docs: 7.5.2 md5 sums

Build was broken by commit 9666529b5a

I'm not certain that this is entirely the correct fix since the demo
from bug #23774 seemed to work before the commit that broke the build.

Signed-off-by: Robert Noland <rnoland@2hip.net>
Signed-off-by: Brian Paul <brianp@vmware.com>

Makefile

@@ -182,7 +182,7 @@ ultrix-gcc:
 
 # Rules for making release tarballs
 
-VERSION=7.5
+VERSION=7.5.2
 DIRECTORY = Mesa-$(VERSION)
 LIB_NAME = MesaLib-$(VERSION)
 DEMO_NAME = MesaDemos-$(VERSION)
@@ -388,7 +388,17 @@ GLW_FILES = \
 	$(DIRECTORY)/src/glw/glw.pc.in			\
 	$(DIRECTORY)/src/glw/depend
 
+GLEW_FILES = \
+	$(DIRECTORY)/include/GL/glew.h			\
+	$(DIRECTORY)/include/GL/glxew.h			\
+	$(DIRECTORY)/include/GL/wglew.h			\
+	$(DIRECTORY)/src/glew/*.c			\
+	$(DIRECTORY)/src/glew/Makefile			\
+	$(DIRECTORY)/src/glew/SConscript		\
+	$(DIRECTORY)/src/glew/LICENSE.txt
+
 DEMO_FILES = \
+	$(GLEW_FILES)					\
 	$(DIRECTORY)/progs/beos/*.cpp			\
 	$(DIRECTORY)/progs/beos/Makefile		\
 	$(DIRECTORY)/progs/images/*.rgb			\

bin/mklib

@@ -971,5 +971,6 @@ esac
 #
 if [ ${INSTALLDIR} != "." ] ; then
     echo "mklib: Installing" ${FINAL_LIBS} "in" ${INSTALLDIR}
+    test -d ${INSTALLDIR} || mkdir -p ${INSTALLDIR}
     mv ${FINAL_LIBS} ${INSTALLDIR}/
 fi

common.py

@@ -33,9 +33,9 @@ else:
 	default_machine = _platform.machine()
 default_machine = _machine_map.get(default_machine, 'generic')
 
-if default_platform in ('linux', 'freebsd', 'darwin'):
+if default_platform in ('linux', 'freebsd'):
 	default_dri = 'yes'
-elif default_platform in ('winddk', 'windows', 'wince'):
+elif default_platform in ('winddk', 'windows', 'wince', 'darwin'):
 	default_dri = 'no'
 else:
 	default_dri = 'no'

configs/default

@@ -10,7 +10,7 @@ CONFIG_NAME = default
 # Version info
 MESA_MAJOR=7
 MESA_MINOR=5
-MESA_TINY=0
+MESA_TINY=2
 MESA_VERSION = $(MESA_MAJOR).$(MESA_MINOR).$(MESA_TINY)
 
 # external projects.  This should be useless now that we use libdrm.
@@ -23,6 +23,7 @@ HOST_CC = $(CC)
 CFLAGS = -O
 CXXFLAGS = -O
 LDFLAGS =
+HOST_CFLAGS = $(CFLAGS)
 GLU_CFLAGS = 
 
 # Compiler for building demos/tests/etc

configure.ac

@@ -121,6 +121,7 @@ AC_ARG_ENABLE([32-bit],
 if test "x$enable_32bit" = xyes; then
     if test "x$GCC" = xyes; then
         CFLAGS="$CFLAGS -m32"
+        ARCH_FLAGS="$ARCH_FLAGS -m32"
     fi
     if test "x$GXX" = xyes; then
         CXXFLAGS="$CXXFLAGS -m32"
@@ -1145,6 +1146,21 @@ AC_ARG_WITH([xorg-driver-dir],
     [XORG_DRIVER_INSTALL_DIR="${libdir}/xorg/modules/drivers"])
 AC_SUBST([XORG_DRIVER_INSTALL_DIR])
 
+AC_ARG_WITH([max-width],
+    [AS_HELP_STRING([--with-max-width=N],
+                    [Maximum framebuffer width (4096)])],
+    [DEFINES="${DEFINES} -DMAX_WIDTH=${withval}";
+     AS_IF([test "${withval}" -gt "4096"],
+           [AC_MSG_WARN([Large framebuffer: see s_tritemp.h comments.])])]
+)
+AC_ARG_WITH([max-height],
+    [AS_HELP_STRING([--with-max-height=N],
+                    [Maximum framebuffer height (4096)])],
+    [DEFINES="${DEFINES} -DMAX_HEIGHT=${withval}";
+     AS_IF([test "${withval}" -gt "4096"],
+           [AC_MSG_WARN([Large framebuffer: see s_tritemp.h comments.])])]
+)
+
 dnl
 dnl Gallium Intel configuration
 dnl
@@ -1172,7 +1188,7 @@ if test "x$enable_gallium_radeon" = xyes; then
 fi
 
 dnl
-dnl Gallium Radeon configuration
+dnl Gallium Nouveau configuration
 dnl
 AC_ARG_ENABLE([gallium-nouveau],
     [AS_HELP_STRING([--enable-gallium-nouveau],

docs/contents.html

@@ -39,7 +39,6 @@ a:visited {
 <ul>
 <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>
 
@@ -68,6 +67,7 @@ a:visited {
 <li><a href="repository.html" target="MainFrame">Source Code Repository</a>
 <li><a href="memory.html" target="MainFrame">DRI Memory Management</a>
 <li><a href="shading.html" target="MainFrame">Shading Language</a>
+<li><a href="glu.html" target="MainFrame">SGI's GLU</a>
 <li><a href="utilities.html" target="MainFrame">Utilities</a>
 <li><a href="helpwanted.html" target="MainFrame">Help Wanted</a>
 <li><a href="devinfo.html" target="MainFrame">Development Notes</a>

docs/download.html

@@ -9,14 +9,14 @@
 <H1>Downloading</H1>
 
 <p>
-Primary download site:
-<a href="http://sourceforge.net/project/showfiles.php?group_id=3"
-target="_parent">SourceForge</a>
+Primary Mesa download site:
+<a href="ftp://ftp.freedesktop.org/pub/mesa/"
+target="_parent">freedesktop.org</a> (FTP)
 </p>
 
 <p>
-When a new release is coming, release candidates (betas) can be found
-<a href="http://www.mesa3d.org/beta/">here</a>.
+When a new release is coming, release candidates (betas) may be found
+<a href="ftp://ftp.freedesktop.org/pub/mesa/beta/" target="_parent">here</a>.
 </p>
 
 

docs/faq.html

@@ -316,6 +316,19 @@ Basically, applying a translation of (0.375, 0.375, 0.0) to your coordinates
 will fix the problem.
 </p>
 
+<h2>3.6 How can I change the maximum framebuffer size in Mesa's
+<tt>swrast</tt> backend?</h2>
+<p>
+These can be overridden by using the <tt>--with-max-width</tt> and
+<tt>--with-max-height</tt> options.  The two need not be equal.
+</p><p>
+Do note that Mesa uses these values to size some internal buffers,
+so increasing these sizes will cause Mesa to require additional
+memory.  Furthermore, increasing these limits beyond <tt>4096</tt>
+may introduce rasterization artifacts; see the leading comments in
+<tt>src/mesa/swrast/s_tritemp.h</tt>.
+</p>
+
 <br>
 <br>
 

docs/news.html

@@ -11,6 +11,13 @@
 <H1>News</H1>
 
 
+<h2>September 3, 2009</h2>
+<p>
+<a href="relnotes-7.5.1.html">Mesa 7.5.1</a> is released.
+This is a bug-fix release which fixes bugs found in version 7.5.
+</p>
+
+
 <h2>July 17, 2009</h2>
 <p>
 <a href="relnotes-7.5.html">Mesa 7.5</a> is released.

docs/precompiled.html

@@ -9,17 +9,11 @@
 <H1>Precompiled Libraries</H1>
 
 <p>
-In general, precompiled libraries are not available.
-However, people occasionally prepare packages of precompiled libraries
-for some systems.
+In general, precompiled Mesa libraries are not available.
 </p>
-
-<H2>Mesa-6.0 for Solaris</H2>
-
 <p>
-Steve Christensen has submitted precompiled Mesa-6.0 libraries for
-Solaris at <a href="http://sunfreeware.com/" target="_parent">
-sunfreeware.com</a>.
+However, some Linux distros (such as Ubuntu) seem to closely track
+Mesa and often have the latest Mesa release available as an update.
 </p>
 
 </BODY>

docs/relnotes-7.5.1.html

@@ -0,0 +1,72 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.5.1 Release Notes, 3 September 2009</H1>
+
+<p>
+Mesa 7.5.1 is a bug-fix release fixing issues found since the 7.5 release.
+</p>
+<p>
+The main new feature of Mesa 7.5.x is the
+<a href="http://wiki.freedesktop.org/wiki/Software/gallium"
+target="_parent">Gallium3D</a> infrastructure.
+</p>
+<p>
+Mesa 7.5.1 implements the OpenGL 2.1 API, but the version reported by
+glGetString(GL_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 2.1.
+</p>
+<p>
+See the <a href="install.html">Compiling/Installing page</a> for prerequisites
+for DRI hardware acceleration.
+</p>
+
+
+<h2>MD5 checksums</h2>
+<pre>
+d7269e93bc7484430637d54ced250876  MesaLib-7.5.1.tar.gz
+877d6a4b24efc2b1d02aa553f262cba8  MesaLib-7.5.1.tar.bz2
+23f4fb757a05c8396425681234ae20e5  MesaLib-7.5.1.zip
+5af4bd113652108f5cec5113dad813f2  MesaDemos-7.5.1.tar.gz
+785402e3b9f0e335538fcc6bf19f6987  MesaDemos-7.5.1.tar.bz2
+950058cc6d6106e9c7d5876a03789fe9  MesaDemos-7.5.1.zip
+cb52ce2c93389c2711cbe8d857ec5303  MesaGLUT-7.5.1.tar.gz
+e3a9892e056d625c5353617a7c5b7e9c  MesaGLUT-7.5.1.tar.bz2
+da1de364df148c94b4994006191a1e69  MesaGLUT-7.5.1.zip
+</pre>
+
+
+<h2>New features</h2>
+<ul>
+<li>Added configure --with-max-width=W, --with-max-height=H options to specify
+    max framebuffer, viewport size.
+</ul>
+
+
+<h2>Bug fixes</h2>
+<ul>
+<li>Added missing GLEW library to MesaDemos tarballs.
+<li>Fixed swapbuffers jerkiness in Doom3/etc in Intel drivers.
+<li>Fixed front buffer rendering bug in Intel drivers.
+<li>Fixed minor GLX memory leaks.
+<li>Fixed some texture env / fragment program state bugs.
+<li>Fixed some Gallium glBlitFramebuffer() bugs
+<li>Empty glBegin/glEnd() pair could cause divide by zero (bug 23489)
+<li>Fixed Gallium glBitmap() Z position bug
+<li>Setting arrays of sampler uniforms did not work
+<li>Selection/Feedback mode didn't handle polygon culling correctly (bug 16866)
+<li>Fixed 32/64-bit cross compilation issue in gen_matypes.c
+<li>Fixed glXCreateGLXPixmap() for direct rendering.
+<li>Fixed Gallium glCopyPixels(GL_STENCIL_INDEX) mispositioned image bug.
+</ul>
+
+
+</body>
+</html>

docs/relnotes-7.5.2.html

@@ -0,0 +1,67 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.5.2 Release Notes, 28 September 2009</H1>
+
+<p>
+Mesa 7.5.2 is a bug-fix release fixing issues found since the 7.5.1 release.
+</p>
+<p>
+The main new feature of Mesa 7.5.x is the
+<a href="http://wiki.freedesktop.org/wiki/Software/gallium"
+target="_parent">Gallium3D</a> infrastructure.
+</p>
+<p>
+Mesa 7.5.2 implements the OpenGL 2.1 API, but the version reported by
+glGetString(GL_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 2.1.
+</p>
+<p>
+See the <a href="install.html">Compiling/Installing page</a> for prerequisites
+for DRI hardware acceleration.
+</p>
+
+
+<h2>MD5 checksums</h2>
+<pre>
+43a90191dd9f76cd65dcc1ac91f3be70  MesaLib-7.5.2.tar.gz
+94e47a499f1226803869c2e37a6a8e3a  MesaLib-7.5.2.tar.bz2
+1ecb822b567ad67a0617361d45206b67  MesaLib-7.5.2.zip
+2718fdce7e075911d6147beb8f27104b  MesaDemos-7.5.2.tar.gz
+4e0f5ccd58afe21eddcd94327d926e86  MesaDemos-7.5.2.tar.bz2
+f621f8c223b278d7c8e49a012d56ca25  MesaDemos-7.5.2.zip
+83c16c1d6bcfcc3f97aab5d2fe430b4c  MesaGLUT-7.5.2.tar.gz
+e5d03bedae369ea3705783573bb33813  MesaGLUT-7.5.2.tar.bz2
+e82ba28e00d653e6f437d32be8ca8481  MesaGLUT-7.5.2.zip
+</pre>
+
+
+<h2>New features</h2>
+<ul>
+<li>Detect B43 chipset in Intel driver
+</ul>
+
+
+<h2>Bug fixes</h2>
+<ul>
+<li>Assorted bug fixes for i965/i945 drivers
+<li>Fixed Gallium glDrawPixels(GL_STENCIL_INDEX) failure.
+<li>Fixed GLSL linker/preprocessor version directive issue seen in Wine
+    (such as bug 23946)
+<li>glUseProgram() is now compiled into display lists (bug 23746).
+<li>glUniform functions are now compiled into display lists
+<li>Auto mipmap generation didn't work reliably with Gallium.
+<li>Fixed random number usage in GLX code.
+<li>Fixed invalid GL_OUT_OF_MEMORY error sometimes raised by glTexSubImage2D
+    when using Gallium.
+</ul>
+
+</body>
+</html>

docs/relnotes-7.5.html

@@ -40,7 +40,15 @@ If you're especially concerned with stability you should probably look for
 
 <h2>MD5 checksums</h2>
 <pre>
-tbd
+553fd956e544727f30fbe249619b6286  MesaLib-7.5.tar.gz
+459f332551f6ebb86f384d21dd15e1f0  MesaLib-7.5.tar.bz2
+8c02c0e17a9025250d20424ae32f5163  MesaLib-7.5.zip
+a188da2886fa5496ea0c2cda602b2eeb  MesaDemos-7.5.tar.gz
+398ee8801814a00e47f6c2314e3dfddc  MesaDemos-7.5.tar.bz2
+15a0c8ae013c54335a26335e1a98d609  MesaDemos-7.5.zip
+81010147def5a644ba14f9bbb7a49a2a  MesaGLUT-7.5.tar.gz
+baa7a1e850b6e39bae58868fd0684004  MesaGLUT-7.5.tar.bz2
+265228418e4423fa328f2f5b7970cf08  MesaGLUT-7.5.zip
 </pre>
 
 

docs/relnotes.html

@@ -13,6 +13,8 @@ The release notes summarize what's new or changed in each Mesa release.
 </p>
 
 <UL>
+<LI><A HREF="relnotes-7.5.2.html">7.5.2 release notes</A>
+<LI><A HREF="relnotes-7.5.1.html">7.5.1 release notes</A>
 <LI><A HREF="relnotes-7.5.html">7.5 release notes</A>
 <LI><A HREF="relnotes-7.4.4.html">7.4.4 release notes</A>
 <LI><A HREF="relnotes-7.4.3.html">7.4.3 release notes</A>

include/EGL/eglplatform.h

@@ -69,7 +69,8 @@ typedef HWND	NativeWindowType;
 typedef HBITMAP NativePixmapType;
 /** END Added for Windows **/
 
-#elif defined(__gnu_linux__) || defined(__FreeBSD__) || defined(__sun)
+#elif defined(__gnu_linux__) || defined(__FreeBSD__) || defined(__sun) || defined(__APPLE__)
+
 
 /** BEGIN Added for X (Mesa) **/
 #ifndef EGLAPI

progs/demos/fbotexture.c

@@ -498,7 +498,7 @@ SetupFunctionPointers(void)
  * Make FBO to render into given texture.
  */
 static GLuint
-MakeFBO_RenderTexture(GLuint TexObj)
+MakeFBO_RenderTexture(GLuint texObj)
 {
    GLuint fb;
    GLint sizeFudge = 0;
@@ -507,7 +507,7 @@ MakeFBO_RenderTexture(GLuint TexObj)
    glBindFramebuffer_func(GL_FRAMEBUFFER_EXT, fb);
    /* Render color to texture */
    glFramebufferTexture2D_func(GL_FRAMEBUFFER_EXT, GL_COLOR_ATTACHMENT0_EXT,
-                             TexTarget, TexObj, TextureLevel);
+                             TexTarget, texObj, TextureLevel);
 
    if (Use_ARB_fbo) {
       /* use a smaller depth buffer to see what happens */
@@ -541,7 +541,7 @@ MakeFBO_RenderTexture(GLuint TexObj)
 
