add support for more than 2 texture units (max 6, default 4). use hang workarounds only for r200, not derivatives.

13 files changed, 292 insertions, 145 deletions

diff --git a/src/mesa/drivers/dri/r200/r200_context.c b/src/mesa/drivers/dri/r200/r200_context.c
index 235522cfd6..3751ad47b2 100644
--- a/src/mesa/drivers/dri/r200/r200_context.c
+++ b/src/mesa/drivers/dri/r200/r200_context.c
@@ -330,9 +330,10 @@ GLboolean r200CreateContext( 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,
@@ -391,11 +392,11 @@ GLboolean r200CreateContext( const __GLcontextModes *glVisual,
    _tnl_allow_vertex_fog( ctx, GL_TRUE );
 
 
-   _math_matrix_ctr( &rmesa->TexGenMatrix[0] );
-   _math_matrix_ctr( &rmesa->TexGenMatrix[1] );
+   for ( i = 0 ; i < R200_MAX_TEXTURE_UNITS ; i++ ) {
+      _math_matrix_ctr( &rmesa->TexGenMatrix[i] );
+      _math_matrix_set_identity( &rmesa->TexGenMatrix[i] );
+   }
    _math_matrix_ctr( &rmesa->tmpmat );
-   _math_matrix_set_identity( &rmesa->TexGenMatrix[0] );
-   _math_matrix_set_identity( &rmesa->TexGenMatrix[1] );
    _math_matrix_set_identity( &rmesa->tmpmat );
 
    driInitExtensions( ctx, card_extensions, GL_TRUE );
diff --git a/src/mesa/drivers/dri/r200/r200_context.h b/src/mesa/drivers/dri/r200/r200_context.h
index 1ed1dfb922..8f90c15831 100644
--- a/src/mesa/drivers/dri/r200/r200_context.h
+++ b/src/mesa/drivers/dri/r200/r200_context.h
@@ -131,7 +131,11 @@ struct r200_stipple_state {
 
 #define TEX_0   0x1
 #define TEX_1   0x2
-#define TEX_ALL 0x3
+#define TEX_2	0x4
+#define TEX_3	0x8
+#define TEX_4	0x10
+#define TEX_5	0x20
+#define TEX_ALL 0x3f
 
 typedef struct r200_tex_obj r200TexObj, *r200TexObjPtr;
 
@@ -173,7 +177,7 @@ struct r200_texture_env_state {
    GLenum envMode;
 };
 
-#define R200_MAX_TEXTURE_UNITS 3
+#define R200_MAX_TEXTURE_UNITS 6
 
 struct r200_texture_state {
    struct r200_texture_env_state unit[R200_MAX_TEXTURE_UNITS];
@@ -512,11 +516,11 @@ struct r200_hw_state {
    struct r200_state_atom cst;
    struct r200_state_atom tam;
    struct r200_state_atom tf;
-   struct r200_state_atom tex[2];
-   struct r200_state_atom cube[2];
+   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[5]; 
+   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 */
@@ -769,29 +773,33 @@ struct r200_prim {
    GLuint prim;
 };
 
-struct r200_vbinfo {
-   GLint counter, initial_counter;
-   GLint *dmaptr;
-   void (*notify)( void );
-   GLint vertex_size;
-
-   /* A maximum total of 17 elements per vertex:  3 floats for position, 3
+   /* 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 (6 floats total).
+    * 3 floats for each texture unit (18 floats total).
     * 
-    * As soon as the 3rd through 6th TMUs are supported, this value will grow.
+    * 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.
     */
-   union { float f; int i; r200_color_t color; } vertex[17];
+
+#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[2];
+   GLfloat *texcoordptr[8];	/* 6 (TMU) + 2 for r200_vtxfmt_c.c when GL_TEXTURE6/7 */
 
 
    GLenum *prim;		/* &ctx->Driver.CurrentExecPrimitive */
diff --git a/src/mesa/drivers/dri/r200/r200_maos_arrays.c b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
index 7bfa316836..3429cf6400 100644
--- a/src/mesa/drivers/dri/r200/r200_maos_arrays.c
+++ b/src/mesa/drivers/dri/r200/r200_maos_arrays.c
@@ -420,9 +420,6 @@ void r200EmitArrays( GLcontext *ctx, GLuint inputs )
       vfmt0 |= R200_VTX_FP_RGB << R200_VTX_COLOR_1_SHIFT; 
       component[nr++] = &rmesa->tcl.spec;
    }
-
-/*    vtx = (rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] & */
-/* 	  ~(R200_TCL_VTX_Q0|R200_TCL_VTX_Q1|R200_TCL_VTX_Q2|R200_TCL_VTX_Q3|R200_TCL_VTX_Q4|R200_TCL_VTX_Q5)); */
       
    re_cntl = rmesa->hw.set.cmd[SET_RE_CNTL] & ~(R200_VTX_STQ0_D3D |
 						R200_VTX_STQ1_D3D |
diff --git a/src/mesa/drivers/dri/r200/r200_screen.c b/src/mesa/drivers/dri/r200/r200_screen.c
index 688c2cad2a..b34ba89e2d 100644
--- a/src/mesa/drivers/dri/r200/r200_screen.c
+++ b/src/mesa/drivers/dri/r200/r200_screen.c
@@ -64,6 +64,7 @@ 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(4,2,6)
     DRI_CONF_SECTION_END
     DRI_CONF_SECTION_QUALITY
         DRI_CONF_TEXTURE_DEPTH(DRI_CONF_TEXTURE_DEPTH_FB)
@@ -77,13 +78,13 @@ DRI_CONF_BEGIN
         DRI_CONF_NO_RAST(false)
     DRI_CONF_SECTION_END
 DRI_CONF_END;
-static const GLuint __driNConfigOptions = 10;
+static const GLuint __driNConfigOptions = 11;
 
