6158 lines
279 KiB
ReStructuredText
6158 lines
279 KiB
ReStructuredText
Mesa 24.3.0 Release Notes / 2024-11-21
|
||
======================================
|
||
|
||
Mesa 24.3.0 is a new development release. People who are concerned
|
||
with stability and reliability should stick with a previous release or
|
||
wait for Mesa 24.3.1.
|
||
|
||
Mesa 24.3.0 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 24.3.0 implements the Vulkan 1.3 API, but the version reported by
|
||
the apiVersion property of the VkPhysicalDeviceProperties struct
|
||
depends on the particular driver being used.
|
||
|
||
SHA checksums
|
||
-------------
|
||
|
||
::
|
||
|
||
SHA256: 97813fe65028ef21b4d4e54164563059e8408d8fee3489a2323468d198bf2efc mesa-24.3.0.tar.xz
|
||
SHA512: 20168ae4c278776a60d5febf53b3367cf08bffffb40ef2054821e68d7a8c37a07871d097ab17555f41a4fe716f0de7df95ad7d452b1ed57db6527838eb839ba4 mesa-24.3.0.tar.xz
|
||
|
||
|
||
New features
|
||
------------
|
||
|
||
- Expose Vulkan 1.3 on v3dv, both rpi4 and rpi5
|
||
- VK_EXT_descriptor_buffer on nvk
|
||
- VK_EXT_post_depth_coverage on nvk
|
||
- VK_KHR_video_maintenance1 on radv
|
||
- VK_EXT_legacy_vertex_attributes on nvk
|
||
- GL_KHR_shader_subgroup on radeonsi
|
||
- VK_KHR_maintenance7 on nvk
|
||
- VK_KHR_dynamic_rendering_local_read on nvk
|
||
- GL_ARB_timer_query on Panfrost
|
||
- GL_EXT_disjoint_timer_query on Panfrost
|
||
- VK_KHR_pipeline_binary on RADV
|
||
- VK_KHR_compute_shader_derivatives on anv
|
||
- VK_NV_compute_shader_derivatives on nvk
|
||
- VK_KHR_compute_shader_derivatives on nvk
|
||
- VK_KHR_compute_shader_derivatives on radv
|
||
- VK_KHR_shader_relaxed_extended_instruction on anv, hasvk, hk, nvk, radv, tu, v3dv, lvp
|
||
- GL_OVR_multiview and GL_OVR_multiview2 on zink
|
||
- VK_KHR_shader_float_controls2 on radv
|
||
- VK_KHR_shader_float_controls2 on nvk
|
||
- VK_EXT_device_generated_commands on nvk, radv
|
||
- VK_EXT_host_image_copy on nvk/Turing+
|
||
- VK_EXT_depth_clamp_control on anv, hasvk, nvk, radv
|
||
- VK_KHR_shader_quad_control on nvk
|
||
- GL_EXT_draw_buffers2 on etnaviv/HALTI5+
|
||
- GL_ARB_draw_buffers_blend on etnaviv/HALTI5+
|
||
- VK_KHR_fragment_shading_rate on NVK
|
||
- GL_ARB_draw_indirect on etnaviv/HALTI5+
|
||
- VK_EXT_depth_clamp_zero_one on NVK
|
||
- GL_ARB_framebuffer_no_attachments on etnaviv
|
||
|
||
|
||
Bug fixes
|
||
---------
|
||
|
||
- anv: Use-after-free detected by AddressSanitizer while running dEQP-VK
|
||
- Confidential issue #11770
|
||
- radv: DCC causes glitches in Red Dead Redemption 2
|
||
- A5xx rendering issues with firefox
|
||
- [ANV][Regression] Broken rendering in Flycast + Per-Pixel Alpha Sorting
|
||
- vtn: OpTypeStruct in kernel parameters trigger assertion in glsl_types.h
|
||
- anv: Assertion failure in \`dEQP-VK.image.extended_usage_bit_compatibility.image_format_list.s8_uint_optimal_transfer_src_bit`
|
||
- radv: Resident Evil 6 Benchmark Tool has artifacts on 7900 XTX when DCC is enabled, game launched on 4K monitor without scaling and with FullHD settings
|
||
- [AMD RX 6700 XT] Artifacts while upscaling games in fullscreen mode
|
||
- Distorted pixelated graphics with Radeon RX 7900 XT with some games
|
||
- Total War Warhammer 2 Graphical Glitch
|
||
- Glitching artifacts in tile shaped patterns on 6700 XT, when using upscaled fullscreen game on labwc
|
||
- anv: Page fault when using MTL simulator in dEQP-VK.ray_tracing_pipeline.data_spill.report_intersection.float32
|
||
- mesa_cache_db.c:316:33: error: call to undeclared function 'mremap'
|
||
- [trunk] shaders fail hard in openmw after cbfc225e2bda2c8627a4580fa3a9b63bfb7133e0
|
||
- u_perfetto.h:33:9: error: unknown type name 'clockid_t'; did you mean 'clock_t'?
|
||
- brw_fs_opt_copy_propagation incorrectly handles size changes of uniforms
|
||
- nouveau paraview msaa corruption 23.1 bisected regression
|
||
- mesa fails to build due to missing SPV_ENV_UNIVERSAL_1_6 symbol
|
||
- Missing xshmfence dependency for X11 WSI on macOS
|
||
- [r300g] Unigine Tropics: black stripes between center of the screen and horizon
|
||
- [AMDGPU RDNA3] Black square artifacts in viewport renders in Blender
|
||
- radv: test_sm67_sample_cmp_level test in vkd3d-proton fails
|
||
- [ANV] LNL tiled corruption in background traci Blackops3-trace-dx11-1080p-high
|
||
- [anv] Possible regression in e3814dee1ac0f90771b921a4f6f5aed10f06e8d4
|
||
- Confidential issue #12092
|
||
- radv: significantly less performance than amdgpu-pro in the AMD FSR2 sample
|
||
- radv: wrong index value in radv_skip_graphics_pipeline_compile
|
||
- Android: IMapper4 / HIDL is being deprecated
|
||
- d3d12_video_screen.cpp undefined reference to ConvertHEVCSupportFromProfile(D3D12_VIDEO_ENCODER_PROFILE_HEVC, D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC1*)
|
||
- anv: Reduce the number of partial resolves for resource transitions with CCS_E
|
||
- Silent Hill 2 Remake performance regression with f5805bcb
|
||
- Disk Cache DB file descriptors are not closed on exec
|
||
- Shader cache takes too many fds
|
||
- Shader cache takes too many fds
|
||
- Disk Cache DB file descriptors are not closed on exec
|
||
- radv: \`test_instruction_msad_dxil` from vkd3d-proton fails on hawaii
|
||
- [anv] functional regression with 02294961eeeec003fd6113c0406a80ea1a2fec72 on DG2
|
||
- [vulkan-intel][regression] War Thunder causes GPU hang
|
||
- RFE: rusticl: out-of-order queues
|
||
- [nak][nouveau] - error[E0425]: cannot find value \`DRM_RS_IOCTL_NOUVEAU_EXEC` in this scope --> ../src/nouveau/compiler/nak/hw_runner.rs:193:17
|
||
- [radeonsi] glLinkProgram terminates the application when Shader is passing Bindless Texture into function
|
||
- gallium: crash when using images obtained from a texture with a bindless handle
|
||
- Using DXT1 textures fails in GL ES 1, even with EXT_texture_compression_dxt1 supported
|
||
- zink: discrepancy between reported max LOD bias and the actual allowed range
|
||
- gallium: Crash when writing to writeonly image3D passed in via parameter
|
||
- vulkan/queue: heap-use-after-free when enabling address sanitizer
|
||
- r300 regression : Upside down and broken redering for Source games
|
||
- vk_texcompress_astc_finish() may crash because astc is null
|
||
- intel A770 dg2 silent hill 2 doesn't run
|
||
- radv: vkGetPipelineKeyKHR global key is bugged
|
||
- Mesa's \`intel_hang_replay` tool fails to work
|
||
- Mesa's \`intel_hang_replay` tool fails to work
|
||
- radv: fix dEQP-VK.texture.mipmap.3d.image_view_min_lod.base_level.* regressions on GFX6-8
|
||
- ci: Uprev Piglit failed
|
||
- PanVK-v10: Missing support for vkCmdDispatchIndirect
|
||
- radv: slow packed math in nested loops due to copies
|
||
- nvk,nak: Implement VK_KHR_shader_quad_control
|
||
- [Turnip] Detroit Become Human gets stuck perpetually compiling a shader
|
||
- Faaast Penguin requires force_vk_vendor set
|
||
- [clover] linking of libMesaOpenCL.so.1.0.0 fails with LLVM/Clang 20: undefined reference to \`clang::api_notes::APINotesManager::loadCurrentModuleAPINotes(...)`
|
||
- zink missing dep on spirv_info_h?
|
||
- ANV: Deadlock(game) bugged invisible player/NPC models and missing textures
|
||
- anv: dEQP-VK.mesh_shader.ext.misc.clip_geom_provoking_last and dEQP-VK.mesh_shader.ext.misc.clip_geom_and_task_shader_provoking_last fail on MTL
|
||
- anv: dEQP-VK.mesh_shader.ext.misc.clip_geom_provoking_last and dEQP-VK.mesh_shader.ext.misc.clip_geom_and_task_shader_provoking_last fail on MTL
|
||
- anv: dEQP-VK.subgroups.clustered.compute.subgroupclusteredmul_i64vec4_requiredsubgroupsize and friends time out due to register spilling
|
||
- [clc] [FTBFS] [llvm20+] no matching function for DataLayout
|
||
- building lavapipe with LLVM-19 and running any vulkan app on a Zen4 CPU shows text on terminal..
|
||
- PanVK-v10: Missing support for vkCmdDraw[Indexed]Indirect
|
||
- Adreno 740v3 (Quest 3) Image Corruption
|
||
- [regression] "anv: Change params of anv_can_fast_clear_color_view" breaks a bunch of Sparse tests on LNL
|
||
- Update defaults for launch crash with XESS in Satisfactory 1.0
|
||
- [bisected][zink] dEQP-GLES3.functional.fbo.multiview.samples_1 failing after feature enabled on zink
|
||
- ACO Error: Unsupported opcode: v4 -- AMD Radeon RX 590 Series (Black Myth Wukong Benchmark Tool)
|
||
- !31075 cause graphical artifact with Warhammer 40,000: Space Marine 2
|
||
- PanVK-v10: VulkanCTS 1.3.9.2 - 3D Copy-and-blit tests hit GPU fault on Rock5B (RK3588)
|
||
- [regression] paraview regressed on T860
|
||
- PanVK-v10: Assert in cs_to_reg_tuple for panvk_v10_CmdWaitEvents2
|
||
- PanVK-v10: VulkanCTS 1.3.9.2 - Segfault in panvk_v10_link_shaders for nocolor-tests on Rock5B (RK3588)
|
||
- ANV: scalar FS validation failure is always hit when playing the steam game "Black Myth: Wukong Benchmark Tool"
|
||
- PanVK-v10: VulkanCTS 1.3.9.2 - Multiple tests hit "Unknown sampler dim type"-assert on Rock5B (RK3588)
|
||
- [REG 24.1.7->24.2.1, bisected] RADV: uVkCompute's mad_throughput benchmark hangs
|
||
- [panfrost] rk3588 Ion Fury game (from gog via box64) no image
|
||
- [panfrost] rk3588 Ion Fury game (from gog via box64) no image
|
||
- [panfrost] rk3588 Ion Fury game (from gog via box64) no image
|
||
- [panfrost] rk3588 Ion Fury game (from gog via box64) no image
|
||
- [v3d] performance regression in GTK's NGL renderer from 24.1 to 24.2
|
||
- Color attribute not restored properly when using glPushAttrib/glPopAttrib calls
|
||
- 69c772e4ea8f (egl: link with libgallium directly) breaks VAAPI - OpenGL interop on radeonsi
|
||
- Select VA-API device under Wayland in Multi-GPU setup
|
||
- nine: intel-whl: ERROR: src0 is null
|
||
- [clc] [FTBFS] [llvm20+] no matching function for GetResourcesPath
|
||
- [AMDGPU RDNA3] Viewport compositing generates severe graphical artifacts in Blender
|
||
- v3d gbm_bo_unmap() crashes when used by TigerVNC
|
||
- !30610 broke piglit program\@execute\@bswap with rusticl on radeonsi
|
||
- [Regression] [BISECTED] Genshin Impact: GPU HANG: ecode 12:1:85def5fb
|
||
- Arma Reforger causes amdgpu crash on 24.2
|
||
- [Regression][Vulkan][TGL]Failed to render to depth image at non-zero mip level on Mesa 24.1 and later
|
||
- [ANV] Breaking Limit benchmark does not run
|
||
- [anv][dg2,mtl][bisected][regression] dEQP-VK.ray_tracing_pipeline.acceleration_structures.device_compability_khr.gpu_built.bottom failure
|
||
- Artifacts on Black Myth Wukong
|
||
- DCS world radeon rx 7900 flicker
|
||
- radv/aco: Scheduler bug in ACO somewhere in parallel-gs
|
||
- Once human black artifacts consume the screen
|
||
- [Regression] Wayland-only MESA fails to build without libx11-xcb-dev, libxcb-dri3-dev, libxcb-present-dev, etc.
|
||
- Possible race condition when generating enums.h
|
||
- Rusticl: unsigned int division results in wrong value
|
||
- a6xx spec\@arb_texture_view\@rendering-formats and spec\@arb_shader_image_load_store\@bitcast regression
|
||
- [Bisected] Chromium GPU Process Crashes with Vulkan Backend
|
||
- radv: weird dEQP-GLES31.functional.blend_equation_advanced.msaa.* failures if DRLR is enabled with RADV
|
||
- Anv/Zink ICL hang
|
||
- [anv] possible regression found on 982106e6768e6f69d71710e314325c813591452d
|
||
- xe2: Non-compiler assertion failure replaying cp2077 fossil
|
||
- iris: GfxBench5's Car Chase slow-clears LOD0 when using HIZ_CCS_WT
|
||
- Mesa doesn't build with the following config options
|
||
- Intel: Enable Compression on depth buffers (Xe2)
|
||
- mpv fails with VA-API for Radeon RX 6600, Mesa 24.3.0-devel
|
||
- Failed to cross-compile Mesa-24.2 & Rusticl with bindgen 0.65.1 and rustc 1.80.0
|
||
- Confidential issue #11711
|
||
- Spectacle crashes recording h264 video with 24.2-rc3
|
||
- [nak][nouveau][nvk] FAILED: src/nouveau/compiler/nak
|
||
- u_debug_stack_test.capture_not_overwritten fails on i386
|
||
- [bdw,bsw][bisected] \*.spec.nv_compute_shader_derivatives.execution.derivatives*
|
||
- a6xx spec\@arb_texture_view\@rendering-formats and spec\@arb_shader_image_load_store\@bitcast regression
|
||
- turnip: vulkaninfo crashed
|
||
- turnip-freereno Build error (/usr/local/include/xf86drm.h:40:10: fatal error: drm.h: No such file or directory)
|
||
- tu: compilation failing when compiling turnip with only kgsl and X11 enabled
|
||
- samplerCube constructor in fragment shader no longer converting uvec2 bindless texture handles ( segmentation fault , core dumped )
|
||
- ci/llvm18: AMD unitialized value
|
||
- anv: gen9.5 flakiness in dEQP-VK.multiview.dynamic_rendering.depth.*
|
||
- Turnip: shaderInt8 support
|
||
- intel/brw: q2rtx RT pipeline fails EU validation
|
||
- vaapi decoding corruption with green blocks
|
||
- interpolateAt precision lowering unhandled by glsl_to_nir()
|
||
- [anv] CS2 crashes on LNL
|
||
- [anv] Dota2 does not start on LNL
|
||
- Raspberry Pi 5 Vulkan 1.3 missing API checklist
|
||
- [radeonsi][bisected][regression] glClientWaitSync() quickly times out with INT64_MAX timeout
|
||
- d3d10umd: Build regression in 24.2.0-devel
|
||
- zink/tu: glcts flake on a750
|
||
- nouveau: advertises GL_EXT_memory_object without implementing \`*UID` callbacks
|
||
- LIBGL_DRIVERS_PATH gone
|
||
- radv: GPU hang with Tauray
|
||
- [Regression][Vulkan][TGL][Bisected]vkCmdCopyQueryPoolResults failed to write buffer with compute pipeline on Mesa 24.1
|
||
- Worms Revolution: not rendering explosion effects
|
||
- crash on pushbuf_validate nvc0_blit do_blit_framebuffer
|
||
- ci_run_n_monitor.py won't continue enabling the jobs
|
||
- ci_run_n_monitor.py doesn't monitor manual jobs
|
||
- piglit: cl-api-build-program crashes
|
||
- nir: regression with bindless textures (bisected)
|
||
- i915g: glGenerateMipmap() fails with 2048×2048 textures
|
||
- gallium-va: VA_RC_QVBR support
|
||
- nir/glsl: vectorize_io unpredictably vectorizes mismatched types
|
||
- [radeonsi] Asterix & Obelix XXLRomastered: river misrendered (completely black)
|
||
- Build fails without Vulkan
|
||
- No dependency check for PyYAML in meson.build
|
||
- GPU Hang in Metal Gear Rising Revengeance
|
||
- VK_ERROR_DEVICE_LOST A770 DXVK Fallout 3
|
||
|
||
|
||
Changes
|
||
-------
|
||
|
||
Aaron Germuth (1):
|
||
|
||
- Add VK_EXT_host_query_reset to guest's allowed device extensions
|
||
|
||
Aaron Ruby (15):
|
||
|
||
- gfxstream: Advertise VK_KHR_external_semaphore_fd always on Linux/Android guests.
|
||
- gfxstream: logging on Linux guests
|
||
- gfxstream: Fix createResource call for Linux-guest VirtGpu layer
|
||
- gfxstream: translate physicalDevices list in VkPhysicalDeviceGroupProperties to their gfxstream objects
|
||
- gfxstream: Use vk_find_struct for other struct searches in gfxstream_vk_CreateDevice
|
||
- Revert "Reland "Partial revert of aosp/2858589 to avoid Mesa layer for Android""
|
||
- gfxstream: Add "transformList" method for VkFences, in addition to VkSemaphores
|
||
- gfxstream: Clean up VK_KHR_external_memory_fd support for Linux
|
||
- gfxstream: Revert all unnecessary gfxstream_vk_* object types and corresponding translations in functable.
|
||
- gfxstream: Remove VkImage from functable handle translation
|
||
- gfxstream: Change createResource to use stride instead of bpp
|
||
- gfxstream: Switch to using scanout path with native DRM images for Linux WSI
|
||
- gfxstream: Add linearPeerImage for Linux guests, to use for imageLayout queries
|
||
- gfxstream: Add vkCmdPipelineBarrier2 implementation in VkDecoderGlobalState
|
||
- gfxstream: ResourceTracker: Fix bug in return-code check for blob wait
|
||
|
||
Adam Jackson (8):
|
||
|
||
- meson: Introduce the legacy-x11 option
|
||
- glx: Move DRI2 client support behind legacy-x11
|
||
- egl/x11: Move DRI2 client support behind legacy-x11
|
||
- gallium/video: Move DRI2 support behind legacy-x11
|
||
- rusticl: Add clCreateSubDevices stub
|
||
- vtn: Handle SPV_INTEL_optnone
|
||
- rusticl: Enable out-of-order execution
|
||
- glx: Fix the GLX_EXT_swap_control_tear drawable attributes
|
||
|
||
Aditya Swarup (3):
|
||
|
||
- iris: Disable fast clear when surface height is 16k
|
||
- anv: Disable fast clear when surface height is 16k
|
||
- anv: Use RCS engine for copying stencil resource for gfx125
|
||
|
||
Agate, Jesse (2):
|
||
|
||
- amd/vpelib: Refactor structs for API change
|
||
- amd/vpelib: Refactor frontend and backend config callback
|
||
|
||
Alan Liu (2):
|
||
|
||
- amd/vpelib: Amend log for tone map support check
|
||
- amd/vpelib: Remove unused structs
|
||
|
||
Alejandro Piñeiro (5):
|
||
|
||
- v3dv/device: set value for maxInlineUniformTotalSize
|
||
- v3dv: update conformanceVersion to 1.3.8.3
|
||
- v3dv: expose Vulkan 1.3
|
||
- panfrost: add support for DRM_PANTHOR_DEV_QUERY_TIMESTAMP_INFO on drm-shim
|
||
- panfrost: make DRM version panfrost/panthor conditional on drm-shim
|
||
|
||
Aleksi Sapon (16):
|
||
|
||
- gallivm: Fix LLVMPipe codegen issues discovered on Apple Silicon
|
||
- util: const hash_table for some readonly operations
|
||
- llvmpipe: much faster sample function cache using RCU pointers
|
||
- llvmpipe: half_to_float use LLVM fpext when fp16 is available
|
||
- llvmpipe: improvements to JIT assembly dump
|
||
- llvmpipe: FlameGraph profiling support
|
||
- wsi: Metal support
|
||
- wsi: Fix Metal WSI CAMetalDrawable leak
|
||
- llvmpipe: fix quad group helper invocation masking
|
||
- llvmpipe: correctly implement output variables loads
|
||
- llvmpipe: implement alpha-to-coverage dithering
|
||
- lavapipe: enable alpha-to-coverage dithering
|
||
- llvmpipe: Fix anisotropic sampling aliasing bug
|
||
- llvmpipe: Fix pmin calculation
|
||
- softpipe: Fix anisotropic sampling aliasing bug
|
||
- llvmpipe: update traces for aniso filtering fix
|
||
|
||
Alessandro Astone (2):
|
||
|
||
- egl/gbm: Walk device list to initialize DRM platform
|
||
- panvk: Add missing headers to android platform
|
||
|
||
Ali Homafar (1):
|
||
|
||
- zink: Optimize descriptor buffers struct filling
|
||
|
||
Alvin Wong (1):
|
||
|
||
- hasvk: Conditionally expose VK_KHR_present_wait
|
||
|
||
Alyssa Rosenzweig (403):
|
||
|
||
- vk/meta: add buffer view tracking
|
||
- vk/meta: add VK_IMAGE_VIEW_CREATE_INTERNAL_MESA flag
|
||
- asahi: drop dead code
|
||
- asahi: drop unused patch index buffer lowering
|
||
- asahi: refactor prolog lowering
|
||
- asahi: drop pointless conversion
|
||
- asahi: drop #includes
|
||
- libagx: fix prefix sum kernel
|
||
- asahi: make GS inputs explicit
|
||
- asahi: add TES->compute lowering
|
||
- asahi: split up agx_nir_lower_vs_before_gs
|
||
- libagx: hoist code out of loop
|
||
- asahi: offset buffer images in software
|
||
- ail: constify more arguments
|
||
- ail: add ail_get_level_size_B helper
|
||
- libagx: add statistic increment kernel
|
||
- libagx: mask counters to 32-bit
|
||
- libagx: rework tess i/o
|
||
- hk: add Vulkan driver for Apple GPUs
|
||
- agx: fix ballot extend packing
|
||
- asahi: don't count helper invs in pipeline stats query
|
||
- nir/opt_uniform_atomics: add fs atomics predicated? flag
|
||
- nir/divergence_analysis: handle AGX stuff
|
||
- agx: add some SRs
|
||
- agx: use opt_uniform_atomics
|
||
- nir/opt_peephole_select: allow speculatable load constant
|
||
- nir: add derivative intrinsics
|
||
- nir/opt_peephole_select: allow derivatives
|
||
- nir/gather_info: handle derivative intrinsics
|
||
- nir/lower_wpos_ytransform: handle intrinsic ddx
|
||
- nir/opt_constant_folding: handle derivative intrinsics
|
||
- nir/divergence_analysis: handle derivative intrinsics
|
||
- nir/instr_set: allow derivative intrinsics
|
||
- nir/schedule: handle derivative intrinsics
|
||
- treewide: use new-style derivative builders
|
||
- vtn: use new derivative builders
|
||
- radv: use glsl function name for dFdxfine
|
||
- amd: switch to derivative intrinsics
|
||
- agx: convert to ddx intrinsics
|
||
- agx: fix clang-format
|
||
- r600: remove 64-bit ddx handling
|
||
- r600: switch to derivative intrinsics
|
||
- dxil: switch to derivative intrinsics
|
||
- broadcom: switch to derivative intrinsics
|
||
- vc4: switch to derivative intrinsics
|
||
- brw: switch to derivative intrinsics
|
||
- elk: switch to derivative intrinsics
|
||
- ir3: switch to derivative intrinsics
|
||
- nir: add nir_shader_alu_pass
|
||
- treewide: use nir_shader_alu_pass
|
||
- nir/lower_alu: use alu_pass
|
||
- nir/scale_fdiv: use alu_pass
|
||
- nir/opt_idiv_const: use alu_pass
|
||
- nir/lower_packing: use alu_pass
|
||
- nir/opt_reassociate_bfi: use alu_pass
|
||
- asahi: fix warning
|
||
- agx: factor out agx_device_key
|
||
- agx: exploit soft fault info
|
||
- hk,asahi: pass through soft fault to VBO lower
|
||
- asahi: optimize out vbo clamp with soft fault
|
||
- hk: optimize bounded loads with soft fault
|
||
- asahi: remove agx_bo::dev
|
||
- asahi: drop segment list parsing
|
||
- asahi: drop mprotect in decoding
|
||
- asahi: drop agx_bo::type
|
||
- asahi: stop fixing up USC addresses
|
||
- asahi: drop iThing logic
|
||
- asahi: always pass align parameter for BOs
|
||
- asahi: introduce agx_va data structure
|
||
- hk: fix BDA alignment
|
||
- asahi,hk: generalize bo_bind hook
|
||
- hk: fix restart index overemit
|
||
- nir,agx: add "active threads in subgroup" intrinsic
|
||
- libagx: add drawIndirectCount clone&patch kernel
|
||
- hk: implement drawIndirectCount
|
||
- hk: fix vkd3d-proton descriptor-performance
|
||
- agx: optimize popcount(ballot(true))
|
||
- agx: handle min_lod sources
|
||
- hk: advertise shaderResourceMinLod
|
||
- hk: advertise more eds features
|
||
- agx: tweak jmp_exec_none heuristic
|
||
- agx: fuse not into if
|
||
- agx: let if-fusing opts interact
|
||
- hk: denouveau
|
||
- hk: advertise KHR_sampler_ycbcr_conversion
|
||
- agx: fix block_image_store indirect ts
|
||
- nir: model AGX explicit coordinate intrinsics
|
||
- nir: add image_block_size_per_thread_agx info
|
||
- agx: plumb imageblock stride
|
||
- agx: allow st_tile in compute shaders
|
||
- agx: handle explicit coordinate tib instructions
|
||
- agx: add a comment about an unknown bit
|
||
- asahi: pack imageblock words for compute
|
||
- asahi: switch to block based compute blitter
|
||
- agx: lower exact frcp
|
||
- elk: fix compute shader derivatives
|
||
- ntt: switch to derivative intrinsics
|
||
- r300: switch to derivative intrinsics
|
||
- pan/bi: switch to derivative intrinsics
|
||
- pan/mdg: convert to derivative intrinsics
|
||
- nak: switch to derivative intrinsics
|
||
- vtn: infer SpvCapabilityLinkage
|
||
- asahi/clc: drop caps list
|
||
- nir/divergence_analysis: handle load_agx
|
||
- nir,agx: make block image store an image() intrinsic
|
||
- agx: support bindless block image store
|
||
- agx: fix wait_pix scheduling with eMRT
|
||
- ail: swallow up formats
|
||
- ail: consider format for compressability
|
||
- ail: clarify metadata buffer calculation
|
||
- ail: add metadata width/height in tile helpers
|
||
- ail: add enums/queries for AGX compression modes
|
||
- asahi/clc: add bindless image intrinsics
|
||
- libagx: add in-place decompression kernel
|
||
- asahi,ail: pull texture view logic into ail
|
||
- asahi,hk: deduplicate txf sampler settings
|
||
- asahi: skip unmapped bo in fault dbg
|
||
- asahi: optimize aligned blits
|
||
- asahi: allow rgb9e5 rendering internally
|
||
- asahi: factor out agx_usc_shared_non_fragment
|
||
- asahi: allow compute blitter to handle rgb9e5 blits
|
||
- asahi: fix fbfetch with eMRT
|
||
- asahi: add agx_fill_decompress_push helper
|
||
- asahi: handle cross-process eMRT
|
||
- asahi: add missing upcast for eMRT MSAA
|
||
- hk: be stricter about formats in copies
|
||
- hk: rework meta for compression
|
||
- hk: prep for block based blitter
|
||
- hk: remove texel buffers from meta
|
||
- hk: fix bg key with eMRT
|
||
- hk: handle compressed eMRT
|
||
- hk: lock device-wide scratch access
|
||
- hk: drop irrelevant comments
|
||
- hk: shrink cmd bo
|
||
- hk: disable submission merging
|
||
- hk: wire up compression
|
||
- lima: set no_integers
|
||
- etnaviv: set no_integers
|
||
- freedreno/a2xx: set no_integers
|
||
- svga: set no_integers
|
||
- virgl: set no_integers
|
||
- nir: use Horner's method for atan
|
||
- nir: correct comment for atan range reduction
|
||
- nir: simplify atan range reduction
|
||
- nir: push down fabs for atan
|
||
- nir: extend copysign for no-integer hw
|
||
- nir: use copysign for atan
|
||
- nir: simplify atan range reduction fixup
|
||
- nir: negate an expression in atan
|
||
- nir: fuse ffma for atan range fixup
|
||
- nir: push up abs in atan2 calculation
|
||
- nir: drop NaN fixup for atan
|
||
- nir: CSE comparisons in atan2
|
||
- ci: update traces
|
||
- virtio/vdrm: don't spam
|
||
- tu/virtio: silence startup spam on asahi
|
||
- zink: fix buffer view clamp overflow
|
||
- vulkan: merge driver internal bits
|
||
- vulkan: export vk_robustness_disabled
|
||
- hk: add missing dep
|
||
- ail: pull in DRM modifier helper
|
||
- hk: handle compression modifiers
|
||
- hk: disable mods on funny float formats
|
||
- hk: advertise modifiers
|
||
- hk: switch to vk_meta for buffer copies
|
||
- hk: drop some dead code
|
||
- hk: drop bad comment
|
||
- hk: quiet gcc warning
|
||
- asahi: add tib sample setter helper
|
||
