Merge commit 'origin/gallium-0.2' into gallium-xlib-rework

Conflicts: progs/glsl/Makefile
author: Keith Whitwell <[email protected]> 2009-01-09 10:08:06 +0000
committer: Keith Whitwell <[email protected]> 2009-01-09 10:08:06 +0000
commit: e3734593aea202e99e77febea7b86c904080939f (patch)
tree: 69856674e2062c55ec5eec95eb0e31225a943084
parent: 221352bbd79a0ea92ce31cffb65537f62ee5668e (diff)
parent: 5cad143e545775acacee294049345c6a3868c51d (diff)
126 files changed, 3178 insertions, 1489 deletions
diff --git a/Makefile b/Makefile
index 52772381713..3e41173ce5e 100644
--- a/Makefile
+++ b/Makefile
@@ -324,7 +324,7 @@ MAIN_FILES = \
 
 DRI_FILES = \
 	$(DIRECTORY)/include/GL/internal/dri_interface.h		\
-	$(DIRECTORY)/include/GL/internal/dri_sarea.h			\
+	$(DIRECTORY)/include/GL/internal/glcore.h			\
 	$(DIRECTORY)/include/GL/internal/sarea.h			\
 	$(DIRECTORY)/src/glx/Makefile					\
 	$(DIRECTORY)/src/glx/x11/Makefile				\
diff --git a/docs/contents.html b/docs/contents.html
index 257b13f68a7..b348d3d17f3 100644
--- a/docs/contents.html
+++ b/docs/contents.html
@@ -37,8 +37,8 @@ a:visited {
 
 <b>Download / Install</b>
 <ul>
-<li><a href="download.html" target="MainFrame">Downloading/Unpacking</a>
-<li><a href="install.html" target="MainFrame">Compilation/Installation</a>
+<li><a href="download.html" target="MainFrame">Downloading / Unpacking</a>
+<li><a href="install.html" target="MainFrame">Compiling / Installing</a>
 <li><a href="glu.html" target="MainFrame">SGI's GLU</a>
 <li><a href="precompiled.html" target="MainFrame">Precompiled Libraries</a>
 </ul>
diff --git a/docs/install.html b/docs/install.html
index 16ef013688a..2d72506f67d 100644
--- a/docs/install.html
+++ b/docs/install.html
@@ -1,58 +1,72 @@
 <HTML>
 
-<TITLE>Compilation and Installation</TITLE>
+<TITLE>Compiling and Installing</TITLE>
 
 <link rel="stylesheet" type="text/css" href="mesa.css"></head>
 
 <BODY>
 
 
-<H1>Compilation and Installation</H1>
+<H1>Compiling and Installing</H1>
 
 <ol>
 <li><a href="#unix-x11">Unix / X11</a>
+  <ul>
+  <li><a href="#prereq">Prerequisites for DRI and hardware acceleration</a>
+  <li><a href="#autoconf">Building with autoconf</a>
+  <li><a href="#traditional">Building with traditional Makefiles</a>
+  <li><a href="#libs">The Libraries</a>
+  <li><a href="#demos">Running the demos
+  <li><a href="#install">Installing the header and library files
+  <li><a href="#pkg-config">Building OpenGL programs with pkg-config
+  </ul>
 <li><a href="#windows">Windows</a>
-<li><a href="#vms">VMS</a>
 <li><a href="#other">Other</a>
 </ol>
-
+<br>
 
 
 <a name="unix-x11">
 <H2>1. Unix/X11 Compilation and Installation</H1>
 
 
-<h3>1.1 Prerequisites for DRI and Hardware Acceleration</h3>
+<a name="prereq">
+<h3>1.1 Prerequisites for DRI and hardware acceleration</h3>
 
 <p>
-To build Mesa 7.1 with DRI-based hardware acceleration you must first have
-the <a href="http://dri.freedesktop.org/libdrm/" target="_parent">DRM version 2.3.1</a>.
+The following are required for DRI-based hardware acceleration with Mesa 7.3:
 </p>
 
-<p>
-You should also be using the Xorg server version 1.4 or 1.5.
+<ul>
+<li><a href="http://xorg.freedesktop.org/releases/individual/proto/">driproto2</a> version 1.99.3 or later
+<li><a href="http://dri.freedesktop.org/libdrm/" target="_parent">DRM</a>
+version 2.4.3 or later
+<li>Xorg server version 1.4 or 1.5.
+</ul>
 </p>
 
 
-
-<h3>1.2 Autoconf Compilation</h3>
+<a name="autoconf">
+<h3>1.2 Building with Autoconf</h3>
 
 <p>
 Mesa may be <a href="autoconf.html">built using autoconf</a>.
 This should work well on most GNU-based systems.
-When that fails, the traditional Mesa build system is available.
+If that fails the traditional Mesa build system is available.
 
 
 
-<h3>1.3 Traditional Compilation</h3>
+<a name="traditional">
+<h3>1.3 Building with traditional Makefiles</h3>
 
 <p>
 The traditional Mesa build system is based on a collection of pre-defined
 system configurations.
 </p>
 <p>
-To see the list of configurations, type <b>make</b> alone.
-Then choose a configuration from the list and type <b>make configname</b>.
+To see the list of configurations, just type <code>make</code>.
+Then choose a configuration from the list and type <code>make</code>
+<em>configname</em>.
 </p>
 
 <p>
@@ -66,7 +80,7 @@ allow you to run OpenGL/GLX applications on any X server (regardless of
 whether it supports the GLX X server extension).
 You will <em>not</em> be able to use hardware 3D acceleration.
 <p>
-To compile stand-alone Mesa type <b>make</b> in the top-level directory.
+To compile stand-alone Mesa type <code>make</code> in the top-level directory.
 You'll see a list of supported system configurations.
 Choose one from the list (such as linux-x86), and type:
 </p>
@@ -109,6 +123,7 @@ Later, if you want to rebuild for a different configuration run
 </p>
 
 
+<a name="libs">
 <h3>1.4 The libraries</h3>
 
 <p>
@@ -169,6 +184,7 @@ If you built the DRI hardware drivers, you'll also see the DRI drivers:
 </pre>
 
 
+<a name="demos">
 <h3>1.5 Running the demos</h3>
 
 <p>
@@ -244,6 +260,7 @@ Retrace your steps if this doesn't look right.
 </p>
 
 
+<a name="install">
 <H3>1.6 Installing the header and library files</H3>
 
 <p>
@@ -284,7 +301,8 @@ This is a handy way to compare multiple OpenGL implementations.
 </p>
 
 
-<H3>1.7 Building OpenGL Programs With pkg-config</H3>
+<a name="pkg-config">
+<H3>1.7 Building OpenGL programs with pkg-config</H3>
 
 <p>
 Running <code>make install</code> will install package configuration files
@@ -314,25 +332,15 @@ Please see the <a href="README.WIN32">README.WIN32</a> file.
 
 
 
-
-<a name="vms">
-<H2>3. VMS Compilation and Installation</H1>
-
-<p>
-Please see the <a href="README.VMS">README.VMS</a> file.
-</p>
-
-
-
-
 <a name="other">
-<H2>4. Other systems</H1>
+<H2>3. Other systems</H1>
 
 <p>
 Documentation for other environments (some may be very out of date):
 </p>
 
 <UL>
+<li><A HREF="README.VMS">README.VMS</A> - VMS
 <LI><A HREF="README.GGI">README.GGI</A> - GGI
 <LI><A HREF="README.3DFX">README.3DFX</A> - 3Dfx/Glide driver
 <LI><A HREF="README.AMIWIN">README.AMIWIN</A> - Amiga Amiwin
diff --git a/progs/glsl/.gitignore b/progs/glsl/.gitignore
index 5d954519b24..40df5bb9780 100644
--- a/progs/glsl/.gitignore
+++ b/progs/glsl/.gitignore
@@ -21,3 +21,5 @@ texdemo1
 toyball
 trirast
 twoside
+vert-or-frag-only
+vert-tex
diff --git a/progs/glsl/Makefile b/progs/glsl/Makefile
index 34da3a5a709..a39170b8c9c 100644
--- a/progs/glsl/Makefile
+++ b/progs/glsl/Makefile
@@ -23,6 +23,7 @@ PROGS = \
 	noise \
 	points \
 	pointcoord \
+	samplers \
 	skinning \
 	texdemo1 \
 	toyball \
@@ -119,10 +120,10 @@ identity: identity.o shaderutil.o
 
 
 fragcoord.o: fragcoord.c extfuncs.h shaderutil.h
-	$(CC) -c -I$(INCDIR) $(CFLAGS) fragcoord.c
+	$(APP_CC) -c -I$(INCDIR) $(CFLAGS) fragcoord.c
 
 fragcoord: fragcoord.o shaderutil.o
-	$(CC) -I$(INCDIR) $(CFLAGS) $(LDFLAGS) fragcoord.o shaderutil.o $(LIBS) -o $@
+	$(APP_CC) -I$(INCDIR) $(CFLAGS) $(LDFLAGS) fragcoord.o shaderutil.o $(LIBS) -o $@
 
 
 mandelbrot.o: mandelbrot.c extfuncs.h shaderutil.h
@@ -159,6 +160,13 @@ pointcoord: pointcoord.o readtex.o shaderutil.o
 	$(APP_CC) -I$(INCDIR) $(CFLAGS) $(LDFLAGS) pointcoord.o readtex.o shaderutil.o $(LIBS) -o $@
 
 
+samplers.o: samplers.c readtex.h extfuncs.h shaderutil.h
+	$(APP_CC) -c -I$(INCDIR) $(CFLAGS) samplers.c
+
+samplers: samplers.o readtex.o shaderutil.o
+	$(APP_CC) -I$(INCDIR) $(CFLAGS) $(LDFLAGS) samplers.o readtex.o shaderutil.o $(LIBS) -o $@
+
+
 skinning.o: skinning.c readtex.h extfuncs.h shaderutil.h
 	$(APP_CC) -c -I$(INCDIR) $(CFLAGS) skinning.c
 
@@ -201,8 +209,6 @@ vert-or-frag-only: vert-or-frag-only.o shaderutil.o
 	$(APP_CC) -I$(INCDIR) $(CFLAGS) $(LDFLAGS) vert-or-frag-only.o shaderutil.o $(LIBS) -o $@
 
 
-
-
 vert-tex.o: vert-tex.c extfuncs.h shaderutil.h
 	$(APP_CC) -c -I$(INCDIR) $(CFLAGS) vert-tex.c
 
diff --git a/progs/glsl/samplers.c b/progs/glsl/samplers.c
new file mode 100644
index 00000000000..d2140097297
--- /dev/null
+++ b/progs/glsl/samplers.c
@@ -0,0 +1,357 @@
+/**
+ * Exercise all available GLSL texture samplers.
+ *
+ * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.
+ * 
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * 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.
+ */
+
+/**
+ * We generate a fragment shader which uses the maximum number of supported
+ * texture samplers.
+ * For each sampler we create a separate texture.  Each texture has a
+ * single strip of color at a different intensity.  The fragment shader
+ * samples all the textures at the same coordinate and sums the values.
+ * The result should be a quad with rows of colors of increasing intensity
+ * from bottom to top.
+ *
+ * Brian Paul
+ * 1 Jan 2009
+ */
+
+#include <assert.h>
+#include <math.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "GL/glut.h"
+#include "readtex.h"
+#include "extfuncs.h"
+#include "shaderutil.h"
+
+
+#define MAX_SAMPLERS 128
+
+
+static const char *Demo = "samplers";
+
+static GLuint Program;
+static GLint NumSamplers;
+static GLuint Textures[MAX_SAMPLERS];
+static GLfloat Xrot = 0.0, Yrot = .0, Zrot = 0.0;
+static GLfloat EyeDist = 10;
+static GLboolean Anim = GL_FALSE;
+
+
+static void
+DrawPolygon(GLfloat size)
+{
+   glPushMatrix();
+   glNormal3f(0, 0, 1);
+   glBegin(GL_POLYGON);
+
+   glMultiTexCoord2f(GL_TEXTURE0, 0, 0);
+   glVertex2f(-size, -size);
+
+   glMultiTexCoord2f(GL_TEXTURE0, 1, 0);
+   glVertex2f( size, -size);
+
+   glMultiTexCoord2f(GL_TEXTURE0, 1, 1);
+   glVertex2f( size,  size);
+
+   glMultiTexCoord2f(GL_TEXTURE0, 0, 1);
+   glVertex2f(-size,  size);
+
+   glEnd();
+   glPopMatrix();
+}
+
+
+static void
+draw(void)
+{
+   glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
+
+   glPushMatrix();
+      glTranslatef(0.0, 0.0, -EyeDist);
+      glRotatef(Zrot, 0, 0, 1);
+      glRotatef(Yrot, 0, 1, 0);
+      glRotatef(Xrot, 1, 0, 0);
+
+      DrawPolygon(3.0);
+
+   glPopMatrix();
+
+   glutSwapBuffers();
+}
+
+
+static void
+idle(void)
+{
+   GLfloat t = 0.05 * glutGet(GLUT_ELAPSED_TIME);
+   Yrot = t;
+   glutPostRedisplay();
+}
+
+
+static void
+key(unsigned char k, int x, int y)
+{
+   (void) x;
+   (void) y;
+   switch (k) {
+   case ' ':
+   case 'a':
+      Anim = !Anim;
+      if (Anim)
+         glutIdleFunc(idle);
+      else
+         glutIdleFunc(NULL);
+      break;
+   case 'z':
+      EyeDist -= 0.5;
+      if (EyeDist < 3.0)
+         EyeDist = 3.0;
+      break;
+   case 'Z':
+      EyeDist += 0.5;
+      if (EyeDist > 90.0)
+         EyeDist = 90;
+      break;
+   case 27:
+      exit(0);
+   }
+   glutPostRedisplay();
+}
+
+
+static void
+specialkey(int key, int x, int y)
+{
+   GLfloat step = 2.0;
+   (void) x;
+   (void) y;
+   switch (key) {
+   case GLUT_KEY_UP:
+      Xrot += step;
+      break;
+   case GLUT_KEY_DOWN:
+      Xrot -= step;
+      break;
+   case GLUT_KEY_LEFT:
+      Yrot -= step;
+      break;
+   case GLUT_KEY_RIGHT:
+      Yrot += step;
+      break;
+   }
+   glutPostRedisplay();
+}
+
+
+/* new window size or exposure */
+static void
+Reshape(int width, int height)
+{
+   GLfloat ar = (float) width / (float) height;
+   glViewport(0, 0, (GLint)width, (GLint)height);
+   glMatrixMode(GL_PROJECTION);
+   glLoadIdentity();
+   glFrustum(-2.0*ar, 2.0*ar, -2.0, 2.0, 4.0, 100.0);
+   glMatrixMode(GL_MODELVIEW);
+   glLoadIdentity();
+}
+
+
+static void
+InitTextures(void)
+{
+   const GLint size = MAX_SAMPLERS;
+   GLubyte *texImage;
+   GLenum filter = GL_NEAREST;
+   GLint stripeSize;
+   GLint s;
+
+   texImage = (GLubyte *) malloc(size * size * 4);
+
+   glGenTextures(NumSamplers, Textures);
+
+   /* size of texels stripe */
+   stripeSize = size / NumSamplers;
+
+   /* create a texture for each sampler */
+   for (s = 0; s < NumSamplers; s++) {
+      GLint x, y, ypos;
+      GLubyte intensity = 31 + s * (256-32) / (NumSamplers - 1);
+
+      printf("Texture %d: color = %d, %d, %d\n", s,
+             (int) intensity, 0, (int) intensity );
+
+      /* initialize the texture to black */
+      memset(texImage, 0, size * size * 4);
+
+      /* set a stripe of texels to the intensity value */
+      ypos = s * stripeSize;
+      for (y = 0; y < stripeSize; y++) {
+         for (x = 0; x < size; x++) {
+            GLint k = 4 * ((ypos + y) * size + x);
+            texImage[k + 0] = intensity;
+            texImage[k + 1] = intensity;
+            texImage[k + 2] = 0;
+            texImage[k + 3] = 255;
+         }
+      }
+
+      glActiveTexture(GL_TEXTURE0 + s);
+      glBindTexture(GL_TEXTURE_2D, Textures[s]);
+      gluBuild2DMipmaps(GL_TEXTURE_2D, 4, size, size,
+                        GL_RGBA, GL_UNSIGNED_BYTE, texImage);
+     
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, filter);
+      glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, filter);
+   }
+
+   free(texImage);
+}
+
+
+/**
+ * Generate a fragment shader that uses the given number of samplers.
+ */
+static char *
+GenFragmentShader(GLint numSamplers)
+{
+   const int maxLen = 10 * 1000;
+   char *prog = (char *) malloc(maxLen);
+   char *p = prog;
+   int s;
+
+   p += sprintf(p, "// Generated fragment shader:\n");
+   for (s = 0; s < numSamplers; s++) {
+      p += sprintf(p, "uniform sampler2D tex%d;\n", s);
+   }
+   p += sprintf(p, "void main()\n");
+   p += sprintf(p, "{\n");
+   p += sprintf(p, "   vec4 color = vec4(0.0);\n");
+   for (s = 0; s < numSamplers; s++) {
+      p += sprintf(p, "   color += texture2D(tex%d, gl_TexCoord[0].xy);\n", s);
+   }
+   p += sprintf(p, "   gl_FragColor = color;\n");
+   p += sprintf(p, "}\n");
+
+   assert(p - prog < maxLen);
+   return prog;
+}
+
+
+/** Create & bind shader program */
+static GLuint
+CreateProgram(void)
+{
+   GLuint fragShader, vertShader, program;
+   const char *vertShaderText = 
+      "void main() \n"
+      "{ \n"
+      "   gl_TexCoord[0] = gl_MultiTexCoord0; \n"
+      "   gl_Position = ftransform(); \n"
+      "} \n";
+   char *fragShaderText = GenFragmentShader(NumSamplers);
+
+   printf("%s", fragShaderText);
+
+   vertShader = CompileShaderText(GL_VERTEX_SHADER, vertShaderText);
+   fragShader = CompileShaderText(GL_FRAGMENT_SHADER, fragShaderText);
+   assert(vertShader);
+   program = LinkShaders(vertShader, fragShader);
+
+   glUseProgram_func(program);
+
+   free(fragShaderText);
+
+   return program;
+}
+
+
+static void
+InitProgram(void)
+{
+   GLint s;
+
+   Program = CreateProgram();
+
+   /* init sampler uniforms */
+   for (s = 0; s < NumSamplers; s++) {
+      char uname[10];
+      GLint loc;
+
+      sprintf(uname, "tex%d", s);
+      loc = glGetUniformLocation_func(Program, uname);
+      assert(loc >= 0);
+
+      glUniform1i_func(loc, s);
+   }
+}
+
+
+static void
+InitGL(void)
+{
+   if (!ShadersSupported()) {
+      printf("GLSL not supported!\n");
+      exit(1);
+   }
+
+   printf("GL_RENDERER = %s\n", (const char *) glGetString(GL_RENDERER));
+
+   GetExtensionFuncs();
+
+   glGetIntegerv(GL_MAX_TEXTURE_IMAGE_UNITS, &NumSamplers);
+   if (NumSamplers > MAX_SAMPLERS)
+      NumSamplers = MAX_SAMPLERS;
+   printf("Testing %d samplers\n", NumSamplers);
+
+   InitTextures();
+   InitProgram();
+
+   glClearColor(.6, .6, .9, 0);
+   glColor3f(1.0, 1.0, 1.0);
+
+   printf("Each color corresponds to a separate sampler/texture.\n");
+}
+
+
+int
+main(int argc, char *argv[])
+{
+   glutInit(&argc, argv);
+   glutInitWindowSize(500, 400);
+   glutInitDisplayMode(GLUT_RGB | GLUT_DEPTH | GLUT_DOUBLE);
+   glutCreateWindow(Demo);
+   glutReshapeFunc(Reshape);
+   glutKeyboardFunc(key);
+   glutSpecialFunc(specialkey);
+   glutDisplayFunc(draw);
+   if (Anim)
+      glutIdleFunc(idle);
+   InitGL();
+   glutMainLoop();
+   return 0;
+}
diff --git a/progs/glsl/twoside.c b/progs/glsl/twoside.c
index 672a00491e6..06488bd1759 100644
--- a/progs/glsl/twoside.c
+++ b/progs/glsl/twoside.c
@@ -26,27 +26,39 @@ static GLuint fragShader;
 static GLuint vertShader;
 static GLuint program;
 static GLint win = 0;
-static GLboolean anim = 0*GL_TRUE;
-static GLboolean DetermineInFragProg = GL_TRUE;
-static GLfloat Xrot = 30.0f;
+static GLboolean anim;
+static GLboolean DetermineFacingInFragProg;
+static GLfloat Xrot;
 static GLint u_fragface;
-static GLenum FrontWinding = GL_CCW;
+static GLenum FrontWinding;
 static int prevTime = 0;
 
 
-static const GLfloat Red[4] = {1, 0, 0, 0};
+static const GLfloat Red[4] = {1, 0, 0, 1};
 static const GLfloat Green[4] = {0, 1, 0, 0};
 
 
 static void
+SetDefaults(void)
+{
+   DetermineFacingInFragProg = GL_TRUE;
+   FrontWinding = GL_CCW;
+   Xrot = 30;
+   anim = 0;
+   glutIdleFunc(NULL);
+}
+
+
+static void
 Redisplay(void)
 {
+   const int sections = 20;
    int i;
    float radius = 2;
 
    glFrontFace(FrontWinding);
 
-   if (DetermineInFragProg) {
+   if (DetermineFacingInFragProg) {
       glUniform1i_func(u_fragface, 1);
       glDisable(GL_VERTEX_PROGRAM_TWO_SIDE);
    }
@@ -64,8 +76,8 @@ Redisplay(void)
    glBegin(GL_TRIANGLE_STRIP);
    glColor4fv(Red);
    glSecondaryColor3fv_func(Green);
-   for (i = 0; i < 20; i++) {
-      float a = i / 19.0 * M_PI * 2.0;
+   for (i = 0; i <= sections; i++) {
+      float a = (float) i / (sections) * M_PI * 2.0;
       float x = radius * cos(a);
       float y = radius * sin(a);
       glVertex3f(x, -1, y);
@@ -139,17 +151,15 @@ Key(unsigned char key, int x, int y)
       break;
    case 'f':
       printf("Using frag shader gl_FrontFacing\n");
-      DetermineInFragProg = GL_TRUE;
+      DetermineFacingInFragProg = GL_TRUE;
       break;
    case 'v':
       printf("Using vert shader Two-sided lighting\n");
-      DetermineInFragProg = GL_FALSE;
+      DetermineFacingInFragProg = GL_FALSE;
       break;
    case 'r':
       /* reset */
-      Xrot = 30;
-      anim = 0;
-      glutIdleFunc(NULL);
+      SetDefaults();
       break;
    case 's':
       Xrot += 5;
@@ -182,14 +192,16 @@ Init(void)
    static const char *fragShaderText =
       "uniform bool fragface; \n"
       "void main() { \n"
-#if 0
+#if 1
       "   if (!fragface || gl_FrontFacing) { \n"
       "      gl_FragColor = gl_Color; \n"
       "   } \n"
       "   else { \n"
+      "      // note: dim green to help debug \n"
       "      gl_FragColor = 0.8 * gl_SecondaryColor; \n"
       "   } \n"
 #else
+      /* DEBUG CODE */
       "   bool f = gl_FrontFacing; \n"
       "   if (f) { \n"
       "      gl_FragColor = vec4(1.0, 0.0, 0.0, 0.0); \n"
@@ -197,8 +209,6 @@ Init(void)
       "   else { \n"
       "      gl_FragColor = vec4(0.0, 1.0, 0.0, 0.0); \n"
       "   } \n"
-      "   //float g = float(gl_FrontFacing) * 0.5 + 0.5; \n"
-      "   //gl_FragColor = vec4(g); \n"
 #endif
       "} \n";
    static const char *vertShaderText =
@@ -241,6 +251,8 @@ Init(void)
    assert(glIsShader_func(vertShader));
 
    glEnable(GL_DEPTH_TEST);
+
+   SetDefaults();
 }
 
 
diff --git a/scons/gallium.py b/scons/gallium.py
index 6dbce62b8ed..3d5a0532eca 100644
--- a/scons/gallium.py
+++ b/scons/gallium.py
@@ -313,6 +313,7 @@ def generate(env):
             '-Wmissing-prototypes',
             '-Wno-long-long',
             '-ffast-math',
+            '-std=c99',
             '-pedantic',
             '-fmessage-length=0', # be nice to Eclipse
         ]
@@ -391,6 +392,16 @@ def generate(env):
     env.Append(CFLAGS = cflags)
     env.Append(CXXFLAGS = cflags)
 
+    if env['platform'] == 'windows' and msvc:
+        # Choose the appropriate MSVC CRT
+        # http://msdn.microsoft.com/en-us/library/2kzt1wy3.aspx
+        if env['debug']:
+            env.Append(CCFLAGS = ['/MTd'])
+            env.Append(SHCCFLAGS = ['/LDd'])
+        else:
+            env.Append(CCFLAGS = ['/MT'])
+            env.Append(SHCCFLAGS = ['/LD'])
+    
     # Assembler options
     if gcc:
         if env['machine'] == 'x86':
diff --git a/scons/generic.py b/scons/generic.py
index f0bb3de9fcd..0ed21a4e6ce 100644
--- a/scons/generic.py
+++ b/scons/generic.py
@@ -458,6 +458,16 @@ def generate(env):
     env.Append(CFLAGS = cflags)
     env.Append(CXXFLAGS = cflags)
 
+    if env['platform'] == 'windows' and msvc:
+        # Choose the appropriate MSVC CRT
+        # http://msdn.microsoft.com/en-us/library/2kzt1wy3.aspx
+        if env['debug']:
+            env.Append(CCFLAGS = ['/MTd'])
+            env.Append(SHCCFLAGS = ['/LDd'])
+        else:
+            env.Append(CCFLAGS = ['/MT'])
+            env.Append(SHCCFLAGS = ['/LD'])
+    
     # Assembler options
     if gcc:
         if env['machine'] == 'x86':
diff --git a/src/gallium/auxiliary/draw/draw_pipe_validate.c b/src/gallium/auxiliary/draw/draw_pipe_validate.c
index f34c68728ef..03e842ce082 100644
--- a/src/gallium/auxiliary/draw/draw_pipe_validate.c
+++ b/src/gallium/auxiliary/draw/draw_pipe_validate.c
@@ -33,6 +33,7 @@
 #include "draw_private.h"
 #include "draw_pipe.h"
 #include "draw_context.h"
+#include "draw_vbuf.h"
 
 static boolean points( unsigned prim )
 {
@@ -52,16 +53,28 @@ static boolean triangles( unsigned prim )
 }
 
 /**
- * Check if we need any special pipeline stages, or whether
- * prims/verts can go through untouched.  Don't test for bypass
- * clipping or vs modes, this function is just about the primitive
- * pipeline stages.
+ * Default version of a function to check if we need any special
+ * pipeline stages, or whether prims/verts can go through untouched.
+ * Don't test for bypass clipping or vs modes, this function is just
+ * about the primitive pipeline stages.
+ *
+ * This can be overridden by the driver.
  */
 boolean
 draw_need_pipeline(const struct draw_context *draw,
                    const struct pipe_rasterizer_state *rasterizer,
                    unsigned int prim )
 {
+   /* If the driver has overridden this, use that version: 
+    */
+   if (draw->render &&
+       draw->render->need_pipeline) 
+   {
+      return draw->render->need_pipeline( draw->render,
+                                          rasterizer,
+                                          prim );
+   }
+
    /* Don't have to worry about triangles turning into lines/points
     * and triggering the pipeline, because we have to trigger the
     * pipeline *anyway* if unfilled mode is active.
diff --git a/src/gallium/auxiliary/draw/draw_vbuf.h b/src/gallium/auxiliary/draw/draw_vbuf.h
index b0aa2df3099..7e1df88f0b8 100644
--- a/src/gallium/auxiliary/draw/draw_vbuf.h
+++ b/src/gallium/auxiliary/draw/draw_vbuf.h
@@ -30,7 +30,7 @@
  * Vertex buffer drawing stage.
  * 
  * \author Keith Whitwell <[email protected]>
- * \author Jos� Fonseca <[email protected]>
+ * \author Jose Fonseca <[email protected]>
  */
 
 #ifndef DRAW_VBUF_H_
@@ -38,6 +38,7 @@
 
 
 
+struct pipe_rasterizer_state;
 struct draw_context;
 struct vertex_info;
 
@@ -55,6 +56,17 @@ struct vbuf_render {
    unsigned max_vertex_buffer_bytes;
 
    /**
+    * Query if the hardware driver needs assistance for a particular
+    * combination of rasterizer state and primitive.
+    *
+    * Currently optional.
+    */
+   boolean (*need_pipeline)(const struct vbuf_render *render,
+                            const struct pipe_rasterizer_state *rasterizer,
+                            unsigned int prim );
+
+
+   /**
     * Get the hardware vertex format.
     * 
     * XXX: have this in draw_context instead?
diff --git a/src/gallium/auxiliary/draw/draw_vertex.h b/src/gallium/auxiliary/draw/draw_vertex.h
index a943607d7ed..c143cf23723 100644
--- a/src/gallium/auxiliary/draw/draw_vertex.h
+++ b/src/gallium/auxiliary/draw/draw_vertex.h
@@ -81,9 +81,9 @@ struct vertex_info
     * memcmp() comparisons.
     */
    struct {
-      ubyte interp_mode:4;      /**< INTERP_x */
-      ubyte emit:4;             /**< EMIT_x */
-      ubyte src_index;          /**< map to post-xform attribs */
+      unsigned interp_mode:4;      /**< INTERP_x */
+      unsigned emit:4;             /**< EMIT_x */
+      unsigned src_index:8;          /**< map to post-xform attribs */
    } attrib[PIPE_MAX_SHADER_INPUTS];
 };
 
diff --git a/src/gallium/auxiliary/draw/draw_vs.c b/src/gallium/auxiliary/draw/draw_vs.c
index 7f305304ff7..c057cd67fde 100644
--- a/src/gallium/auxiliary/draw/draw_vs.c
+++ b/src/gallium/auxiliary/draw/draw_vs.c
@@ -50,7 +50,7 @@ void draw_vs_set_constants( struct draw_context *draw,
                             const float (*constants)[4],
                             unsigned size )
 {
-   if (((unsigned)constants) & 0xf) {
+   if (((uintptr_t)constants) & 0xf) {
       if (size > draw->vs.const_storage_size) {
          if (draw->vs.aligned_constant_storage)
             align_free((void *)draw->vs.aligned_constant_storage);
diff --git a/src/gallium/auxiliary/rtasm/rtasm_ppc.c b/src/gallium/auxiliary/rtasm/rtasm_ppc.c
index b65bfa7bbdf..e9015ec2eb8 100644
--- a/src/gallium/auxiliary/rtasm/rtasm_ppc.c
+++ b/src/gallium/auxiliary/rtasm/rtasm_ppc.c
@@ -261,7 +261,7 @@ emit_vx(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB)
    inst.inst.vB = vB;
    inst.inst.op2 = op2;
    emit_instruction(p, inst.bits);
-};
+}
 
 
 union vxr_inst {
@@ -287,7 +287,7 @@ emit_vxr(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB)
    inst.inst.rC = 0;
    inst.inst.op2 = op2;
    emit_instruction(p, inst.bits);
-};
+}
 
 
 union va_inst {
@@ -313,7 +313,7 @@ emit_va(struct ppc_function *p, uint op2, uint vD, uint vA, uint vB, uint vC)
    inst.inst.vC = vC;
    inst.inst.op2 = op2;
    emit_instruction(p, inst.bits);
-};
+}
 
 
 union i_inst {
@@ -430,7 +430,7 @@ emit_d(struct ppc_function *p, uint op, uint rt, uint ra, int si)
    inst.inst.ra = ra;
    inst.inst.si = (unsigned) (si & 0xffff);
    emit_instruction(p, inst.bits);
-};
+}
 
 
 union a_inst {
@@ -459,7 +459,7 @@ emit_a(struct ppc_function *p, uint op, uint frt, uint fra, uint frb, uint op2,
    inst.inst.op2 = op2;
    inst.inst.rc = rc;
    emit_instruction(p, inst.bits);
-};
+}
 
 
 union xo_inst {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index c2a0ac5aff8..2ed8c2bf07b 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -180,14 +180,16 @@ _dump_register_ind(
    uint file,
    int index,
    uint ind_file,
-   int ind_index )
+   int ind_index,
+   uint ind_swizzle )
 {
    ENM( file, file_names );
    CHR( '[' );
    ENM( ind_file, file_names );
    CHR( '[' );
    SID( ind_index );
-   CHR( ']' );
+   TXT( "]." );
+   ENM( ind_swizzle, swizzle_names );
    if (index != 0) {
       if (index > 0)
          CHR( '+' );
@@ -385,7 +387,8 @@ iter_instruction(
             src->SrcRegister.File,
             src->SrcRegister.Index,
             src->SrcRegisterInd.File,
-            src->SrcRegisterInd.Index );
+            src->SrcRegisterInd.Index,
+            src->SrcRegisterInd.SwizzleX );
       }
       else {
          _dump_register(
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
index be25cb45a0a..c575b6c3e1f 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump_c.c
@@ -646,7 +646,6 @@ tgsi_dump_c(
    struct tgsi_full_declaration fd;
    uint ignored = flags & TGSI_DUMP_C_IGNORED;
    uint deflt = flags & TGSI_DUMP_C_DEFAULT;
-   uint instno = 0;
 
    tgsi_parse_init( &parse, tokens );
 
diff --git a/src/gallium/auxiliary/tgsi/tgsi_scan.c b/src/gallium/auxiliary/tgsi/tgsi_scan.c
index cfc7ea8e898..1239f6c0765 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_scan.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_scan.c
@@ -124,7 +124,7 @@ tgsi_scan_shader(const struct tgsi_token *tokens,
                /* only first 32 regs will appear in this bitfield */
                info->file_mask[file] |= (1 << reg);
                info->file_count[file]++;
-               info->file_max[file] = MAX2(info->file_max[file], (int)i);
+               info->file_max[file] = MAX2(info->file_max[file], (int)reg);
 
                if (file == TGSI_FILE_INPUT) {
                   info->input_semantic_name[reg] = (ubyte)fulldecl->Semantic.SemanticName;
diff --git a/src/gallium/auxiliary/util/SConscript b/src/gallium/auxiliary/util/SConscript
index f5bd308083b..5df932c0f71 100644
--- a/src/gallium/auxiliary/util/SConscript
+++ b/src/gallium/auxiliary/util/SConscript
@@ -23,6 +23,7 @@ util = env.ConvenienceLibrary(
 		'u_stream_wd.c',
 		'u_tile.c',
 		'u_time.c',
+		'u_timed_winsys.c',
 	])
 
 auxiliaries.insert(0, util)
diff --git a/src/gallium/auxiliary/util/u_gen_mipmap.c b/src/gallium/auxiliary/util/u_gen_mipmap.c
index 30f161fe3b3..b5eb896b7a1 100644
--- a/src/gallium/auxiliary/util/u_gen_mipmap.c
+++ b/src/gallium/auxiliary/util/u_gen_mipmap.c
@@ -1198,7 +1198,7 @@ make_3d_mipmap(struct gen_mipmap_state *ctx,
 {
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
-   uint dstLevel, zslice;
+   uint dstLevel, zslice = 0;
 
    assert(pt->block.width == 1);
    assert(pt->block.height == 1);
diff --git a/src/gallium/drivers/cell/ppu/cell_context.c b/src/gallium/drivers/cell/ppu/cell_context.c
index 22d552d8e3d..8f502823f99 100644
--- a/src/gallium/drivers/cell/ppu/cell_context.c
+++ b/src/gallium/drivers/cell/ppu/cell_context.c
@@ -162,7 +162,7 @@ cell_create_context(struct pipe_screen *screen,
     */
    /* This call only works with SDK 3.0.  Anyone still using 2.1??? */
    cell->num_cells = spe_cpu_info_get(SPE_COUNT_PHYSICAL_CPU_NODES, -1);
-   cell->num_spus = spe_cpu_info_get(SPE_COUNT_USABLE_SPES, 0);
+   cell->num_spus = spe_cpu_info_get(SPE_COUNT_USABLE_SPES, -1);
    if (cell->debug_flags) {
       printf("Cell: found %d Cell(s) with %u SPUs\n",
              cell->num_cells, cell->num_spus);
diff --git a/src/gallium/drivers/cell/ppu/cell_gen_fp.c b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
index 96a1743fc10..8f3deb482e6 100644
--- a/src/gallium/drivers/cell/ppu/cell_gen_fp.c
+++ b/src/gallium/drivers/cell/ppu/cell_gen_fp.c
@@ -2,6 +2,7 @@
  * 
  * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
  * All Rights Reserved.
+ * Copyright 2009 VMware, Inc.  All rights reserved.
  * 
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
@@ -75,13 +76,25 @@ struct codegen
 
    int one_reg;         /**< register containing {1.0, 1.0, 1.0, 1.0} */
 
+   int addr_reg;        /**< address register, integer values */
+
    /** Per-instruction temps / intermediate temps */
    int num_itemps;
    int itemps[12];
 
    /** Current IF/ELSE/ENDIF nesting level */
    int if_nesting;
-   /** Index of execution mask register */
+   /** Current BGNLOOP/ENDLOOP nesting level */
+   int loop_nesting;
+   /** Location of start of current loop */
+   int loop_start;
+
+   /** Index of if/conditional mask register */
+   int cond_mask_reg;
+   /** Index of loop mask register */
+   int loop_mask_reg;
+
+   /** Index of master execution mask register */
    int exec_mask_reg;
 
    /** KIL mask: indicates which fragments have been killed */
@@ -145,10 +158,33 @@ get_const_one_reg(struct codegen *gen)
 
 
 /**
- * Return index of the pixel execution mask.
+ * Return index of the address register.
+ * Used for indirect register loads/stores.
+ */
+static int
+get_address_reg(struct codegen *gen)
+{
+   if (gen->addr_reg <= 0) {
+      gen->addr_reg = spe_allocate_available_register(gen->f);
+
+      spe_indent(gen->f, 4);
+      spe_comment(gen->f, -4, "INIT CONSTANT 1.0:");
+
+      /* init addr = {0, 0, 0, 0} */
+      spe_zero(gen->f, gen->addr_reg);
+
+      spe_indent(gen->f, -4);
+   }
+
+   return gen->addr_reg;
+}
+
+
+/**
+ * Return index of the master execution mask.
  * The register is allocated an initialized upon the first call.
  *
- * The pixel execution mask controls which pixels in a quad are
+ * The master execution mask controls which pixels in a quad are
  * modified, according to surrounding conditionals, loops, etc.
  */
 static int
@@ -157,19 +193,40 @@ get_exec_mask_reg(struct codegen *gen)
    if (gen->exec_mask_reg <= 0) {
       gen->exec_mask_reg = spe_allocate_available_register(gen->f);
 
-      spe_indent(gen->f, 4);
-      spe_comment(gen->f, -4, "INIT EXEC MASK = ~0:");
-
-      /* exec_mask = {~0, ~0, ~0, ~0} */
+      /* XXX this may not be needed */
+      spe_comment(gen->f, 0*-4, "initialize master execution mask = ~0");
       spe_load_int(gen->f, gen->exec_mask_reg, ~0);
-
-      spe_indent(gen->f, -4);
    }
 
    return gen->exec_mask_reg;
 }
 
 
+/** Return index of the conditional (if/else) execution mask register */
+static int
+get_cond_mask_reg(struct codegen *gen)
+{
+   if (gen->cond_mask_reg <= 0) {
+      gen->cond_mask_reg = spe_allocate_available_register(gen->f);
+   }
+
+   return gen->cond_mask_reg;
+}
+
+
+/** Return index of the loop execution mask register */
+static int
+get_loop_mask_reg(struct codegen *gen)
+{
+   if (gen->loop_mask_reg <= 0) {
+      gen->loop_mask_reg = spe_allocate_available_register(gen->f);
+   }
+
+   return gen->loop_mask_reg;
+}
+
+
+
 static boolean
 is_register_src(struct codegen *gen, int channel,
                 const struct tgsi_full_src_register *src)
@@ -231,16 +288,22 @@ get_src_reg(struct codegen *gen,
       spe_xor(gen->f, reg, reg, reg);
    }
    else {
+      int index = src->SrcRegister.Index;
+
       assert(swizzle < 4);
 