    /* queries */
    {
-      GLint bits, w, h;
+      GLint bits, w, h, name;
 
       glBindRenderbuffer_func(GL_RENDERBUFFER_EXT, DepthRB);
       glGetRenderbufferParameteriv_func(GL_RENDERBUFFER_EXT,
@@ -559,8 +559,28 @@ MakeFBO_RenderTexture(GLuint TexObj)
       glGetRenderbufferParameteriv_func(GL_RENDERBUFFER_EXT,
                                       GL_RENDERBUFFER_STENCIL_SIZE_EXT, &bits);
       printf("Stencil renderbuffer size = %d bits\n", bits);
-   }
 
+      glGetFramebufferAttachmentParameteriv_func(GL_FRAMEBUFFER_EXT,
+                                      GL_COLOR_ATTACHMENT0,
+                                      GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT,
+                                      &name);
+      printf("Render to texture name: %d\n", texObj);
+      printf("Color attachment[0] name: %d\n", name);
+      assert(texObj == name);
+
+      glGetFramebufferAttachmentParameteriv_func(GL_FRAMEBUFFER_EXT,
+                                      GL_STENCIL_ATTACHMENT,
+                                      GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT,
+                                      &name);
+      printf("Stencil attachment name: %d\n", name);
+
+      glGetFramebufferAttachmentParameteriv_func(GL_FRAMEBUFFER_EXT,
+                                      GL_DEPTH_ATTACHMENT,
+                                      GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME_EXT,
+                                      &name);
+      printf("Depth attachment name: %d\n", name);
+
+   }
    /* bind the regular framebuffer */
    glBindFramebuffer_func(GL_FRAMEBUFFER_EXT, 0);
 

progs/glsl/Makefile

@@ -32,6 +32,7 @@ DEMO_SOURCES = \
 	pointcoord.c \
 	points.c \
 	samplers.c \
+	samplers_array.c \
 	shadow_sampler.c \
 	skinning.c \
 	texaaline.c \
@@ -188,7 +189,8 @@ samplers.o: $(UTIL_HEADERS)
 samplers: samplers.o $(UTIL_OBJS)
 
 
-samplers_array.o: $(UTIL_HEADERS)
+samplers_array.o: samplers.c $(UTIL_HEADERS)
+	$(APP_CC) -I$(INCDIR) $(CFLAGS) -DSAMPLERS_ARRAY $< -c -o $@
 
 samplers_array: samplers_array.o $(UTIL_OBJS)
 

progs/glsl/samplers.c

@@ -211,10 +211,18 @@ InitTextures(void)
       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;
+            if (x < size / 2) {
+               texImage[k + 0] = intensity;
+               texImage[k + 1] = intensity;
+               texImage[k + 2] = 0;
+               texImage[k + 3] = 255;
+            }
+            else {
+               texImage[k + 0] = 255 - intensity;
+               texImage[k + 1] = 0;
+               texImage[k + 2] = 0;
+               texImage[k + 3] = 255;
+            }
          }
       }
 

progs/tests/fbotest1.c

@@ -127,7 +127,7 @@ Init( void )
 
    if (!glutExtensionSupported("GL_EXT_framebuffer_object")) {
       printf("GL_EXT_framebuffer_object not found!\n");
-      /*exit(0);*/
+      exit(0);
    }
    printf("GL_RENDERER = %s\n", (char *) glGetString(GL_RENDERER));
 

progs/tests/texwrap.c

@@ -258,8 +258,26 @@ static void Init( void )
    }
 
    glBindTexture(GL_TEXTURE_2D, BORDER_TEXTURE);
+#ifdef TEST_PBO_DLIST
+   /* test fetching teximage from PBO in display list */
+   {
+      GLuint b = 42, l = 10;
+
+      glBindBufferARB(GL_PIXEL_UNPACK_BUFFER, b);
+      glBufferDataARB(GL_PIXEL_UNPACK_BUFFER, sizeof(BorderImage),
+                      BorderImage, GL_STREAM_DRAW);
+
+      glNewList(l, GL_COMPILE);
+      glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, SIZE+2, SIZE+2, 1,
+                GL_RGBA, GL_UNSIGNED_BYTE, (void *) 0/* BorderImage*/);
+      glEndList();
+      glCallList(l);
+      glBindBufferARB(GL_PIXEL_UNPACK_BUFFER, 0);
+   }
+#else
    glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, SIZE+2, SIZE+2, 1,
                 GL_RGBA, GL_UNSIGNED_BYTE, (void *) BorderImage);
+#endif
 
    for (i = 0; i < SIZE; i++) {
       for (j = 0; j < SIZE; j++) {

progs/trivial/.gitignore

@@ -5,10 +5,17 @@ clear-random
 clear-repeat
 clear-scissor
 clear-undefined
+createwin
+dlist-begin-call-end
 dlist-dangling
 dlist-degenerate
 dlist-edgeflag
 dlist-edgeflag-dangling
+dlist-flat-tri
+dlist-mat-tri
+dlist-recursive-call
+dlist-tri-flat-tri
+dlist-tri-mat-tri
 draw2arrays
 drawarrays
 drawelements
@@ -30,6 +37,7 @@ lineloop
 lineloop-clip
 lineloop-elts
 linestrip
+linestrip-clip
 linestrip-flat-stipple
 linestrip-stipple
 linestrip-stipple-wide
@@ -70,8 +78,10 @@ quadstrip-cont
 quadstrip-flat
 readtex.c
 readtex.h
+readpixels
 tri
 tri-alpha
+tri-alpha-tex
 tri-array-interleaved
 tri-blend
 tri-blend-color
@@ -79,6 +89,7 @@ tri-blend-max
 tri-blend-min
 tri-blend-revsub
 tri-blend-sub
+tri-clear
 tri-clip
 tri-cull
 tri-cull-both
@@ -93,6 +104,7 @@ tri-fog
 tri-fp
 tri-fp-const-imm
 tri-lit
+tri-lit-material
 tri-logicop-none
 tri-logicop-xor
 tri-mask-tri
@@ -101,6 +113,7 @@ tri-orig
 tri-query
 tri-repeat
 tri-scissor-tri
+tri-square
 tri-stencil
 tri-stipple
 tri-tex
@@ -110,6 +123,7 @@ tri-unfilled
 tri-unfilled-clip
 tri-unfilled-edgeflag
 tri-unfilled-fog
+tri-unfilled-point
 tri-unfilled-smooth
 tri-unfilled-tri
 tri-unfilled-tri-lit
@@ -118,6 +132,7 @@ tri-unfilled-userclip-stip
 tri-userclip
 tri-viewport
 tri-z
+tri-z-9
 tri-z-eq
 trifan
 trifan-flat

progs/trivial/Makefile

@@ -18,6 +18,7 @@ SOURCES = \
 	clear-repeat.c \
 	clear-random.c \
 	clear.c \
+	createwin.c \
 	dlist-dangling.c \
 	dlist-flat-tri.c \
 	dlist-mat-tri.c \
@@ -48,6 +49,7 @@ SOURCES = \
 	lineloop-clip.c \
 	lineloop-elts.c \
 	lineloop.c \
+	linestrip-clip.c \
 	linestrip-flat-stipple.c \
 	linestrip-stipple-wide.c \
 	linestrip-stipple.c \
@@ -87,7 +89,9 @@ SOURCES = \
 	quadstrip-cont.c \
 	quadstrip-flat.c \
 	quadstrip.c \
+	readpixels.c \
 	tri-alpha.c \
+	tri-alpha-tex.c \
 	tri-array-interleaved.c \
 	tri-blend-color.c \
 	tri-blend-max.c \
@@ -95,6 +99,7 @@ SOURCES = \
 	tri-blend-revsub.c \
 	tri-blend-sub.c \
 	tri-blend.c \
+	tri-clear.c \
 	tri-clip.c \
 	tri-cull-both.c \
 	tri-cull.c \
@@ -117,6 +122,7 @@ SOURCES = \
 	tri-query.c \
 	tri-repeat.c \
 	tri-scissor-tri.c \
+	tri-square.c \
 	tri-stencil.c \
 	tri-stipple.c \
 	tri-multitex-vbo.c \
@@ -126,6 +132,7 @@ SOURCES = \
 	tri-unfilled-fog.c \
 	tri-unfilled-edgeflag.c \
 	tri-unfilled-clip.c \
+	tri-unfilled-point.c \
 	tri-unfilled-smooth.c \
 	tri-unfilled-tri.c \
 	tri-unfilled-tri-lit.c \
@@ -134,6 +141,7 @@ SOURCES = \
 	tri-unfilled.c \
 	tri-userclip.c \
 	tri-viewport.c \
+	tri-z-9.c \
 	tri-z-eq.c \
 	tri-z.c \
 	tri.c \

progs/vp/vp-tris.c

@@ -119,6 +119,12 @@ static void Init( void )
       glBindProgramARB(GL_VERTEX_PROGRAM_ARB, prognum);
       glProgramStringARB(GL_VERTEX_PROGRAM_ARB, GL_PROGRAM_FORMAT_ASCII_ARB,
 		        sz, (const GLubyte *) buf);
+      if (glGetError()) {
+         printf("Program failed to compile:\n%s\n", buf);
+         printf("Error: %s\n",
+                (char *) glGetString(GL_PROGRAM_ERROR_STRING_ARB));
+         exit(1);
+      }
       assert(glIsProgramARB(prognum));
    }
 

progs/xdemos/glxcontexts.c

@@ -378,6 +378,9 @@ make_window( Display *dpy, const char *name,
    scrnum = DefaultScreen( dpy );
    root = RootWindow( dpy, scrnum );
 
+   if (visinfo)
+	   XFree(visinfo);
+
    visinfo = glXChooseVisual( dpy, scrnum, attribs );
    if (!visinfo) {
       printf("Error: couldn't get an RGB, Double-buffered visual\n");

progs/xdemos/glxswapcontrol.c

@@ -587,6 +587,9 @@ event_loop(Display *dpy, Window win)
       angle += 2.0;
 
       draw();
+
+      glXSwapBuffers(dpy, win);
+
       if ( get_frame_usage != NULL ) {
 	 GLfloat   temp;
 	 
@@ -594,8 +597,6 @@ event_loop(Display *dpy, Window win)
 	 frame_usage += temp;
       }
 
-      glXSwapBuffers(dpy, win);
-
       /* calc framerate */
       {
          static int t0 = -1;

src/egl/drivers/demo/Makefile

@@ -4,7 +4,7 @@ TOP = ../../../..
 include $(TOP)/configs/current
 
 
-INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/egl/main
+INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/egl/main $(X11_INCLUDES)
 
 
 SOURCES = demo.c

src/egl/main/Makefile

@@ -4,7 +4,7 @@ TOP = ../../..
 include $(TOP)/configs/current
 
 
-INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/mesa/glapi
+INCLUDE_DIRS = -I$(TOP)/include -I$(TOP)/src/mesa/glapi $(X11_INCLUDES)
 
 HEADERS = \
 	eglconfig.h \

src/gallium/auxiliary/draw/draw_pipe.c

@@ -158,6 +158,60 @@ static void do_triangle( struct draw_context *draw,
 
 
 
+#define QUAD(i0,i1,i2,i3)                       \
+   do_triangle( draw,                           \
+                ( DRAW_PIPE_RESET_STIPPLE |     \
+                  DRAW_PIPE_EDGE_FLAG_0 |       \
+                  DRAW_PIPE_EDGE_FLAG_2 ),      \
+                verts + stride * elts[i0],      \
+                verts + stride * elts[i1],      \
+                verts + stride * elts[i3]);     \
+   do_triangle( draw,                           \
+                ( DRAW_PIPE_EDGE_FLAG_0 |       \
+                  DRAW_PIPE_EDGE_FLAG_1 ),      \
+                verts + stride * elts[i1],      \
+                verts + stride * elts[i2],      \
+                verts + stride * elts[i3])
+
+#define TRIANGLE(flags,i0,i1,i2)                                        \
+   do_triangle( draw,                                                   \
+                elts[i0],  /* flags */                          \
+                verts + stride * (elts[i0] & ~DRAW_PIPE_FLAG_MASK),     \
+                verts + stride * elts[i1],                              \
+                verts + stride * elts[i2])
+
+#define LINE(flags,i0,i1)                                       \
+   do_line( draw,                                               \
+            elts[i0],                                   \
+            verts + stride * (elts[i0] & ~DRAW_PIPE_FLAG_MASK), \
+            verts + stride * elts[i1])
+
+#define POINT(i0)                               \
+   do_point( draw,                              \
+             verts + stride * elts[i0] )
+
+#define FUNC pipe_run
+#define ARGS                                    \
+    struct draw_context *draw,                  \
+    unsigned prim,                              \
+    struct vertex_header *vertices,             \
+    unsigned stride,                            \
+    const ushort *elts
+
+#define LOCAL_VARS                                           \
+   char *verts = (char *)vertices;                           \
+   boolean flatfirst = (draw->rasterizer->flatshade &&       \
+                        draw->rasterizer->flatshade_first);  \
+   unsigned i;                                               \
+   ushort flags
+
+#define FLUSH
+
+#include "draw_pt_decompose.h"
+#undef ARGS
+#undef LOCAL_VARS
+
+
 
 /* Code to run the pipeline on a fairly arbitary collection of vertices.
  *
@@ -178,34 +232,12 @@ void draw_pipeline_run( struct draw_context *draw,
                         unsigned count )
 {
    char *verts = (char *)vertices;
-   unsigned i;
 
    draw->pipeline.verts = verts;
    draw->pipeline.vertex_stride = stride;
    draw->pipeline.vertex_count = vertex_count;
    
-   switch (prim) {
-   case PIPE_PRIM_POINTS:
-      for (i = 0; i < count; i++) 
-         do_point( draw, 
-                   verts + stride * elts[i] );
-      break;
-   case PIPE_PRIM_LINES:
-      for (i = 0; i+1 < count; i += 2) 
-         do_line( draw, 
-                  elts[i+0],  /* flags */
-                  verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK),
-                  verts + stride * elts[i+1]);
-      break;
-   case PIPE_PRIM_TRIANGLES:
-      for (i = 0; i+2 < count; i += 3)
-         do_triangle( draw, 
-                      elts[i+0],  /* flags */
-                      verts + stride * (elts[i+0] & ~DRAW_PIPE_FLAG_MASK),
-                      verts + stride * elts[i+1],
-                      verts + stride * elts[i+2]);
-      break;
-   }
+   pipe_run(draw, prim, vertices, stride, elts, count);
    
    draw->pipeline.verts = NULL;
    draw->pipeline.vertex_count = 0;

src/gallium/auxiliary/tgsi/tgsi_build.c

@@ -477,6 +477,8 @@ tgsi_default_full_instruction( void )
       full_instruction.FullSrcRegisters[i] = tgsi_default_full_src_register();
    }
 
+   full_instruction.Flags = 0x0;
+
    return full_instruction;
 }
 

src/gallium/auxiliary/tgsi/tgsi_exec.c

@@ -62,6 +62,9 @@
 
 #define FAST_MATH 1
 
+/** for tgsi_full_instruction::Flags */
+#define SOA_DEPENDENCY_FLAG 0x1
+
 #define TILE_TOP_LEFT     0
 #define TILE_TOP_RIGHT    1
 #define TILE_BOTTOM_LEFT  2
@@ -182,7 +185,7 @@ print_temp(const struct tgsi_exec_machine *mach, uint index)
  *   MOV t3, t2;
  * The second instruction will have the wrong value for t0 if executed as-is.
  */
-static boolean
+boolean
 tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst)
 {
    uint i, chan;
@@ -328,20 +331,25 @@ tgsi_exec_machine_bind_shader(
                                    * sizeof(struct tgsi_full_instruction));
             maxInstructions += 10;
          }
+
+         if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) {
+            uint opcode = parse.FullToken.FullInstruction.Instruction.Opcode;
+            parse.FullToken.FullInstruction.Flags = SOA_DEPENDENCY_FLAG;
+            /* XXX we only handle SOA dependencies properly for MOV/SWZ
+             * at this time!
+             */
+            if (opcode != TGSI_OPCODE_MOV && opcode != TGSI_OPCODE_SWZ) {
+               debug_printf("Warning: SOA dependency in instruction"
+                            " is not handled:\n");
+               tgsi_dump_instruction(&parse.FullToken.FullInstruction,
+                                     numInstructions);
+            }
+         }
+
          memcpy(instructions + numInstructions,
                 &parse.FullToken.FullInstruction,
                 sizeof(instructions[0]));
 
-#if 0
-         if (tgsi_check_soa_dependencies(&parse.FullToken.FullInstruction)) {
-            debug_printf("SOA dependency in instruction:\n");
-            tgsi_dump_instruction(&parse.FullToken.FullInstruction,
-                                  numInstructions);
-         }
-#else
-         (void) tgsi_check_soa_dependencies;
-#endif
-
          numInstructions++;
          break;
 