- hk: fix attachmentless multisampling tile sizes
|
||
- hk: ban compression with feedback loops
|
||
- hk: lower memory model
|
||
- agx: pull out the big hammers for device-barriers
|
||
- agx: plumb COHERENT
|
||
- hk: free large_bos
|
||
- hk: don't leak NIR with VS/GS
|
||
- hk: don't leak GS rast with rast disc
|
||
- hk: don't leak GS subparts in an error path
|
||
- libagx: cleanup loop vars
|
||
- libagx: simplify tcs out calc
|
||
- libagx: drop unused arg
|
||
- asahi: fix UB affecting x86 builds
|
||
- asahi: identify factor buffer size
|
||
- asahi/decode: fix ptr casts
|
||
- asahi/virtio: fix ptr cats
|
||
- asahi/virtio: fix allocate-with-alignment
|
||
- agx: fix spilling in i686 builds
|
||
- agx: fix =spill
|
||
- agx: clarify scratch size units
|
||
- agx: allocate scratch in spilling unit test
|
||
- agx: clarify spill lowering math
|
||
- hk: fix weird formatting
|
||
- hk: quiesce gcc uninitialized var warning
|
||
- hk: fix debug cache key
|
||
- hk: fix ptr cast
|
||
- hk: shush gcc warning
|
||
- hk: fix loop submit with NONE
|
||
- hk: fix vertex clustering
|
||
- hk: fix oq availability set
|
||
- hk: fix phi ordering
|
||
- hk: drop dead
|
||
- hk: clarify spilling
|
||
- hk: mutex shader compiles when debugging
|
||
- hk: add notess perftest
|
||
- hk: add noborder perf test
|
||
- hk: soup up =perf
|
||
- hk: optimize empty tiles
|
||
- hk: optimize out empty VDM batches
|
||
- hk: assert batches are nonempty
|
||
- hk: parallelize after-graphics available sets
|
||
- hk: run opt_access
|
||
- nir: add bounds_agx opcode
|
||
- asahi: carveout lower VAs
|
||
- hk: use 0 pointer for 0-sized buffers
|
||
- agx: handle bounds_agx
|
||
- hk: clean up before lowering bounds checks
|
||
- hk: smarten bounds check lowering
|
||
- agx: optimize robust access
|
||
- agx: clean up packs from bounds check
|
||
- asahi/lower_vbo: optimize robustness2
|
||
- asahi: fix speculation of rgb32 loads
|
||
- asahi,hk: disable rgb32 buffer textures
|
||
- asahi: optimize trivial load_invocation_id
|
||
- agx: handle hw tess eval shaders
|
||
- asahi: add tess assert
|
||
- asahi: add nosoft debug
|
||
- agx: increase lower vars to scratch thresh
|
||
- hk: fix =spill with tess
|
||
- agx: add spilling debug helpers
|
||
- agx: fix spilling around exports
|
||
- agx: add helper to pad 16-bit to 32-bit
|
||
- agx: use padding helper
|
||
- agx: factor out reserved_region
|
||
- agx: fix RA shuffling algorithm
|
||
- agx: add helper to visualize reg file
|
||
- asahi/genxml: add nicer error checking
|
||
- asahi: fix AGX_MAX_OCCLUSION_QUERIES
|
||
- asahi,hk: fix st_tile mask size
|
||
- agx: validate phi sources are defined
|
||
- agx: validate sizes are consistent in the IR
|
||
- agx: add pass manager
|
||
- agx: split up optimizer
|
||
- agx: use reserved_size
|
||
- asahi: #define the ABI
|
||
- asahi: move sample mask to r1l
|
||
- agx: quiesche assert with r1l preload
|
||
- agx: fix corner case scratch allocation
|
||
- agx: don't reserve regs if we won't use them
|
||
- agx: fix shuffling with large reserved region
|
||
- agx: print pack assertions to stderr
|
||
- agx: don't propagate constants from trivial phis
|
||
- hk: bump max input/output components
|
||
- agx: expand macro
|
||
- agx: fix backwards copyprop with phis
|
||
- agx: add unit test for tricky backwards phi case
|
||
- hk: fix tess with notess
|
||
- hk: drop dead
|
||
- hk: flush with events
|
||
- agx: gimme c23
|
||
- agx: fix corner with uniform source lowering
|
||
- agx: use reserved_size more
|
||
- agx: commonize reg_to_ssa update
|
||
- agx: factor out insert_copy
|
||
- agx: add another RA torture mode
|
||
- agx: fix shuffling with partial clobbered vector
|
||
- agx: add reg to agx_index
|
||
- agx: reset kill bits in liveness
|
||
- agx: strengthen assert
|
||
- agx: assert a subtle invariant
|
||
- agx: produce regful SSA in RA
|
||
- agx: lower swaps late
|
||
- agx: commonize ra_class
|
||
- agx: fix shuffles of vectors
|
||
- agx: validate RA
|
||
- hk: fix shadowing
|
||
- hk: fix partial EOT program
|
||
- hk: fix ZLS across split CRs
|
||
- hk: add helper to dump draw
|
||
- agx: fix load_first_vertex
|
||
- hk: split out HK_PERFTEST
|
||
- hk: add nobarrier test
|
||
- hk: add =batch perftest
|
||
- hk: add norobust test
|
||
- hk: parametrize sysmem heap fraction
|
||
- hk: reduce heap size
|
||
- asahi: optimize !robustness with soft fault
|
||
- asahi: optimize !robust VBO
|
||
- hk: optimize !robust VBO
|
||
- hk: use vk_meta for compressed MSAA copies
|
||
- hk: fix eMRT decompress of msaa
|
||
- hk: end pregfx even if there's no gfx
|
||
- hk: fix emrt compressed msaa arrays
|
||
- hk: enable compressed MSAA
|
||
- hk: reformat driconf
|
||
- util,hk: add border colour driconf
|
||
- hk: use real opaque black with integers
|
||
- hk: drop #include
|
||
- hk: perf debug resolves
|
||
- hk: perf debug some meta
|
||
- hk: Reset rendering attachment locations in BeginCommandBuffer/Rendering
|
||
- asahi,hk: implement remapping in epilog
|
||
- hk: advertise KHR_dynamic_rendering_local_read
|
||
- asahi: rm include
|
||
- agx: inline texture states into EOT
|
||
- hk: fix store_op_none
|
||
- hk: support geom/tess subgroups
|
||
- hk: drop FSR todo
|
||
- hk: check arch in pipeline cache
|
||
- hk: fail probing silently
|
||
- hk: drop gunk
|
||
- nir: extract round component helper
|
||
- nir/opt_algebraic: optimize patterns hit with OpenCL
|
||
- util/ralloc: add total_size helper
|
||
- nir: correct comment
|
||
- nir/serialize: fix name
|
||
- asahi: split out genxml header
|
||
- asahi: fix spacing
|
||
- asahi: assume compiler is not stupid
|
||
- asahi: normalize enum formatting
|
||
- asahi: migrate f-strings
|
||
- asahi: fix indentation
|
||
- asahi: cleanup validation
|
||
- asahi: cleanup generate valid
|
||
- asahi: drop extra new line
|
||
- asahi: guard some apple code
|
||
- asahi: guard darwin stuff
|
||
- asahi: fix newlines
|
||
- asahi: add exact fields to genxml
|
||
- asahi: convert to exact
|
||
- asahi: drop trailing
|
||
- asahi: be explicit about LOD clamps
|
||
- asahi: drop zero defaults
|
||
- asahi: drop macOS defaults
|
||
- asahi: drop PBE default
|
||
- asahi: drop fragment face 2 defaults
|
||
- asahi: set exact on some fixed? unknown fields
|
||
- asahi: drop comment
|
||
- asahi: drop comment about blob weirdness
|
||
- asahi: drop dated comments
|
||
- asahi: idetnify Clamp W
|
||
- asahi: set clamp W? explicitly
|
||
- asahi: drop unused default
|
||
- asahi: be explicit about usc reg unk
|
||
- asahi: be explicit about frag properties
|
||
- asahi: automatically set default to encode to zero
|
||
- asahi: drop implied defaults
|
||
- asahi: get rid of default from the XML
|
||
- asahi: make wrapping less silly
|
||
- asahi: don't do extra runtime validation for exact
|
||
- asahi: defer fixed->float to tess
|
||
- asahi: rm tessellator branch
|
||
- asahi,agx: clang-format
|
||
- agx: split up f16 ops
|
||
- agx: special case mov_imm
|
||
- agx: drop encoding_16
|
||
- agx: encoding_32 -> encoding
|
||
- agx: consistent ffma name
|
||
- asahi: presumably fix indirect indexed tess
|
||
- libagx: extend indirect tess to handle indexed
|
||
- hk: plumb indirect_local
|
||
- hk: add mechanism to test indirects
|
||
- hk: wire up indirect tess
|
||
- hk: drop store_op_dontcare w/a
|
||
- hk: switch to 64-bit queries
|
||
- hk: use push size macro
|
||
- hk: bump max push size
|
||
- asahi: fix no16 flag
|
||
- agx: fix uniform packing with local_load
|
||
- agx: pack agx_cf_binding
|
||
- agx: move binary_size into info
|
||
- hk: fix alpha-to-coverage with sample shading
|
||
- hk: allow tess modes in either stage
|
||
- hk: make tess partitioning dynamic
|
||
- libagx: fix missing statics
|
||
- libagx: don't rely on loop unroll in txs
|
||
- libagx: drop generated VDM tess path (for now?)
|
||
- libagx: make points mode dynamic
|
||
- libagx: always tessellate clockwise
|
||
- asahi/clc: strip nir
|
||
- asahi: drop unused
|
||
- asahi: pad blit key
|
||
- asahi: use indirect ptr helper more
|
||
- asahi: add the BO even for internal dispatches
|
||
- asahi: fix indirect buffer UAF
|
||
- asahi: fix extremely subtle UAF
|
||
- asahi: assert guard previously-subtle code
|
||
- asahi: don't leak blit shaders
|
||
- asahi: don't leak binaries
|
||
- asahi: don't leak linked shaders
|
||
- asahi: don't leak drm version
|
||
- nir,agx: lower fmin/fmax in NIR
|
||
- agx: optimize wait_pix a bit
|
||
- agx: negate iadd/imsub constants
|
||
- agx: test immediate packing opt
|
||
- agx: run algebraic later
|
||
- asahi,agx: rework constant promotion upload
|
||
- agx: don't upload constant padding at the start
|
||
- agx: fold more inots
|
||
- hk: enable constant promotion
|
||
- nir/divergence_analysis: add AGX atomics
|
||
- clc: add mesa_clc tool
|
||
- asahi/clc: ingest spir-v
|
||
- asahi: fix a2c with sample shading, harder
|
||
|
||
Amber (4):
|
||
|
||
- ir3: add encoding support for 64-bit atomics introduced in a7xx.
|
||
- ir3: add support for 64 bit atomics
|
||
- nir: add pass to lower atomic arithmetic to a loop with cmpxchg.
|
||
- tu, ir3: Implement VK_KHR_shader_atomic_int64 for a7xx.
|
||
|
||
Andrew Woloszyn (2):
|
||
|
||
- Added some fixes for zircon tests in the Android CTS.
|
||
- Fix for dynamic rendering.
|
||
|
||
Anil Hiranniah (1):
|
||
|
||
- panfrost: Fix a memory leak in the CSF backend
|
||
|
||
Antonio Ospite (2):
|
||
|
||
- android: simplify building libgallium_dri on Android
|
||
- android: fix build by removing references to dri-search-path
|
||
|
||
Asahi Lina (9):
|
||
|
||
- asahi: Add 1queue debug option
|
||
- asahi: Print queue ID with batch debug
|
||
- asahi: Add more batch debug
|
||
- asahi: Fix non-async flush
|
||
- asahi: Relax cross-context flush
|
||
- asahi: Update to UAPI 10011
|
||
- asahi: Do not sync against our own queue
|
||
- agx: Fix implicit sync for virtgpu
|
||
- agx: Fix queue destroy op for virtgpu
|
||
|
||
Assadian, Navid (8):
|
||
|
||
- amd/vpelib: Change Max DS support to 4:1
|
||
- amd/vpelib: Add API function to get taps
|
||
- amd/vpelib: Fix CS translation for geometric downscaling
|
||
- amd/vpelib: Fix whitepoint for geometric downscaling
|
||
- amd/vpelib: Add 420 semi-planar 12bit handling
|
||
- amd/vpelib: Document public API structures
|
||
- amd/vpelib: Add better rotation and mirror capability check
|
||
- amd/vpelib: Add mirror & rotation capability check function for vpe1.1
|
||
|
||
Bas Nieuwenhuizen (1):
|
||
|
||
- radv: Disable EXT BDA capture and replay.
|
||
|
||
Benjamin Cheng (1):
|
||
|
||
- radv/video: add event support for VCN4
|
||
|
||
Benjamin Herrenschmidt (1):
|
||
|
||
- dril: Fixup order of pixel formats in drilConfigs
|
||
|
||
Benjamin Lee (6):
|
||
|
||
- nvk: implement vkCmdDispatchIndirect on pre-turing
|
||
- panvk: don't dirty preloaded tiles for writeback
|
||
- panvk: fix combined image/sampler descriptor arrays
|
||
- panvk: allow resuming secondary cmdbufs with dynamic rendering
|
||
- panvk: flush sync point before executing secondary cmdbufs
|
||
- panvk: hack to improve depth clipping with small viewport depth range
|
||
|
||
Benjamin Otte (2):
|
||
|
||
- nvk: Don't emit critical messages during init
|
||
- pvr: Don't emit critical warning all the time
|
||
|
||
Bernhard Rosenkränzer (1):
|
||
|
||
- dril: Fix return types of peglInitialize and peglTerminate
|
||
|
||
Bo Hu (2):
|
||
|
||
- goldfish: handle null handle
|
||
- gfxstream: global_state_wrapped_decoding of vkCreateComputePipelines
|
||
|
||
Boris Brezillon (157):
|
||
|
||
- panvk: Adjust RGB component order for fixed-function blending
|
||
- util/hexdump: Squeeze repeated patterns instead of zero-ed chunks
|
||
- panvk: Make sure we emit blend descriptors even if no FS shader is provided
|
||
- pan/genxml: Fix Resource descriptor alignment
|
||
- pan/decode: Select the right registers when dumping local storage descriptors
|
||
- pan/decode: Interpret LOAD_MULTIPLE opcode when decoding a command stream
|
||
- pan/cs: Allow lazy root chunk allocation
|
||
- pan/cs: Provide cs_root_chunk_{gpu_addr,size}() and cs_is_empty() helpers
|
||
- pan/shader: Fix FS varyings.input_count initialization on v9+
|
||
- pan/bi: Support native layer_id store/load on Valhall
|
||
- pan/tex: Allow format reinterpretation on v9+
|
||
- vk/sync: Use the proper type in vk_filter_{src,dst}_access_flags2()
|
||
- nir/format_convert: Promote input to 32-bit before packing integers
|
||
- vk/format: Add missing R10G10B10A2_SINT entry in the pipe -> vk table
|
||
- vk/image: Add a vk_image_buffer_range() helper
|
||
- vk/image: Fix the view extent of uncompressed views of compressed images
|
||
- vk/meta: Let meta users pass the color attachment write mask explicitly
|
||
- panvk: Don't spam the error output if the device is not for us
|
||
- panvk: Rename the error labels in create_device()
|
||
- panvk: Add missing vk_device_finish() calls
|
||
- pan/texture: Allow uncompressed view of compressed image
|
||
- panvk: Rework the NULL test in panvk_buffer_{gpu_ptr,range}()
|
||
- panvk: Constify the shader pointers in the graphics state
|
||
- panvk: Add helpers to save/restore the compute/graphics state for meta operations
|
||
- panvk: Rework the modifier selection
|
||
- panvk: Switch to vk_meta
|
||
- panvk: Lower load_multisampled_pan
|
||
- panvk: Add support for render-pass resolve operations
|
||
- panvk: Fix depth-stencil AFBC support
|
||
- panvk: Allow compressed formats
|
||
- panvk: Allow ZS formats on 1D/3D images
|
||
- panvk: Lower var copies before lowering IOs
|
||
- panvk/ci: Enable clearing tests
|
||
- panvk: Prepare the blend logic for Valhall
|
||
- panvk: Collect allocated push sets at the command level
|
||
- panvk: Prepare panvk_cmd_pool for CSF
|
||
- panvk: Make panvk_queue_finish() a per-arch function
|
||
- panvk: Allow more descriptor sets on Valhall
|
||
- panvk: Move panvk_debug_adjust_bo_flags() to panvk_device.h
|
||
- panvk: Conditionally register an host address when tracking user memory
|
||
- panvk: Prepare the cmd_desc_state logic for Valhall
|
||
- panvk: Prepare cmd_meta for Valhall
|
||
- pan/cs: Make sure we're not passed cs_now() to asynchronous instructions
|
||
- pan/cs: Accept val=undef when cs_branch_label() condition is ALWAYS
|
||
- pan/cs: Add helpers to emit match (AKA switch-case) blocks
|
||
- pan/cs: Allow sparse register set passed to loads/stores
|
||
- pan/cs: Add an optional load/store tracker to make sure we don't omit waits
|
||
- pan/cs: Add inline specifiers to a bunch of cs\_ helpers
|
||
- pan/cs: Add a facility to restrict register access
|
||
- pan/kmod: Don't cap VM bind operations to one
|
||
- pan/desc: Add a struct for valhall/bifrost to the union in pan_tiler_context
|
||
- pan/desc: Make vertex_count a Midgard field
|
||
- panvk: Fix the varying pool label
|
||
- panvk: Kill the set_dirty() macro
|
||
- panvk: Drop the NO_MMAP flag when PANVK_DEBUG=trace
|
||
- panvk: Fix valhall_pack_buf_idx()
|
||
- panvk: Don't open-code vk_index_type_to_bytes()
|
||
- panvk: Don't make panvk_device size gen dependent
|
||
- panvk: Make sure push sets are reset when picked from the CommandPool
|
||
- panvk: Prepare the panvk_device logic for CSF
|
||
- panvk: Set the submit mode to THREADED_ON_DEMAND
|
||
- panvk: Move Cmd{Bind,Push}DescriptorSets implementation to cmd_desc_state.c
|
||
- panvk: Move cmd_bind_shaders() to panvk_vX_shader.c
|
||
- panvk: Move CmdPushConstants2KHR to panvk_vX_cmd_push_constant.c
|
||
- panvk: Be robust against allocation failures in a command buffer context
|
||
- panvk: Don't pretend we can compile for v9
|
||
- panvk: Transition to explicit VA assignment on v10+
|
||
- panvk: Add a read-write non-cached memory pool for CSF events
|
||
- panvk: Add a CSF backend for panvk_queue/cmd_buffer
|
||
- panvk: Advertize v10 support
|
||
- panvk/csf: Fix missing TSD pointer initialization on RUN_IDVS
|
||
- panvk/csf: Fix inverted address/destination register in CmdSetEvent2()
|
||
- pan/va: Accept DIM_SUBPASS[_MS] in valhall_tex_dimension()
|
||
- panvk: Temporarily disable VK_FORMAT_D32_SFLOAT_S8_UINT on Valhall
|
||
- panvk: Fix valhall_pack_buf_idx() for SSBO store/atomic operations
|
||
- panfrost: Emit an SSBO table on v9+
|
||
- nir/lower_ssbo: Add an option to conditionally lower loads
|
||
- nir/lower_ssbo: Extend the load_ssbo_address intrinsic to pass an offset
|
||
- panfrost: Pass the architecture to panfrost_nir_lower_sysvals()
|
||
- pan/va: Optimize SSBO lowering on Valhall
|
||
- panvk/csf: Unconditionally clean L2 and LS caches when closing a CS
|
||
- pan/cs: Fix buffer overflow in cs_block_end()
|
||
- pan/cs: Make sure we don't overflow the chunk capacity
|
||
- pan/cs: Make sure cs_alloc_ins() is never passed num_instrs=0
|
||
- pan/cs: Make the CS loop helpers more robust
|
||
- pan/cs: Pass the block to end to cs_block_end()
|
||
- pan/cs: Drop the cs_builder::blocks::cur field
|
||
- pan/cs: Don't use a list for our block stack
|
||
- pan/cs: Add helpers for if/else blocks
|
||
- panvk/csf: Replace a cs_while() by a cs_if()
|
||
- panvk/csf: Pass a non-zero flush-id to benefit from cache flush reduction
|
||
- panvk/csf: Fix TLS pointer copy
|
||
- panvk: Use nir_metadata_control_flow
|
||
- panvk: Lower get_ssbo_size() on Valhall
|
||
- panvk: Pass the robustness state to nir_lower_descriptors()
|
||
- panvk: Hook up descriptor array bounds checking
|
||
- vk/meta: Make sure texel is 32-bit in build_buffer_to_image_cs()
|
||
- panfrost: Don't turn 3D/cube images into 2D arrays
|
||
- pan/decode: Interpret CS_BRANCH instructions
|
||
- pan/desc: Add layer_offset field to pan_tiler_context::valhall
|
||
- pan/genxml: Fix layer_offset definition on v9+
|
||
- panvk/csf: Fix multi-layer rendering
|
||
- panvk: Protect access to the virtual address heap
|
||
- pan/va: Fix nir_op_pack_uvec4_to_uint
|
||
- pan/genxml: Remove surface descriptors on v6/v7
|
||
- pan/genxml: Fix surface descriptor alignment on v6
|
||
- pan/genxml: Fix resource table alignment constraint on v9
|
||
- pan/kmod: Try to use local storage in panthor_kmod_vm_bind()
|
||
- panvk: Don't reset errno in panvk_kmod_zalloc()
|
||
- panvk: Stop including pan_blend.h from panvk_device.h
|
||
- panvk: Use the proper alignment for texture payloads
|
||
- panvk: Check allocation failures in panvk_shader_upload()
|
||
- panvk: Report allocation failures in the CreateImageView path
|
||
- panvk: Fix FB initialization when using non-compressed views of compressed images
|
||
- panvk: Fix stencil attachment resolve on JM
|
||
- panvk/v10: Add missing cleanup_queue() in queue_finish()
|
||
- panvk: Don't call queue_finish() on non-initialized queues
|
||
- panvk: Don't leak vertex shader program descriptors
|
||
- panvk: Let panvk_priv_bo_create() return a VkResult
|
||
- panvk: Catch indirect allocation failures
|
||
- panvk: Make sure RUN_IDVS has a valid tiler descriptor
|
||
- panvk: Fix bounds checking in upload_shader_desc_info()
|
||
- panvk/csf: Fix prepare_vp() when viewport->x is negative
|
||
- panvk: Advertise KHR_create_renderpass2 support
|
||
- vk/meta: Fix buffer -> image copy using a graphics pipeline
|
||
- zink: Adjust the surface size of non-compressed views
|
||
- vk/image: Fix the extent adjustment of non-compressed views
|
||
- vk/meta: Support VkShaderExt objects to allow tracking internal shaders
|
||
- vk/meta: Allow tracking of driver-specific objects in the meta list
|
||
- vk/meta: Make some helpers public
|
||
- panvk: Add a debug flag to force image copies through the gfx pipeline
|
||
- panvk: Add an helper to create internal shaders
|
||
- panvk: Keep our blend shaders in vk_meta_device
|
||
- panvk: Keep our copy_desc shader in vk_meta_device
|
||
- panvk: Emit textures needed for FB preload at image view creation time
|
||
- panvk: Store attachment image views in the graphics state
|
||
- panvk: Implement a custom FB preload logic
|
||
- panfrost: Move pan_blitter.{c,h} to the gallium driver
|
||
- panvk: Make panvk_pool_free_mem() error proof
|
||
- panvk/csf: Fix no-fragment IDVS
|
||
- panvk/csf: Fix dirty checking in prepare_ds()
|
||
- panvk/csf: Fix cmd_emit_dcd() in the FB preload logic
|
||
- panvk/csf: Fix a buffer/stack-overflow when PANVK_DEBUG=sync
|
||
- panvk: Advertise dynamic rendering support
|
||
- panvk/csf: Fix the clear-only RUN_FRAGMENT case
|
||
- pan/util: Discard depth/stencil writes when early fragment test is forced
|
||
- panvk: Set .is_blit=true when compiling preload shaders
|
||
- panvk: Fix the ZStencil descriptor initialization in the preload logic
|
||
- panvk: Fix the hierarchy_mask selection
|
||
- panvk: Fix the maximum sampler LOD bias
|
||
- pan/va: Don't set bit 26:27 on texture instructions
|
||
- pan/va: Add support for nir_texop_lod
|
||
- pan/bi: Add support for nir_texop_lod
|
||
- pan/desc: Fix the CRC validity update when clean tiles are not written
|
||
- pan/cs: Fix lazy allocation support
|
||
- pan/bi: Expose the packed TextureOperationDescriptor in bifrost_texture_operation
|
||
- panvk/csf: Fix a wait-LS operation in finish_cs()
|
||
|
||
Boyuan Zhang (3):
|
||
|
||
- frontends/va: get cu_qp_delta from vaapi
|
||
- radeonsi/vcn/enc:enable hevc cu_qp_delta for vcn2+
|
||
- radeonsi/vcn: fix loop filter across slices
|
||
|
||
Brendan (5):
|
||
|
||
- amd/vpelib: Create virtual stream concept
|
||
- amd/vpelib: Multiple instance support in caching framework
|
||
- amd/vpelib: Update configs to be per-pipe specific
|
||
- amd/vpelib: Visual confirm on input
|
||
- amd/vpelib: Fix duplicate config writer bug
|
||
|
||
Caio Oliveira (91):
|
||
|
||
- intel/brw: Simplify \@file annotations
|
||
- intel/brw: Move remaining compile stages to their own files
|
||
- intel/brw: Move and reduce scope of run_*() functions
|
||
- intel/brw: Reduce scope of some VS specific functions
|
||
- intel/brw: Reduce scope of some FS specific functions
|
||
- intel/brw: Reduce scope of some GS specific functions
|
||
- intel/brw: Reduce scope of some TCS specific functions
|
||
- intel/brw: Reduce scope of some TES specific functions
|
||
- intel/brw: Reduce scope of some MESH specific functions
|
||
- intel/brw: Move printing functions to its own file
|
||
- intel/brw: Move out of fs_visitor and rename print instructions
|
||
- intel/brw: Remove workgroup_size() helper from fs_visitor
|
||
- intel/brw: Move regalloc and scheduling functions out of fs_visitor
|
||
- intel/brw: Move calculate_cfg out of fs_visitor
|
||
- intel/brw: Move VARYING_PULL_CONSTANT_LOAD from fs_visitor to fs_builder
|
||
- intel/brw: Move interp_reg and per_primitive_reg out of fs_visitor
|
||
- intel: Let compiler set indirect_ubos_use_sampler
|
||
- intel/brw: Don't retype load_subgroup_invocation result to signed
|
||
- intel/brw: Use CSE for LOAD_SUBGROUP_INVOCATION
|
||
- meson: Update Lua dependency version to 5.3
|
||
- intel/brw: Also return the size of the assembled shader
|
||
- intel: Add executor tool
|
||
- intel/executor: Use PRIx64 to fix building in 32-bit
|
||
- intel/brw: Use %td format for pointer difference
|
||
- meson: Only use fallback for Lua if building freedreno tools
|
||
- intel/brw: Avoid modifying the shader in assign_curb_setup if not needed
|
||
- intel/brw: Remove unused variable
|
||
- intel/brw: Use a helper for common VEC pattern
|
||
- intel/brw: Remove more uses of variable length arrays
|
||
- intel/brw: Move encoding of Gfx11 3-src inside the inst helpers
|
||
- intel/brw: Explicitly map brw_reg_file into hardware values
|
||
- intel/brw: Make BAD_FILE the zero value for brw_reg_file
|
||
- intel/brw: Remove unused ARF values and helpers
|
||
- intel/brw: Remove long register file names
|
||
- intel/brw: Pack brw_reg struct
|
||
- intel/brw: Simplify fs_inst annotation
|
||
- intel/brw: Remove special treatment for 2-src in emit() helper
|
||
- intel/brw: Use linear walk when shader requires DERIVATIVE_GROUP_LINEAR
|
||
- intel/executor: Fix SWSB for sync.nop
|
||
- compiler: Allow derivative_group to be used for all stages in shader_info
|
||
- vk/runtime: Allow "require full subgroups" for Mesh/Task
|
||
- intel/executor: Small fixes to the help message
|
||
- spirv: Rename vtn_type::deref to vtn_type::pointed
|
||
- spirv: Remove (pointed) type from vtn_pointer
|
||
- spirv: Rename vtn_pointer::ptr_type to vtn_pointer::type
|
||
- vulkan: Update XML and headers to 1.3.295
|
||
- spirv: Update headers and metadata from latest Khronos commit
|
||
- spirv: Prefer symbols from Khronos extension for compute shader derivatives
|
||
- nir: Allow Mesh/Task to use implicit LOD when DERIVATIVE_GROUP is set
|
||
- spirv: Allow Mesh/Task to use derivative execution modes
|
||
- anv: Advertise VK_KHR_compute_shader_derivatives
|
||
- intel/executor: Fix a couple of memory leaks in the tool
|
||
- intel/brw: Only validate GRF boundary crossing restriction for GRFs
|
||
- intel/brw: Consider pipe when comparing SWSB in tests
|
||
- intel/brw: Add SWSB MATH pipe to assembler
|
||
- intel/brw: Add DUMP flag to brw_assemble
|
||
- intel/executor: Dump both pre-processed source and assembly
|
||
- intel/brw: Skip per-primitive inputs when computing flat input mask
|
||
