From a71d524ecad48837e0124a03124bc05f59a48be7 Mon Sep 17 00:00:00 2001 From: marha Date: Thu, 30 Apr 2015 23:24:30 +0200 Subject: fontconfig pixman libX11 mesa git update 30 Apr 2015 libX11 commit d3415d1f052530760b4617db45affcb984cfe35c pixman commit e0c0153d8e5d42c08c2b9bd2cf2123bff2c48d75 fontconfig commit 4a6f5efd5f6a468e1872d58e589bcf30ba88e2fd mesa commit 1ac7db07b363207e8ded9259f84bbcaa084b8667 --- fontconfig/conf.d/10-scale-bitmap-fonts.conf | 32 +- fontconfig/doc/fontconfig-user.sgml | 2 +- fontconfig/src/fcfreetype.c | 84 +---- libX11/nls/locale.alias.pre | 6 +- mesalib/Android.common.mk | 18 +- mesalib/Android.mk | 10 +- mesalib/CleanSpec.mk | 8 + mesalib/docs/GL3.txt | 6 +- mesalib/docs/VERSIONS | 4 +- mesalib/docs/index.html | 8 +- mesalib/docs/libGL.txt | 2 +- mesalib/docs/relnotes.html | 1 + mesalib/docs/relnotes/10.0.3.html | 2 +- mesalib/docs/relnotes/10.5.4.html | 125 +++++++ mesalib/docs/relnotes/10.6.0.html | 12 +- mesalib/docs/relnotes/7.10.html | 2 +- mesalib/docs/versions.html | 4 +- mesalib/docs/viewperf.html | 7 + mesalib/include/GL/internal/dri_interface.h | 60 ++++ mesalib/scons/llvm.py | 33 +- mesalib/src/gallium/SConscript | 5 + mesalib/src/gallium/auxiliary/Android.mk | 5 +- mesalib/src/gallium/auxiliary/hud/hud_context.c | 163 ++++++++- mesalib/src/gallium/auxiliary/hud/hud_private.h | 4 + mesalib/src/gallium/auxiliary/util/u_tile.c | 4 +- mesalib/src/gallium/auxiliary/util/u_video.h | 36 ++ mesalib/src/glsl/Android.gen.mk | 66 +++- mesalib/src/glsl/Android.mk | 13 +- mesalib/src/glsl/ast_function.cpp | 2 +- mesalib/src/glsl/ast_to_hir.cpp | 4 +- mesalib/src/glsl/glcpp/glcpp-parse.y | 2 +- mesalib/src/glsl/glsl_types.cpp | 23 +- mesalib/src/glsl/glsl_types.h | 7 +- mesalib/src/glsl/link_uniforms.cpp | 2 +- mesalib/src/glsl/linker.cpp | 3 +- mesalib/src/glsl/lower_packed_varyings.cpp | 8 +- mesalib/src/glsl/lower_packing_builtins.cpp | 8 +- mesalib/src/glsl/nir/nir.c | 112 ++++-- mesalib/src/glsl/nir/nir.h | 17 +- mesalib/src/glsl/nir/nir_array.h | 96 +++++ mesalib/src/glsl/nir/nir_from_ssa.c | 2 +- mesalib/src/glsl/nir/nir_lower_locals_to_regs.c | 148 ++++++-- mesalib/src/glsl/nir/nir_lower_to_source_mods.c | 7 + mesalib/src/glsl/nir/nir_lower_var_copies.c | 24 +- mesalib/src/glsl/nir/nir_lower_vars_to_ssa.c | 120 ++----- mesalib/src/glsl/nir/nir_opt_algebraic.py | 1 + mesalib/src/glsl/nir/nir_print.c | 2 + mesalib/src/glsl/nir/nir_types.cpp | 2 +- mesalib/src/glsl/opt_algebraic.cpp | 20 ++ mesalib/src/glsl/opt_dead_builtin_varyings.cpp | 8 +- mesalib/src/glsl/opt_function_inlining.cpp | 2 +- mesalib/src/glsl/standalone_scaffolding.cpp | 3 - mesalib/src/loader/Android.mk | 2 +- mesalib/src/loader/loader.h | 2 +- mesalib/src/mapi/Android.mk | 2 +- mesalib/src/mapi/glapi/gen/glX_proto_send.py | 2 +- mesalib/src/mapi/glapi/gen/glX_proto_size.py | 2 +- mesalib/src/mapi/glapi/gen/gl_x86-64_asm.py | 2 +- mesalib/src/mapi/glapi/glapi_entrypoint.c | 2 + mesalib/src/mesa/Android.gen.mk | 23 +- mesalib/src/mesa/Android.libmesa_dricore.mk | 15 +- mesalib/src/mesa/Android.libmesa_glsl_utils.mk | 10 +- mesalib/src/mesa/Android.libmesa_st_mesa.mk | 9 +- mesalib/src/mesa/Android.mesa_gen_matypes.mk | 1 - mesalib/src/mesa/drivers/common/meta.c | 1 - mesalib/src/mesa/drivers/dri/Android.mk | 11 +- mesalib/src/mesa/drivers/dri/common/Android.mk | 38 +- mesalib/src/mesa/drivers/dri/common/drirc | 4 + mesalib/src/mesa/main/compiler.h | 5 - mesalib/src/mesa/main/context.c | 6 +- mesalib/src/mesa/main/extensions.c | 2 + mesalib/src/mesa/main/fbobject.c | 9 +- mesalib/src/mesa/main/ffvertex_prog.c | 2 +- mesalib/src/mesa/main/formats.c | 2 +- mesalib/src/mesa/main/macros.h | 5 +- mesalib/src/mesa/main/mtypes.h | 13 - mesalib/src/mesa/main/shader_query.cpp | 56 +-- mesalib/src/mesa/main/shaderapi.c | 9 - mesalib/src/mesa/main/texgetimage.c | 54 ++- mesalib/src/mesa/main/teximage.c | 7 +- mesalib/src/mesa/main/texobj.c | 1 - mesalib/src/mesa/main/texturebarrier.c | 6 + mesalib/src/mesa/main/uniform_query.cpp | 20 +- mesalib/src/mesa/main/version.c | 63 +++- mesalib/src/mesa/math/m_matrix.c | 2 +- mesalib/src/mesa/program/Android.mk | 14 +- mesalib/src/mesa/program/hash_table.h | 4 +- mesalib/src/mesa/program/prog_cache.c | 2 +- mesalib/src/mesa/program/prog_to_nir.c | 25 +- mesalib/src/mesa/state_tracker/st_atom_texture.c | 14 +- mesalib/src/mesa/state_tracker/st_cb_flush.c | 2 +- mesalib/src/mesa/state_tracker/st_cb_texture.c | 2 +- mesalib/src/mesa/state_tracker/st_extensions.c | 9 +- mesalib/src/mesa/state_tracker/st_format.c | 19 +- mesalib/src/mesa/swrast/s_texcombine.c | 2 +- mesalib/src/mesa/tnl/t_vertex.h | 2 +- mesalib/src/mesa/vbo/vbo_exec_api.c | 2 +- mesalib/src/mesa/vbo/vbo_save_draw.c | 2 +- mesalib/src/util/Android.mk | 42 +-- mesalib/src/util/macros.h | 2 + pixman/.gitignore | 1 + pixman/test/Makefile.sources | 1 + pixman/test/affine-bench.c | 436 +++++++++++++++++++++++ 103 files changed, 1783 insertions(+), 531 deletions(-) create mode 100644 mesalib/docs/relnotes/10.5.4.html create mode 100644 mesalib/src/glsl/nir/nir_array.h create mode 100644 pixman/test/affine-bench.c diff --git a/fontconfig/conf.d/10-scale-bitmap-fonts.conf b/fontconfig/conf.d/10-scale-bitmap-fonts.conf index e3bcd46cd..b4e9cb4f2 100644 --- a/fontconfig/conf.d/10-scale-bitmap-fonts.conf +++ b/fontconfig/conf.d/10-scale-bitmap-fonts.conf @@ -2,8 +2,13 @@ + - + false @@ -13,10 +18,20 @@ + + + false + false + + true + @@ -30,20 +45,7 @@ - - - false - - - 64 - - - true - true @@ -55,7 +57,7 @@ If we *are* going to scale, go ahead and do it. --> - + false diff --git a/fontconfig/doc/fontconfig-user.sgml b/fontconfig/doc/fontconfig-user.sgml index 7b4ad7408..5803929f7 100644 --- a/fontconfig/doc/fontconfig-user.sgml +++ b/fontconfig/doc/fontconfig-user.sgml @@ -322,7 +322,7 @@ will be used to store the cache files. If it starts with '~', it refers to a directory in the users home directory. If 'prefix' is set to "xdg", the value in the XDG_CACHE_HOME environment variable will be added as the path prefix. please see XDG Base Directory Specification for more details. The default directory is ``$XDG_CACHE_HOME/fontconfig'' and it contains the cache files named ``<hash value>-<architecture>.cache-<version>'', -where <version> is the font configureation file +where <version> is the font configuration file version number (currently 5). <literal><include ignore_missing="no" prefix="default"></literal> diff --git a/fontconfig/src/fcfreetype.c b/fontconfig/src/fcfreetype.c index 50ff3489c..afbd9ac1e 100644 --- a/fontconfig/src/fcfreetype.c +++ b/fontconfig/src/fcfreetype.c @@ -978,76 +978,6 @@ FcNoticeFoundry(const FT_String *notice) return 0; } -static FcBool -FcVendorMatch(const FT_Char vendor[4], const FT_Char *vendor_string) -{ - /* vendor is not necessarily NUL-terminated. */ - int i, len; - - len = strlen((char *) vendor_string); - if (memcmp(vendor, vendor_string, len) != 0) - return FcFalse; - for (i = len; i < 4; i++) - if (vendor[i] != ' ' && vendor[i] != '\0') - return FcFalse; - return FcTrue; -} - -/* This table is partly taken from ttmkfdir by Joerg Pommnitz. */ - -/* It should not contain useless entries (such as UNKN) nor duplicate - entries for padding both with spaces and NULs. */ - -static const struct { - const FT_Char vendor[5]; - const FcChar8 foundry[13]; -} FcVendorFoundries[] = { - { "ADBE", "adobe"}, - { "AGFA", "agfa"}, - { "ALTS", "altsys"}, - { "APPL", "apple"}, - { "ARPH", "arphic"}, - { "ATEC", "alltype"}, - { "B&H", "b&h"}, - { "BITS", "bitstream"}, - { "CANO", "cannon"}, - { "CLM", "culmus"}, - { "DYNA", "dynalab"}, - { "EPSN", "epson"}, - { "FJ", "fujitsu"}, - { "IBM", "ibm"}, - { "ITC", "itc"}, - { "IMPR", "impress"}, - { "LARA", "larabiefonts"}, - { "LEAF", "interleaf"}, - { "LETR", "letraset"}, - { "LINO", "linotype"}, - { "MACR", "macromedia"}, - { "MONO", "monotype"}, - { "MS", "microsoft"}, - { "MT", "monotype"}, - { "NEC", "nec"}, - { "PARA", "paratype"}, - { "QMSI", "qms"}, - { "RICO", "ricoh"}, - { "URW", "urw"}, - { "Y&Y", "y&y"} -}; - -#define NUM_VENDOR_FOUNDRIES (int) (sizeof (FcVendorFoundries) / sizeof (FcVendorFoundries[0])) - -static const FcChar8 * -FcVendorFoundry(const FT_Char vendor[4]) -{ - int i; - - if (vendor) - for(i = 0; i < NUM_VENDOR_FOUNDRIES; i++) - if (FcVendorMatch (vendor, FcVendorFoundries[i].vendor)) - return FcVendorFoundries[i].foundry; - return 0; -} - typedef struct _FcStringConst { const FcChar8 *name; int value; @@ -1239,7 +1169,7 @@ FcFreeTypeQueryFace (const FT_Face face, #if 0 FcChar8 *family = 0; #endif - FcChar8 *complex_; + FcChar8 *complex_, *foundry_ = NULL; const FcChar8 *foundry = 0; int spacing; TT_OS2 *os2; @@ -1315,7 +1245,15 @@ FcFreeTypeQueryFace (const FT_Face face, */ if (os2 && os2->version >= 0x0001 && os2->version != 0xffff) - foundry = FcVendorFoundry(os2->achVendID); + { + if (os2->achVendID && os2->achVendID[0] != 0) + { + foundry_ = (FcChar8 *) malloc (sizeof (os2->achVendID) + 1); + memcpy ((void *)foundry_, os2->achVendID, sizeof (os2->achVendID)); + foundry_[sizeof (os2->achVendID)] = 0; + foundry = foundry_; + } + } if (FcDebug () & FC_DBG_SCANV) printf ("\n"); @@ -1946,6 +1884,8 @@ bail2: FcCharSetDestroy (cs); bail1: FcPatternDestroy (pat); + if (foundry_) + free (foundry_); bail0: return NULL; } diff --git a/libX11/nls/locale.alias.pre b/libX11/nls/locale.alias.pre index 63d224612..d9f078551 100644 --- a/libX11/nls/locale.alias.pre +++ b/libX11/nls/locale.alias.pre @@ -982,9 +982,9 @@ rw_RW.ISO-8859-1: rw_RW.ISO8859-1 rw_RW.utf8: rw_RW.UTF-8 sd: sd_IN.UTF-8 sd_IN.utf8: sd_IN.UTF-8 -sd@devanagari: sd_INUTF-8@devanagari -sd_IN@devanagari: sd_INUTF-8@devanagari -sd_IN.utf8@devanagari: sd_INUTF-8@devanagari +sd@devanagari: sd_IN.UTF-8@devanagari +sd_IN@devanagari: sd_IN.UTF-8@devanagari +sd_IN.utf8@devanagari: sd_IN.UTF-8@devanagari se_NO: se_NO.UTF-8 se_NO.utf8: se_NO.UTF-8 XCOMM sh was the old ISO code for Serbo-Croatian (now individual sr and hr). diff --git a/mesalib/Android.common.mk b/mesalib/Android.common.mk index 3e6d4c398..edf52d6fa 100644 --- a/mesalib/Android.common.mk +++ b/mesalib/Android.common.mk @@ -24,16 +24,17 @@ # use c99 compiler by default ifeq ($(LOCAL_CC),) ifeq ($(LOCAL_IS_HOST_MODULE),true) -LOCAL_CC := $(HOST_CC) -std=c99 +LOCAL_CC := $(HOST_CC) -std=c99 -D_GNU_SOURCE else LOCAL_CC := $(TARGET_CC) -std=c99 endif endif LOCAL_C_INCLUDES += \ + $(MESA_TOP)/src \ $(MESA_TOP)/include -MESA_VERSION=$(shell cat $(MESA_TOP)/VERSION) +MESA_VERSION := $(shell cat $(MESA_TOP)/VERSION) # define ANDROID_VERSION (e.g., 4.0.x => 0x0400) LOCAL_CFLAGS += \ -DPACKAGE_VERSION=\"$(MESA_VERSION)\" \ @@ -41,6 +42,19 @@ LOCAL_CFLAGS += \ -DANDROID_VERSION=0x0$(MESA_ANDROID_MAJOR_VERSION)0$(MESA_ANDROID_MINOR_VERSION) LOCAL_CFLAGS += \ + -DHAVE___BUILTIN_EXPECT \ + -DHAVE___BUILTIN_FFS \ + -DHAVE___BUILTIN_FFSLL \ + -DHAVE_FUNC_ATTRIBUTE_FLATTEN \ + -DHAVE_FUNC_ATTRIBUTE_UNUSED \ + -DHAVE_FUNC_ATTRIBUTE_FORMAT \ + -DHAVE_FUNC_ATTRIBUTE_PACKED \ + -DHAVE___BUILTIN_CTZ \ + -DHAVE___BUILTIN_POPCOUNT \ + -DHAVE___BUILTIN_POPCOUNTLL \ + -DHAVE___BUILTIN_CLZ \ + -DHAVE___BUILTIN_CLZLL \ + -DHAVE___BUILTIN_UNREACHABLE \ -DHAVE_PTHREAD=1 \ -fvisibility=hidden \ -Wno-sign-compare diff --git a/mesalib/Android.mk b/mesalib/Android.mk index e190f749b..b19419ba7 100644 --- a/mesalib/Android.mk +++ b/mesalib/Android.mk @@ -34,6 +34,13 @@ MESA_TOP := $(call my-dir) MESA_ANDROID_MAJOR_VERSION := $(word 1, $(subst ., , $(PLATFORM_VERSION))) MESA_ANDROID_MINOR_VERSION := $(word 2, $(subst ., , $(PLATFORM_VERSION))) MESA_ANDROID_VERSION := $(MESA_ANDROID_MAJOR_VERSION).$(MESA_ANDROID_MINOR_VERSION) +ifeq ($(filter 1 2 3 4,$(MESA_ANDROID_MAJOR_VERSION)),) +MESA_LOLLIPOP_BUILD := true +else +define local-generated-sources-dir +$(call local-intermediates-dir) +endef +endif MESA_COMMON_MK := $(MESA_TOP)/Android.common.mk MESA_PYTHON2 := python @@ -94,7 +101,6 @@ ifeq ($(strip $(MESA_BUILD_GALLIUM)),true) SUBDIRS += src/gallium endif -mkfiles := $(patsubst %,$(MESA_TOP)/%/Android.mk,$(SUBDIRS)) -include $(mkfiles) +include $(call all-named-subdir-makefiles,$(SUBDIRS)) endif diff --git a/mesalib/CleanSpec.mk b/mesalib/CleanSpec.mk index 820a1c7f0..20681638e 100644 --- a/mesalib/CleanSpec.mk +++ b/mesalib/CleanSpec.mk @@ -5,3 +5,11 @@ $(call add-clean-step, rm -rf $(PRODUCT_OUT)/obj/SHARED_LIBRARIES/libGLES_mesa_i $(call add-clean-step, rm -rf $(OUT_DIR)/host/$(HOST_OS)-$(HOST_ARCH)/obj/EXECUTABLES/mesa_*_intermediates) $(call add-clean-step, rm -rf $(OUT_DIR)/host/$(HOST_OS)-$(HOST_ARCH)/obj/EXECUTABLES/glsl_compiler_intermediates) $(call add-clean-step, rm -rf $(OUT_DIR)/host/$(HOST_OS)-$(HOST_ARCH)/obj/STATIC_LIBRARIES/libmesa_glsl_utils_intermediates) + +$(call add-clean-step, rm -rf $(PRODUCT_OUT)/*/STATIC_LIBRARIES/libmesa_*_intermediates) +$(call add-clean-step, rm -rf $(PRODUCT_OUT)/*/SHARED_LIBRARIES/i9?5_dri_intermediates) +$(call add-clean-step, rm -rf $(PRODUCT_OUT)/*/SHARED_LIBRARIES/libglapi_intermediates) +$(call add-clean-step, rm -rf $(PRODUCT_OUT)/*/SHARED_LIBRARIES/libGLES_mesa_intermediates) +$(call add-clean-step, rm -rf $(HOST_OUT_release)/*/EXECUTABLES/mesa_*_intermediates) +$(call add-clean-step, rm -rf $(HOST_OUT_release)/*/EXECUTABLES/glsl_compiler_intermediates) +$(call add-clean-step, rm -rf $(HOST_OUT_release)/*/STATIC_LIBRARIES/libmesa_*_intermediates) diff --git a/mesalib/docs/GL3.txt b/mesalib/docs/GL3.txt index 2dbd987de..617ff3447 100644 --- a/mesalib/docs/GL3.txt +++ b/mesalib/docs/GL3.txt @@ -163,7 +163,7 @@ GL 4.3, GLSL 4.30: GL_ARB_multi_draw_indirect DONE (i965, nvc0, r600, radeonsi, llvmpipe, softpipe) GL_ARB_program_interface_query DONE (all drivers) GL_ARB_robust_buffer_access_behavior not started - GL_ARB_shader_image_size not started + GL_ARB_shader_image_size in progress (Martin Peres) GL_ARB_shader_storage_buffer_object not started GL_ARB_stencil_texturing DONE (i965/gen8+, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe) GL_ARB_texture_buffer_range DONE (nv50, nvc0, i965, r600, radeonsi, llvmpipe) @@ -182,7 +182,7 @@ GL 4.4, GLSL 4.40: GL_ARB_multi_bind DONE (all drivers) GL_ARB_query_buffer_object not started GL_ARB_texture_mirror_clamp_to_edge DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe) - GL_ARB_texture_stencil8 not started + GL_ARB_texture_stencil8 DONE (nv50, nvc0, r600, radeonsi, llvmpipe, softpipe) GL_ARB_vertex_type_10f_11f_11f_rev DONE (i965, nv50, nvc0, r600, radeonsi, llvmpipe, softpipe) GL 4.5, GLSL 4.50: @@ -220,7 +220,7 @@ GLES3.1, GLSL ES 3.1 GL_ARB_program_interface_query DONE (all drivers) GL_ARB_shader_atomic_counters DONE (i965) GL_ARB_shader_image_load_store in progress (curro) - GL_ARB_shader_image_size not started + GL_ARB_shader_image_size in progress (Martin Peres) GL_ARB_shader_storage_buffer_object not started GL_ARB_shading_language_packing DONE (all drivers) GL_ARB_separate_shader_objects DONE (all drivers) diff --git a/mesalib/docs/VERSIONS b/mesalib/docs/VERSIONS index 0ebcd1469..9af1546a8 100644 --- a/mesalib/docs/VERSIONS +++ b/mesalib/docs/VERSIONS @@ -103,7 +103,7 @@ Mesa Version History - Stencil-related functions now work in display lists Changes: - renamed aux.h as glaux.h (MS-DOS names can't start with aux) - - most filenames are in 8.3 format to accomodate MS-DOS + - most filenames are in 8.3 format to accommodate MS-DOS - use GLubytes to store arrays of colors instead of GLints 1.2.2 August 2, 1995 @@ -1007,7 +1007,7 @@ Mesa Version History - glGetTexImage was using pixel unpacking instead of packing params - auto-mipmap generation for cube maps was incorrect Changes: - - max texture units reduced to six to accomodate texture rectangles + - max texture units reduced to six to accommodate texture rectangles - removed unfinished GL_MESA_sprite_point extension code diff --git a/mesalib/docs/index.html b/mesalib/docs/index.html index 9e4644caa..e01790cd7 100644 --- a/mesalib/docs/index.html +++ b/mesalib/docs/index.html @@ -16,6 +16,12 @@

News

+

April 24, 2015

+

+Mesa 10.5.4 is released. +This is a bug-fix release. +

+

April 12, 2015

Mesa 10.5.3 is released. @@ -1360,7 +1366,7 @@ The Mesa FAQ has been rewritten. - glGetTexImage was using pixel unpacking instead of packing params - auto-mipmap generation for cube maps was incorrect Changes: - - max texture units reduced to six to accomodate texture rectangles + - max texture units reduced to six to accommodate texture rectangles - removed unfinished GL_MESA_sprite_point extension code diff --git a/mesalib/docs/libGL.txt b/mesalib/docs/libGL.txt index d06b4e62a..839e330cf 100644 --- a/mesalib/docs/libGL.txt +++ b/mesalib/docs/libGL.txt @@ -49,7 +49,7 @@ stderr if the LIBGL_DEBUG environment variable is defined. libGL.so is thread safe. The overhead of thread safety for common, single-thread clients is negligible. However, the overhead of thread safety for multi-threaded clients is significant. Each GL API call -requires two calls to pthread_get_specific() which can noticably +requires two calls to pthread_get_specific() which can noticeably impact performance. Warning: libGL.so is thread safe but individual DRI drivers may not be. Please consult the documentation for a driver to learn if it is thread safe. diff --git a/mesalib/docs/relnotes.html b/mesalib/docs/relnotes.html index 6ec35d11f..7f2e1d851 100644 --- a/mesalib/docs/relnotes.html +++ b/mesalib/docs/relnotes.html @@ -21,6 +21,7 @@ The release notes summarize what's new or changed in each Mesa release.

