Import vtx-0-2-branch

10 files changed, 189 insertions, 218 deletions

diff --git a/src/mesa/drivers/dri/radeon/Makefile.X11 b/src/mesa/drivers/dri/radeon/Makefile.X11
index 19dabce262c..fd379e9a0d6 100644
--- a/src/mesa/drivers/dri/radeon/Makefile.X11
+++ b/src/mesa/drivers/dri/radeon/Makefile.X11
@@ -1,4 +1,3 @@
-# $Id: Makefile.X11,v 1.3 2003/10/21 06:05:49 jonsmirl Exp $
 
 # Mesa 3-D graphics library
 # Version:  5.0
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index 5604d9c8c21..02bfb23ca59 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -165,11 +165,11 @@ static const char * const card_extensions[] =
     NULL
 };
 
-extern const struct gl_pipeline_stage _radeon_texrect_stage;
-extern const struct gl_pipeline_stage _radeon_render_stage;
-extern const struct gl_pipeline_stage _radeon_tcl_stage;
+extern const struct tnl_pipeline_stage _radeon_texrect_stage;
+extern const struct tnl_pipeline_stage _radeon_render_stage;
+extern const struct tnl_pipeline_stage _radeon_tcl_stage;
 
-static const struct gl_pipeline_stage *radeon_pipeline[] = {
+static const struct tnl_pipeline_stage *radeon_pipeline[] = {
 
    /* Try and go straight to t&l
     */
@@ -412,6 +412,9 @@ radeonCreateContext( const __GLcontextModes *glVisual,
    radeonInitState( rmesa );
    radeonInitSwtcl( ctx );
 
+   _mesa_vector4f_alloc( &rmesa->tcl.ObjClean, 0, 
+			 ctx->Const.MaxArrayLockSize, 32 );
+
    fthrottle_mode = driQueryOptioni(&rmesa->optionCache, "fthrottle_mode");
    rmesa->iw.irq_seq = -1;
    rmesa->irqsEmitted = 0;
@@ -505,6 +508,8 @@ void radeonDestroyContext( __DRIcontextPrivate *driContextPriv )
       rmesa->glCtx->DriverCtx = NULL;
       _mesa_destroy_context( rmesa->glCtx );
 
+      _mesa_vector4f_free( &rmesa->tcl.ObjClean );
+
       if (rmesa->state.scissor.pClipRects) {
 	 FREE(rmesa->state.scissor.pClipRects);
 	 rmesa->state.scissor.pClipRects = 0;
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.h b/src/mesa/drivers/dri/radeon/radeon_context.h
index 69808837ea4..b0c3c1919e1 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -57,6 +57,8 @@ typedef struct radeon_context *radeonContextPtr;
 #include "radeon_screen.h"
 #include "mm.h"
 
+#include "math/m_vector.h"
+
 /* Flags for software fallback cases */
 /* See correponding strings in radeon_swtcl.c */
 #define RADEON_FALLBACK_TEXTURE		0x0001
@@ -503,6 +505,11 @@ struct radeon_tcl_info {
    GLint last_offset;
    GLuint hw_primitive;
 
+   /* Temporary for cases where incoming vertex data is incompatible
+    * with maos code.
+    */
+   GLvector4f ObjClean;
+
    struct radeon_dma_region *aos_components[8];
    GLuint nr_aos_components;
 
@@ -699,12 +706,6 @@ struct radeon_context {
    GLuint Fallback;
    GLuint NewGLState;
 
-   
-   /* Temporaries for translating away float colors:
-    */
-   struct gl_client_array UbyteColor;
-   struct gl_client_array UbyteSecondaryColor;
-
    /* Vertex buffers
     */
    struct radeon_ioctl ioctl;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
index b379bad985f..c16234a9437 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
@@ -39,13 +39,14 @@ static void TAG(emit)( GLcontext *ctx,
 		       void *dest )
 {
    LOCALVARS
-   struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
+      struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    GLuint (*tc0)[4], (*tc1)[4], (*tc2)[4];
+   GLfloat (*col)[4], (*spec)[4];
    GLfloat (*fog)[4];
    GLuint (*norm)[4];
-   GLubyte (*col)[4], (*spec)[4];
    GLuint tc0_stride, tc1_stride, col_stride, spec_stride, fog_stride;
    GLuint tc2_stride, norm_stride;
+   GLuint fill_tex = 0;
    GLuint (*coord)[4];
    GLuint coord_stride; /* object coordinates */
    GLubyte dummy[4];
@@ -56,36 +57,20 @@ static void TAG(emit)( GLcontext *ctx,
    if (RADEON_DEBUG & DEBUG_VERTS)
       fprintf(stderr, "%s\n", __FUNCTION__); 
 
