enable 3rd texture unit on radeon (default still 2 enabled units). Disable some multitexcoord codegen stuff noone understands to make it work. Replace most code testing explicitly for unit 0 and 1 with loops instead of adding test for unit 2, smaller/more readable code at the cost of maybe some slight performance hit. (Code provided by Andreas Stenglein, some adjustments by me.)

17 files changed, 285 insertions, 251 deletions

diff --git a/src/mesa/drivers/dri/radeon/radeon_context.c b/src/mesa/drivers/dri/radeon/radeon_context.c
index abb2c72e79..9ab6e67678 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.c
+++ b/src/mesa/drivers/dri/radeon/radeon_context.c
@@ -69,7 +69,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define need_GL_EXT_secondary_color
 #include "extension_helper.h"
 
-#define DRIVER_DATE	"20050831"
+#define DRIVER_DATE	"20051008"
 
 #include "vblank.h"
 #include "utils.h"
@@ -327,9 +327,10 @@ radeonCreateContext( const __GLcontextModes *glVisual,
     */
 
    ctx = rmesa->glCtx;
-   ctx->Const.MaxTextureUnits = 2;
-   ctx->Const.MaxTextureImageUnits = 2;
-   ctx->Const.MaxTextureCoordUnits = 2;
+   ctx->Const.MaxTextureUnits = driQueryOptioni (&rmesa->optionCache,
+						 "texture_units");
+   ctx->Const.MaxTextureImageUnits = ctx->Const.MaxTextureUnits;
+   ctx->Const.MaxTextureCoordUnits = ctx->Const.MaxTextureUnits;
 
    driCalculateMaxTextureLevels( rmesa->texture_heaps,
 				 rmesa->nr_heaps,
@@ -403,14 +404,12 @@ radeonCreateContext( const __GLcontextModes *glVisual,
    _tnl_allow_vertex_fog( ctx, GL_TRUE );
 
 
-   _math_matrix_ctr( &rmesa->TexGenMatrix[0] );
-   _math_matrix_ctr( &rmesa->TexGenMatrix[1] );
-   _math_matrix_ctr( &rmesa->tmpmat[0] );
-   _math_matrix_ctr( &rmesa->tmpmat[1] );
-   _math_matrix_set_identity( &rmesa->TexGenMatrix[0] );
-   _math_matrix_set_identity( &rmesa->TexGenMatrix[1] );
-   _math_matrix_set_identity( &rmesa->tmpmat[0] );
-   _math_matrix_set_identity( &rmesa->tmpmat[1] );
+   for ( i = 0 ; i < RADEON_MAX_TEXTURE_UNITS ; i++ ) {
+      _math_matrix_ctr( &rmesa->TexGenMatrix[i] );
+      _math_matrix_ctr( &rmesa->tmpmat[i] );
+      _math_matrix_set_identity( &rmesa->TexGenMatrix[i] );
+      _math_matrix_set_identity( &rmesa->tmpmat[i] );
+   }
 
    driInitExtensions( ctx, card_extensions, GL_TRUE );
    if (rmesa->glCtx->Mesa_DXTn) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_context.h b/src/mesa/drivers/dri/radeon/radeon_context.h
index 3bd8f6ebc6..09bb96861e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_context.h
+++ b/src/mesa/drivers/dri/radeon/radeon_context.h
@@ -122,11 +122,17 @@ struct radeon_stipple_state {
    GLuint mask[32];
 };
 
+/* used for both tcl_vtx and vc_frmt tex bits (they are identical) */
+#define RADEON_ST_BIT(unit) \
+(unit == 0 ? RADEON_CP_VC_FRMT_ST0 : (RADEON_CP_VC_FRMT_ST1 >> 2) << (2 * unit))
 
+#define RADEON_Q_BIT(unit) \
+(unit == 0 ? RADEON_CP_VC_FRMT_Q0 : (RADEON_CP_VC_FRMT_Q1 >> 2) << (2 * unit))
 
 #define TEX_0   0x1
 #define TEX_1   0x2
-#define TEX_ALL 0x3
+#define TEX_2   0x4
+#define TEX_ALL 0x7
 
 typedef struct radeon_tex_obj radeonTexObj, *radeonTexObjPtr;
 
@@ -408,17 +414,17 @@ struct radeon_hw_state {
    struct radeon_state_atom vpt;
    struct radeon_state_atom tcl;
    struct radeon_state_atom msc;
-   struct radeon_state_atom tex[2];
+   struct radeon_state_atom tex[3];
    struct radeon_state_atom zbs;
    struct radeon_state_atom mtl; 
-   struct radeon_state_atom mat[5]; 
+   struct radeon_state_atom mat[6];
    struct radeon_state_atom lit[8]; /* includes vec, scl commands */
    struct radeon_state_atom ucp[6];
    struct radeon_state_atom eye; /* eye pos */
    struct radeon_state_atom grd; /* guard band clipping */
    struct radeon_state_atom fog; 
    struct radeon_state_atom glt; 
-   struct radeon_state_atom txr[2]; /* for NPOT */
+   struct radeon_state_atom txr[3]; /* for NPOT */
 
    int max_state_size;	/* Number of bytes necessary for a full state emit. */
    GLboolean is_dirty, all_dirty;
@@ -646,30 +652,29 @@ struct radeon_prim {
    GLuint prim;
 };
 
+/* A maximum total of 20 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 (9 floats total).
+ * 
+ * The position data is never actually stored here, so 3 elements could be
+ * trimmed out of the buffer. This number is only valid for vtxfmt!
+ */
+#define RADEON_MAX_VERTEX_SIZE 20
+
 struct radeon_vbinfo {
    GLint counter, initial_counter;
    GLint *dmaptr;
    void (*notify)( void );
    GLint vertex_size;
 
-   /* A maximum total of 15 elements per vertex:  3 floats for position, 3
-    * floats for normal, 4 floats for color, 4 bytes for secondary color,
-    * 2 floats for each texture unit (4 floats total).
-    * 
-    * As soon as the 3rd TMU is supported or cube maps (or 3D textures) are
-    * supported, this value will grow.
-    * 
-    * The position data is never actually stored here, so 3 elements could be
-    * trimmed out of the buffer.
-    */
-   union { float f; int i; radeon_color_t color; } vertex[15];
+   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[2];
+   GLfloat *texcoordptr[4];	/* 3 (TMU) + 1 for radeon_vtxfmt_c.c when GL_TEXTURE3 */
 
