docs: Add change log to 7.9.1 release notes

This workaround fixes rendering of kwin thumbnails.

NOTE: This is a candidate for the 7.9 and 7.10 branches.
(cherry picked from commit 8543902bfb)

Conflicts:

	src/mesa/drivers/dri/r300/compiler/radeon_rename_regs.c

Makefile

@@ -180,7 +180,7 @@ ultrix-gcc:
 
 # Rules for making release tarballs
 
-VERSION=7.9-rc2
+VERSION=7.9.1
 DIRECTORY = Mesa-$(VERSION)
 LIB_NAME = MesaLib-$(VERSION)
 GLUT_NAME = MesaGLUT-$(VERSION)
@@ -329,6 +329,8 @@ GALLIUM_FILES = \
 	$(DIRECTORY)/src/gallium/Makefile.template			\
 	$(DIRECTORY)/src/gallium/SConscript				\
 	$(DIRECTORY)/src/gallium/targets/Makefile.dri			\
+	$(DIRECTORY)/src/gallium/targets/Makefile.xorg			\
+	$(DIRECTORY)/src/gallium/targets/SConscript.dri			\
 	$(DIRECTORY)/src/gallium/*/Makefile				\
 	$(DIRECTORY)/src/gallium/*/SConscript				\
 	$(DIRECTORY)/src/gallium/*/*/Makefile				\
@@ -345,23 +347,19 @@ GALLIUM_FILES = \
 
 DRI_FILES = \
 	$(DIRECTORY)/include/GL/internal/dri_interface.h		\
-	$(DIRECTORY)/include/GL/internal/glcore.h			\
 	$(DIRECTORY)/include/GL/internal/sarea.h			\
 	$(DIRECTORY)/src/glx/Makefile					\
-	$(DIRECTORY)/src/glx/Makefile					\
 	$(DIRECTORY)/src/glx/*.[ch]					\
 	$(DIRECTORY)/src/mesa/drivers/dri/Makefile			\
 	$(DIRECTORY)/src/mesa/drivers/dri/Makefile.template		\
 	$(DIRECTORY)/src/mesa/drivers/dri/dri.pc.in			\
-	$(DIRECTORY)/src/mesa/drivers/dri/common/xmlpool/*.[ch]		\
 	$(DIRECTORY)/src/mesa/drivers/dri/common/xmlpool/*.po		\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/*.[chS]			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/*.cpp			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/*/*.[chS]			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/Makefile			\
 	$(DIRECTORY)/src/mesa/drivers/dri/*/*/Makefile			\
-	$(DIRECTORY)/src/mesa/drivers/dri/*/Doxyfile			\
-	$(DIRECTORY)/src/mesa/drivers/dri/*/server/*.[ch]
+	$(DIRECTORY)/src/mesa/drivers/dri/*/Doxyfile
 
 SGI_GLU_FILES = \
 	$(DIRECTORY)/src/glu/Makefile					\

configs/autoconf.in

@@ -53,7 +53,7 @@ MKDEP_OPTIONS = @MKDEP_OPTIONS@
 INSTALL = @INSTALL@
 
 # Python and flags (generally only needed by the developers)
-PYTHON2 = python
+PYTHON2 = @PYTHON2@
 PYTHON_FLAGS = -t -O -O
 
 # Library names (base name)

configure.ac

@@ -30,6 +30,7 @@ AC_PROG_CPP
 AC_PROG_CC
 AC_PROG_CXX
 AC_CHECK_PROGS([MAKE], [gmake make])
+AC_CHECK_PROGS([PYTHON2], [python2 python])
 AC_PATH_PROG([MKDEP], [makedepend])
 AC_PATH_PROG([SED], [sed])
 
@@ -1687,6 +1688,8 @@ echo ""
 echo "        CFLAGS:          $cflags"
 echo "        CXXFLAGS:        $cxxflags"
 echo "        Macros:          $defines"
+echo ""
+echo "        PYTHON2:         $PYTHON2"
 
 echo ""
 echo "        Run '${MAKE-make}' to build Mesa"

docs/devinfo.html

@@ -145,7 +145,7 @@ Make sure the values in src/mesa/main/version.h are correct.
 </p>
 
 <p>
-Update the docs/news.html file and docs/download.html files.
+Update docs/news.html.
 </p>
 
 <p>
@@ -208,10 +208,11 @@ sftp USERNAME,mesa3d@web.sourceforge.net
 
 <p>
 Make an announcement on the mailing lists:
-<em>m</em><em>e</em><em>s</em><em>a</em><em>3</em><em>d</em><em>-</em><em>d</em><em>e</em><em>v</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>s</em><em>f</em><em>.</em><em>n</em><em>e</em><em>t</em>,
-<em>m</em><em>e</em><em>s</em><em>a</em><em>3</em><em>d</em><em>-</em><em>u</em><em>s</em><em>e</em><em>r</em><em>s</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>s</em><em>f</em><em>.</em><em>n</em><em>e</em><em>t</em>
+
+<em>m</em><em>e</em><em>s</em><em>a</em><em>-</em><em>d</em><em>e</em><em>v</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>f</em><em>r</em><em>e</em><em>e</em><em>d</em><em>e</em><em>s</em><em>k</em><em>t</em><em>o</em><em>p</em><em>.</em><em>o</em><em>r</em><em>g</em>,
+<em>m</em><em>e</em><em>s</em><em>a</em><em>-</em><em>u</em><em>s</em><em>e</em><em>r</em><em>s</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>f</em><em>r</em><em>e</em><em>e</em><em>d</em><em>e</em><em>s</em><em>k</em><em>t</em><em>o</em><em>p</em><em>.</em><em>o</em><em>r</em><em>g</em>
 and
-<em>m</em><em>e</em><em>s</em><em>a</em><em>3</em><em>d</em><em>-</em><em>a</em><em>n</em><em>n</em><em>o</em><em>u</em><em>n</em><em>c</em><em>e</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>s</em><em>f</em><em>.</em><em>n</em><em>e</em><em>t</em>
+<em>m</em><em>e</em><em>s</em><em>a</em><em>-</em><em>a</em><em>n</em><em>n</em><em>o</em><em>u</em><em>n</em><em>c</em><em>e</em><em>@</em><em>l</em><em>i</em><em>s</em><em>t</em><em>s</em><em>.</em><em>f</em><em>r</em><em>e</em><em>e</em><em>d</em><em>e</em><em>s</em><em>k</em><em>t</em><em>o</em><em>p</em><em>.</em><em>o</em><em>r</em><em>g</em>
 </p>
 
 

docs/news.html

@@ -11,10 +11,18 @@
 <H1>News</H1>
 
 
+<h2>October 4, 2010</h2>
+
+<p>
+<a href="relnotes-7.9.html">Mesa 7.9</a> (final) is released.  This is a new
+development release.
+</p>
+
+
 <h2>September 27, 2010</h2>
 
 <p>
-<a href="relnotes-7.9.0.html">Mesa 7.9.0-rc1</a> is released.  This is a
+<a href="relnotes-7.9.html">Mesa 7.9.0-rc1</a> is released.  This is a
 release candidate for the 7.9 development release.
 </p>
 

docs/relnotes-7.8.2.html

@@ -26,7 +26,15 @@ for DRI hardware acceleration.
 
 <h2>MD5 checksums</h2>
 <pre>
-tbd
+c89b63d253605ed40e8ac370d25a833c  MesaLib-7.8.2.tar.gz
+6be2d343a0089bfd395ce02aaf8adb57  MesaLib-7.8.2.tar.bz2
+a04ad3b06ac5ff3969a003fa7bbf7d5b  MesaLib-7.8.2.zip
+7c213f92efeb471f0331670d5079d4c0  MesaDemos-7.8.2.tar.gz
+757d9e2e06f48b1a52848be9b0307ced  MesaDemos-7.8.2.tar.bz2
+8d0e5cfe68b8ebf90265d350ae2c48b1  MesaDemos-7.8.2.zip
+b74482e3f44f35ed395c4aada4fd8240  MesaGLUT-7.8.2.tar.gz
+a471807b65e49c325808ba4551be93ed  MesaGLUT-7.8.2.tar.bz2
+9f190268c42be582ef66e47365ee61e3  MesaGLUT-7.8.2.zip
 </pre>
 
 
@@ -44,10 +52,95 @@ tbd
 <ul>
 <li>Fixed Gallium glDrawPixels(GL_DEPTH_COMPONENT).
 <li>Fixed Gallium Cell driver to buildable, runable state
+<li>Fixed bad error checking for glFramebufferRenderbuffer(attachment=GL_DEPTH_STENCIL_ATTACHMENT).
+<li>Fixed incorrect Z coordinate handling in "meta" glDraw/CopyPixels.
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=23670">Bug
+    #23670</a>.</li>
+
+<li>Assorted i965 driver fixes.
+    Including but not limited to:
+  <ul>
+    <li>Fix scissoring when width or height is
+    0.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27643">Bug
+    #27643</a>.
+    <li>Fix bit allocation for number of color regions for
+    ARB_draw_buffers.</li>
+    <li>Set the correct provoking vertex for clipped first-mode
+    trifans.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=24470">Bug
+    #24470</a>.</li>
+    <li>Use <code>R16G16B16A16_FLOAT</code> for 3-component half-float.</li>
+    <li>Fix assertion for surface tile offset usage on Ironlake.</li>
+    <li>Fix cube map layouts on Ironlake.</li>
+    <li>When an RB gets a new region, clear the old from the state
+    cache.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=24119">Bug
+    #24119</a>.</li>
+    <li>Reject shaders with uninlined function calls instead of hanging.</li>
+  </ul>
+</li>
+
+<li>Assorted i915 driver fixes.  Including but not limited to:
+  <ul>
+    <li>Fixed texture LOD clamping in i915 driver.
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=24846">Bug
+    #24846</a>.</li>
+    <li>Fix off-by-one for drawing rectangle.
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27408">Bug
+    #27408</a>.</li>
+  </ul>
+</li>
+
+<li>Fixed hangs in etracer on 830 and 845
+    chipsets.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=26557">Bug
+    #26557</a>.</li>
+<li>Fixed tiling of small textures on all Intel drivers.</li>
+<li>Fixed crash in Savage driver when using <code>_mesa_CopyTexImage2D</code>.
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27652">Bug
+    #27652</a>.</li>
+
+<li>Assorted GLX fixes.  Including but not limited to:
+  <ul>
+    <li>Fixed <code>__glXInitializeVisualConfigFromTags</code>'s handling of
+        unrecognized fbconfig tags.</li>
+    <li>Fixed regression with <code>GLX_USE_GL</code>.
+    <li>Fixed config chooser logic for 'mask' matching.</li>
+    <li>Report swap events correctly in direct rendered case (DRI2)</li>
+    <li>Fixed build with dri2proto which doesn't define
+        <code>X_DRI2SwapInterval</code>.</li>
+    <li>Get <code>GLX_SCREEN</code> first in <code>__glXQueryContextInfo</code>.
+        <a href="https://bugs.freedesktop.org/show_bug.cgi?id=14245">Bug
+	#14245</a>.</li>
+  </ul>
+</li>
+
+<li>Assorted GLSL fixes.  Including but not limited to:
+  <ul>
+    <li>Change variable declared assertion into conditional in GLSL
+        compiler.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27921">Bug
+        #27921</a>.</li>
+    <li>Fix instruction indexing
+        bugs.  <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27566">Bug
+        #27566</a>.</li>
+    <li>Updated uniform location / offset encoding to be more like
+        other implementations.</li>
+    <li>Don't overwrite a driver's shader infolog with generic failure
+        message.</li>
+  </ul>
+</li>
+
+<li>Fixed OSMesa build for 16 and 32-bit color channel depth.
+<li>Fixed OSMesa build with hidden symbol visibility. libOSMesa no longer links to libGL.
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=28305">Bug
+    #28305</a>.
+<li>Fixed handling of multiple render targets in fixed-function
+    texture envrionmnent programs.</li>
+<li>Fixed conversion errors in <code>signed_rgba8888[rev]</code> texel
+    fetch.</li>
+<li>Don't set srcLevel on <code>GL_TEXTURE_RECTANGLE_ARB</code> targets.</li>
+<li>Various build fixes for OpenBSD.</li>
+<li>Various build fixes for OS X.</li>
+<li>Various build fixes for GCC 3.3.</li>
 </ul>
 
 
-<h2>Changes</h2>
-<p>None.</p>
 </body>
 </html>

docs/relnotes-7.8.3.html

@@ -0,0 +1,89 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.8.3 Release Notes / (date tbd)</H1>
+
+<p>
+Mesa 7.8.3 is a bug fix release which fixes bugs found since the 7.8.2 release.
+</p>
+<p>
+Mesa 7.8.3 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>
+x  MesaLib-7.8.3.tar.gz
+x  MesaLib-7.8.3.tar.bz2
+x  MesaLib-7.8.3.zip
+x  MesaDemos-7.8.3.tar.gz
+x  MesaDemos-7.8.3.tar.bz2
+x  MesaDemos-7.8.3.zip
+x  MesaGLUT-7.8.3.tar.gz
+x  MesaGLUT-7.8.3.tar.bz2
+x  MesaGLUT-7.8.3.zip
+</pre>
+
+
+<h2>New features</h2>
+<p>None.</p>
+
+
+<h2>Changes</h2>
+<ul>
+<li>The radeon driver should use less memory when searching for a valid mip
+image.</li>
+</ul>
+
+
+<h2>Bug fixes</h2>
+<ul>
+<li>Fix unsupported FB with D24S8 (bug 
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=23670">29116</a>)</li>
+<li>Fix ReadPixels crash when reading depth/stencil from an FBO</li>
+<li>Fixed a bug rendering to 16-bit buffers using swrast.</li>
+<li>Fixed a state tracker/TGSI bug that caused crashes when using Windows'
+    memory debugging features.</li>
+<li>Fixed an issue rendering to 32-bit channels with swrast (bug
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=29487">29487</a>)</li>
+<li>GLSL: fix indirect <TT>gl_TextureMatrix</TT> addressing (bug
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=28967">28967</a>)</li>
+<li>GLSL: fix for bug
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=27216">27216</a></li>
+<li>GLSL: fix zw fragcoord entries in some cases (bug
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=29183">29183</a>)</li>
+<li>Fix texture env generation in some cases (bug
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=28169">28169</a>)</li>
+<li>osmesa: a fix for calling <TT>OSMesaMakeCurrent</TT> twice was applied (bug 
+    <a href="https://bugs.freedesktop.org/show_bug.cgi?id=10966">10966</a></li>
+<li>A bug was fixed which could cause Mesa to ignore the
+    <TT>MESA_EXTENSION_OVERRIDE</TT> environment variable.</li>
+<li>A bug related to specular highlights on backfaces was fixed.</li>
+<li>A radeon-specific issue with <TT>glCopyTex(Sub)Image</TT> was
+    corrected.</li>
+<li>radeon/wine: flush command stream in more cases, fixing wine d3d9
+    tests.</li>
+<li>r600: fix sin+cos normalization.</li>
+<li>r600: (properly) ignore <TT>GL_COORD_REPLACE</TT> when point sprites are
+    disabled.</li>
+<li>radeon: avoid flushing when the context is not current.</li>
+<li>r300c: a bug affecting unaligned BOs was fixed.</li>
+<li>r300c: a hardlock caused by ARB_half_float_vertex incorrectly advertised on some chipsets.</li>
+</ul>
+
+
+</body>
+</html>

docs/relnotes-7.9.1.html

@@ -0,0 +1,399 @@
+<HTML>
+
+<TITLE>Mesa Release Notes</TITLE>
+
+<head><link rel="stylesheet" type="text/css" href="mesa.css"></head>
+
+<BODY>
+
+<body bgcolor="#eeeeee">
+
+<H1>Mesa 7.9.1 Release Notes / January 7, 2011</H1>
+
+<p>
+Mesa 7.9.1 is a bug fix release which fixes bugs found since the 7.9 release.
+</p>
+<p>
+Mesa 7.9.1 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>
+TBD.
+</pre>
+
+
+<h2>New features</h2>
+<p>None.</p>
+
+<h2>Bug fixes</h2>
+<p>This list is likely incomplete.</p>
+<ul>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28800">Bug 28800</a> - [r300c, r300g] Texture corruption with World of Warcraft</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29420">Bug 29420</a> - Amnesia / HPL2 RendererFeatTest - not rendering correctly</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29946">Bug 29946</a> - [swrast] piglit valgrind glsl-array-bounds-04 fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30261">Bug 30261</a> - [GLSL 1.20] allowing inconsistent invariant declaration between two vertex shaders</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30632">Bug 30632</a> - [softpipe] state_tracker/st_manager.c:489: st_context_notify_invalid_framebuffer: Assertion `stfb &amp;&amp; stfb-&gt;iface == stfbi' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30694">Bug 30694</a> - wincopy will crash on Gallium drivers when going to front buffer</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30787">Bug 30787</a> - Invalid asm shader does not generate draw-time error when used with GLSL shader</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30993">Bug 30993</a> - getFramebufferAttachmentParameteriv wrongly generates error</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31101">Bug 31101</a> -  [glsl2] abort() in ir_validate::visit_enter(ir_assignment *ir)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31193">Bug 31193</a> -  [regression] aa43176e break water reflections</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31194">Bug 31194</a> - The mesa meta save/restore code doesn't ref the current GLSL program</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31371">Bug 31371</a> - glslparsertest: ir.cpp:358: ir_constant::ir_constant(const glsl_type*, const ir_constant_data*): Assertion `(type->base_type &gt;= 0) &amp;&amp; (type->base_type &lt;= 3)' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31439">Bug 31439</a> - Crash in glBufferSubData() with size == 0</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31495">Bug 31495</a> - [i965 gles2c bisected] OpenGL ES 2.0 conformance GL2Tests_GetBIFD_input.run regressed</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31514">Bug 31514</a> - isBuffer returns true for unbound buffers</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31560">Bug 31560</a> - [tdfx] tdfx_tex.c:702: error: ‘const struct gl_color_table’ has no member named ‘Format’</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31617">Bug 31617</a> - Radeon/Compiz: 'failed to attach dri2 front buffer', error case not handled</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31648">Bug 31648</a> -  [GLSL] array-struct-array gets assertion: `(size &gt;= 1) && (size &lt;= 4)' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31650">Bug 31650</a> - [GLSL] varying gl_TexCoord fails to be re-declared to different size in the second shader</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31673">Bug 31673</a> - GL_FRAGMENT_PRECISION_HIGH preprocessor macro undefined in GLSL ES</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31690">Bug 31690</a> -  i915 shader compiler fails to flatten if in Aquarium webgl demo.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31832">Bug 31832</a> - [i915] Bad renderbuffer format: 21</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31841">Bug 31841</a> - [drm:radeon_cs_ioctl] *ERROR* Invalid command stream !</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31894">Bug 31894</a> - Writing to gl_PointSize with GLES2 corrupts other varyings</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31909">Bug 31909</a> - [i965] brw_fs.cpp:1461: void fs_visitor::emit_bool_to_cond_code(ir_rvalue*): Assertion `expr-&gt;operands[i]-&gt;type-&gt;is_scalar()' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31934">Bug 31934</a> - [gallium] Mapping empty buffer object causes SIGSEGV</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31983">Bug 31983</a> -  [i915 gles2] "if (expression with builtin/varying variables) discard" breaks linkage</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31985">Bug 31985</a> - [GLSL 1.20] initialized uniform array considered as "unsized"</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=31987">Bug 31987</a> - [gles2] if input a wrong pname(GL_NONE) to glGetBoolean, it will not case GL_INVALID_ENUM</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32035">Bug 32035</a> - [GLSL bisected] comparing unsized array gets segfault</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32070">Bug 32070</a> - llvmpipe renders stencil demo incorrectly</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32273">Bug 32273</a> - assertion fails when starting vdrift 2010 release with shaders enabled</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32287">Bug 32287</a> - [bisected GLSL] float-int failure</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32311">Bug 32311</a> - [965 bisected] Array look-ups broken on GM45</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32520">Bug 32520</a> -  [gles2] glBlendFunc(GL_ZERO, GL_DST_COLOR) will result in GL_INVALID_ENUM</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=32825">Bug 32825</a> - egl_glx driver completely broken in 7.9 branch [fix in master]</li>
+</ul>
+
+
+<h2>Changes</h2>
+<p>The full set of changes can be viewed by using the following GIT command:</p>
+
+<pre>
+  git log mesa-7.9..mesa-7.9.1
+</pre>
+
+<p>Alex Deucher (5):
+<ul>
+  <li>r100: revalidate after radeon_update_renderbuffers</li>
+  <li>r600c: add missing radeon_prepare_render() call on evergreen</li>
+  <li>r600c: properly align mipmaps to group size</li>
+  <li>gallium/egl: fix r300 vs r600 loading</li>
+  <li>r600c: fix some opcodes on evergreen</li>
+</ul></p>
+
+<p>Aras Pranckevicius (2):
+<ul>
+  <li>glsl: fix crash in loop analysis when some controls can't be determined</li>
+  <li>glsl: fix matrix type check in ir_algebraic</li>
+</ul></p>
+
+<p>Brian Paul (27):
+<ul>
+  <li>swrast: fix choose_depth_texture_level() to respect mipmap filtering state</li>
+  <li>st/mesa: replace assertion w/ conditional in framebuffer invalidation</li>
+  <li>egl/i965: include inline_wrapper_sw_helper.h</li>
+  <li>mesa: Add missing else in do_row_3D</li>
+  <li>mesa: add missing formats in _mesa_format_to_type_and_comps()</li>
+  <li>mesa: handle more pixel types in mipmap generation code</li>
+  <li>mesa: make glIsBuffer() return false for never bound buffers</li>
+  <li>mesa: fix glDeleteBuffers() regression</li>
+  <li>swrast: init alpha value to 1.0 in opt_sample_rgb_2d()</li>
+  <li>meta: Mask Stencil.Clear against stencilMax in _mesa_meta_Clear</li>
+  <li>st/mesa: fix mapping of zero-sized buffer objects</li>
+  <li>mesa: check for posix_memalign() errors</li>
+  <li>llvmpipe: fix broken stencil writemask</li>
+  <li>mesa: fix GL_FRAMEBUFFER_ATTACHMENT_OBJECT_NAME query</li>
+  <li>mesa: return GL_FRAMEBUFFER_DEFAULT as FBO attachment type</li>
+  <li>mesa: make glGet*(GL_NONE) generate GL_INVALID_ENUM</li>
+  <li>mesa: test for cube map completeness in glGenerateMipmap()</li>
+  <li>tnl: Initialize gl_program_machine memory in run_vp.</li>
+  <li>tnl: a better way to initialize the gl_program_machine memory</li>
+  <li>mesa, st/mesa: disable GL_ARB_geometry_shader4</li>
+  <li>glsl: fix off by one in register index assertion</li>
+  <li>st/mesa: fix mipmap generation bug</li>
+  <li>glsl: new glsl_strtod() wrapper to fix decimal point interpretation</li>
+  <li>mesa: no-op glBufferSubData() on size==0</li>
+  <li>tdfx: s/Format/_BaseFormat/</li>
+  <li>st/mesa: fix renderbuffer pointer check in st_Clear()</li>
+  <li>mesa: Bump the number of bits in the register index.</li>
+</ul></p>
+
+<p>Chad Versace (5):
+<ul>
+  <li>glsl: Fix lexer rule for ^=</li>
+  <li>glsl: Fix ast-to-hir for ARB_fragment_coord_conventions</li>
+  <li>glsl: Fix ir_expression::constant_expression_value()</li>
+  <li>glsl: Fix erroneous cast in ast_jump_statement::hir()</li>
+  <li>glsl: Fix linker bug in cross_validate_globals()</li>
+</ul></p>
+
+<p>Chia-I Wu (10):
+<ul>
+  <li>targets/egl: Fix linking with libdrm.</li>
+  <li>st/vega: Fix version check in context creation.</li>
+  <li>st/egl: Do not finish a fence that is NULL.</li>
+  <li>egl: Fix a false negative check in _eglCheckMakeCurrent.</li>
+  <li>st/mesa: Unreference the sampler view in st_bind_surface.</li>
+  <li>egl_dri2: Fix __DRI_DRI2 version 1 support.</li>
+  <li>st/vega: Do not wait NULL fences.</li>
+  <li>mesa: Do not advertise GL_OES_texture_3D.</li>
+  <li>egl_glx: Fix borken driver.</li>
+  <li>egl: Check extensions.</li>
+</ul></p>
+
+<p>Daniel Lichtenberger (1):
+<ul>
+  <li>radeon: fix potential segfault in renderbuffer update</li>
+</ul></p>
+
+<p>Daniel Vetter (1):
+<ul>
+  <li>r200: revalidate after radeon_update_renderbuffers</li>
+</ul></p>
+
+<p>Dave Airlie (1):
+<ul>
+  <li>r300g: fixup rs690 tiling stride alignment calculations.</li>
+</ul></p>
+
+<p>Eric Anholt (13):
+<ul>
+  <li>intel: Allow CopyTexSubImage to InternalFormat 3/4 textures, like RGB/RGBA.</li>
+  <li>glsl: Free the loop state context when we free the loop state.</li>
+  <li>i965: Allow OPCODE_SWZ to put immediates in the first arg.</li>
+  <li>i965: Add support for rendering to SARGB8 FBOs.</li>
+  <li>glsl: Add a helper constructor for expressions that works out result type.</li>
+  <li>glsl: Fix structure and array comparisions.</li>
+  <li>glsl: Quiet unreachable no-return-from-function warning.</li>
+  <li>glsl: Mark the array access for whole-array comparisons.</li>
+  <li>glsl: Fix flipped return of has_value() for array constants.</li>
+  <li>mesa: Add getters for the rest of the supported draw buffers.</li>
+  <li>mesa: Add getters for ARB_copy_buffer's attachment points.</li>
+  <li>i965: Correct the dp_read message descriptor setup on g4x.</li>
+  <li>glsl: Correct the marking of InputsRead/OutputsWritten on in/out matrices.</li>
+</ul></p>
+
+<p>Fabian Bieler (1):
+<ul>
+  <li>glsl: fix lowering conditional returns in subroutines</li>
+</ul></p>
+
+<p>Francisco Jerez (3):
+<ul>
+  <li>meta: Don't leak alpha function/reference value changes.</li>
+  <li>meta: Fix incorrect rendering of the bitmap alpha component.</li>
+  <li>meta: Don't try to disable cube maps if the driver doesn't expose the extension.</li>
+</ul></p>
+
+<p>Henri Verbeet (2):
+<ul>
+  <li>r600: Evergreen has two extra frac_bits for the sampler LOD state.</li>
+  <li>st/mesa: Handle wrapped depth buffers in st_copy_texsubimage().</li>
+</ul></p>
+
+<p>Ian Romanick (33):
+<ul>
+  <li>Add 7.9 md5sums</li>
+  <li>docs: Import 7.8.x release notes from 7.8 branch.</li>
+  <li>docs: download.html does not need to be updated for each release</li>
+  <li>docs: Update mailing lines from sf.net to freedesktop.org</li>
+  <li>docs: added news item for 7.9 release</li>
+  <li>mesa: Validate assembly shaders when GLSL shaders are used</li>
+  <li>linker: Reject shaders that have unresolved function calls</li>
+  <li>mesa: Refactor validation of shader targets</li>
+  <li>glsl: Slightly change the semantic of _LinkedShaders</li>
+  <li>linker: Improve handling of unread/unwritten shader inputs/outputs</li>
+  <li>glsl: Commit lexer files changed by previous cherry picking</li>
+  <li>mesa: Make metaops use program refcounts instead of names.</li>
+  <li>glsl: Fix incorrect gl_type of sampler2DArray and sampler1DArrayShadow</li>
+  <li>mesa: Allow query of MAX_SAMPLES with EXT_framebuffer_multisample</li>
+  <li>glsl: better handling of linker failures</li>
+  <li>mesa: Fix glGet of ES2's GL_MAX_*_VECTORS properties.</li>
+  <li>i915: Disallow alpha, red, RG, and sRGB as render targets</li>
+  <li>glsl/linker: Free any IR discarded by optimization passes.</li>
+  <li>glsl: Add an optimization pass to simplify discards.</li>
+  <li>glsl: Add a lowering pass to move discards out of if-statements.</li>
+  <li>i915: Correctly generate unconditional KIL instructions</li>
+  <li>glsl: Add unary ir_expression constructor</li>
+  <li>glsl: Ensure that equality comparisons don't return a NULL IR tree</li>
+  <li>glcpp: Commit changes in generated files cause by previous commit</li>
+  <li>glsl: Inherrit type of declared variable from initializer</li>
+  <li>glsl: Inherrit type of declared variable from initializer after processing assignment</li>
+  <li>linker: Ensure that unsized arrays have a size after linking</li>
+  <li>linker: Fix regressions caused by previous commit</li>
+  <li>linker: Allow built-in arrays to have different sizes between shader stages</li>
+  <li>ir_to_mesa: Don't generate swizzles for record derefs of non-scalar/vectors</li>
+  <li>Refresh autogenerated file builtin_function.cpp.</li>
+  <li>docs: Initial set of release notes for 7.9.1</li>
+  <li>mesa: set version string to 7.9.1</li>
+</ul></p>
+
+<p>Julien Cristau (1):
+<ul>
+  <li>Makefile: don't include the same files twice in the tarball</li>
+</ul></p>
+
+<p>Kenneth Graunke (19):
+<ul>
+  <li>glcpp: Return NEWLINE token for newlines inside multi-line comments.</li>
+  <li>generate_builtins.py: Output large strings as arrays of characters.</li>
+  <li>glsl: Fix constant component count in vector constructor emitting.</li>
+  <li>ir_dead_functions: Actually free dead functions and signatures.</li>
+  <li>glcpp: Define GL_FRAGMENT_PRECISION_HIGH if GLSL version &gt;= 1.30.</li>
+  <li>glsl: Unconditionally define GL_FRAGMENT_PRECISION_HIGH in ES2 shaders.</li>
+  <li>glsl: Fix constant expression handling for &lt, &gt;, &lt=, &gt;= on vectors.</li>
+  <li>glsl: Use do_common_optimization in the standalone compiler.</li>
+  <li>glsl: Don't inline function prototypes.</li>
+  <li>glsl: Add a virtual as_discard() method.</li>
+  <li>glsl: Remove "discard" support from lower_jumps.</li>
+  <li>glsl: Refactor get_num_operands.</li>
+  <li>glcpp: Don't emit SPACE tokens in conditional_tokens production.</li>
+  <li>glsl: Clean up code by adding a new is_break() function.</li>
+  <li>glsl: Consider the "else" branch when looking for loop breaks.</li>
+  <li>Remove OES_compressed_paletted_texture from the ES2 extension list.</li>
+  <li>glsl/builtins: Compute the correct value for smoothstep(vec, vec, vec).</li>
+  <li>Fix build on systems where "python" is python 3.</li>
+  <li>i965: Internally enable GL_NV_blend_square on ES2.</li>
+</ul></p>
+
+<p>Kristian Høgsberg (1):
+<ul>
+  <li>i965: Don't write mrf assignment for pointsize output</li>
+</ul></p>
+
+<p>Luca Barbieri (1):
+<ul>
+  <li>glsl: Unroll loops with conditional breaks anywhere (not just the end)</li>
+</ul></p>
+
+<p>Marek Olšák (17):
+<ul>
+  <li>r300g: fix microtiling for 16-bits-per-channel formats</li>
+  <li>r300g: fix texture border for 16-bits-per-channel formats</li>
+  <li>r300g: add a default channel ordering of texture border for unhandled formats</li>
+  <li>r300g: fix texture border color for all texture formats</li>
+  <li>r300g: fix rendering with no vertex elements</li>
+  <li>r300/compiler: fix rc_rewrite_depth_out for it to work with any instruction</li>
+  <li>r300g: fix texture border color once again</li>
+  <li>r300g: fix texture swizzling with compressed textures on r400-r500</li>
+  <li>r300g: disable ARB_texture_swizzle if S3TC is enabled on r3xx-only</li>
+  <li>mesa, st/mesa: fix gl_FragCoord with FBOs in Gallium</li>
+  <li>st/mesa: initialize key in st_vp_varient</li>
+  <li>r300/compiler: fix swizzle lowering with a presubtract source operand</li>
+  <li>r300g: fix rendering with a vertex attrib having a zero stride</li>
+  <li>ir_to_mesa: Add support for conditional discards.</li>
+  <li>r300g: finally fix the texture corruption on r3xx-r4xx</li>
+  <li>mesa: fix texel store functions for some float formats</li>
+  <li>r300/compiler: disable the rename_regs pass for loops</li>
+</ul></p>
+
+<p>Mario Kleiner (1):
+<ul>
+  <li>mesa/r300classic: Fix dri2Invalidate/radeon_prepare_render for page flipping.</li>
+</ul></p>
+
+<p>Peter Clifton (1):
+<ul>
+  <li>intel: Fix emit_linear_blit to use DWORD aligned width blits</li>
+</ul></p>
+
+<p>Robert Hooker (2):
+<ul>
+  <li>intel: Add a new B43 pci id.</li>
+  <li>egl_dri2: Add missing intel chip ids.</li>
+</ul></p>
+
+<p>Roland Scheidegger (1):
+<ul>
+  <li>r200: fix r200 large points</li>
+</ul></p>
+
+<p>Thomas Hellstrom (17):
+<ul>
+  <li>st/xorg: Don't try to use option values before processing options</li>
+  <li>xorg/vmwgfx: Make vmwarectrl work also on 64-bit servers</li>
+  <li>st/xorg: Add a customizer option to get rid of annoying cursor update flicker</li>
+  <li>xorg/vmwgfx: Don't hide HW cursors when updating them</li>
+  <li>st/xorg: Don't try to remove invalid fbs</li>
+  <li>st/xorg: Fix typo</li>
+  <li>st/xorg, xorg/vmwgfx: Be a bit more frendly towards cross-compiling environments</li>
+  <li>st/xorg: Fix compilation errors for Xservers compiled without Composite</li>
+  <li>st/xorg: Don't use deprecated x*alloc / xfree functions</li>
+  <li>xorg/vmwgfx: Don't use deprecated x*alloc / xfree functions</li>
+  <li>st/xorg: Fix compilation for Xservers &gt;= 1.10</li>
+  <li>mesa: Make sure we have the talloc cflags when using the talloc headers</li>
+  <li>egl: Add an include for size_t</li>
+  <li>mesa: Add talloc includes for gles</li>
+  <li>st/egl: Fix build for include files in nonstandard places</li>
+  <li>svga/drm: Optionally resolve calls to powf during link-time</li>
+  <li>gallium/targets: Trivial crosscompiling fix</li>
+</ul></p>
+
+<p>Tom Stellard (7):
+<ul>
+  <li>r300/compiler: Make sure presubtract sources use supported swizzles</li>
+  <li>r300/compiler: Fix register allocator's handling of loops</li>
+  <li>r300/compiler: Fix instruction scheduling within IF blocks</li>
+  <li>r300/compiler: Use zero as the register index for unused sources</li>
+  <li>r300/compiler: Ignore alpha dest register when replicating the result</li>
+  <li>r300/compiler: Use correct swizzles for all presubtract sources</li>
+  <li>r300/compiler: Don't allow presubtract sources to be remapped twice</li>
+</ul></p>
+
+<p>Vinson Lee (1):
+<ul>
+  <li>glsl: Fix 'control reaches end of non-void function' warning.</li>
+</ul></p>
+
+<p>richard (1):
+<ul>
+  <li>r600c : inline vertex format is not updated in an app, switch to use vfetch constants. For the 7.9 and 7.10 branches as well.</li>
+</ul></p>
+
+</body>
+</html>