 #if 1
 /* Including xf86PciInfo.h introduces a bunch of errors...
  */
-#define PCI_CHIP_R200_QD	0x5144
-#define PCI_CHIP_R200_QE	0x5145
+#define PCI_CHIP_R200_QD	0x5144 /* why do they have r200 names? */
+#define PCI_CHIP_R200_QE	0x5145 /* Those are all standard radeons */
 #define PCI_CHIP_R200_QF	0x5146
 #define PCI_CHIP_R200_QG	0x5147
 #define PCI_CHIP_R200_QY	0x5159
@@ -92,6 +93,7 @@ static const GLuint __driNConfigOptions = 10;
 #define PCI_CHIP_R200_LY	0x4C59
 #define PCI_CHIP_R200_LZ	0x4C5A
 #define PCI_CHIP_RV200_QW	0x5157 /* Radeon 7500 - not an R200 at all */
+#define PCI_CHIP_RV200_QX       0x5158
 #define PCI_CHIP_RS100_4136     0x4136 /* IGP RS100, RS200, RS250 are not R200 */
 #define PCI_CHIP_RS200_4137     0x4137
 #define PCI_CHIP_RS250_4237     0x4237
@@ -102,6 +104,19 @@ static const GLuint __driNConfigOptions = 10;
 #define PCI_CHIP_RS300_5835     0x5835
 #define PCI_CHIP_RS300_5836     0x5836
 #define PCI_CHIP_RS300_5837     0x5837
+#define PCI_CHIP_R200_BB        0x4242 /* r200 (non-derived) start */
+#define PCI_CHIP_R200_BC        0x4243
+#define PCI_CHIP_R200_QH        0x5148
+#define PCI_CHIP_R200_QI        0x5149
+#define PCI_CHIP_R200_QJ        0x514A
+#define PCI_CHIP_R200_QK        0x514B
+#define PCI_CHIP_R200_QL        0x514C
+#define PCI_CHIP_R200_QM        0x514D
+#define PCI_CHIP_R200_QN        0x514E
+#define PCI_CHIP_R200_QO        0x514F /* r200 (non-derived) end */
+/* are the R200 Qh (0x5168) and following needed too? They are not in xf86PciInfo.h
+   but in the pci database. Maybe just secondary ports or something ? */
+
 #endif
 
 #ifdef USE_NEW_INTERFACE
@@ -277,6 +292,7 @@ r200CreateScreen( __DRIscreenPrivate *sPriv )
    case PCI_CHIP_R200_QY:
    case PCI_CHIP_R200_QZ:
    case PCI_CHIP_RV200_QW:
+   case PCI_CHIP_RV200_QX:
    case PCI_CHIP_R200_LW:
    case PCI_CHIP_R200_LY:
    case PCI_CHIP_R200_LZ:
@@ -296,6 +312,18 @@ r200CreateScreen( __DRIscreenPrivate *sPriv )
    case PCI_CHIP_RS300_5837:
       break;
 
+   case PCI_CHIP_R200_BB:
+   case PCI_CHIP_R200_BC:
+   case PCI_CHIP_R200_QH:
+   case PCI_CHIP_R200_QI:
+   case PCI_CHIP_R200_QJ:
+   case PCI_CHIP_R200_QK:
+   case PCI_CHIP_R200_QL:
+   case PCI_CHIP_R200_QM:
+   case PCI_CHIP_R200_QN:
+   case PCI_CHIP_R200_QO:
+      screen->chipset |= R200_CHIPSET_REAL_R200;
+   /* fallthrough */
    default:
       screen->chipset |= R200_CHIPSET_TCL;
       break;
diff --git a/src/mesa/drivers/dri/r200/r200_screen.h b/src/mesa/drivers/dri/r200/r200_screen.h
index 845570daf4..95633b8ebe 100644
--- a/src/mesa/drivers/dri/r200/r200_screen.h
+++ b/src/mesa/drivers/dri/r200/r200_screen.h
@@ -53,6 +53,7 @@ typedef struct {
 
 /* chipset features */
 #define R200_CHIPSET_TCL	(1 << 0)
+#define R200_CHIPSET_REAL_R200  (1 << 1)
 