-   /* The vertex code expects Obj to be clean to element 3.  To fix
-    * this, add more vertex code (for obj-2, obj-3) or preferably move
-    * to maos.  
-    */
-   if (VB->ObjPtr->size < 3) {
-      if (VB->ObjPtr->flags & VEC_NOT_WRITEABLE) {
-	 VB->import_data( ctx, VERT_BIT_POS, VEC_NOT_WRITEABLE );
-      }
-      _mesa_vector4f_clean_elem( VB->ObjPtr, VB->Count, 2 );
-   }
-
-   if (DO_W && VB->ObjPtr->size < 4) {
-      if (VB->ObjPtr->flags & VEC_NOT_WRITEABLE) {
-	 VB->import_data( ctx, VERT_BIT_POS, VEC_NOT_WRITEABLE );
-      }
-      _mesa_vector4f_clean_elem( VB->ObjPtr, VB->Count, 3 );
-   }
-
    coord = (GLuint (*)[4])VB->ObjPtr->data;
    coord_stride = VB->ObjPtr->stride;
 
    if (DO_TEX2) {
-      const GLuint t2 = GET_TEXSOURCE(2);
-      tc2 = (GLuint (*)[4])VB->TexCoordPtr[t2]->data;
-      tc2_stride = VB->TexCoordPtr[t2]->stride;
-      if (DO_PTEX && VB->TexCoordPtr[t2]->size < 4) {
-	 if (VB->TexCoordPtr[t2]->flags & VEC_NOT_WRITEABLE) {
-	    VB->import_data( ctx, VERT_BIT_TEX2, VEC_NOT_WRITEABLE );
+      if (VB->TexCoordPtr[2]) {
+	 const GLuint t2 = GET_TEXSOURCE(2);
+	 tc2 = (GLuint (*)[4])VB->TexCoordPtr[t2]->data;
+	 tc2_stride = VB->TexCoordPtr[t2]->stride;
+	 if (DO_PTEX && VB->TexCoordPtr[t2]->size < 4) {
+	    fill_tex |= (1<<2);
 	 }
-	 _mesa_vector4f_clean_elem( VB->TexCoordPtr[t2], VB->Count, 3 );
+      } else {
+	 tc2 = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_TEX2];
+	 tc2_stride = 0;
       }
    }
 
@@ -95,13 +80,10 @@ static void TAG(emit)( GLcontext *ctx,
 	 tc1 = (GLuint (*)[4])VB->TexCoordPtr[t1]->data;
 	 tc1_stride = VB->TexCoordPtr[t1]->stride;
 	 if (DO_PTEX && VB->TexCoordPtr[t1]->size < 4) {
-	    if (VB->TexCoordPtr[t1]->flags & VEC_NOT_WRITEABLE) {
-	       VB->import_data( ctx, VERT_BIT_TEX1, VEC_NOT_WRITEABLE );
-	    }
-	    _mesa_vector4f_clean_elem( VB->TexCoordPtr[t1], VB->Count, 3 );
+	    fill_tex |= (1<<1);
 	 }
       } else {
-	 tc1 = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_TEX1]; /* could be anything, really */
+	 tc1 = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_TEX1];
 	 tc1_stride = 0;
       }
    }
@@ -112,13 +94,10 @@ static void TAG(emit)( GLcontext *ctx,
 	 tc0_stride = VB->TexCoordPtr[t0]->stride;
 	 tc0 = (GLuint (*)[4])VB->TexCoordPtr[t0]->data;
 	 if (DO_PTEX && VB->TexCoordPtr[t0]->size < 4) {
-	    if (VB->TexCoordPtr[t0]->flags & VEC_NOT_WRITEABLE) {
-	       VB->import_data( ctx, VERT_BIT_TEX0, VEC_NOT_WRITEABLE );
-	    }
-	    _mesa_vector4f_clean_elem( VB->TexCoordPtr[t0], VB->Count, 3 );
+	    fill_tex |= (1<<0);
 	 }
       } else {
-	 tc0 = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_TEX0]; /* could be anything, really */
+	 tc0 = (GLuint (*)[4])&ctx->Current.Attrib[VERT_ATTRIB_TEX0];
 	 tc0_stride = 0;
       }
 	 
