From cf3f601682297d94482a1448eff3f36b26514ab1 Mon Sep 17 00:00:00 2001
From: Brian <brian.paul@tungstengraphics.com>
Date: Wed, 31 Oct 2007 12:03:55 -0600
Subject: Lift fixed function vertex program generation up from tnl module.

---
 src/mesa/main/ffvertex_prog.c | 1550 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 1550 insertions(+)
 create mode 100644 src/mesa/main/ffvertex_prog.c

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
new file mode 100644
index 0000000000..8fcb9e5b14
--- /dev/null
+++ b/src/mesa/main/ffvertex_prog.c
@@ -0,0 +1,1550 @@
+/**************************************************************************
+ * 
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ * 
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ * 
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ * 
+ **************************************************************************/
+
+/**
+ * \file ffvertex_prog.
+ *
+ * Create a vertex program to execute the current fixed function T&L pipeline.
+ * \author Keith Whitwell
+ */
+
+
+#include "main/glheader.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/enums.h"
+#include "main/ffvertex_prog.h"
+#include "shader/program.h"
+#include "shader/prog_cache.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
+#include "shader/prog_statevars.h"
+
+
+struct state_key {
+   unsigned light_global_enabled:1;
+   unsigned light_local_viewer:1;
+   unsigned light_twoside:1;
+   unsigned light_color_material:1;
+   unsigned light_color_material_mask:12;
+   unsigned light_material_mask:12;
+
+   unsigned normalize:1;
+   unsigned rescale_normals:1;
+   unsigned fog_source_is_depth:1;
+   unsigned tnl_do_vertex_fog:1;
+   unsigned separate_specular:1;
+   unsigned fog_mode:2;
+   unsigned point_attenuated:1;
+   unsigned texture_enabled_global:1;
+   unsigned fragprog_inputs_read:12;
+
+   struct {
+      unsigned light_enabled:1;
+      unsigned light_eyepos3_is_zero:1;
+      unsigned light_spotcutoff_is_180:1;
+      unsigned light_attenuated:1;      
+      unsigned texunit_really_enabled:1;
+      unsigned texmat_enabled:1;
+      unsigned texgen_enabled:4;
+      unsigned texgen_mode0:4;
+      unsigned texgen_mode1:4;
+      unsigned texgen_mode2:4;
+      unsigned texgen_mode3:4;
+   } unit[8];
+};
+
+
+
+#define FOG_NONE   0
+#define FOG_LINEAR 1
+#define FOG_EXP    2
+#define FOG_EXP2   3
+
+static GLuint translate_fog_mode( GLenum mode )
+{
+   switch (mode) {
+   case GL_LINEAR: return FOG_LINEAR;
+   case GL_EXP: return FOG_EXP;
+   case GL_EXP2: return FOG_EXP2;
+   default: return FOG_NONE;
+   }
+}
+
+#define TXG_NONE           0
+#define TXG_OBJ_LINEAR     1
+#define TXG_EYE_LINEAR     2
+#define TXG_SPHERE_MAP     3
+#define TXG_REFLECTION_MAP 4
+#define TXG_NORMAL_MAP     5
+
+static GLuint translate_texgen( GLboolean enabled, GLenum mode )
+{
+   if (!enabled)
+      return TXG_NONE;
+
+   switch (mode) {
+   case GL_OBJECT_LINEAR: return TXG_OBJ_LINEAR;
+   case GL_EYE_LINEAR: return TXG_EYE_LINEAR;
+   case GL_SPHERE_MAP: return TXG_SPHERE_MAP;
+   case GL_REFLECTION_MAP_NV: return TXG_REFLECTION_MAP;
+   case GL_NORMAL_MAP_NV: return TXG_NORMAL_MAP;
+   default: return TXG_NONE;
+   }
+}
+
+
+/**
+ * Returns bitmask of flags indicating which materials are set per-vertex
+ * in the current VB.
+ * XXX get these from the VBO...
+ */
+static GLbitfield
+tnl_get_per_vertex_materials(GLcontext *ctx)
+{
+   GLbitfield mask = 0x0;
+#if 0
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   struct vertex_buffer *VB = &tnl->vb;
+   GLuint i;
+
+   for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) 
+      if (VB->AttribPtr[i] && VB->AttribPtr[i]->stride) 
+         mask |= 1 << (i - _TNL_FIRST_MAT);
+#endif
+   return mask;
+}
+
+/**
+ * Should fog be computed per-vertex?
+ */
+static GLboolean
+tnl_get_per_vertex_fog(GLcontext *ctx)
+{
+#if 0
+   TNLcontext *tnl = TNL_CONTEXT(ctx);
+   return tnl->_DoVertexFog;
+#else
+   return GL_FALSE;
+#endif
+}
+
+
+static struct state_key *make_state_key( GLcontext *ctx )
+{
+   const struct gl_fragment_program *fp;
+   struct state_key *key = CALLOC_STRUCT(state_key);
+   GLuint i;
+
+   fp = ctx->FragmentProgram._Current;
+
+   /* This now relies on texenvprogram.c being active:
+    */
+   assert(fp);
+
+   key->fragprog_inputs_read = fp->Base.InputsRead;
+
+   if (ctx->RenderMode == GL_FEEDBACK) {
+      /* make sure the vertprog emits color and tex0 */
+      key->fragprog_inputs_read |= (FRAG_BIT_COL0 | FRAG_BIT_TEX0);
+   }
+
+   key->separate_specular = (ctx->Light.Model.ColorControl ==
+			     GL_SEPARATE_SPECULAR_COLOR);
+
+   if (ctx->Light.Enabled) {
+      key->light_global_enabled = 1;
+
+      if (ctx->Light.Model.LocalViewer)
+	 key->light_local_viewer = 1;
+
+      if (ctx->Light.Model.TwoSide)
+	 key->light_twoside = 1;
+
+      if (ctx->Light.ColorMaterialEnabled) {
+	 key->light_color_material = 1;
+	 key->light_color_material_mask = ctx->Light.ColorMaterialBitmask;
+      }
+
+      key->light_material_mask = tnl_get_per_vertex_materials(ctx);
+
+      for (i = 0; i < MAX_LIGHTS; i++) {
+	 struct gl_light *light = &ctx->Light.Light[i];
+
+	 if (light->Enabled) {
+	    key->unit[i].light_enabled = 1;
+
+	    if (light->EyePosition[3] == 0.0)
+	       key->unit[i].light_eyepos3_is_zero = 1;
+	    
+	    if (light->SpotCutoff == 180.0)
+	       key->unit[i].light_spotcutoff_is_180 = 1;
+
+	    if (light->ConstantAttenuation != 1.0 ||
+		light->LinearAttenuation != 0.0 ||
+		light->QuadraticAttenuation != 0.0)
+	       key->unit[i].light_attenuated = 1;
+	 }
+      }
+   }
+
+   if (ctx->Transform.Normalize)
+      key->normalize = 1;
+
+   if (ctx->Transform.RescaleNormals)
+      key->rescale_normals = 1;
+
+   key->fog_mode = translate_fog_mode(fp->FogOption);
+   
+   if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT)
+      key->fog_source_is_depth = 1;
+   
+   key->tnl_do_vertex_fog = tnl_get_per_vertex_fog(ctx);
+
+   if (ctx->Point._Attenuated)
+      key->point_attenuated = 1;
+
+   if (ctx->Texture._TexGenEnabled ||
+       ctx->Texture._TexMatEnabled ||
+       ctx->Texture._EnabledUnits)
+      key->texture_enabled_global = 1;
+      
+   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+      struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
+
+      if (texUnit->_ReallyEnabled)
+	 key->unit[i].texunit_really_enabled = 1;
+
+      if (ctx->Texture._TexMatEnabled & ENABLE_TEXMAT(i))      
+	 key->unit[i].texmat_enabled = 1;
+      
+      if (texUnit->TexGenEnabled) {
+	 key->unit[i].texgen_enabled = 1;
+      
+	 key->unit[i].texgen_mode0 = 
+	    translate_texgen( texUnit->TexGenEnabled & (1<<0),
+			      texUnit->GenModeS );
+	 key->unit[i].texgen_mode1 = 
+	    translate_texgen( texUnit->TexGenEnabled & (1<<1),
+			      texUnit->GenModeT );
+	 key->unit[i].texgen_mode2 = 
+	    translate_texgen( texUnit->TexGenEnabled & (1<<2),
+			      texUnit->GenModeR );
+	 key->unit[i].texgen_mode3 = 
+	    translate_texgen( texUnit->TexGenEnabled & (1<<3),
+			      texUnit->GenModeQ );
+      }
+   }
+   
+   return key;
+}
+
+
+   
+/* Very useful debugging tool - produces annotated listing of
+ * generated program with line/function references for each
+ * instruction back into this file:
+ */
+#define DISASSEM (MESA_VERBOSE&VERBOSE_DISASSEM)
+
+/* Should be tunable by the driver - do we want to do matrix
+ * multiplications with DP4's or with MUL/MAD's?  SSE works better
+ * with the latter, drivers may differ.
+ */
+#define PREFER_DP4 0
+
+#define MAX_INSN 256
+
+/* Use uregs to represent registers internally, translate to Mesa's
+ * expected formats on emit.  
+ *
+ * NOTE: These are passed by value extensively in this file rather
+ * than as usual by pointer reference.  If this disturbs you, try
+ * remembering they are just 32bits in size.
+ *
+ * GCC is smart enough to deal with these dword-sized structures in
+ * much the same way as if I had defined them as dwords and was using
+ * macros to access and set the fields.  This is much nicer and easier
+ * to evolve.
+ */
+struct ureg {
+   GLuint file:4;
+   GLint idx:8;      /* relative addressing may be negative */
+   GLuint negate:1;
+   GLuint swz:12;
+   GLuint pad:7;
+};
+
+
+struct tnl_program {
+   const struct state_key *state;
+   struct gl_vertex_program *program;
+   
+   GLuint temp_in_use;
+   GLuint temp_reserved;
+   
+   struct ureg eye_position;
+   struct ureg eye_position_normalized;
+   struct ureg eye_normal;
+   struct ureg identity;
+
+   GLuint materials;
+   GLuint color_materials;
+};
+
+
+static const struct ureg undef = { 
+   PROGRAM_UNDEFINED,
+   ~0,
+   0,
+   0,
+   0
+};
+
+/* Local shorthand:
+ */
+#define X    SWIZZLE_X
+#define Y    SWIZZLE_Y
+#define Z    SWIZZLE_Z
+#define W    SWIZZLE_W
+
+
+/* Construct a ureg:
+ */
+static struct ureg make_ureg(GLuint file, GLint idx)
+{
+   struct ureg reg;
+   reg.file = file;
+   reg.idx = idx;
+   reg.negate = 0;
+   reg.swz = SWIZZLE_NOOP;
+   reg.pad = 0;
+   return reg;
+}
+
+
+
+static struct ureg negate( struct ureg reg )
+{
+   reg.negate ^= 1;
+   return reg;
+} 
+
+
+static struct ureg swizzle( struct ureg reg, int x, int y, int z, int w )
+{
+   reg.swz = MAKE_SWIZZLE4(GET_SWZ(reg.swz, x),
+			   GET_SWZ(reg.swz, y),
+			   GET_SWZ(reg.swz, z),
+			   GET_SWZ(reg.swz, w));
+
+   return reg;
+}
+
+static struct ureg swizzle1( struct ureg reg, int x )
+{
+   return swizzle(reg, x, x, x, x);
+}
+
+static struct ureg get_temp( struct tnl_program *p )
+{
+   int bit = _mesa_ffs( ~p->temp_in_use );
+   if (!bit) {
+      _mesa_problem(NULL, "%s: out of temporaries\n", __FILE__);
+      _mesa_exit(1);
+   }
+
+   if ((GLuint) bit > p->program->Base.NumTemporaries)
+      p->program->Base.NumTemporaries = bit;
+
+   p->temp_in_use |= 1<<(bit-1);
+   return make_ureg(PROGRAM_TEMPORARY, bit-1);
+}
+
+static struct ureg reserve_temp( struct tnl_program *p )
+{
+   struct ureg temp = get_temp( p );
+   p->temp_reserved |= 1<<temp.idx;
+   return temp;
+}
+
+static void release_temp( struct tnl_program *p, struct ureg reg )
+{
+   if (reg.file == PROGRAM_TEMPORARY) {
+      p->temp_in_use &= ~(1<<reg.idx);
+      p->temp_in_use |= p->temp_reserved; /* can't release reserved temps */
+   }
+}
+
+static void release_temps( struct tnl_program *p )
+{
+   p->temp_in_use = p->temp_reserved;
+}
+
+
+
+static struct ureg register_input( struct tnl_program *p, GLuint input )
+{
+   p->program->Base.InputsRead |= (1<<input);
+   return make_ureg(PROGRAM_INPUT, input);
+}
+
+static struct ureg register_output( struct tnl_program *p, GLuint output )
+{
+   p->program->Base.OutputsWritten |= (1<<output);
+   return make_ureg(PROGRAM_OUTPUT, output);
+}
+
+static struct ureg register_const4f( struct tnl_program *p, 
+			      GLfloat s0,
+			      GLfloat s1,
+			      GLfloat s2,
+			      GLfloat s3)
+{
+   GLfloat values[4];
+   GLint idx;
+   GLuint swizzle;
+   values[0] = s0;
+   values[1] = s1;
+   values[2] = s2;
+   values[3] = s3;
+   idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
+                                     &swizzle );
+   ASSERT(swizzle == SWIZZLE_NOOP);
+   return make_ureg(PROGRAM_STATE_VAR, idx);
+}
+
+#define register_const1f(p, s0)         register_const4f(p, s0, 0, 0, 1)
+#define register_scalar_const(p, s0)    register_const4f(p, s0, s0, s0, s0)
+#define register_const2f(p, s0, s1)     register_const4f(p, s0, s1, 0, 1)
+#define register_const3f(p, s0, s1, s2) register_const4f(p, s0, s1, s2, 1)
+
+static GLboolean is_undef( struct ureg reg )
+{
+   return reg.file == PROGRAM_UNDEFINED;
+}
+
+static struct ureg get_identity_param( struct tnl_program *p )
+{
+   if (is_undef(p->identity)) 
+      p->identity = register_const4f(p, 0,0,0,1);
+
+   return p->identity;
+}
+
+static struct ureg register_param5(struct tnl_program *p, 
+				   GLint s0,
+				   GLint s1,
+				   GLint s2,
+				   GLint s3,
+                                   GLint s4)
+{
+   gl_state_index tokens[STATE_LENGTH];
+   GLint idx;
+   tokens[0] = s0;
+   tokens[1] = s1;
+   tokens[2] = s2;
+   tokens[3] = s3;
+   tokens[4] = s4;
+   idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
+   return make_ureg(PROGRAM_STATE_VAR, idx);
+}
+
+
+#define register_param1(p,s0)          register_param5(p,s0,0,0,0,0)
+#define register_param2(p,s0,s1)       register_param5(p,s0,s1,0,0,0)
+#define register_param3(p,s0,s1,s2)    register_param5(p,s0,s1,s2,0,0)
+#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
+
+
+static void register_matrix_param5( struct tnl_program *p,
+				    GLint s0, /* modelview, projection, etc */
+				    GLint s1, /* texture matrix number */
+				    GLint s2, /* first row */
+				    GLint s3, /* last row */
+				    GLint s4, /* inverse, transpose, etc */
+				    struct ureg *matrix )
+{
+   GLint i;
+
+   /* This is a bit sad as the support is there to pull the whole
+    * matrix out in one go:
+    */
+   for (i = 0; i <= s3 - s2; i++) 
+      matrix[i] = register_param5( p, s0, s1, i, i, s4 );
+}
+
+
+static void emit_arg( struct prog_src_register *src,
+		      struct ureg reg )
+{
+   src->File = reg.file;
+   src->Index = reg.idx;
+   src->Swizzle = reg.swz;
+   src->NegateBase = reg.negate ? NEGATE_XYZW : 0;
+   src->Abs = 0;
+   src->NegateAbs = 0;
+   src->RelAddr = 0;
+}
+
+static void emit_dst( struct prog_dst_register *dst,
+		      struct ureg reg, GLuint mask )
+{
+   dst->File = reg.file;
+   dst->Index = reg.idx;
+   /* allow zero as a shorthand for xyzw */
+   dst->WriteMask = mask ? mask : WRITEMASK_XYZW; 
+   dst->CondMask = COND_TR;  /* always pass cond test */
+   dst->CondSwizzle = SWIZZLE_NOOP;
+   dst->CondSrc = 0;
+   dst->pad = 0;
+}
+
+static void debug_insn( struct prog_instruction *inst, const char *fn,
+			GLuint line )
+{
+   if (DISASSEM) {
+      static const char *last_fn;
+   
+      if (fn != last_fn) {
+	 last_fn = fn;
+	 _mesa_printf("%s:\n", fn);
+      }
+	 
+      _mesa_printf("%d:\t", line);
+      _mesa_print_instruction(inst);
+   }
+}
+
+
+static void emit_op3fn(struct tnl_program *p,
+                       enum prog_opcode op,
+		       struct ureg dest,
+		       GLuint mask,
+		       struct ureg src0,
+		       struct ureg src1,
+		       struct ureg src2,
+		       const char *fn,
+		       GLuint line)
+{
+   GLuint nr = p->program->Base.NumInstructions++;
+   struct prog_instruction *inst = &p->program->Base.Instructions[nr];
+      
+   if (p->program->Base.NumInstructions > MAX_INSN) {
+      _mesa_problem(0, "Out of instructions in emit_op3fn\n");
+      return;
+   }
+      
+   inst->Opcode = (enum prog_opcode) op; 
+   inst->StringPos = 0;
+   inst->Data = 0;
+   
+   emit_arg( &inst->SrcReg[0], src0 );
+   emit_arg( &inst->SrcReg[1], src1 );
+   emit_arg( &inst->SrcReg[2], src2 );   
+
+   emit_dst( &inst->DstReg, dest, mask );
+
+   debug_insn(inst, fn, line);
+}
+
+
+#define emit_op3(p, op, dst, mask, src0, src1, src2) \
+   emit_op3fn(p, op, dst, mask, src0, src1, src2, __FUNCTION__, __LINE__)
+
+#define emit_op2(p, op, dst, mask, src0, src1) \
+    emit_op3fn(p, op, dst, mask, src0, src1, undef, __FUNCTION__, __LINE__)
+
+#define emit_op1(p, op, dst, mask, src0) \
+    emit_op3fn(p, op, dst, mask, src0, undef, undef, __FUNCTION__, __LINE__)
+
+
+static struct ureg make_temp( struct tnl_program *p, struct ureg reg )
+{
+   if (reg.file == PROGRAM_TEMPORARY && 
+       !(p->temp_reserved & (1<<reg.idx)))
+      return reg;
+   else {
+      struct ureg temp = get_temp(p);
+      emit_op1(p, OPCODE_MOV, temp, 0, reg);
+      return temp;
+   }
+}
+
+
+/* Currently no tracking performed of input/output/register size or
+ * active elements.  Could be used to reduce these operations, as
+ * could the matrix type.
+ */
+static void emit_matrix_transform_vec4( struct tnl_program *p,
+					struct ureg dest,
+					const struct ureg *mat,
+					struct ureg src)
+{
+   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_X, src, mat[0]);
+   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_Y, src, mat[1]);
+   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_Z, src, mat[2]);
+   emit_op2(p, OPCODE_DP4, dest, WRITEMASK_W, src, mat[3]);
+}
+
+/* This version is much easier to implement if writemasks are not
+ * supported natively on the target or (like SSE), the target doesn't
+ * have a clean/obvious dotproduct implementation.
+ */
+static void emit_transpose_matrix_transform_vec4( struct tnl_program *p,
+						  struct ureg dest,
+						  const struct ureg *mat,
+						  struct ureg src)
+{
+   struct ureg tmp;
+
+   if (dest.file != PROGRAM_TEMPORARY)
+      tmp = get_temp(p);
+   else
+      tmp = dest;
+
+   emit_op2(p, OPCODE_MUL, tmp, 0, swizzle1(src,X), mat[0]);
+   emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Y), mat[1], tmp);
+   emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Z), mat[2], tmp);
+   emit_op3(p, OPCODE_MAD, dest, 0, swizzle1(src,W), mat[3], tmp);
+
+   if (dest.file != PROGRAM_TEMPORARY)
+      release_temp(p, tmp);
+}
+
+static void emit_matrix_transform_vec3( struct tnl_program *p,
+					struct ureg dest,
+					const struct ureg *mat,
+					struct ureg src)
+{
+   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_X, src, mat[0]);
+   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_Y, src, mat[1]);
+   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_Z, src, mat[2]);
+}
+
+
+static void emit_normalize_vec3( struct tnl_program *p,
+				 struct ureg dest,
+				 struct ureg src )
+{
+   struct ureg tmp = get_temp(p);
+   emit_op2(p, OPCODE_DP3, tmp, 0, src, src);
+   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp);
+   emit_op2(p, OPCODE_MUL, dest, 0, src, tmp);
+   release_temp(p, tmp);
+}
+
+static void emit_passthrough( struct tnl_program *p, 
+			      GLuint input,
+			      GLuint output )
+{
+   struct ureg out = register_output(p, output);
+   emit_op1(p, OPCODE_MOV, out, 0, register_input(p, input)); 
+}
+
+static struct ureg get_eye_position( struct tnl_program *p )
+{
+   if (is_undef(p->eye_position)) {
+      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+      struct ureg modelview[4];
+
+      p->eye_position = reserve_temp(p);
+
+      if (PREFER_DP4) {
+	 register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
+                                 0, modelview );
+
+	 emit_matrix_transform_vec4(p, p->eye_position, modelview, pos);
+      }
+      else {
+	 register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
+				 STATE_MATRIX_TRANSPOSE, modelview );
+
+	 emit_transpose_matrix_transform_vec4(p, p->eye_position, modelview, pos);
+      }
+   }
+   
+   return p->eye_position;
+}
+
+
+static struct ureg get_eye_position_normalized( struct tnl_program *p )
+{
+   if (is_undef(p->eye_position_normalized)) {
+      struct ureg eye = get_eye_position(p);
+      p->eye_position_normalized = reserve_temp(p);
+      emit_normalize_vec3(p, p->eye_position_normalized, eye);
+   }
+   
+   return p->eye_position_normalized;
+}
+
+
+static struct ureg get_eye_normal( struct tnl_program *p )
+{
+   if (is_undef(p->eye_normal)) {
+      struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
+      struct ureg mvinv[3];
+
+      register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
+			      STATE_MATRIX_INVTRANS, mvinv );
+
+      p->eye_normal = reserve_temp(p);
+
+      /* Transform to eye space:
+       */
+      emit_matrix_transform_vec3( p, p->eye_normal, mvinv, normal );
+
+      /* Normalize/Rescale:
+       */
+      if (p->state->normalize) {
+	 emit_normalize_vec3( p, p->eye_normal, p->eye_normal );
+      }
+      else if (p->state->rescale_normals) {
+	 struct ureg rescale = register_param2(p, STATE_INTERNAL,
+					       STATE_NORMAL_SCALE);
+
+	 emit_op2( p, OPCODE_MUL, p->eye_normal, 0, p->eye_normal,
+		   swizzle1(rescale, X));
+      }
+   }
+
+   return p->eye_normal;
+}
+
+
+
+static void build_hpos( struct tnl_program *p )
+{
+   struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+   struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
+   struct ureg mvp[4];
+
+   if (PREFER_DP4) {
+      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
+			      0, mvp );
+      emit_matrix_transform_vec4( p, hpos, mvp, pos );
+   }
+   else {
+      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
+			      STATE_MATRIX_TRANSPOSE, mvp );
+      emit_transpose_matrix_transform_vec4( p, hpos, mvp, pos );
+   }
+}
+
+
+static GLuint material_attrib( GLuint side, GLuint property )
+{
+   return ((property - STATE_AMBIENT) * 2 + 
+	   side);
+}
+
+/* Get a bitmask of which material values vary on a per-vertex basis.
+ */
+static void set_material_flags( struct tnl_program *p )
+{
+   p->color_materials = 0;
+   p->materials = 0;
+
+   if (p->state->light_color_material) {
+      p->materials = 
+	 p->color_materials = p->state->light_color_material_mask;
+   }
+
+   p->materials |= p->state->light_material_mask;
+}
+
+
+/* XXX temporary!!! */
+#define _TNL_ATTRIB_MAT_FRONT_AMBIENT 32
+
+static struct ureg get_material( struct tnl_program *p, GLuint side, 
+				 GLuint property )
+{
+   GLuint attrib = material_attrib(side, property);
+
+   if (p->color_materials & (1<<attrib))
+      return register_input(p, VERT_ATTRIB_COLOR0);
+   else if (p->materials & (1<<attrib)) 
+      return register_input( p, attrib + _TNL_ATTRIB_MAT_FRONT_AMBIENT );
+   else
+      return register_param3( p, STATE_MATERIAL, side, property );
+}
+
+#define SCENE_COLOR_BITS(side) (( MAT_BIT_FRONT_EMISSION | \
+				   MAT_BIT_FRONT_AMBIENT | \
+				   MAT_BIT_FRONT_DIFFUSE) << (side))
+
+/* Either return a precalculated constant value or emit code to
+ * calculate these values dynamically in the case where material calls
+ * are present between begin/end pairs.
+ *
+ * Probably want to shift this to the program compilation phase - if
+ * we always emitted the calculation here, a smart compiler could
+ * detect that it was constant (given a certain set of inputs), and
+ * lift it out of the main loop.  That way the programs created here
+ * would be independent of the vertex_buffer details.
+ */
+static struct ureg get_scenecolor( struct tnl_program *p, GLuint side )
+{
+   if (p->materials & SCENE_COLOR_BITS(side)) {
+      struct ureg lm_ambient = register_param1(p, STATE_LIGHTMODEL_AMBIENT);
+      struct ureg material_emission = get_material(p, side, STATE_EMISSION);
+      struct ureg material_ambient = get_material(p, side, STATE_AMBIENT);
+      struct ureg material_diffuse = get_material(p, side, STATE_DIFFUSE);
+      struct ureg tmp = make_temp(p, material_diffuse);
+      emit_op3(p, OPCODE_MAD, tmp,  WRITEMASK_XYZ, lm_ambient, 
+	       material_ambient, material_emission);
+      return tmp;
+   }
+   else
+      return register_param2( p, STATE_LIGHTMODEL_SCENECOLOR, side );
+}
+
+
+static struct ureg get_lightprod( struct tnl_program *p, GLuint light, 
+				  GLuint side, GLuint property )
+{
+   GLuint attrib = material_attrib(side, property);
+   if (p->materials & (1<<attrib)) {
+      struct ureg light_value = 
+	 register_param3(p, STATE_LIGHT, light, property);
+      struct ureg material_value = get_material(p, side, property);
+      struct ureg tmp = get_temp(p);
+      emit_op2(p, OPCODE_MUL, tmp,  0, light_value, material_value);
+      return tmp;
+   }
+   else
+      return register_param4(p, STATE_LIGHTPROD, light, side, property);
+}
+
+static struct ureg calculate_light_attenuation( struct tnl_program *p,
+						GLuint i, 
+						struct ureg VPpli,
+						struct ureg dist )
+{
+   struct ureg attenuation = register_param3(p, STATE_LIGHT, i,
+					     STATE_ATTENUATION);
+   struct ureg att = get_temp(p);
+
+   /* Calculate spot attenuation:
+    */
+   if (!p->state->unit[i].light_spotcutoff_is_180) {
+      struct ureg spot_dir_norm = register_param3(p, STATE_INTERNAL,
+						  STATE_SPOT_DIR_NORMALIZED, i);
+      struct ureg spot = get_temp(p);
+      struct ureg slt = get_temp(p);
+
+      emit_op2(p, OPCODE_DP3, spot, 0, negate(VPpli), spot_dir_norm);
+      emit_op2(p, OPCODE_SLT, slt, 0, swizzle1(spot_dir_norm,W), spot);
+      emit_op2(p, OPCODE_POW, spot, 0, spot, swizzle1(attenuation, W));
+      emit_op2(p, OPCODE_MUL, att, 0, slt, spot);
+
+      release_temp(p, spot);
+      release_temp(p, slt);
+   }
+
+   /* Calculate distance attenuation:
+    */
+   if (p->state->unit[i].light_attenuated) {
+
+      /* 1/d,d,d,1/d */
+      emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist); 
+      /* 1,d,d*d,1/d */
+      emit_op2(p, OPCODE_MUL, dist, WRITEMASK_XZ, dist, swizzle1(dist,Y)); 
+      /* 1/dist-atten */
+      emit_op2(p, OPCODE_DP3, dist, 0, attenuation, dist); 
+
+      if (!p->state->unit[i].light_spotcutoff_is_180) {
+	 /* dist-atten */
+	 emit_op1(p, OPCODE_RCP, dist, 0, dist); 
+	 /* spot-atten * dist-atten */
+	 emit_op2(p, OPCODE_MUL, att, 0, dist, att);	
+      } else {
+	 /* dist-atten */
+	 emit_op1(p, OPCODE_RCP, att, 0, dist); 
+      }
+   }
+
+   return att;
+}
+						
+
+
+
+
+/* Need to add some addtional parameters to allow lighting in object
+ * space - STATE_SPOT_DIRECTION and STATE_HALF_VECTOR implicitly assume eye
+ * space lighting.
+ */
+static void build_lighting( struct tnl_program *p )
+{
+   const GLboolean twoside = p->state->light_twoside;
+   const GLboolean separate = p->state->separate_specular;
+   GLuint nr_lights = 0, count = 0;
+   struct ureg normal = get_eye_normal(p);
+   struct ureg lit = get_temp(p);
+   struct ureg dots = get_temp(p);
+   struct ureg _col0 = undef, _col1 = undef;
+   struct ureg _bfc0 = undef, _bfc1 = undef;
+   GLuint i;
+
+   for (i = 0; i < MAX_LIGHTS; i++) 
+      if (p->state->unit[i].light_enabled)
+	 nr_lights++;
+   
+   set_material_flags(p);
+
+   {
+      struct ureg shininess = get_material(p, 0, STATE_SHININESS);
+      emit_op1(p, OPCODE_MOV, dots,  WRITEMASK_W, swizzle1(shininess,X));
+      release_temp(p, shininess);
+
+      _col0 = make_temp(p, get_scenecolor(p, 0));
+      if (separate)
+	 _col1 = make_temp(p, get_identity_param(p));
+      else
+	 _col1 = _col0;
+
+   }
+
+   if (twoside) {
+      struct ureg shininess = get_material(p, 1, STATE_SHININESS);
+      emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
+	       negate(swizzle1(shininess,X)));
+      release_temp(p, shininess);
+
+      _bfc0 = make_temp(p, get_scenecolor(p, 1));
+      if (separate)
+	 _bfc1 = make_temp(p, get_identity_param(p));
+      else
+	 _bfc1 = _bfc0;
+   }
+
+
+   /* If no lights, still need to emit the scenecolor.
+    */
+      {
+	 struct ureg res0 = register_output( p, VERT_RESULT_COL0 );
+	 emit_op1(p, OPCODE_MOV, res0, 0, _col0);
+      }
+
+      if (separate) {
+	 struct ureg res1 = register_output( p, VERT_RESULT_COL1 );
+	 emit_op1(p, OPCODE_MOV, res1, 0, _col1);
+      }
+
+      if (twoside) {
+	 struct ureg res0 = register_output( p, VERT_RESULT_BFC0 );
+	 emit_op1(p, OPCODE_MOV, res0, 0, _bfc0);
+      }
+      
+      if (twoside && separate) {
+	 struct ureg res1 = register_output( p, VERT_RESULT_BFC1 );
+	 emit_op1(p, OPCODE_MOV, res1, 0, _bfc1);
+      }
+      
+   if (nr_lights == 0) {
+      release_temps(p);
+      return;
+   }
+
+
+   for (i = 0; i < MAX_LIGHTS; i++) {
+      if (p->state->unit[i].light_enabled) {
+	 struct ureg half = undef;
+	 struct ureg att = undef, VPpli = undef;
+	  
+	 count++;
+
+	 if (p->state->unit[i].light_eyepos3_is_zero) {
+	    /* Can used precomputed constants in this case.
+	     * Attenuation never applies to infinite lights.
+	     */
+	    VPpli = register_param3(p, STATE_LIGHT, i, 
+				    STATE_POSITION_NORMALIZED); 
+            if (p->state->light_local_viewer) {
+                struct ureg eye_hat = get_eye_position_normalized(p);
+                half = get_temp(p);
+                emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+                emit_normalize_vec3(p, half, half);
+            } else {
+                half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+            }
+	 } 
+	 else {
+	    struct ureg Ppli = register_param3(p, STATE_LIGHT, i, 
+					       STATE_POSITION); 
+	    struct ureg V = get_eye_position(p);
+	    struct ureg dist = get_temp(p);
+
+	    VPpli = get_temp(p); 
+	    half = get_temp(p);
+ 
+	    /* Calulate VPpli vector
+	     */
+	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); 
+
+	    /* Normalize VPpli.  The dist value also used in
+	     * attenuation below.
+	     */
+	    emit_op2(p, OPCODE_DP3, dist, 0, VPpli, VPpli);
+	    emit_op1(p, OPCODE_RSQ, dist, 0, dist);
+	    emit_op2(p, OPCODE_MUL, VPpli, 0, VPpli, dist);
+
+
+	    /* Calculate  attenuation:
+	     */ 
+	    if (!p->state->unit[i].light_spotcutoff_is_180 ||
+		p->state->unit[i].light_attenuated) {
+	       att = calculate_light_attenuation(p, i, VPpli, dist);
+	    }
+	 
+      
+	    /* Calculate viewer direction, or use infinite viewer:
+	     */
+	    if (p->state->light_local_viewer) {
+	       struct ureg eye_hat = get_eye_position_normalized(p);
+	       emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+	    }
+	    else {
+	       struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z); 
+	       emit_op2(p, OPCODE_ADD, half, 0, VPpli, z_dir);
+	    }
+
+	    emit_normalize_vec3(p, half, half);
+
+	    release_temp(p, dist);
+	 }
+
+	 /* Calculate dot products:
+	  */
+	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
+	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
+
+	
+	 /* Front face lighting:
+	  */
+	 {
+	    struct ureg ambient = get_lightprod(p, i, 0, STATE_AMBIENT);
+	    struct ureg diffuse = get_lightprod(p, i, 0, STATE_DIFFUSE);
+	    struct ureg specular = get_lightprod(p, i, 0, STATE_SPECULAR);
+	    struct ureg res0, res1;
+	    GLuint mask0, mask1;
+
+	    emit_op1(p, OPCODE_LIT, lit, 0, dots);
+   
+	    if (!is_undef(att)) 
+	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
+   
+	    
+	    if (count == nr_lights) {
+	       if (separate) {
+		  mask0 = WRITEMASK_XYZ;
+		  mask1 = WRITEMASK_XYZ;
+		  res0 = register_output( p, VERT_RESULT_COL0 );
+		  res1 = register_output( p, VERT_RESULT_COL1 );
+	       }
+	       else {
+		  mask0 = 0;
+		  mask1 = WRITEMASK_XYZ;
+		  res0 = _col0;
+		  res1 = register_output( p, VERT_RESULT_COL0 );
+	       }
+	    } else {
+	       mask0 = 0;
+	       mask1 = 0;
+	       res0 = _col0;
+	       res1 = _col1;
+	    }
+
+	    emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
+	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _col0);
+	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _col1);
+      
+	    release_temp(p, ambient);
+	    release_temp(p, diffuse);
+	    release_temp(p, specular);
+	 }
+
+	 /* Back face lighting:
+	  */
+	 if (twoside) {
+	    struct ureg ambient = get_lightprod(p, i, 1, STATE_AMBIENT);
+	    struct ureg diffuse = get_lightprod(p, i, 1, STATE_DIFFUSE);
+	    struct ureg specular = get_lightprod(p, i, 1, STATE_SPECULAR);
+	    struct ureg res0, res1;
+	    GLuint mask0, mask1;
+	       
+	    emit_op1(p, OPCODE_LIT, lit, 0, negate(swizzle(dots,X,Y,W,Z)));
+
+	    if (!is_undef(att)) 
+	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
+
+	    if (count == nr_lights) {
+	       if (separate) {
+		  mask0 = WRITEMASK_XYZ;
+		  mask1 = WRITEMASK_XYZ;
+		  res0 = register_output( p, VERT_RESULT_BFC0 );
+		  res1 = register_output( p, VERT_RESULT_BFC1 );
+	       }
+	       else {
+		  mask0 = 0;
+		  mask1 = WRITEMASK_XYZ;
+		  res0 = _bfc0;
+		  res1 = register_output( p, VERT_RESULT_BFC0 );
+	       }
+	    } else {
+	       res0 = _bfc0;
+	       res1 = _bfc1;
+	       mask0 = 0;
+	       mask1 = 0;
+	    }
+
+	    emit_op3(p, OPCODE_MAD, _bfc0, 0, swizzle1(lit,X), ambient, _bfc0);
+	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
+	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
+
+	    release_temp(p, ambient);
+	    release_temp(p, diffuse);
+	    release_temp(p, specular);
+	 }
+
+	 release_temp(p, half);
+	 release_temp(p, VPpli);
+	 release_temp(p, att);
+      }
+   }
+
+   release_temps( p );
+}
+
+
+static void build_fog( struct tnl_program *p )
+{
+   struct ureg fog = register_output(p, VERT_RESULT_FOGC);
+   struct ureg input;
+
+   if (p->state->fog_source_is_depth) {
+      input = swizzle1(get_eye_position(p), Z);
+   }
+   else {
+      input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
+   }
+
+   if (p->state->fog_mode && p->state->tnl_do_vertex_fog) {
+      struct ureg params = register_param2(p, STATE_INTERNAL,
+					   STATE_FOG_PARAMS_OPTIMIZED);
+      struct ureg tmp = get_temp(p);
+      GLboolean useabs = (p->state->fog_mode != FOG_EXP2);
+
+      if (useabs) {
+	 emit_op1(p, OPCODE_ABS, tmp, 0, input);
+      }
+
+      switch (p->state->fog_mode) {
+      case FOG_LINEAR: {
+	 struct ureg id = get_identity_param(p);
+	 emit_op3(p, OPCODE_MAD, tmp, 0, useabs ? tmp : input,
+			swizzle1(params,X), swizzle1(params,Y));
+	 emit_op2(p, OPCODE_MAX, tmp, 0, tmp, swizzle1(id,X)); /* saturate */
+	 emit_op2(p, OPCODE_MIN, fog, WRITEMASK_X, tmp, swizzle1(id,W));
+	 break;
+      }
+      case FOG_EXP:
+	 emit_op2(p, OPCODE_MUL, tmp, 0, useabs ? tmp : input,
+			swizzle1(params,Z));
+	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, negate(tmp));
+	 break;
+      case FOG_EXP2:
+	 emit_op2(p, OPCODE_MUL, tmp, 0, input, swizzle1(params,W));
+	 emit_op2(p, OPCODE_MUL, tmp, 0, tmp, tmp);
+	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, negate(tmp));
+	 break;
+      }
+
+      release_temp(p, tmp);
+   }
+   else {
+      /* results = incoming fog coords (compute fog per-fragment later) 
+       *
+       * KW:  Is it really necessary to do anything in this case?
+       * BP: Yes, we always need to compute the absolute value, unless
+       * we want to push that down into the fragment program...
+       */
+      GLboolean useabs = GL_TRUE;
+      emit_op1(p, useabs ? OPCODE_ABS : OPCODE_MOV, fog, WRITEMASK_X, input);
+   }
+}
+ 
+static void build_reflect_texgen( struct tnl_program *p,
+				  struct ureg dest,
+				  GLuint writemask )
+{
+   struct ureg normal = get_eye_normal(p);
+   struct ureg eye_hat = get_eye_position_normalized(p);
+   struct ureg tmp = get_temp(p);
+
+   /* n.u */
+   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
+   /* 2n.u */
+   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
+   /* (-2n.u)n + u */
+   emit_op3(p, OPCODE_MAD, dest, writemask, negate(tmp), normal, eye_hat);
+
+   release_temp(p, tmp);
+}
+
+static void build_sphere_texgen( struct tnl_program *p,
+				 struct ureg dest,
+				 GLuint writemask )
+{
+   struct ureg normal = get_eye_normal(p);
+   struct ureg eye_hat = get_eye_position_normalized(p);
+   struct ureg tmp = get_temp(p);
+   struct ureg half = register_scalar_const(p, .5);
+   struct ureg r = get_temp(p);
+   struct ureg inv_m = get_temp(p);
+   struct ureg id = get_identity_param(p);
+
+   /* Could share the above calculations, but it would be
+    * a fairly odd state for someone to set (both sphere and
+    * reflection active for different texture coordinate
+    * components.  Of course - if two texture units enable
+    * reflect and/or sphere, things start to tilt in favour
+    * of seperating this out:
+    */
+
+   /* n.u */
+   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
+   /* 2n.u */
+   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
+   /* (-2n.u)n + u */
+   emit_op3(p, OPCODE_MAD, r, 0, negate(tmp), normal, eye_hat); 
+   /* r + 0,0,1 */
+   emit_op2(p, OPCODE_ADD, tmp, 0, r, swizzle(id,X,Y,W,Z)); 
+   /* rx^2 + ry^2 + (rz+1)^2 */
+   emit_op2(p, OPCODE_DP3, tmp, 0, tmp, tmp); 
+   /* 2/m */
+   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp); 
+   /* 1/m */
+   emit_op2(p, OPCODE_MUL, inv_m, 0, tmp, half); 
+   /* r/m + 1/2 */
+   emit_op3(p, OPCODE_MAD, dest, writemask, r, inv_m, half); 
+	       
+   release_temp(p, tmp);
+   release_temp(p, r);
+   release_temp(p, inv_m);
+}
+
+
+static void build_texture_transform( struct tnl_program *p )
+{
+   GLuint i, j;
+
+   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+
+      if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))
+	 continue;
+							     
+      if (p->state->unit[i].texgen_enabled || 
+	  p->state->unit[i].texmat_enabled) {
+	 
+	 GLuint texmat_enabled = p->state->unit[i].texmat_enabled;
+	 struct ureg out = register_output(p, VERT_RESULT_TEX0 + i);
+	 struct ureg out_texgen = undef;
+
+	 if (p->state->unit[i].texgen_enabled) {
+	    GLuint copy_mask = 0;
+	    GLuint sphere_mask = 0;
+	    GLuint reflect_mask = 0;
+	    GLuint normal_mask = 0;
+	    GLuint modes[4];
+	 
+	    if (texmat_enabled) 
+	       out_texgen = get_temp(p);
+	    else
+	       out_texgen = out;
+
+	    modes[0] = p->state->unit[i].texgen_mode0;
+	    modes[1] = p->state->unit[i].texgen_mode1;
+	    modes[2] = p->state->unit[i].texgen_mode2;
+	    modes[3] = p->state->unit[i].texgen_mode3;
+
+	    for (j = 0; j < 4; j++) {
+	       switch (modes[j]) {
+	       case TXG_OBJ_LINEAR: {
+		  struct ureg obj = register_input(p, VERT_ATTRIB_POS);
+		  struct ureg plane = 
+		     register_param3(p, STATE_TEXGEN, i,
+				     STATE_TEXGEN_OBJECT_S + j);
+
+		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
+			   obj, plane );
+		  break;
+	       }
+	       case TXG_EYE_LINEAR: {
+		  struct ureg eye = get_eye_position(p);
+		  struct ureg plane = 
+		     register_param3(p, STATE_TEXGEN, i, 
+				     STATE_TEXGEN_EYE_S + j);
+
+		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
+			   eye, plane );
+		  break;
+	       }
+	       case TXG_SPHERE_MAP: 
+		  sphere_mask |= WRITEMASK_X << j;
+		  break;
+	       case TXG_REFLECTION_MAP:
+		  reflect_mask |= WRITEMASK_X << j;
+		  break;
+	       case TXG_NORMAL_MAP: 
+		  normal_mask |= WRITEMASK_X << j;
+		  break;
+	       case TXG_NONE:
+		  copy_mask |= WRITEMASK_X << j;
+	       }
+
+	    }
+
+	 
+	    if (sphere_mask) {
+	       build_sphere_texgen(p, out_texgen, sphere_mask);
+	    }
+
+	    if (reflect_mask) {
+	       build_reflect_texgen(p, out_texgen, reflect_mask);
+	    }
+
+	    if (normal_mask) {
+	       struct ureg normal = get_eye_normal(p);
+	       emit_op1(p, OPCODE_MOV, out_texgen, normal_mask, normal );
+	    }
+
+	    if (copy_mask) {
+	       struct ureg in = register_input(p, VERT_ATTRIB_TEX0+i);
+	       emit_op1(p, OPCODE_MOV, out_texgen, copy_mask, in );
+	    }
+	 }
+
+	 if (texmat_enabled) {
+	    struct ureg texmat[4];
+	    struct ureg in = (!is_undef(out_texgen) ? 
+			      out_texgen : 
+			      register_input(p, VERT_ATTRIB_TEX0+i));
+	    if (PREFER_DP4) {
+	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+				       0, texmat );
+	       emit_matrix_transform_vec4( p, out, texmat, in );
+	    }
+	    else {
+	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+				       STATE_MATRIX_TRANSPOSE, texmat );
+	       emit_transpose_matrix_transform_vec4( p, out, texmat, in );
+	    }
+	 }
+
+	 release_temps(p);
+      } 
+      else {
+	 emit_passthrough(p, VERT_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
+      }
+   }
+}
+
+
+static void build_pointsize( struct tnl_program *p )
+{
+   struct ureg eye = get_eye_position(p);
+   struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
+   struct ureg state_attenuation = register_param1(p, STATE_POINT_ATTENUATION);
+   struct ureg out = register_output(p, VERT_RESULT_PSIZ);
+   struct ureg ut = get_temp(p);
+
+   /* dist = |eyez| */
+   emit_op1(p, OPCODE_ABS, ut, WRITEMASK_Y, swizzle1(eye, Z));
+   /* p1 + dist * (p2 + dist * p3); */
+   emit_op3(p, OPCODE_MAD, ut, WRITEMASK_X, swizzle1(ut, Y),
+		swizzle1(state_attenuation, Z), swizzle1(state_attenuation, Y));
+   emit_op3(p, OPCODE_MAD, ut, WRITEMASK_X, swizzle1(ut, Y),
+		ut, swizzle1(state_attenuation, X));
+
+   /* 1 / sqrt(factor) */
+   emit_op1(p, OPCODE_RSQ, ut, WRITEMASK_X, ut );
+
+#if 1
+   /* out = pointSize / sqrt(factor) */
+   emit_op2(p, OPCODE_MUL, out, WRITEMASK_X, ut, state_size);
+#else
+   /* not sure, might make sense to do clamping here,
+      but it's not done in t_vb_points neither */
+   emit_op2(p, OPCODE_MUL, ut, WRITEMASK_X, ut, state_size);
+   emit_op2(p, OPCODE_MAX, ut, WRITEMASK_X, ut, swizzle1(state_size, Y));
+   emit_op2(p, OPCODE_MIN, out, WRITEMASK_X, ut, swizzle1(state_size, Z));
+#endif
+
+   release_temp(p, ut);
+}
+
+/**
+ * Emit constant point size.
+ */
+static void constant_pointsize( struct tnl_program *p )
+{
+   struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
+   struct ureg out = register_output(p, VERT_RESULT_PSIZ);
+   emit_op1(p, OPCODE_MOV, out, WRITEMASK_X, state_size);
+}
+
+static void build_tnl_program( struct tnl_program *p )
+{   /* Emit the program, starting with modelviewproject:
+    */
+   build_hpos(p);
+
+   /* Lighting calculations:
+    */
+   if (p->state->fragprog_inputs_read & (FRAG_BIT_COL0|FRAG_BIT_COL1)) {
+      if (p->state->light_global_enabled)
+	 build_lighting(p);
+      else {
+	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL0)
+	    emit_passthrough(p, VERT_ATTRIB_COLOR0, VERT_RESULT_COL0);
+
+	 if (p->state->fragprog_inputs_read & FRAG_BIT_COL1)
+	    emit_passthrough(p, VERT_ATTRIB_COLOR1, VERT_RESULT_COL1);
+      }
+   }
+
+   if ((p->state->fragprog_inputs_read & FRAG_BIT_FOGC) ||
+       p->state->fog_mode != FOG_NONE)
+      build_fog(p);
+
+   if (p->state->fragprog_inputs_read & FRAG_BITS_TEX_ANY)
+      build_texture_transform(p);
+
+   if (p->state->point_attenuated)
+      build_pointsize(p);
+#if 0
+   else
+      constant_pointsize(p);
+#endif
+
+   /* Finish up:
+    */
+   emit_op1(p, OPCODE_END, undef, 0, undef);
+
+   /* Disassemble:
+    */
+   if (DISASSEM) {
+      _mesa_printf ("\n");
+   }
+}
+
+
+static void
+create_new_program( const struct state_key *key,
+                    struct gl_vertex_program *program,
+                    GLuint max_temps)
+{
+   struct tnl_program p;
+
+   _mesa_memset(&p, 0, sizeof(p));
+   p.state = key;
+   p.program = program;
+   p.eye_position = undef;
+   p.eye_position_normalized = undef;
+   p.eye_normal = undef;
+   p.identity = undef;
+   p.temp_in_use = 0;
+   
+   if (max_temps >= sizeof(int) * 8)
+      p.temp_reserved = 0;
+   else
+      p.temp_reserved = ~((1<<max_temps)-1);
+
+   p.program->Base.Instructions = _mesa_alloc_instructions(MAX_INSN);
+   p.program->Base.String = NULL;
+   p.program->Base.NumInstructions =
+   p.program->Base.NumTemporaries =
+   p.program->Base.NumParameters =
+   p.program->Base.NumAttributes = p.program->Base.NumAddressRegs = 0;
+   p.program->Base.Parameters = _mesa_new_parameter_list();
+   p.program->Base.InputsRead = 0;
+   p.program->Base.OutputsWritten = 0;
+
+   build_tnl_program( &p );
+}
+
+
+/**
+ * Return a vertex program which implements the current fixed-function
+ * transform/lighting/texgen operations.
+ * XXX move this into core mesa (main/)
+ */
+struct gl_vertex_program *
+_mesa_get_fixed_func_vertex_program(GLcontext *ctx)
+{
+   struct gl_vertex_program *prog;
+   struct state_key *key;
+
+   /* Grab all the relevent state and put it in a single structure:
+    */
+   key = make_state_key(ctx);
+
+   /* Look for an already-prepared program for this state:
+    */
+   prog = (struct gl_vertex_program *)
+      _mesa_search_program_cache(ctx->VertexProgram.Cache, key, sizeof(*key));
+   
+   if (!prog) {
+      /* OK, we'll have to build a new one */
+      if (0)
+         _mesa_printf("Build new TNL program\n");
+	 
+      prog = (struct gl_vertex_program *)
+         ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); 
+
+      create_new_program( key, prog,
+                          ctx->Const.VertexProgram.MaxTemps );
+
+#if 0
+      if (ctx->Driver.ProgramStringNotify)
+         ctx->Driver.ProgramStringNotify( ctx, GL_VERTEX_PROGRAM_ARB, 
+                                          &prog->Base );
+#endif
+      _mesa_program_cache_insert(ctx, ctx->VertexProgram.Cache,
+                                 key, sizeof(*key), &prog->Base);
+   }
+   else {
+      /* use cached program */
+      _mesa_free(key);
+   }
+
+   return prog;
+}
-- 
cgit v1.2.3