docs/relnotes-7.9.html

@@ -8,7 +8,7 @@
 
 <body bgcolor="#eeeeee">
 
-<H1>Mesa 7.9 Release Notes / date TBD</H1>
+<H1>Mesa 7.9 Release Notes / October 4, 2010</H1>
 
 <p>
 Mesa 7.9 is a new development release.
@@ -28,12 +28,12 @@ for DRI hardware acceleration.
 
 <h2>MD5 checksums</h2>
 <pre>
-f1f01a7baec255f13e9468fb4b05922a  MesaLib-7.9-rc1.tar.gz
-7ffbda3b7056c60b8f87e3082d853af1  MesaLib-7.9-rc1.tar.bz2
-9d4650df4e5b530178d6fde840f76664  MesaLib-7.9-rc1.zip
-a81c2e7a0c7832e67c768d6f209f2c8f  MesaGLUT-7.9-rc1.tar.gz
-b4c1c2f0b47a07be10fa2dd42e6a63d7  MesaGLUT-7.9-rc1.tar.bz2
-c9dd7419a19bcb24a1fe556ec2e78451  MesaGLUT-7.9-rc1.zip
+ed65ab425b25895c7f473d0a5e6e64f8  MesaLib-7.9.tar.gz
+82c740c49d572baa6da2b1a1eee90bca  MesaLib-7.9.tar.bz2
+cd2b6ecec759b0457475e94bbb38fedb  MesaLib-7.9.zip
+7b54af9fb9b1f6a1a65db2520f50848f  MesaGLUT-7.9.tar.gz
+20d07419d1929f833fdb36bced290ad5  MesaGLUT-7.9.tar.bz2
+62a7edecd7c92675cd6029b05217eb0a  MesaGLUT-7.9.zip
 </pre>
 
 
@@ -42,16 +42,85 @@ c9dd7419a19bcb24a1fe556ec2e78451  MesaGLUT-7.9-rc1.zip
 <li>New, improved GLSL compiler written by Intel.
     See the <a href="shading.html"> Shading Language</a> page for
     more information.
-<li>GL_EXT_timer_query extension (i965 driver only)
+<li>New, very experimental Gallium driver for R600-R700 Radeons.
+<li>Support for AMD Evergreen-based Radeons (HD 5xxx)
+<li>GL_EXT_timer_query extension (i965 driver and softpipe only)
 <li>GL_EXT_framebuffer_multisample extension (intel drivers, MAX_SAMPLES = 1)
 <li>GL_ARB_texture_swizzle extension (alias of GL_EXT_texture_swizzle)
-<li>GL_ARB_draw_elements_base_vertex, GL_ARB_fragment_program_shadow
-    and GL_EXT_draw_buffers2 in Gallium drivers
+<li>GL_ARB_draw_elements_base_vertex, GL_ARB_fragment_program_shadow,
+    GL_ARB_window_pos, GL_EXT_gpu_program_parameters,
+    GL_ATI_texture_env_combine3, GL_MESA_pack_invert, and GL_OES_EGL_image
+    extensions in Gallium drivers
+<li>GL_ARB_depth_clamp and GL_NV_depth_clamp extensions (in nv50 and r600
+    Gallium drivers)
+<li>GL_ARB_half_float_vertex extension (in nvfx, r300, r600, softpipe,
+    and llvmpipe Gallium drivers)
+<li>GL_EXT_draw_buffers2 (in nv50, r600, softpipe, and llvmpipe Gallium
+    drivers)
+<li>GL_EXT_texture_swizzle (in nvfx, r300, r600, softpipe, and llvmpipe
+    Gallium drivers)
+<li>GL_ATI_texture_mirror_once (in nvfx, nv50, r300, r600, softpipe, and
+    llvmpipe Gallium drivers)
+<li>GL_NV_conditional_render (in r300 Gallium driver)
+<li>Initial "signs of life" support for Sandybridge hardware in i965 DRI
+    driver.
 </ul>
 
 
 <h2>Bug fixes</h2>
+<p>This list is likely incomplete.</p>
 <ul>
+<li>Massive improvements to the Gallium driver for R300-R500 Radeons; this
+    driver is now considered stable for use as a DRI (OpenGL) driver.
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=10908">Bug 10908</a> - GLSL: gl_FogParamaters gl_Fog built-in uniform not functioning</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=13753">Bug 13753</a> - Numerous bugs in GLSL uniform handling</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=16854">Bug 16854</a> - GLSL function call at global scope causes SEGV</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=16856">Bug 16856</a> - GLSL indexing of unsized array results in assertion failure</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=18659">Bug 18659</a> - Crash in shader/slang/slang_codegen.c _slang_gen_function_call_name()</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=19089">Bug 19089</a> - [GLSL] glsl1/shadow2D() cases fail</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=22622">Bug 22622</a> - [GM965 GLSL] noise*() cause GPU lockup</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=23743">Bug 23743</a> - For loop from 0 to 0 not optimized out</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=24553">Bug 24553</a> - shader compilation times explode when using more () pairs</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25664">Bug 25664</a> - [GLSL] re-declaring an empty array fails to compile</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25769">Bug 25769</a> - [GLSL] "float" can be implicitly converted to "int"</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25808">Bug 25808</a> - [GLSL] const variable is modified successfully</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25826">Bug 25826</a> - [GLSL] declaring an unsized array then re-declaring with a size fails</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25827">Bug 25827</a> - [GLSL] vector constructor accepts too many arguments successfully</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25829">Bug 25829</a> - [GLSL] allowing non-void function without returning value</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25830">Bug 25830</a> - [GLSL] allowing non-constant-expression as const declaration initializer</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25877">Bug 25877</a> - [GLSL 1.10] implicit conversion from "int" to "float" should not be allowed</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25878">Bug 25878</a> - [GLSL] sampler is converted to int successfully</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25994">Bug 25994</a> - [GM45][GLSL] 'return' statement in vertex shader unsupported</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=25999">Bug 25999</a> - [GLSL] embedded structure constructor fails to compile</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=26000">Bug 26000</a> - [GLSL] allowing different parameter qualifier between the function definition and declaration</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=26001">Bug 26001</a> - [GLSL 1.10] constructing matrix from matrix succeeds</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=26224">Bug 26224</a> - [GLSL] Cannot get location of a uniform struct member</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=26990">Bug 26990</a> - [GLSL] variable declaration in "while" fails to compile</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27004">Bug 27004</a> - [GLSL] allowing macro redefinition</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27060">Bug 27060</a> - [965] piglit glsl-fs-raytrace failure due to lack of function calls.</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27216">Bug 27216</a> - Assignment with a function call in an if statement causes an assertion failure</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27261">Bug 27261</a> - GLSL Compiler fails on the following vertex shader</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27265">Bug 27265</a> - GLSL Compiler doesnt link the attached vertex shader</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27388">Bug 27388</a> - [i965] piglit glsl-vs-arrays failure</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27403">Bug 27403</a> - GLSL struct causing "Invalid src register file ..." error</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27914">Bug 27914</a> - GLSL compiler uses MUL+ADD where it could use MAD</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28055">Bug 28055</a> - glsl-texcoord-array fails GLSL compilation</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28374">Bug 28374</a> - SIGSEGV shader/slang/slang_typeinfo.c:534</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28748">Bug 28748</a> - [i965] uninlined function calls support</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28833">Bug 28833</a> - piglit/shaders/glsl-texcoord-array fail</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28834">Bug 28834</a> - Add support for system fpclassify to GL_OES_query_matrix function for OpenBSD / NetBSD</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28837">Bug 28837</a> - varying vec4 index support</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28845">Bug 28845</a> - The GLU tesselator code has some warnings</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28889">Bug 28889</a> - [regression] wine game crash</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28894">Bug 28894</a> - slang build fails if absolute path contains spaces</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28913">Bug 28913</a> - [GLSL] allowing two version statements</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28931">Bug 28931</a> - Floating Point Exception in Warzone2100 Trunk version</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28966">Bug 28966</a> - [r300g] Dynamic branching 3 demo does not run</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28967">Bug 28967</a> - slang/slang_emit.c:350: storage_to_src_reg: Assertion `index &gt;= 0' failed.</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29013">Bug 29013</a> - [r300g] translate_rgb_op: unknown opcode ILLEGAL OPCODE</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29020">Bug 29020</a> - [r300g] Wine d3d9 tests hardlock</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=29910">Bug 29910</a> - Mesa advertises bogus GL_ARB_shading_language_120</li>
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=30196">Bug 30196</a> - [GLSL] gl_TextureMatrix{Inverse,Transpose,InverseTranspose} unsupported</li>
 </ul>
 
 

docs/relnotes.html

@@ -14,6 +14,7 @@ The release notes summarize what's new or changed in each Mesa release.
 
 <UL>
 <LI><A HREF="relnotes-7.9.html">7.9 release notes</A>
+<LI><A HREF="relnotes-7.8.3.html">7.8.3 release notes</A>
 <LI><A HREF="relnotes-7.8.2.html">7.8.2 release notes</A>
 <LI><A HREF="relnotes-7.8.1.html">7.8.1 release notes</A>
 <LI><A HREF="relnotes-7.8.html">7.8 release notes</A>

src/egl/drivers/dri2/egl_dri2.c

@@ -741,7 +741,7 @@ dri2_create_screen(_EGLDisplay *disp)
    if (dri2_dpy->dri2->base.version >= 2)
       api_mask = dri2_dpy->dri2->getAPIMask(dri2_dpy->dri_screen);
    else
-      api_mask = __DRI_API_OPENGL;
+      api_mask = 1 << __DRI_API_OPENGL;
 
    disp->ClientAPIsMask = 0;
    if (api_mask & (1 <<__DRI_API_OPENGL))
@@ -888,10 +888,20 @@ const int i915_chip_ids[] = {
    0x29b2, /* PCI_CHIP_Q35_G */
    0x29c2, /* PCI_CHIP_G33_G */
    0x29d2, /* PCI_CHIP_Q33_G */
+   0xa001, /* PCI_CHIP_IGD_G */
    0xa011, /* Pineview */
 };
 
 const int i965_chip_ids[] = {
+   0x0042, /* PCI_CHIP_ILD_G */
+   0x0046, /* PCI_CHIP_ILM_G */
+   0x0102, /* PCI_CHIP_SANDYBRIDGE_GT1 */
+   0x0106, /* PCI_CHIP_SANDYBRIDGE_M_GT1 */
+   0x010a, /* PCI_CHIP_SANDYBRIDGE_S */
+   0x0112, /* PCI_CHIP_SANDYBRIDGE_GT2 */
+   0x0116, /* PCI_CHIP_SANDYBRIDGE_M_GT2 */
+   0x0122, /* PCI_CHIP_SANDYBRIDGE_GT2_PLUS */
+   0x0126, /* PCI_CHIP_SANDYBRIDGE_M_GT2_PLUS */
    0x29a2, /* PCI_CHIP_I965_G */
    0x2992, /* PCI_CHIP_I965_Q */
    0x2982, /* PCI_CHIP_I965_G_1 */
@@ -903,6 +913,8 @@ const int i965_chip_ids[] = {
    0x2e12, /* PCI_CHIP_Q45_G */
    0x2e22, /* PCI_CHIP_G45_G */
    0x2e32, /* PCI_CHIP_G41_G */
+   0x2e42, /* PCI_CHIP_B43_G */
+   0x2e92, /* PCI_CHIP_B43_G1 */
 };
 
 const struct dri2_driver_map driver_map[] = {

src/egl/drivers/glx/egl_glx.c

@@ -142,7 +142,6 @@ static const struct {
    { GLX_STENCIL_SIZE,			EGL_STENCIL_SIZE },
    { GLX_SAMPLE_BUFFERS,		EGL_SAMPLE_BUFFERS },
    { GLX_SAMPLES,			EGL_SAMPLES },
-   { GLX_RENDER_TYPE,			EGL_RENDERABLE_TYPE },
    { GLX_X_RENDERABLE,			EGL_NATIVE_RENDERABLE },
    { GLX_X_VISUAL_TYPE,			EGL_NATIVE_VISUAL_TYPE },
    { GLX_CONFIG_CAVEAT,			EGL_CONFIG_CAVEAT },
@@ -218,6 +217,16 @@ convert_fbconfig(Display *dpy, GLXFBConfig fbconfig,
 
    _eglSetConfigKey(&GLX_conf->Base, EGL_SURFACE_TYPE, surface_type);
 
+   _eglSetConfigKey(&GLX_conf->Base, EGL_RENDERABLE_TYPE, EGL_OPENGL_BIT);
+
+   glXGetFBConfigAttrib(dpy, fbconfig, GLX_TRANSPARENT_TYPE, &val);
+   if (val == GLX_NONE) {
+      _eglSetConfigKey(&GLX_conf->Base, EGL_TRANSPARENT_TYPE, EGL_NONE);
+      _eglSetConfigKey(&GLX_conf->Base, EGL_TRANSPARENT_RED_VALUE, 0);
+      _eglSetConfigKey(&GLX_conf->Base, EGL_TRANSPARENT_GREEN_VALUE, 0);
+      _eglSetConfigKey(&GLX_conf->Base, EGL_TRANSPARENT_BLUE_VALUE, 0);
+   }
+
    return EGL_TRUE;
 }
 
@@ -294,6 +303,7 @@ convert_visual(Display *dpy, XVisualInfo *vinfo,
       surface_type |= EGL_PIXMAP_BIT;
 
    _eglSetConfigKey(&GLX_conf->Base, EGL_SURFACE_TYPE, surface_type);
+   _eglSetConfigKey(&GLX_conf->Base, EGL_RENDERABLE_TYPE, EGL_OPENGL_BIT);
 
    _eglSetConfigKey(&GLX_conf->Base, EGL_NATIVE_RENDERABLE, EGL_TRUE);
 

src/egl/main/eglapi.c

@@ -402,10 +402,15 @@ eglCreateContext(EGLDisplay dpy, EGLConfig config, EGLContext share_list,
    _EGLContext *context;
    EGLContext ret;
 
-   if (config)
-      _EGL_CHECK_CONFIG(disp, conf, EGL_NO_CONTEXT, drv);
-   else
-      _EGL_CHECK_DISPLAY(disp, EGL_NO_CONTEXT, drv);
+   _EGL_CHECK_DISPLAY(disp, EGL_NO_CONTEXT, drv);
+
+   if (!config) {
+      /* config may be NULL if surfaceless */
+      if (!disp->Extensions.KHR_surfaceless_gles1 &&
+          !disp->Extensions.KHR_surfaceless_gles2 &&
+          !disp->Extensions.KHR_surfaceless_opengl)
+         RETURN_EGL_ERROR(disp, EGL_BAD_CONFIG, EGL_NO_CONTEXT);
+   }
 
    if (!share && share_list != EGL_NO_CONTEXT)
       RETURN_EGL_ERROR(disp, EGL_BAD_CONTEXT, EGL_NO_CONTEXT);
@@ -459,9 +464,19 @@ eglMakeCurrent(EGLDisplay dpy, EGLSurface draw, EGLSurface read,
 
    if (!context && ctx != EGL_NO_CONTEXT)
       RETURN_EGL_ERROR(disp, EGL_BAD_CONTEXT, EGL_FALSE);
-   if ((!draw_surf && draw != EGL_NO_SURFACE) ||
-       (!read_surf && read != EGL_NO_SURFACE))
-      RETURN_EGL_ERROR(disp, EGL_BAD_SURFACE, EGL_FALSE);
+   if (!draw_surf || !read_surf) {
+      /* surfaces may be NULL if surfaceless */
+      if (!disp->Extensions.KHR_surfaceless_gles1 &&
+          !disp->Extensions.KHR_surfaceless_gles2 &&
+          !disp->Extensions.KHR_surfaceless_opengl)
+         RETURN_EGL_ERROR(disp, EGL_BAD_SURFACE, EGL_FALSE);
+
+      if ((!draw_surf && draw != EGL_NO_SURFACE) ||
+          (!read_surf && read != EGL_NO_SURFACE))
+         RETURN_EGL_ERROR(disp, EGL_BAD_SURFACE, EGL_FALSE);
+      if (draw_surf || read_surf)
+         RETURN_EGL_ERROR(disp, EGL_BAD_MATCH, EGL_FALSE);
+   }
 
    ret = drv->API.MakeCurrent(drv, disp, draw_surf, read_surf, context);
 
@@ -1276,6 +1291,8 @@ eglCreateImageKHR(EGLDisplay dpy, EGLContext ctx, EGLenum target,
    EGLImageKHR ret;
 
    _EGL_CHECK_DISPLAY(disp, EGL_NO_IMAGE_KHR, drv);
+   if (!disp->Extensions.KHR_image_base)
+      RETURN_EGL_EVAL(disp, EGL_NO_IMAGE_KHR);
    if (!context && ctx != EGL_NO_CONTEXT)
       RETURN_EGL_ERROR(disp, EGL_BAD_CONTEXT, EGL_NO_IMAGE_KHR);
 
@@ -1296,6 +1313,8 @@ eglDestroyImageKHR(EGLDisplay dpy, EGLImageKHR image)
    EGLBoolean ret;
 
    _EGL_CHECK_DISPLAY(disp, EGL_FALSE, drv);
+   if (!disp->Extensions.KHR_image_base)
+      RETURN_EGL_EVAL(disp, EGL_FALSE);
    if (!img)
       RETURN_EGL_ERROR(disp, EGL_BAD_PARAMETER, EGL_FALSE);
 
@@ -1321,6 +1340,8 @@ eglCreateSyncKHR(EGLDisplay dpy, EGLenum type, const EGLint *attrib_list)
    EGLSyncKHR ret;
 
    _EGL_CHECK_DISPLAY(disp, EGL_NO_SYNC_KHR, drv);
+   if (!disp->Extensions.KHR_reusable_sync)
+      RETURN_EGL_EVAL(disp, EGL_NO_SYNC_KHR);
 
    sync = drv->API.CreateSyncKHR(drv, disp, type, attrib_list);
    ret = (sync) ? _eglLinkSync(sync, disp) : EGL_NO_SYNC_KHR;
@@ -1338,6 +1359,8 @@ eglDestroySyncKHR(EGLDisplay dpy, EGLSyncKHR sync)
    EGLBoolean ret;
 
    _EGL_CHECK_SYNC(disp, s, EGL_FALSE, drv);
+   assert(disp->Extensions.KHR_reusable_sync);
+
    _eglUnlinkSync(s);
    ret = drv->API.DestroySyncKHR(drv, disp, s);
 
@@ -1354,6 +1377,7 @@ eglClientWaitSyncKHR(EGLDisplay dpy, EGLSyncKHR sync, EGLint flags, EGLTimeKHR t
    EGLint ret;
 
    _EGL_CHECK_SYNC(disp, s, EGL_FALSE, drv);
+   assert(disp->Extensions.KHR_reusable_sync);
    ret = drv->API.ClientWaitSyncKHR(drv, disp, s, flags, timeout);
 
    RETURN_EGL_EVAL(disp, ret);
@@ -1369,6 +1393,7 @@ eglSignalSyncKHR(EGLDisplay dpy, EGLSyncKHR sync, EGLenum mode)
    EGLBoolean ret;
 
    _EGL_CHECK_SYNC(disp, s, EGL_FALSE, drv);
+   assert(disp->Extensions.KHR_reusable_sync);
    ret = drv->API.SignalSyncKHR(drv, disp, s, mode);
 
    RETURN_EGL_EVAL(disp, ret);
@@ -1384,6 +1409,7 @@ eglGetSyncAttribKHR(EGLDisplay dpy, EGLSyncKHR sync, EGLint attribute, EGLint *v
    EGLBoolean ret;
 
