2019-01-29 23:32:16 +00:00
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
< html lang = "en" >
< head >
< meta http-equiv = "content-type" content = "text/html; charset=utf-8" >
< title > Mesa Release Notes< / title >
< link rel = "stylesheet" type = "text/css" href = "../mesa.css" >
< / head >
< body >
< div class = "header" >
2019-05-06 12:26:47 +01:00
The Mesa 3D Graphics Library
2019-01-29 23:32:16 +00:00
< / div >
< iframe src = "../contents.html" > < / iframe >
< div class = "content" >
2019-06-11 16:07:39 +01:00
< h1 > Mesa 19.1.0 Release Notes / June 11, 2019< / h1 >
2019-01-29 23:32:16 +00:00
< p >
Mesa 19.1.0 is a new development release. People who are concerned
with stability and reliability should stick with a previous release or
wait for Mesa 19.1.1.
< / p >
< p >
Mesa 19.1.0 implements the OpenGL 4.5 API, but the version reported by
glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
Some drivers don't support all the features required in OpenGL 4.5. OpenGL
4.5 is < strong > only< / strong > available if requested at context creation.
Compatibility contexts may report a lower version depending on each driver.
< / p >
< h2 > SHA256 checksums< / h2 >
< pre >
2019-06-11 16:25:40 +01:00
2a6c3af3a803389183168e449c536304cf03e0f82c4c9333077933543b9d02f3 mesa-19.1.0.tar.xz
2019-01-29 23:32:16 +00:00
< / pre >
< h2 > New features< / h2 >
< ul >
2019-04-02 15:47:37 +01:00
< li > GL_ARB_parallel_shader_compile on all drivers.< / li >
2018-08-07 23:32:31 +01:00
< li > GL_EXT_gpu_shader4 on all GL 3.1 drivers.< / li >
2019-04-15 18:03:13 +01:00
< li > GL_EXT_shader_image_load_formatted on radeonsi.< / li >
2018-08-07 23:32:31 +01:00
< li > GL_EXT_texture_buffer_object on all GL 3.1 drivers.< / li >
2019-01-31 16:47:47 +00:00
< li > GL_EXT_texture_compression_s3tc_srgb on Gallium drivers and i965 (ES extension).< / li >
2019-04-07 06:34:05 +01:00
< li > GL_NV_compute_shader_derivatives on iris and i965.< / li >
2019-04-02 15:47:37 +01:00
< li > GL_KHR_parallel_shader_compile on all drivers.< / li >
2019-01-24 01:06:27 +00:00
< li > VK_EXT_buffer_device_address on Intel and RADV.< / li >
2019-04-20 15:44:57 +01:00
< li > VK_EXT_depth_clip_enable on Intel and RADV.< / li >
< li > VK_KHR_ycbcr_image_arrays on Intel.< / li >
< li > VK_EXT_inline_uniform_block on Intel and RADV.< / li >
< li > VK_EXT_external_memory_host on Intel.< / li >
< li > VK_EXT_host_query_reset on Intel and RADV.< / li >
< li > VK_KHR_surface_protected_capabilities on Intel and RADV.< / li >
< li > VK_EXT_pipeline_creation_feedback on Intel and RADV.< / li >
< li > VK_KHR_8bit_storage on RADV.< / li >
< li > VK_AMD_gpu_shader_int16 on RADV.< / li >
< li > VK_AMD_gpu_shader_half_float on RADV.< / li >
2019-04-07 06:34:05 +01:00
< li > VK_NV_compute_shader_derivatives on Intel.< / li >
2019-04-20 15:44:57 +01:00
< li > VK_KHR_shader_float16_int8 on Intel and RADV (RADV only supports int8).< / li >
< li > VK_KHR_shader_atomic_int64 on Intel.< / li >
2019-02-27 22:08:20 +00:00
< li > VK_EXT_descriptor_indexing on Intel.< / li >
2019-04-18 20:00:37 +01:00
< li > VK_KHR_shader_float16_int8 on Intel and RADV.< / li >
2019-04-18 18:21:56 +01:00
< li > GL_INTEL_conservative_rasterization on iris.< / li >
2019-01-08 12:45:38 +00:00
< li > VK_EXT_memory_budget on Intel.< / li >
2019-01-29 23:32:16 +00:00
< / ul >
< h2 > Bug fixes< / h2 >
< ul >
2019-06-11 16:07:39 +01:00
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=81843" > Bug 81843< / a > - [SNB IVB HSW] ETC2 textures are not returned as compressed images< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=99781" > Bug 99781< / a > - Some Unity games fail assertion on startup in glXCreateContextAttribsARB< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=100239" > Bug 100239< / a > - Incorrect rendering in CS:GO< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=100316" > Bug 100316< / a > - Linking GLSL 1.30 shaders with invariant and deprecated variables triggers an 'mismatching invariant qualifiers' error< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=104272" > Bug 104272< / a > - [OpenGL CTS] [HSW] KHR-GL46.direct_state_access.textures_compressed_subimage assert fails< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=104355" > Bug 104355< / a > - Ivy Bridge ignores component mappings in texture views< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=104602" > Bug 104602< / a > - [apitrace] Graphical artifacts in Civilization VI on RX Vega< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=107052" > Bug 107052< / a > - [Regression][bisected]. Crookz - The Big Heist Demo can't be launched despite the " true" flag in " drirc" < / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=107505" > Bug 107505< / a > - [lars] dEQP-GLES31.functional.geometry_shading.layered#render_with_default_layer_3d failure< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=107510" > Bug 107510< / a > - [GEN8+] up to 10% perf drop on several 3D benchmarks< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=107563" > Bug 107563< / a > - [RADV] Broken rendering in Unity demos< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=107987" > Bug 107987< / a > - [Debug mesa only]. Crash happens when calling drawArrays< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108250" > Bug 108250< / a > - [GLSL] layout-location-struct.shader_test fails to link< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108457" > Bug 108457< / a > - [OpenGL CTS] KHR-GL46.tessellation_shader.single.xfb_captures_data_from_correct_stage fails< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108540" > Bug 108540< / a > - vkAcquireNextImageKHR blocks when timeout=0 in Wayland< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108766" > Bug 108766< / a > - Mesa built with meson has RPATH entries< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108824" > Bug 108824< / a > - Invalid handling when GL buffer is bound on one context and invalidated on another< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108841" > Bug 108841< / a > - [RADV] SPIRV's control flow attributes do not propagate to LLVM< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108879" > Bug 108879< / a > - [CIK] [regression] All opencl apps hangs indefinitely in si_create_context< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=108999" > Bug 108999< / a > - Calculating the scissors fields when the y is flipped (0 on top) can generate negative numbers that will cause assertion failure later on.< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109057" > Bug 109057< / a > - texelFetch from GL_TEXTURE_2D_MULTISAMPLE with integer format fails< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109107" > Bug 109107< / a > - gallium/st/va: change va max_profiles when using Radeon VCN Hardware< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109216" > Bug 109216< / a > - 4-27% performance drop in Vulkan benchmarks< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109326" > Bug 109326< / a > - mesa: Meson configuration summary should be printed< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109328" > Bug 109328< / a > - [BSW BXT GLK] dEQP-VK.subgroups.arithmetic.subgroup regressions< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109391" > Bug 109391< / a > - LTO Build fails< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109401" > Bug 109401< / a > - [DXVK] Project Cars rendering problems< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109404" > Bug 109404< / a > - [ANV] The Witcher 3 shadows flickering< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109443" > Bug 109443< / a > - Build failure with MSVC when using Scons > = 3.0.2< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109451" > Bug 109451< / a > - [IVB,SNB] LINE_STRIPs following a TRIANGLE_FAN fail to use primitive restart< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109543" > Bug 109543< / a > - After upgrade mesa to 19.0.0~rc1 all vulkan based application stop working [" vulkan-cube" received SIGSEGV in radv_pipeline_init_blend_state at ../src/amd/vulkan/radv_pipeline.c:699]< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109561" > Bug 109561< / a > - [regression, bisected] code re-factor causing games to stutter or lock-up system< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109573" > Bug 109573< / a > - dEQP-VK.spirv_assembly.instruction.graphics.module.same_module< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109575" > Bug 109575< / a > - Mesa-19.0.0-rc1 : Computer Crashes trying to run anything Vulkan< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109581" > Bug 109581< / a > - [BISECTED] Nothing is Rendered on Sascha Willem's " subpasses" demo< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109594" > Bug 109594< / a > - totem assert failure: totem: src/intel/genxml/gen9_pack.h:72: __gen_uint: La declaración `v < = max' no se cumple.< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109597" > Bug 109597< / a > - wreckfest issues with transparent objects & skybox< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109601" > Bug 109601< / a > - [Regression] RuneLite GPU rendering broken on 18.3.x< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109603" > Bug 109603< / a > - nir_instr_as_deref: Assertion `parent & & parent-> type == nir_instr_type_deref' failed.< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109645" > Bug 109645< / a > - build error on arm64: tegra_screen.c:33: /usr/include/xf86drm.h:41:10: fatal error: drm.h: No such file or directory< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109646" > Bug 109646< / a > - New video compositor compute shader render glitches mpv< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109647" > Bug 109647< / a > - /usr/include/xf86drm.h:40:10: fatal error: drm.h: No such file or directory< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109648" > Bug 109648< / a > - AMD Raven hang during va-api decoding< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109659" > Bug 109659< / a > - Missing OpenGL symbols in OSMesa Gallium when building with meson< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109698" > Bug 109698< / a > - dri.pc contents invalid when built with meson< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109717" > Bug 109717< / a > - [regression] Cull distance tests asserting< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109735" > Bug 109735< / a > - [Regression] broken font with mesa_vulkan_overlay< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109738" > Bug 109738< / a > - Child of Light shows only a black screen< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109739" > Bug 109739< / a > - Mesa build fails when vulkan-overlay-layer option is enabled< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109742" > Bug 109742< / a > - vdpau state tracker on nv92 started to hit assert after vl compute work< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109743" > Bug 109743< / a > - Test fails: piglit.spec.arb_sample_shading.arb_sample_shading-builtin-gl-sample-mask-mrt-alpha< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109747" > Bug 109747< / a > - Add framerate to vulkan-overlay-layer< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109759" > Bug 109759< / a > - [BISECTED][REGRESSION][IVB, HSW] Font rendering problem in OpenGL< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109788" > Bug 109788< / a > - vulkan-overlay-layer: Only installs 64bit version< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109810" > Bug 109810< / a > - nir_opt_copy_prop_vars.c:454: error: unknown field ‘ ssa’ specified in initializer< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109929" > Bug 109929< / a > - tgsi_to_nir.c:2111: undefined reference to `gl_nir_lower_samplers_as_deref'< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109944" > Bug 109944< / a > - [bisected] Android build test fails with: utils.c: error: use of undeclared identifier 'PACKAGE_VERSION'< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109945" > Bug 109945< / a > - pan_assemble.c:51:46: error: passing argument 2 of ‘ tgsi_to_nir’ from incompatible pointer type [-Werror=incompatible-pointer-types]< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109980" > Bug 109980< / a > - [i915 CI][HSW] spec@ arb_fragment_shader_interlock@ arb_fragment_shader_interlock-image-load-store - fail< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=109984" > Bug 109984< / a > - unhandled VkStructureType VK_STRUCTURE_TYPE_RENDER_PASS_INPUT_ATTACHMENT_ASPECT_CREATE_INFO< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110134" > Bug 110134< / a > - SIGSEGV while playing large hevc video in mpv< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110143" > Bug 110143< / a > - Doom 3: BFG Edition - Steam and GOG.com - white flickering screen< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110201" > Bug 110201< / a > - [ivb] mesa 19.0.0 breaks rendering in kitty< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110211" > Bug 110211< / a > - If DESTDIR is set to an empty string, the dri drivers are not installed< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110216" > Bug 110216< / a > - radv: Segfault when compiling compute shaders from Assassin's Creed Odyssey (regression, bisected)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110221" > Bug 110221< / a > - build error with meson< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110239" > Bug 110239< / a > - Mesa SIGABRT: src/intel/genxml/gen9_pack.h:72: __gen_uint: Assertion `v < = max' failed< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110257" > Bug 110257< / a > - Major artifacts in mpeg2 vaapi hw decoding< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110259" > Bug 110259< / a > - radv: Sampling depth-stencil image in GENERAL layout returns nothing but zero (regression, bisected)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110291" > Bug 110291< / a > - Vega 64 GPU hang running Space Engineers< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110302" > Bug 110302< / a > - [bisected][regression] piglit egl-create-pbuffer-surface and egl-gl-colorspace regressions< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110305" > Bug 110305< / a > - Iris driver fails ext_packed_depth_stencil-getteximage test< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110311" > Bug 110311< / a > - [IVB HSW SNB][regression][bisected] regressions on vec4 deqp/gl{es}cts tests< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110349" > Bug 110349< / a > - radv: Dragon Quest XI (DXVK) has a graphical glitch (regression, bisected)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110353" > Bug 110353< / a > - weird colors seen in valley< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110355" > Bug 110355< / a > - radeonsi: GTK elements become invisible in some applications (GIMP, LibreOffice)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110356" > Bug 110356< / a > - install_megadrivers.py creates new dangling symlink [bisected]< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110404" > Bug 110404< / a > - Iris fails piglit.spec.ext_transform_feedback.immediate-reuse test< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110422" > Bug 110422< / a > - AMD_DEBUG=forcedma will crash OpenGL aps with SIGFAULT on VegaM 8706G< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110441" > Bug 110441< / a > - [llvmpipe] complex-loop-analysis-bug regression< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110443" > Bug 110443< / a > - vaapi/vpp: wrong output for non 64-bytes align width (ex: 1200)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110454" > Bug 110454< / a > - [llvmpipe] piglit arb_color_buffer_float-render GL_RGBA8_SNORM failure with llvm-9< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110462" > Bug 110462< / a > - Epic Games Launcher renders nothing with " -opengl" option< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110474" > Bug 110474< / a > - [bisected][regression] vk cts fp16 arithmetic failures< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110497" > Bug 110497< / a > - [DXVK][Regression][Bisected][SKL] Project Cars 2 crashes with Bug Splat when loading finishes< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110526" > Bug 110526< / a > - [CTS] dEQP-VK.ycbcr.{conversion,format}.* fail< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110530" > Bug 110530< / a > - [CTS] dEQP-VK.ycbcr.format.g8_b8_r8_3plane_420* reports VM faults on Vega10< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110535" > Bug 110535< / a > - [bisected] [icl] GPU hangs on crucible func.miptree.r8g8b8a8-unorm.aspect-color.view-2d.levels01.array01.extent-512x512.upload-copy-with-draw tests< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110540" > Bug 110540< / a > - [AMD TAHITI XT] valve artifact broken< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110573" > Bug 110573< / a > - Mesa vulkan-radeon 19.0.3 system freeze and visual artifacts (RADV)< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110590" > Bug 110590< / a > - [Regression][Bisected] GTAⅣ under wine fails with GLXBadFBConfig< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110632" > Bug 110632< / a > - " glx: Fix synthetic error generation in __glXSendError" broke wine games on 32-bit< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110648" > Bug 110648< / a > - Dota2 will not open using vulkan since 19.0 series< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110655" > Bug 110655< / a > - VK_LAYER_MESA_OVERLAY_CONFIG=draw,fps renders sporadically< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110698" > Bug 110698< / a > - tu_device.c:900:4: error: initializer element is not constant< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110701" > Bug 110701< / a > - GPU faults in in Unigine Valley 1.0< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110721" > Bug 110721< / a > - graphics corruption on steam client with mesa 19.1.0 rc3 on polaris< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110761" > Bug 110761< / a > - Huge problems between Mesa and Electron engine apps< / li >
< li > < a href = "https://bugs.freedesktop.org/show_bug.cgi?id=110784" > Bug 110784< / a > - [regression][bisected] Reverting 'expose 0 shader binary formats for compat profiles for Qt' causes get_program_binary failures on Iris< / li >
2019-01-29 23:32:16 +00:00
< / ul >
2019-06-11 16:07:39 +01:00
2019-01-29 23:32:16 +00:00
< h2 > Changes< / h2 >
2019-06-11 16:07:39 +01:00
< p > Adam Jackson (1):< / p >
< ul >
< li > drisw: Try harder to probe whether MIT-SHM works< / li >
< / ul >
< p > Albert Pal (1):< / p >
< ul >
< li > Fix link release notes for 19.0.0.< / li >
< / ul >
< p > Alejandro Piñeiro (12):< / p >
< ul >
< li > blorp: introduce helper method blorp_nir_init_shader< / li >
< li > nir, glsl: move pixel_center_integer/origin_upper_left to shader_info.fs< / li >
< li > nir/xfb: add component_offset at nir_xfb_info< / li >
< li > nir_types: add glsl_varying_count helper< / li >
< li > nir/xfb: adding varyings on nir_xfb_info and gather_info< / li >
< li > nir/xfb: sort varyings too< / li >
< li > nir_types: add glsl_type_is_struct helper< / li >
< li > nir/xfb: handle arrays and AoA of basic types< / li >
< li > nir/linker: use nir_gather_xfb_info< / li >
< li > nir/linker: fix ARRAY_SIZE query with xfb varyings< / li >
< li > nir/xfb: move varyings info out of nir_xfb_info< / li >
< li > docs: document MESA_GLSL=errors keyword< / li >
< / ul >
< p > Alexander von Gluck IV (1):< / p >
< ul >
< li > haiku: Fix hgl dispatch build. Tested under meson/scons.< / li >
< / ul >
< p > Alexandros Frantzis (1):< / p >
< ul >
< li > virgl: Fake MSAA when max samples is 1< / li >
< / ul >
< p > Alok Hota (32):< / p >
< ul >
< li > swr/rast: update SWR rasterizer shader stats< / li >
< li > gallium/swr: Param defaults for unhandled PIPE_CAPs< / li >
< li > gallium/aux: add PIPE_CAP_MAX_VARYINGS to u_screen< / li >
< li > swr/rast: Convert system memory pointers to gfxptr_t< / li >
< li > swr/rast: Disable use of __forceinline by default< / li >
< li > swr/rast: Correctly align 64-byte spills/fills< / li >
< li > swr/rast: Flip BitScanReverse index calculation< / li >
< li > swr/rast: Move knob defaults to generated cpp file< / li >
< li > swr/rast: FP consistency between POSH/RENDER pipes< / li >
< li > swr/rast: Refactor scratch space variable names< / li >
< li > swr/rast: convert DWORD-> uint32_t, QWORD-> uint64_t< / li >
< li > swr/rast: simdlib cleanup, clipper stack space fixes< / li >
< li > swr/rast: Add translation support to streamout< / li >
< li > swr/rast: bypass size limit for non-sampled textures< / li >
< li > swr/rast: Cleanup and generalize gen_archrast< / li >
< li > swr/rast: Add initial SWTag proto definitions< / li >
< li > swr/rast: Add string handling to AR event framework< / li >
< li > swr/rast: Add general SWTag statistics< / li >
< li > swr/rast: Fix autotools and scons codegen< / li >
< li > swr/rast: Remove deprecated 4x2 backend code< / li >
< li > swr/rast: AVX512 support compiled in by default< / li >
< li > swr/rast: enforce use of tile offsets< / li >
< li > swr/rast: add more llvm intrinsics< / li >
< li > swr/rast: update guardband rects at draw setup< / li >
< li > swr/rast: add SWR_STATIC_ASSERT() macro< / li >
< li > swr/rast: add flat shading< / li >
< li > swr/rast: add guards for cpuid on Linux< / li >
< li > swr/rast: early exit on empty triangle mask< / li >
< li > swr/rast: Cleanup and generalize gen_archrast< / li >
< li > swr/rast: Add initial SWTag proto definitions< / li >
< li > swr/rast: Add string handling to AR event framework< / li >
< li > swr/rast: Add general SWTag statistics< / li >
< / ul >
< p > Alyssa Rosenzweig (192):< / p >
< ul >
< li > panfrost: Initial stub for Panfrost driver< / li >
< li > panfrost: Implement Midgard shader toolchain< / li >
< li > meson: Remove panfrost from default driver list< / li >
< li > kmsro: Move DRM entrypoints to shared block< / li >
< li > panfrost: Use u_pipe_screen_get_param_defaults< / li >
< li > panfrost: Check in sources for command stream< / li >
< li > panfrost: Include glue for out-of-tree legacy code< / li >
< li > kmsro: Silence warning if missing< / li >
< li > panfrost: Clean-up one-argument passing quirk< / li >
< li > panfrost: Don't hardcode number of nir_ssa_defs< / li >
< li > panfrost: Add kernel-agnostic resource management< / li >
< li > panfrost: Remove if 0'd dead code< / li >
< li > panfrost: Remove speculative if 0'd format bit code< / li >
< li > panfrost: Elucidate texture op scheduling comment< / li >
< li > panfrost: Specify supported draw modes per-context< / li >
< li > panfrost: Fix build; depend on libdrm< / li >
< li > panfrost: Backport driver to Mali T600/T700< / li >
< li > panfrost: Identify MALI_OCCLUSION_PRECISE bit< / li >
< li > panfrost: Implement PIPE_QUERY_OCCLUSION_COUNTER< / li >
< li > panfrost: Don't align framebuffer dims< / li >
< li > panfrost: Improve logging and patch memory leaks< / li >
< li > panfrost: Fix various leaks unmapping resources< / li >
< li > panfrost: Free imported BOs< / li >
< li > panfrost: Swap order of tiled texture (de)alloc< / li >
< li > panfrost: Cleanup mali_viewport (clipping) code< / li >
< li > panfrost: Preserve w sign in perspective division< / li >
< li > panfrost: Fix clipping region< / li >
< li > panfrost: Stub out separate stencil functions< / li >
< li > panfrost: Add pandecode (command stream debugger)< / li >
< li > panfrost: Implement pantrace (command stream dump)< / li >
< li > panfrost/midgard: Refactor tag lookahead code< / li >
< li > panfrost/midgard: Fix nested/chained if-else< / li >
< li > panfrost: Rectify doubleplusungood extended branch< / li >
< li > panfrost/midgard: Emit extended branches< / li >
< li > panfrost: Dynamically set discard branch targets< / li >
< li > panfrost: Verify and print brx condition in disasm< / li >
< li > panfrost: Use tiler fast path (performance boost)< / li >
< li > panfrost/meson: Remove subdir for nondrm< / li >
< li > panfrost/nondrm: Flag CPU-invisible regions< / li >
< li > panfrost/nondrm: Make COHERENT_LOCAL explicit< / li >
< li > panfrost/nondrm: Split out dump_counters< / li >
< li > panfrost/midgard: Add fround(_even), ftrunc, ffma< / li >
< li > panfrost: Decode render target swizzle/channels< / li >
< li > panfrost: Add RGB565, RGB5A1 texture formats< / li >
< li > panfrost: Identify 4-bit channel texture formats< / li >
< li > panfrost: Expose perf counters in environment< / li >
< li > panfrost/midgard: Allow flt to run on most units< / li >
< li > panfrost: Import job data structures from v3d< / li >
< li > panfrost: Decouple Gallium clear from FBD clear< / li >
< li > panfrost: Cleanup cruft related to clears< / li >
< li > panfrost/midgard: Don't force constant on VLUT< / li >
< li > panfrost: Flush with offscreen rendering< / li >
< li > panfrost/midgard: Promote smul to vmul< / li >
< li > panfrost/midgard: Preview for data hazards< / li >
< li > panfrost: List primitive restart enable bit< / li >
< li > panfrost/drm: Cast pointer to u64 to fix warning< / li >
< li > panfrost: Cleanup needless if in create_bo< / li >
< li > panfrost: Combine has_afbc/tiled in layout enum< / li >
< li > panfrost: Delay color buffer setup< / li >
< li > panfrost: Determine framebuffer format bits late< / li >
< li > panfrost: Allocate dedicated slab for linear BOs< / li >
< li > panfrost: Support linear depth textures< / li >
< li > panfrost: Document "depth-buffer writeback" bit< / li >
< li > panfrost: Identify fragment_extra flags< / li >
< li > util: Add a drm_find_modifier helper< / li >
< li > v3d: Use shared drm_find_modifier util< / li >
< li > vc4: Use shared drm_find_modifier util< / li >
< li > freedreno: Use shared drm_find_modifier util< / li >
< li > panfrost: Break out fragment to SFBD/MFBD files< / li >
< li > panfrost: Remove staging SFBD for pan_context< / li >
< li > panfrost: Remove staging MFBD< / li >
< li > panfrost: Minor comment cleanup (version detection)< / li >
< li > panfrost/mfbd: Implement linear depth buffers< / li >
< li > panfrost/mfbd: Respect per-job depth write flag< / li >
< li > panfrost: Comment spelling fix< / li >
< li > panfrost: Allocate extra data for depth buffer< / li >
< li > panfrost; Disable AFBC for depth buffers< / li >
< li > panfrost: Compute viewport state on the fly< / li >
< li > panfrost/midgard: Implement fpow< / li >
< li > panfrost: Workaround buffer overrun with mip level< / li >
< li > panfrost: Fix primconvert check< / li >
< li > panfrost: Disable PIPE_CAP_TGSI_TEXCOORD< / li >
< li > panfrost/decode: Respect primitive size pointers< / li >
< li > panfrost: Replay more varying buffers< / li >
< li > panfrost: Rewrite varying assembly< / li >
< li > panfrost/midgard: Fix b2f32 swizzle for vectors< / li >
< li > panfrost: Fix viewports< / li >
< li > panfrost: Implement scissor test< / li >
< li > panfrost/midgard: Add fcsel_i opcode< / li >
< li > panfrost/midgard: Schedule ball/bany to vectors< / li >
< li > panfrost/midgard: Add more ball/bany, iabs ops< / li >
< li > panfrost/midgard: Map more bany/ball opcodes< / li >
< li > panfrost/midgard: Lower bool_to_int32< / li >
< li > panfrost/midgard: Lower f2b32 to fne< / li >
< li > panfrost/midgard: Lower i2b32< / li >
< li > panfrost/midgard: Implement b2i; improve b2f/f2b< / li >
< li > panfrost/midgard: Lower source modifiers for ints< / li >
< li > panfrost/midgard: Cleanup midgard_nir_algebraic.py< / li >
< li > panfrost: Stub out ES3 caps/callbacks< / li >
< li > panfrost/midgard: Add ult/ule ops< / li >
< li > panfrost/midgard: Expand fge lowering to more types< / li >
< li > panfrost/midgard: Handle i2b constant< / li >
< li > panfrost/midgard: fpow is a two-part operation< / li >
< li > panfrost: Preliminary work for mipmaps< / li >
< li > panfrost: Fix vertex buffer corruption< / li >
< li > panfrost/midgard: Disassemble `cube` texture op< / li >
< li > panfrost/midgard: Add L/S op for writing cubemap coordinates< / li >
< li > panfrost: Preliminary work for cubemaps< / li >
< li > panfrost/decode: Decode all cubemap faces< / li >
< li > panfrost: Include all cubemap faces in bitmap list< / li >
< li > panfrost/midgard: Emit cubemap coordinates< / li >
< li > panfrost: Implement command stream for linear cubemaps< / li >
< li > panfrost: Extend tiling for cubemaps< / li >
< li > panfrost: Implement missing texture formats< / li >
< li > panfrost/decode: Print negative_start< / li >
< li > panfrost: Clean index state between indexed draws< / li >
< li > panfrost: Fix index calculation types and asserts< / li >
< li > panfrost: Implement FIXED formats< / li >
< li > panfrost: Remove support for legacy kernels< / li >
< li > nir: Add "viewport vector" system values< / li >
< li > panfrost: Implement system values< / li >
< li > panfrost: Cleanup some indirection in pan_resource< / li >
< li > panfrost: Respect box-> width in tiled stores< / li >
< li > panfrost: Size tiled temp buffers correctly< / li >
< li > panfrost/decode: Add flags for tilebuffer readback< / li >
< li > panfrost: Add tilebuffer load? branch< / li >
< li > panfrost/midgard: Add umin/umax opcodes< / li >
< li > panfrost/midgard: Add ilzcnt op< / li >
< li > panfrost/midgard: Add ibitcount8 op< / li >
< li > panfrost/midgard: Enable lower_find_lsb< / li >
< li > panfrost: Remove "mali_unknown6" nonsense< / li >
< li > panfrost/midgard: Drop dependence on mesa/st< / li >
< li > panfrost: Cleanup indexed draw handling< / li >
< li > nir: Add nir_lower_viewport_transform< / li >
< li > panfrost/midgard: Use shared nir_lower_viewport_transform< / li >
< li > panfrost: Track BO lifetime with jobs and reference counts< / li >
< li > panfrost: Fixup vertex offsets to prevent shadow copy< / li >
< li > panfrost/mdg: Use shared fsign lowering< / li >
< li > panfrost/mdg/disasm: Print raw varying_parameters< / li >
< li > panfrost/midgard: Pipe through varying arrays< / li >
< li > panfrost/midgard: Implement indirect loads of varyings/UBOs< / li >
< li > panfrost/midgard: Respect component of bcsel condition< / li >
< li > panfrost/midgard: Remove useless MIR dump< / li >
< li > panfrost: Respect backwards branches in RA< / li >
< li > panfrost/midgard: Don't try to inline constants on branches< / li >
< li > panfrost/midgard: imul can only run on *mul< / li >
< li > panfrost: Disable indirect outputs for now< / li >
< li > panfrost: Use actual imov instruction< / li >
< li > panfrost/midgard: Dead code eliminate MIR< / li >
< li > panfrost/midgard: Track loop depth< / li >
< li > panfrost/midgard: Fix off-by-one in successor analysis< / li >
< li > panfrost/midgard: Remove unused mir_next_block< / li >
< li > panfrost/midgard: Update integer op list< / li >
< li > panfrost/midgard: Document sign-extension/zero-extension bits (vector)< / li >
< li > panfrost/midgard: Set integer mods< / li >
< li > panfrost/midgard: Implement copy propagation< / li >
< li > panfrost/midgard: Optimize MIR in progress loop< / li >
< li > panfrost/midgard: Refactor opcode tables< / li >
< li > panfrost/midgard: Add "op commutes?" property< / li >
< li > panfrost/midgard: Remove assembler< / li >
< li > panfrost/midgard: Reduce fmax(a, 0.0) to fmov.pos< / li >
< li > panfrost/midgard: Extend copy propagation pass< / li >
< li > panfrost/midgard: Optimize csel involving 0< / li >
< li > panfrost/midgard: Copy prop for texture registers< / li >
< li > panfrost/midgard: Identify inand< / li >
< li > panfrost/midgard: Add new bitwise ops< / li >
< li > Revert "panfrost/midgard: Extend copy propagation pass"< / li >
< li > panfrost/midgard: Only copyprop without an outmod< / li >
< li > panfrost/midgard: Fix regressions in -bjellyfish< / li >
< li > panfrost/midgard: Fix tex propogation< / li >
< li > panfrost/midgard: imov workaround< / li >
< li > panfrost: Use fp32 (not fp16) varyings< / li >
< li > panfrost/midgard: Safety check immediate precision degradations< / li >
< li > panfrost: Workaround -bshadow regression< / li >
< li > panfrost: Remove shader dump< / li >
< li > panfrost/decode: Hit MRT blend shader enable bits< / li >