    GLenum *prim;		/* &ctx->Driver.CurrentExecPrimitive */
    GLuint primflags;
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos.c b/src/mesa/drivers/dri/radeon/radeon_maos.c
index c62edd715c..ea1e8934de 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos.c
@@ -4,7 +4,7 @@
  * Otherwise, must use verts.
  */
 #include "radeon_context.h"
-#define RADEON_MAOS_VERTS 1
+#define RADEON_MAOS_VERTS 0
 #if (RADEON_MAOS_VERTS) || (RADEON_OLD_PACKETS)
 #include "radeon_maos_verts.c"
 #else
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
index 5aa77b8afa..1602464011 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_arrays.c
@@ -443,7 +443,7 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
    GLuint nr = 0;
    GLuint vfmt = 0;
    GLuint count = VB->Count;
-   GLuint vtx;
+   GLuint vtx, unit;
    
 #if 0
    if (RADEON_DEBUG & DEBUG_VERTS) 
@@ -526,58 +526,36 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
    }
 
    vtx = (rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] &
-	  ~(RADEON_TCL_VTX_Q0|RADEON_TCL_VTX_Q1));
-
-   if (inputs & VERT_BIT_TEX0) {
-      if (!rmesa->tcl.tex[0].buf)
-	 emit_tex_vector( ctx,
-			  &(rmesa->tcl.tex[0]),
-			  (char *)VB->TexCoordPtr[0]->data,
-			  VB->TexCoordPtr[0]->size,
-			  VB->TexCoordPtr[0]->stride,
-			  count );
-
-      vfmt |= RADEON_CP_VC_FRMT_ST0;
-      /* assume we need the 3rd coord if texgen is active for r/q OR at least 3
-         coords are submitted. This may not be 100% correct */
-      if (VB->TexCoordPtr[0]->size >= 3) {
-	 vtx |= RADEON_TCL_VTX_Q0;
-	 vfmt |= RADEON_CP_VC_FRMT_Q0;
+	  ~(RADEON_TCL_VTX_Q0|RADEON_TCL_VTX_Q1|RADEON_TCL_VTX_Q2));
+      
+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
+      if (inputs & VERT_BIT_TEX(unit)) {
+	 if (!rmesa->tcl.tex[unit].buf)
+	    emit_tex_vector( ctx,
+			     &(rmesa->tcl.tex[unit]),
+			     (char *)VB->TexCoordPtr[unit]->data,
+			     VB->TexCoordPtr[unit]->size,
+			     VB->TexCoordPtr[unit]->stride,
+			     count );
+
+	 vfmt |= RADEON_ST_BIT(unit);
+         /* assume we need the 3rd coord if texgen is active for r/q OR at least
+	    3 coords are submitted. This may not be 100% correct */
+         if (VB->TexCoordPtr[unit]->size >= 3) {
+	 /* tcl_vtx and vc_frmt values are identical */
+	    vtx |= RADEON_Q_BIT(unit);
+	    vfmt |= RADEON_Q_BIT(unit);
+	 }
+	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
+	    vtx |= RADEON_Q_BIT(unit);
+	 else if (VB->TexCoordPtr[unit]->size >= 3) {
+	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
+		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
+	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
+	 }
+	 component[nr++] = &rmesa->tcl.tex[unit];
       }
-      if ( (ctx->Texture.Unit[0].TexGenEnabled & (R_BIT | Q_BIT)) )
-	 vtx |= RADEON_TCL_VTX_Q0;
-      else if (VB->TexCoordPtr[0]->size >= 3) {
-	 GLuint swaptexmatcol = (VB->TexCoordPtr[0]->size - 3);
-	 if ((rmesa->NeedTexMatrix & 1) &&
-		(swaptexmatcol != (rmesa->TexMatColSwap & 1)))
-	    radeonUploadTexMatrix( rmesa, 0, swaptexmatcol ) ;
-      }
-      component[nr++] = &rmesa->tcl.tex[0];
-   }
-
-   if (inputs & VERT_BIT_TEX1) {
-      if (!rmesa->tcl.tex[1].buf)
-	 emit_tex_vector( ctx,
-			  &(rmesa->tcl.tex[1]),
-			  (char *)VB->TexCoordPtr[1]->data,
-			  VB->TexCoordPtr[1]->size,
-			  VB->TexCoordPtr[1]->stride,
-			  count );
-	 
-      vfmt |= RADEON_CP_VC_FRMT_ST1;
-      if (VB->TexCoordPtr[1]->size >= 3) {
-	 vtx |= RADEON_TCL_VTX_Q1;
-	 vfmt |= RADEON_CP_VC_FRMT_Q1;
-      }
-      if ( (ctx->Texture.Unit[1].TexGenEnabled & (R_BIT | Q_BIT)) )
-	 vtx |= RADEON_TCL_VTX_Q1;
-      else if (VB->TexCoordPtr[1]->size >= 3) {
-	 GLuint swaptexmatcol = (VB->TexCoordPtr[1]->size - 3);
-	 if (((rmesa->NeedTexMatrix >> 1) & 1) &&
-		(swaptexmatcol != ((rmesa->TexMatColSwap >> 1) & 1)))
-	    radeonUploadTexMatrix( rmesa, 1, swaptexmatcol ) ;
-      }
-      component[nr++] = &rmesa->tcl.tex[1];
    }
 
    if (vtx != rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT]) {
@@ -593,6 +571,7 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
 void radeonReleaseArrays( GLcontext *ctx, GLuint newinputs )
 {
    radeonContextPtr rmesa = RADEON_CONTEXT( ctx );
+   GLuint unit;
 
 #if 0
    if (RADEON_DEBUG & DEBUG_VERTS) 
@@ -611,9 +590,8 @@ void radeonReleaseArrays( GLcontext *ctx, GLuint newinputs )
    if (newinputs & VERT_BIT_COLOR1) 
       radeonReleaseDmaRegion( rmesa, &rmesa->tcl.spec, __FUNCTION__ );
 
-   if (newinputs & VERT_BIT_TEX0)
-      radeonReleaseDmaRegion( rmesa, &rmesa->tcl.tex[0], __FUNCTION__ );
-
-   if (newinputs & VERT_BIT_TEX1)
-      radeonReleaseDmaRegion( rmesa, &rmesa->tcl.tex[1], __FUNCTION__ );
+   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
+      if (newinputs & VERT_BIT_TEX(unit))
+         radeonReleaseDmaRegion( rmesa, &rmesa->tcl.tex[unit], __FUNCTION__ );
+   }
 }
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
index fc31aff47a..265d061b36 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_vbtmp.h
@@ -255,6 +255,7 @@ static void TAG(emit)( GLcontext *ctx,
 	 if (DO_TEX2) {
 	    v[0].ui = tc2[0][0];
 	    v[1].ui = tc2[0][1];
+	    if (TCL_DEBUG) fprintf(stderr, "t2: %.2f %.2f ", v[0].f, v[1].f);
 	    if (DO_PTEX) {
 	       if (fill_tex & (1<<2))
 		  v[2].f = 1.0;
@@ -262,6 +263,7 @@ static void TAG(emit)( GLcontext *ctx,
 		  v[2].ui = tc2[0][2];
 	       else
 		  v[2].ui = tc2[0][3];
+	       if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[2].f);
 	       v += 3;
 	    } 
 	    else
diff --git a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
index 3d004adda6..57aa90321c 100644
--- a/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
+++ b/src/mesa/drivers/dri/radeon/radeon_maos_verts.c
@@ -51,7 +51,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "radeon_maos.h"
 