+      if (src->SrcRegister.Indirect) {
+         /* XXX unfinished */
+      }
+
       switch (src->SrcRegister.File) {
       case TGSI_FILE_TEMPORARY:
-         reg = gen->temp_regs[src->SrcRegister.Index][swizzle];
+         reg = gen->temp_regs[index][swizzle];
          break;
       case TGSI_FILE_INPUT:
          {
             /* offset is measured in quadwords, not bytes */
-            int offset = src->SrcRegister.Index * 4 + swizzle;
+            int offset = index * 4 + swizzle;
             reg = get_itemp(gen);
             reg_is_itemp = TRUE;
             /* Load:  reg = memory[(machine_reg) + offset] */
@@ -248,12 +311,12 @@ get_src_reg(struct codegen *gen,
          }
          break;
       case TGSI_FILE_IMMEDIATE:
-         reg = gen->imm_regs[src->SrcRegister.Index][swizzle];
+         reg = gen->imm_regs[index][swizzle];
          break;
       case TGSI_FILE_CONSTANT:
          {
             /* offset is measured in quadwords, not bytes */
-            int offset = src->SrcRegister.Index * 4 + swizzle;
+            int offset = index * 4 + swizzle;
             reg = get_itemp(gen);
             reg_is_itemp = TRUE;
             /* Load:  reg = memory[(machine_reg) + offset] */
@@ -322,7 +385,7 @@ get_dst_reg(struct codegen *gen,
 
    switch (dest->DstRegister.File) {
    case TGSI_FILE_TEMPORARY:
-      if (gen->if_nesting > 0)
+      if (gen->if_nesting > 0 || gen->loop_nesting > 0)
          reg = get_itemp(gen);
       else
          reg = gen->temp_regs[dest->DstRegister.Index][channel];
@@ -367,7 +430,7 @@ store_dest_reg(struct codegen *gen,
 
    switch (dest->DstRegister.File) {
    case TGSI_FILE_TEMPORARY:
-      if (gen->if_nesting > 0) {
+      if (gen->if_nesting > 0 || gen->loop_nesting > 0) {
          int d_reg = gen->temp_regs[dest->DstRegister.Index][channel];
          int exec_reg = get_exec_mask_reg(gen);
          /* Mix d with new value according to exec mask:
@@ -384,7 +447,7 @@ store_dest_reg(struct codegen *gen,
       {
          /* offset is measured in quadwords, not bytes */
          int offset = dest->DstRegister.Index * 4 + channel;
-         if (gen->if_nesting > 0) {
+         if (gen->if_nesting > 0 || gen->loop_nesting > 0) {
             int exec_reg = get_exec_mask_reg(gen);
             int curval_reg = get_itemp(gen);
             /* First read the current value from memory:
@@ -488,96 +551,118 @@ emit_epilogue(struct codegen *gen)
 }
 
 
+#define FOR_EACH_ENABLED_CHANNEL(inst, ch) \
+   for (ch = 0; ch < 4; ch++) \
+      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch))
+
+
+static boolean
+emit_ARL(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int ch = 0, src_reg, addr_reg;
+
+   spe_comment(gen->f, -4, "ARL:");
+
+   src_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+   addr_reg = get_address_reg(gen);
+
+   /* convert float to int */
+   spe_cflts(gen->f, addr_reg, src_reg, 0);
+
+   free_itemps(gen);
+
+   return TRUE;
+}
+
+
 static boolean
 emit_MOV(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch, src_reg[4], dst_reg[4];
 
    spe_comment(gen->f, -4, "MOV:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         src_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         dst_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-      }
+
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      src_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      dst_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
    }
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         if (is_register_src(gen, ch, &inst->FullSrcRegisters[0]) &&
-             is_memory_dst(gen, ch, &inst->FullDstRegisters[0])) {
-            /* special-case: register to memory store */
-            store_dest_reg(gen, src_reg[ch], ch, &inst->FullDstRegisters[0]);
-         }
-         else {
-            spe_move(gen->f, dst_reg[ch], src_reg[ch]);
-            store_dest_reg(gen, dst_reg[ch], ch, &inst->FullDstRegisters[0]);
-         }
-         free_itemps(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      if (is_register_src(gen, ch, &inst->FullSrcRegisters[0]) &&
+          is_memory_dst(gen, ch, &inst->FullDstRegisters[0])) {
+         /* special-case: register to memory store */
+         store_dest_reg(gen, src_reg[ch], ch, &inst->FullDstRegisters[0]);
+      }
+      else {
+         spe_move(gen->f, dst_reg[ch], src_reg[ch]);
+         store_dest_reg(gen, dst_reg[ch], ch, &inst->FullDstRegisters[0]);
       }
    }
-   return true;
+
+   free_itemps(gen);
+
+   return TRUE;
 }
 
 /**
- * Emit addition instructions.  Recall that a single TGSI_OPCODE_ADD
- * becomes (up to) four SPU "fa" instructions because we're doing SOA
- * processing.
+ * Emit binary operation
  */
 static boolean
-emit_ADD(struct codegen *gen, const struct tgsi_full_instruction *inst)
+emit_binop(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch, s1_reg[4], s2_reg[4], d_reg[4];
 
-   spe_comment(gen->f, -4, "ADD:");
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_ADD:
+      spe_comment(gen->f, -4, "ADD:");
+      break;
+   case TGSI_OPCODE_SUB:
+      spe_comment(gen->f, -4, "SUB:");
+      break;
+   case TGSI_OPCODE_MUL:
+      spe_comment(gen->f, -4, "MUL:");
+      break;
+   default:
+      assert(0);
+   }
+
    /* Loop over Red/Green/Blue/Alpha channels, fetch src operands */
-   for (ch = 0; ch < 4; ch++) {
-      /* If the dest R, G, B or A writemask is enabled... */
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
    }
-   /* Loop over Red/Green/Blue/Alpha channels, do the add, store results */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         /* Emit actual SPE instruction: d = s1 + s2 */
+
+   /* Loop over Red/Green/Blue/Alpha channels, do the op, store results */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      /* Emit actual SPE instruction: d = s1 + s2 */
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_ADD:
          spe_fa(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
-         /* Store the result (a no-op for TGSI_FILE_TEMPORARY dests) */
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-         /* Free any intermediate temps we allocated */
-         free_itemps(gen);
+         break;
+      case TGSI_OPCODE_SUB:
+         spe_fs(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      case TGSI_OPCODE_MUL:
+         spe_fm(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      default:
+         ;
       }
    }
-   return true;
-}
 
-/**
- * Emit subtract.  See emit_ADD for comments.
- */
-static boolean
-emit_SUB(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch, s1_reg[4], s2_reg[4], d_reg[4];
-   spe_comment(gen->f, -4, "SUB:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-      }
+   /* Store the result (a no-op for TGSI_FILE_TEMPORARY dests) */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         /* d = s1 - s2 */
-         spe_fs(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
-   }
-   return true;
+
+   /* Free any intermediate temps we allocated */
+   free_itemps(gen);
+
+   return TRUE;
 }
 
+
 /**
  * Emit multiply add.  See emit_ADD for comments.
  */
@@ -586,23 +671,20 @@ emit_MAD(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch, s1_reg[4], s2_reg[4], s3_reg[4], d_reg[4];
    spe_comment(gen->f, -4, "MAD:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         /* d = s1 * s2 + s3 */
-         spe_fma(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch], s3_reg[ch]);
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fma(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch], s3_reg[ch]);
+   }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
-   return true;
+   free_itemps(gen);
+   return TRUE;
 }
 
 
@@ -615,132 +697,108 @@ emit_LERP(struct codegen *gen, const struct tgsi_full_instruction *inst)
    int ch, s1_reg[4], s2_reg[4], s3_reg[4], d_reg[4], tmp_reg[4];
    spe_comment(gen->f, -4, "LERP:");
    /* setup/get src/dst/temp regs */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         tmp_reg[ch] = get_itemp(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      s3_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      tmp_reg[ch] = get_itemp(gen);
    }
 
    /* d = s3 + s1(s2 - s3) */
    /* do all subtracts, then all fma, then all stores to better pipeline */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         spe_fs(gen->f, tmp_reg[ch], s2_reg[ch], s3_reg[ch]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fs(gen->f, tmp_reg[ch], s2_reg[ch], s3_reg[ch]);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         spe_fma(gen->f, d_reg[ch], tmp_reg[ch], s1_reg[ch], s3_reg[ch]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fma(gen->f, d_reg[ch], tmp_reg[ch], s1_reg[ch], s3_reg[ch]);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
+
+
 /**
- * Emit multiply.  See emit_ADD for comments.
+ * Emit reciprocal or recip sqrt.
  */
 static boolean
-emit_MUL(struct codegen *gen, const struct tgsi_full_instruction *inst)
+emit_RCP_RSQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch, s1_reg[4], s2_reg[4], d_reg[4];
-   spe_comment(gen->f, -4, "MUL:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-      }
+   int ch, s1_reg[4], d_reg[4], tmp_reg[4];
+
+   if (inst->Instruction.Opcode == TGSI_OPCODE_RCP) {
+      spe_comment(gen->f, -4, "RCP:");
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         /* d = s1 * s2 */
-         spe_fm(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   else {
+      assert(inst->Instruction.Opcode == TGSI_OPCODE_RSQ);
+      spe_comment(gen->f, -4, "RSQ:");
    }
-   return true;
-}
 
-/**
- * Emit reciprocal.  See emit_ADD for comments.
- */
-static boolean
-emit_RCP(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-   spe_comment(gen->f, -4, "RCP:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         /* d = 1/s1 */
-         spe_frest(gen->f, d_reg, s1_reg);
-         spe_fi(gen->f, d_reg, s1_reg, d_reg);
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      tmp_reg[ch] = get_itemp(gen);
    }
-   return true;
-}
 
-/**
- * Emit reciprocal sqrt.  See emit_ADD for comments.
- */
-static boolean
-emit_RSQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-   spe_comment(gen->f, -4, "RSQ:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         /* d = 1/s1 */
-         spe_frsqest(gen->f, d_reg, s1_reg);
-         spe_fi(gen->f, d_reg, s1_reg, d_reg);
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      if (inst->Instruction.Opcode == TGSI_OPCODE_RCP) {
+         /* tmp = 1/s1 */
+         spe_frest(gen->f, tmp_reg[ch], s1_reg[ch]);
+      }
+      else {
+         /* tmp = 1/sqrt(s1) */
+         spe_frsqest(gen->f, tmp_reg[ch], s1_reg[ch]);
       }
    }
-   return true;
+
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      /* d = float_interp(s1, tmp) */
+      spe_fi(gen->f, d_reg[ch], s1_reg[ch], tmp_reg[ch]);
+   }
+
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+   }
+
+   free_itemps(gen);
+   return TRUE;
 }
 
+
 /**
  * Emit absolute value.  See emit_ADD for comments.
  */
 static boolean
 emit_ABS(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, s1_reg[4], d_reg[4];
+   const int bit31mask_reg = get_itemp(gen);
+
    spe_comment(gen->f, -4, "ABS:");
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         const int bit31mask_reg = get_itemp(gen);
 
-         /* mask with bit 31 set, the rest cleared */  
-         spe_load_uint(gen->f, bit31mask_reg, (1 << 31));
+   /* mask with bit 31 set, the rest cleared */  
+   spe_load_uint(gen->f, bit31mask_reg, (1 << 31));
 
-         /* d = sign bit cleared in s1 */
-         spe_andc(gen->f, d_reg, s1_reg, bit31mask_reg);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+   }
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   /* d = sign bit cleared in s1 */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_andc(gen->f, d_reg[ch], s1_reg[ch], bit31mask_reg);
    }
-   return true;
+
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
+   }
+
+   free_itemps(gen);
+   return TRUE;
 }
 
 /**
@@ -775,16 +833,14 @@ emit_DP3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    /* t0 = t0 + t1 */
    spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         spe_move(gen->f, d_reg, t0_reg);
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      spe_move(gen->f, d_reg, t0_reg);
+      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
 /**
@@ -824,16 +880,14 @@ emit_DP4(struct codegen *gen, const struct tgsi_full_instruction *inst)
    /* t0 = t0 + t1 */
    spe_fa(gen->f, t0_reg, t0_reg, t1_reg);
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         spe_move(gen->f, d_reg, t0_reg);
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      spe_move(gen->f, d_reg, t0_reg);
+      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
 /**
@@ -867,16 +921,14 @@ emit_DPH(struct codegen *gen, const struct tgsi_full_instruction *inst)
    /* t = w1 + t */
    spe_fa(gen->f, tmp_reg, s2_reg, tmp_reg);
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         spe_move(gen->f, d_reg, tmp_reg);
-         store_dest_reg(gen, tmp_reg, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      spe_move(gen->f, d_reg, tmp_reg);
+      store_dest_reg(gen, tmp_reg, ch, &inst->FullDstRegisters[0]);
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
 /**
@@ -911,17 +963,15 @@ emit_NRM3(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_frsqest(gen->f, t1_reg, t0_reg);
    spe_fi(gen->f, t1_reg, t0_reg, t1_reg);
 
-   for (ch = 0; ch < 3; ch++) {  /* NOTE: omit W channel */
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         /* dst = src[ch] * t1 */
-         spe_fm(gen->f, d_reg, src_reg[ch], t1_reg);
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      /* dst = src[ch] * t1 */
+      spe_fm(gen->f, d_reg, src_reg[ch], t1_reg);
+      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
 
@@ -978,201 +1028,101 @@ emit_XPD(struct codegen *gen, const struct tgsi_full_instruction *inst)
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
+
 /**
- * Emit set-if-greater-than.
+ * Emit inequality instruction.
  * Note that the SPE fcgt instruction produces 0x0 and 0xffffffff as
  * the result but OpenGL/TGSI needs 0.0 and 1.0 results.
  * We can easily convert 0x0/0xffffffff to 0.0/1.0 with a bitwise AND.
  */
 static boolean
-emit_SGT(struct codegen *gen, const struct tgsi_full_instruction *inst)
+emit_inequality(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, s1_reg[4], s2_reg[4], d_reg[4], one_reg;
+   bool complement = FALSE;
 
-   spe_comment(gen->f, -4, "SGT:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 > s2) */
-         spe_fcgt(gen->f, d_reg, s1_reg, s2_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = d & one_reg */
-         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_SGT:
+      spe_comment(gen->f, -4, "SGT:");
+      break;
+   case TGSI_OPCODE_SLT:
+      spe_comment(gen->f, -4, "SLT:");
+      break;
+   case TGSI_OPCODE_SGE:
+      spe_comment(gen->f, -4, "SGE:");
+      complement = TRUE;
+      break;
+   case TGSI_OPCODE_SLE:
+      spe_comment(gen->f, -4, "SLE:");
+      complement = TRUE;
+      break;
+   case TGSI_OPCODE_SEQ:
+      spe_comment(gen->f, -4, "SEQ:");
+      break;
+   case TGSI_OPCODE_SNE:
+      spe_comment(gen->f, -4, "SNE:");
+      complement = TRUE;
+      break;
+   default:
+      ;
    }
 
-   return true;
-}
+   one_reg = get_const_one_reg(gen);
 
-/**
- * Emit set-if_less-then.  See emit_SGT for comments.
- */
-static boolean
-emit_SLT(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-
-   spe_comment(gen->f, -4, "SLT:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 < s2) */
-         spe_fcgt(gen->f, d_reg, s2_reg, s1_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = d & one_reg */
-         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s2_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
    }
 
-   return true;
-}
-
-/**
- * Emit set-if_greater-then-or-equal.  See emit_SGT for comments.
- */
-static boolean
-emit_SGE(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-
-   spe_comment(gen->f, -4, "SGE:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 >= s2) */
-         spe_fcgt(gen->f, d_reg, s2_reg, s1_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = ~d & one_reg */
-         spe_andc(gen->f, d_reg, get_const_one_reg(gen), d_reg);
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      switch (inst->Instruction.Opcode) {
+      case TGSI_OPCODE_SGT:
+         spe_fcgt(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      case TGSI_OPCODE_SLT:
+         spe_fcgt(gen->f, d_reg[ch], s2_reg[ch], s1_reg[ch]);
+         break;
+      case TGSI_OPCODE_SGE:
+         spe_fcgt(gen->f, d_reg[ch], s2_reg[ch], s1_reg[ch]);
+         break;
+      case TGSI_OPCODE_SLE:
+         spe_fcgt(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      case TGSI_OPCODE_SEQ:
+         spe_fceq(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      case TGSI_OPCODE_SNE:
+         spe_fceq(gen->f, d_reg[ch], s1_reg[ch], s2_reg[ch]);
+         break;
+      default:
+         assert(0);
       }
    }
 
-   return true;
-}
-
-/**
- * Emit set-if_less-then-or-equal.  See emit_SGT for comments.
- */
-static boolean
-emit_SLE(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-
-   spe_comment(gen->f, -4, "SLE:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 <= s2) */
-         spe_fcgt(gen->f, d_reg, s1_reg, s2_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = ~d & one_reg */
-         spe_andc(gen->f, d_reg, get_const_one_reg(gen), d_reg);
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   /* convert d from 0x0/0xffffffff to 0.0/1.0 */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      /* d = d & one_reg */
+      if (complement)
+         spe_andc(gen->f, d_reg[ch], one_reg, d_reg[ch]);
+      else
+         spe_and(gen->f, d_reg[ch], one_reg, d_reg[ch]);
    }
 
-   return true;
-}
-
-/**
- * Emit set-if_equal.  See emit_SGT for comments.
- */
-static boolean
-emit_SEQ(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-
-   spe_comment(gen->f, -4, "SEQ:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 == s2) */
-         spe_fceq(gen->f, d_reg, s1_reg, s2_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = d & one_reg */
-         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
 
-   return true;
+   free_itemps(gen);
+   return TRUE;
 }
 
-/**
- * Emit set-if_not_equal.  See emit_SGT for comments.
- */
-static boolean
-emit_SNE(struct codegen *gen, const struct tgsi_full_instruction *inst)
-{
-   int ch;
-
-   spe_comment(gen->f, -4, "SNE:");
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-
-         /* d = (s1 != s2) */
-         spe_fceq(gen->f, d_reg, s1_reg, s2_reg);
-         spe_nor(gen->f, d_reg, d_reg, d_reg);
-
-         /* convert d from 0x0/0xffffffff to 0.0/1.0 */
-         /* d = d & one_reg */
-         spe_and(gen->f, d_reg, d_reg, get_const_one_reg(gen));
-
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
-   }
-
-   return true;
-}
 
 /**
- * Emit compare.  See emit_SGT for comments.
+ * Emit compare.
  */
 static boolean
 emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
@@ -1181,26 +1131,24 @@ emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    spe_comment(gen->f, -4, "CMP:");
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         int s3_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         int zero_reg = get_itemp(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      int s2_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      int s3_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[2]);
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      int zero_reg = get_itemp(gen);
    
-         spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+      spe_zero(gen->f, zero_reg);
 
-         /* d = (s1 < 0) ? s2 : s3 */
-         spe_fcgt(gen->f, d_reg, zero_reg, s1_reg);
-         spe_selb(gen->f, d_reg, s3_reg, s2_reg, d_reg);
+      /* d = (s1 < 0) ? s2 : s3 */
+      spe_fcgt(gen->f, d_reg, zero_reg, s1_reg);
+      spe_selb(gen->f, d_reg, s3_reg, s2_reg, d_reg);
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      free_itemps(gen);
    }
 
-   return true;
+   return TRUE;
 }
 
 /**
@@ -1211,29 +1159,34 @@ emit_CMP(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_TRUNC(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, s1_reg[4], d_reg[4];
 
    spe_comment(gen->f, -4, "TRUNC:");
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+   }
 
-         /* Convert float to int */
-         spe_cflts(gen->f, d_reg, s1_reg, 0);
+   /* Convert float to int */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_cflts(gen->f, d_reg[ch], s1_reg[ch], 0);
+   }
 
-         /* Convert int to float */
-         spe_csflt(gen->f, d_reg, d_reg, 0);
+   /* Convert int to float */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_csflt(gen->f, d_reg[ch], d_reg[ch], 0);
+   }
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
 
-   return true;
+   free_itemps(gen);
+   return TRUE;
 }
 
+
 /**
  * Emit floor.  
  * If negative int subtract one
@@ -1243,77 +1196,103 @@ emit_TRUNC(struct codegen *gen, const struct tgsi_full_instruction *inst)
 static boolean
 emit_FLR(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, s1_reg[4], d_reg[4], tmp_reg[4], zero_reg, one_reg;
 
    spe_comment(gen->f, -4, "FLR:");
 
-   int zero_reg = get_itemp(gen);
-   spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+   zero_reg = get_itemp(gen);
+   spe_zero(gen->f, zero_reg);
+   one_reg = get_const_one_reg(gen);
    
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         int tmp_reg = get_itemp(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      tmp_reg[ch] = get_itemp(gen);
+   }
 
-         /* If negative, subtract 1.0 */
-         spe_fcgt(gen->f, tmp_reg, zero_reg, s1_reg);
-         spe_selb(gen->f, tmp_reg, zero_reg, get_const_one_reg(gen), tmp_reg);
-         spe_fs(gen->f, tmp_reg, s1_reg, tmp_reg);
+   /* If negative, subtract 1.0 */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fcgt(gen->f, tmp_reg[ch], zero_reg, s1_reg[ch]);
+   }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_selb(gen->f, tmp_reg[ch], zero_reg, one_reg, tmp_reg[ch]);
+   }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fs(gen->f, tmp_reg[ch], s1_reg[ch], tmp_reg[ch]);
+   }
 
-         /* Convert float to int */
-         spe_cflts(gen->f, tmp_reg, tmp_reg, 0);
+   /* Convert float to int */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_cflts(gen->f, tmp_reg[ch], tmp_reg[ch], 0);
+   }
 
-         /* Convert int to float */
-         spe_csflt(gen->f, d_reg, tmp_reg, 0);
+   /* Convert int to float */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_csflt(gen->f, d_reg[ch], tmp_reg[ch], 0);
+   }
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
 
-   return true;
+   free_itemps(gen);
+   return TRUE;
 }
 
+
 /**
  * Compute frac = Input - FLR(Input)
  */
 static boolean
 emit_FRC(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   int ch;
+   int ch, s1_reg[4], d_reg[4], tmp_reg[4], zero_reg, one_reg;
 
    spe_comment(gen->f, -4, "FRC:");
 
-   int zero_reg = get_itemp(gen);
-   spe_xor(gen->f, zero_reg, zero_reg, zero_reg);
+   zero_reg = get_itemp(gen);
+   spe_zero(gen->f, zero_reg);
+   one_reg = get_const_one_reg(gen);
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s1_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         int tmp_reg = get_itemp(gen);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      tmp_reg[ch] = get_itemp(gen);
+   }
 
-         /* If negative, subtract 1.0 */
-         spe_fcgt(gen->f, tmp_reg, zero_reg, s1_reg);
-         spe_selb(gen->f, tmp_reg, zero_reg, get_const_one_reg(gen), tmp_reg);
-         spe_fs(gen->f, tmp_reg, s1_reg, tmp_reg);
+   /* If negative, subtract 1.0 */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fcgt(gen->f, tmp_reg[ch], zero_reg, s1_reg[ch]);
+   }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_selb(gen->f, tmp_reg[ch], zero_reg, one_reg, tmp_reg[ch]);
+   }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fs(gen->f, tmp_reg[ch], s1_reg[ch], tmp_reg[ch]);
+   }
 
-         /* Convert float to int */
-         spe_cflts(gen->f, tmp_reg, tmp_reg, 0);
+   /* Convert float to int */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_cflts(gen->f, tmp_reg[ch], tmp_reg[ch], 0);
+   }
 
-         /* Convert int to float */
-         spe_csflt(gen->f, tmp_reg, tmp_reg, 0);
+   /* Convert int to float */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_csflt(gen->f, tmp_reg[ch], tmp_reg[ch], 0);
+   }
 
-         /* d = s1 - FLR(s1) */
-         spe_fs(gen->f, d_reg, s1_reg, tmp_reg);
+   /* d = s1 - FLR(s1) */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_fs(gen->f, d_reg[ch], s1_reg[ch], tmp_reg[ch]);
+   }
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   /* store result */
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
 
-   return true;
+   free_itemps(gen);
+   return TRUE;
 }
 
 
@@ -1379,73 +1358,71 @@ emit_function_call(struct codegen *gen,
       retval_reg = spe_allocate_available_register(gen->f);
    }
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int d_reg;
-         ubyte usedRegs[SPE_NUM_REGS];
-         uint i, numUsed;
-
-         if (!scalar) {
-            for (a = 0; a < num_args; a++) {
-               s_regs[a] = get_src_reg(gen, ch, &inst->FullSrcRegisters[a]);
-            }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int d_reg;
+      ubyte usedRegs[SPE_NUM_REGS];
+      uint i, numUsed;
+
+      if (!scalar) {
+         for (a = 0; a < num_args; a++) {
+            s_regs[a] = get_src_reg(gen, ch, &inst->FullSrcRegisters[a]);
          }
+      }
 
-         d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
 
-         if (!scalar || !func_called) {
-            /* for a scalar function, we'll really only call the function once */
+      if (!scalar || !func_called) {
+         /* for a scalar function, we'll really only call the function once */
 
-            numUsed = spe_get_registers_used(gen->f, usedRegs);
-            assert(numUsed < gen->frame_size / 16 - 2);
+         numUsed = spe_get_registers_used(gen->f, usedRegs);
+         assert(numUsed < gen->frame_size / 16 - 2);
 
-            /* save registers to stack */
-            for (i = 0; i < numUsed; i++) {
-               uint reg = usedRegs[i];
-               int offset = 2 + i;
-               spe_stqd(gen->f, reg, SPE_REG_SP, 16 * offset);
-            }
+         /* save registers to stack */
+         for (i = 0; i < numUsed; i++) {
+            uint reg = usedRegs[i];
+            int offset = 2 + i;
+            spe_stqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+         }
 
-            /* setup function arguments */
-            for (a = 0; a < num_args; a++) {
-               spe_move(gen->f, 3 + a, s_regs[a]);
-            }
+         /* setup function arguments */
+         for (a = 0; a < num_args; a++) {
+            spe_move(gen->f, 3 + a, s_regs[a]);
+         }
 
-            /* branch to function, save return addr */
-            spe_brasl(gen->f, SPE_REG_RA, addr);
-
-            /* save function's return value */
-            if (scalar)
-               spe_move(gen->f, retval_reg, 3);
-            else
-               spe_move(gen->f, d_reg, 3);
-
-            /* restore registers from stack */
-            for (i = 0; i < numUsed; i++) {
-               uint reg = usedRegs[i];
-               if (reg != d_reg && reg != retval_reg) {
-                  int offset = 2 + i;
-                  spe_lqd(gen->f, reg, SPE_REG_SP, 16 * offset);
-               }
-            }
+         /* branch to function, save return addr */
+         spe_brasl(gen->f, SPE_REG_RA, addr);
 
-            func_called = TRUE;
-         }
+         /* save function's return value */
+         if (scalar)
+            spe_move(gen->f, retval_reg, 3);
+         else
+            spe_move(gen->f, d_reg, 3);
 
-         if (scalar) {
-            spe_move(gen->f, d_reg, retval_reg);
+         /* restore registers from stack */
+         for (i = 0; i < numUsed; i++) {
+            uint reg = usedRegs[i];
+            if (reg != d_reg && reg != retval_reg) {
+               int offset = 2 + i;
+               spe_lqd(gen->f, reg, SPE_REG_SP, 16 * offset);
+            }
          }
 
-         store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
+         func_called = TRUE;
+      }
+
+      if (scalar) {
+         spe_move(gen->f, d_reg, retval_reg);
       }
+
+      store_dest_reg(gen, d_reg, ch, &inst->FullDstRegisters[0]);
+      free_itemps(gen);
    }
 
    if (scalar) {
       spe_release_register(gen->f, retval_reg);
    }
 
-   return true;
+   return TRUE;
 }
 
 
@@ -1525,11 +1502,9 @@ emit_TEX(struct codegen *gen, const struct tgsi_full_instruction *inst)
       }
    }
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         store_dest_reg(gen, d_regs[ch], ch, &inst->FullDstRegisters[0]);
-         free_itemps(gen);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_regs[ch], ch, &inst->FullDstRegisters[0]);
+      free_itemps(gen);
    }
 
    return TRUE;
@@ -1549,35 +1524,31 @@ emit_KIL(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
    /* zero = {0,0,0,0} */
    zero_reg = get_itemp(gen);
-   spe_load_uint(gen->f, zero_reg, 0);
+   spe_zero(gen->f, zero_reg);
 
    cmp_reg = get_itemp(gen);
 
    /* get src regs */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s_regs[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
    }
 
    /* test if any src regs are < 0 */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         if (kil_reg >= 0) {
-            /* cmp = 0 > src ? : ~0 : 0 */
-            spe_fcgt(gen->f, cmp_reg, zero_reg, s_regs[ch]);
-            /* kil = kil | cmp */
-            spe_or(gen->f, kil_reg, kil_reg, cmp_reg);
-         }
-         else {
-            kil_reg = get_itemp(gen);
-            /* kil = 0 > src ? : ~0 : 0 */
-            spe_fcgt(gen->f, kil_reg, zero_reg, s_regs[ch]);
-         }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      if (kil_reg >= 0) {
+         /* cmp = 0 > src ? : ~0 : 0 */
+         spe_fcgt(gen->f, cmp_reg, zero_reg, s_regs[ch]);
+         /* kil = kil | cmp */
+         spe_or(gen->f, kil_reg, kil_reg, cmp_reg);
+      }
+      else {
+         kil_reg = get_itemp(gen);
+         /* kil = 0 > src ? : ~0 : 0 */
+         spe_fcgt(gen->f, kil_reg, zero_reg, s_regs[ch]);
       }
    }
 
-   if (gen->if_nesting) {
+   if (gen->if_nesting || gen->loop_nesting) {
       /* may have been a conditional kil */
       spe_and(gen->f, kil_reg, kil_reg, gen->exec_mask_reg);
    }
@@ -1599,96 +1570,92 @@ emit_KIL(struct codegen *gen, const struct tgsi_full_instruction *inst)
 
 
 /**
- * Emit max.  See emit_SGT for comments.
+ * Emit min or max.
  */
 static boolean
-emit_MAX(struct codegen *gen, const struct tgsi_full_instruction *inst)
+emit_MIN_MAX(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    int ch, s0_reg[4], s1_reg[4], d_reg[4], tmp_reg[4];
 
    spe_comment(gen->f, -4, "MAX:");
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         tmp_reg[ch] = get_itemp(gen);         
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
+      d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+      tmp_reg[ch] = get_itemp(gen);         
    }
 
    /* d = (s0 > s1) ? s0 : s1 */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      if (inst->Instruction.Opcode == TGSI_OPCODE_MAX)
          spe_fcgt(gen->f, tmp_reg[ch], s0_reg[ch], s1_reg[ch]);
-      }
+      else
+         spe_fcgt(gen->f, tmp_reg[ch], s1_reg[ch], s0_reg[ch]);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         spe_selb(gen->f, d_reg[ch], s1_reg[ch], s0_reg[ch], tmp_reg[ch]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      spe_selb(gen->f, d_reg[ch], s1_reg[ch], s0_reg[ch], tmp_reg[ch]);
    }
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-      }
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
    }
 
    free_itemps(gen);
-   return true;
+   return TRUE;
 }
 
+
 /**
- * Emit max.  See emit_SGT for comments.
+ * Emit code to update the execution mask.
+ * This needs to be done whenever the execution status of a conditional
+ * or loop is changed.
  */
-static boolean
-emit_MIN(struct codegen *gen, const struct tgsi_full_instruction *inst)
+static void
+emit_update_exec_mask(struct codegen *gen)
 {
-   int ch, s0_reg[4], s1_reg[4], d_reg[4], tmp_reg[4];
+   const int exec_reg = get_exec_mask_reg(gen);
+   const int cond_reg = gen->cond_mask_reg;
+   const int loop_reg = gen->loop_mask_reg;
 
-   spe_comment(gen->f, -4, "MIN:");
+   spe_comment(gen->f, 0, "Update master execution mask");
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         s0_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         s1_reg[ch] = get_src_reg(gen, ch, &inst->FullSrcRegisters[1]);
-         d_reg[ch] = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
-         tmp_reg[ch] = get_itemp(gen);         
-      }
+   if (gen->if_nesting > 0 && gen->loop_nesting > 0) {
+      /* exec_mask = cond_mask & loop_mask */
+      assert(cond_reg > 0);
+      assert(loop_reg > 0);
+      spe_and(gen->f, exec_reg, cond_reg, loop_reg);
    }
-
-   /* d = (s1 > s0) ? s0 : s1 */
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         spe_fcgt(gen->f, tmp_reg[ch], s1_reg[ch], s0_reg[ch]);
-      }
+   else if (gen->if_nesting > 0) {
+      assert(cond_reg > 0);
+      spe_move(gen->f, exec_reg, cond_reg);
    }
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         spe_selb(gen->f, d_reg[ch], s1_reg[ch], s0_reg[ch], tmp_reg[ch]);
-      }
+   else if (gen->loop_nesting > 0) {
+      assert(loop_reg > 0);
+      spe_move(gen->f, exec_reg, loop_reg);
    }
-
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         store_dest_reg(gen, d_reg[ch], ch, &inst->FullDstRegisters[0]);
-      }
+   else {
+      spe_load_int(gen->f, exec_reg, ~0x0);
    }
-
-   free_itemps(gen);
-   return true;
 }
 
+
 static boolean
 emit_IF(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
    const int channel = 0;
-   const int exec_reg = get_exec_mask_reg(gen);
+   int cond_reg;
 
    spe_comment(gen->f, -4, "IF:");
 
-   /* update execution mask with the predicate register */
+   cond_reg = get_cond_mask_reg(gen);
+
+   /* XXX push cond exec mask */
+
+   spe_comment(gen->f,  0, "init conditional exec mask = ~0:");
+   spe_load_int(gen->f, cond_reg, ~0);
+
+   /* update conditional execution mask with the predicate register */
    int tmp_reg = get_itemp(gen);
    int s1_reg = get_src_reg(gen, channel, &inst->FullSrcRegisters[0]);
 
@@ -1696,44 +1663,126 @@ emit_IF(struct codegen *gen, const struct tgsi_full_instruction *inst)
    spe_ceqi(gen->f, tmp_reg, s1_reg, 0);
    /* tmp = !tmp */
    spe_complement(gen->f, tmp_reg, tmp_reg);
-   /* exec_mask = exec_mask & tmp */
-   spe_and(gen->f, exec_reg, exec_reg, tmp_reg);
+   /* cond_mask = cond_mask & tmp */
+   spe_and(gen->f, cond_reg, cond_reg, tmp_reg);
 
    gen->if_nesting++;
 
+   /* update the master execution mask */
+   emit_update_exec_mask(gen);
+
    free_itemps(gen);
 
-   return true;
+   return TRUE;
 }
 
 
 static boolean
 emit_ELSE(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
-   const int exec_reg = get_exec_mask_reg(gen);
+   const int cond_reg = get_cond_mask_reg(gen);
 
    spe_comment(gen->f, -4, "ELSE:");
 
-   /* exec_mask = !exec_mask */
-   spe_complement(gen->f, exec_reg, exec_reg);
+   spe_comment(gen->f, 0, "cond exec mask = !cond exec mask");
+   spe_complement(gen->f, cond_reg, cond_reg);
+   emit_update_exec_mask(gen);
 
-   return true;
+   return TRUE;
 }
 
 
 static boolean
 emit_ENDIF(struct codegen *gen, const struct tgsi_full_instruction *inst)
 {
+   spe_comment(gen->f, -4, "ENDIF:");
+
+   /* XXX todo: pop cond exec mask */
+
+   gen->if_nesting--;
+
+   emit_update_exec_mask(gen);
+
+   return TRUE;
+}
+
+
+static boolean
+emit_BGNLOOP(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   int exec_reg, loop_reg;
+
+   spe_comment(gen->f, -4, "BGNLOOP:");
+
+   exec_reg = get_exec_mask_reg(gen);
+   loop_reg = get_loop_mask_reg(gen);
+
+   /* XXX push loop_exec mask */
+
+   spe_comment(gen->f,  0*-4, "initialize loop exec mask = ~0");
+   spe_load_int(gen->f, loop_reg, ~0x0);
+
+   gen->loop_nesting++;
+   gen->loop_start = spe_code_size(gen->f);  /* in bytes */
+
+   return TRUE;
+}
+
+
+static boolean
+emit_ENDLOOP(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   const int loop_reg = get_loop_mask_reg(gen);
+   const int tmp_reg = get_itemp(gen);
+   int offset;
+
+   spe_comment(gen->f, -4, "ENDLOOP:");
+
+   /* tmp_reg = exec[0] | exec[1] | exec[2] | exec[3] */
+   spe_orx(gen->f, tmp_reg, loop_reg);
+
+   offset = gen->loop_start - spe_code_size(gen->f); /* in bytes */
+
+   /* branch back to top of loop if tmp_reg != 0 */
+   spe_brnz(gen->f, tmp_reg, offset / 4);
+
+   /* XXX pop loop_exec mask */
+
+   gen->loop_nesting--;
+
+   emit_update_exec_mask(gen);
+
+   return TRUE;
+}
+
+
+static boolean
+emit_BRK(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
    const int exec_reg = get_exec_mask_reg(gen);
+   const int loop_reg = get_loop_mask_reg(gen);
 
-   spe_comment(gen->f, -4, "ENDIF:");
+   spe_comment(gen->f, -4, "BREAK:");
 
-   /* XXX todo: pop execution mask */
+   assert(gen->loop_nesting > 0);
 
-   spe_load_int(gen->f, exec_reg, ~0x0);
+   spe_comment(gen->f, 0, "loop exec mask &= ~master exec mask");
+   spe_andc(gen->f, loop_reg, loop_reg, exec_reg);
 
-   gen->if_nesting--;
-   return true;
+   emit_update_exec_mask(gen);
+
+   return TRUE;
+}
+
+
+static boolean
+emit_CONT(struct codegen *gen, const struct tgsi_full_instruction *inst)
+{
+   spe_comment(gen->f, -4, "CONT:");
+
+   assert(gen->loop_nesting > 0);
+
+   return TRUE;
 }
 
 
@@ -1745,28 +1794,26 @@ emit_DDX_DDY(struct codegen *gen, const struct tgsi_full_instruction *inst,
 
    spe_comment(gen->f, -4, ddx ? "DDX:" : "DDY:");
 
-   for (ch = 0; ch < 4; ch++) {
-      if (inst->FullDstRegisters[0].DstRegister.WriteMask & (1 << ch)) {
-         int s_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
-         int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
+   FOR_EACH_ENABLED_CHANNEL(inst, ch) {
+      int s_reg = get_src_reg(gen, ch, &inst->FullSrcRegisters[0]);
+      int d_reg = get_dst_reg(gen, ch, &inst->FullDstRegisters[0]);
 
-         int t1_reg = get_itemp(gen);
-         int t2_reg = get_itemp(gen);
+      int t1_reg = get_itemp(gen);
+      int t2_reg = get_itemp(gen);
 