@@ -2018,9 +2026,23 @@ exec_instruction(
 
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:
-      FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
-         FETCH( &r[0], 0, chan_index );
-         STORE( &r[0], 0, chan_index );
+      if (inst->Flags & SOA_DEPENDENCY_FLAG) {
+         /* Do all fetches into temp regs, then do all stores to avoid
+          * intermediate/accidental clobbering.  This could be done all the
+          * time for MOV but for other instructions we'll need more temps...
+          */
+         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+            FETCH( &r[chan_index], 0, chan_index );
+         }
+         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+            STORE( &r[chan_index], 0, chan_index );
+         }
+      }
+      else {
+         FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) {
+            FETCH( &r[0], 0, chan_index );
+            STORE( &r[0], 0, chan_index );
+         }
       }
       break;
 

src/gallium/auxiliary/tgsi/tgsi_exec.h

@@ -272,6 +272,10 @@ void
 tgsi_exec_machine_free_data(struct tgsi_exec_machine *mach);
 
 
+boolean
+tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst);
+
+
 static INLINE void
 tgsi_set_kill_mask(struct tgsi_exec_machine *mach, unsigned mask)
 {

src/gallium/auxiliary/tgsi/tgsi_parse.h

@@ -91,6 +91,7 @@ struct tgsi_full_instruction
    struct tgsi_instruction_ext_texture InstructionExtTexture;
    struct tgsi_full_dst_register       FullDstRegisters[TGSI_FULL_MAX_DST_REGISTERS];
    struct tgsi_full_src_register       FullSrcRegisters[TGSI_FULL_MAX_SRC_REGISTERS];
+   uint Flags;  /**< user-defined usage */
 };
 
 union tgsi_full_token

src/gallium/auxiliary/tgsi/tgsi_ppc.c

@@ -1107,6 +1107,15 @@ static int
 emit_instruction(struct gen_context *gen,
                  struct tgsi_full_instruction *inst)
 {
+
+   /* we don't handle saturation/clamping yet */
+   if (inst->Instruction.Saturate != TGSI_SAT_NONE)
+      return 0;
+
+   /* need to use extra temps to fix SOA dependencies : */
+   if (tgsi_check_soa_dependencies(inst))
+      return FALSE;
+
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_SWZ:

src/gallium/auxiliary/tgsi/tgsi_sse2.c

@@ -1502,6 +1502,14 @@ emit_instruction(
    if (indirect_temp_reference(inst))
       return FALSE;
 
+   /* we don't handle saturation/clamping yet */
+   if (inst->Instruction.Saturate != TGSI_SAT_NONE)
+      return FALSE;
+
+   /* need to use extra temps to fix SOA dependencies : */
+   if (tgsi_check_soa_dependencies(inst))
+      return FALSE;
+
    switch (inst->Instruction.Opcode) {
    case TGSI_OPCODE_ARL:
       FOR_EACH_DST0_ENABLED_CHANNEL( *inst, chan_index ) {

src/gallium/auxiliary/util/u_blit.c

@@ -45,6 +45,7 @@
 #include "util/u_math.h"
 #include "util/u_memory.h"
 #include "util/u_simple_shaders.h"
+#include "util/u_surface.h"
 
 #include "cso_cache/cso_context.h"
 
@@ -61,7 +62,7 @@ struct blit_state
    struct pipe_viewport_state viewport;
 
    void *vs;
-   void *fs;
+   void *fs[TGSI_WRITEMASK_XYZW + 1];
 
    struct pipe_buffer *vbuf;  /**< quad vertices */
    unsigned vbuf_slot;
@@ -124,7 +125,7 @@ util_create_blit(struct pipe_context *pipe, struct cso_context *cso)
    }
 
    /* fragment shader */
-   ctx->fs = util_make_fragment_tex_shader(pipe);
+   ctx->fs[TGSI_WRITEMASK_XYZW] = util_make_fragment_tex_shader(pipe);
    ctx->vbuf = NULL;
 
    /* init vertex data that doesn't change */
@@ -145,9 +146,13 @@ void
 util_destroy_blit(struct blit_state *ctx)
 {
    struct pipe_context *pipe = ctx->pipe;
+   unsigned i;
 
    pipe->delete_vs_state(pipe, ctx->vs);
-   pipe->delete_fs_state(pipe, ctx->fs);
+
+   for (i = 0; i < Elements(ctx->fs); i++)
+      if (ctx->fs[i])
+         pipe->delete_fs_state(pipe, ctx->fs[i]);
 
    pipe_buffer_reference(&ctx->vbuf, NULL);
 
@@ -155,7 +160,11 @@ util_destroy_blit(struct blit_state *ctx)
 }
 
 
-static unsigned get_next_slot( struct blit_state *ctx )
+/**
+ * Get offset of next free slot in vertex buffer for quad vertices.
+ */
+static unsigned
+get_next_slot( struct blit_state *ctx )
 {
    const unsigned max_slots = 4096 / sizeof ctx->vertices;
 
@@ -173,7 +182,6 @@ static unsigned get_next_slot( struct blit_state *ctx )
 }
                                
 
-
 /**
  * Setup vertex data for the textured quad we'll draw.
  * Note: y=0=top
@@ -260,20 +268,50 @@ setup_vertex_data_tex(struct blit_state *ctx,
 
    return offset;
 }
+
+
+/**
+ * \return TRUE if two regions overlap, FALSE otherwise
+ */
+static boolean
+regions_overlap(int srcX0, int srcY0,
+                int srcX1, int srcY1,
+                int dstX0, int dstY0,
+                int dstX1, int dstY1)
+{
+   if (MAX2(srcX0, srcX1) < MIN2(dstX0, dstX1))
+      return FALSE; /* src completely left of dst */
+
+   if (MAX2(dstX0, dstX1) < MIN2(srcX0, srcX1))
+      return FALSE; /* dst completely left of src */
+
+   if (MAX2(srcY0, srcY1) < MIN2(dstY0, dstY1))
+      return FALSE; /* src completely above dst */
+
+   if (MAX2(dstY0, dstY1) < MIN2(srcY0, srcY1))
+      return FALSE; /* dst completely above src */
+
+   return TRUE; /* some overlap */
+}
+
+
 /**
  * Copy pixel block from src surface to dst surface.
  * Overlapping regions are acceptable.
+ * Flipping and stretching are supported.
+ * XXX what about clipping???
  * XXX need some control over blitting Z and/or stencil.
  */
 void
-util_blit_pixels(struct blit_state *ctx,
-                 struct pipe_surface *src,
-                 int srcX0, int srcY0,
-                 int srcX1, int srcY1,
-                 struct pipe_surface *dst,
-                 int dstX0, int dstY0,
-                 int dstX1, int dstY1,
-                 float z, uint filter)
+util_blit_pixels_writemask(struct blit_state *ctx,
+                           struct pipe_surface *src,
+                           int srcX0, int srcY0,
+                           int srcX1, int srcY1,
+                           struct pipe_surface *dst,
+                           int dstX0, int dstY0,
+                           int dstX1, int dstY1,
+                           float z, uint filter,
+                           uint writemask)
 {
    struct pipe_context *pipe = ctx->pipe;
    struct pipe_screen *screen = pipe->screen;
@@ -285,10 +323,41 @@ util_blit_pixels(struct blit_state *ctx,
    const int srcLeft = MIN2(srcX0, srcX1);
    const int srcTop = MIN2(srcY0, srcY1);
    unsigned offset;
+   boolean overlap;
 
    assert(filter == PIPE_TEX_MIPFILTER_NEAREST ||
           filter == PIPE_TEX_MIPFILTER_LINEAR);
 
+   assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE_2D,
+                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
+   assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
+                                      PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
+
+   /* do the regions overlap? */
+   overlap = util_same_surface(src, dst) &&
+      regions_overlap(srcX0, srcY0, srcX1, srcY1,
+                      dstX0, dstY0, dstX1, dstY1);
+
+   /*
+    * Check for simple case:  no format conversion, no flipping, no stretching,
+    * no overlapping.
+    * Filter mode should not matter since there's no stretching.
+    */
+   if (dst->format == src->format &&
+       srcX0 < srcX1 &&
+       dstX0 < dstX1 &&
+       srcY0 < srcY1 &&
+       dstY0 < dstY1 &&
+       (dstX1 - dstX0) == (srcX1 - srcX0) &&
+       (dstY1 - dstY0) == (srcY1 - srcY0) &&
+       !overlap) {
+      pipe->surface_copy(pipe,
+			 dst, dstX0, dstY0, /* dest */
+			 src, srcX0, srcY0, /* src */
+			 srcW, srcH);       /* size */
+      return;
+   }
+   
    if (srcLeft != srcX0) {
       /* left-right flip */
       int tmp = dstX0;
@@ -303,20 +372,6 @@ util_blit_pixels(struct blit_state *ctx,
       dstY1 = tmp;
    }
 
-   assert(screen->is_format_supported(screen, src->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
-   assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
-                                      PIPE_TEXTURE_USAGE_SAMPLER, 0));
-
-   if(dst->format == src->format && (dstX1 - dstX0) == srcW && (dstY1 - dstY0) == srcH) {
-      /* FIXME: this will most surely fail for overlapping rectangles */
-      pipe->surface_copy(pipe,
-			 dst, dstX0, dstY0,   /* dest */
-			 src, srcX0, srcY0, /* src */
-			 srcW, srcH);     /* size */
-      return;
-   }
-   
    assert(screen->is_format_supported(screen, dst->format, PIPE_TEXTURE_2D,
                                       PIPE_TEXTURE_USAGE_RENDER_TARGET, 0));
 
@@ -376,8 +431,11 @@ util_blit_pixels(struct blit_state *ctx,
    /* texture */
    cso_set_sampler_textures(ctx->cso, 1, &tex);
 
+   if (ctx->fs[writemask] == NULL)
+      ctx->fs[writemask] = util_make_fragment_tex_shader_writemask(pipe, writemask);
+
    /* shaders */
-   cso_set_fragment_shader_handle(ctx->cso, ctx->fs);
+   cso_set_fragment_shader_handle(ctx->cso, ctx->fs[writemask]);
    cso_set_vertex_shader_handle(ctx->cso, ctx->vs);
 
    /* drawing dest */
@@ -412,6 +470,27 @@ util_blit_pixels(struct blit_state *ctx,
 }
 
 
+void
+util_blit_pixels(struct blit_state *ctx,
+                 struct pipe_surface *src,
+                 int srcX0, int srcY0,
+                 int srcX1, int srcY1,
+                 struct pipe_surface *dst,
+                 int dstX0, int dstY0,
+                 int dstX1, int dstY1,
+                 float z, uint filter )
+{
+   util_blit_pixels_writemask( ctx, src, 
+                               srcX0, srcY0,
+                               srcX1, srcY1,
+                               dst,
+                               dstX0, dstY0,
+                               dstX1, dstY1,
+                               z, filter,
+                               TGSI_WRITEMASK_XYZW );
+}
+
+
 /* Release vertex buffer at end of frame to avoid synchronous
  * rendering.
  */
@@ -485,7 +564,7 @@ util_blit_pixels_tex(struct blit_state *ctx,
    cso_set_sampler_textures(ctx->cso, 1, &tex);
 
    /* shaders */
-   cso_set_fragment_shader_handle(ctx->cso, ctx->fs);
+   cso_set_fragment_shader_handle(ctx->cso, ctx->fs[TGSI_WRITEMASK_XYZW]);
    cso_set_vertex_shader_handle(ctx->cso, ctx->vs);
 
    /* drawing dest */

src/gallium/auxiliary/util/u_blit.h

@@ -60,6 +60,17 @@ util_blit_pixels(struct blit_state *ctx,
                  int dstX1, int dstY1,
                  float z, uint filter);
 
+void
+util_blit_pixels_writemask(struct blit_state *ctx,
+                           struct pipe_surface *src,
+                           int srcX0, int srcY0,
+                           int srcX1, int srcY1,
+                           struct pipe_surface *dst,
+                           int dstX0, int dstY0,
+                           int dstX1, int dstY1,
+                           float z, uint filter,
+                           uint writemask);
+
 extern void
 util_blit_pixels_tex(struct blit_state *ctx,
                      struct pipe_texture *tex,

src/gallium/auxiliary/util/u_debug.c

@@ -143,11 +143,9 @@ void _debug_vprintf(const char *format, va_list ap)
 #elif defined(PIPE_SUBSYSTEM_WINDOWS_MINIPORT)
    /* TODO */
 #else /* !PIPE_SUBSYSTEM_WINDOWS */
-#ifdef DEBUG
    fflush(stdout);
    vfprintf(stderr, format, ap);
 #endif
-#endif
 }
 
 

src/gallium/auxiliary/util/u_gen_mipmap.c

@@ -921,11 +921,19 @@ static void
 format_to_type_comps(enum pipe_format pformat,
                      enum dtype *datatype, uint *comps)
 {
+   /* XXX I think this could be implemented in terms of the pf_*() functions */
    switch (pformat) {
    case PIPE_FORMAT_A8R8G8B8_UNORM:
    case PIPE_FORMAT_X8R8G8B8_UNORM:
    case PIPE_FORMAT_B8G8R8A8_UNORM:
    case PIPE_FORMAT_B8G8R8X8_UNORM:
+   case PIPE_FORMAT_R8G8B8A8_SRGB:
+   case PIPE_FORMAT_R8G8B8X8_SRGB:
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+   case PIPE_FORMAT_X8R8G8B8_SRGB:
+   case PIPE_FORMAT_B8G8R8A8_SRGB:
+   case PIPE_FORMAT_B8G8R8X8_SRGB:
+   case PIPE_FORMAT_R8G8B8_SRGB:
       *datatype = DTYPE_UBYTE;
       *comps = 4;
       return;
@@ -942,12 +950,14 @@ format_to_type_comps(enum pipe_format pformat,
       *comps = 3;
       return;
    case PIPE_FORMAT_L8_UNORM:
+   case PIPE_FORMAT_L8_SRGB:
    case PIPE_FORMAT_A8_UNORM:
    case PIPE_FORMAT_I8_UNORM:
       *datatype = DTYPE_UBYTE;
       *comps = 1;
       return;
    case PIPE_FORMAT_A8L8_UNORM:
+   case PIPE_FORMAT_A8L8_SRGB:
       *datatype = DTYPE_UBYTE;
       *comps = 2;
       return;
@@ -1509,6 +1519,17 @@ util_gen_mipmap(struct gen_mipmap_state *ctx,
    uint zslice = 0;
    uint offset;
 
+   /* The texture object should have room for the levels which we're
+    * about to generate.
+    */
+   assert(lastLevel <= pt->last_level);
+
+   /* If this fails, why are we here? */
+   assert(lastLevel > baseLevel);
+
+   assert(filter == PIPE_TEX_FILTER_LINEAR ||
+          filter == PIPE_TEX_FILTER_NEAREST);
+
    /* check if we can render in the texture's format */
    if (!screen->is_format_supported(screen, pt->format, PIPE_TEXTURE_2D,
                                     PIPE_TEXTURE_USAGE_RENDER_TARGET, 0)) {

src/gallium/auxiliary/util/u_simple_shaders.c

@@ -152,11 +152,14 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
 
 /**
  * Make simple fragment texture shader:
- *  TEX OUT[0], IN[0], SAMP[0], 2D;
+ *  IMM {0,0,0,1}                         // (if writemask != 0xf)
+ *  MOV OUT[0], IMM[0]                    // (if writemask != 0xf)
+ *  TEX OUT[0].writemask, IN[0], SAMP[0], 2D;
  *  END;
  */
 void *
-util_make_fragment_tex_shader(struct pipe_context *pipe)
+util_make_fragment_tex_shader_writemask(struct pipe_context *pipe,
+                                        unsigned writemask )
 {
    struct pipe_shader_state shader;
    struct tgsi_token tokens[100];
@@ -217,12 +220,43 @@ util_make_fragment_tex_shader(struct pipe_context *pipe)
                                      header,
                                      Elements(tokens) - ti);
 
+
+   if (writemask != TGSI_WRITEMASK_XYZW) {
+      struct tgsi_full_immediate imm;
+      static const float value[4] = { 0, 0, 0, 1 };
+
+      imm = tgsi_default_full_immediate();
+      imm.Immediate.NrTokens += 4;
+      imm.Immediate.DataType = TGSI_IMM_FLOAT32;
+      imm.u.Pointer = value;
+
+      ti += tgsi_build_full_immediate(&imm,
+                                      &tokens[ti],
+                                      header,
+                                      Elements(tokens) - ti );
+
+      /* MOV instruction */
+      inst = tgsi_default_full_instruction();
+      inst.Instruction.Opcode = TGSI_OPCODE_MOV;
+      inst.Instruction.NumDstRegs = 1;
+      inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
+      inst.FullDstRegisters[0].DstRegister.Index = 0;
+      inst.Instruction.NumSrcRegs = 1;
+      inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_IMMEDIATE;
+      inst.FullSrcRegisters[0].SrcRegister.Index = 0;
+      ti += tgsi_build_full_instruction(&inst,
+                                        &tokens[ti],
+                                        header,
+                                        Elements(tokens) - ti );
+   }
+
    /* TEX instruction */
    inst = tgsi_default_full_instruction();
    inst.Instruction.Opcode = TGSI_OPCODE_TEX;
    inst.Instruction.NumDstRegs = 1;
    inst.FullDstRegisters[0].DstRegister.File = TGSI_FILE_OUTPUT;
    inst.FullDstRegisters[0].DstRegister.Index = 0;
+   inst.FullDstRegisters[0].DstRegister.WriteMask = writemask;
    inst.Instruction.NumSrcRegs = 2;
    inst.InstructionExtTexture.Texture = TGSI_TEXTURE_2D;
    inst.FullSrcRegisters[0].SrcRegister.File = TGSI_FILE_INPUT;
@@ -253,6 +287,13 @@ util_make_fragment_tex_shader(struct pipe_context *pipe)
    return pipe->create_fs_state(pipe, &shader);
 }
 