    _EGL_CHECK_SYNC(disp, s, EGL_FALSE, drv);
+   assert(disp->Extensions.KHR_reusable_sync);
    ret = drv->API.GetSyncAttribKHR(drv, disp, s, attribute, value);
 
    RETURN_EGL_EVAL(disp, ret);
@@ -1407,14 +1433,14 @@ eglSwapBuffersRegionNOK(EGLDisplay dpy, EGLSurface surface,
 
    _EGL_CHECK_SURFACE(disp, surf, EGL_FALSE, drv);
 
+   if (!disp->Extensions.NOK_swap_region)
+      RETURN_EGL_EVAL(disp, EGL_FALSE);
+
    /* surface must be bound to current context in EGL 1.4 */
    if (!ctx || !_eglIsContextLinked(ctx) || surf != ctx->DrawSurface)
       RETURN_EGL_ERROR(disp, EGL_BAD_SURFACE, EGL_FALSE);
 
-   if (drv->API.SwapBuffersRegionNOK)
-      ret = drv->API.SwapBuffersRegionNOK(drv, disp, surf, numRects, rects);
-   else
-      ret = drv->API.SwapBuffers(drv, disp, surf);
+   ret = drv->API.SwapBuffersRegionNOK(drv, disp, surf, numRects, rects);
 
    RETURN_EGL_EVAL(disp, ret);
 }
@@ -1433,6 +1459,8 @@ eglCreateDRMImageMESA(EGLDisplay dpy, const EGLint *attr_list)
    EGLImageKHR ret;
 
    _EGL_CHECK_DISPLAY(disp, EGL_NO_IMAGE_KHR, drv);
+   if (!disp->Extensions.MESA_drm_image)
+      RETURN_EGL_EVAL(disp, EGL_NO_IMAGE_KHR);
 
    img = drv->API.CreateDRMImageMESA(drv, disp, attr_list);
    ret = (img) ? _eglLinkImage(img, disp) : EGL_NO_IMAGE_KHR;
@@ -1450,6 +1478,8 @@ eglExportDRMImageMESA(EGLDisplay dpy, EGLImageKHR image,
    EGLBoolean ret;
 
    _EGL_CHECK_DISPLAY(disp, EGL_FALSE, drv);
+   assert(disp->Extensions.MESA_drm_image);
+
    if (!img)
       RETURN_EGL_ERROR(disp, EGL_BAD_PARAMETER, EGL_FALSE);
 

src/egl/main/eglcontext.c

@@ -272,10 +272,6 @@ _eglCheckMakeCurrent(_EGLContext *ctx, _EGLSurface *draw, _EGLSurface *read)
    if (!surfaceless && (draw == NULL || read == NULL))
       return _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
 
-   /* context stealing from another thread is not allowed */
-   if (ctx->Binding && ctx->Binding != t)
-      return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
-
    /*
     * The spec says
     *
@@ -283,16 +279,23 @@ _eglCheckMakeCurrent(_EGLContext *ctx, _EGLSurface *draw, _EGLSurface *read)
     * bound to contexts in another thread, an EGL_BAD_ACCESS error is
     * generated."
     *
-    * But it also says
+    * and
     *
     * "at most one context may be bound to a particular surface at a given
     * time"
-    *
-    * The latter is more restrictive so we can check only the latter case.
     */
-   if ((draw && draw->CurrentContext && draw->CurrentContext != ctx) ||
-       (read && read->CurrentContext && read->CurrentContext != ctx))
+   if (ctx->Binding && ctx->Binding != t)
       return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
+   if (draw && draw->CurrentContext && draw->CurrentContext != ctx) {
+      if (draw->CurrentContext->Binding != t ||
+          draw->CurrentContext->ClientAPI != ctx->ClientAPI)
+         return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
+   }
+   if (read && read->CurrentContext && read->CurrentContext != ctx) {
+      if (read->CurrentContext->Binding != t ||
+          read->CurrentContext->ClientAPI != ctx->ClientAPI)
+         return _eglError(EGL_BAD_ACCESS, "eglMakeCurrent");
+   }
 
    /* simply require the configs to be equal */
    if ((draw && draw->Config != ctx->Config) ||

src/egl/main/egldriver.h

@@ -4,7 +4,7 @@
 
 #include "egltypedefs.h"
 #include "eglapi.h"
-
+#include <stddef.h>
 
 /**
  * Define an inline driver typecast function.

src/gallium/auxiliary/Makefile

@@ -206,16 +206,16 @@ include ../Makefile.template
 
 
 indices/u_indices_gen.c: indices/u_indices_gen.py
-	python $< > $@
+	$(PYTHON2) $< > $@
 
 indices/u_unfilled_gen.c: indices/u_unfilled_gen.py
-	python $< > $@
+	$(PYTHON2) $< > $@
 
 util/u_format_srgb.c: util/u_format_srgb.py
-	python $< > $@
+	$(PYTHON2) $< > $@
 
 util/u_format_table.c: util/u_format_table.py util/u_format_pack.py util/u_format_parse.py util/u_format.csv
-	python util/u_format_table.py util/u_format.csv > $@
+	$(PYTHON2) util/u_format_table.py util/u_format.csv > $@
 
 util/u_half.c: util/u_half.py
-	python util/u_half.py > $@
+	$(PYTHON2) util/u_half.py > $@

src/gallium/auxiliary/tgsi/tgsi_exec.c

@@ -605,8 +605,10 @@ tgsi_check_soa_dependencies(const struct tgsi_full_instruction *inst)
    for (i = 0; i < inst->Instruction.NumSrcRegs; i++) {
       if ((inst->Src[i].Register.File ==
            inst->Dst[0].Register.File) &&
-          (inst->Src[i].Register.Index ==
-           inst->Dst[0].Register.Index)) {
+          ((inst->Src[i].Register.Index ==
+            inst->Dst[0].Register.Index) ||
+	   inst->Src[i].Register.Indirect ||
+	   inst->Dst[0].Register.Indirect)) {
          /* loop over dest channels */
          uint channelsWritten = 0x0;
          FOR_EACH_ENABLED_CHANNEL(*inst, chan) {

src/gallium/auxiliary/tgsi/tgsi_sse2.c

@@ -2830,31 +2830,52 @@ static void soa_to_aos( struct x86_function *func,
  * Check if the instructions dst register is the same as any src
  * register and warn if there's a posible SOA dependency.
  */
-static void
+static boolean
 check_soa_dependencies(const struct tgsi_full_instruction *inst)
 {
-   switch (inst->Instruction.Opcode) {
+   uint opcode = inst->Instruction.Opcode;
+
+   /* XXX: we only handle src/dst aliasing in a few opcodes currently.
+    * Need to use an additional temporay to hold the result in the
+    * cases where the code is too opaque to fix.
+    */
+
+   switch (opcode) {
    case TGSI_OPCODE_ADD:
    case TGSI_OPCODE_MOV:
    case TGSI_OPCODE_MUL:
+   case TGSI_OPCODE_RCP:
+   case TGSI_OPCODE_RSQ:
+   case TGSI_OPCODE_EXP:
+   case TGSI_OPCODE_LOG:
+   case TGSI_OPCODE_DP3:
+   case TGSI_OPCODE_DP4:
+   case TGSI_OPCODE_DP2A:
+   case TGSI_OPCODE_EX2:
+   case TGSI_OPCODE_LG2:
+   case TGSI_OPCODE_POW:
    case TGSI_OPCODE_XPD:
+   case TGSI_OPCODE_DPH:
+   case TGSI_OPCODE_COS:
+   case TGSI_OPCODE_SIN:
+   case TGSI_OPCODE_TEX:
+   case TGSI_OPCODE_TXB:
+   case TGSI_OPCODE_TXP:
+   case TGSI_OPCODE_NRM:
+   case TGSI_OPCODE_NRM4:
+   case TGSI_OPCODE_DP2:
       /* OK - these opcodes correctly handle SOA dependencies */
-      break;
+      return TRUE;
    default:
-      if (tgsi_check_soa_dependencies(inst)) {
-         uint opcode = inst->Instruction.Opcode;
+      if (!tgsi_check_soa_dependencies(inst))
+         return TRUE;
 
-         /* XXX: we only handle src/dst aliasing in a few opcodes
-          * currently.  Need to use an additional temporay to hold
-          * the result in the cases where the code is too opaque to
-          * fix.
-          */
-         if (opcode != TGSI_OPCODE_MOV) {
-            debug_printf("Warning: src/dst aliasing in instruction"
-                         " is not handled:\n");
-            tgsi_dump_instruction(inst, 1);
-         }
-      }
+      debug_printf("Warning: src/dst aliasing in instruction"
+                   " is not handled:\n");
+      debug_printf("Warning: ");
+      tgsi_dump_instruction(inst, 1);
+
+      return FALSE;
    }
 }
 
@@ -2954,7 +2975,8 @@ tgsi_emit_sse2(
                          tgsi_get_processor_name(proc));
 	 }
 
-         check_soa_dependencies(&parse.FullToken.FullInstruction);
+         if (ok)
+            ok = check_soa_dependencies(&parse.FullToken.FullInstruction);
          break;
 
       case TGSI_TOKEN_TYPE_IMMEDIATE:

src/gallium/auxiliary/util/u_atomic.h

@@ -29,6 +29,8 @@
 #define PIPE_ATOMIC_ASM_MSVC_X86                
 #elif (defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86))
 #define PIPE_ATOMIC_ASM_GCC_X86
+#elif (defined(PIPE_CC_GCC) && defined(PIPE_ARCH_X86_64))
+#define PIPE_ATOMIC_ASM_GCC_X86_64
 #elif defined(PIPE_CC_GCC) && (PIPE_CC_GCC_VERSION >= 401)
 #define PIPE_ATOMIC_GCC_INTRINSIC
 #else
@@ -36,6 +38,51 @@
 #endif
 
 
+#if defined(PIPE_ATOMIC_ASM_GCC_X86_64)
+#define PIPE_ATOMIC "GCC x86_64 assembly"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define p_atomic_set(_v, _i) (*(_v) = (_i))
+#define p_atomic_read(_v) (*(_v))
+
+static INLINE boolean
+p_atomic_dec_zero(int32_t *v)
+{
+   unsigned char c;
+
+   __asm__ __volatile__("lock; decl %0; sete %1":"+m"(*v), "=qm"(c)
+			::"memory");
+
+   return c != 0;
+}
+
+static INLINE void
+p_atomic_inc(int32_t *v)
+{
+   __asm__ __volatile__("lock; incl %0":"+m"(*v));
+}
+
+static INLINE void
+p_atomic_dec(int32_t *v)
+{
+   __asm__ __volatile__("lock; decl %0":"+m"(*v));
+}
+
+static INLINE int32_t
+p_atomic_cmpxchg(int32_t *v, int32_t old, int32_t _new)
+{
+   return __sync_val_compare_and_swap(v, old, _new);
+}
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* PIPE_ATOMIC_ASM_GCC_X86_64 */
+
 
 #if defined(PIPE_ATOMIC_ASM_GCC_X86)
 

src/gallium/drivers/llvmpipe/Makefile

@@ -70,7 +70,7 @@ PROGS_DEPS := ../../auxiliary/libgallium.a
 include ../../Makefile.template
 
 lp_tile_soa.c: lp_tile_soa.py ../../auxiliary/util/u_format_parse.py ../../auxiliary/util/u_format_pack.py ../../auxiliary/util/u_format.csv
-	python lp_tile_soa.py ../../auxiliary/util/u_format.csv > $@
+	$(PYTHON2) $(PYTHON_FLAGS) lp_tile_soa.py ../../auxiliary/util/u_format.csv > $@
 
 LDFLAGS += $(LLVM_LDFLAGS)
 LIBS += -L../../auxiliary/ -lgallium libllvmpipe.a $(LLVM_LIBS) $(GL_LIB_DEPS)

src/gallium/drivers/llvmpipe/lp_bld_depth.c

@@ -262,7 +262,7 @@ lp_build_stencil_op_single(struct lp_build_context *bld,
       LLVMValueRef writemask = lp_build_const_int_vec(type, stencil->writemask);
       mask = LLVMBuildAnd(bld->builder, mask, writemask, "");
       /* res = (res & mask) | (stencilVals & ~mask) */
-      res = lp_build_select_bitwise(bld, writemask, res, stencilVals);
+      res = lp_build_select_bitwise(bld, mask, res, stencilVals);
    }
    else {
       /* res = mask ? res : stencilVals */

src/gallium/drivers/r300/r300_chipset.c

@@ -424,4 +424,5 @@ void r300_parse_chipset(struct r300_capabilities* caps)
     }
 
     caps->is_rv350 = caps->family >= CHIP_FAMILY_RV350;
+    caps->dxtc_swizzle = caps->is_r400 || caps->is_r500;
 }

src/gallium/drivers/r300/r300_chipset.h

@@ -79,6 +79,8 @@ struct r300_capabilities {
     boolean is_r500;
     /* Whether or not the second pixel pipe is accessed with the high bit */
     boolean high_second_pipe;
+    /* DXTC texture swizzling. */
+    boolean dxtc_swizzle;
 };
 
 /* Enumerations for legibility and telling which card we're running on. */

src/gallium/drivers/r300/r300_context.c

@@ -79,6 +79,9 @@ static void r300_release_referenced_objects(struct r300_context *r300)
                 NULL);
     }
 
+    /* The dummy VBO. */
+    pipe_resource_reference(&r300->dummy_vb, NULL);
+
     /* The SWTCL VBO. */
     pipe_resource_reference(&r300->vbo, NULL);
 
@@ -488,6 +491,7 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
         rtempl.target = PIPE_TEXTURE_2D;
         rtempl.format = PIPE_FORMAT_I8_UNORM;
         rtempl.bind = PIPE_BIND_SAMPLER_VIEW;
+        rtempl.usage = PIPE_USAGE_IMMUTABLE;
         rtempl.width0 = 1;
         rtempl.height0 = 1;
         rtempl.depth0 = 1;
@@ -501,6 +505,19 @@ struct pipe_context* r300_create_context(struct pipe_screen* screen,
         pipe_resource_reference(&tex, NULL);
     }
 
+    {
+        struct pipe_resource vb = {};
+        vb.target = PIPE_BUFFER;
+        vb.format = PIPE_FORMAT_R8_UNORM;
+        vb.bind = PIPE_BIND_VERTEX_BUFFER;
+        vb.usage = PIPE_USAGE_IMMUTABLE;
+        vb.width0 = sizeof(float) * 16;
+        vb.height0 = 1;
+        vb.depth0 = 1;
+
+        r300->dummy_vb = screen->resource_create(screen, &vb);
+    }
+
     return &r300->context;
 
  fail:

src/gallium/drivers/r300/r300_context.h

@@ -470,6 +470,10 @@ struct r300_context {
      * dummy texture there. */
     struct r300_sampler_view *texkill_sampler;
 
+    /* When no vertex buffer is set, this one is used instead to prevent
+     * hardlocks. */
+    struct pipe_resource *dummy_vb;
+
     /* The currently active query. */
     struct r300_query *query_current;
     /* The saved query for blitter operations. */

src/gallium/drivers/r300/r300_query.c

@@ -158,7 +158,7 @@ static void r300_render_condition(struct pipe_context *pipe,
                                   uint mode)
 {
     struct r300_context *r300 = r300_context(pipe);
-    uint64_t result;
+    uint64_t result = 0;
     boolean wait;
 
     if (query) {
@@ -167,9 +167,9 @@ static void r300_render_condition(struct pipe_context *pipe,
 
         if (!r300_get_query_result(pipe, query, wait, &result)) {
             r300->skip_rendering = FALSE;
+        } else {
+            r300->skip_rendering = result == 0;
         }
-
-        r300->skip_rendering = result == 0;
     } else {
         r300->skip_rendering = FALSE;
     }

src/gallium/drivers/r300/r300_reg.h

@@ -1520,11 +1520,11 @@ USE OR OTHER DEALINGS IN THE SOFTWARE.
 #	define R300_TX_TRI_PERF_3_8            (3<<15)
 #	define R300_ANISO_THRESHOLD_MASK       (7<<17)
 
+#       define R400_DXTC_SWIZZLE_ENABLE        (1<<21)
 #	define R500_MACRO_SWITCH               (1<<22)
 #       define R500_TX_MAX_ANISO(x)            ((x) << 23)
 #       define R500_TX_MAX_ANISO_MASK          (63 << 23)
 #       define R500_TX_ANISO_HIGH_QUALITY      (1 << 30)
-
 #	define R500_BORDER_FIX                 (1<<31)
 
 #define R300_TX_FORMAT0_0                   0x4480

src/gallium/drivers/r300/r300_render.c

@@ -676,9 +676,6 @@ static void r300_draw_vbo(struct pipe_context* pipe,
 {
     struct r300_context* r300 = r300_context(pipe);
 
-    if (!r300->velems->count || !r300->vertex_buffer_count)
-	    return;
-
     if (info->indexed && r300->index_buffer.buffer) {
         unsigned offset;
 

src/gallium/drivers/r300/r300_screen.c

@@ -114,8 +114,9 @@ static int r300_get_param(struct pipe_screen* pscreen, enum pipe_cap param)
         case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
         case PIPE_CAP_TEXTURE_MIRROR_REPEAT:
         case PIPE_CAP_BLEND_EQUATION_SEPARATE:
-        case PIPE_CAP_TEXTURE_SWIZZLE:
             return 1;
+        case PIPE_CAP_TEXTURE_SWIZZLE:
+            return util_format_s3tc_enabled ? r300screen->caps.dxtc_swizzle : 1;
 
         /* Unsupported features (boolean caps). */
         case PIPE_CAP_TIMER_QUERY:

src/gallium/drivers/r300/r300_state.c

@@ -1313,29 +1313,27 @@ static void r300_set_fragment_sampler_views(struct pipe_context* pipe,
     }
 
     for (i = 0; i < count; i++) {
-        if (&state->sampler_views[i]->base != views[i]) {
-            pipe_sampler_view_reference(
-                    (struct pipe_sampler_view**)&state->sampler_views[i],
-                    views[i]);
+        pipe_sampler_view_reference(
+                (struct pipe_sampler_view**)&state->sampler_views[i],
+                views[i]);
 
-            if (!views[i]) {
-                continue;
-            }
-
-            /* A new sampler view (= texture)... */
-            dirty_tex = TRUE;
-
-            /* Set the texrect factor in the fragment shader.
-             * Needed for RECT and NPOT fallback. */
-            texture = r300_texture(views[i]->texture);
-            if (texture->desc.is_npot) {
-                r300->fs_rc_constant_state.dirty = TRUE;
-            }
-
-            state->sampler_views[i]->texcache_region =
-                r300_assign_texture_cache_region(view_index, real_num_views);
-            view_index++;
+        if (!views[i]) {
+            continue;
         }
+
+        /* A new sampler view (= texture)... */
+        dirty_tex = TRUE;
+
+        /* Set the texrect factor in the fragment shader.
+         * Needed for RECT and NPOT fallback. */
+        texture = r300_texture(views[i]->texture);
+        if (texture->desc.is_npot) {
+            r300->fs_rc_constant_state.dirty = TRUE;
+        }
+
+        state->sampler_views[i]->texcache_region =
+                r300_assign_texture_cache_region(view_index, real_num_views);
+        view_index++;
     }
 
     for (i = count; i < tex_units; i++) {
@@ -1363,6 +1361,7 @@ r300_create_sampler_view(struct pipe_context *pipe,
     struct r300_sampler_view *view = CALLOC_STRUCT(r300_sampler_view);
     struct r300_texture *tex = r300_texture(texture);
     boolean is_r500 = r300_screen(pipe->screen)->caps.is_r500;
+    boolean dxtc_swizzle = r300_screen(pipe->screen)->caps.dxtc_swizzle;
 
     if (view) {
         view->base = *templ;
@@ -1379,7 +1378,8 @@ r300_create_sampler_view(struct pipe_context *pipe,
         view->format = tex->tx_format;
         view->format.format1 |= r300_translate_texformat(templ->format,
                                                          view->swizzle,
-                                                         is_r500);
+                                                         is_r500,
+                                                         dxtc_swizzle);
         if (is_r500) {
             view->format.format2 |= r500_tx_format_msb_bit(templ->format);
         }
@@ -1464,6 +1464,15 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
     struct pipe_vertex_buffer *vbo;
     unsigned i, max_index = (1 << 24) - 1;
     boolean any_user_buffer = FALSE;
+    struct pipe_vertex_buffer dummy_vb = {0};
+
+    /* There must be at least one vertex buffer set, otherwise it locks up. */
+    if (!count) {
+        dummy_vb.buffer = r300->dummy_vb;
+        dummy_vb.max_index = r300->dummy_vb->width0 / 4;
+        buffers = &dummy_vb;
+        count = 1;
+    }
 
     if (count == r300->vertex_buffer_count &&
         memcmp(r300->vertex_buffer, buffers,
@@ -1499,14 +1508,14 @@ static void r300_set_vertex_buffers(struct pipe_context* pipe,
                 any_user_buffer = TRUE;
             }
 
+            /* The stride of zero means we will be fetching only the first
+             * vertex, so don't care about max_index. */
+            if (!vbo->stride)
+                continue;
+
             if (vbo->max_index == ~0) {
-                /* if no VBO stride then only one vertex value so max index is 1 */
-                /* should think about converting to VS constants like svga does */
-                if (!vbo->stride)
-                    vbo->max_index = 1;
-                else
-                    vbo->max_index =
-                             (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride;
+                vbo->max_index =
+                        (vbo->buffer->width0 - vbo->buffer_offset) / vbo->stride;
             }
 
             max_index = MIN2(vbo->max_index, max_index);
@@ -1617,6 +1626,14 @@ static void* r300_create_vertex_elements_state(struct pipe_context* pipe,
     struct r300_vertex_element_state *velems;
     unsigned i;
     enum pipe_format *format;
+    struct pipe_vertex_element dummy_attrib = {0};
+
+    /* R300 Programmable Stream Control (PSC) doesn't support 0 vertex elements. */
+    if (!count) {
+        dummy_attrib.src_format = PIPE_FORMAT_R8G8B8A8_UNORM;
+        attribs = &dummy_attrib;
+        count = 1;
+    }
 
     assert(count <= PIPE_MAX_ATTRIBS);
     velems = CALLOC_STRUCT(r300_vertex_element_state);
@@ -1683,7 +1700,8 @@ static void* r300_create_vertex_elements_state(struct pipe_context* pipe,
              * swizzles are already set up.
              * Also compute the vertex size. */
             for (i = 0; i < count; i++) {
-                /* This is OK because we check for aligned strides too. */
+                /* This is OK because we check for aligned strides too
+                 * elsewhere. */
                 velems->hw_format_size[i] =
                     align(util_format_get_blocksize(velems->hw_format[i]), 4);
                 velems->vertex_size_dwords += velems->hw_format_size[i] / 4;

src/gallium/drivers/r300/r300_state_derived.c

@@ -25,6 +25,7 @@
 
 #include "util/u_math.h"
 #include "util/u_memory.h"
+#include "util/u_pack_color.h"
 
 #include "r300_context.h"
 #include "r300_fs.h"
@@ -568,55 +569,93 @@ static void r300_update_rs_block(struct r300_context *r300)
 }
 
 static uint32_t r300_get_border_color(enum pipe_format format,
-                                      const float border[4])
+                                      const float border[4],
+                                      boolean is_r500)
 {
     const struct util_format_description *desc;
-    float border_swizzled[4] = {
-        border[2],
-        border[1],
-        border[0],
-        border[3]
-    };
-    uint32_t r;
+    float border_swizzled[4] = {0};
+    unsigned i;
+    union util_color uc = {0};
 
     desc = util_format_description(format);
 
-    /* We don't use util_pack_format because it does not handle the formats
-     * we want, e.g. R4G4B4A4 is non-existent in Gallium. */
+    /* Do depth formats first. */
+    if (util_format_is_depth_or_stencil(format)) {
+        switch (format) {
+        case PIPE_FORMAT_Z16_UNORM:
+            return util_pack_z(PIPE_FORMAT_Z16_UNORM, border[0]);
+        case PIPE_FORMAT_X8Z24_UNORM:
+        case PIPE_FORMAT_S8_USCALED_Z24_UNORM:
+            if (is_r500) {
+                return util_pack_z(PIPE_FORMAT_X8Z24_UNORM, border[0]);
+            } else {
+                return util_pack_z(PIPE_FORMAT_Z16_UNORM, border[0]) << 16;
+            }
+        default:
+            assert(0);
+            return 0;
+        }
+    }
+
+    /* Apply inverse swizzle of the format. */
+    for (i = 0; i < 4; i++) {
+        switch (desc->swizzle[i]) {
+        case UTIL_FORMAT_SWIZZLE_X:
+            border_swizzled[2] = border[i];
+            break;
+        case UTIL_FORMAT_SWIZZLE_Y:
+            border_swizzled[1] = border[i];
+            break;
+        case UTIL_FORMAT_SWIZZLE_Z:
+            border_swizzled[0] = border[i];
+            break;
+        case UTIL_FORMAT_SWIZZLE_W:
+            border_swizzled[3] = border[i];
+            break;
+        }
+    }
+
+    /* Compressed formats. */
+    if (util_format_is_compressed(format)) {
+        util_pack_color(border_swizzled, PIPE_FORMAT_R8G8B8A8_UNORM, &uc);
+        return uc.ui;
+    }
+
     switch (desc->channel[0].size) {
         case 4:
-            r = ((float_to_ubyte(border_swizzled[0]) & 0xf0) >> 4) |
-                ((float_to_ubyte(border_swizzled[1]) & 0xf0) << 0) |
-                ((float_to_ubyte(border_swizzled[2]) & 0xf0) << 4) |
-                ((float_to_ubyte(border_swizzled[3]) & 0xf0) << 8);
+            util_pack_color(border_swizzled, PIPE_FORMAT_B4G4R4A4_UNORM, &uc);
             break;
 
         case 5:
             if (desc->channel[1].size == 5) {
-                r = ((float_to_ubyte(border_swizzled[0]) & 0xf8) >> 3) |
-                    ((float_to_ubyte(border_swizzled[1]) & 0xf8) << 2) |
-                    ((float_to_ubyte(border_swizzled[2]) & 0xf8) << 7) |
-                    ((float_to_ubyte(border_swizzled[3]) & 0x80) << 8);
+                util_pack_color(border_swizzled, PIPE_FORMAT_B5G5R5A1_UNORM, &uc);
             } else if (desc->channel[1].size == 6) {
-                r = ((float_to_ubyte(border_swizzled[0]) & 0xf8) >> 3) |
-                    ((float_to_ubyte(border_swizzled[1]) & 0xfc) << 3) |
-                    ((float_to_ubyte(border_swizzled[2]) & 0xf8) << 8);
+                util_pack_color(border_swizzled, PIPE_FORMAT_B5G6R5_UNORM, &uc);
             } else {
                 assert(0);
             }
             break;
 
         default:
-            /* I think the fat formats (16, 32) are specified
-             * as the 8-bit ones. I am not sure how compressed formats
-             * work here. */
-            r = ((float_to_ubyte(border_swizzled[0]) & 0xff) << 0) |
-                ((float_to_ubyte(border_swizzled[1]) & 0xff) << 8) |
-                ((float_to_ubyte(border_swizzled[2]) & 0xff) << 16) |
-                ((float_to_ubyte(border_swizzled[3]) & 0xff) << 24);
+        case 8:
+            util_pack_color(border_swizzled, PIPE_FORMAT_B8G8R8A8_UNORM, &uc);
+            break;
+
+        case 10:
+            util_pack_color(border_swizzled, PIPE_FORMAT_B10G10R10A2_UNORM, &uc);
+            break;
+
+        case 16:
+            if (desc->nr_channels <= 2) {
+                border_swizzled[0] = border_swizzled[2];
+                util_pack_color(border_swizzled, PIPE_FORMAT_R16G16_UNORM, &uc);
+            } else {
+                util_pack_color(border_swizzled, PIPE_FORMAT_B8G8R8A8_UNORM, &uc);
+            }
+            break;
     }
 