@@ -136,28 +115,20 @@ static void TAG(emit)( GLcontext *ctx,
 
    if (DO_RGBA) {
       if (VB->ColorPtr[0]) {
-	 /* This is incorrect when colormaterial is enabled:
-	  */
-	 if (VB->ColorPtr[0]->Type != GL_UNSIGNED_BYTE) {
-	    if (0) fprintf(stderr, "IMPORTING FLOAT COLORS\n");
-	    IMPORT_FLOAT_COLORS( ctx );
-	 }
-	 col = (GLubyte (*)[4])VB->ColorPtr[0]->Ptr;
-	 col_stride = VB->ColorPtr[0]->StrideB;
+	 col = VB->ColorPtr[0]->data;
+	 col_stride = VB->ColorPtr[0]->stride;
       } else {
-	 col = &dummy; /* any old memory is fine */
+	 col = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR0];
 	 col_stride = 0;
       }
    }
 
    if (DO_SPEC) {
       if (VB->SecondaryColorPtr[0]) {
-	 if (VB->SecondaryColorPtr[0]->Type != GL_UNSIGNED_BYTE)
-	    IMPORT_FLOAT_SPEC_COLORS( ctx );
-	 spec = (GLubyte (*)[4])VB->SecondaryColorPtr[0]->Ptr;
-	 spec_stride = VB->SecondaryColorPtr[0]->StrideB;
+	 spec = VB->SecondaryColorPtr[0]->data;
+	 spec_stride = VB->SecondaryColorPtr[0]->stride;
       } else {
-	 spec = &dummy;
+	 spec = (GLfloat (*)[4])ctx->Current.Attrib[VERT_ATTRIB_COLOR1];
 	 spec_stride = 0;
       }
    }
@@ -173,33 +144,33 @@ static void TAG(emit)( GLcontext *ctx,
    }
    
    
-   if (VB->importable_data) {
-      if (start) {
-	 coord =  (GLuint (*)[4])((GLubyte *)coord + start * coord_stride);
-	 if (DO_TEX0)
-	    tc0 =  (GLuint (*)[4])((GLubyte *)tc0 + start * tc0_stride);
-	 if (DO_TEX1) 
-	    tc1 =  (GLuint (*)[4])((GLubyte *)tc1 + start * tc1_stride);
-	 if (DO_TEX2) 
-	    tc2 =  (GLuint (*)[4])((GLubyte *)tc2 + start * tc2_stride);
-	 if (DO_NORM) 
-	    norm =  (GLuint (*)[4])((GLubyte *)norm + start * norm_stride);
-	 if (DO_RGBA) 
-	    STRIDE_4UB(col, start * col_stride);
-	 if (DO_SPEC)
-	    STRIDE_4UB(spec, start * spec_stride);
-	 if (DO_FOG)
-	    fog =  (GLfloat (*)[4])((GLubyte *)fog + start * fog_stride);
-      }
+   if (start) {
+      coord =  (GLuint (*)[4])((GLubyte *)coord + start * coord_stride);
+      if (DO_TEX0)
+	 tc0 =  (GLuint (*)[4])((GLubyte *)tc0 + start * tc0_stride);
+      if (DO_TEX1) 
+	 tc1 =  (GLuint (*)[4])((GLubyte *)tc1 + start * tc1_stride);
+      if (DO_TEX2) 
+	 tc2 =  (GLuint (*)[4])((GLubyte *)tc2 + start * tc2_stride);
+      if (DO_NORM) 
+	 norm =  (GLuint (*)[4])((GLubyte *)norm + start * norm_stride);
+      if (DO_RGBA) 
+	 STRIDE_4F(col, start * col_stride);
+      if (DO_SPEC)
+	 STRIDE_4F(spec, start * spec_stride);
+      if (DO_FOG)
+	 STRIDE_4F(fog, start * fog_stride);
+   }
 