-         spe_splat_word(gen->f, t1_reg, s_reg, 0); /* upper-left pixel */
-         if (ddx) {
-            spe_splat_word(gen->f, t2_reg, s_reg, 1); /* upper-right pixel */
-         }
-         else {
-            spe_splat_word(gen->f, t2_reg, s_reg, 2); /* lower-left pixel */
-         }
-         spe_fs(gen->f, d_reg, t2_reg, t1_reg);
-
-         free_itemps(gen);
+      spe_splat_word(gen->f, t1_reg, s_reg, 0); /* upper-left pixel */
+      if (ddx) {
+         spe_splat_word(gen->f, t2_reg, s_reg, 1); /* upper-right pixel */
+      }
+      else {
+         spe_splat_word(gen->f, t2_reg, s_reg, 2); /* lower-left pixel */
       }
+      spe_fs(gen->f, d_reg, t2_reg, t1_reg);
+
+      free_itemps(gen);
    }
 
-   return true;
+   return TRUE;
 }
 
 
@@ -1784,7 +1831,7 @@ emit_END(struct codegen *gen)
 {
    spe_comment(gen->f, -4, "END:");
    emit_epilogue(gen);
-   return true;
+   return TRUE;
 }
 
 
@@ -1796,15 +1843,15 @@ emit_instruction(struct codegen *gen,
                  const struct tgsi_full_instruction *inst)
 {
    switch (inst->Instruction.Opcode) {
+   case TGSI_OPCODE_ARL:
+      return emit_ARL(gen, inst);
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:
       return emit_MOV(gen, inst);
-   case TGSI_OPCODE_MUL:
-      return emit_MUL(gen, inst);
    case TGSI_OPCODE_ADD:
-      return emit_ADD(gen, inst);
    case TGSI_OPCODE_SUB:
-      return emit_SUB(gen, inst);
+   case TGSI_OPCODE_MUL:
+      return emit_binop(gen, inst);
    case TGSI_OPCODE_MAD:
       return emit_MAD(gen, inst);
    case TGSI_OPCODE_LERP:
@@ -1820,29 +1867,22 @@ emit_instruction(struct codegen *gen,
    case TGSI_OPCODE_XPD:
       return emit_XPD(gen, inst);
    case TGSI_OPCODE_RCP:
-      return emit_RCP(gen, inst);
    case TGSI_OPCODE_RSQ:
-      return emit_RSQ(gen, inst);
+      return emit_RCP_RSQ(gen, inst);
    case TGSI_OPCODE_ABS:
       return emit_ABS(gen, inst);
    case TGSI_OPCODE_SGT:
-      return emit_SGT(gen, inst);
    case TGSI_OPCODE_SLT:
-      return emit_SLT(gen, inst);
    case TGSI_OPCODE_SGE:
-      return emit_SGE(gen, inst);
    case TGSI_OPCODE_SLE:
-      return emit_SLE(gen, inst);
    case TGSI_OPCODE_SEQ:
-      return emit_SEQ(gen, inst);
    case TGSI_OPCODE_SNE:
-      return emit_SNE(gen, inst);
+      return emit_inequality(gen, inst);
    case TGSI_OPCODE_CMP:
       return emit_CMP(gen, inst);
-   case TGSI_OPCODE_MAX:
-      return emit_MAX(gen, inst);
    case TGSI_OPCODE_MIN:
-      return emit_MIN(gen, inst);
+   case TGSI_OPCODE_MAX:
+      return emit_MIN_MAX(gen, inst);
    case TGSI_OPCODE_TRUNC:
       return emit_TRUNC(gen, inst);
    case TGSI_OPCODE_FLR:
@@ -1882,20 +1922,29 @@ emit_instruction(struct codegen *gen,
    case TGSI_OPCODE_ENDIF:
       return emit_ENDIF(gen, inst);
 
+   case TGSI_OPCODE_BGNLOOP2:
+      return emit_BGNLOOP(gen, inst);
+   case TGSI_OPCODE_ENDLOOP2:
+      return emit_ENDLOOP(gen, inst);
+   case TGSI_OPCODE_BRK:
+      return emit_BRK(gen, inst);
+   case TGSI_OPCODE_CONT:
+      return emit_CONT(gen, inst);
+
    case TGSI_OPCODE_DDX:
-      return emit_DDX_DDY(gen, inst, true);
+      return emit_DDX_DDY(gen, inst, TRUE);
    case TGSI_OPCODE_DDY:
-      return emit_DDX_DDY(gen, inst, false);
+      return emit_DDX_DDY(gen, inst, FALSE);
 
    /* XXX lots more cases to do... */
 
    default:
       fprintf(stderr, "Cell: unimplemented TGSI instruction %d!\n",
               inst->Instruction.Opcode);
-      return false;
+      return FALSE;
    }
 
-   return true;
+   return TRUE;
 }
 
 
@@ -1923,10 +1972,14 @@ emit_immediate(struct codegen *gen, const struct tgsi_full_immediate *immed)
          gen->imm_regs[gen->num_imm][ch] = gen->imm_regs[gen->num_imm][ch - 1];
       }
       else {
+         char str[100];
          int reg = spe_allocate_available_register(gen->f);
 
          if (reg < 0)
-            return false;
+            return FALSE;
+
+         sprintf(str, "init $%d = %f", reg, val);
+         spe_comment(gen->f, 0, str);
 
          /* update immediate map */
          gen->imm_regs[gen->num_imm][ch] = reg;
@@ -1938,7 +1991,7 @@ emit_immediate(struct codegen *gen, const struct tgsi_full_immediate *immed)
 
    gen->num_imm++;
 
-   return true;
+   return TRUE;
 }
 
 
@@ -1963,7 +2016,7 @@ emit_declaration(struct cell_context *cell,
          for (ch = 0; ch < 4; ch++) {
             gen->temp_regs[i][ch] = spe_allocate_available_register(gen->f);
             if (gen->temp_regs[i][ch] < 0)
-               return false; /* out of regs */
+               return FALSE; /* out of regs */
          }
 
          /* XXX if we run out of SPE registers, we need to spill
@@ -1983,7 +2036,7 @@ emit_declaration(struct cell_context *cell,
       ; /* ignore */
    }
 
-   return true;
+   return TRUE;
 }
 
 
@@ -2019,7 +2072,7 @@ cell_gen_fragment_program(struct cell_context *cell,
    spe_allocate_register(f, gen.constants_reg);
 
    if (cell->debug_flags & CELL_DEBUG_ASM) {
-      spe_print_code(f, true);
+      spe_print_code(f, TRUE);
       spe_indent(f, 8);
       printf("Begin %s\n", __FUNCTION__);
       tgsi_dump(tokens, 0);
@@ -2035,17 +2088,17 @@ cell_gen_fragment_program(struct cell_context *cell,
       switch (parse.FullToken.Token.Type) {
       case TGSI_TOKEN_TYPE_IMMEDIATE:
          if (!emit_immediate(&gen, &parse.FullToken.FullImmediate))
-            gen.error = true;
+            gen.error = TRUE;
          break;
 
       case TGSI_TOKEN_TYPE_DECLARATION:
          if (!emit_declaration(cell, &gen, &parse.FullToken.FullDeclaration))
-            gen.error = true;
+            gen.error = TRUE;
          break;
 
       case TGSI_TOKEN_TYPE_INSTRUCTION:
          if (!emit_instruction(&gen, &parse.FullToken.FullInstruction))
-            gen.error = true;
+            gen.error = TRUE;
          break;
 
       default:
diff --git a/src/gallium/drivers/cell/ppu/cell_screen.c b/src/gallium/drivers/cell/ppu/cell_screen.c
index d2235579507..6fc2257e2a3 100644
--- a/src/gallium/drivers/cell/ppu/cell_screen.c
+++ b/src/gallium/drivers/cell/ppu/cell_screen.c
@@ -81,8 +81,12 @@ cell_get_param(struct pipe_screen *screen, int param)
       return 8;  /* max 128x128x128 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
       return CELL_MAX_TEXTURE_LEVELS;
+   case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
+      return 1; /* XXX not really true */
+   case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
+      return 0; /* XXX to do */
    default:
-      return 10;
+      return 0;
    }
 }
 
@@ -108,7 +112,7 @@ cell_get_paramf(struct pipe_screen *screen, int param)
       return 16.0; /* arbitrary */
 
    default:
-      return 10;
+      return 0;
    }
 }
 
diff --git a/src/gallium/drivers/cell/spu/spu_shuffle.h b/src/gallium/drivers/cell/spu/spu_shuffle.h
new file mode 100644
index 00000000000..7cbdb814d28
--- /dev/null
+++ b/src/gallium/drivers/cell/spu/spu_shuffle.h
@@ -0,0 +1,186 @@
+#ifndef SPU_SHUFFLE_H
+#define SPU_SHUFFLE_H
+
+/*
+ * Generate shuffle patterns with minimal fuss.
+ *
+ * Based on ideas from 
+ * http://www.insomniacgames.com/tech/articles/0408/files/shuffles.pdf
+ *
+ * A-P indicates 0-15th position in first vector
+ * a-p indicates 0-15th position in second vector
+ *
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * |00|01|02|03|04|05|06|07|08|09|0a|0b|0c|0d|0e|0f|
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * |          A|          B|          C|          D|
+ * +-----+-----+-----+-----+-----+-----+-----+-----+
+ * |    A|    B|    C|    D|    E|    F|    G|    H|
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ * | A| B| C| D| E| F| G| H| I| J| K| L| M| N| O| P|
+ * +--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+--+
+ *
+ * x or X indicates 0xff
+ * 8 indicates 0x80
+ * 0 indicates 0x00
+ *
+ * The macros SHUFFLE4() SHUFFLE8() and SHUFFLE16() provide a const vector 
+ * unsigned char literal suitable for use with spu_shuffle().
+ *
+ * The macros SHUFB4() SHUFB8() and SHUFB16() provide a const qword vector 
+ * literal suitable for use with si_shufb().
+ *
+ *
+ * For example :
+ * SHUFB4(A,A,A,A)
+ * expands to :
+ * ((const qword){0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3})
+ * 
+ * SHUFFLE8(A,B,a,b,C,c,8,8)
+ * expands to :
+ * ((const vector unsigned char){0x00,0x01,0x02,0x03,0x10,0x11,0x12,0x13,
+ *				 0x04,0x05,0x14,0x15,0xe0,0xe0,0xe0,0xe0})
+ *
+ */
+
+#include <spu_intrinsics.h>
+
+#define SHUFFLE_PATTERN_4_A__  0x00, 0x01, 0x02, 0x03
+#define SHUFFLE_PATTERN_4_B__  0x04, 0x05, 0x06, 0x07
+#define SHUFFLE_PATTERN_4_C__  0x08, 0x09, 0x0a, 0x0b
+#define SHUFFLE_PATTERN_4_D__  0x0c, 0x0d, 0x0e, 0x0f
+#define SHUFFLE_PATTERN_4_a__  0x10, 0x11, 0x12, 0x13
+#define SHUFFLE_PATTERN_4_b__  0x14, 0x15, 0x16, 0x17
+#define SHUFFLE_PATTERN_4_c__  0x18, 0x19, 0x1a, 0x1b
+#define SHUFFLE_PATTERN_4_d__  0x1c, 0x1d, 0x1e, 0x1f
+#define SHUFFLE_PATTERN_4_X__  0xc0, 0xc0, 0xc0, 0xc0
+#define SHUFFLE_PATTERN_4_x__  0xc0, 0xc0, 0xc0, 0xc0
+#define SHUFFLE_PATTERN_4_0__  0x80, 0x80, 0x80, 0x80
+#define SHUFFLE_PATTERN_4_8__  0xe0, 0xe0, 0xe0, 0xe0
+
+#define SHUFFLE_VECTOR_4__(A, B, C, D) \
+   SHUFFLE_PATTERN_4_##A##__, \
+   SHUFFLE_PATTERN_4_##B##__, \
+   SHUFFLE_PATTERN_4_##C##__, \
+   SHUFFLE_PATTERN_4_##D##__
+
+#define SHUFFLE4(A, B, C, D) \
+   ((const vector unsigned char){ \
+      SHUFFLE_VECTOR_4__(A, B, C, D) \
+   })
+
+#define SHUFB4(A, B, C, D) \
+   ((const qword){ \
+      SHUFFLE_VECTOR_4__(A, B, C, D) \
+   })
+
+
+#define SHUFFLE_PATTERN_8_A__  0x00, 0x01
+#define SHUFFLE_PATTERN_8_B__  0x02, 0x03
+#define SHUFFLE_PATTERN_8_C__  0x04, 0x05
+#define SHUFFLE_PATTERN_8_D__  0x06, 0x07
+#define SHUFFLE_PATTERN_8_E__  0x08, 0x09
+#define SHUFFLE_PATTERN_8_F__  0x0a, 0x0b
+#define SHUFFLE_PATTERN_8_G__  0x0c, 0x0d
+#define SHUFFLE_PATTERN_8_H__  0x0e, 0x0f
+#define SHUFFLE_PATTERN_8_a__  0x10, 0x11
+#define SHUFFLE_PATTERN_8_b__  0x12, 0x13
+#define SHUFFLE_PATTERN_8_c__  0x14, 0x15
+#define SHUFFLE_PATTERN_8_d__  0x16, 0x17
+#define SHUFFLE_PATTERN_8_e__  0x18, 0x19
+#define SHUFFLE_PATTERN_8_f__  0x1a, 0x1b
+#define SHUFFLE_PATTERN_8_g__  0x1c, 0x1d
+#define SHUFFLE_PATTERN_8_h__  0x1e, 0x1f
+#define SHUFFLE_PATTERN_8_X__  0xc0, 0xc0
+#define SHUFFLE_PATTERN_8_x__  0xc0, 0xc0
+#define SHUFFLE_PATTERN_8_0__  0x80, 0x80
+#define SHUFFLE_PATTERN_8_8__  0xe0, 0xe0
+
+
+#define SHUFFLE_VECTOR_8__(A, B, C, D, E, F, G, H) \
+   SHUFFLE_PATTERN_8_##A##__, \
+   SHUFFLE_PATTERN_8_##B##__, \
+   SHUFFLE_PATTERN_8_##C##__, \
+   SHUFFLE_PATTERN_8_##D##__, \
+   SHUFFLE_PATTERN_8_##E##__, \
+   SHUFFLE_PATTERN_8_##F##__, \
+   SHUFFLE_PATTERN_8_##G##__, \
+   SHUFFLE_PATTERN_8_##H##__
+
+#define SHUFFLE8(A, B, C, D, E, F, G, H) \
+   ((const vector unsigned char){ \
+      SHUFFLE_VECTOR_8__(A, B, C, D, E, F, G, H) \
+   })
+
+#define SHUFB8(A, B, C, D, E, F, G, H) \
+   ((const qword){ \
+      SHUFFLE_VECTOR_8__(A, B, C, D, E, F, G, H) \
+   })
+
+
+#define SHUFFLE_PATTERN_16_A__  0x00
+#define SHUFFLE_PATTERN_16_B__  0x01
+#define SHUFFLE_PATTERN_16_C__  0x02
+#define SHUFFLE_PATTERN_16_D__  0x03
+#define SHUFFLE_PATTERN_16_E__  0x04
+#define SHUFFLE_PATTERN_16_F__  0x05
+#define SHUFFLE_PATTERN_16_G__  0x06
+#define SHUFFLE_PATTERN_16_H__  0x07
+#define SHUFFLE_PATTERN_16_I__  0x08
+#define SHUFFLE_PATTERN_16_J__  0x09
+#define SHUFFLE_PATTERN_16_K__  0x0a
+#define SHUFFLE_PATTERN_16_L__  0x0b
+#define SHUFFLE_PATTERN_16_M__  0x0c
+#define SHUFFLE_PATTERN_16_N__  0x0d
+#define SHUFFLE_PATTERN_16_O__  0x0e
+#define SHUFFLE_PATTERN_16_P__  0x0f
+#define SHUFFLE_PATTERN_16_a__  0x10
+#define SHUFFLE_PATTERN_16_b__  0x11
+#define SHUFFLE_PATTERN_16_c__  0x12
+#define SHUFFLE_PATTERN_16_d__  0x13
+#define SHUFFLE_PATTERN_16_e__  0x14
+#define SHUFFLE_PATTERN_16_f__  0x15
+#define SHUFFLE_PATTERN_16_g__  0x16
+#define SHUFFLE_PATTERN_16_h__  0x17
+#define SHUFFLE_PATTERN_16_i__  0x18
+#define SHUFFLE_PATTERN_16_j__  0x19
+#define SHUFFLE_PATTERN_16_k__  0x1a
+#define SHUFFLE_PATTERN_16_l__  0x1b
+#define SHUFFLE_PATTERN_16_m__  0x1c
+#define SHUFFLE_PATTERN_16_n__  0x1d
+#define SHUFFLE_PATTERN_16_o__  0x1e
+#define SHUFFLE_PATTERN_16_p__  0x1f
+#define SHUFFLE_PATTERN_16_X__  0xc0
+#define SHUFFLE_PATTERN_16_x__  0xc0
+#define SHUFFLE_PATTERN_16_0__  0x80
+#define SHUFFLE_PATTERN_16_8__  0xe0
+
+#define SHUFFLE_VECTOR_16__(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P) \
+   SHUFFLE_PATTERN_16_##A##__, \
+   SHUFFLE_PATTERN_16_##B##__, \
+   SHUFFLE_PATTERN_16_##C##__, \
+   SHUFFLE_PATTERN_16_##D##__, \
+   SHUFFLE_PATTERN_16_##E##__, \
+   SHUFFLE_PATTERN_16_##F##__, \
+   SHUFFLE_PATTERN_16_##G##__, \
+   SHUFFLE_PATTERN_16_##H##__, \
+   SHUFFLE_PATTERN_16_##I##__, \
+   SHUFFLE_PATTERN_16_##J##__, \
+   SHUFFLE_PATTERN_16_##K##__, \
+   SHUFFLE_PATTERN_16_##L##__, \
+   SHUFFLE_PATTERN_16_##M##__, \
+   SHUFFLE_PATTERN_16_##N##__, \
+   SHUFFLE_PATTERN_16_##O##__, \
+   SHUFFLE_PATTERN_16_##P
+
+#define SHUFFLE16(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P) \
+   ((const vector unsigned char){ \
+      SHUFFLE_VECTOR_16__(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P) \
+   })
+
+#define SHUFB16(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P) \
+   ((const qword){ \
+      SHUFFLE_VECTOR_16__(A, B, C, D, E, F, G, H, I, J, K, L, M, N, O, P) \
+   })
+
+#endif
diff --git a/src/gallium/drivers/cell/spu/spu_tri.c b/src/gallium/drivers/cell/spu/spu_tri.c
index 22e51a86ae5..322be1252e9 100644
--- a/src/gallium/drivers/cell/spu/spu_tri.c
+++ b/src/gallium/drivers/cell/spu/spu_tri.c
@@ -35,6 +35,7 @@
 #include "util/u_math.h"
 #include "spu_colorpack.h"
 #include "spu_main.h"
+#include "spu_shuffle.h"
 #include "spu_texture.h"
 #include "spu_tile.h"
 #include "spu_tri.h"
@@ -76,8 +77,13 @@ struct vertex_header {
  * Triangle edge info
  */
 struct edge {
-   float dx;		/**< X(v1) - X(v0), used only during setup */
-   float dy;		/**< Y(v1) - Y(v0), used only during setup */
+   union {
+      struct {
+         float dx;	/**< X(v1) - X(v0), used only during setup */
+         float dy;	/**< Y(v1) - Y(v0), used only during setup */
+      };
+      vec_float4 ds;    /**< vector accessor for dx and dy */
+   };
    float dxdy;		/**< dx/dy */
    float sx, sy;	/**< first sample point coord */
    int lines;		/**< number of lines on this edge */
@@ -102,10 +108,15 @@ struct setup_stage {
     * turn.  Currently fixed at 4 floats, but should change in time.
     * Codegen will help cope with this.
     */
-   const struct vertex_header *vmax;
-   const struct vertex_header *vmid;
-   const struct vertex_header *vmin;
-   const struct vertex_header *vprovoke;
+   union {
+      struct {
+         const struct vertex_header *vmin;
+         const struct vertex_header *vmid;
+         const struct vertex_header *vmax;
+         const struct vertex_header *vprovoke;
+      };
+      qword vertex_headers;
+   };
 
    struct edge ebot;
    struct edge etop;
@@ -122,8 +133,7 @@ struct setup_stage {
    struct interp_coef coef[PIPE_MAX_SHADER_INPUTS];
 
    struct {
-      int left[2];   /**< [0] = row0, [1] = row1 */
-      int right[2];
+      vec_int4 quad; /**< [0] = row0, [1] = row1; {left[0],left[1],right[0],right[1]} */
       int y;
       unsigned y_flags;
       unsigned mask;     /**< mask of MASK_BOTTOM/TOP_LEFT/RIGHT bits */
@@ -306,52 +316,35 @@ block(int x)
 
 
 /**
- * Compute mask which indicates which pixels in the 2x2 quad are actually inside
- * the triangle's bounds.
- * The mask is a uint4 vector and each element will be 0 or 0xffffffff.
- */
-static INLINE mask_t
-calculate_mask(int x)
-{
-   /* This is a little tricky.
-    * Use & instead of && to avoid branches.
-    * Use negation to convert true/false to ~0/0 values.
-    */
-   mask_t mask;
-   mask = spu_insert(-((x   >= setup.span.left[0]) & (x   < setup.span.right[0])), mask, 0);
-   mask = spu_insert(-((x+1 >= setup.span.left[0]) & (x+1 < setup.span.right[0])), mask, 1);
-   mask = spu_insert(-((x   >= setup.span.left[1]) & (x   < setup.span.right[1])), mask, 2);
-   mask = spu_insert(-((x+1 >= setup.span.left[1]) & (x+1 < setup.span.right[1])), mask, 3);
-   return mask;
-}
-
-
-/**
  * Render a horizontal span of quads
  */
 static void
 flush_spans(void)
 {
    int minleft, maxright;
-   int x;
+
+   const int l0 = spu_extract(setup.span.quad, 0);
+   const int l1 = spu_extract(setup.span.quad, 1);
+   const int r0 = spu_extract(setup.span.quad, 2);
+   const int r1 = spu_extract(setup.span.quad, 3);
 
    switch (setup.span.y_flags) {
    case 0x3:
       /* both odd and even lines written (both quad rows) */
-      minleft = MIN2(setup.span.left[0], setup.span.left[1]);
-      maxright = MAX2(setup.span.right[0], setup.span.right[1]);
+      minleft = MIN2(l0, l1);
+      maxright = MAX2(r0, r1);
       break;
 
    case 0x1:
       /* only even line written (quad top row) */
-      minleft = setup.span.left[0];
-      maxright = setup.span.right[0];
+      minleft = l0;
+      maxright = r0;
       break;
 
    case 0x2:
       /* only odd line written (quad bottom row) */
-      minleft = setup.span.left[1];
-      maxright = setup.span.right[1];
+      minleft = l1;
+      maxright = r1;
       break;
 
    default:
@@ -389,17 +382,42 @@ flush_spans(void)
       ASSERT(spu.cur_ztile_status != TILE_STATUS_DEFINED);
    }
 
-   /* XXX this loop could be moved into the above switch cases and
-    * calculate_mask() could be simplified a bit...
-    */
-   for (x = block(minleft); x <= block(maxright); x += 2) {
-      emit_quad( x, setup.span.y, calculate_mask( x ));
+   /* XXX this loop could be moved into the above switch cases... */
+   
+   /* Setup for mask calculation */
+   const vec_int4 quad_LlRr = setup.span.quad;
+   const vec_int4 quad_RrLl = spu_rlqwbyte(quad_LlRr, 8);
+   const vec_int4 quad_LLll = spu_shuffle(quad_LlRr, quad_LlRr, SHUFFLE4(A,A,B,B));
+   const vec_int4 quad_RRrr = spu_shuffle(quad_RrLl, quad_RrLl, SHUFFLE4(A,A,B,B));
+
+   const vec_int4 twos = spu_splats(2);
+
+   const int x = block(minleft);
+   vec_int4 xs = {x, x+1, x, x+1};
+
+   for (; spu_extract(xs, 0) <= block(maxright); xs += twos) {
+      /**
+       * Computes mask to indicate which pixels in the 2x2 quad are actually
+       * inside the triangle's bounds.
+       */
+      
+      /* Calculate ({x,x+1,x,x+1} >= {l[0],l[0],l[1],l[1]}) */
+      const mask_t gt_LLll_xs = spu_cmpgt(quad_LLll, xs);
+      const mask_t gte_xs_LLll = spu_nand(gt_LLll_xs, gt_LLll_xs); 
+      
+      /* Calculate ({r[0],r[0],r[1],r[1]} > {x,x+1,x,x+1}) */
+      const mask_t gt_RRrr_xs = spu_cmpgt(quad_RRrr, xs);
+
+      /* Combine results to create mask */
+      const mask_t mask = spu_and(gte_xs_LLll, gt_RRrr_xs);
+
+      emit_quad(spu_extract(xs, 0), setup.span.y, mask);
    }
 
    setup.span.y = 0;
    setup.span.y_flags = 0;
-   setup.span.right[0] = 0;
-   setup.span.right[1] = 0;
+   /* Zero right elements */
+   setup.span.quad = spu_shuffle(setup.span.quad, setup.span.quad, SHUFFLE4(A,B,0,0));
 }
 
 
@@ -444,55 +462,39 @@ setup_sort_vertices(const struct vertex_header *v0,
 
    /* determine bottom to top order of vertices */
    {
-      float y0 = spu_extract(v0->data[0], 1);
-      float y1 = spu_extract(v1->data[0], 1);
-      float y2 = spu_extract(v2->data[0], 1);
-      if (y0 <= y1) {
-	 if (y1 <= y2) {
-	    /* y0<=y1<=y2 */
-	    setup.vmin = v0;   
-	    setup.vmid = v1;   
-	    setup.vmax = v2;
-            sign = -1.0f;
-	 }
-	 else if (y2 <= y0) {
-	    /* y2<=y0<=y1 */
-	    setup.vmin = v2;   
-	    setup.vmid = v0;   
-	    setup.vmax = v1;   
-            sign = -1.0f;
-	 }
-	 else {
-	    /* y0<=y2<=y1 */
-	    setup.vmin = v0;   
-	    setup.vmid = v2;   
-	    setup.vmax = v1;  
-            sign = 1.0f;
-	 }
-      }
-      else {
-	 if (y0 <= y2) {
-	    /* y1<=y0<=y2 */
-	    setup.vmin = v1;   
-	    setup.vmid = v0;   
-	    setup.vmax = v2;  
-            sign = 1.0f;
-	 }
-	 else if (y2 <= y1) {
-	    /* y2<=y1<=y0 */
-	    setup.vmin = v2;   
-	    setup.vmid = v1;   
-	    setup.vmax = v0;  
-            sign = 1.0f;
-	 }
-	 else {
-	    /* y1<=y2<=y0 */
-	    setup.vmin = v1;   
-	    setup.vmid = v2;   
-	    setup.vmax = v0;
-            sign = -1.0f;
-	 }
-      }
+      /* A table of shuffle patterns for putting vertex_header pointers into
+         correct order.  Quite magical. */
+      const vec_uchar16 sort_order_patterns[] = {
+         SHUFFLE4(A,B,C,C),
+         SHUFFLE4(C,A,B,C),
+         SHUFFLE4(A,C,B,C),
+         SHUFFLE4(B,C,A,C),
+         SHUFFLE4(B,A,C,C),
+         SHUFFLE4(C,B,A,C) };
+
+      /* The vertex_header pointers, packed for easy shuffling later */
+      const vec_uint4 vs = {(unsigned)v0, (unsigned)v1, (unsigned)v2};
+
+      /* Collate y values into two vectors for comparison.
+         Using only one shuffle constant! ;) */
+      const vec_float4 y_02_ = spu_shuffle(v0->data[0], v2->data[0], SHUFFLE4(0,B,b,C));
+      const vec_float4 y_10_ = spu_shuffle(v1->data[0], v0->data[0], SHUFFLE4(0,B,b,C));
+      const vec_float4 y_012 = spu_shuffle(y_02_, v1->data[0], SHUFFLE4(0,B,b,C));
+      const vec_float4 y_120 = spu_shuffle(y_10_, v2->data[0], SHUFFLE4(0,B,b,C));
+
+      /* Perform comparison: {y0,y1,y2} > {y1,y2,y0} */
+      const vec_uint4 compare = spu_cmpgt(y_012, y_120);
+      /* Compress the result of the comparison into 4 bits */
+      const vec_uint4 gather = spu_gather(compare);
+      /* Subtract one to attain the index into the LUT.  Magical. */
+      const unsigned int index = spu_extract(gather, 0) - 1;
+
+      /* Load the appropriate pattern and construct the desired vector. */
+      setup.vertex_headers = (qword)spu_shuffle(vs, vs, sort_order_patterns[index]);
+
+      /* Using the result of the comparison, set sign.
+         Very magical. */
+      sign = ((si_to_uint(si_cntb((qword)gather)) == 2) ? 1.0f : -1.0f);
    }
 
    /* Check if triangle is completely outside the tile bounds */
@@ -509,12 +511,9 @@ setup_sort_vertices(const struct vertex_header *v0,
        spu_extract(setup.vmax->data[0], 0) > setup.cliprect_maxx)
       return FALSE;
 
-   setup.ebot.dx = spu_extract(setup.vmid->data[0], 0) - spu_extract(setup.vmin->data[0], 0);
-   setup.ebot.dy = spu_extract(setup.vmid->data[0], 1) - spu_extract(setup.vmin->data[0], 1);
-   setup.emaj.dx = spu_extract(setup.vmax->data[0], 0) - spu_extract(setup.vmin->data[0], 0);
-   setup.emaj.dy = spu_extract(setup.vmax->data[0], 1) - spu_extract(setup.vmin->data[0], 1);
-   setup.etop.dx = spu_extract(setup.vmax->data[0], 0) - spu_extract(setup.vmid->data[0], 0);
-   setup.etop.dy = spu_extract(setup.vmax->data[0], 1) - spu_extract(setup.vmid->data[0], 1);
+   setup.ebot.ds = spu_sub(setup.vmid->data[0], setup.vmin->data[0]);
+   setup.emaj.ds = spu_sub(setup.vmax->data[0], setup.vmin->data[0]);
+   setup.etop.ds = spu_sub(setup.vmax->data[0], setup.vmid->data[0]);
 
    /*
     * Compute triangle's area.  Use 1/area to compute partial
@@ -535,8 +534,6 @@ setup_sort_vertices(const struct vertex_header *v0,
    setup.facing = (area * sign > 0.0f)
       ^ (spu.rasterizer.front_winding == PIPE_WINDING_CW);
 
-   setup.vprovoke = v2;
-
    return TRUE;
 }
 
@@ -746,9 +743,11 @@ subtriangle(struct edge *eleft, struct edge *eright, unsigned lines)
             setup.span.y = block(_y);
          }
 
-         setup.span.left[_y&1] = left;
-         setup.span.right[_y&1] = right;
-         setup.span.y_flags |= 1<<(_y&1);
+         int offset = _y&1;
+         vec_int4 quad_LlRr = {left, left, right, right};
+         /* Store left and right in 0 or 1 row of quad based on offset */
+         setup.span.quad = spu_sel(quad_LlRr, setup.span.quad, spu_maskw(5<<offset));
+         setup.span.y_flags |= 1<<offset;
       }
    }
 
@@ -790,8 +789,8 @@ tri_draw(const float *v0, const float *v1, const float *v2,
 
    setup.span.y = 0;
    setup.span.y_flags = 0;
-   setup.span.right[0] = 0;
-   setup.span.right[1] = 0;
+   /* Zero right elements */
+   setup.span.quad = spu_shuffle(setup.span.quad, setup.span.quad, SHUFFLE4(A,B,0,0));
 
    if (setup.oneOverArea < 0.0) {
       /* emaj on left */
diff --git a/src/gallium/drivers/i915simple/i915_prim_vbuf.c b/src/gallium/drivers/i915simple/i915_prim_vbuf.c
index 4fda1ab64f5..a8e97e7c306 100644
--- a/src/gallium/drivers/i915simple/i915_prim_vbuf.c
+++ b/src/gallium/drivers/i915simple/i915_prim_vbuf.c
@@ -197,9 +197,7 @@ i915_vbuf_render_set_primitive( struct vbuf_render *render,
       i915_render->fallback = 0;
       return TRUE;
    default:
-      assert((int)"Error unkown primtive type" & 0);
-      /* Actually, can handle a lot more just fine...  Fixme.
-       */
+      /* FIXME: Actually, can handle a lot more just fine... */
       return FALSE;
    }
 }
diff --git a/src/gallium/drivers/nv30/nv30_query.c b/src/gallium/drivers/nv30/nv30_query.c
index d40d75f2640..2f974cf5c40 100644
--- a/src/gallium/drivers/nv30/nv30_query.c
+++ b/src/gallium/drivers/nv30/nv30_query.c
@@ -50,7 +50,7 @@ nv30_query_begin(struct pipe_context *pipe, struct pipe_query *pq)
 	 * the existing query to notify completion, but it could be better.
 	 */
 	if (q->object) {
-		uint64 tmp;
+		uint64_t tmp;
 		pipe->get_query_result(pipe, pq, 1, &tmp);
 	}
 
@@ -80,7 +80,7 @@ nv30_query_end(struct pipe_context *pipe, struct pipe_query *pq)
 
 static boolean
 nv30_query_result(struct pipe_context *pipe, struct pipe_query *pq,
-		  boolean wait, uint64 *result)
+		  boolean wait, uint64_t *result)
 {
 	struct nv30_context *nv30 = nv30_context(pipe);
 	struct nv30_query *q = nv30_query(pq);
diff --git a/src/gallium/drivers/nv30/nv30_state_emit.c b/src/gallium/drivers/nv30/nv30_state_emit.c
index 40fed621b24..9480695d6e5 100644
--- a/src/gallium/drivers/nv30/nv30_state_emit.c
+++ b/src/gallium/drivers/nv30/nv30_state_emit.c
@@ -49,7 +49,7 @@ nv30_state_emit(struct nv30_context *nv30)
 	struct nv30_state *state = &nv30->state;
 	struct nv30_screen *screen = nv30->screen;
 	unsigned i, samplers;
-	uint64 states;
+	uint64_t states;
 
 	if (nv30->pctx_id != screen->cur_pctx) {
 		for (i = 0; i < NV30_STATE_MAX; i++) {
diff --git a/src/gallium/drivers/nv40/nv40_query.c b/src/gallium/drivers/nv40/nv40_query.c
index 57f39cfab0c..9b9a43f49df 100644
--- a/src/gallium/drivers/nv40/nv40_query.c
+++ b/src/gallium/drivers/nv40/nv40_query.c
@@ -50,7 +50,7 @@ nv40_query_begin(struct pipe_context *pipe, struct pipe_query *pq)
 	 * the existing query to notify completion, but it could be better.
 	 */
 	if (q->object) {
-		uint64 tmp;
+		uint64_t tmp;
 		pipe->get_query_result(pipe, pq, 1, &tmp);
 	}
 
@@ -80,7 +80,7 @@ nv40_query_end(struct pipe_context *pipe, struct pipe_query *pq)
 
 static boolean
 nv40_query_result(struct pipe_context *pipe, struct pipe_query *pq,
-		  boolean wait, uint64 *result)
+		  boolean wait, uint64_t *result)
 {
 	struct nv40_context *nv40 = nv40_context(pipe);
 	struct nv40_query *q = nv40_query(pq);
diff --git a/src/gallium/drivers/nv40/nv40_state_emit.c b/src/gallium/drivers/nv40/nv40_state_emit.c
index ab88dc416e5..52ec4c044b4 100644
--- a/src/gallium/drivers/nv40/nv40_state_emit.c
+++ b/src/gallium/drivers/nv40/nv40_state_emit.c
@@ -65,7 +65,7 @@ nv40_state_emit(struct nv40_context *nv40)
 	struct nv40_state *state = &nv40->state;
 	struct nv40_screen *screen = nv40->screen;
 	unsigned i, samplers;
-	uint64 states;
+	uint64_t states;
 
 	if (nv40->pctx_id != screen->cur_pctx) {
 		for (i = 0; i < NV40_STATE_MAX; i++) {
diff --git a/src/gallium/drivers/nv50/nv50_context.h b/src/gallium/drivers/nv50/nv50_context.h
index 5d377f2d067..0958bba334a 100644
--- a/src/gallium/drivers/nv50/nv50_context.h
+++ b/src/gallium/drivers/nv50/nv50_context.h
@@ -70,6 +70,10 @@ struct nv50_rasterizer_stateobj {
 struct nv50_miptree {
 	struct pipe_texture base;
 	struct pipe_buffer *buffer;
+
+	int *image_offset;
+	int image_nr;
+	int total_size;
 };
 
 static INLINE struct nv50_miptree *
diff --git a/src/gallium/drivers/nv50/nv50_miptree.c b/src/gallium/drivers/nv50/nv50_miptree.c
index 28a8bdc0fab..24973712324 100644
--- a/src/gallium/drivers/nv50/nv50_miptree.c
+++ b/src/gallium/drivers/nv50/nv50_miptree.c
@@ -31,7 +31,8 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 {
 	struct pipe_winsys *ws = pscreen->winsys;
 	struct nv50_miptree *mt = CALLOC_STRUCT(nv50_miptree);
-	unsigned usage, pitch;
+	unsigned usage, width = pt->width[0], height = pt->height[0];
+	int i;
 
 	mt->base = *pt;
 	mt->base.refcount = 1;
@@ -47,11 +48,31 @@ nv50_miptree_create(struct pipe_screen *pscreen, const struct pipe_texture *pt)
 		break;
 	}
 
-	pitch = ((pt->width[0] + 63) & ~63) * pt->block.size;
-	/*XXX*/
-	pitch *= 2;
+	switch (pt->target) {
+	case PIPE_TEXTURE_3D:
+		mt->image_nr = pt->depth[0];
+		break;
+	case PIPE_TEXTURE_CUBE:
+		mt->image_nr = 6;
+		break;
+	default:
+		mt->image_nr = 1;
+		break;
+	}
+	mt->image_offset = CALLOC(mt->image_nr, sizeof(int));
 
-	mt->buffer = ws->buffer_create(ws, 256, usage, pitch * pt->height[0]);
+	for (i = 0; i < mt->image_nr; i++) {
+		int image_size;
+
+		image_size  = align(width, 8) * pt->block.size;
+		image_size  = align(image_size, 64);
+		image_size *= align(height, 8) * pt->block.size;
+
+		mt->image_offset[i] = mt->total_size;
+		mt->total_size += image_size;
+	}
+
+	mt->buffer = ws->buffer_create(ws, 256, usage, mt->total_size);
 	if (!mt->buffer) {
 		FREE(mt);
 		return NULL;
@@ -83,6 +104,15 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 	struct nv50_miptree *mt = nv50_miptree(pt);
 	struct nv50_surface *s;
 	struct pipe_surface *ps;
+	int img;
+
+	if (pt->target == PIPE_TEXTURE_CUBE)
+		img = face;
+	else
+	if (pt->target == PIPE_TEXTURE_3D)
+		img = zslice;
+	else
+		img = 0;
 
 	s = CALLOC_STRUCT(nv50_surface);
 	if (!s)
@@ -98,7 +128,7 @@ nv50_miptree_surface_new(struct pipe_screen *pscreen, struct pipe_texture *pt,
 	ps->nblocksx = pt->nblocksx[level];
 	ps->nblocksy = pt->nblocksy[level];
 	ps->stride = ps->width * ps->block.size;
-	ps->offset = 0;
+	ps->offset = mt->image_offset[img];
 	ps->usage = flags;
 	ps->status = PIPE_SURFACE_STATUS_DEFINED;
 
diff --git a/src/gallium/drivers/nv50/nv50_program.c b/src/gallium/drivers/nv50/nv50_program.c
index d6fbdd18243..d66e1d0949d 100644
--- a/src/gallium/drivers/nv50/nv50_program.c
+++ b/src/gallium/drivers/nv50/nv50_program.c
@@ -179,6 +179,38 @@ free_temp(struct nv50_pc *pc, struct nv50_reg *r)
 	}
 }
 
