-Fix first frame -bug

-Use 16-bit elts in vtxfmt_a path if possible
-Optimize VSF param uploading
-return in r300DepthMask looks suspicious, use r300Enable instead
-Dont use r300ResetHwState in invalidate state(disabled, missing hooks, possible instabilities)

6 files changed, 109 insertions, 81 deletions

diff --git a/src/mesa/drivers/dri/r300/r300_context.h b/src/mesa/drivers/dri/r300/r300_context.h
index 0ece8f8d1f..c03dd03daf 100644
--- a/src/mesa/drivers/dri/r300/r300_context.h
+++ b/src/mesa/drivers/dri/r300/r300_context.h
@@ -826,6 +826,7 @@ struct r300_context {
 	 */
 	struct r300_dma dma;
 	GLboolean save_on_next_unlock;
+	GLuint NewGLState;
 
 	/* Texture object bookkeeping
 	 */
@@ -839,17 +840,6 @@ struct r300_context {
 	 */
 	GLuint prefer_gart_client_texturing;
 
-	/* TCL stuff
-	 */
-	GLmatrix TexGenMatrix[R300_MAX_TEXTURE_UNITS];
-	GLboolean recheck_texgen[R300_MAX_TEXTURE_UNITS];
-	GLboolean TexGenNeedNormals[R300_MAX_TEXTURE_UNITS];
-	GLuint TexMatEnabled;
-	GLuint TexMatCompSel;
-	GLuint TexGenEnabled;
-	GLuint TexGenInputs;
-	GLuint TexGenCompSel;
-	GLmatrix tmpmat;
 #ifdef USER_BUFFERS
 	key_t mm_ipc_key;
 	int mm_shm_id;
@@ -885,7 +875,7 @@ extern GLboolean r300CreateContext(const __GLcontextModes * glVisual,
 
 void translate_vertex_shader(struct r300_vertex_program *vp);
 extern void r300InitShaderFuncs(struct dd_function_table *functions);
-extern void r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp);
+extern int r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp, float *dst);
 
 #ifdef RADEON_VTXFMT_A
 extern void radeon_init_vtxfmt_a(r300ContextPtr rmesa);
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 1599f0f6af..98ddd82106 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -543,10 +543,13 @@ static GLboolean r300_run_vb_render(GLcontext *ctx,
 	if (RADEON_DEBUG & DEBUG_PRIMS)
 		fprintf(stderr, "%s\n", __FUNCTION__);
 	
-
+	
+	r300UpdateShaders(rmesa);
+	
    	r300ReleaseArrays(ctx);
 	r300EmitArrays(ctx, GL_FALSE);
 	
+	r300UpdateShaderStates(rmesa);
 //	LOCK_HARDWARE(&(rmesa->radeon));
 
 	reg_start(R300_RB3D_DSTCACHE_CTLSTAT,0);
@@ -692,6 +695,8 @@ void dump_dt(struct dt *dt, int count)
 	if (rmesa->state.VB.LockCount == 0) {
  	  	r300ReleaseArrays(ctx);
 		r300EmitArraysVtx(ctx, GL_FALSE);
+		
+		r300UpdateShaderStates(rmesa);
 	} else {
 		/* TODO: Figure out why do we need these. */
 		R300_STATECHANGE(rmesa, vir[0]);
@@ -891,7 +896,8 @@ static GLboolean r300_run_tcl_render(GLcontext *ctx,
 	if(hw_tcl_on == GL_FALSE)
 		return GL_TRUE;
 	