 
-#define RADEON_TCL_MAX_SETUP 13
+#define RADEON_TCL_MAX_SETUP 19
 
 union emit_union { float f; GLuint ui; radeon_color_t rgba; };
 
@@ -67,10 +67,10 @@ static struct {
 #define DO_FOG  (IND & RADEON_CP_VC_FRMT_PKSPEC)
 #define DO_TEX0 (IND & RADEON_CP_VC_FRMT_ST0)
 #define DO_TEX1 (IND & RADEON_CP_VC_FRMT_ST1)
+#define DO_TEX2 (IND & RADEON_CP_VC_FRMT_ST2)
 #define DO_PTEX (IND & RADEON_CP_VC_FRMT_Q0)
 #define DO_NORM (IND & RADEON_CP_VC_FRMT_N0)
 
-#define DO_TEX2 0
 #define DO_TEX3 0
 
 #define GET_TEXSOURCE(n)  n
@@ -202,6 +202,77 @@ static struct {
 #define TAG(x) x##_w_rgba_spec_stq_stq_n
 #include "radeon_maos_vbtmp.h"
 
+#define IDX 13
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_PKCOLOR|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_ST2)
+#define TAG(x) x##_rgba_st_st_st
+#include "radeon_maos_vbtmp.h"
+
+#define IDX 14
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_PKCOLOR|		\
+	     RADEON_CP_VC_FRMT_PKSPEC|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_ST2)
+#define TAG(x) x##_rgba_spec_st_st_st
+#include "radeon_maos_vbtmp.h"
+
+#define IDX 15
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_ST2|		\
+	     RADEON_CP_VC_FRMT_N0)
+#define TAG(x) x##_st_st_st_n
+#include "radeon_maos_vbtmp.h"
+
+#define IDX 16
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_PKCOLOR|		\
+	     RADEON_CP_VC_FRMT_PKSPEC|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_ST2|		\
+	     RADEON_CP_VC_FRMT_N0)
+#define TAG(x) x##_rgba_spec_st_st_st_n
+#include "radeon_maos_vbtmp.h"
+
+#define IDX 17
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_PKCOLOR|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_Q0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_Q1|		\
+	     RADEON_CP_VC_FRMT_ST2|		\
+	     RADEON_CP_VC_FRMT_Q2)
+#define TAG(x) x##_rgba_stq_stq_stq
+#include "radeon_maos_vbtmp.h"
+
+#define IDX 18
+#define IND (RADEON_CP_VC_FRMT_XY|		\
+	     RADEON_CP_VC_FRMT_Z|		\
+	     RADEON_CP_VC_FRMT_W0|		\
+	     RADEON_CP_VC_FRMT_PKCOLOR|		\
+	     RADEON_CP_VC_FRMT_PKSPEC|		\
+	     RADEON_CP_VC_FRMT_ST0|		\
+	     RADEON_CP_VC_FRMT_Q0|		\
+	     RADEON_CP_VC_FRMT_ST1|		\
+	     RADEON_CP_VC_FRMT_Q1|		\
+	     RADEON_CP_VC_FRMT_ST2|		\
+	     RADEON_CP_VC_FRMT_Q2|		\
+	     RADEON_CP_VC_FRMT_N0)
+#define TAG(x) x##_w_rgba_spec_stq_stq_stq_n
+#include "radeon_maos_vbtmp.h"
 
 
 
@@ -226,6 +297,12 @@ static void init_tcl_verts( void )
    init_rgba_stq();
    init_rgba_stq_stq();
    init_w_rgba_spec_stq_stq_n();
+   init_rgba_st_st_st();
+   init_rgba_spec_st_st_st();
+   init_st_st_st_n();
+   init_rgba_spec_st_st_st_n();
+   init_rgba_stq_stq_stq();
+   init_w_rgba_spec_stq_stq_stq_n();
 }
 
 
@@ -234,8 +311,9 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
    struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb;
    GLuint req = 0;
+   GLuint unit;
    GLuint vtx = (rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] &
-		 ~(RADEON_TCL_VTX_Q0|RADEON_TCL_VTX_Q1));
+		 ~(RADEON_TCL_VTX_Q0|RADEON_TCL_VTX_Q1|RADEON_TCL_VTX_Q2));
    int i;
    static int firsttime = 1;
 
@@ -263,39 +341,23 @@ void radeonEmitArrays( GLcontext *ctx, GLuint inputs )
       req |= RADEON_CP_VC_FRMT_PKSPEC;
    }
 