- intel/brw: Remove Gfx8- from test_eu_validate.c
|
||
- intel/brw: Update DPAS validation tests for Xe2
|
||
- intel/brw: Enable EU validation and compaction tests for Xe2
|
||
- intel/brw: Enable mi_builder test for Xe2
|
||
- spirv: Update SPIR-V grammar to use aliases
|
||
- intel/brw: Add ERROR helper variant that returns to EU validation
|
||
- intel/brw: Create a struct to hold a decoded brw_inst in eu_validation
|
||
- intel/brw: Add opcode to brw_hw_decoded_inst
|
||
- intel/brw: Add num_sources to brw_hw_decoded_inst
|
||
- intel/brw: Add has_dst to brw_hw_decoded_inst
|
||
- intel/brw: Add exec_size to brw_hw_decoded_inst
|
||
- intel/brw: Add access_mode to brw_hw_decoded_inst
|
||
- intel/brw: Add dst/srcs reg file to brw_hw_decoded_inst
|
||
- intel/brw: Add dst/srcs type to brw_hw_decoded_inst
|
||
- intel/brw: Add pred/cmod/sat to brw_hw_decoded_inst
|
||
- intel/brw: Add dst/src0 address_mode to brw_hw_decoded_inst
|
||
- intel/brw: Add abs/negate to brw_hw_decoded_inst
|
||
- intel/brw: Add dst/srcs register numbers to brw_hw_decoded_inst
|
||
- intel/brw: Add vstride/width/hstride to brw_hw_decoded_inst
|
||
- intel/brw: Remove the remaining DO_SRC macro from EU validation
|
||
- intel/compiler: Rename shader_stats structs
|
||
- intel: Add statistic for Non SSA registers after NIR to BRW
|
||
- intel/brw: Print Non-SSA regs after NIR in debug output
|
||
- intel/brw: Fix Gfx9 3-src validation to handle FIXED_GRF
|
||
- intel/brw: Add phases to backend
|
||
- intel/brw: Validate some instructions exists only up until some phases
|
||
- intel/brw: Reduce scope of has_source_and_destination_hazard
|
||
- intel/brw: Add SHADER_OPCODE_REDUCE
|
||
- intel/brw: Add SHADER_OPCODE_*_SCAN
|
||
- intel/brw: Move emit_scan/emit_scan_step near its usage
|
||
- intel/brw: Add fs_builder::BROADCAST() helper
|
||
- intel/brw: Ensure BROADCAST() value respect register alignment
|
||
- intel/brw: Add SHADER_OPCODE_VOTE_*
|
||
|
||
Carlos Santa (2):
|
||
|
||
- intel/hang_replay: fix the typo in the ioctl call
|
||
- intel/hang_replay: remove EXEC_OBJECT_WRITE
|
||
|
||
Casey Bowman (5):
|
||
|
||
- vulkan/screenshot-layer: Add Vulkan screenshot layer
|
||
- vulkan/screenshot-layer: Fix image index selection
|
||
- vulkan/screenshot-layer: Increase buffer sizes
|
||
- vulkan/screenshot-layer: Fix memory leaks
|
||
- vulkan/screenshot-layer: Remove extraneous malloc operations
|
||
|
||
Chan, Roy (2):
|
||
|
||
- amd/vpelib: Fix compilation warnings
|
||
- amd/vpelib: Revise the config sharing handling
|
||
|
||
Chang, Tomson (2):
|
||
|
||
- amd/vpelib: Add and fix collaborate sync data
|
||
- amd/vpelib: Fix color fill performance issue on VPE1.1 (#419)
|
||
|
||
Charmaine Lee (1):
|
||
|
||
- svga: sync up with the latest svga include files
|
||
|
||
Chenyu Chen (1):
|
||
|
||
- amd/vpelib: Remove unused define macro
|
||
|
||
Chia-I Wu (32):
|
||
|
||
- panvk: remove unused panvk_physical_device::master_fd
|
||
- panvk: refactor panthor_kmod_dev creation
|
||
- panvk: query device ids of drm device nodes
|
||
- panvk: advertise VK_EXT_physical_device_drm
|
||
- panvk: fill in VkExternalBufferProperties
|
||
- panvk: fill in VkExternalImageFormatProperties
|
||
- panvk: remove panvk_GetPhysicalDeviceImageFormatProperties
|
||
- panvk: advertise VK_KHR_external_memory_fd
|
||
- panvk: advertise VK_EXT_external_memory_dma_buf
|
||
- panvk: advertise VK_KHR_external_{fence,semaphore}_fd
|
||
- panvk: remove panvk_GetImageDrmFormatModifierPropertiesEXT
|
||
- panvk: fix VkDrmFormatModifierPropertiesListEXT query
|
||
- panvk: check VkPhysicalDeviceImageDrmFormatModifierInfoEXT
|
||
- panvk: advertise VK_KHR_image_format_list
|
||
- panvk: advertise VK_EXT_image_drm_format_modifier
|
||
- panvk: remove unused is_sampler and is_texture
|
||
- panvk: fix descriptor set layout hash
|
||
- panvk: fix texture size/levels/samples query
|
||
- panvk: fix gl_VertexIndex
|
||
- panvk: fix scissor box
|
||
- panvk: disable depth write when depth test is disabled
|
||
- panvk: remove an incorrect assert in collect_cs_deps
|
||
- Revert "panvk: Set the submit mode to THREADED_ON_DEMAND"
|
||
- panvk/jm: assert that the submit mode is not threaded
|
||
- panvk: add support for VK_KHR_timeline_semaphore
|
||
- panvk: fix truncated access flags in collect_cs_deps
|
||
- panvk: expand meta stage and access flags
|
||
- panvk: use force_fb_preload for unaligned preload
|
||
- panvk: force_fb_preload should insert a barrier
|
||
- panvk: fix missing same-subqueue wait for CmdWaitEvents2
|
||
- panvk: fix dummy sampler handle for vs
|
||
- panvk: ensure res table is restored after meta
|
||
|
||
Christian Gmeiner (70):
|
||
|
||
- etnaviv: isa: left shift is 3 bit long
|
||
- dri: fix driver names
|
||
- etnaviv: Rework uniform handling for UBO addresses
|
||
- meson: Add missing inc's to idep_nir_headers
|
||
- nir_lower_mem_access_bit_sizes: Support load_kernel_input
|
||
- ci: austriancoder farm maintenance
|
||
- nak: Move bitset to compiler crate
|
||
- nak: Move cfg to compiler crate
|
||
- etnaviv: Move halti determination to drm
|
||
- etnaviv: Switch etna_compiler_create(..) to etna_core_info
|
||
- etnaviv: Extend shader structs with etna_core_info
|
||
- etnaviv: Drop halti from etna_specs
|
||
- etnaviv: isa: Add conv instruction
|
||
- etnaviv: Drop has_sin_cos_sqrt and has_sign_floor_ceil
|
||
- etnaviv: Switch to shader_core_count from etna_core_info
|
||
- etnaviv: Switch to vertex_cache_size from etna_core_info
|
||
- etnaviv: Switch to vertex_output_buffer_size from etna_core_info
|
||
- etnaviv: npu: Switch to use etna_core_info
|
||
- etnaviv: npu: Drop not used spec values
|
||
- etnaviv: Switch to num_constants from etna_core_info
|
||
- etnaviv: Switch to max_registers from etna_core_info
|
||
- etnaviv: Switch to stream_count from etna_core_info
|
||
- etnaviv: Switch to etna_core_has_feature(..) for npot_tex_any_wrap
|
||
- etnaviv: Switch to etna_core_has_feature(..) for has_halti2_instructions
|
||
- etnaviv: Drop assignment of get_timestamp
|
||
- broadcom/simulator: Add support for asahi based systems
|
||
- compiler/rust: Copy MappedInstrs from NAK
|
||
- etnaviv: Update headers from rnndb
|
||
- etnaviv: nir: Make use of nir_lower_fragcolor
|
||
- etnaviv: Report number of render targets
|
||
- etnaviv: nir: Make use of reported number of render targets
|
||
- etnaviv: Call etna_resource_level_mark_changed(..) for all rts
|
||
- etnaviv: Extend frag_rb_swap for MRTs
|
||
- etnaviv: Extend compiled_shader_state for MRTs
|
||
- etnaviv: Extend compiler for MRTs
|
||
- etnaviv: Extend etna_record_flush_resources(..) for MRTs
|
||
- etnaviv: Extend etna_update_blend_color(..) for MRTs
|
||
- etnaviv: Extend etna_update_zsa(..) for MRTs
|
||
- etnaviv: Extend etna_set_framebuffer_state(..) for MRTs
|
||
- etnaviv: Extend etna_update_ts_config(..) for MRTs
|
||
- etnaviv: Emit MRT states
|
||
- etnaviv: Emit PS_OUTPUT_REG2 if num_rt is 8
|
||
- etnaviv: blend: Store information per render target
|
||
- etnaviv: blend: Move logicop_enable assignment
|
||
- etnaviv: blend: Add support for MRTs
|
||
- etnaviv: blt/rs: Just pass the color to the callee
|
||
- etnaviv: rs: Extend to support MRTs
|
||
- etnaviv: blt: Extend to support MRTs
|
||
- etnaviv: shader: Apply output remapping
|
||
- etnaviv: shader: Apply output saturation
|
||
- etnaviv: Relink shader when framebuffer state changed
|
||
- etnaviv: TS usage for MRT needs HALTI2
|
||
- etnaviv: Limit max supported render targets
|
||
- etnaviv: Report correct PIPE_CAP_MAX_RENDER_TARGETS value
|
||
- etnaviv: Enable EXT_draw_buffers2 for HALTI5+
|
||
- etnaviv: Enable ARB_draw_buffers_blend for HALTI5+
|
||
- etnaviv: Improve split sampler check
|
||
- etnaviv: blt: Convert assert to real if
|
||
- etnaviv: Implement ARB_draw_indirect
|
||
- v3d: Add a few function traces
|
||
- vc4: Add a few function traces
|
||
- docs: Update the list of drivers with CPU tracepoints
|
||
- etnaviv: nir: Enforce stricter swizzle for virtual scalar x register
|
||
- etnaviv: nir: Switch to derivative intrinsics
|
||
- compiler/rust: Copy NirInstrPrinter from NAK
|
||
- etnaviv: isa: Add clamp0_max instruction
|
||
- etnaviv: isa: Add img_load instruction
|
||
- etnaviv: Enable ARB_framebuffer_no_attachments
|
||
- etnaviv: isa: Add img_store instruction
|
||
- etnaviv: Fix incorrect pipe_nn creation
|
||
|
||
Colin Marc (2):
|
||
|
||
- vulkan/video: set HEVC nuh_temporal_id_plus1 and nal_unit_type correctly
|
||
- radv/video: set TemporalId correctly
|
||
|
||
Collabora's Gfx CI Team (10):
|
||
|
||
- Uprev Piglit to 4a62c26721a47552a96416a134b789a813dd51a6
|
||
- Uprev Piglit to 0453436872b6e4d502c2e87817addb95e0d77e3b
|
||
- Uprev Piglit to f11abb664bfcad09586f32f411b90331e23be2e5
|
||
- Uprev Piglit to 1d8fab9b49cc487ebcc949119eeb166fb3fa7591
|
||
- Uprev Piglit to 6533b66878c8cb9361c554b7ff940518ab12c816
|
||
- Uprev Piglit to a3826de3c26a279599d15b018a9a3e75ca46f4f8
|
||
- Uprev Piglit to 93b4bd2e0aaab1c22ae3e1a23f9e057a8f7451b2
|
||
- Uprev Piglit to 7ce69da1199d12ed0ddaa251ed489750523798fb
|
||
- Uprev Piglit to 791e420b2628c1e35eea81b3bafdb1c904a141e8
|
||
- Uprev Piglit to c2b31333926a6171c3c02d182b756efad7770410
|
||
|
||
Connor Abbott (60):
|
||
|
||
- egl/x11: pass render fd to screen create for zink
|
||
- freedreno/ci: Combine and document failures due to test bug
|
||
- ci: Move two failing loader-related tests to all-skips.txt
|
||
- ci: Uprev VK-CTS to 1.3.9.0
|
||
- tu: Fix off-by-one in UBO CP_LOAD_STATE size
|
||
- ir3, tu: Use a UBO for VS primitive params on a750+
|
||
- Revert "tu/a750: Disable HW binning when there is GS"
|
||
- ir3: Fix nir_op_vec8/vec16
|
||
- freedreno/computerator: Fix enabling wave128
|
||
- freedreno/a7xx: Document compute dispatch tiling registers
|
||
- freedreno, tu, ir3: Enable tiled workgroup item dispatch on a7xx
|
||
- ir3: Move reconvergence backedge handling
|
||
- tu: Fix passthrough D24S8 attachments
|
||
- tu: Treat partially-bound depth/stencil attachments as passthrough
|
||
- tu: Add a750 flush workaround and re-enable UBWC for storage images
|
||
- freedreno: Rename CP_SET_CTXSWITCH_IB to CP_SET_AMBLE
|
||
- ir3: Increase compute const size on a7xx
|
||
- tu: Initial support for VK_KHR_calibrated_timestamps on a750
|
||
- freedreno/a6xx+: Add CP_CONTEXT_SWITCH_CNTL bitfields
|
||
- ir3: Ban conversions with mismatching sizes
|
||
- Reapply "ci/lima: Temporarily disable"
|
||
- ir3, turnip: Support VK_*_compute_shader_derivatives on a7xx
|
||
- tu: Don't WFI after initializing RB_CCU_CNTL on a7xx
|
||
- tu: Split out register setting from tu6_init_hw()
|
||
- tu: Write bin size for each bin
|
||
- tu: Re-emit LRZ state before each bin
|
||
- tu: Simplify device startup CS creation
|
||
- tu: Emit CP_SET_AMBLE packets
|
||
- freedreno: Bump kernel uapi
|
||
- tu/virtio: Make virtio_simple_ioctl take a vdrm_device
|
||
- tu: Allow being preempted on a7xx
|
||
- tu: Add TU_DEBUG=hiprio
|
||
- vulkan/state: Handle NULL in DS input attachment mapping correctly
|
||
- vulkan/state: Track the input attachment count
|
||
- v3dv: Don't misuse nir_variable::data.index
|
||
- spirv: Make the default input attachment index ~0
|
||
- nir/lower_input_attachments: Handle unscaled input attachments with no index
|
||
- ir3: Fix non-bindless s2en texture/sampler order
|
||
- tu: Support color attachment remapping
|
||
- tu: Support dynamic input attachments
|
||
- tu: Make input attachments always contain a real descriptor
|
||
- tu: Fix flushes for feedback_invalidate case
|
||
- tu: Track possible feedback loops for dynamic renderpasses
|
||
- tu: Expose VK_KHR_dynamic_rendering_local_read
|
||
- ir3, tu: Plumb through support for per-shader robustness
|
||
- tu: Implement VK_EXT_pipeline_robustness
|
||
- freedreno/fdl: Extend 2bpp UBWC special case to 1bpp
|
||
- freedreno/a6xx: Remove dead fd6_get_ubwc_blockwidth() call
|
||
- freedreno/fdl: Add UBWC config struct
|
||
- freedreno: Add default UBWC config values
|
||
- virtio/drm: Update header
|
||
- tu: Gather UBWC config
|
||
- freedreno/fdl: Add tiling/untiling implementation for a6xx/a7xx
|
||
- tu: Add a flag for cached non-coherent BOs
|
||
- tu: Implement VK_EXT_host_image_copy
|
||
- freedreno: Add compute constlen quirk for X1-85
|
||
- tu: Implement VK_PIPELINE_CREATE_2_VIEW_INDEX_FROM_DEVICE_INDEX_BIT_KHR
|
||
- tu: Don't invalidate CS state for 3D blits
|
||
- ir3: Fix detection of nontrivial continues
|
||
- tu: Re-emit visibility stream before each render pass
|
||
|
||
Constantine Shablia (4):
|
||
|
||
- vk/meta: Add copy/fill/update helpers
|
||
- pan/va: Define the TEX_GRADIENT instruction
|
||
- pan/va: Lower nir_texop_txd to TEX_GRADIENT with derivs followed by TEX_SINGLE
|
||
- pan/bi: Lower nir_texop_txd to TEXC in GRDESC_DER mode followed by sampling TEXC
|
||
|
||
Corentin Noël (2):
|
||
|
||
- ci: Allow to pass the PIGLIT_RUNNER_OPTIONS variable
|
||
- virgl: Avoid a race condition on handle removal
|
||
|
||
Craig Stout (18):
|
||
|
||
- util: fix fossilize_db compile error
|
||
- util: fix 32bit compiler error in fossilize_db
|
||
- [guest] Check that the singleton VirtGpuDevice matches desired capset
|
||
- Reapply "[guest] Check that the singleton VirtGpuDevice matches desired capset"
|
||
- Fix dispatchable object layout on Fuchsia
|
||
- Exclude Android entry points on Fuchsia
|
||
- [guest] Fix Fuchsia build
|
||
- [guest] Add skeleton Fuchsia VirtGpu implementation
|
||
- [guest] Build fixes for Fuchsia
|
||
- [guest] Fuchsia: include missing Vulkan entry points
|
||
- [guest] Remove Fuchsia syslog_static dependency
|
||
- [guest] Fix Fuchsia build, add missing header
|
||
- [guest] Fuchsia: open magma device
|
||
- [guest] Fuchsia: initialize queries
|
||
- [guest] Fix Fuchsia build
|
||
- tu: add OS guards to drm_format_mod
|
||
- u_gralloc: include dep_android_mapper4 as needed
|
||
- meson: add log dependency to dep_android
|
||
|
||
Daniel Almeida (6):
|
||
|
||
- nouveau/headers: Run rustfmt on generated files
|
||
- nouveau/headers: Add a Rust struct for each method
|
||
- nouveau/headers: Add an nv_push crate in Rust
|
||
- util: memstream: add fflush support
|
||
- nak: memstream: move into common code
|
||
- util: u_memstream: add tests
|
||
|
||
Daniel Schürmann (55):
|
||
|
||
- aco/reindex_ssa: free memory of previous live variable sets
|
||
- aco: implement aco::validate_live_vars()
|
||
- aco/scheduler: enable live variables validation when ACO_DEBUG=validate-livevars is set
|
||
- aco: don't attempt to spill dead phis
|
||
- aco/cssa: fix kill flags during lowering to CSSA
|
||
- aco: change signature of get_live_changes() and get_temp_registers()
|
||
- aco/cssa: update RegisterDemand and validate live variable information
|
||
- aco/cssa: short-cut some trivial case
|
||
- aco: call nir_copy_prop() and nir_opt_dce() before instruction selection
|
||
- nir: add nir_block \*nir_src_get_block(src) helper
|
||
- nir/opt_if: optimize phis between similar IFs
|
||
- nir/opt_if: merge IFs which have phis between them
|
||
- aco/ra: add add_rename() helper
|
||
- aco/ra: avoid rename map lookups in most cases
|
||
- aco/ra: use auto for std container iterators
|
||
- aco/ra: use arena allocator for hash maps
|
||
- aco/live_var_analysis: Don't attempt to re-insert the same temporary twice into live set
|
||
- aco: introduce new Operand flag 'Clobbered'
|
||
- aco/live_var_analysis: use Clobbered flag to calculate additional operand demand
|
||
- aco: introduce Operand flag 'CopyKill'
|
||
- radv: promote VK_NV_compute_shader_derivatives -> VK_KHR_compute_shader_derivatives
|
||
- aco/live_var_analysis: don't set lateKill for p_interp_gfx11 m0 operand
|
||
- aco: simplify Definition constructors
|
||
- aco/lower_to_hw: don't allocate new temporaries
|
||
- aco/reindex_ssa: remove update_live_out parameter
|
||
- aco: remove Program::allocationId
|
||
- aco/live_var_analysis: don't test whether phis are assigned to VCC
|
||
- aco/ra: add RegisterFile::fill_killed_operands(Instruction*) helper
|
||
- aco/ra: manually fill killed operands when required
|
||
- aco/ra: remove unnecessary check for duplicate precolored operands
|
||
- aco: introduce 'isPrecolored' flag for Operand and Definition
|
||
- aco: set Precolored flag before register allocation
|
||
- aco/live_var_analysis: check for isPrecolored flag rather than isFixed
|
||
- aco/ra: make use of Precolored flag
|
||
- aco/ra: always block register file for precolored operands
|
||
- aco/live_var_analysis: check isFixed() for definitions in order to set needs_vcc
|
||
- aco/spill: fix faulty assertions
|
||
- aco/spill: use float division for score() calculation rather than integers
|
||
- nir/divergence: separately indicate whether loops have divergent continues or breaks
|
||
- nir/opt_uniform_atomics: don't preserve divergence information
|
||
- nir: remove nir_update_instr_divergence()
|
||
- nir: change signature of nir_src_is_divergent()
|
||
- treewide: use nir_src_is_divergent() rather than checking the divergence of the SSA
|
||
- nir/divergence: require nir_metadata_block_index
|
||
- nir/divergence: Introduce and set nir_def::loop_invariant
|
||
- nir: consider loop invariance in nir_src_is_divergent()
|
||
- nir/divergence: calculate divergence without requiring LCSSA form
|
||
- nir/divergence: skip expensive nir_src_is_divergent() check in most cases
|
||
- nir/divergence: Don't assume that LCSSA phis are not loop-invariant
|
||
- treewide: don't lower to LCSSA before calling nir_divergence_analysis()
|
||
- aco: Respect addressible SGPR limit in VS prologs
|
||
- aco/ssa_elimination: perform jump threading after parallelcopy insertion
|
||
- aco/ssa_elimination: remove some redundant checks during jump threading
|
||
- aco: move jump threading optimization into separate pass
|
||
- aco/README: add descriptions of recently added passes
|
||
|
||
Daniel Stone (146):
|
||
|
||
- build: Check for PyYAML in Meson build
|
||
- dri: Allow INVALID for modifier-less drivers
|
||
- gbm/dri: Remove erroneous assert
|
||
- ci: Create new 'performance' stage
|
||
- ci/amd: Move manual/nightly jobs to postmerge stage
|
||
- ci/intel: Move manual/nightly jobs to postmerge stage
|
||
- ci/broadcom: Move manual/nightly jobs to postmerge stage
|
||
- ci/zink: Move manual/nightly jobs to postmerge stage
|
||
- ci/llvmpipe: Move manual/nightly jobs to postmerge stage
|
||
- ci/nouveau: Move manual/nightly jobs to postmerge stage
|
||
- ci/etnaviv: Move manual/nightly jobs to postmerge stage
|
||
- ci/panfrost: Move manual/nightly jobs to postmerge stage
|
||
- ci/run_n_monitor: Add --include-stage argument
|
||
- ci/run_n_monitor: Add --exclude-stage filtering
|
||
- ci/run_n_monitor: Remove --force-manual
|
||
- ci/run_n_monitor: Add --dry-run argument
|
||
- ci: Make per-build dependencies optional
|
||
- doc/llvmpipe: Update URL to fix linkcheck
|
||
- ci/fdno: Add some flaky tests for a618
|
||
- ci/dozen: Add a couple more skips
|
||
- ci/windows: Don't upload shader cache to artifacts
|
||
- ci/{lvp,venus}: Skip massive-alloc tests
|
||
- intel/tests: Reduce load from anv_tests
|
||
- util/tests: Reduce load from sparse-array test
|
||
- ci/angle: Clean up stray open-coded path
|
||
- ci/lava: Fix path to structured logger
|
||
- ci/lava: Make it easier to run pytest locally
|
||
- ci/lava: Don't dump venv into .venv
|
||
- ci/lava: Allow passing more args to pytest
|
||
- ci/lava: Fix pytest not passing farm value
|
||
- ci/lava: Fix fastboot tests
|
||
- ci/alpine: Fix shellcheck errors
|
||
- ci: Fix shellcheck warnings in firmware download
|
||
- ci: Bring back yamllint and shellcheck
|
||
- ci: Run pytest once for both LAVA and bin/ci
|
||
- ci: Bring back shader-db and drop LTO
|
||
- ci/intel-gpu-freq: Fix shellcheck warning
|
||
- ci/vulkan: Backport dEQP commit to make it less OOMy
|
||
- ci/lima: Temporarily disable
|
||
- ci: Add Python requirements include
|
||
- ci: Add pyutils container
|
||
- ci/python: Allow empty PYTHONPATH
|
||
- ci/lava: Use new pyutils container
|
||
- doc/llvmpipe: Fix broken link syntax
|
||
- docs/gallium: Fix broken GL extension link
|
||
- doc/linkcheck: Exclude cgit.fd.o
|
||
- doc/vc4: Remove reference to old shader-db
|
||
- ci/windows: Increase dozen-deqp fraction
|
||
- ci/svga: Add two Piglit flakes seen in the wild
|
||
- ci/svga: Only run half the Piglit tests
|
||
- ci/virgl: Special-case llvmpipe parallelisation
|
||
- ci: Avoid subshell for executing HWCI_TEST_SCRIPT
|
||
- ci/bare-metal: Don't move structured log file
|
||
- ci: Always create results dir from init
|
||
- ci/android: Use common $RESULTS_DIR for cuttlefish
|
||
- ci/deqp: Use common $RESULTS_DIR
|
||
- ci/dxvk: Use common results dir
|
||
- ci/piglit: Use common $RESULTS_DIR
|
||
- ci/trace: Move trace cleanup to Piglit runner
|
||
- ci/gtest: Use common $RESULTS_DIR
|
||
- ci/kdl: Use common $RESULTS_DIR
|
||
- ci/devcoredump: Use common $RESULTS_DIR
|
||
- ci/xorg: Capture Xorg log in results artifacts
|
||
- ci/kdl: Fix KDL install location
|
||
- ci: Upgrade ci-kdl
|
||
- ci: Use new arguments to ci-kdl to avoid child management
|
||
- ci: Make token parsing more robust
|
||
- ci: Make common rules common
|
||
- ci: Make setup-test-env safe for set -u
|
||
- ci/piglit: Make build-piglit safe for set -u
|
||
- ci/deqp: Make build-deqp safe for set -u
|
||
- ci/apitrace: Make build-apitrace safe for set -u
|
||
- ci/va: Make build-va-tools safe for set -u
|
||
- ci/crosvm: Make build-crosvm safe for set -u
|
||
- ci/wayland: Make build-wayland safe for set -u
|
||
- ci/libdrm: Make build-libdrm safe for set -u
|
||
- ci/windows: Make build-directx-headers safe for set -u
|
||
- ci/wine: Make build-wine safe for set -u
|
||
- ci/libclc: Make build-libclc safe for set -u
|
||
- ci/vulkan: Make build-vulkan-validation safe for set -u
|
||
- ci/vkd3d: Make Proton build safe for set -u
|
||
- ci/angle: Make build-angle safe for set -u
|
||
- ci/deqp-runner: Make build-deqp-runner safe for set -u
|
||
- ci/skqp: Make build-skqp safe for set -u
|
||
- ci/skqp: Reorder some things
|
||
- ci/skqp: Don't hardcode LLVM version
|
||
- ci/kernel: Make download-prebuilt-kernel safe for set -u
|
||
- ci/swrast: Fix library paths for llvmpipe/softpipe
|
||
- ci/swrast: Archive Xvfb logs
|
||
- ci/zink: Fix library path
|
||
- ci/virgl: Fix library path
|
||
- ci/{b2c,swrast,layered}: Strip unnecessary things from runtime containers
|
||
- egl/x11: Downgrade warning to info
|
||
- ci: Disable trace jobs using old Wine
|
||
- ci: Remove non-Proton Wine
|
||
- ci/vk: Strip and optimise validation layers
|
||
- ci: Strip yet more unnecessary things from the rootfs
|
||
- ci/deqp: Flatten fraction/shard sed into a single pass
|
||
- ci/deqp: Compress caselists with zstd
|
||
- ci/angle: Update ANGLE, reduce build times
|
||
- ci: Don't carry static libraries around for tests
|
||
- ci/shellcheck: Don't overwrite SCRIPTS_DIR with relative path
|
||
- ci/shellcheck: Don't exit on first failure
|
||
- ci/python: Fix section emission
|
||
- ci: Don't move results directory
|
||
- ci: Don't print structured log data URL
|
||
- ci/lava: Remove pointless messages
|
||
- ci/lava: Quieten timestamp warnings from tar
|
||
- ci/lava: Break section-header print into separate function
|
||
- ci/lava: Add section for device wait
|
||
- ci/lava: Rename lava_boot section
|
||
- ci/lava: Hide more boot details into sections
|
||
- ci/lava: Explicitly pass UTC timezone
|
||
- ci/lava: Truncate printed times
|
||
- ci/lava: Fix colour definitions
|
||
- ci/lava: Be a little less enthusiastic with bold
|
||
- ci/lava: Change default section colour to cyan
|
||
- ci/hw: Wrap pre-test setup in collapsed section
|
||
- ci/lava: Print relative timestamps in sections
|
||
- ci/lava: Coalesce post-processed job information
|
||
- ci/bare-metal: Move preparation steps into section
|
||
- ci/bare-metal: Use section for boot messages from cros-servo
|
||
- ci/bare-metal: Truncate printed times and prefixes
|
||
- ci/bare-metal: Use section for boot messages from poe-powered
|
||
- ci/crosvm: Make crosvm-runner safe for set +u
|
||
- ci/crosvm: Replace SCRIPT_DIR with SCRIPTS_DIR
|
||
- ci/traces: Use sections to quiet execution
|
||
- ci: Quieten post-test cleanup
|
||
- ci: Make section emission really quiet
|
||
- ci: Quieten test execution
|
||
- ci/lava: Source setup-test-env
|
||
- ci/mold: Use appropriate parallelism and strip binaries
|
||
- ci: Really remove Mesa drivers from rootfs
|
||
- ci/firmware: Allow calling firmware as sourced script
|
||
- ci/angle: Use relative paths
|
||
- ci/skqp: Fix relative paths
|
||
- ci: Use sections for test container builds
|
||
- ci/skqp: Set $SKQP_ARCH from build-skqp.sh
|
||
- ci/rpi: Add flakes seen in the wild
|
||
- ci/fdno: Add flake seen in the wild
|
||
- ci/lava: Remove duplicate build download
|
||
- ci/lava: Use LAVA rootfs overlays for build/per-job
|
||
- ci/lava: Provide LAVA rootfs URL directly
|
||
- ci/lava: Provide list of overlays to submitter
|
||
- ci/lava: Refactor is-rootfs-uploaded function
|
||
- build: Don't run wayland-protocols tests
|
||
|
||
Daniel Svensson (1):
|
||
|
||
- zink: add spirv_info_h dep to libzink.
