Merge branch 'master' of ssh://git.freedesktop.org/git/mesa/mesa into r600_state_predict

25 files changed, 293 insertions, 217 deletions

diff --git a/src/mesa/drivers/dri/r300/r300_context.c b/src/mesa/drivers/dri/r300/r300_context.c
index 15c1eeb0d4d..0bf002caeab 100644
--- a/src/mesa/drivers/dri/r300/r300_context.c
+++ b/src/mesa/drivers/dri/r300/r300_context.c
@@ -238,8 +238,8 @@ static void r300_emit_query_finish(radeonContextPtr radeon)
 	struct radeon_query_object *query = radeon->query.current;
 	BATCH_LOCALS(radeon);
 
-	BEGIN_BATCH_NO_AUTOSTATE(3 * 2 *r300->num_z_pipes + 2);
-	switch (r300->num_z_pipes) {
+	BEGIN_BATCH_NO_AUTOSTATE(3 * 2 *r300->radeon.radeonScreen->num_gb_pipes + 2);
+	switch (r300->radeon.radeonScreen->num_gb_pipes) {
 	case 4:
 		OUT_BATCH_REGVAL(R300_SU_REG_DEST, R300_RASTER_PIPE_SELECT_3);
 		OUT_BATCH_REGSEQ(R300_ZB_ZPASS_ADDR, 1);
@@ -265,7 +265,7 @@ static void r300_emit_query_finish(radeonContextPtr radeon)
 	}
 	OUT_BATCH_REGVAL(R300_SU_REG_DEST, R300_RASTER_PIPE_SELECT_ALL);
 	END_BATCH();
-	query->curr_offset += r300->num_z_pipes * sizeof(uint32_t);
+	query->curr_offset += r300->radeon.radeonScreen->num_gb_pipes * sizeof(uint32_t);
 	assert(query->curr_offset < RADEON_QUERY_PAGE_SIZE);
 	query->emitted_begin = GL_FALSE;
 }
@@ -287,14 +287,12 @@ static void rv530_emit_query_finish_single_z(radeonContextPtr radeon)
 	query->emitted_begin = GL_FALSE;
 }
 
-#if 0
 static void rv530_emit_query_finish_double_z(radeonContextPtr radeon)
 {
-	r300ContextPtr r300 = (r300ContextPtr)radeon;
 	BATCH_LOCALS(radeon);
 	struct radeon_query_object *query = radeon->query.current;
 
-	BEGIN_BATCH_NO_AUTOSTATE(6);
+	BEGIN_BATCH_NO_AUTOSTATE(14);
 	OUT_BATCH_REGVAL(RV530_FG_ZBREG_DEST, RV530_FG_ZBREG_DEST_PIPE_SELECT_0);
 	OUT_BATCH_REGSEQ(R300_ZB_ZPASS_ADDR, 1);
 	OUT_BATCH_RELOC(0, query->bo, query->curr_offset, 0, RADEON_GEM_DOMAIN_GTT, 0);
@@ -308,7 +306,6 @@ static void rv530_emit_query_finish_double_z(radeonContextPtr radeon)
 	assert(query->curr_offset < RADEON_QUERY_PAGE_SIZE);
 	query->emitted_begin = GL_FALSE;
 }
-#endif
 
 static void r300_init_vtbl(radeonContextPtr radeon)
 {
@@ -318,11 +315,12 @@ static void r300_init_vtbl(radeonContextPtr radeon)
 	radeon->vtbl.swtcl_flush = r300_swtcl_flush;
 	radeon->vtbl.pre_emit_atoms = r300_vtbl_pre_emit_atoms;
 	radeon->vtbl.fallback = r300_fallback;
-	if (radeon->radeonScreen->chip_family == CHIP_FAMILY_RV530)
-		/* single Z gives me correct results on my hw need to check if we ever need
-		 * double z */
-		radeon->vtbl.emit_query_finish = rv530_emit_query_finish_single_z;
-	else
+	if (radeon->radeonScreen->chip_family == CHIP_FAMILY_RV530) {
+		if (radeon->radeonScreen->num_z_pipes == 2)
+			radeon->vtbl.emit_query_finish = rv530_emit_query_finish_double_z;
+		else
+			radeon->vtbl.emit_query_finish = rv530_emit_query_finish_single_z;
+	} else
 		radeon->vtbl.emit_query_finish = r300_emit_query_finish;
 }
 
@@ -396,10 +394,6 @@ static void r300InitConstValues(GLcontext *ctx, radeonScreenPtr screen)
 		ctx->Const.FragmentProgram.MaxNativeAddressRegs = 0;
 	}
 
-	if (r300->radeon.radeonScreen->chip_family == CHIP_FAMILY_RV530)
-		r300->num_z_pipes = 2;
-	else
-		r300->num_z_pipes = r300->radeon.radeonScreen->num_gb_pipes;
 }
 
 static void r300ParseOptions(r300ContextPtr r300, radeonScreenPtr screen)
@@ -436,7 +430,7 @@ static void r300InitGLExtensions(GLcontext *ctx)
 	if (r300->options.stencil_two_side_disabled)
 		_mesa_disable_extension(ctx, "GL_EXT_stencil_two_side");
 
-	if (ctx->Mesa_DXTn && !r300->options.s3tc_force_enabled) {
+	if (r300->options.s3tc_force_enabled) {
 		_mesa_enable_extension(ctx, "GL_EXT_texture_compression_s3tc");
 		_mesa_enable_extension(ctx, "GL_S3_s3tc");
 	} else if (r300->options.s3tc_force_disabled) {
diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 3202c96c2ed..1dadcc0a697 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -529,7 +529,6 @@ struct r300_context {
 	uint32_t fallback;
 
 	DECLARE_RENDERINPUTS(render_inputs_bitset);
-	int num_z_pipes;
 };
 