+
+   {
       for (i=start; i < end; i++) {
+	 
 	 v[0].ui = coord[0][0];
 	 v[1].ui = coord[0][1];
 	 v[2].ui = coord[0][2];
-	 if (TCL_DEBUG) fprintf(stderr, "%d: %.2f %.2f %.2f ", i, v[0].f, v[1].f, v[2].f);
 	 if (DO_W) {
 	    v[3].ui = coord[0][3];
-	    if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[3].f);
 	    v += 4;
 	 } 
 	 else
@@ -210,26 +181,27 @@ static void TAG(emit)( GLcontext *ctx,
 	    v[0].ui = norm[0][0];
 	    v[1].ui = norm[0][1];
 	    v[2].ui = norm[0][2];
-	    if (TCL_DEBUG) fprintf(stderr, "norm: %.2f %.2f %.2f ", v[0].f, v[1].f, v[2].f);
 	    v += 3;
 	    norm =  (GLuint (*)[4])((GLubyte *)norm +  norm_stride);
 	 }
 	 if (DO_RGBA) {
-	    v[0].ui = LE32_TO_CPU(*(GLuint *)&col[0]);
-	    STRIDE_4UB(col, col_stride);
-	    if (TCL_DEBUG) fprintf(stderr, "%x ", v[0].ui);
+	    UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.red, col[0][0]);
+	    UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.green, col[0][1]);
+	    UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.blue, col[0][2]);
+	    UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.alpha, col[0][3]);
+	    STRIDE_4F(col, col_stride);
 	    v++;
 	 }
 	 if (DO_SPEC || DO_FOG) {
 	    if (DO_SPEC) {
-	       v[0].specular.red   = spec[0][0];
-	       v[0].specular.green = spec[0][1];
-	       v[0].specular.blue  = spec[0][2];
-	       STRIDE_4UB(spec, spec_stride);
+	       UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.red, spec[0][0]);
+	       UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.green, spec[0][1]);
+	       UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.blue, spec[0][2]);
+	       STRIDE_4F(spec, spec_stride);
 	    }
 	    if (DO_FOG) {
-	       v[0].specular.alpha = fog[0][0] * 255.0;
-               fog = (GLfloat (*)[4])((GLubyte *)fog + fog_stride);
+	       UNCLAMPED_FLOAT_TO_UBYTE(v[0].rgba.alpha, fog[0][0]);
+	       fog = (GLfloat (*)[4])((GLubyte *)fog + fog_stride);
 	    }
 	    if (TCL_DEBUG) fprintf(stderr, "%x ", v[0].ui);
 	    v++;
@@ -239,7 +211,10 @@ static void TAG(emit)( GLcontext *ctx,
 	    v[1].ui = tc0[0][1];
 	    if (TCL_DEBUG) fprintf(stderr, "t0: %.2f %.2f ", v[0].f, v[1].f);
 	    if (DO_PTEX) {
-	       v[2].ui = tc0[0][3];
+	       if (fill_tex & (1<<0))
+		  v[2].f = 1.0;
+	       else
+		  v[2].ui = tc0[0][3];
 	       if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[2].f);
 	       v += 3;
 	    } 
@@ -252,7 +227,10 @@ static void TAG(emit)( GLcontext *ctx,
 	    v[1].ui = tc1[0][1];
 	    if (TCL_DEBUG) fprintf(stderr, "t1: %.2f %.2f ", v[0].f, v[1].f);
 	    if (DO_PTEX) {
-	       v[2].ui = tc1[0][3];
+	       if (fill_tex & (1<<1))
+		  v[2].f = 1.0;
+	       else
+		  v[2].ui = tc1[0][3];
 	       if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[2].f);
 	       v += 3;
 	    } 
@@ -264,7 +242,10 @@ static void TAG(emit)( GLcontext *ctx,
 	    v[0].ui = tc2[0][0];
 	    v[1].ui = tc2[0][1];
 	    if (DO_PTEX) {
-	       v[2].ui = tc2[0][3];
+	       if (fill_tex & (1<<2))
+		  v[2].f = 1.0;
+	       else
+		  v[2].ui = tc2[0][3];
 	       v += 3;
 	    } 
 	    else
@@ -273,71 +254,6 @@ static void TAG(emit)( GLcontext *ctx,
 	 } 
 	 if (TCL_DEBUG) fprintf(stderr, "\n");
       }
-   } else {
-      for (i=start; i < end; i++) {
-	 v[0].ui = coord[i][0];
-	 v[1].ui = coord[i][1];
-	 v[2].ui = coord[i][2];
-	 if (DO_W) {
-	    v[3].ui = coord[i][3];
-	    v += 4;
-	 } 
-	 else
-	    v += 3;
-
-	 if (DO_NORM) {
-	    v[0].ui = norm[i][0];
-	    v[1].ui = norm[i][1];
-	    v[2].ui = norm[i][2];
-	    v += 3;
-	 }
-	 if (DO_RGBA) {
-	    v[0].ui = LE32_TO_CPU(*(GLuint *)&col[i]);
-	    v++;
-	 }
-	 if (DO_SPEC || DO_FOG) {
-	    if (DO_SPEC) {
-	       v[0].specular.red   = spec[i][0];
-	       v[0].specular.green = spec[i][1];
-	       v[0].specular.blue  = spec[i][2];
-	    }
-	    if (DO_FOG) {
-               GLfloat *f = (GLfloat *) ((GLubyte *)fog + fog_stride);
-               v[0].specular.alpha = *f * 255.0;
-	    }
-	    v++;
-	 }
-	 if (DO_TEX0) {
-	    v[0].ui = tc0[i][0];
-	    v[1].ui = tc0[i][1];
-	    if (DO_PTEX) {
-	       v[2].ui = tc0[i][3];
-	       v += 3;
-	    } 
-	    else
-	       v += 2;
-	 }
-	 if (DO_TEX1) {
-	    v[0].ui = tc1[i][0];
-	    v[1].ui = tc1[i][1];
-	    if (DO_PTEX) {
-	       v[2].ui = tc1[i][3];
-	       v += 3;
-	    } 
-	    else
-	       v += 2;
-	 } 
-	 if (DO_TEX2) {
-	    v[0].ui = tc2[i][0];
-	    v[1].ui = tc2[i][1];
-	    if (DO_PTEX) {
-	       v[2].ui = tc2[i][3];
-	       v += 3;
-	    } 
-	    else
-	       v += 2;
-	 } 
-      }
    }
 }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index b32fd624fa3..5e63e6e880f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -40,7 +40,6 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "array_cache/acache.h"
 #include "tnl/tnl.h"
 #include "tnl/t_pipeline.h"
-#include "tnl/t_imm_debug.h"
 