-   if (inputs & VERT_BIT_TEX0) {
-      req |= RADEON_CP_VC_FRMT_ST0;
-      /* assume we need the 3rd coord if texgen is active for r/q OR at least 3
-         coords are submitted. This may not be 100% correct */
-      if (VB->TexCoordPtr[0]->size >= 3) {
-	 req |= RADEON_CP_VC_FRMT_Q0;
-	 vtx |= RADEON_TCL_VTX_Q0;
-      }
-      if ( (ctx->Texture.Unit[0].TexGenEnabled & (R_BIT | Q_BIT)) )
-	 vtx |= RADEON_TCL_VTX_Q0;
-      else if (VB->TexCoordPtr[0]->size >= 3) {
-	 GLuint swaptexmatcol = (VB->TexCoordPtr[0]->size - 3);
-	 if ((rmesa->NeedTexMatrix & 1) &&
-		(swaptexmatcol != (rmesa->TexMatColSwap & 1)))
-	    radeonUploadTexMatrix( rmesa, 0, swaptexmatcol ) ;
-      }
-   }
-
-
-   if (inputs & VERT_BIT_TEX1) {
-      req |= RADEON_CP_VC_FRMT_ST1;
-
-      if (VB->TexCoordPtr[1]->size >= 3) {
-	 req |= RADEON_CP_VC_FRMT_Q1;
-	 vtx |= RADEON_TCL_VTX_Q1;
-      }
-      if ( (ctx->Texture.Unit[1].TexGenEnabled & (R_BIT | Q_BIT)) )
-	 vtx |= RADEON_TCL_VTX_Q1;
-      else if (VB->TexCoordPtr[1]->size >= 3) {
-	 GLuint swaptexmatcol = (VB->TexCoordPtr[1]->size - 3);
-	 if (((rmesa->NeedTexMatrix >> 1) & 1) &&
-		(swaptexmatcol != ((rmesa->TexMatColSwap >> 1) & 1)))
-	    radeonUploadTexMatrix( rmesa, 1, swaptexmatcol ) ;
+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
+      if (inputs & VERT_BIT_TEX(unit)) {
+	 req |= RADEON_ST_BIT(unit);
+	 /* assume we need the 3rd coord if texgen is active for r/q OR at least
+	    3 coords are submitted. This may not be 100% correct */
+	 if (VB->TexCoordPtr[unit]->size >= 3) {
+	    req |= RADEON_Q_BIT(unit);
+	    vtx |= RADEON_Q_BIT(unit);
+	 }
+	 if ( (ctx->Texture.Unit[unit].TexGenEnabled & (R_BIT | Q_BIT)) )
+	    vtx |= RADEON_Q_BIT(unit);
+	 else if (VB->TexCoordPtr[unit]->size >= 3) {
+	    GLuint swaptexmatcol = (VB->TexCoordPtr[unit]->size - 3);
+	    if (((rmesa->NeedTexMatrix >> unit) & 1) &&
+		 (swaptexmatcol != ((rmesa->TexMatColSwap >> unit) & 1)))
+	       radeonUploadTexMatrix( rmesa, unit, swaptexmatcol ) ;
+	 }
       }
    }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 9f25de44e2..c79f406f68 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -65,6 +65,7 @@ DRI_CONF_BEGIN
         DRI_CONF_TCL_MODE(DRI_CONF_TCL_CODEGEN)
         DRI_CONF_FTHROTTLE_MODE(DRI_CONF_FTHROTTLE_IRQS)
         DRI_CONF_VBLANK_MODE(DRI_CONF_VBLANK_DEF_INTERVAL_0)
+        DRI_CONF_MAX_TEXTURE_UNITS(2,2,3)
         DRI_CONF_HYPERZ(false)
     DRI_CONF_SECTION_END
     DRI_CONF_SECTION_QUALITY
@@ -81,7 +82,7 @@ DRI_CONF_BEGIN
         DRI_CONF_NO_RAST(false)
     DRI_CONF_SECTION_END
 DRI_CONF_END;
-static const GLuint __driNConfigOptions = 13;
+static const GLuint __driNConfigOptions = 14;
 
 extern const struct dri_extension card_extensions[];
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_state.c b/src/mesa/drivers/dri/radeon/radeon_state.c
index 6457fcac5d..a79fefee50 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state.c
@@ -2145,7 +2145,7 @@ static void update_texturematrix( GLcontext *ctx )
    rmesa->NeedTexMatrix = 0;
    rmesa->TexMatColSwap = 0;
 
-   for (unit = 0 ; unit < 2; unit++) {
+   for (unit = 0 ; unit < ctx->Const.MaxTextureUnits; unit++) {
       if (ctx->Texture.Unit[unit]._ReallyEnabled) {
 	 GLboolean needMatrix = GL_FALSE;
 	 if (ctx->TextureMatrixStack[unit].Top->type != MATRIX_IDENTITY) {
@@ -2189,18 +2189,17 @@ static void update_texturematrix( GLcontext *ctx )
 
    tpc = (texMatEnabled | rmesa->TexGenEnabled);
 
-   vs &= ~((0xf << RADEON_TCL_TEX_0_OUTPUT_SHIFT) |
-	   (0xf << RADEON_TCL_TEX_1_OUTPUT_SHIFT));
-
-   if (tpc & RADEON_TEXGEN_TEXMAT_0_ENABLE)
-      vs |= RADEON_TCL_TEX_COMPUTED_TEX_0 << RADEON_TCL_TEX_0_OUTPUT_SHIFT;
-   else
-      vs |= RADEON_TCL_TEX_INPUT_TEX_0 << RADEON_TCL_TEX_0_OUTPUT_SHIFT;
-
-   if (tpc & RADEON_TEXGEN_TEXMAT_1_ENABLE)
-      vs |= RADEON_TCL_TEX_COMPUTED_TEX_1 << RADEON_TCL_TEX_1_OUTPUT_SHIFT;
-   else
-      vs |= RADEON_TCL_TEX_INPUT_TEX_1 << RADEON_TCL_TEX_1_OUTPUT_SHIFT;
+   /* TCL_TEX_COMPUTED_x is TCL_TEX_INPUT_x | 0x8 */
+   vs &= ~((RADEON_TCL_TEX_COMPUTED_TEX_0 << RADEON_TCL_TEX_0_OUTPUT_SHIFT) |
+	   (RADEON_TCL_TEX_COMPUTED_TEX_0 << RADEON_TCL_TEX_1_OUTPUT_SHIFT) |
+	   (RADEON_TCL_TEX_COMPUTED_TEX_0 << RADEON_TCL_TEX_2_OUTPUT_SHIFT));
+
+   vs |= (((tpc & RADEON_TEXGEN_TEXMAT_0_ENABLE) <<
+	 (RADEON_TCL_TEX_0_OUTPUT_SHIFT + 3)) |
+      ((tpc & RADEON_TEXGEN_TEXMAT_1_ENABLE) <<
+	 (RADEON_TCL_TEX_1_OUTPUT_SHIFT + 2)) |
+      ((tpc & RADEON_TEXGEN_TEXMAT_2_ENABLE) <<
+	 (RADEON_TCL_TEX_2_OUTPUT_SHIFT + 1)));
 
