Get nv10_swtcl.c working enough for glxgears on NV40.

5 files changed, 96 insertions, 6 deletions

diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.c b/src/mesa/drivers/dri/nouveau/nouveau_context.c
index 1e25062c1db..22c1f588742 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.c
@@ -187,11 +187,9 @@ GLboolean nouveauCreateContext( const __GLcontextModes *glVisual,
 	nmesa->current_primitive = -1;
 
 	nouveauShaderInitFuncs(ctx);
-	/* Install Mesa's fixed-function shader support */
-	if (nmesa->screen->card->type >= NV_40) {
-		ctx->_MaintainTnlProgram    = GL_TRUE;
+	/* Install Mesa's fixed-function texenv shader support */
+	if (nmesa->screen->card->type >= NV_40)
 		ctx->_MaintainTexEnvProgram = GL_TRUE;
-	}
 
 	/* Initialize the swrast */
 	_swrast_CreateContext( ctx );
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_context.h b/src/mesa/drivers/dri/nouveau/nouveau_context.h
index 211d4e0a6d3..d7730bd7964 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_context.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_context.h
@@ -144,6 +144,7 @@ typedef struct nouveau_context {
 	nvsFunc FPfunc;
 	nouveauShader *current_fragprog;
 	nouveauShader *current_vertprog;
+	nouveauShader *passthrough_vp;
 
 	nouveauScreenRec *screen;
 	drm_nouveau_sarea_t *sarea;
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.c b/src/mesa/drivers/dri/nouveau/nouveau_shader.c
index e3082ebc69b..9a09f43d581 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.c
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader.c
@@ -37,6 +37,7 @@
 
 #include "program.h"
 #include "tnl/tnl.h"
+#include "shader/arbprogparse.h"
 
 #include "nouveau_context.h"
 #include "nouveau_shader.h"
@@ -161,6 +162,63 @@ nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs)
    return GL_TRUE;
 }
 
+nouveauShader *
+nvsBuildTextShader(GLcontext *ctx, GLenum target, const char *text)
+{
+   nouveauShader *nvs;
+
+   nvs = CALLOC_STRUCT(_nouveauShader);
+   if (!nvs)
+      return NULL;
+
+   if (target == GL_VERTEX_PROGRAM_ARB) {
+      _mesa_init_vertex_program(ctx, &nvs->mesa.vp, GL_VERTEX_PROGRAM_ARB, 0);
+      _mesa_parse_arb_vertex_program(ctx,
+	    			     GL_VERTEX_PROGRAM_ARB,
+				     text,
+				     strlen(text),
+				     &nvs->mesa.vp);
+   } else if (target == GL_FRAGMENT_PROGRAM_ARB) {
+      _mesa_init_fragment_program(ctx, &nvs->mesa.fp, GL_VERTEX_PROGRAM_ARB, 0);
+      _mesa_parse_arb_fragment_program(ctx,
+	    			       GL_FRAGMENT_PROGRAM_ARB,
+				       text,
+				       strlen(text),
+				       &nvs->mesa.fp);
+   }
+
+   nouveau_shader_pass0_arb(ctx, nvs);
+   nouveau_shader_pass1(nvs);
+   nouveau_shader_pass2(nvs);
+
+   return nvs;
+}
+
+static void
+nvsBuildPassthroughVP(GLcontext *ctx)
+{
+   nouveauContextPtr nmesa = NOUVEAU_CONTEXT(ctx);
+
+   const char *vp_text =
+      "!!ARBvp1.0\n"
+      "OPTION ARB_position_invariant;"
+      ""
+      "MOV result.color, vertex.color;\n"
+      "MOV result.texcoord[0], vertex.texcoord[0];\n"
+      "MOV result.texcoord[1], vertex.texcoord[1];\n"
+      "MOV result.texcoord[2], vertex.texcoord[2];\n"
+      "MOV result.texcoord[3], vertex.texcoord[3];\n"
+      "MOV result.texcoord[4], vertex.texcoord[4];\n"
+      "MOV result.texcoord[5], vertex.texcoord[5];\n"
+      "MOV result.texcoord[6], vertex.texcoord[6];\n"
+      "MOV result.texcoord[7], vertex.texcoord[7];\n"
+      "END";
+
+   nmesa->passthrough_vp = nvsBuildTextShader(ctx,
+	 				      GL_VERTEX_PROGRAM_ARB,
+					      vp_text);
+}
+
 void
 nouveauShaderInitFuncs(GLcontext * ctx)
 {
@@ -184,6 +242,11 @@ nouveauShaderInitFuncs(GLcontext * ctx)
       return;
    }
 
