mesa pixman git update 5 aug 2011

9 files changed, 2748 insertions, 2848 deletions

diff --git a/mesalib/src/mesa/main/compiler.h b/mesalib/src/mesa/main/compiler.h
index d736fdfc5..ee7d0b2f8 100644
--- a/mesalib/src/mesa/main/compiler.h
+++ b/mesalib/src/mesa/main/compiler.h
@@ -60,29 +60,7 @@ extern "C" {
 /**
  * Get standard integer types
  */
-#if defined(_MSC_VER)
-   typedef __int8             int8_t;
-   typedef unsigned __int8    uint8_t;
-   typedef __int16            int16_t;
-   typedef unsigned __int16   uint16_t;
-   typedef __int32            int32_t;
-   typedef unsigned __int32   uint32_t;
-   typedef __int64            int64_t;
-   typedef unsigned __int64   uint64_t;
-
-#  if defined(_WIN64)
-     typedef __int64            intptr_t;
-     typedef unsigned __int64   uintptr_t;
-#  else
-     typedef __int32            intptr_t;
-     typedef unsigned __int32   uintptr_t;
-#  endif
-
-#  define INT64_C(__val) __val##i64
-#  define UINT64_C(__val) __val##ui64
-#else
-#  include <stdint.h>
-#endif
+#include <stdint.h>
 
 
 /**
diff --git a/mesalib/src/mesa/main/enable.c b/mesalib/src/mesa/main/enable.c
index aac8b9c5e..3ba4df634 100644
--- a/mesalib/src/mesa/main/enable.c
+++ b/mesalib/src/mesa/main/enable.c
@@ -5,7 +5,6 @@
 
 /*
  * Mesa 3-D graphics library
- * Version:  7.0.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
  *
@@ -560,7 +559,6 @@ _mesa_set_enable(struct gl_context *ctx, GLenum cap, GLboolean state)
          ctx->Polygon.OffsetLine = state;
          break;
       case GL_POLYGON_OFFSET_FILL:
-         /*case GL_POLYGON_OFFSET_EXT:*/
          if (ctx->Polygon.OffsetFill == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_POLYGON);
@@ -643,9 +641,7 @@ _mesa_set_enable(struct gl_context *ctx, GLenum cap, GLboolean state)
          break;
 #endif
 
-      /*
-       * CLIENT STATE!!!
-       */
+      /* client-side state */
       case GL_VERTEX_ARRAY:
       case GL_NORMAL_ARRAY:
       case GL_COLOR_ARRAY:
@@ -1174,7 +1170,6 @@ _mesa_IsEnabled( GLenum cap )
       case GL_POLYGON_OFFSET_LINE:
 	 return ctx->Polygon.OffsetLine;
       case GL_POLYGON_OFFSET_FILL:
-      /*case GL_POLYGON_OFFSET_EXT:*/
 	 return ctx->Polygon.OffsetFill;
       case GL_RESCALE_NORMAL_EXT:
          return ctx->Transform.RescaleNormals;
@@ -1213,9 +1208,7 @@ _mesa_IsEnabled( GLenum cap )
          }
 #endif
 
-      /*
-       * CLIENT STATE!!!
-       */
+      /* client-side state */
       case GL_VERTEX_ARRAY:
          return (ctx->Array.ArrayObj->Vertex.Enabled != 0);
       case GL_NORMAL_ARRAY:
diff --git a/mesalib/src/mesa/main/ff_fragment_shader.cpp b/mesalib/src/mesa/main/ff_fragment_shader.cpp
index dbfa6b57d..7cc172168 100644
--- a/mesalib/src/mesa/main/ff_fragment_shader.cpp
+++ b/mesalib/src/mesa/main/ff_fragment_shader.cpp
@@ -874,7 +874,8 @@ static struct ureg register_const4f( struct texenv_fragment_program *p,
    values[1] = s1;
    values[2] = s2;
    values[3] = s3;
-   idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
+   idx = _mesa_add_unnamed_constant( p->program->Base.Parameters,
+                                     (gl_constant_value *) values, 4,
                                      &swizzle );
    r = make_ureg(PROGRAM_CONSTANT, idx);
    r.swz = swizzle;
diff --git a/mesalib/src/mesa/main/ffvertex_prog.c b/mesalib/src/mesa/main/ffvertex_prog.c
index 281229e34..2d2485c9e 100644
--- a/mesalib/src/mesa/main/ffvertex_prog.c
+++ b/mesalib/src/mesa/main/ffvertex_prog.c
@@ -1,1680 +1,1680 @@
-/**************************************************************************

- *

- * 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

- * 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.c

- *

- * 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/mfeatures.h"

-#include "main/enums.h"

-#include "main/ffvertex_prog.h"

-#include "program/program.h"

-#include "program/prog_cache.h"

-#include "program/prog_instruction.h"

-#include "program/prog_parameter.h"

-#include "program/prog_print.h"

-#include "program/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;

-   unsigned light_local_viewer:1;

-   unsigned light_twoside:1;

-   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 separate_specular:1;

-   unsigned point_attenuated:1;

-   unsigned point_array:1;

-   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;

-      unsigned light_spotcutoff_is_180:1;

-      unsigned light_attenuated:1;

-      unsigned texunit_really_enabled:1;

-      unsigned texmat_enabled:1;

-      unsigned coord_replace:1;

-      unsigned texgen_enabled:4;

-      unsigned texgen_mode0:4;

-      unsigned texgen_mode1:4;

-      unsigned texgen_mode2:4;

-      unsigned texgen_mode3:4;

-   } unit[NUM_UNITS];

-};

-

-

-#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;

-   }

-}

-

-

-

-static GLboolean check_active_shininess( struct gl_context *ctx,

-                                         const struct state_key *key,

-                                         GLuint side )

-{

-   GLuint bit = 1 << (MAT_ATTRIB_FRONT_SHININESS + side);

-

-   if ((key->varying_vp_inputs & VERT_BIT_COLOR0) &&

-       (key->light_color_material_mask & bit))

-      return GL_TRUE;

-

-   if (key->varying_vp_inputs & (bit << 16))

-      return GL_TRUE;

-

-   if (ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS + side][0] != 0.0F)

-      return GL_TRUE;

-

-   return GL_FALSE;

-}

-

-

-static void make_state_key( struct gl_context *ctx, struct state_key *key )

-{

-   const struct gl_fragment_program *fp;

-   GLuint i;

-

-   memset(key, 0, sizeof(struct state_key));

-   fp = ctx->FragmentProgram._Current;

-

-   /* This now relies on texenvprogram.c being active:

-    */

-   assert(fp);

-

-   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 */

-      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_mask = ctx->Light.ColorMaterialBitmask;

-      }

-

-      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 (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)

-      key->normalize = 1;

-

-   if (ctx->Transform.RescaleNormals)

-      key->rescale_normals = 1;

-

-   if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT)

-      key->fog_source_is_depth = 1;

-

-   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 ||

-       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->Point.PointSprite)

-	 if (ctx->Point.CoordReplace[i])

-	    key->unit[i].coord_replace = 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->GenS.Mode );

-	 key->unit[i].texgen_mode1 =

-	    translate_texgen( texUnit->TexGenEnabled & (1<<1),

-			      texUnit->GenT.Mode );

-	 key->unit[i].texgen_mode2 =

-	    translate_texgen( texUnit->TexGenEnabled & (1<<2),

-			      texUnit->GenR.Mode );

-	 key->unit[i].texgen_mode3 =

-	    translate_texgen( texUnit->TexGenEnabled & (1<<3),

-			      texUnit->GenQ.Mode );

-      }

-   }

-}

-

-

-

-/* Very useful debugging tool - produces annotated listing of

- * generated program with line/function references for each

- * instruction back into this file:

- */

-#define DISASSEM 0

-

-

-/* 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:9;      /* relative addressing may be negative */

-                     /* sizeof(idx) should == sizeof(prog_src_reg::Index) */

-   GLuint negate:1;

-   GLuint swz:12;

-   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 */

-   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;

-   struct ureg transformed_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__);

-      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_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 )

-{

-   assert(input < 32);

-

-   if (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 );

-   }

-}

-

-

-/**

- * \param input  one of VERT_RESULT_x tokens.

- */

-static struct ureg register_output( struct tnl_program *p, GLuint output )

-{

-   p->program->Base.OutputsWritten |= BITFIELD64_BIT(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_CONSTANT, 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 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->Negate = reg.negate ? NEGATE_XYZW : NEGATE_NONE;

-   src->Abs = 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,

-		      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;

-   /* Check that bitfield sizes aren't exceeded */

-   ASSERT(dst->Index == reg.idx);

-}

-

-

-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;

-	 printf("%s:\n", fn);

-      }

-

-      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;

-   struct prog_instruction *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 */

-      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->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 )

-{

-#if 0

-   /* XXX use this when drivers are ready for NRM3 */

-   emit_op1(p, OPCODE_NRM3, dest, WRITEMASK_XYZ, src);

-#else

-   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);

-#endif

-}

-

-

-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 (p->mvp_with_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_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)) {

-      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_transformed_normal( struct tnl_program *p )

-{

-   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);

-

-      if (p->state->need_eye_coords) {

-         register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,

-                                 STATE_MATRIX_INVTRANS, mvinv );

-

-         /* 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, transformed_normal, normal );

-         normal = transformed_normal;

-      }

-      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);

-

-	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal, rescale );

-         normal = transformed_normal;

-      }

-

-      assert(normal.file == PROGRAM_TEMPORARY);

-      p->transformed_normal = normal;

-   }

-

-   return p->transformed_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 (p->mvp_with_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->varying_vp_inputs & VERT_BIT_COLOR0) {

-      p->materials =

-	 p->color_materials = p->state->light_color_material_mask;

-   }

-

-   p->materials |= (p->state->varying_vp_inputs >> 16);

-}

-

-

-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)) {

-      /* 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 );

-}

-

-#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_LIGHT_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;

-}

-

-

-/**

- * 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);  /* id = {0,0,0,1} */

-

-   /* Note that lit.x & lit.w will not be examined.  Note also that

-    * dots.xyzw == dots.xxxx.

-    */

-

-   /* MAX lit, 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

-    */

-   emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, swizzle1(id,Z), dots);

-}

-

-

-/* 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_transformed_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;

-

-   /*

-    * NOTE:

-    * 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++)

-      if (p->state->unit[i].light_enabled)

-	 nr_lights++;

-

-   set_material_flags(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));

-         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) {

-      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)));

-         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_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);

-                  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 {

-	    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);

-

-	    /* Calculate 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->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);

-	 }

-

-	 /* 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_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;

-

-	    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;

-	    }

-

-	    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);

-            }

-

-	    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;

-

-	    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;

-	    }

-

-            /* 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)) {

-               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, _bfc0, 0, ambient, _bfc0); /**/

-            }

-            else {

-               emit_degenerate_lit(p, lit, dots);

-               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);

-            /* 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);

-	    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 = get_eye_position_z(p);

-   }

-   else {

-      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));

-}

-

-

-static void build_reflect_texgen( struct tnl_program *p,

-				  struct ureg dest,

-				  GLuint writemask )

-{

-   struct ureg normal = get_transformed_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_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);

-   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_COORD_UNITS; i++) {

-

-      if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))

-	 continue;

-

-      if (p->state->unit[i].coord_replace)

-  	 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_transformed_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 (p->mvp_with_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);

-      }

-   }

-}

-

-

-/**

- * 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_param2(p, STATE_INTERNAL, STATE_POINT_SIZE_CLAMPED);

-   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 0

-   /* out = pointSize / sqrt(factor) */

-   emit_op2(p, OPCODE_MUL, out, WRITEMASK_X, ut, state_size);

-#else

-   /* 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));

-#endif

-

-   release_temp(p, ut);

-}

-

-

-/**

- * 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:

-    */

-   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)

-      build_fog(p);

-

-   if (p->state->fragprog_inputs_read & FRAG_BITS_TEX_ANY)

-      build_texture_transform(p);

-

-   if (p->state->point_attenuated)

-      build_atten_pointsize(p);

-   else if (p->state->point_array)

