+        for(i = 0x2c8; i <= 0x2fc; i += 4)
+        {
+            BEGIN_RING_SIZE(NvSub3D, i, 1);
+            OUT_RING(0x0);
+        }
+
+	BEGIN_RING_SIZE(NvSub3D, 0x0220, 1);
+	OUT_RING(1);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x03b0, 1);
+	OUT_RING(0x00100000);
+	BEGIN_RING_SIZE(NvSub3D, 0x1454, 1);
+	OUT_RING(0);
+	BEGIN_RING_SIZE(NvSub3D, 0x1d80, 1);
+	OUT_RING(3);
+	
+	/* NEW */
+       	BEGIN_RING_SIZE(NvSub3D, 0x1e98, 1);
+        OUT_RING(0);
+        BEGIN_RING_SIZE(NvSub3D, 0x17e0, 3);
+        OUT_RING(0);
+        OUT_RING(0);
+        OUT_RING(0x3f800000);
+        BEGIN_RING_SIZE(NvSub3D, 0x1f80, 16);
+        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
+        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
+        OUT_RING(0x0000ffff);
+        OUT_RING(0); OUT_RING(0); OUT_RING(0); OUT_RING(0); 
+        OUT_RING(0); OUT_RING(0); OUT_RING(0); 
+/*
+        BEGIN_RING_SIZE(NvSub3D, 0x100, 2);
+        OUT_RING(0);
+        OUT_RING(0);
+*/
+        BEGIN_RING_SIZE(NvSub3D, 0x120, 3);
+        OUT_RING(0);
+        OUT_RING(1);
+        OUT_RING(2);
+
+        BEGIN_RING_SIZE(NvSub3D, 0x1d88, 1);
+        OUT_RING(0x00001200);
+
+	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_RC_ENABLE, 1);
+	OUT_RING       (0);
+
+	return GL_TRUE;
 }
 
 static GLboolean nv40InitCard(nouveauContextPtr nmesa)
@@ -750,43 +880,45 @@ static GLboolean nv40InitCard(nouveauContextPtr nmesa)
 	BEGIN_RING_SIZE(NvSub3D, 0x1e94, 1);
 	OUT_RING(0x00000001);
 
-	BEGIN_RING_SIZE(NvSub3D, 0x1d60, 1);
-	OUT_RING(0x03008000);
-
 	return GL_TRUE;
 }
 
 static GLboolean nv30BindBuffers(nouveauContextPtr nmesa, int num_color,
-      				 nouveau_renderbuffer **color,
-				 nouveau_renderbuffer *depth)
-{
-   GLuint x, y, w, h;
-
-   w = color[0]->mesa.Width;
-   h = color[0]->mesa.Height;
-   x = nmesa->drawX;
-   y = nmesa->drawY;
-
-   if (num_color != 1)
-      return GL_FALSE;
-   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0, 5);
-   OUT_RING        (((w+x)<<16)|x);
-   OUT_RING        (((h+y)<<16)|y);
-   if (color[0]->mesa._ActualFormat == GL_RGBA8)
-      OUT_RING        (0x148);
-   else
-      OUT_RING        (0x143);
-   OUT_RING        (color[0]->pitch);
-   OUT_RING        (color[0]->offset);
-
-   if (depth) {
-      BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET, 1);
-      OUT_RING        (depth->offset);
-      BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH, 1);
-      OUT_RING        (depth->pitch);
-   }
-
-   return GL_TRUE;
+		nouveau_renderbuffer **color,
+		nouveau_renderbuffer *depth)
+{
+	GLuint x, y, w, h;
+
+	w = color[0]->mesa.Width;
+	h = color[0]->mesa.Height;
+	x = nmesa->drawX;
+	y = nmesa->drawY;
+
+	if (num_color != 1)
+		return GL_FALSE;
+	BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VIEWPORT_COLOR_BUFFER_DIM0, 5);
+	OUT_RING        (((w+x)<<16)|x);
+	OUT_RING        (((h+y)<<16)|y);
+	if (color[0]->mesa._ActualFormat == GL_RGBA8)
+		OUT_RING        (0x148);
+	else
+		OUT_RING        (0x143);
+	if (nmesa->screen->card->type >= NV_40)
+		OUT_RING        (color[0]->pitch);
+	else
+		OUT_RING        (color[0]->pitch | (depth ? (depth->pitch << 16): 0));
+	OUT_RING        (color[0]->offset);
+
+	if (depth) {
+		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DEPTH_OFFSET, 1);
+		OUT_RING        (depth->offset);
+		if (nmesa->screen->card->type >= NV_40) {
+			BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_LMA_DEPTH_BUFFER_PITCH, 1);
+			OUT_RING        (depth->pitch);
+		}
+	}
+
+	return GL_TRUE;
 }
 
 void nv30InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
@@ -808,6 +940,7 @@ void nv30InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
 	func->FrontFace			= nv30FrontFace;
 	func->DepthFunc			= nv30DepthFunc;
 	func->DepthMask			= nv30DepthMask;
+	func->DepthRange                = nv30DepthRange;
 	func->Enable			= nv30Enable;
 	func->Fogfv			= nv30Fogfv;
 	func->Hint			= nv30Hint;
diff --git a/src/mesa/drivers/dri/nouveau/nv30_vertprog.c b/src/mesa/drivers/dri/nouveau/nv30_vertprog.c
index 6ba8e35d55f..d023e8439e3 100644
--- a/src/mesa/drivers/dri/nouveau/nv30_vertprog.c
+++ b/src/mesa/drivers/dri/nouveau/nv30_vertprog.c
@@ -29,6 +29,13 @@ NV30VPUploadToHW(GLcontext *ctx, nouveauShader *nvs)
    }
    BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_PROGRAM_START_ID, 1);
    OUT_RING(0);
+
+   BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_IN_REG, 2);
+   OUT_RING(nvs->card_priv.NV30VP.vp_in_reg);
+   OUT_RING(nvs->card_priv.NV30VP.vp_out_reg);
+
+   BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_SET_CLIPPING_PLANES, 1);
+   OUT_RING_CACHE  (nvs->card_priv.NV30VP.clip_enables);
 }
 
 static void
@@ -48,6 +55,12 @@ NV30VPUpdateConst(GLcontext *ctx, nouveauShader *nvs, int id)
 /*****************************************************************************
  * Assembly routines
  */
+static void
+NV30VPSetBranchTarget(nvsFunc *shader, int addr)
+{
+	shader->inst[2] &= ~NV30_VP_INST_IADDR_MASK;
+	shader->inst[2] |= (addr << NV30_VP_INST_IADDR_SHIFT);
+}
 
 /*****************************************************************************
  * Disassembly routines
@@ -349,5 +362,6 @@ NV30VPInitShaderFuncs(nvsFunc * shader)
    shader->GetCondRegID		= NV30VPGetCondRegID;
 
    shader->GetBranch		= NV30VPGetBranch;
+   shader->SetBranchTarget	= NV30VPSetBranchTarget;
 }
 
diff --git a/src/mesa/drivers/dri/nouveau/nv40_fragprog.c b/src/mesa/drivers/dri/nouveau/nv40_fragprog.c
index 3d58d6b6667..3e4ae0496e4 100644
--- a/src/mesa/drivers/dri/nouveau/nv40_fragprog.c
+++ b/src/mesa/drivers/dri/nouveau/nv40_fragprog.c
@@ -5,6 +5,71 @@
 unsigned int NVFP_TX_BOP_COUNT = 5;
 struct _op_xlat NVFP_TX_BOP[64];
 
+
+/*****************************************************************************
+ * Assembly routines
+ * 	- These extend the NV30 routines, which are almost identical.  NV40
+ * 	  just has branching hacked into the instruction set.
+ */
+static int
+NV40FPSupportsResultScale(nvsFunc *shader, nvsScale scale)
+{
+	switch (scale) {
+	case NVS_SCALE_1X:
+	case NVS_SCALE_2X:
+	case NVS_SCALE_4X:
+	case NVS_SCALE_8X:
+	case NVS_SCALE_INV_2X:
+	case NVS_SCALE_INV_4X:
+	case NVS_SCALE_INV_8X:
+		return 1;
+	default:
+		return 0;
+	}
+}
+
+static void
+NV40FPSetResultScale(nvsFunc *shader, nvsScale scale)
+{
+	shader->inst[2] &= ~NV40_FP_OP_DST_SCALE_MASK;
+	shader->inst[2] |= ((unsigned int)scale << NV40_FP_OP_DST_SCALE_SHIFT);
+}
+
+static void
+NV40FPSetBranchTarget(nvsFunc *shader, int addr)
+{
+	shader->inst[2] &= ~NV40_FP_OP_IADDR_MASK;
+	shader->inst[2] |= (addr << NV40_FP_OP_IADDR_SHIFT);
+}
+
+static void
+NV40FPSetBranchElse(nvsFunc *shader, int addr)
+{
+	shader->inst[2] &= ~NV40_FP_OP_ELSE_ID_MASK;
+	shader->inst[2] |= (addr << NV40_FP_OP_ELSE_ID_SHIFT);
+}
+
+static void
+NV40FPSetBranchEnd(nvsFunc *shader, int addr)
+{
+	shader->inst[3] &= ~NV40_FP_OP_END_ID_MASK;
+	shader->inst[3] |= (addr << NV40_FP_OP_END_ID_SHIFT);
+}
+
+static void
+NV40FPSetLoopParams(nvsFunc *shader, int count, int initial, int increment)
+{
+	shader->inst[2] &= ~(NV40_FP_OP_LOOP_COUNT_MASK |
+			     NV40_FP_OP_LOOP_INDEX_MASK |
+			     NV40_FP_OP_LOOP_INCR_MASK);
+	shader->inst[2] |= ((count     << NV40_FP_OP_LOOP_COUNT_SHIFT) |
+			    (initial   << NV40_FP_OP_LOOP_INDEX_SHIFT) |
+			    (increment << NV40_FP_OP_LOOP_INCR_SHIFT));
+}
+
+/*****************************************************************************
+ * Disassembly routines
+ */
 static struct _op_xlat *
 NV40FPGetOPTXRec(nvsFunc * shader, int merged)
 {
@@ -138,6 +203,9 @@ NV40FPInitShaderFuncs(nvsFunc * shader)
    MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_REP , NVS_OP_REP , -1, -1, -1);
    MOD_OPCODE(NVFP_TX_BOP, NV40_FP_OP_BRA_OPCODE_RET , NVS_OP_RET , -1, -1, -1);
 
+   shader->SupportsResultScale	= NV40FPSupportsResultScale;
+   shader->SetResultScale	= NV40FPSetResultScale;
+
    /* fragment.facing */
    shader->GetSourceID		= NV40FPGetSourceID;
 
@@ -149,4 +217,8 @@ NV40FPInitShaderFuncs(nvsFunc * shader)
    shader->GetLoopCount		= NV40FPGetLoopCount;
    shader->GetLoopInitial	= NV40FPGetLoopInitial;
    shader->GetLoopIncrement	= NV40FPGetLoopIncrement;
+   shader->SetBranchTarget	= NV40FPSetBranchTarget;
+   shader->SetBranchElse	= NV40FPSetBranchElse;
+   shader->SetBranchEnd		= NV40FPSetBranchEnd;
+   shader->SetLoopParams	= NV40FPSetLoopParams;
 }
diff --git a/src/mesa/drivers/dri/nouveau/nv40_shader.h b/src/mesa/drivers/dri/nouveau/nv40_shader.h
index 2a2b5639b6c..584f4c23e08 100644
--- a/src/mesa/drivers/dri/nouveau/nv40_shader.h
+++ b/src/mesa/drivers/dri/nouveau/nv40_shader.h
@@ -399,8 +399,8 @@
 
 /* high order bits of SRC1 */
 #define NV40_FP_OP_OPCODE_IS_BRANCH                                      (1<<31)
-#define NV40_FP_OP_SRC_SCALE_SHIFT                                            28
-#define NV40_FP_OP_SRC_SCALE_MASK                                      (3 << 28)
+#define NV40_FP_OP_DST_SCALE_SHIFT                                            28
+#define NV40_FP_OP_DST_SCALE_MASK                                      (3 << 28)
 
 /* SRC1 LOOP */
 #define NV40_FP_OP_LOOP_INCR_SHIFT                                            19
diff --git a/src/mesa/drivers/dri/nouveau/nv40_vertprog.c b/src/mesa/drivers/dri/nouveau/nv40_vertprog.c
index 0493e184033..d054140bcd0 100644
--- a/src/mesa/drivers/dri/nouveau/nv40_vertprog.c
+++ b/src/mesa/drivers/dri/nouveau/nv40_vertprog.c
@@ -66,6 +66,103 @@ NV40VPSetCondition(nvsFunc *shader, int on, nvsCond cond, int reg,
    shader->inst[0] |= (swizzle[NVS_SWZ_W] << NV40_VP_INST_COND_SWZ_W_SHIFT);
 }
 
+/* these just exist here until nouveau_reg.h has them. */
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL0	(1<<0)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1	(1<<1)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0	(1<<2)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1	(1<<3)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC	(1<<4)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ	(1<<5)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0	(1<<6)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1	(1<<7)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2	(1<<8)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3	(1<<9)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4	(1<<10)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5	(1<<11)
+#define NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0   (1<<14)
+
+static unsigned int
+NV40VPTranslateResultReg(nvsFunc *shader, nvsFixedReg result,
+					  unsigned int *mask_ret)
+{
+	unsigned int *out_reg = &shader->card_priv->NV30VP.vp_out_reg;
+	unsigned int *clip_en = &shader->card_priv->NV30VP.clip_enables;
+
+	*mask_ret = 0xf;
+
+	switch (result) {
+	case NVS_FR_POSITION:
+		/* out_reg POS implied */
+		return NV40_VP_INST_DEST_POS;
+	case NVS_FR_COL0:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL0;
+		return NV40_VP_INST_DEST_COL0;
+	case NVS_FR_COL1:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_COL1;
+		return NV40_VP_INST_DEST_COL1;
+	case NVS_FR_BFC0:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC0;
+		return NV40_VP_INST_DEST_BFC0;
+	case NVS_FR_BFC1:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_BFC1;
+		return NV40_VP_INST_DEST_BFC1;
+	case NVS_FR_FOGCOORD:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_FOGC;
+		*mask_ret = 0x8;
+		return NV40_VP_INST_DEST_FOGC;
+	case NVS_FR_CLIP0:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP0;
+		(*clip_en) |= 0x00000002;
+		*mask_ret = 0x4;
+		return NV40_VP_INST_DEST_FOGC;
+	case NVS_FR_CLIP1:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP1;
+		(*clip_en) |= 0x00000020;
+		*mask_ret = 0x2;
+		return NV40_VP_INST_DEST_FOGC;
+	case NVS_FR_CLIP2:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP2;
+		(*clip_en) |= 0x00000200;
+		*mask_ret = 0x1;
+		return NV40_VP_INST_DEST_FOGC;
+	case NVS_FR_POINTSZ:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_PSZ;
+		*mask_ret = 0x8;
+		return NV40_VP_INST_DEST_PSZ;
+	case NVS_FR_CLIP3:
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP3;
+		(*clip_en) |= 0x00002000;
+		*mask_ret = 0x4;
+		return NV40_VP_INST_DEST_PSZ;
+	case NVS_FR_CLIP4:
+		(*clip_en) |= 0x00020000;
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP4;
+		*mask_ret = 0x2;
+		return NV40_VP_INST_DEST_PSZ;
+	case NVS_FR_CLIP5:
+		(*clip_en) |= 0x00200000;
+		(*out_reg) |= NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_CLP5;
+		*mask_ret = 0x1;
+		return NV40_VP_INST_DEST_PSZ;
+	case NVS_FR_TEXCOORD0:
+	case NVS_FR_TEXCOORD1:
+	case NVS_FR_TEXCOORD2:
+	case NVS_FR_TEXCOORD3:
+	case NVS_FR_TEXCOORD4:
+	case NVS_FR_TEXCOORD5:
+	case NVS_FR_TEXCOORD6:
+	case NVS_FR_TEXCOORD7:
+	{
+		int unit = result - NVS_FR_TEXCOORD0;
+		(*out_reg) |= (NV30_TCL_PRIMITIVE_3D_VP_OUT_REG_TEX0 << unit);
+		return NV40_VP_INST_DEST_TC(unit);
+	}
+	default:
+		WARN_ONCE("unknown vp output %d\n", result);
+		return NV40_VP_INST_DEST_POS;
+	}
+}
+
 static void
 NV40VPSetResult(nvsFunc *shader, nvsRegister * dest, unsigned int mask,
       		int slot)
@@ -78,29 +175,14 @@ NV40VPSetResult(nvsFunc *shader, nvsRegister * dest, unsigned int mask,
    if (mask & SMASK_W) hwmask |= (1 << 0);
 
    if (dest->file == NVS_FILE_RESULT) {
+      unsigned int valid_mask;
       int hwidx;
 
-      switch (dest->index) {
-      case NVS_FR_POSITION : hwidx = NV40_VP_INST_DEST_POS; break;
-      case NVS_FR_COL0     : hwidx = NV40_VP_INST_DEST_COL0; break;
-      case NVS_FR_COL1     : hwidx = NV40_VP_INST_DEST_COL1; break;
-      case NVS_FR_BFC0     : hwidx = NV40_VP_INST_DEST_BFC0; break;
-      case NVS_FR_BFC1     : hwidx = NV40_VP_INST_DEST_BFC1; break;
-      case NVS_FR_FOGCOORD : hwidx = NV40_VP_INST_DEST_FOGC; break;
-      case NVS_FR_POINTSZ  : hwidx = NV40_VP_INST_DEST_PSZ; break;
-      case NVS_FR_TEXCOORD0: hwidx = NV40_VP_INST_DEST_TC(0); break;
-      case NVS_FR_TEXCOORD1: hwidx = NV40_VP_INST_DEST_TC(1); break;
-      case NVS_FR_TEXCOORD2: hwidx = NV40_VP_INST_DEST_TC(2); break;
-      case NVS_FR_TEXCOORD3: hwidx = NV40_VP_INST_DEST_TC(3); break;
-      case NVS_FR_TEXCOORD4: hwidx = NV40_VP_INST_DEST_TC(4); break;
-      case NVS_FR_TEXCOORD5: hwidx = NV40_VP_INST_DEST_TC(5); break;
-      case NVS_FR_TEXCOORD6: hwidx = NV40_VP_INST_DEST_TC(6); break;
-      case NVS_FR_TEXCOORD7: hwidx = NV40_VP_INST_DEST_TC(7); break;
-      default:
-	WARN_ONCE("unknown vtxprog output %d\n", dest->index);
-	hwidx = 0;
-	break;
-      }
+      hwidx = NV40VPTranslateResultReg(shader, dest->index, &valid_mask);
+      if (hwmask & ~valid_mask)
+	      WARN_ONCE("writing invalid components of result reg\n");
+      hwmask &= valid_mask;
+
       shader->inst[3] &= ~NV40_VP_INST_DEST_MASK;
       shader->inst[3] |= (hwidx << NV40_VP_INST_DEST_SHIFT);
 
@@ -174,6 +256,7 @@ NV40VPSetSource(nvsFunc *shader, nvsRegister * src, int pos)
 
       shader->inst[1] &= ~NV40_VP_INST_INPUT_SRC_MASK;
       shader->inst[1] |= (src->index << NV40_VP_INST_INPUT_SRC_SHIFT);
+      shader->card_priv->NV30VP.vp_in_reg |= (1 << src->index);
       if (src->indexed) {
 	 shader->inst[0] |= NV40_VP_INST_INDEX_INPUT;
 	 if (src->addr_reg)
@@ -228,6 +311,15 @@ NV40VPSetSource(nvsFunc *shader, nvsRegister * src, int pos)
 }
 
 static void
+NV40VPSetBranchTarget(nvsFunc *shader, int addr)
+{
+	shader->inst[2] &= ~NV40_VP_INST_IADDRH_MASK;
+	shader->inst[2] |= ((addr & 0xf8) >> 3) << NV40_VP_INST_IADDRH_SHIFT;
+	shader->inst[3] &= ~NV40_VP_INST_IADDRL_MASK;
+	shader->inst[3] |= ((addr & 0x07) << NV40_VP_INST_IADDRL_SHIFT);
+}
+
+static void
 NV40VPInitInstruction(nvsFunc *shader)
 {
    unsigned int hwsrc = 0;
@@ -657,6 +749,7 @@ NV40VPInitShaderFuncs(nvsFunc * shader)
    shader->SetResult		= NV40VPSetResult;
    shader->SetSource		= NV40VPSetSource;
    shader->SetLastInst		= NV40VPSetLastInst;
+   shader->SetBranchTarget	= NV40VPSetBranchTarget;
 
    shader->HasMergedInst	= NV40VPHasMergedInst;
    shader->GetOpcodeHW		= NV40VPGetOpcodeHW;
diff --git a/src/mesa/drivers/dri/nouveau/nv50_state.c b/src/mesa/drivers/dri/nouveau/nv50_state.c
new file mode 100644
index 00000000000..818e268615c
--- /dev/null
+++ b/src/mesa/drivers/dri/nouveau/nv50_state.c
@@ -0,0 +1,641 @@
+/**************************************************************************
+
+Copyright 2006 Nouveau
+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
+ERIC ANHOLT OR SILICON INTEGRATED SYSTEMS CORP 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 "nouveau_context.h"
+#include "nouveau_object.h"
+#include "nouveau_fifo.h"
+#include "nouveau_reg.h"
+#include "nouveau_state.h"
+
+#include "tnl/t_pipeline.h"
+
+#include "mtypes.h"
+#include "colormac.h"
+
+static void nv50AlphaFunc(GLcontext *ctx, GLenum func, GLfloat ref)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLubyte ubRef;
+	CLAMPED_FLOAT_TO_UBYTE(ubRef, ref);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_REF, 2);
+	OUT_RING_CACHE(ubRef);
+	OUT_RING_CACHE(func);
+}
+
+static void nv50BlendColor(GLcontext *ctx, const GLfloat color[4])
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx); 
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_COLOR_R, 4);
+	OUT_RING_CACHEf(color[0]);
+	OUT_RING_CACHEf(color[1]);
+	OUT_RING_CACHEf(color[2]);
+	OUT_RING_CACHEf(color[3]);
+}
+
+static void nv50BlendEquationSeparate(GLcontext *ctx, GLenum modeRGB, GLenum modeA)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_RGB, 1);
+	OUT_RING_CACHE(modeRGB);
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_EQUATION_ALPHA, 1);
+	OUT_RING_CACHE(modeA);
+}
+
+
+static void nv50BlendFuncSeparate(GLcontext *ctx, GLenum sfactorRGB, GLenum dfactorRGB,
+		GLenum sfactorA, GLenum dfactorA)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_RGB, 2);
+	OUT_RING_CACHE(sfactorRGB);	/* FIXME, sometimes has |0x4000 */
+	OUT_RING_CACHE(dfactorRGB);	/* FIXME, sometimes has |0x4000 */
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_BLEND_FUNC_SRC_ALPHA, 2);
+	OUT_RING_CACHE(sfactorA);	/* FIXME, sometimes has |0x4000 */
+	OUT_RING_CACHE(dfactorA);	/* FIXME, sometimes has |0x4000 */
+}
+
+static void nv50Clear(GLcontext *ctx, GLbitfield mask)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	GLuint hw_bufs = 0;
+
+	if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_BACK_LEFT))
+		hw_bufs |= 0x3c;
+	if (mask & (BUFFER_BIT_STENCIL))
+		hw_bufs |= 0x02;
+	if (mask & (BUFFER_BIT_DEPTH))
+		hw_bufs |= 0x01;
+
+	if (hw_bufs) {
+		BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_BUFFERS, 1);
+		OUT_RING(hw_bufs);
+	}
+}
+
+static void nv50ClearColor(GLcontext *ctx, const GLfloat color[4])
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_COLOR_R, 4);
+	OUT_RING_CACHEf(color[0]);
+	OUT_RING_CACHEf(color[1]);
+	OUT_RING_CACHEf(color[2]);
+	OUT_RING_CACHEf(color[3]);
+}
+
+static void nv50ClearDepth(GLcontext *ctx, GLclampd d)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_DEPTH, 1);
+	OUT_RING_CACHEf(d);
+}
+
+/* we're don't support indexed buffers
+   void (*ClearIndex)(GLcontext *ctx, GLuint index)
+ */
+
+static void nv50ClearStencil(GLcontext *ctx, GLint s)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CLEAR_STENCIL, 1);
+	OUT_RING_CACHE(s);
+}
+
+static void nv50ClipPlane(GLcontext *ctx, GLenum plane, const GLfloat *equation)
+{
+	/* Only using shaders */
+}
+
+static void nv50ColorMask(GLcontext *ctx, GLboolean rmask, GLboolean gmask,
+		GLboolean bmask, GLboolean amask )
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	int i;
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_COLOR_MASK(0), 8);
+	for (i=0; i<8; i++) {
+		OUT_RING_CACHE(((amask && 0x01) << 12) | ((bmask && 0x01) << 8) | ((gmask && 0x01)<< 4) | ((rmask && 0x01) << 0));
+	}
+}
+
+static void nv50ColorMaterial(GLcontext *ctx, GLenum face, GLenum mode)
+{
+	// TODO I need love
+}
+
+static void nv50CullFace(GLcontext *ctx, GLenum mode)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CULL_FACE, 1);
+	OUT_RING_CACHE(mode);
+}
+
+static void nv50FrontFace(GLcontext *ctx, GLenum mode)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_FRONT_FACE, 1);
+	OUT_RING_CACHE(mode);
+}
+
+static void nv50DepthFunc(GLcontext *ctx, GLenum func)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_FUNC, 1);
+	OUT_RING_CACHE(func);
+}
+
+static void nv50DepthMask(GLcontext *ctx, GLboolean flag)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_WRITE_ENABLE, 1);
+	OUT_RING_CACHE(flag);
+}
+
+static void nv50DepthRange(GLcontext *ctx, GLclampd nearval, GLclampd farval)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_RANGE_NEAR, 2);
+	OUT_RING_CACHEf(nearval);
+	OUT_RING_CACHEf(farval);
+}
+
+/** Specify the current buffer for writing */
+//void (*DrawBuffer)( GLcontext *ctx, GLenum buffer );
+/** Specify the buffers for writing for fragment programs*/
+//void (*DrawBuffers)( GLcontext *ctx, GLsizei n, const GLenum *buffers );
+
+static void nv50Enable(GLcontext *ctx, GLenum cap, GLboolean state)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	switch(cap)
+	{
+		case GL_ALPHA_TEST:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_ALPHA_FUNC_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_AUTO_NORMAL:
+//		case GL_BLEND:
+//		case GL_CLIP_PLANE0:
+//		case GL_CLIP_PLANE1:
+//		case GL_CLIP_PLANE2:
+//		case GL_CLIP_PLANE3:
+//		case GL_CLIP_PLANE4:
+//		case GL_CLIP_PLANE5:
+		case GL_COLOR_LOGIC_OP:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LOGIC_OP_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_COLOR_MATERIAL:
+//		case GL_COLOR_SUM_EXT:
+//		case GL_COLOR_TABLE:
+//		case GL_CONVOLUTION_1D:
+//		case GL_CONVOLUTION_2D:
+		case GL_CULL_FACE:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_CULL_FACE_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_DEPTH_TEST:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_DEPTH_TEST_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_DITHER:
+//		case GL_FOG:
+//		case GL_HISTOGRAM:
+//		case GL_INDEX_LOGIC_OP:
+//		case GL_LIGHT0:
+//		case GL_LIGHT1:
+//		case GL_LIGHT2:
+//		case GL_LIGHT3:
+//		case GL_LIGHT4:
+//		case GL_LIGHT5:
+//		case GL_LIGHT6:
+//		case GL_LIGHT7:
+//		case GL_LIGHTING:
+		case GL_LINE_SMOOTH:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_SMOOTH_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_LINE_STIPPLE:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_MAP1_COLOR_4:
+//		case GL_MAP1_INDEX:
+//		case GL_MAP1_NORMAL:
+//		case GL_MAP1_TEXTURE_COORD_1:
+//		case GL_MAP1_TEXTURE_COORD_2:
+//		case GL_MAP1_TEXTURE_COORD_3:
+//		case GL_MAP1_TEXTURE_COORD_4:
+//		case GL_MAP1_VERTEX_3:
+//		case GL_MAP1_VERTEX_4:
+//		case GL_MAP2_COLOR_4:
+//		case GL_MAP2_INDEX:
+//		case GL_MAP2_NORMAL:
+//		case GL_MAP2_TEXTURE_COORD_1:
+//		case GL_MAP2_TEXTURE_COORD_2:
+//		case GL_MAP2_TEXTURE_COORD_3:
+//		case GL_MAP2_TEXTURE_COORD_4:
+//		case GL_MAP2_VERTEX_3:
+//		case GL_MAP2_VERTEX_4:
+//		case GL_MINMAX:
+//		case GL_NORMALIZE:
+//		case GL_POINT_SMOOTH:
+		case GL_POLYGON_OFFSET_POINT:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_POINT_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_POLYGON_OFFSET_LINE:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_LINE_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_POLYGON_OFFSET_FILL:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FILL_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_POLYGON_SMOOTH:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_SMOOTH_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+		case GL_POLYGON_STIPPLE:
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_POST_COLOR_MATRIX_COLOR_TABLE:
+//		case GL_POST_CONVOLUTION_COLOR_TABLE:
+//		case GL_RESCALE_NORMAL:
+		case GL_SCISSOR_TEST:
+			/* No enable bit, nv50Scissor will adjust to max range */
+			ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+			      			 ctx->Scissor.Width, ctx->Scissor.Height);
+			break;
+//		case GL_SEPARABLE_2D:
+		case GL_STENCIL_TEST:
+			// TODO BACK and FRONT ?
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_ENABLE, 1);
+			OUT_RING_CACHE(state);
+			break;
+//		case GL_TEXTURE_GEN_Q:
+//		case GL_TEXTURE_GEN_R:
+//		case GL_TEXTURE_GEN_S:
+//		case GL_TEXTURE_GEN_T:
+//		case GL_TEXTURE_1D:
+//		case GL_TEXTURE_2D:
+//		case GL_TEXTURE_3D:
+	}
+}
+
+static void nv50Fogfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
+{
+	/* Only using shaders */
+}
+   
+static void nv50Hint(GLcontext *ctx, GLenum target, GLenum mode)
+{
+	// TODO I need love (fog and line_smooth hints)
+}
+
+// void (*IndexMask)(GLcontext *ctx, GLuint mask);
+
+static void nv50Lightfv(GLcontext *ctx, GLenum light, GLenum pname, const GLfloat *params )
+{
+	/* Only with shaders */
+}
+
+/** Set the lighting model parameters */
+void (*LightModelfv)(GLcontext *ctx, GLenum pname, const GLfloat *params);
+
+
+static void nv50LineStipple(GLcontext *ctx, GLint factor, GLushort pattern )
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_STIPPLE_PATTERN, 1);
+	OUT_RING_CACHE((pattern << 8) | factor);
+}
+
+static void nv50LineWidth(GLcontext *ctx, GLfloat width)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LINE_WIDTH, 1);
+	OUT_RING_CACHEf(width);
+}
+
+static void nv50LogicOpcode(GLcontext *ctx, GLenum opcode)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_LOGIC_OP_OP, 1);
+	OUT_RING_CACHE(opcode);
+}
+
+static void nv50PointParameterfv(GLcontext *ctx, GLenum pname, const GLfloat *params)
+{
+	/*TODO: not sure what goes here. */
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+	
+}
+
+/** Specify the diameter of rasterized points */
+static void nv50PointSize(GLcontext *ctx, GLfloat size)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POINT_SIZE, 1);
+	OUT_RING_CACHEf(size);
+}
+
+/** Select a polygon rasterization mode */
+static void nv50PolygonMode(GLcontext *ctx, GLenum face, GLenum mode)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_FRONT, 1);
+		OUT_RING_CACHE(mode);
+	}
+	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_MODE_BACK, 1);
+		OUT_RING_CACHE(mode);
+	}
+}
+
+/** Set the scale and units used to calculate depth values */
+static void nv50PolygonOffset(GLcontext *ctx, GLfloat factor, GLfloat units)
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+ 
+        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_FACTOR, 1);
+        OUT_RING_CACHEf(factor);
+        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_OFFSET_UNITS, 1);
+        OUT_RING_CACHEf(units);
+}
+
+/** Set the polygon stippling pattern */
+static void nv50PolygonStipple(GLcontext *ctx, const GLubyte *mask )
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_POLYGON_STIPPLE_PATTERN(0), 32);
+        OUT_RING_CACHEp(mask, 32);
+}
+
+/* Specifies the current buffer for reading */
+void (*ReadBuffer)( GLcontext *ctx, GLenum buffer );
+/** Set rasterization mode */
+void (*RenderMode)(GLcontext *ctx, GLenum mode );
+
+/** Define the scissor box */
+static void nv50Scissor(GLcontext *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
+{
+        nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	/* There's no scissor enable bit, so adjust the scissor to cover the
+	 * maximum draw buffer bounds
+	 */
+	if (!ctx->Scissor.Enabled) {
+	   x = y = 0;
+	   w = h = 8191;
+	} else {
+	   x += nmesa->drawX;
+	   y += nmesa->drawY;
+	}
+
+        BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SCISSOR_WIDTH_XPOS, 2);
+        OUT_RING_CACHE(((w) << 16) | x);
+        OUT_RING_CACHE(((h) << 16) | y);
+}
+
+/** Select flat or smooth shading */
+static void nv50ShadeModel(GLcontext *ctx, GLenum mode)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SHADE_MODEL, 1);
+	OUT_RING_CACHE(mode);
+}
+
+/** OpenGL 2.0 two-sided StencilFunc */
+static void nv50StencilFuncSeparate(GLcontext *ctx, GLenum face, GLenum func,
+		GLint ref, GLuint mask)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_FUNC, 1);
+		OUT_RING_CACHE(func);
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_REF, 1);
+		OUT_RING_CACHE(ref);
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_FUNC_MASK, 1);
+		OUT_RING_CACHE(mask);
+	}
+	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_FUNC, 2);
+		OUT_RING_CACHE(func);
+		OUT_RING_CACHE(ref);
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_FUNC_MASK, 1);
+		OUT_RING_CACHE(mask);
+	}
+}
+
+/** OpenGL 2.0 two-sided StencilMask */
+static void nv50StencilMaskSeparate(GLcontext *ctx, GLenum face, GLuint mask)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_MASK, 1);
+		OUT_RING_CACHE(mask);
+	}
+	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_MASK, 1);
+		OUT_RING_CACHE(mask);
+	}
+}
+
+/** OpenGL 2.0 two-sided StencilOp */
+static void nv50StencilOpSeparate(GLcontext *ctx, GLenum face, GLenum fail,
+		GLenum zfail, GLenum zpass)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	if (face == GL_FRONT || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_FRONT_OP_FAIL, 3);
+		OUT_RING_CACHE(fail);
+		OUT_RING_CACHE(zfail);
+		OUT_RING_CACHE(zpass);
+	}
+	if (face == GL_BACK || face == GL_FRONT_AND_BACK) {
+		BEGIN_RING_CACHE(NvSub3D, NV50_TCL_PRIMITIVE_3D_STENCIL_BACK_OP_FAIL, 3);
+		OUT_RING_CACHE(fail);
+		OUT_RING_CACHE(zfail);
+		OUT_RING_CACHE(zpass);
+	}
+}
+
+/** Control the generation of texture coordinates */
+void (*TexGen)(GLcontext *ctx, GLenum coord, GLenum pname,
+		const GLfloat *params);
+/** Set texture environment parameters */
+void (*TexEnv)(GLcontext *ctx, GLenum target, GLenum pname,
+		const GLfloat *param);
+/** Set texture parameters */
+void (*TexParameter)(GLcontext *ctx, GLenum target,
+		struct gl_texture_object *texObj,
+		GLenum pname, const GLfloat *params);
+
+static void nv50TextureMatrix(GLcontext *ctx, GLuint unit, const GLmatrix *mat)
+{
+	/* Only with shaders */
+}
+
+static void nv50WindowMoved(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;
+
+	BEGIN_RING_CACHE(NvSub3D,
+	      NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(0), 2);
+        OUT_RING_CACHE((8191 << 16) | 0);
+        OUT_RING_CACHE((8191 << 16) | 0);
+	for (i=1; i<8; i++) {
+		BEGIN_RING_CACHE(NvSub3D,
+		      NV50_TCL_PRIMITIVE_3D_VIEWPORT_CLIP_HORIZ(i), 2);
+        	OUT_RING_CACHE(0);
+	        OUT_RING_CACHE(0);
+	}
+
+	ctx->Driver.Scissor(ctx, ctx->Scissor.X, ctx->Scissor.Y,
+	      		    ctx->Scissor.Width, ctx->Scissor.Height);
+}
+
+static GLboolean nv50InitCard(nouveauContextPtr nmesa)
+{
+	int i,j;
+
+	nouveauObjectOnSubchannel(nmesa, NvSub3D, Nv3D);
+
+	BEGIN_RING_SIZE(NvSub3D, 0x1558, 1);
+	OUT_RING(1);
+
+	BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SET_OBJECT_1(0), 8);
+	for (i=0; i<8; i++) {
+		OUT_RING(NvDmaFB);
+	}
+
+	BEGIN_RING_SIZE(NvSub3D, NV50_TCL_PRIMITIVE_3D_SET_OBJECT_0(0), 12);
+	for (i=0; i<12; i++) {
+		OUT_RING(NvDmaFB);
+	}
+
+	BEGIN_RING_SIZE(NvSub3D, 0x121c, 1);
+	OUT_RING(1);
+
+	for (i=0; i<8; i++) {
+		BEGIN_RING_SIZE(NvSub3D, 0x0200 + (i*0x20), 5);
+		for (j=0; j<5; j++) {
+			OUT_RING(0);
+		}
+	}
+
+	BEGIN_RING_SIZE(NvSub3D, 0x0fe0, 5);
+	OUT_RING(0);
+	OUT_RING(0);
+	OUT_RING(0x16);
+	OUT_RING(0);
+	OUT_RING(0);
+
+	return GL_FALSE;
+}
+
+static GLboolean nv50BindBuffers(nouveauContextPtr nmesa, int num_color,
+      				 nouveau_renderbuffer **color,
+				 nouveau_renderbuffer *depth)
+{
+	return GL_FALSE;
+}
+
+void nv50InitStateFuncs(GLcontext *ctx, struct dd_function_table *func)
+{
+	nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+	func->AlphaFunc			= nv50AlphaFunc;
+	func->BlendColor		= nv50BlendColor;
+	func->BlendEquationSeparate	= nv50BlendEquationSeparate;
+	func->BlendFuncSeparate		= nv50BlendFuncSeparate;
+	func->Clear			= nv50Clear;
+	func->ClearColor		= nv50ClearColor;
+	func->ClearDepth		= nv50ClearDepth;
+	func->ClearStencil		= nv50ClearStencil;
+	func->ClipPlane			= nv50ClipPlane;
+	func->ColorMask			= nv50ColorMask;
+	func->ColorMaterial		= nv50ColorMaterial;
+	func->CullFace			= nv50CullFace;
+	func->FrontFace			= nv50FrontFace;
+	func->DepthFunc			= nv50DepthFunc;
+	func->DepthMask			= nv50DepthMask;
+	func->DepthRange		= nv50DepthRange;
+	func->Enable			= nv50Enable;
+	func->Fogfv			= nv50Fogfv;
+	func->Hint			= nv50Hint;
+	func->Lightfv			= nv50Lightfv;
+/*	func->LightModelfv		= nv50LightModelfv;	 */
+	func->LineStipple		= nv50LineStipple;
+	func->LineWidth			= nv50LineWidth;
+	func->LogicOpcode		= nv50LogicOpcode;
+	func->PointParameterfv		= nv50PointParameterfv;
+	func->PointSize			= nv50PointSize;
+	func->PolygonMode		= nv50PolygonMode;
+	func->PolygonOffset		= nv50PolygonOffset;
+	func->PolygonStipple		= nv50PolygonStipple;
+/*	func->ReadBuffer		= nv50ReadBuffer;	*/
+/*	func->RenderMode		= nv50RenderMode;	*/
+	func->Scissor			= nv50Scissor;
+	func->ShadeModel		= nv50ShadeModel;
+	func->StencilFuncSeparate	= nv50StencilFuncSeparate;
+	func->StencilMaskSeparate	= nv50StencilMaskSeparate;
+	func->StencilOpSeparate		= nv50StencilOpSeparate;
+/*	func->TexGen			= nv50TexGen;		*/
+/*	func->TexParameter		= nv50TexParameter;	*/
+	func->TextureMatrix		= nv50TextureMatrix;
+
+	nmesa->hw_func.InitCard		= nv50InitCard;
+	nmesa->hw_func.BindBuffers	= nv50BindBuffers;
+	nmesa->hw_func.WindowMoved	= nv50WindowMoved;
+}
diff --git a/src/mesa/drivers/dri/r128/r128_context.c b/src/mesa/drivers/dri/r128/r128_context.c
index 8ec027542ad..89ddafa02a5 100644
--- a/src/mesa/drivers/dri/r128/r128_context.c
+++ b/src/mesa/drivers/dri/r128/r128_context.c
@@ -42,7 +42,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -235,7 +235,7 @@ GLboolean r128CreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -293,7 +293,7 @@ void r128DestroyContext( __DRIcontextPrivate *driContextPriv  )
 
       _swsetup_DestroyContext( rmesa->glCtx );
       _tnl_DestroyContext( rmesa->glCtx );
-      _ac_DestroyContext( rmesa->glCtx );
+      _vbo_DestroyContext( rmesa->glCtx );
       _swrast_DestroyContext( rmesa->glCtx );
 
       if ( release_texture_heaps ) {
diff --git a/src/mesa/drivers/dri/r128/r128_state.c b/src/mesa/drivers/dri/r128/r128_state.c
index 1bfd3709377..e476afa5d82 100644
--- a/src/mesa/drivers/dri/r128/r128_state.c
+++ b/src/mesa/drivers/dri/r128/r128_state.c
@@ -44,7 +44,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "enums.h"
 #include "colormac.h"
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -1250,7 +1250,7 @@ static void r128DDInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    R128_CONTEXT(ctx)->NewGLState |= new_state;
 }
diff --git a/src/mesa/drivers/dri/r200/Makefile b/src/mesa/drivers/dri/r200/Makefile
index 75c09ff867e..e9144ac75ce 100644
--- a/src/mesa/drivers/dri/r200/Makefile
+++ b/src/mesa/drivers/dri/r200/Makefile
@@ -26,10 +26,6 @@ DRIVER_SOURCES = r200_context.c \
 		 r200_span.c \
 		 r200_maos.c \
 		 r200_sanity.c \
-		 r200_vtxfmt.c \
-		 r200_vtxfmt_c.c \
-		 r200_vtxfmt_sse.c \
-		 r200_vtxfmt_x86.c \
 		 r200_fragshader.c \
 		 r200_vertprog.c \
 		 radeon_screen.c \
@@ -37,7 +33,7 @@ DRIVER_SOURCES = r200_context.c \
 
 C_SOURCES = $(COMMON_SOURCES) $(DRIVER_SOURCES)
 
-X86_SOURCES = r200_vtxtmp_x86.S
+X86_SOURCES = 
 
 DRIVER_DEFINES = -DRADEON_COMMON=1 -DRADEON_COMMON_FOR_R200
 
diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c
index bb7a16c9327..fc6eb93daa8 100644
--- a/src/mesa/drivers/dri/r200/r200_context.c
+++ b/src/mesa/drivers/dri/r200/r200_context.c
@@ -45,7 +45,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -60,7 +60,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_tex.h"
 #include "r200_swtcl.h"
 #include "r200_tcl.h"
-#include "r200_vtxfmt.h"
 #include "r200_maos.h"
 #include "r200_vertprog.h"
 
@@ -415,7 +414,7 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
    _ae_create_context( ctx );
@@ -424,11 +423,10 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
     */
    _tnl_destroy_pipeline( ctx );
    _tnl_install_pipeline( ctx, r200_pipeline );
-   ctx->Driver.FlushVertices = r200FlushVertices;
 
    /* Try and keep materials and vertices separate:
     */
-   _tnl_isolate_materials( ctx, GL_TRUE );
+/*    _tnl_isolate_materials( ctx, GL_TRUE ); */
 
 
    /* Configure swrast and TNL to match hardware characteristics:
@@ -533,12 +531,6 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
       TCL_FALLBACK(rmesa->glCtx, R200_TCL_FALLBACK_TCL_DISABLE, 1);
    }
 
-   if (rmesa->r200Screen->chip_flags & RADEON_CHIPSET_TCL) {
-      if (tcl_mode >= DRI_CONF_TCL_VTXFMT)
-	 r200VtxfmtInit( ctx, tcl_mode >= DRI_CONF_TCL_CODEGEN );
-
-      _tnl_need_dlist_norm_lengths( ctx, GL_FALSE );
-   }
    return GL_TRUE;
 }
 
@@ -568,7 +560,7 @@ void r200DestroyContext( __DRIcontextPrivate *driContextPriv )
       release_texture_heaps = (rmesa->glCtx->Shared->RefCount == 1);
       _swsetup_DestroyContext( rmesa->glCtx );
       _tnl_DestroyContext( rmesa->glCtx );
-      _ac_DestroyContext( rmesa->glCtx );
+      _vbo_DestroyContext( rmesa->glCtx );
       _swrast_DestroyContext( rmesa->glCtx );
 
       r200DestroySwtcl( rmesa->glCtx );
@@ -579,12 +571,6 @@ void r200DestroyContext( __DRIcontextPrivate *driContextPriv )
 	 r200FlushCmdBuf( rmesa, __FUNCTION__ );
       }
 
-      if (!(rmesa->TclFallback & R200_TCL_FALLBACK_TCL_DISABLE)) {
-	 int tcl_mode = driQueryOptioni(&rmesa->optionCache, "tcl_mode");
-	 if (tcl_mode >= DRI_CONF_TCL_VTXFMT)
-	    r200VtxfmtDestroy( rmesa->glCtx );
-      }
-
       if (rmesa->state.scissor.pClipRects) {
 	 FREE(rmesa->state.scissor.pClipRects);
 	 rmesa->state.scissor.pClipRects = NULL;
@@ -700,9 +686,6 @@ r200MakeCurrent( __DRIcontextPrivate *driContextPriv,
 			  (GLframebuffer *) driDrawPriv->driverPrivate,
 			  (GLframebuffer *) driReadPriv->driverPrivate );
 
-      if (newCtx->vb.enabled)
-	 r200VtxfmtMakeCurrent( newCtx->glCtx );
-
       _mesa_update_state( newCtx->glCtx );
       r200ValidateState( newCtx->glCtx );
 
@@ -727,6 +710,5 @@ r200UnbindContext( __DRIcontextPrivate *driContextPriv )
    if (R200_DEBUG & DEBUG_DRI)
       fprintf(stderr, "%s ctx %p\n", __FUNCTION__, (void *)rmesa->glCtx);
 
-   r200VtxfmtUnbindContext( rmesa->glCtx );
    return GL_TRUE;
 }
diff --git a/src/mesa/drivers/dri/r200/r200_context.h b/src/mesa/drivers/dri/r200/r200_context.h
index 44c67b68cbc..e840a502c0b 100644
--- a/src/mesa/drivers/dri/r200/r200_context.h
+++ b/src/mesa/drivers/dri/r200/r200_context.h
@@ -107,8 +107,7 @@ struct r200_vertex_program {
         VERTEX_SHADER_INSTRUCTION instr[R200_VSF_MAX_INST + 6];
         int pos_end;
         int inputs[VERT_ATTRIB_MAX];
-        int rev_inputs[16];
-        int gen_inputs_mapped;
+        GLubyte inputmap_rev[16];
         int native;
         int fogpidx;
         int fogmode;
@@ -724,8 +723,6 @@ struct r200_store {
 /* r200_tcl.c
  */
 struct r200_tcl_info {
-   GLuint vertex_format;
-   GLint last_offset;
    GLuint hw_primitive;
 
 /* hw can handle 12 components max */
@@ -735,14 +732,7 @@ struct r200_tcl_info {
    GLuint *Elts;
 
    struct r200_dma_region indexed_verts;
-   struct r200_dma_region weight;
-   struct r200_dma_region obj;
-   struct r200_dma_region rgba;
-   struct r200_dma_region spec;
-   struct r200_dma_region fog;
-   struct r200_dma_region tex[R200_MAX_TEXTURE_UNITS];
-   struct r200_dma_region norm;
-   struct r200_dma_region generic[16];
+   struct r200_dma_region vertex_data[15];
 };
 
 
@@ -812,87 +802,6 @@ struct r200_ioctl {
 #define R200_MAX_PRIMS 64
 
 
-/* Want to keep a cache of these around.  Each is parameterized by
- * only a single value which has only a small range.  Only expect a
- * few, so just rescan the list each time?
- */
-struct dynfn {
-   struct dynfn *next, *prev;
-   int key[2];
-   char *code;
-};
-
-struct dfn_lists {
-   struct dynfn Vertex2f;
-   struct dynfn Vertex2fv;
-   struct dynfn Vertex3f;
-   struct dynfn Vertex3fv;
-   struct dynfn Color4ub;
-   struct dynfn Color4ubv;
-   struct dynfn Color3ub;
-   struct dynfn Color3ubv;
-   struct dynfn Color4f;
-   struct dynfn Color4fv;
-   struct dynfn Color3f;
-   struct dynfn Color3fv;
-   struct dynfn SecondaryColor3ubEXT;
-   struct dynfn SecondaryColor3ubvEXT;
-   struct dynfn SecondaryColor3fEXT;
-   struct dynfn SecondaryColor3fvEXT;
-   struct dynfn Normal3f;
-   struct dynfn Normal3fv;
-   struct dynfn TexCoord3f;
-   struct dynfn TexCoord3fv;
-   struct dynfn TexCoord2f;
-   struct dynfn TexCoord2fv;
-   struct dynfn TexCoord1f;
-   struct dynfn TexCoord1fv;
-   struct dynfn MultiTexCoord3fARB;
-   struct dynfn MultiTexCoord3fvARB;
-   struct dynfn MultiTexCoord2fARB;
-   struct dynfn MultiTexCoord2fvARB;
-   struct dynfn MultiTexCoord1fARB;
-   struct dynfn MultiTexCoord1fvARB;
-   struct dynfn FogCoordfEXT;
-   struct dynfn FogCoordfvEXT;
-};
-
-struct dfn_generators {
-   struct dynfn *(*Vertex2f)( GLcontext *, const int * );
-   struct dynfn *(*Vertex2fv)( GLcontext *, const int * );
-   struct dynfn *(*Vertex3f)( GLcontext *, const int * );
-   struct dynfn *(*Vertex3fv)( GLcontext *, const int * );
-   struct dynfn *(*Color4ub)( GLcontext *, const int * );
-   struct dynfn *(*Color4ubv)( GLcontext *, const int * );
-   struct dynfn *(*Color3ub)( GLcontext *, const int * );
-   struct dynfn *(*Color3ubv)( GLcontext *, const int * );
-   struct dynfn *(*Color4f)( GLcontext *, const int * );
-   struct dynfn *(*Color4fv)( GLcontext *, const int * );
-   struct dynfn *(*Color3f)( GLcontext *, const int * );
-   struct dynfn *(*Color3fv)( GLcontext *, const int * );
-   struct dynfn *(*SecondaryColor3ubEXT)( GLcontext *, const int * );
-   struct dynfn *(*SecondaryColor3ubvEXT)( GLcontext *, const int * );
-   struct dynfn *(*SecondaryColor3fEXT)( GLcontext *, const int * );
-   struct dynfn *(*SecondaryColor3fvEXT)( GLcontext *, const int * );
-   struct dynfn *(*Normal3f)( GLcontext *, const int * );
-   struct dynfn *(*Normal3fv)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord3f)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord3fv)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord2f)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord2fv)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord1f)( GLcontext *, const int * );
-   struct dynfn *(*TexCoord1fv)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord3fARB)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord3fvARB)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord2fARB)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord2fvARB)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord1fARB)( GLcontext *, const int * );
-   struct dynfn *(*MultiTexCoord1fvARB)( GLcontext *, const int * );
-   struct dynfn *(*FogCoordfEXT)( GLcontext *, const int * );
-   struct dynfn *(*FogCoordfvEXT)( GLcontext *, const int * );
-};
-
-
 
 struct r200_prim {
    GLuint start;
@@ -913,43 +822,6 @@ struct r200_prim {
 
 #define R200_MAX_VERTEX_SIZE ((3*6)+11)
 
-struct r200_vbinfo {
-   GLint counter, initial_counter;
-   GLint *dmaptr;
-   void (*notify)( void );
-   GLint vertex_size;
-
-   union { float f; int i; r200_color_t color; } vertex[R200_MAX_VERTEX_SIZE];
-
-   GLfloat *normalptr;
-   GLfloat *floatcolorptr;
-   GLfloat *fogptr;
-   r200_color_t *colorptr;
-   GLfloat *floatspecptr;
-   r200_color_t *specptr;
-   GLfloat *texcoordptr[8];	/* 6 (TMU) + 2 for r200_vtxfmt_c.c when GL_TEXTURE6/7 */
-
-
-   GLenum *prim;		/* &ctx->Driver.CurrentExecPrimitive */
-   GLuint primflags;
-   GLboolean enabled;		/* *_NO_VTXFMT / *_NO_TCL env vars */
-   GLboolean installed;
-   GLboolean fell_back;
-   GLboolean recheck;
-   GLint nrverts;
-   GLuint vtxfmt_0, vtxfmt_1;
-
-   GLuint installed_vertex_format;
-   GLuint installed_color_3f_sz;
-
-   struct r200_prim primlist[R200_MAX_PRIMS];
-   int nrprims;
-
-   struct dfn_lists dfn_cache;
-   struct dfn_generators codegen;
-   GLvertexformat vtxfmt;
-};
-
 
 struct r200_context {
    GLcontext *glCtx;			/* Mesa context */
@@ -1041,10 +913,6 @@ struct r200_context {
     */
    struct r200_swtcl_info swtcl;
 
-   /* r200_vtxfmt.c
-    */
-   struct r200_vbinfo vb;
-
    /* Mirrors of some DRI state
     */
    struct r200_dri_mirror dri;
diff --git a/src/mesa/drivers/dri/r200/r200_maos.h b/src/mesa/drivers/dri/r200/r200_maos.h
index b9e4d3c2392..4998f67445c 100644
--- a/src/mesa/drivers/dri/r200/r200_maos.h
+++ b/src/mesa/drivers/dri/r200/r200_maos.h
@@ -38,7 +38,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "r200_context.h"
 
-extern void r200EmitArrays( GLcontext *ctx, GLuint inputs );
+extern void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev );
 extern void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs );
 
 #endif
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index 270dc35a46a..3162b508c23 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -376,7 +376,7 @@ static void emit_vector( GLcontext *ctx,
 /* Emit any changed arrays to new GART memory, re-emit a packet to
  * update the arrays.  
  */
-void r200EmitArrays( GLcontext *ctx, GLuint inputs )
+void r200EmitArrays( GLcontext *ctx, GLubyte *vimap_rev )
 {
    r200ContextPtr rmesa = R200_CONTEXT( ctx );
    struct vertex_buffer *VB = &TNL_CONTEXT( ctx )->vb;
@@ -384,335 +384,130 @@ void r200EmitArrays( GLcontext *ctx, GLuint inputs )
    GLuint nr = 0;
    GLuint vfmt0 = 0, vfmt1 = 0;
    GLuint count = VB->Count;
-   GLuint i;
-   GLuint generic_in_mapped = 0;
-   struct r200_vertex_program *vp = NULL;
-
-   /* this looks way more complicated than necessary... */
-   if (ctx->VertexProgram._Enabled) {
-      vp = rmesa->curr_vp_hw;
-      generic_in_mapped = vp->gen_inputs_mapped;
-   }
-
-   if (inputs & VERT_BIT_POS) {
-      if (!rmesa->tcl.obj.buf) 
-	 emit_vector( ctx, 
-		      &rmesa->tcl.obj, 
-		      (char *)VB->ObjPtr->data,
-		      VB->ObjPtr->size,
-		      VB->ObjPtr->stride,
-		      count);
-
-      switch( VB->ObjPtr->size ) {
-      case 4: vfmt0 |= R200_VTX_W0;
-      case 3: vfmt0 |= R200_VTX_Z0;
-      case 2: 
-      default:
-	 break;
-      }
-      component[nr++] = &rmesa->tcl.obj;
-   }
-   else if (generic_in_mapped & (1 << 0)) {
-      int geninput = vp->rev_inputs[0] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_W0 | R200_VTX_Z0;
-   }
-
-   if (inputs & VERT_BIT_WEIGHT) {
-      if (!rmesa->tcl.weight.buf)
-	 emit_vector( ctx, 
-		      &rmesa->tcl.weight, 
-		      (char *)VB->AttribPtr[VERT_ATTRIB_WEIGHT]->data,
-		      VB->AttribPtr[VERT_ATTRIB_WEIGHT]->size,
-		      VB->AttribPtr[VERT_ATTRIB_WEIGHT]->stride,
-		      count);
-
-      assert(VB->AttribPtr[VERT_ATTRIB_WEIGHT]->size <= 4);
-      vfmt0 |= VB->AttribPtr[VERT_ATTRIB_WEIGHT]->size << R200_VTX_WEIGHT_COUNT_SHIFT;
-      component[nr++] = &rmesa->tcl.weight;
-   }
-
-   if (inputs & VERT_BIT_NORMAL) {
-      if (!rmesa->tcl.norm.buf)
-	 emit_vector( ctx, 
-		      &(rmesa->tcl.norm), 
-		      (char *)VB->NormalPtr->data,
-		      3,
-		      VB->NormalPtr->stride,
-		      count);
-
-      vfmt0 |= R200_VTX_N0;
-      component[nr++] = &rmesa->tcl.norm;
-   }
-
-   if (inputs & VERT_BIT_FOG) {
-      if (!rmesa->tcl.fog.buf) {
-	 if (ctx->VertexProgram._Enabled)
-	    emit_vector( ctx, 
-			 &(rmesa->tcl.fog), 
-			 (char *)VB->FogCoordPtr->data,
+   GLuint i, emitsize;
+
+   for ( i = 0; i < 15; i++ ) {
+      GLubyte attrib = vimap_rev[i];
+      if (attrib != 255) {
+	 switch (i) {
+	 case 0:
+	    emitsize = (VB->AttribPtr[attrib]->size);
+	    switch (emitsize) {
+	    case 4:
+	       vfmt0 |= R200_VTX_W0;
+	       /* fallthrough */
+	    case 3:
+	       vfmt0 |= R200_VTX_Z0;
+	       break;
+	    case 2:
+	       break;
+	    default: assert(0);
+	    }
+	    break;
+	 case 1:
+	    assert(attrib == VERT_ATTRIB_WEIGHT);
+	    emitsize = (VB->AttribPtr[attrib]->size);
+	    vfmt0 |= emitsize << R200_VTX_WEIGHT_COUNT_SHIFT;
+	    break;
+	 case 2:
+	    assert(attrib == VERT_ATTRIB_NORMAL);
+	    emitsize = 3;
+	    vfmt0 |= R200_VTX_N0;
+	    break;
+	 case 3:
+	    /* special handling to fix up fog. Will get us into trouble with vbos...*/
+	    assert(attrib == VERT_ATTRIB_FOG);
+	    if (!rmesa->tcl.vertex_data[i].buf) {
+	       if (ctx->VertexProgram._Enabled)
+		  emit_vector( ctx,
+			 &(rmesa->tcl.vertex_data[i]),
+			 (char *)VB->AttribPtr[attrib]->data,
 			 1,
-			 VB->FogCoordPtr->stride,
+			 VB->AttribPtr[attrib]->stride,
 			 count);
-	 else
-	    emit_vecfog( ctx, 
-			 &(rmesa->tcl.fog), 
-			 (char *)VB->FogCoordPtr->data,
-			 VB->FogCoordPtr->stride,
+	       else
+		  emit_vecfog( ctx,
+			 &(rmesa->tcl.vertex_data[i]),
+			 (char *)VB->AttribPtr[attrib]->data,
+			 VB->AttribPtr[attrib]->stride,
 			 count);
-      }
-
-      vfmt0 |= R200_VTX_DISCRETE_FOG;
-      component[nr++] = &rmesa->tcl.fog;
-   }
- 
-   if (inputs & VERT_BIT_COLOR0) {
-      int emitsize;
-
-      if (VB->ColorPtr[0]->size == 4 &&
-	  (VB->ColorPtr[0]->stride != 0 ||
-	   VB->ColorPtr[0]->data[0][3] != 1.0)) { 
-	 vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT; 
-	 emitsize = 4;
-      }
-      else { 
-	 vfmt0 |= R200_VTX_FP_RGB << R200_VTX_COLOR_0_SHIFT; 
-	 emitsize = 3;
-      }
-
-      if (!rmesa->tcl.rgba.buf)
-	 emit_vector( ctx, 
-		      &(rmesa->tcl.rgba), 
-		      (char *)VB->ColorPtr[0]->data,
-		      emitsize,
-		      VB->ColorPtr[0]->stride,
-		      count);
-
-      component[nr++] = &rmesa->tcl.rgba;
-   }
-/*	vfmt0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT;
-	emit_ubyte_rgba( ctx, &rmesa->tcl.rgba, 
-		(char *)VB->ColorPtr[0]->data, 4,
-		      VB->ColorPtr[0]->stride, count);*/
-   else if (generic_in_mapped & (1 << 2)) {
-      int geninput = vp->rev_inputs[2] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT;
-   }
-
-
-   if (inputs & VERT_BIT_COLOR1) {
-      if (!rmesa->tcl.spec.buf) {
-	 emit_vector( ctx, 
-		      &rmesa->tcl.spec, 
-		      (char *)VB->SecondaryColorPtr[0]->data,
-		      3,
-		      VB->SecondaryColorPtr[0]->stride,
-		      count);
-      }
-
-      /* How does this work?
-       */
-      vfmt0 |= R200_VTX_FP_RGB << R200_VTX_COLOR_1_SHIFT; 
-      component[nr++] = &rmesa->tcl.spec;
-   }
-   else if (generic_in_mapped & (1 << 3)) {
-      int geninput = vp->rev_inputs[3] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_1_SHIFT;
-   }
-
-   if (generic_in_mapped & (1 << 4)) {
-      int geninput = vp->rev_inputs[4] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_2_SHIFT;
-   }
-
-   if (generic_in_mapped & (1 << 5)) {
-      int geninput = vp->rev_inputs[5] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_3_SHIFT;
-   }
-
-   for ( i = 0 ; i < 6 ; i++ ) {
-      if (inputs & (VERT_BIT_TEX0 << i)) {
-	 if (!rmesa->tcl.tex[i].buf)
-	     emit_vector( ctx, 
-			  &(rmesa->tcl.tex[i]),
-			  (char *)VB->TexCoordPtr[i]->data,
-			  VB->TexCoordPtr[i]->size,
-			  VB->TexCoordPtr[i]->stride,
-			  count );
-
-	 vfmt1 |= VB->TexCoordPtr[i]->size << (i * 3);
-	 component[nr++] = &rmesa->tcl.tex[i];
-      }
-      else if (generic_in_mapped & (1 << (i + 6))) {
-	 int geninput = vp->rev_inputs[i + 6] - VERT_ATTRIB_GENERIC0;
-	 if (!rmesa->tcl.generic[geninput].buf) {
-            emit_vector( ctx,
-			 &(rmesa->tcl.generic[geninput]),
-			(char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-			4,
-			VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-			count );
-	 }
-	 component[nr++] = &rmesa->tcl.generic[geninput];
-	 vfmt1 |= 4 << (R200_VTX_TEX0_COMP_CNT_SHIFT + (i * 3));
-      }
-   }
-
-   if (generic_in_mapped & (1 << 13)) {
-      int geninput = vp->rev_inputs[13] - VERT_ATTRIB_GENERIC0;
-      if (!rmesa->tcl.generic[geninput].buf) {
-         emit_vector( ctx,
-		      &(rmesa->tcl.generic[geninput]),
-		      (char *)VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->data,
-		      4,
-		      VB->AttribPtr[geninput + VERT_ATTRIB_GENERIC0]->stride,
-		      count );
-      }
-      component[nr++] = &rmesa->tcl.generic[geninput];
-      vfmt0 |= R200_VTX_XY1 | R200_VTX_Z1 | R200_VTX_W1;
-   }
-
-/* doesn't work. Wrong order with mixed generic & conventional! */
-/*
-   if (ctx->VertexProgram._Enabled) {
-      int *vp_inputs = rmesa->curr_vp_hw->inputs;
-      for ( i = VERT_ATTRIB_GENERIC0; i < VERT_ATTRIB_MAX; i++ ) {
-	 if (inputs & (1 << i)) {
-	    int geninput = i - VERT_ATTRIB_GENERIC0;
-	    if (!rmesa->tcl.generic[geninput].buf) {
-	       emit_vector( ctx,
-			  &(rmesa->tcl.generic[geninput]),
-			  (char *)VB->AttribPtr[i]->data,
-			  4,
-			  VB->AttribPtr[i]->stride,
-			  count );
 	    }
-	    component[nr++] = &rmesa->tcl.generic[geninput];
-	    switch (vp_inputs[i]) {
-	    case 0:
-	       vfmt0 |=  R200_VTX_W0 | R200_VTX_Z0;
-	       break;
+	    vfmt0 |= R200_VTX_DISCRETE_FOG;
+	    goto after_emit;
+	    break;
+	 case 4:
+	 case 5:
+	 case 6:
+	 case 7:
+	    if (VB->AttribPtr[attrib]->size == 4 &&
+	       (VB->AttribPtr[attrib]->stride != 0 ||
+		VB->AttribPtr[attrib]->data[0][3] != 1.0)) emitsize = 4;
+	    else emitsize = 3;
+	    if (emitsize == 4)
+	       vfmt0 |= R200_VTX_FP_RGBA << (R200_VTX_COLOR_0_SHIFT + (i - 4) * 2);
+	    else {
+	       vfmt0 |= R200_VTX_FP_RGB << (R200_VTX_COLOR_0_SHIFT + (i - 4) * 2);
+	    }
+	    break;
+	 case 8:
+	 case 9:
+	 case 10:
+	 case 11:
+	 case 12:
+	 case 13:
+	    emitsize = VB->AttribPtr[attrib]->size;
+	    vfmt1 |= emitsize << (R200_VTX_TEX0_COMP_CNT_SHIFT + (i - 8) * 3);
+	    break;
+	 case 14:
+	    emitsize = VB->AttribPtr[attrib]->size >= 2 ? VB->AttribPtr[attrib]->size : 2;
+	    switch (emitsize) {
 	    case 2:
+	       vfmt0 |= R200_VTX_XY1;
+	       /* fallthrough */
 	    case 3:
+	       vfmt0 |= R200_VTX_Z1;
+	       /* fallthrough */
 	    case 4:
-	    case 5:
-	       vfmt0 |= R200_VTX_FP_RGBA << (R200_VTX_COLOR_0_SHIFT + (vp_inputs[i] - 2) * 2);
-	       break;
-	    case 6:
-	    case 7:
-	    case 8:
-	    case 9:
-	    case 10:
-	    case 11:
-	       vfmt1 |= 4 << (R200_VTX_TEX0_COMP_CNT_SHIFT + (vp_inputs[i] - 6) * 3);
-	       break;
-	    case 13:
-	       vfmt0 |= R200_VTX_XY1 | R200_VTX_Z1 | R200_VTX_W1;
-	       break;
-	    case 1:
-	    case 12:
-	    default:
-	       assert(0);
+	       vfmt0 |= R200_VTX_W1;
+	    break;
 	    }
+	 default:
+	    assert(0);
 	 }
+	 if (!rmesa->tcl.vertex_data[i].buf) {
+	    emit_vector( ctx,
+			 &(rmesa->tcl.vertex_data[i]),
+			 (char *)VB->AttribPtr[attrib]->data,
+			 emitsize,
+			 VB->AttribPtr[attrib]->stride,
+			 count );
+	 }
+after_emit:
+	 assert(nr < 12);
+	 component[nr++] = &rmesa->tcl.vertex_data[i];
       }
    }
-*/
 
    if (vfmt0 != rmesa->hw.vtx.cmd[VTX_VTXFMT_0] ||
-       vfmt1 != rmesa->hw.vtx.cmd[VTX_VTXFMT_1]) { 
-      R200_STATECHANGE( rmesa, vtx ); 
+       vfmt1 != rmesa->hw.vtx.cmd[VTX_VTXFMT_1]) {
+      R200_STATECHANGE( rmesa, vtx );
       rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = vfmt0;
       rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = vfmt1;
-   } 
+   }
 
    rmesa->tcl.nr_aos_components = nr;
-   rmesa->tcl.vertex_format = vfmt0;
 }
 
 
 void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs )
 {
-   GLuint unit;
    r200ContextPtr rmesa = R200_CONTEXT( ctx );
 