|
||
|
||
Danylo Piliaiev (20):
|
||
|
||
- tu/a750: Allow mutable images to have UBWC with all compatible formats
|
||
- tu: Enable UBWC for D24S8 with USAGE_SAMPLED and formatless border color
|
||
- freedreno: Add enum for fd_lrzfc_layout::dir_track values
|
||
- tu/u_trace: Add support for u_trace inidirect params
|
||
- tu/u_trace: dispatch indirect dims and LRZ status as indirect params
|
||
- freedreno: Clarify RB_BLIT_INFO::TYPE field
|
||
- tu/a7xx: Use generic clear for LOAD_OP_CLEAR
|
||
- tu/a7xx: Use generic clear for CmdClearAttachments
|
||
- tu/a7xx: Use generic clear for image clears
|
||
- tu/a7xx: Use BLIT_EVENT_STORE_AND_CLEAR when appropriate
|
||
- tu: Fix ZPASS_DONE having stale value with generic clear+early frag test
|
||
- ir3: Respect aligment of required consts when allocating ubo,preamble
|
||
- u_trace: Fix trace_payload_as_extra_func desync between drivers
|
||
- tu/a750: Workaround GPU fault when fast-clearing R8G8 formats
|
||
- freedreno/rnndec: Consider array length when finding by reg name
|
||
- freedreno/cffdec: Add option to dump bindless descriptors
|
||
- nir/opt_16b_tex_image: Sign extension should matter for texel buffer txf
|
||
- util/vma: Fix util_vma_heap_get_max_free_continuous_size calculation
|
||
- ir3: Fix cat5 parsing with a1.x src present
|
||
- nir/nir_opt_offsets: Do not fold load/store with const offset > max
|
||
|
||
Dave Airlie (27):
|
||
|
||
- draw/orcjit: supply stub function for tcs coro
|
||
- llvmpipe/cs/orcjit: add stub function name for coro
|
||
- gallivm/sample: fix sampling indirect from vertex shaders
|
||
- nvc0: fix null ptr deref on fermi due to debug changes.
|
||
- anv/video: add mode costs for h264 encoding
|
||
- vulkan/video: handle KHR_video_maintenance1 lack of profile list.
|
||
- radv/video: handle inline queries for vulkan video encode.
|
||
- radv/video: add KHR_video_maintenance1 support
|
||
- radv/video: fix reporting video format props for encode.
|
||
- radv: move video decode enable test into a flag
|
||
- radv: turn video decode/encode on for VCN4 with latest fw
|
||
- llvmpipe: handle stride properly on lvp udmabuf imports
|
||
- llvmpipe: make sure to duplicate the fd handle before giving out
|
||
- vulkan/video: fix vui encoding
|
||
- vl/bitstream: use an int32_t for se encoding.
|
||
- radv/anv/video: handling encoding both sps and pps in same buffer
|
||
- radv: Fix radeon_enc_code_ue with values over 2^16
|
||
- radv/video: fix encode reference slot counting
|
||
- radv/video/enc: report pps overrides in feedback for h265
|
||
- radv/video: use the h264 defines for macroblock w/h
|
||
- radv/video/enc: report pictureAccessGranularity of CTB size.
|
||
- radv/video: add encode field for vcn4
|
||
- radv/video: handle missing h265 feedback struct.
|
||
- zink: free modifiers
|
||
- zink: handle mutable check without crashing if format list not found
|
||
- zink: whitespace fixups
|
||
- va: add some h264 fields
|
||
|
||
David Heidelberg (66):
|
||
|
||
- freedreno: Enable Adreno 306A
|
||
- freedreno: support lua pkgconfig variants
|
||
- freedreno: drop hack for old dEQP
|
||
- freedreno: Introduce emit_blit_fini
|
||
- freedreno: introduce color format parameter to emit_blit_buffer_dst
|
||
- freedreno: Implement fast clear_buffer for Adreno 6xx and 7xx
|
||
- freedreno: drop dEQP hacks
|
||
- intel/genxml: fix length of HCP_FQM_STATE for gen20 and 125
|
||
- llvmpipe: Silence "possibly uninitialized value" warning for ssbo_limit (cont)
|
||
- meson/addrlib: allow unintialized callbacks
|
||
- ci/alpine: use llvm variables
|
||
- ci/alpine: build test with LLVM 18
|
||
- ci/alpine: use packaged spirv-llvm-translator-dev
|
||
- ci/alpine: enable ORCJIT
|
||
- ci/alpine: libclc on Alpine offering necessary targets spirv-mesa3d-;spirv64-mesa3d-
|
||
- freedreno: Implement get_compute_state_info for Adreno 6xx/7xx
|
||
- freedreno: update night expectations
|
||
- etnaviv: rename enums_h appropriately
|
||
- etnaviv: build dependency for the etnaviv tests
|
||
- freedreno: Convert to SPDX-License-Identifier instead of pasting whole license
|
||
- freedreno: use unicode © instead of DOS-like (C) copyright sign
|
||
- bin/gen_release_notes: adjust checksums section
|
||
- ci/freedreno: add Adreno 306/307 occasional timeout
|
||
- ci/iris: implement nightly CL testing using piglit on ADL
|
||
- panfrost: drop leftover definition after pan_nir_lower_64bit_intrin removal
|
||
- nir_lower_mem_access_bit_sizes: Assert when 0 components or bits are requested
|
||
- freedreno/ir3: Use nir_lower_mem_access_bit_sizes instead custom lowering
|
||
- ci/intel: Officially switch intel-adl-cl to pre-merge
|
||
- ci/freedreno: move disabled a530 entries back to main gitlab-ci.yml
|
||
- ci/freedreno: mark jobs to be retested with patched 6.11 kernel
|
||
- ci/freedreno: re-enable piglit tests for Adreno 530 in nightly runs
|
||
- ci/freedreno: mark a530_piglit as postmerge job
|
||
- panfrost/midgard: Implement nir_lower_mem_access_bit_sizes pass
|
||
- panfrost/midgard: Support 64bit pack/unpack
|
||
- panfrost/midgard: Lower fisnormal
|
||
- panfrost/midgard: Lower \*hadd instructions
|
||
- ci/panfrost: Implement basic nightly OpenCL testing on T860
|
||
- ci/panfrost: switch to 6.11 with timestamp patches for traces
|
||
- ci/panfrost: update furmark checksum for T860 traces
|
||
- ci/panfrost: update rock-5b from rc5 to rc7
|
||
- freedreno/ir3: mad.x24 is not safe to lower
|
||
- freedreno/ir3: Do not allow 16-bit mad.x24
|
||
- pvr: unify code into one condition
|
||
- ci/piglit: keep the include_test.h used for OpenCL testing
|
||
- ci/fedora: when doing non-LTO build, we need C_ARGS too
|
||
- ci: enable IPO/LTO only for nightly builds
|
||
- ci: restrict number of maximum threads
|
||
- util: add aarch64 fast half-float to float conversion
|
||
- amd: Pass addrlib cpp args to the tests
|
||
- osmesa: Fix OSMesaGetDepthBuffer() test without depth
|
||
- ci/etnaviv: unify job naming with the rest of the CI
|
||
- ci/freedreno: switch to dash instead of underscore, same as rest of the CI
|
||
- ci/alpine: stay trendy, Alpine flies on LLVM 19 now
|
||
- build: Fix LTO with Android NDK
|
||
- ci/build: debian-build-testing should follow same limit as other jobs
|
||
- ci/fedora: add workaround for buggy GCC warning triggered by r600
|
||
- ci/build: Do not do LTO on any pre-merge job
|
||
- freedreno/ir3: Lower fisnormal op
|
||
- ci/freedreno: Introduce OpenCL testing for Adreno 618, 660, and 750
|
||
- freedreno: python fixes
|
||
- util: Drop 3Dnow optimisation leftovers
|
||
- util: Remove MMX/MMXext detection code
|
||
- util: Drop ancient Intel CPU detection
|
||
- util: drop XOP detection code
|
||
- llvmpipe: align with u_cpu_detect struct changes
|
||
- compiler/rust: drop duplicated bindgen check
|
||
|
||
David Rosca (147):
|
||
|
||
- radeonsi/vcn: Only enable filler data for CBR
|
||
- radeonsi/vcn: Bump enc FW interface version for VCN3 and VCN4
|
||
- radeonsi/vcn: Add support for QVBR rate control mode
|
||
- radeonsi/vcn: Fix MB/CTB calculation for slice encoding
|
||
- dri: Use --whole-archive when linking libdri
|
||
- radeonsi/vcn: Allow dynamic change of QVBR quality level
|
||
- radeonsi/vcn: Add decode DPB buffers as CS dependency
|
||
- gallium: Add header_flags to pipe_h2645_enc_picture_desc
|
||
- frontends/va: Don't check header type for packed header buffers
|
||
- frontends/va: Keep track if VPS/SPS/PPS/AUD was sent
|
||
- frontends/omx: Request SPS PPS for IDR pictures
|
||
- radeonsi/vcn: Only insert headers when requested for H264/5
|
||
- frontends/va: Fix use after free with EFC
|
||
- radeonsi: Don't allow DCC for encode in is_video_target_buffer_supported
|
||
- vulkan/format: Add VK_FORMAT_G12X4_B12X4R12X4_2PLANE_420_UNORM_3PACK16
|
||
- radv/video: Add support for 12-bit AV1 decode
|
||
- gallium: Change pipe_h265_enc_rate_control to array
|
||
- frontends/va: Assert maximum number of temporal layers
|
||
- frontends/va: Get per temporal layer params for HEVC
|
||
- frontends/va: Parse packed header slice for HEVC TemporalId
|
||
- frontends/va: Fix leaks with multiple coded buffer segments
|
||
- util/rbsp: Fill bits again if reading more than 16 leading zeros
|
||
- frontends/va: Implement DPB management for H264/5 encode
|
||
- frontends/va: Only use parameters for first slice in H264/5 encode
|
||
- frontends/va: Parse H264 slice packed header
|
||
- frontends/va: Parse HEVC slice packed header
|
||
- frontends/omx: Adjust to encoding changes
|
||
- radeonsi: Add GPU copy path to si_video_resize_buffer
|
||
- radeonsi/vcn: Switch to app DPB management for H264 and HEVC encode
|
||
- radeonsi/vcn: Deduplicate header encoding functions
|
||
- radeonsi/vcn: Fix radeon_enc_code_ue with values over 2^16
|
||
- radeonsi/vcn: Use pipe header params in HEVC header encoder
|
||
- radeonsi/vcn: Use pipe header params in H264 header encoder
|
||
- radeonsi: Enable transform_skip HEVC feature support
|
||
- radeonsi: Implement buf_ofst_info in si_video_resize_buffer GPU copy path
|
||
- radeonsi/vcn: Support H264 constrained intra prediction
|
||
- d3d12: Remove get_feedback_fence implementation
|
||
- frontends/va: Implement sync buffer/surface timeout for encode feedback
|
||
- pipe: Add int return value to video end_frame
|
||
- frontends/va: Check for errors from end_frame in vlVaEndPicture
|
||
- radeonsi/vcn: Implement get_feedback_fence
|
||
- pipe: Use correct struct for h264/h265 dpb entry
|
||
- frontends/va: Allocate surface buffers on demand
|
||
- frontends/va: Allow drivers to allocate and use encode DPB surface buffers
|
||
- egl/wayland: Fix flush after blit in swap buffers multi-GPU case
|
||
- frontends/va: Fix locking in vlVaDeriveImage
|
||
- frontends/va: Fix locking in vlVaQueryVideoProcPipelineCaps
|
||
- vulkan/wsi: Fix stack-use-after-scope in x11_surface_create_swapchain
|
||
- frontends/va: Parse more H264 PPS for transform_8x8_mode_flag
|
||
- radeonsi/vcn: Stop hardcoding values in VCN version overrides
|
||
- radeonsi/vcn: Add support for H264 8x8 transform on VCN5
|
||
- radv/video: Don't use op_close in encode
|
||
- radv/video: Don't use multiple tasks for one encode operation
|
||
- ci: Disable building OMX driver
|
||
- Remove OMX
|
||
- docs: Remove OMX frontend
|
||
- glx: Destroy base screen after deinitScreen
|
||
- pipe: Add pipe_h264_nal_unit_type and pipe_h265_nal_unit_type enums
|
||
- frontends/va: Support raw packed headers for H264 and HEVC
|
||
- frontends/va: Allow multiple NALUs in coded buffer segments
|
||
- radeonsi/vcn: Support raw packed headers for H264 and HEVC
|
||
- radeonsi: Remove unused radeon_temporal.h
|
||
- radeonsi/vcn: Don't copy the pipe enc structs
|
||
- pipe: Remove video enc header_flags
|
||
- meson/megadriver: Add megadriver_libdir argument
|
||
- targets/va: Build va driver into libgallium when building with dri
|
||
- targets/vdpau: Build vdpau driver into libgallium when building with dri
|
||
- radeonsi/vce,uvd_enc: Stop using obsolete rate control params
|
||
- radeonsi/vce: Stop using base.level and base.max_references
|
||
- radeonsi/uvd_enc: Stop using base.level
|
||
- radeonsi/vcn: Stop using base.level for H264 level_idc
|
||
- d3d12: Stop using base.level for H264 level_idc
|
||
- frontends/va: Set default encoding parameters at context creation
|
||
- Revert "frontends/va: Process VAEncSequenceParameterBufferType first in vaRenderPicture"
|
||
- frontends/va: Create encoder at context creation
|
||
- pipe: Remove unused fields in video rate control
|
||
- frontends/va: Reset intra refresh in beginPicture
|
||
- radeonsi: Fix reported minimum width for video encode
|
||
- frontends/va: Remove fallback if app doesn't enable packed headers
|
||
- radeonsi: Disable EFC on VCN 2.2
|
||
- radeonsi/vcn: Add rc_per_pic_ex encode command
|
||
- ac: Fix typo RENCDOE -> RENCODE
|
||
- ac: Add ac_vcn_init_enc_cmds
|
||
- ac: Add remaining VCN encode defines
|
||
- radeonsi/vcn: Use ac_vcn_enc_init_cmds and AV1 defines from ac
|
||
- ac: Add VCN IB parser
|
||
- radeonsi/vcn: Enable IB parsing with AMD_DEBUG=ib
|
||
- radeonsi/vcn: Don't reuse context with multiple VCN instances
|
||
- frontends/va: Fix AV1 packed header parsing
|
||
- pipe: Add video fence_wait
|
||
- radeonsi/uvd: Implement fence_wait
|
||
- radeonsi/vcn: Implement fence_wait
|
||
- radeonsi/vpe: Implement fence_wait
|
||
- d3d12: Implement fence_wait
|
||
- r600/uvd: Implement fence_wait
|
||
- frontends/va: Use fence_wait instead of get_*_fence
|
||
- pipe: Remove video get_*_fence
|
||
- radeonsi/vcn: Allow per-frame QP and max frame size changes
|
||
- radeonsi/vcn: Use correct initial DPB size for HEVC encode
|
||
- radeonsi/vcn: Don't hardcode nal_ref_idc
|
||
- radeonsi/vcn: Fix some small coverity issues
|
||
- radeonsi/vcn: Fix out of bounds write when invalidating QP map regions
|
||
- radeonsi/vcn: Fix out of bounds read in H264 decode
|
||
- frontends/va: Implement DPB management for AV1
|
||
- frontends/va: Parse more AV1 frame and sequence header params
|
||
- frontends/va: Support raw packed headers for AV1
|
||
- radeonsi/vcn: Switch to app DPB management for AV1
|
||
- frontends/va: Remove now unused ref fields for AV1 encode
|
||
- radeonsi/vcn: Support raw packed headers for AV1
|
||
- radeonsi/vcn: Fix coded size in sequence header on VCN5
|
||
- radeonsi/vcn: Add support for VCN5 AV1 compound
|
||
- radeonsi/vcn: Add support for VCN5 dpb tier2
|
||
- frontends/va: Fix parsing HEVC st_ref_pic_set
|
||
- frontends/va: Parse H264 prefix for temporal_id
|
||
- frontends/va: Calculate VBV params for other layers relative to base layer
|
||
- radeonsi/vcn: Implement per-layer rate control for HEVC
|
||
- radeonsi/vcn: Select layer for rate control
|
||
- frontends/va: Fix some small coverity issues
|
||
- frontends/va: Fix NULL check in intra refresh params
|
||
- frontends/va: Fix parsing leb128 when using more than 4 bytes
|
||
- frontends/va: Don't relock mutex in vlVaEndPicture
|
||
- pipe: Fix video enc max DPB size for H264
|
||
- frontends/va: Evict unused surfaces from encode DPB
|
||
- radeonsi/vcn: Only enable skip mode with matching references
|
||
- frontends/va: Add support for AV1 still picture encode
|
||
- radeonsi/vcn: Add support for AV1 still picture encode
|
||
- radeonsi/uvd_enc: Implement fence_wait
|
||
- radeonsi/vce: Implement fence_wait
|
||
- radeonsi: Enable VIDEO_CAP_ENC_SUPPORTS_ASYNC_OPERATION on VCE/UVD
|
||
- radeonsi/uvd_enc: Allocate session buffer in VRAM
|
||
- radeonsi: Support HEVC features and block sizes for UVD
|
||
- radeonsi/vcn: Enable VCN4 AV1 encode WA
|
||
- radeonsi/vce: Remove dual instance support
|
||
- frontends/va: Remove H264 encode delayed flush
|
||
- frontends/va: Stop using PIPE_VIDEO_CAP_ENC_SUPPORTS_ASYNC_OPERATION
|
||
- pipe: Remove PIPE_VIDEO_CAP_ENC_SUPPORTS_ASYNC_OPERATION
|
||
- frontends/va: Move encode fence to coded buffer
|
||
- radeonsi/vcn: Don't try to override HEVC SPS conformance window
|
||
- radeonsi/vcn: Gracefully handle encode errors and report to frontend
|
||
- radeonsi/vcn: Rework encode padding and session init params
|
||
- radv/video: Fix H264 slice control
|
||
- radv/video: Fix HEVC slice control
|
||
- radv/video: Report correct encodeInputPictureGranularity
|
||
- radv/video: Avoid selecting rc layer over maximum
|
||
- radeonsi/vcn: Use correct frame context buffer for preencode on VCN5
|
||
- frontends/va: Create surfaces with correct fourcc for RT format
|
||
- frontends/va: Use compositor blit with different number of planes
|
||
|
||
Deborah Brouwer (15):
|
||
|
||
- ci/lava: Detect a6xx gpu recovery failures
|
||
- ci: add firmware files to rootfs
|
||
- ci/intel: add i915/MTL firmware to rootfs
|
||
- ci: remove xfail program\@build\@include-directories
|
||
- ci: remove obsolete build-hang-detection script
|
||
- ci/deqp-runner: fix option to build from a commit
|
||
- ci/deqp-runner: print version info
|
||
- ci/deqp-runner: update instructions for uprevving
|
||
- ci/deqp-runner: uprev from 0.20.0 to 0.20.2
|
||
- ci/deqp-runner: implement max-fails
|
||
- freedreno/ci: remove redundant skip files
|
||
- ci: simplify .baremetal-arm32-asan-test
|
||
- ci: simplify .baremetal-arm64-asan-test
|
||
- freedreno/ci: convert a630-gles-asan to toml suite
|
||
- ci/b2c: update RESULTS_DIR for .b2c-test jobs
|
||
|
||
Denis (1):
|
||
|
||
- freedreno: Enable A505
|
||
|
||
Derek Foreman (5):
|
||
|
||
- meson: Update wayland-protocols to 1.38
|
||
- vulkan/wsi/wayland: Use presentation timing v2 when available
|
||
- vulkan/wsi/wayland: Use fifo protocol for FIFO
|
||
- vulkan/wsi/wayland: Pace frames with commit-timing-v1
|
||
- adv+zink/ci: Add a recent flake
|
||
|
||
Dmitry Baryshkov (3):
|
||
|
||
- freedreno/ir3: Treat MESA_SHADER_KERNEL in the same way as compute
|
||
- freedreno/ir3: Lower the hadd operations
|
||
- rusticl: Enable on freedreno
|
||
|
||
Dmitry Osipenko (6):
|
||
|
||
- util/mesa-db: Fix missing O_CLOEXEC
|
||
- util/mesa-db-multipart: Open one cache part at a time
|
||
- util/mesa-db: Open DB files during access time
|
||
- util/mesa-db: Fix crash on compacting empty DB
|
||
- util/cache_test: Add mesa-db test for adding cache entry bigger than empty cache
|
||
- ci/zink: Mark glx-multithread-clearbuffer flake on ADL
|
||
|
||
Donald Robson (2):
|
||
|
||
- pvr: Stop creating waits when there are no waits
|
||
- pvr: Fixed creation of waits in queue submission
|
||
|
||
Dylan Baker (35):
|
||
|
||
- compilers/clc: Add missing break statements.
|
||
- mesa: fix memory leak when using shader cache
|
||
- util/glsl2spirv: fixup the generated depfile when copying sources
|
||
- util/glsl2spirv: only copy the input file if we're going to change it
|
||
- tgsi_to_nir: free disk cache value if the size is wrong
|
||
- crocus: properly free resources on BO allocation failure
|
||
- crocus: check for depth+stencil before creating resource
|
||
- mesa/program: remove memcpy where src == dst
|
||
- docs: add release notes for 24.2.2
|
||
- docs: update sha sums for 24.2.2
|
||
- docs: update sha sums for 24.2.2
|
||
- iris: Run checks that do not require resources before creating them
|
||
- anv: if queue is NULL in vm_bind return early
|
||
- intel: replace \`(uint64_t - uint64_t) > 0` with \`uint64_t > uint64_t`
|
||
- anv: assert we don't write past the end of an array
|
||
- anv: enforce state->cmd_buffer is never null in emit_Simpler_shader_init_fragment
|
||
- intel/elk: delete copy constructor and copy-assignment-operator
|
||
- compiler/glsl: explicitly delete copy constructor and copy-assign
|
||
- util/fossilize_db: don't use check_files_opened_successfully for foz_prepare
|
||
- anv: remove useless uint >= 0 check
|
||
- hasvk: remove useless uint >= 0 check
|
||
- intel/perf: delete dead code.
