docs: add release notes for 10.6.4

Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

VERSION

@@ -1 +1 @@
-10.6.0-devel
+10.6.4

bin/bugzilla_mesa.sh

@@ -15,17 +15,14 @@
 # $ DRYRUN=yes bin/bugzilla_mesa.sh mesa-9.0.2..mesa-9.0.3 | wc -l
 
 
-# regex pattern: trim before url
-trim_before='s/.*\(http\)/\1/'
+# regex pattern: trim before bug number
+trim_before='s/.*show_bug.cgi?id=\([0-9]*\).*/\1/'
 
-# regex pattern: trim after url
-trim_after='s/\(show_bug.cgi?id=[0-9]*\).*/\1/'
-
-# regex pattern: always use https
-use_https='s/http:/https:/'
+# regex pattern: reconstruct the url
+use_after='s,^,https://bugs.freedesktop.org/show_bug.cgi?id=,'
 
 # extract fdo urls from commit log
-urls=$(git log $* | grep 'bugs.freedesktop.org/show_bug' | sed -e $trim_before -e $trim_after -e $use_https | sort | uniq)
+urls=$(git log $* | grep 'bugs.freedesktop.org/show_bug' | sed -e $trim_before | sort -n -u | sed -e $use_after)
 
 # if DRYRUN is set to "yes", simply print the URLs and don't fetch the
 # details from fdo bugzilla.

configure.ac

@@ -9,7 +9,6 @@ dnl Copyright © 2009-2014 Jon TURNEY
 dnl Copyright © 2011-2012 Benjamin Franzke
 dnl Copyright © 2008-2014 David Airlie
 dnl Copyright © 2009-2013 Brian Paul
-dnl Copyright © 2003-2007 Keith Packard, Daniel Stone
 dnl
 dnl Permission is hereby granted, free of charge, to any person obtaining a
 dnl copy of this software and associated documentation files (the "Software"),
@@ -64,6 +63,8 @@ m4_ifdef([AM_PROG_AR], [AM_PROG_AR])
 dnl Set internal versions
 OSMESA_VERSION=8
 AC_SUBST([OSMESA_VERSION])
+OPENCL_VERSION=1
+AC_SUBST([OPENCL_VERSION])
 
 dnl Versions for external dependencies
 LIBDRM_REQUIRED=2.4.38
@@ -649,6 +650,7 @@ if test "x$enable_asm" = xyes; then
 fi
 
 AC_CHECK_HEADER([xlocale.h], [DEFINES="$DEFINES -DHAVE_XLOCALE_H"])
+AC_CHECK_HEADER([sys/sysctl.h], [DEFINES="$DEFINES -DHAVE_SYS_SYSCTL_H"])
 AC_CHECK_FUNC([strtof], [DEFINES="$DEFINES -DHAVE_STRTOF"])
 
 dnl Check to see if dlopen is in default libraries (like Solaris, which
@@ -909,6 +911,13 @@ fi
 AM_CONDITIONAL(HAVE_DRI_GLX, test "x$enable_glx" = xyes -a \
                                   "x$enable_dri" = xyes)
 
+# Check for libdrm
+PKG_CHECK_MODULES([LIBDRM], [libdrm >= $LIBDRM_REQUIRED],
+                  [have_libdrm=yes], [have_libdrm=no])
+if test "x$have_libdrm" = xyes; then
+	DEFINES="$DEFINES -DHAVE_LIBDRM"
+fi
+
 # Select which platform-dependent DRI code gets built
 case "$host_os" in
 darwin*)
@@ -921,8 +930,8 @@ esac
 
 AM_CONDITIONAL(HAVE_DRICOMMON, test "x$enable_dri" = xyes )
 AM_CONDITIONAL(HAVE_DRISW, test "x$enable_dri" = xyes )
-AM_CONDITIONAL(HAVE_DRI2, test "x$enable_dri" = xyes -a "x$dri_platform" = xdrm )
-AM_CONDITIONAL(HAVE_DRI3, test "x$enable_dri3" = xyes -a "x$dri_platform" = xdrm )
+AM_CONDITIONAL(HAVE_DRI2, test "x$enable_dri" = xyes -a "x$dri_platform" = xdrm -a "x$have_libdrm" = xyes )
+AM_CONDITIONAL(HAVE_DRI3, test "x$enable_dri3" = xyes -a "x$dri_platform" = xdrm -a "x$have_libdrm" = xyes )
 AM_CONDITIONAL(HAVE_APPLEDRI, test "x$enable_dri" = xyes -a "x$dri_platform" = xapple )
 
 AC_ARG_ENABLE([shared-glapi],
@@ -941,7 +950,7 @@ x*yes*yes*)
 esac
 
 # Building Xlib-GLX requires shared glapi to be disabled.
-if test "x$enable_xlib_glx" = xyes; then
+if test "x$enable_shared_glapi$enable_xlib_glx" = xyesyes; then
     AC_MSG_NOTICE([Shared GLAPI should not used with Xlib-GLX, disabling])
     enable_shared_glapi=no
 fi
@@ -978,151 +987,6 @@ fi
 
 AC_SUBST([MESA_LLVM])
 
-# SHA1 hashing
-AC_ARG_WITH([sha1],
-        [AS_HELP_STRING([--with-sha1=libc|libmd|libnettle|libgcrypt|libcrypto|libsha1|CommonCrypto|CryptoAPI],
-        [choose SHA1 implementation])])
-case "x$with_sha1" in
-x | xlibc | xlibmd | xlibnettle | xlibgcrypt | xlibcrypto | xlibsha1 | xCommonCrypto | xCryptoAPI)
-  ;;
-*)
-        AC_MSG_ERROR([Illegal value for --with-sha1: $with_sha1])
-esac
-
-AC_CHECK_FUNC([SHA1Init], [HAVE_SHA1_IN_LIBC=yes])
-if test "x$with_sha1" = x && test "x$HAVE_SHA1_IN_LIBC" = xyes; then
-	with_sha1=libc
-fi
-if test "x$with_sha1" = xlibc && test "x$HAVE_SHA1_IN_LIBC" != xyes; then
-	AC_MSG_ERROR([sha1 in libc requested but not found])
-fi
-if test "x$with_sha1" = xlibc; then
-	AC_DEFINE([HAVE_SHA1_IN_LIBC], [1],
-		[Use libc SHA1 functions])
-	SHA1_LIBS=""
-fi
-AC_CHECK_FUNC([CC_SHA1_Init], [HAVE_SHA1_IN_COMMONCRYPTO=yes])
-if test "x$with_sha1" = x && test "x$HAVE_SHA1_IN_COMMONCRYPTO" = xyes; then
-	with_sha1=CommonCrypto
-fi
-if test "x$with_sha1" = xCommonCrypto && test "x$HAVE_SHA1_IN_COMMONCRYPTO" != xyes; then
-	AC_MSG_ERROR([CommonCrypto requested but not found])
-fi
-if test "x$with_sha1" = xCommonCrypto; then
-	AC_DEFINE([HAVE_SHA1_IN_COMMONCRYPTO], [1],
-		[Use CommonCrypto SHA1 functions])
-	SHA1_LIBS=""
-fi
-dnl stdcall functions cannot be tested with AC_CHECK_LIB
-AC_CHECK_HEADER([wincrypt.h], [HAVE_SHA1_IN_CRYPTOAPI=yes], [], [#include <windows.h>])
-if test "x$with_sha1" = x && test "x$HAVE_SHA1_IN_CRYPTOAPI" = xyes; then
-	with_sha1=CryptoAPI
-fi
-if test "x$with_sha1" = xCryptoAPI && test "x$HAVE_SHA1_IN_CRYPTOAPI" != xyes; then
-	AC_MSG_ERROR([CryptoAPI requested but not found])
-fi
-if test "x$with_sha1" = xCryptoAPI; then
-	AC_DEFINE([HAVE_SHA1_IN_CRYPTOAPI], [1],
-		[Use CryptoAPI SHA1 functions])
-	SHA1_LIBS=""
-fi
-AC_CHECK_LIB([md], [SHA1Init], [HAVE_LIBMD=yes])
-if test "x$with_sha1" = x && test "x$HAVE_LIBMD" = xyes; then
-	with_sha1=libmd
-fi
-if test "x$with_sha1" = xlibmd && test "x$HAVE_LIBMD" != xyes; then
-	AC_MSG_ERROR([libmd requested but not found])
-fi
-if test "x$with_sha1" = xlibmd; then
-	AC_DEFINE([HAVE_SHA1_IN_LIBMD], [1],
-	          [Use libmd SHA1 functions])
-	SHA1_LIBS=-lmd
-fi
-PKG_CHECK_MODULES([LIBSHA1], [libsha1], [HAVE_LIBSHA1=yes], [HAVE_LIBSHA1=no])
-if test "x$with_sha1" = x && test "x$HAVE_LIBSHA1" = xyes; then
-   with_sha1=libsha1
-fi
-if test "x$with_sha1" = xlibsha1 && test "x$HAVE_LIBSHA1" != xyes; then
-	AC_MSG_ERROR([libsha1 requested but not found])
-fi
-if test "x$with_sha1" = xlibsha1; then
-	AC_DEFINE([HAVE_SHA1_IN_LIBSHA1], [1],
-	          [Use libsha1 for SHA1])
-	SHA1_LIBS=-lsha1
-fi
-AC_CHECK_LIB([nettle], [nettle_sha1_init], [HAVE_LIBNETTLE=yes])
-if test "x$with_sha1" = x && test "x$HAVE_LIBNETTLE" = xyes; then
-	with_sha1=libnettle
-fi
-if test "x$with_sha1" = xlibnettle && test "x$HAVE_LIBNETTLE" != xyes; then
-	AC_MSG_ERROR([libnettle requested but not found])
-fi
-if test "x$with_sha1" = xlibnettle; then
-	AC_DEFINE([HAVE_SHA1_IN_LIBNETTLE], [1],
-	          [Use libnettle SHA1 functions])
-	SHA1_LIBS=-lnettle
-fi
-AC_CHECK_LIB([gcrypt], [gcry_md_open], [HAVE_LIBGCRYPT=yes])
-if test "x$with_sha1" = x && test "x$HAVE_LIBGCRYPT" = xyes; then
-	with_sha1=libgcrypt
-fi
-if test "x$with_sha1" = xlibgcrypt && test "x$HAVE_LIBGCRYPT" != xyes; then
-	AC_MSG_ERROR([libgcrypt requested but not found])
-fi
-if test "x$with_sha1" = xlibgcrypt; then
-	AC_DEFINE([HAVE_SHA1_IN_LIBGCRYPT], [1],
-	          [Use libgcrypt SHA1 functions])
-	SHA1_LIBS=-lgcrypt
-fi
-# We don't need all of the OpenSSL libraries, just libcrypto
-AC_CHECK_LIB([crypto], [SHA1_Init], [HAVE_LIBCRYPTO=yes])
-PKG_CHECK_MODULES([OPENSSL], [openssl], [HAVE_OPENSSL_PKC=yes],
-                  [HAVE_OPENSSL_PKC=no])
-if test "x$HAVE_LIBCRYPTO" = xyes || test "x$HAVE_OPENSSL_PKC" = xyes; then
-	if test "x$with_sha1" = x; then
-		with_sha1=libcrypto
-	fi
-else
-	if test "x$with_sha1" = xlibcrypto; then
-		AC_MSG_ERROR([OpenSSL libcrypto requested but not found])
-	fi
-fi
-if test "x$with_sha1" = xlibcrypto; then
-	if test "x$HAVE_LIBCRYPTO" = xyes; then
-		SHA1_LIBS=-lcrypto
-	else
-		SHA1_LIBS="$OPENSSL_LIBS"
-		SHA1_CFLAGS="$OPENSSL_CFLAGS"
-	fi
-fi
-AC_MSG_CHECKING([for SHA1 implementation])
-AC_MSG_RESULT([$with_sha1])
-AC_SUBST(SHA1_LIBS)
-AC_SUBST(SHA1_CFLAGS)
-
-# Allow user to configure out the shader-cache feature
-AC_ARG_ENABLE([shader-cache],
-    AS_HELP_STRING([--disable-shader-cache], [Disable binary shader cache]),
-    [enable_shader_cache="$enableval"],
-    [if test "x$with_sha1" != "x"; then
-        enable_shader_cache=yes
-     else
-        enable_shader_cache=no
-     fi])
-if test "x$with_sha1" = "x"; then
-    if test "x$enable_shader_cache" = "xyes"; then
-        AC_MSG_ERROR([Cannot enable shader cache (no SHA-1 implementation found)])
-    fi
-fi
-AM_CONDITIONAL([ENABLE_SHADER_CACHE], [test x$enable_shader_cache = xyes])
-
-# Check for libdrm
-PKG_CHECK_MODULES([LIBDRM], [libdrm >= $LIBDRM_REQUIRED],
-                  [have_libdrm=yes], [have_libdrm=no])
-if test "x$have_libdrm" = xyes; then
-	DEFINES="$DEFINES -DHAVE_LIBDRM"
-fi
-
 case "$host_os" in
 linux*)
     need_pci_id=yes ;;
@@ -1547,8 +1411,15 @@ if test "x$enable_egl" = xyes; then
 
     if test "$enable_static" != yes; then
         if test "x$enable_dri" = xyes; then
-	    HAVE_EGL_DRIVER_DRI2=1
-	fi
+            HAVE_EGL_DRIVER_DRI2=1
+            if test "x$enable_shared_glapi" = xno; then
+                AC_MSG_ERROR([egl_dri2 requires --enable-shared-glapi])
+            fi
+        else
+            # Avoid building an "empty" libEGL. Drop/update this
+            # when other backends (haiku?) come along.
+            AC_MSG_ERROR([egl requires --enable-dri])
+        fi
 
     fi
 fi
@@ -1758,6 +1629,9 @@ egl_platforms=`IFS=', '; echo $with_egl_platforms`
 for plat in $egl_platforms; do
 	case "$plat" in
 	wayland)
+		test "x$have_libdrm" != xyes &&
+			AC_MSG_ERROR([EGL platform wayland requires libdrm >= $LIBDRM_REQUIRED])
+
 		PKG_CHECK_MODULES([WAYLAND], [wayland-client >= $WAYLAND_REQUIRED wayland-server >= $WAYLAND_REQUIRED])
 
 		if test "x$WAYLAND_SCANNER" = x; then
@@ -2369,6 +2243,7 @@ AC_CONFIG_FILES([Makefile
 		src/gallium/targets/libgl-xlib/Makefile
 		src/gallium/targets/omx/Makefile
 		src/gallium/targets/opencl/Makefile
+		src/gallium/targets/opencl/mesa.icd
 		src/gallium/targets/osmesa/Makefile
 		src/gallium/targets/osmesa/osmesa.pc
 		src/gallium/targets/pipe-loader/Makefile
@@ -2504,12 +2379,6 @@ else
     echo "        Gallium:         no"
 fi
 
-dnl Shader cache
-echo ""
-echo "        Shader cache:    $enable_shader_cache"
-if test "x$enable_shader_cache" = "xyes"; then
-    echo "        With SHA1 from:  $with_sha1"
-fi
 
 dnl Libraries
 echo ""

docs/relnotes/10.6.0.html

@@ -14,7 +14,7 @@
 <iframe src="../contents.html"></iframe>
 <div class="content">
 
-<h1>Mesa 10.6.0 Release Notes / TBD</h1>
+<h1>Mesa 10.6.0 Release Notes / June 14, 2015</h1>
 
 <p>
 Mesa 10.6.0 is a new development release.
@@ -31,9 +31,10 @@ because compatibility contexts are not supported.
 </p>
 
 
-<h2>MD5 checksums</h2>
+<h2>SHA256 checksums</h2>
 <pre>
-TBD.
+9bc659abdba26202509304f259723aaa4343dba6aac4bd87d5baea11d23c8c63  mesa-10.6.0.tar.gz
+f37e2633978deed02ff0522abc36c709586e2b555fd439a82ab71dce2c866c76  mesa-10.6.0.tar.xz
 </pre>
 
 
@@ -71,7 +72,246 @@ Note: some of the new features are only available with certain drivers.
 
 <h2>Bug fixes</h2>
 
-TBD.
+<p>This list is likely incomplete.</p>
+
+<ul>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=15006">Bug 15006</a> - translate &amp; rotate the line cause Aliasing</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=27007">Bug 27007</a> - Lines disappear with GL_LINE_SMOOTH</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=28832">Bug 28832</a> - piglit/general/line-aa-width fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=45348">Bug 45348</a> - [swrast] piglit fbo-drawbuffers-arbfp regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=60797">Bug 60797</a> - 1px lines in octave plot aliased to 0</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=67564">Bug 67564</a> - HiZ buffers are much larger than necessary</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=69226">Bug 69226</a> - Cannot enable basic shaders with Second Life aborts attempt</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=71591">Bug 71591</a> - Second Life shaders fail to compile (extension declared in middle of shader)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=79202">Bug 79202</a> - valgrind errors in glsl-fs-uniform-array-loop-unroll.shader_test; random code generation</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=81025">Bug 81025</a> - [IVB/BYT Bisected]Piglit spec_ARB_draw_indirect_arb_draw_indirect-draw-elements-prim-restart-ugly fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=82477">Bug 82477</a> - [softpipe] piglit fp-long-alu regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=82668">Bug 82668</a> - Can't set int attributes to certain values on 32-bit</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=82831">Bug 82831</a> - i965: Support GL_ARB_blend_func_extended in SIMD16</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=83962">Bug 83962</a> - [HSW/BYT]Piglit spec_ARB_gpu_shader5_arb_gpu_shader5-emitstreamvertex_nodraw fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=84613">Bug 84613</a> - [G965, bisected] piglit regressions : glslparsertest.glsl2</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86747">Bug 86747</a> - Noise in Football Manager 2014 textures</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86792">Bug 86792</a> - [NVC0] Portal 2 Crashes in Wine</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86811">Bug 86811</a> - [BDW/BSW Bisected]Piglit spec_arb_shading_language_packing_execution_built-in-functions_vs-unpackSnorm4x8 fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86837">Bug 86837</a> - kodi segfault since auxiliary/vl: rework the build of the VL code</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86944">Bug 86944</a> - glsl_parser_extras.cpp&quot;, line 1455: Error: Badly formed expression. (Oracle Studio)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86974">Bug 86974</a> - INTEL_DEBUG=shader_time always asserts in fs_generator::generate_code() when Mesa is built with --enable-debug (= with asserts)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=86980">Bug 86980</a> - [swrast] piglit fp-rfl regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=87258">Bug 87258</a> - [BDW/BSW Bisected]Piglit spec_ARB_shader_atomic_counters_array-indexing fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88246">Bug 88246</a> - Commit 2881b12 causes 43 DrawElements test regressions</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88248">Bug 88248</a> - Calling glClear while there is an occlusion query in progress messes up the results</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88521">Bug 88521</a> - GLBenchmark 2.7 TRex renders with artifacts on Gen8 with !UXA</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88534">Bug 88534</a> - include/c11/threads_posix.h PTHREAD_MUTEX_RECURSIVE_NP not defined</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88561">Bug 88561</a> - [radeonsi][regression,bisected] Depth test/buffer issues in Portal</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88793">Bug 88793</a> - [BDW/BSW Bisected]Piglit/shaders_glsl-max-varyings fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88815">Bug 88815</a> - Incorrect handling of GLSL #line directive</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88883">Bug 88883</a> - ir-a2xx.c: variable changed in assert statement</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88885">Bug 88885</a> - Transform feedback uses incorrect interleaving if a previous draw did not write gl_Position</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88905">Bug 88905</a> - [SNB+ Bisected]Ogles3conform ES3-CTS.gtf.GL3Tests.packed_pixels.packed_pixels fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=88999">Bug 88999</a> - [SKL] Compiz crashes after opening unity dash</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89014">Bug 89014</a> - PIPE_QUERY_GPU_FINISHED is not acting as expected on SI</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89026">Bug 89026</a> - Renderbuffer layered state used for framebuffer completeness test</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89032">Bug 89032</a> - [BDW/BSW/SKL Bisected]Piglit spec_OpenGL_1.1_infinite-spot-light fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89037">Bug 89037</a> - [SKL]Piglit spec_EXT_texture_array_copyteximage_1D_ARRAY_samples=2 sporadically causes GPU hang</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89039">Bug 89039</a> - [SKL]etqw system hang</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89058">Bug 89058</a> - [SKL]Render error in some games (etqw-demo, nexuiz, portal)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89068">Bug 89068</a> - glTexImage2D regression by texstore_rgba switch to _mesa_format_convert</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89069">Bug 89069</a> - Lack of grass in The Talos Principle on radeonsi (native\wine\nine)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89094">Bug 89094</a> - [SNB/IVB/HSW/BYT Bisected]Ogles3conform ES3-CTS.gtf.GL3Tests.shadow.shadow_execution_vert fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89095">Bug 89095</a> - [SNB/IVB/BYT Bisected]Webglc conformance/glsl/functions/glsl-function-mix-float.html fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89112">Bug 89112</a> - u_atomic_test: u_atomic_test.c:124: test_atomic_8bits_bool: Assertion `r == 65 &amp;&amp; &quot;p_atomic_add&quot;' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89118">Bug 89118</a> - [SKL Bisected]many Ogles3conform cases core dumped</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89131">Bug 89131</a> - [Bisected] Graphical corruption in Weston,  shows old framebuffer pieces</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89156">Bug 89156</a> - r300g: GL_COMPRESSED_RED_RGTC1 / ATI1N support broken</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89180">Bug 89180</a> - [IVB regression] Rendering issues in Mass Effect through VMware Workstation</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89210">Bug 89210</a> - GS statistics fail on SNB</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89218">Bug 89218</a> - lower_instructions.cpp:648:48: error: invalid suffix 'd' on floating constant</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89224">Bug 89224</a> - Incorrect rendering of Unigine Valley running in VM on VMware Workstation</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89260">Bug 89260</a> - macros.h:34:25: fatal error: util/u_math.h: No such file or directory</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89292">Bug 89292</a> - [regression,bisected] incomplete screenshots in some cases</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89311">Bug 89311</a> - [regression, bisected] dEQP: Added entry points for glCompressedTextureSubImage*D.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89312">Bug 89312</a> - [regression, bisected] main: Added entry points for CopyTextureSubImage*D. (d6b7c40cecfe01)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89315">Bug 89315</a> - [HSW, regression, bisected] i965/fs: Emit MAD instructions when possible.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89317">Bug 89317</a> - [HSW, regression, bisected] i965: Add LINTERP/CINTERP to can_do_cmod() (d91390634)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89328">Bug 89328</a> - python required to build Mesa release tarballs</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89342">Bug 89342</a> - main/light.c:159:62: error: 'M_PI' undeclared (first use in this function)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89343">Bug 89343</a> - compiler/tests/radeon_compiler_optimize_tests.c:43:3: error: implicit declaration of function ‘fprintf’ [-Werror=implicit-function-declaration]</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89345">Bug 89345</a> - imports.h:452:58: error: expected declaration specifiers or '...' before 'va_list'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89364">Bug 89364</a> - c99_alloca.h:40:22: fatal error: alloca.h: No such file or directory</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89372">Bug 89372</a> - [softpipe] piglit glsl-1.50 generate-zero-primitives regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89387">Bug 89387</a> - Double delete in lp_bld_misc.cpp</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89416">Bug 89416</a> - UE4Editor crash after load project</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89430">Bug 89430</a> - [g965][bisected] arb_copy_image-targets gl_texture* tests fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89433">Bug 89433</a> - GCC 4.2 does not support -Wvla</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89455">Bug 89455</a> - [NVC0/Gallium] Unigine Heaven black and white boxes</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89457">Bug 89457</a> - [BSW Bisected]ogles3conform ES3-CTS.gtf.GL3Tests.shadow.shadow_execution_vert fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89477">Bug 89477</a> - include/no_extern_c.h:47:1: error: template with C linkage</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89508">Bug 89508</a> - Bad int(floatBitsToInt(vec4))</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89530">Bug 89530</a> - FTBFS in loader: missing fstat</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89569">Bug 89569</a> - Papo &amp; Yo crash on startup [HSW]</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89590">Bug 89590</a> - Crash in glLinkProgram with shaders with multiple constant arrays</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89662">Bug 89662</a> - context.c:943: undefined reference to `_glapi_new_nop_table'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89670">Bug 89670</a> - cmod_propagation_test.andnz_one regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89679">Bug 89679</a> - [NV50] Portal/Half-Life 2 will not start (native Steam)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89689">Bug 89689</a> - [Regression] Weston on DRM backend won't start with new version of mesa</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89722">Bug 89722</a> - [ILK Bisected]Ogles2conform/ES2-CTS.gtf.GL.equal.equal_vec2_frag fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89726">Bug 89726</a> - [Bisected] dEQP-GLES3: uniform linking logic in the presence of structs</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89746">Bug 89746</a> - Mesa and LLVM 3.6+ break opengl for genymotion</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89754">Bug 89754</a> - vertexAttrib fails WebGL Conformance test with mesa drivers</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89758">Bug 89758</a> - pow WebGL Conformance test with mesa drivers</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89759">Bug 89759</a> - WebGL OGL ES GLSL conformance test with mesa drivers fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89831">Bug 89831</a> - [r600] r600_asm.c:310:assign_alu_units: Assertion `0' failed.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89899">Bug 89899</a> - nir/nir_lower_tex_projector.c:112: error: unknown field ‘ssa’ specified in initializer</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89957">Bug 89957</a> - vm protection faults in piglit lest: texsubimage cube_map_array pbo</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89960">Bug 89960</a> - [softpipe] piglit copy-pixels regreession</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89961">Bug 89961</a> - [BDW/BSW Bisected]Synmark2_v6 OglDrvRes/OglDrvShComp/OglDrvState/OglPSPom Image Validation fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89963">Bug 89963</a> - lp_bld_debug.cpp:100:31: error: no matching function for call to ‘llvm::raw_ostream::raw_ostream()’</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90000">Bug 90000</a> - [i965 Bisected NIR] Piglit/gglean_fragprog1-z-write_test fail</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90109">Bug 90109</a> - [SNB+ Bisected]Ogles3conform ES3-CTS.shaders.uniform_block.random.basic_arrays.3 fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90114">Bug 90114</a> - [SNB+ Bisected]Ogles3conform ES3-CTS.shaders.struct.uniform.sampler_array_fragment fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90130">Bug 90130</a> - gl_PrimitiveId seems to reset at 340</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90147">Bug 90147</a> - swrast: build error undeclared _SC_PHYS_PAGES on osx</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90149">Bug 90149</a> - [SNB+ Bisected]ES3-CTS.gtf.GL3Tests.uniform_buffer_object.uniform_buffer_object_getactiveuniformsiv_for_nonexistent_uniform_indices fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90153">Bug 90153</a> - [SKL Bisected]ES3-CTS.gtf.GL3Tests.uniform_buffer_object.uniform_buffer_object_all_valid_basic_types fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90167">Bug 90167</a> - [softpipe] piglit depthstencil-default_fb-drawpixels-32f_24_8_rev regression</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90207">Bug 90207</a> - [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90213">Bug 90213</a> - glDrawPixels with GL_COLOR_INDEX never returns.</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90243">Bug 90243</a> - [bisected] regression: spec.!opengl 3_2.get-active-attrib-returns-all-inputs</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90258">Bug 90258</a> - [IVB] spec.glsl-1_10.execution.fs-dfdy-accuracy fails intermittently</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90310">Bug 90310</a> - Fails to build gallium_dri.so at linking stage with clang because of multiple redefinitions</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90350">Bug 90350</a> - [G96] Portal's portal are incorrectly rendered</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90363">Bug 90363</a> - [nv50] HW state is not reset correctly when using a new GL context</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90397">Bug 90397</a> - ARB_program_interface_query: glGetProgramResourceiv() returns wrong value for GL_REFERENCED_BY_*_SHADER prop for GL_UNIFORM for members of an interface block with an instance name</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90466">Bug 90466</a> - arm: linker error ndefined reference to `nir_metadata_preserve'</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90520">Bug 90520</a> - Register spilling clobbers registers used elsewhere in the shader</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90547">Bug 90547</a> - [BDW/BSW/SKL Bisected]Piglit/glean&#64;vertprog1-rsq_test_2_(reciprocal_square_root_of_negative_value) fais</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90580">Bug 90580</a> - [HSW bisected] integer multiplication bug</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90629">Bug 90629</a> - [i965] SIMD16 dual_source_blend assertion `src[i].file != GRF || src[i].width == dst.width' failed</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90749">Bug 90749</a> - [BDW Bisected]dEQP-GLES3.functional.rasterization.fbo.rbo_multisample_max.primitives.lines_wide fails</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90830">Bug 90830</a> - [bsw bisected regression] GPU hang for spec.arb_gpu_shader5.execution.sampler_array_indexing.vs-nonzero-base</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90839">Bug 90839</a> - [10.5.5/10.6 regression, bisected] PBO glDrawPixels no longer using blit fastpath</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90905">Bug 90905</a> - mesa: Finish subdir-objects transition</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=9951">Bug 9951</a> - GL_LINE_SMOOTH and GL_POLYGON_SMOOTH with i965 driver</li>
+
+</ul>
+
 