 #define R300_CONTEXT(ctx)		((r300ContextPtr)(ctx->DriverCtx))
diff --git a/src/mesa/drivers/dri/r600/r600_context.c b/src/mesa/drivers/dri/r600/r600_context.c
index 0b0c4f5049b..6a066f35105 100644
--- a/src/mesa/drivers/dri/r600/r600_context.c
+++ b/src/mesa/drivers/dri/r600/r600_context.c
@@ -185,13 +185,7 @@ static void r600_vtbl_emit_cs_header(struct radeon_cs *cs, radeonContextPtr rmes
 
 static void r600_vtbl_pre_emit_atoms(radeonContextPtr radeon)
 {
-	context_t *context = (context_t *)radeon;
-
-	/* always emit CB base to prevent
-	 * lock ups on some chips.
-	 */
-	R600_STATECHANGE(context, cb_target);
-	r700Start3D(context);
+	r700Start3D((context_t *)radeon);
 }
 
 static void r600_fallback(GLcontext *ctx, GLuint bit, GLboolean mode)
diff --git a/src/mesa/drivers/dri/r600/r700_chip.c b/src/mesa/drivers/dri/r600/r700_chip.c
index 3f11cf2c981..16f1a3df366 100644
--- a/src/mesa/drivers/dri/r600/r700_chip.c
+++ b/src/mesa/drivers/dri/r600/r700_chip.c
@@ -230,6 +230,9 @@ static void r700SendVTXState(GLcontext *ctx, struct radeon_state_atom *atom)
     unsigned int i, j = 0;
     BATCH_LOCALS(&context->radeon);
 
+    if (context->radeon.tcl.aos_count == 0)
+	    return;
+
     BEGIN_BATCH_NO_AUTOSTATE(6);
     R600_OUT_BATCH(CP_PACKET3(R600_IT_SET_CTL_CONST, 1));
     R600_OUT_BATCH(mmSQ_VTX_BASE_VTX_LOC - ASIC_CTL_CONST_BASE_INDEX);
@@ -989,11 +992,60 @@ static int check_always(GLcontext *ctx, struct radeon_state_atom *atom)
 	return atom->cmd_size;
 }
 
+static int check_cb(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+	context_t *context = R700_CONTEXT(ctx);
+	int count = 7;
+
+	if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
+		count += 11;
+
+	return count;
+}
+
+static int check_blnd(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+	context_t *context = R700_CONTEXT(ctx);
+	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+	unsigned int ui;
+	int count = 3;
+
+	if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
+		count += 3;
+
+	if (context->radeon.radeonScreen->chip_family > CHIP_FAMILY_R600) {
+		for (ui = 0; ui < R700_MAX_RENDER_TARGETS; ui++) {
+                        if (r700->render_target[ui].enabled)
+				count += 3;
+		}
+	}
+
+	return count;
+}
+
+static int check_ucp(GLcontext *ctx, struct radeon_state_atom *atom)
+{
+	context_t *context = R700_CONTEXT(ctx);
+	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+	int i;
+	int count = 0;
+
+	for (i = 0; i < R700_MAX_UCP; i++) {
+		if (r700->ucp[i].enabled)
+			count += 6;
+	}
+	return count;
+}
+
 static int check_vtx(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	context_t *context = R700_CONTEXT(ctx);
+	int count = context->radeon.tcl.aos_count * 18;
+
+	if (count)
+		count += 6;
 
-	return context->radeon.tcl.aos_count * 18;
+	return count;
 }
 
 static int check_tx(GLcontext *ctx, struct radeon_state_atom *atom)
@@ -1014,16 +1066,24 @@ static int check_ps_consts(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	context_t *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+	int count = r700->ps.num_consts * 4;
 
-	return 2 + (r700->ps.num_consts * 4);
+	if (count)
+		count += 2;
+
+	return count;
 }
 
 static int check_vs_consts(GLcontext *ctx, struct radeon_state_atom *atom)
 {
 	context_t *context = R700_CONTEXT(ctx);
 	R700_CHIP_CONTEXT *r700 = (R700_CHIP_CONTEXT*)(&context->hw);
+	int count = r700->vs.num_consts * 4;
+
+	if (count)
+		count += 2;
 
-	return 2 + (r700->vs.num_consts * 4);
+	return count;
 }
 
 #define ALLOC_STATE( ATOM, CHK, SZ, EMIT )				\
@@ -1056,12 +1116,12 @@ void r600InitAtoms(context_t *context)
 	ALLOC_STATE(aa, always, 12, r700SendAAState);
 	ALLOC_STATE(cl, always, 12, r700SendCLState);
 	ALLOC_STATE(gb, always, 6, r700SendGBState);
-	ALLOC_STATE(ucp, always, 36, r700SendUCPState);
+	ALLOC_STATE(ucp, ucp, (R700_MAX_UCP * 6), r700SendUCPState);
 	ALLOC_STATE(su, always, 9, r700SendSUState);
 	ALLOC_STATE(poly, always, 10, r700SendPolyState);
-	ALLOC_STATE(cb, always, 18, r700SendCBState);
+	ALLOC_STATE(cb, cb, 18, r700SendCBState);
 	ALLOC_STATE(clrcmp, always, 6, r700SendCBCLRCMPState);
-	ALLOC_STATE(blnd, always, 30, r700SendCBBlendState);
+	ALLOC_STATE(blnd, blnd, (6 + (R700_MAX_RENDER_TARGETS * 3)), r700SendCBBlendState);
 	ALLOC_STATE(blnd_clr, always, 6, r700SendCBBlendColorState);
 	ALLOC_STATE(cb_target, always, 25, r700SendRenderTargetState);
 	ALLOC_STATE(sx, always, 9, r700SendSXState);
@@ -1073,7 +1133,7 @@ void r600InitAtoms(context_t *context)
 	ALLOC_STATE(ps, always, 21, r700SendPSState);
 	ALLOC_STATE(vs_consts, vs_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendVSConsts);
 	ALLOC_STATE(ps_consts, ps_consts, (2 + (R700_MAX_DX9_CONSTS * 4)), r700SendPSConsts);
-	ALLOC_STATE(vtx, vtx, (VERT_ATTRIB_MAX * 18), r700SendVTXState);
+	ALLOC_STATE(vtx, vtx, (6 + (VERT_ATTRIB_MAX * 18)), r700SendVTXState);
 	ALLOC_STATE(tx, tx, (R700_TEXTURE_NUMBERUNITS * 20), r700SendTexState);
 	ALLOC_STATE(tx_smplr, tx, (R700_TEXTURE_NUMBERUNITS * 5), r700SendTexSamplerState);
 	ALLOC_STATE(tx_brdr_clr, tx, (R700_TEXTURE_NUMBERUNITS * 6), r700SendTexBorderColorState);
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.c b/src/mesa/drivers/dri/r600/r700_fragprog.c
index 098b420dfcc..e4a6d4cedf8 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.c
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.c
@@ -121,13 +121,13 @@ void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
 		pAsm->pR700Shader->depthIsExported = 1;
 	}
 