 
 #define R200_NR_TEX_HEAPS 2
diff --git a/src/mesa/drivers/dri/r200/r200_state_init.c b/src/mesa/drivers/dri/r200/r200_state_init.c
index 749cef6835..ed53c5d2cf 100644
--- a/src/mesa/drivers/dri/r200/r200_state_init.c
+++ b/src/mesa/drivers/dri/r200/r200_state_init.c
@@ -134,7 +134,9 @@ static GLboolean check_##NM( GLcontext *ctx, int idx )	\
 CHECK( always, GL_TRUE )
 CHECK( never, GL_FALSE )
 CHECK( tex_any, ctx->Texture._EnabledUnits )
+CHECK( tex_pair, (ctx->Texture.Unit[idx]._ReallyEnabled | ctx->Texture.Unit[idx & ~1]._ReallyEnabled))
 CHECK( tex, ctx->Texture.Unit[idx]._ReallyEnabled )
+CHECK( tex_cube, ctx->Texture.Unit[idx]._ReallyEnabled & TEXTURE_CUBE_BIT)
 CHECK( fog, ctx->Fog.Enabled )
 TCL_CHECK( tcl, GL_TRUE )
 TCL_CHECK( tcl_tex, ctx->Texture.Unit[idx]._ReallyEnabled )
@@ -232,18 +234,37 @@ void r200InitState( r200ContextPtr rmesa )
    ALLOC_STATE( msc, always, MSC_STATE_SIZE, "MSC/misc", 0 );
    ALLOC_STATE( cst, always, CST_STATE_SIZE, "CST/constant", 0 );
    ALLOC_STATE( zbs, always, ZBS_STATE_SIZE, "ZBS/zbias", 0 );
-   ALLOC_STATE( tam, tex_any, TAM_STATE_SIZE, "TAM/tam", 0 );
    ALLOC_STATE( tf, tex_any, TF_STATE_SIZE, "TF/tfactor", 0 );
-   ALLOC_STATE( tex[0], tex_any, TEX_STATE_SIZE, "TEX/tex-0", 0 );
-   ALLOC_STATE( tex[1], tex_any, TEX_STATE_SIZE, "TEX/tex-1", 1 );
-
+   if (rmesa->r200Screen->chipset & R200_CHIPSET_REAL_R200) {
+   /* make sure texture units 0/1 are emitted pair-wise for r200 t0 hang workaround */
+      ALLOC_STATE( tex[0], tex_pair, TEX_STATE_SIZE, "TEX/tex-0", 0 );
+      ALLOC_STATE( tex[1], tex_pair, TEX_STATE_SIZE, "TEX/tex-1", 1 );
+      ALLOC_STATE( tam, tex_any, TAM_STATE_SIZE, "TAM/tam", 0 );
+   }
+   else {
+      ALLOC_STATE( tex[0], tex, TEX_STATE_SIZE, "TEX/tex-0", 0 );
+      ALLOC_STATE( tex[1], tex, TEX_STATE_SIZE, "TEX/tex-1", 1 );
+      ALLOC_STATE( tam, never, TAM_STATE_SIZE, "TAM/tam", 0 );
+   }
+   ALLOC_STATE( tex[2], tex, TEX_STATE_SIZE, "TEX/tex-2", 2 );
+   ALLOC_STATE( tex[3], tex, TEX_STATE_SIZE, "TEX/tex-3", 3 );
+   ALLOC_STATE( tex[4], tex, TEX_STATE_SIZE, "TEX/tex-4", 4 );
+   ALLOC_STATE( tex[5], tex, TEX_STATE_SIZE, "TEX/tex-5", 5 );
    if (rmesa->r200Screen->drmSupportsCubeMaps) {
-      ALLOC_STATE( cube[0], tex_any, CUBE_STATE_SIZE, "CUBE/tex-0", 0 );
-      ALLOC_STATE( cube[1], tex_any, CUBE_STATE_SIZE, "CUBE/tex-1", 1 );
+      ALLOC_STATE( cube[0], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-0", 0 );
+      ALLOC_STATE( cube[1], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-1", 1 );
+      ALLOC_STATE( cube[2], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-2", 2 );
+      ALLOC_STATE( cube[3], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-3", 3 );
+      ALLOC_STATE( cube[4], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-4", 4 );
+      ALLOC_STATE( cube[5], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-5", 5 );
    }
    else {
       ALLOC_STATE( cube[0], never, CUBE_STATE_SIZE, "CUBE/tex-0", 0 );
       ALLOC_STATE( cube[1], never, CUBE_STATE_SIZE, "CUBE/tex-1", 1 );
+      ALLOC_STATE( cube[2], never, CUBE_STATE_SIZE, "CUBE/tex-2", 2 );
+      ALLOC_STATE( cube[3], never, CUBE_STATE_SIZE, "CUBE/tex-3", 3 );
+      ALLOC_STATE( cube[4], never, CUBE_STATE_SIZE, "CUBE/tex-4", 4 );
+      ALLOC_STATE( cube[5], never, CUBE_STATE_SIZE, "CUBE/tex-5", 5 );
    }
 
    ALLOC_STATE( tcl, tcl, TCL_STATE_SIZE, "TCL/tcl", 0 );
@@ -260,6 +281,10 @@ void r200InitState( r200ContextPtr rmesa )
    ALLOC_STATE( mat[R200_MTX_MVP], tcl, MAT_STATE_SIZE, "MAT/modelproject", 0 );
    ALLOC_STATE( mat[R200_MTX_TEX0], tcl_tex, MAT_STATE_SIZE, "MAT/texmat0", 0 );
    ALLOC_STATE( mat[R200_MTX_TEX1], tcl_tex, MAT_STATE_SIZE, "MAT/texmat1", 1 );
+   ALLOC_STATE( mat[R200_MTX_TEX2], tcl_tex, MAT_STATE_SIZE, "MAT/texmat2", 2 );
+   ALLOC_STATE( mat[R200_MTX_TEX3], tcl_tex, MAT_STATE_SIZE, "MAT/texmat3", 3 );
+   ALLOC_STATE( mat[R200_MTX_TEX4], tcl_tex, MAT_STATE_SIZE, "MAT/texmat4", 4 );
+   ALLOC_STATE( mat[R200_MTX_TEX5], tcl_tex, MAT_STATE_SIZE, "MAT/texmat5", 5 );
    ALLOC_STATE( ucp[0], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-0", 0 );
    ALLOC_STATE( ucp[1], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-1", 1 );
    ALLOC_STATE( ucp[2], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-2", 2 );
@@ -276,6 +301,10 @@ void r200InitState( r200ContextPtr rmesa )
    ALLOC_STATE( lit[7], tcl_light, LIT_STATE_SIZE, "LIT/light-7", 7 );
    ALLOC_STATE( pix[0], always, PIX_STATE_SIZE, "PIX/pixstage-0", 0 );
    ALLOC_STATE( pix[1], tex, PIX_STATE_SIZE, "PIX/pixstage-1", 1 );
+   ALLOC_STATE( pix[2], tex, PIX_STATE_SIZE, "PIX/pixstage-2", 2 );
+   ALLOC_STATE( pix[3], tex, PIX_STATE_SIZE, "PIX/pixstage-3", 3 );
+   ALLOC_STATE( pix[4], tex, PIX_STATE_SIZE, "PIX/pixstage-4", 4 );
+   ALLOC_STATE( pix[5], tex, PIX_STATE_SIZE, "PIX/pixstage-5", 5 );
 