< li > panfrost: Fix blend shader upload< / li >
< li > panfrost/midgard: reg_mode_full -> reg_mode_32, etc< / li >
< li > panfrost/midgard/disasm: Catch mask errors< / li >
< li > panfrost/midgard/disasm: Extend print_reg to 8-bit< / li >
< li > panfrost/midgard/disasm: Fill in .int mod< / li >
< li > panfrost/midgard: Fix crash on unknown op< / li >
< li > panfrost/midgard: Rename ilzcnt8 -> iclz< / li >
< li > panfrost/midgard/disasm: Support 8-bit destination< / li >
< li > panfrost/midgard/disasm: Print 8-bit sources< / li >
< li > panfrost/midgard/disasm: Stub out 64-bit< / li >
< li > panfrost/midgard/disasm: Handle dest_override generalized< / li >
< li > panfrost: Support RGB565 FBOs< / li >
< li > panfrost/midgard: Fix integer selection< / li >
< li > panfrost/midgard: Fix RA when temp_count = 0< / li >
< li > panfrost/midgard: Lower mixed csel (NIR)< / li >
< li > panfrost/midgard: iabs cannot run on mul< / li >
< / ul >
< p > Alyssa Ross (1):< / p >
< ul >
< li > get_reviewer.pl: improve portability< / li >
< / ul >
< p > Amit Pundir (1):< / p >
< ul >
< li > mesa: android: freedreno: build libfreedreno_{drm,ir3} static libs< / li >
< / ul >
< p > Andre Heider (5):< / p >
2019-01-29 23:32:16 +00:00
< ul >
2019-06-11 16:07:39 +01:00
< li > iris: fix build with gallium nine< / li >
< li > iris: improve PIPE_CAP_VIDEO_MEMORY bogus value< / li >
< li > iris: add support for tgsi_to_nir< / li >
< li > st/nine: enable csmt per default on iris< / li >
< li > st/nine: skip position checks in SetCursorPosition()< / li >
2019-01-29 23:32:16 +00:00
< / ul >
2019-06-11 16:07:39 +01:00
< p > Andreas Baierl (2):< / p >
< ul >
< li > nir: add rcp(w) lowering for gl_FragCoord< / li >
< li > lima/ppir: Add gl_FragCoord handling< / li >
< / ul >
< p > Andres Gomez (12):< / p >
< ul >
< li > mesa: INVALID_VALUE for wrong type or format in Clear*Buffer*Data< / li >
< li > gitlab-ci: install distro's ninja< / li >
< li > glsl: correctly validate component layout qualifier for dvec{3,4}< / li >
< li > glsl/linker: always validate explicit location among inputs< / li >
< li > glsl/linker: don't fail non static used inputs without matching outputs< / li >
< li > glsl/linker: simplify xfb_offset vs xfb_stride overflow check< / li >
< li > Revert "glsl: relax input-> output validation for SSO programs"< / li >
< li > glsl/linker: location aliasing requires types to have the same width< / li >
< li > docs: drop Andres Gomez from the release cycles< / li >
< li > glsl/linker: always validate explicit locations for first and last interfaces< / li >
< li > docs/relnotes: add support for VK_KHR_shader_float16_int8< / li >
< li > glsl/linker: check for xfb_offset aliasing< / li >
< / ul >
< p > Andrii Simiklit (5):< / p >
< ul >
< li > i965: consider a 'base level' when calculating width0, height0, depth0< / li >
< li > i965: re-emit index buffer state on a reset option change.< / li >
< li > util: clean the 24-bit unused field to avoid an issues< / li >
< li > iris: make the TFB result visible to others< / li >
< li > egl: return correct error code for a case req ver < 3 with forward-compatible< / li >
< / ul >
< p > Antia Puentes (1):< / p >
< ul >
< li > nir/linker: Fix TRANSFORM_FEEDBACK_BUFFER_INDEX< / li >
< / ul >
< p > Anuj Phogat (7):< / p >
< ul >
< li > i965/icl: Add WA_2204188704 to disable pixel shader panic dispatch< / li >
< li > anv/icl: Add WA_2204188704 to disable pixel shader panic dispatch< / li >
< li > intel: Add Elkhart Lake device info< / li >
< li > intel: Add Elkhart Lake PCI-IDs< / li >
< li > iris/icl: Set Enabled Texel Offset Precision Fix bit< / li >
< li > iris/icl: Add WA_2204188704 to disable pixel shader panic dispatch< / li >
< li > intel: Add support for Comet Lake< / li >
< / ul >
< p > Axel Davy (49):< / p >
< ul >
< li > st/nine: Ignore window size if error< / li >
< li > st/nine: Ignore multisample quality level if no ms< / li >
< li > st/nine: Disable depth write when nothing gets updated< / li >
< li > st/nine: Do not advertise support for D15S1 and D24X4S4< / li >
< li > st/nine: Do not advertise CANMANAGERESOURCE< / li >
< li > st/nine: Change a few advertised caps< / li >
< li > Revert "d3dadapter9: Support software renderer on any DRI device"< / li >
< li > st/nine: Fix D3DWindowBuffer_release for old wine nine support< / li >
< li > st/nine: Use FLT_MAX/2 for RCP clamping< / li >
< li > st/nine: Upload managed textures only at draw using them< / li >
< li > st/nine: Upload managed buffers only at draw using them< / li >
< li > st/nine: Fix buffer/texture unbinding in nine_state_clear< / li >
< li > st/nine: Finish if nooverwrite after normal mapping< / li >
< li > st/nine: Always return OK on SetSoftwareVertexProcessing< / li >
< li > st/nine: Enable modifiers on ps 1.X texcoords< / li >
< li > st/nine: Ignore nooverwrite for systemmem< / li >
< li > st/nine: Fix SINCOS input< / li >
< li > st/nine: Optimize surface upload with conversion< / li >
< li > st/nine: Optimize volume upload with conversion< / li >
< li > st/nine: rename *_conversion to *_internal< / li >
< li > st/nine: Refactor surface GetSystemMemPointer< / li >
< li > st/nine: Refactor volume GetSystemMemPointer< / li >
< li > st/nine: Support internal compressed format for surfaces< / li >
< li > st/nine: Support internal compressed format for volumes< / li >
< li > st/nine: Add drirc option to use data_internal for dynamic textures< / li >
< li > drirc: Add Gallium nine workaround for Rayman Legends< / li >
< li > st/nine: Recompile optimized shaders based on b/i consts< / li >
< li > st/nine: Control shader constant inlining with drirc< / li >
< li > st/nine: Regroup param-> rel tests< / li >
< li > st/nine: Refactor param-> rel< / li >
< li > st/nine: Compact nine_ff_get_projected_key< / li >
< li > st/nine: Compact pixel shader key< / li >
< li > st/nine: use helper ureg_DECL_sampler everywhere< / li >
< li > st/nine: Manually upload vs and ps constants< / li >
< li > st/nine: Refactor shader constants ureg_src computation< / li >
< li > st/nine: Make swvp_on imply IS_VS< / li >
< li > st/nine: Refactor ct_ctor< / li >
< li > st/nine: Track constant slots used< / li >
< li > st/nine: Refactor counting of constants< / li >
< li > st/nine: Prepare constant compaction in nine_shader< / li >
< li > st/nine: Propagate const_range to context< / li >
< li > st/nine: Cache constant buffer size< / li >
< li > st/nine: Handle const_ranges in nine_state< / li >
< li > st/nine: Enable computing const_ranges< / li >
< li > st/nine: Use TGSI_SEMANTIC_GENERIC for fog< / li >
< li > st/nine: Optimize a bit writeonly buffers< / li >
< li > st/nine: Throttle rendering similarly for thread_submit< / li >
< li > st/nine: Check discard_delayed_release is set before allocating more< / li >
< li > d3dadapter9: Revert to old throttling limit value< / li >
< / ul >
< p > Bart Oldeman (1):< / p >
< ul >
< li > gallium-xlib: query MIT-SHM before using it.< / li >
< / ul >
< p > Bas Nieuwenhuizen (105):< / p >
< ul >
< li > radv: Only look at pImmutableSamples if the descriptor has a sampler.< / li >
< li > amd/common: Add gep helper for pointer increment.< / li >
< li > amd/common: Implement ptr-> int casts in ac_to_integer.< / li >
< li > radv: Fix the shader info pass for not having the variable.< / li >
< li > amd/common: Use correct writemask for shared memory stores.< / li >
< li > amd/common: Fix stores to derefs with unknown variable.< / li >
< li > amd/common: Handle nir_deref_type_ptr_as_array for shared memory.< / li >
< li > amd/common: handle nir_deref_cast for shared memory from integers.< / li >
< li > amd/common: Do not use 32-bit loads for shared memory.< / li >
< li > amd/common: Implement global memory accesses.< / li >
< li > radv: Do not use the bo list for local buffers.< / li >
< li > radv: Implement VK_EXT_buffer_device_address.< / li >
< li > radv: Use correct num formats to detect whether we should be use 1.0 or 1.< / li >
< li > radv: Sync ETC2 whitelisted devices.< / li >
< li > radv: Clean up a bunch of compiler warnings.< / li >
< li > radv: Handle clip+cull distances more generally as compact arrays.< / li >
< li > radv: Implement VK_EXT_depth_clip_enable.< / li >
< li > radv: Disable depth clamping even without EXT_depth_range_unrestricted.< / li >
< li > radv: Fix float16 interpolation set up.< / li >
< li > radv: Allow interpolation on non-float types.< / li >
< li > radv: Interpolate less aggressively.< / li >
< li > turnip: Add driver skeleton (v2)< / li >
< li > turnip: Fix up detection of device.< / li >
< li > turnip: Gather some device info.< / li >
< li > turnip: Remove abort.< / li >
< li > turnip: Fix newly introduced warning.< / li >
< li > turnip: Add buffer allocation & mapping support.< / li >
< li > turnip: Report a memory type and heap.< / li >
< li > turnip: Cargo cult the Intel heap size functionality.< / li >
< li > turnip: Initialize memory type in requirements.< / li >
< li > turnip: Disable more features.< / li >
< li > turnip: Add 630 to the list.< / li >
< li > turnip: Fix bo allocation after we stopped using libdrm_freedreno ...< / li >
< li > turnip: Fix memory mapping.< / li >
< li > turnip: Add image layout calculations.< / li >
< li > turnip: Stop hardcoding the msm version check.< / li >
< li > turnip: move tu_gem.c to tu_drm.c< / li >
< li > turnip: Implement pipe-less param query.< / li >
< li > turnip: Implement some format properties for RGBA8.< / li >
< li > turnip: Remove some radv leftovers.< / li >
< li > turnip: clean up TODO.< / li >
< li > turnip: Implement some UUIDs.< / li >
< li > turnip: Implement a slow bo list< / li >
< li > turnip: Add a command stream.< / li >
< li > turnip: Add msm queue support.< / li >
< li > turnip: Make bo_list functions not static< / li >
< li > turnip: Implement submission.< / li >
< li > turnip: Fill command buffer< / li >
< li > turnip: Shorten primary_cmd_stream name.< / li >
< li > turnip: Add emit functions in a header.< / li >
< li > turnip: Move stream functions to tu_cs.c< / li >
< li > turnip: Add buffer memory binding.< / li >
< li > turnip: Make tu6_emit_event_write shared.< / li >
< li > turnip: Add tu6_rb_fmt_to_ifmt.< / li >
< li > turnip: Implement buffer-> buffer DMA copies.< / li >
< li > turnip: Add image-> buffer DMA copies.< / li >
< li > turnip: Add buffer-> image DMA copies.< / li >
< li > turnip: Add todo for copies.< / li >
< li > turnip: Fix GCC compiles.< / li >
< li > turnip: Deconflict vk_format_table regeneration< / li >
< li > gitlab-ci: Build turnip.< / li >
< li > radeonsi: Remove implicit const cast.< / li >
< li > radv: Allow fast clears with concurrent queue mask for some layouts.< / li >
< li > vulkan/util: Handle enums that are in platform-specific headers.< / li >
< li > vulkan: Update the XML and headers to 1.1.104< / li >
< li > radv: Implement VK_EXT_host_query_reset.< / li >
< li > radv: Use correct image view comparison for fast clears.< / li >
< li > radv: Implement VK_EXT_pipeline_creation_feedback.< / li >
< li > ac/nir: Return frag_coord as integer.< / li >
< li > nir: Add access qualifiers on load_ubo intrinsic.< / li >
< li > radv: Add non-uniform indexing lowering.< / li >
< li > radv: Add bolist RADV_PERFTEST flag.< / li >
< li > ac: Move has_local_buffers disable to radeonsi.< / li >
< li > radv: Use local buffers for the global bo list.< / li >
< li > radv: Support VK_EXT_inline_uniform_block.< / li >
< li > radv: Add support for driconf.< / li >
< li > vulkan/wsi: Add X11 adaptive sync support based on dri options.< / li >
< li > radv: Add adaptive_sync driconfig option and enable it by default.< / li >
< li > radv: Add logic for subsampled format descriptions.< / li >
< li > radv: Add logic for multisample format descriptions.< / li >
< li > radv: Add multiple planes to images.< / li >
< li > radv: Add single plane image views & meta operations.< / li >
< li > radv: Support different source & dest aspects for planar images in blit2d.< / li >
< li > radv: Add ycbcr conversion structs.< / li >
< li > radv: Add support for image views with multiple planes.< / li >
< li > radv: Allow mixed src/dst aspects in copies.< / li >
< li > ac/nir: Add support for planes.< / li >
< li > radv: Add ycbcr samplers in descriptor set layouts.< / li >
< li > radv: Update descriptor sets for multiple planes.< / li >
< li > radv: Add ycbcr lowering pass.< / li >
< li > radv: Run the new ycbcr lowering pass.< / li >
< li > radv: Add hashing for the ycbcr samplers.< / li >
< li > radv: Add ycbcr format features.< / li >
< li > radv: Add ycbcr subsampled & multiplane formats to csv.< / li >
< li > radv: Enable YCBCR conversion feature.< / li >
< li > radv: Expose VK_EXT_ycbcr_image_arrays.< / li >
< li > radv: Expose Vulkan 1.1 for Android.< / li >
< li > radv: Fix hang width YCBCR array textures.< / li >
< li > radv: Set is_array in lowered ycbcr tex instructions.< / li >
< li > radv: Restrict YUVY formats to 1 layer.< / li >
< li > radv: Disable subsampled formats.< / li >
< li > radv: Implement cosited_even sampling.< / li >
< li > radv: Do not use extra descriptor space for the 3rd plane.< / li >
< li > nir: Actually propagate progress in nir_opt_move_load_ubo.< / li >
< li > radv: Prevent out of bound shift on 32-bit builds.< / li >
< / ul >
< p > Benjamin Gordon (1):< / p >
< ul >
< li > configure.ac/meson.build: Add options for library suffixes< / li >
< / ul >
< p > Benjamin Tissoires (1):< / p >
< ul >
< li > CI: use wayland ci-templates repo to create the base image< / li >
< / ul >
< p > Boyan Ding (3):< / p >
< ul >
< li > gk110/ir: Add rcp f64 implementation< / li >
< li > gk110/ir: Add rsq f64 implementation< / li >
< li > gk110/ir: Use the new rcp/rsq in library< / li >
< / ul >
< p > Boyuan Zhang (1):< / p >
< ul >
< li > st/va: reverse qt matrix back to its original order< / li >
< / ul >
< p > Brian Paul (51):< / p >
< ul >
< li > st/mesa: whitespace/formatting fixes in st_cb_texture.c< / li >
< li > svga: assorted whitespace and formatting fixes< / li >
< li > svga: fix dma.pending > 0 test< / li >
< li > mesa: fix display list corner case assertion< / li >
< li > st/mesa: whitespace fixes in st_sampler_view.c< / li >
< li > st/mesa: line wrapping, whitespace fixes in st_cb_texture.c< / li >
< li > st/mesa: whitespace fixes in st_texture.h< / li >
< li > svga: init fill variable to avoid compiler warning< / li >
< li > svga: silence array out of bounds warning< / li >
< li > st/wgl: init a variable to silence MinGW warning< / li >
< li > gallium/util: whitespace cleanups in u_bitmask.[ch]< / li >
< li > gallium/util: add some const qualifiers in u_bitmask.c< / li >
< li > pipebuffer: use new pb_usage_flags enum type< / li >
< li > pipebuffer: whitespace fixes in pb_buffer.h< / li >
< li > winsys/svga: use new pb_usage_flags enum type< / li >
< li > st/mesa: move, clean-up shader variant key decls/inits< / li >
< li > st/mesa: whitespace, formatting fixes in st_cb_flush.c< / li >
< li > svga: refactor draw_vgpu10() function< / li >
< li > svga: remove SVGA_RELOC_READ flag in SVGA3D_BindGBSurface()< / li >
< li > pipebuffer: s/PB_ALL_USAGE_FLAGS/PB_USAGE_ALL/< / li >
< li > st/mesa: init hash keys with memset(), not designated initializers< / li >
< li > intel/decoders: silence uninitialized variable warnings in gen_print_batch()< / li >
< li > intel/compiler: silence unitialized variable warning in opt_vector_float()< / li >
< li > st/mesa: move utility functions, macros into new st_util.h file< / li >
< li > st/mesa: move around some code in st_context.c< / li >
< li > st/mesa: add/improve sampler view comments< / li >
< li > st/mesa: rename st_texture_release_sampler_view()< / li >
< li > st/mesa: minor refactoring of texture/sampler delete code< / li >
< li > docs: try to improve the Meson documentation (v2)< / li >
< li > drisw: fix incomplete type compilation failure< / li >
< li > gallium/winsys/kms: fix incomplete type compilation failure< / li >
< li > nir: silence a couple new compiler warnings< / li >
< li > docs: separate information for compiler selection and compiler options< / li >
< li > docs: link to the meson_options.txt file gitlab.freedesktop.org< / li >
< li > st/mesa: implement "zombie" sampler views (v2)< / li >
< li > st/mesa: implement "zombie" shaders list< / li >
< li > st/mesa: stop using pipe_sampler_view_release()< / li >
< li > svga: stop using pipe_sampler_view_release()< / li >
< li > llvmpipe: stop using pipe_sampler_view_release()< / li >
< li > swr: remove call to pipe_sampler_view_release()< / li >
< li > i915g: remove calls to pipe_sampler_view_release()< / li >
< li > gallium/util: remove pipe_sampler_view_release()< / li >
< li > nir: fix a few signed/unsigned comparison warnings< / li >
< li > st/mesa: fix texture deletion context mix-up issues (v2)< / li >
< li > nir: use {0} initializer instead of {} to fix MSVC build< / li >
< li > util: no-op __builtin_types_compatible_p() for non-GCC compilers< / li >
< li > docs: s/Aptril/April/< / li >
< li > llvmpipe: init some vars to NULL to silence MinGW compiler warnings< / li >
< li > glsl: work around MinGW 7.x compiler bug< / li >
< li > svga: add SVGA_NO_LOGGING env var (v2)< / li >
< li > glsl: fix typo in #warning message< / li >
< / ul >
< p > Caio Marcelo de Oliveira Filho (61):< / p >
< ul >
< li > nir: keep the phi order when splitting blocks< / li >
< li > i965: skip bit6 swizzle detection in Gen8+< / li >
< li > anv: skip bit6 swizzle detection in Gen8+< / li >
< li > isl: assert that Gen8+ don't have bit6_swizzling< / li >
< li > intel/compiler: use 0 as sampler in emit_mcs_fetch< / li >
< li > nir: fix example in opt_peel_loop_initial_if description< / li >
< li > iris: Fix uses of gl_TessLevel*< / li >
< li > iris: Add support for TCS passthrough< / li >
< li > iris: always include an extra constbuf0 if using UBOs< / li >
< li > nir/copy_prop_vars: don't get confused by array_deref of vectors< / li >
< li > nir/copy_prop_vars: add debug helpers< / li >
< li > nir/copy_prop_vars: keep track of components in copy_entry< / li >
< li > nir/copy_prop_vars: change test helper to get intrinsics< / li >
< li > nir: nir_build_deref_follower accept array derefs of vectors< / li >
< li > nir/copy_prop_vars: add tests for load/store elements of vectors< / li >
< li > nir: fix MSVC build< / li >
< li > st/nir: count num_uniforms for FS bultin shader< / li >
< li > nir/copy_prop_vars: rename/refactor store_to_entry helper< / li >
< li > nir/copy_prop_vars: use NIR_MAX_VEC_COMPONENTS< / li >
< li > nir/copy_prop_vars: handle load/store of vector elements< / li >
< li > nir/copy_prop_vars: add tests for indirect array deref< / li >
< li > nir/copy_prop_vars: prefer using entries from equal derefs< / li >
< li > nir/copy_prop_vars: handle indirect vector elements< / li >
< li > anv: Implement VK_EXT_external_memory_host< / li >
< li > nir: Add a pass to combine store_derefs to same vector< / li >
< li > intel/nir: Combine store_derefs after vectorizing IO< / li >
< li > intel/nir: Combine store_derefs to improve code from SPIR-V< / li >
< li > nir: Handle array-deref-of-vector case in loop analysis< / li >
< li > spirv: Add an execution environment to the options< / li >
< li > intel/compiler: handle GLSL_TYPE_INTERFACE as GLSL_TYPE_STRUCT< / li >
< li > spirv: Use interface type for block and buffer block< / li >
< li > iris: Clean up compiler warnings about unused< / li >
< li > nir: Take if_uses into account when repairing SSA< / li >
< li > mesa: Extension boilerplate for NV_compute_shader_derivatives< / li >
< li > glsl: Remove redundant conditions when asserting in_qualifier< / li >
< li > glsl: Enable derivative builtins for NV_compute_shader_derivatives< / li >
< li > glsl: Enable texture builtins for NV_compute_shader_derivatives< / li >
< li > glsl: Parse and propagate derivative_group to shader_info< / li >
< li > nir/algebraic: Lower CS derivatives to zero when no group defined< / li >
< li > nir: Don't set LOD=0 for compute shader that has derivative group< / li >
< li > intel/fs: Use TEX_LOGICAL whenever implicit lod is supported< / li >
< li > intel/fs: Add support for CS to group invocations in quads< / li >
< li > intel/fs: Don't loop when lowering CS intrinsics< / li >
< li > intel/fs: Use NIR_PASS_V when lowering CS intrinsics< / li >
< li > i965: Advertise NV_compute_shader_derivatives< / li >
< li > gallium: Add PIPE_CAP_COMPUTE_SHADER_DERIVATIVES< / li >
< li > iris: Enable NV_compute_shader_derivatives< / li >
< li > spirv: Add support for DerivativeGroup capabilities< / li >
< li > anv: Implement VK_NV_compute_shader_derivatives< / li >
< li > docs: Add NV_compute_shader_derivatives to 19.1.0 relnotes< / li >
< li > spirv: Add more to_string helpers< / li >
< li > spirv: Tell which opcode or value is unhandled when failing< / li >
< li > spirv: Rename vtn_decoration literals to operands< / li >
< li > spirv: Handle SpvOpDecorateId< / li >
< li > nir: Add option to lower tex to txl when shader don't support implicit LOD< / li >
< li > intel/fs: Don't handle texop_tex for shaders without implicit LOD< / li >
< li > spirv: Properly handle SpvOpAtomicCompareExchangeWeak< / li >
< li > intel/fs: Assert when brw_fs_nir sees a nir_deref_instr< / li >
< li > anv: Fix limits when VK_EXT_descriptor_indexing is used< / li >
< li > nir: Fix nir_opt_idiv_const when negatives are involved< / li >
< li > nir: Fix clone of nir_variable state slots< / li >
< / ul >
< p > Carlos Garnacho (1):< / p >
< ul >
< li > wayland/egl: Ensure EGL surface is resized on DRI update_buffers()< / li >
< / ul >
< p > Chad Versace (17):< / p >
< ul >
< li > turnip: Drop Makefile.am and Android.mk< / li >
< li > turnip: Fix indentation in function signatures< / li >
< li > turnip: Fix result of vkEnumerate*LayerProperties< / li >
< li > turnip: Fix result of vkEnumerate*ExtensionProperties< / li >
< li > turnip: Use vk_outarray in all relevant public functions< / li >
< li > turnip: Fix a real -Wmaybe-uninitialized< / li >
< li > turnip: Fix indentation< / li >
< li > turnip: Require DRM device version > = 1.3< / li >
< li > turnip: Add TODO for Android logging< / li >
< li > turnip: Use vk_errorf() for initialization error messages< / li >
< li > turnip: Replace fd_bo with tu_bo< / li >
< li > turnip: Add TODO file< / li >
< li > turnip: Fix 'unused' warnings< / li >
< li > turnip: Don't return from tu_stub funcs< / li >
< li > turnip: Annotate vkGetImageSubresourceLayout with tu_stub< / li >
< li > turnip: Fix error behavior for VkPhysicalDeviceExternalImageFormatInfo< / li >
< li > turnip: Use Vulkan 1.1 names instead of KHR< / li >
< / ul >
< p > Charmaine Lee (5):< / p >
< ul >
< li > svga: add svga shader type in the shader variant< / li >
< li > svga: move host logging to winsys< / li >
< li > st/mesa: purge framebuffers with current context after unbinding winsys buffers< / li >
< li > mesa: unreference current winsys buffers when unbinding winsys buffers< / li >
< li > svga: Remove unnecessary check for the pre flush bit for setting vertex buffers< / li >
< / ul >
< p > Chenglei Ren (1):< / p >
< ul >
< li > anv/android: fix missing dependencies issue during parallel build< / li >
< / ul >
< p > Chia-I Wu (78):< / p >
< ul >
< li > egl: fix KHR_partial_update without EXT_buffer_age< / li >
< li > turnip: add .clang-format< / li >
< li > turnip: use msm_drm.h from inc_freedreno< / li >
< li > turnip: remove unnecessary libfreedreno_drm dep< / li >
< li > turnip: add wrappers around DRM_MSM_GET_PARAM< / li >
< li > turnip: add wrappers around DRM_MSM_SUBMITQUEUE_*< / li >
< li > turnip: constify tu_device in tu_gem_*< / li >
< li > turnip: preliminary support for tu_QueueWaitIdle< / li >
< li > turnip: run sed and clang-format on tu_cs< / li >
< li > turnip: document tu_cs< / li >
< li > turnip: add tu_cs_add_bo< / li >
< li > turnip: minor cleanup to tu_cs_end< / li >
< li > turnip: update cs-> start in tu_cs_end< / li >
< li > turnip: inline tu_cs_check_space< / li >
< li > turnip: add more tu_cs helpers< / li >
< li > turnip: build drm_msm_gem_submit_bo array directly< / li >
< li > turnip: add tu_bo_list_merge< / li >
< li > turnip: add cmdbuf-> bo_list to bo_list in queue submit< / li >
< li > turnip: preliminary support for tu_BindImageMemory2< / li >
< li > turnip: preliminary support for tu_image_view_init< / li >
< li > turnip: preliminary support for tu_CmdBeginRenderPass< / li >
< li > turnip: add tu_cs_reserve_space(_assert)< / li >
< li > turnip: emit HW init in tu_BeginCommandBuffer< / li >
< li > turnip: preliminary support for tu_GetRenderAreaGranularity< / li >
< li > turnip: add tu_tiling_config< / li >
< li > turnip: add internal helpers for tu_cs< / li >
< li > turnip: add tu_cs_{reserve,add}_entry< / li >
< li > turnip: specify initial size in tu_cs_init< / li >
< li > turnip: never fail tu_cs_begin/tu_cs_end< / li >
< li > turnip: add tu_cs_sanity_check< / li >
< li > turnip: provide both emit_ib and emit_call< / li >
< li > turnip: add tu_cs_mode< / li >
< li > turnip: add TU_CS_MODE_SUB_STREAM< / li >
< li > turnip: preliminary support for loadOp and storeOp< / li >
< li > turnip: add a more complete format table< / li >
< li > turnip: add functions to import/export prime fd< / li >
< li > turnip: advertise VK_KHR_external_memory_capabilities< / li >
< li > turnip: advertise VK_KHR_external_memory< / li >
< li > turnip: add support for VK_KHR_external_memory_{fd,dma_buf}< / li >
< li > turnip: fix VkClearValue packing< / li >
< li > turnip: preliminary support for fences< / li >
< li > turnip: respect color attachment formats< / li >
< li > turnip: mark IBs for dumping< / li >
< li > turnip: use 32-bit offset in tu_cs_entry< / li >
< li > turnip: more/better asserts for tu_cs< / li >
< li > turnip: add tu_cs_discard_entries< / li >
< li > turnip: tu_cs_emit_array< / li >
< li > turnip: fix tu_cs sub-streams< / li >
< li > turnip: simplify tu_cs sub-streams usage< / li >
< li > turnip: create a less dummy pipeline< / li >
< li > turnip: parse VkPipelineDynamicStateCreateInfo< / li >
< li > turnip: parse VkPipelineInputAssemblyStateCreateInfo< / li >
< li > turnip: parse VkPipelineViewportStateCreateInfo< / li >
< li > turnip: parse VkPipelineRasterizationStateCreateInfo< / li >
< li > turnip: parse VkPipelineDepthStencilStateCreateInfo< / li >
< li > turnip: parse VkPipeline{Multisample,ColorBlend}StateCreateInfo< / li >
< li > turnip: preliminary support for shader modules< / li >
< li > turnip: compile VkPipelineShaderStageCreateInfo< / li >
< li > turnip: parse VkPipelineShaderStageCreateInfo< / li >
< li > turnip: parse VkPipelineVertexInputStateCreateInfo< / li >
< li > turnip: add draw_cs to tu_cmd_buffer< / li >
< li > turnip: preliminary support for draw state binding< / li >
< li > turnip: preliminary support for tu_CmdDraw< / li >
< li > turnip: guard -Dvulkan-driver=freedreno< / li >
< li > turnip: preliminary support for tu_GetImageSubresourceLayout< / li >
< li > turnip: preliminary support for Wayland WSI< / li >
< li > vulkan/wsi: move modifier array into wsi_wl_swapchain< / li >
< li > vulkan/wsi: create wl_drm wrapper as needed< / li >
< li > vulkan/wsi: refactor drm_handle_format< / li >
< li > vulkan/wsi: add wsi_wl_display_drm< / li >
< li > vulkan/wsi: add wsi_wl_display_dmabuf< / li >
< li > vulkan/wsi: make wl_drm optional< / li >
< li > virgl: handle fence_server_sync in winsys< / li >
< li > virgl: hide fence internals from the driver< / li >
< li > virgl: introduce virgl_drm_fence< / li >
< li > virgl: fix fence fd version check< / li >
< li > virgl: clear vertex_array_dirty< / li >
< li > virgl: skip empty cmdbufs< / li >
< / ul >
< p > Chris Forbes (3):< / p >
< ul >
< li > glsl: add scaffolding for EXT_gpu_shader4< / li >
< li > glsl: enable noperspective|flat|centroid for EXT_gpu_shader4< / li >
< li > glsl: enable types for EXT_gpu_shader4< / li >
< / ul >
< p > Chris Wilson (19):< / p >
< ul >
< li > i965: Assert the execobject handles match for this device< / li >
< li > iris: fix import from dri2/3< / li >
< li > iris: IndexFormat = size/2< / li >
< li > iris: Set resource modifier on handle< / li >
< li > iris: Wrap userptr for creating bo< / li >
< li > iris: AMD_pinned_memory< / li >
< li > iris: Record reusability of bo on construction< / li >
< li > iris: fix memzone_for_address since multibinder changes< / li >
< li > iris: Tidy exporting the flink handle< / li >
< li > iris: Fix assigning the output handle for exporting for KMS< / li >
< li > iris: Merge two walks of the exec_bos list< / li >
< li > iris: Tag each submitted batch with a syncobj< / li >
< li > iris: Add fence support using drm_syncobj< / li >
< li > iris: Wire up EGL_IMG_context_priority< / li >
< li > iris: Use PIPE_BUFFER_STAGING for the query objects< / li >
< li > iris: Use coherent allocation for PIPE_RESOURCE_STAGING< / li >
< li > iris: Use streaming loads to read from tiled surfaces< / li >
< li > iris: Push heavy memchecker code to DEBUG< / li >
< li > iris: Adapt to variable ppGTT size< / li >
< / ul >
< p > Christian Gmeiner (12):< / p >
< ul >
< li > etnaviv: rs: mark used src resource as read from< / li >
< li > etnaviv: blt: mark used src resource as read from< / li >
< li > etnaviv: implement ETC2 block patching for HALTI0< / li >
< li > etnaviv: keep track of mapped bo address< / li >
< li > etnaviv: hook-up etc2 patching< / li >
< li > etnaviv: enable ETC2 texture compression support for HALTI0 GPUs< / li >
< li > etnaviv: fix resource usage tracking across different pipe_context's< / li >
< li > etnaviv: fix compile warnings< / li >
< li > st/dri: allow direct UYVY import< / li >
< li > etnaviv: shrink struct etna_3d_state< / li >
< li > nir: add lower_ftrunc< / li >
< li > etnaviv: use the correct uniform dirty bits< / li >
< / ul >
< p > Chuck Atkins (1):< / p >
< ul >
< li > meson: Fix missing glproto dependency for gallium-glx< / li >
< / ul >
< p > Connor Abbott (6):< / p >
< ul >
< li > nir/serialize: Prevent writing uninitialized state_slot data< / li >
< li > nir: Add a stripping pass for improved cacheability< / li >
< li > radeonsi/nir: Use nir stripping pass< / li >
< li > nir/search: Add automaton-based pre-searching< / li >
< li > nir/search: Add debugging code to dump the pattern matched< / li >
< li > nir/algebraic: Don't emit empty initializers for MSVC< / li >
< / ul >
< p > Daniel Schürmann (2):< / p >
< ul >
< li > nir: Define shifts according to SM5 specification.< / li >
< li > nir: Use SM5 properties to optimize shift(a@32, iand(31, b))< / li >
< / ul >
< p > Daniel Stone (2):< / p >
< ul >
< li > panfrost: Properly align stride< / li >
< li > vulkan/wsi/wayland: Respect non-blocking AcquireNextImage< / li >
< / ul >
< p > Danylo Piliaiev (13):< / p >
< ul >
< li > anv: Handle VK_ATTACHMENT_UNUSED in colorAttachment< / li >
< li > radv: Handle VK_ATTACHMENT_UNUSED in CmdClearAttachment< / li >
< li > anv: Fix VK_EXT_transform_feedback working with varyings packed in PSIZ< / li >
< li > anv: Fix destroying descriptor sets when pool gets reset< / li >
< li > anv: Treat zero size XFB buffer as disabled< / li >
< li > glsl: Cross validate variable's invariance by explicit invariance only< / li >
< li > i965,iris,anv: Make alpha to coverage work with sample mask< / li >
< li > intel/fs: Make alpha test work with MRT and sample mask< / li >
< li > st/mesa: Fix GL_MAP_COLOR with glDrawPixels GL_COLOR_INDEX< / li >
< li > iris: Fix assert when using vertex attrib without buffer binding< / li >
< li > intel/compiler: Do not reswizzle dst if instruction writes to flag register< / li >
< li > drirc: Add workaround for Epic Games Launcher< / li >
< li > anv: Do not emulate texture swizzle for INPUT_ATTACHMENT, STORAGE_IMAGE< / li >
< / ul >
< p > Dave Airlie (63):< / p >
< ul >
< li > virgl: enable elapsed time queries< / li >
< li > virgl: ARB_query_buffer_object support< / li >
< li > docs: update qbo support for virgl< / li >
< li > glsl: glsl to nir fix uninit class member.< / li >
< li > radv/llvm: initialise passes member.< / li >
< li > radv: remove alloc parameter from pipeline init< / li >
< li > iris: fix some hangs around null framebuffers< / li >
< li > iris: fix crash in sparse vertex array< / li >
< li > iris: add initial transform feedback overflow query paths (V3)< / li >
< li > iris: fix cube texture view< / li >