 <h2>Changes</h2>
 

docs/relnotes/10.6.1.html

@@ -0,0 +1,104 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+  <meta http-equiv="content-type" content="text/html; charset=utf-8">
+  <title>Mesa Release Notes</title>
+  <link rel="stylesheet" type="text/css" href="../mesa.css">
+</head>
+<body>
+
+<div class="header">
+  <h1>The Mesa 3D Graphics Library</h1>
+</div>
+
+<iframe src="../contents.html"></iframe>
+<div class="content">
+
+<h1>Mesa 10.6.1 Release Notes / June 29, 2015</h1>
+
+<p>
+Mesa 10.6.1 is a bug fix release which fixes bugs found since the 10.6.0 release.
+</p>
+<p>
+Mesa 10.6.1 implements the OpenGL 3.3 API, but the version reported by
+glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
+glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 3.3.  OpenGL
+3.3 is <strong>only</strong> available if requested at context creation
+because compatibility contexts are not supported.
+</p>
+
+
+<h2>SHA256 checksums</h2>
+<pre>
+b4cccd4d0eabcc2bca00c3175d3ad88fdda57ffdb883a7998525b873a21fe607  mesa-10.6.1.tar.gz
+6c80a2b647e57c85dc36e609d9aed17f878f0d8e0cf9ace86d14cf604101e1eb  mesa-10.6.1.tar.xz
+</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=90347">Bug 90347</a> - [NVE0+] Failure to insert texbar under some circumstances (causing bad colors in Terasology)</li>
+
+</ul>
+
+
+<h2>Changes</h2>
+
+<p>Anuj Phogat (4):</p>
+<ul>
+  <li>mesa: Handle integer formats in need_rgb_to_luminance_conversion()</li>
+  <li>mesa: Use helper function need_rgb_to_luminance_conversion()</li>
+  <li>mesa: Turn need_rgb_to_luminance_conversion() in to a global function</li>
+  <li>meta: Abort meta path if ReadPixels need rgb to luminance conversion</li>
+</ul>
+
+<p>Ben Widawsky (1):</p>
+<ul>
+  <li>i965/gen9: Implement Push Constant Buffer workaround</li>
+</ul>
+
+<p>Boyan Ding (2):</p>
+<ul>
+  <li>egl/x11: Set version of swrastLoader to 2</li>
+  <li>egl/x11: Remove duplicate call to dri2_x11_add_configs_for_visuals</li>
+</ul>
+
+<p>Emil Velikov (6):</p>
+<ul>
+  <li>docs: Add sha256sums for the 10.6.0 release</li>
+  <li>configure: warn about shared_glapi &amp; xlib-glx only when both are set</li>
+  <li>configure: error out when building backend-less libEGL</li>
+  <li>configure: error out when building libEGL without shared-glapi</li>
+  <li>gbm: do not (over)link against libglapi.so</li>
+  <li>Update version to 10.6.1</li>
+</ul>
+
+<p>Frank Henigman (1):</p>
+<ul>
+  <li>gbm: dlopen libglapi so gbm_create_device works</li>
+</ul>
+
+<p>Ilia Mirkin (9):</p>
+<ul>
+  <li>nvc0/ir: fix collection of first uses for texture barrier insertion</li>
+  <li>nv50,nvc0: clamp uniform size to 64k</li>
+  <li>nvc0/ir: can't have a join on a load with an indirect source</li>
+  <li>glsl: handle conversions to double when comparing param matches</li>
+  <li>glsl: add version checks to conditionals for builtin variable enablement</li>
+  <li>mesa: add GL_PROGRAM_PIPELINE support in KHR_debug calls</li>
+  <li>glsl: binding point is a texture unit, which is a combined space</li>
+  <li>nvc0: always put all tfb bufs into bufctx</li>
+  <li>nv50,nvc0: make sure to pushbuf_refn before putting bo into pushbuf_data</li>
+</ul>
+
+
+</div>
+</body>
+</html>

docs/relnotes/10.6.2.html

@@ -0,0 +1,165 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+  <meta http-equiv="content-type" content="text/html; charset=utf-8">
+  <title>Mesa Release Notes</title>
+  <link rel="stylesheet" type="text/css" href="../mesa.css">
+</head>
+<body>
+
+<div class="header">
+  <h1>The Mesa 3D Graphics Library</h1>
+</div>
+
+<iframe src="../contents.html"></iframe>
+<div class="content">
+
+<h1>Mesa 10.6.2 Release Notes / July 11, 2015</h1>
+
+<p>
+Mesa 10.6.2 is a bug fix release which fixes bugs found since the 10.6.1 release.
+</p>
+<p>
+Mesa 10.6.2 implements the OpenGL 3.3 API, but the version reported by
+glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
+glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 3.3.  OpenGL
+3.3 is <strong>only</strong> available if requested at context creation
+because compatibility contexts are not supported.
+</p>
+
+
+<h2>SHA256 checksums</h2>
+<pre>
+9c7ab9300dda6c912faaaff97995ec1820ba21d114d9cf555f145cbad90995f4  mesa-10.6.2.tar.gz
+05753d3db4212900927b9894221a1669a10f56786e86a7e818b6e18a0817dca9  mesa-10.6.2.tar.xz
+</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=73528">Bug 73528</a> - Deferred lighting in Second Life causes system hiccups and screen flickering</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=80500">Bug 80500</a> - Flickering shadows in unreleased title trace</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=82186">Bug 82186</a> - [r600g] BARTS GPU lockup with minecraft shaders</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=84225">Bug 84225</a> - Allow constant-index-expression sampler array indexing with GLSL-ES &lt; 300</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90537">Bug 90537</a> - radeonsi bo/va conflict on RADEON_GEM_VA (rscreen-&gt;ws-&gt;buffer_from_handle returns NULL)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90873">Bug 90873</a> - Kernel hang, TearFree On, Mate desktop environment</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91022">Bug 91022</a> - [g45 g965 bisected] assertions generated from textureGrad cube samplers fix</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91047">Bug 91047</a> - [SNB Bisected] Messed up Fog in Super Smash Bros. Melee in Dolphin</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91056">Bug 91056</a> - The Bard's Tale (2005, native)  has rendering issues</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91117">Bug 91117</a> - Nimbus (running in wine) has rendering issues, objects are semi-transparent</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91124">Bug 91124</a> - Civilization V (in Wine) has rendering issues: text missing, menu bar corrupted</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91173">Bug 91173</a> - Oddworld: Stranger's Wrath HD: disfigured models in wrong colors</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91226">Bug 91226</a> - Crash in glLinkProgram (NEW)</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91231">Bug 91231</a> - [NV92] Psychonauts (native) segfaults on start when DRI3 enabled</li>
+
+</ul>
+
+
+<h2>Changes</h2>
+
+<p>Chris Wilson (1):</p>
+<ul>
+  <li>loader: Look for any version of currently linked libudev.so</li>
+</ul>
+
+<p>Emil Velikov (2):</p>
+<ul>
+  <li>docs: Add sha256 checksums for the 10.6.1 release</li>
+  <li>Update version to 10.6.2</li>
+</ul>
+
+<p>Ilia Mirkin (8):</p>
+<ul>
+  <li>nv50/ir: propagate modifier to right arg when const-folding mad</li>
+  <li>nv50/ir: fix emission of address reg in 3rd source</li>
+  <li>nv50/ir: copy joinAt when splitting both before and after</li>
+  <li>mesa: reset the source packing when creating temp transfer image</li>
+  <li>nv50/ir: don't emit src2 in immediate form</li>
+  <li>mesa/prog: relative offsets into constbufs are not constant</li>
+  <li>nv50/ir: UCMP arguments are float, so make sure modifiers are applied</li>
+  <li>nvc0: turn sample counts off during blit</li>
+</ul>
+
+<p>Kenneth Graunke (5):</p>
+<ul>
+  <li>i965/fs: Fix ir_txs in emit_texture_gen4_simd16().</li>
+  <li>i965: Reserve more batch space to accomodate Gen6 perfmonitors.</li>
+  <li>i965/vs: Fix matNxM vertex attributes where M != 4.</li>
+  <li>Revert "glsl: clone inputs and outputs during linking"</li>
+  <li>Revert "i965: Delete linked GLSL IR when using NIR."</li>
+</ul>
+
+<p>Marek Olšák (3):</p>
+<ul>
+  <li>r600g: disable single-sample fast color clear due to hangs</li>
+  <li>radeonsi: fix a hang with DrawTransformFeedback on 4 SE chips</li>
+  <li>st/dri: don't set PIPE_BIND_SCANOUT for MSAA surfaces</li>
+</ul>
+
+<p>Mario Kleiner (2):</p>
+<ul>
+  <li>nouveau: Use dup fd as key in drm-winsys hash table to fix ZaphodHeads.</li>
+  <li>winsys/radeon: Use dup fd as key in drm-winsys hash table to fix ZaphodHeads.</li>
+</ul>
+
+<p>Matt Turner (2):</p>
+<ul>
+  <li>i965/fs: Don't mess up stride for uniform integer multiplication.</li>
+  <li>Revert SHA1 additions.</li>
+</ul>
+
+<p>Michel Dänzer (1):</p>
+<ul>
+  <li>winsys/radeon: Unmap GPU VM address range when destroying BO</li>
+</ul>
+
+<p>Mike Stroyan (2):</p>
+<ul>
+  <li>meta: Only change and restore viewport 0 in mesa meta mode</li>
+  <li>i965: allocate at least 1 BLEND_STATE element</li>
+</ul>
+
+<p>Neil Roberts (4):</p>
+<ul>
+  <li>i965/skl: Set the pulls bary bit in 3DSTATE_PS_EXTRA</li>
+  <li>glsl: Add missing check for whether an expression is an add operation</li>
+  <li>glsl: Make sure not to dereference NULL</li>
+  <li>i965: Don't try to print the GLSL IR if it has been freed</li>
+</ul>
+
+<p>Tapani Pälli (8):</p>
+<ul>
+  <li>glsl: clone inputs and outputs during linking</li>
+  <li>i965: Delete linked GLSL IR when using NIR.</li>
+  <li>glsl: Allow dynamic sampler array indexing with GLSL ES &lt; 3.00</li>
+  <li>mesa/glsl: new compiler option EmitNoIndirectSampler</li>
+  <li>i965: use EmitNoIndirectSampler for gen &lt; 7</li>
+  <li>i915: use EmitNoIndirectSampler</li>
+  <li>mesa/st: use EmitNoIndirectSampler if !ARB_gpu_shader5</li>
+  <li>glsl: validate sampler array indexing for 'constant-index-expression'</li>
+</ul>
+
+
+</div>
+</body>
+</html>

docs/relnotes/10.6.3.html

@@ -0,0 +1,106 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+  <meta http-equiv="content-type" content="text/html; charset=utf-8">
+  <title>Mesa Release Notes</title>
+  <link rel="stylesheet" type="text/css" href="../mesa.css">
+</head>
+<body>
+
+<div class="header">
+  <h1>The Mesa 3D Graphics Library</h1>
+</div>
+
+<iframe src="../contents.html"></iframe>
+<div class="content">
+
+<h1>Mesa 10.6.3 Release Notes / July 26, 2015</h1>
+
+<p>
+Mesa 10.6.3 is a bug fix release which fixes bugs found since the 10.6.2 release.
+</p>
+<p>
+Mesa 10.6.3 implements the OpenGL 3.3 API, but the version reported by
+glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
+glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 3.3.  OpenGL
+3.3 is <strong>only</strong> available if requested at context creation
+because compatibility contexts are not supported.
+</p>
+
+
+<h2>SHA256 checksums</h2>
+<pre>
+c27e1e33798e69a6d2d2425aee8ac7b4c0b243066a65dd76cbb182ea31b1c7f2  mesa-10.6.3.tar.gz
+58592e07c350cd2e8969b73fa83048c657a39fe2f13f3b88f5e5818fe2e4676d  mesa-10.6.3.tar.xz
+</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=90728">Bug 90728</a> - dvd playback with vlc and vdpau causes segmentation fault</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91337">Bug 91337</a> - OSMesaGetProcAdress(&quot;OSMesaPixelStore&quot;) returns nil</li>
+
+</ul>
+
+
+<h2>Changes</h2>
+
+<p>Brian Paul (1):</p>
+<ul>
+  <li>osmesa: fix OSMesaPixelsStore typo</li>
+</ul>
+
+<p>Chad Versace (1):</p>
+<ul>
+  <li>mesa: Fix generation of git_sha1.h.tmp for gitlinks</li>
+</ul>
+
+<p>Christian König (2):</p>
+<ul>
+  <li>vl: cleanup video buffer private when the decoder is destroyed</li>
+  <li>st/vdpau: fix mixer size checks</li>
+</ul>
+
+<p>Emil Velikov (3):</p>
+<ul>
+  <li>docs: Add sha256 checksums for the 10.6.2 release</li>
+  <li>auxiliary/vl: use the correct screen index</li>
+  <li>Update version to 10.6.3</li>
+</ul>
+
+<p>Francisco Jerez (1):</p>
+<ul>
+  <li>i965/gen9: Use custom MOCS entries set up by the kernel.</li>
+</ul>
+
+<p>Ilia Mirkin (5):</p>
+<ul>
+  <li>nv50, nvc0: enable at least one color RT if alphatest is enabled</li>
+  <li>nvc0/ir: fix txq on indirect samplers</li>
+  <li>nvc0/ir: don't worry about sampler in txq handling</li>
+  <li>gm107/ir: fix indirect txq emission</li>
+  <li>nv50: fix max level clamping on G80</li>
+</ul>
+
+<p>Kenneth Graunke (1):</p>
+<ul>
+  <li>program: Allow redundant OPTION ARB_fog_* directives.</li>
+</ul>
+
+<p>Rob Clark (1):</p>
+<ul>
+  <li>xa: don't leak fences</li>
+</ul>
+
+
+</div>
+</body>
+</html>

docs/relnotes/10.6.4.html

@@ -0,0 +1,136 @@
+<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
+<html lang="en">
+<head>
+  <meta http-equiv="content-type" content="text/html; charset=utf-8">
+  <title>Mesa Release Notes</title>
+  <link rel="stylesheet" type="text/css" href="../mesa.css">
+</head>
+<body>
+
+<div class="header">
+  <h1>The Mesa 3D Graphics Library</h1>
+</div>
+
+<iframe src="../contents.html"></iframe>
+<div class="content">
+
+<h1>Mesa 10.6.4 Release Notes / August 11, 2015</h1>
+
+<p>
+Mesa 10.6.4 is a bug fix release which fixes bugs found since the 10.6.3 release.
+</p>
+<p>
+Mesa 10.6.4 implements the OpenGL 3.3 API, but the version reported by
+glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
+glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
+Some drivers don't support all the features required in OpenGL 3.3.  OpenGL
+3.3 is <strong>only</strong> available if requested at context creation
+because compatibility contexts are not supported.
+</p>
+
+
+<h2>SHA256 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=73512">Bug 73512</a> - [clover] mesa.icd. should contain full path</li>
+
+<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91290">Bug 91290</a> - SIGSEGV glcpp/glcpp-parse.y:1077</li>
+
+</ul>
+
+
+<h2>Changes</h2>
+
+<p>Anuj Phogat (6):</p>
+<ul>
+  <li>mesa: Turn get_readpixels_transfer_ops() in to a global function</li>
+  <li>meta: Fix transfer operations check in meta pbo path for readpixels</li>
+  <li>meta: Abort meta pbo path if readpixels need signed-unsigned conversion</li>
+  <li>meta: Don't do fragment color clamping in _mesa_meta_pbo_GetTexSubImage</li>
+  <li>mesa: Add a helper function _mesa_need_luminance_to_rgb_conversion()</li>
+  <li>meta: Fix reading luminance texture as rgba in _mesa_meta_pbo_GetTexSubImage()</li>
+</ul>
+
+<p>Ben Widawsky (1):</p>
+<ul>
+  <li>i965/skl: Add production thread counts and URB size</li>
+</ul>
+
+<p>Eduardo Lima Mitev (3):</p>
+<ul>
+  <li>mesa: Fix errors values returned by glShaderBinary()</li>
+  <li>mesa: Validate target before resolving tex obj in glTex(ture)SubImageXD</li>
+  <li>mesa: Fix error returned by glCopyTexImage2D() upon an invalid internal format</li>
+</ul>
+
+<p>Emil Velikov (6):</p>
+<ul>
+  <li>docs: Add checksums for mesa 10.6.3 tarballs</li>
+  <li>configure.ac: do not set HAVE_DRI(23) when libdrm is missing</li>
+  <li>egl/wayland: libdrm is a hard requirement, treat it as such</li>
+  <li>winsys/radeon: don't leak the fd when it is 0</li>
+  <li>bugzilla_mesa.sh: sort the bugs list by number</li>
+  <li>Update version to 10.6.4</li>
+</ul>
+
+<p>Francisco Jerez (1):</p>
+<ul>
+  <li>i965/fs: Fix fs_inst::regs_read() for sources in the ATTR file.</li>
+</ul>
+
+<p>Frank Binns (2):</p>
+<ul>
+  <li>egl/dri: Add error info needed for EGL_EXT_image_dma_buf_import extension</li>
+  <li>egl: Add eglQuerySurface surface type check for EGL_LARGEST_PBUFFER attrib</li>
+</ul>
+
+<p>Igor Gnatenko (1):</p>
+<ul>
+  <li>opencl: use versioned .so in mesa.icd</li>
+</ul>
+
+<p>Ilia Mirkin (1):</p>
+<ul>
+  <li>nvc0: fix geometry program revalidation of clipping params</li>
+</ul>
+
+<p>Kenneth Graunke (1):</p>
+<ul>
+  <li>glsl: Fix a bug where LHS swizzles of swizzles were too small.</li>
+</ul>
+
+<p>Marek Olšák (6):</p>
+<ul>
+  <li>st/mesa: don't call st_validate_state in BlitFramebuffer</li>
+  <li>radeonsi: upload shader rodata after updating scratch relocations</li>
+  <li>st/mesa: don't ignore texture buffer state changes</li>
+  <li>radeonsi: rework how shader pointers to descriptors are set</li>
+  <li>radeonsi: completely rework updating descriptors without CP DMA</li>
+  <li>r600g: fix the CB_SHADER_MASK setup</li>
+</ul>
+
+<p>Samuel Iglesias Gonsalvez (1):</p>
+<ul>
+  <li>glsl/glcpp: fix SIGSEGV when checking error condition for macro redefinition</li>
+</ul>
+
+<p>Samuel Pitoiset (1):</p>
+<ul>
+  <li>nv50: avoid segfault with enabled but unbound vertex attrib</li>
+</ul>
+
+
+</div>
+</body>
+</html>

include/GL/internal/dri_interface.h

@@ -1180,7 +1180,8 @@ enum __DRIChromaSiting {
 };
 
 /**
- * \name Reasons that __DRIimageExtensionRec::createImageFromTexture might fail
+ * \name Reasons that __DRIimageExtensionRec::createImageFromTexture or
+ * __DRIimageExtensionRec::createImageFromDmaBufs might fail
  */
 /*@{*/
 /** Success! */
@@ -1189,11 +1190,14 @@ enum __DRIChromaSiting {
 /** Memory allocation failure */
 #define __DRI_IMAGE_ERROR_BAD_ALLOC     1
 
-/** Client requested an invalid attribute for a texture object  */
+/** Client requested an invalid attribute */
 #define __DRI_IMAGE_ERROR_BAD_MATCH     2
 
 /** Client requested an invalid texture object */
 #define __DRI_IMAGE_ERROR_BAD_PARAMETER 3
+
+/** Client requested an invalid pitch and/or offset */
+#define __DRI_IMAGE_ERROR_BAD_ACCESS    4
 /*@}*/
 
 /**

src/Makefile.am

@@ -72,4 +72,5 @@ noinst_LTLIBRARIES = libglsl_util.la
 libglsl_util_la_SOURCES = \
 	mesa/main/imports.c \
 	mesa/program/prog_hash_table.c \
-	mesa/program/symbol_table.c
+	mesa/program/symbol_table.c \
+	mesa/program/dummy_errors.c

src/egl/drivers/dri2/egl_dri2.c

@@ -1510,6 +1510,10 @@ dri2_create_image_khr_texture_error(int dri_error)
       egl_error = EGL_BAD_PARAMETER;
       break;
 
+   case __DRI_IMAGE_ERROR_BAD_ACCESS:
+      egl_error = EGL_BAD_ACCESS;
+      break;
+
    default:
       assert(0);
       egl_error = EGL_BAD_MATCH;
@@ -2123,13 +2127,11 @@ dri2_bind_wayland_display_wl(_EGLDriver *drv, _EGLDisplay *disp,
    wl_drm_callbacks.authenticate =
       (int(*)(void *, uint32_t)) dri2_dpy->vtbl->authenticate;
 
-#ifdef HAVE_LIBDRM
    if (drmGetCap(dri2_dpy->fd, DRM_CAP_PRIME, &cap) == 0 &&
        cap == (DRM_PRIME_CAP_IMPORT | DRM_PRIME_CAP_EXPORT) &&
        dri2_dpy->image->base.version >= 7 &&
        dri2_dpy->image->createImageFromFds != NULL)
       flags |= WAYLAND_DRM_PRIME;
-#endif
 
    dri2_dpy->wl_server_drm =
 	   wayland_drm_init(wl_dpy, dri2_dpy->device_name,

src/egl/drivers/dri2/platform_x11.c

@@ -1112,7 +1112,7 @@ dri2_initialize_x11_swrast(_EGLDriver *drv, _EGLDisplay *disp)
       goto cleanup_conn;
 
    dri2_dpy->swrast_loader_extension.base.name = __DRI_SWRAST_LOADER;
-   dri2_dpy->swrast_loader_extension.base.version = __DRI_SWRAST_LOADER_VERSION;
+   dri2_dpy->swrast_loader_extension.base.version = 2;
    dri2_dpy->swrast_loader_extension.getDrawableInfo = swrastGetDrawableInfo;
    dri2_dpy->swrast_loader_extension.putImage = swrastPutImage;
    dri2_dpy->swrast_loader_extension.getImage = swrastGetImage;
@@ -1283,11 +1283,6 @@ dri2_initialize_x11_dri2(_EGLDriver *drv, _EGLDisplay *disp)
 
    dri2_x11_setup_swap_interval(dri2_dpy);
 