-    return r;
+    return uc.ui;
 }
 
 static void r300_merge_textures_and_samplers(struct r300_context* r300)
@@ -655,7 +694,8 @@ static void r300_merge_textures_and_samplers(struct r300_context* r300)
             /* Set the border color. */
             texstate->border_color =
                 r300_get_border_color(view->base.format,
-                                      sampler->state.border_color);
+                                      sampler->state.border_color,
+                                      r300->screen->caps.is_r500);
 
             /* determine min/max levels */
             max_level = MIN3(sampler->max_lod + view->base.first_level,
@@ -701,13 +741,18 @@ static void r300_merge_textures_and_samplers(struct r300_context* r300)
                 if (sampler->state.compare_mode == PIPE_TEX_COMPARE_NONE) {
                     texstate->format.format1 |=
                         r300_get_swizzle_combined(depth_swizzle,
-                                                  view->swizzle);
+                                                  view->swizzle, FALSE);
                 } else {
                     texstate->format.format1 |=
-                        r300_get_swizzle_combined(depth_swizzle, 0);
+                        r300_get_swizzle_combined(depth_swizzle, 0, FALSE);
                 }
             }
 
+            if (r300->screen->caps.dxtc_swizzle &&
+                util_format_is_compressed(tex->desc.b.b.format)) {
+                texstate->filter1 |= R400_DXTC_SWIZZLE_ENABLE;
+            }
+
             /* to emulate 1D textures through 2D ones correctly */
             if (tex->desc.b.b.target == PIPE_TEXTURE_1D) {
                 texstate->filter0 &= ~R300_TX_WRAP_T_MASK;

src/gallium/drivers/r300/r300_texture.c

@@ -40,7 +40,8 @@
 #include "pipe/p_screen.h"
 
 unsigned r300_get_swizzle_combined(const unsigned char *swizzle_format,
-                                   const unsigned char *swizzle_view)
+                                   const unsigned char *swizzle_view,
+                                   boolean dxtc_swizzle)
 {
     unsigned i;
     unsigned char swizzle[4];
@@ -51,10 +52,10 @@ unsigned r300_get_swizzle_combined(const unsigned char *swizzle_format,
         R300_TX_FORMAT_B_SHIFT,
         R300_TX_FORMAT_A_SHIFT
     };
-    const uint32_t swizzle_bit[4] = {
-        R300_TX_FORMAT_X,
+    uint32_t swizzle_bit[4] = {
+        dxtc_swizzle ? R300_TX_FORMAT_Z : R300_TX_FORMAT_X,
         R300_TX_FORMAT_Y,
-        R300_TX_FORMAT_Z,
+        dxtc_swizzle ? R300_TX_FORMAT_X : R300_TX_FORMAT_Z,
         R300_TX_FORMAT_W
     };
 
@@ -107,7 +108,8 @@ unsigned r300_get_swizzle_combined(const unsigned char *swizzle_format,
  * makes available X, Y, Z, W, ZERO, and ONE for swizzling. */
 uint32_t r300_translate_texformat(enum pipe_format format,
                                   const unsigned char *swizzle_view,
-                                  boolean is_r500)
+                                  boolean is_r500,
+                                  boolean dxtc_swizzle)
 {
     uint32_t result = 0;
     const struct util_format_description *desc;
@@ -169,7 +171,8 @@ uint32_t r300_translate_texformat(enum pipe_format format,
             }
     }
 
-    result |= r300_get_swizzle_combined(desc->swizzle, swizzle_view);
+    result |= r300_get_swizzle_combined(desc->swizzle, swizzle_view,
+                    util_format_is_compressed(format) && dxtc_swizzle);
 
     /* S3TC formats. */
     if (desc->layout == UTIL_FORMAT_LAYOUT_S3TC) {
@@ -538,7 +541,7 @@ boolean r300_is_zs_format_supported(enum pipe_format format)
 
 boolean r300_is_sampler_format_supported(enum pipe_format format)
 {
-    return r300_translate_texformat(format, 0, TRUE) != ~0;
+    return r300_translate_texformat(format, 0, TRUE, FALSE) != ~0;
 }
 
 void r300_texture_setup_format_state(struct r300_screen *screen,
@@ -862,7 +865,7 @@ struct pipe_surface* r300_get_tex_surface(struct pipe_screen* screen,
                                                tex->desc.b.b.nr_samples,
                                                tex->desc.microtile,
                                                tex->desc.macrotile[level],
-                                               DIM_HEIGHT);
+                                               DIM_HEIGHT, 0);
 
         surface->cbzb_height = align((surface->base.height + 1) / 2,
                                      tile_height);

src/gallium/drivers/r300/r300_texture.h

@@ -35,11 +35,13 @@ struct r300_texture;
 struct r300_screen;
 
 unsigned r300_get_swizzle_combined(const unsigned char *swizzle_format,
-                                   const unsigned char *swizzle_view);
+                                   const unsigned char *swizzle_view,
+                                   boolean dxtc_swizzle);
 
 uint32_t r300_translate_texformat(enum pipe_format format,
                                   const unsigned char *swizzle_view,
-                                  boolean is_r500);
+                                  boolean is_r500,
+                                  boolean dxtc_swizzle);
 
 uint32_t r500_tx_format_msb_bit(enum pipe_format format);
 

src/gallium/drivers/r300/r300_texture_desc.c

@@ -34,7 +34,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format,
                                   unsigned num_samples,
                                   enum r300_buffer_tiling microtile,
                                   enum r300_buffer_tiling macrotile,
-                                  enum r300_dim dim)
+                                  enum r300_dim dim, boolean is_rs690)
 {
     static const unsigned table[2][5][3][2] =
     {
@@ -44,7 +44,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format,
             {{ 32, 1}, { 8,  4}, { 0,  0}}, /*   8 bits per pixel */
             {{ 16, 1}, { 8,  2}, { 4,  4}}, /*  16 bits per pixel */
             {{  8, 1}, { 4,  2}, { 0,  0}}, /*  32 bits per pixel */
-            {{  4, 1}, { 0,  0}, { 2,  2}}, /*  64 bits per pixel */
+            {{  4, 1}, { 2,  2}, { 0,  0}}, /*  64 bits per pixel */
             {{  2, 1}, { 0,  0}, { 0,  0}}  /* 128 bits per pixel */
         },
         {
@@ -53,10 +53,11 @@ unsigned r300_get_pixel_alignment(enum pipe_format format,
             {{256, 8}, {64, 32}, { 0,  0}}, /*   8 bits per pixel */
             {{128, 8}, {64, 16}, {32, 32}}, /*  16 bits per pixel */
             {{ 64, 8}, {32, 16}, { 0,  0}}, /*  32 bits per pixel */
-            {{ 32, 8}, { 0,  0}, {16, 16}}, /*  64 bits per pixel */
+            {{ 32, 8}, {16, 16}, { 0,  0}}, /*  64 bits per pixel */
             {{ 16, 8}, { 0,  0}, { 0,  0}}  /* 128 bits per pixel */
         }
     };
+
     static const unsigned aa_block[2] = {4, 8};
     unsigned tile = 0;
     unsigned pixsize = util_format_get_blocksize(format);
@@ -74,6 +75,14 @@ unsigned r300_get_pixel_alignment(enum pipe_format format,
     } else {
         /* Standard alignment. */
         tile = table[macrotile][util_logbase2(pixsize)][microtile][dim];
+        if (macrotile == 0 && is_rs690 && dim == DIM_WIDTH) {
+            int align;
+            int h_tile;
+            h_tile = table[macrotile][util_logbase2(pixsize)][microtile][DIM_HEIGHT];
+            align = 64 / (pixsize * h_tile);
+            if (tile < align)
+                tile = align;
+        }
     }
 
     assert(tile);
@@ -89,7 +98,7 @@ static boolean r300_texture_macro_switch(struct r300_texture_desc *desc,
     unsigned tile, texdim;
 
     tile = r300_get_pixel_alignment(desc->b.b.format, desc->b.b.nr_samples,
-                                    desc->microtile, R300_BUFFER_TILED, dim);
+                                    desc->microtile, R300_BUFFER_TILED, dim, 0);
     if (dim == DIM_WIDTH) {
         texdim = u_minify(desc->b.b.width0, level);
     } else {
@@ -113,6 +122,9 @@ static unsigned r300_texture_get_stride(struct r300_screen *screen,
                                         unsigned level)
 {
     unsigned tile_width, width, stride;
+    boolean is_rs690 = (screen->caps.family == CHIP_FAMILY_RS600 ||
+                        screen->caps.family == CHIP_FAMILY_RS690 ||
+                        screen->caps.family == CHIP_FAMILY_RS740);
 
     if (desc->stride_in_bytes_override)
         return desc->stride_in_bytes_override;
@@ -131,38 +143,14 @@ static unsigned r300_texture_get_stride(struct r300_screen *screen,
                                               desc->b.b.nr_samples,
                                               desc->microtile,
                                               desc->macrotile[level],
-                                              DIM_WIDTH);
+                                              DIM_WIDTH, is_rs690);
         width = align(width, tile_width);
 
         stride = util_format_get_stride(desc->b.b.format, width);
-
-        /* Some IGPs need a minimum stride of 64 bytes, hmm... */
-        if (!desc->macrotile[level] &&
-            (screen->caps.family == CHIP_FAMILY_RS600 ||
-             screen->caps.family == CHIP_FAMILY_RS690 ||
-             screen->caps.family == CHIP_FAMILY_RS740)) {
-            unsigned min_stride;
-
-            if (desc->microtile) {
-                unsigned tile_height =
-                        r300_get_pixel_alignment(desc->b.b.format,
-                                                 desc->b.b.nr_samples,
-                                                 desc->microtile,
-                                                 desc->macrotile[level],
-                                                 DIM_HEIGHT);
-
-                min_stride = 64 / tile_height;
-            } else {
-                min_stride = 64;
-            }
-
-            return stride < min_stride ? min_stride : stride;
-        }
-
         /* The alignment to 32 bytes is sort of implied by the layout... */
         return stride;
     } else {
-        return align(util_format_get_stride(desc->b.b.format, width), 32);
+        return align(util_format_get_stride(desc->b.b.format, width), is_rs690 ? 64 : 32);
     }
 }
 
@@ -179,7 +167,7 @@ static unsigned r300_texture_get_nblocksy(struct r300_texture_desc *desc,
                                                desc->b.b.nr_samples,
                                                desc->microtile,
                                                desc->macrotile[level],
-                                               DIM_HEIGHT);
+                                               DIM_HEIGHT, 0);
         height = align(height, tile_height);
 
         /* This is needed for the kernel checker, unfortunately. */
@@ -368,11 +356,11 @@ static void r300_setup_tiling(struct r300_screen *screen,
     switch (util_format_get_blocksize(format)) {
         case 1:
         case 4:
+        case 8:
             desc->microtile = R300_BUFFER_TILED;
             break;
 
         case 2:
-        case 8:
             if (rws->get_value(rws, R300_VID_SQUARE_TILING_SUPPORT)) {
                 desc->microtile = R300_BUFFER_SQUARETILED;
             }

src/gallium/drivers/r300/r300_texture_desc.h

@@ -41,7 +41,7 @@ unsigned r300_get_pixel_alignment(enum pipe_format format,
                                   unsigned num_samples,
                                   enum r300_buffer_tiling microtile,
                                   enum r300_buffer_tiling macrotile,
-                                  enum r300_dim dim);
+                                  enum r300_dim dim, boolean is_rs690);
 
 boolean r300_texture_desc_init(struct r300_screen *rscreen,
                                struct r300_texture_desc *desc,

src/gallium/state_trackers/egl/Makefile

@@ -17,7 +17,7 @@ x11_INCLUDES = \
 	-I$(TOP)/src/mapi \
 	-I$(TOP)/src/mesa \
 	$(X11_CFLAGS) \
-	$(shell pkg-config --cflags-only-I libdrm)
+	$(shell pkg-config --cflags-only-I libdrm dri2proto)
 
 x11_SOURCES = $(wildcard x11/*.c) \
 	      $(TOP)/src/glx/dri2.c

src/gallium/state_trackers/egl/common/egl_g3d_api.c

@@ -609,8 +609,10 @@ egl_g3d_wait_client(_EGLDriver *drv, _EGLDisplay *dpy, _EGLContext *ctx)
 
    gctx->stctxi->flush(gctx->stctxi,
          PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_FRAME, &fence);
-   screen->fence_finish(screen, fence, 0);
-   screen->fence_reference(screen, &fence, NULL);
+   if (fence) {
+      screen->fence_finish(screen, fence, 0);
+      screen->fence_reference(screen, &fence, NULL);
+   }
 
    return EGL_TRUE;
 }

src/gallium/state_trackers/egl/common/egl_g3d_st.c

@@ -51,7 +51,6 @@ egl_g3d_st_manager(struct st_manager *smapi)
 
 static boolean
 egl_g3d_st_manager_get_egl_image(struct st_manager *smapi,
-                                 struct st_context_iface *stctx,
                                  void *egl_image,
                                  struct st_egl_image *out)
 {

src/gallium/state_trackers/egl/x11/native_x11.h

@@ -27,6 +27,7 @@
 #define _NATIVE_X11_H_
 
 #include "common/native.h"
+#include <X11/Xlib.h>
 
 struct native_display *
 x11_create_ximage_display(Display *dpy,

src/gallium/state_trackers/vega/api_context.c

@@ -73,7 +73,8 @@ void vegaFinish(void)
    pipe = ctx->pipe;
 
    pipe->flush(pipe, PIPE_FLUSH_RENDER_CACHE | PIPE_FLUSH_FRAME, &fence);
-
-   pipe->screen->fence_finish(pipe->screen, fence, 0);
-   pipe->screen->fence_reference(pipe->screen, &fence, NULL);
+   if (fence) {
+      pipe->screen->fence_finish(pipe->screen, fence, 0);
+      pipe->screen->fence_reference(pipe->screen, &fence, NULL);
+   }
 }

src/gallium/state_trackers/vega/vg_manager.c

@@ -352,7 +352,7 @@ vg_api_create_context(struct st_api *stapi, struct st_manager *smapi,
       return NULL;
 
    /* only 1.0 is supported */
-   if (attribs->major != 1 || attribs->minor > 0)
+   if (attribs->major > 1 || (attribs->major == 1 && attribs->minor > 0))
       return NULL;
 
    pipe = smapi->screen->context_create(smapi->screen, NULL);

src/gallium/state_trackers/xorg/Makefile

@@ -10,7 +10,7 @@ LIBRARY_INCLUDES = \
 	$(shell pkg-config libkms --atleast-version=1.0 \
 				&& echo "-DHAVE_LIBKMS") \
 	$(shell pkg-config libkms --silence-errors --cflags-only-I) \
-	$(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto) \
+	$(shell pkg-config --cflags-only-I pixman-1 xorg-server libdrm xproto dri2proto) \
 	-I$(TOP)/src/gallium/include \
 	-I$(TOP)/src/gallium/auxiliary \
 	-I$(TOP)/include \

src/gallium/state_trackers/xorg/xorg_crtc.c

@@ -234,6 +234,10 @@ crtc_load_cursor_argb_ga3d(xf86CrtcPtr crtc, CARD32 * image)
 		   64, 64, (void*)image, 64 * 4, 0, 0);
     ms->ctx->transfer_unmap(ms->ctx, transfer);
     ms->ctx->transfer_destroy(ms->ctx, transfer);
+
+    if (crtc->cursor_shown)
+	drmModeSetCursor(ms->fd, crtcp->drm_crtc->crtc_id,
+			 crtcp->cursor_handle, 64, 64);
 }
 
 #if HAVE_LIBKMS
@@ -271,6 +275,10 @@ crtc_load_cursor_argb_kms(xf86CrtcPtr crtc, CARD32 * image)
     memcpy(ptr, image, 64*64*4);
     kms_bo_unmap(crtcp->cursor_bo);
 
+    if (crtc->cursor_shown)
+	drmModeSetCursor(ms->fd, crtcp->drm_crtc->crtc_id,
+			 crtcp->cursor_handle, 64, 64);
+
     return;
 
 err_bo_destroy:
@@ -353,7 +361,7 @@ crtc_destroy(xf86CrtcPtr crtc)
 
     drmModeFreeCrtc(crtcp->drm_crtc);
 
-    xfree(crtcp);
+    free(crtcp);
     crtc->driver_private = NULL;
 }
 
@@ -401,7 +409,7 @@ xorg_crtc_init(ScrnInfoPtr pScrn)
 	if (crtc == NULL)
 	    goto out;
 
-	crtcp = xcalloc(1, sizeof(struct crtc_private));
+	crtcp = calloc(1, sizeof(struct crtc_private));
 	if (!crtcp) {
 	    xf86CrtcDestroy(crtc);
 	    goto out;

src/gallium/state_trackers/xorg/xorg_dri2.c

@@ -201,11 +201,11 @@ dri2_create_buffer(DrawablePtr pDraw, unsigned int attachment, unsigned int form
     DRI2Buffer2Ptr buffer;
     BufferPrivatePtr private;
 
-    buffer = xcalloc(1, sizeof *buffer);
+    buffer = calloc(1, sizeof *buffer);
     if (!buffer)
 	return NULL;
 
-    private = xcalloc(1, sizeof *private);
+    private = calloc(1, sizeof *private);
     if (!private) {
 	goto fail;
     }
@@ -217,9 +217,9 @@ dri2_create_buffer(DrawablePtr pDraw, unsigned int attachment, unsigned int form
     if (dri2_do_create_buffer(pDraw, (DRI2BufferPtr)buffer, format))
 	return buffer;
 
-    xfree(private);
+    free(private);
 fail:
-    xfree(buffer);
+    free(buffer);
     return NULL;
 }
 
@@ -229,8 +229,8 @@ dri2_destroy_buffer(DrawablePtr pDraw, DRI2Buffer2Ptr buffer)
     /* So far it is safe to downcast a DRI2Buffer2Ptr to DRI2BufferPtr */
     dri2_do_destroy_buffer(pDraw, (DRI2BufferPtr)buffer);
 
-    xfree(buffer->driverPrivate);
-    xfree(buffer);
+    free(buffer->driverPrivate);
+    free(buffer);
 }
 
 #endif /* DRI2INFOREC_VERSION >= 2 */
@@ -244,11 +244,11 @@ dri2_create_buffers(DrawablePtr pDraw, unsigned int *attachments, int count)
     DRI2BufferPtr buffers;
     int i;
 
-    buffers = xcalloc(count, sizeof *buffers);
+    buffers = calloc(count, sizeof *buffers);
     if (!buffers)
 	goto fail_buffers;
 
-    privates = xcalloc(count, sizeof *privates);
+    privates = calloc(count, sizeof *privates);
     if (!privates)
 	goto fail_privates;
 
@@ -263,9 +263,9 @@ dri2_create_buffers(DrawablePtr pDraw, unsigned int *attachments, int count)
     return buffers;
 
 fail:
-    xfree(privates);
+    free(privates);
 fail_privates:
-    xfree(buffers);
+    free(buffers);
 fail_buffers:
     return NULL;
 }
@@ -280,8 +280,8 @@ dri2_destroy_buffers(DrawablePtr pDraw, DRI2BufferPtr buffers, int count)
     }
 
     if (buffers) {
-	xfree(buffers[0].driverPrivate);
-	xfree(buffers);
+	free(buffers[0].driverPrivate);
+	free(buffers);
     }
 }
 

src/gallium/state_trackers/xorg/xorg_driver.c

@@ -45,6 +45,7 @@
 #include "miscstruct.h"
 #include "dixstruct.h"
 #include "xf86xv.h"
+#include "xorgVersion.h"
 #ifndef XSERVER_LIBPCIACCESS
 #error "libpciaccess needed"
 #endif
@@ -122,7 +123,7 @@ xorg_tracker_set_functions(ScrnInfoPtr scrn)
 Bool
 xorg_tracker_have_modesetting(ScrnInfoPtr pScrn, struct pci_device *device)
 {
-    char *BusID = xalloc(64);
+    char *BusID = malloc(64);
     sprintf(BusID, "pci:%04x:%02x:%02x.%d",
 	    device->domain, device->bus,
 	    device->dev, device->func);
@@ -130,14 +131,14 @@ xorg_tracker_have_modesetting(ScrnInfoPtr pScrn, struct pci_device *device)
     if (drmCheckModesettingSupported(BusID)) {
 	xf86DrvMsgVerb(pScrn->scrnIndex, X_INFO, 0,
 		       "Drm modesetting not supported %s\n", BusID);
-	xfree(BusID);
+	free(BusID);
 	return FALSE;
     }
 
     xf86DrvMsgVerb(pScrn->scrnIndex, X_INFO, 0,
 		   "Drm modesetting supported on %s\n", BusID);
 
-    xfree(BusID);
+    free(BusID);
     return TRUE;
 }
 
@@ -174,7 +175,7 @@ drv_free_rec(ScrnInfoPtr pScrn)
     if (!pScrn->driverPrivate)
 	return;
 
-    xfree(pScrn->driverPrivate);
+    free(pScrn->driverPrivate);
 
     pScrn->driverPrivate = NULL;
 }
@@ -274,7 +275,7 @@ drv_init_drm(ScrnInfoPtr pScrn)
     if (ms->fd < 0) {
 	char *BusID;
 
-	BusID = xalloc(64);
+	BusID = malloc(64);
 	sprintf(BusID, "PCI:%d:%d:%d",
 		((ms->PciInfo->domain << 8) | ms->PciInfo->bus),
 		ms->PciInfo->dev, ms->PciInfo->func
@@ -283,7 +284,7 @@ drv_init_drm(ScrnInfoPtr pScrn)
 
 	ms->fd = drmOpen(driver_descriptor.driver_name, BusID);
 	ms->isMaster = TRUE;
-	xfree(BusID);
+	free(BusID);
 
 	if (ms->fd >= 0)
 	    return TRUE;
@@ -369,6 +370,7 @@ drv_pre_init(ScrnInfoPtr pScrn, int flags)
     ms = modesettingPTR(pScrn);
     ms->pEnt = pEnt;
     ms->cust = cust;
+    ms->fb_id = -1;
 
     pScrn->displayWidth = 640;	       /* default it */
 
@@ -402,19 +404,6 @@ drv_pre_init(ScrnInfoPtr pScrn, int flags)
     if (!drv_init_drm(pScrn))
 	return FALSE;
 
-    use3D = cust ? !cust->no_3d : TRUE;
-    ms->from_3D = xf86GetOptValBool(ms->Options, OPTION_3D_ACCEL,
-				    &use3D) ?
-	X_CONFIG : X_PROBED;
-
-    ms->no3D = !use3D;
-
-    if (!drv_init_resource_management(pScrn)) {
-	xf86DrvMsg(pScrn->scrnIndex, X_ERROR, "Could not init "
-					       "Gallium3D or libKMS.");
-	return FALSE;
-    }
-
     pScrn->monitor = pScrn->confScreen->monitor;
     pScrn->progClock = TRUE;
     pScrn->rgbBits = 8;
@@ -444,11 +433,24 @@ drv_pre_init(ScrnInfoPtr pScrn, int flags)
 
     /* Process the options */
     xf86CollectOptions(pScrn, NULL);
-    if (!(ms->Options = xalloc(sizeof(drv_options))))
+    if (!(ms->Options = malloc(sizeof(drv_options))))
 	return FALSE;
     memcpy(ms->Options, drv_options, sizeof(drv_options));
     xf86ProcessOptions(pScrn->scrnIndex, pScrn->options, ms->Options);
 
+    use3D = cust ? !cust->no_3d : TRUE;
+    ms->from_3D = xf86GetOptValBool(ms->Options, OPTION_3D_ACCEL,
+				    &use3D) ?
+	X_CONFIG : X_PROBED;
+
+    ms->no3D = !use3D;
+
+    if (!drv_init_resource_management(pScrn)) {
+	xf86DrvMsg(pScrn->scrnIndex, X_ERROR, "Could not init "
+					       "Gallium3D or libKMS.");
+	return FALSE;
+    }
+
     /* Allocate an xf86CrtcConfig */
     xf86CrtcConfigInit(pScrn, &crtc_config_funcs);
     xf86_config = XF86_CRTC_CONFIG_PTR(pScrn);
@@ -791,7 +793,9 @@ drv_screen_init(int scrnIndex, ScreenPtr pScreen, int argc, char **argv)
     if (!ms->SWCursor)
 	xf86_cursors_init(pScreen, 64, 64,
 			  HARDWARE_CURSOR_SOURCE_MASK_INTERLEAVE_64 |
-			  HARDWARE_CURSOR_ARGB);
+			  HARDWARE_CURSOR_ARGB |
+			  ((cust && cust->unhidden_hw_cursor_update) ?
+			   HARDWARE_CURSOR_UPDATE_UNHIDDEN : 0));
 
     /* Must force it before EnterVT, so we are in control of VT and
      * later memory should be bound when allocating, e.g rotate_mem */
@@ -862,8 +866,10 @@ drv_leave_vt(int scrnIndex, int flags)
 	}
     }
 
-    drmModeRmFB(ms->fd, ms->fb_id);
-    ms->fb_id = -1;
+    if (ms->fb_id != -1) {
+	drmModeRmFB(ms->fd, ms->fb_id);
+	ms->fb_id = -1;
+    }
 
     /* idle hardware */
     if (!ms->kms)
@@ -944,7 +950,6 @@ drv_close_screen(int scrnIndex, ScreenPtr pScreen)
     }
 #endif
 
-    drmModeRmFB(ms->fd, ms->fb_id);
     ms->destroy_front_buffer(pScrn);
 
     if (ms->exa)
@@ -1178,6 +1183,8 @@ drv_bind_front_buffer_kms(ScrnInfoPtr pScrn)
 				stride,
 				ptr);
 
+#if (XORG_VERSION_CURRENT < XORG_VERSION_NUMERIC(1, 9, 99, 1, 0))
+
     /* This a hack to work around EnableDisableFBAccess setting the pointer
      * the real fix would be to replace pScrn->EnableDisableFBAccess hook
      * and set the rootPixmap->devPrivate.ptr to something valid before that.
@@ -1187,6 +1194,8 @@ drv_bind_front_buffer_kms(ScrnInfoPtr pScrn)
      */
     pScrn->pixmapPrivate.ptr = ptr;
 
+#endif
+
     return TRUE;
 
 err_destroy:

src/gallium/state_trackers/xorg/xorg_exa.c

@@ -720,7 +720,7 @@ ExaCreatePixmap(ScreenPtr pScreen, int size, int align)
 {
     struct exa_pixmap_priv *priv;
 
-    priv = xcalloc(1, sizeof(struct exa_pixmap_priv));
+    priv = calloc(1, sizeof(struct exa_pixmap_priv));
     if (!priv)
 	return NULL;
 
@@ -737,7 +737,7 @@ ExaDestroyPixmap(ScreenPtr pScreen, void *dPriv)
 
     pipe_resource_reference(&priv->tex, NULL);
 
-    xfree(priv);
+    free(priv);
 }
 
 static Bool
@@ -975,7 +975,7 @@ xorg_exa_close(ScrnInfoPtr pScrn)
    ms->ctx = NULL;
 
    exaDriverFini(pScrn->pScreen);
-   xfree(exa);
+   free(exa);
    ms->exa = NULL;
 }
 
@@ -987,7 +987,7 @@ xorg_exa_init(ScrnInfoPtr pScrn, Bool accel)
    ExaDriverPtr pExa;
    CustomizerPtr cust = ms->cust;
 