< li > iris: execute compute related query on compute batch.< / li >
< li > iris: iris add load register reg32/64< / li >
< li > iris: add conditional render support< / li >
< li > iris: fix gpu calcs for timestamp queries< / li >
< li > iris/WIP: add broadwell support< / li >
< li > iris: limit gen8 to 8 samples< / li >
< li > iris: setup gen8 caps< / li >
< li > iris: add fs invocations query workaround for broadwell< / li >
< li > iris: handle qbo fragment shader invocation workaround< / li >
< li > st/mesa: add support for lowering fp64/int64 for nir drivers< / li >
< li > softpipe: fix texture view crashes< / li >
< li > nir/spirv: don't use bare types, remove assert in split vars for testing< / li >
< li > nir/deref: remove casts of casts which are likely redundant (v3)< / li >
< li > softpipe: fix 32-bit bitfield extract< / li >
< li > softpipe: handle 32-bit bitfield inserts< / li >
< li > softpipe: remove shadow_ref assert.< / li >
< li > softpipe: fix integer texture swizzling for 1 vs 1.0f< / li >
< li > nir/split_vars: fixup some more explicit_stride related issues.< / li >
< li > draw: bail instead of assert on instance count (v2)< / li >
< li > draw/gs: fix point size outputs from geometry shader.< / li >
< li > draw/vs: partly fix basevertex/vertex id< / li >
< li > softpipe: fix clears to only clear specified color buffers.< / li >
< li > softpipe/draw: fix vertex id in soft paths.< / li >
< li > softpipe: add indirect store buffer/image unit< / li >
< li > nir/deref: fix struct wrapper casts. (v3)< / li >
< li > nir: use proper array sizing define for vectors< / li >
< li > intel/compiler: use defined size for vector components< / li >
< li > iris: avoid use after free in shader destruction< / li >
< li > ddebug: add compute functions to help hang detection< / li >
< li > draw: add stream member to stats callback< / li >
< li > tgsi: add support for geometry shader streams.< / li >
< li > softpipe: add support for indexed queries.< / li >
< li > draw: add support to tgsi paths for geometry streams. (v2)< / li >
< li > softpipe: add support for vertex streams (v2)< / li >
< li > virgl: add support for missing command buffer binding.< / li >
< li > virgl: add support for ARB_multi_draw_indirect< / li >
< li > virgl: add support for ARB_indirect_parameters< / li >
< li > draw: fix undefined shift of (1 < < 31)< / li >
< li > swrast: fix undefined shift of 1 < < 31< / li >
< li > llvmpipe: fix undefined shift 1 < < 31.< / li >
< li > virgl/drm: cleanup buffer from handle creation (v2)< / li >
< li > virgl/drm: handle flink name better.< / li >
< li > virgl/drm: insert correct handles into the table. (v3)< / li >
< li > intel/compiler: fix uninit non-static variable. (v2)< / li >
< li > nir: fix bit_size in lower indirect derefs.< / li >
< li > r600: reset tex array override even when no view bound< / li >
< li > spirv: fix SpvOpBitSize return value.< / li >
< li > nir: fix lower vars to ssa for larger vector sizes.< / li >
< li > util/tests: add basic unit tests for bitset< / li >
< li > util/bitset: fix bitset range mask calculations.< / li >
< li > kmsro: add _dri.so to two of the kmsro drivers.< / li >
< li > glsl: init packed in more constructors.< / li >
< li > Revert "mesa: unreference current winsys buffers when unbinding winsys buffers"< / li >
< / ul >
< p > David Riley (3):< / p >
< ul >
< li > virgl: Store mapped hw resource with transfer object.< / li >
< li > virgl: Allow transfer queue entries to be found and extended.< / li >
< li > virgl: Re-use and extend queue transfers for intersecting buffer subdatas.< / li >
< / ul >
< p > David Shao (1):< / p >
< ul >
< li > meson: ensure that xmlpool_options.h is generated for gallium targets that need it< / li >
< / ul >
< p > Deepak Rawat (2):< / p >
< ul >
< li > winsys/drm: Fix out of scope variable usage< / li >
< li > winsys/svga/drm: Fix 32-bit RPCI send message< / li >
< / ul >
< p > Dominik Drees (1):< / p >
< ul >
< li > Add no_aos_sampling GALLIVM_PERF option< / li >
< / ul >
< p > Drew Davenport (1):< / p >
< ul >
< li > util: Don't block SIGSYS for new threads< / li >
< / ul >
< p > Dylan Baker (40):< / p >
< ul >
< li > bump version for 19.0 branch< / li >
< li > docs: Add relnotes stub for 19.1< / li >
< li > gallium: wrap u_screen in extern "C" for c++< / li >
< li > automake: Add --enable-autotools to distcheck flags< / li >
< li > android,autotools,i965: Fix location of float64_glsl.h< / li >
< li > meson: remove build_by_default : true< / li >
< li > meson: fix style in intel/tools< / li >
< li > meson: remove -std=c++11 from intel/tools< / li >
< li > get-pick-list: Add --pretty=medium to the arguments for Cc patches< / li >
< li > meson: Add dependency on genxml to anvil< / li >
< li > meson/iris: Use current coding style< / li >
< li > docs: Add release notes for 19.0.0< / li >
< li > docs: Add SHA256 sums for 19.0.0< / li >
< li > docs: update calendar, add news item, and link release notes for 19.0.0< / li >
< li > bin/install_megadrivers.py: Correctly handle DESTDIR=''< / li >
< li > bin/install_megadrivers.py: Fix regression for set DESTDIR< / li >
< li > docs: Add release notes for 19.0.1< / li >
< li > docs: Add SHA256 sums for mesa 19.0.1< / li >
< li > docs: update calendar, add news item and link release notes for 19.0.1< / li >
< li > meson: Error if LLVM doesn't have rtti when building clover< / li >
< li > meson: Error if LLVM is turned off but clover it turned on< / li >
< li > docs: Add release notes for 19.0.2< / li >
< li > docs: Add sha256 sums for 19.0.2< / li >
< li > docs: update calendar, and news item and link release notes for 19.0.2< / li >
< li > Delete autotools< / li >
< li > docs: drop most autoconf references< / li >
< li > ci: Delete autotools build jobs< / li >
< li > docs: add relnotes for 19.0.3< / li >
< li > docs: Add SHA256 sums for mesa 19.0.3< / li >
< li > docs: update calendar, and news item and link release notes for 19.0.3< / li >
< li > meson: always define libglapi< / li >
< li > glsl: fix general_ir_test with mingw< / li >
< li > meson: switch gles1 and gles2 to auto options< / li >
< li > meson: Make shader-cache a trillean instead of boolean< / li >
< li > meson: make nm binary optional< / li >
< li > util/tests: Use define instead of VLA< / li >
< li > glsl/tests: define ssize_t on windows< / li >
< li > tests/vma: fix build with MSVC< / li >
< li > meson: Don't build glsl cache_test when shader cache is disabled< / li >
< li > meson: Force the use of config-tool for llvm< / li >
< / ul >
< p > Eduardo Lima Mitev (5):< / p >
< ul >
< li > freedreno/a6xx: Silence compiler warnings< / li >
< li > nir: Add ir3-specific version of most SSBO intrinsics< / li >
< li > ir3/nir: Add a new pass 'ir3_nir_lower_io_offsets'< / li >
< li > ir3/compiler: Enable lower_io_offsets pass and handle new SSBO intrinsics< / li >
< li > ir3/lower_io_offsets: Try propagate SSBO's SHR into a previous shift instruction< / li >
< / ul >
< p > El Christianito (1):< / p >
< ul >
< li > drirc: add Budgie WM to adaptive-sync blacklist< / li >
< / ul >
< p > Eleni Maria Stea (6):< / p >
< ul >
< li > i965: Faking the ETC2 compression on Gen < 8 GPUs using two miptrees.< / li >
< li > i965: Fixed the CopyImageSubData for ETC2 on Gen < 8< / li >
< li > i965: Enabled the OES_copy_image extension on Gen 7 GPUs< / li >
< li > i965: Removed the field etc_format from the struct intel_mipmap_tree< / li >
< li > i965: fixed clamping in set_scissor_bits when the y is flipped< / li >
< li > radv: consider MESA_VK_VERSION_OVERRIDE when setting the api version< / li >
< / ul >
< p > Elie Tournier (3):< / p >
< ul >
< li > virgl: Add a caps to advertise GLES backend< / li >
< li > virgl: Set PIPE_CAP_DOUBLES when running on GLES This is a lie but no known app use fp64.< / li >
< li > virgl: Return an error if we use fp64 on top of GLES< / li >
< / ul >
< p > Emil Velikov (30):< / p >
< ul >
< li > vc4: Declare the last cpu pointer as being modified in NEON asm.< / li >
< li > docs: add release notes for 18.3.3< / li >
< li > docs: add sha256 checksums for 18.3.3< / li >
< li > docs: update calendar, add news item and link release notes for 18.3.3< / li >
< li > anv: wire up the state_pool_padding test< / li >
< li > docs: add release notes for 18.3.4< / li >
< li > docs: add sha256 checksums for 18.3.4< / li >
< li > docs: update calendar, add news item and link release notes for 18.3.4< / li >
< li > egl/dri: de-duplicate dri2_load_driver*< / li >
< li > meson: egl: correctly manage loader/xmlconfig< / li >
< li > loader: use loader_open_device() to handle O_CLOEXEC< / li >
< li > egl/android: bump the number of drmDevices to 64< / li >
< li > docs: mention "Allow commits from members who can merge..."< / li >
< li > egl/sl: split out swrast probe into separate function< / li >
< li > egl/sl: use drmDevice API to enumerate available devices< / li >
< li > egl/sl: use kms_swrast with vgem instead of a random GPU< / li >
< li > docs: add release notes for 18.3.5< / li >
< li > docs: add sha256 checksums for 18.3.5< / li >
< li > docs: update calendar, add news item and link release notes for 18.3.5< / li >
< li > docs: add release notes for 18.3.6< / li >
< li > docs: add sha256 checksums for 18.3.6< / li >
< li > docs: update calendar, add news item and link release notes for 18.3.6< / li >
< li > turnip: drop dead close(master_fd)< / li >
< li > vulkan/wsi: check if the display_fd given is master< / li >
< li > vulkan/wsi: don't use DUMB_CLOSE for normal GEM handles< / li >
< li > llvmpipe: add lp_fence_timedwait() helper< / li >
< li > llvmpipe: correctly handle waiting in llvmpipe_fence_finish< / li >
< li > egl/dri: flesh out and use dri2_create_drawable()< / li >
< li > mapi: add static_date offset to MaxShaderCompilerThreadsKHR< / li >
< li > mapi: correctly handle the full offset table< / li >
< / ul >
< p > Emmanuel Gil Peyrot (1):< / p >
< ul >
< li > docs: make bugs.html easier to find< / li >
< / ul >
< p > Eric Anholt (121):< / p >
< ul >
< li > v3d: Always enable the NEON utile load/store code.< / li >
< li > v3d: Fix a release build set-but-unused compiler warning.< / li >
< li > mesa: Skip partial InvalidateFramebuffer of packed depth/stencil.< / li >
< li > v3d: Fix image_load_store clamping of signed integer stores.< / li >
< li > nir: Move V3D's "the shader was TGSI, ignore FS output types" flag to NIR.< / li >
< li > v3d: Fix precompile of FRAG_RESULT_DATA1 and higher outputs.< / li >
< li > v3d: Store the actual mask of color buffers present in the key.< / li >
< li > v3d: Fix dumping of shaders with alpha test.< / li >
< li > v3d: Fix pack/unpack of VFPACK operand unpacks.< / li >
< li > v3d: Fix input packing of .l for rounding/fdx/fdy.< / li >
< li > v3d: Fix copy-propagation of input unpacks.< / li >
< li > v3d: Whitespace consistency fix.< / li >
< li > nir: Move panfrost's isign lowering to nir_opt_algebraic.< / li >
< li > v3d: Use the NIR lowering for isign instead of rolling our own.< / li >
< li > intel: Use the NIR lowering for isign.< / li >
< li > freedreno: Use the NIR lowering for isign.< / li >
< li > v3d: Clear the GMP on initialization of the simulator.< / li >
< li > v3d: Sync indirect draws on the last rendering.< / li >
< li > v3d: Use the early_fragment_tests flag for the shader's disable-EZ field.< / li >
< li > v3d: Fix incorrect flagging of ldtmu as writing r4 on v3d 4.x.< / li >
< li > v3d: Drop a perf note about merging unpack_half_*, which has been implemented.< / li >
< li > v3d: Drop our hand-lowered nir_op_ffract.< / li >
< li > v3d: Add a helper function for getting a nop register.< / li >
< li > v3d: Refactor bcsel and if condition handling.< / li >
< li > v3d: Do bool-to-cond for discard_if as well.< / li >
< li > v3d: Kill off vir_PF(), which is hard to use right.< / li >
< li > v3d: Fix f2b32 behavior.< / li >
< li > v3d: Fix the check for "is the last thrsw inside control flow"< / li >
< li > v3d: Add a function to describe what the c-> execute.file check means.< / li >
< li > v3d: Stop tracking num_inputs for VPM loads.< / li >
< li > v3d: Delay emitting ldvpm on V3D 4.x until it's actually used.< / li >
< li > v3d: Emit a simpler negate for the iabs implementation.< / li >
< li > v3d: Move i2b and f2b support into emit_comparison.< / li >
< li > kmsro: Add the rest of the current set of tinydrm drivers.< / li >
< li > nir: Just return when asked to rewrite uses of an SSA def to itself.< / li >
< li > v3d: Fix vir_is_raw_mov() for input unpacks.< / li >
< li > v3d: Dump the VIR after register spilling if we were forced to.< / li >
< li > v3d: Rematerialize MOVs of uniforms instead of spilling them.< / li >
< li > v3d: Fix build of NEON code with Mesa's cflags not targeting NEON.< / li >
< li > v3d: Restrict live intervals to the blocks reachable from any def.< / li >
< li > v3d: Stop treating exec masking specially.< / li >
< li > nir: Improve printing of load_input/store_output variable names.< / li >
< li > v3d: Translate f2i(fround_even) as FTOIN.< / li >
< li > v3d: Move the stores for fixed function VS output reads into NIR.< / li >
< li > v3d: Fix temporary leaks of temp_registers and when spilling.< / li >
< li > v3d: Do uniform rematerialization spilling before dropping threadcount< / li >
< li > v3d: Switch implicit uniforms over to being any qinst-> uniform != ~0.< / li >
< li > v3d: Add support for vir-to-qpu of ldunif instructions to a temp.< / li >
< li > v3d: Drop the old class bits splitting up the accumulators.< / li >
< li > v3d: Add support for register-allocating a ldunif to a QFILE_TEMP.< / li >
< li > v3d: Use ldunif instructions for uniforms.< / li >
< li > v3d: Eliminate the TLB and TLBU files.< / li >
< li > v3d: Drop the V3D 3.x vpm read dead code elimination.< / li >
< li > v3d: Include a count of register pressure in the RA failure dumps.< / li >
< li > st/dri: Set the PIPE_BIND_SHARED flag on create_image_with_modifiers.< / li >
< li > util: Add a DAG datastructure.< / li >
< li > vc4: Switch over to using the DAG datastructure for QIR scheduling.< / li >
< li > v3d: Reuse list_for_each_entry_rev().< / li >
< li > vc4: Reuse list_for_each_entry_rev().< / li >
< li > v3d: Use the DAG datastructure for QPU instruction scheduling.< / li >
< li > vc4: Switch the post-RA scheduler over to the DAG datastructure.< / li >
< li > v3d: Disable PIPE_CAP_BLIT_BASED_TEXTURE_TRANSFER.< / li >
< li > v3d: Fix leak of the mem_ctx after the DAG refactor.< / li >
< li > v3d: Fix leak of the renderonly struct on screen destruction.< / li >
< li > mesa/st: Make sure that prog_to_nir NIR gets freed.< / li >
< li > mesa/st: Fix leaks of TGSI tokens in VP variants.< / li >
< li > v3d: Always lay out shared tiled buffers with UIF_TOP set.< / li >
< li > v3d: Allow the UIF modifier with renderonly.< / li >
< li > v3d: Expose the dma-buf modifiers query.< / li >
< li > v3d: Rename v3d_tmu_config_data to v3d_unit_data.< / li >
< li > v3d: Move constant offsets to UBO addresses into the main uniform stream.< / li >
< li > v3d: Upload all of UBO[0] if any indirect load occurs.< / li >
< li > v3d: Remove some dead members of struct v3d_compile.< / li >
< li > egl: Add a 565 pbuffer-only EGL config under X11.< / li >
< li > dri3: Return the current swap interval from glXGetSwapIntervalMESA().< / li >
< li > v3d: Add support for handling OOM signals from the simulator.< / li >
< li > v3d: Bump the maximum texture size to 4k for V3D 4.x.< / li >
< li > v3d: Don't try to use the TFU blit path if a scissor is enabled.< / li >
< li > v3d: Add some more new packets for V3D 4.x.< / li >
< li > st: Lower uniforms in st in the !PIPE_CAP_PACKED_UNIFORMS case as well.< / li >
< li > vc4: Don't forget to set the range when scalarizing our uniforms.< / li >
< li > vc4: Split UBO0 and UBO1 address uniform handling.< / li >
< li > vc4: Upload CS/VS UBO uniforms together.< / li >
< li > v3d: Add an optimization pass for redundant flags updates.< / li >
< li > nir: Drop comments about the constant_index slots for load/stores.< / li >
< li > nir: Drop remaining references to const_index in favor of the call to use.< / li >
< li > nir: Add a comment about how intrinsic definitions work.< / li >
< li > v3d: Add and use a define for the number of channels in a QPU invocation.< / li >
< li > v3d: Drop a note for the future about PIPE_CAP_PACKED_UNIFORMS.< / li >
< li > v3d: Include the number of max temps used in the shader-db output.< / li >
< li > v3d: Replace the old shader-db env var output with the ARB_debug_output.< / li >
< li > v3d: Add Compute Shader compilation support.< / li >
< li > v3d: Add missing base offset to CS shared memory accesses.< / li >
< li > v3d: Add missing dumping for the spill offset/size uniforms.< / li >
< li > v3d: Detect the correct number of QPUs and use it to fix the spill size.< / li >
< li > v3d: Use the new lower_to_scratch implementation for indirects on temps.< / li >
< li > v3d: Only look up the 3rd texture gather offset for non-arrays.< / li >
< li > v3d: Always set up the qregs for CSD payload.< / li >
< li > v3d: Fix an invalid reuse of flags generation from before a thrsw.< / li >
< li > v3d: Fix atomic cmpxchg in shaders on hardware.< / li >
< li > nir: Fix deref offset calculation for structs.< / li >
< li > nir: Use the nir_builder _imm helpers in setting up deref offsets.< / li >
< li > gallium: Remove the pool pipebuffer manager.< / li >
< li > gallium: Remove the ondemand pipebuffer manager.< / li >
< li > gallium: Remove the "alt" pipebuffer manager interface.< / li >
< li > gallium: Remove the malloc pipebuffer manager.< / li >
< li > st/mesa: Don't set atomic counter size != 0 if MAX_SHADER_BUFFERS == 0.< / li >
< li > v3d: Disable SSBOs and atomic counters on vertex shaders.< / li >
< li > v3d: Fill in the ignored segment size fields to appease new simulator.< / li >
< li > v3d: Apply the GFXH-930 workaround to the case where the VS loads attrs.< / li >
< li > v3d: Assert that we do request the normal texturing return data.< / li >
< li > v3d: Use _mesa_hash_table_remove_key() where appropriate.< / li >
< li > vc4: Use _mesa_hash_table_remove_key() where appropriate.< / li >
< li > v3d: Add a note about i/o indirection for future performance work.< / li >
< li > v3d: Don't try to update the shadow texture for separate stencil.< / li >
< li > Revert "v3d: Disable PIPE_CAP_BLIT_BASED_TEXTURE_TRANSFER."< / li >
< li > v3d: Re-add support for memory_barrier_shared.< / li >
< li > v3d: Fix detection of the last ldtmu before a new TMU op.< / li >
< li > v3d: Fix detection of TMU write sequences in register spilling.< / li >
< li > kmsro: Add support for V3D.< / li >
< li > vc4: Fall back to renderonly if the vc4 driver doesn't have v3d.< / li >
< / ul >
< p > Eric Engestrom (142):< / p >
< ul >
< li > wsi/display: add comment< / li >
< li > egl: use coherent variable names< / li >
< li > gitlab-ci: add ubuntu container< / li >
< li > gitlab-ci: add a meson vulkan build< / li >
< li > gitlab-ci: add a make vulkan build< / li >
< li > gitlab-ci: add a scons no-llvm build< / li >
< li > gitlab-ci: add scons llvm 3.5 build< / li >
< li > gitlab-ci: add scons SWR build< / li >
< li > gitlab-ci: add meson loader/classic DRI build< / li >
< li > gitlab-ci: add meson gallium SWR build< / li >
< li > gitlab-ci: add meson gallium RadeonSI build< / li >
< li > gitlab-ci: add meson gallium "other drivers" build< / li >
< li > gitlab-ci: add meson gallium ST Clover (LLVM 5.0) build< / li >
< li > gitlab-ci: add meson gallium ST Clover (LLVM 6.0) build< / li >
< li > gitlab-ci: add meson gallium ST Clover (LLVM 7.0) build< / li >
< li > gitlab-ci: add meson gallium ST "Other" build< / li >
< li > gitlab-ci: add make loaders/classic DRI build< / li >
< li > gitlab-ci: add make Gallium Drivers SWR build< / li >
< li > gitlab-ci: add make Gallium Drivers RadeonSI build< / li >
< li > gitlab-ci: add make Gallium Drivers "Other" build< / li >
< li > gitlab-ci: add make Gallium ST Clover LLVM-3.9 build< / li >
< li > gitlab-ci: add make Gallium ST Clover LLVM-4.0 build< / li >
< li > gitlab-ci: add make Gallium ST Clover LLVM-5.0 build< / li >
< li > gitlab-ci: add make Gallium ST Clover LLVM-6.0 build< / li >
< li > gitlab-ci: add make Gallium ST Clover LLVM-7 build< / li >
< li > gitlab-ci: add make Gallium ST Other build< / li >
< li > travis: remove unused linux code path< / li >
< li > travis: remove unused scons code path< / li >
< li > gitlab-ci: add meson glvnd build< / li >
< li > xvmc: fix string comparison< / li >
< li > xvmc: fix string comparison< / li >
< li > meson: add script to print the options before configuring a builddir< / li >
< li > driconf: drop unused macro< / li >
< li > travis: fix osx make build< / li >
< li > gitlab-ci: workaround docker bug for users with uppercase characters< / li >
< li > wsi: query the ICD's max dimensions instead of hard-coding them< / li >
< li > gitlab-ci: limit ninja to 4 threads max< / li >
< li > drm-uapi/README: remove explicit list of driver names< / li >
< li > drm-uapi: use local files, not system libdrm< / li >
< li > gbm: drop duplicate #defines< / li >
< li > st/dri: drop duplicate #define< / li >
< li > etnaviv: drop duplicate #define< / li >
< li > anv/tests: compile to something sensible in release builds< / li >
< li > util/tests: compile to something sensible in release builds< / li >
< li > gitlab-ci: use ccache to speed up builds< / li >
< li > tegra/meson: add missing dep_libdrm< / li >
< li > tegra/autotools: add missing libdrm cflags< / li >
< li > gitlab-ci: limit the automatic CI to master and MRs< / li >
< li > gitlab-ci: automatically run the CI on pushes to `ci/*` branches< / li >
< li > anv: sort extensions alphabetically< / li >
< li > anv: sort vendors extensions after KHR and EXT< / li >
< li > anv: make sure the extensions stay sorted< / li >
< li > anv: drop unused imports< / li >
< li > anv: use anv_shader_bin_write_to_blob()'s return value< / li >
< li > gitlab-ci: always run the containers build< / li >
< li > dri_interface: add missing #include< / li >
< li > driinfo: add DTD to allow the xml to be validated< / li >
< li > meson/swr: replace hard-coded path with current_build_dir()< / li >
< li > egl/android: replace magic 0=CbCr,1=CrCb with simple enum< / li >
< li > vulkan: use VkBase{In,Out}Structure instead of a custom struct< / li >
< li > driconf: add DTD to allow the drirc xml (00-mesa-defaults.conf) to be validated< / li >
< li > gitlab-ci: install xmllint to validate 00-mesa-defaults.conf< / li >
< li > anv: simplify chained comparison< / li >
< li > anv: drop unused parameter< / li >
< li > anv: remove spaces around kwargs assignment< / li >
< li > anv: fix typo< / li >
< li > Revert "swr/rast: Archrast codegen updates"< / li >
< li > meson: avoid going back up the tree with include_directories()< / li >
< li > anv: use the platform defines in vk.xml instead of hard-coding them< / li >
< li > radv: use the platform defines in vk.xml instead of hard-coding them< / li >
< li > util: #define PATH_MAX when undefined (eg. Hurd)< / li >
< li > vulkan: import missing file from Khronos< / li >
< li > egl: fix libdrm-less builds< / li >
< li > vulkan: import vk_layer.h from Khronos< / li >
< li > gitlab-ci: drop job prefixes< / li >
< li > meson: fix with_dri2 definition for GNU Hurd< / li >
< li > meson: remove unused include_directories(vulkan)< / li >
< li > vulkan/util: use the platform defines in vk.xml instead of hard-coding them< / li >
< li > vulkan/overlay: fix missing var rename in previous commit< / li >
< li > meson: don't build libGLES*.so with GLVND< / li >
< li > autotools: don't build libGLES*.so with GLVND< / li >
< li > travis: fix meson build by letting `auto` do its job< / li >
< li > travis: drop unused vars< / li >
< li > travis: clean up< / li >
< li > gitlab-ci: only build the default (=latest) and oldest llvm versions< / li >
< li > gitlab-ci: autotools needs to be told which llvm version to use< / li >
< li > r600: cast pointer to expected type< / li >
< li > build: make passing an incorrect pointer type a hard error< / li >
< li > gitlab-ci: fix llvm version (7 doesn't have a ".0")< / li >
< li > hgl/meson: drop unused include directory< / li >
< li > glx/meson: use full include path for dri_interface.h< / li >
< li > android: fix missing backspace for line continuation< / li >
< li > panfrost: fix tgsi_to_nir() call< / li >
< li > panfrost: move #include to fix compilation< / li >
< li > gitlab-ci: add panfrost to the gallium drivers build< / li >
< li > wsi: deduplicate get_current_time() functions between display and x11< / li >
< li > wsi/display: s/#if/#ifdef/ to fix -Wundef< / li >
< li > wsi/wayland: fix pointer casting warning on 32bit< / li >
< li > wsi/x11: use WSI_FROM_HANDLE() instead of pointer casts< / li >
< li > turnip: use the platform defines in vk.xml instead of hard-coding them< / li >
< li > travis: fix osx meson build< / li >
< li > nir: const `nir_call_instr::callee`< / li >
< li > gitlab-ci: add clang build< / li >
< li > gitlab-ci: drop most autotools builds< / li >
< li > util/disk_cache: close fd in the fallback path< / li >
< li > egl: hide entrypoints that shouldn't be exported when using glvnd< / li >
< li > meson: strip rpath from megadrivers< / li >
< li > gallium/hud: fix memory leaks< / li >
< li > gallium/hud: prevent buffer overflow< / li >
< li > gallium/hud: fix rounding error in nic bps computation< / li >
< li > simplify LLVM version string printing< / li >
< li > util/process: document memory leak< / li >
< li > vk/util: remove unneeded array index< / li >
< li > bin: drop unused import from install_megadrivers.py< / li >
< li > meson: remove meson-created megadrivers symlinks< / li >
< li > gitlab-ci: build gallium extra hud< / li >
< li > gitlab-ci: add lima to the build< / li >
< li > delete autotools .gitignore files< / li >
< li > delete autotools input files< / li >
< li > docs: remove unsupported GL function name mangling< / li >
< li > docs: drop autotools python information< / li >
< li > docs: replace autotools intructions with meson equivalent< / li >
< li > docs: use past tense when talking about autotools< / li >
< li > docs: haiku can be built using meson< / li >
< li > egl: fixup autotools-specific wording< / li >
< li > util: add os_read_file() helper< / li >
< li > anv: add support for VK_EXT_memory_budget< / li >
< li > radv: update to use the new features struct names< / li >
< li > turnip: update to use the new features struct names< / li >
< li > gitlab-ci: build vulkan drivers in clang build< / li >
< li > util: move #include out of #if linux< / li >
< li > wsi/wayland: document lack of vkAcquireNextImageKHR timeout support< / li >
< li > egl: hard-code destroy function instead of passing it around as a pointer< / li >
< li > gitlab-ci: add scons windows build using mingw< / li >
< li > gitlab-ci: merge several meson jobs< / li >
< li > gitlab-ci: meson-gallium-radeonsi was a subset of meson-gallium-clover-llvm< / li >
< li > gitlab-ci: simplify meson job names< / li >
< li > gitlab-ci: merge meson-glvnd into meson-swr< / li >
< li > travis: fix syntax, and drop unused stuff< / li >
< li > util/os_file: always use the 'grow' mechanism< / li >
< li > meson: expose glapi through osmesa< / li >
< li > util/os_file: actually return the error read() gave us< / li >
< / ul >
< p > Erico Nunes (5):< / p >
< ul >
< li > lima/ppir: support ppir_op_ceil< / li >
< li > nir/algebraic: add lowering for fsign< / li >
< li > lima: enable nir fsign lowering in ppir< / li >
< li > lima/gpir: add limit of max 512 instructions< / li >
< li > lima/ppir: support nir_op_ftrunc< / li >
< / ul >
< p > Erik Faye-Lund (79):< / p >
< ul >
< li > mesa: expose NV_conditional_render on GLES< / li >
< li > st/mesa: remove unused header-file< / li >
< li > swr/codegen: fix autotools build< / li >
< li > virgl: remove unused variables< / li >
< li > virgl: remove unused variable< / li >
< li > virgl: remove unused variable< / li >
< li > virgl: remove unused variable< / li >
< li > virgl: do not allow compressed formats for buffers< / li >
< li > virgl: stricter usage of compressed 3d textures< / li >
< li > virgl: also destroy all read-transfers< / li >
< li > virgl: use debug_printf instead of fprintf< / li >
< li > virgl: unsigned int -> unsigned< / li >
< li > virgl: only warn about unchecked flags< / li >
< li > virgl: do not warn about display-target binding< / li >
< li > virgl: use debug_printf instead of fprintf< / li >
< li > virgl: remove pointless transfer-counter< / li >
< li > virgl: tmp_resource -> templ< / li >
< li > virgl: track full virgl_resource instead of just virgl_hw_res< / li >
< li > virgl: simplify virgl_texture_transfer_unmap logic< / li >
< li > virgl: make unmap queuing a bit more straight-forward< / li >
< li > virgl: check for readback on correct resource< / li >
< li > virgl: wait for the right resource< / li >
< li > virgl: return error if allocating resolve_tmp fails< / li >
< li > virgl: rewrite core of virgl_texture_transfer_map< / li >
< li > virgl: use pipe_box for blit dst-rect< / li >
< li > virgl: support write-back with staged transfers< / li >
< li > virgl: make sure bind is set for non-buffers< / li >
< li > gallium/util: support translating between uint and sint formats< / li >
< li > virgl: get readback-formats from host< / li >
< li > virgl: only blit if resource is read< / li >
< li > virgl: do color-conversion during when mapping transfer< / li >
< li > virgl: document potentially failing blit< / li >
< li > mesa/st: remove impossible error-check< / li >
< li > gallium/u_vbuf: support NULL-resources< / li >
< li > i915: support NULL-resources< / li >
< li > nouveau: support NULL-resources< / li >
< li > swr: support NULL-resources< / li >
< li > mesa/st: accept NULL and empty buffer objects< / li >
< li > mesa/st: remove always-false state< / li >
< li > softpipe: setup pixel_offset for all primitive types< / li >
< li > docs: normaize css-indent style< / li >
< li > docs: remove non-existent css attribute< / li >
< li > docs: remove long commented out css< / li >
< li > docs: add missing semicolon< / li >
< li > docs: avoid repeating the font< / li >
< li > docs: avoid repeating the color< / li >
< li > docs: remove spurious newline< / li >
< li > docs: use multiple background-images for header< / li >
< li > docs: simplify css-centering< / li >
< li > docs: do not hard-code header-height< / li >
< li > docs: properly escape '> '< / li >
< li > docs: properly escape ampersand< / li >
< li > docs: remove stray paragraph-close< / li >
< li > docs: use h2 instead of b-tag for headings< / li >
< li > docs: use dl/dd instead of blockquote for freedesktop link< / li >
< li > docs: open list-item before closing it< / li >
< li > docs: close paragraphs before lists< / li >
< li > docs: close lists< / li >
< li > docs: remove stray paragraph-close< / li >
< li > docs: close paragraphs before preformatted text< / li >
< li > docs: start paragraph before closing it< / li >
< li > docs: drop paragraph around preformatted text< / li >
< li > docs: fix incorrectly closed paragraph< / li >
< li > docs: don't pointlessly close and re-start definition lists< / li >
< li > docs: remove stray list-start< / li >
< li > docs: fixup bad paragraphing< / li >
< li > docs: add missing lists< / li >
< li > docs: fix closing of paragraphs< / li >
< li > docs: fixup list-item tags< / li >
< li > docs: fix closing of list-items< / li >
< li > docs: replace empty list with a none-paragraph< / li >
< li > docs: turn faq-index into an ordered list< / li >
< li > docs: drop centered heading for faq< / li >
< li > docs: reorder heading and notice< / li >
< li > meson: lift driver-collection out into parent build-file< / li >
< li > meson: give dri- and gallium-drivers separate vars< / li >
< li > meson: add build-summary< / li >
< li > docs: fixup mistake in contents< / li >
< li > draw: flush when setting stream-out targets< / li >
< / ul >
< p > Ernestas Kulik (2):< / p >
< ul >
< li > vc4: Fix leak in HW queries error path< / li >
< li > v3d: Fix leak in resource setup error path< / li >
< / ul >
< p > Francisco Jerez (6):< / p >
< ul >
< li > intel/dump_gpu: Disambiguate between BOs from different GEM handle spaces.< / li >