    +
  • 10.5.4 release notes
  • 10.5.3 release notes
  • 10.5.2 release notes
  • 10.4.7 release notes diff --git a/mesalib/docs/relnotes/10.0.3.html b/mesalib/docs/relnotes/10.0.3.html index ad7b50864..ce0dcf033 100644 --- a/mesalib/docs/relnotes/10.0.3.html +++ b/mesalib/docs/relnotes/10.0.3.html @@ -104,7 +104,7 @@ a07b4b6b9eb449b88a6cb5061e51c331 MesaLib-10.0.3.zip
  • Add md5sums for 10.0.2. release.
  • cherry-ignore: Ignore several patches not yet ready for the stable branch
  • Drop another couple of patches.
  • -
  • cherry-ignore: Ignore 4 patches at teh request of the author, (Anuj).
  • +
  • cherry-ignore: Ignore 4 patches at the request of the author, (Anuj).
  • Update version to 10.0.3
diff --git a/mesalib/docs/relnotes/10.5.4.html b/mesalib/docs/relnotes/10.5.4.html new file mode 100644 index 000000000..4c466d0b3 --- /dev/null +++ b/mesalib/docs/relnotes/10.5.4.html @@ -0,0 +1,125 @@ + + + + + Mesa Release Notes + + + + +
+

The Mesa 3D Graphics Library

+
+ + +
+ +

Mesa 10.5.4 Release Notes / April 24, 2015

+ +

+Mesa 10.5.4 is a bug fix release which fixes bugs found since the 10.5.3 release. +

+

+Mesa 10.5.4 implements the OpenGL 3.3 API, but the version reported by +glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) / +glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used. +Some drivers don't support all the features required in OpenGL 3.3. OpenGL +3.3 is only available if requested at context creation +because compatibility contexts are not supported. +

+ + +

SHA256 checksums

+
+e1089567fc7bf8d9b2d8badcc9f2fc3b758701c8c0ccfe7af1805549fea53f11  mesa-10.5.4.tar.gz
+b51e723f3a20d842c88a92d809435b229fc4744ca0dbec0317d9d4a3ac4c6803  mesa-10.5.4.tar.xz
+
+ + +

New features

+

None

+ +

Bug fixes

+ +

This list is likely incomplete.

+ +
    + +
  • Bug 69226 - Cannot enable basic shaders with Second Life aborts attempt
  • + +
  • Bug 71591 - Second Life shaders fail to compile (extension declared in middle of shader)
  • + +
  • Bug 81025 - [IVB/BYT Bisected]Piglit spec_ARB_draw_indirect_arb_draw_indirect-draw-elements-prim-restart-ugly fails
  • + +
  • Bug 89457 - [BSW Bisected]ogles3conform ES3-CTS.gtf.GL3Tests.shadow.shadow_execution_vert fails
  • + +
  • Bug 89957 - vm protection faults in piglit lest: texsubimage cube_map_array pbo
  • + +
+ + +

Changes

+ +

Brian Paul (1):

+
    +
  • glsl: rewrite glsl_type::record_key_hash() to avoid buffer overflow
  • +
+ +

Dave Airlie (2):

+
    +
  • st/mesa: convert sub image for cube map arrays to 2d arrays for upload
  • +
  • st/mesa: align cube map arrays layers
  • +
+ +

Emil Velikov (11):

+
    +
  • docs: Add 256 sums for the 10.5.3 release
  • +
  • radeonsi: remove unused si_dump_key()
  • +
  • android: use LOCAL_SHARED_LIBRARIES over TARGET_OUT_HEADERS
  • +
  • android: add $(mesa_top)/src include to the whole of mesa
  • +
  • android: egl: add libsync_cflags to the build
  • +
  • android: dri/common: conditionally include drm_cflags/set __NOT_HAVE_DRM_H
  • +
  • android: add HAVE__BUILTIN_* and HAVE_FUNC_ATTRIBUTE_* defines
  • +
  • android: add $(mesa_top)/src/mesa/main to the includes list
  • +
  • android: dri: link against libmesa_util
  • +
  • android: mesa: fix the path of the SSE4_1 optimisations
  • +
  • Update version to 10.5.4
  • +
+ +

Ian Romanick (1):

+
    +
  • nir: Fix typo in "ushr by 0" algebraic replacement
  • +
+ +

Kenneth Graunke (2):

+
    +
  • i965: Fix software primitive restart with indirect draws.
  • +
  • drirc: Add "Second Life" quirk (allow_glsl_extension_directive_midshader).
  • +
+ +

Kristian Høgsberg (1):

+
    +
  • i965: Rewrite ir_tex to ir_txl with lod 0 for vertex shaders
  • +
+ +

Marek Olšák (2):

+
    +
  • glsl_to_tgsi: fix out-of-bounds constant access and crash for uniforms
  • +
  • glsl_to_tgsi: don't use a potentially-undefined immediate for ir_query_levels
  • +
+ +

Mathias Froehlich (1):

+
    +
  • i965: Flush batchbuffer containing the query on glQueryCounter.
  • +
+ +

Mauro Rossi (2):

+
    +
  • android: mesa: generate the format_{un,}pack.[ch] sources
  • +
  • android: add inital NIR build
  • +
+ + +
+ + diff --git a/mesalib/docs/relnotes/10.6.0.html b/mesalib/docs/relnotes/10.6.0.html index 82aea5cf9..fbc4d5be9 100644 --- a/mesalib/docs/relnotes/10.6.0.html +++ b/mesalib/docs/relnotes/10.6.0.html @@ -45,15 +45,23 @@ Note: some of the new features are only available with certain drivers.
  • GL_AMD_pinned_memory on r600, radeonsi
  • +
  • GL_ARB_clip_control on i965
  • +
  • GL_ARB_depth_buffer_float on freedreno
  • +
  • GL_ARB_depth_clamp on freedreno
  • GL_ARB_draw_indirect, GL_ARB_multi_draw_indirect on r600
  • GL_ARB_draw_instanced on freedreno
  • GL_ARB_gpu_shader_fp64 on nvc0, softpipe
  • +
  • GL_ARB_gpu_shader5 on i965/gen8+
  • GL_ARB_instanced_arrays on freedreno
  • GL_ARB_pipeline_statistics_query on i965, nv50, nvc0, r600, radeonsi, softpipe
  • +
  • GL_ARB_program_interface_query (all drivers)
  • +
  • GL_ARB_texture_stencil8 on nv50, nvc0, r600, radeonsi, softpipe
  • GL_ARB_uniform_buffer_object on freedreno
  • GL_EXT_draw_buffers2 on freedreno
  • -
  • GL_ARB_clip_control on i965
  • -
  • GL_ARB_program_interface_query (all drivers)
  • +
  • GL_OES_EGL_sync on all drivers
  • +
  • EGL_KHR_fence_sync on freedreno, nv50, nvc0, r600, radeonsi
  • +
  • EGL_KHR_wait_sync on freedreno, nv50, nvc0, r600, radeonsi
  • +
  • EGL_KHR_cl_event2 on freedreno, nv50, nvc0, r600, radeonsi

Bug fixes

diff --git a/mesalib/docs/relnotes/7.10.html b/mesalib/docs/relnotes/7.10.html index a88811e50..447465e71 100644 --- a/mesalib/docs/relnotes/7.10.html +++ b/mesalib/docs/relnotes/7.10.html @@ -1693,7 +1693,7 @@ bc644be551ed585fc4f66c16b64a91c9 MesaGLUT-7.10.tar.gz
  • llvmpipe: Special case complementary and identify blend factors in SoA.
  • llvmpipe: Make rgb/alpha bland func/factors match, when there is no alpha.
  • draw: Prevent clipped vertices overflow.
  • -
  • draw: Fullfil the new min_lod/max_lod/lod_bias/border_color dynamic state
  • +
  • draw: Fulfil the new min_lod/max_lod/lod_bias/border_color dynamic state
  • gallivm: Fetch the lod from the dynamic state when min_lod == max_lod.
  • gallivm: Remove dead experimental code.
  • llvmpipe: Decouple sampler view and sampler state updates.
  • diff --git a/mesalib/docs/versions.html b/mesalib/docs/versions.html index 30c1817e1..c7f94c513 100644 --- a/mesalib/docs/versions.html +++ b/mesalib/docs/versions.html @@ -150,7 +150,7 @@ New features: Changes:
    • renamed aux.h as glaux.h (MS-DOS names can't start with aux) -
    • most filenames are in 8.3 format to accomodate MS-DOS +
    • most filenames are in 8.3 format to accommodate MS-DOS
    • use GLubytes to store arrays of colors instead of GLints
    @@ -1224,7 +1224,7 @@ Bug fixes: Changes:
      -
    • max texture units reduced to six to accomodate texture rectangles +
    • max texture units reduced to six to accommodate texture rectangles
    • removed unfinished GL_MESA_sprite_point extension code
    diff --git a/mesalib/docs/viewperf.html b/mesalib/docs/viewperf.html index 6b63b94d8..a60a44f36 100644 --- a/mesalib/docs/viewperf.html +++ b/mesalib/docs/viewperf.html @@ -329,6 +329,13 @@ array initializer statement, but it neglects to specify #version 120 at the top of the shader code. So, the shader does not compile and all that's rendered is plain white polygons.

    +

    +Also, the test tries to create a very large 3D texture that may exceed +the device driver's limit. +When this happens, the glTexImage3D call fails and all that's rendered is +a white box. +

