| Mesa 21.2.0 Release Notes / 2021-08-04 |
| ====================================== |
| |
| Mesa 21.2.0 is a new development release. People who are concerned |
| with stability and reliability should stick with a previous release or |
| wait for Mesa 21.2.1. |
| |
| Mesa 21.2.0 implements the OpenGL 4.6 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 4.6. OpenGL |
| 4.6 is **only** available if requested at context creation. |
| Compatibility contexts may report a lower version depending on each driver. |
| |
| Mesa 21.2.0 implements the Vulkan 1.2 API, but the version reported by |
| the apiVersion property of the VkPhysicalDeviceProperties struct |
| depends on the particular driver being used. |
| |
| SHA256 checksum |
| --------------- |
| |
| :: |
| |
| 0cb3c802f4b8e7699b1602c08c29d06a4d532ab5b8f7a64676c4ca6bb8f4d426 mesa-21.2.0.tar.xz |
| |
| |
| New features |
| ------------ |
| |
| - zink supports GL_ARB_texture_filter_minmax, GL_ARB_shader_clock |
| |
| - VK_EXT_provoking_vertex on RADV. |
| |
| - VK_EXT_extended_dynamic_state2 on RADV. |
| |
| - VK_EXT_global_priority_query on RADV. |
| |
| - VK_EXT_physical_device_drm on RADV. |
| |
| - VK_KHR_shader_subgroup_uniform_control_flow on Intel and RADV. |
| |
| - VK_EXT_color_write_enable on RADV. |
| |
| - 32-bit x86 builds now default disable x87 math and use sse2. |
| |
| - GL ES 3.1 on GT21x hardware. |
| |
| - VK_EXT_acquire_drm_display on RADV and ANV. |
| |
| - VK_EXT_vertex_input_dynamic_state on lavapipe |
| |
| - wideLines on lavapipe |
| |
| - VK_EXT_line_rasterization on lavapipe |
| |
| - VK_EXT_multi_draw on ANV, lavapipe, and RADV |
| |
| - VK_KHR_separate_depth_stencil_layouts on lavapipe |
| |
| - VK_EXT_separate_stencil_usage on lavapipe |
| |
| - VK_EXT_extended_dynamic_state2 on lavapipe |
| |
| - NGG shader based primitive culling is now supported by RADV. |
| |
| - Panfrost supports OpenGL ES 3.1 |
| |
| - New Asahi driver for the Apple M1 |
| |
| - GL_ARB_sample_locations on zink |
| |
| - GL_ARB_sparse_buffer on zink |
| |
| - GL_ARB_shader_group_vote on zink |
| |
| - DRM format modifiers on zink |
| |
| - freedreno+turnip: Initial support for a6xx gen4 (a660, a635) |
| |
| - None |
| |
| |
| Bug fixes |
| --------- |
| |
| - The image is distorted while use iGPU(Intel GPU) rendering and output via dGPU (AMD GPU) |
| - lima: regression in plbu scissors cmd |
| - freedreno: regression in org.skia.skqp.SkQPRunner#gles_multipicturedraw_*_tiled |
| - Incorrect rendering |
| - intel/isl: Wrong surface format name in batch |
| - [RADV] FSR in Resident Evil: Village looks very pixelated on Polaris |
| - 21.2.0rc1 Build Failure - GCC6.3 |
| - Crash in update_buffers after closing KDE "splash screen" downloader |
| - Firefox (wayland) crash in wayland_platform |
| - Crash in update_buffers after closing KDE "splash screen" downloader |
| - Firefox (wayland) crash in wayland_platform |
| - radeonsi: persistent, read-only buffer maps are slow to read |
| - substance painter flickering with jagged texture and masks shown black |
| - radv: FP16 mode in FidelityFX FSR doesn't look right |
| - Regression in Turnip with KGSL and Zink running opengl in proot |
| - Validation crash on wlroots after wl_shm appeared |
| - [RADV] Blocky corruption in Scarlet Nexus and vkd3d-proton 2.4 |
| - Use out encoding for float immediates |
| - Radeon RX580 and 5700 XT: Reloading ARB assembly shaders causes very glitchy rendering |
| - i915g: dEQP-GLES2.functional.fragment_ops.depth_stencil..* failures |
| - i915g: dEQP-GLES2.functional.fragment_ops.interaction.basic_shader.* and fragment_ops.random.* failures |
| - KHR-GL46.shader_ballot_tests.ShaderBallot* tests fails on the main |
| - i965 nir compiler is lowering fdiv twice or initializing struct twice |
| - [radv] Textures missing in Doom (2016) w/ any temporal AA setting enabled |
| - Drop Android.mk |
| - [build error] macros.h:88:26: error: size of unnamed array is negative |
| - Game Issue: Nuclear Throne crashes in RadeonSI |
| - iris: reduce shader storage buffer object alignment |
| - agx_compile.h:29:10: fatal error: asahi/lib/agx_pack.h: No such file or directory |
| - radv: VBO range check issues with odd strides and sizes |
| - Crash in glLinkProgram while trying to craft the link error |
| - i915g: wide point failures |
| - Wolfenstein II: The New Colossus - Screen goes black in some cases |
| - [radv] [regression] Textures missing in Doom (2016) |
| - Shader compilation memory leaks |
| - radv: fd leak in Android WSI radv_AcquireImageANDROID |
| - SpaceEngine in Steam Proton cannot start with Mesa >=20.3 |
| - [bisected] KDE plasma menu text renders like stretched strangely [amdgpu] |
| - radeonsi: glitches in Euro Truck Simulator 2 |
| - White box for Webrender Firefox with R600_DEBUG=nir on Evergreen GPU |
| - radv_AllocateDescriptorSets: validation on variable description count is too strict |
| - Luna Sky Crashes on Launch |
| - Mesa crashes on undefined texture behaviour |
| - Mesa crashes on undefined texture behaviour |
| - cache_test uses uninitialized stack memory |
| - nir/opt_load_store_vectorize: check_for_robustness() crashes on derefs |
| - [anv] GravityMark (benchmark) crashes on ANV |
| - turnip: corrupted geometry after tesselation shaders in GTAV |
| - [opengl] We happy few not being rendered correctly |
| - anv: dynamic state prim type is hard |
| - [iris][bisected] piglit test ...ext_external_objects.vk-image-display-muliple-textures failing after enabled |
| - Factorio: GPU hang when opening machine inventory |
| - RuneScape on Mesa 21.1.1 (VEGA10) has bad performance and leaks memory |
| - LLVM12 breaks atomicCompSwap tests with radeonsi |
| - freedreno: tex-miplevel-selection causes a creation of too many BOs without flushes, causing a crash |
| - RADV: Resident Evil Village Freezes during a specific cutscene |
| - Supraland: flickering black bars on ground |
| - u_queue.c:173:7: error: implicit declaration of function 'timespec_get' is invalid in C99 |
| - radv: front face and cull mode dynamic state can desync |
| - radv: GPU hang in Cyberpunk 2077 on Big Navi |
| - Cyberpunk 1.22 crashes with amdgpu ring gfx_0.0.0 timeout |
| - [amdgpu][renoir][rx5500m]: [drm:0xffffffff8198ad5e] \*ERROR* ring gfx timeout, signaled seq=10952, emitted seq=10954 |
| - [spirv-fuzz] SPIR-V parsing FAILED: Invalid back or cross-edge in the CFG |
| - anv: dEQP-VK.graphicsfuzz.spv-stable-bubblesort-flag-complex-conditionals fails |
| - panfrost Mount and Blade: Warband (glitches) |
| - anv: dEQP-VK.robustness.robustness2*no_fmt_qual.null_descriptor.samples* fail |
| - KHR-GLES31.core.shader_image_load_store.basic-glsl-earlyFragTests may be flakey on RPI4 |
| - SPIRV AMD Driver compiler memory leak/usage - 8Gb+ to compile single 41Kb SPIRV file, Ubuntu 21.04 |
| - Regression of !10941: mutter Wayland on bare metal crashes when closing HW accelerated windows |
| - radv: ACO miscompiles a specific DIRT 5 shader |
| - Copy paste bug in v3dv_cmd_buffer.c |
| - Segfault in mtx_unlock/amdgpu_bo_slab_destroy |
| - [radeonsi] Incorrect rendering when using glDisable(GL_MULTISAMPLE) with multisample backbuffer |
| - virtio(vulkan): Error building on Android/arm32 |
| - [i965] regression: piglit.spec.arb_depth_buffer_float.fbo-depthstencil-gl_depth32f_stencil8-drawpixels |
| - iris: Select memory map cacheability settings at BO allocation time |
| - zink: regression for primitive-restart on ANV |
| - zink: Unimplemented ALU {un,}pack_half_2x16 in nir_to_spirv |
| - venus: dEQP-VK.api.command_buffers.bad_inheritance_info_random test failure |
| - [radeonsi] glClearTexImage on 1D array only clears first slice |
| - zink: Expected Image Operand ConstOffset to be a const object |
| - docs: bullet-lists no longer show any bullets |
| - [RADV] - Path of Exile (238960) - Ground decals are missing or corrupted using the Vulkan renderer. |
| - [ADL-S / TGL-U / TGL-H] Pixels missing / flickering when render some app on weston |
| - [iris][biected] piglit.spec.glsl-1_50.execution.geometry.clip-distance-vs-gs-out |
| - panfrost ff9a0000.gpu: Unhandled Page fault in AS1 at VA 0x0000000009801200 |
| - black stripes in X/Xwayland applications under panfrost/midgard |
| - radv: significant overhead from radv_pipeline_has_ngg() |
| - mesa-21.1.1/src/gallium/drivers/vc4/vc4_resource.c:790: reading from uninitialised array |
| - [iris,i965][hsw,ivb,snb,bdw][bisected][regression] wflinfo crashes |
| - [iris,i965][hsw,ivb,snb,bdw][bisected][regression] wflinfo crashes |
| - The image is distorted while use iGPU(Intel GPU) rendering and output via dGPU (AMD GPU) |
| - [radv] Revert !7207 once BG3 is fixed. |
| - piglit test 'spec.glsl-1_30.execution.range_analysis_fsat_of_nan' failure |
| - iris: Rework iris_transfer_mapto better use iris_has_color_unresolved |
| - ir3_cf breaks nir_op_fquantize2f16 |
| - [i915g] implicit fallthrough |
| - Add asahi to CI's build |
| - RADV - Regression - CPU side 'hang' in metro exodus |
| - Assassin's Creed Odyssey crash on launch |
| - Metro Exodus not starting under xwayland |
| - [i915g] PIPE_SHADER_CAP_MAX_HW_ATOMIC_COUNTERS is Unknown cap 38 |
| - Elite Dangerous: Odyssey alpha crashes GPU on launch |
| - glmark2-es2 -b terrain crashes since Bifrost FP16 |
| - [iris][bisected][regression] dEQP-GLES31.functional.texture.multisample.* tests crash on multiple platforms |
| - gallium: tc regression |
| - panfrost: Look into invalidate_resource() |
| - [ivb,hsw][i965][bisected] dEQP-GLES3.functional.shaders.texture_functions.texture.sampler2darrayshadow_vertex failing |
| - ci: Explicitly test shader caching |
| - radv: RoTR fails on Raven APU |
| - Tungsten Graphics links in Gallium docs |
| - Obs Studio Broken on Latest Mesa Git(Regression)(Bisected) |
| - Graphics corruption and GPU hang with RADV/LLVM |
| - old kernels (4.19) support in radv |
| - Elite Dangerous: Odyssey alpha crashes GPU on launch |
| - CSGO: Some default variables can cause problems with trust mode |
| - freedreno: dEQP-GLES3.functional.fence_sync.client_wait_sync_finish flakes |
| - glxclient.h:56:10: fatal error: 'loader.h' file not found |
| - mesa git started to break wine + UnrealTournament.exe (old dx6 game) |
| - SuperTuxKart artifacting on RK3399 |
| - [amdgpu]: Golf With Your Friends (431240): ERROR Waiting for fences timed out |
| - don't flush for each blit/grid |
| - No sRGB capable visuals/fbconfigs reported in glx |
| - Documentation request: AMD debug variables |
| - docs: versions is out-of-date |
| - Strange results when trying to read from VK_FORMAT_R64_SFLOAT in compute shader |
| - anv: dEQP-VK.binding_model.buffer_device_address.set3.depth3.basessbo.convertcheck* slow |
| - Iris doesn't support INTEL_performance_query anymore |
| - [iris][bisected][regression] dEQP-GLES3.functional.texture.specification.teximage2d* failures |
| - RADV: TRUNC_COORD breaks gather operations |
| - [RADV] corruption in avatar after dying in Heroes of the Storm |
| - drm-shim build error with glibc 2.33 |
| - Metro Exodus crashing due to memory overflow |
| - Sauerbraten shader rendering broken on RV530 (r300g) |
| - texture glitches on CS:GO on Tiger Lake |
| - Incorrect texture blitting/mapping when running Unigine-Heaven 4.0 on ADL-S / TGL-H, TGL-U |
| - Build fail due to "parameter name omitted" on Gallium Nine |
| - v3dv: DiligentEngine fail to run with SRGB swapchain |
| - Non-DRI builds broken by recent cleanups in Mesa core |
| - Cinnamon core dump after installing latest oibaf mesa build (165a69d2) |
| - yuv sampler lowering regression |
| - clover llvm build failure ‘class llvm::VectorType’ has no member named ‘getNumElements’; |
| - Possible corruption for apps using multiple Z-buffers on TGL |
| |
| |
| Changes |
| ------- |
| |
| Aaron Liu (1): |
| |
| - amd: add Yellow Carp support |
| |
| Abel García Dorta (2): |
| |
| - i915g: add HW atomic counters as unsupported |
| - i915g: fix implicit fallthrough |
| |
| Adam Jackson (53): |
| |
| - Revert "glx: Lift sending the MakeCurrent request to top-level code" |
| - gallium/xlib: Fix for recent gl_config changes |
| - glx/drisw: Enable GLX_ARB_create_context_no_error |
| - glx: Remove silly __glXGetGLVersion() indirection |
| - glx: Remove some truly ancient unused code |
| - glx: Remove major/minor version tracking from extension table |
| - glx: Mark GLX_{ATI_pixel_format_float,NV_float_buffer} as supported |
| - glx: Remove some non-functional GL extension from the table |
| - glx: Generalize __glXGetStringFromTable a little |
| - glx: Remove redundant client_support field from extension table |
| - glx: Enable pure-client-library extensions explicitly |
| - glx: Fold client_gl_only array into its one real user |
| - glx: Remove some ancient backwards-compatibility typedefs |
| - zink: Learn about VK_KHR_swapchain |
| - zink: Fix format query for minmax reduction support |
| - gallium: Fix PIPE_BIND_SAMPLER_REDUCTION_MINMAX definition to be unique |
| - dri: Try harder to infer the drawable fbconfig if needed |
| - glx: Return the right GLX opcode in synthetic MakeCurrent errors |
| - glx: s/dri_message/glx_message/ |
| - glx: Add and use DebugMessageF convenience macro |
| - glx: Convert undocumented LIBGL_DIAGNOSTIC to LIBGL_DEBUG=verbose |
| - glx: Remove unused debugging printfs |
| - glx: Implement GLX_EXT_no_config_context |
| - glx: Stop pretending the GLX major number isn't 1 |
| - glx: Stop force-enabling extensions "implied" by GLX 1.3 |
| - glx: Simplify glXIsDirect |
| - glx: Remove warn-once-ery around GLX 1.3 functions |
| - glx: Remove unused opcode argument to __glX{Get,QueryServer}String |
| - glx: Stash a copy of the XExtCodes in the glx_display |
| - glx: Simplify some overuse of GetGLXScreenConfigs |
| - glx: Move server GLX vendor and version strings to glx_screen |
| - glx: s/Display \*/struct glx_display \*/ over internal API |
| - glx: Remove some dead declarations from glxclient.h |
| - Revert "glx: s/Display \*/struct glx_display \*/ over internal API" |
| - include: Remove unused i810_pci_ids.h |
| - i915c: Add a symlink for i830_dri.so |
| - mesa: Remove unused _mesa_{create,destroy}_visual |
| - mesa: Ignore the depth buffer when computing framebuffer floatness |
| - gallium: Reset attachments to ST_ATTACHMENT_INVALID when revalidating |
| - format/fxt1: Clean up fxt1_variance's argument list |
| - mesa: s/malloc/calloc/ to silence a warning |
| - gallium: Remove unused st_visual::render_buffer |
| - gallivm: Fix a signature mismatch warning |
| - zink/ntv: Don't call free() on ralloc'd memory |
| - gallium/dri: Remove unused dri_drawable::drisw_surface |
| - drisw: Don't bzero displaytarget pixels |
| - intel: properly constify isl_format_layouts |
| - classic/xlib: Fix the build after !9817 |
| - swrast: Fix a warning from gcc 11 |
| - loader/dri3: Properly initialize the XFIXES extension |
| - loader/dri3: Don't churn through xfixes regions in SwapBuffers |
| - vl/dri3: Don't leak regions on the X server |
| - meson: Make prefer-{crocus,iris} always take effect |
| |
| Alejandro Piñeiro (30): |
| |
| - v3dv/debug: print correct stage name |
| - v3dv/debug: use gl stage when checking debug flag |
| - v3dv/pipeline: track descriptor maps per stage, not per pipeline |
| - v3dv: remove custom icd json generation |
| - v3dv: move extensions table to v3dv_device |
| - v3dv: don't use typedef enum with broadcom stages |
| - v3dv: remove unused v3dv_zs_buffer_from_vk_format |
| - broadcom/compiler: use proper type field for atomic operations |
| - v3d/simulator: capture hub interrupts |
| - v3d/simulator: add a cache flush mode enum |
| - v3d/simulator: wait for cache flushes |
| - v3d/simulator: use the proper register when waiting on a CSD submit |
| - v3d/simulator: use BFC/RFC registers to wait for bin/render to complete |
| - broadcom/common: move v3d_tiling to common |
| - v3d/simulator: hw mem is now an v3d_size_t, typedef to uint32_t |
| - v3d/simulator: get rid of has_gca wrapper |
| - v3dv: rename v3dv_pack for v3dvx_pack |
| - v3dv/cmd_buffer: add helper job_emit_binning_prolog |
| - v3dv/cmd_buffer: move cl_emit calls for Draw methods to helpers |
| - v3dv: start to move and wrap hw-version code with v3dv_queue |
| - v3dv: split v3dv_pipeline hw version dependant code to a new source file |
| - v3dv: split v3dv_image hw version dependant code to a new source file |
| - v3dv: split v3dv_format hw version dependant code to a new source file |
| - v3dv: split v3dv_device hw version dependant code to a new source file |
| - v3dv: move several hw version dependant code to their own (v3dvx) source file |
| - v3dv: split v3dv_descriptor hw version dependant to a new source file. |
| - v3dv: don't use cl_packet_length for prepacked data |
| - v3dv: remove gen-dependant includes from v3dv_private |
| - v3dv/build: meson infrastructure for multi-hw-version support |
| - v3dv/format: expose properly that some formats are not filterable |
| |
| Alexander Monakov (2): |
| |
| - freedreno/drm-shim: pretend to offer DRM 1.6.0 |
| - freedreno/drm-shim: keep GEM buffers page-aligned |
| |
| Alexander Shi (1): |
| |
| - mesa: texparam: Add a clamping macro to handle out-of-range floats returned as integers. |
| |
| Alexey Nurmukhametov (1): |
| |
| - tu/kgsl: Fix file descriptor double close |
| |
| Alyssa Rosenzweig (668): |
| |
| - nir: Update some comments referring to imov |
| - panfrost: Don't allow_forward_pixel_to_kill for Z/S blit |
| - panfrost: Set allow_forward_pixel_to_be_killed for blit |
| - panfrost: Set clean_fragment_write for blits |
| - panfrost: Invert blend_reads_dest logic |
| - panfrost: Don't allow FPK if a RT is missing |
| - panfrost: Allow FPK when there are no side effects |
| - panfrost: Keep Bifrost blendable -> pixel in table |
| - panfrost: Specialize blendable formats for sRGB |
| - panfrost: Simplify format_to_bifrost_blend prototype |
| - panfrost: Drop blendable format accessor |
| - panfrost: Always pick dithered tb formats |
| - panfrost: Remove padded unorm blendable formats |
| - docs/macos: Explain Apple GLX versus OSMesa on macOS |
| - nir/lower_fragcolor: Use shader_instructions_pass |
| - nir/lower_fragcolor: Handle fp16 outputs |
| - panfrost: Fix formats converting uninit from AFBC |
| - nir/lower_fragcolor: Fix driver_location assignment |
| - nir/lower_fragcolor: Take max cbufs as argument |
| - d3d12: Switch to nir_lower_fragcolor |
| - util/bitset: Add BITSET_COUNT helper |
| - nir: Add fsin_agx opcode |
| - asahi: Stub command-line compiler for AGX G13B |
| - agx: Add opcode descriptions as Python |
| - agx: Generate opcode list |
| - agx: Generate runtime-accessible opcode table |
| - agx: Generate builder routines |
| - agx: Stub NIR backend compiler |
| - agx: Remap varyings to match AGX ABI |
| - agx: Stub control flow walking |
| - agx: Stub NIR instruction iteration |
| - agx: Stub emit_intrinsic |
| - agx: Implement load_const as mov |
| - agx: Implement direct st_vary |
| - agx: Add agx_alu_src_index helper for emit_alu |
| - agx: Implement vec2/vec3/vec4 ops |
| - agx: Implement fragment_out |
| - agx: Add instruction printing |
| - agx: Add a trivial register allocator |
| - agx: Add instruction packing |
| - agx: Add packing for memory loads/stores |
| - agx: Add st_vary(_final) instruction packing |
| - agx: Terminate programs with stop and traps |
| - agx: Implement ld_vary |
| - agx: Implement simple floating point ops |
| - agx: Implement fsin/fcos |
| - agx: Add 8-bit AGX minifloat routines |
| - agx: Add minifloat tests |
| - agx: Implement native float->int conversions |
| - agx: Implement native int->float conversions |
| - agx: Add bitwise operations |
| - agx: Add iadd/imad integer arithmetic |
| - agx: Add saturated integer add/subtract support |
| - agx: Add 32-bit bitwise shifts |
| - agx: Add forward optimizing pass for fmov |
| - agx: Add dead code eliminator |
| - agx: Propagate fmov backwards as well |
| - agx: Propagate immediates |
| - agx: Implement limited case of i2i16/i2i32 as iadd |
| - agx: Add sysval management helper |
| - agx: Implement load_ubo/kernel_input |
| - agx: Set flag on last st_vary instruction |
| - agx: Lower load_attr to device memory accesses |
| - agx: Implement vertex_id |
| - agx: Add agx_tex_dim helper |
| - agx: Emit texture ops |
| - agx: Pack texture ops |
| - agx: Add min/max support |
| - agx: Support 1-bit booleans |
| - agx: Implement b2f |
| - agx: Add b2i implementation |
| - agx: Pack cmpsel |
| - agx: Support bcsel |
| - asahi: Add hexdump utility |
| - asahi: Add command buffer XML definitions |
| - asahi: Add allocation data structure |
| - asahi: Add a GenXML fork |
| - asahi: Add (clean room) IOKit uABI header |
| - asahi: Add command buffer decode helpers |
| - asahi: Add tiling routines |
| - asahi: Add device abstraction |
| - asahi: Add pool data structure |
| - asahi: Add uniform upload routines |
| - asahi: Add some magic IOGPU routines |
| - asahi: Add vertex formats table |
| - asahi: Add Gallium driver |
| - nir/opcodes: Reword confusing comment |
| - pan/bi: Add missing sr_count to pseudo-atomics |
| - pan/bi: Don't reference uninit source in ATOM_C1 |
| - pan/bi: Add simple constant folding pass |
| - pan/bi: Don't reference nir_lower_mediump_outputs |
| - pan/bi: Simplify Python expression |
| - pan/bi: Union modifiers from across variants |
| - pan/bi: Support 16-bit load_interpolated_input |
| - pan/bi: Emit int CSEL instead of float by default |
| - pan/bi: Implement vectorized f32_to_f16 |
| - pan/bi: Fix 16-bit fsat |
| - pan/bi: Improve assert for vector size errors |
| - pan/bi: Implement vectorized int downcasts |
| - pan/bi: Fix loads and stores smaller than 32 bits |
| - pan/bi: Lower swizzles on CLPER |
| - pan/bi: Add and use bi_negzero helper |
| - pan/bi: Don't schedule clamps to +FADD.v2f16 |
| - pan/bi: Workaround \*V2F32_TO_V2F16 erratum |
| - panfrost: Don't unroll loops in GLSL |
| - panfrost: Remove old dEQP workaround |
| - pan/bi: Track dual-src blend type |
| - pan/bi: Handle different sizes of LD_TILE |
| - pan/bi: Add single-component 8-bit mkvec lowering |
| - pan/bi: Handle swizzles in i2i8 |
| - pan/bi: Lower 8-bit fragment input |
| - panfrost: Make comment less confusing |
| - panfrost: Support alpha_to_one |
| - panfrost: Minor cleanup of blend CSO |
| - panfrost: Don't clobber RT0 if RTn is disabled |
| - pan/lower_blend: Clean up type size handling |
| - pan/lower_blend: Use NIR helpers |
| - pan/lower_blend: Rename is_bifrost->scalar |
| - panfrost/blend: Fix outdated comments |
| - panfrost/blend: Workaround a v7 implementation-detail |
| - panfrost/blend: Distribute to_c_factor |
| - panfrost/blend: Prepare for lower_fragcolor |
| - panfrost: Call nir_lower_fragcolor based on key |
| - panfrost: Assume lower_fragcolor has been called |
| - panfrost/lower_framebufffer: Don't use i2imp |
| - pan/blend: Emit explicit conversions for all types |
| - panfrost: Key blend shaders to the input types |
| - pan/mdg: Hide units behind MIDGARD_MESA_DEBUG=verbose |
| - pan/mdg: More concise RMU name |
| - pan/mdg: Don't print zero shifts |
| - pan/mdg: Suppress most attribute tables |
| - pan/mdg: Don't print explicit .rte |
| - pan/mdg: Don't print mem addr brackets |
| - pan/mdg: Reduced printed parens |
| - pan/mdg: Don't print zero |
| - pan/bi: Add imm_uintN helper |
| - pan/bi: Handle integer min/max ourselves |
| - pan/bi: Handle ineg |
| - pan/bi: Handle b2f ourselves |
| - pan/bi: Handle b2i8/16 |
| - pan/bi: Track scalarness of 16-bit ALU |
| - pan/bi: Don't swizzle scalars |
| - pan/bi: Switch to 1-bit bools |
| - pan/bi: Use nir_lower_to_bit_size |
| - pan/mdg: Use _output_ type for outmod printing |
| - pan/mdg: Remove midgard_opt_copy_prop_reg |
| - pan/mdg: Enable nir_opt_{move, sink} |
| - panfrost/blend: Inline blend constants |
| - pan/mdg: Model blend shader interference |
| - panfrost: Fix typo handling blend types |
| - pan/bi: Change swizzled scalars to identity |
| - pan/bi: Adapt branching for 1-bit bools |
| - pan/bi: Handle make_vec with 1-bit bools |
| - pan/bi: Temporarily switch back to 0/~0 bools |
| - pan/bi: Enable NIR vectorization |
| - pan/bi: Fix int<-->float size converts |
| - pan/bi: Copyprop constants |
| - pan/bi: Garbage collect bifrost_nir.h |
| - pan/bi: Enable mediump BLEND lowering |
| - panfrost: Enable 16-bit support on Bifrost |
| - pan/lower_framebuffer: Fix bitsize mismatch |
| - nir: Add blend lowering pass |
| - panfrost: Use common blend lowering |
| - nir/divergence_anlysis: Add intrinsics for Bifrost |
| - pan/bi: Drop load_sampler_lod_parameters_pan |
| - pan/bi: Map load_subgroup_invocation to FAU |
| - pan/bi: Add "lanes per warp" accessor |
| - pan/bi: Add divergent intrinsic lowering pass |
| - asahi: Translate blend CSO to lower_blend options |
| - asahi: Augment Gallium key with blend state |
| - asahi: Call nir_lower_blend with selected key |
| - asahi: Garbage collect bind_state |
| - asahi: Implement set_blend_color |
| - asahi: Add blend constant system value |
| - asahi: Call nir_lower_fragcolor |
| - asahi: Fix shader key hash function |
| - asahi: Pass through "reads tilebuffer?" bit |
| - agx: Return agx_instr* from emit_intrinsic |
| - agx: Implement blend constant color sysvals |
| - agx: Rename blend -> st_tile |
| - agx: Add ld_tile opcode |
| - agx: Assume lower_fragcolor has been called |
| - agx: Condition writeout ops on already being emitted |
| - agx: Implement load_output |
| - agx: Set reads_tib appropriately |
| - panfrost: Drop panfrost_fence in favour of pipe_fence_handle |
| - docs: Simplify now that kmsro is autoenabled |
| - pan/bi: Add first_vertex to vertex ID |
| - panfrost: Track buffers needing resolve |
| - panfrost: Set discard based on the resolve set |
| - panfrost: Implement framebuffer invalidation |
| - panfrost: Hide CAP_INT16 behind is_deqp |
| - panfrost: Don't translate compare funcs |
| - panfrost: Remove spurious assignment |
| - panfrost: Clean up cases for emit_fbd |
| - panfrost: Don't upload empty push uniform table |