From 573b4414b926bb86e9a1e8f3ffad64426aa4bda4 Mon Sep 17 00:00:00 2001
From: Brian <brian.paul@tungstengraphics.com>
Date: Wed, 26 Dec 2007 06:56:42 -0700
Subject: fix mem leak (free key)

---
 src/mesa/main/ffvertex_prog.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 8fcb9e5b14..4a9a0cd975 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1,6 +1,6 @@
 /**************************************************************************
  * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
@@ -1529,6 +1529,8 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
 	 
       prog = (struct gl_vertex_program *)
          ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); 
+      if (!prog)
+         return NULL;
 
       create_new_program( key, prog,
                           ctx->Const.VertexProgram.MaxTemps );
@@ -1541,10 +1543,8 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
       _mesa_program_cache_insert(ctx, ctx->VertexProgram.Cache,
                                  key, sizeof(*key), &prog->Base);
    }
-   else {
-      /* use cached program */
-      _mesa_free(key);
-   }
+
+   _mesa_free(key);
 
    return prog;
 }
-- 
cgit v1.2.3


From fdc8636bdc65deb0d95a62a51c8d9bca05bc6bb8 Mon Sep 17 00:00:00 2001
From: Brian <brian.paul@tungstengraphics.com>
Date: Fri, 18 Jan 2008 12:45:27 -0700
Subject: use PROGRAM_CONSTANT instead of PROGRAM_STATE_VAR when generating
 immediates/literals

---
 src/mesa/main/ffvertex_prog.c | 2 +-
 src/mesa/main/texenvprogram.c | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 4a9a0cd975..04ece4cda8 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -436,7 +436,7 @@ static struct ureg register_const4f( struct tnl_program *p,
    idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
                                      &swizzle );
    ASSERT(swizzle == SWIZZLE_NOOP);
-   return make_ureg(PROGRAM_STATE_VAR, idx);
+   return make_ureg(PROGRAM_CONSTANT, idx);
 }
 
 #define register_const1f(p, s0)         register_const4f(p, s0, 0, 0, 1)
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index d866d10017..644b1f39c7 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -582,7 +582,7 @@ static struct ureg register_const4f( struct texenv_fragment_program *p,
    idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
                                      &swizzle );
    ASSERT(swizzle == SWIZZLE_NOOP);
-   return make_ureg(PROGRAM_STATE_VAR, idx);
+   return make_ureg(PROGRAM_CONSTANT, idx);
 }
 
 #define register_scalar_const(p, s0)    register_const4f(p, s0, s0, s0, s0)
-- 
cgit v1.2.3


From c2bf23b8372607a5507a44ea3654c8af1083529a Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Fri, 14 Mar 2008 17:47:11 -0600
Subject: mesa: clamp point size in vertex program when computing attenuated
 size

---
 src/mesa/main/ffvertex_prog.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 04ece4cda8..f648d081c0 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1398,12 +1398,13 @@ static void build_pointsize( struct tnl_program *p )
    /* 1 / sqrt(factor) */
    emit_op1(p, OPCODE_RSQ, ut, WRITEMASK_X, ut );
 
-#if 1
+#if 0
    /* out = pointSize / sqrt(factor) */
    emit_op2(p, OPCODE_MUL, out, WRITEMASK_X, ut, state_size);
 #else
-   /* not sure, might make sense to do clamping here,
-      but it's not done in t_vb_points neither */
+   /* this is a good place to clamp the point size since there's likely
+    * no hardware registers to clamp point size at rasterization time.
+    */
    emit_op2(p, OPCODE_MUL, ut, WRITEMASK_X, ut, state_size);
    emit_op2(p, OPCODE_MAX, ut, WRITEMASK_X, ut, swizzle1(state_size, Y));
    emit_op2(p, OPCODE_MIN, out, WRITEMASK_X, ut, swizzle1(state_size, Z));
-- 
cgit v1.2.3


From 10f6ae0355937615d137c79c060b9e5a923f0d65 Mon Sep 17 00:00:00 2001
From: Brian <brian.paul@tungstengraphics.com>
Date: Thu, 1 May 2008 19:29:25 -0600
Subject: mesa: comments, whitespace

---
 src/mesa/main/ffvertex_prog.c | 43 +++++++++++++++++++------------------------
 1 file changed, 19 insertions(+), 24 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index f648d081c0..810af9e33e 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -946,35 +946,33 @@ static void build_lighting( struct tnl_program *p )
 	 _bfc1 = _bfc0;
    }
 
-
    /* If no lights, still need to emit the scenecolor.
     */
-      {
-	 struct ureg res0 = register_output( p, VERT_RESULT_COL0 );
-	 emit_op1(p, OPCODE_MOV, res0, 0, _col0);
-      }
+   {
+      struct ureg res0 = register_output( p, VERT_RESULT_COL0 );
+      emit_op1(p, OPCODE_MOV, res0, 0, _col0);
+   }
 
-      if (separate) {
-	 struct ureg res1 = register_output( p, VERT_RESULT_COL1 );
-	 emit_op1(p, OPCODE_MOV, res1, 0, _col1);
-      }
+   if (separate) {
+      struct ureg res1 = register_output( p, VERT_RESULT_COL1 );
+      emit_op1(p, OPCODE_MOV, res1, 0, _col1);
+   }
 
-      if (twoside) {
-	 struct ureg res0 = register_output( p, VERT_RESULT_BFC0 );
-	 emit_op1(p, OPCODE_MOV, res0, 0, _bfc0);
-      }
+   if (twoside) {
+      struct ureg res0 = register_output( p, VERT_RESULT_BFC0 );
+      emit_op1(p, OPCODE_MOV, res0, 0, _bfc0);
+   }
       
-      if (twoside && separate) {
-	 struct ureg res1 = register_output( p, VERT_RESULT_BFC1 );
-	 emit_op1(p, OPCODE_MOV, res1, 0, _bfc1);
-      }
+   if (twoside && separate) {
+      struct ureg res1 = register_output( p, VERT_RESULT_BFC1 );
+      emit_op1(p, OPCODE_MOV, res1, 0, _bfc1);
+   }
       
    if (nr_lights == 0) {
       release_temps(p);
       return;
    }
 
-
    for (i = 0; i < MAX_LIGHTS; i++) {
       if (p->state->unit[i].light_enabled) {
 	 struct ureg half = undef;
@@ -1006,7 +1004,7 @@ static void build_lighting( struct tnl_program *p )
 	    VPpli = get_temp(p); 
 	    half = get_temp(p);
  
-	    /* Calulate VPpli vector
+	    /* Calculate VPpli vector
 	     */
 	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); 
 
@@ -1017,15 +1015,13 @@ static void build_lighting( struct tnl_program *p )
 	    emit_op1(p, OPCODE_RSQ, dist, 0, dist);
 	    emit_op2(p, OPCODE_MUL, VPpli, 0, VPpli, dist);
 
-
-	    /* Calculate  attenuation:
+	    /* Calculate attenuation:
 	     */ 
 	    if (!p->state->unit[i].light_spotcutoff_is_180 ||
 		p->state->unit[i].light_attenuated) {
 	       att = calculate_light_attenuation(p, i, VPpli, dist);
 	    }
-	 
-      
+
 	    /* Calculate viewer direction, or use infinite viewer:
 	     */
 	    if (p->state->light_local_viewer) {
@@ -1047,7 +1043,6 @@ static void build_lighting( struct tnl_program *p )
 	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
 	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
 
-	
 	 /* Front face lighting:
 	  */
 	 {
-- 
cgit v1.2.3


From adc1f88fc9278bdbb3b24a6d48f91a0bd98e9f1c Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Fri, 23 May 2008 09:10:59 +0100
Subject: mesa: do object-space lighting in ffvertex_prog.c

Start pulling over some of the optimizations from the fixed function
paths.
---
 src/mesa/main/ffvertex_prog.c    | 79 +++++++++++++++++++++++++---------------
 src/mesa/shader/prog_statevars.c | 42 +++++++++++++++++----
 src/mesa/shader/prog_statevars.h |  6 ++-
 3 files changed, 88 insertions(+), 39 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 810af9e33e..adf15b03c2 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -54,6 +54,7 @@ struct state_key {
    unsigned light_color_material_mask:12;
    unsigned light_material_mask:12;
 
+   unsigned need_eye_coords:1;
    unsigned normalize:1;
    unsigned rescale_normals:1;
    unsigned fog_source_is_depth:1;
@@ -167,6 +168,8 @@ static struct state_key *make_state_key( GLcontext *ctx )
     */
    assert(fp);
 
+   key->need_eye_coords = ctx->_NeedEyeCoords;
+
    key->fragprog_inputs_read = fp->Base.InputsRead;
 
    if (ctx->RenderMode == GL_FEEDBACK) {
@@ -310,7 +313,7 @@ struct tnl_program {
    
    struct ureg eye_position;
    struct ureg eye_position_normalized;
-   struct ureg eye_normal;
+   struct ureg transformed_normal;
    struct ureg identity;
 
    GLuint materials;
@@ -653,9 +656,9 @@ static void emit_normalize_vec3( struct tnl_program *p,
 				 struct ureg src )
 {
    struct ureg tmp = get_temp(p);
-   emit_op2(p, OPCODE_DP3, tmp, 0, src, src);
-   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp);
-   emit_op2(p, OPCODE_MUL, dest, 0, src, tmp);
+   emit_op2(p, OPCODE_DP3, tmp, WRITEMASK_X, src, src);
+   emit_op1(p, OPCODE_RSQ, tmp, WRITEMASK_X, tmp);
+   emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(tmp, X));
    release_temp(p, tmp);
 }
 
@@ -705,36 +708,53 @@ static struct ureg get_eye_position_normalized( struct tnl_program *p )
 }
 
 
-static struct ureg get_eye_normal( struct tnl_program *p )
+static struct ureg get_transformed_normal( struct tnl_program *p )
 {
-   if (is_undef(p->eye_normal)) {
+   if (is_undef(p->transformed_normal) &&
+       !p->state->need_eye_coords &&
+       !p->state->normalize &&
+       !(p->state->need_eye_coords == p->state->rescale_normals))
+   {
+      p->transformed_normal = register_input(p, VERT_ATTRIB_NORMAL );
+   }
+   else if (is_undef(p->transformed_normal)) 
+   {
       struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
       struct ureg mvinv[3];
+      struct ureg transformed_normal = reserve_temp(p);
 
-      register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
-			      STATE_MATRIX_INVTRANS, mvinv );
+      if (p->state->need_eye_coords) {
+         register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
+                                 STATE_MATRIX_INVTRANS, mvinv );
 
-      p->eye_normal = reserve_temp(p);
-
-      /* Transform to eye space:
-       */
-      emit_matrix_transform_vec3( p, p->eye_normal, mvinv, normal );
+         /* Transform to eye space:
+          */
+         emit_matrix_transform_vec3( p, transformed_normal, mvinv, normal );
+         normal = transformed_normal;
+      }
 
       /* Normalize/Rescale:
        */
       if (p->state->normalize) {
-	 emit_normalize_vec3( p, p->eye_normal, p->eye_normal );
+	 emit_normalize_vec3( p, transformed_normal, normal );
+         normal = transformed_normal;
       }
-      else if (p->state->rescale_normals) {
+      else if (p->state->need_eye_coords == p->state->rescale_normals) {
+         /* This is already adjusted for eye/non-eye rendering:
+          */
 	 struct ureg rescale = register_param2(p, STATE_INTERNAL,
-					       STATE_NORMAL_SCALE);
+                                               STATE_NORMAL_SCALE);
 
-	 emit_op2( p, OPCODE_MUL, p->eye_normal, 0, p->eye_normal,
+	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal,
 		   swizzle1(rescale, X));
+         normal = transformed_normal;
       }
+      
+      assert(normal.file == PROGRAM_TEMPORARY);
+      p->transformed_normal = normal;
    }
 
-   return p->eye_normal;
+   return p->transformed_normal;
 }
 
 
@@ -856,7 +876,7 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
     */
    if (!p->state->unit[i].light_spotcutoff_is_180) {
       struct ureg spot_dir_norm = register_param3(p, STATE_INTERNAL,
-						  STATE_SPOT_DIR_NORMALIZED, i);
+						  STATE_LIGHT_SPOT_DIR_NORMALIZED, i);
       struct ureg spot = get_temp(p);
       struct ureg slt = get_temp(p);
 
@@ -907,7 +927,7 @@ static void build_lighting( struct tnl_program *p )
    const GLboolean twoside = p->state->light_twoside;
    const GLboolean separate = p->state->separate_specular;
    GLuint nr_lights = 0, count = 0;
-   struct ureg normal = get_eye_normal(p);
+   struct ureg normal = get_transformed_normal(p);
    struct ureg lit = get_temp(p);
    struct ureg dots = get_temp(p);
    struct ureg _col0 = undef, _col1 = undef;
@@ -984,20 +1004,21 @@ static void build_lighting( struct tnl_program *p )
 	    /* Can used precomputed constants in this case.
 	     * Attenuation never applies to infinite lights.
 	     */
-	    VPpli = register_param3(p, STATE_LIGHT, i, 
-				    STATE_POSITION_NORMALIZED); 
+	    VPpli = register_param3(p, STATE_INTERNAL, 
+				    STATE_LIGHT_POSITION_NORMALIZED, i); 
             if (p->state->light_local_viewer) {
                 struct ureg eye_hat = get_eye_position_normalized(p);
                 half = get_temp(p);
                 emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
                 emit_normalize_vec3(p, half, half);
             } else {
-                half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+                half = register_param3(p, STATE_INTERNAL, 
+                                       STATE_LIGHT_HALF_VECTOR, i);
             }
 	 } 
 	 else {
-	    struct ureg Ppli = register_param3(p, STATE_LIGHT, i, 
-					       STATE_POSITION); 
+	    struct ureg Ppli = register_param3(p, STATE_INTERNAL, 
+					       STATE_LIGHT_POSITION, i); 
 	    struct ureg V = get_eye_position(p);
 	    struct ureg dist = get_temp(p);
 
@@ -1201,7 +1222,7 @@ static void build_reflect_texgen( struct tnl_program *p,
 				  struct ureg dest,
 				  GLuint writemask )
 {
-   struct ureg normal = get_eye_normal(p);
+   struct ureg normal = get_transformed_normal(p);
    struct ureg eye_hat = get_eye_position_normalized(p);
    struct ureg tmp = get_temp(p);
 
@@ -1219,7 +1240,7 @@ static void build_sphere_texgen( struct tnl_program *p,
 				 struct ureg dest,
 				 GLuint writemask )
 {
-   struct ureg normal = get_eye_normal(p);
+   struct ureg normal = get_transformed_normal(p);
    struct ureg eye_hat = get_eye_position_normalized(p);
    struct ureg tmp = get_temp(p);
    struct ureg half = register_scalar_const(p, .5);
@@ -1338,7 +1359,7 @@ static void build_texture_transform( struct tnl_program *p )
 	    }
 
 	    if (normal_mask) {
-	       struct ureg normal = get_eye_normal(p);
+	       struct ureg normal = get_transformed_normal(p);
 	       emit_op1(p, OPCODE_MOV, out_texgen, normal_mask, normal );
 	    }
 
@@ -1475,7 +1496,7 @@ create_new_program( const struct state_key *key,
    p.program = program;
    p.eye_position = undef;
    p.eye_position_normalized = undef;
-   p.eye_normal = undef;
+   p.transformed_normal = undef;
    p.identity = undef;
    p.temp_in_use = 0;
    
diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c
index ba3c988445..37bd17ba4a 100644
--- a/src/mesa/shader/prog_statevars.c
+++ b/src/mesa/shader/prog_statevars.c
@@ -134,10 +134,6 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
 	       value[3] = 1.0;
             }						  
             return;
-	 case STATE_POSITION_NORMALIZED:
-            COPY_4V(value, ctx->Light.Light[ln].EyePosition);
-	    NORMALIZE_3FV( value );
-            return;
          default:
             _mesa_problem(ctx, "Invalid light state in fetch_state");
             return;
@@ -431,15 +427,46 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
          value[2] = ctx->Fog.Density * ONE_DIV_LN2;
          value[3] = ctx->Fog.Density * ONE_DIV_SQRT_LN2;
          return;
-      case STATE_SPOT_DIR_NORMALIZED: {
+
+      case STATE_LIGHT_SPOT_DIR_NORMALIZED: {
          /* here, state[2] is the light number */
          /* pre-normalize spot dir */
          const GLuint ln = (GLuint) state[2];
-         COPY_3V(value, ctx->Light.Light[ln].EyeDirection);
-         NORMALIZE_3FV(value);
+         COPY_3V(value, ctx->Light.Light[ln]._NormDirection);
          value[3] = ctx->Light.Light[ln]._CosCutoff;
          return;
       }
+
+      case STATE_LIGHT_POSITION: {
+         const GLuint ln = (GLuint) state[2];
+         COPY_4V(value, ctx->Light.Light[ln]._Position);
+         return;
+      }
+
+      case STATE_LIGHT_POSITION_NORMALIZED: {
+         const GLuint ln = (GLuint) state[2];
+         COPY_4V(value, ctx->Light.Light[ln]._Position);
+         NORMALIZE_3FV( value );
+         return;
+      }
+
+      case STATE_LIGHT_HALF_VECTOR: {
+         const GLuint ln = (GLuint) state[2];
+         GLfloat p[3];
+         /* Compute infinite half angle vector:
+          *   halfVector = normalize(normalize(lightPos) + (0, 0, 1))
+          * light.EyePosition.w should be 0 for infinite lights.
+          */
+         COPY_3V(p, ctx->Light.Light[ln]._Position);
+         NORMALIZE_3FV(p);
+         ADD_3V(value, p, ctx->_EyeZDir);
+         NORMALIZE_3FV(value);
+         value[3] = 1.0;
+         return;
+      }						  
+
+
+
       case STATE_PT_SCALE:
          value[0] = ctx->Pixel.RedScale;
          value[1] = ctx->Pixel.GreenScale;
@@ -696,7 +723,6 @@ append_token(char *dst, gl_state_index k)
       append(dst, "normalScale");
       break;
    case STATE_INTERNAL:
-   case STATE_POSITION_NORMALIZED:
       append(dst, "(internal)");
       break;
    case STATE_PT_SCALE:
diff --git a/src/mesa/shader/prog_statevars.h b/src/mesa/shader/prog_statevars.h
index d12142055f..a515fda3aa 100644
--- a/src/mesa/shader/prog_statevars.h
+++ b/src/mesa/shader/prog_statevars.h
@@ -106,9 +106,11 @@ typedef enum gl_state_index_ {
    STATE_INTERNAL,		/* Mesa additions */
    STATE_NORMAL_SCALE,
    STATE_TEXRECT_SCALE,
-   STATE_POSITION_NORMALIZED,   /* normalized light position */
    STATE_FOG_PARAMS_OPTIMIZED,  /* for faster fog calc */
-   STATE_SPOT_DIR_NORMALIZED,   /* pre-normalized spot dir */
+   STATE_LIGHT_SPOT_DIR_NORMALIZED,   /* pre-normalized spot dir */
+   STATE_LIGHT_POSITION,              /* object vs eye space */
+   STATE_LIGHT_POSITION_NORMALIZED,   /* object vs eye space */
+   STATE_LIGHT_HALF_VECTOR,           /* object vs eye space */
    STATE_PT_SCALE,              /**< Pixel transfer RGBA scale */
    STATE_PT_BIAS,               /**< Pixel transfer RGBA bias */
    STATE_PCM_SCALE,             /**< Post color matrix RGBA scale */
-- 
cgit v1.2.3


From 0ac2f7955c01749e122f67ff03e79a0d8bd0f8e5 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Fri, 23 May 2008 19:17:02 +0100
Subject: mesa: don't emit LIT instruction when mat shininess known to be zero

Use a faster path in that case & make gears go faster.
---
 src/mesa/main/ffvertex_prog.c | 133 ++++++++++++++++++++++++++++++++----------
 1 file changed, 102 insertions(+), 31 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index adf15b03c2..623c2a64b5 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -53,6 +53,7 @@ struct state_key {
    unsigned light_color_material:1;
    unsigned light_color_material_mask:12;
    unsigned light_material_mask:12;
+   unsigned material_shininess_is_zero:1;
 
    unsigned need_eye_coords:1;
    unsigned normalize:1;
@@ -155,6 +156,26 @@ tnl_get_per_vertex_fog(GLcontext *ctx)
 #endif
 }
 
+static GLboolean check_active_shininess( GLcontext *ctx,
+                                         const struct state_key *key,
+                                         GLuint side )
+{
+   GLuint bit = 1 << (MAT_ATTRIB_FRONT_SHININESS + side);
+
+   if (key->light_color_material_mask & bit)
+      return GL_TRUE;
+
+   if (key->light_material_mask & bit)
+      return GL_TRUE;
+
+   if (ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS + side][0] != 0.0F)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+     
+
+
 
 static struct state_key *make_state_key( GLcontext *ctx )
 {
@@ -214,6 +235,17 @@ static struct state_key *make_state_key( GLcontext *ctx )
 	       key->unit[i].light_attenuated = 1;
 	 }
       }
+
+      if (check_active_shininess(ctx, key, 0)) {
+         key->material_shininess_is_zero = 0;
+      }
+      else if (key->light_twoside &&
+               check_active_shininess(ctx, key, 1)) {
+         key->material_shininess_is_zero = 0;
+      }
+      else {
+         key->material_shininess_is_zero = 1;
+      }
    }
 
    if (ctx->Transform.Normalize)
@@ -915,7 +947,26 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
 }
 						
 
+static void emit_degenerate_lit( struct tnl_program *p,
+                                 struct ureg lit,
+                                 struct ureg dots )
+{
+   struct ureg id = get_identity_param(p);
+   
+   /* 1, 0, 0, 1 
+    */
+   emit_op1(p, OPCODE_MOV, lit, 0, swizzle(id, Z, X, X, Z)); 
 
+   /* 1, MAX2(in[0], 0), 0, 1
+    */
+   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_Y, lit, swizzle1(dots, X)); 
+
+   /* 1, MAX2(in[0], 0), (in[0] > 0 ? 1 : 0), 1
+    */
+   emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, 
+            lit,                /* 0 */
+            swizzle1(dots, X)); /* in[0] */
+}
 
 
 /* Need to add some addtional parameters to allow lighting in object
@@ -941,9 +992,11 @@ static void build_lighting( struct tnl_program *p )
    set_material_flags(p);
 
    {
-      struct ureg shininess = get_material(p, 0, STATE_SHININESS);
-      emit_op1(p, OPCODE_MOV, dots,  WRITEMASK_W, swizzle1(shininess,X));
-      release_temp(p, shininess);
+      if (!p->state->material_shininess_is_zero) {
+         struct ureg shininess = get_material(p, 0, STATE_SHININESS);
+         emit_op1(p, OPCODE_MOV, dots,  WRITEMASK_W, swizzle1(shininess,X));
+         release_temp(p, shininess);
+      }
 
       _col0 = make_temp(p, get_scenecolor(p, 0));
       if (separate)
@@ -954,10 +1007,12 @@ static void build_lighting( struct tnl_program *p )
    }
 
    if (twoside) {
-      struct ureg shininess = get_material(p, 1, STATE_SHININESS);
-      emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
-	       negate(swizzle1(shininess,X)));
-      release_temp(p, shininess);
+      if (!p->state->material_shininess_is_zero) {
+         struct ureg shininess = get_material(p, 1, STATE_SHININESS);
+         emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
+                  negate(swizzle1(shininess,X)));
+         release_temp(p, shininess);
+      }
 
       _bfc0 = make_temp(p, get_scenecolor(p, 1));
       if (separate)
@@ -1006,14 +1061,17 @@ static void build_lighting( struct tnl_program *p )
 	     */
 	    VPpli = register_param3(p, STATE_INTERNAL, 
 				    STATE_LIGHT_POSITION_NORMALIZED, i); 
-            if (p->state->light_local_viewer) {
-                struct ureg eye_hat = get_eye_position_normalized(p);
-                half = get_temp(p);
-                emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
-                emit_normalize_vec3(p, half, half);
-            } else {
-                half = register_param3(p, STATE_INTERNAL, 
-                                       STATE_LIGHT_HALF_VECTOR, i);
+            
+            if (!p->state->material_shininess_is_zero) {
+               if (p->state->light_local_viewer) {
+                  struct ureg eye_hat = get_eye_position_normalized(p);
+                  half = get_temp(p);
+                  emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+                  emit_normalize_vec3(p, half, half);
+               } else {
+                  half = register_param3(p, STATE_INTERNAL, 
+                                         STATE_LIGHT_HALF_VECTOR, i);
+               }
             }
 	 } 
 	 else {
@@ -1023,7 +1081,6 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg dist = get_temp(p);
 
 	    VPpli = get_temp(p); 
-	    half = get_temp(p);
  
 	    /* Calculate VPpli vector
 	     */
@@ -1045,16 +1102,20 @@ static void build_lighting( struct tnl_program *p )
 
 	    /* Calculate viewer direction, or use infinite viewer:
 	     */
-	    if (p->state->light_local_viewer) {
-	       struct ureg eye_hat = get_eye_position_normalized(p);
-	       emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
-	    }
-	    else {
-	       struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z); 
-	       emit_op2(p, OPCODE_ADD, half, 0, VPpli, z_dir);
-	    }
-
-	    emit_normalize_vec3(p, half, half);
+            if (!p->state->material_shininess_is_zero) {
+               half = get_temp(p);
+
+               if (p->state->light_local_viewer) {
+                  struct ureg eye_hat = get_eye_position_normalized(p);
+                  emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+               }
+               else {
+                  struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z); 
+                  emit_op2(p, OPCODE_ADD, half, 0, VPpli, z_dir);
+               }
+
+               emit_normalize_vec3(p, half, half);
+            }
 
 	    release_temp(p, dist);
 	 }
@@ -1062,7 +1123,9 @@ static void build_lighting( struct tnl_program *p )
 	 /* Calculate dot products:
 	  */
 	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
-	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
+
+         if (!p->state->material_shininess_is_zero)
+            emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
 
 	 /* Front face lighting:
 	  */
@@ -1073,7 +1136,11 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
 
-	    emit_op1(p, OPCODE_LIT, lit, 0, dots);
+            if (p->state->material_shininess_is_zero) {
+               emit_degenerate_lit(p, lit, dots);
+            } else {
+               emit_op1(p, OPCODE_LIT, lit, 0, dots);
+            }
    
 	    if (!is_undef(att)) 
 	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
@@ -1099,7 +1166,7 @@ static void build_lighting( struct tnl_program *p )
 	       res1 = _col1;
 	    }
 
-	    emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
+	    emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _col0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _col1);
       
@@ -1117,7 +1184,11 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
 	       
-	    emit_op1(p, OPCODE_LIT, lit, 0, negate(swizzle(dots,X,Y,W,Z)));
+            if (p->state->material_shininess_is_zero) {
+               emit_degenerate_lit(p, lit, negate(swizzle(dots,X,Y,W,Z)));
+            } else {
+               emit_op1(p, OPCODE_LIT, lit, 0, negate(swizzle(dots,X,Y,W,Z)));
+            }
 
 	    if (!is_undef(att)) 
 	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
@@ -1142,7 +1213,7 @@ static void build_lighting( struct tnl_program *p )
 	       mask1 = 0;
 	    }
 
-	    emit_op3(p, OPCODE_MAD, _bfc0, 0, swizzle1(lit,X), ambient, _bfc0);
+	    emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
 
-- 
cgit v1.2.3


From e841b92d9c8bf48085b4996df828ae745977f931 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Fri, 23 May 2008 20:05:36 +0100
Subject: mesa: further degenerate the special case lit substitute

---
 src/mesa/main/ffvertex_prog.c | 21 ++++++++++++---------
 1 file changed, 12 insertions(+), 9 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 623c2a64b5..90b156f812 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -953,19 +953,19 @@ static void emit_degenerate_lit( struct tnl_program *p,
 {
    struct ureg id = get_identity_param(p);
    
-   /* 1, 0, 0, 1 
+   /* Note that result.x & result.w will not be examined.  Note also that
+    * dots.xyzw == dots.xxxx.
     */
-   emit_op1(p, OPCODE_MOV, lit, 0, swizzle(id, Z, X, X, Z)); 
 
-   /* 1, MAX2(in[0], 0), 0, 1
+   /* result[1] = MAX2(in, 0)
     */
-   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_Y, lit, swizzle1(dots, X)); 
+   emit_op2(p, OPCODE_MAX, lit, 0, id, dots); 
 
-   /* 1, MAX2(in[0], 0), (in[0] > 0 ? 1 : 0), 1
+   /* result[2] = (in > 0 ? 1 : 0)
     */
    emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, 
             lit,                /* 0 */
-            swizzle1(dots, X)); /* in[0] */
+            dots); /* in[0] */
 }
 
 
@@ -1122,10 +1122,13 @@ static void build_lighting( struct tnl_program *p )
 
 	 /* Calculate dot products:
 	  */
-	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
-
-         if (!p->state->material_shininess_is_zero)
+         if (p->state->material_shininess_is_zero) {
+            emit_op2(p, OPCODE_DP3, dots, 0, normal, VPpli);
+         }
+         else {
+            emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
             emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
+         }
 
 	 /* Front face lighting:
 	  */
-- 
cgit v1.2.3


From feceb43948f76cc4d4c8ecbb86b1b1f438c6daee Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Fri, 23 May 2008 20:37:50 +0100
Subject: mesa: save a temp on normalizes

---
 src/mesa/main/ffvertex_prog.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 90b156f812..e36f1f69a4 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -305,7 +305,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
  * generated program with line/function references for each
  * instruction back into this file:
  */
-#define DISASSEM (MESA_VERBOSE&VERBOSE_DISASSEM)
+#define DISASSEM 1
 
 /* Should be tunable by the driver - do we want to do matrix
  * multiplications with DP4's or with MUL/MAD's?  SSE works better
@@ -687,11 +687,9 @@ static void emit_normalize_vec3( struct tnl_program *p,
 				 struct ureg dest,
 				 struct ureg src )
 {
-   struct ureg tmp = get_temp(p);
-   emit_op2(p, OPCODE_DP3, tmp, WRITEMASK_X, src, src);
-   emit_op1(p, OPCODE_RSQ, tmp, WRITEMASK_X, tmp);
-   emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(tmp, X));
-   release_temp(p, tmp);
+   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_X, src, src);
+   emit_op1(p, OPCODE_RSQ, dest, WRITEMASK_X, dest);
+   emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(dest, X));
 }
 
 static void emit_passthrough( struct tnl_program *p, 
-- 
cgit v1.2.3


From e1590abb17f1effd92c136207f363de6cf52df18 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Sat, 24 May 2008 13:23:06 +0100
Subject: mesa: pre-swizzle normal scale state value

---
 src/mesa/main/ffvertex_prog.c    | 3 +--
 src/mesa/shader/prog_statevars.c | 6 +++++-
 2 files changed, 6 insertions(+), 3 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index e36f1f69a4..7a099b2376 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -775,8 +775,7 @@ static struct ureg get_transformed_normal( struct tnl_program *p )
 	 struct ureg rescale = register_param2(p, STATE_INTERNAL,
                                                STATE_NORMAL_SCALE);
 