-/*    if (R200_DEBUG & DEBUG_VERTS)  */
-/*       _tnl_print_vert_flags( __FUNCTION__, newinputs ); */
-
-   if (newinputs & VERT_BIT_POS) 
-     r200ReleaseDmaRegion( rmesa, &rmesa->tcl.obj, __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_WEIGHT) 
-     r200ReleaseDmaRegion( rmesa, &rmesa->tcl.weight, __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_NORMAL) 
-      r200ReleaseDmaRegion( rmesa, &rmesa->tcl.norm, __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_FOG) 
-      r200ReleaseDmaRegion( rmesa, &rmesa->tcl.fog, __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_COLOR0) 
-      r200ReleaseDmaRegion( rmesa, &rmesa->tcl.rgba, __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_COLOR1) 
-      r200ReleaseDmaRegion( rmesa, &rmesa->tcl.spec, __FUNCTION__ );
-
-   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (newinputs & VERT_BIT_TEX(unit))
-	 r200ReleaseDmaRegion( rmesa, &rmesa->tcl.tex[unit], __FUNCTION__ );
-   }
-
-   if (ctx->VertexProgram._Enabled) {
-      int i;
-      for (i = VERT_ATTRIB_GENERIC0; i < VERT_ATTRIB_MAX; i++) {
-	 if (newinputs & (1 << i))
-	    r200ReleaseDmaRegion( rmesa,
-	       &rmesa->tcl.generic[i - VERT_ATTRIB_GENERIC0], __FUNCTION__ );
-      }
+   /* only do it for changed inputs ? */
+   int i;
+   for (i = 0; i < 15; i++) {
+      if (newinputs & (1 << i))
+	 r200ReleaseDmaRegion( rmesa,
+	    &rmesa->tcl.vertex_data[i], __FUNCTION__ );
    }
-
 }
diff --git a/src/mesa/drivers/dri/r200/r200_state.c b/src/mesa/drivers/dri/r200/r200_state.c
index dc1fbef72ee..bdb487f2b93 100644
--- a/src/mesa/drivers/dri/r200/r200_state.c
+++ b/src/mesa/drivers/dri/r200/r200_state.c
@@ -43,7 +43,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "framebuffer.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
@@ -54,7 +54,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_tcl.h"
 #include "r200_tex.h"
 #include "r200_swtcl.h"
-#include "r200_vtxfmt.h"
 #include "r200_vertprog.h"
 
 #include "drirenderbuffer.h"
@@ -2536,23 +2535,24 @@ static void r200InvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    _ae_invalidate_state( ctx, new_state );
    R200_CONTEXT(ctx)->NewGLState |= new_state;
-   r200VtxfmtInvalidate( ctx );
 }
 
 /* A hack.  The r200 can actually cope just fine with materials
- * between begin/ends, so fix this. But how ?
+ * between begin/ends, so fix this.
+ * Should map to inputs just like the generic vertex arrays for vertex progs.
+ * In theory there could still be too many and we'd still need a fallback.
  */
 static GLboolean check_material( GLcontext *ctx )
 {
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    GLint i;
 
-   for (i = _TNL_ATTRIB_MAT_FRONT_AMBIENT; 
-	i < _TNL_ATTRIB_MAT_BACK_INDEXES; 
+   for (i = _TNL_ATTRIB_MAT_FRONT_AMBIENT;
+	i < _TNL_ATTRIB_MAT_BACK_INDEXES;
 	i++)
       if (tnl->vb.AttribPtr[i] &&
 	  tnl->vb.AttribPtr[i]->stride)
@@ -2560,7 +2560,7 @@ static GLboolean check_material( GLcontext *ctx )
 
    return GL_FALSE;
 }
-      
+
 static void r200WrapRunPipeline( GLcontext *ctx )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
@@ -2574,7 +2574,7 @@ static void r200WrapRunPipeline( GLcontext *ctx )
    if (rmesa->NewGLState)
       r200ValidateState( ctx );
 
-   has_material = (ctx->Light.Enabled && check_material( ctx ));
+   has_material = !ctx->VertexProgram._Enabled && ctx->Light.Enabled && check_material( ctx );
 
    if (has_material) {
       TCL_FALLBACK( ctx, R200_TCL_FALLBACK_MATERIAL, GL_TRUE );
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index d95a80c7bbc..b40d0bdcb7c 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -39,7 +39,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "api_arrayelt.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
@@ -50,7 +50,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_tcl.h"
 #include "r200_tex.h"
 #include "r200_swtcl.h"
-#include "r200_vtxfmt.h"
 
 #include "xmlpool.h"
 
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index c14a275f7a3..25d229d8ed6 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -48,7 +48,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
-#include "tnl/t_vtx_api.h"
 
 #include "r200_context.h"
 #include "r200_ioctl.h"
@@ -936,13 +935,6 @@ r200PointsBitmap( GLcontext *ctx, GLint px, GLint py,
 }
 
 
-void r200FlushVertices( GLcontext *ctx, GLuint flags )
-{
-   _tnl_FlushVertices( ctx, flags );
-
-   if (flags & FLUSH_STORED_VERTICES)
-      R200_NEWPRIM( R200_CONTEXT( ctx ) );
-}
 
 /**********************************************************************/
 /*                            Initialization.                         */
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.h b/src/mesa/drivers/dri/r200/r200_swtcl.h
index ce2b6b5f06b..ccf817988c5 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.h
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.h
@@ -42,7 +42,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 extern void r200InitSwtcl( GLcontext *ctx );
 extern void r200DestroySwtcl( GLcontext *ctx );
 
-extern void r200FlushVertices( GLcontext *ctx, GLuint flags );
 extern void r200ChooseRenderState( GLcontext *ctx );
 extern void r200ChooseVertexState( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index 86edf7132ce..e0c32b26d92 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -40,7 +40,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "colormac.h"
 #include "light.h"
 
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 
@@ -384,8 +384,14 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
    struct vertex_buffer *VB = &tnl->vb;
-   GLuint inputs = 0;
    GLuint i;
+   GLubyte *vimap_rev;
+/* use hw fixed order for simplicity, pos 0, weight 1, normal 2, fog 3, 
+   color0 - color3 4-7, texcoord0 - texcoord5 8-13, pos 1 14. Must not use
+   more than 12 of those at the same time. */
+   GLubyte map_rev_fixed[15] = {255, 255, 255, 255, 255, 255, 255, 255,
+			    255, 255, 255, 255, 255, 255, 255};
+
 
    /* TODO: separate this from the swtnl pipeline 
     */
@@ -404,30 +410,40 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
       r200ValidateState( ctx );
 
    if (!ctx->VertexProgram._Enabled) {
-      inputs = VERT_BIT_POS | VERT_BIT_COLOR0;
    /* NOTE: inputs != tnl->render_inputs - these are the untransformed
     * inputs.
     */
+      map_rev_fixed[0] = VERT_ATTRIB_POS;
+      /* technically there is no reason we always need VA_COLOR0. In theory
+         could disable it depending on lighting, color materials, texturing... */
+      map_rev_fixed[4] = VERT_ATTRIB_COLOR0;
+
       if (ctx->Light.Enabled) {
-	 inputs |= VERT_BIT_NORMAL;
+	 map_rev_fixed[2] = VERT_ATTRIB_NORMAL;
       }
 
+      /* this also enables VA_COLOR1 when using separate specular
+         lighting model, which is unnecessary.
+         FIXME: OTOH, we're missing the case where a ATI_fragment_shader accesses
+         the secondary color (if lighting is disabled). The chip seems
+         misconfigured for that though elsewhere (tcl output, might lock up) */
       if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) {
-	 inputs |= VERT_BIT_COLOR1;
+	 map_rev_fixed[5] = VERT_ATTRIB_COLOR1;
       }
 
       if ( (ctx->Fog.FogCoordinateSource == GL_FOG_COORD) && ctx->Fog.Enabled ) {
-	 inputs |= VERT_BIT_FOG;
+	 map_rev_fixed[3] = VERT_ATTRIB_FOG;
       }
 
       for (i = 0 ; i < ctx->Const.MaxTextureUnits; i++) {
 	 if (ctx->Texture.Unit[i]._ReallyEnabled) {
 	    if (rmesa->TexGenNeedNormals[i]) {
-	       inputs |= VERT_BIT_NORMAL;
+	       map_rev_fixed[2] = VERT_ATTRIB_NORMAL;
 	    }
-	    inputs |= VERT_BIT_TEX(i);
+	    map_rev_fixed[8 + i] = VERT_ATTRIB_TEX0 + i;
 	 }
       }
+      vimap_rev = &map_rev_fixed[0];
    }
    else {
       /* vtx_tcl_output_vtxfmt_0/1 need to match configuration of "fragment
@@ -437,14 +453,8 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
 	 We only need to change compsel. */
       GLuint out_compsel = 0;
       GLuint vp_out = rmesa->curr_vp_hw->mesa_program.Base.OutputsWritten;
-#if 0
-      /* can't handle other inputs, generic attribs etc. currently - should never arrive here */
-      assert ((rmesa->curr_vp_hw->mesa_program.Base.InputsRead &
-	 ~(VERT_BIT_POS | VERT_BIT_NORMAL | VERT_BIT_COLOR0 | VERT_BIT_COLOR1 |
-	  VERT_BIT_FOG | VERT_BIT_TEX0 | VERT_BIT_TEX1 | VERT_BIT_TEX2 |
-	  VERT_BIT_TEX3 | VERT_BIT_TEX4 | VERT_BIT_TEX5)) == 0);
-#endif
-      inputs |= rmesa->curr_vp_hw->mesa_program.Base.InputsRead;
+
+      vimap_rev = &rmesa->curr_vp_hw->inputmap_rev[0];
       assert(vp_out & (1 << VERT_RESULT_HPOS));
       out_compsel = R200_OUTPUT_XYZW;
       if (vp_out & (1 << VERT_RESULT_COL0)) {
@@ -473,7 +483,7 @@ static GLboolean r200_run_tcl_render( GLcontext *ctx,
    /* Do the actual work:
     */
    r200ReleaseArrays( ctx, ~0 /* stage->changed_inputs */ );
-   r200EmitArrays( ctx, inputs );
+   r200EmitArrays( ctx, vimap_rev );
 
    rmesa->tcl.Elts = VB->Elts;
 
diff --git a/src/mesa/drivers/dri/r200/r200_vertprog.c b/src/mesa/drivers/dri/r200/r200_vertprog.c
index 491701b796e..4960d481d5d 100644
--- a/src/mesa/drivers/dri/r200/r200_vertprog.c
+++ b/src/mesa/drivers/dri/r200/r200_vertprog.c
@@ -405,7 +405,6 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
    int dofogfix = 0;
    int fog_temp_i = 0;
    int free_inputs;
-   int free_inputs_conv;
    int array_count = 0;
 
    vp->native = GL_FALSE;
@@ -461,7 +460,7 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 
    /* for fogc, can't change mesa_vp, as it would hose swtnl, and exp with
       base e isn't directly available neither. */
-   if (mesa_vp->Base.OutputsWritten & VERT_RESULT_FOGC && !vp->fogpidx) {
+   if ((mesa_vp->Base.OutputsWritten & (1 << VERT_RESULT_FOGC)) && !vp->fogpidx) {
       struct gl_program_parameter_list *paramList;
       GLint tokens[6] = { STATE_FOG_PARAMS, 0, 0, 0, 0, 0 };
       paramList = mesa_vp->Base.Parameters;
@@ -477,6 +476,8 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 
    for(i = 0; i < VERT_ATTRIB_MAX; i++)
       vp->inputs[i] = -1;
+   for(i = 0; i < 15; i++)
+      vp->inputmap_rev[i] = 255;
    free_inputs = 0x2ffd;
 
 /* fglrx uses fixed inputs as follows for conventional attribs.
@@ -499,38 +500,45 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 /* may look different when using idx buf / input_route instead of se_vtx_fmt? */
    if (mesa_vp->Base.InputsRead & VERT_BIT_POS) {
       vp->inputs[VERT_ATTRIB_POS] = 0;
+      vp->inputmap_rev[0] = VERT_ATTRIB_POS;
       free_inputs &= ~(1 << 0);
       array_count++;
    }
    if (mesa_vp->Base.InputsRead & VERT_BIT_WEIGHT) {
       vp->inputs[VERT_ATTRIB_WEIGHT] = 12;
+      vp->inputmap_rev[1] = VERT_ATTRIB_WEIGHT;
       array_count++;
    }
    if (mesa_vp->Base.InputsRead & VERT_BIT_NORMAL) {
       vp->inputs[VERT_ATTRIB_NORMAL] = 1;
+      vp->inputmap_rev[2] = VERT_ATTRIB_NORMAL;
       array_count++;
    }
    if (mesa_vp->Base.InputsRead & VERT_BIT_COLOR0) {
       vp->inputs[VERT_ATTRIB_COLOR0] = 2;
+      vp->inputmap_rev[4] = VERT_ATTRIB_COLOR0;
       free_inputs &= ~(1 << 2);
       array_count++;
    }
    if (mesa_vp->Base.InputsRead & VERT_BIT_COLOR1) {
       vp->inputs[VERT_ATTRIB_COLOR1] = 3;
+      vp->inputmap_rev[5] = VERT_ATTRIB_COLOR1;
       free_inputs &= ~(1 << 3);
       array_count++;
    }
    if (mesa_vp->Base.InputsRead & VERT_BIT_FOG) {
       vp->inputs[VERT_ATTRIB_FOG] = 15; array_count++;
+      vp->inputmap_rev[3] = VERT_ATTRIB_FOG;
+      array_count++;
    }
    for (i = VERT_ATTRIB_TEX0; i <= VERT_ATTRIB_TEX5; i++) {
       if (mesa_vp->Base.InputsRead & (1 << i)) {
 	 vp->inputs[i] = i - VERT_ATTRIB_TEX0 + 6;
+	 vp->inputmap_rev[8 + i - VERT_ATTRIB_TEX0] = i;
 	 free_inputs &= ~(1 << (i - VERT_ATTRIB_TEX0 + 6));
 	 array_count++;
       }
    }
-   free_inputs_conv = free_inputs;
    /* using VERT_ATTRIB_TEX6/7 would be illegal */
    /* completely ignore aliasing? */
    for (i = VERT_ATTRIB_GENERIC0; i < VERT_ATTRIB_MAX; i++) {
@@ -549,13 +557,14 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 	    if (free_inputs & (1 << j)) {
 	       free_inputs &= ~(1 << j);
 	       vp->inputs[i] = j;
-	       vp->rev_inputs[j] = i;
+	       if (j == 0) vp->inputmap_rev[j] = i; /* mapped to pos */
+	       else if (j < 12) vp->inputmap_rev[j + 2] = i; /* mapped to col/tex */
+	       else vp->inputmap_rev[j + 1] = i; /* mapped to pos1 */
 	       break;
 	    }
 	 }
       }
    }
-   vp->gen_inputs_mapped = free_inputs ^ free_inputs_conv;
 