 
    /* Fill in the packet headers:
@@ -304,12 +333,32 @@ void r200InitState( r200ContextPtr rmesa )
    rmesa->hw.tex[0].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_0);
    rmesa->hw.tex[1].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_1);
    rmesa->hw.tex[1].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_1);
+   rmesa->hw.tex[2].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_2);
+   rmesa->hw.tex[2].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_2);
+   rmesa->hw.tex[3].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_3);
+   rmesa->hw.tex[3].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_3);
+   rmesa->hw.tex[4].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_4);
+   rmesa->hw.tex[4].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_4);
+   rmesa->hw.tex[5].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_5);
+   rmesa->hw.tex[5].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_5);
    rmesa->hw.cube[0].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_0);
    rmesa->hw.cube[0].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_0);
    rmesa->hw.cube[1].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_1);
    rmesa->hw.cube[1].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_1);
+   rmesa->hw.cube[2].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_2);
+   rmesa->hw.cube[2].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_2);
+   rmesa->hw.cube[3].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_3);
+   rmesa->hw.cube[3].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_3);
+   rmesa->hw.cube[4].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_4);
+   rmesa->hw.cube[4].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_4);
+   rmesa->hw.cube[5].cmd[CUBE_CMD_0] = cmdpkt(R200_EMIT_PP_CUBIC_FACES_5);
+   rmesa->hw.cube[5].cmd[CUBE_CMD_1] = cmdpkt(R200_EMIT_PP_CUBIC_OFFSETS_5);
    rmesa->hw.pix[0].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_0);
    rmesa->hw.pix[1].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_1);
+   rmesa->hw.pix[2].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_2);
+   rmesa->hw.pix[3].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_3);
+   rmesa->hw.pix[4].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_4);
+   rmesa->hw.pix[5].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_5);
    rmesa->hw.zbs.cmd[ZBS_CMD_0] = cmdpkt(RADEON_EMIT_SE_ZBIAS_FACTOR);
    rmesa->hw.tcl.cmd[TCL_CMD_0] = cmdpkt(R200_EMIT_TCL_LIGHT_MODEL_CTL_0);
    rmesa->hw.tcl.cmd[TCL_CMD_1] = cmdpkt(R200_EMIT_TCL_UCP_VERT_BLEND_CTL);
@@ -348,6 +397,14 @@ void r200InitState( r200ContextPtr rmesa )
       cmdvec( R200_VS_MATRIX_3_TEX0, 1, 16);
    rmesa->hw.mat[R200_MTX_TEX1].cmd[MAT_CMD_0] = 
       cmdvec( R200_VS_MATRIX_4_TEX1, 1, 16);
+   rmesa->hw.mat[R200_MTX_TEX2].cmd[MAT_CMD_0] = 
+      cmdvec( R200_VS_MATRIX_5_TEX2, 1, 16);
+   rmesa->hw.mat[R200_MTX_TEX3].cmd[MAT_CMD_0] = 
+      cmdvec( R200_VS_MATRIX_6_TEX3, 1, 16);
+   rmesa->hw.mat[R200_MTX_TEX4].cmd[MAT_CMD_0] = 
+      cmdvec( R200_VS_MATRIX_7_TEX4, 1, 16);
+   rmesa->hw.mat[R200_MTX_TEX5].cmd[MAT_CMD_0] = 
+      cmdvec( R200_VS_MATRIX_8_TEX5, 1, 16);
 
    for (i = 0 ; i < 8; i++) {
       rmesa->hw.lit[i].cmd[LIT_CMD_0] = 
diff --git a/src/mesa/drivers/dri/r200/r200_swtcl.c b/src/mesa/drivers/dri/r200/r200_swtcl.c
index 80f6f5f91c..5c5c5bd75e 100644
--- a/src/mesa/drivers/dri/r200/r200_swtcl.c
+++ b/src/mesa/drivers/dri/r200/r200_swtcl.c
@@ -910,8 +910,9 @@ void r200InitSwtcl( GLcontext *ctx )
    tnl->Driver.Render.CopyPV = _tnl_copy_pv;
    tnl->Driver.Render.Interp = _tnl_interp;
 
+   /* FIXME: what are these numbers? */
    _tnl_init_vertices( ctx, ctx->Const.MaxArrayLockSize + 12, 
-		       22 * sizeof(GLfloat) );
+		       36 * sizeof(GLfloat) );
    
    rmesa->swtcl.verts = (GLubyte *)tnl->clipspace.vertex_buf;
    rmesa->swtcl.RenderIndex = ~0;
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.c b/src/mesa/drivers/dri/r200/r200_tcl.c
index 1090666dd3..85f4bc1f7d 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.c
+++ b/src/mesa/drivers/dri/r200/r200_tcl.c
@@ -419,10 +419,13 @@ static void transition_to_swtnl( GLcontext *ctx )
 
    R200_STATECHANGE( rmesa, set );
    rmesa->hw.set.cmd[SET_RE_CNTL] |= (R200_VTX_STQ0_D3D |
-				      R200_VTX_STQ1_D3D);
+				      R200_VTX_STQ1_D3D |
+				      R200_VTX_STQ2_D3D |
+				      R200_VTX_STQ3_D3D |
+				      R200_VTX_STQ4_D3D |
+				      R200_VTX_STQ5_D3D);
 }
 