< li > intel/fs: Exclude control sources from execution type and region alignment calculations.< / li >
< li > intel/fs: Lower integer multiply correctly when destination stride equals 4.< / li >
< li > intel/fs: Cap dst-aligned region stride to maximum representable hstride value.< / li >
< li > intel/fs: Implement extended strides greater than 4 for IR source regions.< / li >
< li > intel/fs: Rely on undocumented unrestricted regioning for 32x16-bit integer multiply.< / li >
< / ul >
< p > Fritz Koenig (4):< / p >
< ul >
< li > freedreno: pass count to query_dmabuf_modifiers< / li >
< li > freedreno/a6xx: UBWC support< / li >
< li > freedreno: UBWC allocator< / li >
< li > freedreno/a6xx: Enable UBWC modifier< / li >
< / ul >
< p > Gert Wollny (35):< / p >
< ul >
< li > mesa/core: Enable EXT_texture_sRGB_R8 also for desktop GL< / li >
< li > radeonsi: release tokens after creating the shader program< / li >
< li > mesa: release references to image textures when a context is destroyed< / li >
< li > virgl: Enable mixed color FBO attachemnets only when the host supports it< / li >
< li > mesa/core: Enable EXT_depth_clamp for GLES > = 2.0< / li >
< li > nir: Add posibility to not lower to source mod 'abs' for ops with three sources< / li >
< li > mesa: Expose EXT_texture_query_lod and add support for its use shaders< / li >
< li > softpipe: Enable PIPE_CAP_MIXED_COLORBUFFER_FORMATS It seems softpipe actually supports this. This change enables the following piglits as passing without regressions in the gpu test set:< / li >
< li > virgl: Add a caps feature check version< / li >
< li > softpipe: Implement ATOMFADD and enable cap TGSI_ATOMFADD< / li >
< li > virgl: define MAX_VERTEX_STREAMS based on availability of TF3< / li >
< li > softpipe: Use mag texture filter also for clamped lod == 0< / li >
< li > softpipe: Don't use mag filter for gather op< / li >
< li > softpipe: raise number of bits used for X coordinate texture lookup< / li >
< li > softpipe: Add an extra code path for the buffer texel lookup< / li >
< li > softpipe: Enable PIPE_CAP_TEXTURE_BUFFER_OFFSET_ALIGNMENT< / li >
< li > Gallium: Add new CAP that indicated whether IO array definitions can be shriked< / li >
< li > virgl: Enable passing arrays as input to fragment shaders< / li >
< li > doc/features: Add a few extensions to the feature matrix< / li >
< li > softpipe: Factor gradient evaluation out of the lambda evaluation< / li >
< li > softpipe: Prepare handling explicit gradients< / li >
< li > softpipe: Pipe gather_comp through from st_tgsi_get_samples< / li >
< li > softpipe: Move selection of shadow values up and clean parameter list< / li >
< li > softpipe: tie in new code path for lod evaluation< / li >
< li > softpipe: keep input lod for explicite derivatives< / li >
< li > softpipe: evaluate cube the faces on a per sample bases< / li >
< li > softpipe: Factor out evaluation of the source indices< / li >
< li > softpipe: Add an per-input array for interpolator correctors to machine< / li >
< li > softpipe: Add (fake) support for TGSI_OPCODE_INTERP_SAMPLE< / li >
< li > softpipe: Add support for TGSI_OPCODE_INTERP_OFFSET< / li >
< li > softpipe: Add support for TGSI_OPCODE_INTERP_CENTROID< / li >
< li > softpipe: Increase the GLSL feature level< / li >
< li > doc: Update feature matrix< / li >
< li > softpipe/buffer: load only as many components as the the buffer resource type provides< / li >
< li > Revert "softpipe/buffer: load only as many components as the the buffer resource type provides"< / li >
< / ul >
< p > Greg V (3):< / p >
< ul >
< li > util: emulate futex on FreeBSD using umtx< / li >
< li > gallium/hud: add CPU usage support for FreeBSD< / li >
< li > gallium: enable dmabuf on BSD as well< / li >
< / ul >
< p > Grigori Goronzy (1):< / p >
< ul >
< li > glx: add support for GLX_ARB_create_context_no_error (v3)< / li >
< / ul >
< p > Guido Günther (4):< / p >
< ul >
< li > docs: Fix 19.0.x version numbers< / li >
< li > gallium: ddebug: Add missing fence related wrappers< / li >
< li > gallium/u_dump: util_dump_sampler_view: Dump u.tex.first_level< / li >
< li > gallium: trace: Add missing fence related wrappers< / li >
< / ul >
< p > Gurchetan Singh (44):< / p >
< ul >
< li > mesa/main: Expose EXT_texture_compression_s3tc_srgb< / li >
< li > i965: Set flag for EXT_texture_compression_s3tc_srgb< / li >
< li > st/mesa: expose EXT_texture_compression_s3tc_srgb< / li >
< li > docs: add GL_EXT_texture_compression_s3tc_srgb to release notes< / li >
< li > virgl: add ability to do finer grain dirty tracking< / li >
< li > virgl: use virgl_resource_dirty helper< / li >
< li > virgl: don't mark unclean after a flush< / li >
< li > virgl: track level cleanliness rather than resource cleanliness< / li >
< li > virgl: make alignment smaller when uploading index user buffers< / li >
< li > virgl: unmap uploader at flush time< / li >
< li > virgl: when creating / freeing transfers, pass slab pool directly< / li >
< li > virgl: add protocol for resource transfers< / li >
< li > virgl: use virgl_transfer in inline write< / li >
< li > virgl: limit command length to 16 bits< / li >
< li > virgl: keep track of number of computations< / li >
< li > virgl: pass virgl transfer to virgl_res_needs_flush_wait< / li >
< li > virgl: add extra checks in virgl_res_needs_flush_wait< / li >
< li > virgl: make winsys modifications for encoded transfers< / li >
< li > virgl: add encoder functions for new protocol< / li >
< li > virgl: introduce transfer queue< / li >
< li > virgl: use transfer queue< / li >
< li > virgl: use virgl_transfer_inline_write even less< / li >
< li > virgl/vtest: deprecate protocol version 1< / li >
< li > egl/sl: also allow virtgpu to fallback to kms_swrast< / li >
< li > virgl: use uint16_t mask instead of separate booleans< / li >
< li > configure.ac / meson: depend on libnativewindow when appropriate< / li >
< li > anv: move anv_GetMemoryAndroidHardwareBufferANDROID up a bit< / li >
< li > anv: fix build on Nougat< / li >
< li > egl/android: move droid_image_loader_extension down a bit< / li >
< li > egl/android: move droid_open_device_drm_gralloc down a bit< / li >
< li > egl/android: droid_open_device_drm_gralloc --> droid_open_device< / li >
< li > egl/android: refactor droid_load_driver a bit< / li >
< li > egl/android: plumb swrast option< / li >
< li > egl/android: use swrast option in droid_load_driver< / li >
< li > egl/android: use software rendering when appropriate< / li >
< li > egl/android: chose node type based on swrast and preprocessor flags< / li >
< li > virgl: wait after a flush< / li >
< li > virgl/vtest: execute a transfer_get when flushing the front buffer< / li >
< li > virgl/vtest: add utilities for receiving fds< / li >
< li > virgl/vtest: plumb support for shared memory< / li >
< li > virgl/vtest: receive and handle shared memory fd< / li >
< li > virgl/vtest: modify sending and receiving data for shared memory< / li >
< li > virgl/vtest: wait after issuing a transfer get< / li >
< li > virgl/vtest: bump up protocol version + support encoded transfers< / li >
< / ul >
< p > Guttula, Suresh (1):< / p >
< ul >
< li > st/va:Add support for indirect manner by returning VA_STATUS_ERROR_OPERATION_FAILED< / li >
< / ul >
< p > Hal Gentz (1):< / p >
< ul >
< li > glx: Fix synthetic error generation in __glXSendError< / li >
< / ul >
< p > Heinrich (1):< / p >
< ul >
< li > gbm: Improve documentation of BO import< / li >
< / ul >
< p > Iago Toral Quiroga (39):< / p >
< ul >
< li > compiler/nir: add an is_conversion field to nir_op_info< / li >
< li > compiler/nir: add lowering option for 16-bit fmod< / li >
< li > compiler/nir: add lowering for 16-bit flrp< / li >
< li > compiler/nir: add lowering for 16-bit ldexp< / li >
< li > intel/compiler: add a NIR pass to lower conversions< / li >
< li > intel/compiler: split float to 64-bit opcodes from int to 64-bit< / li >
< li > intel/compiler: handle b2i/b2f with other integer conversion opcodes< / li >
< li > intel/compiler: assert restrictions on conversions to half-float< / li >
< li > intel/compiler: lower some 16-bit float operations to 32-bit< / li >
< li > intel/compiler: handle extended math restrictions for half-float< / li >
< li > intel/compiler: implement 16-bit fsign< / li >
< li > intel/compiler: drop unnecessary temporary from 32-bit fsign implementation< / li >
< li > intel/compiler: add instruction setters for Src1Type and Src2Type.< / li >
< li > intel/compiler: add new half-float register type for 3-src instructions< / li >
< li > intel/compiler: don't compact 3-src instructions with Src1Type or Src2Type bits< / li >
< li > intel/compiler: allow half-float on 3-source instructions since gen8< / li >
< li > intel/compiler: set correct precision fields for 3-source float instructions< / li >
< li > intel/compiler: fix ddx and ddy for 16-bit float< / li >
< li > intel/compiler: fix ddy for half-float in Broadwell< / li >
< li > intel/compiler: workaround for SIMD8 half-float MAD in gen8< / li >
< li > intel/compiler: split is_partial_write() into two variants< / li >
< li > intel/compiler: activate 16-bit bit-size lowerings also for 8-bit< / li >
< li > intel/compiler: rework conversion opcodes< / li >
< li > intel/compiler: ask for an integer type if requesting an 8-bit type< / li >
< li > intel/eu: force stride of 2 on NULL register for Byte instructions< / li >
< li > intel/compiler: generalize the combine constants pass< / li >
< li > intel/compiler: implement is_zero, is_one, is_negative_one for 8-bit/16-bit< / li >
< li > intel/compiler: add a brw_reg_type_is_integer helper< / li >
< li > intel/compiler: fix cmod propagation for non 32-bit types< / li >
< li > intel/compiler: remove inexact algebraic optimizations from the backend< / li >
< li > intel/compiler: skip MAD algebraic optimization for half-float or mixed mode< / li >
< li > intel/compiler: implement SIMD16 restrictions for mixed-float instructions< / li >
< li > intel/compiler: also set F execution type for mixed float mode in BDW< / li >
< li > intel/compiler: validate region restrictions for half-float conversions< / li >
< li > intel/compiler: validate conversions between 64-bit and 8-bit types< / li >
< li > intel/compiler: validate region restrictions for mixed float mode< / li >
< li > compiler/spirv: move the check for Int8 capability< / li >
< li > anv/pipeline: support Float16 and Int8 SPIR-V capabilities in gen8+< / li >
< li > anv/device: expose VK_KHR_shader_float16_int8 in gen8+< / li >
< / ul >
< p > Ian Romanick (55):< / p >
< ul >
< li > nir: Silence zillions of unused parameter warnings in release builds< / li >
< li > intel/compiler: Silence warning about value that may be used uninitialized< / li >
< li > nir: Document some fields of nir_loop_terminator< / li >
< li > nir: Refactor code that checks phi nodes in opt_peel_loop_initial_if< / li >
< li > nir: Select phi nodes using prev_block instead of continue_block< / li >
< li > nir: Split ALU instructions in loops that read phis< / li >
< li > nir: Convert a bcsel with only phi node sources to a phi node< / li >
< li > spirv: Add missing break< / li >
< li > nir/algebraic: Convert some f2u to f2i< / li >
< li > nir/algebraic: Simplify comparison with sequential integers starting with 0< / li >
< li > intel/vec4: Emit constants for some ALU sources as immediate values< / li >
< li > nir/algebraic: Replace i2b used by bcsel or if-statement with comparison< / li >
< li > intel/fs: Relax type matching rules in cmod propagation from MOV instructions< / li >
< li > intel/fs: Handle OR source modifiers in algebraic optimization< / li >
< li > intel/fs: Refactor ALU source and destination handling to a separate function< / li >
< li > intel/fs: Emit logical-not of operands on Gen8+< / li >
< li > intel/fs: Use De Morgan's laws to avoid logical-not of a logic result on Gen8+< / li >
< li > intel/fs: Emit better code for b2f(inot(a)) and b2i(inot(a))< / li >
< li > nir/algebraic: Replace a bcsel of a b2f sources with a b2f(!(a || b))< / li >
< li > intel/fs: Generate if instructions with inverted conditions< / li >
< li > nir/algebraic: Replace a-fract(a) with floor(a)< / li >
< li > intel/fs: Don't assert on b2f with a saturate modifier< / li >
< li > nir/algebraic: Optimize away an fsat of a b2f< / li >
< li > intel/compiler: Silence many unused parameter warnings in brw_eu.h< / li >
< li > intel/compiler: Silence unused parameter warning in brw_interpolation_map.c< / li >
< li > intel/fs: nir_op_extract_i8 extracts a byte, not a word< / li >
< li > intel/fs: Fix extract_u8 of an odd byte from a 64-bit integer< / li >
< li > nir/algebraic: Fix up extract_[iu]8 after loop unrolling< / li >
< li > nir/algebraic: Remove redundant extract_[iu]8 patterns< / li >
< li > nir/algebraic: Add missing 64-bit extract_[iu]8 patterns< / li >
< li > nir/algebraic: Add missing 16-bit extract_[iu]8 patterns< / li >
< li > nir/algebraic: Fix up extract_[iu]8 after loop unrolling< / li >
< li > nir/algebraic: Remove redundant extract_[iu]8 patterns< / li >
< li > nir/algebraic: Add missing 64-bit extract_[iu]8 patterns< / li >
< li > nir/algebraic: Add missing 16-bit extract_[iu]8 patterns< / li >
< li > nir: Add nir_const_value_negative_equal< / li >
< li > nir: Add nir_alu_srcs_negative_equal< / li >
< li > nir: Add partial redundancy elimination for compares< / li >
< li > intel/compiler: Use partial redundancy elimination for compares< / li >
< li > intel/fs: Eliminate dead code first< / li >
< li > intel/fs: Refactor code generation for nir_op_fsign to its own function< / li >
< li > intel/fs: Add a scale factor to emit_fsign< / li >
< li > intel/fs: Generate better code for fsign multiplied by a value< / li >
< li > nir/algebraic: Recognize open-coded copysign(1.0, a)< / li >
< li > nir/algebraic: Replace a pattern where iand with a Boolean is used as a bcsel< / li >
< li > nir/algebraic: Fix some 1-bit Boolean weirdness< / li >
< li > nir/algebraic: Strength reduce some compares of x and -x< / li >
< li > intel/fs: Add support for float16 to the fsign optimizations< / li >
< li > glsl: Silence may unused parameter warnings in glsl/ir.h< / li >
< li > intel/compiler: Don't have sepearate, per-Gen nir_options< / li >
< li > intel/compiler: Lower ffma on Gen4 and Gen5< / li >
< li > intel/fs: Fix D to W conversion in opt_combine_constants< / li >
< li > mesa: Add missing display list support for GL_FOG_COORDINATE_SOURCE< / li >
< li > nir: Saturating integer arithmetic is not associative< / li >
< li > Revert "nir: add late opt to turn inot/b2f combos back to bcsel"< / li >
< / ul >
< p > Icenowy Zheng (5):< / p >
< ul >
< li > lima: add dummy set_sample_mask function< / li >
< li > lima: make lima_context_framebuffer subtype of pipe_framebuffer_state< / li >
< li > lima: implement blit with util_blitter< / li >
< li > lima: lower bool to float when building shaders< / li >
< li > lima: add Android build< / li >
< / ul >
< p > Ilia Mirkin (14):< / p >
< ul >
< li > nv50,nvc0: add explicit settings for recent caps< / li >
< li > nvc0: add support for handling indirect draws with attrib conversion< / li >
< li > nvc0/ir: always use CG mode for loads from atomic-only buffers< / li >
< li > nvc0/ir: fix second tex argument after levelZero optimization< / li >
< li > nvc0: fix 3d images on kepler< / li >
< li > nv50,nvc0: use condition for occlusion queries when already complete< / li >
< li > nvc0: stick zero values for the compute invocation counts< / li >
< li > nvc0: we have 16k-sized framebuffers, fix default scissors< / li >
< li > swr: set PIPE_CAP_MAX_VARYINGS correctly< / li >
< li > mesa: add explicit enable for EXT_float_blend, and error condition< / li >
< li > st/mesa: enable GL_EXT_float_blend when possible< / li >
< li > i965: always enable EXT_float_blend< / li >
< li > nv50: disable compute< / li >
< li > glsl: fix recording of variables for XFB in TCS shaders< / li >
< / ul >
< p > Illia Iorin (1):< / p >
< ul >
< li > mesa/main: Fix multisample texture initialize< / li >
< / ul >
< p > James Zhu (12):< / p >
< ul >
< li > gallium/auxiliary/vl: Move dirty define to header file< / li >
< li > gallium/auxiliary/vl: Split vl_compositor graphic shaders from vl_compositor API< / li >
< li > gallium/auxiliary/vl: Rename csc_matrix and increase its size.< / li >
< li > gallium/auxiliary/vl: Add compute shader to support video compositor render< / li >
< li > gallium/auxiliary/vl: Add video compositor compute shader render< / li >
< li > gallium/auxiliary/vl: Fix transparent issue on compute shader with rgba< / li >
< li > gallium/auxiliary/vl: Increase shader_params size< / li >
< li > gallium/auxiliary/vl: Change grid setting< / li >
< li > gallium/auxiliary/vl: Change weave compute shader implementation< / li >
< li > gallium/auxiliary/vl: Fixed blur issue with weave compute shader< / li >
< li > gallium/auxiliary/vl: Fixed blank issue with compute shader< / li >
< li > gallium/auxiliary/vl: Add barrier/unbind after compute shader launch.< / li >
< / ul >
< p > Jan Vesely (2):< / p >
< ul >
< li > Partially revert "gallium: fix autotools build of pipe_msm.la"< / li >
< li > gallium/aux: Report error if loading of a pipe driver fails.< / li >
< / ul >
< p > Jan Zielinski (1):< / p >
< ul >
< li > swr/rast: fix 32-bit compilation on Linux< / li >
< / ul >
< p > Jason Ekstrand (212):< / p >
< ul >
< li > spirv: Replace vtn_constant_value with vtn_constant_uint< / li >
< li > spirv: Rework handling of spec constant workgroup size built-ins< / li >
< li > spirv: Handle constants and types before execution modes< / li >
< li > spirv: Handle OpExecutionModeId< / li >
< li > spirv: Support LocalSizeId and LocalSizeHintId execution modes< / li >
< li > intel/nir: Add global support to lower_mem_access_bit_sizes< / li >
< li > intel/fs/cse: Split create_copy_instr into three cases< / li >
< li > intel/fs: Properly handle 64-bit types in LOAD_PAYLOAD< / li >
< li > intel/fs: Do the grf127 hack on SIMD8 instructions in SIMD16 mode< / li >
< li > intel/fs: Implement load/store_global with A64 untyped messages< / li >
< li > intel/fs: Use SENDS for A64 writes on gen9+< / li >
< li > intel/fs: Implement nir_intrinsic_global_atomic_*< / li >
< li > anv: Implement VK_EXT_buffer_device_address< / li >
< li > relnotes: Add VK_EXT_buffer_device_address< / li >
< li > nir/deref: Drop zero ptr_as_array derefs< / li >
< li > README: Drop the badges from the readme< / li >
< li > intel/fs: Use enumerated array assignments in fb read TXF setup< / li >
< li > nir/deref: Rematerialize parents in rematerialize_derefs_in_use_blocks< / li >
< li > nir: Silence a couple of warnings in release builds< / li >
< li > anv/blorp: Delete a pointless assert< / li >
< li > anv: Silence some compiler warnings in release builds< / li >
< li > intel/fs: Silence a compiler warning< / li >
< li > intel/fs: Bail in optimize_extract_to_float if we have modifiers< / li >
< li > nir/dead_cf: Inline cf_node_has_side_effects< / li >
< li > nir/dead_cf: Stop relying on liveness analysis< / li >
< li > compiler/types: Add a contains_64bit helper< / li >
< li > nir/xfb: Properly align 64-bit values< / li >
< li > nir: Rewrite lower_clip_cull_distance_arrays to do a lot less lowering< / li >
< li > nir/xfb: Work in terms of components rather than slots< / li >
< li > nir/xfb: Handle compact arrays in gather_xfb_info< / li >
< li > nir: Fix a compile warning< / li >
< li > nir/lower_clip_cull: Fix an incorrect assert< / li >
< li > iris: Don't lower image formats for write-only images< / li >
< li > iris/compute: Don't increment the grid size offset< / li >
< li > iris/compute: Zero out the last grid size on indirect dispatches< / li >
< li > iris: Configure the L3$ on the compute context< / li >
< li > iris: Don't set constant read lengths at upload time< / li >
< li > iris: Allocate buffer resources separately< / li >
< li > iris: Copy anv's MI_MATH helpers for multiplication and division< / li >
< li > nir/split_vars: Don't compact vectors unnecessarily< / li >
< li > nir/builder: Don't emit no-op swizzles< / li >
< li > intel/eu: Add an EOT parameter to send_indirect_[split]_message< / li >
< li > intel/fs: Add an enum type for logical sampler inst sources< / li >
< li > intel/fs: Re-order logical surface arguments< / li >
< li > intel/fs: Drop the fs_surface_builder< / li >
< li > intel/vec4: Drop dead code for handling typed surface messages< / li >
< li > intel/fs: Get rid of the IMAGE_SIZE opcode< / li >
< li > intel/compiler: Drop unused surface opcodes< / li >
< li > intel/schedule_instructions: Move some comments< / li >
< li > intel/compiler: Re-prefix non-logical surface opcodes with VEC4< / li >
< li > anv: Count surfaces for non-YCbCr images in GetDescriptorSetLayoutSupport< / li >
< li > spirv: OpImageQueryLod requires a sampler< / li >
< li > intel,nir: Lower TXD with min_lod when the sampler index is not < 16< / li >
< li > anv: Use an actual binding for gl_NumWorkgroups< / li >
< li > anv/pipeline: Drop anv_fill_binding_table< / li >
< li > anv/descriptor_set: Refactor alloc/free of descriptor sets< / li >
< li > anv: Rework arguments to anv_descriptor_set_write_*< / li >
< li > anv: Stop allocating buffer views for dynamic buffers< / li >
< li > anv: Count image param entries rather than images< / li >
< li > anv: Clean up descriptor set layouts< / li >
< li > anv: drop add_var_binding from anv_nir_apply_pipeline_layout.c< / li >
< li > anv: Refactor descriptor pushing a bit< / li >
< li > anv: Take references to push descriptor set layouts< / li >
< li > anv: Add a concept of a descriptor buffer< / li >
< li > spirv: Pull offset/stride from the pointer for OpArrayLength< / li >
< li > spirv: Use the generic dereference function for OpArrayLength< / li >
< li > spirv: Use the same types for resource indices as pointers< / li >
< li > anv: Implement VK_EXT_inline_uniform_block< / li >
< li > nir: Expose double and int64 op_to_options_mask helpers< / li >
< li > nir: Teach loop unrolling about 64-bit instruction lowering< / li >
< li > i965: Compile the fp64 program based on nir options< / li >
< li > intel/debug: Add a debug flag to force software fp64< / li >
< li > intel/nir: Drop an unneeded lower_constant_initializers call< / li >
< li > glsl/nir: Add a shared helper for building float64 shaders< / li >
< li > glsl/nir: Inline functions in float64_funcs_to_nir< / li >
< li > nir/inline_functions: Break inlining into a builder helper< / li >
< li > nir/deref: Expose nir_opt_deref_impl< / li >
< li > nir/lower_doubles: Inline functions directly in lower_doubles< / li >
< li > intel/nir: Move 64-bit lowering later< / li >
< li > st/nir: Move 64-bit lowering later< / li >
< li > nir/builder: Emit better code for iadd/imul_imm< / li >
< li > nir/builder: Cast array indices in build_deref_follower< / li >
< li > nir/builder: Add a build_deref_array_imm helper< / li >
< li > intel/nir: Move lower_mem_access_bit_sizes to postprocess_nir< / li >
< li > anv/pipeline: Move lower_explicit_io much later< / li >
< li > nir: Add a pass for lowering IO back to vector when possible< / li >
< li > intel/nir: Vectorize all IO< / li >
< li > anv: Ignore VkRenderPassInputAttachementAspectCreateInfo< / li >
< li > nir/loop_unroll: Fix out-of-bounds access handling< / li >
< li > glsl/list: Add a list variant of insert_after< / li >
< li > glsl/lower_vector_derefs: Don't use a temporary for TCS outputs< / li >
< li > anv: Stop using VK_TRUE/FALSE< / li >
< li > anv/pass: Flag the need for a RT flush for resolve attachments< / li >
< li > anv: Only set 3DSTATE_PS::VectorMaskEnable on gen8+< / li >
< li > nir/algebraic: Add a couple optimizations for iabs and ishr< / li >
< li > nir/validate: Only require bare types to match for copy_deref< / li >
< li > nir/validate: Allow 32-bit boolean load/store intrinsics< / li >
< li > compiler/types: Add a new is_interface C wrapper< / li >
< li > compiler/types: Add a C wrapper to get full struct field data< / li >
< li > compiler/types: Add helpers to get explicit types for standard layouts< / li >
< li > nir/deref: Consider COHERENT decorated var derefs as aliasing< / li >
< li > nir: Rename nir_address_format_vk_index_offset to not be vk< / li >
< li > nir/lower_io: Add a new buffer_array_length intrinsic and lowering< / li >
< li > glsl: Don't lower vector derefs for SSBOs, UBOs, and shared< / li >
< li > glsl/nir: Set explicit types on UBO/SSBO variables< / li >
< li > glsl/nir: Handle unlowered SSBO atomic and array_length intrinsics< / li >
< li > glsl/nir: Add a pass to lower UBO and SSBO access< / li >
< li > i965: Stop setting LowerBuferInterfaceBlocks< / li >
< li > st/mesa: Let NIR lower UBO and SSBO access when we have it< / li >
< li > nir/builder: Add a vector extract helper< / li >
< li > nir: Add a new pass to lower array dereferences on vectors< / li >
< li > intel/nir: Lower array-deref-of-vector UBO and SSBO loads< / li >
< li > anv: Implement VK_EXT_host_query_reset< / li >
< li > anv,radv: Implement VK_KHR_surface_capability_protected< / li >
< li > Revert "nir: const `nir_call_instr::callee`"< / li >
< li > anv: Bump maxComputeWorkgroupInvocations< / li >
< li > nir: Constant values are per-column not per-component< / li >
< li > anv,radv,turnip: Lower TG4 offsets with nir_lower_tex< / li >
< li > spirv: Drop inline tg4 lowering< / li >
< li > nir/lower_io: Add a bounds-checked 64-bit global address format< / li >
< li > nir: Add a lowering pass for non-uniform resource access< / li >
< li > nir: Add texture sources and intrinsics for bindless< / li >
< li > nir: Add access flags to deref and SSBO atomics< / li >
< li > spirv: Handle the NonUniformEXT decoration< / li >
< li > Revert "anv/radv: release memory allocated by glsl types during spirv_to_nir"< / li >
< li > nir: Lock around validation fail shader dumping< / li >
< li > nir/algebraic: Drop some @bool specifiers< / li >
< li > nir/algebraic: Add some logical OR and AND patterns< / li >
< li > vc4: Prefer nir_src_comp_as_uint over nir_src_as_const_value< / li >
< li > nir/search: Search for all combinations of commutative ops< / li >
< li > nir: Get rid of nir_register::is_packed< / li >
< li > nir: Get rid of global registers< / li >
< li > intel/common: Add a MI command builder< / li >
< li > intel/common: Add unit tests for gen_mi_builder< / li >
< li > anv: Use gen_mi_builder for CmdDrawIndirectByteCount< / li >
< li > anv: Use gen_mi_builder for computing resolve predicates< / li >
< li > anv: Use gen_mi_builder for indirect draw parameters< / li >
< li > anv: Use gen_mi_builder for indirect dispatch< / li >
< li > anv: Use gen_mi_builder for conditional rendering< / li >
< li > anv: Use gen_mi_builder for queries< / li >
< li > anv: Move mi_memcpy and mi_memset to gen_mi_builder< / li >
< li > anv/cmd_buffer: Use gen_mi_sub instead of gen_mi_add with a negative< / li >
< li > intel/common: Support bigger right-shifts with mi_builder< / li >
< li > anv/pipeline: Fix MEDIA_VFE_STATE::PerThreadScratchSpace on gen7< / li >
< li > nir: Add a pass for selectively lowering variables to scratch space< / li >
< li > intel/nir: Take a nir_tex_instr and src index in brw_texture_offset< / li >
< li > nir/builder: Add a nir_imm_zero helper< / li >
< li > nir/print: Use nir_src_as_int for array indices< / li >
< li > nir/constant_folding: Get rid of a bit size switch statement< / li >
< li > spirv: Drop some unneeded bit size switch statements< / li >
< li > nir/load_const_to_scalar: Get rid of a bit size switch statement< / li >
< li > nir/validate: Require unused bits of nir_const_value to be zero< / li >
< li > vulkan: Update the XML and headers to 1.1.106< / li >
< li > anv: Update to use the new features struct names< / li >
< li > nir/algebraic: Move the template closer to the render function< / li >
< li > nir/algebraic: Use a cache to avoid re-emitting structs< / li >
< li > intel/mi_builder: Re-order an initializer< / li >
< li > intel/mi_builder: Disable mem_mem tests on IVB< / li >
< li > nir: Drop "struct" from some nir_* declarations< / li >
< li > nir: Rework nir_src_as_alu_instr to not take a pointer< / li >
< li > nir: Add a nir_src_as_intrinsic() helper< / li >
< li > anv: Re-sort the GetPhysicalDeviceFeatures2 switch statement< / li >
< li > anv: Drop some unneeded ANV_FROM_HANDLE for physical devices< / li >
< li > intel/fs: Account for live range lengths in spill costs< / li >
< li > anv: Make all VkDeviceMemory BOs resident permanently< / li >
< li > anv: Put image params in the descriptor set buffer on gen8 and earlier< / li >
< li > anv: Add a #define for the max binding table size< / li >
< li > anv/pipeline: Sort bindings by most used first< / li >
< li > anv/pipeline: Add skeleton support for spilling to bindless< / li >
< li > nir/lower_io: Expose some explicit I/O lowering helpers< / li >
< li > intel/nir: Re-run int64 lowering in postprocess_nir< / li >
< li > anv: Add a has_a64_buffer_access to anv_physical_device< / li >
< li > anv: Lower some SSBO operations in apply_pipeline_layout< / li >
< li > anv: Implement SSBOs bindings with GPU addresses in the descriptor BO< / li >
< li > anv: Implement VK_KHR_shader_atomic_int64< / li >
< li > intel,nir: Lower TXD with a bindless sampler< / li >
< li > intel/fs: Add support for bindless texture ops< / li >
< li > anv: Count the number of planes in each descriptor binding< / li >
< li > anv: Use write_image_view to initialize immutable samplers< / li >
< li > anv: Pass the plane into lower_tex_deref< / li >
< li > anv: Use bindless textures and samplers< / li >
< li > intel/fs: Add support for bindless image load/store/atomic< / li >
< li > anv: Use bindless handles for images< / li >
< li > anv: Put binding flags in descriptor set layouts< / li >
< li > anv: Implement VK_EXT_descriptor_indexing< / li >
< li > nir: Add helpers for getting the type of an address format< / li >
< li > anv/nir: Add a central helper for figuring out SSBO address formats< / li >
< li > anv: Ignore descriptor binding flags if bindingCount == 0< / li >
< li > anv: Rework the descriptor set layout create loop< / li >
< li > anv,radv: Update release notes for newly implemented extensiosn< / li >
< li > nir: Use the NIR_SRC_AS_ macro to define nir_src_as_deref< / li >
< li > anv/descriptor_set: Unlink sets from the pool in set_destroy< / li >
< li > anv/descriptor_set: Destroy sets before pool finalization< / li >
< li > anv/descriptor_set: Only vma_heap_finish if we have a descriptor buffer< / li >
< li > anv/descriptor_set: Properly align descriptor buffer to a page< / li >
< li > anv: Better handle 32-byte alignment of descriptor set buffers< / li >
< li > anv/descriptor_set: Don't fully destroy sets in pool destroy/reset< / li >
< li > nir/algebraic: Optimize integer cast-of-cast< / li >
< li > util/bitset: Return an actual bool from test macros< / li >
< li > anv: Stop including POS in FS input limits< / li >
< li > anv,i965: Stop warning about incomplete gen11 support< / li >
< li > nir: Add a SSA type gathering pass< / li >
< li > intel/fs/ra: Only add dest interference to sources that exist< / li >
< li > intel/fs/ra: Stop adding RA interference to too many SENDS nodes< / li >
< li > anv: Emulate texture swizzle in the shader when needed< / li >
< li > anv: Stop forcing bindless for images< / li >
< li > anv: Only consider minSampleShading when sampleShadingEnable is set< / li >
< li > iris: Don't assume UBO indices are constant< / li >
< li > intel/fs,vec4: Use g0 as the header for MFENCE< / li >
< li > intel/fs: Do a stalling MFENCE in endInvocationInterlock()< / li >
< li > nir/dead_cf: Call instructions aren't dead< / li >
< li > nir/propagate_invariant: Don't add NULL vars to the hash table< / li >
< / ul >
< p > Jian-Hong Pan (1):< / p >
< ul >
< li > intel: Fix the description of Coffeelake pci-id 0x3E98< / li >
< / ul >
< p > Jiang, Sonny (1):< / p >
< ul >
< li > va: use a compute shader for the blit< / li >
< / ul >
< p > John Stultz (3):< / p >
< ul >
< li > mesa: android: freedreno: Fix build failure due to path change< / li >
< li > mesa: Makefile.sources: Add ir3_nir_lower_load_barycentric_at_sample/offset to Makefile.sources< / li >
< li > mesa: Makefile.sources: Add nir_lower_fb_read.c to Makefile.sources list< / li >
< / ul >
< p > Jon Turney (1):< / p >
< ul >