-   if (dri2_dpy->conn) {
-      if (!dri2_x11_add_configs_for_visuals(dri2_dpy, disp))
-	 goto cleanup_configs;
-   }
-
    disp->Extensions.KHR_image_pixmap = EGL_TRUE;
    disp->Extensions.NOK_swap_region = EGL_TRUE;
    disp->Extensions.NOK_texture_from_pixmap = EGL_TRUE;

src/egl/main/eglcontext.c

@@ -131,7 +131,7 @@ _eglParseContextAttribList(_EGLContext *ctx, _EGLDisplay *dpy,
             break;
          }
 
-         ctx->Flags = val;
+         ctx->Flags |= val;
          break;
 
       case EGL_CONTEXT_OPENGL_PROFILE_MASK_KHR:
@@ -194,7 +194,8 @@ _eglParseContextAttribList(_EGLContext *ctx, _EGLDisplay *dpy,
             break;
          }
 
-         ctx->Flags = EGL_CONTEXT_OPENGL_ROBUST_ACCESS_BIT_KHR;
+         if (val == EGL_TRUE)
+            ctx->Flags |= EGL_CONTEXT_OPENGL_ROBUST_ACCESS_BIT_KHR;
          break;
 
       default:

src/egl/main/eglsurface.c

@@ -309,7 +309,8 @@ _eglQuerySurface(_EGLDriver *drv, _EGLDisplay *dpy, _EGLSurface *surface,
       *value = surface->Config->ConfigID;
       break;
    case EGL_LARGEST_PBUFFER:
-      *value = surface->LargestPbuffer;
+      if (surface->Type == EGL_PBUFFER_BIT)
+         *value = surface->LargestPbuffer;
       break;
    case EGL_TEXTURE_FORMAT:
       /* texture attributes: only for pbuffers, no error otherwise */

src/gallium/auxiliary/cso_cache/cso_context.c

@@ -300,6 +300,8 @@ void cso_destroy_context( struct cso_context *ctx )
    unsigned i, shader;
 
    if (ctx->pipe) {
+      ctx->pipe->set_index_buffer(ctx->pipe, NULL);
+
       ctx->pipe->bind_blend_state( ctx->pipe, NULL );
       ctx->pipe->bind_rasterizer_state( ctx->pipe, NULL );
 
@@ -326,13 +328,18 @@ void cso_destroy_context( struct cso_context *ctx )
 
       ctx->pipe->bind_depth_stencil_alpha_state( ctx->pipe, NULL );
       ctx->pipe->bind_fs_state( ctx->pipe, NULL );
+      ctx->pipe->set_constant_buffer(ctx->pipe, PIPE_SHADER_FRAGMENT, 0, NULL);
       ctx->pipe->bind_vs_state( ctx->pipe, NULL );
+      ctx->pipe->set_constant_buffer(ctx->pipe, PIPE_SHADER_VERTEX, 0, NULL);
       if (ctx->has_geometry_shader) {
          ctx->pipe->bind_gs_state(ctx->pipe, NULL);
+         ctx->pipe->set_constant_buffer(ctx->pipe, PIPE_SHADER_GEOMETRY, 0, NULL);
       }
       if (ctx->has_tessellation) {
          ctx->pipe->bind_tcs_state(ctx->pipe, NULL);
+         ctx->pipe->set_constant_buffer(ctx->pipe, PIPE_SHADER_TESS_CTRL, 0, NULL);
          ctx->pipe->bind_tes_state(ctx->pipe, NULL);
+         ctx->pipe->set_constant_buffer(ctx->pipe, PIPE_SHADER_TESS_EVAL, 0, NULL);
       }
       ctx->pipe->bind_vertex_elements_state( ctx->pipe, NULL );
 

src/gallium/auxiliary/draw/draw_llvm.c

@@ -1965,7 +1965,7 @@ draw_llvm_set_sampler_state(struct draw_context *draw,
       for (i = 0; i < draw->num_samplers[PIPE_SHADER_VERTEX]; i++) {
          struct draw_jit_sampler *jit_sam = &draw->llvm->jit_context.samplers[i];
 
-         if (draw->samplers[i]) {
+         if (draw->samplers[PIPE_SHADER_VERTEX][i]) {
             const struct pipe_sampler_state *s
                = draw->samplers[PIPE_SHADER_VERTEX][i];
             jit_sam->min_lod = s->min_lod;
@@ -1978,7 +1978,7 @@ draw_llvm_set_sampler_state(struct draw_context *draw,
       for (i = 0; i < draw->num_samplers[PIPE_SHADER_GEOMETRY]; i++) {
          struct draw_jit_sampler *jit_sam = &draw->llvm->gs_jit_context.samplers[i];
 
-         if (draw->samplers[i]) {
+         if (draw->samplers[PIPE_SHADER_GEOMETRY][i]) {
             const struct pipe_sampler_state *s
                = draw->samplers[PIPE_SHADER_GEOMETRY][i];
             jit_sam->min_lod = s->min_lod;

src/gallium/auxiliary/util/u_math.h

@@ -42,6 +42,7 @@
 #include "pipe/p_compiler.h"
 
 #include "c99_math.h"
+#include <assert.h>
 #include <float.h>
 #include <stdarg.h>
 

src/gallium/auxiliary/vl/vl_mpeg12_decoder.c

@@ -84,6 +84,9 @@ static const unsigned const_empty_block_mask_420[3][2][2] = {
 
 struct video_buffer_private
 {
+   struct list_head list;
+   struct pipe_video_buffer *video_buffer;
+
    struct pipe_sampler_view *sampler_view_planes[VL_NUM_COMPONENTS];
    struct pipe_surface      *surfaces[VL_MAX_SURFACES];
 
@@ -99,6 +102,8 @@ destroy_video_buffer_private(void *private)
    struct video_buffer_private *priv = private;
    unsigned i;
 
+   list_del(&priv->list);
+
    for (i = 0; i < VL_NUM_COMPONENTS; ++i)
       pipe_sampler_view_reference(&priv->sampler_view_planes[i], NULL);
 
@@ -126,6 +131,9 @@ get_video_buffer_private(struct vl_mpeg12_decoder *dec, struct pipe_video_buffer
 
    priv = CALLOC_STRUCT(video_buffer_private);
 
+   list_add(&priv->list, &dec->buffer_privates);
+   priv->video_buffer = buf;
+
    sv = buf->get_sampler_view_planes(buf);
    for (i = 0; i < VL_NUM_COMPONENTS; ++i)
       if (sv[i])
@@ -141,6 +149,18 @@ get_video_buffer_private(struct vl_mpeg12_decoder *dec, struct pipe_video_buffer
    return priv;
 }
 
+static void
+free_video_buffer_privates(struct vl_mpeg12_decoder *dec)
+{
+   struct video_buffer_private *priv, *next;
+
+   LIST_FOR_EACH_ENTRY_SAFE(priv, next, &dec->buffer_privates, list) {
+      struct pipe_video_buffer *buf = priv->video_buffer;
+
+      vl_video_buffer_set_associated_data(buf, &dec->base, NULL, NULL);
+   }
+}
+
 static bool
 init_zscan_buffer(struct vl_mpeg12_decoder *dec, struct vl_mpeg12_buffer *buffer)
 {
@@ -464,6 +484,8 @@ vl_mpeg12_destroy(struct pipe_video_codec *decoder)
 
    assert(decoder);
 
+   free_video_buffer_privates(dec);
+
    /* Asserted in softpipe_delete_fs_state() for some reason */
    dec->context->bind_vs_state(dec->context, NULL);
    dec->context->bind_fs_state(dec->context, NULL);
@@ -1187,6 +1209,8 @@ vl_create_mpeg12_decoder(struct pipe_context *context,
    if (!init_pipe_state(dec))
       goto error_pipe_state;
 
+   list_inithead(&dec->buffer_privates);
+
    return &dec->base;
 
 error_pipe_state:

src/gallium/auxiliary/vl/vl_mpeg12_decoder.h

@@ -30,6 +30,8 @@
 
 #include "pipe/p_video_codec.h"
 
+#include "util/list.h"
+
 #include "vl_mpeg12_bitstream.h"
 #include "vl_zscan.h"
 #include "vl_idct.h"
@@ -77,6 +79,8 @@ struct vl_mpeg12_decoder
 
    unsigned current_buffer;
    struct vl_mpeg12_buffer *dec_buffers[4];
+
+   struct list_head buffer_privates;
 };
 
 struct vl_mpeg12_buffer

src/gallium/auxiliary/vl/vl_winsys_dri.c

@@ -293,6 +293,16 @@ vl_screen_get_private(struct vl_screen *vscreen)
    return vscreen;
 }
 
+static xcb_screen_t *
+get_xcb_screen(xcb_screen_iterator_t iter, int screen)
+{
+    for (; iter.rem; --screen, xcb_screen_next(&iter))
+        if (screen == 0)
+            return iter.data;
+
+    return NULL;
+}
+
 struct vl_screen*
 vl_screen_create(Display *display, int screen)
 {
@@ -334,8 +344,7 @@ vl_screen_create(Display *display, int screen)
       goto free_query;
 
    s = xcb_setup_roots_iterator(xcb_get_setup(scrn->conn));
-   while (screen--)
-	xcb_screen_next(&s);
+
    driverType = XCB_DRI2_DRIVER_TYPE_DRI;
 #ifdef DRI2DriverPrimeShift
    {
@@ -351,7 +360,7 @@ vl_screen_create(Display *display, int screen)
    }
 #endif
 
-   connect_cookie = xcb_dri2_connect_unchecked(scrn->conn, s.data->root, driverType);
+   connect_cookie = xcb_dri2_connect_unchecked(scrn->conn, get_xcb_screen(s, screen)->root, driverType);
    connect = xcb_dri2_connect_reply(scrn->conn, connect_cookie, NULL);
    if (connect == NULL || connect->driver_name_length + connect->device_name_length == 0)
       goto free_connect;
@@ -370,7 +379,7 @@ vl_screen_create(Display *display, int screen)
    if (drmGetMagic(fd, &magic))
       goto free_connect;
 
-   authenticate_cookie = xcb_dri2_authenticate_unchecked(scrn->conn, s.data->root, magic);
+   authenticate_cookie = xcb_dri2_authenticate_unchecked(scrn->conn, get_xcb_screen(s, screen)->root, magic);
    authenticate = xcb_dri2_authenticate_reply(scrn->conn, authenticate_cookie, NULL);
 
    if (authenticate == NULL || !authenticate->authenticated)

src/gallium/drivers/freedreno/Makefile.am

@@ -21,6 +21,8 @@ libfreedreno_la_SOURCES = \
 
 noinst_PROGRAMS = ir3_compiler
 
+# XXX: Required due to the C++ sources in libnir/libglsl_util
+nodist_EXTRA_ir3_compiler_SOURCES = dummy.cpp
 ir3_compiler_SOURCES = \
 	ir3/ir3_cmdline.c
 
@@ -29,7 +31,6 @@ ir3_compiler_LDADD = \
 	../../auxiliary/libgallium.la \
 	$(top_builddir)/src/glsl/libnir.la \
 	$(top_builddir)/src/libglsl_util.la \
-	-lstdc++ \
 	$(top_builddir)/src/util/libmesautil.la \
 	$(GALLIUM_COMMON_LIB_DEPS) \
 	$(FREEDRENO_LIBS)

src/gallium/drivers/freedreno/a3xx/fd3_program.c

@@ -413,12 +413,15 @@ fd3_program_emit(struct fd_ringbuffer *ring, struct fd3_emit *emit,
 				}
 			}
 
-			/* TODO: Figure out if there's a way to make it spit out 0's and
-			 * 1's for the .z and .w components.
+			/* Replace the .xy coordinates with S/T from the point sprite. Set
+			 * interpolation bits for .zw such that they become .01
 			 */
-			if (emit->sprite_coord_enable & (1 << sem2idx(fp->inputs[j].semantic)))
+			if (emit->sprite_coord_enable & (1 << sem2idx(fp->inputs[j].semantic))) {
 				vpsrepl[inloc / 16] |= (emit->sprite_coord_mode ? 0x0d : 0x09)
 					<< ((inloc % 16) * 2);
+				vinterp[(inloc + 2) / 16] |= 2 << (((inloc + 2) % 16) * 2);
+				vinterp[(inloc + 3) / 16] |= 3 << (((inloc + 3) % 16) * 2);
+			}
 		}
 
 		OUT_PKT0(ring, REG_A3XX_VPC_ATTR, 2);

src/gallium/drivers/freedreno/ir3/ir3_cmdline.c

@@ -360,11 +360,3 @@ int main(int argc, char **argv)
 	}
 	dump_info(&v, info);
 }
-
-void _mesa_error_no_memory(const char *caller);
-
-void
-_mesa_error_no_memory(const char *caller)
-{
-	fprintf(stderr, "Mesa error: out of memory in %s", caller);
-}

src/gallium/drivers/freedreno/ir3/ir3_compiler.c

@@ -1615,7 +1615,7 @@ trans_samp(const struct instr_translater *t,
 			instr->cat1.src_type = type_mov;
 			instr->cat1.dst_type = type_mov;
 			add_dst_reg(ctx, instr, &tmp_dst, i);
-			add_src_reg(ctx, instr, &zero, 0);
+			add_src_reg(ctx, instr, &zero, zero.SwizzleX);
 			i++;
 		}
 		if (tgt->array) {
@@ -1669,15 +1669,18 @@ trans_samp(const struct instr_translater *t,
 	 */
 	if (inst->Instruction.Opcode == TGSI_OPCODE_TXD) {
 		while (collect->regs_count < 5)
-			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), &zero, 0);
+			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),
+					&zero, zero.SwizzleX);
 		for (i = 0; i < tgt->dims; i++)
 			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), dpdx, i);
 		if (tgt->dims < 2)
-			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), &zero, 0);
+			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),
+					&zero, zero.SwizzleX);
 		for (i = 0; i < tgt->dims; i++)
 			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), dpdy, i);
 		if (tgt->dims < 2)
-			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), &zero, 0);
+			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),
+					&zero, zero.SwizzleX);
 		tinf.src_wrmask |= ((1 << (2 * MAX2(tgt->dims, 2))) - 1) << 4;
 	}
 
@@ -1700,7 +1703,8 @@ trans_samp(const struct instr_translater *t,
 			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),
 					offset, i);
 		if (tgt->dims < 2)
-			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA), &zero, 0);
+			ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),
+					&zero, zero.SwizzleX);
 	}
 	if (inst->Instruction.Opcode == TGSI_OPCODE_TXB2)
 		ssa_src(ctx, ir3_reg_create(collect, 0, IR3_REG_SSA),

src/gallium/drivers/llvmpipe/lp_query.c

@@ -315,7 +315,7 @@ llvmpipe_check_render_cond(struct llvmpipe_context *lp)
 
    b = pipe->get_query_result(pipe, lp->render_cond_query, wait, (void*)&result);
    if (b)
-      return (!result == lp->render_cond_cond);
+      return ((!result) == lp->render_cond_cond);
    else
       return TRUE;
 }

src/gallium/drivers/nouveau/codegen/nv50_ir_bb.cpp

@@ -332,6 +332,9 @@ BasicBlock::splitBefore(Instruction *insn, bool attach)
    BasicBlock *bb = new BasicBlock(func);
    assert(!insn || insn->op != OP_PHI);
 
+   bb->joinAt = joinAt;
+   joinAt = NULL;
+
    splitCommon(insn, bb, attach);
    return bb;
 }

src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gm107.cpp

@@ -2437,8 +2437,14 @@ CodeEmitterGM107::emitTXQ()
       break;
    }
 
-   emitInsn (0xdf4a0000);
-   emitField(0x24, 13, insn->tex.r);
+   if (insn->tex.rIndirectSrc >= 0) {
+      emitInsn (0xdf500000);
+   } else {
+      emitInsn (0xdf480000);
+      emitField(0x24, 13, insn->tex.r);
+   }
+
+   emitField(0x31, 1, insn->tex.liveOnly);
    emitField(0x1f, 4, insn->tex.mask);
    emitField(0x16, 6, type);
    emitGPR  (0x08, insn->src(0));

src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nv50.cpp

@@ -499,10 +499,14 @@ CodeEmitterNV50::emitForm_MAD(const Instruction *i)
    setSrc(i, 2, 2);
 
    if (i->getIndirect(0, 0)) {
-      assert(!i->getIndirect(1, 0));
+      assert(!i->srcExists(1) || !i->getIndirect(1, 0));
+      assert(!i->srcExists(2) || !i->getIndirect(2, 0));
       setAReg16(i, 0);
-   } else {
+   } else if (i->srcExists(1) && i->getIndirect(1, 0)) {
+      assert(!i->srcExists(2) || !i->getIndirect(2, 0));
       setAReg16(i, 1);
+   } else {
+      setAReg16(i, 2);
    }
 }
 
@@ -546,7 +550,7 @@ CodeEmitterNV50::emitForm_MUL(const Instruction *i)
 }
 
 // usual immediate form
-// - 1 to 3 sources where last is immediate (rir, gir)
+// - 1 to 3 sources where second is immediate (rir, gir)
 // - no address or predicate possible
 void
 CodeEmitterNV50::emitForm_IMM(const Instruction *i)
@@ -562,7 +566,7 @@ CodeEmitterNV50::emitForm_IMM(const Instruction *i)
    if (Target::operationSrcNr[i->op] > 1) {
       setSrc(i, 0, 0);
       setImmediate(i, 1);
-      setSrc(i, 2, 1);
+      // If there is another source, it has to be the same as the dest reg.
    } else {
       setImmediate(i, 0);
    }

src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp

@@ -434,7 +434,6 @@ nv50_ir::DataType Instruction::inferSrcType() const
    case TGSI_OPCODE_USLT:
    case TGSI_OPCODE_USNE:
    case TGSI_OPCODE_USHR:
-   case TGSI_OPCODE_UCMP:
    case TGSI_OPCODE_ATOMUADD:
    case TGSI_OPCODE_ATOMXCHG:
    case TGSI_OPCODE_ATOMCAS:
@@ -1356,18 +1355,20 @@ Converter::srcToSym(tgsi::Instruction::SrcRegister src, int c)
 {
    const int swz = src.getSwizzle(c);
 
+   /* TODO: Use Array ID when it's available for the index */
    return makeSym(src.getFile(),
                   src.is2D() ? src.getIndex(1) : 0,
-                  src.isIndirect(0) ? -1 : src.getIndex(0), swz,
+                  src.getIndex(0), swz,
                   src.getIndex(0) * 16 + swz * 4);
 }
 
 Symbol *
 Converter::dstToSym(tgsi::Instruction::DstRegister dst, int c)
 {
+   /* TODO: Use Array ID when it's available for the index */
    return makeSym(dst.getFile(),
                   dst.is2D() ? dst.getIndex(1) : 0,
-                  dst.isIndirect(0) ? -1 : dst.getIndex(0), c,
+                  dst.getIndex(0), c,
                   dst.getIndex(0) * 16 + c * 4);
 }
 
@@ -1696,6 +1697,7 @@ Converter::insertConvergenceOps(BasicBlock *conv, BasicBlock *fork)
    join->fixed = 1;
    conv->insertHead(join);
 
+   assert(!fork->joinAt);
    fork->joinAt = new_FlowInstruction(func, OP_JOINAT, conv);
    fork->insertBefore(fork->getExit(), fork->joinAt);
 }
@@ -1737,7 +1739,7 @@ Converter::handleTXQ(Value *dst0[4], enum TexQuery query)
    }
    tex->setSrc((c = 0), fetchSrc(0, 0)); // mip level
 
-   setTexRS(tex, c, 1, -1);
+   setTexRS(tex, ++c, 1, -1);
 
    bb->insertTail(tex);
 }
@@ -2578,6 +2580,8 @@ Converter::handleInstruction(const struct tgsi_full_instruction *insn)
       }
       break;
    case TGSI_OPCODE_UCMP:
+      srcTy = TYPE_U32;
+      /* fallthrough */
    case TGSI_OPCODE_CMP:
       FOR_EACH_DST_ENABLED_CHANNEL(0, c, tgsi) {
          src0 = fetchSrc(0, c);

src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_gm107.cpp

@@ -240,6 +240,7 @@ GM107LoweringPass::visit(Instruction *i)
             Value *ptr = bld.mkOp2v(OP_SHL, TYPE_U32, bld.getSSA(),
                                     i->getIndirect(0, 0), bld.mkImm(4));
             i->setIndirect(0, 0, ptr);
+            i->op = OP_VFETCH;
          } else {
             i->op = OP_VFETCH;
             assert(prog->getType() != Program::TYPE_FRAGMENT); // INTERP

src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nv50.cpp

@@ -871,6 +871,7 @@ NV50LoweringPreSSA::handleTXL(TexInstruction *i)
    BasicBlock *joinBB = i->bb->splitAfter(i);
 
    bld.setPosition(currBB, true);
+   assert(!currBB->joinAt);
    currBB->joinAt = bld.mkFlow(OP_JOINAT, joinBB, CC_ALWAYS, NULL);
 
    for (int l = 0; l <= 3; ++l) {

src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp

@@ -100,8 +100,7 @@ void
 NVC0LegalizeSSA::handleFTZ(Instruction *i)
 {
    // Only want to flush float inputs
-   if (i->sType != TYPE_F32)
-      return;
+   assert(i->sType == TYPE_F32);
 
    // If we're already flushing denorms (and NaN's) to zero, no need for this.
    if (i->dnz)
@@ -129,7 +128,7 @@ NVC0LegalizeSSA::visit(BasicBlock *bb)
    Instruction *next;
    for (Instruction *i = bb->getEntry(); i; i = next) {
       next = i->next;
-      if (i->dType == TYPE_F32) {
+      if (i->sType == TYPE_F32) {
          if (prog->getType() != Program::TYPE_COMPUTE)
             handleFTZ(i);
          continue;
@@ -169,7 +168,7 @@ NVC0LegalizePostRA::insnDominatedBy(const Instruction *later,
 
 void
 NVC0LegalizePostRA::addTexUse(std::list<TexUse> &uses,
-                              Instruction *usei, const Instruction *insn)
+                              Instruction *usei, const Instruction *texi)
 {
    bool add = true;
    for (std::list<TexUse>::iterator it = uses.begin();
@@ -184,7 +183,7 @@ NVC0LegalizePostRA::addTexUse(std::list<TexUse> &uses,
          ++it;
    }
    if (add)
-      uses.push_back(TexUse(usei, insn));
+      uses.push_back(TexUse(usei, texi));
 }
 
 void
@@ -196,7 +195,8 @@ NVC0LegalizePostRA::findOverwritingDefs(const Instruction *texi,
    while (insn->op == OP_MOV && insn->getDef(0)->equals(insn->getSrc(0)))
       insn = insn->getSrc(0)->getUniqueInsn();
 
-   if (!insn->bb->reachableBy(texi->bb, term))
+   // NOTE: the tex itself is, of course, not an overwriting definition
+   if (insn == texi || !insn->bb->reachableBy(texi->bb, term))
       return;
 
    switch (insn->op) {
@@ -244,7 +244,12 @@ NVC0LegalizePostRA::findFirstUses(
          visited.insert(usei);
 
          if (usei->op == OP_PHI || usei->op == OP_UNION) {
-            // need a barrier before WAW cases
+            // need a barrier before WAW cases, like:
+            //   %r0 = tex
+            //   if ...
+            //     texbar <- is required or tex might replace x again
+            //     %r1 = x <- overwriting def
+            //   %r2 = phi %r0, %r1
             for (int s = 0; usei->srcExists(s); ++s) {
                Instruction *defi = usei->getSrc(s)->getUniqueInsn();
                if (defi && &usei->src(s) != *u)
@@ -263,7 +268,7 @@ NVC0LegalizePostRA::findFirstUses(
              usei->subOp != NV50_IR_SUBOP_MOV_FINAL) {
             findFirstUses(texi, usei, uses, visited);
          } else {
-            addTexUse(uses, usei, insn);
+            addTexUse(uses, usei, texi);
          }
       }
    }
@@ -951,7 +956,47 @@ NVC0LoweringPass::handleTXD(TexInstruction *txd)
 bool
 NVC0LoweringPass::handleTXQ(TexInstruction *txq)
 {
-   // TODO: indirect resource/sampler index
+   if (txq->tex.rIndirectSrc < 0)
+      return true;
+
+   Value *ticRel = txq->getIndirectR();
+   const int chipset = prog->getTarget()->getChipset();
+
+   txq->setIndirectS(NULL);
+   txq->tex.sIndirectSrc = -1;
+
+   assert(ticRel);
+
+   if (chipset < NVISA_GK104_CHIPSET) {
+      LValue *src = new_LValue(func, FILE_GPR); // 0xttxsaaaa
+
+      txq->setSrc(txq->tex.rIndirectSrc, NULL);
+      if (txq->tex.r)
+         ticRel = bld.mkOp2v(OP_ADD, TYPE_U32, bld.getScratch(),
+                             ticRel, bld.mkImm(txq->tex.r));
+
+      bld.mkOp2(OP_SHL, TYPE_U32, src, ticRel, bld.mkImm(0x17));
+
+      txq->moveSources(0, 1);
+      txq->setSrc(0, src);
+   } else {
+      Value *hnd = loadTexHandle(
+            bld.mkOp2v(OP_SHL, TYPE_U32, bld.getSSA(),
+                       txq->getIndirectR(), bld.mkImm(2)),
+            txq->tex.r);
+      txq->tex.r = 0xff;
+      txq->tex.s = 0x1f;
+
+      if (chipset < NVISA_GM107_CHIPSET) {
+         txq->setIndirectR(NULL);
+         txq->moveSources(0, 1);
+         txq->setSrc(0, hnd);
+         txq->tex.rIndirectSrc = 0;
+      } else {
+         txq->setIndirectR(hnd);
+      }
+   }
+
    return true;
 }
 