-      build_array_pointsize(p);

-

-   /* Finish up:

-    */

-   emit_op1(p, OPCODE_END, undef, 0, undef);

-

-   /* Disassemble:

-    */

-   if (DISASSEM) {

-      printf ("\n");

-   }

-}

-

-

-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;

-

-   memset(&p, 0, sizeof(p));

-   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;

-   p.temp_in_use = 0;

-   p.mvp_with_dp4 = mvp_with_dp4;

-

-   if (max_temps >= sizeof(int) * 8)

-      p.temp_reserved = 0;

-   else

-      p.temp_reserved = ~((1<<max_temps)-1);

-

-   /* 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 =

-   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(struct gl_context *ctx)

-{

-   struct gl_vertex_program *prog;

-   struct state_key key;

-

-   /* Grab all the relevent state and put it in a single structure:

-    */

-   make_state_key(ctx, &key);

-

-   /* 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)

-         printf("Build new TNL program\n");

-

-      prog = (struct gl_vertex_program *)

-         ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0);

-      if (!prog)

-         return NULL;

-

-      create_new_program( &key, prog,

-                          ctx->mvp_with_dp4,

-                          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);

-   }

-

-   return prog;

-}

+/**************************************************************************
+ *
+ * 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
+ * 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.c
+ *
+ * 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/mfeatures.h"
+#include "main/enums.h"
+#include "main/ffvertex_prog.h"
+#include "program/program.h"
+#include "program/prog_cache.h"
+#include "program/prog_instruction.h"
+#include "program/prog_parameter.h"
+#include "program/prog_print.h"
+#include "program/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;
+   unsigned light_local_viewer:1;
+   unsigned light_twoside:1;
+   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 separate_specular:1;
+   unsigned point_attenuated:1;
+   unsigned point_array:1;
+   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;
+      unsigned light_spotcutoff_is_180:1;
+      unsigned light_attenuated:1;
+      unsigned texunit_really_enabled:1;
+      unsigned texmat_enabled:1;
+      unsigned coord_replace:1;
+      unsigned texgen_enabled:4;
+      unsigned texgen_mode0:4;
+      unsigned texgen_mode1:4;
+      unsigned texgen_mode2:4;
+      unsigned texgen_mode3:4;
+   } unit[NUM_UNITS];
+};
+
+
+#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;
+   }
+}
+
+
+
+static GLboolean check_active_shininess( struct gl_context *ctx,
+                                         const struct state_key *key,
+                                         GLuint side )
+{
+   GLuint bit = 1 << (MAT_ATTRIB_FRONT_SHININESS + side);
+
+   if ((key->varying_vp_inputs & VERT_BIT_COLOR0) &&
+       (key->light_color_material_mask & bit))
+      return GL_TRUE;
+
+   if (key->varying_vp_inputs & (bit << 16))
+      return GL_TRUE;
+
+   if (ctx->Light.Material.Attrib[MAT_ATTRIB_FRONT_SHININESS + side][0] != 0.0F)
+      return GL_TRUE;
+
+   return GL_FALSE;
+}
+
+
+static void make_state_key( struct gl_context *ctx, struct state_key *key )
+{
+   const struct gl_fragment_program *fp;
+   GLuint i;
+
+   memset(key, 0, sizeof(struct state_key));
+   fp = ctx->FragmentProgram._Current;
+
+   /* This now relies on texenvprogram.c being active:
+    */
+   assert(fp);
+
+   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 */
+      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_mask = ctx->Light.ColorMaterialBitmask;
+      }
+
+      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 (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)
+      key->normalize = 1;
+
+   if (ctx->Transform.RescaleNormals)
+      key->rescale_normals = 1;
+
+   if (ctx->Fog.FogCoordinateSource == GL_FRAGMENT_DEPTH_EXT)
+      key->fog_source_is_depth = 1;
+
+   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 ||
+       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->Point.PointSprite)
+	 if (ctx->Point.CoordReplace[i])
+	    key->unit[i].coord_replace = 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->GenS.Mode );
+	 key->unit[i].texgen_mode1 =
+	    translate_texgen( texUnit->TexGenEnabled & (1<<1),
+			      texUnit->GenT.Mode );
+	 key->unit[i].texgen_mode2 =
+	    translate_texgen( texUnit->TexGenEnabled & (1<<2),
+			      texUnit->GenR.Mode );
+	 key->unit[i].texgen_mode3 =
+	    translate_texgen( texUnit->TexGenEnabled & (1<<3),
+			      texUnit->GenQ.Mode );
+      }
+   }
+}
+
+
+
+/* Very useful debugging tool - produces annotated listing of
+ * generated program with line/function references for each
+ * instruction back into this file:
+ */
+#define DISASSEM 0
+
+
+/* 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:9;      /* relative addressing may be negative */
+                     /* sizeof(idx) should == sizeof(prog_src_reg::Index) */
+   GLuint negate:1;
+   GLuint swz:12;
+   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 */
+   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;
+   struct ureg transformed_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__);
+      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_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 )
+{
+   assert(input < 32);
+
+   if (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 );
+   }
+}
+
+
+/**
+ * \param input  one of VERT_RESULT_x tokens.
+ */
+static struct ureg register_output( struct tnl_program *p, GLuint output )
+{
+   p->program->Base.OutputsWritten |= BITFIELD64_BIT(output);
+   return make_ureg(PROGRAM_OUTPUT, output);
+}
+
+
+static struct ureg register_const4f( struct tnl_program *p,
+			      GLfloat s0,
+			      GLfloat s1,
+			      GLfloat s2,
+			      GLfloat s3)
+{
+   gl_constant_value values[4];
+   GLint idx;
+   GLuint swizzle;
+   values[0].f = s0;
+   values[1].f = s1;
+   values[2].f = s2;
+   values[3].f = s3;
+   idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
+                                     &swizzle );
+   ASSERT(swizzle == SWIZZLE_NOOP);
+   return make_ureg(PROGRAM_CONSTANT, 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 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->Negate = reg.negate ? NEGATE_XYZW : NEGATE_NONE;
+   src->Abs = 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,
+		      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;
+   /* Check that bitfield sizes aren't exceeded */
+   ASSERT(dst->Index == reg.idx);
+}
+
+
+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;
+	 printf("%s:\n", fn);
+      }
+
+      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;
+   struct prog_instruction *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 */
+      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->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 )
+{
+#if 0
+   /* XXX use this when drivers are ready for NRM3 */
+   emit_op1(p, OPCODE_NRM3, dest, WRITEMASK_XYZ, src);
+#else
+   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);
+#endif
+}
+
+
+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 (p->mvp_with_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_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)) {
+      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_transformed_normal( struct tnl_program *p )
+{
+   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);
+
+      if (p->state->need_eye_coords) {
+         register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
+                                 STATE_MATRIX_INVTRANS, mvinv );
+
+         /* 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, transformed_normal, normal );
+         normal = transformed_normal;
+      }
+      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);
+
+	 emit_op2( p, OPCODE_MUL, transformed_normal, 0, normal, rescale );
+         normal = transformed_normal;
+      }
+
+      assert(normal.file == PROGRAM_TEMPORARY);
+      p->transformed_normal = normal;
+   }
+
+   return p->transformed_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 (p->mvp_with_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->varying_vp_inputs & VERT_BIT_COLOR0) {
+      p->materials =
+	 p->color_materials = p->state->light_color_material_mask;
+   }
+
+   p->materials |= (p->state->varying_vp_inputs >> 16);
+}
+
+
+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)) {
+      /* 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 );
+}
+
+#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_LIGHT_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;
+}
+
+
+/**
+ * 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);  /* id = {0,0,0,1} */
+
+   /* Note that lit.x & lit.w will not be examined.  Note also that
+    * dots.xyzw == dots.xxxx.
+    */
+
+   /* MAX lit, 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
+    */
+   emit_op2(p, OPCODE_SLT, lit, WRITEMASK_Z, swizzle1(id,Z), dots);
+}
+
+
+/* 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_transformed_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;
+
+   /*
+    * NOTE:
+    * 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++)
+      if (p->state->unit[i].light_enabled)
+	 nr_lights++;
+
+   set_material_flags(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));
+         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) {
+      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)));
+         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_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);
+                  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 {
+	    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);
+
+	    /* Calculate 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->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);
+	 }
+
+	 /* 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_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;
+
+	    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;
+	    }
+
+	    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);
+            }
+
+	    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;
+
+	    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;
+	    }
+
+            /* 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)) {
+               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, _bfc0, 0, ambient, _bfc0); /**/
+            }
+            else {
+               emit_degenerate_lit(p, lit, dots);
+               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);
+            /* 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);
+	    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 = get_eye_position_z(p);
+   }
+   else {
+      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));
+}
+
+
+static void build_reflect_texgen( struct tnl_program *p,
+				  struct ureg dest,
+				  GLuint writemask )
+{
+   struct ureg normal = get_transformed_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_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);
+   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_COORD_UNITS; i++) {
+
+      if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))
+	 continue;
+
+      if (p->state->unit[i].coord_replace)
+  	 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_transformed_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 (p->mvp_with_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);
+      }
+   }
+}
+
+
+/**
+ * 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_param2(p, STATE_INTERNAL, STATE_POINT_SIZE_CLAMPED);
+   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 0
+   /* out = pointSize / sqrt(factor) */
+   emit_op2(p, OPCODE_MUL, out, WRITEMASK_X, ut, state_size);
+#else
+   /* 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));
+#endif
+
+   release_temp(p, ut);
+}
+
+
+/**
+ * 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:
+    */
+   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)
+      build_fog(p);
+
+   if (p->state->fragprog_inputs_read & FRAG_BITS_TEX_ANY)
+      build_texture_transform(p);
+
+   if (p->state->point_attenuated)
+      build_atten_pointsize(p);
+   else if (p->state->point_array)
+      build_array_pointsize(p);
+
+   /* Finish up:
+    */
+   emit_op1(p, OPCODE_END, undef, 0, undef);
+
+   /* Disassemble:
+    */
+   if (DISASSEM) {
+      printf ("\n");
+   }
+}
+
+
+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;
+
+   memset(&p, 0, sizeof(p));
+   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;
+   p.temp_in_use = 0;
+   p.mvp_with_dp4 = mvp_with_dp4;
+
+   if (max_temps >= sizeof(int) * 8)
+      p.temp_reserved = 0;
+   else
+      p.temp_reserved = ~((1<<max_temps)-1);
+
+   /* 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 =
+   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(struct gl_context *ctx)
+{
+   struct gl_vertex_program *prog;
+   struct state_key key;
+
+   /* Grab all the relevent state and put it in a single structure:
+    */
+   make_state_key(ctx, &key);
+
+   /* 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)
+         printf("Build new TNL program\n");
+
+      prog = (struct gl_vertex_program *)
+         ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0);
+      if (!prog)
+         return NULL;
+
+      create_new_program( &key, prog,
+                          ctx->mvp_with_dp4,
+                          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);
+   }
+
+   return prog;
+}
diff --git a/mesalib/src/mesa/main/nvprogram.c b/mesalib/src/mesa/main/nvprogram.c
index 6b20fdae6..7ff7645b7 100644
--- a/mesalib/src/mesa/main/nvprogram.c
+++ b/mesalib/src/mesa/main/nvprogram.c
@@ -1,920 +1,920 @@
-/*

- * Mesa 3-D graphics library

- * Version:  6.5.2

- *

- * Copyright (C) 1999-2006  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.

- */

-

-/**

- * \file nvprogram.c

- * NVIDIA vertex/fragment program state management functions.

- * \author Brian Paul

- */

-

-/*

- * Regarding GL_NV_fragment/vertex_program, GL_NV_vertex_program1_1, etc:

- *

- * Portions of this software may use or implement intellectual

- * property owned and licensed by NVIDIA Corporation. NVIDIA disclaims

- * any and all warranties with respect to such intellectual property,

- * including any use thereof or modifications thereto.