< li > meson: Force '.so' extension for DRI drivers< / li >
< / ul >
< p > Jonathan Marek (22):< / p >
< ul >
< li > nir: add missing vec opcodes in lower_bool_to_float< / li >
< li > freedreno: a2xx: fix fast clear< / li >
< li > freedreno: a2xx: don't write 4th vertex in mem2gmem< / li >
< li > freedreno: a2xx: add use_hw_binning function< / li >
< li > freedreno: a2xx: fix fast clear for some gmem configurations< / li >
< li > freedreno: a2xx: fix mipmapping for NPOT textures< / li >
< li > freedreno: use renderonly path for buffers allocated with modifiers< / li >
< li > freedreno: catch failing fd_blit and fallback to software blit< / li >
< li > mesa: add GL_AMD_compressed_ATC_texture support< / li >
< li > gallium: add ATC format support< / li >
< li > llvmpipe, softpipe: no support for ATC textures< / li >
< li > st/mesa: add ATC support< / li >
< li > freedreno: a3xx: add GL_AMD_compressed_ATC_texture support< / li >
< li > freedreno: a2xx: add GL_AMD_compressed_ATC_texture support< / li >
< li > svga: add new ATC formats to the format conversion table< / li >
< li > freedreno: a2xx: fix builtin blit program compilation< / li >
< li > freedreno: a2xx: disable PIPE_CAP_PACKED_UNIFORMS< / li >
< li > freedreno: a2xx: use nir_lower_io for TGSI shaders< / li >
< li > freedreno: a2xx: enable batch reordering< / li >
< li > freedreno: a2xx: same gmem2mem sequence for all tiles< / li >
< li > nir: improve convert_yuv_to_rgb< / li >
< li > freedreno/ir3: fix input ncomp for vertex shaders< / li >
< / ul >
< p > Jordan Justen (22):< / p >
< ul >
< li > iris: Set num_uniforms in bytes< / li >
< li > iris/compute: Set mask bits on PIPELINE_SELECT< / li >
< li > iris: Add IRIS_DIRTY_CONSTANTS_CS< / li >
< li > iris: Add iris_restore_compute_saved_bos< / li >
< li > iris/compute: Add MEDIA_STATE_FLUSH following WALKER< / li >
< li > iris/compute: Flush compute batches< / li >
< li > iris/compute: Get group counts from grid-> grid< / li >
< li > iris/program: Don't try to push ubo ranges for compute< / li >
< li > iris/compute: Wait on compute batch when mapping< / li >
< li > iris/compute: Provide binding table entry for gl_NumWorkGroups< / li >
< li > iris/compute: Flush compute batch on memory-barriers< / li >
< li > iris/compute: Push subgroup-id< / li >
< li > iris/compute: Support indirect compute dispatch< / li >
< li > iris: Emit default L3 config for the render pipeline< / li >
< li > genxml/gen_bits_header.py: Use regex to strip no alphanum chars< / li >
< li > genxml: Remove extra space in gen4/45/5 field name< / li >
< li > iris: Add gitlab-ci build testing< / li >
< li > iris: Always use in-tree i915_drm.h< / li >
< li > nir: Add int64/doubles options into nir_shader_compiler_options< / li >
< li > intel/compiler: Move int64/doubles lowering options< / li >
< li > scons: Generate float64_glsl.h for glsl_to_nir fp64 lowering< / li >
< li > intel/genxml: Support base-16 in value & start fields in gen_sort_tags.py< / li >
< / ul >
< p > Jose Maria Casanova Crespo (4):< / p >
< ul >
< li > iris: Enable ARB_shader_draw_parameters support< / li >
< li > glsl: fix typos in comments "transfor" -> "transform"< / li >
< li > glsl: TCS outputs can not be transform feedback candidates on GLES< / li >
< li > iris: setup EdgeFlag Vertex Element when needed.< / li >
< / ul >
< p > José Fonseca (1):< / p >
< ul >
< li > scons: Workaround failures with MSVC when using SCons 3.0.[2-4].< / li >
< / ul >
< p > Juan A. Suarez Romero (22):< / p >
< ul >
< li > anv/cmd_buffer: check for NULL framebuffer< / li >
< li > nir: move ALU instruction before the jump instruction< / li >
< li > nir: remove jump from two merging jump-ending blocks< / li >
< li > genxml: add missing field values for 3DSTATE_SF< / li >
< li > anv: advertise 8 subpixel precision bits< / li >
< li > nir/spirv: return after emitting a branch in block< / li >
< li > anv: destroy descriptor sets when pool gets reset< / li >
< li > nir: deref only for OpTypePointer< / li >
< li > anv: advertise 8 subtexel/mipmap precision bits< / li >
< li > nir/xfb: do not use bare interface type< / li >
< li > meson: Add dependency on genxml to anvil genfiles< / li >
< li > Revert "intel/compiler: split is_partial_write() into two variants"< / li >
< li > spirv: add missing SPV_EXT_descriptor_indexing capabilities< / li >
< li > radv: enable descriptor indexing capabilities< / li >
< li > anv: enable descriptor indexing capabilities< / li >
< li > Update version to 19.1.0-rc1< / li >
< li > Update version to 19.1.0-rc2< / li >
< li > cherry-ignore: radeonsi: update buffer descriptors in all contexts after buffer invalidation< / li >
< li > Update version to 19.1.0-rc3< / li >
< li > Update version to 19.1.0-rc4< / li >
< li > Update version to 19.1.0-rc5< / li >
< li > Update version to 19.1.0< / li >
< / ul >
< p > Julien Isorce (5):< / p >
< ul >
< li > gallium: add resource_get_info to pipe_screen< / li >
< li > radeonsi: implement resource_get_info< / li >
< li > st/va: properly set stride and offset in vlVaDeriveImage< / li >
< li > r600: implement resource_get_info< / li >
< li > st/va: check resource_get_info nullity in vlVaDeriveImage< / li >
< / ul >
< p > Józef Kucia (3):< / p >
< ul >
< li > mesa: Fix GL_NUM_DEVICE_UUIDS_EXT< / li >
< li > radv: Fix driverUUID< / li >
< li > radv: clear vertex bindings while resetting command buffer< / li >
< / ul >
< p > Karol Herbst (82):< / p >
< ul >
< li > nvc0/ir: replace cvt instructions with add to improve shader performance< / li >
< li > gk104/ir: Use the new rcp/rsq in library< / li >
< li > gm107/ir: add fp64 rcp< / li >
< li > gm107/ir: add fp64 rsq< / li >
< li > gallium: add PIPE_CAP_MAX_VARYINGS< / li >
< li > st/mesa: require RGBA2, RGB4, and RGBA4 to be renderable< / li >
< li > glsl_type: initialize offset and location to -1 for glsl_struct_field< / li >
< li > nir/opt_if: don't mark progress if nothing changes< / li >
< li > clover: update ICD table to support everything up to 2.2< / li >
< li > nir: replace magic numbers with M_PI< / li >
< li > nir/spirv: improve parsing of the memory model< / li >
< li > nir: add support for address bit sized system values< / li >
< li > nir/vtn: add support for SpvBuiltInGlobalLinearId< / li >
< li > nir/spirv: initial handling of OpenCL.std extension opcodes< / li >
< li > prog_to_nir: fix write from vps to FOG< / li >
< li > nvc0: print the shader type when dumping headers< / li >
< li > nv50/ir: move common converter code in base class< / li >
< li > nv50/ir: add lowering helper< / li >
< li > nouveau: add support for nir< / li >
< li > nouveau: fix nir and TGSI shader cache collision< / li >
< li > nv50/ir/nir: run some passes to make the conversion easier< / li >
< li > nv50/ir/nir: track defs and provide easy access functions< / li >
< li > nv50/ir/nir: add nir type helper functions< / li >
< li > nv50/ir/nir: run assignSlots< / li >
< li > nv50/ir/nir: add loadFrom and storeTo helpler< / li >
< li > nv50/ir/nir: parse NIR shader info< / li >
< li > nv50/ir/nir: implement nir_load_const_instr< / li >
< li > nv50/ir/nir: add skeleton for nir_intrinsic_instr< / li >
< li > nv50/ir/nir: implement nir_alu_instr handling< / li >
< li > nv50/ir/nir: implement nir_intrinsic_load_uniform< / li >
< li > nv50/ir/nir: implement nir_intrinsic_store_(per_vertex_)output< / li >
< li > nv50/ir/nir: implement load_(interpolated_)input/output< / li >
< li > nv50/ir/nir: implement intrinsic_discard(_if)< / li >
< li > nv50/ir/nir: implement loading system values< / li >
< li > nv50/ir/nir: implement nir_ssa_undef_instr< / li >
< li > nv50/ir/nir: implement nir_instr_type_tex< / li >
< li > nv50/ir/nir: add skeleton getOperation for intrinsics< / li >
< li > nv50/ir/nir: implement vote and ballot< / li >
< li > nv50/ir/nir: implement variable indexing< / li >
< li > nv50/ir/nir: implement geometry shader nir_intrinsics< / li >
< li > nv50/ir/nir: implement nir_intrinsic_load_ubo< / li >
< li > nv50/ir/nir: implement ssbo intrinsics< / li >
< li > nv50/ir/nir: implement images< / li >
< li > nv50/ir/nir: add memory barriers< / li >
< li > nv50/ir/nir: implement load_per_vertex_output< / li >
< li > nv50/ir/nir: implement intrinsic shader_clock< / li >
< li > nv50/ir/nir: handle user clip planes for each emitted vertex< / li >
< li > nv50ir/nir: move immediates before use< / li >
< li > glsl: add packed for struct types< / li >
< li > glsl: add cl_size and cl_alignment< / li >
< li > nir/lower_locals_to_regs: cast array index to 32 bit< / li >
< li > nir/spirv: handle kernel function parameters< / li >
< li > nir/spirv: support physical pointers< / li >
< li > nir: add support for gather offsets< / li >
< li > nv50/ir/nir: support gather offsets< / li >
< li > nir/lower_tex: Add support for tg4 offsets lowering< / li >
< li > nir/print: fix printing the image_array intrinsic index< / li >
< li > nir/validate: validate that tex deref sources are actually derefs< / li >
< li > v3d: prefer using nir_src_comp_as_int over nir_src_as_const_value< / li >
< li > panfrost/midgard: use nir_src_is_const and nir_src_as_uint< / li >
< li > glsl/standalone: add GLES3.1 and GLES3.2 compatibility< / li >
< li > nir: move brw_nir_rewrite_image_intrinsic into common code< / li >
< li > glsl_to_nir: handle bindless textures< / li >
< li > glsl/nir: fetch the type for images from the deref instruction< / li >
< li > glsl/nir: add support for lowering bindless images_derefs< / li >
< li > nv50/ir/nir: handle bindless texture< / li >
< li > nv50/ir/nir: add support for bindless images< / li >
< li > nvc0/nir: enable bindless texture< / li >
< li > lima: add bool parameter to type_size function< / li >
< li > amd/nir: some cleanups< / li >
< li > radv: use nir constant helpers< / li >
< li > intel/nir: use nir_src_is_const and nir_src_as_uint< / li >
< li > freedreno/ir3: use nir_src_as_uint in a few places< / li >
< li > lima: use nir_src_as_float< / li >
< li > nir/builder: Move nir_imm_vec2 from blorp into the builder< / li >
< li > nir/loop_analyze: use nir_const_value.b for boolean results, not u32< / li >
< li > spirv: reduce array size in vtn_handle_constant< / li >
< li > nir: make nir_const_value scalar< / li >
< li > vtn: handle bitcast with pointer src/dest< / li >
< li > nir: Add a nir_builder_alu variant which takes an array of components< / li >
< li > nir: Add nir_op_vec helper< / li >
< li > spirv/cl: support vload/vstore< / li >
< / ul >
< p > Kasireddy, Vivek (3):< / p >
< ul >
< li > nir/lower_tex: Add support for XYUV lowering< / li >
< li > dri: Add XYUV8888 format< / li >
< li > i965: Add support for sampling from XYUV images< / li >
< / ul >
< p > Kenneth Graunke (872):< / p >
< ul >
< li > st/mesa: Set pipe_image_view::shader_access in PBO readpixels.< / li >
< li > st/nir: Move varying setup code to a helper function.< / li >
< li > st/nir: Make new helpers for constructing built-in NIR shaders.< / li >
< li > st/mesa: Add a NIR version of the drawpixels/bitmap VS copy shader.< / li >
< li > st/mesa: Add NIR versions of the drawpixels Z/stencil fragment shaders.< / li >
< li > st/mesa: Add NIR versions of the clear shaders.< / li >
< li > st/mesa: Add a NIR version of the OES_draw_texture built-in shaders.< / li >
< li > st/mesa: Add NIR versions of the PBO upload/download shaders.< / li >
< li > program: Use u_bit_scan64 in prog_to_nir.< / li >
< li > program: Extend prog_to_nir handle system values.< / li >
< li > nir: Record info-> fs.pixel_center_integer in lower_system_values< / li >
< li > compiler: Mark clip/cull distance arrays as compact before lowering.< / li >
< li > nir: Bail on clip/cull distance lowering if GLSL IR already did it.< / li >
< li > nir: Avoid clip/cull distance lowering multiple times.< / li >
< li > nir: Avoid splitting compact arrays into per-element variables.< / li >
< li > st/nir: Call nir_lower_clip_cull_distance_arrays().< / li >
< li > gallium: Add a PIPE_CAP_NIR_COMPACT_ARRAYS capability bit.< / li >
< li > nouveau: Silence unhandled cap warnings< / li >
< li > st/mesa: Limit GL_MAX_[NATIVE_]PROGRAM_PARAMETERS_ARB to 2048< / li >
< li > glsl: Allow gl_nir_lower_samplers*() without a gl_shader_program< / li >
< li > glsl: Don't look at sampler uniform storage for internal vars< / li >
< li > i965: Call nir_lower_samplers for ARB programs.< / li >
< li > st/nir: Pull sampler lowering into a helper function.< / li >
< li > st/nir: Lower sampler derefs for builtin shaders.< / li >
< li > st/nir: Use sampler derefs in built-in shaders.< / li >
< li > program: Make prog_to_nir create texture/sampler derefs.< / li >
< li > nir: Use sampler derefs in drawpixels and bitmap lowering.< / li >
< li > nir: Gather texture bitmasks in gl_nir_lower_samplers_as_deref.< / li >
< li > i965: Drop unnecessary 'and' with prog-> SamplerUnits< / li >
< li > i965: Use info-> textures_used instead of prog-> SamplersUsed.< / li >
< li > mesa: Advertise EXT_float_blend in ES 3.0+ contexts.< / li >
< li > anv: Put MOCS in the correct location< / li >
< li > spirv: Eliminate dead input/output variables after translation.< / li >
< li > nir: Don't reassociate add/mul chains containing only constants< / li >
< li > compiler: Make is_64bit(GL_*) helper more broadly available< / li >
< li > mesa: Align doubles to a 64-bit starting boundary, even if packing.< / li >
< li > radeonsi: Go back to using llvm.pow intrinsic for nir_op_fpow< / li >
< li > st/mesa: Copy VP TGSI tokens if they exist, even for NIR shaders.< / li >
< li > nir: Don't forget if-uses in new nir_opt_dead_cf liveness check< / li >
< li > iris: Initial commit of a new 'iris' driver for Intel Gen8+ GPUs.< / li >
< li > iris: viewport state, sort of< / li >
< li > iris: port over batchbuffer updates< / li >
< li > iris: initial render state upload< / li >
< li > iris: packing with valgrind.< / li >
< li > iris: merge pack< / li >
< li > iris: initial gpu state, merges< / li >
< li > iris: RASTER + SF + some CLIP, fix DIRTY vs. NEW< / li >
< li > iris: scissors< / li >
< li > iris: SF_CLIP_VIEWPORT< / li >
< li > iris: Surfaces!< / li >
< li > iris: sampler views< / li >
< li > iris: stipples and vertex elements< / li >
< li > iris: framebuffers< / li >
< li > iris: don't segfault on !old_cso< / li >
< li > iris: fix SF_CL length< / li >
< li > iris: a bit of depth< / li >
< li > iris: some draw info, vbs, sample mask< / li >
< li > iris: fix crash - CSO binding can be NULL (when destroying context)< / li >
< li > iris: COLOR_CALC_STATE< / li >
< li > iris: sampler states< / li >
< li > iris: emit 3DSTATE_SAMPLER_STATE_POINTERS< / li >
< li > iris: basic push constant alloc< / li >
< li > iris: some program code< / li >
< li > iris: linear resources< / li >
< li > iris: maps< / li >
< li > iris: shader debug log< / li >
< li > iris: drop unused field< / li >
< li > iris: make an ice-> render_batch field< / li >
< li > iris: disable execbuf for now< / li >
< li > iris: delete iris_pipe.c, shuffle code around< / li >
< li > iris: init the batch!< / li >
< li > iris: fix/rework line stipple< / li >
< li > iris: actually save VBs< / li >
< li > iris: msaa sample count packing problems< / li >
< li > iris: fix prim type< / li >
< li > iris: fix bogus index buffer reference< / li >
< li > iris: draw-> restart_index is uninitialized if PR is not enabled< / li >
< li > iris: parse INTEL_DEBUG< / li >
< li > iris: reworks, FS compile pieces< / li >
< li > iris: import program cache code< / li >
< li > iris: do the FS...asserts because we don't lower uniforms yet< / li >
< li > iris: lower io< / li >
< li > iris: make iris_batch target a particular ring< / li >
< li > iris: kill iris_new_batch< / li >
< li > iris: move MAX defines to iris_batch.h< / li >
< li > iris: bit of SBA code< / li >
< li > iris: flag SBA updates when instruction BO changes< / li >
< li > iris: try and have an iris address< / li >
< li > iris: so, sba then.< / li >
< li > iris: reference VB BOs< / li >
< li > iris: VB addresses< / li >
< li > iris: DEBUG=bat< / li >
< li > iris: VB fixes< / li >
< li > iris: actually APPEND commands, not stomp over the top and never incr< / li >
< li > iris: actually flush the commands< / li >
< li > iris: actually advance forward when emitting commands< / li >
< li > iris: initialize dirty bits to ~0ull< / li >
< li > iris: hack to stop crashing on samplers for now< / li >
< li > iris: fix indentation< / li >
< li > iris: fix assert< / li >
< li > iris: fix VBs< / li >
< li > iris: vertex packet fixes< / li >
< li > iris: fix VF instancing length so we don't get garbage in batch< / li >
< li > iris: 3DPRIMITIVE fields< / li >
< li > iris: bind_state -> compute state< / li >
< li > iris: scissor slots< / li >
< li > iris: some shader bits< / li >
< li > iris: promote iris_program_cache_item to iris_compiled_shader< / li >
< li > iris: actually save derived state< / li >
< li > iris: emit shader packets< / li >
< li > iris: convert IRIS_DIRTY_* to #defines< / li >
< li > iris: don't forget about TE< / li >
< li > iris: reorganize commands to match brw< / li >
< li > iris: initial gpu state< / li >
< li > iris: WM.< / li >
< li > iris: index buffer BO< / li >
< li > iris: more comes from bits filled in< / li >
< li > iris: drop const from prog data parameters< / li >
< li > iris: softpin some things< / li >
< li > iris: use vtbl to avoid multiple symbols, fix state base address< / li >
< li > iris: fix SBA< / li >
< li > iris: move key pop to state module< / li >
< li > iris: bits of WM key< / li >
< li > iris: shuffle comments< / li >
< li > iris: no NEW_SBA< / li >
< li > iris: rewrite program cache to use u_upload_mgr< / li >
< li > iris: actually destroy the cache< / li >
< li > iris: actually softpin at an address< / li >
< li > iris: actually set KSP offsets< / li >
< li > iris: URB configs.< / li >
< li > iris: dummy constants< / li >
< li > iris: blend state< / li >
< li > iris: alpha testing in PSB< / li >
< li > iris: basic SBE code< / li >
< li > iris: warning fixes< / li >
< li > iris: fix silly unused batch with addr macro< / li >
< li > iris: render targets!< / li >
< li > iris: don't do samplers for disabled stages< / li >
< li > iris: smaller blend state< / li >
< li > iris: actually pin the instruction cache buffers< / li >
< li > iris: compctrl< / li >
< li > iris: more sketchy SBE< / li >
< li > iris: fix dmabuf retval comparisons< / li >
< li > iris: more SF CL VPs< / li >
< li > iris: catastrophic state pointer mistake< / li >
< li > iris: fix extents< / li >
< li > iris: write DISABLES are not write ENABLES...whoops< / li >
< li > iris: sample mask...not 0.< / li >
< li > iris: uniform bits...badly< / li >
< li > iris: warn if execbuf fails< / li >
< li > iris: NOOP pad batches correctly< / li >
< li > iris: decode batches if they fail to submit< / li >
< li > iris: enable a few more formats< / li >
< li > iris: set strides on transfers< / li >
< li > iris: stop adding 9 to our varyings< / li >
< li > iris: bufmgr updates.< / li >
< li > iris: some thinking about binding tables< / li >
< li > iris: Soft-pin the universe< / li >
< li > iris: fix icache memzone< / li >
< li > iris: dump gtt offset in dump_validation_list< / li >
< li > iris: Also set SUPPORTS_48B? Not sure if necessary.< / li >
< li > iris: more uploaders< / li >
< li > iris: rewrite to use memzones and not relocs< / li >
< li > iris: set EXEC_OBJECT_WRITE< / li >
< li > iris: include p_defines.h in iris_bufmgr.h< / li >
< li > iris: binders< / li >
< li > iris: hook up batch decoder< / li >
< li > iris: binder fixes< / li >
< li > iris: decoder fixes< / li >
< li > iris: update vb BO handling now that we have softpin< / li >
< li > iris: validation dumping improvements< / li >
< li > iris: canonicalize addresses.< / li >
< li > iris: delete more trash< / li >
< li > iris: allocate SURFACE_STATEs up front and stop streaming them< / li >
< li > iris: same treatment for sampler views< / li >
< li > iris: assemble SAMPLER_STATE table at bind time< / li >
< li > iris: fix a scissor bug< / li >
< li > iris: SBA once at context creation, not per batch< / li >
< li > iris: TES stash< / li >
< li > iris: isv freeing fixes< / li >
< li > iris: set sampler views< / li >
< li > iris: decoder fixes< / li >
< li > iris: better BT asserts< / li >
< li > iris: increase allocator alignment< / li >
< li > iris: fix index< / li >
< li > iris: port bug fix from i965< / li >
< li > iris: fixes from i965< / li >
< li > iris: fixes< / li >
< li > iris: crazy pipe control code< / li >
< li > iris: bo reuse< / li >
< li > iris: vma fixes - don't free binder address< / li >
< li > iris: vma - fix assert< / li >
< li > iris: better SBE< / li >
< li > iris: fix texturing!< / li >
< li > iris: Move get_command_space to iris_batch.c< / li >
< li > iris: Defines for base addresses rather than numbers everywhere< / li >
< li > iris: pull in newer comments< / li >
< li > iris: copy over i965's cache tracking< / li >
< li > iris: move bo_offset_from_sba< / li >
< li > iris: bits of blorp code< / li >
< li > iris: more blitting code to make readpixels work< / li >
< li > iris: drop bogus binder free< / li >
< li > iris: fix sampler view crashes< / li >
< li > iris: more blorp< / li >
< li > iris: fix blorp prog data crashes< / li >
< li > iris: add INTEL_DEBUG=reemit< / li >
< li > iris: drop the 48b printout, we never use anything else< / li >
< li > iris: hacky flushing for now< / li >
< li > iris: linear staging buffers - fast CPU access...< / li >
< li > iris: make blorp pin the binder< / li >
< li > iris: blorp URB< / li >
< li > iris: no more drawing rectangle in blorp< / li >
< li > iris: assert surf init< / li >
< li > iris: some depth stuff :(< / li >
< li > iris: bump GL version to 4.2< / li >
< li > iris: uniforms for VS< / li >
< li > iris: proper length for VE packet?< / li >
< li > iris: proper # of uniforms< / li >
< li > iris: properly reject formats, fixes RGB32 rendering with texture float< / li >
< li > iris: blorp bug fixes< / li >
< li > iris: delete growing code and just die for now< / li >
< li > iris: just turn batch reset_and_clear_caches into reset< / li >
< li > iris: chaining not growing< / li >
< li > iris: caps< / li >
< li > iris: fix batch chaining...< / li >
< li > iris: fix decoding and undo testing code< / li >
< li > iris: Lower the max number of decoded VBO lines< / li >
< li > iris: fix whitespace< / li >
< li > iris: fix 3DSTATE_VERTEX_ELEMENTS length< / li >
< li > iris: more depth stuffs...< / li >
< li > iris: fix VF INSTANCING length< / li >
< li > iris: util_copy_framebuffer_state (ported from Rob's v3d patches)< / li >
< li > iris: transfers< / li >
< li > iris: flush always< / li >
< li > iris: maybe slightly less boats uniforms< / li >
< li > iris: fix constant packet length to match i965< / li >
< li > iris: better ubo handling< / li >
< li > iris: completely rewrite binder< / li >
< li > iris: have more than one const_offset< / li >
< li > iris: make surface states for cbufs< / li >
< li > iris: fill out pull constant buffers< / li >
< li > iris: fix pull bufs that aren't the first user upload< / li >
< li > iris: use u_transfer helpers for now< / li >
< li > iris: better VFI< / li >
< li > iris: fix release builds< / li >
< li > iris: drop assert for now< / li >
< li > iris: disable __gen_validate_value in release mode< / li >
< li > iris: allow mapped buffers during execution (faster)< / li >
< li > iris: comment about reemitting and flushing< / li >
< li > iris: state cleaning< / li >
< li > iris: untested index buffer upload< / li >
< li > iris: delete some pointless STATIC_ASSERTS< / li >
< li > iris: untested SAMPLER_STATE pin BO fix< / li >
< li > iris: put back the always flush - fixes some things :(< / li >
< li > iris: save pointers to streamed state resources< / li >
< li > iris: fix the validation list on new batches< / li >
< li > iris: flag DIRTY_WM properly< / li >
< li > iris: bindings dirty tracking< / li >
< li > iris: some dirty fixes< / li >
< li > iris: clear dirty< / li >
< li > iris: plug leaks< / li >
< li > iris: more leak fixes< / li >
< li > iris: pc fixes< / li >
< li > iris: remove 4 bytes of padding in iris_compiled_shader< / li >
< li > iris: rzalloc iris_compiled_shader so memcmp works even if padding creeps in< / li >
< li > iris: don't leak sampler state table resources< / li >
< li > iris: don't leak keyboxes when searching for an existing program< / li >
< li > iris: indentation< / li >
< li > iris: use pipe resources not direct BOs< / li >
< li > iris: clean up some warnings so I can see through the noise< / li >
< li > iris: print binder utilization in INTEL_DEBUG=submit< / li >
< li > iris: redo VB CSO a bit< / li >
< li > iris: print refcounts in INTEL_DEBUG=submit< / li >
< li > iris: support signed vertex buffer offsets< / li >
< li > iris: fix major refcounting bug with resources< / li >
< li > iris: fix caps so tests run again< / li >
< li > iris: avoid crashing on unbound constant resources< / li >
< li > iris: emit 3DSTATE_SBE_SWIZ< / li >
< li > iris: max VP index< / li >
< li > iris: fix viewport counts and settings< / li >
< li > iris: fix num viewports to be based on programs< / li >
< li > iris: fix VP iteration< / li >
< li > iris: scissor count fixes< / li >
< li > iris: actually init num_viewports< / li >
< li > iris: print second batch size separately< / li >
< li > iris: don't always flush< / li >
< li > iris: Handle batch submission failure "better"< / li >
< li > iris: bad inherited comments< / li >
< li > iris: colorize batchbuffer failures to make them stand out< / li >
< li > iris: iris - fix QWord aligned endings after batch chaining rework< / li >
< li > iris: tidy comments about mirroring modes< / li >
< li > iris: Disable unsupported mirror clamp modes< / li >
< li > iris: fix fragcoord ytransform< / li >
< li > iris: better boxing on maps< / li >
< li > iris: clears< / li >
< li > iris: rework DEBUG_REEMIT< / li >
< li > iris: shader dirty bits< / li >
< li > iris: clear fix< / li >
< li > iris: fall back to u_generate_mipmap< / li >
< li > iris: implement copy image< / li >
< li > iris: lightmodel flat< / li >
< li > iris: maybe-flush before blorp operations< / li >
< li > iris: fix provoking vertex ordering< / li >
< li > iris: larger polygon offset< / li >
< li > iris: TES uniform fixes< / li >
< li > iris: geometry shader support< / li >
< li > iris: don't emit garbage 3DSTATE_VERTEX_BUFFERS when there aren't any< / li >
< li > iris: fix 3DSTATE_VERTEX_ELEMENTS / VF_INSTANCING for 0 elements< / li >
< li > iris: fix GS dispatch mode< / li >
< li > iris: depth clears< / li >
< li > iris: null surface for unbound textures< / li >
< li > iris: state ref tuple< / li >
< li > iris: don't include binder in surface VMA range< / li >
< li > iris: border color memory zone :(< / li >
< li > iris: implement border color, fix other sampler nonsense< / li >
< li > iris: dead pointer< / li >
< li > iris: just malloc one iris_genx_state instead of a bunch of oddball pieces< / li >
< li > iris: SBE change stash< / li >
< li > iris: fix zoffset asserts with 2DArray/Cube< / li >
< li > iris: rename map-> stride< / li >
< li > iris: actually set cube bit properly< / li >
< li > iris: keep DISCARD_RANGE< / li >
< li > iris: actually handle array layers in blits< / li >
< li > iris: comment out l/a/i/la< / li >
< li > iris: fix clip flagging on fb changes< / li >
< li > iris: fix depth bounds clamp enables< / li >
< li > iris: don't crash on shader perf logs< / li >
< li > iris: slab allocate transfers< / li >
< li > iris: rearrange iris_resource.h< / li >
< li > iris: Implement 3DSTATE_SO_DECL_LIST< / li >
< li > iris: SO buffers< / li >
< li > iris: streamout< / li >
< li > iris: set even if no outputs< / li >
< li > iris: bother setting program_string_id...< / li >
< li > iris: fix SO_DECL_LIST< / li >
< li > iris: actually pin the buffers< / li >
< li > iris: fix sample mask for MSAA-off< / li >
< li > iris: disable 6x MSAA support< / li >
< li > iris: multislice transfer maps< / li >
< li > iris: fix CC_VIEWPORT< / li >
< li > iris: draw indirect support?< / li >
< li > iris: save query type< / li >
< li > iris: bits of multisample program key< / li >
< li > iris: s/hwcso/state/g< / li >
< li > iris: bind state helper function< / li >
< li > iris: NOS mechanics< / li >
< li > iris: record FS NOS< / li >
< li > iris: fix crash< / li >
< li > iris: fix sampler views of TBOs< / li >
< li > iris: fix texture buffer stride< / li >
< li > iris: TES program key inputs< / li >
< li > iris: compile a TCS...don't bother with passthrough yet< / li >
< li > iris: don't emit SO_BUFFERS and SO_DECL_LIST unless streamout is enabled< / li >
< li > iris: vertex ID, instance ID< / li >
< li > iris: fix SGVS when there are no valid vertex elements< / li >
< li > iris: fill out MAX_PATCH_VERTICES< / li >
< li > iris: assert about passthrough shaders to make this easier to detect< / li >
< li > iris: fix EmitNoIndirect< / li >
< li > iris: fix Z24< / li >
< li > iris: reemit blend state for alpha test function changes< / li >
< li > iris: point sprite enables< / li >
< li > iris: hack around samples confusion< / li >
< li > iris: fix blorp filters< / li >
< li > iris: expose more things that we already support< / li >
< li > iris: fix msaa flipping filters< / li >
< li > iris: export get_shader_info< / li >
< li > iris: implement set_shader_buffers< / li >
< li > iris: emit binding table for atomic counters and SSBOs< / li >
< li > iris: shorten loop< / li >
< li > iris: unbind compiled shaders if none are present< / li >
< li > iris: fix TBO alignment to match 965< / li >
< li > iris: enable SSBOs< / li >
< li > iris: fix SSBO indexing< / li >
< li > iris: fix for disabling ssbos< / li >
< li > iris: update bindings when changing programs< / li >
< li > iris: drop unused bo parameter< / li >
< li > iris: implement texture/memory barriers< / li >
< li > iris: Don't reserve new binding table section unless things are dirty< / li >
< li > iris: update a todo comment< / li >
< li > iris: BIG OL' HACK for UBO updates< / li >
< li > iris: enable texture gather< / li >
< li > iris: Avoid croaking when trying to create FBO surfaces with bad formats< / li >
< li > iris: fix GS output component limit< / li >
< li > iris: drop pipe_shader_state< / li >
< li > iris: fix sample mask< / li >
< li > iris: cube arrays are cubes too< / li >
< li > iris: we don't support textureGatherOffsets, need it lowered< / li >
< li > iris: add minor comments< / li >
< li > iris: comment everything< / li >
< li > iris: sync bugfixes from brw_bufmgr< / li >
< li > iris: remember to set bo-> userptr< / li >
< li > iris: rename ring to engine< / li >
< li > iris: simplify batch len qword alignment< / li >
< li > iris: get angry about execbuf failures< / li >
< li > iris: fill out more caps< / li >
< li > iris: depth or stencil fixes< / li >
< li > iris: clear stencil< / li >
< li > iris: actually emit stencil packets< / li >
< li > iris: allow S8 as a stencil format< / li >
< li > iris: WTF transfers< / li >
< li > iris: use u_transfer_helper for depth stencil packing/unpacking< / li >
< li > iris: drop stencil handling now that u_transfer_helper does it< / li >
< li > iris: refcounting, who needs it?< / li >
< li > iris: actually do stencil blits< / li >
< li > iris: say no to more formats< / li >
< li > iris: deal with Marek's new MSAA caps< / li >
< li > iris: we can do multisample Z resolves< / li >
< li > iris: Convert RGBX to RGBA for rendering.< / li >
< li > iris: disallow RGB32 formats too< / li >
< li > iris: Fix tiled memcpy for cubes...and for array slices< / li >
< li > iris: blorp blit multiple slices< / li >
< li > iris: assert depth is 1 in resource_copy_region< / li >
< li > iris: call maybe_flush for each blorp operation< / li >
< li > iris: implement ARB_clear_texture< / li >
< li > iris: last VUE map NOS, handle > 16 FS inputs< / li >
< li > iris: drop dead assignments< / li >
< li > iris: drop pwrite< / li >
< li > iris: port non-bucket alignment bugfix< / li >
< li > iris: don't emit SBE all the time< / li >
< li > iris: rename pipe to base< / li >
< li > iris: Drop bogus sampler state saving< / li >
< li > iris: move iris_shader_state from ice-> shaders.state to ice-> state.shaders< / li >
< li > iris: Move things to iris_shader_state< / li >
< li > iris: Move iris_sampler_view declaration to iris_resource.h< / li >
