glsl: make sure UBO arrays are sized in ES

This check was removed in 5b2675093e add it back in.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Cc: "12.0" <mesa-stable@lists.freedesktop.org>
https://bugs.freedesktop.org/show_bug.cgi?id=96349

.travis.yml

@@ -1,7 +1,6 @@
 language: c
 
-sudo: true
-dist: trusty
+sudo: false
 
 cache:
   directories:
@@ -16,11 +15,7 @@ addons:
       - libexpat1-dev
       - libxcb-dri2-0-dev
       - libx11-xcb-dev
-      - llvm-3.5-dev
-      # llvm-config is not in the dev package?
-      - llvm-3.5
-      # LLVM packaging is broken and misses this dep.
-      - libedit-dev
+      - llvm-3.4-dev
       - scons
 
 env:
@@ -46,16 +41,6 @@ install:
   - export PATH="/usr/lib/ccache:$PATH"
   - pip install --user mako
 
-  # Since libdrm gets updated in configure.ac regularly, try to pick up the
-  # latest version from there.
-  - for line in `grep "^LIBDRM_.*_REQUIRED=" configure.ac`; do
-      old_ver=`echo $LIBDRM_VERSION | sed 's/libdrm-//'`;
-      new_ver=`echo $line | sed 's/.*REQUIRED=//'`;
-      if `echo "$old_ver,$new_ver" | tr ',' '\n' | sort -Vc 2> /dev/null`; then
-        export LIBDRM_VERSION="libdrm-$new_ver";
-      fi;
-    done
-
   # Install dependencies where we require specific versions (or where
   # disallowed by Travis CI's package whitelisting).
 
@@ -93,19 +78,22 @@ install:
 
   - wget http://dri.freedesktop.org/libdrm/$LIBDRM_VERSION.tar.bz2
   - tar -jxvf $LIBDRM_VERSION.tar.bz2
-  - (cd $LIBDRM_VERSION && ./configure --prefix=$HOME/prefix --enable-vc4 && make install)
+  - (cd $LIBDRM_VERSION && ./configure --prefix=$HOME/prefix && make install)
 
   - wget $XORG_RELEASES/lib/$LIBXSHMFENCE_VERSION.tar.bz2
   - tar -jxvf $LIBXSHMFENCE_VERSION.tar.bz2
   - (cd $LIBXSHMFENCE_VERSION && ./configure --prefix=$HOME/prefix && make install)
 
+# Disabled LLVM (and therefore r300 and r600) because the build fails
+# with "undefined reference to `clock_gettime'" and "undefined
+# reference to `setupterm'" in llvmpipe.
 script:
   - if test "x$BUILD" = xmake; then
       ./autogen.sh --enable-debug
+        --disable-gallium-llvm
         --with-egl-platforms=x11,drm
         --with-dri-drivers=i915,i965,radeon,r200,swrast,nouveau
-        --with-gallium-drivers=svga,swrast,vc4,virgl,r300,r600
-        --disable-llvm-shared-libs
+        --with-gallium-drivers=svga,swrast,vc4,virgl
         ;
       make && make check;
     elif test x$BUILD = xscons; then

Android.common.mk

@@ -34,6 +34,10 @@ MESA_VERSION := $(shell cat $(MESA_TOP)/VERSION)
 LOCAL_CFLAGS += \
 	-Wno-unused-parameter \
 	-Wno-date-time \
+	-Wno-pointer-arith \
+	-Wno-missing-field-initializers \
+	-Wno-initializer-overrides \
+	-Wno-mismatched-tags \
 	-DPACKAGE_VERSION=\"$(MESA_VERSION)\" \
 	-DPACKAGE_BUGREPORT=\"https://bugs.freedesktop.org/enter_bug.cgi?product=Mesa\" \
 	-DANDROID_VERSION=0x0$(MESA_ANDROID_MAJOR_VERSION)0$(MESA_ANDROID_MINOR_VERSION)
@@ -78,6 +82,12 @@ LOCAL_CFLAGS += \
 	-D__STDC_LIMIT_MACROS
 endif
 
+# add libdrm if there are hardware drivers
+ifneq ($(filter-out swrast,$(MESA_GPU_DRIVERS)),)
+LOCAL_CFLAGS += -DHAVE_LIBDRM
+LOCAL_SHARED_LIBRARIES += libdrm
+endif
+
 LOCAL_CPPFLAGS += \
 	$(if $(filter true,$(MESA_LOLLIPOP_BUILD)),-D_USING_LIBCXX) \
 	-Wno-error=non-virtual-dtor \

Makefile.am

@@ -40,7 +40,7 @@ AM_DISTCHECK_CONFIGURE_FLAGS = \
 	--enable-vdpau \
 	--enable-xa \
 	--enable-xvmc \
-	--enable-llvm-shared-libs \
+	--disable-llvm-shared-libs \
 	--with-egl-platforms=x11,wayland,drm,surfaceless \
 	--with-dri-drivers=i915,i965,nouveau,radeon,r200,swrast \
 	--with-gallium-drivers=i915,ilo,nouveau,r300,r600,radeonsi,freedreno,svga,swrast,vc4,virgl,swr \
@@ -62,7 +62,6 @@ noinst_HEADERS = \
 	include/c99_math.h \
 	include/c11 \
 	include/D3D9 \
-	include/GL/wglext.h \
 	include/HaikuGL \
 	include/no_extern_c.h \
 	include/pci_ids

VERSION

@@ -1 +1 @@
-12.0.6
+12.1.0-devel

appveyor.yml

@@ -37,8 +37,6 @@ cache:
 - win_flex_bison-2.4.5.zip
 - llvm-3.3.1-msvc2013-mtd.7z
 
-os: Visual Studio 2013
-
 environment:
   WINFLEXBISON_ARCHIVE: win_flex_bison-2.4.5.zip
   LLVM_ARCHIVE: llvm-3.3.1-msvc2013-mtd.7z
@@ -49,13 +47,11 @@ install:
 - python -m pip --version
 # Install Mako
 - python -m pip install --egg Mako
-# Install pywin32 extensions, needed by SCons
-- python -m pip install pypiwin32
 # Install SCons
 - python -m pip install --egg scons==2.4.1
 - scons --version
 # Install flex/bison
-- if not exist "%WINFLEXBISON_ARCHIVE%" appveyor DownloadFile "https://downloads.sourceforge.net/project/winflexbison/old_versions/%WINFLEXBISON_ARCHIVE%"
+- if not exist "%WINFLEXBISON_ARCHIVE%" appveyor DownloadFile "http://downloads.sourceforge.net/project/winflexbison/%WINFLEXBISON_ARCHIVE%"
 - 7z x -y -owinflexbison\ "%WINFLEXBISON_ARCHIVE%" > nul
 - set Path=%CD%\winflexbison;%Path%
 - win_flex --version

bin/.cherry-ignore

@@ -1,28 +0,0 @@
-# The offending commit that this patch (part) reverts isn't in 12.0
-be32a2132785fbc119f17e62070e007ee7d17af7 i965/compiler: Bring back the INTEL_PRECISE_TRIG environment variable
-
-# The patch depends on the batch_cache work at least.
-89f00f749fda4c1beca38f362c7f86bdc6e32785 a4xx: make sure to actually clamp depth as requested
-
-# The patch depends on the 'generic' interoplation and location
-# implementation introduced with 2d6dd30a9b30
-114874b22beafb2d07006b197c62d717fc7f80cc i965/fs: Use sample interpolation for interpolateAtCentroid in persample mode
-
-# VAAPI encode landed after the branch point.
-a5993022275c20061ac025d9adc26c5f9d02afee st/va Avoid VBR bitrate calculation overflow v2
-
-# EGL_KHR_debug landed after the branch point.
-17084b6f9340f798111e53e08f5d35c7630cee48 egl: Fix missing unlock in eglGetSyncAttribKHR
-
-# Depends on update_renderbuffer_read_surfaces at least
-f2b9b0c730e345bcffa9eadabb25af3ab02642f2 i965: Add missing BRW_NEW_FS_PROG_DATA to render target reads.
-
-# The commit in question hasn't landed in branch
-1ef787339774bc7f1cc9c1615722f944005e070c Revert "egl/android: Set EGL_MAX_PBUFFER_WIDTH and EGL_MAX_PBUFFER_HEIGHT"
-
-# Patches depend on the fence_finish() gallium API change and corresponding driver work
-f240ad98bc05281ea7013d91973cb5f932ae9434 st/mesa: unduplicate st_check_sync code
-b687f766fddb7b39479cd9ee0427984029ea3559 st/mesa: allow multiple concurrent waiters in ClientWaitSync
-
-# Commit was reverted shortly after it landed in master
-a39ad185932eab4f25a0cb2b112c10d8700ef242 configure.ac: honour LLVM_LIBDIR when linking against LLVM

bin/bugzilla_mesa.sh

@@ -40,7 +40,7 @@ else
 	for i in $urls
 	do
 		id=$(echo $i | cut -d'=' -f2)
-		summary=$(wget --quiet -O - $i | grep -e '<title>.*</title>' | sed -e 's/ *<title>[0-9]\+ &ndash; \(.*\)<\/title>/\1/')
+		summary=$(wget --quiet -O - $i | grep -e '<title>.*</title>' | sed -e 's/ *<title>Bug [0-9]\+ &ndash; \(.*\)<\/title>/\1/')
 		echo "<li><a href=\"$i\">Bug $id</a> - $summary</li>"
 		echo ""
 	done

bin/get-pick-list.sh

@@ -14,7 +14,7 @@ git log --reverse --grep="cherry picked from commit" origin/master..HEAD |\
 	sed -e 's/^[[:space:]]*(cherry picked from commit[[:space:]]*//' -e 's/)//' > already_picked
 
 # Grep for commits that were marked as a candidate for the stable tree.
-git log --reverse --pretty=%H -i --grep='^\([[:space:]]*NOTE: .*[Cc]andidate\|CC:.*12\.0.*mesa-stable\)' HEAD..origin/master |\
+git log --reverse --pretty=%H -i --grep='^\([[:space:]]*NOTE: .*[Cc]andidate\|CC:.*mesa-stable\)' HEAD..origin/master |\
 while read sha
 do
 	# Check to see whether the patch is on the ignore list.

bin/get-typod-pick-list.sh

@@ -1,39 +0,0 @@
-#!/bin/sh
-
-# Script for generating a list of candidates which have typos in the nomination line
-#
-# Usage examples:
-#
-# $ bin/get-typod-pick-list.sh
-# $ bin/get-typod-pick-list.sh > picklist
-# $ bin/get-typod-pick-list.sh | tee picklist
-
-# NB:
-# This script intentionally _never_ checks for specific version tag
-# Should we consider folding it with the original get-pick-list.sh
-
-# Grep for commits with "cherry picked from commit" in the commit message.
-git log --reverse --grep="cherry picked from commit" origin/master..HEAD |\
-	grep "cherry picked from commit" |\
-	sed -e 's/^[[:space:]]*(cherry picked from commit[[:space:]]*//' -e 's/)//' > already_picked
-
-# Grep for commits that were marked as a candidate for the stable tree.
-git log --reverse --pretty=%H -i --grep='^CC:.*mesa-dev' HEAD..origin/master |\
-while read sha
-do
-	# Check to see whether the patch is on the ignore list.
-	if [ -f bin/.cherry-ignore ] ; then
-		if grep -q ^$sha bin/.cherry-ignore ; then
-			continue
-		fi
-	fi
-
-	# Check to see if it has already been picked over.
-	if grep -q ^$sha already_picked ; then
-		continue
-	fi
-
-	git log -n1 --pretty=oneline $sha | cat
-done
-
-rm -f already_picked

configure.ac

@@ -225,7 +225,6 @@ AX_GCC_FUNC_ATTRIBUTE([packed])
 AX_GCC_FUNC_ATTRIBUTE([pure])
 AX_GCC_FUNC_ATTRIBUTE([returns_nonnull])
 AX_GCC_FUNC_ATTRIBUTE([unused])
-AX_GCC_FUNC_ATTRIBUTE([visibility])
 AX_GCC_FUNC_ATTRIBUTE([warn_unused_result])
 AX_GCC_FUNC_ATTRIBUTE([weak])
 
@@ -784,7 +783,6 @@ if test "x$enable_asm" = xyes; then
     esac
 fi
 
-AC_HEADER_MAJOR
 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"])
@@ -1062,7 +1060,6 @@ xno)
     ;;
 esac
 
-AM_CONDITIONAL(HAVE_GLX, test "x$enable_glx" != xno)
 AM_CONDITIONAL(HAVE_DRI_GLX, test "x$enable_glx" = xdri)
 AM_CONDITIONAL(HAVE_XLIB_GLX, test "x$enable_glx" = xxlib)
 AM_CONDITIONAL(HAVE_GALLIUM_XLIB_GLX, test "x$enable_glx" = xgallium-xlib)
@@ -1641,9 +1638,9 @@ esac
 
 AC_ARG_WITH([vulkan-icddir],
     [AS_HELP_STRING([--with-vulkan-icddir=DIR],
-        [directory for the Vulkan driver icd files @<:@${datarootdir}/vulkan/icd.d@:>@])],
+        [directory for the Vulkan driver icd files @<:@${sysconfdir}/vulkan/icd.d@:>@])],
     [VULKAN_ICD_INSTALL_DIR="$withval"],
-    [VULKAN_ICD_INSTALL_DIR='${datarootdir}/vulkan/icd.d'])
+    [VULKAN_ICD_INSTALL_DIR='${sysconfdir}/vulkan/icd.d'])
 AC_SUBST([VULKAN_ICD_INSTALL_DIR])
 
 if test -n "$with_vulkan_drivers"; then
@@ -1999,8 +1996,8 @@ if test "x$with_egl_platforms" != "x" -a "x$enable_egl" != xyes; then
     AC_MSG_ERROR([cannot build egl state tracker without EGL library])
 fi
 
-PKG_CHECK_MODULES([WAYLAND_SCANNER], [wayland-scanner],
-        WAYLAND_SCANNER=`$PKG_CONFIG --variable=wayland_scanner wayland-scanner`,
+PKG_CHECK_MODULES([WAYLAND_SCANNER], [wayland_scanner],
+        WAYLAND_SCANNER=`$PKG_CONFIG --variable=wayland_scanner wayland_scanner`,
         WAYLAND_SCANNER='')
 if test "x$WAYLAND_SCANNER" = x; then
     AC_PATH_PROG([WAYLAND_SCANNER], [wayland-scanner])
@@ -2184,10 +2181,6 @@ if test "x$enable_gallium_llvm" = xyes; then
 
         LLVM_COMPONENTS="engine bitwriter mcjit mcdisassembler"
 
-        if $LLVM_CONFIG --components | grep -q inteljitevents ; then
-            LLVM_COMPONENTS="${LLVM_COMPONENTS} inteljitevents"
-        fi
-
         if test "x$enable_opencl" = xyes; then
             llvm_check_version_for "3" "5" "0" "opencl"
 
@@ -2337,45 +2330,6 @@ swr_llvm_check() {
     fi
 }
 
-swr_require_cxx_feature_flags() {
-    feature_name="$1"
-    preprocessor_test="$2"
-    option_list="$3"
-    output_var="$4"
-
-    AC_MSG_CHECKING([whether $CXX supports $feature_name])
-    AC_LANG_PUSH([C++])
-    save_CXXFLAGS="$CXXFLAGS"
-    save_IFS="$IFS"
-    IFS=","
-    found=0
-    for opts in $option_list
-    do
-        unset IFS
-        CXXFLAGS="$opts $save_CXXFLAGS"
-        AC_COMPILE_IFELSE(
-            [AC_LANG_PROGRAM(
-                [   #if !($preprocessor_test)
-                    #error
-                    #endif
-                ])],
-            [found=1; break],
-            [])
-        IFS=","
-    done
-    IFS="$save_IFS"
-    CXXFLAGS="$save_CXXFLAGS"
-    AC_LANG_POP([C++])
-    if test $found -eq 1; then
-        AC_MSG_RESULT([$opts])
-        eval "$output_var=\$opts"
-        return 0
-    fi
-    AC_MSG_RESULT([no])
-    AC_MSG_ERROR([swr requires $feature_name support])
-    return 1
-}
-
 dnl Duplicates in GALLIUM_DRIVERS_DIRS are removed by sorting it after this block
 if test -n "$with_gallium_drivers"; then
     gallium_drivers=`IFS=', '; echo $with_gallium_drivers`
@@ -2445,20 +2399,29 @@ if test -n "$with_gallium_drivers"; then
         xswr)
             swr_llvm_check "swr"
 
-            swr_require_cxx_feature_flags "C++11" "__cplusplus >= 201103L" \
-                ",-std=c++11" \
-                SWR_CXX11_CXXFLAGS
-            AC_SUBST([SWR_CXX11_CXXFLAGS])
+            AC_MSG_CHECKING([whether $CXX supports c++11/AVX/AVX2])
+            AVX_CXXFLAGS="-march=core-avx-i"
+            AVX2_CXXFLAGS="-march=core-avx2"
 
-            swr_require_cxx_feature_flags "AVX" "defined(__AVX__)" \
-                ",-mavx,-march=core-avx" \
-                SWR_AVX_CXXFLAGS
-            AC_SUBST([SWR_AVX_CXXFLAGS])
+            AC_LANG_PUSH([C++])
+            save_CXXFLAGS="$CXXFLAGS"
+            CXXFLAGS="-std=c++11 $CXXFLAGS"
+            AC_COMPILE_IFELSE([AC_LANG_PROGRAM()],[],
+                              [AC_MSG_ERROR([c++11 compiler support not detected])])
+            CXXFLAGS="$save_CXXFLAGS"
 
-            swr_require_cxx_feature_flags "AVX2" "defined(__AVX2__)" \
-                ",-mavx2 -mfma -mbmi2 -mf16c,-march=core-avx2" \
-                SWR_AVX2_CXXFLAGS
-            AC_SUBST([SWR_AVX2_CXXFLAGS])
+            save_CXXFLAGS="$CXXFLAGS"
+            CXXFLAGS="$AVX_CXXFLAGS $CXXFLAGS"
+            AC_COMPILE_IFELSE([AC_LANG_PROGRAM()],[],
+                              [AC_MSG_ERROR([AVX compiler support not detected])])
+            CXXFLAGS="$save_CXXFLAGS"
+
+            save_CFLAGS="$CXXFLAGS"
+            CXXFLAGS="$AVX2_CXXFLAGS $CXXFLAGS"
+            AC_COMPILE_IFELSE([AC_LANG_PROGRAM()],[],
+                              [AC_MSG_ERROR([AVX2 compiler support not detected])])
+            CXXFLAGS="$save_CXXFLAGS"
+            AC_LANG_POP([C++])
 
             HAVE_GALLIUM_SWR=yes
             ;;
@@ -2596,8 +2559,6 @@ fi
 AM_CONDITIONAL(HAVE_LIBDRM, test "x$have_libdrm" = xyes)
 AM_CONDITIONAL(HAVE_OSMESA, test "x$enable_osmesa" = xyes)
 AM_CONDITIONAL(HAVE_GALLIUM_OSMESA, test "x$enable_gallium_osmesa" = xyes)
-AM_CONDITIONAL(HAVE_COMMON_OSMESA, test "x$enable_osmesa" = xyes -o \
-                                        "x$enable_gallium_osmesa" = xyes)
 
 AM_CONDITIONAL(HAVE_X86_ASM, test "x$asm_arch" = xx86 -o "x$asm_arch" = xx86_64)
 AM_CONDITIONAL(HAVE_X86_64_ASM, test "x$asm_arch" = xx86_64)
@@ -2629,8 +2590,6 @@ AC_SUBST([XA_MINOR], $XA_MINOR)
 AC_SUBST([XA_TINY], $XA_TINY)
 AC_SUBST([XA_VERSION], "$XA_MAJOR.$XA_MINOR.$XA_TINY")
 
-AC_SUBST([TIMESTAMP_CMD], '`test $(SOURCE_DATE_EPOCH) && echo $(SOURCE_DATE_EPOCH) || date +%s`')
-
 AC_ARG_ENABLE(valgrind,
               [AS_HELP_STRING([--enable-valgrind],
                              [Build mesa with valgrind support (default: auto)])],

docs/GL3.txt

@@ -146,45 +146,45 @@ GL 4.1, GLSL 4.10 --- all DONE: nvc0, r600, radeonsi
   GL_ARB_viewport_array                                 DONE (i965, nv50, llvmpipe, softpipe)
 
 
-GL 4.2, GLSL 4.20 -- all DONE: radeonsi
+GL 4.2, GLSL 4.20 -- all DONE: nvc0, radeonsi
 
-  GL_ARB_texture_compression_bptc                       DONE (i965, nvc0, r600, radeonsi)
+  GL_ARB_texture_compression_bptc                       DONE (i965, r600)
   GL_ARB_compressed_texture_pixel_storage               DONE (all drivers)
-  GL_ARB_shader_atomic_counters                         DONE (i965, nvc0, radeonsi, softpipe)
+  GL_ARB_shader_atomic_counters                         DONE (i965, softpipe)
   GL_ARB_texture_storage                                DONE (all drivers)
-  GL_ARB_transform_feedback_instanced                   DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
-  GL_ARB_base_instance                                  DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
-  GL_ARB_shader_image_load_store                        DONE (i965, nvc0, radeonsi, softpipe)
+  GL_ARB_transform_feedback_instanced                   DONE (i965, nv50, r600, llvmpipe, softpipe, swr)
+  GL_ARB_base_instance                                  DONE (i965, nv50, r600, llvmpipe, softpipe, swr)
+  GL_ARB_shader_image_load_store                        DONE (i965, softpipe)
   GL_ARB_conservative_depth                             DONE (all drivers that support GLSL 1.30)
   GL_ARB_shading_language_420pack                       DONE (all drivers that support GLSL 1.30)
   GL_ARB_shading_language_packing                       DONE (all drivers)
-  GL_ARB_internalformat_query                           DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
+  GL_ARB_internalformat_query                           DONE (i965, nv50, r600, llvmpipe, softpipe, swr)
   GL_ARB_map_buffer_alignment                           DONE (all drivers)
 
 
-GL 4.3, GLSL 4.30:
+GL 4.3, GLSL 4.30 -- all DONE: nvc0, radeonsi
 
   GL_ARB_arrays_of_arrays                               DONE (all drivers that support GLSL 1.30)
   GL_ARB_ES3_compatibility                              DONE (all drivers that support GLSL 3.30)
   GL_ARB_clear_buffer_object                            DONE (all drivers)
-  GL_ARB_compute_shader                                 DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_copy_image                                     DONE (i965, nv50, nvc0, r600, radeonsi)
+  GL_ARB_compute_shader                                 DONE (i965, softpipe)
+  GL_ARB_copy_image                                     DONE (i965, nv50, r600, softpipe, llvmpipe)
   GL_KHR_debug                                          DONE (all drivers)
   GL_ARB_explicit_uniform_location                      DONE (all drivers that support GLSL)
-  GL_ARB_fragment_layer_viewport                        DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe)
-  GL_ARB_framebuffer_no_attachments                     DONE (i965, nvc0, r600, radeonsi, softpipe)
+  GL_ARB_fragment_layer_viewport                        DONE (i965, nv50, r600, llvmpipe)
+  GL_ARB_framebuffer_no_attachments                     DONE (i965, r600, softpipe)
   GL_ARB_internalformat_query2                          DONE (all drivers)
   GL_ARB_invalidate_subdata                             DONE (all drivers)
-  GL_ARB_multi_draw_indirect                            DONE (i965, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
+  GL_ARB_multi_draw_indirect                            DONE (i965, r600, llvmpipe, softpipe, swr)
   GL_ARB_program_interface_query                        DONE (all drivers)
-  GL_ARB_robust_buffer_access_behavior                  DONE (i965, nvc0, radeonsi)
-  GL_ARB_shader_image_size                              DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_shader_storage_buffer_object                   DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_stencil_texturing                              DONE (i965/gen8+, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
-  GL_ARB_texture_buffer_range                           DONE (nv50, nvc0, i965, r600, radeonsi, llvmpipe)
+  GL_ARB_robust_buffer_access_behavior                  DONE (i965)
+  GL_ARB_shader_image_size                              DONE (i965, softpipe)
+  GL_ARB_shader_storage_buffer_object                   DONE (i965, softpipe)
+  GL_ARB_stencil_texturing                              DONE (i965/gen8+, nv50, r600, llvmpipe, softpipe, swr)
+  GL_ARB_texture_buffer_range                           DONE (nv50, i965, r600, llvmpipe)
   GL_ARB_texture_query_levels                           DONE (all drivers that support GLSL 1.30)
   GL_ARB_texture_storage_multisample                    DONE (all drivers that support GL_ARB_texture_multisample)
-  GL_ARB_texture_view                                   DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
+  GL_ARB_texture_view                                   DONE (i965, nv50, r600, llvmpipe, softpipe, swr)
   GL_ARB_vertex_attrib_binding                          DONE (all drivers)
 
 
@@ -211,7 +211,7 @@ GL 4.5, GLSL 4.50:
   GL_ARB_ES3_1_compatibility                            DONE (nvc0, radeonsi)
   GL_ARB_clip_control                                   DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
   GL_ARB_conditional_render_inverted                    DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
-  GL_ARB_cull_distance                                  DONE (i965, nv50, nvc0, llvmpipe, softpipe)
+  GL_ARB_cull_distance                                  DONE (i965, nv50, nvc0, llvmpipe, softpipe, swr)
   GL_ARB_derivative_control                             DONE (i965, nv50, nvc0, r600, radeonsi)
   GL_ARB_direct_state_access                            DONE (all drivers)
   GL_ARB_get_texture_sub_image                          DONE (all drivers)
@@ -222,32 +222,32 @@ GL 4.5, GLSL 4.50:
   GL_EXT_shader_integer_mix                             DONE (all drivers that support GLSL)
 
 These are the extensions cherry-picked to make GLES 3.1
-GLES3.1, GLSL ES 3.1
+GLES3.1, GLSL ES 3.1 -- all DONE: nvc0, radeonsi
   GL_ARB_arrays_of_arrays                               DONE (all drivers that support GLSL 1.30)
-  GL_ARB_compute_shader                                 DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_draw_indirect                                  DONE (i965, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
+  GL_ARB_compute_shader                                 DONE (i965, softpipe)
+  GL_ARB_draw_indirect                                  DONE (i965, r600, llvmpipe, softpipe, swr)
   GL_ARB_explicit_uniform_location                      DONE (all drivers that support GLSL)
-  GL_ARB_framebuffer_no_attachments                     DONE (i965, nvc0, r600, radeonsi, softpipe)
+  GL_ARB_framebuffer_no_attachments                     DONE (i965, r600, softpipe)
   GL_ARB_program_interface_query                        DONE (all drivers)
-  GL_ARB_shader_atomic_counters                         DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_shader_image_load_store                        DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_shader_image_size                              DONE (i965, nvc0, radeonsi, softpipe)
-  GL_ARB_shader_storage_buffer_object                   DONE (i965, nvc0, radeonsi, softpipe)
+  GL_ARB_shader_atomic_counters                         DONE (i965, softpipe)
+  GL_ARB_shader_image_load_store                        DONE (i965, softpipe)
+  GL_ARB_shader_image_size                              DONE (i965, softpipe)
+  GL_ARB_shader_storage_buffer_object                   DONE (i965, softpipe)
   GL_ARB_shading_language_packing                       DONE (all drivers)
   GL_ARB_separate_shader_objects                        DONE (all drivers)
-  GL_ARB_stencil_texturing                              DONE (i965/gen8+, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe, swr)
-  GL_ARB_texture_multisample (Multisample textures)     DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe)
+  GL_ARB_stencil_texturing                              DONE (i965/gen8+, nv50, r600, llvmpipe, softpipe, swr)
+  GL_ARB_texture_multisample (Multisample textures)     DONE (i965, nv50, r600, llvmpipe, softpipe)
   GL_ARB_texture_storage_multisample                    DONE (all drivers that support GL_ARB_texture_multisample)
   GL_ARB_vertex_attrib_binding                          DONE (all drivers)
-  GS5 Enhanced textureGather                            DONE (i965, nvc0, r600, radeonsi)
-  GS5 Packing/bitfield/conversion functions             DONE (i965, nvc0, r600, radeonsi)
+  GS5 Enhanced textureGather                            DONE (i965, r600)
+  GS5 Packing/bitfield/conversion functions             DONE (i965, r600)
   GL_EXT_shader_integer_mix                             DONE (all drivers that support GLSL)
 
   Additional functionality not covered above:
       glMemoryBarrierByRegion                           DONE
       glGetTexLevelParameter[fi]v - needs updates       DONE
       glGetBooleani_v - restrict to GLES enums
-      gl_HelperInvocation support                       DONE (i965, nvc0, r600, radeonsi)
+      gl_HelperInvocation support                       DONE (i965, r600)
 
 GLES3.2, GLSL ES 3.2
   GL_EXT_color_buffer_float                             DONE (all drivers)

docs/devinfo.html

@@ -684,9 +684,11 @@ To add a new GL extension to Mesa you have to do at least the following.
 </li>
 <li>
    Add a new entry to the <code>gl_extensions</code> struct in mtypes.h
+   if the extension requires driver capabilities not already exposed by
+   another extension.
 </li>
 <li>
-   Update the <code>extensions.c</code> file.
+   Add a new entry to the src/mesa/main/extensions_table.h file.
 </li>
 <li>
    From this point, the best way to proceed is to find another extension,
@@ -697,12 +699,18 @@ To add a new GL extension to Mesa you have to do at least the following.
    If the new extension adds new GL state, the functions in get.c, enable.c
    and attrib.c will most likely require new code.
 </li>
+<li>
+   To determine if the new extension is active in the current context,
+   use the auto-generated _mesa_has_##name_str() function defined in
+   src/mesa/main/extensions.h.
+</li>
 <li>
    The dispatch tests check_table.cpp and dispatch_sanity.cpp
    should be updated with details about the new extensions functions. These
    tests are run using 'make check'
 </li>
 </ul>
+</p>
 
 
 

docs/relnotes/12.0.0.html

@@ -14,7 +14,7 @@
 <iframe src="../contents.html"></iframe>
 <div class="content">
 
-<h1>Mesa 12.0.0 Release Notes / July 8, 2016</h1>
+<h1>Mesa 12.0.0 Release Notes / TBD</h1>
 
 <p>
 Mesa 12.0.0 is a new development release.
@@ -33,8 +33,7 @@ because compatibility contexts are not supported.
 
 <h2>SHA256 checksums</h2>
 <pre>
-3b8fa4d86d78f8f6ec86055b92ad1afe869001483593b3dd4531184b8bc4fcfb  mesa-12.0.0.tar.gz
-0090c025219318935124292b482e3439bc43e8c074ad01086449fcad88547dc6  mesa-12.0.0.tar.xz
+TBD.
 </pre>
 
 
@@ -79,256 +78,11 @@ Note: some of the new features are only available with certain drivers.
 