- */

-

-#include "main/glheader.h"

-#include "main/context.h"

-#include "main/hash.h"

-#include "main/imports.h"

-#include "main/macros.h"

-#include "main/mtypes.h"

-#include "main/nvprogram.h"

-#include "program/arbprogparse.h"

-#include "program/nvfragparse.h"

-#include "program/nvvertparse.h"

-#include "program/program.h"

-#include "program/prog_instruction.h"

-#include "program/prog_parameter.h"

-

-

-

-/**

- * Execute a vertex state program.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params)

-{

-   struct gl_vertex_program *vprog;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target != GL_VERTEX_STATE_PROGRAM_NV) {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glExecuteProgramNV");

-      return;

-   }

-

-   FLUSH_VERTICES(ctx, _NEW_PROGRAM);

-

-   vprog = (struct gl_vertex_program *) _mesa_lookup_program(ctx, id);

-

-   if (!vprog || vprog->Base.Target != GL_VERTEX_STATE_PROGRAM_NV) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glExecuteProgramNV");

-      return;

-   }

-   

-   _mesa_problem(ctx, "glExecuteProgramNV() not supported");

-}

-

-

-/**

- * Determine if a set of programs is resident in hardware.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-GLboolean GLAPIENTRY

-_mesa_AreProgramsResidentNV(GLsizei n, const GLuint *ids,

-                            GLboolean *residences)

-{

-   GLint i, j;

-   GLboolean allResident = GL_TRUE;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);

-

-   if (n < 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV(n)");

-      return GL_FALSE;

-   }

-

-   for (i = 0; i < n; i++) {

-      const struct gl_program *prog;

-      if (ids[i] == 0) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV");

-         return GL_FALSE;

-      }

-      prog = _mesa_lookup_program(ctx, ids[i]);

-      if (!prog) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV");

-         return GL_FALSE;

-      }

-      if (prog->Resident) {

-	 if (!allResident)

-	    residences[i] = GL_TRUE;

-      }

-      else {

-         if (allResident) {

-	    allResident = GL_FALSE;

-	    for (j = 0; j < i; j++)

-	       residences[j] = GL_TRUE;

-	 }

-	 residences[i] = GL_FALSE;

-      }

-   }

-

-   return allResident;

-}

-

-

-/**

- * Request that a set of programs be resident in hardware.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_RequestResidentProgramsNV(GLsizei n, const GLuint *ids)

-{

-   GLint i;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (n < 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(n)");

-      return;

-   }

-

-   /* just error checking for now */

-   for (i = 0; i < n; i++) {

-      struct gl_program *prog;

-

-      if (ids[i] == 0) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(id)");

-         return;

-      }

-

-      prog = _mesa_lookup_program(ctx, ids[i]);

-      if (!prog) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(id)");

-         return;

-      }

-

-      /* XXX this is really a hardware thing we should hook out */

-      prog->Resident = GL_TRUE;

-   }

-}

-

-

-/**

- * Get a program parameter register.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetProgramParameterfvNV(GLenum target, GLuint index,

-                              GLenum pname, GLfloat *params)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target == GL_VERTEX_PROGRAM_NV) {

-      if (pname == GL_PROGRAM_PARAMETER_NV) {

-         if (index < MAX_NV_VERTEX_PROGRAM_PARAMS) {

-            COPY_4V(params, ctx->VertexProgram.Parameters[index]);

-         }

-         else {

-            _mesa_error(ctx, GL_INVALID_VALUE,

-                        "glGetProgramParameterfvNV(index)");

-            return;

-         }

-      }

-      else {

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterfvNV(pname)");

-         return;

-      }

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterfvNV(target)");

-      return;

-   }

-}

-

-

-/**

- * Get a program parameter register.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetProgramParameterdvNV(GLenum target, GLuint index,

-                              GLenum pname, GLdouble *params)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target == GL_VERTEX_PROGRAM_NV) {

-      if (pname == GL_PROGRAM_PARAMETER_NV) {

-         if (index < MAX_NV_VERTEX_PROGRAM_PARAMS) {

-            COPY_4V(params, ctx->VertexProgram.Parameters[index]);

-         }

-         else {

-            _mesa_error(ctx, GL_INVALID_VALUE,

-                        "glGetProgramParameterdvNV(index)");

-            return;

-         }

-      }

-      else {

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterdvNV(pname)");

-         return;

-      }

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterdvNV(target)");

-      return;

-   }

-}

-

-

-/**

- * Get a program attribute.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetProgramivNV(GLuint id, GLenum pname, GLint *params)

-{

-   struct gl_program *prog;

-   GET_CURRENT_CONTEXT(ctx);

-

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   prog = _mesa_lookup_program(ctx, id);

-   if (!prog) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramivNV");

-      return;

-   }

-

-   switch (pname) {

-      case GL_PROGRAM_TARGET_NV:

-         *params = prog->Target;

-         return;

-      case GL_PROGRAM_LENGTH_NV:

-         *params = prog->String ?(GLint) strlen((char *) prog->String) : 0;

-         return;

-      case GL_PROGRAM_RESIDENT_NV:

-         *params = prog->Resident;

-         return;

-      default:

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramivNV(pname)");

-         return;

-   }

-}

-

-

-/**

- * Get the program source code.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetProgramStringNV(GLuint id, GLenum pname, GLubyte *program)

-{

-   struct gl_program *prog;

-   GET_CURRENT_CONTEXT(ctx);

-

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (pname != GL_PROGRAM_STRING_NV) {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramStringNV(pname)");

-      return;

-   }

-

-   prog = _mesa_lookup_program(ctx, id);

-   if (!prog) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramStringNV");

-      return;

-   }

-

-   if (prog->String) {

-      memcpy(program, prog->String, strlen((char *) prog->String));

-   }

-   else {

-      program[0] = 0;

-   }

-}

-

-

-/**

- * Get matrix tracking information.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetTrackMatrixivNV(GLenum target, GLuint address,

-                         GLenum pname, GLint *params)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target == GL_VERTEX_PROGRAM_NV

-       && ctx->Extensions.NV_vertex_program) {

-      GLuint i;

-

-      if ((address & 0x3) || address >= MAX_NV_VERTEX_PROGRAM_PARAMS) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glGetTrackMatrixivNV(address)");

-         return;

-      }

-

-      i = address / 4;

-

-      switch (pname) {

-         case GL_TRACK_MATRIX_NV:

-            params[0] = (GLint) ctx->VertexProgram.TrackMatrix[i];

-            return;

-         case GL_TRACK_MATRIX_TRANSFORM_NV:

-            params[0] = (GLint) ctx->VertexProgram.TrackMatrixTransform[i];

-            return;

-         default:

-            _mesa_error(ctx, GL_INVALID_ENUM, "glGetTrackMatrixivNV");

-            return;

-      }

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTrackMatrixivNV");

-      return;

-   }

-}

-

-

-/**

- * Get a vertex (or vertex array) attribute.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetVertexAttribdvNV(GLuint index, GLenum pname, GLdouble *params)

-{

-   const struct gl_client_array *array;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");

-      return;

-   }

-

-   array = &ctx->Array.ArrayObj->VertexAttrib[index];

-

-   switch (pname) {

-      case GL_ATTRIB_ARRAY_SIZE_NV:

-         params[0] = array->Size;

-         break;

-      case GL_ATTRIB_ARRAY_STRIDE_NV:

-         params[0] = array->Stride;

-         break;

-      case GL_ATTRIB_ARRAY_TYPE_NV:

-         params[0] = array->Type;

-         break;

-      case GL_CURRENT_ATTRIB_NV:

-         if (index == 0) {

-            _mesa_error(ctx, GL_INVALID_OPERATION,

-                        "glGetVertexAttribdvNV(index == 0)");

-            return;

-         }

-	 FLUSH_CURRENT(ctx, 0);

-         COPY_4V(params, ctx->Current.Attrib[index]);

-         break;

-      default:

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");

-         return;

-   }

-}

-

-/**

- * Get a vertex (or vertex array) attribute.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetVertexAttribfvNV(GLuint index, GLenum pname, GLfloat *params)

-{

-   const struct gl_client_array *array;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");

-      return;

-   }

-

-   array = &ctx->Array.ArrayObj->VertexAttrib[index];

-

-   switch (pname) {

-      case GL_ATTRIB_ARRAY_SIZE_NV:

-         params[0] = (GLfloat) array->Size;

-         break;

-      case GL_ATTRIB_ARRAY_STRIDE_NV:

-         params[0] = (GLfloat) array->Stride;

-         break;

-      case GL_ATTRIB_ARRAY_TYPE_NV:

-         params[0] = (GLfloat) array->Type;

-         break;

-      case GL_CURRENT_ATTRIB_NV:

-         if (index == 0) {

-            _mesa_error(ctx, GL_INVALID_OPERATION,

-                        "glGetVertexAttribfvNV(index == 0)");

-            return;

-         }

-	 FLUSH_CURRENT(ctx, 0);

-         COPY_4V(params, ctx->Current.Attrib[index]);

-         break;

-      default:

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");

-         return;

-   }

-}

-

-/**

- * Get a vertex (or vertex array) attribute.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetVertexAttribivNV(GLuint index, GLenum pname, GLint *params)

-{

-   const struct gl_client_array *array;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");

-      return;

-   }

-

-   array = &ctx->Array.ArrayObj->VertexAttrib[index];

-

-   switch (pname) {

-      case GL_ATTRIB_ARRAY_SIZE_NV:

-         params[0] = array->Size;

-         break;

-      case GL_ATTRIB_ARRAY_STRIDE_NV:

-         params[0] = array->Stride;

-         break;

-      case GL_ATTRIB_ARRAY_TYPE_NV:

-         params[0] = array->Type;

-         break;

-      case GL_CURRENT_ATTRIB_NV:

-         if (index == 0) {

-            _mesa_error(ctx, GL_INVALID_OPERATION,

-                        "glGetVertexAttribivNV(index == 0)");

-            return;

-         }

-	 FLUSH_CURRENT(ctx, 0);

-         params[0] = (GLint) ctx->Current.Attrib[index][0];

-         params[1] = (GLint) ctx->Current.Attrib[index][1];

-         params[2] = (GLint) ctx->Current.Attrib[index][2];

-         params[3] = (GLint) ctx->Current.Attrib[index][3];

-         break;

-      case GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB:

-         params[0] = array->BufferObj->Name;

-         break;

-      default:

-         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");

-         return;

-   }

-}

-

-

-/**

- * Get a vertex array attribute pointer.

- * \note Not compiled into display lists.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_GetVertexAttribPointervNV(GLuint index, GLenum pname, GLvoid **pointer)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribPointerNV(index)");

-      return;

-   }

-

-   if (pname != GL_ATTRIB_ARRAY_POINTER_NV) {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribPointerNV(pname)");

-      return;

-   }

-

-   *pointer = (GLvoid *) ctx->Array.ArrayObj->VertexAttrib[index].Ptr;

-}

-

-void

-_mesa_emit_nv_temp_initialization(struct gl_context *ctx,

-				  struct gl_program *program)

-{

-   struct prog_instruction *inst;

-   GLuint i;

-   struct gl_shader_compiler_options* options =

-         &ctx->ShaderCompilerOptions[_mesa_program_target_to_index(program->Target)];

-

-   if (!options->EmitNVTempInitialization)

-      return;

-

-   /* We'll swizzle up a zero temporary so we can use it for the

-    * ARL.