    if (!(mesa_vp->Base.OutputsWritten & (1 << VERT_RESULT_HPOS))) {
       if (R200_DEBUG & DEBUG_FALLBACKS) {
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt.c b/src/mesa/drivers/dri/r200/r200_vtxfmt.c
deleted file mode 100644
index d73fbbafd5f..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt.c
+++ /dev/null
@@ -1,1234 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.c,v 1.4 2003/05/06 23:52:08 daenzer Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "imports.h"
-#include "r200_context.h"
-#include "r200_state.h"
-#include "r200_ioctl.h"
-#include "r200_tex.h"
-#include "r200_tcl.h"
-#include "r200_swtcl.h"
-#include "r200_vtxfmt.h"
-
-#include "api_noop.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "mtypes.h"
-#include "enums.h"
-#include "glapi.h"
-#include "colormac.h"
-#include "light.h"
-#include "state.h"
-#include "vtxfmt.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-#include "tnl/t_array_api.h"
-#include "tnl/t_save_api.h"
-
-#include "dispatch.h"
-
-static void r200VtxFmtFlushVertices( GLcontext *, GLuint );
-
-static void count_func( const char *name,  struct dynfn *l )
-{
-   int i = 0;
-   struct dynfn *f;
-   foreach (f, l) i++;
-   if (i) fprintf(stderr, "%s: %d\n", name, i );
-}
-
-static void count_funcs( r200ContextPtr rmesa )
-{
-   count_func( "Vertex2f", &rmesa->vb.dfn_cache.Vertex2f );
-   count_func( "Vertex2fv", &rmesa->vb.dfn_cache.Vertex2fv );
-   count_func( "Vertex3f", &rmesa->vb.dfn_cache.Vertex3f );
-   count_func( "Vertex3fv", &rmesa->vb.dfn_cache.Vertex3fv );
-   count_func( "Color4ub", &rmesa->vb.dfn_cache.Color4ub ); 
-   count_func( "Color4ubv", &rmesa->vb.dfn_cache.Color4ubv ); 
-   count_func( "Color3ub", &rmesa->vb.dfn_cache.Color3ub ); 
-   count_func( "Color3ubv", &rmesa->vb.dfn_cache.Color3ubv ); 
-   count_func( "Color4f", &rmesa->vb.dfn_cache.Color4f );
-   count_func( "Color4fv", &rmesa->vb.dfn_cache.Color4fv );
-   count_func( "Color3f", &rmesa->vb.dfn_cache.Color3f );
-   count_func( "Color3fv", &rmesa->vb.dfn_cache.Color3fv );
-   count_func( "SecondaryColor3f", &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   count_func( "SecondaryColor3fv", &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   count_func( "SecondaryColor3ub", &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   count_func( "SecondaryColor3ubv", &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   count_func( "Normal3f", &rmesa->vb.dfn_cache.Normal3f );
-   count_func( "Normal3fv", &rmesa->vb.dfn_cache.Normal3fv );
-   count_func( "TexCoord3f", &rmesa->vb.dfn_cache.TexCoord3f );
-   count_func( "TexCoord3fv", &rmesa->vb.dfn_cache.TexCoord3fv );
-   count_func( "TexCoord2f", &rmesa->vb.dfn_cache.TexCoord2f );
-   count_func( "TexCoord2fv", &rmesa->vb.dfn_cache.TexCoord2fv );
-   count_func( "TexCoord1f", &rmesa->vb.dfn_cache.TexCoord1f );
-   count_func( "TexCoord1fv", &rmesa->vb.dfn_cache.TexCoord1fv );
-   count_func( "MultiTexCoord3fARB", &rmesa->vb.dfn_cache.MultiTexCoord3fARB );
-   count_func( "MultiTexCoord3fvARB", &rmesa->vb.dfn_cache.MultiTexCoord3fvARB );
-   count_func( "MultiTexCoord2fARB", &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   count_func( "MultiTexCoord2fvARB", &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   count_func( "MultiTexCoord1fARB", &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   count_func( "MultiTexCoord1fvARB", &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-/*   count_func( "FogCoordfEXT", &rmesa->vb.dfn_cache.FogCoordfEXT );
-   count_func( "FogCoordfvEXT", &rmesa->vb.dfn_cache.FogCoordfvEXT );*/
-}
-
-static void r200NewList( GLcontext *ctx, GLuint list, GLenum mode )
-{
-   VFMT_FALLBACK( __FUNCTION__ );
-   _tnl_NewList( ctx, list, mode );
-   return;
-}
-
-void r200_copy_to_current( GLcontext *ctx ) 
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   unsigned i;
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT);
-
-   if (rmesa->vb.vtxfmt_0 & R200_VTX_N0) {
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][0] = rmesa->vb.normalptr[0];
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][1] = rmesa->vb.normalptr[1];
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][2] = rmesa->vb.normalptr[2];
-   }
-
-   if (rmesa->vb.vtxfmt_0 & R200_VTX_DISCRETE_FOG) {
-      ctx->Current.Attrib[VERT_ATTRIB_FOG][0] = rmesa->vb.fogptr[0];
-   }
-
-   switch( VTX_COLOR(rmesa->vb.vtxfmt_0, 0) ) {
-   case R200_VTX_PK_RGBA:
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->red );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->green );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->blue );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->alpha );
-      break;
-
-   case R200_VTX_FP_RGB:
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = rmesa->vb.floatcolorptr[0];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = rmesa->vb.floatcolorptr[1];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = rmesa->vb.floatcolorptr[2];
-      break;
-
-   case R200_VTX_FP_RGBA:
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = rmesa->vb.floatcolorptr[0];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = rmesa->vb.floatcolorptr[1];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = rmesa->vb.floatcolorptr[2];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = rmesa->vb.floatcolorptr[3];
-      break;
-      
-   default:
-      break;
-   }
-      
-   if (VTX_COLOR(rmesa->vb.vtxfmt_0, 1) == R200_VTX_PK_RGBA) {
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][0] = UBYTE_TO_FLOAT( rmesa->vb.specptr->red );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][1] = UBYTE_TO_FLOAT( rmesa->vb.specptr->green );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] = UBYTE_TO_FLOAT( rmesa->vb.specptr->blue );
-   } 
-
-   for ( i = 0 ; i < ctx->Const.MaxTextureUnits ; i++ ) {
-      const unsigned count = VTX_TEXn_COUNT( rmesa->vb.vtxfmt_1, i );
-      GLfloat * const src = rmesa->vb.texcoordptr[i];
-
-      if ( count != 0 ) {
-	 switch( count ) {
-	 case 3:
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][1] = src[1];
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][2] = src[2];
-	    break;
-	 case 2:
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][1] = src[1];
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][2] = 0.0F;
-	    break;
-	 case 1:
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][1] = 0.0F;
-	    ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][2] = 0.0F;
-	    break;
-	 }
-
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][0] = src[0];
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX0+i][3] = 1.0F;
-      }
-   }
-
-   ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-}
-
-static GLboolean discreet_gl_prim[GL_POLYGON+1] = {
-   1,				/* 0 points */
-   1,				/* 1 lines */
-   0,				/* 2 line_strip */
-   0,				/* 3 line_loop */
-   1,				/* 4 tris */
-   0,				/* 5 tri_fan */
-   0,				/* 6 tri_strip */
-   1,				/* 7 quads */
-   0,				/* 8 quadstrip */
-   0,				/* 9 poly */
-};
-
-static void flush_prims( r200ContextPtr rmesa )
-{
-   int i,j;
-   struct r200_dma_region tmp = rmesa->dma.current;
-   
-   tmp.buf->refcount++;
-   tmp.aos_size = rmesa->vb.vertex_size;
-   tmp.aos_stride = rmesa->vb.vertex_size;
-   tmp.aos_start = GET_START(&tmp);
-
-   rmesa->dma.current.ptr = rmesa->dma.current.start += 
-      (rmesa->vb.initial_counter - rmesa->vb.counter) * 
-      rmesa->vb.vertex_size * 4; 
-
-   rmesa->tcl.vertex_format = rmesa->vb.vtxfmt_0;
-   rmesa->tcl.aos_components[0] = &tmp;
-   rmesa->tcl.nr_aos_components = 1;
-   rmesa->dma.flush = NULL;
-
-   /* Optimize the primitive list:
-    */
-   if (rmesa->vb.nrprims > 1) {
-      for (j = 0, i = 1 ; i < rmesa->vb.nrprims; i++) {
-	 int pj = rmesa->vb.primlist[j].prim & 0xf;
-	 int pi = rmesa->vb.primlist[i].prim & 0xf;
-      
-	 if (pj == pi && discreet_gl_prim[pj] &&
-	     rmesa->vb.primlist[i].start == rmesa->vb.primlist[j].end) {
-	    rmesa->vb.primlist[j].end = rmesa->vb.primlist[i].end;
-	 }
-	 else {
-	    j++;
-	    if (j != i) rmesa->vb.primlist[j] = rmesa->vb.primlist[i];
-	 }
-      }
-      rmesa->vb.nrprims = j+1;
-   }
-
-   if (rmesa->vb.vtxfmt_0 != rmesa->hw.vtx.cmd[VTX_VTXFMT_0] ||
-       rmesa->vb.vtxfmt_1 != rmesa->hw.vtx.cmd[VTX_VTXFMT_1]) { 
-      R200_STATECHANGE( rmesa, vtx ); 
-      rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = rmesa->vb.vtxfmt_0;
-      rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = rmesa->vb.vtxfmt_1;
-   } 
-
-
-   for (i = 0 ; i < rmesa->vb.nrprims; i++) {
-      if (R200_DEBUG & DEBUG_PRIMS)
-	 fprintf(stderr, "vtxfmt prim %d: %s %d..%d\n", i,
-		 _mesa_lookup_enum_by_nr( rmesa->vb.primlist[i].prim & 
-					  PRIM_MODE_MASK ),
-		 rmesa->vb.primlist[i].start,
-		 rmesa->vb.primlist[i].end);
-
-      if (rmesa->vb.primlist[i].start < rmesa->vb.primlist[i].end)
-	 r200EmitPrimitive( rmesa->glCtx,
-			    rmesa->vb.primlist[i].start,
-			    rmesa->vb.primlist[i].end,
-			    rmesa->vb.primlist[i].prim );
-   }
-
-   rmesa->vb.nrprims = 0;
-   r200ReleaseDmaRegion( rmesa, &tmp, __FUNCTION__ );
-}
-
-
-static void start_prim( r200ContextPtr rmesa, GLuint mode )
-{
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, 
-	      rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   rmesa->vb.primlist[rmesa->vb.nrprims].start = 
-      rmesa->vb.initial_counter - rmesa->vb.counter;
-   rmesa->vb.primlist[rmesa->vb.nrprims].prim = mode;
-}
-
-static void note_last_prim( r200ContextPtr rmesa, GLuint flags )
-{
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, 
-	      rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      rmesa->vb.primlist[rmesa->vb.nrprims].prim |= flags;
-      rmesa->vb.primlist[rmesa->vb.nrprims].end = 
-	 rmesa->vb.initial_counter - rmesa->vb.counter;
-
-      if (++(rmesa->vb.nrprims) == R200_MAX_PRIMS)
-	 flush_prims( rmesa );
-   }
-}
-
-
-static void copy_vertex( r200ContextPtr rmesa, GLuint n, GLfloat *dst )
-{
-   GLuint i;
-   GLfloat *src = (GLfloat *)(rmesa->dma.current.address + 
-			      rmesa->dma.current.ptr + 
-			      (rmesa->vb.primlist[rmesa->vb.nrprims].start + n) * 
-			      rmesa->vb.vertex_size * 4);
-
-   if (R200_DEBUG & DEBUG_VFMT) 
-      fprintf(stderr, "copy_vertex %d\n", rmesa->vb.primlist[rmesa->vb.nrprims].start + n);
-
-   for (i = 0 ; i < rmesa->vb.vertex_size; i++) {
-      dst[i] = src[i];
-   }
-}
-
-/* NOTE: This actually reads the copied vertices back from uncached
- * memory.  Could also use the counter/notify mechanism to populate
- * tmp on the fly as vertices are generated.  
- */
-static GLuint copy_dma_verts( r200ContextPtr rmesa, GLfloat (*tmp)[R200_MAX_VERTEX_SIZE] )
-{
-   GLuint ovf, i;
-   GLuint nr = (rmesa->vb.initial_counter - rmesa->vb.counter) - 
-      rmesa->vb.primlist[rmesa->vb.nrprims].start;
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d verts\n", __FUNCTION__, nr);
-
-   switch( rmesa->vb.prim[0] )
-   {
-   case GL_POINTS:
-      return 0;
-   case GL_LINES:
-      ovf = nr&1;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_TRIANGLES:
-      ovf = nr%3;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_QUADS:
-      ovf = nr&3;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_LINE_STRIP:
-      if (nr == 0) 
-	 return 0;
-      copy_vertex( rmesa, nr-1, tmp[0] );
-      return 1;
-   case GL_LINE_LOOP:
-   case GL_TRIANGLE_FAN:
-   case GL_POLYGON:
-      if (nr == 0) 
-	 return 0;
-      else if (nr == 1) {
-	 copy_vertex( rmesa, 0, tmp[0] );
-	 return 1;
-      } else {
-	 copy_vertex( rmesa, 0, tmp[0] );
-	 copy_vertex( rmesa, nr-1, tmp[1] );
-	 return 2;
-      }
-   case GL_TRIANGLE_STRIP:
-      ovf = MIN2( nr, 2 );
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_QUAD_STRIP:
-      switch (nr) {
-      case 0: ovf = 0; break;
-      case 1: ovf = 1; break;
-      default: ovf = 2 + (nr&1); break;
-      }
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-static void VFMT_FALLBACK_OUTSIDE_BEGIN_END( const char *caller )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
-      fprintf(stderr, "%s from %s\n", __FUNCTION__, caller);
-
-   if (ctx->Driver.NeedFlush) 
-      r200VtxFmtFlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   if (ctx->NewState)
-      _mesa_update_state( ctx ); /* clear state so fell_back sticks */
-
-   _tnl_wakeup_exec( ctx );
-   ctx->Driver.FlushVertices = r200FlushVertices;
-   ctx->Driver.NewList = _tnl_NewList;
-
-   assert( rmesa->dma.flush == 0 );
-   rmesa->vb.fell_back = GL_TRUE;
-   rmesa->vb.installed = GL_FALSE;
-}
-
-
-/**
- * \todo
- * An interesting optimization of this function would be to have 3 element
- * table with the dispatch offsets of the TexCoord?fv functions, use count
- * to look-up the table, and a specialized version of GL_CALL that used the
- * offset number instead of the name.
- */
-static void dispatch_multitexcoord( GLuint count, GLuint unit, GLfloat * f )
-{
-   switch( count ) {
-   case 3:
-      CALL_MultiTexCoord3fvARB(GET_DISPATCH(), (GL_TEXTURE0+unit, f));
-      break;
-   case 2:
-      CALL_MultiTexCoord2fvARB(GET_DISPATCH(), (GL_TEXTURE0+unit, f));
-      break;
-   case 1:
-      CALL_MultiTexCoord1fvARB(GET_DISPATCH(), (GL_TEXTURE0+unit, f));
-      break;
-   default:
-      assert( count == 0 );
-      break;
-   }
-}
-
-void VFMT_FALLBACK( const char *caller )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat tmp[3][R200_MAX_VERTEX_SIZE];
-   GLuint i, prim;
-   GLuint ind0 = rmesa->vb.vtxfmt_0;
-   GLuint ind1 = rmesa->vb.vtxfmt_1;
-   GLuint nrverts;
-   GLfloat alpha = 1.0;
-   GLuint count;
-   GLuint unit;
-
-   if (R200_DEBUG & (DEBUG_FALLBACKS|DEBUG_VFMT))
-      fprintf(stderr, "%s from %s\n", __FUNCTION__, caller);
-
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) {
-      VFMT_FALLBACK_OUTSIDE_BEGIN_END( __FUNCTION__ );
-      return;
-   }
-
-   /* Copy vertices out of dma:
-    */
-   nrverts = copy_dma_verts( rmesa, tmp );
-
-   /* Finish the prim at this point:
-    */
-   note_last_prim( rmesa, 0 );
-   flush_prims( rmesa );
-
-   /* Update ctx->Driver.CurrentExecPrimitive and swap in swtnl. 
-    */
-   prim = rmesa->vb.prim[0];
-   ctx->Driver.CurrentExecPrimitive = GL_POLYGON+1;
-   _tnl_wakeup_exec( ctx );
-   ctx->Driver.FlushVertices = r200FlushVertices;
-
-   assert(rmesa->dma.flush == 0);
-   rmesa->vb.fell_back = GL_TRUE;
-   rmesa->vb.installed = GL_FALSE;
-   CALL_Begin(GET_DISPATCH(), (prim));
-
-   if (rmesa->vb.installed_color_3f_sz == 4)
-      alpha = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3];
-
-   /* Replay saved vertices
-    */
-   for (i = 0 ; i < nrverts; i++) {
-      GLuint offset = 3;
-
-      if (ind0 & R200_VTX_N0) {
-	 CALL_Normal3fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset += 3;
-      }
-
-      if (ind0 & R200_VTX_DISCRETE_FOG) {
-	 CALL_FogCoordfvEXT(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset++;
-      }
-
-      if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) {
-	 CALL_Color4ubv(GET_DISPATCH(), ((GLubyte *)&tmp[i][offset]));
-	 offset++;
-      }
-      else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) {
-	 CALL_Color4fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset+=4;
-      } 
-      else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) {
-	 CALL_Color3fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset+=3;
-      }
-
-      if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) {
-	 CALL_SecondaryColor3ubvEXT(GET_DISPATCH(), ((GLubyte *)&tmp[i][offset]));
-	 offset++;
-      }
-
-      for ( unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
-	 count = VTX_TEXn_COUNT( ind1, unit );
-	 dispatch_multitexcoord( count, unit, &tmp[i][offset] );
-	 offset += count;
-      }
-
-      CALL_Vertex3fv(GET_DISPATCH(), (&tmp[i][0]));
-   }
-
-   /* Replay current vertex
-    */
-   if (ind0 & R200_VTX_N0) 
-      CALL_Normal3fv(GET_DISPATCH(), (rmesa->vb.normalptr));
-   if (ind0 & R200_VTX_DISCRETE_FOG) {
-      CALL_FogCoordfvEXT(GET_DISPATCH(), (rmesa->vb.fogptr));
-   }
-
-   if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) {
-      CALL_Color4ub(GET_DISPATCH(), (rmesa->vb.colorptr->red,
-				     rmesa->vb.colorptr->green,
-				     rmesa->vb.colorptr->blue,
-				     rmesa->vb.colorptr->alpha));
-   }
-   else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) {
-      CALL_Color4fv(GET_DISPATCH(), (rmesa->vb.floatcolorptr));
-   }
-   else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) {
-      if (rmesa->vb.installed_color_3f_sz == 4 && alpha != 1.0) {
-	 CALL_Color4f(GET_DISPATCH(), (rmesa->vb.floatcolorptr[0],
-				       rmesa->vb.floatcolorptr[1],
-				       rmesa->vb.floatcolorptr[2],
-				       alpha));
-      }
-      else {
-	 CALL_Color3fv(GET_DISPATCH(), (rmesa->vb.floatcolorptr));
-      }
-   }
-
-   if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) 
-       CALL_SecondaryColor3ubEXT(GET_DISPATCH(), (rmesa->vb.specptr->red, 
-						  rmesa->vb.specptr->green,
-						  rmesa->vb.specptr->blue));
-
-   for ( unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
-      count = VTX_TEXn_COUNT( ind1, unit );
-      dispatch_multitexcoord( count, unit, rmesa->vb.texcoordptr[unit] );
-   }
-}
-
-
-
-static void wrap_buffer( void )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat tmp[3][R200_MAX_VERTEX_SIZE];
-   GLuint i, nrverts;
-
-   if (R200_DEBUG & (DEBUG_VFMT|DEBUG_PRIMS))
-      fprintf(stderr, "%s %d\n", __FUNCTION__,
-	      rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   /* Don't deal with parity.
-    */
-   if ((((rmesa->vb.initial_counter - rmesa->vb.counter) -  
-	 rmesa->vb.primlist[rmesa->vb.nrprims].start) & 1)) {
-      rmesa->vb.counter++;
-      rmesa->vb.initial_counter++;
-      return;
-   }
-
-   /* Copy vertices out of dma:
-    */
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) 
-      nrverts = 0;
-   else {
-      nrverts = copy_dma_verts( rmesa, tmp );
-
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%d vertices to copy\n", nrverts);
-   
-      /* Finish the prim at this point:
-       */
-      note_last_prim( rmesa, 0 );
-   }
-
-   /* Fire any buffered primitives
-    */
-   flush_prims( rmesa );
-
-   /* Get new buffer
-    */
-   r200RefillCurrentDmaRegion( rmesa );
-
-   /* Reset counter, dmaptr
-    */
-   rmesa->vb.dmaptr = (int *)(rmesa->dma.current.ptr + rmesa->dma.current.address);
-   rmesa->vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / 
-      (rmesa->vb.vertex_size * 4);
-   rmesa->vb.counter--;
-   rmesa->vb.initial_counter = rmesa->vb.counter;
-   rmesa->vb.notify = wrap_buffer;
-
-   rmesa->dma.flush = flush_prims;
-
-   /* Restart wrapped primitive:
-    */
-   if (rmesa->vb.prim[0] != GL_POLYGON+1)
-      start_prim( rmesa, rmesa->vb.prim[0] );
-
-
-   /* Reemit saved vertices
-    */
-   for (i = 0 ; i < nrverts; i++) {
-      if (R200_DEBUG & DEBUG_VERTS) {
-	 int j;
-	 fprintf(stderr, "re-emit vertex %d to %p\n", i, 
-		 (void *)rmesa->vb.dmaptr);
-	 if (R200_DEBUG & DEBUG_VERBOSE)
-	    for (j = 0 ; j < rmesa->vb.vertex_size; j++) 
-	       fprintf(stderr, "\t%08x/%f\n", *(int*)&tmp[i][j], tmp[i][j]);
-      }
-
-      memcpy( rmesa->vb.dmaptr, tmp[i], rmesa->vb.vertex_size * 4 );
-      rmesa->vb.dmaptr += rmesa->vb.vertex_size;
-      rmesa->vb.counter--;
-   }
-}
-
-
-/**
- * Determines the hardware vertex format based on the current state vector.
- * 
- * \returns
- * If the hardware TCL unit is capable of handling the current state vector,
- * \c GL_TRUE is returned.  Otherwise, \c GL_FALSE is returned.
- *
- * \todo
- * Make this color format selection data driven.  If we receive only ubytes,
- * send color as ubytes.  Also check if converting (with free checking for
- * overflow) is cheaper than sending floats directly.
- *
- * \todo
- * When intializing texture coordinates, it might be faster to just copy the
- * entire \c VERT_ATTRIB_TEX0 vector into the vertex buffer.  It may mean that
- * some of the data (i.e., the last texture coordinate components) get copied
- * over, but that still may be faster than the conditional branching.  If
- * nothing else, the code will be smaller and easier to follow.
- */
-static GLboolean check_vtx_fmt( GLcontext *ctx )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLuint ind0 = R200_VTX_Z0;
-   GLuint ind1 = 0;
-   GLuint i;
-   GLuint count[R200_MAX_TEXTURE_UNITS];
-
-   if (rmesa->TclFallback || rmesa->vb.fell_back || ctx->CompileFlag ||
-      (ctx->Fog.Enabled && (ctx->Fog.FogCoordinateSource == GL_FOG_COORD)) ||
-      /* TODO: set tcl out fmt/compsel and reenable vtxfmt code */
-      ctx->VertexProgram._Enabled)
-      return GL_FALSE;
-
-   if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) 
-      ctx->Driver.FlushVertices( ctx, FLUSH_UPDATE_CURRENT );
-   
-   /* Make all this event-driven:
-    */
-   if (ctx->Light.Enabled) {
-      ind0 |= R200_VTX_N0;
-
-      if (ctx->Light.ColorMaterialEnabled) 
-	 ind0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT;
-      else
-	 ind0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT;
-   }
-   else {
-      /* TODO: make this data driven?
-       */
-      ind0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT;
-	 
-      if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) {
-	 ind0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT;
-      }
-   }
-
-   if ( ctx->Fog.FogCoordinateSource == GL_FOG_COORD ) {
-      ind0 |= R200_VTX_DISCRETE_FOG;
-   }
-
-   for ( i = 0 ; i < ctx->Const.MaxTextureUnits ; i++ ) {
-      count[i] = 0;
-
-      if (ctx->Texture.Unit[i]._ReallyEnabled) {
-	 if (rmesa->TexGenNeedNormals[i]) {
-	    ind0 |= R200_VTX_N0;
-	 }
-	 else {
-	    switch( ctx->Texture.Unit[i]._ReallyEnabled ) {
-	    case TEXTURE_CUBE_BIT:
-	    case TEXTURE_3D_BIT:
-	       count[i] = 3;
-	       break;
-	    case TEXTURE_2D_BIT:
-	    case TEXTURE_RECT_BIT:
-	       count[i] = 2;
-	       break;
-	    case TEXTURE_1D_BIT:
-	       count[i] = 1;
-	       break;
-	    }
-
-	    ind1 |= count[i] << (3 * i);
-	 }
-      }
-   }
-
-   if (R200_DEBUG & (DEBUG_VFMT|DEBUG_STATE))
-      fprintf(stderr, "%s: format: 0x%x, 0x%x\n", __FUNCTION__, ind0, ind1 );
-
-   R200_NEWPRIM(rmesa);
-   rmesa->vb.vtxfmt_0 = ind0;
-   rmesa->vb.vtxfmt_1 = ind1;
-   rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive;
-
-   rmesa->vb.vertex_size = 3;
-   rmesa->vb.normalptr = ctx->Current.Attrib[VERT_ATTRIB_NORMAL];
-   rmesa->vb.colorptr = NULL;
-   rmesa->vb.floatcolorptr = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
-   rmesa->vb.fogptr = ctx->Current.Attrib[VERT_ATTRIB_FOG];
-   rmesa->vb.specptr = NULL;
-   rmesa->vb.floatspecptr = ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
-   rmesa->vb.texcoordptr[0] = ctx->Current.Attrib[VERT_ATTRIB_TEX0];
-   rmesa->vb.texcoordptr[1] = ctx->Current.Attrib[VERT_ATTRIB_TEX1];
-   rmesa->vb.texcoordptr[2] = ctx->Current.Attrib[VERT_ATTRIB_TEX2];
-   rmesa->vb.texcoordptr[3] = ctx->Current.Attrib[VERT_ATTRIB_TEX3];
-   rmesa->vb.texcoordptr[4] = ctx->Current.Attrib[VERT_ATTRIB_TEX4];
-   rmesa->vb.texcoordptr[5] = ctx->Current.Attrib[VERT_ATTRIB_TEX5];
-   rmesa->vb.texcoordptr[6] = ctx->Current.Attrib[VERT_ATTRIB_TEX0];	/* dummy */
-   rmesa->vb.texcoordptr[7] = ctx->Current.Attrib[VERT_ATTRIB_TEX0];	/* dummy */
-
-   /* Run through and initialize the vertex components in the order
-    * the hardware understands:
-    */
-   if (ind0 & R200_VTX_N0) {
-      rmesa->vb.normalptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 3;
-      rmesa->vb.normalptr[0] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][0];
-      rmesa->vb.normalptr[1] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][1];
-      rmesa->vb.normalptr[2] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][2];
-   }
-
-   if (ind0 & R200_VTX_DISCRETE_FOG) {
-      rmesa->vb.fogptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 1;
-      rmesa->vb.fogptr[0] = ctx->Current.Attrib[VERT_ATTRIB_FOG][0];
-   }
-
-   if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) {
-      rmesa->vb.colorptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].color;
-      rmesa->vb.vertex_size += 1;
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->red,   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->green, ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->blue,  ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->alpha, ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] );
-   }
-   else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) {
-      rmesa->vb.floatcolorptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 4;
-      rmesa->vb.floatcolorptr[0] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0];
-      rmesa->vb.floatcolorptr[1] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1];
-      rmesa->vb.floatcolorptr[2] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2];
-      rmesa->vb.floatcolorptr[3] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3];
-   }
-   else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) {
-      rmesa->vb.floatcolorptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 3;
-      rmesa->vb.floatcolorptr[0] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0];
-      rmesa->vb.floatcolorptr[1] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1];
-      rmesa->vb.floatcolorptr[2] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2];
-   }   
-   
-   if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) {
-      rmesa->vb.specptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].color;
-      rmesa->vb.vertex_size += 1;
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->red,   ctx->Current.Attrib[VERT_ATTRIB_COLOR1][0] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->green, ctx->Current.Attrib[VERT_ATTRIB_COLOR1][1] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->blue,  ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] );
-   }
-
-
-   for ( i = 0 ; i < ctx->Const.MaxTextureUnits ; i++ ) {
-      if ( count[i] != 0 ) {
-	 float * const attr = ctx->Current.Attrib[VERT_ATTRIB_TEX0+i];
-	 unsigned  j;
-
-	 rmesa->vb.texcoordptr[i] = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-
-	 for ( j = 0 ; j < count[i] ; j++ ) {
-	    rmesa->vb.texcoordptr[i][j] = attr[j];
-	 }
-
-	 rmesa->vb.vertex_size += count[i];
-      }
-   }
-
-   if (rmesa->vb.installed_vertex_format != rmesa->vb.vtxfmt_0) {
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "reinstall on vertex_format change\n");
-      _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-      rmesa->vb.installed_vertex_format = rmesa->vb.vtxfmt_0;
-   }
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s -- success\n", __FUNCTION__);
-
-   return GL_TRUE;
-}
-
-
-void r200VtxfmtInvalidate( GLcontext *ctx )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   rmesa->vb.recheck = GL_TRUE;
-   rmesa->vb.fell_back = GL_FALSE;
-}
-
-
-static void r200VtxfmtValidate( GLcontext *ctx )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (ctx->Driver.NeedFlush)
-      ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   rmesa->vb.recheck = GL_FALSE;
-
-   if (check_vtx_fmt( ctx )) {
-      if (!rmesa->vb.installed) {
-	 if (R200_DEBUG & DEBUG_VFMT)
-	    fprintf(stderr, "reinstall (new install)\n");
-
-	 _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-	 ctx->Driver.FlushVertices = r200VtxFmtFlushVertices;
-	 ctx->Driver.NewList = r200NewList;
-	 rmesa->vb.installed = GL_TRUE;
-      }
-      else if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: already installed", __FUNCTION__);
-   } 
-   else {
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: failed\n", __FUNCTION__);
-
-      if (rmesa->vb.installed) {
-	 if (rmesa->dma.flush)
-	    rmesa->dma.flush( rmesa );
-	 _tnl_wakeup_exec( ctx );
-	 ctx->Driver.FlushVertices = r200FlushVertices;
-	 ctx->Driver.NewList =_tnl_NewList;
-	 rmesa->vb.installed = GL_FALSE;
-      }
-   }      
-}
-
-
-
-/* Materials:
- */
-static void r200_Materialfv( GLenum face, GLenum pname, 
-			       const GLfloat *params )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      VFMT_FALLBACK( __FUNCTION__ );
-      CALL_Materialfv(GET_DISPATCH(), (face, pname, params));
-      return;
-   }
-   _mesa_noop_Materialfv( face, pname, params );
-   r200UpdateMaterial( ctx );
-}
-
-
-/* Begin/End
- */
-static void r200_Begin( GLenum mode )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s( %s )\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr( mode ));
-
-   if (mode > GL_POLYGON) {
-      _mesa_error( ctx, GL_INVALID_ENUM, "glBegin" );
-      return;
-   }
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" );
-      return;
-   }
-   
-   if (ctx->NewState) 
-      _mesa_update_state( ctx );
-
-   if (rmesa->NewGLState)
-      r200ValidateState( ctx );
-
-   if (rmesa->vb.recheck) 
-      r200VtxfmtValidate( ctx );
-
-   if (!rmesa->vb.installed) {
-      CALL_Begin(GET_DISPATCH(), (mode));
-      return;
-   }
-
-
-   if (rmesa->dma.flush && rmesa->vb.counter < 12) {
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: flush almost-empty buffers\n", __FUNCTION__);
-      flush_prims( rmesa );
-   }
-
-   /* Need to arrange to save vertices here?  Or always copy from dma (yuk)?
-    */
-   if (!rmesa->dma.flush) {
-      if (rmesa->dma.current.ptr + 12*rmesa->vb.vertex_size*4 > 
-	  rmesa->dma.current.end) {
-	 R200_NEWPRIM( rmesa );
-	 r200RefillCurrentDmaRegion( rmesa );
-      }
-
-      rmesa->vb.dmaptr = (int *)(rmesa->dma.current.address + rmesa->dma.current.ptr);
-      rmesa->vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / 
-	 (rmesa->vb.vertex_size * 4);
-      rmesa->vb.counter--;
-      rmesa->vb.initial_counter = rmesa->vb.counter;
-      rmesa->vb.notify = wrap_buffer;
-      rmesa->dma.flush = flush_prims;
-      ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;
-   }
-   
-   
-   rmesa->vb.prim[0] = mode;
-   start_prim( rmesa, mode | PRIM_BEGIN );
-}
-
-
-
-static void r200_End( void )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) {
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" );
-      return;
-   }
-
-   note_last_prim( rmesa, PRIM_END );
-   rmesa->vb.prim[0] = GL_POLYGON+1;
-}
-
-
-/* Fallback on difficult entrypoints:
- */
-#define PRE_LOOPBACK( FUNC )			\
-do {						\
-   if (R200_DEBUG & DEBUG_VFMT) 		\
-      fprintf(stderr, "%s\n", __FUNCTION__);	\
-   VFMT_FALLBACK( __FUNCTION__ );		\
-} while (0)
-#define TAG(x) r200_fallback_##x
-#include "vtxfmt_tmp.h"
-
-
-
-static GLboolean r200NotifyBegin( GLcontext *ctx, GLenum p )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-   
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(!rmesa->vb.installed);
-
-   if (ctx->NewState) 
-      _mesa_update_state( ctx );
-
-   if (rmesa->NewGLState)
-      r200ValidateState( ctx );
-
-   if (ctx->Driver.NeedFlush)
-      ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   if (rmesa->vb.recheck) 
-      r200VtxfmtValidate( ctx );
-
-   if (!rmesa->vb.installed) {
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s -- failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   r200_Begin( p );
-   return GL_TRUE;
-}
-
-static void r200VtxFmtFlushVertices( GLcontext *ctx, GLuint flags )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   if (R200_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(rmesa->vb.installed);
-
-   if (flags & FLUSH_UPDATE_CURRENT) {
-      r200_copy_to_current( ctx );
-      if (R200_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "reinstall on update_current\n");
-      _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-      ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-   }
-
-   if (flags & FLUSH_STORED_VERTICES) {
-      assert (rmesa->dma.flush == 0 ||
-	      rmesa->dma.flush == flush_prims);
-      if (rmesa->dma.flush == flush_prims)
-	 flush_prims( rmesa );
-      ctx->Driver.NeedFlush &= ~FLUSH_STORED_VERTICES;
-   }
-}
-
-
-
-/* At this point, don't expect very many versions of each function to
- * be generated, so not concerned about freeing them?
- */
-
-
-/**
- * Called once during context creation.
- */
-void r200VtxfmtInit( GLcontext *ctx, GLboolean useCodegen )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-   GLvertexformat *vfmt = &(rmesa->vb.vtxfmt);
-
-   /* start by initializing to no-op functions */
-   _mesa_noop_vtxfmt_init(vfmt);
-
-   /* Hook in chooser functions for codegen, etc:
-    */
-   r200VtxfmtInitChoosers( vfmt );
-
-   /* Handled fully in supported states, but no codegen:
-    */
-   vfmt->Materialfv = r200_Materialfv;
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
-   vfmt->Begin = r200_Begin;
-   vfmt->End = r200_End;
-
-   /* Fallback for performance reasons:  (Fix with cva/elt path here and
-    * dmatmp2.h style primitive-merging)
-    *
-    * These should call NotifyBegin(), as should _tnl_EvalMesh, to allow
-    * a driver-hook.
-    */
-   vfmt->DrawArrays = r200_fallback_DrawArrays;
-   vfmt->DrawElements = r200_fallback_DrawElements;
-   vfmt->DrawRangeElements = r200_fallback_DrawRangeElements; 
-
-   /* Active but unsupported -- fallback if we receive these:
-    */
-   vfmt->CallList = r200_fallback_CallList;
-   vfmt->CallLists = r200_fallback_CallLists;
-   vfmt->EvalCoord1f = r200_fallback_EvalCoord1f;
-   vfmt->EvalCoord1fv = r200_fallback_EvalCoord1fv;
-   vfmt->EvalCoord2f = r200_fallback_EvalCoord2f;
-   vfmt->EvalCoord2fv = r200_fallback_EvalCoord2fv;
-   vfmt->EvalMesh1 = r200_fallback_EvalMesh1;
-   vfmt->EvalMesh2 = r200_fallback_EvalMesh2;
-   vfmt->EvalPoint1 = r200_fallback_EvalPoint1;
-   vfmt->EvalPoint2 = r200_fallback_EvalPoint2;
-   vfmt->TexCoord4f = r200_fallback_TexCoord4f;
-   vfmt->TexCoord4fv = r200_fallback_TexCoord4fv;
-   vfmt->MultiTexCoord4fARB = r200_fallback_MultiTexCoord4fARB;
-   vfmt->MultiTexCoord4fvARB = r200_fallback_MultiTexCoord4fvARB;
-   vfmt->Vertex4f = r200_fallback_Vertex4f;
-   vfmt->Vertex4fv = r200_fallback_Vertex4fv;
-   vfmt->VertexAttrib1fNV  = r200_fallback_VertexAttrib1fNV;
-   vfmt->VertexAttrib1fvNV = r200_fallback_VertexAttrib1fvNV;
-   vfmt->VertexAttrib2fNV  = r200_fallback_VertexAttrib2fNV;
-   vfmt->VertexAttrib2fvNV = r200_fallback_VertexAttrib2fvNV;
-   vfmt->VertexAttrib3fNV  = r200_fallback_VertexAttrib3fNV;
-   vfmt->VertexAttrib3fvNV = r200_fallback_VertexAttrib3fvNV;
-   vfmt->VertexAttrib4fNV  = r200_fallback_VertexAttrib4fNV;
-   vfmt->VertexAttrib4fvNV = r200_fallback_VertexAttrib4fvNV;
-   vfmt->FogCoordfEXT = r200_fallback_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = r200_fallback_FogCoordfvEXT;
-   
-   (void)r200_fallback_vtxfmt;
-
-   TNL_CONTEXT(ctx)->Driver.NotifyBegin = r200NotifyBegin;
-
-   rmesa->vb.enabled = 1;
-   rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive;
-   rmesa->vb.primflags = 0;
-
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex2f );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex2fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4ub );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4ubv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3ub );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3ubv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4f );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.Normal3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Normal3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord3f );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord2f );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord2fv );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord1f );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord1fv );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord3fARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord3fvARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-/*   make_empty_list( &rmesa->vb.dfn_cache.FogCoordfEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.FogCoordfvEXT );*/
-
-   r200InitCodegen( &rmesa->vb.codegen, useCodegen );
-}
-
-static void free_funcs( struct dynfn *l )
-{
-   struct dynfn *f, *tmp;
-   foreach_s (f, tmp, l) {
-      remove_from_list( f );
-      _mesa_exec_free( f->code );
-      _mesa_free( f );
-   }
-}
-
-void r200VtxfmtUnbindContext( GLcontext *ctx )
-{
-}
-
-
-void r200VtxfmtMakeCurrent( GLcontext *ctx )
-{
-}
-
-
-void r200VtxfmtDestroy( GLcontext *ctx )
-{
-   r200ContextPtr rmesa = R200_CONTEXT( ctx );
-
-   count_funcs( rmesa );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex2f );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex2fv );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex3f );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex3fv );
-   free_funcs( &rmesa->vb.dfn_cache.Color4ub );
-   free_funcs( &rmesa->vb.dfn_cache.Color4ubv );
-   free_funcs( &rmesa->vb.dfn_cache.Color3ub );
-   free_funcs( &rmesa->vb.dfn_cache.Color3ubv );
-   free_funcs( &rmesa->vb.dfn_cache.Color4f );
-   free_funcs( &rmesa->vb.dfn_cache.Color4fv );
-   free_funcs( &rmesa->vb.dfn_cache.Color3f );
-   free_funcs( &rmesa->vb.dfn_cache.Color3fv );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   free_funcs( &rmesa->vb.dfn_cache.Normal3f );
-   free_funcs( &rmesa->vb.dfn_cache.Normal3fv );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord3f );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord3fv );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord2f );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord2fv );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord1f );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord1fv );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord3fARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord3fvARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-/*   free_funcs( &rmesa->vb.dfn_cache.FogCoordfEXT );
-   free_funcs( &rmesa->vb.dfn_cache.FogCoordfvEXT );*/
-}
-
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt.h b/src/mesa/drivers/dri/r200/r200_vtxfmt.h
deleted file mode 100644
index 46999191e04..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt.h
+++ /dev/null
@@ -1,123 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.h,v 1.1 2002/10/30 12:51:53 alanh Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#ifndef __R200_VTXFMT_H__
-#define __R200_VTXFMT_H__
-
-#include "r200_context.h"
-
-
-
-extern void r200VtxfmtUpdate( GLcontext *ctx );
-extern void r200VtxfmtInit( GLcontext *ctx, GLboolean useCodegen );
-extern void r200VtxfmtInvalidate( GLcontext *ctx );
-extern void r200VtxfmtDestroy( GLcontext *ctx );
-extern void r200VtxfmtInitChoosers( GLvertexformat *vfmt );
-
-extern void r200VtxfmtMakeCurrent( GLcontext *ctx );
-extern void r200VtxfmtUnbindContext( GLcontext *ctx );
-
-extern void r200_copy_to_current( GLcontext *ctx );
-extern void VFMT_FALLBACK( const char *caller );
-
-#define DFN( FUNC, CACHE)				\
-do {							\
-   char *start = (char *)&FUNC;				\
-   char *end = (char *)&FUNC##_end;			\
-   insert_at_head( &CACHE, dfn );			\
-   dfn->key[0] = key[0];				\
-   dfn->key[1] = key[1];				\
-   dfn->code = _mesa_exec_malloc( end - start );	\
-   _mesa_memcpy(dfn->code, start, end - start);		\
-}							\
-while ( 0 )
-
-#define FIXUP( CODE, OFFSET, CHECKVAL, NEWVAL )	\
-do {						\
-   int *icode = (int *)(CODE+OFFSET);		\
-   assert (*icode == CHECKVAL);			\
-   *icode = (int)NEWVAL;			\
-} while (0)
-
-
-/* Useful for figuring out the offsets:
- */
-#define FIXUP2( CODE, OFFSET, CHECKVAL, NEWVAL )		\
-do {								\
-   while (*(int *)(CODE+OFFSET) != CHECKVAL) OFFSET++;		\
-   /*fprintf(stderr, "%s/%d CVAL %x OFFSET %d VAL %x\n", __FUNCTION__,*/ \
-   /*	   __LINE__, CHECKVAL, OFFSET, (int)(NEWVAL));*/		\
-   *(int *)(CODE+OFFSET) = (int)(NEWVAL);				\
-   OFFSET += 4;							\
-} while (0)
-
-/* 
- */
-void r200InitCodegen( struct dfn_generators *gen, GLboolean useCodegen );
-void r200InitX86Codegen( struct dfn_generators *gen );
-void r200InitSSECodegen( struct dfn_generators *gen );
-
-
-
-/* Defined in r200_vtxfmt_x86.c
- */
-struct dynfn *r200_makeX86Vertex2f( GLcontext *, const int * );
-struct dynfn *r200_makeX86Vertex2fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86Vertex3f( GLcontext *, const int * );
-struct dynfn *r200_makeX86Vertex3fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color4ub( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color4ubv( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color3ub( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color3ubv( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color4f( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color4fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color3f( GLcontext *, const int * );
-struct dynfn *r200_makeX86Color3fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86SecondaryColor3ubEXT( GLcontext *, const int * );
-struct dynfn *r200_makeX86SecondaryColor3ubvEXT( GLcontext *, const int * );
-struct dynfn *r200_makeX86SecondaryColor3fEXT( GLcontext *, const int * );
-struct dynfn *r200_makeX86SecondaryColor3fvEXT( GLcontext *, const int * );
-struct dynfn *r200_makeX86Normal3f( GLcontext *, const int * );
-struct dynfn *r200_makeX86Normal3fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86TexCoord2f( GLcontext *, const int * );
-struct dynfn *r200_makeX86TexCoord2fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86TexCoord1f( GLcontext *, const int * );
-struct dynfn *r200_makeX86TexCoord1fv( GLcontext *, const int * );
-struct dynfn *r200_makeX86MultiTexCoord2fARB( GLcontext *, const int * );
-struct dynfn *r200_makeX86MultiTexCoord2fvARB( GLcontext *, const int * );
-struct dynfn *r200_makeX86MultiTexCoord1fARB( GLcontext *, const int * );
-struct dynfn *r200_makeX86MultiTexCoord1fvARB( GLcontext *, const int * );
-
-#endif
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c b/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c
deleted file mode 100644
index 1db5950c8ff..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c
+++ /dev/null
@@ -1,1002 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_c.c,v 1.2 2002/12/16 16:18:56 dawes Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "simple_list.h"
-#include "api_noop.h"
-#include "vtxfmt.h"
-
-#include "r200_vtxfmt.h"
-#include "r200_tcl.h"
-
-#include "dispatch.h"
-
-/* Fallback versions of all the entrypoints for situations where
- * codegen isn't available.  This is still a lot faster than the
- * vb/pipeline implementation in Mesa.
- */
-static void r200_Vertex3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&x;
-   *rmesa->vb.dmaptr++ = *(int *)&y;
-   *rmesa->vb.dmaptr++ = *(int *)&z;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void r200_Vertex3fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&v[0];
-   *rmesa->vb.dmaptr++ = *(int *)&v[1];
-   *rmesa->vb.dmaptr++ = *(int *)&v[2];
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void r200_Vertex2f( GLfloat x, GLfloat y )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   int i;
-   
-   *rmesa->vb.dmaptr++ = *(int *)&x;
-   *rmesa->vb.dmaptr++ = *(int *)&y;
-   *rmesa->vb.dmaptr++ = 0;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void r200_Vertex2fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&v[0];
-   *rmesa->vb.dmaptr++ = *(int *)&v[1];
-   *rmesa->vb.dmaptr++ = 0;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-
-/* Color for ubyte (packed) color formats:
- */
-#if 0
-static void r200_Color3ub_ub( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= 0xff;
-}
-
-static void r200_Color3ubv_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= v[0];
-   dest->green	= v[1];
-   dest->blue	= v[2];
-   dest->alpha	= 0xff;
-}
-
-static void r200_Color4ub_ub( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= a;
-}
-
-static void r200_Color4ubv_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   *(GLuint *)rmesa->vb.colorptr = LE32_TO_CPU(*(GLuint *)v);
-}
-#endif /* 0 */
-
-static void r200_Color3f_ub( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   dest->alpha = 255;
-}
-
-static void r200_Color3fv_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   dest->alpha = 255;
-}
-
-static void r200_Color4f_ub( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->alpha, a );
-}
-
-static void r200_Color4fv_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->alpha, v[3] );
-}
-
-
-/* Color for float color+alpha formats:
- */
-#if 0
-static void r200_Color3ub_4f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = 1.0;
-}
-
-static void r200_Color3ubv_4f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = 1.0;
-}
-
-static void r200_Color4ub_4f( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = UBYTE_TO_FLOAT(a);
-}
-
-static void r200_Color4ubv_4f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = UBYTE_TO_FLOAT(v[3]);
-}
-#endif /* 0 */
-
-
-static void r200_Color3f_4f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = 1.0;		
-}
-
-static void r200_Color3fv_4f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = 1.0;
-}
-
-static void r200_Color4f_4f( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = a;
-}
-
-static void r200_Color4fv_4f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = v[3];
-}
-
-
-/* Color for float color formats:
- */
-#if 0
-static void r200_Color3ub_3f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-}
-
-static void r200_Color3ubv_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-}
-
-static void r200_Color4ub_3f( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT(a);
-}
-
-static void r200_Color4ubv_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT(v[3]);
-}
-#endif /* 0 */
-
-
-static void r200_Color3f_3f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-}
-
-static void r200_Color3fv_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-}
-
-static void r200_Color4f_3f( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = a;
-}
-
-static void r200_Color4fv_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = v[3]; 
-}
-
-
-/* Secondary Color:
- */
-#if 0
-static void r200_SecondaryColor3ubEXT_ub( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.specptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= 0xff;
-}
-
-static void r200_SecondaryColor3ubvEXT_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.specptr;
-   dest->red	= v[0];
-   dest->green	= v[1];
-   dest->blue	= v[2];
-   dest->alpha	= 0xff;
-}
-#endif /* 0 */
-
-static void r200_SecondaryColor3fEXT_ub( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.specptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   dest->alpha = 255;
-}
-
-static void r200_SecondaryColor3fvEXT_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   r200_color_t *dest = rmesa->vb.specptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   dest->alpha = 255;
-}
-
-#if 0
-static void r200_SecondaryColor3ubEXT_3f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = 1.0;
-}
-
-static void r200_SecondaryColor3ubvEXT_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = 1.0;
-}
-#endif /* 0 */
-
-static void r200_SecondaryColor3fEXT_3f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = 1.0;
-}
-
-static void r200_SecondaryColor3fvEXT_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = 1.0;
-}
-
-
-
-/* Normal
- */
-static void r200_Normal3f( GLfloat n0, GLfloat n1, GLfloat n2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.normalptr;
-   dest[0] = n0;
-   dest[1] = n1;
-   dest[2] = n2;
-}
-
-static void r200_Normal3fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.normalptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-}
-
-
-/* FogCoord
- */
-static void r200_FogCoordfEXT( GLfloat f )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.fogptr;
-   dest[0] = r200ComputeFogBlendFactor( ctx, f );
-/*   ctx->Current.Attrib[VERT_ATTRIB_FOG][0] = f;*/
-}
-
-static void r200_FogCoordfvEXT( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.fogptr;
-   dest[0] = r200ComputeFogBlendFactor( ctx, v[0] );
-/*   ctx->Current.Attrib[VERT_ATTRIB_FOG][0] = v[0];*/
-}
-
-
-/* TexCoord
- */
-
-/* \todo maybe (target & 4 ? target & 5 : target & 3) is more save than (target & 7) */
-static void r200_MultiTexCoord1fARB(GLenum target, GLfloat s)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLint unit = (target & 7);
-   GLfloat * const dest = rmesa->vb.texcoordptr[unit];
-
-   switch( ctx->Texture.Unit[unit]._ReallyEnabled ) {
-   case TEXTURE_CUBE_BIT:
-   case TEXTURE_3D_BIT:
-      dest[2] = 0.0;
-      /* FALLTHROUGH */
-   case TEXTURE_2D_BIT:
-   case TEXTURE_RECT_BIT:
-      dest[1] = 0.0;
-      /* FALLTHROUGH */
-   case TEXTURE_1D_BIT:
-      dest[0] = s;
-   }
-}
-
-static void r200_MultiTexCoord2fARB(GLenum target, GLfloat s, GLfloat t)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLint unit = (target & 7);
-   GLfloat * const dest = rmesa->vb.texcoordptr[unit];
-
-   switch( ctx->Texture.Unit[unit]._ReallyEnabled ) {
-   case TEXTURE_CUBE_BIT:
-   case TEXTURE_3D_BIT:
-      dest[2] = 0.0;
-      /* FALLTHROUGH */
-   case TEXTURE_2D_BIT:
-   case TEXTURE_RECT_BIT:
-      dest[1] = t;
-      dest[0] = s;
-      break;
-   default:
-      VFMT_FALLBACK(__FUNCTION__);
-      CALL_MultiTexCoord2fARB(GET_DISPATCH(), (target, s, t));
-      return;	
-   }
-}
-
-static void r200_MultiTexCoord3fARB(GLenum target, GLfloat s, GLfloat t, GLfloat r)
-{
-   GET_CURRENT_CONTEXT(ctx);
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLint unit = (target & 7);
-   GLfloat * const dest = rmesa->vb.texcoordptr[unit];
-
-   switch( ctx->Texture.Unit[unit]._ReallyEnabled ) {
-   case TEXTURE_CUBE_BIT:
-   case TEXTURE_3D_BIT:
-      dest[2] = r;
-      dest[1] = t;
-      dest[0] = s;
-      break;
-   default:
-      VFMT_FALLBACK(__FUNCTION__);
-      CALL_MultiTexCoord3fARB(GET_DISPATCH(), (target, s, t, r));
-      return;	
-   }
-}
-
-static void r200_TexCoord1f(GLfloat s)
-{
-   r200_MultiTexCoord1fARB(GL_TEXTURE0, s);
-}
-
-static void r200_TexCoord2f(GLfloat s, GLfloat t)
-{
-   r200_MultiTexCoord2fARB(GL_TEXTURE0, s, t);
-}
-
-static void r200_TexCoord3f(GLfloat s, GLfloat t, GLfloat r)
-{
-   r200_MultiTexCoord3fARB(GL_TEXTURE0, s, t, r);
-}
-
-static void r200_TexCoord1fv(const GLfloat *v)
-{
-   r200_MultiTexCoord1fARB(GL_TEXTURE0, v[0]);
-}
-
-static void r200_TexCoord2fv(const GLfloat *v)
-{
-   r200_MultiTexCoord2fARB(GL_TEXTURE0, v[0], v[1]);
-}
-
-static void r200_TexCoord3fv(const GLfloat *v)
-{
-   r200_MultiTexCoord3fARB(GL_TEXTURE0, v[0], v[1], v[2]);
-}
-
-static void r200_MultiTexCoord1fvARB(GLenum target, const GLfloat *v)
-{
-   r200_MultiTexCoord1fARB(target, v[0]);
-}
-
-static void r200_MultiTexCoord2fvARB(GLenum target, const GLfloat *v)
-{
-   r200_MultiTexCoord2fARB(target, v[0], v[1]);
-}
-
-static void r200_MultiTexCoord3fvARB(GLenum target, const GLfloat *v)
-{
-   r200_MultiTexCoord3fARB(target, v[0], v[1], v[2]);
-}
-
-
-static struct dynfn *lookup( struct dynfn *l, const int *key )
-{
-   struct dynfn *f;
-
-   foreach( f, l ) {
-      if (f->key[0] == key[0] && f->key[1] == key[1]) 
-	 return f;
-   }
-
-   return NULL;
-}
-
-/* Can't use the loopback template for this:
- */
-
-#define CHOOSE(FN, FNTYPE, MASK0, MASK1, ARGS1, ARGS2 )			\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx);						\
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);				\
-   int key[2];								\
-   struct dynfn *dfn;							\
-									\
-   key[0] = rmesa->vb.vtxfmt_0 & MASK0;					\
-   key[1] = rmesa->vb.vtxfmt_1 & MASK1;					\
-									\
-   dfn = lookup( &rmesa->vb.dfn_cache.FN, key );			\
-   if (dfn == 0)							\
-      dfn = rmesa->vb.codegen.FN( ctx, key );				\
-   else if (R200_DEBUG & DEBUG_CODEGEN)					\
-      fprintf(stderr, "%s -- cached codegen\n", __FUNCTION__ );		\
-									\
-   if (dfn)								\
-      SET_ ## FN (ctx->Exec, (FNTYPE)(dfn->code));			\
-   else {								\
-      if (R200_DEBUG & DEBUG_CODEGEN)					\
-	 fprintf(stderr, "%s -- generic version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, r200_##FN);				\
-   }									\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;			\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-/* For the _3f case, only allow one color function to be hooked in at
- * a time.  Eventually, use a similar mechanism to allow selecting the
- * color component of the vertex format based on client behaviour.  
- *
- * Note:  Perform these actions even if there is a codegen or cached 
- * codegen version of the chosen function.
- */
-#define CHOOSE_COLOR(FN, FNTYPE, NR, MASK0, MASK1, ARGS1, ARGS2 )	\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx);						\
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);			\
-   int key[2];								\
-   struct dynfn *dfn;							\
-									\
-   key[0] = rmesa->vb.vtxfmt_0 & MASK0;					\
-   key[1] = rmesa->vb.vtxfmt_1 & MASK1;					\
-									\
-   if (VTX_COLOR(rmesa->vb.vtxfmt_0,0) == R200_VTX_PK_RGBA) {		\
-      SET_ ## FN (ctx->Exec, r200_##FN##_ub);				\
-   }									\
-   else if (VTX_COLOR(rmesa->vb.vtxfmt_0,0) == R200_VTX_FP_RGB) {	\
-									\
-      if (rmesa->vb.installed_color_3f_sz != NR) {			\
-         rmesa->vb.installed_color_3f_sz = NR;				\
-         if (NR == 3) ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = 1.0;	\
-         if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) {		\
-            r200_copy_to_current( ctx );				\
-            _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );	\
-            CALL_ ## FN (ctx->Exec, ARGS2);				\
-            return;							\
-         }								\
-      }									\
-									\
-      SET_ ## FN (ctx->Exec, r200_##FN##_3f);				\
-   }									\
-   else {								\
-      SET_ ## FN (ctx->Exec, r200_##FN##_4f);				\
-   }									\
-									\
-									\
-   dfn = lookup( &rmesa->vb.dfn_cache.FN, key );			\
-   if (!dfn) dfn = rmesa->vb.codegen.FN( ctx, key );			\
-									\
-   if (dfn) {								\
-      if (R200_DEBUG & DEBUG_CODEGEN)					\
-         fprintf(stderr, "%s -- codegen version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, (FNTYPE)dfn->code);			\
-   }									\
-   else if (R200_DEBUG & DEBUG_CODEGEN)					\
-         fprintf(stderr, "%s -- 'c' version\n", __FUNCTION__ );		\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;			\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-/* Right now there are both _ub and _3f versions of the secondary color
- * functions.  Currently, we only set-up the hardware to use the _ub versions.
- * The _3f versions are needed for the cases where secondary color isn't used
- * in the vertex format, but it still needs to be stored in the context
- * state vector.
- */
-#define CHOOSE_SECONDARY_COLOR(FN, FNTYPE, MASK0, MASK1, ARGS1, ARGS2 )	\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx);						\
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);			\
-   int key[2];								\
-   struct dynfn *dfn;							\
-									\
-   key[0] = rmesa->vb.vtxfmt_0 & MASK0;					\
-   key[1] = rmesa->vb.vtxfmt_1 & MASK1;					\
-									\
-   dfn = lookup( &rmesa->vb.dfn_cache.FN, key );			\
-   if (dfn == 0)							\
-      dfn = rmesa->vb.codegen.FN( ctx, key );			\
-   else  if (R200_DEBUG & DEBUG_CODEGEN)				\
-      fprintf(stderr, "%s -- cached version\n", __FUNCTION__ );		\
-									\
-   if (dfn)								\
-      SET_ ## FN (ctx->Exec, (FNTYPE)(dfn->code));			\
-   else {								\
-      if (R200_DEBUG & DEBUG_CODEGEN)					\
-         fprintf(stderr, "%s -- generic version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, (VTX_COLOR(rmesa->vb.vtxfmt_0,1) == R200_VTX_PK_RGBA) \
-	  ? r200_##FN##_ub : r200_##FN##_3f);				\
-   }									\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;		\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-
-
-
-
-/* VTXFMT_0
- */
-#define MASK_XYZW  (R200_VTX_W0|R200_VTX_Z0)
-#define MASK_NORM  (MASK_XYZW|R200_VTX_N0)
-#define MASK_FOG   (MASK_NORM |R200_VTX_DISCRETE_FOG)
-#define MASK_COLOR (MASK_FOG |(R200_VTX_COLOR_MASK<<R200_VTX_COLOR_0_SHIFT))
-#define MASK_SPEC  (MASK_COLOR|(R200_VTX_COLOR_MASK<<R200_VTX_COLOR_1_SHIFT))
-
-/* VTXFMT_1
- */
-#define MASK_ST0 (0x7 << R200_VTX_TEX0_COMP_CNT_SHIFT)
-/* FIXME: maybe something like in the radeon driver is needed here? */
-
-
-typedef void (*p4f)( GLfloat, GLfloat, GLfloat, GLfloat );
-typedef void (*p3f)( GLfloat, GLfloat, GLfloat );
-typedef void (*p2f)( GLfloat, GLfloat );
-typedef void (*p1f)( GLfloat );
-typedef void (*pe3f)( GLenum, GLfloat, GLfloat, GLfloat );
-typedef void (*pe2f)( GLenum, GLfloat, GLfloat );
-typedef void (*pe1f)( GLenum, GLfloat );
-typedef void (*p4ub)( GLubyte, GLubyte, GLubyte, GLubyte );
-typedef void (*p3ub)( GLubyte, GLubyte, GLubyte );
-typedef void (*pfv)( const GLfloat * );
-typedef void (*pefv)( GLenum, const GLfloat * );
-typedef void (*pubv)( const GLubyte * );
-
-
-CHOOSE(Normal3f, p3f, MASK_NORM, 0, 
-       (GLfloat a,GLfloat b,GLfloat c), (a,b,c))
-CHOOSE(Normal3fv, pfv, MASK_NORM, 0, 
-       (const GLfloat *v), (v))
-
-#if 0
-CHOOSE_COLOR(Color4ub, p4ub, 4, MASK_COLOR, 0,
-	(GLubyte a,GLubyte b, GLubyte c, GLubyte d), (a,b,c,d))
-CHOOSE_COLOR(Color4ubv, pubv, 4, MASK_COLOR, 0, 
-	(const GLubyte *v), (v))
-CHOOSE_COLOR(Color3ub, p3ub, 3, MASK_COLOR, 0, 
-	(GLubyte a,GLubyte b, GLubyte c), (a,b,c))
-CHOOSE_COLOR(Color3ubv, pubv, 3, MASK_COLOR, 0, 
-	(const GLubyte *v), (v))
-CHOOSE_SECONDARY_COLOR(SecondaryColor3ubEXT, p3ub, MASK_SPEC, 0, 
-	(GLubyte a,GLubyte b, GLubyte c), (a,b,c))
-CHOOSE_SECONDARY_COLOR(SecondaryColor3ubvEXT, pubv, MASK_SPEC, 0, 
-	(const GLubyte *v), (v))
-#endif
-
-CHOOSE_COLOR(Color4f, p4f, 4, MASK_COLOR, 0, 
-	(GLfloat a,GLfloat b, GLfloat c, GLfloat d), (a,b,c,d))
-CHOOSE_COLOR(Color4fv, pfv, 4, MASK_COLOR, 0, 
-	(const GLfloat *v), (v))
-CHOOSE_COLOR(Color3f, p3f, 3, MASK_COLOR, 0,
-	(GLfloat a,GLfloat b, GLfloat c), (a,b,c))
-CHOOSE_COLOR(Color3fv, pfv, 3, MASK_COLOR, 0,
-	(const GLfloat *v), (v))
-
-
-CHOOSE_SECONDARY_COLOR(SecondaryColor3fEXT, p3f, MASK_SPEC, 0,
-	(GLfloat a,GLfloat b, GLfloat c), (a,b,c))
-CHOOSE_SECONDARY_COLOR(SecondaryColor3fvEXT, pfv, MASK_SPEC, 0,
-	(const GLfloat *v), (v))
-
-CHOOSE(TexCoord3f, p3f, ~0, MASK_ST0, 
-       (GLfloat a,GLfloat b,GLfloat c), (a,b,c))
-CHOOSE(TexCoord3fv, pfv, ~0, MASK_ST0, 
-       (const GLfloat *v), (v))
-CHOOSE(TexCoord2f, p2f, ~0, MASK_ST0, 
-       (GLfloat a,GLfloat b), (a,b))
-CHOOSE(TexCoord2fv, pfv, ~0, MASK_ST0, 
-       (const GLfloat *v), (v))
-CHOOSE(TexCoord1f, p1f, ~0, MASK_ST0, 
-       (GLfloat a), (a))
-CHOOSE(TexCoord1fv, pfv, ~0, MASK_ST0, 
-       (const GLfloat *v), (v))
-
-CHOOSE(MultiTexCoord3fARB, pe3f, ~0, ~0,
-	 (GLenum u,GLfloat a,GLfloat b,GLfloat c), (u,a,b,c))
-CHOOSE(MultiTexCoord3fvARB, pefv, ~0, ~0,
-	(GLenum u,const GLfloat *v), (u,v))
-CHOOSE(MultiTexCoord2fARB, pe2f, ~0, ~0,
-	 (GLenum u,GLfloat a,GLfloat b), (u,a,b))
-CHOOSE(MultiTexCoord2fvARB, pefv, ~0, ~0,
-	(GLenum u,const GLfloat *v), (u,v))
-CHOOSE(MultiTexCoord1fARB, pe1f, ~0, ~0,
-	 (GLenum u,GLfloat a), (u,a))
-CHOOSE(MultiTexCoord1fvARB, pefv, ~0, ~0,
-	(GLenum u,const GLfloat *v), (u,v))
-
-CHOOSE(Vertex3f, p3f, ~0, ~0, 
-       (GLfloat a,GLfloat b,GLfloat c), (a,b,c))
-CHOOSE(Vertex3fv, pfv, ~0, ~0, 
-       (const GLfloat *v), (v))
-CHOOSE(Vertex2f, p2f, ~0, ~0, 
-       (GLfloat a,GLfloat b), (a,b))
-CHOOSE(Vertex2fv, pfv, ~0, ~0, 
-       (const GLfloat *v), (v))
-
-CHOOSE(FogCoordfEXT, p1f, MASK_FOG, ~0, 
-       (GLfloat f), (f))
-CHOOSE(FogCoordfvEXT, pfv, MASK_FOG, ~0, 
-       (const GLfloat *f), (f))
-
-
-
-
-void r200VtxfmtInitChoosers( GLvertexformat *vfmt )
-{
-   vfmt->Color3f = choose_Color3f;
-   vfmt->Color3fv = choose_Color3fv;
-   vfmt->Color4f = choose_Color4f;
-   vfmt->Color4fv = choose_Color4fv;
-   vfmt->SecondaryColor3fEXT = choose_SecondaryColor3fEXT;
-   vfmt->SecondaryColor3fvEXT = choose_SecondaryColor3fvEXT;
-   vfmt->MultiTexCoord1fARB = choose_MultiTexCoord1fARB;
-   vfmt->MultiTexCoord1fvARB = choose_MultiTexCoord1fvARB;
-   vfmt->MultiTexCoord2fARB = choose_MultiTexCoord2fARB;
-   vfmt->MultiTexCoord2fvARB = choose_MultiTexCoord2fvARB;
-   vfmt->MultiTexCoord3fARB = choose_MultiTexCoord3fARB;
-   vfmt->MultiTexCoord3fvARB = choose_MultiTexCoord3fvARB;
-   vfmt->Normal3f = choose_Normal3f;
-   vfmt->Normal3fv = choose_Normal3fv;
-   vfmt->TexCoord1f = choose_TexCoord1f;
-   vfmt->TexCoord1fv = choose_TexCoord1fv;
-   vfmt->TexCoord2f = choose_TexCoord2f;
-   vfmt->TexCoord2fv = choose_TexCoord2fv;
-   vfmt->TexCoord3f = choose_TexCoord3f;
-   vfmt->TexCoord3fv = choose_TexCoord3fv;
-   vfmt->Vertex2f = choose_Vertex2f;
-   vfmt->Vertex2fv = choose_Vertex2fv;
-   vfmt->Vertex3f = choose_Vertex3f;
-   vfmt->Vertex3fv = choose_Vertex3fv;
-/*   vfmt->FogCoordfEXT = choose_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = choose_FogCoordfvEXT;*/
-
-   /* TODO: restore ubyte colors to vtxfmt.
-    */
-#if 0
-   vfmt->Color3ub = choose_Color3ub;
-   vfmt->Color3ubv = choose_Color3ubv;
-   vfmt->Color4ub = choose_Color4ub;
-   vfmt->Color4ubv = choose_Color4ubv;
-   vfmt->SecondaryColor3ubEXT = choose_SecondaryColor3ubEXT;
-   vfmt->SecondaryColor3ubvEXT = choose_SecondaryColor3ubvEXT;
-#endif
-}
-
-
-static struct dynfn *codegen_noop( GLcontext *ctx, const int *key )
-{
-   (void) ctx; (void) key;
-   return NULL;
-}
-
-void r200InitCodegen( struct dfn_generators *gen, GLboolean useCodegen )
-{
-   gen->Vertex3f = codegen_noop;
-   gen->Vertex3fv = codegen_noop;
-   gen->Color4ub = codegen_noop;
-   gen->Color4ubv = codegen_noop;
-   gen->Normal3f = codegen_noop;
-   gen->Normal3fv = codegen_noop;
-
-   gen->TexCoord3f = codegen_noop;
-   gen->TexCoord3fv = codegen_noop;
-   gen->TexCoord2f = codegen_noop;
-   gen->TexCoord2fv = codegen_noop;
-   gen->TexCoord1f = codegen_noop;
-   gen->TexCoord1fv = codegen_noop;
-
-   gen->MultiTexCoord3fARB = codegen_noop;
-   gen->MultiTexCoord3fvARB = codegen_noop;
-   gen->MultiTexCoord2fARB = codegen_noop;
-   gen->MultiTexCoord2fvARB = codegen_noop;
-   gen->MultiTexCoord1fARB = codegen_noop;
-   gen->MultiTexCoord1fvARB = codegen_noop;
-/*   gen->FogCoordfEXT = codegen_noop;
-   gen->FogCoordfvEXT = codegen_noop;*/
-
-   gen->Vertex2f = codegen_noop;
-   gen->Vertex2fv = codegen_noop;
-   gen->Color3ub = codegen_noop;
-   gen->Color3ubv = codegen_noop;
-   gen->Color4f = codegen_noop;
-   gen->Color4fv = codegen_noop;
-   gen->Color3f = codegen_noop;
-   gen->Color3fv = codegen_noop;
-   gen->SecondaryColor3fEXT = codegen_noop;
-   gen->SecondaryColor3fvEXT = codegen_noop;
-   gen->SecondaryColor3ubEXT = codegen_noop;
-   gen->SecondaryColor3ubvEXT = codegen_noop;
-
-   if (useCodegen) {
-#if defined(USE_X86_ASM)
-      r200InitX86Codegen( gen );
-#endif
-
-#if defined(USE_SSE_ASM)
-      r200InitSSECodegen( gen );
-#endif
-   }
-}
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt_sse.c b/src/mesa/drivers/dri/r200/r200_vtxfmt_sse.c
deleted file mode 100644
index 5901730494a..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt_sse.c
+++ /dev/null
@@ -1,234 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_sse.c,v 1.1 2002/10/30 12:51:53 alanh Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "imports.h"
-#include "simple_list.h" 
-#include "r200_vtxfmt.h"
-
-#if defined(USE_SSE_ASM)
-#include "x86/common_x86_asm.h"
-
-#define EXTERN( FUNC )		\
-extern const char *FUNC;	\
-extern const char *FUNC##_end
-
-EXTERN( _sse_Attribute2fv );
-EXTERN( _sse_Attribute2f );
-EXTERN( _sse_Attribute3fv );
-EXTERN( _sse_Attribute3f );
-EXTERN( _sse_MultiTexCoord2fv );
-EXTERN( _sse_MultiTexCoord2f );
-EXTERN( _sse_MultiTexCoord2fv_2 );
-EXTERN( _sse_MultiTexCoord2f_2 );
-
-/* Build specialized versions of the immediate calls on the fly for
- * the current state.
- */
-
-static struct dynfn *r200_makeSSEAttribute2fv( struct dynfn * cache, const int * key,
-					       const char * name, void * dest)
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _sse_Attribute2fv, (*cache) );
-   FIXUP(dfn->code, 10, 0x0, (int)dest);
-   return dfn;
-}
-
-static struct dynfn *r200_makeSSEAttribute2f( struct dynfn * cache, const int * key,
-					      const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _sse_Attribute2f, (*cache) );
-   FIXUP(dfn->code, 8, 0x0, (int)dest); 
-   return dfn;
-}
-
-static struct dynfn *r200_makeSSEAttribute3fv( struct dynfn * cache, const int * key,
-					       const char * name, void * dest)
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _sse_Attribute3fv, (*cache) );
-   FIXUP(dfn->code, 13, 0x0, (int)dest);
-   FIXUP(dfn->code, 18, 0x8, 8+(int)dest);
-   return dfn;
-}
-
-static struct dynfn *r200_makeSSEAttribute3f( struct dynfn * cache, const int * key,
-					      const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _sse_Attribute3f, (*cache) );
-   FIXUP(dfn->code, 12, 0x0, (int)dest); 
-   FIXUP(dfn->code, 17, 0x8, 8+(int)dest); 
-   return dfn;
-}
-
-static struct dynfn *r200_makeSSENormal3fv( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeSSEAttribute3fv( & rmesa->vb.dfn_cache.Normal3fv, key,
-				    __FUNCTION__, rmesa->vb.normalptr );
-}
-
-static struct dynfn *r200_makeSSENormal3f( GLcontext *ctx, const int * key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeSSEAttribute3f( & rmesa->vb.dfn_cache.Normal3f, key,
-				   __FUNCTION__, rmesa->vb.normalptr );
-}
-
-static struct dynfn *r200_makeSSEColor3fv( GLcontext *ctx, const int * key )
-{
-   if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) 
-      return NULL;
-   else
-   {
-      r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-      return r200_makeSSEAttribute3fv( & rmesa->vb.dfn_cache.Color3fv, key,
-				       __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-static struct dynfn *r200_makeSSEColor3f( GLcontext *ctx, const int * key )
-{
-   if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) 
-      return NULL;
-   else
-   {
-      r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-      return r200_makeSSEAttribute3f( & rmesa->vb.dfn_cache.Color3f, key,
-				      __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-#if 0 /* Temporarily disabled as it is broken w/the new cubemap code. - idr */
-static struct dynfn *r200_makeSSETexCoord2fv( GLcontext *ctx, const int * key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeSSEAttribute2fv( & rmesa->vb.dfn_cache.TexCoord2fv, key,
-				    __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-static struct dynfn *r200_makeSSETexCoord2f( GLcontext *ctx, const int * key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeSSEAttribute2f( & rmesa->vb.dfn_cache.TexCoord2f, key,
-				   __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-static struct dynfn *r200_makeSSEMultiTexCoord2fv( GLcontext *ctx, const int * key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] );
-
-   if (rmesa->vb.texcoordptr[1] == rmesa->vb.texcoordptr[0]+4) {
-      DFN ( _sse_MultiTexCoord2fv, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 18, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);	
-   } else {
-      DFN ( _sse_MultiTexCoord2fv_2, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 14, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-
-static struct dynfn *r200_makeSSEMultiTexCoord2f( GLcontext *ctx, const int * key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] );
-
-   if (rmesa->vb.texcoordptr[1] == rmesa->vb.texcoordptr[0]+4) {
-      DFN ( _sse_MultiTexCoord2f, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 16, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);	
-   } else {
-      DFN ( _sse_MultiTexCoord2f_2, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 15, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-#endif
-
-void r200InitSSECodegen( struct dfn_generators *gen )
-{
-   if ( cpu_has_xmm ) {
-      gen->Normal3fv = (void *) r200_makeSSENormal3fv;
-      gen->Normal3f = (void *) r200_makeSSENormal3f;
-      gen->Color3fv = (void *) r200_makeSSEColor3fv;
-      gen->Color3f = (void *) r200_makeSSEColor3f;
-#if 0 /* Temporarily disabled as it is broken w/the new cubemap code. - idr */
-      gen->TexCoord2fv = (void *) r200_makeSSETexCoord2fv;
-      gen->TexCoord2f = (void *) r200_makeSSETexCoord2f;
-      gen->MultiTexCoord2fvARB = (void *) r200_makeSSEMultiTexCoord2fv;
-      gen->MultiTexCoord2fARB = (void *) r200_makeSSEMultiTexCoord2f;
-#endif
-   }
-}
-
-#else 
-
-void r200InitSSECodegen( struct dfn_generators *gen )
-{
-   (void) gen;
-}
-
-#endif
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt_x86.c b/src/mesa/drivers/dri/r200/r200_vtxfmt_x86.c
deleted file mode 100644
index b78a55c31f6..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt_x86.c
+++ /dev/null
@@ -1,440 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_x86.c,v 1.2 2002/12/16 16:18:56 dawes Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#include "glheader.h"
-#include "imports.h"
-#include "simple_list.h" 
-#include "r200_vtxfmt.h"
-
-#if defined(USE_X86_ASM)
-
-#define EXTERN( FUNC )		\
-extern const char *FUNC;	\
-extern const char *FUNC##_end
-
-EXTERN ( _x86_Attribute2fv );
-EXTERN ( _x86_Attribute2f );
-EXTERN ( _x86_Attribute3fv );
-EXTERN ( _x86_Attribute3f );
-EXTERN ( _x86_Vertex3fv_6 );
-EXTERN ( _x86_Vertex3fv_8 );
-EXTERN ( _x86_Vertex3fv );
-EXTERN ( _x86_Vertex3f_4 );
-EXTERN ( _x86_Vertex3f_6 );
-EXTERN ( _x86_Vertex3f );
-EXTERN ( _x86_Color4ubv_ub );
-EXTERN ( _x86_Color4ubv_4f );
-EXTERN ( _x86_Color4ub_ub );
-EXTERN ( _x86_MultiTexCoord2fv );
-EXTERN ( _x86_MultiTexCoord2fv_2 );
-EXTERN ( _x86_MultiTexCoord2f );
-EXTERN ( _x86_MultiTexCoord2f_2 );
-
-
-/* Build specialized versions of the immediate calls on the fly for
- * the current state.  Generic x86 versions.
- */
-
-struct dynfn *r200_makeX86Vertex3f( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x 0x%08x %d\n", __FUNCTION__, 
-	      key[0], key[1], rmesa->vb.vertex_size );
-
-   switch (rmesa->vb.vertex_size) {
-   case 4: {
-
-      DFN ( _x86_Vertex3f_4, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 2, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 25, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 36, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 46, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 51, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 60, 0x0, (int)&rmesa->vb.notify);
-      break;
-   }
-   case 6: {
-
-      DFN ( _x86_Vertex3f_6, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 3, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 28, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 34, 0x0, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 40, 0x0, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 57, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 63, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 70, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 79, 0x0, (int)&rmesa->vb.notify);
-      break;
-   }
-   default: {
-
-      DFN ( _x86_Vertex3f, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 3, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 9, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 37, 0x0, rmesa->vb.vertex_size-3);
-      FIXUP(dfn->code, 44, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 50, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 56, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x0, (int)&rmesa->vb.notify);
-   break;
-   }
-   }
-
-   return dfn;
-}
-
-
-
-struct dynfn *r200_makeX86Vertex3fv( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x 0x%08x %d\n", __FUNCTION__, 
-	      key[0], key[1], rmesa->vb.vertex_size );
-
-   switch (rmesa->vb.vertex_size) {
-   case 6: {
-
-      DFN ( _x86_Vertex3fv_6, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 1, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 27, 0x0000001c, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 33, 0x00000020, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 45, 0x00000024, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 56, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 61, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 76, 0x00000008, (int)&rmesa->vb.notify);
-      break;
-   }
-   
-
-   case 8: {
-
-      DFN ( _x86_Vertex3fv_8, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 1, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 27, 0x0000001c, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 33, 0x00000020, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 45, 0x0000001c, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 51, 0x00000020, (int)&rmesa->vb.vertex[6]);
-      FIXUP(dfn->code, 63, 0x00000024, (int)&rmesa->vb.vertex[7]);
-      FIXUP(dfn->code, 74, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 79, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 85, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 94, 0x00000008, (int)&rmesa->vb.notify);
-      break;
-   }
-   
-
-
-   default: {
-
-      DFN ( _x86_Vertex3fv, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 8, 0x01010101, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 32, 0x00000006, rmesa->vb.vertex_size-3);
-      FIXUP(dfn->code, 37, 0x00000058, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 45, 0x01010101, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 50, 0x02020202, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 58, 0x02020202, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x0, (int)&rmesa->vb.notify);
-   break;
-   }
-   }
-
-   return dfn;
-}
-
-static struct dynfn *
-r200_makeX86Attribute2fv( struct dynfn * cache, const int *key,
-			  const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _x86_Attribute2fv, (*cache) );
-   FIXUP(dfn->code, 11, 0x0, (int)dest); 
-   FIXUP(dfn->code, 16, 0x4, 4+(int)dest); 
-
-   return dfn;
-}
-
-static struct dynfn *
-r200_makeX86Attribute2f( struct dynfn * cache, const int *key,
-			 const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _x86_Attribute2f, (*cache) );
-   FIXUP(dfn->code, 1, 0x0, (int)dest); 
-
-   return dfn;
-}
-
-
-static struct dynfn *
-r200_makeX86Attribute3fv( struct dynfn * cache, const int *key,
-			  const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _x86_Attribute3fv, (*cache) );
-   FIXUP(dfn->code, 14, 0x0, (int)dest); 
-   FIXUP(dfn->code, 20, 0x4, 4+(int)dest); 
-   FIXUP(dfn->code, 25, 0x8, 8+(int)dest);
-
-   return dfn;
-}
-
-static struct dynfn *
-r200_makeX86Attribute3f( struct dynfn * cache, const int *key,
-			 const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key[0] );
-
-   DFN ( _x86_Attribute3f, (*cache) );
-   FIXUP(dfn->code, 14, 0x0, (int)dest); 
-   FIXUP(dfn->code, 20, 0x4, 4+(int)dest); 
-   FIXUP(dfn->code, 25, 0x8, 8+(int)dest);
-
-   return dfn;
-}
-
-struct dynfn *r200_makeX86Normal3fv( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeX86Attribute3fv( & rmesa->vb.dfn_cache.Normal3fv, key,
-				    __FUNCTION__, rmesa->vb.normalptr );
-}
-
-struct dynfn *r200_makeX86Normal3f( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeX86Attribute3f( & rmesa->vb.dfn_cache.Normal3f, key,
-				   __FUNCTION__, rmesa->vb.normalptr );
-}
-
-struct dynfn *r200_makeX86Color4ubv( GLcontext *ctx, const int *key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] );
-
-   if (VTX_COLOR(key[0],0) == R200_VTX_PK_RGBA) {
-      DFN ( _x86_Color4ubv_ub, rmesa->vb.dfn_cache.Color4ubv);
-      FIXUP(dfn->code, 5, 0x12345678, (int)rmesa->vb.colorptr); 
-      return dfn;
-   } 
-   else {
-
-      DFN ( _x86_Color4ubv_4f, rmesa->vb.dfn_cache.Color4ubv);
-      FIXUP(dfn->code, 2, 0x00000000, (int)_mesa_ubyte_to_float_color_tab); 
-      FIXUP(dfn->code, 27, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr); 
-      FIXUP(dfn->code, 33, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+4); 
-      FIXUP(dfn->code, 55, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+8); 
-      FIXUP(dfn->code, 61, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+12); 
-      return dfn;
-   }
-}
-
-struct dynfn *r200_makeX86Color4ub( GLcontext *ctx, const int *key )
-{
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] );
-
-   if (VTX_COLOR(key[0],0) == R200_VTX_PK_RGBA) {
-      struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-      r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-      DFN ( _x86_Color4ub_ub, rmesa->vb.dfn_cache.Color4ub );
-      FIXUP(dfn->code, 18, 0x0, (int)rmesa->vb.colorptr); 
-      FIXUP(dfn->code, 24, 0x0, (int)rmesa->vb.colorptr+1); 
-      FIXUP(dfn->code, 30, 0x0, (int)rmesa->vb.colorptr+2); 
-      FIXUP(dfn->code, 36, 0x0, (int)rmesa->vb.colorptr+3); 
-      return dfn;
-   }
-   else
-      return NULL;
-}
-
-
-struct dynfn *r200_makeX86Color3fv( GLcontext *ctx, const int *key )
-{
-   if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) 
-      return NULL;
-   else
-   {
-      r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-      return r200_makeX86Attribute3fv( & rmesa->vb.dfn_cache.Color3fv, key,
-				       __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-struct dynfn *r200_makeX86Color3f( GLcontext *ctx, const int *key )
-{
-   if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) 
-      return NULL;
-   else
-   {
-      r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-      return r200_makeX86Attribute3f( & rmesa->vb.dfn_cache.Color3f, key,
-				      __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-
-
-#if 0 /* Temporarily disabled as it is broken w/the new cubemap code. - idr */
-struct dynfn *r200_makeX86TexCoord2fv( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeX86Attribute2fv( & rmesa->vb.dfn_cache.TexCoord2fv, key,
-				    __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-struct dynfn *r200_makeX86TexCoord2f( GLcontext *ctx, const int *key )
-{
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   return r200_makeX86Attribute2f( & rmesa->vb.dfn_cache.TexCoord2f, key,
-				   __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-struct dynfn *r200_makeX86MultiTexCoord2fvARB( GLcontext *ctx, const int *key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] );
-
-   if (rmesa->vb.texcoordptr[1] == rmesa->vb.texcoordptr[0]+4) {
-      DFN ( _x86_MultiTexCoord2fv, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 21, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);	
-      FIXUP(dfn->code, 27, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]+4);
-   } else {
-      DFN ( _x86_MultiTexCoord2fv_2, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 14, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-
-struct dynfn *r200_makeX86MultiTexCoord2fARB( GLcontext *ctx, 
-					      const int *key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   r200ContextPtr rmesa = R200_CONTEXT(ctx);
-
-   if (R200_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] );
-
-   if (rmesa->vb.texcoordptr[1] == rmesa->vb.texcoordptr[0]+4) {
-      DFN ( _x86_MultiTexCoord2f, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 20, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]); 
-      FIXUP(dfn->code, 26, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]+4); 
-   }
-   else {
-      /* Note: this might get generated multiple times, even though the
-       * actual emitted code is the same.
-       */
-      DFN ( _x86_MultiTexCoord2f_2, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 18, 0x0, (int)rmesa->vb.texcoordptr); 
-   }      
-   return dfn;
-}
-#endif
-
-void r200InitX86Codegen( struct dfn_generators *gen )
-{
-   gen->Vertex3f = r200_makeX86Vertex3f;
-   gen->Vertex3fv = r200_makeX86Vertex3fv;
-   gen->Color4ub = r200_makeX86Color4ub; /* PKCOLOR only */
-   gen->Color4ubv = r200_makeX86Color4ubv; /* PKCOLOR only */
-   gen->Normal3f = r200_makeX86Normal3f;
-   gen->Normal3fv = r200_makeX86Normal3fv;
-#if 0 /* Temporarily disabled as it is broken w/the new cubemap code. - idr */
-   gen->TexCoord2f = r200_makeX86TexCoord2f;
-   gen->TexCoord2fv = r200_makeX86TexCoord2fv;
-   gen->MultiTexCoord2fARB = r200_makeX86MultiTexCoord2fARB;
-   gen->MultiTexCoord2fvARB = r200_makeX86MultiTexCoord2fvARB;
-#endif
-   gen->Color3f = r200_makeX86Color3f;
-   gen->Color3fv = r200_makeX86Color3fv;
-
-   /* Not done:
-    */
-/*     gen->Vertex2f = r200_makeX86Vertex2f; */
-/*     gen->Vertex2fv = r200_makeX86Vertex2fv; */
-/*     gen->Color3ub = r200_makeX86Color3ub; */
-/*     gen->Color3ubv = r200_makeX86Color3ubv; */
-/*     gen->Color4f = r200_makeX86Color4f; */
-/*     gen->Color4fv = r200_makeX86Color4fv; */
-/*     gen->TexCoord1f = r200_makeX86TexCoord1f; */
-/*     gen->TexCoord1fv = r200_makeX86TexCoord1fv; */
-/*     gen->MultiTexCoord1fARB = r200_makeX86MultiTexCoord1fARB; */
-/*     gen->MultiTexCoord1fvARB = r200_makeX86MultiTexCoord1fvARB; */
-}
-
-
-#else 
-
-void r200InitX86Codegen( struct dfn_generators *gen )
-{
-   (void) gen;
-}
-
-#endif
diff --git a/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S b/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S
deleted file mode 100644
index 5e33c7bdeea..00000000000
--- a/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S
+++ /dev/null
@@ -1,499 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_vtxtmp_x86.S,v 1.2 2002/11/07 18:31:59 tsi Exp $ */
-/**************************************************************************
-
-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, TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-#define GLOBL( x )	\
-.globl x;		\
-x:
-
-.data
-.align 4
-
-/*
-	vertex 3f vertex size 4
-*/
-	
-GLOBL ( _x86_Vertex3f_4 )
-	movl	(0), %ecx
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	%eax, (%ecx)
-	movl	%edx, 4(%ecx)
-	movl	12(%esp), %eax
-	movl	(0), %edx
-	movl	%eax, 8(%ecx)
-	movl	%edx, 12(%ecx)
-	movl	(0), %eax
-	addl	$16, %ecx
-	dec 	%eax
-	movl	%ecx, (0)
-	movl	%eax, (0)
-	je	.1 
-	ret
-.1:	jmp	*0
-	
-GLOBL ( _x86_Vertex3f_4_end )
-
-/*
-	vertex 3f vertex size 6
-*/
-GLOBL ( _x86_Vertex3f_6 )
-	push	%edi
-	movl	(0), %edi
-	movl	8(%esp), %eax
-	movl	12(%esp), %edx
-	movl	16(%esp), %ecx
-	movl	%eax, (%edi)
-	movl	%edx, 4(%edi)
-	movl	%ecx, 8(%edi)
-	movl	(0), %eax
-	movl	(0), %edx
-	movl	(0), %ecx
-	movl	%eax, 12(%edi)
-	movl	%edx, 16(%edi)
-	movl	%ecx, 20(%edi)
-	addl	$24, %edi
-	movl	(0), %eax
-	movl	%edi, (0)
-	dec 	%eax
-	pop 	%edi
-	movl	%eax, (0)
-	je	.2
-	ret
-.2:	jmp	*0
-GLOBL ( _x86_Vertex3f_6_end )
-/*
-	vertex 3f generic size
-*/
-GLOBL ( _x86_Vertex3f )
-	push	%edi
-	push	%esi
-	movl	$0, %esi
-	movl	(0), %edi
-	movl	12(%esp), %eax
-	movl	16(%esp), %edx
-	movl	20(%esp), %ecx
-	movl	%eax, (%edi)
-	movl	%edx, 4(%edi)
-	movl	%ecx, 8(%edi)
-	addl	$12, %edi
-	movl	$0, %ecx
-	repz
-	movsl %ds:(%esi), %es:(%edi)
-	movl	(0), %eax
-	movl	%edi, (0)
-	dec 	%eax
-	movl	%eax, (0)
-	pop 	%esi
-	pop 	%edi
-	je  	.3
-	ret
-.3:	jmp	*0
-
-GLOBL ( _x86_Vertex3f_end )
-
-/*
-	Vertex 3fv vertex size 6
-*/
-GLOBL ( _x86_Vertex3fv_6 )
-	movl	(0), %eax
-	movl	4(%esp), %ecx
-	movl	(%ecx), %edx
-	movl	%edx, (%eax)
-	movl	4(%ecx), %edx
-	movl	8(%ecx), %ecx
-	movl	%edx, 4(%eax)
-	movl	%ecx, 8(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 12(%eax)
-	movl	%ecx, 16(%eax)
-	movl	(36), %edx
-	movl	%edx, 20(%eax)
-	addl	$24, %eax
-	movl	%eax, 0
-	movl	4, %eax
-	dec 	%eax
-	movl	%eax, 4
-	je	.4
-	ret
-.4:	jmp    *8
-	
-GLOBL ( _x86_Vertex3fv_6_end )
-
-/*
-	Vertex 3fv vertex size 8
-*/
-GLOBL ( _x86_Vertex3fv_8 )
-	movl	(0), %eax
-	movl	4(%esp), %ecx
-	movl	(%ecx), %edx
-	movl	%edx ,(%eax)
-	movl	4(%ecx) ,%edx
-	movl	8(%ecx) ,%ecx
-	movl	%edx, 4(%eax)
-	movl	%ecx, 8(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 12(%eax)
-	movl	%ecx, 16(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 20(%eax)
-	movl	%ecx, 24(%eax)
-	movl	(36), %edx
-	movl	%edx, 28(%eax)
-	addl	$32, %eax
-	movl	%eax, (0)
-	movl	4, %eax
-	dec	%eax
-	movl    %eax, (4)
-	je	.5
-	ret
-.5:	jmp    *8
-	
-GLOBL ( _x86_Vertex3fv_8_end )
-
-/*
-	Vertex 3fv generic vertex size
-*/
-GLOBL ( _x86_Vertex3fv )
-	movl	4(%esp), %edx
-	push	%edi
-	push	%esi
-	movl	(0x1010101), %edi
-	movl	(%edx), %eax
-	movl	4(%edx), %ecx
-	movl	8(%edx), %esi
-	movl	%eax, (%edi)
-	movl	%ecx, 4(%edi)
-	movl	%esi, 8(%edi)
-	addl	$12, %edi
-	movl	$6, %ecx
-	movl	$0x58, %esi
-	repz
-	movsl %ds:(%esi), %es:(%edi)
-	movl	%edi, (0x1010101)
-	movl	(0x2020202), %eax
-	pop	%esi
-	pop	%edi
-	dec	%eax
-	movl	%eax, (0x2020202)
-	je	.6
-	ret
-.6:	jmp    *0
-GLOBL ( _x86_Vertex3fv_end )
-
-
-/**
- * Generic handler for 2 float format data.  This can be used for
- * TexCoord2f and possibly other functions.
- */
-
-GLOBL ( _x86_Attribute2f )
-	movl	$0x0, %edx
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	movl	%eax, (%edx)
-	movl	%ecx, 4(%edx)
-	ret
-GLOBL ( _x86_Attribute2f_end )
-
-
-/**
- * Generic handler for 2 float vector format data.  This can be used for
- * TexCoord2fv and possibly other functions.
- */
-
-GLOBL( _x86_Attribute2fv)
-	movl 4(%esp), %eax      /* load 'v' off stack */
-	movl (%eax), %ecx       /* load v[0] */
-	movl 4(%eax), %eax      /* load v[1] */
-	movl %ecx, 0      	/* store v[0] to current vertex */
-	movl %eax, 4      	/* store v[1] to current vertex */
-	ret
-GLOBL ( _x86_Attribute2fv_end )
-
-
-/**
- * Generic handler for 3 float format data.  This can be used for
- * Normal3f, Color3f (when the color target is also float), or
- * TexCoord3f.
- */
-
-GLOBL ( _x86_Attribute3f )
-	movl	4(%esp), %ecx
-	movl	8(%esp), %edx
-	movl	12(%esp), %eax
-	movl	%ecx, 0
-	movl	%edx, 4
-	movl	%eax, 8
-	ret
-GLOBL ( _x86_Attribute3f_end )
-
-/**
- * Generic handler for 3 float vector format data.  This can be used for
- * Normal3f, Color3f (when the color target is also float), or
- * TexCoord3f.
- */
-
-GLOBL( _x86_Attribute3fv)
-	movl 4(%esp), %eax      /* load 'v' off stack */
-	movl (%eax), %ecx       /* load v[0] */
-	movl 4(%eax), %edx      /* load v[1] */
-	movl 8(%eax), %eax      /* load v[2] */
-	movl %ecx, 0      	/* store v[0] to current vertex */
-	movl %edx, 4      	/* store v[1] to current vertex */
-	movl %eax, 8      	/* store v[2] to current vertex */
-	ret
-GLOBL ( _x86_Attribute3fv_end )
-
-
-/*
-	Color 4ubv_ub
-*/
-GLOBL ( _x86_Color4ubv_ub )
-	movl 4(%esp), %eax
-	movl $0x12345678, %edx
-	movl (%eax), %eax
-	movl %eax, (%edx)
-	ret
-GLOBL ( _x86_Color4ubv_ub_end )
-
-/*
-	Color 4ubv 4f
-*/
-GLOBL ( _x86_Color4ubv_4f )
-	push	%ebx
-	movl	$0, %edx
-	xor	%eax, %eax
-	xor	%ecx, %ecx
-	movl	8(%esp), %ebx
-	movl	(%ebx), %ebx
-	mov	%bl, %al
-	mov	%bh, %cl
-	movl	(%edx,%eax,4),%eax
-	movl	(%edx,%ecx,4),%ecx
-	movl	%eax, (0xdeadbeaf)
-	movl	%ecx, (0xdeadbeaf)
-	xor	%eax, %eax
-	xor	%ecx, %ecx
-	shr	$16, %ebx
-	mov	%bl, %al
-	mov	%bh, %cl
-	movl	(%edx,%eax,4), %eax
-	movl	(%edx,%ecx,4), %ecx
-	movl	%eax, (0xdeadbeaf)
-	movl	%ecx, (0xdeadbeaf)
-	pop	%ebx
-	ret
-GLOBL ( _x86_Color4ubv_4f_end )
-
-/*
-
-	Color4ub_ub
-*/
-GLOBL( _x86_Color4ub_ub )
-	push	%ebx
-	movl	8(%esp), %eax
-	movl	12(%esp), %edx
-	movl	16(%esp), %ecx
-	movl	20(%esp), %ebx
-	mov	%al, (0)
-	mov	%dl, (0)
-	mov	%cl, (0)
-	mov	%bl, (0)
-	pop	%ebx
-	ret
-GLOBL( _x86_Color4ub_ub_end )
-
-
-/* \todo: change the "and $7, %eax" to something like "target & 4 ? target & 5 : target & 3)" */
-/*
-	MultiTexCoord2fv st0/st1
-*/
-GLOBL( _x86_MultiTexCoord2fv )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$7, %eax
-	movl	(%ecx), %edx
-	shl	$3, %eax
-	movl	4(%ecx), %ecx
-	movl	%edx, 0xdeadbeef(%eax)
-	movl	%ecx, 0xdeadbeef(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2fv_end )
-
-/*
-	MultiTexCoord2fv
-*/
-
-GLOBL( _x86_MultiTexCoord2fv_2 )
-	movl	4(%esp,1), %eax
-	movl	8(%esp,1), %ecx
-	and	$0x7, %eax
-	movl	0(,%eax,4), %edx
-	movl	(%ecx), %eax
-	movl	%eax, (%edx)
-	movl	4(%ecx), %eax
-	movl	%eax, 4(%edx)
-	ret
-GLOBL( _x86_MultiTexCoord2fv_2_end )
-
-/*
-	MultiTexCoord2f st0/st1
-*/
-GLOBL( _x86_MultiTexCoord2f )
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	12(%esp), %ecx
-	and	$7, %eax
-	shl	$3, %eax
-	movl	%edx, 0xdeadbeef(%eax)
-	movl	%ecx, 0xdeadbeef(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2f_end )
-
-/*
-	MultiTexCoord2f
-*/
-GLOBL( _x86_MultiTexCoord2f_2 )
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	12(%esp,1), %ecx
-	and	$7,%eax
-	movl	0(,%eax,4), %eax
-	movl	%edx, (%eax)
-	movl	%ecx, 4(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2f_2_end )
-
-#if defined(USE_SSE_ASM)
-/**
- * This can be used as a template for either Color3fv (when the color
- * target is also a 3f) or Normal3fv.
- */
-
-GLOBL( _sse_Attribute3fv )
-	movl	4(%esp), %eax
-	movlps	(%eax), %xmm0
-	movl	8(%eax), %eax
-	movlps	%xmm0, 0
-	movl	%eax, 8
-	ret
-GLOBL( _sse_Attribute3fv_end )
-
-/**
- * This can be used as a template for either Color3f (when the color
- * target is also a 3f) or Normal3f.
- */
-
-GLOBL( _sse_Attribute3f )
-	movlps	4(%esp), %xmm0
-	movl	12(%esp), %eax
-	movlps	%xmm0, 0
-	movl	%eax, 8
-	ret
-GLOBL( _sse_Attribute3f_end )
-
-
-/**
- * Generic handler for 2 float vector format data.  This can be used for
- * TexCoord2fv and possibly other functions.
- */
-
-GLOBL( _sse_Attribute2fv )
-	movl	4(%esp), %eax
-	movlps	(%eax), %xmm0
-	movlps	%xmm0, 0
-	ret
-GLOBL( _sse_Attribute2fv_end )
-
-
-/**
- * Generic handler for 2 float format data.  This can be used for
- * TexCoord2f and possibly other functions.
- */
-
-GLOBL( _sse_Attribute2f )
-	movlps	4(%esp), %xmm0
-	movlps	%xmm0, 0
-	ret
-GLOBL( _sse_Attribute2f_end )
-
-/*
-	MultiTexCoord2fv st0/st1
-*/
-GLOBL( _sse_MultiTexCoord2fv )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$7, %eax
-	movlps	(%ecx), %xmm0
-	movlps	%xmm0, 0xdeadbeef(,%eax,8)
-	ret
-GLOBL( _sse_MultiTexCoord2fv_end )
-
-/*
-	MultiTexCoord2fv
-*/
-GLOBL( _sse_MultiTexCoord2fv_2 )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$0x7, %eax
-	movl	0(,%eax,4), %edx
-	movlps	(%ecx), %xmm0
-	movlps	%xmm0, (%edx)
-	ret
-GLOBL( _sse_MultiTexCoord2fv_2_end )
-
-/*
-	MultiTexCoord2f st0/st1
-*/
-GLOBL( _sse_MultiTexCoord2f )
-	movl	4(%esp), %eax
-	and	$7, %eax
-	movlps	8(%esp), %xmm0
-	movlps	%xmm0, 0xdeadbeef(,%eax,8)
-	ret
-GLOBL( _sse_MultiTexCoord2f_end )
-
-/*
-	MultiTexCoord2f
-*/
-GLOBL( _sse_MultiTexCoord2f_2 )
-	movl	4(%esp), %eax
-	movlps	8(%esp), %xmm0
-	and	$7,%eax
-	movl	0(,%eax,4), %eax
-	movlps	%xmm0, (%eax)
-	ret
-GLOBL( _sse_MultiTexCoord2f_2_end )
-#endif
-
-#if defined (__ELF__) && defined (__linux__)
-	.section .note.GNU-stack,"",%progbits
-#endif
diff --git a/src/mesa/drivers/dri/r300/Makefile b/src/mesa/drivers/dri/r300/Makefile
index 5abb91d9876..2ee23289341 100644
--- a/src/mesa/drivers/dri/r300/Makefile
+++ b/src/mesa/drivers/dri/r300/Makefile
@@ -90,9 +90,11 @@ COMMON_SYMLINKS = \
 
 include ../Makefile.template
 
-$(SYMLINKS):
+server:
 	mkdir -p server
-	for i in $(SYMLINKS) ; do rm -f $$i && test -f ../radeon/$$i && ln -s ../../radeon/$$i $$i ; done
+
+$(SYMLINKS): server
+	@[ -e $@ ] || ln -sf ../../radeon/$@ server/
 