-    pAsm->pucOutMask = (unsigned char*) MALLOC(pAsm->number_of_exports);    
+    pAsm->pucOutMask = (unsigned char*) MALLOC(pAsm->number_of_exports);
     for(ui=0; ui<pAsm->number_of_exports; ui++)
     {
         pAsm->pucOutMask[ui] = 0x0;
     }
-	
-	pAsm->uFirstHelpReg = pAsm->number_used_registers;
+
+    pAsm->uFirstHelpReg = pAsm->number_used_registers;
 }
 
 GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
@@ -258,6 +258,19 @@ GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
 	return GL_TRUE;
 }
 
+void r700SelectFragmentShader(GLcontext *ctx)
+{
+    context_t *context = R700_CONTEXT(ctx);
+    struct r700_fragment_program *fp = (struct r700_fragment_program *)
+	    (ctx->FragmentProgram._Current);
+    if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
+    {
+	    fp->r700AsmCode.bR6xx = 1;
+    }
+
+    r700TranslateFragmentShader(fp, &(fp->mesa_program));
+}
+
 void * r700GetActiveFpShaderBo(GLcontext * ctx)
 {
     struct r700_fragment_program *fp = (struct r700_fragment_program *)
@@ -283,7 +296,7 @@ GLboolean r700SetupFragmentProgram(GLcontext * ctx)
 
     if(GL_FALSE == fp->loaded)
     {
-        if(fp->r700Shader.bNeedsAssembly == GL_TRUE)
+	    if(fp->r700Shader.bNeedsAssembly == GL_TRUE)
 	    {
 		    Assemble( &(fp->r700Shader) );
 	    }
diff --git a/src/mesa/drivers/dri/r600/r700_fragprog.h b/src/mesa/drivers/dri/r600/r700_fragprog.h
index 9c7813e9084..cbb108d2128 100644
--- a/src/mesa/drivers/dri/r600/r700_fragprog.h
+++ b/src/mesa/drivers/dri/r600/r700_fragprog.h
@@ -49,13 +49,16 @@ struct r700_fragment_program
 
 /* Internal */
 void Map_Fragment_Program(r700_AssemblerBase         *pAsm,
-						  struct gl_fragment_program *mesa_fp);
+			  struct gl_fragment_program *mesa_fp);
 GLboolean Find_Instruction_Dependencies_fp(struct r700_fragment_program *fp,
-					                	   struct gl_fragment_program   *mesa_fp);
+					   struct gl_fragment_program   *mesa_fp);
+
+GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
+				      struct gl_fragment_program   *mesa_vp);
 
 /* Interface */
-extern GLboolean r700TranslateFragmentShader(struct r700_fragment_program *fp,
-							                 struct gl_fragment_program   *mesa_vp);
+extern void r700SelectFragmentShader(GLcontext *ctx);
+
 extern GLboolean r700SetupFragmentProgram(GLcontext * ctx);
 
 extern void *    r700GetActiveFpShaderBo(GLcontext * ctx);
diff --git a/src/mesa/drivers/dri/r600/r700_render.c b/src/mesa/drivers/dri/r600/r700_render.c
index 9b0b5443d6a..d64e921bdad 100644
--- a/src/mesa/drivers/dri/r600/r700_render.c
+++ b/src/mesa/drivers/dri/r600/r700_render.c
@@ -347,6 +347,13 @@ static GLboolean r700RunRender(GLcontext * ctx,
         fprintf(stderr, "%s: cs begin at %d\n",
                 __func__, context->radeon.cmdbuf.cs->cdw);
 
+    /* always emit CB base to prevent
+     * lock ups on some chips.
+     */
+    R600_STATECHANGE(context, cb_target);
+    /* mark vtx as dirty since it changes per-draw */
+    R600_STATECHANGE(context, vtx);
+
     r700UpdateShaders(ctx);
     r700SetScissor(context);
     r700SetupVertexProgram(ctx);
diff --git a/src/mesa/drivers/dri/r600/r700_state.c b/src/mesa/drivers/dri/r600/r700_state.c
index 15f40b2771e..efa1daf2d3c 100644
--- a/src/mesa/drivers/dri/r600/r700_state.c
+++ b/src/mesa/drivers/dri/r600/r700_state.c
@@ -71,65 +71,37 @@ void r700SetDefaultStates(context_t *context) //--------------------
 void r700UpdateShaders (GLcontext * ctx)  //----------------------------------
 {
     context_t *context = R700_CONTEXT(ctx);
-
     GLvector4f dummy_attrib[_TNL_ATTRIB_MAX];
     GLvector4f *temp_attrib[_TNL_ATTRIB_MAX];
     int i;
 
-    if (ctx->FragmentProgram._Current) {
-	    struct r700_fragment_program *fp = (struct r700_fragment_program *)
-		    (ctx->FragmentProgram._Current);
-	    if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
-	    {
-		    fp->r700AsmCode.bR6xx = 1;
-	    }
-
-	    if(GL_FALSE == fp->translated)
-	    {
-		    if( GL_FALSE == r700TranslateFragmentShader(fp, &(fp->mesa_program)) )
-		    {
-			    //return GL_TRUE;
-		    }
-	    }
+    /* should only happenen once, just after context is created */
+    /* TODO: shouldn't we fallback to sw here? */
+    if (!ctx->FragmentProgram._Current) {
+	    _mesa_fprintf(stderr, "No ctx->FragmentProgram._Current!!\n");
+	    return;
     }
 
-    if (context->radeon.NewGLState) 
-    {
-	struct r700_vertex_program *vp;
-        context->radeon.NewGLState = 0;
-
-        for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) 
-        {
-            /* mat states from state var not array for sw */
-            dummy_attrib[i].stride = 0;
+    r700SelectFragmentShader(ctx);
 
-            temp_attrib[i] = TNL_CONTEXT(ctx)->vb.AttribPtr[i];
-            TNL_CONTEXT(ctx)->vb.AttribPtr[i] = &(dummy_attrib[i]);
-        }
-
-        _tnl_UpdateFixedFunctionProgram(ctx);
-
-        for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) 
-        {
-            TNL_CONTEXT(ctx)->vb.AttribPtr[i] = temp_attrib[i];
-        }
+    if (context->radeon.NewGLState) {
+	    for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {
+		    /* mat states from state var not array for sw */
+		    dummy_attrib[i].stride = 0;
+	            temp_attrib[i] = TNL_CONTEXT(ctx)->vb.AttribPtr[i];
+		    TNL_CONTEXT(ctx)->vb.AttribPtr[i] = &(dummy_attrib[i]);
+	    }
 
