X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fdrivers%2Fdri%2Fr300%2Fr300_render.c;h=93fdc575885e2b32f0f5778a3db422187ad47b15;hb=42f16aa4e0d9f1c5f016919ed04c55430507234e;hp=7af9691cdeeb444e6fad449e816a9843c1457e50;hpb=88a409fa8e85a45eb71f99e5926340a5fe6a8071;p=mesa.git

diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 7af9691cdee..93fdc575885 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -66,16 +66,13 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #include "tnl/t_vp_build.h"
 #include "radeon_reg.h"
 #include "radeon_macros.h"
-#include "radeon_ioctl.h"
-#include "radeon_state.h"
 #include "r300_context.h"
 #include "r300_ioctl.h"
 #include "r300_state.h"
 #include "r300_reg.h"
 #include "r300_tex.h"
 #include "r300_emit.h"
-#include "r300_fragprog.h"
-extern int future_hw_tcl_on;
+#include "r300_fragprog_common.h"
 
 /**
  * \brief Convert a OpenGL primitive type into a R300 primitive type.
@@ -177,12 +174,12 @@ static void r300EmitElts(GLcontext * ctx, void *elts, unsigned long n_elts)
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	void *out;
 
-	radeonAllocDmaRegion(&rmesa->radeon, &rmesa->state.elt_dma_bo,
-			     &rmesa->state.elt_dma_offset, n_elts * 4, 4);
-	radeon_bo_map(rmesa->state.elt_dma_bo, 1);
-	out = rmesa->state.elt_dma_bo->ptr + rmesa->state.elt_dma_offset;
+	radeonAllocDmaRegion(&rmesa->radeon, &rmesa->radeon.tcl.elt_dma_bo,
+			     &rmesa->radeon.tcl.elt_dma_offset, n_elts * 4, 4);
+	radeon_bo_map(rmesa->radeon.tcl.elt_dma_bo, 1);
+	out = rmesa->radeon.tcl.elt_dma_bo->ptr + rmesa->radeon.tcl.elt_dma_offset;
 	memcpy(out, elts, n_elts * 4);
-	radeon_bo_unmap(rmesa->state.elt_dma_bo);
+	radeon_bo_unmap(rmesa->radeon.tcl.elt_dma_bo);
 }
 
 static void r300FireEB(r300ContextPtr rmesa, int vertex_count, int type)
@@ -190,28 +187,30 @@ static void r300FireEB(r300ContextPtr rmesa, int vertex_count, int type)
 	BATCH_LOCALS(&rmesa->radeon);
 
 	if (vertex_count > 0) {
-		BEGIN_BATCH(8);
+		BEGIN_BATCH(10);
 		OUT_BATCH_PACKET3(R300_PACKET3_3D_DRAW_INDX_2, 0);
 		OUT_BATCH(R300_VAP_VF_CNTL__PRIM_WALK_INDICES |
 			  ((vertex_count + 0) << 16) |
 			  type |
 			  R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
-		
+
 		if (!rmesa->radeon.radeonScreen->kernel_mm) {
 			OUT_BATCH_PACKET3(R300_PACKET3_INDX_BUFFER, 2);
-			OUT_BATCH(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
-			OUT_BATCH_RELOC(rmesa->state.elt_dma_offset,
-					rmesa->state.elt_dma_bo,
-					rmesa->state.elt_dma_offset,
+			OUT_BATCH(R300_INDX_BUFFER_ONE_REG_WR | (0 << R300_INDX_BUFFER_SKIP_SHIFT) |
+	    			 (R300_VAP_PORT_IDX0 >> 2));
+			OUT_BATCH_RELOC(rmesa->radeon.tcl.elt_dma_offset,
+					rmesa->radeon.tcl.elt_dma_bo,
+					rmesa->radeon.tcl.elt_dma_offset,
 					RADEON_GEM_DOMAIN_GTT, 0, 0);
 			OUT_BATCH(vertex_count);
 		} else {
 			OUT_BATCH_PACKET3(R300_PACKET3_INDX_BUFFER, 2);
-			OUT_BATCH(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
-			OUT_BATCH(rmesa->state.elt_dma_offset);
+			OUT_BATCH(R300_INDX_BUFFER_ONE_REG_WR | (0 << R300_INDX_BUFFER_SKIP_SHIFT) |
+	    			 (R300_VAP_PORT_IDX0 >> 2));
+			OUT_BATCH(rmesa->radeon.tcl.elt_dma_offset);
 			OUT_BATCH(vertex_count);
 			radeon_cs_write_reloc(rmesa->radeon.cmdbuf.cs,
-					      rmesa->state.elt_dma_bo,
+					      rmesa->radeon.tcl.elt_dma_bo,
 					      RADEON_GEM_DOMAIN_GTT, 0, 0);
 		}
 		END_BATCH();
@@ -224,46 +223,46 @@ static void r300EmitAOS(r300ContextPtr rmesa, GLuint nr, GLuint offset)
 	uint32_t voffset;
 	int sz = 1 + (nr >> 1) * 3 + (nr & 1) * 2;
 	int i;
-	
+
 	if (RADEON_DEBUG & DEBUG_VERTS)
 		fprintf(stderr, "%s: nr=%d, ofs=0x%08x\n", __FUNCTION__, nr,
 			offset);
 