 $(COMMON_SYMLINKS):
 	@[ -e $@ ] || ln -sf ../radeon/$@ ./
diff --git a/src/mesa/drivers/dri/r300/r200_context.h b/src/mesa/drivers/dri/r300/r200_context.h
deleted file mode 100644
index a06d7152d78..00000000000
--- a/src/mesa/drivers/dri/r300/r200_context.h
+++ /dev/null
@@ -1,822 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_context.h,v 1.2 2002/12/16 16:18:54 dawes Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#ifndef __R200_CONTEXT_H__
-#define __R200_CONTEXT_H__
-
-#ifdef GLX_DIRECT_RENDERING
-
-#include "tnl/t_vertex.h"
-#include "drm.h"
-#include "radeon_drm.h"
-#include "dri_util.h"
-#include "texmem.h"
-
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "r200_reg.h"
-#include "radeon_context.h"
-
-#define ENABLE_HW_3D_TEXTURE 1	/* XXX this is temporary! */
-
-struct r200_context;
-typedef struct r200_context r200ContextRec;
-typedef struct r200_context *r200ContextPtr;
-
-#include "mm.h"
-
-/* The blit width for texture uploads
- */
-#define BLIT_WIDTH_BYTES 1024
-
-/* Use the templated vertex format:
- */
-#define COLOR_IS_RGBA
-#define TAG(x) r200##x
-#include "tnl_dd/t_dd_vertex.h"
-#undef TAG
-
-typedef void (*r200_tri_func) (r200ContextPtr,
-			       r200Vertex *, r200Vertex *, r200Vertex *);
-
-typedef void (*r200_line_func) (r200ContextPtr, r200Vertex *, r200Vertex *);
-
-typedef void (*r200_point_func) (r200ContextPtr, r200Vertex *);
-
-struct r200_depthbuffer_state {
-	GLfloat scale;
-};
-
-struct r200_stencilbuffer_state {
-	GLboolean hwBuffer;
-	GLuint clear;		/* rb3d_stencilrefmask value */
-};
-
-struct r200_stipple_state {
-	GLuint mask[32];
-};
-
-typedef struct r200_tex_obj r200TexObj, *r200TexObjPtr;
-
-/* Texture object in locally shared texture space.
- */
-struct r200_tex_obj {
-	driTextureObject base;
-
-	GLuint bufAddr;		/* Offset to start of locally
-				   shared texture block */
-
-	GLuint dirty_state;	/* Flags (1 per texunit) for
-				   whether or not this texobj
-				   has dirty hardware state
-				   (pp_*) that needs to be
-				   brought into the
-				   texunit. */
-
-	drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
-	/* Six, for the cube faces */
-
-	GLuint pp_txfilter;	/* hardware register values */
-	GLuint pp_txformat;
-	GLuint pp_txformat_x;
-	GLuint pp_txoffset;	/* Image location in texmem.
-				   All cube faces follow. */
-	GLuint pp_txsize;	/* npot only */
-	GLuint pp_txpitch;	/* npot only */
-	GLuint pp_border_color;
-	GLuint pp_cubic_faces;	/* cube face 1,2,3,4 log2 sizes */
-
-	GLboolean border_fallback;
-};
-
-struct r200_texture_env_state {
-	r200TexObjPtr texobj;
-	GLenum format;
-	GLenum envMode;
-};
-
-#define R200_MAX_TEXTURE_UNITS 6
-
-struct r200_texture_state {
-	struct r200_texture_env_state unit[R200_MAX_TEXTURE_UNITS];
-};
-
-struct r200_state_atom {
-	struct r200_state_atom *next, *prev;
-	const char *name;	/* for debug */
-	int cmd_size;		/* size in bytes */
-	GLuint idx;
-	int *cmd;		/* one or more cmd's */
-	int *lastcmd;		/* one or more cmd's */
-	int *savedcmd;		/* one or more cmd's */
-	GLboolean dirty;
-	 GLboolean(*check) (GLcontext *, int);	/* is this state active? */
-};
-
-/* Trying to keep these relatively short as the variables are becoming
- * extravagently long.  Drop the driver name prefix off the front of
- * everything - I think we know which driver we're in by now, and keep the
- * prefix to 3 letters unless absolutely impossible.
- */
-
-#define CTX_CMD_0             0
-#define CTX_PP_MISC           1
-#define CTX_PP_FOG_COLOR      2
-#define CTX_RE_SOLID_COLOR    3
-#define CTX_RB3D_BLENDCNTL    4
-#define CTX_RB3D_DEPTHOFFSET  5
-#define CTX_RB3D_DEPTHPITCH   6
-#define CTX_RB3D_ZSTENCILCNTL 7
-#define CTX_CMD_1             8
-#define CTX_PP_CNTL           9
-#define CTX_RB3D_CNTL         10
-#define CTX_RB3D_COLOROFFSET  11
-#define CTX_CMD_2             12	/* why */
-#define CTX_RB3D_COLORPITCH   13	/* why */
-#define CTX_STATE_SIZE_OLDDRM 14
-#define CTX_CMD_3             14
-#define CTX_RB3D_BLENDCOLOR   15
-#define CTX_RB3D_ABLENDCNTL   16
-#define CTX_RB3D_CBLENDCNTL   17
-#define CTX_STATE_SIZE_NEWDRM 18
-
-#define SET_CMD_0               0
-#define SET_SE_CNTL             1
-#define SET_RE_CNTL             2	/* replace se_coord_fmt */
-#define SET_STATE_SIZE          3
-
-#define VTE_CMD_0               0
-#define VTE_SE_VTE_CNTL         1
-#define VTE_STATE_SIZE          2
-
-#define LIN_CMD_0               0
-#define LIN_RE_LINE_PATTERN     1
-#define LIN_RE_LINE_STATE       2
-#define LIN_CMD_1               3
-#define LIN_SE_LINE_WIDTH       4
-#define LIN_STATE_SIZE          5
-
-#define MSK_CMD_0               0
-#define MSK_RB3D_STENCILREFMASK 1
-#define MSK_RB3D_ROPCNTL        2
-#define MSK_RB3D_PLANEMASK      3
-#define MSK_STATE_SIZE          4
-
-#define VPT_CMD_0           0
-#define VPT_SE_VPORT_XSCALE          1
-#define VPT_SE_VPORT_XOFFSET         2
-#define VPT_SE_VPORT_YSCALE          3
-#define VPT_SE_VPORT_YOFFSET         4
-#define VPT_SE_VPORT_ZSCALE          5
-#define VPT_SE_VPORT_ZOFFSET         6
-#define VPT_STATE_SIZE      7
-
-#define ZBS_CMD_0               0
-#define ZBS_SE_ZBIAS_FACTOR     1
-#define ZBS_SE_ZBIAS_CONSTANT   2
-#define ZBS_STATE_SIZE          3
-
-#define MSC_CMD_0               0
-#define MSC_RE_MISC             1
-#define MSC_STATE_SIZE          2
-
-#define TAM_CMD_0               0
-#define TAM_DEBUG3              1
-#define TAM_STATE_SIZE          2
-
-#define TEX_CMD_0                   0
-#define TEX_PP_TXFILTER             1	/*2c00 */
-#define TEX_PP_TXFORMAT             2	/*2c04 */
-#define TEX_PP_TXFORMAT_X           3	/*2c08 */
-#define TEX_PP_TXSIZE               4	/*2c0c */
-#define TEX_PP_TXPITCH              5	/*2c10 */
-#define TEX_PP_BORDER_COLOR         6	/*2c14 */
-#define TEX_CMD_1                   7
-#define TEX_PP_TXOFFSET             8	/*2d00 */
-#define TEX_STATE_SIZE              9
-
-#define CUBE_CMD_0                  0	/* 1 register follows */
-#define CUBE_PP_CUBIC_FACES         1	/* 0x2c18 */
-#define CUBE_CMD_1                  2	/* 5 registers follow */
-#define CUBE_PP_CUBIC_OFFSET_F1     3	/* 0x2d04 */
-#define CUBE_PP_CUBIC_OFFSET_F2     4	/* 0x2d08 */
-#define CUBE_PP_CUBIC_OFFSET_F3     5	/* 0x2d0c */
-#define CUBE_PP_CUBIC_OFFSET_F4     6	/* 0x2d10 */
-#define CUBE_PP_CUBIC_OFFSET_F5     7	/* 0x2d14 */
-#define CUBE_STATE_SIZE             8
-
-#define PIX_CMD_0                   0
-#define PIX_PP_TXCBLEND             1
-#define PIX_PP_TXCBLEND2            2
-#define PIX_PP_TXABLEND             3
-#define PIX_PP_TXABLEND2            4
-#define PIX_STATE_SIZE              5
-
-#define TF_CMD_0                    0
-#define TF_TFACTOR_0                1
-#define TF_TFACTOR_1                2
-#define TF_TFACTOR_2                3
-#define TF_TFACTOR_3                4
-#define TF_TFACTOR_4                5
-#define TF_TFACTOR_5                6
-#define TF_STATE_SIZE               7
-
-#define TCL_CMD_0                 0
-#define TCL_LIGHT_MODEL_CTL_0     1
-#define TCL_LIGHT_MODEL_CTL_1     2
-#define TCL_PER_LIGHT_CTL_0       3
-#define TCL_PER_LIGHT_CTL_1       4
-#define TCL_PER_LIGHT_CTL_2       5
-#define TCL_PER_LIGHT_CTL_3       6
-#define TCL_CMD_1                 7
-#define TCL_UCP_VERT_BLEND_CTL    8
-#define TCL_STATE_SIZE            9
-
-#define MSL_CMD_0                     0
-#define MSL_MATRIX_SELECT_0           1
-#define MSL_MATRIX_SELECT_1           2
-#define MSL_MATRIX_SELECT_2           3
-#define MSL_MATRIX_SELECT_3           4
-#define MSL_MATRIX_SELECT_4           5
-#define MSL_STATE_SIZE                6
-
-#define TCG_CMD_0                 0
-#define TCG_TEX_PROC_CTL_2            1
-#define TCG_TEX_PROC_CTL_3            2
-#define TCG_TEX_PROC_CTL_0            3
-#define TCG_TEX_PROC_CTL_1            4
-#define TCG_TEX_CYL_WRAP_CTL      5
-#define TCG_STATE_SIZE            6
-
-#define MTL_CMD_0            0
-#define MTL_EMMISSIVE_RED    1
-#define MTL_EMMISSIVE_GREEN  2
-#define MTL_EMMISSIVE_BLUE   3
-#define MTL_EMMISSIVE_ALPHA  4
-#define MTL_AMBIENT_RED      5
-#define MTL_AMBIENT_GREEN    6
-#define MTL_AMBIENT_BLUE     7
-#define MTL_AMBIENT_ALPHA    8
-#define MTL_DIFFUSE_RED      9
-#define MTL_DIFFUSE_GREEN    10
-#define MTL_DIFFUSE_BLUE     11
-#define MTL_DIFFUSE_ALPHA    12
-#define MTL_SPECULAR_RED     13
-#define MTL_SPECULAR_GREEN   14
-#define MTL_SPECULAR_BLUE    15
-#define MTL_SPECULAR_ALPHA   16
-#define MTL_CMD_1            17
-#define MTL_SHININESS        18
-#define MTL_STATE_SIZE       19
-
-#define VAP_CMD_0                   0
-#define VAP_SE_VAP_CNTL             1
-#define VAP_STATE_SIZE              2
-
-/* Replaces a lot of packet info from radeon
- */
-#define VTX_CMD_0                   0
-#define VTX_VTXFMT_0            1
-#define VTX_VTXFMT_1            2
-#define VTX_TCL_OUTPUT_VTXFMT_0 3
-#define VTX_TCL_OUTPUT_VTXFMT_1 4
-#define VTX_CMD_1               5
-#define VTX_TCL_OUTPUT_COMPSEL  6
-#define VTX_CMD_2               7
-#define VTX_STATE_CNTL          8
-#define VTX_STATE_SIZE          9
-
-#define VTX_COLOR(v,n)   (((v)>>(R200_VTX_COLOR_0_SHIFT+(n)*2))&\
-                         R200_VTX_COLOR_MASK)
-
-/**
- * Given the \c R200_SE_VTX_FMT_1 for the current vertex state, determine
- * how many components are in texture coordinate \c n.
- */
-#define VTX_TEXn_COUNT(v,n)   (((v) >> (3 * n)) & 0x07)
-
-#define MAT_CMD_0              0
-#define MAT_ELT_0              1
-#define MAT_STATE_SIZE         17
-
-#define GRD_CMD_0                  0
-#define GRD_VERT_GUARD_CLIP_ADJ    1
-#define GRD_VERT_GUARD_DISCARD_ADJ 2
-#define GRD_HORZ_GUARD_CLIP_ADJ    3
-#define GRD_HORZ_GUARD_DISCARD_ADJ 4
-#define GRD_STATE_SIZE             5
-
-/* position changes frequently when lighting in modelpos - separate
- * out to new state item?
- */
-#define LIT_CMD_0                  0
-#define LIT_AMBIENT_RED            1
-#define LIT_AMBIENT_GREEN          2
-#define LIT_AMBIENT_BLUE           3
-#define LIT_AMBIENT_ALPHA          4
-#define LIT_DIFFUSE_RED            5
-#define LIT_DIFFUSE_GREEN          6
-#define LIT_DIFFUSE_BLUE           7
-#define LIT_DIFFUSE_ALPHA          8
-#define LIT_SPECULAR_RED           9
-#define LIT_SPECULAR_GREEN         10
-#define LIT_SPECULAR_BLUE          11
-#define LIT_SPECULAR_ALPHA         12
-#define LIT_POSITION_X             13
-#define LIT_POSITION_Y             14
-#define LIT_POSITION_Z             15
-#define LIT_POSITION_W             16
-#define LIT_DIRECTION_X            17
-#define LIT_DIRECTION_Y            18
-#define LIT_DIRECTION_Z            19
-#define LIT_DIRECTION_W            20
-#define LIT_ATTEN_QUADRATIC        21
-#define LIT_ATTEN_LINEAR           22
-#define LIT_ATTEN_CONST            23
-#define LIT_ATTEN_XXX              24
-#define LIT_CMD_1                  25
-#define LIT_SPOT_DCD               26
-#define LIT_SPOT_DCM               27
-#define LIT_SPOT_EXPONENT          28
-#define LIT_SPOT_CUTOFF            29
-#define LIT_SPECULAR_THRESH        30
-#define LIT_RANGE_CUTOFF           31	/* ? */
-#define LIT_ATTEN_CONST_INV        32
-#define LIT_STATE_SIZE             33
-
-/* Fog
- */
-#define FOG_CMD_0      0
-#define FOG_R          1
-#define FOG_C          2
-#define FOG_D          3
-#define FOG_PAD        4
-#define FOG_STATE_SIZE 5
-
-/* UCP
- */
-#define UCP_CMD_0      0
-#define UCP_X          1
-#define UCP_Y          2
-#define UCP_Z          3
-#define UCP_W          4
-#define UCP_STATE_SIZE 5
-
-/* GLT - Global ambient
- */
-#define GLT_CMD_0      0
-#define GLT_RED        1
-#define GLT_GREEN      2
-#define GLT_BLUE       3
-#define GLT_ALPHA      4
-#define GLT_STATE_SIZE 5
-
-/* EYE
- */
-#define EYE_CMD_0          0
-#define EYE_X              1
-#define EYE_Y              2
-#define EYE_Z              3
-#define EYE_RESCALE_FACTOR 4
-#define EYE_STATE_SIZE     5
-
-/* CST - constant state
- */
-#define CST_CMD_0                             0
-#define CST_PP_CNTL_X                         1
-#define CST_CMD_1                             2
-#define CST_RB3D_DEPTHXY_OFFSET               3
-#define CST_CMD_2                             4
-#define CST_RE_AUX_SCISSOR_CNTL               5
-#define CST_CMD_3                             6
-#define CST_RE_SCISSOR_TL_0                   7
-#define CST_RE_SCISSOR_BR_0                   8
-#define CST_CMD_4                             9
-#define CST_SE_VAP_CNTL_STATUS                10
-#define CST_CMD_5                             11
-#define CST_RE_POINTSIZE                      12
-#define CST_CMD_6                             13
-#define CST_SE_TCL_INPUT_VTX_0                14
-#define CST_SE_TCL_INPUT_VTX_1                15
-#define CST_SE_TCL_INPUT_VTX_2                16
-#define CST_SE_TCL_INPUT_VTX_3                17
-#define CST_STATE_SIZE                        18
-
-struct r200_hw_state {
-	/* Head of the linked list of state atoms. */
-	struct r200_state_atom atomlist;
-
-	/* Hardware state, stored as cmdbuf commands:
-	 *   -- Need to doublebuffer for
-	 *           - reviving state after loss of context
-	 *           - eliding noop statechange loops? (except line stipple count)
-	 */
-	struct r200_state_atom ctx;
-	struct r200_state_atom set;
-	struct r200_state_atom vte;
-	struct r200_state_atom lin;
-	struct r200_state_atom msk;
-	struct r200_state_atom vpt;
-	struct r200_state_atom vap;
-	struct r200_state_atom vtx;
-	struct r200_state_atom tcl;
-	struct r200_state_atom msl;
-	struct r200_state_atom tcg;
-	struct r200_state_atom msc;
-	struct r200_state_atom cst;
-	struct r200_state_atom tam;
-	struct r200_state_atom tf;
-	struct r200_state_atom tex[6];
-	struct r200_state_atom cube[6];
-	struct r200_state_atom zbs;
-	struct r200_state_atom mtl[2];
-	struct r200_state_atom mat[9];
-	struct r200_state_atom lit[8];	/* includes vec, scl commands */
-	struct r200_state_atom ucp[6];
-	struct r200_state_atom pix[6];	/* pixshader stages */
-	struct r200_state_atom eye;	/* eye pos */
-	struct r200_state_atom grd;	/* guard band clipping */
-	struct r200_state_atom fog;
-	struct r200_state_atom glt;
-
-	int max_state_size;	/* Number of bytes necessary for a full state emit. */
-	GLboolean is_dirty, all_dirty;
-};
-
-struct r200_colorbuffer_state {
-	int roundEnable;
-};
-
-struct r200_state {
-	/* Derived state for internal purposes:
-	 */
-	struct r200_colorbuffer_state color;
-	struct r200_depthbuffer_state depth;
-	struct r200_stencilbuffer_state stencil;
-	struct r200_stipple_state stipple;
-	struct r200_texture_state texture;
-};
-
-/* Need refcounting on dma buffers:
- */
-struct r200_dma_buffer {
-	int refcount;		/* the number of retained regions in buf */
-	drmBufPtr buf;
-};
-
-#define GET_START(rvb) (rmesa->radeon.radeonScreen->gart_buffer_offset +		\
-			(rvb)->address - rmesa->dma.buf0_address +	\
-			(rvb)->start)
-
-/* A retained region, eg vertices for indexed vertices.
- */
-struct r200_dma_region {
-	struct r200_dma_buffer *buf;
-	char *address;		/* == buf->address */
-	int start, end, ptr;	/* offsets from start of buf */
-	int aos_start;
-	int aos_stride;
-	int aos_size;
-};
-
-struct r200_dma {
-	/* Active dma region.  Allocations for vertices and retained
-	 * regions come from here.  Also used for emitting random vertices,
-	 * these may be flushed by calling flush_current();
-	 */
-	struct r200_dma_region current;
-
-	void (*flush) (r200ContextPtr);
-
-	char *buf0_address;	/* start of buf[0], for index calcs */
-	GLuint nr_released_bufs;	/* flush after so many buffers released */
-};
-
-#define R200_CMD_BUF_SZ  (8*1024)
-
-struct r200_store {
-	GLuint statenr;
-	GLuint primnr;
-	char cmd_buf[R200_CMD_BUF_SZ];
-	int cmd_used;
-	int elts_start;
-};
-
-/* r200_tcl.c
- */
-struct r200_tcl_info {
-	GLuint vertex_format;
-	GLint last_offset;
-	GLuint hw_primitive;
-
-	struct r200_dma_region *aos_components[8];
-	GLuint nr_aos_components;
-
-	GLuint *Elts;
-
-	struct r200_dma_region indexed_verts;
-	struct r200_dma_region obj;
-	struct r200_dma_region rgba;
-	struct r200_dma_region spec;
-	struct r200_dma_region fog;
-	struct r200_dma_region tex[R200_MAX_TEXTURE_UNITS];
-	struct r200_dma_region norm;
-};
-
-/* r200_swtcl.c
- */
-struct r200_swtcl_info {
-	GLuint RenderIndex;
-
-   /**
-    * Size of a hardware vertex.  This is calculated when \c ::vertex_attrs is
-    * installed in the Mesa state vector.
-    */
-	GLuint vertex_size;
-
-   /**
-    * Attributes instructing the Mesa TCL pipeline where / how to put vertex
-    * data in the hardware buffer.
-    */
-	struct tnl_attr_map vertex_attrs[VERT_ATTRIB_MAX];
-
-   /**
-    * Number of elements of \c ::vertex_attrs that are actually used.
-    */
-	GLuint vertex_attr_count;
-
-   /**
-    * Cached pointer to the buffer where Mesa will store vertex data.
-    */
-	GLubyte *verts;
-
-	/* Fallback rasterization functions
-	 */
-	r200_point_func draw_point;
-	r200_line_func draw_line;
-	r200_tri_func draw_tri;
-
-	GLuint hw_primitive;
-	GLenum render_primitive;
-	GLuint numverts;
-
-   /**
-    * Offset of the 4UB color data within a hardware (swtcl) vertex.
-    */
-	GLuint coloroffset;
-
-   /**
-    * Offset of the 3UB specular color data within a hardware (swtcl) vertex.
-    */
-	GLuint specoffset;
-
-   /**
-    * Should Mesa project vertex data or will the hardware do it?
-    */
-	GLboolean needproj;
-
-	struct r200_dma_region indexed_verts;
-};
-
-struct r200_ioctl {
-	GLuint vertex_offset;
-	GLuint vertex_size;
-};
-
-#define R200_MAX_PRIMS 64
-
-/* Want to keep a cache of these around.  Each is parameterized by
- * only a single value which has only a small range.  Only expect a
- * few, so just rescan the list each time?
- */
-struct dynfn {
-	struct dynfn *next, *prev;
-	int key[2];
-	char *code;
-};
-
-struct dfn_lists {
-	struct dynfn Vertex2f;
-	struct dynfn Vertex2fv;
-	struct dynfn Vertex3f;
-	struct dynfn Vertex3fv;
-	struct dynfn Color4ub;
-	struct dynfn Color4ubv;
-	struct dynfn Color3ub;
-	struct dynfn Color3ubv;
-	struct dynfn Color4f;
-	struct dynfn Color4fv;
-	struct dynfn Color3f;
-	struct dynfn Color3fv;
-	struct dynfn SecondaryColor3ubEXT;
-	struct dynfn SecondaryColor3ubvEXT;
-	struct dynfn SecondaryColor3fEXT;
-	struct dynfn SecondaryColor3fvEXT;
-	struct dynfn Normal3f;
-	struct dynfn Normal3fv;
-	struct dynfn TexCoord3f;
-	struct dynfn TexCoord3fv;
-	struct dynfn TexCoord2f;
-	struct dynfn TexCoord2fv;
-	struct dynfn TexCoord1f;
-	struct dynfn TexCoord1fv;
-	struct dynfn MultiTexCoord3fARB;
-	struct dynfn MultiTexCoord3fvARB;
-	struct dynfn MultiTexCoord2fARB;
-	struct dynfn MultiTexCoord2fvARB;
-	struct dynfn MultiTexCoord1fARB;
-	struct dynfn MultiTexCoord1fvARB;
-};
-
-struct dfn_generators {
-	struct dynfn *(*Vertex2f) (GLcontext *, const int *);
-	struct dynfn *(*Vertex2fv) (GLcontext *, const int *);
-	struct dynfn *(*Vertex3f) (GLcontext *, const int *);
-	struct dynfn *(*Vertex3fv) (GLcontext *, const int *);
-	struct dynfn *(*Color4ub) (GLcontext *, const int *);
-	struct dynfn *(*Color4ubv) (GLcontext *, const int *);
-	struct dynfn *(*Color3ub) (GLcontext *, const int *);
-	struct dynfn *(*Color3ubv) (GLcontext *, const int *);
-	struct dynfn *(*Color4f) (GLcontext *, const int *);
-	struct dynfn *(*Color4fv) (GLcontext *, const int *);
-	struct dynfn *(*Color3f) (GLcontext *, const int *);
-	struct dynfn *(*Color3fv) (GLcontext *, const int *);
-	struct dynfn *(*SecondaryColor3ubEXT) (GLcontext *, const int *);
-	struct dynfn *(*SecondaryColor3ubvEXT) (GLcontext *, const int *);
-	struct dynfn *(*SecondaryColor3fEXT) (GLcontext *, const int *);
-	struct dynfn *(*SecondaryColor3fvEXT) (GLcontext *, const int *);
-	struct dynfn *(*Normal3f) (GLcontext *, const int *);
-	struct dynfn *(*Normal3fv) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord3f) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord3fv) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord2f) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord2fv) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord1f) (GLcontext *, const int *);
-	struct dynfn *(*TexCoord1fv) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord3fARB) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord3fvARB) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord2fARB) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord2fvARB) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord1fARB) (GLcontext *, const int *);
-	struct dynfn *(*MultiTexCoord1fvARB) (GLcontext *, const int *);
-};
-
-struct r200_prim {
-	GLuint start;
-	GLuint end;
-	GLuint prim;
-};
-
-   /* A maximum total of 29 elements per vertex:  3 floats for position, 3
-    * floats for normal, 4 floats for color, 4 bytes for secondary color,
-    * 3 floats for each texture unit (18 floats total).
-    *
-    * we maybe need add. 4 to prevent segfault if someone specifies
-    * GL_TEXTURE6/GL_TEXTURE7 (esp. for the codegen-path) (FIXME: )
-    *
-    * The position data is never actually stored here, so 3 elements could be
-    * trimmed out of the buffer.
-    */
-
-#define R200_MAX_VERTEX_SIZE ((3*6)+11)
-
-struct r200_vbinfo {
-	GLint counter, initial_counter;
-	GLint *dmaptr;
-	void (*notify) (void);
-	GLint vertex_size;
-
-	union {
-		float f;
-		int i;
-		r200_color_t color;
-	} vertex[R200_MAX_VERTEX_SIZE];
-
-	GLfloat *normalptr;
-	GLfloat *floatcolorptr;
-	r200_color_t *colorptr;
-	GLfloat *floatspecptr;
-	r200_color_t *specptr;
-	GLfloat *texcoordptr[8];	/* 6 (TMU) + 2 for r200_vtxfmt_c.c when GL_TEXTURE6/7 */
-
-	GLenum *prim;		/* &ctx->Driver.CurrentExecPrimitive */
-	GLuint primflags;
-	GLboolean enabled;	/* *_NO_VTXFMT / *_NO_TCL env vars */
-	GLboolean installed;
-	GLboolean fell_back;
-	GLboolean recheck;
-	GLint nrverts;
-	GLuint vtxfmt_0, vtxfmt_1;
-
-	GLuint installed_vertex_format;
-	GLuint installed_color_3f_sz;
-
-	struct r200_prim primlist[R200_MAX_PRIMS];
-	int nrprims;
-
-	struct dfn_lists dfn_cache;
-	struct dfn_generators codegen;
-	GLvertexformat vtxfmt;
-};
-
-/**
- * R200 context structure.
- */
-struct r200_context {
-	struct radeon_context radeon; /* parent class, must be first */
-
-	/* Driver and hardware state management
-	 */
-	struct r200_hw_state hw;
-	struct r200_state state;
-
-	/* Texture object bookkeeping
-	 */
-	unsigned nr_heaps;
-	driTexHeap *texture_heaps[RADEON_NR_TEX_HEAPS];
-	driTextureObject swapped;
-	int texture_depth;
-	float initialMaxAnisotropy;
-
-	/* Rasterization and vertex state:
-	 */
-	GLuint NewGLState;
-
-	/* Vertex buffers
-	 */
-	struct r200_ioctl ioctl;
-	struct r200_dma dma;
-	struct r200_store store;
-	GLboolean save_on_next_unlock;
-
-	/* Clientdata textures;
-	 */
-	GLuint prefer_gart_client_texturing;
-
-	/* TCL stuff
-	 */
-	GLmatrix TexGenMatrix[R200_MAX_TEXTURE_UNITS];
-	GLboolean recheck_texgen[R200_MAX_TEXTURE_UNITS];
-	GLboolean TexGenNeedNormals[R200_MAX_TEXTURE_UNITS];
-	GLuint TexMatEnabled;
-	GLuint TexMatCompSel;
-	GLuint TexGenEnabled;
-	GLuint TexGenInputs;
-	GLuint TexGenCompSel;
-	GLmatrix tmpmat;
-
-	/* r200_tcl.c
-	 */
-	struct r200_tcl_info tcl;
-
-	/* r200_swtcl.c
-	 */
-	struct r200_swtcl_info swtcl;
-
-	/* r200_vtxfmt.c
-	 */
-	struct r200_vbinfo vb;
-};
-
-#define R200_CONTEXT(ctx)		((r200ContextPtr)(ctx->DriverCtx))
-
-extern void r200DestroyContext(__DRIcontextPrivate * driContextPriv);
-extern GLboolean r200CreateContext(const __GLcontextModes * glVisual,
-				   __DRIcontextPrivate * driContextPriv,
-				   void *sharedContextPrivate);
-
-#endif
-#endif				/* __R200_CONTEXT_H__ */
diff --git a/src/mesa/drivers/dri/r300/r200_ioctl.h b/src/mesa/drivers/dri/r300/r200_ioctl.h
deleted file mode 100644
index db7bd7697f1..00000000000
--- a/src/mesa/drivers/dri/r300/r200_ioctl.h
+++ /dev/null
@@ -1,204 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#ifndef __R200_IOCTL_H__
-#define __R200_IOCTL_H__
-
-#include "simple_list.h"
-#include "radeon_dri.h"
-#include "radeon_lock.h"
-
-#include "xf86drm.h"
-#include "drm.h"
-#include "r200_context.h"
-#include "radeon_drm.h"
-
-extern void r200EmitState(r200ContextPtr rmesa);
-extern void r200EmitVertexAOS(r200ContextPtr rmesa,
-			      GLuint vertex_size, GLuint offset);
-
-extern void r200EmitVbufPrim(r200ContextPtr rmesa,
-			     GLuint primitive, GLuint vertex_nr);
-
-extern void r200FlushElts(r200ContextPtr rmesa);
-
-extern GLushort *r200AllocEltsOpenEnded(r200ContextPtr rmesa,
-					GLuint primitive, GLuint min_nr);
-
-extern void r200EmitAOS(r200ContextPtr rmesa,
-			struct r200_dma_region **regions,
-			GLuint n, GLuint offset);
-
-extern void r200EmitBlit(r200ContextPtr rmesa,
-			 GLuint color_fmt,
-			 GLuint src_pitch,
-			 GLuint src_offset,
-			 GLuint dst_pitch,
-			 GLuint dst_offset,
-			 GLint srcx, GLint srcy,
-			 GLint dstx, GLint dsty, GLuint w, GLuint h);
-
-extern void r200EmitWait(r200ContextPtr rmesa, GLuint flags);
-
-extern void r200FlushCmdBuf(r200ContextPtr rmesa, const char *);
-extern int r200FlushCmdBufLocked(r200ContextPtr rmesa, const char *caller);
-extern void r200Flush(GLcontext * ctx);
-
-extern void r200RefillCurrentDmaRegion(r200ContextPtr rmesa);
-
-extern void r200AllocDmaRegion(r200ContextPtr rmesa,
-			       struct r200_dma_region *region,
-			       int bytes, int alignment);
-
-extern void r200AllocDmaRegionVerts(r200ContextPtr rmesa,
-				    struct r200_dma_region *region,
-				    int numverts, int vertsize, int alignment);
-
-extern void r200ReleaseDmaRegion(r200ContextPtr rmesa,
-				 struct r200_dma_region *region,
-				 const char *caller);
-
-extern void r200WaitForVBlank(r200ContextPtr rmesa);
-extern void r200InitIoctlFuncs(struct dd_function_table *functions);
-
-extern void *r200AllocateMemoryMESA(__DRInativeDisplay * dpy, int scrn,
-				    GLsizei size, GLfloat readfreq,
-				    GLfloat writefreq, GLfloat priority);
-extern void r200FreeMemoryMESA(__DRInativeDisplay * dpy, int scrn,
-			       GLvoid * pointer);
-extern GLuint r200GetMemoryOffsetMESA(__DRInativeDisplay * dpy, int scrn,
-				      const GLvoid * pointer);
-
-extern GLboolean r200IsGartMemory(r200ContextPtr rmesa, const GLvoid * pointer,
-				  GLint size);
-
-extern GLuint r200GartOffsetFromVirtual(r200ContextPtr rmesa,
-					const GLvoid * pointer);
-
-void r200SaveHwState(r200ContextPtr radeon);
-void r200SetUpAtomList(r200ContextPtr rmesa);
-
-/* ================================================================
- * Helper macros:
- */
-
-/* Close off the last primitive, if it exists.
- */
-#define R200_NEWPRIM( rmesa )			\
-do {						\
-   if ( rmesa->dma.flush )			\
-      rmesa->dma.flush( rmesa );	\
-} while (0)
-
-/* Can accomodate several state changes and primitive changes without
- * actually firing the buffer.
- */
-#define R200_STATECHANGE( rmesa, ATOM )			\
-do {								\
-   R200_NEWPRIM( rmesa );					\
-   rmesa->hw.ATOM.dirty = GL_TRUE;				\
-   rmesa->hw.is_dirty = GL_TRUE;				\
-} while (0)
-
-#define R200_DB_STATE( ATOM )			        \
-   memcpy( rmesa->hw.ATOM.lastcmd, rmesa->hw.ATOM.cmd,	\
-	   rmesa->hw.ATOM.cmd_size * 4)
-
-static __inline int R200_DB_STATECHANGE(r200ContextPtr rmesa,
-					struct r200_state_atom *atom)
-{
-	if (memcmp(atom->cmd, atom->lastcmd, atom->cmd_size * 4)) {
-		int *tmp;
-		R200_NEWPRIM(rmesa);
-		atom->dirty = GL_TRUE;
-		rmesa->hw.is_dirty = GL_TRUE;
-		tmp = atom->cmd;
-		atom->cmd = atom->lastcmd;
-		atom->lastcmd = tmp;
-		return 1;
-	} else
-		return 0;
-}
-
-/* Fire the buffered vertices no matter what.
- */
-#define R200_FIREVERTICES( r200 )			\
-do {							\
-   if ( (r200)->store.cmd_used || (r200)->dma.flush ) {	\
-      radeonFlush( (r200)->radeon.glCtx );		\
-   }							\
-} while (0)
-
-/* Command lengths.  Note that any time you ensure ELTS_BUFSZ or VBUF_BUFSZ
- * are available, you will also be adding an rmesa->state.max_state_size because
- * r200EmitState is called from within r200EmitVbufPrim and r200FlushElts.
- */
-#define AOS_BUFSZ(nr)	((3 + ((nr / 2) * 3) + ((nr & 1) * 2)) * sizeof(int))
-#define VERT_AOS_BUFSZ	(5 * sizeof(int))
-#define ELTS_BUFSZ(nr)	(12 + nr * 2)
-#define VBUF_BUFSZ	(3 * sizeof(int))
-
-/* Ensure that a minimum amount of space is available in the command buffer.
- * This is used to ensure atomicity of state updates with the rendering requests
- * that rely on them.
- *
- * An alternative would be to implement a "soft lock" such that when the buffer
- * wraps at an inopportune time, we grab the lock, flush the current buffer,
- * and hang on to the lock until the critical section is finished and we flush
- * the buffer again and unlock.
- */
-static __inline void r200EnsureCmdBufSpace(r200ContextPtr rmesa, int bytes)
-{
-	if (rmesa->store.cmd_used + bytes > R200_CMD_BUF_SZ)
-		r200FlushCmdBuf(rmesa, __FUNCTION__);
-	assert(bytes <= R200_CMD_BUF_SZ);
-}
-
-/* Alloc space in the command buffer
- */
-static __inline char *r200AllocCmdBuf(r200ContextPtr rmesa,
-				      int bytes, const char *where)
-{
-	char *head;
-
-	if (rmesa->store.cmd_used + bytes > R200_CMD_BUF_SZ)
-		r200FlushCmdBuf(rmesa, where);
-
-	head = rmesa->store.cmd_buf + rmesa->store.cmd_used;
-	rmesa->store.cmd_used += bytes;
-	assert(rmesa->store.cmd_used <= R200_CMD_BUF_SZ);
-	return head;
-}
-
-#endif				/* __R200_IOCTL_H__ */
diff --git a/src/mesa/drivers/dri/r300/r200_reg.h b/src/mesa/drivers/dri/r300/r200_reg.h
deleted file mode 100644
index 1336e961ac6..00000000000
--- a/src/mesa/drivers/dri/r300/r200_reg.h
+++ /dev/null
@@ -1,1423 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/r200/r200_reg.h,v 1.2 2002/12/16 16:18:54 dawes Exp $ */
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-*/
-
-#ifndef _R200_REG_H_
-#define _R200_REG_H_
-
-#define R200_PP_MISC                      0x1c14
-#define     R200_REF_ALPHA_MASK        0x000000ff
-#define     R200_ALPHA_TEST_FAIL       (0 << 8)
-#define     R200_ALPHA_TEST_LESS       (1 << 8)
-#define     R200_ALPHA_TEST_LEQUAL     (2 << 8)
-#define     R200_ALPHA_TEST_EQUAL      (3 << 8)
-#define     R200_ALPHA_TEST_GEQUAL     (4 << 8)
-#define     R200_ALPHA_TEST_GREATER    (5 << 8)
-#define     R200_ALPHA_TEST_NEQUAL     (6 << 8)
-#define     R200_ALPHA_TEST_PASS       (7 << 8)
-#define     R200_ALPHA_TEST_OP_MASK    (7 << 8)
-#define     R200_CHROMA_FUNC_FAIL      (0 << 16)
-#define     R200_CHROMA_FUNC_PASS      (1 << 16)
-#define     R200_CHROMA_FUNC_NEQUAL    (2 << 16)
-#define     R200_CHROMA_FUNC_EQUAL     (3 << 16)
-#define     R200_CHROMA_KEY_NEAREST    (0 << 18)
-#define     R200_CHROMA_KEY_ZERO       (1 << 18)
-#define     R200_RIGHT_HAND_CUBE_D3D   (0 << 24)
-#define     R200_RIGHT_HAND_CUBE_OGL   (1 << 24)
-#define R200_PP_FOG_COLOR                 0x1c18
-#define     R200_FOG_COLOR_MASK        0x00ffffff
-#define     R200_FOG_VERTEX            (0 << 24)
-#define     R200_FOG_TABLE             (1 << 24)
-#define     R200_FOG_USE_DEPTH         (0 << 25)
-#define     R200_FOG_USE_W             (1 << 25)
-#define     R200_FOG_USE_DIFFUSE_ALPHA (2 << 25)
-#define     R200_FOG_USE_SPEC_ALPHA    (3 << 25)
-#define     R200_FOG_USE_VTX_FOG       (4 << 25)
-#define     R200_FOG_USE_MASK          (7 << 25)
-#define R200_RE_SOLID_COLOR               0x1c1c
-#define R200_RB3D_BLENDCNTL               0x1c20
-#define     R200_COMB_FCN_MASK                    (7  << 12)
-#define     R200_COMB_FCN_ADD_CLAMP               (0  << 12)
-#define     R200_COMB_FCN_ADD_NOCLAMP             (1  << 12)
-#define     R200_COMB_FCN_SUB_CLAMP               (2  << 12)
-#define     R200_COMB_FCN_SUB_NOCLAMP             (3  << 12)
-#define     R200_COMB_FCN_MIN                     (4  << 12)
-#define     R200_COMB_FCN_MAX                     (5  << 12)
-#define     R200_COMB_FCN_RSUB_CLAMP              (6  << 12)
-#define     R200_COMB_FCN_RSUB_NOCLAMP            (7  << 12)
-#define       R200_BLEND_GL_ZERO                  (32)
-#define       R200_BLEND_GL_ONE                   (33)
-#define       R200_BLEND_GL_SRC_COLOR             (34)
-#define       R200_BLEND_GL_ONE_MINUS_SRC_COLOR   (35)
-#define       R200_BLEND_GL_DST_COLOR             (36)
-#define       R200_BLEND_GL_ONE_MINUS_DST_COLOR   (37)
-#define       R200_BLEND_GL_SRC_ALPHA             (38)
-#define       R200_BLEND_GL_ONE_MINUS_SRC_ALPHA   (39)
-#define       R200_BLEND_GL_DST_ALPHA             (40)
-#define       R200_BLEND_GL_ONE_MINUS_DST_ALPHA   (41)
-#define       R200_BLEND_GL_SRC_ALPHA_SATURATE    (42)	/* src factor only */
-#define       R200_BLEND_GL_CONST_COLOR           (43)
-#define       R200_BLEND_GL_ONE_MINUS_CONST_COLOR (44)
-#define       R200_BLEND_GL_CONST_ALPHA           (45)
-#define       R200_BLEND_GL_ONE_MINUS_CONST_ALPHA (46)
-#define       R200_BLEND_MASK                     (63)
-#define     R200_SRC_BLEND_SHIFT                  (16)
-#define     R200_DST_BLEND_SHIFT                  (24)
-#define R200_RB3D_DEPTHOFFSET             0x1c24
-#define R200_RB3D_DEPTHPITCH              0x1c28
-#define     R200_DEPTHPITCH_MASK         0x00001ff8
-#define     R200_DEPTH_ENDIAN_NO_SWAP    (0 << 18)
-#define     R200_DEPTH_ENDIAN_WORD_SWAP  (1 << 18)
-#define     R200_DEPTH_ENDIAN_DWORD_SWAP (2 << 18)
-#define R200_RB3D_ZSTENCILCNTL            0x1c2c
-#define     R200_DEPTH_FORMAT_MASK          (0xf << 0)
-#define     R200_DEPTH_FORMAT_16BIT_INT_Z   (0  <<  0)
-#define     R200_DEPTH_FORMAT_24BIT_INT_Z   (2  <<  0)
-#define     R200_DEPTH_FORMAT_24BIT_FLOAT_Z (3  <<  0)
-#define     R200_DEPTH_FORMAT_32BIT_INT_Z   (4  <<  0)
-#define     R200_DEPTH_FORMAT_32BIT_FLOAT_Z (5  <<  0)
-#define     R200_DEPTH_FORMAT_24BIT_FLOAT_W (9  <<  0)
-#define     R200_DEPTH_FORMAT_32BIT_FLOAT_W (11 <<  0)
-#define     R200_Z_TEST_NEVER               (0  <<  4)
-#define     R200_Z_TEST_LESS                (1  <<  4)
-#define     R200_Z_TEST_LEQUAL              (2  <<  4)
-#define     R200_Z_TEST_EQUAL               (3  <<  4)
-#define     R200_Z_TEST_GEQUAL              (4  <<  4)
-#define     R200_Z_TEST_GREATER             (5  <<  4)
-#define     R200_Z_TEST_NEQUAL              (6  <<  4)
-#define     R200_Z_TEST_ALWAYS              (7  <<  4)
-#define     R200_Z_TEST_MASK                (7  <<  4)
-#define     R200_STENCIL_TEST_NEVER         (0  << 12)
-#define     R200_STENCIL_TEST_LESS          (1  << 12)
-#define     R200_STENCIL_TEST_LEQUAL        (2  << 12)
-#define     R200_STENCIL_TEST_EQUAL         (3  << 12)
-#define     R200_STENCIL_TEST_GEQUAL        (4  << 12)
-#define     R200_STENCIL_TEST_GREATER       (5  << 12)
-#define     R200_STENCIL_TEST_NEQUAL        (6  << 12)
-#define     R200_STENCIL_TEST_ALWAYS        (7  << 12)
-#define     R200_STENCIL_TEST_MASK          (0x7 << 12)
-#define     R200_STENCIL_FAIL_KEEP          (0  << 16)
-#define     R200_STENCIL_FAIL_ZERO          (1  << 16)
-#define     R200_STENCIL_FAIL_REPLACE       (2  << 16)
-#define     R200_STENCIL_FAIL_INC           (3  << 16)
-#define     R200_STENCIL_FAIL_DEC           (4  << 16)
-#define     R200_STENCIL_FAIL_INVERT        (5  << 16)
-#define     R200_STENCIL_FAIL_INC_WRAP      (6  << 16)
-#define     R200_STENCIL_FAIL_DEC_WRAP      (7  << 16)
-#define     R200_STENCIL_FAIL_MASK          (0x7 << 16)
-#define     R200_STENCIL_ZPASS_KEEP         (0  << 20)
-#define     R200_STENCIL_ZPASS_ZERO         (1  << 20)
-#define     R200_STENCIL_ZPASS_REPLACE      (2  << 20)
-#define     R200_STENCIL_ZPASS_INC          (3  << 20)
-#define     R200_STENCIL_ZPASS_DEC          (4  << 20)
-#define     R200_STENCIL_ZPASS_INVERT       (5  << 20)
-#define     R200_STENCIL_ZPASS_INC_WRAP     (6  << 20)
-#define     R200_STENCIL_ZPASS_DEC_WRAP     (7  << 20)
-#define     R200_STENCIL_ZPASS_MASK         (0x7 << 20)
-#define     R200_STENCIL_ZFAIL_KEEP         (0  << 24)
-#define     R200_STENCIL_ZFAIL_ZERO         (1  << 24)
-#define     R200_STENCIL_ZFAIL_REPLACE      (2  << 24)
-#define     R200_STENCIL_ZFAIL_INC          (3  << 24)
-#define     R200_STENCIL_ZFAIL_DEC          (4  << 24)
-#define     R200_STENCIL_ZFAIL_INVERT       (5  << 24)
-#define     R200_STENCIL_ZFAIL_INC_WRAP     (6  << 24)
-#define     R200_STENCIL_ZFAIL_DEC_WRAP     (7  << 24)
-#define     R200_STENCIL_ZFAIL_MASK         (0x7 << 24)
-#define     R200_Z_WRITE_ENABLE             (1  << 30)
-/*gap*/
-#define R200_PP_CNTL                      0x1c38
-#define     R200_TEX_0_ENABLE                         0x00000010
-#define     R200_TEX_1_ENABLE                         0x00000020
-#define     R200_TEX_2_ENABLE                         0x00000040
-#define     R200_TEX_3_ENABLE                         0x00000080
-#define     R200_TEX_4_ENABLE                         0x00000100
-#define     R200_TEX_5_ENABLE                         0x00000200
-#define     R200_TEX_ENABLE_MASK                      0x000003f0
-#define     R200_FILTER_ROUND_MODE_MASK               0x00000400
-#define     R200_TEX_BLEND_7_ENABLE                   0x00000800
-#define     R200_TEX_BLEND_0_ENABLE                   0x00001000
-#define     R200_TEX_BLEND_1_ENABLE                   0x00002000
-#define     R200_TEX_BLEND_2_ENABLE                   0x00004000
-#define     R200_TEX_BLEND_3_ENABLE                   0x00008000
-#define     R200_TEX_BLEND_4_ENABLE                   0x00010000
-#define     R200_TEX_BLEND_5_ENABLE                   0x00020000
-#define     R200_TEX_BLEND_6_ENABLE                   0x00040000
-#define     R200_MULTI_PASS_ENABLE                    0x00080000
-#define     R200_SPECULAR_ENABLE                      0x00200000
-#define     R200_FOG_ENABLE                           0x00400000
-#define     R200_ALPHA_TEST_ENABLE                    0x00800000
-#define     R200_ANTI_ALIAS_NONE                       0x00000000
-#define     R200_ANTI_ALIAS_LINE                       0x01000000
-#define     R200_ANTI_ALIAS_POLY                       0x02000000
-#define     R200_ANTI_ALIAS_MASK                       0x03000000
-#define R200_RB3D_CNTL                    0x1c3c
-#define     R200_ALPHA_BLEND_ENABLE       (1  <<  0)
-#define     R200_PLANE_MASK_ENABLE        (1  <<  1)
-#define     R200_DITHER_ENABLE            (1  <<  2)
-#define     R200_ROUND_ENABLE             (1  <<  3)
-#define     R200_SCALE_DITHER_ENABLE      (1  <<  4)
-#define     R200_DITHER_INIT              (1  <<  5)
-#define     R200_ROP_ENABLE               (1  <<  6)
-#define     R200_STENCIL_ENABLE           (1  <<  7)
-#define     R200_Z_ENABLE                 (1  <<  8)
-#define     R200_DEPTH_XZ_OFFEST_ENABLE   (1  <<  9)
-#define     R200_COLOR_FORMAT_ARGB1555    (3  << 10)
-#define     R200_COLOR_FORMAT_RGB565      (4  << 10)
-#define     R200_COLOR_FORMAT_ARGB8888    (6  << 10)
-#define     R200_COLOR_FORMAT_RGB332      (7  << 10)
-#define     R200_COLOR_FORMAT_Y8          (8  << 10)
-#define     R200_COLOR_FORMAT_RGB8        (9  << 10)
-#define     R200_COLOR_FORMAT_YUV422_VYUY (11 << 10)
-#define     R200_COLOR_FORMAT_YUV422_YVYU (12 << 10)
-#define     R200_COLOR_FORMAT_aYUV444     (14 << 10)
-#define     R200_COLOR_FORMAT_ARGB4444    (15 << 10)
-#define     R200_CLRCMP_FLIP_ENABLE       (1  << 14)
-#define     R200_SEPARATE_ALPHA_ENABLE    (1  << 16)
-#define R200_RB3D_COLOROFFSET             0x1c40
-#define     R200_COLOROFFSET_MASK      0xfffffff0
-#define R200_RE_WIDTH_HEIGHT              0x1c44
-#define     R200_RE_WIDTH_SHIFT        0
-#define     R200_RE_HEIGHT_SHIFT       16
-#define R200_RB3D_COLORPITCH              0x1c48
-#define     R200_COLORPITCH_MASK         0x000001ff8
-#define     R200_COLOR_ENDIAN_NO_SWAP    (0 << 18)
-#define     R200_COLOR_ENDIAN_WORD_SWAP  (1 << 18)
-#define     R200_COLOR_ENDIAN_DWORD_SWAP (2 << 18)
-#define R200_SE_CNTL                      0x1c4c
-#define     R200_FFACE_CULL_CW          (0 <<  0)
-#define     R200_FFACE_CULL_CCW         (1 <<  0)
-#define     R200_FFACE_CULL_DIR_MASK    (1 <<  0)
-#define     R200_BFACE_CULL             (0 <<  1)
-#define     R200_BFACE_SOLID            (3 <<  1)
-#define     R200_FFACE_CULL             (0 <<  3)
-#define     R200_FFACE_SOLID            (3 <<  3)
-#define     R200_FFACE_CULL_MASK        (3 <<  3)
-#define     R200_FLAT_SHADE_VTX_0       (0 <<  6)
-#define     R200_FLAT_SHADE_VTX_1       (1 <<  6)
-#define     R200_FLAT_SHADE_VTX_2       (2 <<  6)
-#define     R200_FLAT_SHADE_VTX_LAST    (3 <<  6)
-#define     R200_DIFFUSE_SHADE_SOLID    (0 <<  8)
-#define     R200_DIFFUSE_SHADE_FLAT     (1 <<  8)
-#define     R200_DIFFUSE_SHADE_GOURAUD  (2 <<  8)
-#define     R200_DIFFUSE_SHADE_MASK     (3 <<  8)
-#define     R200_ALPHA_SHADE_SOLID      (0 << 10)
-#define     R200_ALPHA_SHADE_FLAT       (1 << 10)
-#define     R200_ALPHA_SHADE_GOURAUD    (2 << 10)
-#define     R200_ALPHA_SHADE_MASK       (3 << 10)
-#define     R200_SPECULAR_SHADE_SOLID   (0 << 12)
-#define     R200_SPECULAR_SHADE_FLAT    (1 << 12)
-#define     R200_SPECULAR_SHADE_GOURAUD (2 << 12)
-#define     R200_SPECULAR_SHADE_MASK    (3 << 12)
-#define     R200_FOG_SHADE_SOLID        (0 << 14)
-#define     R200_FOG_SHADE_FLAT         (1 << 14)
-#define     R200_FOG_SHADE_GOURAUD      (2 << 14)
-#define     R200_FOG_SHADE_MASK         (3 << 14)
-#define     R200_ZBIAS_ENABLE_POINT     (1 << 16)
-#define     R200_ZBIAS_ENABLE_LINE      (1 << 17)
-#define     R200_ZBIAS_ENABLE_TRI       (1 << 18)
-#define     R200_WIDELINE_ENABLE        (1 << 20)
-#define     R200_VTX_PIX_CENTER_D3D     (0 << 27)
-#define     R200_VTX_PIX_CENTER_OGL     (1 << 27)
-#define     R200_ROUND_MODE_TRUNC       (0 << 28)
-#define     R200_ROUND_MODE_ROUND       (1 << 28)
-#define     R200_ROUND_MODE_ROUND_EVEN  (2 << 28)
-#define     R200_ROUND_MODE_ROUND_ODD   (3 << 28)
-#define     R200_ROUND_PREC_16TH_PIX    (0 << 30)
-#define     R200_ROUND_PREC_8TH_PIX     (1 << 30)
-#define     R200_ROUND_PREC_4TH_PIX     (2 << 30)
-#define     R200_ROUND_PREC_HALF_PIX    (3 << 30)
-#define R200_RE_CNTL                      0x1c50
-#define     R200_STIPPLE_ENABLE                     0x1
-#define     R200_SCISSOR_ENABLE                     0x2
-#define     R200_PATTERN_ENABLE                     0x4
-#define     R200_PERSPECTIVE_ENABLE                 0x8
-#define     R200_POINT_SMOOTH                       0x20
-#define     R200_VTX_STQ0_D3D                       0x00010000
-#define     R200_VTX_STQ1_D3D                       0x00040000
-#define     R200_VTX_STQ2_D3D                       0x00100000
-#define     R200_VTX_STQ3_D3D                       0x00400000
-#define     R200_VTX_STQ4_D3D                       0x01000000
-#define     R200_VTX_STQ5_D3D                       0x04000000
-/* gap */
-#define R200_RE_STIPPLE_ADDR              0x1cc8
-#define R200_RE_STIPPLE_DATA              0x1ccc
-#define R200_RE_LINE_PATTERN              0x1cd0
-#define     R200_LINE_PATTERN_MASK             0x0000ffff
-#define     R200_LINE_REPEAT_COUNT_SHIFT       16
-#define     R200_LINE_PATTERN_START_SHIFT      24
-#define     R200_LINE_PATTERN_LITTLE_BIT_ORDER (0 << 28)
-#define     R200_LINE_PATTERN_BIG_BIT_ORDER    (1 << 28)
-#define     R200_LINE_PATTERN_AUTO_RESET       (1 << 29)
-#define R200_RE_LINE_STATE                0x1cd4
-#define     R200_LINE_CURRENT_PTR_SHIFT       0
-#define     R200_LINE_CURRENT_COUNT_SHIFT     8
-#define R200_RE_SCISSOR_TL_0              0x1cd8
-#define R200_RE_SCISSOR_BR_0              0x1cdc
-#define R200_RE_SCISSOR_TL_1              0x1ce0
-#define R200_RE_SCISSOR_BR_1              0x1ce4
-#define R200_RE_SCISSOR_TL_2              0x1ce8
-#define R200_RE_SCISSOR_BR_2              0x1cec
-/* gap */
-#define R200_RB3D_DEPTHXY_OFFSET          0x1d60
-#define     R200_DEPTHX_SHIFT  0
-#define     R200_DEPTHY_SHIFT  16
-/* gap */
-#define R200_RB3D_STENCILREFMASK          0x1d7c
-#define     R200_STENCIL_REF_SHIFT           0
-#define     R200_STENCIL_REF_MASK            (0xff << 0)
-#define     R200_STENCIL_MASK_SHIFT          16
-#define     R200_STENCIL_VALUE_MASK          (0xff << 16)
-#define     R200_STENCIL_WRITEMASK_SHIFT     24
-#define     R200_STENCIL_WRITE_MASK          (0xff << 24)
-#define R200_RB3D_ROPCNTL                 0x1d80
-#define     R200_ROP_MASK                    (15 << 8)
-#define     R200_ROP_CLEAR                   (0  << 8)
-#define     R200_ROP_NOR                     (1  << 8)
-#define     R200_ROP_AND_INVERTED            (2  << 8)
-#define     R200_ROP_COPY_INVERTED           (3  << 8)
-#define     R200_ROP_AND_REVERSE             (4  << 8)
-#define     R200_ROP_INVERT                  (5  << 8)
-#define     R200_ROP_XOR                     (6  << 8)
-#define     R200_ROP_NAND                    (7  << 8)
-#define     R200_ROP_AND                     (8  << 8)
-#define     R200_ROP_EQUIV                   (9  << 8)
-#define     R200_ROP_NOOP                    (10 << 8)
-#define     R200_ROP_OR_INVERTED             (11 << 8)
-#define     R200_ROP_COPY                    (12 << 8)
-#define     R200_ROP_OR_REVERSE              (13 << 8)
-#define     R200_ROP_OR                      (14 << 8)
-#define     R200_ROP_SET                     (15 << 8)
-#define R200_RB3D_PLANEMASK               0x1d84
-/* gap */
-#define R200_SE_VPORT_XSCALE              0x1d98
-#define R200_SE_VPORT_XOFFSET             0x1d9c
-#define R200_SE_VPORT_YSCALE              0x1da0
-#define R200_SE_VPORT_YOFFSET             0x1da4
-#define R200_SE_VPORT_ZSCALE              0x1da8
-#define R200_SE_VPORT_ZOFFSET             0x1dac
-#define R200_SE_ZBIAS_FACTOR              0x1db0
-#define R200_SE_ZBIAS_CONSTANT            0x1db4
-#define R200_SE_LINE_WIDTH                0x1db8
-#define	    R200_LINE_WIDTH_SHIFT                   0x00000000
-#define	    R200_MINPOINTSIZE_SHIFT                 0x00000010
-/* gap */
-#define R200_SE_VAP_CNTL                           0x2080
-#define     R200_VAP_TCL_ENABLE                       0x00000001
-#define     R200_VAP_SINGLE_BUF_STATE_ENABLE          0x00000010
-#define     R200_VAP_FORCE_W_TO_ONE                   0x00010000
-#define     R200_VAP_D3D_TEX_DEFAULT                  0x00020000
-#define     R200_VAP_VF_MAX_VTX_NUM__SHIFT            18
-#define     R200_VAP_DX_CLIP_SPACE_DEF                0x00400000
-#define R200_SE_VF_CNTL                           0x2084
-#define     R200_VF_PRIM_NONE                         0x00000000
-#define     R200_VF_PRIM_POINTS                       0x00000001
-#define     R200_VF_PRIM_LINES                        0x00000002
-#define     R200_VF_PRIM_LINE_STRIP                   0x00000003
-#define     R200_VF_PRIM_TRIANGLES                    0x00000004
-#define     R200_VF_PRIM_TRIANGLE_FAN                 0x00000005
-#define     R200_VF_PRIM_TRIANGLE_STRIP               0x00000006
-#define     R200_VF_PRIM_RECT_LIST                    0x00000008
-#define     R200_VF_PRIM_3VRT_POINTS                  0x00000009
-#define     R200_VF_PRIM_3VRT_LINES                   0x0000000a
-#define     R200_VF_PRIM_POINT_SPRITES                0x0000000b
-#define     R200_VF_PRIM_LINE_LOOP                    0x0000000c
-#define     R200_VF_PRIM_QUADS                        0x0000000d
-#define     R200_VF_PRIM_QUAD_STRIP                   0x0000000e
-#define     R200_VF_PRIM_POLYGON                      0x0000000f
-#define     R200_VF_PRIM_MASK                         0x0000000f
-#define     R200_VF_PRIM_WALK_IND                     0x00000010
-#define     R200_VF_PRIM_WALK_LIST                    0x00000020
-#define     R200_VF_PRIM_WALK_RING                    0x00000030
-#define     R200_VF_PRIM_WALK_MASK                    0x00000030
-#define     R200_VF_COLOR_ORDER_RGBA                  0x00000040
-#define     R200_VF_TCL_OUTPUT_VTX_ENABLE             0x00000200
-#define     R200_VF_INDEX_SZ_4                        0x00000800
-#define     R200_VF_VERTEX_NUMBER_MASK                0xffff0000
-#define     R200_VF_VERTEX_NUMBER_SHIFT               16
-#define R200_SE_VTX_FMT_0                 0x2088
-#define     R200_VTX_XY                     0	/* always have xy */
-#define     R200_VTX_Z0                     (1<<0)
-#define     R200_VTX_W0                     (1<<1)
-#define     R200_VTX_WEIGHT_COUNT_SHIFT     (2)
-#define     R200_VTX_PV_MATRIX_SEL          (1<<5)
-#define     R200_VTX_N0                     (1<<6)
-#define     R200_VTX_POINT_SIZE             (1<<7)
-#define     R200_VTX_DISCRETE_FOG           (1<<8)
-#define     R200_VTX_SHININESS_0            (1<<9)
-#define     R200_VTX_SHININESS_1            (1<<10)
-#define       R200_VTX_COLOR_NOT_PRESENT      0
-#define       R200_VTX_PK_RGBA          1
-#define       R200_VTX_FP_RGB           2
-#define       R200_VTX_FP_RGBA          3
-#define       R200_VTX_COLOR_MASK             3
-#define     R200_VTX_COLOR_0_SHIFT          11
-#define     R200_VTX_COLOR_1_SHIFT          13
-#define     R200_VTX_COLOR_2_SHIFT          15
-#define     R200_VTX_COLOR_3_SHIFT          17
-#define     R200_VTX_COLOR_4_SHIFT          19
-#define     R200_VTX_COLOR_5_SHIFT          21
-#define     R200_VTX_COLOR_6_SHIFT          23
-#define     R200_VTX_COLOR_7_SHIFT          25
-#define     R200_VTX_XY1                    (1<<28)
-#define     R200_VTX_Z1                     (1<<29)
-#define     R200_VTX_W1                     (1<<30)
-#define     R200_VTX_N1                     (1<<31)
-#define R200_SE_VTX_FMT_1                 0x208c
-#define     R200_VTX_TEX0_COMP_CNT_SHIFT        0
-#define     R200_VTX_TEX1_COMP_CNT_SHIFT        3
-#define     R200_VTX_TEX2_COMP_CNT_SHIFT        6
-#define     R200_VTX_TEX3_COMP_CNT_SHIFT        9
-#define     R200_VTX_TEX4_COMP_CNT_SHIFT        12
-#define     R200_VTX_TEX5_COMP_CNT_SHIFT        15
-#define R200_SE_TCL_OUTPUT_VTX_FMT_0      0x2090
-#define R200_SE_TCL_OUTPUT_VTX_FMT_1      0x2094
-/* gap */
-#define R200_SE_VTE_CNTL                  0x20b0
-#define     R200_VPORT_X_SCALE_ENA                0x00000001
-#define     R200_VPORT_X_OFFSET_ENA               0x00000002
-#define     R200_VPORT_Y_SCALE_ENA                0x00000004
-#define     R200_VPORT_Y_OFFSET_ENA               0x00000008
-#define     R200_VPORT_Z_SCALE_ENA                0x00000010
-#define     R200_VPORT_Z_OFFSET_ENA               0x00000020
-#define     R200_VTX_XY_FMT                       0x00000100
-#define     R200_VTX_Z_FMT                        0x00000200
-#define     R200_VTX_W0_FMT                       0x00000400
-#define     R200_VTX_W0_NORMALIZE                 0x00000800
-#define     R200_VTX_ST_DENORMALIZED              0x00001000
-/* gap */
-#define R200_SE_VTX_NUM_ARRAYS            0x20c0
-#define R200_SE_VTX_AOS_ATTR01            0x20c4
-#define R200_SE_VTX_AOS_ADDR0             0x20c8
-#define R200_SE_VTX_AOS_ADDR1             0x20cc
-#define R200_SE_VTX_AOS_ATTR23            0x20d0
-#define R200_SE_VTX_AOS_ADDR2             0x20d4
-#define R200_SE_VTX_AOS_ADDR3             0x20d8
-#define R200_SE_VTX_AOS_ATTR45            0x20dc
-#define R200_SE_VTX_AOS_ADDR4             0x20e0
-#define R200_SE_VTX_AOS_ADDR5             0x20e4
-#define R200_SE_VTX_AOS_ATTR67            0x20e8
-#define R200_SE_VTX_AOS_ADDR6             0x20ec
-#define R200_SE_VTX_AOS_ADDR7             0x20f0
-#define R200_SE_VTX_AOS_ATTR89            0x20f4
-#define R200_SE_VTX_AOS_ADDR8             0x20f8
-#define R200_SE_VTX_AOS_ADDR9             0x20fc
-#define R200_SE_VTX_AOS_ATTR1011          0x2100
-#define R200_SE_VTX_AOS_ADDR10            0x2104
-#define R200_SE_VTX_AOS_ADDR11            0x2108
-#define R200_SE_VF_MAX_VTX_INDX           0x210c
-#define R200_SE_VF_MIN_VTX_INDX           0x2110
-/* gap */
-#define R200_SE_VAP_CNTL_STATUS           0x2140
-#define     R200_VC_NO_SWAP                  (0 << 0)
-#define     R200_VC_16BIT_SWAP               (1 << 0)
-#define     R200_VC_32BIT_SWAP               (2 << 0)
-/* gap */
-#define R200_SE_VTX_STATE_CNTL                     0x2180
-#define     R200_VSC_COLOR_0_ASSEMBLY_CNTL_SHIFT    0x00000000
-#define     R200_VSC_COLOR_1_ASSEMBLY_CNTL_SHIFT    0x00000002
-#define     R200_VSC_COLOR_2_ASSEMBLY_CNTL_SHIFT    0x00000004
-#define     R200_VSC_COLOR_3_ASSEMBLY_CNTL_SHIFT    0x00000006
-#define     R200_VSC_COLOR_4_ASSEMBLY_CNTL_SHIFT    0x00000008
-#define     R200_VSC_COLOR_5_ASSEMBLY_CNTL_SHIFT    0x0000000a
-#define     R200_VSC_COLOR_6_ASSEMBLY_CNTL_SHIFT    0x0000000c
-#define     R200_VSC_COLOR_7_ASSEMBLY_CNTL_SHIFT    0x0000000e
-#define     R200_VSC_UPDATE_USER_COLOR_0_ENABLE    0x00010000
-#define     R200_VSC_UPDATE_USER_COLOR_1_ENABLE    0x00020000
-/* gap */
-#define R200_SE_TCL_VECTOR_INDX_REG                0x2200
-#define R200_SE_TCL_VECTOR_DATA_REG                0x2204
-#define R200_SE_TCL_SCALAR_INDX_REG                0x2208
-#define R200_SE_TCL_SCALAR_DATA_REG                0x220c
-/* gap */
-#define R200_SE_TCL_MATRIX_SEL_0                   0x2230
-#define     R200_MODELVIEW_0_SHIFT           (0)
-#define     R200_MODELVIEW_1_SHIFT           (8)
-#define     R200_MODELVIEW_2_SHIFT           (16)
-#define     R200_MODELVIEW_3_SHIFT           (24)
-#define R200_SE_TCL_MATRIX_SEL_1                   0x2234
-#define     R200_IT_MODELVIEW_0_SHIFT        (0)
-#define     R200_IT_MODELVIEW_1_SHIFT        (8)
-#define     R200_IT_MODELVIEW_2_SHIFT        (16)
-#define     R200_IT_MODELVIEW_3_SHIFT        (24)
-#define R200_SE_TCL_MATRIX_SEL_2                   0x2238
-#define     R200_MODELPROJECT_0_SHIFT         (0)
-#define     R200_MODELPROJECT_1_SHIFT         (8)
-#define     R200_MODELPROJECT_2_SHIFT         (16)
-#define     R200_MODELPROJECT_3_SHIFT         (24)
-#define R200_SE_TCL_MATRIX_SEL_3                   0x223c
-#define     R200_TEXMAT_0_SHIFT    0
-#define     R200_TEXMAT_1_SHIFT    8
-#define     R200_TEXMAT_2_SHIFT    16
-#define     R200_TEXMAT_3_SHIFT    24
-#define R200_SE_TCL_MATRIX_SEL_4                   0x2240
-#define     R200_TEXMAT_4_SHIFT    0
-#define     R200_TEXMAT_5_SHIFT    8
-/* gap */
-#define R200_SE_TCL_OUTPUT_VTX_COMP_SEL     0x2250
-#define     R200_OUTPUT_XYZW                    (1<<0)
-#define     R200_OUTPUT_COLOR_0                 (1<<8)
-#define     R200_OUTPUT_COLOR_1                 (1<<9)
-#define     R200_OUTPUT_TEX_0                   (1<<16)
-#define     R200_OUTPUT_TEX_1                   (1<<17)
-#define     R200_OUTPUT_TEX_2                   (1<<18)
-#define     R200_OUTPUT_TEX_3                   (1<<19)
-#define     R200_OUTPUT_TEX_4                   (1<<20)
-#define     R200_OUTPUT_TEX_5                   (1<<21)
-#define     R200_OUTPUT_TEX_MASK                (0x3f<<16)
-#define     R200_OUTPUT_DISCRETE_FOG            (1<<24)
-#define     R200_OUTPUT_PT_SIZE                 (1<<25)
-#define     R200_FORCE_INORDER_PROC             (1<<31)
-#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0  0x2254
-#define	    R200_VERTEX_POSITION_ADDR__SHIFT     0x00000000
-#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_1  0x2258
-#define	    R200_VTX_COLOR_0_ADDR__SHIFT         0x00000000
-#define	    R200_VTX_COLOR_1_ADDR__SHIFT         0x00000008
-#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_2  0x225c
-#define	    R200_VTX_TEX_0_ADDR__SHIFT           0x00000000
-#define	    R200_VTX_TEX_1_ADDR__SHIFT           0x00000008
-#define	    R200_VTX_TEX_2_ADDR__SHIFT           0x00000010
-#define	    R200_VTX_TEX_3_ADDR__SHIFT           0x00000018
-#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_3  0x2260
-#define	    R200_VTX_TEX_4_ADDR__SHIFT           0x00000000
-#define	    R200_VTX_TEX_5_ADDR__SHIFT           0x00000008
-
-/* gap */
-#define R200_SE_TCL_LIGHT_MODEL_CTL_0       0x2268
-#define     R200_LIGHTING_ENABLE                (1<<0)
-#define     R200_LIGHT_IN_MODELSPACE            (1<<1)
-#define     R200_LOCAL_VIEWER                   (1<<2)
-#define     R200_NORMALIZE_NORMALS              (1<<3)
-#define     R200_RESCALE_NORMALS                (1<<4)
-#define     R200_SPECULAR_LIGHTS                (1<<5)
-#define     R200_DIFFUSE_SPECULAR_COMBINE       (1<<6)
-#define     R200_LIGHT_ALPHA                    (1<<7)
-#define     R200_LOCAL_LIGHT_VEC_GL             (1<<8)
-#define     R200_LIGHT_NO_NORMAL_AMBIENT_ONLY   (1<<9)
-#define     R200_LIGHT_TWOSIDE                  (1<<10)
-#define     R200_FRONT_SHININESS_SOURCE_SHIFT       (0xb)
-#define     R200_BACK_SHININESS_SOURCE_SHIFT        (0xd)
-#define       R200_LM0_SOURCE_MATERIAL_0           (0)
-#define       R200_LM0_SOURCE_MATERIAL_1           (1)
-#define       R200_LM0_SOURCE_VERTEX_SHININESS_0   (2)
-#define       R200_LM0_SOURCE_VERTEX_SHININESS_1   (3)
-#define R200_SE_TCL_LIGHT_MODEL_CTL_1       0x226c
-#define       R200_LM1_SOURCE_LIGHT_PREMULT        (0)
-#define       R200_LM1_SOURCE_MATERIAL_0           (1)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_0       (2)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_1       (3)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_2       (4)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_3       (5)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_4       (6)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_5       (7)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_6       (8)
-#define       R200_LM1_SOURCE_VERTEX_COLOR_7       (9)
-#define       R200_LM1_SOURCE_MATERIAL_1           (0xf)
-#define     R200_FRONT_EMISSIVE_SOURCE_SHIFT        (0)
-#define     R200_FRONT_AMBIENT_SOURCE_SHIFT         (4)
-#define     R200_FRONT_DIFFUSE_SOURCE_SHIFT         (8)
-#define     R200_FRONT_SPECULAR_SOURCE_SHIFT        (12)
-#define     R200_BACK_EMISSIVE_SOURCE_SHIFT         (16)
-#define     R200_BACK_AMBIENT_SOURCE_SHIFT          (20)
-#define     R200_BACK_DIFFUSE_SOURCE_SHIFT          (24)
-#define     R200_BACK_SPECULAR_SOURCE_SHIFT         (28)
-#define R200_SE_TCL_PER_LIGHT_CTL_0       0x2270
-#define     R200_LIGHT_0_ENABLE                    (1<<0)
-#define     R200_LIGHT_0_ENABLE_AMBIENT            (1<<1)
-#define     R200_LIGHT_0_ENABLE_SPECULAR           (1<<2)
-#define     R200_LIGHT_0_IS_LOCAL                  (1<<3)
-#define     R200_LIGHT_0_IS_SPOT                   (1<<4)
-#define     R200_LIGHT_0_DUAL_CONE                 (1<<5)
-#define     R200_LIGHT_0_ENABLE_RANGE_ATTEN        (1<<6)
-#define     R200_LIGHT_0_CONSTANT_RANGE_ATTEN      (1<<7)
-#define     R200_LIGHT_1_ENABLE                    (1<<16)
-#define     R200_LIGHT_1_ENABLE_AMBIENT            (1<<17)
-#define     R200_LIGHT_1_ENABLE_SPECULAR           (1<<18)
-#define     R200_LIGHT_1_IS_LOCAL                  (1<<19)
-#define     R200_LIGHT_1_IS_SPOT                   (1<<20)
-#define     R200_LIGHT_1_DUAL_CONE                 (1<<21)
-#define     R200_LIGHT_1_ENABLE_RANGE_ATTEN        (1<<22)
-#define     R200_LIGHT_1_CONSTANT_RANGE_ATTEN      (1<<23)
-#define     R200_LIGHT_0_SHIFT                   (0)
-#define     R200_LIGHT_1_SHIFT                   (16)
-#define R200_SE_TCL_PER_LIGHT_CTL_1       0x2274
-#define     R200_LIGHT_2_SHIFT                   (0)
-#define     R200_LIGHT_3_SHIFT                   (16)
-#define R200_SE_TCL_PER_LIGHT_CTL_2       0x2278
-#define     R200_LIGHT_4_SHIFT                   (0)
-#define     R200_LIGHT_5_SHIFT                   (16)
-#define R200_SE_TCL_PER_LIGHT_CTL_3       0x227c
-#define     R200_LIGHT_6_SHIFT                   (0)
-#define     R200_LIGHT_7_SHIFT                   (16)
-/* gap */
-#define R200_SE_TCL_TEX_PROC_CTL_2        0x22a8
-#define     R200_TEXGEN_0_COMP_MASK_SHIFT        (0)
-#define     R200_TEXGEN_1_COMP_MASK_SHIFT        (4)
-#define     R200_TEXGEN_2_COMP_MASK_SHIFT        (8)
-#define     R200_TEXGEN_3_COMP_MASK_SHIFT        (12)
-#define     R200_TEXGEN_4_COMP_MASK_SHIFT        (16)
-#define     R200_TEXGEN_5_COMP_MASK_SHIFT        (20)
-#define R200_SE_TCL_TEX_PROC_CTL_3        0x22ac
-#define     R200_TEXGEN_0_INPUT_TEX_SHIFT        (0)
-#define     R200_TEXGEN_1_INPUT_TEX_SHIFT        (4)
-#define     R200_TEXGEN_2_INPUT_TEX_SHIFT        (8)
-#define     R200_TEXGEN_3_INPUT_TEX_SHIFT        (12)
-#define     R200_TEXGEN_4_INPUT_TEX_SHIFT        (16)
-#define     R200_TEXGEN_5_INPUT_TEX_SHIFT        (20)
-#define R200_SE_TCL_TEX_PROC_CTL_0        0x22b0
-#define     R200_TEXGEN_TEXMAT_0_ENABLE         (1<<0)
-#define     R200_TEXGEN_TEXMAT_1_ENABLE         (1<<1)
-#define     R200_TEXGEN_TEXMAT_2_ENABLE         (1<<2)
-#define     R200_TEXGEN_TEXMAT_3_ENABLE         (1<<3)
-#define     R200_TEXGEN_TEXMAT_4_ENABLE         (1<<4)
-#define     R200_TEXGEN_TEXMAT_5_ENABLE         (1<<5)
-#define     R200_TEXMAT_0_ENABLE                (1<<8)
-#define     R200_TEXMAT_1_ENABLE                (1<<9)
-#define     R200_TEXMAT_2_ENABLE                (1<<10)
-#define     R200_TEXMAT_3_ENABLE                (1<<11)
-#define     R200_TEXMAT_4_ENABLE                (1<<12)
-#define     R200_TEXMAT_5_ENABLE                (1<<13)
-#define     R200_TEXGEN_FORCE_W_TO_ONE          (1<<16)
-#define R200_SE_TCL_TEX_PROC_CTL_1        0x22b4
-#define       R200_TEXGEN_INPUT_MASK           (0xf)
-#define       R200_TEXGEN_INPUT_TEXCOORD_0     (0)
-#define       R200_TEXGEN_INPUT_TEXCOORD_1     (1)
-#define       R200_TEXGEN_INPUT_TEXCOORD_2     (2)
-#define       R200_TEXGEN_INPUT_TEXCOORD_3     (3)
-#define       R200_TEXGEN_INPUT_TEXCOORD_4     (4)
-#define       R200_TEXGEN_INPUT_TEXCOORD_5     (5)
-#define       R200_TEXGEN_INPUT_OBJ            (8)
-#define       R200_TEXGEN_INPUT_EYE            (9)
-#define       R200_TEXGEN_INPUT_EYE_NORMAL     (0xa)
-#define       R200_TEXGEN_INPUT_EYE_REFLECT    (0xb)
-#define       R200_TEXGEN_INPUT_SPHERE         (0xd)
-#define     R200_TEXGEN_0_INPUT_SHIFT        (0)
-#define     R200_TEXGEN_1_INPUT_SHIFT        (4)
-#define     R200_TEXGEN_2_INPUT_SHIFT        (8)
-#define     R200_TEXGEN_3_INPUT_SHIFT        (12)
-#define     R200_TEXGEN_4_INPUT_SHIFT        (16)
-#define     R200_TEXGEN_5_INPUT_SHIFT        (20)
-#define R200_SE_TC_TEX_CYL_WRAP_CTL       0x22b8
-/* gap */
-#define R200_SE_TCL_UCP_VERT_BLEND_CTL    0x22c0
-#define     R200_UCP_IN_CLIP_SPACE              (1<<0)
-#define     R200_UCP_IN_MODEL_SPACE             (1<<1)
-#define     R200_UCP_ENABLE_0                   (1<<2)
-#define     R200_UCP_ENABLE_1                   (1<<3)
-#define     R200_UCP_ENABLE_2                   (1<<4)
-#define     R200_UCP_ENABLE_3                   (1<<5)
-#define     R200_UCP_ENABLE_4                   (1<<6)
-#define     R200_UCP_ENABLE_5                   (1<<7)
-#define     R200_TCL_FOG_MASK                   (3<<8)
-#define     R200_TCL_FOG_DISABLE                (0<<8)
-#define     R200_TCL_FOG_EXP                    (1<<8)
-#define     R200_TCL_FOG_EXP2                   (2<<8)
-#define     R200_TCL_FOG_LINEAR                 (3<<8)
-#define     R200_RNG_BASED_FOG                  (1<<10)
-#define     R200_CLIP_DISABLE                   (1<<11)
-#define     R200_CULL_FRONT_IS_CW               (0<<28)
-#define     R200_CULL_FRONT_IS_CCW              (1<<28)
-#define     R200_CULL_FRONT                     (1<<29)
-#define     R200_CULL_BACK                      (1<<30)
-#define R200_SE_TCL_POINT_SPRITE_CNTL     0x22c4
-/* gap */
-#define R200_SE_VTX_ST_POS_0_X_4                   0x2300
-#define R200_SE_VTX_ST_POS_0_Y_4                   0x2304
-#define R200_SE_VTX_ST_POS_0_Z_4                   0x2308
-#define R200_SE_VTX_ST_POS_0_W_4                   0x230c
-#define R200_SE_VTX_ST_NORM_0_X                    0x2310
-#define R200_SE_VTX_ST_NORM_0_Y                    0x2314
-#define R200_SE_VTX_ST_NORM_0_Z                    0x2318
-#define R200_SE_VTX_ST_PVMS                        0x231c
-#define R200_SE_VTX_ST_CLR_0_R                     0x2320
-#define R200_SE_VTX_ST_CLR_0_G                     0x2324
-#define R200_SE_VTX_ST_CLR_0_B                     0x2328
-#define R200_SE_VTX_ST_CLR_0_A                     0x232c
-#define R200_SE_VTX_ST_CLR_1_R                     0x2330
-#define R200_SE_VTX_ST_CLR_1_G                     0x2334
-#define R200_SE_VTX_ST_CLR_1_B                     0x2338
-#define R200_SE_VTX_ST_CLR_1_A                     0x233c
-#define R200_SE_VTX_ST_CLR_2_R                     0x2340
-#define R200_SE_VTX_ST_CLR_2_G                     0x2344
-#define R200_SE_VTX_ST_CLR_2_B                     0x2348
-#define R200_SE_VTX_ST_CLR_2_A                     0x234c
-#define R200_SE_VTX_ST_CLR_3_R                     0x2350
-#define R200_SE_VTX_ST_CLR_3_G                     0x2354
-#define R200_SE_VTX_ST_CLR_3_B                     0x2358
-#define R200_SE_VTX_ST_CLR_3_A                     0x235c
-#define R200_SE_VTX_ST_CLR_4_R                     0x2360
-#define R200_SE_VTX_ST_CLR_4_G                     0x2364
-#define R200_SE_VTX_ST_CLR_4_B                     0x2368
-#define R200_SE_VTX_ST_CLR_4_A                     0x236c
-#define R200_SE_VTX_ST_CLR_5_R                     0x2370
-#define R200_SE_VTX_ST_CLR_5_G                     0x2374
-#define R200_SE_VTX_ST_CLR_5_B                     0x2378
-#define R200_SE_VTX_ST_CLR_5_A                     0x237c
-#define R200_SE_VTX_ST_CLR_6_R                     0x2380
-#define R200_SE_VTX_ST_CLR_6_G                     0x2384
-#define R200_SE_VTX_ST_CLR_6_B                     0x2388
-#define R200_SE_VTX_ST_CLR_6_A                     0x238c
-#define R200_SE_VTX_ST_CLR_7_R                     0x2390
-#define R200_SE_VTX_ST_CLR_7_G                     0x2394
-#define R200_SE_VTX_ST_CLR_7_B                     0x2398
-#define R200_SE_VTX_ST_CLR_7_A                     0x239c
-#define R200_SE_VTX_ST_TEX_0_S                     0x23a0
-#define R200_SE_VTX_ST_TEX_0_T                     0x23a4
-#define R200_SE_VTX_ST_TEX_0_R                     0x23a8
-#define R200_SE_VTX_ST_TEX_0_Q                     0x23ac
-#define R200_SE_VTX_ST_TEX_1_S                     0x23b0
-#define R200_SE_VTX_ST_TEX_1_T                     0x23b4
-#define R200_SE_VTX_ST_TEX_1_R                     0x23b8
-#define R200_SE_VTX_ST_TEX_1_Q                     0x23bc
-#define R200_SE_VTX_ST_TEX_2_S                     0x23c0
-#define R200_SE_VTX_ST_TEX_2_T                     0x23c4
-#define R200_SE_VTX_ST_TEX_2_R                     0x23c8
-#define R200_SE_VTX_ST_TEX_2_Q                     0x23cc
-#define R200_SE_VTX_ST_TEX_3_S                     0x23d0
-#define R200_SE_VTX_ST_TEX_3_T                     0x23d4
-#define R200_SE_VTX_ST_TEX_3_R                     0x23d8
-#define R200_SE_VTX_ST_TEX_3_Q                     0x23dc
-#define R200_SE_VTX_ST_TEX_4_S                     0x23e0
-#define R200_SE_VTX_ST_TEX_4_T                     0x23e4
-#define R200_SE_VTX_ST_TEX_4_R                     0x23e8
-#define R200_SE_VTX_ST_TEX_4_Q                     0x23ec
-#define R200_SE_VTX_ST_TEX_5_S                     0x23f0
-#define R200_SE_VTX_ST_TEX_5_T                     0x23f4
-#define R200_SE_VTX_ST_TEX_5_R                     0x23f8
-#define R200_SE_VTX_ST_TEX_5_Q                     0x23fc
-#define R200_SE_VTX_ST_PNT_SPRT_SZ                 0x2400
-#define R200_SE_VTX_ST_DISC_FOG                    0x2404
-#define R200_SE_VTX_ST_SHININESS_0                 0x2408
-#define R200_SE_VTX_ST_SHININESS_1                 0x240c
-#define R200_SE_VTX_ST_BLND_WT_0                   0x2410
-#define R200_SE_VTX_ST_BLND_WT_1                   0x2414
-#define R200_SE_VTX_ST_BLND_WT_2                   0x2418
-#define R200_SE_VTX_ST_BLND_WT_3                   0x241c
-#define R200_SE_VTX_ST_POS_1_X                     0x2420
-#define R200_SE_VTX_ST_POS_1_Y                     0x2424
-#define R200_SE_VTX_ST_POS_1_Z                     0x2428
-#define R200_SE_VTX_ST_POS_1_W                     0x242c
-#define R200_SE_VTX_ST_NORM_1_X                    0x2430
-#define R200_SE_VTX_ST_NORM_1_Y                    0x2434
-#define R200_SE_VTX_ST_NORM_1_Z                    0x2438
-#define R200_SE_VTX_ST_USR_CLR_0_R                 0x2440
-#define R200_SE_VTX_ST_USR_CLR_0_G                 0x2444
-#define R200_SE_VTX_ST_USR_CLR_0_B                 0x2448
-#define R200_SE_VTX_ST_USR_CLR_0_A                 0x244c
-#define R200_SE_VTX_ST_USR_CLR_1_R                 0x2450
-#define R200_SE_VTX_ST_USR_CLR_1_G                 0x2454
-#define R200_SE_VTX_ST_USR_CLR_1_B                 0x2458
-#define R200_SE_VTX_ST_USR_CLR_1_A                 0x245c
-#define R200_SE_VTX_ST_CLR_0_PKD                   0x2460
-#define R200_SE_VTX_ST_CLR_1_PKD                   0x2464
-#define R200_SE_VTX_ST_CLR_2_PKD                   0x2468
-#define R200_SE_VTX_ST_CLR_3_PKD                   0x246c
-#define R200_SE_VTX_ST_CLR_4_PKD                   0x2470
-#define R200_SE_VTX_ST_CLR_5_PKD                   0x2474
-#define R200_SE_VTX_ST_CLR_6_PKD                   0x2478
-#define R200_SE_VTX_ST_CLR_7_PKD                   0x247c
-#define R200_SE_VTX_ST_POS_0_X_2                   0x2480
-#define R200_SE_VTX_ST_POS_0_Y_2                   0x2484
-#define R200_SE_VTX_ST_PAR_CLR_LD                  0x2488
-#define R200_SE_VTX_ST_USR_CLR_PKD                 0x248c
-#define R200_SE_VTX_ST_POS_0_X_3                   0x2490
-#define R200_SE_VTX_ST_POS_0_Y_3                   0x2494
-#define R200_SE_VTX_ST_POS_0_Z_3                   0x2498
-#define R200_SE_VTX_ST_END_OF_PKT                  0x249c
-/* gap */
-#define R200_RE_POINTSIZE                          0x2648
-#define     R200_POINTSIZE_SHIFT                       0
-#define     R200_MAXPOINTSIZE_SHIFT                    16
-/* gap */
-#define R200_RE_TOP_LEFT                  0x26c0
-#define     R200_RE_LEFT_SHIFT         0
-#define     R200_RE_TOP_SHIFT          16
-#define R200_RE_MISC                      0x26c4
-#define     R200_STIPPLE_COORD_MASK           0x1f
-#define     R200_STIPPLE_X_OFFSET_SHIFT       0
-#define     R200_STIPPLE_X_OFFSET_MASK        (0x1f << 0)
-#define     R200_STIPPLE_Y_OFFSET_SHIFT       8
-#define     R200_STIPPLE_Y_OFFSET_MASK        (0x1f << 8)
-#define     R200_STIPPLE_LITTLE_BIT_ORDER     (0 << 16)
-#define     R200_STIPPLE_BIG_BIT_ORDER        (1 << 16)
-/* gap */
-#define R200_RE_AUX_SCISSOR_CNTL                   0x26f0
-#define     R200_EXCLUSIVE_SCISSOR_0      0x01000000
-#define     R200_EXCLUSIVE_SCISSOR_1      0x02000000
-#define     R200_EXCLUSIVE_SCISSOR_2      0x04000000
-#define     R200_SCISSOR_ENABLE_0         0x10000000
-#define     R200_SCISSOR_ENABLE_1         0x20000000
-#define     R200_SCISSOR_ENABLE_2         0x40000000
-/* gap */
-#define R200_PP_TXFILTER_0                0x2c00
-#define     R200_MAG_FILTER_NEAREST                   (0  <<  0)
-#define     R200_MAG_FILTER_LINEAR                    (1  <<  0)
-#define     R200_MAG_FILTER_MASK                      (1  <<  0)
-#define     R200_MIN_FILTER_NEAREST                   (0  <<  1)
-#define     R200_MIN_FILTER_LINEAR                    (1  <<  1)
-#define     R200_MIN_FILTER_NEAREST_MIP_NEAREST       (2  <<  1)
-#define     R200_MIN_FILTER_NEAREST_MIP_LINEAR        (3  <<  1)
-#define     R200_MIN_FILTER_LINEAR_MIP_NEAREST        (6  <<  1)
-#define     R200_MIN_FILTER_LINEAR_MIP_LINEAR         (7  <<  1)
-#define     R200_MIN_FILTER_ANISO_NEAREST             (8  <<  1)
-#define     R200_MIN_FILTER_ANISO_LINEAR              (9  <<  1)
-#define     R200_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (10 <<  1)
-#define     R200_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR  (11 <<  1)
-#define     R200_MIN_FILTER_MASK                      (15 <<  1)
-#define     R200_MAX_ANISO_1_TO_1                     (0  <<  5)
-#define     R200_MAX_ANISO_2_TO_1                     (1  <<  5)
-#define     R200_MAX_ANISO_4_TO_1                     (2  <<  5)
-#define     R200_MAX_ANISO_8_TO_1                     (3  <<  5)
-#define     R200_MAX_ANISO_16_TO_1                    (4  <<  5)
-#define     R200_MAX_ANISO_MASK                       (7  <<  5)
-#define     R200_MAX_MIP_LEVEL_MASK                   (0x0f << 16)
-#define     R200_MAX_MIP_LEVEL_SHIFT                  16
-#define     R200_YUV_TO_RGB                           (1  << 20)
-#define     R200_YUV_TEMPERATURE_COOL                 (0  << 21)
-#define     R200_YUV_TEMPERATURE_HOT                  (1  << 21)
-#define     R200_YUV_TEMPERATURE_MASK                 (1  << 21)
-#define     R200_WRAPEN_S                             (1  << 22)
-#define     R200_CLAMP_S_WRAP                         (0  << 23)
-#define     R200_CLAMP_S_MIRROR                       (1  << 23)
-#define     R200_CLAMP_S_CLAMP_LAST                   (2  << 23)
-#define     R200_CLAMP_S_MIRROR_CLAMP_LAST            (3  << 23)
-#define     R200_CLAMP_S_CLAMP_BORDER                 (4  << 23)
-#define     R200_CLAMP_S_MIRROR_CLAMP_BORDER          (5  << 23)
-#define     R200_CLAMP_S_CLAMP_GL                     (6  << 23)
-#define     R200_CLAMP_S_MIRROR_CLAMP_GL              (7  << 23)
-#define     R200_CLAMP_S_MASK                         (7  << 23)
-#define     R200_WRAPEN_T                             (1  << 26)
-#define     R200_CLAMP_T_WRAP                         (0  << 27)
-#define     R200_CLAMP_T_MIRROR                       (1  << 27)
-#define     R200_CLAMP_T_CLAMP_LAST                   (2  << 27)
-#define     R200_CLAMP_T_MIRROR_CLAMP_LAST            (3  << 27)
-#define     R200_CLAMP_T_CLAMP_BORDER                 (4  << 27)
-#define     R200_CLAMP_T_MIRROR_CLAMP_BORDER          (5  << 27)
-#define     R200_CLAMP_T_CLAMP_GL                     (6  << 27)
-#define     R200_CLAMP_T_MIRROR_CLAMP_GL              (7  << 27)
-#define     R200_CLAMP_T_MASK                         (7  << 27)
-#define     R200_KILL_LT_ZERO                         (1  << 30)
-#define     R200_BORDER_MODE_OGL                      (0  << 31)
-#define     R200_BORDER_MODE_D3D                      (1  << 31)
-#define R200_PP_TXFORMAT_0                0x2c04
-#define     R200_TXFORMAT_I8                 (0  <<  0)
-#define     R200_TXFORMAT_AI88               (1  <<  0)
-#define     R200_TXFORMAT_RGB332             (2  <<  0)
-#define     R200_TXFORMAT_ARGB1555           (3  <<  0)
-#define     R200_TXFORMAT_RGB565             (4  <<  0)
-#define     R200_TXFORMAT_ARGB4444           (5  <<  0)
-#define     R200_TXFORMAT_ARGB8888           (6  <<  0)
-#define     R200_TXFORMAT_RGBA8888           (7  <<  0)
-#define     R200_TXFORMAT_Y8                 (8  <<  0)
-#define     R200_TXFORMAT_AVYU4444           (9  <<  0)
-#define     R200_TXFORMAT_VYUY422            (10  <<  0)
-#define     R200_TXFORMAT_YVYU422            (11  <<  0)
-#define     R200_TXFORMAT_DXT1               (12  <<  0)
-#define     R200_TXFORMAT_DXT23              (14  <<  0)
-#define     R200_TXFORMAT_DXT45              (15  <<  0)
-#define     R200_TXFORMAT_FORMAT_MASK        (31 <<  0)
-#define     R200_TXFORMAT_FORMAT_SHIFT       0
-#define     R200_TXFORMAT_ALPHA_IN_MAP       (1  <<  6)
-#define     R200_TXFORMAT_NON_POWER2         (1  <<  7)
-#define     R200_TXFORMAT_WIDTH_MASK         (15 <<  8)
-#define     R200_TXFORMAT_WIDTH_SHIFT        8
-#define     R200_TXFORMAT_HEIGHT_MASK        (15 << 12)
-#define     R200_TXFORMAT_HEIGHT_SHIFT       12
-#define     R200_TXFORMAT_F5_WIDTH_MASK      (15 << 16)	/* cube face 5 */
-#define     R200_TXFORMAT_F5_WIDTH_SHIFT     16
-#define     R200_TXFORMAT_F5_HEIGHT_MASK     (15 << 20)
-#define     R200_TXFORMAT_F5_HEIGHT_SHIFT    20
-#define     R200_TXFORMAT_ST_ROUTE_STQ0      (0  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_STQ1      (1  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_STQ2      (2  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_STQ3      (3  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_STQ4      (4  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_STQ5      (5  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_MASK      (7  << 24)
-#define     R200_TXFORMAT_ST_ROUTE_SHIFT     24
-#define     R200_TXFORMAT_ALPHA_MASK_ENABLE  (1  << 28)
-#define     R200_TXFORMAT_CHROMA_KEY_ENABLE  (1  << 29)
-#define     R200_TXFORMAT_CUBIC_MAP_ENABLE   (1  << 30)
-#define R200_PP_TXFORMAT_X_0              0x2c08
-#define     R200_DEPTH_LOG2_MASK                      (0xf << 0)
-#define     R200_DEPTH_LOG2_SHIFT                     0
-#define     R200_VOLUME_FILTER_SHIFT                  4
-#define     R200_VOLUME_FILTER_MASK                   (1 << 4)
-#define     R200_VOLUME_FILTER_NEAREST                (0 << 4)
-#define     R200_VOLUME_FILTER_LINEAR                 (1 << 4)
-#define     R200_WRAPEN_Q                             (1  << 8)
-#define     R200_CLAMP_Q_WRAP                         (0  << 9)
-#define     R200_CLAMP_Q_MIRROR                       (1  << 9)
-#define     R200_CLAMP_Q_CLAMP_LAST                   (2  << 9)
-#define     R200_CLAMP_Q_MIRROR_CLAMP_LAST            (3  << 9)
-#define     R200_CLAMP_Q_CLAMP_BORDER                 (4  << 9)
-#define     R200_CLAMP_Q_MIRROR_CLAMP_BORDER          (5  << 9)
-#define     R200_CLAMP_Q_CLAMP_GL                     (6  << 9)
-#define     R200_CLAMP_Q_MIRROR_CLAMP_GL              (7  << 9)
-#define     R200_CLAMP_Q_MASK                         (7  << 9)
-#define     R200_MIN_MIP_LEVEL_MASK                   (0xff << 12)
-#define     R200_MIN_MIP_LEVEL_SHIFT                  12
-#define     R200_TEXCOORD_NONPROJ                     (0  << 16)
-#define     R200_TEXCOORD_CUBIC_ENV                   (1  << 16)
-#define     R200_TEXCOORD_VOLUME                      (2  << 16)
-#define     R200_TEXCOORD_PROJ                        (3  << 16)
-#define     R200_TEXCOORD_DEPTH                       (4  << 16)
-#define     R200_TEXCOORD_1D_PROJ                     (5  << 16)
-#define     R200_TEXCOORD_1D                          (6  << 16)
-#define     R200_TEXCOORD_ZERO                        (7  << 16)
-#define     R200_TEXCOORD_MASK                        (7  << 16)
-#define     R200_LOD_BIAS_MASK                        (0xfff80000)
-#define     R200_LOD_BIAS_SHIFT                       19
-#define R200_PP_TXSIZE_0                  0x2c0c	/* NPOT only */
-#define R200_PP_TXPITCH_0                 0x2c10	/* NPOT only */
-#define R200_PP_BORDER_COLOR_0            0x2c14
-#define R200_PP_CUBIC_FACES_0             0x2c18
-#define     R200_FACE_WIDTH_1_SHIFT                   0
-#define     R200_FACE_HEIGHT_1_SHIFT                  4
-#define     R200_FACE_WIDTH_1_MASK                   (0xf << 0)
-#define     R200_FACE_HEIGHT_1_MASK                  (0xf << 4)
-#define     R200_FACE_WIDTH_2_SHIFT                   8
-#define     R200_FACE_HEIGHT_2_SHIFT                 12
-#define     R200_FACE_WIDTH_2_MASK                   (0xf << 8)
-#define     R200_FACE_HEIGHT_2_MASK                  (0xf << 12)
-#define     R200_FACE_WIDTH_3_SHIFT                  16
-#define     R200_FACE_HEIGHT_3_SHIFT                 20
-#define     R200_FACE_WIDTH_3_MASK                   (0xf << 16)
-#define     R200_FACE_HEIGHT_3_MASK                  (0xf << 20)
-#define     R200_FACE_WIDTH_4_SHIFT                  24
-#define     R200_FACE_HEIGHT_4_SHIFT                 28
-#define     R200_FACE_WIDTH_4_MASK                   (0xf << 24)
-#define     R200_FACE_HEIGHT_4_MASK                  (0xf << 28)
-#define R200_PP_TXFILTER_1                0x2c20
-#define R200_PP_TXFORMAT_1                0x2c24
-#define R200_PP_TXFORMAT_X_1              0x2c28
-#define R200_PP_TXSIZE_1                  0x2c2c
-#define R200_PP_TXPITCH_1                 0x2c30
-#define R200_PP_BORDER_COLOR_1            0x2c34
-#define R200_PP_CUBIC_FACES_1             0x2c38
-#define R200_PP_TXFILTER_2                0x2c40
-#define R200_PP_TXFORMAT_2                0x2c44
-#define R200_PP_TXSIZE_2                  0x2c4c
-#define R200_PP_TXFORMAT_X_2              0x2c48
-#define R200_PP_TXPITCH_2                 0x2c50
-#define R200_PP_BORDER_COLOR_2            0x2c54
-#define R200_PP_CUBIC_FACES_2             0x2c58
-#define R200_PP_TXFILTER_3                0x2c60
-#define R200_PP_TXFORMAT_3                0x2c64
-#define R200_PP_TXSIZE_3                  0x2c6c
-#define R200_PP_TXFORMAT_X_3              0x2c68
-#define R200_PP_TXPITCH_3                 0x2c70
-#define R200_PP_BORDER_COLOR_3            0x2c74
-#define R200_PP_CUBIC_FACES_3             0x2c78
-#define R200_PP_TXFILTER_4                0x2c80
-#define R200_PP_TXFORMAT_4                0x2c84
-#define R200_PP_TXSIZE_4                  0x2c8c
-#define R200_PP_TXFORMAT_X_4              0x2c88
-#define R200_PP_TXPITCH_4                 0x2c90
-#define R200_PP_BORDER_COLOR_4            0x2c94
-#define R200_PP_CUBIC_FACES_4             0x2c98
-#define R200_PP_TXFILTER_5                0x2ca0
-#define R200_PP_TXFORMAT_5                0x2ca4
-#define R200_PP_TXSIZE_5                  0x2cac
-#define R200_PP_TXFORMAT_X_5              0x2ca8
-#define R200_PP_TXPITCH_5                 0x2cb0
-#define R200_PP_BORDER_COLOR_5            0x2cb4
-#define R200_PP_CUBIC_FACES_5             0x2cb8
-/* gap */
-#define R200_PP_CNTL_X             0x2cc4
-/* gap */
-#define R200_PP_TXOFFSET_0                0x2d00
-#define     R200_TXO_ENDIAN_NO_SWAP     (0 << 0)
-#define     R200_TXO_ENDIAN_BYTE_SWAP   (1 << 0)
-#define     R200_TXO_ENDIAN_WORD_SWAP   (2 << 0)
-#define     R200_TXO_ENDIAN_HALFDW_SWAP (3 << 0)
-#define     R200_TXO_OFFSET_MASK        0xffffffe0
-#define     R200_TXO_OFFSET_SHIFT       5
-#define R200_PP_CUBIC_OFFSET_F1_0         0x2d04
-#define R200_PP_CUBIC_OFFSET_F2_0         0x2d08
-#define R200_PP_CUBIC_OFFSET_F3_0         0x2d0c
-#define R200_PP_CUBIC_OFFSET_F4_0         0x2d10
-#define R200_PP_CUBIC_OFFSET_F5_0         0x2d14
-#define R200_PP_TXOFFSET_1                0x2d18
-#define R200_PP_CUBIC_OFFSET_F1_1         0x2d1c
-#define R200_PP_CUBIC_OFFSET_F2_1         0x2d20
-#define R200_PP_CUBIC_OFFSET_F3_1         0x2d24
-#define R200_PP_CUBIC_OFFSET_F4_1         0x2d28
-#define R200_PP_CUBIC_OFFSET_F5_1         0x2d2c
-#define R200_PP_TXOFFSET_2                0x2d30
-#define R200_PP_CUBIC_OFFSET_F1_2         0x2d34
-#define R200_PP_CUBIC_OFFSET_F2_2         0x2d38
-#define R200_PP_CUBIC_OFFSET_F3_2         0x2d3c
-#define R200_PP_CUBIC_OFFSET_F4_2         0x2d40
-#define R200_PP_CUBIC_OFFSET_F5_2         0x2d44
-#define R200_PP_TXOFFSET_3                0x2d48
-#define R200_PP_CUBIC_OFFSET_F1_3         0x2d4c
-#define R200_PP_CUBIC_OFFSET_F2_3         0x2d50
-#define R200_PP_CUBIC_OFFSET_F3_3         0x2d54
-#define R200_PP_CUBIC_OFFSET_F4_3         0x2d58
-#define R200_PP_CUBIC_OFFSET_F5_3         0x2d5c
-#define R200_PP_TXOFFSET_4                0x2d60
-#define R200_PP_CUBIC_OFFSET_F1_4         0x2d64
-#define R200_PP_CUBIC_OFFSET_F2_4         0x2d68
-#define R200_PP_CUBIC_OFFSET_F3_4         0x2d6c
-#define R200_PP_CUBIC_OFFSET_F4_4         0x2d70
-#define R200_PP_CUBIC_OFFSET_F5_4         0x2d74
-#define R200_PP_TXOFFSET_5                0x2d78
-#define R200_PP_CUBIC_OFFSET_F1_5         0x2d7c
-#define R200_PP_CUBIC_OFFSET_F2_5         0x2d80
-#define R200_PP_CUBIC_OFFSET_F3_5         0x2d84
-#define R200_PP_CUBIC_OFFSET_F4_5         0x2d88
-#define R200_PP_CUBIC_OFFSET_F5_5         0x2d8c
-/* gap */
-#define R200_PP_TAM_DEBUG3                0x2d9c
-/* gap */
-#define R200_PP_TFACTOR_0                 0x2ee0
-#define R200_PP_TFACTOR_1                 0x2ee4
-#define R200_PP_TFACTOR_2                 0x2ee8
-#define R200_PP_TFACTOR_3                 0x2eec
-#define R200_PP_TFACTOR_4                 0x2ef0
-#define R200_PP_TFACTOR_5                 0x2ef4
-/* gap */
-#define R200_PP_TXCBLEND_0                0x2f00
-#define     R200_TXC_ARG_A_ZERO                (0)
-#define     R200_TXC_ARG_A_CURRENT_COLOR       (2)
-#define     R200_TXC_ARG_A_CURRENT_ALPHA       (3)
-#define     R200_TXC_ARG_A_DIFFUSE_COLOR       (4)
-#define     R200_TXC_ARG_A_DIFFUSE_ALPHA       (5)
-#define     R200_TXC_ARG_A_SPECULAR_COLOR      (6)
-#define     R200_TXC_ARG_A_SPECULAR_ALPHA      (7)
-#define     R200_TXC_ARG_A_TFACTOR_COLOR       (8)
-#define     R200_TXC_ARG_A_TFACTOR_ALPHA       (9)
-#define     R200_TXC_ARG_A_R0_COLOR            (10)
-#define     R200_TXC_ARG_A_R0_ALPHA            (11)
-#define     R200_TXC_ARG_A_R1_COLOR            (12)
-#define     R200_TXC_ARG_A_R1_ALPHA            (13)
-#define     R200_TXC_ARG_A_R2_COLOR            (14)
-#define     R200_TXC_ARG_A_R2_ALPHA            (15)
-#define     R200_TXC_ARG_A_R3_COLOR            (16)
-#define     R200_TXC_ARG_A_R3_ALPHA            (17)
-#define     R200_TXC_ARG_A_R4_COLOR            (18)
-#define     R200_TXC_ARG_A_R4_ALPHA            (19)
-#define     R200_TXC_ARG_A_R5_COLOR            (20)
-#define     R200_TXC_ARG_A_R5_ALPHA            (21)
-#define     R200_TXC_ARG_A_TFACTOR1_COLOR      (26)
-#define     R200_TXC_ARG_A_TFACTOR1_ALPHA      (27)
-#define     R200_TXC_ARG_A_MASK			(31 << 0)
-#define     R200_TXC_ARG_A_SHIFT			0
-#define     R200_TXC_ARG_B_ZERO                (0<<5)
-#define     R200_TXC_ARG_B_CURRENT_COLOR       (2<<5)
-#define     R200_TXC_ARG_B_CURRENT_ALPHA       (3<<5)
-#define     R200_TXC_ARG_B_DIFFUSE_COLOR       (4<<5)
-#define     R200_TXC_ARG_B_DIFFUSE_ALPHA       (5<<5)
-#define     R200_TXC_ARG_B_SPECULAR_COLOR      (6<<5)
-#define     R200_TXC_ARG_B_SPECULAR_ALPHA      (7<<5)
-#define     R200_TXC_ARG_B_TFACTOR_COLOR       (8<<5)
-#define     R200_TXC_ARG_B_TFACTOR_ALPHA       (9<<5)
-#define     R200_TXC_ARG_B_R0_COLOR            (10<<5)
-#define     R200_TXC_ARG_B_R0_ALPHA            (11<<5)
-#define     R200_TXC_ARG_B_R1_COLOR            (12<<5)
-#define     R200_TXC_ARG_B_R1_ALPHA            (13<<5)
-#define     R200_TXC_ARG_B_R2_COLOR            (14<<5)
-#define     R200_TXC_ARG_B_R2_ALPHA            (15<<5)
-#define     R200_TXC_ARG_B_R3_COLOR            (16<<5)
-#define     R200_TXC_ARG_B_R3_ALPHA            (17<<5)
-#define     R200_TXC_ARG_B_R4_COLOR            (18<<5)
-#define     R200_TXC_ARG_B_R4_ALPHA            (19<<5)
-#define     R200_TXC_ARG_B_R5_COLOR            (20<<5)
-#define     R200_TXC_ARG_B_R5_ALPHA            (21<<5)
-#define     R200_TXC_ARG_B_TFACTOR1_COLOR      (26<<5)
-#define     R200_TXC_ARG_B_TFACTOR1_ALPHA      (27<<5)
-#define     R200_TXC_ARG_B_MASK			(31 << 5)
-#define     R200_TXC_ARG_B_SHIFT			5
-#define     R200_TXC_ARG_C_ZERO                (0<<10)
-#define     R200_TXC_ARG_C_CURRENT_COLOR       (2<<10)
-#define     R200_TXC_ARG_C_CURRENT_ALPHA       (3<<10)
-#define     R200_TXC_ARG_C_DIFFUSE_COLOR       (4<<10)
-#define     R200_TXC_ARG_C_DIFFUSE_ALPHA       (5<<10)
-#define     R200_TXC_ARG_C_SPECULAR_COLOR      (6<<10)
-#define     R200_TXC_ARG_C_SPECULAR_ALPHA      (7<<10)
-#define     R200_TXC_ARG_C_TFACTOR_COLOR       (8<<10)
-#define     R200_TXC_ARG_C_TFACTOR_ALPHA       (9<<10)
-#define     R200_TXC_ARG_C_R0_COLOR            (10<<10)
-#define     R200_TXC_ARG_C_R0_ALPHA            (11<<10)
-#define     R200_TXC_ARG_C_R1_COLOR            (12<<10)
-#define     R200_TXC_ARG_C_R1_ALPHA            (13<<10)
-#define     R200_TXC_ARG_C_R2_COLOR            (14<<10)
-#define     R200_TXC_ARG_C_R2_ALPHA            (15<<10)
-#define     R200_TXC_ARG_C_R3_COLOR            (16<<10)
-#define     R200_TXC_ARG_C_R3_ALPHA            (17<<10)
-#define     R200_TXC_ARG_C_R4_COLOR            (18<<10)
-#define     R200_TXC_ARG_C_R4_ALPHA            (19<<10)
-#define     R200_TXC_ARG_C_R5_COLOR            (20<<10)
-#define     R200_TXC_ARG_C_R5_ALPHA            (21<<10)
-#define     R200_TXC_ARG_C_TFACTOR1_COLOR      (26<<10)
-#define     R200_TXC_ARG_C_TFACTOR1_ALPHA      (27<<10)
-#define     R200_TXC_ARG_C_MASK			(31 << 10)
-#define     R200_TXC_ARG_C_SHIFT			10
-#define     R200_TXC_COMP_ARG_A                    (1 << 16)
-#define     R200_TXC_COMP_ARG_A_SHIFT              (16)
-#define     R200_TXC_BIAS_ARG_A                    (1 << 17)
-#define     R200_TXC_SCALE_ARG_A                   (1 << 18)
-#define     R200_TXC_NEG_ARG_A                     (1 << 19)
-#define     R200_TXC_COMP_ARG_B                    (1 << 20)
-#define     R200_TXC_COMP_ARG_B_SHIFT              (20)
-#define     R200_TXC_BIAS_ARG_B                    (1 << 21)
-#define     R200_TXC_SCALE_ARG_B                   (1 << 22)
-#define     R200_TXC_NEG_ARG_B                     (1 << 23)
-#define     R200_TXC_COMP_ARG_C                    (1 << 24)
-#define     R200_TXC_COMP_ARG_C_SHIFT              (24)
-#define     R200_TXC_BIAS_ARG_C                    (1 << 25)
-#define     R200_TXC_SCALE_ARG_C                   (1 << 26)
-#define     R200_TXC_NEG_ARG_C                     (1 << 27)
-#define     R200_TXC_OP_MADD                        (0 << 28)
-#define     R200_TXC_OP_CND0                       (2 << 28)
-#define     R200_TXC_OP_LERP                       (3 << 28)
-#define     R200_TXC_OP_DOT3                       (4 << 28)
-#define     R200_TXC_OP_DOT4                       (5 << 28)
-#define     R200_TXC_OP_CONDITIONAL                (6 << 28)
-#define     R200_TXC_OP_DOT2_ADD                   (7 << 28)
-#define     R200_TXC_OP_MASK                       (7 << 28)
-#define R200_PP_TXCBLEND2_0                0x2f04
-#define     R200_TXC_TFACTOR_SEL_SHIFT             0
-#define     R200_TXC_TFACTOR_SEL_MASK              0x7
-#define     R200_TXC_TFACTOR1_SEL_SHIFT            4
-#define     R200_TXC_TFACTOR1_SEL_MASK             (0x7 << 4)
-#define     R200_TXC_SCALE_SHIFT                   8
-#define     R200_TXC_SCALE_MASK                    (7 << 8)
-#define     R200_TXC_SCALE_1X                      (0 << 8)
-#define     R200_TXC_SCALE_2X                      (1 << 8)
-#define     R200_TXC_SCALE_4X                      (2 << 8)
-#define     R200_TXC_SCALE_8X                      (3 << 8)
-#define     R200_TXC_SCALE_INV2                    (5 << 8)
-#define     R200_TXC_SCALE_INV4                    (6 << 8)
-#define     R200_TXC_SCALE_INV8                    (7 << 8)
-#define     R200_TXC_CLAMP_SHIFT                   12
-#define     R200_TXC_CLAMP_MASK                    (3 << 12)
-#define     R200_TXC_CLAMP_WRAP                    (0 << 12)
-#define     R200_TXC_CLAMP_0_1                     (1 << 12)
-#define     R200_TXC_CLAMP_8_8                     (2 << 12)
-#define     R200_TXC_OUTPUT_REG_MASK               (7 << 16)
-#define     R200_TXC_OUTPUT_REG_NONE               (0 << 16)
-#define     R200_TXC_OUTPUT_REG_R0                 (1 << 16)
-#define     R200_TXC_OUTPUT_REG_R1                 (2 << 16)
-#define     R200_TXC_OUTPUT_REG_R2                 (3 << 16)
-#define     R200_TXC_OUTPUT_REG_R3                 (4 << 16)
-#define     R200_TXC_OUTPUT_REG_R4                 (5 << 16)
-#define     R200_TXC_OUTPUT_REG_R5                 (6 << 16)
-#define     R200_TXC_OUTPUT_MASK_MASK              (7 << 20)
-#define     R200_TXC_OUTPUT_MASK_RGB               (0 << 20)
-#define     R200_TXC_OUTPUT_MASK_RG                (1 << 20)
-#define     R200_TXC_OUTPUT_MASK_RB                (2 << 20)
-#define     R200_TXC_OUTPUT_MASK_R                 (3 << 20)
-#define     R200_TXC_OUTPUT_MASK_GB                (4 << 20)
-#define     R200_TXC_OUTPUT_MASK_G                 (5 << 20)
-#define     R200_TXC_OUTPUT_MASK_B                 (6 << 20)
-#define     R200_TXC_OUTPUT_MASK_NONE              (7 << 20)
-#define     R200_TXC_REPL_NORMAL                   0
-#define     R200_TXC_REPL_RED                      1
-#define     R200_TXC_REPL_GREEN                    2
-#define     R200_TXC_REPL_BLUE                     3
-#define     R200_TXC_REPL_ARG_A_SHIFT              26
-#define     R200_TXC_REPL_ARG_A_MASK               (3 << 26)
-#define     R200_TXC_REPL_ARG_B_SHIFT              28
-#define     R200_TXC_REPL_ARG_B_MASK               (3 << 28)
-#define     R200_TXC_REPL_ARG_C_SHIFT              30
-#define     R200_TXC_REPL_ARG_C_MASK               (3 << 30)
-#define R200_PP_TXABLEND_0                0x2f08
-#define     R200_TXA_ARG_A_ZERO              (0)
-#define     R200_TXA_ARG_A_CURRENT_ALPHA     (2)	/* guess */
-#define     R200_TXA_ARG_A_CURRENT_BLUE      (3)	/* guess */
-#define     R200_TXA_ARG_A_DIFFUSE_ALPHA     (4)
-#define     R200_TXA_ARG_A_DIFFUSE_BLUE      (5)
-#define     R200_TXA_ARG_A_SPECULAR_ALPHA    (6)
-#define     R200_TXA_ARG_A_SPECULAR_BLUE     (7)
-#define     R200_TXA_ARG_A_TFACTOR_ALPHA     (8)
-#define     R200_TXA_ARG_A_TFACTOR_BLUE      (9)
-#define     R200_TXA_ARG_A_R0_ALPHA          (10)
-#define     R200_TXA_ARG_A_R0_BLUE           (11)
-#define     R200_TXA_ARG_A_R1_ALPHA          (12)
-#define     R200_TXA_ARG_A_R1_BLUE           (13)
-#define     R200_TXA_ARG_A_R2_ALPHA          (14)
-#define     R200_TXA_ARG_A_R2_BLUE           (15)
-#define     R200_TXA_ARG_A_R3_ALPHA          (16)
-#define     R200_TXA_ARG_A_R3_BLUE           (17)
-#define     R200_TXA_ARG_A_R4_ALPHA          (18)
-#define     R200_TXA_ARG_A_R4_BLUE           (19)
-#define     R200_TXA_ARG_A_R5_ALPHA          (20)
-#define     R200_TXA_ARG_A_R5_BLUE           (21)
-#define     R200_TXA_ARG_A_TFACTOR1_ALPHA    (26)
-#define     R200_TXA_ARG_A_TFACTOR1_BLUE     (27)
-#define     R200_TXA_ARG_A_MASK			(31 << 0)
-#define     R200_TXA_ARG_A_SHIFT			0
-#define     R200_TXA_ARG_B_ZERO              (0<<5)
-#define     R200_TXA_ARG_B_CURRENT_ALPHA     (2<<5)	/* guess */
-#define     R200_TXA_ARG_B_CURRENT_BLUE      (3<<5)	/* guess */
-#define     R200_TXA_ARG_B_DIFFUSE_ALPHA     (4<<5)
-#define     R200_TXA_ARG_B_DIFFUSE_BLUE      (5<<5)
-#define     R200_TXA_ARG_B_SPECULAR_ALPHA    (6<<5)
-#define     R200_TXA_ARG_B_SPECULAR_BLUE     (7<<5)
-#define     R200_TXA_ARG_B_TFACTOR_ALPHA     (8<<5)
-#define     R200_TXA_ARG_B_TFACTOR_BLUE      (9<<5)
-#define     R200_TXA_ARG_B_R0_ALPHA          (10<<5)
-#define     R200_TXA_ARG_B_R0_BLUE           (11<<5)
-#define     R200_TXA_ARG_B_R1_ALPHA          (12<<5)
-#define     R200_TXA_ARG_B_R1_BLUE           (13<<5)
-#define     R200_TXA_ARG_B_R2_ALPHA          (14<<5)
-#define     R200_TXA_ARG_B_R2_BLUE           (15<<5)
-#define     R200_TXA_ARG_B_R3_ALPHA          (16<<5)
-#define     R200_TXA_ARG_B_R3_BLUE           (17<<5)
-#define     R200_TXA_ARG_B_R4_ALPHA          (18<<5)
-#define     R200_TXA_ARG_B_R4_BLUE           (19<<5)
-#define     R200_TXA_ARG_B_R5_ALPHA          (20<<5)
-#define     R200_TXA_ARG_B_R5_BLUE           (21<<5)
-#define     R200_TXA_ARG_B_TFACTOR1_ALPHA    (26<<5)
-#define     R200_TXA_ARG_B_TFACTOR1_BLUE     (27<<5)
-#define     R200_TXA_ARG_B_MASK			(31 << 5)
-#define     R200_TXA_ARG_B_SHIFT			5
-#define     R200_TXA_ARG_C_ZERO              (0<<10)
-#define     R200_TXA_ARG_C_CURRENT_ALPHA     (2<<10)	/* guess */
-#define     R200_TXA_ARG_C_CURRENT_BLUE      (3<<10)	/* guess */
-#define     R200_TXA_ARG_C_DIFFUSE_ALPHA     (4<<10)
-#define     R200_TXA_ARG_C_DIFFUSE_BLUE      (5<<10)
-#define     R200_TXA_ARG_C_SPECULAR_ALPHA    (6<<10)
-#define     R200_TXA_ARG_C_SPECULAR_BLUE     (7<<10)
-#define     R200_TXA_ARG_C_TFACTOR_ALPHA     (8<<10)
-#define     R200_TXA_ARG_C_TFACTOR_BLUE      (9<<10)
-#define     R200_TXA_ARG_C_R0_ALPHA          (10<<10)
-#define     R200_TXA_ARG_C_R0_BLUE           (11<<10)
-#define     R200_TXA_ARG_C_R1_ALPHA          (12<<10)
-#define     R200_TXA_ARG_C_R1_BLUE           (13<<10)
-#define     R200_TXA_ARG_C_R2_ALPHA          (14<<10)
-#define     R200_TXA_ARG_C_R2_BLUE           (15<<10)
-#define     R200_TXA_ARG_C_R3_ALPHA          (16<<10)
-#define     R200_TXA_ARG_C_R3_BLUE           (17<<10)
-#define     R200_TXA_ARG_C_R4_ALPHA          (18<<10)
-#define     R200_TXA_ARG_C_R4_BLUE           (19<<10)
-#define     R200_TXA_ARG_C_R5_ALPHA          (20<<10)
-#define     R200_TXA_ARG_C_R5_BLUE           (21<<10)
-#define     R200_TXA_ARG_C_TFACTOR1_ALPHA    (26<<10)
-#define     R200_TXA_ARG_C_TFACTOR1_BLUE     (27<<10)
-#define     R200_TXA_ARG_C_MASK			(31 << 10)
-#define     R200_TXA_ARG_C_SHIFT			10
-#define     R200_TXA_COMP_ARG_A                    (1 << 16)
-#define     R200_TXA_COMP_ARG_A_SHIFT              (16)
-#define     R200_TXA_BIAS_ARG_A                    (1 << 17)
-#define     R200_TXA_SCALE_ARG_A                   (1 << 18)
-#define     R200_TXA_NEG_ARG_A                     (1 << 19)
-#define     R200_TXA_COMP_ARG_B                    (1 << 20)
-#define     R200_TXA_COMP_ARG_B_SHIFT              (20)
-#define     R200_TXA_BIAS_ARG_B                    (1 << 21)
-#define     R200_TXA_SCALE_ARG_B                   (1 << 22)
-#define     R200_TXA_NEG_ARG_B                     (1 << 23)
-#define     R200_TXA_COMP_ARG_C                    (1 << 24)
-#define     R200_TXA_COMP_ARG_C_SHIFT              (24)
-#define     R200_TXA_BIAS_ARG_C                    (1 << 25)
-#define     R200_TXA_SCALE_ARG_C                   (1 << 26)
-#define     R200_TXA_NEG_ARG_C                     (1 << 27)
-#define     R200_TXA_OP_MADD                       (0 << 28)
-#define     R200_TXA_OP_CND0                       (2 << 28)
-#define     R200_TXA_OP_LERP                       (3 << 28)
-#define     R200_TXA_OP_CONDITIONAL                (6 << 28)
-#define     R200_TXA_OP_MASK                       (7 << 28)
-#define R200_PP_TXABLEND2_0                0x2f0c
-#define     R200_TXA_TFACTOR_SEL_SHIFT             0
-#define     R200_TXA_TFACTOR_SEL_MASK              0x7
-#define     R200_TXA_TFACTOR1_SEL_SHIFT            4
-#define     R200_TXA_TFACTOR1_SEL_MASK             (0x7 << 4)
-#define     R200_TXA_SCALE_SHIFT                   8
-#define     R200_TXA_SCALE_MASK                    (7 << 8)
-#define     R200_TXA_SCALE_1X                      (0 << 8)
-#define     R200_TXA_SCALE_2X                      (1 << 8)
-#define     R200_TXA_SCALE_4X                      (2 << 8)
-#define     R200_TXA_SCALE_8X                      (3 << 8)
-#define     R200_TXA_SCALE_INV2                    (5 << 8)
-#define     R200_TXA_SCALE_INV4                    (6 << 8)
-#define     R200_TXA_SCALE_INV8                    (7 << 8)
-#define     R200_TXA_CLAMP_SHIFT                   12
-#define     R200_TXA_CLAMP_MASK                    (3 << 12)
-#define     R200_TXA_CLAMP_WRAP                    (0 << 12)
-#define     R200_TXA_CLAMP_0_1                     (1 << 12)
-#define     R200_TXA_CLAMP_8_8                     (2 << 12)
-#define     R200_TXA_OUTPUT_REG_MASK               (7 << 16)
-#define     R200_TXA_OUTPUT_REG_NONE               (0 << 16)
-#define     R200_TXA_OUTPUT_REG_R0                 (1 << 16)
-#define     R200_TXA_OUTPUT_REG_R1                 (2 << 16)
-#define     R200_TXA_OUTPUT_REG_R2                 (3 << 16)
-#define     R200_TXA_OUTPUT_REG_R3                 (4 << 16)
-#define     R200_TXA_OUTPUT_REG_R4                 (5 << 16)
-#define     R200_TXA_OUTPUT_REG_R5                 (6 << 16)
-#define     R200_TXA_DOT_ALPHA                     (1 << 20)
-#define     R200_TXA_REPL_NORMAL                   0
-#define     R200_TXA_REPL_RED                      1
-#define     R200_TXA_REPL_GREEN                    2
-#define     R200_TXA_REPL_ARG_A_SHIFT              26
-#define     R200_TXA_REPL_ARG_A_MASK               (3 << 26)
-#define     R200_TXA_REPL_ARG_B_SHIFT              28
-#define     R200_TXA_REPL_ARG_B_MASK               (3 << 28)
-#define     R200_TXA_REPL_ARG_C_SHIFT              30
-#define     R200_TXA_REPL_ARG_C_MASK               (3 << 30)
-#define R200_PP_TXCBLEND_1                0x2f10
-#define R200_PP_TXCBLEND2_1               0x2f14
-#define R200_PP_TXABLEND_1                0x2f18
-#define R200_PP_TXABLEND2_1               0x2f1c
-#define R200_PP_TXCBLEND_2                0x2f20
-#define R200_PP_TXCBLEND2_2               0x2f24
-#define R200_PP_TXABLEND_2                0x2f28
-#define R200_PP_TXABLEND2_2               0x2f2c
-#define R200_PP_TXCBLEND_3                0x2f30
-#define R200_PP_TXCBLEND2_3               0x2f34
-#define R200_PP_TXABLEND_3                0x2f38
-#define R200_PP_TXABLEND2_3               0x2f3c
-#define R200_PP_TXCBLEND_4                0x2f40
-#define R200_PP_TXCBLEND2_4               0x2f44
-#define R200_PP_TXABLEND_4                0x2f48
-#define R200_PP_TXABLEND2_4               0x2f4c
-#define R200_PP_TXCBLEND_5                0x2f50
-#define R200_PP_TXCBLEND2_5               0x2f54
-#define R200_PP_TXABLEND_5                0x2f58
-#define R200_PP_TXABLEND2_5               0x2f5c
-#define R200_PP_TXCBLEND_6                0x2f60
-#define R200_PP_TXCBLEND2_6               0x2f64
-#define R200_PP_TXABLEND_6                0x2f68
-#define R200_PP_TXABLEND2_6               0x2f6c
-#define R200_PP_TXCBLEND_7                0x2f70
-#define R200_PP_TXCBLEND2_7               0x2f74
-#define R200_PP_TXABLEND_7                0x2f78
-#define R200_PP_TXABLEND2_7               0x2f7c
-/* gap */
-#define R200_RB3D_BLENDCOLOR               0x3218	/* ARGB 8888 */
-#define R200_RB3D_ABLENDCNTL               0x321C	/* see BLENDCTL */
-#define R200_RB3D_CBLENDCNTL               0x3220	/* see BLENDCTL */
-
-/*
- * Offsets in TCL vector state.  NOTE: Hardwiring matrix positions.
- * Multiple contexts could collaberate to eliminate state bouncing.
- */
-#define R200_VS_LIGHT_AMBIENT_ADDR          0x00000028
-#define R200_VS_LIGHT_DIFFUSE_ADDR          0x00000030
-#define R200_VS_LIGHT_SPECULAR_ADDR         0x00000038
-#define R200_VS_LIGHT_DIRPOS_ADDR           0x00000040
-#define R200_VS_LIGHT_HWVSPOT_ADDR          0x00000048
-#define R200_VS_LIGHT_ATTENUATION_ADDR      0x00000050
-#define R200_VS_SPOT_DUAL_CONE              0x00000058
-#define R200_VS_GLOBAL_AMBIENT_ADDR         0x0000005C
-#define R200_VS_FOG_PARAM_ADDR              0x0000005D
-#define R200_VS_EYE_VECTOR_ADDR             0x0000005E
-#define R200_VS_UCP_ADDR                    0x00000060
-#define R200_VS_PNT_SPRITE_VPORT_SCALE      0x00000068
-#define R200_VS_MATRIX_0_MV                 0x00000080
-#define R200_VS_MATRIX_1_INV_MV        	    0x00000084
-#define R200_VS_MATRIX_2_MVP        	    0x00000088
-#define R200_VS_MATRIX_3_TEX0        	    0x0000008C
-#define R200_VS_MATRIX_4_TEX1        	    0x00000090
-#define R200_VS_MATRIX_5_TEX2        	    0x00000094
-#define R200_VS_MATRIX_6_TEX3        	    0x00000098
-#define R200_VS_MATRIX_7_TEX4        	    0x0000009C
-#define R200_VS_MATRIX_8_TEX5        	    0x000000A0
-#define R200_VS_MAT_0_EMISS                 0x000000B0
-#define R200_VS_MAT_0_AMB                   0x000000B1
-#define R200_VS_MAT_0_DIF                   0x000000B2
-#define R200_VS_MAT_0_SPEC                  0x000000B3
-#define R200_VS_MAT_1_EMISS                 0x000000B4
-#define R200_VS_MAT_1_AMB                   0x000000B5
-#define R200_VS_MAT_1_DIF                   0x000000B6
-#define R200_VS_MAT_1_SPEC                  0x000000B7
-#define R200_VS_EYE2CLIP_MTX                0x000000B8
-#define R200_VS_PNT_SPRITE_ATT_CONST        0x000000BC
-#define R200_VS_PNT_SPRITE_EYE_IN_MODEL     0x000000BD
-#define R200_VS_PNT_SPRITE_CLAMP            0x000000BE
-#define R200_VS_MAX                         0x000001C0
-
-/*
- * Offsets in TCL scalar state
- */
-#define R200_SS_LIGHT_DCD_ADDR              0x00000000
-#define R200_SS_LIGHT_DCM_ADDR              0x00000008
-#define R200_SS_LIGHT_SPOT_EXPONENT_ADDR    0x00000010
-#define R200_SS_LIGHT_SPOT_CUTOFF_ADDR      0x00000018
-#define R200_SS_LIGHT_SPECULAR_THRESH_ADDR  0x00000020
-#define R200_SS_LIGHT_RANGE_CUTOFF_SQRD     0x00000028
-#define R200_SS_LIGHT_RANGE_ATT_CONST       0x00000030
-#define R200_SS_VERT_GUARD_CLIP_ADJ_ADDR    0x00000080
-#define R200_SS_VERT_GUARD_DISCARD_ADJ_ADDR 0x00000081
-#define R200_SS_HORZ_GUARD_CLIP_ADJ_ADDR    0x00000082
-#define R200_SS_HORZ_GUARD_DISCARD_ADJ_ADDR 0x00000083
-#define R200_SS_MAT_0_SHININESS             0x00000100
-#define R200_SS_MAT_1_SHININESS             0x00000101
-
-/*
- * Matrix indices
- */
-#define R200_MTX_MV                        0
-#define R200_MTX_IMV                       1
-#define R200_MTX_MVP                       2
-#define R200_MTX_TEX0                      3
-#define R200_MTX_TEX1                      4
-#define R200_MTX_TEX2                      5
-#define R200_MTX_TEX3                      6
-#define R200_MTX_TEX4                      7
-#define R200_MTX_TEX5                      8
-
-/* Color formats for 2d packets
- */
-#define R200_CP_COLOR_FORMAT_CI8	2
-#define R200_CP_COLOR_FORMAT_ARGB1555	3
-#define R200_CP_COLOR_FORMAT_RGB565	4
-#define R200_CP_COLOR_FORMAT_ARGB8888	6
-#define R200_CP_COLOR_FORMAT_RGB332	7
-#define R200_CP_COLOR_FORMAT_RGB8	9
-#define R200_CP_COLOR_FORMAT_ARGB4444	15
-
-/*
- * CP type-3 packets
- */
-#define R200_CP_CMD_NOP                 0xC0001000
-#define R200_CP_CMD_NEXT_CHAR           0xC0001900
-#define R200_CP_CMD_PLY_NEXTSCAN        0xC0001D00
-#define R200_CP_CMD_SET_SCISSORS        0xC0001E00
-#define R200_CP_CMD_LOAD_MICROCODE      0xC0002400
-#define R200_CP_CMD_WAIT_FOR_IDLE       0xC0002600
-#define R200_CP_CMD_3D_DRAW_VBUF        0xC0002800
-#define R200_CP_CMD_3D_DRAW_IMMD        0xC0002900
-#define R200_CP_CMD_3D_DRAW_INDX        0xC0002A00
-#define R200_CP_CMD_LOAD_PALETTE        0xC0002C00
-#define R200_CP_CMD_3D_LOAD_VBPNTR      0xC0002F00
-#define R200_CP_CMD_INDX_BUFFER         0xC0003300
-#define R200_CP_CMD_3D_DRAW_VBUF_2      0xC0003400
-#define R200_CP_CMD_3D_DRAW_IMMD_2      0xC0003500
-#define R200_CP_CMD_3D_DRAW_INDX_2      0xC0003600
-#define R200_CP_CMD_PAINT		0xC0009100
-#define R200_CP_CMD_BITBLT		0xC0009200
-#define R200_CP_CMD_SMALLTEXT		0xC0009300
-#define R200_CP_CMD_HOSTDATA_BLT	0xC0009400
-#define R200_CP_CMD_POLYLINE		0xC0009500
-#define R200_CP_CMD_POLYSCANLINES	0xC0009800
-#define R200_CP_CMD_PAINT_MULTI		0xC0009A00
-#define R200_CP_CMD_BITBLT_MULTI	0xC0009B00
-#define R200_CP_CMD_TRANS_BITBLT	0xC0009C00
-
-#endif
diff --git a/src/mesa/drivers/dri/r300/r200_state.h b/src/mesa/drivers/dri/r300/r200_state.h
deleted file mode 100644
index 3e1a9c8ba19..00000000000
--- a/src/mesa/drivers/dri/r300/r200_state.h
+++ /dev/null
@@ -1,58 +0,0 @@
-/*
-Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
-
-The Weather Channel (TM) funded Tungsten Graphics to develop the
-initial release of the Radeon 8500 driver under the XFree86 license.
-This notice must be preserved.
-
-Permission is hereby granted, free of charge, to any person obtaining
-a copy of this software and associated documentation files (the
-"Software"), to deal in the Software without restriction, including
-without limitation the rights to use, copy, modify, merge, publish,
-distribute, sublicense, and/or sell copies of the Software, and to
-permit persons to whom the Software is furnished to do so, subject to
-the following conditions:
-
-The above copyright notice and this permission notice (including the
-next paragraph) shall be included in all copies or substantial
-portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
-EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
-IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
-LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
-OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
-WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
-
-**************************************************************************/
-
-/*
- * Authors:
- *   Keith Whitwell <[email protected]>
- */
-
-#ifndef __R200_STATE_H__
-#define __R200_STATE_H__
-
-#ifdef GLX_DIRECT_RENDERING
-
-#include "r200_context.h"
-
-extern void r200InitState(r200ContextPtr rmesa);
-extern void r200InitStateFuncs(struct dd_function_table *functions);
-extern void r200InitTnlFuncs(GLcontext * ctx);
-
-extern void r200UpdateMaterial(GLcontext * ctx);
-
-extern void r200UpdateViewportOffset(GLcontext * ctx);
-extern void r200UpdateWindow(GLcontext * ctx);
-
-extern void r200ValidateState(GLcontext * ctx);
-
-extern void r200PrintDirty(r200ContextPtr rmesa, const char *msg);
-
-extern void r200LightingSpaceChange(GLcontext * ctx);
-
-#endif
-#endif
diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 54eb081d055..d10a9d87d31 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -44,7 +44,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -73,6 +73,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 int future_hw_tcl_on=1;
 int hw_tcl_on=1;
 