-    */

-   if (program->NumTemporaries == 0)

-      program->NumTemporaries = 1;

-

-   _mesa_insert_instructions(program, 0, program->NumTemporaries + 1);

-

-   for (i = 0; i < program->NumTemporaries; i++) {

-      struct prog_instruction *inst = &program->Instructions[i];

-

-      inst->Opcode = OPCODE_SWZ;

-      inst->DstReg.File = PROGRAM_TEMPORARY;

-      inst->DstReg.Index = i;

-      inst->DstReg.WriteMask = WRITEMASK_XYZW;

-      inst->SrcReg[0].File = PROGRAM_TEMPORARY;

-      inst->SrcReg[0].Index = 0;

-      inst->SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ZERO,

-					      SWIZZLE_ZERO,

-					      SWIZZLE_ZERO,

-					      SWIZZLE_ZERO);

-   }

-

-   inst = &program->Instructions[i];

-   inst->Opcode = OPCODE_ARL;

-   inst->DstReg.File = PROGRAM_ADDRESS;

-   inst->DstReg.Index = 0;

-   inst->DstReg.WriteMask = WRITEMASK_XYZW;

-   inst->SrcReg[0].File = PROGRAM_TEMPORARY;

-   inst->SrcReg[0].Index = 0;

-   inst->SrcReg[0].Swizzle = SWIZZLE_XXXX;

-

-   if (program->NumAddressRegs == 0)

-      program->NumAddressRegs = 1;

-}

-

-void

-_mesa_setup_nv_temporary_count(struct gl_context *ctx, struct gl_program *program)

-{

-   GLuint i;

-

-   program->NumTemporaries = 0;

-   for (i = 0; i < program->NumInstructions; i++) {

-      struct prog_instruction *inst = &program->Instructions[i];

-

-      if (inst->DstReg.File == PROGRAM_TEMPORARY) {

-	 program->NumTemporaries = MAX2(program->NumTemporaries,

-					inst->DstReg.Index + 1);

-      }

-      if (inst->SrcReg[0].File == PROGRAM_TEMPORARY) {

-	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,

-					inst->SrcReg[0].Index + 1);

-      }

-      if (inst->SrcReg[1].File == PROGRAM_TEMPORARY) {

-	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,

-					inst->SrcReg[1].Index + 1);

-      }

-      if (inst->SrcReg[2].File == PROGRAM_TEMPORARY) {

-	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,

-					inst->SrcReg[2].Index + 1);

-      }

-   }

-}

-

-/**

- * Load/parse/compile a program.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_LoadProgramNV(GLenum target, GLuint id, GLsizei len,

-                    const GLubyte *program)

-{

-   struct gl_program *prog;

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (!ctx->Extensions.NV_vertex_program

-       && !ctx->Extensions.NV_fragment_program) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV()");

-      return;

-   }

-

-   if (id == 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glLoadProgramNV(id)");

-      return;

-   }

-

-   if (len < 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glLoadProgramNV(len)");

-      return;

-   }

-

-   FLUSH_VERTICES(ctx, _NEW_PROGRAM);

-

-   prog = _mesa_lookup_program(ctx, id);

-

-   if (prog && prog->Target != 0 && prog->Target != target) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV(target)");

-      return;

-   }

-

-   if ((target == GL_VERTEX_PROGRAM_NV ||

-        target == GL_VERTEX_STATE_PROGRAM_NV)

-       && ctx->Extensions.NV_vertex_program) {

-      struct gl_vertex_program *vprog = (struct gl_vertex_program *) prog;

-      if (!vprog || prog == &_mesa_DummyProgram) {

-         vprog = (struct gl_vertex_program *)

-            ctx->Driver.NewProgram(ctx, target, id);

-         if (!vprog) {

-            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");

-            return;

-         }

-         _mesa_HashInsert(ctx->Shared->Programs, id, vprog);

-      }

-

-      if (ctx->Extensions.ARB_vertex_program

-	  && (strncmp((char *) program, "!!ARB", 5) == 0)) {

-	 _mesa_parse_arb_vertex_program(ctx, target, program, len, vprog);

-      } else {

-	 _mesa_parse_nv_vertex_program(ctx, target, program, len, vprog);

-      }

-   }

-   else if (target == GL_FRAGMENT_PROGRAM_NV

-            && ctx->Extensions.NV_fragment_program) {

-      struct gl_fragment_program *fprog = (struct gl_fragment_program *) prog;

-      if (!fprog || prog == &_mesa_DummyProgram) {

-         fprog = (struct gl_fragment_program *)

-            ctx->Driver.NewProgram(ctx, target, id);

-         if (!fprog) {

-            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");

-            return;

-         }

-         _mesa_HashInsert(ctx->Shared->Programs, id, fprog);

-      }

-      _mesa_parse_nv_fragment_program(ctx, target, program, len, fprog);

-   }

-   else if (target == GL_FRAGMENT_PROGRAM_ARB

-            && ctx->Extensions.ARB_fragment_program) {

-      struct gl_fragment_program *fprog = (struct gl_fragment_program *) prog;

-      if (!fprog || prog == &_mesa_DummyProgram) {

-         fprog = (struct gl_fragment_program *)

-            ctx->Driver.NewProgram(ctx, target, id);

-         if (!fprog) {

-            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");

-            return;

-         }

-         _mesa_HashInsert(ctx->Shared->Programs, id, fprog);

-      }

-      _mesa_parse_arb_fragment_program(ctx, target, program, len, fprog);

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glLoadProgramNV(target)");

-   }

-}

-

-

-

-/**

- * Set a sequence of program parameter registers.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_ProgramParameters4dvNV(GLenum target, GLuint index,

-                             GLsizei num, const GLdouble *params)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {

-      GLint i;

-      if (index + num > MAX_NV_VERTEX_PROGRAM_PARAMS) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glProgramParameters4dvNV");

-         return;

-      }

-      for (i = 0; i < num; i++) {

-         ctx->VertexProgram.Parameters[index + i][0] = (GLfloat) params[0];

-         ctx->VertexProgram.Parameters[index + i][1] = (GLfloat) params[1];

-         ctx->VertexProgram.Parameters[index + i][2] = (GLfloat) params[2];

-         ctx->VertexProgram.Parameters[index + i][3] = (GLfloat) params[3];

-         params += 4;

-      };

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glProgramParameters4dvNV");

-      return;

-   }

-}

-

-

-/**

- * Set a sequence of program parameter registers.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_ProgramParameters4fvNV(GLenum target, GLuint index,

-                             GLsizei num, const GLfloat *params)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {

-      GLint i;

-      if (index + num > MAX_NV_VERTEX_PROGRAM_PARAMS) {

-         _mesa_error(ctx, GL_INVALID_VALUE, "glProgramParameters4fvNV");

-         return;

-      }

-      for (i = 0; i < num; i++) {

-         COPY_4V(ctx->VertexProgram.Parameters[index + i], params);

-         params += 4;

-      }

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glProgramParameters4fvNV");

-      return;

-   }

-}

-

-

-

-/**

- * Setup tracking of matrices into program parameter registers.

- * \note Called from the GL API dispatcher.

- */

-void GLAPIENTRY

-_mesa_TrackMatrixNV(GLenum target, GLuint address,

-                    GLenum matrix, GLenum transform)

-{

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   FLUSH_VERTICES(ctx, _NEW_PROGRAM);

-

-   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {

-      if (address & 0x3) {

-         /* addr must be multiple of four */

-         _mesa_error(ctx, GL_INVALID_VALUE, "glTrackMatrixNV(address)");

-         return;

-      }

-

-      switch (matrix) {

-         case GL_NONE:

-         case GL_MODELVIEW:

-         case GL_PROJECTION:

-         case GL_TEXTURE:

-         case GL_COLOR:

-         case GL_MODELVIEW_PROJECTION_NV:

-         case GL_MATRIX0_NV:

-         case GL_MATRIX1_NV:

-         case GL_MATRIX2_NV:

-         case GL_MATRIX3_NV:

-         case GL_MATRIX4_NV:

-         case GL_MATRIX5_NV:

-         case GL_MATRIX6_NV:

-         case GL_MATRIX7_NV:

-            /* OK, fallthrough */

-            break;

-         default:

-            _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(matrix)");

-            return;

-      }

-

-      switch (transform) {

-         case GL_IDENTITY_NV:

-         case GL_INVERSE_NV:

-         case GL_TRANSPOSE_NV:

-         case GL_INVERSE_TRANSPOSE_NV:

-            /* OK, fallthrough */

-            break;

-         default:

-            _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(transform)");

-            return;

-      }

-

-      ctx->VertexProgram.TrackMatrix[address / 4] = matrix;

-      ctx->VertexProgram.TrackMatrixTransform[address / 4] = transform;

-   }

-   else {

-      _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(target)");

-      return;

-   }

-}

-

-

-void GLAPIENTRY

-_mesa_ProgramNamedParameter4fNV(GLuint id, GLsizei len, const GLubyte *name,

-                                GLfloat x, GLfloat y, GLfloat z, GLfloat w)

-{

-   struct gl_program *prog;

-   struct gl_fragment_program *fragProg;

-   GLfloat *v;

-

-   GET_CURRENT_CONTEXT(ctx);

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   FLUSH_VERTICES(ctx, _NEW_PROGRAM_CONSTANTS);

-

-   prog = _mesa_lookup_program(ctx, id);

-   if (!prog || prog->Target != GL_FRAGMENT_PROGRAM_NV) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glProgramNamedParameterNV");

-      return;

-   }

-

-   if (len <= 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glProgramNamedParameterNV(len)");

-      return;

-   }

-

-   fragProg = (struct gl_fragment_program *) prog;

-   v = _mesa_lookup_parameter_value(fragProg->Base.Parameters, len,

-                                    (char *) name);

-   if (v) {

-      v[0] = x;

-      v[1] = y;

-      v[2] = z;

-      v[3] = w;

-      return;

-   }

-

-   _mesa_error(ctx, GL_INVALID_VALUE, "glProgramNamedParameterNV(name)");

-}

-

-

-void GLAPIENTRY

-_mesa_ProgramNamedParameter4fvNV(GLuint id, GLsizei len, const GLubyte *name,

-                                 const float v[])

-{

-   _mesa_ProgramNamedParameter4fNV(id, len, name, v[0], v[1], v[2], v[3]);

-}

-

-

-void GLAPIENTRY

-_mesa_ProgramNamedParameter4dNV(GLuint id, GLsizei len, const GLubyte *name,

-                                GLdouble x, GLdouble y, GLdouble z, GLdouble w)

-{

-   _mesa_ProgramNamedParameter4fNV(id, len, name, (GLfloat)x, (GLfloat)y, 

-                                   (GLfloat)z, (GLfloat)w);

-}

-

-

-void GLAPIENTRY

-_mesa_ProgramNamedParameter4dvNV(GLuint id, GLsizei len, const GLubyte *name,

-                                 const double v[])

-{

-   _mesa_ProgramNamedParameter4fNV(id, len, name,

-                                   (GLfloat)v[0], (GLfloat)v[1],

-                                   (GLfloat)v[2], (GLfloat)v[3]);

-}

-

-

-void GLAPIENTRY

-_mesa_GetProgramNamedParameterfvNV(GLuint id, GLsizei len, const GLubyte *name,

-                                   GLfloat *params)

-{

-   struct gl_program *prog;

-   struct gl_fragment_program *fragProg;

-   const GLfloat *v;

-

-   GET_CURRENT_CONTEXT(ctx);

-

-   ASSERT_OUTSIDE_BEGIN_END(ctx);

-

-   prog = _mesa_lookup_program(ctx, id);

-   if (!prog || prog->Target != GL_FRAGMENT_PROGRAM_NV) {

-      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramNamedParameterNV");

-      return;

-   }

-

-   if (len <= 0) {

-      _mesa_error(ctx, GL_INVALID_VALUE, "glGetProgramNamedParameterNV");

-      return;

-   }

-

-   fragProg = (struct gl_fragment_program *) prog;

-   v = _mesa_lookup_parameter_value(fragProg->Base.Parameters,

-                                    len, (char *) name);

-   if (v) {

-      params[0] = v[0];

-      params[1] = v[1];

-      params[2] = v[2];

-      params[3] = v[3];

-      return;

-   }

-

-   _mesa_error(ctx, GL_INVALID_VALUE, "glGetProgramNamedParameterNV");

-}

-

-

-void GLAPIENTRY

-_mesa_GetProgramNamedParameterdvNV(GLuint id, GLsizei len, const GLubyte *name,

-                                   GLdouble *params)

-{

-   GLfloat floatParams[4];

-   _mesa_GetProgramNamedParameterfvNV(id, len, name, floatParams);

-   COPY_4V(params, floatParams);