-	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal,
-		   swizzle1(rescale, X));
+	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal, rescale );
          normal = transformed_normal;
       }
       
diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c
index 37bd17ba4a..44fbfdcd04 100644
--- a/src/mesa/shader/prog_statevars.c
+++ b/src/mesa/shader/prog_statevars.c
@@ -397,7 +397,11 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
    case STATE_INTERNAL:
       switch (state[1]) {
       case STATE_NORMAL_SCALE:
-         ASSIGN_4V(value, ctx->_ModelViewInvScale, 0, 0, 1);
+         ASSIGN_4V(value, 
+                   ctx->_ModelViewInvScale, 
+                   ctx->_ModelViewInvScale, 
+                   ctx->_ModelViewInvScale, 
+                   1);
          return;
       case STATE_TEXRECT_SCALE:
          {
-- 
cgit v1.2.3


From 48a24f0ff7e3aad000b8acc55c16bbeaca58abe6 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Sat, 24 May 2008 16:32:08 +0100
Subject: Revert "mesa: save a temp on normalizes"

This reverts commit feceb43948f76cc4d4c8ecbb86b1b1f438c6daee.
---
 src/mesa/main/ffvertex_prog.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 7a099b2376..a627a21f65 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -305,7 +305,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
  * generated program with line/function references for each
  * instruction back into this file:
  */
-#define DISASSEM 1
+#define DISASSEM (MESA_VERBOSE&VERBOSE_DISASSEM)
 
 /* Should be tunable by the driver - do we want to do matrix
  * multiplications with DP4's or with MUL/MAD's?  SSE works better
@@ -687,9 +687,11 @@ static void emit_normalize_vec3( struct tnl_program *p,
 				 struct ureg dest,
 				 struct ureg src )
 {
-   emit_op2(p, OPCODE_DP3, dest, WRITEMASK_X, src, src);
-   emit_op1(p, OPCODE_RSQ, dest, WRITEMASK_X, dest);
-   emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(dest, X));
+   struct ureg tmp = get_temp(p);
+   emit_op2(p, OPCODE_DP3, tmp, WRITEMASK_X, src, src);
+   emit_op1(p, OPCODE_RSQ, tmp, WRITEMASK_X, tmp);
+   emit_op2(p, OPCODE_MUL, dest, 0, src, swizzle1(tmp, X));
+   release_temp(p, tmp);
 }
 
 static void emit_passthrough( struct tnl_program *p, 
-- 
cgit v1.2.3


From dc1537bc25c7cbff0a41034ece0830146616f036 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Tue, 27 May 2008 09:48:32 +0100
Subject: ffvertex: don't compute whole eye vector if only eye.z is required

---
 src/mesa/main/ffvertex_prog.c | 30 +++++++++++++++++++++++++++---
 1 file changed, 27 insertions(+), 3 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index a627a21f65..a790d4b142 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -305,7 +305,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
  * generated program with line/function references for each
  * instruction back into this file:
  */
-#define DISASSEM (MESA_VERBOSE&VERBOSE_DISASSEM)
+#define DISASSEM 1
 
 /* Should be tunable by the driver - do we want to do matrix
  * multiplications with DP4's or with MUL/MAD's?  SSE works better
@@ -344,6 +344,7 @@ struct tnl_program {
    GLuint temp_reserved;
    
    struct ureg eye_position;
+   struct ureg eye_position_z;
    struct ureg eye_position_normalized;
    struct ureg transformed_normal;
    struct ureg identity;
@@ -728,6 +729,28 @@ static struct ureg get_eye_position( struct tnl_program *p )
 }
 
 
+static struct ureg get_eye_position_z( struct tnl_program *p )
+{
+   if (!is_undef(p->eye_position)) 
+      return swizzle1(p->eye_position, Z);
+
+   if (is_undef(p->eye_position_z)) {
+      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+      struct ureg modelview[4];
+
+      p->eye_position_z = reserve_temp(p);
+
+      register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
+                              0, modelview );
+
+      emit_op2(p, OPCODE_DP4, p->eye_position_z, 0, pos, modelview[2]);
+   }
+   
+   return p->eye_position_z;
+}
+   
+
+
 static struct ureg get_eye_position_normalized( struct tnl_program *p )
 {
    if (is_undef(p->eye_position_normalized)) {
@@ -1240,7 +1263,7 @@ static void build_fog( struct tnl_program *p )
    struct ureg input;
 
    if (p->state->fog_source_is_depth) {
-      input = swizzle1(get_eye_position(p), Z);
+      input = get_eye_position_z(p);
    }
    else {
       input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
@@ -1470,7 +1493,7 @@ static void build_texture_transform( struct tnl_program *p )
 
 static void build_pointsize( struct tnl_program *p )
 {
-   struct ureg eye = get_eye_position(p);
+   struct ureg eye = get_eye_position_z(p);
    struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
    struct ureg state_attenuation = register_param1(p, STATE_POINT_ATTENUATION);
    struct ureg out = register_output(p, VERT_RESULT_PSIZ);
@@ -1568,6 +1591,7 @@ create_new_program( const struct state_key *key,
    p.state = key;
    p.program = program;
    p.eye_position = undef;
+   p.eye_position_z = undef;
    p.eye_position_normalized = undef;
    p.transformed_normal = undef;
    p.identity = undef;
-- 
cgit v1.2.3


From 2109ddafefde26dd20a1c6a25f594984143944a3 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Tue, 27 May 2008 10:35:33 +0100
Subject: ffvertex: emit full LIT when attenuating (needs the 1 in X position)

---
 src/mesa/main/ffvertex_prog.c | 50 +++++++++++++++++++++++++++----------------
 src/mesa/main/light.c         |  1 +
 2 files changed, 32 insertions(+), 19 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index a790d4b142..2ef3286e57 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1161,15 +1161,6 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
 
-            if (p->state->material_shininess_is_zero) {
-               emit_degenerate_lit(p, lit, dots);
-            } else {
-               emit_op1(p, OPCODE_LIT, lit, 0, dots);
-            }
-   
-	    if (!is_undef(att)) 
-	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
-   
 	    
 	    if (count == nr_lights) {
 	       if (separate) {
@@ -1191,7 +1182,21 @@ static void build_lighting( struct tnl_program *p )
 	       res1 = _col1;
 	    }
 
-	    emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
+
+	    if (!is_undef(att)) {
+               emit_op1(p, OPCODE_LIT, lit, 0, dots);
+               emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
+               emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
+            } 
+            else if (!p->state->material_shininess_is_zero) {
+               emit_op1(p, OPCODE_LIT, lit, 0, dots);
+               emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
+            } 
+            else {
+               emit_degenerate_lit(p, lit, dots);
+               emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
+            }
+
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _col0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _col1);
       
@@ -1209,15 +1214,6 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
 	       
-            if (p->state->material_shininess_is_zero) {
-               emit_degenerate_lit(p, lit, negate(swizzle(dots,X,Y,W,Z)));
-            } else {
-               emit_op1(p, OPCODE_LIT, lit, 0, negate(swizzle(dots,X,Y,W,Z)));
-            }
-
-	    if (!is_undef(att)) 
-	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
-
 	    if (count == nr_lights) {
 	       if (separate) {
 		  mask0 = WRITEMASK_XYZ;
@@ -1238,6 +1234,22 @@ static void build_lighting( struct tnl_program *p )
 	       mask1 = 0;
 	    }
 
+            dots = negate(swizzle(dots,X,Y,W,Z));
+
+	    if (!is_undef(att)) {
+               emit_op1(p, OPCODE_LIT, lit, 0, dots);
+	       emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
+               emit_op3(p, OPCODE_MAD, _bfc0, 0, swizzle1(lit,X), ambient, _bfc0);
+            }
+            else if (!p->state->material_shininess_is_zero) {
+               emit_op1(p, OPCODE_LIT, lit, 0, dots);
+               emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
+            } 
+            else {
+               emit_degenerate_lit(p, lit, dots);
+               emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
+            }
+
 	    emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
diff --git a/src/mesa/main/light.c b/src/mesa/main/light.c
index 6e057614ba..0af647e13c 100644
--- a/src/mesa/main/light.c
+++ b/src/mesa/main/light.c
@@ -1357,6 +1357,7 @@ _mesa_init_lighting( GLcontext *ctx )
    /* Miscellaneous */
    ctx->Light._NeedEyeCoords = GL_FALSE;
    ctx->_NeedEyeCoords = GL_FALSE;
+   ctx->_ForceEyeCoords = GL_TRUE;
    ctx->_ModelViewInvScale = 1.0;
 }
 
-- 
cgit v1.2.3


From e4fda51404a6a05c4047a639de4ccc3ea9678c2c Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Fri, 6 Jun 2008 15:57:37 +0100
Subject: mesa: turn off ffvertex prog debug

---
 src/mesa/main/ffvertex_prog.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 2ef3286e57..06710f405d 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -305,7 +305,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
  * generated program with line/function references for each
  * instruction back into this file:
  */
-#define DISASSEM 1
+#define DISASSEM 0
 
 /* Should be tunable by the driver - do we want to do matrix
  * multiplications with DP4's or with MUL/MAD's?  SSE works better
-- 
cgit v1.2.3


From 34ff12ca1fe7153671eea2fe084f3991094ec3ce Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Wed, 11 Jun 2008 20:50:26 -0600
Subject: Revert "mesa: further degenerate the special case lit substitute"

This reverts commit e841b92d9c8bf48085b4996df828ae745977f931.

This fixes two specular lighting conform failures.
---
 src/mesa/main/ffvertex_prog.c | 21 +++++++++------------
 1 file changed, 9 insertions(+), 12 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 06710f405d..d71e0c00fd 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -975,19 +975,19 @@ static void emit_degenerate_lit( struct tnl_program *p,
 {
    struct ureg id = get_identity_param(p);
    
-   /* Note that result.x & result.w will not be examined.  Note also that
-    * dots.xyzw == dots.xxxx.
+   /* 1, 0, 0, 1 
     */
+   emit_op1(p, OPCODE_MOV, lit, 0, swizzle(id, Z, X, X, Z)); 
 
-   /* result[1] = MAX2(in, 0)
+   /* 1, MAX2(in[0], 0), 0, 1
     */
-   emit_op2(p, OPCODE_MAX, lit, 0, id, dots); 
+   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_Y, lit, swizzle1(dots, X)); 
 
-   /* result[2] = (in > 0 ? 1 : 0)
+   /* 1, MAX2(in[0], 0), (in[0] > 0 ? 1 : 0), 1
     */
    emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, 
             lit,                /* 0 */
-            dots); /* in[0] */
+            swizzle1(dots, X)); /* in[0] */
 }
 
 
@@ -1144,13 +1144,10 @@ static void build_lighting( struct tnl_program *p )
 
 	 /* Calculate dot products:
 	  */
-         if (p->state->material_shininess_is_zero) {
-            emit_op2(p, OPCODE_DP3, dots, 0, normal, VPpli);
-         }
-         else {
-            emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
+	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
+
+         if (!p->state->material_shininess_is_zero)
             emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
-         }
 
 	 /* Front face lighting:
 	  */
-- 
cgit v1.2.3


From 5ecb2f2d0fca0c5ea847d1968459aa0dd8138f14 Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Thu, 12 Jun 2008 11:17:20 -0600
Subject: mesa: restore and fix Keith's "further degenerate the special case
 lit substitute"

There was a bug in emit_degenerate_lit() that caused the SLT to produce
unpredictable results in lit.z

Plus, added a bunch of new comments.
---
 src/mesa/main/ffvertex_prog.c | 44 ++++++++++++++++++++++++++++++-------------
 1 file changed, 31 insertions(+), 13 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index d71e0c00fd..e6c7c1040f 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -969,25 +969,29 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
 }
 						
 
+/**
+ * Compute:
+ *   lit.y = MAX(0, dots.x)
+ *   lit.z = SLT(0, dots.x)
+ */
 static void emit_degenerate_lit( struct tnl_program *p,
                                  struct ureg lit,
                                  struct ureg dots )
 {
-   struct ureg id = get_identity_param(p);
-   
-   /* 1, 0, 0, 1 
+   struct ureg id = get_identity_param(p);  /* id = {0,0,0,1} */
+
+   /* Note that lit.x & lit.w will not be examined.  Note also that
+    * dots.xyzw == dots.xxxx.
     */
-   emit_op1(p, OPCODE_MOV, lit, 0, swizzle(id, Z, X, X, Z)); 
 
-   /* 1, MAX2(in[0], 0), 0, 1
+   /* MAX lit, id, dots;
     */
-   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_Y, lit, swizzle1(dots, X)); 
+   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_XYZW, id, dots); 
 
-   /* 1, MAX2(in[0], 0), (in[0] > 0 ? 1 : 0), 1
+   /* result[2] = (in > 0 ? 1 : 0)
+    * SLT lit.z, id.z, dots;   # lit.z = (0 < dots.z) ? 1 : 0
     */
-   emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, 
-            lit,                /* 0 */
-            swizzle1(dots, X)); /* in[0] */
+   emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, swizzle1(id,Z), dots);
 }
 
 
@@ -1007,6 +1011,14 @@ static void build_lighting( struct tnl_program *p )
    struct ureg _bfc0 = undef, _bfc1 = undef;
    GLuint i;
 
+   /*
+    * NOTE:
+    * dot.x = dot(normal, VPpli)
+    * dot.y = dot(normal, halfAngle)
+    * dot.z = back.shininess
+    * dot.w = front.shininess
+    */
+
    for (i = 0; i < MAX_LIGHTS; i++) 
       if (p->state->unit[i].light_enabled)
 	 nr_lights++;
@@ -1144,10 +1156,13 @@ static void build_lighting( struct tnl_program *p )
 
 	 /* Calculate dot products:
 	  */
-	 emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
-
-         if (!p->state->material_shininess_is_zero)
+         if (p->state->material_shininess_is_zero) {
+            emit_op2(p, OPCODE_DP3, dots, 0, normal, VPpli);
+         }
+         else {
+            emit_op2(p, OPCODE_DP3, dots, WRITEMASK_X, normal, VPpli);
             emit_op2(p, OPCODE_DP3, dots, WRITEMASK_Y, normal, half);
+         }
 
 	 /* Front face lighting:
 	  */
@@ -1181,15 +1196,18 @@ static void build_lighting( struct tnl_program *p )
 
 
 	    if (!is_undef(att)) {
+               /* light is attenuated by distance */
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
                emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
                emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
             } 
             else if (!p->state->material_shininess_is_zero) {
+               /* there's a non-zero specular term */
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
                emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
             } 
             else {
+               /* no attenutation, no specular */
                emit_degenerate_lit(p, lit, dots);
                emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
             }
-- 
cgit v1.2.3


From d378f7b3dfda3b549e4b02380e492671cc34bb59 Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Wed, 25 Jun 2008 08:45:14 -0600
Subject: mesa: point size arrays

---
 src/mesa/main/arrayobj.c            |  9 +++++++++
 src/mesa/main/enable.c              | 12 ++++++++++++
 src/mesa/main/ffvertex_prog.c       | 34 ++++++++++++++++++++++++++++++----
 src/mesa/main/glheader.h            | 10 ++++++++++
 src/mesa/main/mfeatures.h           |  1 +
 src/mesa/main/mtypes.h              |  4 ++++
 src/mesa/main/state.c               |  2 +-
 src/mesa/main/varray.c              | 31 +++++++++++++++++++++++++++++++
 src/mesa/main/varray.h              |  4 ++++
 src/mesa/state_tracker/st_program.c |  4 ++++
 src/mesa/vbo/vbo_exec_array.c       |  4 ++++
 11 files changed, 110 insertions(+), 5 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/arrayobj.c b/src/mesa/main/arrayobj.c
index f08f99d8e1..d62661e2b5 100644
--- a/src/mesa/main/arrayobj.c
+++ b/src/mesa/main/arrayobj.c
@@ -164,6 +164,15 @@ _mesa_initialize_array_object( GLcontext *ctx,
       obj->VertexAttrib[i].Normalized = GL_FALSE;
    }
 
+#if FEATURE_point_size_array
+   obj->PointSize.Type = GL_FLOAT;
+   obj->PointSize.Stride = 0;
+   obj->PointSize.StrideB = 0;
+   obj->PointSize.Ptr = NULL;
+   obj->PointSize.Enabled = GL_FALSE;
+   obj->PointSize.BufferObj = ctx->Array.NullBufferObj;
+#endif
+
 #if FEATURE_ARB_vertex_buffer_object
    /* Vertex array buffers */
    obj->Vertex.BufferObj = ctx->Array.NullBufferObj;
diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c
index 6b4ad8eea3..9dc55d4e69 100644
--- a/src/mesa/main/enable.c
+++ b/src/mesa/main/enable.c
@@ -92,6 +92,13 @@ client_state(GLcontext *ctx, GLenum cap, GLboolean state)
          flag = _NEW_ARRAY_COLOR1;
          break;
 
+#if FEATURE_point_size_array
+      case GL_POINT_SIZE_ARRAY_OES:
+         var = &ctx->Array.ArrayObj->PointSize.Enabled;
+         flag = _NEW_ARRAY_POINT_SIZE;
+         break;
+#endif
+
 #if FEATURE_NV_vertex_program
       case GL_VERTEX_ATTRIB_ARRAY0_NV:
       case GL_VERTEX_ATTRIB_ARRAY1_NV:
@@ -652,6 +659,7 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
       case GL_EDGE_FLAG_ARRAY:
       case GL_FOG_COORDINATE_ARRAY_EXT:
       case GL_SECONDARY_COLOR_ARRAY_EXT:
+      case GL_POINT_SIZE_ARRAY_OES:
          client_state( ctx, cap, state );
          return;
 
@@ -1174,6 +1182,10 @@ _mesa_IsEnabled( GLenum cap )
       case GL_SECONDARY_COLOR_ARRAY_EXT:
          CHECK_EXTENSION(EXT_secondary_color);
          return (ctx->Array.ArrayObj->SecondaryColor.Enabled != 0);
+#if FEATURE_point_size_array
+      case GL_POINT_SIZE_ARRAY_OES:
+         return (ctx->Array.ArrayObj->PointSize.Enabled != 0);
+#endif
 
       /* GL_EXT_histogram */
       case GL_HISTOGRAM:
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index e6c7c1040f..2baae77570 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -63,6 +63,7 @@ struct state_key {
    unsigned separate_specular:1;
    unsigned fog_mode:2;
    unsigned point_attenuated:1;
+   unsigned point_array:1;
    unsigned texture_enabled_global:1;
    unsigned fragprog_inputs_read:12;
 
@@ -264,6 +265,9 @@ static struct state_key *make_state_key( GLcontext *ctx )
    if (ctx->Point._Attenuated)
       key->point_attenuated = 1;
 
+   if (ctx->Array.ArrayObj->PointSize.Enabled)
+      key->point_array = 1;
+
    if (ctx->Texture._TexGenEnabled ||
        ctx->Texture._TexMatEnabled ||
        ctx->Texture._EnabledUnits)
@@ -444,12 +448,18 @@ static void release_temps( struct tnl_program *p )
 
 
 
+/**
+ * \param input  one of VERT_ATTRIB_x tokens.
+ */
 static struct ureg register_input( struct tnl_program *p, GLuint input )
 {
    p->program->Base.InputsRead |= (1<<input);
    return make_ureg(PROGRAM_INPUT, input);
 }
 
+/**
+ * \param input  one of VERT_RESULT_x tokens.
+ */
 static struct ureg register_output( struct tnl_program *p, GLuint output )
 {
    p->program->Base.OutputsWritten |= (1<<output);
@@ -1518,7 +1528,10 @@ static void build_texture_transform( struct tnl_program *p )
 }
 
 
-static void build_pointsize( struct tnl_program *p )
+/**
+ * Point size attenuation computation.
+ */
+static void build_atten_pointsize( struct tnl_program *p )
 {
    struct ureg eye = get_eye_position_z(p);
    struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
@@ -1555,13 +1568,24 @@ static void build_pointsize( struct tnl_program *p )
 /**
  * Emit constant point size.
  */
-static void constant_pointsize( struct tnl_program *p )
+static void build_constant_pointsize( struct tnl_program *p )
 {
    struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
    struct ureg out = register_output(p, VERT_RESULT_PSIZ);
    emit_op1(p, OPCODE_MOV, out, WRITEMASK_X, state_size);
 }
 
+/**
+ * Pass-though per-vertex point size, from user's point size array.
+ */
+static void build_array_pointsize( struct tnl_program *p )
+{
+   struct ureg in = register_input(p, VERT_ATTRIB_POINT_SIZE);
+   struct ureg out = register_output(p, VERT_RESULT_PSIZ);
+   emit_op1(p, OPCODE_MOV, out, WRITEMASK_X, in);
+}
+
+
 static void build_tnl_program( struct tnl_program *p )
 {   /* Emit the program, starting with modelviewproject:
     */
@@ -1589,10 +1613,12 @@ static void build_tnl_program( struct tnl_program *p )
       build_texture_transform(p);
 
    if (p->state->point_attenuated)
-      build_pointsize(p);
+      build_atten_pointsize(p);
+   else if (p->state->point_array)
+      build_array_pointsize(p);
 #if 0
    else
-      constant_pointsize(p);
+      build_constant_pointsize(p);
 #endif
 
    /* Finish up:
diff --git a/src/mesa/main/glheader.h b/src/mesa/main/glheader.h
index 57d7e60ad3..3131a356b8 100644
--- a/src/mesa/main/glheader.h
+++ b/src/mesa/main/glheader.h
@@ -167,6 +167,16 @@
 #endif
 
 
+#ifndef GL_OES_point_size_array
+#define GL_POINT_SIZE_ARRAY_OES                                 0x8B9C
+#define GL_POINT_SIZE_ARRAY_TYPE_OES                            0x898A
+#define GL_POINT_SIZE_ARRAY_STRIDE_OES                          0x898B
+#define GL_POINT_SIZE_ARRAY_POINTER_OES                         0x898C
+#define GL_POINT_SIZE_ARRAY_BUFFER_BINDING_OES                  0x8B9F
+#endif
+
+
+
 #if !defined(CAPI) && defined(WIN32) && !defined(BUILD_FOR_SNAP)
 #define CAPI _cdecl
 #endif
diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h
index a305bbd605..c3c337ea90 100644
--- a/src/mesa/main/mfeatures.h
+++ b/src/mesa/main/mfeatures.h
@@ -49,6 +49,7 @@
 #define FEATURE_fixedpt 0
 #define FEATURE_histogram  _HAVE_FULL_GL
 #define FEATURE_pixel_transfer  _HAVE_FULL_GL
+#define FEATURE_point_size_array 0
 #define FEATURE_texgen  _HAVE_FULL_GL
 #define FEATURE_texture_fxt1  _HAVE_FULL_GL
 #define FEATURE_texture_s3tc  _HAVE_FULL_GL
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 8a6c84368a..dceb7611f2 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -151,6 +151,7 @@ enum
    VERT_ATTRIB_COLOR1 = 4,
    VERT_ATTRIB_FOG = 5,
    VERT_ATTRIB_COLOR_INDEX = 6,
+   VERT_ATTRIB_POINT_SIZE = 6,  /*alias*/
    VERT_ATTRIB_EDGEFLAG = 7,
    VERT_ATTRIB_TEX0 = 8,
    VERT_ATTRIB_TEX1 = 9,
@@ -1707,6 +1708,7 @@ struct gl_array_object
    struct gl_client_array Index;
    struct gl_client_array EdgeFlag;
    struct gl_client_array TexCoord[MAX_TEXTURE_COORD_UNITS];
+   struct gl_client_array PointSize;
    /*@}*/
 
    /** Generic arrays for vertex programs/shaders */
@@ -2748,6 +2750,7 @@ struct gl_matrix_stack
 #define _NEW_ARRAY_FOGCOORD         VERT_BIT_FOG
 #define _NEW_ARRAY_INDEX            VERT_BIT_COLOR_INDEX
 #define _NEW_ARRAY_EDGEFLAG         VERT_BIT_EDGEFLAG
+#define _NEW_ARRAY_POINT_SIZE       VERT_BIT_COLOR_INDEX  /* aliased */
 #define _NEW_ARRAY_TEXCOORD_0       VERT_BIT_TEX0
 #define _NEW_ARRAY_TEXCOORD_1       VERT_BIT_TEX1
 #define _NEW_ARRAY_TEXCOORD_2       VERT_BIT_TEX2
@@ -2765,6 +2768,7 @@ struct gl_matrix_stack
 /*@}*/
 
 
+
 /**
  * \name A bunch of flags that we think might be useful to drivers.
  * 
diff --git a/src/mesa/main/state.c b/src/mesa/main/state.c
index 4d1fdbf47c..315253d90a 100644
--- a/src/mesa/main/state.c
+++ b/src/mesa/main/state.c
@@ -448,7 +448,7 @@ _mesa_update_state_locked( GLcontext *ctx )
       prog_flags |= (_NEW_TEXTURE | _NEW_FOG | _DD_NEW_SEPARATE_SPECULAR);
    }
    if (ctx->VertexProgram._MaintainTnlProgram) {
-      prog_flags |= (_NEW_TEXTURE | _NEW_TEXTURE_MATRIX |
+      prog_flags |= (_NEW_ARRAY | _NEW_TEXTURE | _NEW_TEXTURE_MATRIX |
                      _NEW_TRANSFORM | _NEW_POINT |
                      _NEW_FOG | _NEW_LIGHT |
                      _MESA_NEW_NEED_EYE_COORDS);
diff --git a/src/mesa/main/varray.c b/src/mesa/main/varray.c
index 220db35855..50fe874556 100644
--- a/src/mesa/main/varray.c
+++ b/src/mesa/main/varray.c
@@ -468,6 +468,37 @@ _mesa_EdgeFlagPointer(GLsizei stride, const GLvoid *ptr)
 }
 
 
+void GLAPIENTRY
+_mesa_PointSizePointer(GLenum type, GLsizei stride, const GLvoid *ptr)
+{
+   GLsizei elementSize;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
+
+   if (stride < 0) {
+      _mesa_error( ctx, GL_INVALID_VALUE, "glPointSizePointer(stride)" );
+      return;
+   }
+
+   switch (type) {
+      case GL_FLOAT:
+         elementSize = sizeof(GLfloat);
+         break;
+#if FEATURE_fixedpt
+      case GL_FIXED:
+         elementSize = sizeof(GLfixed);
+         break;
+#endif
+      default:
+         _mesa_error( ctx, GL_INVALID_ENUM, "glPointSizePointer(type)" );
+         return;
+   }
+
+   update_array(ctx, &ctx->Array.ArrayObj->PointSize, _NEW_ARRAY_POINT_SIZE,
+                elementSize, 1, type, stride, GL_FALSE, ptr);
+}
+
+
 #if FEATURE_NV_vertex_program
 void GLAPIENTRY
 _mesa_VertexAttribPointerNV(GLuint index, GLint size, GLenum type,
diff --git a/src/mesa/main/varray.h b/src/mesa/main/varray.h
index ba91ecf5f6..f557940738 100644
--- a/src/mesa/main/varray.h
+++ b/src/mesa/main/varray.h
@@ -111,6 +111,10 @@ _mesa_SecondaryColorPointerEXT(GLint size, GLenum type,
 			       GLsizei stride, const GLvoid *ptr);
 
 
+extern void GLAPIENTRY
+_mesa_PointSizePointer(GLenum type, GLsizei stride, const GLvoid *ptr);
+
+
 extern void GLAPIENTRY
 _mesa_VertexAttribPointerNV(GLuint index, GLint size, GLenum type,
                             GLsizei stride, const GLvoid *pointer);
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index d450c30694..958096f68e 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -136,6 +136,10 @@ st_translate_vertex_program(struct st_context *st,
             vs_input_semantic_name[slot] = TGSI_SEMANTIC_FOG;
             vs_input_semantic_index[slot] = 0;
             break;
+         case VERT_ATTRIB_POINT_SIZE:
+            vs_input_semantic_name[slot] = TGSI_SEMANTIC_PSIZE;
+            vs_input_semantic_index[slot] = 0;
+            break;
          case VERT_ATTRIB_TEX0:
          case VERT_ATTRIB_TEX1:
          case VERT_ATTRIB_TEX2:
diff --git a/src/mesa/vbo/vbo_exec_array.c b/src/mesa/vbo/vbo_exec_array.c
index e3d2fc51cb..bf97956286 100644
--- a/src/mesa/vbo/vbo_exec_array.c
+++ b/src/mesa/vbo/vbo_exec_array.c
@@ -107,6 +107,10 @@ static void bind_array_obj( GLcontext *ctx )
    exec->array.legacy_array[VERT_ATTRIB_COLOR1] = &ctx->Array.ArrayObj->SecondaryColor;
    exec->array.legacy_array[VERT_ATTRIB_FOG] = &ctx->Array.ArrayObj->FogCoord;
    exec->array.legacy_array[VERT_ATTRIB_COLOR_INDEX] = &ctx->Array.ArrayObj->Index;
+   if (ctx->Array.ArrayObj->PointSize.Enabled) {
+      /* this aliases COLOR_INDEX */
+      exec->array.legacy_array[VERT_ATTRIB_POINT_SIZE] = &ctx->Array.ArrayObj->PointSize;
+   }
    exec->array.legacy_array[VERT_ATTRIB_EDGEFLAG] = &ctx->Array.ArrayObj->EdgeFlag;
 
    for (i = 0; i < 8; i++)
-- 
cgit v1.2.3


From a1ec6efce0f614dfc2fb7af2cab68eca3be43850 Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Sat, 28 Jun 2008 16:15:03 -0600
Subject: mesa: check FEATURE_point_size_array

---
 src/mesa/main/ffvertex_prog.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 2baae77570..5f3def257d 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -265,8 +265,10 @@ static struct state_key *make_state_key( GLcontext *ctx )
    if (ctx->Point._Attenuated)
       key->point_attenuated = 1;
 
+#if FEATURE_point_size_array
    if (ctx->Array.ArrayObj->PointSize.Enabled)
       key->point_array = 1;
+#endif
 
    if (ctx->Texture._TexGenEnabled ||
        ctx->Texture._TexMatEnabled ||
-- 
cgit v1.2.3


From 4e4cb0274050f300da9801256b66d3be528d549f Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Thu, 28 Aug 2008 08:38:27 -0600
Subject: mesa: bump MAX_INSN to 300

---
 src/mesa/main/ffvertex_prog.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 5f3def257d..b93376a564 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -319,7 +319,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
  */
 #define PREFER_DP4 0
 
-#define MAX_INSN 256
+#define MAX_INSN 300
 
 /* Use uregs to represent registers internally, translate to Mesa's
  * expected formats on emit.  
-- 
cgit v1.2.3


From cacf5f21efcbf0ddde90bf8b7fc2b9ba1873302e Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Thu, 28 Aug 2008 15:11:04 -0600
Subject: mesa: dynamically grow the fixed function vertex program as needed

Don't use a fixed-size array.  Saves memory in most cases and avoids
potential overflow for long programs.
---
 src/mesa/main/ffvertex_prog.c | 56 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 46 insertions(+), 10 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index b93376a564..e9ddc06980 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -319,7 +319,6 @@ static struct state_key *make_state_key( GLcontext *ctx )
  */
 #define PREFER_DP4 0
 
-#define MAX_INSN 300
 
 /* Use uregs to represent registers internally, translate to Mesa's
  * expected formats on emit.  
@@ -335,16 +334,18 @@ static struct state_key *make_state_key( GLcontext *ctx )
  */
 struct ureg {
    GLuint file:4;
-   GLint idx:8;      /* relative addressing may be negative */
+   GLint idx:9;      /* relative addressing may be negative */
+                     /* sizeof(idx) should == sizeof(prog_src_reg::Index) */
    GLuint negate:1;
    GLuint swz:12;
-   GLuint pad:7;
+   GLuint pad:6;
 };
 
 
 struct tnl_program {
    const struct state_key *state;
    struct gl_vertex_program *program;
+   GLint max_inst;  /** number of instructions allocated for program */
    
    GLuint temp_in_use;
    GLuint temp_reserved;
@@ -362,7 +363,7 @@ struct tnl_program {
 
 static const struct ureg undef = { 
    PROGRAM_UNDEFINED,
-   ~0,
+   0,
    0,
    0,
    0
@@ -558,6 +559,8 @@ static void emit_arg( struct prog_src_register *src,
    src->Abs = 0;
    src->NegateAbs = 0;
    src->RelAddr = 0;
+   /* Check that bitfield sizes aren't exceeded */
+   ASSERT(src->Index == reg.idx);
 }
 
 static void emit_dst( struct prog_dst_register *dst,
@@ -571,6 +574,8 @@ static void emit_dst( struct prog_dst_register *dst,
    dst->CondSwizzle = SWIZZLE_NOOP;
    dst->CondSrc = 0;
    dst->pad = 0;
+   /* Check that bitfield sizes aren't exceeded */
+   ASSERT(dst->Index == reg.idx);
 }
 
 static void debug_insn( struct prog_instruction *inst, const char *fn,
@@ -600,14 +605,37 @@ static void emit_op3fn(struct tnl_program *p,
 		       const char *fn,
 		       GLuint line)
 {
-   GLuint nr = p->program->Base.NumInstructions++;
-   struct prog_instruction *inst = &p->program->Base.Instructions[nr];
+   GLuint nr;
+   struct prog_instruction *inst;
       
-   if (p->program->Base.NumInstructions > MAX_INSN) {
-      _mesa_problem(0, "Out of instructions in emit_op3fn\n");
-      return;
+   assert(p->program->Base.NumInstructions <= p->max_inst);
+
+   if (p->program->Base.NumInstructions == p->max_inst) {
+      /* need to extend the program's instruction array */
+      struct prog_instruction *newInst;
+
+      /* double the size */
+      p->max_inst *= 2;
+
+      newInst = _mesa_alloc_instructions(p->max_inst);
+      if (!newInst) {
+         _mesa_error(NULL, GL_OUT_OF_MEMORY, "vertex program build");
+         return;
+      }
+
+      _mesa_copy_instructions(newInst,
+                              p->program->Base.Instructions,
+                              p->program->Base.NumInstructions);
+
+      _mesa_free_instructions(p->program->Base.Instructions,
+                              p->program->Base.NumInstructions);
+
+      p->program->Base.Instructions = newInst;
    }
       
+   nr = p->program->Base.NumInstructions++;
+
+   inst = &p->program->Base.Instructions[nr];
    inst->Opcode = (enum prog_opcode) op; 
    inst->StringPos = 0;
    inst->Data = 0;
@@ -1621,12 +1649,16 @@ static void build_tnl_program( struct tnl_program *p )
 #if 0
    else
       build_constant_pointsize(p);
+#else
+   (void) build_constant_pointsize;
 #endif
 
    /* Finish up:
     */
    emit_op1(p, OPCODE_END, undef, 0, undef);
 
+   _mesa_print_program(&p->program->Base);
+
    /* Disassemble:
     */
    if (DISASSEM) {
@@ -1657,7 +1689,11 @@ create_new_program( const struct state_key *key,
    else
       p.temp_reserved = ~((1<<max_temps)-1);
 
-   p.program->Base.Instructions = _mesa_alloc_instructions(MAX_INSN);
+   /* Start by allocating 32 instructions.
+    * If we need more, we'll grow the instruction array as needed.
+    */
+   p.max_inst = 32;
+   p.program->Base.Instructions = _mesa_alloc_instructions(p.max_inst);
    p.program->Base.String = NULL;
    p.program->Base.NumInstructions =
    p.program->Base.NumTemporaries =
-- 
cgit v1.2.3


From ee87d7d9ecc5eca261ce58115ae1ed2c273ceda6 Mon Sep 17 00:00:00 2001
From: Brian Paul <brian.paul@tungstengraphics.com>
Date: Thu, 28 Aug 2008 15:21:44 -0600
Subject: mesa: remove debug code

---
 src/mesa/main/ffvertex_prog.c | 2 --
 1 file changed, 2 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index e9ddc06980..62802ffe5d 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1657,8 +1657,6 @@ static void build_tnl_program( struct tnl_program *p )
     */
    emit_op1(p, OPCODE_END, undef, 0, undef);
 
-   _mesa_print_program(&p->program->Base);
-
    /* Disassemble:
     */
    if (DISASSEM) {
-- 
cgit v1.2.3


From 978f07bebdd5a368b8f900ed9cf77d3464a77ec0 Mon Sep 17 00:00:00 2001
From: Michal Krol <michal@tungstengraphics.com>
Date: Thu, 4 Sep 2008 13:37:15 +0200
Subject: mesa: Silence compiler warnings on Windows.