+#define need_GL_EXT_stencil_two_side
 #define need_GL_ARB_multisample
 #define need_GL_ARB_texture_compression
 #define need_GL_ARB_vertex_buffer_object
@@ -126,6 +127,10 @@ const struct dri_extension card_extensions[] = {
   {NULL,				NULL}
 };
 
+const struct dri_extension stencil_two_side[] = {
+  {"GL_EXT_stencil_two_side",		GL_EXT_stencil_two_side_functions},
+};
+
 extern struct tnl_pipeline_stage _r300_render_stage;
 extern const struct tnl_pipeline_stage _r300_tcl_stage;
 extern const struct tnl_pipeline_stage _r300_texrect_stage;
@@ -195,6 +200,8 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	 */
 	driParseConfigFiles(&r300->radeon.optionCache, &screen->optionCache,
 			    screen->driScreen->myNum, "r300");
+	r300->initialMaxAnisotropy = driQueryOptionf(&r300->radeon.optionCache,
+						     "def_max_anisotropy");
 
 	//r300->texmicrotile = GL_TRUE;
 	
@@ -287,7 +294,7 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 	/* Initialize the software rasterizer and helper modules.
 	 */
 	_swrast_CreateContext(ctx);
-	_ac_CreateContext(ctx);
+	_vbo_CreateContext(ctx);
 	_tnl_CreateContext(ctx);
 	_swsetup_CreateContext(ctx);
 	_swsetup_Wakeup(ctx);
@@ -300,7 +307,7 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 
 	/* Try and keep materials and vertices separate:
 	 */
-	_tnl_isolate_materials(ctx, GL_TRUE);
+/* 	_tnl_isolate_materials(ctx, GL_TRUE); */
 
 	/* Configure swrast and TNL to match hardware characteristics:
 	 */
@@ -331,6 +338,9 @@ GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 
 	driInitExtensions(ctx, card_extensions, GL_TRUE);
 	
+	if (driQueryOptionb(&r300->radeon.optionCache, "disable_stencil_two_side") == 0)
+		driInitSingleExtension(ctx, stencil_two_side);
+	
 	if (r300->radeon.glCtx->Mesa_DXTn && !driQueryOptionb (&r300->radeon.optionCache, "disable_s3tc")) {
 	  _mesa_enable_extension( ctx, "GL_EXT_texture_compression_s3tc" );
 	  _mesa_enable_extension( ctx, "GL_S3_s3tc" );
@@ -478,7 +488,7 @@ void r300DestroyContext(__DRIcontextPrivate * driContextPriv)
 		_swsetup_DestroyContext(r300->radeon.glCtx);
 		_tnl_ProgramCacheDestroy(r300->radeon.glCtx);
 		_tnl_DestroyContext(r300->radeon.glCtx);
-		_ac_DestroyContext(r300->radeon.glCtx);
+		_vbo_DestroyContext(r300->radeon.glCtx);
 		_swrast_DestroyContext(r300->radeon.glCtx);
 		
 		if (r300->dma.current.buf) {
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index dd3ecbb2357..a339b2f9c16 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -48,7 +48,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_context.h"
 
 #define USER_BUFFERS
-#define RADEON_VTXFMT_A
+/* KW: Disable this code.  Driver should hook into vbo module
+ * directly, see i965 driver for example.
+ */
+/* #define RADEON_VTXFMT_A */
 #define HW_VBOS
 
 /* We don't handle 16 bits elts swapping yet */
@@ -57,7 +60,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #endif
 
 //#define OPTIMIZE_ELTS
-#define CB_DPATH
 
 struct r300_context;
 typedef struct r300_context r300ContextRec;
@@ -726,6 +728,11 @@ struct r300_fragment_program {
 	GLboolean params_uptodate;
 
 	int max_temp_idx;
+
+	/* the index of the sin constant is stored here */
+	GLint const_sin[2];
+	
+	GLuint optimization;
 };
 
 #define R300_MAX_AOS_ARRAYS		16
@@ -754,7 +761,7 @@ struct radeon_vertex_buffer {
 	
 	struct dt AttribPtr[VERT_ATTRIB_MAX];
 	
-	struct tnl_prim  *Primitive;	              
+	const struct _mesa_prim  *Primitive;	              
 	GLuint      PrimitiveCount;
 	GLint LockFirst;
 	GLsizei LockCount;
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.c b/src/mesa/drivers/dri/r300/r300_fragprog.c
index 179bc58e9e9..8e45bd54030 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.c
@@ -33,7 +33,6 @@
 
 /*TODO'S
  *
- * - COS/SIN/SCS instructions
  * - Depth write, WPOS/FOGC inputs
  * - FogOption
  * - Verify results of opcodes for accuracy, I've only checked them
@@ -187,6 +186,10 @@ static const struct {
 #define SLOT_VECTOR	(1<<0)
 #define SLOT_SCALAR	(1<<3)
 #define SLOT_BOTH	(SLOT_VECTOR | SLOT_SCALAR)
+
+/* mapping from SWIZZLE_* to r300 native values for scalar insns */
+#define SWIZZLE_HALF 6
+
 #define MAKE_SWZ3(x, y, z) (MAKE_SWIZZLE4(SWIZZLE_##x, \
 					  SWIZZLE_##y, \
 					  SWIZZLE_##z, \
@@ -208,7 +211,7 @@ static const struct r300_pfs_swizzle {
 	{ MAKE_SWZ3(W, Z, Y), R300_FPI0_ARGC_SRC0CA_WZY, 1, SLOT_BOTH },
 	{ MAKE_SWZ3(ONE, ONE, ONE), R300_FPI0_ARGC_ONE, 0, 0},
 	{ MAKE_SWZ3(ZERO, ZERO, ZERO), R300_FPI0_ARGC_ZERO, 0, 0},
-	{ PFS_INVAL, R300_FPI0_ARGC_HALF, 0, 0},
+	{ MAKE_SWZ3(HALF, HALF, HALF), R300_FPI0_ARGC_HALF, 0, 0},
 	{ PFS_INVAL, 0, 0, 0},
 };
 
@@ -232,8 +235,6 @@ static const struct {
 	{ PFS_INVAL, PFS_INVAL, PFS_INVAL}
 };
 
-/* mapping from SWIZZLE_* to r300 native values for scalar insns */
-#define SWIZZLE_HALF 6
 static const struct {
 	int base;	/* hw value of swizzle */
 	int stride;	/* difference between SRC0/1/2 */
@@ -463,7 +464,7 @@ static int swz_native(struct r300_fragment_program *rp,
 		      GLuint arbneg)
 {
 	/* Native swizzle, handle negation */
-	src = (src & ~REG_NEGS_SHIFT) |
+	src = (src & ~REG_NEGS_MASK) |
 		(((arbneg >> 3) & 1) << REG_NEGS_SHIFT);
 
 	if ((arbneg & 0x7) == 0x0) {
@@ -590,6 +591,7 @@ static GLuint do_swizzle(struct r300_fragment_program *rp,
 	/* If swizzling from something without an XYZW native swizzle,
 	 * emit result to a temp, and do new swizzle from the temp.
 	 */
+#if 0
 	if (REG_GET_VSWZ(src) != SWIZZLE_XYZ ||
 	    REG_GET_SSWZ(src) != SWIZZLE_W) {
 		GLuint temp = get_temp_reg(rp);
@@ -603,10 +605,30 @@ static GLuint do_swizzle(struct r300_fragment_program *rp,
 			   0);
 		src = temp;
 	}
+#endif
+
+	if (REG_GET_VSWZ(src) != SWIZZLE_XYZ ||
+	    REG_GET_SSWZ(src) != SWIZZLE_W) {
+	    GLuint vsrcswz = (v_swiz[REG_GET_VSWZ(src)].hash & (SWZ_X_MASK|SWZ_Y_MASK|SWZ_Z_MASK)) | REG_GET_SSWZ(src) << 9;
+	    GLint i;
 
-	/* set scalar swizzling */
-	REG_SET_SSWZ(src, GET_SWZ(arbswz, 3));
+	    GLuint newswz = 0;
+	    GLuint offset;
+	    for(i=0; i < 4; ++i){
+		offset = GET_SWZ(arbswz, i);
+		
+		newswz |= (offset <= 3)?GET_SWZ(vsrcswz, offset) << i*3:offset << i*3;
+	    }
+
+	    arbswz = newswz & (SWZ_X_MASK|SWZ_Y_MASK|SWZ_Z_MASK);
+	    REG_SET_SSWZ(src, GET_SWZ(newswz, 3));
+	}
+	else
+	{
+	    /* set scalar swizzling */
+	    REG_SET_SSWZ(src, GET_SWZ(arbswz, 3));
 
+	}
 	do {
 		vswz = REG_GET_VSWZ(src);
 		do {
@@ -1058,7 +1080,7 @@ static void emit_arith(struct r300_fragment_program *rp,
 				break;
 			}
 			if (emit_sop &&
-			    (s_swiz[REG_GET_VSWZ(src[i])].flags & SLOT_VECTOR)) {
+			    (s_swiz[REG_GET_SSWZ(src[i])].flags & SLOT_VECTOR)) {
 				vpos = spos = MAX2(vpos, spos);
 				break;
 			}
@@ -1181,6 +1203,25 @@ static GLuint get_attrib(struct r300_fragment_program *rp, GLuint attr)
 }
 #endif
 
+static void make_sin_const(struct r300_fragment_program *rp)
+{
+	if(rp->const_sin[0] == -1){
+	    GLfloat cnstv[4];
+
+	    cnstv[0] = 1.273239545; // 4/PI
+	    cnstv[1] =-0.405284735; // -4/(PI*PI)
+	    cnstv[2] = 3.141592654; // PI
+	    cnstv[3] = 0.2225;      // weight
+	    rp->const_sin[0] = emit_const4fv(rp, cnstv);
+
+	    cnstv[0] = 0.5;
+	    cnstv[1] = -1.5;
+	    cnstv[2] = 0.159154943; // 1/(2*PI)
+	    cnstv[3] = 6.283185307; // 2*PI
+	    rp->const_sin[1] = emit_const4fv(rp, cnstv);
+	}
+}
+
 static GLboolean parse_program(struct r300_fragment_program *rp)
 {	
 	struct gl_fragment_program *mp = &rp->mesa_program;
@@ -1234,62 +1275,71 @@ static GLboolean parse_program(struct r300_fragment_program *rp)
 			break;
 		case OPCODE_COS:
 			/*
-			 * cos using taylor serie:
-			 * cos(x) = 1 - x^2/2! + x^4/4! - x^6/6!
+			 * cos using a parabola (see SIN):
+			 * cos(x):
+			 *   x += PI/2
+			 *   x = (x/(2*PI))+0.5
+			 *   x = frac(x)
+			 *   x = (x*2*PI)-PI
+			 *   result = sin(x)
 			 */
 			temp = get_temp_reg(rp);
-			cnstv[0] = 0.5;
-			cnstv[1] = 0.041666667;
-			cnstv[2] = 0.001388889;
-			cnstv[4] = 0.0;
-			cnst = emit_const4fv(rp, cnstv);
+			make_sin_const(rp);
 			src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
 