@@ -1751,6 +1796,7 @@ NVC0LoweringPass::visit(Instruction *i)
             Value *ptr = bld.mkOp2v(OP_SHL, TYPE_U32, bld.getSSA(),
                                     i->getIndirect(0, 0), bld.mkImm(4));
             i->setIndirect(0, 0, ptr);
+            i->op = OP_VFETCH;
          } else {
             i->op = OP_VFETCH;
             assert(prog->getType() != Program::TYPE_FRAGMENT); // INTERP

src/gallium/drivers/nouveau/codegen/nv50_ir_peephole.cpp

@@ -236,6 +236,9 @@ LoadPropagation::visit(BasicBlock *bb)
       if (i->op == OP_CALL) // calls have args as sources, they must be in regs
          continue;
 
+      if (i->op == OP_PFETCH) // pfetch expects arg1 to be a reg
+         continue;
+
       if (i->srcExists(1))
          checkSwapSrc01(i);
 
@@ -574,6 +577,11 @@ ConstantFolding::expr(Instruction *i,
    case OP_POPCNT:
       res.data.u32 = util_bitcount(a->data.u32 & b->data.u32);
       break;
+   case OP_PFETCH:
+      // The two arguments to pfetch are logically added together. Normally
+      // the second argument will not be constant, but that can happen.
+      res.data.u32 = a->data.u32 + b->data.u32;
+      break;
    default:
       return;
    }
@@ -588,12 +596,17 @@ ConstantFolding::expr(Instruction *i,
 
    i->getSrc(0)->reg.data = res.data;
 
-   if (i->op == OP_MAD || i->op == OP_FMA) {
+   switch (i->op) {
+   case OP_MAD:
+   case OP_FMA: {
       i->op = OP_ADD;
 
+      /* Move the immediate to the second arg, otherwise the ADD operation
+       * won't be emittable
+       */
       i->setSrc(1, i->getSrc(0));
-      i->src(1).mod = i->src(2).mod;
       i->setSrc(0, i->getSrc(2));
+      i->src(0).mod = i->src(2).mod;
       i->setSrc(2, NULL);
 
       ImmediateValue src0;
@@ -603,8 +616,14 @@ ConstantFolding::expr(Instruction *i,
          bld.setPosition(i, false);
          i->setSrc(1, bld.loadImm(NULL, res.data.u32));
       }
-   } else {
+      break;
+   }
+   case OP_PFETCH:
+      // Leave PFETCH alone... we just folded its 2 args into 1.
+      break;
+   default:
       i->op = i->saturate ? OP_SAT : OP_MOV; /* SAT handled by unary() */
+      break;
    }
    i->subOp = 0;
 }
@@ -2216,7 +2235,7 @@ FlatteningPass::visit(BasicBlock *bb)
              insn->op != OP_LINTERP && // probably just nve4
              insn->op != OP_PINTERP && // probably just nve4
              ((insn->op != OP_LOAD && insn->op != OP_STORE) ||
-              typeSizeof(insn->dType) <= 4) &&
+              (typeSizeof(insn->dType) <= 4 && !insn->src(0).isIndirect(0))) &&
              !insn->isNop()) {
             insn->join = 1;
             bb->remove(bb->getExit());

src/gallium/drivers/nouveau/nv30/nv30_clear.c

@@ -58,7 +58,7 @@ nv30_clear(struct pipe_context *pipe, unsigned buffers,
    struct pipe_framebuffer_state *fb = &nv30->framebuffer;
    uint32_t colr = 0, zeta = 0, mode = 0;
 
-   if (!nv30_state_validate(nv30, TRUE))
+   if (!nv30_state_validate(nv30, NV30_NEW_FRAMEBUFFER | NV30_NEW_SCISSOR, TRUE))
       return;
 
    if (buffers & PIPE_CLEAR_COLOR && fb->nr_cbufs) {

src/gallium/drivers/nouveau/nv30/nv30_context.h

@@ -204,7 +204,7 @@ void
 nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info);
 
 boolean
-nv30_state_validate(struct nv30_context *nv30, boolean hwtnl);
+nv30_state_validate(struct nv30_context *nv30, uint32_t mask, boolean hwtnl);
 
 void
 nv30_state_release(struct nv30_context *nv30);

src/gallium/drivers/nouveau/nv30/nv30_draw.c

@@ -71,12 +71,12 @@ nv30_render_allocate_vertices(struct vbuf_render *render,
    struct nv30_render *r = nv30_render(render);
    struct nv30_context *nv30 = r->nv30;
 
-   r->length = vertex_size * nr_vertices;
+   r->length = (uint32_t)vertex_size * (uint32_t)nr_vertices;
 
    if (r->offset + r->length >= render->max_vertex_buffer_bytes) {
       pipe_resource_reference(&r->buffer, NULL);
       r->buffer = pipe_buffer_create(&nv30->screen->base.base,
-                                     PIPE_BIND_VERTEX_BUFFER, 0,
+                                     PIPE_BIND_VERTEX_BUFFER, PIPE_USAGE_STREAM,
                                      render->max_vertex_buffer_bytes);
       if (!r->buffer)
          return FALSE;
@@ -91,10 +91,14 @@ static void *
 nv30_render_map_vertices(struct vbuf_render *render)
 {
    struct nv30_render *r = nv30_render(render);
-   char *map = pipe_buffer_map(&r->nv30->base.pipe, r->buffer,
-                               PIPE_TRANSFER_WRITE |
-                               PIPE_TRANSFER_UNSYNCHRONIZED, &r->transfer);
-   return map + r->offset;
+   char *map = pipe_buffer_map_range(
+         &r->nv30->base.pipe, r->buffer,
+         r->offset, r->length,
+         PIPE_TRANSFER_WRITE |
+         PIPE_TRANSFER_DISCARD_RANGE,
+         &r->transfer);
+   assert(map);
+   return map;
 }
 
 static void
@@ -103,6 +107,7 @@ nv30_render_unmap_vertices(struct vbuf_render *render,
 {
    struct nv30_render *r = nv30_render(render);
    pipe_buffer_unmap(&r->nv30->base.pipe, r->transfer);
+   r->transfer = NULL;
 }
 
 static void
@@ -126,10 +131,10 @@ nv30_render_draw_elements(struct vbuf_render *render,
    for (i = 0; i < r->vertex_info.num_attribs; i++) {
       PUSH_RESRC(push, NV30_3D(VTXBUF(i)), BUFCTX_VTXTMP,
                        nv04_resource(r->buffer), r->offset + r->vtxptr[i],
-                       NOUVEAU_BO_LOW | NOUVEAU_BO_RD, 0, 0);
+                       NOUVEAU_BO_LOW | NOUVEAU_BO_RD, 0, NV30_3D_VTXBUF_DMA1);
    }
 
-   if (!nv30_state_validate(nv30, FALSE))
+   if (!nv30_state_validate(nv30, ~0, FALSE))
       return;
 
    BEGIN_NV04(push, NV30_3D(VERTEX_BEGIN_END), 1);
@@ -171,10 +176,10 @@ nv30_render_draw_arrays(struct vbuf_render *render, unsigned start, uint nr)
    for (i = 0; i < r->vertex_info.num_attribs; i++) {
       PUSH_RESRC(push, NV30_3D(VTXBUF(i)), BUFCTX_VTXTMP,
                        nv04_resource(r->buffer), r->offset + r->vtxptr[i],
-                       NOUVEAU_BO_LOW | NOUVEAU_BO_RD, 0, 0);
+                       NOUVEAU_BO_LOW | NOUVEAU_BO_RD, 0, NV30_3D_VTXBUF_DMA1);
    }
 
-   if (!nv30_state_validate(nv30, FALSE))
+   if (!nv30_state_validate(nv30, ~0, FALSE))
       return;
 
    BEGIN_NV04(push, NV30_3D(VERTEX_BEGIN_END), 1);
@@ -213,22 +218,24 @@ static const struct {
    [TGSI_SEMANTIC_BCOLOR  ] = { EMIT_4F, INTERP_LINEAR     , 1, 3, 0x00000004 },
    [TGSI_SEMANTIC_FOG     ] = { EMIT_4F, INTERP_PERSPECTIVE, 5, 5, 0x00000010 },
    [TGSI_SEMANTIC_PSIZE   ] = { EMIT_1F_PSIZE, INTERP_POS  , 6, 6, 0x00000020 },
-   [TGSI_SEMANTIC_GENERIC ] = { EMIT_4F, INTERP_PERSPECTIVE, 8, 7, 0x00004000 }
+   [TGSI_SEMANTIC_TEXCOORD] = { EMIT_4F, INTERP_PERSPECTIVE, 8, 7, 0x00004000 },
 };
 
 static boolean
 vroute_add(struct nv30_render *r, uint attrib, uint sem, uint *idx)
 {
-   struct pipe_screen *pscreen = &r->nv30->screen->base.base;
+   struct nv30_screen *screen = r->nv30->screen;
    struct nv30_fragprog *fp = r->nv30->fragprog.program;
    struct vertex_info *vinfo = &r->vertex_info;
    enum pipe_format format;
    uint emit = EMIT_OMIT;
    uint result = *idx;
 
-   if (sem == TGSI_SEMANTIC_GENERIC && result >= 8) {
-      for (result = 0; result < 8; result++) {
-         if (fp->texcoord[result] == *idx) {
+   if (sem == TGSI_SEMANTIC_GENERIC) {
+      uint num_texcoords = (screen->eng3d->oclass < NV40_3D_CLASS) ? 8 : 10;
+      for (result = 0; result < num_texcoords; result++) {
+         if (fp->texcoord[result] == *idx + 8) {
+            sem = TGSI_SEMANTIC_TEXCOORD;
             emit = vroute[sem].emit;
             break;
          }
@@ -243,11 +250,11 @@ vroute_add(struct nv30_render *r, uint attrib, uint sem, uint *idx)
    draw_emit_vertex_attr(vinfo, emit, vroute[sem].interp, attrib);
    format = draw_translate_vinfo_format(emit);
 
-   r->vtxfmt[attrib] = nv30_vtxfmt(pscreen, format)->hw;
-   r->vtxptr[attrib] = vinfo->size | NV30_3D_VTXBUF_DMA1;
+   r->vtxfmt[attrib] = nv30_vtxfmt(&screen->base.base, format)->hw;
+   r->vtxptr[attrib] = vinfo->size;
    vinfo->size += draw_translate_vinfo_size(emit);
 
-   if (nv30_screen(pscreen)->eng3d->oclass < NV40_3D_CLASS) {
+   if (screen->eng3d->oclass < NV40_3D_CLASS) {
       r->vtxprog[attrib][0] = 0x001f38d8;
       r->vtxprog[attrib][1] = 0x0080001b | (attrib << 9);
       r->vtxprog[attrib][2] = 0x0836106c;
@@ -259,7 +266,12 @@ vroute_add(struct nv30_render *r, uint attrib, uint sem, uint *idx)
       r->vtxprog[attrib][3] = 0x6041ff80 | (result + vroute[sem].vp40) << 2;
    }
 
-   *idx = vroute[sem].ow40 << result;
+   if (result < 8)
+      *idx = vroute[sem].ow40 << result;
+   else {
+      assert(sem == TGSI_SEMANTIC_TEXCOORD);
+      *idx = 0x00001000 << (result - 8);
+   }
    return TRUE;
 }
 
@@ -313,7 +325,7 @@ nv30_render_validate(struct nv30_context *nv30)
 
    while (pntc && attrib < 16) {
       uint index = ffs(pntc) - 1; pntc &= ~(1 << index);
-      if (vroute_add(r, attrib, TGSI_SEMANTIC_GENERIC, &index)) {
+      if (vroute_add(r, attrib, TGSI_SEMANTIC_TEXCOORD, &index)) {
          vp_attribs |= (1 << attrib++);
          vp_results |= index;
       }
@@ -398,17 +410,17 @@ nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
       if (nv30->vertprog.constbuf) {
          void *map = nv04_resource(nv30->vertprog.constbuf)->data;
          draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX, 0,
-                                         map, nv30->vertprog.constbuf_nr);
+                                         map, nv30->vertprog.constbuf_nr * 16);
+      } else {
+         draw_set_mapped_constant_buffer(draw, PIPE_SHADER_VERTEX, 0, NULL, 0);
       }
    }
 
    for (i = 0; i < nv30->num_vtxbufs; i++) {
       const void *map = nv30->vtxbuf[i].user_buffer;
       if (!map) {
-         if (!nv30->vtxbuf[i].buffer) {
-            continue;
-         }
-         map = pipe_buffer_map(pipe, nv30->vtxbuf[i].buffer,
+         if (nv30->vtxbuf[i].buffer)
+            map = pipe_buffer_map(pipe, nv30->vtxbuf[i].buffer,
                                   PIPE_TRANSFER_UNSYNCHRONIZED |
                                   PIPE_TRANSFER_READ, &transfer[i]);
       }
@@ -418,9 +430,9 @@ nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    if (info->indexed) {
       const void *map = nv30->idxbuf.user_buffer;
       if (!map)
-         pipe_buffer_map(pipe, nv30->idxbuf.buffer,
-                                  PIPE_TRANSFER_UNSYNCHRONIZED |
-                                  PIPE_TRANSFER_READ, &transferi);
+         map = pipe_buffer_map(pipe, nv30->idxbuf.buffer,
+                               PIPE_TRANSFER_UNSYNCHRONIZED |
+                               PIPE_TRANSFER_READ, &transferi);
       draw_set_indexes(draw,
                        (ubyte *) map + nv30->idxbuf.offset,
                        nv30->idxbuf.index_size, ~0);
@@ -444,6 +456,12 @@ nv30_render_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
 static void
 nv30_render_destroy(struct vbuf_render *render)
 {
+   struct nv30_render *r = nv30_render(render);
+
+   if (r->transfer)
+      pipe_buffer_unmap(&r->nv30->base.pipe, r->transfer);
+   pipe_resource_reference(&r->buffer, NULL);
+   nouveau_heap_free(&r->vertprog);
    FREE(render);
 }
 

src/gallium/drivers/nouveau/nv30/nv30_fragprog.c

@@ -23,6 +23,7 @@
  *
  */
 
+#include "draw/draw_context.h"
 #include "tgsi/tgsi_parse.h"
 
 #include "nv_object.xml.h"
@@ -147,8 +148,12 @@ nv30_fp_state_delete(struct pipe_context *pipe, void *hwcso)
 
    pipe_resource_reference(&fp->buffer, NULL);
 
+   if (fp->draw)
+      draw_delete_fragment_shader(nv30_context(pipe)->draw, fp->draw);
+
    FREE((void *)fp->pipe.tokens);
    FREE(fp->insn);
+   FREE(fp->consts);
    FREE(fp);
 }
 

src/gallium/drivers/nouveau/nv30/nv30_state_validate.c

@@ -272,15 +272,13 @@ nv30_validate_clip(struct nv30_context *nv30)
    uint32_t clpd_enable = 0;
 
    for (i = 0; i < 6; i++) {
-      if (nv30->rast->pipe.clip_plane_enable & (1 << i)) {
-         if (nv30->dirty & NV30_NEW_CLIP) {
-            BEGIN_NV04(push, NV30_3D(VP_UPLOAD_CONST_ID), 5);
-            PUSH_DATA (push, i);
-            PUSH_DATAp(push, nv30->clip.ucp[i], 4);
-         }
-
-         clpd_enable |= 1 << (1 + 4*i);
+      if (nv30->dirty & NV30_NEW_CLIP) {
+         BEGIN_NV04(push, NV30_3D(VP_UPLOAD_CONST_ID), 5);
+         PUSH_DATA (push, i);
+         PUSH_DATAp(push, nv30->clip.ucp[i], 4);
       }
+      if (nv30->rast->pipe.clip_plane_enable & (1 << i))
+         clpd_enable |= 2 << (4*i);
    }
 
    BEGIN_NV04(push, NV30_3D(VP_CLIP_PLANES_ENABLE), 1);
@@ -389,7 +387,7 @@ static struct state_validate hwtnl_validate_list[] = {
     { nv30_validate_stipple,       NV30_NEW_STIPPLE },
     { nv30_validate_scissor,       NV30_NEW_SCISSOR | NV30_NEW_RASTERIZER },
     { nv30_validate_viewport,      NV30_NEW_VIEWPORT },
-    { nv30_validate_clip,          NV30_NEW_CLIP },
+    { nv30_validate_clip,          NV30_NEW_CLIP | NV30_NEW_RASTERIZER },
     { nv30_fragprog_validate,      NV30_NEW_FRAGPROG | NV30_NEW_FRAGCONST },
     { nv30_vertprog_validate,      NV30_NEW_VERTPROG | NV30_NEW_VERTCONST |
                                    NV30_NEW_FRAGPROG | NV30_NEW_RASTERIZER },
@@ -456,7 +454,7 @@ nv30_state_context_switch(struct nv30_context *nv30)
 }
 
 boolean
-nv30_state_validate(struct nv30_context *nv30, boolean hwtnl)
+nv30_state_validate(struct nv30_context *nv30, uint32_t mask, boolean hwtnl)
 {
    struct nouveau_screen *screen = &nv30->screen->base;
    struct nouveau_pushbuf *push = nv30->base.pushbuf;
@@ -481,14 +479,16 @@ nv30_state_validate(struct nv30_context *nv30, boolean hwtnl)
    else
       validate = swtnl_validate_list;
 
-   if (nv30->dirty) {
+   mask &= nv30->dirty;
+
+   if (mask) {
       while (validate->func) {
-         if (nv30->dirty & validate->mask)
+         if (mask & validate->mask)
             validate->func(nv30);
          validate++;
       }
 
-      nv30->dirty = 0;
+      nv30->dirty &= ~mask;
    }
 
    nouveau_pushbuf_bufctx(push, bctx);

src/gallium/drivers/nouveau/nv30/nv30_vbo.c

@@ -564,7 +564,7 @@ nv30_draw_vbo(struct pipe_context *pipe, const struct pipe_draw_info *info)
    if (nv30->vbo_user && !(nv30->dirty & (NV30_NEW_VERTEX | NV30_NEW_ARRAYS)))
       nv30_update_user_vbufs(nv30);
 
-   nv30_state_validate(nv30, TRUE);
+   nv30_state_validate(nv30, ~0, TRUE);
    if (nv30->draw_flags) {
       nv30_render_vbo(pipe, info);
       return;

src/gallium/drivers/nouveau/nv30/nv30_vertprog.c

@@ -23,6 +23,7 @@
  *
  */
 
+#include "draw/draw_context.h"
 #include "util/u_dynarray.h"
 #include "tgsi/tgsi_parse.h"
 
@@ -237,6 +238,10 @@ nv30_vp_state_delete(struct pipe_context *pipe, void *hwcso)
 
    if (vp->translated)
       nv30_vertprog_destroy(vp);
+
+   if (vp->draw)
+      draw_delete_vertex_shader(nv30_context(pipe)->draw, vp->draw);
+
    FREE((void *)vp->pipe.tokens);
    FREE(vp);
 }

src/gallium/drivers/nouveau/nv30/nvfx_vertprog.c

@@ -872,9 +872,8 @@ nvfx_vertprog_parse_decl_output(struct nvfx_vpc *vpc,
       }
       break;
    case TGSI_SEMANTIC_EDGEFLAG:
-      /* not really an error just a fallback */
-      NOUVEAU_ERR("cannot handle edgeflag output\n");
-      return FALSE;
+      vpc->r_result[idx] = nvfx_reg(NVFXSR_NONE, 0);
+      return TRUE;
    default:
       NOUVEAU_ERR("bad output semantic\n");
       return FALSE;

src/gallium/drivers/nouveau/nv50/nv50_miptree.c

@@ -419,7 +419,7 @@ nv50_miptree_from_handle(struct pipe_screen *pscreen,
       FREE(mt);
       return NULL;
    }
-   mt->base.domain = NOUVEAU_BO_VRAM;
+   mt->base.domain = mt->base.bo->flags & NOUVEAU_BO_APER;
    mt->base.address = mt->base.bo->offset;
 
    mt->base.base = *templ;

src/gallium/drivers/nouveau/nv50/nv50_query.c

@@ -434,6 +434,7 @@ nv50_query_pushbuf_submit(struct nouveau_pushbuf *push,
    /* XXX: does this exist ? */
 #define NV50_IB_ENTRY_1_NO_PREFETCH (0 << (31 - 8))
 
+   PUSH_REFN(push, q->bo, NOUVEAU_BO_RD | NOUVEAU_BO_GART);
    nouveau_pushbuf_space(push, 0, 0, 1);
    nouveau_pushbuf_data(push, q->bo, q->offset + result_offset, 4 |
                         NV50_IB_ENTRY_1_NO_PREFETCH);

src/gallium/drivers/nouveau/nv50/nv50_state.c

@@ -811,12 +811,12 @@ nv50_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
    nv50->constbuf[s][i].user = (cb && cb->user_buffer) ? TRUE : FALSE;
    if (nv50->constbuf[s][i].user) {
       nv50->constbuf[s][i].u.data = cb->user_buffer;
-      nv50->constbuf[s][i].size = cb->buffer_size;
+      nv50->constbuf[s][i].size = MIN2(cb->buffer_size, 0x10000);
       nv50->constbuf_valid[s] |= 1 << i;
    } else
    if (res) {
       nv50->constbuf[s][i].offset = cb->buffer_offset;
-      nv50->constbuf[s][i].size = align(cb->buffer_size, 0x100);
+      nv50->constbuf[s][i].size = MIN2(align(cb->buffer_size, 0x100), 0x10000);
       nv50->constbuf_valid[s] |= 1 << i;
    } else {
       nv50->constbuf_valid[s] &= ~(1 << i);

src/gallium/drivers/nouveau/nv50/nv50_state_validate.c

@@ -296,6 +296,23 @@ nv50_check_program_ucps(struct nv50_context *nv50,
    nv50_fp_linkage_validate(nv50);
 }
 
+/* alpha test is disabled if there are no color RTs, so make sure we have at
+ * least one if alpha test is enabled. Note that this must run after
+ * nv50_validate_fb, otherwise that will override the RT count setting.
+ */
+static void
+nv50_validate_derived_2(struct nv50_context *nv50)
+{
+   struct nouveau_pushbuf *push = nv50->base.pushbuf;
+
+   if (nv50->zsa && nv50->zsa->pipe.alpha.enabled &&
+       nv50->framebuffer.nr_cbufs == 0) {
+      nv50_fb_set_null_rt(push, 0);
+      BEGIN_NV04(push, NV50_3D(RT_CONTROL), 1);
+      PUSH_DATA (push, (076543210 << 4) | 1);
+   }
+}
+
 static void
 nv50_validate_clip(struct nv50_context *nv50)
 {
@@ -456,6 +473,7 @@ static struct state_validate {
     { nv50_gp_linkage_validate,    NV50_NEW_GMTYPROG | NV50_NEW_VERTPROG },
     { nv50_validate_derived_rs,    NV50_NEW_FRAGPROG | NV50_NEW_RASTERIZER |
                                    NV50_NEW_VERTPROG | NV50_NEW_GMTYPROG },
+    { nv50_validate_derived_2,     NV50_NEW_ZSA | NV50_NEW_FRAMEBUFFER },
     { nv50_validate_clip,          NV50_NEW_CLIP | NV50_NEW_RASTERIZER |
                                    NV50_NEW_VERTPROG | NV50_NEW_GMTYPROG },
     { nv50_constbufs_validate,     NV50_NEW_CONSTBUF },

src/gallium/drivers/nouveau/nv50/nv50_tex.c

@@ -71,6 +71,7 @@ nv50_create_texture_view(struct pipe_context *pipe,
                          uint32_t flags,
                          enum pipe_texture_target target)
 {
+   const uint32_t class_3d = nouveau_context(pipe)->screen->class_3d;
    const struct util_format_description *desc;
    uint64_t addr;
    uint32_t *tic;
@@ -201,11 +202,17 @@ nv50_create_texture_view(struct pipe_context *pipe,
 
    tic[5] = (mt->base.base.height0 << mt->ms_y) & 0xffff;
    tic[5] |= depth << 16;
-   tic[5] |= mt->base.base.last_level << NV50_TIC_5_LAST_LEVEL__SHIFT;
+   if (class_3d > NV50_3D_CLASS)
+      tic[5] |= mt->base.base.last_level << NV50_TIC_5_LAST_LEVEL__SHIFT;
+   else
+      tic[5] |= view->pipe.u.tex.last_level << NV50_TIC_5_LAST_LEVEL__SHIFT;
 
    tic[6] = (mt->ms_x > 1) ? 0x88000000 : 0x03000000; /* sampling points */
 
-   tic[7] = (view->pipe.u.tex.last_level << 4) | view->pipe.u.tex.first_level;
+   if (class_3d > NV50_3D_CLASS)
+      tic[7] = (view->pipe.u.tex.last_level << 4) | view->pipe.u.tex.first_level;
+   else
+      tic[7] = 0;
 
    if (unlikely(!(tic[2] & NV50_TIC_2_NORMALIZED_COORDS)))
       if (mt->base.base.last_level)

src/gallium/drivers/nouveau/nv50/nv50_vbo.c

@@ -382,6 +382,11 @@ nv50_vertex_arrays_validate(struct nv50_context *nv50)
       if (nv50->vbo_user & (1 << b)) {
          address = addrs[b] + ve->pipe.src_offset;
          limit = addrs[b] + limits[b];
+      } else
+      if (!vb->buffer) {
+         BEGIN_NV04(push, NV50_3D(VERTEX_ARRAY_FETCH(i)), 1);
+         PUSH_DATA (push, 0);
+         continue;
       } else {
          struct nv04_resource *buf = nv04_resource(vb->buffer);
          if (!(refd & (1 << b))) {
@@ -628,6 +633,7 @@ nv50_draw_elements(struct nv50_context *nv50, boolean shorten,
          BEGIN_NV04(push, NV50_3D(VERTEX_BEGIN_GL), 1);
          PUSH_DATA (push, prim);
 