---
 src/mesa/main/ffvertex_prog.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 62802ffe5d..787672be9f 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -608,7 +608,7 @@ static void emit_op3fn(struct tnl_program *p,
    GLuint nr;
    struct prog_instruction *inst;
       
-   assert(p->program->Base.NumInstructions <= p->max_inst);
+   assert((GLint) p->program->Base.NumInstructions <= p->max_inst);
 
    if (p->program->Base.NumInstructions == p->max_inst) {
       /* need to extend the program's instruction array */
-- 
cgit v1.2.3


From 568e96b4533c5135f4d7f568a81cdfc0a6dcd7eb Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keith@tungstengraphics.com>
Date: Tue, 14 Oct 2008 14:15:26 +0100
Subject: mesa: modify fixed function vertex programs not to reference constant
 attributes

---
 src/mesa/main/context.c       |  1 +
 src/mesa/main/ffvertex_prog.c | 70 ++++++++++++++++++++++++-------------------
 2 files changed, 41 insertions(+), 30 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index 144da61384..e5ec35c77f 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -1064,6 +1064,7 @@ init_attrib_groups(GLcontext *ctx)
    ctx->NewState = _NEW_ALL;
    ctx->ErrorValue = (GLenum) GL_NO_ERROR;
    ctx->_Facing = 0;
+   ctx->varying_vp_inputs = ~0;
 
    return GL_TRUE;
 }
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 787672be9f..dff4306322 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -47,17 +47,17 @@
 
 
 struct state_key {
+   unsigned light_color_material_mask:12;
+   unsigned light_material_mask:12;
    unsigned light_global_enabled:1;
    unsigned light_local_viewer:1;
    unsigned light_twoside:1;
    unsigned light_color_material:1;
-   unsigned light_color_material_mask:12;
-   unsigned light_material_mask:12;
    unsigned material_shininess_is_zero:1;
-
    unsigned need_eye_coords:1;
    unsigned normalize:1;
    unsigned rescale_normals:1;
+
    unsigned fog_source_is_depth:1;
    unsigned tnl_do_vertex_fog:1;
    unsigned separate_specular:1;
@@ -67,6 +67,8 @@ struct state_key {
    unsigned texture_enabled_global:1;
    unsigned fragprog_inputs_read:12;
 
+   unsigned varying_vp_inputs;
+
    struct {
       unsigned light_enabled:1;
       unsigned light_eyepos3_is_zero:1;
@@ -193,6 +195,7 @@ static struct state_key *make_state_key( GLcontext *ctx )
    key->need_eye_coords = ctx->_NeedEyeCoords;
 
    key->fragprog_inputs_read = fp->Base.InputsRead;
+   key->varying_vp_inputs = ctx->varying_vp_inputs;
 
    if (ctx->RenderMode == GL_FEEDBACK) {
       /* make sure the vertprog emits color and tex0 */
@@ -450,14 +453,46 @@ static void release_temps( struct tnl_program *p )
 }
 
 
+static struct ureg register_param5(struct tnl_program *p, 
+				   GLint s0,
+				   GLint s1,
+				   GLint s2,
+				   GLint s3,
+                                   GLint s4)
+{
+   gl_state_index tokens[STATE_LENGTH];
+   GLint idx;
+   tokens[0] = s0;
+   tokens[1] = s1;
+   tokens[2] = s2;
+   tokens[3] = s3;
+   tokens[4] = s4;
+   idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
+   return make_ureg(PROGRAM_STATE_VAR, idx);
+}
+
+
+#define register_param1(p,s0)          register_param5(p,s0,0,0,0,0)
+#define register_param2(p,s0,s1)       register_param5(p,s0,s1,0,0,0)
+#define register_param3(p,s0,s1,s2)    register_param5(p,s0,s1,s2,0,0)
+#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
+
+
 
 /**
  * \param input  one of VERT_ATTRIB_x tokens.
  */
 static struct ureg register_input( struct tnl_program *p, GLuint input )
 {
-   p->program->Base.InputsRead |= (1<<input);
-   return make_ureg(PROGRAM_INPUT, input);
+   /* Material attribs are passed here as inputs >= 32
+    */
+   if (input >= 32 || (p->state->varying_vp_inputs & (1<<input))) {
+      p->program->Base.InputsRead |= (1<<input);
+      return make_ureg(PROGRAM_INPUT, input);
+   }
+   else {
+      return register_param3( p, STATE_INTERNAL, STATE_CURRENT_ATTRIB, input );
+   }
 }
 
 /**
@@ -506,31 +541,6 @@ static struct ureg get_identity_param( struct tnl_program *p )
    return p->identity;
 }
 
-static struct ureg register_param5(struct tnl_program *p, 
-				   GLint s0,
-				   GLint s1,
-				   GLint s2,
-				   GLint s3,
-                                   GLint s4)
-{
-   gl_state_index tokens[STATE_LENGTH];
-   GLint idx;
-   tokens[0] = s0;
-   tokens[1] = s1;
-   tokens[2] = s2;
-   tokens[3] = s3;
-   tokens[4] = s4;
-   idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
-   return make_ureg(PROGRAM_STATE_VAR, idx);
-}
-
-
-#define register_param1(p,s0)          register_param5(p,s0,0,0,0,0)
-#define register_param2(p,s0,s1)       register_param5(p,s0,s1,0,0,0)
-#define register_param3(p,s0,s1,s2)    register_param5(p,s0,s1,s2,0,0)
-#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
-
-
 static void register_matrix_param5( struct tnl_program *p,
 				    GLint s0, /* modelview, projection, etc */
 				    GLint s1, /* texture matrix number */
-- 
cgit v1.2.3


From 9705cff2033f1771a39ac3bb78eb5fcea522218a Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Sat, 21 Feb 2009 13:23:04 -0700
Subject: mesa: re-org texgen state

New gl_texgen struct allows quite a bit of code reduction.
---
 src/mesa/drivers/dri/r200/r200_texstate.c     |  42 +-
 src/mesa/drivers/dri/radeon/radeon_texstate.c |  24 +-
 src/mesa/main/attrib.c                        |  32 +-
 src/mesa/main/ffvertex_prog.c                 |   8 +-
 src/mesa/main/mtypes.h                        |  40 +-
 src/mesa/main/texgen.c                        | 588 +++++++-------------------
 src/mesa/main/texstate.c                      |  65 ++-
 src/mesa/shader/prog_statevars.c              |  16 +-
 src/mesa/tnl/t_rasterpos.c                    |  24 +-
 src/mesa/tnl/t_vb_texgen.c                    |  24 +-
 10 files changed, 295 insertions(+), 568 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/drivers/dri/r200/r200_texstate.c b/src/mesa/drivers/dri/r200/r200_texstate.c
index 3f9a2f4ac1..0ad5651cd4 100644
--- a/src/mesa/drivers/dri/r200/r200_texstate.c
+++ b/src/mesa/drivers/dri/r200/r200_texstate.c
@@ -1366,27 +1366,27 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
       fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit);
 
    if (texUnit->TexGenEnabled & S_BIT) {
-      mode = texUnit->GenModeS;
+      mode = texUnit->GenS.Mode;
    } else {
       tgcm |= R200_TEXGEN_COMP_S << (unit * 4);
    }
 
    if (texUnit->TexGenEnabled & T_BIT) {
-      if (texUnit->GenModeT != mode)
+      if (texUnit->GenT.Mode != mode)
 	 mixed_fallback = GL_TRUE;
    } else {
       tgcm |= R200_TEXGEN_COMP_T << (unit * 4);
    }
 
    if (texUnit->TexGenEnabled & R_BIT) {
-      if (texUnit->GenModeR != mode)
+      if (texUnit->GenR.Mode != mode)
 	 mixed_fallback = GL_TRUE;
    } else {
       tgcm |= R200_TEXGEN_COMP_R << (unit * 4);
    }
 
    if (texUnit->TexGenEnabled & Q_BIT) {
-      if (texUnit->GenModeQ != mode)
+      if (texUnit->GenQ.Mode != mode)
 	 mixed_fallback = GL_TRUE;
    } else {
       tgcm |= R200_TEXGEN_COMP_Q << (unit * 4);
@@ -1395,8 +1395,8 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
    if (mixed_fallback) {
       if (R200_DEBUG & DEBUG_FALLBACKS)
 	 fprintf(stderr, "fallback mixed texgen, 0x%x (0x%x 0x%x 0x%x 0x%x)\n",
-		 texUnit->TexGenEnabled, texUnit->GenModeS, texUnit->GenModeT,
-		 texUnit->GenModeR, texUnit->GenModeQ);
+		 texUnit->TexGenEnabled, texUnit->GenS.Mode, texUnit->GenT.Mode,
+		 texUnit->GenR.Mode, texUnit->GenQ.Mode);
       return GL_FALSE;
    }
 
@@ -1414,8 +1414,10 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
    switch (mode) {
    case GL_OBJECT_LINEAR: {
       GLuint needtgenable = r200_need_dis_texgen( texUnit->TexGenEnabled,
-				texUnit->ObjectPlaneS, texUnit->ObjectPlaneT,
-				texUnit->ObjectPlaneR, texUnit->ObjectPlaneQ );
+                                                  texUnit->GenS.ObjectPlane,
+                                                  texUnit->GenT.ObjectPlane,
+                                                  texUnit->GenR.ObjectPlane,
+                                                  texUnit->GenQ.ObjectPlane );
       if (needtgenable & (S_BIT | T_BIT)) {
 	 if (R200_DEBUG & DEBUG_FALLBACKS)
 	 fprintf(stderr, "fallback mixed texgen / obj plane, 0x%x\n",
@@ -1431,17 +1433,19 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
 
       tgi |= R200_TEXGEN_INPUT_OBJ << inputshift;
       set_texgen_matrix( rmesa, unit, 
-	 (texUnit->TexGenEnabled & S_BIT) ? texUnit->ObjectPlaneS : I,
-	 (texUnit->TexGenEnabled & T_BIT) ? texUnit->ObjectPlaneT : I + 4,
-	 (texUnit->TexGenEnabled & R_BIT) ? texUnit->ObjectPlaneR : I + 8,
-	 (texUnit->TexGenEnabled & Q_BIT) ? texUnit->ObjectPlaneQ : I + 12);
+	 (texUnit->TexGenEnabled & S_BIT) ? texUnit->GenS.ObjectPlane : I,
+	 (texUnit->TexGenEnabled & T_BIT) ? texUnit->GenT.ObjectPlane : I + 4,
+	 (texUnit->TexGenEnabled & R_BIT) ? texUnit->GenR.ObjectPlane : I + 8,
+	 (texUnit->TexGenEnabled & Q_BIT) ? texUnit->GenQ.ObjectPlane : I + 12);
       }
       break;
 
    case GL_EYE_LINEAR: {
       GLuint needtgenable = r200_need_dis_texgen( texUnit->TexGenEnabled,
-				texUnit->EyePlaneS, texUnit->EyePlaneT,
-				texUnit->EyePlaneR, texUnit->EyePlaneQ );
+                                                  texUnit->GenS.EyePlane,
+                                                  texUnit->GenT.EyePlane,
+                                                  texUnit->GenR.EyePlane,
+                                                  texUnit->GenQ.EyePlane );
       if (needtgenable & (S_BIT | T_BIT)) {
 	 if (R200_DEBUG & DEBUG_FALLBACKS)
 	 fprintf(stderr, "fallback mixed texgen / eye plane, 0x%x\n",
@@ -1456,10 +1460,10 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
       }
       tgi |= R200_TEXGEN_INPUT_EYE << inputshift;
       set_texgen_matrix( rmesa, unit,
-	 (texUnit->TexGenEnabled & S_BIT) ? texUnit->EyePlaneS : I,
-	 (texUnit->TexGenEnabled & T_BIT) ? texUnit->EyePlaneT : I + 4,
-	 (texUnit->TexGenEnabled & R_BIT) ? texUnit->EyePlaneR : I + 8,
-	 (texUnit->TexGenEnabled & Q_BIT) ? texUnit->EyePlaneQ : I + 12);
+	 (texUnit->TexGenEnabled & S_BIT) ? texUnit->GenS.EyePlane : I,
+	 (texUnit->TexGenEnabled & T_BIT) ? texUnit->GenT.EyePlane : I + 4,
+	 (texUnit->TexGenEnabled & R_BIT) ? texUnit->GenR.EyePlane : I + 8,
+	 (texUnit->TexGenEnabled & Q_BIT) ? texUnit->GenQ.EyePlane : I + 12);
       }
       break;
 
@@ -1495,7 +1499,7 @@ static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit )
        */
       if (R200_DEBUG & DEBUG_FALLBACKS)
 	 fprintf(stderr, "fallback unsupported texgen, %d\n",
-		 texUnit->GenModeS);
+		 texUnit->GenS.Mode);
       return GL_FALSE;
    }
 
diff --git a/src/mesa/drivers/dri/radeon/radeon_texstate.c b/src/mesa/drivers/dri/radeon/radeon_texstate.c
index 1e2f654add..b165205c09 100644
--- a/src/mesa/drivers/dri/radeon/radeon_texstate.c
+++ b/src/mesa/drivers/dri/radeon/radeon_texstate.c
@@ -1023,11 +1023,11 @@ static GLboolean radeon_validate_texgen( GLcontext *ctx, GLuint unit )
     */
    else if ( (texUnit->TexGenEnabled & S_BIT) &&
 	     (texUnit->TexGenEnabled & T_BIT) &&
-	     (texUnit->GenModeS == texUnit->GenModeT) ) {
+	     (texUnit->GenS.Mode == texUnit->GenT.Mode) ) {
       if ( ((texUnit->TexGenEnabled & R_BIT) &&
-	    (texUnit->GenModeS != texUnit->GenModeR)) ||
+	    (texUnit->GenS.Mode != texUnit->GenR.Mode)) ||
 	   ((texUnit->TexGenEnabled & Q_BIT) &&
-	    (texUnit->GenModeS != texUnit->GenModeQ)) ) {
+	    (texUnit->GenS.Mode != texUnit->GenQ.Mode)) ) {
 	 /* Mixed modes, fallback:
 	  */
 	 if (RADEON_DEBUG & DEBUG_FALLBACKS)
@@ -1051,23 +1051,23 @@ static GLboolean radeon_validate_texgen( GLcontext *ctx, GLuint unit )
       rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] |= RADEON_Q_BIT(unit);
    }
 
-   switch (texUnit->GenModeS) {
+   switch (texUnit->GenS.Mode) {
    case GL_OBJECT_LINEAR:
       rmesa->TexGenEnabled |= RADEON_TEXGEN_INPUT_OBJ << inputshift;
       set_texgen_matrix( rmesa, unit,
-			 texUnit->ObjectPlaneS,
-			 texUnit->ObjectPlaneT,
-			 texUnit->ObjectPlaneR,
-			 texUnit->ObjectPlaneQ);
+			 texUnit->GenS.ObjectPlane,
+			 texUnit->GenT.ObjectPlane,
+			 texUnit->GenR.ObjectPlane,
+			 texUnit->GenQ.ObjectPlane);
       break;
 
    case GL_EYE_LINEAR:
       rmesa->TexGenEnabled |= RADEON_TEXGEN_INPUT_EYE << inputshift;
       set_texgen_matrix( rmesa, unit,
-			 texUnit->EyePlaneS,
-			 texUnit->EyePlaneT,
-			 texUnit->EyePlaneR,
-			 texUnit->EyePlaneQ);
+			 texUnit->GenS.EyePlane,
+			 texUnit->GenT.EyePlane,
+			 texUnit->GenR.EyePlane,
+			 texUnit->GenQ.EyePlane);
       break;
 
    case GL_REFLECTION_MAP_NV:
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index add0045959..dc33eb1fac 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -699,26 +699,26 @@ pop_texture_group(GLcontext *ctx, struct texture_state *texstate)
       }
       _mesa_TexEnvi(GL_TEXTURE_ENV, GL_TEXTURE_ENV_MODE, unit->EnvMode);
       _mesa_TexEnvfv(GL_TEXTURE_ENV, GL_TEXTURE_ENV_COLOR, unit->EnvColor);
-      _mesa_TexGeni(GL_S, GL_TEXTURE_GEN_MODE, unit->GenModeS);
-      _mesa_TexGeni(GL_T, GL_TEXTURE_GEN_MODE, unit->GenModeT);
-      _mesa_TexGeni(GL_R, GL_TEXTURE_GEN_MODE, unit->GenModeR);
-      _mesa_TexGeni(GL_Q, GL_TEXTURE_GEN_MODE, unit->GenModeQ);
-      _mesa_TexGenfv(GL_S, GL_OBJECT_PLANE, unit->ObjectPlaneS);
-      _mesa_TexGenfv(GL_T, GL_OBJECT_PLANE, unit->ObjectPlaneT);
-      _mesa_TexGenfv(GL_R, GL_OBJECT_PLANE, unit->ObjectPlaneR);
-      _mesa_TexGenfv(GL_Q, GL_OBJECT_PLANE, unit->ObjectPlaneQ);
+      _mesa_TexGeni(GL_S, GL_TEXTURE_GEN_MODE, unit->GenS.Mode);
+      _mesa_TexGeni(GL_T, GL_TEXTURE_GEN_MODE, unit->GenT.Mode);
+      _mesa_TexGeni(GL_R, GL_TEXTURE_GEN_MODE, unit->GenR.Mode);
+      _mesa_TexGeni(GL_Q, GL_TEXTURE_GEN_MODE, unit->GenQ.Mode);
+      _mesa_TexGenfv(GL_S, GL_OBJECT_PLANE, unit->GenS.ObjectPlane);
+      _mesa_TexGenfv(GL_T, GL_OBJECT_PLANE, unit->GenT.ObjectPlane);
+      _mesa_TexGenfv(GL_R, GL_OBJECT_PLANE, unit->GenR.ObjectPlane);
+      _mesa_TexGenfv(GL_Q, GL_OBJECT_PLANE, unit->GenQ.ObjectPlane);
       /* Eye plane done differently to avoid re-transformation */
       {
          struct gl_texture_unit *destUnit = &ctx->Texture.Unit[u];
-         COPY_4FV(destUnit->EyePlaneS, unit->EyePlaneS);
-         COPY_4FV(destUnit->EyePlaneT, unit->EyePlaneT);
-         COPY_4FV(destUnit->EyePlaneR, unit->EyePlaneR);
-         COPY_4FV(destUnit->EyePlaneQ, unit->EyePlaneQ);
+         COPY_4FV(destUnit->GenS.EyePlane, unit->GenS.EyePlane);
+         COPY_4FV(destUnit->GenT.EyePlane, unit->GenT.EyePlane);
+         COPY_4FV(destUnit->GenR.EyePlane, unit->GenR.EyePlane);
+         COPY_4FV(destUnit->GenQ.EyePlane, unit->GenQ.EyePlane);
          if (ctx->Driver.TexGen) {
-            ctx->Driver.TexGen(ctx, GL_S, GL_EYE_PLANE, unit->EyePlaneS);
-            ctx->Driver.TexGen(ctx, GL_T, GL_EYE_PLANE, unit->EyePlaneT);
-            ctx->Driver.TexGen(ctx, GL_R, GL_EYE_PLANE, unit->EyePlaneR);
-            ctx->Driver.TexGen(ctx, GL_Q, GL_EYE_PLANE, unit->EyePlaneQ);
+            ctx->Driver.TexGen(ctx, GL_S, GL_EYE_PLANE, unit->GenS.EyePlane);
+            ctx->Driver.TexGen(ctx, GL_T, GL_EYE_PLANE, unit->GenT.EyePlane);
+            ctx->Driver.TexGen(ctx, GL_R, GL_EYE_PLANE, unit->GenR.EyePlane);
+            ctx->Driver.TexGen(ctx, GL_Q, GL_EYE_PLANE, unit->GenQ.EyePlane);
          }
       }
       _mesa_set_enable(ctx, GL_TEXTURE_GEN_S,
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index d70b78f258..42c8cc97c0 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -293,16 +293,16 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
       
 	 key->unit[i].texgen_mode0 = 
 	    translate_texgen( texUnit->TexGenEnabled & (1<<0),
-			      texUnit->GenModeS );
+			      texUnit->GenS.Mode );
 	 key->unit[i].texgen_mode1 = 
 	    translate_texgen( texUnit->TexGenEnabled & (1<<1),
-			      texUnit->GenModeT );
+			      texUnit->GenT.Mode );
 	 key->unit[i].texgen_mode2 = 
 	    translate_texgen( texUnit->TexGenEnabled & (1<<2),
-			      texUnit->GenModeR );
+			      texUnit->GenR.Mode );
 	 key->unit[i].texgen_mode3 = 
 	    translate_texgen( texUnit->TexGenEnabled & (1<<3),
-			      texUnit->GenModeQ );
+			      texUnit->GenQ.Mode );
       }
    }
 }
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 235d6b6bb5..92fe4cb77d 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1492,6 +1492,18 @@ struct gl_tex_env_combine_state
 };
 
 
+/**
+ * Texture coord generation state.
+ */
+struct gl_texgen
+{
+   GLenum Mode;         /**< GL_EYE_LINEAR, GL_SPHERE_MAP, etc */
+   GLbitfield _ModeBit; /**< TEXGEN_x bit corresponding to Mode */
+   GLfloat ObjectPlane[4];
+   GLfloat EyePlane[4];
+};
+
+
 /**
  * Texture unit state.  Contains enable flags, texture environment/function/
  * combiners, texgen state, pointers to current texture objects and
@@ -1504,28 +1516,14 @@ struct gl_texture_unit
 
    GLenum EnvMode;              /**< GL_MODULATE, GL_DECAL, GL_BLEND, etc. */
    GLfloat EnvColor[4];
+
+   struct gl_texgen GenS;
+   struct gl_texgen GenT;
+   struct gl_texgen GenR;
+   struct gl_texgen GenQ;
    GLbitfield TexGenEnabled;	/**< Bitwise-OR of [STRQ]_BIT values */
-   /** \name Tex coord generation mode
-    * Either GL_OBJECT_LINEAR, GL_EYE_LINEAR or GL_SPHERE_MAP. */
-   /*@{*/
-   GLenum GenModeS;		
-   GLenum GenModeT;
-   GLenum GenModeR;
-   GLenum GenModeQ;
-   /*@}*/
-   GLbitfield _GenBitS;
-   GLbitfield _GenBitT;
-   GLbitfield _GenBitR;
-   GLbitfield _GenBitQ;
-   GLbitfield _GenFlags;	/**< bitwise or of _GenBit[STRQ] */
-   GLfloat ObjectPlaneS[4];
-   GLfloat ObjectPlaneT[4];
-   GLfloat ObjectPlaneR[4];
-   GLfloat ObjectPlaneQ[4];
-   GLfloat EyePlaneS[4];
-   GLfloat EyePlaneT[4];
-   GLfloat EyePlaneR[4];
-   GLfloat EyePlaneQ[4];
+   GLbitfield _GenFlags;	/**< Bitwise-OR of Gen[STRQ]._ModeBit */
+
    GLfloat LodBias;		/**< for biasing mipmap levels */
 
    /** 
diff --git a/src/mesa/main/texgen.c b/src/mesa/main/texgen.c
index d2baecb92a..e3feb024c3 100644
--- a/src/mesa/main/texgen.c
+++ b/src/mesa/main/texgen.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.5
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -37,12 +38,33 @@
 #include "math/m_matrix.h"
 
 
+/**
+ * Return texgen state for given coordinate
+ */
+static struct gl_texgen *
+get_texgen(struct gl_texture_unit *texUnit, GLenum coord)
+{
+   switch (coord) {
+   case GL_S:
+      return &texUnit->GenS;
+   case GL_T:
+      return &texUnit->GenT;
+   case GL_R:
+      return &texUnit->GenR;
+   case GL_Q:
+      return &texUnit->GenQ;
+   default:
+      return NULL;
+   }
+}
+
 
 void GLAPIENTRY
 _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params )
 {
-   GET_CURRENT_CONTEXT(ctx);
    struct gl_texture_unit *texUnit;
+   struct gl_texgen *texgen;
+   GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    if (MESA_VERBOSE&(VERBOSE_API|VERBOSE_TEXTURE))
@@ -59,210 +81,79 @@ _mesa_TexGenfv( GLenum coord, GLenum pname, const GLfloat *params )
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
 
-   switch (coord) {
-      case GL_S:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-	    GLenum mode = (GLenum) (GLint) *params;
-	    GLbitfield bits;
-	    switch (mode) {
-	    case GL_OBJECT_LINEAR:
-	       bits = TEXGEN_OBJ_LINEAR;
-	       break;
-	    case GL_EYE_LINEAR:
-	       bits = TEXGEN_EYE_LINEAR;
-	       break;
-	    case GL_REFLECTION_MAP_NV:
-	       bits = TEXGEN_REFLECTION_MAP_NV;
-	       break;
-	    case GL_NORMAL_MAP_NV:
-	       bits = TEXGEN_NORMAL_MAP_NV;
-	       break;
-	    case GL_SPHERE_MAP:
-	       bits = TEXGEN_SPHERE_MAP;
-	       break;
-	    default:
-	       _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(param)" );
-	       return;
-	    }
-	    if (texUnit->GenModeS == mode)
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    texUnit->GenModeS = mode;
-	    texUnit->_GenBitS = bits;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-	    if (TEST_EQ_4V(texUnit->ObjectPlaneS, params))
-		return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-            COPY_4FV(texUnit->ObjectPlaneS, params);
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-	    GLfloat tmp[4];
-            /* Transform plane equation by the inverse modelview matrix */
-            if (_math_matrix_is_dirty(ctx->ModelviewMatrixStack.Top)) {
-               _math_matrix_analyse( ctx->ModelviewMatrixStack.Top );
-            }
-            _mesa_transform_vector( tmp, params, ctx->ModelviewMatrixStack.Top->inv );
-	    if (TEST_EQ_4V(texUnit->EyePlaneS, tmp))
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    COPY_4FV(texUnit->EyePlaneS, tmp);
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_T:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-	    GLenum mode = (GLenum) (GLint) *params;
-	    GLbitfield bitt;
-	    switch (mode) {
-               case GL_OBJECT_LINEAR:
-                  bitt = TEXGEN_OBJ_LINEAR;
-                  break;
-               case GL_EYE_LINEAR:
-                  bitt = TEXGEN_EYE_LINEAR;
-                  break;
-               case GL_REFLECTION_MAP_NV:
-                  bitt = TEXGEN_REFLECTION_MAP_NV;
-                  break;
-               case GL_NORMAL_MAP_NV:
-                  bitt = TEXGEN_NORMAL_MAP_NV;
-                  break;
-               case GL_SPHERE_MAP:
-                  bitt = TEXGEN_SPHERE_MAP;
-                  break;
-               default:
-                  _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(param)" );
-                  return;
-	    }
-	    if (texUnit->GenModeT == mode)
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    texUnit->GenModeT = mode;
-	    texUnit->_GenBitT = bitt;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-	    if (TEST_EQ_4V(texUnit->ObjectPlaneT, params))
-		return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-            COPY_4FV(texUnit->ObjectPlaneT, params);
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-	    GLfloat tmp[4];
-            /* Transform plane equation by the inverse modelview matrix */
-	    if (_math_matrix_is_dirty(ctx->ModelviewMatrixStack.Top)) {
-               _math_matrix_analyse( ctx->ModelviewMatrixStack.Top );
-            }
-            _mesa_transform_vector( tmp, params, ctx->ModelviewMatrixStack.Top->inv );
-	    if (TEST_EQ_4V(texUnit->EyePlaneT, tmp))
-		return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    COPY_4FV(texUnit->EyePlaneT, tmp);
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_R:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-	    GLenum mode = (GLenum) (GLint) *params;
-	    GLbitfield bitr;
-	    switch (mode) {
-	    case GL_OBJECT_LINEAR:
-	       bitr = TEXGEN_OBJ_LINEAR;
-	       break;
-	    case GL_REFLECTION_MAP_NV:
-	       bitr = TEXGEN_REFLECTION_MAP_NV;
-	       break;
-	    case GL_NORMAL_MAP_NV:
-	       bitr = TEXGEN_NORMAL_MAP_NV;
-	       break;
-	    case GL_EYE_LINEAR:
-	       bitr = TEXGEN_EYE_LINEAR;
-	       break;
-	    default:
-	       _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(param)" );
-	       return;
-	    }
-	    if (texUnit->GenModeR == mode)
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    texUnit->GenModeR = mode;
-	    texUnit->_GenBitR = bitr;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-	    if (TEST_EQ_4V(texUnit->ObjectPlaneR, params))
-		return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    COPY_4FV(texUnit->ObjectPlaneR, params);
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-	    GLfloat tmp[4];
-            /* Transform plane equation by the inverse modelview matrix */
-            if (_math_matrix_is_dirty(ctx->ModelviewMatrixStack.Top)) {
-               _math_matrix_analyse( ctx->ModelviewMatrixStack.Top );
-            }
-            _mesa_transform_vector( tmp, params, ctx->ModelviewMatrixStack.Top->inv );
-	    if (TEST_EQ_4V(texUnit->EyePlaneR, tmp))
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    COPY_4FV(texUnit->EyePlaneR, tmp);
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_Q:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-	    GLenum mode = (GLenum) (GLint) *params;
-	    GLbitfield bitq;
-	    switch (mode) {
-	    case GL_OBJECT_LINEAR:
-	       bitq = TEXGEN_OBJ_LINEAR;
-	       break;
-	    case GL_EYE_LINEAR:
-	       bitq = TEXGEN_EYE_LINEAR;
-	       break;
-	    default:
-	       _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(param)" );
-	       return;
-	    }
-	    if (texUnit->GenModeQ == mode)
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    texUnit->GenModeQ = mode;
-	    texUnit->_GenBitQ = bitq;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-	    if (TEST_EQ_4V(texUnit->ObjectPlaneQ, params))
-		return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-            COPY_4FV(texUnit->ObjectPlaneQ, params);
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-	    GLfloat tmp[4];
-            /* Transform plane equation by the inverse modelview matrix */
-            if (_math_matrix_is_dirty(ctx->ModelviewMatrixStack.Top)) {
-               _math_matrix_analyse( ctx->ModelviewMatrixStack.Top );
-            }
-            _mesa_transform_vector( tmp, params, ctx->ModelviewMatrixStack.Top->inv );
-	    if (TEST_EQ_4V(texUnit->EyePlaneQ, tmp))
-	       return;
-	    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-	    COPY_4FV(texUnit->EyePlaneQ, tmp);
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      default:
-         _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(coord)" );
-	 return;
+   texgen = get_texgen(texUnit, coord);
+   if (!texgen) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glTexGen(coord)");
+      return;
+   }
+
+   switch (pname) {
+   case GL_TEXTURE_GEN_MODE:
+      {
+         GLenum mode = (GLenum) (GLint) params[0];
+         GLbitfield bit = 0x0;
+         if (texgen->Mode == mode)
+            return;
+         switch (mode) {
+         case GL_OBJECT_LINEAR:
+            bit = TEXGEN_OBJ_LINEAR;
+            break;
+         case GL_EYE_LINEAR:
+            bit = TEXGEN_EYE_LINEAR;
+            break;
+         case GL_SPHERE_MAP:
+            if (coord == GL_S || coord == GL_T)
+               bit = TEXGEN_SPHERE_MAP;
+            break;
+         case GL_REFLECTION_MAP_NV:
+            if (coord != GL_Q)
+               bit = TEXGEN_REFLECTION_MAP_NV;
+            break;
+         case GL_NORMAL_MAP_NV:
+            if (coord != GL_Q)
+               bit = TEXGEN_NORMAL_MAP_NV;
+            break;
+         default:
+            ; /* nop */
+         }
+         if (!bit) {
+            _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(param)" );
+            return;
+         }
+         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+         texgen->Mode = mode;
+         texgen->_ModeBit = bit;
+      }
+      break;
+
+   case GL_OBJECT_PLANE:
+      {
+         if (TEST_EQ_4V(texgen->ObjectPlane, params))
+            return;
+         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+         COPY_4FV(texgen->ObjectPlane, params);
+      }
+      break;
+
+   case GL_EYE_PLANE:
+      {
+         GLfloat tmp[4];
+         /* Transform plane equation by the inverse modelview matrix */
+         if (_math_matrix_is_dirty(ctx->ModelviewMatrixStack.Top)) {
+            _math_matrix_analyse(ctx->ModelviewMatrixStack.Top);
+         }
+         _mesa_transform_vector(tmp, params,
+                                ctx->ModelviewMatrixStack.Top->inv);
+         if (TEST_EQ_4V(texgen->EyePlane, tmp))
+            return;
+         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+         COPY_4FV(texgen->EyePlane, tmp);
+      }
+      break;
+
+   default:
+      _mesa_error( ctx, GL_INVALID_ENUM, "glTexGenfv(pname)" );
+      return;
    }
 
    if (ctx->Driver.TexGen)
@@ -330,7 +221,8 @@ _mesa_TexGeni( GLenum coord, GLenum pname, GLint param )
 void GLAPIENTRY
 _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
 {
-   const struct gl_texture_unit *texUnit;
+   struct gl_texture_unit *texUnit;
+   struct gl_texgen *texgen;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -341,70 +233,24 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
 
-   switch (coord) {
-      case GL_S:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_DOUBLE(texUnit->GenModeS);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneS );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneS );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_T:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_DOUBLE(texUnit->GenModeT);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneT );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneT );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_R:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_DOUBLE(texUnit->GenModeR);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneR );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneR );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_Q:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_DOUBLE(texUnit->GenModeQ);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneQ );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneQ );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(pname)" );
-	    return;
-	 }
-	 break;
-      default:
-         _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(coord)" );
-	 return;
+   texgen = get_texgen(texUnit, coord);
+   if (!texgen) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexGendv(coord)");
+      return;
+   }
+
+   switch (pname) {
+   case GL_TEXTURE_GEN_MODE:
+      params[0] = ENUM_TO_DOUBLE(texgen->Mode);
+      break;
+   case GL_OBJECT_PLANE:
+      COPY_4V(params, texgen->ObjectPlane);
+      break;
+   case GL_EYE_PLANE:
+      COPY_4V(params, texgen->EyePlane);
+      break;
+   default:
+      _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGendv(pname)" );
    }
 }
 
@@ -413,7 +259,8 @@ _mesa_GetTexGendv( GLenum coord, GLenum pname, GLdouble *params )
 void GLAPIENTRY
 _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
 {
-   const struct gl_texture_unit *texUnit;
+   struct gl_texture_unit *texUnit;
+   struct gl_texgen *texgen;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -424,70 +271,24 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
 
-   switch (coord) {
-      case GL_S:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_FLOAT(texUnit->GenModeS);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneS );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneS );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_T:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_FLOAT(texUnit->GenModeT);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneT );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneT );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_R:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_FLOAT(texUnit->GenModeR);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneR );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneR );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_Q:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = ENUM_TO_FLOAT(texUnit->GenModeQ);
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            COPY_4V( params, texUnit->ObjectPlaneQ );
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            COPY_4V( params, texUnit->EyePlaneQ );
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(pname)" );
-	    return;
-	 }
-	 break;
-      default:
-         _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(coord)" );
-	 return;
+   texgen = get_texgen(texUnit, coord);
+   if (!texgen) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexGenfv(coord)");
+      return;
+   }
+
+   switch (pname) {
+   case GL_TEXTURE_GEN_MODE:
+      params[0] = ENUM_TO_FLOAT(texgen->Mode);
+      break;
+   case GL_OBJECT_PLANE:
+      COPY_4V(params, texgen->ObjectPlane);
+      break;
+   case GL_EYE_PLANE:
+      COPY_4V(params, texgen->EyePlane);
+      break;
+   default:
+      _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGenfv(pname)" );
    }
 }
 
@@ -496,7 +297,8 @@ _mesa_GetTexGenfv( GLenum coord, GLenum pname, GLfloat *params )
 void GLAPIENTRY
 _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params )
 {
-   const struct gl_texture_unit *texUnit;
+   struct gl_texture_unit *texUnit;
+   struct gl_texgen *texgen;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -507,94 +309,30 @@ _mesa_GetTexGeniv( GLenum coord, GLenum pname, GLint *params )
 
    texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
 
-   switch (coord) {
-      case GL_S:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = texUnit->GenModeS;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            params[0] = (GLint) texUnit->ObjectPlaneS[0];
-            params[1] = (GLint) texUnit->ObjectPlaneS[1];
-            params[2] = (GLint) texUnit->ObjectPlaneS[2];
-            params[3] = (GLint) texUnit->ObjectPlaneS[3];
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            params[0] = (GLint) texUnit->EyePlaneS[0];
-            params[1] = (GLint) texUnit->EyePlaneS[1];
-            params[2] = (GLint) texUnit->EyePlaneS[2];
-            params[3] = (GLint) texUnit->EyePlaneS[3];
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_T:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = texUnit->GenModeT;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            params[0] = (GLint) texUnit->ObjectPlaneT[0];
-            params[1] = (GLint) texUnit->ObjectPlaneT[1];
-            params[2] = (GLint) texUnit->ObjectPlaneT[2];
-            params[3] = (GLint) texUnit->ObjectPlaneT[3];
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            params[0] = (GLint) texUnit->EyePlaneT[0];
-            params[1] = (GLint) texUnit->EyePlaneT[1];
-            params[2] = (GLint) texUnit->EyePlaneT[2];
-            params[3] = (GLint) texUnit->EyePlaneT[3];
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_R:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = texUnit->GenModeR;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            params[0] = (GLint) texUnit->ObjectPlaneR[0];
-            params[1] = (GLint) texUnit->ObjectPlaneR[1];
-            params[2] = (GLint) texUnit->ObjectPlaneR[2];
-            params[3] = (GLint) texUnit->ObjectPlaneR[3];
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            params[0] = (GLint) texUnit->EyePlaneR[0];
-            params[1] = (GLint) texUnit->EyePlaneR[1];
-            params[2] = (GLint) texUnit->EyePlaneR[2];
-            params[3] = (GLint) texUnit->EyePlaneR[3];
-	 }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(pname)" );
-	    return;
-	 }
-	 break;
-      case GL_Q:
-         if (pname==GL_TEXTURE_GEN_MODE) {
-            params[0] = texUnit->GenModeQ;
-	 }
-	 else if (pname==GL_OBJECT_PLANE) {
-            params[0] = (GLint) texUnit->ObjectPlaneQ[0];
-            params[1] = (GLint) texUnit->ObjectPlaneQ[1];
-            params[2] = (GLint) texUnit->ObjectPlaneQ[2];
-            params[3] = (GLint) texUnit->ObjectPlaneQ[3];
-	 }
-	 else if (pname==GL_EYE_PLANE) {
-            params[0] = (GLint) texUnit->EyePlaneQ[0];
-            params[1] = (GLint) texUnit->EyePlaneQ[1];
-            params[2] = (GLint) texUnit->EyePlaneQ[2];
-            params[3] = (GLint) texUnit->EyePlaneQ[3];
-         }
-	 else {
-	    _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(pname)" );
-	    return;
-	 }
-	 break;
-      default:
-         _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(coord)" );
-	 return;
+   texgen = get_texgen(texUnit, coord);
+   if (!texgen) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexGeniv(coord)");
+      return;
+   }
+
+   switch (pname) {
+   case GL_TEXTURE_GEN_MODE:
+      params[0] = texgen->Mode;
+      break;
+   case GL_OBJECT_PLANE:
+      params[0] = (GLint) texgen->ObjectPlane[0];
+      params[1] = (GLint) texgen->ObjectPlane[1];
+      params[2] = (GLint) texgen->ObjectPlane[2];
+      params[3] = (GLint) texgen->ObjectPlane[3];
+      break;
+   case GL_EYE_PLANE:
+      params[0] = (GLint) texgen->EyePlane[0];
+      params[1] = (GLint) texgen->EyePlane[1];
+      params[2] = (GLint) texgen->EyePlane[2];
+      params[3] = (GLint) texgen->EyePlane[3];
+      break;
+   default:
+      _mesa_error( ctx, GL_INVALID_ENUM, "glGetTexGeniv(pname)" );
    }
 }
 
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index 6586a48646..053097cec0 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -85,23 +85,10 @@ _mesa_copy_texture_state( const GLcontext *src, GLcontext *dst )
       dst->Texture.Unit[i].EnvMode = src->Texture.Unit[i].EnvMode;
       COPY_4V(dst->Texture.Unit[i].EnvColor, src->Texture.Unit[i].EnvColor);
       dst->Texture.Unit[i].TexGenEnabled = src->Texture.Unit[i].TexGenEnabled;
-      dst->Texture.Unit[i].GenModeS = src->Texture.Unit[i].GenModeS;
-      dst->Texture.Unit[i].GenModeT = src->Texture.Unit[i].GenModeT;
-      dst->Texture.Unit[i].GenModeR = src->Texture.Unit[i].GenModeR;
-      dst->Texture.Unit[i].GenModeQ = src->Texture.Unit[i].GenModeQ;
-      dst->Texture.Unit[i]._GenBitS = src->Texture.Unit[i]._GenBitS;
-      dst->Texture.Unit[i]._GenBitT = src->Texture.Unit[i]._GenBitT;
-      dst->Texture.Unit[i]._GenBitR = src->Texture.Unit[i]._GenBitR;
-      dst->Texture.Unit[i]._GenBitQ = src->Texture.Unit[i]._GenBitQ;
-      dst->Texture.Unit[i]._GenFlags = src->Texture.Unit[i]._GenFlags;
-      COPY_4V(dst->Texture.Unit[i].ObjectPlaneS, src->Texture.Unit[i].ObjectPlaneS);
-      COPY_4V(dst->Texture.Unit[i].ObjectPlaneT, src->Texture.Unit[i].ObjectPlaneT);
-      COPY_4V(dst->Texture.Unit[i].ObjectPlaneR, src->Texture.Unit[i].ObjectPlaneR);
-      COPY_4V(dst->Texture.Unit[i].ObjectPlaneQ, src->Texture.Unit[i].ObjectPlaneQ);
-      COPY_4V(dst->Texture.Unit[i].EyePlaneS, src->Texture.Unit[i].EyePlaneS);
-      COPY_4V(dst->Texture.Unit[i].EyePlaneT, src->Texture.Unit[i].EyePlaneT);
-      COPY_4V(dst->Texture.Unit[i].EyePlaneR, src->Texture.Unit[i].EyePlaneR);
-      COPY_4V(dst->Texture.Unit[i].EyePlaneQ, src->Texture.Unit[i].EyePlaneQ);
+      dst->Texture.Unit[i].GenS = src->Texture.Unit[i].GenS;
+      dst->Texture.Unit[i].GenT = src->Texture.Unit[i].GenT;
+      dst->Texture.Unit[i].GenR = src->Texture.Unit[i].GenR;
+      dst->Texture.Unit[i].GenQ = src->Texture.Unit[i].GenQ;
       dst->Texture.Unit[i].LodBias = src->Texture.Unit[i].LodBias;
 
       /* GL_EXT_texture_env_combine */
@@ -655,16 +642,16 @@ update_texture_state( GLcontext *ctx )
 
       if (texUnit->TexGenEnabled) {
 	 if (texUnit->TexGenEnabled & S_BIT) {
-	    texUnit->_GenFlags |= texUnit->_GenBitS;
+	    texUnit->_GenFlags |= texUnit->GenS._ModeBit;
 	 }
 	 if (texUnit->TexGenEnabled & T_BIT) {
-	    texUnit->_GenFlags |= texUnit->_GenBitT;
-	 }
-	 if (texUnit->TexGenEnabled & Q_BIT) {
-	    texUnit->_GenFlags |= texUnit->_GenBitQ;
+	    texUnit->_GenFlags |= texUnit->GenT._ModeBit;
 	 }
 	 if (texUnit->TexGenEnabled & R_BIT) {
-	    texUnit->_GenFlags |= texUnit->_GenBitR;
+	    texUnit->_GenFlags |= texUnit->GenR._ModeBit;
+	 }
+	 if (texUnit->TexGenEnabled & Q_BIT) {
+	    texUnit->_GenFlags |= texUnit->GenQ._ModeBit;
 	 }
 
 	 ctx->Texture._TexGenEnabled |= ENABLE_TEXGEN(unit);
@@ -756,24 +743,24 @@ init_texture_unit( GLcontext *ctx, GLuint unit )
    texUnit->_CurrentCombine = & texUnit->_EnvMode;
 
    texUnit->TexGenEnabled = 0;
-   texUnit->GenModeS = GL_EYE_LINEAR;
-   texUnit->GenModeT = GL_EYE_LINEAR;
-   texUnit->GenModeR = GL_EYE_LINEAR;
-   texUnit->GenModeQ = GL_EYE_LINEAR;
-   texUnit->_GenBitS = TEXGEN_EYE_LINEAR;
-   texUnit->_GenBitT = TEXGEN_EYE_LINEAR;
-   texUnit->_GenBitR = TEXGEN_EYE_LINEAR;
-   texUnit->_GenBitQ = TEXGEN_EYE_LINEAR;
+   texUnit->GenS.Mode = GL_EYE_LINEAR;
+   texUnit->GenT.Mode = GL_EYE_LINEAR;
+   texUnit->GenR.Mode = GL_EYE_LINEAR;
+   texUnit->GenQ.Mode = GL_EYE_LINEAR;
+   texUnit->GenS._ModeBit = TEXGEN_EYE_LINEAR;
+   texUnit->GenT._ModeBit = TEXGEN_EYE_LINEAR;
+   texUnit->GenR._ModeBit = TEXGEN_EYE_LINEAR;
+   texUnit->GenQ._ModeBit = TEXGEN_EYE_LINEAR;
 
    /* Yes, these plane coefficients are correct! */
-   ASSIGN_4V( texUnit->ObjectPlaneS, 1.0, 0.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->ObjectPlaneT, 0.0, 1.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->ObjectPlaneR, 0.0, 0.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->ObjectPlaneQ, 0.0, 0.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->EyePlaneS, 1.0, 0.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->EyePlaneT, 0.0, 1.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->EyePlaneR, 0.0, 0.0, 0.0, 0.0 );
-   ASSIGN_4V( texUnit->EyePlaneQ, 0.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenS.ObjectPlane, 1.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenT.ObjectPlane, 0.0, 1.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenR.ObjectPlane, 0.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenQ.ObjectPlane, 0.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenS.EyePlane, 1.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenT.EyePlane, 0.0, 1.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenR.EyePlane, 0.0, 0.0, 0.0, 0.0 );
+   ASSIGN_4V( texUnit->GenQ.EyePlane, 0.0, 0.0, 0.0, 0.0 );
 