+void *
+util_make_fragment_tex_shader(struct pipe_context *pipe )
+{
+   return util_make_fragment_tex_shader_writemask( pipe,
+                                                   TGSI_WRITEMASK_XYZW );
+}
+
 
 
 

src/gallium/auxiliary/util/u_simple_shaders.h

@@ -49,6 +49,10 @@ util_make_vertex_passthrough_shader(struct pipe_context *pipe,
                                     const uint *semantic_indexes);
 
 
+extern void *
+util_make_fragment_tex_shader_writemask(struct pipe_context *pipe, 
+                                        unsigned writemask );
+
 extern void *
 util_make_fragment_tex_shader(struct pipe_context *pipe);
 

src/gallium/auxiliary/util/u_surface.h

@@ -37,6 +37,23 @@ struct pipe_texture;
 struct pipe_surface;
 
 
+/**
+ * Are s1 and s2 the same surface?
+ * Surfaces are basically views into textures so check if the two surfaces
+ * name the same part of the same texture.
+ */
+static INLINE boolean
+util_same_surface(const struct pipe_surface *s1, const struct pipe_surface *s2)
+{
+   return (s1->texture == s2->texture &&
+           s1->face == s2->face &&
+           s1->level == s2->level &&
+           s1->zslice == s2->zslice);
+}
+
+
+
+
 extern boolean
 util_create_rgba_surface(struct pipe_screen *screen,
                          uint width, uint height,

src/gallium/auxiliary/util/u_tile.c

@@ -153,7 +153,7 @@ a8r8g8b8_put_tile_rgba(unsigned *dst,
 }
 
 
-/*** PIPE_FORMAT_A8R8G8B8_UNORM ***/
+/*** PIPE_FORMAT_X8R8G8B8_UNORM ***/
 
 static void
 x8r8g8b8_get_tile_rgba(const unsigned *src,
@@ -596,6 +596,184 @@ r16g16b16a16_put_tile_rgba(short *dst,
 }
 
 
+/*** PIPE_FORMAT_R8G8B8A8_SRGB ***/
+
+/**
+ * Convert an 8-bit sRGB value from non-linear space to a
+ * linear RGB value in [0, 1].
+ * Implemented with a 256-entry lookup table.
+ */
+static INLINE float
+srgb_to_linear(ubyte cs8)
+{
+   static float table[256];
+   static boolean tableReady = FALSE;
+   if (!tableReady) {
+      /* compute lookup table now */
+      uint i;
+      for (i = 0; i < 256; i++) {
+         const float cs = ubyte_to_float(i);
+         if (cs <= 0.04045) {
+            table[i] = cs / 12.92f;
+         }
+         else {
+            table[i] = (float) powf((cs + 0.055) / 1.055, 2.4);
+         }
+      }
+      tableReady = TRUE;
+   }
+   return table[cs8];
+}
+
+
+/**
+ * Convert linear float in [0,1] to an srgb ubyte value in [0,255].
+ * XXX this hasn't been tested (render to srgb surface).
+ * XXX this needs optimization.
+ */
+static INLINE ubyte
+linear_to_srgb(float cl)
+{
+   if (cl >= 1.0F)
+      return 255;
+   else if (cl >= 0.0031308F)
+      return float_to_ubyte(1.055F * powf(cl, 0.41666F) - 0.055F);
+   else if (cl > 0.0F)
+      return float_to_ubyte(12.92F * cl);
+   else
+      return 0.0;
+}
+
+
+static void
+a8r8g8b8_srgb_get_tile_rgba(const unsigned *src,
+                            unsigned w, unsigned h,
+                            float *p,
+                            unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         const unsigned pixel = *src++;
+         pRow[0] = srgb_to_linear((pixel >> 16) & 0xff);
+         pRow[1] = srgb_to_linear((pixel >>  8) & 0xff);
+         pRow[2] = srgb_to_linear((pixel >>  0) & 0xff);
+         pRow[3] = ubyte_to_float((pixel >> 24) & 0xff);
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+a8r8g8b8_srgb_put_tile_rgba(unsigned *dst,
+                            unsigned w, unsigned h,
+                            const float *p,
+                            unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r, g, b, a;
+         r = linear_to_srgb(pRow[0]);
+         g = linear_to_srgb(pRow[1]);
+         b = linear_to_srgb(pRow[2]);
+         a = float_to_ubyte(pRow[3]);
+         *dst++ = (a << 24) | (r << 16) | (g << 8) | b;
+      }
+      p += src_stride;
+   }
+}
+
+
+/*** PIPE_FORMAT_A8L8_SRGB ***/
+
+static void
+a8l8_srgb_get_tile_rgba(const ushort *src,
+                        unsigned w, unsigned h,
+                        float *p,
+                        unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         ushort p = *src++;
+         pRow[0] =
+         pRow[1] =
+         pRow[2] = srgb_to_linear(p & 0xff);
+         pRow[3] = ubyte_to_float(p >> 8);
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+a8l8_srgb_put_tile_rgba(ushort *dst,
+                        unsigned w, unsigned h,
+                        const float *p,
+                        unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r, a;
+         r = linear_to_srgb(pRow[0]);
+         a = float_to_ubyte(pRow[3]);
+         *dst++ = (a << 8) | r;
+      }
+      p += src_stride;
+   }
+}
+
+
+/*** PIPE_FORMAT_L8_SRGB ***/
+
+static void
+l8_srgb_get_tile_rgba(const ubyte *src,
+                      unsigned w, unsigned h,
+                      float *p,
+                      unsigned dst_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      float *pRow = p;
+      for (j = 0; j < w; j++, src++, pRow += 4) {
+         pRow[0] =
+         pRow[1] =
+         pRow[2] = srgb_to_linear(*src);
+         pRow[3] = 1.0;
+      }
+      p += dst_stride;
+   }
+}
+
+static void
+l8_srgb_put_tile_rgba(ubyte *dst,
+                      unsigned w, unsigned h,
+                      const float *p,
+                      unsigned src_stride)
+{
+   unsigned i, j;
+
+   for (i = 0; i < h; i++) {
+      const float *pRow = p;
+      for (j = 0; j < w; j++, pRow += 4) {
+         unsigned r;
+         r = linear_to_srgb(pRow[0]);
+         *dst++ = (ubyte) r;
+      }
+      p += src_stride;
+   }
+}
+
 
 /*** PIPE_FORMAT_I8_UNORM ***/
 
@@ -946,6 +1124,15 @@ pipe_tile_raw_to_rgba(enum pipe_format format,
    case PIPE_FORMAT_R16G16B16A16_SNORM:
       r16g16b16a16_get_tile_rgba((short *) src, w, h, dst, dst_stride);
       break;
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      a8r8g8b8_srgb_get_tile_rgba((unsigned *) src, w, h, dst, dst_stride);
+      break;
+   case PIPE_FORMAT_A8L8_SRGB:
+      a8l8_srgb_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
+      break;
+   case PIPE_FORMAT_L8_SRGB:
+      l8_srgb_get_tile_rgba((ubyte *) src, w, h, dst, dst_stride);
+      break;
    case PIPE_FORMAT_Z16_UNORM:
       z16_get_tile_rgba((ushort *) src, w, h, dst, dst_stride);
       break;
@@ -1059,6 +1246,15 @@ pipe_put_tile_rgba(struct pipe_transfer *pt,
    case PIPE_FORMAT_R16G16B16A16_SNORM:
       r16g16b16a16_put_tile_rgba((short *) packed, w, h, p, src_stride);
       break;
+   case PIPE_FORMAT_A8R8G8B8_SRGB:
+      a8r8g8b8_srgb_put_tile_rgba((unsigned *) packed, w, h, p, src_stride);
+      break;
+   case PIPE_FORMAT_A8L8_SRGB:
+      a8l8_srgb_put_tile_rgba((ushort *) packed, w, h, p, src_stride);
+      break;
+   case PIPE_FORMAT_L8_SRGB:
+      l8_srgb_put_tile_rgba((ubyte *) packed, w, h, p, src_stride);
+      break;
    case PIPE_FORMAT_Z16_UNORM:
       /*z16_put_tile_rgba((ushort *) packed, w, h, p, src_stride);*/
       break;

src/gallium/auxiliary/util/u_upload_mgr.c

@@ -123,21 +123,25 @@ static enum pipe_error
 u_upload_alloc_buffer( struct u_upload_mgr *upload,
                        unsigned min_size )
 {
+   unsigned size;
+
    /* Release old buffer, if present:
     */
    u_upload_flush( upload );
 
    /* Allocate a new one: 
     */
-   upload->size = align(MAX2(upload->default_size, min_size), 4096);
+   size = align(MAX2(upload->default_size, min_size), 4096);
 
    upload->buffer = pipe_buffer_create( upload->screen,
                                         upload->alignment,
                                         upload->usage | PIPE_BUFFER_USAGE_CPU_WRITE,
-                                        upload->size );
+                                        size );
    if (upload->buffer == NULL) 
       goto fail;
    
+   upload->size = size;
+
    upload->offset = 0;
    return 0;
 

src/gallium/drivers/softpipe/sp_clear.c

@@ -85,5 +85,7 @@ softpipe_clear(struct pipe_context *pipe, unsigned buffers, const float *rgba,
       /* non-cached surface */
       pipe->surface_fill(pipe, ps, 0, 0, ps->width, ps->height, cv);
 #endif
-      }
+   }
+
+   softpipe->dirty_render_cache = TRUE;
 }

src/gallium/drivers/softpipe/sp_screen.c

@@ -84,7 +84,7 @@ softpipe_get_param(struct pipe_screen *screen, int param)
    case PIPE_CAP_MAX_TEXTURE_2D_LEVELS:
       return 12; /* max 2Kx2K */
    case PIPE_CAP_MAX_TEXTURE_3D_LEVELS:
-      return 8;  /* max 128x128x128 */
+      return 9;  /* max 256x256x256 */
    case PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS:
       return 12; /* max 2Kx2K */
    case PIPE_CAP_TGSI_CONT_SUPPORTED:

src/gallium/drivers/softpipe/sp_tex_sample.c

@@ -464,7 +464,7 @@ choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
    unsigned face;
    float sc, tc, ma;
 
-   if (arx > ary && arx > arz) {
+   if (arx >= ary && arx >= arz) {
       if (rx >= 0.0F) {
          face = PIPE_TEX_FACE_POS_X;
          sc = -rz;
@@ -478,7 +478,7 @@ choose_cube_face(float rx, float ry, float rz, float *newS, float *newT)
          ma = arx;
       }
    }
-   else if (ary > arx && ary > arz) {
+   else if (ary >= arx && ary >= arz) {
       if (ry >= 0.0F) {
          face = PIPE_TEX_FACE_POS_Y;
          sc = rx;

src/gallium/drivers/softpipe/sp_texture.c

@@ -42,6 +42,7 @@
 #include "sp_texture.h"
 #include "sp_tile_cache.h"
 #include "sp_screen.h"
+#include "sp_winsys.h"
 
 
 /* Simple, maximally packed layout.

src/gallium/drivers/trace/tr_dump.c

@@ -351,7 +351,7 @@ void trace_dump_call_begin_locked(const char *klass, const char *method)
    trace_dump_indent(1);
    trace_dump_writes("<call no=\'");
    trace_dump_writef("%lu", call_no);
-   trace_dump_writes("\' class =\'");
+   trace_dump_writes("\' class=\'");
    trace_dump_escape(klass);
    trace_dump_writes("\' method=\'");
    trace_dump_escape(method);

src/gallium/include/pipe/p_inlines.h

@@ -63,6 +63,13 @@ pipe_buffer_map(struct pipe_screen *screen,
    if(screen->buffer_map_range) {
       unsigned offset = 0;
       unsigned length = buf->size;
+
+      /* XXX: Actually we should be using/detecting DISCARD
+       * instead of assuming that WRITE implies discard */
+      if((usage & PIPE_BUFFER_USAGE_CPU_WRITE) &&
+         !(usage & PIPE_BUFFER_USAGE_DISCARD))
+         usage |= PIPE_BUFFER_USAGE_CPU_READ;
+
       return screen->buffer_map_range(screen, buf, offset, length, usage);
    }
    else

src/gallium/state_trackers/glx/xlib/xm_api.c

@@ -743,7 +743,7 @@ XMesaContext XMesaCreateContext( XMesaVisual v, XMesaContext share_list )
 {
    static GLboolean firstTime = GL_TRUE;
    struct pipe_screen *screen;
-   struct pipe_context *pipe;
+   struct pipe_context *pipe = NULL;
    XMesaContext c;
    GLcontext *mesaCtx;
    uint pf;

src/gallium/state_trackers/python/retrace/parse.py

@@ -371,7 +371,7 @@ class Main:
                     stream = GzipFile(arg, 'rt')
                 elif arg.endswith('.bz2'):
                     from bz2 import BZ2File
-                    stream = BZ2File(arg, 'rt')
+                    stream = BZ2File(arg, 'rU')
                 else:
                     stream = open(arg, 'rt')
                 self.process_arg(stream, options)

src/gallium/winsys/xlib/xlib_brw_screen.c

@@ -364,7 +364,7 @@ void xlib_brw_buffer_subdata_typed( struct pipe_winsys *pws,
                                     unsigned data_type )
 {
    unsigned aub_type = DW_GENERAL_STATE;
-   unsigned aub_sub_type;
+   unsigned aub_sub_type = 0;
 
    switch (data_type) {
    case BRW_CC_VP:

src/gallium/winsys/xlib/xlib_softpipe.c

@@ -303,6 +303,7 @@ xm_flush_frontbuffer(struct pipe_winsys *pws,
     */
    XMesaContext xmctx = (XMesaContext) context_private;
    xlib_softpipe_display_surface(xmctx->xm_buffer, surf);
+   xmesa_check_and_update_buffer_size(xmctx, xmctx->xm_buffer);
 }
 
 

src/glx/x11/dri2_glx.c

@@ -481,6 +481,8 @@ static __GLXDRIscreen *dri2CreateScreen(__GLXscreenConfigs *psc, int screen,
     psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
     psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
 
+    psc->driver_configs = driver_configs;
+
     psp->destroyScreen = dri2DestroyScreen;
     psp->createContext = dri2CreateContext;
     psp->createDrawable = dri2CreateDrawable;

src/glx/x11/dri_glx.c

@@ -418,6 +418,8 @@ CallCreateNewScreen(Display *dpy, int scrn, __GLXscreenConfigs *psc,
     psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
     psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
 
+    psc->driver_configs = driver_configs;
+
     /* Visuals with depth != screen depth are subject to automatic compositing
      * in the X server, so DRI1 can't render to them properly. Mark them as
      * non-conformant to prevent apps from picking them up accidentally.

src/glx/x11/drisw_glx.c

@@ -401,7 +401,7 @@ driCreateScreen(__GLXscreenConfigs * psc, int screen,
    psc->configs = driConvertConfigs(psc->core, psc->configs, driver_configs);
    psc->visuals = driConvertConfigs(psc->core, psc->visuals, driver_configs);
 
-   free(driver_configs);
+   psc->driver_configs = driver_configs;
 
    psp->destroyScreen = driDestroyScreen;
    psp->createContext = driCreateContext;

src/glx/x11/glx_pbuffer.c

@@ -39,6 +39,30 @@
 #include "glxextensions.h"
 #include "glcontextmodes.h"
 
+#define WARN_ONCE_GLX_1_3(a, b) {		\
+		static int warned=1;		\
+		if(warned) {			\
+			warn_GLX_1_3((a), b );	\
+			warned=0;		\
+		}				\
+	}
+
+/**
+ * Emit a warning when clients use GLX 1.3 functions on pre-1.3 systems.
+ */
+static void
+warn_GLX_1_3(Display *dpy, const char *function_name)
+{
+   __GLXdisplayPrivate *priv = __glXInitialize(dpy);
+
+   if (priv->minorVersion < 3) {
+      fprintf(stderr, 
+	      "WARNING: Application calling GLX 1.3 function \"%s\" "
+	      "when GLX 1.3 is not supported!  This is an application bug!\n",
+	      function_name);
+   }
+}
+
 
 /**
  * Change a drawable's attribute.
@@ -559,6 +583,8 @@ glXCreatePbuffer(Display * dpy, GLXFBConfig config, const int *attrib_list)
    width = 0;
    height = 0;
 