-   exa = xcalloc(1, sizeof(struct exa_context));
+   exa = calloc(1, sizeof(struct exa_context));
    if (!exa)
       return NULL;
 

src/gallium/state_trackers/xorg/xorg_output.c

@@ -128,7 +128,7 @@ output_get_modes(xf86OutputPtr output)
     for (i = 0; i < drm_connector->count_modes; i++) {
 	drm_mode = &drm_connector->modes[i];
 	if (drm_mode) {
-	    mode = xcalloc(1, sizeof(DisplayModeRec));
+	    mode = calloc(1, sizeof(DisplayModeRec));
 	    if (!mode)
 		continue;
 	    mode->Clock = drm_mode->clock;
@@ -195,7 +195,7 @@ output_destroy(xf86OutputPtr output)
 {
     struct output_private *priv = output->driver_private;
     drmModeFreeConnector(priv->drm_connector);
-    xfree(priv);
+    free(priv);
     output->driver_private = NULL;
 }
 
@@ -262,14 +262,14 @@ xorg_output_init(ScrnInfoPtr pScrn)
 		 drm_connector->connector_type_id);
 
 
-	priv = xcalloc(sizeof(*priv), 1);
+	priv = calloc(sizeof(*priv), 1);
 	if (!priv) {
 	    continue;
 	}
 
 	output = xf86OutputCreate(pScrn, &output_funcs, name);
 	if (!output) {
-	    xfree(priv);
+	    free(priv);
 	    continue;
 	}
 

src/gallium/state_trackers/xorg/xorg_tracker.h

@@ -76,6 +76,7 @@ typedef struct _CustomizerRec
     Bool dirty_throttling;
     Bool swap_throttling;
     Bool no_3d;
+    Bool unhidden_hw_cursor_update;
     Bool (*winsys_pre_init) (struct _CustomizerRec *cust, int fd);
     Bool (*winsys_screen_init)(struct _CustomizerRec *cust);
     Bool (*winsys_screen_close)(struct _CustomizerRec *cust);

src/gallium/state_trackers/xorg/xorg_xv.c

@@ -536,8 +536,10 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
    dst_surf = xorg_gpu_surface(pPriv->r->pipe->screen, dst);
    hdtv = ((src_w >= RES_720P_X) && (src_h >= RES_720P_Y));
 
+#ifdef COMPOSITE
    REGION_TRANSLATE(pScrn->pScreen, dstRegion, -pPixmap->screen_x,
                     -pPixmap->screen_y);
+#endif
 
    dxo = dstRegion->extents.x1;
    dyo = dstRegion->extents.y1;
@@ -562,11 +564,16 @@ display_video(ScrnInfoPtr pScrn, struct xorg_xv_port_priv *pPriv, int id,
       int box_y2 = pbox->y2;
       float diff_x = (float)src_w / (float)dst_w;
       float diff_y = (float)src_h / (float)dst_h;
-      float offset_x = box_x1 - dstX + pPixmap->screen_x;
-      float offset_y = box_y1 - dstY + pPixmap->screen_y;
+      float offset_x = box_x1 - dstX;
+      float offset_y = box_y1 - dstY;
       float offset_w;
       float offset_h;
 
+#ifdef COMPOSITE
+      offset_x += pPixmap->screen_x;
+      offset_y += pPixmap->screen_y;
+#endif
+
       x = box_x1;
       y = box_y1;
       w = box_x2 - box_x1;

src/gallium/targets/Makefile.dri

@@ -80,7 +80,7 @@ $(LIBNAME): $(OBJECTS) $(MESA_MODULES) $(PIPE_DRIVERS) Makefile \
 		$(OBJECTS) $(PIPE_DRIVERS) \
                 -Wl,--start-group $(MESA_MODULES) -Wl,--end-group \
                  $(DRI_LIB_DEPS) $(DRIVER_EXTRAS)
-	$(CXX) $(CFLAGS) -o $@.test $(TOP)/src/mesa/drivers/dri/common/dri_test.o $@.tmp $(DRI_LIB_DEPS);
+	$(CXX) $(CFLAGS) -o $@.test $(TOP)/src/mesa/drivers/dri/common/dri_test.o $@.tmp $(DRI_LIB_DEPS) $(LDFLAGS);
 	@rm -f $@.test
 	mv -f $@.tmp $@
 

src/gallium/targets/Makefile.xorg

@@ -29,7 +29,7 @@ INCLUDES = \
 LIBNAME_STAGING = $(TOP)/$(LIB_DIR)/gallium/$(TARGET)
 
 ifeq ($(MESA_LLVM),1)
-LD = g++
+LD = $(CXX)
 LDFLAGS += $(LLVM_LDFLAGS)
 USE_CXX=1
 DRIVER_PIPES += $(TOP)/src/gallium/drivers/llvmpipe/libllvmpipe.a
@@ -42,7 +42,7 @@ endif
 default: depend $(TOP)/$(LIB_DIR)/gallium $(LIBNAME) $(LIBNAME_STAGING)
 
 $(LIBNAME): $(OBJECTS) Makefile ../Makefile.xorg $(LIBS) $(DRIVER_PIPES)
-	$(MKLIB) -noprefix -o $@ $(LDFLAGS) $(OBJECTS) $(DRIVER_PIPES) $(GALLIUM_AUXILIARIES) $(DRIVER_LINKS)
+	$(MKLIB) -linker $(CC) -noprefix -o $@ $(LDFLAGS) $(OBJECTS) $(DRIVER_PIPES) $(GALLIUM_AUXILIARIES) $(DRIVER_LINKS)
 
 depend: $(C_SOURCES) $(CPP_SOURCES) $(ASM_SOURCES) $(SYMLINKS) $(GENERATED_SOURCES)
 	rm -f depend

src/gallium/targets/dri-vmwgfx/Makefile

@@ -12,6 +12,7 @@ PIPE_DRIVERS = \
 
 C_SOURCES = \
 	target.c \
+	vmw_powf.c \
 	$(COMMON_GALLIUM_SOURCES)
 
 DRIVER_DEFINES = \

src/gallium/targets/dri-vmwgfx/vmw_powf.c

@@ -0,0 +1,17 @@
+/**
+ * Powf may leave an unresolved symbol pointing to a libstdc++.so powf.
+ * However, not all libstdc++.so include this function, so optionally
+ * replace the powf function with calls to expf and logf.
+ */
+
+#ifdef VMW_RESOLVE_POWF
+
+extern float expf(float x);
+extern float logf(float x);
+extern float powf(float x, float y);
+
+float powf(float x, float y) {
+    return expf(logf(x)*y);
+}
+
+#endif

src/gallium/targets/egl/Makefile

@@ -24,7 +24,9 @@ common_CPPFLAGS := \
 	-I$(TOP)/src/gallium/auxiliary \
 	-I$(TOP)/src/gallium/drivers \
 	-I$(TOP)/src/gallium/include \
-	-I$(TOP)/src/gallium/winsys
+	-I$(TOP)/src/gallium/winsys \
+	$(LIBDRM_CFLAGS)
+
 common_SYS :=
 common_LIBS := \
 	$(TOP)/src/gallium/drivers/identity/libidentity.a \
@@ -41,11 +43,11 @@ egl_SYS := -lm $(DLOPEN_LIBS) -L$(TOP)/$(LIB_DIR) -lEGL
 egl_LIBS := $(TOP)/src/gallium/state_trackers/egl/libegl.a
 
 ifneq ($(findstring x11, $(EGL_PLATFORMS)),)
-egl_SYS += -lX11 -lXext -lXfixes
+egl_SYS += -lX11 -lXext -lXfixes $(LIBDRM_LIB)
 egl_LIBS += $(TOP)/src/gallium/winsys/sw/xlib/libws_xlib.a
 endif
-ifneq ($(findstring kms, $(EGL_PLATFORMS)),)
-egl_SYS += -ldrm
+ifneq ($(findstring drm, $(EGL_PLATFORMS)),)
+egl_SYS += $(LIBDRM_LIB)
 endif
 ifneq ($(findstring fbdev, $(EGL_PLATFORMS)),)
 egl_LIBS += $(TOP)/src/gallium/winsys/sw/fbdev/libfbdev.a

src/gallium/targets/egl/pipe_i965.c

@@ -1,6 +1,7 @@
 
 #include "target-helpers/inline_wrapper_sw_helper.h"
 #include "target-helpers/inline_debug_helper.h"
+#include "target-helpers/inline_wrapper_sw_helper.h"
 #include "state_tracker/drm_driver.h"
 #include "i965/drm/i965_drm_public.h"
 #include "i965/brw_public.h"

src/gallium/targets/xorg-vmwgfx/vmw_ctrl.c

@@ -32,6 +32,7 @@
  *      allows X clients to communicate with the driver.
  */
 
+#include <xorg-server.h>
 #include "dixstruct.h"
 #include "extnsionst.h"
 #include <X11/X.h>
@@ -211,7 +212,7 @@ VMwareCtrlDoSetTopology(ScrnInfoPtr pScrn,
    struct vmw_customizer *vmw = vmw_customizer(xorg_customizer(pScrn));
    int i;
 
-   rects = xcalloc(number, sizeof(*rects));
+   rects = calloc(number, sizeof(*rects));
    if (!rects)
       return FALSE;
 
@@ -224,7 +225,7 @@ VMwareCtrlDoSetTopology(ScrnInfoPtr pScrn,
 
    vmw_ioctl_update_layout(vmw, number, rects);
 
-   xfree(rects);
+   free(rects);
    return TRUE;
 }
 

src/gallium/targets/xorg-vmwgfx/vmw_ioctl.c

@@ -165,7 +165,7 @@ vmw_ioctl_buffer_create(struct vmw_customizer *vmw, uint32_t size, unsigned *han
     struct drm_vmw_dmabuf_rep *rep = &arg.rep;
     int ret;
 
-    buf = xcalloc(1, sizeof(*buf));
+    buf = calloc(1, sizeof(*buf));
     if (!buf)
 	goto err;
 
@@ -192,7 +192,7 @@ vmw_ioctl_buffer_create(struct vmw_customizer *vmw, uint32_t size, unsigned *han
     return buf;
 
 err_free:
-    xfree(buf);
+    free(buf);
 err:
     return NULL;
 }
@@ -211,7 +211,7 @@ vmw_ioctl_buffer_destroy(struct vmw_customizer *vmw, struct vmw_dma_buffer *buf)
     arg.handle = buf->handle; 
     drmCommandWrite(vmw->fd, DRM_VMW_UNREF_DMABUF, &arg, sizeof(arg)); 
 
-    xfree(buf); 
+    free(buf);
 } 
 
 void *

src/gallium/targets/xorg-vmwgfx/vmw_screen.c

@@ -245,6 +245,7 @@ vmw_screen_pre_init(ScrnInfoPtr pScrn, int flags)
     cust->winsys_enter_vt = vmw_screen_enter_vt;
     cust->winsys_leave_vt = vmw_screen_leave_vt;
     cust->no_3d = TRUE;
+    cust->unhidden_hw_cursor_update = TRUE;
     vmw->pScrn = pScrn;
 
     pScrn->driverPrivate = cust;

src/gallium/targets/xorg-vmwgfx/vmw_video.c

@@ -300,7 +300,7 @@ vmw_video_init(struct vmw_customizer *vmw)
         numAdaptors = 1;
         overlayAdaptors = &newAdaptor;
     } else {
-         newAdaptors = xalloc((numAdaptors + 1) *
+         newAdaptors = malloc((numAdaptors + 1) *
                               sizeof(XF86VideoAdaptorPtr*));
          if (!newAdaptors) {
             xf86XVFreeVideoAdaptorRec(newAdaptor);
@@ -320,7 +320,7 @@ vmw_video_init(struct vmw_customizer *vmw)
     }
 
     if (newAdaptors) {
-        xfree(newAdaptors);
+        free(newAdaptors);
     }
 
     debug_printf("Initialized VMware Xv extension successfully\n");
@@ -438,7 +438,7 @@ vmw_video_init_adaptor(ScrnInfoPtr pScrn, struct vmw_customizer *vmw)
         return NULL;
     }
 
-    video = xcalloc(1, sizeof(*video));
+    video = calloc(1, sizeof(*video));
     if (!video) {
         debug_printf("Not enough memory.\n");
         xf86XVFreeVideoAdaptorRec(adaptor);
@@ -742,7 +742,7 @@ vmw_video_buffer_alloc(struct vmw_customizer *vmw, int size,
     }
 
     out->size = size;
-    out->extra_data = xcalloc(1, size);
+    out->extra_data = calloc(1, size);
 
     debug_printf("\t\t%s: allocated buffer %p of size %i\n", __func__, out, size);
 
@@ -773,7 +773,7 @@ vmw_video_buffer_free(struct vmw_customizer *vmw,
     if (out->size == 0)
 	return Success;
 
-    xfree(out->extra_data);
+    free(out->extra_data);
     vmw_ioctl_buffer_unmap(vmw, out->buf);
     vmw_ioctl_buffer_destroy(vmw, out->buf);
 

src/gallium/winsys/radeon/drm/radeon_drm.h

@@ -37,7 +37,214 @@
  * I believe that this check is valid, but I haven't been exhaustive. */
 static INLINE boolean is_r3xx(int pciid)
 {
-    return (pciid > 0x3150) && (pciid < 0x796f);
+      switch (pciid) {
+      case 0x4144: /* PCI_CHIP_R300_AD */
+      case 0x4145: /* PCI_CHIP_R300_AE */
+      case 0x4146: /* PCI_CHIP_R300_AF */
+      case 0x4147: /* PCI_CHIP_R300_AG */
+      case 0x4E44: /* PCI_CHIP_R300_ND */
+      case 0x4E45: /* PCI_CHIP_R300_NE */
+      case 0x4E46: /* PCI_CHIP_R300_NF */
+      case 0x4E47: /* PCI_CHIP_R300_NG */
+      case 0x4E48: /* PCI_CHIP_R350_NH */
+      case 0x4E49: /* PCI_CHIP_R350_NI */
+      case 0x4E4B: /* PCI_CHIP_R350_NK */
+      case 0x4148: /* PCI_CHIP_R350_AH */
+      case 0x4149: /* PCI_CHIP_R350_AI */
+      case 0x414A: /* PCI_CHIP_R350_AJ */
+      case 0x414B: /* PCI_CHIP_R350_AK */
+      case 0x4E4A: /* PCI_CHIP_R360_NJ */
+      case 0x4150: /* PCI_CHIP_RV350_AP */
+      case 0x4151: /* PCI_CHIP_RV350_AQ */
+      case 0x4152: /* PCI_CHIP_RV350_AR */
+      case 0x4153: /* PCI_CHIP_RV350_AS */
+      case 0x4154: /* PCI_CHIP_RV350_AT */
+      case 0x4155: /* PCI_CHIP_RV350_AU */
+      case 0x4156: /* PCI_CHIP_RV350_AV */
+      case 0x4E50: /* PCI_CHIP_RV350_NP */
+      case 0x4E51: /* PCI_CHIP_RV350_NQ */
+      case 0x4E52: /* PCI_CHIP_RV350_NR */
+      case 0x4E53: /* PCI_CHIP_RV350_NS */
+      case 0x4E54: /* PCI_CHIP_RV350_NT */
+      case 0x4E56: /* PCI_CHIP_RV350_NV */
+      case 0x5460: /* PCI_CHIP_RV370_5460 */
+      case 0x5462: /* PCI_CHIP_RV370_5462 */
+      case 0x5464: /* PCI_CHIP_RV370_5464 */
+      case 0x5B60: /* PCI_CHIP_RV370_5B60 */
+      case 0x5B62: /* PCI_CHIP_RV370_5B62 */
+      case 0x5B63: /* PCI_CHIP_RV370_5B63 */
+      case 0x5B64: /* PCI_CHIP_RV370_5B64 */
+      case 0x5B65: /* PCI_CHIP_RV370_5B65 */
+      case 0x3150: /* PCI_CHIP_RV380_3150 */
+      case 0x3152: /* PCI_CHIP_RV380_3152 */
+      case 0x3154: /* PCI_CHIP_RV380_3154 */
+      case 0x3155: /* PCI_CHIP_RV380_3155 */
+      case 0x3E50: /* PCI_CHIP_RV380_3E50 */
+      case 0x3E54: /* PCI_CHIP_RV380_3E54 */
+      case 0x4A48: /* PCI_CHIP_R420_JH */
+      case 0x4A49: /* PCI_CHIP_R420_JI */
+      case 0x4A4A: /* PCI_CHIP_R420_JJ */
+      case 0x4A4B: /* PCI_CHIP_R420_JK */
+      case 0x4A4C: /* PCI_CHIP_R420_JL */
+      case 0x4A4D: /* PCI_CHIP_R420_JM */
+      case 0x4A4E: /* PCI_CHIP_R420_JN */
+      case 0x4A4F: /* PCI_CHIP_R420_JO */
+      case 0x4A50: /* PCI_CHIP_R420_JP */
+      case 0x4A54: /* PCI_CHIP_R420_JT */
+      case 0x5548: /* PCI_CHIP_R423_UH */
+      case 0x5549: /* PCI_CHIP_R423_UI */
+      case 0x554A: /* PCI_CHIP_R423_UJ */
+      case 0x554B: /* PCI_CHIP_R423_UK */
+      case 0x5550: /* PCI_CHIP_R423_5550 */
+      case 0x5551: /* PCI_CHIP_R423_UQ */
+      case 0x5552: /* PCI_CHIP_R423_UR */
+      case 0x5554: /* PCI_CHIP_R423_UT */
+      case 0x5D57: /* PCI_CHIP_R423_5D57 */
+      case 0x554C: /* PCI_CHIP_R430_554C */
+      case 0x554D: /* PCI_CHIP_R430_554D */
+      case 0x554E: /* PCI_CHIP_R430_554E */
+      case 0x554F: /* PCI_CHIP_R430_554F */
+      case 0x5D48: /* PCI_CHIP_R430_5D48 */
+      case 0x5D49: /* PCI_CHIP_R430_5D49 */
+      case 0x5D4A: /* PCI_CHIP_R430_5D4A */
+      case 0x5D4C: /* PCI_CHIP_R480_5D4C */
+      case 0x5D4D: /* PCI_CHIP_R480_5D4D */
+      case 0x5D4E: /* PCI_CHIP_R480_5D4E */
+      case 0x5D4F: /* PCI_CHIP_R480_5D4F */
+      case 0x5D50: /* PCI_CHIP_R480_5D50 */
+      case 0x5D52: /* PCI_CHIP_R480_5D52 */
+      case 0x4B49: /* PCI_CHIP_R481_4B49 */
+      case 0x4B4A: /* PCI_CHIP_R481_4B4A */
+      case 0x4B4B: /* PCI_CHIP_R481_4B4B */
+      case 0x4B4C: /* PCI_CHIP_R481_4B4C */
+      case 0x564A: /* PCI_CHIP_RV410_564A */
+      case 0x564B: /* PCI_CHIP_RV410_564B */
+      case 0x564F: /* PCI_CHIP_RV410_564F */
+      case 0x5652: /* PCI_CHIP_RV410_5652 */
+      case 0x5653: /* PCI_CHIP_RV410_5653 */
+      case 0x5657: /* PCI_CHIP_RV410_5657 */
+      case 0x5E48: /* PCI_CHIP_RV410_5E48 */
+      case 0x5E4A: /* PCI_CHIP_RV410_5E4A */
+      case 0x5E4B: /* PCI_CHIP_RV410_5E4B */
+      case 0x5E4C: /* PCI_CHIP_RV410_5E4C */
+      case 0x5E4D: /* PCI_CHIP_RV410_5E4D */
+      case 0x5E4F: /* PCI_CHIP_RV410_5E4F */
+      case 0x5A41: /* PCI_CHIP_RS400_5A41 */
+      case 0x5A42: /* PCI_CHIP_RS400_5A42 */
+      case 0x5A61: /* PCI_CHIP_RC410_5A61 */
+      case 0x5A62: /* PCI_CHIP_RC410_5A62 */
+      case 0x5954: /* PCI_CHIP_RS480_5954 */
+      case 0x5955: /* PCI_CHIP_RS480_5955 */
+      case 0x5974: /* PCI_CHIP_RS482_5974 */
+      case 0x5975: /* PCI_CHIP_RS482_5975 */
+      case 0x7100: /* PCI_CHIP_R520_7100 */
+      case 0x7101: /* PCI_CHIP_R520_7101 */
+      case 0x7102: /* PCI_CHIP_R520_7102 */
+      case 0x7103: /* PCI_CHIP_R520_7103 */
+      case 0x7104: /* PCI_CHIP_R520_7104 */
+      case 0x7105: /* PCI_CHIP_R520_7105 */
+      case 0x7106: /* PCI_CHIP_R520_7106 */
+      case 0x7108: /* PCI_CHIP_R520_7108 */
+      case 0x7109: /* PCI_CHIP_R520_7109 */
+      case 0x710A: /* PCI_CHIP_R520_710A */
+      case 0x710B: /* PCI_CHIP_R520_710B */
+      case 0x710C: /* PCI_CHIP_R520_710C */
+      case 0x710E: /* PCI_CHIP_R520_710E */
+      case 0x710F: /* PCI_CHIP_R520_710F */
+      case 0x7140: /* PCI_CHIP_RV515_7140 */
+      case 0x7141: /* PCI_CHIP_RV515_7141 */
+      case 0x7142: /* PCI_CHIP_RV515_7142 */
+      case 0x7143: /* PCI_CHIP_RV515_7143 */
+      case 0x7144: /* PCI_CHIP_RV515_7144 */
+      case 0x7145: /* PCI_CHIP_RV515_7145 */
+      case 0x7146: /* PCI_CHIP_RV515_7146 */
+      case 0x7147: /* PCI_CHIP_RV515_7147 */
+      case 0x7149: /* PCI_CHIP_RV515_7149 */
+      case 0x714A: /* PCI_CHIP_RV515_714A */
+      case 0x714B: /* PCI_CHIP_RV515_714B */
+      case 0x714C: /* PCI_CHIP_RV515_714C */
+      case 0x714D: /* PCI_CHIP_RV515_714D */
+      case 0x714E: /* PCI_CHIP_RV515_714E */
+      case 0x714F: /* PCI_CHIP_RV515_714F */
+      case 0x7151: /* PCI_CHIP_RV515_7151 */
+      case 0x7152: /* PCI_CHIP_RV515_7152 */
+      case 0x7153: /* PCI_CHIP_RV515_7153 */
+      case 0x715E: /* PCI_CHIP_RV515_715E */
+      case 0x715F: /* PCI_CHIP_RV515_715F */
+      case 0x7180: /* PCI_CHIP_RV515_7180 */
+      case 0x7181: /* PCI_CHIP_RV515_7181 */
+      case 0x7183: /* PCI_CHIP_RV515_7183 */
+      case 0x7186: /* PCI_CHIP_RV515_7186 */
+      case 0x7187: /* PCI_CHIP_RV515_7187 */
+      case 0x7188: /* PCI_CHIP_RV515_7188 */
+      case 0x718A: /* PCI_CHIP_RV515_718A */
+      case 0x718B: /* PCI_CHIP_RV515_718B */
+      case 0x718C: /* PCI_CHIP_RV515_718C */
+      case 0x718D: /* PCI_CHIP_RV515_718D */
+      case 0x718F: /* PCI_CHIP_RV515_718F */
+      case 0x7193: /* PCI_CHIP_RV515_7193 */
+      case 0x7196: /* PCI_CHIP_RV515_7196 */
+      case 0x719B: /* PCI_CHIP_RV515_719B */
+      case 0x719F: /* PCI_CHIP_RV515_719F */
+      case 0x7200: /* PCI_CHIP_RV515_7200 */
+      case 0x7210: /* PCI_CHIP_RV515_7210 */
+      case 0x7211: /* PCI_CHIP_RV515_7211 */
+      case 0x71C0: /* PCI_CHIP_RV530_71C0 */
+      case 0x71C1: /* PCI_CHIP_RV530_71C1 */
+      case 0x71C2: /* PCI_CHIP_RV530_71C2 */
+      case 0x71C3: /* PCI_CHIP_RV530_71C3 */
+      case 0x71C4: /* PCI_CHIP_RV530_71C4 */
+      case 0x71C5: /* PCI_CHIP_RV530_71C5 */
+      case 0x71C6: /* PCI_CHIP_RV530_71C6 */
+      case 0x71C7: /* PCI_CHIP_RV530_71C7 */
+      case 0x71CD: /* PCI_CHIP_RV530_71CD */
+      case 0x71CE: /* PCI_CHIP_RV530_71CE */
+      case 0x71D2: /* PCI_CHIP_RV530_71D2 */
+      case 0x71D4: /* PCI_CHIP_RV530_71D4 */
+      case 0x71D5: /* PCI_CHIP_RV530_71D5 */
+      case 0x71D6: /* PCI_CHIP_RV530_71D6 */
+      case 0x71DA: /* PCI_CHIP_RV530_71DA */
+      case 0x71DE: /* PCI_CHIP_RV530_71DE */
+      case 0x7281: /* PCI_CHIP_RV560_7281 */
+      case 0x7283: /* PCI_CHIP_RV560_7283 */
+      case 0x7287: /* PCI_CHIP_RV560_7287 */
+      case 0x7290: /* PCI_CHIP_RV560_7290 */
+      case 0x7291: /* PCI_CHIP_RV560_7291 */
+      case 0x7293: /* PCI_CHIP_RV560_7293 */
+      case 0x7297: /* PCI_CHIP_RV560_7297 */
+      case 0x7280: /* PCI_CHIP_RV570_7280 */
+      case 0x7288: /* PCI_CHIP_RV570_7288 */
+      case 0x7289: /* PCI_CHIP_RV570_7289 */
+      case 0x728B: /* PCI_CHIP_RV570_728B */
+      case 0x728C: /* PCI_CHIP_RV570_728C */
+      case 0x7240: /* PCI_CHIP_R580_7240 */
+      case 0x7243: /* PCI_CHIP_R580_7243 */
+      case 0x7244: /* PCI_CHIP_R580_7244 */
+      case 0x7245: /* PCI_CHIP_R580_7245 */
+      case 0x7246: /* PCI_CHIP_R580_7246 */
+      case 0x7247: /* PCI_CHIP_R580_7247 */
+      case 0x7248: /* PCI_CHIP_R580_7248 */
+      case 0x7249: /* PCI_CHIP_R580_7249 */
+      case 0x724A: /* PCI_CHIP_R580_724A */
+      case 0x724B: /* PCI_CHIP_R580_724B */
+      case 0x724C: /* PCI_CHIP_R580_724C */
+      case 0x724D: /* PCI_CHIP_R580_724D */
+      case 0x724E: /* PCI_CHIP_R580_724E */
+      case 0x724F: /* PCI_CHIP_R580_724F */
+      case 0x7284: /* PCI_CHIP_R580_7284 */
+      case 0x793F: /* PCI_CHIP_RS600_793F */
+      case 0x7941: /* PCI_CHIP_RS600_7941 */
+      case 0x7942: /* PCI_CHIP_RS600_7942 */
+      case 0x791E: /* PCI_CHIP_RS690_791E */
+      case 0x791F: /* PCI_CHIP_RS690_791F */
+      case 0x796C: /* PCI_CHIP_RS740_796C */
+      case 0x796D: /* PCI_CHIP_RS740_796D */
+      case 0x796E: /* PCI_CHIP_RS740_796E */
+      case 0x796F: /* PCI_CHIP_RS740_796F */
+	      return TRUE;
+      default:
+	      return FALSE;
+      }
 }
 