 <h2>Bug fixes</h2>
 
-<ul>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=42187">Bug 42187</a> - ES 1.1 conformance pntszary.c fail</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=71789">Bug 71789</a> - [r300g] Visuals not found in (default) depth = 24</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=81585">Bug 81585</a> - piglit spec_glsl-1.10_compiler_literals_invalid-float-suffix-capital-f.vert fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=83036">Bug 83036</a> - [ILK]Piglit spec_ARB_copy_image_arb_copy_image-formats fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89607">Bug 89607</a> - Assertion hit in opt_array_splitting with recursive array indexing</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=90513">Bug 90513</a> - Odd gray and red flicker in The Talos Principle on GK104</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91526">Bug 91526</a> - World of Warcraft (on Wine) has UI corruption with nouveau</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=92363">Bug 92363</a> - [BSW/BDW] ogles1conform Gets test fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=92628">Bug 92628</a> - HTTP site for Mesa downloads</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=92743">Bug 92743</a> - Centroid shouldn't have to match between the FS and the VS</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=92850">Bug 92850</a> - Segfault loading War Thunder</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93054">Bug 93054</a> - [BDW] DiRT Showdown and Bioshock Infinite only render half the screen (bottom left triangle)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93524">Bug 93524</a> - Clover doesn't build</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93551">Bug 93551</a> - Divinity: Original Sin Enhanced Edition(Native) crash on start</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93667">Bug 93667</a> - Crash in eglCreateImageKHR with huge texture size</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93767">Bug 93767</a> - Glitches with soft shadows and MSAA in Knights of the Old Republic 2</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93840">Bug 93840</a> - [i965] Alien: Isolation fails with GL_ARB_compute_shader enabled</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=93962">Bug 93962</a> - [HSW, regression, bisected, CTS] ES2-CTS.gtf.GL2FixedTests.scissor.scissor - segfault/asserts</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94081">Bug 94081</a> - [HSW] compute shader shared var + atomic op = fail</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94086">Bug 94086</a> - Multiple conflicting libGL libraries installed</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94116">Bug 94116</a> - program interface queries not returning right data for UBO / GL_BLOCK_INDEX</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94129">Bug 94129</a> - Mesa's compiler should warn about undefined values</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94181">Bug 94181</a> - [regression] piglit.spec.ext_framebuffer_object.getteximage-formats init-by-clear-and-render</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94193">Bug 94193</a> - [llvmpipe] Line antialiasing looks different when GL_LINE_STIPPLE is enabled with pattern 0xffff</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94198">Bug 94198</a> - [HSW] segfault in copy image when copying from cubemap to 2d</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94199">Bug 94199</a> - Shader abort/crash</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94253">Bug 94253</a> - [llvmpipe] piglit gl-1.0-swapbuffers-behavior regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94254">Bug 94254</a> - [llvmpipe] [softpipe] piglit read-front regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94257">Bug 94257</a> - [softpipe] piglit glx-copy-sub-buffer regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94274">Bug 94274</a> - [swrast] piglit arb_occlusion_query2-render regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94284">Bug 94284</a> - [radeonsi] outlast segfault on start</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94291">Bug 94291</a> - llvmpipe tests fail if built on skylake i7-6700k</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94348">Bug 94348</a> - vkBindImageMemory doesn't take into account the offset when the image is used as a depth buffer</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94383">Bug 94383</a> - build error on i386 when enabling swr</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94388">Bug 94388</a> - r600_blit.c:281: r600_decompress_depth_textures: Assertion `tex-&gt;is_depth &amp;&amp; !tex-&gt;is_flushing_texture' failed.</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94412">Bug 94412</a> - Trine 3 misrender</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94447">Bug 94447</a> - glsl/glcpp/tests/glcpp-test-cr-lf regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94453">Bug 94453</a> - dEQP-GLES3.functional.clipping.line.wide_line_clip_viewport_{center,corner} fail</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94454">Bug 94454</a> - dEQP-GLES3.functional.clipping.point.wide_point_clip* fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94456">Bug 94456</a> - dEQP-GLES3.functional.state_query.floats.{blend_color,color_clear_value,depth_clear_value}_getinteger64 fail</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94458">Bug 94458</a> - dEQP-GLES3.functional.state_query.fbo.framebuffer_attachment_x_size_initial fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94468">Bug 94468</a> - [HSW, regression, bisected] numerous Sascha demos render incorrectly</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94481">Bug 94481</a> - softpipe - access violation in img_filter_2d_nearest</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94485">Bug 94485</a> - dEQP-GLES3.functional.negative_api.shader.compile_shader and delete_shader broken by Meta</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94524">Bug 94524</a> - Wrong gl_TessLevelOuter interpretation for isolines</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94595">Bug 94595</a> - [Mesa AMD&amp;swrast] Texture views attached as framebuffers return their viewed tecture's color encoding and render incorrectly</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94657">Bug 94657</a> - [llvmpipe] [softpipe] piglit arb_texture_view-getteximage-srgb regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94661">Bug 94661</a> - [bdw, skl] vk-cts: new test failing</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94671">Bug 94671</a> - [radeonsi] Blue-ish textures in Shadow of Mordor</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94713">Bug 94713</a> - [Gen8+] ES 3.1 Stencil texturing broken for 2DArray/Cubes</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94747">Bug 94747</a> - Convert phi nodes to logical operations</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94835">Bug 94835</a> - Increase fragment shader sample limits from 16 to 32 (AMD Linux - Mesa/RadeonSi)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94847">Bug 94847</a> - [ES3.1CTS] es31-cts.draw_buffers_indexed.color_masks fails</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94896">Bug 94896</a> - [vulkan] new CTS tests fail on i965</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94904">Bug 94904</a> - [vulkan, BSW] dEQP-VK.api.object_management.multithreaded_per_thread_device intermittent crash</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94907">Bug 94907</a> - codegen/nv50_ir_ra.cpp:1330:29: error: ‘isinf’ was not declared in this scope</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94909">Bug 94909</a> - [llvmpipe] piglit fs-roundEven-float regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94917">Bug 94917</a> - radeonsi supports GL_ARB_shader_storage_buffer_object with 0 GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94924">Bug 94924</a> - [GEN8] Ungine Valley fails to run due to &quot;intel_do_flush_locked failed: Input/output error&quot;</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94925">Bug 94925</a> - Crash in egl_dri3_get_dri_context with Dolphin EGL/X11 in single-core mode</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94944">Bug 94944</a> - [regression, hswgt1] gpu hang on arb_shader_image_load_store</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94955">Bug 94955</a> - Uninitialized variables leads to random segfaults (valgrind log, apitrace attached)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94969">Bug 94969</a> - build fails because install-data-local doesn't follow $DESTDIR</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94972">Bug 94972</a> - blend failures on llvmpipe with llvm 3.7 due to vector selects</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94979">Bug 94979</a> - dolphin-emu rendering broken on gallium/SWR + crashing often</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94984">Bug 94984</a> - XCom2 crashes with SIGSEGV on radeonsi</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94994">Bug 94994</a> - OSMesaGetProcAdress always fails on mangled OSMesa</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94997">Bug 94997</a> - [vulkan, SKL,BDW,HSW] deqp-vk.spirv_assembly.instruction.compute.opcopymemory.array regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94998">Bug 94998</a> - [vulkan] deqp-vk.pipeline.push_constant.graphics_pipeline.count_3shader_vgf regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95001">Bug 95001</a> - [vulkan] deqp-vk.binding_model.shader_access regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95005">Bug 95005</a> - Unreal engine demos segfault after shader compilation error with OpenGL 4.3</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95026">Bug 95026</a> - Alien Isolation segfault after initial loading screen/video</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95034">Bug 95034</a> - vkResetCommandPool should not destroy the command buffers.</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95071">Bug 95071</a> - [bisected] Wrong colors in KDE/Qt applications</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95133">Bug 95133</a> - X-COM Enemy Within crashes when entering tactical mission with Bonaire</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95138">Bug 95138</a> - [deqp, 32bit, gen8+] deqp-gles31.functional.draw_indirect.negative</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95142">Bug 95142</a> - [ES3.1CTS,GEN8] ESEXT-CTS.draw_elements_base_vertex_tests.invalid_mapped_bos assertion</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95158">Bug 95158</a> - glx-test compilation fails in `make check`</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95164">Bug 95164</a> - GLSL compiler (linker I think) emits assertion upon call to glAttachShader</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95180">Bug 95180</a> - rasterizer/memory/Convert.h:170:9: error: ‘__builtin_isnan’ is not a member of ‘std’</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95198">Bug 95198</a> - Shadow of Mordor beta has missing geometry with gl 4.3</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95203">Bug 95203</a> - Tonga GST/OMX/VCE encode broken since mesa: st/omx: Fix resource leak on OMX_ErrorNone</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95211">Bug 95211</a> - scons TypeError: 'tuple' object is not callable</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95246">Bug 95246</a> - Segfault in glBindFramebuffer()</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95251">Bug 95251</a> - vdpau decoder capabilities: not supported</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95252">Bug 95252</a> - [deqp] deqp-gles31.functional.debug.object_labels.query_length_only crashes</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95292">Bug 95292</a> - [IVB,SKL] vulkan: stride/tiling issue with vkCmdCopyBufferToImage from larger source buffer into destination image</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95296">Bug 95296</a> - nir_lower_double_packing.c:79:4: error: void function 'lower_double_pack_impl' should not return a value [-Wreturn-type]</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95324">Bug 95324</a> - GL33-CTS.gtf32.GL3Tests.packed_pixels.packed_pixels_pbo fails in one case on Haswell</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95370">Bug 95370</a> - [965GM] piglit fails many tests after a5d7e144</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95373">Bug 95373</a> - Suspicious warning in brw_blorp_clear.cpp</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95403">Bug 95403</a> - [GK110] misaligned_gpr spamming dmesg when playing victor vran</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95419">Bug 95419</a> - [HSW][regression][bisect] RPG Maker game gives &quot;invalid floating point operation&quot; at startup</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95456">Bug 95456</a> - glXGetFBConfigs has invalid screen bounds</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95462">Bug 95462</a> - [BXT,BSW] arb_gpu_shader_fp64 causes gpu hang</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95529">Bug 95529</a> - [regression, bisected] Image corruption in Chrome</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95537">Bug 95537</a> - Invalid argument  in anv_ioctl called from anv_physical_device_init</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96221">Bug 96221</a> - nir/nir_lower_tex.c:202: error: unknown field ‘f32’ specified in initializer</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96228">Bug 96228</a> - SSBO test regressions from mesa 5b267509</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96236">Bug 96236</a> - dri_interface.h:404: error: redefinition of typedef ‘mesa_glinterop_device_info’</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96238">Bug 96238</a> - swr fails to build outside of the main directory</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96239">Bug 96239</a> - [radeonsi tessellation] [R9 290/390] Random &quot;texture flickering&quot; (Shadow of Mordor, Tomb Raider, Unigine Heaven 4.0)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96258">Bug 96258</a> - [NVC0] Hang when running compute program</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96285">Bug 96285</a> - Mesa build broken</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96299">Bug 96299</a> - [vulkan] 64 regressions due to mesa d5f2f32</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96346">Bug 96346</a> - [SNB,CTS] es2-cts.gtf.gl.atan regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96349">Bug 96349</a> - [CTS,SKL,BSW,BDW,KBL,BXT] es31-cts.arrays_of_arrays.interactionuniformbuffers3</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96351">Bug 96351</a> - [CTS,SKL,KBL,BXT] es2-cts.gtf.gl2extensiontests.egl_image.egl_image</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96358">Bug 96358</a> - SSO: wrong interface validation between GS and VS (regresion due to latest gles 3.1)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96425">Bug 96425</a> - [bisected] occasional dark render in The Talos Principle</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96504">Bug 96504</a> - [vulkancts] compute tests crash</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96516">Bug 96516</a> - [bisected: 482526] &quot;clover: Update OpenCL version string to match OpenGL&quot;: clover's build fails because of missing git_sha1.h</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96565">Bug 96565</a> - Clive Barker's Jericho displays strange,vivid colors when motion blur enabled</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96607">Bug 96607</a> - [bisected] texture misrender / flicker in The Talos Principle on SKL</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96617">Bug 96617</a> - gl_SecondaryFragDataEXT doesn't work for extended blend func</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96629">Bug 96629</a> - dEQP-GLES2.functional.texture.completeness.cube.not_positive_level_0: Assertion `width &gt;= 1' failed.</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96639">Bug 96639</a> - st/mesa: transfer_map with too-high level with dEQP-GLES2.functional.texture.completeness.cube.extra_level</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96674">Bug 96674</a> - [SNB, ILK] spec.ext_image_dma_buf_import.ext_image_dma_buf_import-sample_nv1</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96765">Bug 96765</a> - BindFragDataLocationIndexed on array fragment shader output.</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96791">Bug 96791</a> - Cannot use image from swapchains for sampling</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96825">Bug 96825</a> - anv_device.c:31:27: fatal error: anv_timestamp.h: No such file or directory</li>
-
-</ul>
-
+TBD.
 
 <h2>Changes</h2>
 
-Radeon drivers (r600 and radeonsi) now require LLVm 3.6 as a minimum.
+TBD.
 
 </div>
 </body>

docs/relnotes/12.0.2.html

@@ -1,403 +0,0 @@
-<!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 12.0.2 Release Notes / September 2, 2016</h1>
-
-<p>
-Mesa 12.0.2 is a bug fix release which fixes bugs found since the 12.0.1 release.
-</p>
-<p>
-Mesa 12.0.2 implements the OpenGL 4.3 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 4.3.  OpenGL
-4.3 is <strong>only</strong> available if requested at context creation
-because compatibility contexts are not supported.
-</p>
-
-
-<h2>SHA256 checksums</h2>
-<pre>
-a08565ab1273751ebe2ffa928cbf785056594c803077c9719d0763da780f2918  mesa-12.0.2.tar.gz
-d957a5cc371dcd7ff2aa0d87492f263aece46f79352f4520039b58b1f32552cb  mesa-12.0.2.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=69622">Bug 69622</a> - eglTerminate then eglMakeCurrent crahes</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=89599">Bug 89599</a> - symbol 'x86_64_entry_start' is already defined when building with LLVM/clang</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=91342">Bug 91342</a> - Very dark textures on some objects in indoors environments in Postal 2</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=92306">Bug 92306</a> - GL Excess demo renders incorrectly on nv43</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94148">Bug 94148</a> - Framebuffer considered invalid when a draw call is done before glCheckFramebufferStatus</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96274">Bug 96274</a> - [NVC0] Failure when compiling compute shader: Assertion `bb-&gt;getFirst()-&gt;serial &lt;= bb-&gt;getExit()-&gt;serial' failed</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96358">Bug 96358</a> - SSO: wrong interface validation between GS and VS (regresion due to latest gles 3.1)</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96381">Bug 96381</a> - Texture artifacts with immutable texture storage and mipmaps</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96762">Bug 96762</a> - [radeonsi,apitrace] Firewatch: nothing rendered in scrollable (text) areas</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96835">Bug 96835</a> - &quot;gallium: Force blend color to 16-byte alignment&quot; crash with &quot;-march=native -O3&quot; causes some 32bit games to crash</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96850">Bug 96850</a> - Crucible tests fail for 32bit mesa</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96908">Bug 96908</a> - [radeonsi] MSAA causes graphical artifacts</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96911">Bug 96911</a> - webgl2 conformance2/textures/misc/tex-mipmap-levels.html crashes 12.1 Intel driver</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96971">Bug 96971</a> - invariant qualifier is not valid for shader inputs</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97039">Bug 97039</a> - The Talos Principle and Serious Sam 3 GPU faults</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97207">Bug 97207</a> - [IVY BRIDGE] Fragment shader discard writing to depth</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97214">Bug 97214</a> - X not running with error &quot;Failed to make EGL context current&quot;</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97225">Bug 97225</a> - [i965 on HD4600 Haswell] xcom switch to ingame cinematics cause segmentation fault</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97231">Bug 97231</a> - GL_DEPTH_CLAMP doesn't clamp to the far plane</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97307">Bug 97307</a> - glsl/glcpp/tests/glcpp-test regression</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97331">Bug 97331</a> - glDrawElementsBaseVertex doesn't work in display list on i915</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97351">Bug 97351</a> - DrawElementsBaseVertex with VBO ignores base vertex on Intel GMA 9xx in some cases</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97426">Bug 97426</a> - glScissor gives vertically inverted result</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97476">Bug 97476</a> - Shader binaries should not be stored in the PipelineCache</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97567">Bug 97567</a> - [SNB, ILK] ctl, piglit regressions in mesa 12.0.2rc1</li>
-
-</ul>
-
-
-<h2>Changes</h2>
-
-<p>Andreas Boll (1):</p>
-<ul>
-  <li>configure.ac: Use ${datarootdir} for --with-vulkan-icddir help string too</li>
-</ul>
-
-<p>Bernard Kilarski (1):</p>
-<ul>
-  <li>glx: fix error code when there is no context bound</li>
-</ul>
-
-<p>Brian Paul (4):</p>
-<ul>
-  <li>svga: handle mismatched number of samplers, sampler views</li>
-  <li>mesa: use _mesa_clear_texture_image() in clear_texture_fields()</li>
-  <li>swrast: fix incorrectly positioned putImage() in swrast driver</li>
-  <li>mesa: fix format conversion bug in get_tex_rgba_uncompressed()</li>
-</ul>
-
-<p>Chad Versace (2):</p>
-<ul>
-  <li>i965: Fix miptree layout for EGLImage-based renderbuffers</li>
-  <li>i965: Respect miptree offsets in intel_readpixels_tiled_memcpy()</li>
-</ul>
-
-<p>Christian König (1):</p>
-<ul>
-  <li>st/mesa: fix reference counting bug in st_vdpau</li>
-</ul>
-
-<p>Chuck Atkins (1):</p>
-<ul>
-  <li>swr: Refactor checks for compiler feature flags</li>
-</ul>
-
-<p>Daniel Scharrer (1):</p>
-<ul>
-  <li>mesa: Fix fixed function spot lighting on newer hardware (again)</li>
-</ul>
-
-<p>Dave Airlie (2):</p>
-<ul>
-  <li>anv: fix writemask on blit fragment shader.</li>
-  <li>st/glsl_to_tgsi: fix st_src_reg_for_double constant.</li>
-</ul>
-
-<p>Emil Velikov (15):</p>
-<ul>
-  <li>docs: add sha256 checksums for 12.0.1</li>
-  <li>mesa: automake: list builddir before srcdir</li>
-  <li>mesa: scons: list builddir before srcdir</li>
-  <li>i965: store reference to the context within struct brw_fence (v2)</li>
-  <li>anv: remove internal 'validate' layer</li>
-  <li>anv: automake: use VISIBILITY_CFLAGS to restrict symbol visibility</li>
-  <li>anv: automake: build with -Bsymbolic</li>
-  <li>anv: do not export the Vulkan API</li>
-  <li>anv: remove dummy VK_DEBUG_MARKER_EXT entry points</li>
-  <li>isl: automake: use VISIBILITY_CFLAGS to restrict symbol visibility</li>
-  <li>cherry-ignore: temporary(?) drop "a4xx: make sure to actually clamp depth"</li>
-  <li>i915: Check return value of screen-&gt;image.loader-&gt;getBuffers</li>
-  <li>Revert "i965/miptree: Set logical_depth0 == 6 for cube maps"</li>
-  <li>glx/glvnd: list the strcmp arguments in correct order</li>
-  <li>Update version to 12.0.2</li>
-</ul>
-
-<p>Eric Anholt (4):</p>
-<ul>
-  <li>vc4: Close our screen's fd on screen close.</li>
-  <li>vc4: Disable early Z with computed depth.</li>
-  <li>vc4: Fix a leak of the src[] array of VPM reads in optimization.</li>
-  <li>vc4: Fix leak of the bo_handles table.</li>
-</ul>
-
-<p>Francisco Jerez (3):</p>
-<ul>
-  <li>i965: Emit SKL VF cache invalidation W/A from brw_emit_pipe_control_flush.</li>
-  <li>i965: Make room in the batch epilogue for three more pipe controls.</li>
-  <li>i965: Fix remaining flush vs invalidate race conditions in brw_emit_pipe_control_flush.</li>
-</ul>
-
-<p>Haixia Shi (1):</p>
-<ul>
-  <li>platform_android: prevent deadlock in droid_swap_buffers</li>
-</ul>
-
-<p>Ian Romanick (5):</p>
-<ul>
-  <li>mesa: Strip arrayness from interface block names in some IO validation</li>
-  <li>glsl: Pack integer and double varyings as flat even if interpolation mode is none</li>
-  <li>glcpp: Track the actual version instead of just the version_resolved flag</li>
-  <li>glcpp: Only disallow #undef of pre-defined macros on GLSL ES &gt;= 3.00 shaders</li>
-  <li>glsl: Mark cube map array sampler types as reserved in GLSL ES 3.10</li>
-</ul>
-
-<p>Ilia Mirkin (16):</p>
-<ul>
-  <li>mesa: etc2 online compression is unsupported, don't attempt it</li>
-  <li>st/mesa: return appropriate mesa format for ETC texture formats</li>
-  <li>mesa: set _NEW_BUFFERS when updating texture bound to current buffers</li>
-  <li>nv50,nvc0: srgb rendering is only available for rgba/bgra</li>
-  <li>vbo: allow DrawElementsBaseVertex in display lists</li>
-  <li>gallium/util: add helper to compute zmin/zmax for a viewport state</li>
-  <li>nv50,nvc0: fix depth range when halfz is enabled</li>
-  <li>nv50/ir: fix bb positions after exit instructions</li>
-  <li>vbo: add basevertex when looking up elements for vbo splitting</li>
-  <li>a4xx: only disable depth clipping, not all clipping, when requested</li>
-  <li>nv50/ir: make sure cfg iterator always hits all blocks</li>
-  <li>main: add missing EXTRA_END in OES_sample_variables get check</li>
-  <li>nouveau: always enable at least one RC</li>
-  <li>nv30: only bail on color/depth bpp mismatch when surfaces are swizzled</li>
-  <li>a4xx: make sure to actually clamp depth as requested</li>
-  <li>gk110/ir: fix quadop dall emission</li>
-</ul>
-
-<p>Jan Ziak (2):</p>
-<ul>
-  <li>egl/x11: avoid using freed memory if dri2 init fails</li>
-  <li>loader: fix memory leak in loader_dri3_open</li>
-</ul>
-
-<p>Jason Ekstrand (31):</p>
-<ul>
-  <li>nir/spirv: Don't multiply the push constant block size by 4</li>
-  <li>anv: Add a stub for CmdCopyQueryPoolResults on Ivy Bridge</li>
-  <li>glsl/types: Fix function type comparison function</li>
-  <li>glsl/types: Use _mesa_hash_data for hashing function types</li>
-  <li>genxml: Make gen6-7 blending look more like gen8</li>
-  <li>anv/pipeline: Unify blend state setup between gen7 and gen8</li>
-  <li>anv: Enable independentBlend on gen7</li>
-  <li>anv: Add an align_down_npot_u32 helper</li>
-  <li>anv: Handle VK_WHOLE_SIZE properly for buffer views</li>
-  <li>i965/miptree: Enforce that height == 1 for 1-D array textures</li>
-  <li>i965/miptree: Set logical_depth0 == 6 for cube maps</li>
-  <li>nir: Add a nir_deref_foreach_leaf helper</li>
-  <li>nir/inline: Constant-initialize local variables in the callee if needed</li>
-  <li>anv/pipeline: Set up point coord enables</li>
-  <li>i965/miptree: Stop multiplying cube depth by 6 in HiZ calculations</li>
-  <li>i965/vec4: Make opt_vector_float reset at the top of each block</li>
-  <li>anv/blit2d: Add a format parameter to bind_dst and create_iview</li>
-  <li>anv/blit2d: Add support for RGB destinations</li>
-  <li>anv/clear: Make cmd_clear_image take an actual VkClearValue</li>
-  <li>anv/clear: Clear E5B9G9R9 images as R32_UINT</li>
-  <li>anv: Include the pipeline layout in the shader hash</li>
-  <li>isl: Allow multisampled array textures</li>
-  <li>anv/descriptor_set: memset anv_descriptor_set_layout</li>
-  <li>anv/pipeline: Fix bind maps for fragment output arrays</li>
-  <li>anv/allocator: Correctly set the number of buckets</li>
-  <li>anv/pipeline: Properly handle OOM during shader compilation</li>
-  <li>anv: Remove unused fields from anv_pipeline_bind_map</li>
-  <li>anv: Add pipeline_has_stage guards a few places</li>
-  <li>anv: Add a struct for storing a compiled shader</li>
-  <li>anv/pipeline: Add support for caching the push constant map</li>
-  <li>anv: Rework pipeline caching</li>
-</ul>
-
-<p>José Fonseca (2):</p>
-<ul>
-  <li>appveyor: Install pywin32 extensions.</li>
-  <li>appveyor: Force Visual Studio 2013 image.</li>
-</ul>
-
-<p>Kenneth Graunke (21):</p>
-<ul>
-  <li>genxml: Add CLIPMODE_* prefix to 3DSTATE_CLIP's "Clip Mode" enum values.</li>
-  <li>genxml: Add APIMODE_D3D missing enum values and improve consistency.</li>
-  <li>anv: Fix near plane clipping on Gen7/7.5.</li>
-  <li>anv: Enable early culling on Gen7.</li>
-  <li>anv: Unify 3DSTATE_CLIP code across generations.</li>
-  <li>genxml: Rename "API Rendering Disable" to "Rendering Disable".</li>
-  <li>anv: Properly call gen75_emit_state_base_address on Haswell.</li>
-  <li>i965: Include VUE handles for GS with invocations &gt; 1.</li>
-  <li>nir: Add a base const_index to shared atomic intrinsics.</li>
-  <li>i965: Fix shared atomic intrinsics to pay attention to base.</li>
-  <li>mesa: Add GL_BGRA_EXT to the list of GenerateMipmap internal formats.</li>
-  <li>mesa: Don't call GenerateMipmap if Width or Height == 0.</li>
-  <li>glsl: Delete bogus ir_set_program_inouts assert.</li>
-  <li>glsl: Fix the program resource names of gl_TessLevelOuter/Inner[].</li>
-  <li>glsl: Fix location bias for patch variables.</li>
-  <li>glsl: Fix invariant matching in GLSL 4.30 and GLSL ES 1.00.</li>
-  <li>mesa: Fix uf10_to_f32() scale factor in the E == 0 and M != 0 case.</li>
-  <li>nir/builder: Add bany_inequal and bany helpers.</li>
-  <li>i965: Implement the WaPreventHSTessLevelsInterference workaround.</li>
-  <li>i965: Fix execution size of scalar TCS barrier setup code.</li>
-  <li>i965: Fix barrier count shift in scalar TCS backend.</li>
-</ul>
-
-<p>Leo Liu (2):</p>
-<ul>
-  <li>st/omx/enc: check uninitialized list from task release</li>
-  <li>vl/dri3: fix a memory leak from front buffer</li>
-</ul>
-
-<p>Marek Olšák (7):</p>
-<ul>
-  <li>glsl_to_tgsi: don't use the negate modifier in integer ops after bitcast</li>
-  <li>radeonsi: add a workaround for a compute VGPR-usage LLVM bug</li>
-  <li>winsys/amdgpu: disallow DCC with mipmaps</li>
-  <li>gallium/util: fix align64</li>
-  <li>radeonsi: only set dual source blending for MRT0</li>
-  <li>radeonsi: fix VM faults due NULL internal const buffers on CIK</li>
-  <li>radeonsi: disable SDMA texture copying on Carrizo</li>
-</ul>
-
-<p>Matt Turner (4):</p>
-<ul>
-  <li>mapi: Massage code to allow clang to compile.</li>
-  <li>i965/vec4: Ignore swizzle of VGRF for use by var_range_end().</li>
-  <li>mesa: Use AC_HEADER_MAJOR to include correct header for major().</li>
-  <li>nir: Walk blocks in source code order in lower_vars_to_ssa.</li>
-</ul>
-
-<p>Michel Dänzer (1):</p>
-<ul>
-  <li>glx: Don't use current context in __glXSendError</li>
-</ul>
-
-<p>Miklós Máté (1):</p>
-<ul>
-  <li>vbo: set draw_id</li>
-</ul>
-
-<p>Nanley Chery (5):</p>
-<ul>
-  <li>anv/descriptor_set: Fix binding partly undefined descriptor sets</li>
-  <li>isl: Fix assert on raw buffer surface state size</li>
-  <li>anv/device: Fix max buffer range limits</li>
-  <li>isl: Fix isl_tiling_is_any_y()</li>
-  <li>anv/gen7_pipeline: Set PixelShaderKillPixel for discards</li>
-</ul>
-
-<p>Nicolai Hähnle (7):</p>
-<ul>
-  <li>radeonsi: explicitly choose center locations for 1xAA on Polaris</li>
-  <li>radeonsi: fix Polaris MSAA regression</li>
-  <li>radeonsi: ensure sample locations are set for line and polygon smoothing</li>
-  <li>st_glsl_to_tgsi: only skip over slots of an input array that are present</li>
-  <li>glsl: fix optimization of discard nested multiple levels</li>
-  <li>radeonsi: flush TC L2 cache for indirect draw data</li>
-  <li>radeonsi: add si_set_rw_buffer to be used for internal descriptors</li>
-</ul>
-
-<p>Nicolas Boichat (6):</p>
-<ul>
-  <li>egl/dri2: dri2_make_current: Set EGL error if bindContext fails</li>
-  <li>egl/wayland: Set disp-&gt;DriverData to NULL on error</li>
-  <li>egl/surfaceless: Set disp-&gt;DriverData to NULL on error</li>
-  <li>egl/drm: Set disp-&gt;DriverData to NULL on error</li>
-  <li>egl/android: Set dpy-&gt;DriverData to NULL on error</li>
-  <li>egl/dri2: Add reference count for dri2_egl_display</li>
-</ul>
-
-<p>Rob Herring (3):</p>
-<ul>
-  <li>Android: add missing u_math.h include path for libmesa_isl</li>
-  <li>vc4: fix vc4_resource_from_handle() stride calculation</li>
-  <li>vc4: add hash table look-up for exported dmabufs</li>
-</ul>
-
-<p>Samuel Pitoiset (7):</p>
-<ul>
-  <li>nvc0/ir: fix images indirect access on Fermi</li>
-  <li>nvc0: fix the driver cb size when draw parameters are used</li>
-  <li>gm107/ir: add missing NEG modifier for IADD32I</li>
-  <li>gm107/ir: make use of ADD32I for all immediates</li>
-  <li>nvc0: upload sample locations on GM20x</li>
-  <li>nvc0: invalidate textures/samplers on GK104+</li>
-  <li>nv50/ir: always emit the NDV bit for OP_QUADOP</li>
-</ul>
-
-<p>Stefan Dirsch (1):</p>
-<ul>
-  <li>Avoid overflow in 'last' variable of FindGLXFunction(...)</li>
-</ul>
-
-<p>Stencel, Joanna (1):</p>
-<ul>
-  <li>egl/wayland-egl: Fix for segfault in dri2_wl_destroy_surface.</li>
-</ul>
-
-<p>Tim Rowley (2):</p>
-<ul>
-  <li>Revert "gallium: Force blend color to 16-byte alignment"</li>
-  <li>swr: switch from overriding -march to selecting features</li>
-</ul>
-
-<p>Tomasz Figa (8):</p>
-<ul>
-  <li>gallium/dri: Add shared glapi to LIBADD on Android</li>
-  <li>egl/android: Remove unused variables</li>
-  <li>egl/android: Check return value of dri2_get_dri_config()</li>
-  <li>egl/android: Stop leaking DRI images</li>
-  <li>gallium/winsys/kms: Fix double refcount when importing from prime FD (v2)</li>
-  <li>gallium/winsys/kms: Fully initialize kms_sw_dt at prime import time (v2)</li>
-  <li>gallium/winsys/kms: Move display target handle lookup to separate function</li>
-  <li>gallium/winsys/kms: Look up the GEM handle after importing a prime FD</li>
-</ul>
-
-
-</div>
-</body>
-</html>

docs/relnotes/12.0.3.html

@@ -1,71 +0,0 @@
-<!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 12.0.3 Release Notes / September 15, 2016</h1>
-
-<p>
-Mesa 12.0.3 is a bug fix release which fixes bugs found since the 12.0.3 release.
-</p>
-<p>
-Mesa 12.0.3 implements the OpenGL 4.3 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 4.3.  OpenGL
-4.3 is <strong>only</strong> available if requested at context creation
-because compatibility contexts are not supported.
-</p>
-
-
-<h2>SHA256 checksums</h2>
-<pre>
-79abcfab3de30dbd416d1582a3cf6b1be308466231488775f1b7bb43be353602 mesa-12.0.3.tar.gz
-1dc86dd9b51272eee1fad3df65e18cda2e556ef1bc0b6e07cd750b9757f493b1 mesa-12.0.3.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=97781">Bug 97781</a> - [HSW, BYT, IVB] es2-cts.gtf.gl2extensiontests.depth_texture_cube_map.depth_texture_cube_map</li>
-
-</ul>
-
-
-<h2>Changes</h2>
-
-<p>Emil Velikov (3):</p>
-<ul>
-  <li>docs: add sha256 checksums for 12.0.2</li>
-  <li>Revert "i965/miptree: Stop multiplying cube depth by 6 in HiZ calculations"</li>
-  <li>Update version to 12.0.3</li>
-</ul>
-
-<p>José Fonseca (1):</p>
-<ul>
-  <li>appveyor: Update winflexbison download URL.</li>
-</ul>
-
-
-</div>
-</body>
-</html>