< li > iris: track depth/stencil writes enabled< / li >
< li > iris: use consistent copyright formatting< / li >
< li > iris: Move cache tracking to iris_resolve.c< / li >
< li > iris: proper cache tracking< / li >
< li > iris: precompute hashes for cache tracking< / li >
< li > iris: Reduce binder alignment from 64 to 32< / li >
< li > iris: reenable R32G32B32 texture buffers< / li >
< li > iris: z_res -> s_res< / li >
< li > iris: implement get_sample_position< / li >
< li > iris: fix line-aa-width< / li >
< li > iris: try to hack around binder issue< / li >
< li > iris: fix sampler state setting< / li >
< li > iris: big old hack for tex-miplevel-selection< / li >
< li > iris: use linear for 1D textures< / li >
< li > iris: handle level/layer in direct maps< / li >
< li > iris: fix crash when binding optional shader for the first time< / li >
< li > iris: Skip primitive ID overrides if the shader wrote a custom value< / li >
< li > iris: fix blend state memcpy< / li >
< li > iris: new caps< / li >
< li > iris: use Eric's new caps helper< / li >
< li > iris: Allow inlining of require/get_command_space< / li >
< li > iris: skip over whole function if dirty == 0< / li >
< li > iris: don't unconditionally emit 3DSTATE_VF / 3DSTATE_VF_TOPOLOGY< / li >
< li > iris: fix constant buffer 0 to be absolute< / li >
< li > iris: set EXEC_OBJECT_CAPTURE on all driver internal buffers< / li >
< li > iris: fix null FB and unbound tex surface state addresses< / li >
< li > iris: Support multiple binder BOs, update Surface State Base Address< / li >
< li > iris: fix SO offset writes for multiple streams< / li >
< li > iris: update comments for multibinder< / li >
< li > iris: move binder pinning outside the dirty == 0 check< / li >
< li > iris: re-pin binding table contents if we didn't re-emit them< / li >
< li > iris: enable ARB_enhanced_layouts< / li >
< li > iris: refactor LRIs in context setup< / li >
< li > iris: initialize "don't suck" bits, as Ben likes to call them< / li >
< li > iris: totally untested icelake support< / li >
< li > iris: refactor program CSO stuff< / li >
< li > iris: silence const warning< / li >
< li > iris: fix context restore of 3DSTATE_CONSTANT ranges< / li >
< li > iris: properly re-pin stencil buffers< / li >
< li > iris: delete bogus comment< / li >
< li > iris: inherit the index buffer properly< / li >
< li > iris: use 0 for TCS passthrough program string ID< / li >
< li > iris: rw_bo for pipe controls< / li >
< li > iris: LRM/SRM/SDI hooks< / li >
< li > iris: initial query code< / li >
< li > iris: gen10+ workarounds and break fix< / li >
< li > iris: results write< / li >
< li > iris: flush batch when asking for result via QBO< / li >
< li > iris: fix random failures via CS stall...but why?< / li >
< li > iris: gpr0 to bool< / li >
< li > iris: play chicken with timer queries for now< / li >
< li > iris: pipeline stats< / li >
< li > iris: primitives generated query support< / li >
< li > iris: drop explicit pinning< / li >
< li > iris: timestamps< / li >
< li > iris: ...and SO prims emitted queries< / li >
< li > iris: glGet timestamps, more correct timestamps< / li >
< li > iris: Need to | 1 when asking for timestamps< / li >
< li > iris: 36-bit overflow fixes< / li >
< li > iris: early return properly< / li >
< li > iris: better query file comment< / li >
< li > iris: magic number 36 -> #define< / li >
< li > iris: Enable ARB_shader_vote< / li >
< li > iris: just mark snapshots_landed from the CPU< / li >
< li > iris: drop a bunch of pipe_sampler_state stuff we don't need< / li >
< li > iris: vma_free bo-> size, not bo_size< / li >
< li > iris: don't mark contains_draw = false when chaining batches< / li >
< li > iris: fix Z32_S8 depth sampling< / li >
< li > iris: stencil texturing< / li >
< li > iris: force persample interp cap< / li >
< li > iris: pipe to scs -> iris_pipe.h< / li >
< li > iris: inline stage_from_pipe to avoid unused warnings< / li >
< li > iris: add gen11 to genX_call< / li >
< li > iris: Allow PIPE_CONTROL with Stall at Scoreboard and RT flush< / li >
< li > iris: rework format translation apis< / li >
< li > iris: Use R/RG instead of I/L/A when sampling< / li >
< li > iris: enable I/L formats< / li >
< li > iris: X32_S8X24 :/< / li >
< li > iris: set the binding table size< / li >
< li > iris: lower storage image derefs< / li >
< li > iris: implement set_shader_images hook< / li >
< li > iris: bother with BTIs< / li >
< li > iris: set image access correctly< / li >
< li > iris: actually set image access< / li >
< li > iris: null for non-existent cbufs< / li >
< li > iris: move images next to textures in binding table< / li >
< li > iris: advertise GL_ARB_shader_texture_image_samples< / li >
< li > iris: Enable fb fetch< / li >
< li > iris: initial compute caps< / li >
< li > iris: yes< / li >
< li > iris: drop dead format //'s< / li >
< li > iris: drop XXX's about swizzling< / li >
< li > iris: little bits of compute basics< / li >
< li > iris: drop XXX that Jordan handled< / li >
< li > iris: drop unnecessary #ifdefs< / li >
< li > iris: leave XXX about unnecessary binding table uploads< / li >
< li > iris: bail if SLM is needed< / li >
< li > iris: fix whitespace< / li >
< li > iris: XXX for compute state tracking :/< / li >
< li > iris: rewrite grid surface handling< / li >
< li > iris: better dirty checking< / li >
< li > iris: don't let render/compute contexts stomp each other's dirty bits< / li >
< li > iris: hack to avoid memorybarriers out the wazoo< / li >
< li > iris: do PIPELINE_SELECT for render engine, add flushes, GLK hacks< / li >
< li > iris: fix SBA flushing by refactoring code< / li >
< li > iris: try and avoid pointless compute submissions< / li >
< li > iris: fix UBOs with bindings that have an offset< / li >
< li > iris: flag CC_VIEWPORT when changing num viewports< / li >
< li > iris: fix SF_CLIP_VIEWPORT array indexing with multiple VPs< / li >
< li > iris: Fix texture buffer / image buffer sizes.< / li >
< li > iris: Clamp UBO and SSBO access to the actual BO size, for safety< / li >
< li > iris: Move snapshots_landed to the front.< / li >
< li > iris: Fix off by one in scissoring, empty scissors, default scissors< / li >
< li > iris: Fall back to 1x1x1 null surface if no framebuffer supplied< / li >
< li > iris: SO_DECL_LIST fix< / li >
< li > iris: Fix refcounting of grid surface< / li >
< li > iris: delete dead code< / li >
< li > iris: fix overhead regression from "don't stomp each other's dirty bits"< / li >
< li > iris: allow binding a null vertex buffer< / li >
< li > iris: Flag constants dirty on program changes< / li >
< li > iris: Disable a PIPE_CONTROL workaround on Icelake< / li >
< li > iris: Enable ARB_shader_stencil_export< / li >
< li > iris: Enable A8/A16_UNORM in an inefficient manner< / li >
< li > iris: Drop B5G5R5X1 support< / li >
< li > iris: Use at least 1x1 size for null FB surface state.< / li >
< li > iris: Cross-link iris_batches so they can potentially flush each other< / li >
< li > iris: cross batch flushing< / li >
< li > iris: Don't leak the compute batch< / li >
< li > iris: Actually create/destroy HW contexts< / li >
< li > iris: Enable msaa_map transfer helpers< / li >
< li > iris: tidy more warnings< / li >
< li > iris: implement scratch space!< / li >
< li > iris: Fix MSAA smooth points< / li >
< li > iris: Fix TextureBarrier< / li >
< li > iris: Fix multiple RTs with non-independent blending< / li >
< li > iris: partial set_query_active_state< / li >
< li > iris: Print the batch name when decoding< / li >
< li > iris: Clone the NIR< / li >
< li > iris: Defer cbuf0 upload to draw time< / li >
< li > iris: drop unnecessary param[] setup from iris_setup_uniforms< / li >
< li > iris: add param domain defines< / li >
< li > iris: fill out params array with built-ins, like clip planes< / li >
< li > iris: only bother with params if there are any...< / li >
< li > iris: lower user clip planes< / li >
< li > iris: hook up key stuff for clip plane lowering< / li >
< li > iris: fix system value remapping< / li >
< li > iris: dodge backend UCP lowering< / li >
< li > iris: bypass params and do it ourselves< / li >
< li > iris: actually upload clip planes.< / li >
< li > iris: fix num clip plane consts< / li >
< li > iris: fix more uniform setup< / li >
< li > iris: drop iris_setup_push_uniform_range< / li >
< li > iris: enable push constants if we have sysvals but no uniforms< / li >
< li > iris: regather info so we get CLIP_DIST slots, not CLIP_VERTEX< / li >
< li > iris: don't support pull constants.< / li >
< li > iris: don't trip on param asserts< / li >
< li > iris: drop param stuffs< / li >
< li > iris: don't forget to upload CS consts< / li >
< li > iris: fix sysval only binding tables< / li >
< li > iris: only clip lower if there's something to clip against< / li >
< li > iris: leave another TODO< / li >
< li > iris: Fix SourceAlphaBlendFactor< / li >
< li > iris: "Fix" transfer maps of buffers< / li >
< li > iris: Fix independent alpha blending.< / li >
< li > iris: more TODO< / li >
< li > iris: scissored and mirrored blits< / li >
< li > iris: more todo notes< / li >
< li > iris: Fix TCS/TES slot unification< / li >
< li > iris: properly pin stencil buffers< / li >
< li > iris: Fix SLM< / li >
< li > iris: Use iris_use_pinned_bo rather than add_exec_bo directly< / li >
< li > iris: Combine iris_use_pinned_bo and add_exec_bo< / li >
< li > iris: Avoid cross-batch synchronization on read/reads< / li >
< li > iris: Avoid synchronizing due to the workaround BO< / li >
< li > iris: replace vestiges of fence fds with newer exec_fence API< / li >
< li > iris: Drop vestiges of throttling code< / li >
< li > iris: Hang on to the last batch's sync-point, so we can wait on it< / li >
< li > iris: Add wait fences to properly sync between render/compute< / li >
< li > iris: leave a TODO< / li >
< li > iris: flush the compute batch too if border pool is redone< / li >
< li > iris: put render batch first in fence code< / li >
< li > iris: Put batches in an array< / li >
< li > iris: PIPE_CONTROL workarounds for GPGPU mode< / li >
< li > iris: RT flush for memorybarrier with texture bit< / li >
< li > iris: update comment< / li >
< li > iris: Enable ctx-> Const.UseSTD430AsDefaultPacking< / li >
< li > iris: Lie about indirects< / li >
< li > iris: Fix buffer -> buffer copy_region< / li >
< li > iris: Fix VIEWPORT/LAYER in stream output info< / li >
< li > iris: Do the 48-bit vertex buffer address invalidation workaround< / li >
< li > iris: drop long dead XXX comment< / li >
< li > iris: Track a binding history for buffer resources< / li >
< li > iris: add iris_flush_and_dirty_for_history< / li >
< li > iris: Flush for history at various moments< / li >
< li > iris: Re-pin even if nothing is dirty< / li >
< li > iris: fix prototype warning< / li >
< li > iris: export iris_upload_shader< / li >
< li > iris: fix comment location< / li >
< li > iris: Use wrappers for create_xs_state rather than a switch statement< / li >
< li > iris: rework program cache interface< / li >
< li > iris: Enable precompiles< / li >
< li > iris: Use program's num textures not the state tracker's bound< / li >
< li > iris: drop pull constant binding table entry< / li >
< li > iris: add assertions about binding table starts< / li >
< li > iris: add an extra BT assert from Chris Wilson< / li >
< li > iris: actually flush for storage images< / li >
< li > iris: fix some SO overflow query bugs and tidy the code a bit< / li >
< li > iris: drop key_size_for_cache< / li >
< li > iris: for BLORP, only use the predicate enable bit when USE_BIT< / li >
< li > iris: check query first< / li >
< li > iris: fix conditional compute, don't stomp predicate for pipelined queries< / li >
< li > iris: Rework tiling/modifiers handling< / li >
< li > iris: Fix failed to compile TCS message< / li >
< li > iris: Destroy transfer helper on screen teardown< / li >
< li > iris: Destroy the border color pool< / li >
< li > iris: Unref unbound_tex resource< / li >
< li > iris: Fix IRIS_MEMZONE_COUNT to exclude the border color pool< / li >
< li > iris: Destroy the bufmgr< / li >
< li > iris: Stop leaking iris_uncompiled_shaders like mad< / li >
< li > iris: move some non-buffer case code in a bit< / li >
< li > iris: Don't bother considering if the underlying surface is a cube< / li >
< li > iris: fix alpha channel for RGB BC1 formats< / li >
< li > iris: fix dma buf import strides< / li >
< li > iris: CS stall for stream out -> VB< / li >
< li > iris: make clipper statistics dynamic< / li >
< li > iris: reject all clipping when we can't use streamout render disabled< / li >
< li > iris: omask can kill< / li >
< li > iris: reemit SBE when sprite coord origin changes< / li >
< li > iris: re-pin inherited streamout buffers< / li >
< li > iris: Fix NOS mechanism< / li >
< li > iris: fix overhead regression from flushing for storage images< / li >
< li > iris: fix set_sampler_views to not unbind, be better about bounds< / li >
< li > iris: Fix set_sampler_views with start > 0< / li >
< li > iris: Replace num_textures etc with a bitmask we can scan< / li >
< li > iris: Drop continues in resolve< / li >
< li > iris: Fix clear dimensions< / li >
< li > iris: Clamp viewport extents to the framebuffer dimensions< / li >
< li > iris: Enable guardband clipping< / li >
< li > iris: Fix primitive generated query active flag< / li >
< li > iris: Always do rasterizer discard in clipper< / li >
< li > iris: override alpha to one src1 blend factors< / li >
< li > iris: handle PatchVerticesIn as a system value.< / li >
< li > iris: rewrite set_vertex_buffer and VB handling< / li >
< li > iris: Reorder LRR parameters to have dst first.< / li >
< li > iris: Add _MI_ALU helpers that don't paste< / li >
< li > iris: Don't bother packing 3DSTATE_SO_BUFFER at create time< / li >
< li > iris: Move iris_stream_output_target def to iris_context.h< / li >
< li > iris: only get space for one offset in stream output targets< / li >
< li > iris: Implement DrawTransformFeedback()< / li >
< li > iris: drop unnecessary genx-> streamout field< / li >
< li > iris: Fix for PIPE_CAP_SIGNED_VERTEX_BUFFER_OFFSET< / li >
< li > iris: Fix the prototype for iris_bo_alloc_tiled< / li >
< li > iris: don't print the pointer in INTEL_DEBUG=submit< / li >
< li > iris: Use a surface state fill helper< / li >
< li > iris: Make a alloc_surface_state helper< / li >
< li > iris: whitespace fixes< / li >
< li > iris: Track blend enables, save outbound for resolve code< / li >
< li > iris: always pin the binder...in the compute context, too.< / li >
< li > iris: delete finished comments< / li >
< li > iris: pin and re-pin the scratch BO< / li >
< li > iris: more dead comments< / li >
< li > iris: only mark depth/stencil as writable if writes are actually enabled< / li >
< li > iris: better MOCS< / li >
< li > iris: Fix scratch space allocation on Icelake.< / li >
< li > iris: Only resolve inputs for actual shader stages< / li >
< li > iris: Add a more long term TODO about timebase scaling< / li >
< li > iris: Fix compute scratch pinning< / li >
< li > iris: Delete bogus comment about cube array counting.< / li >
< li > iris: Fix framebuffer layer count< / li >
< li > iris: Don't enable push constants just because there are system values< / li >
< li > iris: Don't make duplicate system values< / li >
< li > iris: Fill out brw_image_params for storage images on Broadwell< / li >
< li > iris: Fix surface states for Gen8 lowered-to-untype images< / li >
< li > iris: Leave a comment about why Broadwell images are broken< / li >
< li > iris: Implement multi-slice copy_region< / li >
< li > iris: Flush the render cache in flush_and_dirty_for_history< / li >
< li > iris: Handle PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE somewhat< / li >
< li > iris: Don't check other batches for our batch BO< / li >
< li > iris: Drop a dead comment< / li >
< li > iris: Delete genx-> bound_vertex_buffers< / li >
< li > iris: Fix Broadwell WaDividePSInvocationCountBy4< / li >
< li > iris: Use new PIPE_STAT_QUERY enums rather than hardcoded numbers.< / li >
< li > iris: Switch to the new PIPELINE_STATISTICS_QUERY_SINGLE capability< / li >
< li > iris: fail to create screen for older unsupported HW< / li >
< li > iris: Allow sample mask of 0< / li >
< li > iris: Don't enable smooth points when point sprites are enabled< / li >
< li > iris: Assert about blits with color masking< / li >
< li > iris: Pay attention to blit masks< / li >
< li > iris: CS stall on VF cache invalidate workarounds< / li >
< li > iris: Fix SO issue with INTEL_DEBUG=reemit, set fewer bits< / li >
< li > iris: Don't whack SO dirty bits when finishing a BLORP op< / li >
< li > iris: Fix memzone_for_address for the surface and binder zones< / li >
< li > iris: Do binder address allocations per-context, not globally.< / li >
< li > iris: Zero the compute predicate when changing the render condition< / li >
< li > iris: Remap stream output indexes back to VARYING_SLOT_*.< / li >
< li > iris: Enable PIPE_CAP_COMPACT_ARRAYS< / li >
< li > iris: Drop comment about ISP_DIS< / li >
< li > iris: Drop dead state_size hash table< / li >
< li > iris: Unreference some more things on state module teardown< / li >
< li > iris: minor tidying< / li >
< li > iris: Fix bug in bound vertex buffer tracking< / li >
< li > iris: Implement ALT mode for ARB_{vertex,fragment}_shader< / li >
< li > iris: Add a timeout_nsec parameter, rename check_syncpt to wait_syncpt< / li >
< li > iris: Fix accidental busy-looping in query waits< / li >
< li > iris: Use READ_ONCE and WRITE_ONCE for snapshots_landed< / li >
< li > iris: Make a iris_batch_reference_signal_syncpt helper function.< / li >
< li > iris: Add PIPE_CAP_MAX_VARYINGS< / li >
< li > iris: rework num textures to util_lastbit< / li >
< li > iris: Stop chopping off the first nine characters of the renderer string< / li >
< li > iris: Drop XXX about alpha testing< / li >
< li > iris: Set 3DSTATE_WM::ForceThreadDispatchEnable< / li >
< li > iris: Set HasWriteableRT correctly< / li >
< li > iris: Drop XXX about checking for swizzling< / li >
< li > iris: Move create and bind driver hooks to the end of iris_program.c< / li >
< li > iris: Make an IRIS_MAX_MIPLEVELS define< / li >
< li > iris: Simplify iris_get_depth_stencil_resources< / li >
< li > iris: Add missing depth cache flushes< / li >
< li > iris: Always emit at least one BLEND_STATE< / li >
< li > iris: Add iris_resource fields for aux surfaces< / li >
< li > iris: Fill out res-> aux.possible_usages< / li >
< li > iris: Fill out SURFACE_STATE entries for each possible aux usage< / li >
< li > iris: create aux surface if needed< / li >
< li > iris: Initial import of resolve code< / li >
< li > iris: blorp using resolve hooks< / li >
< li > iris: add some draw resolve hooks< / li >
< li > iris: actually use the multiple surf states for aux modes< / li >
< li > iris: try to fix copyimage vs copybuffers< / li >
< li > iris: be sure to skip buffers in resolve code< / li >
< li > iris: resolve before transfer maps< / li >
< li > iris: pin the buffers< / li >
< li > iris: store modifier info in res< / li >
< li > iris: Make blit code use actual aux usages< / li >
< li > iris: consider framebuffer parameter for aux usages< / li >
< li > iris: Resolves for compute< / li >
< li > iris: disable aux for external things< / li >
< li > iris: some initial HiZ bits< / li >
< li > iris: don't use hiz for MSAA buffers< / li >
< li > iris: Set program key fields for MCS< / li >
< li > iris: make surface states for CCS_D too< / li >
< li > iris: do flush for buffers still< / li >
< li > iris: Allow disabling aux via INTEL_DEBUG options< / li >
< li > iris: Fix aux usage in render resolve code< / li >
< li > iris: Only resolve compute resources for compute shaders< / li >
< li > iris: Enable auxiliary buffer support< / li >
< li > iris: Enable -msse2 and -mstackrealign< / li >
< li > Revert "iris: Enable auxiliary buffer support"< / li >
< li > vulkan: Fix 32-bit build for the new overlay layer< / li >
< li > mesa: Fix RGBBuffers for renderbuffers with sized internal formats< / li >
< li > iris: Drop RGBX -> RGBA for storage image usages< / li >
< li > iris: Properly allow rendering to RGBX formats.< / li >
< li > i965: Implement threaded GL support.< / li >
< li > tgsi_to_nir: use sampler variables and derefs< / li >
< li > iris: Fix MOCS for blits and clears< / li >
< li > isl: Add a swizzle parameter to isl_buffer_fill_state()< / li >
< li > iris: Plumb through ISL_SWIZZLE_IDENTITY in buffer surface emitters< / li >
< li > iris: Defer uploading sampler state tables until draw time< / li >
< li > iris: Properly support alpha and luminance-alpha formats< / li >
< li > iris: Drop PIPE_CAP_BUFFER_SAMPLER_VIEW_RGBA_ONLY< / li >
< li > iris: Spruce up "are we using this engine?" checks for flushing< / li >
< li > iris: Export a copy_region helper that doesn't flush< / li >
< li > iris: Use copy_region and staging resources to avoid transfer stalls< / li >
< li > Revert MR 369 (Fix extract_i8 and extract_u8 for 64-bit integers)< / li >
< li > iris: Fix backface stencil write condition< / li >
< li > iris: Rework default tessellation level uploads< / li >
< li > iris: Fix TES gl_PatchVerticesIn handling.< / li >
< li > iris: Move depth/stencil flushes so they actually do something< / li >
< li > iris: Refactor depth/stencil buffer pinning into a helper.< / li >
< li > iris: Fix write enable in pinning of depth/stencil resources< / li >
< li > i965: Move some genX infrastructure to genX_boilerplate.h.< / li >
< li > i965: Rename ISP_DIS to INDIRECT_STATE_POINTERS_DISABLE.< / li >
< li > i965: Use genxml for emitting PIPE_CONTROL.< / li >
< li > i965: Reimplement all the PIPE_CONTROL rules.< / li >
< li > intel/fs: Fix opt_peephole_csel to not throw away saturates.< / li >
< li > iris: Don't mutate box in transfer map code< / li >
< li > iris: Don't flush the batch for unsynchronized mappings< / li >
< li > iris: Slightly better bounds on buffer sizes< / li >
< li > gallium: Add PIPE_BARRIER_UPDATE_BUFFER and UPDATE_TEXTURE bits.< / li >
< li > nvc0: Skip new update barrier bits< / li >
< li > nir: Record non-vector/scalar varyings as unmovable when compacting< / li >
< li > iris: Fix util_vma_heap_init size for IRIS_MEMZONE_SHADER< / li >
< li > iris: Skip input resolve handling if bindings haven't changed< / li >
< li > iris: Skip framebuffer resolve tracking if framebuffer isn't dirty< / li >
< li > iris: Skip resolves and flushes altogether if unnecessary< / li >
< li > iris: Fix batch chaining map_next increment.< / li >
< li > iris: Actually advertise some modifiers< / li >
< li > st/nir: Free the GLSL IR after linking.< / li >
< li > st/mesa: Fix blitting from GL_DEPTH_STENCIL to GL_STENCIL_INDEX< / li >
< li > iris: Fix blits with S8_UINT destination< / li >
< li > iris: Print the memzone name when allocating BOs with INTEL_DEBUG=buf< / li >
< li > iris: Save/restore MI_PREDICATE_RESULT, not MI_PREDICATE_DATA.< / li >
< li > iris: Silence unused variable warnings in release mode< / li >
< li > gallium/util: Add const to u_range_intersect< / li >
< li > iris: Actually pin the scratch BO.< / li >
< li > glsl: Set location on structure-split sampler uniform variables< / li >
< li > intel: Emit 3DSTATE_VF_STATISTICS dynamically< / li >
< li > iris: Actually mark blorp_copy_buffer destinations as written.< / li >
< li > iris: Preserve all PIPE_TRANSFER flags in xfer-> usage< / li >
< li > iris: Fix FLUSH_EXPLICIT handling with staging buffers.< / li >
< li > iris: Make shader_perf_log print to stderr if INTEL_DEBUG=perf is set< / li >
< li > i965: Move program key debugging to the compiler.< / li >
< li > iris: Print the reason for shader recompiles.< / li >
< li > iris: Move iris_debug_recompile calls before uploading.< / li >
< li > iris: Change vendor and renderer strings< / li >
< li > iris: Add texture cache flushing hacks for blit and resource_copy_region< / li >
< li > iris: Be less aggressive at postdraw work skipping< / li >
< li > iris: Add mechanism for iris-specific driconf options< / li >
< li > iris: Enable the dual_color_blend_by_location driconf option.< / li >
< li > iris: Track bound and writable SSBOs< / li >
< li > Revert "glsl: Set location on structure-split sampler uniform variables"< / li >
< li > i965: Ignore uniform storage for samplers or images, use binding info< / li >
< li > i965: Tidy bogus indentation left by previous commit< / li >
< li > iris: Mark constants dirty on transfer unmap even if no flushes occur< / li >
< li > iris: Track bound constant buffers< / li >
< li > iris: Rework UBOs and SSBOs to use pipe_shader_buffer< / li >
< li > iris: Rework image views to store pipe_image_view.< / li >
< li > iris: Make a gl_shader_stage -> pipe_shader_stage helper function< / li >
< li > iris: Make memzone_for_address non-static< / li >
< li > iris: Replace buffer backing storage and rebind to update addresses.< / li >
< li > iris: Make a resource_is_busy() helper< / li >
< li > iris: Track valid data range and infer unsynchronized mappings.< / li >
< li > iris: Make some offset math helpers take a const isl_surf pointer< / li >
< li > iris: Fix DrawTransformFeedback math when there's a buffer offset< / li >
< li > iris: Prefer staging blits when destination supports CCS_E.< / li >
< li > iris: Actually put Mesa in GL_RENDERER string< / li >
< li > iris: Split iris_flush_and_dirty_for_history into two helpers.< / li >
< li > iris: Enable GL_AMD_depth_clamp_separate< / li >
< li > iris: Advertise EXT_texture_sRGB_R8 support< / li >
< li > iris: Some tidying for preemption support< / li >
< li > iris: Silence unused function warning< / li >
< li > iris: Fix zeroing of transform feedback offsets in strange cases.< / li >
< li > glsl/list: Add an exec_list_is_singular() helper.< / li >
< li > nir: Add a new nir_cf_list_is_empty_block() helper.< / li >
< li > intel/fs: Don't emit empty ELSE blocks.< / li >
< li > iris: Set XY Clipping correctly.< / li >
< li > iris: Only enable GL_AMD_depth_clamp_separate on Gen9+< / li >
< li > iris: Fix imageBuffer and PBO download.< / li >
< li > iris: Disable dual source blending when shader doesn't handle it< / li >
< li > iris: Resolve textures used by the program, not merely bound textures< / li >
< li > iris: Fix 4GB memory zone heap sizes.< / li >
< li > iris: leave the top 4Gb of the high heap VMA unused< / li >
< li > iris: Force VMA alignment to be a multiple of the page size.< / li >
< li > iris: Delete bucketing allocators< / li >
< li > i965: Fix BRW_MEMZONE_LOW_4G heap size.< / li >
< li > i965: Force VMA alignment to be a multiple of the page size.< / li >
< li > i965: leave the top 4Gb of the high heap VMA unused< / li >
< li > i965: Fix memory leaks in brw_upload_cs_work_groups_surface().< / li >
< li > iris: Use full ways for L3 cache setup on Icelake.< / li >
< li > egl/x11: calloc dri2_surf so it's properly zeroed< / li >
< / ul >
< p > Kevin Strasser (1):< / p >
< ul >
< li > egl/dri: Avoid out of bounds array access< / li >
< / ul >
< p > Khaled Emara (1):< / p >
< ul >
< li > freedreno: PIPE_CAP_SHADER_BUFFER_OFFSET_ALIGNMENT unreachable statement< / li >
< / ul >
< p > Khem Raj (1):< / p >
< ul >
< li > winsys/svga/drm: Include sys/types.h< / li >
< / ul >
< p > Kishore Kadiyala (1):< / p >
< ul >
< li > android: static link with libexpat with Android O+< / li >
< / ul >
< p > Konstantin Kharlamov (1):< / p >
< ul >
< li > mapi: work around GCC LTO dropping assembly-defined functions< / li >
< / ul >
< p > Kristian Høgsberg (49):< / p >
< ul >
< li > st/nir: Use src/ relative include path for autotools< / li >
< li > freedreno/a6xx: Emit blitter dst with OUT_RELOCW< / li >
< li > freedreno/a6xx: Use tiling for all resources< / li >
< li > freedreno/a6xx: regen headers< / li >
< li > freedreno/a6xx: Drop render condition check in blitter< / li >
< li > freedreno: Log number of draw for sysmem passes< / li >
< li > freedreno/a6xx: Use the right resource for separate stencil stride< / li >
< li > freedreno/a6xx: Combine emit_blit and fd6_blit< / li >
< li > freedreno: Consolidate u_blitter functions in freedreno_blitter.c< / li >
< li > freedreno: Don't tell the blitter what it can't do< / li >
< li > freedreno/a6xx: Move blit check so as to restore comment< / li >
< li > freedreno/a6xx: Support some depth/stencil blits on blitter< / li >
< li > freedreno/a6xx: Support y-inverted blits< / li >
< li > freedreno/a6xx: Add format argument to fd6_tex_swiz()< / li >
< li > freedreno/a6xx: Fall back to masked RGBA blits for depth/stencil< / li >
< li > freedreno/a6xx: Clean up mixed use of swap and swizzle for texture state< / li >
< li > freedreno/a6xx: Update headers< / li >
< li > freedreno/a6xx: Front facing needs UNK3 bit< / li >
< li > freedreno/a6xx: Fix point coord< / li >
< li > .mailmap: Add a few more alises for myself< / li >
< li > freedreno: Update headers< / li >
< li > freedreno/a6xx: Copy stencil as R8_UINT< / li >
< li > freedreno/a6xx: Support MSAA resolve blits on blitter< / li >
< li > freedreno/a6xx: Only output MRT control for used framebuffers< / li >
< li > freedreno/a6xx: Don't zero SO buffer addresses< / li >
< li > freedreno: Fix a couple of warnings< / li >
< li > turnip: Only get bo offset when we need to mmap< / li >
< li > freedreno: Use c_vis_args and no_override_init_args< / li >
< li > freedreno/a6xx: Remove extra parens< / li >
< li > freedreno/ir3: Track whether shader needs derivatives< / li >
< li > freedreno/ir3: Fix operand order for DSX/DSY< / li >
< li > st/glsl_to_nir: Calculate num_uniforms from NumParameterValues< / li >
< li > freedreno/ir3: Enable PIPE_CAP_PACKED_UNIFORMS< / li >
< li > freedreno/ir3: Push UBOs to constant file< / li >
< li > freedreno/ir3: Don't access beyond available regs< / li >
< li > freedreno/ir3: Add workaround for VS samgq< / li >
< li > freedreno/ir3: Mark ir3_context_error() as NORETURN< / li >
< li > freedreno/a2xx: Fix redundant if statement< / li >
< li > freedreno: Use enum values from matching enum< / li >
< li > freedreno/a6xx: Add helper for incrementing regid< / li >
< li > freedreno: Fix format string warning< / li >
< li > .gitignore: Remove autotool artifacts< / li >
< li > tgsi: Mark tgsi_strings_check() unused< / li >
< li > glsl_to_nir: Initialize debug variable< / li >
< li > nir_opcodes.py: Saturate to expression that doesn't overflow< / li >
< li > ralloc: Fully qualify non-virtual destructor call< / li >
< li > egl/dri2: Mark potentially unused 'display' variable with MAYBE_UNUSED< / li >
< li > gallium/auxiliary/vl: Fix a couple of warnings< / li >
< li > freedreno/drm: Quiet pointer to u64 conversion warning< / li >
< / ul >
< p > Leo Liu (6):< / p >
< ul >
< li > st/va: fix the incorrect max profiles report< / li >
< li > st/va/vp9: set max reference as default of VP9 reference number< / li >
< li > vl/dri3: remove the wait before getting back buffer< / li >
< li > radeon/vcn: add H.264 constrained baseline support< / li >
< li > radeon/vcn/vp9: search the render target from the whole list< / li >
< li > winsys/amdgpu: add VCN JPEG to no user fence group< / li >
< / ul >
< p > Lepton Wu (2):< / p >
< ul >
< li > virgl: close drm fd when destroying virgl screen.< / li >
< li > virgl: Set bind when creating temp resource.< / li >
< / ul >
< p > Lionel Landwerlin (127):< / p >
< ul >
< li > anv: assert that color attachment are valid< / li >
< li > radv: assert that colorAttachment is valid for CmdClearAttachment< / li >
< li > i965: scale factor changes should trigger recompile< / li >
< li > vulkan: Update the XML and headers to 1.1.101< / li >
< li > anv: implement VK_EXT_depth_clip_enable< / li >
< li > build: move imgui out of src/intel/tools to be reused< / li >
< li > imgui: bump copy< / li >
< li > imgui: make sure our copy of imgui doesn't clash with others in the same process< / li >
< li > vulkan: add an overlay layer< / li >
< li > intel: fix urb size for CFL GT1< / li >
< li > anv: add support for INTEL_DEBUG=bat< / li >
< li > Revert "anv: add support for INTEL_DEBUG=bat"< / li >
< li > intel/aub_viewer: printout 48bits addresses< / li >
< li > intel/aub_viewer: silence compiler warning< / li >
< li > intel/aub_viewer: silence more compiler warnings< / li >