| - pan/mdg: Use smaller LD_UNIFORM instructions |
| - ci: Build asahi in meson-gallium job |
| - panfrost: Fix major flaw in BO cache |
| - panfrost: Drop random #define |
| - panfrost: Use natural shader limits |
| - panfrost: Make clear which limits are arbitrary |
| - panfrost: Garbage collect comment |
| - panfrost: Shorten iffy comment |
| - pan/mdg: Remove unused midgard_int_alu_op_prefix |
| - pan/mdg: Fix output types for scalar fields |
| - pan/mdg: Fix spills to TLS |
| - pan/mdg: Set lower_uniforms_to_ubo |
| - panfrost: Add unowned mode to pan_pool |
| - panfrost: Label all BOs in userspace |
| - panfrost: Label pools |
| - panfrost: Make pool slab size configurable |
| - panfrost: Add reference type for unowned pool |
| - panfrost: Pool shaders |
| - panfrost: Pool texture views |
| - panfrost: Reduce blitter pool size |
| - panfrost: Fix blending for unbacked MRT |
| - panfrost: Fix the reads_dest prototype |
| - panfrost: Fix is_opaque prototype |
| - panfrost: Fix blend constant fetch prototype |
| - panfrost: Fix blend fixed-function prototype |
| - panfrost: Fix pan_blend_to_fixed_function_equation prototype |
| - panfrost: Move blend properties to CSO create |
| - panfrost: Translate fixed-function blend at CSO create |
| - panfrost: Garbage collect Gallium blend includes |
| - panfrost: Pack blend equations at CSO create time |
| - panfrost: Distribute out constant colour code |
| - panfrost: Simplify blend_final |
| - panfrost: Pass batch to panfrost_get_blend |
| - panfrost: Streamline fixed-function get_blend path |
| - panfrost: Remove unused dither flag |
| - panfrost: Split Bifrost BLEND emit by word |
| - panfrost: Precompute bifrost_blend_type_from_nir |
| - panfrost: Add draw-time merge helper |
| - panfrost: Prepack partial RSD at compile time |
| - panfrost: Move depth/stencil/alpha to CSO create |
| - panfrost: Preset evaluate_per_sample |
| - panfrost: Correct the type of sample_mask |
| - panfrost: Fill out the rasterizer CSO |
| - panfrost: Move early-z decision earlier |
| - panfrost: Streamline the !fs_required case |
| - panfrost: Hoist allow_forward_pixel_to_be_killed |
| - panfrost: Partially determine FPK state |
| - panfrost: Distribute masks for FPK selection |
| - panfrost: Pull erratum workaround into own function |
| - panfrost: Hoist part of shader_reads_tilebuffer |
| - panfrost: Pack draw-time RSD all-at-once |
| - panfrost: Move batch_set_requirements to the CSO |
| - panfrost: Deduplicate some code from indirect/direct draws |
| - panfrost: Pass batch to panfrost_get_index_buffer_bounded |
| - panfrost: Remove silly assertion |
| - panfrost: Mark job_index > 10000 as unlikely |
| - panfrost: Simplify panfrost_bind_sampler_states |
| - panfrost: Express viewport in terms of the batch |
| - asahi: Set PACKED_STREAM_OUTPUT |
| - glsl: Fix subscripted arrays with no XFB packing |
| - glsl: Fix packing of matrices for XFB |
| - panfrost: Streamline varying linking code |
| - panfrost: Define dirty tracking flags |
| - panfrost: Add the usual clean/dirty helpers |
| - panfrost: Dirty all state when batch is set |
| - panfrost: Dirty track RSDs |
| - panfrost: Dirty track textures/samplers |
| - panfrost: Dirty track viewport descriptor |
| - panfrost: Dirty track fragment images |
| - panfrost: Add PAN_MESA_DEBUG=dirty option |
| - panfrost/ci: Disable GLES2 jobs when we run GLES3 |
| - panfrost/ci: Disable G72 jobs for now |
| - panfrost/ci: Split rules by ISA |
| - ci: Condition ppc64-el on specific drivers |
| - ci: Condition s390x on specific drivers |
| - panfrost: Only link varyings once in good conditions |
| - panfrost: Lower max inputs again |
| - panfrost: Abort on faults in SYNC mode |
| - panfrost: Remove minimal mode |
| - panfrost: Increase tiler_heap max allocation to 64MB |
| - panfrost/ci: Disable terrain trace |
| - panfrost/ci: Remove reference to dated flag |
| - panfrost/ci: Run jobs with PAN_MESA_DEBUG=sync |
| - panfrost: Add Message Preload descriptor XML |
| - panfrost: Add message preload to pan_shader_info |
| - panfrost: Inline pan_prepare_shader_descriptor |
| - panfrost: Don't take ctx in panfrost_shader_compile |
| - panfrost: Expose PIPE_CAP_SHAREABLE_SHADERS |
| - asahi: Fix meson.build definition to depend on agx_pack.h |
| - agx: Drop cmdline version back to ES3.0 |
| - agx: Pack ld_var Dx |
| - agx: Enable 1-bit load_const |
| - agx: Implement boolean mov |
| - agx: Track current_block |
| - agx: Track block offsets |
| - agx: Add nest field to IR |
| - agx: Add invert_cond (ccn) to IR |
| - agx: Add branch target to IR |
| - agx: Add inner loop nesting count field |
| - agx: Model control flow instructions |
| - agx: Model pop_exec |
| - agx: Add push_exec alias |
| - agx: Pack control flow instructions |
| - agx: Model jump instructions |
| - agx: Fix up branch offsets at pack time |
| - agx: Implement emit_if the simplest way |
| - agx: Optimize out empty else blocks |
| - agx: Implement loops in the simplest way |
| - agx: Add break/continue support |
| - agx: Zero r0l before first use of control flow |
| - asahi: Fix scissor descriptor definition |
| - asahi: Add "set scissor" command |
| - asahi: Add scissor enable bit |
| - asahi: Defer viewport pack |
| - asahi: Dirty track viewport descriptor |
| - asahi: Track scissor states |
| - asahi: Mark scissor dirty if rast->scissor changes |
| - asahi: Skip draws if the scissor culls everything |
| - agx: Add scissor upload BO |
| - asahi: Expose PIPE_CAP_CLIP_HALFZ |
| - asahi: Add unknown bits seen with the GL driver |
| - asahi: Enable depth culling |
| - asahi: Update viewport descriptor depth fields |
| - asahi: Implement scissors and scissor to viewport |
| - asahi: Fix off-by-one in viewport scissoring |
| - asahi: Implement wide lines |
| - asahi: Determine tiling vs linear for internal textures |
| - asahi: Use dt_stride for line_stride where needed |
| - asahi: Add layout enum to XML |
| - asahi: Translate layouts for texture and RTs |
| - asahi: Identify line stride in texture/RT XML |
| - asahi: Respect linear strides |
| - asahi: Handle linear display targets as well as tiled |
| - asahi: Note that "render target" lacks an sRGB bit |
| - asahi: Align strides to 16 bytes |
| - asahi: Print unknown enum values |
| - asahi: Add format enums |
| - asahi: Hide pixel formats behind an opaque type |
| - asahi: Scaffold format table |
| - asahi: Use pixel table in is_format_supported |
| - asahi: Respect render target format swizzle |
| - asahi: Add ETC2 formats to table |
| - asahi: Add "hacks for dEQP" flag |
| - asahi: Lift streamout scaffolding from Panfrost |
| - asahi: Fake CAPs for ES3 with AGX_MESA_DEBUG=deqp |
| - asahi: Flesh out the formats table |
| - asahi: Allow half-float vertex buffers |
| - asahi: Make data_valid a bitset to save memory |
| - asahi: Abort on blit() |
| - asahi: Add mipmapping state to the XML |
| - asahi: Set levels in texture descriptor |
| - asahi: Allocate slices for mipmapping |
| - panfrost: Update comment |
| - panfrost: Shrink pan_draw_mode return type |
| - panfrost: Add draw parameters dirty flags |
| - panfrost: Analyze sysval dirty flags |
| - panfrost: Dirty track constant buffers |
| - panfrost: Don't allocate empty varying buffer |
| - panfrost: Dirty track stack sizes |
| - panfrost: Write translate_index_size better |
| - panfrost: Minor changes to draw_vbo |
| - panfrost: Bubble up errors |
| - panfrost: Elucidate thread group split field |
| - panfrost: Eliminate reserve_* functions |
| - panfrost/ci: Report flakes on IRC |
| - vc4: Use Rn_UINT instead of In_UINT for index buffers |
| - v3d: Use Rn_UINT instead of In_UINT for index buffers |
| - etnaviv: Use Rn_UINT instead of In_UINT for index buffers |
| - freedreno: Use Rn_UINT instead of In_UINT for index buffers |
| - lima: Use Rn_UINT instead of In_UINT for index buffers |
| - si: Use Rn_UINT instead of In_UINT for index buffers |
| - docs/gallium: Document the index buffer format convention |
| - nir: Add nir_intrinsic_load_back_face_agx |
| - asahi: Mark special fragment inputs as sysvals |
| - agx: Model get_sr |
| - agx: Generate enums from Python |
| - agx: List sr enum in Python |
| - agx: Pack SR immediate |
| - agx: Lower front face to back face |
| - agx: Handle load_back_face_agx |
| - ci: Disable the iris APL jobs |
| - nir/lower_fragcolor: Avoid redundant load_output |
| - pan/bi: Pull out bi_count_write_registers |
| - pan/bi: Use TEXS_2D for rect textures |
| - pan/bi: Simplify TEXC codegen for sr_count=0 |
| - pan/bi: Fix bi_rewrite_passthrough ordering |
| - pan/bi: Bundle after RA |
| - pan/bi: Add post-RA optimizer |
| - pan/bi: Track liveness while scheduling |
| - pan/bi: Allow IADD.u32 on FMA as \*IADDC |
| - pan/bi: Use explicit affinities in RA |
| - pan/bi: Inline spilling in RA |
| - pan/bi: Explicit zero reg_live_{in, out} when needed |
| - pan/bi: Model interference with preloaded regs |
| - pan/bi: Allow move/sink in blend shaders |
| - pan/bi: Don't restrict the register file in non-blend shaders |
| - pan/bi: Model +BLEND clobbering of r48 |
| - pan/bi: Handle images in vertex shaders |
| - pan/bi: Lower loads with component > 0 |
| - pan/bi: Lower stores with component != 0 |
| - pan/bi: Lower 64-bit ints again |
| - pan/bi: Emit a dummy ATEST if needed |
| - pan/bi: Simplify spill code |
| - pan/bi: Track words instead of bytes in RA |
| - pan/bi: Don't allocate past the end of the reg file |
| - panfrost: Remove AFBC format fixups |
| - panfrost: Add missing 'Reverse issue order flag' |
| - panfrost: Disable AFBC on v7 |
| - panfrost: Don't duplicate attribute buffers |
| - panfrost: Separate image attribute and buffer emit |
| - panfrost: Be explicit in image modifier handling |
| - panfrost: Use util_last_bit for images |
| - panfrost: Default indirect attributes to 1D type |
| - pan/indirect: Factor out is_power_of_two_or_zero |
| - pan/indirect_draw: Use unsigned comparisons |
| - pan/indirect_draw: Fix 1 instance, nonzero divisor |
| - panfrost: Correctly size varyings |
| - panfrost: Use varying format from frag shader |
| - pan/bi: Force u32 for flat varyings |
| - panfrost: Fix vertex image attribute overrun |
| - panfrost: Simplify compute_checksum_size formula |
| - panfrost: Fix crc_valid condition |
| - panfrost: Zero r_dimension for buffer textures |
| - panfrost: Add util_draw_indirect() debug path |
| - panfrost: Align NPOT divisor records |
| - panfrost: Fix src_offset data type |
| - panfrost: Make instancing code more obvious |
| - panfrost: Assert alignment of indirect records |
| - pan/mdg: Use consistent casing in midgard_print |
| - pan/mdg: Make -Wswitch happy |
| - pan/mdg: Stub memory_barrier{_image} |
| - panfrost: Clarify how fs_sidefx works with oq |
| - panfrost: Simplify Midgard blend disable |
| - panfrost: Don't force early-z with occlusion query |
| - panfrost: Respect early-Z force on Midgard |
| - pan/mdg: Fix units for SUBSAT |
| - pan/mdg: Handle {i,u}{add,sub}_sat |
| - pan/mdg: Update r1.w comment |
| - pan/mdg: Fix incorrect rewrite in Midgard scheduler |
| - panfrost: Mark 16/32_UNORM as non-renderable (v5) |
| - panfrost: Don't allocate WLS when not needed |
| - pan/mdg: Wire in PAN_SYSVAL_VERTEX_INSTANCE_OFFSETS |
| - pan/mdg: Lower away gl_VertexID offset |
| - pan/mdg: Use more accurate ld/st reg estimates |
| - pan/mdg: Don't skip unit-based checks in choose_instruction |
| - pan/mdg: Assert scheduled instructions are reasonable |
| - pan/mdg: Insert moves to load/store registers |
| - panfrost: Fix dirty state emission |
| - panfrost: Emulate indirect draws on Midgard |
| - panfrost: Add some missing BGRA formats |
| - panfrost: Remove scissor_culls_everything |
| - panfrost: Don't set a blend shader for no_colour |
| - panfrost: Allocate XFB buffers per-instance |
| - panfrost: Fix BUFFER image handling |
| - panfrost: Make image buffers robust |
| - panfrost: Lower max compute size |
| - panfrost: Set PIPE_COMPUTE_CAP_SUBGROUP_SIZE |
| - panfrost: Set PIPE_COMPUTE_CAP_MAX_THREADS_PER_BLOCK |
| - panfrost: Drop todo on PIPE_COMPUTE_CAP_IMAGES_SUPPORTED |
| - panfrost: Don't CRC mipmapped textures |
| - panfrost: Reduce pan_image_state indirection |
| - pan/indirect_dispatch: Indent NIR blocks |
| - pan/indirect_dispatch: Simplify empty command case |
| - pan/indirect_dispatch: Distinguish minus-1 defs |
| - pan/indirect_dispatch: Expand split expressions |
| - pan/indirect_dispatch: Use extracted values |
| - panfrost: Use direct dispatch with shared memory |
| - panfrost: Don't clobber indirect dispatch fields |
| - panfrost: Make data_valid a bitset |
| - panfrost: Remove pan_image_state |
| - panfrost: Set valid_buffer_range for GPU writes |
| - panfrost: Add XML for vertex/instance ID records |
| - panfrost: Clean up vertex/instance ID on Midgard |
| - panfrost: Flush everything for glMemoryBarrier |
| - panfrost: Flush before compute jobs |
| - panfrost: Set vertex job_barrier |
| - panfrost: Add "Cache Flush" job XML |
| - panfrost: Advertise GLES3.1 |
| - pan/decode: Fix image attribute counting |
| - pan/decode: Handle cache flush jobs |
| - panfrost/ci: Blank G52 flakes file |
| - panfrost/ci: Don't skip SSBO tests on G52 |
| - panfrost/ci: Do fractional dEQP-GLES31 run on Midgard |
| - docs/features: Mark GLES3.1 as done on Panfrost |
| - docs/panfrost: Update API versions |
| - pan/bi: Include modifier info in opcode table |
| - pan/bi: Move bi_word_node to common code |
| - pan/bi: Move typesize to common code |
| - pan/bi: Track instruction size in opcode table |
| - pan/bi: Handle fsat_signed and fclamp_pos |
| - pan/bi: Report tuples, not nops, in shader-db |
| - pan/bi: Propagate fabs/neg/sat |
| - pan/bi: Add back custom algebraic opts |
| - pan/bi: Fuse fclamp_pos and fsat_signed |
| - pan/bi: Schedule FCMP.v2f16 with abs modifier |
| - pan/bi: Fuse abs into FCMP/FMIN/FMAX.v2f16 |
| - nir: Fix constant folding for irhadd/urhadd |
| - agx: Mark components as ASSERTED |
| - agx: Add agx_immediate_f helper |
| - agx: Add perspective bit to ld_var |
| - agx: Update ld_vary encoding mask |
| - agx: Add ld_vary_flat opcode |
| - asahi: Identify varying descriptor fields |
| - agx: Rename remap_varyings -> remap_varyings_vs |
| - agx: Implement nir_intrinsic_load_frag_coord |
| - agx: Implement ld_vary_flat |
| - agx: Rename agx_pack to agx_pack_binary |
| - agx: Remap fragment shader varyings explicitly |
| - asahi: Unify varying linking code with vertex shaders |
| - agx: Pull out agx_write_components |
| - agx: Add agx_exit_block helper |
| - agx: Add liveness analysis pass |
| - agx: Mark sources that kill |
| - agx: Count write registers, not components |
| - agx: Lift agx_block_add_successor from Panfrost |
| - agx: Track logical control flow graph |
| - asahi: Wire in tgsi_to_nir |
| - asahi: Fix random \*2 |
| - asahi: Guard for overflow when packing |
| - asahi: Always flush when setting framebuffer state |
| - asahi: Handle Z16_UNORM textures |
| - asahi: Add zsbuf to batch |
| - asahi: Save zsbuf ptr |
| - asahi: Add internal (renderable) formats to the table |
| - asahi: Set fragment key for non-U8NORM render targets |
| - asahi: Implement colour buffer reloads |
| - asahi: Remove spurious assignment |
| - asahi: Remove spurious varying assignment |
| - asahi: Generalize varying linking |
| - asahi: Add ASAHI_MESA_DEBUG=no16 option |
| - agx: Fix 32-bit bitwise shifts |
| - agx: Fix LOD_MIN enum |
| - agx: Pack LOD descriptors |
| - agx: Fix lod_mode shift |
| - agx: Legalize LOD sources to be 16-bit only |
| - agx: Handle txl |
| - asahi: Fail on LOD clamps/bias |
| - asahi: Identify texture/sampler count fields |
| - asahi: Identify vertex texture/sampler counts |
| - asahi: Set vertex texture/sampler counts |
| - asahi: Track more Gallium state |
| - asahi: Wire in u_blitter |
| - asahi: Handle nonzero first_level |
| - asahi: Fix meson dependency on packing in compiler |
| - asahi: Prepack rasterizer faces |
| - asahi: Implement the stencil test |
| - asahi: Flush for accesses to Z/S buffer |
| - asahi: Comment on an embedded data structure |
| - asahi: Skip over holes in the vbufs |
| - asahi: Add XML for the attachment structure |
| - asahi: Sync attachment magic with asahi demo |
| - asahi: Parametrize software "command buffer" size |
| - asahi: Identify "command buffer" size field in map |
| - asahi: Move IOGPU header to XML |
| - asahi: Extend IOGPU header to contain encoder |
| - asahi: Use GenXML for main bind fragment |
| - asahi: Identify attachment length field |
| - asahi: Set data_valid for the depth buffer |
| - asahi: Enable primitive restart |
| - asahi: Use XML for interpolation packet |
| - panfrost: Express dependencies as resources, not BOs |
| - panfrost: Wrap occlusion query in pipe_resource |
| - panfrost: Split "flush writer" from "flush accessing" |
| - panfrost: Eliminate redundant flushes with AFBC |
| - panfrost: Add secondary shader XML fields |
| - pan/decode: Handle IDVS jobs on Bifrost |
| - agx: Don't choke on registers in the optimizer |
| - agx: Count read registers as well |
| - agx: Assign registers locally |
| - agx: Pipe in nir_register |
| - agx: Ensure we don't overallocate registers |
| - panfrost: Move draw_vbo to pan_cmdstream.c |
| - panfrost: Move most CSO creates to pan_cmdstream.c |
| - panfrost: Split out prepare_rsd into a vtbl |
| - panfrost: Move blend CSO to cmdstream/context |
| - panfrost: Don't ralloc panfrost_blend_state |
| - panfrost: Move launch_grid to pan_cmdstream |
| - panfrost: Move panfrost_emit_tile_map to pan_job |
| - panfrost: Use vtable for fragment descriptor functions |
| - panfrost: Clean up pan_cmdstream.h |
| - panfrost: Move sample accessor to pan_cmdstream |
| - panfrost: Remove pan_cmdstream.h |
| - panfrost: Remove unused midgard-pack.h includes |
| - docs: Update relnotes for panfrost/asahi |
| - pan/bi: Improve clause printing |
| - pan/bi: Fix skip/lod_mode aliasing with VAR_TEX |
| - pan/bi: Add bi_foreach_instr_global_rev_safe helper |
| - pan/bi: Pack staging_barrier for the -next- clause |
| - pan/bi: Try to hit full occupancy on v7 |
| - pan/bi: Only spill nodes that could progress in RA |
| - pan/bi: Report cycle counts |
| - pan/bi: Track LOD mode even for TEXC |
| - pan/bi: Analyze helper invocations |
| - pan/bi: Fuse LD_VAR+TEXS_2D -> VAR_TEX |
| - pan/bi: Add a constant subexpression elimination pass |
| - pan/bi: Workaround widen restrictions on +FADD.f32 |
| - pan/bi: Simplify cube map descriptor generation |
| - pan/bi: Comment the fexp2 implementation |
| - pan/bi: Factor out exp2/log2 code |
| - pan/bi: Don't lower fpow |
| - panfrost: Fix FPK enable condition |
| - panfrost: Add a performance counter dump utility |
| - panfrost: Don't set zs_update_operation in vertex shaders |
| - panfrost: Zero depth_source in vertex shaders |
| - panfrost: Query tiler features |
| - panfrost: Enable more tiler levels if we can |
| - panfrost: Generalize pan_blitter's reg count assert |
| - panfrost: Set register allocation in the v7 RSD |
| - asahi: Move fixed internal shaders to agx_blit.c |
| - asahi: Add missing copyright/guards for magic.c/h |
| - asahi: Remove unused bo_access property |
| - asahi/decode: Only dump mapped allocations |
| - asahi: Make track_free safer |
| - asahi/decode: Check fewer zeroes after a command buffer |
| - asahi: Reserve more space to stop a command buffer |
| - asahi: Identify more unknown fields in the memmap |
| - asahi/decode: Fix up high word |
| - asahi/decode: Handle CULL packets |
| - asahi/decode: Fix decoding of draw calls |
| - asahi: Allow specifying an encoder ID |
| - asahi: Allocate global IDs |
| - asahi: Consolidate some magic numbers |
| - asahi: Garbage collect senseless cmdbuf struct |
| - asahi/decode: Print clear/store pipelines |
| - asahi/decode: Print some IOGPU stuff |
| - asahi: Set bits in UNK11 needed for points |
| - asahi: Set point magic bit in rasterizer |
| - asahi: Set bit for psiz |
| - asahi: Lower PIPE_CAPF_MAX_POINT_WIDTH to hw limit |
| - asahi: Unpack varying descriptors (1x) |
| - asahi: Identify triangle/lines vs point varyings |
| - asahi: Handle point coordinates |
| - agx: Flip point coordinates because OpenGL |
| - panfrost: Inline flip_compare_func into pan_encoder.h |
| - panfrost: Move panfrost_vertex/instance_id to per-gen |
| - panfrost: Inline away pan_pool.c |
| - panfrost: Express pack_work_groups more concisely |
| - panfrost: Inline away pan_invocation.c |
| - panfrost: Assert that injected jobs are for blits |
| - panfrost: Inline panfrost_get_z_internal_format |
| - panfrost: Move arch-independent pan_format code |
| - panvk: Don't use panfrost_bifrost_swizzle |
| - panfrost: Remove panfrost_bifrost_swizzle |
| - panfrost: Add GenXML macros |
| - panfrost: Compile format table multiple times |
| - panfrost: Specialize blendable_formats for v6 |
| - panfrost: Use smaller sizes in blend table |
| - panfrost: Give WLS Instances a default |
| - panfrost: Pin an architecture for blending |
| - panfrost: Use generic delete for ZSA |
| - panfrost: Remove reference to mali_blend_equation_packed |
| - panfrost: Avoid GenXML enum dependences |
| - panfrost: Remove pan_blitter integration |
| - panfrost: Init/destroy blitter from per-gen file |
| - panfrost: Only access blitter from per-gen |
| - pan/bi: Refuse to CSE non-SSA sources |
| - pan/bi: Make bi_foreach_instr_in_tuple safer |
| - pan/bi: Update ins->link after scheduling |
| - pan/bi: Do helper termination analysis on clauses |
| - pan/bi: Handle multiple destinations in scheduler |
| - pan/bi: Add bi_before_tuple convenience method |
| - pan/bi: Handle 4-src instructions in scheduler |
| - pan/bi: Calculate dependency graph when bundling |
| - pan/bi: Add a bundling heuristic |
| - panfrost: Fix format swizzles on G72 |
| - targets/graw-xlib: Add missing dep_x11 |
| - pan/mdg: Garbage collect silly quirk |
| - asahi: Fix sampler filtering flag |
| - agx: Fix mismatched units in load_ubo |
| - agx: Plug memory leak in register allocator |
| - pan/bi: Restrict swizzles on same cycle temporaries |
| - pan/bi: Remove incorrect errata workaround |
| |
| Andres Gomez (25): |
| |
| - ci: Uprev piglit to 9d87cc3d79e ("framework/replay: send backend's subprocess stderr to sys.stderr") |
| - ci: Add test which occasionally times out to lavapipe-vk skips |
| - ci: add xorg to the x86_test-vk container |
| - ci: allow starting xorg for piglit run |
| - ci: remove results directory content only with piglit runners |
| - ci: make sure we only read the first line from install/VERSION |
| - ci: update some radv trace checksums |
| - ci: update some radv trace checksums |
| - ci: update radv's trace job tag for Raven |
| - ci: remove radv's trace job for Polaris10 |
| - ci: uprev apitrace to 10.0 |
| - ci: uprev DXVK to 1.8.1 |
| - ci: add radv's trace job for Navy Flounder |
| - ci: include VKD3D-Proton tests into the VK test container |
| - ci: add VKD3D-Proton testsuite runner |
| - ci: add VKD3D-Proton testsuite job for radv's Navy Flounder |
| - ci: disentangle tags for containers and artifacts produced by them |
| - ci: remove glslangValidator installation from the VK test container |
| - ci: replace glslangValidator with glslang-tools |
| - ci: fix the vkd3d-proton runner |
| - ci: build the hang-detection tool into x86_test-vk |
| - ci: update some radv trace checksums |
| - ci: bump x86_test-base tag |
| - ci: remove unzip from several containers that don't use it at all |
| - ci: use bash with download-git-cache.sh |
| |
| Andrii Simiklit (1): |
| |
| - Remove redundant assignment |
| |
| Antonio Caggiano (15): |
| |
| - panfrost: Fix invalid conversions |
| - panfrost: Meson dependency |
| - util: Perfetto SDK v15.0 |
| - pps: Gfx-pps v0.3.0 |
| - pps: Gfx-pps config tool |
| - pps: Documentation |
| - intel/perf: Extern C |
| - pps: Intel pps driver |
| - pps: Intel documentation |
| - ci: Add a manual job for tracking the performance of Freedreno |
| - panfrost: Counter definitions |
| - panfrost: Performance configuration |
| - panfrost: Fix pan_pool_ref construction |
| - pps: Panfrost pps driver |
| - pps: Panfrost documentation |
| |
| Anuj Phogat (39): |
| |
| - intel: Rename files with gen_debug prefix |
| - intel: Rename gen_debug prefix to intel_debug |
| - intel: Rename GEN_DEBUG prefix to INTEL_DEBUG |
| - intel: Rename intel_device_info.c to intel_dev_info.c |
| - intel: Rename gen_device prefix in filenames |
| - intel: Rename gen_device prefix to intel_device |
| - intel: Fix alignment and line wrapping due to gen_device renaming |
| - intel: Rename GEN_DEVICE prefix in macros to INTEL_DEVICE |
| - intel: Rename gen_get_device prefix to intel_get_device |
| - intel: Rename gen_get_aperture_size to intel_get_aperture_size |
| - intel: Drop gen prefix in gen_has_get_tiling() |
| - intel: Rename gen_context.h to intel_context.h |
| - intel: Rename gen_context prefix to intel_context |
| - intel: Rename gen_perf prefix in filenames to intel_perf |
| - intel: Rename gen_perf prefix to intel_perf in source files |
| - intel: Fix alignment and line wrapping due to gen_perf renaming |
| - intel: Rename GEN_PERF prefix to INTEL_PERF in build files |
| - intel: Rename GEN_PERF prefix to INTEL_PERF in source files |
| - intel: Rename gen_{pipeline, oa, counter, hw} to intel_{..} |
| - intel: Rename brw_gen_enum.h to brw_gfx_ver_enum.h |
| - intel: Rename gen enum to gfx_ver |
| - intel: Rename gen keyword in test_eu_validate.cpp |
| - intel: Rename gens keyword to gfx_vers |
| - intel: Rename index_gen keyword to index_ver |
| - intel: Rename eu compact instruction tests |
| - intel: Rename gen_{mapped, clflush, invalidate} prefix to intel_{..} |
| - intel: Remove devinfo_to_gen() helper function |
| - intel: Rename isl_to_gen keyword to isl_encode |
| - intel: Rename vk_to_gen keyword to vk_to_intel |
| - intel: Rename gen_10 to ver_10 |
| - intel: Rename calculate_gen_slm_size to intel_calculate_slm_size |
| - intel: Rename _gen_{program, part, batch, freq} to _intel_{..} |
| - intel: Rename GEN_PART to INTEL_PART |