-    
+
 	if (!rmesa->radeon.radeonScreen->kernel_mm) {
-		BEGIN_BATCH(sz+2);
+		BEGIN_BATCH(sz+2+(nr * 2));
 		OUT_BATCH_PACKET3(R300_PACKET3_3D_LOAD_VBPNTR, sz - 1);
 		OUT_BATCH(nr);
 
 		for (i = 0; i + 1 < nr; i += 2) {
-			OUT_BATCH((rmesa->state.aos[i].components << 0) |
-				  (rmesa->state.aos[i].stride << 8) |
-				  (rmesa->state.aos[i + 1].components << 16) |
-				  (rmesa->state.aos[i + 1].stride << 24));
-			
-			voffset =  rmesa->state.aos[i + 0].offset +
-				offset * 4 * rmesa->state.aos[i + 0].stride;
+			OUT_BATCH((rmesa->radeon.tcl.aos[i].components << 0) |
+				  (rmesa->radeon.tcl.aos[i].stride << 8) |
+				  (rmesa->radeon.tcl.aos[i + 1].components << 16) |
+				  (rmesa->radeon.tcl.aos[i + 1].stride << 24));
+
+			voffset =  rmesa->radeon.tcl.aos[i + 0].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[i + 0].stride;
 			OUT_BATCH_RELOC(voffset,
-					rmesa->state.aos[i].bo,
+					rmesa->radeon.tcl.aos[i].bo,
 					voffset,
 					RADEON_GEM_DOMAIN_GTT,
 					0, 0);
-			voffset =  rmesa->state.aos[i + 1].offset +
-			  offset * 4 * rmesa->state.aos[i + 1].stride;
+			voffset =  rmesa->radeon.tcl.aos[i + 1].offset +
+			  offset * 4 * rmesa->radeon.tcl.aos[i + 1].stride;
 			OUT_BATCH_RELOC(voffset,
-					rmesa->state.aos[i+1].bo,
+					rmesa->radeon.tcl.aos[i+1].bo,
 					voffset,
 					RADEON_GEM_DOMAIN_GTT,
 					0, 0);
 		}
-		
+
 		if (nr & 1) {
-			OUT_BATCH((rmesa->state.aos[nr - 1].components << 0) |
-				  (rmesa->state.aos[nr - 1].stride << 8));
-			voffset =  rmesa->state.aos[nr - 1].offset +
-				offset * 4 * rmesa->state.aos[nr - 1].stride;
+			OUT_BATCH((rmesa->radeon.tcl.aos[nr - 1].components << 0) |
+				  (rmesa->radeon.tcl.aos[nr - 1].stride << 8));
+			voffset =  rmesa->radeon.tcl.aos[nr - 1].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[nr - 1].stride;
 			OUT_BATCH_RELOC(voffset,
-					rmesa->state.aos[nr - 1].bo,
+					rmesa->radeon.tcl.aos[nr - 1].bo,
 					voffset,
 					RADEON_GEM_DOMAIN_GTT,
 					0, 0);
@@ -276,45 +275,45 @@ static void r300EmitAOS(r300ContextPtr rmesa, GLuint nr, GLuint offset)
 		OUT_BATCH(nr);
 