|
||
- meson: use \`add_project_arguments` instead of \`global`
|
||
- anv: enable VK_EXT_ycbcr_2plane_444_formats
|
||
- release: push 24.3 out two weeks
|
||
- anv: assert that we don't read off the end color_att array
|
||
- anv: when handling descriptor sets, layout cannot be NULL if dynamic_offets are used
|
||
- VERSION: bump for 24.3.0-rc1 release
|
||
- .pick_status.json: Update to ced2404cb433deaa84cf6cf9edce212733117c0b
|
||
- .pick_status.json: Update to 5e0b81413de588803c9a0736acd8decd40d19ab6
|
||
- .pick_status.json: Update to b0c9789bc1ed808d29f642e9445599dc85896444
|
||
- VERSION: bump for 24.3.0-rc2 release
|
||
- .pick_status.json: Update to b8d253b30c0a8c1878f2429f0a4df0a068bbf0aa
|
||
- .pick_status.json: Update to 8845255881f03df9a695fae164a64201c4211d35
|
||
- .pick_status.json: Update to 9b779068c3a7a8ded00f1d2519fd01dd1d05f3d8
|
||
|
||
Echo J (3):
|
||
|
||
- d3d10umd: Use pipe_resource_usage enum in translate_resource_usage()
|
||
- util: Fix the integer addition in os_time_get_absolute_timeout()
|
||
- compiler/rust: Force native build for the compiler_proc library
|
||
|
||
Eric Engestrom (249):
|
||
|
||
- VERSION: bump to 24.3
|
||
- docs: reset new_features.txt
|
||
- loader: gc loader_get_extensions_name() and __DRI_DRIVER_{GET\_,}EXTENSIONS defines
|
||
- radeonsi/ci: skip timing out test
|
||
- freedreno/ci: double job timeout for a306
|
||
- freedreno/ci: document extra variants of failing tests on a618 and a630
|
||
- anv+zink/ci: mark some tests as fixed
|
||
- anv+zink/ci: document two tests, one failing and one crashing
|
||
- anv+zink/ci: mark a couple of tests as flaky
|
||
- venus/ci: skip timing out test
|
||
- nak: fix meson typo
|
||
- venus: initialize bitset in CreateDescriptorPool()
|
||
- v3d/ci: mark spec\@amd_performance_monitor\@vc4 tests as flaky
|
||
- meson: xcb & xcb-randr are needed by the loader whenever x11 is built
|
||
- docs: add stub header for u_format_gen.h
|
||
- v3d,v3dv: reuse dep_v3d_hw from simulator/meson.build
|
||
- v3d,v3dv: figure out whether we're using the simulator only once
|
||
- v3d,v3dv: simplify USE_V3D_SIMULATOR/using_v3d_simulator logic
|
||
- vc4,v3d,v3dv: avoid compiling in unused \`sim_file` field in the {vc4,v3d}_screen/v3dv_physical_device struct
|
||
- ci/baremetal: fix logic for retrying boot when it failed
|
||
- v3d/ci: mark spec\@amd_performance_monitor\@api as flaky on rpi4
|
||
- v3d/ci: mark glx\@glx-visuals-depth as flaky on rpi4, just like the -stencil variant
|
||
- meson: don't select the deprecated \`swrast` option ourselves
|
||
- meson: improve wording of "incompatible llvm options" error
|
||
- ci: remove llvmpipe in the job that disables llvm
|
||
- rpi4/ci: mark fs variant of arb_texture_buffer_object as flaky too
|
||
- docs/meson: add missing newline at the begining of code-block
|
||
- bin/ci: update python-gitlab to 4.x
|
||
- ci/prepare-artifacts: use find's -exec instead of iterating over its output with xargs
|
||
- ci/prepare-artifacts: check if we're about to overwrite ci config/expectations files
|
||
- llvmpipe/ci: bump timeout for nightly job to 1h
|
||
- ci: drop dead VKD3D_CONFIG
|
||
- ci/vkd3d: rename vkd3d runner script to vkd3d-runner.sh to match the other ones
|
||
- ci/vkd3d: store vkd3d "errors" in the results file instead of spamming the job log
|
||
- ci/vkd3d: use GPU_VERSION to identify the list of failures, unifying with deqp-runner.sh
|
||
- ci/vkd3d: move test expectations file to the results folder
|
||
- ci/vkd3d: move expectation file creation logic out of the test results handling
|
||
- ci/vkd3d: be more specific for what to grep
|
||
- ci/vkd3d: rename vkd3d expected failures list to be more explicit
|
||
- ci/vkd3d: add support for \*-skips.txt list files
|
||
- ci/vkd3d: add support for \*-flakes.txt list files
|
||
- pick-ui: move NominationType values around so that None is 0
|
||
- ci/poe-powered: rename --test-timeout to --test-timeout-minutes to be more explicit
|
||
- ci/bare-metal: rename fastboot & cros-servo TEST_PHASE_TIMEOUT to TEST_PHASE_TIMEOUT_MINUTES to be coherent
|
||
- ci/poe-powered: allow setting a different boot timeout than 5 minutes
|
||
- rpi/ci: shorten the boot timeout to 90 seconds
|
||
- docs: add release notes for 24.1.5
|
||
- docs: update calendar for 24.1.5
|
||
- docs: add sha256sum for 24.1.5
|
||
- nvk/ci: bump vkcts job timeout
|
||
- meson,ci: remove dead \`kmsro` option in \`gallium-drivers`
|
||
- ci: remove dead start-x.sh script
|
||
- rpi3/ci: remove {version,render}_check from [piglit.env]
|
||
- ci/deqp-runner: fix --fraction-start in non-parallel jobs
|
||
- ci: pass MESA_SPIRV_LOG_LEVEL from job to the test
|
||
- android: fix build in multiple ways
|
||
- docs: update calendar for 24.2
|
||
- docs: add release notes for 24.2.0
|
||
- docs: add sha256sum for 24.2.0
|
||
- docs: update calendar for 24.1.6
|
||
- docs: add release notes for 24.1.6
|
||
- docs: add sha256sum for 24.1.6
|
||
- ci/container: define LLVM_VERSION in the alpine container job
|
||
- ci/build: reuse alpine llvm version to make sure it stays coherent
|
||
- ci/build: add comment to explain why s390x is useful
|
||
- ci/build: document \@ajax as a stakeholder for s390x
|
||
- ci/build: drop softpipe from s390x job
|
||
- nvk/ci: add vkd3d job on the ga106
|
||
- llvmpipe/ci: mark now-skipping test as no longer failing
|
||
- nvk+zink/ci: mark \`spec\@arb_sample_shading\@samplemask .*\@0\\..* partition` as fixed
|
||
- nvk/ci: document regressions
|
||
- turnip/ci: document regressions
|
||
- turnip/ci: skip timing out tests
|
||
- turnip/ci: document flakes
|
||
- nvk/ci: mark a bunch of tests as fixed
|
||
- turnip/ci: add a750_vk flakes
|
||
- docs/envvars: fix reference
|
||
- broadcom/ci: set default TEST_PHASE_TIMEOUT_MINUTES so that jobs never timeout without uploading results
|
||
- broadcom/ci: increase the timeout of the nightly job v3d-rpi4-gl:arm32 by 10min
|
||
- ci/build: add honeykrisp (asahi vk) to several builds
|
||
- intel/ci: don't trigger anv-jsl-full & anv-tgl-full on GL changes
|
||
- nvk/ci: document CTS bug
|
||
- ci: add check for misleading indentation in ci toml files
|
||
- ci: don't run hardware tests if the toml files don't pass validation
|
||
- ci: fix toml-lint rules to not block nightly pipelines
|
||
- r300/ci: move postmerge jobs to -postmerge stage
|
||
- lavapipe/ci: move postmerge jobs to -postmerge stage
|
||
- venus/ci: move postmerge jobs to -postmerge stage
|
||
- ci/rules: make every job exist as manual in fork pipelines
|
||
- kopper: only include loader_dri3_helper.h when xcb is enabled
|
||
- zink+nvk/ci: document new variant of test failing
|
||
- zink+nvk/ci: document all the flakes seen in the last week
|
||
- turnip/ci: document all the a750 flakes seen in the last week
|
||
- vc4: Add missing libvc4_neon build dependencies
|
||
- vc4/meson: simplify neon build now that the android build system doesn't exist anymore
|
||
- ci/deqp-runner: restore CC after temporarily overriding it
|
||
- ci/deqp: simplify command to list local deqp patches
|
||
- ci/deqp-runner: be less verbose in the loop printing the deqp builds info
|
||
- ci/deqp-runner: build from git checkout even on linux
|
||
- ci/deqp-runner: add infra to apply patches
|
||
- ci/image-tags: re-generate all the images building deqp-runner
|
||
- ci/build: fix ppc64le and s390x jobs rules
|
||
- ci: take igalia farm offline
|
||
- docs: add release notes for 24.2.1
|
||
- docs: add sha sum for 24.2.1
|
||
- docs: update calendar for 24.2.1
|
||
- nvk/ci: drop skip of fixed vkd3d test
|
||
- ci/vkd3d: make the job output less spammy, focus on what is useful to print
|
||
- ci/vkd3d: use upstream test-runner.sh for process isolation
|
||
- ci: run only one vkd3d test at a time
|
||
- ci/vkd3d: check for tests listed in fails but not actually failing
|
||
- v3dv/ci: fix test timeout for v3dv-rpi5-vk-full:arm64
|
||
- etnaviv/ci: fix gc2000_piglit test timeout
|
||
- freedreno/ci: fix a630_*_full tests timeout
|
||
- nouveau/ci: fix gk20a-gles test timeout
|
||
- freedreno/ci: drop unused TEST_PHASE_TIMEOUT_MINUTES in LAVA jobs
|
||
- ci/bare-metal/poe: change the default TEST_PHASE_TIMEOUT_MINUTES to \`CI_JOB_TIMEOUT - 5min`
|
||
- broadcom/ci: drop TEST_PHASE_TIMEOUT_MINUTES that match the default value
|
||
- broadcom/ci: use TEST_SETUP_AND_UPLOAD_MARGIN_MINUTES
|
||
- etnaviv/ci: drop TEST_PHASE_TIMEOUT_MINUTES that match the default value
|
||
- freedreno/ci: drop TEST_PHASE_TIMEOUT_MINUTES that match the default value
|
||
- nouveau/ci: drop TEST_PHASE_TIMEOUT_MINUTES that match the default value
|
||
- ci/build: untangle debian-s390x from debian-ppc64el
|
||
- freedreno/ci: add a630 flake
|
||
- docs: add release notes for 24.1.7
|
||
- docs: update calendar for 24.1.7
|
||
- docs: add sha sum for 24.1.7
|
||
- ci: move build jobs that are used by tests to an earlier stage
|
||
- ci: merge build-x86_64 & build-misc
|
||
- ci: do not start build-only jobs until the critical build-for-tests jobs are done
|
||
- nvk/ci: mark -dEQP-VK.drm_format_modifiers.export_import* as fixed
|
||
- radeonsi/ci: bump timeout for nightly job glcts-vangogh-valve
|
||
- Revert "freedreno/ci: drop TEST_PHASE_TIMEOUT_MINUTES that match the default value"
|
||
- freedreno/ci: fix test timeout for a306_piglit
|
||
- radeonsi/ci: mark a bunch of subgroups tests as failing
|
||
- venus/ci: drop redundant flakes definitions
|
||
- venus/ci: add flake and skip timing out test
|
||
- zink+nvk/ci: bump zink-nvk-ga106-valve timeout as more tests are being run
|
||
- zink+nvk/ci: mark a ton of tests as fixed
|
||
- zink+turnip/ci: fix .zink-turnip-valve-manual-rules
|
||
- turnip/ci: add vkd3d job on the a750
|
||
- ci: add virgl & venus to the s390 build
|
||
- docs: Fix linkcheck failures
|
||
- zink+nvk/ci: mark test as fixed
|
||
- zink+nvk/ci: add flakes seen recently
|
||
- docs: add release notes for 24.2.3
|
||
- docs: update calendar for 24.2.3
|
||
- docs: add sha sum for 24.2.3
|
||
- amd/ci: track changes to the global driver \`*-skips.txt` files
|
||
- broadcom/ci: track changes to the global driver \`*-skips.txt` files
|
||
- llvmpipe/ci: track changes to the global driver \`*-skips.txt` files
|
||
- lavapipe/ci: track changes to the global driver \`*-skips.txt` files
|
||
- intel/ci: track changes to the global driver \`*-skips.txt` files
|
||
- v3dv/ci: drop skip of fixed test
|
||
- docs/release-calendar: add 24.3 branchpoint and rcs
|
||
- zink+nvk/ci: document spec\@egl_ext_surface_compression\@create as crashing
|
||
- zink+nvk/ci: add flakes seen recently
|
||
- lavapipe/ci: document regression
|
||
- llvmpipe/ci: document regression
|
||
- radeonsi/ci: document spec\@egl_ext_surface_compression\@create as crashing
|
||
- radeonsi/ci: mark KHR-GL46.shader_image_load_store.basic-allTargets-atomic as fixed
|
||
- ci/build: move debian-clang-release to priority build-for-tests stage
|
||
- lavapipe/ci: document ray tracing regression
|
||
- mr-label-maker: add gfxstream
|
||
- ci/deqp: make sure the git commit hash of deqp is stable across rebuilds
|
||
- ci: fix windows container jobs being missing on forks
|
||
- ci/build: don't wait on the windows build anymore
|
||
- turnip/ci: mark vkd3d test_planar_video_formats as fixed
|
||
- zink+nvk/ci: mark spec\@egl 1.4\@egl-ext_egl_image_storage as fixed
|
||
- zink+nvk/ci: document more flakes seen lately
|
||
- egl: fix dri2_from_names() call
|
||
- docs: add release notes for 24.2.4
|
||
- docs: update calendar for 24.2.4
|
||
- docs: add sha sum for 24.2.4
|
||
- Revert "ci/build: move debian-clang-release to priority build-for-tests stage"
|
||
- ci/build: move debian-build-testing to priority build-for-tests stage
|
||
- ci/build: mention the .build-for-tests-jobs exception
|
||
- docs: add release notes for 24.2.5
|
||
- docs: update calendar for 24.2.5
|
||
- docs: add sha sum for 24.2.5
|
||
- ci: drop spec\@egl_ext_surface_compression\@create from the rest of the expected fails
|
||
- zink+nvk/ci: document regression
|
||
- zink+nvk/ci: document more flakes seen
|
||
- zink+radv/ci: document more flakes seen
|
||
- nvk/ci: document more flakes seen
|
||
- ci: run shader-db only when one of the drivers it tests is modified
|
||
- zink+nvk/ci: mark a few tests as fixed
|
||
- zink+nvk/ci: document more flakes seen
|
||
- ci/{deqp,piglit}-runner: make zstd respect FDO_CI_CONCURRENT when set, and fallback to the current "all the threads"
|
||
- ci/prepare-artifacts: use more than one thread when compressing the install tarball
|
||
- freedreno/ci: add {freedreno,turnip}{,-manual}-rules to simplify the rest of the code
|
||
- freedreno/ci: use {freedreno,turnip}{,-manual}-rules
|
||
- ci: rename "freedreno" farm to "google-freedreno"
|
||
- {freedreno,intel}/ci: rename "premerge-skips.txt" to "merge-skips.txt" to accurately reflect what they are
|
||
- {freedreno,intel}/ci: add missing tracking of \`merge-skips.txt` files
|
||
- ci: simplify ci-tron priority variable
|
||
- ci: simplify setting lava job priority
|
||
- ci: fix shader-db rules
|
||
- ci/b2c: allow B2C_DTB_URL to be set to an empty string to unset it
|
||
- ci/b2c: make initramfs & kernel urls optional
|
||
- ci: move yaml-toml-shell-test out of build-for-tests stage
|
||
- ci: move comment into the appropriate section
|
||
- zink+turnip/ci: update a750 results
|
||
- docs: update link to \`perf` wiki
|
||
- nvk/ci: add back a crash that was mistakenly removed from the expectations
|
||
- zink+nvk/ci: document ext_egl_image_storage regression in c06a55fd...2fb4aed9 (likely !31585)
|
||
- all-drivers/ci: drop duplicate flakes lines
|
||
- zink+nvk/ci: drop duplicate flakes lines
|
||
- zink+nvk/ci: add flakes seen recently
|
||
- turnip/ci: add more dEQP-VK.renderpass2.fragment_density_map.* flakes seen recently
|
||
- nvk/ci: add flakes seen recently
|
||
- lavapipe/ci: group & sort fails
|
||
- lavapipe/ci: group & sort skips
|
||
- lavapipe/ci: skip builtin ray query tests that take too long and time out
|
||
- lavapipe/ci: document regression in the commit range 765d1c47...366f63fd
|
||
- ci: move shellcheck options to .shellcheckrc
|
||
- ci: rename "merge-skips" to "slow-skips" as they're about to be used outside of merge piplines
|
||
- ci: skip slow tests on all non-"full" jobs
|
||
- ci: drop dead "load jwt in the environment just before exiting the job" code
|
||
- egl: error out during setup if the configuration is invalid
|
||
- freedreno/ci: abort a750 testing when a hang is detected
|
||
- freedreno/ci: add nightly freedreno gl testing on a750
|
||
- ci/build: deduplicate meson build command
|
||
- ci/build: handle build timeout ourselves to give more time to nightly LTO builds
|
||
- ci/build: lower most job's timeout from 30 to 15 min
|
||
- docs: add release notes for 24.2.6
|
||
- docs: update calendar for 24.2.6
|
||
- docs: add sha sum for 24.2.6
|
||
- ci: only end current section if there is a current section
|
||
- ci: initialize state_x (tracking \`set -x`)
|
||
- ci: consistently restore \`-x` after temporarily disabling it
|
||
- ci: rename state_x to previous_state_x to be clear about what it contains
|
||
- ci/b2c: properly escape double-quoted container command
|
||
- ci/b2c: make dut boot quieter on ci-tron jobs
|
||
- ci: properly quote ccache dir and path
|
||
- ci: don't replace ld with ld.gold if ld.gold doesn't exist
|
||
- ci: bump ci-templates
|
||
- amd/ci: document flakes seen lately
|
||
- nvk/ci: document flakes seen lately
|
||
- freedreno/ci: document flakes seen lately
|
||
- broadcom/ci: document flakes seen lately
|
||
- meson: drop variable initialized twice
|
||
- meson: add dependencies needed by wsi_common_x11.c even on non-drm platforms
|
||
- ci: drop unused extra args in build-vkd3d-proton.sh
|
||
- ci: replace plain \`meson` with explicit \`meson setup`
|
||
- meson: bump spirv-tools version needed to v2022.1
|
||
- ci: move error handling functions at the end
|
||
- ci: use quiet alias for commands
|
||
- ci: raise priority of release manager pipelines
|
||
|
||
Eric R. Smith (19):
|
||
|
||
- panfrost: make PAN_MESA_DEBUG=dump really show all GPU memory
|
||
- panfrost: fix texture.border_clamp regression for valhall
|
||
- panfrost: use RGB1 component ordering for R5G6B5 pixel formats
|
||
- panfrost: Make the unit to use for instructions explicit
|
||
- panfrost: Update bifrost_isa.py to handle some Valhall constructs
|
||
- panfrost: refactor valhall ISA parser
|
||
- panfrost: split pseudo instructions from Bifrost and Valhall
|
||
- panfrost: fix crash when disassembling shader for trace
|
||
- panfrost: Add b2i16 conversion to table
|
||
- panfrost: move pan_arch definition out of genxml to pan_props.h
|
||
- panfrost: move lower_image_ms to after lower_atomics
|
||
- panfrost: implement image2dms support for valhall
|
||
- panfrost: add support for image2DMSArray on bifrost
|
||
- panfrost: improve swizzle inversion
|
||
- panfrost: print human readable versions of some swizzle fields
|
||
- panfrost: Add back A8_UNORM format for valhall
|
||
- panfrost: fix SAMPLE_ALPHA_TO_ONE
|
||
- panvk: Support blend shaders for alpha_to_one_enable
|
||
- panfrost: fix earlyzs settings for alpha_to_coverage
|
||
|
||
Erico Nunes (13):
|
||
|
||
- dri: fix sun4i-drm driver name
|
||
- lima/ppir: add folding lowering for src modifiers
|
||
- lima/ppir: add folding lowering for dest modifiers
|
||
- lima/ppir: remove nir_legacy
|
||
- lima/ppir: add support and folding for fclamp_pos
|
||
- lima/ppir: add support and folding trunc to a modifier
|
||
- lima/ppir: switch to derivative intrinsics
|
||
- docs/ci: pass -D platforms=[] to avoid dependencies
|
||
- v3dv: unconditionally expose VK_EXT_queue_family_foreign
|
||
- v3dv: match render and display device for wsi present
|
||
- lima/ci: define FDO_HTTP_CACHE_URI for caching-proxy
|
||
- lima/ci: update piglit ci expectations
|
||
- Revert "ci/lima: Temporarily disable"
|
||
|
||
Erik Faye-Lund (54):
|
||
|
||
- docs/panfrost: mention that Mali-G610 is conformant
|
||
- mesa/main: use extension-helper
|
||
- mesa/main: validate GL_UNSIGNED_INT_5_9_9_9_REV
|
||
- mesa/main: validate GL_UNSIGNED_INT_2_10_10_10_REV
|
||
- mesa/main: validate GL_UNSIGNED_INT_10_10_10_2
|
||
- mesa/main: check depth/stencil formats
|
||
- mesa/main: validate desktop gl format/types
|
||
- mesa/main: validate bgr formats
|
||
- mesa/main: validate abgr format
|
||
- mesa/main: validate integer-formats
|
||
- mesa/main: validate half-float types
|
||
- mesa/main: switch from deny-list to allow-list
|
||
- mesa/main: add gles-compatible check helpers
|
||
- mesa/main: tighten glsl-version checks
|
||
- mesa/main: tighten gl-version checks
|
||
- panvk: remove out-of-date detail
|
||
- panvk: drop unused define
|
||
- panvk: error-check get_fb_descs calls
|
||
- mesa: make compressed-format emulation optional
|
||
- panfrost: unify compressed formats
|
||
- panfrost: store texfeat_bit in panfrost_format
|
||
- panfrost: check fmt.bitfeat_bit for compressed-support
|
||
- panfrost: do not re-fetch compressed formats
|
||
- panvk: check against texfeat_bit
|
||
- panvk: implement KHR_get_memory_requirements2
|
||
- panvk: drop needless function implementation
|
||
- panvk: expose already-implemented KHR_maintenance1
|
||
- i915: do not calculate native instructions
|
||
- mesa: remove superfluous "native" limits and stats
|
||
- mesa: clean up setup of MaxAddressRegs constant
|
||
- panvk: implement and expose KHR_maintenance2
|
||
- panvk: expose KHR_relaxed_block_layout support
|
||
- panvk: expose KHR_bind_memory2 support
|
||
- docs/features: fix bad sorting of drivers
|
||
- panvk: expose KHR_external_semaphore and KHR_external_semaphore_capabilities
|
||
- panvk: expose KHR_external_memory and KHR_external_memory_capabilities
|
||
- panvk: expose KHR_external_fence and KHR_external_fence_capabilities
|
||
- panvk: fix broken wsi
|
||
- panvk: avoid signed integer underflow
|
||
- panvk: drop needless assert
|
||
- panvk: put conditional outside of define
|
||
- panvk: assert on missing vs
|
||
- panvk/csf: only look at fs if it's required
|
||
- panfrost: drop needless assign
|
||
- panfrost: add an assert in render-target setup
|
||
- panvk: drop duplicate dirty-test
|
||
- panvk/jm: respect depth-clip state
|
||
- panvk/csf: respect depth-clip state
|
||
- panvk: expose EXT_depth_clip_enable
|
||
- docs/features: claim vk 1.0 for panvk
|
||
- docs/features: mark off more panvk extensions
|
||
- panvk: enable KHR_16bit_storage
|
||
- docs/features: mark off missing panvk feature
|
||
- glx: avoid null-deref
|
||
|
||
Erwin Jansen (1):
|
||
|
||
- Enable GFXStream bazel build
|
||
|
||
Evan (3):
|
||
|
||
- amd/vpelib: Color Keyer Implementation
|
||
- amd/vpelib: Luma AND Color Keyer Full Support
|
||
- amd/vpelib: Input Format Adjustment
|
||
|
||
Faith Ekstrand (309):
|
||
|
||
- nvk: Fix indirect cbuf binds pre-Turing
|
||
- nvk: Don't advertise sparse residency on Maxwell A
|
||
- nak/sm50: Split IAdd2 into IAdd2 and IAdd2X
|
||
- nak: Implement ineg on sm50
|
||
- nak: Add an AtomCmpSrc to AtomOp::CmpExch
|
||
- nak/sm50: Support AtomOp::CmpExch
|
||
- nak/sm50: Improve encoding of OpFFma
|
||
- nak/sm50: Various encoding fixes
|
||
- nak/sm50: Implement OpPixLd
|
||
- nak: Lower fsqrt in NIR on Maxwell A and earlier
|
||
- nak/sm50: Implement OpCCtl
|
||
- nak/nir: Split 64-bit conversions pre-Volta
|
||
- nak: Drop the nvfuzz binary
|
||
- nak: Plumb tessellation parameters through ShaderStageInfo
|
||
- nak: Rework fragment shader stage info
|
||
- nak: Move a few more things to VtgIoInfo
|
||
- nak: Move creation of nak_shader_info to ShaderBin::new()
|
||
- nvk: Reject sparse images on Maxwell A and earlier
|
||
- nil: Add a helper for populating the null descriptor
|
||
- nvk: Create the zero page earlier in device creation
|
||
- nvk: Call into NIL to populate the null descriptor
|
||
- nouveau/winsys: Uninline nouveau_ws_bo_unmap()
|
||
- nak: Pull the QMD code out into its own crate
|
||
- nak: Make ShaderBin a public struct
|
||
- nak/meson: Rename libnak_bindings_gen to _libnak_bindings_rs
|
||
- nak: Add a bare HW shader runner
|
||
- nak: Add an implementation of the ACORN random number generator
|
||
- nak: Add dst_type decorations
|
||
- nak: Add a Foldable trait
|
||
- nak: Implement Foldable for OpIAdd3 and OpIAdd3X
|
||
- nak: Initialize DEBUG on-demand
|
||
- nak: Add a framework for running hardware tests
|
||
- nak: Add some opcode tests
|
||
- nak: Implement Foldable for OpPrmt
|
||
- nak: Implement Foldable for OpShf
|
||
- nak: Fix shf.l.high on Maxwell
|
||
- nak: Add 64-bit shift helpers
|
||
- nak: Add tests for 64-bit shifts
|
||
- nak: Use .wrap for 64-bit shifts
|
||
- nak: Fix shl64 for pre-Volta
|
||
- meson/megadriver: Don't invoke the megadriver script with no drivers
|
||
- nak/spill_values: Don't assume no trivial phis
|
||
- nak: Drop the old encode_sm*.rs files
|
||
- nak/sm50: Add control-flow sync ops
|
||
- nak: Add more NIR wrappers for walking the NIR CFG
|
||
- nak/sm50: Emit sync instructions for control-flow
|
||
- nak: Disable opt_jump_thread pre-Volta
|
||
- nak: Don't emit barrier ops for shared memory pre-Volta
|
||
- nak/sm70: Implement OpAtom with Dst::None as RED
|
||
- nak/sm50: Implement OpAtom with Dst::None as RED
|
||
- nak: Use the RED path for atomics with unused destinations
|
||
- nak: IMul and IMad are variable latency pre-Volta
|
||
- nak/calc_instr_deps: Set a delay of 13 for flow ops pre-Volta
|
||
- nak: Align shader data to the min CBuf alignment
|
||
- nak: Add tests for nak_builder::isetp64
|
||
- nak: Add tests for nak_builder::iadd64
|
||
- nak: Add tests for OpISetP
|
||
- nak/hw_tests: Add debug printing
|
||
- nak/sm50: Set the .x bit for isetp
|
||
- nak/builder: Fix isetp64 for pre-Volta
|
||
- nvk: Squash txq results to 0 for null descriptors
|
||
- nak: Fix OpCCtl encoding on Maxwell
|
||
- nak/nir: Don't emit CCtl for release barriers on Maxwell
|
||
- nvk: Don't push bound UBOs on Volta and earlier
|
||
- nak: Implement From<bool> for Pred
|
||
- nak/sm50: OpIAdd2 only supports one negate modifier
|
||
- nak/sm50: OpLop2 does not have .X
|
||
- nak: Add tests for OpLop2 and OpLop3
|
||
- nak/sm50: Fix the encoding of ffma with cbuf in src2
|
||
- nak: Implement depth clip control pre-Volta
|
||
- nvk: Use nvk_buffer_addr_range() for index buffers
|
||
- nvk: Fix zero-size vertex/index buffer bindings pre-Turing
|
||
- nak/sm50: Add support for OpAL2P
|
||
- nak: Move the runner into the main nak crate
|
||
- nak: Plumb through the call/return stack size
|
||
- nvk: Take CRS into account when allocating the SLM area