docs/relnotes/12.0.4.html

@@ -1,321 +0,0 @@
-<!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 12.0.4 Release Notes / November 10, 2016</h1>
-
-<p>
-Mesa 12.0.4 is a bug fix release which fixes bugs found since the 12.0.4 release.
-</p>
-<p>
-Mesa 12.0.4 implements the OpenGL 4.3 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 4.3.  OpenGL
-4.3 is <strong>only</strong> available if requested at context creation
-because compatibility contexts are not supported.
-</p>
-
-
-<h2>SHA256 checksums</h2>
-<pre>
-22026ce4f1c6a7908b0d10ff057decec0a5633afe7f38a0cef5c08d0689f02a6 mesa-12.0.4.tar.gz
-5d6003da867d3f54e5000b4acdfc37e6cce5b6a4459274fdad73e24bd2f0065e mesa-12.0.4.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=71759">Bug 71759</a> - Intel driver fails with &quot;intel_do_flush_locked failed: No such file or directory&quot; if buffer imported with EGL_NATIVE_PIXMAP_KHR</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=94354">Bug 94354</a> - R9285 Unigine Valley perf regression since radeonsi: use re-Z</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96770">Bug 96770</a> - include/GL/mesa_glinterop.h:62: error: redefinition of typedef ‘GLXContext’</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97231">Bug 97231</a> - GL_DEPTH_CLAMP doesn't clamp to the far plane</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97233">Bug 97233</a> - vkQuake VkSpecializationMapEntry related bug</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97260">Bug 97260</a> - R9 290 low performance in Linux 4.7</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97549">Bug 97549</a> - [SNB, BXT] up to 40% perf drop from &quot;loader/dri3: Overhaul dri3_update_num_back&quot; commit</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97887">Bug 97887</a> - llvm segfault in janusvr -render vive</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=98025">Bug 98025</a> - [radeonsi] incorrect primitive restart index used</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=98134">Bug 98134</a> - dEQP-GLES31.functional.debug.negative_coverage.get_error.buffer.draw_buffers wants a different GL error code</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=98326">Bug 98326</a> - [dEQP, EGL] pbuffer depth/stencil tests fail</li>
-
-</ul>
-
-
-<h2>Changes</h2>
-
-<p>Axel Davy (4):</p>
-<ul>
-  <li>gallium/util: Really allow aliasing of dst for u_box_union_*</li>
-  <li>st/nine: Fix the calculation of the number of vs inputs</li>
-  <li>st/nine: Fix mistake in Volume9 UnlockBox</li>
-  <li>st/nine: Fix locking CubeTexture surfaces.</li>
-</ul>
-
-<p>Brendan King (1):</p>
-<ul>
-  <li>configure.ac: fix the name of the Wayland Scanner pc file</li>
-</ul>
-
-<p>Brian Paul (1):</p>
-<ul>
-  <li>st/mesa: fix swizzle issue in st_create_sampler_view_from_stobj()</li>
-</ul>
-
-<p>Chad Versace (3):</p>
-<ul>
-  <li>egl: Fix truncation error in _eglParseSyncAttribList64</li>
-  <li>i965/sync: Fix uninitalized usage and leak of mutex</li>
-  <li>egl: Don't advertise unsupported platform extensions</li>
-</ul>
-
-<p>Chuanbo Weng (1):</p>
-<ul>
-  <li>gbm: fix potential NULL deref of mapImage/unmapImage.</li>
-</ul>
-
-<p>Chuck Atkins (1):</p>
-<ul>
-  <li>autoconf: Make header install distinct for various APIs (v2)</li>
-</ul>
-
-<p>Dave Airlie (3):</p>
-<ul>
-  <li>anv: initialise and increment send_sbc</li>
-  <li>anv/wsi: fix apps that acquire multiple images up front</li>
-  <li>Revert "st/vdpau: use linear layout for output surfaces"</li>
-</ul>
-
-<p>Emil Velikov (12):</p>
-<ul>
-  <li>docs: add sha256 checksums for 12.0.3</li>
-  <li>cherry-ignore: add non-applicable i965 commit</li>
-  <li>cherry-ignore: add vaapi encode fix</li>
-  <li>cherry-ignore: add EGL_KHR_debug fix</li>
-  <li>cherry-ignore: add update_renderbuffer_read_surfaces()</li>
-  <li>isl/gen6: correctly check msaa layout samples count</li>
-  <li>egl/x11: don't crash if dri2_dpy-&gt;conn is NULL</li>
-  <li>get-pick-list.sh: Require explicit "12.0" for nominating stable patches</li>
-  <li>automake: don't forget to pick wglext.h in the tarball</li>
-  <li>cherry-ignore: add N/A EGL revert</li>
-  <li>cherry-ignore: add ClientWaitSync fixes</li>
-  <li>Update version to 12.0.4</li>
-</ul>
-
-<p>Eric Anholt (5):</p>
-<ul>
-  <li>travis: Parse configure.ac to pick an updated LIBDRM_VERSION.</li>
-  <li>travis: Update to the Ubuntu Trusty image.</li>
-  <li>travis: Enable vc4 in libdrm to satisfy vc4 test build dependency.</li>
-  <li>travis: Upgrade LLVM dependency to 3.5 and enable LLVM drivers.</li>
-  <li>gallium: Fix install-gallium-links.mk on non-bash /bin/sh</li>
-</ul>
-
-<p>Hans de Goede (1):</p>
-<ul>
-  <li>pipe_loader_sw: Fix fd leak when instantiated via pipe_loader_sw_probe_kms</li>
-</ul>
-
-<p>Ian Romanick (1):</p>
-<ul>
-  <li>glsl: Fix cut-and-paste bug in hierarchical visitor ir_expression::accept</li>
-</ul>
-
-<p>Ilia Mirkin (16):</p>
-<ul>
-  <li>nv30: set usage to staging so that the buffer is allocated in GART</li>
-  <li>a3xx: make sure to actually clamp depth as requested</li>
-  <li>a3xx: make use of software clipping when hw can't handle it</li>
-  <li>a3xx: use window scissor to simulate viewport xy clip</li>
-  <li>main: GL_RGB10_A2UI does not come with GL 3.0/EXT_texture_integer</li>
-  <li>mesa/formatquery: limit ES target support, fix core context support</li>
-  <li>nir: fix definition of pack_uvec2_to_uint</li>
-  <li>gm107/ir: AL2P writes to a predicate register</li>
-  <li>st/mesa: fix is_scissor_enabled when X/Y are negative</li>
-  <li>nvc0/ir: fix overwriting of value backing non-constant gather offset</li>
-  <li>nv50/ir: copy over value's register id when resolving merge of a phi</li>
-  <li>nvc0/ir: fix textureGather with a single offset</li>
-  <li>gm107/ir: fix texturing with indirect samplers</li>
-  <li>gm107/ir: fix bit offset of tex lod setting for indirect texturing</li>
-  <li>nv50,nvc0: avoid reading out of bounds when getting bogus so info</li>
-  <li>nv50/ir: process texture offset sources as regular sources</li>
-</ul>
-
-<p>James Legg (1):</p>
-<ul>
-  <li>radeonsi: Fix primitive restart when index changes</li>
-</ul>
-
-<p>Jason Ekstrand (9):</p>
-<ul>
-  <li>nir/spirv: Swap the argument order for AtomicCompareExchange</li>
-  <li>nir/spirv: Use the correct sources for CompareExchange on images</li>
-  <li>nir/spirv: Break variable decoration handling into a helper</li>
-  <li>nir/spirv: Refactor variable deocration handling</li>
-  <li>nir/spirv/cfg: Handle switches whose break block is a loop continue</li>
-  <li>nir/spirv/cfg: Detect switch_break after loop_break/continue</li>
-  <li>nir: Add a nop intrinsic</li>
-  <li>nir/spirv/cfg: Use a nop intrinsic for tagging the ends of blocks</li>
-  <li>intel/blorp: Rework our usage of ralloc when compiling shaders</li>
-</ul>
-
-<p>Jonathan Gray (3):</p>
-<ul>
-  <li>genxml: add generated headers to EXTRA_DIST</li>
-  <li>mapi: automake: set VISIBILITY_CFLAGS for shared glapi</li>
-  <li>mesa: automake: include mesa_glinterop.h in distfile</li>
-</ul>
-
-<p>Julien Isorce (1):</p>
-<ul>
-  <li>st/va: also honors interlaced preference when providing a video format</li>
-</ul>
-
-<p>Kenneth Graunke (8):</p>
-<ul>
-  <li>nir: Call nir_metadata_preserve from nir_lower_alu_to_scalar().</li>
-  <li>mesa: Expose RESET_NOTIFICATION_STRATEGY with KHR_robustness.</li>
-  <li>i965: Fix missing _NEW_TRANSFORM in Gen8+ 3DSTATE_DS atom.</li>
-  <li>i965: Add missing BRW_NEW_VS_PROG_DATA to 3DSTATE_CLIP.</li>
-  <li>i965: Move BRW_NEW_FRAGMENT_PROGRAM from 3DSTATE_PS to PS_EXTRA.</li>
-  <li>i965: Add missing BRW_NEW_CS_PROG_DATA to compute constant atom.</li>
-  <li>i965: Add missing BRW_CS_PROG_DATA to CS work group surface atom.</li>
-  <li>i965: Fix gl_InvocationID in dual object GS where invocations == 1.</li>
-</ul>
-
-<p>Marek Olšák (12):</p>
-<ul>
-  <li>radeonsi: fix cubemaps viewed as 2D</li>
-  <li>radeonsi: take compute shader and dispatch indirect memory usage into account</li>
-  <li>radeonsi: fix FP64 UBO loads with indirect uniform block indexing</li>
-  <li>mesa: fix glGetFramebufferAttachmentParameteriv w/ on-demand FRONT_BACK alloc</li>
-  <li>radeonsi: fix interpolateAt opcodes for .zw components</li>
-  <li>radeonsi: fix texture border colors for compute shaders</li>
-  <li>radeonsi: disable ReZ</li>
-  <li>gallium/radeon: make sure the address of separate CMASK is aligned properly</li>
-  <li>winsys/amdgpu: fix radeon_surf::macro_tile_index for imported textures</li>
-  <li>egl: use util/macros.h</li>
-  <li>egl: make interop ABI visible again</li>
-  <li>glx: make interop ABI visible again</li>
-</ul>
-
-<p>Mario Kleiner (1):</p>
-<ul>
-  <li>glx: Perform check for valid fbconfig against proper X-Screen.</li>
-</ul>
-
-<p>Martin Peres (2):</p>
-<ul>
-  <li>loader/dri3: add get_dri_screen() to the vtable</li>
-  <li>loader/dri3: import prime buffers in the currently-bound screen</li>
-</ul>
-
-<p>Matt Whitlock (5):</p>
-<ul>
-  <li>egl/android: replace call to dup(2) with fcntl(F_DUPFD_CLOEXEC)</li>
-  <li>gallium/auxiliary: replace call to dup(2) with fcntl(F_DUPFD_CLOEXEC)</li>
-  <li>st/dri: replace calls to dup(2) with fcntl(F_DUPFD_CLOEXEC)</li>
-  <li>st/xa: replace call to dup(2) with fcntl(F_DUPFD_CLOEXEC)</li>
-  <li>gallium/winsys: replace calls to dup(2) with fcntl(F_DUPFD_CLOEXEC)</li>
-</ul>
-
-<p>Max Staudt (1):</p>
-<ul>
-  <li>r300g: Set R300_VAP_CNTL on RSxxx to avoid triangle flickering</li>
-</ul>
-
-<p>Michel Dänzer (1):</p>
-<ul>
-  <li>loader/dri3: Overhaul dri3_update_num_back</li>
-</ul>
-
-<p>Nicholas Bishop (2):</p>
-<ul>
-  <li>gbm: return appropriate error when queryImage() fails</li>
-  <li>st/dri: check pipe_screen-&gt;resource_get_handle() return value</li>
-</ul>
-
-<p>Nicolai Hähnle (10):</p>
-<ul>
-  <li>gallium/radeon: cleanup and fix branch emits</li>
-  <li>st/glsl_to_tgsi: disable on-the-fly peephole for 64-bit operations</li>
-  <li>st/glsl_to_tgsi: simplify translate_tex_offset</li>
-  <li>st/glsl_to_tgsi: fix textureGatherOffset with indirectly loaded offsets</li>
-  <li>st/mesa: fix vertex elements setup for doubles</li>
-  <li>radeonsi: fix indirect loads of 64 bit constants</li>
-  <li>st/glsl_to_tgsi: fix atomic counter addressing</li>
-  <li>st/glsl_to_tgsi: fix block copies of arrays of doubles</li>
-  <li>st/mesa: only set primitive_restart when the restart index is in range</li>
-  <li>radeonsi: fix 64-bit loads from LDS</li>
-</ul>
-
-<p>Samuel Pitoiset (4):</p>
-<ul>
-  <li>nvc0/ir: fix subops for IMAD</li>
-  <li>gk110/ir: fix wrong emission of OP_NOT</li>
-  <li>nvc0: use correct bufctx when invalidating CP textures</li>
-  <li>nvc0/ir: fix emission of IMAD with NEG modifiers</li>
-</ul>
-
-<p>Stencel, Joanna (1):</p>
-<ul>
-  <li>egl/wayland: add missing destroy_window callback</li>
-</ul>
-
-<p>Tapani Pälli (5):</p>
-<ul>
-  <li>egl: stop claiming support for pbuffer + msaa</li>
-  <li>egl/dri2: set max values for pbuffer width and height</li>
-  <li>egl: add check that eglCreateContext gets a valid config</li>
-  <li>mesa: fix error handling in DrawBuffers</li>
-  <li>egl: set preserved behavior for surface only if config supports it</li>
-</ul>
-
-<p>Tim Rowley (1):</p>
-<ul>
-  <li>configure.ac: add llvm inteljitevents component if enabled</li>
-</ul>
-
-<p>Vedran Miletić (1):</p>
-<ul>
-  <li>clover: Fix build against clang SVN &gt;= r273191</li>
-</ul>
-
-<p>Vinson Lee (1):</p>
-<ul>
-  <li>Revert "mesa_glinterop: remove inclusion of GLX header"</li>
-</ul>
-
-
-</div>
-</body>
-</html>

docs/relnotes/12.0.5.html

@@ -1,138 +0,0 @@
-<!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 12.0.5 Release Notes / December 5, 2016</h1>
-
-<p>
-Mesa 12.0.5 is a bug fix release which fixes bugs found since the 12.0.5 release.
-</p>
-<p>
-Mesa 12.0.5 implements the OpenGL 4.3 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 4.3.  OpenGL
-4.3 is <strong>only</strong> available if requested at context creation
-because compatibility contexts are not supported.
-</p>
-
-
-<h2>SHA256 checksums</h2>
-<pre>
-44d08a27d98bfeacd864381189e434d98afbf451689d01f80380dc1d66450e5b  mesa-12.0.5.tar.gz
-2b0a972d8282860a11291c09c3ef01ac45171405951eb21a83c45ed2b4321924  mesa-12.0.5.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=77662">Bug 77662</a> - Fail to render to different faces of depth-stencil cube map</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=97779">Bug 97779</a> - [regression, bisected][BDW, GPU hang] stuck on render ring, always reproducible</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=98415">Bug 98415</a> - Vulkan Driver JSON file contains incorrect field</li>
-
-</ul>
-
-
-<h2>Changes</h2>
-
-<p>Adam Jackson (2):</p>
-<ul>
-  <li>glx/glvnd: Don't modify the dummy slot in the dispatch table</li>
-  <li>glx/glvnd: Fix dispatch function names and indices</li>
-</ul>
-
-<p>Anuj Phogat (1):</p>
-<ul>
-  <li>i965: Fix GPU hang related to multiple render targets and alpha testing</li>
-</ul>
-
-<p>Emil Velikov (4):</p>
-<ul>
-  <li>docs: add release notes for 12.0.4</li>
-  <li>docs: add sha256 checksums for 12.0.4</li>
-  <li>cherry-ignore: add reverted LLVM_LIBDIR patch</li>
-  <li>Update version to 12.0.5</li>
-</ul>
-
-<p>Haixia Shi (1):</p>
-<ul>
-  <li>mesa: change state query return value for RGB565</li>
-</ul>
-
-<p>Jason Ekstrand (3):</p>
-<ul>
-  <li>i965/fs/generator: Don't use the address immediate for MOV_INDIRECT</li>
-  <li>anv/cmd_buffer: Take a command buffer instead of a batch in two helpers</li>
-  <li>anv/cmd_buffer: Enable a CS stall workaround for Sky Lake gt4</li>
-</ul>
-
-<p>Kenneth Graunke (1):</p>
-<ul>
-  <li>intel: Fix pixel shader scratch space allocation on Gen9+ platforms.</li>
-</ul>
-
-<p>Marek Olšák (13):</p>
-<ul>
-  <li>gallium/radeon: fix behavior of GLSL findLSB(0)</li>
-  <li>gallium/radeon: make sure HTILE address is aligned properly</li>
-  <li>radeonsi: fix an assertion failure in si_decompress_sampler_color_textures</li>
-  <li>gallium/radeon: unify viewport emission code</li>
-  <li>gallium/radeon: set VPORT_ZMIN/MAX registers correctly</li>
-  <li>radeonsi: fix gl_PatchVerticesIn for tessellation evaluation shader</li>
-  <li>radeonsi: fix a crash in imageSize for cubemap arrays</li>
-  <li>radeonsi: emit TA_CS_BC_BASE_ADDR on SI only if the kernel allows it</li>
-  <li>gallium/radeon: add support for sharing textures with DCC between processes</li>
-  <li>radeonsi: always set all blend registers</li>
-  <li>radeonsi: set CB_BLEND1_CONTROL.ENABLE for dual source blending</li>
-  <li>radeonsi: disable RB+ blend optimizations for dual source blending</li>
-  <li>radeonsi: silence runtime warnings with LLVM 3.9</li>
-</ul>
-
-<p>Matt Turner (1):</p>
-<ul>
-  <li>anv: Replace "abi_versions" with correct "api_version".</li>
-</ul>
-
-<p>Nanley Chery (1):</p>
-<ul>
-  <li>mesa/fbobject: Update CubeMapFace when reusing textures</li>
-</ul>
-
-<p>Steinar H. Gunderson (1):</p>
-<ul>
-  <li>Fix races during _mesa_HashWalk().</li>
-</ul>
-
-<p>Tim Rowley (3):</p>
-<ul>
-  <li>swr: [rasterizer jitter] cleanup supporting different llvm versions</li>
-  <li>swr: [rasterizer jitter] fix llvm-3.7 compile</li>
-  <li>swr: [rasterizer] add support for llvm-3.9</li>
-</ul>
-
-
-</div>
-</body>
-</html>

docs/relnotes/12.0.6.html

@@ -1,148 +0,0 @@
-<!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 12.0.6 Release Notes / January 23, 2017</h1>
-
-<p>
-Mesa 12.0.6 is a bug fix release which fixes bugs found since the 12.0.5 release.
-</p>
-<p>
-Mesa 12.0.6 implements the OpenGL 4.3 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 4.3.  OpenGL
-4.3 is <strong>only</strong> available if requested at context creation
-because compatibility contexts are not supported.
-</p>
-
-
-<h2>SHA256 checksums</h2>
-<pre>
-65339ba5d76a45225b8b56f9a1da9db15c569e1d163760faa2921da0a8461741  mesa-12.0.6.tar.gz
-7d6da9744c1022a4c2ab6ad01a206984d00443fb691568011d01b3dd97e36448  mesa-12.0.6.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=92234">Bug 92234</a> - [BDW] GPU hang in Shogun2</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=95130">Bug 95130</a> - Derivatives of gl_Color wrong when helper pixels used</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=98329">Bug 98329</a> - [dEQP, EGL, SKL, BDW, BSW] dEQP-EGL.functional.image.render_multiple_contexts.gles2_renderbuffer_depth16_depth_buffer</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=99030">Bug 99030</a> - [HSW, regression] transform feedback fails on Linux 4.8</li>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=99354">Bug 99354</a> - [G71] &quot;Assertion `bkref' failed&quot; reproducible with glmark2</li>
-
-</ul>
-
-<h2>Changes</h2>
-
-<p>Chad Versace (3):</p>
-<ul>
-  <li>i965/mt: Disable aux surfaces after making miptree shareable</li>
-  <li>i965/mt: Disable HiZ when sharing depth buffer externally (v2)</li>
-  <li>anv: Handle vkGetPhysicalDeviceQueueFamilyProperties with count == 0</li>
-</ul>
-
-<p>Emil Velikov (5):</p>
-<ul>
-  <li>docs: add sha256 checksums for 12.0.5</li>
-  <li>get-typod-pick-list.sh: add new script</li>
-  <li>automake: use shared llvm libs for make distcheck</li>
-  <li>egl/wayland: use the destroy_window_callback for swrast</li>
-  <li>Update version to 12.0.6</li>
-</ul>
-
-<p>Fredrik Höglund (1):</p>
-<ul>
-  <li>dri3: Fix MakeCurrent without a default framebuffer</li>
-</ul>
-
-<p>Ilia Mirkin (1):</p>
-<ul>
-  <li>nouveau: take extra push space into account for pushbuf_space calls</li>
-</ul>
-
-<p>Jason Ekstrand (19):</p>
-<ul>
-  <li>spirv/nir: Fix some texture opcode asserts</li>
-  <li>spirv/nir: Add support for shadow samplers that return vec4</li>
-  <li>spirv/nir: Properly handle gather components</li>
-  <li>anv/pipeline: Set binding_table.gather_texture_start</li>
-  <li>nir: Add a helper for determining the type of a texture source</li>
-  <li>nir/lower_tex: Add some helpers for working with tex sources</li>
-  <li>nir/lower_tex: Add support for lowering coordinate offsets</li>
-  <li>i965/nir: Enable NIR lowering of txf and rect offsets</li>
-  <li>i965: Get rid of the do_lower_unnormalized_offsets pass</li>
-  <li>spirv/nir: Don't increment coord_components for array lod queries</li>
-  <li>anv/image: Assert that the image format is actually supported</li>
-  <li>spirv/nir: Move opcode selection higher up in handle_texture</li>
-  <li>spirv/nir: Refactor type handling in handle_texture</li>
-  <li>nir/spirv: Refactor coordinate handling in handle_texture</li>
-  <li>spirv/nir: Handle texture projectors</li>
-  <li>spirv/nir: Add support for ImageQuerySamples</li>
-  <li>anv/device: Return the right error for failed maps</li>
-  <li>anv/device: Implicitly unmap memory objects in FreeMemory</li>
-  <li>anv/descriptor_set: Write the state offset in the surface state free list.</li>
-</ul>
-
-<p>Kenneth Graunke (2):</p>
-<ul>
-  <li>spirv: Move cursor before calling vtn_ssa_value() in phi 2nd pass.</li>
-  <li>i965: Properly flush in hsw_pause_transform_feedback().</li>
-</ul>
-
-<p>Marek Olšák (6):</p>
-<ul>
-  <li>cso: don't release sampler states that are bound</li>
-  <li>radeonsi: always restore sampler states when unbinding sampler views</li>
-  <li>radeonsi: fix incorrect FMASK checking in bind_sampler_states</li>
-  <li>radeonsi: disable CE on SI + AMDGPU</li>
-  <li>radeonsi: disable the constant engine (CE) on Carrizo and Stoney</li>
-  <li>gallium/radeon: fix the draw-calls HUD query</li>
-</ul>
-
-<p>Matt Turner (3):</p>
-<ul>
-  <li>i965/fs: Rename opt_copy_propagate -&gt; opt_copy_propagation.</li>
-  <li>i965/fs: Add unit tests for copy propagation pass.</li>
-  <li>i965/fs: Reject copy propagation into SEL if not min/max.</li>
-</ul>
-
-<p>Michel Dänzer (1):</p>
-<ul>
-  <li>cso: Don't restore nr_samplers in cso_restore_fragment_samplers</li>
-</ul>
-
-<p>Nicolai Hähnle (1):</p>
-<ul>
-  <li>radeonsi: enable WQM in PS prolog when needed</li>
-</ul>
-
-
-</div>
-</body>
-</html>

docs/relnotes/12.1.0.html

@@ -14,15 +14,15 @@
 <iframe src="../contents.html"></iframe>
 <div class="content">
 
-<h1>Mesa 12.0.1 Release Notes / July 8, 2016</h1>
-
-<h1>Mesa 12.0.1 Release Notes / July 8, 2016</h1>
+<h1>Mesa 12.1.0 Release Notes / TBD</h1>
 
 <p>
-Mesa 12.0.1 is a bug fix release which fixes bugs found since the 12.0.1 release.
+Mesa 12.1.0 is a new development release.
+People who are concerned with stability and reliability should stick
+with a previous release or wait for Mesa 12.1.1.
 </p>
 <p>
-Mesa 12.0.1 implements the OpenGL 4.3 API, but the version reported by
+Mesa 12.1.0 implements the OpenGL 4.3 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 4.3.  OpenGL
@@ -33,34 +33,27 @@ because compatibility contexts are not supported.
 
 <h2>SHA256 checksums</h2>
 <pre>
-28dff9c045f4305c96a875a487b9f06c7e88d910511cd6016dbddcd1f53ade0d  mesa-12.0.1.tar.gz
-bab24fb79f78c876073527f515ed871fc9c81d816f66c8a0b051d8d653896389  mesa-12.0.1.tar.xz
+TBD.
 </pre>
 
 
 <h2>New features</h2>
-<p>None</p>
 
+<p>
+Note: some of the new features are only available with certain drivers.
+</p>
+
+<ul>
+<li>GL_ARB_shader_group_vote on nvc0</li>
+</ul>
 
 <h2>Bug fixes</h2>
 
-<ul>
-
-<li><a href="https://bugs.freedesktop.org/show_bug.cgi?id=96864">Bug 96864</a> - Mesa 12.0 radeon build broken</li>
-
-</ul>
-
+TBD.
 
 <h2>Changes</h2>
 
-<p>Emil Velikov (4):</p>
-<ul>
-  <li>docs: add sha256 checksums for 12.0.0</li>
-  <li>radeon: reference the correct cdw/max_dw</li>
-  <li>Update version to 12.0.1</li>
-  <li>docs: add release notes for 12.0.1</li>
-</ul>
-
+TBD.
 
 </div>
 </body>

include/GL/mesa_glinterop.h

@@ -58,8 +58,8 @@ extern "C" {
 #endif
 
 /* Forward declarations to avoid inclusion of GL/glx.h */
-struct _XDisplay;
-struct __GLXcontextRec;
+typedef struct _XDisplay Display;
+typedef struct __GLXcontextRec *GLXContext;
 
 /* Forward declarations to avoid inclusion of EGL/egl.h */
 typedef void *EGLDisplay;
@@ -246,7 +246,7 @@ struct mesa_glinterop_export_out {
  * \return MESA_GLINTEROP_SUCCESS or MESA_GLINTEROP_* != 0 on error
  */
 int
-MesaGLInteropGLXQueryDeviceInfo(struct _XDisplay *dpy, struct __GLXcontextRec *context,
+MesaGLInteropGLXQueryDeviceInfo(Display *dpy, GLXContext context,
                                 struct mesa_glinterop_device_info *out);
 
 
@@ -271,7 +271,7 @@ MesaGLInteropEGLQueryDeviceInfo(EGLDisplay dpy, EGLContext context,
  * \return MESA_GLINTEROP_SUCCESS or MESA_GLINTEROP_* != 0 on error
  */
 int
-MesaGLInteropGLXExportObject(struct _XDisplay *dpy, struct __GLXcontextRec *context,
+MesaGLInteropGLXExportObject(Display *dpy, GLXContext context,
                              struct mesa_glinterop_export_in *in,
                              struct mesa_glinterop_export_out *out);
 
@@ -286,11 +286,11 @@ MesaGLInteropEGLExportObject(EGLDisplay dpy, EGLContext context,
                              struct mesa_glinterop_export_out *out);
 
 
-typedef int (PFNMESAGLINTEROPGLXQUERYDEVICEINFOPROC)(struct _XDisplay *dpy, struct __GLXcontextRec *context,
+typedef int (PFNMESAGLINTEROPGLXQUERYDEVICEINFOPROC)(Display *dpy, GLXContext context,
                                                      struct mesa_glinterop_device_info *out);
 typedef int (PFNMESAGLINTEROPEGLQUERYDEVICEINFOPROC)(EGLDisplay dpy, EGLContext context,
                                                      struct mesa_glinterop_device_info *out);
-typedef int (PFNMESAGLINTEROPGLXEXPORTOBJECTPROC)(struct _XDisplay *dpy, struct __GLXcontextRec *context,
+typedef int (PFNMESAGLINTEROPGLXEXPORTOBJECTPROC)(Display *dpy, GLXContext context,
                                                   struct mesa_glinterop_export_in *in,
                                                   struct mesa_glinterop_export_out *out);
 typedef int (PFNMESAGLINTEROPEGLEXPORTOBJECTPROC)(EGLDisplay dpy, EGLContext context,

include/c99_compat.h

@@ -36,8 +36,8 @@
  */
 #if defined(_MSC_VER)
 
-#  if _MSC_VER < 1800 || (_MSC_FULL_VER < 180031101 && !defined(__clang__))
-#    error "Microsoft Visual Studio 2013 Update 4 or higher required"
+#  if _MSC_VER < 1800
+#    error "Microsoft Visual Studio 2013 or higher required"
 #  endif
 
    /*

install-gallium-links.mk

@@ -13,8 +13,8 @@ all-local : .install-gallium-links
 	fi;							\
 	$(MKDIR_P) $$link_dir;					\
 	file_list="$(dri_LTLIBRARIES:%.la=.libs/%.so)";		\
-	file_list="$$file_list$(egl_LTLIBRARIES:%.la=.libs/%.$(LIB_EXT)*)";	\
-	file_list="$$file_list$(lib_LTLIBRARIES:%.la=.libs/%.$(LIB_EXT)*)";	\
+	file_list+="$(egl_LTLIBRARIES:%.la=.libs/%.$(LIB_EXT)*)";	\
+	file_list+="$(lib_LTLIBRARIES:%.la=.libs/%.$(LIB_EXT)*)";	\
 	for f in $$file_list; do 				\
 		if test -h .libs/$$f; then			\
 			cp -d $$f $$link_dir;			\

src/Makefile.am

@@ -25,13 +25,15 @@ git_sha1.h.tmp:
 	@# a gitlink file if $(top_srcdir) is a submodule checkout or a linked
 	@# worktree.
 	@# If we are building from a release tarball copy the bundled header.
-	@touch git_sha1.h.tmp
 	@if test -e $(top_srcdir)/.git; then \
 		if which git > /dev/null; then \
 		    git --git-dir=$(top_srcdir)/.git log -n 1 --oneline | \
 			sed 's/^\([^ ]*\) .*/#define MESA_GIT_SHA1 "git-\1"/' \
 			> git_sha1.h.tmp ; \
 		fi \
+	else \
+		cp $(srcdir)/git_sha1.h git_sha1.h.tmp ;\
+		chmod u+w git_sha1.h.tmp; \
 	fi
 
 git_sha1.h: git_sha1.h.tmp
@@ -43,34 +45,15 @@ git_sha1.h: git_sha1.h.tmp
 	fi
 
 BUILT_SOURCES = git_sha1.h
-CLEANFILES = $(BUILT_SOURCES)
+
+# We want to keep the srcdir file since we need it on rebuild from tarball.
+# At the same time `make distclean' gets angry at us if we don't cleanup the
+# builddir one.
+distclean-local:
+	( test $(top_srcdir) != $(top_builddir) && rm $(builddir)/git_sha1.h ) || true
 
 SUBDIRS = . gtest util mapi/glapi/gen mapi
 
-if HAVE_OPENGL
-gldir = $(includedir)/GL
-gl_HEADERS = \
-  $(top_srcdir)/include/GL/gl.h \
-  $(top_srcdir)/include/GL/glext.h \
-  $(top_srcdir)/include/GL/glcorearb.h \
-  $(top_srcdir)/include/GL/gl_mangle.h
-endif
-
-if HAVE_GLX
-glxdir = $(includedir)/GL
-glx_HEADERS = \
-  $(top_srcdir)/include/GL/glx.h \
-  $(top_srcdir)/include/GL/glxext.h \
-  $(top_srcdir)/include/GL/glx_mangle.h
-pkgconfigdir = $(libdir)/pkgconfig
-pkgconfig_DATA = mesa/gl.pc
-endif
-
-if HAVE_COMMON_OSMESA
-osmesadir = $(includedir)/GL
-osmesa_HEADERS = $(top_srcdir)/include/GL/osmesa.h
-endif
-
 # include only conditionally ?
 SUBDIRS += compiler
 
@@ -117,8 +100,7 @@ SUBDIRS += gallium
 endif
 
 EXTRA_DIST = \
-	getopt hgl SConscript \
-	$(top_srcdir)/include/GL/mesa_glinterop.h
+	getopt hgl SConscript git_sha1.h
 