| - intel: Rename {i965, iris, anv, isl}_gen prefix in build files |
| - intel: Rename since_gen to since ver |
| - intel: Rename _gen keyword to _gfx_ver in few build files |
| - intel: Fix GEN_GEN macro checks |
| - intel/gfx12+: Add Wa_14013840143 |
| - intel: Rename GFX 12.5 to XE_HP |
| |
| Axel Davy (1): |
| |
| - st/nine: Fix compilation error on non-x86 platforms |
| |
| Bas Nieuwenhuizen (41): |
| |
| - radv: Fix memory leak on descriptor pool reset with layout_size=0. |
| - amd/common: Use cap to test kernel modifier support. |
| - radv: Only require DRM 3.23. |
| - radeon/vcn: Use the correct pitch for chroma surface. |
| - nir: Add load_sbt_amd intrinsic. |
| - radv: Add sbt descriptors user SGPR input. |
| - aco: Add load_sbt_amd intrinsic implementation. |
| - radv: Use global BO list with raytracing. |
| - radv: Add support for RT bind point. |
| - radv: Add RT pipeline bind. |
| - radv: Implement vkCmdTraceRays. |
| - radv: Use correct border swizzle on GFX9+. |
| - nir: Add bvh64_intersect_ray_amd intrinsic. |
| - aco: Implement bvh64_intersect_ray_amd intrinsic. |
| - nir/lower_returns: Deal with single-arg phis after if. |
| - radv: Don't skip barriers that only change queues. |
| - radv: Actually return correct value for read-only DCC compressedness. |
| - radv: Allow DCC images to be compressed with foreign queues. |
| - gallium/vl: Use format plane count for sampler view creation. |
| - gallium/va: Add support for PRIME_2 import. |
| - radv: Use the global BO list for acceleration structures. |
| - radv: Add initial CPU BVH building. |
| - radv: Implement device-side BVH building. |
| - radv: Add acceleration structure descriptor set support. |
| - radv: Convert lower_intrinsics to a switch statement |
| - radv: Implement load_vulkan_descriptor for acceleration structures. |
| - radv: Expose formats for acceleration structure. |
| - radv: Add rt perftest flag. |
| - radv: Enable VK_KHR_acceleration_structure with RADV_PERFTEST=rt. |
| - radv: Add -Wpointer-arith. |
| - util/fossilize_db: Pull seek into lock. |
| - util/fossilize_db: Split out reading the index. |
| - util/fossilize_db: Do not lock the fossilize db permanently. |
| - util/fossilize_db: Only lock the db file, not the index. |
| - nir: Add lowered vendor independent raytracing intrinsics. |
| - nir: Add raytracing shader call lowering pass. |
| - meson: Bump libdrm for amdgpu to 2.4.107. |
| - radv/winsys: Return vulkan errors for buffer creation. |
| - radv/winsys: Add support for a fixed VA address for replay. |
| - radv: Support address capture and replay. |
| - ac/surface: Handle non-retiled displayable DCC correctly for modifiers. |
| |
| Bastian Beranek (1): |
| |
| - glx: Assign unique serial number to GLXBadFBConfig error |
| |
| BillKristiansen (2): |
| |
| - d3d12: Fixes stale context bindings after copy, resolve, and clear |
| - d3d12: Sets all SRV descriptors as data-static |
| |
| Billy Laws (1): |
| |
| - meson: Increase Android Platform SDK version limit |
| |
| Boris Brezillon (60): |
| |
| - panfrost: Don't advertise AFBC mods when the format is not supported |
| - panfrost: Reserve thread storage descriptor in panfrost_launch_grid() |
| - panfrost: Fix RSD emission on Bifrost v6 |
| - panfrost: Fix indirect draws |
| - pan/bi: Don't set the EOS flag if there's at least one successor |
| - panfrost: Keep panfrost_batch_reserve_framebuffer() private |
| - panfrost: Fix ZS reloading on Bifrost v6 |
| - pan/midg: Fix 2 memory leaks |
| - pan/bi: Expand pseudo instructions when nosched is set |
| - pan/midg: Fix midgard_pack_common_store_mask() |
| - pan/midg: Make sure the constant offset is in range in mir_match_iadd() |
| - panfrost: Make sure pack_work_groups_compute() is passed valid dimensions |
| - panfrost: Add helpers to emit indirect dispatch jobs |
| - panfrost: Hook-up indirect dispatch support |
| - panfrost: Only advertise INDIRECT_DRAW if the kernel supports HEAP BOs |
| - ci: Update to a kernel that has the panfrost MMU fixes |
| - panfrost/ci: Test GLES 3.1 on Bifrost |
| - panfrost/ci: Skip draw_indirect.compute_interop.large.* |
| - panfrost/ci: Run the full deqp-gles3 testsuite |
| - panfrost: Fix format definitions to match gallium expectations |
| - Revert "gallium/util: Fix depth/stencil blit shaders" |
| - panfrost: Pass an image view to panfrost_estimate_texture_payload_size() |
| - panfrost: Fix blit shader names |
| - panfrost: Pack pan_blit_surface fields |
| - panfrost: Get rid of the vertex_count arg in pan_preload_emit_varying() |
| - panfrost: Make pan_preload_emit_*_sampler() applicable to blits |
| - panfrost: Stop assigning ->position in pan_preload_emit_varying() |
| - panfrost: Make pan_preload_emit_*_textures() applicable to blits |
| - panfrost: Make pan_preload_emit_viewport() applicable to blits |
| - panfrost: Rename pan_preload_emit_varying() |
| - panfrost: Shrink the number of args passed to prepare_{bifrost,midgard}_rsd() |
| - panfrost: Don't select the blit shader fragout type twice |
| - panfrost: Stop assuming the viewport will always cover the framebuffer |
| - panfrost: Extend pan_blitter to support blit/resolve operations |
| - panfrost: Use pan_blit() when PAN_MESA_DEBUG=panblit |
| - panfrost: Split the indexed and !indexed indirect draw info structs |
| - pan/bi: Add support for gl_{BaseVertex,BaseInstance} |
| - pan/bi: Add support for gl_DrawID |
| - panfrost: Expose the DRAW_PARAMETERS cap on Bifrost |
| - panfrost: Flag indirect draw/dispatch shaders as internal |
| - panfrost: Relax the stride check when importing resources |
| - panfrost: Try to align scanout resource stride on 64 bytes |
| - panfrost: Don't freeze blit batches |
| - panfrost: Avoid duplicate entries in access->readers |
| - panfrost: Simplify the dependency tracking logic |
| - panfrost: Limit the number of active batch to 32 |
| - ci: Update to a new kernel fixing a bug in the panfrost driver |
| - panfrost: Constify the constants pointers passed to pan_blend functions |
| - panfrost: Make panfrost_scoreboard_initialize_tiler() return the job pointer |
| - pan/midg: Add a flag to dump internal shaders |
| - panfrost: Add a Vulkan driver for Midgard/Bifrost GPUs |
| - panfrost: Don't add blit context BOs twice |
| - panfrost: Pass a memory pool to pan_blit_ctx_init() |
| - panfrost: Add alignment info to write-value and cache-flush jobs |
| - panfrost: Allocate WRITE_VALUE jobs with panfrost_pool_alloc_desc() |
| - panvk: Use the desc alloctor when we can |
| - panfrost: Start splitting the panfrost pool logic |
| - panvk: Support returning BOs allocated by panvk_pool to a 'free BO' pool |
| - panfrost: Replace the batch->bos hashmap by a sparse array |
| - panfrost: Do tracking of resources, not BOs |
| |
| Boyuan Zhang (8): |
| |
| - vl: add st_rps_bits for HEVC decode |
| - frontends/va: get st_rps_bits from VA pic param hevc |
| - frontends/vdpau: disable UseStRpsBits for vdpau hevc |
| - radeon/vcn: enable parsing support for st_rps_bits |
| - frontends/omx: use pipe buffer map instead of texture map |
| - radeon/vcn: move calc_dpb_size into create_decoder |
| - radeon/vcn: allocate non-tmz context buffer for VCN2+ |
| - radeon/vcn: use st_rps_bits only when it's set |
| |
| Caio Marcelo de Oliveira Filho (25): |
| |
| - spirv: Don't replicate patch bool in vtn_variable |
| - nir: Remove now unnecessary conditions from emit_load/store helpers |
| - intel/compiler: Add common function for CS dispatch info |
| - iris: Use brw_cs_get_dispatch_info() |
| - anv: Use brw_cs_get_dispatch_info() |
| - i965: Use brw_cs_get_dispatch_info() |
| - intel/compiler: Remove unused exported functions |
| - nir: Move shared_memory_explicit_layout bit into common shader_info |
| - intel/compiler: Clarify why VUE is recomputed by FS |
| - nir: Rename nir_is_per_vertex_io to nir_is_arrayed_io |
| - compiler: Rename local_size to workgroup_size |
| - compiler: Rename SYSTEM_VALUE_LOCAL_GROUP_SIZE to SYSTEM_VALUE_WORKGROUP_SIZE |
| - nir: Rename nir_intrinsic_load_local_group_size to nir_intrinsic_load_workgroup_size |
| - nir: Rename WORK_GROUP (and similar) to WORKGROUP |
| - nir: Move zero_initialize_shared_memory into common shader_info |
| - nir: Move workgroup_size and workgroup_variable_size into common shader_info |
| - anv: Support workgroup memory in other shaders |
| - nir/lower_io: Rename vertex_index to array_index in helpers |
| - nir/gather_info: Rename per_vertex to is_arrayed |
| - spirv: Fix handling of OpBranchConditional with same THEN and ELSE |
| - nir/opt_if: Don't split ALU for single block infinite loops |
| - nir: Add test to check edge case in Split ALU optimization |
| - spirv: Update headers and metadata from latest Khronos commit |
| - spirv: Support SPV_KHR_subgroup_uniform_control_flow |
| - anv: Advertise VK_KHR_shader_subgroup_uniform_control_flow |
| |
| Carsten Haitzler (Rasterman) (1): |
| |
| - panfrost: Fix Bo imports to not take the process down if fd is invalid |
| |
| Charlie (10): |
| |
| - v3dv: enable KHR_image_format_list |
| - v3dv: enable KHR_sampler_mirror_clamp_to_edge |
| - v3dv: enable KHR_incremental_present |
| - v3dv: enable KHR_uniform_buffer_standard_layout |
| - v3dv: clamp srgb render targets |
| - v3dv: remove sRGB blending workaround |
| - v3dv: add the unswizzled RGBA4444 format |
| - v3dv: divide by block size in copy_image_blit |
| - v3dv: add ASTC formats to get_compatible_tlb_format |
| - v3dv: enable ASTC formats |
| |
| Charlie Birks (1): |
| |
| - v3dv: document two supported extensions |
| |
| Charlie Turner (5): |
| |
| - ci: Remove obsolete reference to DEQP_SKIPS |
| - radv: Merge dEQP default skips into all generation-specific skip lists |
| - radv: Add a STONEY baseline for dEQP. |
| - radv: Provide a toggle to avoid warnings about unsupported devices. |
| - spirv_to_nir: Add environment variable to change default log level |
| |
| Charmaine Lee (1): |
| |
| - svga: fix texture rectangle sampling when no sampler view declaration is defined |
| |
| Chia-I Wu (93): |
| |
| - venus: stop using vn_renderer_sync in vn_fence |
| - venus: stop using vn_renderer_sync in vn_semaphore |
| - venus: stop using vn_renderer_sync in vn_queue |
| - venus: remove vn_renderer_sync support from vn_queue_submission |
| - venus: remove VN_SYNC_TYPE_SYNC |
| - venus: remove vn_queue::sync_queue_index |
| - venus: remove vn_ring_wait_all |
| - venus: wait on vkQueuePresentKHR |
| - venus: remove vn_renderer_info::has_timeline_sync |
| - venus: add vn_image_init_memory_requirements |
| - venus: add vn_image_create |
| - venus: add vn_wsi_create_scanout_image |
| - venus: refactor vn_queue_submission_count_semaphores |
| - venus: clang-format clean |
| - venus: change SpaceBeforeParens style option |
| - venus: provide accessors for vn_instance_submit_command |
| - venus: update venus-protocol headers to use accessors |
| - venus: rename VN_CS_ENCODER_INITIALIZER |
| - venus: add vn_renderer_shmem |
| - venus: use vn_renderer_shmem |
| - venus: add dev->renderer pointer |
| - venus: pass vn_renderer in vn_renderer_bo functions |
| - venus: move vn_renderer_bo_ops to vn_renderer |
| - venus: merge bo create and init ops |
| - venus: move some common members to vn_renderer_bo |
| - venus: use sparse array to manage vn_renderer_bo |
| - venus: make sure gem_handle and vn_renderer_bo are 1:1 |
| - venus: update venus-protocol for external memory |
| - venus: rework external memory capability queries |
| - venus: enable external memory support |
| - venus: fix render pass without attachments |
| - venus: fix dmabuf import mmap_size check |
| - venus: fix dmabuf import fail path |
| - venus: add VN_MAX_API_VERSION |
| - venus: rename vn_instance::renderer_version |
| - venus: clarify/fix instance renderer versions |
| - venus: clarify/fix device renderer version |
| - venus: refactor vn_physical_device_init_extensions |
| - venus: avoid strcmp for spec version override |
| - venus: refactor vn_physical_device_init_supported_extensions |
| - venus: init supported extensions in one place |
| - venus: add extension check for ANDROID_native_buffer |
| - venus: clean up vn_device_fix_create_info |
| - venus: get rid of #ifdef's in vn_CreateImage |
| - pps: fix a missing include in Intel pps driver |
| - util/u_thread: fix u_thread_setname for long names |
| - venus: add struct vn_command_buffer_builder |
| - venus: remember cmd buffer level and queue family |
| - venus: ignore pInheritanceInfo when we should |
| - docs: add basic documentation for venus |
| - vulkan/wsi: provide more info in wsi_image_create_info |
| - venus: add vn_device_memory_alloc as a helper |
| - venus: fix asserts on mem bo |
| - venus: fix opaque fd re-import |
| - venus: move wsi_image_create_info parsing |
| - venus: remember image wsi states |
| - venus: handle VN_COMMAND_BUFFER_STATE_INVALID |
| - venus: remember render pass PRESENT_SRC attachments |
| - venus: remember render pass PRESENT_SRC barriers |
| - venus: remember image view image |
| - venus: remember framebuffer attachments |
| - venus: remember cmd buffer render pass and framebuffer |
| - venus: remember cmd buffer fb attachments |
| - venus: add vn_image_memory_barrier_has_present_src |
| - venus: add vn_cmd_get_image_memory_barriers |
| - venus: clean up vn_get_intercepted_barriers |
| - venus: undo wsi iamge ownership transfer for Android |
| - venus: reland wsi image ownership transfer for Android |
| - venus: enable wsi image ownership transfer for common wsi |
| - vulkan/util: add vk_default_allocator |
| - radv: use vk_default_allocator |
| - v3dv: use vk_default_allocator |
| - tu: use vk_default_allocator |
| - anv: use vk_default_allocator |
| - venus: use vk_default_allocator |
| - venus: silence compiler warnings |
| - venus: query experimental features in one call |
| - venus: document the darkest corner of venus |
| - venus: move vn_renderer_sync_ops to vn_renderer |
| - venus: simplify vn_renderer_sync creation |
| - venus: update venus-protocol headers |
| - venus: add support for external fence on Android |
| - venus: add support for external semaphores on Android |
| - venus: clean up vn_physical_device_get_native_extensions |
| - venus: fix compatibility with older host drivers |
| - venus: be verbose about which physical devices are skipped |
| - vulkan/wsi: fix select_memory_type when all MTs are local |
| - venus: fix empty submits with BOs |
| - egl/surfaceless: try kms_swrast before swrast |
| - meson: allow egl_native_platform to be specified |
| - venus: clean up vn_AllocateMemory |
| - venus: suballocate memory in more cases |
| - vulkan/wsi/x11: do not inherit last_present_mode |
| |
| Christian Gmeiner (3): |
| |
| - ci: disable initrd support |
| - drm-shim: fix compile with glibc >= 2.33 |
| - ci: bare-metal: drop webdav support |
| |
| Connor Abbott (133): |
| |
| - ir3: Fix list corruption in legalize_block() |
| - ir3: Reduce max const file indirect offset base to 9 bits |
| - ir3, tu: Add compiler flag for robust UBO behavior |
| - tu: Correctly preserve old push descriptor contents |
| - tu: Handle robust UBO behavior for pushed UBO ranges |
| - tu: Handle null descriptors |
| - tu: Expose VK_EXT_robustness2 |
| - ir3/parser: Fix oob write with immediates array |
| - ir3: Improve cat1 modifier disassembly |
| - ir3: Assemble and disassemble swz/gat/sct |
| - ir3: Prevent oob writes to inputs/outputs array |
| - nir/lower_clip_disable: Fix store writemask |
| - ir3, tu: Cleanup indirect i/o lowering |
| - freedreno: Don't lower indirects in GLSL IR |
| - freedreno/a6xx: Better document SP_GS_PRIM_SIZE |
| - freedreno/a6xx: Fix SP_GS_PRIM_SIZE for large sizes |
| - tu: Fix SP_GS_PRIM_SIZE for large sizes |
| - ir3/postsched: Fix dependencies for a0.x/p0.x |
| - ir3/cp: Clone registers for compare-folding optimization |
| - ir3/sched: Use correct src index |
| - ir3/postsched: Use correct src index |
| - ir3/delay: Remove special case for array deps |
| - ir3/postsched: Fix ir3_postsched_node::delay calculation |
| - ir3/cp_postsched: Fixup SSA use pointer for direct reads |
| - ir3: Refactor nir->ir3 block handling |
| - ir3: Make predecessors an array |
| - ir3: Rework outputs |
| - ir3: Don't assume regs[1] exists in ir3_fixup_src_type() |
| - nir/lower_phis_to_scalar: Add "lower_all" option |
| - ir3/cf: Rewrite pass |
| - ir3: Use round-to-nearest-even for fquantize2f16 |
| - ir3: Call nir_lower_wrmask() again after lowering scratch |
| - ir3: Only use per-wave pvtmem layout for compute |
| - ir3: Introduce phi and parallelcopy instructions |
| - ir3: Add ir3_start_block() |
| - ir3: Readd support for translating NIR phi nodes |
| - ir3: Prepare for instructions with multiple destinations |
| - ir3: Improve register printing for SSA |
| - ir3: Add ir3_register::array.base |
| - ir3/delay: Fix full->half and half->full delay |
| - ir3: Add reg_elems(), reg_elem_size(), and reg_size() |
| - ir3: Make branch conditions non-SSA |
| - ir3: Rewrite delay calculation |
| - ir3/delay: Delete pre-RA repeat handling |
| - ir3/postsched: Don't use SSA source information |
| - ir3: Remove unused check_src_cond() |
| - ir3: Add dominance infrastructure |
| - ir3: Add pass to lower arrays to SSA |
| - ir3: Expose occupancy calculation functions |
| - ir3: Rewrite register allocation |
| - ir3/ra: Add a validation pass |
| - ir3: Remove right and left copy prop restrictions |
| - ir3/sched: Don't schedule collect early |
| - ir3/sched: Make collects count against tex/sfu limits |
| - ir3/sched: Consider unused destinations when computing live effect |
| - ir3: Add simple CSE pass |
| - ir3: Insert output collects in the main shader |
| - ir3: Copy propagate immed/const to meta instructions |
| - ir3: Improve printing of array parallelcopies/phis |
| - ir3/ra: Fix array parallelcopy confusion |
| - ir3: Make tied sources/destinations part of the IR |
| - ir3: Split read-modify-write array dests in two |
| - ir3: Update ir3_register::instr when cloning instructions |
| - ir3: Validate that ir3_register::instr is correct |
| - ir3: Add is_reg_special() |
| - ir3: Make ir3_instruction::address a normal register |
| - ir3: Split ir3_reg_create() into ir3_{src,dst}_create() |
| - ir3: Add separate src/dst count in ir3_instr |
| - ir3/legalize: Construct branch properly |
| - ir3: Add srcs/dsts arrays to ir3_instruction |
| - freedreno/isa: Convert to srcs/dsts |
| - freedreno/tests: Convert to srcs/dsts |
| - ir3/sched: Convert to srcs/dsts arrays |
| - ir3/core: Switch to srcs/dsts arrays |
| - ir3/ra: Switch to srcs/dsts arrays |
| - ir3/parser: Switch to srcs/dsts arrays |
| - ir3/array_to_ssa: Switch to srcs/dsts arrays |
| - ir3/legalize: Switch to srcs/dsts arrays |
| - ir3/print: Switch to srcs/dsts arrays |
| - ir3/validate: Switch to srcs/dsts arrays |
| - ir3/opts: Switch to srcs/dsts arrays |
| - ir3/frontend: Switch to srcs/dsts arrays |
| - ir3: Remove regs array |
| - ir3: Remove IR3_REG_DEST |
| - ir3/ra: Fix corner case in collect handling |
| - freedreno/a6xx: Make SP_XS_PVT_MEM_HW_STACK_OFFSET non-inline |
| - freedreno, tu: Set SP_XS_PVT_MEM_HW_STACK_OFFSET |
| - freedreno/computerator: Fix local_size typo |
| - ir3/sched: Speed up live_effect |
| - ir3: Stop creating dummy dest registers |
| - ir3: Prepare dest helpers for multi-dst instructions |
| - ir3: Add foreach_dst/foreach_dst_n |
| - ir3: Support multi-mov instructions |
| - ir3/delay: Support multi-mov instructions |
| - ir3/postsched: Support multi-mov instructions |
| - ir3/legalize: Support multi-mov instructions |
| - ir3: Use correct flags for movmsk & multi-mov |
| - ir3/validate: Support multi-mov instructions |
| - ir3: Print multi-mov instructions |
| - ir3: Add min gen for multi-mov instructions |
| - ir3/lower_parallelcopy: Use SWZ |
| - nir/subgroups: Replace lower_vote_eq_to_ballot with lower_vote_eq |
| - nir/subgroups: Support > 1 ballot components |
| - nir: Add read_invocation_cond_ir3 intrinsic |
| - tu, ir3: Plumb through support for CS subgroup size/id |
| - ir3/nir: Call nir_lower_subgroups |
| - ir3: Handle shared register liveness correctly |
| - ir3: Handle unreachable blocks |
| - ir3: Prevent propagating shared regs out of loops |
| - ir3: Better valid flags for shared regs |
| - ir3: Actually allow shared reg moves to be folded |
| - ir3: Fix shared reg delay |
| - ir3: Make MOVMSK use repeat |
| - ir3: Fix infinite loop in scheduler when splitting |
| - ir3/sched: Handle branch condition in split_pred() |
| - ir3: Cleanup ir3_legalize jump optimization |
| - ir3: Support any/all/getone branches |
| - ir3: Add subgroup pseudoinstructions |
| - ir3: Handle shared registers in lower_parallelcopy |
| - ir3: Implement nir subgroup intrinsics |
| - ir3: Fix convergence behavior for loops with continues |
| - ir3/legalize: Fix loop convergence behavior |
| - tu: Update subgroup properties |
| - ir3/nir: Lower indirect references of compact variables |
| - ir3: Add missing include to ir3_parser.y |
| - ir3: Add ir3_collect() for fixed-size collects |
| - ir3/lower_parallelcopy: Don't manually set wrmask |
| - ir3: Update .editorconfig and .dir-locals.el |
| - ir3: Manually reformat some places |
| - freedreno: Add some options to .clang-format |
| - ir3: Reformat source with clang-format |
| - ir3/print: Manual formatting fixups |
| - ir3: Preserve gl_ViewportIndex in the binning shader |
| |
| Corentin Noël (4): |
| |
| - ci: Use the caching proxy for Mesa artifacts |
| - ci: Re-enable virgl tesselation shader |
| - ci: Bump virglrenderer |
| - ci: actually run piglit tests with virgl |
| |
| Daniel Schürmann (33): |
| |
| - aco: fix additional register requirements for spilling |
| - aco: relax validation rules for p_reduce dst RegType |
| - driconf: set vk_x11_strict_image_count for Metro: Exodus |
| - aco/ra: prevent underflow register for p_create_vector operands |
| - radv: call nir_copy_prop() after load-store vectorization |
| - aco/ra: also prevent overflow register for p_create_vector operands |
| - aco: remove condition operand from branch in invert block |
| - radv,aco: scalarize all phis via nir_lower_phis_to_scalar() |
| - aco: simplify Phi RegClass selection |
| - aco/ra: only create phi-affinities for killed operands |
| - aco/ra: refactor affinity coalescing |
| - aco/ra: refactor register assignment for vector operands |
| - amd/ci: add hawaii-specific skip and fail lists |
| - aco/ra: handle copies of definition registers |
| - aco/ra: handle copies of copies better |
| - aco/util: replace DIV_ROUND_UP(n+1,m) by n/m+1 |
| - aco: reorder and cleanup #includes |
| - aco: add missing Licenses and remove Authors from files |
| - aco: add 'common/' and 'llvm/' prefix to #includes |
| - aco/meson: remove unnecessary dependencies |
| - aco: refactor SDWA opcode validation |
| - aco: remove (wrong) GCC array-bounds warning |
| - util/meson: include inc_gallium |
| - aco: add .clang-format file |
| - aco: Format. |
| - aco/meson: remove inc_gallium from include_directories |
| - aco: fix self-intersecting register swaps |
| - aco: fix extract_vector optimization |
| - aco/isel: avoid unnecessary calls to nir_unsigned_upper_bound() |
| - aco/insert_waitcnt: Remove many unnecessary wait_imm.combine() |
| - aco/live_var_analysis: change worklist to a single integer |
| - aco/optimizer: ensure to not erase high bits when propagating packed constants |
| - aco: include <cstddef> in aco_util.h |
| |
| Daniel Stone (69): |
| |
| - CI: Disable Panfrost and radeonsi |
| - CI: Disable all Panfrost/AMD/Iris automatic jobs |
| - CI: Disable rk3399-gru-kevin jobs for now |
| - doc: Gratuituous promotion of Wayland |
| - docs: Even more gratutious nitpicks |
| - Revert "CI: Disable rk3399-gru-kevin jobs for now" |
| - CI: Fix path confusion in OpenCL Piglit execution |
| - ci/zink: Skip flaky GLX test |
| - ci/radeonsi: Skip flaky glx-swap-copy test |
| - ci/windows: Artifact Meson build and test logs |
| - ci/windows: Re-enable Windows build |
| - ci: Add Piglit gl-1.0-blend-func to everyone's skips |
| - ci/lava: Iterate all job results, not just the first |
| - ci/lava: Handle proxy download failures |
| - ci/lava: Add validate-only mode to job submitter |
| - ci/lava: Add --dump-yaml option to submitter |
| - ci/bare-metal: Factor out environment to a separate script |
| - ci/bare-metal: Don't leak JWT into logs |
| - ci/lava: Move LAVA files to lava/ |
| - ci/lava: Pass JWT separately from environment variables |
| - ci/lava: Cosmetic reordering of job init |
| - ci/lava: Wrap submission in a shell script |
| - ci/lava: Clean up variable naming, document them |
| - ci: Make PIPELINE_ARTIFACTS_BASE a common variable |
| - ci: Add JOB_ARTIFACTS_BASE variable |
| - ci: Use JOB_ARTIFACTS_BASE for Piglit fails |
| - ci/lava: Use per-job rootfs overlay for environment |
| - ci/panfrost: Remove useless variable |
| - ci/lava: Generate job name from lava-submit.sh |
| - ci/lava: Remove unused arguments |
| - ci/lava: Add explicit fatal-error handler |
| - ci/lava: Disable stdout/stderr buffering |
| - ci/lava: Dump and artifact YAML again |
| - ci/lava: Avoid tee as it ruins exit status |
| - ci/piglit: Fix path to uploaded images |
| - ci/lava: Always upload Piglit replay images to MinIO |
| - ci/lava: Set PIGLIT_NO_WINDOW |
| - ci/lava: Explicitly start Xorg for Iris EGL tests |
| - ci/bare-metal: Rename BM_KERNEL_MODULES to HWCI_KERNEL_MODULES |
| - ci/lava: Use HWCI_KERNEL_MODULES to load modules |
| - ci/lava: Rename environment variable script |
| - ci/bare-metal: Try harder to do NTP |
| - ci/bare-metal: Reorder init so network comes first |
| - ci: Move bare-metal init script to common directory |
| - ci: Be consistent about install path |
| - ci/bare-metal: Consistently set library paths |
| - ci/bare-metal: Split init script into two stages |
| - ci/bare-metal: Move devcoredump capture to CI common |
| - ci/lava: Start using devcoredump captures |
| - ci: Consistent pass/fail result output |
| - ci: Unify {BM,LAVA}_START_XORG environment |