+         PUSH_REFN(push, buf->bo, NOUVEAU_BO_RD | buf->domain);
          nouveau_pushbuf_space(push, 8, 0, 1);
 
          switch (index_size) {

src/gallium/drivers/nouveau/nvc0/nvc0_program.c

@@ -392,7 +392,7 @@ nvc0_gp_gen_header(struct nvc0_program *gp, struct nv50_ir_prog_info *info)
       break;
    }
 
-   gp->hdr[4] = info->prop.gp.maxVertices & 0x1ff;
+   gp->hdr[4] = MIN2(info->prop.gp.maxVertices, 1024);
 
    return nvc0_vtgp_gen_header(gp, info);
 }

src/gallium/drivers/nouveau/nvc0/nvc0_query.c

@@ -617,6 +617,7 @@ nvc0_query_pushbuf_submit(struct nouveau_pushbuf *push,
 
 #define NVC0_IB_ENTRY_1_NO_PREFETCH (1 << (31 - 8))
 
+   PUSH_REFN(push, q->bo, NOUVEAU_BO_RD | NOUVEAU_BO_GART);
    nouveau_pushbuf_space(push, 0, 0, 1);
    nouveau_pushbuf_data(push, q->bo, q->offset + result_offset, 4 |
                         NVC0_IB_ENTRY_1_NO_PREFETCH);

src/gallium/drivers/nouveau/nvc0/nvc0_shader_state.c

@@ -262,11 +262,13 @@ nvc0_tfb_validate(struct nvc0_context *nvc0)
       if (tfb)
          targ->stride = tfb->stride[b];
 
+      buf = nv04_resource(targ->pipe.buffer);
+
+      BCTX_REFN(nvc0->bufctx_3d, TFB, buf, WR);
+
       if (!(nvc0->tfbbuf_dirty & (1 << b)))
          continue;
 
-      buf = nv04_resource(targ->pipe.buffer);
-
       if (!targ->clean)
          nvc0_query_fifo_wait(push, targ->pq);
       BEGIN_NVC0(push, NVC0_3D(TFB_BUFFER_ENABLE(b)), 5);
@@ -280,7 +282,6 @@ nvc0_tfb_validate(struct nvc0_context *nvc0)
          PUSH_DATA(push, 0); /* TFB_BUFFER_OFFSET */
          targ->clean = FALSE;
       }
-      BCTX_REFN(nvc0->bufctx_3d, TFB, buf, WR);
    }
    for (; b < 4; ++b)
       IMMED_NVC0(push, NVC0_3D(TFB_BUFFER_ENABLE(b)), 0);

src/gallium/drivers/nouveau/nvc0/nvc0_state.c

@@ -793,12 +793,12 @@ nvc0_set_constant_buffer(struct pipe_context *pipe, uint shader, uint index,
    nvc0->constbuf[s][i].user = (cb && cb->user_buffer) ? TRUE : FALSE;
    if (nvc0->constbuf[s][i].user) {
       nvc0->constbuf[s][i].u.data = cb->user_buffer;
-      nvc0->constbuf[s][i].size = cb->buffer_size;
+      nvc0->constbuf[s][i].size = MIN2(cb->buffer_size, 0x10000);
       nvc0->constbuf_valid[s] |= 1 << i;
    } else
    if (cb) {
       nvc0->constbuf[s][i].offset = cb->buffer_offset;
-      nvc0->constbuf[s][i].size = align(cb->buffer_size, 0x100);
+      nvc0->constbuf[s][i].size = MIN2(align(cb->buffer_size, 0x100), 0x10000);
       nvc0->constbuf_valid[s] |= 1 << i;
    }
    else {

src/gallium/drivers/nouveau/nvc0/nvc0_state_validate.c

@@ -339,7 +339,7 @@ nvc0_check_program_ucps(struct nvc0_context *nvc0,
       nvc0_vertprog_validate(nvc0);
    else
    if (likely(vp == nvc0->gmtyprog))
-      nvc0_vertprog_validate(nvc0);
+      nvc0_gmtyprog_validate(nvc0);
    else
       nvc0_tevlprog_validate(nvc0);
 }
@@ -535,6 +535,23 @@ nvc0_validate_derived_1(struct nvc0_context *nvc0)
    }
 }
 
+/* alpha test is disabled if there are no color RTs, so make sure we have at
+ * least one if alpha test is enabled. Note that this must run after
+ * nvc0_validate_fb, otherwise that will override the RT count setting.
+ */
+static void
+nvc0_validate_derived_2(struct nvc0_context *nvc0)
+{
+   struct nouveau_pushbuf *push = nvc0->base.pushbuf;
+
+   if (nvc0->zsa && nvc0->zsa->pipe.alpha.enabled &&
+       nvc0->framebuffer.nr_cbufs == 0) {
+      nvc0_fb_set_null_rt(push, 0);
+      BEGIN_NVC0(push, NVC0_3D(RT_CONTROL), 1);
+      PUSH_DATA (push, (076543210 << 4) | 1);
+   }
+}
+
 static void
 nvc0_switch_pipe_context(struct nvc0_context *ctx_to)
 {
@@ -597,6 +614,7 @@ static struct state_validate {
     { nvc0_fragprog_validate,      NVC0_NEW_FRAGPROG },
     { nvc0_validate_derived_1,     NVC0_NEW_FRAGPROG | NVC0_NEW_ZSA |
                                    NVC0_NEW_RASTERIZER },
+    { nvc0_validate_derived_2,     NVC0_NEW_ZSA | NVC0_NEW_FRAMEBUFFER },
     { nvc0_validate_clip,          NVC0_NEW_CLIP | NVC0_NEW_RASTERIZER |
                                    NVC0_NEW_VERTPROG |
                                    NVC0_NEW_TEVLPROG |

src/gallium/drivers/nouveau/nvc0/nvc0_surface.c

@@ -1376,6 +1376,7 @@ static void
 nvc0_blit(struct pipe_context *pipe, const struct pipe_blit_info *info)
 {
    struct nvc0_context *nvc0 = nvc0_context(pipe);
+   struct nouveau_pushbuf *push = nvc0->base.pushbuf;
    boolean eng3d = FALSE;
 
    if (util_format_is_depth_or_stencil(info->dst.resource->format)) {
@@ -1439,11 +1440,17 @@ nvc0_blit(struct pipe_context *pipe, const struct pipe_blit_info *info)
         info->src.box.height != -info->dst.box.height))
       eng3d = TRUE;
 
+   if (nvc0->screen->num_occlusion_queries_active)
+      IMMED_NVC0(push, NVC0_3D(SAMPLECNT_ENABLE), 0);
+
    if (!eng3d)
       nvc0_blit_eng2d(nvc0, info);
    else
       nvc0_blit_3d(nvc0, info);
 
+   if (nvc0->screen->num_occlusion_queries_active)
+      IMMED_NVC0(push, NVC0_3D(SAMPLECNT_ENABLE), 1);
+
    NOUVEAU_DRV_STAT(&nvc0->screen->base, tex_blit_count, 1);
 }
 

src/gallium/drivers/nouveau/nvc0/nvc0_vbo.c

@@ -829,6 +829,7 @@ nvc0_draw_indirect(struct nvc0_context *nvc0, const struct pipe_draw_info *info)
    }
    PUSH_DATA(push, nvc0_prim_gl(info->mode));
 #define NVC0_IB_ENTRY_1_NO_PREFETCH (1 << (31 - 8))
+   PUSH_REFN(push, buf->bo, NOUVEAU_BO_RD | buf->domain);
    nouveau_pushbuf_space(push, 0, 0, 1);
    nouveau_pushbuf_data(push,
                         buf->bo, offset, NVC0_IB_ENTRY_1_NO_PREFETCH | size);

src/gallium/drivers/r600/evergreen_state.c

@@ -1732,10 +1732,10 @@ static void evergreen_emit_cb_misc_state(struct r600_context *rctx, struct r600_
 
 	r600_write_context_reg_seq(cs, R_028238_CB_TARGET_MASK, 2);
 	radeon_emit(cs, a->blend_colormask & fb_colormask); /* R_028238_CB_TARGET_MASK */
-	/* Always enable the first colorbuffer in CB_SHADER_MASK. This
-	 * will assure that the alpha-test will work even if there is
-	 * no colorbuffer bound. */
-	radeon_emit(cs, 0xf | (a->dual_src_blend ? ps_colormask : 0) | fb_colormask); /* R_02823C_CB_SHADER_MASK */
+	/* This must match the used export instructions exactly.
+	 * Other values may lead to undefined behavior and hangs.
+	 */
+	radeon_emit(cs, ps_colormask); /* R_02823C_CB_SHADER_MASK */
 }
 
 static void evergreen_emit_db_state(struct r600_context *rctx, struct r600_atom *atom)

src/gallium/drivers/r600/r600_blit.c

@@ -393,7 +393,12 @@ static void r600_clear(struct pipe_context *ctx, unsigned buffers,
 	struct r600_context *rctx = (struct r600_context *)ctx;
 	struct pipe_framebuffer_state *fb = &rctx->framebuffer.state;
 
-	if (buffers & PIPE_CLEAR_COLOR && rctx->b.chip_class >= EVERGREEN) {
+	/* Single-sample fast color clear is broken on r600g:
+	 *   https://bugs.freedesktop.org/show_bug.cgi?id=73528
+	 *   https://bugs.freedesktop.org/show_bug.cgi?id=82186
+	 */
+	if (buffers & PIPE_CLEAR_COLOR && rctx->b.chip_class >= EVERGREEN &&
+	    rctx->framebuffer.nr_samples > 1) {
 		evergreen_do_fast_color_clear(&rctx->b, fb, &rctx->framebuffer.atom,
 					      &buffers, color);
 	}

src/gallium/drivers/r600/r600_shader.c

@@ -2482,6 +2482,7 @@ static int r600_shader_from_tgsi(struct r600_context *rctx,
 			output[j].array_base = 0;
 			output[j].op = CF_OP_EXPORT;
 			j++;
+			shader->nr_ps_color_exports++;
 		}
 
 		noutput = j;

src/gallium/drivers/radeonsi/si_descriptors.c

@@ -24,14 +24,23 @@
  *      Marek Olšák <marek.olsak@amd.com>
  */
 
-/* Resource binding slots and sampler states (each described with 8 or 4 dwords)
- * live in memory on SI.
+/* Resource binding slots and sampler states (each described with 8 or
+ * 4 dwords) are stored in lists in memory which is accessed by shaders
+ * using scalar load instructions.
  *
- * This file is responsible for managing lists of resources and sampler states
- * in memory and binding them, which means updating those structures in memory.
+ * This file is responsible for managing such lists. It keeps a copy of all
+ * descriptors in CPU memory and re-uploads a whole list if some slots have
+ * been changed.
  *
- * There is also code for updating shader pointers to resources and sampler
- * states. CP DMA functions are here too.
+ * This code is also reponsible for updating shader pointers to those lists.
+ *
+ * Note that CP DMA can't be used for updating the lists, because a GPU hang
+ * could leave the list in a mid-IB state and the next IB would get wrong
+ * descriptors and the whole context would be unusable at that point.
+ * (Note: The register shadowing can't be used due to the same reason)
+ *
+ * Also, uploading descriptors to newly allocated memory doesn't require
+ * a KCACHE flush.
  */
 
 #include "radeon/r600_cs.h"
@@ -42,7 +51,6 @@
 #include "util/u_memory.h"
 #include "util/u_upload_mgr.h"
 
-#define SI_NUM_CONTEXTS 16
 
 /* NULL image and buffer descriptor.
  *
@@ -139,209 +147,62 @@ static void si_emit_cp_dma_clear_buffer(struct si_context *sctx,
 	}
 }
 
-static void si_init_descriptors(struct si_context *sctx,
-				struct si_descriptors *desc,
-				unsigned shader_userdata_reg,
+static void si_init_descriptors(struct si_descriptors *desc,
+				unsigned shader_userdata_index,
 				unsigned element_dw_size,
-				unsigned num_elements,
-				void (*emit_func)(struct si_context *ctx, struct r600_atom *state))
+				unsigned num_elements)
 {
-	assert(num_elements <= sizeof(desc->enabled_mask)*8);
-	assert(num_elements <= sizeof(desc->dirty_mask)*8);
+	int i;
 
-	desc->atom.emit = (void*)emit_func;
-	desc->shader_userdata_reg = shader_userdata_reg;
+	assert(num_elements <= sizeof(desc->enabled_mask)*8);
+
+	desc->list = CALLOC(num_elements, element_dw_size * 4);
 	desc->element_dw_size = element_dw_size;
 	desc->num_elements = num_elements;
-	desc->context_size = num_elements * element_dw_size * 4;
+	desc->list_dirty = true; /* upload the list before the next draw */
+	desc->shader_userdata_offset = shader_userdata_index * 4;
 
-	desc->buffer = (struct r600_resource*)
-		pipe_buffer_create(sctx->b.b.screen, PIPE_BIND_CUSTOM,
-				   PIPE_USAGE_DEFAULT,
-				   SI_NUM_CONTEXTS * desc->context_size);
-
-	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx, desc->buffer,
-			      RADEON_USAGE_READWRITE, RADEON_PRIO_SHADER_DATA);
-
-	/* We don't check for CS space here, because this should be called
-	 * only once at context initialization. */
-	si_emit_cp_dma_clear_buffer(sctx, desc->buffer->gpu_address,
-				    desc->buffer->b.b.width0, 0,
-				    R600_CP_DMA_SYNC | CIK_CP_DMA_USE_L2);
+	/* Initialize the array to NULL descriptors if the element size is 8. */
+	if (element_dw_size == 8)
+		for (i = 0; i < num_elements; i++)
+			memcpy(desc->list + i*element_dw_size, null_descriptor,
+			       sizeof(null_descriptor));
 }
 
 static void si_release_descriptors(struct si_descriptors *desc)
 {
 	pipe_resource_reference((struct pipe_resource**)&desc->buffer, NULL);
+	FREE(desc->list);
 }
 
-static void si_update_descriptors(struct si_context *sctx,
+static bool si_upload_descriptors(struct si_context *sctx,
 				  struct si_descriptors *desc)
 {
-	if (desc->dirty_mask) {
-		desc->atom.num_dw =
-			7 + /* copy */
-			(4 + desc->element_dw_size) * util_bitcount64(desc->dirty_mask) + /* update */
-			4; /* pointer update */
+	unsigned list_size = desc->num_elements * desc->element_dw_size * 4;
+	void *ptr;
 
-		if (desc->shader_userdata_reg >= R_00B130_SPI_SHADER_USER_DATA_VS_0 &&
-		    desc->shader_userdata_reg < R_00B230_SPI_SHADER_USER_DATA_GS_0)
-			desc->atom.num_dw += 4; /* second pointer update */
+	if (!desc->list_dirty)
+		return true;
 
-		desc->atom.dirty = true;
+	u_upload_alloc(sctx->b.uploader, 0, list_size,
+		       &desc->buffer_offset,
+		       (struct pipe_resource**)&desc->buffer, &ptr);
+	if (!desc->buffer)
+		return false; /* skip the draw call */
 
-		/* TODO: Investigate if these flushes can be removed after
-		 * adding CE support. */
+	util_memcpy_cpu_to_le32(ptr, desc->list, list_size);
 
-		/* The descriptors are read with the K cache. */
-		sctx->b.flags |= SI_CONTEXT_INV_KCACHE;
+	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx, desc->buffer,
+			      RADEON_USAGE_READ, RADEON_PRIO_SHADER_DATA);
 
-		/* Since SI uses uncached CP DMA to update descriptors,
-		 * we have to flush TC L2, which is used to fetch constants
-		 * along with KCACHE. */
-		if (sctx->b.chip_class == SI)
-			sctx->b.flags |= SI_CONTEXT_INV_TC_L2;
-	} else {
-		desc->atom.dirty = false;
-	}
-}
-
-static void si_emit_shader_pointer(struct si_context *sctx,
-				   struct r600_atom *atom)
-{
-	struct si_descriptors *desc = (struct si_descriptors*)atom;
-	struct radeon_winsys_cs *cs = sctx->b.rings.gfx.cs;
-	uint64_t va = desc->buffer->gpu_address +
-		      desc->current_context_id * desc->context_size +
-		      desc->buffer_offset;
-
-	radeon_emit(cs, PKT3(PKT3_SET_SH_REG, 2, 0));
-	radeon_emit(cs, (desc->shader_userdata_reg - SI_SH_REG_OFFSET) >> 2);
-	radeon_emit(cs, va);
-	radeon_emit(cs, va >> 32);
-
-	if (desc->shader_userdata_reg >= R_00B130_SPI_SHADER_USER_DATA_VS_0 &&
-	    desc->shader_userdata_reg < R_00B230_SPI_SHADER_USER_DATA_GS_0) {
-		radeon_emit(cs, PKT3(PKT3_SET_SH_REG, 2, 0));
-		radeon_emit(cs, (desc->shader_userdata_reg +
-				 (R_00B330_SPI_SHADER_USER_DATA_ES_0 -
-				  R_00B130_SPI_SHADER_USER_DATA_VS_0) -
-				 SI_SH_REG_OFFSET) >> 2);
-		radeon_emit(cs, va);
-		radeon_emit(cs, va >> 32);
-	}
-}
-
-static void si_emit_descriptors(struct si_context *sctx,
-				struct si_descriptors *desc,
-				uint32_t **descriptors)
-{
-	struct radeon_winsys_cs *cs = sctx->b.rings.gfx.cs;
-	uint64_t va_base;
-	int packet_start = 0;
-	int packet_size = 0;
-	int last_index = desc->num_elements; /* point to a non-existing element */
-	uint64_t dirty_mask = desc->dirty_mask;
-	unsigned new_context_id = (desc->current_context_id + 1) % SI_NUM_CONTEXTS;
-
-	assert(dirty_mask);
-
-	va_base = desc->buffer->gpu_address;
-
-	/* Copy the descriptors to a new context slot. */
-	si_emit_cp_dma_copy_buffer(sctx,
-				   va_base + new_context_id * desc->context_size,
-				   va_base + desc->current_context_id * desc->context_size,
-				   desc->context_size, R600_CP_DMA_SYNC | CIK_CP_DMA_USE_L2);
-
-	va_base += new_context_id * desc->context_size;
-
-	/* Update the descriptors.
-	 * Updates of consecutive descriptors are merged to one WRITE_DATA packet.
-	 *
-	 * XXX When unbinding lots of resources, consider clearing the memory
-	 *     with CP DMA instead of emitting zeros.
-	 */
-	while (dirty_mask) {
-		int i = u_bit_scan64(&dirty_mask);
-
-		assert(i < desc->num_elements);
-
-		if (last_index+1 == i && packet_size) {
-			/* Append new data at the end of the last packet. */
-			packet_size += desc->element_dw_size;
-			cs->buf[packet_start] = PKT3(PKT3_WRITE_DATA, packet_size, 0);
-		} else {
-			/* Start a new packet. */
-			uint64_t va = va_base + i * desc->element_dw_size * 4;
-
-			packet_start = cs->cdw;
-			packet_size = 2 + desc->element_dw_size;
-
-			radeon_emit(cs, PKT3(PKT3_WRITE_DATA, packet_size, 0));
-			radeon_emit(cs, PKT3_WRITE_DATA_DST_SEL(sctx->b.chip_class == SI ?
-						PKT3_WRITE_DATA_DST_SEL_MEM_SYNC :
-						PKT3_WRITE_DATA_DST_SEL_TC_L2) |
-					     PKT3_WRITE_DATA_WR_CONFIRM |
-					     PKT3_WRITE_DATA_ENGINE_SEL(PKT3_WRITE_DATA_ENGINE_SEL_ME));
-			radeon_emit(cs, va & 0xFFFFFFFFUL);
-			radeon_emit(cs, (va >> 32UL) & 0xFFFFFFFFUL);
-		}
-
-		radeon_emit_array(cs, descriptors[i], desc->element_dw_size);
-
-		last_index = i;
-	}
-
-	desc->dirty_mask = 0;
-	desc->current_context_id = new_context_id;
-
-	/* Now update the shader userdata pointer. */
-	si_emit_shader_pointer(sctx, &desc->atom);
-}
-
-static unsigned si_get_shader_user_data_base(unsigned shader)
-{
-	switch (shader) {
-	case PIPE_SHADER_VERTEX:
-		return R_00B130_SPI_SHADER_USER_DATA_VS_0;
-	case PIPE_SHADER_GEOMETRY:
-		return R_00B230_SPI_SHADER_USER_DATA_GS_0;
-	case PIPE_SHADER_FRAGMENT:
-		return R_00B030_SPI_SHADER_USER_DATA_PS_0;
-	default:
-		assert(0);
-		return 0;
-	}
+	desc->list_dirty = false;
+	desc->pointer_dirty = true;
+	sctx->shader_userdata.atom.dirty = true;
+	return true;
 }
 
 /* SAMPLER VIEWS */
 
-static void si_emit_sampler_views(struct si_context *sctx, struct r600_atom *atom)
-{
-	struct si_sampler_views *views = (struct si_sampler_views*)atom;
-
-	si_emit_descriptors(sctx, &views->desc, views->desc_data);
-}
-
-static void si_init_sampler_views(struct si_context *sctx,
-				  struct si_sampler_views *views,
-				  unsigned shader)
-{
-	int i;
-
-	si_init_descriptors(sctx, &views->desc,
-			    si_get_shader_user_data_base(shader) +
-			    SI_SGPR_RESOURCE * 4,
-			    8, SI_NUM_SAMPLER_VIEWS, si_emit_sampler_views);
-
-	for (i = 0; i < views->desc.num_elements; i++) {
-		views->desc_data[i] = null_descriptor;
-		views->desc.dirty_mask |= 1llu << i;
-	}
-	si_update_descriptors(sctx, &views->desc);
-}
-
 static void si_release_sampler_views(struct si_sampler_views *views)
 {
 	int i;
@@ -382,10 +243,10 @@ static void si_sampler_views_begin_new_cs(struct si_context *sctx,
 				      si_get_resource_ro_priority(rview->resource));
 	}
 
+	if (!views->desc.buffer)
+		return;
 	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx, views->desc.buffer,
 			      RADEON_USAGE_READWRITE, RADEON_PRIO_SHADER_DATA);
-
-	si_emit_shader_pointer(sctx, &views->desc.atom);
 }
 
 static void si_set_sampler_view(struct si_context *sctx, unsigned shader,
@@ -406,17 +267,16 @@ static void si_set_sampler_view(struct si_context *sctx, unsigned shader,
 				rview->resource, RADEON_USAGE_READ,
 				si_get_resource_ro_priority(rview->resource));
 
-
 		pipe_sampler_view_reference(&views->views[slot], view);
-		views->desc_data[slot] = view_desc;
+		memcpy(views->desc.list + slot*8, view_desc, 8*4);
 		views->desc.enabled_mask |= 1llu << slot;
 	} else {
 		pipe_sampler_view_reference(&views->views[slot], NULL);
-		views->desc_data[slot] = null_descriptor;
+		memcpy(views->desc.list + slot*8, null_descriptor, 8*4);
 		views->desc.enabled_mask &= ~(1llu << slot);
 	}
 
-	views->desc.dirty_mask |= 1llu << slot;
+	views->desc.list_dirty = true;
 }
 
 static void si_set_sampler_views(struct pipe_context *ctx,
@@ -475,25 +335,17 @@ static void si_set_sampler_views(struct pipe_context *ctx,
 					    NULL, NULL);
 		}
 	}
-
-	si_update_descriptors(sctx, &samplers->views.desc);
 }
 
 /* SAMPLER STATES */
 
-static void si_emit_sampler_states(struct si_context *sctx, struct r600_atom *atom)
-{
-	struct si_sampler_states *states = (struct si_sampler_states*)atom;
-
-	si_emit_descriptors(sctx, &states->desc, states->desc_data);
-}
-
 static void si_sampler_states_begin_new_cs(struct si_context *sctx,
 					   struct si_sampler_states *states)
 {
+	if (!states->desc.buffer)
+		return;
 	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx, states->desc.buffer,
 			      RADEON_USAGE_READWRITE, RADEON_PRIO_SHADER_DATA);
-	si_emit_shader_pointer(sctx, &states->desc.atom);
 }
 
 void si_set_sampler_descriptors(struct si_context *sctx, unsigned shader,
@@ -513,66 +365,39 @@ void si_set_sampler_descriptors(struct si_context *sctx, unsigned shader,
 	for (i = 0; i < count; i++) {
 		unsigned slot = start + i;
 