    if (tpc != rmesa->hw.tcl.cmd[TCL_TEXTURE_PROC_CTL] ||
        vs != rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXSEL]) {
diff --git a/src/mesa/drivers/dri/radeon/radeon_state_init.c b/src/mesa/drivers/dri/radeon/radeon_state_init.c
index be3635a380..7e7dc0f782 100644
--- a/src/mesa/drivers/dri/radeon/radeon_state_init.c
+++ b/src/mesa/drivers/dri/radeon/radeon_state_init.c
@@ -115,10 +115,12 @@ static GLboolean check_##NM( GLcontext *ctx )		\
 CHECK( always, GL_TRUE )
 CHECK( tex0, ctx->Texture.Unit[0]._ReallyEnabled )
 CHECK( tex1, ctx->Texture.Unit[1]._ReallyEnabled )
+CHECK( tex2, ctx->Texture.Unit[2]._ReallyEnabled )
 CHECK( fog, ctx->Fog.Enabled )
 TCL_CHECK( tcl, GL_TRUE )
 TCL_CHECK( tcl_tex0, ctx->Texture.Unit[0]._ReallyEnabled )
 TCL_CHECK( tcl_tex1, ctx->Texture.Unit[1]._ReallyEnabled )
+TCL_CHECK( tcl_tex2, ctx->Texture.Unit[2]._ReallyEnabled )
 TCL_CHECK( tcl_lighting, ctx->Light.Enabled )
 TCL_CHECK( tcl_eyespace_or_lighting, ctx->_NeedEyeCoords || ctx->Light.Enabled )
 TCL_CHECK( tcl_lit0, ctx->Light.Enabled && ctx->Light.Light[0].Enabled )
@@ -139,6 +141,7 @@ TCL_CHECK( tcl_eyespace_or_fog, ctx->_NeedEyeCoords || ctx->Fog.Enabled )
 
 CHECK( txr0, (ctx->Texture.Unit[0]._ReallyEnabled & TEXTURE_RECT_BIT))
 CHECK( txr1, (ctx->Texture.Unit[1]._ReallyEnabled & TEXTURE_RECT_BIT))
+CHECK( txr2, (ctx->Texture.Unit[2]._ReallyEnabled & TEXTURE_RECT_BIT))
 
 
 
@@ -229,11 +232,13 @@ void radeonInitState( radeonContextPtr rmesa )
    ALLOC_STATE( eye, tcl_lighting, EYE_STATE_SIZE, "EYE/eye-vector", 1 );
    ALLOC_STATE( tex[0], tex0, TEX_STATE_SIZE, "TEX/tex-0", 0 );
    ALLOC_STATE( tex[1], tex1, TEX_STATE_SIZE, "TEX/tex-1", 0 );
+   ALLOC_STATE( tex[2], tex2, TEX_STATE_SIZE, "TEX/tex-2", 0 );
    ALLOC_STATE( mat[0], tcl, MAT_STATE_SIZE, "MAT/modelproject", 1 );
    ALLOC_STATE( mat[1], tcl_eyespace_or_fog, MAT_STATE_SIZE, "MAT/modelview", 1 );
    ALLOC_STATE( mat[2], tcl_eyespace_or_lighting, MAT_STATE_SIZE, "MAT/it-modelview", 1 );
    ALLOC_STATE( mat[3], tcl_tex0, MAT_STATE_SIZE, "MAT/texmat0", 1 );
    ALLOC_STATE( mat[4], tcl_tex1, MAT_STATE_SIZE, "MAT/texmat1", 1 );
+   ALLOC_STATE( mat[5], tcl_tex2, MAT_STATE_SIZE, "MAT/texmat2", 1 );
    ALLOC_STATE( ucp[0], tcl_ucp0, UCP_STATE_SIZE, "UCP/userclip-0", 1 );
    ALLOC_STATE( ucp[1], tcl_ucp1, UCP_STATE_SIZE, "UCP/userclip-1", 1 );
    ALLOC_STATE( ucp[2], tcl_ucp2, UCP_STATE_SIZE, "UCP/userclip-2", 1 );
@@ -250,6 +255,7 @@ void radeonInitState( radeonContextPtr rmesa )
    ALLOC_STATE( lit[7], tcl_lit7, LIT_STATE_SIZE, "LIT/light-7", 1 );
    ALLOC_STATE( txr[0], txr0, TXR_STATE_SIZE, "TXR/txr-0", 0 );
    ALLOC_STATE( txr[1], txr1, TXR_STATE_SIZE, "TXR/txr-1", 0 );
+   ALLOC_STATE( txr[2], txr2, TXR_STATE_SIZE, "TXR/txr-2", 0 );
 
    radeonSetUpAtomList( rmesa );
 
@@ -269,12 +275,15 @@ void radeonInitState( radeonContextPtr rmesa )
    rmesa->hw.tex[0].cmd[TEX_CMD_1] = cmdpkt(RADEON_EMIT_PP_BORDER_COLOR_0);
    rmesa->hw.tex[1].cmd[TEX_CMD_0] = cmdpkt(RADEON_EMIT_PP_TXFILTER_1);
    rmesa->hw.tex[1].cmd[TEX_CMD_1] = cmdpkt(RADEON_EMIT_PP_BORDER_COLOR_1);
+   rmesa->hw.tex[2].cmd[TEX_CMD_0] = cmdpkt(RADEON_EMIT_PP_TXFILTER_2);
+   rmesa->hw.tex[2].cmd[TEX_CMD_1] = cmdpkt(RADEON_EMIT_PP_BORDER_COLOR_2);
    rmesa->hw.zbs.cmd[ZBS_CMD_0] = cmdpkt(RADEON_EMIT_SE_ZBIAS_FACTOR);
    rmesa->hw.tcl.cmd[TCL_CMD_0] = cmdpkt(RADEON_EMIT_SE_TCL_OUTPUT_VTX_FMT);
    rmesa->hw.mtl.cmd[MTL_CMD_0] = 
       cmdpkt(RADEON_EMIT_SE_TCL_MATERIAL_EMMISSIVE_RED);
    rmesa->hw.txr[0].cmd[TXR_CMD_0] = cmdpkt(RADEON_EMIT_PP_TEX_SIZE_0);
    rmesa->hw.txr[1].cmd[TXR_CMD_0] = cmdpkt(RADEON_EMIT_PP_TEX_SIZE_1);
+   rmesa->hw.txr[2].cmd[TXR_CMD_0] = cmdpkt(RADEON_EMIT_PP_TEX_SIZE_2);
    rmesa->hw.grd.cmd[GRD_CMD_0] = 
       cmdscl( RADEON_SS_VERT_GUARD_CLIP_ADJ_ADDR, 1, 4 );
    rmesa->hw.fog.cmd[FOG_CMD_0] = 
@@ -284,7 +293,7 @@ void radeonInitState( radeonContextPtr rmesa )
    rmesa->hw.eye.cmd[EYE_CMD_0] = 
       cmdvec( RADEON_VS_EYE_VECTOR_ADDR, 1, 4 );
 