-	r300UpdateShaderStates(rmesa);
+	//r300UpdateShaders(rmesa);
+	//r300UpdateShaderStates(rmesa);
 	
 	return r300_run_vb_render(ctx, stage);
 }
diff --git a/src/mesa/drivers/dri/r300/r300_state.c b/src/mesa/drivers/dri/r300/r300_state.c
index e3aa74b102..343acc426a 100644
--- a/src/mesa/drivers/dri/r300/r300_state.c
+++ b/src/mesa/drivers/dri/r300/r300_state.c
@@ -621,14 +621,7 @@ static void r300DepthFunc(GLcontext* ctx, GLenum func)
 static void r300DepthMask(GLcontext* ctx, GLboolean mask)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-
-	if (!ctx->Depth.Test)
-		return;
-
-	R300_STATECHANGE(r300, zs);
-	r300->hw.zs.cmd[R300_ZS_CNTL_0] &= R300_RB3D_STENCIL_ENABLE;
-	r300->hw.zs.cmd[R300_ZS_CNTL_0] |= mask 
-	    ? R300_RB3D_Z_TEST_AND_WRITE : R300_RB3D_Z_TEST;
+	r300Enable(ctx, GL_DEPTH_TEST, ctx->Depth.Test);
 }
 
 
@@ -1054,7 +1047,7 @@ void r300_setup_textures(GLcontext *ctx)
 			
 			if(t == NULL){
 				fprintf(stderr, "Texture unit %d enabled, but corresponding texobj is NULL, using default object.\n", i);
-				//exit(-1);
+				exit(-1);
 				t=&default_tex_obj;
 			}
 			
@@ -1506,24 +1499,24 @@ void r300SetupVertexProgram(r300ContextPtr rmesa)
 	struct r300_vertex_program *prog=(struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
 			
 
-	/* Reset state, in case we don't use something */
 	((drm_r300_cmd_header_t*)rmesa->hw.vpp.cmd)->vpu.count = 0;
+	R300_STATECHANGE(rmesa, vpp);
+	param_count = r300VertexProgUpdateParams(ctx, prog, (float *)&rmesa->hw.vpp.cmd[R300_VPP_PARAM_0]);
+	bump_vpu_count(rmesa->hw.vpp.cmd, param_count);
+	param_count /= 4;
+	
+	/* Reset state, in case we don't use something */
 	((drm_r300_cmd_header_t*)rmesa->hw.vpi.cmd)->vpu.count = 0;
 	((drm_r300_cmd_header_t*)rmesa->hw.vps.cmd)->vpu.count = 0;
 
-	r300VertexProgUpdateParams(ctx, prog);
-
 	setup_vertex_shader_fragment(rmesa, VSF_DEST_PROGRAM, &(prog->program));
 
-	setup_vertex_shader_fragment(rmesa, VSF_DEST_MATRIX0, &(prog->params));
-
 #if 0
 	setup_vertex_shader_fragment(rmesa, VSF_DEST_UNKNOWN1, &(rmesa->state.vertex_shader.unknown1));
 	setup_vertex_shader_fragment(rmesa, VSF_DEST_UNKNOWN2, &(rmesa->state.vertex_shader.unknown2));
 #endif
 
 	inst_count=prog->program.length/4 - 1;
-	param_count=prog->params.length/4;
 
 	R300_STATECHANGE(rmesa, pvs);
 	rmesa->hw.pvs.cmd[R300_PVS_CNTL_1]=(0 << R300_PVS_CNTL_1_PROGRAM_START_SHIFT)
@@ -1545,28 +1538,42 @@ void r300SetupVertexProgram(r300ContextPtr rmesa)
 extern void _tnl_UpdateFixedFunctionProgram( GLcontext *ctx );
 
 extern int future_hw_tcl_on;
-void r300UpdateShaderStates(r300ContextPtr rmesa)
+void r300UpdateShaders(r300ContextPtr rmesa)
 {
 	GLcontext *ctx;
 	struct r300_vertex_program *vp;
 	
-	ctx = rmesa->radeon.glCtx; 
+	ctx = rmesa->radeon.glCtx;
 	