+   WARN_ONCE_GLX_1_3(dpy, __func__);
+
    for (i = 0; attrib_list[i * 2]; i++) {
       switch (attrib_list[i * 2]) {
       case GLX_PBUFFER_WIDTH:
@@ -592,6 +618,7 @@ PUBLIC void
 glXQueryDrawable(Display * dpy, GLXDrawable drawable,
                  int attribute, unsigned int *value)
 {
+   WARN_ONCE_GLX_1_3(dpy, __func__);
    GetDrawableAttribute(dpy, drawable, attribute, value);
 }
 
@@ -645,6 +672,8 @@ PUBLIC GLXPixmap
 glXCreatePixmap(Display * dpy, GLXFBConfig config, Pixmap pixmap,
                 const int *attrib_list)
 {
+   WARN_ONCE_GLX_1_3(dpy, __func__);
+
    return CreateDrawable(dpy, (__GLcontextModes *) config,
                          (Drawable) pixmap, attrib_list, X_GLXCreatePixmap);
 }
@@ -654,6 +683,8 @@ PUBLIC GLXWindow
 glXCreateWindow(Display * dpy, GLXFBConfig config, Window win,
                 const int *attrib_list)
 {
+   WARN_ONCE_GLX_1_3(dpy, __func__);
+
    return CreateDrawable(dpy, (__GLcontextModes *) config,
                          (Drawable) win, attrib_list, X_GLXCreateWindow);
 }
@@ -662,6 +693,8 @@ glXCreateWindow(Display * dpy, GLXFBConfig config, Window win,
 PUBLIC void
 glXDestroyPixmap(Display * dpy, GLXPixmap pixmap)
 {
+   WARN_ONCE_GLX_1_3(dpy, __func__);
+
    DestroyDrawable(dpy, (GLXDrawable) pixmap, X_GLXDestroyPixmap);
 }
 
@@ -669,6 +702,8 @@ glXDestroyPixmap(Display * dpy, GLXPixmap pixmap)
 PUBLIC void
 glXDestroyWindow(Display * dpy, GLXWindow win)
 {
+   WARN_ONCE_GLX_1_3(dpy, __func__);
+
    DestroyDrawable(dpy, (GLXDrawable) win, X_GLXDestroyWindow);
 }
 
@@ -689,3 +724,4 @@ GLX_ALIAS_VOID(glXGetSelectedEventSGIX,
                (Display * dpy, GLXDrawable drawable,
                 unsigned long *mask), (dpy, drawable, mask),
                glXGetSelectedEvent)
+

src/glx/x11/glxclient.h

@@ -501,6 +501,8 @@ struct __GLXscreenConfigsRec {
 
     __GLXDRIscreen *driScreen;
 
+    const __DRIconfig** driver_configs;
+
 #ifdef __DRI_COPY_SUB_BUFFER
     const __DRIcopySubBufferExtension *driCopySubBuffer;
 #endif

src/glx/x11/glxcmds.c

@@ -845,6 +845,34 @@ PUBLIC GLXPixmap glXCreateGLXPixmap(Display *dpy, XVisualInfo *vis,
     req->glxpixmap = xid = XAllocID(dpy);
     UnlockDisplay(dpy);
     SyncHandle();
+
+#ifdef GLX_DIRECT_RENDERING
+    do {
+	/* FIXME: Maybe delay __DRIdrawable creation until the drawable
+	 * is actually bound to a context... */
+
+	__GLXdisplayPrivate *const priv = __glXInitialize(dpy);
+	__GLXDRIdrawable *pdraw;
+	__GLXscreenConfigs *psc;
+	__GLcontextModes *modes;
+
+	psc = &priv->screenConfigs[vis->screen];
+	if (psc->driScreen == NULL)
+	    break;
+	modes = _gl_context_modes_find_visual(psc->visuals, vis->visualid);
+	pdraw = psc->driScreen->createDrawable(psc, pixmap, req->glxpixmap, modes);
+	if (pdraw == NULL) {
+	    fprintf(stderr, "failed to create pixmap\n");
+	    break;
+	}
+
+	if (__glxHashInsert(psc->drawHash, req->glxpixmap, pdraw)) {
+	    (*pdraw->destroyDrawable) (pdraw);
+	    return None;           /* FIXME: Check what we're supposed to do here... */
+	}
+    } while (0);
+#endif
+
     return xid;
 }
 

src/glx/x11/glxext.c

@@ -149,6 +149,12 @@ FreeScreenConfigs(__GLXdisplayPrivate * priv)
       Xfree((char *) psc->serverGLXexts);
 
 #ifdef GLX_DIRECT_RENDERING
+      if (psc->driver_configs) {
+         for(unsigned int i = 0; psc->driver_configs[i]; i++)
+            free((__DRIconfig*)psc->driver_configs[i]);
+         free(psc->driver_configs);
+         psc->driver_configs = NULL;
+      }
       if (psc->driScreen) {
          psc->driScreen->destroyScreen(psc);
          __glxHashDestroy(psc->drawHash);

src/glx/x11/glxhash.c

@@ -77,6 +77,7 @@
 
 #include <stdio.h>
 #include <stdlib.h>
+#include <string.h>
 
 #define HASH_MAGIC 0xdeadbeef
 #define HASH_DEBUG 0
@@ -87,10 +88,21 @@
 
 #define HASH_ALLOC malloc
 #define HASH_FREE  free
-#define HASH_RANDOM_DECL
-#define HASH_RANDOM_INIT(seed)  srandom(seed)
-#define HASH_RANDOM             random()
+#ifndef __GLIBC__
+#define HASH_RANDOM_DECL	char *ps, rs[256]
+#define HASH_RANDOM_INIT(seed)	ps = initstate(seed, rs, sizeof(rs))
+#define HASH_RANDOM		random()
+#define HASH_RANDOM_DESTROY	setstate(ps)
+#else
+#define HASH_RANDOM_DECL	struct random_data rd; int32_t rv; char rs[256]
+#define HASH_RANDOM_INIT(seed)					\
+   do {								\
+      (void) memset(&rd, 0, sizeof(rd));			\
+      (void) initstate_r(seed, rs, sizeof(rs), &rd);		\
+   } while(0)
+#define HASH_RANDOM             ((void) random_r(&rd, &rv), rv)
 #define HASH_RANDOM_DESTROY
+#endif
 
 typedef struct __glxHashBucket
 {

src/mesa/drivers/dri/common/dri_util.c

@@ -763,7 +763,7 @@ dri2CreateNewScreen(int scrn, int fd,
     if (driDriverAPI.InitScreen2 == NULL)
         return NULL;
 
-    psp = _mesa_malloc(sizeof(*psp));
+    psp = _mesa_calloc(sizeof(*psp));
     if (!psp)
 	return NULL;
 

src/mesa/drivers/dri/i915/intel_pixel_read.c

@@ -1,306 +0,0 @@
-/**************************************************************************
- * 
- * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
- * All Rights Reserved.
- * 
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the
- * "Software"), to deal in the Software without restriction, including
- * without limitation the rights to use, copy, modify, merge, publish,
- * distribute, sub license, and/or sell copies of the Software, and to
- * permit persons to whom the Software is furnished to do so, subject to
- * the following conditions:
- * 
- * The above copyright notice and this permission notice (including the
- * next paragraph) shall be included in all copies or substantial portions
- * of the Software.
- * 
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
- * 
- **************************************************************************/
-
-#include "main/glheader.h"
-#include "main/enums.h"
-#include "main/mtypes.h"
-#include "main/macros.h"
-#include "main/image.h"
-#include "main/bufferobj.h"
-#include "swrast/swrast.h"
-
-#include "intel_screen.h"
-#include "intel_context.h"
-#include "intel_batchbuffer.h"
-#include "intel_blit.h"
-#include "intel_buffers.h"
-#include "intel_regions.h"
-#include "intel_pixel.h"
-#include "intel_buffer_objects.h"
-
-/* For many applications, the new ability to pull the source buffers
- * back out of the GTT and then do the packing/conversion operations
- * in software will be as much of an improvement as trying to get the
- * blitter and/or texture engine to do the work. 
- *
- * This step is gated on private backbuffers.
- * 
- * Obviously the frontbuffer can't be pulled back, so that is either
- * an argument for blit/texture readpixels, or for blitting to a
- * temporary and then pulling that back.
- *
- * When the destination is a pbo, however, it's not clear if it is
- * ever going to be pulled to main memory (though the access param
- * will be a good hint).  So it sounds like we do want to be able to
- * choose between blit/texture implementation on the gpu and pullback
- * and cpu-based copying.
- *
- * Unless you can magically turn client memory into a PBO for the
- * duration of this call, there will be a cpu-based copying step in
- * any case.
- */
-
-
-static GLboolean
-do_texture_readpixels(GLcontext * ctx,
-                      GLint x, GLint y, GLsizei width, GLsizei height,
-                      GLenum format, GLenum type,
-                      const struct gl_pixelstore_attrib *pack,
-                      struct intel_region *dest_region)
-{
-#if 0
-   struct intel_context *intel = intel_context(ctx);
-   intelScreenPrivate *screen = intel->intelScreen;
-   GLint pitch = pack->RowLength ? pack->RowLength : width;
-   __DRIdrawablePrivate *dPriv = intel->driDrawable;
-   int textureFormat;
-   GLenum glTextureFormat;
-   int destFormat, depthFormat, destPitch;
-   drm_clip_rect_t tmp;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-
-   if (ctx->_ImageTransferState ||
-       pack->SwapBytes || pack->LsbFirst || !pack->Invert) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   intel->vtbl.meta_texrect_source(intel, intel_readbuf_region(intel));
-
-   if (!intel->vtbl.meta_render_dest(intel, dest_region, type, format)) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         fprintf(stderr, "%s: couldn't set dest %s/%s\n",
-                 __FUNCTION__,
-                 _mesa_lookup_enum_by_nr(type),
-                 _mesa_lookup_enum_by_nr(format));
-      return GL_FALSE;
-   }
-
-   LOCK_HARDWARE(intel);
-
-   if (intel->driDrawable->numClipRects) {
-      intel->vtbl.install_meta_state(intel);
-      intel->vtbl.meta_no_depth_write(intel);
-      intel->vtbl.meta_no_stencil_write(intel);
-
-      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
-         UNLOCK_HARDWARE(intel);
-         SET_STATE(i830, state);
-         if (INTEL_DEBUG & DEBUG_PIXEL)
-            fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
-         return GL_TRUE;
-      }
-
-      y = dPriv->h - y - height;
-      x += dPriv->x;
-      y += dPriv->y;
-
-
-      /* Set the frontbuffer up as a large rectangular texture.
-       */
-      intel->vtbl.meta_tex_rect_source(intel, src_region, textureFormat);
-
-
-      intel->vtbl.meta_texture_blend_replace(i830, glTextureFormat);
-
-
-      /* Set the 3d engine to draw into the destination region:
-       */
-
-      intel->vtbl.meta_draw_region(intel, dest_region);
-      intel->vtbl.meta_draw_format(intel, destFormat, depthFormat);     /* ?? */
-
-
-      /* Draw a single quad, no cliprects:
-       */
-      intel->vtbl.meta_disable_cliprects(intel);
-
-      intel->vtbl.draw_quad(intel,
-                            0, width, 0, height,
-                            0x00ff00ff, x, x + width, y, y + height);
-
-      intel->vtbl.leave_meta_state(intel);
-   }
-   UNLOCK_HARDWARE(intel);
-
-   intel_region_wait_fence(ctx, dest_region);   /* required by GL */
-   return GL_TRUE;
-#endif
-
-   return GL_FALSE;
-}
-
-
-
-
-static GLboolean
-do_blit_readpixels(GLcontext * ctx,
-                   GLint x, GLint y, GLsizei width, GLsizei height,
-                   GLenum format, GLenum type,
-                   const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
-{
-   struct intel_context *intel = intel_context(ctx);
-   struct intel_region *src = intel_readbuf_region(intel);
-   struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj);
-   GLuint dst_offset;
-   GLuint rowLength;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("%s\n", __FUNCTION__);
-
-   if (!src)
-      return GL_FALSE;
-
-   if (dst) {
-      /* XXX This validation should be done by core mesa:
-       */
-      if (!_mesa_validate_pbo_access(2, pack, width, height, 1,
-                                     format, type, pixels)) {
-         _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels");
-         return GL_TRUE;
-      }
-   }
-   else {
-      /* PBO only for now:
-       */
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         _mesa_printf("%s - not PBO\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-
-   if (ctx->_ImageTransferState ||
-       !intel_check_blit_format(src, format, type)) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         _mesa_printf("%s - bad format for blit\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         _mesa_printf("%s: bad packing params\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-
-   if (pack->RowLength > 0)
-      rowLength = pack->RowLength;
-   else
-      rowLength = width;
-
-   if (pack->Invert) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         _mesa_printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
-      return GL_FALSE;
-   }
-   else {
-      rowLength = -rowLength;
-   }
-
-   /* XXX 64-bit cast? */
-   dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height,
-                                             format, type, 0, 0, 0);
-
-
-   /* Although the blits go on the command buffer, need to do this and
-    * fire with lock held to guarentee cliprects are correct.
-    */
-   intelFlush(&intel->ctx);
-   LOCK_HARDWARE(intel);
-
-   if (intel->driDrawable->numClipRects) {
-      GLboolean all = (width * height * src->cpp == dst->Base.Size &&
-                       x == 0 && dst_offset == 0);
-
-      dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
-						  all ? INTEL_WRITE_FULL :
-						  INTEL_WRITE_PART);
-      __DRIdrawablePrivate *dPriv = intel->driDrawable;
-      int nbox = dPriv->numClipRects;
-      drm_clip_rect_t *box = dPriv->pClipRects;
-      drm_clip_rect_t rect;
-      drm_clip_rect_t src_rect;
-      int i;
-
-      src_rect.x1 = dPriv->x + x;
-      src_rect.y1 = dPriv->y + dPriv->h - (y + height);
-      src_rect.x2 = src_rect.x1 + width;
-      src_rect.y2 = src_rect.y1 + height;
-
-
-
-      for (i = 0; i < nbox; i++) {
-         if (!intel_intersect_cliprects(&rect, &src_rect, &box[i]))
-            continue;
-
-         intelEmitCopyBlit(intel,
-                           src->cpp,
-                           src->pitch, src->buffer, 0, src->tiling,
-                           rowLength, dst_buffer, dst_offset, GL_FALSE,
-                           rect.x1,
-                           rect.y1,
-                           rect.x1 - src_rect.x1,
-                           rect.y2 - src_rect.y2,
-                           rect.x2 - rect.x1, rect.y2 - rect.y1,
-			   GL_COPY);
-      }
-   }
-   UNLOCK_HARDWARE(intel);
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("%s - DONE\n", __FUNCTION__);
-
-   return GL_TRUE;
-}
-
-void
-intelReadPixels(GLcontext * ctx,
-                GLint x, GLint y, GLsizei width, GLsizei height,
-                GLenum format, GLenum type,
-                const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
-{
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
-
-   intelFlush(ctx);
-
-   if (do_blit_readpixels
-       (ctx, x, y, width, height, format, type, pack, pixels))
-      return;
-
-   if (do_texture_readpixels
-       (ctx, x, y, width, height, format, type, pack, pixels))
-      return;
-
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
-
-   _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels);
-}

src/mesa/drivers/dri/i915/intel_pixel_read.c

@@ -0,0 +1 @@
+../intel/intel_pixel_read.c

src/mesa/drivers/dri/i965/Makefile

@@ -22,6 +22,7 @@ DRIVER_SOURCES = \
 	intel_pixel_bitmap.c \
 	intel_pixel_copy.c \
 	intel_pixel_draw.c \
+	intel_pixel_read.c \
 	intel_state.c \
 	intel_swapbuffers.c \
 	intel_tex.c \

src/mesa/drivers/dri/i965/brw_clip.c

@@ -152,7 +152,8 @@ static void upload_clip_prog(struct brw_context *brw)
 
    /* _NEW_POLYGON */
    if (key.primitive == GL_TRIANGLES) {
-      if (ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK) 
+      if (ctx->Polygon.CullFlag &&
+	  ctx->Polygon.CullFaceMode == GL_FRONT_AND_BACK)
 	 key.clip_mode = BRW_CLIPMODE_REJECT_ALL;
       else {
 	 GLuint fill_front = CLIP_CULL;

src/mesa/drivers/dri/i965/brw_context.h

@@ -386,6 +386,8 @@ struct brw_cached_batch_item {
 struct brw_vertex_element {
    const struct gl_client_array *glarray;
 
+   /** The corresponding Mesa vertex attribute */
+   gl_vert_attrib attrib;
    /** Size of a complete element */
    GLuint element_size;
    /** Number of uploaded elements for this input. */
@@ -477,6 +479,9 @@ struct brw_context
    struct {
       struct brw_vertex_element inputs[VERT_ATTRIB_MAX];
 