 #include "radeon_context.h"
 #include "radeon_state.h"
@@ -53,7 +52,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 
 #define RADEON_TCL_MAX_SETUP 13
 
-union emit_union { float f; GLuint ui; radeon_color_t specular; };
+union emit_union { float f; GLuint ui; radeon_color_t rgba; };
 
 static struct {
    void   (*emit)( GLcontext *, GLuint, GLuint, void * );
@@ -308,6 +307,41 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 			      setup_tab[i].vertex_size * 4, 
 			      4);
 
+   /* The vertex code expects Obj to be clean to element 3.  To fix
+    * this, add more vertex code (for obj-2, obj-3) or preferably move
+    * to maos.  
+    */
+   if (VB->ObjPtr->size < 3 || 
+       (VB->ObjPtr->size == 3 && 
+	(setup_tab[i].vertex_format & RADEON_CP_VC_FRMT_W0))) {
+
+      _math_trans_4f( rmesa->tcl.ObjClean.data,
+		      VB->ObjPtr->data,
+		      VB->ObjPtr->stride,
+		      GL_FLOAT,
+		      VB->ObjPtr->size,
+		      0,
+		      VB->Count );
+
+      switch (VB->ObjPtr->size) {
+      case 1:
+	    _mesa_vector4f_clean_elem(&rmesa->tcl.ObjClean, VB->Count, 1);
+      case 2:
+	    _mesa_vector4f_clean_elem(&rmesa->tcl.ObjClean, VB->Count, 2);
+      case 3:
+	 if (setup_tab[i].vertex_format & RADEON_CP_VC_FRMT_W0) {
+	    _mesa_vector4f_clean_elem(&rmesa->tcl.ObjClean, VB->Count, 3);
+	 }
+      case 4:
+      default:
+	 break;
+      }
+
+      VB->ObjPtr = &rmesa->tcl.ObjClean;
+   }
+
+
+
    setup_tab[i].emit( ctx, 0, VB->Count, 
 		      rmesa->tcl.indexed_verts.address + 
 		      rmesa->tcl.indexed_verts.start );
diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c
index 7b1bbe75fd2..c15f3417bb9 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state.c
@@ -2115,10 +2115,30 @@ static void radeonInvalidateState( GLcontext *ctx, GLuint new_state )
    radeonVtxfmtInvalidate( ctx );
 }
 