-        r700SelectVertexShader(ctx);
-        vp = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+	    _tnl_UpdateFixedFunctionProgram(ctx);
 
-        if (vp->translated == GL_FALSE) 
-        {
-            // TODO
-            //fprintf(stderr, "Failing back to sw-tcl\n");
-            //hw_tcl_on = future_hw_tcl_on = 0;
-            //r300ResetHwState(rmesa);
-            //
-            r700UpdateStateParameters(ctx, _NEW_PROGRAM);
-            return;
-        }
+	    for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) {
+		    TNL_CONTEXT(ctx)->vb.AttribPtr[i] = temp_attrib[i];
+	    }
     }
 
-    r700UpdateStateParameters(ctx, _NEW_PROGRAM);
+    r700SelectVertexShader(ctx);
+    r700UpdateStateParameters(ctx, _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS);
+    context->radeon.NewGLState = 0;
 }
 
 /*
@@ -176,45 +148,25 @@ void r700UpdateDrawBuffer(GLcontext * ctx) /* TODO */ //---------------------
 	r700SetDepthTarget(context);
 }
 
-static void r700FetchStateParameter(GLcontext * ctx,
-			                        const gl_state_index state[STATE_LENGTH],
-			                        GLfloat * value)
-{
-    /* TODO */
-}
-
 void r700UpdateStateParameters(GLcontext * ctx, GLuint new_state) //--------------------
 {
-	struct r700_fragment_program *fp;
+	struct r700_fragment_program *fp =
+		(struct r700_fragment_program *)ctx->FragmentProgram._Current;
 	struct gl_program_parameter_list *paramList;
-	GLuint i;
 
-	if (!(new_state & (_NEW_BUFFERS | _NEW_PROGRAM)))
+	if (!(new_state & (_NEW_BUFFERS | _NEW_PROGRAM | _NEW_PROGRAM_CONSTANTS)))
 		return;
 
-	fp = (struct r700_fragment_program *)ctx->FragmentProgram._Current;
-	if (!fp)
-    {
+	if (!ctx->FragmentProgram._Current || !fp)
 		return;
-    }
 
-	paramList = fp->mesa_program.Base.Parameters;
+	paramList = ctx->FragmentProgram._Current->Base.Parameters;
 
 	if (!paramList)
-    {
 		return;
-    }
 
-	for (i = 0; i < paramList->NumParameters; i++) 
-    {
-		if (paramList->Parameters[i].Type == PROGRAM_STATE_VAR) 
-        {
-			r700FetchStateParameter(ctx,
-						paramList->Parameters[i].
-						StateIndexes,
-						paramList->ParameterValues[i]);
-		}
-	}
+	_mesa_load_state_parameters(ctx, paramList);
+
 }
 
 /**
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.c b/src/mesa/drivers/dri/r600/r700_vertprog.c
index 550594e9df8..f8f862b33aa 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.c
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.c
@@ -296,16 +296,16 @@ void r700SelectVertexShader(GLcontext *ctx)
     context_t *context = R700_CONTEXT(ctx);
     struct r700_vertex_program *vpc
              = (struct r700_vertex_program *)ctx->VertexProgram._Current;
+    TNLcontext *tnl = TNL_CONTEXT(ctx);
+    struct vertex_buffer *vb = &tnl->vb;
+    unsigned int unBit;
+    unsigned int i;
+
     if (context->radeon.radeonScreen->chip_family < CHIP_FAMILY_RV770)
     {
         vpc->r700AsmCode.bR6xx = 1;
     }
-    
-    TNLcontext *tnl = TNL_CONTEXT(ctx);
-	struct vertex_buffer *vb = &tnl->vb;
 
-    unsigned int unBit;
-    unsigned int i;
 	for(i=0; i<VERT_ATTRIB_MAX; i++)
 	{
 		unBit = 1 << i;
@@ -317,11 +317,9 @@ void r700SelectVertexShader(GLcontext *ctx)
 		}
 	}
 
-    if(GL_FALSE == vpc->translated)
-    {
-        r700TranslateVertexShader(vpc,
-		    					  &(vpc->mesa_program) );
-    }
+	if(GL_FALSE == vpc->translated) {
+		r700TranslateVertexShader(vpc, &(vpc->mesa_program) );
+	}
 }
 
 void * r700GetActiveVpShaderBo(GLcontext * ctx)
@@ -345,17 +343,17 @@ GLboolean r700SetupVertexProgram(GLcontext * ctx)
 
     if(GL_FALSE == vp->loaded)
     {
-        if(vp->r700Shader.bNeedsAssembly == GL_TRUE)
+	    if(vp->r700Shader.bNeedsAssembly == GL_TRUE)
 	    {
 		    Assemble( &(vp->r700Shader) );
 	    }
 
         /* Load vp to gpu */
-        r600EmitShader(ctx, 
-                       &(vp->shaderbo), 
+        r600EmitShader(ctx,
+                       &(vp->shaderbo),
                        (GLvoid *)(vp->r700Shader.pProgram),
                        vp->r700Shader.uShaderBinaryDWORDSize,
-                       "VS"); 
+                       "VS");
 
         vp->loaded = GL_TRUE;
     }
diff --git a/src/mesa/drivers/dri/r600/r700_vertprog.h b/src/mesa/drivers/dri/r600/r700_vertprog.h
index 6a9726a3d0f..4c3b7ee453d 100644
--- a/src/mesa/drivers/dri/r600/r700_vertprog.h
+++ b/src/mesa/drivers/dri/r600/r700_vertprog.h
@@ -62,23 +62,23 @@ struct r700_vertex_program
 
 //Internal
 unsigned int Map_Vertex_Output(r700_AssemblerBase       *pAsm, 
-					           struct gl_vertex_program *mesa_vp,
-					           unsigned int unStart);
+			       struct gl_vertex_program *mesa_vp,
+			       unsigned int unStart);
 unsigned int Map_Vertex_Input(r700_AssemblerBase       *pAsm, 
-					          struct gl_vertex_program *mesa_vp,
-					          unsigned int unStart);
+			      struct gl_vertex_program *mesa_vp,
+			      unsigned int unStart);
 GLboolean Process_Vertex_Program_Vfetch_Instructions(
-						struct r700_vertex_program *vp,
-						struct gl_vertex_program   *mesa_vp);
+	struct r700_vertex_program *vp,
+	struct gl_vertex_program   *mesa_vp);
 void Map_Vertex_Program(struct r700_vertex_program *vp,
-						struct gl_vertex_program   *mesa_vp);
+			struct gl_vertex_program   *mesa_vp);
 GLboolean Find_Instruction_Dependencies_vp(struct r700_vertex_program *vp,
-					                	   struct gl_vertex_program   *mesa_vp);
+					   struct gl_vertex_program   *mesa_vp);
 