+static int
+alloc_temp4(struct nv50_pc *pc, struct nv50_reg *dst[4], int idx)
+{
+	int i;
+
+	if ((idx + 4) >= NV50_SU_MAX_TEMP)
+		return 1;
+
+	if (pc->r_temp[idx] || pc->r_temp[idx + 1] ||
+	    pc->r_temp[idx + 2] || pc->r_temp[idx + 3])
+		return alloc_temp4(pc, dst, idx + 1);
+
+	for (i = 0; i < 4; i++) {
+		dst[i] = CALLOC_STRUCT(nv50_reg);
+		dst[i]->type = P_TEMP;
+		dst[i]->index = -1;
+		dst[i]->hw = idx + i;
+		pc->r_temp[idx + i] = dst[i];
+	}
+
+	return 0;
+}
+
+static void
+free_temp4(struct nv50_pc *pc, struct nv50_reg *reg[4])
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		free_temp(pc, reg[i]);
+}
+
 static struct nv50_reg *
 temp_temp(struct nv50_pc *pc)
 {
@@ -902,7 +934,7 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 {
 	const struct tgsi_full_instruction *inst = &tok->FullInstruction;
 	struct nv50_reg *rdst[4], *dst[4], *src[3][4], *temp;
-	unsigned mask, sat;
+	unsigned mask, sat, unit;
 	int i, c;
 
 	mask = inst->FullDstRegisters[0].DstRegister.WriteMask;
@@ -916,8 +948,13 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 	}
 
 	for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
+		struct tgsi_full_src_register *fs = &inst->FullSrcRegisters[i];
+
+		if (fs->SrcRegister.File == TGSI_FILE_SAMPLER)
+			unit = fs->SrcRegister.Index;
+
 		for (c = 0; c < 4; c++)
-			src[i][c] = tgsi_src(pc, c, &inst->FullSrcRegisters[i]);
+			src[i][c] = tgsi_src(pc, c, fs);
 	}
 
 	if (sat) {
@@ -1155,35 +1192,30 @@ nv50_program_tx_insn(struct nv50_pc *pc, const union tgsi_full_token *tok)
 		}
 		break;
 	case TGSI_OPCODE_TEX:
-		{
-			struct nv50_reg *t0, *t1, *t2, *t3;
-			struct nv50_program_exec *e;
+	case TGSI_OPCODE_TXP:
+	{
+		struct nv50_reg *t[4];
+		struct nv50_program_exec *e;
 
-			t0 = alloc_temp(pc, NULL);
-			t0 = alloc_temp(pc, NULL);
-			t1 = alloc_temp(pc, NULL);
-			t2 = alloc_temp(pc, NULL);
-			t3 = alloc_temp(pc, NULL);
-			emit_mov(pc, t0, src[0][0]);
-			emit_mov(pc, t1, src[0][1]);
+		alloc_temp4(pc, t, 0);
+		emit_mov(pc, t[0], src[0][0]);
+		emit_mov(pc, t[1], src[0][1]);
 
-			e = exec(pc);
-			e->inst[0] = 0xf6400000;
-			set_long(pc, e);
-			e->inst[1] |= 0x0000c004;
-			set_dst(pc, t0, e);
-			emit(pc, e);
+		e = exec(pc);
+		e->inst[0] = 0xf6400000;
+		e->inst[0] |= (unit << 9);
+		set_long(pc, e);
+		e->inst[1] |= 0x0000c004;
+		set_dst(pc, t[0], e);
+		emit(pc, e);
 
-			if (mask & (1 << 0)) emit_mov(pc, dst[0], t0);
-			if (mask & (1 << 1)) emit_mov(pc, dst[1], t1);
-			if (mask & (1 << 2)) emit_mov(pc, dst[2], t2);
-			if (mask & (1 << 3)) emit_mov(pc, dst[3], t3);
+		if (mask & (1 << 0)) emit_mov(pc, dst[0], t[0]);
+		if (mask & (1 << 1)) emit_mov(pc, dst[1], t[1]);
+		if (mask & (1 << 2)) emit_mov(pc, dst[2], t[2]);
+		if (mask & (1 << 3)) emit_mov(pc, dst[3], t[3]);
 
-			free_temp(pc, t0);
-			free_temp(pc, t1);
-			free_temp(pc, t2);
-			free_temp(pc, t3);
-		}
+		free_temp4(pc, t);
+	}
 		break;
 	case TGSI_OPCODE_XPD:
 		temp = alloc_temp(pc, NULL);
@@ -1570,8 +1602,13 @@ nv50_program_validate_code(struct nv50_context *nv50, struct nv50_program *p)
 	if (!upload)
 		return;
 
+	NOUVEAU_ERR("-------\n");
 	up = ptr = MALLOC(p->exec_size * 4);
 	for (e = p->exec_head; e; e = e->next) {
+		NOUVEAU_ERR("0x%08x\n", e->inst[0]);
+		if (is_long(e))
+			NOUVEAU_ERR("0x%08x\n", e->inst[1]);
+
 		*(ptr++) = e->inst[0];
 		if (is_long(e))
 			*(ptr++) = e->inst[1];
@@ -1687,7 +1724,7 @@ nv50_fragprog_validate(struct nv50_context *nv50)
 void
 nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
 {
-	struct pipe_winsys *ws = nv50->pipe.winsys;
+	struct pipe_screen *pscreen = nv50->pipe.screen;
 
 	while (p->exec_head) {
 		struct nv50_program_exec *e = p->exec_head;
@@ -1699,7 +1736,7 @@ nv50_program_destroy(struct nv50_context *nv50, struct nv50_program *p)
 	p->exec_size = 0;
 
 	if (p->buffer)
-		pipe_buffer_reference(ws, &p->buffer, NULL);
+		pipe_buffer_reference(pscreen, &p->buffer, NULL);
 
 	nv50->screen->nvws->res_free(&p->data);
 
diff --git a/src/gallium/drivers/nv50/nv50_query.c b/src/gallium/drivers/nv50/nv50_query.c
index 26bd90ccc5e..777e77906d5 100644
--- a/src/gallium/drivers/nv50/nv50_query.c
+++ b/src/gallium/drivers/nv50/nv50_query.c
@@ -51,7 +51,7 @@ nv50_query_end(struct pipe_context *pipe, struct pipe_query *q)
 
 static boolean
 nv50_query_result(struct pipe_context *pipe, struct pipe_query *q,
-		  boolean wait, uint64 *result)
+		  boolean wait, uint64_t *result)
 {
 	NOUVEAU_ERR("unimplemented\n");
 	*result = 0xdeadcafe;
diff --git a/src/gallium/drivers/softpipe/sp_context.h b/src/gallium/drivers/softpipe/sp_context.h
index 7ab12a6d702..e2451c6ecb5 100644
--- a/src/gallium/drivers/softpipe/sp_context.h
+++ b/src/gallium/drivers/softpipe/sp_context.h
@@ -94,7 +94,7 @@ struct softpipe_context {
    /* Counter for occlusion queries.  Note this supports overlapping
     * queries.
     */
-   uint64 occlusion_count;
+   uint64_t occlusion_count;
 
    /*
     * Mapped vertex buffers
diff --git a/src/gallium/drivers/softpipe/sp_quad_fs.c b/src/gallium/drivers/softpipe/sp_quad_fs.c
index 40329a95627..5dacbbe55f8 100644
--- a/src/gallium/drivers/softpipe/sp_quad_fs.c
+++ b/src/gallium/drivers/softpipe/sp_quad_fs.c
@@ -171,7 +171,6 @@ static void shade_destroy(struct quad_stage *qs)
 struct quad_stage *sp_quad_shade_stage( struct softpipe_context *softpipe )
 {
    struct quad_shade_stage *qss = CALLOC_STRUCT(quad_shade_stage);
-   uint i;
 
    /* allocate storage for program inputs/outputs, aligned to 16 bytes */
    qss->inputs = MALLOC(PIPE_MAX_ATTRIBS * sizeof(*qss->inputs) + 16);
diff --git a/src/gallium/drivers/softpipe/sp_query.c b/src/gallium/drivers/softpipe/sp_query.c
index 2106ee1d235..b0d8e01426d 100644
--- a/src/gallium/drivers/softpipe/sp_query.c
+++ b/src/gallium/drivers/softpipe/sp_query.c
@@ -37,8 +37,8 @@
 #include "sp_query.h"
 
 struct softpipe_query {
-   uint64 start;
-   uint64 end;
+   uint64_t start;
+   uint64_t end;
 };
 
 
@@ -87,7 +87,7 @@ static boolean
 softpipe_get_query_result(struct pipe_context *pipe, 
 			  struct pipe_query *q,
 			  boolean wait,
-			  uint64 *result )
+			  uint64_t *result )
 {
    struct softpipe_query *sq = softpipe_query(q);
    *result = sq->end - sq->start;
diff --git a/src/gallium/drivers/softpipe/sp_screen.c b/src/gallium/drivers/softpipe/sp_screen.c
index 12f98c32f53..11b08b3a82d 100644
--- a/src/gallium/drivers/softpipe/sp_screen.c
+++ b/src/gallium/drivers/softpipe/sp_screen.c
@@ -55,9 +55,9 @@ softpipe_get_param(struct pipe_screen *screen, int param)
 {
    switch (param) {
    case PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS:
-      return 8;
+      return PIPE_MAX_SAMPLERS;
    case PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS:
-      return 8;
+      return PIPE_MAX_SAMPLERS;
    case PIPE_CAP_NPOT_TEXTURES:
       return 1;
    case PIPE_CAP_TWO_SIDED_STENCIL:
diff --git a/src/gallium/drivers/softpipe/sp_texture.c b/src/gallium/drivers/softpipe/sp_texture.c
index 0cb4b2f03c3..a64dc89f432 100644
--- a/src/gallium/drivers/softpipe/sp_texture.c
+++ b/src/gallium/drivers/softpipe/sp_texture.c
@@ -94,31 +94,50 @@ softpipe_texture_layout(struct pipe_screen *screen,
    return spt->buffer != NULL;
 }
 
+/* Hack it up to use the old winsys->surface_alloc_storage()
+ * method for now:
+ */
 static boolean
 softpipe_displaytarget_layout(struct pipe_screen *screen,
                               struct softpipe_texture * spt)
 {
    struct pipe_winsys *ws = screen->winsys;
-   size_t tex_size;
-   unsigned cpp;
-
-   switch (spt->base.format) {
-   case PIPE_FORMAT_R5G6B5_UNORM:
-      cpp = 2;
-      break;
-   case PIPE_FORMAT_Z24S8_UNORM: 
-   case PIPE_FORMAT_A8R8G8B8_UNORM:
-   default:
-      cpp = 4;
-      break;
+   struct pipe_surface surf;
+   unsigned flags = (PIPE_BUFFER_USAGE_CPU_READ |
+                     PIPE_BUFFER_USAGE_CPU_WRITE |
+                     PIPE_BUFFER_USAGE_GPU_READ |
+                     PIPE_BUFFER_USAGE_GPU_WRITE);
+   int ret;
+
+
+   memset(&surf, 0, sizeof(surf));
+
+   ret =ws->surface_alloc_storage( ws, 
+                                   &surf,
+                                   spt->base.width[0], 
+                                   spt->base.height[0],
+                                   spt->base.format,
+                                   flags,
+                                   spt->base.tex_usage);
+   if(ret != 0)
+      return FALSE;
+
+   if (!surf.buffer) {
+      /* allocation failed */
+      return FALSE;
    }
-   tex_size = spt->base.width[0] * cpp * spt->base.height[0];
-   spt->buffer = ws->buffer_create(ws, 64, PIPE_BUFFER_USAGE_PIXEL, tex_size);
+
    /* Now extract the goodies: 
     */
    spt->base.nblocksx[0] = pf_get_nblocksx(&spt->base.block, spt->base.width[0]);  
    spt->base.nblocksy[0] = pf_get_nblocksy(&spt->base.block, spt->base.height[0]);  
-   spt->stride[0] = spt->base.width[0] * cpp;
+   spt->stride[0] = surf.stride;
+
+   /* Transfer the reference:
+    */
+   spt->buffer = surf.buffer;
+   surf.buffer = NULL;
+
    return spt->buffer != NULL;
 }
 
@@ -220,10 +239,8 @@ softpipe_get_tex_surface(struct pipe_screen *screen,
 
    ps = CALLOC_STRUCT(pipe_surface);
    ps->refcount = 1;
-   ps->winsys = ws;
    if (ps) {
       assert(ps->refcount);
-      assert(ps->winsys);
       pipe_texture_reference(&ps->texture, pt);
       pipe_buffer_reference(screen, &ps->buffer, spt->buffer);
       ps->format = pt->format;
diff --git a/src/gallium/drivers/trace/tr_context.c b/src/gallium/drivers/trace/tr_context.c
index 1dd77193791..f0d51ad82ef 100644
--- a/src/gallium/drivers/trace/tr_context.c
+++ b/src/gallium/drivers/trace/tr_context.c
@@ -274,11 +274,11 @@ static INLINE boolean
 trace_context_get_query_result(struct pipe_context *_pipe, 
                                struct pipe_query *query,
                                boolean wait,
-                               uint64 *presult)
+                               uint64_t *presult)
 {
    struct trace_context *tr_ctx = trace_context(_pipe);
    struct pipe_context *pipe = tr_ctx->pipe;
-   uint64 result;
+   uint64_t result;
    boolean _result;
 
    trace_dump_call_begin("pipe_context", "get_query_result");
diff --git a/src/gallium/include/pipe/p_compiler.h b/src/gallium/include/pipe/p_compiler.h
index 7bcebd3d6b6..bc2a0a7ef3a 100644
--- a/src/gallium/include/pipe/p_compiler.h
+++ b/src/gallium/include/pipe/p_compiler.h
@@ -96,7 +96,6 @@ typedef int     _Bool;
 typedef unsigned int       uint;
 typedef unsigned char      ubyte;
 typedef unsigned short     ushort;
-typedef uint64_t           uint64;
 
 #if 0
 #define boolean bool
@@ -112,20 +111,22 @@ typedef unsigned char boolean;
 
 
 /* Function inlining */
-#ifdef __cplusplus
-#  define INLINE inline
-#elif defined(__GNUC__)
-#  define INLINE __inline__
-#elif defined(_MSC_VER)
-#  define INLINE __inline
-#elif defined(__ICL)
-#  define INLINE __inline
-#elif defined(__INTEL_COMPILER)
-#  define INLINE inline
-#elif defined(__WATCOMC__) && (__WATCOMC__ >= 1100)
-#  define INLINE __inline
-#else
-#  define INLINE
+#ifndef INLINE
+#  ifdef __cplusplus
+#    define INLINE inline
+#  elif defined(__GNUC__)
+#    define INLINE __inline__
+#  elif defined(_MSC_VER)
+#    define INLINE __inline
+#  elif defined(__ICL)
+#    define INLINE __inline
+#  elif defined(__INTEL_COMPILER)
+#    define INLINE inline
+#  elif defined(__WATCOMC__) && (__WATCOMC__ >= 1100)
+#    define INLINE __inline
+#  else
+#    define INLINE
+#  endif
 #endif
 
 
diff --git a/src/gallium/include/pipe/p_context.h b/src/gallium/include/pipe/p_context.h
index 2646706ff23..166c6b6b7e0 100644
--- a/src/gallium/include/pipe/p_context.h
+++ b/src/gallium/include/pipe/p_context.h
@@ -109,7 +109,7 @@ struct pipe_context {
    boolean (*get_query_result)(struct pipe_context *pipe, 
                                struct pipe_query *q,
                                boolean wait,
-                               uint64 *result);
+                               uint64_t *result);
    /*@}*/
 
    /**
diff --git a/src/gallium/state_trackers/python/p_context.i b/src/gallium/state_trackers/python/p_context.i
index 1fdcec639f7..7b8b64b5923 100644
--- a/src/gallium/state_trackers/python/p_context.i
+++ b/src/gallium/state_trackers/python/p_context.i
@@ -245,7 +245,7 @@ struct st_context {
       memcpy(map, vertices, size);
       pipe_buffer_unmap(screen, vbuf);
       
-      util_draw_vertex_buffer(pipe, vbuf, prim, num_verts, num_attribs);
+      util_draw_vertex_buffer(pipe, vbuf, 0, prim, num_verts, num_attribs);
       
 error2:
       pipe_buffer_reference(screen, &vbuf, NULL);
diff --git a/src/gallium/winsys/gdi/SConscript b/src/gallium/winsys/gdi/SConscript
index fdcdde101ac..e68d5db7f4a 100644
--- a/src/gallium/winsys/gdi/SConscript
+++ b/src/gallium/winsys/gdi/SConscript
@@ -19,7 +19,7 @@ if env['platform'] == 'windows':
 
 	env.Append(LIBS = [
 		'gdi32',
-		'user32',ss
+		'user32',
 		'kernel32',
 	])
 
diff --git a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
index e66ce48f2dc..bd5aa10a20f 100644
--- a/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
+++ b/src/gallium/winsys/gdi/gdi_softpipe_winsys.c
@@ -320,7 +320,7 @@ gdi_softpipe_flush_frontbuffer(struct pipe_winsys *winsys,
     memset(&bmi, 0, sizeof(BITMAPINFO));
     bmi.bmiHeader.biSize = sizeof(BITMAPINFOHEADER);
     bmi.bmiHeader.biWidth = surface->stride / pf_get_size(surface->format);
-    bmi.bmiHeader.biHeight= -surface->height;
+    bmi.bmiHeader.biHeight= -(long)surface->height;
     bmi.bmiHeader.biPlanes = 1;
     bmi.bmiHeader.biBitCount = pf_get_bits(surface->format);
     bmi.bmiHeader.biCompression = BI_RGB;
@@ -337,8 +337,23 @@ gdi_softpipe_flush_frontbuffer(struct pipe_winsys *winsys,
 }
 
 
-const struct stw_winsys stw_winsys = {
+static const struct stw_winsys stw_winsys = {
    &gdi_softpipe_screen_create,
    &gdi_softpipe_context_create,
    &gdi_softpipe_flush_frontbuffer
 };
+
+
+BOOL WINAPI
+DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpReserved)
+{
+   switch (fdwReason) {
+   case DLL_PROCESS_ATTACH:
+      return st_init(&stw_winsys);
+
+   case DLL_PROCESS_DETACH:
+      st_cleanup();
+      break;
+   }
+   return TRUE;
+}
diff --git a/src/glu/sgi/glu.exports b/src/glu/sgi/glu.exports
index 1d1b6da24fe..aeb72729166 100644
--- a/src/glu/sgi/glu.exports
+++ b/src/glu/sgi/glu.exports
@@ -57,3 +57,62 @@
 	gluTessVertex
 	gluUnProject
 	gluUnProject4
+	mgluBeginCurve
+	mgluBeginPolygon
+	mgluBeginSurface
+	mgluBeginTrim
+	mgluBuild1DMipmapLevels
+	mgluBuild1DMipmaps
+	mgluBuild2DMipmapLevels
+	mgluBuild2DMipmaps
+	mgluBuild3DMipmapLevels
+	mgluBuild3DMipmaps
+	mgluCheckExtension
+	mgluCylinder
+	mgluDeleteNurbsRenderer
+	mgluDeleteQuadric
+	mgluDeleteTess
+	mgluDisk
+	mgluEndCurve
+	mgluEndPolygon
+	mgluEndSurface
+	mgluEndTrim
+	mgluErrorString
+	mgluGetNurbsProperty
+	mgluGetString
+	mgluGetTessProperty
+	mgluLoadSamplingMatrices
+	mgluLookAt
+	mgluNewNurbsRenderer
+	mgluNewQuadric
+	mgluNewTess
+	mgluNextContour
+	mgluNurbsCallback
+	mgluNurbsCallbackData
+	mgluNurbsCallbackDataEXT
+	mgluNurbsCurve
+	mgluNurbsProperty
+	mgluNurbsSurface
+	mgluOrtho2D
+	mgluPartialDisk
+	mgluPerspective
+	mgluPickMatrix
+	mgluProject
+	mgluPwlCurve
+	mgluQuadricCallback
+	mgluQuadricDrawStyle
+	mgluQuadricNormals
+	mgluQuadricOrientation
+	mgluQuadricTexture
+	mgluScaleImage
+	mgluSphere
+	mgluTessBeginContour
+	mgluTessBeginPolygon
+	mgluTessCallback
+	mgluTessEndContour
+	mgluTessEndPolygon
+	mgluTessNormal
+	mgluTessProperty
+	mgluTessVertex
+	mgluUnProject
+	mgluUnProject4
diff --git a/src/glut/glx/SConscript b/src/glut/glx/SConscript
index 99b3bb7df64..cb442ba027f 100644
--- a/src/glut/glx/SConscript
+++ b/src/glut/glx/SConscript
@@ -11,7 +11,6 @@ env.Replace(CPPDEFINES = [
     'BUILD_GLUT32', 
     'GLUT_BUILDING_LIB', 
     'MESA', 
-    '_DLL',
     'NDEBUG',
     'GLUT_NO_WARNING_DISABLE',
 ])
@@ -38,7 +37,6 @@ sources = [
     'glut_dstr.c',
     'glut_event.c',
     'glut_ext.c',
-    'glut_fcb.c',
     'glut_fullscrn.c',
     'glut_gamemode.c',
     'glut_get.c',
@@ -51,6 +49,7 @@ sources = [
     'glut_mesa.c',
     'glut_modifier.c',
     'glut_overlay.c',
+    'glut_ppm.c',
     'glut_shapes.c',
     'glut_space.c',
     'glut_stroke.c',
diff --git a/src/mesa/Makefile b/src/mesa/Makefile
index 274010c54f5..552c20e0b9b 100644
--- a/src/mesa/Makefile
+++ b/src/mesa/Makefile
@@ -34,14 +34,10 @@ libmesa.a: $(MESA_OBJECTS)
 
 # Make archive of gl* API dispatcher functions only
 libglapi.a: $(GLAPI_OBJECTS)
-	@ $(MKLIB) -o glapi -static $(GLAPI_OBJECTS)
-
-# Make archive of gl* API dispatcher functions only
-$(GLAPI_LIB): $(GLAPI_OBJECTS)
 	@if [ "${WINDOW_SYSTEM}" = "dri" ] ; then \
 		touch libglapi.a ; \
 	else \
-		$(TOP)/bin/mklib -o glapi -static $(GLAPI_OBJECTS) ; \
+		$(MKLIB) -o glapi -static $(GLAPI_OBJECTS) ; \
 	fi
 
 ######################################################################
diff --git a/src/mesa/SConscript b/src/mesa/SConscript
index abd64030b63..01620ee6147 100644
--- a/src/mesa/SConscript
+++ b/src/mesa/SConscript
@@ -12,10 +12,11 @@ if env['platform'] != 'winddk':
 		'#/src/mesa',
 	])
 	
-	if gcc:
-		env.Append(CFLAGS = [
-			'-std=c99',
-		])
+	if env['platform'] == 'windows':
+		env.Append(CPPDEFINES = [
+			'_GDI32_', # prevent gl* being declared __declspec(dllimport) in MS headers
+			'BUILD_GL32', # declare gl* as __declspec(dllexport) in Mesa headers 
+		])	 
 
 	#
 	# Source files
@@ -165,6 +166,7 @@ if env['platform'] != 'winddk':
 		'state_tracker/st_context.c',
 		'state_tracker/st_debug.c',
 		'state_tracker/st_draw.c',
+		'state_tracker/st_draw_feedback.c',
 		'state_tracker/st_extensions.c',
 		'state_tracker/st_format.c',
 		'state_tracker/st_framebuffer.c',
diff --git a/src/mesa/drivers/dri/common/dri_util.c b/src/mesa/drivers/dri/common/dri_util.c
index a2316e2662b..ae790554055 100644
--- a/src/mesa/drivers/dri/common/dri_util.c
+++ b/src/mesa/drivers/dri/common/dri_util.c
@@ -314,10 +314,28 @@ static void driReportDamage(__DRIdrawable *pdp,
 static void driSwapBuffers(__DRIdrawable *dPriv)
 {
     __DRIscreen *psp = dPriv->driScreenPriv;
+    drm_clip_rect_t *rects;
+    int i;
+    
+    if (!dPriv->numClipRects)
+        return;
 
     psp->DriverAPI.SwapBuffers(dPriv);
 
-    driReportDamage(dPriv, dPriv->pClipRects, dPriv->numClipRects);
+    rects = _mesa_malloc(sizeof(*rects) * dPriv->numClipRects);
+
+    if (!rects)
+        return;
+
+    for (i = 0; i < dPriv->numClipRects; i++) {
+        rects[i].x1 = dPriv->pClipRects[i].x1 - dPriv->x;
+        rects[i].y1 = dPriv->pClipRects[i].y1 - dPriv->y;
+        rects[i].x2 = dPriv->pClipRects[i].x2 - dPriv->x;
+        rects[i].y2 = dPriv->pClipRects[i].y2 - dPriv->y;
+    }
+
+    driReportDamage(dPriv, rects, dPriv->numClipRects);
+    _mesa_free(rects);
 }
 
 static int driDrawableGetMSC( __DRIscreen *sPriv, __DRIdrawable *dPriv,
diff --git a/src/mesa/drivers/dri/common/vblank.c b/src/mesa/drivers/dri/common/vblank.c
index d610253fe6f..12aeaa108f7 100644
--- a/src/mesa/drivers/dri/common/vblank.c
+++ b/src/mesa/drivers/dri/common/vblank.c
@@ -130,9 +130,8 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
 
 
    if ( divisor != 0 ) {
-      unsigned int target = (unsigned int)target_msc;
-      unsigned int next = target;
-      unsigned int r;
+      int64_t next = target_msc;
+      int64_t r;
       int dont_wait = (target_msc == 0);
 
       do {
@@ -154,9 +153,9 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
 
 	 *msc = vblank_to_msc(priv, vbl.reply.sequence);
 
-         dont_wait = 0;
-         if (target_msc != 0 && *msc == target)
+         if (!dont_wait && *msc == next)
             break;
+         dont_wait = 0;
 
          /* Assuming the wait-done test fails, the next refresh to wait for
           * will be one that satisfies (MSC % divisor) == remainder.  The
@@ -165,11 +164,12 @@ int driWaitForMSC32( __DRIdrawablePrivate *priv,
           * If this refresh has already happened, we add divisor to obtain 
           * the next refresh after the current one that will satisfy it.
           */
-         r = (*msc % (unsigned int)divisor);
-         next = (*msc - r + (unsigned int)remainder);
-         if (next <= *msc) next += (unsigned int)divisor;
+         r = ((uint64_t)*msc % divisor);
+         next = (*msc - r + remainder);
+         if (next <= *msc)
+	    next += divisor;
 
-      } while ( r != (unsigned int)remainder );
+      } while (r != remainder);
    }
    else {
       /* If the \c divisor is zero, just wait until the MSC is greater
diff --git a/src/mesa/drivers/dri/i915/i915_context.c b/src/mesa/drivers/dri/i915/i915_context.c
index e0ddc7fd613..9bff74294d8 100644
--- a/src/mesa/drivers/dri/i915/i915_context.c
+++ b/src/mesa/drivers/dri/i915/i915_context.c
@@ -55,6 +55,7 @@ static const struct dri_extension i915_extensions[] = {
    {"GL_ARB_fragment_program", NULL},
    {"GL_ARB_shadow", NULL},
    {"GL_ARB_texture_non_power_of_two", NULL},
+   {"GL_ATI_texture_env_combine3",       NULL},
    {"GL_EXT_shadow_funcs", NULL},
    {NULL, NULL}
 };
diff --git a/src/mesa/drivers/dri/i915/i915_fragprog.c b/src/mesa/drivers/dri/i915/i915_fragprog.c
index 8bd761ec6a1..4760906a7ed 100644
--- a/src/mesa/drivers/dri/i915/i915_fragprog.c
+++ b/src/mesa/drivers/dri/i915/i915_fragprog.c
@@ -1105,30 +1105,14 @@ i915ValidateFragmentProgram(struct i915_context *i915)
       EMIT_ATTR(_TNL_ATTRIB_COLOR0, EMIT_4UB_4F_BGRA, S4_VFMT_COLOR, 4);
    }
 
-   if ((inputsRead & (FRAG_BIT_COL1 | FRAG_BIT_FOGC)) ||
-       i915->vertex_fog != I915_FOG_NONE) {
-
-      if (inputsRead & FRAG_BIT_COL1) {
-         intel->specoffset = offset / 4;
-         EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_3UB_3F_BGR, S4_VFMT_SPEC_FOG, 3);
-      }
-      else
-         EMIT_PAD(3);
-
-      if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE)
-         EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1UB_1F, S4_VFMT_SPEC_FOG, 1);
-      else
-         EMIT_PAD(1);
+   if (inputsRead & FRAG_BIT_COL1) {
+       intel->specoffset = offset / 4;
+       EMIT_ATTR(_TNL_ATTRIB_COLOR1, EMIT_4UB_4F_BGRA, S4_VFMT_SPEC_FOG, 4);
    }
 
-   /* XXX this was disabled, but enabling this code helped fix the Glean
-    * tfragprog1 fog tests.
-    */
-#if 1
    if ((inputsRead & FRAG_BIT_FOGC) || i915->vertex_fog != I915_FOG_NONE) {
       EMIT_ATTR(_TNL_ATTRIB_FOG, EMIT_1F, S4_VFMT_FOG_PARAM, 4);
    }
-#endif
 
    for (i = 0; i < p->ctx->Const.MaxTextureCoordUnits; i++) {
       if (inputsRead & FRAG_BIT_TEX(i)) {
diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
index 1d6ac2cea68..a415e378fff 100644
--- a/src/mesa/drivers/dri/i965/brw_context.c
+++ b/src/mesa/drivers/dri/i965/brw_context.c
@@ -32,6 +32,7 @@
 
 #include "main/imports.h"
 #include "main/api_noop.h"
+#include "main/macros.h"
 #include "main/vtxfmt.h"
 #include "main/simple_list.h"
 #include "shader/shader_api.h"
@@ -128,9 +129,10 @@ GLboolean brwCreateContext( const __GLcontextModes *mesaVis,
 
    TNL_CONTEXT(ctx)->Driver.RunPipeline = _tnl_run_pipeline;
 
-   ctx->Const.MaxTextureUnits = BRW_MAX_TEX_UNIT;
    ctx->Const.MaxTextureImageUnits = BRW_MAX_TEX_UNIT;
-   ctx->Const.MaxTextureCoordUnits = BRW_MAX_TEX_UNIT;
+   ctx->Const.MaxTextureCoordUnits = 8; /* Mesa limit */
+   ctx->Const.MaxTextureUnits = MIN2(ctx->Const.MaxTextureCoordUnits,
+                                     ctx->Const.MaxTextureImageUnits);
    ctx->Const.MaxVertexTextureImageUnits = 0; /* no vertex shader textures */
 
    /* Advertise the full hardware capabilities.  The new memory
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index 77980109cdf..5d3f99e025e 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -238,7 +238,7 @@ struct brw_vs_ouput_sizes {
 };
 
 
-#define BRW_MAX_TEX_UNIT 8
+#define BRW_MAX_TEX_UNIT 16
 #define BRW_WM_MAX_SURF BRW_MAX_TEX_UNIT + MAX_DRAW_BUFFERS
 
 enum brw_cache_id {
diff --git a/src/mesa/drivers/dri/i965/brw_eu.h b/src/mesa/drivers/dri/i965/brw_eu.h
index 49b422ee2ff..9e2b39af9bb 100644
--- a/src/mesa/drivers/dri/i965/brw_eu.h
+++ b/src/mesa/drivers/dri/i965/brw_eu.h
@@ -129,17 +129,28 @@ static INLINE int type_sz( GLuint type )
    }
 }
 
+/**
+ * Construct a brw_reg.
+ * \param file  one of the BRW_x_REGISTER_FILE values
+ * \param nr  register number/index
+ * \param subnr  register sub number
+ * \param type  one of BRW_REGISTER_TYPE_x
+ * \param vstride  one of BRW_VERTICAL_STRIDE_x
+ * \param width  one of BRW_WIDTH_x
+ * \param hstride  one of BRW_HORIZONTAL_STRIDE_x
+ * \param swizzle  one of BRW_SWIZZLE_x
+ * \param writemask  WRITEMASK_X/Y/Z/W bitfield
+ */
 static INLINE struct brw_reg brw_reg( GLuint file,
-					GLuint nr,
-					GLuint subnr,
-					GLuint type,
-					GLuint vstride,
-					GLuint width,
-					GLuint hstride,
-					GLuint swizzle,
-					GLuint writemask)
-{
-      
+                                      GLuint nr,
+                                      GLuint subnr,
+                                      GLuint type,
+                                      GLuint vstride,
+                                      GLuint width,
+                                      GLuint hstride,
+                                      GLuint swizzle,
+                                      GLuint writemask )
+{
    struct brw_reg reg;
    reg.type = type;
    reg.file = file;
@@ -166,6 +177,7 @@ static INLINE struct brw_reg brw_reg( GLuint file,
    return reg;
 }
 
+/** Construct float[16] register */
 static INLINE struct brw_reg brw_vec16_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -181,6 +193,7 @@ static INLINE struct brw_reg brw_vec16_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
+/** Construct float[8] register */
 static INLINE struct brw_reg brw_vec8_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -196,7 +209,7 @@ static INLINE struct brw_reg brw_vec8_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
-
+/** Construct float[4] register */
 static INLINE struct brw_reg brw_vec4_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -212,7 +225,7 @@ static INLINE struct brw_reg brw_vec4_reg( GLuint file,
 		  WRITEMASK_XYZW);
 }
 
-
+/** Construct float[2] register */
 static INLINE struct brw_reg brw_vec2_reg( GLuint file,
 					      GLuint nr,
 					      GLuint subnr )
@@ -228,6 +241,7 @@ static INLINE struct brw_reg brw_vec2_reg( GLuint file,
 		  WRITEMASK_XY);
 }
 
+/** Construct float[1] register */
 static INLINE struct brw_reg brw_vec1_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -277,6 +291,7 @@ static INLINE struct brw_reg byte_offset( struct brw_reg reg,
 }
    
 
+/** Construct unsigned word[16] register */
 static INLINE struct brw_reg brw_uw16_reg( GLuint file,
 					     GLuint nr,
 					     GLuint subnr )
@@ -284,6 +299,7 @@ static INLINE struct brw_reg brw_uw16_reg( GLuint file,
    return suboffset(retype(brw_vec16_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
+/** Construct unsigned word[8] register */
 static INLINE struct brw_reg brw_uw8_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
@@ -291,6 +307,7 @@ static INLINE struct brw_reg brw_uw8_reg( GLuint file,
    return suboffset(retype(brw_vec8_reg(file, nr, 0), BRW_REGISTER_TYPE_UW), subnr);
 }
 
+/** Construct unsigned word[1] register */
 static INLINE struct brw_reg brw_uw1_reg( GLuint file,
 					    GLuint nr,
 					    GLuint subnr )
@@ -311,6 +328,7 @@ static INLINE struct brw_reg brw_imm_reg( GLuint type )
 		   0);      
 }
 
+/** Construct float immediate register */
 static INLINE struct brw_reg brw_imm_f( GLfloat f )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_F);
@@ -318,6 +336,7 @@ static INLINE struct brw_reg brw_imm_f( GLfloat f )
    return imm;
 }
 
+/** Construct integer immediate register */
 static INLINE struct brw_reg brw_imm_d( GLint d )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_D);
@@ -325,6 +344,7 @@ static INLINE struct brw_reg brw_imm_d( GLint d )
    return imm;
 }
 
+/** Construct uint immediate register */
 static INLINE struct brw_reg brw_imm_ud( GLuint ud )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UD);
@@ -332,6 +352,7 @@ static INLINE struct brw_reg brw_imm_ud( GLuint ud )
    return imm;
 }
 
+/** Construct ushort immediate register */
 static INLINE struct brw_reg brw_imm_uw( GLushort uw )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_UW);
@@ -339,6 +360,7 @@ static INLINE struct brw_reg brw_imm_uw( GLushort uw )
    return imm;
 }
 
+/** Construct short immediate register */
 static INLINE struct brw_reg brw_imm_w( GLshort w )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_W);
@@ -350,8 +372,7 @@ static INLINE struct brw_reg brw_imm_w( GLshort w )
  * numbers alias with _V and _VF below:
  */
 
-/* Vector of eight signed half-byte values: 
- */
+/** Construct vector of eight signed half-byte values */
 static INLINE struct brw_reg brw_imm_v( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_V);
@@ -362,8 +383,7 @@ static INLINE struct brw_reg brw_imm_v( GLuint v )
    return imm;
 }
 
-/* Vector of four 8-bit float values:
- */
+/** Construct vector of four 8-bit float values */
 static INLINE struct brw_reg brw_imm_vf( GLuint v )
 {
    struct brw_reg imm = brw_imm_reg(BRW_REGISTER_TYPE_VF);
@@ -400,44 +420,43 @@ static INLINE struct brw_reg brw_address( struct brw_reg reg )
    return brw_imm_uw(reg.nr * REG_SIZE + reg.subnr);
 }
 
-
-static INLINE struct brw_reg brw_vec1_grf( GLuint nr,
-					       GLuint subnr )
+/** Construct float[1] general-purpose register */
+static INLINE struct brw_reg brw_vec1_grf( GLuint nr, GLuint subnr )
 {
    return brw_vec1_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_vec8_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[2] general-purpose register */
+static INLINE struct brw_reg brw_vec2_grf( GLuint nr, GLuint subnr )
 {
-   return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
+   return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_vec4_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[4] general-purpose register */
+static INLINE struct brw_reg brw_vec4_grf( GLuint nr, GLuint subnr )
 {
    return brw_vec4_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-
-static INLINE struct brw_reg brw_vec2_grf( GLuint nr,
-					     GLuint subnr )
+/** Construct float[8] general-purpose register */
+static INLINE struct brw_reg brw_vec8_grf( GLuint nr, GLuint subnr )
 {
-   return brw_vec2_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
+   return brw_vec8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_uw8_grf( GLuint nr,
-					    GLuint subnr )
+
+static INLINE struct brw_reg brw_uw8_grf( GLuint nr, GLuint subnr )
 {
    return brw_uw8_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
-static INLINE struct brw_reg brw_uw16_grf( GLuint nr,
-					    GLuint subnr )
+static INLINE struct brw_reg brw_uw16_grf( GLuint nr, GLuint subnr )
 {
    return brw_uw16_reg(BRW_GENERAL_REGISTER_FILE, nr, subnr);
 }
 
+
+/** Construct null register (usually used for setting condition codes) */
 static INLINE struct brw_reg brw_null_reg( void )
 {
    return brw_vec8_reg(BRW_ARCHITECTURE_REGISTER_FILE, 
@@ -524,13 +543,13 @@ static INLINE struct brw_reg stride( struct brw_reg reg,
 				       GLuint width,
 				       GLuint hstride )
 {
-   
    reg.vstride = cvt(vstride);
    reg.width = cvt(width) - 1;
    reg.hstride = cvt(hstride);
    return reg;
 }
 
+
 static INLINE struct brw_reg vec16( struct brw_reg reg )
 {
    return stride(reg, 16,16,1);
@@ -556,6 +575,7 @@ static INLINE struct brw_reg vec1( struct brw_reg reg )
    return stride(reg, 0,1,0);
 }
 
+
 static INLINE struct brw_reg get_element( struct brw_reg reg, GLuint elt )
 {
    return vec1(suboffset(reg, elt));
@@ -687,7 +707,7 @@ static INLINE struct brw_indirect brw_indirect( GLuint addr_subnr, GLint offset
 
 static INLINE struct brw_instruction *current_insn( struct brw_compile *p)
 {
-	return &p->store[p->nr_insn];
+   return &p->store[p->nr_insn];
 }
 
 void brw_pop_insn_state( struct brw_compile *p );
@@ -733,6 +753,7 @@ ALU2(ADD)
 ALU2(MUL)
 ALU1(FRC)
 ALU1(RNDD)
+ALU1(RNDZ)
 ALU2(MAC)
 ALU2(MACH)
 ALU1(LZD)
diff --git a/src/mesa/drivers/dri/i965/brw_eu_emit.c b/src/mesa/drivers/dri/i965/brw_eu_emit.c
index ce4cf46cfa6..4e099b5945c 100644
--- a/src/mesa/drivers/dri/i965/brw_eu_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_eu_emit.c
@@ -439,6 +439,7 @@ ALU2(ADD)
 ALU2(MUL)
 ALU1(FRC)
 ALU1(RNDD)
+ALU1(RNDZ)
 ALU2(MAC)
 ALU2(MACH)
 ALU1(LZD)
diff --git a/src/mesa/drivers/dri/i965/brw_vs_emit.c b/src/mesa/drivers/dri/i965/brw_vs_emit.c
index 4a9541378f7..71e2a95bfd9 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_emit.c
@@ -73,8 +73,6 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 
    c->prog_data.curb_read_length = reg - 1;
 