-	if(ctx->VertexProgram._Enabled == GL_FALSE){
-		_tnl_UpdateFixedFunctionProgram(ctx);
-	}
-	vp = (struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
-	if(vp->translated == GL_FALSE)
-		translate_vertex_shader(vp);
-	if(vp->translated == GL_FALSE){
-		fprintf(stderr, "Failing back to sw-tcl\n");
-		debug_vp(ctx, &vp->mesa_program);
-		hw_tcl_on=future_hw_tcl_on=0;
-		r300ResetHwState(rmesa);
-		
-		return ;
+	if (rmesa->NewGLState && hw_tcl_on) {
+		rmesa->NewGLState = 0;
+		if (ctx->VertexProgram._Enabled == GL_FALSE)
+			_tnl_UpdateFixedFunctionProgram(ctx);
+	
+		vp = (struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
+		if (vp->translated == GL_FALSE)
+			translate_vertex_shader(vp);
+		if (vp->translated == GL_FALSE) {
+			fprintf(stderr, "Failing back to sw-tcl\n");
+			debug_vp(ctx, &vp->mesa_program);
+			hw_tcl_on = future_hw_tcl_on = 0;
+			r300ResetHwState(rmesa);
+
+			return ;
+		}
 	}
-		
+	
+}
+
+void r300UpdateShaderStates(r300ContextPtr rmesa)
+{
+	GLcontext *ctx;
+	ctx = rmesa->radeon.glCtx;
+	
+#ifdef CB_DPATH
+	r300UpdateTextureState(ctx);
+#endif
+	
 	r300_setup_textures(ctx);
 	r300_setup_rs_unit(ctx);
 
@@ -1865,12 +1872,15 @@ static void r300InvalidateState(GLcontext * ctx, GLuint new_state)
 	_tnl_InvalidateState(ctx, new_state);
 	_ae_invalidate_state(ctx, new_state);
 
+#ifndef CB_DPATH
 	/* Go inefficiency! */
 	r300ResetHwState(r300);
+#endif
 #ifdef HW_VBOS
 	if(new_state & _NEW_ARRAY)
 		r300->state.VB.lock_uptodate = GL_FALSE;
 #endif
+	r300->NewGLState |= new_state;
 }
 
 /**
diff --git a/src/mesa/drivers/dri/r300/r300_state.h b/src/mesa/drivers/dri/r300/r300_state.h
index ee54e4b3ef..9564c0716c 100644
--- a/src/mesa/drivers/dri/r300/r300_state.h
+++ b/src/mesa/drivers/dri/r300/r300_state.h
@@ -68,6 +68,7 @@ extern void r300SetupPixelShader(r300ContextPtr rmesa);
 
 extern void r300_setup_textures(GLcontext *ctx);
 extern void r300_setup_rs_unit(GLcontext *ctx);
+extern void r300UpdateShaders(r300ContextPtr rmesa);
 extern void r300UpdateShaderStates(r300ContextPtr rmesa);
 
 extern void r300_print_state_atom(r300ContextPtr r300, struct r300_state_atom *state);
diff --git a/src/mesa/drivers/dri/r300/r300_vertexprog.c b/src/mesa/drivers/dri/r300/r300_vertexprog.c
index 7526bb86e6..9bf92cde55 100644
--- a/src/mesa/drivers/dri/r300/r300_vertexprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertexprog.c
@@ -242,11 +242,11 @@ void debug_vp(GLcontext *ctx, struct vertex_program *vp)
 	
 }
 
-void r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp)
+int r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp, float *dst)
 {
 	int pi;
 	struct vertex_program *mesa_vp=(void *)vp;
-	int dst_index;
+	float *dst_o=dst;
 	
 	_mesa_load_state_parameters(ctx, mesa_vp->Parameters);
 	
@@ -255,7 +255,7 @@ void r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp)
 		fprintf(stderr, "%s:Params exhausted\n", __FUNCTION__);
 		exit(-1);
 	}
-	dst_index=0;
+	
 	for(pi=0; pi < mesa_vp->Parameters->NumParameters; pi++){
 		switch(mesa_vp->Parameters->Parameters[pi].Type){
 			