 		for (i = 0; i + 1 < nr; i += 2) {
-			OUT_BATCH((rmesa->state.aos[i].components << 0) |
-				  (rmesa->state.aos[i].stride << 8) |
-				  (rmesa->state.aos[i + 1].components << 16) |
-				  (rmesa->state.aos[i + 1].stride << 24));
-			
-			voffset =  rmesa->state.aos[i + 0].offset +
-				offset * 4 * rmesa->state.aos[i + 0].stride;
+			OUT_BATCH((rmesa->radeon.tcl.aos[i].components << 0) |
+				  (rmesa->radeon.tcl.aos[i].stride << 8) |
+				  (rmesa->radeon.tcl.aos[i + 1].components << 16) |
+				  (rmesa->radeon.tcl.aos[i + 1].stride << 24));
+
+			voffset =  rmesa->radeon.tcl.aos[i + 0].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[i + 0].stride;
 			OUT_BATCH(voffset);
-			voffset =  rmesa->state.aos[i + 1].offset +
-				offset * 4 * rmesa->state.aos[i + 1].stride;
+			voffset =  rmesa->radeon.tcl.aos[i + 1].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[i + 1].stride;
 			OUT_BATCH(voffset);
 		}
-		
+
 		if (nr & 1) {
-			OUT_BATCH((rmesa->state.aos[nr - 1].components << 0) |
-			  (rmesa->state.aos[nr - 1].stride << 8));
-			voffset =  rmesa->state.aos[nr - 1].offset +
-				offset * 4 * rmesa->state.aos[nr - 1].stride;
+			OUT_BATCH((rmesa->radeon.tcl.aos[nr - 1].components << 0) |
+			  (rmesa->radeon.tcl.aos[nr - 1].stride << 8));
+			voffset =  rmesa->radeon.tcl.aos[nr - 1].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[nr - 1].stride;
 			OUT_BATCH(voffset);
 		}
 		for (i = 0; i + 1 < nr; i += 2) {
-			voffset =  rmesa->state.aos[i + 0].offset +
-				offset * 4 * rmesa->state.aos[i + 0].stride;
+			voffset =  rmesa->radeon.tcl.aos[i + 0].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[i + 0].stride;
 			radeon_cs_write_reloc(rmesa->radeon.cmdbuf.cs,
-					      rmesa->state.aos[i+0].bo,
+					      rmesa->radeon.tcl.aos[i+0].bo,
 					      RADEON_GEM_DOMAIN_GTT,
 					      0, 0);
-			voffset =  rmesa->state.aos[i + 1].offset +
-				offset * 4 * rmesa->state.aos[i + 1].stride;
+			voffset =  rmesa->radeon.tcl.aos[i + 1].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[i + 1].stride;
 			radeon_cs_write_reloc(rmesa->radeon.cmdbuf.cs,
-					      rmesa->state.aos[i+1].bo,
+					      rmesa->radeon.tcl.aos[i+1].bo,
 					      RADEON_GEM_DOMAIN_GTT,
 					      0, 0);
 		}
 		if (nr & 1) {
-			voffset =  rmesa->state.aos[nr - 1].offset +
-				offset * 4 * rmesa->state.aos[nr - 1].stride;
+			voffset =  rmesa->radeon.tcl.aos[nr - 1].offset +
+				offset * 4 * rmesa->radeon.tcl.aos[nr - 1].stride;
 			radeon_cs_write_reloc(rmesa->radeon.cmdbuf.cs,
-					      rmesa->state.aos[nr-1].bo,
+					      rmesa->radeon.tcl.aos[nr-1].bo,
 					      RADEON_GEM_DOMAIN_GTT,
 					      0, 0);
 		}
@@ -336,7 +335,6 @@ static void r300FireAOS(r300ContextPtr rmesa, int vertex_count, int type)
 static void r300RunRenderPrimitive(r300ContextPtr rmesa, GLcontext * ctx,
 				   int start, int end, int prim)
 {
-	BATCH_LOCALS(&rmesa->radeon);
 	int type, num_verts;
 	TNLcontext *tnl = TNL_CONTEXT(ctx);
 	struct vertex_buffer *vb = &tnl->vb;
@@ -371,10 +369,10 @@ static void r300RunRenderPrimitive(r300ContextPtr rmesa, GLcontext * ctx,
 		 * arrays. *sigh*
 		 */
 		r300EmitElts(ctx, vb->Elts, num_verts);
-		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
+		r300EmitAOS(rmesa, rmesa->radeon.tcl.aos_count, start);
 		r300FireEB(rmesa, num_verts, type);
 	} else {
-		r300EmitAOS(rmesa, rmesa->state.aos_count, start);
+		r300EmitAOS(rmesa, rmesa->radeon.tcl.aos_count, start);
 		r300FireAOS(rmesa, num_verts, type);
 	}
 	COMMIT_BATCH();
@@ -398,7 +396,7 @@ static GLboolean r300RunRender(GLcontext * ctx,
 	r300UpdateShaderStates(rmesa);
 
 	r300EmitCacheFlush(rmesa);
-	r300EmitState(rmesa);
+	radeonEmitState(&rmesa->radeon);
 
 	for (i = 0; i < vb->PrimitiveCount; i++) {
 		GLuint prim = _tnl_translate_prim(&vb->Primitive[i]);
@@ -409,7 +407,7 @@ static GLboolean r300RunRender(GLcontext * ctx,
 
 	r300EmitCacheFlush(rmesa);
 