-
-
    /* Allocate input regs:  
     */
    c->nr_inputs = 0;
@@ -84,8 +82,7 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 	 c->regs[PROGRAM_INPUT][i] = brw_vec8_grf(reg, 0);
 	 reg++;
       }
-   }     
-
+   }
 
    /* Allocate outputs: TODO: could organize the non-position outputs
     * to go straight into message regs.
@@ -339,6 +336,7 @@ static void emit_math1( struct brw_vs_compile *c,
    }
 }
 
+
 static void emit_math2( struct brw_vs_compile *c, 
 			GLuint function,
 			struct brw_reg dst,
@@ -370,7 +368,6 @@ static void emit_math2( struct brw_vs_compile *c,
       release_tmp(c, tmp);
    }
 }
-		     
 
 
 static void emit_exp_noalias( struct brw_vs_compile *c,
@@ -521,8 +518,6 @@ static void emit_log_noalias( struct brw_vs_compile *c,
 }
 
 
-      
-
 /* Need to unalias - consider swizzles:   r0 = DST r0.xxxx r1
  */
 static void emit_dst_noalias( struct brw_vs_compile *c, 
@@ -544,6 +539,7 @@ static void emit_dst_noalias( struct brw_vs_compile *c,
       brw_MOV(p, brw_writemask(dst, WRITEMASK_W), arg1);
 }
 
+
 static void emit_xpd( struct brw_compile *p,
 		      struct brw_reg dst,
 		      struct brw_reg t,
@@ -554,7 +550,6 @@ static void emit_xpd( struct brw_compile *p,
 }
 
 
-
 static void emit_lit_noalias( struct brw_vs_compile *c, 
 			      struct brw_reg dst,
 			      struct brw_reg arg0 )
@@ -596,7 +591,29 @@ static void emit_lit_noalias( struct brw_vs_compile *c,
 }
 
 
+/** 3 or 4-component vector normalization */
+static void emit_nrm( struct brw_vs_compile *c, 
+                      struct brw_reg dst,
+                      struct brw_reg arg0,
+                      int num_comps)
+{
+   struct brw_compile *p = &c->func;
+   struct brw_reg tmp = get_tmp(c);
 
+   /* tmp = dot(arg0, arg0) */
+   if (num_comps == 3)
+      brw_DP3(p, tmp, arg0, arg0);
+   else
+      brw_DP4(p, tmp, arg0, arg0);
+
+   /* tmp = 1 / sqrt(tmp) */
+   emit_math1(c, BRW_MATH_FUNCTION_RSQ, tmp, tmp, BRW_MATH_PRECISION_FULL);
+
+   /* dst = arg0 * tmp */
+   brw_MUL(p, dst, arg0, tmp);
+
+   release_tmp(c, tmp);
+}
 
 
 /* TODO: relative addressing!
@@ -634,7 +651,6 @@ static struct brw_reg get_reg( struct brw_vs_compile *c,
 }
 
 
-
 static struct brw_reg deref( struct brw_vs_compile *c,
 			     struct brw_reg arg,
 			     GLint offset)
@@ -728,8 +744,6 @@ static struct brw_reg get_dst( struct brw_vs_compile *c,
 }
 
 
-
-
 static void emit_swz( struct brw_vs_compile *c, 
 		      struct brw_reg dst,
 		      struct prog_src_register src )
@@ -801,8 +815,8 @@ static void emit_swz( struct brw_vs_compile *c,
 }
 
 
-
-/* Post-vertex-program processing.  Send the results to the URB.
+/**
+ * Post-vertex-program processing.  Send the results to the URB.
  */
 static void emit_vertex_write( struct brw_vs_compile *c)
 {
@@ -817,7 +831,6 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	      get_reg(c, PROGRAM_INPUT, VERT_ATTRIB_EDGEFLAG));
    }
 
-
    /* Build ndc coords */
    if (!c->key.know_w_is_one) {
       ndc = get_tmp(c);
@@ -848,7 +861,6 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	 brw_AND(p, brw_writemask(header1, WRITEMASK_W), header1, brw_imm_ud(0x7ff<<8));
       }
 
-
       for (i = 0; i < c->key.nr_userclip; i++) {
 	 brw_set_conditionalmod(p, BRW_CONDITIONAL_L);
 	 brw_DP4(p, brw_null_reg(), pos, c->userplane[i]);
@@ -856,7 +868,6 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 	 brw_set_predicate_control(p, BRW_PREDICATE_NONE);
       }
 
-
       /* i965 clipping workaround: 
        * 1) Test for -ve rhw
        * 2) If set, 
@@ -888,14 +899,12 @@ static void emit_vertex_write( struct brw_vs_compile *c)
       brw_MOV(p, retype(brw_message_reg(1), BRW_REGISTER_TYPE_UD), brw_imm_ud(0));
    }
 
-
    /* Emit the (interleaved) headers for the two vertices - an 8-reg
     * of zeros followed by two sets of NDC coordinates:
     */
    brw_set_access_mode(p, BRW_ALIGN_1);
    brw_MOV(p, offset(m0, 2), ndc);
    brw_MOV(p, offset(m0, 3), pos);
-   
 
    brw_urb_WRITE(p, 
 		 brw_null_reg(), /* dest */
@@ -909,9 +918,9 @@ static void emit_vertex_write( struct brw_vs_compile *c)
 		 1, 		/* writes complete */
 		 0, 		/* urb destination offset */
 		 BRW_URB_SWIZZLE_INTERLEAVE);
-
 }
 
+
 static void 
 post_vs_emit( struct brw_vs_compile *c, struct brw_instruction *end_inst )
 {
@@ -1035,6 +1044,12 @@ void brw_vs_emit(struct brw_vs_compile *c )
       case OPCODE_DPH:
 	 brw_DPH(p, dst, args[0], args[1]);
 	 break;
+      case OPCODE_NRM3:
+	 emit_nrm(c, dst, args[0], 3);
+	 break;
+      case OPCODE_NRM4:
+	 emit_nrm(c, dst, args[0], 4);
+	 break;
       case OPCODE_DST:
 	 unalias2(c, dst, args[0], args[1], emit_dst_noalias); 
 	 break;
@@ -1102,7 +1117,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 break;
       case OPCODE_SGT:
          emit_sgt(p, dst, args[0], args[1]);
-        break;
+         break;
       case OPCODE_SLT:
 	 emit_slt(p, dst, args[0], args[1]);
 	 break;
@@ -1118,6 +1133,10 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	  */
 	 emit_swz(c, dst, inst->SrcReg[0] );
 	 break;
+      case OPCODE_TRUNC:
+         /* round toward zero */
+	 brw_RNDZ(p, dst, args[0]);
+	 break;
       case OPCODE_XPD:
 	 emit_xpd(p, dst, args[0], args[1]);
 	 break;
@@ -1136,7 +1155,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
          brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
          brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
          brw_set_predicate_control_flag_value(p, 0xff);
-        break;
+         break;
       case OPCODE_CAL:
 	 brw_set_access_mode(p, BRW_ALIGN_1);
 	 brw_ADD(p, deref_1d(stack_index, 0), brw_ip_reg(), brw_imm_d(3*16));
@@ -1145,7 +1164,7 @@ void brw_vs_emit(struct brw_vs_compile *c )
 			 get_addr_reg(stack_index), brw_imm_d(4));
 	 inst->Data = &p->store[p->nr_insn];
 	 brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-        break;
+         break;
       case OPCODE_RET:
 	 brw_ADD(p, get_addr_reg(stack_index),
 			 get_addr_reg(stack_index), brw_imm_d(-4));
@@ -1154,17 +1173,17 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 brw_set_access_mode(p, BRW_ALIGN_16);
       case OPCODE_END:	
          brw_ADD(p, brw_ip_reg(), brw_ip_reg(), brw_imm_d(1*16));
-        break;
+         break;
       case OPCODE_PRINT:
       case OPCODE_BGNSUB:
       case OPCODE_ENDSUB:
+         /* no-op instructions */
 	 break;
       default:
-	 _mesa_printf("Unsupported opcode %i (%s) in vertex shader\n",
-		      inst->Opcode, inst->Opcode < MAX_OPCODE ?
+	 _mesa_problem(NULL, "Unsupported opcode %i (%s) in vertex shader",
+                       inst->Opcode, inst->Opcode < MAX_OPCODE ?
 				    _mesa_opcode_string(inst->Opcode) :
 				    "unknown");
-	 break;
       }
 
       if ((inst->DstReg.File == PROGRAM_OUTPUT)
diff --git a/src/mesa/drivers/dri/i965/brw_wm.c b/src/mesa/drivers/dri/i965/brw_wm.c
index bad76793af7..5b4ee20ecb5 100644
--- a/src/mesa/drivers/dri/i965/brw_wm.c
+++ b/src/mesa/drivers/dri/i965/brw_wm.c
@@ -36,6 +36,7 @@
 #include "brw_state.h"
 
 
+/** Return number of src args for given instruction */
 GLuint brw_wm_nr_args( GLuint opcode )
 {
    switch (opcode) {
@@ -58,6 +59,8 @@ GLuint brw_wm_nr_args( GLuint opcode )
    case OPCODE_TXP:	
    case OPCODE_KIL:
    case OPCODE_LIT: 
+   case OPCODE_NRM3:
+   case OPCODE_NRM4:
    case WM_CINTERP: 
    case WM_WPOSXY: 
       return 1;
diff --git a/src/mesa/drivers/dri/i965/brw_wm_emit.c b/src/mesa/drivers/dri/i965/brw_wm_emit.c
index 58c78c4b2c5..b5050a3e40b 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_emit.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_emit.c
@@ -194,7 +194,7 @@ static void emit_linterp( struct brw_compile *p,
    interp[2] = brw_vec1_grf(nr+1, 0);
    interp[3] = brw_vec1_grf(nr+1, 4);
 
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
 	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
@@ -219,42 +219,40 @@ static void emit_pinterp( struct brw_compile *p,
    interp[2] = brw_vec1_grf(nr+1, 0);
    interp[3] = brw_vec1_grf(nr+1, 4);
 
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_LINE(p, brw_null_reg(), interp[i], deltas[0]);
 	 brw_MAC(p, dst[i], suboffset(interp[i],1), deltas[1]);
       }
    }
-   for(i = 0; i < 4; i++ ) {
+   for (i = 0; i < 4; i++) {
       if (mask & (1<<i)) {
 	 brw_MUL(p, dst[i], dst[i], w[3]);
       }
    }
 }
 
+
 static void emit_cinterp( struct brw_compile *p, 
 			 const struct brw_reg *dst,
 			 GLuint mask,
 			 const struct brw_reg *arg0 )
 {
-	struct brw_reg interp[4];
-	GLuint nr = arg0[0].nr;
-	GLuint i;
-
-	interp[0] = brw_vec1_grf(nr, 0);
-	interp[1] = brw_vec1_grf(nr, 4);
-	interp[2] = brw_vec1_grf(nr+1, 0);
-	interp[3] = brw_vec1_grf(nr+1, 4);
-
-	for(i = 0; i < 4; i++ ) {
-		if (mask & (1<<i)) {
-			brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
-		}
-	}
-}
-
+   struct brw_reg interp[4];
+   GLuint nr = arg0[0].nr;
+   GLuint i;
 
+   interp[0] = brw_vec1_grf(nr, 0);
+   interp[1] = brw_vec1_grf(nr, 4);
+   interp[2] = brw_vec1_grf(nr+1, 0);
+   interp[3] = brw_vec1_grf(nr+1, 4);
 
+   for (i = 0; i < 4; i++) {
+      if (mask & (1<<i)) {
+         brw_MOV(p, dst[i], suboffset(interp[i],3));	/* TODO: optimize away like other moves */
+      }
+   }
+}
 
 
 static void emit_alu1( struct brw_compile *p, 
@@ -280,6 +278,7 @@ static void emit_alu1( struct brw_compile *p,
       brw_set_saturate(p, 0);
 }
 
+
 static void emit_alu2( struct brw_compile *p, 
 		       struct brw_instruction *(*func)(struct brw_compile *, 
 						       struct brw_reg, 
@@ -351,6 +350,7 @@ static void emit_lrp( struct brw_compile *p,
       }
    }
 }
+
 static void emit_sop( struct brw_compile *p, 
 		      const struct brw_reg *dst,
 		      GLuint mask,
@@ -376,7 +376,7 @@ static void emit_slt( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_L, arg0, arg1);
 }
 
 static void emit_sle( struct brw_compile *p, 
@@ -385,7 +385,7 @@ static void emit_sle( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_LE, arg0, arg1);
 }
 
 static void emit_sgt( struct brw_compile *p, 
@@ -394,7 +394,7 @@ static void emit_sgt( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_G, arg0, arg1);
 }
 
 static void emit_sge( struct brw_compile *p, 
@@ -403,7 +403,7 @@ static void emit_sge( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_GE, arg0, arg1);
 }
 
 static void emit_seq( struct brw_compile *p, 
@@ -412,7 +412,7 @@ static void emit_seq( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_EQ, arg0, arg1);
 }
 
 static void emit_sne( struct brw_compile *p, 
@@ -421,7 +421,7 @@ static void emit_sne( struct brw_compile *p,
 		      const struct brw_reg *arg0,
 		      const struct brw_reg *arg1 )
 {
-	 emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
+   emit_sop(p, dst, mask, BRW_CONDITIONAL_NEQ, arg0, arg1);
 }
 
 static void emit_cmp( struct brw_compile *p, 
@@ -505,7 +505,7 @@ static void emit_dp3( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -525,7 +525,7 @@ static void emit_dp4( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -546,7 +546,7 @@ static void emit_dph( struct brw_compile *p,
 		      const struct brw_reg *arg1 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -592,7 +592,7 @@ static void emit_math1( struct brw_compile *p,
 			const struct brw_reg *arg0 )
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    //assert((mask & WRITEMASK_XYZW) == WRITEMASK_X ||
    //	  function == BRW_MATH_FUNCTION_SINCOS);
@@ -619,7 +619,7 @@ static void emit_math2( struct brw_compile *p,
 			const struct brw_reg *arg1)
 {
    if (!(mask & WRITEMASK_XYZW))
-      return; /* Do not emit dead code*/
+      return; /* Do not emit dead code */
 
    assert((mask & WRITEMASK_XYZW) == WRITEMASK_X);
 
@@ -760,7 +760,6 @@ static void emit_txb( struct brw_wm_compile *c,
    brw_MOV(p, brw_message_reg(8), arg[3]);
    msgLength = 9;
 
-
    brw_SAMPLE(p, 
 	      retype(vec16(dst[0]), BRW_REGISTER_TYPE_UW),
 	      1,
@@ -772,7 +771,6 @@ static void emit_txb( struct brw_wm_compile *c,
 	      8,		/* responseLength */
 	      msgLength,
 	      0);	
-
 }
 
 
@@ -823,7 +821,6 @@ static void emit_kil( struct brw_wm_compile *c,
    struct brw_reg r0uw = retype(brw_vec1_grf(0, 0), BRW_REGISTER_TYPE_UW);
    GLuint i;
    
-
    /* XXX - usually won't need 4 compares!
     */
    for (i = 0; i < 4; i++) {
@@ -836,6 +833,7 @@ static void emit_kil( struct brw_wm_compile *c,
    }
 }
 
+
 static void fire_fb_write( struct brw_wm_compile *c,
 			   GLuint base_reg,
 			   GLuint nr,
@@ -869,6 +867,7 @@ static void fire_fb_write( struct brw_wm_compile *c,
 		eot);
 }
 
+
 static void emit_aa( struct brw_wm_compile *c,
 		     struct brw_reg *arg1,
 		     GLuint reg )
@@ -962,7 +961,6 @@ static void emit_fb_write( struct brw_wm_compile *c,
       nr += 2;
    }
 
-
    if (!c->key.runtime_check_aads_emit) {
       if (c->key.aa_dest_stencil_reg)
 	 emit_aa(c, arg1, 2);
@@ -996,8 +994,6 @@ static void emit_fb_write( struct brw_wm_compile *c,
 }
 
 
-
-
 /* Post-fragment-program processing.  Send the results to the
  * framebuffer.
  */
@@ -1022,6 +1018,7 @@ static void emit_spill( struct brw_wm_compile *c,
 		   slot);
 }
 
+
 static void emit_unspill( struct brw_wm_compile *c,
 			  struct brw_reg reg,
 			  GLuint slot )
@@ -1047,7 +1044,6 @@ static void emit_unspill( struct brw_wm_compile *c,
 }
 
 
-
 /**
  * Retrieve upto 4 GEN4 register pairs for the given wm reg:
  */
@@ -1073,6 +1069,7 @@ static void get_argument_regs( struct brw_wm_compile *c,
    }
 }
 
+
 static void spill_values( struct brw_wm_compile *c,
 			  struct brw_wm_value *values,
 			  GLuint nr )
@@ -1085,7 +1082,6 @@ static void spill_values( struct brw_wm_compile *c,
 }
 
 
-
 /* Emit the fragment program instructions here.
  */
 void brw_wm_emit( struct brw_wm_compile *c )
@@ -1176,7 +1172,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_alu1(p, brw_RNDD, dst, dst_flags, args[0]);
 	 break;
 
-      case OPCODE_DP3:	/*  */
+      case OPCODE_DP3:
 	 emit_dp3(p, dst, dst_flags, args[0], args[1]);
 	 break;
 
@@ -1188,7 +1184,7 @@ void brw_wm_emit( struct brw_wm_compile *c )
 	 emit_dph(p, dst, dst_flags, args[0], args[1]);
 	 break;
 
-      case OPCODE_LRP:	/*  */
+      case OPCODE_LRP:
 	 emit_lrp(p, dst, dst_flags, args[0], args[1], args[2]);
 	 break;
 
@@ -1315,8 +1311,3 @@ void brw_wm_emit( struct brw_wm_compile *c )
 		      inst->dst[i]->spill_slot);
    }
 }
-
-
-
-
-
diff --git a/src/mesa/drivers/dri/i965/brw_wm_glsl.c b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
index baecfdcb799..d43e326f7d1 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_glsl.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_glsl.c
@@ -267,7 +267,7 @@ static void emit_trunc( struct brw_wm_compile *c,
 	    struct brw_reg src, dst;
 	    dst = get_dst_reg(c, inst, i, 1) ;
 	    src = get_src_reg(c, &inst->SrcReg[0], i, 1);
-	    brw_RNDD(p, dst, src);
+	    brw_RNDZ(p, dst, src);
 	}
     }
     brw_set_saturate(p, 0);
diff --git a/src/mesa/drivers/dri/intel/intel_buffers.c b/src/mesa/drivers/dri/intel/intel_buffers.c
index f8f009c6a30..4d036dee42e 100644
--- a/src/mesa/drivers/dri/intel/intel_buffers.c
+++ b/src/mesa/drivers/dri/intel/intel_buffers.c
@@ -181,7 +181,7 @@ intelUpdatePageFlipping(struct intel_context *intel,
    intel_fb->pf_current_page = (intel->sarea->pf_current_page >>
 				(intel_fb->pf_planes & 0x2)) & 0x3;
 
-   intel_fb->pf_num_pages = intel->intelScreen->third.handle ? 3 : 2;
+   intel_fb->pf_num_pages = 2;
 
    pf_active = pf_planes && (pf_planes & intel->sarea->pf_active) == pf_planes;
 
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 6c625b428c0..44b276a123e 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -412,6 +412,7 @@ static const struct dri_extension brw_extensions[] = {
    { "GL_EXT_shadow_funcs",               NULL },
    { "GL_EXT_texture_sRGB",		  NULL },
    { "GL_ATI_separate_stencil",           GL_ATI_separate_stencil_functions },
+   { "GL_ATI_texture_env_combine3",       NULL },
    { NULL,                                NULL }
 };
 
@@ -775,7 +776,6 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
 
       intel_region_release(&intel->front_region);
       intel_region_release(&intel->back_region);
-      intel_region_release(&intel->third_region);
       intel_region_release(&intel->depth_region);
 
       driDestroyOptionCache(&intel->optionCache);
@@ -825,12 +825,7 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 	    intel_renderbuffer_set_region(intel_fb->color_rb[1],
 					  intel->back_region);
          }
-#if 0
-         if (intel_fb->color_rb[2]) {
-	    intel_renderbuffer_set_region(intel_fb->color_rb[2],
-					  intel->third_region);
-         }
-#endif
+
          if (irbDepth) {
 	    intel_renderbuffer_set_region(irbDepth, intel->depth_region);
          }
@@ -867,7 +862,7 @@ intelMakeCurrent(__DRIcontextPrivate * driContextPriv,
 	       driDrawableInitVBlank(driDrawPriv);
 	       intel_fb->vbl_waited = driDrawPriv->vblSeq;
 
-	       for (i = 0; i < (intel->intelScreen->third.handle ? 3 : 2); i++) {
+	       for (i = 0; i < 2; i++) {
 		  if (intel_fb->color_rb[i])
 		     intel_fb->color_rb[i]->vbl_pending = driDrawPriv->vblSeq;
 	       }
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index ee43ed7e833..048286c196b 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -157,6 +157,19 @@ struct intel_context
       void (*debug_batch)(struct intel_context *intel);
    } vtbl;
 
+   struct {
+      struct gl_fragment_program *bitmap_fp;
+      struct gl_vertex_program *passthrough_vp;
+
+      struct gl_fragment_program *saved_fp;
+      GLboolean saved_fp_enable;
+      struct gl_vertex_program *saved_vp;
+      GLboolean saved_vp_enable;
+
+      GLint saved_vp_x, saved_vp_y;
+      GLsizei saved_vp_width, saved_vp_height;
+   } meta;
+
    GLint refcount;
    GLuint Fallback;
    GLuint NewGLState;
@@ -166,7 +179,6 @@ struct intel_context
 
    struct intel_region *front_region;
    struct intel_region *back_region;
-   struct intel_region *third_region;
    struct intel_region *depth_region;
 
    /**
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
index 5702ad9bb57..cf2f32d3845 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -27,7 +27,12 @@
 
 #include "main/enums.h"
 #include "main/state.h"
+#include "main/context.h"
+#include "main/enable.h"
+#include "main/matrix.h"
 #include "swrast/swrast.h"
+#include "shader/arbprogram.h"
+#include "shader/program.h"
 
 #include "intel_context.h"
 #include "intel_pixel.h"
@@ -167,6 +172,159 @@ intel_check_blit_format(struct intel_region * region,
    return GL_FALSE;
 }
 
+void
+intel_meta_set_passthrough_transform(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   intel->meta.saved_vp_x = ctx->Viewport.X;
+   intel->meta.saved_vp_y = ctx->Viewport.Y;
+   intel->meta.saved_vp_width = ctx->Viewport.Width;
+   intel->meta.saved_vp_height = ctx->Viewport.Height;
+
+   _mesa_Viewport(0, 0, ctx->DrawBuffer->Width, ctx->DrawBuffer->Height);
+
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
+
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PushMatrix();
+   _mesa_LoadIdentity();
+}
+
+void
+intel_meta_restore_transform(struct intel_context *intel)
+{
+   _mesa_MatrixMode(GL_PROJECTION);
+   _mesa_PopMatrix();
+   _mesa_MatrixMode(GL_MODELVIEW);
+   _mesa_PopMatrix();
+
+   _mesa_Viewport(intel->meta.saved_vp_x, intel->meta.saved_vp_y,
+		  intel->meta.saved_vp_width, intel->meta.saved_vp_height);
+}
+
+/**
+ * Set up a vertex program to pass through the position and first texcoord
+ * for pixel path.
+ */
+void
+intel_meta_set_passthrough_vertex_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+   static const char *vp =
+      "!!ARBvp1.0\n"
+      "TEMP vertexClip;\n"
+      "DP4 vertexClip.x, state.matrix.mvp.row[0], vertex.position;\n"
+      "DP4 vertexClip.y, state.matrix.mvp.row[1], vertex.position;\n"
+      "DP4 vertexClip.z, state.matrix.mvp.row[2], vertex.position;\n"
+      "DP4 vertexClip.w, state.matrix.mvp.row[3], vertex.position;\n"
+      "MOV result.position, vertexClip;\n"
+      "MOV result.texcoord[0], vertex.texcoord[0];\n"
+      "MOV result.color, vertex.color;\n"
+      "END\n";
+
+   assert(intel->meta.saved_vp == NULL);
+
+   _mesa_reference_vertprog(ctx, &intel->meta.saved_vp,
+			    ctx->VertexProgram.Current);
+   if (intel->meta.passthrough_vp == NULL) {
+      GLuint prog_name;
+      _mesa_GenPrograms(1, &prog_name);
+      _mesa_BindProgram(GL_VERTEX_PROGRAM_ARB, prog_name);
+      _mesa_ProgramStringARB(GL_VERTEX_PROGRAM_ARB,
+			     GL_PROGRAM_FORMAT_ASCII_ARB,
+			     strlen(vp), (const GLubyte *)vp);
+      _mesa_reference_vertprog(ctx, &intel->meta.passthrough_vp,
+			       ctx->VertexProgram.Current);
+      _mesa_DeletePrograms(1, &prog_name);
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current,
+			    intel->meta.passthrough_vp);
+   ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
+			   &intel->meta.passthrough_vp->Base);
+
+   intel->meta.saved_vp_enable = ctx->VertexProgram.Enabled;
+   _mesa_Enable(GL_VERTEX_PROGRAM_ARB);
+}
+
+/**
+ * Restores the previous vertex program after
+ * intel_meta_set_passthrough_vertex_program()
+ */
+void
+intel_meta_restore_vertex_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_vertprog(ctx, &ctx->VertexProgram.Current,
+			    intel->meta.saved_vp);
+   _mesa_reference_vertprog(ctx, &intel->meta.saved_vp, NULL);
+   ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
+			   &ctx->VertexProgram.Current->Base);
+
+   if (!intel->meta.saved_vp_enable)
+      _mesa_Disable(GL_VERTEX_PROGRAM_ARB);
+}
+
+/**
+ * Binds the given program string to GL_FRAGMENT_PROGRAM_ARB, caching the
+ * program object.
+ */
+void
+intel_meta_set_fragment_program(struct intel_context *intel,
+				struct gl_fragment_program **prog,
+				const char *prog_string)
+{
+   GLcontext *ctx = &intel->ctx;
+   assert(intel->meta.saved_fp == NULL);
+
+   _mesa_reference_fragprog(ctx, &intel->meta.saved_fp,
+			    ctx->FragmentProgram.Current);
+   if (*prog == NULL) {
+      GLuint prog_name;
+      _mesa_GenPrograms(1, &prog_name);
+      _mesa_BindProgram(GL_FRAGMENT_PROGRAM_ARB, prog_name);
+      _mesa_ProgramStringARB(GL_FRAGMENT_PROGRAM_ARB,
+			     GL_PROGRAM_FORMAT_ASCII_ARB,
+			     strlen(prog_string), (const GLubyte *)prog_string);
+      _mesa_reference_fragprog(ctx, prog, ctx->FragmentProgram.Current);
+      /* Note that DeletePrograms unbinds the program on us */
+      _mesa_DeletePrograms(1, &prog_name);
+   }
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current, *prog);
+   ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, &((*prog)->Base));
+
+   intel->meta.saved_fp_enable = ctx->FragmentProgram.Enabled;
+   _mesa_Enable(GL_FRAGMENT_PROGRAM_ARB);
+}
+
+/**
+ * Restores the previous fragment program after
+ * intel_meta_set_fragment_program()
+ */
+void
+intel_meta_restore_fragment_program(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   FLUSH_VERTICES(ctx, _NEW_PROGRAM);
+   _mesa_reference_fragprog(ctx, &ctx->FragmentProgram.Current,
+			    intel->meta.saved_fp);
+   _mesa_reference_fragprog(ctx, &intel->meta.saved_fp, NULL);
+   ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB,
+			   &ctx->FragmentProgram.Current->Base);
+
+   if (!intel->meta.saved_fp_enable)
+      _mesa_Disable(GL_FRAGMENT_PROGRAM_ARB);
+}
 
 void
 intelInitPixelFuncs(struct dd_function_table *functions)
@@ -181,3 +339,13 @@ intelInitPixelFuncs(struct dd_function_table *functions)
 #endif
    }
 }
+
+void
+intel_free_pixel_state(struct intel_context *intel)
+{
+   GLcontext *ctx = &intel->ctx;
+
+   _mesa_reference_vertprog(ctx, &intel->meta.passthrough_vp, NULL);
+   _mesa_reference_fragprog(ctx, &intel->meta.bitmap_fp, NULL);
+}
+
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.h b/src/mesa/drivers/dri/intel/intel_pixel.h
index 6fa6effe835..76b8781316b 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.h
+++ b/src/mesa/drivers/dri/intel/intel_pixel.h
@@ -31,6 +31,15 @@
 #include "main/mtypes.h"
 
 void intelInitPixelFuncs(struct dd_function_table *functions);
+void intel_meta_set_passthrough_transform(struct intel_context *intel);
+void intel_meta_restore_transform(struct intel_context *intel);
+void intel_meta_set_passthrough_vertex_program(struct intel_context *intel);
+void intel_meta_restore_vertex_program(struct intel_context *intel);
+void intel_meta_set_fragment_program(struct intel_context *intel,
+				     struct gl_fragment_program **prog,
+				     const char *prog_string);
+void intel_meta_restore_fragment_program(struct intel_context *intel);
+void intel_free_pixel_state(struct intel_context *intel);
 
 GLboolean intel_check_blit_fragment_ops(GLcontext * ctx,
 					GLboolean src_alpha_is_one);
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index fb1a051cdc8..1d7f15f10a5 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -32,7 +32,18 @@
 #include "main/mtypes.h"
 #include "main/macros.h"
 #include "main/bufferobj.h"
+#include "main/pixelstore.h"
 #include "main/state.h"
+#include "main/teximage.h"
+#include "main/texenv.h"
+#include "main/texobj.h"
+#include "main/texstate.h"
+#include "main/texparam.h"
+#include "main/varray.h"
+#include "main/attrib.h"
+#include "main/enable.h"
+#include "shader/arbprogram.h"
+#include "glapi/dispatch.h"
 #include "swrast/swrast.h"
 
 #include "intel_screen.h"
@@ -87,6 +98,11 @@ static GLboolean test_bit( const GLubyte *src,
    return (src[bit/8] & (1<<(bit % 8))) ? 1 : 0;
 }
 
+static GLboolean test_msb_bit(const GLubyte *src, GLuint bit)
+{
+   return (src[bit/8] & (1<<(7 - (bit % 8)))) ? 1 : 0;
+}
+
 static void set_bit( GLubyte *dest,
 			  GLuint bit )
 {
@@ -244,8 +260,8 @@ do_blit_bitmap( GLcontext *ctx,
 	 /* Clip to drawable cliprect */
          if (!_mesa_clip_to_region(cliprects[i].x1,
 				   cliprects[i].y1,
-				   cliprects[i].x2 - cliprects[i].x1,
-				   cliprects[i].y2 - cliprects[i].y1,
+				   cliprects[i].x2,
+				   cliprects[i].y2,
 				   &box_x, &box_y, &box_w, &box_h))
 	    continue;
 
@@ -317,9 +333,174 @@ out:
    return GL_TRUE;
 }
 
+static GLboolean
+intel_texture_bitmap(GLcontext * ctx,
+		     GLint dst_x, GLint dst_y,
+		     GLsizei width, GLsizei height,
+		     const struct gl_pixelstore_attrib *unpack,
+		     const GLubyte *bitmap)
+{
+   struct intel_context *intel = intel_context(ctx);
+   static const char *fp =
+      "!!ARBfp1.0\n"
+      "TEMP val;\n"
+      "PARAM color=program.local[0];\n"
+      "TEX val, fragment.texcoord[0], texture[0], 2D;\n"
+      "ADD val, val.wwww, {-.5, -.5, -.5, -.5};\n"
+      "KIL val;\n"
+      "MOV result.color, color;\n"
+      "END\n";
+   GLuint texname;
+   GLfloat vertices[4][4];
+   GLfloat texcoords[4][2];
+   GLint old_active_texture;
+   GLubyte *unpacked_bitmap;
+   GLubyte *a8_bitmap;
+   int x, y;
+
+   /* We need a fragment program for the KIL effect */
+   if (!ctx->Extensions.ARB_fragment_program ||
+       !ctx->Extensions.ARB_vertex_program) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr,
+		 "glBitmap fallback: No fragment/vertex program support\n");
+      return GL_FALSE;
+   }
+
+   /* We're going to mess with texturing with no regard to existing texture
+    * state, so if there is some set up we have to bail.
+    */
+   if (ctx->Texture._EnabledUnits != 0) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: texturing enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Can't do textured DrawPixels with a fragment program, unless we were
+    * to generate a new program that sampled our texture and put the results
+    * in the fragment color before the user's program started.
+    */
+   if (ctx->FragmentProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: fragment program enabled\n");
+      return GL_FALSE;
+   }
+
+   if (ctx->VertexProgram.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: vertex program enabled\n");
+      return GL_FALSE;
+   }
+
+   /* Check that we can load in a texture this big. */
+   if (width > (1 << (ctx->Const.MaxTextureLevels - 1)) ||
+       height > (1 << (ctx->Const.MaxTextureLevels - 1))) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap fallback: bitmap too large (%dx%d)\n",
+		 width, height);
+      return GL_FALSE;
+   }
 
+   /* Convert the A1 bitmap to an A8 format suitable for glTexImage */
+   if (unpack->BufferObj->Name) {
+      bitmap = map_pbo(ctx, width, height, unpack, bitmap);
+      if (bitmap == NULL)
+	 return GL_TRUE;	/* even though this is an error, we're done */
+   }
+   unpacked_bitmap = _mesa_unpack_bitmap(width, height, bitmap,
+					 unpack);
+   a8_bitmap = _mesa_calloc(width * height);
+   for (y = 0; y < height; y++) {
+      for (x = 0; x < width; x++) {
+	 if (test_msb_bit(unpacked_bitmap, ALIGN(width, 8) * y + x))
+	    a8_bitmap[y * width + x] = 0xff;
+      }
+   }
+   _mesa_free(unpacked_bitmap);
+   if (unpack->BufferObj->Name) {
+      /* done with PBO so unmap it now */
+      ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
+                              unpack->BufferObj);
+   }
 
+   /* Save GL state before we start setting up our drawing */
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_CURRENT_BIT |
+		    GL_VIEWPORT_BIT);
+   _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT |
+			  GL_CLIENT_PIXEL_STORE_BIT);
+   old_active_texture = ctx->Texture.CurrentUnit;
+
+   _mesa_Disable(GL_POLYGON_STIPPLE);
+
+   /* Upload our bitmap data to an alpha texture */
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB);
+   _mesa_Enable(GL_TEXTURE_2D);
+   _mesa_GenTextures(1, &texname);
+   _mesa_BindTexture(GL_TEXTURE_2D, texname);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+   _mesa_TexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+
+   _mesa_PixelStorei(GL_UNPACK_SWAP_BYTES, GL_FALSE);
+   _mesa_PixelStorei(GL_UNPACK_LSB_FIRST, GL_FALSE);
+   _mesa_PixelStorei(GL_UNPACK_ROW_LENGTH, 0);
+   _mesa_PixelStorei(GL_UNPACK_SKIP_PIXELS, 0);
+   _mesa_PixelStorei(GL_UNPACK_SKIP_ROWS, 0);
+   _mesa_PixelStorei(GL_UNPACK_ALIGNMENT, 1);
+   _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_ALPHA, width, height, 0,
+		    GL_ALPHA, GL_UNSIGNED_BYTE, a8_bitmap);
+   _mesa_free(a8_bitmap);
+
+   intel_meta_set_fragment_program(intel, &intel->meta.bitmap_fp, fp);
+   _mesa_ProgramLocalParameter4fvARB(GL_FRAGMENT_PROGRAM_ARB, 0,
+				     ctx->Current.RasterColor);
+   intel_meta_set_passthrough_vertex_program(intel);
+   intel_meta_set_passthrough_transform(intel);
+
+   vertices[0][0] = dst_x;
+   vertices[0][1] = dst_y;
+   vertices[0][2] = ctx->Current.RasterPos[2];
+   vertices[0][3] = 1.0;
+   vertices[1][0] = dst_x + width;
+   vertices[1][1] = dst_y;
+   vertices[1][2] = ctx->Current.RasterPos[2];
+   vertices[1][3] = 1.0;
+   vertices[2][0] = dst_x + width;
+   vertices[2][1] = dst_y + height;
+   vertices[2][2] = ctx->Current.RasterPos[2];
+   vertices[2][3] = 1.0;
+   vertices[3][0] = dst_x;
+   vertices[3][1] = dst_y + height;
+   vertices[3][2] = ctx->Current.RasterPos[2];
+   vertices[3][3] = 1.0;
+
+   texcoords[0][0] = 0.0;
+   texcoords[0][1] = 0.0;
+   texcoords[1][0] = 1.0;
+   texcoords[1][1] = 0.0;
+   texcoords[2][0] = 1.0;
+   texcoords[2][1] = 1.0;
+   texcoords[3][0] = 0.0;
+   texcoords[3][1] = 1.0;
+
+   _mesa_VertexPointer(4, GL_FLOAT, 4 * sizeof(GLfloat), &vertices);
+   _mesa_TexCoordPointer(2, GL_FLOAT, 2 * sizeof(GLfloat), &texcoords);
+   _mesa_Enable(GL_VERTEX_ARRAY);
+   _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
+   CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
+
+   intel_meta_restore_transform(intel);
+   intel_meta_restore_fragment_program(intel);
+   intel_meta_restore_vertex_program(intel);
+
+   _mesa_PopClientAttrib();
+   _mesa_Disable(GL_TEXTURE_2D); /* asserted that it was disabled at entry */
+   _mesa_ActiveTextureARB(GL_TEXTURE0_ARB + old_active_texture);
+   _mesa_PopAttrib();
+
+   _mesa_DeleteTextures(1, &texname);
 
+   return GL_TRUE;
+}
 
 /* There are a large number of possible ways to implement bitmap on
  * this hardware, most of them have some sort of drawback.  Here are a
@@ -352,6 +533,10 @@ intelBitmap(GLcontext * ctx,
                           unpack, pixels))
       return;
 
+   if (intel_texture_bitmap(ctx, x, y, width, height,
+			    unpack, pixels))
+      return;
+
    if (INTEL_DEBUG & DEBUG_PIXEL)
       _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
 
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_draw.c b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
index 8ebbc95a1d0..0d66935ad27 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_draw.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_draw.c
@@ -36,7 +36,6 @@
 #include "main/texobj.h"
 #include "main/texstate.h"
 #include "main/texparam.h"
-#include "main/matrix.h"
 #include "main/varray.h"
 #include "main/attrib.h"
 #include "main/enable.h"
@@ -68,6 +67,7 @@ intel_texture_drawpixels(GLcontext * ctx,
 			 const struct gl_pixelstore_attrib *unpack,
 			 const GLvoid *pixels)
 {
+   struct intel_context *intel = intel_context(ctx);
    GLuint texname;
    GLfloat vertices[4][4];
    GLfloat texcoords[4][2];
@@ -117,7 +117,7 @@ intel_texture_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT |
 		    GL_CURRENT_BIT);
    _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
 
@@ -138,14 +138,7 @@ intel_texture_drawpixels(GLcontext * ctx,
    _mesa_TexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, format,
 		    type, pixels);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
-   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
-
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
+   intel_meta_set_passthrough_transform(intel);
 