+
+/* A hack.  Need a faster way to find this out.
+ */
+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; 
+	i++)
+      if (tnl->vb.AttribPtr[i] &&
+	  tnl->vb.AttribPtr[i]->stride)
+	 return GL_TRUE;
+
+   return GL_FALSE;
+}
+      
+
 static void radeonWrapRunPipeline( GLcontext *ctx )
 {
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
+   GLboolean has_material;
 
    if (0)
       fprintf(stderr, "%s, newstate: %x\n", __FUNCTION__, rmesa->NewGLState);
@@ -2128,7 +2148,9 @@ static void radeonWrapRunPipeline( GLcontext *ctx )
    if (rmesa->NewGLState)
       radeonValidateState( ctx );
 
-   if (tnl->vb.Material) {
+   has_material = (ctx->Light.Enabled && check_material( ctx ));
+
+   if (has_material) {
       TCL_FALLBACK( ctx, RADEON_TCL_FALLBACK_MATERIAL, GL_TRUE );
    }
 
@@ -2136,7 +2158,7 @@ static void radeonWrapRunPipeline( GLcontext *ctx )
     */ 
    _tnl_run_pipeline( ctx );
 
-   if (tnl->vb.Material) {
+   if (has_material) {
       TCL_FALLBACK( ctx, RADEON_TCL_FALLBACK_MATERIAL, GL_FALSE );
       radeonUpdateMaterial( ctx ); /* not needed any more? */
    }
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index 89db677392d..157a863e328 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -44,8 +44,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "math/m_translate.h"
 #include "tnl/tnl.h"
 #include "tnl/t_context.h"
-#include "tnl/t_imm_exec.h"
 #include "tnl/t_pipeline.h"
+#include "tnl/t_vtx_api.h"	/* for _tnl_FlushVertices */
 
 #include "radeon_context.h"
 #include "radeon_ioctl.h"
@@ -521,7 +521,7 @@ static void VERT_FALLBACK( GLcontext *ctx,
    tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK );
    tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 );
    tnl->Driver.Render.PrimTabVerts[flags&PRIM_MODE_MASK]( ctx, start, count, flags );
-   RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_CLIP;
+   RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_POS;
 }
 
 static void ELT_FALLBACK( GLcontext *ctx,
@@ -533,7 +533,7 @@ static void ELT_FALLBACK( GLcontext *ctx,
    tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK );
    tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 );
    tnl->Driver.Render.PrimTabElts[flags&PRIM_MODE_MASK]( ctx, start, count, flags );
-   RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_CLIP;
+   RADEON_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_BIT_POS;
 }
 
 
@@ -621,7 +621,7 @@ do {									\
 
 
 static GLboolean radeon_run_render( GLcontext *ctx,
-				    struct gl_pipeline_stage *stage )
+				    struct tnl_pipeline_stage *stage )
 {
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
@@ -646,18 +646,22 @@ static GLboolean radeon_run_render( GLcontext *ctx,
 	    return GL_TRUE;	/* too many vertices */
    }
 
-   for (i = 0 ; !(flags & PRIM_LAST) ; i += length)
+   for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      flags = VB->Primitive[i];
-      length = VB->PrimitiveLength[i];
+      GLuint prim = VB->Primitive[i].mode;
+      GLuint start = VB->Primitive[i].start;
+      GLuint length = VB->Primitive[i].count;
+
+      if (!length)
+	 continue;
 
       if (RADEON_DEBUG & DEBUG_PRIMS)
-	 fprintf(stderr, "radeon_render.c: prim %s %d..%d\n", 
-		 _mesa_lookup_enum_by_nr(flags & PRIM_MODE_MASK), 
-		 i, i+length);
+	 fprintf(stderr, "r200_render.c: prim %s %d..%d\n", 
+		 _mesa_lookup_enum_by_nr(prim & PRIM_MODE_MASK), 
+		 start, start+length);
 
       if (length)
-	 tab[flags & PRIM_MODE_MASK]( ctx, i, i + length, flags );
+	 tab[prim & PRIM_MODE_MASK]( ctx, start, start + length, flags );
    }
 
    tnl->Driver.Render.Finish( ctx );
@@ -668,9 +672,9 @@ static GLboolean radeon_run_render( GLcontext *ctx,
 
 
 static void radeon_check_render( GLcontext *ctx,
-				 struct gl_pipeline_stage *stage )
+				 struct tnl_pipeline_stage *stage )
 {
-   GLuint inputs = VERT_BIT_POS | VERT_BIT_CLIP | VERT_BIT_COLOR0;
+   GLuint inputs = VERT_BIT_POS | VERT_BIT_COLOR0;
 
    if (ctx->RenderMode == GL_RENDER) {
       if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)
@@ -690,13 +694,13 @@ static void radeon_check_render( GLcontext *ctx,
 }
 