    +

    showcase-01

    diff --git a/mesalib/include/GL/internal/dri_interface.h b/mesalib/include/GL/internal/dri_interface.h index eb7da23f7..c827bb640 100644 --- a/mesalib/include/GL/internal/dri_interface.h +++ b/mesalib/include/GL/internal/dri_interface.h @@ -85,6 +85,7 @@ typedef struct __DRIdri2ExtensionRec __DRIdri2Extension; typedef struct __DRIdri2LoaderExtensionRec __DRIdri2LoaderExtension; typedef struct __DRI2flushExtensionRec __DRI2flushExtension; typedef struct __DRI2throttleExtensionRec __DRI2throttleExtension; +typedef struct __DRI2fenceExtensionRec __DRI2fenceExtension; typedef struct __DRIimageLoaderExtensionRec __DRIimageLoaderExtension; @@ -339,6 +340,65 @@ struct __DRI2throttleExtensionRec { enum __DRI2throttleReason reason); }; + +/** + * Extension for fences / synchronization objects. + */ + +#define __DRI2_FENCE "DRI2_Fence" +#define __DRI2_FENCE_VERSION 1 + +#define __DRI2_FENCE_TIMEOUT_INFINITE 0xffffffffffffffffllu + +#define __DRI2_FENCE_FLAG_FLUSH_COMMANDS (1 << 0) + +struct __DRI2fenceExtensionRec { + __DRIextension base; + + /** + * Create and insert a fence into the command stream of the context. + */ + void *(*create_fence)(__DRIcontext *ctx); + + /** + * Get a fence associated with the OpenCL event object. + * This can be NULL, meaning that OpenCL interoperability is not supported. + */ + void *(*get_fence_from_cl_event)(__DRIscreen *screen, intptr_t cl_event); + + /** + * Destroy a fence. + */ + void (*destroy_fence)(__DRIscreen *screen, void *fence); + + /** + * This function waits and doesn't return until the fence is signalled + * or the timeout expires. It returns true if the fence has been signaled. + * + * \param ctx the context where commands are flushed + * \param fence the fence + * \param flags a combination of __DRI2_FENCE_FLAG_xxx flags + * \param timeout the timeout in ns or __DRI2_FENCE_TIMEOUT_INFINITE + */ + GLboolean (*client_wait_sync)(__DRIcontext *ctx, void *fence, + unsigned flags, uint64_t timeout); + + /** + * This function enqueues a wait command into the command stream of + * the context and then returns. When the execution reaches the wait + * command, no further execution will be done in the context until + * the fence is signaled. This is a no-op if the device doesn't support + * parallel execution of contexts. + * + * \param ctx the context where the waiting is done + * \param fence the fence + * \param flags a combination of __DRI2_FENCE_FLAG_xxx flags that make + * sense with this function (right now there are none) + */ + void (*server_wait_sync)(__DRIcontext *ctx, void *fence, unsigned flags); +}; + + /*@}*/ /** diff --git a/mesalib/scons/llvm.py b/mesalib/scons/llvm.py index be7df9fad..17278df88 100644 --- a/mesalib/scons/llvm.py +++ b/mesalib/scons/llvm.py @@ -72,18 +72,25 @@ def generate(env): return # Try to determine the LLVM version from llvm/Config/config.h - llvm_config = os.path.join(llvm_dir, 'include/llvm/Config/config.h') + llvm_config = os.path.join(llvm_dir, 'include/llvm/Config/llvm-config.h') if not os.path.exists(llvm_config): print 'scons: could not find %s' % llvm_config return - llvm_version_re = re.compile(r'^#define PACKAGE_VERSION "([^"]*)"') + llvm_version_major_re = re.compile(r'^#define LLVM_VERSION_MAJOR ([0-9]+)') + llvm_version_minor_re = re.compile(r'^#define LLVM_VERSION_MINOR ([0-9]+)') llvm_version = None + llvm_version_major = None + llvm_version_minor = None for line in open(llvm_config, 'rt'): - mo = llvm_version_re.match(line) + mo = llvm_version_major_re.match(line) if mo: - llvm_version = mo.group(1) - llvm_version = distutils.version.LooseVersion(llvm_version) - break + llvm_version_major = mo.group(1) + mo = llvm_version_minor_re.match(line) + if mo: + llvm_version_minor = mo.group(1) + if llvm_version_major is not None and llvm_version_minor is not None: + llvm_version = distutils.version.LooseVersion('%s.%s' % (llvm_version_major, llvm_version_minor)) + if llvm_version is None: print 'scons: could not determine the LLVM version from %s' % llvm_config return @@ -99,7 +106,19 @@ def generate(env): ]) env.Prepend(LIBPATH = [os.path.join(llvm_dir, 'lib')]) # LIBS should match the output of `llvm-config --libs engine mcjit bitwriter x86asmprinter` - if llvm_version >= distutils.version.LooseVersion('3.5'): + if llvm_version >= distutils.version.LooseVersion('3.6'): + env.Prepend(LIBS = [ + 'LLVMBitWriter', 'LLVMX86Disassembler', 'LLVMX86AsmParser', + 'LLVMX86CodeGen', 'LLVMSelectionDAG', 'LLVMAsmPrinter', + 'LLVMCodeGen', 'LLVMScalarOpts', 'LLVMProfileData', + 'LLVMInstCombine', 'LLVMTransformUtils', 'LLVMipa', + 'LLVMAnalysis', 'LLVMX86Desc', 'LLVMMCDisassembler', + 'LLVMX86Info', 'LLVMX86AsmPrinter', 'LLVMX86Utils', + 'LLVMMCJIT', 'LLVMTarget', 'LLVMExecutionEngine', + 'LLVMRuntimeDyld', 'LLVMObject', 'LLVMMCParser', + 'LLVMBitReader', 'LLVMMC', 'LLVMCore', 'LLVMSupport' + ]) + elif llvm_version >= distutils.version.LooseVersion('3.5'): env.Prepend(LIBS = [ 'LLVMBitWriter', 'LLVMMCJIT', 'LLVMRuntimeDyld', 'LLVMX86Disassembler', 'LLVMX86AsmParser', 'LLVMX86CodeGen', diff --git a/mesalib/src/gallium/SConscript b/mesalib/src/gallium/SConscript index 680ad925f..eeb1c780f 100644 --- a/mesalib/src/gallium/SConscript +++ b/mesalib/src/gallium/SConscript @@ -60,6 +60,11 @@ SConscript([ ]) if not env['embedded']: + SConscript([ + 'state_trackers/osmesa/SConscript', + 'targets/osmesa/SConscript', + ]) + if env['x11']: SConscript([ 'state_trackers/glx/xlib/SConscript', diff --git a/mesalib/src/gallium/auxiliary/Android.mk b/mesalib/src/gallium/auxiliary/Android.mk index 0bc183170..96a2125de 100644 --- a/mesalib/src/gallium/auxiliary/Android.mk +++ b/mesalib/src/gallium/auxiliary/Android.mk @@ -33,14 +33,13 @@ LOCAL_SRC_FILES := \ $(VL_STUB_SOURCES) LOCAL_C_INCLUDES := \ - $(GALLIUM_TOP)/auxiliary/util \ - $(MESA_TOP)/src + $(GALLIUM_TOP)/auxiliary/util LOCAL_MODULE := libmesa_gallium # generate sources LOCAL_MODULE_CLASS := STATIC_LIBRARIES -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) LOCAL_GENERATED_SOURCES := $(addprefix $(intermediates)/, $(GENERATED_SOURCES)) $(LOCAL_GENERATED_SOURCES): PRIVATE_PYTHON := $(MESA_PYTHON2) diff --git a/mesalib/src/gallium/auxiliary/hud/hud_context.c b/mesalib/src/gallium/auxiliary/hud/hud_context.c index e46c68cdd..00ec20589 100644 --- a/mesalib/src/gallium/auxiliary/hud/hud_context.c +++ b/mesalib/src/gallium/auxiliary/hud/hud_context.c @@ -569,9 +569,36 @@ hud_pane_set_max_value(struct hud_pane *pane, uint64_t value) pane->yscale = -(int)pane->inner_height / (float)pane->max_value; } +static void +hud_pane_update_dyn_ceiling(struct hud_graph *gr, struct hud_pane *pane) +{ + unsigned i; + float tmp = 0.0f; + + if (pane->dyn_ceil_last_ran != gr->index) { + LIST_FOR_EACH_ENTRY(gr, &pane->graph_list, head) { + for (i = 0; i < gr->num_vertices; ++i) { + tmp = gr->vertices[i * 2 + 1] > tmp ? + gr->vertices[i * 2 + 1] : tmp; + } + } + + /* Avoid setting it lower than the initial starting height. */ + tmp = tmp > pane->initial_max_value ? tmp : pane->initial_max_value; + hud_pane_set_max_value(pane, tmp); + } + + /* + * Mark this adjustment run so we could avoid repeating a full update + * again needlessly in case the pane has more than one graph. + */ + pane->dyn_ceil_last_ran = gr->index; +} + static struct hud_pane * hud_pane_create(unsigned x1, unsigned y1, unsigned x2, unsigned y2, - unsigned period, uint64_t max_value) + unsigned period, uint64_t max_value, uint64_t ceiling, + boolean dyn_ceiling) { struct hud_pane *pane = CALLOC_STRUCT(hud_pane); @@ -590,6 +617,10 @@ hud_pane_create(unsigned x1, unsigned y1, unsigned x2, unsigned y2, pane->inner_height = pane->inner_y2 - pane->inner_y1; pane->period = period; pane->max_num_vertices = (x2 - x1 + 2) / 2; + pane->ceiling = ceiling; + pane->dyn_ceiling = dyn_ceiling; + pane->dyn_ceil_last_ran = 0; + pane->initial_max_value = max_value; hud_pane_set_max_value(pane, max_value); LIST_INITHEAD(&pane->graph_list); return pane; @@ -633,6 +664,9 @@ hud_pane_add_graph(struct hud_pane *pane, struct hud_graph *gr) void hud_graph_add_value(struct hud_graph *gr, uint64_t value) { + gr->current_value = value; + value = value > gr->pane->ceiling ? gr->pane->ceiling : value; + if (gr->index == gr->pane->max_num_vertices) { gr->vertices[0] = 0; gr->vertices[1] = gr->vertices[(gr->index-1)*2+1]; @@ -646,7 +680,9 @@ hud_graph_add_value(struct hud_graph *gr, uint64_t value) gr->num_vertices++; } - gr->current_value = value; + if (gr->pane->dyn_ceiling == true) { + hud_pane_update_dyn_ceiling(gr, gr->pane); + } if (value > gr->pane->max_value) { hud_pane_set_max_value(gr->pane, value); } @@ -683,6 +719,69 @@ parse_string(const char *s, char *out) return i; } +static char * +read_pane_settings(char *str, unsigned * const x, unsigned * const y, + unsigned * const width, unsigned * const height, + uint64_t * const ceiling, boolean * const dyn_ceiling) +{ + char *ret = str; + unsigned tmp; + + while (*str == '.') { + ++str; + switch (*str) { + case 'x': + ++str; + *x = strtoul(str, &ret, 10); + str = ret; + break; + + case 'y': + ++str; + *y = strtoul(str, &ret, 10); + str = ret; + break; + + case 'w': + ++str; + tmp = strtoul(str, &ret, 10); + *width = tmp > 80 ? tmp : 80; /* 80 is chosen arbitrarily */ + str = ret; + break; + + /* + * Prevent setting height to less than 50. If the height is set to less, + * the text of the Y axis labels on the graph will start overlapping. + */ + case 'h': + ++str; + tmp = strtoul(str, &ret, 10); + *height = tmp > 50 ? tmp : 50; + str = ret; + break; + + case 'c': + ++str; + tmp = strtoul(str, &ret, 10); + *ceiling = tmp > 10 ? tmp : 10; + str = ret; + break; + + case 'd': + ++str; + ret = str; + *dyn_ceiling = true; + break; + + default: + fprintf(stderr, "gallium_hud: syntax error: unexpected '%c'\n", *str); + } + + } + + return ret; +} + static boolean has_occlusion_query(struct pipe_screen *screen) { @@ -705,11 +804,15 @@ static void hud_parse_env_var(struct hud_context *hud, const char *env) { unsigned num, i; - char name[256], s[256]; + char name_a[256], s[256]; + char *name; struct hud_pane *pane = NULL; unsigned x = 10, y = 10; unsigned width = 251, height = 100; unsigned period = 500 * 1000; /* default period (1/2 second) */ + uint64_t ceiling = UINT64_MAX; + unsigned column_width = 251; + boolean dyn_ceiling = false; const char *period_env; /* @@ -725,11 +828,23 @@ hud_parse_env_var(struct hud_context *hud, const char *env) } } - while ((num = parse_string(env, name)) != 0) { + while ((num = parse_string(env, name_a)) != 0) { env += num; + /* check for explicit location, size and etc. settings */ + name = read_pane_settings(name_a, &x, &y, &width, &height, &ceiling, + &dyn_ceiling); + + /* + * Keep track of overall column width to avoid pane overlapping in case + * later we create a new column while the bottom pane in the current + * column is less wide than the rest of the panes in it. + */ + column_width = width > column_width ? width : column_width; + if (!pane) { - pane = hud_pane_create(x, y, x + width, y + height, period, 10); + pane = hud_pane_create(x, y, x + width, y + height, period, 10, + ceiling, dyn_ceiling); if (!pane) return; } @@ -807,6 +922,7 @@ hud_parse_env_var(struct hud_context *hud, const char *env) if (num && sscanf(s, "%u", &i) == 1) { hud_pane_set_max_value(pane, i); + pane->initial_max_value = i; } else { fprintf(stderr, "gallium_hud: syntax error: unexpected '%c' (%i) " @@ -826,6 +942,7 @@ hud_parse_env_var(struct hud_context *hud, const char *env) case ',': env++; y += height + hud->font.glyph_height * (pane->num_graphs + 2); + height = 100; if (pane && pane->num_graphs) { LIST_ADDTAIL(&pane->head, &hud->pane_list); @@ -836,17 +953,27 @@ hud_parse_env_var(struct hud_context *hud, const char *env) case ';': env++; y = 10; - x += width + hud->font.glyph_width * 7; + x += column_width + hud->font.glyph_width * 7; + height = 100; if (pane && pane->num_graphs) { LIST_ADDTAIL(&pane->head, &hud->pane_list); pane = NULL; } + + /* Starting a new column; reset column width. */ + column_width = 251; break; default: fprintf(stderr, "gallium_hud: syntax error: unexpected '%c'\n", *env); } + + /* Reset to defaults for the next pane in case these were modified. */ + width = 251; + ceiling = UINT64_MAX; + dyn_ceiling = false; + } if (pane) { @@ -878,6 +1005,30 @@ print_help(struct pipe_screen *screen) puts(""); puts(" Example: GALLIUM_HUD=\"cpu,fps;primitives-generated\""); puts(""); + puts(" Additionally, by prepending '.[identifier][value]' modifiers to"); + puts(" a name, it is possible to explicitly set the location and size"); + puts(" of a pane, along with limiting overall maximum value of the"); + puts(" Y axis and activating dynamic readjustment of the Y axis."); + puts(" Several modifiers may be applied to the same pane simultaneously."); + puts(""); + puts(" 'x[value]' sets the location of the pane on the x axis relative"); + puts(" to the upper-left corner of the viewport, in pixels."); + puts(" 'y[value]' sets the location of the pane on the y axis relative"); + puts(" to the upper-left corner of the viewport, in pixels."); + puts(" 'w[value]' sets width of the graph pixels."); + puts(" 'h[value]' sets height of the graph in pixels."); + puts(" 'c[value]' sets the ceiling of the value of the Y axis."); + puts(" If the graph needs to draw values higher than"); + puts(" the ceiling allows, the value is clamped."); + puts(" 'd' activates dynamic Y axis readjustment to set the value of"); + puts(" the Y axis to match the highest value still visible in the graph."); + puts(""); + puts(" If 'c' and 'd' modifiers are used simultaneously, both are in effect:"); + puts(" the Y axis does not go above the restriction imposed by 'c' while"); + puts(" still adjusting the value of the Y axis down when appropriate."); + puts(""); + puts(" Example: GALLIUM_HUD=\".w256.h64.x1600.y520.d.c1000fps+cpu,.datom-count\""); + puts(""); puts(" Available names:"); puts(" fps"); puts(" cpu"); diff --git a/mesalib/src/gallium/auxiliary/hud/hud_private.h b/mesalib/src/gallium/auxiliary/hud/hud_private.h index 1606ada4a..230f02664 100644 --- a/mesalib/src/gallium/auxiliary/hud/hud_private.h +++ b/mesalib/src/gallium/auxiliary/hud/hud_private.h @@ -62,6 +62,10 @@ struct hud_pane { float yscale; unsigned max_num_vertices; uint64_t max_value; + uint64_t initial_max_value; + uint64_t ceiling; + unsigned dyn_ceil_last_ran; + boolean dyn_ceiling; boolean uses_byte_units; uint64_t period; /* in microseconds */ diff --git a/mesalib/src/gallium/auxiliary/util/u_tile.c b/mesalib/src/gallium/auxiliary/util/u_tile.c index 8e199200f..b91bb413a 100644 --- a/mesalib/src/gallium/auxiliary/util/u_tile.c +++ b/mesalib/src/gallium/auxiliary/util/u_tile.c @@ -341,13 +341,13 @@ x32_s8_get_tile_rgba(const unsigned *src, unsigned i, j; for (i = 0; i < h; i++) { - float *pRow = p; + uint32_t *pRow = (uint32_t *)p; for (j = 0; j < w; j++, pRow += 4) { src++; pRow[0] = pRow[1] = pRow[2] = - pRow[3] = (float)(*src++ & 0xff); + pRow[3] = (*src++ & 0xff); } p += dst_stride; } diff --git a/mesalib/src/gallium/auxiliary/util/u_video.h b/mesalib/src/gallium/auxiliary/util/u_video.h index 45b2d6e76..b4743d13f 100644 --- a/mesalib/src/gallium/auxiliary/util/u_video.h +++ b/mesalib/src/gallium/auxiliary/util/u_video.h @@ -38,6 +38,7 @@ extern "C" { /* u_reduce_video_profile() needs these */ #include "pipe/p_compiler.h" #include "util/u_debug.h" +#include "util/u_math.h" static INLINE enum pipe_video_format u_reduce_video_profile(enum pipe_video_profile profile) @@ -146,6 +147,41 @@ u_copy_swap422_packed(void *const *destination_data, } } +static INLINE uint32_t +u_get_h264_level(uint32_t width, uint32_t height, uint32_t *max_reference) +{ + uint32_t max_dpb_mbs; + + width = align(width, 16); + height = align(height, 16); + + /* Max references will be used for caculation of number of DPB buffers + in the UVD driver, limitation of max references is 16. Some client + like mpv application for VA-API, it requires references more than that, + so we have to set max of references to 16 here. */ + *max_reference = MIN2(*max_reference, 16); + max_dpb_mbs = (width / 16) * (height / 16) * *max_reference; + + /* The calculation is based on "Decoded picture buffering" section + from http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC */ + if (max_dpb_mbs <= 8100) + return 30; + else if (max_dpb_mbs <= 18000) + return 31; + else if (max_dpb_mbs <= 20480) + return 32; + else if (max_dpb_mbs <= 32768) + return 41; + else if (max_dpb_mbs <= 34816) + return 42; + else if (max_dpb_mbs <= 110400) + return 50; + else if (max_dpb_mbs <= 184320) + return 51; + else + return 52; +} + #ifdef __cplusplus } #endif diff --git a/mesalib/src/glsl/Android.gen.mk b/mesalib/src/glsl/Android.gen.mk index 7ec56d4f2..0835871f9 100644 --- a/mesalib/src/glsl/Android.gen.mk +++ b/mesalib/src/glsl/Android.gen.mk @@ -27,17 +27,29 @@ ifeq ($(LOCAL_MODULE_CLASS),) LOCAL_MODULE_CLASS := STATIC_LIBRARIES endif -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) sources := \ glsl_lexer.cpp \ glsl_parser.cpp \ glcpp/glcpp-lex.c \ - glcpp/glcpp-parse.c + glcpp/glcpp-parse.c \ + nir/nir_builder_opcodes.h \ + nir/nir_constant_expressions.c \ + nir/nir_opcodes.c \ + nir/nir_opcodes.h \ + nir/nir_opt_algebraic.c LOCAL_SRC_FILES := $(filter-out $(sources), $(LOCAL_SRC_FILES)) -LOCAL_C_INCLUDES += $(intermediates) $(intermediates)/glcpp $(MESA_TOP)/src/glsl/glcpp +LOCAL_C_INCLUDES += \ + $(intermediates)/glcpp \ + $(intermediates)/nir \ + $(MESA_TOP)/src/glsl/glcpp \ + $(MESA_TOP)/src/glsl/nir + +LOCAL_EXPORT_C_INCLUDE_DIRS += \ + $(intermediates)/nir sources := $(addprefix $(intermediates)/, $(sources)) LOCAL_GENERATED_SOURCES += $(sources) @@ -77,3 +89,51 @@ $(intermediates)/glcpp/glcpp-lex.c: $(LOCAL_PATH)/glcpp/glcpp-lex.l $(intermediates)/glcpp/glcpp-parse.c: $(LOCAL_PATH)/glcpp/glcpp-parse.y $(call glsl_local-y-to-c-and-h) + +nir_builder_opcodes_gen := $(LOCAL_PATH)/nir/nir_builder_opcodes_h.py +nir_builder_opcodes_deps := \ + $(LOCAL_PATH)/nir/nir_opcodes.py \ + $(LOCAL_PATH)/nir/nir_builder_opcodes_h.py + +$(intermediates)/nir/nir_builder_opcodes.h: $(nir_builder_opcodes_deps) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(nir_builder_opcodes_gen) $< > $@ + +nir_constant_expressions_gen := $(LOCAL_PATH)/nir/nir_constant_expressions.py +nir_constant_expressions_deps := \ + $(LOCAL_PATH)/nir/nir_opcodes.py \ + $(LOCAL_PATH)/nir/nir_constant_expressions.py \ + $(LOCAL_PATH)/nir/nir_constant_expressions.h + +$(intermediates)/nir/nir_constant_expressions.c: $(nir_constant_expressions_deps) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(nir_constant_expressions_gen) $< > $@ + +nir_opcodes_h_gen := $(LOCAL_PATH)/nir/nir_opcodes_h.py +nir_opcodes_h_deps := \ + $(LOCAL_PATH)/nir/nir_opcodes.py \ + $(LOCAL_PATH)/nir/nir_opcodes_h.py + +$(intermediates)/nir/nir_opcodes.h: $(nir_opcodes_h_deps) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(nir_opcodes_h_gen) $< > $@ + +$(LOCAL_PATH)/nir/nir.h: $(intermediates)/nir/nir_opcodes.h + +nir_opcodes_c_gen := $(LOCAL_PATH)/nir/nir_opcodes_c.py +nir_opcodes_c_deps := \ + $(LOCAL_PATH)/nir/nir_opcodes.py \ + $(LOCAL_PATH)/nir/nir_opcodes_c.py + +$(intermediates)/nir/nir_opcodes.c: $(nir_opcodes_c_deps) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(nir_opcodes_c_gen) $< > $@ + +nir_opt_algebraic_gen := $(LOCAL_PATH)/nir/nir_opt_algebraic.py +nir_opt_algebraic_deps := \ + $(LOCAL_PATH)/nir/nir_opt_algebraic.py \ + $(LOCAL_PATH)/nir/nir_algebraic.py + +$(intermediates)/nir/nir_opt_algebraic.c: $(nir_opt_algebraic_deps) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(nir_opt_algebraic_gen) $< > $@ diff --git a/mesalib/src/glsl/Android.mk b/mesalib/src/glsl/Android.mk index 38c2087a4..f20741e0d 100644 --- a/mesalib/src/glsl/Android.mk +++ b/mesalib/src/glsl/Android.mk @@ -35,12 +35,14 @@ include $(CLEAR_VARS) LOCAL_SRC_FILES := \ $(LIBGLCPP_FILES) \ - $(LIBGLSL_FILES) + $(LIBGLSL_FILES) \ + $(NIR_FILES) LOCAL_C_INCLUDES := \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/mapi \ - $(MESA_TOP)/src/mesa + $(MESA_TOP)/src/mesa \ + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_MODULE := libmesa_glsl @@ -59,9 +61,10 @@ LOCAL_SRC_FILES := \ $(GLSL_COMPILER_CXX_FILES) LOCAL_C_INCLUDES := \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/mapi \ - $(MESA_TOP)/src/mesa + $(MESA_TOP)/src/mesa \ + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_STATIC_LIBRARIES := libmesa_glsl libmesa_glsl_utils libmesa_util diff --git a/mesalib/src/glsl/ast_function.cpp b/mesalib/src/glsl/ast_function.cpp index 87df93e68..758361324 100644 --- a/mesalib/src/glsl/ast_function.cpp +++ b/mesalib/src/glsl/ast_function.cpp @@ -1191,7 +1191,7 @@ emit_inline_matrix_constructor(const glsl_type *type, * * - Construct a matrix from an arbirary combination of vectors and * scalars. The components of the constructor parameters are assigned - * to the matrix in colum-major order until the matrix is full. + * to the matrix in column-major order until the matrix is full. * * - Construct a matrix from a single matrix. The source matrix is copied * to the upper left portion of the constructed matrix, and the remaining diff --git a/mesalib/src/glsl/ast_to_hir.cpp b/mesalib/src/glsl/ast_to_hir.cpp index 78369360f..18b82e3be 100644 --- a/mesalib/src/glsl/ast_to_hir.cpp +++ b/mesalib/src/glsl/ast_to_hir.cpp @@ -3532,9 +3532,7 @@ ast_declarator_list::hir(exec_list *instructions, * vectors. Vertex shader inputs cannot be arrays or * structures." */ - const glsl_type *check_type = var->type; - while (check_type->is_array()) - check_type = check_type->element_type(); + const glsl_type *check_type = var->type->without_array(); switch (check_type->base_type) { case GLSL_TYPE_FLOAT: diff --git a/mesalib/src/glsl/glcpp/glcpp-parse.y b/mesalib/src/glsl/glcpp/glcpp-parse.y index c2f5223a9..cfceca66b 100644 --- a/mesalib/src/glsl/glcpp/glcpp-parse.y +++ b/mesalib/src/glsl/glcpp/glcpp-parse.y @@ -1463,7 +1463,7 @@ _arguments_parse (argument_list_t *arguments, else { if (argument->head == NULL) { /* Don't treat initial whitespace as - * part of the arguement. */ + * part of the argument. */ if (node->token->type == SPACE) continue; } diff --git a/mesalib/src/glsl/glsl_types.cpp b/mesalib/src/glsl/glsl_types.cpp index 4aa36a794..9c9b7efcb 100644 --- a/mesalib/src/glsl/glsl_types.cpp +++ b/mesalib/src/glsl/glsl_types.cpp @@ -738,24 +738,27 @@ glsl_type::record_key_compare(const void *a, const void *b) } +/** + * Generate an integer hash value for a glsl_type structure type. + */ unsigned glsl_type::record_key_hash(const void *a) { const glsl_type *const key = (glsl_type *) a; - char hash_key[128]; - unsigned size = 0; - - size = snprintf(hash_key, sizeof(hash_key), "%08x", key->length); + uintptr_t hash = key->length; + unsigned retval; 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); + /* casting pointer to uintptr_t */ + hash = (hash * 13 ) + (uintptr_t) key->fields.structure[i].type; } - return hash_table_string_hash(& hash_key); + if (sizeof(hash) == 8) + retval = (hash & 0xffffffff) ^ ((uint64_t) hash >> 32); + else + retval = hash; + + return retval; } diff --git a/mesalib/src/glsl/glsl_types.h b/mesalib/src/glsl/glsl_types.h index d383dd5be..5645dcd50 100644 --- a/mesalib/src/glsl/glsl_types.h +++ b/mesalib/src/glsl/glsl_types.h @@ -540,7 +540,12 @@ struct glsl_type { */ const glsl_type *without_array() const { - return this->is_array() ? this->fields.array : this; + const glsl_type *t = this; + + while (t->is_array()) + t = t->fields.array; + + return t; } /** diff --git a/mesalib/src/glsl/link_uniforms.cpp b/mesalib/src/glsl/link_uniforms.cpp index 59adc298b..d5ca23a38 100644 --- a/mesalib/src/glsl/link_uniforms.cpp +++ b/mesalib/src/glsl/link_uniforms.cpp @@ -274,7 +274,7 @@ namespace { * * As uniforms are added to the active set the number of active uniforms and * the storage requirements for those uniforms are accumulated. The active - * uniforms are added the the hash table supplied to the constructor. + * uniforms are added to the hash table supplied to the constructor. * * If the same uniform is added multiple times (i.e., once for each shader * target), it will only be accounted once. diff --git a/mesalib/src/glsl/linker.cpp b/mesalib/src/glsl/linker.cpp index b6baa5d36..21fde9444 100644 --- a/mesalib/src/glsl/linker.cpp +++ b/mesalib/src/glsl/linker.cpp @@ -2570,7 +2570,8 @@ add_interface_variables(struct gl_shader_program *shProg, if (var->data.location != SYSTEM_VALUE_VERTEX_ID && var->data.location != SYSTEM_VALUE_VERTEX_ID_ZERO_BASE && var->data.location != SYSTEM_VALUE_INSTANCE_ID) - continue; + continue; + /* FALLTHROUGH */ case ir_var_shader_in: if (programInterface != GL_PROGRAM_INPUT) continue; diff --git a/mesalib/src/glsl/lower_packed_varyings.cpp b/mesalib/src/glsl/lower_packed_varyings.cpp index 2c9a1c495..d8bebb522 100644 --- a/mesalib/src/glsl/lower_packed_varyings.cpp +++ b/mesalib/src/glsl/lower_packed_varyings.cpp @@ -645,13 +645,7 @@ lower_packed_varyings_visitor::needs_lowering(ir_variable *var) if (var->data.explicit_location) return false; - const glsl_type *type = var->type; - if (this->gs_input_vertices != 0) { - assert(type->is_array()); - type = type->element_type(); - } - if (type->is_array()) - type = type->fields.array; + const glsl_type *type = var->type->without_array(); if (type->vector_elements == 4 && !type->is_double()) return false; return true; diff --git a/mesalib/src/glsl/lower_packing_builtins.cpp b/mesalib/src/glsl/lower_packing_builtins.cpp index db73c7b0f..a6fb8a883 100644 --- a/mesalib/src/glsl/lower_packing_builtins.cpp +++ b/mesalib/src/glsl/lower_packing_builtins.cpp @@ -569,8 +569,8 @@ private: * return pack_uvec2_to_uint(uvec2( * round(clamp(VEC2_RVALUE, 0.0f, 1.0f) * 65535.0f))); * - * Here it is safe to directly convert the vec2 to uvec2 because the the - * vec2 has been clamped to a non-negative range. + * Here it is safe to directly convert the vec2 to uvec2 because the vec2 + * has been clamped to a non-negative range. */ assert(vec2_rval->type == glsl_type::vec2_type); @@ -613,8 +613,8 @@ private: * return pack_uvec4_to_uint(uvec4( * round(clamp(VEC2_RVALUE, 0.0f, 1.0f) * 255.0f))); * - * Here it is safe to directly convert the vec4 to uvec4 because the the - * vec4 has been clamped to a non-negative range. + * Here it is safe to directly convert the vec4 to uvec4 because the vec4 + * has been clamped to a non-negative range. */ assert(vec4_rval->type == glsl_type::vec4_type); diff --git a/mesalib/src/glsl/nir/nir.c b/mesalib/src/glsl/nir/nir.c index c6e53612b..4cc074b80 100644 --- a/mesalib/src/glsl/nir/nir.c +++ b/mesalib/src/glsl/nir/nir.c @@ -589,6 +589,66 @@ nir_copy_deref(void *mem_ctx, nir_deref *deref) return NULL; } +/* 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. + */ +nir_load_const_instr * +nir_deref_get_const_initializer_load(nir_shader *shader, nir_deref_var *deref) +{ + nir_constant *constant = deref->var->constant_initializer; + assert(constant); + + const nir_deref *tail = &deref->deref; + unsigned matrix_offset = 0; + while (tail->child) { + switch (tail->child->deref_type) { + case nir_deref_type_array: { + nir_deref_array *arr = nir_deref_as_array(tail->child); + assert(arr->deref_array_type == nir_deref_array_type_direct); + if (glsl_type_is_matrix(tail->type)) { + assert(arr->deref.child == NULL); + matrix_offset = arr->base_offset; + } else { + constant = constant->elements[arr->base_offset]; + } + break; + } + + case nir_deref_type_struct: { + constant = constant->elements[nir_deref_as_struct(tail->child)->index]; + break; + } + + default: + unreachable("Invalid deref child type"); + } + + tail = tail->child; + } + + nir_load_const_instr *load = + nir_load_const_instr_create(shader, glsl_get_vector_elements(tail->type)); + + matrix_offset *= load->def.num_components; + for (unsigned i = 0; i < load->def.num_components; i++) { + switch (glsl_get_base_type(tail->type)) { + case GLSL_TYPE_FLOAT: + case GLSL_TYPE_INT: + case GLSL_TYPE_UINT: + load->value.u[i] = constant->value.u[matrix_offset + i]; + break; + case GLSL_TYPE_BOOL: + load->value.u[i] = constant->value.b[matrix_offset + i] ? + NIR_TRUE : NIR_FALSE; + break; + default: + unreachable("Invalid immediate type"); + } + } + + return load; +} /** * \name Control flow modification @@ -1800,33 +1860,37 @@ src_does_not_use_reg(nir_src *src, void *void_reg) void nir_instr_rewrite_src(nir_instr *instr, nir_src *src, nir_src new_src) { - if (src->is_ssa) { - nir_ssa_def *old_ssa = src->ssa; - *src = new_src; - if (old_ssa && nir_foreach_src(instr, src_does_not_use_def, old_ssa)) { - struct set_entry *entry = _mesa_set_search(old_ssa->uses, instr); - assert(entry); - _mesa_set_remove(old_ssa->uses, entry); - } - } else { - if (src->reg.indirect) - nir_instr_rewrite_src(instr, src->reg.indirect, new_src); - - nir_register *old_reg = src->reg.reg; - *src = new_src; - if (old_reg && nir_foreach_src(instr, src_does_not_use_reg, old_reg)) { - struct set_entry *entry = _mesa_set_search(old_reg->uses, instr); - assert(entry); - _mesa_set_remove(old_reg->uses, entry); + nir_src old_src = *src; + *src = new_src; + + for (nir_src *iter_src = &old_src; iter_src; + iter_src = iter_src->is_ssa ? NULL : iter_src->reg.indirect) { + if (iter_src->is_ssa) { + nir_ssa_def *ssa = iter_src->ssa; + if (ssa && nir_foreach_src(instr, src_does_not_use_def, ssa)) { + struct set_entry *entry = _mesa_set_search(ssa->uses, instr); + assert(entry); + _mesa_set_remove(ssa->uses, entry); + } + } else { + nir_register *reg = iter_src->reg.reg; + if (reg && nir_foreach_src(instr, src_does_not_use_reg, reg)) { + struct set_entry *entry = _mesa_set_search(reg->uses, instr); + assert(entry); + _mesa_set_remove(reg->uses, entry); + } } } - if (new_src.is_ssa) { - if (new_src.ssa) - _mesa_set_add(new_src.ssa->uses, instr); - } else { - if (new_src.reg.reg) - _mesa_set_add(new_src.reg.reg->uses, instr); + for (nir_src *iter_src = &new_src; iter_src; + iter_src = iter_src->is_ssa ? NULL : iter_src->reg.indirect) { + if (iter_src->is_ssa) { + if (iter_src->ssa) + _mesa_set_add(iter_src->ssa->uses, instr); + } else { + if (iter_src->reg.reg) + _mesa_set_add(iter_src->reg.reg->uses, instr); + } } } diff --git a/mesalib/src/glsl/nir/nir.h b/mesalib/src/glsl/nir/nir.h index 74772c798..98b0ec328 100644 --- a/mesalib/src/glsl/nir/nir.h +++ b/mesalib/src/glsl/nir/nir.h @@ -960,7 +960,8 @@ typedef struct { static inline unsigned nir_tex_instr_dest_size(nir_tex_instr *instr) { - if (instr->op == nir_texop_txs) { + switch (instr->op) { + case nir_texop_txs: { unsigned ret; switch (instr->sampler_dim) { case GLSL_SAMPLER_DIM_1D: @@ -985,13 +986,18 @@ nir_tex_instr_dest_size(nir_tex_instr *instr) return ret; } - if (instr->op == nir_texop_query_levels) + case nir_texop_lod: return 2; - if (instr->is_shadow && instr->is_new_style_shadow) + case nir_texop_query_levels: return 1; - return 4; + default: + if (instr->is_shadow && instr->is_new_style_shadow) + return 1; + + return 4; + } } static inline unsigned @@ -1514,6 +1520,9 @@ nir_deref_struct *nir_deref_struct_create(void *mem_ctx, unsigned field_index); nir_deref *nir_copy_deref(void *mem_ctx, nir_deref *deref); +nir_load_const_instr * +nir_deref_get_const_initializer_load(nir_shader *shader, nir_deref_var *deref); + void nir_instr_insert_before(nir_instr *instr, nir_instr *before); void nir_instr_insert_after(nir_instr *instr, nir_instr *after); diff --git a/mesalib/src/glsl/nir/nir_array.h b/mesalib/src/glsl/nir/nir_array.h new file mode 100644 index 000000000..1db4e8cea --- /dev/null +++ b/mesalib/src/glsl/nir/nir_array.h @@ -0,0 +1,96 @@ +/* + * Copyright © 2015 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. + * + * Authors: + * Jason Ekstrand (jason@jlekstrand.net) + * + */ + +#pragma once + +#ifdef __cplusplus +extern "C" { +#endif + +typedef struct { + void *mem_ctx; + size_t size; + size_t alloc; + void *data; +} nir_array; + +static inline void +nir_array_init(nir_array *arr, void *mem_ctx) +{ + arr->mem_ctx = mem_ctx; + arr->size = 0; + arr->alloc = 0; + arr->data = NULL; +} + +static inline void +nir_array_fini(nir_array *arr) +{ + if (arr->mem_ctx) + ralloc_free(arr->data); + else + free(arr->data); +} + +#define NIR_ARRAY_INITIAL_SIZE 64 + +/* Increments the size of the array by the given ammount and returns a + * pointer to the beginning of the newly added space. + */ +static inline void * +nir_array_grow(nir_array *arr, size_t additional) +{ + size_t new_size = arr->size + additional; + if (new_size > arr->alloc) { + if (arr->alloc == 0) + arr->alloc = NIR_ARRAY_INITIAL_SIZE; + + while (new_size > arr->alloc) + arr->alloc *= 2; + + if (arr->mem_ctx) + arr->data = reralloc_size(arr->mem_ctx, arr->data, arr->alloc); + else + arr->data = realloc(arr->data, arr->alloc); + } + + void *ptr = (void *)((char *)arr->data + arr->size); + arr->size = new_size; + + return ptr; +} + +#define nir_array_add(arr, type, elem) \ + *(type *)nir_array_grow(arr, sizeof(type)) = (elem) + +#define nir_array_foreach(arr, type, elem) \ + for (type *elem = (type *)(arr)->data; \ + elem < (type *)((char *)(arr)->data + (arr)->size); elem++) + +#ifdef __cplusplus +} /* extern "C" */ +#endif diff --git a/mesalib/src/glsl/nir/nir_from_ssa.c b/mesalib/src/glsl/nir/nir_from_ssa.c index 184698abd..6a3b141bd 100644 --- a/mesalib/src/glsl/nir/nir_from_ssa.c +++ b/mesalib/src/glsl/nir/nir_from_ssa.c @@ -642,7 +642,7 @@ emit_copy(nir_parallel_copy_instr *pcopy, nir_src src, nir_src dest_src, * * The algorithm works by playing this little shell game with the values. * We start by recording where every source value is and which source value - * each destination value should recieve. We then grab any copy whose + * each destination value should receive. We then grab any copy whose * destination is "empty", i.e. not used as a source, and do the following: * - Find where its source value currently lives * - Emit the move instruction diff --git a/mesalib/src/glsl/nir/nir_lower_locals_to_regs.c b/mesalib/src/glsl/nir/nir_lower_locals_to_regs.c index 8c5df7be6..bc6a3d320 100644 --- a/mesalib/src/glsl/nir/nir_lower_locals_to_regs.c +++ b/mesalib/src/glsl/nir/nir_lower_locals_to_regs.c @@ -26,13 +26,20 @@ */ #include "nir.h" +#include "nir_array.h" struct locals_to_regs_state { - void *mem_ctx; + nir_shader *shader; nir_function_impl *impl; /* A hash table mapping derefs to registers */ struct hash_table *regs_table; + + /* A growing array of derefs that we have encountered. There is exactly + * one element of this array per element in the hash table. This is + * used to make adding register initialization code deterministic. + */ + nir_array derefs_array; }; /* The following two functions implement a hash and equality check for @@ -100,15 +107,8 @@ get_reg_for_deref(nir_deref_var *deref, struct locals_to_regs_state *state) unsigned array_size = 1; nir_deref *tail = &deref->deref; while (tail->child) { - if (tail->child->deref_type == nir_deref_type_array) { - /* Multiply by the parent's type. */ - if (glsl_type_is_matrix(tail->type)) { - array_size *= glsl_get_matrix_columns(tail->type); - } else { - assert(glsl_get_length(tail->type) > 0); - array_size *= glsl_get_length(tail->type); - } - } + if (tail->child->deref_type == nir_deref_type_array) + array_size *= glsl_get_length(tail->type); tail = tail->child; } @@ -119,6 +119,7 @@ get_reg_for_deref(nir_deref_var *deref, struct locals_to_regs_state *state) reg->num_array_elems = array_size > 1 ? array_size : 0; _mesa_hash_table_insert_pre_hashed(state->regs_table, hash, deref, reg); + nir_array_add(&state->derefs_array, nir_deref_var *, deref); return reg; } @@ -134,6 +135,14 @@ get_deref_reg_src(nir_deref_var *deref, nir_instr *instr, src.reg.base_offset = 0; src.reg.indirect = NULL; + /* It is possible for a user to create a shader that has an array with a + * single element and then proceed to access it indirectly. Indirectly + * accessing a non-array register is not allowed in NIR. In order to + * handle this case we just convert it to a direct reference. + */ + if (src.reg.reg->num_array_elems == 0) + return src; + nir_deref *tail = &deref->deref; while (tail->child != NULL) { const struct glsl_type *parent_type = tail->type; @@ -149,11 +158,11 @@ get_deref_reg_src(nir_deref_var *deref, nir_instr *instr, if (src.reg.indirect) { nir_load_const_instr *load_const = - nir_load_const_instr_create(state->mem_ctx, 1); + nir_load_const_instr_create(state->shader, 1); load_const->value.u[0] = glsl_get_length(parent_type); nir_instr_insert_before(instr, &load_const->instr); - nir_alu_instr *mul = nir_alu_instr_create(state->mem_ctx, nir_op_imul); + nir_alu_instr *mul = nir_alu_instr_create(state->shader, nir_op_imul); mul->src[0].src = *src.reg.indirect; mul->src[1].src.is_ssa = true; mul->src[1].src.ssa = &load_const->def; @@ -167,15 +176,15 @@ get_deref_reg_src(nir_deref_var *deref, nir_instr *instr, if (deref_array->deref_array_type == nir_deref_array_type_indirect) { if (src.reg.indirect == NULL) { - src.reg.indirect = ralloc(state->mem_ctx, nir_src); + src.reg.indirect = ralloc(state->shader, nir_src); nir_src_copy(src.reg.indirect, &deref_array->indirect, - state->mem_ctx); + state->shader); } else { - nir_alu_instr *add = nir_alu_instr_create(state->mem_ctx, + nir_alu_instr *add = nir_alu_instr_create(state->shader, nir_op_iadd); add->src[0].src = *src.reg.indirect; nir_src_copy(&add->src[1].src, &deref_array->indirect, - state->mem_ctx); + state->shader); add->dest.write_mask = 1; nir_ssa_dest_init(&add->instr, &add->dest.dest, 1, NULL); nir_instr_insert_before(instr, &add->instr); @@ -205,7 +214,7 @@ lower_locals_to_regs_block(nir_block *block, void *void_state) if (intrin->variables[0]->var->data.mode != nir_var_local) continue; - nir_alu_instr *mov = nir_alu_instr_create(state->mem_ctx, nir_op_imov); + nir_alu_instr *mov = nir_alu_instr_create(state->shader, nir_op_imov); mov->src[0].src = get_deref_reg_src(intrin->variables[0], &intrin->instr, state); mov->dest.write_mask = (1 << intrin->num_components) - 1; @@ -214,9 +223,9 @@ lower_locals_to_regs_block(nir_block *block, void *void_state) intrin->num_components, NULL); nir_ssa_def_rewrite_uses(&intrin->dest.ssa, nir_src_for_ssa(&mov->dest.dest.ssa), - state->mem_ctx); + state->shader); } else { - nir_dest_copy(&mov->dest.dest, &intrin->dest, state->mem_ctx); + nir_dest_copy(&mov->dest.dest, &intrin->dest, state->shader); } nir_instr_insert_before(&intrin->instr, &mov->instr); @@ -231,8 +240,8 @@ lower_locals_to_regs_block(nir_block *block, void *void_state) nir_src reg_src = get_deref_reg_src(intrin->variables[0], &intrin->instr, state); - nir_alu_instr *mov = nir_alu_instr_create(state->mem_ctx, nir_op_imov); - nir_src_copy(&mov->src[0].src, &intrin->src[0], state->mem_ctx); + nir_alu_instr *mov = nir_alu_instr_create(state->shader, nir_op_imov); + nir_src_copy(&mov->src[0].src, &intrin->src[0], state->shader); mov->dest.write_mask = (1 << intrin->num_components) - 1; mov->dest.dest.is_ssa = false; mov->dest.dest.reg.reg = reg_src.reg.reg; @@ -257,20 +266,115 @@ lower_locals_to_regs_block(nir_block *block, void *void_state) return true; } +static nir_block * +compute_reg_usedef_lca(nir_register *reg) +{ + struct set_entry *entry; + nir_block *lca = NULL; + + set_foreach(reg->defs, entry) + lca = nir_dominance_lca(lca, ((nir_instr *)entry->key)->block); + + set_foreach(reg->uses, entry) + lca = nir_dominance_lca(lca, ((nir_instr *)entry->key)->block); + + set_foreach(reg->if_uses, entry) { + nir_if *if_stmt = (nir_if *)entry->key; + nir_cf_node *prev_node = nir_cf_node_prev(&if_stmt->cf_node); + assert(prev_node->type == nir_cf_node_block); + lca = nir_dominance_lca(lca, nir_cf_node_as_block(prev_node)); + } + + return lca; +} + +static void +insert_constant_initializer(nir_deref_var *deref_head, nir_deref *deref_tail, + nir_block *block, + struct locals_to_regs_state *state) +{ + if (deref_tail->child) { + switch (deref_tail->child->deref_type) { + case nir_deref_type_array: { + unsigned array_elems = glsl_get_length(deref_tail->type); + + nir_deref_array arr_deref; + arr_deref.deref = *deref_tail->child; + arr_deref.deref_array_type = nir_deref_array_type_direct; + + nir_deref *old_child = deref_tail->child; + deref_tail->child = &arr_deref.deref; + for (unsigned i = 0; i < array_elems; i++) { + arr_deref.base_offset = i; + insert_constant_initializer(deref_head, &arr_deref.deref, + block, state); + } + deref_tail->child = old_child; + return; + } + + case nir_deref_type_struct: + insert_constant_initializer(deref_head, deref_tail->child, + block, state); + return; + + default: + unreachable("Invalid deref child type"); + } + } + + assert(deref_tail->child == NULL); + + nir_load_const_instr *load = + nir_deref_get_const_initializer_load(state->shader, deref_head); + nir_instr_insert_before_block(block, &load->instr); + + nir_src reg_src = get_deref_reg_src(deref_head, &load->instr, state); + + nir_alu_instr *mov = nir_alu_instr_create(state->shader, nir_op_imov); + mov->src[0].src = nir_src_for_ssa(&load->def); + mov->dest.write_mask = (1 << load->def.num_components) - 1; + mov->dest.dest.is_ssa = false; + mov->dest.dest.reg.reg = reg_src.reg.reg; + mov->dest.dest.reg.base_offset = reg_src.reg.base_offset; + mov->dest.dest.reg.indirect = reg_src.reg.indirect; + + nir_instr_insert_after(&load->instr, &mov->instr); +} + static void nir_lower_locals_to_regs_impl(nir_function_impl *impl) { struct locals_to_regs_state state; - state.mem_ctx = ralloc_parent(impl); + state.shader = impl->overload->function->shader; state.impl = impl; state.regs_table = _mesa_hash_table_create(NULL, hash_deref, derefs_equal); + nir_array_init(&state.derefs_array, NULL); + + nir_metadata_require(impl, nir_metadata_dominance); nir_foreach_block(impl, lower_locals_to_regs_block, &state); + nir_array_foreach(&state.derefs_array, nir_deref_var *, deref_ptr) { + nir_deref_var *deref = *deref_ptr; + struct hash_entry *deref_entry = + _mesa_hash_table_search(state.regs_table, deref); + assert(deref_entry && deref_entry->key == deref); + nir_register *reg = (nir_register *)deref_entry->data; + + if (deref->var->constant_initializer == NULL) + continue; + + nir_block *usedef_lca = compute_reg_usedef_lca(reg); + + insert_constant_initializer(deref, &deref->deref, usedef_lca, &state); + } + nir_metadata_preserve(impl, nir_metadata_block_index | nir_metadata_dominance); + nir_array_fini(&state.derefs_array); _mesa_hash_table_destroy(state.regs_table, NULL); } diff --git a/mesalib/src/glsl/nir/nir_lower_to_source_mods.c b/mesalib/src/glsl/nir/nir_lower_to_source_mods.c index d6bf77f17..7b4a0f657 100644 --- a/mesalib/src/glsl/nir/nir_lower_to_source_mods.c +++ b/mesalib/src/glsl/nir/nir_lower_to_source_mods.c @@ -67,6 +67,13 @@ nir_lower_to_source_mods_block(nir_block *block, void *state) continue; } + /* We can only do a rewrite if the source we are copying is SSA. + * Otherwise, moving the read might invalidly reorder reads/writes + * on a register. + */ + if (!parent->src[0].src.is_ssa) + continue; + nir_instr_rewrite_src(instr, &alu->src[i].src, parent->src[0].src); if (alu->src[i].abs) { /* abs trumps both neg and abs, do nothing */ diff --git a/mesalib/src/glsl/nir/nir_lower_var_copies.c b/mesalib/src/glsl/nir/nir_lower_var_copies.c index 58389a7c7..21672901f 100644 --- a/mesalib/src/glsl/nir/nir_lower_var_copies.c +++ b/mesalib/src/glsl/nir/nir_lower_var_copies.c @@ -64,26 +64,6 @@ get_deref_tail(nir_deref *deref) return deref; } -static int -type_get_length(const struct glsl_type *type) -{ - switch (glsl_get_base_type(type)) { - case GLSL_TYPE_STRUCT: - case GLSL_TYPE_ARRAY: - return glsl_get_length(type); - case GLSL_TYPE_FLOAT: - case GLSL_TYPE_INT: - case GLSL_TYPE_UINT: - case GLSL_TYPE_BOOL: - if (glsl_type_is_matrix(type)) - return glsl_get_matrix_columns(type); - else - return glsl_get_vector_elements(type); - default: - unreachable("Invalid deref base type"); - } -} - /* This function recursively walks the given deref chain and replaces the * given copy instruction with an equivalent sequence load/store * operations. @@ -121,9 +101,9 @@ emit_copy_load_store(nir_intrinsic_instr *copy_instr, nir_deref_array *src_arr = nir_deref_as_array(src_arr_parent->child); nir_deref_array *dest_arr = nir_deref_as_array(dest_arr_parent->child); - unsigned length = type_get_length(src_arr_parent->type); + unsigned length = glsl_get_length(src_arr_parent->type); /* The wildcards should represent the same number of elements */ - assert(length == type_get_length(dest_arr_parent->type)); + assert(length == glsl_get_length(dest_arr_parent->type)); assert(length > 0); /* Walk over all of the elements that this wildcard refers to and diff --git a/mesalib/src/glsl/nir/nir_lower_vars_to_ssa.c b/mesalib/src/glsl/nir/nir_lower_vars_to_ssa.c index 2ca74d71b..bb60f4601 100644 --- a/mesalib/src/glsl/nir/nir_lower_vars_to_ssa.c +++ b/mesalib/src/glsl/nir/nir_lower_vars_to_ssa.c @@ -55,7 +55,7 @@ struct deref_node { }; struct lower_variables_state { - void *mem_ctx; + nir_shader *shader; void *dead_ctx; nir_function_impl *impl; @@ -90,34 +90,14 @@ struct lower_variables_state { struct hash_table *phi_table; }; -static int -type_get_length(const struct glsl_type *type) -{ - switch (glsl_get_base_type(type)) { - case GLSL_TYPE_STRUCT: - case GLSL_TYPE_ARRAY: - return glsl_get_length(type); - case GLSL_TYPE_FLOAT: - case GLSL_TYPE_INT: - case GLSL_TYPE_UINT: - case GLSL_TYPE_BOOL: - if (glsl_type_is_matrix(type)) - return glsl_get_matrix_columns(type); - else - return glsl_get_vector_elements(type); - default: - unreachable("Invalid deref base type"); - } -} - static struct deref_node * deref_node_create(struct deref_node *parent, - const struct glsl_type *type, void *mem_ctx) + const struct glsl_type *type, nir_shader *shader) { size_t size = sizeof(struct deref_node) + - type_get_length(type) * sizeof(struct deref_node *); + glsl_get_length(type) * sizeof(struct deref_node *); - struct deref_node *node = rzalloc_size(mem_ctx, size); + struct deref_node *node = rzalloc_size(shader, size); node->type = type; node->parent = parent; node->deref = NULL; @@ -165,7 +145,7 @@ get_deref_node(nir_deref_var *deref, struct lower_variables_state *state) case nir_deref_type_struct: { nir_deref_struct *deref_struct = nir_deref_as_struct(tail); - assert(deref_struct->index < type_get_length(node->type)); + assert(deref_struct->index < glsl_get_length(node->type)); if (node->children[deref_struct->index] == NULL) node->children[deref_struct->index] = @@ -184,7 +164,7 @@ get_deref_node(nir_deref_var *deref, struct lower_variables_state *state) * out-of-bounds offset. We need to handle this at least * somewhat gracefully. */ - if (arr->base_offset >= type_get_length(node->type)) + if (arr->base_offset >= glsl_get_length(node->type)) return NULL; if (node->children[arr->base_offset] == NULL) @@ -317,6 +297,10 @@ deref_may_be_aliased_node(struct deref_node *node, nir_deref *deref, if (arr->deref_array_type == nir_deref_array_type_indirect) return true; + /* If there is an indirect at this level, we're aliased. */ + if (node->indirect) + return true; + assert(arr->deref_array_type == nir_deref_array_type_direct); if (node->children[arr->base_offset] && @@ -465,7 +449,7 @@ lower_copies_to_load_store(struct deref_node *node, set_foreach(node->copies, copy_entry) { nir_intrinsic_instr *copy = (void *)copy_entry->key; - nir_lower_var_copy_instr(copy, state->mem_ctx); + nir_lower_var_copy_instr(copy, state->shader); for (unsigned i = 0; i < 2; ++i) { struct deref_node *arg_node = @@ -485,67 +469,6 @@ lower_copies_to_load_store(struct deref_node *node, return true; } -/* 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. - */ -static nir_load_const_instr * -get_const_initializer_load(const nir_deref_var *deref, - struct lower_variables_state *state) -{ - nir_constant *constant = deref->var->constant_initializer; - const nir_deref *tail = &deref->deref; - unsigned matrix_offset = 0; - while (tail->child) { - switch (tail->child->deref_type) { - case nir_deref_type_array: { - nir_deref_array *arr = nir_deref_as_array(tail->child); - assert(arr->deref_array_type == nir_deref_array_type_direct); - if (glsl_type_is_matrix(tail->type)) { - assert(arr->deref.child == NULL); - matrix_offset = arr->base_offset; - } else { - constant = constant->elements[arr->base_offset]; - } - break; - } - - case nir_deref_type_struct: { - constant = constant->elements[nir_deref_as_struct(tail->child)->index]; - break; - } - - default: - unreachable("Invalid deref child type"); - } - - tail = tail->child; - } - - nir_load_const_instr *load = - nir_load_const_instr_create(state->mem_ctx, - glsl_get_vector_elements(tail->type)); - - matrix_offset *= load->def.num_components; - for (unsigned i = 0; i < load->def.num_components; i++) { - switch (glsl_get_base_type(tail->type)) { - case GLSL_TYPE_FLOAT: - case GLSL_TYPE_INT: - case GLSL_TYPE_UINT: - load->value.u[i] = constant->value.u[matrix_offset + i]; - break; - case GLSL_TYPE_BOOL: - load->value.u[i] = constant->value.b[matrix_offset + i] ? - NIR_TRUE : NIR_FALSE; - break; - default: - unreachable("Invalid immediate type"); - } - } - - return load; -} - /** Pushes an SSA def onto the def stack for the given node * * Each node is potentially associated with a stack of SSA definitions. @@ -614,7 +537,7 @@ get_ssa_def_for_block(struct deref_node *node, nir_block *block, * given block. This means that we need to add an undef and use that. */ nir_ssa_undef_instr *undef = - nir_ssa_undef_instr_create(state->mem_ctx, + nir_ssa_undef_instr_create(state->shader, glsl_get_vector_elements(node->type)); nir_instr_insert_before_cf_list(&state->impl->body, &undef->instr); def_stack_push(node, &undef->def, state); @@ -694,7 +617,7 @@ rename_variables_block(nir_block *block, struct lower_variables_state *state) * should result in an undefined value. */ nir_ssa_undef_instr *undef = - nir_ssa_undef_instr_create(state->mem_ctx, + nir_ssa_undef_instr_create(state->shader, intrin->num_components); nir_instr_insert_before(&intrin->instr, &undef->instr); @@ -702,14 +625,14 @@ rename_variables_block(nir_block *block, struct lower_variables_state *state) nir_ssa_def_rewrite_uses(&intrin->dest.ssa, nir_src_for_ssa(&undef->def), - state->mem_ctx); + state->shader); continue; } if (!node->lower_to_ssa) continue; - nir_alu_instr *mov = nir_alu_instr_create(state->mem_ctx, + nir_alu_instr *mov = nir_alu_instr_create(state->shader, nir_op_imov); mov->src[0].src.is_ssa = true; mov->src[0].src.ssa = get_ssa_def_for_block(node, block, state); @@ -727,7 +650,7 @@ rename_variables_block(nir_block *block, struct lower_variables_state *state) nir_ssa_def_rewrite_uses(&intrin->dest.ssa, nir_src_for_ssa(&mov->dest.dest.ssa), - state->mem_ctx); + state->shader); break; } @@ -750,7 +673,7 @@ rename_variables_block(nir_block *block, struct lower_variables_state *state) assert(intrin->src[0].is_ssa); - nir_alu_instr *mov = nir_alu_instr_create(state->mem_ctx, + nir_alu_instr *mov = nir_alu_instr_create(state->shader, nir_op_imov); mov->src[0].src.is_ssa = true; mov->src[0].src.ssa = intrin->src[0].ssa; @@ -887,7 +810,7 @@ insert_phi_nodes(struct lower_variables_state *state) continue; if (has_already[next->index] < iter_count) { - nir_phi_instr *phi = nir_phi_instr_create(state->mem_ctx); + nir_phi_instr *phi = nir_phi_instr_create(state->shader); nir_ssa_dest_init(&phi->instr, &phi->dest, glsl_get_vector_elements(node->type), NULL); nir_instr_insert_before_block(next, &phi->instr); @@ -938,8 +861,8 @@ nir_lower_vars_to_ssa_impl(nir_function_impl *impl) { struct lower_variables_state state; - state.mem_ctx = ralloc_parent(impl); - state.dead_ctx = ralloc_context(state.mem_ctx); + state.shader = impl->overload->function->shader; + state.dead_ctx = ralloc_context(state.shader); state.impl = impl; state.deref_var_nodes = _mesa_hash_table_create(state.dead_ctx, @@ -983,7 +906,8 @@ nir_lower_vars_to_ssa_impl(nir_function_impl *impl) progress = true; if (deref->var->constant_initializer) { - nir_load_const_instr *load = get_const_initializer_load(deref, &state); + nir_load_const_instr *load = + nir_deref_get_const_initializer_load(state.shader, deref); nir_ssa_def_init(&load->instr, &load->def, glsl_get_vector_elements(node->type), NULL); nir_instr_insert_before_cf_list(&impl->body, &load->instr); diff --git a/mesalib/src/glsl/nir/nir_opt_algebraic.py b/mesalib/src/glsl/nir/nir_opt_algebraic.py index cdb19241c..2a2b9561e 100644 --- a/mesalib/src/glsl/nir/nir_opt_algebraic.py +++ b/mesalib/src/glsl/nir/nir_opt_algebraic.py @@ -144,6 +144,7 @@ optimizations = [ (('fexp', ('fmul', ('flog', a), b)), ('fpow', a, b), '!options->lower_fpow'), # e^(ln(a)*b) = a^b (('fpow', a, 1.0), a), (('fpow', a, 2.0), ('fmul', a, a)), + (('fpow', a, 4.0), ('fmul', ('fmul', a, a), ('fmul', a, a))), (('fpow', 2.0, a), ('fexp2', a)), (('fsqrt', ('fexp2', a)), ('fexp2', ('fmul', 0.5, a))), (('fsqrt', ('fexp', a)), ('fexp', ('fmul', 0.5, a))), diff --git a/mesalib/src/glsl/nir/nir_print.c b/mesalib/src/glsl/nir/nir_print.c index fb8c9344c..eb4045cec 100644 --- a/mesalib/src/glsl/nir/nir_print.c +++ b/mesalib/src/glsl/nir/nir_print.c @@ -533,6 +533,8 @@ print_load_const_instr(nir_load_const_instr *instr, unsigned tabs, FILE *fp) fprintf(fp, "0x%08x /* %f */", instr->value.u[i], instr->value.f[i]); } + + fprintf(fp, ")"); } static void diff --git a/mesalib/src/glsl/nir/nir_types.cpp b/mesalib/src/glsl/nir/nir_types.cpp index f0d0b46d2..62176f508 100644 --- a/mesalib/src/glsl/nir/nir_types.cpp +++ b/mesalib/src/glsl/nir/nir_types.cpp @@ -103,7 +103,7 @@ glsl_get_matrix_columns(const struct glsl_type *type) unsigned glsl_get_length(const struct glsl_type *type) { - return type->length; + return type->is_matrix() ? type->matrix_columns : type->length; } const char * diff --git a/mesalib/src/glsl/opt_algebraic.cpp b/mesalib/src/glsl/opt_algebraic.cpp index 3d2f2ca0b..fa5db70f2 100644 --- a/mesalib/src/glsl/opt_algebraic.cpp +++ b/mesalib/src/glsl/opt_algebraic.cpp @@ -98,6 +98,12 @@ is_vec_two(ir_constant *ir) return (ir == NULL) ? false : ir->is_value(2.0, 2); } +static inline bool +is_vec_four(ir_constant *ir) +{ + return (ir == NULL) ? false : ir->is_value(4.0, 4); +} + static inline bool is_vec_negative_one(ir_constant *ir) { @@ -774,6 +780,20 @@ ir_algebraic_visitor::handle_expression(ir_expression *ir) return mul(x, x); } + if (is_vec_four(op_const[1])) { + ir_variable *x = new(ir) ir_variable(ir->operands[1]->type, "x", + ir_var_temporary); + base_ir->insert_before(x); + base_ir->insert_before(assign(x, ir->operands[0])); + + ir_variable *squared = new(ir) ir_variable(ir->operands[1]->type, + "squared", + ir_var_temporary); + base_ir->insert_before(squared); + base_ir->insert_before(assign(squared, mul(x, x))); + return mul(squared, squared); + } + break; case ir_binop_min: diff --git a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp index 50c8aa763..92f20c71d 100644 --- a/mesalib/src/glsl/opt_dead_builtin_varyings.cpp +++ b/mesalib/src/glsl/opt_dead_builtin_varyings.cpp @@ -35,7 +35,7 @@ * the built-in varyings have pre-assigned locations. Also, the elimination * of unused gl_TexCoord elements requires its own lowering pass anyway. * - * It's implemented by replacing all occurences of dead varyings with + * It's implemented by replacing all occurrences of dead varyings with * temporary variables, which creates dead code. It is recommended to run * a dead-code elimination pass after this. * @@ -280,7 +280,7 @@ public: * * We're going to break down the gl_TexCoord array into separate * variables. First, add declarations of the new variables all - * occurences of gl_TexCoord will be replaced with. + * occurrences of gl_TexCoord will be replaced with. */ if (info->lower_texcoord_array) { prepare_array(ir, this->new_texcoord, ARRAY_SIZE(this->new_texcoord), @@ -411,7 +411,7 @@ public: * variable dereference representing gl_TexCoord[i]. */ if (this->info->lower_texcoord_array) { - /* gl_TexCoord[i] occurence */ + /* gl_TexCoord[i] occurrence */ ir_dereference_array *const da = (*rvalue)->as_dereference_array(); if (da && da->variable_referenced() == @@ -425,7 +425,7 @@ public: /* Same for gl_FragData. */ if (this->info->lower_fragdata_array) { - /* gl_FragData[i] occurence */ + /* gl_FragData[i] occurrence */ ir_dereference_array *const da = (*rvalue)->as_dereference_array(); if (da && da->variable_referenced() == this->info->fragdata_array) { diff --git a/mesalib/src/glsl/opt_function_inlining.cpp b/mesalib/src/glsl/opt_function_inlining.cpp index 64b4907ba..84a9e4fa0 100644 --- a/mesalib/src/glsl/opt_function_inlining.cpp +++ b/mesalib/src/glsl/opt_function_inlining.cpp @@ -130,7 +130,7 @@ ir_call::generate_inline(ir_instruction *next_ir) parameters[i] = sig_param->clone(ctx, ht); parameters[i]->data.mode = ir_var_auto; - /* Remove the read-only decoration becuase we're going to write + /* Remove the read-only decoration because we're going to write * directly to this variable. If the cloned variable is left * read-only and the inlined function is inside a loop, the loop * analysis code will get confused. diff --git a/mesalib/src/glsl/standalone_scaffolding.cpp b/mesalib/src/glsl/standalone_scaffolding.cpp index 6f5a27f85..a109c4e92 100644 --- a/mesalib/src/glsl/standalone_scaffolding.cpp +++ b/mesalib/src/glsl/standalone_scaffolding.cpp @@ -189,9 +189,6 @@ void initialize_context_to_defaults(struct gl_context *ctx, gl_api api) options.MaxUnrollIterations = 32; options.MaxIfDepth = UINT_MAX; - /* Default pragma settings */ - options.DefaultPragmas.Optimize = true; - for (int sh = 0; sh < MESA_SHADER_STAGES; ++sh) memcpy(&ctx->Const.ShaderCompilerOptions[sh], &options, sizeof(options)); } diff --git a/mesalib/src/loader/Android.mk b/mesalib/src/loader/Android.mk index 8daeedf5d..8e215de3c 100644 --- a/mesalib/src/loader/Android.mk +++ b/mesalib/src/loader/Android.mk @@ -37,7 +37,7 @@ LOCAL_SRC_FILES := \ ifeq ($(MESA_GPU_DRIVERS),swrast) LOCAL_CFLAGS += -D__NOT_HAVE_DRM_H else -LOCAL_C_INCLUDES += $(TARGET_OUT_HEADERS)/libdrm +LOCAL_SHARED_LIBRARIES := libdrm endif LOCAL_MODULE := libmesa_loader diff --git a/mesalib/src/loader/loader.h b/mesalib/src/loader/loader.h index fa57950de..810e7da7f 100644 --- a/mesalib/src/loader/loader.h +++ b/mesalib/src/loader/loader.h @@ -43,7 +43,7 @@ loader_get_device_name_for_fd(int fd); /* Function to get a different device than the one we are to use by default, * if the user requests so and it is possible. The initial fd will be closed - * if neccessary. The returned fd is potentially a render-node. + * if necessary. The returned fd is potentially a render-node. */ int diff --git a/mesalib/src/mapi/Android.mk b/mesalib/src/mapi/Android.mk index f10437874..44452181e 100644 --- a/mesalib/src/mapi/Android.mk +++ b/mesalib/src/mapi/Android.mk @@ -53,7 +53,7 @@ LOCAL_C_INCLUDES := \ LOCAL_MODULE := libglapi LOCAL_MODULE_CLASS := SHARED_LIBRARIES -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) abi_header := $(intermediates)/$(abi_header) LOCAL_GENERATED_SOURCES := $(abi_header) diff --git a/mesalib/src/mapi/glapi/gen/glX_proto_send.py b/mesalib/src/mapi/glapi/gen/glX_proto_send.py index 1419a7cc1..b93989f75 100644 --- a/mesalib/src/mapi/glapi/gen/glX_proto_send.py +++ b/mesalib/src/mapi/glapi/gen/glX_proto_send.py @@ -574,7 +574,7 @@ generic_%u_byte( GLint rop, const void * ptr ) print ' struct glx_context * const gc = __glXGetCurrentContext();' # The only reason that single and vendor private commands need - # a variable called 'dpy' is becuase they use the SyncHandle + # a variable called 'dpy' is because they use the SyncHandle # macro. For whatever brain-dead reason, that macro is hard- # coded to use a variable called 'dpy' instead of taking a # parameter. diff --git a/mesalib/src/mapi/glapi/gen/glX_proto_size.py b/mesalib/src/mapi/glapi/gen/glX_proto_size.py index fdb355d4d..4737fbf71 100644 --- a/mesalib/src/mapi/glapi/gen/glX_proto_size.py +++ b/mesalib/src/mapi/glapi/gen/glX_proto_size.py @@ -57,7 +57,7 @@ class glx_enum_function(object): # Fill self.count and self.enums using the dictionary of enums # that was passed in. The generic Get functions (e.g., # GetBooleanv and friends) are handled specially here. In - # the data the generic Get functions are refered to as "Get". + # the data the generic Get functions are referred to as "Get". if func_name in ["GetIntegerv", "GetBooleanv", "GetFloatv", "GetDoublev"]: match_name = "Get" diff --git a/mesalib/src/mapi/glapi/gen/gl_x86-64_asm.py b/mesalib/src/mapi/glapi/gen/gl_x86-64_asm.py index 5a69e0343..7afc2b108 100644 --- a/mesalib/src/mapi/glapi/gen/gl_x86-64_asm.py +++ b/mesalib/src/mapi/glapi/gen/gl_x86-64_asm.py @@ -187,7 +187,7 @@ class PrintGenericStubs(gl_XML.gl_print_base): # The x86-64 ABI divides function parameters into a couple # classes. For the OpenGL interface, the only ones that are - # relevent are INTEGER and SSE. Basically, the first 8 + # relevant are INTEGER and SSE. Basically, the first 8 # GLfloat or GLdouble parameters are placed in %xmm0 - %xmm7, # the first 6 non-GLfloat / non-GLdouble parameters are placed # in registers listed in int_parameters. diff --git a/mesalib/src/mapi/glapi/glapi_entrypoint.c b/mesalib/src/mapi/glapi/glapi_entrypoint.c index 7facd8aca..f1f7f5e59 100644 --- a/mesalib/src/mapi/glapi/glapi_entrypoint.c +++ b/mesalib/src/mapi/glapi/glapi_entrypoint.c @@ -29,6 +29,8 @@ */ +#include + #include "c11/threads.h" #include "glapi/glapi_priv.h" #include "u_execmem.h" diff --git a/mesalib/src/mesa/Android.gen.mk b/mesalib/src/mesa/Android.gen.mk index caae2c107..cc979547e 100644 --- a/mesalib/src/mesa/Android.gen.mk +++ b/mesalib/src/mesa/Android.gen.mk @@ -27,13 +27,16 @@ ifeq ($(LOCAL_MODULE_CLASS),) LOCAL_MODULE_CLASS := STATIC_LIBRARIES endif -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) # This is the list of auto-generated files: sources and headers sources := \ main/enums.c \ main/api_exec.c \ main/dispatch.h \ + main/format_pack.c \ + main/format_unpack.c \ + main/format_info.h \ main/remap_helper.h \ main/get_hash.h @@ -124,3 +127,21 @@ format_info_deps := \ $(intermediates)/main/format_info.h: $(format_info_deps) @$(MESA_PYTHON2) $(FORMAT_INFO) $< > $@ + +FORMAT_PACK := $(LOCAL_PATH)/main/format_pack.py +format_pack_deps := \ + $(LOCAL_PATH)/main/formats.csv \ + $(LOCAL_PATH)/main/format_parser.py \ + $(FORMAT_PACK) + +$(intermediates)/main/format_pack.c: $(format_pack_deps) + $(hide) $(MESA_PYTHON2) $(FORMAT_PACK) $< > $@ + +FORMAT_UNPACK := $(LOCAL_PATH)/main/format_unpack.py +format_unpack_deps := \ + $(LOCAL_PATH)/main/formats.csv \ + $(LOCAL_PATH)/main/format_parser.py \ + $(FORMAT_UNPACK) + +$(intermediates)/main/format_unpack.c: $(format_unpack_deps) + $(hide) $(MESA_PYTHON2) $(FORMAT_UNPACK) $< > $@ diff --git a/mesalib/src/mesa/Android.libmesa_dricore.mk b/mesalib/src/mesa/Android.libmesa_dricore.mk index e4a52677c..2e308b837 100644 --- a/mesalib/src/mesa/Android.libmesa_dricore.mk +++ b/mesalib/src/mesa/Android.libmesa_dricore.mk @@ -49,21 +49,18 @@ endif # MESA_ENABLE_ASM ifeq ($(ARCH_X86_HAVE_SSE4_1),true) LOCAL_SRC_FILES += \ - $(SRCDIR)main/streaming-load-memcpy.c \ - $(SRCDIR)main/sse_minmax.c -LOCAL_CFLAGS := -msse4.1 -endif - -ifeq ($(ARCH_X86_HAVE_SSE4_1),true) -LOCAL_CFLAGS += \ + main/streaming-load-memcpy.c \ + mesa/main/sse_minmax.c +LOCAL_CFLAGS := \ + -msse4.1 \ -DUSE_SSE41 endif LOCAL_C_INCLUDES := \ - $(call intermediates-dir-for STATIC_LIBRARIES,libmesa_program,,) \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/mapi \ + $(MESA_TOP)/src/mesa/main \ $(MESA_TOP)/src/glsl \ + $(MESA_TOP)/src/gallium/include \ $(MESA_TOP)/src/gallium/auxiliary LOCAL_WHOLE_STATIC_LIBRARIES := \ diff --git a/mesalib/src/mesa/Android.libmesa_glsl_utils.mk b/mesalib/src/mesa/Android.libmesa_glsl_utils.mk index a9f6ff59a..3497377af 100644 --- a/mesalib/src/mesa/Android.libmesa_glsl_utils.mk +++ b/mesalib/src/mesa/Android.libmesa_glsl_utils.mk @@ -36,9 +36,10 @@ include $(CLEAR_VARS) LOCAL_MODULE := libmesa_glsl_utils LOCAL_C_INCLUDES := \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/glsl \ - $(MESA_TOP)/src/mapi + $(MESA_TOP)/src/mapi \ + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_SRC_FILES := \ main/imports.c \ @@ -59,9 +60,10 @@ LOCAL_IS_HOST_MODULE := true LOCAL_CFLAGS := -D_POSIX_C_SOURCE=199309L LOCAL_C_INCLUDES := \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/glsl \ - $(MESA_TOP)/src/mapi + $(MESA_TOP)/src/mapi \ + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_SRC_FILES := \ main/imports.c \ diff --git a/mesalib/src/mesa/Android.libmesa_st_mesa.mk b/mesalib/src/mesa/Android.libmesa_st_mesa.mk index a08366d53..b4b7fd977 100644 --- a/mesalib/src/mesa/Android.libmesa_st_mesa.mk +++ b/mesalib/src/mesa/Android.libmesa_st_mesa.mk @@ -52,12 +52,11 @@ LOCAL_CFLAGS := \ endif LOCAL_C_INCLUDES := \ - $(call intermediates-dir-for STATIC_LIBRARIES,libmesa_program,,) \ - $(MESA_TOP)/src/gallium/auxiliary \ - $(MESA_TOP)/src/gallium/include \ - $(MESA_TOP)/src \ + $(MESA_TOP)/src/mapi \ + $(MESA_TOP)/src/mesa/main \ $(MESA_TOP)/src/glsl \ - $(MESA_TOP)/src/mapi + $(MESA_TOP)/src/gallium/auxiliary \ + $(MESA_TOP)/src/gallium/include LOCAL_WHOLE_STATIC_LIBRARIES := \ libmesa_program diff --git a/mesalib/src/mesa/Android.mesa_gen_matypes.mk b/mesalib/src/mesa/Android.mesa_gen_matypes.mk index 5521087c4..6e301f904 100644 --- a/mesalib/src/mesa/Android.mesa_gen_matypes.mk +++ b/mesalib/src/mesa/Android.mesa_gen_matypes.mk @@ -33,7 +33,6 @@ include $(CLEAR_VARS) LOCAL_MODULE := mesa_gen_matypes LOCAL_IS_HOST_MODULE := true -LOCAL_CFLAGS := -D_POSIX_C_SOURCE=199309L LOCAL_C_INCLUDES := \ $(MESA_TOP)/src/mapi \ diff --git a/mesalib/src/mesa/drivers/common/meta.c b/mesalib/src/mesa/drivers/common/meta.c index cf99d9563..d2ab7b8de 100644 --- a/mesalib/src/mesa/drivers/common/meta.c +++ b/mesalib/src/mesa/drivers/common/meta.c @@ -41,7 +41,6 @@ #include "main/bufferobj.h" #include "main/buffers.h" #include "main/clear.h" -#include "main/colortab.h" #include "main/condrender.h" #include "main/depth.h" #include "main/enable.h" diff --git a/mesalib/src/mesa/drivers/dri/Android.mk b/mesalib/src/mesa/drivers/dri/Android.mk index 935722c53..42bfd77d4 100644 --- a/mesalib/src/mesa/drivers/dri/Android.mk +++ b/mesalib/src/mesa/drivers/dri/Android.mk @@ -35,17 +35,17 @@ MESA_DRI_CFLAGS := \ -DHAVE_ANDROID_PLATFORM MESA_DRI_C_INCLUDES := \ - $(MESA_TOP)/src \ - $(call intermediates-dir-for,STATIC_LIBRARIES,libmesa_dri_common) \ $(addprefix $(MESA_TOP)/, $(mesa_dri_common_INCLUDES)) \ - $(TARGET_OUT_HEADERS)/libdrm \ + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary \ external/expat/lib MESA_DRI_WHOLE_STATIC_LIBRARIES := \ libmesa_glsl \ libmesa_megadriver_stub \ libmesa_dri_common \ - libmesa_dricore + libmesa_dricore \ + libmesa_util MESA_DRI_SHARED_LIBRARIES := \ libcutils \ @@ -55,9 +55,6 @@ MESA_DRI_SHARED_LIBRARIES := \ libglapi \ liblog -# All DRI modules must add this to LOCAL_GENERATED_SOURCES. -MESA_DRI_OPTIONS_H := $(call intermediates-dir-for,STATIC_LIBRARIES,libmesa_dri_common)/xmlpool/options.h - #----------------------------------------------- # Build drivers and libmesa_dri_common diff --git a/mesalib/src/mesa/drivers/dri/common/Android.mk b/mesalib/src/mesa/drivers/dri/common/Android.mk index b95feb679..a7fcd6d57 100644 --- a/mesalib/src/mesa/drivers/dri/common/Android.mk +++ b/mesalib/src/mesa/drivers/dri/common/Android.mk @@ -34,16 +34,24 @@ include $(LOCAL_PATH)/Makefile.sources LOCAL_MODULE := libmesa_dri_common LOCAL_MODULE_CLASS := STATIC_LIBRARIES -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) LOCAL_C_INCLUDES := \ - $(intermediates) \ $(MESA_DRI_C_INCLUDES) +LOCAL_EXPORT_C_INCLUDE_DIRS := $(intermediates) + +# swrast only +ifeq ($(MESA_GPU_DRIVERS),swrast) +LOCAL_CFLAGS := -D__NOT_HAVE_DRM_H +else +LOCAL_SHARED_LIBRARIES := libdrm +endif + LOCAL_SRC_FILES := $(DRI_COMMON_FILES) -LOCAL_GENERATED_SOURCES := \ - $(intermediates)/xmlpool/options.h +MESA_DRI_OPTIONS_H := $(intermediates)/xmlpool/options.h +LOCAL_GENERATED_SOURCES := $(MESA_DRI_OPTIONS_H) # # Generate options.h from gettext translations. @@ -73,15 +81,14 @@ $(intermediates)/xmlpool/%/LC_MESSAGES/options.mo: $(intermediates)/xmlpool/%.po mkdir -p $(dir $@) msgfmt -o $@ $< -$(intermediates)/xmlpool/options.h: PRIVATE_SCRIPT := $(LOCAL_PATH)/xmlpool/gen_xmlpool.py -$(intermediates)/xmlpool/options.h: PRIVATE_LOCALEDIR := $(intermediates)/xmlpool -$(intermediates)/xmlpool/options.h: PRIVATE_TEMPLATE_HEADER := $(LOCAL_PATH)/xmlpool/t_options.h -$(intermediates)/xmlpool/options.h: PRIVATE_MO_FILES := $(MESA_DRI_OPTIONS_LANGS:%=$(intermediates)/xmlpool/%/LC_MESSAGES/options.mo) +$(MESA_DRI_OPTIONS_H): PRIVATE_SCRIPT := $(LOCAL_PATH)/xmlpool/gen_xmlpool.py +$(MESA_DRI_OPTIONS_H): PRIVATE_LOCALEDIR := $(intermediates)/xmlpool +$(MESA_DRI_OPTIONS_H): PRIVATE_TEMPLATE_HEADER := $(LOCAL_PATH)/xmlpool/t_options.h +$(MESA_DRI_OPTIONS_H): PRIVATE_MO_FILES := $(MESA_DRI_OPTIONS_LANGS:%=$(intermediates)/xmlpool/%/LC_MESSAGES/options.mo) .SECONDEXPANSION: -$(intermediates)/xmlpool/options.h: $$(PRIVATE_SCRIPT) $$(PRIVATE_TEMPLATE_HEADER) $$(PRIVATE_MO_FILES) - mkdir -p $(dir $@) - mkdir -p $(PRIVATE_LOCALEDIR) - $(MESA_PYTHON2) $(PRIVATE_SCRIPT) $(PRIVATE_TEMPLATE_HEADER) \ +$(MESA_DRI_OPTIONS_H): $$(PRIVATE_SCRIPT) $$(PRIVATE_TEMPLATE_HEADER) $$(PRIVATE_MO_FILES) + @mkdir -p $(dir $@) + $(hide) $(MESA_PYTHON2) $(PRIVATE_SCRIPT) $(PRIVATE_TEMPLATE_HEADER) \ $(PRIVATE_LOCALEDIR) $(MESA_DRI_OPTIONS_LANGS) > $@ include $(MESA_COMMON_MK) @@ -99,6 +106,13 @@ LOCAL_MODULE_CLASS := STATIC_LIBRARIES LOCAL_C_INCLUDES := \ $(MESA_DRI_C_INCLUDES) +# swrast only +ifeq ($(MESA_GPU_DRIVERS),swrast) +LOCAL_CFLAGS := -D__NOT_HAVE_DRM_H +else +LOCAL_SHARED_LIBRARIES := libdrm +endif + LOCAL_SRC_FILES := $(megadriver_stub_FILES) include $(MESA_COMMON_MK) diff --git a/mesalib/src/mesa/drivers/dri/common/drirc b/mesalib/src/mesa/drivers/dri/common/drirc index cecd6a953..145e707a6 100644 --- a/mesalib/src/mesa/drivers/dri/common/drirc +++ b/mesalib/src/mesa/drivers/dri/common/drirc @@ -91,5 +91,9 @@ TODO: document the other workarounds. + + + diff --git a/mesalib/src/mesa/main/compiler.h b/mesalib/src/mesa/main/compiler.h index 55152fdef..93b4e6f12 100644 --- a/mesalib/src/mesa/main/compiler.h +++ b/mesalib/src/mesa/main/compiler.h @@ -60,11 +60,6 @@ extern "C" { #endif -/* XXX: Use standard `__func__` instead */ -#ifndef __FUNCTION__ -# define __FUNCTION__ __func__ -#endif - /** * Either define MESA_BIG_ENDIAN or MESA_LITTLE_ENDIAN, and CPU_TO_LE32. * Do not use these unless absolutely necessary! diff --git a/mesalib/src/mesa/main/context.c b/mesalib/src/mesa/main/context.c index adf64976e..4aaf8b1af 100644 --- a/mesalib/src/mesa/main/context.c +++ b/mesalib/src/mesa/main/context.c @@ -894,14 +894,14 @@ update_default_objects(struct gl_context *ctx) * If there's no current OpenGL context for the calling thread, we can * print a message to stderr. * - * \param name the name of the OpenGL function, without the "gl" prefix + * \param name the name of the OpenGL function */ static void nop_handler(const char *name) { GET_CURRENT_CONTEXT(ctx); if (ctx) { - _mesa_error(ctx, GL_INVALID_OPERATION, "gl%s(invalid call)", name); + _mesa_error(ctx, GL_INVALID_OPERATION, "%s(invalid call)", name); } #if defined(DEBUG) else if (getenv("MESA_DEBUG") || getenv("LIBGL_DEBUG")) { @@ -936,7 +936,7 @@ alloc_dispatch_table(void) { /* Find the larger of Mesa's dispatch table and libGL's dispatch table. * In practice, this'll be the same for stand-alone Mesa. But for DRI - * Mesa we do this to accomodate different versions of libGL and various + * Mesa we do this to accommodate different versions of libGL and various * DRI drivers. */ GLint numEntries = MAX2(_glapi_get_dispatch_table_size(), _gloffset_COUNT); diff --git a/mesalib/src/mesa/main/extensions.c b/mesalib/src/mesa/main/extensions.c index 861b15006..9be8993b0 100644 --- a/mesalib/src/mesa/main/extensions.c +++ b/mesalib/src/mesa/main/extensions.c @@ -186,6 +186,7 @@ static const struct extension extension_table[] = { { "GL_ARB_texture_rectangle", o(NV_texture_rectangle), GL, 2004 }, { "GL_ARB_texture_rgb10_a2ui", o(ARB_texture_rgb10_a2ui), GL, 2009 }, { "GL_ARB_texture_rg", o(ARB_texture_rg), GL, 2008 }, + { "GL_ARB_texture_stencil8", o(ARB_texture_stencil8), GL, 2013 }, { "GL_ARB_texture_storage", o(dummy_true), GL, 2011 }, { "GL_ARB_texture_storage_multisample", o(ARB_texture_multisample), GL, 2012 }, { "GL_ARB_texture_view", o(ARB_texture_view), GL, 2012 }, @@ -297,6 +298,7 @@ static const struct extension extension_table[] = { { "GL_OES_depth_texture", o(ARB_depth_texture), ES2, 2006 }, { "GL_OES_depth_texture_cube_map", o(OES_depth_texture_cube_map), ES2, 2012 }, { "GL_OES_draw_texture", o(OES_draw_texture), ES1, 2004 }, + { "GL_OES_EGL_sync", o(dummy_true), ES1 | ES2, 2010 }, /* FIXME: Mesa expects GL_OES_EGL_image to be available in OpenGL contexts. */ { "GL_OES_EGL_image", o(OES_EGL_image), GL | ES1 | ES2, 2006 }, { "GL_OES_EGL_image_external", o(OES_EGL_image_external), ES1 | ES2, 2010 }, diff --git a/mesalib/src/mesa/main/fbobject.c b/mesalib/src/mesa/main/fbobject.c index 8032585ab..27cf97f17 100644 --- a/mesalib/src/mesa/main/fbobject.c +++ b/mesalib/src/mesa/main/fbobject.c @@ -813,8 +813,10 @@ test_attachment_completeness(const struct gl_context *ctx, GLenum format, if (ctx->Extensions.ARB_depth_texture && baseFormat == GL_DEPTH_STENCIL) { /* OK */ - } - else { + } else if (ctx->Extensions.ARB_texture_stencil8 && + baseFormat == GL_STENCIL_INDEX) { + /* OK */ + } else { /* no such thing as stencil-only textures */ att_incomplete("illegal stencil texture"); att->Complete = GL_FALSE; @@ -978,7 +980,8 @@ _mesa_test_framebuffer_completeness(struct gl_context *ctx, if (!is_format_color_renderable(ctx, attFormat, texImg->InternalFormat) && - !is_legal_depth_format(ctx, f)) { + !is_legal_depth_format(ctx, f) && + f != GL_STENCIL_INDEX) { fb->_Status = GL_FRAMEBUFFER_INCOMPLETE_ATTACHMENT; fbo_incomplete(ctx, "texture attachment incomplete", -1); return; diff --git a/mesalib/src/mesa/main/ffvertex_prog.c b/mesalib/src/mesa/main/ffvertex_prog.c index edf7e3390..7fdd9bab6 100644 --- a/mesalib/src/mesa/main/ffvertex_prog.c +++ b/mesalib/src/mesa/main/ffvertex_prog.c @@ -1657,7 +1657,7 @@ _mesa_get_fixed_func_vertex_program(struct gl_context *ctx) struct gl_vertex_program *prog; struct state_key key; - /* Grab all the relevent state and put it in a single structure: + /* Grab all the relevant state and put it in a single structure: */ make_state_key(ctx, &key); diff --git a/mesalib/src/mesa/main/formats.c b/mesalib/src/mesa/main/formats.c index 2bc8bcad9..8af44e905 100644 --- a/mesalib/src/mesa/main/formats.c +++ b/mesalib/src/mesa/main/formats.c @@ -811,7 +811,7 @@ _mesa_format_image_size(mesa_format format, GLsizei width, /** * Same as _mesa_format_image_size() but returns a 64-bit value to - * accomodate very large textures. + * accommodate very large textures. */ uint64_t _mesa_format_image_size64(mesa_format format, GLsizei width, diff --git a/mesalib/src/mesa/main/macros.h b/mesalib/src/mesa/main/macros.h index 3344ec841..0608650ae 100644 --- a/mesalib/src/mesa/main/macros.h +++ b/mesalib/src/mesa/main/macros.h @@ -31,6 +31,7 @@ #ifndef MACROS_H #define MACROS_H +#include "util/macros.h" #include "util/u_math.h" #include "imports.h" @@ -800,10 +801,6 @@ DIFFERENT_SIGNS(GLfloat x, GLfloat y) } -/** Compute ceiling of integer quotient of A divided by B. */ -#define DIV_ROUND_UP( A, B ) ( (A) % (B) == 0 ? (A)/(B) : (A)/(B)+1 ) - - /** casts to silence warnings with some compilers */ #define ENUM_TO_INT(E) ((GLint)(E)) #define ENUM_TO_FLOAT(E) ((GLfloat)(GLint)(E)) diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h index 1c751cfff..fb4143086 100644 --- a/mesalib/src/mesa/main/mtypes.h +++ b/mesalib/src/mesa/main/mtypes.h @@ -2365,16 +2365,6 @@ struct gl_ati_fragment_shader_state }; -/** Set by #pragma directives */ -struct gl_sl_pragmas -{ - GLboolean IgnoreOptimize; /**< ignore #pragma optimize(on/off) ? */ - GLboolean IgnoreDebug; /**< ignore #pragma debug(on/off) ? */ - GLboolean Optimize; /**< defaults on */ - GLboolean Debug; /**< defaults off */ -}; - - /** * A GLSL vertex or fragment shader object. */ @@ -2397,7 +2387,6 @@ struct gl_shader struct gl_program *Program; /**< Post-compile assembly code */ GLchar *InfoLog; - struct gl_sl_pragmas Pragmas; unsigned Version; /**< GLSL version used for linking */ @@ -2903,8 +2892,6 @@ struct gl_shader_compiler_options */ GLboolean OptimizeForAOS; - struct gl_sl_pragmas DefaultPragmas; /**< Default #pragma settings */ - const struct nir_shader_compiler_options *NirOptions; }; diff --git a/mesalib/src/mesa/main/shader_query.cpp b/mesalib/src/mesa/main/shader_query.cpp index b5f1d082c..d2ca49b43 100644 --- a/mesalib/src/mesa/main/shader_query.cpp +++ b/mesalib/src/mesa/main/shader_query.cpp @@ -291,7 +291,6 @@ _mesa_GetAttribLocation(GLhandleARB program, const GLcharARB * name) return (loc >= 0) ? loc : -1; } - unsigned _mesa_count_active_attribs(struct gl_shader_program *shProg) { @@ -300,19 +299,15 @@ _mesa_count_active_attribs(struct gl_shader_program *shProg) return 0; } - exec_list *const ir = shProg->_LinkedShaders[MESA_SHADER_VERTEX]->ir; - unsigned i = 0; - - foreach_in_list(ir_instruction, node, ir) { - const ir_variable *const var = node->as_variable(); - - if (!is_active_attrib(var)) - continue; - - i++; + struct gl_program_resource *res = shProg->ProgramResourceList; + unsigned count = 0; + for (unsigned j = 0; j < shProg->NumProgramResourceList; j++, res++) { + if (res->Type == GL_PROGRAM_INPUT && + res->StageReferences & (1 << MESA_SHADER_VERTEX) && + is_active_attrib(RESOURCE_VAR(res))) + count++; } - - return i; + return count; } @@ -324,20 +319,16 @@ _mesa_longest_attribute_name_length(struct gl_shader_program *shProg) return 0; } - exec_list *const ir = shProg->_LinkedShaders[MESA_SHADER_VERTEX]->ir; + struct gl_program_resource *res = shProg->ProgramResourceList; size_t longest = 0; + for (unsigned j = 0; j < shProg->NumProgramResourceList; j++, res++) { + if (res->Type == GL_PROGRAM_INPUT && + res->StageReferences & (1 << MESA_SHADER_VERTEX)) { - foreach_in_list(ir_instruction, node, ir) { - const ir_variable *const var = node->as_variable(); - - if (var == NULL - || var->data.mode != ir_var_shader_in - || var->data.location == -1) - continue; - - const size_t len = strlen(var->name); - if (len >= longest) - longest = len + 1; + const size_t length = strlen(RESOURCE_VAR(res)->name); + if (length >= longest) + longest = length + 1; + } } return longest; @@ -537,6 +528,7 @@ array_index_of_resource(struct gl_program_resource *res, return get_matching_index(RESOURCE_VAR(res), name); default: assert(!"support for resource type not implemented"); + return -1; } } @@ -634,7 +626,7 @@ _mesa_program_resource_find_index(struct gl_shader_program *shProg, case GL_ATOMIC_COUNTER_BUFFER: if (_mesa_program_resource_index(shProg, res) == index) return res; - + break; case GL_TRANSFORM_FEEDBACK_VARYING: case GL_PROGRAM_INPUT: case GL_PROGRAM_OUTPUT: @@ -860,13 +852,23 @@ get_buffer_property(struct gl_shader_program *shProg, *val = RESOURCE_UBO(res)->UniformBufferSize; return 1; case GL_NUM_ACTIVE_VARIABLES: - *val = RESOURCE_UBO(res)->NumUniforms; + *val = 0; + for (unsigned i = 0; i < RESOURCE_UBO(res)->NumUniforms; i++) { + const char *iname = RESOURCE_UBO(res)->Uniforms[i].IndexName; + struct gl_program_resource *uni = + _mesa_program_resource_find_name(shProg, GL_UNIFORM, iname); + if (!uni) + continue; + (*val)++; + } return 1; case GL_ACTIVE_VARIABLES: for (unsigned i = 0; i < RESOURCE_UBO(res)->NumUniforms; i++) { const char *iname = RESOURCE_UBO(res)->Uniforms[i].IndexName; struct gl_program_resource *uni = _mesa_program_resource_find_name(shProg, GL_UNIFORM, iname); + if (!uni) + continue; *val++ = _mesa_program_resource_index(shProg, uni); } diff --git a/mesalib/src/mesa/main/shaderapi.c b/mesalib/src/mesa/main/shaderapi.c index 77e2b8745..a04b28711 100644 --- a/mesalib/src/mesa/main/shaderapi.c +++ b/mesalib/src/mesa/main/shaderapi.c @@ -115,9 +115,6 @@ _mesa_init_shader_state(struct gl_context *ctx) options.MaxUnrollIterations = 32; options.MaxIfDepth = UINT_MAX; - /* Default pragma settings */ - options.DefaultPragmas.Optimize = GL_TRUE; - for (sh = 0; sh < MESA_SHADER_STAGES; ++sh) memcpy(&ctx->Const.ShaderCompilerOptions[sh], &options, sizeof(options)); @@ -864,17 +861,11 @@ static void compile_shader(struct gl_context *ctx, GLuint shaderObj) { struct gl_shader *sh; - struct gl_shader_compiler_options *options; sh = _mesa_lookup_shader_err(ctx, shaderObj, "glCompileShader"); if (!sh) return; - options = &ctx->Const.ShaderCompilerOptions[sh->Stage]; - - /* set default pragma state for shader */ - sh->Pragmas = options->DefaultPragmas; - if (!sh->Source) { /* If the user called glCompileShader without first calling * glShaderSource, we should fail to compile, but not raise a GL_ERROR. diff --git a/mesalib/src/mesa/main/texgetimage.c b/mesalib/src/mesa/main/texgetimage.c index 255d36559..92b4d6795 100644 --- a/mesalib/src/mesa/main/texgetimage.c +++ b/mesalib/src/mesa/main/texgetimage.c @@ -175,6 +175,51 @@ get_tex_depth_stencil(struct gl_context *ctx, GLuint dimensions, } } +/** + * glGetTexImage for stencil pixels. + */ +static void +get_tex_stencil(struct gl_context *ctx, GLuint dimensions, + GLenum format, GLenum type, GLvoid *pixels, + struct gl_texture_image *texImage) +{ + const GLint width = texImage->Width; + const GLint height = texImage->Height; + const GLint depth = texImage->Depth; + GLint img, row; + + assert(format == GL_STENCIL_INDEX); + + for (img = 0; img < depth; img++) { + GLubyte *srcMap; + GLint rowstride; + + /* map src texture buffer */ + ctx->Driver.MapTextureImage(ctx, texImage, img, + 0, 0, width, height, GL_MAP_READ_BIT, + &srcMap, &rowstride); + + if (srcMap) { + for (row = 0; row < height; row++) { + const GLubyte *src = srcMap + row * rowstride; + void *dest = _mesa_image_address(dimensions, &ctx->Pack, pixels, + width, height, format, type, + img, row, 0); + _mesa_unpack_ubyte_stencil_row(texImage->TexFormat, + width, + (const GLuint *) src, + dest); + } + + ctx->Driver.UnmapTextureImage(ctx, texImage, img); + } + else { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glGetTexImage"); + break; + } + } +} + /** * glGetTexImage for YCbCr pixels. @@ -285,7 +330,7 @@ get_tex_rgba_compressed(struct gl_context *ctx, GLuint dimensions, } /* Depending on the base format involved we may need to apply a rebase - * tranaform (for example: if we download to a Luminance format we want + * transform (for example: if we download to a Luminance format we want * G=0 and B=0). */ if (baseFormat == GL_LUMINANCE || @@ -388,7 +433,7 @@ get_tex_rgba_uncompressed(struct gl_context *ctx, GLuint dimensions, } /* Depending on the base format involved we may need to apply a rebase - * tranaform (for example: if we download to a Luminance format we want + * transform (for example: if we download to a Luminance format we want * G=0 and B=0). */ if (texImage->_BaseFormat == GL_LUMINANCE || @@ -684,6 +729,9 @@ _mesa_GetTexImage_sw(struct gl_context *ctx, else if (format == GL_DEPTH_STENCIL_EXT) { get_tex_depth_stencil(ctx, dimensions, format, type, pixels, texImage); } + else if (format == GL_STENCIL_INDEX) { + get_tex_stencil(ctx, dimensions, format, type, pixels, texImage); + } else if (format == GL_YCBCR_MESA) { get_tex_ycbcr(ctx, dimensions, format, type, pixels, texImage); } @@ -879,7 +927,7 @@ getteximage_error_check(struct gl_context *ctx, "glGetTex%sImage(format mismatch)", suffix); return GL_TRUE; } - else if (_mesa_is_enum_format_integer(format) != + else if (!_mesa_is_stencil_format(format) && _mesa_is_enum_format_integer(format) != _mesa_is_format_integer(texImage->TexFormat)) { _mesa_error(ctx, GL_INVALID_OPERATION, "glGetTex%sImage(format mismatch)", suffix); diff --git a/mesalib/src/mesa/main/teximage.c b/mesalib/src/mesa/main/teximage.c index 8d9d7cfc1..7bc1da7f8 100644 --- a/mesalib/src/mesa/main/teximage.c +++ b/mesalib/src/mesa/main/teximage.c @@ -1997,7 +1997,8 @@ _mesa_legal_texture_base_format_for_target(struct gl_context *ctx, const char *caller) { if (_mesa_base_tex_format(ctx, internalFormat) == GL_DEPTH_COMPONENT - || _mesa_base_tex_format(ctx, internalFormat) == GL_DEPTH_STENCIL) { + || _mesa_base_tex_format(ctx, internalFormat) == GL_DEPTH_STENCIL + || _mesa_base_tex_format(ctx, internalFormat) == GL_STENCIL_INDEX) { /* Section 3.8.3 (Texture Image Specification) of the OpenGL 3.3 Core * Profile spec says: * @@ -3319,7 +3320,9 @@ teximage(struct gl_context *ctx, GLboolean compressed, GLuint dims, if (!sizeOK) { _mesa_error(ctx, GL_OUT_OF_MEMORY, - "glTexImage%uD(image too large)", dims); + "glTexImage%uD(image too large: %d x %d x %d, %s format)", + dims, width, height, depth, + _mesa_lookup_enum_by_nr(internalFormat)); return; } diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c index e018ab910..c563f1e74 100644 --- a/mesalib/src/mesa/main/texobj.c +++ b/mesalib/src/mesa/main/texobj.c @@ -30,7 +30,6 @@ #include #include "bufferobj.h" -#include "colortab.h" #include "context.h" #include "enums.h" #include "fbobject.h" diff --git a/mesalib/src/mesa/main/texturebarrier.c b/mesalib/src/mesa/main/texturebarrier.c index 08ff56150..d879eed57 100644 --- a/mesalib/src/mesa/main/texturebarrier.c +++ b/mesalib/src/mesa/main/texturebarrier.c @@ -49,5 +49,11 @@ _mesa_TextureBarrierNV(void) { GET_CURRENT_CONTEXT(ctx); + if (!ctx->Extensions.NV_texture_barrier) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glTextureBarrier(not supported)"); + return; + } + ctx->Driver.TextureBarrier(ctx); } diff --git a/mesalib/src/mesa/main/uniform_query.cpp b/mesalib/src/mesa/main/uniform_query.cpp index 4e77b3284..3e857ed10 100644 --- a/mesalib/src/mesa/main/uniform_query.cpp +++ b/mesalib/src/mesa/main/uniform_query.cpp @@ -129,14 +129,26 @@ _mesa_GetActiveUniformsiv(GLuint program, res_prop = resource_prop_from_uniform_prop(pname); + /* We need to first verify that each entry exists as active uniform. If + * not, generate error and do not cause any other side effects. + * + * In the case of and error condition, Page 16 (section 2.3.1 Errors) + * of the OpenGL 4.5 spec says: + * + * "If the generating command modifies values through a pointer argu- + * ment, no change is made to these values." + */ for (int i = 0; i < uniformCount; i++) { - res = _mesa_program_resource_find_index(shProg, GL_UNIFORM, - uniformIndices[i]); - if (!res) { + if (!_mesa_program_resource_find_index(shProg, GL_UNIFORM, + uniformIndices[i])) { _mesa_error(ctx, GL_INVALID_VALUE, "glGetActiveUniformsiv(index)"); - break; + return; } + } + for (int i = 0; i < uniformCount; i++) { + res = _mesa_program_resource_find_index(shProg, GL_UNIFORM, + uniformIndices[i]); if (!_mesa_program_resource_prop(shProg, res, uniformIndices[i], res_prop, ¶ms[i], "glGetActiveUniformsiv")) diff --git a/mesalib/src/mesa/main/version.c b/mesalib/src/mesa/main/version.c index 7c6d99418..5b8ac0a5c 100644 --- a/mesalib/src/mesa/main/version.c +++ b/mesalib/src/mesa/main/version.c @@ -295,7 +295,51 @@ compute_version(const struct gl_extensions *extensions, extensions->EXT_texture_swizzle); /* ARB_sampler_objects is always enabled in mesa */ - if (ver_3_3) { + const GLboolean ver_4_0 = (ver_3_3 && + consts->GLSLVersion >= 400 && + extensions->ARB_draw_buffers_blend && + extensions->ARB_draw_indirect && + extensions->ARB_gpu_shader5 && + extensions->ARB_gpu_shader_fp64 && + extensions->ARB_sample_shading && + 0/*extensions->ARB_shader_subroutine*/ && + extensions->ARB_tessellation_shader && + extensions->ARB_texture_buffer_object_rgb32 && + extensions->ARB_texture_cube_map_array && + extensions->ARB_texture_query_lod && + extensions->ARB_transform_feedback2 && + extensions->ARB_transform_feedback3); + const GLboolean ver_4_1 = (ver_4_0 && + consts->GLSLVersion >= 410 && + extensions->ARB_ES2_compatibility && + extensions->ARB_shader_precision && + 0/*extensions->ARB_vertex_attrib_64bit*/ && + extensions->ARB_viewport_array); + const GLboolean ver_4_2 = (ver_4_1 && + consts->GLSLVersion >= 420 && + extensions->ARB_base_instance && + extensions->ARB_conservative_depth && + extensions->ARB_internalformat_query && + extensions->ARB_shader_atomic_counters && + extensions->ARB_shader_image_load_store && + extensions->ARB_shading_language_420pack && + extensions->ARB_shading_language_packing && + extensions->ARB_texture_compression_bptc && + extensions->ARB_transform_feedback_instanced); + + if (ver_4_2) { + major = 4; + minor = 2; + } + else if (ver_4_1) { + major = 4; + minor = 1; + } + else if (ver_4_0) { + major = 4; + minor = 0; + } + else if (ver_3_3) { major = 3; minor = 3; } @@ -439,6 +483,23 @@ _mesa_compute_version(struct gl_context *ctx) ctx->Version = _mesa_get_version(&ctx->Extensions, &ctx->Const, ctx->API); + /* Make sure that the GLSL version lines up with the GL version. In some + * cases it can be too high, e.g. if an extension is missing. + */ + if (ctx->API == API_OPENGL_CORE) { + switch (ctx->Version) { + case 31: + ctx->Const.GLSLVersion = 140; + break; + case 32: + ctx->Const.GLSLVersion = 150; + break; + default: + ctx->Const.GLSLVersion = ctx->Version * 10; + break; + } + } + switch (ctx->API) { case API_OPENGL_COMPAT: case API_OPENGL_CORE: diff --git a/mesalib/src/mesa/math/m_matrix.c b/mesalib/src/mesa/math/m_matrix.c index 0475a7ab2..ecf564c00 100644 --- a/mesalib/src/mesa/math/m_matrix.c +++ b/mesalib/src/mesa/math/m_matrix.c @@ -50,7 +50,7 @@ /*@{*/ #define MAT_FLAG_IDENTITY 0 /**< is an identity matrix flag. * (Not actually used - the identity - * matrix is identified by the absense + * matrix is identified by the absence * of all other flags.) */ #define MAT_FLAG_GENERAL 0x1 /**< is a general matrix flag */ diff --git a/mesalib/src/mesa/program/Android.mk b/mesalib/src/mesa/program/Android.mk index a237b65bc..ccb0fa5f3 100644 --- a/mesalib/src/mesa/program/Android.mk +++ b/mesalib/src/mesa/program/Android.mk @@ -44,8 +44,9 @@ include $(CLEAR_VARS) LOCAL_MODULE := libmesa_program LOCAL_MODULE_CLASS := STATIC_LIBRARIES +LOCAL_STATIC_LIBRARIES := libmesa_glsl -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) # TODO(chadv): In Makefile.sources, move these vars to a different list so we can # remove this kludge. @@ -55,7 +56,8 @@ generated_sources_basenames := \ program_parse.tab.h LOCAL_SRC_FILES := \ - $(filter-out $(generated_sources_basenames),$(subst program/,,$(PROGRAM_FILES))) + $(filter-out $(generated_sources_basenames),$(subst program/,,$(PROGRAM_FILES))) \ + $(subst program/,,$(PROGRAM_NIR_FILES)) LOCAL_GENERATED_SOURCES := \ $(addprefix $(intermediates)/program/,$(generated_sources_basenames)) @@ -70,11 +72,13 @@ $(intermediates)/program/lex.yy.c: $(LOCAL_PATH)/program_lexer.l $(local-l-to-c) LOCAL_C_INCLUDES := \ - $(intermediates) \ - $(MESA_TOP)/src \ $(MESA_TOP)/src/mapi \ $(MESA_TOP)/src/mesa \ - $(MESA_TOP)/src/glsl + $(MESA_TOP)/src/glsl \ + $(MESA_TOP)/src/gallium/auxiliary \ + $(MESA_TOP)/src/gallium/include + +LOCAL_EXPORT_C_INCLUDE_DIRS := $(intermediates) include $(MESA_COMMON_MK) include $(BUILD_STATIC_LIBRARY) diff --git a/mesalib/src/mesa/program/hash_table.h b/mesalib/src/mesa/program/hash_table.h index eed2e55dc..e85a836a0 100644 --- a/mesalib/src/mesa/program/hash_table.h +++ b/mesalib/src/mesa/program/hash_table.h @@ -116,8 +116,8 @@ extern void hash_table_insert(struct hash_table *ht, void *data, * Add an element to a hash table with replacement * * \return - * 1 if it did replace the the value (in which case the old key is kept), 0 if - * it did not replace the value (in which case the new key is kept). + * 1 if it did replace the value (in which case the old key is kept), 0 if it + * did not replace the value (in which case the new key is kept). * * \warning * If \c key is already in the hash table, \c data will \b replace the most diff --git a/mesalib/src/mesa/program/prog_cache.c b/mesalib/src/mesa/program/prog_cache.c index 34609f056..ed93af7f1 100644 --- a/mesalib/src/mesa/program/prog_cache.c +++ b/mesalib/src/mesa/program/prog_cache.c @@ -77,7 +77,7 @@ hash_key(const void *key, GLuint key_size) /** - * Rebuild/expand the hash table to accomodate more entries + * Rebuild/expand the hash table to accommodate more entries */ static void rehash(struct gl_program_cache *cache) diff --git a/mesalib/src/mesa/program/prog_to_nir.c b/mesalib/src/mesa/program/prog_to_nir.c index c738f5073..ff3d9f3be 100644 --- a/mesalib/src/mesa/program/prog_to_nir.c +++ b/mesalib/src/mesa/program/prog_to_nir.c @@ -222,12 +222,23 @@ ptn_get_src(struct ptn_compile *c, const struct prog_src_register *prog_src) } nir_ssa_def *def; - if (!HAS_EXTENDED_SWIZZLE(prog_src->Swizzle)) { + if (!HAS_EXTENDED_SWIZZLE(prog_src->Swizzle) && + (prog_src->Negate == NEGATE_NONE || prog_src->Negate == NEGATE_XYZW)) { + /* The simple non-SWZ case. */ for (int i = 0; i < 4; i++) src.swizzle[i] = GET_SWZ(prog_src->Swizzle, i); def = nir_fmov_alu(b, src, 4); + + if (prog_src->Abs) + def = nir_fabs(b, def); + + if (prog_src->Negate) + def = nir_fneg(b, def); } else { + /* The SWZ instruction allows per-component zero/one swizzles, and also + * per-component negation. + */ nir_ssa_def *chans[4]; for (int i = 0; i < 4; i++) { int swizzle = GET_SWZ(prog_src->Swizzle, i); @@ -246,16 +257,16 @@ ptn_get_src(struct ptn_compile *c, const struct prog_src_register *prog_src) chans[i] = &mov->dest.dest.ssa; } + + if (prog_src->Abs) + chans[i] = nir_fabs(b, chans[i]); + + if (prog_src->Negate & (1 << i)) + chans[i] = nir_fneg(b, chans[i]); } def = nir_vec4(b, chans[0], chans[1], chans[2], chans[3]); } - if (prog_src->Abs) - def = nir_fabs(b, def); - - if (prog_src->Negate) - def = nir_fneg(b, def); - return def; } diff --git a/mesalib/src/mesa/state_tracker/st_atom_texture.c b/mesalib/src/mesa/state_tracker/st_atom_texture.c index eff28fc6f..04ba86448 100644 --- a/mesalib/src/mesa/state_tracker/st_atom_texture.c +++ b/mesalib/src/mesa/state_tracker/st_atom_texture.c @@ -287,16 +287,22 @@ st_get_texture_sampler_view_from_stobj(struct st_context *st, enum pipe_format format) { struct pipe_sampler_view **sv; - + const struct st_texture_image *firstImage; if (!stObj || !stObj->pt) { return NULL; } sv = st_texture_get_sampler_view(st, stObj); - if (stObj->base.StencilSampling && - util_format_is_depth_and_stencil(format)) - format = util_format_stencil_only(format); + if (util_format_is_depth_and_stencil(format)) { + if (stObj->base.StencilSampling) + format = util_format_stencil_only(format); + else { + firstImage = st_texture_image_const(_mesa_base_tex_image(&stObj->base)); + if (firstImage->base._BaseFormat == GL_STENCIL_INDEX) + format = util_format_stencil_only(format); + } + } /* if sampler view has changed dereference it */ if (*sv) { diff --git a/mesalib/src/mesa/state_tracker/st_cb_flush.c b/mesalib/src/mesa/state_tracker/st_cb_flush.c index 6df430049..ca51eeee3 100644 --- a/mesalib/src/mesa/state_tracker/st_cb_flush.c +++ b/mesalib/src/mesa/state_tracker/st_cb_flush.c @@ -152,7 +152,7 @@ void st_init_flush_functions(struct dd_function_table *functions) * Finish() is identical to Flush() in all cases - no differences in * rendering or ReadPixels are visible if we opt not to wait here. * - * Only set this up on windows to avoid suprise elsewhere. + * Only set this up on Windows to avoid surprise elsewhere. */ #ifdef PIPE_OS_WINDOWS functions->Finish = st_glFlush; diff --git a/mesalib/src/mesa/state_tracker/st_cb_texture.c b/mesalib/src/mesa/state_tracker/st_cb_texture.c index bdf236e82..7ea3846ff 100644 --- a/mesalib/src/mesa/state_tracker/st_cb_texture.c +++ b/mesalib/src/mesa/state_tracker/st_cb_texture.c @@ -950,7 +950,7 @@ st_GetTexImage(struct gl_context * ctx, /* XXX Fallback to _mesa_GetTexImage_sw for depth-stencil formats * due to an incomplete stencil blit implementation in some drivers. */ - if (format == GL_DEPTH_STENCIL) { + if (format == GL_DEPTH_STENCIL || format == GL_STENCIL_INDEX) { goto