    /* Create the vertex buffer based on the current raster pos.  The x and y
     * we're handed are ctx->Current.RasterPos[0,1] rounded to integers.
@@ -184,10 +177,7 @@ intel_texture_drawpixels(GLcontext * ctx,
    _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
    CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PopMatrix();
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PopMatrix();
+   intel_meta_restore_transform(intel);
    _mesa_PopClientAttrib();
    _mesa_PopAttrib();
 
@@ -205,6 +195,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
 			 const struct gl_pixelstore_attrib *unpack,
 			 const GLvoid *pixels)
 {
+   struct intel_context *intel = intel_context(ctx);
    GLuint texname, rb_name, fb_name, old_fb_name;
    GLfloat vertices[4][2];
    GLfloat texcoords[4][2];
@@ -267,7 +258,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
       return GL_FALSE;
    }
 
-   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TRANSFORM_BIT | GL_TEXTURE_BIT |
+   _mesa_PushAttrib(GL_ENABLE_BIT | GL_TEXTURE_BIT |
 		    GL_CURRENT_BIT | GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
    _mesa_PushClientAttrib(GL_CLIENT_VERTEX_ARRAY_BIT);
    old_fb_name = ctx->DrawBuffer->Name;
@@ -335,14 +326,7 @@ intel_stencil_drawpixels(GLcontext * ctx,
    ctx->Unpack = old_unpack;
    _mesa_free(stencil_pixels);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
-   _mesa_Ortho(0, ctx->DrawBuffer->Width, 0, ctx->DrawBuffer->Height, 1, -1);
-
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PushMatrix();
-   _mesa_LoadIdentity();
+   intel_meta_set_passthrough_transform(intel);
 
    vertices[0][0] = x;
    vertices[0][1] = y;
@@ -368,12 +352,10 @@ intel_stencil_drawpixels(GLcontext * ctx,
    _mesa_Enable(GL_TEXTURE_COORD_ARRAY);
    CALL_DrawArrays(ctx->Exec, (GL_TRIANGLE_FAN, 0, 4));
 
+   intel_meta_restore_transform(intel);
+
    _mesa_BindFramebufferEXT(GL_FRAMEBUFFER_EXT, old_fb_name);
 
-   _mesa_MatrixMode(GL_PROJECTION);
-   _mesa_PopMatrix();
-   _mesa_MatrixMode(GL_MODELVIEW);
-   _mesa_PopMatrix();
    _mesa_PopClientAttrib();
    _mesa_PopAttrib();
 
diff --git a/src/mesa/drivers/dri/intel/intel_regions.c b/src/mesa/drivers/dri/intel/intel_regions.c
index 8dbcc3050ee..51ce32a9679 100644
--- a/src/mesa/drivers/dri/intel/intel_regions.c
+++ b/src/mesa/drivers/dri/intel/intel_regions.c
@@ -550,15 +550,6 @@ intel_recreate_static_regions(struct intel_context *intel)
 			    intel->back_region,
 			    &intelScreen->back);
 
-#ifdef I915
-   if (intelScreen->third.handle) {
-      intel->third_region =
-	 intel_recreate_static(intel, "third",
-			       intel->third_region,
-			       &intelScreen->third);
-   }
-#endif /* I915 */
-
    /* Still assumes front.cpp == depth.cpp.  We can kill this when we move to
     * private buffers.
     */
diff --git a/src/mesa/drivers/dri/intel/intel_screen.c b/src/mesa/drivers/dri/intel/intel_screen.c
index 61b55b97b51..fc4e82b56c9 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.c
+++ b/src/mesa/drivers/dri/intel/intel_screen.c
@@ -177,13 +177,6 @@ intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
    intelScreen->back.size = sarea->back_size;
    intelScreen->back.tiled = sarea->back_tiled;
 
-   if (intelScreen->driScrnPriv->ddx_version.minor >= 8) {
-      intelScreen->third.offset = sarea->third_offset;
-      intelScreen->third.handle = sarea->third_handle;
-      intelScreen->third.size = sarea->third_size;
-      intelScreen->third.tiled = sarea->third_tiled;
-   }
-
    intelScreen->depth.offset = sarea->depth_offset;
    intelScreen->depth.handle = sarea->depth_handle;
    intelScreen->depth.size = sarea->depth_size;
@@ -192,12 +185,10 @@ intelUpdateScreenFromSAREA(intelScreenPrivate * intelScreen,
    if (intelScreen->driScrnPriv->ddx_version.minor >= 9) {
       intelScreen->front.bo_handle = sarea->front_bo_handle;
       intelScreen->back.bo_handle = sarea->back_bo_handle;
-      intelScreen->third.bo_handle = sarea->third_bo_handle;
       intelScreen->depth.bo_handle = sarea->depth_bo_handle;
    } else {
       intelScreen->front.bo_handle = -1;
       intelScreen->back.bo_handle = -1;
-      intelScreen->third.bo_handle = -1;
       intelScreen->depth.bo_handle = -1;
    }
 
@@ -353,12 +344,6 @@ intelCreateBuffer(__DRIscreenPrivate * driScrnPriv,
          _mesa_add_renderbuffer(&intel_fb->Base, BUFFER_BACK_LEFT,
 				&intel_fb->color_rb[1]->Base);
 
-	 if (screen->third.handle) {
-	    struct gl_renderbuffer *tmp_rb = NULL;
-
-	    intel_fb->color_rb[2] = intel_create_renderbuffer(rgbFormat);
-	    _mesa_reference_renderbuffer(&tmp_rb, &intel_fb->color_rb[2]->Base);
-	 }
       }
 
       if (mesaVis->depthBits == 24) {
diff --git a/src/mesa/drivers/dri/intel/intel_screen.h b/src/mesa/drivers/dri/intel/intel_screen.h
index 91f0d6d1ae8..cf5359baaeb 100644
--- a/src/mesa/drivers/dri/intel/intel_screen.h
+++ b/src/mesa/drivers/dri/intel/intel_screen.h
@@ -56,7 +56,6 @@ typedef struct
 {
    intelRegion front;
    intelRegion back;
-   intelRegion third;
    intelRegion depth;
    intelRegion tex;
 
diff --git a/src/mesa/drivers/dri/r300/r300_reg.h b/src/mesa/drivers/dri/r300/r300_reg.h
index 778db96cc1f..7c6485ef60b 100644
--- a/src/mesa/drivers/dri/r300/r300_reg.h
+++ b/src/mesa/drivers/dri/r300/r300_reg.h
@@ -64,7 +64,7 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define R300_SE_VPORT_ZSCALE                0x1DA8
 #define R300_SE_VPORT_ZOFFSET               0x1DAC
 
-
+#define R300_VAP_PORT_IDX0		    0x2040
 /*
  * Vertex Array Processing (VAP) Control
  */
@@ -3201,9 +3201,9 @@ enum {
 #define R300_PACKET3_3D_LOAD_VBPNTR         0x00002F00
 
 #define R300_PACKET3_INDX_BUFFER            0x00003300
-#    define R300_EB_UNK1_SHIFT                      24
-#    define R300_EB_UNK1                    (0x80<<24)
-#    define R300_EB_UNK2                        0x0810
+#    define R300_INDX_BUFFER_DST_SHIFT          0
+#    define R300_INDX_BUFFER_SKIP_SHIFT         16
+#    define R300_INDX_BUFFER_ONE_REG_WR		(1<<31)
 
 /* Same as R300_PACKET3_3D_DRAW_VBUF but without VAP_VTX_FMT */
 #define R300_PACKET3_3D_DRAW_VBUF_2         0x00003400
diff --git a/src/mesa/drivers/dri/r300/r300_render.c b/src/mesa/drivers/dri/r300/r300_render.c
index 292f87a2b1b..f9266e44c13 100644
--- a/src/mesa/drivers/dri/r300/r300_render.c
+++ b/src/mesa/drivers/dri/r300/r300_render.c
@@ -208,7 +208,8 @@ static void r300FireEB(r300ContextPtr rmesa, unsigned long addr,
 	e32(R300_VAP_VF_CNTL__PRIM_WALK_INDICES | (vertex_count << 16) | type | R300_VAP_VF_CNTL__INDEX_SIZE_32bit);
 
 	start_packet3(CP_PACKET3(R300_PACKET3_INDX_BUFFER, 2), 2);
-	e32(R300_EB_UNK1 | (0 << 16) | R300_EB_UNK2);
+	e32(R300_INDX_BUFFER_ONE_REG_WR | (0 << R300_INDX_BUFFER_SKIP_SHIFT) |
+	    (R300_VAP_PORT_IDX0 >> 2));
 	e32(addr);
 	e32(vertex_count);
 }
diff --git a/src/mesa/main/arrayobj.c b/src/mesa/main/arrayobj.c
index 14612393176..f3f482f8c87 100644
--- a/src/mesa/main/arrayobj.c
+++ b/src/mesa/main/arrayobj.c
@@ -142,7 +142,7 @@ _mesa_initialize_array_object( GLcontext *ctx,
    obj->Index.StrideB = 0;
    obj->Index.Ptr = NULL;
    obj->Index.Enabled = GL_FALSE;
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       obj->TexCoord[i].Size = 4;
       obj->TexCoord[i].Type = GL_FLOAT;
       obj->TexCoord[i].Stride = 0;
@@ -181,7 +181,7 @@ _mesa_initialize_array_object( GLcontext *ctx,
    obj->SecondaryColor.BufferObj = ctx->Array.NullBufferObj;
    obj->FogCoord.BufferObj = ctx->Array.NullBufferObj;
    obj->Index.BufferObj = ctx->Array.NullBufferObj;
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       obj->TexCoord[i].BufferObj = ctx->Array.NullBufferObj;
    }
    obj->EdgeFlag.BufferObj = ctx->Array.NullBufferObj;
@@ -335,7 +335,7 @@ _mesa_DeleteVertexArraysAPPLE(GLsizei n, const GLuint *ids)
 	 unbind_buffer_object( ctx, obj->SecondaryColor.BufferObj );
 	 unbind_buffer_object( ctx, obj->FogCoord.BufferObj );
 	 unbind_buffer_object( ctx, obj->Index.BufferObj );
-	 for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+	 for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
 	    unbind_buffer_object( ctx, obj->TexCoord[i].BufferObj );
 	 }
 	 unbind_buffer_object( ctx, obj->EdgeFlag.BufferObj );
diff --git a/src/mesa/main/attrib.c b/src/mesa/main/attrib.c
index dc85da2518f..825c841ee21 100644
--- a/src/mesa/main/attrib.c
+++ b/src/mesa/main/attrib.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.2
+ * Version:  7.3
  *
  * Copyright (C) 1999-2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -224,7 +225,7 @@ _mesa_PushAttrib(GLbitfield mask)
       attr->SampleAlphaToOne = ctx->Multisample.SampleAlphaToOne;
       attr->SampleCoverage = ctx->Multisample.SampleCoverage;
       attr->SampleCoverageInvert = ctx->Multisample.SampleCoverageInvert;
-      for (i=0; i<MAX_TEXTURE_UNITS; i++) {
+      for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
          attr->Texture[i] = ctx->Texture.Unit[i].Enabled;
          attr->TexGen[i] = ctx->Texture.Unit[i].TexGenEnabled;
          attr->TextureColorTable[i] = ctx->Texture.Unit[i].ColorTableEnabled;
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index 190e6ab5641..016543da01f 100644
--- a/src/mesa/main/bufferobj.c
+++ b/src/mesa/main/bufferobj.c
@@ -808,7 +808,7 @@ _mesa_DeleteBuffersARB(GLsizei n, const GLuint *ids)
          unbind(ctx, &ctx->Array.ArrayObj->FogCoord.BufferObj, bufObj);
          unbind(ctx, &ctx->Array.ArrayObj->Index.BufferObj, bufObj);
          unbind(ctx, &ctx->Array.ArrayObj->EdgeFlag.BufferObj, bufObj);
-         for (j = 0; j < MAX_TEXTURE_UNITS; j++) {
+         for (j = 0; j < MAX_TEXTURE_COORD_UNITS; j++) {
             unbind(ctx, &ctx->Array.ArrayObj->TexCoord[j].BufferObj, bufObj);
          }
          for (j = 0; j < VERT_ATTRIB_MAX; j++) {
diff --git a/src/mesa/main/config.h b/src/mesa/main/config.h
index e29964a1e86..c3feffda986 100644
--- a/src/mesa/main/config.h
+++ b/src/mesa/main/config.h
@@ -1,13 +1,9 @@
-/**
- * \file config.h
- * Tunable configuration parameters.
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2008  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -27,6 +23,10 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+/**
+ * \file config.h
+ * Tunable configuration parameters.
+ */
 
 #ifndef MESA_CONFIG_H_INCLUDED
 #define MESA_CONFIG_H_INCLUDED
@@ -115,27 +115,28 @@
 /** Maximum number of layers in a 1D or 2D array texture - GL_MESA_texture_array */
 #define MAX_ARRAY_TEXTURE_LAYERS 64
 
-/** Number of texture units - GL_ARB_multitexture
- * This needs to be the larger of MAX_TEXTURE_COORD_UNITS and
- * MAX_TEXTURE_IMAGE_UNITS seen below, since MAX_TEXTURE_UNITS is used
- * to dimension some arrays that store both coord and image data.
-*/
-#define MAX_TEXTURE_UNITS 8
-
-/*@}*/
+/**
+ * Max number of texture coordinate units.  This mainly just applies to
+ * the fixed-function vertex code.  This will be difficult to raise above
+ * eight because of various vertex attribute bitvectors.
+ */
+#define MAX_TEXTURE_COORD_UNITS 8
 
+/**
+ * Max number of texture image units.  Also determines number of texture
+ * samplers in shaders.
+ */
+#define MAX_TEXTURE_IMAGE_UNITS 16
 
 /**
- * \name Separate numbers of texture coordinates and texture image units.
- *
- * These values will eventually replace most instances of MAX_TEXTURE_UNITS.
- * We should always have MAX_TEXTURE_COORD_UNITS <= MAX_TEXTURE_IMAGE_UNITS.
- * And, GL_MAX_TEXTURE_UNITS <= MAX_TEXTURE_COORD_UNITS.
+ * Larger of MAX_TEXTURE_COORD_UNITS and MAX_TEXTURE_IMAGE_UNITS.
+ * This value is only used for dimensioning arrays.
+ * Either MAX_TEXTURE_COORD_UNITS or MAX_TEXTURE_IMAGE_UNITS (or the
+ * corresponding ctx->Const.MaxTextureCoord/ImageUnits fields) should be
+ * used almost everywhere else.
  */
-/*@{*/
-#define MAX_TEXTURE_COORD_UNITS 8
-#define MAX_TEXTURE_IMAGE_UNITS 8
-/*@}*/
+#define MAX_TEXTURE_UNITS ((MAX_TEXTURE_COORD_UNITS > MAX_TEXTURE_IMAGE_UNITS) ? MAX_TEXTURE_COORD_UNITS : MAX_TEXTURE_IMAGE_UNITS)
+
 
 /** 
  * Maximum viewport/image width. Must accomodate all texture sizes too. 
@@ -181,16 +182,16 @@
 /** For any program target/extension */
 /*@{*/
 #define MAX_PROGRAM_INSTRUCTIONS  (16 * 1024)
-#define MAX_PROGRAM_LOCAL_PARAMS 128 /* KW: power of two */
+#define MAX_PROGRAM_LOCAL_PARAMS 256 /**< per-program constants (power of two) */
 #define MAX_PROGRAM_ENV_PARAMS 128
 #define MAX_PROGRAM_MATRICES 8
 #define MAX_PROGRAM_MATRIX_STACK_DEPTH 4
 #define MAX_PROGRAM_CALL_DEPTH 8
 #define MAX_PROGRAM_TEMPS 128
 #define MAX_PROGRAM_ADDRESS_REGS 2
-#define MAX_UNIFORMS 128   /**< number of float components */
+#define MAX_UNIFORMS 256   /**< number of vec4 uniforms */
 #define MAX_VARYING 8      /**< number of float[4] vectors */
-#define MAX_SAMPLERS 8
+#define MAX_SAMPLERS MAX_TEXTURE_IMAGE_UNITS
 #define MAX_PROGRAM_INPUTS 32
 #define MAX_PROGRAM_OUTPUTS 32
 /*@}*/
@@ -218,8 +219,8 @@
 /** For GL_ARB_vertex_shader */
 /*@{*/
 #define MAX_VERTEX_ATTRIBS 16
-#define MAX_VERTEX_TEXTURE_IMAGE_UNITS MAX_TEXTURE_UNITS
-#define MAX_COMBINED_TEXTURE_IMAGE_UNITS (MAX_TEXTURE_IMAGE_UNITS + MAX_VERTEX_TEXTURE_IMAGE_UNITS)
+#define MAX_VERTEX_TEXTURE_IMAGE_UNITS MAX_TEXTURE_IMAGE_UNITS
+#define MAX_COMBINED_TEXTURE_IMAGE_UNITS MAX_TEXTURE_IMAGE_UNITS
 /*@}*/
 
 
diff --git a/src/mesa/main/context.c b/src/mesa/main/context.c
index 61c0861cbd4..ea52b26f0f7 100644
--- a/src/mesa/main/context.c
+++ b/src/mesa/main/context.c
@@ -1,14 +1,9 @@
-/**
- * \file context.c
- * Mesa context/visual/framebuffer management functions.
- * \author Brian Paul
- */
-
 /*
  * Mesa 3-D graphics library
- * Version:  7.1
+ * Version:  7.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2008  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -28,6 +23,11 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
+/**
+ * \file context.c
+ * Mesa context/visual/framebuffer management functions.
+ * \author Brian Paul
+ */
 
 /**
  * \mainpage Mesa Main Module
@@ -870,9 +870,6 @@ _mesa_init_constants(GLcontext *ctx)
    assert(MAX_TEXTURE_LEVELS >= MAX_3D_TEXTURE_LEVELS);
    assert(MAX_TEXTURE_LEVELS >= MAX_CUBE_TEXTURE_LEVELS);
 
-   assert(MAX_TEXTURE_UNITS >= MAX_TEXTURE_COORD_UNITS);
-   assert(MAX_TEXTURE_UNITS >= MAX_TEXTURE_IMAGE_UNITS);
-
    /* Constants, may be overriden (usually only reduced) by device drivers */
    ctx->Const.MaxTextureLevels = MAX_TEXTURE_LEVELS;
    ctx->Const.Max3DTextureLevels = MAX_3D_TEXTURE_LEVELS;
@@ -963,6 +960,9 @@ check_context_limits(GLcontext *ctx)
    assert(ctx->Const.MaxTextureUnits <= MAX_TEXTURE_IMAGE_UNITS);
    assert(ctx->Const.MaxTextureUnits <= MAX_TEXTURE_COORD_UNITS);
 
+   /* number of coord units cannot be greater than number of image units */
+   assert(ctx->Const.MaxTextureCoordUnits <= ctx->Const.MaxTextureImageUnits);
+
    assert(ctx->Const.MaxViewportWidth <= MAX_WIDTH);
    assert(ctx->Const.MaxViewportHeight <= MAX_WIDTH);
 
diff --git a/src/mesa/main/dlopen.c b/src/mesa/main/dlopen.c
index becef8173ef..8bc83c094fd 100644
--- a/src/mesa/main/dlopen.c
+++ b/src/mesa/main/dlopen.c
@@ -48,7 +48,7 @@ _mesa_dlopen(const char *libname, int flags)
    flags = RTLD_LAZY | RTLD_GLOBAL; /* Overriding flags at this time */
    return dlopen(libname, flags);
 #elif defined(__MINGW32__)
-   return LoadLibrary(libname);
+   return LoadLibraryA(libname);
 #else
    return NULL;
 #endif
diff --git a/src/mesa/main/enable.c b/src/mesa/main/enable.c
index 248df1badca..72ed50808c5 100644
--- a/src/mesa/main/enable.c
+++ b/src/mesa/main/enable.c
@@ -201,6 +201,26 @@ _mesa_DisableClientState( GLenum cap )
    }
 
 
+
+/**
+ * Return pointer to current texture unit for setting/getting coordinate
+ * state.
+ * Note that we'll set GL_INVALID_OPERATION if the active texture unit is
+ * higher than the number of supported coordinate units.  And we'll return NULL.
+ */
+static struct gl_texture_unit *
+get_texcoord_unit(GLcontext *ctx)
+{
+   if (ctx->Texture.CurrentUnit >= ctx->Const.MaxTextureCoordUnits) {
+      _mesa_error(ctx, GL_INVALID_OPERATION, "glEnable/Disable(texcoord unit)");
+      return NULL;
+   }
+   else {
+      return &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
+   }
+}
+
+
 /**
  * Helper function to enable or disable a texture target.
  */
@@ -612,54 +632,62 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
             return;
          }
          break;
-      case GL_TEXTURE_GEN_Q: {
-         GLuint unit = ctx->Texture.CurrentUnit;
-         struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-         GLuint newenabled = texUnit->TexGenEnabled & ~Q_BIT;
-         if (state)
-            newenabled |= Q_BIT;
-         if (texUnit->TexGenEnabled == newenabled)
-            return;
-         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-         texUnit->TexGenEnabled = newenabled;
+      case GL_TEXTURE_GEN_Q:
+         {
+            struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               GLuint newenabled = texUnit->TexGenEnabled & ~Q_BIT;
+               if (state)
+                  newenabled |= Q_BIT;
+               if (texUnit->TexGenEnabled == newenabled)
+                  return;
+               FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+               texUnit->TexGenEnabled = newenabled;
+            }
+         }
          break;
-      }
-      case GL_TEXTURE_GEN_R: {
-         GLuint unit = ctx->Texture.CurrentUnit;
-         struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-         GLuint newenabled = texUnit->TexGenEnabled & ~R_BIT;
-         if (state)
-            newenabled |= R_BIT;
-         if (texUnit->TexGenEnabled == newenabled)
-            return;
-         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-         texUnit->TexGenEnabled = newenabled;
+      case GL_TEXTURE_GEN_R:
+         {
+            struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               GLuint newenabled = texUnit->TexGenEnabled & ~R_BIT;
+               if (state)
+                  newenabled |= R_BIT;
+               if (texUnit->TexGenEnabled == newenabled)
+                  return;
+               FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+               texUnit->TexGenEnabled = newenabled;
+            }
+         }
          break;
-      }
-      case GL_TEXTURE_GEN_S: {
-         GLuint unit = ctx->Texture.CurrentUnit;
-         struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-         GLuint newenabled = texUnit->TexGenEnabled & ~S_BIT;
-         if (state)
-            newenabled |= S_BIT;
-         if (texUnit->TexGenEnabled == newenabled)
-            return;
-         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-         texUnit->TexGenEnabled = newenabled;
+      case GL_TEXTURE_GEN_S:
+         {
+            struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               GLuint newenabled = texUnit->TexGenEnabled & ~S_BIT;
+               if (state)
+                  newenabled |= S_BIT;
+               if (texUnit->TexGenEnabled == newenabled)
+                  return;
+               FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+               texUnit->TexGenEnabled = newenabled;
+            }
+         }
          break;
-      }
-      case GL_TEXTURE_GEN_T: {
-         GLuint unit = ctx->Texture.CurrentUnit;
-         struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-         GLuint newenabled = texUnit->TexGenEnabled & ~T_BIT;
-         if (state)
-            newenabled |= T_BIT;
-         if (texUnit->TexGenEnabled == newenabled)
-            return;
-         FLUSH_VERTICES(ctx, _NEW_TEXTURE);
-         texUnit->TexGenEnabled = newenabled;
+      case GL_TEXTURE_GEN_T:
+         {
+            struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               GLuint newenabled = texUnit->TexGenEnabled & ~T_BIT;
+               if (state)
+                  newenabled |= T_BIT;
+               if (texUnit->TexGenEnabled == newenabled)
+                  return;
+               FLUSH_VERTICES(ctx, _NEW_TEXTURE);
+               texUnit->TexGenEnabled = newenabled;
+            }
+         }
          break;
-      }
 
       /*
        * CLIENT STATE!!!
@@ -1155,28 +1183,36 @@ _mesa_IsEnabled( GLenum cap )
          return is_texture_enabled(ctx, TEXTURE_3D_BIT);
       case GL_TEXTURE_GEN_Q:
          {
-            const struct gl_texture_unit *texUnit;
-            texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-            return (texUnit->TexGenEnabled & Q_BIT) ? GL_TRUE : GL_FALSE;
+            const struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               return (texUnit->TexGenEnabled & Q_BIT) ? GL_TRUE : GL_FALSE;
+            }
          }
+         return GL_FALSE;
       case GL_TEXTURE_GEN_R:
          {
-            const struct gl_texture_unit *texUnit;
-            texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-            return (texUnit->TexGenEnabled & R_BIT) ? GL_TRUE : GL_FALSE;
+            const struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               return (texUnit->TexGenEnabled & R_BIT) ? GL_TRUE : GL_FALSE;
+            }
          }
+         return GL_FALSE;
       case GL_TEXTURE_GEN_S:
          {
-            const struct gl_texture_unit *texUnit;
-            texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-            return (texUnit->TexGenEnabled & S_BIT) ? GL_TRUE : GL_FALSE;
+            const struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               return (texUnit->TexGenEnabled & S_BIT) ? GL_TRUE : GL_FALSE;
+            }
          }
+         return GL_FALSE;
       case GL_TEXTURE_GEN_T:
          {
-            const struct gl_texture_unit *texUnit;
-            texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
-            return (texUnit->TexGenEnabled & T_BIT) ? GL_TRUE : GL_FALSE;
+            const struct gl_texture_unit *texUnit = get_texcoord_unit(ctx);
+            if (texUnit) {
+               return (texUnit->TexGenEnabled & T_BIT) ? GL_TRUE : GL_FALSE;
+            }
          }
+         return GL_FALSE;
 
       /*
        * CLIENT STATE!!!
diff --git a/src/mesa/main/ffvertex_prog.c b/src/mesa/main/ffvertex_prog.c
index ec0a5e3896d..d70b78f2586 100644
--- a/src/mesa/main/ffvertex_prog.c
+++ b/src/mesa/main/ffvertex_prog.c
@@ -26,7 +26,7 @@
  **************************************************************************/
 
 /**
- * \file ffvertex_prog.
+ * \file ffvertex_prog.c
  *
  * Create a vertex program to execute the current fixed function T&L pipeline.
  * \author Keith Whitwell
@@ -101,6 +101,7 @@ static GLuint translate_fog_mode( GLenum mode )
    }
 }
 
+
 #define TXG_NONE           0
 #define TXG_OBJ_LINEAR     1
 #define TXG_EYE_LINEAR     2
@@ -145,6 +146,7 @@ tnl_get_per_vertex_materials(GLcontext *ctx)
    return mask;
 }
 
+
 /**
  * Should fog be computed per-vertex?
  */
@@ -159,6 +161,7 @@ tnl_get_per_vertex_fog(GLcontext *ctx)
 #endif
 }
 
+
 static GLboolean check_active_shininess( GLcontext *ctx,
                                          const struct state_key *key,
                                          GLuint side )
@@ -176,8 +179,6 @@ static GLboolean check_active_shininess( GLcontext *ctx,
 
    return GL_FALSE;
 }
-     
-
 
 
 static void make_state_key( GLcontext *ctx, struct state_key *key )
@@ -278,7 +279,7 @@ static void make_state_key( GLcontext *ctx, struct state_key *key )
        ctx->Texture._EnabledUnits)
       key->texture_enabled_global = 1;
       
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
 
       if (texUnit->_ReallyEnabled)
@@ -410,11 +411,13 @@ static struct ureg swizzle( struct ureg reg, int x, int y, int z, int 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 );
@@ -430,6 +433,7 @@ static struct ureg get_temp( struct tnl_program *p )
    return make_ureg(PROGRAM_TEMPORARY, bit-1);
 }
 
+
 static struct ureg reserve_temp( struct tnl_program *p )
 {
    struct ureg temp = get_temp( p );
@@ -437,6 +441,7 @@ static struct ureg reserve_temp( struct tnl_program *p )
    return temp;
 }
 
+
 static void release_temp( struct tnl_program *p, struct ureg reg )
 {
    if (reg.file == PROGRAM_TEMPORARY) {
@@ -493,6 +498,7 @@ static struct ureg register_input( struct tnl_program *p, GLuint input )
    }
 }
 
+
 /**
  * \param input  one of VERT_RESULT_x tokens.
  */
@@ -502,6 +508,7 @@ static struct ureg register_output( struct tnl_program *p, GLuint output )
    return make_ureg(PROGRAM_OUTPUT, output);
 }
 
+
 static struct ureg register_const4f( struct tnl_program *p, 
 			      GLfloat s0,
 			      GLfloat s1,
@@ -531,6 +538,7 @@ 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)) 
@@ -571,6 +579,7 @@ static void emit_arg( struct prog_src_register *src,
    ASSERT(src->Index == reg.idx);
 }
 
+
 static void emit_dst( struct prog_dst_register *dst,
 		      struct ureg reg, GLuint mask )
 {
@@ -586,6 +595,7 @@ static void emit_dst( struct prog_dst_register *dst,
    ASSERT(dst->Index == reg.idx);
 }
 
+
 static void debug_insn( struct prog_instruction *inst, const char *fn,
 			GLuint line )
 {
@@ -696,6 +706,7 @@ static void emit_matrix_transform_vec4( struct tnl_program *p,
    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.
@@ -721,6 +732,7 @@ static void emit_transpose_matrix_transform_vec4( struct tnl_program *p,
       release_temp(p, tmp);
 }
 
+
 static void emit_matrix_transform_vec3( struct tnl_program *p,
 					struct ureg dest,
 					const struct ureg *mat,
@@ -748,6 +760,7 @@ static void emit_normalize_vec3( struct tnl_program *p,
 #endif
 }
 
+
 static void emit_passthrough( struct tnl_program *p, 
 			      GLuint input,
 			      GLuint output )
@@ -756,6 +769,7 @@ static void emit_passthrough( struct tnl_program *p,
    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)) {
@@ -803,7 +817,6 @@ static struct ureg get_eye_position_z( struct tnl_program *p )
 }
    
 
-
 static struct ureg get_eye_position_normalized( struct tnl_program *p )
 {
    if (is_undef(p->eye_position_normalized)) {
@@ -865,7 +878,6 @@ static struct ureg get_transformed_normal( struct tnl_program *p )
 }
 
 
-
 static void build_hpos( struct tnl_program *p )
 {
    struct ureg pos = register_input( p, VERT_ATTRIB_POS ); 
@@ -891,7 +903,9 @@ static GLuint material_attrib( GLuint side, GLuint property )
 	   side);
 }
 
-/* Get a bitmask of which material values vary on a per-vertex basis.
+
+/**
+ * Get a bitmask of which material values vary on a per-vertex basis.
  */
 static void set_material_flags( struct tnl_program *p )
 {
@@ -927,7 +941,9 @@ static struct ureg get_material( struct tnl_program *p, GLuint side,
 				   MAT_BIT_FRONT_AMBIENT | \
 				   MAT_BIT_FRONT_DIFFUSE) << (side))
 
-/* Either return a precalculated constant value or emit code to
+
+/**
+ * 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.
  *
@@ -970,6 +986,7 @@ static struct ureg get_lightprod( struct tnl_program *p, GLuint light,
       return register_param4(p, STATE_LIGHTPROD, light, side, property);
 }
 
+
 static struct ureg calculate_light_attenuation( struct tnl_program *p,
 						GLuint i, 
 						struct ureg VPpli,
@@ -1226,7 +1243,6 @@ static void build_lighting( struct tnl_program *p )
 	    struct ureg res0, res1;
 	    GLuint mask0, mask1;
 
-	    
 	    if (count == nr_lights) {
 	       if (separate) {
 		  mask0 = WRITEMASK_XYZ;
@@ -1247,7 +1263,6 @@ static void build_lighting( struct tnl_program *p )
 	       res1 = _col1;
 	    }
 
-
 	    if (!is_undef(att)) {
                /* light is attenuated by distance */
                emit_op1(p, OPCODE_LIT, lit, 0, dots);
@@ -1320,7 +1335,6 @@ static void build_lighting( struct tnl_program *p )
 
 	    emit_op3(p, OPCODE_MAD, res0, mask0, swizzle1(lit,Y), diffuse, _bfc0);
 	    emit_op3(p, OPCODE_MAD, res1, mask1, swizzle1(lit,Z), specular, _bfc1);
-
             /* restore negate flag for next lighting */
             dots = negate(dots);
 
@@ -1395,6 +1409,7 @@ static void build_fog( struct tnl_program *p )
       emit_op1(p, useabs ? OPCODE_ABS : OPCODE_MOV, fog, WRITEMASK_X, input);
    }
 }
+
  
 static void build_reflect_texgen( struct tnl_program *p,
 				  struct ureg dest,
@@ -1414,6 +1429,7 @@ static void build_reflect_texgen( struct tnl_program *p,
    release_temp(p, tmp);
 }
 
+
 static void build_sphere_texgen( struct tnl_program *p,
 				 struct ureg dest,
 				 GLuint writemask )
@@ -1461,7 +1477,7 @@ static void build_texture_transform( struct tnl_program *p )
 {
    GLuint i, j;
 
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
 
       if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))
 	 continue;
@@ -1524,10 +1540,8 @@ static void build_texture_transform( struct tnl_program *p )
 	       case TXG_NONE:
 		  copy_mask |= WRITEMASK_X << j;
 	       }
-
 	    }
 
-	 
 	    if (sphere_mask) {
 	       build_sphere_texgen(p, out_texgen, sphere_mask);
 	    }
@@ -1610,6 +1624,7 @@ static void build_atten_pointsize( struct tnl_program *p )
    release_temp(p, ut);
 }
 
+
 /**
  * Emit constant point size.
  */
@@ -1620,6 +1635,7 @@ static void build_constant_pointsize( struct tnl_program *p )
    emit_op1(p, OPCODE_MOV, out, WRITEMASK_X, state_size);
 }
 
+
 /**
  * Pass-though per-vertex point size, from user's point size array.
  */
diff --git a/src/mesa/main/get.c b/src/mesa/main/get.c
index f72aa6a2882..8ce9b0ae69a 100644
--- a/src/mesa/main/get.c
+++ b/src/mesa/main/get.c
@@ -289,7 +289,7 @@ _mesa_GetBooleanv( GLenum pname, GLboolean *params )
          params[0] = INT_TO_BOOLEAN(ctx->DrawBuffer->Visual.depthBits);
          break;
       case GL_DEPTH_CLEAR_VALUE:
-         params[0] = FLOAT_TO_BOOLEAN(ctx->Depth.Clear);
+         params[0] = FLOAT_TO_BOOLEAN(((GLfloat) ctx->Depth.Clear));
          break;
       case GL_DEPTH_FUNC:
          params[0] = ENUM_TO_BOOLEAN(ctx->Depth.Func);
@@ -2137,7 +2137,7 @@ _mesa_GetFloatv( GLenum pname, GLfloat *params )
          params[0] = (GLfloat)(ctx->DrawBuffer->Visual.depthBits);
          break;
       case GL_DEPTH_CLEAR_VALUE:
-         params[0] = ctx->Depth.Clear;
+         params[0] = ((GLfloat) ctx->Depth.Clear);
          break;
       case GL_DEPTH_FUNC:
          params[0] = ENUM_TO_FLOAT(ctx->Depth.Func);
@@ -3985,7 +3985,7 @@ _mesa_GetIntegerv( GLenum pname, GLint *params )
          params[0] = ctx->DrawBuffer->Visual.depthBits;
          break;
       case GL_DEPTH_CLEAR_VALUE:
-         params[0] = FLOAT_TO_INT(ctx->Depth.Clear);
+         params[0] = FLOAT_TO_INT(((GLfloat) ctx->Depth.Clear));
          break;
       case GL_DEPTH_FUNC:
          params[0] = ENUM_TO_INT(ctx->Depth.Func);
diff --git a/src/mesa/main/get_gen.py b/src/mesa/main/get_gen.py
index 152e378b4fe..a191b045d33 100644
--- a/src/mesa/main/get_gen.py
+++ b/src/mesa/main/get_gen.py
@@ -180,7 +180,7 @@ StateVars = [
 	( "GL_DEPTH_BIAS", GLfloat, ["ctx->Pixel.DepthBias"], "", None ),
 	( "GL_DEPTH_BITS", GLint, ["ctx->DrawBuffer->Visual.depthBits"],
 	  "", None ),
-	( "GL_DEPTH_CLEAR_VALUE", GLfloatN, ["ctx->Depth.Clear"], "", None ),
+	( "GL_DEPTH_CLEAR_VALUE", GLfloatN, ["((GLfloat) ctx->Depth.Clear)"], "", None ),
 	( "GL_DEPTH_FUNC", GLenum, ["ctx->Depth.Func"], "", None ),
 	( "GL_DEPTH_RANGE", GLfloatN,
 	  [ "ctx->Viewport.Near", "ctx->Viewport.Far" ], "", None ),
diff --git a/src/mesa/main/image.c b/src/mesa/main/image.c
index c205b4b766d..4d86c547775 100644
--- a/src/mesa/main/image.c
+++ b/src/mesa/main/image.c
@@ -61,7 +61,7 @@
 /**
  * \return GL_TRUE if type is packed pixel type, GL_FALSE otherwise.
  */
-static GLboolean
+GLboolean
 _mesa_type_is_packed(GLenum type)
 {
    switch (type) {
diff --git a/src/mesa/main/image.h b/src/mesa/main/image.h
index 38e1374c20c..0e0bbd96d85 100644
--- a/src/mesa/main/image.h
+++ b/src/mesa/main/image.h
@@ -36,6 +36,9 @@ _mesa_swap2( GLushort *p, GLuint n );
 extern void
 _mesa_swap4( GLuint *p, GLuint n );
 
+extern GLboolean
+_mesa_type_is_packed(GLenum type);
+
 extern GLint
 _mesa_sizeof_type( GLenum type );
 
diff --git a/src/mesa/main/imports.c b/src/mesa/main/imports.c
index 6cfd7ccc723..69d55923c37 100644
--- a/src/mesa/main/imports.c
+++ b/src/mesa/main/imports.c
@@ -930,6 +930,18 @@ _mesa_sprintf( char *str, const char *fmt, ... )
    return r;
 }
 
+/** Wrapper around vsnprintf() */
+int
+_mesa_snprintf( char *str, size_t size, const char *fmt, ... )
+{
+   int r;
+   va_list args;
+   va_start( args, fmt );  
+   r = vsnprintf( str, size, fmt, args );
+   va_end( args );
+   return r;
+}
+
 /** Wrapper around printf(), using vsprintf() for the formatting. */
 void
 _mesa_printf( const char *fmtString, ... )
diff --git a/src/mesa/main/imports.h b/src/mesa/main/imports.h
index bab00420710..92d5d0bb4a8 100644
--- a/src/mesa/main/imports.h
+++ b/src/mesa/main/imports.h
@@ -763,6 +763,9 @@ _mesa_strtod( const char *s, char **end );
 extern int
 _mesa_sprintf( char *str, const char *fmt, ... );
 
+extern int
+_mesa_snprintf( char *str, size_t size, const char *fmt, ... );
+
 extern void
 _mesa_printf( const char *fmtString, ... );
 
diff --git a/src/mesa/main/mipmap.c b/src/mesa/main/mipmap.c
index 9e051ace25b..3dd4b3391b9 100644
--- a/src/mesa/main/mipmap.c
+++ b/src/mesa/main/mipmap.c
@@ -41,7 +41,11 @@ bytes_per_pixel(GLenum datatype, GLuint comps)
 {
    GLint b = _mesa_sizeof_packed_type(datatype);
    assert(b >= 0);
-   return b * comps;
+
+   if (_mesa_type_is_packed(datatype))
+       return b;
+   else
+       return b * comps;
 }
 
 
diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
index 2a48b7c5179..a17bf9c7b0c 100644
--- a/src/mesa/main/mtypes.h
+++ b/src/mesa/main/mtypes.h
@@ -1,15 +1,9 @@
-/**
- * \file mtypes.h
- * Main Mesa data structures.
- *
- * Please try to mark derived values with a leading underscore ('_').
- */
-
 /*
  * Mesa 3-D graphics library
  * Version:  7.3
  *
  * Copyright (C) 1999-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 1999-2008  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -29,7 +23,12 @@
  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  */
 