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_XYZ,
-				   src[0],
-				   src[0],
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_Y | WRITEMASK_Z,
-				   temp, temp,
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_Z,
-				   temp,
-				   swizzle(temp, X, X, X, W),
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_XYZ,
-				   temp, cnst,
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   pfs_one,
-				   pfs_one,
-				   negate(temp),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   temp,
-				   pfs_one,
-				   swizzle(temp, Y, Y, Y, W),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   temp,
-				   pfs_one,
-				   negate(swizzle(temp, Z, Z, Z, W)),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, dest, mask,
+			/* add 0.5*PI and do range reduction */
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+				   swizzle(rp->const_sin[0], Z, Z, Z, Z), //PI
+				   pfs_half,
+				   swizzle(keep(src[0]), X, X, X, X),
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
 				   swizzle(temp, X, X, X, X),
-				   pfs_one,
+				   swizzle(rp->const_sin[1], Z, Z, Z, Z),
+				   pfs_half,
+				   0);
+
+			emit_arith(rp, PFS_OP_FRC, temp, WRITEMASK_X,
+				   swizzle(temp, X, X, X, X),
+				   undef,
+				   undef,
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Z,
+				   swizzle(temp, X, X, X, X),
+				   swizzle(rp->const_sin[1], W, W, W, W), //2*PI
+				   negate(swizzle(rp->const_sin[0], Z, Z, Z, Z)), //-PI
+				   0);
+
+			/* SIN */
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X | WRITEMASK_Y,
+				   swizzle(temp, Z, Z, Z, Z),
+				   rp->const_sin[0],
 				   pfs_zero,
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+				   swizzle(temp, Y, Y, Y, Y),
+				   absolute(swizzle(temp, Z, Z, Z, Z)),
+				   swizzle(temp, X, X, X, X),
+				   0);
+			
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Y,
+				   swizzle(temp, X, X, X, X),
+				   absolute(swizzle(temp, X, X, X, X)),
+				   negate(swizzle(temp, X, X, X, X)),
+				   0);
+
+
+	    		emit_arith(rp, PFS_OP_MAD, dest, mask,
+				   swizzle(temp, Y, Y, Y, Y),
+				   swizzle(rp->const_sin[0], W, W, W, W),
+				   swizzle(temp, X, X, X, X),
 				   flags);
+
 			free_temp(rp, temp);
 			break;
 		case OPCODE_DP3:
@@ -1398,7 +1448,7 @@ static GLboolean parse_program(struct r300_fragment_program *rp)
 			 * change the compare to (t.x + 0.5) > 0.5 we may
 			 * save one instruction by doing CMP -t.x 
 			 */
-			cnstv[0] = cnstv[1] = cnstv[2] = cnstv[4] = 0.50001;
+			cnstv[0] = cnstv[1] = cnstv[2] = cnstv[3] = 0.50001;
 			src[0] = t_src(rp, fpi->SrcReg[0]);
 			temp = get_temp_reg(rp);
 			cnst = emit_const4fv(rp, cnstv);
@@ -1529,7 +1579,93 @@ static GLboolean parse_program(struct r300_fragment_program *rp)
 				   flags);
 			break;
 		case OPCODE_SCS:
-			ERROR("SCS not implemented\n");
+			/*
+			 * cos using a parabola (see SIN):
+			 * cos(x):
+			 *   x += PI/2
+			 *   x = (x/(2*PI))+0.5
+			 *   x = frac(x)
+			 *   x = (x*2*PI)-PI
+			 *   result = sin(x)
+			 */
+			temp = get_temp_reg(rp);
+			make_sin_const(rp);
+			src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
+
+			/* add 0.5*PI and do range reduction */
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X|WRITEMASK_Y,
+				   swizzle(rp->const_sin[0], Z, Z, Z, Z),
+				   rp->const_sin[1],
+				   swizzle(keep(src[0]), X, X, X, X),
+				   0);
+
+			emit_arith(rp, PFS_OP_CMP, temp, WRITEMASK_W,
+				   swizzle(rp->const_sin[0], Z, Z, Z, Z),
+				   negate(pfs_half),
+				   swizzle(keep(src[0]), X, X, X, X),
+				   0);
+
+			emit_arith(rp, PFS_OP_CMP, temp, WRITEMASK_Z,
+				   swizzle(temp, X, X, X, X),
+				   swizzle(temp, Y, Y, Y, Y),
+				   swizzle(temp, W, W, W, W),
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X | WRITEMASK_Y,
+			           swizzle(temp, Z, Z, Z, Z),
+				   rp->const_sin[0],
+			           pfs_zero,
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_W,
+				   swizzle(temp, Y, Y, Y, Y),
+				   absolute(swizzle(temp, Z, Z, Z, Z)),
+				   swizzle(temp, X, X, X, X),
+				   0);
+
+			if(mask & WRITEMASK_Y)
+			{
+			    emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X | WRITEMASK_Y,
+				       swizzle(keep(src[0]), X, X, X, X),
+				       rp->const_sin[0],
+				       pfs_zero,
+				       0);
+
+			    emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+				       swizzle(temp, Y, Y, Y, Y),
+				       absolute(swizzle(keep(src[0]), X, X, X, X)),
+				       swizzle(temp, X, X, X, X),
+				       0);
+			}
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Z,
+				   swizzle(temp, W, W, W, W),
+				   absolute(swizzle(temp, W, W, W, W)),
+				   negate(swizzle(temp, W, W, W, W)),
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, dest, WRITEMASK_X,
+				   swizzle(temp, Z, Z, Z, Z),
+				   swizzle(rp->const_sin[0], W, W, W, W),
+				   swizzle(temp, W, W, W, W),
+				   flags);
+
+			if(mask & WRITEMASK_Y)
+			{
+			    emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_W,
+				       swizzle(temp, X, X, X, X),
+				       absolute(swizzle(temp, X, X, X, X)),
+				       negate(swizzle(temp, X, X, X, X)),
+				       0);
+
+	    		    emit_arith(rp, PFS_OP_MAD, dest, WRITEMASK_Y,
+				       swizzle(temp, W, W, W, W),
+				       swizzle(rp->const_sin[0], W, W, W, W),
+				       swizzle(temp, X, X, X, X),
+				       flags);
+			}
+			free_temp(rp, temp);
 			break;
 		case OPCODE_SGE:
 			src[0] = t_src(rp, fpi->SrcReg[0]);
@@ -1548,68 +1684,63 @@ static GLboolean parse_program(struct r300_fragment_program *rp)
 			break;
 		case OPCODE_SIN:
 			/*
-			 * sin using taylor serie:
-			 * sin(x) = x - x^3/3! + x^5/5! - x^7/7!
+			 *  using a parabola:
+			 * sin(x) = 4/pi * x + -4/(pi*pi) * x * abs(x)
+			 * extra precision is obtained by weighting against
+			 * itself squared.
 			 */
+
 			temp = get_temp_reg(rp);
-			cnstv[0] = 0.333333333;
-			cnstv[1] = 0.008333333;
-			cnstv[2] = 0.000198413;
-			cnstv[4] = 0.0;
-			cnst = emit_const4fv(rp, cnstv);
+			make_sin_const(rp);
 			src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
 
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_XYZ,
-				   src[0],
-				   src[0],
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_Y | WRITEMASK_Z,
-				   temp, temp,
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_Z,
-				   temp,
-				   swizzle(temp, X, X, X, W),
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_XYZ,
-				   src[0],
-				   temp,
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_XYZ,
-				   temp, cnst,
-				   pfs_zero,
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   src[0],
-				   pfs_one,
-				   negate(temp),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   temp,
-				   pfs_one,
-				   swizzle(temp, Y, Y, Y, W),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, temp,
-				   WRITEMASK_X,
-				   temp,
-				   pfs_one,
-				   negate(swizzle(temp, Z, Z, Z, W)),
-				   flags);
-			emit_arith(rp, PFS_OP_MAD, dest, mask,
+			/* do range reduction */
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+				   swizzle(keep(src[0]), X, X, X, X),
+				   swizzle(rp->const_sin[1], Z, Z, Z, Z),
+				   pfs_half,
+				   0);
+
+			emit_arith(rp, PFS_OP_FRC, temp, WRITEMASK_X,
 				   swizzle(temp, X, X, X, X),
-				   pfs_one,
+				   undef,
+				   undef,
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Z,
+				   swizzle(temp, X, X, X, X),
+				   swizzle(rp->const_sin[1], W, W, W, W), //2*PI
+				   negate(swizzle(rp->const_sin[0], Z, Z, Z, Z)), //PI
+				   0);
+
+			/* SIN */
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X | WRITEMASK_Y,
+				   swizzle(temp, Z, Z, Z, Z),
+				   rp->const_sin[0],
 				   pfs_zero,
+				   0);
+
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+				   swizzle(temp, Y, Y, Y, Y),
+				   absolute(swizzle(temp, Z, Z, Z, Z)),
+				   swizzle(temp, X, X, X, X),
+				   0);
+			
+			emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Y,
+				   swizzle(temp, X, X, X, X),
+				   absolute(swizzle(temp, X, X, X, X)),
+				   negate(swizzle(temp, X, X, X, X)),
+				   0);
+
+
+	    		emit_arith(rp, PFS_OP_MAD, dest, mask,
+				   swizzle(temp, Y, Y, Y, Y),
+				   swizzle(rp->const_sin[0], W, W, W, W),
+				   swizzle(temp, X, X, X, X),
 				   flags);
+
 			free_temp(rp, temp);
 			break;
 		case OPCODE_SLT:
@@ -1681,7 +1812,7 @@ static GLboolean parse_program(struct r300_fragment_program *rp)
 /* - Init structures
  * - Determine what hwregs each input corresponds to
  */
-static void init_program(struct r300_fragment_program *rp)
+static void init_program(r300ContextPtr r300, struct r300_fragment_program *rp)
 {
 	struct r300_pfs_compile_state *cs = NULL;
 	struct gl_fragment_program *mp = &rp->mesa_program;	
@@ -1691,6 +1822,7 @@ static void init_program(struct r300_fragment_program *rp)
 	int i,j;
 
 	/* New compile, reset tracking data */
+	rp->optimization = driQueryOptioni(&r300->radeon.optionCache, "fp_optimization");
 	rp->translated = GL_FALSE;
 	rp->error      = GL_FALSE;
 	rp->cs = cs	   = &(R300_CONTEXT(rp->ctx)->state.pfs_compile);
@@ -1703,6 +1835,7 @@ static void init_program(struct r300_fragment_program *rp)
 	rp->max_temp_idx = 0;
 	rp->node[0].alu_end = -1;
 	rp->node[0].tex_end = -1;
+	rp->const_sin[0] = -1;
 	
 	_mesa_memset(cs, 0, sizeof(*rp->cs));
 	for (i=0;i<PFS_MAX_ALU_INST;i++) {
@@ -1816,13 +1949,13 @@ static void update_params(struct r300_fragment_program *rp)
 	rp->params_uptodate = GL_TRUE;
 }
 
-void r300_translate_fragment_shader(struct r300_fragment_program *rp)
+void r300_translate_fragment_shader(r300ContextPtr r300, struct r300_fragment_program *rp)
 {
 	struct r300_pfs_compile_state *cs = NULL;
 
 	if (!rp->translated) {
 		
-		init_program(rp);
+		init_program(r300, rp);
 		cs = rp->cs;
 
 		if (parse_program(rp) == GL_FALSE) {
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.h b/src/mesa/drivers/dri/r300/r300_fragprog.h
index b0cebe60bb0..73986abc3cc 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.h
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.h
@@ -112,8 +112,11 @@ typedef struct r300_fragment_program_swizzle {
 		((0 | SRC_CONST) << R300_FPI3_SRC1A_SHIFT) | \
 		((0 | SRC_CONST) << R300_FPI3_SRC2A_SHIFT))
 
+#define DRI_CONF_FP_OPTIMIZATION_SPEED   0
+#define DRI_CONF_FP_OPTIMIZATION_QUALITY 1
+
 struct r300_fragment_program;
 
-extern void r300_translate_fragment_shader(struct r300_fragment_program *rp);
+extern void r300_translate_fragment_shader(r300ContextPtr r300, struct r300_fragment_program *rp);
 
 #endif
diff --git a/src/mesa/drivers/dri/r300/r300_ioctl.c b/src/mesa/drivers/dri/r300/r300_ioctl.c
index 0774a2af6b4..c5ad0260457 100644
--- a/src/mesa/drivers/dri/r300/r300_ioctl.c
+++ b/src/mesa/drivers/dri/r300/r300_ioctl.c
@@ -70,10 +70,6 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 	drm_radeon_cmd_header_t *cmd = NULL;
 	r300ContextPtr rmesa=r300;
 
-#ifndef CB_DPATH
-	int i;
-#endif
-	
 	if (RADEON_DEBUG & DEBUG_IOCTL)
 		fprintf(stderr, "%s: %s buffer (%i,%i %ix%i)\n",
 			__FUNCTION__, buffer ? "back" : "front",
@@ -89,168 +85,8 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 
 	cboffset += r300->radeon.radeonScreen->fbLocation;
 
-#ifndef CB_DPATH
-	R300_STATECHANGE(r300, vir[0]);
-	((drm_r300_cmd_header_t*)r300->hw.vir[0].cmd)->packet0.count = 1;
-	r300->hw.vir[0].cmd[1] = 0x21030003;
-
-	R300_STATECHANGE(r300, vir[1]);
-	((drm_r300_cmd_header_t*)r300->hw.vir[1].cmd)->packet0.count = 1;
-	r300->hw.vir[1].cmd[1] = 0xF688F688;
-
-	R300_STATECHANGE(r300, vic);
-	r300->hw.vic.cmd[R300_VIC_CNTL_0] = 0x00000001;
-	r300->hw.vic.cmd[R300_VIC_CNTL_1] = 0x00000405;
-	
-	R300_STATECHANGE(r300, vof);
-	r300->hw.vof.cmd[R300_VOF_CNTL_0] = R300_VAP_OUTPUT_VTX_FMT_0__POS_PRESENT
-				| R300_VAP_OUTPUT_VTX_FMT_0__COLOR_PRESENT;
-	r300->hw.vof.cmd[R300_VOF_CNTL_1] = 0; /* no textures */
-	
-	R300_STATECHANGE(r300, txe);
-	r300->hw.txe.cmd[R300_TXE_ENABLE] = 0;
-	
-	R300_STATECHANGE(r300, vpt);
-	r300->hw.vpt.cmd[R300_VPT_XSCALE] = r300PackFloat32(1.0);
-	r300->hw.vpt.cmd[R300_VPT_XOFFSET] = r300PackFloat32(dPriv->x);
-	r300->hw.vpt.cmd[R300_VPT_YSCALE] = r300PackFloat32(1.0);
-	r300->hw.vpt.cmd[R300_VPT_YOFFSET] = r300PackFloat32(dPriv->y);
-	r300->hw.vpt.cmd[R300_VPT_ZSCALE] = r300PackFloat32(1.0);
-	r300->hw.vpt.cmd[R300_VPT_ZOFFSET] = r300PackFloat32(0.0);
-
-	R300_STATECHANGE(r300, at);
-	r300->hw.at.cmd[R300_AT_ALPHA_TEST] = 0;
-	
-	R300_STATECHANGE(r300, bld);
-	r300->hw.bld.cmd[R300_BLD_CBLEND] = 0;
-	r300->hw.bld.cmd[R300_BLD_ABLEND] = 0;
-	
-	if (r300->radeon.radeonScreen->cpp == 4)
-		cbpitch |= R300_COLOR_FORMAT_ARGB8888;
-	else
-		cbpitch |= R300_COLOR_FORMAT_RGB565;
-	
-	if (r300->radeon.sarea->tiling_enabled)
-		cbpitch |= R300_COLOR_TILE_ENABLE;
-	
-	R300_STATECHANGE(r300, cb);
-	r300->hw.cb.cmd[R300_CB_OFFSET] = cboffset;
-	r300->hw.cb.cmd[R300_CB_PITCH] = cbpitch;
-
-	R300_STATECHANGE(r300, unk221C);
-	r300->hw.unk221C.cmd[1] = R300_221C_CLEAR;
-
-	R300_STATECHANGE(r300, ps);
-	r300->hw.ps.cmd[R300_PS_POINTSIZE] =
-		((dPriv->w * 6) << R300_POINTSIZE_X_SHIFT) |
-		((dPriv->h * 6) << R300_POINTSIZE_Y_SHIFT);
-
-	R300_STATECHANGE(r300, ri);
-	for(i = 1; i <= 8; ++i)
-		r300->hw.ri.cmd[i] = R300_RS_INTERP_USED;
-
-	R300_STATECHANGE(r300, rc);
-	/* The second constant is needed to get glxgears display anything .. */
-	r300->hw.rc.cmd[1] = (1 << R300_RS_CNTL_CI_CNT_SHIFT) | R300_RS_CNTL_0_UNKNOWN_18;
-	r300->hw.rc.cmd[2] = 0;
-	
-	R300_STATECHANGE(r300, rr);
-	((drm_r300_cmd_header_t*)r300->hw.rr.cmd)->packet0.count = 1;
-	r300->hw.rr.cmd[1] = 0x00004000;
-
-	R300_STATECHANGE(r300, cmk);
-	if (flags & CLEARBUFFER_COLOR) {
-		r300->hw.cmk.cmd[R300_CMK_COLORMASK] =
-			(ctx->Color.ColorMask[BCOMP] ? R300_COLORMASK0_B : 0) |
-			(ctx->Color.ColorMask[GCOMP] ? R300_COLORMASK0_G : 0) |
-			(ctx->Color.ColorMask[RCOMP] ? R300_COLORMASK0_R : 0) |
-			(ctx->Color.ColorMask[ACOMP] ? R300_COLORMASK0_A : 0);
-	} else {
-		r300->hw.cmk.cmd[R300_CMK_COLORMASK] = 0;
-	}
-
-	R300_STATECHANGE(r300, fp);
-	r300->hw.fp.cmd[R300_FP_CNTL0] = 0; /* 1 pass, no textures */
-	r300->hw.fp.cmd[R300_FP_CNTL1] = 0; /* no temporaries */
-	r300->hw.fp.cmd[R300_FP_CNTL2] = 0; /* no offset, one ALU instr */
-	r300->hw.fp.cmd[R300_FP_NODE0] = 0;
-	r300->hw.fp.cmd[R300_FP_NODE1] = 0;
-	r300->hw.fp.cmd[R300_FP_NODE2] = 0;
-	r300->hw.fp.cmd[R300_FP_NODE3] = R300_PFS_NODE_OUTPUT_COLOR;
-
-	R300_STATECHANGE(r300, fpi[0]);
-	R300_STATECHANGE(r300, fpi[1]);
-	R300_STATECHANGE(r300, fpi[2]);
-	R300_STATECHANGE(r300, fpi[3]);
-	((drm_r300_cmd_header_t*)r300->hw.fpi[0].cmd)->packet0.count = 1;
-	((drm_r300_cmd_header_t*)r300->hw.fpi[1].cmd)->packet0.count = 1;
-	((drm_r300_cmd_header_t*)r300->hw.fpi[2].cmd)->packet0.count = 1;
-	((drm_r300_cmd_header_t*)r300->hw.fpi[3].cmd)->packet0.count = 1;
-
-	/* MOV o0, t0 */
-	r300->hw.fpi[0].cmd[1] = FP_INSTRC(MAD, FP_ARGC(SRC0C_XYZ), FP_ARGC(ONE), FP_ARGC(ZERO));
-	r300->hw.fpi[1].cmd[1] = FP_SELC(0,NO,XYZ,FP_TMP(0),0,0);
-	r300->hw.fpi[2].cmd[1] = FP_INSTRA(MAD, FP_ARGA(SRC0A), FP_ARGA(ONE), FP_ARGA(ZERO));
-	r300->hw.fpi[3].cmd[1] = FP_SELA(0,NO,W,FP_TMP(0),0,0);
-
-	R300_STATECHANGE(r300, pvs);
-	r300->hw.pvs.cmd[R300_PVS_CNTL_1] =
-		(0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT) |
-		(0 << R300_PVS_CNTL_1_POS_END_SHIFT) |
-		(1 << R300_PVS_CNTL_1_PROGRAM_END_SHIFT);
-	r300->hw.pvs.cmd[R300_PVS_CNTL_2] = 0; /* no parameters */
-	r300->hw.pvs.cmd[R300_PVS_CNTL_3] =
-		(1 << R300_PVS_CNTL_3_PROGRAM_UNKNOWN_SHIFT);
-
-	R300_STATECHANGE(r300, vpi);
-	((drm_r300_cmd_header_t*)r300->hw.vpi.cmd)->packet0.count = 8;
-
-	/* MOV o0, i0; */
-	r300->hw.vpi.cmd[1] = VP_OUT(ADD,OUT,0,XYZW);
-	r300->hw.vpi.cmd[2] = VP_IN(IN,0);
-	r300->hw.vpi.cmd[3] = VP_ZERO();
-	r300->hw.vpi.cmd[4] = 0;
-
-	/* MOV o1, i1; */
-	r300->hw.vpi.cmd[5] = VP_OUT(ADD,OUT,1,XYZW);
-	r300->hw.vpi.cmd[6] = VP_IN(IN,1);
-	r300->hw.vpi.cmd[7] = VP_ZERO();
-	r300->hw.vpi.cmd[8] = 0;
-
-	R300_STATECHANGE(r300, zs);
-	r300->hw.zs.cmd[R300_ZS_CNTL_0] = 0;
-	r300->hw.zs.cmd[R300_ZS_CNTL_1] = 0;
-	if (flags & CLEARBUFFER_DEPTH) {
-		r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_RB3D_Z_WRITE_ONLY;
-		r300->hw.zs.cmd[R300_ZS_CNTL_1] |= (R300_ZS_ALWAYS<<R300_RB3D_ZS1_DEPTH_FUNC_SHIFT);
-	} else {
-		r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_RB3D_Z_DISABLED_1; // disable
-	}
-	
-	R300_STATECHANGE(r300, zs);
-	if (flags & CLEARBUFFER_STENCIL) {
-		r300->hw.zs.cmd[R300_ZS_CNTL_0] |= R300_RB3D_STENCIL_ENABLE;
-		r300->hw.zs.cmd[R300_ZS_CNTL_1] |= 
-		    (R300_ZS_ALWAYS<<R300_RB3D_ZS1_FRONT_FUNC_SHIFT) | 
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_FRONT_FAIL_OP_SHIFT) |
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT) |
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_FRONT_ZFAIL_OP_SHIFT) |
-		    (R300_ZS_ALWAYS<<R300_RB3D_ZS1_BACK_FUNC_SHIFT) |
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT) |
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT) |
-		    (R300_ZS_REPLACE<<R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT) ;
-		r300->hw.zs.cmd[R300_ZS_CNTL_2] = r300->state.stencil.clear;
-	}
-			
-	/* Make sure we have enough space */
-	r300EnsureCmdBufSpace(r300, r300->hw.max_state_size + 9+8, __FUNCTION__);
-
-	r300EmitState(r300);
-#else
-#if 1
 	cp_wait(r300, R300_WAIT_3D | R300_WAIT_3D_CLEAN);
 	end_3d(rmesa);
-#endif
 	
 	R300_STATECHANGE(r300, cb);
 	reg_start(R300_RB3D_COLOROFFSET0, 0);
@@ -313,8 +149,6 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 	e32(r300->state.stencil.clear);
 	}
 	
-#endif
-
 	cmd2 = (drm_r300_cmd_header_t*)r300AllocCmdBuf(r300, 9, __FUNCTION__);
 	cmd2[0].packet3.cmd_type = R300_CMD_PACKET3;
 	cmd2[0].packet3.packet = R300_CMD_PACKET3_CLEAR;
@@ -327,7 +161,6 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 	cmd2[7].u = r300PackFloat32(ctx->Color.ClearColor[2]);
 	cmd2[8].u = r300PackFloat32(ctx->Color.ClearColor[3]);
 
-#if 1
 	reg_start(R300_RB3D_DSTCACHE_CTLSTAT,0);
 	e32(0x0000000a);
 	  
@@ -335,10 +168,8 @@ static void r300ClearBuffer(r300ContextPtr r300, int flags, int buffer)
 	reg_start(0x4f18,0);
 	e32(0x00000003);
 	cp_wait(rmesa, R300_WAIT_3D | R300_WAIT_3D_CLEAN);
-#endif
 }
 
-#ifdef CB_DPATH
 static void r300EmitClearState(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
@@ -472,7 +303,6 @@ static void r300EmitClearState(GLcontext * ctx)
 	/*reg_start(0x4500,0);
 	e32(2560-1);*/
 }
-#endif
 
 /**
  * Buffer clear
@@ -524,12 +354,10 @@ static void r300Clear(GLcontext * ctx, GLbitfield mask)
 
 	swapped = r300->radeon.doPageFlip && (r300->radeon.sarea->pfCurrentPage == 1);
 
-#ifdef CB_DPATH
 	/* Make sure it fits there. */
 	r300EnsureCmdBufSpace(r300, 421*3, __FUNCTION__);
 	if(flags || bits)
 		r300EmitClearState(ctx);
-#endif
 
 	if (flags & BUFFER_BIT_FRONT_LEFT) {
 		r300ClearBuffer(r300, bits | CLEARBUFFER_COLOR, swapped);
@@ -544,15 +372,6 @@ static void r300Clear(GLcontext * ctx, GLbitfield mask)
 	if (bits)
 		r300ClearBuffer(r300, bits, 0);
 
-#ifndef CB_DPATH
-	/* Recalculate the hardware state. This could be done more efficiently,
-	 * but do keep it like this for now.
-	 */
-	r300ResetHwState(r300);
-	
-	/* r300ClearBuffer has trampled all over the hardware state.. */
-	r300->hw.all_dirty=GL_TRUE;
-#endif
 }
 
 
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 03f168365dc..211c451f661 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -42,7 +42,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "api_arrayelt.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_vp_build.h"
 
@@ -352,7 +352,7 @@ GLboolean r300_run_vb_render(GLcontext *ctx,
 	r300EmitState(rmesa);
 	
 	for(i=0; i < VB->PrimitiveCount; i++){
-		GLuint prim = VB->Primitive[i].mode;
+		GLuint prim = _tnl_translate_prim(&VB->Primitive[i]);
 		GLuint start = VB->Primitive[i].start;
 		GLuint length = VB->Primitive[i].count;
 		
@@ -385,8 +385,18 @@ GLboolean r300_run_vb_render(GLcontext *ctx,
 int r300Fallback(GLcontext *ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
+	struct r300_fragment_program *rp =
+		(struct r300_fragment_program *)
+		(char *)ctx->FragmentProgram._Current;
 	int i;
 
+	if (rp) {
+		if (!rp->translated)
+			r300_translate_fragment_shader(r300, rp);
+
+		FALLBACK_IF(!rp->translated);
+	}
+
 	/* We do not do SELECT or FEEDBACK (yet ?)
 	 * Is it worth doing them ?
 	 */
@@ -406,6 +416,10 @@ int r300Fallback(GLcontext *ctx)
 	 */
 	FALLBACK_IF(ctx->Fog.Enabled);
 #endif
+	FALLBACK_IF(ctx->Stencil._TestTwoSide &&
+		    (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[1] ||
+		     ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[1] ||
+		     ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[1]));
 
 	if(!r300->disable_lowimpact_fallback){
 		/* GL_POLYGON_OFFSET_POINT */
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index 72255066d59..b30ece160a0 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -46,7 +46,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "api_arrayelt.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "texformat.h"
 
@@ -509,7 +509,6 @@ static void r300Enable(GLcontext* ctx, GLenum cap, GLboolean state)
 		if (r300->state.stencil.hw_stencil) {
 			R300_STATECHANGE(r300, zs);
 			if (state) {
-				WARN_ONCE("TODO - double side stencil !\n");
 				r300->hw.zs.cmd[R300_ZS_CNTL_0] |=
 				    R300_RB3D_STENCIL_ENABLE;
 			} else {
@@ -863,9 +862,12 @@ static void r300StencilFuncSeparate(GLcontext * ctx, GLenum face,
 						(R300_RB3D_ZS2_STENCIL_MASK << R300_RB3D_ZS2_STENCIL_MASK_SHIFT));
 	
 	flag = translate_func(ctx->Stencil.Function[0]);
-
-	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |= (flag << R300_RB3D_ZS1_FRONT_FUNC_SHIFT)
-					  | (flag << R300_RB3D_ZS1_BACK_FUNC_SHIFT);
+	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |= (flag << R300_RB3D_ZS1_FRONT_FUNC_SHIFT);
+	
+	if (ctx->Stencil._TestTwoSide)
+		flag = translate_func(ctx->Stencil.Function[1]);
+	
+	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |= (flag << R300_RB3D_ZS1_BACK_FUNC_SHIFT);
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_2] |= refmask;
 }
 
@@ -894,10 +896,19 @@ static void r300StencilOpSeparate(GLcontext * ctx, GLenum face, GLenum fail,
 	rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
 		 (translate_stencil_op(ctx->Stencil.FailFunc[0]) << R300_RB3D_ZS1_FRONT_FAIL_OP_SHIFT)
 		|(translate_stencil_op(ctx->Stencil.ZFailFunc[0]) << R300_RB3D_ZS1_FRONT_ZFAIL_OP_SHIFT)
-		|(translate_stencil_op(ctx->Stencil.ZPassFunc[0]) << R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT)
-		|(translate_stencil_op(ctx->Stencil.FailFunc[0]) << R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT)
-		|(translate_stencil_op(ctx->Stencil.ZFailFunc[0]) << R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT)
-		|(translate_stencil_op(ctx->Stencil.ZPassFunc[0]) << R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT);
+		|(translate_stencil_op(ctx->Stencil.ZPassFunc[0]) << R300_RB3D_ZS1_FRONT_ZPASS_OP_SHIFT);
+	
+	if (ctx->Stencil._TestTwoSide) {
+		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
+			 (translate_stencil_op(ctx->Stencil.FailFunc[1]) << R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT)
+			|(translate_stencil_op(ctx->Stencil.ZFailFunc[1]) << R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT)
+			|(translate_stencil_op(ctx->Stencil.ZPassFunc[1]) << R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT);
+	} else {
+		rmesa->hw.zs.cmd[R300_ZS_CNTL_1] |=
+			 (translate_stencil_op(ctx->Stencil.FailFunc[0]) << R300_RB3D_ZS1_BACK_FAIL_OP_SHIFT)
+			|(translate_stencil_op(ctx->Stencil.ZFailFunc[0]) << R300_RB3D_ZS1_BACK_ZFAIL_OP_SHIFT)
+			|(translate_stencil_op(ctx->Stencil.ZPassFunc[0]) << R300_RB3D_ZS1_BACK_ZPASS_OP_SHIFT);
+	}
 }
 
 static void r300ClearStencil(GLcontext * ctx, GLint s)
@@ -1757,9 +1768,7 @@ void r300UpdateShaderStates(r300ContextPtr rmesa)
 	GLcontext *ctx;
 	ctx = rmesa->radeon.glCtx;
 	
-#ifdef CB_DPATH
 	r300UpdateTextureState(ctx);
-#endif
 
 	r300SetupPixelShader(rmesa);
 	r300_setup_textures(ctx);
@@ -1809,10 +1818,10 @@ void r300SetupPixelShader(r300ContextPtr rmesa)
 	if (!rp)	/* should only happenen once, just after context is created */
 		return;
 	
-	r300_translate_fragment_shader(rp);
+	r300_translate_fragment_shader(rmesa, rp);
 	if (!rp->translated) {
 		fprintf(stderr, "%s: No valid fragment shader, exiting\n", __func__);
-		exit(-1);
+		return;
 	}
 	
 #define OUTPUT_FIELD(st, reg, field)  \
@@ -1874,7 +1883,7 @@ static void r300InvalidateState(GLcontext * ctx, GLuint new_state)
 	
 	_swrast_InvalidateState(ctx, new_state);
 	_swsetup_InvalidateState(ctx, new_state);
-	_ac_InvalidateState(ctx, new_state);
+	_vbo_InvalidateState(ctx, new_state);
 	_tnl_InvalidateState(ctx, new_state);
 	_ae_invalidate_state(ctx, new_state);
 
@@ -1884,10 +1893,6 @@ static void r300InvalidateState(GLcontext * ctx, GLuint new_state)
 
 	r300UpdateStateParameters(ctx, new_state);
 
-#ifndef CB_DPATH
-	/* Go inefficiency! */
-	r300ResetHwState(r300);
-#endif
 #ifdef HW_VBOS
 	if(new_state & _NEW_ARRAY)
 		r300->state.VB.lock_uptodate = GL_FALSE;
diff --git a/src/mesa/drivers/dri/r300/r300_vertexprog.c b/src/mesa/drivers/dri/r300/r300_vertexprog.c
index 2492a4a3a0a..2ff92e13289 100644
--- a/src/mesa/drivers/dri/r300/r300_vertexprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertexprog.c
@@ -960,26 +960,23 @@ static void position_invariant(struct gl_program *prog)
 
 static void insert_wpos(struct r300_vertex_program *vp,
 		       struct gl_program *prog,
-		       GLint pos)
+		       GLuint temp_index)
 {
 
 	GLint tokens[6] = { STATE_INTERNAL, STATE_R300_WINDOW_DIMENSION, 0, 0, 0, 0 };
 	struct prog_instruction *vpi;
 	struct prog_instruction *vpi_insert;
-	GLuint temp_index;
 	GLuint window_index;
 	int i = 0;
 	
 	vpi = malloc((prog->NumInstructions + 5) * sizeof(struct prog_instruction));
-	memcpy(vpi, prog->Instructions, (pos+1) * sizeof(struct prog_instruction));
+	/* all but END */
+	memcpy(vpi, prog->Instructions, (prog->NumInstructions - 1) * sizeof(struct prog_instruction));
+	/* END */
+	memcpy(&vpi[prog->NumInstructions + 4], &prog->Instructions[prog->NumInstructions - 1],
+		sizeof(struct prog_instruction));
 	
-	vpi_insert = &vpi[pos];
-
-	/* make a copy before outputting VERT_RESULT_HPOS */
-	vpi_insert->DstReg.File = vpi_insert->SrcReg[2].File;
-	vpi_insert->DstReg.Index = temp_index = vpi_insert->SrcReg[2].Index;
-	
-	vpi_insert++;
+	vpi_insert = &vpi[prog->NumInstructions - 1];
 	memset(vpi_insert, 0, 5 * sizeof(struct prog_instruction));
 
 	vpi_insert[i].Opcode = OPCODE_MOV;
@@ -1062,8 +1059,6 @@ static void insert_wpos(struct r300_vertex_program *vp,
 	vpi_insert[i].SrcReg[1].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_ONE, SWIZZLE_ONE);
 	i++;
 
-	memcpy(&vpi_insert[i], &prog->Instructions[pos+1], (prog->NumInstructions-(pos+1)) * sizeof(struct prog_instruction));
-
 	free(prog->Instructions);
 
 	prog->Instructions = vpi;
@@ -1078,16 +1073,18 @@ static void pos_as_texcoord(struct r300_vertex_program *vp,
 			    struct gl_program *prog)
 {
 	struct prog_instruction *vpi;
-	int pos = 0;
-	
-	for(vpi = prog->Instructions; vpi->Opcode != OPCODE_END; vpi++, pos++){
+	GLuint tempregi = prog->NumTemporaries;
+	/* should do something else if no temps left... */
+	prog->NumTemporaries++;
+
+	for(vpi = prog->Instructions; vpi->Opcode != OPCODE_END; vpi++){
 		if( vpi->DstReg.File == PROGRAM_OUTPUT &&
 		    vpi->DstReg.Index == VERT_RESULT_HPOS ){
-			insert_wpos(vp, prog, pos);
-			break;
+			vpi->DstReg.File = PROGRAM_TEMPORARY;
+			vpi->DstReg.Index = tempregi;
 		}
 	}
-
+	insert_wpos(vp, prog, tempregi);
 }
 
 static struct r300_vertex_program *build_program(struct r300_vertex_program_key *wanted_key,
@@ -1101,8 +1098,9 @@ static struct r300_vertex_program *build_program(struct r300_vertex_program_key
 
 	vp->wpos_idx = wpos_idx;
 
-	if(mesa_vp->IsPositionInvariant)
+	if(mesa_vp->IsPositionInvariant) {
 		position_invariant(&mesa_vp->Base);
+	}
 
 	if(wpos_idx > -1)
 		pos_as_texcoord(vp, &mesa_vp->Base);
@@ -1158,6 +1156,10 @@ void r300_select_vertex_shader(r300ContextPtr r300)
 			wanted_key.OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i);
 
 	wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
+	if(vpc->mesa_program.IsPositionInvariant) {
+		/* we wan't position don't we ? */
+		wanted_key.InputsRead |= (1 << VERT_ATTRIB_POS);
+	}
 
 	for (vp = vpc->progs; vp; vp = vp->next)
 		if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key)) == 0) {
@@ -1170,6 +1172,5 @@ void r300_select_vertex_shader(r300ContextPtr r300)
 	vp = build_program(&wanted_key, &vpc->mesa_program, wpos_idx);
 	vp->next = vpc->progs;
 	vpc->progs = vp;
-
 	r300->selected_vp = vp;
 }
diff --git a/src/mesa/drivers/dri/r300/radeon_state.c b/src/mesa/drivers/dri/r300/radeon_state.c
index 0ceaf7ebb7d..ddadf83a003 100644
--- a/src/mesa/drivers/dri/r300/radeon_state.c
+++ b/src/mesa/drivers/dri/r300/radeon_state.c
@@ -41,7 +41,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "light.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
diff --git a/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c b/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
index 72c03c53ad9..0625e5bc571 100644
--- a/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
+++ b/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
@@ -46,6 +46,8 @@
 #include "state.h"
 #include "image.h"
 
+#include "vbo/vbo_context.h"
+
 #define CONV_VB(a, b) rvb->AttribPtr[(a)].size = vb->b->size, \
 			rvb->AttribPtr[(a)].type = GL_FLOAT, \
 			rvb->AttribPtr[(a)].stride = vb->b->stride, \
@@ -129,15 +131,7 @@ static int setup_arrays(r300ContextPtr rmesa, GLint start)
 			CONV(i, VertexAttrib[i]);
 	
 	for (i=0; i < VERT_ATTRIB_MAX; i++) {
-		if (enabled & (1 << i)) {
-			rmesa->state.VB.AttribPtr[i].data += rmesa->state.VB.AttribPtr[i].stride * start;
-		} else {
-			def.data = ctx->Current.Attrib[i];
-			memcpy(&rmesa->state.VB.AttribPtr[i], &def, sizeof(struct dt));
-		}
-		
-		/*if(rmesa->state.VB.AttribPtr[i].data == ctx->Current.Attrib[i])
-			fprintf(stderr, "%d is default coord\n", i);*/
+	   rmesa->state.VB.AttribPtr[i].data += rmesa->state.VB.AttribPtr[i].stride * start;
 	}
 	
 	for(i=0; i < VERT_ATTRIB_MAX; i++){
@@ -177,177 +171,18 @@ static int setup_arrays(r300ContextPtr rmesa, GLint start)
 
 void radeon_init_vtxfmt_a(r300ContextPtr rmesa);
 
-static void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *c_indices )
-{
-	GET_CURRENT_CONTEXT(ctx);
-	r300ContextPtr rmesa = R300_CONTEXT(ctx);
-	int elt_size;
-	int i;
-	unsigned int min = ~0, max = 0;
-	struct tnl_prim prim;
-	static void *ptr = NULL;
-	struct r300_dma_region rvb;
-	const GLvoid *indices = c_indices;
-	
-	if (count > 65535) {
-		WARN_ONCE("Too many verts!\n");
-		goto fallback;
-	}
-	
-	if (ctx->Array.ElementArrayBufferObj->Name) {
-		/* use indices in the buffer object */
-		if (!ctx->Array.ElementArrayBufferObj->Data) {
-			_mesa_warning(ctx, "DrawRangeElements with empty vertex elements buffer!");
-			return;
-		}
-		/* actual address is the sum of pointers */
-		indices = (GLvoid *)
-		ADD_POINTERS(ctx->Array.ElementArrayBufferObj->Data, (const GLubyte *) c_indices);
-	}
-	
-	if (!_mesa_validate_DrawElements( ctx, mode, count, type, indices ))
-		return;
-	
-	FLUSH_CURRENT( ctx, 0 );
-	
-	memset(&rvb, 0, sizeof(rvb));
-	switch (type) {
-	case GL_UNSIGNED_BYTE:
-		for (i=0; i < count; i++) {
-			if(((unsigned char *)indices)[i] < min)
-				min = ((unsigned char *)indices)[i];
-			if(((unsigned char *)indices)[i] > max)
-				max = ((unsigned char *)indices)[i];
-		}
-		
-#ifdef FORCE_32BITS_ELTS
-		elt_size = 4;
-#else
-		elt_size = 2;
-#endif		
-		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
-		rvb.aos_offset = GET_START(&rvb);
-		ptr = rvb.address + rvb.start;
-			
-#ifdef FORCE_32BITS_ELTS
-		for (i=0; i < count; i++)
-			((unsigned int *)ptr)[i] = ((unsigned char *)indices)[i] - min;
-#else
-		for (i=0; i < count; i++)
-			((unsigned short int *)ptr)[i] = ((unsigned char *)indices)[i] - min;
-#endif
-	break;
-		
-	case GL_UNSIGNED_SHORT:
-		for (i=0; i < count; i++) {
-			if(((unsigned short int *)indices)[i] < min)
-				min = ((unsigned short int *)indices)[i];
-			if(((unsigned short int *)indices)[i] > max)
-				max = ((unsigned short int *)indices)[i];
-		}
-		
-#ifdef FORCE_32BITS_ELTS
-		elt_size = 4;
-#else
-		elt_size = 2;
-#endif
-		
-		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
-		rvb.aos_offset = GET_START(&rvb);
-		ptr = rvb.address + rvb.start;
-		
-#ifdef FORCE_32BITS_ELTS
-		for (i=0; i < count; i++)
-			((unsigned int *)ptr)[i] = ((unsigned short int *)indices)[i] - min;
-#else
-		for (i=0; i < count; i++)
-			((unsigned short int *)ptr)[i] = ((unsigned short int *)indices)[i] - min;
-#endif
-	break;
-	
-	case GL_UNSIGNED_INT:
-		for (i=0; i < count; i++) {
-			if(((unsigned int *)indices)[i] < min)
-				min = ((unsigned int *)indices)[i];
-			if(((unsigned int *)indices)[i] > max)
-				max = ((unsigned int *)indices)[i];
-		}
-		
-#ifdef FORCE_32BITS_ELTS
-		elt_size = 4;
-#else
-		if (max - min <= 65535)
-			elt_size = 2;
-		else 
-			elt_size = 4;
-#endif
-		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
-		rvb.aos_offset = GET_START(&rvb);
-		ptr = rvb.address + rvb.start;
-		
-		
-		if (elt_size == 2)
-			for (i=0; i < count; i++)
-				((unsigned short int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
-		else
-			for (i=0; i < count; i++)
-				((unsigned int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
-	break;
-	
-	default:
-		WARN_ONCE("Unknown elt type!\n");
-	goto fallback;
-	}
-	
-	if (ctx->NewState) 
-		_mesa_update_state( ctx );
-	
-	r300UpdateShaders(rmesa);
-	
-	if (setup_arrays(rmesa, min) >= R300_FALLBACK_TCL) {
-		r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-		goto fallback;
-	}
-	
-	rmesa->state.VB.Count = max - min + 1;
-	
-	r300UpdateShaderStates(rmesa);
-	
-	rmesa->state.VB.Primitive = &prim;
-	rmesa->state.VB.PrimitiveCount = 1;
-	
-	prim.mode = mode | PRIM_BEGIN | PRIM_END;
-	if (rmesa->state.VB.LockCount)
-		prim.start = min - rmesa->state.VB.LockFirst;
-	else
-		prim.start = 0;
-	prim.count = count;
-	
-	rmesa->state.VB.Elts = ptr;
-	rmesa->state.VB.elt_size = elt_size;
-	
-	if (r300_run_vb_render(ctx, NULL)) {
-		r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-		goto fallback;
-	}
-	
-	if(rvb.buf)
-		radeon_mm_use(rmesa, rvb.buf->id);
-	
-	r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-	return;
-	
-	fallback:
-	_tnl_array_init(ctx);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
-	CALL_DrawElements(GET_DISPATCH(), (mode, count, type, c_indices));
-	radeon_init_vtxfmt_a(rmesa);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
-}
 
-static void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count, GLenum type, const GLvoid *c_indices)
+static void radeonDrawRangeElements(GLcontext *ctx,
+				    GLenum mode,
+				    GLuint min,
+				    GLuint max,
+				    GLsizei count, 
+				    GLenum type,
+				    const GLvoid *c_indices)
 {
-	GET_CURRENT_CONTEXT(ctx);
+#if 1
+	return GL_FALSE;
+#else
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct tnl_prim prim;
 	int elt_size;
@@ -371,26 +206,23 @@ static void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei
 				indices += i * _mesa_sizeof_type(type);
 				count -= i;
 			}
-			return ;
+			return GL_TRUE;
 		}
 		WARN_ONCE("Too many verts!\n");
-		goto fallback;
+		return GL_FALSE;
 	}
 	
 	if (ctx->Array.ElementArrayBufferObj->Name) {
 		/* use indices in the buffer object */
 		if (!ctx->Array.ElementArrayBufferObj->Data) {
 			_mesa_warning(ctx, "DrawRangeElements with empty vertex elements buffer!");
-			return;
+			return GL_TRUE;
 		}
 		/* actual address is the sum of pointers */
 		indices = (GLvoid *)
 		ADD_POINTERS(ctx->Array.ElementArrayBufferObj->Data, (const GLubyte *) c_indices);
 	}
 	
-	if (!_mesa_validate_DrawRangeElements( ctx, mode, min, max, count, type, indices ))
-		return;
-	
 	FLUSH_CURRENT( ctx, 0 );
 #ifdef OPTIMIZE_ELTS
 	min = 0;
@@ -465,7 +297,7 @@ static void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei
 	
 	default:
 		WARN_ONCE("Unknown elt type!\n");
-	goto fallback;
+		return GL_FALSE;
 	}
 	
 	/* XXX: setup_arrays before state update? */
@@ -477,7 +309,7 @@ static void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei
 
 	if (setup_arrays(rmesa, min) >= R300_FALLBACK_TCL) {
 		r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-		goto fallback;
+		return GL_FALSE;
 	}
 
 	rmesa->state.VB.Count = max - min + 1;
@@ -501,37 +333,34 @@ static void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei
 	
 	if (r300_run_vb_render(ctx, NULL)) {
 		r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-		goto fallback;
+		return GL_FALSE;
 	}
 	
 	if(rvb.buf)
 		radeon_mm_use(rmesa, rvb.buf->id);
 	
 	r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
-	return ;
-	
-	fallback:
-	_tnl_array_init(ctx);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
-	CALL_DrawRangeElements(GET_DISPATCH(), (mode, min, max, count, type, c_indices));
-	radeon_init_vtxfmt_a(rmesa);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
+	return GL_TRUE;
+#endif
 }
 
-static void radeonDrawArrays( GLenum mode, GLint start, GLsizei count )
+static GLboolean radeonDrawArrays( GLcontext *ctx,
+				   GLenum mode, GLint start, GLsizei count )
 {
+#if 1
+	return GL_FALSE;
+#else
 	GET_CURRENT_CONTEXT(ctx);
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct tnl_prim prim;
 	
 	if (count > 65535) {
+	   /* TODO: split into multiple draws.
+	    */
 		WARN_ONCE("Too many verts!\n");
-		goto fallback;
+		return GL_FALSE;
 	}
 	
-	if (!_mesa_validate_DrawArrays( ctx, mode, start, count ))
-		return;
-	
 	FLUSH_CURRENT( ctx, 0 );
 	
 	if (ctx->NewState) 
@@ -542,7 +371,7 @@ static void radeonDrawArrays( GLenum mode, GLint start, GLsizei count )
 	r300UpdateShaders(rmesa);
 
 	if (setup_arrays(rmesa, start) >= R300_FALLBACK_TCL)
-		goto fallback;
+		return GL_FALSE;
 
 	rmesa->state.VB.Count = count;
 
@@ -564,31 +393,70 @@ static void radeonDrawArrays( GLenum mode, GLint start, GLsizei count )
 	rmesa->state.VB.elt_max = 0;
 	
 	if (r300_run_vb_render(ctx, NULL))
-		goto fallback;
+	   return GL_FALSE;
 
-	return ;
-	
-	fallback:
-	_tnl_array_init(ctx);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
-	CALL_DrawArrays(GET_DISPATCH(), (mode, start, count));
-	radeon_init_vtxfmt_a(rmesa);
-	_mesa_install_exec_vtxfmt( ctx, &TNL_CONTEXT(ctx)->exec_vtxfmt );
+	return GL_TRUE;
+#endif
 }
 
+static void radeon_draw_prims( GLcontext *ctx,
+			       const struct gl_client_array *arrays[],
+			       const struct _mesa_prim *prim,
+			       GLuint nr_prims,
+			       const struct _mesa_index_buffer *ib,
+			       GLuint min_index,
+			       GLuint max_index)
+{
+	if (ib == NULL) {
+		for (i = 0; i < nr_prims; i++) {
+			if (!radeonDrawArrays(ctx,
+					      prim->mode,
+					      prim->start,
+					      prim->count)) {
+				/* Fallback
+				 */
+				_tnl_draw_prims(ctx,
+						arrays,
+						prim + i,
+						nr_prims - i,
+						ib, 
+						min_index,
+						max_index);
+				return;
+			}
+		}
+	} else {
+		for (i = 0; i < nr_prims; i++) {
+			if (!radeonDrawRangeElements(ctx,
+						     prim->mode,
+						     min_index,
+						     max_index,
+						     prim->count,
+						     ib->types,
+						     ib->ptr)) {
+				/* Fallback
+				 */
+				_tnl_draw_prims(ctx,
+						arrays,
+						prim + i,
+						nr_prims - i,
+						ib, 
+						min_index,
+						max_index);
+				return;
+			}
+		}
+	}
+}
+	   
 void radeon_init_vtxfmt_a(r300ContextPtr rmesa)
 {
 	GLcontext *ctx;
-	GLvertexformat *vfmt;
-	
-	ctx = rmesa->radeon.glCtx; 
-	vfmt = (GLvertexformat *)ctx->TnlModule.Current;
-   
-	vfmt->DrawElements = radeonDrawElements;
-	vfmt->DrawArrays = radeonDrawArrays;
-	vfmt->DrawRangeElements = radeonDrawRangeElements;
+	struct vbo_context *vbo = vbo_context(ctx);
 	
+	vbo->draw_prims = radeon_draw_prims;
 }
+
 #endif
 
 #ifdef HW_VBOS
diff --git a/src/mesa/drivers/dri/radeon/Makefile b/src/mesa/drivers/dri/radeon/Makefile
index 6e46620b397..f223b2d9228 100644
--- a/src/mesa/drivers/dri/radeon/Makefile
+++ b/src/mesa/drivers/dri/radeon/Makefile
@@ -22,11 +22,7 @@ DRIVER_SOURCES = \
 	radeon_swtcl.c \
 	radeon_span.c \
 	radeon_maos.c \
-	radeon_sanity.c \
-	radeon_vtxfmt.c \
-	radeon_vtxfmt_c.c \
-	radeon_vtxfmt_sse.c \
-	radeon_vtxfmt_x86.c 
+	radeon_sanity.c 
 
 C_SOURCES = \
 	$(COMMON_SOURCES) \
@@ -34,8 +30,7 @@ C_SOURCES = \
 
 DRIVER_DEFINES = -DRADEON_COMMON=0
 
-X86_SOURCES = \
-	radeon_vtxtmp_x86.S 
+X86_SOURCES = 
 
 include ../Makefile.template
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index 8b92e3fe65f..6bc2c4aa5c5 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -46,7 +46,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -60,7 +60,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_tex.h"
 #include "radeon_swtcl.h"
 #include "radeon_tcl.h"
-#include "radeon_vtxfmt.h"
 #include "radeon_maos.h"
 
 #define need_GL_ARB_multisample
@@ -361,7 +360,7 @@ radeonCreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
    _ae_create_context( ctx );
@@ -370,13 +369,10 @@ radeonCreateContext( const __GLcontextModes *glVisual,
     */
    _tnl_destroy_pipeline( ctx );
    _tnl_install_pipeline( ctx, radeon_pipeline );
-   ctx->Driver.FlushVertices = radeonFlushVertices;
 
    /* Try and keep materials and vertices separate:
     */
-   _tnl_isolate_materials( ctx, GL_TRUE );
-
-/*     _mesa_allow_light_in_model( ctx, GL_FALSE ); */
+/*    _tnl_isolate_materials( ctx, GL_TRUE ); */
 
    /* Configure swrast and T&L to match hardware characteristics:
     */
@@ -450,10 +446,7 @@ radeonCreateContext( const __GLcontextModes *glVisual,
    }
 
    if (rmesa->radeonScreen->chip_flags & RADEON_CHIPSET_TCL) {
-      if (tcl_mode >= DRI_CONF_TCL_VTXFMT)
-	 radeonVtxfmtInit( ctx, tcl_mode >= DRI_CONF_TCL_CODEGEN );
-
-      _tnl_need_dlist_norm_lengths( ctx, GL_FALSE );
+/*       _tnl_need_dlist_norm_lengths( ctx, GL_FALSE ); */
    }
    return GL_TRUE;
 }
@@ -484,7 +477,7 @@ void radeonDestroyContext( __DRIcontextPrivate *driContextPriv )
       release_texture_heaps = (rmesa->glCtx->Shared->RefCount == 1);
       _swsetup_DestroyContext( rmesa->glCtx );
       _tnl_DestroyContext( rmesa->glCtx );
-      _ac_DestroyContext( rmesa->glCtx );
+      _vbo_DestroyContext( rmesa->glCtx );
       _swrast_DestroyContext( rmesa->glCtx );
 
       radeonDestroySwtcl( rmesa->glCtx );
@@ -494,12 +487,6 @@ void radeonDestroyContext( __DRIcontextPrivate *driContextPriv )
 	 radeonFlushCmdBuf( rmesa, __FUNCTION__ );
       }
 
-      if (!(rmesa->TclFallback & RADEON_TCL_FALLBACK_TCL_DISABLE)) {
-	 int tcl_mode = driQueryOptioni(&rmesa->optionCache, "tcl_mode");
-	 if (tcl_mode >= DRI_CONF_TCL_VTXFMT)
-	    radeonVtxfmtDestroy( rmesa->glCtx );
-      }
-
       _mesa_vector4f_free( &rmesa->tcl.ObjClean );
 
       if (rmesa->state.scissor.pClipRects) {
@@ -621,9 +608,6 @@ radeonMakeCurrent( __DRIcontextPrivate *driContextPriv,
 			  (GLframebuffer *) driDrawPriv->driverPrivate,
 			  (GLframebuffer *) driReadPriv->driverPrivate );
 
-      if (newCtx->vb.enabled)
-	 radeonVtxfmtMakeCurrent( newCtx->glCtx );
-
    } else {
       if (RADEON_DEBUG & DEBUG_DRI)
 	 fprintf(stderr, "%s ctx is null\n", __FUNCTION__);
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.h b/src/mesa/drivers/dri/radeon/radeon_context.h
index 0a7c3b2f544..02cea2f4e37 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -529,7 +529,6 @@ struct radeon_store {
  */
 struct radeon_tcl_info {
    GLuint vertex_format;
-   GLint last_offset;
    GLuint hw_primitive;
 
    /* Temporary for cases where incoming vertex data is incompatible
@@ -600,75 +599,6 @@ struct radeon_ioctl {
 #define RADEON_MAX_PRIMS 64
 
 
-/* Want to keep a cache of these around.  Each is parameterized by
- * only a single value which has only a small range.  Only expect a
- * few, so just rescan the list each time?
- */
-struct dynfn {
-   struct dynfn *next, *prev;
-   int key;
-   char *code;
-};
-
-struct dfn_lists {
-   struct dynfn Vertex2f;
-   struct dynfn Vertex2fv;
-   struct dynfn Vertex3f;
-   struct dynfn Vertex3fv;
-   struct dynfn Color4ub;
-   struct dynfn Color4ubv;
-   struct dynfn Color3ub;
-   struct dynfn Color3ubv;
-   struct dynfn Color4f;
-   struct dynfn Color4fv;
-   struct dynfn Color3f;
-   struct dynfn Color3fv;
-   struct dynfn SecondaryColor3ubEXT;
-   struct dynfn SecondaryColor3ubvEXT;
-   struct dynfn SecondaryColor3fEXT;
-   struct dynfn SecondaryColor3fvEXT;
-   struct dynfn Normal3f;
-   struct dynfn Normal3fv;
-   struct dynfn TexCoord2f;
-   struct dynfn TexCoord2fv;
-   struct dynfn TexCoord1f;
-   struct dynfn TexCoord1fv;
-   struct dynfn MultiTexCoord2fARB;
-   struct dynfn MultiTexCoord2fvARB;
-   struct dynfn MultiTexCoord1fARB;
-   struct dynfn MultiTexCoord1fvARB;
-};
-
-struct dfn_generators {
-   struct dynfn *(*Vertex2f)( GLcontext *, int );
-   struct dynfn *(*Vertex2fv)( GLcontext *, int );
-   struct dynfn *(*Vertex3f)( GLcontext *, int );
-   struct dynfn *(*Vertex3fv)( GLcontext *, int );
-   struct dynfn *(*Color4ub)( GLcontext *, int );
-   struct dynfn *(*Color4ubv)( GLcontext *, int );
-   struct dynfn *(*Color3ub)( GLcontext *, int );
-   struct dynfn *(*Color3ubv)( GLcontext *, int );
-   struct dynfn *(*Color4f)( GLcontext *, int );
-   struct dynfn *(*Color4fv)( GLcontext *, int );
-   struct dynfn *(*Color3f)( GLcontext *, int );
-   struct dynfn *(*Color3fv)( GLcontext *, int );
-   struct dynfn *(*SecondaryColor3ubEXT)( GLcontext *, int );
-   struct dynfn *(*SecondaryColor3ubvEXT)( GLcontext *, int );
-   struct dynfn *(*SecondaryColor3fEXT)( GLcontext *, int );
-   struct dynfn *(*SecondaryColor3fvEXT)( GLcontext *, int );
-   struct dynfn *(*Normal3f)( GLcontext *, int );
-   struct dynfn *(*Normal3fv)( GLcontext *, int );
-   struct dynfn *(*TexCoord2f)( GLcontext *, int );
-   struct dynfn *(*TexCoord2fv)( GLcontext *, int );
-   struct dynfn *(*TexCoord1f)( GLcontext *, int );
-   struct dynfn *(*TexCoord1fv)( GLcontext *, int );
-   struct dynfn *(*MultiTexCoord2fARB)( GLcontext *, int );
-   struct dynfn *(*MultiTexCoord2fvARB)( GLcontext *, int );
-   struct dynfn *(*MultiTexCoord1fARB)( GLcontext *, int );
-   struct dynfn *(*MultiTexCoord1fvARB)( GLcontext *, int );
-};
-
-
 
 struct radeon_prim {
    GLuint start;
@@ -685,43 +615,6 @@ struct radeon_prim {
  */
 #define RADEON_MAX_VERTEX_SIZE 20
 
-struct radeon_vbinfo {
-   GLint counter, initial_counter;
-   GLint *dmaptr;
-   void (*notify)( void );
-   GLint vertex_size;
-
-   union { float f; int i; radeon_color_t color; } vertex[RADEON_MAX_VERTEX_SIZE];
-
-   GLfloat *normalptr;
-   GLfloat *floatcolorptr;
-   radeon_color_t *colorptr;
-   GLfloat *floatspecptr;
-   radeon_color_t *specptr;
-   GLfloat *texcoordptr[4];	/* 3 (TMU) + 1 for radeon_vtxfmt_c.c when GL_TEXTURE3 */
-
-   GLenum *prim;		/* &ctx->Driver.CurrentExecPrimitive */
-   GLuint primflags;
-   GLboolean enabled;		/* *_NO_VTXFMT / *_NO_TCL env vars */
-   GLboolean installed;
-   GLboolean fell_back;
-   GLboolean recheck;
-   GLint nrverts;
-   GLuint vertex_format;
-
-   GLuint installed_vertex_format;
-   GLuint installed_color_3f_sz;
-
-   struct radeon_prim primlist[RADEON_MAX_PRIMS];
-   int nrprims;
-
-   struct dfn_lists dfn_cache;
-   struct dfn_generators codegen;
-   GLvertexformat vtxfmt;
-};
-
-
-
 
 struct radeon_context {
    GLcontext *glCtx;			/* Mesa context */
@@ -808,10 +701,6 @@ struct radeon_context {
     */
    struct radeon_swtcl_info swtcl;
 
-   /* radeon_vtxfmt.c
-    */
-   struct radeon_vbinfo vb;
-
    /* Mirrors of some DRI state
     */
    struct radeon_dri_mirror dri;
diff --git a/src/mesa/drivers/dri/radeon/radeon_ioctl.c b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
index 737490fcde0..4c64bc201ae 100644
--- a/src/mesa/drivers/dri/radeon/radeon_ioctl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_ioctl.c
@@ -1087,10 +1087,6 @@ static void radeonClear( GLcontext *ctx, GLbitfield mask )
       }
    }
 
-   /* Flip top to bottom */
-   cx += dPriv->x;
-   cy  = dPriv->y + dPriv->h - cy - ch;
-
    LOCK_HARDWARE( rmesa );
 
    /* compute region after locking: */
@@ -1099,6 +1095,10 @@ static void radeonClear( GLcontext *ctx, GLbitfield mask )
    cw = ctx->DrawBuffer->_Xmax - cx;
    ch = ctx->DrawBuffer->_Ymax - cy;
 
+   /* Flip top to bottom */
+   cx += dPriv->x;
+   cy  = dPriv->y + dPriv->h - cy - ch;
+
    /* Throttle the number of clear ioctls we do.
     */
    while ( 1 ) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index 91a60bb9f15..65dbecf7a6c 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -37,7 +37,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "imports.h"
 #include "mtypes.h"
 
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "math/m_translate.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 279357ab590..fc5aa114621 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -55,6 +55,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "r200_span.h"
 #elif RADEON_COMMON && defined(RADEON_COMMON_FOR_R300)
 #include "r300_context.h"
+#include "r300_fragprog.h"
 #include "radeon_span.h"
 #endif
 
@@ -163,6 +164,18 @@ DRI_CONF_OPT_BEGIN(disable_lowimpact_fallback,bool,def) \
         DRI_CONF_DESC(en,"Disable Low-impact fallback") \
 DRI_CONF_OPT_END
 
+#define DRI_CONF_DISABLE_DOUBLE_SIDE_STENCIL(def) \
+DRI_CONF_OPT_BEGIN(disable_stencil_two_side,bool,def) \
+        DRI_CONF_DESC(en,"Disable GL_EXT_stencil_two_side") \
+DRI_CONF_OPT_END
+
+#define DRI_CONF_FP_OPTIMIZATION(def) \
+DRI_CONF_OPT_BEGIN_V(fp_optimization,enum,def,"0:1") \
+	DRI_CONF_DESC_BEGIN(en,"Fragment Program optimization") \
+                DRI_CONF_ENUM(0,"Optimize for Speed") \
+                DRI_CONF_ENUM(1,"Optimize for Quality") \
+        DRI_CONF_DESC_END \
+DRI_CONF_OPT_END
 
 const char __driConfigOptions[] =
 DRI_CONF_BEGIN
@@ -174,6 +187,7 @@ DRI_CONF_BEGIN
 		DRI_CONF_MAX_TEXTURE_COORD_UNITS(8, 2, 8)
 		DRI_CONF_COMMAND_BUFFER_SIZE(8, 8, 32)
 		DRI_CONF_DISABLE_FALLBACK(false)
+		DRI_CONF_DISABLE_DOUBLE_SIDE_STENCIL(false)
 	DRI_CONF_SECTION_END
 	DRI_CONF_SECTION_QUALITY
 		DRI_CONF_TEXTURE_DEPTH(DRI_CONF_TEXTURE_DEPTH_FB)
@@ -184,12 +198,13 @@ DRI_CONF_BEGIN
 		DRI_CONF_COLOR_REDUCTION(DRI_CONF_COLOR_REDUCTION_DITHER)
 		DRI_CONF_ROUND_MODE(DRI_CONF_ROUND_TRUNC)
 		DRI_CONF_DITHER_MODE(DRI_CONF_DITHER_XERRORDIFF)
+		DRI_CONF_FP_OPTIMIZATION(DRI_CONF_FP_OPTIMIZATION_SPEED)
 	DRI_CONF_SECTION_END
 	DRI_CONF_SECTION_DEBUG
 		DRI_CONF_NO_RAST(false)
 	DRI_CONF_SECTION_END
 DRI_CONF_END;
-static const GLuint __driNConfigOptions = 16;
+static const GLuint __driNConfigOptions = 18;
 
 #ifndef RADEON_DEBUG
 int RADEON_DEBUG = 0;
diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c
index 216d7dca174..e19202fa443 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state.c
@@ -42,7 +42,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "context.h"
 #include "framebuffer.h"
 
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
@@ -53,7 +53,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_tcl.h"
 #include "radeon_tex.h"
 #include "radeon_swtcl.h"
-#include "radeon_vtxfmt.h"
 #include "drirenderbuffer.h"
 
 static void radeonUpdateSpecular( GLcontext *ctx );
@@ -2303,11 +2302,10 @@ static void radeonInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    _ae_invalidate_state( ctx, new_state );
    RADEON_CONTEXT(ctx)->NewGLState |= new_state;
-   radeonVtxfmtInvalidate( ctx );
 }
 
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_state_init.c b/src/mesa/drivers/dri/radeon/radeon_state_init.c
index 5682d95ae10..5fc34f09331 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -33,7 +33,7 @@
 #include "api_arrayelt.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
@@ -44,7 +44,6 @@
 #include "radeon_tcl.h"
 #include "radeon_tex.h"
 #include "radeon_swtcl.h"
-#include "radeon_vtxfmt.h"
 
 #include "xmlpool.h"
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index e36a710d33a..7ce1fa67cf6 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -45,7 +45,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
-#include "tnl/t_vtx_api.h"	/* for _tnl_FlushVertices */
 
 #include "radeon_context.h"
 #include "radeon_ioctl.h"
@@ -848,14 +847,6 @@ void radeonFallback( GLcontext *ctx, GLuint bit, GLboolean mode )
 }
 
 
-void radeonFlushVertices( GLcontext *ctx, GLuint flags )
-{
-   _tnl_FlushVertices( ctx, flags );
-
-   if (flags & FLUSH_STORED_VERTICES)
-      RADEON_NEWPRIM( RADEON_CONTEXT( ctx ) );
-}
-
 /**********************************************************************/
 /*                            Initialization.                         */
 /**********************************************************************/
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.h b/src/mesa/drivers/dri/radeon/radeon_swtcl.h
index f95a52c0b8e..64f9019513d 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.h
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.h
@@ -43,7 +43,6 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 extern void radeonInitSwtcl( GLcontext *ctx );
 extern void radeonDestroySwtcl( GLcontext *ctx );
 