-		if (!sstates[i]) {
-			samplers->desc.dirty_mask &= ~(1llu << slot);
+		if (!sstates[i])
 			continue;
-		}
 
-		samplers->desc_data[slot] = sstates[i]->val;
-		samplers->desc.dirty_mask |= 1llu << slot;
+		memcpy(samplers->desc.list + slot*4, sstates[i]->val, 4*4);
+		samplers->desc.list_dirty = true;
 	}
-
-	si_update_descriptors(sctx, &samplers->desc);
 }
 
 /* BUFFER RESOURCES */
 
-static void si_emit_buffer_resources(struct si_context *sctx, struct r600_atom *atom)
-{
-	struct si_buffer_resources *buffers = (struct si_buffer_resources*)atom;
-
-	si_emit_descriptors(sctx, &buffers->desc, buffers->desc_data);
-}
-
-static void si_init_buffer_resources(struct si_context *sctx,
-				     struct si_buffer_resources *buffers,
-				     unsigned num_buffers, unsigned shader,
+static void si_init_buffer_resources(struct si_buffer_resources *buffers,
+				     unsigned num_buffers,
 				     unsigned shader_userdata_index,
 				     enum radeon_bo_usage shader_usage,
 				     enum radeon_bo_priority priority)
 {
-	int i;
-
-	buffers->num_buffers = num_buffers;
 	buffers->shader_usage = shader_usage;
 	buffers->priority = priority;
 	buffers->buffers = CALLOC(num_buffers, sizeof(struct pipe_resource*));
-	buffers->desc_storage = CALLOC(num_buffers, sizeof(uint32_t) * 4);
 
-	/* si_emit_descriptors only accepts an array of arrays.
-	 * This adds such an array. */
-	buffers->desc_data = CALLOC(num_buffers, sizeof(uint32_t*));
-	for (i = 0; i < num_buffers; i++) {
-		buffers->desc_data[i] = &buffers->desc_storage[i*4];
-	}
-
-	si_init_descriptors(sctx, &buffers->desc,
-			    si_get_shader_user_data_base(shader) +
-			    shader_userdata_index*4, 4, num_buffers,
-			    si_emit_buffer_resources);
+	si_init_descriptors(&buffers->desc, shader_userdata_index, 4,
+			    num_buffers);
 }
 
 static void si_release_buffer_resources(struct si_buffer_resources *buffers)
 {
 	int i;
 
-	for (i = 0; i < buffers->num_buffers; i++) {
+	for (i = 0; i < buffers->desc.num_elements; i++) {
 		pipe_resource_reference(&buffers->buffers[i], NULL);
 	}
 
 	FREE(buffers->buffers);
-	FREE(buffers->desc_storage);
-	FREE(buffers->desc_data);
 	si_release_descriptors(&buffers->desc);
 }
 
@@ -590,11 +415,11 @@ static void si_buffer_resources_begin_new_cs(struct si_context *sctx,
 				      buffers->shader_usage, buffers->priority);
 	}
 
+	if (!buffers->desc.buffer)
+		return;
 	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 			      buffers->desc.buffer, RADEON_USAGE_READWRITE,
 			      RADEON_PRIO_SHADER_DATA);
-
-	si_emit_shader_pointer(sctx, &buffers->desc.atom);
 }
 
 /* VERTEX BUFFERS */
@@ -617,14 +442,15 @@ static void si_vertex_buffers_begin_new_cs(struct si_context *sctx)
 				      (struct r600_resource*)sctx->vertex_buffer[vb].buffer,
 				      RADEON_USAGE_READ, RADEON_PRIO_SHADER_BUFFER_RO);
 	}
+
+	if (!desc->buffer)
+		return;
 	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 			      desc->buffer, RADEON_USAGE_READ,
 			      RADEON_PRIO_SHADER_DATA);
-
-	si_emit_shader_pointer(sctx, &desc->atom);
 }
 
-void si_update_vertex_buffers(struct si_context *sctx)
+static bool si_upload_vertex_buffer_descriptors(struct si_context *sctx)
 {
 	struct si_descriptors *desc = &sctx->vertex_buffers;
 	bool bound[SI_NUM_VERTEX_BUFFERS] = {};
@@ -632,8 +458,10 @@ void si_update_vertex_buffers(struct si_context *sctx)
 	uint64_t va;
 	uint32_t *ptr;
 
+	if (!sctx->vertex_buffers_dirty)
+		return true;
 	if (!count || !sctx->vertex_elements)
-		return;
+		return true;
 
 	/* Vertex buffer descriptors are the only ones which are uploaded
 	 * directly through a staging buffer and don't go through
@@ -641,13 +469,14 @@ void si_update_vertex_buffers(struct si_context *sctx)
 	 */
 	u_upload_alloc(sctx->b.uploader, 0, count * 16, &desc->buffer_offset,
 		       (struct pipe_resource**)&desc->buffer, (void**)&ptr);
+	if (!desc->buffer)
+		return false;
 
 	r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 			      desc->buffer, RADEON_USAGE_READ,
 			      RADEON_PRIO_SHADER_DATA);
 
 	assert(count <= SI_NUM_VERTEX_BUFFERS);
-	assert(desc->current_context_id == 0);
 
 	for (i = 0; i < count; i++) {
 		struct pipe_vertex_element *ve = &sctx->vertex_elements->elements[i];
@@ -693,13 +522,14 @@ void si_update_vertex_buffers(struct si_context *sctx)
 		}
 	}
 
-	desc->atom.num_dw = 8; /* update 2 shader pointers (VS+ES) */
-	desc->atom.dirty = true;
-
 	/* Don't flush the const cache. It would have a very negative effect
 	 * on performance (confirmed by testing). New descriptors are always
 	 * uploaded to a fresh new buffer, so I don't think flushing the const
 	 * cache is needed. */
+	desc->pointer_dirty = true;
+	sctx->shader_userdata.atom.dirty = true;
+	sctx->vertex_buffers_dirty = false;
+	return true;
 }
 
 
@@ -724,7 +554,7 @@ static void si_set_constant_buffer(struct pipe_context *ctx, uint shader, uint s
 	if (shader >= SI_NUM_SHADERS)
 		return;
 
-	assert(slot < buffers->num_buffers);
+	assert(slot < buffers->desc.num_elements);
 	pipe_resource_reference(&buffers->buffers[slot], NULL);
 
 	/* CIK cannot unbind a constant buffer (S_BUFFER_LOAD is buggy
@@ -751,7 +581,7 @@ static void si_set_constant_buffer(struct pipe_context *ctx, uint shader, uint s
 		}
 
 		/* Set the descriptor. */
-		uint32_t *desc = buffers->desc_data[slot];
+		uint32_t *desc = buffers->desc.list + slot*4;
 		desc[0] = va;
 		desc[1] = S_008F04_BASE_ADDRESS_HI(va >> 32) |
 			  S_008F04_STRIDE(0);
@@ -770,12 +600,11 @@ static void si_set_constant_buffer(struct pipe_context *ctx, uint shader, uint s
 		buffers->desc.enabled_mask |= 1llu << slot;
 	} else {
 		/* Clear the descriptor. */
-		memset(buffers->desc_data[slot], 0, sizeof(uint32_t) * 4);
+		memset(buffers->desc.list + slot*4, 0, sizeof(uint32_t) * 4);
 		buffers->desc.enabled_mask &= ~(1llu << slot);
 	}
 
-	buffers->desc.dirty_mask |= 1llu << slot;
-	si_update_descriptors(sctx, &buffers->desc);
+	buffers->desc.list_dirty = true;
 }
 
 /* RING BUFFERS */
@@ -795,7 +624,7 @@ void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
 	/* The stride field in the resource descriptor has 14 bits */
 	assert(stride < (1 << 14));
 
-	assert(slot < buffers->num_buffers);
+	assert(slot < buffers->desc.num_elements);
 	pipe_resource_reference(&buffers->buffers[slot], NULL);
 
 	if (buffer) {
@@ -840,7 +669,7 @@ void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
 		}
 
 		/* Set the descriptor. */
-		uint32_t *desc = buffers->desc_data[slot];
+		uint32_t *desc = buffers->desc.list + slot*4;
 		desc[0] = va;
 		desc[1] = S_008F04_BASE_ADDRESS_HI(va >> 32) |
 			  S_008F04_STRIDE(stride) |
@@ -863,12 +692,11 @@ void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
 		buffers->desc.enabled_mask |= 1llu << slot;
 	} else {
 		/* Clear the descriptor. */
-		memset(buffers->desc_data[slot], 0, sizeof(uint32_t) * 4);
+		memset(buffers->desc.list + slot*4, 0, sizeof(uint32_t) * 4);
 		buffers->desc.enabled_mask &= ~(1llu << slot);
 	}
 
-	buffers->desc.dirty_mask |= 1llu << slot;
-	si_update_descriptors(sctx, &buffers->desc);
+	buffers->desc.list_dirty = true;
 }
 
 /* STREAMOUT BUFFERS */
@@ -930,7 +758,7 @@ static void si_set_streamout_targets(struct pipe_context *ctx,
 			uint64_t va = r600_resource(buffer)->gpu_address;
 
 			/* Set the descriptor. */
-			uint32_t *desc = buffers->desc_data[bufidx];
+			uint32_t *desc = buffers->desc.list + bufidx*4;
 			desc[0] = va;
 			desc[1] = S_008F04_BASE_ADDRESS_HI(va >> 32);
 			desc[2] = 0xffffffff;
@@ -948,24 +776,22 @@ static void si_set_streamout_targets(struct pipe_context *ctx,
 			buffers->desc.enabled_mask |= 1llu << bufidx;
 		} else {
 			/* Clear the descriptor and unset the resource. */
-			memset(buffers->desc_data[bufidx], 0,
+			memset(buffers->desc.list + bufidx*4, 0,
 			       sizeof(uint32_t) * 4);
 			pipe_resource_reference(&buffers->buffers[bufidx],
 						NULL);
 			buffers->desc.enabled_mask &= ~(1llu << bufidx);
 		}
-		buffers->desc.dirty_mask |= 1llu << bufidx;
 	}
 	for (; i < old_num_targets; i++) {
 		bufidx = SI_SO_BUF_OFFSET + i;
 		/* Clear the descriptor and unset the resource. */
-		memset(buffers->desc_data[bufidx], 0, sizeof(uint32_t) * 4);
+		memset(buffers->desc.list + bufidx*4, 0, sizeof(uint32_t) * 4);
 		pipe_resource_reference(&buffers->buffers[bufidx], NULL);
 		buffers->desc.enabled_mask &= ~(1llu << bufidx);
-		buffers->desc.dirty_mask |= 1llu << bufidx;
 	}
 
-	si_update_descriptors(sctx, &buffers->desc);
+	buffers->desc.list_dirty = true;
 }
 
 static void si_desc_reset_buffer_offset(struct pipe_context *ctx,
@@ -1034,22 +860,19 @@ static void si_invalidate_buffer(struct pipe_context *ctx, struct pipe_resource
 	/* Read/Write buffers. */
 	for (shader = 0; shader < SI_NUM_SHADERS; shader++) {
 		struct si_buffer_resources *buffers = &sctx->rw_buffers[shader];
-		bool found = false;
 		uint64_t mask = buffers->desc.enabled_mask;
 
 		while (mask) {
 			i = u_bit_scan64(&mask);
 			if (buffers->buffers[i] == buf) {
-				si_desc_reset_buffer_offset(ctx, buffers->desc_data[i],
+				si_desc_reset_buffer_offset(ctx, buffers->desc.list + i*4,
 							    old_va, buf);
+				buffers->desc.list_dirty = true;
 
 				r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 						      rbuffer, buffers->shader_usage,
 						      buffers->priority);
 
-				buffers->desc.dirty_mask |= 1llu << i;
-				found = true;
-
 				if (i >= SI_SO_BUF_OFFSET && shader == PIPE_SHADER_VERTEX) {
 					/* Update the streamout state. */
 					if (sctx->b.streamout.begin_emitted) {
@@ -1061,34 +884,25 @@ static void si_invalidate_buffer(struct pipe_context *ctx, struct pipe_resource
 				}
 			}
 		}
-		if (found) {
-			si_update_descriptors(sctx, &buffers->desc);
-		}
 	}
 
 	/* Constant buffers. */
 	for (shader = 0; shader < SI_NUM_SHADERS; shader++) {
 		struct si_buffer_resources *buffers = &sctx->const_buffers[shader];
-		bool found = false;
 		uint64_t mask = buffers->desc.enabled_mask;
 
 		while (mask) {
 			unsigned i = u_bit_scan64(&mask);
 			if (buffers->buffers[i] == buf) {
-				si_desc_reset_buffer_offset(ctx, buffers->desc_data[i],
+				si_desc_reset_buffer_offset(ctx, buffers->desc.list + i*4,
 							    old_va, buf);
+				buffers->desc.list_dirty = true;
 
 				r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 						      rbuffer, buffers->shader_usage,
 						      buffers->priority);
-
-				buffers->desc.dirty_mask |= 1llu << i;
-				found = true;
 			}
 		}
-		if (found) {
-			si_update_descriptors(sctx, &buffers->desc);
-		}
 	}
 
 	/* Texture buffers - update virtual addresses in sampler view descriptors. */
@@ -1100,23 +914,20 @@ static void si_invalidate_buffer(struct pipe_context *ctx, struct pipe_resource
 	/* Texture buffers - update bindings. */
 	for (shader = 0; shader < SI_NUM_SHADERS; shader++) {
 		struct si_sampler_views *views = &sctx->samplers[shader].views;
-		bool found = false;
 		uint64_t mask = views->desc.enabled_mask;
 
 		while (mask) {
 			unsigned i = u_bit_scan64(&mask);
 			if (views->views[i]->texture == buf) {
+				si_desc_reset_buffer_offset(ctx, views->desc.list + i*8+4,
+							    old_va, buf);
+				views->desc.list_dirty = true;
+
 				r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
 						      rbuffer, RADEON_USAGE_READ,
 						      RADEON_PRIO_SHADER_BUFFER_RO);
-
-				views->desc.dirty_mask |= 1llu << i;
-				found = true;
 			}
 		}
-		if (found) {
-			si_update_descriptors(sctx, &views->desc);
-		}
 	}
 }
 
@@ -1277,39 +1088,133 @@ void si_copy_buffer(struct si_context *sctx,
 		r600_resource(dst)->TC_L2_dirty = true;
 }
 
-/* INIT/DEINIT */
+/* SHADER USER DATA */
+
+static void si_mark_shader_pointers_dirty(struct si_context *sctx,
+					  unsigned shader)
+{
+	sctx->const_buffers[shader].desc.pointer_dirty = true;
+	sctx->rw_buffers[shader].desc.pointer_dirty = true;
+	sctx->samplers[shader].views.desc.pointer_dirty = true;
+	sctx->samplers[shader].states.desc.pointer_dirty = true;
+
+	if (shader == PIPE_SHADER_VERTEX)
+		sctx->vertex_buffers.pointer_dirty = true;
+
+	sctx->shader_userdata.atom.dirty = true;
+}
+
+static void si_shader_userdata_begin_new_cs(struct si_context *sctx)
+{
+	int i;
+
+	for (i = 0; i < SI_NUM_SHADERS; i++) {
+		si_mark_shader_pointers_dirty(sctx, i);
+	}
+}
+
+/* Set a base register address for user data constants in the given shader.
+ * This assigns a mapping from PIPE_SHADER_* to SPI_SHADER_USER_DATA_*.
+ */
+static void si_set_user_data_base(struct si_context *sctx,
+				  unsigned shader, uint32_t new_base)
+{
+	uint32_t *base = &sctx->shader_userdata.sh_base[shader];
+
+	if (*base != new_base) {
+		*base = new_base;
+
+		if (new_base)
+			si_mark_shader_pointers_dirty(sctx, shader);
+	}
+}
+
+/* This must be called when these shaders are changed from non-NULL to NULL
+ * and vice versa:
+ * - geometry shader
+ * - tessellation control shader
+ * - tessellation evaluation shader
+ */
+void si_shader_change_notify(struct si_context *sctx)
+{
+	/* VS can be bound as VS or ES. */
+	if (sctx->gs_shader)
+		si_set_user_data_base(sctx, PIPE_SHADER_VERTEX,
+				      R_00B330_SPI_SHADER_USER_DATA_ES_0);
+	else
+		si_set_user_data_base(sctx, PIPE_SHADER_VERTEX,
+				      R_00B130_SPI_SHADER_USER_DATA_VS_0);
+}
+
+static void si_emit_shader_pointer(struct si_context *sctx,
+				   struct si_descriptors *desc,
+				   unsigned sh_base, bool keep_dirty)
+{
+	struct radeon_winsys_cs *cs = sctx->b.rings.gfx.cs;
+	uint64_t va;
+
+	if (!desc->pointer_dirty || !desc->buffer)
+		return;
+
+	va = desc->buffer->gpu_address +
+	     desc->buffer_offset;
+
+	radeon_emit(cs, PKT3(PKT3_SET_SH_REG, 2, 0));
+	radeon_emit(cs, (sh_base + desc->shader_userdata_offset - SI_SH_REG_OFFSET) >> 2);
+	radeon_emit(cs, va);
+	radeon_emit(cs, va >> 32);
+
+	desc->pointer_dirty = keep_dirty;
+}
+
+static void si_emit_shader_userdata(struct si_context *sctx,
+				    struct r600_atom *atom)
+{
+	unsigned i;
+	uint32_t *sh_base = sctx->shader_userdata.sh_base;
+
+	/* The VS copy shader needs these for clipping, streamout, and rings. */
+	if (sctx->gs_shader) {
+		unsigned base = R_00B130_SPI_SHADER_USER_DATA_VS_0;
+		unsigned i = PIPE_SHADER_VERTEX;
+
+		si_emit_shader_pointer(sctx, &sctx->const_buffers[i].desc, base, true);
+		si_emit_shader_pointer(sctx, &sctx->rw_buffers[i].desc, base, true);
+	}
+
+	for (i = 0; i < SI_NUM_SHADERS; i++) {
+		unsigned base = sh_base[i];
+
+		si_emit_shader_pointer(sctx, &sctx->const_buffers[i].desc, base, false);
+		si_emit_shader_pointer(sctx, &sctx->rw_buffers[i].desc, base, false);
+		si_emit_shader_pointer(sctx, &sctx->samplers[i].views.desc, base, false);
+		si_emit_shader_pointer(sctx, &sctx->samplers[i].states.desc, base, false);
+	}
+	si_emit_shader_pointer(sctx, &sctx->vertex_buffers, sh_base[PIPE_SHADER_VERTEX], false);
+}
+
+/* INIT/DEINIT/UPLOAD */
 
 void si_init_all_descriptors(struct si_context *sctx)
 {
 	int i;
 
 	for (i = 0; i < SI_NUM_SHADERS; i++) {
-		si_init_buffer_resources(sctx, &sctx->const_buffers[i],
-					 SI_NUM_CONST_BUFFERS, i, SI_SGPR_CONST,
+		si_init_buffer_resources(&sctx->const_buffers[i],
+					 SI_NUM_CONST_BUFFERS, SI_SGPR_CONST,
 					 RADEON_USAGE_READ, RADEON_PRIO_SHADER_BUFFER_RO);
-		si_init_buffer_resources(sctx, &sctx->rw_buffers[i],
-					 i == PIPE_SHADER_VERTEX ?
-					 SI_NUM_RW_BUFFERS : SI_NUM_RING_BUFFERS,
-					 i, SI_SGPR_RW_BUFFERS,
+		si_init_buffer_resources(&sctx->rw_buffers[i],
+					 SI_NUM_RW_BUFFERS, SI_SGPR_RW_BUFFERS,
 					 RADEON_USAGE_READWRITE, RADEON_PRIO_SHADER_RESOURCE_RW);
 
-		si_init_sampler_views(sctx, &sctx->samplers[i].views, i);
-
-		si_init_descriptors(sctx, &sctx->samplers[i].states.desc,
-				    si_get_shader_user_data_base(i) + SI_SGPR_SAMPLER * 4,
-				    4, SI_NUM_SAMPLER_STATES, si_emit_sampler_states);
-
-		sctx->atoms.s.const_buffers[i] = &sctx->const_buffers[i].desc.atom;
-		sctx->atoms.s.rw_buffers[i] = &sctx->rw_buffers[i].desc.atom;
-		sctx->atoms.s.sampler_views[i] = &sctx->samplers[i].views.desc.atom;
-		sctx->atoms.s.sampler_states[i] = &sctx->samplers[i].states.desc.atom;
+		si_init_descriptors(&sctx->samplers[i].views.desc,
+				    SI_SGPR_RESOURCE, 8, SI_NUM_SAMPLER_VIEWS);
+		si_init_descriptors(&sctx->samplers[i].states.desc,
+				    SI_SGPR_SAMPLER, 4, SI_NUM_SAMPLER_STATES);
 	}
 
-	si_init_descriptors(sctx, &sctx->vertex_buffers,
-			    si_get_shader_user_data_base(PIPE_SHADER_VERTEX) +
-			    SI_SGPR_VERTEX_BUFFER*4, 4, SI_NUM_VERTEX_BUFFERS,
-			    si_emit_shader_pointer);
-	sctx->atoms.s.vertex_buffers = &sctx->vertex_buffers.atom;
+	si_init_descriptors(&sctx->vertex_buffers, SI_SGPR_VERTEX_BUFFER,
+			    4, SI_NUM_VERTEX_BUFFERS);
 
 	/* Set pipe_context functions. */
 	sctx->b.b.set_constant_buffer = si_set_constant_buffer;
@@ -1317,6 +1222,32 @@ void si_init_all_descriptors(struct si_context *sctx)
 	sctx->b.b.set_stream_output_targets = si_set_streamout_targets;
 	sctx->b.clear_buffer = si_clear_buffer;
 	sctx->b.invalidate_buffer = si_invalidate_buffer;
+
+	/* Shader user data. */
+	sctx->atoms.s.shader_userdata = &sctx->shader_userdata.atom;
+	sctx->shader_userdata.atom.emit = (void*)si_emit_shader_userdata;
+
+	/* Upper bound, 4 pointers per shader, +1 for vertex buffers, +2 for the VS copy shader. */
+	sctx->shader_userdata.atom.num_dw = (SI_NUM_SHADERS * 4 + 1 + 2) * 4;
+
+	/* Set default and immutable mappings. */
+	si_set_user_data_base(sctx, PIPE_SHADER_VERTEX, R_00B130_SPI_SHADER_USER_DATA_VS_0);
+	si_set_user_data_base(sctx, PIPE_SHADER_GEOMETRY, R_00B230_SPI_SHADER_USER_DATA_GS_0);
+	si_set_user_data_base(sctx, PIPE_SHADER_FRAGMENT, R_00B030_SPI_SHADER_USER_DATA_PS_0);
+}
+
+bool si_upload_shader_descriptors(struct si_context *sctx)
+{
+	int i;
+
+	for (i = 0; i < SI_NUM_SHADERS; i++) {
+		if (!si_upload_descriptors(sctx, &sctx->const_buffers[i].desc) ||
+		    !si_upload_descriptors(sctx, &sctx->rw_buffers[i].desc) ||
+		    !si_upload_descriptors(sctx, &sctx->samplers[i].views.desc) ||
+		    !si_upload_descriptors(sctx, &sctx->samplers[i].states.desc))
+			return false;
+	}
+	return si_upload_vertex_buffer_descriptors(sctx);
 }
 
 void si_release_all_descriptors(struct si_context *sctx)
@@ -1343,4 +1274,5 @@ void si_all_descriptors_begin_new_cs(struct si_context *sctx)
 		si_sampler_states_begin_new_cs(sctx, &sctx->samplers[i].states);
 	}
 	si_vertex_buffers_begin_new_cs(sctx);
+	si_shader_userdata_begin_new_cs(sctx);
 }

src/gallium/drivers/radeonsi/si_pipe.h

@@ -125,8 +125,6 @@ struct si_framebuffer {
 
 #define SI_NUM_ATOMS(sctx) (sizeof((sctx)->atoms)/sizeof((sctx)->atoms.array[0]))
 
-#define SI_NUM_SHADERS (PIPE_SHADER_GEOMETRY+1)
-
 struct si_context {
 	struct r600_common_context	b;
 	struct blitter_context		*blitter;
@@ -141,13 +139,6 @@ struct si_context {
 	union {
 		struct {
 			/* The order matters. */
-			struct r600_atom *vertex_buffers;
-			struct r600_atom *const_buffers[SI_NUM_SHADERS];
-			struct r600_atom *rw_buffers[SI_NUM_SHADERS];
-			struct r600_atom *sampler_views[SI_NUM_SHADERS];
-			struct r600_atom *sampler_states[SI_NUM_SHADERS];
-			/* Caches must be flushed after resource descriptors are
-			 * updated in memory. */
 			struct r600_atom *cache_flush;
 			struct r600_atom *streamout_begin;
 			struct r600_atom *streamout_enable; /* must be after streamout_begin */
@@ -156,6 +147,7 @@ struct si_context {
 			struct r600_atom *db_render_state;
 			struct r600_atom *msaa_config;
 			struct r600_atom *clip_regs;
+			struct r600_atom *shader_userdata;
 		} s;
 		struct r600_atom *array[0];
 	} atoms;
@@ -169,6 +161,7 @@ struct si_context {
 	struct si_shader_selector	*gs_shader;
 	struct si_shader_selector	*vs_shader;
 	struct si_cs_shader_state	cs_shader_state;
+	struct si_shader_data		shader_userdata;
 	/* shader information */
 	unsigned			sprite_coord_enable;
 	bool				flatshade;

src/gallium/drivers/radeonsi/si_shader.c

@@ -2647,18 +2647,43 @@ void si_shader_apply_scratch_relocs(struct si_context *sctx,
 	}
 }
 