    /* initialize current texture object ptrs to the shared default objects */
    _mesa_reference_texobj(&texUnit->Current1D, ctx->Shared->Default1D);
diff --git a/src/mesa/shader/prog_statevars.c b/src/mesa/shader/prog_statevars.c
index e1db30b78f..1f7d87ccc9 100644
--- a/src/mesa/shader/prog_statevars.c
+++ b/src/mesa/shader/prog_statevars.c
@@ -206,28 +206,28 @@ _mesa_fetch_state(GLcontext *ctx, const gl_state_index state[],
          /* state[2] is the texgen attribute */
          switch (state[2]) {
          case STATE_TEXGEN_EYE_S:
-            COPY_4V(value, ctx->Texture.Unit[unit].EyePlaneS);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenS.EyePlane);
             return;
          case STATE_TEXGEN_EYE_T:
-            COPY_4V(value, ctx->Texture.Unit[unit].EyePlaneT);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenT.EyePlane);
             return;
          case STATE_TEXGEN_EYE_R:
-            COPY_4V(value, ctx->Texture.Unit[unit].EyePlaneR);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenR.EyePlane);
             return;
          case STATE_TEXGEN_EYE_Q:
-            COPY_4V(value, ctx->Texture.Unit[unit].EyePlaneQ);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenQ.EyePlane);
             return;
          case STATE_TEXGEN_OBJECT_S:
-            COPY_4V(value, ctx->Texture.Unit[unit].ObjectPlaneS);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenS.ObjectPlane);
             return;
          case STATE_TEXGEN_OBJECT_T:
-            COPY_4V(value, ctx->Texture.Unit[unit].ObjectPlaneT);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenT.ObjectPlane);
             return;
          case STATE_TEXGEN_OBJECT_R:
-            COPY_4V(value, ctx->Texture.Unit[unit].ObjectPlaneR);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenR.ObjectPlane);
             return;
          case STATE_TEXGEN_OBJECT_Q:
-            COPY_4V(value, ctx->Texture.Unit[unit].ObjectPlaneQ);
+            COPY_4V(value, ctx->Texture.Unit[unit].GenQ.ObjectPlane);
             return;
          default:
             _mesa_problem(ctx, "Invalid texgen state in fetch_state");
diff --git a/src/mesa/tnl/t_rasterpos.c b/src/mesa/tnl/t_rasterpos.c
index 14d79eab54..04fb1d8f8c 100644
--- a/src/mesa/tnl/t_rasterpos.c
+++ b/src/mesa/tnl/t_rasterpos.c
@@ -301,12 +301,12 @@ compute_texgen(GLcontext *ctx, const GLfloat vObj[4], const GLfloat vEye[4],
       mInv = 0.0F;
 
    if (texUnit->TexGenEnabled & S_BIT) {
-      switch (texUnit->GenModeS) {
+      switch (texUnit->GenS.Mode) {
          case GL_OBJECT_LINEAR:
-            texcoord[0] = DOT4(vObj, texUnit->ObjectPlaneS);
+            texcoord[0] = DOT4(vObj, texUnit->GenS.ObjectPlane);
             break;
          case GL_EYE_LINEAR:
-            texcoord[0] = DOT4(vEye, texUnit->EyePlaneS);
+            texcoord[0] = DOT4(vEye, texUnit->GenS.EyePlane);
             break;
          case GL_SPHERE_MAP:
             texcoord[0] = rx * mInv + 0.5F;
@@ -324,12 +324,12 @@ compute_texgen(GLcontext *ctx, const GLfloat vObj[4], const GLfloat vEye[4],
    }
 
    if (texUnit->TexGenEnabled & T_BIT) {
-      switch (texUnit->GenModeT) {
+      switch (texUnit->GenT.Mode) {
          case GL_OBJECT_LINEAR:
-            texcoord[1] = DOT4(vObj, texUnit->ObjectPlaneT);
+            texcoord[1] = DOT4(vObj, texUnit->GenT.ObjectPlane);
             break;
          case GL_EYE_LINEAR:
-            texcoord[1] = DOT4(vEye, texUnit->EyePlaneT);
+            texcoord[1] = DOT4(vEye, texUnit->GenT.EyePlane);
             break;
          case GL_SPHERE_MAP:
             texcoord[1] = ry * mInv + 0.5F;
@@ -347,12 +347,12 @@ compute_texgen(GLcontext *ctx, const GLfloat vObj[4], const GLfloat vEye[4],
    }
 
    if (texUnit->TexGenEnabled & R_BIT) {
-      switch (texUnit->GenModeR) {
+      switch (texUnit->GenR.Mode) {
          case GL_OBJECT_LINEAR:
-            texcoord[2] = DOT4(vObj, texUnit->ObjectPlaneR);
+            texcoord[2] = DOT4(vObj, texUnit->GenR.ObjectPlane);
             break;
          case GL_EYE_LINEAR:
-            texcoord[2] = DOT4(vEye, texUnit->EyePlaneR);
+            texcoord[2] = DOT4(vEye, texUnit->GenR.EyePlane);
             break;
          case GL_REFLECTION_MAP:
             texcoord[2] = rz;
@@ -367,12 +367,12 @@ compute_texgen(GLcontext *ctx, const GLfloat vObj[4], const GLfloat vEye[4],
    }
 
    if (texUnit->TexGenEnabled & Q_BIT) {
-      switch (texUnit->GenModeQ) {
+      switch (texUnit->GenQ.Mode) {
          case GL_OBJECT_LINEAR:
-            texcoord[3] = DOT4(vObj, texUnit->ObjectPlaneQ);
+            texcoord[3] = DOT4(vObj, texUnit->GenQ.ObjectPlane);
             break;
          case GL_EYE_LINEAR:
-            texcoord[3] = DOT4(vEye, texUnit->EyePlaneQ);
+            texcoord[3] = DOT4(vEye, texUnit->GenQ.EyePlane);
             break;
          default:
             _mesa_problem(ctx, "Bad Q texgen in compute_texgen()");
diff --git a/src/mesa/tnl/t_vb_texgen.c b/src/mesa/tnl/t_vb_texgen.c
index 14d3876e54..7c1819b223 100644
--- a/src/mesa/tnl/t_vb_texgen.c
+++ b/src/mesa/tnl/t_vb_texgen.c
@@ -367,16 +367,16 @@ static void texgen( GLcontext *ctx,
 
    if (texUnit->TexGenEnabled & S_BIT) {
       GLuint i;
-      switch (texUnit->GenModeS) {
+      switch (texUnit->GenS.Mode) {
       case GL_OBJECT_LINEAR:
 	 _mesa_dotprod_tab[obj->size]( (GLfloat *)out->data,
 				       sizeof(out->data[0]), obj,
-				       texUnit->ObjectPlaneS );
+				       texUnit->GenS.ObjectPlane );
 	 break;
       case GL_EYE_LINEAR:
 	 _mesa_dotprod_tab[eye->size]( (GLfloat *)out->data,
 				       sizeof(out->data[0]), eye,
-				       texUnit->EyePlaneS );
+				       texUnit->GenS.EyePlane );
 	 break;
       case GL_SPHERE_MAP:
          for (i = 0; i < count; i++)
@@ -400,16 +400,16 @@ static void texgen( GLcontext *ctx,
 
    if (texUnit->TexGenEnabled & T_BIT) {
       GLuint i;
-      switch (texUnit->GenModeT) {
+      switch (texUnit->GenT.Mode) {
       case GL_OBJECT_LINEAR:
 	 _mesa_dotprod_tab[obj->size]( &(out->data[0][1]),
 				       sizeof(out->data[0]), obj,
-				       texUnit->ObjectPlaneT );
+				       texUnit->GenT.ObjectPlane );
 	 break;
       case GL_EYE_LINEAR:
 	 _mesa_dotprod_tab[eye->size]( &(out->data[0][1]),
 				       sizeof(out->data[0]), eye,
-				       texUnit->EyePlaneT );
+				       texUnit->GenT.EyePlane );
 	 break;
       case GL_SPHERE_MAP:
          for (i = 0; i < count; i++)
@@ -433,16 +433,16 @@ static void texgen( GLcontext *ctx,
 
    if (texUnit->TexGenEnabled & R_BIT) {
       GLuint i;
-      switch (texUnit->GenModeR) {
+      switch (texUnit->GenR.Mode) {
       case GL_OBJECT_LINEAR:
 	 _mesa_dotprod_tab[obj->size]( &(out->data[0][2]),
 				       sizeof(out->data[0]), obj,
-				       texUnit->ObjectPlaneR );
+				       texUnit->GenR.ObjectPlane );
 	 break;
       case GL_EYE_LINEAR:
 	 _mesa_dotprod_tab[eye->size]( &(out->data[0][2]),
 				       sizeof(out->data[0]), eye,
-				       texUnit->EyePlaneR );
+				       texUnit->GenR.EyePlane );
 	 break;
       case GL_REFLECTION_MAP_NV:
 	 for (i=0;i<count;i++)
@@ -461,16 +461,16 @@ static void texgen( GLcontext *ctx,
    }
 
    if (texUnit->TexGenEnabled & Q_BIT) {
-      switch (texUnit->GenModeQ) {
+      switch (texUnit->GenQ.Mode) {
       case GL_OBJECT_LINEAR:
 	 _mesa_dotprod_tab[obj->size]( &(out->data[0][3]),
 				       sizeof(out->data[0]), obj,
-				       texUnit->ObjectPlaneQ );
+				       texUnit->GenQ.ObjectPlane );
 	 break;
       case GL_EYE_LINEAR:
 	 _mesa_dotprod_tab[eye->size]( &(out->data[0][3]),
 				       sizeof(out->data[0]), eye,
-				       texUnit->EyePlaneQ );
+				       texUnit->GenQ.EyePlane );
 	 break;
       default:
 	 _mesa_problem(ctx, "Bad Q texgen");
-- 
cgit v1.2.3


From 69e07bdeb42f2454f5052f86119adfb68f253098 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Sat, 7 Mar 2009 11:53:18 -0700
Subject: mesa: remove GL_MESA_program_debug extension

This was never fully fleshed out and hasn't been used.
---
 docs/MESA_shader_debug.spec              |   2 +-
 src/mesa/SConscript                      |   1 -
 src/mesa/drivers/common/driverfuncs.c    |   3 -
 src/mesa/drivers/dri/tdfx/tdfx_context.c |   4 -
 src/mesa/drivers/glide/fxdd.c            |   1 -
 src/mesa/main/dd.h                       |   3 -
 src/mesa/main/enable.c                   |  22 ---
 src/mesa/main/extensions.c               |   4 -
 src/mesa/main/ffvertex_prog.c            |   1 -
 src/mesa/main/get.c                      |  48 ------
 src/mesa/main/get_gen.py                 |  10 --
 src/mesa/main/getstring.c                |  30 ----
 src/mesa/main/mfeatures.h                |   1 -
 src/mesa/main/mtypes.h                   |  15 --
 src/mesa/shader/arbprogparse.c           |   9 --
 src/mesa/shader/nvfragparse.c            |   4 -
 src/mesa/shader/nvvertparse.c            |   7 -
 src/mesa/shader/prog_debug.c             | 255 -------------------------------
 src/mesa/shader/prog_debug.h             |  44 ------
 src/mesa/shader/prog_execute.c           |  41 -----
 src/mesa/shader/prog_instruction.h       |   5 -
 src/mesa/sources.mak                     |   1 -
 src/mesa/swrast/swrast.h                 |   6 -
 23 files changed, 1 insertion(+), 516 deletions(-)
 delete mode 100644 src/mesa/shader/prog_debug.c
 delete mode 100644 src/mesa/shader/prog_debug.h

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/docs/MESA_shader_debug.spec b/docs/MESA_shader_debug.spec
index 1f7d42ac91..fab92abc76 100644
--- a/docs/MESA_shader_debug.spec
+++ b/docs/MESA_shader_debug.spec
@@ -13,7 +13,7 @@ Contact
 
 Status
 
-    XXX - Not complete yet!!!
+    Obsolete.
 
 Version
 
diff --git a/src/mesa/SConscript b/src/mesa/SConscript
index 273ce11280..2e0168e0bc 100644
--- a/src/mesa/SConscript
+++ b/src/mesa/SConscript
@@ -189,7 +189,6 @@ if env['platform'] != 'winddk':
 		'shader/nvvertparse.c',
 		'shader/program.c',
 		'shader/prog_cache.c',
-		'shader/prog_debug.c',
 		'shader/prog_execute.c',
 		'shader/prog_instruction.c',
 		'shader/prog_noise.c',
diff --git a/src/mesa/drivers/common/driverfuncs.c b/src/mesa/drivers/common/driverfuncs.c
index 9bee212c6a..44adaf8682 100644
--- a/src/mesa/drivers/common/driverfuncs.c
+++ b/src/mesa/drivers/common/driverfuncs.c
@@ -134,9 +134,6 @@ _mesa_init_driver_functions(struct dd_function_table *driver)
    driver->BindProgram = NULL;
    driver->NewProgram = _mesa_new_program;
    driver->DeleteProgram = _mesa_delete_program;
-#if FEATURE_MESA_program_debug
-   driver->GetProgramRegister = _mesa_get_program_register;
-#endif /* FEATURE_MESA_program_debug */
 
    /* simple state commands */
    driver->AlphaFunc = NULL;
diff --git a/src/mesa/drivers/dri/tdfx/tdfx_context.c b/src/mesa/drivers/dri/tdfx/tdfx_context.c
index 20046fcb3a..68b5027561 100644
--- a/src/mesa/drivers/dri/tdfx/tdfx_context.c
+++ b/src/mesa/drivers/dri/tdfx/tdfx_context.c
@@ -67,7 +67,6 @@
 #define need_GL_EXT_fog_coord
 #define need_GL_EXT_paletted_texture
 /* #define need_GL_EXT_secondary_color */
-/* #define need_GL_MESA_program_debug */
 /* #define need_GL_NV_vertex_program */
 #include "extension_helper.h"
 
@@ -101,9 +100,6 @@ const struct dri_extension card_extensions[] =
 #ifdef need_GL_NV_vertex_program
     { "GL_NV_vertex_program",              GL_NV_vertex_program_functions }
     { "GL_NV_vertex_program1_1",           NULL },
-#endif
-#ifdef need_GL_MESA_program_debug
-    { "GL_MESA_program_debug",             GL_MESA_program_debug_functions },
 #endif
     { NULL,                                NULL }
 };
diff --git a/src/mesa/drivers/glide/fxdd.c b/src/mesa/drivers/glide/fxdd.c
index 1bcf2512a6..2bc60399ea 100644
--- a/src/mesa/drivers/glide/fxdd.c
+++ b/src/mesa/drivers/glide/fxdd.c
@@ -1922,7 +1922,6 @@ fxDDInitExtensions(GLcontext * ctx)
       _mesa_enable_extension(ctx, "GL_ARB_vertex_program");
       _mesa_enable_extension(ctx, "GL_NV_vertex_program");
       _mesa_enable_extension(ctx, "GL_NV_vertex_program1_1");
-      _mesa_enable_extension(ctx, "GL_MESA_program_debug");
    }
 #if 0
    /* this requires _tnl_vertex_cull_stage in the pipeline */
diff --git a/src/mesa/main/dd.h b/src/mesa/main/dd.h
index a335f77479..d994401e55 100644
--- a/src/mesa/main/dd.h
+++ b/src/mesa/main/dd.h
@@ -602,9 +602,6 @@ struct dd_function_table {
    /** Notify driver that a program string has been specified. */
    void (*ProgramStringNotify)(GLcontext *ctx, GLenum target, 
 			       struct gl_program *prog);
-   /** Get value of a program register during program execution. */
-   void (*GetProgramRegister)(GLcontext *ctx, enum register_file file,
-                              GLuint index, GLfloat val[4]);
 
    /** Query if program can be loaded onto hardware */
    GLboolean (*IsProgramNative)(GLcontext *ctx, GLenum target, 
diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c
index a824705bdc..f432be183c 100644
--- a/src/mesa/main/enable.c
+++ b/src/mesa/main/enable.c
@@ -944,18 +944,6 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
          ctx->Depth.BoundsTest = state;
          break;
 
-      /* GL_MESA_program_debug */
-#if FEATURE_MESA_program_debug
-      case GL_FRAGMENT_PROGRAM_CALLBACK_MESA:
-         CHECK_EXTENSION(MESA_program_debug, cap);
-         ctx->FragmentProgram.CallbackEnabled = state;
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_MESA:
-         CHECK_EXTENSION(MESA_program_debug, cap);
-         ctx->VertexProgram.CallbackEnabled = state;
-         break;
-#endif
-
 #if FEATURE_ATI_fragment_shader
       case GL_FRAGMENT_SHADER_ATI:
         CHECK_EXTENSION(ATI_fragment_shader, cap);
@@ -1398,16 +1386,6 @@ _mesa_IsEnabled( GLenum cap )
          CHECK_EXTENSION(EXT_depth_bounds_test);
          return ctx->Depth.BoundsTest;
 
-      /* GL_MESA_program_debug */
-#if FEATURE_MESA_program_debug
-      case GL_FRAGMENT_PROGRAM_CALLBACK_MESA:
-         CHECK_EXTENSION(MESA_program_debug);
-         return ctx->FragmentProgram.CallbackEnabled;
-      case GL_VERTEX_PROGRAM_CALLBACK_MESA:
-         CHECK_EXTENSION(MESA_program_debug);
-         return ctx->VertexProgram.CallbackEnabled;
-#endif
-
 #if FEATURE_ATI_fragment_shader
       case GL_FRAGMENT_SHADER_ATI:
 	 CHECK_EXTENSION(ATI_fragment_shader);
diff --git a/src/mesa/main/extensions.c b/src/mesa/main/extensions.c
index 9c8bd13e69..fbca924ad3 100644
--- a/src/mesa/main/extensions.c
+++ b/src/mesa/main/extensions.c
@@ -147,7 +147,6 @@ static const struct {
    { OFF, "GL_INGR_blend_func_separate",       F(EXT_blend_func_separate) },
    { OFF, "GL_MESA_pack_invert",               F(MESA_pack_invert) },
    { OFF, "GL_MESA_packed_depth_stencil",      F(MESA_packed_depth_stencil) },
-   { OFF, "GL_MESA_program_debug",             F(MESA_program_debug) },
    { OFF, "GL_MESA_resize_buffers",            F(MESA_resize_buffers) },
    { OFF, "GL_MESA_texture_array",             F(MESA_texture_array) },
    { OFF, "GL_MESA_ycbcr_texture",             F(MESA_ycbcr_texture) },
@@ -276,9 +275,6 @@ _mesa_enable_sw_extensions(GLcontext *ctx)
    ctx->Extensions.EXT_vertex_array_bgra = GL_TRUE;
    /*ctx->Extensions.IBM_multimode_draw_arrays = GL_TRUE;*/
    ctx->Extensions.MESA_pack_invert = GL_TRUE;
-#if FEATURE_MESA_program_debug
-   ctx->Extensions.MESA_program_debug = GL_TRUE;
-#endif
    ctx->Extensions.MESA_resize_buffers = GL_TRUE;
    ctx->Extensions.MESA_texture_array = GL_TRUE;
    ctx->Extensions.MESA_ycbcr_texture = GL_TRUE;
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 42c8cc97c0..72b880e28e 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -655,7 +655,6 @@ static void emit_op3fn(struct tnl_program *p,
 
    inst = &p->program->Base.Instructions[nr];
    inst->Opcode = (enum prog_opcode) op; 
-   inst->StringPos = 0;
    inst->Data = 0;
    
    emit_arg( &inst->SrcReg[0], src0 );
diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c
index 3a8d56140f..0937fd053c 100644
--- a/src/mesa/main/get.c
+++ b/src/mesa/main/get.c
@@ -1715,22 +1715,6 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params )
          params[0] = FLOAT_TO_BOOLEAN(ctx->Depth.BoundsMin);
          params[1] = FLOAT_TO_BOOLEAN(ctx->Depth.BoundsMax);
          break;
-      case GL_FRAGMENT_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetBooleanv");
-         params[0] = ctx->FragmentProgram.CallbackEnabled;
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetBooleanv");
-         params[0] = ctx->VertexProgram.CallbackEnabled;
-         break;
-      case GL_FRAGMENT_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetBooleanv");
-         params[0] = INT_TO_BOOLEAN(ctx->FragmentProgram.CurrentPosition);
-         break;
-      case GL_VERTEX_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetBooleanv");
-         params[0] = INT_TO_BOOLEAN(ctx->VertexProgram.CurrentPosition);
-         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = INT_TO_BOOLEAN(ctx->Const.MaxDrawBuffers);
          break;
@@ -3541,22 +3525,6 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params )
          params[0] = ctx->Depth.BoundsMin;
          params[1] = ctx->Depth.BoundsMax;
          break;
-      case GL_FRAGMENT_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetFloatv");
-         params[0] = BOOLEAN_TO_FLOAT(ctx->FragmentProgram.CallbackEnabled);
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetFloatv");
-         params[0] = BOOLEAN_TO_FLOAT(ctx->VertexProgram.CallbackEnabled);
-         break;
-      case GL_FRAGMENT_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetFloatv");
-         params[0] = (GLfloat)(ctx->FragmentProgram.CurrentPosition);
-         break;
-      case GL_VERTEX_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetFloatv");
-         params[0] = (GLfloat)(ctx->VertexProgram.CurrentPosition);
-         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = (GLfloat)(ctx->Const.MaxDrawBuffers);
          break;
@@ -5367,22 +5335,6 @@ _mesa_GetIntegerv( GLenum pname, GLint *params )
          params[0] = IROUND(ctx->Depth.BoundsMin);
          params[1] = IROUND(ctx->Depth.BoundsMax);
          break;
-      case GL_FRAGMENT_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetIntegerv");
-         params[0] = BOOLEAN_TO_INT(ctx->FragmentProgram.CallbackEnabled);
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetIntegerv");
-         params[0] = BOOLEAN_TO_INT(ctx->VertexProgram.CallbackEnabled);
-         break;
-      case GL_FRAGMENT_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetIntegerv");
-         params[0] = ctx->FragmentProgram.CurrentPosition;
-         break;
-      case GL_VERTEX_PROGRAM_POSITION_MESA:
-         CHECK_EXT1(MESA_program_debug, "GetIntegerv");
-         params[0] = ctx->VertexProgram.CurrentPosition;
-         break;
       case GL_MAX_DRAW_BUFFERS_ARB:
          params[0] = ctx->Const.MaxDrawBuffers;
          break;
diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py
index 729f382b4b..fa695c48f1 100644
--- a/src/mesa/main/get_gen.py
+++ b/src/mesa/main/get_gen.py
@@ -902,16 +902,6 @@ StateVars = [
 	  ["ctx->Depth.BoundsMin", "ctx->Depth.BoundsMax"],
 	  "", ["EXT_depth_bounds_test"] ),
 
-	# GL_MESA_program_debug
-	( "GL_FRAGMENT_PROGRAM_CALLBACK_MESA", GLboolean,
-	  ["ctx->FragmentProgram.CallbackEnabled"], "", ["MESA_program_debug"] ),
-	( "GL_VERTEX_PROGRAM_CALLBACK_MESA", GLboolean,
-	  ["ctx->VertexProgram.CallbackEnabled"], "", ["MESA_program_debug"] ),
-	( "GL_FRAGMENT_PROGRAM_POSITION_MESA", GLint,
-	  ["ctx->FragmentProgram.CurrentPosition"], "", ["MESA_program_debug"] ),
-	( "GL_VERTEX_PROGRAM_POSITION_MESA", GLint,
-	  ["ctx->VertexProgram.CurrentPosition"], "", ["MESA_program_debug"] ),
-
 	# GL_ARB_draw_buffers
 	( "GL_MAX_DRAW_BUFFERS_ARB", GLint,
 	  ["ctx->Const.MaxDrawBuffers"], "", None ),
diff --git a/src/mesa/main/getstring.c b/src/mesa/main/getstring.c
index a42c44686d..41fd786d7d 100644
--- a/src/mesa/main/getstring.c
+++ b/src/mesa/main/getstring.c
@@ -242,36 +242,6 @@ _mesa_GetPointerv( GLenum pname, GLvoid **params )
       case GL_SELECTION_BUFFER_POINTER:
          *params = ctx->Select.Buffer;
          break;
-#if FEATURE_MESA_program_debug
-      case GL_FRAGMENT_PROGRAM_CALLBACK_FUNC_MESA:
-         if (!ctx->Extensions.MESA_program_debug) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetPointerv");
-            return;
-         }
-         *params = *(GLvoid **) &ctx->FragmentProgram.Callback;
-         break;
-      case GL_FRAGMENT_PROGRAM_CALLBACK_DATA_MESA:
-         if (!ctx->Extensions.MESA_program_debug) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetPointerv");
-            return;
-         }
-         *params = ctx->FragmentProgram.CallbackData;
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_FUNC_MESA:
-         if (!ctx->Extensions.MESA_program_debug) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetPointerv");
-            return;
-         }
-         *params = *(GLvoid **) &ctx->VertexProgram.Callback;
-         break;
-      case GL_VERTEX_PROGRAM_CALLBACK_DATA_MESA:
-         if (!ctx->Extensions.MESA_program_debug) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetPointerv");
-            return;
-         }
-         *params = ctx->VertexProgram.CallbackData;
-         break;
-#endif
       default:
          _mesa_error( ctx, GL_INVALID_ENUM, "glGetPointerv" );
          return;
diff --git a/src/mesa/main/mfeatures.h b/src/mesa/main/mfeatures.h
index 8fb32dd7e9..f570647942 100644
--- a/src/mesa/main/mfeatures.h
+++ b/src/mesa/main/mfeatures.h
@@ -75,7 +75,6 @@
 #define FEATURE_EXT_texture_sRGB _HAVE_FULL_GL
 #define FEATURE_EXT_timer_query  _HAVE_FULL_GL
 #define FEATURE_ATI_fragment_shader _HAVE_FULL_GL
-#define FEATURE_MESA_program_debug  _HAVE_FULL_GL
 #define FEATURE_NV_fence  _HAVE_FULL_GL
 #define FEATURE_NV_fragment_program  _HAVE_FULL_GL
 #define FEATURE_NV_vertex_program  _HAVE_FULL_GL
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index baf5850b83..17f6241e04 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1856,13 +1856,6 @@ struct gl_vertex_program_state
    /** Cache of fixed-function programs */
    struct gl_program_cache *Cache;
 
-#if FEATURE_MESA_program_debug
-   GLprogramcallbackMESA Callback;
-   GLvoid *CallbackData;
-   GLboolean CallbackEnabled;
-   GLuint CurrentPosition;
-#endif
-
    GLboolean _Overriden;
 };
 
@@ -1892,13 +1885,6 @@ struct gl_fragment_program_state
 
    /** Cache of fixed-function programs */
    struct gl_program_cache *Cache;
-
-#if FEATURE_MESA_program_debug
-   GLprogramcallbackMESA Callback;
-   GLvoid *CallbackData;
-   GLboolean CallbackEnabled;
-   GLuint CurrentPosition;
-#endif
 };
 
 
@@ -2502,7 +2488,6 @@ struct gl_extensions
    GLboolean IBM_multimode_draw_arrays;
    GLboolean MESA_pack_invert;
    GLboolean MESA_packed_depth_stencil;
-   GLboolean MESA_program_debug;
    GLboolean MESA_resize_buffers;
    GLboolean MESA_ycbcr_texture;
    GLboolean MESA_texture_array;
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index ccc0318a53..75398cda90 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -2752,9 +2752,6 @@ parse_fp_instruction (GLcontext * ctx, const GLubyte ** inst,
 
    _mesa_init_instructions(fp, 1);
 
-   /* Record the position in the program string for debugging */
-   fp->StringPos = Program->Position;
-
    /* OP_ALU_INST or OP_TEX_INST */
    instClass = *(*inst)++;
 
@@ -3194,8 +3191,6 @@ parse_vp_instruction (GLcontext * ctx, const GLubyte ** inst,
    code = *(*inst)++;
 
    _mesa_init_instructions(vp, 1);
-   /* Record the position in the program string for debugging */
-   vp->StringPos = Program->Position;
 
    switch (type) {
          /* XXX: */
@@ -3557,10 +3552,6 @@ parse_instructions(GLcontext * ctx, const GLubyte * inst,
       const GLuint numInst = Program->Base.NumInstructions;
       _mesa_init_instructions(Program->Base.Instructions + numInst, 1);
       Program->Base.Instructions[numInst].Opcode = OPCODE_END;
-      /* YYY Wrong Position in program, whatever, at least not random -> crash
-	 Program->Position = parse_position (&inst);
-      */
-      Program->Base.Instructions[numInst].StringPos = Program->Position;
    }
    Program->Base.NumInstructions++;
 
diff --git a/src/mesa/shader/nvfragparse.c b/src/mesa/shader/nvfragparse.c
index 37418ffb6e..b935cb562a 100644
--- a/src/mesa/shader/nvfragparse.c
+++ b/src/mesa/shader/nvfragparse.c
@@ -1311,8 +1311,6 @@ Parse_InstructionSequence(struct parse_state *parseState,
       }
       else if (Parse_String(parseState, "END")) {
          inst->Opcode = OPCODE_END;
-         inst->StringPos = parseState->curLine - parseState->start;
-         assert(inst->StringPos >= 0);
          parseState->numInst++;
          if (Parse_Token(parseState, token)) {
             RETURN_ERROR1("Code after END opcode.");
@@ -1339,8 +1337,6 @@ Parse_InstructionSequence(struct parse_state *parseState,
          inst->SaturateMode = (instMatch.suffixes & (_S))
             ? SATURATE_ZERO_ONE : SATURATE_OFF;
          inst->CondUpdate = (instMatch.suffixes & (_C)) ? GL_TRUE : GL_FALSE;
-         inst->StringPos = parseState->curLine - parseState->start;
-         assert(inst->StringPos >= 0);
 
          /*
           * parse the input and output operands
diff --git a/src/mesa/shader/nvvertparse.c b/src/mesa/shader/nvvertparse.c
index 08538c0ee4..268b577aec 100644
--- a/src/mesa/shader/nvvertparse.c
+++ b/src/mesa/shader/nvvertparse.c
@@ -799,7 +799,6 @@ Parse_UnaryOpInstruction(struct parse_state *parseState,
       RETURN_ERROR1("ABS illegal for vertex program 1.0");
 
    inst->Opcode = opcode;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* dest reg */
    if (!Parse_MaskedDstReg(parseState, &inst->DstReg))
@@ -832,7 +831,6 @@ Parse_BiOpInstruction(struct parse_state *parseState,
       RETURN_ERROR1("SUB illegal for vertex program 1.0");
 
    inst->Opcode = opcode;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* dest reg */
    if (!Parse_MaskedDstReg(parseState, &inst->DstReg))
@@ -880,7 +878,6 @@ Parse_TriOpInstruction(struct parse_state *parseState,
                        enum prog_opcode opcode)
 {
    inst->Opcode = opcode;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* dest reg */
    if (!Parse_MaskedDstReg(parseState, &inst->DstReg))
@@ -951,7 +948,6 @@ Parse_ScalarInstruction(struct parse_state *parseState,
       RETURN_ERROR1("RCC illegal for vertex program 1.0");
 
    inst->Opcode = opcode;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* dest reg */
    if (!Parse_MaskedDstReg(parseState, &inst->DstReg))
@@ -977,7 +973,6 @@ static GLboolean
 Parse_AddressInstruction(struct parse_state *parseState, struct prog_instruction *inst)
 {
    inst->Opcode = OPCODE_ARL;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* Make ARB_vp backends happy */
    inst->DstReg.File = PROGRAM_ADDRESS;
@@ -1010,7 +1005,6 @@ Parse_EndInstruction(struct parse_state *parseState, struct prog_instruction *in
    GLubyte token[100];
 
    inst->Opcode = OPCODE_END;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* this should fail! */
    if (Parse_Token(parseState, token))
@@ -1044,7 +1038,6 @@ Parse_PrintInstruction(struct parse_state *parseState, struct prog_instruction *
    GLint idx;
 
    inst->Opcode = OPCODE_PRINT;
-   inst->StringPos = parseState->curLine - parseState->start;
 
    /* The first argument is a literal string 'just like this' */
    if (!Parse_String(parseState, "'"))
diff --git a/src/mesa/shader/prog_debug.c b/src/mesa/shader/prog_debug.c
deleted file mode 100644
index 35ce37d5ce..0000000000
--- a/src/mesa/shader/prog_debug.c
+++ /dev/null
@@ -1,255 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#include "main/glheader.h"
-#include "main/context.h"
-#include "main/macros.h"
-#include "nvfragparse.h"
-#include "nvvertparse.h"
-#include "program.h"
-#include "prog_debug.h"
-#include "prog_parameter.h"
-#include "prog_instruction.h"
-
-
-
-/**
- * Functions for the experimental GL_MESA_program_debug extension.
- */
-
-
-/* XXX temporary */
-GLAPI void GLAPIENTRY
-glProgramCallbackMESA(GLenum target, GLprogramcallbackMESA callback,
-                      GLvoid *data)
-{
-   _mesa_ProgramCallbackMESA(target, callback, data);
-}
-
-
-void
-_mesa_ProgramCallbackMESA(GLenum target, GLprogramcallbackMESA callback,
-                          GLvoid *data)
-{
-   GET_CURRENT_CONTEXT(ctx);
-
-   switch (target) {
-      case GL_FRAGMENT_PROGRAM_ARB:
-         if (!ctx->Extensions.ARB_fragment_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glProgramCallbackMESA(target)");
-            return;
-         }
-         ctx->FragmentProgram.Callback = callback;
-         ctx->FragmentProgram.CallbackData = data;
-         break;
-      case GL_FRAGMENT_PROGRAM_NV:
-         if (!ctx->Extensions.NV_fragment_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glProgramCallbackMESA(target)");
-            return;
-         }
-         ctx->FragmentProgram.Callback = callback;
-         ctx->FragmentProgram.CallbackData = data;
-         break;
-      case GL_VERTEX_PROGRAM_ARB: /* == GL_VERTEX_PROGRAM_NV */
-         if (!ctx->Extensions.ARB_vertex_program &&
-             !ctx->Extensions.NV_vertex_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM, "glProgramCallbackMESA(target)");
-            return;
-         }
-         ctx->VertexProgram.Callback = callback;
-         ctx->VertexProgram.CallbackData = data;
-         break;
-      default:
-         _mesa_error(ctx, GL_INVALID_ENUM, "glProgramCallbackMESA(target)");
-         return;
-   }
-}
-
-
-/* XXX temporary */
-GLAPI void GLAPIENTRY
-glGetProgramRegisterfvMESA(GLenum target,
-                           GLsizei len, const GLubyte *registerName,
-                           GLfloat *v)
-{
-   _mesa_GetProgramRegisterfvMESA(target, len, registerName, v);
-}
-
-
-void
-_mesa_GetProgramRegisterfvMESA(GLenum target,
-                               GLsizei len, const GLubyte *registerName,
-                               GLfloat *v)
-{
-   char reg[1000];
-   GET_CURRENT_CONTEXT(ctx);
-
-   /* We _should_ be inside glBegin/glEnd */
-#if 0
-   if (ctx->Driver.CurrentExecPrimitive == PRIM_OUTSIDE_BEGIN_END) {
-      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramRegisterfvMESA");
-      return;
-   }
-#endif
-
-   /* make null-terminated copy of registerName */
-   len = MIN2((unsigned int) len, sizeof(reg) - 1);
-   _mesa_memcpy(reg, registerName, len);
-   reg[len] = 0;
-
-   switch (target) {
-      case GL_VERTEX_PROGRAM_ARB: /* == GL_VERTEX_PROGRAM_NV */
-         if (!ctx->Extensions.ARB_vertex_program &&
-             !ctx->Extensions.NV_vertex_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetProgramRegisterfvMESA(target)");
-            return;
-         }
-         if (!ctx->VertexProgram._Enabled) {
-            _mesa_error(ctx, GL_INVALID_OPERATION,
-                        "glGetProgramRegisterfvMESA");
-            return;
-         }
-         /* GL_NV_vertex_program */
-         if (reg[0] == 'R') {
-            /* Temp register */
-            GLint i = _mesa_atoi(reg + 1);
-            if (i >= (GLint)ctx->Const.VertexProgram.MaxTemps) {
-               _mesa_error(ctx, GL_INVALID_VALUE,
-                           "glGetProgramRegisterfvMESA(registerName)");
-               return;
-            }
-            ctx->Driver.GetProgramRegister(ctx, PROGRAM_TEMPORARY, i, v);
-         }
-         else if (reg[0] == 'v' && reg[1] == '[') {
-            /* Vertex Input attribute */
-            GLuint i;
-            for (i = 0; i < ctx->Const.VertexProgram.MaxAttribs; i++) {
-               const char *name = _mesa_nv_vertex_input_register_name(i);
-               char number[10];
-               _mesa_sprintf(number, "%d", i);
-               if (_mesa_strncmp(reg + 2, name, 4) == 0 ||
-                   _mesa_strncmp(reg + 2, number, _mesa_strlen(number)) == 0) {
-                  ctx->Driver.GetProgramRegister(ctx, PROGRAM_INPUT, i, v);
-                  return;
-               }
-            }
-            _mesa_error(ctx, GL_INVALID_VALUE,
-                        "glGetProgramRegisterfvMESA(registerName)");
-            return;
-         }
-         else if (reg[0] == 'o' && reg[1] == '[') {
-            /* Vertex output attribute */
-         }
-         /* GL_ARB_vertex_program */
-         else if (_mesa_strncmp(reg, "vertex.", 7) == 0) {
-
-         }
-         else {
-            _mesa_error(ctx, GL_INVALID_VALUE,
-                        "glGetProgramRegisterfvMESA(registerName)");
-            return;
-         }
-         break;
-      case GL_FRAGMENT_PROGRAM_ARB:
-         if (!ctx->Extensions.ARB_fragment_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetProgramRegisterfvMESA(target)");
-            return;
-         }
-         if (!ctx->FragmentProgram._Enabled) {
-            _mesa_error(ctx, GL_INVALID_OPERATION,
-                        "glGetProgramRegisterfvMESA");
-            return;
-         }
-         /* XXX to do */
-         break;
-      case GL_FRAGMENT_PROGRAM_NV:
-         if (!ctx->Extensions.NV_fragment_program) {
-            _mesa_error(ctx, GL_INVALID_ENUM,
-                        "glGetProgramRegisterfvMESA(target)");
-            return;
-         }
-         if (!ctx->FragmentProgram._Enabled) {
-            _mesa_error(ctx, GL_INVALID_OPERATION,
-                        "glGetProgramRegisterfvMESA");
-            return;
-         }
-         if (reg[0] == 'R') {
-            /* Temp register */
-            GLint i = _mesa_atoi(reg + 1);
-            if (i >= (GLint)ctx->Const.FragmentProgram.MaxTemps) {
-               _mesa_error(ctx, GL_INVALID_VALUE,
-                           "glGetProgramRegisterfvMESA(registerName)");
-               return;
-            }
-            ctx->Driver.GetProgramRegister(ctx, PROGRAM_TEMPORARY,
-                                                   i, v);
-         }
-         else if (reg[0] == 'f' && reg[1] == '[') {
-            /* Fragment input attribute */
-            GLuint i;
-            for (i = 0; i < ctx->Const.FragmentProgram.MaxAttribs; i++) {
-               const char *name = _mesa_nv_fragment_input_register_name(i);
-               if (_mesa_strncmp(reg + 2, name, 4) == 0) {
-                  ctx->Driver.GetProgramRegister(ctx, PROGRAM_INPUT, i, v);
-                  return;
-               }
-            }
-            _mesa_error(ctx, GL_INVALID_VALUE,
-                        "glGetProgramRegisterfvMESA(registerName)");
-            return;
-         }
-         else if (_mesa_strcmp(reg, "o[COLR]") == 0 ||
-                  _mesa_strcmp(reg, "o[COLH]") == 0) {
-            /* Fragment output color */
-            ctx->Driver.GetProgramRegister(ctx, PROGRAM_OUTPUT,
-                                           FRAG_RESULT_COLOR, v);
-         }
-         else if (_mesa_strcmp(reg, "o[DEPR]") == 0) {
-            /* Fragment output depth */
-            ctx->Driver.GetProgramRegister(ctx, PROGRAM_OUTPUT,
-                                           FRAG_RESULT_DEPTH, v);
-         }
-         else {
-            /* try user-defined identifiers */
-            const GLfloat *value = _mesa_lookup_parameter_value(
-                       ctx->FragmentProgram.Current->Base.Parameters, -1, reg);
-            if (value) {
-               COPY_4V(v, value);
-            }
-            else {
-               _mesa_error(ctx, GL_INVALID_VALUE,
-                           "glGetProgramRegisterfvMESA(registerName)");
-               return;
-            }
-         }
-         break;
-      default:
-         _mesa_error(ctx, GL_INVALID_ENUM,
-                     "glGetProgramRegisterfvMESA(target)");
-         return;
-   }
-}
diff --git a/src/mesa/shader/prog_debug.h b/src/mesa/shader/prog_debug.h
deleted file mode 100644
index fc400e19de..0000000000
--- a/src/mesa/shader/prog_debug.h
+++ /dev/null
@@ -1,44 +0,0 @@
-/*
- * Mesa 3-D graphics library
- * Version:  6.5.3
- *
- * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice shall be included
- * in all copies or substantial portions of the Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
- * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- */
-
-
-#ifndef PROG_DEBUG_H
-#define PROG_DEBUG_H 1
-
-
-/*
- * GL_MESA_program_debug
- */
-
-extern void
-_mesa_ProgramCallbackMESA(GLenum target, GLprogramcallbackMESA callback,
-                          GLvoid *data);
-
-extern void
-_mesa_GetProgramRegisterfvMESA(GLenum target, GLsizei len,
-                               const GLubyte *registerName, GLfloat *v);
-
-
-
-#endif /* PROG_DEBUG_H */
diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c
index a93733c085..a60cda674b 100644
--- a/src/mesa/shader/prog_execute.c
+++ b/src/mesa/shader/prog_execute.c
@@ -186,30 +186,6 @@ get_dst_register_pointer(const struct prog_dst_register *dest,
 