|
||
- nak/nir: Fix the reverse execlist iterator
|
||
- nak/nir: Add a few more NIR helpers
|
||
- nak/nir: Rework CRS handling
|
||
- nak: Add target labels to CRS pop instructions
|
||
- nak: Add an optimization for CRS ops
|
||
- nak: OpKill also needs a delay on Maxwell
|
||
- nak/sm50: Add execution delays for more ops
|
||
- nak: SrcMod::INeg cannot be folded
|
||
- nak: Add tests for nak_builder::ineg64
|
||
- nak/builder: Allow source modifiers in ineg64()
|
||
- nak: Lower iadd64(x, -y) to isub64(x, y) in NIR
|
||
- nak/hw_tests: Explicitly test equal cases in test_isetp64
|
||
- nak: Add tests for OpFlo
|
||
- nak: Assert no overflow when legalizing iadd with too many ineg modifiers
|
||
- nak: Add source and destination types for carry values
|
||
- nak/hw_tests: Respect src_mod and src_swizzle in op auto-tests
|
||
- nak/hw_tests: Use better test values for iadd tests
|
||
- nak: Add tests for OpIAdd2 and OpIAdd2X
|
||
- nak/sm50: Set the not bit for OpLop2::srcs[0]
|
||
- nak: Use OpFoldData::get_u32_bnot_src() more places
|
||
- nak: Add tests for OpPopC and enable SrcMod::BNot
|
||
- nak/nir: Preserve rounding modes in split_64bit_conversions()
|
||
- nak: Rename OpFSOut to OpRegOut
|
||
- nak: pre-Volta uses r0 to output the final GS handle
|
||
- nak: Add a PredSetOp::eval() helper
|
||
- nak: Add tests for OpPSetP
|
||
- nak: Non-constant offsets are allowed on patch loads
|
||
- nak: Rework isetp64 on sm50 again
|
||
- nak: Disallow isetp.x pre-Volta
|
||
- nak: Manually flush denorms for nir_op_fquantize2f16 pre-Volta
|
||
- nak/sm50: Set f2f.high
|
||
- nvk: Default to NAK on Maxwell+
|
||
- nvk: Disable sparse D32S8 cubes pre-Turing
|
||
- nak: Move the QMD code back into the main nak crate
|
||
- nak/qmd: Return the complete QMD dispatch size layout
|
||
- mme/fermi: Don't try to access zero-size std::vector
|
||
- nouveau/mme: Add support for MUL on Fermi
|
||
- nouveau/mme: Add support for multiplication on Fermi
|
||
- nak/hw_runner: Use byte_offset()
|
||
- nak/hw_tests: Use ptr::from_ref() and ptr::cast()
|
||
- nak: Use nak_builder::fsetp() in op_fquantize2f16
|
||
- nvk: Fix indirect dispatch on Turing+
|
||
- vulkan: Update XML and headers to 1.3.292
|
||
- nvk: Move nvk_sample_location to NAK
|
||
- nak: Sample locations are byte-aligned
|
||
- nak,nir: Rework gl_SampleMaskIn[] handling
|
||
- nvk: Don't force sample shading when gl_SampleMaskIn[] is read
|
||
- nvk: Set POST_PS_INITIAL_COVERAGE
|
||
- compiler/rust: Copy the Rust NIR wrappers from NAK
|
||
- compiler/rust/nir: Implement NIR types directly
|
||
- compiler/rust/nir: Add explicit lifetimes
|
||
- nak: Switch to the common NIR wrappers
|
||
- nak: Move DstsAsSlice::is_uniform() to its own trait
|
||
- nak: Replace the guts of Srcs/DstsAsSlice with a new AsSlice trait
|
||
- nak,compiler: Move AsSlice to common code
|
||
- nvk: Require color or depth/stencil attachment support for input attachments
|
||
- nvk: Support STORAGE_READ_WITHOUT_FORMAT on buffers
|
||
- nvk: Move descrptor structs into a separate header
|
||
- nvk: s/device/dev/ in nvk_buffer_view.c
|
||
- nvk/nvkmd: Advertise the usable VA range
|
||
- nvk: Add a VK_EXT_descriptor_buffer buffer view cache
|
||
- nvk: Refactor image intrinsic lowering a bit
|
||
- nvk: Lower descriptors for VK_EXT_descriptor_buffer buffer views
|
||
- nvk: Use nvk_edb_buffer_view_descriptor for EDB descriptor set layouts
|
||
- nvk: Use the EDB buffer view path with NVK_DEBUG=edb_bview
|
||
- nvk: Implement GetDescriptorLayoutSize/BindingOffsetEXT()
|
||
- nvk: Refactor some descriptor set helpers
|
||
- nvk: Implement GetDescriptorEXT
|
||
- nvk/descriptor_table: Add support for requesting a specific index
|
||
- nvk: Implement descriptor capture/replay
|
||
- nvk: Properly indent a comment
|
||
- nvk: Rework descriptor set bindings
|
||
- nvk: Implement descriptor buffer binding
|
||
- nvk: Add support for embedded immutable samplers
|
||
- zink: Align descriptor buffers to descriptorBufferOffsetAlignment
|
||
- nvk: Advertise VK_EXT_descriptor_buffer
|
||
- nvk: Invalidate MME data for descriptor buffers
|
||
- nak: Pass -C default-linker-libraries when building tests
|
||
- ci: Add NVK to debian-arm64-build-test
|
||
- vulkan: Add null descriptor bits to vk_pipeline_robustness_state
|
||
- nvk: Plumb the whole vk_pipeline_robustness_state through to nvk_ubo/ssbo_addr_format
|
||
- nvk: Enable shader bounds checking when nullDescriptor is enabled
|
||
- nouveau/headers: Add more nv_push asserts
|
||
- nouveau/winsys: Fix an undefined shift
|
||
- nouveau/mme: Fix add64 of immediates on Fermi
|
||
- nouveau/mme: Better handle the too many immedaties case on Turing+
|
||
- nouveau/mme: Don't allow PRMT and MTHD in the same op on Turing+
|
||
- nouveau/mme: Disassemble MERGE more like the mme_merge()
|
||
- nouveau/mme: Add NOT and AND_NOT ops
|
||
- nouveau/mme: Allow shifting immediates on Fermi
|
||
- nouveau/mme: Use a designated initializer in mme_fermi_decode()
|
||
- nouveau/mme/tu104: Break the state simulator into a pluggable component
|
||
- nouveau/mme/fermi: Break the state simulator into a pluggable component
|
||
- nouveau/mme: Add a pluggable state interface
|
||
- nouveau/mme/tu104: Properly handle mthd increment
|
||
- nouveau/mme: Add a generic simulator function
|
||
- nvk: Reserve MME scratch area for communicating with FALCON
|
||
- nvk/mme: Add a unit test framework for driver macros
|
||
- nvk: Emit NVK_MME_SCRATCH_DRAW_BEGIN as part of flush_ia_state()
|
||
- nvk: Increase the context init size
|
||
- nvk: Move the guts of BindIndexBuffer2KHR to a macro
|
||
- nvk: Move the guts of BindVertexBuffers2 into a macro
|
||
- nvk: Move setting VB enables and strides into macros
|
||
- nvk/mme: Add helpers for masked scratch updates
|
||
- nvk: Add macros to sort out tess parameters
|
||
- nvk: Move the ANTI_ALIAS_CONTROL logic to the MME
|
||
- nvk: Stop leaking the descriptor table in-use set
|
||
- nvk: Disable conditional rendering around CopyQueryPoolResults
|
||
- nvk: Expose a new nvk_compile_nir_shader() helper
|
||
- nvk: Refactor compute shader dispatch
|
||
- nvk: Add a helper for dispatching compute shaders
|
||
- nvk: Use the new dispatch helper for queries
|
||
- nvk: Treat NULL vertex/index buffer pointers as null descriptors
|
||
- nil,nvk: Disable modifiers for B10G11R11_UFLOAT and E5B9G9R9_UFLOAT
|
||
- nir: Preserve fp_fast_math in nir_opt_vectorize()
|
||
- v3dv: Update a debug message
|
||
- vtn: Move initialization of mesa_spirv_debug to a helper
|
||
- vtn: Allow SPIR-V debug options in release builds
|
||
- vtn: Add a debug flag to dump SPIR-V assembly
|
||
- v3dv: Drop the SPIR-V dumper
|
||
- radv: Use the SPIR-V printer in spirv_to_nir
|
||
- vulkan/pipeline: Handle VIEW_INDEX_FROM_DEVICE_INDEX_BIT
|
||
- lvp: Stash the PipelineCreateFlags2KHR in the pipeline
|
||
- dzn: Use vk_pipeline_shader_stage_to_nir()
|
||
- vulkan: Take a VkPipelineCreateFlags2KHR in vk_pipeline_*shader_stage*()
|
||
- vulkan: Handle VIEW_INDEX_FROM_DEVICE_INDEX_BIT in the runtime
|
||
- nvk: Fix a typo in a comment
|
||
- nvk: Use 4 bits per value for the anti alias values
|
||
- nvk: Hash minSampleShading in nvk_hash_graphics_state()
|
||
- nvk: Advertise VK_KHR_maintenance7
|
||
- nvk/nvkmd: Add an nvkmd_mem_init() helper
|
||
- nvk/nvkmd: Plumb map_out through to the back-end nvkmd_mem_ops::map
|
||
- nvk/nvkmd: Plumb map pointers through to back-end unmap funcs
|
||
- nvk/nvkmd: Plumb map_flags through to unmap/overmap
|
||
- nvk/nvkmd: Don't inline nvkmd_mem_[un]map()
|
||
- nvk/nvkmd: Split client and internal mappings
|
||
- nvk/nvkmd: Reference count internal mappings
|
||
- nvk: Don't do linear<->tiled copies for rendering suspend/resume
|
||
- nvk: Take depth image layer counts from the VkRenderingInfo
|
||
- nvk: Add all_linear to nvk_rendering_state
|
||
- vulkan: Add a helper for getting VkRenderingAttachmentLocaiontInfoKHR
|
||
- vulkan: Allow pColorAttachmentLocations == NULL in CmdSetRenderingAttachmentLocationsKHR()
|
||
- vulkan: Expose a vk_cmd_set_rendering_attachment_locations() helper
|
||
- nvk: Reset rendering attachment locations in BeginCommandBuffer/Rendering
|
||
- nvk: Emit SET_CT_SELECT based on the dynamic color location map
|
||
- nvk: Adverties VK_KHR_dynamic_rendering_local_read
|
||
- nak: Add a NAK_SUBGROUP_SIZE #define
|
||
- nak: Move subgroup_id and num_subgroups to lower_system_values
|
||
- nak: Shuffle gl_LocalInvocationID for quad derivatives
|
||
- nvk: Advertise VK_NV/KHR_compute_shader_derivatives
|
||
- nvk: Only advertise compute_shader_derivatives when NAK is used
|
||
- vulkan: Handle variable-length property arrays more generically
|
||
- nil: Rework calculations of the sizes of miplevels
|
||
- nil: Return a more accurate size in level_size_B
|
||
- nil: Assert array_len == 0 in Extent4D<Bytes>::size_B()
|
||
- nvk: Advertise VK_KHR_shader_float_controls2
|
||
- nvk: Silence a maybe-uninitialized warning
|
||
- nvk: Add an NVK_DEBUG=gart flag
|
||
- vulkan: Add a stages field to vk_pipeline
|
||
- vulkan: Add a vk_pipeline_ops::get_shader method
|
||
- vulkan/pipeline: Patch through INDIRECT_BINDABLE_BIT
|
||
- nvk: Fill out shader push data at compile time
|
||
- nvk: Use VkShaderStageFlags for shaders_dirty
|
||
- nvk: Disable CBuf pushing when INDIRECT_BINDABLE_BIT is set
|
||
- nvk: Add an implementation of VkIndirectExecutionSetEXT
|
||
- nvk/nvkmd: Always map read/write for internal maps
|
||
- nvk/nvkmd: Add a mechanism to track and look up nvkmd_mems
|
||
- nvk: Track and dump device-generated commands if possible
|
||
- nvk: Move flush_push_descriptors to the front of flush_state()
|
||
- nvk: Export graphics state flushing helpers
|
||
- nvk: Expose an nvk_cmd_flush_cs_qmd() helper
|
||
- nvk: Implement CmdProcess/ExecuteGeneratedCommandsEXT
|
||
- nvk: Handle indirect dispatch in pipeline barriers
|
||
- nvk: Advertise VK_EXT_device_generated_commands
|
||
- nvk: Handle aspects in D32_S8_UINT copies
|
||
- nvk: Assume a GOB height of 8 for copies
|
||
- nvk,nil: Replace gob_height_is_8 and is_tiled with a new GOBType enum
|
||
- vulkan: Use SEVERITY_WARNING_BIT for vk_error*()
|
||
- nil: Add a couple Extent4D and Offset4D methods
|
||
- nil: Add tiled memcpy helpers
|
||
- vulkan: Add a comment explainint SEVERITY_WARNING_BIT for vk_error*()
|
||
- nvk/nvkmd: Stop leaking the primary FD
|
||
- nir: Add an option to lower quad vote
|
||
- nak,nvk: Implement VK_KHR_shader_quad_control
|
||
- nvk: Fix a comment in SET_VIEWPORT_CLIP_CONTROL
|
||
- vulkan/queue: Don't use vk_semaphore in threaded payload stealing
|
||
- vulkan/queue: Move has_binary_permanent_semaphore_wait into the sumbit struct
|
||
- vulkan/queue: Use _mem_signal_temp instead of signal_mem_sync
|
||
- vulkan: Add a vk_queue_submit_has_bind() helper
|
||
- vulkan/queue: Use a builder pattern for vk_queue_submit
|
||
- vulkan/queue: Split vk_queue_submit into create and submit
|
||
- vulkan/queue: Merge submits when possible
|
||
- nouveau/mme: Test ifs with immediate parameters
|
||
- nouveau/mme: Evaluate methods immediately in the Turing sim
|
||
- nouveau/mme: Turing can't have emit and state in the same op
|
||
- nil: Add a helper to get samples from a SampleLayout
|
||
- nvk: Add an nvk_cmd_emit_sample_layout() helper
|
||
- nvk: Rework setup of sample masks
|
||
- nouveau/class_parser: Add more method arrays
|
||
- nil: Add D3D sample layouts
|
||
- nvk: Support D3D sample modes
|
||
- nil: Use D3D sample modes by default
|
||
- nvk: Implement GetPhysicalDeviceFragmentShadingRatesKHR()
|
||
- nvk: Initialize VARIABLE_PIXEL_RATE_SAMPLE_ORDER
|
||
- vulkan: Add a vk_fragment_shading_rate_is_disabled() helper
|
||
- nvk: Rework setting of min_sample_shading
|
||
- nvk: Emit FSR state
|
||
- nvk: Add support for binding fragment shading rate images
|
||
- nak: Get rid of some dead code warnings
|
||
- nak: Rename SV_VERTEX_COUNT to PRIM_TYPE
|
||
- nak: Add support for gl_ShadingRateEXT
|
||
- nak: Pass a nak_compiler into sysval and attrib helpers
|
||
- nak: Add support for gl_PrimitiveShadingRateEXT
|
||
- nvk: Set VARIABLE_PIXEL_RATE_SHADING_TABLE_SELECT
|
||
- nvk: Advertise VK_KHR_fragment_shading_rate
|
||
- vulkan/queue: Check for _mem_signal_temp before we submit
|
||
- nvk: Add an nvkmd_engines bitfield to nvk_queue
|
||
- nvk: Only set texture/sampler tables and SLM for enabled engines
|
||
- nvk: Re-sort the features table
|
||
- nvk: Advertise 64-bit atomics on buffer views
|
||
- nak: Remove annotations before calc_instr_deps()
|
||
- nak: Handle annotations in legalization
|
||
- compiler/rust: Enable unit tests
|
||
- compiler/rust: Add a unit test for the memstream abstraction
|
||
- compiler/rust: Fix a bad cast in the memstream abstraction
|
||
- nvk: Only wait on the upload queue if there are commands
|
||
- nvk: Only submit the state pushbuf if there are commands
|
||
- nvk: move nvk_queue_state_update to nvk_queue_submit_exec
|
||
- nvk: Use nvk_queue_submit_simple() for nvk_queue_state_update()
|
||
- nak/bindings: Use an enum for IOCTL numbers
|
||
- nvk: Fix host copies for array images
|
||
- nil: Add a nil_sample_offset() helper
|
||
- nvk: Put a sample map in the descriptor for MSAA storage images
|
||
|
||
Felix DeGrood (5):
|
||
|
||
- anv: fix src_hash dumping for compute shaders
|
||
- anv/measure: ignore events from reused command buffers
|
||
- intel/measure: add nogl feature
|
||
- intel/measure: defer file open until first write
|
||
- intel/measure: increase size of filename malloc to account for \\0
|
||
|
||
Francisco Jerez (8):
|
||
|
||
- intel/dev: Add devinfo flag for TBIMR push constant workaround.
|
||
- intel/brw: Implement null push constant workaround.
|
||
- anv/gfx12.5: Pass non-empty push constant data to PS stage for TBIMR workaround.
|
||
- iris/gfx12.5: Pass non-empty push constant data to PS stage for TBIMR workaround.
|
||
- iris: Pin pixel hashing table BO from iris_batch submission instead of from iris_state.
|
||
- intel/brw/gfx12.5+: Fix IR of sub-dword atomic LSC operations.
|
||
- intel/brw/xe2+: Adjust performance analysis divergence weight due to EU fusion removal.
|
||
- intel/fs/xe2: Fix up subdword integer region restriction with strided byte src and packed byte dst.
|
||
|
||
Frank Binns (9):
|
||
|
||
- pvr: minor vulkan driver formatting fixes
|
||
- pvr: drop 'Imagination' from deviceName
|
||
- pvr: fix image size calculation when mipLevels is 1
|
||
- pvr: fix bo refcounting for imported buffers
|
||
- pvr: fix use of uninitialised value when setting up tex info swizzle
|
||
- pvr: remove set but not used struct member
|
||
- pvr: add handling for retry error in pvr_srv_rgx_submit_transfer2()
|
||
- pvr: add handling for retry error in pvr_srv_rgx_kick_compute2()
|
||
- pvr: ensure stencil clear value fits TA_STATE_ISPA.sref field
|
||
|
||
Friedrich Vock (5):
|
||
|
||
- radv/rt: Only do ploc atomicCompSwap once per workgroup
|
||
- radv/rt: Don't atomicAdd local prefix sums
|
||
- aco: Fix 1D->2D dispatch conversion on <gfx9
|
||
- radv/rt: Skip all AABB code when no_skip_aabbs is not set
|
||
- vulkan/rmv: Correctly set heap size
|
||
|
||
GKraats (8):
|
||
|
||
- i915g: fix generation of large mipmaps
|
||
- i915g: fix mipmap-layout for npots
|
||
- i915g: fix max_lod at mipmap-sampling
|
||
- i915g: Screen corruption with ENOBUFS caused by fence register shortage
|
||
- i915g: fix count of buffers at i915_drm_batchbuffer_validate_buffers
|
||
- X11: fix crash of gnome-shell if mesa is compiled with legacy-x11=dri2
|
||
- i915g: fix texture3d npot mipmaps
|
||
- i915g: fix GL_TEXTURE_BASE_LEVEL MAX_LEVEL MIN_LOD and MAX_LOD
|
||
|
||
Ganesh Belgur Ramachandra (7):
|
||
|
||
- radeonsi: fix eptich on chips without image opcodes (e.g. gfx940)
|
||
- amd/common: skip lane size determination for chips without image opcodes (e.g. gfx940)
|
||
- amd: fix incorrect PIPE_INTERLEAVE_BYTES size for CDNA chips
|
||
- radeonsi: fix decode corruptions with AMD_IMAGE_OPCODES=false on non-CDNA based chips
|
||
- amd,radeonsi: use new pass manager to handle midend optimizations
|
||
- amd,radeonsi: reduce legacy::PassManager use to only run backend passes
|
||
- amd: remove the redundant target library info instance in LLVM compiler
|
||
|
||
Georg Lehmann (126):
|
||
|
||
- radv/rt: remove one VALU from traversal loop
|
||
- spirv: ignore more function param decorations
|
||
- aco/optimizer: update temp_rc when converting to uniform bool alu
|
||
- aco: add CompilationProgress::after_lower_to_hw
|
||
- aco: validate temp_rc
|
||
- aco/gfx11+: apply neg to vinterp
|
||
- aco: reorder dpp for ddx/ddy
|
||
- aco/gfx11.5+: use vinterp for fddx/fddy
|
||
- aco/gfx11+: don't use VOP3 v_swap_b16
|
||
- aco: test xor swap16 path
|
||
- nir/peephole_select: ignore masked/quad swizzle without fetch_inactive
|
||
- aco: optimize 64bit find_lsb/find_msb
|
||
- aco: optimize dd[xy]_fine if it's only used by abs
|
||
- nir/lower_int64: replace uadd_sat with ior for find_lsb64 and ufind_msb64
|
||
- aco/gfx10+: set lateKill for sgprs used by wave64 VALU writing a mask
|
||
- aco: only set latekill in live_var_analysis
|
||
- aco/tests: don't test dpp constant propagation with row shift
|
||
- aco/tests: do not use add to tests neg modifer
|
||
- aco/print_ir: use neg() for constants
|
||
- aco/tests: parse neg(constant) in vop3p test
|
||
- aco/tests: allow literals with resolved swizzles in vop3p test
|
||
- aco/tests: do not use mul with constant to tests neg modifier
|
||
- nir: optimize pack_uint_2x16 of pack_half(a, 0)
|
||
- zink: switch to derivative intrinsics
|
||
- aco/ra: unconditionally replace literal with sgpr when promoting to VOP3
|
||
- aco/ra: use parallelcopy to copy literal instead of s_mov
|
||
- aco/ra: fix copying 64bit literal to sgprs
|
||
- aco/ra: always reuse def register for literal copy
|
||
- aco: use v_cvt_pk_u8_f32 for f2u8
|
||
- nir/opt_algebraic: reassociate constants in ior(iand) chains
|
||
- radv/ci: update trace checksums
|
||
- aco/ra: do not reuse killed vcc
|
||
- nir/opt_sink: do not sink load_ubo_vec4 out of loops
|
||
- nir/opt_sink: do not sink inverse_ballot out of loops
|
||
- nir: merge out of loop decision with nir_can_move_instr logic
|
||
- nir/instr_set: fix fp_fast_math
|
||
- aco/isel: add function to create builder for alu
|
||
- aco/ir: add float control definition flags
|
||
- aco/vn: handle sz/inf/nan preserve
|
||
- aco/isel: set per instruction float control modes
|
||
- aco/tests: use new float control flags
|
||
- aco/optimizer: use new per definition float control flags
|
||
- aco: remove per block inf/nan/sz control
|
||
- aco/tests: fix omod signed zero tests
|
||
- radv: advertise VK_KHR_shader_float_controls2
|
||
- nir: add amd shared append/consume
|
||
- amd/nir: add ac_nir_opt_shared_append
|
||
- aco/assembler: support ds_append/ds_*_addtid
|
||
- aco: implement nir_shared_append/consume_amd
|
||
- aco: use ac_nir_opt_shared_append
|
||
- aco/isel: use upper bound for v_mul_u32_u24
|
||
- aco/gfx10+: work around non uniform ds_append wave64 result
|
||
- spirv: explicitly lower derivatives to zero
|
||
- amd: add offset to is_subgroup_invocation_lt_amd
|
||
- ac/lower_ngg: use is_subgroup_invocation_lt_amd offset
|
||
- aco/ssa_elimination: don't check for VALU limitation when optimizing branching sequence
|
||
- aco/ssa_elimination: optimize branching sequence with SALU that has multiple definitions
|
||
- aco: optimize lanecount_to_mask
|
||
- nir: sink is_subgroup_invocation_lt_amd
|
||
- nir: replace nir_opt_remove_phis_block with a single source version
|
||
- nir: make nir_instr_clone usable with load_const and undef
|
||
- nir/opt_remove_phis: rematerialize constants
|
||
- radv: remove nir_opt_reuse_constants call
|
||
- nir: delete nir_opt_reuse_constants
|
||
- vc4/ci: documment new failure
|
||
- aco: use s_pack_hl for shfr16
|
||
- aco: use s_pack_lh for bitfield_select(0xffff)
|
||
- radv: call nir_opt_remove_phis in radv_optimize_nir_algebraic
|
||
- nir/opt_move_discards_to_top: handle ddx/ddy intrinsics
|
||
- elk: remove alu fddx/fddy check
|
||
- ir3: remove alu fddx/fddy check
|
||
- spirv: remove alu fddx/fddy from comment
|
||
- mesa/prog_to_nir: use derivative builder
|
||
- nir: always emit ddx intrinsics
|
||
- nir: remove uses_fddx_fddy
|
||
- nir: remove alu ddx/ddy
|
||
- nir: remove nir_op_is_derivative
|
||
- aco/ir: add function to parse depctr waits
|
||
- aco: do not track ALU delay across jumps
|
||
- aco/insert_delay_alu: consider more implicit waits
|
||
- aco/insert_delay_alu: do not delay lane mask fast forwarding
|
||
- aco/statistics: update branch issue cycles
|
||
- aco/print_ir: use parse_depctr_wait
|
||
- meson: remove selinux option
|
||
- ci: remove selinux from containers
|
||
- aco: fix 64bit extract_i8/extract_i16
|
||
- aco/insert_exec: don't pretend WQMState is a bit mask
|
||
- aco/insert_exec: replace pair with a named struct
|
||
- aco/insert_exec: remove stray break_cond variable
|
||
- aco/insert_exec: untangle add_branch_code control flow
|
||
- aco/insert_exec: remove unused includes
|
||
- aco/ir: rework Operand equality to return true for equal fixed non-temp ops
|
||
- aco/insert_exec: remove get_exec_op
|
||
- nir/opt_algebraic: optimize -0.0 + a
|
||
- aco/ssa_elimination: also optimize branching sequence with s_and without saveexec
|
||
- aco/insert_exec: avoid phis for masks in exec
|
||
- aco/insert_exec: don't always reset top exec
|
||
- aco/insert_exec: reuse old exec temp in loop pre-header
|
||
- aco/insert_exec: reuse old exec temp instead using s_and_saveexec
|
||
- aco/ssa_elimination: don't avoid saving exec when optimizing branching sequence
|
||
- radv: don't use v_mqsad_u32_u8 on gfx7
|
||
- aco: remove heuristic that restricts VOP2/C with 2 sgprs
|
||
- aco/ssa_elimination: add instr_accesses helper
|
||
- aco/ssa_elimination: don't assume exec writes can be removed based on block kind
|
||
- aco: optimize conditional divergent breaks at the end of loops
|
||
- aco: move exec copy out of waterfall loops
|
||
- ac/llvm: handle shared atomic base offset
|
||
- nir,radv: optimize shared atomic offsets
|
||
- nir/opt_algebraic: create ubfe with non constant mask
|
||
- radv: use nir_opt_fragdepth
|
||
- nir: add SYSTEM_VALUE_PIXEL_COORD
|
||
- radv: add support for load_pixel_coord
|
||
- ac/llvm: implement load_pixel_coord
|
||
- aco: implement load_pixel_coord
|
||
- nir: add nir_opt_frag_coord_to_pixel_coord
|
||
- radv: use nir_opt_frag_coord_to_pixel_coord
|
||
- nir/opt_frag_coord_to_pixel_coord: optimize trunc/floor
|
||
- nir/lower_wpos_ytransform: remove unnecessary state variable
|
||
- nir/lower_wpos_ytransform: remove redundant state shader
|
||
- nir/lower_wpos_ytransform: clean up baryc_at_offset
|
||
- nir/lower_wpos_ytransform: clean up sample_pos
|
||
- nir/lower_wpos_ytransform: clean up wpos_adjustment
|
||
- nir/lower_wpos_ytransform: use more typical pass structure
|
||
- nir/lower_wpos_ytransform: remove reference to long removed TGSI code
|
||
- nir/lower_fragcoord_wtrans: use intrinsics_pass
|
||
- amd: lower gl_FragCoord.w rcp in NIR
|
||
|
||
George Ouzounoudis (3):
|
||
|
||
- d3d12: Split memory size info to device and system memory
|
||
- d3d12: Keep track of size and number of evictions
|
||
- d3d12: Add support for GL_NVX_gpu_memory_info and GL_ATI_meminfo
|
||
|
||
Gert Wollny (2):
|
||
|
||
- nir/opt_algebraic: Allow two-step lowering of ftrunc\@64 to use ffract\@64
|
||
- Revert: r600/sfn: call nir_lower_doubles explicitely"
|
||
|
||
Guilherme Gallo (26):
|
||
|
||
- bin/ci: crnm: Deal with \`created` status
|
||
- bin/ci: crnm: Fix pipeline monitoring skipping
|
||
- bin/ci: crnm: Reduce trace cluttering
|
||
- bin/ci: crnm: Don't overwrite the global sets
|
||
- bin/ci: crnm: Use frozen set for statuses
|
||
- ci/a618: Fix zink-tu-a618-full rules
|
||
- ci/freedreno: Rebalance limozeen jobs
|
||
- ci/freedreno: Replace 2 limozeen with kingoftown
|
||
- ci/freedreno: Rebalance jobs via fraction and parallel
|
||
- ci/freedreno: Add a618_gl_full
|
||
- ci/amd: Rebalance radeonsi-stoney-gl:x86_64
|
||
- ci/amd: Rebalance jobs via DEQP_FRACTION
|
||
- ci/intel: Rebalance jobs via parallel
|
||
- ci/anv: Introduce missing farm var for ADL jobs
|
||
- ci/anv: Update xfiles
|
||
- ci/lava: Make perf jobs use pyutils artifact
|
||
- ci/zink: Add skips for zink-anv-(adl|tgl)
|
||
- ci/radv: Rebalance stoney-vkcts
|
||
- ci: Use merge-skips files during merge pipelines
|
||
- ci/amd: Track merge-skips.txt changes
|
||
- ci/amd: Reassign radv-raven-traces from zork to dalboz
|
||
- ci/panfrost: Increment panfrost-g52-gl parallel no.