-
+/**
+ * \file mtypes.h
+ * Main Mesa data structures.
+ *
+ * Please try to mark derived values with a leading underscore ('_').
+ */
 
 #ifndef TYPES_H
 #define TYPES_H
@@ -641,7 +640,7 @@ struct gl_current_attrib
    GLfloat RasterColor[4];
    GLfloat RasterSecondaryColor[4];
    GLfloat RasterIndex;
-   GLfloat RasterTexCoords[MAX_TEXTURE_COORD_UNITS][4];
+   GLfloat RasterTexCoords[MAX_TEXTURE_UNITS][4];
    GLboolean RasterPosValid;
    /*@}*/
 };
@@ -725,10 +724,10 @@ struct gl_enable_attrib
    GLboolean SampleCoverage;          /* GL_ARB_multisample */
    GLboolean SampleCoverageInvert;    /* GL_ARB_multisample */
    GLboolean RasterPositionUnclipped; /* GL_IBM_rasterpos_clip */
-   GLuint Texture[MAX_TEXTURE_IMAGE_UNITS];
-   GLuint TexGen[MAX_TEXTURE_COORD_UNITS];
+   GLuint Texture[MAX_TEXTURE_UNITS];
+   GLuint TexGen[MAX_TEXTURE_UNITS];
    /* SGI_texture_color_table */
-   GLboolean TextureColorTable[MAX_TEXTURE_IMAGE_UNITS];
+   GLboolean TextureColorTable[MAX_TEXTURE_UNITS];
    /* GL_ARB_vertex_program / GL_NV_vertex_program */
    GLboolean VertexProgram;
    GLboolean VertexProgramPointSize;
@@ -1071,7 +1070,7 @@ struct gl_point_attrib
    GLfloat Threshold;		/**< GL_EXT_point_parameters */
    GLboolean _Attenuated;	/**< True if Params != [1, 0, 0] */
    GLboolean PointSprite;	/**< GL_NV/ARB_point_sprite */
-   GLboolean CoordReplace[MAX_TEXTURE_COORD_UNITS]; /**< GL_ARB_point_sprite */
+   GLboolean CoordReplace[MAX_TEXTURE_UNITS]; /**< GL_ARB_point_sprite */
    GLenum SpriteRMode;		/**< GL_NV_point_sprite (only!) */
    GLenum SpriteOrigin;		/**< GL_ARB_point_sprite */
 };
@@ -1559,7 +1558,7 @@ struct gl_texture_attrib
     * name multitexture 
     */
    /**@{*/
-   GLuint CurrentUnit;	       /**< Active texture unit */
+   GLuint CurrentUnit;   /**< Active texture unit [0, MaxTextureImageUnits-1] */
    GLbitfield _EnabledUnits;  /**< one bit set for each really-enabled unit */
    GLbitfield _EnabledCoordUnits;   /**< one bit per enabled coordinate unit */
    GLbitfield _GenFlags;            /**< for texgen */
@@ -1876,7 +1875,7 @@ struct gl_program
    GLbitfield OutputsWritten; /**< Bitmask of which output regs are written to */
    GLbitfield InputFlags[MAX_PROGRAM_INPUTS];   /**< PROG_PARAM_BIT_x flags */
    GLbitfield OutputFlags[MAX_PROGRAM_OUTPUTS]; /**< PROG_PARAM_BIT_x flags */
-   GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS];  /**< TEXTURE_x_BIT bitmask */
+   GLbitfield TexturesUsed[MAX_TEXTURE_UNITS];  /**< TEXTURE_x_BIT bitmask */
    GLbitfield SamplersUsed;   /**< Bitfield of which samplers are used */
    GLbitfield ShadowSamplers; /**< Texture units used for shadow sampling. */
 
@@ -2473,9 +2472,9 @@ struct gl_constants
    GLint MaxTextureRectSize;            /* GL_NV_texture_rectangle */
    GLuint MaxTextureCoordUnits;
    GLuint MaxTextureImageUnits;
-   GLuint MaxTextureUnits;              /* = MIN(CoordUnits, ImageUnits) */
-   GLfloat MaxTextureMaxAnisotropy;	/* GL_EXT_texture_filter_anisotropic */
-   GLfloat MaxTextureLodBias;           /* GL_EXT_texture_lod_bias */
+   GLuint MaxTextureUnits;              /**< = MIN(CoordUnits, ImageUnits) */
+   GLfloat MaxTextureMaxAnisotropy;     /**< GL_EXT_texture_filter_anisotropic */
+   GLfloat MaxTextureLodBias;           /**< GL_EXT_texture_lod_bias */
    GLuint MaxArrayLockSize;
    GLint SubPixelBits;
    GLfloat MinPointSize, MaxPointSize;		/* aliased */
@@ -2951,7 +2950,7 @@ struct __GLcontextRec
    struct gl_matrix_stack ModelviewMatrixStack;
    struct gl_matrix_stack ProjectionMatrixStack;
    struct gl_matrix_stack ColorMatrixStack;
-   struct gl_matrix_stack TextureMatrixStack[MAX_TEXTURE_COORD_UNITS];
+   struct gl_matrix_stack TextureMatrixStack[MAX_TEXTURE_UNITS];
    struct gl_matrix_stack ProgramMatrixStack[MAX_PROGRAM_MATRICES];
    struct gl_matrix_stack *CurrentStack; /**< Points to one of the above stacks */
    /*@}*/
diff --git a/src/mesa/main/points.c b/src/mesa/main/points.c
index 1fe697033f1..4c8fc1f72e0 100644
--- a/src/mesa/main/points.c
+++ b/src/mesa/main/points.c
@@ -257,7 +257,7 @@ _mesa_init_point(GLcontext *ctx)
    ctx->Point.PointSprite = GL_FALSE; /* GL_ARB/NV_point_sprite */
    ctx->Point.SpriteRMode = GL_ZERO; /* GL_NV_point_sprite (only!) */
    ctx->Point.SpriteOrigin = GL_UPPER_LEFT; /* GL_ARB_point_sprite */
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       ctx->Point.CoordReplace[i] = GL_FALSE; /* GL_ARB/NV_point_sprite */
    }
 }
diff --git a/src/mesa/main/rastpos.c b/src/mesa/main/rastpos.c
index 9842172f460..9f309d6ab84 100644
--- a/src/mesa/main/rastpos.c
+++ b/src/mesa/main/rastpos.c
@@ -500,7 +500,7 @@ void _mesa_init_rastpos( GLcontext * ctx )
    ASSIGN_4V( ctx->Current.RasterColor, 1.0, 1.0, 1.0, 1.0 );
    ASSIGN_4V( ctx->Current.RasterSecondaryColor, 0.0, 0.0, 0.0, 1.0 );
    ctx->Current.RasterIndex = 1.0;
-   for (i=0; i<MAX_TEXTURE_UNITS; i++)
+   for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++)
       ASSIGN_4V( ctx->Current.RasterTexCoords[i], 0.0, 0.0, 0.0, 1.0 );
    ctx->Current.RasterPosValid = GL_TRUE;
 }
diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c
index 2f90a342202..48abf51d890 100644
--- a/src/mesa/main/texenvprogram.c
+++ b/src/mesa/main/texenvprogram.c
@@ -38,6 +38,21 @@
 #include "texenvprogram.h"
 
 
+/*
+ * Note on texture units:
+ *
+ * The number of texture units supported by fixed-function fragment
+ * processing is MAX_TEXTURE_COORD_UNITS, not MAX_TEXTURE_IMAGE_UNITS.
+ * That's because there's a one-to-one correspondence between texture
+ * coordinates and samplers in fixed-function processing.
+ *
+ * Since fixed-function vertex processing is limited to MAX_TEXTURE_COORD_UNITS
+ * sets of texcoords, so is fixed-function fragment processing.
+ *
+ * We can safely use ctx->Const.MaxTextureUnits for loop bounds.
+ */
+
+
 struct texenvprog_cache_item
 {
    GLuint hash;
@@ -50,13 +65,13 @@ struct texenvprog_cache_item
 /**
  * Up to nine instructions per tex unit, plus fog, specular color.
  */
-#define MAX_INSTRUCTIONS ((MAX_TEXTURE_UNITS * 9) + 12)
+#define MAX_INSTRUCTIONS ((MAX_TEXTURE_COORD_UNITS * 9) + 12)
 
 #define DISASSEM (MESA_VERBOSE & VERBOSE_DISASSEM)
 
 struct mode_opt {
-   GLubyte Source:4;
-   GLubyte Operand:3;
+   GLuint Source:4;
+   GLuint Operand:3;
 };
 
 struct state_key {
@@ -313,7 +328,7 @@ static void make_state_key( GLcontext *ctx,  struct state_key *key )
 
    memset(key, 0, sizeof(*key));
 
-   for (i=0;i<MAX_TEXTURE_UNITS;i++) {
+   for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
       const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
       GLenum format;
 
@@ -416,7 +431,7 @@ struct texenv_fragment_program {
    GLbitfield temp_in_use;	/**< Tracks temporary regs which are in use. */
    GLboolean error;
 
-   struct ureg src_texture[MAX_TEXTURE_UNITS];   
+   struct ureg src_texture[MAX_TEXTURE_COORD_UNITS];   
    /* Reg containing each texture unit's sampled texture color,
     * else undef.
     */
@@ -1201,7 +1216,7 @@ create_new_program(GLcontext *ctx, struct state_key *key,
    p.program->Base.InputsRead = 0;
    p.program->Base.OutputsWritten = 1 << FRAG_RESULT_COLR;
 
-   for (unit = 0; unit < MAX_TEXTURE_UNITS; unit++)
+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++)
       p.src_texture[unit] = undef;
 
    p.src_previous = undef;
diff --git a/src/mesa/main/texformat.c b/src/mesa/main/texformat.c
index db3525666a2..50e8d7a3b87 100644
--- a/src/mesa/main/texformat.c
+++ b/src/mesa/main/texformat.c
@@ -1685,7 +1685,7 @@ _mesa_format_to_type_and_comps(const struct gl_texture_format *format,
    case MESA_FORMAT_ARGB1555:
    case MESA_FORMAT_ARGB1555_REV:
       *datatype = GL_UNSIGNED_SHORT_1_5_5_5_REV;
-      *comps = 3;
+      *comps = 4;
       return;
 
    case MESA_FORMAT_AL88:
diff --git a/src/mesa/main/texrender.c b/src/mesa/main/texrender.c
index 163bda45014..4ae13a7b9b5 100644
--- a/src/mesa/main/texrender.c
+++ b/src/mesa/main/texrender.c
@@ -49,6 +49,14 @@ texture_get_row(GLcontext *ctx, struct gl_renderbuffer *rb, GLuint count,
          trb->TexImage->FetchTexelc(trb->TexImage, x + i, y, z, rgbaOut + 4 * i);
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      GLushort *zValues = (GLushort *) values;
+      for (i = 0; i < count; i++) {
+         GLfloat flt;
+         trb->TexImage->FetchTexelf(trb->TexImage, x + i, y, z, &flt);
+         zValues[i] = (GLushort) (flt * 0xffff);
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT) {
       GLuint *zValues = (GLuint *) values;
       /*
@@ -96,6 +104,15 @@ texture_get_values(GLcontext *ctx, struct gl_renderbuffer *rb, GLuint count,
 				    z, rgbaOut + 4 * i);
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      GLushort *zValues = (GLushort *) values;
+      for (i = 0; i < count; i++) {
+         GLfloat flt;
+         trb->TexImage->FetchTexelf(trb->TexImage, x[i], y[i] + trb->Yoffset,
+				    z, &flt);
+         zValues[i] = (GLushort) (flt * 0xffff);
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT) {
       GLuint *zValues = (GLuint *) values;
       for (i = 0; i < count; i++) {
@@ -147,6 +164,14 @@ texture_put_row(GLcontext *ctx, struct gl_renderbuffer *rb, GLuint count,
          rgba += 4;
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      const GLushort *zValues = (const GLushort *) values;
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            trb->Store(trb->TexImage, x + i, y, z, zValues + i);
+         }
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT) {
       const GLuint *zValues = (const GLuint *) values;
       for (i = 0; i < count; i++) {
@@ -189,6 +214,14 @@ texture_put_mono_row(GLcontext *ctx, struct gl_renderbuffer *rb, GLuint count,
          }
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      const GLushort zValue = *((const GLushort *) value);
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            trb->Store(trb->TexImage, x + i, y, z, &zValue);
+         }
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT) {
       const GLuint zValue = *((const GLuint *) value);
       for (i = 0; i < count; i++) {
@@ -231,12 +264,19 @@ texture_put_values(GLcontext *ctx, struct gl_renderbuffer *rb, GLuint count,
          rgba += 4;
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      const GLushort *zValues = (const GLushort *) values;
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            trb->Store(trb->TexImage, x[i], y[i] + trb->Yoffset, z, zValues + i);
+         }
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT) {
       const GLuint *zValues = (const GLuint *) values;
       for (i = 0; i < count; i++) {
          if (!mask || mask[i]) {
-            trb->Store(trb->TexImage, x[i], y[i] + trb->Yoffset, z,
-		       zValues + i);
+            trb->Store(trb->TexImage, x[i], y[i] + trb->Yoffset, z, zValues + i);
          }
       }
    }
@@ -281,6 +321,14 @@ texture_put_mono_values(GLcontext *ctx, struct gl_renderbuffer *rb,
          }
       }
    }
+   else if (rb->DataType == GL_UNSIGNED_SHORT) {
+      const GLushort zValue = *((const GLushort *) value);
+      for (i = 0; i < count; i++) {
+         if (!mask || mask[i]) {
+            trb->Store(trb->TexImage, x[i], y[i] + trb->Yoffset, z, &zValue);
+         }
+      }
+   }
    else if (rb->DataType == GL_UNSIGNED_INT_24_8_EXT) {
       const GLuint zValue = *((const GLuint *) value);
       const GLfloat flt = (GLfloat) ((zValue >> 8) * (1.0 / 0xffffff));
diff --git a/src/mesa/main/texstate.c b/src/mesa/main/texstate.c
index f0193770412..29955d76cbe 100644
--- a/src/mesa/main/texstate.c
+++ b/src/mesa/main/texstate.c
@@ -81,7 +81,7 @@ _mesa_copy_texture_state( const GLcontext *src, GLcontext *dst )
    dst->Texture.SharedPalette = src->Texture.SharedPalette;
 
    /* per-unit state */
-   for (i = 0; i < src->Const.MaxTextureUnits; i++) {
+   for (i = 0; i < src->Const.MaxTextureImageUnits; i++) {
       dst->Texture.Unit[i].Enabled = src->Texture.Unit[i].Enabled;
       dst->Texture.Unit[i].EnvMode = src->Texture.Unit[i].EnvMode;
       COPY_4V(dst->Texture.Unit[i].EnvColor, src->Texture.Unit[i].EnvColor);
@@ -307,8 +307,7 @@ _mesa_ActiveTextureARB(GLenum texture)
       _mesa_debug(ctx, "glActiveTexture %s\n",
                   _mesa_lookup_enum_by_nr(texture));
 
-   /* XXX error-check against max(coordunits, imageunits) */
-   if (texUnit >= ctx->Const.MaxTextureUnits) {
+   if (texUnit >= ctx->Const.MaxTextureImageUnits) {
       _mesa_error(ctx, GL_INVALID_ENUM, "glActiveTexture(texture)");
       return;
    }
@@ -369,7 +368,7 @@ update_texture_matrices( GLcontext *ctx )
 
    ctx->Texture._TexMatEnabled = 0;
 
-   for (i=0; i < ctx->Const.MaxTextureUnits; i++) {
+   for (i=0; i < ctx->Const.MaxTextureCoordUnits; i++) {
       if (_math_matrix_is_dirty(ctx->TextureMatrixStack[i].Top)) {
 	 _math_matrix_analyse( ctx->TextureMatrixStack[i].Top );
 
@@ -491,7 +490,7 @@ update_texture_state( GLcontext *ctx )
    /*
     * Update texture unit state.
     */
-   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
+   for (unit = 0; unit < ctx->Const.MaxTextureImageUnits; unit++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
       GLbitfield enableBits;
 
@@ -621,7 +620,7 @@ update_texture_state( GLcontext *ctx )
    }
 
    /* Setup texgen for those texture coordinate sets that are in use */
-   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
+   for (unit = 0; unit < ctx->Const.MaxTextureCoordUnits; unit++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 
       if (!(ctx->Texture._EnabledCoordUnits & (1 << unit)))
diff --git a/src/mesa/shader/prog_parameter.c b/src/mesa/shader/prog_parameter.c
index e6653157d46..8ae961241f6 100644
--- a/src/mesa/shader/prog_parameter.c
+++ b/src/mesa/shader/prog_parameter.c
@@ -290,7 +290,8 @@ _mesa_use_uniform(struct gl_program_parameter_list *paramList,
    GLuint i;
    for (i = 0; i < paramList->NumParameters; i++) {
       struct gl_program_parameter *p = paramList->Parameters + i;
-      if (p->Type == PROGRAM_UNIFORM && _mesa_strcmp(p->Name, name) == 0) {
+      if ((p->Type == PROGRAM_UNIFORM || p->Type == PROGRAM_SAMPLER) &&
+          _mesa_strcmp(p->Name, name) == 0) {
          p->Used = GL_TRUE;
          /* Note that large uniforms may occupy several slots so we're
           * not done searching yet.
diff --git a/src/mesa/shader/prog_statevars.h b/src/mesa/shader/prog_statevars.h
index 72e51f40314..d5358a1d042 100644
--- a/src/mesa/shader/prog_statevars.h
+++ b/src/mesa/shader/prog_statevars.h
@@ -79,10 +79,10 @@ typedef enum gl_state_index_ {
    STATE_SHININESS,
    STATE_HALF_VECTOR,
 
-   STATE_POSITION,
-   STATE_ATTENUATION,
-   STATE_SPOT_DIRECTION,
-   STATE_SPOT_CUTOFF,
+   STATE_POSITION,       /**< xyzw = position */
+   STATE_ATTENUATION,    /**< xyz = attenuation, w = spot exponent */
+   STATE_SPOT_DIRECTION, /**< xyz = direction, w = cos(cutoff) */
+   STATE_SPOT_CUTOFF,    /**< x = cutoff, yzw = undefined */
 
    STATE_TEXGEN_EYE_S,
    STATE_TEXGEN_EYE_T,
diff --git a/src/mesa/shader/slang/library/slang_common_builtin.gc b/src/mesa/shader/slang/library/slang_common_builtin.gc
index a051c53eea3..230c57cea8d 100644
--- a/src/mesa/shader/slang/library/slang_common_builtin.gc
+++ b/src/mesa/shader/slang/library/slang_common_builtin.gc
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5
+ * Version:  7.3
  *
  * Copyright (C) 2006  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2008  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -26,8 +27,9 @@
 // From Shader Spec, ver. 1.10, rev. 59
 //
 
-//bp: XXX these will probably go away since the value needs to be
-//determined at runtime and may vary from one GLcontext to another...
+// Note: the values assigned to these constants here aren't actually used.
+// They're set by the compiler according to the GL context limits.
+// See slang_simplify.c
 const int gl_MaxLights = 8;
 const int gl_MaxClipPlanes = 6;
 const int gl_MaxTextureUnits = 8;
@@ -98,6 +100,9 @@ struct gl_MaterialParameters {
 uniform gl_MaterialParameters gl_FrontMaterial;
 uniform gl_MaterialParameters gl_BackMaterial;
 
+/* NOTE: the order of these fields is significant!
+ * See the definition of the lighting state vars such as STATE_SPOT_DIRECTION.
+ */
 struct gl_LightSourceParameters {
     vec4 ambient;
     vec4 diffuse;
@@ -105,12 +110,14 @@ struct gl_LightSourceParameters {
     vec4 position;
     vec4 halfVector;
     vec3 spotDirection;
-    float spotExponent;
-    float spotCutoff;
     float spotCosCutoff;
+
     float constantAttenuation;
     float linearAttenuation;
     float quadraticAttenuation;
+    float spotExponent;
+
+    float spotCutoff;
 };
 
 uniform gl_LightSourceParameters gl_LightSource[gl_MaxLights];
diff --git a/src/mesa/shader/slang/library/slang_common_builtin_gc.h b/src/mesa/shader/slang/library/slang_common_builtin_gc.h
index 46632010815..759bf247d80 100644
--- a/src/mesa/shader/slang/library/slang_common_builtin_gc.h
+++ b/src/mesa/shader/slang/library/slang_common_builtin_gc.h
@@ -66,58 +66,58 @@
 116,83,111,117,114,99,101,80,97,114,97,109,101,116,101,114,115,0,12,0,97,109,98,105,101,110,116,0,
 0,0,1,12,0,100,105,102,102,117,115,101,0,0,0,1,12,0,115,112,101,99,117,108,97,114,0,0,0,1,12,0,112,
 111,115,105,116,105,111,110,0,0,0,1,12,0,104,97,108,102,86,101,99,116,111,114,0,0,0,1,11,0,115,112,
-111,116,68,105,114,101,99,116,105,111,110,0,0,0,1,9,0,115,112,111,116,69,120,112,111,110,101,110,
-116,0,0,0,1,9,0,115,112,111,116,67,117,116,111,102,102,0,0,0,1,9,0,115,112,111,116,67,111,115,67,
-117,116,111,102,102,0,0,0,1,9,0,99,111,110,115,116,97,110,116,65,116,116,101,110,117,97,116,105,
-111,110,0,0,0,1,9,0,108,105,110,101,97,114,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,0,
-113,117,97,100,114,97,116,105,99,65,116,116,101,110,117,97,116,105,111,110,0,0,0,0,0,0,0,2,2,90,95,
-4,0,25,103,108,95,76,105,103,104,116,83,111,117,114,99,101,80,97,114,97,109,101,116,101,114,115,0,
-0,1,103,108,95,76,105,103,104,116,83,111,117,114,99,101,0,3,18,103,108,95,77,97,120,76,105,103,104,
-116,115,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,77,111,100,101,108,80,97,114,97,109,
-101,116,101,114,115,0,12,0,97,109,98,105,101,110,116,0,0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,76,
-105,103,104,116,77,111,100,101,108,80,97,114,97,109,101,116,101,114,115,0,0,1,103,108,95,76,105,
-103,104,116,77,111,100,101,108,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,77,111,100,101,
-108,80,114,111,100,117,99,116,115,0,12,0,115,99,101,110,101,67,111,108,111,114,0,0,0,0,0,0,0,2,2,
-90,95,4,0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,0,1,
-103,108,95,70,114,111,110,116,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,
-0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,
-0,0,1,103,108,95,66,97,99,107,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,
-0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,12,0,97,109,98,
-105,101,110,116,0,0,0,1,12,0,100,105,102,102,117,115,101,0,0,0,1,12,0,115,112,101,99,117,108,97,
-114,0,0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,0,
-1,103,108,95,70,114,111,110,116,76,105,103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,
-97,120,76,105,103,104,116,115,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,80,114,111,100,
-117,99,116,115,0,0,1,103,108,95,66,97,99,107,76,105,103,104,116,80,114,111,100,117,99,116,0,3,18,
-103,108,95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,84,101,120,116,
-117,114,101,69,110,118,67,111,108,111,114,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,
-73,109,97,103,101,85,110,105,116,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,
+111,116,68,105,114,101,99,116,105,111,110,0,0,0,1,9,0,115,112,111,116,67,111,115,67,117,116,111,
+102,102,0,0,0,1,9,0,99,111,110,115,116,97,110,116,65,116,116,101,110,117,97,116,105,111,110,0,0,0,
+1,9,0,108,105,110,101,97,114,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,0,113,117,97,100,
+114,97,116,105,99,65,116,116,101,110,117,97,116,105,111,110,0,0,0,1,9,0,115,112,111,116,69,120,112,
+111,110,101,110,116,0,0,0,1,9,0,115,112,111,116,67,117,116,111,102,102,0,0,0,0,0,0,0,2,2,90,95,4,0,
+25,103,108,95,76,105,103,104,116,83,111,117,114,99,101,80,97,114,97,109,101,116,101,114,115,0,0,1,
+103,108,95,76,105,103,104,116,83,111,117,114,99,101,0,3,18,103,108,95,77,97,120,76,105,103,104,116,
+115,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,77,111,100,101,108,80,97,114,97,109,101,
+116,101,114,115,0,12,0,97,109,98,105,101,110,116,0,0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,
+103,104,116,77,111,100,101,108,80,97,114,97,109,101,116,101,114,115,0,0,1,103,108,95,76,105,103,
+104,116,77,111,100,101,108,0,0,0,2,2,90,95,0,0,24,103,108,95,76,105,103,104,116,77,111,100,101,108,
+80,114,111,100,117,99,116,115,0,12,0,115,99,101,110,101,67,111,108,111,114,0,0,0,0,0,0,0,2,2,90,95,
+4,0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,0,1,103,
+108,95,70,114,111,110,116,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,
+2,90,95,4,0,25,103,108,95,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,115,0,0,
+1,103,108,95,66,97,99,107,76,105,103,104,116,77,111,100,101,108,80,114,111,100,117,99,116,0,0,0,2,
+2,90,95,0,0,24,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,12,0,97,109,98,105,
+101,110,116,0,0,0,1,12,0,100,105,102,102,117,115,101,0,0,0,1,12,0,115,112,101,99,117,108,97,114,0,
+0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,80,114,111,100,117,99,116,115,0,0,1,103,
+108,95,70,114,111,110,116,76,105,103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,95,77,97,120,
+76,105,103,104,116,115,0,0,0,2,2,90,95,4,0,25,103,108,95,76,105,103,104,116,80,114,111,100,117,99,
+116,115,0,0,1,103,108,95,66,97,99,107,76,105,103,104,116,80,114,111,100,117,99,116,0,3,18,103,108,
+95,77,97,120,76,105,103,104,116,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,84,101,120,116,117,114,
+101,69,110,118,67,111,108,111,114,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,73,109,97,
+103,101,85,110,105,116,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,110,101,83,0,
+3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,
+0,1,103,108,95,69,121,101,80,108,97,110,101,84,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,
+101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,110,101,82,0,
+3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,
+0,1,103,108,95,69,121,101,80,108,97,110,101,81,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,
+101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,80,108,97,
 110,101,83,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,
-90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,110,101,84,0,3,18,103,108,95,77,97,120,84,101,120,
-116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,
-110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,
-90,95,4,0,12,0,1,103,108,95,69,121,101,80,108,97,110,101,81,0,3,18,103,108,95,77,97,120,84,101,120,
-116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,
-80,108,97,110,101,83,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,
-0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,84,0,3,18,103,108,95,
-77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,
-79,98,106,101,99,116,80,108,97,110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,
-67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,80,108,97,110,
-101,81,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,
-95,0,0,24,103,108,95,70,111,103,80,97,114,97,109,101,116,101,114,115,0,12,0,99,111,108,111,114,0,0,
-0,1,9,0,100,101,110,115,105,116,121,0,0,0,1,9,0,115,116,97,114,116,0,0,0,1,9,0,101,110,100,0,0,0,1,
-9,0,115,99,97,108,101,0,0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,70,111,103,80,97,114,97,109,101,
-116,101,114,115,0,0,1,103,108,95,70,111,103,0,0,0,1,90,95,0,0,9,0,0,114,97,100,105,97,110,115,0,1,
-1,0,0,9,0,100,101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,
-48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,
-108,0,0,18,100,101,103,0,0,18,99,0,0,0,0,1,90,95,0,0,10,0,0,114,97,100,105,97,110,115,0,1,1,0,0,10,
-0,100,101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,
-0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,
-59,120,121,0,0,18,100,101,103,0,59,120,121,0,0,18,99,0,59,120,120,0,0,0,0,1,90,95,0,0,11,0,0,114,
-97,100,105,97,110,115,0,1,1,0,0,11,0,100,101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,
-49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,
-0,18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,100,101,103,0,59,120,121,122,0,0,18,99,0,
-59,120,120,120,0,0,0,0,1,90,95,0,0,12,0,0,114,97,100,105,97,110,115,0,1,1,0,0,12,0,100,101,103,0,0,
-0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,
+90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,84,0,3,18,103,108,95,77,97,120,
+84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,
+101,99,116,80,108,97,110,101,82,0,3,18,103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,
+114,100,115,0,0,0,2,2,90,95,4,0,12,0,1,103,108,95,79,98,106,101,99,116,80,108,97,110,101,81,0,3,18,
+103,108,95,77,97,120,84,101,120,116,117,114,101,67,111,111,114,100,115,0,0,0,2,2,90,95,0,0,24,103,
+108,95,70,111,103,80,97,114,97,109,101,116,101,114,115,0,12,0,99,111,108,111,114,0,0,0,1,9,0,100,
+101,110,115,105,116,121,0,0,0,1,9,0,115,116,97,114,116,0,0,0,1,9,0,101,110,100,0,0,0,1,9,0,115,99,
+97,108,101,0,0,0,0,0,0,0,2,2,90,95,4,0,25,103,108,95,70,111,103,80,97,114,97,109,101,116,101,114,
+115,0,0,1,103,108,95,70,111,103,0,0,0,1,90,95,0,0,9,0,0,114,97,100,105,97,110,115,0,1,1,0,0,9,0,
+100,101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,
+0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,
+18,100,101,103,0,0,18,99,0,0,0,0,1,90,95,0,0,10,0,0,114,97,100,105,97,110,115,0,1,1,0,0,10,0,100,
+101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,
+49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,59,
+120,121,0,0,18,100,101,103,0,59,120,121,0,0,18,99,0,59,120,120,0,0,0,0,1,90,95,0,0,11,0,0,114,97,
+100,105,97,110,115,0,1,1,0,0,11,0,100,101,103,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,
+53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,101,99,52,95,109,117,108,116,105,112,108,121,0,
+18,95,95,114,101,116,86,97,108,0,59,120,121,122,0,0,18,100,101,103,0,59,120,121,122,0,0,18,99,0,59,
+120,120,120,0,0,0,0,1,90,95,0,0,12,0,0,114,97,100,105,97,110,115,0,1,1,0,0,12,0,100,101,103,0,0,0,
+1,3,2,90,95,1,0,9,0,1,99,0,2,17,51,0,49,52,49,53,57,50,54,0,0,17,49,56,48,0,48,0,0,49,0,0,4,118,
 101,99,52,95,109,117,108,116,105,112,108,121,0,18,95,95,114,101,116,86,97,108,0,0,18,100,101,103,0,
 0,18,99,0,59,120,120,120,120,0,0,0,0,1,90,95,0,0,9,0,0,100,101,103,114,101,101,115,0,1,1,0,0,9,0,
 114,97,100,0,0,0,1,3,2,90,95,1,0,9,0,1,99,0,2,17,49,56,48,0,48,0,0,17,51,0,49,52,49,53,57,50,54,0,
diff --git a/src/mesa/shader/slang/slang_builtin.c b/src/mesa/shader/slang/slang_builtin.c
index db00c54b8a2..c0f4c79e136 100644
--- a/src/mesa/shader/slang/slang_builtin.c
+++ b/src/mesa/shader/slang/slang_builtin.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  6.5.3
+ * Version:  7.3
  *
  * Copyright (C) 2005-2007  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2008  VMware, Inc.   All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -39,6 +40,10 @@
 #include "shader/slang/slang_builtin.h"
 
 
+/** special state token (see below) */
+#define STATE_ARRAY ((gl_state_index) 0xfffff)
+
+
 /**
  * Lookup GL state given a variable name, 0, 1 or 2 indexes and a field.
  * Allocate room for the state in the given param list and return position
@@ -132,6 +137,8 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
       }
    }
    else if (strcmp(var, "gl_ClipPlane") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_CLIPPLANE;
       tokens[1] = index1;
    }
@@ -196,8 +203,12 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
       }
    }
    else if (strcmp(var, "gl_LightSource") == 0) {
+      if (!field || index1 < 0)
+         return -1;
+
       tokens[0] = STATE_LIGHT;
       tokens[1] = index1;
+
       if (strcmp(field, "ambient") == 0) {
          tokens[2] = STATE_AMBIENT;
       }
@@ -272,6 +283,9 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
    }
    else if (strcmp(var, "gl_FrontLightProduct") == 0 ||
             strcmp(var, "gl_BackLightProduct") == 0) {
+      if (index1 < 0 || !field)
+         return -1;
+
       tokens[0] = STATE_LIGHTPROD;
       tokens[1] = index1; /* light number */
       if (strcmp(var, "gl_FrontLightProduct") == 0) {
@@ -294,45 +308,63 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
       }
    }
    else if (strcmp(var, "gl_TextureEnvColor") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXENV_COLOR;
       tokens[1] = index1;
    }
    else if (strcmp(var, "gl_EyePlaneS") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_EYE_S;
    }
    else if (strcmp(var, "gl_EyePlaneT") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_EYE_T;
    }
    else if (strcmp(var, "gl_EyePlaneR") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_EYE_R;
    }
    else if (strcmp(var, "gl_EyePlaneQ") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_EYE_Q;
    }
    else if (strcmp(var, "gl_ObjectPlaneS") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_OBJECT_S;
    }
    else if (strcmp(var, "gl_ObjectPlaneT") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_OBJECT_T;
    }
    else if (strcmp(var, "gl_ObjectPlaneR") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_OBJECT_R;
    }
    else if (strcmp(var, "gl_ObjectPlaneQ") == 0) {
+      if (index1 < 0)
+         return -1;
       tokens[0] = STATE_TEXGEN;
       tokens[1] = index1; /* tex unit */
       tokens[2] = STATE_TEXGEN_OBJECT_Q;
@@ -386,13 +418,222 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
 }
 
 
+
+/**
+ * Given a variable name and datatype, emit uniform/constant buffer
+ * entries which will store that state variable.
+ * For example, if name="gl_LightSource" we'll emit 64 state variable
+ * vectors/references and return position where that data starts.  This will
+ * allow run-time array indexing into the light source array.
+ *
+ * Note that this is a recursive function.
+ *
+ * \return -1 if error, else index of start of data in the program parameter list
+ */
+static GLint
+emit_statevars(const char *name, int array_len,
+               const slang_type_specifier *type,
+               gl_state_index tokens[STATE_LENGTH],
+               struct gl_program_parameter_list *paramList)
+{
+   if (type->type == SLANG_SPEC_ARRAY) {
+      GLint i, pos;
+      assert(array_len > 0);
+      if (strcmp(name, "gl_ClipPlane") == 0) {
+         tokens[0] = STATE_CLIPPLANE;
+      }
+      else if (strcmp(name, "gl_LightSource") == 0) {
+         tokens[0] = STATE_LIGHT;
+      }
+      else if (strcmp(name, "gl_FrontLightProduct") == 0) {
+         tokens[0] = STATE_LIGHTPROD;
+         tokens[2] = 0; /* front */
+      }
+      else if (strcmp(name, "gl_BackLightProduct") == 0) {
+         tokens[0] = STATE_LIGHTPROD;
+         tokens[2] = 1; /* back */
+      }
+      else if (strcmp(name, "gl_TextureEnvColor") == 0) {
+         tokens[0] = STATE_TEXENV_COLOR;
+      }
+      else if (strcmp(name, "gl_EyePlaneS") == 0) {
+         tokens[0] = STATE_TEXGEN_EYE_S;
+      }
+      else if (strcmp(name, "gl_EyePlaneT") == 0) {
+         tokens[0] = STATE_TEXGEN_EYE_T;
+      }
+      else if (strcmp(name, "gl_EyePlaneR") == 0) {
+         tokens[0] = STATE_TEXGEN_EYE_R;
+      }
+      else if (strcmp(name, "gl_EyePlaneQ") == 0) {
+         tokens[0] = STATE_TEXGEN_EYE_Q;
+      }
+      else if (strcmp(name, "gl_ObjectPlaneS") == 0) {
+         tokens[0] = STATE_TEXGEN_OBJECT_S;
+      }
+      else if (strcmp(name, "gl_ObjectPlaneT") == 0) {
+         tokens[0] = STATE_TEXGEN_OBJECT_T;
+      }
+      else if (strcmp(name, "gl_ObjectPlaneR") == 0) {
+         tokens[0] = STATE_TEXGEN_OBJECT_R;
+      }
+      else if (strcmp(name, "gl_ObjectPlaneQ") == 0) {
+         tokens[0] = STATE_TEXGEN_OBJECT_Q;
+      }
+      else {
+         return -1; /* invalid array name */
+      }
+      for (i = 0; i < array_len; i++) {
+         GLint p;
+         tokens[1] = i;
+         p = emit_statevars(NULL, 0, type->_array, tokens, paramList);
+         if (i == 0)
+            pos = p;
+      }
+      return pos;
+   }
+   else if (type->type == SLANG_SPEC_STRUCT) {
+      const slang_variable_scope *fields = type->_struct->fields;
+      GLuint i, pos;
+      for (i = 0; i < fields->num_variables; i++) {
+         const slang_variable *var = fields->variables[i];
+         GLint p = emit_statevars(var->a_name, 0, &var->type.specifier,
+                                  tokens, paramList);
+         if (i == 0)
+            pos = p;
+      }
+      return pos;
+   }
+   else {
+      GLint pos;
+      assert(type->type == SLANG_SPEC_VEC4 ||
+             type->type == SLANG_SPEC_VEC3 ||
+             type->type == SLANG_SPEC_VEC2 ||
+             type->type == SLANG_SPEC_FLOAT ||
+             type->type == SLANG_SPEC_IVEC4 ||
+             type->type == SLANG_SPEC_IVEC3 ||
+             type->type == SLANG_SPEC_IVEC2 ||
+             type->type == SLANG_SPEC_INT);
+      if (name) {
+         GLint t;
+
+         if (tokens[0] == STATE_LIGHT)
+            t = 2;
+         else if (tokens[0] == STATE_LIGHTPROD)
+            t = 3;
+         else
+            return -1; /* invalid array name */
+
+         if (strcmp(name, "ambient") == 0) {
+            tokens[t] = STATE_AMBIENT;
+         }
+         else if (strcmp(name, "diffuse") == 0) {
+            tokens[t] = STATE_DIFFUSE;
+         }
+         else if (strcmp(name, "specular") == 0) {
+            tokens[t] = STATE_SPECULAR;
+         }
+         else if (strcmp(name, "position") == 0) {
+            tokens[t] = STATE_POSITION;
+         }
+         else if (strcmp(name, "halfVector") == 0) {
+            tokens[t] = STATE_HALF_VECTOR;
+         }
+         else if (strcmp(name, "spotDirection") == 0) {
+            tokens[t] = STATE_SPOT_DIRECTION; /* xyz components */
+         }
+         else if (strcmp(name, "spotCosCutoff") == 0) {
+            tokens[t] = STATE_SPOT_DIRECTION; /* w component */
+         }
+
+         else if (strcmp(name, "constantAttenuation") == 0) {
+            tokens[t] = STATE_ATTENUATION; /* x component */
+         }
+         else if (strcmp(name, "linearAttenuation") == 0) {
+            tokens[t] = STATE_ATTENUATION; /* y component */
+         }
+         else if (strcmp(name, "quadraticAttenuation") == 0) {
+            tokens[t] = STATE_ATTENUATION; /* z component */
+         }
+         else if (strcmp(name, "spotExponent") == 0) {
+            tokens[t] = STATE_ATTENUATION; /* w = spot exponent */
+         }
+
+         else if (strcmp(name, "spotCutoff") == 0) {
+            tokens[t] = STATE_SPOT_CUTOFF; /* x component */
+         }
+
+         else {
+            return -1; /* invalid field name */
+         }
+      }
+
+      pos = _mesa_add_state_reference(paramList, tokens);
+      return pos;
+   }
+
+   return 1;
+}
+
+
+/**
+ * Unroll the named built-in uniform variable into a sequence of state
+ * vars in the given parameter list.
+ */
+static GLint
+alloc_state_var_array(const slang_variable *var,
+                      struct gl_program_parameter_list *paramList)
+{
+   gl_state_index tokens[STATE_LENGTH];
+   GLuint i;
+   GLint pos;
+
+   /* Initialize the state tokens array.  This is very important.
+    * When we call _mesa_add_state_reference() it'll searches the parameter
+    * list to see if the given statevar token sequence is already present.
+    * This is normally a good thing since it prevents redundant values in the
+    * constant buffer.
+    *
+    * But when we're building arrays of state this can be bad.  For example,
+    * consider this fragment of GLSL code:
+    *   foo = gl_LightSource[3].diffuse;
+    *   ...
+    *   bar = gl_LightSource[i].diffuse;
+    *
+    * When we unroll the gl_LightSource array (for "bar") we want to re-emit
+    * gl_LightSource[3].diffuse and not re-use the first instance (from "foo")
+    * since that would upset the array layout.  We handle this situation by
+    * setting the last token in the state var token array to the special
+    * value STATE_ARRAY.
+    * This token will only be set for array state.  We can hijack the last
+    * element in the array for this since it's never used for light, clipplane
+    * or texture env array state.
+    */
+   for (i = 0; i < STATE_LENGTH; i++)
+      tokens[i] = 0;
+   tokens[STATE_LENGTH - 1] = STATE_ARRAY;
+
+   pos = emit_statevars(var->a_name, var->array_len, &var->type.specifier,
+                        tokens, paramList);
+
+   return pos;
+}
+
+
+
 /**
  * Allocate storage for a pre-defined uniform (a GL state variable).
  * As a memory-saving optimization, we try to only allocate storage for
  * state vars that are actually used.
- * For example, the "gl_LightSource" uniform is huge.  If we only use
- * a handful of gl_LightSource fields, we don't want to allocate storage
- * for all of gl_LightSource.
+ *
+ * Arrays such as gl_LightSource are handled specially.  For an expression
+ * like "gl_LightSource[2].diffuse", we can allocate a single uniform/constant
+ * slot and return the index.  In this case, we return direct=TRUE.
+ *
+ * Buf for something like "gl_LightSource[i].diffuse" we don't know the value
+ * of 'i' at compile time so we need to "unroll" the gl_LightSource array
+ * into a consecutive sequence of uniform/constant slots so it can be indexed
+ * at runtime.  In this case, we return direct=FALSE.
  *
  * Currently, all pre-defined uniforms are in one of these forms:
  *   var
@@ -401,52 +642,62 @@ lookup_statevar(const char *var, GLint index1, GLint index2, const char *field,
  *   var[i].field
  *   var[i][j]
  *
- * \return -1 upon error, else position in paramList of the state var/data
+ * \return -1 upon error, else position in paramList of the state variable/data
  */
 GLint
 _slang_alloc_statevar(slang_ir_node *n,
-                      struct gl_program_parameter_list *paramList)
+                      struct gl_program_parameter_list *paramList,
+                      GLboolean *direct)
 {
    slang_ir_node *n0 = n;
-   const char *field = NULL, *var;
-   GLint index1 = -1, index2 = -1, pos;
+   const char *field = NULL;
+   GLint index1 = -1, index2 = -1;
    GLuint swizzle;
 