| - ci: Unify {BARE_METAL,LAVA}_TEST_SCRIPT environment |
| - ci/bare-metal: Set CPU and GPU governors to max, disable GPU runtime PM |
| - ci/lava: Pass MinIO path on the command line |
| - ci/lava: Use common stage-2 init |
| - ci/lava: Drop bitrotten fastboot support |
| - ci/lava: Make kernel image type a normal argument |
| - ci/lava: Generate YAML from Python, not Jinja |
| - llvmpipe: Add handle export for resource_get_param |
| - Revert "ci: disable panfrost t760 jobs" |
| - CI: Disable LAVA devices for maintenance |
| - Revert "CI: Disable LAVA devices for maintenance" |
| - util/disk_cache: Don't leak when cache is empty |
| - panfrost/genxml: Decode Bifrost index-driven vertex jobs |
| - ci/panfrost: Temporarily disable sun50i/RK3288 |
| - Revert "ci/panfrost: Temporarily disable sun50i/RK3288" |
| - vulkan/wsi/wayland: Initialise wl_shm pointer in VkImage |
| - egl/wayland: Error on invalid native window |
| - egl/wayland: Allow EGLSurface to outlive wl_egl_window |
| |
| Danylo Piliaiev (36): |
| |
| - nir: add lowering pass for helperInvocationEXT() |
| - turnip: implement VK_EXT_shader_demote_to_helper_invocation |
| - turnip: implement VK_KHR_shader_terminate_invocation |
| - ir3: treat 16b imul as mul.s24 |
| - turnip: enable shaderInt16 |
| - ir3: do not double threadsize when exceeding branchstack limit |
| - ir3: make possible to specify branchstack up to 64 |
| - tu: do not corrupt unwritten render targets |
| - ir3: do not move varying inputs that depend on unmovable instrs |
| - ir3: do not fold cmps from different blocks with non-null address |
| - ir3: memory_barrier also controls shared memory access order |
| - ir3: update bar/fence bits in accordance to blob |
| - turnip: implement VK_KHR_vulkan_memory_model |
| - docs: mark off VK_KHR_vulkan_memory_model for turnip |
| - turnip,freedreno/a6xx: SP_BLEND_CNTL has per-mrt blend enable bit |
| - freedreno/a5xx: SP_BLEND_CNTL has per-mrt blend enable bit |
| - turnip: copy all layers specified in vkCmdCopyImage |
| - ci/turnip: drop fail annotation for float_control tests |
| - ci/turnip: drop fail annotation for image.extend_operands_spirv1p4.* |
| - turnip: do not ignore early_fragment_tests |
| - turnip: make possible to create read-only bo with tu_bo_init_new |
| - turnip: make cmdstream bo's read-only to GPU |
| - turnip: place a limit on the growth of BOs |
| - freedreno: reduce the upper bound of IB size by one |
| - turnip: reset push descriptor set on command buffer reset |
| - turnip: emit vb stride dynamic state when it is dirty |
| - turnip: fix register_index calculations of xfb outputs |
| - turnip: implement VK_EXT_provoking_vertex |
| - turnip: do not re-emit same vs params |
| - turnip: early exit in tu6_draw_common to save cpu cycles |
| - freedreno/computerator: pass iova of buffer to const register |
| - freedreno/isa: add uoffset type to print positive-only offsets |
| - ir3: add ldg.a,stg.a which allow complex in-place offset calculation |
| - glsl: Prohibit implicit conversion of mem parameter in atomicOP functions |
| - ir3: add newly found shlg.b16 instruction |
| - freedreno: fix wrong tile aligment for 3 CCU gpu |
| |
| Dave Airlie (146): |
| |
| - iris: move get_time into a static in bufmgr code. |
| - iris: move target to isl dim translate to inline. |
| - lavapipe: add support for non-dri loader on linux |
| - llvmpipe: split screen init up. |
| - llvmpipe: wrap late screen init with a mutex. |
| - llvmpipe: delay late screen creation until context init. |
| - lavapipe: fix mipmapped resolves. |
| - lavapipe: mark event_storage as volatile |
| - intel: move brw_ff_gs_prog_key/data to compiler. |
| - intel/compiler: add support for compiling fixed function gs |
| - i965: port fixed function geom shader to use compiler paths |
| - i965: drop old brw ff gs code. |
| - intel/genxml: align gen4/5 xml for store data immediate |
| - intel/genxml: rewrite the prefilterop xml to be more consistent. |
| - intel/gemxml: move blitter command to render on gen4/5 |
| - intel/genxml: fix raster op fields on gen4/5 |
| - intel/decoder: fixup batch decoder for binding tables on gen4/5 |
| - intel/decoder: add gen4/5 geometry state decode |
| - gallivm: handle texture arrays in non-fragment shaders with lod. |
| - llvmpipe: fix non-multisampled rendering to multisampled framebuffer |
| - llvmpipe: add the interesting bit of cpu detection to the cache. |
| - st/nir: always revectorise if scalarising happens. |
| - intel/gfx6: move xfb_setup outside the gs compiler into the driver. |
| - intel/isl: decrease isl_format_layouts size by 36k |
| - intel/isl: convert null surface fill to a struct. |
| - intel/isl: add levels and minimum array element to null fill |
| - intel/isl: add blend enable flag to gen4/5 |
| - u_blitter: fix fs used when no color emitted |
| - u_blitter: fix stencil blit fallback for crocus. |
| - iris: drop unused function declaration |
| - nir/edgeflags: update outputs written when lowering edge flags. |
| - st/mesa: also disable other int textures |
| - intel/decode: handle gen4/5 WM state fragment shaders |
| - intel: reorder base program key. |
| - intel/compiler: add flag to indicate edge flags vertex input is last |
| - crocus: initial gallium driver for Intel gfx 4-7 |
| - ci: add crocus to the build tests |
| - crocus: Don't call SET_TILING for dmabuf imports |
| - crocus: Make iris_bo_import_dmabuf take a modifier |
| - crocus: introduce main resource configuration helper. |
| - crocus: Drop buffer support in resource_from_handle |
| - crocus: hook up memory object creation from handle |
| - crocus: hook up resource creation from memory object |
| - crocus: plumb device/driver UUID generators |
| - crocus: enable GL_EXT_memory_object feature on gen7 |
| - crocus: fix scanout tiling so glamor/modesetting can work. |
| - crocus: fixed some missing WM dirtys. |
| - crocus: fixup render aux usage function. |
| - crocus: disable Z16 |
| - crocus/gen6: fix depth blit blorp regression. |
| - i965: fix regression in pipe control on g45 |
| - crocus: drop dead gen prototypes. |
| - crocus: fixup stray tab |
| - crocus: rename genX proto functions to avoid iris conflicts. |
| - crocus: fixup workaround_bo to match 965. |
| - crocus: convert a bunch of is_haswell into verx10 checks. |
| - crocus: refactor blend state code. |
| - crocus/gen8: limit some pipe controls to gen7/hsw |
| - crocus: limit texture gather workarounds to gen7/hsw |
| - crocus/stencil: limit stencil workaround to gen7 |
| - crocus/query: add gen8 support to queries by extending hsw checks |
| - crocus: extend l3 config setup to gen8 |
| - crocus/gen8: add push constant support (extend hsw) |
| - crocus/gen8: extend some compute + state functions to gen8 |
| - crocus/gen8: extend image support to gen8 |
| - crocus: extend hsw cut index to gen8 |
| - crocus/gen8: extend predicate handling to gen8. |
| - crocus/gen8: add sampler / border color support for gen8 |
| - crocus/gen8: add l3 config support |
| - crocus/gen8: extending gen7 binding table pointers |
| - crocus/gen8: limit vertex buffer workarounds to ivb |
| - crocus/gen8: add raw pipe control support for gen8 workarounds |
| - crocus/gen8: add support for vertex instancing and index buffers. |
| - crocus/gen8: state base address + misc setup state. |
| - crocus/gen8: add VF topology support |
| - crocus/gen8: add PMA fix from iris |
| - crocus/gen8: add streamout support |
| - crocus/gen8: add SBE swiz support |
| - crocus/gen8: add VF SGVS support. |
| - crocus/gen8: add PS blend command support. |
| - crocus/gen8: refactor blend state for gen8 |
| - crocus/gen8: add rasterizer state changes. |
| - crocus/gen8: add viewport support |
| - crocus/gen8: add depth stencil state support |
| - crocus/gen8: port over vs/gs/ds state changes. |
| - crocus/gen8: port over ps/wm state changes from iris. |
| - crocus/gen8: port over VFE/compute state changes |
| - croucs/gen8: handle gfx8 surface aux addr reloc. |
| - crocus/gen8: handle sampler differences |
| - crocus/gen8: hookup gen8 state generators |
| - crocus/gen8: add support for cherryview (env var for bdw) |
| - croucs: limit stencil swizzle change to older generations |
| - crocus/bufmgr: fix userptr left over fail |
| - crocus: Explicitly cast value to uint64_t |
| - crocus: free context state properly. |
| - crocus: fix vertex buffer leak on screen end. |
| - crocus: fix batch state bo leak |
| - meson/crocus: add prefer-crocus option. |
| - crocus/query: poll the syncobj in the no wait situation |
| - intel/genxml: fix gfx6 GS SVB_INDEX encoding |
| - crocus/gfx6: fix sampler view first level. |
| - crocus: dirty blend state more often. |
| - crocus: Avoid replacing backing storage for buffers with no contents |
| - crocus/gfx6: always be dirtying gs attachments for xfb |
| - crocus: fix another printf specifier. |
| - crocus/gen8: add back z16 support for gen8 |
| - crocus: disable Z16 unorm textures on pre-gen8 as well. |
| - gallium/sw: add sw_vk bit to avoid having to futz with env vars for lavapipe |
| - zink: drop getenv hacking now that gallium is fixed. |
| - iris: make iris_bind_reserve_3d and Wa_1604061319 only check for dirty render bindings |
| - crocus: cleanup some deadcode in the gen5 blend emit |
| - crocus: expose ARB_blend_func_extended on gen 45/50 |
| - crocus/gen5: enable support for GL_EXT_gpu_shader4 |
| - crocus: fix crash on index buffer rebinding. |
| - crocus: fixup index buffer dirtying. |
| - draw: fix tessellation output vertex size calculation |
| - draw/tess: write correct primitive id into vertices |
| - crocus: inline the d/s resource handling functions |
| - crocus: don't update draw parameters unless needed |
| - crocus: optimise bo_unref path a little. |
| - crocus: inline group_index<->bti |
| - crocus: reorder version checks on indirect xfb |
| - crocus: restrict prim_restart on index buffer check to pre-hsw |
| - crocus: support rebinding streamout target buffers |
| - crocus: use threaded context base classes |
| - crocus/tc: init/deinit threaded resource |
| - crocus: add unsync transfer pool |
| - crocus: enable threaded context support |
| - ac: fix win32 build |
| - crocus/gen8: fix wrap mode needs border color. |
| - crocus: add GL_CLAMP emulation in driver again. |
| - vulkan/wsi/wl: add wl_shm support for lavapipe. |
| - lavapipe: add the separate depth/stencil layout enable. |
| - crocus: use simple_mtx in the bufmgr |
| - lvp: fixup multi draw memcpys |
| - draw: handle resetting draw_id between instances. |
| - softpipe/aniso: move DDQ calculation to after scaling. |
| - crocus/gen4-5: fix ff gs emit on VS vue map change. |
| - llvmpipe: add support for time elapsed queries. |
| - draw/llvmpipe: multiply polygon offset units by 2 |
| - teximage: return correct desktop GL error for compressedteximage |
| - crocus/gen4: restrict memcpy mapping to gen5 |
| - intel/fs: restrict max push length on older GPUs to a smaller amount |
| - crocus/gen45: fix mapping compressed textures |
| - intel/genxml: fix raster operation field in blt genxml |
| - crocus: add support for set alpha to one with blt. |
| |
| Dmitry Baryshkov (2): |
| |
| - freedreno/regs: split DSI PHY registers to separate xml files. |
| - freedreno/regs: split old/not used phy registers to separate DB |
| |
| Drew Davenport (1): |
| |
| - radeonsi: Report multi-plane formats as unsupported |
| |
| Duncan Hopkins (3): |
| |
| - zink: Correct compiler issue with have_moltenvk member having been moved. |
| - gallium/dri: Guard DRI driver global variables on MacOS if Zink is enabled. |
| - zink: Fix MacOS compiling issues |
| |
| Dylan Baker (27): |
| |
| - meson: OpenMP is supposed to be optional |
| - docs: add release notes for 21.0.3 |
| - docs: update sha256 sum for mesa 21.0.3 |
| - docs: update calendar and link releases notes for 21.0.3 |
| - docs: update calendar for 21.1.0-rc1 |
| - docs: update calendar for 21.1.0-rc2 |
| - docs: update calendar for 21.1.0-rc3 |
| - meson/vulkan: fix linkage on windows |
| - docs: Add calendar entries for 21.2 release candidates. |
| - VERSION: bump for 21.2-rc1 |
| - .pick_status.json: Update to f40a08d25c91256cd3dff0211b8e10d5bbb3734e |
| - .pick_status.json: Update to a62973580b7846f2213cbd2589e9473c26596683 |
| - .pick_status.json: Update to 27534a49cf3872646cb8ef9371707d74a81b1986 |
| - VERSION: bump for 21.2-rc2 |
| - .pick_status.json: Update to b45cddda183230232937387f91d009500b2372c9 |
| - .pick_status.json: Update to 49908c602ffd2d84063effa7ddd0ee842be41a89 |
| - VERSION: bump for 21.2.0-rc3 |
| - .pick_status.json: Update to dff0d9911d176802b54890c796e19f56c50f24e1 |
| - .pick_status.json: Update to b8e29e89366a5264391dc7c10e778330b7add66a |
| - freedreno/ir3: Add build id to the disassembler test |
| - .pick_status.json: Mark 8cb795b4772f882024b20c4d4b051b2411dd1a8c as denominated |
| - .pick_status.json: Update to 87b0962fef4e447a2ea9c76a611aa20b109a259d |
| - .pick_status.json: Update to 842b8c8965327615f4692384a905dd63f1fba63d |
| - .pick_status.json: Update to 97be8e42e42f3b739c3de808553094f86ad8879f |
| - bin/gen_release_notes: Add basic tests for parsing issues |
| - bin/gen_release_notes: Don't consider issues for other projects |
| - bin/gen_release_notes: Fix commits with multiple Closes: |
| |
| Eleni Maria Stea (5): |
| |
| - egl: fix in expected type |
| - util: replaced ENODATA with ENOATTR for non-Linux systems |
| - util: Removed unused statement from FreeBSD build |
| - intel: struct bitset is renamed to brw_bitset |
| - intel: PAGE_SIZE used in allocators shouldn't be defined on FreeBSD |
| |
| Ella-0 (1): |
| |
| - anv: expose primary node to VK_EXT_physical_device_drm even when VK_KHR_display is not enabled |
| |
| Emil Velikov (1): |
| |
| - gbm: list to stderr all the missing extension |
| |
| Emma Anholt (251): |
| |
| - ci/freedreno: Merge a630 piglit to a single job. |
| - freedreno: Fix YUV sampler regression. |
| - ci/virgl: Mark a couple of new Crash tests as flakes. |
| - ci/freedreno: Skip some precision tests on a530. |
| - nir_to_tgsi: Use ARL instead of UARL in the !native_integers case. |
| - nir: Generate load_ubo_vec4 directly for !PIPE_CAP_NATIVE_INTEGERS |
| - ci/lavapipe: Don't include deqp's shader_cache in the artifacts. |
| - ci/lava: Return the run's results/ artifacts from the DUTs. |
| - ci/piglit: Always include the HTML summary in a run. |
| - ci/lava: Point the shader cache at tmpfs. |
| - mesa: Remove dead _mesa_unpack_rgba_block(). |
| - util: Switch the non-block formats to unpacking rgba rows instead of rects. |
| - util/format: Add some NEON intrinsics-based u_format_unpack. |
| - panfrost: Enable packed uniforms. |
| - zink: Enable PIPE_CAP_PACKED_UNIFORMS. |
| - ci: Build deqp-egl targeting x11_egl_glx |
| - ci/llvmpipe: Test dEQP-EGL against Xvfb. |
| - ci/freedreno: Test dEQP-EGL against Xorg. |
| - mapi: Respect MESA_DEBUG=silent for no-op debug output. |
| - freedreno: Mark glsl-fs-fogscale as a Fail. |
| - freedreno/a6xx: Don't try to do Z-as-RGBA blits for mismatched formats. |
| - util: Fix big-endian handling of z/s formats. |
| - mesa: Deduplicate _mesa_pack_ubyte_stencil_row() |
| - mesa: Deduplicate _mesa_pack_float_z_row(). |
| - mesa: Deduplicate _mesa_pack_uint_z_row(). |
| - mesa: Remove dead _mesa_get_pack_float_z_func(). |
| - msea: Move z24s8-to-z24s8 packing fastpath to swrast. |
| - mesa: Move per-pixel Z pack functions to swrast. |
| - mesa: Remove dead _mesa_pack_ubyte_rgba_rect(). |
| - mesa: Replace _mesa_pack_ubyte_rgba_row() with pack_ubyte_rgba_8unorm(). |
| - ci/radeonsi: Mark a glx_arb_sync_control/timing flake. |
| - turnip: Only write the tu_RegisterDeviceEXT() out fence on success. |
| - ci: Add missing vulkan dep for freedreno (turnip) and v3dv test jobs. |
| - u_format: Fix z32_s8x24 s8 unpacking on big-endian. |
| - u_format: Add missing BE swizzles for R8SG8SB8UX8U_NORM |
| - ci/freedreno: Mark dEQP-EGL flakes reported on IRC since its introduction. |
| - ci/freedreno: Mark new flakes from the go-fast branch. |
| - ci/freedreno: Mark another recent piglit flake. |
| - ci/freedreno: Fix the recent-a5xx-texture-flakes matches. |
| - ci/freedreno: Add another db820c flake that's appeared in the last few months. |
| - tgsi: Mark the tgsi_exec_channel and tgsi_double_channel ALIGN16. |
| - tunrip: Add support for VK_EXT_separate_stencil_usage. |
| - ci/freedreno: Mark a5xx texture gather as flaky. |
| - turnip: Demote API version to 1.1. |
| - ci/llvmpipe: Add testing of gles3/31/gl. |
| - ci/lavapipe: Add fractional NIR stress test job. |
| - freedreno/a5xx: Fix up border color pointers. |
| - gallium/tgsi_exec: Drop the unused dst_datatypes from dest stores. |
| - tgsi_exec: Drop unused destination dimension support. |
| - tgsi_exec: Mark the store file default case as unreachable. |
| - gallium/tgsi_exec: Simplify bounds checks on the const file. |
| - turnip: Switch to the shared vulkan ICD generator. |
| - turnip: Move the extension tables to tu_device.c |
| - ci/freedreno: Add another daily dose of a530 flakes. |
| - turnip: Drop wideLines properties since we don't support wide lines. |
| - turnip: Claim 2 discrete queue priorities. |
| - freedreno: Update editorconfig and emacs settings for freedreno reformat. |
| - ci/turnip: Clean up some stale fail annotations. |
| - ci/turnip: Add some links to issues and MRs for some test failures. |
| - turnip: Drop fail annotation for driver_properties. |
| - ci: Switch to apitraces for glmark2 |
| - ci/panfrost: Add some more traces to replay. |
| - ci/iris: Add some more traces to replay. |
| - ci/freedreno: Skip refract on a306 now that it hangchecks sometimes. |
| - midgard: Fix type for vertex_builtin_arg() and compute_builtin_arg(). |
| - ci/freedreno: Skip a test that's taking out the a530 boards. |
| - ci/freedreno: Mark two more recent intermittent a530 flakes. |
| - ci/deqp: Make DEQP_EXPECTED_RENDERER a required regex for VK like for GLES. |
| - ci/intel: Add test jobs for dEQP. |
| - vulkan: Avoid stomping array padding in the MemoryProperties wrapper. |
| - mesa/st: Only use 16-bit ints or floats in the NIR path. |
| - i915g: Disable 3D-pipeline clears. |
| - i915g: Switch batchbuffer dumping to mesa_logi(). |
| - i915g: Fix dumping of the FS in batchbuffers. |
| - ci/i915g: Introduce manual testing of i915g using anholt's runner. |
| - i915g: Make the FS for compile failures write red instead of DIFFUSE. |
| - i915g: Add support for the .Absolute flag on TGSI srcs. |
| - i915g: Stop advertising support for indirect addressing in the FS. |
| - i915g: Fix writing of undefined depth value if not writing any outputs. |
| - i915g: Fix undefined results for TGSI_OPCODE_KILL |
| - ci/iris: Switch GLK back to manual testing. |
| - ci/freedreno: Clear compswap flake annotation. |
| - ci/freedreno: Clear stale validation failure flake annotation. |
| - ci/freedreno: Drop a630 flake annotation from the go-fast changes. |
| - ci/freedreno: Add a link explaining get_display_plane_capabilities |
| - ci/freedreno: Drop VK flake annotations not seen in the last ~year. |
| - ci/freedreno: Consolidate ssbo.fragment_binding_array flake annotation. |
| - ci/freedreno: Mark a630 glx-visuals-depth/stencil as piglit flakes. |
| - ci/freedreno: Also mark waitformsc as flaky. |
| - ci/freedreno: Add glx-copy-sub-buffer to flakes on a530 and a630. |
| - mesa/st: Fix iris regression with clip distances. |
| - ci/freedreno: Add another a630 piglit flake. |
| - ci/freedreno: Turn off default a530 quick_gl testing, do full quick_shader. |
| - turnip: Reorganize copy_format()'s switch statement. |
| - turnip: Make sure that SNORM blits don't clamp ambiguous -1.0 values. |
| - Revert "ci: Configure DUTs for max performance" |
| - ci: Add known-flake handling for the IRC flake reports |
| - ci: Move the flakes channels to OFTC |
| - util: Add a helper macro for defining initial-exec variables. |
| - android: Fix ELF TLS support. |
| - ci/android: Update to building for SDK 29 by default. |
| - u_format: Fix some pep8 in u_format_parse.py. |
| - u_format: Drop redundant .name init. |
| - u_format: Move the BE swizzle computation into Format init. |
| - u_format: Use the nice helper for reversing an array. |
| - u_format: Assert that array formats don't include BE swizzles. |
| - u_format: Define tests for r3g3b2 formats and fix BE swizzles for them. |
| - u_format: Fix the BE channel ordering for R5G5B5A1_UINT. |
| - u_format: Sanity check the BE channels for all bitmask formats. |
| - u_format: Sanity check that BE swizzles are appropriately mapped from LE. |
| - u_format: Use the computed BE channels/swizzles for bitmask formats. |
| - ci/freedreno: Add some more known flakes from recent marge runs. |
| - docs/freedreno: Update for the fanin/fanout -> collect/split rename. |
| - docs/freedreno: Rewrite the section on array access. |
| - tgsi_exec: Garbage-collect the FAST_MATH path. |
| - u_math: Reduce fast-log2 table size from 65k entries back to 256. |
| - llvmpipe: Don't call util_init_math(). |
| - ra: Add a unit test. |
| - ra: Document that class index is allocated in order, use that in r300. |
| - ra: Use struct ra_class in the public API. |
| - ra: Add fast-path support for register classes of contiguous regs. |
| - vc4: Use the ra_alloc_contig_reg_class() function to speed up RA. |
| - v3d: Use the ra_alloc_contig_reg_class() function to speed up RA. |
| - intel/fs: Use ra_alloc_contig_reg_class() to speed up RA. |
| - intel/vec4: Use ra_alloc_contig_reg_class() to reduce RA overhead. |
| - lima: Use ra_alloc_contig_reg_class(). |
| - util/ra: Use the conflicting neighbor to skip unavailable registers. |
| - ci/i915g: Fix incorrect expectation. |
| - i915g: Make sure we don't try to texture from the const file. |
| - ci/lava: Finish garbage-collecting the TEST_SUITE variable |
| - ci: Update piglit and deqp/piglit-runner. |
| - ci/freedreno: Enable running all of piglit_gl for a530's manual test. |
| - ci/piglit: Skip WGL on all the Linux runs. |
| - ci/fastboot: Add a serial timeout to catch fastboot prompt failure. |
| - ci/fastboot: Consistently restart the run on intermittent conditions. |
| - ci/iris: Enable piglit testing on AML-Y. |
| - ci: Disable Xorg's screensaver entirely. |
| - ci/deqp: Drop stress/perf skips lists. |
| - ci/deqp: Skip flush_finish on all CI jobs. |
| - ci/softpipe: Move the flake to the flakes list. |
| - ci: Add a flakes IRC channel for llvmpipe/softpipe. |
| - ci/deqp: Skip dEQP-VK.wsi.display.get_display_plane_capabilities |
| - ci/piglit: Move the WGL skip to a common skips file. |
| - ci/piglit: Skip glx_arb_sync_control@timing.* on all systems. |
| - freedreno: Drop batch-cache orphan tracking. |
| - freedreno: Make a bunch of the batch cache take ctx as the arg. |
| - freedreno: Drop a bit of indirection around the batch cache flush path. |
| - freedreno: Be more strict about QUERY_AVAILABLE to simplify the code. |
| - freedreno: Fix batch reference handling in flush_resource(). |
| - freedreno: Move the !MAP_WRITE write batch refcounting to the branch. |
| - freedreno: Remove broken back_blit optimization. |
| - freedreno: Flush batches upon destroying the ctx. |
| - freedreno: Add perf_debug() for our software conditional rendering. |
| - freedreno: Move FD_MESA_DEBUG=msgs output to mesa_logi. |
| - freedreno/fdl: Give the tiling mode a nice name in debug dumps. |
| - freedreno: Add more detailed blit debug in FD_MESA_DEBUG=msgs. |
| - freedreno: Skip staging blits from uninitialized resources. |
| - freedreno: Add some cheza flakes from the last week. |
| - i915: Disable vertex texturing and delete the code. |
| - i915: Drop assertion failure about seeing each const decled once. |
| - i915g: Allow fragment coord conventions TGSI properties to be set. |
| - nir/lower_int_to_float: Make sure the cursor is in the right spot. |
| - nir: Do peephole select on other instructions if the limit is ~0. |
| - nir_to_tgsi: Fix internal handling of NIR uints for !CAP_INTEGERS |
| - nir_to_tgsi: Support integer sysvals on !CAP_INTEGERS hardware. |
| - i915g: Handle fragment depth being in OUT[1] not OUT[0]. |
| - i915g: Switch to using nir-to-tgsi. |
| - i915g: Add triangle provoking vertex support. |
| - freedreno: Move some driver debug printfs to mesa_logd. |
| - freedreno/ir3: Move the assert output to mesa_loge(). |
| - util/log: Add a streaming printf interface. |
| - freedreno/ir3: Use mesa_log_stream() for ir3 disassembly. |
| - freedreno/ir3: Move the native code output to mesa_log as well. |
| - nir: Add an interface for logging shaders with mesa_log*. |
| - freedreno/ir3: Move NIR printing to mesa_log. |
| - gallium/draw: Garbage collect draw_set_force_passthrough |
| - gallium/draw: Garbage collect draw_pt_fetch_emit. |
| - gallium/util: Introduce a helper for finding whole-resource blits. |
| - freedreno: Move the rsc-based batch flushing to helper functions. |
| - freedreno: Handle full blit discards by invalidating the resource. |
| - freedreno: Cooperate with tc to stop checking the BC for resource_busy(). |
| - ci/llvmpipe: Mark two more multithread program link flakes. |
| - i915g: Remove nr_cbufs loop. |
| - i915g: Create an i915_surface for our pipe_surfaces. |
| - i915g: Compute 3DSTATE_BUF_INFO flags at surface create time. |
| - i915g: Move cbuf color swizzle lookup to CSO creation time. |
| - i915g: Simplify color write mask setup. |
| - i915g: Use the color swizzle to reshuffle the blend const color. |
| - i915g: Clear xfails for vertex texturing. |
| - freedreno: Fix leak of the screen hash table. |
| - i915g: Fix GL_ARB_copy_buffer assertion fails. |
| - i915g: Fix bad naming of depth texture formats. |
| - i915g: Finish out blend factor overrides for both RGBx and A8. |
| - ci/i915g: Skip the piglit glx tests since we're not running X. |
| - freedreno/ir3: Report RA failure with mesa_loge(). |
| - turnip: Link more MRs and issues related to our xfails. |
| - turnip: Use vk_startup_errorf() in more startup paths. |
| - ci/turnip: Document create_instance_device_intentional_alloc_fail's fail. |