< li > vulkan/overlay: fix missing installation of layer< / li >
< li > vulkan/overlay: fix includes< / li >
< li > imgui: update commit< / li >
< li > imgui: update memory editor< / li >
< li > vulkan/overlay: install layer binary in libdir< / li >
< li > intel/compiler: use correct swizzle for replacement< / li >
< li > vulkan/overlay: fix min/max computations< / li >
< li > vulkan/overlay: rework option parsing< / li >
< li > vulkan/overlay: add support for fps output in file< / li >
< li > anv: add support for INTEL_DEBUG=bat< / li >
< li > vulkan: update headers/registry to 1.1.102< / li >
< li > anv: update supported patch version< / li >
< li > radv: set num_components on vulkan_resource_index intrinsic< / li >
< li > vulkan/util: make header available from c++< / li >
< li > vulkan/util: generate instance/device dispatch tables< / li >
< li > vulkan/overlay: drop dependency on validation layer headers< / li >
< li > intel/decoders: add address space indicator to get BOs< / li >
< li > intel/decoders: handle decoding MI_BBS from ring< / li >
< li > intel/decoders: limit number of decoded batchbuffers< / li >
< li > intel/aub_read: reuse defines from gen_context< / li >
< li > intel/aub_write: split comment section from HW setup< / li >
< li > intel/aub_write: write header in init< / li >
< li > intel/aub_write: break execlist write in 2< / li >
< li > intel/aub_write: switch to use i915_drm engine classes< / li >
< li > intel/aub_write: log mmio writes< / li >
< li > intel/aub_write: store the physical page allocator in struct< / li >
< li > intel/aub_write: turn context images arrays into functions< / li >
< li > intel/aub_write: factorize context image/pphwsp/ring creation< / li >
< li > iris: fix decoder call< / li >
< li > iris: fix decode_get_bo callback< / li >
< li > intel/error2aub: build a list of BOs before writing them< / li >
< li > intel/error2aub: identify buffers by engine< / li >
< li > intel/error2aub: strenghten batchbuffer identifier marker< / li >
< li > intel/error2aub: parse other buffer types< / li >
< li > intel/error2aub: annotate buffer with their address space< / li >
< li > intel/error2aub: store engine last ring buffer head/tail pointers< / li >
< li > intel/error2aub: write GGTT buffers into the aub file< / li >
< li > intel/error2aub: add a verbose option< / li >
< li > intel/error2aub: deal with GuC log buffer< / li >
< li > intel/error2aub: support older style engine names< / li >
< li > vulkan: factor out wsi dependencies< / li >
< li > anv: implement VK_EXT_pipeline_creation_feedback< / li >
< li > vulkan/overlay: properly register layer object with loader< / li >
< li > vulkan/overlay: silence validation layer warnings< / li >
< li > vulkan/overlay: check return value of swapchain get images< / li >
< li > vulkan/overlay: improve error reporting< / li >
< li > i965: perf: sklgt2: update a priority for register programming< / li >
< li > i965: perf: sklgt2: update compute metrics config< / li >
< li > i965: perf: sklgt2: update memory write config< / li >
< li > i965: perf: add PMA stall metrics< / li >
< li > i965: perf: chv: fixup counters names< / li >
< li > i965: perf: hsw: drop register programming not needed on HSW< / li >
< li > i965: perf: sklgt2: drop programming of an unused NOA register< / li >
< li > i965: perf: add Icelake metrics< / li >
< li > i965: perf: enable Icelake metrics< / li >
< li > i965: perf: add ring busyness metric for cfl gt2< / li >
< li > i965: perf: update render basic configs for big core gen9/gen10< / li >
< li > anv: implement VK_KHR_swapchain revision 70< / li >
< li > intel: add dependency on genxml generated files< / li >
< li > genxml: add a sorting script< / li >
< li > genxml: sort xml files using new script< / li >
< li > anv: don't use default pipeline cache for hits for VK_EXT_pipeline_creation_feedback< / li >
< li > anv: store heap address bounds when initializing physical device< / li >
< li > anv: leave the top 4Gb of the high heap VMA unused< / li >
< li > i965: store device revision in gen_device_info< / li >
< li > i965: extract performance query metrics< / li >
< li > i965: move mdapi data structure to intel/perf< / li >
< li > i965: move OA accumulation code to intel/perf< / li >
< li > i965: move brw_timebase_scale to device info< / li >
< li > i965: move mdapi result data format to intel/perf< / li >
< li > i965: move mdapi guid into intel/perf< / li >
< li > intel/perf: stub gen10/11 missing definitions< / li >
< li > i965: perf: add mdapi pipeline statistics queries on gen10/11< / li >
< li > intel/perf: drop counter size field< / li >
< li > intel/perf: constify accumlator parameter< / li >
< li > iris: implement WaEnableStateCacheRedirectToCS< / li >
< li > i965: implement WaEnableStateCacheRedirectToCS< / li >
< li > anv: implement WaEnableStateCacheRedirectToCS< / li >
< li > anv: fix uninitialized pthread cond clock domain< / li >
< li > intel/devinfo: fix missing num_thread_per_eu on ICL< / li >
< li > intel/devinfo: add basic sanity tests on device database< / li >
< li > anv: limit URB reconfigurations when using blorp< / li >
< li > intel: workaround VS fixed function issue on Gen9 GT1 parts< / li >
< li > anv: fix argument name for vkCmdEndQuery< / li >
< li > i965: fix icelake performance query enabling< / li >
< li > Revert "anv: limit URB reconfigurations when using blorp"< / li >
< li > vulkan/util: generate a helper function to return pNext struct sizes< / li >
< li > vulkan/overlay: update help printout< / li >
< li > vulkan/overlay: record stats in command buffers and accumulate on exec/submit< / li >
< li > vulkan/overlay: add pipeline statistic & timestamps support< / li >
< li > vulkan/overlay: add no display option< / li >
< li > vulkan/overlay: add a margin to the size of the window< / li >
< li > vulkan/overlay: record all select metrics into output file< / li >
< li > vulkan/overlay: add a frame counter option< / li >
< li > vulkan/overlay: make overlay size configurable< / li >
< li > vulkan/overlay: make overriden functions static< / li >
< li > vulkan/overlay: add TODO list< / li >
< li > anv: fix crash when application does not provide push constants< / li >
< li > anv: rework queries writes to ensure ordering memory writes< / li >
< li > anv: fix use after free< / li >
< li > anv: Use corresponding type from the vector allocation< / li >
< li > vulkan/overlay: keep allocating draw data until it can be reused< / li >
< li > nir: fix lower_non_uniform_access pass< / li >
< li > vulkan/overlay-layer: fix cast errors< / li >
< li > vulkan/overlay: fix truncating error on 32bit platforms< / li >
< li > nir: lower_non_uniform_access: iterate over instructions safely< / li >
< li > vulkan/overlay: fix timestamp query emission with no pipeline stats< / li >
< li > vulkan: fix build dependency issue with generated files< / li >
< li > anv: fix apply_pipeline_layout pass for arrays of YCbCr descriptors< / li >
< li > nir/lower_non_uniform: safely iterate over blocks< / li >
< li > intel/perf: fix EuThreadsCount value in performance equations< / li >
< li > intel/perf: improve dynamic loading config detection< / li >
< / ul >
< p > Lubomir Rintel (3):< / p >
< ul >
< li > kmsro: Extend to include armada-drm< / li >
< li > gallivm: guess CPU features also on ARM< / li >
< li > gallivm: disable NEON instructions if they are not supported< / li >
< / ul >
< p > Lucas Stach (3):< / p >
< ul >
< li > etnaviv: don't flush own context when updating resource use< / li >
< li > etnaviv: flush all pending contexts when accessing a resource with the CPU< / li >
< li > etnaviv: only try to construct scanout resource when on KMS winsys< / li >
< / ul >
< p > Marek Olšák (121):< / p >
< ul >
< li > radeonsi: enable dithered alpha-to-coverage for better quality< / li >
< li > radeonsi: merge & rename texture BO metadata functions< / li >
< li > radeonsi: unify error paths in si_texture_create_object< / li >
< li > winsys/amdgpu: remove amdgpu_drm.h definitions< / li >
< li > r600: add -Wstrict-overflow=0 to meson to silence the warning< / li >
< li > radeonsi: fix a comment typo in si_fine_fence_set< / li >
< li > gallium: allow more PIPE_RESOURCE_ driver flags< / li >
< li > meson: drop the xcb-xrandr version requirement< / li >
< li > radeonsi: handle render_condition_enable in si_compute_clear_render_target< / li >
< li > radeonsi: fix crashing performance counters (division by zero)< / li >
< li > radeonsi: initialize textures using DCC to black when possible< / li >
< li > radeonsi: clear allocator_zeroed_memory with SDMA< / li >
< li > radeonsi: make allocator_zeroed_memory unmappable and use bigger buffers< / li >
< li > radeonsi: don't leak an index buffer if draw_vbo fails< / li >
< li > radeonsi: use local ws variable in si_need_dma_space< / li >
< li > gallium/u_threaded: fix EXPLICIT_FLUSH for flush offsets > 0< / li >
< li > radeonsi: fix EXPLICIT_FLUSH for flush offsets > 0< / li >
< li > winsys/amdgpu: don't drop manually added fence dependencies< / li >
< li > winsys/amdgpu: unify fence list code< / li >
< li > winsys/amdgpu: use a separate fence list for syncobjs< / li >
< li > winsys/amdgpu: remove occurence of INDIRECT_BUFFER_CONST< / li >
< li > winsys/amdgpu: clean up IB buffer size computation< / li >
< li > winsys/amdgpu: cs_check_space sets the minimum IB size for future IBs< / li >
< li > radeonsi: add AMD_DEBUG env var as an alternative to R600_DEBUG< / li >
< li > radeonsi: use MEM instead of MEM_GRBM in COPY_DATA.DST_SEL< / li >
< li > radeonsi: add driconf option radeonsi_enable_nir< / li >
< li > radeonsi: always enable NIR for Civilization 6 to fix corruption< / li >
< li > driconf: add Civ6Sub executable for Civilization 6< / li >
< li > st/mesa: always unmap the uploader in st_atom_array.c< / li >
< li > gallium/u_threaded: always unmap const_uploader< / li >
< li > gallium/u_upload_mgr: allow use of FLUSH_EXPLICIT with persistent mappings< / li >
< li > radeonsi: use SDMA for uploading data through const_uploader< / li >
< li > tgsi: don't set tgsi_info::uses_bindless_images for constbufs and hw atomics< / li >
< li > radeonsi: always use compute rings for clover on CI and newer (v2)< / li >
< li > gallium/u_tests: use a compute-only context to test GCN compute ring< / li >
< li > gallium: add pipe_grid_info::last_block< / li >
< li > omx: clean up enc_LoadImage_common< / li >
< li > omx: add a compute path in enc_LoadImage_common< / li >
< li > radeonsi: fix assertion failure by using the correct type< / li >
< li > mesa: implement ARB/KHR_parallel_shader_compile< / li >
< li > gallium: implement ARB/KHR_parallel_shader_compile< / li >
< li > util/queue: move thread creation into a separate function< / li >
< li > util/queue: add ability to kill a subset of threads< / li >
< li > util/queue: hold a lock when reading num_threads in util_queue_finish< / li >
< li > util/queue: add util_queue_adjust_num_threads< / li >
< li > radeonsi: implement ARB/KHR_parallel_shader_compile callbacks< / li >
< li > radeonsi: don't use PFP_SYNC_ME with compute-only contexts< / li >
< li > docs/relnotes: document parallel_shader_compile changes in 19.1.0, not 19.0.0< / li >
< li > amd/addrlib: fix uninitialized values for Addr2ComputeDccAddrFromCoord< / li >
< li > radeonsi/gfx9: add support for PIPE_ALIGNED=0< / li >
< li > radeonsi: add ability to bind images as image buffers< / li >
< li > radeonsi: add support for displayable DCC for 1 RB chips< / li >
< li > radeonsi: add support for displayable DCC for multi-RB chips< / li >
< li > radeonsi: enable displayable DCC on Ravens< / li >
< li > gallium: add writable_bitmask parameter into set_shader_buffers< / li >
< li > glsl: remember which SSBOs are not read-only and pass it to gallium< / li >
< li > radeonsi: set exact shader buffer read/write usage in CS< / li >
< li > tegra: fix the build after the set_shader_buffers change< / li >
< li > radeonsi: fix a crash when unbinding sampler states< / li >
< li > glsl: fix shader_storage_blocks_write_access for SSBO block arrays< / li >
< li > Revert "glsl: fix shader_storage_blocks_write_access for SSBO block arrays"< / li >
< li > glsl: allow the #extension directive within code blocks for the dri option< / li >
< li > mesa: don't overwrite existing shader files with MESA_SHADER_CAPTURE_PATH< / li >
< li > radeonsi: set AC_FUNC_ATTR_READNONE for image opcodes where it was missing< / li >
< li > ac: use the common helper ac_apply_fmask_to_sample< / li >
< li > ac: fix incorrect bindless atomic code in visit_image_atomic< / li >
< li > radeonsi: enable GL_EXT_shader_image_load_formatted< / li >
< li > nir: optimize gl_SampleMaskIn to gl_HelperInvocation for radeonsi when possible< / li >
< li > winsys/amdgpu: don't set GTT with GDS & OA placements on APUs< / li >
< li > radeonsi/gfx9: use the correct condition for the DPBB + QUANT_MODE workaround< / li >
< li > radeonsi: use CP DMA for the null const buffer clear on CIK< / li >
< li > tgsi/scan: add uses_drawid< / li >
< li > ac: add radeon_info::marketing_name, replacing the winsys callback< / li >
< li > ac: add radeon_info::is_pro_graphics< / li >
< li > ac: add ac_get_i1_sgpr_mask< / li >
< li > ac: add REWIND and GDS registers to register headers< / li >
< li > winsys/amdgpu: make IBs writable and expose their address< / li >
< li > winsys/amdgpu: reorder chunks, make BO_HANDLES first, IB and FENCE last< / li >
< li > winsys/amdgpu: enable chaining for compute IBs< / li >
< li > winsys/amdgpu: clean up and remove nonsensical assertion< / li >
< li > radeonsi: add si_cp_copy_data< / li >
< li > radeonsi: add helper si_get_minimum_num_gfx_cs_dwords< / li >
< li > radeonsi: delay adding BOs at the beginning of IBs until the first draw< / li >
< li > gallium: document conservative rasterization flags< / li >
< li > st/dri: simplify throttling code< / li >
< li > gallium: replace DRM_CONF_THROTTLE with PIPE_CAP_MAX_FRAMES_IN_FLIGHT< / li >
< li > gallium: replace DRM_CONF_SHARE_FD with PIPE_CAP_DMABUF< / li >
< li > gallium: replace drm_driver_descriptor::configuration with driconf_xml< / li >
< li > gallium: set PIPE_CAP_MAX_FRAMES_IN_FLIGHT to 2 for all drivers< / li >
< li > gallium: add PIPE_CAP_PREFER_COMPUTE_BLIT_FOR_MULTIMEDIA< / li >
< li > util: fix a compile failure in u_compute.c on windows< / li >
< li > mesa: enable glGet for EXT_gpu_shader4< / li >
< li > glsl: add `unsigned int` type for EXT_GPU_shader4< / li >
< li > glsl: apply some 1.30 and other rules to EXT_gpu_shader4 as well< / li >
< li > glsl: add builtin variables for EXT_gpu_shader4< / li >
< li > glsl: add arithmetic builtin functions for EXT_gpu_shader4< / li >
< li > glsl: add texture builtin functions for EXT_gpu_shader4< / li >
< li > glsl: allow "varying out" for fragment shader outputs with EXT_gpu_shader4< / li >
< li > mesa: expose EXT_texture_buffer_object< / li >
< li > mesa: only allow EXT_gpu_shader4 in the compatibility profile< / li >
< li > st/mesa: expose EXT_gpu_shader4 if GLSL 1.40 is supported< / li >
< li > glsl: handle interactions between EXT_gpu_shader4 and texture extensions< / li >
< li > radeonsi: add BOs after need_cs_space< / li >
< li > radeonsi/gfx9: set that window_rectangles always roll the context< / li >
< li > radeonsi/gfx9: rework the gfx9 scissor bug workaround (v2)< / li >
< li > radeonsi: remove dirty slot masks from scissor and viewport states< / li >
< li > glsl: fix shader_storage_blocks_write_access for SSBO block arrays (v2)< / li >
< li > radeonsi: don't ignore PIPE_FLUSH_ASYNC< / li >
< li > mesa: rework error handling in glDrawBuffers< / li >
< li > mesa: fix pbuffers because internally they are front buffers< / li >
< li > st/mesa: don't flush the front buffer if it's a pbuffer< / li >
< li > radeonsi: use new atomic LLVM helpers< / li >
< li > radeonsi: set sampler state and view functions for compute-only contexts< / li >
< li > st/dri: decrease input lag by syncing sooner in SwapBuffers< / li >
< li > glsl: fix and clean up NV_compute_shader_derivatives support< / li >
< li > st/mesa: fix 2 crashes in st_tgsi_lower_yuv< / li >
< li > radeonsi: remove old_va parameter from si_rebind_buffer by remembering offsets< / li >
< li > radeonsi: update buffer descriptors in all contexts after buffer invalidation< / li >
< li > radeonsi: fix a regression in si_rebind_buffer< / li >
< li > u_blitter: don't fail mipmap generation for depth formats containing stencil< / li >
< li > ac: fix a typo in ac_build_wg_scan_bottom< / li >
< / ul >
< p > Mario Kleiner (1):< / p >
< ul >
< li > drirc: Add sddm-greeter to adaptive_sync blacklist.< / li >
< / ul >
< p > Mark Janes (5):< / p >
< ul >
< li > mesa: properly report the length of truncated log messages< / li >
< li > mesa: rename logging functions to reflect that they format strings< / li >
< li > mesa: add logging function for formatted string< / li >
< li > intel/common: move gen_debug to intel/dev< / li >
< li > intel/tools: Remove redundant definitions of INTEL_DEBUG< / li >
< / ul >
< p > Mateusz Krzak (2):< / p >
< ul >
< li > panfrost: cast bo_handles pointer to uintptr_t first< / li >
< li > panfrost: use os_mmap and os_munmap< / li >
< / ul >
< p > Mathias Fröhlich (22):< / p >
< ul >
< li > st/mesa: Reduce array updates due to current changes.< / li >
< li > mesa: Track buffer object use also for VAO usage.< / li >
< li > st/mesa: Invalidate the gallium array atom only if needed.< / li >
< li > mesa: Implement helper functions to map and unmap a VAO.< / li >
< li > mesa: Factor out _mesa_array_element.< / li >
< li > mesa: Use _mesa_array_element in dlist save.< / li >
< li > mesa: Replace _ae_{,un}map_vbos with _mesa_vao_{,un}map_arrays< / li >
< li > mesa: Remove _ae_{,un}map_vbos and dependencies.< / li >
< li > mesa: Use mapping tools in debug prints.< / li >
< li > vbo: Fix basevertex handling in display list compiles.< / li >
< li > vbo: Fix GL_PRIMITIVE_RESTART_FIXED_INDEX in display list compiles.< / li >
< li > mesa: Add assert to _mesa_primitive_restart_index.< / li >
< li > mesa: Factor out index function that will have multiple use.< / li >
< li > mesa: Use glVertexAttrib*NV functions for fixed function attribs.< / li >
< li > mesa: Implement _mesa_array_element by walking enabled arrays.< / li >
< li > mesa: Rip out now unused gl_context::aelt_context.< / li >
< li > mesa: Remove the now unused _NEW_ARRAY state change flag.< / li >
< li > mesa: Constify static const array in api_arrayelt.c< / li >
< li > mesa: Remove the _glapi_table argument from _mesa_array_element.< / li >
< li > mesa: Set CurrentSavePrimitive in vbo_save_NotifyBegin.< / li >
< li > mesa: Correct the is_vertex_position decision for dlists.< / li >
< li > mesa: Leave aliasing of vertex and generic0 attribute to the dlist code.< / li >
< / ul >
< p > Matt Turner (7):< / p >
< ul >
< li > intel/compiler/test: Set devinfo-> gen = 7< / li >
< li > intel/compiler: Avoid propagating inequality cmods if types are different< / li >
< li > intel/compiler/test: Add unit test for mismatched signedness comparison< / li >
< li > intel/compiler: Add commas on final values of compaction table arrays< / li >
< li > intel/compiler: Use SIMD16 instructions in fs saturate prop unit test< / li >
< li > intel/compiler: Add unit tests for sat prop for different exec sizes< / li >
< li > intel/compiler: Improve fix_3src_operand()< / li >
< / ul >
< p > Matthias Lorenz (1):< / p >
< ul >
< li > vulkan/overlay: Add fps counter< / li >
< / ul >
< p > Mauro Rossi (6):< / p >
< ul >
< li > android: intel/isl: remove redundant building rules< / li >
< li > android: anv: fix generated files depedencies (v2)< / li >
< li > android: anv: fix libexpat shared dependency< / li >
< li > android: nouveau: add support for nir< / li >
< li > android: fix LLVM version string related building errors< / li >
< li > draw: fix building error in draw_gs_init()< / li >
< / ul >
< p > Maya Rashish (1):< / p >
< ul >
< li > configure: fix test portability< / li >
< / ul >
< p > Michel Dänzer (19):< / p >
< ul >
< li > loader/dri3: Use strlen instead of sizeof for creating VRR property atom< / li >
< li > gitlab-ci: Re-use docker image from the main repo in forked repos< / li >
< li > gitlab-ci: List some longer-running jobs before others of the same stage< / li >
< li > gitlab-ci: Use 8 CPU cores in autotools job< / li >
< li > gitlab-ci: Make sure clang job actually uses ccache< / li >
< li > gitlab-ci: Only pull/push cache contents in build+test stage jobs< / li >
< li > gitlab-ci: Automatically retry jobs after runner system failure< / li >
< li > gitlab-ci: Run CI pipeline for all branches in the main repository< / li >
< li > gitlab-ci: Use Debian stretch instead of Ubuntu bionic< / li >
< li > gitlab-ci: Use HTTPS for APT repositories< / li >
< li > gitlab-ci: Use Debian packages instead of pip ones for meson and scons< / li >
< li > gitlab-ci: Install most packages from Debian buster< / li >
< li > gitlab-ci: Remove unneded (stuff from) APT command lines< / li >
< li > gitlab-ci: Remove unused Debian packages from Docker image< / li >
< li > gitlab-ci: Use clang 8 instead of 7< / li >
< li > gitlab-ci: Drop unused clang 5/6 packages< / li >
< li > gitlab-ci: Do not use subshells for compiling dependencies< / li >
< li > gitlab-ci: Use LLVM 3.4 from Debian jessie for scons-llvm job< / li >
< li > gitlab-ci: Use meson buildtype debug instead of default debugoptimized< / li >
< / ul >
< p > Mike Blumenkrantz (6):< / p >
< ul >
< li > iris: support INTEL_NO_HW environment variable< / li >
< li > gallium: add pipe cap for inner_coverage conservative raster mode< / li >
< li > st/mesa: indicate intel extension support for inner_coverage based on cap< / li >
< li > iris: add support for INTEL_conservative_rasterization< / li >
< li > iris: add preemption support on gen9< / li >
< li > iris: enable preemption support for gen10< / li >
< / ul >
< p > Nanley Chery (3):< / p >
< ul >
< li > i965: Rename intel_mipmap_tree::r8stencil_* -> ::shadow_*< / li >
< li > anv: Fix some depth buffer sampling cases on ICL+< / li >
< li > anv/cmd_buffer: Initalize the clear color struct for CNL+< / li >
< / ul >
< p > Nataraj Deshpande (1):< / p >
< ul >
< li > anv: Fix check for isl_fmt in assert< / li >
< / ul >
< p > Neha Bhende (2):< / p >
< ul >
< li > st/mesa: Fix topogun-1.06-orc-84k-resize.trace crash< / li >
< li > draw: fix memory leak introduced 7720ce32a< / li >
< / ul >
< p > Nicolai Hähnle (9):< / p >
< ul >
< li > amd/surface: provide firstMipIdInTail for metadata surface calculations< / li >
< li > radeonsi: add si_debug_options for convenient adding/removing of options< / li >
< li > util/u_log: flush auto loggers before starting a new page< / li >
< li > ddebug: set thread name< / li >
< li > ddebug: log calls to pipe-> flush< / li >
< li > ddebug: dump driver state into a separate file< / li >
< li > ddebug: expose some helper functions as non-inline< / li >
< li > radeonsi: add radeonsi_aux_debug option for aux context debug dumps< / li >
< li > radeonsi: add radeonsi_sync_compile option< / li >
< / ul >
< p > Oscar Blumberg (3):< / p >
< ul >
< li > intel/fs: Fix memory corruption when compiling a CS< / li >
< li > radeonsi: Fix guardband computation for large render targets< / li >
< li > glsl: Fix function return typechecking< / li >
< / ul >
< p > Patrick Lerda (1):< / p >
< ul >
< li > lima/ppir: fix pointer referenced after a free< / li >
< / ul >
< p > Patrick Rudolph (1):< / p >
< ul >
< li > d3dadapter9: Support software renderer on any DRI device< / li >
< / ul >
< p > Philipp Zabel (1):< / p >
< ul >
< li > etnaviv: fill missing offset in etna_resource_get_handle< / li >
< / ul >
< p > Pierre Moreau (12):< / p >
< ul >
< li > include/CL: Update to the latest OpenCL 2.2 headers< / li >
< li > clover: Avoid warnings from new OpenCL headers< / li >
< li > clover: Remove the TGSI backend as unused< / li >
< li > clover: Add an helper for checking if an IR is supported< / li >
< li > clover/api: Rework the validation of devices for building< / li >
< li > clover/api: Fail if trying to build a non-executable binary< / li >
< li > clover: Disallow creating libraries from other libraries< / li >
< li > clover: Validate program and library linking options< / li >
< li > clover: Move device extensions definitions to core/device.cpp< / li >
< li > clover: Move platform extensions definitions to clover/platform.cpp< / li >
< li > clover: Only use devices supporting IR_NATIVE< / li >
< li > clover: Fix indentation issues< / li >
< / ul >
< p > Pierre-Eric Pelloux-Prayer (1):< / p >
< ul >
< li > radeonsi: init sctx-> dma_copy before using it< / li >
< / ul >
< p > Plamena Manolova (3):< / p >
< ul >
< li > i965: Disable ARB_fragment_shader_interlock for platforms prior to GEN9< / li >
< li > isl: Set ClearColorConversionEnable.< / li >
< li > i965: Re-enable fast color clears for GEN11.< / li >
< / ul >
< p > Qiang Yu (9):< / p >
< ul >
< li > u_math: add ushort_to_float/float_to_ushort< / li >
< li > u_dynarray: add util_dynarray_grow_cap< / li >
< li > gallium/u_vbuf: export u_vbuf_get_minmax_index< / li >
< li > drm-uapi: add lima_drm.h< / li >
< li > gallium: add lima driver< / li >
< li > lima/gpir: fix compile fail when two slot node< / li >
< li > lima/gpir: fix alu check miss last store slot< / li >
< li > lima: fix lima_blit with non-zero level source resource< / li >
< li > lima: fix render to non-zero level texture< / li >
< / ul >
< p > Rafael Antognolli (45):< / p >
< ul >
< li > iris: Store internal_format when getting resource from handle.< / li >
< li > iris: Skip msaa16 on gen < 9.< / li >
< li > iris: Flush before hiz_exec.< / li >
< li > iris: Pin HiZ buffers when rendering.< / li >
< li > iris: Avoid leaking if we fail to allocate the aux buffer.< / li >
< li > iris/clear: Pass on render_condition_enabled.< / li >
< li > iris: Skip resolve if there's no context.< / li >
< li > iris: Flag ALL_DIRTY_BINDINGS on aux state change.< / li >
< li > iris: Add resolve on iris_flush_resource.< / li >
< li > iris: Convert RGBX to RGBA always.< / li >
< li > iris: Enable auxiliary buffer support again< / li >
< li > iris: Enable HiZ for multisampled depth surfaces.< / li >
< li > iris: Make intel_hiz_exec public.< / li >
< li > iris: Allocate buffer space for the fast clear color.< / li >
< li > iris: Use the clear depth when emitting 3DSTATE_CLEAR_PARAMS.< / li >
< li > iris: Fast clear depth buffers.< / li >
< li > iris: Add helper to convert fast clear color.< / li >
< li > iris: Add function to update clear color in surface state.< / li >
< li > iris: Bring back check for srgb and fast clear color.< / li >
< li > intel/isl: Add isl_format_has_color_component() function.< / li >
< li > intel/blorp: Make swizzle_color_value public.< / li >
< li > iris: Implement fast clear color.< / li >
< li > iris: Add iris_resolve_conditional_render().< / li >
< li > iris: Stall on the CPU and resolve predication during fast clears.< / li >
< li > iris: Track fast clear color.< / li >
< li > iris: Let blorp update the clear color for us.< / li >
< li > i965/blorp: Remove unused parameter from blorp_surf_for_miptree.< / li >
< li > iris: Only update clear color for gens 8 and 9.< / li >
< li > iris/gen8: Re-emit the SURFACE_STATE if the clear color changed.< / li >
< li > iris: Manually apply fast clear color channel overrides.< / li >
< li > iris: Do not allocate clear_color_bo for gen8.< / li >
< li > iris: Add aux.sampler_usages.< / li >
< li > iris: Enable fast clears on gen8.< / li >
< li > intel/fs: Only propagate saturation if exec_size is the same.< / li >
< li > intel/fs: Move the scalar-region conversion to the generator.< / li >
< li > intel/fs: Add a lowering pass for linear interpolation.< / li >
< li > intel/fs: Remove fs_generator::generate_linterp from gen11+.< / li >
< li > intel/isl: Resize clear color buffer to full cacheline< / li >
< li > intel/genxml: Update MI_ATOMIC genxml definition.< / li >
< li > intel/blorp: Make blorp update the clear color in gen11.< / li >
< li > iris: Do not advertise multisampled image load/store.< / li >
< li > iris: Support sRGB fast clears even if the colorspaces differ.< / li >
< li > iris: Use the linear version of the surface format during fast clears.< / li >
< li > iris: Update the surface state clear color address when available.< / li >
< li > iris: Enable fast clear colors on gen11.< / li >
< / ul >
< p > Ray Zhang (1):< / p >
< ul >
< li > glx: fix shared memory leak in X11< / li >
< / ul >
< p > Rhys Kidd (1):< / p >
< ul >
< li > iris: Fix assertion in iris_resource_from_handle() tiling usage< / li >
< / ul >
< p > Rhys Perry (28):< / p >
< ul >
< li > nvc0: add compute invocation counter< / li >
< li > radv: bitcast 16-bit outputs to integers< / li >
< li > radv: ensure export arguments are always float< / li >
< li > ac/nir: implement 8-bit nir_load_const_instr< / li >
< li > ac/nir: fix 64-bit nir_op_f2f16_rtz< / li >
< li > ac/nir: make ac_build_clamp work on all bit sizes< / li >
< li > ac/nir: make ac_build_isign work on all bit sizes< / li >
< li > ac/nir: make ac_build_fdiv support 16-bit floats< / li >
< li > ac/nir: implement half-float nir_op_frcp< / li >
< li > ac/nir: implement half-float nir_op_frsq< / li >
< li > ac/nir: implement half-float nir_op_ldexp< / li >
< li > ac/nir: fix 16-bit ssbo stores< / li >
< li > ac/nir: implement 8-bit push constant, ssbo and ubo loads< / li >
< li > ac/nir: implement 8-bit ssbo stores< / li >
< li > ac/nir: add 8-bit types to glsl_base_to_llvm_type< / li >
< li > ac/nir: implement 8-bit conversions< / li >
< li > radv: enable VK_KHR_8bit_storage< / li >
< li > ac/nir: implement 16-bit pack/unpack opcodes< / li >
< li > radv: lower 16-bit flrp< / li >
< li > ac: add 16-bit support to ac_build_ddxy()< / li >
< li > nir,ac/nir: fix cube_face_coord< / li >
< li > gallium: add support for formatted image loads< / li >
< li > mesa, glsl: add support for EXT_shader_image_load_formatted< / li >
< li > st/mesa: add support for EXT_shader_image_load_formatted< / li >
< li > vc4: fix build< / li >
< li > ac,ac/nir: use a better sync scope for shared atomics< / li >
< li > radv: fix set_output_usage_mask() with composite and 64-bit types< / li >
< li > ac/nir: mark some texture intrinsics as convergent< / li >
< / ul >
< p > Rob Clark (135):< / p >
< ul >
< li > freedreno: fix release tarball< / li >
< li > freedreno: more fixing release tarball< / li >
< li > freedreno/a6xx: small compiler warning fix< / li >
< li > freedreno/ir3: fix varying packing vs. tex sharp edge< / li >
< li > freedreno/a6xx: move stream-out emit to helper< / li >
< li > freedreno/a6xx: clean up some open-coded bits< / li >
< li > freedreno/ir3: split out image helpers< / li >
< li > freedreno/ir3: split out a4xx+ instructions< / li >
< li > freedreno/ir3: fix ncomp for _store_image() src< / li >
< li > freedreno/ir3: add image/ssbo < -> ibo/tex mapping< / li >
< li > freedreno/ir3: add a6xx instruction encoding< / li >
< li > freedreno/ir3: add a6xx+ SSBO/image support< / li >
< li > freedreno/ir3: HIGH reg w/a for a6xx< / li >
< li > freedreno/a6xx: border-color offset helper< / li >
< li > freedreno/a6xx: image/ssbo state emit< / li >
< li > freedreno/a6xx: compute support< / li >
< li > freedreno/a6xx: cache flush harder< / li >
< li > freedreno/a6xx: fix helper_invocation (sampler mask/id)< / li >
< li > freedreno/ir3: handle quirky atomic dst for a6xx< / li >
< li > freedreno/ir3: fix legalize for vecN inputs< / li >
< li > freedreno/ir3: fix crash in compile fail case< / li >
< li > freedreno/a6xx: 3d and cube image fixes< / li >
< li > freedreno: fix crash w/ masked non-SSA dst< / li >
< li > freedreno/ir3: rename put_dst()< / li >
< li > freedreno/ir3/a6xx: fix load_ssbo barrier type.< / li >
< li > freedreno/ir3: sync instr/disasm and add ldib encoding< / li >
< li > freedreno/ir3/a6xx: use ldib for ssbo reads< / li >
< li > freedreno/a6xx: samplerBuffer fixes< / li >
< li > freedreno/a6xx: enable tiled images< / li >
< li > freedreno: fix race condition< / li >
< li > freedreno/ir3: don't hardcode wrmask< / li >
< li > freedreno/a6xx: fix border-color offset< / li >
< li > freedreno/a6xx: cube image fix< / li >
< li > freedreno/a6xx: fix hangs with large shaders< / li >
< li > freedreno/ir3: use nopN encoding when possible< / li >