|
||
- ci/freedreno: Rebalance limozeen jobs
|
||
- ci/turnip: Rebalance kingoftown jobs
|
||
- ci/lava: Fix lava-tags parsing
|
||
- ci/lava: Fix LAVA job definition tests
|
||
|
||
Gurchetan Singh (145):
|
||
|
||
- vulkan/runtime: introduce vk_meta_object_list
|
||
- tu: use os_get_total_physical_memory(..)
|
||
- vulkan: fix glibc AOSP build
|
||
- vulkan/android: change vk_enum_defines.h path
|
||
- gfxstream: fix Linux guest build
|
||
- gfxstream-guest: add precise virtio-gpu capabilities
|
||
- gfxstream: use ASG for Vulkan streams only
|
||
- gfxstream: gfxstreamCapset --> vulkanCapset
|
||
- gfxstream: query non-Vulkan caps from the guest
|
||
- gfxstream/guest: add GetRingParamsFromCapset function
|
||
- gfxstream/guest: use GetRingParamsFromCapset
|
||
- gfxstream: nuke unused functions
|
||
- gfxstream: simplify VK_HOST_CONNECTION macro
|
||
- gfxstream: don't use renderControlEnc for Vulkan
|
||
- gfxstream: nuke ResourceTracker::Impl
|
||
- gfxstream: fix meson build system issues
|
||
- gfxstream/guest: pass through Android process names
|
||
- gfxstream: import Vulkan-docs from 463f8c
|
||
- gfxstream: vulkan-Docs: import the cerealgenerator
|
||
- gfxstream: vulkan-docs: track feature type
|
||
- gfxstream: vulkan-docs: add cereal generator
|
||
- gfxstream: genvk: add support for vk_gfxstream.xml
|
||
- gfxstream: vulkan-docs: don't auto-generate vk_android_native_buffer
|
||
- gfxstream: add generate-gfxstream-vulkan.sh
|
||
- gfxstream: add support for VK_KHR_maintenance5 + VK_EXT_host_image_copy
|
||
- gfxstream: use standard Vulkan headers and new auto-gen flow
|
||
- gfxstream: look for common/opengl headers
|
||
- gfxstream: cereal: [moduleType] --> [(moduleType, moduleName)]
|
||
- gfxstream: cereal: SUPPORTED_WRAPPERS --> SUPPORTED_MODULES
|
||
- gfxstream: codegen EXT_swapchain_maintenance1 + KHR_swapchain host only
|
||
- gfxstream: mesa: import Mesa
|
||
- gfxstream: mega-change to support guest Linux WSI with gfxstream
|
||
- gfxstream: nuke goldfish_vk_handlemap
|
||
- gfxstream: use absolute path for some code
|
||
- genvk: options.filename --> args.target
|
||
- gfxstream: codegen: allow generating host/guest autogen separately
|
||
- gfxstream: guest: cpp_args --> guest_cpp_args
|
||
- gfxstream: reduce use of PLATFORM_SDK_VERSION
|
||
- gfxstream: PLATFORM_SDK_VERSION --> ANDROID_API_LEVEL
|
||
- gfxstream: don't forward declare ProcessResources
|
||
- gfxstream: nuke VIRTIO_GPU flag
|
||
- gfxstream: nuke HOST_BUILD
|
||
- gfxstream: add GFXSTREAM_ENABLE_GUEST_GOLDFISH
|
||
- gfxstream: add REQUIRED_TYPES list
|
||
- gfxstream: always initialize rcEncoder and other things
|
||
- gfxstream: build the test layer via Meson
|
||
- gfxstream: generate development ICD for meson build
|
||
- gfxstream: always advertise device memory extension guest side
|
||
- gfxstream: convert some Vulkan to mesa_log(..)
|
||
- gfxstream: valgrind fix
|
||
- gfxstream: add VK_EXT_vertex_attribute_divisor
|
||
- Revert^2 "gfxstream: add VK_EXT_vertex_attribute_divisor"
|
||
- Revert "gfxstream: Add vkCmdPipelineBarrier2 implementation in VkDecoderGlobalState"
|
||
- Revert "Revert "gfxstream: Add vkCmdPipelineBarrier2 implementat..."
|
||
- gfxstream: ResourceTracker: fix mesa_loge(..)
|
||
- gfxstream: guest: add VK_EXT_color_write_enable
|
||
- gfxstream: guest: add XCB Surface support
|
||
- gfxstream: guest: support blobCmd + blobCmdSize
|
||
- gfxstream: don't define hasDedicatedImage + hasDedicatedBuffer multiple times
|
||
- Revert "Add snapshot support for vkBindImageMemory2"
|
||
- gfxstream: guest: enable virtio-gpu kumquat
|
||
- gfxstream: host + guest: bump up version of the protocol
|
||
- gfxstream: linux guest: support client-side blob allocations
|
||
- gfxstream: guest: fix useless return statement
|
||
- gfxstream: guest: simpler get param for kumquat
|
||
- gfxstream: host: introduce vkGetSemaphoreGOOGLE
|
||
- gfxstream: guest: add GFXSTREAM_ACQUIRE_SYNC protocol
|
||
- gfxstream: guest: add experimental fence passing APIs
|
||
- gfxstream: guest: mBlobId -> mAtomicId
|
||
- gfxstream: guest: start using new vkGetSemaphoreGOOGLE apis
|
||
- gfxstream: kumquat: unmap memory when needed
|
||
- gfxstream: guest: fix kumquat build
|
||
- gfxstream: vulkan: fix issue with GCC
|
||
- gfxstream: guest: nuke AndroidLock
|
||
- gfxstream: guest: nuke aemu/base/Optional.h
|
||
- gfxstream: guest: nuke unused functions
|
||
- gfxstream: guest: make sure signalSemaphoreValueCount is correct
|
||
- gfxstream: guest: vk_CmdBeginTransformFeedbackEXT fix
|
||
- gfxstream: host: don't abort when unbox fails
|
||
- gfxstream: guest: add support VK_FORMAT_A2R10G10B10_UNORM_PACK32 Wayland buffers
|
||
- gfxstream: guest: use MESA_SCOPE_TRACE
|
||
- gfxstream: guest: nuke more AEMU base stuff
|
||
- gfxstream: guest: kumquat: fixes in preparation for AOSP Kumquat testing
|
||
- gfxstream: nuke set instance for testing
|
||
- gfxstream: nuke libplatform stub
|
||
- gfxstream: guest: plumb descriptor all the way through the HostConnection
|
||
- gfxstream: connect to the specificied kumquat socket
|
||
- gfxstream: guest: add vkInitializeKumquat
|
||
- gfxstream: guest: nuke Android HealthMonitor
|
||
- gfxstream: guest: plumb the descriptor everywhere
|
||
- gfxstream: end2end: nuke RutabagaLayer, use kumquat for testing
|
||
- gfxstream: simplify GLESv1/GLESv2/EGL/Vulkan builds
|
||
- gfxstream: nuke ErrorLog.h
|
||
- gfxstream: nuke AlignedBuf.h
|
||
- gfxstream: nuke mWorkPool
|
||
- gfxstream: guest: nuke vk_enum_string_helper.h
|
||
- gfxstream: guest: nuke Fuchsia Goldfish Address Space
|
||
- gfxstream: guest: nuke unnecessary Android properties
|
||
- gfxstream: guest: nuke goldfish_address_space_android.impl
|
||
- gfxstream: guest: mesa-ify GoldfishAddressSpace
|
||
- gfxstream: guest: mesa-ify libplatform_virtgpu
|
||
- gfxstream: guest: add virtgpu suffix
|
||
- gfxstream: guest: remove EmulatorFeatureInfo dependency in libgfxstream_vulkan.so
|
||
- gfxstream: guest: simplify Vulkan sequence number logic
|
||
- gfxstream: sync command buffer api with Mesa upstream
|
||
- gfxstream: move EmulatorFeatureInfo and ExtendedRenderControl
|
||
- gfxstream: guest: remove dependency on EmulatorFeatureInfo
|
||
- gfxstream: add processPipeInit and connect to IOStream
|
||
- Reapply "gfxstream: nuke mWorkPool"
|
||
- gfxstream: guest: fix timeout issues
|
||
- gfxstream: guest: introduce Gralloc::getGrallocType()
|
||
- gfxstream: guest: remove renderControl dependency from Gralloc
|
||
- gfxstream: GrallocMinigbm uses platform_virtgpu
|
||
- gfxstream: mesa-ify libgfxstream_android
|
||
- gfxstream: delete duplicate virtgpu_drm.h
|
||
- gfxstream: guest: don't rely on HostConnection for platform helpers
|
||
- gfxstream: guest: introduce GfxStreamConnectionManager
|
||
- gfxstream: guest: vulkan: use connection manager
|
||
- gfxstream: guest: introduce QemuPipeStream stub
|
||
- gfxstream: guest: fully mesa-ify vulkan_enc
|
||
- gfxstream: guest: remove unnecessary includes in VkEncoder
|
||
- gfxstream: guest: remove unneeded dependencies in meson build
|
||
- gfxstream: guest: vulkan: use hardware/google/aemu
|
||
- gfxstream: guest: vulkan: don't assume Android
|
||
- gfxstream: guest: nuke android_compat directory
|
||
- gfxstream: move goldfish sync
|
||
- gfxstream: guest: Gralloc --> GfxStreamGralloc
|
||
- gfxstream: don't assert without host
|
||
- gfxstream: move include outside of ANDROID
|
||
- gfxstream: codegen: don't require clang-format
|
||
- gfxstream: delete unneccessary code
|
||
- gfxstream: build time guest autogeneration
|
||
- gfxstream: re-license files
|
||
- gfxstream: re-license cereal
|
||
- gfxstream: simplify codegen
|
||
- gfxstream: moar relicense
|
||
- mesa: gfxstream: nuke vk_icd_*
|
||
- mesa: add gfxstream-experimental to -Dvulkan-drivers
|
||
- gfxstream: add clang-format
|
||
- gfxstream: nuke util function
|
||
- gfxstream: use util/libsync
|
||
- util: add sync_fence_info
|
||
- gfxstream: use sync_fence_info
|
||
- gfxstream: move generate-gfxstream-vulkan.sh script
|
||
- gfxstream: update Kumquat API
|
||
|
||
Hailin Zhang (1):
|
||
|
||
- Vulkan: fix dstArrayElement index wrap issue
|
||
|
||
Hans-Kristian Arntzen (4):
|
||
|
||
- wsi/common: Do not update present mode with MESA_VK_WSI_PRESENT_MODE.
|
||
- wsi/x11: Bump maximum number of outstanding COMPLETE events.
|
||
- radv: Always make sure to write the pipeline binary key.
|
||
- vulkan/wsi/wayland: Use X11-style image count strategy when using FIFO.
|
||
|
||
Hsieh, Mike (4):
|
||
|
||
- amd/vpelib: Remove deprecated update_3dlut flag
|
||
- amd/vpelib: Refactor OPP registers
|
||
- amd/vpelib: Add cache mechanism for 3D Lut command
|
||
- amd/vpelib: Refactor 3D LUT code
|
||
|
||
Hyunjun Ko (25):
|
||
|
||
- intel/genxml: add a missing value for MFX_SURFACE_STATE
|
||
- intel/genxml: change the length of MFX_QM_STATE
|
||
- intel/genxml: update VDENC instructions
|
||
- vulkan/video: adds a bitstream writer of h264 slice header
|
||
- anv/query: handle VK_QUERY_TYPE_VIDEO_ENCODE_FEEDBACK_KHR
|
||
- anv/video: remove unnecessary macros
|
||
- anv/video: implemnt VkGetEncodedVideoSessionParametersKHR
|
||
- anv/video: initial support for h264 encoding
|
||
- anv: enable VK_KHR_video_encode_queue and VK_KHR_video_encode_h264
|
||
- intel/genxml: Add missing fields for HCP_SLICE_STATE
|
||
- intel/genxml: fix the length of VDENC_DS_REF_SURFACE_STATE
|
||
- intel/genxml: Adds more VDENC commands
|
||
- intel/genxml: fix some length of HCP_FQM_STATE
|
||
- intel/genxml: adds a value of reference pic to HCP_SURFACE_STATE
|
||
- intel/decoder: Handle HCP_PAK_INSERT_OBJECT
|
||
- vulkan/video: adds a bitstream writer of h265 slice header
|
||
- anv/query: consider codec when querying the encoding status.
|
||
- anv/video: initial support for h265 encoding
|
||
- anv: support h265 encoding
|
||
- zink: walk the chain of resources for multi-planar formats.
|
||
- anv/video: support VK_VIDEO_ENCODE_RATE_CONTROL_MODE_DISABLED_BIT_KHR.
|
||
- anv: consider VK_VIDEO_CODEC_OPERATION_ENCODE_H264_BIT_KHR when allocating mv storgae.
|
||
- anv: support for inline query for vulkan video
|
||
- anv: support VK_IMAGE_CREATE_VIDEO_PROFILE_INDEPENDENT_BIT_KHR
|
||
- anv: enable KHR_video_maintenance1
|
||
|
||
Iago Toral Quiroga (22):
|
||
|
||
- v3d: move TLB blit checks to a helper
|
||
- v3d: support blitting straight from tile buffer
|
||
- v3d: add a faster TLB blit path
|
||
- v3d: do not expose EXT_float_blend
|
||
- broadcom/compiler: emit instructions producing flags earlier
|
||
- broadcom: implement nir_op_uclz
|
||
- broadcom: drop backend implementation of nir_op_ufind_msb
|
||
- broadcom/compiler: skip small immediates optimization on vpm instructions
|
||
- broadcom/compiler: rename is_ldunif_dst to try_rf0
|
||
- broadcom/compiler: avoid register conflict with ldunif(a) and ldvary
|
||
- broadcom/compiler: don't add const offset to unifa if it is 0
|
||
- broadcom/compiler: don't use small immediates in geometry stages
|
||
- v3d: support shader precompile for compute shaders
|
||
- nir: make fclamp_pos_mali and fsat_signed_mali opcodes generic
|
||
- broadcom: add missing opcodes for fmov on mul alu for V3D 7.x
|
||
- broadcom: add helpers to identify availability of new unpack modifiers
|
||
- broadcom/compiler: implement NIR mali opcodes for clamping
|
||
- broadcom/compiler: generate mali opcodes for clamping on Pi5
|
||
- broadcom: fix pairing tmu lookup with previous ldtmu
|
||
- v3dv: drop device_id field
|
||
- v3dv: rewrite device identification
|
||
- v3dv: fix leak during device initialization
|
||
|
||
Ian Forbes (1):
|
||
|
||
- driconf: Re-enable GLX_EXT_buffer_age on vmwgfx
|
||
|
||
Ian Romanick (48):
|
||
|
||
- nir/range_analysis: Fix errors in fmin and fmax tables
|
||
- nir/algebraic: Optimize some bit operation nonsense observed in some shaders
|
||
- nir/algebraic: Optimize some masking of extract_u8 operations
|
||
- intel/brw: Fix undefined shift by 64 of uint64_t in brw_compute_first_urb_slot_required
|
||
- intel/brw: Fix undefined left shift of negative value in update_uip_jip
|
||
- intel/brw: Fix undefined left shift of large UW value in brw_imm_uw
|
||
- intel/brw: Fix undefined left shift of negative value in brw_texture_offset
|
||
- intel/elk: Fix undefined shift by 64 of uint64_t in elk_compute_first_urb_slot_required
|
||
- intel/elk: Fix undefined left shift of negative value in update_uip_jip
|
||
- intel/elk: Fix undefined left shift of large UW value in elk_imm_uw
|
||
- intel/elk: Fix undefined left shift of negative value in elk_texture_offset
|
||
- intel/brw: Eliminate dead flag writes
|
||
- intel/brw: Don't propagate saturate to an instruction that writes flags
|
||
- intel/elk: Don't propagate saturate to an instruction that writes flags
|
||
- intel/brw: Small code refactor in brw_fs_opt_saturate_propagation
|
||
- intel/brw: Use def analysis for simple cases of saturate propagation
|
||
- intel/brw: Move fsat instructions closer to the source
|
||
- intel/brw: Make 16-bit ishl, ishr, and ushr SSA friendly
|
||
- intel/brw: Make ifind_msb SSA friendly
|
||
- nir/algebraic: Improve some find_lsb and ifind_msb patterns
|
||
- anv: Protect against OOB access to anv_state_pool::buckets
|
||
- anv: Larger memory pools for huge shaders
|
||
- intel/brw/xe2: Allow int64 conversions
|
||
- nir/algebraic: Remove incorrect bfi of iand pattern
|
||
- nir/divergence: resource_intel is less divergent than you thought
|
||
- intel/brw: Copy prop from raw integer moves with mismatched types
|
||
- intel/brw: Run intel_nir_lower_conversions only after brw_nir_optimize
|
||
- intel/brw: Unconditionally run optimizations after nir_opt_uniform_subgroup
|
||
- intel/brw: Relax is_partial_write check in cmod propagation
|
||
- intel/brw: Enable constant propagation for a couple more logical sends
|
||
- intel/brw: load_offset cannot be constant on this path
|
||
- nir/algebraic: Optimize more comparisons with b2f
|
||
- nir/algebraic: Recognize open-coded bitfield_reverse in XCOM 2
|
||
- nir: Add a pass to generate BFI instructions from logical operations
|
||
- intel/brw: Use nir_opt_generate_bfi
|
||
- r600: Use nir_opt_generate_bfi to generate bitfield_select
|
||
- radeonsi: Use nir_opt_generate_bfi to generate bitfield_select
|
||
- radv: Use nir_opt_generate_bfi to generate bitfield_select
|
||
- brw/copy: Don't remove instructions w/ conditional modifier
|
||
- brw/builder: Add new style ALU3 builder
|
||
- brw/sat: Fix small typos, copy and paste, etc.
|
||
- brw/sat: Convert nearly all tests to use new style builders
|
||
- brw/algebraic: Refactor constant folding out of brw_fs_opt_algebraic
|
||
- brw/copy: Perform constant folding with constant propagation
|
||
- brw: Call brw_fs_opt_algebraic less often
|
||
- brw/emit: Add correct 3-source instruction assertions for each platform
|
||
- brw/copy: Don't copy propagate through smaller entry dest size
|
||
- brw/cse: Don't eliminate instructions that write flags
|
||
|
||
Icenowy Zheng (10):
|
||
|
||
- zink: reject Imagination proprietary driver w/o geometryShader
|
||
- gallivm: orcjit: keep the ownership of tm for LPJit
|
||
- gallivm: orcjit: use atexit to release LPJit singleton at exit
|
||
- util: detect LoongArch architecture
|
||
- gallivm: add LoongArch support to the mattrs setting code
|
||
- llvmpipe: add LoongArch support in ORCJIT
|
||
- zink: do not expose quad support when no geometry shader
|
||
- pvr: enlarge transfer fw_stream buffer for multicore
|
||
- pvr: add tpu_dm_global_registers feature
|
||
- pvr: emit tpu_tag_cdm_ctrl in compute stream when present
|
||
|
||
Igor Chernyshev (4):
|
||
|
||
- Add VK_EXT_debug_utils and VK_EXT_scalar_block_layout
|
||
- Enable VK_KHR_create_renderpass2 in Guest
|
||
- Move vkUpdateDescriptorSets to hand-written mode
|
||
- Do not copy invalid descriptor set image view handles
|
||
|
||
Iliyan Dinev (2):
|
||
|
||
- pvr: fix mipmap alignment for non-32bpp textures
|
||
- pvr: handle PVR_SRV_ERROR_RETRY pvr_srv_rgx_kick_render2()
|
||
|
||
Iván Briano (13):
|
||
|
||
- intel/rt: fix terminateOnFirstHit handling
|
||
- nir: add pass to convert ViewIndex to DeviceIndex
|
||
- anv: handle VK_PIPELINE_CREATE_VIEW_INDEX_FROM_DEVICE_INDEX_BIT
|
||
- anv: fix adding to wa_addr
|
||
- anv: be consistent about aux usage with modifiers
|
||
- anv: free shaders on rt pipeline compile error
|
||
- anv: skip rt pipeline compile if we found all shaders
|
||
- vulkan: use standard sample locations if there's no VkPipelineSampleLocationsStateCreateInfoEXT
|
||
- anv: allocate sparse descriptor buffers from the correct heap
|
||
- hasvk: fix non matching image/view format attachment resolve
|
||
- brw: fix task/mesh push constant loading
|
||
- intel/rt: fix ray_query stack address calculation
|
||
- anv: remove unused/misleading/wrong parameters from the RT trampoline
|
||
|
||
Jami Kettunen (1):
|
||
|
||
- nouveau/headers: Fix build without rustfmt
|
||
|
||
Jason Macnak (49):
|
||
|
||
- Rename imported system directory to guest
|
||
- Handle AHB and swapchain info in VkBindImageMemoryInfo
|
||
- Move platform into guest
|
||
- Move guest/shared/* to guest/*
|
||
- Add error logging to aid debugging
|
||
- Make VirtGpu* interfaces
|
||
- Wrap interactions with sync fds in an interface
|
||
- Update namespace in libandroidemu to gfxstream::guest
|
||
- Update iostream namespace to gfxstream::guest
|
||
- Enable building guest Gfxstream components on host
|
||
- Wrap AHardwareBuffer_* interactions into Gralloc interface
|
||
- Update VirtioGpuPipeStream to use VirtGpu abstraction
|
||
- Move SyncHelper to platform
|
||
- Revert "[guest] Check that the singleton VirtGpuDevice matches desired capset"
|
||
- Move info maps into ResourceTracker
|
||
- Promote testing framework to a proper platform backend
|
||
- Use libbase's property getting functions
|
||
- Add dispatch magic for Gfxstream linux ICD
|
||
- Wrap some classes in "namespace gfxstream::guest {}"
|
||
- Sort and dedup physical device extensions
|
||
- Fix "-Werror=conversion" errors for RanchuHwc
|
||
- Handle AHB R8 format conversions
|
||
- Make RutabagaLayer shared between "guest impls"
|
||
- Introduce Gfxstream Features to decouple Gfxstream from AEMU
|
||
- Ensure glProgramBinary initializes uniform and attrib info
|
||
- Add composition support to the end2end test framework
|
||
- Update emulated gralloc to use AHB format
|
||
- Rename VirtGpuBlob -> VirtGpuResource
|
||
- Explicitly specify target/bind/bpp in resource creation
|
||
- Partial revert of aosp/2858589 to avoid Mesa layer for Android
|
||
- Revert "Partial revert of aosp/2858589 to avoid Mesa layer for Android"
|
||
- Reland "Partial revert of aosp/2858589 to avoid Mesa layer for Android"
|
||
- Update vkMapMemory to not hold lock when calling into enc
|
||
- Remove extra semicolon in VkEncoder
|
||
- Support BGRA render target in CompositorVk
|
||
- Adds helper for tracking command buffers with device lost
|
||
- Adds AHB import and sampling test
|
||
- Add YUV AHB import and sample end2end tests
|
||
- Demote some logs to debug only
|
||
- Disable device groups
|
||
- Ensure mesa struct is translated to gfxstream struct on Android
|
||
- Handle DEPTH_STENCIL_ATTACHMENT_BIT in VK -> AHB usage
|
||
- Handle null handles in vkFreeCommandBuffers
|
||
- Fix vkAllocateCommandBuffers with failing allocation
|
||
- Revert "gfxstream: nuke mWorkPool"
|
||
- Handle external fences in vkGetFenceStatus()
|
||
- Add perfetto tracing to host renderer
|
||
- gfxstream: fix log levels in descriptor handling
|
||
- gfxstream: use gralloc metadata in vkGetAHBPropertiesANDROID
|
||
|
||
Jean-Francois Thibert (4):
|
||
|
||
- Add vkUpdateDescriptorSetWithTemplateKHR to resource tracker
|
||
- gfxstream: Fix usage of encoder after release in vkResetCommandBuffer
|
||
- gfxstream: move staging reset to end of reset command buffer
|
||
- Provided supported formats through gfxstream caps
|
||
|
||
Jean-François Thibert (1):
|
||
|
||
- Add depth formats support to import/export
|
||
|
||
Jesse (1):
|
||
|
||
- amd/vpelib: Config Writer hook and CDC refinement
|
||
|
||
Jesse Natalie (18):
|
||
|
||
- microsoft/clc: Split struct copies before vars_to_ssa in pre-inline optimizations
|
||
- mr-label-maker: Add d3d10umd label rules
|
||
- meson: Add an error message for llvmpipe without llvm draw support
|
||
- wgl: Add missing idep_mesautilformat
|
||
- d3d12: Don't use a vertex re-ordering GS for line primitives
|
||
- ci/windows: Use MSVC v143 build tools
|
||
- ci/windows: Update VK-GL-CTS to d48899f85b486a70d090af59a1453763458611d9
|
||
- ci/windows: Bump image tags to rebuild deps with v143 build tools
|
||
- compiler, vk: Support subgroup size of 4
|
||
- dzn: Drop custom subgroup size handling
|
||
- microsoft/compiler: Handle subgroup size 4 as a required wave size
|
||
- microsoft/compiler: Move nir_lower_undef_to_zero out of the optimization loop
|
||
- d3d12: Fix shader selector hash to hash array instead of pointer-to-array
|
||
- winsys/d3d12: Use the waitable object to decrease latency
|
||
- wgl: Add driconf options for controlling latency and swap interval
|
||
- winsys/d3d12: Implement the latency control option
|
||
- driconf: Disable dzn for bg3.exe
|
||
- dzn: Clean up dri options cache
|
||
|
||
Jessica Clarke (3):
|
||
|
||
- Revert "meson: Do not require libdrm for DRI2 on hurd"
|
||
- Revert "meson: fix with_dri2 definition for GNU Hurd"
|
||
- meson: egl: Build egl_dri2 driver even for plain DRI
|
||
|
||
Jiale Zhao (1):
|
||
|
||
- llvmpipe: add loongarch util_get_cpu_caps function
|
||
|
||
Jiali (1):
|
||
|
||
- amd/vpelib: Enhance output format capabilities
|
||
|
||
Jianxun Zhang (8):
|
||
|
||
- intel/common: Ensure SIMD16 for fast-clear kernel (xe2)
|
||
- intel/common: Remove blank lines in intel_set_ps_dispatch_state() (xe2)
|
||
- anv: Fix assertion failures on BMG (xe2)
|
||
- iris: Fix an assertion failure with compressed format
|
||
- anv: Disable compression on legacy modifiers (xe2)
|
||
- anv: Disable legacy CCS setup in binding (xe2)
|
||
- Revert "anv: Disable PAT-based compression on depth images (xe2)"
|
||
- Revert "iris: Disable PAT-based compression on depth surfaces (xe2)"
|
||
|
||
Job Noorman (88):
|
||
|
||
- ir3/postsched: improve debug state dumping
|
||
- ir3/postsched: fix calculation of max_delay
|
||
- ir3/postsched: remove unused delay field
|
||
- ir3/postsched: include ss/sy delay in max_delay
|
||
- ir3/postched: don't prioritize instructions with soft delays
|
||
- ir3: fix spill/reload split src/dst regs
|
||
- ir3: update merge set affinity in shared RA
|
||
- ir3: fix clearing merge sets after shared RA
|
||
- ir3: fix wrong dstn used in postsched
|
||
- ir3: fix counting of repeated registers
|
||
- ir3: correctly count vectorized instructions for tex prefetch
|
||
- ir3: add debug option to expand rpt instructions
|
||
- ir3: print (sat) modifier of instructions
|
||
- ir3: add backend support for repeated instructions
|
||
- ir3: add builders for repeated instructions
|
||
- ir3: make RA aware of repeat groups
|
||
- ir3: add pre-RA pass to clean up repeat groups
|
||
- ir3: add post-RA pass to merge repeat groups into rptN instructions
|
||
- ir3: lower vectorized NIR instructions
|
||
- nir/opt_vectorize: move rewriting of uses to a function
|
||
- nir/opt_vectorize: prepare for multiple try_combine functions
|
||
- nir/opt_vectorize: process blocks in source-code order
|
||
- nir/opt_vectorize: add support for phi nodes
|
||
- ir3: add support for vectorized NIR phi nodes
|
||
- ir3: replace \@load_uniform by new \@load_const_ir3 intrinsic
|
||
- ir3: rename \@store_uniform_ir3 to \@store_const_ir3
|
||
- nir/load_store_vectorize: support non-byte offset
|
||
- nir/load_store_vectorize: support stores without wrmask
|
||
- nir/load_store_vectorize: add load/store_const_ir3
|
||
- ir3: enable load/store_const_ir3 vectorization
|
||
- ir3: add support for rpt bary.f/flat.b
|
||
- ir3: use rpt instructions for frag coord
|
||
- ir3: use correct bit size for bools in emit_alu
|
||
- ir3/ci: remove fixed tests from a307-fails
|
||
- ir3: make fullsync sync after shared writes
|
||
- ir3/legalize: handle scalar ALU WAR hazards for a0.x
|
||
- zink/ci: add a618 flake
|
||
- freedreno: don't require binning and non-binning inputs to match
|
||
- ir3: remove unused outputs for binning pass in NIR
|
||
- ir3: don't modify const state for the binning variant in ir3_cp
|
||
- ir3: make ir3_const_state less error-prone to use
|
||
- ir3: fix calling ir3_const_state_mut in the binning VS
|
||
- ir3: print predicate inversion for branches
|
||
- ir3: print block divergence info
|
||
- ir3: remove unnecessary reconvergence point in scan_clusters lowering
|
||
- ir3: preserve block divergence info in ir3_lower_subgroups
|
||
- ir3: fix reconvergence points for blocks with two terminators
|
||
- ir3: fix adding physical edges multiple times
|
||
- ir3: fix reconvergence of blocks with multiple divergent predecessors
|
||
- ir3: recalculate reconvergence after opt_jump
|
||
- ir3: use physical cfg in helper_sched
|
||
- ir3: fix recognizing const/imm registers as a0
|
||
- ir3/legalize: don't add WAR dependencies for const/imm regs
|
||
- ir3: add assert to detect getting reg file of const/imm
|
||
- ir3/legalize: add needs_ss_war helper
|
||
- ir3/legalize: don't add (ss) for WAR hazards synced with (sy)
|
||
- ir3: add is_war_hazard_producer helper
|
||
- ir3/postsched: take WAR ss-delay into account
|
||
- ir3/legalize: resolve WAR hazards for stc
|
||
- ir3/isa: add isaspec definition for shfl
|
||
- ir3: add ir3_compiler::has_shfl for shfl support on a6xx+
|
||
- ir3/print: add support for shfl
|
||
- ir3: make backend aware of shfl:
|
||
- ir3: add codegen for rotate
|
||
- tu: advertise VK_KHR_shader_subgroup_rotate
|
||
- ir3: disallow immediates for shfl src1
|
||
- nir/load_store_vectorize: fix division by zero
|
||
- nir/lower_int64: add nir_intrinsic_read_invocation_cond_ir3
|
||
- nir/lower_int64: add nir_intrinsic_rotate
|
||
- ir3: lower 64b intrinsics after generic 64b lowering
|
||
- ir3: run 64b phi lowering together with other 64b lowerings
|
||
- ir3: lower 64b SSBO accesses
|
||
- ir3: make ir3_mem_access_size_align publically available
|
||
- tu: lower 64b push constants before lowering IO
|
||
- ir3: add support for 64b reductions
|
||
- tu: advertise shaderInt64
|
||
- nir: add shuffle_{xor,up,down}_uniform_ir3 intrinsics
|
||
- ir3: add codegen for shuffle_{xor,up,down}_uniform_ir3
|
||
- ir3: add helper to get glsl type for nir_def
|
||
- ir3: optimize subgroup shuffles using shfl
|
||
- nir/lower_subgroups: move up some helper functions
|
||
- nir/lower_subgroups: add build_cluster_mask helper
|
||
- nir/lower_subgroups: scan/reduce for multiple ballot components
|
||
- ir3: fix physical edges of predicated branches
|
||
- freedreno,computerator: add support for local memory
|
||
- ir3: merge is_reg_gpr and reg_gpr
|
||
- ir3/ra: prevent moving source intervals for shared collects
|
||
- ir3,tu: include ir3 debug flags in shader hash key
|
||
|
||
Jocelyn Falempe (3):
|
||
|
||
- loader: Fix typo in __DRI_IMAGE_FORMAT_XBGR16161616 definition
|
||
- gbm/dri: Use PIPE_FORMAT_* instead of using __DRI_IMAGE_*
|
||
- gbm/dri: Fix color format for big endian.