| - turnip: Disable buffer texturing on 422 formats. |
| - Revert "freedreno: Cooperate with tc to stop checking the BC for resource_busy()." |
| - nir: Add a helper for chasing movs with nir_ssa_scalar(). |
| - turnip: Short-circuit if ladder generation for constant index SSBO/UBOs. |
| - i915g: Apply clang-format. |
| - i915g: Bake the decls and program together. |
| - i915g: Allow use of I915_DEBUG= options on non-DEBUG builds. |
| - i915g: Enable dumping of fragment shaders under I915_DEBUG=fs. |
| - i915g: Use the normal compile error path for empty FSes. |
| - i915g: Log program compile errors to mesa_loge(). |
| - i915g: Stop translating the fragment program on the first error. |
| - i915g: Improve logging of unsupported opcodes. |
| - i915g: replace "uint" with normal uint32_t. |
| - i915g: Use stdbool.h instead of custom bools. |
| - i915g: Remove redundant p->error setting. |
| - i915g: Mark program errors on setting up temps, constants, and immediates. |
| - i915g: Fix off-by-one in constant count assertion. |
| - intel: Early exit from inst_is_in_block(). |
| - i915g: Finish the uint -> uint32_t conversion. |
| - i915g: Add the nice cube map layout comments from i915c. |
| - i915g: Fix FS debug dumping for declarations. |
| - i915g: Delete redundant i915_hw_sampler_views atom. |
| - i915g: Add curly braces for normal mesa style (and helps clang-format) |
| - i915g: Set up the cube map texture wrap modes. |
| - freedreno: Update comments about PIPE_BUFFER shadowing. |
| - freedreno: swap ->valid when shadowing resources. |
| - freedreno/a5xx: Make sure to mark blit read/write access in the BC. |
| - freedreno: Stop manually marking blit dst buffers as valid. |
| - freedreno: Swap needs_ubwc_clear when shadowing. |
| - freedreno: Flush the shadowed resource's write batch up front. |
| - i915g: Add support for per-vertex point size. |
| - i915g: whitespace fixup from the cube map fix. |
| - i915g: Force 1D textures to use wrap mode for the Y coordinate. |
| - i915g: Make sure the 1D texture Y channel is initialized. |
| - anv: Fix unused var warning on release builds from an assertion. |
| - nir: Add a nir_instr_remove that recursively removes dead code. |
| - nir: Use remove_and_dce for nir_shader_lower_instructions(). |
| - nir: Free the instructions in a DCE instr removal. |
| - i915g: Fix writemasking of SEQ/SNE/SSG. |
| - nir_to_tgsi: Run copy prop (and thus dce) after lower_bool_to_float. |
| - nir_to_tgsi: Declare immediates as float on non-native-ints hardware. |
| - turnip: Fix allocation size for vkCmdUpdateBuffer. |
| - i915g: Fix dumping of 3DSTATE_BACKFACE_STENCIL_OPS. |
| - i915g: Fix backface stencil when front_ccw is set. |
| - ci: Make sure that we build the piglit dmabuf tests. |
| - freedreno: Suballocate our long-lived ring objects. |
| - freedreno/a6xx: Reduce the size of the config stateobj allocation. |
| - freedrneo/a6xx: Reduce the size of the long-lived texture stateobj. |
| - freedreno/a6xx: Allocate just enough memory for SO state, only if we do SO. |
| - freedreno: Optimize duplicate obj-obj ring relocs. |
| - i915g: Fix release build compiler warnings. |
| - ci: Enable testing of i915g in the the debian -Werror release build. |
| - freedreno: Lock access to msm_pipe for RB object suballocation. |
| |
| Enrico Galli (10): |
| |
| - microsoft/compiler: zero out unused WebGPU system values |
| - microsoft/compiler: Remove de-duplication of arbitrary semantic names |
| - d3d12, microsoft/compiler: Switching semantic names to TEXCOORD |
| - d3d12, microsoft/compiler: Moving driver_location allocation to compiler |
| - util: Add simple test for util_qsort_r |
| - util: Add qsort_r/s args adapter for MSVC and BSD/macOS |
| - nir: Add modes filter to nir_sort_variables |
| - microsoft/compiler: Switch io sort to use nir_sort_variables_with_modes |
| - microsoft/spirv_to_dxil: Add drive_location assignment |
| - microsoft/compiler: Add support for get_ssbo_size to translator |
| |
| Eric Engestrom (15): |
| |
| - VERSION: bump to 21.2.0-devel |
| - docs: reset new_features.txt |
| - egl/x11: don't forget to exit the attrib list loop |
| - docs: add release notes for 21.1.0 |
| - docs: add release notes for 21.1.1 |
| - docs: update calendar and link releases notes for 21.1.0 |
| - docs: update calendar and link releases notes for 21.1.1 |
| - docs/release-calendar: add the schedule for the 21.1 branch |
| - docs: add release notes for 21.1.2 |
| - docs: update calendar and link releases notes for 21.1.2 |
| - docs: add release notes for 21.1.3 |
| - docs: update calendar and link releases notes for 21.1.3 |
| - docs: add release notes for 21.1.4 |
| - docs: update calendar and link releases notes for 21.1.4 |
| - docs/release-calendar: add a few more 21.1 releases |
| |
| Erico Nunes (7): |
| |
| - gallium/hud: create vs_text to match fs_text |
| - gallium/hud: extend check for has_srgb |
| - docs/lima: add an initial page for Lima |
| - lima: enable z16 format |
| - lima: add reload command to the command dump |
| - meson: kmsro: require dri3 for X11 |
| - lima: avoid crash with negative viewport values |
| |
| Erik Faye-Lund (193): |
| |
| - zink: fix stencil-export cap emission |
| - lavapipe: resolve border-color when creating sampler |
| - lavapipe: implement VK_EXT_custom_border_color |
| - nir/lower_tex: do not stumble on 16-bit inputs |
| - zink: document requirement of VK_EXT_custom_border_color |
| - gallivm: handle 16-bit input in i2b32 |
| - gallivm: run nir_opt_algebraic_late |
| - gallivm: add 16-bit integer support |
| - zink: do not require vulkan memory model for shader-images |
| - docs: write basic meta-documentation |
| - zink: do not read outside of array |
| - docs: remove out-of-date gles info |
| - docs: remove documentation of MESA_CI_VISUAL |
| - docs: remove documentation of MESA_PRIVATE_CMAP |
| - docs: remove documentation of MESA_HPCR_CLEAR |
| - docs: nest cherry-pick example under note |
| - docs: use tables instead of pre-formatted text |
| - docs: use math notation for example matrices |
| - docs: use code-block for console-content |
| - docs: use code-block for glsl |
| - docs: use code-block for c |
| - docs: use code-block for ini |
| - zink: only emit extended-formats cap if needed |
| - zink: remove memory-model leftovers |
| - docs: fixup link to extension |
| - docs: fix quoting around a few limits |
| - zink: correct image cap checks |
| - docs: add missing zink-requirement |
| - docs: someome -> someone |
| - zink: enable required instance ext |
| - zink: make zink_binding private |
| - zink: remove stray semicolons |
| - zink: fixup bad indentation |
| - docs: remove out-of-date versions doc |
| - zink: fix shader-image requirements |
| - zink: correct an extension-link |
| - docs: fixup indentation of radeonsi envvar values |
| - docs: document r600 envvars |
| - zink: use UINT32_MAX instead of UINT_MAX |
| - zink: respect bit-size of dref-result |
| - zink: run nir_opt_algebraic_late |
| - zink: always lower function-temp derefs |
| - zink: support emitting 16-bit int types |
| - zink: enable 16-bit int support |
| - zink: support emitting 16-bit float types |
| - zink: perform fp16 texture-lookups as fp32 and then convert |
| - zink: enable 16-bit float support |
| - zink/codegen: prefer first definition of prop/feature structs |
| - zink: also enable float16 from KHR extension |
| - lavapipe: consistently use nir macros |
| - docs: update gallium doxygen docs |
| - zink: handle matrix-types after vectors |
| - zink: cache SpvId for aggregate glsl_types |
| - zink: always enable fixed shader-caps |
| - zink: do not check for varying output for fragment shaders |
| - zink: emit cap early |
| - zink: remove needless shader-info from context |
| - zink: emit sample-shading cap early |
| - zink: emit cap early |
| - zink: only emit ImageBuffer cap if needed |
| - docs: do not generate redirects on error |
| - gallium/u_vbuf: avoid dereferencing NULL pointer |
| - freedreno/a5xx: Remove ppgtt hack |
| - docs: remove doxygen support |
| - zink: remove incorrect border-swizzle assumption |
| - lavapipe: emit correct textures_used for texture-arrays |
| - zink: do not ask glsl-compiler to unroll |
| - lavapipe: fix fsum with swizzle |
| - st/mesa: do not take util_logbase2 of a negative size |
| - zink: check for error when binding memory |
| - gallium: allow to report errors from p_screen::resource_bind_backing |
| - lavapipe: report out-of-memory when binding |
| - llvmpipe: allow calculating size of overly large texture |
| - lavapipe: report allocation-error |
| - lavapipe: correct reported number of UBOs |
| - translate: reserve more vertex-shader outputs |
| - translate: assert that nr_elements is in range |
| - ci: Uprev piglit to 3351e8952 ("max-texture-size: report merged results") |
| - docs/features: document GL_ARB_ES3_2_compatibility support for zink |
| - docs/features: mark a few more extensions as done for zink |
| - zink: fix provoking-vertex cap for quads |
| - docs: promote #dri-devel on oftc over freenode |
| - docs: update link to #zink |
| - docs: update location of #panfrost |
| - docs: update link to #lima |
| - zink: simplify emit_load_const |
| - v3d: use helper to simplify things |
| - ci: downgrade sphinx to v3.x |
| - docs: update another IRC reference |
| - docs: update another IRC reference |
| - docs: drop clayton from intel-ci notice |
| - zink: use actual const for const offset |
| - lavapipe: handle cube-array image-views |
| - lavapipe: do not interpret cube-compatible as cubemap |
| - zink: only mark resources as cube-compatible if supported |
| - zink: mark 2d-arrays as cube-compatible |
| - zink: implement half-float packing |
| - zink: untangle have_EXT_debug_utils and ZINK_DEBUG_VALIDATION |
| - zink: add support for string-markers |
| - util/prim_restart: revert part of bad fix |
| - docs: quote a few defines |
| - docs: fix header-levels in envvars.rst |
| - docs: use file-role for paths |
| - docs: use envvar role for envvars |
| - docs: add the doc-comment for fse-vars |
| - docs: do not list all gles major versions |
| - docs: update list of apis to match website |
| - docs: update llvm requirement |
| - docs: rename vmware-guest article |
| - docs: clean up list of deprecated systems |
| - docs: move swrast to deprecated drivers list |
| - docs: clean up software-drivers list |
| - docs: clean up openswr links |
| - docs: split out layered driver to its own list |
| - docs: clean up freedreno links |
| - docs: add links to documented drivers |
| - r600: explicitly advertise index buffer format support |
| - zink: limit images we mark as cube-compatible |
| - zink: rename spirv_15 bool to spirv_1_4_interfaces |
| - zink: allow to specify any spir-v version to nir_to_spirv |
| - zink: calculate spir-v version based on vk version |
| - zink: only enable vote if we can support it |
| - zink: use a macro for spir-v versions |
| - st/pbo: use correct type for images and textures |
| - docs: update master -> main in edit-links |
| - zink/ci: increase piglit and deqp-runner timeouts |
| - llvmpipe: fix edge-rule logic for lines |
| - llvmpipe: consistently deal with post-rast state |
| - llvmpipe: fix multisample lines again |
| - llvmpipe: do not always use pixel-rounded coordinates for points |
| - zink/ci: re-enable test |
| - zink: reject more illegal blits |
| - zink: limit non-extension version feature to spirv 1.5 |
| - zink: use correct type for u_bit_scan |
| - zink: do not unmap dt-buffers twice |
| - zink: drop paranoid code |
| - zink: add missing compiler-dependency |
| - zink: drop some more vla usage |
| - zink: fix more initializer styles |
| - zink: introduce a define for max descriptors per type |
| - zink: use max-descriptor define |
| - zink: use alloca instead of hard-to-size vlas |
| - zink: correct type of flags to flush |
| - zink: fixup signedness of subtraction |
| - zink: remove unused function |
| - zink: drop repeated usage-bit |
| - zink: do not check buffer-format for usage-bits |
| - docs: remove outdated meson-section |
| - docs: remove outdated clarification |
| - docs: drop historic meson details |
| - docs: use more file-roles |
| - docs: use rst captions |
| - wgl: remove hard limit on pixelformats |
| - zink: drop unused macros |
| - zink: remove unused function-pointers |
| - zink: unbreak moltenvk code |
| - zink: remove unused moltenvk functions |
| - zink: do not store moltenvk functions in screen |
| - zink: remove some needless moltenvk details |
| - libgl-gdi: add missing include |
| - iris/ci: disable amly jobs |
| - aux/trace: fix bool argument |
| - zink: cast pointers to uintptr_t |
| - ci/windows: work around meson encoding issues |
| - ci/windows: enable msvc builds of zink |
| - ci/windows: fix zink msvc build-rules |
| - gallium/u_threaded: do not apply start twice |
| - ci: fix source-deps for radv on windows |
| - zink: hook up line-rasterization ext |
| - zink: use bit-allocation for boolean rasterizer-state |
| - zink: support line stippling |
| - zink: fill in the right line-mode based on state |
| - docs: update zink requirements |
| - llvmpipe: reject unsupported shader-image formats |
| - lavapipe: query formats for shader-image support |
| - llvmpipe: only report supported shader-image formats |
| - lavapipe: expose more storage-image features |
| - lavapipe: do not disable multisampling for smooth lines |
| - lavapipe: fix disable_multisample condition |
| - gallium: explicitly specify line rasterization mode |
| - draw: respect line_rectangular state |
| - llvmpipe: respect rectangular_lines |
| - lavapipe: re-expose line-rasterization extension |
| - lavapipe: expose strict-lines feature |
| - zink: implement support for non-planar DRM modifiers |
| - zink: remove duplicate format-mapping on little-endian |
| - vulkan: do not map zero-sized region of memory |
| - vulkan: allocate host-visible memory for swapchain images |
| - zink: check for right feature |
| - zink: respect line_rectangular state |
| - lavapipe: do not assert on more than 32 samplers |
| - lavapipe: do not mark unsupported tests as crashing |
| - d3d12: split up root parameter update and set |
| |
| Erik Kurzinger (1): |
| |
| - vulkan/device_select: avoid segfault on Wayland if wl_drm is unavailable |
| |
| Ernst Sjöstrand (1): |
| |
| - nv50: Fix use of initializers on older compilers |
| |
| Ezequiel Garcia (2): |
| |
| - panfrost: Add GPU IDs for G52 1-Core-2EE (RK3568/RK3566) |
| - panfrost: Rename G52 product ID 0x7402 as G52r1 |
| |
| Felix DeGrood (16): |
| |
| - intel: add L3 Bypass Disable to gen xml |
| - iris: Cache VB/IB in L3$ for Gen12 |
| - iris: reduce redundant tile cache flushes |
| - intel/blorp: remove tile flush from emit surface state |
| - intel/compiler: Use switch for DERIVATIVE_GROUP logic |
| - intel/compile: refactor DERIVATIVE_GROUP logic |
| - intel/compiler: tileY friendly LID order for CS |
| - intel/compiler: balanced tileY/linear friendly LID order for CS |
| - anv: Cache VB/IB in L3$ for Gfx12 |
| - anv: Add debug messages for DEBUG_PIPE_CONTROL |
| - anv: Clear all pending stall after pipe flush |
| - anv: Remove Tile Cache flush from SBA, Pipe Select |
| - anv: remove unnecessary Tile Cache flushes |
| - anv: Only flush Tile Cache on VK_ACCESS_HOST_R/W |
| - anv: Add ANV_PIPE_HDC_PIPELINE_FLUSH_BIT |
| - anv: Replace DC Flush with HDC Pipeline Flush |
| |
| Francisco Jerez (20): |
| |
| - intel/fs: Implement representation of SWSB cross-pipeline synchronization annotations. |
| - intel/fs: Add helper functions inferring sync and exec pipeline of an instruction. |
| - intel/fs: Represent SWSB in-order dependency addresses as vectors. |
| - intel/fs: Calculate SWSB cross-pipeline synchronization information. |
| - intel/fs: Use CHV/BXT implementation of 64-bit MOV_INDIRECT on XeHP+. |
| - intel/fs: Fix repclear assembly for XeHP+ regioning restrictions. |
| - intel/fs: Handle regioning restrictions of split FP/DP pipelines. |
| - intel/eu: Teach EU validator about FP/DP pipeline regioning restrictions. |
| - intel/compiler: Lower integer division on XeHP. |
| - intel/fs: Introduce lowering pass to implement derivatives in terms of quad swizzles. |
| - intel/fs: Add more efficient fragment coordinate calculation. |
| - iris/gen12: Work around push constant corruption on context switch. |
| - iris/gfx12: Invalidate ISP at the end of every batch. |
| - intel/fs/xehp: Assert that the compiler is sending all 3 coords for cubemaps. |
| - intel/fs: Track single accumulator in scoreboard lowering pass. |
| - intel/fs: Implement Wa_22012725308 for cross-pipe accumulator data hazard. |
| - intel/fs: Add SWSB dependency annotations for cross-pipeline WaR data hazards on XeHP+. |
| - intel/fs: Teach IR about EOT instruction writing the accumulator implicitly on TGL+. |
| - intel/fs: Fix synchronization of accumulator-clearing W/A move on TGL+. |
| - intel/fs: Implement Wa_14013745556 on TGL+. |
| |
| Georg Lehmann (11): |
| |
| - radv: Fix compatible image handle type for dmabufs. |
| - v3dv: use VKAPI_ATTR and VKAPI_CALL. |
| - zink: Add a missing VKAPI_ATTR. |
| - vulkan: Update the XML and headers to 1.2.180 |
| - radv: Implement VK_EXT_global_priority_query. |
| - ac: Check me_fw_feature for 32bit predication on gfx10.3 |
| - ac: Enable 32bit predication on gfx10. |
| - ac: Enable 32bit predication on gfx9 with fw feature version 52. |
| - lavapipe: Use common default allocator. |
| - lavapipe: Add a missing VKAPI_ATTR. |
| - vulkan/wsi/wayland: Add support for more SRGB formats. |
| |
| Gert Wollny (40): |
| |
| - Revert "r600: don't set an index_bias for indirect draw calls" |
| - Revert "r600: Don't advertise support for scaled int16 vertex formats" |
| - r600: don't set an index_bias for indirect draw calls |
| - virgl: use pipe_draw_info::restart_index only when primitive_restart is enabled |
| - r600: update pipe_draw_info::restart_index only when primitive_restart is enabled |
| - nir/opt_algebraic: optimizations for add umax/umin with zero |
| - nir: Add filter callback for lower_to_scalar to the options |
| - gallium: pass lower_to_scalar_filter to lower_to_scalar pass |
| - r600/sfn: lower to scalar with filter applied |
| - mesa: add an extension MESA_bgra |
| - compiler/nir: check whether var is an input in lower_fragcoord_wtrans |
| - nir/linker: add option to ignore the IO precisions for better varying packing |
| - r600/sfn: Ignore precision when linking |
| - r600: don't put INTERP_X and INTERP_Z into one instruction group |
| - r600/sfn: Use valid pixel mode only in fragment shaders |
| - r600/sfn: Use valid pixel mode for SSBO and Image result fetches |
| - r600/sfn: force new CF if fetch through TC would be used in same clause |
| - r600/sfn: Lower FS pos input w-transform in NIR |
| - r600/sfn: Don't check the faction when searching for the input slot |
| - r600/sfn: count only distinct literals per instruction group |
| - r600/sfn: Fix Cayman trans ops |
| - r600/sfn: Use unified index register code for samplers |
| - r600/sfn: Use unified code path for index register load |
| - r600/sfn: Fix texture gather for Cayman |
| - r600/sfn: Fix ssbo/image atomic swap for Cayman |
| - r600/sfn: Fix Cayman SSBO write with more than one value |
| - r600/sfn: Fix Geometry shader for Cayman |
| - r600/sfn: read number of images from shader info |
| - r600/sfn: Fix cube query layer number for indirect access |
| - r600/sfn: Add lowering pass to legalize image access |
| - r600/sfn: legalize image access on Cayman |
| - r600: Enable NIR debug flags also for Cayman |
| - r600/sfn: don't designates initializers, since they are c++20 |
| - r600/sfn: don't read back unused image atomic result values |
| - r600/sfn: Drop method for emit_atomic_add, it is handled in generic code |
| - r600/sfn: Don't read return values of atomic ops that are not used |
| - r600/sfn: Clean up some ALU lowering and move code |
| - r600/sfn: Lower offset in TXF instructions |
| - virgl: Enable ASTC formats also for 3D textures |
| - r600/sfn: initialize all texture lower options |
| |
| Gustavo Padovan (10): |
| |
| - traces-iris: fix expectation for Intel GLK |
| - gitlab-ci: enable Intel AML-Y as experimental |
| - gitlab-ci: rule anchor for experimental devices as manual in MRs |
| - gitlab-ci: enable all 3 intel devices as manual in MR pipelines |
| - iris/ci: disable failing gimark test for now |
| - iris/ci: enable intel devices automatically in MR pipelines |
| - gitlab-ci: add python script to submit lava jobs |
| - gitlab-ci: enable testing on Intel Kaby Lake as experimental |
| - ci/lava: propely report test failure through sys.exit() |
| - ci/lava: do not save lava.yaml in the artifacts |
| |
| Hans-Kristian Arntzen (2): |
| |
| - radv: Take image alignment into account when allocating MUTABLE pool. |
| - radv: Allocate buffer list for MUTABLE descriptor types as well. |
| |
| Heinrich Fink (6): |
| |
| - softpipe: add missing sentinel to debug option array |
| - llvmpipe: unmap display target of shader image/sampler |
| - softpipe: unmap display target of shader sampler |
| - llvmpipe: do not leak map of display target in fs setup |
| - llvmpipe: do not leak display target mapped ptr in cs setup |
| - gbm/dri: Fix leaking bo memory on failure path |
| |
| Hoe Hao Cheng (15): |
| |
| - vulkan/util: generate vk_dispatch_table that combines all dispatch tables |
| - nir: define NIR_ALU_MAX_INPUTS |
| - zink: remove variable length arrays in ntv |
| - zink: introduce vk_dispatch_table |
| - zink/codegen: split commands into three groups |
| - zink/codegen: add zink_verify_*_extensions() |
| - zink: slight refactor of load_device_extensions() |
| - zink: use the dispatch tables |
| - zink/codegen: allow conditional enabling of instance extensions |
| - zink/codegen: clean the constructor of Extension up |
| - zink: do not fail when EXT_calibrated_timestamps is unavailable |
| - zink: move extension function verification to when it is used |
| - zink: zero-init structs with ISO C |
| - zink: standardize zero-init code style |
| - zink: make codegen compatible with python 3.5 |
| |
| Hubert Jasudowicz (1): |
| |
| - docs/egl: Add missing backticks |
| |
| Hyunjun Ko (6): |
| |
| - turnip: prep work for timeline semaphore support |
| - turnip: Implement VK_KHR_timeline_semaphore. |
| - turnip/kgsl: Fix to build on android. |
| - turnip: add missing VKAPI_ATTR/CALL |
| - turnip: Copy command buffers to deferred submit request |
| - turnip/kgsl: new flag TU_USE_KGSL |
| |
| Iago Toral Quiroga (118): |
| |
| - v3dv: avoid redundant BO job additions for textures and samplers |
| - v3dv: avoid redundant BO job additions for UBO/SSBO |
| - v3dv: avoid redundant BO job additions for spill / shared BOs |
| - v3dv: optimize a few cases of BO job additions |
| - v3dv: use a bitfield to implement a quick check for job BO tracking |
| - v3dv: fix descriptor set limits |
| - v3dv: fix array sizes when tracking BOs during uniform setup |
| - v3dv: don't use a dedicated BO for each occlusion query |
| - v3dv: fix sRGB blending workaround |
| - v3dv: improve dirty descriptor set state tracking |
| - v3dv: dirty viewport doesn't affect fragment shaders |
| - v3dv: better tracking of dirty push constant state |
| - vulkan/wsi: give drivers the option to decide if they need to blit |
| - v3dv: implement wsi hook to decide if we can present directly on device |
| - compiler/nir: add a divergence analysis option for non-uniform workgroup id |
| - v3dv: choose a larger CSD supergroup size if possible |
| - broadcom/compiler: track if a shader has control barriers in prog_data |
| - v3dv: limit supergroup size in presence of TSY barriers |
| - broadcom/common: move CSD supergroup sizing to a common helper |
| - v3d: choose a larger CSD supergroup size if possible |
| - broadcom/compiler: add a loop unrolling pass |
| - v3dv: setup loop unrolling |
| - v3d: move NIR compiler options to GL driver |
| - broadcom/compiler: add a compiler strategy to disable loop unrolling |
| - broadcom/compiler: refactor compile strategies |
| - broadcom/compiler: specify maximum thread count in compile strategies |
| - v3d: enable NIR loop unrolling |
| - v3d: re-enable GLSL loop unrolling |
| - broadcom/compiler: change register allocation policy for accumulators |
| - broadcom/compiler: move vertex shader output handling to its own function |
| - broadcom/compiler: implement non-uniform offset on vertex outputs |
| - broadcom/compiler: make vir_VPM_WRITE_indirect handle non-uniform offsets |
| - broacom/compiler: enable PIPE_SHADER_CAP_INDIRECT_OUTPUT_ADDR |
| - broadcom/compiler: don't use nir_src_is_dynamically_uniform |
| - v3dv: don't lower indirect derefs on output variables |
| - broadcom/compiler: don't unroll due to indirect indexing of outputs |
| - v3d: disable GLSL loop unrolling again |
| - broadcom/compiler: clarify PIPE_SHADER_CAP_INDIRECT_INPUT_ADDR setting |
| - broadcom/compiler: don't emit TLB loads for components that don't exist |
| - broadcom/compiler: consider RT component size when lowering logic ops in Vulkan |
| - broadcom/ci: update fail list for v3dv |
| - v3d: take TLB blit framebuffer dimensions from smallest surface dimensions |
| - v3dv: implement VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_POINT_CLIPPING_PROPERTIES |
| - v3dv: fix texture_size() |
| - v3dv: allow creating uncompressed views from compressed images and vice versa |
| - v3dv: expose VK_KHR_maintenance2 |
| - v3dv: define V3D_MAX_BUFFER_RANGE |
| - v3dv: implement VK_KHR_maintenance3 |
| - v3dv: implement VK_KHR_bind_memory2 |
| - v3dv: implement VK_KHR_get_memory_requirements2 |
| - v3dv: keep track of whether an image may be backed by external memory |
| - v3dv: implement VK_KHR_dedicated_allocation |
| - v3dv: trivially handle VK_STRUCTURE_TYPE_EXPORT_MEMORY_ALLOCATE_INFO_KHR |
| - v3dv: add v3dv_GetImageSparseMemoryRequirements back |
| - v3dv: implement vkCmdDispatchBase |
| - v3dv: create a helper for image creation |