+   /* Build a vertex program that simply passes through all attribs.
+    * Needed to do swtcl on nv40
+    */
+   nvsBuildPassthroughVP(ctx);
+
    ctx->Const.VertexProgram.MaxNativeInstructions    = nmesa->VPfunc.MaxInst;
    ctx->Const.VertexProgram.MaxNativeAluInstructions = nmesa->VPfunc.MaxInst;
    ctx->Const.VertexProgram.MaxNativeTexInstructions = nmesa->VPfunc.MaxInst;
@@ -446,7 +509,7 @@ nvsDumpReg(nvsInstruction * inst, nvsRegister * reg)
       printf(")");
 }
 
-void
+static void
 nvsDumpInstruction(nvsInstruction * inst, int slot, int lvl)
 {
    struct _opcode_info *opr = &ops[inst->op];
diff --git a/src/mesa/drivers/dri/nouveau/nouveau_shader.h b/src/mesa/drivers/dri/nouveau/nouveau_shader.h
index dce2e23f46e..652775e6c23 100644
--- a/src/mesa/drivers/dri/nouveau/nouveau_shader.h
+++ b/src/mesa/drivers/dri/nouveau/nouveau_shader.h
@@ -349,6 +349,9 @@ nvsSwizzle(nvsRegister reg, nvsSwzComp x, nvsSwzComp y,
 
 extern GLboolean nvsUpdateShader(GLcontext *ctx, nouveauShader *nvs);
 extern void nvsDisasmHWShader(nvsPtr);
+extern void nvsDumpFragmentList(nvsFragmentList *f, int lvl);
+extern nouveauShader *nvsBuildTextShader(GLcontext *ctx, GLenum target,
+      					 const char *text);
 
 extern void NV20VPInitShaderFuncs(nvsFunc *);
 extern void NV30VPInitShaderFuncs(nvsFunc *);
diff --git a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
index 07b3e666df1..37d9f001d12 100644
--- a/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
+++ b/src/mesa/drivers/dri/nouveau/nv10_swtcl.c
@@ -366,6 +366,15 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 	int i;
 	int slots=0;
 	int total_size=0;
+	/* t_vertex_generic dereferences a NULL pointer if we
+	 * pass NULL as the vp transform...
+	 */
+	const GLfloat ident_vp[16] = {
+	   1.0, 0.0, 0.0, 0.0,
+	   0.0, 1.0, 0.0, 0.0,
+	   0.0, 0.0, 1.0, 0.0,
+	   0.0, 0.0, 0.0, 1.0
+	};
 
 	RENDERINPUTS_COPY(index, nmesa->render_inputs_bitset);
 
@@ -425,10 +434,11 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 			total_size+=attr_size[i];
 		}
 	}
+
 	nmesa->vertex_size=_tnl_install_attrs( ctx,
 			nmesa->vertex_attrs, 
 			nmesa->vertex_attr_count,
-			NULL, 0 );
+			ident_vp, 0 );
 	assert(nmesa->vertex_size==total_size*4);
 
 	/* 
@@ -467,6 +477,8 @@ static inline void nv10OutputVertexFormat(struct nouveau_context* nmesa)
 			OUT_RING_CACHE(NV_VERTEX_ATTRIBUTE_TYPE_FLOAT|(size*0x10));
 		}
 	} else {
+		BEGIN_RING_SIZE(NvSub3D, NV30_TCL_PRIMITIVE_3D_DO_VERTICES, 1);
+		OUT_RING(0);
 		BEGIN_RING_CACHE(NvSub3D,NV30_TCL_PRIMITIVE_3D_VERTEX_ATTR0_POS,slots);
 		for(i=0;i<slots;i++)
 		{
@@ -496,6 +508,19 @@ static void nv10ChooseVertexState( GLcontext *ctx )
 		RENDERINPUTS_COPY(nmesa->render_inputs_bitset, index);
 		nv10OutputVertexFormat(nmesa);
 	}
+
+	if (nmesa->screen->card->type >= NV_40) {
+		/* Ensure passthrough shader is being used, and mvp matrix
+		 * is up to date
+		 */
+		nvsUpdateShader(ctx, nmesa->passthrough_vp);
+		BEGIN_RING_CACHE(NvSub3D, NV30_TCL_PRIMITIVE_3D_VP_IN_REG, 2);
+		OUT_RING_CACHE  (0xff09);   /*IN : POS, COL, TC0-7 */
+		OUT_RING_CACHE  (0x3fc001); /*OUT: COL, TC0-7, POS implied */
+
+		/* Update texenv shader / user fragprog */
+		nvsUpdateShader(ctx, (nouveauShader*)ctx->FragmentProgram._Current);
+	}
 }