-   for (i = 0 ; i < 5; i++) {
+   for (i = 0 ; i < 6; i++) {
       rmesa->hw.mat[i].cmd[MAT_CMD_0] = 
 	 cmdvec( RADEON_VS_MATRIX_0_ADDR + i*4, 1, 16);
    }
@@ -309,8 +318,8 @@ void radeonInitState( radeonContextPtr rmesa )
 				     RADEON_CHROMA_FUNC_FAIL |
 				     RADEON_CHROMA_KEY_NEAREST |
 				     RADEON_SHADOW_FUNC_EQUAL |
-				     RADEON_SHADOW_PASS_1 |
-				     RADEON_RIGHT_HAND_CUBE_OGL);
+				     RADEON_SHADOW_PASS_1 /*|
+				     RADEON_RIGHT_HAND_CUBE_OGL */);
 
    rmesa->hw.ctx.cmd[CTX_PP_FOG_COLOR] = (RADEON_FOG_VERTEX |
 					  RADEON_FOG_USE_DEPTH);
@@ -505,7 +514,8 @@ void radeonInitState( radeonContextPtr rmesa )
    rmesa->hw.tcl.cmd[TCL_MATRIX_SELECT_1] = 
       ((MODEL_PROJ << RADEON_MODELPROJECT_0_SHIFT) |
        (TEXMAT_0 << RADEON_TEXMAT_0_SHIFT) |
-       (TEXMAT_1 << RADEON_TEXMAT_1_SHIFT));
+       (TEXMAT_1 << RADEON_TEXMAT_1_SHIFT) |
+       (TEXMAT_2 << RADEON_TEXMAT_2_SHIFT));
 
    rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] = 
       (RADEON_UCP_IN_CLIP_SPACE |
diff --git a/src/mesa/drivers/dri/radeon/radeon_swtcl.c b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
index 9045619187..7200f71ef6 100644
--- a/src/mesa/drivers/dri/radeon/radeon_swtcl.c
+++ b/src/mesa/drivers/dri/radeon/radeon_swtcl.c
@@ -625,7 +625,7 @@ static void radeonResetLineStipple( GLcontext *ctx );
 
 #define RADEON_TWOSIDE_BIT	0x01
 #define RADEON_UNFILLED_BIT	0x02
-#define RADEON_MAX_TRIFUNC	0x08
+#define RADEON_MAX_TRIFUNC	0x04
 
 
 static struct {
diff --git a/src/mesa/drivers/dri/radeon/radeon_tcl.h b/src/mesa/drivers/dri/radeon/radeon_tcl.h
index 263b803d62..062c9fbb54 100644
--- a/src/mesa/drivers/dri/radeon/radeon_tcl.h
+++ b/src/mesa/drivers/dri/radeon/radeon_tcl.h
@@ -56,7 +56,8 @@ extern void radeonTclFallback( GLcontext *ctx, GLuint bit, GLboolean mode );
 #define RADEON_TCL_FALLBACK_TEXGEN_2          0x40 /* texgen, unit 2 */
 #define RADEON_TCL_FALLBACK_TCL_DISABLE       0x80 /* user disable */
 
-#define RADEON_MAX_TCL_VERTSIZE (15*4)
+/* max maos_verts vertex format has a size of 18 floats */
+#define RADEON_MAX_TCL_VERTSIZE (18*4)
 
 #define TCL_FALLBACK( ctx, bit, mode )	radeonTclFallback( ctx, bit, mode )
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index 56ec2ae398..fe5445bf62 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -921,10 +921,7 @@ static GLboolean radeon_validate_texgen( GLcontext *ctx, GLuint unit )
          this from way too many places, would be much easier if we could leave
          tcl q coord always enabled as on r200) */
       RADEON_STATECHANGE( rmesa, tcl );
-      if (unit == 0)
-	 rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_TCL_VTX_Q0;
-      else
-	 rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_TCL_VTX_Q1;
+      rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_Q_BIT(unit);
    }
 
    switch (texUnit->GenModeS) {
@@ -997,19 +994,8 @@ static void disable_tex( GLcontext *ctx, int unit )
 	  ~((RADEON_TEX_0_ENABLE | RADEON_TEX_BLEND_0_ENABLE) << unit);
 
       RADEON_STATECHANGE( rmesa, tcl );
-      switch (unit) {
-      case 0:
-	 rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] &= ~(RADEON_TCL_VTX_ST0 |
-						   RADEON_TCL_VTX_Q0);
-	    break;
-      case 1:
-	 rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] &= ~(RADEON_TCL_VTX_ST1 |
-						   RADEON_TCL_VTX_Q1);
-	 break;
-      default:
-	 break;
-      }
-
+      rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] &= ~(RADEON_ST_BIT(unit) |
+						RADEON_Q_BIT(unit));
 
       if (rmesa->TclFallback & (RADEON_TCL_FALLBACK_TEXGEN_0<<unit)) {
 	 TCL_FALLBACK( ctx, (RADEON_TCL_FALLBACK_TEXGEN_0<<unit), GL_FALSE);
@@ -1135,10 +1121,7 @@ static GLboolean update_tex_common( GLcontext *ctx, int unit )
 
       RADEON_STATECHANGE( rmesa, tcl );
 
-      if (unit == 0)
-	  rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_TCL_VTX_ST0;
-      else 
-	  rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_TCL_VTX_ST1;
+      rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_ST_BIT(unit);
 
       rmesa->recheck_texgen[unit] = GL_TRUE;
    }
@@ -1199,7 +1182,8 @@ void radeonUpdateTextureState( GLcontext *ctx )
    GLboolean ok;
 
    ok = (radeonUpdateTextureUnit( ctx, 0 ) &&
-	 radeonUpdateTextureUnit( ctx, 1 ));
+	 radeonUpdateTextureUnit( ctx, 1 ) &&
+	 radeonUpdateTextureUnit( ctx, 2 ));
 
    FALLBACK( rmesa, RADEON_FALLBACK_TEXTURE, !ok );
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
index c5ea51cea9..e554aeb330 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt.c
@@ -57,6 +57,8 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #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 );
@@ -102,6 +104,7 @@ static void count_funcs( radeonContextPtr rmesa )
 
 void radeon_copy_to_current( GLcontext *ctx ) 
 {
+   GLuint unit;
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
 
    assert(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT);
@@ -134,18 +137,13 @@ void radeon_copy_to_current( GLcontext *ctx )
       ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] = UBYTE_TO_FLOAT( rmesa->vb.specptr->blue );
    } 
 