-}

+/*
+ * Mesa 3-D graphics library
+ * Version:  6.5.2
+ *
+ * Copyright (C) 1999-2006  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.
+ */
+
+/**
+ * \file nvprogram.c
+ * NVIDIA vertex/fragment program state management functions.
+ * \author Brian Paul
+ */
+
+/*
+ * Regarding GL_NV_fragment/vertex_program, GL_NV_vertex_program1_1, etc:
+ *
+ * Portions of this software may use or implement intellectual
+ * property owned and licensed by NVIDIA Corporation. NVIDIA disclaims
+ * any and all warranties with respect to such intellectual property,
+ * including any use thereof or modifications thereto.
+ */
+
+#include "main/glheader.h"
+#include "main/context.h"
+#include "main/hash.h"
+#include "main/imports.h"
+#include "main/macros.h"
+#include "main/mtypes.h"
+#include "main/nvprogram.h"
+#include "program/arbprogparse.h"
+#include "program/nvfragparse.h"
+#include "program/nvvertparse.h"
+#include "program/program.h"
+#include "program/prog_instruction.h"
+#include "program/prog_parameter.h"
+
+
+
+/**
+ * Execute a vertex state program.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_ExecuteProgramNV(GLenum target, GLuint id, const GLfloat *params)
+{
+   struct gl_vertex_program *vprog;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target != GL_VERTEX_STATE_PROGRAM_NV) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glExecuteProgramNV");
+      return;
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+
+   vprog = (struct gl_vertex_program *) _mesa_lookup_program(ctx, id);
+
+   if (!vprog || vprog->Base.Target != GL_VERTEX_STATE_PROGRAM_NV) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glExecuteProgramNV");
+      return;
+   }
+   
+   _mesa_problem(ctx, "glExecuteProgramNV() not supported");
+}
+
+
+/**
+ * Determine if a set of programs is resident in hardware.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+GLboolean GLAPIENTRY
+_mesa_AreProgramsResidentNV(GLsizei n, const GLuint *ids,
+                            GLboolean *residences)
+{
+   GLint i, j;
+   GLboolean allResident = GL_TRUE;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END_WITH_RETVAL(ctx, GL_FALSE);
+
+   if (n < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV(n)");
+      return GL_FALSE;
+   }
+
+   for (i = 0; i < n; i++) {
+      const struct gl_program *prog;
+      if (ids[i] == 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV");
+         return GL_FALSE;
+      }
+      prog = _mesa_lookup_program(ctx, ids[i]);
+      if (!prog) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glAreProgramsResidentNV");
+         return GL_FALSE;
+      }
+      if (prog->Resident) {
+	 if (!allResident)
+	    residences[i] = GL_TRUE;
+      }
+      else {
+         if (allResident) {
+	    allResident = GL_FALSE;
+	    for (j = 0; j < i; j++)
+	       residences[j] = GL_TRUE;
+	 }
+	 residences[i] = GL_FALSE;
+      }
+   }
+
+   return allResident;
+}
+
+
+/**
+ * Request that a set of programs be resident in hardware.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_RequestResidentProgramsNV(GLsizei n, const GLuint *ids)
+{
+   GLint i;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (n < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(n)");
+      return;
+   }
+
+   /* just error checking for now */
+   for (i = 0; i < n; i++) {
+      struct gl_program *prog;
+
+      if (ids[i] == 0) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(id)");
+         return;
+      }
+
+      prog = _mesa_lookup_program(ctx, ids[i]);
+      if (!prog) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glRequestResidentProgramsNV(id)");
+         return;
+      }
+
+      /* XXX this is really a hardware thing we should hook out */
+      prog->Resident = GL_TRUE;
+   }
+}
+
+
+/**
+ * Get a program parameter register.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetProgramParameterfvNV(GLenum target, GLuint index,
+                              GLenum pname, GLfloat *params)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target == GL_VERTEX_PROGRAM_NV) {
+      if (pname == GL_PROGRAM_PARAMETER_NV) {
+         if (index < MAX_NV_VERTEX_PROGRAM_PARAMS) {
+            COPY_4V(params, ctx->VertexProgram.Parameters[index]);
+         }
+         else {
+            _mesa_error(ctx, GL_INVALID_VALUE,
+                        "glGetProgramParameterfvNV(index)");
+            return;
+         }
+      }
+      else {
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterfvNV(pname)");
+         return;
+      }
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterfvNV(target)");
+      return;
+   }
+}
+
+
+/**
+ * Get a program parameter register.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetProgramParameterdvNV(GLenum target, GLuint index,
+                              GLenum pname, GLdouble *params)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target == GL_VERTEX_PROGRAM_NV) {
+      if (pname == GL_PROGRAM_PARAMETER_NV) {
+         if (index < MAX_NV_VERTEX_PROGRAM_PARAMS) {
+            COPY_4V(params, ctx->VertexProgram.Parameters[index]);
+         }
+         else {
+            _mesa_error(ctx, GL_INVALID_VALUE,
+                        "glGetProgramParameterdvNV(index)");
+            return;
+         }
+      }
+      else {
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterdvNV(pname)");
+         return;
+      }
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramParameterdvNV(target)");
+      return;
+   }
+}
+
+
+/**
+ * Get a program attribute.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetProgramivNV(GLuint id, GLenum pname, GLint *params)
+{
+   struct gl_program *prog;
+   GET_CURRENT_CONTEXT(ctx);
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   prog = _mesa_lookup_program(ctx, id);
+   if (!prog) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramivNV");
+      return;
+   }
+
+   switch (pname) {
+      case GL_PROGRAM_TARGET_NV:
+         *params = prog->Target;
+         return;
+      case GL_PROGRAM_LENGTH_NV:
+         *params = prog->String ?(GLint) strlen((char *) prog->String) : 0;
+         return;
+      case GL_PROGRAM_RESIDENT_NV:
+         *params = prog->Resident;
+         return;
+      default:
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramivNV(pname)");
+         return;
+   }
+}
+
+
+/**
+ * Get the program source code.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetProgramStringNV(GLuint id, GLenum pname, GLubyte *program)
+{
+   struct gl_program *prog;
+   GET_CURRENT_CONTEXT(ctx);
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (pname != GL_PROGRAM_STRING_NV) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetProgramStringNV(pname)");
+      return;
+   }
+
+   prog = _mesa_lookup_program(ctx, id);
+   if (!prog) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramStringNV");
+      return;
+   }
+
+   if (prog->String) {
+      memcpy(program, prog->String, strlen((char *) prog->String));
+   }
+   else {
+      program[0] = 0;
+   }
+}
+
+
+/**
+ * Get matrix tracking information.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetTrackMatrixivNV(GLenum target, GLuint address,
+                         GLenum pname, GLint *params)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target == GL_VERTEX_PROGRAM_NV
+       && ctx->Extensions.NV_vertex_program) {
+      GLuint i;
+
+      if ((address & 0x3) || address >= MAX_NV_VERTEX_PROGRAM_PARAMS) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glGetTrackMatrixivNV(address)");
+         return;
+      }
+
+      i = address / 4;
+
+      switch (pname) {
+         case GL_TRACK_MATRIX_NV:
+            params[0] = (GLint) ctx->VertexProgram.TrackMatrix[i];
+            return;
+         case GL_TRACK_MATRIX_TRANSFORM_NV:
+            params[0] = (GLint) ctx->VertexProgram.TrackMatrixTransform[i];
+            return;
+         default:
+            _mesa_error(ctx, GL_INVALID_ENUM, "glGetTrackMatrixivNV");
+            return;
+      }
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTrackMatrixivNV");
+      return;
+   }
+}
+
+
+/**
+ * Get a vertex (or vertex array) attribute.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetVertexAttribdvNV(GLuint index, GLenum pname, GLdouble *params)
+{
+   const struct gl_client_array *array;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");
+      return;
+   }
+
+   array = &ctx->Array.ArrayObj->VertexAttrib[index];
+
+   switch (pname) {
+      case GL_ATTRIB_ARRAY_SIZE_NV:
+         params[0] = array->Size;
+         break;
+      case GL_ATTRIB_ARRAY_STRIDE_NV:
+         params[0] = array->Stride;
+         break;
+      case GL_ATTRIB_ARRAY_TYPE_NV:
+         params[0] = array->Type;
+         break;
+      case GL_CURRENT_ATTRIB_NV:
+         if (index == 0) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "glGetVertexAttribdvNV(index == 0)");
+            return;
+         }
+	 FLUSH_CURRENT(ctx, 0);
+         COPY_4V(params, ctx->Current.Attrib[index]);
+         break;
+      default:
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");
+         return;
+   }
+}
+
+/**
+ * Get a vertex (or vertex array) attribute.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetVertexAttribfvNV(GLuint index, GLenum pname, GLfloat *params)
+{
+   const struct gl_client_array *array;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");
+      return;
+   }
+
+   array = &ctx->Array.ArrayObj->VertexAttrib[index];
+
+   switch (pname) {
+      case GL_ATTRIB_ARRAY_SIZE_NV:
+         params[0] = (GLfloat) array->Size;
+         break;
+      case GL_ATTRIB_ARRAY_STRIDE_NV:
+         params[0] = (GLfloat) array->Stride;
+         break;
+      case GL_ATTRIB_ARRAY_TYPE_NV:
+         params[0] = (GLfloat) array->Type;
+         break;
+      case GL_CURRENT_ATTRIB_NV:
+         if (index == 0) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "glGetVertexAttribfvNV(index == 0)");
+            return;
+         }
+	 FLUSH_CURRENT(ctx, 0);
+         COPY_4V(params, ctx->Current.Attrib[index]);
+         break;
+      default:
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");
+         return;
+   }
+}
+
+/**
+ * Get a vertex (or vertex array) attribute.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetVertexAttribivNV(GLuint index, GLenum pname, GLint *params)
+{
+   const struct gl_client_array *array;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribdvNV(index)");
+      return;
+   }
+
+   array = &ctx->Array.ArrayObj->VertexAttrib[index];
+
+   switch (pname) {
+      case GL_ATTRIB_ARRAY_SIZE_NV:
+         params[0] = array->Size;
+         break;
+      case GL_ATTRIB_ARRAY_STRIDE_NV:
+         params[0] = array->Stride;
+         break;
+      case GL_ATTRIB_ARRAY_TYPE_NV:
+         params[0] = array->Type;
+         break;
+      case GL_CURRENT_ATTRIB_NV:
+         if (index == 0) {
+            _mesa_error(ctx, GL_INVALID_OPERATION,
+                        "glGetVertexAttribivNV(index == 0)");
+            return;
+         }
+	 FLUSH_CURRENT(ctx, 0);
+         params[0] = (GLint) ctx->Current.Attrib[index][0];
+         params[1] = (GLint) ctx->Current.Attrib[index][1];
+         params[2] = (GLint) ctx->Current.Attrib[index][2];
+         params[3] = (GLint) ctx->Current.Attrib[index][3];
+         break;
+      case GL_VERTEX_ATTRIB_ARRAY_BUFFER_BINDING_ARB:
+         params[0] = array->BufferObj->Name;
+         break;
+      default:
+         _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribdvNV");
+         return;
+   }
+}
+
+
+/**
+ * Get a vertex array attribute pointer.
+ * \note Not compiled into display lists.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_GetVertexAttribPointervNV(GLuint index, GLenum pname, GLvoid **pointer)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (index >= MAX_NV_VERTEX_PROGRAM_INPUTS) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetVertexAttribPointerNV(index)");
+      return;
+   }
+
+   if (pname != GL_ATTRIB_ARRAY_POINTER_NV) {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glGetVertexAttribPointerNV(pname)");
+      return;
+   }
+
+   *pointer = (GLvoid *) ctx->Array.ArrayObj->VertexAttrib[index].Ptr;
+}
+
+void
+_mesa_emit_nv_temp_initialization(struct gl_context *ctx,
+				  struct gl_program *program)
+{
+   struct prog_instruction *inst;
+   GLuint i;
+   struct gl_shader_compiler_options* options =
+         &ctx->ShaderCompilerOptions[_mesa_program_target_to_index(program->Target)];
+
+   if (!options->EmitNVTempInitialization)
+      return;
+
+   /* We'll swizzle up a zero temporary so we can use it for the
+    * ARL.
+    */
+   if (program->NumTemporaries == 0)
+      program->NumTemporaries = 1;
+
+   _mesa_insert_instructions(program, 0, program->NumTemporaries + 1);
+
+   for (i = 0; i < program->NumTemporaries; i++) {
+      struct prog_instruction *inst = &program->Instructions[i];
+
+      inst->Opcode = OPCODE_SWZ;
+      inst->DstReg.File = PROGRAM_TEMPORARY;
+      inst->DstReg.Index = i;