 
 
-#if FEATURE_MESA_program_debug
-static struct gl_program_machine *CurrentMachine = NULL;
-
-/**
- * For GL_MESA_program_debug.
- * Return current value (4*GLfloat) of a program register.
- * Called via ctx->Driver.GetProgramRegister().
- */
-void
-_mesa_get_program_register(GLcontext *ctx, enum register_file file,
-                           GLuint index, GLfloat val[4])
-{
-   if (CurrentMachine) {
-      struct prog_src_register srcReg;
-      const GLfloat *src;
-      srcReg.File = file;
-      srcReg.Index = index;
-      src = get_src_register_pointer(&srcReg, CurrentMachine);
-      COPY_4V(val, src);
-   }
-}
-#endif /* FEATURE_MESA_program_debug */
-
-
 /**
  * Fetch a 4-element float vector from the given source register.
  * Apply swizzling and negating as needed.
@@ -633,10 +609,6 @@ _mesa_execute_program(GLcontext * ctx,
       printf("execute program %u --------------------\n", program->Id);
    }
 
-#if FEATURE_MESA_program_debug
-   CurrentMachine = machine;
-#endif
-
    if (program->Target == GL_VERTEX_PROGRAM_ARB) {
       machine->EnvParams = ctx->VertexProgram.Parameters;
    }
@@ -647,15 +619,6 @@ _mesa_execute_program(GLcontext * ctx,
    for (pc = 0; pc < numInst; pc++) {
       const struct prog_instruction *inst = program->Instructions + pc;
 
-#if FEATURE_MESA_program_debug
-      if (ctx->FragmentProgram.CallbackEnabled &&
-          ctx->FragmentProgram.Callback) {
-         ctx->FragmentProgram.CurrentPosition = inst->StringPos;
-         ctx->FragmentProgram.Callback(program->Target,
-                                       ctx->FragmentProgram.CallbackData);
-      }
-#endif
-
       if (DEBUG_PROG) {
          _mesa_print_instruction(inst);
       }
@@ -1780,9 +1743,5 @@ _mesa_execute_program(GLcontext * ctx,
 
    } /* for pc */
 
-#if FEATURE_MESA_program_debug
-   CurrentMachine = NULL;
-#endif
-
    return GL_TRUE;
 }
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index 95dd7fda7f..4adce11f95 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -421,11 +421,6 @@ struct prog_instruction
 
    /** for driver use (try to remove someday) */
    GLint Aux;
-
-   /* XXX obsolete - remove someday */
-#if FEATURE_MESA_program_debug
-   GLshort StringPos;
-#endif
 };
 
 
diff --git a/src/mesa/sources.mak b/src/mesa/sources.mak
index 08ba345009..d8d48476e8 100644
--- a/src/mesa/sources.mak
+++ b/src/mesa/sources.mak
@@ -220,7 +220,6 @@ SHADER_SOURCES = \
 	shader/nvvertparse.c \
 	shader/program.c \
 	shader/prog_cache.c \
-	shader/prog_debug.c \
 	shader/prog_execute.c \
 	shader/prog_instruction.c \
 	shader/prog_noise.c \
diff --git a/src/mesa/swrast/swrast.h b/src/mesa/swrast/swrast.h
index 015f8a05c3..c319ca62f9 100644
--- a/src/mesa/swrast/swrast.h
+++ b/src/mesa/swrast/swrast.h
@@ -266,12 +266,6 @@ extern void
 _swrast_eject_texture_images(GLcontext *ctx);
 
 
-#if FEATURE_MESA_program_debug
-extern void
-_swrast_get_program_register(GLcontext *, enum register_file file,
-                             GLuint index, GLfloat val[4]);
-#endif /* FEATURE_MESA_program_debug */
-
 
 /**
  * The driver interface for the software rasterizer.
-- 
cgit v1.2.3


From 53d507562b252718d95fd99812f5d21c9ae79713 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Tue, 31 Mar 2009 15:47:05 -0600
Subject: mesa: minor reformatting, whitespace changes

---
 src/mesa/main/ffvertex_prog.c | 51 ++++++++++++++++++++++---------------------
 1 file changed, 26 insertions(+), 25 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 72b880e28e..859f72b458 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -363,7 +363,7 @@ struct tnl_program {
 };
 
 
-static const struct ureg undef = { 
+static const struct ureg undef = {
    PROGRAM_UNDEFINED,
    0,
    0,
@@ -398,7 +398,7 @@ static struct ureg negate( struct ureg reg )
 {
    reg.negate ^= 1;
    return reg;
-} 
+}
 
 
 static struct ureg swizzle( struct ureg reg, int x, int y, int z, int w )
@@ -407,7 +407,6 @@ static struct ureg swizzle( struct ureg reg, int x, int y, int z, int w )
 			   GET_SWZ(reg.swz, y),
 			   GET_SWZ(reg.swz, z),
 			   GET_SWZ(reg.swz, w));
-
    return reg;
 }
 
@@ -898,8 +897,7 @@ static void build_hpos( struct tnl_program *p )
 
 static GLuint material_attrib( GLuint side, GLuint property )
 {
-   return ((property - STATE_AMBIENT) * 2 + 
-	   side);
+   return (property - STATE_AMBIENT) * 2 + side;
 }
 
 
@@ -960,7 +958,7 @@ static struct ureg get_scenecolor( struct tnl_program *p, GLuint side )
       struct ureg material_ambient = get_material(p, side, STATE_AMBIENT);
       struct ureg material_diffuse = get_material(p, side, STATE_DIFFUSE);
       struct ureg tmp = make_temp(p, material_diffuse);
-      emit_op3(p, OPCODE_MAD, tmp,  WRITEMASK_XYZ, lm_ambient, 
+      emit_op3(p, OPCODE_MAD, tmp, WRITEMASK_XYZ, lm_ambient, 
 	       material_ambient, material_emission);
       return tmp;
    }
@@ -978,7 +976,7 @@ static struct ureg get_lightprod( struct tnl_program *p, GLuint light,
 	 register_param3(p, STATE_LIGHT, light, property);
       struct ureg material_value = get_material(p, side, property);
       struct ureg tmp = get_temp(p);
-      emit_op2(p, OPCODE_MUL, tmp,  0, light_value, material_value);
+      emit_op2(p, OPCODE_MUL, tmp, 0, light_value, material_value);
       return tmp;
    }
    else
@@ -1015,7 +1013,6 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
    /* Calculate distance attenuation:
     */
    if (p->state->unit[i].light_attenuated) {
-
       /* 1/d,d,d,1/d */
       emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist); 
       /* 1,d,d*d,1/d */
@@ -1028,7 +1025,8 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
 	 emit_op1(p, OPCODE_RCP, dist, 0, dist); 
 	 /* spot-atten * dist-atten */
 	 emit_op2(p, OPCODE_MUL, att, 0, dist, att);	
-      } else {
+      }
+      else {
 	 /* dist-atten */
 	 emit_op1(p, OPCODE_RCP, att, 0, dist); 
       }
@@ -1082,10 +1080,10 @@ static void build_lighting( struct tnl_program *p )
 
    /*
     * NOTE:
-    * dot.x = dot(normal, VPpli)
-    * dot.y = dot(normal, halfAngle)
-    * dot.z = back.shininess
-    * dot.w = front.shininess
+    * dots.x = dot(normal, VPpli)
+    * dots.y = dot(normal, halfAngle)
+    * dots.z = back.shininess
+    * dots.w = front.shininess
     */
 
    for (i = 0; i < MAX_LIGHTS; i++) 
@@ -1097,7 +1095,7 @@ static void build_lighting( struct tnl_program *p )
    {
       if (!p->state->material_shininess_is_zero) {
          struct ureg shininess = get_material(p, 0, STATE_SHININESS);
-         emit_op1(p, OPCODE_MOV, dots,  WRITEMASK_W, swizzle1(shininess,X));
+         emit_op1(p, OPCODE_MOV, dots, WRITEMASK_W, swizzle1(shininess,X));
          release_temp(p, shininess);
       }
 
@@ -1106,7 +1104,6 @@ static void build_lighting( struct tnl_program *p )
 	 _col1 = make_temp(p, get_identity_param(p));
       else
 	 _col1 = _col0;
-
    }
 
    if (twoside) {
@@ -1171,12 +1168,13 @@ static void build_lighting( struct tnl_program *p )
                   half = get_temp(p);
                   emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
                   emit_normalize_vec3(p, half, half);
-               } else {
+               }
+               else {
                   half = register_param3(p, STATE_INTERNAL, 
                                          STATE_LIGHT_HALF_VECTOR, i);
                }
             }
-	 } 
+	 }
 	 else {
 	    struct ureg Ppli = register_param3(p, STATE_INTERNAL, 
 					       STATE_LIGHT_POSITION, i); 
@@ -1255,7 +1253,8 @@ static void build_lighting( struct tnl_program *p )
 		  res0 = _col0;
 		  res1 = register_output( p, VERT_RESULT_COL0 );
 	       }
-	    } else {
+	    }
+            else {
 	       mask0 = 0;
 	       mask1 = 0;
 	       res0 = _col0;
@@ -1267,12 +1266,12 @@ static void build_lighting( struct tnl_program *p )
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
                emit_op2(p, OPCODE_MUL, lit, 0, lit, att);
                emit_op3(p, OPCODE_MAD, _col0, 0, swizzle1(lit,X), ambient, _col0);
-            } 
+            }
             else if (!p->state->material_shininess_is_zero) {
                /* there's a non-zero specular term */
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
                emit_op2(p, OPCODE_ADD, _col0, 0, ambient, _col0);
-            } 
+            }
             else {
                /* no attenutation, no specular */
                emit_degenerate_lit(p, lit, dots);
@@ -1309,7 +1308,8 @@ static void build_lighting( struct tnl_program *p )
 		  res0 = _bfc0;
 		  res1 = register_output( p, VERT_RESULT_BFC0 );
 	       }
-	    } else {
+	    }
+            else {
 	       res0 = _bfc0;
 	       res1 = _bfc1;
 	       mask0 = 0;
@@ -1325,8 +1325,8 @@ static void build_lighting( struct tnl_program *p )
             }
             else if (!p->state->material_shininess_is_zero) {
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
-               emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
-            } 
+               emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0); /**/
+            }
             else {
                emit_degenerate_lit(p, lit, dots);
                emit_op2(p, OPCODE_ADD, _bfc0, 0, ambient, _bfc0);
@@ -1578,7 +1578,7 @@ static void build_texture_transform( struct tnl_program *p )
 	 }
 
 	 release_temps(p);
-      } 
+      }
       else {
 	 emit_passthrough(p, VERT_ATTRIB_TEX0+i, VERT_RESULT_TEX0+i);
       }
@@ -1647,7 +1647,8 @@ static void build_array_pointsize( struct tnl_program *p )
 
 
 static void build_tnl_program( struct tnl_program *p )
-{   /* Emit the program, starting with modelviewproject:
+{
+   /* Emit the program, starting with modelviewproject:
     */
    build_hpos(p);
 
-- 
cgit v1.2.3


From 919f57078b289a273e0e46ee2214a9f042b11b1f Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Tue, 31 Mar 2009 16:13:14 -0600
Subject: mesa: fix bug in GPU codegen for fixed-function two-sided lighting

The 'dots' register wasn't getting properly un-negated and un-swizzled
after emitting the code for back-face lighting.  So, if more than one
light source was enabled, the specular exponent for the next light source
was wrong.

During execution we were evaluating pow(x, y) where y was negative instead
of positive.  This led to the outcome being zero or NaN.

This fixes the occasional black triangles seen in isosurf when hacked to
enable two-sided lighting.
---
 src/mesa/main/ffvertex_prog.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 859f72b458..03f42704a7 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1108,6 +1108,9 @@ static void build_lighting( struct tnl_program *p )
 
    if (twoside) {
       if (!p->state->material_shininess_is_zero) {
+         /* Note that we negate the back-face specular exponent here.
+          * The negation will be un-done later in the back-face code below.
+          */
          struct ureg shininess = get_material(p, 1, STATE_SHININESS);
          emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
                   negate(swizzle1(shininess,X)));
@@ -1316,6 +1319,11 @@ static void build_lighting( struct tnl_program *p )
 	       mask1 = 0;
 	    }
 
+            /* For the back face we need to negate the X and Y component
+             * dot products.  dots.Z has the negated back-face specular
+             * exponent.  We swizzle that into the W position.  This
+             * negation makes the back-face specular term positive again.
+             */
             dots = negate(swizzle(dots,X,Y,W,Z));
 
 	    if (!is_undef(att)) {
@@ -1334,8 +1342,10 @@ static void build_lighting( struct tnl_program *p )
 
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
-            /* restore negate flag for next lighting */
-            dots = negate(dots);
+            /* restore dots to its original state for subsequent lights
+             * by negating and swizzling again.
+             */
+            dots = negate(swizzle(dots,X,Y,W,Z));
 
 	    release_temp(p, ambient);
 	    release_temp(p, diffuse);
-- 
cgit v1.2.3


From 7db7ff878d3e5a6b345228e6eaee4797bb68b360 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Tue, 14 Apr 2009 22:14:30 -0600
Subject: mesa: merge the prog_src_register::NegateBase and NegateAbs fields

There's really no need for two negation fields.  This came from the
GL_NV_fragment_program extension.  The new, unified Negate bitfield applies
after the absolute value step.
---
 src/mesa/drivers/dri/i915/i915_fragprog.c         | 10 ++---
 src/mesa/drivers/dri/i965/brw_vs_constval.c       |  2 +-
 src/mesa/drivers/dri/i965/brw_vs_emit.c           |  8 ++--
 src/mesa/drivers/dri/i965/brw_wm_fp.c             | 13 +++---
 src/mesa/drivers/dri/i965/brw_wm_glsl.c           |  6 +--
 src/mesa/drivers/dri/i965/brw_wm_pass0.c          |  2 +-
 src/mesa/drivers/dri/r200/r200_vertprog.c         | 34 ++++++++--------
 src/mesa/drivers/dri/r300/r300_fragprog.c         |  4 +-
 src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c | 17 ++++----
 src/mesa/drivers/dri/r300/r300_vertprog.c         | 49 +++++++++--------------
 src/mesa/drivers/dri/r300/r500_fragprog.c         | 19 +++++----
 src/mesa/drivers/dri/r300/radeon_nqssadce.c       |  7 ++--
 src/mesa/drivers/dri/r300/radeon_program_alu.c    | 20 +++++----
 src/mesa/drivers/dri/r300/radeon_program_pair.c   | 11 +++--
 src/mesa/main/ffvertex_prog.c                     |  3 +-
 src/mesa/main/texenvprogram.c                     |  3 +-
 src/mesa/shader/arbprogparse.c                    |  8 ++--
 src/mesa/shader/nvfragparse.c                     | 33 ++++++++-------
 src/mesa/shader/nvvertparse.c                     | 10 ++---
 src/mesa/shader/prog_execute.c                    | 27 ++++---------
 src/mesa/shader/prog_instruction.h                | 32 +++------------
 src/mesa/shader/prog_print.c                      | 20 ++++-----
 src/mesa/shader/programopt.c                      |  2 +-
 src/mesa/shader/slang/slang_emit.c                |  2 +-
 src/mesa/state_tracker/st_cb_bitmap.c             |  2 +-
 src/mesa/state_tracker/st_mesa_to_tgsi.c          | 20 ++++-----
 26 files changed, 156 insertions(+), 208 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 52f09a4b1b..a5158de945 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -162,12 +162,12 @@ src_vector(struct i915_fragment_program *p,
                  GET_SWZ(source->Swizzle, 1),
                  GET_SWZ(source->Swizzle, 2), GET_SWZ(source->Swizzle, 3));
 
-   if (source->NegateBase)
+   if (source->Negate)
       src = negate(src,
-                   GET_BIT(source->NegateBase, 0),
-                   GET_BIT(source->NegateBase, 1),
-                   GET_BIT(source->NegateBase, 2),
-                   GET_BIT(source->NegateBase, 3));
+                   GET_BIT(source->Negate, 0),
+                   GET_BIT(source->Negate, 1),
+                   GET_BIT(source->Negate, 2),
+                   GET_BIT(source->Negate, 3));
 
    return src;
 }
diff --git a/src/mesa/drivers/dri/i965/brw_vs_constval.c b/src/mesa/drivers/dri/i965/brw_vs_constval.c
index d29eb17f8c..2637344b48 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_constval.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_constval.c
@@ -96,7 +96,7 @@ static GLubyte get_active( struct tracker *t,
 			   struct prog_src_register src )
 {
    GLuint i;
-   GLubyte active = src.NegateBase; /* NOTE! */
+   GLubyte active = src.Negate; /* NOTE! */
 
    if (src.RelAddr)
       return 0xf;
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 2ee63129bc..42f6a99142 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -899,7 +899,7 @@ static struct brw_reg get_arg( struct brw_vs_compile *c,
 
    /* Note this is ok for non-swizzle instructions: 
     */
-   reg.negate = src->NegateBase ? 1 : 0;   
+   reg.negate = src->Negate ? 1 : 0;   
 
    return reg;
 }
@@ -945,7 +945,7 @@ static void emit_swz( struct brw_vs_compile *c,
    GLuint ones_mask = 0;
    GLuint src_mask = 0;
    GLubyte src_swz[4];
-   GLboolean need_tmp = (src.NegateBase &&
+   GLboolean need_tmp = (src.Negate &&
 			 dst.file != BRW_GENERAL_REGISTER_FILE);
    struct brw_reg tmp = dst;
    GLuint i;
@@ -997,8 +997,8 @@ static void emit_swz( struct brw_vs_compile *c,
    if (ones_mask) 
       brw_MOV(p, brw_writemask(tmp, ones_mask), brw_imm_f(1));
 
-   if (src.NegateBase)
-      brw_MOV(p, brw_writemask(tmp, src.NegateBase), negate(tmp));
+   if (src.Negate)
+      brw_MOV(p, brw_writemask(tmp, src.Negate), negate(tmp));
    
    if (need_tmp) {
       brw_MOV(p, dst, tmp);
diff --git a/src/mesa/drivers/dri/i965/brw_wm_fp.c b/src/mesa/drivers/dri/i965/brw_wm_fp.c
index a7f5f1b9a2..1798d842c7 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_fp.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_fp.c
@@ -80,9 +80,8 @@ static struct prog_src_register src_reg(GLuint file, GLuint idx)
    reg.Index = idx;
    reg.Swizzle = SWIZZLE_NOOP;
    reg.RelAddr = 0;
-   reg.NegateBase = 0;
+   reg.Negate = NEGATE_NONE;
    reg.Abs = 0;
-   reg.NegateAbs = 0;
    return reg;
 }
 
@@ -569,7 +568,7 @@ static void precalc_dst( struct brw_wm_compile *c,
 		    src_undef(),
 		    src_undef());
       /* Avoid letting negation flag of src0 affect our 1 constant. */
-      swz->SrcReg[0].NegateBase &= ~NEGATE_X;
+      swz->SrcReg[0].Negate &= ~NEGATE_X;
    }
    if (dst.WriteMask & WRITEMASK_W) {
       /* dst.w = mov src1.w
@@ -604,7 +603,7 @@ static void precalc_lit( struct brw_wm_compile *c,
 		    src_undef(),
 		    src_undef());
       /* Avoid letting the negation flag of src0 affect our 1 constant. */
-      swz->SrcReg[0].NegateBase = 0;
+      swz->SrcReg[0].Negate = NEGATE_NONE;
    }
 
    if (dst.WriteMask & WRITEMASK_YZ) {
@@ -651,7 +650,7 @@ static void precalc_tex( struct brw_wm_compile *c,
                      src0,
                      src_undef(),
                      src_undef());
-       out->SrcReg[0].NegateBase = 0;
+       out->SrcReg[0].Negate = NEGATE_NONE;
        out->SrcReg[0].Abs = 1;
 
        /* tmp0 = MAX(coord.X, coord.Y) */
@@ -1050,14 +1049,14 @@ void brw_wm_pass_fp( struct brw_wm_compile *c )
       case OPCODE_ABS:
 	 out = emit_insn(c, inst);
 	 out->Opcode = OPCODE_MOV;
-	 out->SrcReg[0].NegateBase = 0;
+	 out->SrcReg[0].Negate = NEGATE_NONE;
 	 out->SrcReg[0].Abs = 1;
 	 break;
 
       case OPCODE_SUB: 
 	 out = emit_insn(c, inst);
 	 out->Opcode = OPCODE_ADD;
-	 out->SrcReg[1].NegateBase ^= 0xf;
+	 out->SrcReg[1].Negate ^= NEGATE_XYZW;
 	 break;
 
       case OPCODE_SCS: 
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index 49fea2e41a..385efd2dd3 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -340,7 +340,7 @@ get_src_reg_const(struct brw_wm_compile *c,
    const_reg = stride(const_reg, 0, 1, 0);
    const_reg.subnr = component * 4;
 
-   if (src->NegateBase)
+   if (src->Negate & (1 << component))
       const_reg = negate(const_reg);
    if (src->Abs)
       const_reg = brw_abs(const_reg);
@@ -377,7 +377,7 @@ static struct brw_reg get_src_reg(struct brw_wm_compile *c,
     else {
        /* other type of source register */
        return get_reg(c, src->File, src->Index, component, nr, 
-                      src->NegateBase, src->Abs);
+                      src->Negate, src->Abs);
     }
 }
 
@@ -402,7 +402,7 @@ static struct brw_reg get_src_reg_imm(struct brw_wm_compile *c,
        const GLfloat *param =
           c->fp->program.Base.Parameters->ParameterValues[src->Index];
        GLfloat value = param[component];
-       if (src->NegateBase)
+       if (src->Negate & (1 << channel))
           value = -value;
        if (src->Abs)
           value = FABSF(value);
diff --git a/src/mesa/drivers/dri/i965/brw_wm_pass0.c b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
index 2debd0678a..92142764f5 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_pass0.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_pass0.c
@@ -322,7 +322,7 @@ static struct brw_wm_ref *get_new_ref( struct brw_wm_compile *c,
       newref->value->lastuse = newref;
    }
 
-   if (src.NegateBase & (1<<i))
+   if (src.Negate & (1 << i))
       newref->hw_reg.negate ^= 1;
 
    if (src.Abs) {
diff --git a/src/mesa/drivers/dri/r200/r200_vertprog.c b/src/mesa/drivers/dri/r200/r200_vertprog.c
index a2561df579..4ce93b5145 100644
--- a/src/mesa/drivers/dri/r200/r200_vertprog.c
+++ b/src/mesa/drivers/dri/r200/r200_vertprog.c
@@ -290,7 +290,7 @@ static unsigned long t_src(struct r200_vertex_program *vp, struct prog_src_regis
 			t_swizzle(GET_SWZ(src->Swizzle, 2)),
 			t_swizzle(GET_SWZ(src->Swizzle, 3)),
 			t_src_class(src->File),
-			src->NegateBase) | (src->RelAddr << 4);
+			src->Negate) | (src->RelAddr << 4);
 }
 
 static unsigned long t_src_scalar(struct r200_vertex_program *vp, struct prog_src_register *src)
@@ -302,7 +302,7 @@ static unsigned long t_src_scalar(struct r200_vertex_program *vp, struct prog_sr
 			t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			t_src_class(src->File),
-			src->NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src->RelAddr << 4);
+			src->Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src->RelAddr << 4);
 }
 
 static unsigned long t_opcode(enum prog_opcode opcode)
@@ -700,7 +700,7 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 		   t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
 		   SWIZZLE_ZERO,
 		   t_src_class(src[0].File),
-		   src[0].NegateBase) | (src[0].RelAddr << 4);
+		   src[0].Negate) | (src[0].RelAddr << 4);
 	    o_inst->src1 = UNUSED_SRC_0;
 	    o_inst->src2 = UNUSED_SRC_0;
 	 }
@@ -712,12 +712,12 @@ static GLboolean r200_translate_vertex_program(GLcontext *ctx, struct r200_verte
 		   t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 		   SWIZZLE_ZERO, SWIZZLE_ZERO, SWIZZLE_ZERO,
 		   t_src_class(src[0].File),
-		   src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+		   src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
 	    o_inst->src1 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 		   SWIZZLE_ZERO, SWIZZLE_ZERO,
 		   t_swizzle(GET_SWZ(src[1].Swizzle, 0)), SWIZZLE_ZERO,
 		   t_src_class(src[1].File),
-		   src[1].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+		   src[1].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
 	    o_inst->src2 = UNUSED_SRC_1;
 	    o_inst++;
 
@@ -766,11 +766,11 @@ if ((o_inst - vp->instr) == 31) {
 o_inst->src1 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 			SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X,
 			t_src_class(src[1].File),
-			src[1].NegateBase) | (src[1].RelAddr << 4);
+			src[1].Negate) | (src[1].RelAddr << 4);
 o_inst->src2 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 			SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y,
 			t_src_class(src[1].File),
-			src[1].NegateBase) | (src[1].RelAddr << 4);
+			src[1].Negate) | (src[1].RelAddr << 4);
 }
 else {
 	 o_inst->src1 = t_src(vp, &src[1]);
@@ -792,7 +792,7 @@ else {
 		t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
 		SWIZZLE_ZERO,
 		t_src_class(src[0].File),
-		src[0].NegateBase) | (src[0].RelAddr << 4);
+		src[0].Negate) | (src[0].RelAddr << 4);
 
 	 o_inst->src1 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 		t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
@@ -800,7 +800,7 @@ else {
 		t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
 		SWIZZLE_ZERO,
 		t_src_class(src[1].File),
-		src[1].NegateBase) | (src[1].RelAddr << 4);
+		src[1].Negate) | (src[1].RelAddr << 4);
 
 	 o_inst->src2 = UNUSED_SRC_1;
 	 goto next;
@@ -815,7 +815,7 @@ else {
 		t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
 		VSF_IN_COMPONENT_ONE,
 		t_src_class(src[0].File),
-		src[0].NegateBase) | (src[0].RelAddr << 4);
+		src[0].Negate) | (src[0].RelAddr << 4);
 	 o_inst->src1 = t_src(vp, &src[1]);
 	 o_inst->src2 = UNUSED_SRC_1;
 	 goto next;
@@ -831,7 +831,7 @@ else {
 		t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
 		t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
 		t_src_class(src[1].File),
-		(!src[1].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+		(!src[1].Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
 	 o_inst->src2 = UNUSED_SRC_1;
 	 goto next;
 
@@ -846,7 +846,7 @@ else {
 		t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
 		t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
 		t_src_class(src[0].File),
-		(!src[0].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+		(!src[0].Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
 	 o_inst->src2 = UNUSED_SRC_1;
 	 goto next;
 
@@ -874,7 +874,7 @@ else {
 		VSF_IN_COMPONENT_W,
 		VSF_IN_CLASS_TMP,
 		/* Not 100% sure about this */
-		(!src[0].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE/*VSF_FLAG_ALL*/);
+		(!src[0].Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE/*VSF_FLAG_ALL*/);
 
 	 o_inst->src2 = UNUSED_SRC_0;
 	 u_temp_i--;
@@ -899,7 +899,7 @@ else {
 		t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
 		t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
 		t_src_class(src[0].File),
-		src[0].NegateBase) | (src[0].RelAddr << 4);
+		src[0].Negate) | (src[0].RelAddr << 4);
 
 	 o_inst->src1 = MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
 		t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z
@@ -907,7 +907,7 @@ else {
 		t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y
 		t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w
 		t_src_class(src[1].File),
-		src[1].NegateBase) | (src[1].RelAddr << 4);
+		src[1].Negate) | (src[1].RelAddr << 4);
 
 	 o_inst->src2 = UNUSED_SRC_1;
 	 o_inst++;
@@ -922,7 +922,7 @@ else {
 		t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x
 		t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w
 		t_src_class(src[1].File),
-		(!src[1].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+		(!src[1].Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
 
 	 o_inst->src1 = MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
 		t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z
@@ -930,7 +930,7 @@ else {
 		t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
 		t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
 		t_src_class(src[0].File),
-		src[0].NegateBase) | (src[0].RelAddr << 4);
+		src[0].Negate) | (src[0].RelAddr << 4);
 
 	 o_inst->src2 = MAKE_VSF_SOURCE(u_temp_i+1,
 		VSF_IN_COMPONENT_X,
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog.c b/src/mesa/drivers/dri/r300/r300_fragprog.c
index 32182bb667..873cde4414 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog.c
@@ -214,9 +214,9 @@ static GLboolean transform_TEX(
 		 *   r  < tex  <=>      -tex+r < 0
 		 *   r >= tex  <=> not (-tex+r < 0 */
 		if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
-			tgt[1].SrcReg[2].NegateBase = tgt[0].SrcReg[2].NegateBase ^ NEGATE_XYZW;
+			tgt[1].SrcReg[2].Negate = tgt[0].SrcReg[2].Negate ^ NEGATE_XYZW;
 		else
-			tgt[1].SrcReg[0].NegateBase = tgt[0].SrcReg[0].NegateBase ^ NEGATE_XYZW;
+			tgt[1].SrcReg[0].Negate = tgt[0].SrcReg[0].Negate ^ NEGATE_XYZW;
 
 		tgt[2].Opcode = OPCODE_CMP;
 		tgt[2].DstReg = orig_inst->DstReg;
diff --git a/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c
index a86d2bd471..191853ac1f 100644
--- a/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c
+++ b/src/mesa/drivers/dri/r300/r300_fragprog_swizzle.c
@@ -92,7 +92,7 @@ static const struct swizzle_data* lookup_native_swizzle(GLuint swizzle)
 GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
 {
 	if (reg.Abs)
-		reg.NegateBase = 0;
+		reg.Negate = NEGATE_NONE;
 
 	if (opcode == OPCODE_KIL ||
 	    opcode == OPCODE_TEX ||
@@ -100,7 +100,8 @@ GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
 	    opcode == OPCODE_TXP) {
 		int j;
 
-		if (reg.Abs || reg.NegateBase != (15*reg.NegateAbs))
+		if (reg.Abs || (reg.Negate != NEGATE_XYZW &&
+                                reg.Negate != NEGATE_NONE))
 			return GL_FALSE;
 
 		for(j = 0; j < 4; ++j) {
@@ -121,7 +122,7 @@ GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
 		if (GET_SWZ(reg.Swizzle, j) != SWIZZLE_NIL)
 			relevant |= 1 << j;
 
-	if ((reg.NegateBase & relevant) && (reg.NegateBase & relevant) != relevant)
+	if ((reg.Negate & relevant) && (reg.Negate & relevant) != relevant)
 		return GL_FALSE;
 
 	if (!lookup_native_swizzle(reg.Swizzle))
@@ -137,7 +138,7 @@ GLboolean r300FPIsNativeSwizzle(GLuint opcode, struct prog_src_register reg)
 void r300FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst, struct prog_src_register src)
 {
 	if (src.Abs)
-		src.NegateBase = 0;
+		src.Negate = NEGATE_NONE;
 
 	while(dst.WriteMask) {
 		const struct swizzle_data *best_swizzle = 0;
@@ -170,11 +171,11 @@ void r300FPBuildSwizzle(struct nqssadce_state *s, struct prog_dst_register dst,
 			}
 		}
 
-		if ((src.NegateBase & best_matchmask) != 0) {
-			best_matchmask &= src.NegateBase;
-			rgbnegate = !src.NegateAbs;
+		if ((src.Negate & best_matchmask) != 0) {
+			best_matchmask &= src.Negate;
+			rgbnegate = !src.Negate;
 		} else {
-			rgbnegate = src.NegateAbs;
+			rgbnegate = src.Negate;
 		}
 
 		struct prog_instruction *inst;
diff --git a/src/mesa/drivers/dri/r300/r300_vertprog.c b/src/mesa/drivers/dri/r300/r300_vertprog.c
index 50806575ce..146daa367c 100644
--- a/src/mesa/drivers/dri/r300/r300_vertprog.c
+++ b/src/mesa/drivers/dri/r300/r300_vertprog.c
@@ -245,7 +245,7 @@ static unsigned long t_src_index(struct r300_vertex_program *vp,
 static unsigned long t_src(struct r300_vertex_program *vp,
 			   struct prog_src_register *src)
 {
-	/* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
+	/* src->Negate uses the NEGATE_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
 	return PVS_SRC_OPERAND(t_src_index(vp, src),
@@ -254,13 +254,13 @@ static unsigned long t_src(struct r300_vertex_program *vp,
 			       t_swizzle(GET_SWZ(src->Swizzle, 2)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 3)),
 			       t_src_class(src->File),
-			       src->NegateBase) | (src->RelAddr << 4);
+			       src->Negate) | (src->RelAddr << 4);
 }
 
 static unsigned long t_src_scalar(struct r300_vertex_program *vp,
 				  struct prog_src_register *src)
 {
-	/* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
+	/* src->Negate uses the NEGATE_ flags from program_instruction.h,
 	 * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
 	 */
 	return PVS_SRC_OPERAND(t_src_index(vp, src),
@@ -269,8 +269,7 @@ static unsigned long t_src_scalar(struct r300_vertex_program *vp,
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_swizzle(GET_SWZ(src->Swizzle, 0)),
 			       t_src_class(src->File),
-			       src->
-			       NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+			       src->Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src->RelAddr << 4);
 }
 
@@ -307,7 +306,7 @@ static GLuint *r300TranslateOpcodeABS(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
 				  t_src_class(src[0].File),
 				  (!src[0].
-				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				   Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[3] = 0;
 
@@ -369,8 +368,7 @@ static GLuint *r300TranslateOpcodeDP3(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
 				  SWIZZLE_ZERO,
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] =
 	    PVS_SRC_OPERAND(t_src_index(vp, &src[1]),
@@ -378,8 +376,7 @@ static GLuint *r300TranslateOpcodeDP3(struct r300_vertex_program *vp,
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
 			    t_swizzle(GET_SWZ(src[1].Swizzle, 2)), SWIZZLE_ZERO,
 			    t_src_class(src[1].File),
-			    src[1].
-			    NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+			    src[1].Negate ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 	inst[3] = __CONST(1, SWIZZLE_ZERO);
 
@@ -422,8 +419,7 @@ static GLuint *r300TranslateOpcodeDPH(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
 				  PVS_SRC_SELECT_FORCE_1,
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_XYZ : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] = t_src(vp, &src[1]);
 	inst[3] = __CONST(1, SWIZZLE_ZERO);
@@ -519,7 +515,7 @@ static GLuint *r300TranslateOpcodeFLR(struct r300_vertex_program *vp,
 				  PVS_SRC_SELECT_W, PVS_SRC_REG_TEMPORARY,
 				  /* Not 100% sure about this */
 				  (!src[0].
-				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE
+				   Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE
 				  /*VSF_FLAG_ALL */ );
 	inst[3] = __CONST(0, SWIZZLE_ZERO);
 	(*u_temp_i)--;
@@ -564,8 +560,7 @@ static GLuint *r300TranslateOpcodeLG2(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] = __CONST(0, SWIZZLE_ZERO);
 	inst[3] = __CONST(0, SWIZZLE_ZERO);
@@ -592,24 +587,21 @@ static GLuint *r300TranslateOpcodeLIT(struct r300_vertex_program *vp,
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[3] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
 				  PVS_SRC_SELECT_FORCE_0,	// Z
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 
 	return inst;
@@ -837,7 +829,7 @@ static GLuint *r300TranslateOpcodeSUB(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
 				  t_src_class(src[1].File),
 				  (!src[1].
-				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				   Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 	inst[3] = 0;
 #else
@@ -857,7 +849,7 @@ static GLuint *r300TranslateOpcodeSUB(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
 				  t_src_class(src[1].File),
 				  (!src[1].
-				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				   Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 #endif
 
@@ -905,16 +897,14 @@ static GLuint *r300TranslateOpcodeXPD(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[1]), t_swizzle(GET_SWZ(src[1].Swizzle, 2)),	// Z
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 0)),	// X
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// W
 				  t_src_class(src[1].File),
-				  src[1].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[1].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 	inst[3] = __CONST(1, SWIZZLE_ZERO);
 	inst += 4;
@@ -931,15 +921,14 @@ static GLuint *r300TranslateOpcodeXPD(struct r300_vertex_program *vp,
 				  t_swizzle(GET_SWZ(src[1].Swizzle, 3)),	// W
 				  t_src_class(src[1].File),
 				  (!src[1].
-				   NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				   Negate) ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[1].RelAddr << 4);
 	inst[2] = PVS_SRC_OPERAND(t_src_index(vp, &src[0]), t_swizzle(GET_SWZ(src[0].Swizzle, 2)),	// Z
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 0)),	// X
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 1)),	// Y
 				  t_swizzle(GET_SWZ(src[0].Swizzle, 3)),	// W
 				  t_src_class(src[0].File),
-				  src[0].
-				  NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
+				  src[0].Negate ? VSF_FLAG_ALL : VSF_FLAG_NONE) |
 	    (src[0].RelAddr << 4);
 	inst[3] =
 	    PVS_SRC_OPERAND(*u_temp_i, PVS_SRC_SELECT_X, PVS_SRC_SELECT_Y,
diff --git a/src/mesa/drivers/dri/r300/r500_fragprog.c b/src/mesa/drivers/dri/r300/r500_fragprog.c
index 07a2a7b17c..292573de89 100644
--- a/src/mesa/drivers/dri/r300/r500_fragprog.c
+++ b/src/mesa/drivers/dri/r300/r500_fragprog.c
@@ -156,9 +156,9 @@ static GLboolean transform_TEX(
 		 *   r  < tex  <=>      -tex+r < 0
 		 *   r >= tex  <=> not (-tex+r < 0 */
 		if (comparefunc == GL_LESS || comparefunc == GL_GEQUAL)
-			tgt[1].SrcReg[2].NegateBase = tgt[0].SrcReg[2].NegateBase ^ NEGATE_XYZW;
+			tgt[1].SrcReg[2].Negate = tgt[0].SrcReg[2].Negate ^ NEGATE_XYZW;
 		else
-			tgt[1].SrcReg[0].NegateBase = tgt[0].SrcReg[0].NegateBase ^ NEGATE_XYZW;
+			tgt[1].SrcReg[0].Negate = tgt[0].SrcReg[0].Negate ^ NEGATE_XYZW;
 
 		tgt[2].Opcode = OPCODE_CMP;
 		tgt[2].DstReg = orig_inst->DstReg;
@@ -314,8 +314,8 @@ static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
 		if (reg.Abs)
 			return GL_FALSE;
 
-		if (reg.NegateAbs)
-			reg.NegateBase ^= 15;
+		if (reg.Negate)
+			reg.Negate ^= NEGATE_XYZW;
 
 		if (opcode == OPCODE_KIL) {
 			if (reg.Swizzle != SWIZZLE_NOOP)
@@ -324,7 +324,7 @@ static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
 			for(i = 0; i < 4; ++i) {
 				GLuint swz = GET_SWZ(reg.Swizzle, i);
 				if (swz == SWIZZLE_NIL) {
-					reg.NegateBase &= ~(1 << i);
+					reg.Negate &= ~(1 << i);
 					continue;
 				}
 				if (swz >= 4)
@@ -332,15 +332,14 @@ static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
 			}
 		}
 
-		if (reg.NegateBase)
+		if (reg.Negate)
 			return GL_FALSE;
 
 		return GL_TRUE;
 	} else if (opcode == OPCODE_DDX || opcode == OPCODE_DDY) {
 		/* DDX/MDH and DDY/MDV explicitly ignore incoming swizzles;
 		 * if it doesn't fit perfectly into a .xyzw case... */
-		if (reg.Swizzle == SWIZZLE_NOOP && !reg.Abs
-				&& !reg.NegateBase && !reg.NegateAbs)
+		if (reg.Swizzle == SWIZZLE_NOOP && !reg.Abs && !reg.Negate)
 			return GL_TRUE;
 
 		return GL_FALSE;
@@ -355,7 +354,7 @@ static GLboolean is_native_swizzle(GLuint opcode, struct prog_src_register reg)
 			if (swz != SWIZZLE_NIL && swz != SWIZZLE_ZERO)
 				relevant |= 1 << i;
 		}
-		if ((reg.NegateBase & relevant) && ((reg.NegateBase & relevant) != relevant))
+		if ((reg.Negate & relevant) && ((reg.Negate & relevant) != relevant))
 			return GL_FALSE;
 
 		return GL_TRUE;
@@ -379,7 +378,7 @@ static void nqssadce_build_swizzle(struct nqssadce_state *s,
 		GLuint swz = GET_SWZ(src.Swizzle, i);
 		if (swz == SWIZZLE_NIL)
 			continue;
-		negatebase[GET_BIT(src.NegateBase, i)] |= 1 << i;
+		negatebase[GET_BIT(src.Negate, i)] |= 1 << i;
 	}
 