 AM_CFLAGS = $(VISIBILITY_CFLAGS)
 AM_CXXFLAGS = $(VISIBILITY_CXXFLAGS)

src/compiler/Makefile.sources

@@ -231,11 +231,11 @@ NIR_FILES = \
 	nir/nir_phi_builder.c \
 	nir/nir_phi_builder.h \
 	nir/nir_print.c \
-	nir/nir_propagate_invariant.c \
 	nir/nir_remove_dead_variables.c \
 	nir/nir_repair_ssa.c \
 	nir/nir_search.c \
 	nir/nir_search.h \
+	nir/nir_search_helpers.h \
 	nir/nir_split_var_copies.c \
 	nir/nir_sweep.c \
 	nir/nir_to_ssa.c \

src/compiler/glsl/ast_to_hir.cpp

@@ -2278,10 +2278,10 @@ precision_qualifier_allowed(const glsl_type *type)
     * From this, we infer that GLSL 1.30 (and later) should allow precision
     * qualifiers on sampler types just like float and integer types.
     */
-   const glsl_type *const t = type->without_array();
-
-   return (t->is_float() || t->is_integer() || t->contains_opaque()) &&
-          !t->is_record();
+   return (type->is_float()
+       || type->is_integer()
+       || type->contains_opaque())
+       && !type->without_array()->is_record();
 }
 
 const glsl_type *
@@ -3393,7 +3393,7 @@ apply_layout_qualifier_to_variable(const struct ast_type_qualifier *qual,
                 (qual_component + components - 1) > 3) {
                _mesa_glsl_error(loc, state, "component overflow (%u > 3)",
                                 (qual_component + components - 1));
-            } else if (qual_component == 1 && type->is_double()) {
+            } else if (qual_component == 1 && type->is_64bit()) {
                /* We don't bother checking for 3 as it should be caught by the
                 * overflow check above.
                 */
@@ -4697,14 +4697,6 @@ ast_declarator_list::hir(exec_list *instructions,
       apply_layout_qualifier_to_variable(&this->type->qualifier, var, state,
                                          &loc);
 
-      if ((var->data.mode == ir_var_auto || var->data.mode == ir_var_temporary)
-          && (var->type->is_numeric() || var->type->is_boolean())
-          && state->zero_init) {
-         const ir_constant_data data = {0};
-         var->data.has_initializer = true;
-         var->constant_initializer = new(var) ir_constant(var->type, &data);
-      }
-
       if (this->type->qualifier.flags.q.invariant) {
          if (!is_varying_var(var, state->stage)) {
             _mesa_glsl_error(&loc, state,
@@ -5002,8 +4994,13 @@ ast_declarator_list::hir(exec_list *instructions,
          state->check_precision_qualifiers_allowed(&loc);
       }
 
-      if (this->type->qualifier.precision != ast_precision_none &&
-          !precision_qualifier_allowed(var->type)) {
+
+      /* If a precision qualifier is allowed on a type, it is allowed on
+       * an array of that type.
+       */
+      if (!(this->type->qualifier.precision == ast_precision_none
+          || precision_qualifier_allowed(var->type->without_array()))) {
+
          _mesa_glsl_error(&loc, state,
                           "precision qualifiers apply only to floating point"
                           ", integer and opaque types");
@@ -6846,7 +6843,7 @@ ast_process_struct_or_iface_block_members(exec_list *instructions,
             }
          } else {
             if (layout && layout->flags.q.explicit_xfb_offset) {
-               unsigned align = field_type->is_double() ? 8 : 4;
+               unsigned align = field_type->is_64bit() ? 8 : 4;
                fields[i].offset = glsl_align(block_xfb_offset, align);
                block_xfb_offset +=
                   MAX2(xfb_stride, (int) (4 * field_type->component_slots()));

src/compiler/glsl/builtin_functions.cpp

@@ -528,6 +528,12 @@ barrier_supported(const _mesa_glsl_parse_state *state)
           state->stage == MESA_SHADER_TESS_CTRL;
 }
 
+static bool
+vote(const _mesa_glsl_parse_state *state)
+{
+   return state->ARB_shader_group_vote_enable;
+}
+
 /** @} */
 
 /******************************************************************************/
@@ -853,6 +859,8 @@ private:
    ir_function_signature *_shader_clock(builtin_available_predicate avail,
                                         const glsl_type *type);
 
+   ir_function_signature *_vote(enum ir_expression_operation opcode);
+
 #undef B0
 #undef B1
 #undef B2
@@ -2935,6 +2943,10 @@ builtin_builder::create_builtins()
                               glsl_type::uvec2_type),
                 NULL);
 
+   add_function("anyInvocationARB", _vote(ir_unop_vote_any), NULL);
+   add_function("allInvocationsARB", _vote(ir_unop_vote_all), NULL);
+   add_function("allInvocationsEqualARB", _vote(ir_unop_vote_eq), NULL);
+
 #undef F
 #undef FI
 #undef FIUD
@@ -5576,6 +5588,16 @@ builtin_builder::_shader_clock(builtin_available_predicate avail,
    return sig;
 }
 
+ir_function_signature *
+builtin_builder::_vote(enum ir_expression_operation opcode)
+{
+   ir_variable *value = in_var(glsl_type::bool_type, "value");
+
+   MAKE_SIG(glsl_type::bool_type, vote, 1, value);
+   body.emit(ret(expr(opcode, value)));
+   return sig;
+}
+
 /** @} */
 
 /******************************************************************************/

src/compiler/glsl/builtin_variables.cpp

@@ -37,11 +37,6 @@ static const struct gl_builtin_uniform_element gl_NumSamples_elements[] = {
    {NULL, {STATE_NUM_SAMPLES, 0, 0}, SWIZZLE_XXXX}
 };
 
-/* only for TCS */
-static const struct gl_builtin_uniform_element gl_PatchVerticesIn_elements[] = {
-   {NULL, {STATE_INTERNAL, STATE_TCS_PATCH_VERTICES_IN}, SWIZZLE_XXXX}
-};
-
 static const struct gl_builtin_uniform_element gl_DepthRange_elements[] = {
    {"near", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_XXXX},
    {"far", {STATE_DEPTH_RANGE, 0, 0}, SWIZZLE_YYYY},
@@ -239,7 +234,6 @@ static const struct gl_builtin_uniform_element gl_NormalMatrix_elements[] = {
 #define STATEVAR(name) {#name, name ## _elements, ARRAY_SIZE(name ## _elements)}
 
 static const struct gl_builtin_uniform_desc _mesa_builtin_uniform_desc[] = {
-   STATEVAR(gl_PatchVerticesIn),
    STATEVAR(gl_NumSamples),
    STATEVAR(gl_DepthRange),
    STATEVAR(gl_ClipPlane),
@@ -1035,14 +1029,9 @@ void
 builtin_variable_generator::generate_tcs_special_vars()
 {
    add_system_value(SYSTEM_VALUE_PRIMITIVE_ID, int_t, "gl_PrimitiveID");
+   add_system_value(SYSTEM_VALUE_VERTICES_IN, int_t, "gl_PatchVerticesIn");
    add_system_value(SYSTEM_VALUE_INVOCATION_ID, int_t, "gl_InvocationID");
 
-   if (state->ctx->Const.LowerTCSPatchVerticesIn) {
-      add_uniform(int_t, "gl_PatchVerticesIn");
-   } else {
-      add_system_value(SYSTEM_VALUE_VERTICES_IN, int_t, "gl_PatchVerticesIn");
-   }
-
    add_output(VARYING_SLOT_TESS_LEVEL_OUTER, array(float_t, 4),
               "gl_TessLevelOuter")->data.patch = 1;
    add_output(VARYING_SLOT_TESS_LEVEL_INNER, array(float_t, 2),

src/compiler/glsl/glcpp/glcpp-parse.y

@@ -278,34 +278,10 @@ control_line_success:
 	HASH_TOKEN DEFINE_TOKEN define
 |	HASH_TOKEN UNDEF IDENTIFIER NEWLINE {
 		macro_t *macro;
-
-                /* Section 3.4 (Preprocessor) of the GLSL ES 3.00 spec says:
-                 *
-                 *    It is an error to undefine or to redefine a built-in
-                 *    (pre-defined) macro name.
-                 *
-                 * The GLSL ES 1.00 spec does not contain this text.
-                 *
-                 * Section 3.3 (Preprocessor) of the GLSL 1.30 spec says:
-                 *
-                 *    #define and #undef functionality are defined as is
-                 *    standard for C++ preprocessors for macro definitions
-                 *    both with and without macro parameters.
-                 *
-                 * At least as far as I can tell GCC allow '#undef __FILE__'.
-                 * Furthermore, there are desktop OpenGL conformance tests
-                 * that expect '#undef __VERSION__' and '#undef
-                 * GL_core_profile' to work.
-                 *
-                 * Only disallow #undef of pre-defined macros on GLSL ES >=
-                 * 3.00 shaders.
-                 */
-		if (parser->is_gles &&
-                    parser->version >= 300 &&
-                    (strcmp("__LINE__", $3) == 0
-                     || strcmp("__FILE__", $3) == 0
-                     || strcmp("__VERSION__", $3) == 0
-                     || strncmp("GL_", $3, 3) == 0))
+		if (strcmp("__LINE__", $3) == 0
+		    || strcmp("__FILE__", $3) == 0
+		    || strcmp("__VERSION__", $3) == 0
+		    || strncmp("GL_", $3, 3) == 0)
 			glcpp_error(& @1, parser, "Built-in (pre-defined)"
 				    " macro names cannot be undefined.");
 
@@ -420,13 +396,13 @@ control_line_success:
 		_glcpp_parser_skip_stack_pop (parser, & @1);
 	} NEWLINE
 |	HASH_TOKEN VERSION_TOKEN integer_constant NEWLINE {
-		if (parser->version != 0) {
+		if (parser->version_resolved) {
 			glcpp_error(& @1, parser, "#version must appear on the first line");
 		}
 		_glcpp_parser_handle_version_declaration(parser, $3, NULL, true);
 	}
 |	HASH_TOKEN VERSION_TOKEN integer_constant IDENTIFIER NEWLINE {
-		if (parser->version != 0) {
+		if (parser->version_resolved) {
 			glcpp_error(& @1, parser, "#version must appear on the first line");
 		}
 		_glcpp_parser_handle_version_declaration(parser, $3, $4, true);
@@ -1369,7 +1345,7 @@ glcpp_parser_create(const struct gl_extensions *extensions, gl_api api)
 
    parser->extensions = extensions;
    parser->api = api;
-   parser->version = 0;
+   parser->version_resolved = false;
 
    parser->has_new_line_number = 0;
    parser->new_line_number = 1;
@@ -2305,10 +2281,10 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio
 {
    const struct gl_extensions *extensions = parser->extensions;
 
-   if (parser->version != 0)
+   if (parser->version_resolved)
       return;
 
-   parser->version = version;
+   parser->version_resolved = true;
 
    add_builtin_define (parser, "__VERSION__", version);
 
@@ -2491,6 +2467,9 @@ _glcpp_parser_handle_version_declaration(glcpp_parser_t *parser, intmax_t versio
 
          if (extensions->ARB_cull_distance)
             add_builtin_define(parser, "GL_ARB_cull_distance", 1);
+
+         if (extensions->ARB_shader_group_vote)
+            add_builtin_define(parser, "GL_ARB_shader_group_vote", 1);
       }
    }
 

src/compiler/glsl/glcpp/glcpp.h

@@ -196,7 +196,7 @@ struct glcpp_parser {
 	int error;
 	const struct gl_extensions *extensions;
 	gl_api api;
-	unsigned version;
+	bool version_resolved;
 	bool has_new_line_number;
 	int new_line_number;
 	bool has_new_source_number;

src/compiler/glsl/glcpp/tests/120-undef-builtin.c

@@ -1,4 +1,3 @@
-#version 300 es
 #undef __LINE__
 #undef __FILE__
 #undef __VERSION__

src/compiler/glsl/glcpp/tests/120-undef-builtin.c.expected

@@ -1,7 +1,6 @@
+0:1(1): preprocessor error: Built-in (pre-defined) macro names cannot be undefined.
 0:2(1): preprocessor error: Built-in (pre-defined) macro names cannot be undefined.
 0:3(1): preprocessor error: Built-in (pre-defined) macro names cannot be undefined.
-0:4(1): preprocessor error: Built-in (pre-defined) macro names cannot be undefined.
-#version 300 es
 
 
 

src/compiler/glsl/glcpp/tests/147-undef-builtin-allowed.c

@@ -1,4 +0,0 @@
-#version 110
-#undef __LINE__
-#undef __FILE__
-#undef __VERSION__

src/compiler/glsl/glcpp/tests/147-undef-builtin-allowed.c.expected

@@ -1,4 +0,0 @@
-#version 110
-
-
-

src/compiler/glsl/glsl_lexer.ll

@@ -348,10 +348,10 @@ isampler2DMSArray  KEYWORD_WITH_ALT(150, 300, 150, 320, yyextra->ARB_texture_mul
 usampler2DMSArray  KEYWORD_WITH_ALT(150, 300, 150, 320, yyextra->ARB_texture_multisample_enable || yyextra->OES_texture_storage_multisample_2d_array_enable, USAMPLER2DMSARRAY);
 
    /* keywords available with ARB_texture_cube_map_array_enable extension on desktop GLSL */
-samplerCubeArray   KEYWORD_WITH_ALT(400, 310, 400, 0, yyextra->ARB_texture_cube_map_array_enable, SAMPLERCUBEARRAY);
-isamplerCubeArray KEYWORD_WITH_ALT(400, 310, 400, 0, yyextra->ARB_texture_cube_map_array_enable, ISAMPLERCUBEARRAY);
-usamplerCubeArray KEYWORD_WITH_ALT(400, 310, 400, 0, yyextra->ARB_texture_cube_map_array_enable, USAMPLERCUBEARRAY);
-samplerCubeArrayShadow   KEYWORD_WITH_ALT(400, 310, 400, 0, yyextra->ARB_texture_cube_map_array_enable, SAMPLERCUBEARRAYSHADOW);
+samplerCubeArray   KEYWORD_WITH_ALT(400, 0, 400, 0, yyextra->ARB_texture_cube_map_array_enable, SAMPLERCUBEARRAY);
+isamplerCubeArray KEYWORD_WITH_ALT(400, 0, 400, 0, yyextra->ARB_texture_cube_map_array_enable, ISAMPLERCUBEARRAY);
+usamplerCubeArray KEYWORD_WITH_ALT(400, 0, 400, 0, yyextra->ARB_texture_cube_map_array_enable, USAMPLERCUBEARRAY);
+samplerCubeArrayShadow   KEYWORD_WITH_ALT(400, 0, 400, 0, yyextra->ARB_texture_cube_map_array_enable, SAMPLERCUBEARRAYSHADOW);
 
 samplerExternalOES		{
 			  if (yyextra->OES_EGL_image_external_enable)

src/compiler/glsl/glsl_parser.yy

@@ -1784,10 +1784,8 @@ type_qualifier:
        * variables. As only outputs can be declared as invariant, an invariant
        * output from one shader stage will still match an input of a subsequent
        * stage without the input being declared as invariant."
-       *
-       * On the desktop side, this text first appears in GLSL 4.30.
        */
-      if (state->is_version(430, 300) && $$.flags.q.in)
+      if (state->es_shader && state->language_version >= 300 && $$.flags.q.in)
          _mesa_glsl_error(&@1, state, "invariant qualifiers cannot be used with shader inputs");
    }
    | interpolation_qualifier type_qualifier

src/compiler/glsl/glsl_parser_extras.cpp

@@ -74,7 +74,6 @@ _mesa_glsl_parse_state::_mesa_glsl_parse_state(struct gl_context *_ctx,
    /* Set default language version and extensions */
    this->language_version = 110;
    this->forced_language_version = ctx->Const.ForceGLSLVersion;
-   this->zero_init = ctx->Const.GLSLZeroInit;
    this->es_shader = false;
    this->ARB_texture_rectangle_enable = true;
 
@@ -595,6 +594,7 @@ static const _mesa_glsl_extension _mesa_glsl_supported_extensions[] = {
    EXT(ARB_shader_bit_encoding,          true,  false,     ARB_shader_bit_encoding),
    EXT(ARB_shader_clock,                 true,  false,     ARB_shader_clock),
    EXT(ARB_shader_draw_parameters,       true,  false,     ARB_shader_draw_parameters),
+   EXT(ARB_shader_group_vote,            true,  false,     ARB_shader_group_vote),
    EXT(ARB_shader_image_load_store,      true,  false,     ARB_shader_image_load_store),
    EXT(ARB_shader_image_size,            true,  false,     ARB_shader_image_size),
    EXT(ARB_shader_precision,             true,  false,     ARB_shader_precision),
@@ -1603,6 +1603,7 @@ ast_struct_specifier::ast_struct_specifier(const char *identifier,
    name = identifier;
    this->declarations.push_degenerate_list_at_head(&declarator_list->link);
    is_declaration = true;
+   layout = NULL;
 }
 
 void ast_subroutine_list::print(void) const

src/compiler/glsl/glsl_parser_extras.h

@@ -306,7 +306,6 @@ struct _mesa_glsl_parse_state {
    bool es_shader;
    unsigned language_version;
    unsigned forced_language_version;
-   bool zero_init;
    gl_shader_stage stage;
 
    /**
@@ -576,6 +575,8 @@ struct _mesa_glsl_parse_state {
    bool ARB_shader_clock_warn;
    bool ARB_shader_draw_parameters_enable;
    bool ARB_shader_draw_parameters_warn;
+   bool ARB_shader_group_vote_enable;
+   bool ARB_shader_group_vote_warn;
    bool ARB_shader_image_load_store_enable;
    bool ARB_shader_image_load_store_warn;
    bool ARB_shader_image_size_enable;

src/compiler/glsl/glsl_to_nir.cpp

@@ -1284,9 +1284,6 @@ nir_visitor::visit(ir_expression *ir)
           intrin->intrinsic == nir_intrinsic_interp_var_at_sample)
          intrin->src[0] = nir_src_for_ssa(evaluate_rvalue(ir->operands[1]));
 
-      if (intrin->intrinsic == nir_intrinsic_interp_var_at_offset)
-         shader->info.uses_interp_var_at_offset = true;
-
       unsigned bit_size =  glsl_get_bit_size(deref->type);
       add_instr(&intrin->instr, deref->type->vector_elements, bit_size);
 

src/compiler/glsl/ir.cpp

@@ -341,6 +341,12 @@ ir_expression::ir_expression(int op, ir_rvalue *op0)
       this->type = glsl_type::int_type;
       break;
 
+   case ir_unop_vote_any:
+   case ir_unop_vote_all:
+   case ir_unop_vote_eq:
+      this->type = glsl_type::bool_type;
+      break;
+
    default:
       assert(!"not reached: missing automatic type setup for ir_expression");
       this->type = op0->type;
@@ -563,6 +569,9 @@ static const char *const operator_strs[] = {
    "interpolate_at_centroid",
    "get_buffer_size",
    "ssbo_unsized_array_length",
+   "vote_any",
+   "vote_all",
+   "vote_eq",
    "+",
    "-",
    "*",

src/compiler/glsl/ir.h

@@ -537,6 +537,10 @@ public:
       return this->interface_type;
    }
 
+   enum glsl_interface_packing get_interface_type_packing() const
+   {
+     return this->interface_type->get_interface_packing();
+   }
    /**
     * Get the max_ifc_array_access pointer
     *
@@ -586,13 +590,6 @@ public:
       return this->u.state_slots;
    }
 
-   inline bool is_interpolation_flat() const
-   {
-      return this->data.interpolation == INTERP_QUALIFIER_FLAT ||
-             this->type->contains_integer() ||
-             this->type->contains_double();
-   }
-
    inline bool is_name_ralloced() const
    {
       return this->name != ir_variable::tmp_name;
@@ -1484,10 +1481,17 @@ enum ir_expression_operation {
     */
    ir_unop_ssbo_unsized_array_length,
 
+   /**
+    * Vote among threads on the value of the boolean argument.
+    */
+   ir_unop_vote_any,
+   ir_unop_vote_all,
+   ir_unop_vote_eq,
+
    /**
     * A sentinel marking the last of the unary operations.
     */
-   ir_last_unop = ir_unop_ssbo_unsized_array_length,
+   ir_last_unop = ir_unop_vote_eq,
 
    ir_binop_add,
    ir_binop_sub,

src/compiler/glsl/ir_hv_accept.cpp

@@ -147,7 +147,7 @@ ir_expression::accept(ir_hierarchical_visitor *v)
 	 goto done;
 
       case visit_stop:
-	 return visit_stop;
+	 return s;
       }
    }
 

src/compiler/glsl/ir_set_program_inouts.cpp

@@ -119,7 +119,7 @@ mark(struct gl_program *prog, ir_variable *var, int offset, int len,
 
          /* double inputs read is only for vertex inputs */
          if (stage == MESA_SHADER_VERTEX &&
-             var->type->without_array()->is_dual_slot_double())
+             var->type->without_array()->is_dual_slot())
             prog->DoubleInputsRead |= bitfield;
 
          if (stage == MESA_SHADER_FRAGMENT) {
@@ -260,19 +260,15 @@ ir_set_program_inouts_visitor::try_mark_partial_variable(ir_variable *var,
     * lowering passes (do_vec_index_to_swizzle() gets rid of indexing into
     * vectors, and lower_packed_varyings() gets rid of structs that occur in
     * varyings).
-    *
-    * However, we don't use varying packing in all cases - tessellation
-    * shaders bypass it.  This means we'll see varying structs and arrays
-    * of structs here.  For now, we just give up so the caller marks the
-    * entire variable as used.
     */
    if (!(type->is_matrix() ||
         (type->is_array() &&
          (type->fields.array->is_numeric() ||
           type->fields.array->is_boolean())))) {
+      assert(!"Unexpected indexing in ir_set_program_inouts");
 
-      /* If we don't know how to handle this case, give up and let the
-       * caller mark the whole variable as used.
+      /* For safety in release builds, in case we ever encounter unexpected
+       * indexing, give up and let the caller mark the whole variable as used.
        */
       return false;
    }
@@ -310,7 +306,7 @@ ir_set_program_inouts_visitor::try_mark_partial_variable(ir_variable *var,
    /* double element width for double types that takes two slots */
    if (this->shader_stage != MESA_SHADER_VERTEX ||
        var->data.mode != ir_var_shader_in) {
-      if (type->without_array()->is_dual_slot_double())
+      if (type->without_array()->is_dual_slot())
 	 elem_width *= 2;
    }
 

src/compiler/glsl/ir_validate.cpp

@@ -453,6 +453,14 @@ ir_validate::visit_leave(ir_expression *ir)
       assert(ir->operands[0]->type->base_type == GLSL_TYPE_SUBROUTINE);
       assert(ir->type->base_type == GLSL_TYPE_INT);
       break;
+
+   case ir_unop_vote_any:
+   case ir_unop_vote_all:
+   case ir_unop_vote_eq:
+      assert(ir->type == glsl_type::bool_type);
+      assert(ir->operands[0]->type == glsl_type::bool_type);
+      break;
+
    case ir_binop_add:
    case ir_binop_sub:
    case ir_binop_mul:

src/compiler/glsl/link_uniform_block_active_visitor.cpp

@@ -167,8 +167,7 @@ link_uniform_block_active_visitor::visit(ir_variable *var)
     *     also considered active, even if no member of the block is
     *     referenced."
     */
-   if (var->get_interface_type()->interface_packing ==
-       GLSL_INTERFACE_PACKING_PACKED)
+   if (var->get_interface_type_packing() == GLSL_INTERFACE_PACKING_PACKED)
       return visit_continue;
 
    /* Process the block.  Bail if there was an error.
@@ -258,8 +257,7 @@ link_uniform_block_active_visitor::visit_enter(ir_dereference_array *ir)
     * std140 layout qualifier, all its instances have been already marked
     * as used in link_uniform_block_active_visitor::visit(ir_variable *).
     */
-   if (var->get_interface_type()->interface_packing ==
-       GLSL_INTERFACE_PACKING_PACKED) {
+   if (var->get_interface_type_packing() == GLSL_INTERFACE_PACKING_PACKED) {
       b->var = var;
       process_arrays(this->mem_ctx, ir, b);
    }

src/compiler/glsl/link_uniform_blocks.cpp

@@ -70,7 +70,7 @@ private:
    }
 
    virtual void enter_record(const glsl_type *type, const char *,
-                             bool row_major, const unsigned packing) {
+                             bool row_major, const enum glsl_interface_packing packing) {
       assert(type->is_record());
       if (packing == GLSL_INTERFACE_PACKING_STD430)
          this->offset = glsl_align(
@@ -81,7 +81,7 @@ private:
    }
 
    virtual void leave_record(const glsl_type *type, const char *,
-                             bool row_major, const unsigned packing) {
+                             bool row_major, const enum glsl_interface_packing packing) {
       assert(type->is_record());
 
       /* If this is the last field of a structure, apply rule #9.  The
@@ -106,7 +106,7 @@ private:
 
    virtual void visit_field(const glsl_type *type, const char *name,
                             bool row_major, const glsl_type *,
-                            const unsigned packing,
+                            const enum glsl_interface_packing packing,
                             bool last_field)
    {
       assert(this->index < this->num_variables);

src/compiler/glsl/link_uniform_initializers.cpp

@@ -222,7 +222,7 @@ set_uniform_initializer(void *mem_ctx, gl_shader_program *prog,
 	 val->array_elements[0]->type->base_type;
       const unsigned int elements = val->array_elements[0]->type->components();
       unsigned int idx = 0;
-      unsigned dmul = (base_type == GLSL_TYPE_DOUBLE) ? 2 : 1;
+      unsigned dmul = glsl_base_type_is_64bit(base_type) ? 2 : 1;
 
       assert(val->type->length >= storage->array_elements);
       for (unsigned int i = 0; i < storage->array_elements; i++) {

src/compiler/glsl/link_uniforms.cpp

@@ -65,7 +65,7 @@ program_resource_visitor::process(const glsl_type *type, const char *name)
 
    unsigned record_array_count = 1;
    char *name_copy = ralloc_strdup(NULL, name);
-   unsigned packing = type->interface_packing;
+   enum glsl_interface_packing packing = type->get_interface_packing();
 
    recursion(type, &name_copy, strlen(name), false, NULL, packing, false,
              record_array_count, NULL);
@@ -79,9 +79,9 @@ program_resource_visitor::process(ir_variable *var)
    const bool row_major =
       var->data.matrix_layout == GLSL_MATRIX_LAYOUT_ROW_MAJOR;
 
-   const unsigned packing = var->get_interface_type() ?
-      var->get_interface_type()->interface_packing :
-      var->type->interface_packing;
+   const enum glsl_interface_packing packing = var->get_interface_type() ?
+      var->get_interface_type_packing() :
+      var->type->get_interface_packing();
 
    const glsl_type *t =
       var->data.from_named_ifc_block ? var->get_interface_type() : var->type;
@@ -116,7 +116,7 @@ void
 program_resource_visitor::recursion(const glsl_type *t, char **name,
                                     size_t name_length, bool row_major,
                                     const glsl_type *record_type,
-                                    const unsigned packing,
+                                    const enum glsl_interface_packing packing,
                                     bool last_field,
                                     unsigned record_array_count,
                                     const glsl_struct_field *named_ifc_member)
@@ -228,7 +228,7 @@ void
 program_resource_visitor::visit_field(const glsl_type *type, const char *name,
                                       bool row_major,
                                       const glsl_type *,
-                                      const unsigned,
+                                      const enum glsl_interface_packing,
                                       bool /* last_field */)
 {
    visit_field(type, name, row_major);
@@ -243,13 +243,13 @@ program_resource_visitor::visit_field(const glsl_struct_field *field)
 
 void
 program_resource_visitor::enter_record(const glsl_type *, const char *, bool,
-                                       const unsigned)
+                                       const enum glsl_interface_packing)
 {
 }
 
 void
 program_resource_visitor::leave_record(const glsl_type *, const char *, bool,
-                                       const unsigned)
+                                       const enum glsl_interface_packing)
 {
 }
 
@@ -402,7 +402,9 @@ private:
        * uniforms.
        */
       this->num_active_uniforms++;
-      this->num_values += values;
+
+      if(!is_gl_identifier(name) && !is_shader_storage)
+         this->num_values += values;
    }
 
    struct string_to_uint_map *hidden_map;
@@ -660,7 +662,7 @@ private:
    }
 
    virtual void enter_record(const glsl_type *type, const char *,
-                             bool row_major, const unsigned packing) {
+                             bool row_major, const enum glsl_interface_packing packing) {
       assert(type->is_record());
       if (this->buffer_block_index == -1)
          return;
@@ -673,7 +675,7 @@ private:
    }
 
    virtual void leave_record(const glsl_type *type, const char *,
-                             bool row_major, const unsigned packing) {
+                             bool row_major, const enum glsl_interface_packing packing) {
       assert(type->is_record());
       if (this->buffer_block_index == -1)
          return;
@@ -687,7 +689,7 @@ private:
 
    virtual void visit_field(const glsl_type *type, const char *name,
                             bool row_major, const glsl_type * /* record_type */,
-                            const unsigned packing,
+                            const enum glsl_interface_packing packing,
                             bool /* last_field */)
    {
       assert(!type->without_array()->is_record());
@@ -762,13 +764,14 @@ private:
          current_var->data.how_declared == ir_var_hidden;
       this->uniforms[id].builtin = is_gl_identifier(name);
 
-      /* Do not assign storage if the uniform is builtin */
-      if (!this->uniforms[id].builtin)
-         this->uniforms[id].storage = this->values;
-
       this->uniforms[id].is_shader_storage =
          current_var->is_in_shader_storage_block();
 
+      /* Do not assign storage if the uniform is builtin */
+      if (!this->uniforms[id].builtin &&
+          !this->uniforms[id].is_shader_storage)
+         this->uniforms[id].storage = this->values;
+
       if (this->buffer_block_index != -1) {
          this->uniforms[id].block_index = this->buffer_block_index;
 
@@ -819,7 +822,9 @@ private:
          this->uniforms[id].row_major = false;
       }
 
-      this->values += values_for_type(type);
+      if (!this->uniforms[id].builtin &&
+          !this->uniforms[id].is_shader_storage)
+         this->values += values_for_type(type);
    }
 
    /**
@@ -1251,7 +1256,8 @@ link_assign_uniform_locations(struct gl_shader_program *prog,
 
 #ifndef NDEBUG
    for (unsigned i = 0; i < num_uniforms; i++) {
-      assert(uniforms[i].storage != NULL || uniforms[i].builtin);
+      assert(uniforms[i].storage != NULL || uniforms[i].builtin ||
+             uniforms[i].is_shader_storage);
    }
 
    assert(parcel.values == data_end);

src/compiler/glsl/link_varyings.cpp

@@ -308,25 +308,7 @@ cross_validate_types_and_qualifiers(struct gl_shader_program *prog,
       return;
    }
 
-   /* The GLSL 4.30 and GLSL ES 3.00 specifications say:
-    *
-    *    "As only outputs need be declared with invariant, an output from
-    *     one shader stage will still match an input of a subsequent stage
-    *     without the input being declared as invariant."
-    *
-    * while GLSL 4.20 says:
-    *
-    *    "For variables leaving one shader and coming into another shader,
-    *     the invariant keyword has to be used in both shaders, or a link
-    *     error will result."
-    *
-    * and GLSL ES 1.00 section 4.6.4 "Invariance and Linking" says:
-    *
-    *    "The invariance of varyings that are declared in both the vertex
-    *     and fragment shaders must match."
-    */
-   if (input->data.invariant != output->data.invariant &&
-       prog->Version < (prog->IsES ? 300 : 430)) {
+   if (!prog->IsES && input->data.invariant != output->data.invariant) {
       linker_error(prog,
                    "%s shader output `%s' %s invariant qualifier, "
                    "but %s shader input %s invariant qualifier\n",
@@ -415,15 +397,15 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
          unsigned slot_limit = idx + num_elements;
          unsigned last_comp;
 