| - v3dv: implement interactions of VK_KHR_device_group with VK_KHR_swapchain |
| - v3dv: implement VK_KHR_device_group |
| - v3dv: don't keep an open file descriptor for imported fences/semaphores |
| - v3dv: implement external semaphore/fence extensions |
| - v3dv: increase number of supported SSBOs |
| - v3dv: expose KHR_relaxed_block_layout |
| - v3dv: document VK_KHR_relaxed_block_layout as implemented |
| - v3dv: expose VK_KHR_storage_buffer_storage_class |
| - v3dv: refactor descriptor updates |
| - v3dv: implement VK_KHR_descriptor_update_template |
| - v3dv: fix incorrect render area setup |
| - v3dv: expose KHR_variable_pointers |
| - v3dv: don't lower vulkan resource index result to scalar |
| - v3dv: implement VK_KHR_get_display_properties2 |
| - v3dv: handle Vulkan 1.1 feature and property queries |
| - v3dv: don't support VK_IMAGE_CREATE_BLOCK_TEXEL_VIEW_COMPATIBLE_BIT |
| - Revert "v3dv: allow creating uncompressed views from compressed images and vice versa" |
| - v3dv: expose VK_KHR_shader_non_semantic_info |
| - v3dv: implement VK_EXT_index_type_uint8 |
| - v3dv: implement vkCmdBlitImage2KHR |
| - v3dv: implement vkCmdCopyBuffer2KHR |
| - v3dv: implement vkCmdCopyBufferToImage2KHR and vkCmdCopyImageToBuffer2KHR |
| - v3dv: implement vkCmdCopyImage2KHR |
| - v3dv: implement vkCmdResolveImage2KHR |
| - v3dv: expose VK_KHR_copy_commands2 |
| - v3dv: remove const qualifier for resource pointer in view objects |
| - broadcom/compiler: implement nir_intrinsic_load_subgroup_id correctly |
| - broadcom/compiler: lower nir_intrinsic_load_num_subgroups |
| - broadcom/compiler: add FLAFIRST and FLNAFIRST opcodes |
| - broadcom/compiler: implement more subgroup intrinsics |
| - broadcom/compiler: add a ntq_emit_cond_to_bool helper |
| - broadcom/compiler: add a set_a_flags_for_subgroup helper |
| - broadcom/compiler: track if a compute shader uses subgroup functionality |
| - broadcom/util: don't use compute supergroup packing with subgroups |
| - v3dv: expose correct subgroup size |
| - v3dv: expose support for basic subgroup operations |
| - broadcom/compiler: use nir_sort_variables_with_modes |
| - v3dv: account for dst offset of copy query results operations |
| - v3dv: always free pipeline stages after compiling |
| - v3dv: extend broadcom stages to include geometry |
| - v3dv: define a generic helper to create binning pipeline stages |
| - v3dv: add a few more broadcom shader stage helpers |
| - broadcom/compiler: track if geometry shaders write gl_PointSize |
| - v3dv: add support for geometry shaders to pipelines |
| - broadcom/compiler: create a helper for computing VPM config |
| - v3dv: emit state packets for geometry shaders |
| - v3dv: handle QUNIFORM_FB_LAYERS |
| - v3dv: fix copy buffer to image TFU path for 3D images |
| - broadcom/compiler: handle compact input arrays for geometry shaders |
| - broadcom/compiler: don't ignore constant offset on per-vertex input loads |
| - v3dv: implement layered attachment clears |
| - v3dv: remove fallback path for vkCmdClearAttachments |
| - v3dv: remove deferred vkCmdClearAtachments path |
| - broadcom/ci: update expected fails for v3dv after enabling geometry shaders |
| - v3dv: expose geometry shaders |
| - v3dv: fix push constant range for texel buffer copy pipelines |
| - v3dv: implement layered texel buffer copies using a geometry shader |
| - v3dv: allow batching texel buffer copies for 3D images |
| - v3dv: use defines for push constant offsets used by texel buffer copy shaders |
| - v3d: better scissor tracking |
| - broadcom/compiler: implement gl_PrimitiveID in FS without a GS |
| - v3dv: remove more dead clearing code |
| |
| Ian Romanick (49): |
| |
| - tgsi_exec: Fix NaN behavior of saturate |
| - tgsi_exec: Fix NaN behavior of min and max |
| - ci: Uprev piglit to b3a9fa345 ("framework/replay: Quote resource names before signing") |
| - tgsi_exec: Use C99 functions for min and max instead of open coding |
| - gallivm: Fix NaN behavior of min and max |
| - gallivm: Use range analysis to generate better fmin and fmax code |
| - gallivm: Use GALLIVM_NAN_RETURN_OTHER_SECOND_NONNAN for norm clamping |
| - gallivm: Remove unused GALLIVM_NAN_RETURN_NAN |
| - nir/algebraic: Remove some optimizations of comparisons with fsat |
| - nir/algebraic: Tautology replacements require sources be numbers |
| - nir/algebraic: Invert comparisons less often |
| - nir/algebraic: Equality comparison inversions require sources be numbers |
| - nir/algebraic: Mark some more comparison reductions exact |
| - nir/algebraic: Mark some more logic-joined comparison reductions as exact |
| - nir/algebraic: Rearrange some logic-joined comparisons and reduce |
| - nir/algebraic: Add algebraic opt for float comparisons with identical operands. |
| - util/format: Delete trailing whitespace |
| - dri: Fix typo before __DRI_IMAGE_COMPONENTS defines |
| - egl/dri2: Rely on drm-uapi for DRM_FORMAT defines |
| - mesa/st: Always call st_nir_lower_tex_src_plane if samplers were lowered |
| - nir/lower_tex: Add support for lowering Y41x formats |
| - util/format: Add Y41x formats |
| - egl/dri2: Add Y41x formats |
| - gallium/dri: Add Y41x formats |
| - util/format: Add Y21x formats |
| - egl/dri2: Add Y21x formats |
| - gallium/dri: Add Y21x formats |
| - intel/isl: Add mappings for PIPE_FORMAT_R8G8_R8B8_UNORM and PIPE_FORMAT_G8R8_B8R8_UNORM |
| - mesa: Add R8G8_R8B8 and G8R8_B8R8 formats |
| - nir/lower_tex: Add support for lowering YUYV formats |
| - gallium/dri: Allow use of R8G8_R8B8 for YUYV and G8R8_B8R8 for UYVY |
| - iris: Return correct enum names from fmt_swizzle |
| - iris: Silence warnings about implicit enum type conversions |
| - mesa/st: Don't assert !unify_interfaces in the passthrough edge flags case |
| - radeonsi: Use util_cpu_caps to detect number of CPUs |
| - util: Zero out all of mask in util_set_thread_affinity |
| - util: Change order of PIPE_OS_UNIX code in util_cpu_detect_once |
| - util: Trivial cleanup in the BSD code of util_cpu_detect_once |
| - util: Fix setting nr_cpus on some BSD variants |
| - util: Set util_cpu_caps.num_cpu_mask_bits based on total CPUs in the system |
| - util: Use maximum number of CPUs for determining cache topology |
| - util: Consider CPU affinity when detecting number of CPUs |
| - v3d: ci: Add KHR-GLES31.core.shader_image_load_store.basic-glsl-earlyFragTests to flakes |
| - intel/compiler: Add the ability to defer IP updates in backend_instruction::remove |
| - intel/compiler: Add cfg_t::adjust_block_ips() method |
| - intel/compiler: Update block IPs once in dead_code_eliminate |
| - intel/compiler: Update block IPs once in register_coalesce |
| - intel/compiler: Update block IPs once in opt_cmod_propagation |
| - nir/gcm: Clear out pass_flags before starting |
| |
| Icecream95 (38): |
| |
| - panfrost: Assert staging resource allocation was successful |
| - panfrost: Unset shared/scanout binding flags for staging resources |
| - pan/bi: Skip nir_opt_move/sink for blend shaders |
| - panfrost: Fix shader texture count |
| - pan/decode: Allow frame shader DCDs to be in another BO than the FBD |
| - pan/decode: Print errors when closing dump file |
| - pan/mdg: Fix calculation of available work registers |
| - panfrost: Remove incorrect comment |
| - panfrost: Fix viewport scissor for preload draws |
| - panfrost: Split panfrost_batch_submit to prevent stack overflows |
| - pan/bi: Add "lane_dest" modifier |
| - pan/bi: Replace lane0 modifier with lane_dest for load instructions |
| - panfrost: Make pan_select_crc_rt a non-static function |
| - panfrost: Always write reloaded tiles when making CRC data valid |
| - pan/bi: Add two tuples to a clause when needed with NOSCHED |
| - panfrost: Skip blit shader labelling if the buffer has no space |
| - panfrost: Fix polygon list size computations |
| - pan/mdg: Try scheduling load/store ops in pairs |
| - pan/decode: Flush the dump stream after decoding |
| - panfrost: Call abort() when aborting on fault |
| - panfrost: Use first_tiler to check if tiling is needed |
| - pan/mdg: Add a bundle ID to instructions |
| - pan/mdg: Reorder some code in mir_spill_register |
| - pan/mdg: Fill from TLS before spilling non-SSA nodes |
| - pan/mdg: Fix reading a spilt register in the bundle it's written |
| - pan/mdg: Add 16 bytes of padding to the end of shaders |
| - panfrost: Don't set dirty_mask for constant buffers |
| - pan/bi: Create a mask of UBOs that need to be uploaded |
| - pan/mdg: Create a mask of UBOs that need to be uploaded |
| - panfrost: Only upload UBOs when needed |
| - panfrost: Set bound dimensions to framebuffer size |
| - Revert "panfrost: Fix crc_valid condition" |
| - panfrost: Always use a fragment shader when alpha test is enabled |
| - panfrost: Fix GPU ID for t76x in get_perf_config |
| - panfrost: Fix full_threads calculation on v6 |
| - pan/bi: Create a nop clause when the shader starts with ATEST |
| - panfrost: Initialise the blend equation in create_blend_state |
| - pan/mdg: Analyze helper termination after scheduling |
| |
| Ilia Mirkin (29): |
| |
| - nv50/ir: offset accesses to shared memory |
| - nv50/ir: refine limitation on load/store loading offsets, include atomics |
| - nv50/ir: "zero" register does not work with g[] memory |
| - nv50/ir: mark ATOM as having 3 arguments |
| - nv50/ir: wipe any info about memory when seeing a locking op |
| - nv50/ir: optimize shift of 0 bits |
| - nv50: pass surface/buffer parameters to shader via aux buffer |
| - nv50/ir: add surface op lowering |
| - nv50/ir: add lowering for shared atomics |
| - nv50: add compute invocations counter |
| - nv50: add remapping of buffers/images into unified space |
| - nv50: add support for doing membars |
| - nv50: add indirect compute support |
| - intel: fix MI builder for pre-gen7 |
| - nv50: fix streamout queries |
| - nvc0: fix 3d images |
| - vdpau: allow state tracker to report a lower number of macroblocks |
| - nouveau: improve video limit reporting |
| - st/mesa: avoid enabling image/buffer/compute extensions for weak hardware |
| - mesa: relax ES 3.1 compute shader requirements |
| - st/mesa: properly encode OES_geometry_shader requirement |
| - mesa/get: allow image/buffer/atomic variables to be fetched in es3.1 |
| - st/mesa: allow hardware to claim ES 3.1 without hw indirect draws |
| - nv50: expose images/buffers/compute |
| - nv50: expose GL ES 3.1 for nva3+ hardware |
| - mesa: always expose NV_image_formats and OES_shader_image_atomic |
| - mesa: also flush after compute dispatch when debug flag enabled |
| - nv50: use the no-mipmap texture type for 2d ms views |
| - st/mesa: always report the max samples as supported |
| |
| Ishi Tatsuyuki (1): |
| |
| - radv: ignore redundant variable descriptor counts (v2) |
| |
| Italo Nicola (28): |
| |
| - pan/mdg: fix midgard writemask encoding for stores |
| - util: add util_sign_extend |
| - pan/mdg: clean up redundant/unused variables in disassemble.c |
| - pan/mdg: rename dest_override to shrink_mode |
| - pan/mdg: improve outmod printing |
| - pan/mdg: refactor mir_pack_swizzle |
| - pan/mdg: add proper expand_mode enum |
| - pan/mdg: encode/decode expand_mode properly |
| - pan/mdg: add midgard_src_expand_mode validation |
| - pan/mdg: improve input modifier printing |
| - pan/mdg: improve swizzle decoding |
| - pan/mdg: fix/change ALU opcodes descriptions and add some missing ops |
| - pan/mdg: stop querying datatype by reading opcode name |
| - pan/mdg: print input data type for ALU opcodes |
| - pan/mdg: stop using size disambiguation suffixes |
| - pan/mdg: fix midgard.h indentation |
| - pan/mdg: improve mask decoding |
| - pan/mdg: remove register prefixes |
| - pan/mdg: print special alu arg outmods |
| - pan/mdg: misc cleanups |
| - pan/mdg: add helpers for load/store special read regs |
| - pan/mdg: improve ldst opcode names and add missing ops |
| - pan/mdg: print names of non-work registers |
| - pan/mdg: properly encode/decode ldst instructions |
| - pan/mdg: improve tex opcode decoding and add missing ops |
| - panfrost/ci: Improve coverage for T860 |
| - virgl: implement EXT_multisampled_render_to_texture |
| - panfrost: fix GL_EXT_multisampled_render_to_texture regression |
| |
| Iván Briano (2): |
| |
| - intel/nir: Fix txs for null surfaces |
| - anv: fix feature/property/sizes reported for fragment shading rate |
| |
| James Jones (18): |
| |
| - gbm: Remove stat and refcount fields from gbm_device |
| - gbm: Inline load_backend function content |
| - gbm: Create device directly in find_backend |
| - gbm: Consolidate env var and default backend loops |
| - gbm: Give getenv backend override its own function |
| - gbm: Give gbm_device a reference to its backend |
| - gbm: Add gbm_core struct to export code to backends |
| - gbm: Move majority of gbmint.h to gbm_backend_abi.h |
| - gbm: Version the GBM backend interface |
| - gbm: Add backend ABI-check test |
| - gbm: Rename backend description list to builtin_backends |
| - loader: Factor out driver library loading code |
| - meson: Add a GBM backends search path build option |
| - gbm: Rename the DRI backend from gbm_dri.so to dri |
| - gbm: Put common device creation in a helper function |
| - gbm: Support dynamically loading named backends |
| - gbm: Load backend based on DRM device driver name |
| - loader: Handle failure to load DRI driver library |
| |
| James Park (14): |
| |
| - meson: Fix winflexbison warnings |
| - ac/surface: Move drm_fourcc.h to common header |
| - radv: Use ac_drm_fourcc.h |
| - meson: Add wrap for libelf on Windows |
| - meson: Disable libdrm for RADV on Windows |
| - meson: Disable MSVC warning 5105 |
| - amd: Fix warnings around variable sizes |
| - radv: Add _WIN32 guard in radv_check_gpu_hangs |
| - radv: Fix unused label warning on Windows |
| - radv: Add <io.h> on WIndows for missing close() |
| - draw/clip: Use NAN to make MSVC happy |
| - llvmpipe: Remove stray ## operator for MSVC |
| - ci: Update Windows image to build RADV |
| - vulkan: Support 32-bit "weak" symbols on MSVC |
| |
| Jan Beich (1): |
| |
| - anv: adjust headers for non-GNU after e9e1e0362b6c |
| |
| Jason Ekstrand (139): |
| |
| - intel/compiler: Don't insert barriers for NULL sources |
| - anv: Use the same re-order mode for streamout as for GS |
| - vulkan: Update the XML and headers to 1.2.177 |
| - anv: Implement VK_EXT_provoking_vertex |
| - gallium: Add a transcode_astc driconf option |
| - intel/isl: There are seven aux states |
| - intel/isl: Fix isl_color_value_unpack to match the prototype |
| - intel/eu: SVB writes only happen on Gen6 |
| - intel/fs: Stop using brw_dp_read/write_desc in Gen7+ only code |
| - intel/eu: Set message subtype properly for SIMD8 FB fetch |
| - intel/fs: Don't use pixel_z for Gen4-5 source_depth_to_render_target |
| - intel/nir: Set lower txs with non-zero LOD |
| - nir/builder: Move clamp helpers to nir_builder.h |
| - anv: Check offset instead of alloc_size for freeing surface states |
| - anv: Allow storage on all formats that support typed writes |
| - anv: Plumb the shader into push constant helpers |
| - anv: Support pushing shader constants |
| - anv: Push at most 32 regs for vec4 shaders |
| - intel/vec4: Don't spill fp64 registers more than once |
| - intel/vec4: Add some asserts to move_push_to_pull |
| - intel/vec4: Update nr_params in pack_uniform_registers |
| - intel/vec4: Set up push ranges before we emit any code |
| - intel/vec4: Add support for masking pushed data |
| - intel/vec4: Add support for UBO pushing |
| - nir: Add a nir_instr_move helper |
| - nir/gather_info: Expose a nir_intrinsic_writes_external_memory helper |
| - nir: Add a discard optimization pass |
| - intel/fs: Handle non-perspective-correct interpolation on gen4-5 |
| - intel/nir,i965: Move HW generation check for UBO pushing to i965 |
| - intel/vec4: Also use MOV_FOR_SCRATCH for swizzle resolves |
| - intel/isl: Fix isl_format_is_valid |
| - intel/fs/ra: Fix payload node setup for SIMD16 on Gen4-5 |
| - ttn: Stop manually managing system_values_read |
| - anv: Require softpin on Gen8+ |
| - anv: Make use_softpin compile-time in genX code |
| - anv: Handle OOM in the pinned path in anv_reloc_list_add |
| - anv: Add a helper to add a BO to the batch list without a reloc |
| - anv: Make anv_batch_emit_reloc inline and optimize SKL+ |
| - anv: Fast-path surface relocs when we have softpin |
| - anv: Optimize anv_address_physical when ANV_ALWAYS_SOFTPIN |
| - anv/blorp: Optimize addresses/relocations when ANV_ALWAYS_SOFTPIN |
| - iris: Use isl_surf_get_image_surf instead of hand-rolling it |
| - iris: Move target_to_isl_surf_dim to iris_resource.c |
| - intel/isl: Add a isl_surf_get_image_offset_B_tile_el helper |
| - intel/blorp: Use isl_surf_get_image_offset_B_tile_el in ccs_ambiguate |
| - intel/isl: Make the offset helpers four dimensional |
| - intel/isl: Make tile logical extents four dimensional |
| - intel/isl: Use a 4D physical total extent for size calculations |
| - i965: Use nir_lower_passthrough_edgeflags |
| - anv: Agressively no-op Flush/InvalidateMappedMemoryRanges |
| - docs: Begin documenting ISL |
| - isl: Document more members of isl_surf |
| - docs/isl: Document ISL's units |
| - docs/isl: Add detailed documentation about isl formats |
| - docs/isl: Add detailed documentation about tiling on Intel GPUs |
| - docs/isl: Add detailed documentation about CCS compression |
| - util: Move the 4x4 matrix inverse function to u_math |
| - crocus: Drop extra_aux support |
| - nir,amd: Suffix nir_op_cube_face_coord/index with _amd |
| - nir,panfrost: Suffix fsat_signed and fclamp_pos with _mali |
| - nir,vc4: Suffix a bunch of unorm 4x8 opcodes _vc4 |
| - vulkan: Update the XML and headers to 1.2.182 |
| - nir: Require vectorized ALU ops to be all-or-nothing |
| - nir,docs: Add docs for NIR ALU instructions |
| - nir: Document all the ALU opcodes |
| - docs,isl: Document Sandy Bridge HiZ/stencil |
| - editorconfig: Use 3-space tabs for .rst |
| - docs/nir: Use 3-space tabs |
| - docs/isl: Consistently use 3-space tabs |
| - spirv: Create acceleration structure and shader record variables |
| - anv: Add minimal boilerplate for VK_KHR_acceleration_structure |
| - anv: Add stub support for acceleration structures |
| - anv: Add support for binding acceleration structures |
| - anv: Add minimal boilerplate for VK_KHR_ray_tracing_pipeline |
| - anv: Get ready for more pipeline stages |
| - anv: Add a ray-tracing pipeline object |
| - anv: Add support for binding ray-tracing pipelines |
| - anv,iris: Move the SHADER_RELOC enums to brw_compiler.h |
| - intel/compiler: Generalize shader relocations a bit |
| - intel/compiler: Add a U32 reloc type |
| - intel/fs: Add support for compiling bindless shaders with resume shaders |
| - intel/rt: Use reloc constants for the resume SBT |
| - anv: Disallow UBO pushing for bindless shaders |
| - nir/apply_pipeline_layout: Handle bindless shaders |
| - anv: Support fetching descriptor addresses from push constants |
| - anv: Compile ray-tracing shaders |
| - anv: Compile trivial return and trampoline shaders |
| - intel/fs: Don't pull CS push constants if uses_inline_data |
| - anv: Create and return ray-tracing pipeline SBT handles |
| - anv: Compute scratch sizes for ray-tracing pipelines and shader groups |
| - anv: Add support for vkCmdSetRayTracingPipelineStackSizeKHR |
| - anv: Allow _anv_combine_address with a NULL batch |
| - anv: Make anv_address::offset 64-bit |
| - anv: Implement vkCmdTraceRays and vkCmdTraceRaysIndirect |
| - isl: Assert some iris invariants in isl_surf_get_ccs_surf |
| - isl: Take a hiz_or_mcs_surf in isl_surf_supports_ccs |
| - isl,iris: Move the extra_aux_surf logic into iris |
| - isl,docs: Add a chapter on AUX state tracking |
| - docs/isl: Improve the bit[6] swizzling section of the tiling chapter |
| - include/drm-uapi: bump headers |
| - anv: Claim to be a discrete GPU if has_lmem |
| - util: Add an implementation of qsort_r for non-GNU platforms |
| - nir: Add a function for sorting variables |
| - intel/genxml: Add SURFTYPE_SCRATCH on GFX version 12.5 |
| - intel/isl: Add support for scratch buffers |
| - intel/fs: Implement spilling on XeHP |
| - intel/fs: Implement load/store_scratch on XeHP |
| - intel/genxml: Add new ScratchSpaceBuffer fields on GFX version 12.5 |
| - iris: Add a MEMZONE_BINDLESS and uploader |
| - iris: Add support for scratch on XeHP |
| - anv: Add support for scratch on XeHP |
| - intel/genxml: Remove old scratch fields on GFX version 12.5 |
| - iris/bufmgr: Stop changing mapping modes on buffers |
| - intel/devinfo: Add a has_lsc bit |
| - intel/compiler: Add LSC to messages brw_ir_performance |
| - intel/fs: Lower uniform pull constant load message to LSC dataport |
| - docs/isl/tiling: Fix swizzle pattern for X-tiling |
| - intel/isl: Pull the uncompressed surface view code from anv |
| - intel/blorp: Adjust the compressed copy rectangle before convert_to_single_slice |
| - intel/blorp: Use isl_surf_get_uncompressed_surf |
| - intel/isl: Add more cases to isl_surf_get_uncompressed_surf |
| - iris: Don't leak the surface if uncompressed re-interp fails |
| - iris: Use isl_surf_get_uncompressed_surf |
| - nir: Drop nir_ssa_def::name and nir_register::name |
| - android: Drop the Android.mk build system |
| - android: Restore android/Android.mk |
| - nir/lower_subgroups: Pad ballot values before bitcasting |
| - docs: Add docs for running a local Mesa build |
| - mailmap: Update for Emma's new e-mail address |
| - Convert a few files to UTF-8 |
| - mailmap: Add two more lines for Alyssa Rosenzweig |
| - glsl: Delete lower_texture_projection |
| - anv/allocator: Use list->u64 in free_list_push |
| - iris: Re-emit MEDIA_VFE_STATE for variable group size shaders |
| - anv: Handle errors properly in anv_i915_query |
| - intel: Pull anv_i915_query into common code |
| - anv: Use intel_i915_query_alloc for memory regions |
| - iris: Use intel_i915_query for meminfo |
| - nir/lower_tex: Rework invalid implicit LOD lowering |
| |
| Jeremy Huddleston (2): |
| |
| - libgl-xlib: Set darwin-versions |
| - libgl-xlib: Add missing dep_x11 dependency |
| |
| Jeremy Newton (1): |
| |
| - Update libva requirement |
| |
| Jesse Natalie (44): |
| |
| - microsoft/spirv_to_dxil: Lower samplers from deref to index |
| - microsoft/spirv_to_dxil: Lower loads/stores to DXIL |
| - microsoft/compiler: Support raw SRVs/UAVs through dxil_module_get_res_type |
| - microsoft/compiler: Support arrays of UBOs |
| - microsoft/compiler: Emit CBVs via variables for Vulkan |
| - microsoft/compiler: Emit SSBO variables |
| - microsoft/compiler: Split Vulkan resource_index / descriptor processing |
| - microsoft/compiler: Better support UBO/SSBO references to descriptors |
| - microsoft/compiler: Store nir_shader in the ntd_context |
| - microsoft/compiler: Support raw SRVs in addition to typed SRVs |
| - microsoft/compiler: Propagate access when lowering SSBO loads |
| - microsoft/clc: If local size isn't specified either in the shader or at runtime, set it to (1,1,1) |
| - gallium: Define PIPE_ARCH_AARCH64 for MSVC arm64 builds |
| - nir: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - d3d12: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - microsoft/clc: Fix MSVC unreferenced variable warnings |
| - microsoft/clc: Fix undeclared function warning |
| - microsoft/compiler: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - shader_enums: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - gallium/aux: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - llvmpipe: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - xmlconfig: Fix MSVC warning C4334 (32bit shift cast to 64bit) |
| - CI: Windows: Bump warning level to W3 (except for zlib) |
| - microsoft/compiler: Remove hardcoded limits on numbers of resource arrays |
| - microsoft/compiler: Remove assert-only resource size or usage tracking |
| - microsoft/compiler: For Vulkan environment, don't create resource handles upfront |
| - vtn: Propagate access data that's present on all struct members to the struct itself |
| - vtn: Propagate access data from UBO/SSBO/push constant types to variables of that type, not just their pointers |
| - nir: Rename nir_lower_cl_images_to_tex, replace 'cl' with 'readonly' |
| - nir_lower_readonly_images_to_tex: Support non-CL semantics |
| - nir_lower_readonly_images_to_tex: Use nir_shader_lower_instructions |
| - microsoft/compiler: Treat read-only SSBOs as SRVs |
| - microsoft/spirv_to_dxil: Treat read-only storage images as SRVs |
| - d3d12, microsoft/compiler: Use SRV/sampler variable binding data |
| - microsoft/compiler: Rewrite sampler splitting pass to be smarter and handle derefs |
| - microsoft/compiler: Fix function signature for bufferStore to support overloads |
| - microsoft/compiler: Map descriptor set -> binding space |
| - microsoft/compiler: Handle unbounded arrays |
| - llvmpipe: Fix optimization loop to actually loop |
| - nir: Add relaxed 24bit opcodes |
| - vtn: Use relaxed 24bit opcodes for CL 24bit math |
| - microsoft/compiler: Change behavior for emitting inexpressible barriers |
| - nir_lower_readonly_images: Clear variable data when changing the type |
| - mesa/main: Check for fbo attachments when importing EGL images to textures |
| |
| John Bates (1): |
| |
| - add execmem build option |
| |
| Jonathan Marek (4): |
| |
| - freedreno/registers: define REG_DSI_CPHY_MODE_CTRL |
| - tu: remove workaround for conditional rendering + hw binning |
| - freedreno/a6xx: larger gmem_page_align from tile align instead of gpu id |
| - freedreno/common: unhardcode CCU color cache offset |
| |
| Jordan Justen (20): |
| |
| - Revert "intel/compiler: Silence unused parameter warning in update_inst_scoreboard" |
| - intel/eu: Allow 64-bit registers on XeHP. |
| - intel/fs: Disable 3-src immediates on XeHP. |
| - intel/fs: End computer shader with message gateway on XeHP. |
| - intel/compiler: Lower txd for 3D samplers on XeHP. |
| - intel/compiler: Fix INTEL_DEBUG=hex |
| - commit_in_branch_test.py: Rename branch master to main |