-/* Interface */
-extern GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp,
-							   struct gl_vertex_program   *mesa_vp);
+GLboolean r700TranslateVertexShader(struct r700_vertex_program *vp,
+				    struct gl_vertex_program   *mesa_vp);
 
+/* Interface */
 extern void r700SelectVertexShader(GLcontext *ctx);
 
 extern GLboolean r700SetupVertexProgram(GLcontext * ctx);
diff --git a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
index a42870f4a93..4520a7d7d49 100644
--- a/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
+++ b/src/mesa/drivers/dri/radeon/radeon_bocs_wrapper.h
@@ -45,6 +45,10 @@ struct drm_radeon_info {
 #define RADEON_PARAM_DEVICE_ID 16
 #endif
 
+#ifndef RADEON_PARAM_NUM_Z_PIPES
+#define RADEON_PARAM_NUM_Z_PIPES 17
+#endif
+
 #ifndef RADEON_INFO_DEVICE_ID
 #define RADEON_INFO_DEVICE_ID 0
 #endif
@@ -52,6 +56,10 @@ struct drm_radeon_info {
 #define RADEON_INFO_NUM_GB_PIPES 0
 #endif
 
+#ifndef RADEON_INFO_NUM_Z_PIPES
+#define RADEON_INFO_NUM_Z_PIPES 0
+#endif
+
 #ifndef DRM_RADEON_INFO
 #define DRM_RADEON_INFO 0x1
 #endif
diff --git a/src/mesa/drivers/dri/radeon/radeon_fbo.c b/src/mesa/drivers/dri/radeon/radeon_fbo.c
index 3d7c9708e1a..6f0cc08770c 100644
--- a/src/mesa/drivers/dri/radeon/radeon_fbo.c
+++ b/src/mesa/drivers/dri/radeon/radeon_fbo.c
@@ -178,12 +178,13 @@ radeon_alloc_renderbuffer_storage(GLcontext * ctx, struct gl_renderbuffer *rb,
                                              width, height);
    }
    else {
-     uint32_t size = width * height * cpp;
+     uint32_t size;
      uint32_t pitch = ((cpp * width + 63) & ~63) / cpp;
 
      fprintf(stderr,"Allocating %d x %d radeon RBO (pitch %d)\n", width,
 	  height, pitch);
 
+     size = pitch * height * cpp;
      rrb->pitch = pitch * cpp;
      rrb->cpp = cpp;
      rrb->bo = radeon_bo_open(radeon->radeonScreen->bom,
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.c b/src/mesa/drivers/dri/radeon/radeon_screen.c
index 10afe527d3d..bdcfd10c06e 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.c
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.c
@@ -267,6 +267,9 @@ radeonGetParam(__DRIscreenPrivate *sPriv, int param, void *value)
       case RADEON_PARAM_NUM_GB_PIPES:
           info.request = RADEON_INFO_NUM_GB_PIPES;
           break;
+      case RADEON_PARAM_NUM_Z_PIPES:
+          info.request = RADEON_INFO_NUM_Z_PIPES;
+          break;
       default:
           return -EINVAL;
       }
@@ -1171,6 +1174,15 @@ radeonCreateScreen( __DRIscreenPrivate *sPriv )
        default:
 	   break;
        }
+
+       if ( sPriv->drm_version.minor >= 31 ) {
+	       ret = radeonGetParam(sPriv, RADEON_PARAM_NUM_Z_PIPES, &temp);
+	       if (ret)
+		       screen->num_z_pipes = 2;
+	       else
+		       screen->num_z_pipes = temp;
+       } else
+	       screen->num_z_pipes = 2;
    }
 
    if ( sPriv->drm_version.minor >= 10 ) {
@@ -1372,6 +1384,12 @@ radeonCreateScreen2(__DRIscreenPrivate *sPriv)
 	   break;
        }
 
+       ret = radeonGetParam(sPriv, RADEON_PARAM_NUM_Z_PIPES, &temp);
+       if (ret)
+	       screen->num_z_pipes = 2;
+       else
+	       screen->num_z_pipes = temp;
+
    }
 
    i = 0;
diff --git a/src/mesa/drivers/dri/radeon/radeon_screen.h b/src/mesa/drivers/dri/radeon/radeon_screen.h
index f0dd46b0b13..15744e88284 100644
--- a/src/mesa/drivers/dri/radeon/radeon_screen.h
+++ b/src/mesa/drivers/dri/radeon/radeon_screen.h
@@ -108,6 +108,7 @@ typedef struct radeon_screen {
    const __DRIextension *extensions[16];
 