+      inst->DstReg.WriteMask = WRITEMASK_XYZW;
+      inst->SrcReg[0].File = PROGRAM_TEMPORARY;
+      inst->SrcReg[0].Index = 0;
+      inst->SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_ZERO,
+					      SWIZZLE_ZERO,
+					      SWIZZLE_ZERO,
+					      SWIZZLE_ZERO);
+   }
+
+   inst = &program->Instructions[i];
+   inst->Opcode = OPCODE_ARL;
+   inst->DstReg.File = PROGRAM_ADDRESS;
+   inst->DstReg.Index = 0;
+   inst->DstReg.WriteMask = WRITEMASK_XYZW;
+   inst->SrcReg[0].File = PROGRAM_TEMPORARY;
+   inst->SrcReg[0].Index = 0;
+   inst->SrcReg[0].Swizzle = SWIZZLE_XXXX;
+
+   if (program->NumAddressRegs == 0)
+      program->NumAddressRegs = 1;
+}
+
+void
+_mesa_setup_nv_temporary_count(struct gl_context *ctx, struct gl_program *program)
+{
+   GLuint i;
+
+   program->NumTemporaries = 0;
+   for (i = 0; i < program->NumInstructions; i++) {
+      struct prog_instruction *inst = &program->Instructions[i];
+
+      if (inst->DstReg.File == PROGRAM_TEMPORARY) {
+	 program->NumTemporaries = MAX2(program->NumTemporaries,
+					inst->DstReg.Index + 1);
+      }
+      if (inst->SrcReg[0].File == PROGRAM_TEMPORARY) {
+	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,
+					inst->SrcReg[0].Index + 1);
+      }
+      if (inst->SrcReg[1].File == PROGRAM_TEMPORARY) {
+	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,
+					inst->SrcReg[1].Index + 1);
+      }
+      if (inst->SrcReg[2].File == PROGRAM_TEMPORARY) {
+	 program->NumTemporaries = MAX2((GLint)program->NumTemporaries,
+					inst->SrcReg[2].Index + 1);
+      }
+   }
+}
+
+/**
+ * Load/parse/compile a program.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_LoadProgramNV(GLenum target, GLuint id, GLsizei len,
+                    const GLubyte *program)
+{
+   struct gl_program *prog;
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (!ctx->Extensions.NV_vertex_program
+       && !ctx->Extensions.NV_fragment_program) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV()");
+      return;
+   }
+
+   if (id == 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glLoadProgramNV(id)");
+      return;
+   }
+
+   if (len < 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glLoadProgramNV(len)");
+      return;
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+
+   prog = _mesa_lookup_program(ctx, id);
+
+   if (prog && prog->Target != 0 && prog->Target != target) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glLoadProgramNV(target)");
+      return;
+   }
+
+   if ((target == GL_VERTEX_PROGRAM_NV ||
+        target == GL_VERTEX_STATE_PROGRAM_NV)
+       && ctx->Extensions.NV_vertex_program) {
+      struct gl_vertex_program *vprog = (struct gl_vertex_program *) prog;
+      if (!vprog || prog == &_mesa_DummyProgram) {
+         vprog = (struct gl_vertex_program *)
+            ctx->Driver.NewProgram(ctx, target, id);
+         if (!vprog) {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");
+            return;
+         }
+         _mesa_HashInsert(ctx->Shared->Programs, id, vprog);
+      }
+
+      if (ctx->Extensions.ARB_vertex_program
+	  && (strncmp((char *) program, "!!ARB", 5) == 0)) {
+	 _mesa_parse_arb_vertex_program(ctx, target, program, len, vprog);
+      } else {
+	 _mesa_parse_nv_vertex_program(ctx, target, program, len, vprog);
+      }
+   }
+   else if (target == GL_FRAGMENT_PROGRAM_NV
+            && ctx->Extensions.NV_fragment_program) {
+      struct gl_fragment_program *fprog = (struct gl_fragment_program *) prog;
+      if (!fprog || prog == &_mesa_DummyProgram) {
+         fprog = (struct gl_fragment_program *)
+            ctx->Driver.NewProgram(ctx, target, id);
+         if (!fprog) {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");
+            return;
+         }
+         _mesa_HashInsert(ctx->Shared->Programs, id, fprog);
+      }
+      _mesa_parse_nv_fragment_program(ctx, target, program, len, fprog);
+   }
+   else if (target == GL_FRAGMENT_PROGRAM_ARB
+            && ctx->Extensions.ARB_fragment_program) {
+      struct gl_fragment_program *fprog = (struct gl_fragment_program *) prog;
+      if (!fprog || prog == &_mesa_DummyProgram) {
+         fprog = (struct gl_fragment_program *)
+            ctx->Driver.NewProgram(ctx, target, id);
+         if (!fprog) {
+            _mesa_error(ctx, GL_OUT_OF_MEMORY, "glLoadProgramNV");
+            return;
+         }
+         _mesa_HashInsert(ctx->Shared->Programs, id, fprog);
+      }
+      _mesa_parse_arb_fragment_program(ctx, target, program, len, fprog);
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glLoadProgramNV(target)");
+   }
+}
+
+
+
+/**
+ * Set a sequence of program parameter registers.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_ProgramParameters4dvNV(GLenum target, GLuint index,
+                             GLsizei num, const GLdouble *params)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {
+      GLint i;
+      if (index + num > MAX_NV_VERTEX_PROGRAM_PARAMS) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glProgramParameters4dvNV");
+         return;
+      }
+      for (i = 0; i < num; i++) {
+         ctx->VertexProgram.Parameters[index + i][0] = (GLfloat) params[0];
+         ctx->VertexProgram.Parameters[index + i][1] = (GLfloat) params[1];
+         ctx->VertexProgram.Parameters[index + i][2] = (GLfloat) params[2];
+         ctx->VertexProgram.Parameters[index + i][3] = (GLfloat) params[3];
+         params += 4;
+      };
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glProgramParameters4dvNV");
+      return;
+   }
+}
+
+
+/**
+ * Set a sequence of program parameter registers.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_ProgramParameters4fvNV(GLenum target, GLuint index,
+                             GLsizei num, const GLfloat *params)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {
+      GLint i;
+      if (index + num > MAX_NV_VERTEX_PROGRAM_PARAMS) {
+         _mesa_error(ctx, GL_INVALID_VALUE, "glProgramParameters4fvNV");
+         return;
+      }
+      for (i = 0; i < num; i++) {
+         COPY_4V(ctx->VertexProgram.Parameters[index + i], params);
+         params += 4;
+      }
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glProgramParameters4fvNV");
+      return;
+   }
+}
+
+
+
+/**
+ * Setup tracking of matrices into program parameter registers.
+ * \note Called from the GL API dispatcher.
+ */
+void GLAPIENTRY
+_mesa_TrackMatrixNV(GLenum target, GLuint address,
+                    GLenum matrix, GLenum transform)
+{
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+
+   if (target == GL_VERTEX_PROGRAM_NV && ctx->Extensions.NV_vertex_program) {
+      if (address & 0x3) {
+         /* addr must be multiple of four */
+         _mesa_error(ctx, GL_INVALID_VALUE, "glTrackMatrixNV(address)");
+         return;
+      }
+
+      switch (matrix) {
+         case GL_NONE:
+         case GL_MODELVIEW:
+         case GL_PROJECTION:
+         case GL_TEXTURE:
+         case GL_COLOR:
+         case GL_MODELVIEW_PROJECTION_NV:
+         case GL_MATRIX0_NV:
+         case GL_MATRIX1_NV:
+         case GL_MATRIX2_NV:
+         case GL_MATRIX3_NV:
+         case GL_MATRIX4_NV:
+         case GL_MATRIX5_NV:
+         case GL_MATRIX6_NV:
+         case GL_MATRIX7_NV:
+            /* OK, fallthrough */
+            break;
+         default:
+            _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(matrix)");
+            return;
+      }
+
+      switch (transform) {
+         case GL_IDENTITY_NV:
+         case GL_INVERSE_NV:
+         case GL_TRANSPOSE_NV:
+         case GL_INVERSE_TRANSPOSE_NV:
+            /* OK, fallthrough */
+            break;
+         default:
+            _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(transform)");
+            return;
+      }
+
+      ctx->VertexProgram.TrackMatrix[address / 4] = matrix;
+      ctx->VertexProgram.TrackMatrixTransform[address / 4] = transform;
+   }
+   else {
+      _mesa_error(ctx, GL_INVALID_ENUM, "glTrackMatrixNV(target)");
+      return;
+   }
+}
+
+
+void GLAPIENTRY
+_mesa_ProgramNamedParameter4fNV(GLuint id, GLsizei len, const GLubyte *name,
+                                GLfloat x, GLfloat y, GLfloat z, GLfloat w)
+{
+   struct gl_program *prog;
+   struct gl_fragment_program *fragProg;
+   gl_constant_value *v;
+
+   GET_CURRENT_CONTEXT(ctx);
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM_CONSTANTS);
+
+   prog = _mesa_lookup_program(ctx, id);
+   if (!prog || prog->Target != GL_FRAGMENT_PROGRAM_NV) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glProgramNamedParameterNV");
+      return;
+   }
+
+   if (len <= 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glProgramNamedParameterNV(len)");
+      return;
+   }
+
+   fragProg = (struct gl_fragment_program *) prog;
+   v = _mesa_lookup_parameter_value(fragProg->Base.Parameters, len,
+                                    (char *) name);
+   if (v) {
+      v[0].f = x;
+      v[1].f = y;
+      v[2].f = z;
+      v[3].f = w;
+      return;
+   }
+
+   _mesa_error(ctx, GL_INVALID_VALUE, "glProgramNamedParameterNV(name)");
+}
+
+
+void GLAPIENTRY
+_mesa_ProgramNamedParameter4fvNV(GLuint id, GLsizei len, const GLubyte *name,
+                                 const float v[])
+{
+   _mesa_ProgramNamedParameter4fNV(id, len, name, v[0], v[1], v[2], v[3]);
+}
+
+
+void GLAPIENTRY
+_mesa_ProgramNamedParameter4dNV(GLuint id, GLsizei len, const GLubyte *name,
+                                GLdouble x, GLdouble y, GLdouble z, GLdouble w)
+{
+   _mesa_ProgramNamedParameter4fNV(id, len, name, (GLfloat)x, (GLfloat)y, 
+                                   (GLfloat)z, (GLfloat)w);
+}
+
+
+void GLAPIENTRY
+_mesa_ProgramNamedParameter4dvNV(GLuint id, GLsizei len, const GLubyte *name,
+                                 const double v[])
+{
+   _mesa_ProgramNamedParameter4fNV(id, len, name,
+                                   (GLfloat)v[0], (GLfloat)v[1],
+                                   (GLfloat)v[2], (GLfloat)v[3]);
+}
+
+
+void GLAPIENTRY
+_mesa_GetProgramNamedParameterfvNV(GLuint id, GLsizei len, const GLubyte *name,
+                                   GLfloat *params)
+{
+   struct gl_program *prog;
+   struct gl_fragment_program *fragProg;
+   const gl_constant_value *v;
+
+   GET_CURRENT_CONTEXT(ctx);
+
+   ASSERT_OUTSIDE_BEGIN_END(ctx);
+
+   prog = _mesa_lookup_program(ctx, id);
+   if (!prog || prog->Target != GL_FRAGMENT_PROGRAM_NV) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glGetProgramNamedParameterNV");
+      return;
+   }
+
+   if (len <= 0) {
+      _mesa_error(ctx, GL_INVALID_VALUE, "glGetProgramNamedParameterNV");
+      return;
+   }
+
+   fragProg = (struct gl_fragment_program *) prog;
+   v = _mesa_lookup_parameter_value(fragProg->Base.Parameters,
+                                    len, (char *) name);
+   if (v) {
+      params[0] = v[0].f;
+      params[1] = v[1].f;
+      params[2] = v[2].f;
+      params[3] = v[3].f;
+      return;
+   }
+
+   _mesa_error(ctx, GL_INVALID_VALUE, "glGetProgramNamedParameterNV");
+}
+
+
+void GLAPIENTRY
+_mesa_GetProgramNamedParameterdvNV(GLuint id, GLsizei len, const GLubyte *name,
+                                   GLdouble *params)
+{
+   GLfloat floatParams[4];
+   _mesa_GetProgramNamedParameterfvNV(id, len, name, floatParams);
+   COPY_4V(params, floatParams);
+}
diff --git a/mesalib/src/mesa/main/texcompress_rgtc_tmp.h b/mesalib/src/mesa/main/texcompress_rgtc_tmp.h
index c8bf082a1..48bbd374e 100644
--- a/mesalib/src/mesa/main/texcompress_rgtc_tmp.h
+++ b/mesalib/src/mesa/main/texcompress_rgtc_tmp.h
@@ -181,7 +181,7 @@ static void TAG(encode_rgtc_chan)(TYPE *blkaddr, TYPE srccolors[4][4],
       fprintf(stderr, "%d ", alphaenc1[i]);
    }
    fprintf(stderr, "cutVals ");
-   for (i = 0; i < 8; i++) {
+   for (i = 0; i < 7; i++) {
       fprintf(stderr, "%d ", acutValues[i]);
    }
    fprintf(stderr, "srcVals ");
diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c
index 3021716a0..078a43ab1 100644
--- a/mesalib/src/mesa/main/texobj.c
+++ b/mesalib/src/mesa/main/texobj.c
@@ -842,7 +842,7 @@ _mesa_GenTextures( GLsizei n, GLuint *textures )
       struct gl_texture_object *texObj;
       GLuint name = first + i;
       GLenum target = 0;
-      texObj = (*ctx->Driver.NewTextureObject)( ctx, name, target);
+      texObj = ctx->Driver.NewTextureObject(ctx, name, target);
       if (!texObj) {
          _glthread_UNLOCK_MUTEX(ctx->Shared->Mutex);
          _mesa_error(ctx, GL_OUT_OF_MEMORY, "glGenTextures");
@@ -1066,7 +1066,7 @@ _mesa_BindTexture( GLenum target, GLuint texName )
       }
       else {
          /* if this is a new texture id, allocate a texture object now */
-         newTexObj = (*ctx->Driver.NewTextureObject)(ctx, texName, target);
+         newTexObj = ctx->Driver.NewTextureObject(ctx, texName, target);
          if (!newTexObj) {
             _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBindTexture");
             return;
@@ -1108,7 +1108,7 @@ _mesa_BindTexture( GLenum target, GLuint texName )
 