@@ -263,10 +263,10 @@ void r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp)
 		case NAMED_PARAMETER:
 			//fprintf(stderr, "%s", vp->Parameters->Parameters[pi].Name);
 		case CONSTANT:
-			vp->params.body.f[dst_index++]=mesa_vp->Parameters->ParameterValues[pi][0];
-			vp->params.body.f[dst_index++]=mesa_vp->Parameters->ParameterValues[pi][1];
-			vp->params.body.f[dst_index++]=mesa_vp->Parameters->ParameterValues[pi][2];
-			vp->params.body.f[dst_index++]=mesa_vp->Parameters->ParameterValues[pi][3];
+			*dst++=mesa_vp->Parameters->ParameterValues[pi][0];
+			*dst++=mesa_vp->Parameters->ParameterValues[pi][1];
+			*dst++=mesa_vp->Parameters->ParameterValues[pi][2];
+			*dst++=mesa_vp->Parameters->ParameterValues[pi][3];
 		break;
 		
 		default: _mesa_problem(NULL, "Bad param type in %s", __FUNCTION__);
@@ -274,7 +274,7 @@ void r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program *vp)
 	
 	}
 	
-	vp->params.length=dst_index;
+	return dst - dst_o;
 }
 		
 static unsigned long t_dst_mask(GLuint mask)
diff --git a/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c b/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
index b6292bb00e..34fe0952fe 100644
--- a/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
+++ b/src/mesa/drivers/dri/r300/radeon_vtxfmt_a.c
@@ -136,12 +136,6 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 	
 	switch (type) {
 	case GL_UNSIGNED_BYTE:
-		elt_size = 2;
-		
-		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
-		rvb.aos_offset = GET_START(&rvb);
-		ptr = rvb.address + rvb.start;
-			
 		for (i=0; i < count; i++) {
 			if(((unsigned char *)indices)[i] < min)
 				min = ((unsigned char *)indices)[i];
@@ -149,17 +143,17 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 				max = ((unsigned char *)indices)[i];
 		}
 		
-		for (i=0; i < count; i++)
-			((unsigned short int *)ptr)[i] = ((unsigned char *)indices)[i] - min;
-	break;
-		
-	case GL_UNSIGNED_SHORT:
 		elt_size = 2;
 		
 		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
 		rvb.aos_offset = GET_START(&rvb);
 		ptr = rvb.address + rvb.start;
+			
+		for (i=0; i < count; i++)
+			((unsigned short int *)ptr)[i] = ((unsigned char *)indices)[i] - min;
+	break;
 		
+	case GL_UNSIGNED_SHORT:
 		for (i=0; i < count; i++) {
 			if(((unsigned short int *)indices)[i] < min)
 				min = ((unsigned short int *)indices)[i];
@@ -167,17 +161,17 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 				max = ((unsigned short int *)indices)[i];
 		}
 		
-		for (i=0; i < count; i++)
-			((unsigned short int *)ptr)[i] = ((unsigned short int *)indices)[i] - min;
-	break;
-	
-	case GL_UNSIGNED_INT:
-		elt_size = 4;
+		elt_size = 2;
 		
 		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
 		rvb.aos_offset = GET_START(&rvb);
 		ptr = rvb.address + rvb.start;
 		
+		for (i=0; i < count; i++)
+			((unsigned short int *)ptr)[i] = ((unsigned short int *)indices)[i] - min;
+	break;
+	
+	case GL_UNSIGNED_INT:
 		for (i=0; i < count; i++) {
 			if(((unsigned int *)indices)[i] < min)
 				min = ((unsigned int *)indices)[i];
@@ -185,8 +179,22 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 				max = ((unsigned int *)indices)[i];
 		}
 		
-		for (i=0; i < count; i++)
-			((unsigned int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
+		if (max - min <= 65535)
+			elt_size = 2;
+		else 
+			elt_size = 4;
+		
+		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
+		rvb.aos_offset = GET_START(&rvb);
+		ptr = rvb.address + rvb.start;
+		
+		
+		if (max - min <= 65535)
+			for (i=0; i < count; i++)
+				((unsigned short int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
+		else
+			for (i=0; i < count; i++)
+				((unsigned int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
 	break;
 	
 	default:
@@ -199,7 +207,7 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 	if (ctx->NewState) 
 		_mesa_update_state( ctx );
 	