 	_mesa_insert_instructions(s->Program, s->IP, (negatebase[0] ? 1 : 0) + (negatebase[1] ? 1 : 0));
diff --git a/src/mesa/drivers/dri/r300/radeon_nqssadce.c b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
index a083c3d243..4a2e1cba40 100644
--- a/src/mesa/drivers/dri/r300/radeon_nqssadce.c
+++ b/src/mesa/drivers/dri/r300/radeon_nqssadce.c
@@ -61,12 +61,12 @@ static struct prog_src_register lmul_swizzle(GLuint swizzle, struct prog_src_reg
 	struct prog_src_register tmp = srcreg;
 	int i;
 	tmp.Swizzle = 0;
-	tmp.NegateBase = 0;
+	tmp.Negate = NEGATE_NONE;
 	for(i = 0; i < 4; ++i) {
 		GLuint swz = GET_SWZ(swizzle, i);
 		if (swz < 4) {
 			tmp.Swizzle |= GET_SWZ(srcreg.Swizzle, swz) << (i*3);
-			tmp.NegateBase |= GET_BIT(srcreg.NegateBase, swz) << i;
+			tmp.Negate |= GET_BIT(srcreg.Negate, swz) << i;
 		} else {
 			tmp.Swizzle |= swz << (i*3);
 		}
@@ -103,9 +103,8 @@ static struct prog_instruction* track_used_srcreg(struct nqssadce_state* s,
 		inst->SrcReg[src].File = PROGRAM_TEMPORARY;
 		inst->SrcReg[src].Index = dstreg.Index;
 		inst->SrcReg[src].Swizzle = 0;
-		inst->SrcReg[src].NegateBase = 0;
+		inst->SrcReg[src].Negate = NEGATE_NONE;
 		inst->SrcReg[src].Abs = 0;
-		inst->SrcReg[src].NegateAbs = 0;
 		for(i = 0; i < 4; ++i) {
 			if (GET_BIT(sourced, i))
 				inst->SrcReg[src].Swizzle |= i << (3*i);
diff --git a/src/mesa/drivers/dri/r300/radeon_program_alu.c b/src/mesa/drivers/dri/r300/radeon_program_alu.c
index 1ef71e74dc..ebc5c913b2 100644
--- a/src/mesa/drivers/dri/r300/radeon_program_alu.c
+++ b/src/mesa/drivers/dri/r300/radeon_program_alu.c
@@ -89,8 +89,9 @@ static void set_swizzle(struct prog_src_register *SrcReg, int coordinate, int sw
 
 static void set_negate_base(struct prog_src_register *SrcReg, int coordinate, int negate)
 {
-	SrcReg->NegateBase &= ~(1 << coordinate);
-	SrcReg->NegateBase |= (negate << coordinate);
+	/* XXX note sure about this negation logic here */
+	SrcReg->Negate &= ~(1 << coordinate);
+	SrcReg->Negate |= (negate << coordinate);
 }
 
 static struct prog_dst_register dstreg(int file, int index)
@@ -156,15 +157,14 @@ static struct prog_src_register absolute(struct prog_src_register reg)
 {
 	struct prog_src_register newreg = reg;
 	newreg.Abs = 1;
-	newreg.NegateBase = 0;
-	newreg.NegateAbs = 0;
+	newreg.Negate = NEGATE_NONE;
 	return newreg;
 }
 
 static struct prog_src_register negate(struct prog_src_register reg)
 {
 	struct prog_src_register newreg = reg;
-	newreg.NegateAbs = !newreg.NegateAbs;
+	newreg.Negate = newreg.Negate ^ NEGATE_XYZW;
 	return newreg;
 }
 
@@ -189,8 +189,7 @@ static void transform_ABS(struct radeon_transform_context* t,
 {
 	struct prog_src_register src = inst->SrcReg[0];
 	src.Abs = 1;
-	src.NegateBase = 0;
-	src.NegateAbs = 0;
+	src.Negate = NEGATE_NONE;
 	emit1(t->Program, OPCODE_MOV, inst->SaturateMode, inst->DstReg, src);
 }
 
@@ -198,14 +197,13 @@ static void transform_DPH(struct radeon_transform_context* t,
 	struct prog_instruction* inst)
 {
 	struct prog_src_register src0 = inst->SrcReg[0];
-	if (src0.NegateAbs) {
+	if (src0.Negate) {
 		if (src0.Abs) {
 			int tempreg = radeonFindFreeTemporary(t);
 			emit1(t->Program, OPCODE_MOV, 0, dstreg(PROGRAM_TEMPORARY, tempreg), src0);
 			src0 = srcreg(src0.File, src0.Index);
 		} else {
-			src0.NegateAbs = 0;
-			src0.NegateBase ^= NEGATE_XYZW;
+			src0.Negate ^= NEGATE_XYZW;
 		}
 	}
 	set_swizzle(&src0, 3, SWIZZLE_ONE);
@@ -649,7 +647,7 @@ GLboolean radeonTransformDeriv(struct radeon_transform_context* t,
 
 	B.Swizzle = MAKE_SWIZZLE4(SWIZZLE_ONE, SWIZZLE_ONE,
 						SWIZZLE_ONE, SWIZZLE_ONE);
-	B.NegateBase = NEGATE_XYZW;
+	B.Negate = NEGATE_XYZW;
 
 	emit2(t->Program, inst->Opcode, inst->SaturateMode, inst->DstReg,
 		inst->SrcReg[0], B);
diff --git a/src/mesa/drivers/dri/r300/radeon_program_pair.c b/src/mesa/drivers/dri/r300/radeon_program_pair.c
index f398404f9f..ecc82ff8a8 100644
--- a/src/mesa/drivers/dri/r300/radeon_program_pair.c
+++ b/src/mesa/drivers/dri/r300/radeon_program_pair.c
@@ -255,8 +255,7 @@ static void final_rewrite(struct pair_state *s, struct prog_instruction *inst)
 		inst->SrcReg[2] = inst->SrcReg[1];
 		inst->SrcReg[1].File = PROGRAM_BUILTIN;
 		inst->SrcReg[1].Swizzle = SWIZZLE_1111;
-		inst->SrcReg[1].NegateBase = 0;
-		inst->SrcReg[1].NegateAbs = 0;
+		inst->SrcReg[1].Negate = NEGATE_NONE;
 		inst->Opcode = OPCODE_MAD;
 		break;
 	case OPCODE_CMP:
@@ -730,7 +729,7 @@ static GLboolean fill_instruction_into_pair(struct pair_state *s, struct radeon_
 					srcrgb = GL_TRUE;
 				else if (swz < 4)
 					srcalpha = GL_TRUE;
-				if (swz != SWIZZLE_NIL && GET_BIT(inst->SrcReg[i].NegateBase, j))
+				if (swz != SWIZZLE_NIL && GET_BIT(inst->SrcReg[i].Negate, j))
 					negatebase = 1;
 			}
 			source = alloc_pair_source(s, pair, inst->SrcReg[i], srcrgb, srcalpha);
@@ -739,12 +738,12 @@ static GLboolean fill_instruction_into_pair(struct pair_state *s, struct radeon_
 			pair->RGB.Arg[i].Source = source;
 			pair->RGB.Arg[i].Swizzle = inst->SrcReg[i].Swizzle & 0x1ff;
 			pair->RGB.Arg[i].Abs = inst->SrcReg[i].Abs;
-			pair->RGB.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].NegateAbs;
+			pair->RGB.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].Negate;
 		}
 		if (pairinst->NeedAlpha) {
 			GLboolean srcrgb = GL_FALSE;
 			GLboolean srcalpha = GL_FALSE;
-			GLuint negatebase = GET_BIT(inst->SrcReg[i].NegateBase, pairinst->IsTranscendent ? 0 : 3);
+			GLuint negatebase = GET_BIT(inst->SrcReg[i].Negate, pairinst->IsTranscendent ? 0 : 3);
 			GLuint swz = GET_SWZ(inst->SrcReg[i].Swizzle, pairinst->IsTranscendent ? 0 : 3);
 			if (swz < 3)
 				srcrgb = GL_TRUE;
@@ -756,7 +755,7 @@ static GLboolean fill_instruction_into_pair(struct pair_state *s, struct radeon_
 			pair->Alpha.Arg[i].Source = source;
 			pair->Alpha.Arg[i].Swizzle = swz;
 			pair->Alpha.Arg[i].Abs = inst->SrcReg[i].Abs;
-			pair->Alpha.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].NegateAbs;
+			pair->Alpha.Arg[i].Negate = (negatebase & ~pair->RGB.Arg[i].Abs) ^ inst->SrcReg[i].Negate;
 		}
 	}
 
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 03f42704a7..1ce5685af4 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -570,9 +570,8 @@ static void emit_arg( struct prog_src_register *src,
    src->File = reg.file;
    src->Index = reg.idx;
    src->Swizzle = reg.swz;
-   src->NegateBase = reg.negate ? NEGATE_XYZW : 0;
+   src->Negate = reg.negate ? NEGATE_XYZW : NEGATE_NONE;
    src->Abs = 0;
-   src->NegateAbs = 0;
    src->RelAddr = 0;
    /* Check that bitfield sizes aren't exceeded */
    ASSERT(src->Index == reg.idx);
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index 4a124bf27e..a70d069bd9 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -663,9 +663,8 @@ static void emit_arg( struct prog_src_register *reg,
    reg->File = ureg.file;
    reg->Index = ureg.idx;
    reg->Swizzle = ureg.swz;
-   reg->NegateBase = ureg.negatebase ? 0xf : 0x0;
+   reg->Negate = ureg.negatebase ? NEGATE_XYZW : NEGATE_NONE;
    reg->Abs = ureg.abs;
-   reg->NegateAbs = ureg.negateabs;
 }
 
 static void emit_dst( struct prog_dst_register *dst,
diff --git a/src/mesa/shader/arbprogparse.c b/src/mesa/shader/arbprogparse.c
index 35253daa2e..b47bf360cf 100644
--- a/src/mesa/shader/arbprogparse.c
+++ b/src/mesa/shader/arbprogparse.c
@@ -2669,7 +2669,7 @@ parse_vector_src_reg(GLcontext *ctx, const GLubyte **inst,
    reg->File = file;
    reg->Index = index;
    reg->Swizzle = MAKE_SWIZZLE4(swizzle[0], swizzle[1], swizzle[2], swizzle[3]);
-   reg->NegateBase = negateMask;
+   reg->Negate = negateMask;
    reg->RelAddr = isRelOffset;
    return 0;
 }
@@ -2703,7 +2703,7 @@ parse_scalar_src_reg(GLcontext *ctx, const GLubyte **inst,
    reg->File = file;
    reg->Index = index;
    reg->Swizzle = (swizzle[0] << 0);
-   reg->NegateBase = negateMask;
+   reg->Negate = negateMask;
    reg->RelAddr = isRelOffset;
    return 0;
 }
@@ -3019,7 +3019,7 @@ parse_fp_instruction (GLcontext * ctx, const GLubyte ** inst,
 	    parse_extended_swizzle_mask(inst, swizzle, &negateMask);
 	    fp->SrcReg[0].File = file;
 	    fp->SrcReg[0].Index = index;
-	    fp->SrcReg[0].NegateBase = negateMask;
+	    fp->SrcReg[0].Negate = negateMask;
 	    fp->SrcReg[0].Swizzle = MAKE_SWIZZLE4(swizzle[0],
                                                   swizzle[1],
                                                   swizzle[2],
@@ -3363,7 +3363,7 @@ parse_vp_instruction (GLcontext * ctx, const GLubyte ** inst,
 	    parse_extended_swizzle_mask (inst, swizzle, &negateMask);
 	    vp->SrcReg[0].File = file;
 	    vp->SrcReg[0].Index = index;
-	    vp->SrcReg[0].NegateBase = negateMask;
+	    vp->SrcReg[0].Negate = negateMask;
 	    vp->SrcReg[0].Swizzle = MAKE_SWIZZLE4(swizzle[0],
                                                   swizzle[1],
                                                   swizzle[2],
diff --git a/src/mesa/shader/nvfragparse.c b/src/mesa/shader/nvfragparse.c
index 56b7c29bea..0fd55524ab 100644
--- a/src/mesa/shader/nvfragparse.c
+++ b/src/mesa/shader/nvfragparse.c
@@ -957,6 +957,7 @@ Parse_VectorSrc(struct parse_state *parseState,
    GLfloat sign = 1.0F;
    GLubyte token[100];
    GLint idx;
+   GLuint negateBase, negateAbs;
 
    /*
     * First, take care of +/- and absolute value stuff.
@@ -968,21 +969,23 @@ Parse_VectorSrc(struct parse_state *parseState,
 
    if (Parse_String(parseState, "|")) {
       srcReg->Abs = GL_TRUE;
-      srcReg->NegateAbs = (sign < 0.0F) ? GL_TRUE : GL_FALSE;
+      negateAbs = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
 
       if (Parse_String(parseState, "-"))
-         srcReg->NegateBase = NEGATE_XYZW;
+         negateBase = NEGATE_XYZW;
       else if (Parse_String(parseState, "+"))
-         srcReg->NegateBase = NEGATE_NONE;
+         negateBase = NEGATE_NONE;
       else
-         srcReg->NegateBase = NEGATE_NONE;
+         negateBase = NEGATE_NONE;
    }
    else {
       srcReg->Abs = GL_FALSE;
-      srcReg->NegateAbs = GL_FALSE;
-      srcReg->NegateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
+      negateAbs = NEGATE_NONE;
+      negateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
    }
 
+   srcReg->Negate = srcReg->Abs ? negateAbs : negateBase;
+
    /* This should be the real src vector/register name */
    if (!Peek_Token(parseState, token))
       RETURN_ERROR;
@@ -1083,6 +1086,7 @@ Parse_ScalarSrcReg(struct parse_state *parseState,
    GLfloat sign = 1.0F;
    GLboolean needSuffix = GL_TRUE;
    GLint idx;
+   GLuint negateBase, negateAbs;
 
    /*
     * First, take care of +/- and absolute value stuff.
@@ -1094,21 +1098,23 @@ Parse_ScalarSrcReg(struct parse_state *parseState,
 
    if (Parse_String(parseState, "|")) {
       srcReg->Abs = GL_TRUE;
-      srcReg->NegateAbs = (sign < 0.0F) ? GL_TRUE : GL_FALSE;
+      negateAbs = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
 
       if (Parse_String(parseState, "-"))
-         srcReg->NegateBase = NEGATE_XYZW;
+         negateBase = NEGATE_XYZW;
       else if (Parse_String(parseState, "+"))
-         srcReg->NegateBase = NEGATE_NONE;
+         negateBase = NEGATE_NONE;
       else
-         srcReg->NegateBase = NEGATE_NONE;
+         negateBase = NEGATE_NONE;
    }
    else {
       srcReg->Abs = GL_FALSE;
-      srcReg->NegateAbs = GL_FALSE;
-      srcReg->NegateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
+      negateAbs = NEGATE_NONE;
+      negateBase = (sign < 0.0F) ? NEGATE_XYZW : NEGATE_NONE;
    }
 
+   srcReg->Negate = srcReg->Abs ? negateAbs : negateBase;
+
    if (!Peek_Token(parseState, token))
       RETURN_ERROR;
 
@@ -1247,9 +1253,8 @@ Parse_PrintInstruction(struct parse_state *parseState,
    }
 
    inst->SrcReg[0].Swizzle = SWIZZLE_NOOP;
-   inst->SrcReg[0].NegateBase = NEGATE_NONE;
    inst->SrcReg[0].Abs = GL_FALSE;
-   inst->SrcReg[0].NegateAbs = GL_FALSE;
+   inst->SrcReg[0].Negate = NEGATE_NONE;
 
    return GL_TRUE;
 }
diff --git a/src/mesa/shader/nvvertparse.c b/src/mesa/shader/nvvertparse.c
index 624262395b..f5e2df2670 100644
--- a/src/mesa/shader/nvvertparse.c
+++ b/src/mesa/shader/nvvertparse.c
@@ -641,12 +641,12 @@ Parse_SwizzleSrcReg(struct parse_state *parseState, struct prog_src_register *sr
       RETURN_ERROR;
    if (token[0] == '-') {
       (void) Parse_String(parseState, "-");
-      srcReg->NegateBase = NEGATE_XYZW;
+      srcReg->Negate = NEGATE_XYZW;
       if (!Peek_Token(parseState, token))
          RETURN_ERROR;
    }
    else {
-      srcReg->NegateBase = NEGATE_NONE;
+      srcReg->Negate = NEGATE_NONE;
    }
 
    /* Src reg can be R<n>, c[n], c[n +/- offset], or a named vertex attrib */
@@ -734,13 +734,13 @@ Parse_ScalarSrcReg(struct parse_state *parseState, struct prog_src_register *src
    if (!Peek_Token(parseState, token))
       RETURN_ERROR;
    if (token[0] == '-') {
-      srcReg->NegateBase = NEGATE_XYZW;
+      srcReg->Negate = NEGATE_XYZW;
       (void) Parse_String(parseState, "-"); /* consume '-' */
       if (!Peek_Token(parseState, token))
          RETURN_ERROR;
    }
    else {
-      srcReg->NegateBase = NEGATE_NONE;
+      srcReg->Negate = NEGATE_NONE;
    }
 
    /* Src reg can be R<n>, c[n], c[n +/- offset], or a named vertex attrib */
@@ -1062,7 +1062,7 @@ Parse_PrintInstruction(struct parse_state *parseState, struct prog_instruction *
          RETURN_ERROR;
 
       srcReg->RelAddr = GL_FALSE;
-      srcReg->NegateBase = NEGATE_NONE;
+      srcReg->Negate = NEGATE_NONE;
       srcReg->Swizzle = SWIZZLE_NOOP;
 
       /* Register can be R<n>, c[n], c[n +/- offset], a named vertex attrib,
diff --git a/src/mesa/shader/prog_execute.c b/src/mesa/shader/prog_execute.c
index bdac1d4f8a..68a59350a1 100644
--- a/src/mesa/shader/prog_execute.c
+++ b/src/mesa/shader/prog_execute.c
@@ -212,19 +212,14 @@ fetch_vector4(const struct prog_src_register *source,
       result[3] = src[GET_SWZ(source->Swizzle, 3)];
    }
 
-   if (source->NegateBase) {
-      result[0] = -result[0];
-      result[1] = -result[1];
-      result[2] = -result[2];
-      result[3] = -result[3];
-   }
    if (source->Abs) {
       result[0] = FABSF(result[0]);
       result[1] = FABSF(result[1]);
       result[2] = FABSF(result[2]);
       result[3] = FABSF(result[3]);
    }
-   if (source->NegateAbs) {
+   if (source->Negate) {
+      ASSERT(source->Negate == NEGATE_XYZW);
       result[0] = -result[0];
       result[1] = -result[1];
       result[2] = -result[2];
@@ -259,7 +254,7 @@ fetch_vector4ui(const struct prog_src_register *source,
       result[3] = src[GET_SWZ(source->Swizzle, 3)];
    }
 
-   /* Note: no NegateBase, Abs, NegateAbs here */
+   /* Note: no Negate or Abs here */
 }
 
 
@@ -299,19 +294,14 @@ fetch_vector4_deriv(GLcontext * ctx,
       result[2] = deriv[GET_SWZ(source->Swizzle, 2)];
       result[3] = deriv[GET_SWZ(source->Swizzle, 3)];
       
-      if (source->NegateBase) {
-         result[0] = -result[0];
-         result[1] = -result[1];
-         result[2] = -result[2];
-         result[3] = -result[3];
-      }
       if (source->Abs) {
          result[0] = FABSF(result[0]);
          result[1] = FABSF(result[1]);
          result[2] = FABSF(result[2]);
          result[3] = FABSF(result[3]);
       }
-      if (source->NegateAbs) {
+      if (source->Negate) {
+         ASSERT(source->Negate == NEGATE_XYZW);
          result[0] = -result[0];
          result[1] = -result[1];
          result[2] = -result[2];
@@ -336,13 +326,10 @@ fetch_vector1(const struct prog_src_register *source,
 
    result[0] = src[GET_SWZ(source->Swizzle, 0)];
 
-   if (source->NegateBase) {
-      result[0] = -result[0];
-   }
    if (source->Abs) {
       result[0] = FABSF(result[0]);
    }
-   if (source->NegateAbs) {
+   if (source->Negate) {
       result[0] = -result[0];
    }
 }
@@ -1514,7 +1501,7 @@ _mesa_execute_program(GLcontext * ctx,
                   ASSERT(swz <= 3);
                   result[i] = src[swz];
                }
-               if (source->NegateBase & (1 << i))
+               if (source->Negate & (1 << i))
                   result[i] = -result[i];
             }
             store_vector4(inst, machine, result);
diff --git a/src/mesa/shader/prog_instruction.h b/src/mesa/shader/prog_instruction.h
index 4adce11f95..3109f6cbae 100644
--- a/src/mesa/shader/prog_instruction.h
+++ b/src/mesa/shader/prog_instruction.h
@@ -261,37 +261,15 @@ struct prog_src_register
    GLuint Swizzle:12;
    GLuint RelAddr:1;
 
-   /**
-    * \name Source register "sign" control.
-    *
-    * The ARB and NV extensions allow varrying degrees of control over the
-    * sign of the source vector components.  These values allow enough control
-    * for all flavors of the extensions.
-    */
-   /*@{*/
-   /**
-    * Per-component negation for the SWZ instruction.  For non-SWZ
-    * instructions the only possible values are NEGATE_XYZW and NEGATE_NONE.
-    *
-    * \since
-    * ARB_vertex_program, ARB_fragment_program
-    */
-   GLuint NegateBase:4;
-
-   /**
-    * Take the component-wise absolute value.
-    *
-    * \since
-    * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2,
-    * NV_vertex_program2_option.
-    */
+   /** Take the component-wise absolute value */
    GLuint Abs:1;
 
    /**
-    * Post-absolute value negation (all components).
+    * Post-Abs negation.
+    * This will either be NEGATE_NONE or NEGATE_XYZW, except for the SWZ
+    * instruction which allows per-component negation.
     */
-   GLuint NegateAbs:1;
-   /*@}*/
+   GLuint Negate:4;
 };
 
 
diff --git a/src/mesa/shader/prog_print.c b/src/mesa/shader/prog_print.c
index b832ddb477..d73c619fea 100644
--- a/src/mesa/shader/prog_print.c
+++ b/src/mesa/shader/prog_print.c
@@ -325,19 +325,19 @@ reg_string(gl_register_file f, GLint index, gl_prog_print_mode mode,
  * \param extended  if true, also allow 0, 1 values
  */
 const char *
-_mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended)
+_mesa_swizzle_string(GLuint swizzle, GLuint negateMask, GLboolean extended)
 {
    static const char swz[] = "xyzw01!?";  /* See SWIZZLE_x definitions */
    static char s[20];
    GLuint i = 0;
 
-   if (!extended && swizzle == SWIZZLE_NOOP && negateBase == 0)
+   if (!extended && swizzle == SWIZZLE_NOOP && negateMask == 0)
       return ""; /* no swizzle/negation */
 
    if (!extended)
       s[i++] = '.';
 
-   if (negateBase & NEGATE_X)
+   if (negateMask & NEGATE_X)
       s[i++] = '-';
    s[i++] = swz[GET_SWZ(swizzle, 0)];
 
@@ -345,7 +345,7 @@ _mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended)
       s[i++] = ',';
    }
 
-   if (negateBase & NEGATE_Y)
+   if (negateMask & NEGATE_Y)
       s[i++] = '-';
    s[i++] = swz[GET_SWZ(swizzle, 1)];
 
@@ -353,7 +353,7 @@ _mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended)
       s[i++] = ',';
    }
 
-   if (negateBase & NEGATE_Z)
+   if (negateMask & NEGATE_Z)
       s[i++] = '-';
    s[i++] = swz[GET_SWZ(swizzle, 2)];
 
@@ -361,7 +361,7 @@ _mesa_swizzle_string(GLuint swizzle, GLuint negateBase, GLboolean extended)
       s[i++] = ',';
    }
 
-   if (negateBase & NEGATE_W)
+   if (negateMask & NEGATE_W)
       s[i++] = '-';
    s[i++] = swz[GET_SWZ(swizzle, 3)];
 
@@ -465,14 +465,14 @@ fprint_src_reg(FILE *f,
                  reg_string((gl_register_file) srcReg->File,
                             srcReg->Index, mode, srcReg->RelAddr, prog),
                  _mesa_swizzle_string(srcReg->Swizzle,
-                                      srcReg->NegateBase, GL_FALSE),
+                                      srcReg->Negate, GL_FALSE),
                  abs);
 #if 0
    _mesa_fprintf(f, "%s[%d]%s",
                  file_string((gl_register_file) srcReg->File, mode),
                  srcReg->Index,
                  _mesa_swizzle_string(srcReg->Swizzle,
-                                      srcReg->NegateBase, GL_FALSE));
+                                      srcReg->Negate, GL_FALSE));
 #endif
 }
 
@@ -566,7 +566,7 @@ _mesa_fprint_instruction_opt(FILE *f,
                                   mode),
                       inst->SrcReg[0].Index,
                       _mesa_swizzle_string(inst->SrcReg[0].Swizzle,
-                                           inst->SrcReg[0].NegateBase, GL_FALSE));
+                                           inst->SrcReg[0].Negate, GL_FALSE));
       }
       if (inst->Comment)
          _mesa_fprintf(f, "  # %s", inst->Comment);
@@ -583,7 +583,7 @@ _mesa_fprint_instruction_opt(FILE *f,
                                mode),
                    inst->SrcReg[0].Index,
                    _mesa_swizzle_string(inst->SrcReg[0].Swizzle,
-                                        inst->SrcReg[0].NegateBase, GL_TRUE));
+                                        inst->SrcReg[0].Negate, GL_TRUE));
       fprint_comment(f, inst);
       break;
    case OPCODE_TEX:
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index e283f8933b..ecd98dc85c 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -241,7 +241,7 @@ _mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
       inst->DstReg.WriteMask = WRITEMASK_X;
       inst->SrcReg[0].File = PROGRAM_TEMPORARY;
       inst->SrcReg[0].Index = fogFactorTemp;
-      inst->SrcReg[0].NegateBase = NEGATE_XYZW;
+      inst->SrcReg[0].Negate = NEGATE_XYZW;
       inst->SrcReg[0].Swizzle = SWIZZLE_XXXX;
       inst->SaturateMode = SATURATE_ZERO_ONE;
       inst++;
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 8493c490fb..3f455e0640 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -1135,7 +1135,7 @@ emit_negation(slang_emit_info *emitInfo, slang_ir_node *n)
                            n->Children[0]->Store,
                            NULL,
                            NULL);
-   inst->SrcReg[0].NegateBase = NEGATE_XYZW;
+   inst->SrcReg[0].Negate = NEGATE_XYZW;
    return inst;
 }
 
diff --git a/src/mesa/state_tracker/st_cb_bitmap.c b/src/mesa/state_tracker/st_cb_bitmap.c
index 3b2ad00f5c..fa4f4082a7 100644
--- a/src/mesa/state_tracker/st_cb_bitmap.c
+++ b/src/mesa/state_tracker/st_cb_bitmap.c
@@ -147,7 +147,7 @@ make_bitmap_fragment_program(GLcontext *ctx, GLuint samplerIndex)
       p->Instructions[ic].SrcReg[0].Swizzle = SWIZZLE_XXXX;
 
    p->Instructions[ic].SrcReg[0].Index = 0;
-   p->Instructions[ic].SrcReg[0].NegateBase = NEGATE_XYZW;
+   p->Instructions[ic].SrcReg[0].Negate = NEGATE_XYZW;
    ic++;
 
    /* END; */
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index ffa607dd87..43c9afccc3 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -275,8 +275,8 @@ compile_instruction(
       /* swizzle (ext swizzle also depends on negation) */
       {
          GLuint swz[4];
-         GLboolean extended = (inst->SrcReg[i].NegateBase != NEGATE_NONE &&
-                               inst->SrcReg[i].NegateBase != NEGATE_XYZW);
+         GLboolean extended = (inst->SrcReg[i].Negate != NEGATE_NONE &&
+                               inst->SrcReg[i].Negate != NEGATE_XYZW);
          for( j = 0; j < 4; j++ ) {
             swz[j] = GET_SWZ( inst->SrcReg[i].Swizzle, j );
             if (swz[j] > SWIZZLE_W)
@@ -296,20 +296,20 @@ compile_instruction(
          }
       }
 
-      if( inst->SrcReg[i].NegateBase == NEGATE_XYZW ) {
+      if( inst->SrcReg[i].Negate == NEGATE_XYZW ) {
          fullsrc->SrcRegister.Negate = 1;
       }
-      else if( inst->SrcReg[i].NegateBase != NEGATE_NONE ) {
-         if( inst->SrcReg[i].NegateBase & NEGATE_X ) {
+      else if( inst->SrcReg[i].Negate != NEGATE_NONE ) {
+         if( inst->SrcReg[i].Negate & NEGATE_X ) {
             fullsrc->SrcRegisterExtSwz.NegateX = 1;
          }
-         if( inst->SrcReg[i].NegateBase & NEGATE_Y ) {
+         if( inst->SrcReg[i].Negate & NEGATE_Y ) {
             fullsrc->SrcRegisterExtSwz.NegateY = 1;
          }
-         if( inst->SrcReg[i].NegateBase & NEGATE_Z ) {
+         if( inst->SrcReg[i].Negate & NEGATE_Z ) {
             fullsrc->SrcRegisterExtSwz.NegateZ = 1;
          }
-         if( inst->SrcReg[i].NegateBase & NEGATE_W ) {
+         if( inst->SrcReg[i].Negate & NEGATE_W ) {
             fullsrc->SrcRegisterExtSwz.NegateW = 1;
          }
       }
@@ -318,10 +318,6 @@ compile_instruction(
          fullsrc->SrcRegisterExtMod.Absolute = 1;
       }
 
-      if( inst->SrcReg[i].NegateAbs ) {
-         fullsrc->SrcRegisterExtMod.Negate = 1;
-      }
-
       if( inst->SrcReg[i].RelAddr ) {
          fullsrc->SrcRegister.Indirect = 1;
 
-- 
cgit v1.2.3


From 751f73e2812cf8185c775a91c16cf8565b85536d Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Fri, 1 May 2009 18:20:42 +0100
Subject: mesa/main: set PREFER_DP4 to match position_invarient code

This is a quick fix for z fighting in quake4 caused by the mismatch
between vertex transformation here and in the position_invarient code.
Full fix would be to make this driver-tunable and adjust both
position_invarient and ffvertex_prog.c code to respect driver
preferences.
---
 src/mesa/main/ffvertex_prog.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 1ce5685af4..82e1c4af66 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -319,7 +319,7 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
  * multiplications with DP4's or with MUL/MAD's?  SSE works better
  * with the latter, drivers may differ.
  */
-#define PREFER_DP4 0
+#define PREFER_DP4 1
 
 
 /* Use uregs to represent registers internally, translate to Mesa's
-- 
cgit v1.2.3


From b6e8256899a9a93c665c34e10efcc918f2fcc095 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Tue, 5 May 2009 12:12:28 +0100
Subject: mesa: more complete fix for transform_invarient glitches

Add a new flag mvp_with_dp4 in the context, and use that to switch
both ffvertex.c and programopt.c vertex transformation code to
either DP4 or MUL/MAD implementations.
---
 src/mesa/main/context.c             |  13 ++++
 src/mesa/main/context.h             |   4 ++
 src/mesa/main/ffvertex_prog.c       |  16 +++--
 src/mesa/main/mtypes.h              |   6 ++
 src/mesa/shader/programopt.c        | 119 +++++++++++++++++++++++++++++++++++-
 src/mesa/state_tracker/st_context.c |   6 ++
 6 files changed, 153 insertions(+), 11 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index 016284de9a..d780f91f04 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -1522,4 +1522,17 @@ _mesa_Flush(void)
 }
 
 
+/**
+ * Set mvp_with_dp4 flag.  If a driver has a preference for DP4 over
+ * MUL/MAD, or vice versa, call this function to register that.
+ * Otherwise we default to MUL/MAD.
+ */
+void
+_mesa_set_mvp_with_dp4( GLcontext *ctx,
+                        GLboolean flag )
+{
+   ctx->mvp_with_dp4 = flag;
+}
+
+
 /*@}*/
diff --git a/src/mesa/main/context.h b/src/mesa/main/context.h
index ecc1cec779..5b57d88029 100644
--- a/src/mesa/main/context.h
+++ b/src/mesa/main/context.h
@@ -151,6 +151,10 @@ extern struct _glapi_table *
 _mesa_get_dispatch(GLcontext *ctx);
 
 
+void
+_mesa_set_mvp_with_dp4( GLcontext *ctx,
+                        GLboolean flag );
+
 
 /** \name Miscellaneous */
 /*@{*/
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 82e1c4af66..43325b1352 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -315,12 +315,6 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
  */
 #define DISASSEM 0
 
-/* Should be tunable by the driver - do we want to do matrix
- * multiplications with DP4's or with MUL/MAD's?  SSE works better
- * with the latter, drivers may differ.
- */
-#define PREFER_DP4 1
-
 
 /* Use uregs to represent registers internally, translate to Mesa's
  * expected formats on emit.  
@@ -348,6 +342,7 @@ struct tnl_program {
    const struct state_key *state;
    struct gl_vertex_program *program;
    GLint max_inst;  /** number of instructions allocated for program */
+   GLboolean mvp_with_dp4;
    
    GLuint temp_in_use;
    GLuint temp_reserved;
@@ -775,7 +770,7 @@ static struct ureg get_eye_position( struct tnl_program *p )
 
       p->eye_position = reserve_temp(p);
 
-      if (PREFER_DP4) {
+      if (p->mvp_with_dp4) {
 	 register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
                                  0, modelview );
 
@@ -881,7 +876,7 @@ static void build_hpos( struct tnl_program *p )
    struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
    struct ureg mvp[4];
 
-   if (PREFER_DP4) {
+   if (p->mvp_with_dp4) {
       register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
 			      0, mvp );
       emit_matrix_transform_vec4( p, hpos, mvp, pos );
@@ -1574,7 +1569,7 @@ static void build_texture_transform( struct tnl_program *p )
 	    struct ureg in = (!is_undef(out_texgen) ? 
 			      out_texgen : 
 			      register_input(p, VERT_ATTRIB_TEX0+i));
-	    if (PREFER_DP4) {
+	    if (p->mvp_with_dp4) {
 	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
 				       0, texmat );
 	       emit_matrix_transform_vec4( p, out, texmat, in );
@@ -1708,6 +1703,7 @@ static void build_tnl_program( struct tnl_program *p )
 static void
 create_new_program( const struct state_key *key,
                     struct gl_vertex_program *program,
+                    GLboolean mvp_with_dp4,
                     GLuint max_temps)
 {
    struct tnl_program p;
@@ -1721,6 +1717,7 @@ create_new_program( const struct state_key *key,
    p.transformed_normal = undef;
    p.identity = undef;
    p.temp_in_use = 0;
+   p.mvp_with_dp4 = mvp_with_dp4;
    
    if (max_temps >= sizeof(int) * 8)
       p.temp_reserved = 0;
@@ -1776,6 +1773,7 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
          return NULL;
 
       create_new_program( &key, prog,
+                          ctx->mvp_with_dp4,
                           ctx->Const.VertexProgram.MaxTemps );
 
 #if 0
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 30c7cca3b5..ed6b1062bd 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -2977,6 +2977,12 @@ struct __GLcontextRec
    /** software compression/decompression supported or not */
    GLboolean Mesa_DXTn;
 
+   /** 
+    * Use dp4 (rather than mul/mad) instructions for position
+    * transformation?
+    */
+   GLboolean mvp_with_dp4;
+
    /** Core tnl module support */
    struct gl_tnl_module TnlModule;
 
diff --git a/src/mesa/shader/programopt.c b/src/mesa/shader/programopt.c
index ecd98dc85c..f70c75cec8 100644
--- a/src/mesa/shader/programopt.c
+++ b/src/mesa/shader/programopt.c
@@ -45,8 +45,8 @@
  * into a vertex program.
  * May be used to implement the position_invariant option.
  */
-void
-_mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog)
+static void
+_mesa_insert_mvp_dp4_code(GLcontext *ctx, struct gl_vertex_program *vprog)
 {
    struct prog_instruction *newInst;
    const GLuint origLen = vprog->Base.NumInstructions;
@@ -113,6 +113,121 @@ _mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog)
 }
 
 
+static void
+_mesa_insert_mvp_mad_code(GLcontext *ctx, struct gl_vertex_program *vprog)
+{
+   struct prog_instruction *newInst;
+   const GLuint origLen = vprog->Base.NumInstructions;
+   const GLuint newLen = origLen + 4;
+   GLuint hposTemp;
+   GLuint i;
+
+   /*
+    * Setup state references for the modelview/projection matrix.
+    * XXX we should check if these state vars are already declared.
+    */
+   static const gl_state_index mvpState[4][STATE_LENGTH] = {
+      { STATE_MVP_MATRIX, 0, 0, 0, STATE_MATRIX_TRANSPOSE },
+      { STATE_MVP_MATRIX, 0, 1, 1, STATE_MATRIX_TRANSPOSE },
+      { STATE_MVP_MATRIX, 0, 2, 2, STATE_MATRIX_TRANSPOSE },
+      { STATE_MVP_MATRIX, 0, 3, 3, STATE_MATRIX_TRANSPOSE },
+   };
+   GLint mvpRef[4];
+
+   for (i = 0; i < 4; i++) {
+      mvpRef[i] = _mesa_add_state_reference(vprog->Base.Parameters,
+                                            mvpState[i]);
+   }
+
+   /* Alloc storage for new instructions */
+   newInst = _mesa_alloc_instructions(newLen);
+   if (!newInst) {
+      _mesa_error(ctx, GL_OUT_OF_MEMORY,
+                  "glProgramString(inserting position_invariant code)");
+      return;
+   }
+
+   /* TEMP hposTemp; */
+   hposTemp = vprog->Base.NumTemporaries++;
+
+   /*
+    * Generated instructions:
+    *    emit_op2(p, OPCODE_MUL, tmp, 0, swizzle1(src,X), mat[0]);
+    *    emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Y), mat[1], tmp);
+    *    emit_op3(p, OPCODE_MAD, tmp, 0, swizzle1(src,Z), mat[2], tmp);
+    *    emit_op3(p, OPCODE_MAD, dest, 0, swizzle1(src,W), mat[3], tmp);
+    */
+   _mesa_init_instructions(newInst, 4);
+
+   newInst[0].Opcode = OPCODE_MUL;
+   newInst[0].DstReg.File = PROGRAM_TEMPORARY;
+   newInst[0].DstReg.Index = hposTemp;
+   newInst[0].DstReg.WriteMask = WRITEMASK_XYZW;
+   newInst[0].SrcReg[0].File = PROGRAM_INPUT;
+   newInst[0].SrcReg[0].Index = VERT_ATTRIB_POS;
+   newInst[0].SrcReg[0].Swizzle = SWIZZLE_XXXX;
+   newInst[0].SrcReg[1].File = PROGRAM_STATE_VAR;
+   newInst[0].SrcReg[1].Index = mvpRef[0];
+   newInst[0].SrcReg[1].Swizzle = SWIZZLE_NOOP;
+
+   for (i = 1; i <= 2; i++) {
+      newInst[i].Opcode = OPCODE_MAD;
+      newInst[i].DstReg.File = PROGRAM_TEMPORARY;
+      newInst[i].DstReg.Index = hposTemp;
+      newInst[i].DstReg.WriteMask = WRITEMASK_XYZW;
+      newInst[i].SrcReg[0].File = PROGRAM_INPUT;
+      newInst[i].SrcReg[0].Index = VERT_ATTRIB_POS;
+      newInst[i].SrcReg[0].Swizzle = MAKE_SWIZZLE4(i,i,i,i);
+      newInst[i].SrcReg[1].File = PROGRAM_STATE_VAR;
+      newInst[i].SrcReg[1].Index = mvpRef[i];
+      newInst[i].SrcReg[1].Swizzle = SWIZZLE_NOOP;
+      newInst[i].SrcReg[2].File = PROGRAM_TEMPORARY;
+      newInst[i].SrcReg[2].Index = hposTemp;
+      newInst[1].SrcReg[2].Swizzle = SWIZZLE_NOOP;
+   }
+
+   newInst[3].Opcode = OPCODE_MAD;
+   newInst[3].DstReg.File = PROGRAM_OUTPUT;
+   newInst[3].DstReg.Index = VERT_RESULT_HPOS;
+   newInst[3].DstReg.WriteMask = WRITEMASK_XYZW;
+   newInst[3].SrcReg[0].File = PROGRAM_INPUT;
+   newInst[3].SrcReg[0].Index = VERT_ATTRIB_POS;
+   newInst[3].SrcReg[0].Swizzle = SWIZZLE_WWWW;
+   newInst[3].SrcReg[1].File = PROGRAM_STATE_VAR;
+   newInst[3].SrcReg[1].Index = mvpRef[3];
+   newInst[3].SrcReg[1].Swizzle = SWIZZLE_NOOP;
+   newInst[3].SrcReg[2].File = PROGRAM_TEMPORARY;
+   newInst[3].SrcReg[2].Index = hposTemp;
+   newInst[3].SrcReg[2].Swizzle = SWIZZLE_NOOP;
+
+
+   /* Append original instructions after new instructions */
+   _mesa_copy_instructions (newInst + 4, vprog->Base.Instructions, origLen);
+
+   /* free old instructions */
+   _mesa_free_instructions(vprog->Base.Instructions, origLen);
+
+   /* install new instructions */
+   vprog->Base.Instructions = newInst;
+   vprog->Base.NumInstructions = newLen;
+   vprog->Base.InputsRead |= VERT_BIT_POS;
+   vprog->Base.OutputsWritten |= (1 << VERT_RESULT_HPOS);
+}
+
+
+void
+_mesa_insert_mvp_code(GLcontext *ctx, struct gl_vertex_program *vprog)
+{
+   if (ctx->mvp_with_dp4) 
+      _mesa_insert_mvp_dp4_code( ctx, vprog );
+   else
+      _mesa_insert_mvp_mad_code( ctx, vprog );
+}
+      
+
+
+
+
 