-extern void radeonFlushVertices( GLcontext *ctx, GLuint flags );
 extern void radeonChooseRenderState( GLcontext *ctx );
 extern void radeonChooseVertexState( GLcontext *ctx );
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_tcl.c b/src/mesa/drivers/dri/radeon/radeon_tcl.c
index ffd49b525b8..5ad044c2628 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tcl.c
@@ -39,7 +39,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "mtypes.h"
 #include "enums.h"
 
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
deleted file mode 100644
index a5a9eb144b5..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
+++ /dev/null
@@ -1,1086 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt.c,v 1.6 2003/05/06 23:52:08 daenzer Exp $ */
-/**************************************************************************
-
-Copyright 2000, 2001 ATI Technologies Inc., Ontario, Canada, and
-                     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, 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 "glheader.h"
-#include "imports.h"
-#include "api_noop.h"
-#include "api_arrayelt.h"
-#include "context.h"
-#include "mtypes.h"
-#include "enums.h"
-#include "glapi.h"
-#include "colormac.h"
-#include "light.h"
-#include "state.h"
-#include "vtxfmt.h"
-
-#include "tnl/tnl.h"
-#include "tnl/t_context.h"
-#include "tnl/t_array_api.h"
-#include "tnl/t_save_api.h"
-
-#include "radeon_context.h"
-#include "radeon_state.h"
-#include "radeon_ioctl.h"
-#include "radeon_tex.h"
-#include "radeon_tcl.h"
-#include "radeon_swtcl.h"
-#include "radeon_vtxfmt.h"
-
-#define VERT_ATTRIB_TEX(u)	(VERT_ATTRIB_TEX0 + (u))
-
-#include "dispatch.h"
-
-static void radeonVtxfmtFlushVertices( GLcontext *, GLuint );
-
-static void count_func( const char *name,  struct dynfn *l )
-{
-   int i = 0;
-   struct dynfn *f;
-   foreach (f, l) i++;
-   if (i) fprintf(stderr, "%s: %d\n", name, i );
-}
-
-static void count_funcs( radeonContextPtr rmesa )
-{
-   count_func( "Vertex2f", &rmesa->vb.dfn_cache.Vertex2f );
-   count_func( "Vertex2fv", &rmesa->vb.dfn_cache.Vertex2fv );
-   count_func( "Vertex3f", &rmesa->vb.dfn_cache.Vertex3f );
-   count_func( "Vertex3fv", &rmesa->vb.dfn_cache.Vertex3fv );
-   count_func( "Color4ub", &rmesa->vb.dfn_cache.Color4ub );
-   count_func( "Color4ubv", &rmesa->vb.dfn_cache.Color4ubv );
-   count_func( "Color3ub", &rmesa->vb.dfn_cache.Color3ub );
-   count_func( "Color3ubv", &rmesa->vb.dfn_cache.Color3ubv );
-   count_func( "Color4f", &rmesa->vb.dfn_cache.Color4f );
-   count_func( "Color4fv", &rmesa->vb.dfn_cache.Color4fv );
-   count_func( "Color3f", &rmesa->vb.dfn_cache.Color3f );
-   count_func( "Color3fv", &rmesa->vb.dfn_cache.Color3fv );
-   count_func( "SecondaryColor3f", &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   count_func( "SecondaryColor3fv", &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   count_func( "SecondaryColor3ub", &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   count_func( "SecondaryColor3ubv", &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   count_func( "Normal3f", &rmesa->vb.dfn_cache.Normal3f );
-   count_func( "Normal3fv", &rmesa->vb.dfn_cache.Normal3fv );
-   count_func( "TexCoord2f", &rmesa->vb.dfn_cache.TexCoord2f );
-   count_func( "TexCoord2fv", &rmesa->vb.dfn_cache.TexCoord2fv );
-   count_func( "TexCoord1f", &rmesa->vb.dfn_cache.TexCoord1f );
-   count_func( "TexCoord1fv", &rmesa->vb.dfn_cache.TexCoord1fv );
-   count_func( "MultiTexCoord2fARB", &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   count_func( "MultiTexCoord2fvARB", &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   count_func( "MultiTexCoord1fARB", &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   count_func( "MultiTexCoord1fvARB", &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-}
-
-
-void radeon_copy_to_current( GLcontext *ctx ) 
-{
-   GLuint unit;
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   assert(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT);
-
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_N0) {
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][0] = rmesa->vb.normalptr[0];
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][1] = rmesa->vb.normalptr[1];
-      ctx->Current.Attrib[VERT_ATTRIB_NORMAL][2] = rmesa->vb.normalptr[2];
-   }
-
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_PKCOLOR) {
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->red );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->green );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->blue );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT( rmesa->vb.colorptr->alpha );
-   } 
-   
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_FPCOLOR) {
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] = rmesa->vb.floatcolorptr[0];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] = rmesa->vb.floatcolorptr[1];
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] = rmesa->vb.floatcolorptr[2];
-   }
-
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_FPALPHA)
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = rmesa->vb.floatcolorptr[3];
-      
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_PKSPEC) {
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][0] = UBYTE_TO_FLOAT( rmesa->vb.specptr->red );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][1] = UBYTE_TO_FLOAT( rmesa->vb.specptr->green );
-      ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] = UBYTE_TO_FLOAT( rmesa->vb.specptr->blue );
-   } 
-
-   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (rmesa->vb.vertex_format & RADEON_ST_BIT(unit)) {
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][0] = rmesa->vb.texcoordptr[unit][0];
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][1] = rmesa->vb.texcoordptr[unit][1];
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][2] = 0.0F;
-	 ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][3] = 1.0F;
-      }
-   }
-
-   ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-}
-
-static GLboolean discreet_gl_prim[GL_POLYGON+1] = {
-   1,				/* 0 points */
-   1,				/* 1 lines */
-   0,				/* 2 line_strip */
-   0,				/* 3 line_loop */
-   1,				/* 4 tris */
-   0,				/* 5 tri_fan */
-   0,				/* 6 tri_strip */
-   1,				/* 7 quads */
-   0,				/* 8 quadstrip */
-   0,				/* 9 poly */
-};
-
-static void flush_prims( radeonContextPtr rmesa )
-{
-   int i,j;
-   struct radeon_dma_region tmp = rmesa->dma.current;
-   
-   tmp.buf->refcount++;
-   tmp.aos_size = rmesa->vb.vertex_size;
-   tmp.aos_stride = rmesa->vb.vertex_size;
-   tmp.aos_start = GET_START(&tmp);
-
-   rmesa->dma.current.ptr = rmesa->dma.current.start += 
-      (rmesa->vb.initial_counter - rmesa->vb.counter) * rmesa->vb.vertex_size * 4; 
-
-   rmesa->tcl.vertex_format = rmesa->vb.vertex_format;
-   rmesa->tcl.aos_components[0] = &tmp;
-   rmesa->tcl.nr_aos_components = 1;
-   rmesa->dma.flush = NULL;
-
-   /* Optimize the primitive list:
-    */
-   if (rmesa->vb.nrprims > 1) {
-      for (j = 0, i = 1 ; i < rmesa->vb.nrprims; i++) {
-	 int pj = rmesa->vb.primlist[j].prim & 0xf;
-	 int pi = rmesa->vb.primlist[i].prim & 0xf;
-      
-	 if (pj == pi && discreet_gl_prim[pj] &&
-	     rmesa->vb.primlist[i].start == rmesa->vb.primlist[j].end) {
-	    rmesa->vb.primlist[j].end = rmesa->vb.primlist[i].end;
-	 }
-	 else {
-	    j++;
-	    if (j != i) rmesa->vb.primlist[j] = rmesa->vb.primlist[i];
-	 }
-      }
-      rmesa->vb.nrprims = j+1;
-   }
-
-   for (i = 0 ; i < rmesa->vb.nrprims; i++) {
-      if (RADEON_DEBUG & DEBUG_PRIMS)
-	 fprintf(stderr, "vtxfmt prim %d: %s %d..%d\n", i,
-		 _mesa_lookup_enum_by_nr( rmesa->vb.primlist[i].prim & 
-					  PRIM_MODE_MASK ),
-		 rmesa->vb.primlist[i].start,
-		 rmesa->vb.primlist[i].end);
-
-      radeonEmitPrimitive( rmesa->glCtx,
-			   rmesa->vb.primlist[i].start,
-			   rmesa->vb.primlist[i].end,
-			   rmesa->vb.primlist[i].prim );
-   }
-
-   rmesa->vb.nrprims = 0;
-   radeonReleaseDmaRegion( rmesa, &tmp, __FUNCTION__ );
-}
-
-
-static void start_prim( radeonContextPtr rmesa, GLuint mode )
-{
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   rmesa->vb.primlist[rmesa->vb.nrprims].start = rmesa->vb.initial_counter - rmesa->vb.counter;
-   rmesa->vb.primlist[rmesa->vb.nrprims].prim = mode;
-}
-
-static void note_last_prim( radeonContextPtr rmesa, GLuint flags )
-{
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d\n", __FUNCTION__, rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      rmesa->vb.primlist[rmesa->vb.nrprims].prim |= flags;
-      rmesa->vb.primlist[rmesa->vb.nrprims].end = rmesa->vb.initial_counter - rmesa->vb.counter;
-
-      if (++(rmesa->vb.nrprims) == RADEON_MAX_PRIMS)
-	 flush_prims( rmesa );
-   }
-}
-
-
-static void copy_vertex( radeonContextPtr rmesa, GLuint n, GLfloat *dst )
-{
-   GLuint i;
-   GLfloat *src = (GLfloat *)(rmesa->dma.current.address + 
-			      rmesa->dma.current.ptr + 
-			      (rmesa->vb.primlist[rmesa->vb.nrprims].start + n) * 
-			      rmesa->vb.vertex_size * 4);
-
-   if (RADEON_DEBUG & DEBUG_VFMT) 
-      fprintf(stderr, "copy_vertex %d\n", rmesa->vb.primlist[rmesa->vb.nrprims].start + n);
-
-   for (i = 0 ; i < rmesa->vb.vertex_size; i++) {
-      dst[i] = src[i];
-   }
-}
-
-/* NOTE: This actually reads the copied vertices back from uncached
- * memory.  Could also use the counter/notify mechanism to populate
- * tmp on the fly as vertices are generated.  
- */
-static GLuint copy_dma_verts( radeonContextPtr rmesa, GLfloat (*tmp)[RADEON_MAX_VERTEX_SIZE] )
-{
-   GLuint ovf, i;
-   GLuint nr = (rmesa->vb.initial_counter - rmesa->vb.counter) - rmesa->vb.primlist[rmesa->vb.nrprims].start;
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s %d verts\n", __FUNCTION__, nr);
-
-   switch( rmesa->vb.prim[0] )
-   {
-   case GL_POINTS:
-      return 0;
-   case GL_LINES:
-      ovf = nr&1;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_TRIANGLES:
-      ovf = nr%3;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_QUADS:
-      ovf = nr&3;
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_LINE_STRIP:
-      if (nr == 0) 
-	 return 0;
-      copy_vertex( rmesa, nr-1, tmp[0] );
-      return 1;
-   case GL_LINE_LOOP:
-   case GL_TRIANGLE_FAN:
-   case GL_POLYGON:
-      if (nr == 0) 
-	 return 0;
-      else if (nr == 1) {
-	 copy_vertex( rmesa, 0, tmp[0] );
-	 return 1;
-      } else {
-	 copy_vertex( rmesa, 0, tmp[0] );
-	 copy_vertex( rmesa, nr-1, tmp[1] );
-	 return 2;
-      }
-   case GL_TRIANGLE_STRIP:
-      ovf = MIN2(nr, 2);
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   case GL_QUAD_STRIP:
-      switch (nr) {
-      case 0: ovf = 0; break;
-      case 1: ovf = 1; break;
-      default: ovf = 2 + (nr&1); break;
-      }
-      for (i = 0 ; i < ovf ; i++)
-	 copy_vertex( rmesa, nr-ovf+i, tmp[i] );
-      return i;
-   default:
-      assert(0);
-      return 0;
-   }
-}
-
-static void VFMT_FALLBACK_OUTSIDE_BEGIN_END( const char *caller )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
-      fprintf(stderr, "%s from %s\n", __FUNCTION__, caller);
-
-   if (ctx->Driver.NeedFlush) 
-      radeonVtxfmtFlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   if (ctx->NewState)
-      _mesa_update_state( ctx ); /* clear state so fell_back sticks */
-
-   _tnl_wakeup_exec( ctx );
-   ctx->Driver.FlushVertices = radeonFlushVertices;
-   ctx->Driver.NewList =_tnl_NewList;
-
-   assert( rmesa->dma.flush == 0 );
-   rmesa->vb.fell_back = GL_TRUE;
-   rmesa->vb.installed = GL_FALSE;
-}
-
-
-static void VFMT_FALLBACK( const char *caller )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat tmp[3][RADEON_MAX_VERTEX_SIZE];
-   GLuint i, prim;
-   GLuint ind = rmesa->vb.vertex_format;
-   GLuint nrverts;
-   GLfloat alpha = 1.0;
-   GLuint unit;
-
-   if (RADEON_DEBUG & (DEBUG_FALLBACKS|DEBUG_VFMT))
-      fprintf(stderr, "%s from %s\n", __FUNCTION__, caller);
-
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) {
-      VFMT_FALLBACK_OUTSIDE_BEGIN_END( __FUNCTION__ );
-      return;
-   }
-
-   /* Copy vertices out of dma:
-    */
-   nrverts = copy_dma_verts( rmesa, tmp );
-
-   /* Finish the prim at this point:
-    */
-   note_last_prim( rmesa, 0 );
-   flush_prims( rmesa );
-
-   /* Update ctx->Driver.CurrentExecPrimitive and swap in swtnl. 
-    */
-   prim = rmesa->vb.prim[0];
-   ctx->Driver.CurrentExecPrimitive = GL_POLYGON+1;
-   _tnl_wakeup_exec( ctx );
-   ctx->Driver.FlushVertices = radeonFlushVertices;
-
-   assert(rmesa->dma.flush == 0);
-   rmesa->vb.fell_back = GL_TRUE;
-   rmesa->vb.installed = GL_FALSE;
-   CALL_Begin(GET_DISPATCH(), (prim));
-   
-   if (rmesa->vb.installed_color_3f_sz == 4)
-      alpha = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3];
-
-   /* Replay saved vertices
-    */
-   for (i = 0 ; i < nrverts; i++) {
-      GLuint offset = 3;
-      if (ind & RADEON_CP_VC_FRMT_N0) {
-	 CALL_Normal3fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset += 3;
-      }
-
-      if (ind & RADEON_CP_VC_FRMT_PKCOLOR) {
-	 radeon_color_t *col = (radeon_color_t *)&tmp[i][offset];
-	 CALL_Color4ub(GET_DISPATCH(), (col->red, col->green, col->blue, col->alpha));
-	 offset++;
-      }
-      else if (ind & RADEON_CP_VC_FRMT_FPALPHA) {
-	 CALL_Color4fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset+=4;
-      } 
-      else if (ind & RADEON_CP_VC_FRMT_FPCOLOR) {
-	 CALL_Color3fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset+=3;
-      }
-
-      if (ind & RADEON_CP_VC_FRMT_PKSPEC) {
-	 radeon_color_t *spec = (radeon_color_t *)&tmp[i][offset];
-	 CALL_SecondaryColor3ubEXT(GET_DISPATCH(), (spec->red, spec->green, spec->blue));
-	 offset++;
-      }
-
-      for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-	 if (ind & RADEON_ST_BIT(unit)) {
-	    CALL_MultiTexCoord2fvARB(GET_DISPATCH(), ((GL_TEXTURE0 + unit), &tmp[i][offset]));
-	    offset += 2;
-	 }
-      }
-      CALL_Vertex3fv(GET_DISPATCH(), (&tmp[i][0]));
-   }
-
-   /* Replay current vertex
-    */
-   if (ind & RADEON_CP_VC_FRMT_N0) 
-      CALL_Normal3fv(GET_DISPATCH(), (rmesa->vb.normalptr));
-
-   if (ind & RADEON_CP_VC_FRMT_PKCOLOR)
-      CALL_Color4ub(GET_DISPATCH(), (rmesa->vb.colorptr->red, rmesa->vb.colorptr->green,
-      				     rmesa->vb.colorptr->blue, rmesa->vb.colorptr->alpha));
-   else if (ind & RADEON_CP_VC_FRMT_FPALPHA)
-      CALL_Color4fv(GET_DISPATCH(), (rmesa->vb.floatcolorptr));
-   else if (ind & RADEON_CP_VC_FRMT_FPCOLOR) {
-      if (rmesa->vb.installed_color_3f_sz == 4 && alpha != 1.0)
-	 CALL_Color4f(GET_DISPATCH(), (rmesa->vb.floatcolorptr[0],
-				       rmesa->vb.floatcolorptr[1],
-				       rmesa->vb.floatcolorptr[2],
-				       alpha));
-      else
-	 CALL_Color3fv(GET_DISPATCH(), (rmesa->vb.floatcolorptr));
-   }
-
-   if (ind & RADEON_CP_VC_FRMT_PKSPEC) 
-       CALL_SecondaryColor3ubEXT(GET_DISPATCH(), (rmesa->vb.specptr->red,
-       						  rmesa->vb.specptr->green,
-						  rmesa->vb.specptr->blue));
-
-   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (ind & RADEON_ST_BIT(unit)) {
-	 CALL_MultiTexCoord2fvARB(GET_DISPATCH(), ((GL_TEXTURE0 + unit),
-	 			  rmesa->vb.texcoordptr[unit]));
-      }
-   }
-}
-
-static void radeonNewList( GLcontext *ctx, GLuint list, GLenum mode )
-{
-   VFMT_FALLBACK( __FUNCTION__ );
-   _tnl_NewList( ctx, list, mode );
-   return;
-}
-
-
-static void wrap_buffer( void )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat tmp[3][RADEON_MAX_VERTEX_SIZE];
-   GLuint i, nrverts;
-
-   if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_PRIMS))
-      fprintf(stderr, "%s %d\n", __FUNCTION__, rmesa->vb.initial_counter - rmesa->vb.counter);
-
-   /* Don't deal with parity.
-    */
-   if ((((rmesa->vb.initial_counter - rmesa->vb.counter) -  
-	 rmesa->vb.primlist[rmesa->vb.nrprims].start) & 1)) {
-      rmesa->vb.counter++;
-      rmesa->vb.initial_counter++;
-      return;
-   }
-
-   /* Copy vertices out of dma:
-    */
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) 
-      nrverts = 0;
-   else {
-      nrverts = copy_dma_verts( rmesa, tmp );
-
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%d vertices to copy\n", nrverts);
-   
-      /* Finish the prim at this point:
-       */
-      note_last_prim( rmesa, 0 );
-   }
-
-   /* Fire any buffered primitives
-    */
-   flush_prims( rmesa );
-
-   /* Get new buffer
-    */
-   radeonRefillCurrentDmaRegion( rmesa );
-
-   /* Reset counter, dmaptr
-    */
-   rmesa->vb.dmaptr = (int *)(rmesa->dma.current.ptr + rmesa->dma.current.address);
-   rmesa->vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / 
-      (rmesa->vb.vertex_size * 4);
-   rmesa->vb.counter--;
-   rmesa->vb.initial_counter = rmesa->vb.counter;
-   rmesa->vb.notify = wrap_buffer;
-
-   rmesa->dma.flush = flush_prims;
-
-   /* Restart wrapped primitive:
-    */
-   if (rmesa->vb.prim[0] != GL_POLYGON+1)
-      start_prim( rmesa, rmesa->vb.prim[0] );
-
-   /* Reemit saved vertices
-    */
-   for (i = 0 ; i < nrverts; i++) {
-      if (RADEON_DEBUG & DEBUG_VERTS) {
-	 int j;
-	 fprintf(stderr, "re-emit vertex %d to %p\n", i, (void *)rmesa->vb.dmaptr);
-	 if (RADEON_DEBUG & DEBUG_VERBOSE)
-	    for (j = 0 ; j < rmesa->vb.vertex_size; j++) 
-	       fprintf(stderr, "\t%08x/%f\n", *(int*)&tmp[i][j], tmp[i][j]);
-      }
-
-      memcpy( rmesa->vb.dmaptr, tmp[i], rmesa->vb.vertex_size * 4 );
-      rmesa->vb.dmaptr += rmesa->vb.vertex_size;
-      rmesa->vb.counter--;
-   }
-}
-
-
-
-static GLboolean check_vtx_fmt( GLcontext *ctx )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLuint ind = RADEON_CP_VC_FRMT_Z;
-   GLuint unit;
-
-   if (rmesa->TclFallback || rmesa->vb.fell_back || ctx->CompileFlag ||
-      (ctx->Fog.Enabled && (ctx->Fog.FogCoordinateSource == GL_FOG_COORD)))
-      return GL_FALSE;
-
-   if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) 
-      ctx->Driver.FlushVertices( ctx, FLUSH_UPDATE_CURRENT );
-   
-   /* Make all this event-driven:
-    */
-   if (ctx->Light.Enabled) {
-      ind |= RADEON_CP_VC_FRMT_N0;
-
-      /* TODO: make this data driven: If we receive only ubytes, send
-       * color as ubytes.  Also check if converting (with free
-       * checking for overflow) is cheaper than sending floats
-       * directly.
-       */
-      if (ctx->Light.ColorMaterialEnabled) {
-	 ind |= (RADEON_CP_VC_FRMT_FPCOLOR |
-		 RADEON_CP_VC_FRMT_FPALPHA);
-      }
-      else
-	 ind |= RADEON_CP_VC_FRMT_PKCOLOR; /* for alpha? */
-   }
-   else {
-      /* TODO: make this data driven?
-       */
-      ind |= RADEON_CP_VC_FRMT_PKCOLOR;
-	 
-      if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) {
-	 ind |= RADEON_CP_VC_FRMT_PKSPEC;
-      }
-   }
-
-   if ( ctx->Fog.FogCoordinateSource == GL_FOG_COORD ) {
-      ind |= RADEON_CP_VC_FRMT_PKSPEC;
-   }
-
-   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (ctx->Texture.Unit[unit]._ReallyEnabled) {
-	 if (ctx->Texture.Unit[unit].TexGenEnabled) {
-	    if (rmesa->TexGenNeedNormals[unit]) {
-	       ind |= RADEON_CP_VC_FRMT_N0;
-	    }
-	 } else {
-	    if (ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][3] != 1.0) {
-	       if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
-		  fprintf(stderr, "%s: q%u\n", __FUNCTION__, unit);
-	       return GL_FALSE;
-	    }
-	    ind |= RADEON_ST_BIT(unit);
-	 }
-      }
-   }
-
-   if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_STATE))
-      fprintf(stderr, "%s: format: 0x%x\n", __FUNCTION__, ind );
-
-   RADEON_NEWPRIM(rmesa);
-   rmesa->vb.vertex_format = ind;
-   rmesa->vb.vertex_size = 3;
-   rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive;
-
-   rmesa->vb.normalptr = ctx->Current.Attrib[VERT_ATTRIB_NORMAL];
-   rmesa->vb.colorptr = NULL;
-   rmesa->vb.floatcolorptr = ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
-   rmesa->vb.specptr = NULL;
-   rmesa->vb.floatspecptr = ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
-   rmesa->vb.texcoordptr[0] = ctx->Current.Attrib[VERT_ATTRIB_TEX0];
-   rmesa->vb.texcoordptr[1] = ctx->Current.Attrib[VERT_ATTRIB_TEX1];
-   rmesa->vb.texcoordptr[2] = ctx->Current.Attrib[VERT_ATTRIB_TEX2];
-   rmesa->vb.texcoordptr[3] = ctx->Current.Attrib[VERT_ATTRIB_TEX0]; /* dummy */
-
-   /* Run through and initialize the vertex components in the order
-    * the hardware understands:
-    */
-   if (ind & RADEON_CP_VC_FRMT_N0) {
-      rmesa->vb.normalptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 3;
-      rmesa->vb.normalptr[0] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][0];
-      rmesa->vb.normalptr[1] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][1];
-      rmesa->vb.normalptr[2] = ctx->Current.Attrib[VERT_ATTRIB_NORMAL][2];
-   }
-
-   if (ind & RADEON_CP_VC_FRMT_PKCOLOR) {
-      rmesa->vb.colorptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].color;
-      rmesa->vb.vertex_size += 1;
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->red,   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->green, ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->blue,  ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.colorptr->alpha, ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] );
-   }
-
-   if (ind & RADEON_CP_VC_FRMT_FPCOLOR) {
-      assert(!(ind & RADEON_CP_VC_FRMT_PKCOLOR));
-      rmesa->vb.floatcolorptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 3;
-      rmesa->vb.floatcolorptr[0] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][0];
-      rmesa->vb.floatcolorptr[1] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][1];
-      rmesa->vb.floatcolorptr[2] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][2];
-
-      if (ind & RADEON_CP_VC_FRMT_FPALPHA) {
-	 rmesa->vb.vertex_size += 1;
-	 rmesa->vb.floatcolorptr[3] = ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3];
-      }
-   }
-   
-   if (ind & RADEON_CP_VC_FRMT_PKSPEC) {
-      rmesa->vb.specptr = &rmesa->vb.vertex[rmesa->vb.vertex_size].color;
-      rmesa->vb.vertex_size += 1;
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->red,   ctx->Current.Attrib[VERT_ATTRIB_COLOR1][0] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->green, ctx->Current.Attrib[VERT_ATTRIB_COLOR1][1] );
-      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->blue,  ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] );
-      /* fog ??? */
-/*      UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->alpha,
-			       radeonComputeFogFactor(ctx->Current.Attrib[VERT_ATTRIB_FOG][0]) ); */
-   }
-
-   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (ind & RADEON_ST_BIT(unit)) {
-	 rmesa->vb.texcoordptr[unit] = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-	 rmesa->vb.vertex_size += 2;
-	 rmesa->vb.texcoordptr[unit][0] = ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][0];
-	 rmesa->vb.texcoordptr[unit][1] = ctx->Current.Attrib[VERT_ATTRIB_TEX(unit)][1];
-      }
-   }
-
-   if (rmesa->vb.installed_vertex_format != rmesa->vb.vertex_format) {
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "reinstall on vertex_format change\n");
-      _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-      rmesa->vb.installed_vertex_format = rmesa->vb.vertex_format;
-   }
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s -- success\n", __FUNCTION__);
-   
-   return GL_TRUE;
-}
-
-void radeonVtxfmtInvalidate( GLcontext *ctx )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-
-   rmesa->vb.recheck = GL_TRUE;
-   rmesa->vb.fell_back = GL_FALSE;
-}
-
-
-static void radeonVtxfmtValidate( GLcontext *ctx )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (ctx->Driver.NeedFlush)
-      ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   rmesa->vb.recheck = GL_FALSE;
-
-   if (check_vtx_fmt( ctx )) {
-      if (!rmesa->vb.installed) {
-	 if (RADEON_DEBUG & DEBUG_VFMT)
-	    fprintf(stderr, "reinstall (new install)\n");
-
-	 _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-	 ctx->Driver.FlushVertices = radeonVtxfmtFlushVertices;
-	 ctx->Driver.NewList = radeonNewList;
-	 rmesa->vb.installed = GL_TRUE;
-      }
-      else if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: already installed", __FUNCTION__);
-   } 
-   else {
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: failed\n", __FUNCTION__);
-
-      if (rmesa->vb.installed) {
-	 if (rmesa->dma.flush)
-	    rmesa->dma.flush( rmesa );
-	 _tnl_wakeup_exec( ctx );
-	 ctx->Driver.FlushVertices = radeonFlushVertices;
-	 ctx->Driver.NewList =_tnl_NewList;
-	 rmesa->vb.installed = GL_FALSE;
-      }
-   }      
-}
-
-
-
-/* Materials:
- */
-static void radeon_Materialfv( GLenum face, GLenum pname, 
-			       const GLfloat *params )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      VFMT_FALLBACK( __FUNCTION__ );
-      CALL_Materialfv(GET_DISPATCH(), (face, pname, params));
-      return;
-   }
-   _mesa_noop_Materialfv( face, pname, params );
-   radeonUpdateMaterial( ctx );
-}
-
-
-/* Begin/End
- */
-static void radeon_Begin( GLenum mode )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s( %s )\n", __FUNCTION__,
-	      _mesa_lookup_enum_by_nr( mode ));
-
-   if (mode > GL_POLYGON) {
-      _mesa_error( ctx, GL_INVALID_ENUM, "glBegin" );
-      return;
-   }
-
-   if (rmesa->vb.prim[0] != GL_POLYGON+1) {
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" );
-      return;
-   }
-   
-   if (ctx->NewState) 
-      _mesa_update_state( ctx );
-
-   if (rmesa->NewGLState)
-      radeonValidateState( ctx );
-
-   if (rmesa->vb.recheck) 
-      radeonVtxfmtValidate( ctx );
-
-   if (!rmesa->vb.installed) {
-      CALL_Begin(GET_DISPATCH(), (mode));
-      return;
-   }
-
-
-   if (rmesa->dma.flush && rmesa->vb.counter < 12) {
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s: flush almost-empty buffers\n", __FUNCTION__);
-      flush_prims( rmesa );
-   }
-
-   /* Need to arrange to save vertices here?  Or always copy from dma (yuk)?
-    */
-   if (!rmesa->dma.flush) {
-/* FIXME: what are these constants? */
-      if (rmesa->dma.current.ptr + 12*rmesa->vb.vertex_size*4 > 
-	  rmesa->dma.current.end) {
-	 RADEON_NEWPRIM( rmesa );
-	 radeonRefillCurrentDmaRegion( rmesa );
-      }
-
-      rmesa->vb.dmaptr = (int *)(rmesa->dma.current.address + rmesa->dma.current.ptr);
-      rmesa->vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / 
-	 (rmesa->vb.vertex_size * 4);
-      rmesa->vb.counter--;
-      rmesa->vb.initial_counter = rmesa->vb.counter;
-      rmesa->vb.notify = wrap_buffer;
-      rmesa->dma.flush = flush_prims;
-      ctx->Driver.NeedFlush |= FLUSH_STORED_VERTICES;
-   }
-   
-   
-   rmesa->vb.prim[0] = mode;
-   start_prim( rmesa, mode | PRIM_BEGIN );
-}
-
-
-
-static void radeon_End( void )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   if (rmesa->vb.prim[0] == GL_POLYGON+1) {
-      _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" );
-      return;
-   }
-	  
-   note_last_prim( rmesa, PRIM_END );
-   rmesa->vb.prim[0] = GL_POLYGON+1;
-}
-
-
-/* Fallback on difficult entrypoints:
- */
-#define PRE_LOOPBACK( FUNC )			\
-do {						\
-   if (RADEON_DEBUG & DEBUG_VFMT) 		\
-      fprintf(stderr, "%s\n", __FUNCTION__);	\
-   VFMT_FALLBACK( __FUNCTION__ );		\
-} while (0)
-#define TAG(x) radeon_fallback_##x
-#include "vtxfmt_tmp.h"
-
-
-
-static GLboolean radeonNotifyBegin( GLcontext *ctx, GLenum p )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-   
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(!rmesa->vb.installed);
-
-   if (ctx->NewState) 
-      _mesa_update_state( ctx );
-
-   if (rmesa->NewGLState)
-      radeonValidateState( ctx );
-
-   if (ctx->Driver.NeedFlush)
-      ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush );
-
-   if (rmesa->vb.recheck) 
-      radeonVtxfmtValidate( ctx );
-
-   if (!rmesa->vb.installed) {
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "%s -- failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   radeon_Begin( p );
-   return GL_TRUE;
-}
-
-static void radeonVtxfmtFlushVertices( GLcontext *ctx, GLuint flags )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-
-   if (RADEON_DEBUG & DEBUG_VFMT)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   assert(rmesa->vb.installed);
-
-   if (flags & FLUSH_UPDATE_CURRENT) {
-      radeon_copy_to_current( ctx );
-      if (RADEON_DEBUG & DEBUG_VFMT)
-	 fprintf(stderr, "reinstall on update_current\n");
-      _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );
-      ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT;
-   }
-
-   if (flags & FLUSH_STORED_VERTICES) {
-      radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-      assert (rmesa->dma.flush == 0 ||
-	      rmesa->dma.flush == flush_prims);
-      if (rmesa->dma.flush == flush_prims)
-	 flush_prims( RADEON_CONTEXT( ctx ) );
-      ctx->Driver.NeedFlush &= ~FLUSH_STORED_VERTICES;
-   }
-}
-
-
-
-/* At this point, don't expect very many versions of each function to
- * be generated, so not concerned about freeing them?
- */
-
-
-void radeonVtxfmtInit( GLcontext *ctx, GLboolean useCodegen )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-   GLvertexformat *vfmt = &(rmesa->vb.vtxfmt);
-
-   /* start by initializing to no-op functions */
-   _mesa_noop_vtxfmt_init(vfmt);
-
-   /* Hook in chooser functions for codegen, etc:
-    */
-   radeonVtxfmtInitChoosers( vfmt );
-
-   /* Handled fully in supported states, but no codegen:
-    */
-   vfmt->Materialfv = radeon_Materialfv;
-   vfmt->ArrayElement = _ae_loopback_array_elt;	        /* generic helper */
-   vfmt->Begin = radeon_Begin;
-   vfmt->End = radeon_End;
-
-   /* Fallback for performance reasons:  (Fix with cva/elt path here and
-    * dmatmp2.h style primitive-merging)
-    *
-    * These should call NotifyBegin(), as should _tnl_EvalMesh, to allow
-    * a driver-hook.
-    */
-   vfmt->DrawArrays = radeon_fallback_DrawArrays;
-   vfmt->DrawElements = radeon_fallback_DrawElements;
-   vfmt->DrawRangeElements = radeon_fallback_DrawRangeElements; 
-
-   /* Active but unsupported -- fallback if we receive these:
-    */
-   vfmt->CallList = radeon_fallback_CallList;
-   vfmt->CallLists = radeon_fallback_CallLists;
-   vfmt->EvalCoord1f = radeon_fallback_EvalCoord1f;
-   vfmt->EvalCoord1fv = radeon_fallback_EvalCoord1fv;
-   vfmt->EvalCoord2f = radeon_fallback_EvalCoord2f;
-   vfmt->EvalCoord2fv = radeon_fallback_EvalCoord2fv;
-   vfmt->EvalMesh1 = radeon_fallback_EvalMesh1;
-   vfmt->EvalMesh2 = radeon_fallback_EvalMesh2;
-   vfmt->EvalPoint1 = radeon_fallback_EvalPoint1;
-   vfmt->EvalPoint2 = radeon_fallback_EvalPoint2;
-   vfmt->TexCoord3f = radeon_fallback_TexCoord3f;
-   vfmt->TexCoord3fv = radeon_fallback_TexCoord3fv;
-   vfmt->TexCoord4f = radeon_fallback_TexCoord4f;
-   vfmt->TexCoord4fv = radeon_fallback_TexCoord4fv;
-   vfmt->MultiTexCoord3fARB = radeon_fallback_MultiTexCoord3fARB;
-   vfmt->MultiTexCoord3fvARB = radeon_fallback_MultiTexCoord3fvARB;
-   vfmt->MultiTexCoord4fARB = radeon_fallback_MultiTexCoord4fARB;
-   vfmt->MultiTexCoord4fvARB = radeon_fallback_MultiTexCoord4fvARB;
-   vfmt->Vertex4f = radeon_fallback_Vertex4f;
-   vfmt->Vertex4fv = radeon_fallback_Vertex4fv;
-   vfmt->VertexAttrib1fNV  = radeon_fallback_VertexAttrib1fNV;
-   vfmt->VertexAttrib1fvNV = radeon_fallback_VertexAttrib1fvNV;
-   vfmt->VertexAttrib2fNV  = radeon_fallback_VertexAttrib2fNV;
-   vfmt->VertexAttrib2fvNV = radeon_fallback_VertexAttrib2fvNV;
-   vfmt->VertexAttrib3fNV  = radeon_fallback_VertexAttrib3fNV;
-   vfmt->VertexAttrib3fvNV = radeon_fallback_VertexAttrib3fvNV;
-   vfmt->VertexAttrib4fNV  = radeon_fallback_VertexAttrib4fNV;
-   vfmt->VertexAttrib4fvNV = radeon_fallback_VertexAttrib4fvNV;
-   vfmt->FogCoordfEXT = radeon_fallback_FogCoordfEXT;
-   vfmt->FogCoordfvEXT = radeon_fallback_FogCoordfvEXT;
-
-   (void)radeon_fallback_vtxfmt;
-
-   TNL_CONTEXT(ctx)->Driver.NotifyBegin = radeonNotifyBegin;
-
-   rmesa->vb.enabled = 1;
-   rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive;
-   rmesa->vb.primflags = 0;
-
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex2f );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex2fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Vertex3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4ub );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4ubv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3ub );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3ubv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4f );
-   make_empty_list( &rmesa->vb.dfn_cache.Color4fv );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Color3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   make_empty_list( &rmesa->vb.dfn_cache.Normal3f );
-   make_empty_list( &rmesa->vb.dfn_cache.Normal3fv );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord2f );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord2fv );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord1f );
-   make_empty_list( &rmesa->vb.dfn_cache.TexCoord1fv );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-
-   radeonInitCodegen( &rmesa->vb.codegen, useCodegen );
-}
-
-static void free_funcs( struct dynfn *l )
-{
-   struct dynfn *f, *tmp;
-   foreach_s (f, tmp, l) {
-      remove_from_list( f );
-      _mesa_exec_free( f->code );
-      _mesa_free( f );
-   }
-}
-
-
-
-void radeonVtxfmtMakeCurrent( GLcontext *ctx )
-{
-}
-
-
-void radeonVtxfmtDestroy( GLcontext *ctx )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
-
-   count_funcs( rmesa );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex2f );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex2fv );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex3f );
-   free_funcs( &rmesa->vb.dfn_cache.Vertex3fv );
-   free_funcs( &rmesa->vb.dfn_cache.Color4ub );
-   free_funcs( &rmesa->vb.dfn_cache.Color4ubv );
-   free_funcs( &rmesa->vb.dfn_cache.Color3ub );
-   free_funcs( &rmesa->vb.dfn_cache.Color3ubv );
-   free_funcs( &rmesa->vb.dfn_cache.Color4f );
-   free_funcs( &rmesa->vb.dfn_cache.Color4fv );
-   free_funcs( &rmesa->vb.dfn_cache.Color3f );
-   free_funcs( &rmesa->vb.dfn_cache.Color3fv );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fEXT );
-   free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT );
-   free_funcs( &rmesa->vb.dfn_cache.Normal3f );
-   free_funcs( &rmesa->vb.dfn_cache.Normal3fv );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord2f );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord2fv );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord1f );
-   free_funcs( &rmesa->vb.dfn_cache.TexCoord1fv );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fARB );
-   free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB );
-}
-
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.h b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.h
deleted file mode 100644
index a656e49e445..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.h
+++ /dev/null
@@ -1,120 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt.h,v 1.3 2002/12/21 17:02:16 dawes Exp $ */
-/**************************************************************************
-
-Copyright 2000, 2001 ATI Technologies Inc., Ontario, Canada, and
-                     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, 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 __RADEON_VTXFMT_H__
-#define __RADEON_VTXFMT_H__
-
-#include "radeon_context.h"
-
-
-extern void radeonVtxfmtUpdate( GLcontext *ctx );
-extern void radeonVtxfmtInit( GLcontext *ctx, GLboolean useCodegen );
-extern void radeonVtxfmtInvalidate( GLcontext *ctx );
-extern void radeonVtxfmtDestroy( GLcontext *ctx );
-extern void radeonVtxfmtInitChoosers( GLvertexformat *vfmt );
-
-extern void radeonVtxfmtMakeCurrent( GLcontext *ctx );
-extern void radeonVtxfmtUnbindContext( GLcontext *ctx );
-
-extern void radeon_copy_to_current( GLcontext *ctx );
-
-#define DFN( FUNC, CACHE)				\
-do {							\
-   char *start = (char *)&FUNC;				\
-   char *end = (char *)&FUNC##_end;			\
-   insert_at_head( &CACHE, dfn );			\
-   dfn->key = key;					\
-   dfn->code = _mesa_exec_malloc( end - start );	\
-   _mesa_memcpy(dfn->code, start, end - start);		\
-}							\
-while ( 0 )
-
-#define FIXUP( CODE, OFFSET, CHECKVAL, NEWVAL )	\
-do {						\
-   int *icode = (int *)(CODE+OFFSET);		\
-   assert (*icode == CHECKVAL);			\
-   *icode = (int)NEWVAL;			\
-} while (0)
-
-
-/* Useful for figuring out the offsets:
- */
-#define FIXUP2( CODE, OFFSET, CHECKVAL, NEWVAL )		\
-do {								\
-   while (*(int *)(CODE+OFFSET) != CHECKVAL) OFFSET++;		\
-   fprintf(stderr, "%s/%d CVAL %x OFFSET %d VAL %x\n", __FUNCTION__,	\
-	   __LINE__, CHECKVAL, OFFSET, (int)(NEWVAL));			\
-   *(int *)(CODE+OFFSET) = (int)(NEWVAL);				\
-   OFFSET += 4;							\
-} while (0)
-
-/* 
- */
-void radeonInitCodegen( struct dfn_generators *gen, GLboolean useCodegen );
-void radeonInitX86Codegen( struct dfn_generators *gen );
-void radeonInitSSECodegen( struct dfn_generators *gen );
-
-
-
-/* Defined in radeon_vtxfmt_x86.c
- */
-struct dynfn *radeon_makeX86Vertex2f( GLcontext *, int );
-struct dynfn *radeon_makeX86Vertex2fv( GLcontext *, int );
-struct dynfn *radeon_makeX86Vertex3f( GLcontext *, int );
-struct dynfn *radeon_makeX86Vertex3fv( GLcontext *, int );
-struct dynfn *radeon_makeX86Color4ub( GLcontext *, int );
-struct dynfn *radeon_makeX86Color4ubv( GLcontext *, int );
-struct dynfn *radeon_makeX86Color3ub( GLcontext *, int );
-struct dynfn *radeon_makeX86Color3ubv( GLcontext *, int );
-struct dynfn *radeon_makeX86Color4f( GLcontext *, int );
-struct dynfn *radeon_makeX86Color4fv( GLcontext *, int );
-struct dynfn *radeon_makeX86Color3f( GLcontext *, int );
-struct dynfn *radeon_makeX86Color3fv( GLcontext *, int );
-struct dynfn *radeon_makeX86SecondaryColor3ubEXT( GLcontext *, int );
-struct dynfn *radeon_makeX86SecondaryColor3ubvEXT( GLcontext *, int );
-struct dynfn *radeon_makeX86SecondaryColor3fEXT( GLcontext *, int );
-struct dynfn *radeon_makeX86SecondaryColor3fvEXT( GLcontext *, int );
-struct dynfn *radeon_makeX86Normal3f( GLcontext *, int );
-struct dynfn *radeon_makeX86Normal3fv( GLcontext *, int );
-struct dynfn *radeon_makeX86TexCoord2f( GLcontext *, int );
-struct dynfn *radeon_makeX86TexCoord2fv( GLcontext *, int );
-struct dynfn *radeon_makeX86TexCoord1f( GLcontext *, int );
-struct dynfn *radeon_makeX86TexCoord1fv( GLcontext *, int );
-struct dynfn *radeon_makeX86MultiTexCoord2fARB( GLcontext *, int );
-struct dynfn *radeon_makeX86MultiTexCoord2fvARB( GLcontext *, int );
-struct dynfn *radeon_makeX86MultiTexCoord1fARB( GLcontext *, int );
-struct dynfn *radeon_makeX86MultiTexCoord1fvARB( GLcontext *, int );
-
-#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
deleted file mode 100644
index aac029aa0e4..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
+++ /dev/null
@@ -1,924 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt_c.c,v 1.2 2002/12/16 16:18:59 dawes Exp $ */
-/**************************************************************************
-
-Copyright 2002 ATI Technologies Inc., Ontario, Canada, and
-                     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, 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 "glheader.h"
-#include "mtypes.h"
-#include "colormac.h"
-#include "simple_list.h"
-#include "api_noop.h"
-#include "vtxfmt.h"
-
-#include "radeon_vtxfmt.h"
-
-#include "dispatch.h"
-
-/* Fallback versions of all the entrypoints for situations where
- * codegen isn't available.  This is still a lot faster than the
- * vb/pipeline implementation in Mesa.
- */
-static void radeon_Vertex3f( GLfloat x, GLfloat y, GLfloat z )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&x;
-   *rmesa->vb.dmaptr++ = *(int *)&y;
-   *rmesa->vb.dmaptr++ = *(int *)&z;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void radeon_Vertex3fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&v[0];
-   *rmesa->vb.dmaptr++ = *(int *)&v[1];
-   *rmesa->vb.dmaptr++ = *(int *)&v[2];
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void radeon_Vertex2f( GLfloat x, GLfloat y )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&x;
-   *rmesa->vb.dmaptr++ = *(int *)&y;
-   *rmesa->vb.dmaptr++ = 0;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = *(int *)&rmesa->vb.vertex[i];
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-static void radeon_Vertex2fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   int i;
-
-   *rmesa->vb.dmaptr++ = *(int *)&v[0];
-   *rmesa->vb.dmaptr++ = *(int *)&v[1];
-   *rmesa->vb.dmaptr++ = 0;
-
-   for (i = 3; i < rmesa->vb.vertex_size; i++)
-      *rmesa->vb.dmaptr++ = rmesa->vb.vertex[i].i;
-   
-   if (--rmesa->vb.counter == 0)
-      rmesa->vb.notify();
-}
-
-
-#if 0
-/* Color for ubyte (packed) color formats:
- */
-static void radeon_Color3ub_ub( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= 0xff;
-}
-
-static void radeon_Color3ubv_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= v[0];
-   dest->green	= v[1];
-   dest->blue	= v[2];
-   dest->alpha	= 0xff;
-}
-
-static void radeon_Color4ub_ub( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= a;
-}
-
-static void radeon_Color4ubv_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   *(GLuint *)rmesa->vb.colorptr = LE32_TO_CPU(*(GLuint *)v);
-}
-#endif /* 0 */
-
-static void radeon_Color3f_ub( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   dest->alpha = 255;
-}
-
-static void radeon_Color3fv_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   dest->alpha = 255;
-}
-
-static void radeon_Color4f_ub( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,   r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->alpha, a );
-}
-
-static void radeon_Color4fv_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.colorptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->alpha, v[3] );
-}
-
-
-/* Color for float color+alpha formats:
- */
-#if 0
-static void radeon_Color3ub_4f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = 1.0;
-}
-
-static void radeon_Color3ubv_4f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = 1.0;
-}
-
-static void radeon_Color4ub_4f( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = UBYTE_TO_FLOAT(a);
-}
-
-static void radeon_Color4ubv_4f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = UBYTE_TO_FLOAT(v[3]);
-}
-#endif /* 0 */
-
-
-static void radeon_Color3f_4f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = 1.0;		
-}
-
-static void radeon_Color3fv_4f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = 1.0;
-}
-
-static void radeon_Color4f_4f( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = a;
-}
-
-static void radeon_Color4fv_4f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = v[3];
-}
-
-
-/* Color for float color formats:
- */
-#if 0
-static void radeon_Color3ub_3f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-}
-
-static void radeon_Color3ubv_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-}
-
-static void radeon_Color4ub_3f( GLubyte r, GLubyte g, GLubyte b, GLubyte a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT(a);
-}
-
-static void radeon_Color4ubv_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = UBYTE_TO_FLOAT(v[3]);
-}
-#endif /* 0 */
-
-
-static void radeon_Color3f_3f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-}
-
-static void radeon_Color3fv_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-}
-
-static void radeon_Color4f_3f( GLfloat r, GLfloat g, GLfloat b, GLfloat a )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = a;
-}
-
-static void radeon_Color4fv_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatcolorptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = v[3]; 
-}
-
-
-/* Secondary Color:
- */
-#if 0
-static void radeon_SecondaryColor3ubEXT_ub( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.specptr;
-   dest->red	= r;
-   dest->green	= g;
-   dest->blue	= b;
-   dest->alpha	= 0xff;
-}
-
-static void radeon_SecondaryColor3ubvEXT_ub( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.specptr;
-   dest->red	= v[0];
-   dest->green	= v[1];
-   dest->blue	= v[2];
-   dest->alpha	= 0xff;
-}
-#endif /* 0 */
-
-static void radeon_SecondaryColor3fEXT_ub( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.specptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  r );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, g );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  b );
-   dest->alpha = 255;
-}
-
-static void radeon_SecondaryColor3fvEXT_ub( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   radeon_color_t *dest = rmesa->vb.specptr;
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->red,	  v[0] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->green, v[1] );
-   UNCLAMPED_FLOAT_TO_UBYTE( dest->blue,  v[2] );
-   dest->alpha = 255;
-}
-
-#if 0
-static void radeon_SecondaryColor3ubEXT_3f( GLubyte r, GLubyte g, GLubyte b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = UBYTE_TO_FLOAT(r);
-   dest[1] = UBYTE_TO_FLOAT(g);
-   dest[2] = UBYTE_TO_FLOAT(b);
-   dest[3] = 1.0;
-}
-
-static void radeon_SecondaryColor3ubvEXT_3f( const GLubyte *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = UBYTE_TO_FLOAT(v[0]);
-   dest[1] = UBYTE_TO_FLOAT(v[1]);
-   dest[2] = UBYTE_TO_FLOAT(v[2]);
-   dest[3] = 1.0;
-}
-#endif /* 0 */
-
-static void radeon_SecondaryColor3fEXT_3f( GLfloat r, GLfloat g, GLfloat b )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = r;
-   dest[1] = g;
-   dest[2] = b;
-   dest[3] = 1.0;
-}
-
-static void radeon_SecondaryColor3fvEXT_3f( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.floatspecptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-   dest[3] = 1.0;
-}
-
-
-/* Normal
- */
-static void radeon_Normal3f( GLfloat n0, GLfloat n1, GLfloat n2 )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.normalptr;
-   dest[0] = n0;
-   dest[1] = n1;
-   dest[2] = n2;
-}
-
-static void radeon_Normal3fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.normalptr;
-   dest[0] = v[0];
-   dest[1] = v[1];
-   dest[2] = v[2];
-}
-
-
-/* TexCoord
- */
-static void radeon_TexCoord1f( GLfloat s )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[0];
-   dest[0] = s;
-   dest[1] = 0;
-}
-
-static void radeon_TexCoord1fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[0];
-   dest[0] = v[0];
-   dest[1] = 0;
-}
-
-static void radeon_TexCoord2f( GLfloat s, GLfloat t )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[0];
-   dest[0] = s;
-   dest[1] = t;
-}
-
-static void radeon_TexCoord2fv( const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[0];
-   dest[0] = v[0];
-   dest[1] = v[1];
-}
-
-
-/* MultiTexcoord
- * 
- * Technically speaking, these functions should subtract GL_TEXTURE0 from
- * \c target before masking and using it.  The value of GL_TEXTURE0 is 0x84C0,
- * which has the low-order 5 bits 0.  For all possible valid values of 
- * \c target.  Subtracting GL_TEXTURE0 has the net effect of masking \c target
- * with 0x1F.  Masking with 0x1F and then masking with 0x01 is redundant, so
- * the subtraction has been omitted.
- */
-
-static void radeon_MultiTexCoord1fARB( GLenum target, GLfloat s  )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
-   dest[0] = s;
-   dest[1] = 0;
-}
-
-static void radeon_MultiTexCoord1fvARB( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
-   dest[0] = v[0];
-   dest[1] = 0;
-}
-
-static void radeon_MultiTexCoord2fARB( GLenum target, GLfloat s, GLfloat t )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
-   dest[0] = s;
-   dest[1] = t;
-}
-
-static void radeon_MultiTexCoord2fvARB( GLenum target, const GLfloat *v )
-{
-   GET_CURRENT_CONTEXT(ctx);
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
-   dest[0] = v[0];
-   dest[1] = v[1];
-}
-
-static struct dynfn *lookup( struct dynfn *l, int key )
-{
-   struct dynfn *f;
-
-   foreach( f, l ) {
-      if (f->key == key) 
-	 return f;
-   }
-
-   return NULL;
-}
-
-/* Can't use the loopback template for this:
- */
-
-#define CHOOSE(FN, FNTYPE, MASK, ACTIVE, ARGS1, ARGS2 )			\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx);						\
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);			\
-   int key = rmesa->vb.vertex_format & (MASK|ACTIVE);			\
-   struct dynfn *dfn;							\
-									\
-   dfn = lookup( &rmesa->vb.dfn_cache.FN, key );			\
-   if (dfn == 0)							\
-      dfn = rmesa->vb.codegen.FN( ctx, key );				\
-   else if (RADEON_DEBUG & DEBUG_CODEGEN)				\
-      fprintf(stderr, "%s -- cached codegen\n", __FUNCTION__ );		\
-									\
-   if (dfn)								\
-      SET_ ## FN (ctx->Exec, (FNTYPE)(dfn->code));			\
-   else {								\
-      if (RADEON_DEBUG & DEBUG_CODEGEN)					\
-	 fprintf(stderr, "%s -- generic version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, radeon_##FN);				\
-   }									\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;			\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-/* For the _3f case, only allow one color function to be hooked in at
- * a time.  Eventually, use a similar mechanism to allow selecting the
- * color component of the vertex format based on client behaviour.  
- *
- * Note:  Perform these actions even if there is a codegen or cached 
- * codegen version of the chosen function.
- */
-#define CHOOSE_COLOR(FN, FNTYPE, NR, MASK, ACTIVE, ARGS1, ARGS2 )	\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx); \
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);			\
-   int key = rmesa->vb.vertex_format & (MASK|ACTIVE);			\
-   struct dynfn *dfn;							\
-									\
-   if (rmesa->vb.vertex_format & ACTIVE_PKCOLOR) {			\
-      SET_ ## FN (ctx->Exec, radeon_##FN##_ub);				\
-   }									\
-   else if ((rmesa->vb.vertex_format &					\
-            (ACTIVE_FPCOLOR|ACTIVE_FPALPHA)) == ACTIVE_FPCOLOR) {	\
-									\
-      if (rmesa->vb.installed_color_3f_sz != NR) {			\
-         rmesa->vb.installed_color_3f_sz = NR;				\
-         if (NR == 3) ctx->Current.Attrib[VERT_ATTRIB_COLOR0][3] = 1.0;	\
-         if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) {		\
-            radeon_copy_to_current( ctx );				\
-            _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt );	\
-            CALL_ ## FN (ctx->Exec, ARGS2);				\
-            return;							\
-         }								\
-      }									\
-									\
-      SET_ ## FN (ctx->Exec, radeon_##FN##_3f);				\
-   }									\
-   else {								\
-      SET_ ## FN (ctx->Exec, radeon_##FN##_4f);				\
-   }									\
-									\
-									\
-   dfn = lookup( &rmesa->vb.dfn_cache.FN, key );			\
-   if (!dfn) dfn = rmesa->vb.codegen.FN( ctx, key );			\
-									\
-   if (dfn) {								\
-      if (RADEON_DEBUG & DEBUG_CODEGEN)					\
-         fprintf(stderr, "%s -- codegen version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, (FNTYPE)dfn->code);			\
-   }									\
-   else if (RADEON_DEBUG & DEBUG_CODEGEN)				\
-         fprintf(stderr, "%s -- 'c' version\n", __FUNCTION__ );		\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;			\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-/* Right now there are both _ub and _3f versions of the secondary color
- * functions.  Currently, we only set-up the hardware to use the _ub versions.
- * The _3f versions are needed for the cases where secondary color isn't used
- * in the vertex format, but it still needs to be stored in the context
- * state vector.
- */
-#define CHOOSE_SECONDARY_COLOR(FN, FNTYPE, MASK, ACTIVE, ARGS1, ARGS2 )	\
-static void choose_##FN ARGS1						\
-{									\
-   GET_CURRENT_CONTEXT(ctx);						\
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);			\
-   int key = rmesa->vb.vertex_format & (MASK|ACTIVE);			\
-   struct dynfn *dfn = lookup( &rmesa->vb.dfn_cache.FN, key );		\
-									\
-   if (dfn == 0)							\
-      dfn = rmesa->vb.codegen.FN( ctx, key );				\
-   else  if (RADEON_DEBUG & DEBUG_CODEGEN)				\
-      fprintf(stderr, "%s -- cached version\n", __FUNCTION__ );		\
-									\
-   if (dfn)								\
-      SET_ ## FN (ctx->Exec, (FNTYPE)(dfn->code));			\
-   else {								\
-      if (RADEON_DEBUG & DEBUG_CODEGEN)					\
-         fprintf(stderr, "%s -- generic version\n", __FUNCTION__ );	\
-      SET_ ## FN (ctx->Exec, ((rmesa->vb.vertex_format & ACTIVE_PKSPEC) != 0)	\
-	  ? radeon_##FN##_ub : radeon_##FN##_3f);			\
-   }									\
-									\
-   ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT;			\
-   CALL_ ## FN (ctx->Exec, ARGS2);					\
-}
-
-
-
-
-
-/* Shorthands
- */
-#define ACTIVE_XYZW (RADEON_CP_VC_FRMT_W0|RADEON_CP_VC_FRMT_Z)
-#define ACTIVE_NORM RADEON_CP_VC_FRMT_N0
-
-#define ACTIVE_PKCOLOR RADEON_CP_VC_FRMT_PKCOLOR
-#define ACTIVE_FPCOLOR RADEON_CP_VC_FRMT_FPCOLOR
-#define ACTIVE_FPALPHA RADEON_CP_VC_FRMT_FPALPHA
-#define ACTIVE_COLOR (ACTIVE_FPCOLOR|ACTIVE_PKCOLOR)
-
-#define ACTIVE_PKSPEC RADEON_CP_VC_FRMT_PKSPEC
-#define ACTIVE_FPSPEC RADEON_CP_VC_FRMT_FPSPEC
-#define ACTIVE_SPEC   (ACTIVE_FPSPEC|ACTIVE_PKSPEC)
-
-#define ACTIVE_ST0 RADEON_CP_VC_FRMT_ST0
-#define ACTIVE_ST1 RADEON_CP_VC_FRMT_ST1
-#define ACTIVE_ST2 RADEON_CP_VC_FRMT_ST2
-#define ACTIVE_ST_ALL (RADEON_CP_VC_FRMT_ST1|RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST2)
-
-/* Each codegen function should be able to be fully specified by a
- * subsetted version of rmesa->vb.vertex_format.
- */
-#define MASK_NORM    (ACTIVE_XYZW)
-#define MASK_COLOR   (MASK_NORM|ACTIVE_NORM)
-#define MASK_SPEC    (MASK_COLOR|ACTIVE_COLOR)
-#define MASK_ST0     (MASK_SPEC|ACTIVE_SPEC)
-#define MASK_ST1     (MASK_ST0|ACTIVE_ST0)
-#define MASK_ST2     (MASK_ST1|ACTIVE_ST1)
-#define MASK_ST_ALL  (MASK_ST2|ACTIVE_ST2)
-#define MASK_VERTEX  (MASK_ST_ALL|ACTIVE_FPALPHA) 
-
-
-typedef void (*p4f)( GLfloat, GLfloat, GLfloat, GLfloat );
-typedef void (*p3f)( GLfloat, GLfloat, GLfloat );
-typedef void (*p2f)( GLfloat, GLfloat );
-typedef void (*p1f)( GLfloat );
-typedef void (*pe2f)( GLenum, GLfloat, GLfloat );
-typedef void (*pe1f)( GLenum, GLfloat );
-typedef void (*p4ub)( GLubyte, GLubyte, GLubyte, GLubyte );
-typedef void (*p3ub)( GLubyte, GLubyte, GLubyte );
-typedef void (*pfv)( const GLfloat * );
-typedef void (*pefv)( GLenum, const GLfloat * );
-typedef void (*pubv)( const GLubyte * );
-
-
-CHOOSE(Normal3f, p3f, MASK_NORM, ACTIVE_NORM, 
-       (GLfloat a,GLfloat b,GLfloat c), (a,b,c))
-CHOOSE(Normal3fv, pfv, MASK_NORM, ACTIVE_NORM, 
-       (const GLfloat *v), (v))
-
-#if 0
-CHOOSE_COLOR(Color4ub, p4ub, 4, MASK_COLOR, ACTIVE_COLOR,
-	(GLubyte a,GLubyte b, GLubyte c, GLubyte d), (a,b,c,d))
-CHOOSE_COLOR(Color4ubv, pubv, 4, MASK_COLOR, ACTIVE_COLOR, 
-	(const GLubyte *v), (v))
-CHOOSE_COLOR(Color3ub, p3ub, 3, MASK_COLOR, ACTIVE_COLOR, 
-	(GLubyte a,GLubyte b, GLubyte c), (a,b,c))
-CHOOSE_COLOR(Color3ubv, pubv, 3, MASK_COLOR, ACTIVE_COLOR, 
-	(const GLubyte *v), (v))
-#endif
-
-CHOOSE_COLOR(Color4f, p4f, 4, MASK_COLOR, ACTIVE_COLOR, 
-	(GLfloat a,GLfloat b, GLfloat c, GLfloat d), (a,b,c,d))
-CHOOSE_COLOR(Color4fv, pfv, 4, MASK_COLOR, ACTIVE_COLOR, 
-	(const GLfloat *v), (v))
-CHOOSE_COLOR(Color3f, p3f, 3, MASK_COLOR, ACTIVE_COLOR,
-	(GLfloat a,GLfloat b, GLfloat c), (a,b,c))
-CHOOSE_COLOR(Color3fv, pfv, 3, MASK_COLOR, ACTIVE_COLOR,
-	(const GLfloat *v), (v))
-
-
-#if 0
-CHOOSE_SECONDARY_COLOR(SecondaryColor3ubEXT, p3ub, MASK_SPEC, ACTIVE_SPEC,
-	(GLubyte a,GLubyte b, GLubyte c), (a,b,c))
-CHOOSE_SECONDARY_COLOR(SecondaryColor3ubvEXT, pubv, MASK_SPEC, ACTIVE_SPEC,
-	(const GLubyte *v), (v))
-#endif
-CHOOSE_SECONDARY_COLOR(SecondaryColor3fEXT, p3f, MASK_SPEC, ACTIVE_SPEC,
-	(GLfloat a,GLfloat b, GLfloat c), (a,b,c))
-CHOOSE_SECONDARY_COLOR(SecondaryColor3fvEXT, pfv, MASK_SPEC, ACTIVE_SPEC,
-	(const GLfloat *v), (v))
-
-CHOOSE(TexCoord2f, p2f, MASK_ST0, ACTIVE_ST0, 
-       (GLfloat a,GLfloat b), (a,b))
-CHOOSE(TexCoord2fv, pfv, MASK_ST0, ACTIVE_ST0, 
-       (const GLfloat *v), (v))
-CHOOSE(TexCoord1f, p1f, MASK_ST0, ACTIVE_ST0, 
-       (GLfloat a), (a))
-CHOOSE(TexCoord1fv, pfv, MASK_ST0, ACTIVE_ST0, 
-       (const GLfloat *v), (v))
-
-CHOOSE(MultiTexCoord2fARB, pe2f, MASK_ST_ALL, ACTIVE_ST_ALL,
-	 (GLenum u,GLfloat a,GLfloat b), (u,a,b))
-CHOOSE(MultiTexCoord2fvARB, pefv, MASK_ST_ALL, ACTIVE_ST_ALL,
-	(GLenum u,const GLfloat *v), (u,v))
-CHOOSE(MultiTexCoord1fARB, pe1f, MASK_ST_ALL, ACTIVE_ST_ALL,
-	 (GLenum u,GLfloat a), (u,a))
-CHOOSE(MultiTexCoord1fvARB, pefv, MASK_ST_ALL, ACTIVE_ST_ALL,
-	(GLenum u,const GLfloat *v), (u,v))
-
-CHOOSE(Vertex3f, p3f, MASK_VERTEX, MASK_VERTEX, 
-       (GLfloat a,GLfloat b,GLfloat c), (a,b,c))
-CHOOSE(Vertex3fv, pfv, MASK_VERTEX, MASK_VERTEX, 
-       (const GLfloat *v), (v))
-CHOOSE(Vertex2f, p2f, MASK_VERTEX, MASK_VERTEX, 
-       (GLfloat a,GLfloat b), (a,b))
-CHOOSE(Vertex2fv, pfv, MASK_VERTEX, MASK_VERTEX, 
-       (const GLfloat *v), (v))
-
-
-
-
-
-void radeonVtxfmtInitChoosers( GLvertexformat *vfmt )
-{
-   vfmt->Color3f = choose_Color3f;
-   vfmt->Color3fv = choose_Color3fv;
-   vfmt->Color4f = choose_Color4f;
-   vfmt->Color4fv = choose_Color4fv;
-   vfmt->SecondaryColor3fEXT = choose_SecondaryColor3fEXT;
-   vfmt->SecondaryColor3fvEXT = choose_SecondaryColor3fvEXT;
-   vfmt->MultiTexCoord1fARB = choose_MultiTexCoord1fARB;
-   vfmt->MultiTexCoord1fvARB = choose_MultiTexCoord1fvARB;
-   vfmt->MultiTexCoord2fARB = choose_MultiTexCoord2fARB;
-   vfmt->MultiTexCoord2fvARB = choose_MultiTexCoord2fvARB;
-   vfmt->Normal3f = choose_Normal3f;
-   vfmt->Normal3fv = choose_Normal3fv;
-   vfmt->TexCoord1f = choose_TexCoord1f;
-   vfmt->TexCoord1fv = choose_TexCoord1fv;
-   vfmt->TexCoord2f = choose_TexCoord2f;
-   vfmt->TexCoord2fv = choose_TexCoord2fv;
-   vfmt->Vertex2f = choose_Vertex2f;
-   vfmt->Vertex2fv = choose_Vertex2fv;
-   vfmt->Vertex3f = choose_Vertex3f;
-   vfmt->Vertex3fv = choose_Vertex3fv;
-
-#if 0
-   vfmt->Color3ub = choose_Color3ub;
-   vfmt->Color3ubv = choose_Color3ubv;
-   vfmt->Color4ub = choose_Color4ub;
-   vfmt->Color4ubv = choose_Color4ubv;
-   vfmt->SecondaryColor3ubEXT = choose_SecondaryColor3ubEXT;
-   vfmt->SecondaryColor3ubvEXT = choose_SecondaryColor3ubvEXT;
-#endif
-}
-
-
-static struct dynfn *codegen_noop( GLcontext *ctx, int key )
-{
-   (void) ctx; (void) key;
-   return NULL;
-}
-
-void radeonInitCodegen( struct dfn_generators *gen, GLboolean useCodegen )
-{
-   gen->Vertex3f = codegen_noop;
-   gen->Vertex3fv = codegen_noop;
-   gen->Color4ub = codegen_noop;
-   gen->Color4ubv = codegen_noop;
-   gen->Normal3f = codegen_noop;
-   gen->Normal3fv = codegen_noop;
-   gen->TexCoord2f = codegen_noop;
-   gen->TexCoord2fv = codegen_noop;
-   gen->MultiTexCoord2fARB = codegen_noop;
-   gen->MultiTexCoord2fvARB = codegen_noop;
-   gen->Vertex2f = codegen_noop;
-   gen->Vertex2fv = codegen_noop;
-   gen->Color3ub = codegen_noop;
-   gen->Color3ubv = codegen_noop;
-   gen->Color4f = codegen_noop;
-   gen->Color4fv = codegen_noop;
-   gen->Color3f = codegen_noop;
-   gen->Color3fv = codegen_noop;
-   gen->SecondaryColor3fEXT = codegen_noop;
-   gen->SecondaryColor3fvEXT = codegen_noop;
-   gen->SecondaryColor3ubEXT = codegen_noop;
-   gen->SecondaryColor3ubvEXT = codegen_noop;
-   gen->TexCoord1f = codegen_noop;
-   gen->TexCoord1fv = codegen_noop;
-   gen->MultiTexCoord1fARB = codegen_noop;
-   gen->MultiTexCoord1fvARB = codegen_noop;
-
-   if (useCodegen) {
-#if defined(USE_X86_ASM)
-      radeonInitX86Codegen( gen );
-#endif
-
-#if defined(USE_SSE_ASM)
-      radeonInitSSECodegen( gen );
-#endif
-   }
-}
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c
deleted file mode 100644
index 0f0fc9e0653..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c
+++ /dev/null
@@ -1,236 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt_sse.c,v 1.1 2002/10/30 12:51:58 alanh Exp $ */
-/**************************************************************************
-
-Copyright 2000, 2001 ATI Technologies Inc., Ontario, Canada, and
-                     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, 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 "glheader.h"
-#include "imports.h"
-#include "simple_list.h" 
-#include "radeon_vtxfmt.h"
-
-#if defined(USE_SSE_ASM)
-#include "x86/common_x86_asm.h"
-
-#define EXTERN( FUNC )		\
-extern const char *FUNC;	\
-extern const char *FUNC##_end
-
-EXTERN( _sse_Attribute2fv );
-EXTERN( _sse_Attribute2f );
-EXTERN( _sse_Attribute3fv );
-EXTERN( _sse_Attribute3f );
-EXTERN( _sse_MultiTexCoord2fv );
-EXTERN( _sse_MultiTexCoord2f );
-EXTERN( _sse_MultiTexCoord2fv_2 );
-EXTERN( _sse_MultiTexCoord2f_2 );
-
-/* Build specialized versions of the immediate calls on the fly for
- * the current state.
- */
-
-static struct dynfn *radeon_makeSSEAttribute2fv( struct dynfn * cache, int key,
-					       const char * name, void * dest)
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _sse_Attribute2fv, (*cache) );
-   FIXUP(dfn->code, 10, 0x0, (int)dest);
-   return dfn;
-}
-
-static struct dynfn *radeon_makeSSEAttribute2f( struct dynfn * cache, int key,
-					      const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _sse_Attribute2f, (*cache) );
-   FIXUP(dfn->code, 8, 0x0, (int)dest); 
-   return dfn;
-}
-
-static struct dynfn *radeon_makeSSEAttribute3fv( struct dynfn * cache, int key,
-					       const char * name, void * dest)
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _sse_Attribute3fv, (*cache) );
-   FIXUP(dfn->code, 13, 0x0, (int)dest);
-   FIXUP(dfn->code, 18, 0x8, 8+(int)dest);
-   return dfn;
-}
-
-static struct dynfn *radeon_makeSSEAttribute3f( struct dynfn * cache, int key,
-					      const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _sse_Attribute3f, (*cache) );
-   FIXUP(dfn->code, 12, 0x0, (int)dest); 
-   FIXUP(dfn->code, 17, 0x8, 8+(int)dest); 
-   return dfn;
-}
-
-static struct dynfn * radeon_makeSSENormal3fv( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeSSEAttribute3fv( & rmesa->vb.dfn_cache.Normal3fv, key,
-				      __FUNCTION__, rmesa->vb.normalptr );
-}
-
-static struct dynfn *radeon_makeSSENormal3f( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeSSEAttribute3f( & rmesa->vb.dfn_cache.Normal3f, key,
-				     __FUNCTION__, rmesa->vb.normalptr );
-}
-
-static struct dynfn *radeon_makeSSEColor3fv( GLcontext *ctx, int key )
-{
-   if (key & (RADEON_CP_VC_FRMT_PKCOLOR|RADEON_CP_VC_FRMT_FPALPHA))
-      return NULL;
-   else
-   {
-      radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-      return radeon_makeSSEAttribute3fv( & rmesa->vb.dfn_cache.Color3fv, key,
-					 __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-static struct dynfn *radeon_makeSSEColor3f( GLcontext *ctx, int key )
-{
-   if (key & (RADEON_CP_VC_FRMT_PKCOLOR|RADEON_CP_VC_FRMT_FPALPHA))
-      return NULL;
-   else
-   {
-      radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-      return radeon_makeSSEAttribute3f( & rmesa->vb.dfn_cache.Color3f, key,
-					__FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-static struct dynfn *radeon_makeSSETexCoord2fv( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeSSEAttribute2fv( & rmesa->vb.dfn_cache.TexCoord2fv, key,
-				       __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-static struct dynfn *radeon_makeSSETexCoord2f( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeSSEAttribute2f( & rmesa->vb.dfn_cache.TexCoord2f, key,
-				     __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-#if 0 /* Temporarily disabled - probably needs adjustments for more than 2 tex units -rs */
-static struct dynfn *radeon_makeSSEMultiTexCoord2fv( GLcontext *ctx, int key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if ((key & (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) ==
-      (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) {
-      DFN ( _sse_MultiTexCoord2fv, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 18, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);	
-   } else {
-      DFN ( _sse_MultiTexCoord2fv_2, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 14, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-
-static struct dynfn *radeon_makeSSEMultiTexCoord2f( GLcontext *ctx, int key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if ((key & (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) ==
-      (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) {
-      DFN ( _sse_MultiTexCoord2f, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 16, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);	
-   } else {
-      DFN ( _sse_MultiTexCoord2f_2, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 15, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-#endif
-
-void radeonInitSSECodegen( struct dfn_generators *gen )
-{
-   if ( cpu_has_xmm ) {
-      gen->Normal3fv = (void *) radeon_makeSSENormal3fv;
-      gen->Normal3f = (void *) radeon_makeSSENormal3f;
-      gen->Color3fv = (void *) radeon_makeSSEColor3fv;
-      gen->Color3f = (void *) radeon_makeSSEColor3f;
-      gen->TexCoord2fv = (void *) radeon_makeSSETexCoord2fv;
-      gen->TexCoord2f = (void *) radeon_makeSSETexCoord2f;
-#if 0 /* Temporarily disabled - probably needs adjustments for more than 2 tex units -rs */
-      gen->MultiTexCoord2fvARB = (void *) radeon_makeSSEMultiTexCoord2fv;
-      gen->MultiTexCoord2fARB = (void *) radeon_makeSSEMultiTexCoord2f;
-#endif
-   }
-}
-
-#else 
-
-void radeonInitSSECodegen( struct dfn_generators *gen )
-{
-   (void) gen;
-}
-
-#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c
deleted file mode 100644
index 529e79065e9..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c
+++ /dev/null
@@ -1,440 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxfmt_x86.c,v 1.2 2002/12/21 17:02:16 dawes Exp $ */
-/**************************************************************************
-
-Copyright 2000, 2001 ATI Technologies Inc., Ontario, Canada, and
-                     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, 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 "glheader.h"
-#include "imports.h"
-#include "simple_list.h" 
-#include "radeon_vtxfmt.h"
-
-#if defined(USE_X86_ASM)
-
-#define EXTERN( FUNC )		\
-extern const char *FUNC;	\
-extern const char *FUNC##_end
-
-EXTERN ( _x86_Attribute2fv );
-EXTERN ( _x86_Attribute2f );
-EXTERN ( _x86_Attribute3fv );
-EXTERN ( _x86_Attribute3f );
-EXTERN ( _x86_Vertex3fv_6 );
-EXTERN ( _x86_Vertex3fv_8 );
-EXTERN ( _x86_Vertex3fv );
-EXTERN ( _x86_Vertex3f_4 );
-EXTERN ( _x86_Vertex3f_6 );
-EXTERN ( _x86_Vertex3f );
-EXTERN ( _x86_Color4ubv_ub );
-EXTERN ( _x86_Color4ubv_4f );
-EXTERN ( _x86_Color4ub_ub );
-EXTERN ( _x86_MultiTexCoord2fv );
-EXTERN ( _x86_MultiTexCoord2fv_2 );
-EXTERN ( _x86_MultiTexCoord2f );
-EXTERN ( _x86_MultiTexCoord2f_2 );
-
-
-/* Build specialized versions of the immediate calls on the fly for
- * the current state.  Generic x86 versions.
- */
-
-struct dynfn *radeon_makeX86Vertex3f( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x %d\n", __FUNCTION__, key, rmesa->vb.vertex_size );
-
-   switch (rmesa->vb.vertex_size) {
-   case 4: {
-
-      DFN ( _x86_Vertex3f_4, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 2, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 25, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 36, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 46, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 51, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 60, 0x0, (int)&rmesa->vb.notify);
-      break;
-   }
-   case 6: {
-
-      DFN ( _x86_Vertex3f_6, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 3, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 28, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 34, 0x0, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 40, 0x0, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 57, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 63, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 70, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 79, 0x0, (int)&rmesa->vb.notify);
-      break;
-   }
-   default: {
-
-      DFN ( _x86_Vertex3f, rmesa->vb.dfn_cache.Vertex3f );
-      FIXUP(dfn->code, 3, 0x0, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 9, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 37, 0x0, rmesa->vb.vertex_size-3);
-      FIXUP(dfn->code, 44, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 50, 0x0, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 56, 0x0, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x0, (int)&rmesa->vb.notify);
-   break;
-   }
-   }
-
-   return dfn;
-}
-
-
-
-struct dynfn *radeon_makeX86Vertex3fv( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x %d\n", __FUNCTION__, key, rmesa->vb.vertex_size );
-
-   switch (rmesa->vb.vertex_size) {
-   case 6: {
-
-      DFN ( _x86_Vertex3fv_6, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 1, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 27, 0x0000001c, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 33, 0x00000020, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 45, 0x00000024, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 56, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 61, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 76, 0x00000008, (int)&rmesa->vb.notify);
-      break;
-   }
-   
-
-   case 8: {
-
-      DFN ( _x86_Vertex3fv_8, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 1, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 27, 0x0000001c, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 33, 0x00000020, (int)&rmesa->vb.vertex[4]);
-      FIXUP(dfn->code, 45, 0x0000001c, (int)&rmesa->vb.vertex[5]);
-      FIXUP(dfn->code, 51, 0x00000020, (int)&rmesa->vb.vertex[6]);
-      FIXUP(dfn->code, 63, 0x00000024, (int)&rmesa->vb.vertex[7]);
-      FIXUP(dfn->code, 74, 0x00000000, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 79, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 85, 0x00000004, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 94, 0x00000008, (int)&rmesa->vb.notify);
-      break;
-   }
-   
-
-
-   default: {
-
-      DFN ( _x86_Vertex3fv, rmesa->vb.dfn_cache.Vertex3fv );
-      FIXUP(dfn->code, 8, 0x01010101, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 32, 0x00000006, rmesa->vb.vertex_size-3);
-      FIXUP(dfn->code, 37, 0x00000058, (int)&rmesa->vb.vertex[3]);
-      FIXUP(dfn->code, 45, 0x01010101, (int)&rmesa->vb.dmaptr);
-      FIXUP(dfn->code, 50, 0x02020202, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 58, 0x02020202, (int)&rmesa->vb.counter);
-      FIXUP(dfn->code, 67, 0x0, (int)&rmesa->vb.notify);
-   break;
-   }
-   }
-
-   return dfn;
-}
-
-static struct dynfn *
-radeon_makeX86Attribute2fv( struct dynfn * cache, int key,
-			    const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _x86_Attribute2fv, (*cache) );
-   FIXUP(dfn->code, 11, 0x0, (int)dest); 
-   FIXUP(dfn->code, 16, 0x4, 4+(int)dest); 
-
-   return dfn;
-}
-
-static struct dynfn *
-radeon_makeX86Attribute2f( struct dynfn * cache, int key,
-			   const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _x86_Attribute2f, (*cache) );
-   FIXUP(dfn->code, 1, 0x0, (int)dest); 
-
-   return dfn;
-}
-
-
-static struct dynfn *
-radeon_makeX86Attribute3fv( struct dynfn * cache, int key,
-			    const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _x86_Attribute3fv, (*cache) );
-   FIXUP(dfn->code, 14, 0x0, (int)dest); 
-   FIXUP(dfn->code, 20, 0x4, 4+(int)dest); 
-   FIXUP(dfn->code, 25, 0x8, 8+(int)dest);
-
-   return dfn;
-}
-
-static struct dynfn *
-radeon_makeX86Attribute3f( struct dynfn * cache, int key,
-			 const char * name, void * dest )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", name, key );
-
-   DFN ( _x86_Attribute3f, (*cache) );
-   FIXUP(dfn->code, 14, 0x0, (int)dest); 
-   FIXUP(dfn->code, 20, 0x4, 4+(int)dest); 
-   FIXUP(dfn->code, 25, 0x8, 8+(int)dest);
-
-   return dfn;
-}
-
-struct dynfn *radeon_makeX86Normal3fv( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeX86Attribute3fv( & rmesa->vb.dfn_cache.Normal3fv, key,
-				      __FUNCTION__, rmesa->vb.normalptr );
-}
-
-struct dynfn *radeon_makeX86Normal3f( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeX86Attribute3f( & rmesa->vb.dfn_cache.Normal3f, key,
-				     __FUNCTION__, rmesa->vb.normalptr );
-}
-
-struct dynfn *radeon_makeX86Color4ubv( GLcontext *ctx, int key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if (key & RADEON_CP_VC_FRMT_PKCOLOR) {
-      DFN ( _x86_Color4ubv_ub, rmesa->vb.dfn_cache.Color4ubv);
-      FIXUP(dfn->code, 5, 0x12345678, (int)rmesa->vb.colorptr); 
-      return dfn;
-   } 
-   else {
-
-      DFN ( _x86_Color4ubv_4f, rmesa->vb.dfn_cache.Color4ubv);
-      FIXUP(dfn->code, 2, 0x00000000, (int)_mesa_ubyte_to_float_color_tab); 
-      FIXUP(dfn->code, 27, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr); 
-      FIXUP(dfn->code, 33, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+4); 
-      FIXUP(dfn->code, 55, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+8); 
-      FIXUP(dfn->code, 61, 0xdeadbeaf, (int)rmesa->vb.floatcolorptr+12); 
-      return dfn;
-   }
-}
-
-struct dynfn *radeon_makeX86Color4ub( GLcontext *ctx, int key )
-{
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if (key & RADEON_CP_VC_FRMT_PKCOLOR) {
-      struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-      radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-      DFN ( _x86_Color4ub_ub, rmesa->vb.dfn_cache.Color4ub );
-      FIXUP(dfn->code, 18, 0x0, (int)rmesa->vb.colorptr); 
-      FIXUP(dfn->code, 24, 0x0, (int)rmesa->vb.colorptr+1); 
-      FIXUP(dfn->code, 30, 0x0, (int)rmesa->vb.colorptr+2); 
-      FIXUP(dfn->code, 36, 0x0, (int)rmesa->vb.colorptr+3); 
-      return dfn;
-   }
-   else
-      return NULL;
-}
-
-
-struct dynfn *radeon_makeX86Color3fv( GLcontext *ctx, int key )
-{
-   if (key & (RADEON_CP_VC_FRMT_PKCOLOR|RADEON_CP_VC_FRMT_FPALPHA))
-      return NULL;
-   else
-   {
-      radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-      return radeon_makeX86Attribute3fv( & rmesa->vb.dfn_cache.Color3fv, key,
-					 __FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-struct dynfn *radeon_makeX86Color3f( GLcontext *ctx, int key )
-{
-   if (key & (RADEON_CP_VC_FRMT_PKCOLOR|RADEON_CP_VC_FRMT_FPALPHA))
-      return NULL;
-   else
-   {
-      radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-      return radeon_makeX86Attribute3f( & rmesa->vb.dfn_cache.Color3f, key,
-					__FUNCTION__, rmesa->vb.floatcolorptr );
-   }
-}
-
-
-
-struct dynfn *radeon_makeX86TexCoord2fv( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeX86Attribute2fv( & rmesa->vb.dfn_cache.TexCoord2fv, key,
-				      __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-struct dynfn *radeon_makeX86TexCoord2f( GLcontext *ctx, int key )
-{
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   return radeon_makeX86Attribute2f( & rmesa->vb.dfn_cache.TexCoord2f, key,
-				     __FUNCTION__, rmesa->vb.texcoordptr[0] );
-}
-
-#if 0 /* Temporarily disabled - probably needs adjustments for more than 2 tex units -rs */
-struct dynfn *radeon_makeX86MultiTexCoord2fvARB( GLcontext *ctx, int key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if ((key & (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) ==
-      (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) {
-      DFN ( _x86_MultiTexCoord2fv, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 21, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);
-      FIXUP(dfn->code, 27, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]+4);
-   } else {
-      DFN ( _x86_MultiTexCoord2fv_2, rmesa->vb.dfn_cache.MultiTexCoord2fvARB );
-      FIXUP(dfn->code, 14, 0x0, (int)rmesa->vb.texcoordptr);
-   }
-   return dfn;
-}
-
-struct dynfn *radeon_makeX86MultiTexCoord2fARB( GLcontext *ctx, 
-						int key )
-{
-   struct dynfn *dfn = MALLOC_STRUCT( dynfn );
-   radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-
-   if (RADEON_DEBUG & DEBUG_CODEGEN)
-      fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key );
-
-   if ((key & (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) ==
-       (RADEON_CP_VC_FRMT_ST0|RADEON_CP_VC_FRMT_ST1)) {
-      DFN ( _x86_MultiTexCoord2f, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 20, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]);
-      FIXUP(dfn->code, 26, 0xdeadbeef, (int)rmesa->vb.texcoordptr[0]+4); 
-   }
-   else {
-      /* Note: this might get generated multiple times, even though the
-       * actual emitted code is the same.
-       */
-      DFN ( _x86_MultiTexCoord2f_2, rmesa->vb.dfn_cache.MultiTexCoord2fARB );
-      FIXUP(dfn->code, 18, 0x0, (int)rmesa->vb.texcoordptr); 
-   }      
-   return dfn;
-}
-#endif
-
-void radeonInitX86Codegen( struct dfn_generators *gen )
-{
-   gen->Vertex3f = radeon_makeX86Vertex3f;
-   gen->Vertex3fv = radeon_makeX86Vertex3fv;
-   gen->Color4ub = radeon_makeX86Color4ub; /* PKCOLOR only */
-   gen->Color4ubv = radeon_makeX86Color4ubv; /* PKCOLOR only */
-   gen->Normal3f = radeon_makeX86Normal3f;
-   gen->Normal3fv = radeon_makeX86Normal3fv;
-   gen->TexCoord2f = radeon_makeX86TexCoord2f;
-   gen->TexCoord2fv = radeon_makeX86TexCoord2fv;
-#if 0 /* Temporarily disabled - probably needs adjustments for more than 2 tex units -rs */
-   gen->MultiTexCoord2fARB = radeon_makeX86MultiTexCoord2fARB;
-   gen->MultiTexCoord2fvARB = radeon_makeX86MultiTexCoord2fvARB;
-#endif
-   gen->Color3f = radeon_makeX86Color3f;
-   gen->Color3fv = radeon_makeX86Color3fv;
-
-   /* Not done:
-    */
-/*     gen->Vertex2f = radeon_makeX86Vertex2f; */
-/*     gen->Vertex2fv = radeon_makeX86Vertex2fv; */
-/*     gen->Color3ub = radeon_makeX86Color3ub; */
-/*     gen->Color3ubv = radeon_makeX86Color3ubv; */
-/*     gen->Color4f = radeon_makeX86Color4f; */
-/*     gen->Color4fv = radeon_makeX86Color4fv; */
-/*     gen->TexCoord1f = radeon_makeX86TexCoord1f; */
-/*     gen->TexCoord1fv = radeon_makeX86TexCoord1fv; */
-/*     gen->MultiTexCoord1fARB = radeon_makeX86MultiTexCoord1fARB; */
-/*     gen->MultiTexCoord1fvARB = radeon_makeX86MultiTexCoord1fvARB; */
-}
-
-
-#else 
-
-void radeonInitX86Codegen( struct dfn_generators *gen )
-{
-   (void) gen;
-}
-
-#endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S b/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S
deleted file mode 100644
index 1b433491aa8..00000000000
--- a/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S
+++ /dev/null
@@ -1,498 +0,0 @@
-/* $XFree86: xc/lib/GL/mesa/src/drv/radeon/radeon_vtxtmp_x86.S,v 1.1 2002/10/30 12:51:58 alanh Exp $ */
-/**************************************************************************
-
-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, TUNGSTEN GRAPHICS 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.
-
-**************************************************************************/
-
-#define GLOBL( x )	\
-.globl x;		\
-x:
-
-.data
-.align 4
-
-/*
-	vertex 3f vertex size 4
-*/
-	
-GLOBL ( _x86_Vertex3f_4 )
-	movl	(0), %ecx
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	%eax, (%ecx)
-	movl	%edx, 4(%ecx)
-	movl	12(%esp), %eax
-	movl	(0), %edx
-	movl	%eax, 8(%ecx)
-	movl	%edx, 12(%ecx)
-	movl	(0), %eax
-	addl	$16, %ecx
-	dec 	%eax
-	movl	%ecx, (0)
-	movl	%eax, (0)
-	je	.1 
-	ret
-.1:	jmp	*0
-	
-GLOBL ( _x86_Vertex3f_4_end )
-
-/*
-	vertex 3f vertex size 6
-*/
-GLOBL ( _x86_Vertex3f_6 )
-	push	%edi
-	movl	(0), %edi
-	movl	8(%esp), %eax
-	movl	12(%esp), %edx
-	movl	16(%esp), %ecx
-	movl	%eax, (%edi)
-	movl	%edx, 4(%edi)
-	movl	%ecx, 8(%edi)
-	movl	(0), %eax
-	movl	(0), %edx
-	movl	(0), %ecx
-	movl	%eax, 12(%edi)
-	movl	%edx, 16(%edi)
-	movl	%ecx, 20(%edi)
-	addl	$24, %edi
-	movl	(0), %eax
-	movl	%edi, (0)
-	dec 	%eax
-	pop 	%edi
-	movl	%eax, (0)
-	je	.2
-	ret
-.2:	jmp	*0
-GLOBL ( _x86_Vertex3f_6_end )
-/*
-	vertex 3f generic size
-*/
-GLOBL ( _x86_Vertex3f )
-	push	%edi
-	push	%esi
-	movl	$0, %esi
-	movl	(0), %edi
-	movl	12(%esp), %eax
-	movl	16(%esp), %edx
-	movl	20(%esp), %ecx
-	movl	%eax, (%edi)
-	movl	%edx, 4(%edi)
-	movl	%ecx, 8(%edi)
-	addl	$12, %edi
-	movl	$0, %ecx
-	repz
-	movsl %ds:(%esi), %es:(%edi)
-	movl	(0), %eax
-	movl	%edi, (0)
-	dec 	%eax
-	movl	%eax, (0)
-	pop 	%esi
-	pop 	%edi
-	je  	.3
-	ret
-.3:	jmp	*0
-
-GLOBL ( _x86_Vertex3f_end )
-
-/*
-	Vertex 3fv vertex size 6
-*/
-GLOBL ( _x86_Vertex3fv_6 )
-	movl	(0), %eax
-	movl	4(%esp), %ecx
-	movl	(%ecx), %edx
-	movl	%edx, (%eax)
-	movl	4(%ecx), %edx
-	movl	8(%ecx), %ecx
-	movl	%edx, 4(%eax)
-	movl	%ecx, 8(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 12(%eax)
-	movl	%ecx, 16(%eax)
-	movl	(36), %edx
-	movl	%edx, 20(%eax)
-	addl	$24, %eax
-	movl	%eax, 0
-	movl	4, %eax
-	dec 	%eax
-	movl	%eax, 4
-	je	.4
-	ret
-.4:	jmp    *8
-	
-GLOBL ( _x86_Vertex3fv_6_end )
-
-/*
-	Vertex 3fv vertex size 8
-*/
-GLOBL ( _x86_Vertex3fv_8 )
-	movl	(0), %eax
-	movl	4(%esp), %ecx
-	movl	(%ecx), %edx
-	movl	%edx ,(%eax)
-	movl	4(%ecx) ,%edx
-	movl	8(%ecx) ,%ecx
-	movl	%edx, 4(%eax)
-	movl	%ecx, 8(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 12(%eax)
-	movl	%ecx, 16(%eax)
-	movl	(28), %edx
-	movl	(32), %ecx
-	movl	%edx, 20(%eax)
-	movl	%ecx, 24(%eax)
-	movl	(36), %edx
-	movl	%edx, 28(%eax)
-	addl	$32, %eax
-	movl	%eax, (0)
-	movl	4, %eax
-	dec	%eax
-	movl    %eax, (4)
-	je	.5
-	ret
-.5:	jmp    *8
-	
-GLOBL ( _x86_Vertex3fv_8_end )
-
-/*
-	Vertex 3fv generic vertex size
-*/
-GLOBL ( _x86_Vertex3fv )
-	movl	4(%esp), %edx
-	push	%edi
-	push	%esi
-	movl	(0x1010101), %edi
-	movl	(%edx), %eax
-	movl	4(%edx), %ecx
-	movl	8(%edx), %esi
-	movl	%eax, (%edi)
-	movl	%ecx, 4(%edi)
-	movl	%esi, 8(%edi)
-	addl	$12, %edi
-	movl	$6, %ecx
-	movl	$0x58, %esi
-	repz
-	movsl %ds:(%esi), %es:(%edi)
-	movl	%edi, (0x1010101)
-	movl	(0x2020202), %eax
-	pop	%esi
-	pop	%edi
-	dec	%eax
-	movl	%eax, (0x2020202)
-	je	.6
-	ret
-.6:	jmp    *0
-GLOBL ( _x86_Vertex3fv_end )
-
-
-/**
- * Generic handler for 2 float format data.  This can be used for
- * TexCoord2f and possibly other functions.
- */
-
-GLOBL ( _x86_Attribute2f )
-	movl	$0x0, %edx
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	movl	%eax, (%edx)
-	movl	%ecx, 4(%edx)
-	ret
-GLOBL ( _x86_Attribute2f_end )
-
-
-/**
- * Generic handler for 2 float vector format data.  This can be used for
- * TexCoord2fv and possibly other functions.
- */
-
-GLOBL( _x86_Attribute2fv)
-	movl 4(%esp), %eax      /* load 'v' off stack */
-	movl (%eax), %ecx       /* load v[0] */
-	movl 4(%eax), %eax      /* load v[1] */
-	movl %ecx, 0      	/* store v[0] to current vertex */
-	movl %eax, 4      	/* store v[1] to current vertex */
-	ret
-GLOBL ( _x86_Attribute2fv_end )
-
-
-/**
- * Generic handler for 3 float format data.  This can be used for
- * Normal3f, Color3f (when the color target is also float), or
- * TexCoord3f.
- */
-
-GLOBL ( _x86_Attribute3f )
-	movl	4(%esp), %ecx
-	movl	8(%esp), %edx
-	movl	12(%esp), %eax
-	movl	%ecx, 0
-	movl	%edx, 4
-	movl	%eax, 8
-	ret
-GLOBL ( _x86_Attribute3f_end )
-
-/**
- * Generic handler for 3 float vector format data.  This can be used for
- * Normal3f, Color3f (when the color target is also float), or
- * TexCoord3f.
- */
-
-GLOBL( _x86_Attribute3fv)
-	movl 4(%esp), %eax      /* load 'v' off stack */
-	movl (%eax), %ecx       /* load v[0] */
-	movl 4(%eax), %edx      /* load v[1] */
-	movl 8(%eax), %eax      /* load v[2] */
-	movl %ecx, 0      	/* store v[0] to current vertex */
-	movl %edx, 4      	/* store v[1] to current vertex */
-	movl %eax, 8      	/* store v[2] to current vertex */
-	ret
-GLOBL ( _x86_Attribute3fv_end )
-
-
-/*
-	Color 4ubv_ub
-*/
-GLOBL ( _x86_Color4ubv_ub )
-	movl 4(%esp), %eax
-	movl $0x12345678, %edx
-	movl (%eax), %eax
-	movl %eax, (%edx)
-	ret
-GLOBL ( _x86_Color4ubv_ub_end )
-
-/*
-	Color 4ubv 4f
-*/
-GLOBL ( _x86_Color4ubv_4f )
-	push	%ebx
-	movl	$0, %edx
-	xor	%eax, %eax
-	xor	%ecx, %ecx
-	movl	8(%esp), %ebx
-	movl	(%ebx), %ebx
-	mov	%bl, %al
-	mov	%bh, %cl
-	movl	(%edx,%eax,4),%eax
-	movl	(%edx,%ecx,4),%ecx
-	movl	%eax, (0xdeadbeaf)
-	movl	%ecx, (0xdeadbeaf)
-	xor	%eax, %eax
-	xor	%ecx, %ecx
-	shr	$16, %ebx
-	mov	%bl, %al
-	mov	%bh, %cl
-	movl	(%edx,%eax,4), %eax
-	movl	(%edx,%ecx,4), %ecx
-	movl	%eax, (0xdeadbeaf)
-	movl	%ecx, (0xdeadbeaf)
-	pop	%ebx
-	ret
-GLOBL ( _x86_Color4ubv_4f_end )
-
-/*
-
-	Color4ub_ub
-*/
-GLOBL( _x86_Color4ub_ub )
-	push	%ebx
-	movl	8(%esp), %eax
-	movl	12(%esp), %edx
-	movl	16(%esp), %ecx
-	movl	20(%esp), %ebx
-	mov	%al, (0)
-	mov	%dl, (0)
-	mov	%cl, (0)
-	mov	%bl, (0)
-	pop	%ebx
-	ret
-GLOBL( _x86_Color4ub_ub_end )
-
-
-/*
-	MultiTexCoord2fv st0/st1
-*/
-GLOBL( _x86_MultiTexCoord2fv )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$3, %eax
-	movl	(%ecx), %edx
-	shl	$3, %eax
-	movl	4(%ecx), %ecx
-	movl	%edx, 0xdeadbeef(%eax)
-	movl	%ecx, 0xdeadbeef(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2fv_end )
-
-/*
-	MultiTexCoord2fv
-*/
-
-GLOBL( _x86_MultiTexCoord2fv_2 )
-	movl	4(%esp,1), %eax
-	movl	8(%esp,1), %ecx
-	and	$3, %eax
-	movl	0(,%eax,4), %edx
-	movl	(%ecx), %eax
-	movl	%eax, (%edx)
-	movl	4(%ecx), %eax
-	movl	%eax, 4(%edx)
-	ret
-GLOBL( _x86_MultiTexCoord2fv_2_end )
-
-/*
-	MultiTexCoord2f st0/st1
-*/
-GLOBL( _x86_MultiTexCoord2f )
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	12(%esp), %ecx
-	and	$3, %eax
-	shl	$3, %eax
-	movl	%edx, 0xdeadbeef(%eax)
-	movl	%ecx, 0xdeadbeef(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2f_end )
-
-/*
-	MultiTexCoord2f
-*/
-GLOBL( _x86_MultiTexCoord2f_2 )
-	movl	4(%esp), %eax
-	movl	8(%esp), %edx
-	movl	12(%esp,1), %ecx
-	and	$3,%eax
-	movl	0(,%eax,4), %eax
-	movl	%edx, (%eax)
-	movl	%ecx, 4(%eax)
-	ret
-GLOBL( _x86_MultiTexCoord2f_2_end )
-
-#if defined(USE_SSE_ASM)
-/**
- * This can be used as a template for either Color3fv (when the color
- * target is also a 3f) or Normal3fv.
- */
-
-GLOBL( _sse_Attribute3fv )
-	movl	4(%esp), %eax
-	movlps	(%eax), %xmm0
-	movl	8(%eax), %eax
-	movlps	%xmm0, 0
-	movl	%eax, 8
-	ret
-GLOBL( _sse_Attribute3fv_end )
-
-/**
- * This can be used as a template for either Color3f (when the color
- * target is also a 3f) or Normal3f.
- */
-
-GLOBL( _sse_Attribute3f )
-	movlps	4(%esp), %xmm0
-	movl	12(%esp), %eax
-	movlps	%xmm0, 0
-	movl	%eax, 8
-	ret
-GLOBL( _sse_Attribute3f_end )
-
-
-/**
- * Generic handler for 2 float vector format data.  This can be used for
- * TexCoord2fv and possibly other functions.
- */
-
-GLOBL( _sse_Attribute2fv )
-	movl	4(%esp), %eax
-	movlps	(%eax), %xmm0
-	movlps	%xmm0, 0
-	ret
-GLOBL( _sse_Attribute2fv_end )
-
-
-/**
- * Generic handler for 2 float format data.  This can be used for
- * TexCoord2f and possibly other functions.
- */
-
-GLOBL( _sse_Attribute2f )
-	movlps	4(%esp), %xmm0
-	movlps	%xmm0, 0
-	ret
-GLOBL( _sse_Attribute2f_end )
-
-/*
-	MultiTexCoord2fv st0/st1
-*/
-GLOBL( _sse_MultiTexCoord2fv )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$3, %eax
-	movlps	(%ecx), %xmm0
-	movlps	%xmm0, 0xdeadbeef(,%eax,8)
-	ret
-GLOBL( _sse_MultiTexCoord2fv_end )
-
-/*
-	MultiTexCoord2fv
-*/
-GLOBL( _sse_MultiTexCoord2fv_2 )
-	movl	4(%esp), %eax
-	movl	8(%esp), %ecx
-	and	$3, %eax
-	movl	0(,%eax,4), %edx
-	movlps	(%ecx), %xmm0
-	movlps	%xmm0, (%edx)
-	ret
-GLOBL( _sse_MultiTexCoord2fv_2_end )
-
-/*
-	MultiTexCoord2f st0/st1
-*/
-GLOBL( _sse_MultiTexCoord2f )
-	movl	4(%esp), %eax
-	and	$3, %eax
-	movlps	8(%esp), %xmm0
-	movlps	%xmm0, 0xdeadbeef(,%eax,8)
-	ret
-GLOBL( _sse_MultiTexCoord2f_end )
-
-/*
-	MultiTexCoord2f
-*/
-GLOBL( _sse_MultiTexCoord2f_2 )
-	movl	4(%esp), %eax
-	movlps	8(%esp), %xmm0
-	and	$3,%eax
-	movl	0(,%eax,4), %eax
-	movlps	%xmm0, (%eax)
-	ret
-GLOBL( _sse_MultiTexCoord2f_2_end )
-#endif
-
-#if defined (__ELF__) && defined (__linux__)
-	.section .note.GNU-stack,"",%progbits
-#endif
diff --git a/src/mesa/drivers/dri/s3v/s3v_context.c b/src/mesa/drivers/dri/s3v/s3v_context.c
index 7b0aa0daee0..2d2f704ad77 100644
--- a/src/mesa/drivers/dri/s3v/s3v_context.c
+++ b/src/mesa/drivers/dri/s3v/s3v_context.c
@@ -6,7 +6,7 @@
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -139,7 +139,7 @@ GLboolean s3vCreateContext(const __GLcontextModes *glVisual,
 	/* Initialize the software rasterizer and helper modules.
 	 */
 	_swrast_CreateContext( ctx );
-	_ac_CreateContext( ctx );
+	_vbo_CreateContext( ctx );
 	_tnl_CreateContext( ctx );
 	_swsetup_CreateContext( ctx );
 