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_ST0) {
-      ctx->Current.Attrib[VERT_ATTRIB_TEX0][0] = rmesa->vb.texcoordptr[0][0];
-      ctx->Current.Attrib[VERT_ATTRIB_TEX0][1] = rmesa->vb.texcoordptr[0][1];
-      ctx->Current.Attrib[VERT_ATTRIB_TEX0][2] = 0.0F;
-      ctx->Current.Attrib[VERT_ATTRIB_TEX0][3] = 1.0F;
-   }
-
-   if (rmesa->vb.vertex_format & RADEON_CP_VC_FRMT_ST1) {
-      ctx->Current.Attrib[VERT_ATTRIB_TEX1][0] = rmesa->vb.texcoordptr[1][0];
-      ctx->Current.Attrib[VERT_ATTRIB_TEX1][1] = rmesa->vb.texcoordptr[1][1];
-      ctx->Current.Attrib[VERT_ATTRIB_TEX1][2] = 0.0F;
-      ctx->Current.Attrib[VERT_ATTRIB_TEX1][3] = 1.0F;
+   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;
@@ -264,7 +262,7 @@ static void copy_vertex( radeonContextPtr rmesa, GLuint n, GLfloat *dst )
  * 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)[15] )
+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;
@@ -356,11 +354,12 @@ static void VFMT_FALLBACK( const char *caller )
 {
    GET_CURRENT_CONTEXT(ctx);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat tmp[3][15];
+   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);
@@ -423,14 +422,11 @@ static void VFMT_FALLBACK( const char *caller )
 	 offset++;
       }
 
-      if (ind & RADEON_CP_VC_FRMT_ST0) {
-	 CALL_TexCoord2fv(GET_DISPATCH(), (&tmp[i][offset]));
-	 offset += 2;
-      }
-
-      if (ind & RADEON_CP_VC_FRMT_ST1) {
-	 CALL_MultiTexCoord2fvARB(GET_DISPATCH(), (GL_TEXTURE1_ARB, &tmp[i][offset]));
-	 offset += 2;
+      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]));
    }
@@ -441,7 +437,8 @@ static void VFMT_FALLBACK( const char *caller )
       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));
+      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) {
@@ -455,13 +452,16 @@ static void VFMT_FALLBACK( const char *caller )
    }
 
    if (ind & RADEON_CP_VC_FRMT_PKSPEC) 
-       CALL_SecondaryColor3ubEXT(GET_DISPATCH(), (rmesa->vb.specptr->red, rmesa->vb.specptr->green, rmesa->vb.specptr->blue));
-
-   if (ind & RADEON_CP_VC_FRMT_ST0) 
-      CALL_TexCoord2fv(GET_DISPATCH(), (rmesa->vb.texcoordptr[0]));
-
-   if (ind & RADEON_CP_VC_FRMT_ST1) 
-      CALL_MultiTexCoord2fvARB(GET_DISPATCH(), (GL_TEXTURE1_ARB, rmesa->vb.texcoordptr[1]));
+       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]));
+      }
+   }
 }
 
 
@@ -470,7 +470,7 @@ static void wrap_buffer( void )
 {
    GET_CURRENT_CONTEXT(ctx);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat tmp[3][15];
+   GLfloat tmp[3][RADEON_MAX_VERTEX_SIZE];
    GLuint i, nrverts;
 
    if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_PRIMS))
@@ -547,6 +547,7 @@ 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)
       return GL_FALSE;
@@ -581,35 +582,20 @@ static GLboolean check_vtx_fmt( GLcontext *ctx )
       }
    }
 
-   if (ctx->Texture.Unit[0]._ReallyEnabled) {
-      if (ctx->Texture.Unit[0].TexGenEnabled) {
-	 if (rmesa->TexGenNeedNormals[0]) {
-	    ind |= RADEON_CP_VC_FRMT_N0;
-	 }
-      } else {
-	 if (ctx->Current.Attrib[VERT_ATTRIB_TEX0][2] != 0.0F ||
-	     ctx->Current.Attrib[VERT_ATTRIB_TEX0][3] != 1.0) {
-	    if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
-	       fprintf(stderr, "%s: rq0\n", __FUNCTION__);
-	    return GL_FALSE;
-	 }
-	 ind |= RADEON_CP_VC_FRMT_ST0;
-      }
-   }
-
-   if (ctx->Texture.Unit[1]._ReallyEnabled) {
-      if (ctx->Texture.Unit[1].TexGenEnabled) {
-	 if (rmesa->TexGenNeedNormals[1]) {
-	    ind |= RADEON_CP_VC_FRMT_N0;
-	 }
-      } else {
-	 if (ctx->Current.Attrib[VERT_ATTRIB_TEX1][2] != 0.0F ||
-	     ctx->Current.Attrib[VERT_ATTRIB_TEX1][3] != 1.0) {
-	    if (RADEON_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS))
-	       fprintf(stderr, "%s: rq1\n", __FUNCTION__);
-	    return GL_FALSE;
+   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);
 	 }
-	 ind |= RADEON_CP_VC_FRMT_ST1;
       }
    }
 
@@ -628,6 +614,8 @@ static GLboolean check_vtx_fmt( GLcontext *ctx )
    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:
@@ -671,19 +659,14 @@ static GLboolean check_vtx_fmt( GLcontext *ctx )
       UNCLAMPED_FLOAT_TO_CHAN( rmesa->vb.specptr->blue,  ctx->Current.Attrib[VERT_ATTRIB_COLOR1][2] );
    }
 
-   if (ind & RADEON_CP_VC_FRMT_ST0) {
-      rmesa->vb.texcoordptr[0] = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 2;
-      rmesa->vb.texcoordptr[0][0] = ctx->Current.Attrib[VERT_ATTRIB_TEX0][0];
-      rmesa->vb.texcoordptr[0][1] = ctx->Current.Attrib[VERT_ATTRIB_TEX0][1];   
-   } 
-
-   if (ind & RADEON_CP_VC_FRMT_ST1) {
-      rmesa->vb.texcoordptr[1] = &rmesa->vb.vertex[rmesa->vb.vertex_size].f;
-      rmesa->vb.vertex_size += 2;
-      rmesa->vb.texcoordptr[1][0] = ctx->Current.Attrib[VERT_ATTRIB_TEX1][0];
-      rmesa->vb.texcoordptr[1][1] = ctx->Current.Attrib[VERT_ATTRIB_TEX1][1];
-   } 
+   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)
@@ -813,6 +796,7 @@ static void radeon_Begin( GLenum mode )
    /* 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 );
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
index 342b0b39c1..aac029aa0e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_c.c
@@ -563,7 +563,7 @@ static void radeon_MultiTexCoord1fARB( GLenum target, GLfloat s  )
 {
    GET_CURRENT_CONTEXT(ctx);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 1];
+   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
    dest[0] = s;
    dest[1] = 0;
 }
@@ -572,7 +572,7 @@ static void radeon_MultiTexCoord1fvARB( GLenum target, const GLfloat *v )
 {
    GET_CURRENT_CONTEXT(ctx);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 1];
+   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
    dest[0] = v[0];
    dest[1] = 0;
 }
@@ -581,7 +581,7 @@ 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 & 1];
+   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
    dest[0] = s;
    dest[1] = t;
 }
@@ -590,7 +590,7 @@ static void radeon_MultiTexCoord2fvARB( GLenum target, const GLfloat *v )
 {
    GET_CURRENT_CONTEXT(ctx);
    radeonContextPtr rmesa = RADEON_CONTEXT(ctx);
-   GLfloat *dest = rmesa->vb.texcoordptr[target & 1];
+   GLfloat *dest = rmesa->vb.texcoordptr[target & 3];
    dest[0] = v[0];
    dest[1] = v[1];
 }
@@ -746,7 +746,8 @@ static void choose_##FN ARGS1						\
 