    int num_gb_pipes;
+   int num_z_pipes;
    int kernel_mm;
    drm_radeon_sarea_t *sarea;	/* Private SAREA data */
    struct radeon_bo_manager *bom;
diff --git a/src/mesa/glapi/glapi.c b/src/mesa/glapi/glapi.c
index 2b105d0f17d..4bb10df1d84 100644
--- a/src/mesa/glapi/glapi.c
+++ b/src/mesa/glapi/glapi.c
@@ -160,26 +160,20 @@ static GLint NoOpUnused(void)
  * the variables \c _glapi_Dispatch and \c _glapi_Context are used for this
  * purpose.
  *
- * In the "normal" threaded case, the variables \c _glapi_Dispatch and
- * \c _glapi_Context will be \c NULL if an application is detected as being
- * multithreaded.  Single-threaded applications will use \c _glapi_Dispatch
- * and \c _glapi_Context just like the case without any threading support.
- * When \c _glapi_Dispatch and \c _glapi_Context are \c NULL, the thread state
- * data \c _gl_DispatchTSD and \c ContextTSD are used.  Drivers and the
- * static dispatch functions access these variables via \c _glapi_get_dispatch
- * and \c _glapi_get_context.
+ * In the "normal" threaded case, the variable \c _glapi_SingleThreaded will be
+ * \c GL_FALSE if an application is detected as being multithreaded.
+ * Single-threaded applications will use \c _glapi_Dispatch and \c
+ * _glapi_Context just like the case without any threading support.  When \c
+ * _glapi_SingleThreaded is \c GL_FALSE, the thread state data \c
+ * _gl_DispatchTSD and \c ContextTSD are used.  Drivers and the static dispatch
+ * functions access these variables via \c _glapi_get_dispatch and \c
+ * _glapi_get_context.
  *
- * There is a race condition in setting \c _glapi_Dispatch to \c NULL.  It is
- * possible for the original thread to be setting it at the same instant a new
- * thread, perhaps running on a different processor, is clearing it.  Because
- * of that, \c ThreadSafe, which can only ever be changed to \c GL_TRUE, is
- * used to determine whether or not the application is multithreaded.
- * 
  * In the TLS case, the variables \c _glapi_Dispatch and \c _glapi_Context are
  * hardcoded to \c NULL.  Instead the TLS variables \c _glapi_tls_Dispatch and
- * \c _glapi_tls_Context are used.  Having \c _glapi_Dispatch and
- * \c _glapi_Context be hardcoded to \c NULL maintains binary compatability
- * between TLS enabled loaders and non-TLS DRI drivers.
+ * \c _glapi_tls_Context are used.  The variable \c _glapi_SingleThreaded,
+ * though not used, is defined and hardcoded to \c GL_FALSE to maintain binary
+ * compatability between TLS enabled loaders and non-TLS DRI drivers.
  */
 /*@{*/
 #if defined(GLX_USE_TLS)
@@ -194,13 +188,26 @@ PUBLIC __thread void * _glapi_tls_Context
 PUBLIC const struct _glapi_table *_glapi_Dispatch = NULL;
 PUBLIC const void *_glapi_Context = NULL;
 
+/* Unused, but maintain binary compatability with non-TLS DRI drivers */
+GLboolean _glapi_SingleThreaded = GL_FALSE;
+
 #else
 
 #if defined(THREADS)
 
-static GLboolean ThreadSafe = GL_FALSE;  /**< In thread-safe mode? */
-_glthread_TSD _gl_DispatchTSD;           /**< Per-thread dispatch pointer */
-static _glthread_TSD ContextTSD;         /**< Per-thread context pointer */
+#ifdef WIN32_THREADS
+/* _glthread_DECLARE_STATIC_MUTEX is broken on windows.  There will be race! */
+#define CHECK_MULTITHREAD_LOCK()
+#define CHECK_MULTITHREAD_UNLOCK()
+#else
+_glthread_DECLARE_STATIC_MUTEX(ThreadCheckMutex);
+#define CHECK_MULTITHREAD_LOCK() _glthread_LOCK_MUTEX(ThreadCheckMutex)
+#define CHECK_MULTITHREAD_UNLOCK() _glthread_UNLOCK_MUTEX(ThreadCheckMutex)
+#endif
+
+GLboolean _glapi_SingleThreaded = GL_TRUE;  /**< In single-thread mode? */
+_glthread_TSD _gl_DispatchTSD;              /**< Per-thread dispatch pointer */
+static _glthread_TSD ContextTSD;            /**< Per-thread context pointer */
 
 #if defined(WIN32_THREADS)
 void FreeTSD(_glthread_TSD *p);
@@ -231,23 +238,30 @@ void
 _glapi_check_multithread(void)
 {
 #if defined(THREADS) && !defined(GLX_USE_TLS)
-   if (!ThreadSafe) {
-      static unsigned long knownID;
-      static GLboolean firstCall = GL_TRUE;
-      if (firstCall) {
-         knownID = _glthread_GetID();
-         firstCall = GL_FALSE;
-      }
-      else if (knownID != _glthread_GetID()) {
-         ThreadSafe = GL_TRUE;
-         _glapi_set_dispatch(NULL);
-         _glapi_set_context(NULL);
-      }
+   static unsigned long knownID;
+   static GLboolean firstCall = GL_TRUE;
+
+   if (!_glapi_SingleThreaded)
+      return;
+
+   CHECK_MULTITHREAD_LOCK();
+   if (firstCall) {
+      /* initialize TSDs */
+      (void) _glthread_GetTSD(&ContextTSD);
+      (void) _glthread_GetTSD(&_gl_DispatchTSD);
+
+      knownID = _glthread_GetID();
+      firstCall = GL_FALSE;
    }
-   else if (!_glapi_get_dispatch()) {
-      /* make sure that this thread's dispatch pointer isn't null */
-      _glapi_set_dispatch(NULL);
+   else if (knownID != _glthread_GetID()) {
+      /*
+       * switch to thread-safe mode.  _glapi_Context and _glapi_Dispatch are no
+       * longer accessed after this point, except for raced by the first
+       * thread.  Because of the race, they cannot be reset to NULL.
+       */
+      _glapi_SingleThreaded = GL_FALSE;
    }
+   CHECK_MULTITHREAD_UNLOCK();
 #endif
 }
 
@@ -265,8 +279,10 @@ _glapi_set_context(void *context)
 #if defined(GLX_USE_TLS)
    _glapi_tls_Context = context;
 #elif defined(THREADS)
+   if (_glapi_SingleThreaded)
+      _glapi_Context = context;
+   /* always update TSD because we might switch to it at any time */
    _glthread_SetTSD(&ContextTSD, context);
-   _glapi_Context = (ThreadSafe) ? NULL : context;
 #else
    _glapi_Context = context;
 #endif
@@ -285,12 +301,8 @@ _glapi_get_context(void)
 #if defined(GLX_USE_TLS)
    return _glapi_tls_Context;
 #elif defined(THREADS)
-   if (ThreadSafe) {
-      return _glthread_GetTSD(&ContextTSD);
-   }
-   else {
-      return _glapi_Context;
-   }
+   return (_glapi_SingleThreaded)
+      ? _glapi_Context : _glthread_GetTSD(&ContextTSD);
 #else
    return _glapi_Context;
 #endif
@@ -505,8 +517,9 @@ _glapi_set_dispatch(struct _glapi_table *dispatch)
 #if defined(GLX_USE_TLS)
    _glapi_tls_Dispatch = dispatch;
 #elif defined(THREADS)
+   if (_glapi_SingleThreaded)
+      _glapi_Dispatch = dispatch;
    _glthread_SetTSD(&_gl_DispatchTSD, (void *) dispatch);
-   _glapi_Dispatch = (ThreadSafe) ? NULL : dispatch;
 #else /*THREADS*/
    _glapi_Dispatch = dispatch;
 #endif /*THREADS*/
@@ -520,17 +533,15 @@ _glapi_set_dispatch(struct _glapi_table *dispatch)
 PUBLIC struct _glapi_table *
 _glapi_get_dispatch(void)
 {
-   struct _glapi_table * api;
 #if defined(GLX_USE_TLS)
-   api = _glapi_tls_Dispatch;
+   return _glapi_tls_Dispatch;
 #elif defined(THREADS)
-   api = (ThreadSafe)
-     ? (struct _glapi_table *) _glthread_GetTSD(&_gl_DispatchTSD)
-     : _glapi_Dispatch;
+   return (_glapi_SingleThreaded)
+      ? _glapi_Dispatch
+      : (struct _glapi_table *) _glthread_GetTSD(&_gl_DispatchTSD);
 #else
-   api = _glapi_Dispatch;
+   return _glapi_Dispatch;
 #endif
-   return api;
 }
 