+int si_shader_binary_upload(struct si_screen *sscreen, struct si_shader *shader)
+{
+	const struct radeon_shader_binary *binary = &shader->binary;
+	unsigned code_size = binary->code_size + binary->rodata_size;
+	unsigned char *ptr;
+
+	r600_resource_reference(&shader->bo, NULL);
+	shader->bo = si_resource_create_custom(&sscreen->b.b,
+					       PIPE_USAGE_IMMUTABLE,
+					       code_size);
+	if (!shader->bo)
+		return -ENOMEM;
+
+	ptr = sscreen->b.ws->buffer_map(shader->bo->cs_buf, NULL,
+					PIPE_TRANSFER_READ_WRITE);
+	util_memcpy_cpu_to_le32(ptr, binary->code, binary->code_size);
+	if (binary->rodata_size > 0) {
+		ptr += binary->code_size;
+		util_memcpy_cpu_to_le32(ptr, binary->rodata,
+					binary->rodata_size);
+	}
+
+	sscreen->b.ws->buffer_unmap(shader->bo->cs_buf);
+	return 0;
+}
+
 int si_shader_binary_read(struct si_screen *sscreen,
 			struct si_shader *shader,
 			const struct radeon_shader_binary *binary)
 {
 
 	unsigned i;
-	unsigned code_size;
-	unsigned char *ptr;
 	bool dump  = r600_can_dump_shader(&sscreen->b,
 		shader->selector ? shader->selector->tokens : NULL);
 
 	si_shader_binary_read_config(sscreen, shader, 0);
+	si_shader_binary_upload(sscreen, shader);
 
 	if (dump) {
 		if (!binary->disassembled) {
@@ -2676,26 +2701,6 @@ int si_shader_binary_read(struct si_screen *sscreen,
 			shader->num_sgprs, shader->num_vgprs, binary->code_size,
 			shader->lds_size, shader->scratch_bytes_per_wave);
 	}
-
-	/* copy new shader */
-	code_size = binary->code_size + binary->rodata_size;
-	r600_resource_reference(&shader->bo, NULL);
-	shader->bo = si_resource_create_custom(&sscreen->b.b, PIPE_USAGE_IMMUTABLE,
-					       code_size);
-	if (shader->bo == NULL) {
-		return -ENOMEM;
-	}
-
-
-	ptr = sscreen->b.ws->buffer_map(shader->bo->cs_buf, NULL, PIPE_TRANSFER_READ_WRITE);
-	util_memcpy_cpu_to_le32(ptr, binary->code, binary->code_size);
-	if (binary->rodata_size > 0) {
-		ptr += binary->code_size;
-		util_memcpy_cpu_to_le32(ptr, binary->rodata, binary->rodata_size);
-	}
-
-	sscreen->b.ws->buffer_unmap(shader->bo->cs_buf);
-
 	return 0;
 }
 

src/gallium/drivers/radeonsi/si_shader.h

@@ -189,6 +189,7 @@ int si_compile_llvm(struct si_screen *sscreen, struct si_shader *shader,
 		    LLVMTargetMachineRef tm, LLVMModuleRef mod);
 void si_shader_destroy(struct pipe_context *ctx, struct si_shader *shader);
 unsigned si_shader_io_get_unique_index(unsigned semantic_name, unsigned index);
+int si_shader_binary_upload(struct si_screen *sscreen, struct si_shader *shader);
 int si_shader_binary_read(struct si_screen *sscreen, struct si_shader *shader,
 		const struct radeon_shader_binary *binary);
 void si_shader_apply_scratch_relocs(struct si_context *sctx,

src/gallium/drivers/radeonsi/si_state.h

@@ -30,6 +30,8 @@
 #include "si_pm4.h"
 #include "radeon/r600_pipe_common.h"
 
+#define SI_NUM_SHADERS (PIPE_SHADER_GEOMETRY+1)
+
 struct si_screen;
 struct si_shader;
 
@@ -111,6 +113,11 @@ union si_state {
 	struct si_pm4_state	*array[0];
 };
 
+struct si_shader_data {
+	struct r600_atom	atom;
+	uint32_t		sh_base[SI_NUM_SHADERS];
+};
+
 #define SI_NUM_USER_SAMPLERS            16 /* AKA OpenGL textures units per shader */
 #define SI_POLY_STIPPLE_SAMPLER         SI_NUM_USER_SAMPLERS
 #define SI_NUM_SAMPLERS                 (SI_POLY_STIPPLE_SAMPLER + 1)
@@ -144,59 +151,48 @@ union si_state {
 #define SI_NUM_VERTEX_BUFFERS	16
 
 
-/* This represents resource descriptors in memory, such as buffer resources,
+/* This represents descriptors in memory, such as buffer resources,
  * image resources, and sampler states.
  */
 struct si_descriptors {
-	struct r600_atom atom;
-
-	/* The size of one resource descriptor. */
+	/* The list of descriptors in malloc'd memory. */
+	uint32_t *list;
+	/* The size of one descriptor. */
 	unsigned element_dw_size;
-	/* The maximum number of resource descriptors. */
+	/* The maximum number of descriptors. */
 	unsigned num_elements;
+	/* Whether the list has been changed and should be re-uploaded. */
+	bool list_dirty;
 
-	/* The buffer where resource descriptors are stored. */
+	/* The buffer where the descriptors have been uploaded. */
 	struct r600_resource *buffer;
 	unsigned buffer_offset;
 
-	/* The i-th bit is set if that element is dirty (changed but not emitted). */
-	uint64_t dirty_mask;
 	/* The i-th bit is set if that element is enabled (non-NULL resource). */
 	uint64_t enabled_mask;
 
-	/* We can't update descriptors directly because the GPU might be
-	 * reading them at the same time, so we have to update them
-	 * in a copy-on-write manner. Each such copy is called a context,
-	 * which is just another array descriptors in the same buffer. */
-	unsigned current_context_id;
-	/* The size of a context, should be equal to 4*element_dw_size*num_elements. */
-	unsigned context_size;
-
-	/* The shader userdata register where the 64-bit pointer to the descriptor
+	/* The shader userdata offset within a shader where the 64-bit pointer to the descriptor
 	 * array will be stored. */
-	unsigned shader_userdata_reg;
+	unsigned shader_userdata_offset;
+	/* Whether the pointer should be re-emitted. */
+	bool pointer_dirty;
 };
 
 struct si_sampler_views {
 	struct si_descriptors		desc;
 	struct pipe_sampler_view	*views[SI_NUM_SAMPLER_VIEWS];
-	uint32_t			*desc_data[SI_NUM_SAMPLER_VIEWS];
 };
 
 struct si_sampler_states {
 	struct si_descriptors		desc;
-	uint32_t			*desc_data[SI_NUM_SAMPLER_STATES];
 	void				*saved_states[2]; /* saved for u_blitter */
 };
 
 struct si_buffer_resources {
 	struct si_descriptors		desc;
-	unsigned			num_buffers;
 	enum radeon_bo_usage		shader_usage; /* READ, WRITE, or READWRITE */
 	enum radeon_bo_priority		priority;
 	struct pipe_resource		**buffers; /* this has num_buffers elements */
-	uint32_t			*desc_storage; /* this has num_buffers*4 elements */
-	uint32_t			**desc_data; /* an array of pointers pointing to desc_storage */
 };
 
 #define si_pm4_block_idx(member) \
@@ -232,13 +228,13 @@ struct si_buffer_resources {
 /* si_descriptors.c */
 void si_set_sampler_descriptors(struct si_context *sctx, unsigned shader,
 				unsigned start, unsigned count, void **states);
-void si_update_vertex_buffers(struct si_context *sctx);
 void si_set_ring_buffer(struct pipe_context *ctx, uint shader, uint slot,
 			struct pipe_resource *buffer,
 			unsigned stride, unsigned num_records,
 			bool add_tid, bool swizzle,
 			unsigned element_size, unsigned index_stride);
 void si_init_all_descriptors(struct si_context *sctx);
+bool si_upload_shader_descriptors(struct si_context *sctx);
 void si_release_all_descriptors(struct si_context *sctx);
 void si_all_descriptors_begin_new_cs(struct si_context *sctx);
 void si_copy_buffer(struct si_context *sctx,
@@ -246,6 +242,7 @@ void si_copy_buffer(struct si_context *sctx,
 		    uint64_t dst_offset, uint64_t src_offset, unsigned size, bool is_framebuffer);
 void si_upload_const_buffer(struct si_context *sctx, struct r600_resource **rbuffer,
 			    const uint8_t *ptr, unsigned size, uint32_t *const_offset);
+void si_shader_change_notify(struct si_context *sctx);
 
 /* si_state.c */
 struct si_shader_selector;

src/gallium/drivers/radeonsi/si_state_draw.c

@@ -139,6 +139,10 @@ static unsigned si_get_ia_multi_vgt_param(struct si_context *sctx,
 		    (info->indirect || info->instance_count > 1))
 			wd_switch_on_eop = true;
 
+		/* USE_OPAQUE doesn't work when WD_SWITCH_ON_EOP is 0. */
+		if (info->count_from_stream_output)
+			wd_switch_on_eop = true;
+
 		/* If the WD switch is false, the IA switch must be false too. */
 		assert(wd_switch_on_eop || !ia_switch_on_eop);
 	}
@@ -245,8 +249,7 @@ static void si_emit_draw_packets(struct si_context *sctx,
 				 const struct pipe_index_buffer *ib)
 {
 	struct radeon_winsys_cs *cs = sctx->b.rings.gfx.cs;
-	unsigned sh_base_reg = (sctx->gs_shader ? R_00B330_SPI_SHADER_USER_DATA_ES_0 :
-						  R_00B130_SPI_SHADER_USER_DATA_VS_0);
+	unsigned sh_base_reg = sctx->shader_userdata.sh_base[PIPE_SHADER_VERTEX];
 
 	if (info->count_from_stream_output) {
 		struct r600_so_target *t =
@@ -536,11 +539,8 @@ void si_draw_vbo(struct pipe_context *ctx, const struct pipe_draw_info *info)
 		sctx->current_rast_prim = info->mode;
 
 	si_update_shaders(sctx);
-
-	if (sctx->vertex_buffers_dirty) {
-		si_update_vertex_buffers(sctx);
-		sctx->vertex_buffers_dirty = false;
-	}
+	if (!si_upload_shader_descriptors(sctx))
+		return;
 
 	if (info->indexed) {
 		/* Initialize the index buffer struct. */

src/gallium/drivers/radeonsi/si_state_shaders.c

@@ -535,6 +535,7 @@ static void si_bind_gs_shader(struct pipe_context *ctx, void *state)
 {
 	struct si_context *sctx = (struct si_context *)ctx;
 	struct si_shader_selector *sel = state;
+	bool enable_changed = !!sctx->gs_shader != !!sel;
 
 	if (sctx->gs_shader == sel)
 		return;
@@ -542,6 +543,9 @@ static void si_bind_gs_shader(struct pipe_context *ctx, void *state)
 	sctx->gs_shader = sel;
 	sctx->clip_regs.dirty = true;
 	sctx->last_rast_prim = -1; /* reset this so that it gets updated */
+
+	if (enable_changed)
+		si_shader_change_notify(sctx);
 }
 
 static void si_make_dummy_ps(struct si_context *sctx)
@@ -743,7 +747,6 @@ static unsigned si_update_scratch_buffer(struct si_context *sctx,
 {
 	struct si_shader *shader;
 	uint64_t scratch_va = sctx->scratch_buffer->gpu_address;
-	unsigned char *ptr;
 
 	if (!sel)
 		return 0;
@@ -764,12 +767,7 @@ static unsigned si_update_scratch_buffer(struct si_context *sctx,
 	si_shader_apply_scratch_relocs(sctx, shader, scratch_va);
 
 	/* Replace the shader bo with a new bo that has the relocs applied. */
-	r600_resource_reference(&shader->bo, NULL);
-	shader->bo = si_resource_create_custom(&sctx->screen->b.b, PIPE_USAGE_IMMUTABLE,
-					       shader->binary.code_size);
-	ptr = sctx->screen->b.ws->buffer_map(shader->bo->cs_buf, NULL, PIPE_TRANSFER_WRITE);
-	util_memcpy_cpu_to_le32(ptr, shader->binary.code, shader->binary.code_size);
-	sctx->screen->b.ws->buffer_unmap(shader->bo->cs_buf);
+	si_shader_binary_upload(sctx->screen, shader);
 
 	/* Update the shader state to use the new shader bo. */
 	si_shader_init_pm4_state(shader);
@@ -835,8 +833,15 @@ static void si_update_spi_tmpring_size(struct si_context *sctx)
 			si_pm4_bind_state(sctx, ps, sctx->ps_shader->current->pm4);
 		if (si_update_scratch_buffer(sctx, sctx->gs_shader))
 			si_pm4_bind_state(sctx, gs, sctx->gs_shader->current->pm4);
-		if (si_update_scratch_buffer(sctx, sctx->vs_shader))
-			si_pm4_bind_state(sctx, vs, sctx->vs_shader->current->pm4);
+
+		/* VS can be bound as ES or VS. */
+		if (sctx->gs_shader) {
+			if (si_update_scratch_buffer(sctx, sctx->vs_shader))
+				si_pm4_bind_state(sctx, es, sctx->vs_shader->current->pm4);
+		} else {
+			if (si_update_scratch_buffer(sctx, sctx->vs_shader))
+				si_pm4_bind_state(sctx, vs, sctx->vs_shader->current->pm4);
+		}
 	}
 
 	/* The LLVM shader backend should be reporting aligned scratch_sizes. */

src/gallium/state_trackers/clover/api/interop.cpp

@@ -31,7 +31,12 @@ extern "C" {
 PUBLIC bool
 opencl_dri_event_add_ref(cl_event event)
 {
-   return clRetainEvent(event) == CL_SUCCESS;
+   /* This should fail if the event hasn't been created by
+    * clEnqueueReleaseGLObjects or clEnqueueReleaseEGLObjects.
+    *
+    * TODO: implement the CL functions
+    */
+   return false; /*return clRetainEvent(event) == CL_SUCCESS;*/
 }
 
 PUBLIC bool

src/gallium/state_trackers/clover/core/error.hpp

@@ -26,6 +26,7 @@
 #include "CL/cl.h"
 
 #include <stdexcept>
+#include <string>
 
 namespace clover {
    class command_queue;

src/gallium/state_trackers/dri/dri2.c

@@ -554,7 +554,7 @@ dri2_allocate_textures(struct dri_context *ctx,
 
          if (drawable->textures[statt]) {
             templ.format = drawable->textures[statt]->format;
-            templ.bind = drawable->textures[statt]->bind;
+            templ.bind = drawable->textures[statt]->bind & ~PIPE_BIND_SCANOUT;
             templ.nr_samples = drawable->stvis.samples;
 
             /* Try to reuse the resource.
@@ -1483,7 +1483,7 @@ dri2_init_screen(__DRIscreen * sPriv)
       }
    }
 
-   if (pscreen->get_param(pscreen, PIPE_CAP_DEVICE_RESET_STATUS_QUERY)) {
+   if (pscreen && pscreen->get_param(pscreen, PIPE_CAP_DEVICE_RESET_STATUS_QUERY)) {
       sPriv->extensions = dri_robust_screen_extensions;
       screen->has_reset_status_query = true;
    }

src/gallium/state_trackers/dri/dri_context.c

@@ -244,11 +244,10 @@ dri_make_current(__DRIcontext * cPriv,
 
    ctx->stapi->make_current(ctx->stapi, ctx->st, &draw->base, &read->base);
 
-   // This is ok to call here. If they are already init, it's a no-op.
-   if (draw->textures[ST_ATTACHMENT_BACK_LEFT] && draw->textures[ST_ATTACHMENT_DEPTH_STENCIL]
-      && ctx->pp)
-         pp_init_fbos(ctx->pp, draw->textures[ST_ATTACHMENT_BACK_LEFT]->width0,
-            draw->textures[ST_ATTACHMENT_BACK_LEFT]->height0);
+   /* This is ok to call here. If they are already init, it's a no-op. */
+   if (ctx->pp && draw->textures[ST_ATTACHMENT_BACK_LEFT])
+      pp_init_fbos(ctx->pp, draw->textures[ST_ATTACHMENT_BACK_LEFT]->width0,
+                   draw->textures[ST_ATTACHMENT_BACK_LEFT]->height0);
 
    return GL_TRUE;
 }

src/gallium/state_trackers/osmesa/osmesa.c

@@ -886,7 +886,7 @@ static struct name_function functions[] = {
    { "OSMesaDestroyContext", (OSMESAproc) OSMesaDestroyContext },
    { "OSMesaMakeCurrent", (OSMESAproc) OSMesaMakeCurrent },
    { "OSMesaGetCurrentContext", (OSMESAproc) OSMesaGetCurrentContext },
-   { "OSMesaPixelsStore", (OSMESAproc) OSMesaPixelStore },
+   { "OSMesaPixelStore", (OSMESAproc) OSMesaPixelStore },
    { "OSMesaGetIntegerv", (OSMESAproc) OSMesaGetIntegerv },
    { "OSMesaGetDepthBuffer", (OSMESAproc) OSMesaGetDepthBuffer },
    { "OSMesaGetColorBuffer", (OSMESAproc) OSMesaGetColorBuffer },

src/gallium/state_trackers/vdpau/mixer.c

@@ -49,7 +49,8 @@ vlVdpVideoMixerCreate(VdpDevice device,
    vlVdpVideoMixer *vmixer = NULL;
    VdpStatus ret;
    struct pipe_screen *screen;
-   unsigned max_width, max_height, i;
+   uint32_t max_2d_texture_level;
+   unsigned max_size, i;
 
    vlVdpDevice *dev = vlGetDataHTAB(device);
    if (!dev)
@@ -134,18 +135,17 @@ vlVdpVideoMixerCreate(VdpDevice device,
       VDPAU_MSG(VDPAU_WARN, "[VDPAU] Max layers > 4 not supported\n", vmixer->max_layers);
       goto no_params;
    }
-   max_width = screen->get_video_param(screen, PIPE_VIDEO_PROFILE_UNKNOWN,
-                                       PIPE_VIDEO_ENTRYPOINT_BITSTREAM, PIPE_VIDEO_CAP_MAX_WIDTH);
-   max_height = screen->get_video_param(screen, PIPE_VIDEO_PROFILE_UNKNOWN,
-                                        PIPE_VIDEO_ENTRYPOINT_BITSTREAM, PIPE_VIDEO_CAP_MAX_HEIGHT);
-   if (vmixer->video_width < 48 ||
-       vmixer->video_width > max_width) {
-      VDPAU_MSG(VDPAU_WARN, "[VDPAU] 48 < %u < %u not valid for width\n", vmixer->video_width, max_width);
+
+   max_2d_texture_level = screen->get_param(screen, PIPE_CAP_MAX_TEXTURE_2D_LEVELS);
+   max_size = pow(2, max_2d_texture_level-1);
+   if (vmixer->video_width < 48 || vmixer->video_width > max_size) {
+      VDPAU_MSG(VDPAU_WARN, "[VDPAU] 48 < %u < %u not valid for width\n",
+                vmixer->video_width, max_size);
       goto no_params;
    }
-   if (vmixer->video_height < 48 ||
-       vmixer->video_height > max_height) {
-      VDPAU_MSG(VDPAU_WARN, "[VDPAU] 48 < %u < %u  not valid for height\n", vmixer->video_height, max_height);
+   if (vmixer->video_height < 48 || vmixer->video_height > max_size) {
+      VDPAU_MSG(VDPAU_WARN, "[VDPAU] 48 < %u < %u  not valid for height\n",
+                vmixer->video_height, max_size);
       goto no_params;
    }
    vmixer->luma_key_min = 0.f;

src/gallium/state_trackers/xa/xa_context.c

@@ -37,7 +37,11 @@
 XA_EXPORT void
 xa_context_flush(struct xa_context *ctx)
 {
-	ctx->pipe->flush(ctx->pipe, &ctx->last_fence, 0);
+    if (ctx->last_fence) {
+        struct pipe_screen *screen = ctx->xa->screen;
+        screen->fence_reference(screen, &ctx->last_fence, NULL);
+    }
+    ctx->pipe->flush(ctx->pipe, &ctx->last_fence, 0);
 }
 
 XA_EXPORT struct xa_context *

src/gallium/state_trackers/xa/xa_tracker.c

@@ -461,7 +461,7 @@ xa_surface_redefine(struct xa_surface *srf,
 			xa_min(save_height, template->height0), &src_box);
 	pipe->resource_copy_region(pipe, texture,
 				   0, 0, 0, 0, srf->tex, 0, &src_box);
-	pipe->flush(pipe, &xa->default_ctx->last_fence, 0);
+	xa_context_flush(xa->default_ctx);
     }
 
     pipe_resource_reference(&srf->tex, texture);
@@ -535,15 +535,3 @@ xa_surface_format(const struct xa_surface *srf)
 {
     return srf->fdesc.xa_format;
 }
-
-/*
- * _mesa_error_no_memory() is expected by NIR to be provided by the
- * user.  Normally this is in mesa st, but other state trackers
- * must provide their own.
- */
-void _mesa_error_no_memory(const char *caller);
-void
-_mesa_error_no_memory(const char *caller)
-{
-	debug_printf("Mesa error: out of memory in %s", caller);
-}

src/gallium/state_trackers/xa/xa_yuv.c

@@ -154,7 +154,7 @@ xa_yuv_planar_blit(struct xa_context *r,
 	box++;
     }
 
-    r->pipe->flush(r->pipe, &r->last_fence, 0);
+    xa_context_flush(r);
 
     xa_ctx_sampler_views_destroy(r);
     xa_ctx_srf_destroy(r);

src/gallium/targets/d3dadapter9/Makefile.am

@@ -74,6 +74,8 @@ endif # HAVE_LD_VERSION_SCRIPT
 d3dadapter9_la_LIBADD = \
 	$(top_builddir)/src/gallium/auxiliary/libgalliumvl_stub.la \
 	$(top_builddir)/src/gallium/auxiliary/libgallium.la \
+	$(top_builddir)/src/glsl/libnir.la \
+	$(top_builddir)/src/libglsl_util.la \
 	$(top_builddir)/src/gallium/state_trackers/nine/libninetracker.la \
 	$(top_builddir)/src/util/libmesautil.la \
 	$(top_builddir)/src/gallium/winsys/sw/wrapper/libwsw.la \

src/gallium/targets/dri/Makefile.am

@@ -53,12 +53,6 @@ gallium_dri_la_LIBADD = \
 	$(LIBDRM_LIBS) \
 	$(GALLIUM_COMMON_LIB_DEPS)
 
-# XXX: Temporary allow duplicated symbols, as the loader pulls in xmlconfig.c
-# which already provides driParse* and driQuery* amongst others.
-# Remove this hack as we come up with a cleaner solution.
-gallium_dri_la_LDFLAGS += \
-	-Wl,--allow-multiple-definition
-
 EXTRA_gallium_dri_la_DEPENDENCIES = \
 	dri.sym \
 	$(top_srcdir)/src/gallium/targets/dri-vdpau.dyn

src/gallium/targets/opencl/Makefile.am

@@ -7,7 +7,7 @@ lib_LTLIBRARIES = lib@OPENCL_LIBNAME@.la
 lib@OPENCL_LIBNAME@_la_LDFLAGS = \
 	$(LLVM_LDFLAGS) \
 	-no-undefined \
-	-version-number 1:0 \
+	-version-number @OPENCL_VERSION@:0 \
 	$(GC_SECTIONS) \
 	$(LD_NO_UNDEFINED)
 

src/gallium/targets/opencl/mesa.icd

@@ -1 +0,0 @@
-libMesaOpenCL.so

src/gallium/targets/opencl/mesa.icd.in

@@ -0,0 +1 @@
+lib@OPENCL_LIBNAME@.so.@OPENCL_VERSION@

src/gallium/targets/osmesa/Makefile.am

@@ -42,7 +42,6 @@ nodist_EXTRA_lib@OSMESA_LIB@_la_SOURCES = dummy.cpp
 lib@OSMESA_LIB@_la_SOURCES = target.c
 
 lib@OSMESA_LIB@_la_LDFLAGS = \
-	-module \
 	-no-undefined \
 	-version-number @OSMESA_VERSION@ \
 	$(GC_SECTIONS) \

src/gallium/targets/pipe-loader/Makefile.am

@@ -52,6 +52,8 @@ endif
 
 PIPE_LIBS += \
 	$(top_builddir)/src/gallium/auxiliary/libgallium.la \
+	$(top_builddir)/src/glsl/libnir.la \
+	$(top_builddir)/src/libglsl_util.la \
 	$(top_builddir)/src/util/libmesautil.la \
 	$(top_builddir)/src/gallium/drivers/rbug/librbug.la \
 	$(top_builddir)/src/gallium/drivers/trace/libtrace.la \

src/gallium/winsys/nouveau/drm/nouveau_drm_winsys.c

@@ -120,7 +120,11 @@ nouveau_drm_screen_create(int fd)
 	if (!screen)
 		goto err;
 