-         if (var->type->without_array()->is_record()) {
+         if (type->without_array()->is_record()) {
             /* The component qualifier can't be used on structs so just treat
              * all component slots as used.
              */
             last_comp = 4;
          } else {
-            unsigned dmul = var->type->is_double() ? 2 : 1;
+            unsigned dmul = type->without_array()->is_64bit() ? 2 : 1;
             last_comp = var->data.location_frac +
-               var->type->without_array()->vector_elements * dmul;
+               type->without_array()->vector_elements * dmul;
          }
 
          while (idx < slot_limit) {
@@ -443,7 +425,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
                for (unsigned j = 0; j < 4; j++) {
                   if (explicit_locations[idx][j] &&
                       (explicit_locations[idx][j]->type->without_array()
-                       ->base_type != var->type->without_array()->base_type)) {
+                       ->base_type != type->without_array()->base_type)) {
                      linker_error(prog,
                                   "Varyings sharing the same location must "
                                   "have the same underlying numerical type. "
@@ -461,7 +443,7 @@ cross_validate_outputs_to_inputs(struct gl_shader_program *prog,
                 * worry about components beginning at anything other than 0 as
                 * the spec does not allow this for dvec3 and dvec4.
                 */
-               if (i == 3 && last_comp > 4) {
+               if (i == 4 && last_comp > 4) {
                   last_comp = last_comp - 4;
                   /* Bump location index and reset the component index */
                   idx++;
@@ -726,7 +708,7 @@ tfeedback_decl::assign_location(struct gl_context *ctx,
       + this->matched_candidate->toplevel_var->data.location_frac
       + this->matched_candidate->offset;
    const unsigned dmul =
-      this->matched_candidate->type->without_array()->is_double() ? 2 : 1;
+      this->matched_candidate->type->without_array()->is_64bit() ? 2 : 1;
 
    if (this->matched_candidate->type->is_array()) {
       /* Array variable */
@@ -904,7 +886,7 @@ tfeedback_decl::store(struct gl_context *ctx, struct gl_shader_program *prog,
    }
 
    if (explicit_stride && explicit_stride[buffer]) {
-      if (this->is_double() && info->Buffers[buffer].Stride % 2) {
+      if (this->is_64bit() && info->Buffers[buffer].Stride % 2) {
          linker_error(prog, "invalid qualifier xfb_stride=%d must be a "
                       "multiple of 8 as its applied to a type that is or "
                       "contains a double.",
@@ -1628,8 +1610,7 @@ varying_matches::compute_packing_class(const ir_variable *var)
    unsigned packing_class = var->data.centroid | (var->data.sample << 1) |
                             (var->data.patch << 2);
    packing_class *= 4;
-   packing_class += var->is_interpolation_flat()
-      ? unsigned(INTERP_QUALIFIER_FLAT) : var->data.interpolation;
+   packing_class += var->data.interpolation;
    return packing_class;
 }
 
@@ -1956,7 +1937,7 @@ canonicalize_shader_io(exec_list *ir, enum ir_variable_mode io_mode)
  * 64 bit map. Per-vertex and per-patch both have separate location domains
  * with a max of MAX_VARYING.
  */
-static uint64_t
+uint64_t
 reserved_varying_slot(struct gl_shader *stage, ir_variable_mode io_mode)
 {
    assert(io_mode == ir_var_shader_in || io_mode == ir_var_shader_out);
@@ -2018,7 +1999,8 @@ assign_varying_locations(struct gl_context *ctx,
                          struct gl_shader_program *prog,
                          gl_shader *producer, gl_shader *consumer,
                          unsigned num_tfeedback_decls,
-                         tfeedback_decl *tfeedback_decls)
+                         tfeedback_decl *tfeedback_decls,
+                         const uint64_t reserved_slots)
 {
    /* Tessellation shaders treat inputs and outputs as shared memory and can
     * access inputs and outputs of other invocations.
@@ -2196,10 +2178,6 @@ assign_varying_locations(struct gl_context *ctx,
       }
    }
 
-   const uint64_t reserved_slots =
-      reserved_varying_slot(producer, ir_var_shader_out) |
-      reserved_varying_slot(consumer, ir_var_shader_in);
-
    const unsigned slots_used = matches.assign_locations(prog, reserved_slots);
    matches.store_locations();
 
@@ -2282,14 +2260,16 @@ assign_varying_locations(struct gl_context *ctx,
 bool
 check_against_output_limit(struct gl_context *ctx,
                            struct gl_shader_program *prog,
-                           gl_shader *producer)
+                           gl_shader *producer,
+                           unsigned num_explicit_locations)
 {
-   unsigned output_vectors = 0;
+   unsigned output_vectors = num_explicit_locations;
 
    foreach_in_list(ir_instruction, node, producer->ir) {
       ir_variable *const var = node->as_variable();
 
-      if (var && var->data.mode == ir_var_shader_out &&
+      if (var && !var->data.explicit_location &&
+          var->data.mode == ir_var_shader_out &&
           var_counts_against_varying_limit(producer->Stage, var)) {
          /* outputs for fragment shader can't be doubles */
          output_vectors += var->type->count_attribute_slots(false);
@@ -2324,14 +2304,16 @@ check_against_output_limit(struct gl_context *ctx,
 bool
 check_against_input_limit(struct gl_context *ctx,
                           struct gl_shader_program *prog,
-                          gl_shader *consumer)
+                          gl_shader *consumer,
+                          unsigned num_explicit_locations)
 {
-   unsigned input_vectors = 0;
+   unsigned input_vectors = num_explicit_locations;
 
    foreach_in_list(ir_instruction, node, consumer->ir) {
       ir_variable *const var = node->as_variable();
 
-      if (var && var->data.mode == ir_var_shader_in &&
+      if (var && !var->data.explicit_location &&
+          var->data.mode == ir_var_shader_in &&
           var_counts_against_varying_limit(consumer->Stage, var)) {
          /* vertex inputs aren't varying counted */
          input_vectors += var->type->count_attribute_slots(false);

src/compiler/glsl/link_varyings.h

@@ -151,7 +151,7 @@ public:
          return this->size;
       else
          return this->vector_elements * this->matrix_columns * this->size *
-            (this->is_double() ? 2 : 1);
+            (this->is_64bit() ? 2 : 1);
    }
 
    unsigned get_location() const {
@@ -160,7 +160,7 @@ public:
 
 private:
 
-   bool is_double() const
+   bool is_64bit() const
    {
       switch (this->type) {
       case GL_DOUBLE:
@@ -320,16 +320,22 @@ assign_varying_locations(struct gl_context *ctx,
 			 struct gl_shader_program *prog,
 			 gl_shader *producer, gl_shader *consumer,
                          unsigned num_tfeedback_decls,
-                         tfeedback_decl *tfeedback_decls);
+                         tfeedback_decl *tfeedback_decls,
+                         const uint64_t reserved_slots);
+
+uint64_t
+reserved_varying_slot(struct gl_shader *stage, ir_variable_mode io_mode);
 
 bool
 check_against_output_limit(struct gl_context *ctx,
                            struct gl_shader_program *prog,
-                           gl_shader *producer);
+                           gl_shader *producer,
+                           unsigned num_explicit_locations);
 
 bool
 check_against_input_limit(struct gl_context *ctx,
                           struct gl_shader_program *prog,
-                          gl_shader *consumer);
+                          gl_shader *consumer,
+                          unsigned num_explicit_locations);
 
 #endif /* GLSL_LINK_VARYINGS_H */

src/compiler/glsl/linker.cpp

@@ -72,7 +72,6 @@
 #include "ir.h"
 #include "program.h"
 #include "program/hash_table.h"
-#include "program/prog_instruction.h"
 #include "linker.h"
 #include "link_varyings.h"
 #include "ir_optimization.h"
@@ -2486,7 +2485,7 @@ resize_tes_inputs(struct gl_context *ctx,
       ir->accept(&input_resize_visitor);
    }
 
-   if (tcs || ctx->Const.LowerTESPatchVerticesIn) {
+   if (tcs) {
       /* Convert the gl_PatchVerticesIn system value into a constant, since
        * the value is known at this point.
        */
@@ -2495,22 +2494,9 @@ resize_tes_inputs(struct gl_context *ctx,
          if (var && var->data.mode == ir_var_system_value &&
              var->data.location == SYSTEM_VALUE_VERTICES_IN) {
             void *mem_ctx = ralloc_parent(var);
+            var->data.mode = ir_var_auto;
             var->data.location = 0;
-            var->data.explicit_location = false;
-            if (tcs) {
-               var->data.mode = ir_var_auto;
-               var->constant_value = new(mem_ctx) ir_constant(num_vertices);
-            } else {
-               var->data.mode = ir_var_uniform;
-               var->data.how_declared = ir_var_hidden;
-               var->allocate_state_slots(1);
-               ir_state_slot *slot0 = &var->get_state_slots()[0];
-               slot0->swizzle = SWIZZLE_XXXX;
-               slot0->tokens[0] = STATE_INTERNAL;
-               slot0->tokens[1] = STATE_TES_PATCH_VERTICES_IN;
-               for (int i = 2; i < STATE_LENGTH; i++)
-                  slot0->tokens[i] = 0;
-            }
+            var->constant_value = new(mem_ctx) ir_constant(num_vertices);
          }
       }
    }
@@ -2877,7 +2863,7 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
              * issue (3) of the GL_ARB_vertex_attrib_64bit behavior, this
              * is optional behavior, but it seems preferable.
              */
-            if (var->type->without_array()->is_dual_slot_double())
+            if (var->type->without_array()->is_dual_slot())
                double_storage_locations |= (use_mask << attr);
 	 }
 
@@ -2954,7 +2940,7 @@ assign_attribute_or_color_locations(gl_shader_program *prog,
       to_assign[i].var->data.is_unmatched_generic_inout = 0;
       used_locations |= (use_mask << location);
 
-      if (to_assign[i].var->type->without_array()->is_dual_slot_double())
+      if (to_assign[i].var->type->without_array()->is_dual_slot())
          double_storage_locations |= (use_mask << location);
    }
 
@@ -3687,18 +3673,6 @@ create_shader_variable(struct gl_shader_program *shProg,
    if (in->data.mode == ir_var_system_value &&
        in->data.location == SYSTEM_VALUE_VERTEX_ID_ZERO_BASE) {
       out->name = ralloc_strdup(shProg, "gl_VertexID");
-   } else if ((in->data.mode == ir_var_shader_out &&
-               in->data.location == VARYING_SLOT_TESS_LEVEL_OUTER) ||
-              (in->data.mode == ir_var_system_value &&
-               in->data.location == SYSTEM_VALUE_TESS_LEVEL_OUTER)) {
-      out->name = ralloc_strdup(shProg, "gl_TessLevelOuter");
-      type = glsl_type::get_array_instance(glsl_type::float_type, 4);
-   } else if ((in->data.mode == ir_var_shader_out &&
-               in->data.location == VARYING_SLOT_TESS_LEVEL_INNER) ||
-              (in->data.mode == ir_var_system_value &&
-               in->data.location == SYSTEM_VALUE_TESS_LEVEL_INNER)) {
-      out->name = ralloc_strdup(shProg, "gl_TessLevelInner");
-      type = glsl_type::get_array_instance(glsl_type::float_type, 2);
    } else {
       out->name = ralloc_strdup(shProg, name);
    }
@@ -3851,9 +3825,6 @@ add_interface_variables(struct gl_shader_program *shProg,
          continue;
       };
 
-      if (var->data.patch)
-         loc_bias = int(VARYING_SLOT_PATCH0);
-
       /* Skip packed varyings, packed varyings are handled separately
        * by add_packed_varyings.
        */
@@ -4783,7 +4754,6 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
 	 ;
 
       lower_const_arrays_to_uniforms(prog->_LinkedShaders[i]->ir);
-      propagate_invariance(prog->_LinkedShaders[i]->ir);
    }
 
    /* Validation for special cases where we allow sampler array indexing
@@ -4880,9 +4850,12 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
     */
    if (last < MESA_SHADER_FRAGMENT &&
        (num_tfeedback_decls != 0 || prog->SeparateShader)) {
+      const uint64_t reserved_out_slots =
+         reserved_varying_slot(prog->_LinkedShaders[last], ir_var_shader_out);
       if (!assign_varying_locations(ctx, mem_ctx, prog,
                                     prog->_LinkedShaders[last], NULL,
-                                    num_tfeedback_decls, tfeedback_decls))
+                                    num_tfeedback_decls, tfeedback_decls,
+                                    reserved_out_slots))
          goto done;
    }
 
@@ -4900,6 +4873,9 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
 
          gl_shader *const sh = prog->_LinkedShaders[last];
          if (prog->SeparateShader) {
+            const uint64_t reserved_slots =
+               reserved_varying_slot(sh, ir_var_shader_in);
+
             /* Assign input locations for SSO, output locations are already
              * assigned.
              */
@@ -4907,7 +4883,8 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
                                           NULL /* producer */,
                                           sh /* consumer */,
                                           0 /* num_tfeedback_decls */,
-                                          NULL /* tfeedback_decls */))
+                                          NULL /* tfeedback_decls */,
+                                          reserved_slots))
                goto done;
          }
 
@@ -4928,9 +4905,15 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
             gl_shader *const sh_i = prog->_LinkedShaders[i];
             gl_shader *const sh_next = prog->_LinkedShaders[next];
 
+            const uint64_t reserved_out_slots =
+               reserved_varying_slot(sh_i, ir_var_shader_out);
+            const uint64_t reserved_in_slots =
+               reserved_varying_slot(sh_next, ir_var_shader_in);
+
             if (!assign_varying_locations(ctx, mem_ctx, prog, sh_i, sh_next,
                       next == MESA_SHADER_FRAGMENT ? num_tfeedback_decls : 0,
-                      tfeedback_decls))
+                      tfeedback_decls,
+                      reserved_out_slots | reserved_in_slots))
                goto done;
 
             do_dead_builtin_varyings(ctx, sh_i, sh_next,
@@ -4939,11 +4922,14 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
 
             /* This must be done after all dead varyings are eliminated. */
             if (sh_i != NULL) {
-               if (!check_against_output_limit(ctx, prog, sh_i)) {
+               unsigned slots_used = _mesa_bitcount_64(reserved_out_slots);
+               if (!check_against_output_limit(ctx, prog, sh_i, slots_used)) {
                   goto done;
                }
             }
-            if (!check_against_input_limit(ctx, prog, sh_next))
+
+            unsigned slots_used = _mesa_bitcount_64(reserved_in_slots);
+            if (!check_against_input_limit(ctx, prog, sh_next, slots_used))
                goto done;
 
             next = i;

src/compiler/glsl/linker.h

@@ -156,7 +156,7 @@ protected:
     */
    virtual void visit_field(const glsl_type *type, const char *name,
                             bool row_major, const glsl_type *record_type,
-                            const unsigned packing,
+                            const enum glsl_interface_packing packing,
                             bool last_field);
 
    /**
@@ -180,10 +180,10 @@ protected:
    virtual void visit_field(const glsl_struct_field *field);
 
    virtual void enter_record(const glsl_type *type, const char *name,
-                             bool row_major, const unsigned packing);
+                             bool row_major, const enum glsl_interface_packing packing);
 
    virtual void leave_record(const glsl_type *type, const char *name,
-                             bool row_major, const unsigned packing);
+                             bool row_major, const enum glsl_interface_packing packing);
 
    virtual void set_buffer_offset(unsigned offset);
 
@@ -199,7 +199,7 @@ private:
     */
    void recursion(const glsl_type *t, char **name, size_t name_length,
                   bool row_major, const glsl_type *record_type,
-                  const unsigned packing,
+                  const enum glsl_interface_packing packing,
                   bool last_field, unsigned record_array_count,
                   const glsl_struct_field *named_ifc_member);
 };

src/compiler/glsl/lower_buffer_access.cpp

@@ -114,7 +114,7 @@ lower_buffer_access::emit_access(void *mem_ctx,
             /* For a row-major matrix, the next column starts at the next
              * element.
              */
-            int size_mul = deref->type->is_double() ? 8 : 4;
+            int size_mul = deref->type->is_64bit() ? 8 : 4;
             emit_access(mem_ctx, is_write, col_deref, base_offset,
                         deref_offset + i * size_mul,
                         row_major, deref->type->matrix_columns, packing,
@@ -125,7 +125,7 @@ lower_buffer_access::emit_access(void *mem_ctx,
             /* std430 doesn't round up vec2 size to a vec4 size */
             if (packing == GLSL_INTERFACE_PACKING_STD430 &&
                 deref->type->vector_elements == 2 &&
-                !deref->type->is_double()) {
+                !deref->type->is_64bit()) {
                size_mul = 8;
             } else {
                /* std140 always rounds the stride of arrays (and matrices) to a
@@ -137,7 +137,7 @@ lower_buffer_access::emit_access(void *mem_ctx,
                 * machine units, the base alignment is 4N. For vec4, base
                 * alignment is 4N.
                 */
-               size_mul = (deref->type->is_double() &&
+               size_mul = (deref->type->is_64bit() &&
                            deref->type->vector_elements > 2) ? 32 : 16;
             }
 
@@ -159,7 +159,7 @@ lower_buffer_access::emit_access(void *mem_ctx,
          is_write ? write_mask : (1 << deref->type->vector_elements) - 1;
       insert_buffer_access(mem_ctx, deref, deref->type, offset, mask, -1);
    } else {
-      unsigned N = deref->type->is_double() ? 8 : 4;
+      unsigned N = deref->type->is_64bit() ? 8 : 4;
 
       /* We're dereffing a column out of a row-major matrix, so we
        * gather the vector from each stored row.
@@ -328,7 +328,7 @@ lower_buffer_access::setup_buffer_access(void *mem_ctx,
                                          bool *row_major,
                                          int *matrix_columns,
                                          const glsl_struct_field **struct_field,
-                                         unsigned packing)
+                                         enum glsl_interface_packing packing)
 {
    *offset = new(mem_ctx) ir_constant(0u);
    *row_major = is_dereferenced_thing_row_major(deref);
@@ -358,7 +358,7 @@ lower_buffer_access::setup_buffer_access(void *mem_ctx,
              * thread or SIMD channel is modifying the same vector.
              */
             array_stride = 4;
-            if (deref_array->array->type->is_double())
+            if (deref_array->array->type->is_64bit())
                array_stride *= 2;
          } else if (deref_array->array->type->is_matrix() && *row_major) {
             /* When loading a vector out of a row major matrix, the
@@ -367,7 +367,7 @@ lower_buffer_access::setup_buffer_access(void *mem_ctx,
              * vector) is handled below in emit_ubo_loads.
              */
             array_stride = 4;
-            if (deref_array->array->type->is_double())
+            if (deref_array->array->type->is_64bit())
                array_stride *= 2;
             *matrix_columns = deref_array->array->type->matrix_columns;
          } else if (deref_array->type->without_array()->is_interface()) {

src/compiler/glsl/lower_buffer_access.h

@@ -58,7 +58,7 @@ public:
                             ir_rvalue **offset, unsigned *const_offset,
                             bool *row_major, int *matrix_columns,
                             const glsl_struct_field **struct_field,
-                            unsigned packing);
+                            enum glsl_interface_packing packing);
 };
 
 } /* namespace lower_buffer_access */

src/compiler/glsl/lower_const_arrays_to_uniforms.cpp

@@ -57,7 +57,6 @@ public:
       return progress;
    }
 
-   ir_visitor_status visit_enter(ir_texture *);
    void handle_rvalue(ir_rvalue **rvalue);
 
 private:
@@ -65,25 +64,23 @@ private:
    bool progress;
 };
 
-ir_visitor_status
-lower_const_array_visitor::visit_enter(ir_texture *)
-{
-   return visit_continue_with_parent;
-}
-
 void
 lower_const_array_visitor::handle_rvalue(ir_rvalue **rvalue)
 {
    if (!*rvalue)
       return;
 
-   ir_constant *con = (*rvalue)->as_constant();
+   ir_dereference_array *dra = (*rvalue)->as_dereference_array();
+   if (!dra)
+      return;
+
+   ir_constant *con = dra->array->as_constant();
    if (!con || !con->type->is_array())
       return;
 
    void *mem_ctx = ralloc_parent(con);
 
-   char *uniform_name = ralloc_asprintf(mem_ctx, "constarray__%p", con);
+   char *uniform_name = ralloc_asprintf(mem_ctx, "constarray__%p", dra);
 
    ir_variable *uni =
       new(mem_ctx) ir_variable(con->type, uniform_name, ir_var_uniform);
@@ -96,7 +93,8 @@ lower_const_array_visitor::handle_rvalue(ir_rvalue **rvalue)
    uni->data.max_array_access = uni->type->length - 1;
    instructions->push_head(uni);
 
-   *rvalue = new(mem_ctx) ir_dereference_variable(uni);
+   ir_dereference_variable *varref = new(mem_ctx) ir_dereference_variable(uni);
+   *rvalue = new(mem_ctx) ir_dereference_array(varref, dra->array_index);
 
    progress = true;
 }

src/compiler/glsl/lower_packed_varyings.cpp

@@ -273,11 +273,11 @@ lower_packed_varyings_visitor::run(struct gl_shader *shader)
          continue;
 
       /* This lowering pass is only capable of packing floats and ints
-       * together when their interpolation mode is "flat".  Treat integers as
-       * being flat when the interpolation mode is none.
+       * together when their interpolation mode is "flat".  Therefore, to be
+       * safe, caller should ensure that integral varyings always use flat
+       * interpolation, even when this is not required by GLSL.
        */
       assert(var->data.interpolation == INTERP_QUALIFIER_FLAT ||
-             var->data.interpolation == INTERP_QUALIFIER_NONE ||
              !var->type->contains_integer());
 
       /* Clone the variable for program resource list before
@@ -432,7 +432,7 @@ lower_packed_varyings_visitor::lower_rvalue(ir_rvalue *rvalue,
                                             bool gs_input_toplevel,
                                             unsigned vertex_index)
 {
-   unsigned dmul = rvalue->type->is_double() ? 2 : 1;
+   unsigned dmul = rvalue->type->is_64bit() ? 2 : 1;
    /* When gs_input_toplevel is set, we should be looking at a geometry shader
     * input array.
     */
@@ -480,7 +480,7 @@ lower_packed_varyings_visitor::lower_rvalue(ir_rvalue *rvalue,
       char right_swizzle_name[4] = { 0, 0, 0, 0 };
 
       left_components = 4 - fine_location % 4;
-      if (rvalue->type->is_double()) {
+      if (rvalue->type->is_64bit()) {
          /* We might actually end up with 0 left components! */
          left_components /= 2;
       }
@@ -607,7 +607,7 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
    if (this->packed_varyings[slot] == NULL) {
       char *packed_name = ralloc_asprintf(this->mem_ctx, "packed:%s", name);
       const glsl_type *packed_type;
-      if (unpacked_var->is_interpolation_flat())
+      if (unpacked_var->data.interpolation == INTERP_QUALIFIER_FLAT)
          packed_type = glsl_type::ivec4_type;
       else
          packed_type = glsl_type::vec4_type;
@@ -627,8 +627,7 @@ lower_packed_varyings_visitor::get_packed_varying_deref(
       packed_var->data.centroid = unpacked_var->data.centroid;
       packed_var->data.sample = unpacked_var->data.sample;
       packed_var->data.patch = unpacked_var->data.patch;
-      packed_var->data.interpolation = packed_type == glsl_type::ivec4_type
-         ? unsigned(INTERP_QUALIFIER_FLAT) : unpacked_var->data.interpolation;
+      packed_var->data.interpolation = unpacked_var->data.interpolation;
       packed_var->data.location = location;
       packed_var->data.precision = unpacked_var->data.precision;
       packed_var->data.always_active_io = unpacked_var->data.always_active_io;
@@ -677,7 +676,7 @@ lower_packed_varyings_visitor::needs_lowering(ir_variable *var)
       return false;
 
    type = type->without_array();
-   if (type->vector_elements == 4 && !type->is_double())
+   if (type->vector_elements == 4 && !type->is_64bit())
       return false;
    return true;
 }

src/compiler/glsl/lower_shared_reference.cpp

@@ -138,7 +138,7 @@ lower_shared_reference_visitor::handle_rvalue(ir_rvalue **rvalue)
    bool row_major;
    int matrix_columns;
    assert(var->get_interface_type() == NULL);
-   const unsigned packing = GLSL_INTERFACE_PACKING_STD430;
+   const enum glsl_interface_packing packing = GLSL_INTERFACE_PACKING_STD430;
 
    setup_buffer_access(mem_ctx, var, deref,
                        &offset, &const_offset,
@@ -206,7 +206,7 @@ lower_shared_reference_visitor::handle_assignment(ir_assignment *ir)
    bool row_major;
    int matrix_columns;
    assert(var->get_interface_type() == NULL);
-   const unsigned packing = GLSL_INTERFACE_PACKING_STD430;
+   const enum glsl_interface_packing packing = GLSL_INTERFACE_PACKING_STD430;
 
    setup_buffer_access(mem_ctx, var, deref,
                        &offset, &const_offset,
@@ -365,7 +365,7 @@ lower_shared_reference_visitor::lower_shared_atomic_intrinsic(ir_call *ir)
    bool row_major;
    int matrix_columns;
    assert(var->get_interface_type() == NULL);
-   const unsigned packing = GLSL_INTERFACE_PACKING_STD430;
+   const enum glsl_interface_packing packing = GLSL_INTERFACE_PACKING_STD430;
    buffer_access_type = shared_atomic_access;
 
    setup_buffer_access(mem_ctx, var, deref,

src/compiler/glsl/lower_ubo_reference.cpp

@@ -61,7 +61,7 @@ public:
                                 unsigned *const_offset,
                                 bool *row_major,
                                 int *matrix_columns,
-                                unsigned packing);
+                                enum glsl_interface_packing packing);
    uint32_t ssbo_access_params();
    ir_expression *ubo_load(void *mem_ctx, const struct glsl_type *type,
 			   ir_rvalue *offset);
@@ -99,7 +99,7 @@ public:
    ir_expression *emit_ssbo_get_buffer_size(void *mem_ctx);
 
    unsigned calculate_unsized_array_stride(ir_dereference *deref,
-                                           unsigned packing);
+                                           enum glsl_interface_packing packing);
 
    ir_call *lower_ssbo_atomic_intrinsic(ir_call *ir);
    ir_call *check_for_ssbo_atomic_intrinsic(ir_call *ir);
@@ -273,7 +273,7 @@ lower_ubo_reference_visitor::setup_for_load_or_store(void *mem_ctx,
                                                      unsigned *const_offset,
                                                      bool *row_major,
                                                      int *matrix_columns,
-                                                     unsigned packing)
+                                                     enum glsl_interface_packing packing)
 {
    /* Determine the name of the interface block */
    ir_rvalue *nonconst_block_index;
@@ -344,7 +344,7 @@ lower_ubo_reference_visitor::handle_rvalue(ir_rvalue **rvalue)
    unsigned const_offset;
    bool row_major;
    int matrix_columns;
-   unsigned packing = var->get_interface_type()->interface_packing;
+   enum glsl_interface_packing packing = var->get_interface_type_packing();
 
    this->buffer_access_type =
       var->is_in_shader_storage_block() ?
@@ -557,7 +557,7 @@ lower_ubo_reference_visitor::write_to_memory(void *mem_ctx,
    unsigned const_offset;
    bool row_major;
    int matrix_columns;
-   unsigned packing = var->get_interface_type()->interface_packing;
+   enum glsl_interface_packing packing = var->get_interface_type_packing();
 
    this->buffer_access_type = ssbo_store_access;
    this->variable = var;
@@ -666,7 +666,7 @@ lower_ubo_reference_visitor::emit_ssbo_get_buffer_size(void *mem_ctx)
 
 unsigned
 lower_ubo_reference_visitor::calculate_unsized_array_stride(ir_dereference *deref,
-                                                            unsigned packing)
+                                                            enum glsl_interface_packing packing)
 {
    unsigned array_stride = 0;
 
@@ -736,7 +736,7 @@ lower_ubo_reference_visitor::process_ssbo_unsized_array_length(ir_rvalue **rvalu
    unsigned const_offset;
    bool row_major;
    int matrix_columns;
-   unsigned packing = var->get_interface_type()->interface_packing;
+   enum glsl_interface_packing packing = var->get_interface_type_packing();
    int unsized_array_stride = calculate_unsized_array_stride(deref, packing);
 
    this->buffer_access_type = ssbo_unsized_array_length_access;
@@ -970,7 +970,7 @@ lower_ubo_reference_visitor::lower_ssbo_atomic_intrinsic(ir_call *ir)
    unsigned const_offset;
    bool row_major;
    int matrix_columns;
-   unsigned packing = var->get_interface_type()->interface_packing;
+   enum glsl_interface_packing packing = var->get_interface_type_packing();
 
    this->buffer_access_type = ssbo_atomic_access;
    this->variable = var;

src/compiler/glsl/opt_array_splitting.cpp

@@ -93,7 +93,6 @@ public:
    {
       this->mem_ctx = ralloc_context(NULL);
       this->variable_list.make_empty();
-      this->in_whole_array_copy = false;
    }
 
    ~ir_array_reference_visitor(void)
@@ -105,8 +104,6 @@ public:
 
    virtual ir_visitor_status visit(ir_variable *);
    virtual ir_visitor_status visit(ir_dereference_variable *);
-   virtual ir_visitor_status visit_enter(ir_assignment *);
-   virtual ir_visitor_status visit_leave(ir_assignment *);
    virtual ir_visitor_status visit_enter(ir_dereference_array *);
    virtual ir_visitor_status visit_enter(ir_function_signature *);
 
@@ -116,8 +113,6 @@ public:
    exec_list variable_list;
 
    void *mem_ctx;
-
-   bool in_whole_array_copy;
 };
 
 } /* namespace */
@@ -162,34 +157,11 @@ ir_array_reference_visitor::visit(ir_variable *ir)
    return visit_continue;
 }
 
-ir_visitor_status
-ir_array_reference_visitor::visit_enter(ir_assignment *ir)
-{
-   in_whole_array_copy =
-      ir->lhs->type->is_array() && ir->whole_variable_written();
-
-   return visit_continue;
-}
-
-ir_visitor_status
-ir_array_reference_visitor::visit_leave(ir_assignment *ir)
-{
-   in_whole_array_copy = false;
-
-   return visit_continue;
-}
-
 ir_visitor_status
 ir_array_reference_visitor::visit(ir_dereference_variable *ir)
 {
    variable_entry *entry = this->get_variable_entry(ir->var);
 
-   /* Allow whole-array assignments on the LHS.  We can split those
-    * by "unrolling" the assignment into component-wise assignments.
-    */
-   if (in_assignee && in_whole_array_copy)
-      return visit_continue;
-
    /* If we made it to here without seeing an ir_dereference_array,
     * then the dereference of this array didn't have a constant index
     * (see the visit_continue_with_parent below), so we can't split
@@ -378,33 +350,6 @@ ir_array_splitting_visitor::visit_leave(ir_assignment *ir)
     */
    ir_rvalue *lhs = ir->lhs;
 
-   /* "Unroll" any whole array assignments, creating assignments for
-    * each array element.  Then, do splitting on each new assignment.
-    */
-   if (lhs->type->is_array() && ir->whole_variable_written() &&
-       get_splitting_entry(ir->whole_variable_written())) {
-      void *mem_ctx = ralloc_parent(ir);
-
-      for (unsigned i = 0; i < lhs->type->length; i++) {
-         ir_rvalue *lhs_i =
-            new(mem_ctx) ir_dereference_array(ir->lhs->clone(mem_ctx, NULL),
-                                              new(mem_ctx) ir_constant(i));
-         ir_rvalue *rhs_i =
-            new(mem_ctx) ir_dereference_array(ir->rhs->clone(mem_ctx, NULL),
-                                              new(mem_ctx) ir_constant(i));
-         ir_rvalue *condition_i =
-            ir->condition ? ir->condition->clone(mem_ctx, NULL) : NULL;
-
-         ir_assignment *assign_i =
-            new(mem_ctx) ir_assignment(lhs_i, rhs_i, condition_i);
-
-         ir->insert_before(assign_i);
-         assign_i->accept(this);
-      }
-      ir->remove();
-      return visit_continue;
-   }
-
    handle_rvalue(&lhs);
    ir->lhs = lhs->as_dereference();
 

src/compiler/glsl/opt_conditional_discard.cpp

@@ -72,14 +72,7 @@ opt_conditional_discard_visitor::visit_leave(ir_if *ir)
 