 /**
  * Append extra instructions onto the given fragment program to implement
diff --git a/src/mesa/state_tracker/st_context.c b/src/mesa/state_tracker/st_context.c
index 92a630eff9..2a1f21c51c 100644
--- a/src/mesa/state_tracker/st_context.c
+++ b/src/mesa/state_tracker/st_context.c
@@ -177,6 +177,12 @@ struct st_context *st_create_context(struct pipe_context *pipe,
 
    ctx = _mesa_create_context(visual, shareCtx, &funcs, NULL);
 
+   /* XXX: need a capability bit in gallium to query if the pipe
+    * driver prefers DP4 or MUL/MAD for vertex transformation.
+    */
+   if (debug_get_bool_option("MESA_MVP_DP4", FALSE))
+      _mesa_set_mvp_with_dp4( ctx, GL_TRUE );
+
    return st_create_context_priv(ctx, pipe);
 }
 
-- 
cgit v1.2.3


From f6c8ca06f649ae1367f61eca6fb7635862bd5584 Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Sun, 31 May 2009 19:01:55 -0700
Subject: mesa: fix material inputs in ffvertex_prog.c

Varying material inputs were not being picked up from the same slots
where the VBO code is currently placing them (GENERIC0 and above).
Most often they were just being ignored.
---
 src/mesa/main/ffvertex_prog.c | 54 ++++++++++++-------------------------------
 1 file changed, 15 insertions(+), 39 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 43325b1352..b939401de1 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -48,11 +48,9 @@
 
 struct state_key {
    unsigned light_color_material_mask:12;
-   unsigned light_material_mask:12;
    unsigned light_global_enabled:1;
    unsigned light_local_viewer:1;
    unsigned light_twoside:1;
-   unsigned light_color_material:1;
    unsigned material_shininess_is_zero:1;
    unsigned need_eye_coords:1;
    unsigned normalize:1;
@@ -125,27 +123,6 @@ static GLuint translate_texgen( GLboolean enabled, GLenum mode )
 }
 
 
-/**
- * Returns bitmask of flags indicating which materials are set per-vertex
- * in the current VB.
- * XXX get these from the VBO...
- */
-static GLbitfield
-tnl_get_per_vertex_materials(GLcontext *ctx)
-{
-   GLbitfield mask = 0x0;
-#if 0
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   struct vertex_buffer *VB = &tnl->vb;
-   GLuint i;
-
-   for (i = _TNL_FIRST_MAT; i <= _TNL_LAST_MAT; i++) 
-      if (VB->AttribPtr[i] && VB->AttribPtr[i]->stride) 
-         mask |= 1 << (i - _TNL_FIRST_MAT);
-#endif
-   return mask;
-}
-
 
 /**
  * Should fog be computed per-vertex?
@@ -168,10 +145,11 @@ static GLboolean check_active_shininess( GLcontext *ctx,
 {
    GLuint bit = 1 << (MAT_ATTRIB_FRONT_SHININESS + side);
 
-   if (key->light_color_material_mask & bit)
+   if ((key->varying_vp_inputs & VERT_BIT_COLOR0) &&
+       (key->light_color_material_mask & bit))
       return GL_TRUE;
 
-   if (key->light_material_mask & bit)
+   if (key->varying_vp_inputs & (bit << 16))
       return GL_TRUE;
 
    if (ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS + side][0] != 0.0F)
@@ -216,12 +194,9 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
 	 key->light_twoside = 1;
 
       if (ctx->Light.ColorMaterialEnabled) {
-	 key->light_color_material = 1;
 	 key->light_color_material_mask = ctx->Light.ColorMaterialBitmask;
       }
 
-      key->light_material_mask = tnl_get_per_vertex_materials(ctx);
-
       for (i = 0; i < MAX_LIGHTS; i++) {
 	 struct gl_light *light = &ctx->Light.Light[i];
 
@@ -481,9 +456,9 @@ static struct ureg register_param5(struct tnl_program *p,
  */
 static struct ureg register_input( struct tnl_program *p, GLuint input )
 {
-   /* Material attribs are passed here as inputs >= 32
-    */
-   if (input >= 32 || (p->state->varying_vp_inputs & (1<<input))) {
+   assert(input < 32);
+
+   if (p->state->varying_vp_inputs & (1<<input)) {
       p->program->Base.InputsRead |= (1<<input);
       return make_ureg(PROGRAM_INPUT, input);
    }
@@ -903,18 +878,15 @@ static void set_material_flags( struct tnl_program *p )
    p->color_materials = 0;
    p->materials = 0;
 
-   if (p->state->light_color_material) {
-      p->materials = 
+   if (p->state->varying_vp_inputs & VERT_BIT_COLOR0) {
+      p->materials =
 	 p->color_materials = p->state->light_color_material_mask;
    }
 
-   p->materials |= p->state->light_material_mask;
+   p->materials |= (p->state->varying_vp_inputs >> 16);
 }
 
 
-/* XXX temporary!!! */
-#define _TNL_ATTRIB_MAT_FRONT_AMBIENT 32
-
 static struct ureg get_material( struct tnl_program *p, GLuint side, 
 				 GLuint property )
 {
@@ -922,8 +894,12 @@ static struct ureg get_material( struct tnl_program *p, GLuint side,
 
    if (p->color_materials & (1<<attrib))
       return register_input(p, VERT_ATTRIB_COLOR0);
-   else if (p->materials & (1<<attrib)) 
-      return register_input( p, attrib + _TNL_ATTRIB_MAT_FRONT_AMBIENT );
+   else if (p->materials & (1<<attrib)) {
+      /* Put material values in the GENERIC slots -- they are not used
+       * for anything in fixed function mode.
+       */
+      return register_input( p, attrib + VERT_ATTRIB_GENERIC0 );
+   }
    else
       return register_param3( p, STATE_MATERIAL, side, property );
 }
-- 
cgit v1.2.3


From 79047cc1ddf0332e5b64c9f05a19885cc36c76ae Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Sun, 31 May 2009 19:03:02 -0700
Subject: mesa: remove dead vertex fog code from ffvertex_prog.c

---
 src/mesa/main/ffvertex_prog.c | 87 ++-----------------------------------------
 1 file changed, 3 insertions(+), 84 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index b939401de1..a8e48253a5 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -57,9 +57,7 @@ struct state_key {
    unsigned rescale_normals:1;
 
    unsigned fog_source_is_depth:1;
-   unsigned tnl_do_vertex_fog:1;
    unsigned separate_specular:1;
-   unsigned fog_mode:2;
    unsigned point_attenuated:1;
    unsigned point_array:1;
    unsigned texture_enabled_global:1;
@@ -83,23 +81,6 @@ struct state_key {
 };
 
 
-
-#define FOG_NONE   0
-#define FOG_LINEAR 1
-#define FOG_EXP    2
-#define FOG_EXP2   3
-
-static GLuint translate_fog_mode( GLenum mode )
-{
-   switch (mode) {
-   case GL_LINEAR: return FOG_LINEAR;
-   case GL_EXP: return FOG_EXP;
-   case GL_EXP2: return FOG_EXP2;
-   default: return FOG_NONE;
-   }
-}
-
-
 #define TXG_NONE           0
 #define TXG_OBJ_LINEAR     1
 #define TXG_EYE_LINEAR     2
@@ -124,21 +105,6 @@ static GLuint translate_texgen( GLboolean enabled, GLenum mode )
 
 
 
-/**
- * Should fog be computed per-vertex?
- */
-static GLboolean
-tnl_get_per_vertex_fog(GLcontext *ctx)
-{
-#if 0
-   TNLcontext *tnl = TNL_CONTEXT(ctx);
-   return tnl->_DoVertexFog;
-#else
-   return GL_FALSE;
-#endif
-}
-
-
 static GLboolean check_active_shininess( GLcontext *ctx,
                                          const struct state_key *key,
                                          GLuint side )
@@ -234,12 +200,8 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
    if (ctx->Transform.RescaleNormals)
       key->rescale_normals = 1;
 
-   key->fog_mode = translate_fog_mode(fp->FogOption);
-   
    if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT)
       key->fog_source_is_depth = 1;
-   
-   key->tnl_do_vertex_fog = tnl_get_per_vertex_fog(ctx);
 
    if (ctx->Point._Attenuated)
       key->point_attenuated = 1;
@@ -1344,52 +1306,10 @@ static void build_fog( struct tnl_program *p )
       input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
    }
 
-   if (p->state->fog_mode && p->state->tnl_do_vertex_fog) {
-      struct ureg params = register_param2(p, STATE_INTERNAL,
-					   STATE_FOG_PARAMS_OPTIMIZED);
-      struct ureg tmp = get_temp(p);
-      GLboolean useabs = (p->state->fog_mode != FOG_EXP2);
-
-      if (useabs) {
-	 emit_op1(p, OPCODE_ABS, tmp, 0, input);
-      }
-
-      switch (p->state->fog_mode) {
-      case FOG_LINEAR: {
-	 struct ureg id = get_identity_param(p);
-	 emit_op3(p, OPCODE_MAD, tmp, 0, useabs ? tmp : input,
-			swizzle1(params,X), swizzle1(params,Y));
-	 emit_op2(p, OPCODE_MAX, tmp, 0, tmp, swizzle1(id,X)); /* saturate */
-	 emit_op2(p, OPCODE_MIN, fog, WRITEMASK_X, tmp, swizzle1(id,W));
-	 break;
-      }
-      case FOG_EXP:
-	 emit_op2(p, OPCODE_MUL, tmp, 0, useabs ? tmp : input,
-			swizzle1(params,Z));
-	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, negate(tmp));
-	 break;
-      case FOG_EXP2:
-	 emit_op2(p, OPCODE_MUL, tmp, 0, input, swizzle1(params,W));
-	 emit_op2(p, OPCODE_MUL, tmp, 0, tmp, tmp);
-	 emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, negate(tmp));
-	 break;
-      }
-
-      release_temp(p, tmp);
-   }
-   else {
-      /* results = incoming fog coords (compute fog per-fragment later) 
-       *
-       * KW:  Is it really necessary to do anything in this case?
-       * BP: Yes, we always need to compute the absolute value, unless
-       * we want to push that down into the fragment program...
-       */
-      GLboolean useabs = GL_TRUE;
-      emit_op1(p, useabs ? OPCODE_ABS : OPCODE_MOV, fog, WRITEMASK_X, input);
-   }
+   emit_op1(p, OPCODE_ABS, fog, WRITEMASK_X, input);
 }
 
- 
+
 static void build_reflect_texgen( struct tnl_program *p,
 				  struct ureg dest,
 				  GLuint writemask )
@@ -1646,8 +1566,7 @@ static void build_tnl_program( struct tnl_program *p )
       }
    }
 
-   if ((p->state->fragprog_inputs_read & FRAG_BIT_FOGC) ||
-       p->state->fog_mode != FOG_NONE)
+   if (p->state->fragprog_inputs_read & FRAG_BIT_FOGC)
       build_fog(p);
 
    if (p->state->fragprog_inputs_read & FRAG_BITS_TEX_ANY)
-- 
cgit v1.2.3


From cea259f0392766e8a2e4749c6d5277d866abbc3b Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Sun, 31 May 2009 19:03:46 -0700
Subject: mesa: remove dead constant pointsize code from ffvertex_prog.c

---
 src/mesa/main/ffvertex_prog.c | 17 -----------------
 1 file changed, 17 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index a8e48253a5..c38dba04c7 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1524,17 +1524,6 @@ static void build_atten_pointsize( struct tnl_program *p )
 }
 
 
-/**
- * Emit constant point size.
- */
-static void build_constant_pointsize( struct tnl_program *p )
-{
-   struct ureg state_size = register_param1(p, STATE_POINT_SIZE);
-   struct ureg out = register_output(p, VERT_RESULT_PSIZ);
-   emit_op1(p, OPCODE_MOV, out, WRITEMASK_X, state_size);
-}
-
-
 /**
  * Pass-though per-vertex point size, from user's point size array.
  */
@@ -1576,12 +1565,6 @@ static void build_tnl_program( struct tnl_program *p )
       build_atten_pointsize(p);
    else if (p->state->point_array)
       build_array_pointsize(p);
-#if 0
-   else
-      build_constant_pointsize(p);
-#else
-   (void) build_constant_pointsize;
-#endif
 
    /* Finish up:
     */
-- 
cgit v1.2.3


From 6dfb89e4a09a269cbd5c4dc725881fbd42eaa18a Mon Sep 17 00:00:00 2001
From: Keith Whitwell <keithw@vmware.com>
Date: Sun, 31 May 2009 19:04:06 -0700
Subject: mesa: remove whitespace

---
 src/mesa/main/ffvertex_prog.c | 216 +++++++++++++++++++++---------------------
 1 file changed, 108 insertions(+), 108 deletions(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index c38dba04c7..80dde4b5aa 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1,8 +1,8 @@
 /**************************************************************************
- * 
+ *
  * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,11 +10,11 @@
  * distribute, sub license, and/or sell copies of the Software, and to
  * permit persons to whom the Software is furnished to do so, subject to
  * the following conditions:
- * 
+ *
  * The above copyright notice and this permission notice (including the
  * next paragraph) shall be included in all copies or substantial portions
  * of the Software.
- * 
+ *
  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
@@ -22,7 +22,7 @@
  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
+ *
  **************************************************************************/
 
 /**
@@ -69,7 +69,7 @@ struct state_key {
       unsigned light_enabled:1;
       unsigned light_eyepos3_is_zero:1;
       unsigned light_spotcutoff_is_180:1;
-      unsigned light_attenuated:1;      
+      unsigned light_attenuated:1;
       unsigned texunit_really_enabled:1;
       unsigned texmat_enabled:1;
       unsigned texgen_enabled:4;
@@ -171,7 +171,7 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
 
 	    if (light->EyePosition[3] == 0.0)
 	       key->unit[i].light_eyepos3_is_zero = 1;
-	    
+
 	    if (light->SpotCutoff == 180.0)
 	       key->unit[i].light_spotcutoff_is_180 = 1;
 
@@ -215,29 +215,29 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
        ctx->Texture._TexMatEnabled ||
        ctx->Texture._EnabledUnits)
       key->texture_enabled_global = 1;
-      
+
    for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
 
       if (texUnit->_ReallyEnabled)
 	 key->unit[i].texunit_really_enabled = 1;
 
-      if (ctx->Texture._TexMatEnabled & ENABLE_TEXMAT(i))      
+      if (ctx->Texture._TexMatEnabled & ENABLE_TEXMAT(i))
 	 key->unit[i].texmat_enabled = 1;
-      
+
       if (texUnit->TexGenEnabled) {
 	 key->unit[i].texgen_enabled = 1;
-      
-	 key->unit[i].texgen_mode0 = 
+
+	 key->unit[i].texgen_mode0 =
 	    translate_texgen( texUnit->TexGenEnabled & (1<<0),
 			      texUnit->GenS.Mode );
-	 key->unit[i].texgen_mode1 = 
+	 key->unit[i].texgen_mode1 =
 	    translate_texgen( texUnit->TexGenEnabled & (1<<1),
 			      texUnit->GenT.Mode );
-	 key->unit[i].texgen_mode2 = 
+	 key->unit[i].texgen_mode2 =
 	    translate_texgen( texUnit->TexGenEnabled & (1<<2),
 			      texUnit->GenR.Mode );
-	 key->unit[i].texgen_mode3 = 
+	 key->unit[i].texgen_mode3 =
 	    translate_texgen( texUnit->TexGenEnabled & (1<<3),
 			      texUnit->GenQ.Mode );
       }
@@ -245,7 +245,7 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
 }
 
 
-   
+
 /* Very useful debugging tool - produces annotated listing of
  * generated program with line/function references for each
  * instruction back into this file:
@@ -254,7 +254,7 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
 
 
 /* Use uregs to represent registers internally, translate to Mesa's
- * expected formats on emit.  
+ * expected formats on emit.
  *
  * NOTE: These are passed by value extensively in this file rather
  * than as usual by pointer reference.  If this disturbs you, try
@@ -280,10 +280,10 @@ struct tnl_program {
    struct gl_vertex_program *program;
    GLint max_inst;  /** number of instructions allocated for program */
    GLboolean mvp_with_dp4;
-   
+
    GLuint temp_in_use;
    GLuint temp_reserved;
-   
+
    struct ureg eye_position;
    struct ureg eye_position_z;
    struct ureg eye_position_normalized;
@@ -387,7 +387,7 @@ static void release_temps( struct tnl_program *p )
 }
 
 
-static struct ureg register_param5(struct tnl_program *p, 
+static struct ureg register_param5(struct tnl_program *p,
 				   GLint s0,
 				   GLint s1,
 				   GLint s2,
@@ -440,7 +440,7 @@ static struct ureg register_output( struct tnl_program *p, GLuint output )
 }
 
 
-static struct ureg register_const4f( struct tnl_program *p, 
+static struct ureg register_const4f( struct tnl_program *p,
 			      GLfloat s0,
 			      GLfloat s1,
 			      GLfloat s2,
@@ -472,7 +472,7 @@ static GLboolean is_undef( struct ureg reg )
 
 static struct ureg get_identity_param( struct tnl_program *p )
 {
-   if (is_undef(p->identity)) 
+   if (is_undef(p->identity))
       p->identity = register_const4f(p, 0,0,0,1);
 
    return p->identity;
@@ -491,7 +491,7 @@ static void register_matrix_param5( struct tnl_program *p,
    /* This is a bit sad as the support is there to pull the whole
     * matrix out in one go:
     */
-   for (i = 0; i <= s3 - s2; i++) 
+   for (i = 0; i <= s3 - s2; i++)
       matrix[i] = register_param5( p, s0, s1, i, i, s4 );
 }
 
@@ -516,7 +516,7 @@ static void emit_dst( struct prog_dst_register *dst,
    dst->File = reg.file;
    dst->Index = reg.idx;
    /* allow zero as a shorthand for xyzw */
-   dst->WriteMask = mask ? mask : WRITEMASK_XYZW; 
+   dst->WriteMask = mask ? mask : WRITEMASK_XYZW;
    dst->CondMask = COND_TR;  /* always pass cond test */
    dst->CondSwizzle = SWIZZLE_NOOP;
    dst->CondSrc = 0;
@@ -531,12 +531,12 @@ static void debug_insn( struct prog_instruction *inst, const char *fn,
 {
    if (DISASSEM) {
       static const char *last_fn;
-   
+
       if (fn != last_fn) {
 	 last_fn = fn;
 	 _mesa_printf("%s:\n", fn);
       }
-	 
+
       _mesa_printf("%d:\t", line);
       _mesa_print_instruction(inst);
    }
@@ -555,7 +555,7 @@ static void emit_op3fn(struct tnl_program *p,
 {
    GLuint nr;
    struct prog_instruction *inst;
-      
+
    assert((GLint) p->program->Base.NumInstructions <= p->max_inst);
 
    if (p->program->Base.NumInstructions == p->max_inst) {
@@ -580,16 +580,16 @@ static void emit_op3fn(struct tnl_program *p,
 
       p->program->Base.Instructions = newInst;
    }
-      
+
    nr = p->program->Base.NumInstructions++;
 
    inst = &p->program->Base.Instructions[nr];
-   inst->Opcode = (enum prog_opcode) op; 
+   inst->Opcode = (enum prog_opcode) op;
    inst->Data = 0;
-   
+
    emit_arg( &inst->SrcReg[0], src0 );
    emit_arg( &inst->SrcReg[1], src1 );
-   emit_arg( &inst->SrcReg[2], src2 );   
+   emit_arg( &inst->SrcReg[2], src2 );
 
    emit_dst( &inst->DstReg, dest, mask );
 
@@ -609,7 +609,7 @@ static void emit_op3fn(struct tnl_program *p,
 
 static struct ureg make_temp( struct tnl_program *p, struct ureg reg )
 {
-   if (reg.file == PROGRAM_TEMPORARY && 
+   if (reg.file == PROGRAM_TEMPORARY &&
        !(p->temp_reserved & (1<<reg.idx)))
       return reg;
    else {
@@ -690,19 +690,19 @@ static void emit_normalize_vec3( struct tnl_program *p,
 }
 
 
-static void emit_passthrough( struct tnl_program *p, 
+static void emit_passthrough( struct tnl_program *p,
 			      GLuint input,
 			      GLuint output )
 {
    struct ureg out = register_output(p, output);
-   emit_op1(p, OPCODE_MOV, out, 0, register_input(p, input)); 
+   emit_op1(p, OPCODE_MOV, out, 0, register_input(p, input));
 }
 
 
 static struct ureg get_eye_position( struct tnl_program *p )
 {
    if (is_undef(p->eye_position)) {
-      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+      struct ureg pos = register_input( p, VERT_ATTRIB_POS );
       struct ureg modelview[4];
 
       p->eye_position = reserve_temp(p);
@@ -720,18 +720,18 @@ static struct ureg get_eye_position( struct tnl_program *p )
 	 emit_transpose_matrix_transform_vec4(p, p->eye_position, modelview, pos);
       }
    }
-   
+
    return p->eye_position;
 }
 
 
 static struct ureg get_eye_position_z( struct tnl_program *p )
 {
-   if (!is_undef(p->eye_position)) 
+   if (!is_undef(p->eye_position))
       return swizzle1(p->eye_position, Z);
 
    if (is_undef(p->eye_position_z)) {
-      struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+      struct ureg pos = register_input( p, VERT_ATTRIB_POS );
       struct ureg modelview[4];
 
       p->eye_position_z = reserve_temp(p);
@@ -741,10 +741,10 @@ static struct ureg get_eye_position_z( struct tnl_program *p )
 
       emit_op2(p, OPCODE_DP4, p->eye_position_z, 0, pos, modelview[2]);
    }
-   
+
    return p->eye_position_z;
 }
-   
+
 
 static struct ureg get_eye_position_normalized( struct tnl_program *p )
 {
@@ -753,7 +753,7 @@ static struct ureg get_eye_position_normalized( struct tnl_program *p )
       p->eye_position_normalized = reserve_temp(p);
       emit_normalize_vec3(p, p->eye_position_normalized, eye);
    }
-   
+
    return p->eye_position_normalized;
 }
 
@@ -767,7 +767,7 @@ static struct ureg get_transformed_normal( struct tnl_program *p )
    {
       p->transformed_normal = register_input(p, VERT_ATTRIB_NORMAL );
    }
-   else if (is_undef(p->transformed_normal)) 
+   else if (is_undef(p->transformed_normal))
    {
       struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
       struct ureg mvinv[3];
@@ -798,7 +798,7 @@ static struct ureg get_transformed_normal( struct tnl_program *p )
 	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal, rescale );
          normal = transformed_normal;
       }
-      
+
       assert(normal.file == PROGRAM_TEMPORARY);
       p->transformed_normal = normal;
    }
@@ -809,17 +809,17 @@ static struct ureg get_transformed_normal( struct tnl_program *p )
 
 static void build_hpos( struct tnl_program *p )
 {
-   struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
+   struct ureg pos = register_input( p, VERT_ATTRIB_POS );
    struct ureg hpos = register_output( p, VERT_RESULT_HPOS );
    struct ureg mvp[4];
 
    if (p->mvp_with_dp4) {
-      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
+      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
 			      0, mvp );
       emit_matrix_transform_vec4( p, hpos, mvp, pos );
    }
    else {
-      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3, 
+      register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
 			      STATE_MATRIX_TRANSPOSE, mvp );
       emit_transpose_matrix_transform_vec4( p, hpos, mvp, pos );
    }
@@ -849,7 +849,7 @@ static void set_material_flags( struct tnl_program *p )
 }
 
 
-static struct ureg get_material( struct tnl_program *p, GLuint side, 
+static struct ureg get_material( struct tnl_program *p, GLuint side,
 				 GLuint property )
 {
    GLuint attrib = material_attrib(side, property);
@@ -890,7 +890,7 @@ static struct ureg get_scenecolor( struct tnl_program *p, GLuint side )
       struct ureg material_ambient = get_material(p, side, STATE_AMBIENT);
       struct ureg material_diffuse = get_material(p, side, STATE_DIFFUSE);
       struct ureg tmp = make_temp(p, material_diffuse);
-      emit_op3(p, OPCODE_MAD, tmp, WRITEMASK_XYZ, lm_ambient, 
+      emit_op3(p, OPCODE_MAD, tmp, WRITEMASK_XYZ, lm_ambient,
 	       material_ambient, material_emission);
       return tmp;
    }
@@ -899,12 +899,12 @@ static struct ureg get_scenecolor( struct tnl_program *p, GLuint side )
 }
 
 
-static struct ureg get_lightprod( struct tnl_program *p, GLuint light, 
+static struct ureg get_lightprod( struct tnl_program *p, GLuint light,
 				  GLuint side, GLuint property )
 {
    GLuint attrib = material_attrib(side, property);
    if (p->materials & (1<<attrib)) {
-      struct ureg light_value = 
+      struct ureg light_value =
 	 register_param3(p, STATE_LIGHT, light, property);
       struct ureg material_value = get_material(p, side, property);
       struct ureg tmp = get_temp(p);
@@ -917,7 +917,7 @@ static struct ureg get_lightprod( struct tnl_program *p, GLuint light,
 
 
 static struct ureg calculate_light_attenuation( struct tnl_program *p,
-						GLuint i, 
+						GLuint i,
 						struct ureg VPpli,
 						struct ureg dist )
 {
@@ -946,27 +946,27 @@ static struct ureg calculate_light_attenuation( struct tnl_program *p,
     */
    if (p->state->unit[i].light_attenuated) {
       /* 1/d,d,d,1/d */
-      emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist); 
+      emit_op1(p, OPCODE_RCP, dist, WRITEMASK_YZ, dist);
       /* 1,d,d*d,1/d */
-      emit_op2(p, OPCODE_MUL, dist, WRITEMASK_XZ, dist, swizzle1(dist,Y)); 
+      emit_op2(p, OPCODE_MUL, dist, WRITEMASK_XZ, dist, swizzle1(dist,Y));
       /* 1/dist-atten */
-      emit_op2(p, OPCODE_DP3, dist, 0, attenuation, dist); 
+      emit_op2(p, OPCODE_DP3, dist, 0, attenuation, dist);
 
       if (!p->state->unit[i].light_spotcutoff_is_180) {
 	 /* dist-atten */
-	 emit_op1(p, OPCODE_RCP, dist, 0, dist); 
+	 emit_op1(p, OPCODE_RCP, dist, 0, dist);
 	 /* spot-atten * dist-atten */
-	 emit_op2(p, OPCODE_MUL, att, 0, dist, att);	
+	 emit_op2(p, OPCODE_MUL, att, 0, dist, att);
       }
       else {
 	 /* dist-atten */
-	 emit_op1(p, OPCODE_RCP, att, 0, dist); 
+	 emit_op1(p, OPCODE_RCP, att, 0, dist);
       }
    }
 
    return att;
 }
-						
+
 
 /**
  * Compute:
@@ -985,7 +985,7 @@ static void emit_degenerate_lit( struct tnl_program *p,
 
    /* MAX lit, id, dots;
     */
-   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_XYZW, id, dots); 
+   emit_op2(p, OPCODE_MAX, lit, WRITEMASK_XYZW, id, dots);
 
    /* result[2] = (in > 0 ? 1 : 0)
     * SLT lit.z, id.z, dots;   # lit.z = (0 < dots.z) ? 1 : 0
@@ -1018,10 +1018,10 @@ static void build_lighting( struct tnl_program *p )
     * dots.w = front.shininess
     */
 
-   for (i = 0; i < MAX_LIGHTS; i++) 
+   for (i = 0; i < MAX_LIGHTS; i++)
       if (p->state->unit[i].light_enabled)
 	 nr_lights++;
-   
+
    set_material_flags(p);
 
    {
@@ -1044,7 +1044,7 @@ static void build_lighting( struct tnl_program *p )
           * The negation will be un-done later in the back-face code below.
           */
          struct ureg shininess = get_material(p, 1, STATE_SHININESS);
-         emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z, 
+         emit_op1(p, OPCODE_MOV, dots, WRITEMASK_Z,
                   negate(swizzle1(shininess,X)));
          release_temp(p, shininess);
       }
@@ -1072,12 +1072,12 @@ static void build_lighting( struct tnl_program *p )
       struct ureg res0 = register_output( p, VERT_RESULT_BFC0 );
       emit_op1(p, OPCODE_MOV, res0, 0, _bfc0);
    }
-      
+
    if (twoside && separate) {
       struct ureg res1 = register_output( p, VERT_RESULT_BFC1 );
       emit_op1(p, OPCODE_MOV, res1, 0, _bfc1);
    }
-      
+
    if (nr_lights == 0) {
       release_temps(p);
       return;
@@ -1087,16 +1087,16 @@ static void build_lighting( struct tnl_program *p )
       if (p->state->unit[i].light_enabled) {
 	 struct ureg half = undef;
 	 struct ureg att = undef, VPpli = undef;
-	  
+
 	 count++;
 
 	 if (p->state->unit[i].light_eyepos3_is_zero) {
 	    /* Can used precomputed constants in this case.
 	     * Attenuation never applies to infinite lights.
 	     */
-	    VPpli = register_param3(p, STATE_INTERNAL, 
-				    STATE_LIGHT_POSITION_NORMALIZED, i); 
-            
+	    VPpli = register_param3(p, STATE_INTERNAL,
+				    STATE_LIGHT_POSITION_NORMALIZED, i);
+
             if (!p->state->material_shininess_is_zero) {
                if (p->state->light_local_viewer) {
                   struct ureg eye_hat = get_eye_position_normalized(p);
@@ -1105,22 +1105,22 @@ static void build_lighting( struct tnl_program *p )
                   emit_normalize_vec3(p, half, half);
                }
                else {
-                  half = register_param3(p, STATE_INTERNAL, 
+                  half = register_param3(p, STATE_INTERNAL,
                                          STATE_LIGHT_HALF_VECTOR, i);
                }
             }
 	 }
 	 else {
-	    struct ureg Ppli = register_param3(p, STATE_INTERNAL, 
-					       STATE_LIGHT_POSITION, i); 
+	    struct ureg Ppli = register_param3(p, STATE_INTERNAL,
+					       STATE_LIGHT_POSITION, i);
 	    struct ureg V = get_eye_position(p);
 	    struct ureg dist = get_temp(p);
 
-	    VPpli = get_temp(p); 
- 
+	    VPpli = get_temp(p);
+
 	    /* Calculate VPpli vector
 	     */
-	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V); 
+	    emit_op2(p, OPCODE_SUB, VPpli, 0, Ppli, V);
 
 	    /* Normalize VPpli.  The dist value also used in
 	     * attenuation below.
@@ -1130,7 +1130,7 @@ static void build_lighting( struct tnl_program *p )
 	    emit_op2(p, OPCODE_MUL, VPpli, 0, VPpli, dist);
 
 	    /* Calculate attenuation:
-	     */ 
+	     */
 	    if (!p->state->unit[i].light_spotcutoff_is_180 ||
 		p->state->unit[i].light_attenuated) {
 	       att = calculate_light_attenuation(p, i, VPpli, dist);
@@ -1146,7 +1146,7 @@ static void build_lighting( struct tnl_program *p )
                   emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
                }
                else {
-                  struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z); 
+                  struct ureg z_dir = swizzle(get_identity_param(p),X,Y,W,Z);
                   emit_op2(p, OPCODE_ADD, half, 0, VPpli, z_dir);
                }
 
@@ -1215,7 +1215,7 @@ static void build_lighting( struct tnl_program *p )
 
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _col0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _col1);
-      
+
 	    release_temp(p, ambient);
 	    release_temp(p, diffuse);
 	    release_temp(p, specular);
@@ -1229,7 +1229,7 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg specular = get_lightprod(p, i, 1, STATE_SPECULAR);
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
-	       
+
 	    if (count == nr_lights) {
 	       if (separate) {
 		  mask0 = WRITEMASK_XYZ;
@@ -1319,9 +1319,9 @@ static void build_reflect_texgen( struct tnl_program *p,
    struct ureg tmp = get_temp(p);
 
    /* n.u */
-   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
+   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat);
    /* 2n.u */
-   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
+   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp);
    /* (-2n.u)n + u */
    emit_op3(p, OPCODE_MAD, dest, writemask, negate(tmp), normal, eye_hat);
 
@@ -1350,22 +1350,22 @@ static void build_sphere_texgen( struct tnl_program *p,
     */
 
    /* n.u */
-   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat); 
+   emit_op2(p, OPCODE_DP3, tmp, 0, normal, eye_hat);
    /* 2n.u */
-   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp); 
+   emit_op2(p, OPCODE_ADD, tmp, 0, tmp, tmp);
    /* (-2n.u)n + u */
-   emit_op3(p, OPCODE_MAD, r, 0, negate(tmp), normal, eye_hat); 
+   emit_op3(p, OPCODE_MAD, r, 0, negate(tmp), normal, eye_hat);
    /* r + 0,0,1 */
-   emit_op2(p, OPCODE_ADD, tmp, 0, r, swizzle(id,X,Y,W,Z)); 
+   emit_op2(p, OPCODE_ADD, tmp, 0, r, swizzle(id,X,Y,W,Z));
    /* rx^2 + ry^2 + (rz+1)^2 */
-   emit_op2(p, OPCODE_DP3, tmp, 0, tmp, tmp); 
+   emit_op2(p, OPCODE_DP3, tmp, 0, tmp, tmp);
    /* 2/m */
-   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp); 
+   emit_op1(p, OPCODE_RSQ, tmp, 0, tmp);
    /* 1/m */
-   emit_op2(p, OPCODE_MUL, inv_m, 0, tmp, half); 
+   emit_op2(p, OPCODE_MUL, inv_m, 0, tmp, half);
    /* r/m + 1/2 */
-   emit_op3(p, OPCODE_MAD, dest, writemask, r, inv_m, half); 
-	       
+   emit_op3(p, OPCODE_MAD, dest, writemask, r, inv_m, half);
+
    release_temp(p, tmp);
    release_temp(p, r);
    release_temp(p, inv_m);
@@ -1380,10 +1380,10 @@ static void build_texture_transform( struct tnl_program *p )
 
       if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))
 	 continue;
-							     
-      if (p->state->unit[i].texgen_enabled || 
+
+      if (p->state->unit[i].texgen_enabled ||
 	  p->state->unit[i].texmat_enabled) {
-	 
+
 	 GLuint texmat_enabled = p->state->unit[i].texmat_enabled;
 	 struct ureg out = register_output(p, VERT_RESULT_TEX0 + i);
 	 struct ureg out_texgen = undef;
@@ -1394,8 +1394,8 @@ static void build_texture_transform( struct tnl_program *p )
 	    GLuint reflect_mask = 0;
 	    GLuint normal_mask = 0;
 	    GLuint modes[4];
-	 
-	    if (texmat_enabled) 
+
+	    if (texmat_enabled)
 	       out_texgen = get_temp(p);
 	    else
 	       out_texgen = out;
@@ -1409,31 +1409,31 @@ static void build_texture_transform( struct tnl_program *p )
 	       switch (modes[j]) {
 	       case TXG_OBJ_LINEAR: {
 		  struct ureg obj = register_input(p, VERT_ATTRIB_POS);
-		  struct ureg plane = 
+		  struct ureg plane =
 		     register_param3(p, STATE_TEXGEN, i,
 				     STATE_TEXGEN_OBJECT_S + j);
 
-		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
+		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j,
 			   obj, plane );
 		  break;
 	       }
 	       case TXG_EYE_LINEAR: {
 		  struct ureg eye = get_eye_position(p);
-		  struct ureg plane = 
-		     register_param3(p, STATE_TEXGEN, i, 
+		  struct ureg plane =
+		     register_param3(p, STATE_TEXGEN, i,
 				     STATE_TEXGEN_EYE_S + j);
 
-		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j, 
+		  emit_op2(p, OPCODE_DP4, out_texgen, WRITEMASK_X << j,
 			   eye, plane );
 		  break;
 	       }
-	       case TXG_SPHERE_MAP: 
+	       case TXG_SPHERE_MAP:
 		  sphere_mask |= WRITEMASK_X << j;
 		  break;
 	       case TXG_REFLECTION_MAP:
 		  reflect_mask |= WRITEMASK_X << j;
 		  break;
-	       case TXG_NORMAL_MAP: 
+	       case TXG_NORMAL_MAP:
 		  normal_mask |= WRITEMASK_X << j;
 		  break;
 	       case TXG_NONE:
@@ -1462,8 +1462,8 @@ static void build_texture_transform( struct tnl_program *p )
 
 	 if (texmat_enabled) {
 	    struct ureg texmat[4];
-	    struct ureg in = (!is_undef(out_texgen) ? 
-			      out_texgen : 
+	    struct ureg in = (!is_undef(out_texgen) ?
+			      out_texgen :
 			      register_input(p, VERT_ATTRIB_TEX0+i));
 	    if (p->mvp_with_dp4) {
 	       register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
@@ -1596,7 +1596,7 @@ create_new_program( const struct state_key *key,
    p.identity = undef;
    p.temp_in_use = 0;
    p.mvp_with_dp4 = mvp_with_dp4;
-   
+
    if (max_temps >= sizeof(int) * 8)
       p.temp_reserved = 0;
    else
@@ -1639,14 +1639,14 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
     */
    prog = (struct gl_vertex_program *)
       _mesa_search_program_cache(ctx->VertexProgram.Cache, &key, sizeof(key));
-   
+
    if (!prog) {
       /* OK, we'll have to build a new one */
       if (0)
          _mesa_printf("Build new TNL program\n");
-	 
+
       prog = (struct gl_vertex_program *)
-         ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0); 
+         ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0);
       if (!prog)
          return NULL;
 
@@ -1656,7 +1656,7 @@ _mesa_get_fixed_func_vertex_program(GLcontext *ctx)
 
 #if 0
       if (ctx->Driver.ProgramStringNotify)
-         ctx->Driver.ProgramStringNotify( ctx, GL_VERTEX_PROGRAM_ARB, 
+         ctx->Driver.ProgramStringNotify( ctx, GL_VERTEX_PROGRAM_ARB,
                                           &prog->Base );
 #endif
       _mesa_program_cache_insert(ctx, ctx->VertexProgram.Cache,
-- 
cgit v1.2.3


From ee0984e299f7a48fbd3a0172ec9c471e0e8cbb92 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Tue, 18 Aug 2009 17:39:55 -0600
Subject: mesa: when emitting vertex program fog, set yzw=0,0,1

Fixes piglit fp-fog failure with gallium.
---
 src/mesa/main/ffvertex_prog.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 80dde4b5aa..8e21a27f89 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -1306,7 +1306,9 @@ static void build_fog( struct tnl_program *p )
       input = swizzle1(register_input(p, VERT_ATTRIB_FOG), X);
    }
 
+   /* result.fog = {abs(f),0,0,1}; */
    emit_op1(p, OPCODE_ABS, fog, WRITEMASK_X, input);
+   emit_op1(p, OPCODE_MOV, fog, WRITEMASK_YZW, get_identity_param(p));
 }
 
 
-- 
cgit v1.2.3


From a7a3d3c27144595b2134f4271996cba901938d12 Mon Sep 17 00:00:00 2001
From: Brian Paul <brianp@vmware.com>
Date: Wed, 2 Sep 2009 09:43:00 -0600
Subject: mesa: replace 8 with NUM_UNITS

---
 src/mesa/main/ffvertex_prog.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'src/mesa/main/ffvertex_prog.c')

diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index 8e21a27f89..356476e35a 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -46,6 +46,9 @@
 #include "shader/prog_statevars.h"
 
 
+/** Max of number of lights and texture coord units */
+#define NUM_UNITS MAX2(MAX_TEXTURE_COORD_UNITS, MAX_LIGHTS)
+
 struct state_key {
    unsigned light_color_material_mask:12;
    unsigned light_global_enabled:1;
@@ -77,7 +80,7 @@ struct state_key {
       unsigned texgen_mode1:4;
       unsigned texgen_mode2:4;
       unsigned texgen_mode3:4;
-   } unit[8];
+   } unit[NUM_UNITS];
 };
 
 
-- 
cgit v1.2.3