-	r300ReleaseArrays(ctx);
+	radeonReleaseArrays(ctx, ~0);
 
 	return GL_FALSE;
 }
@@ -427,36 +425,21 @@ static GLboolean r300RunRender(GLcontext * ctx,
 static int r300Fallback(GLcontext * ctx)
 {
 	r300ContextPtr r300 = R300_CONTEXT(ctx);
-	/* Do we need to use new-style shaders?
-	 * Also is there a better way to do this? */
-	if (r300->radeon.radeonScreen->chip_family >= CHIP_FAMILY_RV515) {
-		struct r500_fragment_program *fp = (struct r500_fragment_program *)
-	    (char *)ctx->FragmentProgram._Current;
-		if (fp) {
-			if (!fp->translated) {
-				r500TranslateFragmentShader(r300, fp);
-				FALLBACK_IF(!fp->translated);
-			}
-		}
-	} else {
-		struct r300_fragment_program *fp = (struct r300_fragment_program *)
-	    (char *)ctx->FragmentProgram._Current;
-		if (fp) {
-			if (!fp->translated) {
-				r300TranslateFragmentShader(r300, fp);
-				FALLBACK_IF(!fp->translated);
-			}
-		}
+	const unsigned back = ctx->Stencil._BackFace;
+
+	FALLBACK_IF(r300->radeon.Fallback);
+
+	struct r300_fragment_program *fp = (struct r300_fragment_program *) ctx->FragmentProgram._Current;
+	if (fp && !fp->translated) {
+		r300TranslateFragmentShader(ctx, ctx->FragmentProgram._Current);
+		FALLBACK_IF(fp->error);
 	}
 
 	FALLBACK_IF(ctx->RenderMode != GL_RENDER);
 
-	FALLBACK_IF(ctx->Stencil._TestTwoSide
-		    && (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[1]
-			|| ctx->Stencil.ValueMask[0] !=
-			ctx->Stencil.ValueMask[1]
-			|| ctx->Stencil.WriteMask[0] !=
-			ctx->Stencil.WriteMask[1]));
+	FALLBACK_IF(ctx->Stencil.Enabled && (ctx->Stencil.Ref[0] != ctx->Stencil.Ref[back]
+		    || ctx->Stencil.ValueMask[0] != ctx->Stencil.ValueMask[back]
+		    || ctx->Stencil.WriteMask[0] != ctx->Stencil.WriteMask[back]));
 
 	if (ctx->Extensions.NV_point_sprite || ctx->Extensions.ARB_point_sprite)
 		FALLBACK_IF(ctx->Point.PointSprite);
@@ -483,8 +466,11 @@ static GLboolean r300RunNonTCLRender(GLcontext * ctx,
 	if (r300Fallback(ctx) >= R300_FALLBACK_RAST)
 		return GL_TRUE;
 
-	if (!(rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL))
- 	        return GL_TRUE;
+	if (rmesa->options.hw_tcl_enabled == GL_FALSE)
+		return GL_TRUE;
+
+	if (!r300ValidateBuffers(ctx))
+	    return GL_TRUE;
 
 	return r300RunRender(ctx, stage);
 }
@@ -495,27 +481,25 @@ static GLboolean r300RunTCLRender(GLcontext * ctx,
 	r300ContextPtr rmesa = R300_CONTEXT(ctx);
 	struct r300_vertex_program *vp;
 
-	hw_tcl_on = future_hw_tcl_on;
-
 	if (RADEON_DEBUG & DEBUG_PRIMS)
 		fprintf(stderr, "%s\n", __FUNCTION__);
 
-	if (hw_tcl_on == GL_FALSE)
+	if (rmesa->options.hw_tcl_enabled == GL_FALSE)
 		return GL_TRUE;
 
 	if (r300Fallback(ctx) >= R300_FALLBACK_TCL) {
-		hw_tcl_on = GL_FALSE;
+		rmesa->options.hw_tcl_enabled = GL_FALSE;
 		return GL_TRUE;
 	}
 
 	if (!r300ValidateBuffers(ctx))
 	    return GL_TRUE;
-	
+
 	r300UpdateShaders(rmesa);
 
 	vp = (struct r300_vertex_program *)CURRENT_VERTEX_SHADER(ctx);
 	if (vp->native == GL_FALSE) {
-		hw_tcl_on = GL_FALSE;
+		rmesa->options.hw_tcl_enabled = GL_FALSE;
 		return GL_TRUE;
 	}