-
 static void transition_to_hwtnl( GLcontext *ctx )
 {
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
@@ -454,7 +457,11 @@ static void transition_to_hwtnl( GLcontext *ctx )
 
    R200_STATECHANGE( rmesa, set );
    rmesa->hw.set.cmd[SET_RE_CNTL] &= ~(R200_VTX_STQ0_D3D |
-				       R200_VTX_STQ1_D3D);
+				       R200_VTX_STQ1_D3D |
+				       R200_VTX_STQ2_D3D |
+				       R200_VTX_STQ3_D3D |
+				       R200_VTX_STQ4_D3D |
+				       R200_VTX_STQ5_D3D);
 
 
    if (R200_DEBUG & DEBUG_FALLBACKS) 
@@ -470,7 +477,11 @@ static char *fallbackStrings[] = {
    "Texgen unit 0",
    "Texgen unit 1",
    "Texgen unit 2",
-   "User disable"
+   "Texgen unit 3",
+   "Texgen unit 4",
+   "Texgen unit 5",
+   "User disable",
+   "Bitmap as points"
 };
 
 
diff --git a/src/mesa/drivers/dri/r200/r200_tcl.h b/src/mesa/drivers/dri/r200/r200_tcl.h
index 0fdf9c22c7..f06aa8e4ff 100644
--- a/src/mesa/drivers/dri/r200/r200_tcl.h
+++ b/src/mesa/drivers/dri/r200/r200_tcl.h
@@ -55,10 +55,11 @@ extern void r200TclFallback( GLcontext *ctx, GLuint bit, GLboolean mode );
 #define R200_TCL_FALLBACK_TEXGEN_0          0x10 /* texgen, unit 0 */
 #define R200_TCL_FALLBACK_TEXGEN_1          0x20 /* texgen, unit 1 */
 #define R200_TCL_FALLBACK_TEXGEN_2          0x40 /* texgen, unit 2 */
-#define R200_TCL_FALLBACK_TCL_DISABLE       0x80 /* user disable */
-#define R200_TCL_FALLBACK_BITMAP            0x100 /* draw bitmap with points */
-
-#define R200_MAX_TCL_VERTSIZE (4*4) /* using maos now... */
+#define R200_TCL_FALLBACK_TEXGEN_3          0x80 /* texgen, unit 3 */
+#define R200_TCL_FALLBACK_TEXGEN_4          0x100 /* texgen, unit 4 */
+#define R200_TCL_FALLBACK_TEXGEN_5          0x200 /* texgen, unit 5 */
+#define R200_TCL_FALLBACK_TCL_DISABLE       0x400 /* user disable */
+#define R200_TCL_FALLBACK_BITMAP            0x800 /* draw bitmap with points */
 
 #define TCL_FALLBACK( ctx, bit, mode )	r200TclFallback( ctx, bit, mode )
 
diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index 164c17e064..eb89915357 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -290,22 +290,34 @@ static GLuint r200_register_color[][R200_MAX_TEXTURE_UNITS] =
    {
       R200_TXC_ARG_A_R0_COLOR,
       R200_TXC_ARG_A_R1_COLOR,
-      R200_TXC_ARG_A_R2_COLOR
+      R200_TXC_ARG_A_R2_COLOR,
+      R200_TXC_ARG_A_R3_COLOR,
+      R200_TXC_ARG_A_R4_COLOR,
+      R200_TXC_ARG_A_R5_COLOR
    },
    {
       R200_TXC_ARG_A_R0_COLOR | R200_TXC_COMP_ARG_A,
       R200_TXC_ARG_A_R1_COLOR | R200_TXC_COMP_ARG_A,
-      R200_TXC_ARG_A_R2_COLOR | R200_TXC_COMP_ARG_A
+      R200_TXC_ARG_A_R2_COLOR | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R3_COLOR | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R4_COLOR | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R5_COLOR | R200_TXC_COMP_ARG_A
    },
    {
       R200_TXC_ARG_A_R0_ALPHA,
       R200_TXC_ARG_A_R1_ALPHA,
-      R200_TXC_ARG_A_R2_ALPHA
+      R200_TXC_ARG_A_R2_ALPHA,
+      R200_TXC_ARG_A_R3_ALPHA,
+      R200_TXC_ARG_A_R4_ALPHA,
+      R200_TXC_ARG_A_R5_ALPHA
    },
    {
       R200_TXC_ARG_A_R0_ALPHA | R200_TXC_COMP_ARG_A,
       R200_TXC_ARG_A_R1_ALPHA | R200_TXC_COMP_ARG_A,
-      R200_TXC_ARG_A_R2_ALPHA | R200_TXC_COMP_ARG_A
+      R200_TXC_ARG_A_R2_ALPHA | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R3_ALPHA | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R4_ALPHA | R200_TXC_COMP_ARG_A,
+      R200_TXC_ARG_A_R5_ALPHA | R200_TXC_COMP_ARG_A
    },
 };
 