| - bin/pick: Rename master branch to main |
| - .gitlab-ci.yml: Use main branch for gitlab ci |
| - issue_templates/Bug Report: Rename master branch to main |
| - docs/releasing.rst: Rename master branch to main |
| - docs: Rename master branch to main |
| - mesa: NOTE! Default branch is now main |
| - intel/isl: Add Wa_22011186057 to disable CCS on ADL GT2 A0 |
| - intel/dev: Add device info for ADL GT2 |
| - intel: Add 2 ADL-S pci-ids |
| - intel/gen125.xml: Drop GPGPU_WALKER |
| - intel/devinfo: Add has_local_mem |
| - iris/bufmgr: Align vma addresses to 64K for local memory |
| - intel/dev: Set has_local_mem for DG1 |
| |
| Jose Maria Casanova Crespo (5): |
| |
| - v3d: YUV formats at is_dmabuf_modifier_supported are external_only |
| - v3d: YUV formats at query_dmabuf_modifiers are external_only |
| - v3d: DRM_FORMAT_MOD_BROADCOM_SAND128 only available for NV12 format. |
| - ci/v3d: Update piglit expectations. |
| - v3d/driconf: Expose non-MSAA texture limits for mutter and gnome-shell |
| |
| Joshua Ashton (5): |
| |
| - radv: Handle unnormalized samplers in YCbCr lowering |
| - venus: Fix zero-initialized fd causing apps to hang/crash |
| - driconf: Add more workarounds for Teardown |
| - llvmpipe: Handle NULL views in llvmpipe_cleanup_stage_sampling |
| - lavapipe: Use common Vulkan format helpers |
| |
| Joshua Watt (1): |
| |
| - v3d, vc4: Fix dmabuf import for non-scanout buffers |
| |
| José Fonseca (8): |
| |
| - lavapipe: Fix lvp_execute_cmds' pipe_stream_output_target leak. |
| - lavapipe: Fix lvp_pipeline_compile's nir_xfb_info leak. |
| - wgl: Remove opengl32.mingw.def. |
| - draw: Allocate extra padding for extra shader outputs. |
| - draw: Plug leak when combining tessellation with primitive assembly. |
| - d3d10umd,d3d10sw: Initial import. |
| - d3d10sw: Add a sanity test. |
| - d3d10umd: Avoid duplication in CreateDevice. |
| |
| Juan A. Suarez Romero (42): |
| |
| - v3dv: avoid dereferencing null value |
| - ci: support KHR-GLES testing |
| - ci/v3d: add KHR-GLES test jobs |
| - ci/llvmpipe: run KHR-GLES2.* tests |
| - ci/softpipe: run KHR-GLESxx tests |
| - iris: hook up memory object creation from handle |
| - iris: hook up resource creation from memory object |
| - iris: enable GL_EXT_memory_object feature |
| - ci/broadcom: update expected results |
| - ci/vc4: add KHR-GLES2.* job test |
| - ci/broadcom: add EGL testing jobs |
| - v3dv: check returned values |
| - ci/v3d: execute all piglit tests |
| - v3dv/pipeline_cache: bail out in case of error |
| - ci/v3d: fix typo in job name |
| - ci/v3dv: update flakes |
| - ci/baremetal: propagate ASAN_OPTIONS to devices |
| - ci/broadcom: update expected results |
| - v3d: rename header include guards |
| - v3d: rename VC5 enums and definitions |
| - broadcom/qpu: rename from VC5 to V3D |
| - broadcom/simulator: change references to VC5 |
| - v3dv: rename VC5 to V3D |
| - v3dv: check dest bitsize in color blit |
| - util/hash_table: do not leak u64 struct key |
| - ci/broadcom: update expected results |
| - v3d: fix resource leak in error path |
| - st/mesa: fix pipe resource leak |
| - broadcom/compiler: fix dynamic-stack-buffer-overflow error |
| - ci: Update VK-GL-CTS to 1.2.6.1 |
| - ci/broadcom: update expected results |
| - vc4: initialize array |
| - ci/v3dv: update expected results |
| - ci/broadcom: unset manual jobs |
| - ci/v3dv: test v3dv in arm64 environment |
| - broadcom/ci: Report flakes on IRC |
| - ci/vc4: update piglit failures |
| - ci: update VK-GL-CTS to 1.2.6.2 |
| - broadcom/compiler: emit TMU flush before a jump |
| - v3dv: assert job->cmd_buffer is valid |
| - broadcom: remove v3dv3 from neon library |
| - gallium/hud: initialize query |
| |
| Kai-Heng Feng (1): |
| |
| - iris: Avoid abort() if kernel can't allocate memory |
| |
| Karol Herbst (18): |
| |
| - clover/llvm: handle Fixed vs Scalable vectors explicitly starting with llvm-11 |
| - util/format: fix value declarations for big endian |
| - nv50/query: fix stringop-overflow gcc warning |
| - nvc0: fix implicit-fallthrough gcc warning |
| - clover/memory: fix data race in buffer subclasses |
| - nouveau: fix race in nouveau_screen_get_name |
| - nouveau/mm: pass mm_bucket to mm_slab_new |
| - nouveau/mm: remove unused nouveau_mm_allocation.next field |
| - nv50/ir: when constant folding shl(mul, a) we need to copy muls type |
| - nv50/ir: don't optimize shl(mul_hi, a) to mul_hi |
| - nv50/ir/ra: fixes upcoming barrier file |
| - nv50/ir: add barrier and thread_state files |
| - gv100/ir: add support for barrier thread state files for OP_CVT |
| - gm107/ir: emit barrier sources for quadon/pop |
| - gv100/ir: fix quadop/pop lowering |
| - nv50/ir: fix surface lowering when values get shared accross operations |
| - nv50/ir/nir: fix smem size for GL |
| - nv30: fix emulated vertex index buffers |
| |
| Keith Packard (1): |
| |
| - vulkan/x11: Mark present complete using serial instead of MSC |
| |
| Kenneth Graunke (29): |
| |
| - iris: only flush the render cache for aux changes, not format changes |
| - isl: Work around NVIDIA and AMD display pitch requirements |
| - i965: Don't advertise Y-tiled modifiers for scanout buffers on Gfx8- |
| - iris: Don't advertise Y-tiled modifiers for scanout buffers on Gfx8 |
| - iris: Replace no_gpu flag with PIPE_MAP_DIRECTLY |
| - iris: Promote to MAP_DIRECTLY when required before NULL return |
| - iris: Delete a comment suggesting we use tiled staging buffers |
| - iris: Make an iris_bo_is_external() helper and use it in a few places |
| - iris: Track imported vs. exported status separately |
| - iris: Use staging blits for reads from uncached buffers. |
| - iris: Use staging blits for transfers involving imported BOs |
| - iris: Assert on mapping a tiled buffer without MAP_RAW |
| - iris: Drop fallback GEM_MMAP_GTT if GEM_MMAP with I915_MMAP_WC fails |
| - iris: Delete GTT mapping support |
| - iris: Pick a single mmap mode (WB/WC) at BO allocation time |
| - iris: Use bo->mmap_mode in transfer map read check |
| - iris: Add a flags argument to iris_bo_alloc() |
| - iris: Add an alignment parameter to iris_bo_alloc() |
| - iris: Only use SET/GET_TILING when exporting/importing BOs |
| - iris: Add a BO_ALLOC_SMEM flag for allocating from system memory |
| - anv: Fix dynamic primitive topology for tess on Gfx7.x too |
| - iris: Stop calling I915_GEM_SET_CACHING on discrete GPUs |
| - iris: Fail BO allocation if we can't enable snooping properly. |
| - iris: Delete unused bo->cache_coherent flag |
| - vulkan/wsi: Fix prime blits to use system memory for the destination |
| - iris: Reduce SSBO alignment requirements from 64B to 4B |
| - crocus: Reduce SSBO alignment requirements from 64B to 4B. |
| - iris: Force device local memory for u_upload_mgr buffers |
| - iris: Use simple_mtx in the bufmgr. |
| |
| Leo Liu (9): |
| |
| - frontends/va: add VASurfaceAttribUsageHint attribute |
| - frontends/va: fix multi planes for external memeory type |
| - frontends/va: use pipe buffer map instead of texture map |
| - radeon/vcn/enc: use surface swizzle mode instead of linear |
| - radeonsi: add PIPE_FORMAT_P010 for HEVC Main10 profile to encode param |
| - radeonsi: separate video hw info based on HW engine individually |
| - frontends/va: use the correct entrypoint to get config attributes |
| - frontends/va: include the profile queries for encoder as well |
| - frontends/va: use the entrypoint from context instead of the hard-coded one |
| |
| Lepton Wu (3): |
| |
| - virgl: move new added field to the end. |
| - Revert "virgl: Cache depth and stencil buffers" |
| - gallium: Reset {d,r}Priv in dri_unbind_context |
| |
| Lionel Landwerlin (59): |
| |
| - anv: fix 3DSTATE_MULTISAMPLE emission on gen8+ |
| - anv: disable baked in pipeline bits from dynamic emission path |
| - vulkan/util: cast enums to int64_t in switch |
| - spirv: fix uToAccelerationStructure handling |
| - spirv: fixup pointer_to/from_ssa with acceleration structures |
| - vulkan: bump headers/registry to version 1.2.175 |
| - anv: drop extension check for dynamic state |
| - anv: prepare pipeline for delayed emission of color writes |
| - anv: implement VK_EXT_color_write_enable |
| - anv: reuse define for number of render target assert |
| - vulkan/wsi/display: don't report support if there is no drm fd |
| - i965/bufmgr: fix invalid assertion |
| - intel/dev: printout correct subslice/dualsubslice name |
| - intel/genxml: Add coarse pixel shading instructions |
| - intel/decoder: decode CPS_STATE |
| - intel/compiler: make sure we keep the lowest dispatch limit |
| - intel/compiler: rework message descriptors for render targets |
| - intel/compiler: use existing helpers to pull bits of descriptors |
| - intel/compiler: handle coarse pixel in render target writes descriptors |
| - intel/compiler: add support for fragment shading rate variable |
| - intel/compiler: add support for fragment coordinate with coarse pixels |
| - intel/compiler: add coarse pixel offset on Gfx12.5+ |
| - intel/compiler: add restrictions related to coarse pixel shading |
| - anv: implement VK_KHR_fragment_shading_rate |
| - isl: document format fields |
| - intel/fs: use the final destination type for regioning restrictions |
| - intel/mi_builder: fix resolve call |
| - anv: fix perf query pass with command buffer batching |
| - anv: handle spirv parsing failure |
| - iris: fix assert to reflect correct limit for encoded size |
| - intel/perf: allow opening perf stream with no context filtering |
| - intel/perf: allow metric sets to be loaded with on OA reports |
| - anv: fixup physical device properties of fragment shading rate |
| - intel/fs: make sure shuffle is lowered to supported types |
| - intel/perf: update gen9/11 TestOa configs |
| - intel/perf: update Gen11 RenderBasic programming |
| - intel/perf: update Gen11 RenderBasic programming |
| - intel/perf: add EHL availability condition to HDCAndSF counters |
| - intel/perf: update Gen9/11 programming for AsyncCompute |
| - intel/perf: rename metric descriptions |
| - anv: implement VK_EXT_physical_device_drm |
| - blorp: add blorp string in shader keys |
| - anv: cache raytracing trampoline shader |
| - anv: store more RT shader data in pipeline_stage object |
| - anv: move trivial return shader to device |
| - anv: implement caching for ray tracing pipelines |
| - intel/rt: switch to common pass for shader calls lowering |
| - nir: drop the btd_resume_intel intrinsic |
| - nir: use a more fitting index for btd_stack_push_intel |
| - anv: bound checks buffer memory binding in debug builds |
| - anv: allocate bigger batches as we grow command buffers |
| - intel/perf: use the right popcount for 64bits |
| - intel/compiler: Track latency/perf of LSC fences |
| - isl: fix mapping of format->stringname |
| - loader/dri3: create linear buffer with scanout support |
| - nir/lower_shader_calls: adding missing stack offset alignment |
| - anv: fix submission batching with perf queries |
| - drm-shim: implement stat/fstat when xstat variants are not there |
| - intel/disasm: fix missing oword index decoding |
| |
| Lucas Stach (9): |
| |
| - etnaviv: fix vertex sampler setup |
| - dri: add loader_dri_create_image helper |
| - loader/dri3: convert to loader_dri_create_image |
| - loader/dri: hook up createImageWithModifiers2 |
| - gallium/dri: copy image use in dup_image |
| - dri: don't call modifier interfaces when modifiers_count is 0 |
| - frontend/dri: add EXPLICIT_FLUSH hint in dri2_resource_get_param |
| - etnaviv: remove double assigment of surface->texture |
| - etnaviv: flush used render buffers on context flush when neccessary |
| |
| Luis Felipe Strano Moraes (2): |
| |
| - meson: print information about layers being built as part of summary |
| - overlay_layer: add missing undef |
| |
| Maksim Sisov (2): |
| |
| - iris: export GEM handle with RDWR access rights |
| - i965: export GEM handle with RDWR access rights |
| |
| Marcin Ślusarz (28): |
| |
| - intel/tools: remove unused macros |
| - intel/batch_decoder: set foreground color of decoded instructions |
| - i965: fully populate perf_config before using it to initialize perf_context |
| - iris: fully populate perf_config before using it to initialize perf_context |
| - intel/perf: move calculation of period_exponent to perf ctx init |
| - gallium/u_threaded: implement INTEL_performance_query hooks |
| - gallium/u_threaded: offload begin/end_intel_perf_query |
| - nir: handle float atomics in nir_gather_info |
| - nir: handle float atomics in nir_lower_memory_model |
| - intel: simplify is_haswell checks, part 1 |
| - intel: simplify is_haswell checks, part 2 |
| - i965: simplify gfx version checks |
| - intel/isl: replace format_gen by verx10 |
| - intel/disasm: decode/describe more send messages |
| - intel/disasm: remove useless space after "(" |
| - iris: fix error message on I915_GEM_[GS]ET_TILING failure |
| - intel/decoder: add assert for register size |
| - anv: fix potential integer overflows |
| - intel/tools: fix left shift overflow on 32-bit |
| - intel/tools: fix int-to-pointer/pointer-to-int cast warnings on 32-bit |
| - intel/tools: fix invalid type in argument to printf format specifier |
| - intel/tools: fix potential memory leaks |
| - intel/blorp: initialize BLEND_STATE using braced initializer list |
| - intel/fs: use stack for temporary array |
| - anv: keep descriptor set's address directly in anv_descriptor_set |
| - anv: handle push descriptor sets when they are sent with push constants |
| - anv: drop unused argument of anv_descriptor_set_address |
| - intel/compiler: document register types |
| |
| Marek Olšák (190): |
| |
| - ci: don't build clover with LLVM 9 on radeonsi because it's unsupported |
| - amd: drop support for LLVM 9 |
| - amd: drop support for LLVM 10 |
| - amd: remove some references to older LLVM versions in comments |
| - amd/registers: fix the kernel header parser with latest headers |
| - amd/registers: clean up gfx103.json |
| - amd/registers: rename IMG_FORMAT to GFX10_FORMAT to disambiguate the meaning |
| - radeonsi: don't decompress DCC for float formats in si_compute_copy_image |
| - radeonsi: fix incorrect comments in culling code and NIR lowering |
| - radeonsi: fix automatic DCC retiling after DCC clear and DCC decompression |
| - radeonsi: fix automatic DCC retiling after compute image stores |
| - gallium/util: add easy profiling helpers using TIME_ELAPSED queries |
| - Revert "st/pbo: use cso_set_vertex_buffers_and_elements() for st_pbo_draw" |
| - Revert "ci/radeonsi: Add expected failures due to #4674 having slipped in" |
| - ac/surface: document more meta equation dependencies |
| - radeonsi: make the gfx9 DCC MSAA clear shader depend on the number of samples |
| - radeonsi: remove the separate DCC optimization for Stoney |
| - amd: addrlib update for April |
| - gallium: renumber PIPE_MAP_* enums to remove holes |
| - gallium: remove 4 bytes from pipe_transfer |
| - gallium+(u_threaded,r300,r600,radeonsi): move transfer offset into pipe_transfer |
| - util: print CPU caps in release builds too |
| - util: fix (re-enable) L3 cache pinning |
| - Revert "gallium/u_threaded: align batches and call slots to 16 bytes" |
| - gallium/u_threaded: move base_valid_buffer_range to transfer where it belongs |
| - gallium/u_threaded: handle sampler views == NULL better |
| - gallium/u_threaded: rewrite slot layout to reduce wasted space |
| - gallium/u_threaded: don't set resource pointers to NULL after driver calls |
| - gallium/u_threaded: fix 32-bit breakage due to incorrect pointer arithmetic |
| - gallium/u_threaded: pass last into and return call size from execute callbacks |
| - gallium/u_threaded: merge draws in tc_call_draw_single |
| - gallium/u_threaded: add callbacks and documentation for resource busy checking |
| - gallium/u_threaded: track whether TCS, TES, or GS have ever been used |
| - gallium/u_threaded: query shader resource limits |
| - gallium/u_threaded: add buffer lists - tracking of buffers referenced by tc |
| - gallium/u_threaded: add driver-internal flush tracking for buffer lists |
| - gallium/u_threaded: don't invalidate idle buffers |
| - gallium/u_threaded: always map idle buffers unsynchronized |
| - gallium/u_threaded: don't reference resource in pipe_transfer |
| - util: add thread-safe version of idalloc |
| - zink: don't set u_resource_vtbl |
| - gallium: remove unused u_default_transfer_unmap |
| - gallium: remove resource_get_handle from u_resource_vtbl |
| - gallium: remove transfer_flush_region from u_resource_vtbl |
| - radeonsi: stop using u_resource_vtbl::resource_destroy |
| - r300: stop using u_resource_vtbl::resource_destroy |
| - nouveau: stop using u_resource_vtbl::resource_destroy |
| - i915g: stop using u_resource_vtbl::resource_destroy |
| - virgl: stop using u_resource_vtbl::resource_destroy |
| - svga: stop using u_resource_vtbl::resource_destroy |
| - r600: stop using u_resource_vtbl::resource_destroy |
| - gallium: remove u_resource_vtbl::resource_destroy |
| - gallium: split transfer_(un)map into buffer_(un)map and texture_(un)map |
| - gallium: remove u_resource_vtbl::transfer_(un)map |
| - gallium: remove empty structure u_resource_vtbl |
| - gallium: remove structure u_resource |
| - radeonsi: simplify the NGG culling vertex count heuristic |
| - amd: add Beige Goby support |
| - amd/registers: don't generate 32-bit register fields |
| - amd/registers: regenerate json files without 32-bit register fields |
| - amd: fix incorrect addrlib comment for HTILE equations |
| - ac/gpu_info: set has_zero_index_buffer_bug for Navi12 too |
| - ac/llvm: set target features per function instead of per target machine |
| - ac/llvm: expose set_range_metadata to more users |
| - ac/llvm: allow ac_build_optimization_barrier with SGPRs, pointers, and metadata |
| - ac/llvm: set range metadata on mbcnt and deduplicate get_thread_id |
| - ac/llvm: don't draw the primitive for the dummy export workaround for Navi1x |
| - winsys/amdgpu: don't hold a mutex while accessing is_shared |
| - radeonsi: remove unused SI_IMAGE_ACCESS_AS_BUFFER |
| - radeonsi: handle PIPE_CAP_MAX_VERTEX_BUFFERS |
| - radeonsi: add a gfx10 bug workaround for NOT_EOP |
| - radeonsi: fix a coherency issue when VS memory stores are not visible in PS |
| - radeonsi: always use the L2 LRU cache policy for faster clears and copies |
| - radeonsi: don't disable L2 caching for staging textures |
| - radeonsi: don't use GS fast launch with small instances |
| - radeonsi: fix the fast launch vert/prim thread counts if they are trimmed |
| - radeonsi: remove a twice duplicated workaround for VERT_GRP_SIZE |
| - radeonsi: re-enable fast launch with indexed tri strips because it doesn't hang |
| - radeonsi: improve generated culling code by adding optimization barriers |
| - radeonsi: change si_resource::alignment to alignment_log2 for better packing |
| - radeonsi: remove 8 bytes from si_resource, turn other 4 bytes into padding |
| - radeonsi: add a gfx10 hw bug workaround with the barrier before gs_alloc_req |
| - radeonsi: add missing threaded_resource_deinit calls in fail paths |
| - radeonsi: rewrite the prefix sum computation for shader culling |
| - radeonsi: allow changing the NGG subgroup size to 256 but don't change it yet |
| - radeonsi: generate buffer_id_unique for u_threaded_context |
| - radeonsi: implement threaded context callbacks for resource busy checking |
| - radeonsi: disable DFSM on gfx9 by default because it decreases performance a lot |
| - radeonsi: remove DFSM after we discovered how bad it is |
| - gallium/u_vbuf: add a fast path to skip refcounting for uploaded user buffers |
| - mesa: move _mesa_copy_vertex_attrib/buffer functions to their only use |
| - mesa: don't call _mesa_set_draw_vao in glPushClientAttrib |
| - mesa: optimize glPush/PopClientAttrib for GL_CLIENT_VERTEX_ARRAY_BIT |
| - mesa: optimize unreferencing VBOs in glPopClientAttrib |
| - mesa: don't call FLUSH_VERTICES in glPopClientAttrib |
| - mesa: don't save/restore VAO NumUpdates and IsDynamic to fix update tracking |
| - st/mesa: execute glFlush asynchronously if no image has been imported/exported |
| - gallium/u_threaded: don't update valid_buffer_range for read-only shader buffers |
| - gallium/u_threaded: clear valid buffer range only if it's not bound for write |
| - gallium/u_threaded: use tc_drop_resource_reference in call_draw_single_drawid |
| - gallium/u_threaded: merge draws faster by merging indexbuf unreferencing |
| - radeonsi: check is_buffer once instead of 4 times in si_set_sampler_view_desc |
| - radeonsi: use the restrict keyword to set sampler view descriptors faster |
| - radeonsi: don't clear register fields in si_set_mutable_tex_desc_fields |
| - radeonsi: restructure si_set_sampler_views for faster unbinding trailing slots |
| - radeonsi: remove no-op unref in si_set_constant_buffer |
| - radeonsi: set desc[3] of all buffer descriptors at context creation |
| - radeonsi: move a few functions from si_state_draw.cpp into si_gfx_cs.c |
| - radeonsi: compile si_state_draw.cpp for each gfx generation separately |
| - radeonsi: remove the chip_class dimension from the draw_vbo array |
| - radeonsi: remove -Wstrict-overflow=0 since it doesn't seem to be needed |
| - gallium/pb: change alignment to 32 bits |
| - shader_enums: change VERT_BIT back to the 32-bit shift |
| - glthread: change when glFlush flushes asynchronously |
| - st/mesa: fix an incorrect comment in st_context_flush |
| - st/mesa: move the st_flush_bitmap_cache call into st_flush |
| - mesa: add gallium flush_flags param into ctx->Driver.Flush |
| - mesa: move _mesa_notifySwapBuffers into the x11 swrast driver |
| - mesa: execute glFlush asynchronously if no image has been imported/exported |
| - radeonsi: fix compile failures with SI_PRIM_DISCARD_DEBUG enabled |
| - radeonsi: use ac_build_bit_count instead of opencoding it |
| - radeonsi: fix incorrect counting of compute_num_verts_rejected |
| - radeonsi: fix multi draws for the prim discard CS |
| - ac/llvm: add a callback to ac_cull_triangle to generate code in inner-most block |
| - radeonsi: move the accepting code into the bbox cull branch in NGG cull code |
| - ac/surface/tests: fix RB counts |
| - ac/surface: don't set DCC_PIPE_ALIGN modifier bit for gfx10 with 1 RB |
| - radeonsi: restructure si_get_vs_vgpr_comp_cnt for readability |
| - radeonsi: merge 2 conditional blocks with same condition into 1 in culling code |
| - radeonsi: set more precise max_waves in NGG code |
| - radeonsi: remove incorrect comment about PA |
| - radeonsi: try to keep all VS input loads together for better perf |
| - radeonsi: don't compile TES and GS draw_vbo variants for the prim discard CS |
| - radeonsi: remove the Z culling option from the primitive discard CS |
| - radeonsi: drop gfx7 support from the prim discard CS to simplify code |
| - radeonsi: drop support for triangle fans from the prim discard CS |
| - radeonsi: skip buffer_atomic_add(ptr, n) when n=0 in the prim discard CS |
| - radeonsi: cleanup some primitive discard CS TODOs regarding instancing, etc. |
| - ac/llvm: don't set skip-uniform-regions to fix atomic.cmpswap |
| - mesa: unreference zombie buffers when creating buffers to lower memory usage |
| - radeonsi: document why VBO descriptors in user SGPRs are beneficial |
| - radeonsi: if shader culling culls all vertices, cull the primitive exports too |
| - radeonsi: remove incorrect comment about hangs in gfx10_ngg_gs_emit_epilogue |
| - radeonsi: don't use NGG culling on 1 RB chips |
| - ac/gpu_info: adjust the condition for use_late_alloc |
| - radeonsi: optimize set_inlinable_constants when they don't change |
| - st/mesa: don't track VS sampler views for st_draw_feedback in st_context |
| - st/mesa: don't track FS sampler views for bitmap/drawpix in st_context |
| - st/mesa: don't memset the sampler view array, don't init trailing slots to NULL |
| - st/mesa: sink _mesa_get_samplerobj into st_update_single_texture |
| - st/mesa: read Target only once in st_update_single_texture |
| - st/mesa: return sview from st_update_single_texture via return value, not param |
| - st/mesa: remove the const qualifier for a few st_sampler_view instances |
| - st/mesa: sink refcounting from st_get_sampler_views into st_sampler_view.c |
| - st/mesa: add a mechanism to bypass atomics when binding sampler views |
| - st/mesa: remove the sampler min_lod/max_lod value swap |
| - cso: disallow NULL sampler state templates in cso_single_sampler |
| - cso: update max_sampler_seen only once in cso_set_samplers |
| - cso: don't look up a sampler CSO if the last one is identical |
| - mesa: use atomics instead of mutexes for refcounting texture objects |
| - mesa: use atomics instead of mutexes for refcounting sampler objects |
| - mesa: use atomics instead of mutexes for refcounting renderbuffers |
| - mesa: remove mutex locking from a glBindTexture early out path |
| - mesa: translate into pipe_sampler_state in GL functions |
| - mesa: add LodBias quantization from st/mesa |
| - mesa: add IsBorderColorNonZero to skip border color update for st/mesa faster |
| - mesa: lower GL_CLAMP in texture and sampler functions instead of st/mesa |
| - radeonsi: remove the GDS variants of compute-based primitive discard |
| - radeonsi: change how the prim discard CS is enabled and splitting limits |
| - radeonsi: fix issues with draw-level splitting for the prim discard CS |
| - radeonsi: add optimal multi draws and draw-level splitting for prim discard CS |
| - radeonsi: move the accepting code into the bbox cull branch in prim discard CS |
| - radeonsi: drop smoothing quality to 4xAA for better performance |
| - ac/llvm: don't return a status from ac_cull_triangle because it's unused |
| - ac/llvm: rework how negative W affects culling to not call accept_func twice |
| - radeonsi: rewrite a confusing comment in si_upload_and_prefetch_VB_descriptors |
| - ac/surface/tests: fix the ARM build |
| - radeonsi,radv: fix a late alloc deadlock with <= 6 CUs per SA |