< li > freedreno/a6xx: fix ssbo alignment< / li >
< li > freedreno/ir3/a6xx: fix non-ssa atomic dst< / li >
< li > freedreno/a6xx: fix DRAW_IDX_INDIRECT max_indicies< / li >
< li > freedreno/a6xx: vertex_id is not _zero_based< / li >
< li > freedreno/ir3/a6xx: fix atomic shader outputs< / li >
< li > freedreno/ir3: gsampler2DMSArray fixes< / li >
< li > freedreno/ir3: include nopN in expanded instruction count< / li >
< li > freedreno/ir3: add Sethi– Ullman numbering pass< / li >
< li > freedreno/ir3: track register pressure in sched< / li >
< li > freedreno: fix ir3_cmdline build< / li >
< li > freedreno/a6xx: remove astc_srgb workaround< / li >
< li > freedreno/a6xx: refactor fd6_tex_swiz()< / li >
< li > freedreno/a6xx: fix border-color swizzles< / li >
< li > freedreno/a6xx: perfcntrs< / li >
< li > freedreno/ir3: fix ir3_cmdline harder< / li >
< li > freedreno/ir3: turn on [iu]mul_high< / li >
< li > freedreno/a6xx: more bcolor fixes< / li >
< li > freedreno/ir3/cp: fix ldib bug< / li >
< li > freedreno/ir3/a6xx: fix ssbo comp_swap< / li >
< li > freedreno/ir3 better cat6 encoding detection< / li >
< li > freedreno/ir3/ra: fix half-class conflicts< / li >
< li > freedreno/ir3: fix sam.s2en decoding< / li >
< li > freedreno/ir3: fix sam.s2en encoding< / li >
< li > freedreno/ir3: fix regmask for merged regs< / li >
< li > nir: move gls_type_get_{sampler,image}_count()< / li >
< li > freedreno/ir3: find # of samplers from uniform vars< / li >
< li > freedreno/ir3: enable indirect tex/samp (sam.s2en)< / li >
< li > freedreno/ir3: optimize sam.s2en to sam< / li >
< li > freedreno/ir3: additional lowering< / li >
< li > freedreno/ir3: fix bit_count< / li >
< li > freedreno/ir3: dynamic UBO indexing vs 64b pointers< / li >
< li > freedreno/ir3: rename has_kill to no_earlyz< / li >
< li > freedreno/ir3: disable early-z for SSBO/image writes< / li >
< li > gallium: add PIPE_CAP_ESSL_FEATURE_LEVEL< / li >
< li > mesa/st: use ESSL cap top enable gpu_shader5< / li >
< li > freedreno: add ESSL cap< / li >
< li > docs: update freedreno status< / li >
< li > freedreno/a6xx: small cleanup< / li >
< li > freedreno/ir3: sched fix< / li >
< li > freedreno/ir3: reads/writes to unrelated arrays are not dependent< / li >
< li > freedreno/ir3: align const size to vec4< / li >
< li > nir: print var name for load_interpolated_input too< / li >
< li > nir: add lower_all_io_to_elements< / li >
< li > freedreno/ir3: re-indent comment< / li >
< li > freedreno/ir3: rework varying packing< / li >
< li > freedreno/ir3: add pass to move varying loads< / li >
< li > freedreno/ir3: convert to "new style" frag inputs< / li >
< li > gallium/docs: clarify set_sampler_views (v2)< / li >
< li > iris: fix set_sampler_view< / li >
< li > freedreno/ir3: fix const assert< / li >
< li > freedreno/drm: update for robustness< / li >
< li > freedreno: add robustness support< / li >
< li > compiler: rename SYSTEM_VALUE_VARYING_COORD< / li >
< li > freedreno/ir3: fix rgetpos decoding< / li >
< li > freedreno/ir3: more emit-cat5 fixes< / li >
< li > freedreno/ir3: cleanup instruction builder macros< / li >
< li > freedreno: update generated headers< / li >
< li > freedreno/ir3: lower load_barycentric_at_sample< / li >
< li > freedreno/ir3: lower load_barycentric_at_offset< / li >
< li > freedreno/ir3: remove bogus assert< / li >
< li > freedreno/ir3: rename frag_vcoord -> ij_pixel< / li >
< li > freedreno/a6xx: add VALIDREG/CONDREG helper macros< / li >
< li > freedreno/ir3: fix load_interpolated_input slot< / li >
< li > freedreno: wire up core sample-shading support< / li >
< li > freedreno/ir3: sample-shading support< / li >
< li > freedreno/a6xx: sample-shading support< / li >
< li > docs/features: update GL too< / li >
< li > freedreno/ir3: switch fragcoord to sysval< / li >
< li > freedreno/a6xx: small texture emit cleanup< / li >
< li > freedreno/a6xx: pre-bake UBWC flags in texture-view< / li >
< li > freedreno/ir3: fixes for half reg in/out< / li >
< li > freedreno/ir3: fix shader variants vs UBO analysis< / li >
< li > freedreno/ir3: fix lowered ubo region alignment< / li >
< li > freedreno/ir3: add IR3_SHADER_DEBUG flag to disable ubo lowering< / li >
< li > freedreno/ir3: add some ubo range related asserts< / li >
< li > nir: rework tex instruction printing< / li >
< li > nir: fix lower_wpos_ytransform in load_frag_coord case< / li >
< li > nir: add pass to lower fb reads< / li >
< li > freedreno/drm: expose GMEM_BASE address< / li >
< li > freedreno/ir3: fb read support< / li >
< li > freedreno/a6xx: KHR_blend_equation_advanced support< / li >
< li > freedreno/a6xx: smaller hammer for fb barrier< / li >
< li > docs: mark KHR_blend_equation_advanced done on a6xx< / li >
< li > nir: fix nir tex print harder< / li >
< li > freedreno/ir3: remove assert< / li >
< li > freedreno/a6xx: OUT_RELOC vs OUT_RELOCW fixes< / li >
< li > freedreno: update generated headers< / li >
< li > freedreno/a6xx: UBWC fixes< / li >
< li > freedreno/a6xx: UBWC support for images< / li >
< li > freedreno: mark imported resources as valid< / li >
< li > freedreno/a6xx: buffer resources cannot be compressed< / li >
< li > freedreno: move UBWC color offset to fd_resource_offset()< / li >
< li > freedreno: add ubwc_enabled helper< / li >
< li > freedreno/a6xx: deduplicate a few lines< / li >
< li > freedreno: remove unused forward struct declaration< / li >
< li > freedreno/ir3: fix rasterflat/glxgears< / li >
< li > freedreno/ir3: set more barrier bits< / li >
< li > freedreno/a6xx: fix GPU crash on small render targets< / li >
< li > freedreno/a6xx: fix issues with gallium HUD< / li >
< li > freedreno/a6xx: fix hangs with newer sqe fw< / li >
< / ul >
< p > Rob Herring (2):< / p >
< ul >
< li > kmsro: Add lima renderonly support< / li >
< li > kmsro: Add platform support for exynos and sun4i< / li >
< / ul >
< p > Rodrigo Vivi (1):< / p >
< ul >
< li > intel: Add more PCI Device IDs for Coffee Lake and Ice Lake.< / li >
< / ul >
< p > Roland Scheidegger (2):< / p >
< ul >
< li > gallivm: fix bogus assert in get_indirect_index< / li >
< li > gallivm: fix saturated signed add / sub with llvm 9< / li >
< / ul >
< p > Romain Failliot (1):< / p >
< ul >
< li > docs: changed "Done" to "DONE" in features.txt< / li >
< / ul >
< p > Ross Burton (1):< / p >
< ul >
< li > Revert "meson: drop GLESv1 .so version back to 1.0.0"< / li >
< / ul >
< p > Ryan Houdek (1):< / p >
< ul >
< li > panfrost: Adds Bifrost shader disassembler utility< / li >
< / ul >
< p > Sagar Ghuge (10):< / p >
< ul >
< li > iris: Don't allocate a BO per query object< / li >
< li > nir/glsl: Add another way of doing lower_imul64 for gen8+< / li >
< li > glsl: [u/i]mulExtended optimization for GLSL< / li >
< li > nir/algebraic: Optimize low 32 bit extraction< / li >
< li > spirv: Allow [i/u]mulExtended to use new nir opcode< / li >
< li > iris: Refactor code to share 3DSTATE_URB_* packet< / li >
< li > iris: Track last VS URB entry size< / li >
< li > iris: Flag fewer dirty bits in BLORP< / li >
< li > intel/fs: Remove unused condition from opt_algebraic case< / li >
< li > intel/compiler: Fix assertions in brw_alu3< / li >
< / ul >
< p > Samuel Iglesias Gonsálvez (4):< / p >
< ul >
< li > isl: remove the cache line size alignment requirement< / li >
< li > isl: the display engine requires 64B alignment for linear surfaces< / li >
< li > radv: don't overwrite results in VkGetQueryPoolResults() when queries are not available< / li >
< li > radv: write availability status vkGetQueryPoolResults() when the data is not available< / li >
< / ul >
< p > Samuel Pitoiset (147):< / p >
< ul >
< li > radv/winsys: fix hash when adding internal buffers< / li >
< li > radv: fix build< / li >
< li > radv: bail out when no image transitions will be performed< / li >
< li > radv: remove unused radv_render_pass_attachment::view_mask< / li >
< li > radv: remove useless MAYBE_UNUSED in CmdBeginRenderPass()< / li >
< li > radv: add radv_cmd_buffer_begin_subpass() helper< / li >
< li > radv: move subpass image transitions to radv_cmd_buffer_begin_subpass()< / li >
< li > radv: store the list of attachments for every subpass< / li >
< li > radv: use the new attachments array when starting subpasses< / li >
< li > radv: determine the last subpass id for every attachments< / li >
< li > radv: handle final layouts at end of every subpass and render pass< / li >
< li > radv: move some render pass things to radv_render_pass_compile()< / li >
< li > radv: add radv_render_pass_add_subpass_dep() helper< / li >
< li > radv: track if subpasses have color attachments< / li >
< li > radv: handle subpass dependencies correctly< / li >
< li > radv: accumulate all ingoing external dependencies to the first subpass< / li >
< li > radv: execute external subpass barriers after ending subpasses< / li >
< li > radv: drop useless checks when resolving subpass color attachments< / li >
< li > radv: do not set preserveAttachments for internal render passes< / li >
< li > radv: don't flush src stages when dstStageMask == BOTTOM_OF_PIPE< / li >
< li > radv: fix compiler issues with GCC 9< / li >
< li > radv: gather more info about push constants< / li >
< li > radv: gather if shaders load dynamic offsets separately< / li >
< li > radv: keep track of the number of remaining user SGPRs< / li >
< li > radv: add support for push constants inlining when possible< / li >
< li > radv: fix using LOAD_CONTEXT_REG with old GFX ME firmwares on GFX8< / li >
< li > radv/winsys: fix BO list creation when RADV_DEBUG=allbos is set< / li >
< li > radv: always export gl_SampleMask when the fragment shader uses it< / li >
< li > ac: make use of ac_build_expand_to_vec4() in visit_image_store()< / li >
< li > radv: use MAX_{VBS,VERTEX_ATTRIBS} when defining max vertex input limits< / li >
< li > radv: store vertex attribute formats as pipeline keys< / li >
< li > radv: reduce the number of loaded channels for vertex input fetches< / li >
< li > radv: fix radv_fixup_vertex_input_fetches()< / li >
< li > radv: fix invalid element type when filling vertex input default values< / li >
< li > ac: add ac_build_llvm8_tbuffer_load() helper< / li >
< li > ac: use new LLVM 8 intrinsic when loading 16-bit values< / li >
< li > radv: write the alpha channel of MRT0 when alpha coverage is enabled< / li >
< li > radv: remove unused variable in gather_push_constant_info()< / li >
< li > radv: fix writing the alpha channel of MRT0 when alpha coverage is enabled< / li >
< li > radv: fix clearing attachments in secondary command buffers< / li >
< li > radv: fix out-of-bounds access when copying descriptors BO list< / li >
< li > radv: don't copy buffer descriptors list for samplers< / li >
< li > rav: use 32_AR instead of 32_ABGR when alpha coverage is required< / li >
< li > radv: allocate enough space in cmdbuf when starting a subpass< / li >
< li > radv: properly align the fence and EOP bug VA on GFX9< / li >
< li > radv: enable lower_mul_2x32_64< / li >
< li > Revert "radv: execute external subpass barriers after ending subpasses"< / li >
< li > radv: fix pointSizeRange limits< / li >
< li > radv: set the maximum number of IBs per submit to 192< / li >
< li > ac: rework typed buffers loads for LLVM 7< / li >
< li > radv: store more vertex attribute infos as pipeline keys< / li >
< li > radv: use typed buffer loads for vertex input fetches< / li >
< li > ac: add ac_build_{struct,raw}_tbuffer_load() helpers< / li >
< li > ac: use the raw tbuffer version for 16-bit SSBO loads< / li >
< li > radv: always initialize HTILE when the src layout is UNDEFINED< / li >
< li > radv: always load 3 channels for formats that need to be shuffled< / li >
< li > ac: use llvm.amdgcn.fract intrinsic for nir_op_ffract< / li >
< li > radv: fix binding transform feedback buffers< / li >
< li > ac: make use of ac_get_store_intr_attribs() where possible< / li >
< li > ac/nir: set attrib flags for SSBO and image store operations< / li >
< li > ac: add ac_build_buffer_store_format() helper< / li >
< li > ac/nir: remove one useless check in visit_store_ssbo()< / li >
< li > ac/nir: use new LLVM 8 intrinsics for SSBO atomic operations< / li >
< li > ac/nir: use ac_build_buffer_load() for SSBO load operations< / li >
< li > ac/nir: use ac_build_buffer_store_dword() for SSBO store operations< / li >
< li > ac: use new LLVM 8 intrinsics in ac_build_buffer_load()< / li >
< li > ac: add ac_build_{struct,raw}_tbuffer_store() helpers< / li >
< li > ac: use new LLVM 8 intrinsic when storing 16-bit values< / li >
< li > ac: use new LLVM 8 intrinsics in ac_build_buffer_store_dword()< / li >
< li > ac: add various int8 definitions< / li >
< li > ac: add ac_build_tbuffer_load_byte() helper< / li >
< li > ac: add ac_build_tbuffer_store_byte() helper< / li >
< li > radv: add missing initializations since VK_EXT_pipeline_creation_feedback< / li >
< li > ac: add f16_0 and f16_1 constants< / li >
< li > ac: add 16-bit support fo fsign< / li >
< li > ac: add 16-bit support to fract< / li >
< li > ac: fix 16-bit shifts< / li >
< li > ac: fix incorrect argument type for tbuffer.{load,store} with LLVM 7< / li >
< li > nir: use generic float types for frexp_exp and frexp_sig< / li >
< li > spirv,nir: lower frexp_exp/frexp_sig inside a new NIR pass< / li >
< li > nir: add nir_{load,store}_deref_with_access() helpers< / li >
< li > spirv: propagate the access flag for store and load derefs< / li >
< li > ac: use llvm.amdgcn.fmed3 intrinsic for nir_op_fmed3< / li >
< li > ac: add ac_build_frexp_mant() helper and 16-bit/32-bit support< / li >
< li > ac: add ac_build_frex_exp() helper ans 16-bit/32-bit support< / li >
< li > radv: do not lower frexp_exp and frexp_sig< / li >
< li > radv: enable VK_AMD_gpu_shader_int16< / li >
< li > radv: skip updating depth/color metadata for conditional rendering< / li >
< li > radv: do not always initialize HTILE in compressed state< / li >
< li > ac: fix return type for llvm.amdgcn.frexp.exp.i32.64< / li >
< li > ac/nir: fix nir_op_b2i16< / li >
< li > ac: fix ac_build_bit_count() for 16-bit integer type< / li >
< li > ac: fix ac_build_bitfield_reverse() for 16-bit integer type< / li >
< li > ac: fix ac_find_lsb() for 16-bit integer type< / li >
< li > ac: fix ac_build_umsb() for 16-bit integer type< / li >
< li > ac/nir: add support for nir_op_b2i8< / li >
< li > ac: add 8-bit support to ac_build_bit_count()< / li >
< li > ac: add 8-bit support to ac_find_lsb()< / li >
< li > ac: add 8-bit support to ac_build_umsb()< / li >
< li > ac: add 8-bit and 64-bit support to ac_build_bitfield_reverse()< / li >
< li > radv: partially enable VK_KHR_shader_float16_int8< / li >
< li > nir: do not pack varying with different types< / li >
< li > ac/nir: fix intrinsic names for atomic operations with LLVM 9+< / li >
< li > radv: fix getting the vertex strides if the bindings aren't contiguous< / li >
< li > ac/nir: fix nir_op_b2f16< / li >
< li > radv: enable VK_AMD_gpu_shader_half_float< / li >
< li > wsi: allow to override the present mode with MESA_VK_WSI_PRESENT_MODE< / li >
< li > ac/nir: make use of ac_build_imax() where possible< / li >
< li > ac/nir: make use of ac_build_imin() where possible< / li >
< li > ac/nir: make use of ac_build_umin() where possible< / li >
< li > ac: add ac_build_umax() and use it where possible< / li >
< li > ac: add ac_build_ddxy_interp() helper< / li >
< li > ac: add ac_build_load_helper_invocation() helper< / li >
< li > ac/nir: remove useles LLVMGetUndef for nir_op_pack_64_2x32_split< / li >
< li > ac/nir: remove useless integer cast in adjust_sample_index_using_fmask()< / li >
< li > ac/nir: remove useless integer cast in visit_image_load()< / li >
< li > ac/nir: remove some useless integer casts for ALU operations< / li >
< li > spirv: add SpvCapabilityFloat16 support< / li >
< li > radv: enable VK_KHR_shader_float16_int8< / li >
< li > radv: set ACCESS_NON_READABLE on stores for copy/fill/clear meta shaders< / li >
< li > radv: enable shaderInt8 on SI and CIK< / li >
< li > radv: sort the shader capabilities alphabetically< / li >
< li > ac/nir: use new LLVM 8 intrinsics for SSBO atomics except cmpswap< / li >
< li > ac/nir: add 64-bit SSBO atomic operations support< / li >
< li > radv: add VK_KHR_shader_atomic_int64 but disable it for now< / li >
< li > ac: add support for more types with struct/raw LLVM intrinsics< / li >
< li > ac: use struct/raw load intrinsics for 8-bit/16-bit int with LLVM 9+< / li >
< li > ac: use struct/raw store intrinsics for 8-bit/16-bit int with LLVM 9+< / li >
< li > ac/nir: only use the new raw/struct image atomic intrinsics with LLVM 9+< / li >
< li > ac/nir: only use the new raw/struct SSBO atomic intrinsics with LLVM 9+< / li >
< li > ac/nir: use the new raw/struct SSBO atomic intrisics for comp_swap< / li >
< li > radv: add VK_NV_compute_shader_derivates support< / li >
< li > radv: add missing VEGA20 chip in radv_get_device_name()< / li >
< li > radv: do not need to force emit the TCS regs on Vega20< / li >
< li > radv: fix color conversions for normalized uint/sint formats< / li >
< li > radv: implement a workaround for VK_EXT_conditional_rendering< / li >
< li > ac: tidy up ac_build_llvm8_tbuffer_{load,store}< / li >
< li > radv: set WD_SWITCH_ON_EOP=1 when drawing primitives from a stream output buffer< / li >
< li > radv: only need to force emit the TCS regs on Vega10 and Raven1< / li >
< li > radv: fix radv_get_aspect_format() for D+S formats< / li >
< li > radv: apply the indexing workaround for atomic buffer operations on GFX9< / li >
< li > radv: fix setting the number of rectangles when it's dyanmic< / li >
< li > radv: add a workaround for Monster Hunter World and LLVM 7& 8< / li >
< li > radv: allocate more space in the CS when emitting events< / li >
< li > radv: do not use gfx fast depth clears for layered depth/stencil images< / li >
< li > radv: fix alpha-to-coverage when there is unused color attachments< / li >
< li > radv: fix setting CB_SHADER_MASK for dual source blending< / li >
< / ul >
< p > Sergii Romantsov (4):< / p >
< ul >
< li > dri: meson: do not prefix user provided dri-drivers-path< / li >
< li > d3d: meson: do not prefix user provided d3d-drivers-path< / li >
< li > i965,iris/blorp: do not blit 0-sizes< / li >
< li > glsl: Fix input/output structure matching across shader stages< / li >
< / ul >
< p > Sonny Jiang (1):< / p >
< ul >
< li > radeonsi: use compute for clear_render_target when possible< / li >
< / ul >
< p > Tapani Pälli (42):< / p >
< ul >
< li > nir: add option to use scaling factor when sampling planes YUV lowering< / li >
< li > dri: add P010, P012, P016 for 10bit/12bit/16bit YUV420 formats< / li >
< li > intel/compiler: add scale_factors to sampler_prog_key_data< / li >
< li > i965: add P0x formats and propagate required scaling factors< / li >
< li > drirc/i965: add option to disable 565 configs and visuals< / li >
< li > mesa: return NULL if we exceed MaxColorAttachments in get_fb_attachment< / li >
< li > anv: anv: refactor error handling in anv_shader_bin_write_to_blob()< / li >
< li > iris: add Android build< / li >
< li > nir: initialize value in copy_prop_vars_block< / li >
< li > nir: use nir_variable_create instead of open-coding the logic< / li >
< li > android: add liblog to libmesa_intel_common build< / li >
< li > android: make libbacktrace optional on USE_LIBBACKTRACE< / li >
< li > iris: add libmesa_iris_gen8 library to the build< / li >
< li > util: fix a warning when building against clang7 headers< / li >
< li > anv: retain the is_array state in create_plane_tex_instr_implicit< / li >
< li > anv: toggle on support for VK_EXT_ycbcr_image_arrays< / li >
< li > anv: use anv_gem_munmap in block pool cleanup< / li >
< li > anv: call blob_finish when done with it< / li >
< li > nir: free dead_ctx in case of no progress< / li >
< li > anv: destroy descriptor sets when pool gets destroyed< / li >
< li > anv: release memory allocated by bo_heap when descriptor pool is destroyed< / li >
< li > anv: release memory allocated by glsl types during spirv_to_nir< / li >
< li > anv: revert "anv: release memory allocated by glsl types during spirv_to_nir"< / li >
< li > i965: remove scaling factors from P010, P012< / li >
< li > isl: fix automake build when sse41 is not supported< / li >
< li > android: Build fixes for OMR1< / li >
< li > iris: initialize num_cbufs< / li >
< li > iris: mark switch case fallthrough< / li >
< li > anv/radv: release memory allocated by glsl types during spirv_to_nir< / li >
< li > st/mesa: fix compilation warning on storage_flags_to_buffer_flags< / li >
< li > st/mesa: fix warnings about implicit conversion on enumeration type< / li >
< li > spirv: fix a compiler warning< / li >
< li > st/nir: run st_nir_opts after 64bit ops lowering< / li >
< li > iris: move variable to the scope where it is being used< / li >
< li > iris: move iris_flush_resource so we can call it from get_handle< / li >
< li > iris: handle aux properly in iris_resource_get_handle< / li >
< li > egl: setup fds array correctly when exporting dmabuf< / li >
< li > compiler/glsl: handle case where we have multiple users for types< / li >
< li > android/iris: fix driinfo header filename< / li >
< li > nir: use braces around subobject in initializer< / li >
< li > glsl: use empty brace initializer< / li >
< li > anv: expose VK_EXT_queue_family_foreign on Android< / li >
< / ul >
< p > Thomas Hellstrom (5):< / p >
< ul >
< li > winsys/svga: Add an environment variable to force host-backed operation< / li >
< li > winsys/svga: Enable the transfer_from_buffer GPU command for vgpu10< / li >
< li > svga: Avoid bouncing buffer data in malloced buffers< / li >
< li > winsys/svga: Update the drm interface file< / li >
< li > winsys/svga: Don't abort on EBUSY errors from execbuffer< / li >
< / ul >
< p > Timo Aaltonen (1):< / p >
< ul >
< li > util/os_misc: Add check for PIPE_OS_HURD< / li >
< / ul >
< p > Timothy Arceri (72):< / p >
< ul >
< li > st/glsl_to_nir: remove dead local variables< / li >
< li > ac/radv/radeonsi: add ac_get_num_physical_sgprs() helper< / li >
< li > radv: take LDS into account for compute shader occupancy stats< / li >
< li > util: move BITFIELD macros to util/macros.h< / li >
< li > st/glsl_to_nir: call nir_remove_dead_variables() after lowing local indirects< / li >
< li > nir: add support for marking used patches when packing varyings< / li >
< li > nir: add glsl_type_is_32bit() helper< / li >
< li > nir: add is_packing_supported_for_type() helper< / li >
< li > nir: rewrite varying component packing< / li >
< li > nir: prehash instruction in nir_instr_set_add_or_rewrite()< / li >
< li > nir: turn ssa check into an assert< / li >
< li > nir: turn an ssa check in nir_search into an assert< / li >
< li > nir: remove simple dead if detection from nir_opt_dead_cf()< / li >
< li > radeonsi/nir: set input_usage_mask properly< / li >
< li > radeonsi/nir: set colors_read properly< / li >
< li > radeonsi/nir: set shader_buffers_declared properly< / li >
< li > st/nir: use NIR for asm programs< / li >
< li > nir: remove non-ssa support from nir_copy_prop()< / li >
< li > nir: clone instruction set rather than removing individual entries< / li >
< li > nir: allow nir_lower_phis_to_scalar() on more src types< / li >
< li > radeonsi: fix query buffer allocation< / li >
< li > glsl: fix shader cache for packed param list< / li >
< li > radeonsi/nir: move si_lower_nir() call into compiler thread< / li >
< li > glsl: rename is_record() -> is_struct()< / li >
< li > glsl: rename get_record_instance() -> get_struct_instance()< / li >
< li > glsl: rename record_location_offset() -> struct_location_offset()< / li >
< li > glsl: rename record_types -> struct_types< / li >
< li > nir: rename glsl_type_is_struct() -> glsl_type_is_struct_or_ifc()< / li >
< li > glsl/freedreno/panfrost: pass gl_context to the standalone compiler< / li >
< li > glsl: use NIR function inlining for drivers that use glsl_to_nir()< / li >
< li > i965: stop calling nir_lower_returns()< / li >
< li > radeonsi/nir: stop calling nir_lower_returns()< / li >
< li > st/glsl: start spilling out common st glsl conversion code< / li >
< li > anv: add support for dumping shader info via VK_EXT_debug_report< / li >
< li > nir: add guess trip count support to loop analysis< / li >
< li > nir: add new partially_unrolled bool to nir_loop< / li >
< li > nir: add partial loop unrolling support< / li >
< li > nir: calculate trip count for more loops< / li >
< li > nir: unroll some loops with a variable limit< / li >
< li > nir: simplify the loop analysis trip count code a little< / li >
< li > nir: add helper to return inversion op of a comparison< / li >
< li > nir: add get_induction_and_limit_vars() helper to loop analysis< / li >
< li > nir: pass nir_op to calculate_iterations()< / li >
< li > nir: find induction/limit vars in iand instructions< / li >
< li > st/glsl_to_nir: fix incorrect arrary access< / li >
< li > radeonsi/nir: call some more var optimisation passes< / li >
< li > ac/nir_to_llvm: add assert to emit_bcsel()< / li >
< li > nir: only override previous alu during loop analysis if supported< / li >
< li > nir: fix opt_if_loop_last_continue()< / li >
< li > nir: add support for user defined loop control< / li >
< li > spirv: make use of the loop control support in nir< / li >
< li > nir: add support for user defined select control< / li >
< li > spirv: make use of the select control support in nir< / li >
< li > Revert "ac/nir: use new LLVM 8 intrinsics for SSBO atomic operations"< / li >
< li > nir: propagate known constant values into the if-then branch< / li >
< li > Revert "nir: propagate known constant values into the if-then branch"< / li >
< li > nir/radv: remove restrictions on opt_if_loop_last_continue()< / li >
< li > nir: initialise some variables in opt_if_loop_last_continue()< / li >
< li > nir/i965/freedreno/vc4: add a bindless bool to type size functions< / li >
< li > ac/nir_to_llvm: make get_sampler_desc() more generic and pass it the image intrinsic< / li >
< li > ac/nir_to_llvm: add image bindless support< / li >
< li > nir: fix packing components with arrays< / li >
< li > radeonsi/nir: fix scanning of bindless images< / li >
< li > st/mesa/radeonsi: fix race between destruction of types and shader compilation< / li >
< li > nir: fix nir_remove_unused_varyings()< / li >
< li > radeonsi/nir: create si_nir_opts() helper< / li >
< li > radeonsi/nir: call radeonsi nir opts before the scan pass< / li >
< li > util/drirc: add workarounds for bugs in Doom 3: BFG< / li >
< li > radeonsi: add config entry for Counter-Strike Global Offensive< / li >
< li > Revert "glx: Fix synthetic error generation in __glXSendError"< / li >
< li > Revert "st/mesa: expose 0 shader binary formats for compat profiles for Qt"< / li >
< li > st/glsl: make sure to propagate initialisers to driver storage< / li >
< / ul >
< p > Timur Kristóf (19):< / p >
< ul >
< li > radeonsi/nir: Use uniform location when calculating const_file_max.< / li >
< li > iris: implement clearing render target and depth stencil< / li >
< li > nir: Add ability for shaders to use window space coordinates.< / li >
< li > tgsi_to_nir: Fix the TGSI ARR translation by converting the result to int.< / li >
< li > tgsi_to_nir: Fix TGSI LIT translation by using flt.< / li >
< li > tgsi_to_nir: Make the TGSI IF translation code more readable.< / li >
< li > tgsi_to_nir: Split to smaller functions.< / li >
< li > nir: Move nir_lower_uniforms_to_ubo to compiler/nir.< / li >
< li > nir: Add multiplier argument to nir_lower_uniforms_to_ubo.< / li >
< li > freedreno: Plumb pipe_screen through to irX_tgsi_to_nir.< / li >
< li > tgsi_to_nir: Produce optimized NIR for a given pipe_screen.< / li >
< li > tgsi_to_nir: Restructure system value loads.< / li >
< li > tgsi_to_nir: Extract ttn_emulate_tgsi_front_face into its own function.< / li >
< li > tgsi_to_nir: Support FACE and POSITION properly.< / li >
< li > tgsi_to_nir: Improve interpolation modes.< / li >
< li > tgsi_to_nir: Set correct location for uniforms.< / li >
< li > radeonsi/nir: Only set window_space_position for vertex shaders.< / li >
< li > iris: Face should be a system value.< / li >
< li > gallium: fix autotools build of pipe_msm.la< / li >
< / ul >
< p > Tobias Klausmann (1):< / p >
< ul >
< li > vulkan/util: meson build - add wayland client include< / li >
< / ul >
< p > Tomasz Figa (1):< / p >
< ul >
< li > llvmpipe: Always return some fence in flush (v2)< / li >
< / ul >
< p > Tomeu Vizoso (19):< / p >
< ul >
< li > panfrost: Add gem_handle to panfrost_memory and panfrost_bo< / li >
< li > panfrost: Add backend targeting the DRM driver< / li >
< li > panfrost/midgard: Add support for MIDGARD_MESA_DEBUG< / li >
< li > panfrost: Add support for PAN_MESA_DEBUG< / li >
< li > panfrost: Set bo-> size[0] in the DRM backend< / li >
< li > panfrost: Set bo-> gem_handle when creating a linear BO< / li >
< li > panfrost: Adapt to uapi changes< / li >
< li > panfrost: Fix sscanf format options< / li >
< li > panfrost: Set the GEM handle for AFBC buffers< / li >
< li > panfrost: Also tell the kernel about the checksum_slab< / li >
< li > panfrost: Pass the context BOs to the kernel so they aren't unmapped while in use< / li >
< li > panfrost: Wait for last job to finish in force_flush_fragment< / li >
< li > panfrost: split asserts in pandecode< / li >
< li > panfrost: Guard against reading past end of buffer< / li >
< li > panfrost/ci: Initial commit< / li >
< li > panfrost/midgard: Skip register allocation if there's no work to do< / li >
< li > panfrost/midgard: Skip liveness analysis for instructions without dest< / li >
< li > panfrost: Fix two uninitialized accesses in compiler< / li >
< li > panfrost: Only take the fast paths on buffers aligned to block size< / li >
< / ul >
< p > Toni Lönnberg (8):< / p >
< ul >
< li > intel/genxml: Only handle instructions meant for render engine when generating headers< / li >
< li > intel/genxml: Media instructions and structures for gen6< / li >
< li > intel/genxml: Media instructions and structures for gen7< / li >
< li > intel/genxml: Media instructions and structures for gen7.5< / li >
< li > intel/genxml: Media instructions and structures for gen8< / li >
< li > intel/genxml: Media instructions and structures for gen9< / li >
< li > intel/genxml: Media instructions and structures for gen10< / li >
< li > intel/genxml: Media instructions and structures for gen11< / li >
< / ul >
< p > Topi Pohjolainen (2):< / p >
< ul >
< li > intel/compiler/icl: Use tcs barrier id bits 24:30 instead of 24:27< / li >
< li > intel/compiler/fs/icl: Use dummy masked urb write for tess eval< / li >
< / ul >
< p > Vasily Khoruzhick (2):< / p >
< ul >
< li > lima: use individual tile heap for each GP job.< / li >
< li > lima: add support for depth/stencil fbo attachments and textures< / li >
< / ul >
< p > Vinson Lee (5):< / p >
< ul >
< li > gallium/auxiliary/vl: Fix duplicate symbol build errors.< / li >
< li > nir: Fix anonymous union initialization with older GCC.< / li >
< li > swr: Fix build with llvm-9.0.< / li >
< li > gallium: Fix autotools build with libxatracker.la.< / li >
< li > freedreno: Fix GCC build error.< / li >
< / ul >
< p > Vivek Kasireddy (1):< / p >
< ul >
< li > drm-uapi: Update headers from drm-next< / li >
< / ul >
< p > Xavier Bouchoux (1):< / p >
< ul >
< li > nir/spirv: Fix assert when unsampled OpTypeImage has unknown 'Depth'< / li >
< / ul >
< p > Yevhenii Kolesnikov (1):< / p >
< ul >
< li > i965: Fix allow_higher_compat_version workaround limited by OpenGL 3.0< / li >
< / ul >
< p > coypu (1):< / p >
< ul >
< li > gbm: don't return void< / li >
< / ul >
< p > davidbepo (1):< / p >
< ul >
< li > drirc: add Waterfox to adaptive-sync blacklist< / li >
< / ul >
< p > grmat (1):< / p >
< ul >
< li > drirc: add Spectacle, Falkon to a-sync blacklist< / li >
< / ul >
< p > pal1000 (1):< / p >
< ul >
< li > scons: Compatibility with Scons development version string< / li >
< / ul >
< p > suresh guttula (3):< / p >
< ul >
< li > vl: Add cropping flags for H264< / li >
< li > radeon/vce:Add support for frame_cropping_flag of VAEncSequenceParameterBufferH264< / li >
< li > st/va/enc: Add support for frame_cropping_flag of VAEncSequenceParameterBufferH264< / li >
< / ul >
2019-01-29 23:32:16 +00:00
< / div >
< / body >
< / html >