@@ -344,12 +356,18 @@ static GLuint r200_register_alpha[][R200_MAX_TEXTURE_UNITS] =
    {
       R200_TXA_ARG_A_R0_ALPHA,
       R200_TXA_ARG_A_R1_ALPHA,
-      R200_TXA_ARG_A_R2_ALPHA
+      R200_TXA_ARG_A_R2_ALPHA,
+      R200_TXA_ARG_A_R3_ALPHA,
+      R200_TXA_ARG_A_R4_ALPHA,
+      R200_TXA_ARG_A_R5_ALPHA
    },
    {
       R200_TXA_ARG_A_R0_ALPHA | R200_TXA_COMP_ARG_A,
       R200_TXA_ARG_A_R1_ALPHA | R200_TXA_COMP_ARG_A,
-      R200_TXA_ARG_A_R2_ALPHA | R200_TXA_COMP_ARG_A
+      R200_TXA_ARG_A_R2_ALPHA | R200_TXA_COMP_ARG_A,
+      R200_TXA_ARG_A_R3_ALPHA | R200_TXA_COMP_ARG_A,
+      R200_TXA_ARG_A_R4_ALPHA | R200_TXA_COMP_ARG_A,
+      R200_TXA_ARG_A_R5_ALPHA | R200_TXA_COMP_ARG_A
    },
 };
 
@@ -428,11 +446,12 @@ static GLboolean r200UpdateTextureEnv( GLcontext *ctx, int unit )
     * reduces the amount of special-casing we have to do, alpha-only
     * textures being a notable exception.
     */
+   /* Don't cache these results.
+    */
+   rmesa->state.texture.unit[unit].format = 0;
+   rmesa->state.texture.unit[unit].envMode = 0;
+
    if ( !texUnit->_ReallyEnabled ) {
-      /* Don't cache these results.
-       */
-      rmesa->state.texture.unit[unit].format = 0;
-      rmesa->state.texture.unit[unit].envMode = 0;
       if ( unit == 0 ) {
 	 color_combine = R200_TXC_ARG_A_ZERO | R200_TXC_ARG_B_ZERO
 	     | R200_TXC_ARG_C_DIFFUSE_COLOR | R200_TXC_OP_MADD;
@@ -454,11 +473,6 @@ static GLboolean r200UpdateTextureEnv( GLcontext *ctx, int unit )
       GLuint RGBshift = texUnit->_CurrentCombine->ScaleShiftRGB;
       GLuint Ashift = texUnit->_CurrentCombine->ScaleShiftA;
 
-      /* Don't cache these results.
-       */
-      rmesa->state.texture.unit[unit].format = 0;
-      rmesa->state.texture.unit[unit].envMode = 0;
-
 
       /* Step 1:
        * Extract the color and alpha combine function arguments.
@@ -1249,75 +1263,102 @@ void r200UpdateTextureState( GLcontext *ctx )
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
    GLboolean ok;
    GLuint dbg;
+   int i;
 
    ok = (r200UpdateTextureUnit( ctx, 0 ) &&
-	 r200UpdateTextureUnit( ctx, 1 ));
+	 r200UpdateTextureUnit( ctx, 1 ) &&
+	 r200UpdateTextureUnit( ctx, 2 ) &&
+	 r200UpdateTextureUnit( ctx, 3 ) &&
+	 r200UpdateTextureUnit( ctx, 4 ) &&
+	 r200UpdateTextureUnit( ctx, 5 ));
 
    FALLBACK( rmesa, R200_FALLBACK_TEXTURE, !ok );
 
    if (rmesa->TclFallback)
       r200ChooseVertexState( ctx );
 
-   /*
-    * T0 hang workaround -------------
-    */
-#if 1
-   if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_ENABLE_MASK) == R200_TEX_0_ENABLE &&
-       (rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK) > R200_MIN_FILTER_LINEAR) {
-
-      R200_STATECHANGE(rmesa, ctx);
-      R200_STATECHANGE(rmesa, tex[1]);
-      rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_TEX_1_ENABLE;
-      rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~TEXOBJ_TXFORMAT_MASK;
-      rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] |= 0x08000000;
-   }
-   else {
-      if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_1_ENABLE) &&
-	  (rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] & 0x08000000)) {
-	 R200_STATECHANGE(rmesa, tex[1]);
-	 rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~0x08000000;
-      }
-   }
-#endif
-
-#if 1
-   /*
-    * Texture cache LRU hang workaround -------------
-    */
-   dbg = 0x0;
-   if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_0_ENABLE) &&
-	((((rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 
-	  0x04) == 0)))
-   {
-      dbg |= 0x02;
-   }
 
-   if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_1_ENABLE) &&
-	((((rmesa->hw.tex[1].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 
-	  0x04) == 0)))
-   {
-      dbg |= 0x04;
-   }
+   if (rmesa->r200Screen->chipset & R200_CHIPSET_REAL_R200) {
 
-   if (dbg != rmesa->hw.tam.cmd[TAM_DEBUG3]) {
-      R200_STATECHANGE( rmesa, tam );
-      rmesa->hw.tam.cmd[TAM_DEBUG3] = dbg;
-      if (0) printf("TEXCACHE LRU HANG WORKAROUND %x\n", dbg);
-   }
-#endif
-}
+      /*
+       * T0 hang workaround -------------
+       * not needed for r200 derivatives?
+       */
+      if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_ENABLE_MASK) == R200_TEX_0_ENABLE &&
+         (rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK) > R200_MIN_FILTER_LINEAR) {
+
+         R200_STATECHANGE(rmesa, ctx);
+         R200_STATECHANGE(rmesa, tex[1]);
+         rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_TEX_1_ENABLE;
+         rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~TEXOBJ_TXFORMAT_MASK;
+         rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] |= 0x08000000;
+      }
+      else {
+         if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_1_ENABLE) &&
+            (rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] & 0x08000000)) {
+               R200_STATECHANGE(rmesa, tex[1]);
+               rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~0x08000000;
+         }
+      }
 