 #define ACTIVE_ST0 RADEON_CP_VC_FRMT_ST0
 #define ACTIVE_ST1 RADEON_CP_VC_FRMT_ST1
-#define ACTIVE_ST_ALL (RADEON_CP_VC_FRMT_ST1|RADEON_CP_VC_FRMT_ST0)
+#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.
@@ -756,7 +757,8 @@ static void choose_##FN ARGS1						\
 #define MASK_SPEC    (MASK_COLOR|ACTIVE_COLOR)
 #define MASK_ST0     (MASK_SPEC|ACTIVE_SPEC)
 #define MASK_ST1     (MASK_ST0|ACTIVE_ST0)
-#define MASK_ST_ALL  (MASK_ST1|ACTIVE_ST1)
+#define MASK_ST2     (MASK_ST1|ACTIVE_ST1)
+#define MASK_ST_ALL  (MASK_ST2|ACTIVE_ST2)
 #define MASK_VERTEX  (MASK_ST_ALL|ACTIVE_FPALPHA) 
 
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c
index cb5d840116..0f0fc9e065 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_sse.c
@@ -170,6 +170,7 @@ static struct dynfn *radeon_makeSSETexCoord2f( GLcontext *ctx, int 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 );
@@ -207,6 +208,7 @@ static struct dynfn *radeon_makeSSEMultiTexCoord2f( GLcontext *ctx, int key )
    }
    return dfn;
 }
+#endif
 
 void radeonInitSSECodegen( struct dfn_generators *gen )
 {
@@ -217,8 +219,10 @@ void radeonInitSSECodegen( struct dfn_generators *gen )
       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
    }
 }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c
index d1cf1979f1..529e79065e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxfmt_x86.c
@@ -351,6 +351,7 @@ struct dynfn *radeon_makeX86TexCoord2f( GLcontext *ctx, int 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 );
@@ -395,7 +396,7 @@ struct dynfn *radeon_makeX86MultiTexCoord2fARB( GLcontext *ctx,
    }      
    return dfn;
 }
-
+#endif
 
 void radeonInitX86Codegen( struct dfn_generators *gen )
 {
@@ -407,8 +408,10 @@ void radeonInitX86Codegen( struct dfn_generators *gen )
    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;
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S b/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S
index 569d3b9b48..d2bb1d1519 100644
--- a/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S
+++ b/src/mesa/drivers/dri/radeon/radeon_vtxtmp_x86.S
@@ -338,7 +338,7 @@ GLOBL( _x86_Color4ub_ub_end )
 GLOBL( _x86_MultiTexCoord2fv )
 	movl	4(%esp), %eax
 	movl	8(%esp), %ecx
-	and	$1, %eax
+	and	$3, %eax
 	movl	(%ecx), %edx
 	shl	$3, %eax
 	movl	4(%ecx), %ecx
@@ -354,7 +354,7 @@ GLOBL( _x86_MultiTexCoord2fv_end )
 GLOBL( _x86_MultiTexCoord2fv_2 )
 	movl	4(%esp,1), %eax
 	movl	8(%esp,1), %ecx
-	and	$0x1, %eax
+	and	$3, %eax
 	movl	0(,%eax,4), %edx
 	movl	(%ecx), %eax
 	movl	%eax, (%edx)
@@ -370,7 +370,7 @@ GLOBL( _x86_MultiTexCoord2f )
 	movl	4(%esp), %eax
 	movl	8(%esp), %edx
 	movl	12(%esp), %ecx
-	and	$1, %eax
+	and	$3, %eax
 	shl	$3, %eax
 	movl	%edx, 0xdeadbeef(%eax)
 	movl	%ecx, 0xdeadbeef(%eax)
@@ -384,7 +384,7 @@ GLOBL( _x86_MultiTexCoord2f_2 )
 	movl	4(%esp), %eax
 	movl	8(%esp), %edx
 	movl	12(%esp,1), %ecx
-	and	$1,%eax
+	and	$3,%eax
 	movl	0(,%eax,4), %eax
 	movl	%edx, (%eax)
 	movl	%ecx, 4(%eax)
@@ -450,7 +450,7 @@ GLOBL( _sse_Attribute2f_end )
 GLOBL( _sse_MultiTexCoord2fv )
 	movl	4(%esp), %eax
 	movl	8(%esp), %ecx
-	and	$1, %eax
+	and	$3, %eax
 	movlps	(%ecx), %xmm0
 	movlps	%xmm0, 0xdeadbeef(,%eax,8)
 	ret
@@ -462,7 +462,7 @@ GLOBL( _sse_MultiTexCoord2fv_end )
 GLOBL( _sse_MultiTexCoord2fv_2 )
 	movl	4(%esp), %eax
 	movl	8(%esp), %ecx
-	and	$0x1, %eax
+	and	$3, %eax
 	movl	0(,%eax,4), %edx
 	movlps	(%ecx), %xmm0
 	movlps	%xmm0, (%edx)
@@ -474,7 +474,7 @@ GLOBL( _sse_MultiTexCoord2fv_2_end )
 */
 GLOBL( _sse_MultiTexCoord2f )
 	movl	4(%esp), %eax
-	and	$1, %eax
+	and	$3, %eax
 	movlps	8(%esp), %xmm0
 	movlps	%xmm0, 0xdeadbeef(,%eax,8)
 	ret
@@ -486,7 +486,7 @@ GLOBL( _sse_MultiTexCoord2f_end )
 GLOBL( _sse_MultiTexCoord2f_2 )
 	movl	4(%esp), %eax
 	movlps	8(%esp), %xmm0
-	and	$1,%eax
+	and	$3,%eax
 	movl	0(,%eax,4), %eax
 	movlps	%xmm0, (%eax)
 	ret