 #endif

src/glsl/Makefile

@@ -16,6 +16,7 @@ GLCPP_SOURCES = \
 	glcpp/glcpp.c
 
 C_SOURCES = \
+	strtod.c \
 	$(LIBGLCPP_SOURCES)
 
 CXX_SOURCES = \
@@ -75,8 +76,10 @@ CXX_SOURCES = \
 	loop_analysis.cpp \
 	loop_controls.cpp \
 	loop_unroll.cpp \
+	lower_discard.cpp \
 	lower_noise.cpp \
 	lower_variable_index_to_cond_assign.cpp \
+	opt_discard_simplification.cpp \
 	opt_redundant_jumps.cpp \
 	s_expression.cpp
 
@@ -177,7 +180,7 @@ builtins: builtin_function.cpp builtins/profiles/* builtins/ir/* builtins/tools/
 	cp builtins/tools/builtin_function.cpp .
 	make glsl_compiler
 	@echo Regenerating builtin_function.cpp...
-	./builtins/tools/generate_builtins.py > builtin_function.cpp
+	$(PYTHON2) $(PYTHON_FLAGS) builtins/tools/generate_builtins.py > builtin_function.cpp
 	@echo Rebuilding the real compiler...
 	make glsl_compiler
 

src/glsl/SConscript

@@ -72,10 +72,13 @@ sources = [
     'loop_analysis.cpp',
     'loop_controls.cpp',
     'loop_unroll.cpp',
+    'lower_discard.cpp',
     'lower_noise.cpp',
     'lower_variable_index_to_cond_assign.cpp',
+    'opt_discard_simplification.cpp',
     'opt_redundant_jumps.cpp',
     's_expression.cpp',
+    'strtod.c',
 ]
 
 glsl = env.ConvenienceLibrary(

src/glsl/ast_function.cpp

@@ -545,7 +545,7 @@ emit_inline_vector_constructor(const glsl_type *type,
 	    /* Mask of fields to be written in the assignment.
 	     */
 	    constant_mask |= ((1U << rhs_components) - 1) << base_lhs_component;
-	    constant_components++;
+	    constant_components += rhs_components;
 
 	    base_component += rhs_components;
 	 }

src/glsl/ast_to_hir.cpp

@@ -615,6 +615,94 @@ ast_node::hir(exec_list *instructions,
    return NULL;
 }
 
+static void
+mark_whole_array_access(ir_rvalue *access)
+{
+   ir_dereference_variable *deref = access->as_dereference_variable();
+
+   if (deref) {
+      deref->var->max_array_access = deref->type->length - 1;
+   }
+}
+
+static ir_rvalue *
+do_comparison(void *mem_ctx, int operation, ir_rvalue *op0, ir_rvalue *op1)
+{
+   int join_op;
+   ir_rvalue *cmp = NULL;
+
+   if (operation == ir_binop_all_equal)
+      join_op = ir_binop_logic_and;
+   else
+      join_op = ir_binop_logic_or;
+
+   switch (op0->type->base_type) {
+   case GLSL_TYPE_FLOAT:
+   case GLSL_TYPE_UINT:
+   case GLSL_TYPE_INT:
+   case GLSL_TYPE_BOOL:
+      return new(mem_ctx) ir_expression(operation, op0, op1);
+
+   case GLSL_TYPE_ARRAY: {
+      for (unsigned int i = 0; i < op0->type->length; i++) {
+	 ir_rvalue *e0, *e1, *result;
+
+	 e0 = new(mem_ctx) ir_dereference_array(op0->clone(mem_ctx, NULL),
+						new(mem_ctx) ir_constant(i));
+	 e1 = new(mem_ctx) ir_dereference_array(op1->clone(mem_ctx, NULL),
+						new(mem_ctx) ir_constant(i));
+	 result = do_comparison(mem_ctx, operation, e0, e1);
+
+	 if (cmp) {
+	    cmp = new(mem_ctx) ir_expression(join_op, cmp, result);
+	 } else {
+	    cmp = result;
+	 }
+      }
+
+      mark_whole_array_access(op0);
+      mark_whole_array_access(op1);
+      break;
+   }
+
+   case GLSL_TYPE_STRUCT: {
+      for (unsigned int i = 0; i < op0->type->length; i++) {
+	 ir_rvalue *e0, *e1, *result;
+	 const char *field_name = op0->type->fields.structure[i].name;
+
+	 e0 = new(mem_ctx) ir_dereference_record(op0->clone(mem_ctx, NULL),
+						 field_name);
+	 e1 = new(mem_ctx) ir_dereference_record(op1->clone(mem_ctx, NULL),
+						 field_name);
+	 result = do_comparison(mem_ctx, operation, e0, e1);
+
+	 if (cmp) {
+	    cmp = new(mem_ctx) ir_expression(join_op, cmp, result);
+	 } else {
+	    cmp = result;
+	 }
+      }
+      break;
+   }
+
+   case GLSL_TYPE_ERROR:
+   case GLSL_TYPE_VOID:
+   case GLSL_TYPE_SAMPLER:
+      /* I assume a comparison of a struct containing a sampler just
+       * ignores the sampler present in the type.
+       */
+      break;
+
+   default:
+      assert(!"Should not get here.");
+      break;
+   }
+
+   if (cmp == NULL)
+      cmp = new(mem_ctx) ir_constant(true);
+
+   return cmp;
+}
 
 ir_rvalue *
 ast_expression::hir(exec_list *instructions,
@@ -800,11 +888,10 @@ ast_expression::hir(exec_list *instructions,
 	 error_emitted = true;
       }
 
-      result = new(ctx) ir_expression(operations[this->oper], glsl_type::bool_type,
-				      op[0], op[1]);
+      result = do_comparison(ctx, operations[this->oper], op[0], op[1]);
       type = glsl_type::bool_type;
 
-      assert(result->type == glsl_type::bool_type);
+      assert(error_emitted || (result->type == glsl_type::bool_type));
       break;
 
    case ast_bit_and:
@@ -1954,10 +2041,37 @@ ast_declarator_list::hir(exec_list *instructions,
 
 	    /* Never emit code to initialize a uniform.
 	     */
-	    if (!this->type->qualifier.uniform)
+	    const glsl_type *initializer_type;
+	    if (!this->type->qualifier.uniform) {
 	       result = do_assignment(&initializer_instructions, state,
 				      lhs, rhs,
 				      this->get_location());
+	       initializer_type = result->type;
+	    } else
+	       initializer_type = rhs->type;
+
+	    /* If the declared variable is an unsized array, it must inherrit
+	     * its full type from the initializer.  A declaration such as
+	     *
+	     *     uniform float a[] = float[](1.0, 2.0, 3.0, 3.0);
+	     *
+	     * becomes
+	     *
+	     *     uniform float a[4] = float[](1.0, 2.0, 3.0, 3.0);
+	     *
+	     * The assignment generated in the if-statement (below) will also
+	     * automatically handle this case for non-uniforms.
+	     *
+	     * If the declared variable is not an array, the types must
+	     * already match exactly.  As a result, the type assignment
+	     * here can be done unconditionally.  For non-uniforms the call
+	     * to do_assignment can change the type of the initializer (via
+	     * the implicit conversion rules).  For uniforms the initializer
+	     * must be a constant expression, and the type of that expression
+	     * was validated above.
+	     */
+	    var->type = initializer_type;
+
 	    var->read_only = temp;
 	 }
       }
@@ -2021,7 +2135,7 @@ ast_declarator_list::hir(exec_list *instructions,
 	    earlier->type = var->type;
 	    delete var;
 	    var = NULL;
-	 } else if (state->extensions->ARB_fragment_coord_conventions
+	 } else if (state->ARB_fragment_coord_conventions_enable
 		    && strcmp(var->name, "gl_FragCoord") == 0
 		    && earlier->type == var->type
 		    && earlier->mode == var->mode) {
@@ -2447,8 +2561,7 @@ ast_jump_statement::hir(exec_list *instructions,
 			     state->current_function->function_name());
 	 }
 
-	 ir_expression *const ret = (ir_expression *)
-	    opt_return_value->hir(instructions, state);
+	 ir_rvalue *const ret = opt_return_value->hir(instructions, state);
 	 assert(ret != NULL);
 
 	 /* Implicit conversions are not allowed for return values. */

src/glsl/builtin_types.h

@@ -273,9 +273,9 @@ const glsl_type glsl_type::builtin_ARB_texture_rectangle_types[] = {
 const glsl_type glsl_type::builtin_EXT_texture_array_types[] = {
    glsl_type(GL_SAMPLER_1D_ARRAY,
 	     GLSL_SAMPLER_DIM_1D, 0, 1, GLSL_TYPE_FLOAT, "sampler1DArray"),
-   glsl_type(GL_SAMPLER_1D_ARRAY_SHADOW,
-	     GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_FLOAT, "sampler2DArray"),
    glsl_type(GL_SAMPLER_2D_ARRAY,
+	     GLSL_SAMPLER_DIM_2D, 0, 1, GLSL_TYPE_FLOAT, "sampler2DArray"),
+   glsl_type(GL_SAMPLER_1D_ARRAY_SHADOW,
 	     GLSL_SAMPLER_DIM_1D, 1, 1, GLSL_TYPE_FLOAT, "sampler1DArrayShadow"),
    glsl_type(GL_SAMPLER_2D_ARRAY_SHADOW,
 	     GLSL_SAMPLER_DIM_2D, 1, 1, GLSL_TYPE_FLOAT, "sampler2DArrayShadow"),

src/glsl/builtins/ir/smoothstep

@@ -5,40 +5,26 @@
        (declare (in) float edge1)
        (declare (in) float x))
      ((declare () float t)
-
       (assign (constant bool (1)) (x) (var_ref t)
               (expression float max
 	                  (expression float min
 	                              (expression float / (expression float - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
 	                              (constant float (1.0)))
 	                  (constant float (0.0))))
-      (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (var_ref t))))))))
-
+      (return (expression float * (var_ref t) (expression float * (var_ref t) (expression float - (constant float (3.0)) (expression float * (constant float (2.0)) (var_ref t))))))))
    (signature vec2
      (parameters
        (declare (in) float edge0)
        (declare (in) float edge1)
        (declare (in) vec2 x))
      ((declare () vec2 t)
-      (declare () vec2 retval)
-
-      (assign (constant bool (1)) (x) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
+      (assign (constant bool (1)) (xy) (var_ref t)
+              (expression vec2 max
+	                  (expression vec2 min
+	                              (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
 	                              (constant float (1.0)))
 	                  (constant float (0.0))))
-      (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))
-
-      (assign (constant bool (1)) (y) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))
-      (return (var_ref retval))
-      ))
+      (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t))))))))
 
    (signature vec3
      (parameters
@@ -46,33 +32,13 @@
        (declare (in) float edge1)
        (declare (in) vec3 x))
      ((declare () vec3 t)
-      (declare () vec3 retval)
-
-      (assign (constant bool (1)) (x) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
+      (assign (constant bool (1)) (xyz) (var_ref t)
+              (expression vec3 max
+	                  (expression vec3 min
+	                              (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
 	                              (constant float (1.0)))
 	                  (constant float (0.0))))
-      (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))
-
-      (assign (constant bool (1)) (y) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))
-
-      (assign (constant bool (1)) (z) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t)))))))
-      (return (var_ref retval))
-      ))
+      (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t))))))))
 
 
    (signature vec4
@@ -81,73 +47,54 @@
        (declare (in) float edge1)
        (declare (in) vec4 x))
      ((declare () vec4 t)
-      (declare () vec4 retval)
-
-      (assign (constant bool (1)) (x) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz x (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
+      (assign (constant bool (1)) (xyzw) (var_ref t)
+              (expression vec4 max
+	                  (expression vec4 min
+	                              (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
 	                              (constant float (1.0)))
 	                  (constant float (0.0))))
-      (assign (constant bool (1)) (x) (var_ref retval) (expression float * (swiz x (var_ref t)) (expression float * (swiz x (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz x (var_ref t)))))))
-
-      (assign (constant bool (1)) (y) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz y (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (y) (var_ref retval) (expression float * (swiz y (var_ref t)) (expression float * (swiz y (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz y (var_ref t)))))))
-
-      (assign (constant bool (1)) (z) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz z (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (z) (var_ref retval) (expression float * (swiz z (var_ref t)) (expression float * (swiz z (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz z (var_ref t)))))))
-
-      (assign (constant bool (1)) (w) (var_ref t)
-              (expression float max
-	                  (expression float min
-	                              (expression float / (expression float - (swiz w (var_ref x)) (var_ref edge0)) (expression float - (var_ref edge1) (var_ref edge0)))
-	                              (constant float (1.0)))
-	                  (constant float (0.0))))
-      (assign (constant bool (1)) (w) (var_ref retval) (expression float * (swiz w (var_ref t)) (expression float * (swiz w (var_ref t)) (expression float - (constant float (3.000000)) (expression float * (constant float (2.000000)) (swiz w (var_ref t)))))))
-      (return (var_ref retval))
-      ))
+      (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t))))))))
 
    (signature vec2
      (parameters
        (declare (in) vec2 edge0)
        (declare (in) vec2 edge1)
        (declare (in) vec2 x))
-     ((return (expression vec2 max
+     ((declare () vec2 t)
+      (assign (constant bool (1)) (xy) (var_ref t)
+              (expression vec2 max
                           (expression vec2 min
                                       (expression vec2 / (expression vec2 - (var_ref x) (var_ref edge0)) (expression vec2 - (var_ref edge1) (var_ref edge0)))
-                                      (constant vec2 (1.0 1.0)))
-                          (constant vec2 (0.0 0.0))))))
+                                      (constant float (1.0)))
+                          (constant float (0.0))))
+      (return (expression vec2 * (var_ref t) (expression vec2 * (var_ref t) (expression vec2 - (constant float (3.0)) (expression vec2 * (constant float (2.0)) (var_ref t))))))))
 
    (signature vec3
      (parameters
        (declare (in) vec3 edge0)
        (declare (in) vec3 edge1)
        (declare (in) vec3 x))
-     ((return (expression vec3 max
+     ((declare () vec3 t)
+      (assign (constant bool (1)) (xyz) (var_ref t)
+              (expression vec3 max
                           (expression vec3 min
                                       (expression vec3 / (expression vec3 - (var_ref x) (var_ref edge0)) (expression vec3 - (var_ref edge1) (var_ref edge0)))
-                                      (constant vec3 (1.0 1.0 1.0)))
-                          (constant vec3 (0.0 0.0 0.0))))))
+                                      (constant float (1.0)))
+                          (constant float (0.0))))
+      (return (expression vec3 * (var_ref t) (expression vec3 * (var_ref t) (expression vec3 - (constant float (3.0)) (expression vec3 * (constant float (2.0)) (var_ref t))))))))
 
    (signature vec4
      (parameters
        (declare (in) vec4 edge0)
        (declare (in) vec4 edge1)
        (declare (in) vec4 x))
-     ((return (expression vec4 max
+     ((declare () vec4 t)
+      (assign (constant bool (1)) (xyzw) (var_ref t)
+              (expression vec4 max
                           (expression vec4 min
                                       (expression vec4 / (expression vec4 - (var_ref x) (var_ref edge0)) (expression vec4 - (var_ref edge1) (var_ref edge0)))
-                                      (constant vec4 (1.0 1.0 1.0 1.0)))
-                          (constant vec4 (0.0 0.0 0.0 0.0))))))
+                                      (constant float (1.0)))
+                          (constant float (0.0))))
+      (return (expression vec4 * (var_ref t) (expression vec4 * (var_ref t) (expression vec4 - (constant float (3.0)) (expression vec4 * (constant float (2.0)) (var_ref t))))))))
 ))
 

src/glsl/builtins/tools/generate_builtins.py

@@ -25,13 +25,25 @@ def get_builtin_definitions():
     return fs
 
 def stringify(s):
+    # Work around MSVC's 65535 byte limit by outputting an array of characters
+    # rather than actual string literals.
+    if len(s) >= 65535:
+        #t = "/* Warning: length " + repr(len(s)) + " too large */\n"
+	t = ""
+        for c in re.sub('\s\s+', ' ', s):
+            if c == '\n':
+	        t += '\n'
+            else:
+	        t += "'" + c + "',"
+        return '{' + t[:-1] + '}'
+
     t = s.replace('\\', '\\\\').replace('"', '\\"').replace('\n', '\\n"\n   "')
     return '   "' + t + '"\n'
 
 def write_function_definitions():
     fs = get_builtin_definitions()
     for k, v in sorted(fs.iteritems()):
-        print 'static const char *builtin_' + k + ' ='
+        print 'static const char builtin_' + k + '[] ='
         print stringify(v), ';'
 
 def run_compiler(args):
@@ -64,7 +76,7 @@ def write_profile(filename, profile):
     # clutter the diff output.
     proto_ir = re.sub(r'@0x[0-9a-f]+', '', proto_ir);
 
-    print 'static const char *prototypes_for_' + profile + ' ='
+    print 'static const char prototypes_for_' + profile + '[] ='
     print stringify(proto_ir), ';'
 
     # Print a table of all the functions (not signatures) referenced.

src/glsl/glcpp/glcpp-lex.c

@@ -1109,7 +1109,7 @@ case 4:
 /* rule 4 can match eol */
 YY_RULE_SETUP
 #line 85 "glcpp/glcpp-lex.l"
-{ yylineno++; yycolumn = 0; }
+{ yylineno++; yycolumn = 0; return NEWLINE; }
 	YY_BREAK
 case 5:
 YY_RULE_SETUP
@@ -1120,7 +1120,7 @@ case 6:
 /* rule 6 can match eol */
 YY_RULE_SETUP
 #line 87 "glcpp/glcpp-lex.l"
-{ yylineno++; yycolumn = 0; }
+{ yylineno++; yycolumn = 0; return NEWLINE; }
 	YY_BREAK
 case 7:
 YY_RULE_SETUP

src/glsl/glcpp/glcpp-lex.l

@@ -82,9 +82,9 @@ HEXADECIMAL_INTEGER	0[xX][0-9a-fA-F]+[uU]?
 	/* Multi-line comments */
 "/*"                    { yy_push_state(COMMENT, yyscanner); }
 <COMMENT>[^*\n]*
-<COMMENT>[^*\n]*\n      { yylineno++; yycolumn = 0; }
+<COMMENT>[^*\n]*\n      { yylineno++; yycolumn = 0; return NEWLINE; }
 <COMMENT>"*"+[^*/\n]*
-<COMMENT>"*"+[^*/\n]*\n { yylineno++; yycolumn = 0; }
+<COMMENT>"*"+[^*/\n]*\n { yylineno++; yycolumn = 0; return NEWLINE; }
 <COMMENT>"*"+"/"        {
 	yy_pop_state(yyscanner);
 	if (yyextra->space_tokens)

src/glsl/glcpp/glcpp-parse.h

@@ -1,9 +1,10 @@
-/* A Bison parser, made by GNU Bison 2.4.3.  */
+
+/* A Bison parser, made by GNU Bison 2.4.1.  */
 
 /* Skeleton interface for Bison's Yacc-like parsers in C
    
-      Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
-   2009, 2010 Free Software Foundation, Inc.
+      Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006
+   Free Software Foundation, Inc.
    
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by

src/glsl/glcpp/glcpp-parse.y

@@ -320,6 +320,14 @@ control_line:
 		if ($2 == 100)
 			add_builtin_define (parser, "GL_ES", 1);
 
+		/* Currently, all ES2 implementations support highp in the
+		 * fragment shader, so we always define this macro in ES2.
+		 * If we ever get a driver that doesn't support highp, we'll
+		 * need to add a flag to the gl_context and check that here.
+		 */
+		if ($2 >= 130 || $2 == 100)
+			add_builtin_define (parser, "GL_FRAGMENT_PRECISION_HIGH", 1);
+
 		glcpp_printf(parser->output, "#version %" PRIiMAX, $2);
 	}
 |	HASH NEWLINE
@@ -464,7 +472,6 @@ conditional_token:
 conditional_tokens:
 	/* Exactly the same as pp_tokens, but using conditional_token */
 	conditional_token {
-		parser->space_tokens = 1;
 		$$ = _token_list_create (parser);
 		_token_list_append ($$, $1);
 		talloc_unlink (parser, $1);

src/glsl/glcpp/glcpp.c

@@ -29,9 +29,17 @@
 #include <errno.h>
 #include "glcpp.h"
 #include "main/mtypes.h"
+#include "main/shaderobj.h"
 
 extern int yydebug;
 
+void
+_mesa_reference_shader(GLcontext *ctx, struct gl_shader **ptr,
+                       struct gl_shader *sh)
+{
+   *ptr = sh;
+}
+
 /* Read from fd until EOF and return a string of everything read.
  */
 static char *

src/glsl/glsl_lexer.lpp

@@ -22,6 +22,7 @@
  * DEALINGS IN THE SOFTWARE.
  */
 #include <ctype.h>
+#include "strtod.h"
 #include "ast.h"
 #include "glsl_parser_extras.h"
 #include "glsl_parser.h"
@@ -243,7 +244,7 @@ layout		{
 \<\<=		return LEFT_ASSIGN;
 >>=		return RIGHT_ASSIGN;
 &=		return AND_ASSIGN;
-^=		return XOR_ASSIGN;
+"^="		return XOR_ASSIGN;
 \|=		return OR_ASSIGN;
 -=		return SUB_ASSIGN;
 
@@ -261,23 +262,23 @@ layout		{
 			}
 
 [0-9]+\.[0-9]+([eE][+-]?[0-9]+)?[fF]?	{
-			    yylval->real = strtod(yytext, NULL);
+			    yylval->real = glsl_strtod(yytext, NULL);
 			    return FLOATCONSTANT;
 			}
 \.[0-9]+([eE][+-]?[0-9]+)?[fF]?		{
-			    yylval->real = strtod(yytext, NULL);
+			    yylval->real = glsl_strtod(yytext, NULL);
 			    return FLOATCONSTANT;
 			}
 [0-9]+\.([eE][+-]?[0-9]+)?[fF]?		{
-			    yylval->real = strtod(yytext, NULL);
+			    yylval->real = glsl_strtod(yytext, NULL);
 			    return FLOATCONSTANT;
 			}
 [0-9]+[eE][+-]?[0-9]+[fF]?		{
-			    yylval->real = strtod(yytext, NULL);
+			    yylval->real = glsl_strtod(yytext, NULL);
 			    return FLOATCONSTANT;
 			}
 [0-9]+[fF]		{
-			    yylval->real = strtod(yytext, NULL);
+			    yylval->real = glsl_strtod(yytext, NULL);
 			    return FLOATCONSTANT;
 			}
 

src/glsl/glsl_parser.cpp

@@ -1,9 +1,9 @@
-/* A Bison parser, made by GNU Bison 2.4.3.  */
+/* A Bison parser, made by GNU Bison 2.4.2.  */
 
 /* Skeleton implementation for Bison's Yacc-like parsers in C
    
-      Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
-   2009, 2010 Free Software Foundation, Inc.
+      Copyright (C) 1984, 1989-1990, 2000-2006, 2009-2010 Free Software
+   Foundation, Inc.
    
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -45,7 +45,7 @@
 #define YYBISON 1
 
 /* Bison version.  */
-#define YYBISON_VERSION "2.4.3"
+#define YYBISON_VERSION "2.4.2"
 
 /* Skeleton name.  */
 #define YYSKELETON_NAME "yacc.c"
@@ -2621,7 +2621,7 @@ YYLTYPE yylloc;
     YYLTYPE *yylsp;
 
     /* The locations where the error started and ended.  */
-    YYLTYPE yyerror_range[3];
+    YYLTYPE yyerror_range[2];
 
     YYSIZE_T yystacksize;
 
@@ -5084,7 +5084,7 @@ yyerrlab:
 #endif
     }
 
-  yyerror_range[1] = yylloc;
+  yyerror_range[0] = yylloc;
 
   if (yyerrstatus == 3)
     {
@@ -5121,7 +5121,7 @@ yyerrorlab:
   if (/*CONSTCOND*/ 0)
      goto yyerrorlab;
 
-  yyerror_range[1] = yylsp[1-yylen];
+  yyerror_range[0] = yylsp[1-yylen];
   /* Do not reclaim the symbols of the rule which action triggered
      this YYERROR.  */
   YYPOPSTACK (yylen);
@@ -5155,7 +5155,7 @@ yyerrlab1:
       if (yyssp == yyss)
 	YYABORT;
 
-      yyerror_range[1] = *yylsp;
+      yyerror_range[0] = *yylsp;
       yydestruct ("Error: popping",
 		  yystos[yystate], yyvsp, yylsp, state);
       YYPOPSTACK (1);
@@ -5165,10 +5165,10 @@ yyerrlab1:
 
   *++yyvsp = yylval;
 
-  yyerror_range[2] = yylloc;
+  yyerror_range[1] = yylloc;
   /* Using YYLLOC is tempting, but would change the location of
      the lookahead.  YYLOC is available though.  */
-  YYLLOC_DEFAULT (yyloc, yyerror_range, 2);
+  YYLLOC_DEFAULT (yyloc, (yyerror_range - 1), 2);
   *++yylsp = yyloc;
 
   /* Shift the error token.  */

src/glsl/glsl_parser.h

@@ -1,9 +1,9 @@
-/* A Bison parser, made by GNU Bison 2.4.3.  */
+/* A Bison parser, made by GNU Bison 2.4.2.  */
 
 /* Skeleton interface for Bison's Yacc-like parsers in C
    
-      Copyright (C) 1984, 1989, 1990, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
-   2009, 2010 Free Software Foundation, Inc.
+      Copyright (C) 1984, 1989-1990, 2000-2006, 2009-2010 Free Software
+   Foundation, Inc.
    