diff --git a/src/mesa/drivers/dri/s3v/s3v_state.c b/src/mesa/drivers/dri/s3v/s3v_state.c
index fcea901c1c9..b86b618c117 100644
--- a/src/mesa/drivers/dri/s3v/s3v_state.c
+++ b/src/mesa/drivers/dri/s3v/s3v_state.c
@@ -10,7 +10,7 @@
 #include "colormac.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 
 /* #define DEBUG(str) printf str */
@@ -825,7 +825,7 @@ static void s3vDDUpdateState( GLcontext *ctx, GLuint new_state )
 {
 	_swrast_InvalidateState( ctx, new_state );
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 	S3V_CONTEXT(ctx)->new_gl_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/s3v/s3v_xmesa.c b/src/mesa/drivers/dri/s3v/s3v_xmesa.c
index 533424cdeab..c451f7452bf 100644
--- a/src/mesa/drivers/dri/s3v/s3v_xmesa.c
+++ b/src/mesa/drivers/dri/s3v/s3v_xmesa.c
@@ -13,7 +13,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 /* #define DEBUG(str) printf str */
 
@@ -38,7 +38,7 @@ s3vDestroyContext(__DRIcontextPrivate *driContextPriv)
     if (vmesa) {
       _swsetup_DestroyContext( vmesa->glCtx );
       _tnl_DestroyContext( vmesa->glCtx );
-      _ac_DestroyContext( vmesa->glCtx );
+      _vbo_DestroyContext( vmesa->glCtx );
       _swrast_DestroyContext( vmesa->glCtx );
 
       s3vFreeVB( vmesa->glCtx );
diff --git a/src/mesa/drivers/dri/savage/savage_xmesa.c b/src/mesa/drivers/dri/savage/savage_xmesa.c
index f51e321a4a6..ad79b9235c1 100644
--- a/src/mesa/drivers/dri/savage/savage_xmesa.c
+++ b/src/mesa/drivers/dri/savage/savage_xmesa.c
@@ -40,7 +40,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 
@@ -509,7 +509,7 @@ savageCreateContext( const __GLcontextModes *mesaVis,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    
    _swsetup_CreateContext( ctx );
@@ -599,7 +599,7 @@ savageDestroyContext(__DRIcontextPrivate *driContextPriv)
 
       _swsetup_DestroyContext(imesa->glCtx );
       _tnl_DestroyContext( imesa->glCtx );
-      _ac_DestroyContext( imesa->glCtx );
+      _vbo_DestroyContext( imesa->glCtx );
       _swrast_DestroyContext( imesa->glCtx );
 
       /* free the Mesa context */
diff --git a/src/mesa/drivers/dri/savage/savagestate.c b/src/mesa/drivers/dri/savage/savagestate.c
index 2f20dc23b6e..741a9dda4f7 100644
--- a/src/mesa/drivers/dri/savage/savagestate.c
+++ b/src/mesa/drivers/dri/savage/savagestate.c
@@ -41,7 +41,7 @@
 #include "savage_bci.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -1673,7 +1673,7 @@ static void savageDDInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    SAVAGE_CONTEXT(ctx)->new_gl_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/sis/sis6326_state.c b/src/mesa/drivers/dri/sis/sis6326_state.c
index 6bc2c6de3bf..08402fb3e2a 100644
--- a/src/mesa/drivers/dri/sis/sis6326_state.c
+++ b/src/mesa/drivers/dri/sis/sis6326_state.c
@@ -37,7 +37,7 @@
 #include "enums.h"
 #include "colormac.h"
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -645,7 +645,7 @@ sis6326DDInvalidateState( GLcontext *ctx, GLuint new_state )
 
 	_swrast_InvalidateState( ctx, new_state );
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 	smesa->NewGLState |= new_state;
 }
diff --git a/src/mesa/drivers/dri/sis/sis_context.c b/src/mesa/drivers/dri/sis/sis_context.c
index a300a080ec0..89b81da347a 100644
--- a/src/mesa/drivers/dri/sis/sis_context.c
+++ b/src/mesa/drivers/dri/sis/sis_context.c
@@ -53,7 +53,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -306,7 +306,7 @@ sisCreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -353,7 +353,7 @@ sisDestroyContext ( __DRIcontextPrivate *driContextPriv )
    if ( smesa != NULL ) {
       _swsetup_DestroyContext( smesa->glCtx );
       _tnl_DestroyContext( smesa->glCtx );
-      _ac_DestroyContext( smesa->glCtx );
+      _vbo_DestroyContext( smesa->glCtx );
       _swrast_DestroyContext( smesa->glCtx );
 
       if (smesa->using_agp)
diff --git a/src/mesa/drivers/dri/sis/sis_state.c b/src/mesa/drivers/dri/sis/sis_state.c
index 2a10a5fbf62..33a2f089b82 100644
--- a/src/mesa/drivers/dri/sis/sis_state.c
+++ b/src/mesa/drivers/dri/sis/sis_state.c
@@ -42,7 +42,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "enums.h"
 #include "colormac.h"
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -707,7 +707,7 @@ sisDDInvalidateState( GLcontext *ctx, GLuint new_state )
 
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    smesa->NewGLState |= new_state;
 }
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.c b/src/mesa/drivers/dri/tdfx/tdfx_context.c
index bd9dade89b5..a9163f49a8f 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.c
@@ -49,7 +49,7 @@
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -328,7 +328,7 @@ GLboolean tdfxCreateContext( const __GLcontextModes *mesaVis,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -609,7 +609,7 @@ tdfxDestroyContext( __DRIcontextPrivate *driContextPriv )
 
       _swsetup_DestroyContext( fxMesa->glCtx );
       _tnl_DestroyContext( fxMesa->glCtx );
-      _ac_DestroyContext( fxMesa->glCtx );
+      _vbo_DestroyContext( fxMesa->glCtx );
       _swrast_DestroyContext( fxMesa->glCtx );
 
       tdfxFreeVB( fxMesa->glCtx );
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_state.c b/src/mesa/drivers/dri/tdfx/tdfx_state.c
index 59e6549e5ef..42cb5dfaa38 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_state.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_state.c
@@ -46,7 +46,7 @@
 #include "teximage.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
 #include "swrast_setup/swrast_setup.h"
@@ -1234,7 +1234,7 @@ static void tdfxDDInvalidateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    TDFX_CONTEXT(ctx)->new_gl_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/trident/trident_context.c b/src/mesa/drivers/dri/trident/trident_context.c
index d3806586475..dbbd1ac0c69 100644
--- a/src/mesa/drivers/dri/trident/trident_context.c
+++ b/src/mesa/drivers/dri/trident/trident_context.c
@@ -30,7 +30,7 @@
 
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
@@ -143,7 +143,7 @@ tridentCreateContext( const __GLcontextModes *glVisual,
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext( ctx );
-   _ac_CreateContext( ctx );
+   _vbo_CreateContext( ctx );
    _tnl_CreateContext( ctx );
    _swsetup_CreateContext( ctx );
 
@@ -183,7 +183,7 @@ tridentDestroyContext(__DRIcontextPrivate *driContextPriv)
     if (tmesa) {
       _swsetup_DestroyContext( tmesa->glCtx );
       _tnl_DestroyContext( tmesa->glCtx );
-      _ac_DestroyContext( tmesa->glCtx );
+      _vbo_DestroyContext( tmesa->glCtx );
       _swrast_DestroyContext( tmesa->glCtx );
 
       /* free the Mesa context */
diff --git a/src/mesa/drivers/dri/trident/trident_state.c b/src/mesa/drivers/dri/trident/trident_state.c
index 037cce5f862..5303bd422ed 100644
--- a/src/mesa/drivers/dri/trident/trident_state.c
+++ b/src/mesa/drivers/dri/trident/trident_state.c
@@ -26,7 +26,7 @@
  */
 #include "trident_context.h"
 #include "trident_lock.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -468,7 +468,7 @@ tridentDDUpdateState( GLcontext *ctx, GLuint new_state )
 {
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
    TRIDENT_CONTEXT(ctx)->new_gl_state |= new_state;
 }
diff --git a/src/mesa/drivers/dri/unichrome/server/via_dri.c b/src/mesa/drivers/dri/unichrome/server/via_dri.c
index 2fe61e40390..6944bd66f9c 100644
--- a/src/mesa/drivers/dri/unichrome/server/via_dri.c
+++ b/src/mesa/drivers/dri/unichrome/server/via_dri.c
@@ -1179,6 +1179,7 @@ static int viaInitFBDev(DRIDriverContext *ctx)
         pVia->Chipset = VIA_KM400;
         break;
     case PCI_CHIP_VT3204:
+    case PCI_CHIP_VT3344:
         pVia->Chipset = VIA_K8M800;
         break;
     case PCI_CHIP_VT3259:
diff --git a/src/mesa/drivers/dri/unichrome/server/via_regs.h b/src/mesa/drivers/dri/unichrome/server/via_regs.h
index 5c5eecbe831..87e1e9daa92 100644
--- a/src/mesa/drivers/dri/unichrome/server/via_regs.h
+++ b/src/mesa/drivers/dri/unichrome/server/via_regs.h
@@ -44,6 +44,7 @@
 #define PCI_CHIP_VT7205         0x7205
 #define PCI_CHIP_VT3204         0x3108
 #define PCI_CHIP_VT3259         0x3118
+#define PCI_CHIP_VT3344         0x3344
 
 
 #define BIOS_BSIZE              1024
diff --git a/src/mesa/drivers/dri/unichrome/via_context.c b/src/mesa/drivers/dri/unichrome/via_context.c
index 49c24656551..bc5a414df60 100644
--- a/src/mesa/drivers/dri/unichrome/via_context.c
+++ b/src/mesa/drivers/dri/unichrome/via_context.c
@@ -42,7 +42,7 @@
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 #include "tnl/t_pipeline.h"
 
@@ -583,7 +583,7 @@ viaCreateContext(const __GLcontextModes *visual,
     /* Initialize the software rasterizer and helper modules.
      */
     _swrast_CreateContext(ctx);
-    _ac_CreateContext(ctx);
+    _vbo_CreateContext(ctx);
     _tnl_CreateContext(ctx);
     _swsetup_CreateContext(ctx);
 
@@ -708,7 +708,7 @@ viaDestroyContext(__DRIcontextPrivate *driContextPriv)
 	
 	_swsetup_DestroyContext(vmesa->glCtx);
         _tnl_DestroyContext(vmesa->glCtx);
-        _ac_DestroyContext(vmesa->glCtx);
+        _vbo_DestroyContext(vmesa->glCtx);
         _swrast_DestroyContext(vmesa->glCtx);
         /* free the Mesa context */
 	_mesa_destroy_context(vmesa->glCtx);
diff --git a/src/mesa/drivers/dri/unichrome/via_state.c b/src/mesa/drivers/dri/unichrome/via_state.c
index 2c9cfca522e..30b9dc289a7 100644
--- a/src/mesa/drivers/dri/unichrome/via_state.c
+++ b/src/mesa/drivers/dri/unichrome/via_state.c
@@ -40,7 +40,7 @@
 #include "via_3d_reg.h"
 
 #include "swrast/swrast.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "swrast_setup/swrast_setup.h"
 
@@ -1507,7 +1507,7 @@ static void viaInvalidateState(GLcontext *ctx, GLuint newState)
 
     _swrast_InvalidateState(ctx, newState);
     _swsetup_InvalidateState(ctx, newState);
-    _ac_InvalidateState(ctx, newState);
+    _vbo_InvalidateState(ctx, newState);
     _tnl_InvalidateState(ctx, newState);
 }
 
diff --git a/src/mesa/drivers/fbdev/glfbdev.c b/src/mesa/drivers/fbdev/glfbdev.c
index bd9d3ac0ad9..f9d57dbdd81 100644
--- a/src/mesa/drivers/fbdev/glfbdev.c
+++ b/src/mesa/drivers/fbdev/glfbdev.c
@@ -54,7 +54,7 @@
 #include "texformat.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -143,7 +143,7 @@ update_state( GLcontext *ctx, GLuint new_state )
    /* not much to do here - pass it on */
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
 }
 
@@ -776,7 +776,7 @@ glFBDevCreateContext( const GLFBDevVisualPtr visual, GLFBDevContextPtr share )
    /* Create module contexts */
    glctx = (GLcontext *) &ctx->glcontext;
    _swrast_CreateContext( glctx );
-   _ac_CreateContext( glctx );
+   _vbo_CreateContext( glctx );
    _tnl_CreateContext( glctx );
    _swsetup_CreateContext( glctx );
    _swsetup_Wakeup( glctx );
diff --git a/src/mesa/drivers/ggi/ggimesa.c b/src/mesa/drivers/ggi/ggimesa.c
index 47d7f2bdb99..85592d98dab 100644
--- a/src/mesa/drivers/ggi/ggimesa.c
+++ b/src/mesa/drivers/ggi/ggimesa.c
@@ -39,7 +39,7 @@
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "teximage.h"
 #include "texformat.h"
 #include "texstore.h"
@@ -563,7 +563,7 @@ ggi_mesa_context_t ggiMesaCreateContext(ggi_visual_t vis)
         _mesa_enable_sw_extensions(ctx->gl_ctx);
 	
 	_swrast_CreateContext(ctx->gl_ctx);
-	_ac_CreateContext(ctx->gl_ctx);
+	_vbo_CreateContext(ctx->gl_ctx);
 	_tnl_CreateContext(ctx->gl_ctx);
 	_swsetup_CreateContext(ctx->gl_ctx);
 	
diff --git a/src/mesa/drivers/glide/fxdd.c b/src/mesa/drivers/glide/fxdd.c
index ef60602c832..43d3cd94051 100644
--- a/src/mesa/drivers/glide/fxdd.c
+++ b/src/mesa/drivers/glide/fxdd.c
@@ -55,7 +55,7 @@
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 
 
 
@@ -1783,7 +1783,7 @@ fxDDInitFxMesaContext(fxMesaContext fxMesa)
    /* Initialize the software rasterizer and helper modules.
     */
    _swrast_CreateContext(ctx);
-   _ac_CreateContext(ctx);
+   _vbo_CreateContext(ctx);
    _tnl_CreateContext(ctx);
    _swsetup_CreateContext(ctx);
 
@@ -1826,7 +1826,7 @@ fxDDDestroyFxMesaContext(fxMesaContext fxMesa)
 {
    _swsetup_DestroyContext(fxMesa->glCtx);
    _tnl_DestroyContext(fxMesa->glCtx);
-   _ac_DestroyContext(fxMesa->glCtx);
+   _vbo_DestroyContext(fxMesa->glCtx);
    _swrast_DestroyContext(fxMesa->glCtx);
 
    if (fxMesa->state)
@@ -2101,7 +2101,7 @@ fxDDUpdateDDPointers(GLcontext * ctx, GLuint new_state)
    }
 
    _swrast_InvalidateState(ctx, new_state);
-   _ac_InvalidateState(ctx, new_state);
+   _vbo_InvalidateState(ctx, new_state);
    _tnl_InvalidateState(ctx, new_state);
    _swsetup_InvalidateState(ctx, new_state);
 
diff --git a/src/mesa/drivers/osmesa/osmesa.c b/src/mesa/drivers/osmesa/osmesa.c
index 0b027eddec5..c4fc8821e50 100644
--- a/src/mesa/drivers/osmesa/osmesa.c
+++ b/src/mesa/drivers/osmesa/osmesa.c
@@ -41,7 +41,6 @@
 #include "imports.h"
 #include "mtypes.h"
 #include "renderbuffer.h"
-#include "array_cache/acache.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/s_context.h"
@@ -51,6 +50,7 @@
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
 #include "drivers/common/driverfuncs.h"
+#include "vbo/vbo.h"
 
 
 
@@ -110,8 +110,8 @@ osmesa_update_state( GLcontext *ctx, GLuint new_state )
    /* easy - just propogate */
    _swrast_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
 }
 
 
@@ -1261,7 +1261,7 @@ OSMesaCreateContextExt( GLenum format, GLint depthBits, GLint stencilBits,
          TNLcontext *tnl;
 
 	 if (!_swrast_CreateContext( ctx ) ||
-             !_ac_CreateContext( ctx ) ||
+             !_vbo_CreateContext( ctx ) ||
              !_tnl_CreateContext( ctx ) ||
              !_swsetup_CreateContext( ctx )) {
             _mesa_destroy_visual(osmesa->gl_visual);
@@ -1299,7 +1299,7 @@ OSMesaDestroyContext( OSMesaContext osmesa )
    if (osmesa) {
       _swsetup_DestroyContext( &osmesa->mesa );
       _tnl_DestroyContext( &osmesa->mesa );
-      _ac_DestroyContext( &osmesa->mesa );
+      _vbo_DestroyContext( &osmesa->mesa );
       _swrast_DestroyContext( &osmesa->mesa );
 
       _mesa_destroy_visual( osmesa->gl_visual );
diff --git a/src/mesa/drivers/svga/svgamesa15.c b/src/mesa/drivers/svga/svgamesa15.c
index a4c0901e38f..ae5104d0c01 100644
--- a/src/mesa/drivers/svga/svgamesa15.c
+++ b/src/mesa/drivers/svga/svgamesa15.c
@@ -1,4 +1,4 @@
-/* $Id: svgamesa15.c,v 1.12 2006/11/01 19:35:23 brianp Exp $ */
+/* $Id: svgamesa15.c,v 1.11.36.1 2006/11/02 12:02:17 alanh Exp $ */
 
 /*
  * Mesa 3-D graphics library
diff --git a/src/mesa/drivers/svga/svgamesa16.c b/src/mesa/drivers/svga/svgamesa16.c
index 999c6331c9c..a59937bfb4b 100644
--- a/src/mesa/drivers/svga/svgamesa16.c
+++ b/src/mesa/drivers/svga/svgamesa16.c
@@ -1,4 +1,4 @@
-/* $Id: svgamesa16.c,v 1.12 2006/11/01 19:35:23 brianp Exp $ */
+/* $Id: svgamesa16.c,v 1.11.36.1 2006/11/02 12:02:17 alanh Exp $ */
 
 /*
  * Mesa 3-D graphics library
diff --git a/src/mesa/drivers/svga/svgamesa24.c b/src/mesa/drivers/svga/svgamesa24.c
index 93c84167ebb..dd15bf38db2 100644
--- a/src/mesa/drivers/svga/svgamesa24.c
+++ b/src/mesa/drivers/svga/svgamesa24.c
@@ -1,4 +1,4 @@
-/* $Id: svgamesa24.c,v 1.13 2006/11/01 19:35:23 brianp Exp $ */
+/* $Id: svgamesa24.c,v 1.12.36.1 2006/11/02 12:02:17 alanh Exp $ */
 
 /*
  * Mesa 3-D graphics library
diff --git a/src/mesa/drivers/svga/svgamesa32.c b/src/mesa/drivers/svga/svgamesa32.c
index 5a3a6776ce7..4da18795d8c 100644
--- a/src/mesa/drivers/svga/svgamesa32.c
+++ b/src/mesa/drivers/svga/svgamesa32.c
@@ -1,4 +1,4 @@
-/* $Id: svgamesa32.c,v 1.13 2006/11/01 19:35:23 brianp Exp $ */
+/* $Id: svgamesa32.c,v 1.12.36.1 2006/11/02 12:02:17 alanh Exp $ */
 
 /*
  * Mesa 3-D graphics library
diff --git a/src/mesa/drivers/svga/svgamesa8.c b/src/mesa/drivers/svga/svgamesa8.c
index e3f5eaebf6a..4264fcd9599 100644
--- a/src/mesa/drivers/svga/svgamesa8.c
+++ b/src/mesa/drivers/svga/svgamesa8.c
@@ -1,4 +1,4 @@
-/* $Id: svgamesa8.c,v 1.10 2006/11/01 19:35:23 brianp Exp $ */
+/* $Id: svgamesa8.c,v 1.9.10.1 2006/11/02 12:02:17 alanh Exp $ */
 
 /*
  * Mesa 3-D graphics library
diff --git a/src/mesa/drivers/windows/gdi/mesa.def b/src/mesa/drivers/windows/gdi/mesa.def
index f0b275730ca..8d654281215 100644
--- a/src/mesa/drivers/windows/gdi/mesa.def
+++ b/src/mesa/drivers/windows/gdi/mesa.def
@@ -859,9 +859,9 @@ EXPORTS
 	wglGetExtensionsStringARB
 ;
 ; Mesa internals - mostly for OSMESA
-	_ac_CreateContext
-	_ac_DestroyContext
-	_ac_InvalidateState
+	_vbo_CreateContext
+	_vbo_DestroyContext
+	_vbo_InvalidateState
 	_glapi_check_multithread
 	_glapi_get_context
 	_glapi_get_proc_address
diff --git a/src/mesa/drivers/windows/gdi/wmesa.c b/src/mesa/drivers/windows/gdi/wmesa.c
index 8c3229ab14c..078b590af87 100644
--- a/src/mesa/drivers/windows/gdi/wmesa.c
+++ b/src/mesa/drivers/windows/gdi/wmesa.c
@@ -11,7 +11,7 @@
 #include "framebuffer.h"
 #include "renderbuffer.h"
 #include "drivers/common/driverfuncs.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "tnl/tnl.h"
@@ -1125,7 +1125,7 @@ static void wmesa_update_state(GLcontext *ctx, GLuint new_state)
 {
     _swrast_InvalidateState(ctx, new_state);
     _swsetup_InvalidateState(ctx, new_state);
-    _ac_InvalidateState(ctx, new_state);
+    _vbo_InvalidateState(ctx, new_state);
     _tnl_InvalidateState(ctx, new_state);
 
     /* TODO - This code is not complete yet because I 
@@ -1242,7 +1242,7 @@ WMesaContext WMesaCreateContext(HDC hDC,
   
     /* Initialize the software rasterizer and helper modules. */
     if (!_swrast_CreateContext(ctx) ||
-        !_ac_CreateContext(ctx) ||
+        !_vbo_CreateContext(ctx) ||
         !_tnl_CreateContext(ctx) ||
 	!_swsetup_CreateContext(ctx)) {
 	_mesa_free_context_data(ctx);
@@ -1285,7 +1285,7 @@ void WMesaDestroyContext( WMesaContext pwc )
     
     _swsetup_DestroyContext(ctx);
     _tnl_DestroyContext(ctx);
-    _ac_DestroyContext(ctx);
+    _vbo_DestroyContext(ctx);
     _swrast_DestroyContext(ctx);
     
     _mesa_free_context_data(ctx);
diff --git a/src/mesa/drivers/windows/gldirect/dglcontext.c b/src/mesa/drivers/windows/gldirect/dglcontext.c
index 5dafe14518d..4ad7a76e67f 100644
--- a/src/mesa/drivers/windows/gldirect/dglcontext.c
+++ b/src/mesa/drivers/windows/gldirect/dglcontext.c
@@ -1453,7 +1453,7 @@ SkipPrimaryCreate:
 #ifdef _USE_GLD3_WGL
 	// Init Mesa internals
 	_swrast_CreateContext( lpCtx->glCtx );
-	_ac_CreateContext( lpCtx->glCtx );
+	_vbo_CreateContext( lpCtx->glCtx );
 	_tnl_CreateContext( lpCtx->glCtx );
 	_swsetup_CreateContext( lpCtx->glCtx );
 
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_driver_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_driver_dx7.c
index cad299540d7..d5fa6428006 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_driver_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_driver_dx7.c
@@ -52,7 +52,7 @@
 #include "texformat.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
@@ -868,7 +868,7 @@ void gld_update_state_DX7(
 		return;
 
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 
 	// SetupIndex will be used in the pipelines for choosing setup function
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_ext_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_ext_dx7.c
index 6be41a80dd3..ba60980bbe8 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_ext_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_ext_dx7.c
@@ -58,7 +58,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
diff --git a/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c b/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
index 5da25003c08..c99ba0bba55 100644
--- a/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
+++ b/src/mesa/drivers/windows/gldirect/dx7/gld_primitive_dx7.c
@@ -54,7 +54,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_driver_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_driver_dx8.c
index 0a6d9f8555c..7afa9190cdf 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_driver_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_driver_dx8.c
@@ -52,7 +52,7 @@
 #include "texformat.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
@@ -850,7 +850,7 @@ void gld_update_state_DX8(
 		return;
 
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 
 	// SetupIndex will be used in the pipelines for choosing setup function
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_ext_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_ext_dx8.c
index 108f12a9d16..b51bba9b3ca 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_ext_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_ext_dx8.c
@@ -58,7 +58,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
diff --git a/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c b/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
index 700b5200862..a5b5462f039 100644
--- a/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
+++ b/src/mesa/drivers/windows/gldirect/dx8/gld_primitive_dx8.c
@@ -54,7 +54,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_driver_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_driver_dx9.c
index 1b01cb1f7f3..c191564d6eb 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_driver_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_driver_dx9.c
@@ -52,7 +52,7 @@
 #include "texformat.h"
 #include "teximage.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
@@ -876,7 +876,7 @@ void gld_update_state_DX9(
 		return;
 
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 
 	// SetupIndex will be used in the pipelines for choosing setup function
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_ext_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_ext_dx9.c
index 745c987602b..e8c73a6ff80 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_ext_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_ext_dx9.c
@@ -58,7 +58,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
 #include "tnl/tnl.h"
diff --git a/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c b/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
index 65fd821276e..403a9d5f86e 100644
--- a/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
+++ b/src/mesa/drivers/windows/gldirect/dx9/gld_primitive_dx9.c
@@ -54,7 +54,7 @@
 #include "mtypes.h"
 #include "texformat.h"
 #include "texstore.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast_setup/ss_context.h"
diff --git a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
index 0f11b4fe513..b590dc795a0 100644
--- a/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
+++ b/src/mesa/drivers/windows/gldirect/mesasw/gld_wgl_mesasw.c
@@ -54,7 +54,7 @@
 #include "texformat.h"
 #include "texstore.h"
 #include "teximage.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
 #include "swrast/s_context.h"
@@ -1321,7 +1321,7 @@ static void wmesa_update_state(
 {
     _swrast_InvalidateState( ctx, new_state );
 	_swsetup_InvalidateState( ctx, new_state );
-	_ac_InvalidateState( ctx, new_state );
+	_vbo_InvalidateState( ctx, new_state );
 	_tnl_InvalidateState( ctx, new_state );
 }
 
@@ -1631,7 +1631,7 @@ BOOL gldInitialiseMesa_MesaSW(
 	_mesa_enable_1_3_extensions(ctx);
 	
 //	_swrast_CreateContext( ctx );
-//	_ac_CreateContext( ctx );
+//	_vbo_CreateContext( ctx );
 //	_tnl_CreateContext( ctx );
 //	_swsetup_CreateContext( ctx );
 	
diff --git a/src/mesa/drivers/windows/icd/mesa.def b/src/mesa/drivers/windows/icd/mesa.def
index 1826192d307..465b380a0cb 100644
--- a/src/mesa/drivers/windows/icd/mesa.def
+++ b/src/mesa/drivers/windows/icd/mesa.def
@@ -24,9 +24,9 @@ EXPORTS
 
 ;
 ; Mesa internals - mostly for OSMESA
-	_ac_CreateContext
-	_ac_DestroyContext
-	_ac_InvalidateState
+	_vbo_CreateContext
+	_vbo_DestroyContext
+	_vbo_InvalidateState
 	_glapi_get_context
 	_glapi_get_proc_address
 	_mesa_buffer_data
diff --git a/src/mesa/drivers/x11/glxapi.c b/src/mesa/drivers/x11/glxapi.c
index 7945f1c5b06..973f3940451 100644
--- a/src/mesa/drivers/x11/glxapi.c
+++ b/src/mesa/drivers/x11/glxapi.c
@@ -241,7 +241,7 @@ glXGetCurrentContext(void)
 #if defined(GLX_USE_TLS)
    return CurrentContext;
 #elif defined(THREADS)
-   return _glthread_GetTSD(&ContextTSD);
+   return (GLXContext) _glthread_GetTSD(&ContextTSD);
 #else
    return CurrentContext;
 #endif
diff --git a/src/mesa/drivers/x11/xm_api.c b/src/mesa/drivers/x11/xm_api.c
index b50c4757afc..b0ef422de2e 100644
--- a/src/mesa/drivers/x11/xm_api.c
+++ b/src/mesa/drivers/x11/xm_api.c
@@ -74,7 +74,7 @@
 #include "renderbuffer.h"
 #include "swrast/swrast.h"
 #include "swrast_setup/swrast_setup.h"
-#include "array_cache/acache.h"
+#include "vbo/vbo.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
 #include "tnl/t_pipeline.h"
@@ -1545,6 +1545,13 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
     _mesa_enable_extension(mesaCtx, "GL_EXT_timer_query");
 #endif
 
+#ifdef XFree86Server
+   /* If we're running in the X server, do bounds checking to prevent
+    * segfaults and server crashes!
+    */
+   mesaCtx->Const.CheckArrayBounds = GL_TRUE;
+#endif
+
    /* finish up xmesa context initializations */
    c->swapbytes = CHECK_BYTE_ORDER(v) ? GL_FALSE : GL_TRUE;
    c->xm_visual = v;
@@ -1555,7 +1562,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
    /* Initialize the software rasterizer and helper modules.
     */
    if (!_swrast_CreateContext( mesaCtx ) ||
-       !_ac_CreateContext( mesaCtx ) ||
+       !_vbo_CreateContext( mesaCtx ) ||
        !_tnl_CreateContext( mesaCtx ) ||
        !_swsetup_CreateContext( mesaCtx )) {
       _mesa_free_context_data(&c->mesa);
@@ -1587,7 +1594,7 @@ void XMesaDestroyContext( XMesaContext c )
    _swsetup_DestroyContext( mesaCtx );
    _swrast_DestroyContext( mesaCtx );
    _tnl_DestroyContext( mesaCtx );
-   _ac_DestroyContext( mesaCtx );
+   _vbo_DestroyContext( mesaCtx );
    _mesa_free_context_data( mesaCtx );
    _mesa_free( c );
 }
diff --git a/src/mesa/drivers/x11/xm_dd.c b/src/mesa/drivers/x11/xm_dd.c
index d272630d18a..dbac3b8cdb6 100644
--- a/src/mesa/drivers/x11/xm_dd.c
+++ b/src/mesa/drivers/x11/xm_dd.c
@@ -47,7 +47,6 @@
 #include "texstore.h"
 #include "texformat.h"
 #include "xmesaP.h"
-#include "array_cache/acache.h"
 #include "swrast/swrast.h"
 #include "swrast/s_context.h"
 #include "swrast_setup/swrast_setup.h"
@@ -829,8 +828,8 @@ xmesa_update_state( GLcontext *ctx, GLbitfield new_state )
     * modules.  The X11 driver has no internal GL-dependent state.
     */
    _swrast_InvalidateState( ctx, new_state );
-   _ac_InvalidateState( ctx, new_state );
    _tnl_InvalidateState( ctx, new_state );
+   _vbo_InvalidateState( ctx, new_state );
    _swsetup_InvalidateState( ctx, new_state );
 
    if (ctx->DrawBuffer->Name != 0)
author	Nian Wu <nian@graphics.(none)>	2007-02-25 09:40:28 -0800
committer	Nian Wu <nian@graphics.(none)>	2007-02-25 09:40:28 -0800
commit	675f7f627bec92315bf168a9c872ffc05f88c69c (patch)
tree	19c8f9cc509dc7ce73a0709832f34bf13a2c7425 /src/mesa/drivers
parent	51bfb8fc8c78bb066d24e6ecbc20f00af7210386 (diff)
parent	c080123998f28d9317331aec7ddfcd1074b29daf (diff)