+      struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
+      GLuint nr_enabled;
+
 #define BRW_NR_UPLOAD_BUFS 17
 #define BRW_UPLOAD_INIT_SIZE (128*1024)
 

src/mesa/drivers/dri/i965/brw_draw.c

@@ -185,6 +185,7 @@ static void brw_merge_inputs( struct brw_context *brw,
 
    for (i = 0; i < VERT_ATTRIB_MAX; i++) {
       brw->vb.inputs[i].glarray = arrays[i];
+      brw->vb.inputs[i].attrib = (gl_vert_attrib) i;
 
       if (arrays[i]->StrideB != 0)
 	 brw->vb.info.varying |= 1 << i;

src/mesa/drivers/dri/i965/brw_draw_upload.c

@@ -343,16 +343,13 @@ static void brw_prepare_vertices(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = intel_context(ctx);
-   GLuint tmp = brw->vs.prog_data->inputs_read; 
+   GLbitfield vs_inputs = brw->vs.prog_data->inputs_read; 
    GLuint i;
    const unsigned char *ptr = NULL;
    GLuint interleave = 0;
    unsigned int min_index = brw->vb.min_index;
    unsigned int max_index = brw->vb.max_index;
 
-   struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
-   GLuint nr_enabled = 0;
-
    struct brw_vertex_element *upload[VERT_ATTRIB_MAX];
    GLuint nr_uploads = 0;
 
@@ -362,12 +359,13 @@ static void brw_prepare_vertices(struct brw_context *brw)
       _mesa_printf("%s %d..%d\n", __FUNCTION__, min_index, max_index);
 
    /* Accumulate the list of enabled arrays. */
-   while (tmp) {
-      GLuint i = _mesa_ffsll(tmp)-1;
+   brw->vb.nr_enabled = 0;
+   while (vs_inputs) {
+      GLuint i = _mesa_ffsll(vs_inputs) - 1;
       struct brw_vertex_element *input = &brw->vb.inputs[i];
 
-      tmp &= ~(1<<i);
-      enabled[nr_enabled++] = input;
+      vs_inputs &= ~(1 << i);
+      brw->vb.enabled[brw->vb.nr_enabled++] = input;
    }
 
    /* XXX: In the rare cases where this happens we fallback all
@@ -376,13 +374,13 @@ static void brw_prepare_vertices(struct brw_context *brw)
     * cases with > 17 vertex attributes enabled, so it probably
     * isn't an issue at this point.
     */
-   if (nr_enabled >= BRW_VEP_MAX) {
+   if (brw->vb.nr_enabled >= BRW_VEP_MAX) {
       intel->Fallback = 1;
       return;
    }
 
-   for (i = 0; i < nr_enabled; i++) {
-      struct brw_vertex_element *input = enabled[i];
+   for (i = 0; i < brw->vb.nr_enabled; i++) {
+      struct brw_vertex_element *input = brw->vb.enabled[i];
 
       input->element_size = get_size(input->glarray->Type) * input->glarray->Size;
       input->count = input->glarray->StrideB ? max_index + 1 - min_index : 1;
@@ -398,6 +396,20 @@ static void brw_prepare_vertices(struct brw_context *brw)
 	 dri_bo_reference(input->bo);
 	 input->offset = (unsigned long)input->glarray->Ptr;
 	 input->stride = input->glarray->StrideB;
+
+	 /* This is a common place to reach if the user mistakenly supplies
+	  * a pointer in place of a VBO offset.  If we just let it go through,
+	  * we may end up dereferencing a pointer beyond the bounds of the
+	  * GTT.  We would hope that the VBO's max_index would save us, but
+	  * Mesa appears to hand us min/max values not clipped to the
+	  * array object's _MaxElement, and _MaxElement frequently appears
+	  * to be wrong anyway.
+	  *
+	  * The VBO spec allows application termination in this case, and it's
+	  * probably a service to the poor programmer to do so rather than
+	  * trying to just not render.
+	  */
+	 assert(input->offset < input->bo->size);
       } else {
 	 if (input->bo != NULL) {
 	    /* Already-uploaded vertex data is present from a previous
@@ -410,7 +422,7 @@ static void brw_prepare_vertices(struct brw_context *brw)
 	 /* Queue the buffer object up to be uploaded in the next pass,
 	  * when we've decided if we're doing interleaved or not.
 	  */
-	 if (i == 0) {
+	 if (input->attrib == VERT_ATTRIB_POS) {
 	    /* Position array not properly enabled:
 	     */
             if (input->glarray->StrideB == 0) {
@@ -466,8 +478,8 @@ static void brw_prepare_vertices(struct brw_context *brw)
 
    brw_prepare_query_begin(brw);
 
-   for (i = 0; i < nr_enabled; i++) {
-      struct brw_vertex_element *input = enabled[i];
+   for (i = 0; i < brw->vb.nr_enabled; i++) {
+      struct brw_vertex_element *input = brw->vb.enabled[i];
 
       brw_add_validated_bo(brw, input->bo);
    }
@@ -477,34 +489,44 @@ static void brw_emit_vertices(struct brw_context *brw)
 {
    GLcontext *ctx = &brw->intel.ctx;
    struct intel_context *intel = intel_context(ctx);
-   GLuint tmp = brw->vs.prog_data->inputs_read;
-   struct brw_vertex_element *enabled[VERT_ATTRIB_MAX];
    GLuint i;
-   GLuint nr_enabled = 0;
-
-  /* Accumulate the list of enabled arrays. */
-   while (tmp) {
-      i = _mesa_ffsll(tmp)-1;
-      struct brw_vertex_element *input = &brw->vb.inputs[i];
-
-      tmp &= ~(1<<i);
-      enabled[nr_enabled++] = input;
-   }
 
    brw_emit_query_begin(brw);
 
+   /* If the VS doesn't read any inputs (calculating vertex position from
+    * a state variable for some reason, for example), emit a single pad
+    * VERTEX_ELEMENT struct and bail.
+    *
+    * The stale VB state stays in place, but they don't do anything unless
+    * a VE loads from them.
+    */
+   if (brw->vb.nr_enabled == 0) {
+      BEGIN_BATCH(3, IGNORE_CLIPRECTS);
+      OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | 1);
+      OUT_BATCH((0 << BRW_VE0_INDEX_SHIFT) |
+		BRW_VE0_VALID |
+		(BRW_SURFACEFORMAT_R32G32B32A32_FLOAT << BRW_VE0_FORMAT_SHIFT) |
+		(0 << BRW_VE0_SRC_OFFSET_SHIFT));
+      OUT_BATCH((BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_0_SHIFT) |
+		(BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_1_SHIFT) |
+		(BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_2_SHIFT) |
+		(BRW_VE1_COMPONENT_STORE_1_FLT << BRW_VE1_COMPONENT_3_SHIFT));
+      ADVANCE_BATCH();
+      return;
+   }
+
    /* Now emit VB and VEP state packets.
     *
     * This still defines a hardware VB for each input, even if they
     * are interleaved or from the same VBO.  TBD if this makes a
     * performance difference.
     */
-   BEGIN_BATCH(1 + nr_enabled * 4, IGNORE_CLIPRECTS);
+   BEGIN_BATCH(1 + brw->vb.nr_enabled * 4, IGNORE_CLIPRECTS);
    OUT_BATCH((CMD_VERTEX_BUFFER << 16) |
-	     ((1 + nr_enabled * 4) - 2));
+	     ((1 + brw->vb.nr_enabled * 4) - 2));
 
-   for (i = 0; i < nr_enabled; i++) {
-      struct brw_vertex_element *input = enabled[i];
+   for (i = 0; i < brw->vb.nr_enabled; i++) {
+      struct brw_vertex_element *input = brw->vb.enabled[i];
 
       OUT_BATCH((i << BRW_VB0_INDEX_SHIFT) |
 		BRW_VB0_ACCESS_VERTEXDATA |
@@ -517,10 +539,10 @@ static void brw_emit_vertices(struct brw_context *brw)
    }
    ADVANCE_BATCH();
 
-   BEGIN_BATCH(1 + nr_enabled * 2, IGNORE_CLIPRECTS);
-   OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + nr_enabled * 2) - 2));
-   for (i = 0; i < nr_enabled; i++) {
-      struct brw_vertex_element *input = enabled[i];
+   BEGIN_BATCH(1 + brw->vb.nr_enabled * 2, IGNORE_CLIPRECTS);
+   OUT_BATCH((CMD_VERTEX_ELEMENT << 16) | ((1 + brw->vb.nr_enabled * 2) - 2));
+   for (i = 0; i < brw->vb.nr_enabled; i++) {
+      struct brw_vertex_element *input = brw->vb.enabled[i];
       uint32_t format = get_surface_type(input->glarray->Type,
 					 input->glarray->Size,
 					 input->glarray->Format,
@@ -635,7 +657,7 @@ static void brw_emit_indices(struct brw_context *brw)
    if (index_buffer == NULL)
       return;
 
-   ib_size = get_size(index_buffer->type) * index_buffer->count;
+   ib_size = get_size(index_buffer->type) * index_buffer->count - 1;
 
    /* Emit the indexbuffer packet:
     */

src/mesa/drivers/dri/i965/brw_eu.c

@@ -62,7 +62,7 @@ void brw_set_predicate_control( struct brw_compile *p, GLuint pc )
 
 void brw_set_conditionalmod( struct brw_compile *p, GLuint conditional )
 {
-   p->current->header.destreg__conditonalmod = conditional;
+   p->current->header.destreg__conditionalmod = conditional;
 }
 
 void brw_set_access_mode( struct brw_compile *p, GLuint access_mode )

src/mesa/drivers/dri/i965/brw_eu_emit.c

@@ -377,8 +377,8 @@ static struct brw_instruction *next_insn( struct brw_compile *p,
    /* Reset this one-shot flag: 
     */
 
-   if (p->current->header.destreg__conditonalmod) {
-      p->current->header.destreg__conditonalmod = 0;   
+   if (p->current->header.destreg__conditionalmod) {
+      p->current->header.destreg__conditionalmod = 0;
       p->current->header.predicate_control = BRW_PREDICATE_NORMAL;
    }
 
@@ -746,7 +746,7 @@ void brw_CMP(struct brw_compile *p,
 {
    struct brw_instruction *insn = next_insn(p, BRW_OPCODE_CMP);
 
-   insn->header.destreg__conditonalmod = conditional;
+   insn->header.destreg__conditionalmod = conditional;
    brw_set_dest(insn, dest);
    brw_set_src0(insn, src0);
    brw_set_src1(insn, src1);
@@ -790,7 +790,7 @@ void brw_math( struct brw_compile *p,
     * instructions.
     */
    insn->header.predicate_control = 0; 
-   insn->header.destreg__conditonalmod = msg_reg_nr;
+   insn->header.destreg__conditionalmod = msg_reg_nr;
 
    brw_set_dest(insn, dest);
    brw_set_src0(insn, src);
@@ -826,7 +826,7 @@ void brw_math_16( struct brw_compile *p,
    brw_set_compression_control(p, BRW_COMPRESSION_NONE);
 
    insn = next_insn(p, BRW_OPCODE_SEND);
-   insn->header.destreg__conditonalmod = msg_reg_nr;
+   insn->header.destreg__conditionalmod = msg_reg_nr;
 
    brw_set_dest(insn, dest);
    brw_set_src0(insn, src);
@@ -842,7 +842,7 @@ void brw_math_16( struct brw_compile *p,
     */
    insn = next_insn(p, BRW_OPCODE_SEND);
    insn->header.compression_control = BRW_COMPRESSION_2NDHALF;
-   insn->header.destreg__conditonalmod = msg_reg_nr+1;
+   insn->header.destreg__conditionalmod = msg_reg_nr+1;
 
    brw_set_dest(insn, offset(dest,1));
    brw_set_src0(insn, src);
@@ -888,7 +888,7 @@ void brw_dp_WRITE_16( struct brw_compile *p,
    
       insn->header.predicate_control = 0; /* XXX */
       insn->header.compression_control = BRW_COMPRESSION_NONE; 
-      insn->header.destreg__conditonalmod = msg_reg_nr;
+      insn->header.destreg__conditionalmod = msg_reg_nr;
   
       brw_set_dest(insn, dest);
       brw_set_src0(insn, src);
@@ -933,7 +933,7 @@ void brw_dp_READ_16( struct brw_compile *p,
    
       insn->header.predicate_control = 0; /* XXX */
       insn->header.compression_control = BRW_COMPRESSION_NONE; 
-      insn->header.destreg__conditonalmod = msg_reg_nr;
+      insn->header.destreg__conditionalmod = msg_reg_nr;
   
       brw_set_dest(insn, dest);	/* UW? */
       brw_set_src0(insn, retype(brw_vec8_grf(0, 0), BRW_REGISTER_TYPE_UW));
@@ -986,7 +986,7 @@ void brw_dp_READ_4( struct brw_compile *p,
    
       insn->header.predicate_control = BRW_PREDICATE_NONE;
       insn->header.compression_control = BRW_COMPRESSION_NONE; 
-      insn->header.destreg__conditonalmod = msg_reg_nr;
+      insn->header.destreg__conditionalmod = msg_reg_nr;
       insn->header.mask_control = BRW_MASK_DISABLE;
   
       /* cast dest to a uword[8] vector */
@@ -1059,7 +1059,7 @@ void brw_dp_READ_4_vs(struct brw_compile *p,
    
       insn->header.predicate_control = BRW_PREDICATE_NONE;
       insn->header.compression_control = BRW_COMPRESSION_NONE; 
-      insn->header.destreg__conditonalmod = msg_reg_nr;
+      insn->header.destreg__conditionalmod = msg_reg_nr;
       insn->header.mask_control = BRW_MASK_DISABLE;
       /*insn->header.access_mode = BRW_ALIGN_16;*/
   
@@ -1092,7 +1092,7 @@ void brw_fb_WRITE(struct brw_compile *p,
    
    insn->header.predicate_control = 0; /* XXX */
    insn->header.compression_control = BRW_COMPRESSION_NONE; 
-   insn->header.destreg__conditonalmod = msg_reg_nr;
+   insn->header.destreg__conditionalmod = msg_reg_nr;
   
    brw_set_dest(insn, dest);
    brw_set_src0(insn, src0);
@@ -1187,7 +1187,7 @@ void brw_SAMPLE(struct brw_compile *p,
    
       insn->header.predicate_control = 0; /* XXX */
       insn->header.compression_control = BRW_COMPRESSION_NONE;
-      insn->header.destreg__conditonalmod = msg_reg_nr;
+      insn->header.destreg__conditionalmod = msg_reg_nr;
 
       brw_set_dest(insn, dest);
       brw_set_src0(insn, src0);
@@ -1238,7 +1238,7 @@ void brw_urb_WRITE(struct brw_compile *p,
    brw_set_src0(insn, src0);
    brw_set_src1(insn, brw_imm_d(0));
 
-   insn->header.destreg__conditonalmod = msg_reg_nr;
+   insn->header.destreg__conditionalmod = msg_reg_nr;
 
    brw_set_urb_message(insn,
 		       allocate,

src/mesa/drivers/dri/i965/brw_sf_state.c

@@ -233,7 +233,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
    else if (sf.sf6.line_width <= 0x2)
        sf.sf6.line_width = 0;
 
-   /* _NEW_POINT */
+   /* _NEW_BUFFERS */
    key->render_to_fbo = brw->intel.ctx.DrawBuffer->Name != 0;
    if (!key->render_to_fbo) {
       /* Rendering to an OpenGL window */
@@ -263,6 +263,7 @@ sf_unit_create_from_key(struct brw_context *brw, struct brw_sf_unit_key *key,
    }
    /* XXX clamp max depends on AA vs. non-AA */
 
+   /* _NEW_POINT */
    sf.sf7.sprite_point = key->point_sprite;
    sf.sf7.point_size = CLAMP(rint(key->point_size), 1, 255) * (1<<3);
    sf.sf7.use_point_size_state = !key->point_attenuated;
@@ -328,7 +329,8 @@ const struct brw_tracked_state brw_sf_unit = {
       .mesa  = (_NEW_POLYGON | 
 		_NEW_LINE | 
 		_NEW_POINT | 
-		_NEW_SCISSOR),
+		_NEW_SCISSOR |
+		_NEW_BUFFERS),
       .brw   = BRW_NEW_URB_FENCE,
       .cache = (CACHE_NEW_SF_VP |
 		CACHE_NEW_SF_PROG)

src/mesa/drivers/dri/i965/brw_structs.h

@@ -1168,7 +1168,7 @@ struct brw_instruction
       GLuint predicate_control:4;
       GLuint predicate_inverse:1;
       GLuint execution_size:3;
-      GLuint destreg__conditonalmod:4; /* destreg - send, conditionalmod - others */
+      GLuint destreg__conditionalmod:4; /* destreg - send, conditionalmod - others */
       GLuint pad0:2;
       GLuint debug_control:1;
       GLuint saturate:1;

src/mesa/drivers/dri/i965/brw_tex_layout.c

@@ -119,6 +119,16 @@ GLboolean brw_miptree_layout( struct intel_context *intel, struct intel_mipmap_t
     }
 