 
-static void dtr( struct gl_pipeline_stage *stage )
+static void dtr( struct tnl_pipeline_stage *stage )
 {
    (void)stage;
 }
 
 
-const struct gl_pipeline_stage _radeon_render_stage =
+const struct tnl_pipeline_stage _radeon_render_stage =
 {
    "radeon render",
    (_DD_NEW_SEPARATE_SPECULAR |
@@ -729,7 +733,7 @@ struct texrect_stage_data {
 
 
 static GLboolean run_texrect_stage( GLcontext *ctx,
-				    struct gl_pipeline_stage *stage )
+				    struct tnl_pipeline_stage *stage )
 {
    struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
@@ -771,7 +775,7 @@ static GLboolean run_texrect_stage( GLcontext *ctx,
 /* Called the first time stage->run() is invoked.
  */
 static GLboolean alloc_texrect_data( GLcontext *ctx,
-				     struct gl_pipeline_stage *stage )
+				     struct tnl_pipeline_stage *stage )
 {
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    struct texrect_stage_data *store;
@@ -793,7 +797,7 @@ static GLboolean alloc_texrect_data( GLcontext *ctx,
 
 
 static void check_texrect( GLcontext *ctx,
-			   struct gl_pipeline_stage *stage )
+			   struct tnl_pipeline_stage *stage )
 {
    GLuint flags = 0;
 
@@ -809,7 +813,7 @@ static void check_texrect( GLcontext *ctx,
 }
 
 
-static void free_texrect_data( struct gl_pipeline_stage *stage )
+static void free_texrect_data( struct tnl_pipeline_stage *stage )
 {
    struct texrect_stage_data *store = TEXRECT_STAGE_DATA(stage);
    GLuint i;
@@ -824,7 +828,7 @@ static void free_texrect_data( struct gl_pipeline_stage *stage )
 }
 
 
-const struct gl_pipeline_stage _radeon_texrect_stage =
+const struct tnl_pipeline_stage _radeon_texrect_stage =
 {
    "radeon texrect stage",			/* name */
    _NEW_TEXTURE,	/* check_state */
@@ -1193,7 +1197,7 @@ void radeonFallback( GLcontext *ctx, GLuint bit, GLboolean mode )
 
 void radeonFlushVertices( GLcontext *ctx, GLuint flags )
 {
-   _tnl_flush_vertices( ctx, flags );
+   _tnl_FlushVertices( ctx, flags );
 
    if (flags & FLUSH_STORED_VERTICES)
       RADEON_NEWPRIM( RADEON_CONTEXT( ctx ) );
@@ -1242,13 +1246,4 @@ void radeonDestroySwtcl( GLcontext *ctx )
       rmesa->swtcl.verts = 0;
    }
 
-   if (rmesa->UbyteSecondaryColor.Ptr) {
-      ALIGN_FREE(rmesa->UbyteSecondaryColor.Ptr);
-      rmesa->UbyteSecondaryColor.Ptr = 0;
-   }
-
-   if (rmesa->UbyteColor.Ptr) {
-      ALIGN_FREE(rmesa->UbyteColor.Ptr);
-      rmesa->UbyteColor.Ptr = 0;
-   }
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_tcl.c b/src/mesa/drivers/dri/radeon/radeon_tcl.c
index 651194a8047..ffa09d15dac 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_tcl.c
@@ -277,7 +277,7 @@ void radeonTclPrimitive( GLcontext *ctx,
 /* TCL render.
  */
 static GLboolean radeon_run_tcl_render( GLcontext *ctx,
-					struct gl_pipeline_stage *stage )
+					struct tnl_pipeline_stage *stage )
 {
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
    TNLcontext *tnl = TNL_CONTEXT(ctx);
@@ -297,24 +297,19 @@ static GLboolean radeon_run_tcl_render( GLcontext *ctx,
 
    rmesa->tcl.Elts = VB->Elts;
 