    /* Move the condition and replace the ir_if with the ir_discard. */
    ir_discard *discard = (ir_discard *) ir->then_instructions.head;
-   if (!discard->condition)
-      discard->condition = ir->condition;
-   else {
-      void *ctx = ralloc_parent(ir);
-      discard->condition = new(ctx) ir_expression(ir_binop_logic_and,
-                                                  ir->condition,
-                                                  discard->condition);
-   }
+   discard->condition = ir->condition;
    ir->replace_with(discard);
 
    progress = true;

src/compiler/glsl/opt_constant_propagation.cpp

@@ -138,14 +138,14 @@ public:
 void
 ir_constant_propagation_visitor::constant_folding(ir_rvalue **rvalue)
 {
-   if (this->in_assignee || *rvalue == NULL)
+   if (*rvalue == NULL)
       return;
 
    if (ir_constant_fold(rvalue))
       this->progress = true;
 
    ir_dereference_variable *var_ref = (*rvalue)->as_dereference_variable();
-   if (var_ref && !var_ref->type->is_array()) {
+   if (var_ref) {
       ir_constant *constant = var_ref->constant_expression_value();
       if (constant) {
          *rvalue = constant;

src/compiler/glsl/opt_copy_propagation.cpp

@@ -83,6 +83,7 @@ public:
    }
 
    virtual ir_visitor_status visit(class ir_dereference_variable *);
+   void handle_loop(class ir_loop *, bool keep_acp);
    virtual ir_visitor_status visit_enter(class ir_loop *);
    virtual ir_visitor_status visit_enter(class ir_function_signature *);
    virtual ir_visitor_status visit_enter(class ir_function *);
@@ -252,21 +253,24 @@ ir_copy_propagation_visitor::visit_enter(ir_if *ir)
    return visit_continue_with_parent;
 }
 
-ir_visitor_status
-ir_copy_propagation_visitor::visit_enter(ir_loop *ir)
+void
+ir_copy_propagation_visitor::handle_loop(ir_loop *ir, bool keep_acp)
 {
    exec_list *orig_acp = this->acp;
    exec_list *orig_kills = this->kills;
    bool orig_killed_all = this->killed_all;
 
-   /* FINISHME: For now, the initial acp for loops is totally empty.
-    * We could go through once, then go through again with the acp
-    * cloned minus the killed entries after the first run through.
-    */
    this->acp = new(mem_ctx) exec_list;
    this->kills = new(mem_ctx) exec_list;
    this->killed_all = false;
 
+   if (keep_acp) {
+      /* Populate the initial acp with a copy of the original */
+      foreach_in_list(acp_entry, a, orig_acp) {
+         this->acp->push_tail(new(this->acp) acp_entry(a->lhs, a->rhs));
+      }
+   }
+
    visit_list_elements(this, &ir->body_instructions);
 
    if (this->killed_all) {
@@ -284,6 +288,20 @@ ir_copy_propagation_visitor::visit_enter(ir_loop *ir)
    }
 
    ralloc_free(new_kills);
+}
+
+ir_visitor_status
+ir_copy_propagation_visitor::visit_enter(ir_loop *ir)
+{
+   /* Make a conservative first pass over the loop with an empty ACP set.
+    * This also removes any killed entries from the original ACP set.
+    */
+   handle_loop(ir, false);
+
+   /* Then, run it again with the real ACP set, minus any killed entries.
+    * This takes care of propagating values from before the loop into it.
+    */
+   handle_loop(ir, true);
 
    /* already descended into the children. */
    return visit_continue_with_parent;

src/compiler/glsl/opt_copy_propagation_elements.cpp

@@ -106,6 +106,7 @@ public:
       ralloc_free(mem_ctx);
    }
 
+   void handle_loop(ir_loop *, bool keep_acp);
    virtual ir_visitor_status visit_enter(class ir_loop *);
    virtual ir_visitor_status visit_enter(class ir_function_signature *);
    virtual ir_visitor_status visit_leave(class ir_assignment *);
@@ -374,8 +375,8 @@ ir_copy_propagation_elements_visitor::visit_enter(ir_if *ir)
    return visit_continue_with_parent;
 }
 
-ir_visitor_status
-ir_copy_propagation_elements_visitor::visit_enter(ir_loop *ir)
+void
+ir_copy_propagation_elements_visitor::handle_loop(ir_loop *ir, bool keep_acp)
 {
    exec_list *orig_acp = this->acp;
    exec_list *orig_kills = this->kills;
@@ -389,6 +390,13 @@ ir_copy_propagation_elements_visitor::visit_enter(ir_loop *ir)
    this->kills = new(mem_ctx) exec_list;
    this->killed_all = false;
 
+   if (keep_acp) {
+      /* Populate the initial acp with a copy of the original */
+      foreach_in_list(acp_entry, a, orig_acp) {
+         this->acp->push_tail(new(this->acp) acp_entry(a));
+      }
+   }
+
    visit_list_elements(this, &ir->body_instructions);
 
    if (this->killed_all) {
@@ -406,6 +414,13 @@ ir_copy_propagation_elements_visitor::visit_enter(ir_loop *ir)
    }
 
    ralloc_free(new_kills);
+}
+
+ir_visitor_status
+ir_copy_propagation_elements_visitor::visit_enter(ir_loop *ir)
+{
+   handle_loop(ir, false);
+   handle_loop(ir, true);
 
    /* already descended into the children. */
    return visit_continue_with_parent;

src/compiler/glsl/opt_dead_builtin_varyings.cpp

@@ -85,13 +85,10 @@ public:
    {
       ir_variable *var = ir->variable_referenced();
 
-      if (!var || var->data.mode != this->mode || !var->type->is_array() ||
-          !is_gl_identifier(var->name))
+      if (!var || var->data.mode != this->mode || !var->type->is_array())
          return visit_continue;
 
-      /* Only match gl_FragData[], not gl_SecondaryFragDataEXT[] */
-      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0 &&
-          var->data.index == 0) {
+      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0) {
          this->fragdata_array = var;
 
          ir_constant *index = ir->array_index->as_constant();
@@ -146,8 +143,7 @@ public:
       if (var->data.mode != this->mode || !var->type->is_array())
          return visit_continue;
 
-      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0 &&
-          var->data.index == 0) {
+      if (this->find_frag_outputs && var->data.location == FRAG_RESULT_DATA0) {
          /* This is a whole array dereference. */
          this->fragdata_usage |= (1 << var->type->array_size()) - 1;
          this->lower_fragdata_array = false;

src/compiler/glsl/opt_dead_code.cpp

@@ -144,7 +144,7 @@ do_dead_code(exec_list *instructions, bool uniform_locations_assigned)
              * layouts, do not eliminate it.
              */
             if (entry->var->is_in_buffer_block()) {
-               if (entry->var->get_interface_type()->interface_packing !=
+               if (entry->var->get_interface_type_packing() !=
                    GLSL_INTERFACE_PACKING_PACKED)
                   continue;
             }

src/compiler/glsl_types.cpp

@@ -1079,7 +1079,7 @@ function_key_compare(const void *a, const void *b)
    const glsl_type *const key2 = (glsl_type *) b;
 
    if (key1->length != key2->length)
-      return false;
+      return 1;
 
    return memcmp(key1->fields.parameters, key2->fields.parameters,
                  (key1->length + 1) * sizeof(*key1->fields.parameters)) == 0;
@@ -1090,8 +1090,20 @@ static uint32_t
 function_key_hash(const void *a)
 {
    const glsl_type *const key = (glsl_type *) a;
-   return _mesa_hash_data(key->fields.parameters,
-                          (key->length + 1) * sizeof(*key->fields.parameters));
+   char hash_key[128];
+   unsigned size = 0;
+
+   size = snprintf(hash_key, sizeof(hash_key), "%08x", key->length);
+
+   for (unsigned i = 0; i < key->length; i++) {
+      if (size >= sizeof(hash_key))
+	 break;
+
+      size += snprintf(& hash_key[size], sizeof(hash_key) - size,
+		       "%p", (void *) key->fields.structure[i].type);
+   }
+
+   return _mesa_hash_string(hash_key);
 }
 
 const glsl_type *
@@ -1422,7 +1434,7 @@ glsl_type::can_implicitly_convert_to(const glsl_type *desired,
 unsigned
 glsl_type::std140_base_alignment(bool row_major) const
 {
-   unsigned N = is_double() ? 8 : 4;
+   unsigned N = is_64bit() ? 8 : 4;
 
    /* (1) If the member is a scalar consuming <N> basic machine units, the
     *     base alignment is <N>.
@@ -1540,7 +1552,7 @@ glsl_type::std140_base_alignment(bool row_major) const
 unsigned
 glsl_type::std140_size(bool row_major) const
 {
-   unsigned N = is_double() ? 8 : 4;
+   unsigned N = is_64bit() ? 8 : 4;
 
    /* (1) If the member is a scalar consuming <N> basic machine units, the
     *     base alignment is <N>.
@@ -1677,7 +1689,7 @@ unsigned
 glsl_type::std430_base_alignment(bool row_major) const
 {
 
-   unsigned N = is_double() ? 8 : 4;
+   unsigned N = is_64bit() ? 8 : 4;
 
    /* (1) If the member is a scalar consuming <N> basic machine units, the
     *     base alignment is <N>.
@@ -1786,7 +1798,7 @@ glsl_type::std430_base_alignment(bool row_major) const
 unsigned
 glsl_type::std430_array_stride(bool row_major) const
 {
-   unsigned N = is_double() ? 8 : 4;
+   unsigned N = is_64bit() ? 8 : 4;
 
    /* Notice that the array stride of a vec3 is not 3 * N but 4 * N.
     * See OpenGL 4.30 spec, section 7.6.2.2 "Standard Uniform Block Layout"
@@ -1804,7 +1816,7 @@ glsl_type::std430_array_stride(bool row_major) const
 unsigned
 glsl_type::std430_size(bool row_major) const
 {
-   unsigned N = is_double() ? 8 : 4;
+   unsigned N = is_64bit() ? 8 : 4;
 
    /* OpenGL 4.30 spec, section 7.6.2.2 "Standard Uniform Block Layout":
     *

src/compiler/glsl_types.h

@@ -64,6 +64,11 @@ enum glsl_base_type {
    GLSL_TYPE_ERROR
 };
 
+static inline bool glsl_base_type_is_64bit(enum glsl_base_type type)
+{
+   return type == GLSL_TYPE_DOUBLE;
+}
+
 enum glsl_sampler_dim {
    GLSL_SAMPLER_DIM_1D = 0,
    GLSL_SAMPLER_DIM_2D,
@@ -490,11 +495,19 @@ struct glsl_type {
    }
 
    /**
-    * Query whether a double takes two slots.
+    * Query whether a 64-bit type takes two slots.
     */
-   bool is_dual_slot_double() const
+   bool is_dual_slot() const
    {
-      return base_type == GLSL_TYPE_DOUBLE && vector_elements > 2;
+      return is_64bit() && vector_elements > 2;
+   }
+
+   /**
+    * Query whether or not a type is 64-bit
+    */
+   bool is_64bit() const
+   {
+      return glsl_base_type_is_64bit(base_type);
    }
 
    /**
@@ -745,6 +758,14 @@ struct glsl_type {
     */
    bool record_compare(const glsl_type *b, bool match_locations = true) const;
 
+   /**
+    * Get the type interface packing.
+    */
+   enum glsl_interface_packing get_interface_packing() const
+   {
+      return (enum glsl_interface_packing)interface_packing;
+   }
+
 private:
 
    static mtx_t mutex;

src/compiler/nir/nir.c

@@ -659,122 +659,6 @@ nir_copy_deref(void *mem_ctx, nir_deref *deref)
    return NULL;
 }
 
-/* This is the second step in the recursion.  We've found the tail and made a
- * copy.  Now we need to iterate over all possible leaves and call the
- * callback on each one.
- */
-static bool
-deref_foreach_leaf_build_recur(nir_deref_var *deref, nir_deref *tail,
-                               nir_deref_foreach_leaf_cb cb, void *state)
-{
-   unsigned length;
-   union {
-      nir_deref_array arr;
-      nir_deref_struct str;
-   } tmp;
-
-   assert(tail->child == NULL);
-   switch (glsl_get_base_type(tail->type)) {
-   case GLSL_TYPE_UINT:
-   case GLSL_TYPE_INT:
-   case GLSL_TYPE_FLOAT:
-   case GLSL_TYPE_DOUBLE:
-   case GLSL_TYPE_BOOL:
-      if (glsl_type_is_vector_or_scalar(tail->type))
-         return cb(deref, state);
-      /* Fall Through */
-
-   case GLSL_TYPE_ARRAY:
-      tmp.arr.deref.deref_type = nir_deref_type_array;
-      tmp.arr.deref.type = glsl_get_array_element(tail->type);
-      tmp.arr.deref_array_type = nir_deref_array_type_direct;
-      tmp.arr.indirect = NIR_SRC_INIT;
-      tail->child = &tmp.arr.deref;
-
-      length = glsl_get_length(tail->type);
-      for (unsigned i = 0; i < length; i++) {
-         tmp.arr.deref.child = NULL;
-         tmp.arr.base_offset = i;
-         if (!deref_foreach_leaf_build_recur(deref, &tmp.arr.deref, cb, state))
-            return false;
-      }
-      return true;
-
-   case GLSL_TYPE_STRUCT:
-      tmp.str.deref.deref_type = nir_deref_type_struct;
-      tail->child = &tmp.str.deref;
-
-      length = glsl_get_length(tail->type);
-      for (unsigned i = 0; i < length; i++) {
-         tmp.arr.deref.child = NULL;
-         tmp.str.deref.type = glsl_get_struct_field(tail->type, i);
-         tmp.str.index = i;
-         if (!deref_foreach_leaf_build_recur(deref, &tmp.arr.deref, cb, state))
-            return false;
-      }
-      return true;
-
-   default:
-      unreachable("Invalid type for dereference");
-   }
-}
-
-/* This is the first step of the foreach_leaf recursion.  In this step we are
- * walking to the end of the deref chain and making a copy in the stack as we
- * go.  This is because we don't want to mutate the deref chain that was
- * passed in by the caller.  The downside is that this deref chain is on the
- * stack and , if the caller wants to do anything with it, they will have to
- * make their own copy because this one will go away.
- */
-static bool
-deref_foreach_leaf_copy_recur(nir_deref_var *deref, nir_deref *tail,
-                              nir_deref_foreach_leaf_cb cb, void *state)
-{
-   union {
-      nir_deref_array arr;
-      nir_deref_struct str;
-   } c;
-
-   if (tail->child) {
-      switch (tail->child->deref_type) {
-      case nir_deref_type_array:
-         c.arr = *nir_deref_as_array(tail->child);
-         tail->child = &c.arr.deref;
-         return deref_foreach_leaf_copy_recur(deref, &c.arr.deref, cb, state);
-
-      case nir_deref_type_struct:
-         c.str = *nir_deref_as_struct(tail->child);
-         tail->child = &c.str.deref;
-         return deref_foreach_leaf_copy_recur(deref, &c.str.deref, cb, state);
-
-      case nir_deref_type_var:
-      default:
-         unreachable("Invalid deref type for a child");
-      }
-   } else {
-      /* We've gotten to the end of the original deref.  Time to start
-       * building our own derefs.
-       */
-      return deref_foreach_leaf_build_recur(deref, tail, cb, state);
-   }
-}
-
-/**
- * This function iterates over all of the possible derefs that can be created
- * with the given deref as the head.  It then calls the provided callback with
- * a full deref for each one.
- *
- * The deref passed to the callback will be allocated on the stack.  You will
- * need to make a copy if you want it to hang around.
- */
-bool
-nir_deref_foreach_leaf(nir_deref_var *deref,
-                       nir_deref_foreach_leaf_cb cb, void *state)
-{
-   nir_deref_var copy = *deref;
-   return deref_foreach_leaf_copy_recur(&copy, &copy.deref, cb, state);
-}
-
 /* Returns a load_const instruction that represents the constant
  * initializer for the given deref chain.  The caller is responsible for
  * ensuring that there actually is a constant initializer.

src/compiler/nir/nir.h

@@ -1234,50 +1234,6 @@ nir_tex_instr_is_query(nir_tex_instr *instr)
    }
 }
 
-static inline nir_alu_type
-nir_tex_instr_src_type(nir_tex_instr *instr, unsigned src)
-{
-   switch (instr->src[src].src_type) {
-   case nir_tex_src_coord:
-      switch (instr->op) {
-      case nir_texop_txf:
-      case nir_texop_txf_ms:
-      case nir_texop_txf_ms_mcs:
-      case nir_texop_samples_identical:
-         return nir_type_int;
-
-      default:
-         return nir_type_float;
-      }
-
-   case nir_tex_src_lod:
-      switch (instr->op) {
-      case nir_texop_txs:
-      case nir_texop_txf:
-         return nir_type_int;
-
-      default:
-         return nir_type_float;
-      }
-
-   case nir_tex_src_projector:
-   case nir_tex_src_comparitor:
-   case nir_tex_src_bias:
-   case nir_tex_src_ddx:
-   case nir_tex_src_ddy:
-      return nir_type_float;
-
-   case nir_tex_src_offset:
-   case nir_tex_src_ms_index:
-   case nir_tex_src_texture_offset:
-   case nir_tex_src_sampler_offset:
-      return nir_type_int;
-
-   default:
-      unreachable("Invalid texture source type");
-   }
-}
-
 static inline unsigned
 nir_tex_instr_src_size(nir_tex_instr *instr, unsigned src)
 {
@@ -1695,6 +1651,9 @@ typedef struct nir_shader_compiler_options {
    /* lower {slt,sge,seq,sne} to {flt,fge,feq,fne} + b2f: */
    bool lower_scmp;
 
+   /** enables rules to lower idiv by power-of-two: */
+   bool lower_idiv;
+
    /* Does the native fdot instruction replicate its result for four
     * components?  If so, then opt_algebraic_late will turn all fdotN
     * instructions into fdot_replicatedN instructions.
@@ -1764,9 +1723,6 @@ typedef struct nir_shader_info {
    /* Whether or not this shader ever uses textureGather() */
    bool uses_texture_gather;
 
-   /** Whether or not this shader uses nir_intrinsic_interp_var_at_offset */
-   bool uses_interp_var_at_offset;
-
    /* Whether or not this shader uses the gl_ClipDistance output */
    bool uses_clip_distance_out;
 
@@ -1967,10 +1923,6 @@ nir_deref_struct *nir_deref_struct_create(void *mem_ctx, unsigned field_index);
 
 nir_deref *nir_copy_deref(void *mem_ctx, nir_deref *deref);
 
-typedef bool (*nir_deref_foreach_leaf_cb)(nir_deref_var *deref, void *state);
-bool nir_deref_foreach_leaf(nir_deref_var *deref,
-                            nir_deref_foreach_leaf_cb cb, void *state);
-
 nir_load_const_instr *
 nir_deref_get_const_initializer_load(nir_shader *shader, nir_deref_var *deref);
 
@@ -2338,8 +2290,6 @@ bool nir_lower_returns(nir_shader *shader);
 
 bool nir_inline_functions(nir_shader *shader);
 
-bool nir_propagate_invariant(nir_shader *shader);
-
 void nir_lower_var_copy_instr(nir_intrinsic_instr *copy, void *mem_ctx);
 void nir_lower_var_copies(nir_shader *shader);
 
@@ -2388,16 +2338,6 @@ typedef struct nir_lower_tex_options {
     */
    unsigned lower_txp;
 
-   /**
-    * If true, lower away nir_tex_src_offset for all texelfetch instructions.
-    */
-   bool lower_txf_offset;
-
-   /**
-    * If true, lower away nir_tex_src_offset for all rect textures.
-    */
-   bool lower_rect_offset;
-
    /**
     * If true, lower rect textures to 2D, using txs to fetch the
     * texture dimensions and dividing the texture coords by the

src/compiler/nir/nir_algebraic.py

@@ -76,6 +76,7 @@ class Value(object):
          return Constant(val, name_base)
 
    __template = mako.template.Template("""
+#include "compiler/nir/nir_search_helpers.h"
 static const ${val.c_type} ${val.name} = {
    { ${val.type_enum}, ${val.bit_size} },
 % if isinstance(val, Constant):
@@ -84,6 +85,7 @@ static const ${val.c_type} ${val.name} = {
    ${val.index}, /* ${val.var_name} */
    ${'true' if val.is_constant else 'false'},
    ${val.type() or 'nir_type_invalid' },
+   ${val.cond if val.cond else 'NULL'},
 % elif isinstance(val, Expression):
    ${'true' if val.inexact else 'false'},
    nir_op_${val.opcode},
@@ -113,7 +115,7 @@ static const ${val.c_type} ${val.name} = {
                                     Variable=Variable,
                                     Expression=Expression)
 
-_constant_re = re.compile(r"(?P<value>[^@]+)(?:@(?P<bits>\d+))?")
+_constant_re = re.compile(r"(?P<value>[^@\(]+)(?:@(?P<bits>\d+))?")
 
 class Constant(Value):
    def __init__(self, val, name):
@@ -150,7 +152,8 @@ class Constant(Value):
          return "nir_type_float"
 
 _var_name_re = re.compile(r"(?P<const>#)?(?P<name>\w+)"
-                          r"(?:@(?P<type>int|uint|bool|float)?(?P<bits>\d+)?)?")
+                          r"(?:@(?P<type>int|uint|bool|float)?(?P<bits>\d+)?)?"
+                          r"(?P<cond>\([^\)]+\))?")
 
 class Variable(Value):
    def __init__(self, val, name, varset):
@@ -161,6 +164,7 @@ class Variable(Value):
 
       self.var_name = m.group('name')
       self.is_constant = m.group('const') is not None
+      self.cond = m.group('cond')
       self.required_type = m.group('type')
       self.bit_size = int(m.group('bits')) if m.group('bits') else 0
 

src/compiler/nir/nir_builder.h

@@ -317,25 +317,6 @@ nir_fdot(nir_builder *build, nir_ssa_def *src0, nir_ssa_def *src1)
    return NULL;
 }
 
-static inline nir_ssa_def *
-nir_bany_inequal(nir_builder *b, nir_ssa_def *src0, nir_ssa_def *src1)
-{
-   switch (src0->num_components) {
-   case 1: return nir_ine(b, src0, src1);
-   case 2: return nir_bany_inequal2(b, src0, src1);
-   case 3: return nir_bany_inequal3(b, src0, src1);
-   case 4: return nir_bany_inequal4(b, src0, src1);
-   default:
-      unreachable("bad component size");
-   }
-}
-
-static inline nir_ssa_def *
-nir_bany(nir_builder *b, nir_ssa_def *src)
-{
-   return nir_bany_inequal(b, src, nir_imm_int(b, 0));
-}
-
 static inline nir_ssa_def *
 nir_channel(nir_builder *b, nir_ssa_def *def, unsigned c)
 {

src/compiler/nir/nir_gather_info.c

@@ -57,10 +57,6 @@ gather_intrinsic_info(nir_intrinsic_instr *instr, nir_shader *shader)
       shader->info.gs.uses_end_primitive = 1;
       break;
 
-   case nir_intrinsic_interp_var_at_offset:
-      shader->info.uses_interp_var_at_offset = 1;
-      break;
-
    default:
       break;
    }

src/compiler/nir/nir_inline_functions.c

@@ -25,20 +25,6 @@
 #include "nir_builder.h"
 #include "nir_control_flow.h"
 
-static bool
-deref_apply_constant_initializer(nir_deref_var *deref, void *state)
-{
-   struct nir_builder *b = state;
-
-   nir_load_const_instr *initializer =
-      nir_deref_get_const_initializer_load(b->shader, deref);
-   nir_builder_instr_insert(b, &initializer->instr);
-
-   nir_store_deref_var(b, deref, &initializer->def, 0xf);
-
-   return true;
-}
-
 static bool inline_function_impl(nir_function_impl *impl, struct set *inlined);
 
 static void
@@ -188,35 +174,11 @@ inline_functions_block(nir_block *block, nir_builder *b,
       /* Add copies of all in parameters */
       assert(call->num_params == callee_copy->num_params);
 
-      b->cursor = nir_before_instr(&call->instr);
-
-      /* Before we insert the copy of the function, we need to lower away
-       * constant initializers on local variables.  This is because constant
-       * initializers happen (effectively) at the top of the function and,
-       * since these are about to become locals of the calling function,
-       * initialization will happen at the top of the caller rather than at
-       * the top of the callee.  This isn't usually a problem, but if we are
-       * being inlined inside of a loop, it can result in the variable not
-       * getting re-initialized properly for all loop iterations.
-       */
-      nir_foreach_variable(local, &callee_copy->locals) {
-         if (!local->constant_initializer)
-            continue;
-
-         nir_deref_var deref;
-         deref.deref.deref_type = nir_deref_type_var,
-         deref.deref.child = NULL;
-         deref.deref.type = local->type,
-         deref.var = local;
-
-         nir_deref_foreach_leaf(&deref, deref_apply_constant_initializer, b);
-
-         local->constant_initializer = NULL;
-      }
-
       exec_list_append(&b->impl->locals, &callee_copy->locals);
       exec_list_append(&b->impl->registers, &callee_copy->registers);
 
+      b->cursor = nir_before_instr(&call->instr);
+
       /* We now need to tie the two functions together using the
        * parameters.  There are two ways we do this: One is to turn the
        * parameter into a local variable and do a shadow-copy.  The other

src/compiler/nir/nir_intrinsics.h

@@ -41,8 +41,6 @@
 
 #define ARR(...) { __VA_ARGS__ }
 
-INTRINSIC(nop, 0, ARR(0), false, 0, 0, 0, xx, xx, xx,
-          NIR_INTRINSIC_CAN_ELIMINATE)
 
 INTRINSIC(load_var, 0, ARR(0), true, 0, 1, 0, xx, xx, xx, NIR_INTRINSIC_CAN_ELIMINATE)
 INTRINSIC(store_var, 1, ARR(0), false, 0, 1, 1, WRMASK, xx, xx, 0)
@@ -268,16 +266,16 @@ INTRINSIC(ssbo_atomic_comp_swap, 4, ARR(1, 1, 1, 1), true, 1, 0, 0, xx, xx, xx,
  *    in shared_atomic_add, etc).
  * 2: For CompSwap only: the second data parameter.
  */
-INTRINSIC(shared_atomic_add, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_imin, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_umin, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_imax, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_umax, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_and, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_or, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_xor, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_exchange, 2, ARR(1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
-INTRINSIC(shared_atomic_comp_swap, 3, ARR(1, 1, 1), true, 1, 0, 1, BASE, xx, xx, 0)
+INTRINSIC(shared_atomic_add, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_imin, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_umin, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_imax, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_umax, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_and, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_or, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_xor, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_exchange, 2, ARR(1, 1), true, 1, 0, 0, xx, xx, xx, 0)
+INTRINSIC(shared_atomic_comp_swap, 3, ARR(1, 1, 1), true, 1, 0, 0, xx, xx, xx, 0)
 
 #define SYSTEM_VALUE(name, components, num_indices, idx0, idx1, idx2) \
    INTRINSIC(load_##name, 0, ARR(0), true, components, 0, num_indices, \

src/compiler/nir/nir_lower_alu_to_scalar.c

@@ -56,7 +56,6 @@ lower_reduction(nir_alu_instr *instr, nir_op chan_op, nir_op merge_op,
          nir_alu_src_copy(&chan->src[1], &instr->src[1], chan);
          chan->src[1].swizzle[0] = chan->src[1].swizzle[i];
       }
-      chan->exact = instr->exact;
 
       nir_builder_instr_insert(builder, &chan->instr);
 
@@ -230,7 +229,6 @@ lower_alu_instr_scalar(nir_alu_instr *instr, nir_builder *b)
       nir_alu_ssa_dest_init(lower, 1, instr->dest.dest.ssa.bit_size);
       lower->dest.saturate = instr->dest.saturate;
       comps[chan] = &lower->dest.dest.ssa;
-      lower->exact = instr->exact;
 
       nir_builder_instr_insert(b, &lower->instr);
    }
@@ -254,9 +252,6 @@ nir_lower_alu_to_scalar_impl(nir_function_impl *impl)
             lower_alu_instr_scalar(nir_instr_as_alu(instr), &builder);
       }
    }
-
-   nir_metadata_preserve(impl, nir_metadata_block_index |
-                               nir_metadata_dominance);
 }
 
 void

src/compiler/nir/nir_lower_tex.c

@@ -38,39 +38,16 @@
 #include "nir.h"
 #include "nir_builder.h"
 
-static int
-tex_instr_find_src(nir_tex_instr *tex, nir_tex_src_type src_type)
-{
-   for (unsigned i = 0; i < tex->num_srcs; i++) {
-      if (tex->src[i].src_type == src_type)
-         return i;
-   }
-
-   return -1;
-}
-
-static void
-tex_instr_remove_src(nir_tex_instr *tex, unsigned src_idx)
-{
-   assert(src_idx < tex->num_srcs);
-
-   /* First rewrite the source to NIR_SRC_INIT */
-   nir_instr_rewrite_src(&tex->instr, &tex->src[src_idx].src, NIR_SRC_INIT);
-
-   /* Now, move all of the other sources down */
-   for (unsigned i = src_idx + 1; i < tex->num_srcs; i++) {
-      tex->src[i-1].src_type = tex->src[i].src_type;
-      nir_instr_move_src(&tex->instr, &tex->src[i-1].src, &tex->src[i].src);
-   }
-   tex->num_srcs--;
-}
-
 static void
 project_src(nir_builder *b, nir_tex_instr *tex)
 {
    /* Find the projector in the srcs list, if present. */
-   int proj_index = tex_instr_find_src(tex, nir_tex_src_projector);
-   if (proj_index < 0)
+   unsigned proj_index;
+   for (proj_index = 0; proj_index < tex->num_srcs; proj_index++) {
+      if (tex->src[proj_index].src_type == nir_tex_src_projector)
+         break;
+   }
+   if (proj_index == tex->num_srcs)
       return;
 
    b->cursor = nir_before_instr(&tex->instr);
@@ -125,57 +102,18 @@ project_src(nir_builder *b, nir_tex_instr *tex)
                             nir_src_for_ssa(projected));
    }
 
-   tex_instr_remove_src(tex, proj_index);
-}
-
-static bool
-lower_offset(nir_builder *b, nir_tex_instr *tex)
-{
-   int offset_index = tex_instr_find_src(tex, nir_tex_src_offset);
-   if (offset_index < 0)
-      return false;
-
-   int coord_index = tex_instr_find_src(tex, nir_tex_src_coord);
-   assert(coord_index >= 0);
-
-   assert(tex->src[offset_index].src.is_ssa);
-   assert(tex->src[coord_index].src.is_ssa);
-   nir_ssa_def *offset = tex->src[offset_index].src.ssa;
-   nir_ssa_def *coord = tex->src[coord_index].src.ssa;
-
-   b->cursor = nir_before_instr(&tex->instr);
-
-   nir_ssa_def *offset_coord;
-   if (nir_tex_instr_src_type(tex, coord_index) == nir_type_float) {
-      assert(tex->sampler_dim == GLSL_SAMPLER_DIM_RECT);
-      offset_coord = nir_fadd(b, coord, nir_i2f(b, offset));
-   } else {
-      offset_coord = nir_iadd(b, coord, offset);
+   /* Now move the later tex sources down the array so that the projector
+    * disappears.
+    */
+   nir_instr_rewrite_src(&tex->instr, &tex->src[proj_index].src,
+                         NIR_SRC_INIT);
+   for (unsigned i = proj_index + 1; i < tex->num_srcs; i++) {
+      tex->src[i-1].src_type = tex->src[i].src_type;
+      nir_instr_move_src(&tex->instr, &tex->src[i-1].src, &tex->src[i].src);
    }
-
-   if (tex->is_array) {
-      /* The offset is not applied to the array index */
-      if (tex->coord_components == 2) {
-         offset_coord = nir_vec2(b, nir_channel(b, offset_coord, 0),
-                                    nir_channel(b, coord, 1));
-      } else if (tex->coord_components == 3) {
-         offset_coord = nir_vec3(b, nir_channel(b, offset_coord, 0),
-                                    nir_channel(b, offset_coord, 1),
-                                    nir_channel(b, coord, 2));
-      } else {
-         unreachable("Invalid number of components");
-      }
-   }
-
-   nir_instr_rewrite_src(&tex->instr, &tex->src[coord_index].src,
-                         nir_src_for_ssa(offset_coord));
-
-   tex_instr_remove_src(tex, offset_index);
-
-   return true;
+   tex->num_srcs--;
 }
 
-
 static nir_ssa_def *
 get_texture_size(nir_builder *b, nir_tex_instr *tex)
 {
@@ -506,12 +444,6 @@ nir_lower_tex_block(nir_block *block, nir_builder *b,
          progress = true;
       }
 