       }
+      /* The 965's sampler lays cachelines out according to how accesses
+       * in the texture surfaces run, so they may be "vertical" through
+       * memory.  As a result, the docs say in Surface Padding Requirements:
+       * Sampling Engine Surfaces that two extra rows of padding are required.
+       * We don't know of similar requirements for pre-965, but given that
+       * those docs are silent on padding requirements in general, let's play
+       * it safe.
+       */
+      if (mt->target == GL_TEXTURE_CUBE_MAP)
+	 mt->total_height += 2;
       break;
    }
 

src/mesa/drivers/dri/i965/brw_vs_emit.c

@@ -68,6 +68,7 @@ static void release_tmps( struct brw_vs_compile *c )
 static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 {
    GLuint i, reg = 0, mrf;
+   int attributes_in_vue;
 
 #if 0
    if (c->vp->program.Base.Parameters->NumParameters >= 6)
@@ -123,6 +124,11 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
 	 reg++;
       }
    }
+   /* If there are no inputs, we'll still be reading one attribute's worth
+    * because it's required -- see urb_read_length setting.
+    */
+   if (c->nr_inputs == 0)
+      reg++;
 
    /* Allocate outputs: TODO: could organize the non-position outputs
     * to go straight into message regs.
@@ -200,8 +206,20 @@ static void brw_vs_alloc_regs( struct brw_vs_compile *c )
     * vertex urb, so is half the amount:
     */
    c->prog_data.urb_read_length = (c->nr_inputs + 1) / 2;
+   /* Setting this field to 0 leads to undefined behavior according to the
+    * the VS_STATE docs.  Our VUEs will always have at least one attribute
+    * sitting in them, even if it's padding.
+    */
+   if (c->prog_data.urb_read_length == 0)
+      c->prog_data.urb_read_length = 1;
+
+   /* The VS VUEs are shared by VF (outputting our inputs) and VS, so size
+    * them to fit the biggest thing they need to.
+    */
+   attributes_in_vue = MAX2(c->nr_outputs, c->nr_inputs);
+
+   c->prog_data.urb_entry_size = (attributes_in_vue + 2 + 3) / 4;
 
-   c->prog_data.urb_entry_size = (c->nr_outputs + 2 + 3) / 4;
    c->prog_data.total_grf = reg;
 
    if (INTEL_DEBUG & DEBUG_VS) {
@@ -1172,20 +1190,36 @@ post_vs_emit( struct brw_vs_compile *c,
    brw_set_src1(end_inst, brw_imm_d(offset * 16));
 }
 
+static uint32_t
+get_predicate(uint32_t swizzle)
+{
+   switch (swizzle) {
+   case SWIZZLE_XXXX:
+      return BRW_PREDICATE_ALIGN16_REPLICATE_X;
+   case SWIZZLE_YYYY:
+      return BRW_PREDICATE_ALIGN16_REPLICATE_Y;
+   case SWIZZLE_ZZZZ:
+      return BRW_PREDICATE_ALIGN16_REPLICATE_Z;
+   case SWIZZLE_WWWW:
+      return BRW_PREDICATE_ALIGN16_REPLICATE_W;
+   default:
+      _mesa_problem(NULL, "Unexpected predicate: 0x%08x\n", swizzle);
+      return BRW_PREDICATE_NORMAL;
+   }
+}
 
 /* Emit the vertex program instructions here.
  */
 void brw_vs_emit(struct brw_vs_compile *c )
 {
-#define MAX_IFSN 32
+#define MAX_IF_DEPTH 32
    struct brw_compile *p = &c->func;
-   GLuint nr_insns = c->vp->program.Base.NumInstructions;
-   GLuint insn, if_insn = 0;
+   const GLuint nr_insns = c->vp->program.Base.NumInstructions;
+   GLuint insn, if_depth = 0;
    GLuint end_offset = 0;
    struct brw_instruction *end_inst, *last_inst;
-   struct brw_instruction *if_inst[MAX_IFSN];
-   struct brw_indirect stack_index = brw_indirect(0, 0);   
-
+   struct brw_instruction *if_inst[MAX_IF_DEPTH];
+   const struct brw_indirect stack_index = brw_indirect(0, 0);   
    GLuint index;
    GLuint file;
 
@@ -1376,15 +1410,18 @@ void brw_vs_emit(struct brw_vs_compile *c )
 	 emit_xpd(p, dst, args[0], args[1]);
 	 break;
       case OPCODE_IF:
-	 assert(if_insn < MAX_IFSN);
-         if_inst[if_insn++] = brw_IF(p, BRW_EXECUTE_8);
+	 assert(if_depth < MAX_IF_DEPTH);
+	 if_inst[if_depth] = brw_IF(p, BRW_EXECUTE_8);
+	 if_inst[if_depth]->header.predicate_control =
+	    get_predicate(inst->DstReg.CondSwizzle);
+	 if_depth++;
 	 break;
       case OPCODE_ELSE:
-	 if_inst[if_insn-1] = brw_ELSE(p, if_inst[if_insn-1]);
+	 if_inst[if_depth-1] = brw_ELSE(p, if_inst[if_depth-1]);
 	 break;
       case OPCODE_ENDIF:
-         assert(if_insn > 0);
-	 brw_ENDIF(p, if_inst[--if_insn]);
+         assert(if_depth > 0);
+	 brw_ENDIF(p, if_inst[--if_depth]);
 	 break;			
       case OPCODE_BRA:
          brw_set_predicate_control(p, BRW_PREDICATE_NORMAL);
@@ -1430,6 +1467,19 @@ void brw_vs_emit(struct brw_vs_compile *c )
 				    "unknown");
       }
 
+      /* Set the predication update on the last instruction of the native
+       * instruction sequence.
+       *
+       * This would be problematic if it was set on a math instruction,
+       * but that shouldn't be the case with the current GLSL compiler.
+       */
+      if (inst->CondUpdate) {
+	 struct brw_instruction *hw_insn = &p->store[p->nr_insn - 1];
+
+	 assert(hw_insn->header.destreg__conditionalmod == 0);
+	 hw_insn->header.destreg__conditionalmod = BRW_CONDITIONAL_NZ;
+      }
+
       if ((inst->DstReg.File == PROGRAM_OUTPUT)
           && (inst->DstReg.Index != VERT_RESULT_HPOS)
           && c->output_regs[inst->DstReg.Index].used_in_src) {

src/mesa/drivers/dri/i965/brw_wm.c

@@ -202,6 +202,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
    /* BRW_NEW_FRAGMENT_PROGRAM */
    const struct brw_fragment_program *fp = 
       (struct brw_fragment_program *)brw->fragment_program;
+   GLboolean uses_depth = (fp->program.Base.InputsRead & (1 << FRAG_ATTRIB_WPOS)) != 0;
    GLuint lookup = 0;
    GLuint line_aa;
    GLuint i;
@@ -263,6 +264,7 @@ static void brw_wm_populate_key( struct brw_context *brw,
 	 
    brw_wm_lookup_iz(line_aa,
 		    lookup,
+		    uses_depth,
 		    key);
 
 

src/mesa/drivers/dri/i965/brw_wm.h

@@ -286,6 +286,7 @@ void brw_wm_print_program( struct brw_wm_compile *c,
 
 void brw_wm_lookup_iz( GLuint line_aa,
 		       GLuint lookup,
+		       GLboolean ps_uses_depth,
 		       struct brw_wm_prog_key *key );
 
 GLboolean brw_wm_is_glsl(const struct gl_fragment_program *fp);

src/mesa/drivers/dri/i965/brw_wm_fp.c

@@ -705,7 +705,11 @@ static void precalc_tex( struct brw_wm_compile *c,
 	      tmpcoord,
 	      0,
 	      inst->SrcReg[0],
-	      scale,
+	      src_swizzle(scale,
+			  SWIZZLE_X,
+			  SWIZZLE_Y,
+			  SWIZZLE_ONE,
+			  SWIZZLE_ONE),
 	      src_undef());
 
       coord = src_reg_from_dst(tmpcoord);

src/mesa/drivers/dri/i965/brw_wm_iz.c

@@ -118,6 +118,7 @@ const struct {
 
 void brw_wm_lookup_iz( GLuint line_aa,
 		       GLuint lookup,
+		       GLboolean ps_uses_depth,
 		       struct brw_wm_prog_key *key )
 {
    GLuint reg = 2;
@@ -127,7 +128,7 @@ void brw_wm_lookup_iz( GLuint line_aa,
    if (lookup & IZ_PS_COMPUTES_DEPTH_BIT)
       key->computes_depth = 1;
 
-   if (wm_iz_table[lookup].sd_present) {
+   if (wm_iz_table[lookup].sd_present || ps_uses_depth) {
       key->source_depth_reg = reg;
       reg += 2;
    }

src/mesa/drivers/dri/i965/intel_pixel_read.c

@@ -0,0 +1 @@
+../intel/intel_pixel_read.c

src/mesa/drivers/dri/intel/intel_batchbuffer.c

@@ -197,6 +197,11 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
    GLuint used = batch->ptr - batch->map;
    GLboolean was_locked = intel->locked;
 
+   if (intel->first_post_swapbuffers_batch == NULL) {
+      intel->first_post_swapbuffers_batch = intel->batch->buf;
+      drm_intel_bo_reference(intel->first_post_swapbuffers_batch);
+   }
+
    if (used == 0) {
       batch->cliprect_mode = IGNORE_CLIPRECTS;
       return;

src/mesa/drivers/dri/intel/intel_buffer_objects.c

@@ -209,7 +209,10 @@ intel_bufferobj_get_subdata(GLcontext * ctx,
    struct intel_buffer_object *intel_obj = intel_buffer_object(obj);
 
    assert(intel_obj);
-   dri_bo_get_subdata(intel_obj->buffer, offset, size, data);
+   if (intel_obj->sys_buffer)
+      memcpy(data, (char *)intel_obj->sys_buffer + offset, size);
+   else
+      dri_bo_get_subdata(intel_obj->buffer, offset, size, data);
 }
 
 

src/mesa/drivers/dri/intel/intel_buffers.c

@@ -345,6 +345,23 @@ intelDrawBuffer(GLcontext * ctx, GLenum mode)
 static void
 intelReadBuffer(GLcontext * ctx, GLenum mode)
 {
+   if ((ctx->DrawBuffer != NULL) && (ctx->DrawBuffer->Name == 0)) {
+      struct intel_context *const intel = intel_context(ctx);
+      const GLboolean was_front_buffer_reading =
+	intel->is_front_buffer_reading;
+
+      intel->is_front_buffer_reading = (mode == GL_FRONT_LEFT)
+	|| (mode == GL_FRONT);
+
+      /* If we weren't front-buffer reading before but we are now, make sure
+       * that the front-buffer has actually been allocated.
+       */
+      if (!was_front_buffer_reading && intel->is_front_buffer_reading) {
+	 intel_update_renderbuffers(intel->driContext,
+				    intel->driContext->driDrawablePriv);
+      }
+   }
+
    if (ctx->ReadBuffer == ctx->DrawBuffer) {
       /* This will update FBO completeness status.
        * A framebuffer will be incomplete if the GL_READ_BUFFER setting

src/mesa/drivers/dri/intel/intel_chipset.h

@@ -66,6 +66,7 @@
 #define PCI_CHIP_Q45_G                  0x2E12
 #define PCI_CHIP_G45_G                  0x2E22
 #define PCI_CHIP_G41_G                  0x2E32
+#define PCI_CHIP_B43_G                  0x2E42
 
 #define IS_MOBILE(devid)	(devid == PCI_CHIP_I855_GM || \
 				 devid == PCI_CHIP_I915_GM || \
@@ -78,7 +79,8 @@
 #define IS_G45(devid)           (devid == PCI_CHIP_IGD_E_G || \
                                  devid == PCI_CHIP_Q45_G || \
                                  devid == PCI_CHIP_G45_G || \
-                                 devid == PCI_CHIP_G41_G)
+                                 devid == PCI_CHIP_G41_G || \
+                                 devid == PCI_CHIP_B43_G)
 #define IS_GM45(devid)          (devid == PCI_CHIP_GM45_GM)
 #define IS_G4X(devid)		(IS_G45(devid) || IS_GM45(devid))
 

src/mesa/drivers/dri/intel/intel_context.c

@@ -161,6 +161,9 @@ intelGetString(GLcontext * ctx, GLenum name)
       case PCI_CHIP_G41_G:
          chipset = "Intel(R) G41";
          break;
+      case PCI_CHIP_B43_G:
+         chipset = "Intel(R) B43";
+         break;
       default:
          chipset = "Unknown Intel Chipset";
          break;
@@ -220,7 +223,9 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
       struct intel_renderbuffer *stencil_rb;
 
       i = 0;
-      if ((intel->is_front_buffer_rendering || !intel_fb->color_rb[1])
+      if ((intel->is_front_buffer_rendering ||
+	   intel->is_front_buffer_reading ||
+	   !intel_fb->color_rb[1])
 	   && intel_fb->color_rb[0]) {
 	 attachments[i++] = __DRI_BUFFER_FRONT_LEFT;
 	 attachments[i++] = intel_bits_per_pixel(intel_fb->color_rb[0]);
@@ -495,7 +500,8 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush)
 
       if (screen->dri2.loader &&
           (screen->dri2.loader->base.version >= 2)
-	  && (screen->dri2.loader->flushFrontBuffer != NULL)) {
+	  && (screen->dri2.loader->flushFrontBuffer != NULL) &&
+          intel->driDrawable && intel->driDrawable->loaderPrivate) {
 	 (*screen->dri2.loader->flushFrontBuffer)(intel->driDrawable,
 						  intel->driDrawable->loaderPrivate);
 
@@ -505,7 +511,7 @@ intel_flush(GLcontext *ctx, GLboolean needs_mi_flush)
 	  * each of N places that do rendering.  This has worse performances,
 	  * but it is much easier to get correct.
 	  */
-	 if (intel->is_front_buffer_rendering) {
+	 if (!intel->is_front_buffer_rendering) {
 	    intel->front_buffer_dirty = GL_FALSE;
 	 }
       }
@@ -521,7 +527,27 @@ intelFlush(GLcontext * ctx)
 static void
 intel_glFlush(GLcontext *ctx)
 {
+   struct intel_context *intel = intel_context(ctx);
+
    intel_flush(ctx, GL_TRUE);
+
+   /* We're using glFlush as an indicator that a frame is done, which is
+    * what DRI2 does before calling SwapBuffers (and means we should catch
+    * people doing front-buffer rendering, as well)..
+    *
+    * Wait for the swapbuffers before the one we just emitted, so we don't
+    * get too many swaps outstanding for apps that are GPU-heavy but not
+    * CPU-heavy.
+    *
+    * Unfortunately, we don't have a handle to the batch containing the swap,
+    * and getting our hands on that doesn't seem worth it, so we just us the
+    * first batch we emitted after the last swap.
+    */
+   if (intel->first_post_swapbuffers_batch != NULL) {
+      drm_intel_bo_wait_rendering(intel->first_post_swapbuffers_batch);
+      drm_intel_bo_unreference(intel->first_post_swapbuffers_batch);
+      intel->first_post_swapbuffers_batch = NULL;
+   }
 }
 
 void
@@ -787,6 +813,8 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
       intel->prim.vb = NULL;
       dri_bo_unreference(intel->prim.vb_bo);
       intel->prim.vb_bo = NULL;
+      dri_bo_unreference(intel->first_post_swapbuffers_batch);
+      intel->first_post_swapbuffers_batch = NULL;
 
       if (release_texture_heaps) {
          /* This share group is about to go away, free our private
@@ -804,12 +832,23 @@ intelDestroyContext(__DRIcontextPrivate * driContextPriv)
 
       /* free the Mesa context */
       _mesa_free_context_data(&intel->ctx);
+
+      FREE(intel);
+      driContextPriv->driverPrivate = NULL;
    }
 }
 
 GLboolean
 intelUnbindContext(__DRIcontextPrivate * driContextPriv)
 {
+   struct intel_context *intel =
+      (struct intel_context *) driContextPriv->driverPrivate;
+
+   /* Deassociate the context with the drawables.
+    */
+   intel->driDrawable = NULL;
+   intel->driReadDrawable = NULL;
+
    return GL_TRUE;
 }
 

src/mesa/drivers/dri/intel/intel_context.h

@@ -191,6 +191,7 @@ struct intel_context
    GLboolean ttm;
 
    struct intel_batchbuffer *batch;
+   drm_intel_bo *first_post_swapbuffers_batch;
    GLboolean no_batch_wrap;
    unsigned batch_id;
 
@@ -294,6 +295,14 @@ struct intel_context
     * easily.
     */
    GLboolean is_front_buffer_rendering;
+   /**
+    * Track whether front-buffer is the current read target.
+    *
+    * This is closely associated with is_front_buffer_rendering, but may
+    * be set separately.  The DRI2 fake front buffer must be referenced
+    * either way.
+    */
+   GLboolean is_front_buffer_reading;
 
    drm_clip_rect_t fboRect;     /**< cliprect for FBO rendering */
 

src/mesa/drivers/dri/intel/intel_fbo.c

@@ -680,6 +680,11 @@ intel_validate_framebuffer(GLcontext *ctx, struct gl_framebuffer *fb)
       if (rb == NULL)
 	 continue;
 