-	r300UpdateShaderStates(rmesa);
+	r300UpdateShaders(rmesa);
 	
 	if (rmesa->state.VB.LockCount) {
 		if (rmesa->state.VB.lock_uptodate == GL_FALSE) {
@@ -230,6 +238,8 @@ void radeonDrawElements( GLenum mode, GLsizei count, GLenum type, const GLvoid *
 		rmesa->state.VB.Count = max - min + 1;
 	}
 	
+	r300UpdateShaderStates(rmesa);
+	
 	rmesa->state.VB.Primitive = &prim;
 	rmesa->state.VB.PrimitiveCount = 1;
 	
@@ -256,7 +266,7 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 	struct tnl_prim prim;
 	int elt_size;
 	int i;
-	static void *ptr = NULL;
+	void *ptr = NULL;
 	static struct r300_dma_region rvb;
 	
 	if (ctx->Array.ElementArrayBufferObj->Name) {
@@ -275,7 +285,7 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 	
 	FLUSH_CURRENT( ctx, 0 );
 #ifdef OPTIMIZE_ELTS
-	start = 0;
+	min = 0;
 #endif
 	r300ReleaseDmaRegion(rmesa, &rvb, __FUNCTION__);
 	
@@ -295,7 +305,7 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 		elt_size = 2;
 		
 #ifdef OPTIMIZE_ELTS
-		if (start == 0 && ctx->Array.ElementArrayBufferObj->Name){
+		if (min == 0 && ctx->Array.ElementArrayBufferObj->Name){
 			ptr = indices;
 			break;
 		}
@@ -309,14 +319,21 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 	break;
 	
 	case GL_UNSIGNED_INT:
-		elt_size = 4;
+		if (max - min <= 65535)
+			elt_size = 2;
+		else 
+			elt_size = 4;
 		
 		r300AllocDmaRegion(rmesa, &rvb, count * elt_size, elt_size);
 		rvb.aos_offset = GET_START(&rvb);
 		ptr = rvb.address + rvb.start;
 		
-		for(i=0; i < count; i++)
-			((unsigned int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
+		if (max - min <= 65535)
+			for (i=0; i < count; i++)
+				((unsigned short int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
+		else
+			for (i=0; i < count; i++)
+				((unsigned int *)ptr)[i] = ((unsigned int *)indices)[i] - min;
 	break;
 	
 	default:
@@ -330,7 +347,7 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 	if (ctx->NewState) 
 		_mesa_update_state( ctx );
 	
-	r300UpdateShaderStates(rmesa);
+	r300UpdateShaders(rmesa);
 
 	if (rmesa->state.VB.LockCount) {
 		if (rmesa->state.VB.lock_uptodate == GL_FALSE) {
@@ -361,6 +378,8 @@ void radeonDrawRangeElements(GLenum mode, GLuint min, GLuint max, GLsizei count,
 		rmesa->state.VB.Count = max - min + 1;
 	}
 	
+	r300UpdateShaderStates(rmesa);
+	
 	rmesa->state.VB.Primitive = &prim;
 	rmesa->state.VB.PrimitiveCount = 1;
 	
@@ -398,7 +417,7 @@ void radeonDrawArrays( GLenum mode, GLint start, GLsizei count )
 	
 	/* XXX: setup_arrays before state update? */
 	
-	r300UpdateShaderStates(rmesa);
+	r300UpdateShaders(rmesa);
 
 	if (rmesa->state.VB.LockCount) {
 		if (rmesa->state.VB.lock_uptodate == GL_FALSE) {
@@ -429,6 +448,8 @@ void radeonDrawArrays( GLenum mode, GLint start, GLsizei count )
 		rmesa->state.VB.Count = count;
 	}
 
+	r300UpdateShaderStates(rmesa);
+	
 	rmesa->state.VB.Primitive = &prim;
 	rmesa->state.VB.PrimitiveCount = 1;