-/*
-  also tests for higher texunits:
+      /* maybe needs to be done pairwise due to 2 parallel (physical) tex units ?
+         looks like that's not the case, if 8500/9100 owners don't complain remove this...
+      for ( i = 0; i < ctx->Const.MaxTextureUnits; i += 2) {
+         if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & ((R200_TEX_0_ENABLE |
+            R200_TEX_1_ENABLE ) << i)) == (R200_TEX_0_ENABLE << i)) &&
+            ((rmesa->hw.tex[i].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK) >
+            R200_MIN_FILTER_LINEAR)) {
+            R200_STATECHANGE(rmesa, ctx);
+            R200_STATECHANGE(rmesa, tex[i+1]);
+            rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= (R200_TEX_1_ENABLE << i);
+            rmesa->hw.tex[i+1].cmd[TEX_PP_TXFORMAT] &= ~TEXOBJ_TXFORMAT_MASK;
+            rmesa->hw.tex[i+1].cmd[TEX_PP_TXFORMAT] |= 0x08000000;
+         }
+         else {
+            if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & (R200_TEX_1_ENABLE << i)) &&
+               (rmesa->hw.tex[i+1].cmd[TEX_PP_TXFORMAT] & 0x08000000)) {
+               R200_STATECHANGE(rmesa, tex[i+1]);
+               rmesa->hw.tex[i+1].cmd[TEX_PP_TXFORMAT] &= ~0x08000000;
+            }
+         }
+      } */
 
-       ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_2_ENABLE) &&
-	((((rmesa->hw.tex[2].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)) ||
-       ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_4_ENABLE) &&
-	((((rmesa->hw.tex[4].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)))
+      /*
+       * Texture cache LRU hang workaround -------------
+       * not needed for r200 derivatives?
+       */
+      dbg = 0x0;
+
+      if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & (R200_TEX_0_ENABLE )) &&
+         ((((rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)) ||
+         ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_2_ENABLE) &&
+         ((((rmesa->hw.tex[2].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)) ||
+         ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_4_ENABLE) &&
+         ((((rmesa->hw.tex[4].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)))
+      {
+         dbg |= 0x02;
+      }
 
-       ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_3_ENABLE) &&
-	((((rmesa->hw.tex[3].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)) ||
-       ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_5_ENABLE) &&
-	((((rmesa->hw.tex[5].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)))
+      if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & (R200_TEX_1_ENABLE )) &&
+         ((((rmesa->hw.tex[1].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)) ||
+         ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_3_ENABLE) &&
+         ((((rmesa->hw.tex[3].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)) ||
+         ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_5_ENABLE) &&
+         ((((rmesa->hw.tex[5].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) &
+         0x04) == 0)))
+      {
+         dbg |= 0x04;
+      }
 
-*/
+      if (dbg != rmesa->hw.tam.cmd[TAM_DEBUG3]) {
+         R200_STATECHANGE( rmesa, tam );
+         rmesa->hw.tam.cmd[TAM_DEBUG3] = dbg;
+         if (0) printf("TEXCACHE LRU HANG WORKAROUND %x\n", dbg);
+      }
+   }
+}
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt.c b/src/mesa/drivers/dri/r200/r200_vtxfmt.c
index ee313610b9..bd0003e127 100644
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt.c
+++ b/src/mesa/drivers/dri/r200/r200_vtxfmt.c
@@ -305,7 +305,7 @@ static void copy_vertex( r200ContextPtr 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( r200ContextPtr rmesa, GLfloat (*tmp)[15] )
+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) - 
@@ -441,7 +441,7 @@ static void VFMT_FALLBACK( const char *caller )
 {
    GET_CURRENT_CONTEXT(ctx);
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat tmp[3][15];
+   GLfloat tmp[3][R200_MAX_VERTEX_SIZE];
    GLuint i, prim;
    GLuint ind0 = rmesa->vb.vtxfmt_0;
    GLuint ind1 = rmesa->vb.vtxfmt_1;
@@ -510,11 +510,7 @@ static void VFMT_FALLBACK( const char *caller )
 	 offset++;
       }
 
-      count = VTX_TEXn_COUNT( ind1, 0 );
-      dispatch_texcoord( count, &tmp[i][offset] );
-      offset += count;
-
-      for ( unit = 1 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
+      for ( unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
 	 count = VTX_TEXn_COUNT( ind1, unit );
 	 dispatch_multitexcoord( count, unit, &tmp[i][offset] );
 	 offset += count;
@@ -554,10 +550,7 @@ static void VFMT_FALLBACK( const char *caller )
 				      rmesa->vb.specptr->green,
 				      rmesa->vb.specptr->blue ); 
 
-   count = VTX_TEXn_COUNT( ind1, 0 );
-   dispatch_texcoord( count, rmesa->vb.texcoordptr[0] );
-
-   for ( unit = 1 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
+   for ( unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++ ) {
       count = VTX_TEXn_COUNT( ind1, unit );
       dispatch_multitexcoord( count, unit, rmesa->vb.texcoordptr[unit] );
    }
@@ -569,7 +562,7 @@ static void wrap_buffer( void )
 {
    GET_CURRENT_CONTEXT(ctx);
    r200ContextPtr rmesa = R200_CONTEXT(ctx);
-   GLfloat tmp[3][15];
+   GLfloat tmp[3][R200_MAX_VERTEX_SIZE];
    GLuint i, nrverts;
 
    if (R200_DEBUG & (DEBUG_VFMT|DEBUG_PRIMS))
@@ -757,6 +750,12 @@ 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_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:
diff --git a/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c b/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c
index e9f66fb5d7..dcd7daf763 100644
--- a/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c
+++ b/src/mesa/drivers/dri/r200/r200_vtxfmt_c.c
@@ -537,7 +537,7 @@ TEX_to_nF( 3fv, (const GLfloat * v),               v[0], v[1], v[2] )
  * \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
+ * with 0x1F.  Masking with 0x1F and then masking with 0x07 is redundant, so
  * the subtraction has been omitted.
  */
 