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by

src/glsl/glsl_parser_extras.cpp

@@ -702,6 +702,7 @@ do_common_optimization(exec_list *ir, bool linked, unsigned max_unroll_iteration
    }
    progress = do_structure_splitting(ir) || progress;
    progress = do_if_simplification(ir) || progress;
+   progress = do_discard_simplification(ir) || progress;
    progress = do_copy_propagation(ir) || progress;
    if (linked)
       progress = do_dead_code(ir) || progress;

src/glsl/ir.cpp

@@ -176,9 +176,22 @@ ir_assignment::ir_assignment(ir_rvalue *lhs, ir_rvalue *rhs,
 }
 
 
+ir_expression::ir_expression(int op, const struct glsl_type *type,
+			     ir_rvalue *op0)
+{
+   assert(get_num_operands(ir_expression_operation(op)) == 1);
+   this->ir_type = ir_type_expression;
+   this->type = type;
+   this->operation = ir_expression_operation(op);
+   this->operands[0] = op0;
+   this->operands[1] = NULL;
+}
+
 ir_expression::ir_expression(int op, const struct glsl_type *type,
 			     ir_rvalue *op0, ir_rvalue *op1)
 {
+   assert((op1 == NULL) && (get_num_operands(ir_expression_operation(op)) == 1)
+	  || (get_num_operands(ir_expression_operation(op)) == 2));
    this->ir_type = ir_type_expression;
    this->type = type;
    this->operation = ir_expression_operation(op);
@@ -186,81 +199,116 @@ ir_expression::ir_expression(int op, const struct glsl_type *type,
    this->operands[1] = op1;
 }
 
+ir_expression::ir_expression(int op, ir_rvalue *op0)
+{
+   this->ir_type = ir_type_expression;
+
+   this->operation = ir_expression_operation(op);
+   this->operands[0] = op0;
+   this->operands[1] = NULL;
+
+   assert(op <= ir_last_unop);
+
+   switch (this->operation) {
+   case ir_unop_bit_not:
+   case ir_unop_logic_not:
+   case ir_unop_neg:
+   case ir_unop_abs:
+   case ir_unop_sign:
+   case ir_unop_rcp:
+   case ir_unop_rsq:
+   case ir_unop_sqrt:
+   case ir_unop_exp:
+   case ir_unop_log:
+   case ir_unop_exp2:
+   case ir_unop_log2:
+   case ir_unop_trunc:
+   case ir_unop_ceil:
+   case ir_unop_floor:
+   case ir_unop_fract:
+   case ir_unop_cos:
+   case ir_unop_dFdx:
+   case ir_unop_dFdy:
+      this->type = op0->type;
+      break;
+
+   case ir_unop_any:
+      this->type = glsl_type::bool_type;
+      break;
+
+   default:
+      assert(!"not reached: missing automatic type setup for ir_expression");
+      this->type = op0->type;
+      break;
+   }
+}
+
+ir_expression::ir_expression(int op, ir_rvalue *op0, ir_rvalue *op1)
+{
+   this->ir_type = ir_type_expression;
+
+   this->operation = ir_expression_operation(op);
+   this->operands[0] = op0;
+   this->operands[1] = op1;
+
+   assert(op > ir_last_unop);
+
+   switch (this->operation) {
+   case ir_binop_all_equal:
+   case ir_binop_any_nequal:
+      this->type = glsl_type::bool_type;
+      break;
+
+   case ir_binop_add:
+   case ir_binop_sub:
+   case ir_binop_min:
+   case ir_binop_max:
+   case ir_binop_pow:
+   case ir_binop_mul:
+      if (op0->type->is_scalar()) {
+	 this->type = op1->type;
+      } else if (op1->type->is_scalar()) {
+	 this->type = op0->type;
+      } else {
+	 /* FINISHME: matrix types */
+	 assert(!op0->type->is_matrix() && !op1->type->is_matrix());
+	 assert(op0->type == op1->type);
+	 this->type = op0->type;
+      }
+      break;
+
+   case ir_binop_logic_and:
+   case ir_binop_logic_or:
+      if (op0->type->is_scalar()) {
+	 this->type = op1->type;
+      } else if (op1->type->is_scalar()) {
+	 this->type = op0->type;
+      }
+      break;
+
+   case ir_binop_dot:
+      this->type = glsl_type::float_type;
+      break;
+
+   default:
+      assert(!"not reached: missing automatic type setup for ir_expression");
+      this->type = glsl_type::float_type;
+   }
+}
+
 unsigned int
 ir_expression::get_num_operands(ir_expression_operation op)
 {
-/* Update ir_print_visitor.cpp when updating this list. */
-   const int num_operands[] = {
-      1, /* ir_unop_bit_not */
-      1, /* ir_unop_logic_not */
-      1, /* ir_unop_neg */
-      1, /* ir_unop_abs */
-      1, /* ir_unop_sign */
-      1, /* ir_unop_rcp */
-      1, /* ir_unop_rsq */
-      1, /* ir_unop_sqrt */
-      1, /* ir_unop_exp */
-      1, /* ir_unop_log */
-      1, /* ir_unop_exp2 */
-      1, /* ir_unop_log2 */
-      1, /* ir_unop_f2i */
-      1, /* ir_unop_i2f */
-      1, /* ir_unop_f2b */
-      1, /* ir_unop_b2f */
-      1, /* ir_unop_i2b */
-      1, /* ir_unop_b2i */
-      1, /* ir_unop_u2f */
-      1, /* ir_unop_any */
+   assert(op <= ir_last_opcode);
 
-      1, /* ir_unop_trunc */
-      1, /* ir_unop_ceil */
-      1, /* ir_unop_floor */
-      1, /* ir_unop_fract */
+   if (op <= ir_last_unop)
+      return 1;
 
-      1, /* ir_unop_sin */
-      1, /* ir_unop_cos */
+   if (op <= ir_last_binop)
+      return 2;
 
-      1, /* ir_unop_dFdx */
-      1, /* ir_unop_dFdy */
-
-      1, /* ir_unop_noise */
-
-      2, /* ir_binop_add */
-      2, /* ir_binop_sub */
-      2, /* ir_binop_mul */
-      2, /* ir_binop_div */
-      2, /* ir_binop_mod */
-
-      2, /* ir_binop_less */
-      2, /* ir_binop_greater */
-      2, /* ir_binop_lequal */
-      2, /* ir_binop_gequal */
-      2, /* ir_binop_equal */
-      2, /* ir_binop_nequal */
-      2, /* ir_binop_all_equal */
-      2, /* ir_binop_any_nequal */
-
-      2, /* ir_binop_lshift */
-      2, /* ir_binop_rshift */
-      2, /* ir_binop_bit_and */
-      2, /* ir_binop_bit_xor */
-      2, /* ir_binop_bit_or */
-
-      2, /* ir_binop_logic_and */
-      2, /* ir_binop_logic_xor */
-      2, /* ir_binop_logic_or */
-
-      2, /* ir_binop_dot */
-      2, /* ir_binop_cross */
-      2, /* ir_binop_min */
-      2, /* ir_binop_max */
-
-      2, /* ir_binop_pow */
-   };
-
-   assert(sizeof(num_operands) / sizeof(num_operands[0]) == ir_binop_pow + 1);
-
-   return num_operands[op];
+   assert(false);
+   return 0;
 }
 
 static const char *const operator_strs[] = {
@@ -687,7 +735,7 @@ ir_constant::has_value(const ir_constant *c) const
 
    if (this->type->is_array()) {
       for (unsigned i = 0; i < this->type->length; i++) {
-	 if (this->array_elements[i]->has_value(c->array_elements[i]))
+	 if (!this->array_elements[i]->has_value(c->array_elements[i]))
 	    return false;
       }
       return true;

src/glsl/ir.h

@@ -125,6 +125,7 @@ public:
    virtual class ir_if *                as_if()               { return NULL; }
    virtual class ir_swizzle *           as_swizzle()          { return NULL; }
    virtual class ir_constant *          as_constant()         { return NULL; }
+   virtual class ir_discard *           as_discard()          { return NULL; }
    /*@}*/
 
 protected:
@@ -711,6 +712,11 @@ enum ir_expression_operation {
 
    ir_unop_noise,
 
+   /**
+    * A sentinel marking the last of the unary operations.
+    */
+   ir_last_unop = ir_unop_noise,
+
    ir_binop_add,
    ir_binop_sub,
    ir_binop_mul,
@@ -769,13 +775,33 @@ enum ir_expression_operation {
    ir_binop_min,
    ir_binop_max,
 
-   ir_binop_pow
+   ir_binop_pow,
+
+   /**
+    * A sentinel marking the last of the binary operations.
+    */
+   ir_last_binop = ir_binop_pow,
+
+   /**
+    * A sentinel marking the last of all operations.
+    */
+   ir_last_opcode = ir_last_binop
 };
 
 class ir_expression : public ir_rvalue {
 public:
+   /**
+    * Constructor for unary operation expressions
+    */
+   ir_expression(int op, const struct glsl_type *type, ir_rvalue *);
+   ir_expression(int op, ir_rvalue *);
+
+   /**
+    * Constructor for binary operation expressions
+    */
    ir_expression(int op, const struct glsl_type *type,
 		 ir_rvalue *, ir_rvalue *);
+   ir_expression(int op, ir_rvalue *op0, ir_rvalue *op1);
 
    virtual ir_expression *as_expression()
    {
@@ -1044,6 +1070,11 @@ public:
 
    virtual ir_visitor_status accept(ir_hierarchical_visitor *);
 
+   virtual ir_discard *as_discard()
+   {
+      return this;
+   }
+
    ir_rvalue *condition;
 };
 /*@}*/

src/glsl/ir_algebraic.cpp

@@ -187,8 +187,8 @@ ir_algebraic_visitor::reassociate_constant(ir_expression *ir1, int const_index,
 
    /* Don't want to even think about matrices. */
    if (ir1->operands[0]->type->is_matrix() ||
-       ir1->operands[0]->type->is_matrix() ||
-       ir2->operands[1]->type->is_matrix() ||
+       ir1->operands[1]->type->is_matrix() ||
+       ir2->operands[0]->type->is_matrix() ||
        ir2->operands[1]->type->is_matrix())
       return false;
 

src/glsl/ir_constant_expression.cpp

@@ -54,6 +54,9 @@ dot(ir_constant *op0, ir_constant *op1)
 ir_constant *
 ir_expression::constant_expression_value()
 {
+   if (this->type->is_error())
+      return NULL;
+
    ir_constant *op[2] = { NULL, NULL };
    ir_constant_data data;
 
@@ -563,63 +566,75 @@ ir_expression::constant_expression_value()
       break;
 
    case ir_binop_less:
-      switch (op[0]->type->base_type) {
-      case GLSL_TYPE_UINT:
-	 data.b[0] = op[0]->value.u[0] < op[1]->value.u[0];
-	 break;
-      case GLSL_TYPE_INT:
-	 data.b[0] = op[0]->value.i[0] < op[1]->value.i[0];
-	 break;
-      case GLSL_TYPE_FLOAT:
-	 data.b[0] = op[0]->value.f[0] < op[1]->value.f[0];
-	 break;
-      default:
-	 assert(0);
+      assert(op[0]->type == op[1]->type);
+      for (unsigned c = 0; c < op[0]->type->components(); c++) {
+	 switch (op[0]->type->base_type) {
+	 case GLSL_TYPE_UINT:
+	    data.b[0] = op[0]->value.u[0] < op[1]->value.u[0];
+	    break;
+	 case GLSL_TYPE_INT:
+	    data.b[0] = op[0]->value.i[0] < op[1]->value.i[0];
+	    break;
+	 case GLSL_TYPE_FLOAT:
+	    data.b[0] = op[0]->value.f[0] < op[1]->value.f[0];
+	    break;
+	 default:
+	    assert(0);
+	 }
       }
       break;
    case ir_binop_greater:
-      switch (op[0]->type->base_type) {
-      case GLSL_TYPE_UINT:
-	 data.b[0] = op[0]->value.u[0] > op[1]->value.u[0];
-	 break;
-      case GLSL_TYPE_INT:
-	 data.b[0] = op[0]->value.i[0] > op[1]->value.i[0];
-	 break;
-      case GLSL_TYPE_FLOAT:
-	 data.b[0] = op[0]->value.f[0] > op[1]->value.f[0];
-	 break;
-      default:
-	 assert(0);
+      assert(op[0]->type == op[1]->type);
+      for (unsigned c = 0; c < op[0]->type->components(); c++) {
+	 switch (op[0]->type->base_type) {
+	 case GLSL_TYPE_UINT:
+	    data.b[c] = op[0]->value.u[c] > op[1]->value.u[c];
+	    break;
+	 case GLSL_TYPE_INT:
+	    data.b[c] = op[0]->value.i[c] > op[1]->value.i[c];
+	    break;
+	 case GLSL_TYPE_FLOAT:
+	    data.b[c] = op[0]->value.f[c] > op[1]->value.f[c];
+	    break;
+	 default:
+	    assert(0);
+	 }
       }
       break;
    case ir_binop_lequal:
-      switch (op[0]->type->base_type) {
-      case GLSL_TYPE_UINT:
-	 data.b[0] = op[0]->value.u[0] <= op[1]->value.u[0];
-	 break;
-      case GLSL_TYPE_INT:
-	 data.b[0] = op[0]->value.i[0] <= op[1]->value.i[0];
-	 break;
-      case GLSL_TYPE_FLOAT:
-	 data.b[0] = op[0]->value.f[0] <= op[1]->value.f[0];
-	 break;
-      default:
-	 assert(0);
+      assert(op[0]->type == op[1]->type);
+      for (unsigned c = 0; c < op[0]->type->components(); c++) {
+	 switch (op[0]->type->base_type) {
+	 case GLSL_TYPE_UINT:
+	    data.b[0] = op[0]->value.u[0] <= op[1]->value.u[0];
+	    break;
+	 case GLSL_TYPE_INT:
+	    data.b[0] = op[0]->value.i[0] <= op[1]->value.i[0];
+	    break;
+	 case GLSL_TYPE_FLOAT:
+	    data.b[0] = op[0]->value.f[0] <= op[1]->value.f[0];
+	    break;
+	 default:
+	    assert(0);
+	 }
       }
       break;
    case ir_binop_gequal:
-      switch (op[0]->type->base_type) {
-      case GLSL_TYPE_UINT:
-	 data.b[0] = op[0]->value.u[0] >= op[1]->value.u[0];
-	 break;
-      case GLSL_TYPE_INT:
-	 data.b[0] = op[0]->value.i[0] >= op[1]->value.i[0];
-	 break;
-      case GLSL_TYPE_FLOAT:
-	 data.b[0] = op[0]->value.f[0] >= op[1]->value.f[0];
-	 break;
-      default:
-	 assert(0);
+      assert(op[0]->type == op[1]->type);
+      for (unsigned c = 0; c < op[0]->type->components(); c++) {
+	 switch (op[0]->type->base_type) {
+	 case GLSL_TYPE_UINT:
+	    data.b[0] = op[0]->value.u[0] >= op[1]->value.u[0];
+	    break;
+	 case GLSL_TYPE_INT:
+	    data.b[0] = op[0]->value.i[0] >= op[1]->value.i[0];
+	    break;
+	 case GLSL_TYPE_FLOAT:
+	    data.b[0] = op[0]->value.f[0] >= op[1]->value.f[0];
+	    break;
+	 default:
+	    assert(0);
+	 }
       }
       break;
    case ir_binop_equal:

src/glsl/ir_constant_propagation.cpp

@@ -1,5 +1,5 @@
 /*
- * Constantright © 2010 Intel Corporation
+ * Copyright © 2010 Intel Corporation
  *
  * Permission is hereby granted, free of charge, to any person obtaining a
  * constant of this software and associated documentation files (the "Software"),

src/glsl/ir_dead_functions.cpp

@@ -125,6 +125,7 @@ do_dead_functions(exec_list *instructions)
 
       if (!entry->used) {
 	 entry->signature->remove();
+	 delete entry->signature;
 	 progress = true;
       }
       delete(entry);
@@ -143,6 +144,7 @@ do_dead_functions(exec_list *instructions)
 	  * symbol table should be OK.
 	  */
 	 func->remove();
+	 delete func;
 	 progress = true;
       }
    }

src/glsl/ir_function_can_inline.cpp

@@ -60,6 +60,8 @@ can_inline(ir_call *call)
 {
    ir_function_can_inline_visitor v;
    const ir_function_signature *callee = call->get_callee();
+   if (!callee->is_defined)
+      return false;
 
    v.run((exec_list *) &callee->body);
 

src/glsl/ir_lower_jumps.cpp

@@ -36,7 +36,6 @@ enum jump_strength
    strength_continue,
    strength_break,
    strength_return,
-   strength_discard
 };
 
 struct block_record
@@ -202,8 +201,6 @@ struct ir_lower_jumps_visitor : public ir_control_flow_visitor {
 
    virtual void visit(class ir_discard * ir)
    {
-      truncate_after_instruction(ir);
-      this->block.min_strength = strength_discard;
    }
 
    enum jump_strength get_jump_strength(ir_instruction* ir)
@@ -217,8 +214,6 @@ struct ir_lower_jumps_visitor : public ir_control_flow_visitor {
             return strength_continue;
       } else if(ir->ir_type == ir_type_return)
          return strength_return;
-      else if(ir->ir_type == ir_type_discard)
-         return strength_discard;
       else
          return strength_none;
    }
@@ -253,9 +248,6 @@ struct ir_lower_jumps_visitor : public ir_control_flow_visitor {
          else
             lower = lower_sub_return;
          break;
-      case strength_discard:
-         lower = false; /* probably nothing needs this lowered */
-         break;
       }
       return lower;
    }
@@ -313,9 +305,8 @@ retry: /* we get here if we put code after the if inside a branch */
             /* FINISHME: unify returns with identical expressions */
             else if(jump_strengths[0] == strength_return && this->function.signature->return_type->is_void())
                ir->insert_after(new(ir) ir_return(NULL));
-            /* FINISHME: unify discards */
-            else
-               unify = false;
+	    else
+	       unify = false;
 
             if(unify) {
                jumps[0]->remove();
@@ -490,7 +481,11 @@ lower_continue:
       if(this->loop.may_set_return_flag) {
          assert(this->function.return_flag);
          ir_if* return_if = new(ir) ir_if(new(ir) ir_dereference_variable(this->function.return_flag));
-         return_if->then_instructions.push_tail(new(ir) ir_loop_jump(saved_loop.loop ? ir_loop_jump::jump_break : ir_loop_jump::jump_continue));
+         saved_loop.may_set_return_flag = true;
+         if(saved_loop.loop)
+            return_if->then_instructions.push_tail(new(ir) ir_loop_jump(ir_loop_jump::jump_break));
+         else
+            move_outer_block_inside(ir, &return_if->else_instructions);
          ir->insert_after(return_if);
       }
 

src/glsl/ir_optimization.h

@@ -45,6 +45,7 @@ bool do_explog_to_explog2(exec_list *instructions);
 bool do_function_inlining(exec_list *instructions);
 bool do_lower_jumps(exec_list *instructions, bool pull_out_jumps = true, bool lower_sub_return = true, bool lower_main_return = false, bool lower_continue = false, bool lower_break = false);
 bool do_if_simplification(exec_list *instructions);
+bool do_discard_simplification(exec_list *instructions);
 bool do_if_to_cond_assign(exec_list *instructions);
 bool do_mat_op_to_vec(exec_list *instructions);
 bool do_mod_to_fract(exec_list *instructions);
@@ -55,6 +56,7 @@ bool do_swizzle_swizzle(exec_list *instructions);
 bool do_tree_grafting(exec_list *instructions);
 bool do_vec_index_to_cond_assign(exec_list *instructions);
 bool do_vec_index_to_swizzle(exec_list *instructions);
+bool lower_discard(exec_list *instructions);
 bool lower_noise(exec_list *instructions);
 bool lower_variable_index_to_cond_assign(exec_list *instructions,
     bool lower_input, bool lower_output, bool lower_temp, bool lower_uniform);

src/glsl/ir_set_program_inouts.cpp

@@ -66,7 +66,7 @@ public:
 };
 
 static void
-mark(struct gl_program *prog, ir_variable *var, int index)
+mark(struct gl_program *prog, ir_variable *var, int offset, int len)
 {
    /* As of GLSL 1.20, varyings can only be floats, floating-point
     * vectors or matrices, or arrays of them.  For Mesa programs using
@@ -75,25 +75,12 @@ mark(struct gl_program *prog, ir_variable *var, int index)
     * something doing a more clever packing would use something other
     * than InputsRead/OutputsWritten.
     */
-   const glsl_type *element_type;
-   int element_size;
 
-   if (var->type->is_array())
-      element_type = var->type->fields.array;
-   else
-      element_type = var->type;
-
-   if (element_type->is_matrix())
-      element_size = element_type->matrix_columns;
-   else
-      element_size = 1;
-
-   index *= element_size;
-   for (int i = 0; i < element_size; i++) {
+   for (int i = 0; i < len; i++) {
       if (var->mode == ir_var_in)
-	 prog->InputsRead |= BITFIELD64_BIT(var->location + index + i);
+	 prog->InputsRead |= BITFIELD64_BIT(var->location + offset + i);
       else
-	 prog->OutputsWritten |= BITFIELD64_BIT(var->location + index + i);
+	 prog->OutputsWritten |= BITFIELD64_BIT(var->location + offset + i);
    }
 }
 
@@ -106,10 +93,11 @@ ir_set_program_inouts_visitor::visit(ir_dereference_variable *ir)
 
    if (ir->type->is_array()) {
       for (unsigned int i = 0; i < ir->type->length; i++) {
-	 mark(this->prog, ir->var, i);
+	 mark(this->prog, ir->var, i,
+	      ir->type->length * ir->type->fields.array->matrix_columns);
       }
    } else {
-      mark(this->prog, ir->var, 0);
+      mark(this->prog, ir->var, 0, ir->type->matrix_columns);
    }
 
    return visit_continue;
@@ -128,7 +116,14 @@ ir_set_program_inouts_visitor::visit_enter(ir_dereference_array *ir)
       var = (ir_variable *)hash_table_find(this->ht, deref_var->var);
 
    if (index && var) {
-      mark(this->prog, var, index->value.i[0]);
+      int width = 1;
+
+      if (deref_var->type->is_array() &&
+	  deref_var->type->fields.array->is_matrix()) {
+	 width = deref_var->type->fields.array->matrix_columns;
+      }
+
+      mark(this->prog, var, index->value.i[0] * width, width);
       return visit_continue_with_parent;
    }
 

src/glsl/linker.cpp

@@ -80,6 +80,10 @@ extern "C" {
 #include "linker.h"
 #include "ir_optimization.h"
 
+extern "C" {
+#include "main/shaderobj.h"
+}
+
 /**
  * Visitor that determines whether or not a variable is ever written.
  */
@@ -321,6 +325,9 @@ cross_validate_globals(struct gl_shader_program *prog,
     */
    glsl_symbol_table variables;
    for (unsigned i = 0; i < num_shaders; i++) {
+      if (shader_list[i] == NULL)
+	 continue;
+
       foreach_list(node, shader_list[i]->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
@@ -353,8 +360,12 @@ cross_validate_globals(struct gl_shader_program *prog,
 		   && (var->type->fields.array == existing->type->fields.array)
 		   && ((var->type->length == 0)
 		       || (existing->type->length == 0))) {
-		  if (existing->type->length == 0)
+		  if (existing->type->length == 0) {
 		     existing->type = var->type;
+		     existing->max_array_access =
+			MAX2(existing->max_array_access,
+			     var->max_array_access);
+		  }
 	       } else {
 		  linker_error_printf(prog, "%s `%s' declared as type "
 				      "`%s' and type `%s'\n",
@@ -391,6 +402,13 @@ cross_validate_globals(struct gl_shader_program *prog,
 		  existing->constant_value =
 		     var->constant_value->clone(talloc_parent(existing), NULL);
 	    }
+
+	    if (existing->invariant != var->invariant) {
+	       linker_error_printf(prog, "declarations for %s `%s' have "
+	                           "mismatching invariant qualifiers\n",
+	                           mode_string(var), var->name);
+	       return false;
+	    }
 	 } else
 	    variables.add_variable(var->name, var);
       }
@@ -407,7 +425,7 @@ bool
 cross_validate_uniforms(struct gl_shader_program *prog)
 {
    return cross_validate_globals(prog, prog->_LinkedShaders,
-				 prog->_NumLinkedShaders, true);
+				 MESA_SHADER_TYPES, true);
 }
 
 
@@ -456,14 +474,35 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
 	 /* Check that the types match between stages.
 	  */
 	 if (input->type != output->type) {
-	    linker_error_printf(prog,
-				"%s shader output `%s' declared as "
-				"type `%s', but %s shader input declared "
-				"as type `%s'\n",
-				producer_stage, output->name,
-				output->type->name,
-				consumer_stage, input->type->name);
-	    return false;
+	    /* There is a bit of a special case for gl_TexCoord.  This
+	     * built-in is unsized by default.  Appliations that variable
+	     * access it must redeclare it with a size.  There is some
+	     * language in the GLSL spec that implies the fragment shader
+	     * and vertex shader do not have to agree on this size.  Other
+	     * driver behave this way, and one or two applications seem to
+	     * rely on it.
+	     *
+	     * Neither declaration needs to be modified here because the array
+	     * sizes are fixed later when update_array_sizes is called.
+	     *
+	     * From page 48 (page 54 of the PDF) of the GLSL 1.10 spec:
+	     *
+	     *     "Unlike user-defined varying variables, the built-in
+	     *     varying variables don't have a strict one-to-one
+	     *     correspondence between the vertex language and the
+	     *     fragment language."
+	     */
+	    if (!output->type->is_array()
+		|| (strncmp("gl_", output->name, 3) != 0)) {
+	       linker_error_printf(prog,
+				   "%s shader output `%s' declared as "
+				   "type `%s', but %s shader input declared "
+				   "as type `%s'\n",
+				   producer_stage, output->name,
+				   output->type->name,
+				   consumer_stage, input->type->name);
+	       return false;
+	    }
 	 }
 
 	 /* Check that all of the qualifiers match between stages.
@@ -706,7 +745,8 @@ get_main_function_signature(gl_shader *sh)
  * shader is returned.
  */
 static struct gl_shader *
-link_intrastage_shaders(GLcontext *ctx,
+link_intrastage_shaders(void *mem_ctx,
+			GLcontext *ctx,
 			struct gl_shader_program *prog,
 			struct gl_shader **shader_list,
 			unsigned num_shaders)
@@ -780,9 +820,9 @@ link_intrastage_shaders(GLcontext *ctx,
       return NULL;
    }
 
-   gl_shader *const linked = ctx->Driver.NewShader(NULL, 0, main->Type);
+   gl_shader *linked = ctx->Driver.NewShader(NULL, 0, main->Type);
    linked->ir = new(linked) exec_list;
-   clone_ir_list(linked, linked->ir, main->ir);
+   clone_ir_list(mem_ctx, linked->ir, main->ir);
 
    populate_symbol_table(linked);
 
@@ -827,10 +867,40 @@ link_intrastage_shaders(GLcontext *ctx,
 
    assert(idx == num_linking_shaders);
 
-   link_function_calls(prog, linked, linking_shaders, num_linking_shaders);
+   if (!link_function_calls(prog, linked, linking_shaders,
+			    num_linking_shaders)) {
+      ctx->Driver.DeleteShader(ctx, linked);
+      linked = NULL;
+   }
 
    free(linking_shaders);
 
+   /* Make a pass over all global variables to ensure that arrays with
+    * unspecified sizes have a size specified.  The size is inferred from the
+    * max_array_access field.
+    */
+   if (linked != NULL) {
+      foreach_list(node, linked->ir) {
+	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
+
+	 if (var == NULL)
+	    continue;
+
+	 if ((var->mode != ir_var_auto) && (var->mode != ir_var_temporary))
+	    continue;
+
+	 if (!var->type->is_array() || (var->type->length != 0))
+	    continue;
+
+	 const glsl_type *type =
+	    glsl_type::get_array_instance(var->type->fields.array,
+					  var->max_array_access);
+
+	 assert(type != NULL);
+	 var->type = type;
+      }
+   }
+
    return linked;
 }
 
@@ -859,7 +929,10 @@ struct uniform_node {
 static void
 update_array_sizes(struct gl_shader_program *prog)
 {
-   for (unsigned i = 0; i < prog->_NumLinkedShaders; i++) {
+   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+	 if (prog->_LinkedShaders[i] == NULL)
+	    continue;
+
       foreach_list(node, prog->_LinkedShaders[i]->ir) {
 	 ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
@@ -870,7 +943,10 @@ update_array_sizes(struct gl_shader_program *prog)
 	    continue;
 
 	 unsigned int size = var->max_array_access;
-	 for (unsigned j = 0; j < prog->_NumLinkedShaders; j++) {
+	 for (unsigned j = 0; j < MESA_SHADER_TYPES; j++) {
+	       if (prog->_LinkedShaders[j] == NULL)
+		  continue;
+
 	    foreach_list(node2, prog->_LinkedShaders[j]->ir) {
 	       ir_variable *other_var = ((ir_instruction *) node2)->as_variable();
 	       if (!other_var)
@@ -982,7 +1058,10 @@ assign_uniform_locations(struct gl_shader_program *prog)
 				    hash_table_string_compare);
    void *mem_ctx = talloc_new(NULL);
 
-   for (unsigned i = 0; i < prog->_NumLinkedShaders; i++) {
+   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+      if (prog->_LinkedShaders[i] == NULL)
+	 continue;
+
       unsigned next_position = 0;
 
       foreach_list(node, prog->_LinkedShaders[i]->ir) {
@@ -1239,19 +1318,20 @@ assign_attribute_locations(gl_shader_program *prog, unsigned max_attribute_index
 