    /* Pass BindTexture call to device driver */
    if (ctx->Driver.BindTexture)
-      (*ctx->Driver.BindTexture)( ctx, target, newTexObj );
+      ctx->Driver.BindTexture(ctx, target, newTexObj);
 }
 
 
diff --git a/mesalib/src/mesa/main/texparam.c b/mesalib/src/mesa/main/texparam.c
index 134f15346..bbbb306b2 100644
--- a/mesalib/src/mesa/main/texparam.c
+++ b/mesalib/src/mesa/main/texparam.c
@@ -994,28 +994,21 @@ _mesa_GetTexLevelParameteriv( GLenum target, GLint level,
             *params = 0;
          break;
       case GL_TEXTURE_DEPTH_SIZE_ARB:
-         if (ctx->Extensions.ARB_depth_texture)
-            *params = _mesa_get_format_bits(texFormat, pname);
-         else
+         if (!ctx->Extensions.ARB_depth_texture)
             goto invalid_pname;
+         *params = _mesa_get_format_bits(texFormat, pname);
          break;
       case GL_TEXTURE_STENCIL_SIZE_EXT:
-         if (ctx->Extensions.EXT_packed_depth_stencil ||
-             ctx->Extensions.ARB_framebuffer_object) {
-            *params = _mesa_get_format_bits(texFormat, pname);
-         }
-         else {
+         if (!ctx->Extensions.EXT_packed_depth_stencil &&
+             !ctx->Extensions.ARB_framebuffer_object)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, pname);
          break;
       case GL_TEXTURE_SHARED_SIZE:
-         if (ctx->VersionMajor >= 3 ||
-             ctx->Extensions.EXT_texture_shared_exponent) {
-            *params = texFormat == MESA_FORMAT_RGB9_E5_FLOAT ? 5 : 0;
-         }
-         else {
+         if (ctx->VersionMajor < 3 &&
+             !ctx->Extensions.EXT_texture_shared_exponent)
             goto invalid_pname;
-         }
+         *params = texFormat == MESA_FORMAT_RGB9_E5_FLOAT ? 5 : 0;
          break;
 
       /* GL_ARB_texture_compression */
@@ -1036,67 +1029,46 @@ _mesa_GetTexLevelParameteriv( GLenum target, GLint level,
 
       /* GL_ARB_texture_float */
       case GL_TEXTURE_RED_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_RED_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_RED_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_GREEN_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_GREEN_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_GREEN_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_BLUE_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_BLUE_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_BLUE_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_ALPHA_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_ALPHA_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_ALPHA_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_LUMINANCE_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_LUMINANCE_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_LUMINANCE_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_INTENSITY_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_INTENSITY_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_INTENSITY_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
       case GL_TEXTURE_DEPTH_TYPE_ARB:
-         if (ctx->Extensions.ARB_texture_float) {
-            *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_DEPTH_SIZE) ?
-               _mesa_get_format_datatype(texFormat) : GL_NONE;
-         }
-         else {
+         if (!ctx->Extensions.ARB_texture_float)
             goto invalid_pname;
-         }
+         *params = _mesa_get_format_bits(texFormat, GL_TEXTURE_DEPTH_SIZE) ?
+            _mesa_get_format_datatype(texFormat) : GL_NONE;
          break;
 
       default:
@@ -1118,7 +1090,6 @@ void GLAPIENTRY
 _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
 {
    struct gl_texture_object *obj;
-   GLboolean error = GL_FALSE;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
@@ -1144,17 +1115,15 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
          *params = ENUM_TO_FLOAT(obj->Sampler.WrapR);
          break;
       case GL_TEXTURE_BORDER_COLOR:
-         if(ctx->NewState & (_NEW_BUFFERS | _NEW_FRAG_CLAMP))
+         if (ctx->NewState & (_NEW_BUFFERS | _NEW_FRAG_CLAMP))
             _mesa_update_state_locked(ctx);
-         if(ctx->Color._ClampFragmentColor)
-         {
+         if (ctx->Color._ClampFragmentColor) {
             params[0] = CLAMP(obj->Sampler.BorderColor.f[0], 0.0F, 1.0F);
             params[1] = CLAMP(obj->Sampler.BorderColor.f[1], 0.0F, 1.0F);
             params[2] = CLAMP(obj->Sampler.BorderColor.f[2], 0.0F, 1.0F);
             params[3] = CLAMP(obj->Sampler.BorderColor.f[3], 0.0F, 1.0F);
          }
-         else
-         {
+         else {
             params[0] = obj->Sampler.BorderColor.f[0];
             params[1] = obj->Sampler.BorderColor.f[1];
             params[2] = obj->Sampler.BorderColor.f[2];
@@ -1162,14 +1131,8 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
          }
          break;
       case GL_TEXTURE_RESIDENT:
-         {
-            GLboolean resident;
-            if (ctx->Driver.IsTextureResident)
-               resident = ctx->Driver.IsTextureResident(ctx, obj);
-            else
-               resident = GL_TRUE;
-            *params = ENUM_TO_FLOAT(resident);
-         }
+         *params = ctx->Driver.IsTextureResident ?
+            ctx->Driver.IsTextureResident(ctx, obj) : 1.0F;
          break;
       case GL_TEXTURE_PRIORITY:
          *params = obj->Priority;
@@ -1187,49 +1150,37 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
          *params = (GLfloat) obj->MaxLevel;
          break;
       case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-         if (ctx->Extensions.EXT_texture_filter_anisotropic) {
-            *params = obj->Sampler.MaxAnisotropy;
-         }
-	 else
-	    error = GL_TRUE;
+         if (!ctx->Extensions.EXT_texture_filter_anisotropic)
+            goto invalid_pname;
+         *params = obj->Sampler.MaxAnisotropy;
          break;
       case GL_TEXTURE_COMPARE_FAIL_VALUE_ARB:
-         if (ctx->Extensions.ARB_shadow_ambient) {
-            *params = obj->Sampler.CompareFailValue;
-         }
-	 else 
-	    error = GL_TRUE;
+         if (!ctx->Extensions.ARB_shadow_ambient)
+            goto invalid_pname;
+         *params = obj->Sampler.CompareFailValue;
          break;
       case GL_GENERATE_MIPMAP_SGIS:
 	 *params = (GLfloat) obj->GenerateMipmap;
          break;
       case GL_TEXTURE_COMPARE_MODE_ARB:
-         if (ctx->Extensions.ARB_shadow) {
-            *params = (GLfloat) obj->Sampler.CompareMode;
-         }
-	 else 
-	    error = GL_TRUE;
+         if (!ctx->Extensions.ARB_shadow)
+            goto invalid_pname;
+         *params = (GLfloat) obj->Sampler.CompareMode;
          break;
       case GL_TEXTURE_COMPARE_FUNC_ARB:
-         if (ctx->Extensions.ARB_shadow) {
-            *params = (GLfloat) obj->Sampler.CompareFunc;
-         }
-	 else 
-	    error = GL_TRUE;
+         if (!ctx->Extensions.ARB_shadow)
+            goto invalid_pname;
+         *params = (GLfloat) obj->Sampler.CompareFunc;
          break;
       case GL_DEPTH_TEXTURE_MODE_ARB:
-         if (ctx->Extensions.ARB_depth_texture) {
-            *params = (GLfloat) obj->Sampler.DepthMode;
-         }
-	 else 
-	    error = GL_TRUE;
+         if (!ctx->Extensions.ARB_depth_texture)
+            goto invalid_pname;
+         *params = (GLfloat) obj->Sampler.DepthMode;
          break;
       case GL_TEXTURE_LOD_BIAS:
-         if (ctx->Extensions.EXT_texture_lod_bias) {
-            *params = obj->Sampler.LodBias;
-         }
-	 else 
-	    error = GL_TRUE;
+         if (!ctx->Extensions.EXT_texture_lod_bias)
+            goto invalid_pname;
+         *params = obj->Sampler.LodBias;
          break;
 #if FEATURE_OES_draw_texture
       case GL_TEXTURE_CROP_RECT_OES:
@@ -1244,46 +1195,40 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params )
       case GL_TEXTURE_SWIZZLE_G_EXT:
       case GL_TEXTURE_SWIZZLE_B_EXT:
       case GL_TEXTURE_SWIZZLE_A_EXT:
-         if (ctx->Extensions.EXT_texture_swizzle) {
-            GLuint comp = pname - GL_TEXTURE_SWIZZLE_R_EXT;
-            *params = (GLfloat) obj->Swizzle[comp];
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.EXT_texture_swizzle)
+            goto invalid_pname;
+         *params = (GLfloat) obj->Swizzle[pname - GL_TEXTURE_SWIZZLE_R_EXT];
          break;
 
       case GL_TEXTURE_SWIZZLE_RGBA_EXT:
-         if (ctx->Extensions.EXT_texture_swizzle) {
+         if (!ctx->Extensions.EXT_texture_swizzle) {
+            goto invalid_pname;
+         }
+         else {
             GLuint comp;
             for (comp = 0; comp < 4; comp++) {
                params[comp] = (GLfloat) obj->Swizzle[comp];
             }
          }
-         else {
-            error = GL_TRUE;
-         }
          break;
 
       case GL_TEXTURE_CUBE_MAP_SEAMLESS:
-         if (ctx->Extensions.AMD_seamless_cubemap_per_texture) {
-            *params = (GLfloat) obj->Sampler.CubeMapSeamless;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.AMD_seamless_cubemap_per_texture)
+            goto invalid_pname;
+         *params = (GLfloat) obj->Sampler.CubeMapSeamless;
          break;
 
       default:
-	 error = GL_TRUE;
-	 break;
+         goto invalid_pname;
    }
 
-   if (error)
-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexParameterfv(pname=0x%x)",
-		  pname);
+   /* no error if we get here */
+   _mesa_unlock_texture(ctx, obj);
+   return;
 