-   for (i = VB->FirstPrimitive ; !(flags & PRIM_LAST) ; i += length)
+   for (i = 0 ; i < VB->PrimitiveCount ; i++)
    {
-      flags = VB->Primitive[i];
-      length = VB->PrimitiveLength[i];
-
-      if (RADEON_DEBUG & DEBUG_PRIMS)
-	 fprintf(stderr, "%s: prim %s %d..%d\n", 
-		 __FUNCTION__,
-		 _mesa_lookup_enum_by_nr(flags & PRIM_MODE_MASK), 
-		 i, i+length);
+      GLuint prim = VB->Primitive[i].mode;
+      GLuint start = VB->Primitive[i].start;
+      GLuint length = VB->Primitive[i].count;
 
       if (!length)
 	 continue;
 
       if (rmesa->tcl.Elts)
-	 radeonEmitEltPrimitive( ctx, i, i+length, flags );
+	 radeonEmitEltPrimitive( ctx, start, start+length, prim );
       else
-	 radeonEmitPrimitive( ctx, i, i+length, flags );
+	 radeonEmitPrimitive( ctx, start, start+length, prim );
    }
 
    return GL_FALSE;		/* finished the pipe */
@@ -323,7 +318,7 @@ static GLboolean radeon_run_tcl_render( GLcontext *ctx,
 
 
 static void radeon_check_tcl_render( GLcontext *ctx,
-				     struct gl_pipeline_stage *stage )
+				     struct tnl_pipeline_stage *stage )
 {
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
    GLuint inputs = VERT_BIT_POS;
@@ -374,13 +369,13 @@ static void radeon_check_tcl_render( GLcontext *ctx,
 }
 
 static void radeon_init_tcl_render( GLcontext *ctx,
-				    struct gl_pipeline_stage *stage )
+				    struct tnl_pipeline_stage *stage )
 {
    stage->check = radeon_check_tcl_render;
    stage->check( ctx, stage );
 }
 
-static void dtr( struct gl_pipeline_stage *stage )
+static void dtr( struct tnl_pipeline_stage *stage )
 {
    (void)stage;
 }
@@ -388,7 +383,7 @@ static void dtr( struct gl_pipeline_stage *stage )
 
 /* Initial state for tcl stage.  
  */
-const struct gl_pipeline_stage _radeon_tcl_stage =
+const struct tnl_pipeline_stage _radeon_tcl_stage =
 {
    "radeon render",
    (_DD_NEW_SEPARATE_SPECULAR |
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
index 9af0942898d..4e27324807f 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
@@ -973,13 +973,14 @@ void radeonVtxfmtInit( GLcontext *ctx, GLboolean useCodegen )
    vfmt->FogCoordfEXT = _mesa_noop_FogCoordfEXT;
    vfmt->EdgeFlag = _mesa_noop_EdgeFlag;
    vfmt->EdgeFlagv = _mesa_noop_EdgeFlagv;
-   vfmt->Indexi = _mesa_noop_Indexi;
-   vfmt->Indexiv = _mesa_noop_Indexiv;
+   vfmt->Indexf = _mesa_noop_Indexf;
+   vfmt->Indexfv = _mesa_noop_Indexfv;
 
 
    /* 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;
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
index 04cffb77729..0bdb9a8bbd6 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
@@ -831,16 +831,10 @@ void radeonVtxfmtInitChoosers( GLvertexformat *vfmt )
 {
    vfmt->Color3f = choose_Color3f;
    vfmt->Color3fv = choose_Color3fv;
-   vfmt->Color3ub = choose_Color3ub;
-   vfmt->Color3ubv = choose_Color3ubv;
    vfmt->Color4f = choose_Color4f;
    vfmt->Color4fv = choose_Color4fv;
-   vfmt->Color4ub = choose_Color4ub;
-   vfmt->Color4ubv = choose_Color4ubv;
    vfmt->SecondaryColor3fEXT = choose_SecondaryColor3fEXT;
    vfmt->SecondaryColor3fvEXT = choose_SecondaryColor3fvEXT;
-   vfmt->SecondaryColor3ubEXT = choose_SecondaryColor3ubEXT;
-   vfmt->SecondaryColor3ubvEXT = choose_SecondaryColor3ubvEXT;
    vfmt->MultiTexCoord1fARB = choose_MultiTexCoord1fARB;
    vfmt->MultiTexCoord1fvARB = choose_MultiTexCoord1fvARB;
    vfmt->MultiTexCoord2fARB = choose_MultiTexCoord2fARB;
@@ -855,6 +849,15 @@ void radeonVtxfmtInitChoosers( GLvertexformat *vfmt )
    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
 }