-	util_hash_table_set(fd_tab, intptr_to_pointer(fd), screen);
+	/* Use dupfd in hash table, to avoid errors if the original fd gets
+	 * closed by its owner. The hash key needs to live at least as long as
+	 * the screen.
+	 */
+	util_hash_table_set(fd_tab, intptr_to_pointer(dupfd), screen);
 	screen->refcount = 1;
 	pipe_mutex_unlock(nouveau_screen_mutex);
 	return &screen->base;

src/gallium/winsys/radeon/drm/radeon_drm_bo.c

@@ -305,14 +305,34 @@ static void radeon_bo_destroy(struct pb_buffer *_buf)
     if (bo->ptr)
         os_munmap(bo->ptr, bo->base.size);
 
+    if (mgr->va) {
+        if (bo->rws->va_unmap_working) {
+            struct drm_radeon_gem_va va;
+
+            va.handle = bo->handle;
+            va.vm_id = 0;
+            va.operation = RADEON_VA_UNMAP;
+            va.flags = RADEON_VM_PAGE_READABLE |
+                       RADEON_VM_PAGE_WRITEABLE |
+                       RADEON_VM_PAGE_SNOOPED;
+            va.offset = bo->va;
+
+            if (drmCommandWriteRead(bo->rws->fd, DRM_RADEON_GEM_VA, &va,
+				    sizeof(va)) != 0 &&
+		va.operation == RADEON_VA_RESULT_ERROR) {
+                fprintf(stderr, "radeon: Failed to deallocate virtual address for buffer:\n");
+                fprintf(stderr, "radeon:    size      : %d bytes\n", bo->base.size);
+                fprintf(stderr, "radeon:    va        : 0x%016llx\n", (unsigned long long)bo->va);
+            }
+	}
+
+	radeon_bomgr_free_va(mgr, bo->va, bo->base.size);
+    }
+
     /* Close object. */
     args.handle = bo->handle;
     drmIoctl(bo->rws->fd, DRM_IOCTL_GEM_CLOSE, &args);
 
-    if (mgr->va) {
-        radeon_bomgr_free_va(mgr, bo->va, bo->base.size);
-    }
-
     pipe_mutex_destroy(bo->map_mutex);
 
     if (bo->initial_domain & RADEON_DOMAIN_VRAM)

src/gallium/winsys/radeon/drm/radeon_drm_winsys.c

@@ -57,6 +57,8 @@
 #define RADEON_INFO_READ_REG		0x24
 #endif
 
+#define RADEON_INFO_VA_UNMAP_WORKING	0x25
+
 static struct util_hash_table *fd_tab = NULL;
 pipe_static_mutex(fd_tab_mutex);
 
@@ -399,6 +401,8 @@ static boolean do_winsys_init(struct radeon_drm_winsys *ws)
             if (!radeon_get_drm_value(ws->fd, RADEON_INFO_IB_VM_MAX_SIZE, NULL,
                                       &ib_vm_max_size))
                 ws->info.r600_virtual_address = FALSE;
+            radeon_get_drm_value(ws->fd, RADEON_INFO_VA_UNMAP_WORKING, NULL,
+                                 &ws->va_unmap_working);
         }
 	if (ws->gen == DRV_R600 && !debug_get_bool_option("RADEON_VA", FALSE))
 		ws->info.r600_virtual_address = FALSE;
@@ -484,6 +488,10 @@ static void radeon_winsys_destroy(struct radeon_winsys *rws)
     if (ws->gen >= DRV_R600) {
         radeon_surface_manager_free(ws->surf_man);
     }
+
+    if (ws->fd >= 0)
+        close(ws->fd);
+
     FREE(rws);
 }
 
@@ -696,7 +704,7 @@ radeon_drm_winsys_create(int fd, radeon_screen_create_t screen_create)
         return NULL;
     }
 
-    ws->fd = fd;
+    ws->fd = dup(fd);
 
     if (!do_winsys_init(ws))
         goto fail;
@@ -712,7 +720,7 @@ radeon_drm_winsys_create(int fd, radeon_screen_create_t screen_create)
         goto fail;
 
     if (ws->gen >= DRV_R600) {
-        ws->surf_man = radeon_surface_manager_new(fd);
+        ws->surf_man = radeon_surface_manager_new(ws->fd);
         if (!ws->surf_man)
             goto fail;
     }
@@ -753,7 +761,7 @@ radeon_drm_winsys_create(int fd, radeon_screen_create_t screen_create)
         return NULL;
     }
 
-    util_hash_table_set(fd_tab, intptr_to_pointer(fd), ws);
+    util_hash_table_set(fd_tab, intptr_to_pointer(ws->fd), ws);
 
     /* We must unlock the mutex once the winsys is fully initialized, so that
      * other threads attempting to create the winsys from the same fd will
@@ -770,6 +778,9 @@ fail:
         ws->kman->destroy(ws->kman);
     if (ws->surf_man)
         radeon_surface_manager_free(ws->surf_man);
+    if (ws->fd >= 0)
+        close(ws->fd);
+
     FREE(ws);
     return NULL;
 }

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

@@ -74,6 +74,7 @@ struct radeon_drm_winsys {
     enum radeon_generation gen;
     struct radeon_info info;
     uint32_t va_start;
+    uint32_t va_unmap_working;
     uint32_t accel_working2;
 
     struct pb_manager *kman;

src/gbm/Makefile.am

@@ -52,7 +52,8 @@ libgbm_dri_la_CFLAGS = \
 	$(LIBDRM_CFLAGS)
 
 libgbm_la_LIBADD += \
-	libgbm_dri.la $(top_builddir)/src/mapi/shared-glapi/libglapi.la $(LIBDRM_LIBS)
+	libgbm_dri.la \
+	$(LIBDRM_LIBS)
 endif
 
 TESTS = gbm-symbols-check

src/gbm/backends/dri/gbm_dri.c

@@ -311,6 +311,14 @@ dri_open_driver(struct gbm_dri_device *dri)
    if (search_paths == NULL)
       search_paths = DEFAULT_DRIVER_DIR;
 
+   /* Temporarily work around dri driver libs that need symbols in libglapi
+    * but don't automatically link it in.
+    */
+   /* XXX: Library name differs on per platforms basis. Update this as
+    * osx/cygwin/windows/bsd gets support for GBM..
+    */
+   dlopen("libglapi.so.0", RTLD_LAZY | RTLD_GLOBAL);
+
    dri->driver = NULL;
    end = search_paths + strlen(search_paths);
    for (p = search_paths; p < end && dri->driver == NULL; p = next + 1) {

src/glsl/Makefile.am

@@ -89,8 +89,7 @@ tests_general_ir_test_SOURCES =		\
 	tests/builtin_variable_test.cpp			\
 	tests/invalidate_locations_test.cpp		\
 	tests/general_ir_test.cpp			\
-	tests/varyings_test.cpp				\
-	tests/common.c
+	tests/varyings_test.cpp
 tests_general_ir_test_CFLAGS =				\
 	$(PTHREAD_CFLAGS)
 tests_general_ir_test_LDADD =				\
@@ -103,8 +102,7 @@ tests_uniform_initializer_test_SOURCES =		\
 	tests/copy_constant_to_storage_tests.cpp	\
 	tests/set_uniform_initializer_tests.cpp		\
 	tests/uniform_initializer_utils.cpp		\
-	tests/uniform_initializer_utils.h		\
-	tests/common.c
+	tests/uniform_initializer_utils.h
 tests_uniform_initializer_test_CFLAGS =			\
 	$(PTHREAD_CFLAGS)
 tests_uniform_initializer_test_LDADD =			\
@@ -114,8 +112,7 @@ tests_uniform_initializer_test_LDADD =			\
 	$(PTHREAD_LIBS)
 
 tests_sampler_types_test_SOURCES =			\
-	tests/sampler_types_test.cpp			\
-	tests/common.c
+	tests/sampler_types_test.cpp
 tests_sampler_types_test_CFLAGS =			\
 	$(PTHREAD_CFLAGS)
 tests_sampler_types_test_LDADD =			\
@@ -133,8 +130,7 @@ libglcpp_la_SOURCES =					\
 	$(LIBGLCPP_FILES)
 
 glcpp_glcpp_SOURCES =					\
-	glcpp/glcpp.c					\
-	tests/common.c
+	glcpp/glcpp.c
 glcpp_glcpp_LDADD =					\
 	libglcpp.la					\
 	$(top_builddir)/src/libglsl_util.la		\
@@ -164,7 +160,6 @@ glsl_compiler_LDADD =					\
 
 glsl_test_SOURCES = \
 	standalone_scaffolding.cpp \
-	tests/common.c \
 	test.cpp \
 	test_optpass.cpp \
 	test_optpass.h

src/glsl/SConscript

@@ -71,6 +71,7 @@ env.Command('imports.c', '#src/mesa/main/imports.c', Copy('$TARGET', '$SOURCE'))
 env.Prepend(CPPPATH = ['#src/mesa/program'])
 env.Command('prog_hash_table.c', '#src/mesa/program/prog_hash_table.c', Copy('$TARGET', '$SOURCE'))
 env.Command('symbol_table.c', '#src/mesa/program/symbol_table.c', Copy('$TARGET', '$SOURCE'))
+env.Command('dummy_errors.c', '#src/mesa/program/dummy_errors.c', Copy('$TARGET', '$SOURCE'))
 
 compiler_objs = env.StaticObject(source_lists['GLSL_COMPILER_CXX_FILES'])
 
@@ -78,6 +79,7 @@ mesa_objs = env.StaticObject([
     'imports.c',
     'prog_hash_table.c',
     'symbol_table.c',
+    'dummy_errors.c',
 ])
 
 compiler_objs += mesa_objs
@@ -115,6 +117,6 @@ env.Alias('glsl_compiler', glsl_compiler)
 
 glcpp = env.Program(
     target = 'glcpp/glcpp',
-    source = ['glcpp/glcpp.c', 'tests/common.c'] + mesa_objs,
+    source = ['glcpp/glcpp.c'] + mesa_objs,
 )
 env.Alias('glcpp', glcpp)

src/glsl/ast_array_index.cpp

@@ -225,25 +225,25 @@ _mesa_ast_array_index_to_hir(void *mem_ctx,
        * values *do* diverge, then the behavior of the operation requiring a
        * dynamically uniform expression is undefined.
        */
-      if (array->type->element_type()->is_sampler()) {
-	 if (!state->is_version(130, 100)) {
-	    if (state->es_shader) {
-	       _mesa_glsl_warning(&loc, state,
-				  "sampler arrays indexed with non-constant "
-				  "expressions is optional in %s",
-				  state->get_version_string());
-	    } else {
-	       _mesa_glsl_warning(&loc, state,
-				  "sampler arrays indexed with non-constant "
-				  "expressions will be forbidden in GLSL 1.30 "
-				  "and later");
-	    }
-	 } else if (!state->is_version(400, 0) && !state->ARB_gpu_shader5_enable) {
-	    _mesa_glsl_error(&loc, state,
-			     "sampler arrays indexed with non-constant "
-			     "expressions is forbidden in GLSL 1.30 and "
-			     "later");
-	 }
+      if (array->type->without_array()->is_sampler()) {
+         if (!state->is_version(400, 0) && !state->ARB_gpu_shader5_enable) {
+            if (state->is_version(130, 300))
+               _mesa_glsl_error(&loc, state,
+                                "sampler arrays indexed with non-constant "
+                                "expressions are forbidden in GLSL %s "
+                                "and later",
+                                state->es_shader ? "ES 3.00" : "1.30");
+            else if (state->es_shader)
+               _mesa_glsl_warning(&loc, state,
+                                  "sampler arrays indexed with non-constant "
+                                  "expressions will be forbidden in GLSL "
+                                  "3.00 and later");
+            else
+               _mesa_glsl_warning(&loc, state,
+                                  "sampler arrays indexed with non-constant "
+                                  "expressions will be forbidden in GLSL "
+                                  "1.30 and later");
+         }
       }
    }
 

src/glsl/ast_to_hir.cpp

@@ -2086,7 +2086,7 @@ validate_binding_qualifier(struct _mesa_glsl_parse_state *state,
        *  with an array of size N, all elements of the array from binding
        *  through binding + N - 1 must be within this range."
        */
-      unsigned limit = ctx->Const.Program[state->stage].MaxTextureImageUnits;
+      unsigned limit = ctx->Const.MaxCombinedTextureImageUnits;
 
       if (max_index >= limit) {
          _mesa_glsl_error(loc, state, "layout(binding = %d) for %d samplers "

src/glsl/builtin_variables.cpp

@@ -876,9 +876,9 @@ void
 builtin_variable_generator::generate_gs_special_vars()
 {
    add_output(VARYING_SLOT_LAYER, int_t, "gl_Layer");
-   if (state->ARB_viewport_array_enable)
+   if (state->is_version(410, 0) || state->ARB_viewport_array_enable)
       add_output(VARYING_SLOT_VIEWPORT, int_t, "gl_ViewportIndex");
-   if (state->ARB_gpu_shader5_enable)
+   if (state->is_version(400, 0) || state->ARB_gpu_shader5_enable)
       add_system_value(SYSTEM_VALUE_INVOCATION_ID, int_t, "gl_InvocationID");
 
    /* Although gl_PrimitiveID appears in tessellation control and tessellation
@@ -946,7 +946,7 @@ builtin_variable_generator::generate_fs_special_vars()
          var->enable_extension_warning("GL_AMD_shader_stencil_export");
    }
 
-   if (state->ARB_sample_shading_enable) {
+   if (state->is_version(400, 0) || state->ARB_sample_shading_enable) {
       add_system_value(SYSTEM_VALUE_SAMPLE_ID, int_t, "gl_SampleID");
       add_system_value(SYSTEM_VALUE_SAMPLE_POS, vec2_t, "gl_SamplePosition");
       /* From the ARB_sample_shading specification:
@@ -959,11 +959,11 @@ builtin_variable_generator::generate_fs_special_vars()
       add_output(FRAG_RESULT_SAMPLE_MASK, array(int_t, 1), "gl_SampleMask");
    }
 
-   if (state->ARB_gpu_shader5_enable) {
+   if (state->is_version(400, 0) || state->ARB_gpu_shader5_enable) {
       add_system_value(SYSTEM_VALUE_SAMPLE_MASK_IN, array(int_t, 1), "gl_SampleMaskIn");
    }
 
-   if (state->ARB_fragment_layer_viewport_enable) {
+   if (state->is_version(430, 0) || state->ARB_fragment_layer_viewport_enable) {
       add_input(VARYING_SLOT_LAYER, int_t, "gl_Layer");
       add_input(VARYING_SLOT_VIEWPORT, int_t, "gl_ViewportIndex");
    }

src/glsl/glcpp/glcpp-parse.y

@@ -1074,9 +1074,9 @@ _token_list_equal_ignoring_space (token_list_t *a, token_list_t *b)
 		 */
 		if (node_a->token->type == SPACE
 		    && node_b->token->type == SPACE) {
-			while (node_a->token->type == SPACE)
+			while (node_a && node_a->token->type == SPACE)
 				node_a = node_a->next;
-			while (node_b->token->type == SPACE)
+			while (node_b && node_b->token->type == SPACE)
 				node_b = node_b->next;
 			continue;
 		}

src/glsl/ir.cpp

@@ -63,8 +63,6 @@ update_rhs_swizzle(ir_swizzle_mask &m, unsigned from, unsigned to)
    case 3: m.w = from; break;
    default: assert(!"Should not get here.");
    }
-
-   m.num_components = MAX2(m.num_components, (to + 1));
 }
 
 void
@@ -95,6 +93,7 @@ ir_assignment::set_lhs(ir_rvalue *lhs)
 
 	 write_mask |= (((this->write_mask >> i) & 1) << c);
 	 update_rhs_swizzle(rhs_swiz, i, c);
+         rhs_swiz.num_components = swiz->val->type->vector_elements;
       }
 
       this->write_mask = write_mask;
@@ -114,6 +113,7 @@ ir_assignment::set_lhs(ir_rvalue *lhs)
 	 if (write_mask & (1 << i))
 	    update_rhs_swizzle(rhs_swiz, i, rhs_chan++);
       }
+      rhs_swiz.num_components = rhs_chan;
       this->rhs = new(mem_ctx) ir_swizzle(this->rhs, rhs_swiz);
    }
 

src/glsl/ir_function.cpp

@@ -148,9 +148,11 @@ get_parameter_match_type(const ir_variable *param,
    if (from_type == to_type)
       return PARAMETER_EXACT_MATCH;
 
-   /* XXX: When ARB_gpu_shader_fp64 support is added, check for float->double,
-    * and int/uint->double conversions
-    */
+   if (to_type->base_type == GLSL_TYPE_DOUBLE) {
+      if (from_type->base_type == GLSL_TYPE_FLOAT)
+         return PARAMETER_FLOAT_TO_DOUBLE;
+      return PARAMETER_INT_TO_DOUBLE;
+   }
 
    if (to_type->base_type == GLSL_TYPE_FLOAT)
       return PARAMETER_INT_TO_FLOAT;

src/glsl/linker.cpp

@@ -346,6 +346,39 @@ private:
    bool uses_non_zero_stream;
 };
 
+/* Class that finds array derefs and check if indexes are dynamic. */
+class dynamic_sampler_array_indexing_visitor : public ir_hierarchical_visitor
+{
+public:
+   dynamic_sampler_array_indexing_visitor() :
+      dynamic_sampler_array_indexing(false)
+   {
+   }
+
+   ir_visitor_status visit_enter(ir_dereference_array *ir)
+   {
+      if (!ir->variable_referenced())
+         return visit_continue;
+
+      if (!ir->variable_referenced()->type->contains_sampler())
+         return visit_continue;
+
+      if (!ir->array_index->constant_expression_value()) {
+         dynamic_sampler_array_indexing = true;
+         return visit_stop;
+      }
+      return visit_continue;
+   }
+
+   bool uses_dynamic_sampler_array_indexing()
+   {
+      return dynamic_sampler_array_indexing;
+   }
+
+private:
+   bool dynamic_sampler_array_indexing;
+};
+
 } /* anonymous namespace */
 
 void
@@ -2736,6 +2769,40 @@ build_program_resource_list(struct gl_context *ctx,
     */
 }
 
+/**
+ * This check is done to make sure we allow only constant expression
+ * indexing and "constant-index-expression" (indexing with an expression
+ * that includes loop induction variable).
+ */
+static bool
+validate_sampler_array_indexing(struct gl_context *ctx,
+                                struct gl_shader_program *prog)
+{
+   dynamic_sampler_array_indexing_visitor v;
+   for (unsigned i = 0; i < MESA_SHADER_STAGES; i++) {
+      if (prog->_LinkedShaders[i] == NULL)
+	 continue;
+
+      bool no_dynamic_indexing =
+         ctx->Const.ShaderCompilerOptions[i].EmitNoIndirectSampler;
+
+      /* Search for array derefs in shader. */
+      v.run(prog->_LinkedShaders[i]->ir);
+      if (v.uses_dynamic_sampler_array_indexing()) {
+         const char *msg = "sampler arrays indexed with non-constant "
+                           "expressions is forbidden in GLSL %s %u";
+         /* Backend has indicated that it has no dynamic indexing support. */
+         if (no_dynamic_indexing) {
+            linker_error(prog, msg, prog->IsES ? "ES" : "", prog->Version);
+            return false;
+         } else {
+            linker_warning(prog, msg, prog->IsES ? "ES" : "", prog->Version);
+         }
+      }
+   }
+   return true;
+}
+
 
 void
 link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
@@ -2829,8 +2896,11 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
             link_intrastage_shaders(mem_ctx, ctx, prog, shader_list[stage],
                                     num_shaders[stage]);
 
-         if (!prog->LinkStatus)
+         if (!prog->LinkStatus) {
+            if (sh)
+               ctx->Driver.DeleteShader(ctx, sh);
             goto done;
+         }
 
          switch (stage) {
          case MESA_SHADER_VERTEX:
@@ -2843,8 +2913,11 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
             validate_fragment_shader_executable(prog, sh);
             break;
          }
-         if (!prog->LinkStatus)
+         if (!prog->LinkStatus) {
+            if (sh)
+               ctx->Driver.DeleteShader(ctx, sh);
             goto done;
+         }
 
          _mesa_reference_shader(ctx, &prog->_LinkedShaders[stage], sh);
       }
@@ -2948,6 +3021,16 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
       lower_const_arrays_to_uniforms(prog->_LinkedShaders[i]->ir);
    }
 
+   /* Validation for special cases where we allow sampler array indexing
+    * with loop induction variable. This check emits a warning or error
+    * depending if backend can handle dynamic indexing.
+    */
+   if ((!prog->IsES && prog->Version < 130) ||
+       (prog->IsES && prog->Version < 300)) {
+      if (!validate_sampler_array_indexing(ctx, prog))
+         goto done;
+   }
+
    /* Check and validate stream emissions in geometry shaders */
    validate_geometry_shader_emissions(ctx, prog);
 

src/glsl/loop_unroll.cpp

@@ -100,6 +100,18 @@ public:
 
    virtual ir_visitor_status visit_enter(ir_dereference_array *ir)
    {
+      /* Force unroll in case of dynamic indexing with sampler arrays
+       * when EmitNoIndirectSampler is set.
+       */
+      if (options->EmitNoIndirectSampler) {
+         if ((ir->array->type->is_array() &&
+              ir->array->type->contains_sampler()) &&
+             !ir->array_index->constant_expression_value()) {
+            unsupported_variable_indexing = true;
+            return visit_continue;
+         }
+      }
+
       /* Check for arrays variably-indexed by a loop induction variable.
        * Unrolling the loop may convert that access into constant-indexing.
        *

src/glsl/main.cpp

@@ -41,12 +41,6 @@
 
 static int glsl_version = 330;
 
-extern "C" void
-_mesa_error_no_memory(const char *caller)
-{
-   fprintf(stderr, "Mesa error: out of memory in %s", caller);
-}
-
 static void
 initialize_context(struct gl_context *ctx, gl_api api)
 {

src/glsl/opt_algebraic.cpp

@@ -580,7 +580,7 @@ ir_algebraic_visitor::handle_expression(ir_expression *ir)
             continue;
 
          ir_expression *add_expr = floor_expr->operands[0]->as_expression();
-         if (!add_expr)
+         if (!add_expr || add_expr->operation != ir_binop_add)
             continue;
 
          for (int j = 0; j < 2; j++) {
@@ -589,7 +589,7 @@ ir_algebraic_visitor::handle_expression(ir_expression *ir)
                continue;
 
             ir_constant *point_five = add_expr->operands[1 - j]->as_constant();
-            if (!point_five->is_value(0.5, 0))
+            if (!point_five || !point_five->is_value(0.5, 0))
                continue;
 
             if (abs_expr->operands[0]->equals(sign_expr->operands[0])) {

src/loader/Makefile.am

@@ -41,15 +41,11 @@ libloader_la_CPPFLAGS += \
 	-I$(top_builddir)/src/mesa/drivers/dri/common/ \
 	-I$(top_srcdir)/src/mesa/ \
 	-I$(top_srcdir)/src/mapi/ \
-	-DUSE_DRICONF \
-	$(EXPAT_CFLAGS)
+	-DUSE_DRICONF
 
-libloader_la_SOURCES += \
-	$(top_srcdir)/src/mesa/drivers/dri/common/xmlconfig.c
+ libloader_la_LIBADD += \
+	$(top_builddir)/src/mesa/drivers/dri/common/libxmlconfig.la
 
-libloader_la_LIBADD += \
-	-lm \
-	$(EXPAT_LIBS)
 endif
 
 if !HAVE_LIBDRM

src/loader/loader.c

@@ -112,26 +112,36 @@ static void *udev_handle = NULL;
 static void *
 udev_dlopen_handle(void)
 {
-   if (!udev_handle) {
-      udev_handle = dlopen("libudev.so.1", RTLD_LOCAL | RTLD_LAZY);
+   char name[80];
+   unsigned flags = RTLD_NOLOAD | RTLD_LOCAL | RTLD_LAZY;
+   int version;
 
-      if (!udev_handle) {
-         /* libudev.so.1 changed the return types of the two unref functions
-          * from voids to pointers.  We don't use those return values, and the
-          * only ABI I've heard that cares about this kind of change (calling
-          * a function with a void * return that actually only returns void)
-          * might be ia64.
-          */
-         udev_handle = dlopen("libudev.so.0", RTLD_LOCAL | RTLD_LAZY);
+   /* libudev.so.1 changed the return types of the two unref functions
+    * from voids to pointers.  We don't use those return values, and the
+    * only ABI I've heard that cares about this kind of change (calling
+    * a function with a void * return that actually only returns void)
+    * might be ia64.
+    */
 
-         if (!udev_handle) {
-            log_(_LOADER_WARNING, "Couldn't dlopen libudev.so.1 or "
-                 "libudev.so.0, driver detection may be broken.\n");
-         }
+   /* First try opening an already linked libudev, then try loading one */
+   do {
+      for (version = 1; version >= 0; version--) {
+         snprintf(name, sizeof(name), "libudev.so.%d", version);
+         udev_handle = dlopen(name, flags);
+         if (udev_handle)
+            return udev_handle;
       }
-   }
 
-   return udev_handle;
+      if ((flags & RTLD_NOLOAD) == 0)
+         break;
+
+      flags &= ~RTLD_NOLOAD;
+   } while (1);
+
+   log_(_LOADER_WARNING,
+        "Couldn't dlopen libudev.so.1 or "
+        "libudev.so.0, driver detection may be broken.\n");
+   return NULL;
 }
 
 static int dlsym_failed = 0;