+invalid_pname:
    _mesa_unlock_texture(ctx, obj);
+   _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexParameterfv(pname=0x%x)", pname);
 }
 
 
@@ -1291,13 +1236,12 @@ void GLAPIENTRY
 _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
 {
    struct gl_texture_object *obj;
-   GLboolean error = GL_FALSE;
    GET_CURRENT_CONTEXT(ctx);
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
-    obj = get_texobj(ctx, target, GL_TRUE);
-    if (!obj)
-       return;
+   obj = get_texobj(ctx, target, GL_TRUE);
+   if (!obj)
+      return;
 
    _mesa_lock_texture(ctx, obj);
    switch (pname) {
@@ -1330,14 +1274,8 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
          }
          break;;
       case GL_TEXTURE_RESIDENT:
-         {
-            GLboolean resident;
-            if (ctx->Driver.IsTextureResident)
-               resident = ctx->Driver.IsTextureResident(ctx, obj);
-            else
-               resident = GL_TRUE;
-            *params = (GLint) resident;
-         }
+         *params = ctx->Driver.IsTextureResident ?
+            ctx->Driver.IsTextureResident(ctx, obj) : 1;
          break;;
       case GL_TEXTURE_PRIORITY:
          *params = FLOAT_TO_INT(obj->Priority);
@@ -1355,55 +1293,37 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
          *params = obj->MaxLevel;
          break;;
       case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-         if (ctx->Extensions.EXT_texture_filter_anisotropic) {
-            *params = (GLint) obj->Sampler.MaxAnisotropy;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.EXT_texture_filter_anisotropic)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.MaxAnisotropy;
          break;
       case GL_TEXTURE_COMPARE_FAIL_VALUE_ARB:
-         if (ctx->Extensions.ARB_shadow_ambient) {
-            *params = (GLint) FLOAT_TO_INT(obj->Sampler.CompareFailValue);
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.ARB_shadow_ambient)
+            goto invalid_pname;
+         *params = (GLint) FLOAT_TO_INT(obj->Sampler.CompareFailValue);
          break;
       case GL_GENERATE_MIPMAP_SGIS:
 	 *params = (GLint) obj->GenerateMipmap;
          break;
       case GL_TEXTURE_COMPARE_MODE_ARB:
-         if (ctx->Extensions.ARB_shadow) {
-            *params = (GLint) obj->Sampler.CompareMode;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.ARB_shadow)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.CompareMode;
          break;
       case GL_TEXTURE_COMPARE_FUNC_ARB:
-         if (ctx->Extensions.ARB_shadow) {
-            *params = (GLint) obj->Sampler.CompareFunc;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.ARB_shadow)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.CompareFunc;
          break;
       case GL_DEPTH_TEXTURE_MODE_ARB:
-         if (ctx->Extensions.ARB_depth_texture) {
-            *params = (GLint) obj->Sampler.DepthMode;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.ARB_depth_texture)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.DepthMode;
          break;
       case GL_TEXTURE_LOD_BIAS:
-         if (ctx->Extensions.EXT_texture_lod_bias) {
-            *params = (GLint) obj->Sampler.LodBias;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.EXT_texture_lod_bias)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.LodBias;
          break;
 #if FEATURE_OES_draw_texture
       case GL_TEXTURE_CROP_RECT_OES:
@@ -1417,42 +1337,34 @@ _mesa_GetTexParameteriv( GLenum target, GLenum pname, GLint *params )
       case GL_TEXTURE_SWIZZLE_G_EXT:
       case GL_TEXTURE_SWIZZLE_B_EXT:
       case GL_TEXTURE_SWIZZLE_A_EXT:
-         if (ctx->Extensions.EXT_texture_swizzle) {
-            GLuint comp = pname - GL_TEXTURE_SWIZZLE_R_EXT;
-            *params = obj->Swizzle[comp];
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.EXT_texture_swizzle)
+            goto invalid_pname;
+         *params = obj->Swizzle[pname - GL_TEXTURE_SWIZZLE_R_EXT];
          break;
 
       case GL_TEXTURE_SWIZZLE_RGBA_EXT:
-         if (ctx->Extensions.EXT_texture_swizzle) {
-            COPY_4V(params, obj->Swizzle);
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.EXT_texture_swizzle)
+            goto invalid_pname;
+         COPY_4V(params, obj->Swizzle);
          break;
 
       case GL_TEXTURE_CUBE_MAP_SEAMLESS:
-         if (ctx->Extensions.AMD_seamless_cubemap_per_texture) {
-            *params = (GLint) obj->Sampler.CubeMapSeamless;
-         }
-         else {
-            error = GL_TRUE;
-         }
+         if (!ctx->Extensions.AMD_seamless_cubemap_per_texture)
+            goto invalid_pname;
+         *params = (GLint) obj->Sampler.CubeMapSeamless;
          break;
 
       default:
-         ; /* silence warnings */
+         goto invalid_pname;
    }
 
-   if (error)
-      _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexParameteriv(pname=0x%x)",
-		  pname);
+   /* no error if we get here */
+   _mesa_unlock_texture(ctx, obj);
+   return;
 
+invalid_pname:
    _mesa_unlock_texture(ctx, obj);
+   _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexParameteriv(pname=0x%x)", pname);
 }
 
 
@@ -1465,6 +1377,8 @@ _mesa_GetTexParameterIiv(GLenum target, GLenum pname, GLint *params)
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    texObj = get_texobj(ctx, target, GL_TRUE);
+   if (!texObj)
+      return;
    
    switch (pname) {
    case GL_TEXTURE_BORDER_COLOR:
@@ -1485,6 +1399,8 @@ _mesa_GetTexParameterIuiv(GLenum target, GLenum pname, GLuint *params)
    ASSERT_OUTSIDE_BEGIN_END(ctx);
 
    texObj = get_texobj(ctx, target, GL_TRUE);
+   if (!texObj)
+      return;
    
    switch (pname) {
    case GL_TEXTURE_BORDER_COLOR:
diff --git a/mesalib/src/mesa/main/uniforms.c b/mesalib/src/mesa/main/uniforms.c
index dd069a3a4..1329af4cd 100644
--- a/mesalib/src/mesa/main/uniforms.c
+++ b/mesalib/src/mesa/main/uniforms.c
@@ -429,7 +429,7 @@ get_uniform(struct gl_context *ctx, GLuint program, GLint location,
             for (i = 0; i < rows; i++) {
                const int base = paramPos + offset + i;
                for (j = 0; j < cols; j++ ) {
-                  params[k++] = prog->Parameters->ParameterValues[base][j];
+                  params[k++] = prog->Parameters->ParameterValues[base][j].f;
                }
             }
          }
@@ -442,7 +442,7 @@ get_uniform(struct gl_context *ctx, GLuint program, GLint location,
                const int base = paramPos + offset + i;
                for (j = 0; j < cols; j++ ) {
                   params[k++] = (GLdouble)
-                     prog->Parameters->ParameterValues[base][j];
+                     prog->Parameters->ParameterValues[base][j].f;
                }
             }
          }
@@ -454,8 +454,9 @@ get_uniform(struct gl_context *ctx, GLuint program, GLint location,
             for (i = 0; i < rows; i++) {
                const int base = paramPos + offset + i;
                for (j = 0; j < cols; j++ ) {
-                  params[k++] = (GLint)
-                     prog->Parameters->ParameterValues[base][j];
+                  params[k++] = ctx->Const.GLSLVersion <= 120 ? 
+                     (GLint) prog->Parameters->ParameterValues[base][j].f : 
+                     prog->Parameters->ParameterValues[base][j].i;
                }
             }
          }
@@ -467,8 +468,9 @@ get_uniform(struct gl_context *ctx, GLuint program, GLint location,
             for (i = 0; i < rows; i++) {
                const int base = paramPos + offset + i;
                for (j = 0; j < cols; j++ ) {
-                  params[k++] = (GLuint)
-                     prog->Parameters->ParameterValues[base][j];
+                  params[k++] = ctx->Const.GLSLVersion <= 120 ? 
+                     (GLuint) prog->Parameters->ParameterValues[base][j].f : 
+                     prog->Parameters->ParameterValues[base][j].u;
                }
             }
          }
@@ -670,7 +672,7 @@ set_program_uniform(struct gl_context *ctx, struct gl_program *program,
       /* loop over number of samplers to change */
       for (i = 0; i < count; i++) {
          GLuint sampler = (GLuint)
-            program->Parameters->ParameterValues[index + offset + i][0];
+            program->Parameters->ParameterValues[index+offset + i][0].f;
          GLuint texUnit = ((GLuint *) values)[i];
 
          /* check that the sampler (tex unit index) is legal */
@@ -735,42 +737,52 @@ set_program_uniform(struct gl_context *ctx, struct gl_program *program,
 
       /* loop over number of array elements */
       for (k = 0; k < count; k++) {
-         GLfloat *uniformVal;
+         gl_constant_value *uniformVal;
 
          if (offset + k >= slots) {
             /* Extra array data is ignored */
             break;
          }
 
-         /* uniformVal (the destination) is always float[4] */
+         /* uniformVal (the destination) is always gl_constant_value[4] */
          uniformVal = program->Parameters->ParameterValues[index + offset + k];
 
          if (basicType == GL_INT) {
-            /* convert user's ints to floats */
             const GLint *iValues = ((const GLint *) values) + k * elems;
             for (i = 0; i < elems; i++) {
-               uniformVal[i] = (GLfloat) iValues[i];
+               if (ctx->Const.GLSLVersion <= 120)
+                  uniformVal[i].f = (GLfloat) iValues[i];
+               else
+                  uniformVal[i].i = iValues[i];
             }
          }
          else if (basicType == GL_UNSIGNED_INT) {
-            /* convert user's uints to floats */
             const GLuint *iValues = ((const GLuint *) values) + k * elems;
             for (i = 0; i < elems; i++) {
-               uniformVal[i] = (GLfloat) iValues[i];
+               if (ctx->Const.GLSLVersion <= 120)
+                  uniformVal[i].f = (GLfloat)(GLuint) iValues[i];
+               else
+                  uniformVal[i].u = iValues[i];
             }
          }
          else {
             const GLfloat *fValues = ((const GLfloat *) values) + k * elems;
             assert(basicType == GL_FLOAT);
             for (i = 0; i < elems; i++) {
-               uniformVal[i] = fValues[i];
+               uniformVal[i].f = fValues[i];
             }
          }
 
-         /* if the uniform is bool-valued, convert to 1.0 or 0.0 */
+         /* if the uniform is bool-valued, convert to 1 or 0 */
          if (isUniformBool) {
             for (i = 0; i < elems; i++) {
-               uniformVal[i] = uniformVal[i] ? 1.0f : 0.0f;
+               if (basicType == GL_FLOAT)
+                  uniformVal[i].b = uniformVal[i].f != 0.0f ? 1 : 0;
+               else
+                  uniformVal[i].b = uniformVal[i].u ? 1 : 0;
+               
+               if (ctx->Const.GLSLVersion <= 120)
+                  uniformVal[i].f = uniformVal[i].b ? 1.0f : 0.0f;
             }
          }
       }
@@ -936,7 +948,7 @@ set_program_uniform_matrix(struct gl_context *ctx, struct gl_program *program,
             /* Ignore writes beyond the end of (the used part of) an array */
             return;
          }
-         v = program->Parameters->ParameterValues[index + offset];
+         v = (GLfloat *) program->Parameters->ParameterValues[index + offset];
          for (row = 0; row < rows; row++) {
             if (transpose) {
                v[row] = values[src + row * cols + col];