| - radeonsi: move an incorrectly placed comment about late alloc |
| - ac,radeonsi: move late alloc computation into common code and shader states |
| - radeonsi: enable uniform inlining by default |
| - util/idalloc: change num_elements to units of elements instead of bits |
| - util/idalloc: fold the size call into init |
| - util/idalloc: reserving an ID that already exists should be no-op |
| - util/idalloc: hide or remove unused public functions |
| - util/idalloc: add exists and foreach helpers |
| - util/idalloc: add util_idalloc_alloc_range |
| - radeonsi: don't expose no-attachment MSAA 16x on all 1 RB chips due to issues |
| - mesa: fix incorrect comment in draw_gallium_multimode |
| - st/mesa: always use PIPE_USAGE_STAGING for GL_MAP_READ_BIT usage |
| |
| Mark Janes (11): |
| |
| - iris: Increase the size of upload buffers |
| - iris: Upload constant resources for efficient GPU access |
| - iris: Use const_uploader for iris_create_stream_output_target |
| - iris: Use const uploader for blorp vertex data |
| - iris: Use const uploader for draw parameters |
| - iris: Use const uploader for user index data |
| - intel/compiler: Add getter helpers for LSC message descriptor fields |
| - intel/compiler: Add LSC messages to brw_schedule_instructions |
| - intel/fs: Lower DW untyped r/w messages to LSC when available |
| - intel/fs: Lower untyped atomic messages to LSC when available |
| - intel/fs: Lower A64 untyped r/w messages to LSC when available |
| |
| Martin Krastev (1): |
| |
| - compiler/glsl: Use mutex lock while freeing up mem_ctx |
| |
| Martin Peres (1): |
| |
| - ci: add the dEQP expectations for radv on Renoir |
| |
| Matt Turner (10): |
| |
| - intel/eu: Add instruction compaction support on XeHP. |
| - compiler/glsl: Return progress from propagate_invariance() |
| - compiler/glsl: Propagate invariant/precise when splitting arrays |
| - compiler/glsl: Always propagate_invariance() last |
| - freedreno/afuc: Print uintptr_t with PRIxPTR |
| - sparc: Avoid some redefinition warnings |
| - tu: Provide a toggle to avoid warnings about unsupported devices |
| - freedreno/ci: Use TU_IGNORE_CONFORMANCE_WARNING to reduce warnings |
| - ci: Unify on MESA_VK_IGNORE_CONFORMANCE_WARNING |
| - amd/ci: Use MESA_VK_IGNORE_CONFORMANCE_WARNING to reduce warnings |
| |
| Matti Hamalainen (11): |
| |
| - gallium/tools: clean up tracediff.sh a bit |
| - gallium/tools: improve option handling in dump_state.py |
| - gallium/tools: implement better suppression of variants |
| - gallium/tools: implement 'named' pointers option in dump.py |
| - gallium/tools: use left-column output mode of sdiff in tracediff.sh |
| - gallium/tools: improve tracediff.sh argument handling |
| - gallium/tools: implement "high-level" overview mode option in dump scripts |
| - gallium/tools: improve pointer type tracking in parse.py |
| - gallium/tools: add option to use Meld for diffing |
| - aux/trace: add missing return value to trace output |
| - gallium/tools: improve handling of pointer arrays |
| |
| Mauro Rossi (11): |
| |
| - egl/android: include "util/compiler.h" for FALLTHROUGH macro |
| - android: panfrost/lib: add pan_cs.c to Makefile.sources |
| - android: gallium/radeonsi: add nir include path |
| - android: amd/common: add nir include path |
| - android: pan/bi: add bi_opt_constant_fold.c to Makefile.sources |
| - android: nir: add nir_lower_fragcolor.c to Makefile.sources |
| - android: intel/compiler: add brw_compile_ff_gs.c to Makefile.sources |
| - android: i965: remove brw_ff_gs_emit.c from Makefile.sources |
| - android: ac: add ac_nir_lower_ngg.c to Makefile.sources |
| - android: ac: add include src/util path |
| - android: aco: add aco_optimizer_postRA.cpp to Makefile.sources |
| |
| Michael Tang (1): |
| |
| - microsoft/compiler: Maintain sorting of resource type in the context |
| |
| Michael Walle (1): |
| |
| - kmsro: Add mali-dp |
| |
| Michel Dänzer (18): |
| |
| - lima/ppir: Cast pointer to uintptr_t instead of uint64_t |
| - util: Remove unused Android options_tbl_lock |
| - Convert most remaining free-form fall-through comments to FALLTHROUGH |
| - Guard FALLTHROUGH annotations after assert() |
| - llvmpipe: Drop switch with only default case |
| - iris: Drop unneeded default switch case |
| - Use explicit break instead of fall-through to break-only case |
| - ci: Enable -Werror in clang jobs |
| - osmesa: Replace default case FALLTHROUGH annotation by following return |
| - ci: Enable -Werror for the remaining GCC build jobs |
| - ci: Move -Werror enabling from job definitions to meson build script |
| - ci: Add test which occasionally times out to lavapipe-vk skips |
| - Fix up leftover "state_trackers" references to "frontends" |
| - turnip: Mark local variable ASSERTED |
| - ci: Add debian/ prefix to job names for Debian based docker images |
| - ci: Rename Debian based build jobs from meson-* to debian-* |
| - ci: Add Fedora 34 based x86 build docker image |
| - ci: Add Fedora release build job |
| |
| Michel Zou (14): |
| |
| - lavapipe: fix unused variable warning |
| - vulkan: fix duplicate win32 def |
| - gallium: fix uninitialized variable warning |
| - meson: link vulkan_util with link_whole on mingw |
| - docs: list more vulkan extensions |
| - vulkan/wsi: avoid wsi_x11_check_for_dri3 for sw device |
| - zink: fix win32 build |
| - swr: fix uninitialized variable warnings |
| - llvmpipe: restrict optim bug workaround to gcc 10.x |
| - glapi: fix Warray-parameter |
| - zink: Drop useless zink_dispatch_table |
| - zink: Fix win32 build |
| - zink: Fix unused-variable warning |
| - meson: dont use missing dumpbin path |
| |
| Miguel Gomez (1): |
| |
| - i965: Prevent invalid framebuffer usage |
| |
| Mike Blumenkrantz (548): |
| |
| - gallium: add PIPE_BIND_SAMPLER_REDUCTION_MINMAX |
| - gallium: split PIPE_CAP_SAMPLER_REDUCTION_MINMAX into modes |
| - mesa/st: plumb GL_TEXTURE_REDUCTION_MODE_ARB through QueryInternalFormat |
| - zink: hook up VK_EXT_sampler_filter_minmax |
| - zink: support format queries for VK_EXT_sampler_filter_minmax |
| - zink: handle minmax sampler creation for VK_EXT_sampler_filter_minmax |
| - zink: export PIPE_CAP_SAMPLER_REDUCTION_MINMAX_ARB |
| - docs: update GL_ARB_texture_filter_minmax for zink |
| - zink: compare against screen batch id when determining which semaphore to use |
| - zink: always copy the nir shader before compiling |
| - zink: fix tcs slot map eval for user vars |
| - zink: fix tcs input reservation for user vars |
| - st/pbo: use cso_set_vertex_buffers_and_elements() for st_pbo_draw |
| - zink: merge copy-to-scanout path into non-deferred flush path |
| - zink: force scanout sync when mapping scanout resource |
| - util/format: add util_format_is_rgbx_or_bgrx |
| - zink: use undefined layout for first scanout obj transition |
| - Revert "zink: force scanout sync when mapping scanout resource" |
| - zink: move scanout sync to end of batch |
| - zink: add a flag indicating whether scanout object needs updating |
| - zink: move wsi flush info conditional to queue submission |
| - zink: directly set batch->state->flush_res from flush_resource hook |
| - zink: add clear-on-flush mechanic deeper into flush codepath |
| - gallium: when tracing is enabled for threaded drivers, trace the driver thread |
| - nir/lower_fragcolor: set outputs_written for fragdata members |
| - softpipe: fix render condition checking |
| - softpipe: fix streamout queries |
| - softpipe: ci updates |
| - zink: track persistent resource objects, not resources |
| - zink: restore previous semaphore (prev_sem) handling |
| - zink: use cached memory for staging resources |
| - zink: init timeline semaphore on screen creation, not first batch creation |
| - zink: only reset query on suspend if the query has previously been stopped |
| - zink: when performing an implicit reset, sync qbos |
| - lavapipe: implement VK_EXT_provoking_vertex |
| - zink: hook up VK_EXT_provoking_vertex |
| - zink: implement VK_EXT_provoking_vertex |
| - zink: ci updates |
| - zink: update docs |
| - nir/gl_lower_buffers: set access for ssbo load/store instrs |
| - zink: use non-atomic load/store ops if intrinsic is not actually coherent |
| - zink: remove leftover references to flatshading in shader keys |
| - zink: hook up VK_KHR_shader_clock |
| - zink: add conversion util for nir_scope -> SpvScope |
| - zink: add spirv builder for unops with a const operand |
| - zink: support nir_intrinsic_shader_clock |
| - zink: export PIPE_CAP_TGSI_CLOCK |
| - zink: generate spirv 1.5 from ntv when using vk >= 1.2 |
| - zink: create entrypoints for descriptor variables with spirv 1.5 |
| - zink: add fastpath for getting default shader variants |
| - zink: use first-created shader variant as the default |
| - zink: hook up VK_EXT_sample_locations |
| - zink: hook up VK_EXT_conservative_rasterization |
| - zink: hook up VK_EXT_shader_subgroup_ballot |
| - zink: hook up EXT_image_drm_format_modifier |
| - docs: mark off GL_ARB_shader_clock for zink |
| - gallium: rename pipe_draw_start_count -> pipe_draw_start_count_bias |
| - gallium: move pipe_draw_info::index_bias to pipe_draw_start_count_bias |
| - mesa/st: rename DrawGalliumComplex -> DrawGalliumMultiMode |
| - gallium: split drawid out of pipe_draw_info and as a separate draw_vbo param |
| - gallium: remove padding members from pipe_draw_info |
| - util/tc: split out drawid-using draws into a separate call |
| - iris: fix indirect drawid |
| - zink: grab GetPhysicalDeviceMemoryProperties2 from instance |
| - zink: hook up VK_EXT_memory_budget |
| - zink: support PIPE_CAP_QUERY_MEMORY_INFO |
| - zink: minor refactoring of buffer map for read case |
| - zink: add a screen util function for handling VkResults |
| - zink: use zink_screen_handle_vkresult() for fence and timeline waiting |
| - zink: add a ctx function for handling device lost resets |
| - zink: use new ctx device lost checker function |
| - zink: add a pipe_context::resource_commit hook |
| - zink: implement sparse buffer creation/mapping |
| - zink: export PIPE_CAP_SPARSE_BUFFER_PAGE_SIZE |
| - aux/cso_cache: add handling for save/restore of compute states |
| - zink: clamp zs samplers to XXXX swizzle for all non-zero/one swizzles |
| - gallium/inlines: remove atomic set from pipe_reference_init() |
| - nir: add nir_isub_imm |
| - lavapipe: handle buffer sizes better in CmdBindTransformFeedbackBuffersEXT |
| - lavapipe: do not read sampler descriptor info during update if layout has immutables |
| - lavapipe: set events to the unsignalled state on creation |
| - lavapipe: flag renderpasses as having color/zs attachments |
| - lavapipe: update more states on null multisample pipeline info |
| - lavapipe: zero out the dsa state info and flag for updating on null dsa state |
| - lavapipe: zero out the blend state info and flag for updating on null blend state |
| - lavapipe: don't unnecessarily flag dsa states for updating |
| - lavapipe: ignore tess pipeline info if no tess shaders in pipeline |
| - lavapipe: don't access pipeline viewport state when it should be ignored |
| - lavapipe: don't access pipeline dsa state when it should be ignored |
| - lavapipe: don't access pipeline blend state when it should be ignored |
| - zink: split off descriptor layout from descriptor pools |
| - zink: unify pipeline layout creation functions |
| - zink: abstract descriptor init |
| - zink: abstract descriptor usage for programs |
| - zink: abstract descriptor pool usage for programs |
| - zink: use explicit types during descriptor updates |
| - zink: check descriptor layout support before creating it |
| - zink: move more vertex state stuff into the hw state |
| - zink: split vertex state pipeline hashing into its own value |
| - zink: flag pipeline for change more often when vbos change without dynamic state |
| - zink: return current pipeline object if state hasn't changed |
| - zink: hook up dynamic dsa states |
| - zink: start using dynamic front face state |
| - util/hash_table: _mesa_hash_table_create_u32_keys() |
| - zink: add a pipe_context::clear_buffer hook |
| - zink: never use LINEAR for VK_EXT_4444_formats |
| - zink: make ZINK_INLINE_UNIFORMS more standardized in function |
| - zink: clamp 3D surface viewtype to 2D only in the create_surface hook |
| - zink: add a target param to create_ivci() |
| - zink: simplify samplerview surface creation |
| - zink: only set layer info for samplerviews if there are multiple layers |
| - zink: handle in-renderpass clears in fb_clears_apply_internal() |
| - zink: break zs clear loop once both bits are set when beginning renderpass |
| - zink: add debug assert to verify that zink_clear_framebuffer() is accurate |
| - zink: remove compute cruft from resource mapping |
| - zink: break out draw dispatch into separate functions |
| - zink: fix texture barriers for real this time |
| - zink: rework memory_barrier hook again (third time's the charm) |
| - ci: skip glsl-uniform-interstage-limits tests for softpipe jobs |
| - zink: use DONTCARE renderpass when a new scanout fb attachment is set |
| - iris: refcount separate screen objects for resource tracking |
| - zink: stop invalidating descriptor sets on pool destroy |
| - zink: add context-based descriptor info tracking infrastructure |
| - zink: unify resource rebinding |
| - zink: track bind counts for descriptors |
| - zink: update samplerview descriptor layouts when image binds are set |
| - zink: don't track sampler states onto buffer sampler sets |
| - zink: track max slot idx for descriptor types |
| - zink: track number of tbos in shader data |
| - zink: add slot params to zink_context_invalidate_descriptor_state |
| - zink: use better iterating for buffer rebinds |
| - zink: call invalidate on invalid descriptor sets during recycle |
| - zink: make zink_context_update_descriptor_states() static |
| - zink: remove screen param from zink_descriptors_update() |
| - zink: pop descriptor refs when invalidating sets |
| - zink: flush every 100k draws/computes |
| - zink: check for a work_count-based stall in zink_maybe_flush_or_stall() |
| - zink: always do maybe_flush after draw/compute |
| - zink: stop overwriting buffer map pointers for stream uploader |
| - zink: fix DrawParameters shader cap usage |
| - lavapipe: fix fencing when submitting multiple cmdbufs |
| - zink: immediately return false when getting query result if it's not gonna happen |
| - util/queue: don't require a fence when adding a job |
| - zink: split out base renderpass begin into separate function |
| - zink: add a flag for tracking/validating renderpass clears |
| - zink: add flags for determining whether to update framebuffer and renderpass |
| - zink: emit some barriers out of renderpass where possible |
| - nir/builder: add nir_pad_vector and nir_pad_vec4 util functions |
| - zink: don't multiply cube array image layers |
| - zink: populate images with u_blitter if transfer_dst isn't available |
| - zink: add even more validation for linear images before creation |
| - util/primconvert: add C++ guards to header |
| - aux/trace: support pipe_screen::query_memory_info |
| - aux/trace: pipe_screen::query_dmabuf_modifiers |
| - aux/trace: pipe_context::is_dmabuf_modifier_supported |
| - aux/trace: propagate pipe_screen::transfer_helper pointer |
| - aux/trace: pipe_screen::get_dmabuf_modifier_planes |
| - aux/trace: trace pipe_screen::resource_create_with_modifiers |
| - util/prim_restart: fix util_translate_prim_restart_ib |
| - ci: more freedreno flakes |
| - aux/vbuf: prevent uint underflow and assert if no vbs are dirty |
| - aux/trace: add pipe_context::set_debug_callback hook |
| - aux/trace: more effectively unwrap pipe_context params from screen functions |
| - aux/trace: trace transfer ops |
| - aux/trace: stop dumping transfer data for threaded contexts |
| - aux/trace: hook tc methods |
| - aux/trace: fix set_inlinable_constants hook |
| - aux/trace: fix query handling with tc |
| - aux/trace: add a pipe_context::clear_buffer hook |
| - aux/trace: dump 'wait' param for get_query_result |
| - radeonsi: explicitly return support for all index buffer formats |
| - zink: rename ptr_add_usage -> batch_ptr_add_usage |
| - zink: make descriptor_layout_get a public util function |
| - zink: make a public util function for allocating descriptor sets |
| - zink: unify pipeline layout creation and move to descriptor_program_init |
| - zink: pass descriptor type to set layout create() |
| - zink: replace has_descriptors program member with a util function |
| - zink: abstract descriptor functionality and make descriptor structs private |
| - zink: improve samplerview update flagging |
| - zink: emit descriptor barriers and references during bind |
| - zink: add vertex buffer barriers during bind |
| - zink: make timeline_wait use only a screen param |
| - zink: move timeline_wait() to screen function |
| - zink: implement tc idalloc resource id stuff |
| - zink: force streamout rebind when mapping a streamout buffer for writing |
| - zink: implement a tc is_resource_busy hook |
| - zink: call tc_driver_internal_flush_notify() on flush |
| - zink: mark some buffer barrier functions inline/static |
| - zink: switch to memory barriers instead of actual buffer barriers |
| - zink: hook up push descriptor and descriptor template extensions |
| - zink: disable push descriptors on amd |
| - nir/builder: add nir_mask |
| - radv: make radv_pipeline::attrib_ends 32bit |
| - radv: set maxVertexInputAttributeOffset to UINT32_MAX |
| - zink: remove weird lod hack for texturing |
| - zink: ci updates |
| - llvmpipe: remove clamping to [0,1] for tri offset |
| - lavapipe: moar @optimize |
| - llvmpipe: split out scene surface info into separate struct |
| - llvmpipe: split out scene surface init into separate function |
| - llvmpipe: only dump tgsi shaders if they're actually tgsi shaders |
| - llvmpipe: store a screen pointer in resource struct |
| - llvmpipe: stop accessing pipe_resource::screen internally |
| - lavapipe: skip "pipeline barriers" if they're first or last in a cmdbuf |
| - lavapipe: also ignore multiple pipeline barriers in succession |
| - gallium/aux: add helper for pre-clamping clear_buffer value to dword |
| - zink: clamp clear_buffer values |
| - radeonsi: clamp clear_buffer values using new util helper |
| - zink: improve unsupported feature warning message |
| - aux/trace: avoid deadlock in screen::flush_frontbuffer hook |
| - gallivm: fix oob imageLoad with formats that have <4 components |
| - llvmpipe: ci updates |
| - aux/indices: break out primitive type conversion to separate function |
| - aux/indices: break out index size conversion to separate function |
| - aux/indices: break out index count conversion into separate function |
| - aux/indices: employ Delete The Code methodology |
| - lavapipe: add more format mappings for vertex buffer formats |
| - zink: reapply resource/surface refs after app flushes |
| - zink: reapply program refs automatically |
| - zink: remove barriers/refs from descriptor cache |
| - zink: mark some draw functions inline |
| - zink: only rebind pipelines when necessary |
| - zink: handle rebinds for vertex buffers |
| - zink: only rebind vertex buffers when necessary |
| - zink: only update viewport state when necessary |
| - zink: update scissor only when necessary |
| - zink: ref vertex buffers during set_vertex_buffers |
| - zink: stop using util_set_vertex_buffers_mask() |
| - Revert "zink: call tc_driver_internal_flush_notify() on flush" |
| - compiler/spirv: expand_to_vec4 -> nir_pad_vec4 |
| - anv: fix availability for copying timestamp query results |
| - zink: add a second descriptor manager |
| - zink: unify code for updating res->bind_count values |
| - zink: unify more resource bind count tracking code |
| - zink: optimize buffer rebinds |
| - zink: ci updates |
| - aux/trace: dump resource for samplerview and surface |
| - aux/draw: if pipe_draw_info::index_bias_varies is not set, ignore index_bias for N>1 |
| - aux/draw: fix aalines and aapoints for shaders with explicit FragData outputs |
| - radv: declare index_va in a single call for indexed draw packet emit |
| - radv: explicitly load a desc set layout struct member during set allocate |
| - zink: add a util function to create a null surface |
| - zink: replace context-based null framebuffer surfaces with internal ones |
| - zink: create dummy surface/bufferview for null descriptor use |
| - zink: handle null bufferview/imageview descriptors when robustness2 is missing |
| - zink: ci updates |
| - zink: no-op read access buffer barriers if existing access exists for earlier stage |
| - zink: emit fb attachment barriers inline during renderpass start |
| - zink: track number of fb attachment binds on resources |
| - zink: use VK_IMAGE_LAYOUT_DEPTH_STENCIL_READ_ONLY_OPTIMAL when possible |
| - aux/tc: fix ubo unbinding |
| - Revert "Revert "zink: call tc_driver_internal_flush_notify() on flush"" |
| - nouveau: explicitly advertise index buffer format support |
| - r300: explicitly advertise index buffer format support |
| - d3d12: explicitly advertise index buffer format support |
| - zink: explicitly advertise index buffer format support |
| - zink: more accurately handle shader layer/viewport caps |
| - util/prim_restart: assert the index size at the start of the function |
| - util/prim_restart: pre-trim degenerate primitives during draw rewrite |
| - util/prim_restart: store index bounds while rewriting draws |
| - util/prim_restart: store the total index count when rewriting draws |
| - util/prim_restart: update index bounds before draws in util_draw_vbo_without_prim_restart |
| - util/prim_restart: simplify util_draw_vbo_without_prim_restart a bit |
| - zink: populate maxSampleLocationGridSize for all available sample sizes on init |
| - zink: set VK_IMAGE_CREATE_SAMPLE_LOCATIONS_COMPATIBLE_DEPTH_BIT_EXT on zs rts |
| - zink; add a pipe_screen::get_sample_pixel_grid hook |
| - zink: add a pipe_context::set_sample_locations hook |
| - zink: also flag sample_locations_changed if framebuffer samples changes |
| - zink: add a util function for populating VkSampleLocationsInfoEXT |
| - zink: update vk sample location info during framebuffer setup |
| - zink: add a pipe_context::evaluate_depth_buffer hook |
| - zink: use dynamic state to apply sample locations during draw |
| - zink: export PIPE_CAP_PROGRAMMABLE_SAMPLE_LOCATIONS |
| - util/vbuf: fix buffer overrun in attribute conversions |
| - zink: fix caching of shader variants with inlined uniforms |
| - util/blitter: remove duplicated set_sample_mask calls |
| - util/disk_cache: add nocopy variant of disk cache store function |
| - zink: use scissor region for discarding clears during blit |
| - zink: clamp PIPE_CAP_MAX_VIEWPORTS to PIPE_MAX_VIEWPORTS |
| - aux/cso: add flag to disable vbuf |
| - aux/cso: split cso_destroy_context into unbind and a destroy functions |
| - lavapipe: use cso caching |
| - zink: fix typo that's definitely not at all embarrassing or anything like that |
| - aux/cso: store flatshade_first state from rasterizer |
| - util/primconvert: add function for setting flatshade_first |
| - util/vbuf: add flatshade_first to vbuf context and api |
| - aux/cso: set flatshade_first onto vbuf when binding rasterizer |
| - aux/tc: pass rebind count and rebind bitmask with replace_buffer_storage func |
| - util/prim_restart: use more direct conversion for restart index |
| - zink: add a function for creating descriptor layouts for push sets |
| - zink: split lazy sets based on descriptor type |
| - zink: match lazy descriptor set layout in cache mode |
| - zink: modernize cached push ubo descriptor updating |
| - zink: modernize cached ubo descriptor updating |
| - zink: modernize cached ssbo descriptor updating |
| - zink: modernize cached image descriptor updating |
| - zink: remove sorting for dynamic ubo offset updating |
| - zink: move ubo range assert to update_descriptor_state() |
| - zink: unify cached descriptor update code |
| - zink: run lazy batch descriptor functions in cache mode |
| - zink: add is_buffer flag to union zink_descriptor_surface |
| - zink: update null sampler/image descriptor surface with is_buffer during hashing |
| - zink: move shader image descriptor set refs to underlying type |
| - zink: add funcs for descriptor_surface refs |
| - zink: move samplerview descset refs to base objects |
| - zink: enable templated descriptor updates in cache mode |
| - zink: add oob asserts for descriptor set ref setting |
| - zink: skip hash updates for descriptor types which aren't used |
| - zink: unblock last_set cached descriptor reuse when safe to do so |
| - zink: add ZINK_DESCRIPTORS env var to explicitly set a mode |
| - zink: remove zink_batch_state::descs_used |
| - zink: split batch state work_count into separate vars |
| - zink: reorder has_barriers flag in batch state struct |
| - zink: optimize zink_tc_fence struct packing |
| - zink: move batch usage functions to static inlines |
| - zink: remove atomic from batch usage setting |
| - zink: make batch_usage_unset take a batch state param |
| - zink: unset program batch usage on state reset |
| - zink: remove unnecessary conditionals in resource batch tracking |
| - zink: make batch_usage_set take a batch state param |
| - zink: make batch_usage_matches take a batch state param |
| - zink: cache descriptor update templates along with layout |
| - zink: track active use counts for descriptor layouts |
| - zink: destroy lazy descriptor pools during batch reset when unused |
| - zink: slightly refactor program updating during draw |
| - zink: remove return types from program update functions during draw |
| - zink: simplify zink_program_has_descriptors() |
| - zink: mark bind_stage() as inline |
| - zink: unify gfx shader create callbacks |
| - zink: use u_live_shader_cache |
| - zink: remove unnecessary draw checks |
| - zink: move batch decl to top of draw_vbo |
| - zink: stop sanitizing primitive_restart flag in draw info |
| - zink: handle nir_op_pack_64_2x32 |
| -
|