-      if ((tex->op == nir_texop_txf && options->lower_txf_offset) ||
-          (tex->sampler_dim == GLSL_SAMPLER_DIM_RECT &&
-           options->lower_rect_offset)) {
-         progress = lower_offset(b, tex) || progress;
-      }
-
       if ((tex->sampler_dim == GLSL_SAMPLER_DIM_RECT) && options->lower_rect) {
          lower_rect(b, tex);
          progress = true;

src/compiler/nir/nir_lower_vars_to_ssa.c

@@ -471,7 +471,7 @@ lower_copies_to_load_store(struct deref_node *node,
    return true;
 }
 
-/* Performs variable renaming
+/* Performs variable renaming by doing a DFS of the dominance tree
  *
  * This algorithm is very similar to the one outlined in "Efficiently
  * Computing Static Single Assignment Form and the Control Dependence
@@ -479,132 +479,133 @@ lower_copies_to_load_store(struct deref_node *node,
  * SSA def on the stack per block.
  */
 static bool
-rename_variables(struct lower_variables_state *state)
+rename_variables_block(nir_block *block, struct lower_variables_state *state)
 {
    nir_builder b;
    nir_builder_init(&b, state->impl);
 
-   nir_foreach_block(block, state->impl) {
-      nir_foreach_instr_safe(instr, block) {
-         if (instr->type != nir_instr_type_intrinsic)
-            continue;
+   nir_foreach_instr_safe(instr, block) {
+      if (instr->type != nir_instr_type_intrinsic)
+         continue;
 
-         nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
+      nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
 
-         switch (intrin->intrinsic) {
-         case nir_intrinsic_load_var: {
-            struct deref_node *node =
-               get_deref_node(intrin->variables[0], state);
+      switch (intrin->intrinsic) {
+      case nir_intrinsic_load_var: {
+         struct deref_node *node =
+            get_deref_node(intrin->variables[0], state);
 
-            if (node == NULL) {
-               /* If we hit this path then we are referencing an invalid
-                * value.  Most likely, we unrolled something and are
-                * reading past the end of some array.  In any case, this
-                * should result in an undefined value.
-                */
-               nir_ssa_undef_instr *undef =
-                  nir_ssa_undef_instr_create(state->shader,
-                                             intrin->num_components,
-                                             intrin->dest.ssa.bit_size);
+         if (node == NULL) {
+            /* If we hit this path then we are referencing an invalid
+             * value.  Most likely, we unrolled something and are
+             * reading past the end of some array.  In any case, this
+             * should result in an undefined value.
+             */
+            nir_ssa_undef_instr *undef =
+               nir_ssa_undef_instr_create(state->shader,
+                                          intrin->num_components,
+                                          intrin->dest.ssa.bit_size);
 
-               nir_instr_insert_before(&intrin->instr, &undef->instr);
-               nir_instr_remove(&intrin->instr);
-
-               nir_ssa_def_rewrite_uses(&intrin->dest.ssa,
-                                        nir_src_for_ssa(&undef->def));
-               continue;
-            }
-
-            if (!node->lower_to_ssa)
-               continue;
-
-            nir_alu_instr *mov = nir_alu_instr_create(state->shader,
-                                                      nir_op_imov);
-            mov->src[0].src = nir_src_for_ssa(
-               nir_phi_builder_value_get_block_def(node->pb_value, block));
-            for (unsigned i = intrin->num_components; i < 4; i++)
-               mov->src[0].swizzle[i] = 0;
-
-            assert(intrin->dest.is_ssa);
-
-            mov->dest.write_mask = (1 << intrin->num_components) - 1;
-            nir_ssa_dest_init(&mov->instr, &mov->dest.dest,
-                              intrin->num_components,
-                              intrin->dest.ssa.bit_size, NULL);
-
-            nir_instr_insert_before(&intrin->instr, &mov->instr);
+            nir_instr_insert_before(&intrin->instr, &undef->instr);
             nir_instr_remove(&intrin->instr);
 
             nir_ssa_def_rewrite_uses(&intrin->dest.ssa,
-                                     nir_src_for_ssa(&mov->dest.dest.ssa));
-            break;
+                                     nir_src_for_ssa(&undef->def));
+            continue;
          }
 
-         case nir_intrinsic_store_var: {
-            struct deref_node *node =
-               get_deref_node(intrin->variables[0], state);
+         if (!node->lower_to_ssa)
+            continue;
 
-            if (node == NULL) {
-               /* Probably an out-of-bounds array store.  That should be a
-                * no-op. */
-               nir_instr_remove(&intrin->instr);
-               continue;
-            }
+         nir_alu_instr *mov = nir_alu_instr_create(state->shader,
+                                                   nir_op_imov);
+         mov->src[0].src = nir_src_for_ssa(
+            nir_phi_builder_value_get_block_def(node->pb_value, block));
+         for (unsigned i = intrin->num_components; i < 4; i++)
+            mov->src[0].swizzle[i] = 0;
 
-            if (!node->lower_to_ssa)
-               continue;
+         assert(intrin->dest.is_ssa);
 
-            assert(intrin->num_components ==
-                   glsl_get_vector_elements(node->type));
+         mov->dest.write_mask = (1 << intrin->num_components) - 1;
+         nir_ssa_dest_init(&mov->instr, &mov->dest.dest,
+                           intrin->num_components,
+                           intrin->dest.ssa.bit_size, NULL);
 
-            assert(intrin->src[0].is_ssa);
+         nir_instr_insert_before(&intrin->instr, &mov->instr);
+         nir_instr_remove(&intrin->instr);
 
-            nir_ssa_def *new_def;
-            b.cursor = nir_before_instr(&intrin->instr);
+         nir_ssa_def_rewrite_uses(&intrin->dest.ssa,
+                                  nir_src_for_ssa(&mov->dest.dest.ssa));
+         break;
+      }
 
-            unsigned wrmask = nir_intrinsic_write_mask(intrin);
-            if (wrmask == (1 << intrin->num_components) - 1) {
-               /* Whole variable store - just copy the source.  Note that
-                * intrin->num_components and intrin->src[0].ssa->num_components
-                * may differ.
-                */
-               unsigned swiz[4];
-               for (unsigned i = 0; i < 4; i++)
-                  swiz[i] = i < intrin->num_components ? i : 0;
+      case nir_intrinsic_store_var: {
+         struct deref_node *node =
+            get_deref_node(intrin->variables[0], state);
 
-               new_def = nir_swizzle(&b, intrin->src[0].ssa, swiz,
-                                     intrin->num_components, false);
-            } else {
-               nir_ssa_def *old_def =
-                  nir_phi_builder_value_get_block_def(node->pb_value, block);
-               /* For writemasked store_var intrinsics, we combine the newly
-                * written values with the existing contents of unwritten
-                * channels, creating a new SSA value for the whole vector.
-                */
-               nir_ssa_def *srcs[4];
-               for (unsigned i = 0; i < intrin->num_components; i++) {
-                  if (wrmask & (1 << i)) {
-                     srcs[i] = nir_channel(&b, intrin->src[0].ssa, i);
-                  } else {
-                     srcs[i] = nir_channel(&b, old_def, i);
-                  }
-               }
-               new_def = nir_vec(&b, srcs, intrin->num_components);
-            }
-
-            assert(new_def->num_components == intrin->num_components);
-
-            nir_phi_builder_value_set_block_def(node->pb_value, block, new_def);
+         if (node == NULL) {
+            /* Probably an out-of-bounds array store.  That should be a
+             * no-op. */
             nir_instr_remove(&intrin->instr);
-            break;
+            continue;
          }
 
-         default:
-            break;
+         if (!node->lower_to_ssa)
+            continue;
+
+         assert(intrin->num_components ==
+                glsl_get_vector_elements(node->type));
+
+         assert(intrin->src[0].is_ssa);
+
+         nir_ssa_def *new_def;
+         b.cursor = nir_before_instr(&intrin->instr);
+
+         unsigned wrmask = nir_intrinsic_write_mask(intrin);
+         if (wrmask == (1 << intrin->num_components) - 1) {
+            /* Whole variable store - just copy the source.  Note that
+             * intrin->num_components and intrin->src[0].ssa->num_components
+             * may differ.
+             */
+            unsigned swiz[4];
+            for (unsigned i = 0; i < 4; i++)
+               swiz[i] = i < intrin->num_components ? i : 0;
+
+            new_def = nir_swizzle(&b, intrin->src[0].ssa, swiz,
+                                  intrin->num_components, false);
+         } else {
+            nir_ssa_def *old_def =
+               nir_phi_builder_value_get_block_def(node->pb_value, block);
+            /* For writemasked store_var intrinsics, we combine the newly
+             * written values with the existing contents of unwritten
+             * channels, creating a new SSA value for the whole vector.
+             */
+            nir_ssa_def *srcs[4];
+            for (unsigned i = 0; i < intrin->num_components; i++) {
+               if (wrmask & (1 << i)) {
+                  srcs[i] = nir_channel(&b, intrin->src[0].ssa, i);
+               } else {
+                  srcs[i] = nir_channel(&b, old_def, i);
+               }
+            }
+            new_def = nir_vec(&b, srcs, intrin->num_components);
          }
+
+         assert(new_def->num_components == intrin->num_components);
+
+         nir_phi_builder_value_set_block_def(node->pb_value, block, new_def);
+         nir_instr_remove(&intrin->instr);
+         break;
+      }
+
+      default:
+         break;
       }
    }
 
+   for (unsigned i = 0; i < block->num_dom_children; ++i)
+      rename_variables_block(block->dom_children[i], state);
+
    return true;
 }
 
@@ -736,7 +737,7 @@ nir_lower_vars_to_ssa_impl(nir_function_impl *impl)
       }
    }
 
-   rename_variables(&state);
+   rename_variables_block(nir_start_block(impl), &state);
 
    nir_phi_builder_finish(state.phi_builder);
 

src/compiler/nir/nir_opcodes.py

@@ -257,7 +257,7 @@ unpack_4x8("unorm")
 unpack_2x16("half")
 
 unop_horiz("pack_uvec2_to_uint", 1, tuint32, 2, tuint32, """
-dst.x = (src0.x & 0xffff) | (src0.y << 16);
+dst.x = (src0.x & 0xffff) | (src0.y >> 16);
 """)
 
 unop_horiz("pack_uvec4_to_uint", 1, tuint32, 4, tuint32, """

src/compiler/nir/nir_opt_algebraic.py

@@ -45,10 +45,11 @@ d = 'd'
 # however, be used for backend-requested lowering operations as those need to
 # happen regardless of precision.
 #
-# Variable names are specified as "[#]name[@type]" where "#" inicates that
-# the given variable will only match constants and the type indicates that
+# Variable names are specified as "[#]name[@type][(cond)]" where "#" inicates
+# that the given variable will only match constants and the type indicates that
 # the given variable will only match values from ALU instructions with the
-# given output type.
+# given output type, and (cond) specifies an additional condition function
+# (see nir_search_helpers.h).
 #
 # For constants, you have to be careful to make sure that it is the right
 # type because python is unaware of the source and destination types of the
@@ -62,6 +63,14 @@ d = 'd'
 # constructed value should have that bit-size.
 
 optimizations = [
+
+   (('imul', a, '#b@32(is_pos_power_of_two)'), ('ishl', a, ('find_lsb', b))),
+   (('imul', a, '#b@32(is_neg_power_of_two)'), ('ineg', ('ishl', a, ('find_lsb', ('iabs', b))))),
+   (('udiv', a, '#b@32(is_pos_power_of_two)'), ('ushr', a, ('find_lsb', b))),
+   (('idiv', a, '#b@32(is_pos_power_of_two)'), ('imul', ('isign', a), ('ushr', ('iabs', a), ('find_lsb', b))), 'options->lower_idiv'),
+   (('idiv', a, '#b@32(is_neg_power_of_two)'), ('ineg', ('imul', ('isign', a), ('ushr', ('iabs', a), ('find_lsb', ('iabs', b))))), 'options->lower_idiv'),
+   (('umod', a, '#b(is_pos_power_of_two)'),    ('iand', a, ('isub', b, 1))),
+
    (('fneg', ('fneg', a)), a),
    (('ineg', ('ineg', a)), a),
    (('fabs', ('fabs', a)), ('fabs', a)),
@@ -224,6 +233,8 @@ optimizations = [
    (('~flog2', ('frcp', a)), ('fneg', ('flog2', a))),
    (('~flog2', ('frsq', a)), ('fmul', -0.5, ('flog2', a))),
    (('~flog2', ('fpow', a, b)), ('fmul', b, ('flog2', a))),
+   (('~fadd', ('flog2', a), ('flog2', b)), ('flog2', ('fmul', a, b))),
+   (('~fadd', ('flog2', a), ('fneg', ('flog2', b))), ('flog2', ('fdiv', a, b))),
    (('~fmul', ('fexp2', a), ('fexp2', b)), ('fexp2', ('fadd', a, b))),
    # Division and reciprocal
    (('~fdiv', 1.0, a), ('frcp', a)),

src/compiler/nir/nir_phi_builder.h

@@ -44,8 +44,7 @@
  *         var.pb_val = nir_phi_builder_add_value(pb, var.defs)
  *
  *     // Visit each block.  This needs to visit dominators first;
- *     // nir_foreach_block() will be ok.
- *
+ *     // nir_for_each_block() will be ok.
  *     foreach block:
  *         foreach instruction:
  *             foreach use of variable var:

src/compiler/nir/nir_propagate_invariant.c

@@ -1,196 +0,0 @@
-/*
- * Copyright © 2016 Intel Corporation
- *
- * Permission is hereby granted, free of charge, to any person obtaining a
- * copy of this software and associated documentation files (the "Software"),
- * to deal in the Software without restriction, including without limitation
- * the rights to use, copy, modify, merge, publish, distribute, sublicense,
- * and/or sell copies of the Software, and to permit persons to whom the
- * Software is furnished to do so, subject to the following conditions:
- *
- * The above copyright notice and this permission notice (including the next
- * paragraph) shall be included in all copies or substantial portions of the
- * Software.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
- * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
- * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
- * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
- * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
- * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
- * DEALINGS IN THE SOFTWARE.
- */
-
-#include "nir.h"
-
-static void
-add_src(nir_src *src, struct set *invariants)
-{
-   if (src->is_ssa) {
-      _mesa_set_add(invariants, src->ssa);
-   } else {
-      _mesa_set_add(invariants, src->reg.reg);
-   }
-}
-
-static bool
-add_src_cb(nir_src *src, void *state)
-{
-   add_src(src, state);
-   return true;
-}
-
-static bool
-dest_is_invariant(nir_dest *dest, struct set *invariants)
-{
-   if (dest->is_ssa) {
-      return _mesa_set_search(invariants, &dest->ssa);
-   } else {
-      return _mesa_set_search(invariants, dest->reg.reg);
-   }
-}
-
-static void
-add_cf_node(nir_cf_node *cf, struct set *invariants)
-{
-   if (cf->type == nir_cf_node_if) {
-      nir_if *if_stmt = nir_cf_node_as_if(cf);
-      add_src(&if_stmt->condition, invariants);
-   }
-
-   if (cf->parent)
-      add_cf_node(cf->parent, invariants);
-}
-
-static void
-add_var(nir_variable *var, struct set *invariants)
-{
-   _mesa_set_add(invariants, var);
-}
-
-static bool
-var_is_invariant(nir_variable *var, struct set * invariants)
-{
-   return var->data.invariant || _mesa_set_search(invariants, var);
-}
-
-static void
-propagate_invariant_instr(nir_instr *instr, struct set *invariants)
-{
-   switch (instr->type) {
-   case nir_instr_type_alu: {
-      nir_alu_instr *alu = nir_instr_as_alu(instr);
-      if (!dest_is_invariant(&alu->dest.dest, invariants))
-         break;
-
-      alu->exact = true;
-      nir_foreach_src(instr, add_src_cb, invariants);
-      break;
-   }
-
-   case nir_instr_type_tex: {
-      nir_tex_instr *tex = nir_instr_as_tex(instr);
-      if (dest_is_invariant(&tex->dest, invariants))
-         nir_foreach_src(instr, add_src_cb, invariants);
-      break;
-   }
-
-   case nir_instr_type_intrinsic: {
-      nir_intrinsic_instr *intrin = nir_instr_as_intrinsic(instr);
-      switch (intrin->intrinsic) {
-      case nir_intrinsic_copy_var:
-         /* If the destination is invariant then so is the source */
-         if (var_is_invariant(intrin->variables[0]->var, invariants))
-            add_var(intrin->variables[1]->var, invariants);
-         break;
-
-      case nir_intrinsic_load_var:
-         if (dest_is_invariant(&intrin->dest, invariants))
-            add_var(intrin->variables[0]->var, invariants);
-         break;
-
-      case nir_intrinsic_store_var:
-         if (var_is_invariant(intrin->variables[0]->var, invariants))
-            add_src(&intrin->src[0], invariants);
-         break;
-
-      default:
-         /* Nothing to do */
-         break;
-      }
-   }
-
-   case nir_instr_type_jump:
-   case nir_instr_type_ssa_undef:
-   case nir_instr_type_load_const:
-      break; /* Nothing to do */
-
-   case nir_instr_type_phi: {
-      nir_phi_instr *phi = nir_instr_as_phi(instr);
-      if (!dest_is_invariant(&phi->dest, invariants))
-         break;
-
-      nir_foreach_phi_src(src, phi) {
-         add_src(&src->src, invariants);
-         add_cf_node(&src->pred->cf_node, invariants);
-      }
-      break;
-   }
-
-   case nir_instr_type_call:
-      unreachable("This pass must be run after function inlining");
-
-   case nir_instr_type_parallel_copy:
-   default:
-      unreachable("Cannot have this instruction type");
-   }
-}
-
-static bool
-propagate_invariant_impl(nir_function_impl *impl, struct set *invariants)
-{
-   bool progress = false;
-
-   while (true) {
-      uint32_t prev_entries = invariants->entries;
-
-      nir_foreach_block_reverse(block, impl) {
-         nir_foreach_instr_reverse(instr, block)
-            propagate_invariant_instr(instr, invariants);
-      }
-
-      /* Keep running until we make no more progress. */
-      if (invariants->entries > prev_entries) {
-         progress = true;
-         continue;
-      } else {
-         break;
-      }
-   }
-
-   if (progress) {
-      nir_metadata_preserve(impl, nir_metadata_block_index |
-                                  nir_metadata_dominance |
-                                  nir_metadata_live_ssa_defs);
-   }
-
-   return progress;
-}
-
-bool
-nir_propagate_invariant(nir_shader *shader)
-{
-   /* Hash set of invariant things */
-   struct set *invariants = _mesa_set_create(NULL, _mesa_hash_pointer,
-                                             _mesa_key_pointer_equal);
-
-   bool progress = false;
-   nir_foreach_function(function, shader) {
-      if (function->impl && propagate_invariant_impl(function->impl, invariants))
-         progress = true;
-   }
-
-   _mesa_set_destroy(invariants, NULL);
-
-   return progress;
-}

src/compiler/nir/nir_search.c

@@ -127,6 +127,9 @@ match_value(const nir_search_value *value, nir_alu_instr *instr, unsigned src,
              instr->src[src].src.ssa->parent_instr->type != nir_instr_type_load_const)
             return false;
 
+         if (var->cond && !var->cond(instr, src, num_components, new_swizzle))
+            return false;
+
          if (var->type != nir_type_invalid) {
             if (instr->src[src].src.ssa->parent_instr->type != nir_instr_type_alu)
                return false;

src/compiler/nir/nir_search.h

@@ -68,6 +68,16 @@ typedef struct {
     * never match anything.
     */
    nir_alu_type type;
+
+   /** Optional condition fxn ptr
+    *
+    * This is only allowed in search expressions, and allows additional
+    * constraints to be placed on the match.  Typically used for 'is_constant'
+    * variables to require, for example, power-of-two in order for the search
+    * to match.
+    */
+   bool (*cond)(nir_alu_instr *instr, unsigned src,
+                unsigned num_components, const uint8_t *swizzle);
 } nir_search_variable;
 
 typedef struct {

src/compiler/nir/nir_search_helpers.h

@@ -0,0 +1,94 @@
+/*
+ * Copyright © 2016 Red Hat
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ * Authors:
+ *    Rob Clark <robclark@freedesktop.org>
+ */
+
+#ifndef _NIR_SEARCH_HELPERS_
+#define _NIR_SEARCH_HELPERS_
+
+#include "nir.h"
+
+static inline bool
+__is_power_of_two(unsigned int x)
+{
+   return ((x != 0) && !(x & (x - 1)));
+}
+
+static inline bool
+is_pos_power_of_two(nir_alu_instr *instr, unsigned src, unsigned num_components,
+                    const uint8_t *swizzle)
+{
+   nir_const_value *val = nir_src_as_const_value(instr->src[src].src);
+
+   /* only constant src's: */
+   if (!val)
+      return false;
+
+   for (unsigned i = 0; i < num_components; i++) {
+      switch (nir_op_infos[instr->op].input_types[src]) {
+      case nir_type_int:
+         if (val->i32[swizzle[i]] < 0)
+            return false;
+         if (!__is_power_of_two(val->i32[swizzle[i]]))
+            return false;
+         break;
+      case nir_type_uint:
+         if (!__is_power_of_two(val->u32[swizzle[i]]))
+            return false;
+         break;
+      default:
+         return false;
+      }
+   }
+
+   return true;
+}
+
+static inline bool
+is_neg_power_of_two(nir_alu_instr *instr, unsigned src, unsigned num_components,
+                    const uint8_t *swizzle)
+{
+   nir_const_value *val = nir_src_as_const_value(instr->src[src].src);
+
+   /* only constant src's: */
+   if (!val)
+      return false;
+
+   for (unsigned i = 0; i < num_components; i++) {
+      switch (nir_op_infos[instr->op].input_types[src]) {
+      case nir_type_int:
+         if (val->i32[swizzle[i]] > 0)
+            return false;
+         if (!__is_power_of_two(abs(val->i32[swizzle[i]])))
+            return false;
+         break;
+      default:
+         return false;
+      }
+   }
+
+   return true;
+}
+
+#endif /* _NIR_SEARCH_ */

src/compiler/spirv/spirv_to_nir.c

@@ -1335,9 +1335,54 @@ vtn_handle_texture(struct vtn_builder *b, SpvOp opcode,
    } else {
       image_type = sampled.sampler->var->var->interface_type;
    }
-   const enum glsl_sampler_dim sampler_dim = glsl_get_sampler_dim(image_type);
-   const bool is_array = glsl_sampler_type_is_array(image_type);
-   const bool is_shadow = glsl_sampler_type_is_shadow(image_type);
+
+   nir_tex_src srcs[8]; /* 8 should be enough */
+   nir_tex_src *p = srcs;
+
+   unsigned idx = 4;
+
+   bool has_coord = false;
+   switch (opcode) {
+   case SpvOpImageSampleImplicitLod:
+   case SpvOpImageSampleExplicitLod:
+   case SpvOpImageSampleDrefImplicitLod:
+   case SpvOpImageSampleDrefExplicitLod:
+   case SpvOpImageSampleProjImplicitLod:
+   case SpvOpImageSampleProjExplicitLod:
+   case SpvOpImageSampleProjDrefImplicitLod:
+   case SpvOpImageSampleProjDrefExplicitLod:
+   case SpvOpImageFetch:
+   case SpvOpImageGather:
+   case SpvOpImageDrefGather:
+   case SpvOpImageQueryLod: {
+      /* All these types have the coordinate as their first real argument */
+      struct vtn_ssa_value *coord = vtn_ssa_value(b, w[idx++]);
+      has_coord = true;
+      p->src = nir_src_for_ssa(coord->def);
+      p->src_type = nir_tex_src_coord;
+      p++;
+      break;
+   }
+
+   default:
+      break;
+   }
+
+   /* These all have an explicit depth value as their next source */
+   switch (opcode) {
+   case SpvOpImageSampleDrefImplicitLod:
+   case SpvOpImageSampleDrefExplicitLod:
+   case SpvOpImageSampleProjDrefImplicitLod:
+   case SpvOpImageSampleProjDrefExplicitLod:
+      (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_comparitor);
+      break;
+   default:
+      break;
+   }
+
+   /* For OpImageQuerySizeLod, we always have an LOD */
+   if (opcode == SpvOpImageQuerySizeLod)
+      (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_lod);
 
    /* Figure out the base texture operation */
    nir_texop texop;
@@ -1383,108 +1428,10 @@ vtn_handle_texture(struct vtn_builder *b, SpvOp opcode,
       break;
 
    case SpvOpImageQuerySamples:
-      texop = nir_texop_texture_samples;
-      break;
-
    default:
       unreachable("Unhandled opcode");
    }
 
-   nir_tex_src srcs[8]; /* 8 should be enough */
-   nir_tex_src *p = srcs;
-
-   unsigned idx = 4;
-
-   struct nir_ssa_def *coord;
-   unsigned coord_components;
-   switch (opcode) {
-   case SpvOpImageSampleImplicitLod:
-   case SpvOpImageSampleExplicitLod:
-   case SpvOpImageSampleDrefImplicitLod:
-   case SpvOpImageSampleDrefExplicitLod:
-   case SpvOpImageSampleProjImplicitLod:
-   case SpvOpImageSampleProjExplicitLod:
-   case SpvOpImageSampleProjDrefImplicitLod:
-   case SpvOpImageSampleProjDrefExplicitLod:
-   case SpvOpImageFetch:
-   case SpvOpImageGather:
-   case SpvOpImageDrefGather:
-   case SpvOpImageQueryLod: {
-      /* All these types have the coordinate as their first real argument */
-      switch (sampler_dim) {
-      case GLSL_SAMPLER_DIM_1D:
-      case GLSL_SAMPLER_DIM_BUF:
-         coord_components = 1;
-         break;
-      case GLSL_SAMPLER_DIM_2D:
-      case GLSL_SAMPLER_DIM_RECT:
-      case GLSL_SAMPLER_DIM_MS:
-         coord_components = 2;
-         break;
-      case GLSL_SAMPLER_DIM_3D:
-      case GLSL_SAMPLER_DIM_CUBE:
-         coord_components = 3;
-         break;
-      default:
-         assert("Invalid sampler type");
-      }
-
-      if (is_array && texop != nir_texop_lod)
-         coord_components++;
-
-      coord = vtn_ssa_value(b, w[idx++])->def;
-      p->src = nir_src_for_ssa(coord);
-      p->src_type = nir_tex_src_coord;
-      p++;
-      break;
-   }
-
-   default:
-      coord = NULL;
-      coord_components = 0;
-      break;
-   }
-
-   switch (opcode) {
-   case SpvOpImageSampleProjImplicitLod:
-   case SpvOpImageSampleProjExplicitLod:
-   case SpvOpImageSampleProjDrefImplicitLod:
-   case SpvOpImageSampleProjDrefExplicitLod:
-      /* These have the projector as the last coordinate component */
-      p->src = nir_src_for_ssa(nir_channel(&b->nb, coord, coord_components));
-      p->src_type = nir_tex_src_projector;
-      p++;
-      break;
-
-   default:
-      break;
-   }
-
-   unsigned gather_component = 0;
-   switch (opcode) {
-   case SpvOpImageSampleDrefImplicitLod:
-   case SpvOpImageSampleDrefExplicitLod:
-   case SpvOpImageSampleProjDrefImplicitLod:
-   case SpvOpImageSampleProjDrefExplicitLod:
-   case SpvOpImageDrefGather:
-      /* These all have an explicit depth value as their next source */
-      (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_comparitor);
-      break;
-
-   case SpvOpImageGather:
-      /* This has a component as its next source */
-      gather_component =
-         vtn_value(b, w[idx++], vtn_value_type_constant)->constant->value.u[0];
-      break;
-
-   default:
-      break;
-   }
-
-   /* For OpImageQuerySizeLod, we always have an LOD */
-   if (opcode == SpvOpImageQuerySizeLod)
-      (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_lod);
-
    /* Now we need to handle some number of optional arguments */
    if (idx < count) {
       uint32_t operands = w[idx++];
@@ -1497,12 +1444,12 @@ vtn_handle_texture(struct vtn_builder *b, SpvOp opcode,
 
       if (operands & SpvImageOperandsLodMask) {
          assert(texop == nir_texop_txl || texop == nir_texop_txf ||
-                texop == nir_texop_txs);
+                texop == nir_texop_txf_ms || texop == nir_texop_txs);
          (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_lod);
       }
 
       if (operands & SpvImageOperandsGradMask) {
-         assert(texop == nir_texop_txl);
+         assert(texop == nir_texop_tex);
          texop = nir_texop_txd;
          (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_ddx);
          (*p++) = vtn_tex_src(b, w[idx++], nir_tex_src_ddy);
@@ -1529,13 +1476,35 @@ vtn_handle_texture(struct vtn_builder *b, SpvOp opcode,
 
    memcpy(instr->src, srcs, instr->num_srcs * sizeof(*instr->src));
 
-   instr->coord_components = coord_components;
-   instr->sampler_dim = sampler_dim;
-   instr->is_array = is_array;
-   instr->is_shadow = is_shadow;
-   instr->is_new_style_shadow =
-      is_shadow && glsl_get_components(ret_type->type) == 1;
-   instr->component = gather_component;
+   instr->sampler_dim = glsl_get_sampler_dim(image_type);
+   instr->is_array = glsl_sampler_type_is_array(image_type);
+   instr->is_shadow = glsl_sampler_type_is_shadow(image_type);
+   instr->is_new_style_shadow = instr->is_shadow;
+
+   if (has_coord) {
+      switch (instr->sampler_dim) {
+      case GLSL_SAMPLER_DIM_1D:
+      case GLSL_SAMPLER_DIM_BUF:
+         instr->coord_components = 1;
+         break;
+      case GLSL_SAMPLER_DIM_2D:
+      case GLSL_SAMPLER_DIM_RECT:
+      case GLSL_SAMPLER_DIM_MS:
+         instr->coord_components = 2;
+         break;
+      case GLSL_SAMPLER_DIM_3D:
+      case GLSL_SAMPLER_DIM_CUBE:
+         instr->coord_components = 3;
+         break;
+      default:
+         assert("Invalid sampler type");
+      }
+
+      if (instr->is_array)
+         instr->coord_components++;
+   } else {
+      instr->coord_components = 0;
+   }
 
    switch (glsl_get_sampler_result_type(image_type)) {
    case GLSL_TYPE_FLOAT:   instr->dest_type = nir_type_float;     break;
@@ -1749,8 +1718,8 @@ vtn_handle_image(struct vtn_builder *b, SpvOp opcode,
       break;
 
    case SpvOpAtomicCompareExchange:
-      intrin->src[2] = nir_src_for_ssa(vtn_ssa_value(b, w[8])->def);
-      intrin->src[3] = nir_src_for_ssa(vtn_ssa_value(b, w[7])->def);
+      intrin->src[2] = nir_src_for_ssa(vtn_ssa_value(b, w[7])->def);
+      intrin->src[3] = nir_src_for_ssa(vtn_ssa_value(b, w[6])->def);
       break;
 
    case SpvOpAtomicISub:
@@ -1847,8 +1816,8 @@ fill_common_atomic_sources(struct vtn_builder *b, SpvOp opcode,
       break;
 
    case SpvOpAtomicCompareExchange:
-      src[0] = nir_src_for_ssa(vtn_ssa_value(b, w[8])->def);
-      src[1] = nir_src_for_ssa(vtn_ssa_value(b, w[7])->def);
+      src[0] = nir_src_for_ssa(vtn_ssa_value(b, w[7])->def);
+      src[1] = nir_src_for_ssa(vtn_ssa_value(b, w[8])->def);
       break;
       /* Fall through */
 

src/compiler/spirv/vtn_cfg.c

@@ -239,12 +239,12 @@ vtn_get_branch_type(struct vtn_block *block,
              swcase->fallthrough == block->switch_case);
       swcase->fallthrough = block->switch_case;
       return vtn_branch_type_switch_fallthrough;
+   } else if (block == switch_break) {
+      return vtn_branch_type_switch_break;
    } else if (block == loop_break) {
       return vtn_branch_type_loop_break;
    } else if (block == loop_cont) {
       return vtn_branch_type_loop_continue;
-   } else if (block == switch_break) {
-      return vtn_branch_type_switch_break;
    } else {
       return vtn_branch_type_none;
    }
@@ -443,19 +443,6 @@ vtn_cfg_walk_blocks(struct vtn_builder *b, struct list_head *cf_list,
             vtn_order_case(swtch, case_block->switch_case);
          }
 