 
diff --git a/src/mesa/glapi/glapi.h b/src/mesa/glapi/glapi.h
index 8f2cf662183..b2a1fe6ee93 100644
--- a/src/mesa/glapi/glapi.h
+++ b/src/mesa/glapi/glapi.h
@@ -94,7 +94,10 @@ extern void *_glapi_Context;
 extern struct _glapi_table *_glapi_Dispatch;
 
 # ifdef THREADS
-#  define GET_CURRENT_CONTEXT(C)  GLcontext *C = (GLcontext *) (_glapi_Context ? _glapi_Context : _glapi_get_context())
+/* this variable is here only for quick access to current context/dispatch */
+extern GLboolean _glapi_SingleThreaded;
+#  define GET_CURRENT_CONTEXT(C)  GLcontext *C = (GLcontext *) \
+       ((_glapi_SingleThreaded) ? _glapi_Context : _glapi_get_context())
 # else
 #  define GET_CURRENT_CONTEXT(C)  GLcontext *C = (GLcontext *) _glapi_Context
 # endif
diff --git a/src/mesa/glapi/glthread.h b/src/mesa/glapi/glthread.h
index 8ec933a8514..a36bea71762 100644
--- a/src/mesa/glapi/glthread.h
+++ b/src/mesa/glapi/glthread.h
@@ -322,7 +322,7 @@ extern __thread struct _glapi_table * _glapi_tls_Dispatch
 #elif !defined(GL_CALL)
 # if defined(THREADS)
 #  define GET_DISPATCH() \
-   ((__builtin_expect( _glapi_Dispatch != NULL, 1 )) \
+   ((__builtin_expect(_glapi_SingleThreaded, 1)) \
        ? _glapi_Dispatch : _glapi_get_dispatch())
 # else
 #  define GET_DISPATCH() _glapi_Dispatch
diff --git a/src/mesa/shader/arbprogram.c b/src/mesa/shader/arbprogram.c
index 39136efadac..4d8cff07001 100644
--- a/src/mesa/shader/arbprogram.c
+++ b/src/mesa/shader/arbprogram.c
@@ -1001,6 +1001,9 @@ _mesa_GetProgramivARB(GLenum target, GLenum pname, GLint *params)
             _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramivARB(pname)");
             return;
       }
+   } else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramivARB(pname)");
+      return;
    }
 }
 
diff --git a/src/mesa/shader/prog_parameter_layout.c b/src/mesa/shader/prog_parameter_layout.c
index 8f2b3062203..1c37b3a7a5a 100644
--- a/src/mesa/shader/prog_parameter_layout.c
+++ b/src/mesa/shader/prog_parameter_layout.c
@@ -106,7 +106,11 @@ copy_indirect_accessed_array(struct gl_program_parameter_list *src,
 }
 
 
-int
+/**
+ * XXX description???
+ * \return GL_TRUE for success, GL_FALSE for failure
+ */
+GLboolean
 _mesa_layout_parameters(struct asm_parser_state *state)
 {
    struct gl_program_parameter_list *layout;
@@ -128,12 +132,12 @@ _mesa_layout_parameters(struct asm_parser_state *state)
 	     */
 	    if (!inst->SrcReg[i].Symbol->pass1_done) {
 	       const int new_begin =
-		  copy_indirect_accessed_array(state->prog->Parameters, layout, 
+		  copy_indirect_accessed_array(state->prog->Parameters, layout,
 		      inst->SrcReg[i].Symbol->param_binding_begin,
 		      inst->SrcReg[i].Symbol->param_binding_length);
 
 	       if (new_begin < 0) {
-		  return 0;
+		  return GL_FALSE;
 	       }
 
 	       inst->SrcReg[i].Symbol->param_binding_begin = new_begin;
@@ -209,5 +213,5 @@ _mesa_layout_parameters(struct asm_parser_state *state)
    _mesa_free_parameter_list(state->prog->Parameters);
    state->prog->Parameters = layout;
 
-   return 1;
+   return GL_TRUE;
 }
diff --git a/src/mesa/shader/prog_parameter_layout.h b/src/mesa/shader/prog_parameter_layout.h
index 1686170bab5..99a7b6c7266 100644
--- a/src/mesa/shader/prog_parameter_layout.h
+++ b/src/mesa/shader/prog_parameter_layout.h
@@ -36,6 +36,7 @@
 extern unsigned _mesa_combine_swizzles(unsigned base, unsigned applied);
 
 struct asm_parser_state;
-extern int _mesa_layout_parameters(struct asm_parser_state *state);
+
+extern GLboolean _mesa_layout_parameters(struct asm_parser_state *state);
 
 #endif /* PROG_PARAMETER_LAYOUT_H */
diff --git a/src/mesa/shader/program_parse.tab.c b/src/mesa/shader/program_parse.tab.c
index dbed448113e..5c604c2fd1f 100644
--- a/src/mesa/shader/program_parse.tab.c
+++ b/src/mesa/shader/program_parse.tab.c
@@ -5156,12 +5156,12 @@ _mesa_parse_arb_program(GLcontext *ctx, GLenum target, const GLubyte *str,
       ? & ctx->Const.VertexProgram
       : & ctx->Const.FragmentProgram;
 