+      if (irb == NULL) {
+	 fb->_Status = GL_FRAMEBUFFER_UNSUPPORTED_EXT;
+	 continue;
+      }
+
       switch (irb->texformat->MesaFormat) {
       case MESA_FORMAT_ARGB8888:
       case MESA_FORMAT_RGB565:

src/mesa/drivers/dri/intel/intel_pixel.c

@@ -342,10 +342,8 @@ intelInitPixelFuncs(struct dd_function_table *functions)
       functions->Bitmap = intelBitmap;
       functions->CopyPixels = intelCopyPixels;
       functions->DrawPixels = intelDrawPixels;
-#ifdef I915
-      functions->ReadPixels = intelReadPixels;
-#endif
    }
+   functions->ReadPixels = intelReadPixels;
 }
 
 void

src/mesa/drivers/dri/intel/intel_pixel_bitmap.c

@@ -409,6 +409,12 @@ intel_texture_bitmap(GLcontext * ctx,
       return GL_FALSE;
    }
 
+   if (ctx->Fog.Enabled) {
+      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+	 fprintf(stderr, "glBitmap() fallback: fog\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))) {

src/mesa/drivers/dri/intel/intel_pixel_read.c

@@ -0,0 +1,321 @@
+/**************************************************************************
+ *
+ * Copyright 2003 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sub license, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the
+ * next paragraph) shall be included in all copies or substantial portions
+ * of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
+ * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ *
+ **************************************************************************/
+
+#include "main/glheader.h"
+#include "main/enums.h"
+#include "main/mtypes.h"
+#include "main/macros.h"
+#include "main/image.h"
+#include "main/bufferobj.h"
+#include "main/state.h"
+#include "swrast/swrast.h"
+
+#include "intel_screen.h"
+#include "intel_context.h"
+#include "intel_batchbuffer.h"
+#include "intel_blit.h"
+#include "intel_buffers.h"
+#include "intel_regions.h"
+#include "intel_pixel.h"
+#include "intel_buffer_objects.h"
+
+/* For many applications, the new ability to pull the source buffers
+ * back out of the GTT and then do the packing/conversion operations
+ * in software will be as much of an improvement as trying to get the
+ * blitter and/or texture engine to do the work.
+ *
+ * This step is gated on private backbuffers.
+ *
+ * Obviously the frontbuffer can't be pulled back, so that is either
+ * an argument for blit/texture readpixels, or for blitting to a
+ * temporary and then pulling that back.
+ *
+ * When the destination is a pbo, however, it's not clear if it is
+ * ever going to be pulled to main memory (though the access param
+ * will be a good hint).  So it sounds like we do want to be able to
+ * choose between blit/texture implementation on the gpu and pullback
+ * and cpu-based copying.
+ *
+ * Unless you can magically turn client memory into a PBO for the
+ * duration of this call, there will be a cpu-based copying step in
+ * any case.
+ */
+
+
+static GLboolean
+do_texture_readpixels(GLcontext * ctx,
+                      GLint x, GLint y, GLsizei width, GLsizei height,
+                      GLenum format, GLenum type,
+                      const struct gl_pixelstore_attrib *pack,
+                      struct intel_region *dest_region)
+{
+#if 0
+   struct intel_context *intel = intel_context(ctx);
+   intelScreenPrivate *screen = intel->intelScreen;
+   GLint pitch = pack->RowLength ? pack->RowLength : width;
+   __DRIdrawablePrivate *dPriv = intel->driDrawable;
+   int textureFormat;
+   GLenum glTextureFormat;
+   int destFormat, depthFormat, destPitch;
+   drm_clip_rect_t tmp;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+
+   if (ctx->_ImageTransferState ||
+       pack->SwapBytes || pack->LsbFirst || !pack->Invert) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         fprintf(stderr, "%s: check_color failed\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   intel->vtbl.meta_texrect_source(intel, intel_readbuf_region(intel));
+
+   if (!intel->vtbl.meta_render_dest(intel, dest_region, type, format)) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         fprintf(stderr, "%s: couldn't set dest %s/%s\n",
+                 __FUNCTION__,
+                 _mesa_lookup_enum_by_nr(type),
+                 _mesa_lookup_enum_by_nr(format));
+      return GL_FALSE;
+   }
+
+   LOCK_HARDWARE(intel);
+
+   if (intel->driDrawable->numClipRects) {
+      intel->vtbl.install_meta_state(intel);
+      intel->vtbl.meta_no_depth_write(intel);
+      intel->vtbl.meta_no_stencil_write(intel);
+
+      if (!driClipRectToFramebuffer(ctx->ReadBuffer, &x, &y, &width, &height)) {
+         UNLOCK_HARDWARE(intel);
+         SET_STATE(i830, state);
+         if (INTEL_DEBUG & DEBUG_PIXEL)
+            fprintf(stderr, "%s: cliprect failed\n", __FUNCTION__);
+         return GL_TRUE;
+      }
+
+      y = dPriv->h - y - height;
+      x += dPriv->x;
+      y += dPriv->y;
+
+
+      /* Set the frontbuffer up as a large rectangular texture.
+       */
+      intel->vtbl.meta_tex_rect_source(intel, src_region, textureFormat);
+
+
+      intel->vtbl.meta_texture_blend_replace(i830, glTextureFormat);
+
+
+      /* Set the 3d engine to draw into the destination region:
+       */
+
+      intel->vtbl.meta_draw_region(intel, dest_region);
+      intel->vtbl.meta_draw_format(intel, destFormat, depthFormat);     /* ?? */
+
+
+      /* Draw a single quad, no cliprects:
+       */
+      intel->vtbl.meta_disable_cliprects(intel);
+
+      intel->vtbl.draw_quad(intel,
+                            0, width, 0, height,
+                            0x00ff00ff, x, x + width, y, y + height);
+
+      intel->vtbl.leave_meta_state(intel);
+   }
+   UNLOCK_HARDWARE(intel);
+
+   intel_region_wait_fence(ctx, dest_region);   /* required by GL */
+   return GL_TRUE;
+#endif
+
+   return GL_FALSE;
+}
+
+
+
+
+static GLboolean
+do_blit_readpixels(GLcontext * ctx,
+                   GLint x, GLint y, GLsizei width, GLsizei height,
+                   GLenum format, GLenum type,
+                   const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
+{
+   struct intel_context *intel = intel_context(ctx);
+   struct intel_region *src = intel_readbuf_region(intel);
+   struct intel_buffer_object *dst = intel_buffer_object(pack->BufferObj);
+   GLuint dst_offset;
+   GLuint rowLength;
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s\n", __FUNCTION__);
+
+   if (!src)
+      return GL_FALSE;
+
+   if (dst) {
+      /* XXX This validation should be done by core mesa:
+       */
+      if (!_mesa_validate_pbo_access(2, pack, width, height, 1,
+                                     format, type, pixels)) {
+         _mesa_error(ctx, GL_INVALID_OPERATION, "glDrawPixels");
+         return GL_TRUE;
+      }
+   }
+   else {
+      /* PBO only for now:
+       */
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s - not PBO\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+
+   if (ctx->_ImageTransferState ||
+       !intel_check_blit_format(src, format, type)) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s - bad format for blit\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s: bad packing params\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+
+   if (pack->RowLength > 0)
+      rowLength = pack->RowLength;
+   else
+      rowLength = width;
+
+   if (pack->Invert) {
+      if (INTEL_DEBUG & DEBUG_PIXEL)
+         _mesa_printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
+      return GL_FALSE;
+   }
+   else {
+      rowLength = -rowLength;
+   }
+
+   /* XXX 64-bit cast? */
+   dst_offset = (GLuint) _mesa_image_address(2, pack, pixels, width, height,
+                                             format, type, 0, 0, 0);
+
+
+   /* Although the blits go on the command buffer, need to do this and
+    * fire with lock held to guarentee cliprects are correct.
+    */
+   intelFlush(&intel->ctx);
+   LOCK_HARDWARE(intel);
+
+   if (intel->driDrawable->numClipRects) {
+      GLboolean all = (width * height * src->cpp == dst->Base.Size &&
+                       x == 0 && dst_offset == 0);
+
+      dri_bo *dst_buffer = intel_bufferobj_buffer(intel, dst,
+						  all ? INTEL_WRITE_FULL :
+						  INTEL_WRITE_PART);
+      __DRIdrawablePrivate *dPriv = intel->driDrawable;
+      int nbox = dPriv->numClipRects;
+      drm_clip_rect_t *box = dPriv->pClipRects;
+      drm_clip_rect_t rect;
+      drm_clip_rect_t src_rect;
+      int i;
+
+      src_rect.x1 = dPriv->x + x;
+      src_rect.y1 = dPriv->y + dPriv->h - (y + height);
+      src_rect.x2 = src_rect.x1 + width;
+      src_rect.y2 = src_rect.y1 + height;
+
+
+
+      for (i = 0; i < nbox; i++) {
+         if (!intel_intersect_cliprects(&rect, &src_rect, &box[i]))
+            continue;
+
+         intelEmitCopyBlit(intel,
+                           src->cpp,
+                           src->pitch, src->buffer, 0, src->tiling,
+                           rowLength, dst_buffer, dst_offset, GL_FALSE,
+                           rect.x1,
+                           rect.y1,
+                           rect.x1 - src_rect.x1,
+                           rect.y2 - src_rect.y2,
+                           rect.x2 - rect.x1, rect.y2 - rect.y1,
+			   GL_COPY);
+      }
+   }
+   UNLOCK_HARDWARE(intel);
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s - DONE\n", __FUNCTION__);
+
+   return GL_TRUE;
+}
+
+void
+intelReadPixels(GLcontext * ctx,
+                GLint x, GLint y, GLsizei width, GLsizei height,
+                GLenum format, GLenum type,
+                const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
+{
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      fprintf(stderr, "%s\n", __FUNCTION__);
+
+   intelFlush(ctx);
+
+#ifdef I915
+   if (do_blit_readpixels
+       (ctx, x, y, width, height, format, type, pack, pixels))
+      return;
+
+   if (do_texture_readpixels
+       (ctx, x, y, width, height, format, type, pack, pixels))
+      return;
+#else
+   (void)do_blit_readpixels;
+   (void)do_texture_readpixels;
+#endif
+
+   if (INTEL_DEBUG & DEBUG_PIXEL)
+      _mesa_printf("%s: fallback to swrast\n", __FUNCTION__);
+
+   /* Update Mesa state before calling down into _swrast_ReadPixels, as
+    * the spans code requires the computed buffer states to be up to date,
+    * but _swrast_ReadPixels only updates Mesa state after setting up
+    * the spans code.
+    */
+
+   if (ctx->NewState)
+      _mesa_update_state(ctx);
+
+   _swrast_ReadPixels(ctx, x, y, width, height, format, type, pack, pixels);
+}

src/mesa/drivers/dri/intel/intel_regions.c

@@ -114,6 +114,13 @@ intel_region_alloc(struct intel_context *intel,
 {
    dri_bo *buffer;
 
+   /* If we're untiled, we have to align to 2 rows high because the
+    * data port accesses 2x2 blocks even if the bottom row isn't to be
+    * rendered, so failure to align means we could walk off the end of the
+    * GTT and fault.
+    */
+   height = ALIGN(height, 2);
+
    if (expect_accelerated_upload) {
       buffer = drm_intel_bo_alloc_for_render(intel->bufmgr, "region",
 					     pitch * cpp * height, 64);

src/mesa/drivers/dri/intel/intel_screen.c

@@ -305,7 +305,7 @@ intelDestroyScreen(__DRIscreenPrivate * sPriv)
 
    dri_bufmgr_destroy(intelScreen->bufmgr);
    intelUnmapScreenRegions(intelScreen);
-   driDestroyOptionCache(&intelScreen->optionCache);
+   driDestroyOptionInfo(&intelScreen->optionCache);
 
    FREE(intelScreen);
    sPriv->private = NULL;
@@ -617,10 +617,10 @@ intel_init_bufmgr(intelScreenPrivate *intelScreen)
    /* Otherwise, use the classic buffer manager. */
    if (intelScreen->bufmgr == NULL) {
       if (gem_disable) {
-	 fprintf(stderr, "GEM disabled.  Using classic.\n");
+	 _mesa_warning(NULL, "GEM disabled.  Using classic.");
       } else {
-	 fprintf(stderr, "Failed to initialize GEM.  "
-		 "Falling back to classic.\n");
+	 _mesa_warning(NULL,
+                       "Failed to initialize GEM.  Falling back to classic.");
       }
 
       if (intelScreen->tex.size == 0) {

src/mesa/drivers/dri/radeon/radeon_screen.c

@@ -283,12 +283,12 @@ radeonFillInModes( __DRIscreenPrivate *psp,
      * with a stencil buffer.  It will be a sw fallback, but some apps won't
      * care about that.
      */
-    stencil_bits_array[0] = 0;
+    stencil_bits_array[0] = stencil_bits;
     stencil_bits_array[1] = (stencil_bits == 0) ? 8 : stencil_bits;
 
     msaa_samples_array[0] = 0;
 
-    depth_buffer_factor = ((depth_bits != 0) || (stencil_bits != 0)) ? 2 : 1;
+    depth_buffer_factor = (stencil_bits == 0) ? 2 : 1;
     back_buffer_factor  = (have_back_buffer) ? 2 : 1;
 
     if (pixel_bits == 16) {

src/mesa/drivers/windows/gdi/mesa.def

@@ -902,7 +902,6 @@ EXPORTS
 	_mesa_generate_mipmap
 	_mesa_get_compressed_teximage
 	_mesa_get_current_context
-	_mesa_get_program_register
 	_mesa_get_teximage
 	_mesa_init_driver_functions
 	_mesa_init_glsl_driver_functions

src/mesa/main/bufferobj.c

@@ -1040,7 +1040,7 @@ _mesa_MapBufferARB(GLenum target, GLenum access)
    ASSERT(ctx->Driver.MapBuffer);
    bufObj->Pointer = ctx->Driver.MapBuffer( ctx, target, access, bufObj );
    if (!bufObj->Pointer) {
-      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(access)");
+      _mesa_error(ctx, GL_OUT_OF_MEMORY, "glMapBufferARB(map failed)");
    }
 
    bufObj->Access = access;

src/mesa/main/compiler.h

@@ -110,10 +110,8 @@ extern "C" {
 #if defined(_WIN32) && !defined(__WIN32__) && !defined(__CYGWIN__) && !defined(BUILD_FOR_SNAP)
 #  define __WIN32__
 #  define finite _finite
-#endif
-#if defined(__WATCOMC__)
+#elif defined(__WATCOMC__)
 #  define finite _finite
-#  pragma disable_message(201) /* Disable unreachable code warnings */
 #endif
 
 
@@ -135,6 +133,10 @@ extern "C" {
 #    endif
 #  endif
 #endif
+#if defined(__WATCOMC__)
+#  pragma disable_message(201) /* Disable unreachable code warnings */
+#endif
+
 
 
 /**

src/mesa/main/config.h

@@ -138,9 +138,14 @@
 /** 
  * Maximum viewport/image width. Must accomodate all texture sizes too. 
  */
-#define MAX_WIDTH 4096
+
+#ifndef MAX_WIDTH
+#   define MAX_WIDTH 4096
+#endif
 /** Maximum viewport/image height */
-#define MAX_HEIGHT 4096
+#ifndef MAX_HEIGHT
+#   define MAX_HEIGHT 4096
+#endif
 
 /** Maxmimum size for CVA.  May be overridden by the drivers.  */
 #define MAX_ARRAY_LOCK_SIZE 3000

src/mesa/main/enable.c

@@ -231,7 +231,7 @@ enable_texture(GLcontext *ctx, GLboolean state, GLbitfield bit)
    const GLuint newenabled = (!state)
        ? (texUnit->Enabled & ~bit) :  (texUnit->Enabled | bit);
 
-   if (!ctx->DrawBuffer->Visual.rgbMode || texUnit->Enabled == newenabled)
+   if (texUnit->Enabled == newenabled)
        return GL_FALSE;
 
    FLUSH_VERTICES(ctx, _NEW_TEXTURE);
@@ -933,11 +933,6 @@ _mesa_set_enable(GLcontext *ctx, GLenum cap, GLboolean state)
       /* GL_EXT_depth_bounds_test */
       case GL_DEPTH_BOUNDS_TEST_EXT:
          CHECK_EXTENSION(EXT_depth_bounds_test, cap);
-         if (state && ctx->DrawBuffer->Visual.depthBits == 0) {
-            _mesa_warning(ctx,
-                   "glEnable(GL_DEPTH_BOUNDS_TEST_EXT) but no depth buffer");
-            return;
-         }
          if (ctx->Depth.BoundsTest == state)
             return;
          FLUSH_VERTICES(ctx, _NEW_DEPTH);

src/mesa/main/extensions.c

@@ -404,6 +404,7 @@ _mesa_enable_2_0_extensions(GLcontext *ctx)
    ctx->Extensions.ARB_fragment_shader = GL_TRUE;
 #endif
    ctx->Extensions.ARB_point_sprite = GL_TRUE;
+   ctx->Extensions.EXT_blend_equation_separate = GL_TRUE;
    ctx->Extensions.ARB_texture_non_power_of_two = GL_TRUE;
 #if FEATURE_ARB_shader_objects
    ctx->Extensions.ARB_shader_objects = GL_TRUE;