-         enum vtn_branch_type branch_type =
-            vtn_get_branch_type(break_block, switch_case, NULL,
-                                loop_break, loop_cont);
-
-         if (branch_type != vtn_branch_type_none) {
-            /* It is possible that the break is actually the continue block
-             * for the containing loop.  In this case, we need to bail and let
-             * the loop parsing code handle the continue properly.
-             */
-            assert(branch_type == vtn_branch_type_loop_continue);
-            return;
-         }
-
          block = break_block;
          continue;
       }
@@ -527,12 +514,11 @@ vtn_handle_phi_second_pass(struct vtn_builder *b, SpvOp opcode,
    nir_variable *phi_var = phi_entry->data;
 
    for (unsigned i = 3; i < count; i += 2) {
+      struct vtn_ssa_value *src = vtn_ssa_value(b, w[i]);
       struct vtn_block *pred =
          vtn_value(b, w[i + 1], vtn_value_type_block)->block;
 
-      b->nb.cursor = nir_after_instr(&pred->end_nop->instr);
-
-      struct vtn_ssa_value *src = vtn_ssa_value(b, w[i]);
+      b->nb.cursor = nir_after_block_before_jump(pred->end_block);
 
       vtn_local_store(b, src, nir_deref_var_create(b, phi_var));
    }
@@ -590,9 +576,7 @@ vtn_emit_cf_list(struct vtn_builder *b, struct list_head *cf_list,
 
          vtn_foreach_instruction(b, block_start, block_end, handler);
 
-         block->end_nop = nir_intrinsic_instr_create(b->nb.shader,
-                                                     nir_intrinsic_nop);
-         nir_builder_instr_insert(&b->nb, &block->end_nop->instr);
+         block->end_block = nir_cursor_current_block(b->nb.cursor);
 
          if ((*block->branch & SpvOpCodeMask) == SpvOpReturnValue) {
             struct vtn_ssa_value *src = vtn_ssa_value(b, block->branch[1]);

src/compiler/spirv/vtn_private.h

@@ -149,8 +149,8 @@ struct vtn_block {
    /** Points to the switch case started by this block (if any) */
    struct vtn_case *switch_case;
 
-   /** Every block ends in a nop intrinsic so that we can find it again */
-   nir_intrinsic_instr *end_nop;
+   /** The last block in this SPIR-V block. */
+   nir_block *end_block;
 };
 
 struct vtn_function {

src/compiler/spirv/vtn_variables.c

@@ -839,8 +839,8 @@ vtn_get_builtin_location(struct vtn_builder *b,
       assert(*mode == nir_var_shader_in);
       break;
    case SpvBuiltInFrontFacing:
-      *location = SYSTEM_VALUE_FRONT_FACE;
-      set_mode_system_value(mode);
+      *location = VARYING_SLOT_FACE;
+      assert(*mode == nir_var_shader_in);
       break;
    case SpvBuiltInSampleId:
       *location = SYSTEM_VALUE_SAMPLE_ID;
@@ -889,9 +889,81 @@ vtn_get_builtin_location(struct vtn_builder *b,
 }
 
 static void
-apply_var_decoration(struct vtn_builder *b, nir_variable *nir_var,
-                     const struct vtn_decoration *dec)
+var_decoration_cb(struct vtn_builder *b, struct vtn_value *val, int member,
+                  const struct vtn_decoration *dec, void *void_var)
 {
+   struct vtn_variable *vtn_var = void_var;
+
+   /* Handle decorations that apply to a vtn_variable as a whole */
+   switch (dec->decoration) {
+   case SpvDecorationBinding:
+      vtn_var->binding = dec->literals[0];
+      return;
+   case SpvDecorationDescriptorSet:
+      vtn_var->descriptor_set = dec->literals[0];
+      return;
+   default:
+      break;
+   }
+
+   /* Now we handle decorations that apply to a particular nir_variable */
+   nir_variable *nir_var = vtn_var->var;
+   if (val->value_type == vtn_value_type_access_chain) {
+      assert(val->access_chain->length == 0);
+      assert(val->access_chain->var == void_var);
+      assert(member == -1);
+   } else {
+      assert(val->value_type == vtn_value_type_type);
+      if (member != -1)
+         nir_var = vtn_var->members[member];
+   }
+
+   /* Location is odd in that it can apply in three different cases: To a
+    * non-split variable, to a whole split variable, or to one structure
+    * member of a split variable.
+    */
+   if (dec->decoration == SpvDecorationLocation) {
+      unsigned location = dec->literals[0];
+      bool is_vertex_input;
+      if (b->shader->stage == MESA_SHADER_FRAGMENT &&
+          vtn_var->mode == vtn_variable_mode_output) {
+         is_vertex_input = false;
+         location += FRAG_RESULT_DATA0;
+      } else if (b->shader->stage == MESA_SHADER_VERTEX &&
+                 vtn_var->mode == vtn_variable_mode_input) {
+         is_vertex_input = true;
+         location += VERT_ATTRIB_GENERIC0;
+      } else if (vtn_var->mode == vtn_variable_mode_input ||
+                 vtn_var->mode == vtn_variable_mode_output) {
+         is_vertex_input = false;
+         location += VARYING_SLOT_VAR0;
+      } else {
+         assert(!"Location must be on input or output variable");
+      }
+
+      if (nir_var) {
+         /* This handles the member and lone variable cases */
+         nir_var->data.location = location;
+         nir_var->data.explicit_location = true;
+      } else {
+         /* This handles the structure member case */
+         assert(vtn_var->members);
+         unsigned length =
+            glsl_get_length(glsl_without_array(vtn_var->type->type));
+         for (unsigned i = 0; i < length; i++) {
+            vtn_var->members[i]->data.location = location;
+            vtn_var->members[i]->data.explicit_location = true;
+            location +=
+               glsl_count_attribute_slots(vtn_var->members[i]->interface_type,
+                                          is_vertex_input);
+         }
+      }
+      return;
+   }
+
+   if (nir_var == NULL)
+      return;
+
    switch (dec->decoration) {
    case SpvDecorationRelaxedPrecision:
       break; /* FIXME: Do nothing with this for now. */
@@ -1008,99 +1080,6 @@ apply_var_decoration(struct vtn_builder *b, nir_variable *nir_var,
    }
 }
 
-static void
-var_decoration_cb(struct vtn_builder *b, struct vtn_value *val, int member,
-                  const struct vtn_decoration *dec, void *void_var)
-{
-   struct vtn_variable *vtn_var = void_var;
-
-   /* Handle decorations that apply to a vtn_variable as a whole */
-   switch (dec->decoration) {
-   case SpvDecorationBinding:
-      vtn_var->binding = dec->literals[0];
-      return;
-   case SpvDecorationDescriptorSet:
-      vtn_var->descriptor_set = dec->literals[0];
-      return;
-   default:
-      break;
-   }
-
-   if (val->value_type == vtn_value_type_access_chain) {
-      assert(val->access_chain->length == 0);
-      assert(val->access_chain->var == void_var);
-      assert(member == -1);
-   } else {
-      assert(val->value_type == vtn_value_type_type);
-   }
-
-   /* Location is odd.  If applied to a split structure, we have to walk the
-    * whole thing and accumulate the location.  It's easier to handle as a
-    * special case.
-    */
-   if (dec->decoration == SpvDecorationLocation) {
-      unsigned location = dec->literals[0];
-      bool is_vertex_input;
-      if (b->shader->stage == MESA_SHADER_FRAGMENT &&
-          vtn_var->mode == vtn_variable_mode_output) {
-         is_vertex_input = false;
-         location += FRAG_RESULT_DATA0;
-      } else if (b->shader->stage == MESA_SHADER_VERTEX &&
-                 vtn_var->mode == vtn_variable_mode_input) {
-         is_vertex_input = true;
-         location += VERT_ATTRIB_GENERIC0;
-      } else if (vtn_var->mode == vtn_variable_mode_input ||
-                 vtn_var->mode == vtn_variable_mode_output) {
-         is_vertex_input = false;
-         location += VARYING_SLOT_VAR0;
-      } else {
-         assert(!"Location must be on input or output variable");
-      }
-
-      if (vtn_var->var) {
-         /* This handles the member and lone variable cases */
-         vtn_var->var->data.location = location;
-         vtn_var->var->data.explicit_location = true;
-      } else {
-         /* This handles the structure member case */
-         assert(vtn_var->members);
-         unsigned length =
-            glsl_get_length(glsl_without_array(vtn_var->type->type));
-         for (unsigned i = 0; i < length; i++) {
-            vtn_var->members[i]->data.location = location;
-            vtn_var->members[i]->data.explicit_location = true;
-            location +=
-               glsl_count_attribute_slots(vtn_var->members[i]->interface_type,
-                                          is_vertex_input);
-         }
-      }
-      return;
-   } else {
-      if (vtn_var->var) {
-         assert(member <= 0);
-         apply_var_decoration(b, vtn_var->var, dec);
-      } else if (vtn_var->members) {
-         if (member >= 0) {
-            assert(vtn_var->members);
-            apply_var_decoration(b, vtn_var->members[member], dec);
-         } else {
-            unsigned length =
-               glsl_get_length(glsl_without_array(vtn_var->type->type));
-            for (unsigned i = 0; i < length; i++)
-               apply_var_decoration(b, vtn_var->members[i], dec);
-         }
-      } else {
-         /* A few variables, those with external storage, have no actual
-          * nir_variables associated with them.  Fortunately, all decorations
-          * we care about for those variables are on the type only.
-          */
-         assert(vtn_var->mode == vtn_variable_mode_ubo ||
-                vtn_var->mode == vtn_variable_mode_ssbo ||
-                vtn_var->mode == vtn_variable_mode_push_constant);
-      }
-   }
-}
-
 /* Tries to compute the size of an interface block based on the strides and
  * offsets that are provided to us in the SPIR-V source.
  */
@@ -1194,7 +1173,7 @@ vtn_handle_variables(struct vtn_builder *b, SpvOp opcode,
       case SpvStorageClassPushConstant:
          var->mode = vtn_variable_mode_push_constant;
          assert(b->shader->num_uniforms == 0);
-         b->shader->num_uniforms = vtn_type_block_size(var->type);
+         b->shader->num_uniforms = vtn_type_block_size(var->type) * 4;
          break;
       case SpvStorageClassInput:
          var->mode = vtn_variable_mode_input;

src/egl/Android.mk

@@ -61,12 +61,6 @@ ifeq ($(shell echo "$(MESA_ANDROID_VERSION) >= 4.2" | bc),1)
 LOCAL_SHARED_LIBRARIES += libsync
 endif
 
-# add libdrm if there are hardware drivers
-ifneq ($(filter-out swrast,$(MESA_GPU_DRIVERS)),)
-LOCAL_CFLAGS += -DHAVE_LIBDRM
-LOCAL_SHARED_LIBRARIES += libdrm
-endif
-
 ifeq ($(strip $(MESA_BUILD_CLASSIC)),true)
 # require i915_dri and/or i965_dri
 LOCAL_REQUIRED_MODULES += \

src/egl/drivers/dri2/egl_dri2.c

@@ -242,15 +242,6 @@ dri2_add_config(_EGLDisplay *disp, const __DRIconfig *dri_config, int id,
             return NULL;
          break;
 
-      case __DRI_ATTRIB_MAX_PBUFFER_WIDTH:
-         _eglSetConfigKey(&base, EGL_MAX_PBUFFER_WIDTH,
-                          _EGL_MAX_PBUFFER_WIDTH);
-         break;
-      case __DRI_ATTRIB_MAX_PBUFFER_HEIGHT:
-         _eglSetConfigKey(&base, EGL_MAX_PBUFFER_HEIGHT,
-                          _EGL_MAX_PBUFFER_HEIGHT);
-         break;
-
       default:
 	 key = dri2_to_egl_attribute_map[attrib];
 	 if (key != 0)
@@ -329,15 +320,6 @@ dri2_add_config(_EGLDisplay *disp, const __DRIconfig *dri_config, int id,
       surface_type &= ~EGL_PIXMAP_BIT;
    }
 
-   /* No support for pbuffer + MSAA for now.
-    *
-    * XXX TODO: pbuffer + MSAA does not work and causes crashes.
-    * See QT bugreport: https://bugreports.qt.io/browse/QTBUG-47509
-    */
-   if (base.Samples) {
-      surface_type &= ~EGL_PBUFFER_BIT;
-   }
-
    conf->base.SurfaceType |= surface_type;
 
    return conf;
@@ -775,99 +757,64 @@ dri2_create_screen(_EGLDisplay *disp)
 
 /**
  * Called via eglInitialize(), GLX_drv->API.Initialize().
- *
- * This must be guaranteed to be called exactly once, even if eglInitialize is
- * called many times (without a eglTerminate in between).
  */
 static EGLBoolean
 dri2_initialize(_EGLDriver *drv, _EGLDisplay *disp)
 {
-   EGLBoolean ret = EGL_FALSE;
-   struct dri2_egl_display *dri2_dpy = dri2_egl_display(disp);
-
-   /* In the case where the application calls eglMakeCurrent(context1),
-    * eglTerminate, then eglInitialize again (without a call to eglReleaseThread
-    * or eglMakeCurrent(NULL) before that), dri2_dpy structure is still
-    * initialized, as we need it to be able to free context1 correctly.
-    *
-    * It would probably be safest to forcibly release the display with
-    * dri2_display_release, to make sure the display is reinitialized correctly.
-    * However, the EGL spec states that we need to keep a reference to the
-    * current context (so we cannot call dri2_make_current(NULL)), and therefore
-    * we would leak context1 as we would be missing the old display connection
-    * to free it up correctly.
-    */
-   if (dri2_dpy) {
-      dri2_dpy->ref_count++;
-      return EGL_TRUE;
-   }
-
    /* not until swrast_dri is supported */
    if (disp->Options.UseFallback)
       return EGL_FALSE;
 
-   /* Nothing to initialize for a test only display */
-   if (disp->Options.TestOnly)
-      return EGL_TRUE;
-
    switch (disp->Platform) {
 #ifdef HAVE_SURFACELESS_PLATFORM
    case _EGL_PLATFORM_SURFACELESS:
-      ret = dri2_initialize_surfaceless(drv, disp);
-      break;
+      if (disp->Options.TestOnly)
+         return EGL_TRUE;
+      return dri2_initialize_surfaceless(drv, disp);
 #endif
+
 #ifdef HAVE_X11_PLATFORM
    case _EGL_PLATFORM_X11:
-      ret = dri2_initialize_x11(drv, disp);
-      break;
+      if (disp->Options.TestOnly)
+         return EGL_TRUE;
+      return dri2_initialize_x11(drv, disp);
 #endif
+
 #ifdef HAVE_DRM_PLATFORM
    case _EGL_PLATFORM_DRM:
-      ret = dri2_initialize_drm(drv, disp);
-      break;
+      if (disp->Options.TestOnly)
+         return EGL_TRUE;
+      return dri2_initialize_drm(drv, disp);
 #endif
 #ifdef HAVE_WAYLAND_PLATFORM
    case _EGL_PLATFORM_WAYLAND:
-      ret = dri2_initialize_wayland(drv, disp);
-      break;
+      if (disp->Options.TestOnly)
+         return EGL_TRUE;
+      return dri2_initialize_wayland(drv, disp);
 #endif
 #ifdef HAVE_ANDROID_PLATFORM
    case _EGL_PLATFORM_ANDROID:
-      ret = dri2_initialize_android(drv, disp);
-      break;
+      if (disp->Options.TestOnly)
+         return EGL_TRUE;
+      return dri2_initialize_android(drv, disp);
 #endif
+
    default:
       _eglLog(_EGL_WARNING, "No EGL platform enabled.");
       return EGL_FALSE;
    }
-
-   if (ret) {
-      dri2_dpy = dri2_egl_display(disp);
-
-      if (!dri2_dpy) {
-         return EGL_FALSE;
-      }
-
-      dri2_dpy->ref_count++;
-   }
-
-   return ret;
 }
 
 /**
- * Decrement display reference count, and free up display if necessary.
+ * Called via eglTerminate(), drv->API.Terminate().
  */
-static void
-dri2_display_release(_EGLDisplay *disp) {
+static EGLBoolean
+dri2_terminate(_EGLDriver *drv, _EGLDisplay *disp)
+{
    struct dri2_egl_display *dri2_dpy = dri2_egl_display(disp);
    unsigned i;
 
-   assert(dri2_dpy->ref_count > 0);
-   dri2_dpy->ref_count--;
-
-   if (dri2_dpy->ref_count > 0)
-      return;
-
+   _eglReleaseDisplayResources(drv, disp);
    _eglCleanupDisplay(disp);
 
    if (dri2_dpy->own_dri_screen)
@@ -922,21 +869,6 @@ dri2_display_release(_EGLDisplay *disp) {
    }
    free(dri2_dpy);
    disp->DriverData = NULL;
-}
-
-/**
- * Called via eglTerminate(), drv->API.Terminate().
- *
- * This must be guaranteed to be called exactly once, even if eglTerminate is
- * called many times (without a eglInitialize in between).
- */
-static EGLBoolean
-dri2_terminate(_EGLDriver *drv, _EGLDisplay *disp)
-{
-   /* Release all non-current Context/Surfaces. */
-   _eglReleaseDisplayResources(drv, disp);
-
-   dri2_display_release(disp);
 
    return EGL_TRUE;
 }
@@ -1256,16 +1188,10 @@ dri2_make_current(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *dsurf,
    _EGLSurface *tmp_dsurf, *tmp_rsurf;
    __DRIdrawable *ddraw, *rdraw;
    __DRIcontext *cctx;
-   EGLBoolean unbind;
-
-   if (!dri2_dpy)
-      return _eglError(EGL_NOT_INITIALIZED, "eglMakeCurrent");
 
    /* make new bindings */
-   if (!_eglBindContext(ctx, dsurf, rsurf, &old_ctx, &old_dsurf, &old_rsurf)) {
-      /* _eglBindContext already sets the EGL error (in _eglCheckMakeCurrent) */
+   if (!_eglBindContext(ctx, dsurf, rsurf, &old_ctx, &old_dsurf, &old_rsurf))
       return EGL_FALSE;
-   }
 
    /* flush before context switch */
    if (old_ctx && dri2_drv->glFlush)
@@ -1280,21 +1206,14 @@ dri2_make_current(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *dsurf,
       dri2_dpy->core->unbindContext(old_cctx);
    }
 
-   unbind = (cctx == NULL && ddraw == NULL && rdraw == NULL);
-
-   if (unbind || dri2_dpy->core->bindContext(cctx, ddraw, rdraw)) {
+   if ((cctx == NULL && ddraw == NULL && rdraw == NULL) ||
+       dri2_dpy->core->bindContext(cctx, ddraw, rdraw)) {
       if (old_dsurf)
          drv->API.DestroySurface(drv, disp, old_dsurf);
       if (old_rsurf)
          drv->API.DestroySurface(drv, disp, old_rsurf);
-
-      if (!unbind)
-         dri2_dpy->ref_count++;
-      if (old_ctx) {
-         EGLDisplay old_disp = _eglGetDisplayHandle(old_ctx->Resource.Display);
+      if (old_ctx)
          drv->API.DestroyContext(drv, disp, old_ctx);
-         dri2_display_release(old_disp);
-      }
 
       return EGL_TRUE;
    } else {
@@ -1312,11 +1231,7 @@ dri2_make_current(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *dsurf,
       _eglPutSurface(old_rsurf);
       _eglPutContext(old_ctx);
 
-      /* dri2_dpy->core->bindContext failed. We cannot tell for sure why, but
-       * setting the error to EGL_BAD_MATCH is surely better than leaving it
-       * as EGL_SUCCESS.
-       */
-      return _eglError(EGL_BAD_MATCH, "eglMakeCurrent");
+      return EGL_FALSE;
    }
 }
 

src/egl/drivers/dri2/egl_dri2.h

@@ -80,6 +80,8 @@
 #include "eglimage.h"
 #include "eglsync.h"
 
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
+
 struct wl_buffer;
 
 struct dri2_egl_driver
@@ -175,10 +177,6 @@ struct dri2_egl_display
    const __DRI2interopExtension *interop;
    int                       fd;
 
-   /* dri2_initialize/dri2_terminate increment/decrement this count, so does
-    * dri2_make_current (tracks if there are active contexts/surfaces). */
-   int                       ref_count;
-
    int                       own_device;
    int                       swap_available;
    int                       invalidate_available;

src/egl/drivers/dri2/platform_android.c

@@ -29,7 +29,6 @@
 
 #include <errno.h>
 #include <dlfcn.h>
-#include <fcntl.h>
 #include <xf86drm.h>
 
 #if ANDROID_VERSION >= 0x402
@@ -163,8 +162,6 @@ droid_window_dequeue_buffer(struct dri2_egl_surface *dri2_surf)
 static EGLBoolean
 droid_window_enqueue_buffer(_EGLDisplay *disp, struct dri2_egl_surface *dri2_surf)
 {
-   struct dri2_egl_display *dri2_dpy = dri2_egl_display(disp);
-
    /* To avoid blocking other EGL calls, release the display mutex before
     * we enter droid_window_enqueue_buffer() and re-acquire the mutex upon
     * return.
@@ -195,12 +192,6 @@ droid_window_enqueue_buffer(_EGLDisplay *disp, struct dri2_egl_surface *dri2_sur
    dri2_surf->buffer = NULL;
 
    mtx_lock(&disp->Mutex);
-
-   if (dri2_surf->dri_image) {
-      dri2_dpy->image->destroyImage(dri2_surf->dri_image);
-      dri2_surf->dri_image = NULL;
-   }
-
    return EGL_TRUE;
 }
 
@@ -289,8 +280,6 @@ droid_create_surface(_EGLDriver *drv, _EGLDisplay *disp, EGLint type,
 
    config = dri2_get_dri_config(dri2_conf, EGL_WINDOW_BIT,
                                 dri2_surf->base.GLColorspace);
-   if (!config)
-      goto cleanup_surface;
 
    dri2_surf->dri_drawable =
       (*dri2_dpy->dri2->createNewDrawable)(dri2_dpy->dri_screen, config,
@@ -384,9 +373,6 @@ get_back_bo(struct dri2_egl_surface *dri2_surf)
    int fourcc, pitch;
    int offset = 0, fd;
 
-   if (dri2_surf->dri_image)
-	   return 0;
-
    if (!dri2_surf->buffer)
       return -1;
 
@@ -445,8 +431,10 @@ droid_image_get_buffers(__DRIdrawable *driDrawable,
 static EGLBoolean
 droid_swap_buffers(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *draw)
 {
+   struct dri2_egl_driver *dri2_drv = dri2_egl_driver(drv);
    struct dri2_egl_display *dri2_dpy = dri2_egl_display(disp);
    struct dri2_egl_surface *dri2_surf = dri2_egl_surface(draw);
+   _EGLContext *ctx;
 
    if (dri2_surf->base.Type != EGL_WINDOW_BIT)
       return EGL_TRUE;
@@ -737,7 +725,7 @@ droid_open_device(void)
       fd = -1;
    }
 
-   return (fd >= 0) ? fcntl(fd, F_DUPFD_CLOEXEC, 3) : -1;
+   return (fd >= 0) ? dup(fd) : -1;
 }
 
 /* support versions < JellyBean */
@@ -883,7 +871,6 @@ cleanup_device:
    close(dri2_dpy->fd);
 cleanup_display:
    free(dri2_dpy);
-   dpy->DriverData = NULL;
 
    return _eglError(EGL_NOT_INITIALIZED, err);
 }

src/egl/drivers/dri2/platform_drm.c

@@ -726,6 +726,5 @@ cleanup:
       close(fd);
 
    free(dri2_dpy);
-   disp->DriverData = NULL;
    return EGL_FALSE;
 }

src/egl/drivers/dri2/platform_surfaceless.c

@@ -157,7 +157,6 @@ cleanup_driver:
    close(dri2_dpy->fd);
 cleanup_display:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return _eglError(EGL_NOT_INITIALIZED, err);
 }

src/egl/drivers/dri2/platform_wayland.c

@@ -118,13 +118,6 @@ resize_callback(struct wl_egl_window *wl_win, void *data)
    (*dri2_dpy->flush->invalidate)(dri2_surf->dri_drawable);
 }
 
-static void
-destroy_window_callback(void *data)
-{
-   struct dri2_egl_surface *dri2_surf = data;
-   dri2_surf->wl_win = NULL;
-}
-
 /**
  * Called via eglCreateWindowSurface(), drv->API.CreateWindowSurface().
  */
@@ -166,7 +159,6 @@ dri2_wl_create_surface(_EGLDriver *drv, _EGLDisplay *disp,
 
    dri2_surf->wl_win->private = dri2_surf;
    dri2_surf->wl_win->resize_callback = resize_callback;
-   dri2_surf->wl_win->destroy_window_callback = destroy_window_callback;
 
    dri2_surf->base.Width =  -1;
    dri2_surf->base.Height = -1;
@@ -265,11 +257,8 @@ dri2_wl_destroy_surface(_EGLDriver *drv, _EGLDisplay *disp, _EGLSurface *surf)
    if (dri2_surf->throttle_callback)
       wl_callback_destroy(dri2_surf->throttle_callback);
 
-   if (dri2_surf->wl_win) {
-      dri2_surf->wl_win->private = NULL;
-      dri2_surf->wl_win->resize_callback = NULL;
-      dri2_surf->wl_win->destroy_window_callback = NULL;
-   }
+   dri2_surf->wl_win->private = NULL;
+   dri2_surf->wl_win->resize_callback = NULL;
 
    free(surf);
 
@@ -1249,7 +1238,6 @@ dri2_initialize_wayland_drm(_EGLDriver *drv, _EGLDisplay *disp)
    wl_event_queue_destroy(dri2_dpy->wl_queue);
  cleanup_dpy:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return EGL_FALSE;
 }
@@ -1706,8 +1694,6 @@ dri2_wl_swrast_create_window_surface(_EGLDriver *drv, _EGLDisplay *disp,
       dri2_surf->format = WL_SHM_FORMAT_ARGB8888;
 
    dri2_surf->wl_win = window;
-   dri2_surf->wl_win->private = dri2_surf;
-   dri2_surf->wl_win->destroy_window_callback = destroy_window_callback;
 
    dri2_surf->base.Width = -1;
    dri2_surf->base.Height = -1;
@@ -1897,7 +1883,6 @@ dri2_initialize_wayland_swrast(_EGLDriver *drv, _EGLDisplay *disp)
    wl_event_queue_destroy(dri2_dpy->wl_queue);
  cleanup_dpy:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return EGL_FALSE;
 }

src/egl/drivers/dri2/platform_x11.c

@@ -1231,7 +1231,6 @@ dri2_initialize_x11_swrast(_EGLDriver *drv, _EGLDisplay *disp)
       xcb_disconnect(dri2_dpy->conn);
  cleanup_dpy:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return EGL_FALSE;
 }
@@ -1303,13 +1302,15 @@ dri2_initialize_x11_dri3(_EGLDriver *drv, _EGLDisplay *disp)
       dri2_dpy->screen = DefaultScreen(dpy);
    }
 
-   if (!dri2_dpy->conn || xcb_connection_has_error(dri2_dpy->conn)) {
+   if (xcb_connection_has_error(dri2_dpy->conn)) {
       _eglLog(_EGL_WARNING, "DRI3: xcb_connect failed");
       goto cleanup_dpy;
    }
 
-   if (!dri3_x11_connect(dri2_dpy))
-      goto cleanup_conn;
+   if (dri2_dpy->conn) {
+      if (!dri3_x11_connect(dri2_dpy))
+         goto cleanup_conn;
+   }
 
    if (!dri2_load_driver_dri3(disp))
       goto cleanup_conn;
@@ -1337,8 +1338,10 @@ dri2_initialize_x11_dri3(_EGLDriver *drv, _EGLDisplay *disp)
    disp->Extensions.WL_bind_wayland_display = EGL_TRUE;
 #endif
 
-   if (!dri2_x11_add_configs_for_visuals(dri2_dpy, disp, false))
-      goto cleanup_configs;
+   if (dri2_dpy->conn) {
+      if (!dri2_x11_add_configs_for_visuals(dri2_dpy, disp, false))
+         goto cleanup_configs;
+   }
 
    dri2_dpy->loader_dri3_ext.core = dri2_dpy->core;
    dri2_dpy->loader_dri3_ext.image_driver = dri2_dpy->image_driver;
@@ -1367,7 +1370,6 @@ dri2_initialize_x11_dri3(_EGLDriver *drv, _EGLDisplay *disp)
       xcb_disconnect(dri2_dpy->conn);
  cleanup_dpy:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return EGL_FALSE;
 }
@@ -1465,7 +1467,6 @@ dri2_initialize_x11_dri2(_EGLDriver *drv, _EGLDisplay *disp)
       xcb_disconnect(dri2_dpy->conn);
  cleanup_dpy:
    free(dri2_dpy);
-   disp->DriverData = NULL;
 
    return EGL_FALSE;
 }

src/egl/drivers/dri2/platform_x11_dri3.c

@@ -103,17 +103,6 @@ egl_dri3_get_dri_context(struct loader_dri3_drawable *draw)
    return dri2_ctx->dri_context;
 }
 
-static __DRIscreen *
-egl_dri3_get_dri_screen(struct loader_dri3_drawable *draw)
-{
-   _EGLContext *ctx = _eglGetCurrentContext();
-   struct dri2_egl_context *dri2_ctx;
-   if (!ctx)
-      return NULL;
-   dri2_ctx = dri2_egl_context(ctx);
-   return dri2_egl_display(dri2_ctx->base.Resource.Display)->dri_screen;
-}
-
 static void
 egl_dri3_flush_drawable(struct loader_dri3_drawable *draw, unsigned flags)
 {
@@ -130,7 +119,6 @@ static struct loader_dri3_vtable egl_dri3_vtable = {
    .set_drawable_size = egl_dri3_set_drawable_size,
    .in_current_context = egl_dri3_in_current_context,
    .get_dri_context = egl_dri3_get_dri_context,
-   .get_dri_screen = egl_dri3_get_dri_screen,
    .flush_drawable = egl_dri3_flush_drawable,
    .show_fps = NULL,
 };

src/egl/main/eglapi.c

@@ -627,9 +627,7 @@ eglCreateContext(EGLDisplay dpy, EGLConfig config, EGLContext share_list,
 
    _EGL_CHECK_DISPLAY(disp, EGL_NO_CONTEXT, drv);
 
-   if (config)
-      _EGL_CHECK_CONFIG(disp, conf, EGL_NO_CONTEXT, drv);
-   else if (!disp->Extensions.MESA_configless_context)
+   if (!config && !disp->Extensions.MESA_configless_context)
       RETURN_EGL_ERROR(disp, EGL_BAD_CONFIG, EGL_NO_CONTEXT);
 
    if (!share && share_list != EGL_NO_CONTEXT)
@@ -1939,7 +1937,7 @@ _eglLockDisplayInterop(EGLDisplay dpy, EGLContext context,
    return MESA_GLINTEROP_SUCCESS;
 }
 
-PUBLIC int
+int
 MesaGLInteropEGLQueryDeviceInfo(EGLDisplay dpy, EGLContext context,
                                 struct mesa_glinterop_device_info *out)
 {
@@ -1961,7 +1959,7 @@ MesaGLInteropEGLQueryDeviceInfo(EGLDisplay dpy, EGLContext context,
    return ret;
 }
 
-PUBLIC int
+int
 MesaGLInteropEGLExportObject(EGLDisplay dpy, EGLContext context,
                              struct mesa_glinterop_export_in *in,
                              struct mesa_glinterop_export_out *out)

src/egl/main/egldefines.h

@@ -34,8 +34,6 @@
 #ifndef EGLDEFINES_INCLUDED
 #define EGLDEFINES_INCLUDED
 
-#include "util/macros.h"
-
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -50,6 +48,7 @@ extern "C" {
 
 #define _EGL_VENDOR_STRING "Mesa Project"
 
+#define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0]))
 #define MIN2(A, B)  (((A) < (B)) ? (A) : (B))
 
 #ifdef __cplusplus

src/egl/main/eglglobals.c

@@ -53,16 +53,10 @@ struct _egl_global _eglGlobal =
    /* ClientExtensionsString */
    "EGL_EXT_client_extensions"
    " EGL_EXT_platform_base"
-#ifdef HAVE_WAYLAND_PLATFORM
    " EGL_EXT_platform_wayland"
-#endif
-#ifdef HAVE_X11_PLATFORM
    " EGL_EXT_platform_x11"
-#endif
-#ifdef HAVE_DRM_PLATFORM
-   " EGL_MESA_platform_gbm"
-#endif
    " EGL_KHR_client_get_all_proc_addresses"
+   " EGL_MESA_platform_gbm"
 };