-   state->MaxTextureImageUnits = 16;
-   state->MaxTextureCoordUnits = 8;
-   state->MaxTextureUnits = 8;
-   state->MaxClipPlanes = 6;
-   state->MaxLights = 8;
-   state->MaxProgramMatrices = 8;
+   state->MaxTextureImageUnits = ctx->Const.MaxTextureImageUnits;
+   state->MaxTextureCoordUnits = ctx->Const.MaxTextureCoordUnits;
+   state->MaxTextureUnits = ctx->Const.MaxTextureUnits;
+   state->MaxClipPlanes = ctx->Const.MaxClipPlanes;
+   state->MaxLights = ctx->Const.MaxLights;
+   state->MaxProgramMatrices = ctx->Const.MaxProgramMatrices;
 
    state->state_param_enum = (target == GL_VERTEX_PROGRAM_ARB)
       ? STATE_VERTEX_PROGRAM : STATE_FRAGMENT_PROGRAM;
diff --git a/src/mesa/shader/program_parse.y b/src/mesa/shader/program_parse.y
index 89e88502121..e2e83e484f5 100644
--- a/src/mesa/shader/program_parse.y
+++ b/src/mesa/shader/program_parse.y
@@ -2287,12 +2287,12 @@ _mesa_parse_arb_program(GLcontext *ctx, GLenum target, const GLubyte *str,
       ? & ctx->Const.VertexProgram
       : & ctx->Const.FragmentProgram;
 
-   state->MaxTextureImageUnits = 16;
-   state->MaxTextureCoordUnits = 8;
-   state->MaxTextureUnits = 8;
-   state->MaxClipPlanes = 6;
-   state->MaxLights = 8;
-   state->MaxProgramMatrices = 8;
+   state->MaxTextureImageUnits = ctx->Const.MaxTextureImageUnits;
+   state->MaxTextureCoordUnits = ctx->Const.MaxTextureCoordUnits;
+   state->MaxTextureUnits = ctx->Const.MaxTextureUnits;
+   state->MaxClipPlanes = ctx->Const.MaxClipPlanes;
+   state->MaxLights = ctx->Const.MaxLights;
+   state->MaxProgramMatrices = ctx->Const.MaxProgramMatrices;
 
    state->state_param_enum = (target == GL_VERTEX_PROGRAM_ARB)
       ? STATE_VERTEX_PROGRAM : STATE_FRAGMENT_PROGRAM;
diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c
index ad2e306c19e..bef0f856534 100644
--- a/src/mesa/shader/slang/slang_builtin.c
+++ b/src/mesa/shader/slang/slang_builtin.c
@@ -436,7 +436,7 @@ emit_statevars(const char *name, int array_len,
                struct gl_program_parameter_list *paramList)
 {
    if (type->type == SLANG_SPEC_ARRAY) {
-      GLint i, pos;
+      GLint i, pos = -1;
       assert(array_len > 0);
       if (strcmp(name, "gl_ClipPlane") == 0) {
          tokens[0] = STATE_CLIPPLANE;
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index 8709633557c..ccf972f8055 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -94,6 +94,9 @@ struct bitmap_cache
 
    GLfloat color[4];
 
+   /** Bitmap's Z position */
+   GLfloat zpos;
+
    struct pipe_texture *texture;
    struct pipe_transfer *trans;
 
@@ -104,6 +107,8 @@ struct bitmap_cache
 };
 
 
+/** Epsilon for Z comparisons */
+#define Z_EPSILON 1e-06
 
 
 /**
@@ -538,9 +543,7 @@ draw_bitmap_quad(GLcontext *ctx, GLint x, GLint y, GLfloat z,
    }
 
    /* draw textured quad */
-   offset = setup_bitmap_vertex_data(st, x, y, width, height,
-                                     ctx->Current.RasterPos[2],
-                                     color);
+   offset = setup_bitmap_vertex_data(st, x, y, width, height, z, color);
 
    util_draw_vertex_buffer(pipe, st->bitmap.vbuf, offset,
                            PIPE_PRIM_TRIANGLE_FAN,
@@ -647,7 +650,7 @@ st_flush_bitmap_cache(struct st_context *st)
          draw_bitmap_quad(st->ctx,
                           cache->xpos,
                           cache->ypos,
-                          st->ctx->Current.RasterPos[2],
+                          cache->zpos,
                           BITMAP_CACHE_WIDTH, BITMAP_CACHE_HEIGHT,
                           cache->texture,
                           cache->color);
@@ -687,6 +690,7 @@ accum_bitmap(struct st_context *st,
 {
    struct bitmap_cache *cache = st->bitmap.cache;
    int px = -999, py;
+   const GLfloat z = st->ctx->Current.RasterPos[2];
 
    if (width > BITMAP_CACHE_WIDTH ||
        height > BITMAP_CACHE_HEIGHT)
@@ -697,7 +701,8 @@ accum_bitmap(struct st_context *st,
       py = y - cache->ypos;
       if (px < 0 || px + width > BITMAP_CACHE_WIDTH ||
           py < 0 || py + height > BITMAP_CACHE_HEIGHT ||
-          !TEST_EQ_4V(st->ctx->Current.RasterColor, cache->color)) {
+          !TEST_EQ_4V(st->ctx->Current.RasterColor, cache->color) ||
+          ((fabs(z - cache->zpos) > Z_EPSILON))) {
          /* This bitmap would extend beyond cache bounds, or the bitmap
           * color is changing
           * so flush and continue.
@@ -712,6 +717,7 @@ accum_bitmap(struct st_context *st,
       py = (BITMAP_CACHE_HEIGHT - height) / 2;
       cache->xpos = x;
       cache->ypos = y - py;
+      cache->zpos = z;
       cache->empty = GL_FALSE;
       COPY_4FV(cache->color, st->ctx->Current.RasterColor);
    }
diff --git a/src/mesa/vbo/vbo_exec_draw.c b/src/mesa/vbo/vbo_exec_draw.c
index d76c45f3560..625452ac09a 100644
--- a/src/mesa/vbo/vbo_exec_draw.c
+++ b/src/mesa/vbo/vbo_exec_draw.c
@@ -398,7 +398,7 @@ vbo_exec_vtx_flush( struct vbo_exec_context *exec, GLboolean unmap )
       vbo_exec_vtx_unmap( exec );
    }
 
-   if (unmap) 
+   if (unmap || exec->vtx.vertex_size == 0)
       exec->vtx.max_vert = 0;
    else
       exec->vtx.max_vert = ((VBO_VERT_BUFFER_SIZE - exec->vtx.buffer_used) /