 
 /**
- * Demote shader outputs that are not read to being just plain global variables
+ * Demote shader inputs and outputs that are not used in other stages
  */
 void
-demote_unread_shader_outputs(gl_shader *sh)
+demote_shader_inputs_and_outputs(gl_shader *sh, enum ir_variable_mode mode)
 {
    foreach_list(node, sh->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
-      if ((var == NULL) || (var->mode != ir_var_out))
+      if ((var == NULL) || (var->mode != int(mode)))
 	 continue;
 
-      /* An 'out' variable is only really a shader output if its value is read
-       * by the following stage.
+      /* A shader 'in' or 'out' variable is only really an input or output if
+       * its value is used by other shader stages.  This will cause the variable
+       * to have a location assigned.
        */
       if (var->location == -1) {
 	 var->mode = ir_var_auto;
@@ -1317,8 +1397,6 @@ assign_varying_locations(struct gl_shader_program *prog,
       }
    }
 
-   demote_unread_shader_outputs(producer);
-
    foreach_list(node, consumer->ir) {
       ir_variable *const var = ((ir_instruction *) node)->as_variable();
 
@@ -1357,6 +1435,8 @@ assign_varying_locations(struct gl_shader_program *prog,
 void
 link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
 {
+   void *mem_ctx = talloc_init("temporary linker context");
+
    prog->LinkStatus = false;
    prog->Validated = false;
    prog->_Used = false;
@@ -1414,16 +1494,19 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
 
    prog->Version = max_version;
 
-   for (unsigned int i = 0; i < prog->_NumLinkedShaders; i++) {
-      ctx->Driver.DeleteShader(ctx, prog->_LinkedShaders[i]);
+   for (unsigned int i = 0; i < MESA_SHADER_TYPES; i++) {
+      if (prog->_LinkedShaders[i] != NULL)
+	 ctx->Driver.DeleteShader(ctx, prog->_LinkedShaders[i]);
+
+      prog->_LinkedShaders[i] = NULL;
    }
 
    /* Link all shaders for a particular stage and validate the result.
     */
-   prog->_NumLinkedShaders = 0;
    if (num_vert_shaders > 0) {
       gl_shader *const sh =
-	 link_intrastage_shaders(ctx, prog, vert_shader_list, num_vert_shaders);
+	 link_intrastage_shaders(mem_ctx, ctx, prog, vert_shader_list,
+				 num_vert_shaders);
 
       if (sh == NULL)
 	 goto done;
@@ -1431,13 +1514,14 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
       if (!validate_vertex_shader_executable(prog, sh))
 	  goto done;
 
-      prog->_LinkedShaders[prog->_NumLinkedShaders] = sh;
-      prog->_NumLinkedShaders++;
+      _mesa_reference_shader(ctx, &prog->_LinkedShaders[MESA_SHADER_VERTEX],
+			     sh);
    }
 
    if (num_frag_shaders > 0) {
       gl_shader *const sh =
-	 link_intrastage_shaders(ctx, prog, frag_shader_list, num_frag_shaders);
+	 link_intrastage_shaders(mem_ctx, ctx, prog, frag_shader_list,
+				 num_frag_shaders);
 
       if (sh == NULL)
 	 goto done;
@@ -1445,8 +1529,8 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
       if (!validate_fragment_shader_executable(prog, sh))
 	  goto done;
 
-      prog->_LinkedShaders[prog->_NumLinkedShaders] = sh;
-      prog->_NumLinkedShaders++;
+      _mesa_reference_shader(ctx, &prog->_LinkedShaders[MESA_SHADER_FRAGMENT],
+			     sh);
    }
 
    /* Here begins the inter-stage linking phase.  Some initial validation is
@@ -1454,14 +1538,26 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
     * varyings.
     */
    if (cross_validate_uniforms(prog)) {
+      unsigned prev;
+
+      for (prev = 0; prev < MESA_SHADER_TYPES; prev++) {
+	 if (prog->_LinkedShaders[prev] != NULL)
+	    break;
+      }
+
       /* Validate the inputs of each stage with the output of the preceeding
        * stage.
        */
-      for (unsigned i = 1; i < prog->_NumLinkedShaders; i++) {
+      for (unsigned i = prev + 1; i < MESA_SHADER_TYPES; i++) {
+	 if (prog->_LinkedShaders[i] == NULL)
+	    continue;
+
 	 if (!cross_validate_outputs_to_inputs(prog,
-					       prog->_LinkedShaders[i - 1],
+					       prog->_LinkedShaders[prev],
 					       prog->_LinkedShaders[i]))
 	    goto done;
+
+	 prev = i;
       }
 
       prog->LinkStatus = true;
@@ -1471,7 +1567,10 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
     * uniforms, and varyings.  Later optimization could possibly make
     * some of that unused.
     */
-   for (unsigned i = 0; i < prog->_NumLinkedShaders; i++) {
+   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+      if (prog->_LinkedShaders[i] == NULL)
+	 continue;
+
       while (do_common_optimization(prog->_LinkedShaders[i]->ir, true, 32))
 	 ;
    }
@@ -1480,7 +1579,7 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
 
    assign_uniform_locations(prog);
 
-   if (prog->_NumLinkedShaders && prog->_LinkedShaders[0]->Type == GL_VERTEX_SHADER) {
+   if (prog->_LinkedShaders[MESA_SHADER_VERTEX] != NULL) {
       /* FINISHME: The value of the max_attribute_index parameter is
        * FINISHME: implementation dependent based on the value of
        * FINISHME: GL_MAX_VERTEX_ATTRIBS.  GL_MAX_VERTEX_ATTRIBS must be
@@ -1488,18 +1587,55 @@ link_shaders(GLcontext *ctx, struct gl_shader_program *prog)
        */
       if (!assign_attribute_locations(prog, 16))
 	 goto done;
-
-      if (prog->_NumLinkedShaders == 1)
-	 demote_unread_shader_outputs(prog->_LinkedShaders[0]);
    }
 
-   for (unsigned i = 1; i < prog->_NumLinkedShaders; i++)
+   unsigned prev;
+   for (prev = 0; prev < MESA_SHADER_TYPES; prev++) {
+      if (prog->_LinkedShaders[prev] != NULL)
+	 break;
+   }
+
+   for (unsigned i = prev + 1; i < MESA_SHADER_TYPES; i++) {
+      if (prog->_LinkedShaders[i] == NULL)
+	 continue;
+
       assign_varying_locations(prog,
-			       prog->_LinkedShaders[i - 1],
+			       prog->_LinkedShaders[prev],
 			       prog->_LinkedShaders[i]);
+      prev = i;
+   }
+
+   if (prog->_LinkedShaders[MESA_SHADER_VERTEX] != NULL) {
+      demote_shader_inputs_and_outputs(prog->_LinkedShaders[MESA_SHADER_VERTEX],
+				       ir_var_out);
+   }
+
+   if (prog->_LinkedShaders[MESA_SHADER_GEOMETRY] != NULL) {
+      gl_shader *const sh = prog->_LinkedShaders[MESA_SHADER_GEOMETRY];
+
+      demote_shader_inputs_and_outputs(sh, ir_var_in);
+      demote_shader_inputs_and_outputs(sh, ir_var_inout);
+      demote_shader_inputs_and_outputs(sh, ir_var_out);
+   }
+
+   if (prog->_LinkedShaders[MESA_SHADER_FRAGMENT] != NULL) {
+      gl_shader *const sh = prog->_LinkedShaders[MESA_SHADER_FRAGMENT];
+
+      demote_shader_inputs_and_outputs(sh, ir_var_in);
+   }
 
    /* FINISHME: Assign fragment shader output locations. */
 
 done:
    free(vert_shader_list);
+
+   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++) {
+      if (prog->_LinkedShaders[i] == NULL)
+	 continue;
+
+      /* Retain any live IR, but trash the rest. */
+      reparent_ir(prog->_LinkedShaders[i]->ir, prog->_LinkedShaders[i]->ir);
+   }
+
+   talloc_free(mem_ctx);
 }

src/glsl/loop_analysis.cpp

@@ -44,6 +44,7 @@ loop_state::loop_state()
 loop_state::~loop_state()
 {
    hash_table_dtor(this->ht);
+   talloc_free(this->mem_ctx);
 }
 
 

src/glsl/loop_controls.cpp

@@ -85,6 +85,9 @@ int
 calculate_iterations(ir_rvalue *from, ir_rvalue *to, ir_rvalue *increment,
 		     enum ir_expression_operation op)
 {
+   if (from == NULL || to == NULL || increment == NULL)
+      return -1;
+
    void *mem_ctx = talloc_init("%s", __func__);
 
    ir_expression *const sub =

src/glsl/loop_unroll.cpp

@@ -43,6 +43,14 @@ public:
 };
 
 
+static bool
+is_break(ir_instruction *ir)
+{
+   return ir != NULL && ir->ir_type == ir_type_loop_jump
+		     && ((ir_loop_jump *) ir)->is_break();
+}
+
+
 ir_visitor_status
 loop_unroll_visitor::visit_leave(ir_loop *ir)
 {
@@ -73,44 +81,74 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
    if (ls->num_loop_jumps > 1)
       return visit_continue;
    else if (ls->num_loop_jumps) {
-      /* recognize loops in the form produced by ir_lower_jumps */
-      ir_instruction *last_ir =
-	 ((ir_instruction*)ir->body_instructions.get_tail());
-
+      ir_instruction *last_ir = (ir_instruction *) ir->body_instructions.get_tail();
       assert(last_ir != NULL);
 
-      ir_if *last_if = last_ir->as_if();
-      if (last_if) {
-	 bool continue_from_then_branch;
+      if (is_break(last_ir)) {
+         /* If the only loop-jump is a break at the end of the loop, the loop
+          * will execute exactly once.  Remove the break, set the iteration
+          * count, and fall through to the normal unroller.
+          */
+         last_ir->remove();
+         iterations = 1;
 
-	 /* Determine which if-statement branch, if any, ends with a break.
-	  * The branch that did *not* have the break will get a temporary
-	  * continue inserted in each iteration of the loop unroll.
-	  *
-	  * Note that since ls->num_loop_jumps is <= 1, it is impossible for
-	  * both branches to end with a break.
-	  */
-	 ir_instruction *last =
-	    (ir_instruction *) last_if->then_instructions.get_tail();
+         this->progress = true;
+      } else {
+         ir_if *ir_if = NULL;
+         ir_instruction *break_ir = NULL;
+         bool continue_from_then_branch = false;
 
-	 if (last && last->ir_type == ir_type_loop_jump
-	     && ((ir_loop_jump*) last)->is_break()) {
-	    continue_from_then_branch = false;
-	 } else {
-	    last = (ir_instruction *) last_if->then_instructions.get_tail();
+         foreach_list(node, &ir->body_instructions) {
+            /* recognize loops in the form produced by ir_lower_jumps */
+            ir_instruction *cur_ir = (ir_instruction *) node;
 
-	    if (last && last->ir_type == ir_type_loop_jump
-		&& ((ir_loop_jump*) last)->is_break())
-	       continue_from_then_branch = true;
-	    else
-	       /* Bail out if neither if-statement branch ends with a break.
+            ir_if = cur_ir->as_if();
+            if (ir_if != NULL) {
+	       /* Determine which if-statement branch, if any, ends with a
+		* break.  The branch that did *not* have the break will get a
+		* temporary continue inserted in each iteration of the loop
+		* unroll.
+		*
+		* Note that since ls->num_loop_jumps is <= 1, it is impossible
+		* for both branches to end with a break.
 		*/
-	       return visit_continue;
-	 }
+               ir_instruction *ir_if_last =
+                  (ir_instruction *) ir_if->then_instructions.get_tail();
 
-	 /* Remove the break from the if-statement.
-	  */
-	 last->remove();
+               if (is_break(ir_if_last)) {
+                  continue_from_then_branch = false;
+                  break_ir = ir_if_last;
+                  break;
+               } else {
+                  ir_if_last =
+		     (ir_instruction *) ir_if->else_instructions.get_tail();
+
+                  if (is_break(ir_if_last)) {
+                     break_ir = ir_if_last;
+                     continue_from_then_branch = true;
+                     break;
+                  }
+               }
+            }
+         }
+
+         if (break_ir == NULL)
+            return visit_continue;
+
+         /* move instructions after then if in the continue branch */
+         while (!ir_if->get_next()->is_tail_sentinel()) {
+            ir_instruction *move_ir = (ir_instruction *) ir_if->get_next();
+
+            move_ir->remove();
+            if (continue_from_then_branch)
+               ir_if->then_instructions.push_tail(move_ir);
+            else
+               ir_if->else_instructions.push_tail(move_ir);
+         }
+
+         /* Remove the break from the if-statement.
+          */
+         break_ir->remove();
 
          void *const mem_ctx = talloc_parent(ir);
          ir_instruction *ir_to_replace = ir;
@@ -121,8 +159,8 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
             copy_list.make_empty();
             clone_ir_list(mem_ctx, &copy_list, &ir->body_instructions);
 
-            last_if = ((ir_instruction*)copy_list.get_tail())->as_if();
-            assert(last_if);
+            ir_if = ((ir_instruction *) copy_list.get_tail())->as_if();
+            assert(ir_if != NULL);
 
             ir_to_replace->insert_before(&copy_list);
             ir_to_replace->remove();
@@ -132,7 +170,7 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
 	       new(mem_ctx) ir_loop_jump(ir_loop_jump::jump_continue);
 
             exec_list *const list = (continue_from_then_branch)
-	       ? &last_if->then_instructions : &last_if->else_instructions;
+               ? &ir_if->then_instructions : &ir_if->else_instructions;
 
             list->push_tail(ir_to_replace);
          }
@@ -141,18 +179,7 @@ loop_unroll_visitor::visit_leave(ir_loop *ir)
 
          this->progress = true;
          return visit_continue;
-      } else if (last_ir->ir_type == ir_type_loop_jump
-		 && ((ir_loop_jump *)last_ir)->is_break()) {
-	 /* If the only loop-jump is a break at the end of the loop, the loop
-	  * will execute exactly once.  Remove the break, set the iteration
-	  * count, and fall through to the normal unroller.
-	  */
-         last_ir->remove();
-	 iterations = 1;
-
-	 this->progress = true;
-      } else
-         return visit_continue;
+      }
    }
 
    void *const mem_ctx = talloc_parent(ir);

src/glsl/lower_discard.cpp

@@ -0,0 +1,198 @@
+/*
+ * Copyright © 2010 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (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 NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file lower_discard.cpp
+ *
+ * This pass moves discards out of if-statements.
+ *
+ * Case 1: The "then" branch contains a conditional discard:
+ * ---------------------------------------------------------
+ *
+ *    if (cond1) {
+ *	 s1;
+ *	 discard cond2;
+ *	 s2;
+ *    } else {
+ *	 s3;
+ *    }
+ *
+ * becomes:
+ *
+ *    temp = false;
+ *    if (cond1) {
+ *	 s1;
+ *	 temp = cond2;
+ *	 s2;
+ *    } else {
+ *	 s3;
+ *    }
+ *    discard temp;
+ *
+ * Case 2: The "else" branch contains a conditional discard:
+ * ---------------------------------------------------------
+ *
+ *    if (cond1) {
+ *	 s1;
+ *    } else {
+ *	 s2;
+ *	 discard cond2;
+ *	 s3;
+ *    }
+ *
+ * becomes:
+ *
+ *    temp = false;
+ *    if (cond1) {
+ *	 s1;
+ *    } else {
+ *	 s2;
+ *	 temp = cond2;
+ *	 s3;
+ *    }
+ *    discard temp;
+ *
+ * Case 3: Both branches contain a conditional discard:
+ * ----------------------------------------------------
+ *
+ *    if (cond1) {
+ *	 s1;
+ *	 discard cond2;
+ *	 s2;
+ *    } else {
+ *	 s3;
+ *	 discard cond3;
+ *	 s4;
+ *    }
+ *
+ * becomes:
+ *
+ *    temp = false;
+ *    if (cond1) {
+ *	 s1;
+ *	 temp = cond2;
+ *	 s2;
+ *    } else {
+ *	 s3;
+ *	 temp = cond3;
+ *	 s4;
+ *    }
+ *    discard temp;
+ *
+ * If there are multiple conditional discards, we need only deal with one of
+ * them.  Repeatedly applying this pass will take care of the others.
+ *
+ * Unconditional discards are treated as having a condition of "true".
+ */
+
+#include "glsl_types.h"
+#include "ir.h"
+
+class lower_discard_visitor : public ir_hierarchical_visitor {
+public:
+   lower_discard_visitor()
+   {
+      this->progress = false;
+   }
+
+   ir_visitor_status visit_leave(ir_if *);
+
+   bool progress;
+};
+
+
+bool
+lower_discard(exec_list *instructions)
+{
+   lower_discard_visitor v;
+
+   visit_list_elements(&v, instructions);
+
+   return v.progress;
+}
+
+
+static ir_discard *
+find_discard(exec_list &instructions)
+{
+   foreach_list(n, &instructions) {
+      ir_discard *ir = ((ir_instruction *) n)->as_discard();
+      if (ir != NULL)
+	 return ir;
+   }
+   return NULL;
+}
+
+
+static void
+replace_discard(void *mem_ctx, ir_variable *var, ir_discard *ir)
+{
+   ir_rvalue *condition = ir->condition;
+
+   /* For unconditional discards, use "true" as the condition. */
+   if (condition == NULL)
+      condition = new(mem_ctx) ir_constant(true);
+
+   ir_assignment *assignment =
+      new(mem_ctx) ir_assignment(new(mem_ctx) ir_dereference_variable(var),
+				 condition, NULL);
+
+   ir->replace_with(assignment);
+}
+
+
+ir_visitor_status
+lower_discard_visitor::visit_leave(ir_if *ir)
+{
+   ir_discard *then_discard = find_discard(ir->then_instructions);
+   ir_discard *else_discard = find_discard(ir->else_instructions);
+
+   if (then_discard == NULL && else_discard == NULL)
+      return visit_continue;
+
+   void *mem_ctx = talloc_parent(ir);
+
+   ir_variable *temp = new(mem_ctx) ir_variable(glsl_type::bool_type,
+						"discard_cond_temp",
+						ir_var_temporary);
+   ir_assignment *temp_initializer =
+      new(mem_ctx) ir_assignment(new(mem_ctx) ir_dereference_variable(temp),
+				 new(mem_ctx) ir_constant(false), NULL);
+
+   ir->insert_before(temp);
+   ir->insert_before(temp_initializer);
+
+   if (then_discard != NULL)
+      replace_discard(mem_ctx, temp, then_discard);
+
+   if (else_discard != NULL)
+      replace_discard(mem_ctx, temp, else_discard);
+
+   ir_discard *discard = then_discard != NULL ? then_discard : else_discard;
+   discard->condition = new(mem_ctx) ir_dereference_variable(temp);
+   ir->insert_after(discard);
+
+   this->progress = true;
+
+   return visit_continue;
+}

src/glsl/main.cpp

@@ -40,8 +40,19 @@
 extern "C" struct gl_shader *
 _mesa_new_shader(GLcontext *ctx, GLuint name, GLenum type);
 
+extern "C" void
+_mesa_reference_shader(struct gl_context *ctx, struct gl_shader **ptr,
+                       struct gl_shader *sh);
+
 /* Copied from shader_api.c for the stand-alone compiler.
  */
+void
+_mesa_reference_shader(struct gl_context *ctx, struct gl_shader **ptr,
+                       struct gl_shader *sh)
+{
+   *ptr = sh;
+}
+
 struct gl_shader *
 _mesa_new_shader(GLcontext *ctx, GLuint name, GLenum type)
 {
@@ -197,26 +208,7 @@ compile_shader(GLcontext *ctx, struct gl_shader *shader)
    if (!state->error && !shader->ir->is_empty()) {
       bool progress;
       do {
-	 progress = false;
-
-	 progress = do_function_inlining(shader->ir) || progress;
-	 progress = do_if_simplification(shader->ir) || progress;
-	 progress = do_copy_propagation(shader->ir) || progress;
-	 progress = do_dead_code_local(shader->ir) || progress;
-	 progress = do_dead_code_unlinked(shader->ir) || progress;
-	 progress = do_tree_grafting(shader->ir) || progress;
-	 progress = do_constant_propagation(shader->ir) || progress;
-	 progress = do_constant_variable_unlinked(shader->ir) || progress;
-	 progress = do_constant_folding(shader->ir) || progress;
-	 progress = do_algebraic(shader->ir) || progress;
-	 progress = do_vec_index_to_swizzle(shader->ir) || progress;
-	 progress = do_vec_index_to_cond_assign(shader->ir) || progress;
-	 progress = do_swizzle_swizzle(shader->ir) || progress;
-
-	 loop_state *ls = analyze_loop_variables(shader->ir);
-	 progress = set_loop_controls(shader->ir, ls) || progress;
-	 progress = unroll_loops(shader->ir, ls, 32) || progress;
-	 delete ls;
+	 progress = do_common_optimization(shader->ir, false, 32);
       } while (progress);
 
       validate_ir_tree(shader->ir);
@@ -319,7 +311,7 @@ main(int argc, char **argv)
 	 printf("Info log for linking:\n%s\n", whole_program->InfoLog);
    }
 
-   for (unsigned i = 0; i < whole_program->_NumLinkedShaders; i++)
+   for (unsigned i = 0; i < MESA_SHADER_TYPES; i++)
       talloc_free(whole_program->_LinkedShaders[i]);
 
    talloc_free(whole_program);

src/glsl/opt_discard_simplification.cpp

@@ -0,0 +1,180 @@
+/*
+ * Copyright © 2010 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (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 NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
+ * DEALINGS IN THE SOFTWARE.
+ */
+
+/**
+ * \file opt_discard_simplification.cpp
+ *
+ * This pass simplifies if-statements and loops containing unconditional
+ * discards.
+ *
+ * Case 1: Both branches contain unconditional discards:
+ * -----------------------------------------------------
+ *
+ *    if (cond) {
+ *	 s1;
+ *	 discard;
+ *	 s2;
+ *    } else {
+ *	 s3;
+ *	 discard;
+ *	 s4;
+ *    }
+ *
+ * becomes:
+ *
+ *    discard
+ *
+ * Case 2: The "then" clause contains an unconditional discard:
+ * ------------------------------------------------------------
+ *
+ *    if (cond) {
+ *       s1;
+ *       discard;
+ *       s2;
+ *    } else {
+ *	 s3;
+ *    }
+ *
+ * becomes:
+ *
+ *    if (cond) {
+ *	 discard;
+ *    } else {
+ *	 s3;
+ *    }
+ *
+ * Case 3: The "else" clause contains an unconditional discard:
+ * ------------------------------------------------------------
+ *
+ *    if (cond) {
+ *       s1;
+ *    } else {
+ *       s2;
+ *       discard;
+ *	 s3;
+ *    }
+ *
+ * becomes:
+ *
+ *    if (cond) {
+ *	 s1;
+ *    } else {
+ *	 discard;
+ *    }
+ */
+
+#include "glsl_types.h"
+#include "ir.h"
+
+class discard_simplifier : public ir_hierarchical_visitor {
+public:
+   discard_simplifier()
+   {
+      this->progress = false;
+   }
+
+   ir_visitor_status visit_enter(ir_if *);
+   ir_visitor_status visit_enter(ir_loop *);
+
+   bool progress;
+};
+
+static ir_discard *
+find_unconditional_discard(exec_list &instructions)
+{
+   foreach_list(n, &instructions) {
+      ir_discard *ir = ((ir_instruction *) n)->as_discard();
+      if (ir != NULL && ir->condition == NULL)
+	 return ir;
+   }
+   return NULL;
+}
+
+static bool
+is_only_instruction(ir_discard *discard)
+{
+   return (discard->prev->is_head_sentinel() &&
+	   discard->next->is_tail_sentinel());
+}
+
+ir_visitor_status
+discard_simplifier::visit_enter(ir_if *ir)
+{
+   ir_discard *then_discard = find_unconditional_discard(ir->then_instructions);
+   ir_discard *else_discard = find_unconditional_discard(ir->else_instructions);
+
+   if (then_discard == NULL && else_discard == NULL)
+      return visit_continue;
+
+   /* If both branches result in discard, replace whole if with discard. */
+   if (then_discard != NULL && else_discard != NULL) {
+      this->progress = true;
+      ir->replace_with(then_discard);
+      return visit_continue_with_parent;
+   }
+
+   /* Otherwise, one branch has a discard. */
+   if (then_discard != NULL && !is_only_instruction(then_discard)) {
+      this->progress = true;
+      ir->then_instructions.make_empty();
+      ir->then_instructions.push_tail(then_discard);
+   } else if (else_discard != NULL && !is_only_instruction(else_discard)) {
+      this->progress = true;
+      ir->else_instructions.make_empty();
+      ir->else_instructions.push_tail(else_discard);
+   }
+
+   visit_list_elements(this, &ir->then_instructions);
+   return visit_continue_with_parent;
+}
+
+ir_visitor_status
+discard_simplifier::visit_enter(ir_loop *ir)
+{
+   ir_discard *discard = find_unconditional_discard(ir->body_instructions);
+
+   if (discard) {
+      ir->replace_with(discard);
+      return visit_continue_with_parent;
+   }
+
+   return visit_continue;
+}
+
+bool
+do_discard_simplification(exec_list *instructions)
+{
+   /* Look for a top-level unconditional discard */
+   ir_discard *discard = find_unconditional_discard(*instructions);
+   if (discard != NULL) {
+      instructions->make_empty();
+      instructions->push_tail(discard);
+      return true;
+   }
+
+   discard_simplifier v;
+
+   visit_list_elements(&v, instructions);
+
+   return v.progress;
+}

src/glsl/s_expression.cpp

@@ -62,7 +62,7 @@ read_atom(void *ctx, const char *& src)
 
    // Check if the atom is a number.
    char *float_end = NULL;
-   double f = strtod(src, &float_end);
+   double f = glsl_strtod(src, &float_end);
    if (float_end != src) {
       char *int_end = NULL;
       int i = strtol(src, &int_end, 10);

src/glsl/s_expression.h

@@ -26,6 +26,7 @@
 #ifndef S_EXPRESSION_H
 #define S_EXPRESSION_H
 
+#include "strtod.h"
 #include "list.h"
 
 #define SX_AS_(t,x) ((x) && ((s_expression*) x)->is_##t()) ? ((s_##t*) (x)) \

src/glsl/strtod.c

@@ -0,0 +1,56 @@
+/*
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, 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 VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#include <stdlib.h>
+
+#ifdef _GNU_SOURCE
+#include <locale.h>
+#ifdef __APPLE__
+#include <xlocale.h>
+#endif
+#endif
+
+#include "strtod.h"
+
+
+
+/**
+ * Wrapper around strtod which uses the "C" locale so the decimal
+ * point is always '.'
+ */
+double
+glsl_strtod(const char *s, char **end)
+{
+#if defined(_GNU_SOURCE) && !defined(__CYGWIN__) && !defined(__FreeBSD__)
+   static locale_t loc = NULL;
+   if (!loc) {
+      loc = newlocale(LC_CTYPE_MASK, "C", NULL);
+   }
+   return strtod_l(s, end, loc);
+#else
+   return strtod(s, end);
+#endif
+}

src/glsl/strtod.h

@@ -0,0 +1,43 @@
+/*
+ * Copyright 2010 VMware, Inc.
+ * All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, 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 VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR
+ * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#ifndef STRTOD_H
+#define STRTOD_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+extern double
+glsl_strtod(const char *s, char **end);
+
+
+#ifdef __cplusplus
+}
+#endif
+
+
+#endif