fallback; } diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c index bc20f7323..82e4a3009 100644 --- a/mesalib/src/mesa/state_tracker/st_extensions.c +++ b/mesalib/src/mesa/state_tracker/st_extensions.c @@ -650,12 +650,15 @@ void st_init_extensions(struct pipe_screen *screen, ARRAY_SIZE(vertex_mapping), PIPE_BUFFER, PIPE_BIND_VERTEX_BUFFER); + if (extensions->ARB_stencil_texturing) + extensions->ARB_texture_stencil8 = GL_TRUE; + /* Figure out GLSL support. */ glsl_feature_level = screen->get_param(screen, PIPE_CAP_GLSL_FEATURE_LEVEL); consts->GLSLVersion = glsl_feature_level; - if (glsl_feature_level >= 330) - consts->GLSLVersion = 330; + if (glsl_feature_level >= 410) + consts->GLSLVersion = 410; _mesa_override_glsl_version(consts); @@ -666,6 +669,8 @@ void st_init_extensions(struct pipe_screen *screen, if (glsl_feature_level >= 400) extensions->ARB_gpu_shader5 = GL_TRUE; + if (glsl_feature_level >= 410) + extensions->ARB_shader_precision = GL_TRUE; /* This extension needs full OpenGL 3.2, but we don't know if that's * supported at this point. Only check the GLSL version. */ diff --git a/mesalib/src/mesa/state_tracker/st_format.c b/mesalib/src/mesa/state_tracker/st_format.c index 72dbf3bd4..181465dd8 100644 --- a/mesalib/src/mesa/state_tracker/st_format.c +++ b/mesalib/src/mesa/state_tracker/st_format.c @@ -1942,11 +1942,6 @@ st_ChooseTextureFormat(struct gl_context *ctx, GLenum target, GLint internalFormat, GLenum format, GLenum type) { - const boolean want_renderable = - internalFormat == 3 || internalFormat == 4 || - internalFormat == GL_RGB || internalFormat == GL_RGBA || - internalFormat == GL_RGB8 || internalFormat == GL_RGBA8 || - internalFormat == GL_BGRA; struct st_context *st = st_context(ctx); enum pipe_format pFormat; unsigned bindings; @@ -1962,15 +1957,17 @@ st_ChooseTextureFormat(struct gl_context *ctx, GLenum target, } /* GL textures may wind up being render targets, but we don't know - * that in advance. Specify potential render target flags now. + * that in advance. Specify potential render target flags now for formats + * that we know should always be renderable. */ bindings = PIPE_BIND_SAMPLER_VIEW; - if (want_renderable) { - if (_mesa_is_depth_or_stencil_format(internalFormat)) - bindings |= PIPE_BIND_DEPTH_STENCIL; - else + if (_mesa_is_depth_or_stencil_format(internalFormat)) + bindings |= PIPE_BIND_DEPTH_STENCIL; + else if (internalFormat == 3 || internalFormat == 4 || + internalFormat == GL_RGB || internalFormat == GL_RGBA || + internalFormat == GL_RGB8 || internalFormat == GL_RGBA8 || + internalFormat == GL_BGRA) bindings |= PIPE_BIND_RENDER_TARGET; - } /* GLES allows the driver to choose any format which matches * the format+type combo, because GLES only supports unsized internal diff --git a/mesalib/src/mesa/swrast/s_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c index 0adb8e5ba..453bd3636 100644 --- a/mesalib/src/mesa/swrast/s_texcombine.c +++ b/mesalib/src/mesa/swrast/s_texcombine.c @@ -645,7 +645,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span ) /* * Must do all texture sampling before combining in order to - * accomodate GL_ARB_texture_env_crossbar. + * accommodate GL_ARB_texture_env_crossbar. */ for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) { const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; diff --git a/mesalib/src/mesa/tnl/t_vertex.h b/mesalib/src/mesa/tnl/t_vertex.h index 36d8a969e..a84b299e5 100644 --- a/mesalib/src/mesa/tnl/t_vertex.h +++ b/mesalib/src/mesa/tnl/t_vertex.h @@ -40,7 +40,7 @@ struct tnl_clipspace; /* It will probably be necessary to allow drivers to specify new - * emit-styles to cover all the wierd and wacky things out there. + * emit-styles to cover all the weird and wacky things out there. */ enum tnl_attr_format { EMIT_1F, diff --git a/mesalib/src/mesa/vbo/vbo_exec_api.c b/mesalib/src/mesa/vbo/vbo_exec_api.c index 859078f12..48680ec88 100644 --- a/mesalib/src/mesa/vbo/vbo_exec_api.c +++ b/mesalib/src/mesa/vbo/vbo_exec_api.c @@ -685,7 +685,7 @@ static void GLAPIENTRY vbo_exec_Begin( GLenum mode ) return; } - /* Heuristic: attempt to isolate attributes occuring outside + /* Heuristic: attempt to isolate attributes occurring outside * begin/end pairs. */ if (exec->vtx.vertex_size && !exec->vtx.attrsz[0]) diff --git a/mesalib/src/mesa/vbo/vbo_save_draw.c b/mesalib/src/mesa/vbo/vbo_save_draw.c index 74e266d19..de744e0c7 100644 --- a/mesalib/src/mesa/vbo/vbo_save_draw.c +++ b/mesalib/src/mesa/vbo/vbo_save_draw.c @@ -278,7 +278,7 @@ vbo_save_playback_vertex_list(struct gl_context *ctx, void *data) goto end; } else if (save->replay_flags) { - /* Various degnerate cases: translate into immediate mode + /* Various degenerate cases: translate into immediate mode * calls rather than trying to execute in place. */ vbo_save_loopback_vertex_list( ctx, node ); diff --git a/mesalib/src/util/Android.mk b/mesalib/src/util/Android.mk index 6b38cf9af..a39185ae4 100644 --- a/mesalib/src/util/Android.mk +++ b/mesalib/src/util/Android.mk @@ -36,27 +36,22 @@ LOCAL_SRC_FILES := \ LOCAL_C_INCLUDES := \ $(MESA_TOP)/src/mesa \ $(MESA_TOP)/src/mapi \ - $(MESA_TOP)/src + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_MODULE := libmesa_util # Generated sources -ifeq ($(LOCAL_MODULE_CLASS),) LOCAL_MODULE_CLASS := STATIC_LIBRARIES -endif -intermediates := $(call local-intermediates-dir) +intermediates := $(call local-generated-sources-dir) +LOCAL_GENERATED_SOURCES := $(addprefix $(intermediates)/,$(MESA_UTIL_GENERATED_FILES)) -# This is the list of auto-generated files: sources and headers -sources := $(addprefix $(intermediates)/, $(MESA_UTIL_GENERATED_FILES)) - -LOCAL_GENERATED_SOURCES += $(sources) - -FORMAT_SRGB := $(LOCAL_PATH)/format_srgb.py - -$(intermediates)/format_srgb.c: $(FORMAT_SRGB) - @$(MESA_PYTHON2) $(FORMAT_SRGB) $< > $@ +$(LOCAL_GENERATED_SOURCES): PRIVATE_PYTHON := $(MESA_PYTHON2) +$(LOCAL_GENERATED_SOURCES): PRIVATE_CUSTOM_TOOL = $(PRIVATE_PYTHON) $^ > $@ +$(LOCAL_GENERATED_SOURCES): $(intermediates)/%.c: $(LOCAL_PATH)/%.py + $(transform-generated-source) include $(MESA_COMMON_MK) include $(BUILD_STATIC_LIBRARY) @@ -76,27 +71,22 @@ LOCAL_SRC_FILES := \ LOCAL_C_INCLUDES := \ $(MESA_TOP)/src/mesa \ $(MESA_TOP)/src/mapi \ - $(MESA_TOP)/src + $(MESA_TOP)/src/gallium/include \ + $(MESA_TOP)/src/gallium/auxiliary LOCAL_MODULE := libmesa_util # Generated sources -ifeq ($(LOCAL_MODULE_CLASS),) LOCAL_MODULE_CLASS := STATIC_LIBRARIES -endif - -intermediates := $(call local-intermediates-dir) - -# This is the list of auto-generated files: sources and headers -sources := $(addprefix $(intermediates)/, $(MESA_UTIL_GENERATED_FILES)) - -LOCAL_GENERATED_SOURCES += $(sources) -FORMAT_SRGB := $(LOCAL_PATH)/format_srgb.py +intermediates := $(call local-generated-sources-dir) +LOCAL_GENERATED_SOURCES := $(addprefix $(intermediates)/,$(MESA_UTIL_GENERATED_FILES)) -$(intermediates)/format_srgb.c: $(FORMAT_SRGB) - @$(MESA_PYTHON2) $(FORMAT_SRGB) $< > $@ +$(LOCAL_GENERATED_SOURCES): PRIVATE_PYTHON := $(MESA_PYTHON2) +$(LOCAL_GENERATED_SOURCES): PRIVATE_CUSTOM_TOOL = $(PRIVATE_PYTHON) $^ > $@ +$(LOCAL_GENERATED_SOURCES): $(intermediates)/%.c: $(LOCAL_PATH)/%.py + $(transform-generated-source) include $(MESA_COMMON_MK) include $(BUILD_HOST_STATIC_LIBRARY) diff --git a/mesalib/src/util/macros.h b/mesalib/src/util/macros.h index 6c7bda7ae..3b708ed6a 100644 --- a/mesalib/src/util/macros.h +++ b/mesalib/src/util/macros.h @@ -182,5 +182,7 @@ do { \ #define UNUSED #endif +/** Compute ceiling of integer quotient of A divided by B. */ +#define DIV_ROUND_UP( A, B ) ( (A) % (B) == 0 ? (A)/(B) : (A)/(B)+1 ) #endif /* UTIL_MACROS_H */ diff --git a/pixman/.gitignore b/pixman/.gitignore index 0f114966c..7da6b6f79 100644 --- a/pixman/.gitignore +++ b/pixman/.gitignore @@ -46,6 +46,7 @@ demos/tri-test pixman/pixman-srgb.c pixman/pixman-version.h test/a1-trap-test +test/affine-bench test/affine-test test/alpha-loop test/alphamap diff --git a/pixman/test/Makefile.sources b/pixman/test/Makefile.sources index c20c34b17..8b0e855e2 100644 --- a/pixman/test/Makefile.sources +++ b/pixman/test/Makefile.sources @@ -37,6 +37,7 @@ OTHERPROGRAMS = \ radial-perf-test \ check-formats \ scaling-bench \ + affine-bench \ $(NULL) # Utility functions diff --git a/pixman/test/affine-bench.c b/pixman/test/affine-bench.c new file mode 100644 index 000000000..9e0121e77 --- /dev/null +++ b/pixman/test/affine-bench.c @@ -0,0 +1,436 @@ +/* + * Copyright © 2014 RISC OS Open Ltd + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that + * the above copyright notice appear in all copies and that both that + * copyright notice and this permission notice appear in supporting + * documentation, and that the name of the copyright holders not be used in + * advertising or publicity pertaining to distribution of the software without + * specific, written prior permission. The copyright holders make no + * representations about the suitability of this software for any purpose. It + * is provided "as is" without express or implied warranty. + * + * THE COPYRIGHT HOLDERS DISCLAIM ALL WARRANTIES WITH REGARD TO THIS + * SOFTWARE, INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND + * FITNESS, IN NO EVENT SHALL THE COPYRIGHT HOLDERS BE LIABLE FOR ANY + * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN + * AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + * + * Author: Ben Avison (bavison@riscosopen.org) + */ + +#include +#include +#include +#include +#include +#include "utils.h" + +#ifdef HAVE_GETTIMEOFDAY +#include +#else +#include +#endif + +#define WIDTH 1920 +#define HEIGHT 1080 + +/* How much data to read to flush all cached data to RAM */ +#define MAX_L2CACHE_SIZE (8 * 1024 * 1024) + +#define PAGE_SIZE (4 * 1024) + +struct bench_info +{ + pixman_op_t op; + pixman_transform_t transform; + pixman_image_t *src_image; + pixman_image_t *mask_image; + pixman_image_t *dest_image; + int32_t src_x; + int32_t src_y; +}; + +typedef struct bench_info bench_info_t; + +struct box_48_16 +{ + pixman_fixed_48_16_t x1; + pixman_fixed_48_16_t y1; + pixman_fixed_48_16_t x2; + pixman_fixed_48_16_t y2; +}; + +typedef struct box_48_16 box_48_16_t; + +/* This function is copied verbatim from pixman.c. */ +static pixman_bool_t +compute_transformed_extents (pixman_transform_t *transform, + const pixman_box32_t *extents, + box_48_16_t *transformed) +{ + pixman_fixed_48_16_t tx1, ty1, tx2, ty2; + pixman_fixed_t x1, y1, x2, y2; + int i; + + x1 = pixman_int_to_fixed (extents->x1) + pixman_fixed_1 / 2; + y1 = pixman_int_to_fixed (extents->y1) + pixman_fixed_1 / 2; + x2 = pixman_int_to_fixed (extents->x2) - pixman_fixed_1 / 2; + y2 = pixman_int_to_fixed (extents->y2) - pixman_fixed_1 / 2; + + if (!transform) + { + transformed->x1 = x1; + transformed->y1 = y1; + transformed->x2 = x2; + transformed->y2 = y2; + + return TRUE; + } + + tx1 = ty1 = INT64_MAX; + tx2 = ty2 = INT64_MIN; + + for (i = 0; i < 4; ++i) + { + pixman_fixed_48_16_t tx, ty; + pixman_vector_t v; + + v.vector[0] = (i & 0x01)? x1 : x2; + v.vector[1] = (i & 0x02)? y1 : y2; + v.vector[2] = pixman_fixed_1; + + if (!pixman_transform_point (transform, &v)) + return FALSE; + + tx = (pixman_fixed_48_16_t)v.vector[0]; + ty = (pixman_fixed_48_16_t)v.vector[1]; + + if (tx < tx1) + tx1 = tx; + if (ty < ty1) + ty1 = ty; + if (tx > tx2) + tx2 = tx; + if (ty > ty2) + ty2 = ty; + } + + transformed->x1 = tx1; + transformed->y1 = ty1; + transformed->x2 = tx2; + transformed->y2 = ty2; + + return TRUE; +} + +static void +create_image (uint32_t width, + uint32_t height, + pixman_format_code_t format, + pixman_filter_t filter, + uint32_t **bits, + pixman_image_t **image) +{ + uint32_t stride = (width * PIXMAN_FORMAT_BPP (format) + 31) / 32 * 4; + + *bits = aligned_malloc (PAGE_SIZE, stride * height); + memset (*bits, 0xCC, stride * height); + *image = pixman_image_create_bits (format, width, height, *bits, stride); + pixman_image_set_repeat (*image, PIXMAN_REPEAT_NORMAL); + pixman_image_set_filter (*image, filter, NULL, 0); +} + +/* This needs to match the shortest cacheline length we expect to encounter */ +#define CACHE_CLEAN_INCREMENT 32 + +static void +flush_cache (void) +{ + static const char clean_space[MAX_L2CACHE_SIZE]; + volatile const char *x = clean_space; + const char *clean_end = clean_space + sizeof clean_space; + + while (x < clean_end) + { + (void) *x; + x += CACHE_CLEAN_INCREMENT; + } +} + +/* Obtain current time in microseconds modulo 2^32 */ +uint32_t +gettimei (void) +{ +#ifdef HAVE_GETTIMEOFDAY + struct timeval tv; + + gettimeofday (&tv, NULL); + return tv.tv_sec * 1000000 + tv.tv_usec; +#else + return (uint64_t) clock () * 1000000 / CLOCKS_PER_SEC; +#endif +} + +static void +pixman_image_composite_wrapper (const pixman_composite_info_t *info) +{ + pixman_image_composite (info->op, + info->src_image, info->mask_image, info->dest_image, + info->src_x, info->src_y, + info->mask_x, info->mask_y, + info->dest_x, info->dest_y, + info->width, info->height); +} + +static void +pixman_image_composite_empty (const pixman_composite_info_t *info) +{ + pixman_image_composite (info->op, + info->src_image, info->mask_image, info->dest_image, + info->src_x, info->src_y, + info->mask_x, info->mask_y, + info->dest_x, info->dest_y, + 1, 1); +} + +static void +bench (const bench_info_t *bi, + uint32_t max_n, + uint32_t max_time, + uint32_t *ret_n, + uint32_t *ret_time, + void (*func) (const pixman_composite_info_t *info)) +{ + uint32_t n = 0; + uint32_t t0; + uint32_t t1; + uint32_t x = 0; + pixman_transform_t t; + pixman_composite_info_t info; + + t = bi->transform; + info.op = bi->op; + info.src_image = bi->src_image; + info.mask_image = bi->mask_image; + info.dest_image = bi->dest_image; + info.src_x = 0; + info.src_y = 0; + info.mask_x = 0; + info.mask_y = 0; + /* info.dest_x set below */ + info.dest_y = 0; + info.width = WIDTH; + info.height = HEIGHT; + + t0 = gettimei (); + + do + { + + if (++x >= 64) + x = 0; + + info.dest_x = 63 - x; + + t.matrix[0][2] = pixman_int_to_fixed (bi->src_x + x); + t.matrix[1][2] = pixman_int_to_fixed (bi->src_y); + pixman_image_set_transform (bi->src_image, &t); + + if (bi->mask_image) + pixman_image_set_transform (bi->mask_image, &t); + + func (&info); + t1 = gettimei (); + } + while (++n < max_n && (t1 - t0) < max_time); + + if (ret_n) + *ret_n = n; + + *ret_time = t1 - t0; +} + +int +parse_fixed_argument (char *arg, pixman_fixed_t *value) +{ + char *tailptr; + + *value = pixman_double_to_fixed (strtod (arg, &tailptr)); + + return *tailptr == '\0'; +} + +int +parse_arguments (int argc, + char *argv[], + pixman_transform_t *t, + pixman_op_t *op, + pixman_format_code_t *src_format, + pixman_format_code_t *mask_format, + pixman_format_code_t *dest_format) +{ + if (!parse_fixed_argument (*argv, &t->matrix[0][0])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[0][1])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[1][0])) + return 0; + + if (*++argv == NULL) + return 1; + + if (!parse_fixed_argument (*argv, &t->matrix[1][1])) + return 0; + + if (*++argv == NULL) + return 1; + + *op = operator_from_string (*argv); + if (*op == PIXMAN_OP_NONE) + return 0; + + if (*++argv == NULL) + return 1; + + *src_format = format_from_string (*argv); + if (*src_format == PIXMAN_null) + return 0; + + ++argv; + if (argv[0] && argv[1]) + { + *mask_format = format_from_string (*argv); + if (*mask_format == PIXMAN_null) + return 0; + ++argv; + } + if (*argv) + { + *dest_format = format_from_string (*argv); + if (*dest_format == PIXMAN_null) + return 0; + } + return 1; +} + +static void +run_benchmark (const bench_info_t *bi) +{ + uint32_t n; /* number of iterations in at least 5 seconds */ + uint32_t t1; /* time taken to do n iterations, microseconds */ + uint32_t t2; /* calling overhead for n iterations, microseconds */ + + flush_cache (); + bench (bi, UINT32_MAX, 5000000, &n, &t1, pixman_image_composite_wrapper); + bench (bi, n, UINT32_MAX, NULL, &t2, pixman_image_composite_empty); + + /* The result indicates the output rate in megapixels/second */ + printf ("%6.2f\n", (double) n * WIDTH * HEIGHT / (t1 - t2)); +} + + +int +main (int argc, char *argv[]) +{ + bench_info_t binfo; + pixman_filter_t filter = PIXMAN_FILTER_NEAREST; + pixman_format_code_t src_format = PIXMAN_a8r8g8b8; + pixman_format_code_t mask_format = 0; + pixman_format_code_t dest_format = PIXMAN_a8r8g8b8; + pixman_box32_t dest_box = { 0, 0, WIDTH, HEIGHT }; + box_48_16_t transformed = { 0 }; + int32_t xmin, ymin, xmax, ymax; + uint32_t *src, *mask, *dest; + + binfo.op = PIXMAN_OP_SRC; + binfo.mask_image = NULL; + pixman_transform_init_identity (&binfo.transform); + + ++argv; + if (*argv && (*argv)[0] == '-' && (*argv)[1] == 'n') + { + filter = PIXMAN_FILTER_NEAREST; + ++argv; + --argc; + } + + if (*argv && (*argv)[0] == '-' && (*argv)[1] == 'b') + { + filter = PIXMAN_FILTER_BILINEAR; + ++argv; + --argc; + } + + if (argc == 1 || + !parse_arguments (argc, argv, &binfo.transform, &binfo.op, + &src_format, &mask_format, &dest_format)) + { + printf ("Usage: affine-bench [-n] [-b] axx [axy] [ayx] [ayy] [combine type]\n"); + printf (" [src format] [mask format] [dest format]\n"); + printf (" -n : nearest scaling (default)\n"); + printf (" -b : bilinear scaling\n"); + printf (" axx : x_out:x_in factor\n"); + printf (" axy : x_out:y_in factor (default 0)\n"); + printf (" ayx : y_out:x_in factor (default 0)\n"); + printf (" ayy : y_out:y_in factor (default 1)\n"); + printf (" combine type : src, over, in etc (default src)\n"); + printf (" src format : a8r8g8b8, r5g6b5 etc (default a8r8g8b8)\n"); + printf (" mask format : as for src format, but no mask used if omitted\n"); + printf (" dest format : as for src format (default a8r8g8b8)\n"); + printf ("The output is a single number in megapixels/second.\n"); + + return EXIT_FAILURE; + } + + compute_transformed_extents (&binfo.transform, &dest_box, &transformed); + /* The source area is expanded by a tiny bit (8/65536th pixel) + * to match the calculation of the COVER_CLIP flags in analyze_extent() + */ + xmin = pixman_fixed_to_int (transformed.x1 - 8 * pixman_fixed_e - pixman_fixed_1 / 2); + ymin = pixman_fixed_to_int (transformed.y1 - 8 * pixman_fixed_e - pixman_fixed_1 / 2); + xmax = pixman_fixed_to_int (transformed.x2 + 8 * pixman_fixed_e + pixman_fixed_1 / 2); + ymax = pixman_fixed_to_int (transformed.y2 + 8 * pixman_fixed_e + pixman_fixed_1 / 2); + binfo.src_x = -xmin; + binfo.src_y = -ymin; + + /* Always over-allocate width by 64 pixels for all src, mask and dst, + * so that we can iterate over an x-offset 0..63 in bench (). + * This is similar to lowlevel-blt-bench, which uses the same method + * to hit different cacheline misalignments. + */ + create_image (xmax - xmin + 64, ymax - ymin + 1, src_format, filter, + &src, &binfo.src_image); + + if (mask_format) + { + create_image (xmax - xmin + 64, ymax - ymin + 1, mask_format, filter, + &mask, &binfo.mask_image); + + if ((PIXMAN_FORMAT_R(mask_format) || + PIXMAN_FORMAT_G(mask_format) || + PIXMAN_FORMAT_B(mask_format))) + { + pixman_image_set_component_alpha (binfo.mask_image, 1); + } + } + + create_image (WIDTH + 64, HEIGHT, dest_format, filter, + &dest, &binfo.dest_image); + + run_benchmark (&binfo); + + return EXIT_SUCCESS; +} -- cgit v1.2.3