@@ -558,13 +558,14 @@ TEX_to_nF( 3fv, (const GLfloat * v),               v[0], v[1], v[2] )
       }									\
    }
 
-MTEX_to_nF( 1f, (GLenum target, GLfloat s), (target & 1), s, 0.0, 0.0 )
-MTEX_to_nF( 2f, (GLenum target, GLfloat s, GLfloat t), (target & 1), s, t, 0.0 )
-MTEX_to_nF( 3f, (GLenum target, GLfloat s, GLfloat t, GLfloat r), (target & 1), s, t, r )
+/* \todo maybe (target & 4 ? target & 5 : target & 3) is more save than (target & 7) */
+MTEX_to_nF( 1f, (GLenum target, GLfloat s), (target & 7), s, 0.0, 0.0 )
+MTEX_to_nF( 2f, (GLenum target, GLfloat s, GLfloat t), (target & 7), s, t, 0.0 )
+MTEX_to_nF( 3f, (GLenum target, GLfloat s, GLfloat t, GLfloat r), (target & 7), s, t, r )
 
-MTEX_to_nF( 1fv, (GLenum target, const GLfloat *v), (target & 1), v[0], 0.0, 0.0 )
-MTEX_to_nF( 2fv, (GLenum target, const GLfloat *v), (target & 1), v[0], v[1], 0.0 )
-MTEX_to_nF( 3fv, (GLenum target, const GLfloat *v), (target & 1), v[0], v[1], v[2] )
+MTEX_to_nF( 1fv, (GLenum target, const GLfloat *v), (target & 7), v[0], 0.0, 0.0 )
+MTEX_to_nF( 2fv, (GLenum target, const GLfloat *v), (target & 7), v[0], v[1], 0.0 )
+MTEX_to_nF( 3fv, (GLenum target, const GLfloat *v), (target & 7), v[0], v[1], v[2] )
 
 static struct dynfn *lookup( struct dynfn *l, const int *key )
 {
@@ -722,7 +723,7 @@ static void choose_##FN ARGS1						\
 /* 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 );
diff --git a/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S b/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S
index 4ad2b6f1f0..57a35c657b 100644
--- a/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S
+++ b/src/mesa/drivers/dri/r200/r200_vtxtmp_x86.S
@@ -332,13 +332,14 @@ GLOBL( _x86_Color4ub_ub )
 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	$1, %eax
+	and	$7, %eax
 	movl	(%ecx), %edx
 	shl	$3, %eax
 	movl	4(%ecx), %ecx
@@ -354,7 +355,7 @@ GLOBL( _x86_MultiTexCoord2fv_end )
 GLOBL( _x86_MultiTexCoord2fv_2 )
 	movl	4(%esp,1), %eax
 	movl	8(%esp,1), %ecx
-	and	$0x1, %eax
+	and	$0x7, %eax
 	movl	0(,%eax,4), %edx
 	movl	(%ecx), %eax
 	movl	%eax, (%edx)
@@ -370,7 +371,7 @@ GLOBL( _x86_MultiTexCoord2f )
 	movl	4(%esp), %eax
 	movl	8(%esp), %edx
 	movl	12(%esp), %ecx
-	and	$1, %eax
+	and	$7, %eax
 	shl	$3, %eax
 	movl	%edx, 0xdeadbeef(%eax)
 	movl	%ecx, 0xdeadbeef(%eax)
@@ -384,7 +385,7 @@ GLOBL( _x86_MultiTexCoord2f_2 )
 	movl	4(%esp), %eax
 	movl	8(%esp), %edx
 	movl	12(%esp,1), %ecx
-	and	$1,%eax
+	and	$7,%eax
 	movl	0(,%eax,4), %eax
 	movl	%edx, (%eax)
 	movl	%ecx, 4(%eax)
@@ -450,7 +451,7 @@ GLOBL( _sse_Attribute2f_end )
 GLOBL( _sse_MultiTexCoord2fv )
 	movl	4(%esp), %eax
 	movl	8(%esp), %ecx
-	and	$1, %eax
+	and	$7, %eax
 	movlps	(%ecx), %xmm0
 	movlps	%xmm0, 0xdeadbeef(,%eax,8)
 	ret
@@ -462,7 +463,7 @@ GLOBL( _sse_MultiTexCoord2fv_end )
 GLOBL( _sse_MultiTexCoord2fv_2 )
 	movl	4(%esp), %eax
 	movl	8(%esp), %ecx
-	and	$0x1, %eax
+	and	$0x7, %eax
 	movl	0(,%eax,4), %edx
 	movlps	(%ecx), %xmm0
 	movlps	%xmm0, (%edx)
@@ -474,7 +475,7 @@ GLOBL( _sse_MultiTexCoord2fv_2_end )
 */
 GLOBL( _sse_MultiTexCoord2f )
 	movl	4(%esp), %eax
-	and	$1, %eax
+	and	$7, %eax
 	movlps	8(%esp), %xmm0
 	movlps	%xmm0, 0xdeadbeef(,%eax,8)
 	ret
@@ -486,7 +487,7 @@ GLOBL( _sse_MultiTexCoord2f_end )
 GLOBL( _sse_MultiTexCoord2f_2 )
 	movl	4(%esp), %eax
 	movlps	8(%esp), %xmm0
-	and	$1,%eax
+	and	$7,%eax
 	movl	0(,%eax,4), %eax
 	movlps	%xmm0, (%eax)
 	ret