+   *direct = GL_TRUE;
+
    if (n->Opcode == IR_FIELD) {
       field = n->Field;
       n = n->Children[0];
    }
 
    if (n->Opcode == IR_ELEMENT) {
-      /* XXX can only handle constant indexes for now */
       if (n->Children[1]->Opcode == IR_FLOAT) {
          index1 = (GLint) n->Children[1]->Value[0];
-         n = n->Children[0];
       }
       else {
-         return -1;
+         *direct = GL_FALSE;
       }
+      n = n->Children[0];
    }
 
    if (n->Opcode == IR_ELEMENT) {
       /* XXX can only handle constant indexes for now */
-      assert(n->Children[1]->Opcode == IR_FLOAT);
-      index2 = (GLint) n->Children[1]->Value[0];
+      if (n->Children[1]->Opcode == IR_FLOAT) {
+         index2 = (GLint) n->Children[1]->Value[0];
+      }
+      else {
+         *direct = GL_FALSE;
+      }
       n = n->Children[0];
    }
 
    assert(n->Opcode == IR_VAR);
-   var = (char *) n->Var->a_name;
 
-   pos = lookup_statevar(var, index1, index2, field, &swizzle, paramList);
-   assert(pos >= 0);
-   if (pos >= 0) {
-      /* newly resolved storage for the statevar/constant/uniform */
-      n0->Store->File = PROGRAM_STATE_VAR;
-      n0->Store->Index = pos;
-      n0->Store->Swizzle = swizzle;
-      n0->Store->Parent = NULL;
+   if (*direct) {
+      const char *var = (const char *) n->Var->a_name;
+      GLint pos =
+         lookup_statevar(var, index1, index2, field, &swizzle, paramList);
+      if (pos >= 0) {
+         /* newly resolved storage for the statevar/constant/uniform */
+         n0->Store->File = PROGRAM_STATE_VAR;
+         n0->Store->Index = pos;
+         n0->Store->Swizzle = swizzle;
+         n0->Store->Parent = NULL;
+         return pos;
+      }
    }
-   return pos;
-}
 
+   *direct = GL_FALSE;
+   return alloc_state_var_array(n->Var, paramList);
+}
diff --git a/src/mesa/shader/slang/slang_builtin.h b/src/mesa/shader/slang/slang_builtin.h
index 58629f4f7fe..7f6fe80fcca 100644
--- a/src/mesa/shader/slang/slang_builtin.h
+++ b/src/mesa/shader/slang/slang_builtin.h
@@ -33,7 +33,8 @@
 
 extern GLint
 _slang_alloc_statevar(slang_ir_node *n,
-                      struct gl_program_parameter_list *paramList);
+                      struct gl_program_parameter_list *paramList,
+                      GLboolean *direct);
 
 
 #endif /* SLANG_BUILTIN_H */
diff --git a/src/mesa/shader/slang/slang_codegen.c b/src/mesa/shader/slang/slang_codegen.c
index 7d764cb5c1d..66615d3afd6 100644
--- a/src/mesa/shader/slang/slang_codegen.c
+++ b/src/mesa/shader/slang/slang_codegen.c
@@ -391,7 +391,7 @@ _slang_input_index(const char *name, GLenum target, GLuint *swizzleOut)
    const struct input_info *inputs
       = (target == GL_VERTEX_PROGRAM_ARB) ? vertInputs : fragInputs;
 
-   ASSERT(MAX_TEXTURE_UNITS == 8); /* if this fails, fix vertInputs above */
+   ASSERT(MAX_TEXTURE_COORD_UNITS == 8); /* if this fails, fix vertInputs above */
 
    for (i = 0; inputs[i].Name; i++) {
       if (strcmp(inputs[i].Name, name) == 0) {
@@ -1976,7 +1976,7 @@ _slang_make_array_constructor(slang_assemble_ctx *A, slang_operation *oper)
          */
          slang_variable *p = slang_variable_scope_grow(fun->parameters);
          char name[10];
-         snprintf(name, sizeof(name), "p%d", i);
+         _mesa_snprintf(name, sizeof(name), "p%d", i);
          p->a_name = slang_atom_pool_atom(A->atoms, name);
          p->type.qualifier = SLANG_QUAL_CONST;
          p->type.specifier.type = baseType;
@@ -3323,6 +3323,22 @@ is_store_writable(const slang_assemble_ctx *A, const slang_ir_storage *store)
 
 
 /**
+ * Walk up an IR storage path to compute the final swizzle.
+ * This is used when we find an expression such as "foo.xz.yx".
+ */
+static GLuint
+root_swizzle(const slang_ir_storage *st)
+{
+   GLuint swizzle = st->Swizzle;
+   while (st->Parent) {
+      st = st->Parent;
+      swizzle = _slang_swizzle_swizzle(st->Swizzle, swizzle);
+   }
+   return swizzle;
+}
+
+
+/**
  * Generate IR tree for an assignment (=).
  */
 static slang_ir_node *
@@ -3397,9 +3413,9 @@ _slang_gen_assignment(slang_assemble_ctx * A, slang_operation *oper)
       rhs = _slang_gen_operation(A, &oper->children[1]);
       if (lhs && rhs) {
          /* convert lhs swizzle into writemask */
+         const GLuint swizzle = root_swizzle(lhs->Store);
          GLuint writemask, newSwizzle;
-         if (!swizzle_to_writemask(A, lhs->Store->Swizzle,
-                                   &writemask, &newSwizzle)) {
+         if (!swizzle_to_writemask(A, swizzle, &writemask, &newSwizzle)) {
             /* Non-simple writemask, need to swizzle right hand side in
              * order to put components into the right place.
              */
diff --git a/src/mesa/shader/slang/slang_compile.c b/src/mesa/shader/slang/slang_compile.c
index d8aefd64950..b8044b10600 100644
--- a/src/mesa/shader/slang/slang_compile.c
+++ b/src/mesa/shader/slang/slang_compile.c
@@ -1470,9 +1470,9 @@ parse_expression(slang_parse_ctx * C, slang_output_ctx * O,
                   RETURN0;
                }
                else {
+                  slang_operation array_size;
                   array_constructor = GL_TRUE;
                   /* parse the array constructor size */
-                  slang_operation array_size;
                   slang_operation_construct(&array_size);
                   if (!parse_expression(C, O, &array_size)) {
                      slang_operation_destruct(&array_size);
diff --git a/src/mesa/shader/slang/slang_emit.c b/src/mesa/shader/slang/slang_emit.c
index 6587f9cf276..d3b4e64b78d 100644
--- a/src/mesa/shader/slang/slang_emit.c
+++ b/src/mesa/shader/slang/slang_emit.c
@@ -310,24 +310,22 @@ storage_to_dst_reg(struct prog_dst_register *dst, const slang_ir_storage *st)
       dst->WriteMask = swizzle_to_writemask(swizzle);
    }
    else {
-      GLuint writemask;
       switch (size) {
       case 1:
-         writemask = WRITEMASK_X << GET_SWZ(st->Swizzle, 0);
+         dst->WriteMask = WRITEMASK_X << GET_SWZ(st->Swizzle, 0);
          break;
       case 2:
-         writemask = WRITEMASK_XY;
+         dst->WriteMask = WRITEMASK_XY;
          break;
       case 3:
-         writemask = WRITEMASK_XYZ;
+         dst->WriteMask = WRITEMASK_XYZ;
          break;
       case 4:
-         writemask = WRITEMASK_XYZW;
+         dst->WriteMask = WRITEMASK_XYZW;
          break;
       default:
          ; /* error would have been caught above */
       }
-      dst->WriteMask = writemask;
    }
 
    dst->RelAddr = relAddr;
@@ -348,6 +346,10 @@ storage_to_src_reg(struct prog_src_register *src, const slang_ir_storage *st)
    assert(index >= 0);
    while (st->Parent) {
       st = st->Parent;
+      if (st->Index < 0) {
+         /* an error should have been reported already */
+         return;
+      }
       assert(st->Index >= 0);
       index += st->Index;
       swizzle = _slang_swizzle_swizzle(fix_swizzle(st->Swizzle), swizzle);
@@ -1286,6 +1288,7 @@ emit_tex(slang_emit_info *emitInfo, slang_ir_node *n)
 
    /* Child[0] is the sampler (a uniform which'll indicate the texture unit) */
    assert(n->Children[0]->Store);
+   assert(n->Children[0]->Store->File == PROGRAM_SAMPLER);
    /* Store->Index is the sampler index */
    assert(n->Children[0]->Store->Index >= 0);
    /* Store->Size is the texture target */
@@ -1295,6 +1298,10 @@ emit_tex(slang_emit_info *emitInfo, slang_ir_node *n)
    inst->TexSrcTarget = n->Children[0]->Store->Size;
    inst->TexSrcUnit = n->Children[0]->Store->Index; /* i.e. uniform's index */
 
+   /* mark the sampler as being used */
+   _mesa_use_uniform(emitInfo->prog->Parameters,
+                     (char *) n->Children[0]->Var->a_name);
+
    return inst;
 }
 
@@ -1841,9 +1848,17 @@ emit_array_element(slang_emit_info *emitInfo, slang_ir_node *n)
          root = root->Parent;
 
       if (root->File == PROGRAM_STATE_VAR) {
-         GLint index = _slang_alloc_statevar(n, emitInfo->prog->Parameters);
-         assert(n->Store->Index == index);
-         return NULL;
+         GLboolean direct;
+         GLint index =
+            _slang_alloc_statevar(n, emitInfo->prog->Parameters, &direct);
+         if (index < 0) {
+            /* error */
+            return NULL;
+         }
+         if (direct) {
+            n->Store->Index = index;
+            return NULL; /* all done */
+         }
       }
    }
 
@@ -1967,28 +1982,30 @@ emit_struct_field(slang_emit_info *emitInfo, slang_ir_node *n)
     * space for the ones that we actually use!
     */
    if (root->File == PROGRAM_STATE_VAR) {
-      root->Index = _slang_alloc_statevar(n, emitInfo->prog->Parameters);
-      if (root->Index < 0) {
+      GLboolean direct;
+      GLint index = _slang_alloc_statevar(n, emitInfo->prog->Parameters, &direct);
+      if (index < 0) {
          slang_info_log_error(emitInfo->log, "Error parsing state variable");
          return NULL;
       }
-      return NULL;
-   }
-   else {
-      /* do codegen for struct */
-      emit(emitInfo, n->Children[0]);
-      assert(n->Children[0]->Store->Index >= 0);
+      if (direct) {
+         root->Index = index;
+         return NULL; /* all done */
+      }
    }
 
+   /* do codegen for struct */
+   emit(emitInfo, n->Children[0]);
+   assert(n->Children[0]->Store->Index >= 0);
+
+
    fieldOffset = n->Store->Index;
    fieldSize = n->Store->Size;
 
    _slang_copy_ir_storage(n->Store, n->Children[0]->Store);
 
    n->Store->Index = n->Children[0]->Store->Index + fieldOffset / 4;
-   /* XXX test this:
-   n->Store->Index += fieldOffset / 4;
-   */
+   n->Store->Size = fieldSize;
 
    switch (fieldSize) {
    case 1:
@@ -2077,9 +2094,21 @@ emit_var_ref(slang_emit_info *emitInfo, slang_ir_node *n)
    assert(n->Store->File != PROGRAM_UNDEFINED);
 
    if (n->Store->File == PROGRAM_STATE_VAR && n->Store->Index < 0) {
-      n->Store->Index = _slang_alloc_statevar(n, emitInfo->prog->Parameters);
+      GLboolean direct;
+      GLint index = _slang_alloc_statevar(n, emitInfo->prog->Parameters, &direct);
+      if (index < 0) {
+         /* error */
+         char s[100];
+         _mesa_snprintf(s, sizeof(s), "Undefined variable '%s'",
+                        (char *) n->Var->a_name);
+         slang_info_log_error(emitInfo->log, s);
+         return NULL;
+      }
+
+      n->Store->Index = index;
    }
-   else if (n->Store->File == PROGRAM_UNIFORM) {
+   else if (n->Store->File == PROGRAM_UNIFORM ||
+            n->Store->File == PROGRAM_SAMPLER) {
       /* mark var as used */
       _mesa_use_uniform(emitInfo->prog->Parameters, (char *) n->Var->a_name);
    }
diff --git a/src/mesa/shader/slang/slang_link.c b/src/mesa/shader/slang/slang_link.c
index 3e97c8672b6..3f953d86e74 100644
--- a/src/mesa/shader/slang/slang_link.c
+++ b/src/mesa/shader/slang/slang_link.c
@@ -1,8 +1,9 @@
 /*
  * Mesa 3-D graphics library
- * Version:  7.2
+ * Version:  7.3
  *
  * Copyright (C) 2008  Brian Paul   All Rights Reserved.
+ * Copyright (C) 2009  VMware, Inc.  All Rights Reserved.
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the "Software"),
@@ -136,15 +137,15 @@ link_varying_vars(struct gl_shader_program *shProg, struct gl_program *prog)
          }
          if (!bits_agree(var->Flags, v->Flags, PROG_PARAM_BIT_CENTROID)) {
             char msg[100];
-            snprintf(msg, sizeof(msg),
-                     "centroid modifier mismatch for '%s'", var->Name);
+            _mesa_snprintf(msg, sizeof(msg),
+                           "centroid modifier mismatch for '%s'", var->Name);
             link_error(shProg, msg);
             return GL_FALSE;
          }
          if (!bits_agree(var->Flags, v->Flags, PROG_PARAM_BIT_INVARIANT)) {
             char msg[100];
-            snprintf(msg, sizeof(msg),
-                     "invariant modifier mismatch for '%s'", var->Name);
+            _mesa_snprintf(msg, sizeof(msg),
+                           "invariant modifier mismatch for '%s'", var->Name);
             link_error(shProg, msg);
             return GL_FALSE;
          }
@@ -206,13 +207,27 @@ link_varying_vars(struct gl_shader_program *shProg, struct gl_program *prog)
  * Build the shProg->Uniforms list.
  * This is basically a list/index of all uniforms found in either/both of
  * the vertex and fragment shaders.
+ *
+ * About uniforms:
+ * Each uniform has two indexes, one that points into the vertex
+ * program's parameter array and another that points into the fragment
+ * program's parameter array.  When the user changes a uniform's value
+ * we have to change the value in the vertex and/or fragment program's
+ * parameter array.
+ *
+ * This function will be called twice to set up the two uniform->parameter
+ * mappings.
+ *
+ * If a uniform is only present in the vertex program OR fragment program
+ * then the fragment/vertex parameter index, respectively, will be -1.
  */
 static GLboolean
-link_uniform_vars(struct gl_shader_program *shProg,
+link_uniform_vars(GLcontext *ctx,
+                  struct gl_shader_program *shProg,
                   struct gl_program *prog,
                   GLuint *numSamplers)
 {
-   GLuint samplerMap[MAX_SAMPLERS];
+   GLuint samplerMap[200]; /* max number of samplers declared, not used */
    GLuint i;
 
    for (i = 0; i < prog->Parameters->NumParameters; i++) {
@@ -227,30 +242,41 @@ link_uniform_vars(struct gl_shader_program *shProg,
        * Furthermore, we'll need to fix the state-var's size/datatype info.
        */
 
-      if ((p->Type == PROGRAM_UNIFORM && p->Used) ||
-          p->Type == PROGRAM_SAMPLER) {
+      if ((p->Type == PROGRAM_UNIFORM || p->Type == PROGRAM_SAMPLER)
+          && p->Used) {
+         /* add this uniform, indexing into the target's Parameters list */
          struct gl_uniform *uniform =
             _mesa_append_uniform(shProg->Uniforms, p->Name, prog->Target, i);
          if (uniform)
             uniform->Initialized = p->Initialized;
       }
 
-      if (p->Type == PROGRAM_SAMPLER) {
+      /* The samplerMap[] table we build here is used to remap/re-index
+       * sampler references by TEX instructions.
+       */
+      if (p->Type == PROGRAM_SAMPLER && p->Used) {
          /* Allocate a new sampler index */
-         GLuint sampNum = *numSamplers;
          GLuint oldSampNum = (GLuint) prog->Parameters->ParameterValues[i][0];
-         if (oldSampNum >= MAX_SAMPLERS) {
-            link_error(shProg, "Too many texture samplers");
+         GLuint newSampNum = *numSamplers;
+         if (newSampNum >= ctx->Const.MaxTextureImageUnits) {
+            char s[100];
+            sprintf(s, "Too many texture samplers (%u, max is %u)",
+                    newSampNum, ctx->Const.MaxTextureImageUnits);
+            link_error(shProg, s);
             return GL_FALSE;
          }
-         samplerMap[oldSampNum] = sampNum;
+         /* save old->new mapping in the table */
+         if (oldSampNum < Elements(samplerMap))
+            samplerMap[oldSampNum] = newSampNum;
+         /* update parameter's sampler index */
+         prog->Parameters->ParameterValues[i][0] = (GLfloat) newSampNum;
          (*numSamplers)++;
       }
    }
 
-
-   /* OK, now scan the program/shader instructions looking for sampler vars,
-    * replacing the old index with the new index.
+   /* OK, now scan the program/shader instructions looking for texture
+    * instructions using sampler vars.  Replace old sampler indexes with
+    * new ones.
     */
    prog->SamplersUsed = 0x0;
    for (i = 0; i < prog->NumInstructions; i++) {
@@ -261,10 +287,13 @@ link_uniform_vars(struct gl_shader_program *shProg,
                 inst->Sampler, map[ inst->Sampler ]);
          */
          /* here, texUnit is really samplerUnit */
-         assert(inst->TexSrcUnit < MAX_SAMPLERS);
-         inst->TexSrcUnit = samplerMap[inst->TexSrcUnit];
-         prog->SamplerTargets[inst->TexSrcUnit] = inst->TexSrcTarget;
-         prog->SamplersUsed |= (1 << inst->TexSrcUnit);
+         const GLint oldSampNum = inst->TexSrcUnit;
+         if (oldSampNum < Elements(samplerMap)) {
+            const GLuint newSampNum = samplerMap[oldSampNum];
+            inst->TexSrcUnit = newSampNum;
+            prog->SamplerTargets[newSampNum] = inst->TexSrcTarget;
+            prog->SamplersUsed |= (1 << newSampNum);
+         }
       }
    }
 
@@ -564,13 +593,13 @@ _slang_link(GLcontext *ctx,
 
    /* link uniform vars */
    if (shProg->VertexProgram) {
-      if (!link_uniform_vars(shProg, &shProg->VertexProgram->Base,
+      if (!link_uniform_vars(ctx, shProg, &shProg->VertexProgram->Base,
                              &numSamplers)) {
          return;
       }
    }
    if (shProg->FragmentProgram) {
-      if (!link_uniform_vars(shProg, &shProg->FragmentProgram->Base,
+      if (!link_uniform_vars(ctx, shProg, &shProg->FragmentProgram->Base,
                              &numSamplers)) {
          return;
       }
diff --git a/src/mesa/state_tracker/st_atom_texture.c b/src/mesa/state_tracker/st_atom_texture.c
index 18ffc08c6b1..21f7321f97a 100644
--- a/src/mesa/state_tracker/st_atom_texture.c
+++ b/src/mesa/state_tracker/st_atom_texture.c
@@ -53,7 +53,7 @@ update_textures(struct st_context *st)
    st->state.num_textures = 0;
 
    /* loop over sampler units (aka tex image units) */
-   for (su = 0; su < st->ctx->Const.MaxTextureCoordUnits; su++) {
+   for (su = 0; su < st->ctx->Const.MaxTextureImageUnits; su++) {
       struct pipe_texture *pt = NULL;
 
       if (samplersUsed & (1 << su)) {
diff --git a/src/mesa/state_tracker/st_cb_rasterpos.c b/src/mesa/state_tracker/st_cb_rasterpos.c
index 3eaccb74e17..8867ca5652e 100644
--- a/src/mesa/state_tracker/st_cb_rasterpos.c
+++ b/src/mesa/state_tracker/st_cb_rasterpos.c
@@ -162,7 +162,7 @@ rastpos_point(struct draw_stage *stage, struct prim_header *prim)
                  ctx->Current.RasterSecondaryColor,
                  VERT_RESULT_COL1, VERT_ATTRIB_COLOR1);
 
-   for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
+   for (i = 0; i < ctx->Const.MaxTextureCoordUnits; i++) {
       update_attrib(ctx, outputMapping, prim->v[0],
                     ctx->Current.RasterTexCoords[i],
                     VERT_RESULT_TEX0 + i, VERT_ATTRIB_TEX0 + i);
diff --git a/src/mesa/state_tracker/st_extensions.c b/src/mesa/state_tracker/st_extensions.c
index 5ff0c611477..8ec2113f074 100644
--- a/src/mesa/state_tracker/st_extensions.c
+++ b/src/mesa/state_tracker/st_extensions.c
@@ -84,14 +84,18 @@ void st_init_limits(struct st_context *st)
    c->MaxTextureRectSize
       = _min(1 << (c->MaxTextureLevels - 1), MAX_TEXTURE_RECT_SIZE);
 
-   c->MaxTextureUnits
-      = c->MaxTextureImageUnits
-      = c->MaxTextureCoordUnits
+   c->MaxTextureImageUnits
       = _min(screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_IMAGE_UNITS),
             MAX_TEXTURE_IMAGE_UNITS);
 
    c->MaxVertexTextureImageUnits
-      = screen->get_param(screen, PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS);
+      = _min(screen->get_param(screen, PIPE_CAP_MAX_VERTEX_TEXTURE_UNITS),
+             MAX_VERTEX_TEXTURE_IMAGE_UNITS);
+
+   c->MaxTextureCoordUnits
+      = _min(c->MaxTextureImageUnits, MAX_TEXTURE_COORD_UNITS);
+
+   c->MaxTextureUnits = _min(c->MaxTextureImageUnits, c->MaxTextureCoordUnits);
 
    c->MaxDrawBuffers
       = _clamp(screen->get_param(screen, PIPE_CAP_MAX_RENDER_TARGETS),
diff --git a/src/mesa/state_tracker/st_framebuffer.c b/src/mesa/state_tracker/st_framebuffer.c
index 43ac195e67c..ea22a943039 100644
--- a/src/mesa/state_tracker/st_framebuffer.c
+++ b/src/mesa/state_tracker/st_framebuffer.c
@@ -170,7 +170,6 @@ st_set_framebuffer_surface(struct st_framebuffer *stfb,
                            uint surfIndex, struct pipe_surface *surf)
 {
    GET_CURRENT_CONTEXT(ctx);
-   struct st_context *st;
    static const GLuint invalid_size = 9999999;
    struct st_renderbuffer *strb;
    GLuint width, height, i;
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.c b/src/mesa/state_tracker/st_mesa_to_tgsi.c
index ce8b4411940..5b5707fa1c5 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.c
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.c
@@ -745,6 +745,7 @@ find_temporaries(const struct gl_program *program,
  */
 GLuint
 st_translate_mesa_program(
+   GLcontext *ctx,
    uint procType,
    const struct gl_program *program,
    GLuint numInputs,
@@ -992,7 +993,7 @@ st_translate_mesa_program(
    }
 
    /* texture samplers */
-   for (i = 0; i < 8; i++) {
+   for (i = 0; i < ctx->Const.MaxTextureImageUnits; i++) {
       if (program->SamplersUsed & (1 << i)) {
          struct tgsi_full_declaration fulldecl;
 
diff --git a/src/mesa/state_tracker/st_mesa_to_tgsi.h b/src/mesa/state_tracker/st_mesa_to_tgsi.h
index 7b2bee1ab70..b465b3bddcf 100644
--- a/src/mesa/state_tracker/st_mesa_to_tgsi.h
+++ b/src/mesa/state_tracker/st_mesa_to_tgsi.h
@@ -41,6 +41,7 @@ struct gl_program;
 
 GLuint
 st_translate_mesa_program(
+   GLcontext *ctx,
    uint procType,
    const struct gl_program *program,
    GLuint numInputs,
diff --git a/src/mesa/state_tracker/st_program.c b/src/mesa/state_tracker/st_program.c
index cf4b39cee40..442eeed1470 100644
--- a/src/mesa/state_tracker/st_program.c
+++ b/src/mesa/state_tracker/st_program.c
@@ -317,7 +317,8 @@ st_translate_vertex_program(struct st_context *st,
 
    /* XXX: fix static allocation of tokens:
     */
-   num_tokens = st_translate_mesa_program(TGSI_PROCESSOR_VERTEX,
+   num_tokens = st_translate_mesa_program(st->ctx,
+                                          TGSI_PROCESSOR_VERTEX,
                                           &stvp->Base.Base,
                                           /* inputs */
                                           vs_num_inputs,
@@ -503,7 +504,8 @@ st_translate_fragment_program(struct st_context *st,
 
    /* XXX: fix static allocation of tokens:
     */
-   num_tokens = st_translate_mesa_program(TGSI_PROCESSOR_FRAGMENT,
+   num_tokens = st_translate_mesa_program(st->ctx,
+                                          TGSI_PROCESSOR_FRAGMENT,
                                           &stfp->Base.Base,
                                           /* inputs */
                                           fs_num_inputs,
diff --git a/src/mesa/state_tracker/wgl/SConscript b/src/mesa/state_tracker/wgl/SConscript
index cceb8264ead..bb579930f52 100644
--- a/src/mesa/state_tracker/wgl/SConscript
+++ b/src/mesa/state_tracker/wgl/SConscript
@@ -11,8 +11,9 @@ if env['platform'] in ['windows']:
     ])
      
     env.Append(CPPDEFINES = [
+        '_GDI32_', # prevent wgl* being declared __declspec(dllimport)
+        'BUILD_GL32', # declare gl* as __declspec(dllexport) in Mesa headers 
         '__GL_EXPORTS',
-        'BUILD_GL32',
         '_GNU_H_WINDOWS32_DEFINES',
     ])
      
diff --git a/src/mesa/state_tracker/wgl/stw_device.c b/src/mesa/state_tracker/wgl/stw_device.c
index 52907f1a792..e2a17d83ac6 100644
--- a/src/mesa/state_tracker/wgl/stw_device.c
+++ b/src/mesa/state_tracker/wgl/stw_device.c
@@ -37,8 +37,8 @@
 struct stw_device *stw_dev = NULL;
 
 
-static BOOL
-st_init(void)
+boolean
+st_init(const struct stw_winsys *stw_winsys)
 {
    static struct stw_device stw_dev_storage;
 
@@ -47,7 +47,9 @@ st_init(void)
    stw_dev = &stw_dev_storage;
    memset(stw_dev, 0, sizeof(*stw_dev));
 
-   stw_dev->screen = stw_winsys.create_screen();
+   stw_dev->stw_winsys = stw_winsys;
+
+   stw_dev->screen = stw_winsys->create_screen();
    if(!stw_dev->screen)
       goto error1;
 
@@ -61,7 +63,7 @@ error1:
 }
 
 
-static void
+void
 st_cleanup(void)
 {
    DHGLRC dhglrc;
@@ -76,18 +78,3 @@ st_cleanup(void)
 
    stw_dev = NULL;
 }
-
-
-BOOL WINAPI
-DllMain(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpReserved)
-{
-   switch (fdwReason) {
-   case DLL_PROCESS_ATTACH:
-      return st_init();
-
-   case DLL_PROCESS_DETACH:
-      st_cleanup();
-      break;
-   }
-   return TRUE;
-}
diff --git a/src/mesa/state_tracker/wgl/stw_device.h b/src/mesa/state_tracker/wgl/stw_device.h
index 49f79ac9c70..e2020bf0554 100644
--- a/src/mesa/state_tracker/wgl/stw_device.h
+++ b/src/mesa/state_tracker/wgl/stw_device.h
@@ -44,6 +44,8 @@ struct drv_context
 
 struct stw_device
 {
+   const struct stw_winsys *stw_winsys;
+   
    struct pipe_screen *screen;
 
    struct drv_context ctx_array[DRV_CONTEXT_MAX];
diff --git a/src/mesa/state_tracker/wgl/stw_framebuffer.c b/src/mesa/state_tracker/wgl/stw_framebuffer.c
index 57b89eee966..1ecafa451ee 100644
--- a/src/mesa/state_tracker/wgl/stw_framebuffer.c
+++ b/src/mesa/state_tracker/wgl/stw_framebuffer.c
@@ -25,9 +25,8 @@
  * 
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
+
 #include "main/context.h"
 #include "pipe/p_format.h"
 #include "state_tracker/st_context.h"
diff --git a/src/mesa/state_tracker/wgl/stw_icd.c b/src/mesa/state_tracker/wgl/stw_icd.c
index 17bdbd15fac..1dddc242097 100644
--- a/src/mesa/state_tracker/wgl/stw_icd.c
+++ b/src/mesa/state_tracker/wgl/stw_icd.c
@@ -29,12 +29,12 @@
 #include <stdio.h>
 
 #include "GL/gl.h"
-#include "GL/mesa_wgl.h"
 
 #include "pipe/p_debug.h"
 
 #include "stw_device.h"
 #include "stw_icd.h"
+#include "stw_wgl.h"
 
 
 static HGLRC
diff --git a/src/mesa/state_tracker/wgl/stw_icd.h b/src/mesa/state_tracker/wgl/stw_icd.h
index 7e2edca16ec..8e676fb5b74 100644
--- a/src/mesa/state_tracker/wgl/stw_icd.h
+++ b/src/mesa/state_tracker/wgl/stw_icd.h
@@ -31,9 +31,7 @@
 
 #include <windows.h>
 
-
 #include "GL/gl.h"
-#include "GL/mesa_wgl.h"
 
 
 typedef ULONG DHGLRC;
diff --git a/src/mesa/state_tracker/wgl/stw_wgl.c b/src/mesa/state_tracker/wgl/stw_wgl.c
index 6cace957458..0528c369fc8 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl.c
@@ -25,8 +25,6 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
 
 #include "pipe/p_debug.h"
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.h b/src/mesa/state_tracker/wgl/stw_wgl.h
index ee875c7a1d2..b86cc240f28 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.h
+++ b/src/mesa/state_tracker/wgl/stw_wgl.h
@@ -1,8 +1,8 @@
 /**************************************************************************
- * 
- * Copyright 2008 Tungsten Graphics, Inc., Cedar Park, Texas.
+ *
+ * Copyright 2009 VMware, Inc.
  * All Rights Reserved.
- * 
+ *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * copy of this software and associated documentation files (the
  * "Software"), to deal in the Software without restriction, including
@@ -10,26 +10,54 @@
  * 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
+ * IN NO EVENT SHALL VMWARE 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.
- * 
+ *
  **************************************************************************/
 
-#ifndef WGL_PIXELFORMAT_H
-#define WGL_PIXELFORMAT_H
+#ifndef STW_WGL_H_
+#define STW_WGL_H_
 
-WINGDIAPI int APIENTRY
-wglGetPixelFormat(
-   HDC hdc );
 
-#endif /* WGL_PIXELFORMAT_H */
+#include <windows.h>
+
+#include "GL/gl.h"
+
+
+/*
+ * Undeclared APIs exported by opengl32.dll
+ */
+
+WINGDIAPI BOOL WINAPI
+wglSwapBuffers(HDC hdc);
+
+WINGDIAPI int WINAPI
+wglChoosePixelFormat(HDC hdc,
+                     CONST PIXELFORMATDESCRIPTOR *ppfd);
+
+WINGDIAPI int WINAPI
+wglDescribePixelFormat(HDC hdc,
+                       int iPixelFormat,
+                       UINT nBytes,
+                       LPPIXELFORMATDESCRIPTOR ppfd);
+
+WINGDIAPI int WINAPI
+wglGetPixelFormat(HDC hdc);
+
+WINGDIAPI BOOL WINAPI
+wglSetPixelFormat(HDC hdc,
+                  int iPixelFormat,
+                  CONST PIXELFORMATDESCRIPTOR *ppfd);
+
+
+#endif /* STW_WGL_H_ */
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_arbextensionsstring.c b/src/mesa/state_tracker/wgl/stw_wgl_arbextensionsstring.c
index fe3a3c1daa9..04865796ec8 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_arbextensionsstring.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_arbextensionsstring.c
@@ -25,9 +25,8 @@
  * 
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
+
 #include "stw_wgl_arbextensionsstring.h"
 
 WINGDIAPI const char * APIENTRY
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_arbpixelformat.c b/src/mesa/state_tracker/wgl/stw_wgl_arbpixelformat.c
index 14a7c5e1e06..344bb15d3c1 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_arbpixelformat.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_arbpixelformat.c
@@ -25,8 +25,6 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
 
 #include "pipe/p_compiler.h"
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_context.c b/src/mesa/state_tracker/wgl/stw_wgl_context.c
index 59b47200be4..0c13c6b68ab 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_context.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_context.c
@@ -25,8 +25,6 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
 
 #include "main/mtypes.h"
@@ -41,7 +39,7 @@
 #include "stw_pixelformat.h"
 #include "stw_wgl_arbmultisample.h"
 #include "stw_wgl_context.h"
-#include "stw_wgl_pixelformat.h"
+#include "stw_wgl.h"
 
 static struct wgl_context *ctx_head = NULL;
 
@@ -107,7 +105,7 @@ wglCreateContext(
       return NULL;
    }
 
-   pipe = stw_winsys.create_context( stw_dev->screen );
+   pipe = stw_dev->stw_winsys->create_context( stw_dev->screen );
    if (!pipe) {
       _mesa_destroy_visual( visual );
       FREE( ctx );
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_getprocaddress.c b/src/mesa/state_tracker/wgl/stw_wgl_getprocaddress.c
index ec4f1513cb8..ec92d2dfce2 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_getprocaddress.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_getprocaddress.c
@@ -25,9 +25,8 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
+
 #include "glapi/glapi.h"
 #include "stw_wgl_arbextensionsstring.h"
 #include "stw_wgl_arbpixelformat.h"
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.c b/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.c
index bfc085093a8..7a8a2e22e49 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_pixelformat.c
@@ -25,14 +25,12 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
 
 #include "pipe/p_compiler.h"
 #include "pipe/p_debug.h"
 #include "stw_pixelformat.h"
-#include "stw_wgl_pixelformat.h"
+#include "stw_wgl.h"
 
 static uint currentpixelformat = 0;
 
diff --git a/src/mesa/state_tracker/wgl/stw_wgl_swapbuffers.c b/src/mesa/state_tracker/wgl/stw_wgl_swapbuffers.c
index a4dffc5fa0e..bd86501ac04 100644
--- a/src/mesa/state_tracker/wgl/stw_wgl_swapbuffers.c
+++ b/src/mesa/state_tracker/wgl/stw_wgl_swapbuffers.c
@@ -25,9 +25,8 @@
  *
  **************************************************************************/
 
-#define _GDI32_
-
 #include <windows.h>
+
 #include "pipe/p_winsys.h"
 #include "pipe/p_screen.h"
 #include "pipe/p_context.h"
@@ -36,7 +35,7 @@
 #include "stw_winsys.h"
 #include "stw_device.h"
 #include "stw_framebuffer.h"
-#include "stw_wgl_context.h"
+#include "stw_wgl.h"
 
 WINGDIAPI BOOL APIENTRY
 wglSwapBuffers(
@@ -56,9 +55,9 @@ wglSwapBuffers(
 
    surf = st_get_framebuffer_surface( fb->stfb, ST_SURFACE_BACK_LEFT );
 
-   stw_winsys.flush_frontbuffer(stw_dev->screen->winsys,
-                               surf,
-                               hdc );
+   stw_dev->stw_winsys->flush_frontbuffer(stw_dev->screen->winsys,
+                                          surf,
+                                          hdc );
 
    return TRUE;
 }
diff --git a/src/mesa/state_tracker/wgl/stw_winsys.h b/src/mesa/state_tracker/wgl/stw_winsys.h
index 68f1c7b16b7..8557327ccdd 100644
--- a/src/mesa/state_tracker/wgl/stw_winsys.h
+++ b/src/mesa/state_tracker/wgl/stw_winsys.h
@@ -30,9 +30,12 @@
 
 #include <windows.h> /* for HDC */
 
+#include "pipe/p_compiler.h"
+
 struct pipe_screen;
 struct pipe_context;
 struct pipe_winsys;
+struct pipe_surface;
 
 struct stw_winsys
 {
@@ -48,6 +51,10 @@ struct stw_winsys
                          HDC hDC );
 };
 
-extern const struct stw_winsys stw_winsys;
+boolean
+st_init(const struct stw_winsys *stw_winsys);
+
+void
+st_cleanup(void);
 
 #endif /* STW_WINSYS_H */
author	Keith Whitwell <[email protected]>	2009-01-09 10:08:06 +0000
committer	Keith Whitwell <[email protected]>	2009-01-09 10:08:06 +0000
commit	e3734593aea202e99e77febea7b86c904080939f (patch)
tree	69856674e2062c55ec5eec95eb0e31225a943084
parent	221352bbd79a0ea92ce31cffb65537f62ee5668e (diff)
parent	5cad143e545775acacee294049345c6a3868c51d (diff)