|
||
|
||
John Anthony (4):
|
||
|
||
- pan/kmod: Add max_tasks_per_core to kmod props
|
||
- panvk: Add cmd_dispatch
|
||
- panvk: Pull out task axis and increment calculation for dispatch
|
||
- panvk: Add support for CmdDispatchIndirect
|
||
|
||
Jonathan Marek (2):
|
||
|
||
- freedreno: improve a6xx CP_SET_MARKER xml definition
|
||
- freedreno/a6xx: add missing USES_GMEM flag and BIN_RENDER_END markers
|
||
|
||
Jordan Justen (31):
|
||
|
||
- include/drm-uapi/README: Update README for drm uapi header files
|
||
- intel/brw/validate: Simplify grf span validation check by not using a mask
|
||
- intel/brw/validate: Update dst grf crossing check for Xe2
|
||
- intel/brw/validate: Convert access mask to be grf based
|
||
- intel/dev: Update hwconfig => max_threads_per_psd for Xe2
|
||
- intel/dev: Enable BMG PCI IDs (without INTEL_FORCE_PROBE)
|
||
- anv: Drop "not yet supported" warning for Xe2
|
||
- anv: Do hasvk devices check first
|
||
- intel/dev: Add devinfo::probe_forced based on INTEL_FORCE_PROBE
|
||
- anv: Don't warn about unsupported devices if INTEL_FORCE_PROBE was used
|
||
- intel/dev: Fix warning for max_threads_per_psd when devinfo->verx10 == 120
|
||
- intel/dev: Add 0xb640 ARL PCI id
|
||
- intel/dev: Rework DEVINFO_HWCONFIG; add DEVINFO_HWCONFIG_KV macro
|
||
- intel/dev: Simplify DEVINFO_HWCONFIG_KV by adding should_apply_hwconfig_item()
|
||
- intel/dev: Allow specifying a version when to always use hwconfig
|
||
- intel/dev: Use hwconfig for urb min/max entry values
|
||
- intel/dev: Support Xe3 device init (for intel_device_info_test)
|
||
- intel/dev: Add XE3_FEATURES macro
|
||
- intel/genxml: Start Xe3 support
|
||
- intel/isl: Build for Xe3
|
||
- intel/shaders: Build for Xe3
|
||
- iris: Build for Xe3
|
||
- intel/dev: Add Xe3 support to get_l3_list()
|
||
- intel/compiler: Add compiler enum for Xe3
|
||
- intel/dev: Add INTEL_PLATFORM_PTL platform enum
|
||
- intel/dev: Add PTL device info
|
||
- intel/dev: Add PTL PCI IDs (with FORCE_PROBE set)
|
||
- intel/brw: Allow Xe3 in brw_stage_has_packed_dispatch()
|
||
- intel/compiler: Xe2 and Xe3 use the same compaction tables
|
||
- anv: Build for Xe3
|
||
- intel/dev: Set L3 bank count for Xe2+ from Xe KMD
|
||
|
||
Jose Maria Casanova Crespo (5):
|
||
|
||
- v3d: v3d_resource Use LINEAR layout for importing with INVALID modifier
|
||
- v3d: avoid load/store of tile buffer on invalidated framebuffer
|
||
- v3d: initialize job local key with the 8 color buffer available in v7.1+
|
||
- v3dv/ci: Add missing fails on RPi4/5 for uprev VKCTS to 1.3.10.0
|
||
- v3d: Enable Early-Z with discards when depth updates are disabled
|
||
|
||
Joshua Ashton (2):
|
||
|
||
- radv: Enable variableSampleLocations
|
||
- radv: Fix sample locations at 0 for X/Y
|
||
|
||
Joshua Duong (1):
|
||
|
||
- For goldfish pipe, compute colorBufferMemoryIndex the same as host.
|
||
|
||
José Roberto de Souza (50):
|
||
|
||
- intel/brw: Add a maximum scratch size restriction
|
||
- anv: Drop useless 'if (total_scratch > 0) {' block in cmd_buffer_ensure_cfe_state()
|
||
- anv: Wait for Xe exec queue to be idle before destroying it
|
||
- isl: Fix Xe2 protected mask
|
||
- anv: Propagate protected information to blorp_batch_isl_copy_usage()
|
||
- anv: Improve error message when pipeline creation fails during shader compilation
|
||
- anv: Handle internal shader compilation failure
|
||
- intel: Sync xe_drm.h
|
||
- intel/dev: Support new topology type with SIMD16 EUs
|
||
- docs: Move all ANV environment variables to a single section in envvars
|
||
- anv: Disable sparse even on Xe KMD with ANV_SPARSE
|
||
- docs/anv: Document ANV_SPARSE and ANV_SPARSE_USE_TRTT
|
||
- anv: Extend ANV_QUEUE_OVERRIDE to blit count
|
||
- intel: Remove INTEL_ENGINE_CLASS_COMPUTE and INTEL_ENGINE_CLASS_COPY parameters
|
||
- docs/anv: Remove documentation about INTEL_COMPUTE_CLASS and INTEL_COPY_CLASS
|
||
- anv: Add trivial_batch and query-pool to the error capture
|
||
- anv: Use batch_bo_pool in utrace anv_async_submit_init() calls
|
||
- anv: Nuke anv_utrace_submit::trace_bo
|
||
- intel/isl/gfx20: Alow hierarchial depth buffer write through for multi sampled surfaces
|
||
- anv/gfx20: Enable depth buffer write through for multi sampled images
|
||
- iris/gfx20: Enable depth buffer write through for multi sampled images
|
||
- anv: Nuke perf_metric
|
||
- intel/perf: Fix intel_gem.h include
|
||
- anv: Drop useless '>= 0' check over a unsigned
|
||
- anv: Add anv_device_perf_close()
|
||
- anv: Fix context id or exec queue used to open perf stream
|
||
- anv: Add warning about mismatch between query queues
|
||
- anv: Make sure all previous vm binds are done before execute perf query pool
|
||
- anv: Check if vkCreateQueryPool() is being created in a supported queue
|
||
- anv: Add documentation to some fields in anv_query_pool
|
||
- anv/query: Fix batch end value
|
||
- anv: Improve readbility of khr_perf_query_availability_offset() and khr_perf_query_data_offset()
|
||
- anv: Fix condition to clear query pool with blorp
|
||
- intel: Split anv_xe_wait_exec_queue_idle() and move part of it to common/
|
||
- iris: Use xe_queue_get_syncobj_for_idle()
|
||
- anv: Optimize vkQueueWaitIdle() on Xe KMD
|
||
- intel/perf: Sync oa-lnl.xml
|
||
- intel/perf: Add BMG OA support
|
||
- intel/genxml: Append 'Z Async Throttle settings' to gfx125 definition of STATE_COMPUTE_MODE
|
||
- anv: Set all async compute registers in STATE_COMPUTE_MODE
|
||
- intel/genxml: Do small fixes in gfx20 definition of STATE_COMPUTE_MODE
|
||
- anv: Set recommended values for gfx20 async compute registers in STATE_COMPUTE_MODE
|
||
- iris: Program async compute registers of STATE_COMPUTE_MODE in compute engine
|
||
- intel/dev/xe: Set max_eus_per_subslice using topology query
|
||
- intel/dev: Fix max_cs_threads value on simulator
|
||
- intel/perf: Add OA support to ARL
|
||
- drm-uapi: Sync xe_drm.h
|
||
- intel/perf: Extend intel_perf_stream_set_metrics_id() to syncronize metrics id changes
|
||
- intel/perf: Add INTEL_PERF_FEATURE_METRIC_SYNC and check if KMD supports it
|
||
- anv: Enable perf metrics id set syncronization
|
||
|
||
Juan A. Suarez Romero (32):
|
||
|
||
- v3d: use operations to specify what to save in blitter
|
||
- v3d: add new clear blitter op
|
||
- v3d/ci: add failures from piglit all profile
|
||
- v3d/ci: add new flakes
|
||
- vc4: fix typo in simulator conditional declaration
|
||
- v3d/vc4/ci: add address sanitizer jobs
|
||
- v3d/ci: enable full run jobs for rpi5
|
||
- broadcom/ci: annotate some of the failures
|
||
- v3d/ci: add new flakes in rpi5
|
||
- v3d/ci: update number of rpi5 available for testing
|
||
- v3d: free dbuf
|
||
- broadcom/ci: use kernel8 for rpi5 full jobs
|
||
- v3d/ci: update expected results
|
||
- vc4/ci: update expected results
|
||
- Revert "ci: take igalia farm offline"
|
||
- vc4/ci: update expected results
|
||
- vc4: build simulator references conditionally
|
||
- v3d/v3dv: build simulator references conditionally
|
||
- v3d/ci: add new flake for rpi5
|
||
- Revert "v3d: never replace a mapped bo"
|
||
- v3d: do not rebind a sampler view already rebound
|
||
- v3d/ci: add new flake for rpi5
|
||
- v3d/ci: update expected results
|
||
- v3dv/ci: add new flake
|
||
- broadcom: do not include simulator if not available
|
||
- v3d/ci: add new flakes and timeouts
|
||
- vc4: handle nir_op_ult32 intrinsic
|
||
- v3d/v3dv/ci: update expected results
|
||
- v3d/ci: use special kernel for full jobs
|
||
- v3dv/ci: move asan fail to skip
|
||
- v3d: add new flake
|
||
- v3d/ci: add OpenCL failures
|
||
|
||
Jude Shih (2):
|
||
|
||
- amd/vpelib: Dynamic Command List Expansion
|
||
- amd/vpelib: Restructure CDC FE/BE
|
||
|
||
Jules Blok (5):
|
||
|
||
- vulkan: Add support for VK_EXT_depth_clamp_control
|
||
- radv: Add support for VK_EXT_depth_clamp_control
|
||
- anv: Add support for VK_EXT_depth_clamp_control
|
||
- nvk: Add support for VK_EXT_depth_clamp_control
|
||
- docs: Update status of VK_EXT_depth_clamp_control support
|
||
|
||
Julia Zhang (1):
|
||
|
||
- mesa/st: use drawable->ID as hash for drawable_ht
|
||
|
||
Juston Li (5):
|
||
|
||
- anv/android: remove unneeded ANB implicit import flags
|
||
- anv/android: handle R8G8B8X8 as R8G8B8A8
|
||
- anv/android: refactor out u_gralloc tiling query
|
||
- anv/android: resolve ANB swapchain images on bind
|
||
- android: look for debug/vendor prefixed options
|
||
|
||
K900 (5):
|
||
|
||
- gbm: split the DRI backend into a separate library and unify backend handling
|
||
- meson: require libdrm when building gbm
|
||
- meson/gbm: clean up dependencies
|
||
- meson: remove dri-search-path
|
||
- meson/gbm: fix meson devenv for GBM backends
|
||
|
||
Kai Wasserbäch (2):
|
||
|
||
- fix(FTBFS): clover: use \`.getDataLayout()` with LLVM >= 20
|
||
- fix(FTBFS): build: link OpenCL target with \`dep_clang` from top-level
|
||
|
||
Karmjit Mahil (12):
|
||
|
||
- freedreno: Enable the A735
|
||
- tu: Set \`TU_ACCESS_CCHE_READ` for transfer ops with read access
|
||
- freedreno: Define SP_DITHER_CNTL (0xA9AC)
|
||
- tu: Implement VK_EXT_legacy_dithering
|
||
- ir3: Use \`foreach_instr_safe` in ir3_shared_folding
|
||
- pvr: Fix \`barier`->`barrier` typo
|
||
- pvr: Use a pixel size of \`0` for invalid pbe accum formats
|
||
- pvr: Regenerate all descriptor program data sections
|
||
- pvr: Handle DISABLE_PIXELMERGE
|
||
- tu: Fix push_set host memory leak on command buffer reset
|
||
- tu: Fix potential alloc of 0 size
|
||
- nir: Fix \`no_lower_set` leak on early return
|
||
|
||
Karol Herbst (101):
|
||
|
||
- spirv: generate info for FunctionParameterAttribute
|
||
- spirv: initial parsing of function parameter decorations
|
||
- spirv: handle function parameters passed by value
|
||
- nak: allow clippy::not_unsafe_ptr_arg_deref lints
|
||
- nak: fix clippy::if_same_then_else error
|
||
- rusticl/platform: add perf debug option
|
||
- rusticl/memory: remove stale TODOs
|
||
- rusticl/memory: add a couple of performance warnings
|
||
- rusticl: move more kernel related things into kernel.rs
|
||
- rusticl/kernel: make some types and fields private
|
||
- rusticl/kernel: move setting of nir metadata into lower_and_optimize_nir
|
||
- rusticl/program: move attribute_str to the spirv module
|
||
- rusticl: use blob for all serialization
|
||
- rusticl: remove unsued serialization helpers
|
||
- rusticl/kernel: add type for the convert_spirv_to_nir return value
|
||
- rusticl/kernel: move internal_args into NirKernelBuild
|
||
- clc: force linking of spirvs with mismatching pointer types in signatures
|
||
- rusticl/spirv: do not add additional new lines when joining logs
|
||
- rusticl: fix clippy lint having bounds defined in multiple places
|
||
- rusticl/program: protect against 0 length in slice::from_raw_parts
|
||
- rusticl/api: protect against 0 length in slice::from_raw_parts
|
||
- rusticl/spirv: protect against 0 length in slice::from_raw_parts
|
||
- nouveau: handle realloc failure inside cli_kref_set
|
||
- rusticl: bump rust req to 1.76
|
||
- rusticl: use div_ceil
|
||
- rusticl: use next_multiple_of
|
||
- rusticl: use pointer byte offset API
|
||
- rusticl: replace some raw pointer casts
|
||
- rusticl: use ptr::from_ref and ptr::from_mut
|
||
- rusticl/icd: use ptr::addr_eq
|
||
- meson: centralize rust handling
|
||
- docs: move rust update policy
|
||
- meson: centralize bindgen version check
|
||
- meson: centralize checking for new enough meson for rust support
|
||
- nir: add load_global_size intrinsic
|
||
- rusticl/kernel: emit system values with the correct device size
|
||
- rusticl/kernel: handle load_global_size
|
||
- mesa: check for enabled extensions for \*UID enums
|
||
- nouveau/winsys: fix handling of NV_DEVICE_TYPE_IGP
|
||
- nouveau: use nv_devince_info and fill in PCI and type information
|
||
- nouveau: add nv_device_uuid
|
||
- nouveau: implement driver_uuid and device_uuid
|
||
- nvk: use nv_device_uuid
|
||
- nouveau: remove nouveau_check_for_uma
|
||
- nouveau: implement PIPE_PCI caps
|
||
- nouveau: use nv_device_info directly for dumping push buffers
|
||
- zink: lower 64 bit find_lsb, ufind_msb and bit_count
|
||
- zink: lower 8/16 bit alu ops vk spirv doesn't allow
|
||
- rusticl/kernel: properly respect device thread limits per dimension
|
||
- rusticl/device: Make supported 1Dbuffer formats a strict subset of 1D
|
||
- rusticl/memory: Fix memory unmaps after rework
|
||
- rusticl/image: take pitches into account when allocating memory for maps
|
||
- rusticl/image: properly sync mappings content for 1Dbuffer images
|
||
- rusticl/queue: add clSetCommandQueueProperty
|
||
- util/u_printf: do not double print format string with unused arugments
|
||
- rusticl/memory: fix sampler argument size check
|
||
- rusticl/kernel: move kernel size into Constant arg type
|
||
- rusticl/kernel: rename InternalKernelArg to CompiledKernelArg
|
||
- rusticl/kernel: remove CompiledKernelArg::size
|
||
- rusticl/kernel: make CompiledKernelArg::offset a u32
|
||
- rusticl/kernel: make KernelArg::offset and KernelArg::binding proper u32 values
|
||
- rusticl/kernel: rename InternalKernelArgType to CompiledKernelArgType
|
||
- rusticl/kernel: move serialization code of KernelArgType
|
||
- rusticl/kernel: big kernel arg rework
|
||
- rusticl/kernel: add KernelArgType::is_opaque
|
||
- rusticl/kernel: move CompiledKernelArg decleration above the impl block
|
||
- rusticl/kernel: simplify adding internal kernel args
|
||
- rusticl/mem: do not check against image base alignment for 1Dbuffer images
|
||
- rusticl: do not use CL vector types in bindings and code
|
||
- ac/llvm: fix umul_high
|
||
- rusticl/device: limit CL_DEVICE_IMAGE_MAX_BUFFER_SIZE more aggressively
|
||
- vtn: ignore volatile on functions for now
|
||
- clc: add support for more image related extensions
|
||
- compiler/types: Add multisample vimage/vtexture types
|
||
- nir: Support multisampled images in lower_read_only_images_to_tex()
|
||
- vtn, nir: handle OpImageQueryLevels on images
|
||
- vtn: mark ImageMipmap as supported
|
||
- rusticl: reuse PipeContext
|
||
- zink: cache generated vendor and device name inside zink_screen
|
||
- rusticl/mesa: Return CStr for device and vendor names.
|
||
- clc: fix compilation error with llvm-20
|
||
- rusticl/kernel: move dead_var options into constant space
|
||
- rusticl/platform: add env variable to disable kernel variants
|
||
- rusticl/mesa: add more workgroup_Size functions to NirShader
|
||
- rusticl/mesa: implement Clone for NirShader
|
||
- rusticl/kernel: move assign_locations into CompiledKernelArg
|
||
- rusticl/kernel: split up nir compilation into various steps
|
||
- rusticl/kernel: track if arg is dead inside CompiledKernelArg
|
||
- rusticl/kernel: add CompilationResult to hold compilation artifacts
|
||
- rusticl/kernel: add optimized Kernel variant
|
||
- rusticl: add nir debugging option
|
||
- rusticl: allow devices to be enabled by default
|
||
- docs: document gallium-rusticl-enable-drivers
|
||
- asahi: implement get_cl_cts_version
|
||
- rusticl: allow asahi to be enabled by default
|
||
- zink: add CL CTS result
|
||
- radeonsi: move si_compute::global_buffers to si_context
|
||
- nv/codegen: Do not use a zero immediate for tex instructions
|
||
- nvc0: return NULL instead of asserting in nvc0_resource_from_user_memory
|
||
- rusticl/kernel: fix kernel variant selection
|
||
- vtn: handle struct kernel arguments passed by value
|
||
|
||
Kenneth Graunke (48):
|
||
|
||
- intel/brw: Don't force g1's live range to be the entire program
|
||
- intel/brw: Delete fs_reg_alloc::discard_interference_graph()
|
||
- intel/brw: Record that SHADER_OPCODE_SCRATCH_HEADER uses g0
|
||
- intel/brw: Record g0 as live for sends with send_ex_desc_scratch set
|
||
- intel/brw: Only force g0's liveness to be the whole program if spilling
|
||
- intel/brw: Mark all UBO access with a direct buffer index as speculative
|
||
- intel/brw: Delete the brw_fs_opt_peephole_select() pass
|
||
- intel/brw: Delete the brw_fs_opt_dead_control_flow_eliminate() pass
|
||
- intel/brw: Replace predicated break optimization with a simple peephole
|
||
- intel/brw: Set appropriate types for 16-bit sampler trailing components
|
||
- intel/brw: Drop image_{load,store}_raw_intel handling
|
||
- intel/brw: Pass opcode to brw_swsb_encode/decode
|
||
- intel/brw: Fix Xe2+ SWSB encoding/decoding for DPAS instructions
|
||
- intel: Fix bad align_offset on global_constant_uniform_block_intel
|
||
- intel/brw: Fix OOB reads when printing instructions post-reg-alloc
|
||
- intel/brw: Print blocks in brw_print_instructions_to_file()
|
||
- intel/brw: Add a file parameter to idom_tree::dump()
|
||
- intel/brw: Switch from LSC CMASK opcodes to regular LOAD/STORE
|
||
- intel/brw: Get rid of the lsc_msg_desc_wcmask helper
|
||
- intel/brw: Drop misguided sign extension attempts in extract_imm()
|
||
- intel/brw: Fix extract_imm for subregion reads of 64-bit immediates
|
||
- intel/brw: Use NUM_BRW_OPCODES in can_omit_write() check
|
||
- intel/brw: Use size_written for NoMask instructions in is_partial_write
|
||
- intel/brw: Rename lsc_aop_for_nir_intrinsic to "op" instead of "aop"
|
||
- intel/brw: Handle load/stores in lsc_op_for_nir_intrinsic()
|
||
- intel/brw: Introduce new MEMORY_*_LOGICAL opcodes
|
||
- intel/brw: Expose functions to convert LSC enums to strings
|
||
- intel/brw: Pretty-print memory logical opcodes
|
||
- intel/brw: Lower MEMORY_OPCODE_*_LOGICAL to LSC messages
|
||
- intel/brw: Lower MEMORY_OPCODE_*_LOGICAL to HDC messages
|
||
- intel/brw: Switch load_num_workgroups to the new memory intrinsic
|
||
- intel/brw: Switch to emitting MEMORY_*_LOGICAL opcodes
|
||
- intel/brw: Switch load_ubo_uniform_block_intel over to memory intrinsics
|
||
- intel/brw: Delete old-style surface and A64 message opcodes
|
||
- intel/brw: Don't include sync.nop in INTEL_DEBUG instruction counts
|
||
- nir: Don't generate single iteration loops to zero-initialize memory
|
||
- intel/brw: Fix spill/fill count for load/store_scratch in SIMD32
|
||
- intel/brw: Delete Gfx7-8 code from emit_barrier()
|
||
- intel/brw: Make a ubld temporary in emit_barrier()
|
||
- intel/brw: Fix register and builder size in emit_barrier() for Xe2
|
||
- intel/brw: Delete more Gfx8 code from brw_fs_combine_constants
|
||
- intel/brw: Use whole 512-bit registers in constant combining on Xe2
|
||
- brw: Optimize 16-bit texture fetches later
|
||
- intel/brw: Allow immediates in the BFE instruction on Gfx12+
|
||
- brw/validate: Return an error for Align16 access mode on Icelake+
|
||
- brw/emit: Fix align16 3src subregister encodings for HF types
|
||
- intel: Set shader_spilling_rate=11 in intel_clc
|
||
- brw: Fix try_rebuild_source's ult32/ushr handling to use unsigned types
|
||
|
||
Konrad Dybcio (1):
|
||
|
||
- freedreno: Add initial A621 support
|
||
|
||
Konstantin (5):
|
||
|
||
- util: Handle more formats in get_plane_(width|height)
|
||
- vtn: Remove dead shader_call_data from all RT stages
|
||
- radv: Handle instruction encodings > 8 bytes when splitting disassembly
|
||
- radv: Handle repeated instructions when splitting disassembly
|
||
- meson: Allow building lavapipe without specifying llvmpipe
|
||
|
||
Konstantin Seurer (28):
|
||
|
||
- llvmpipe: Improve PIPE_QUERY_PIPELINE_STATISTICS accuracy
|
||
- lavapipe: Fix box sorting during ray traversal
|
||
- aco: print s_delay_alu INSTSKIP>3 correctly
|
||
- gallivm: Use variables for loop exec mask handling
|
||
- gallivm: Skip inactive branches
|
||
- radv: Synchronize shader dumping
|
||
- llvmpipe: Use derivative intrinsics
|
||
- nir: Introduce nir_debug_info_instr
|
||
- spirv: Emit nir_debug_info_instr for OpLine
|
||
- nir/print: Add a helper for generating debug info
|
||
- nir/opt_loop: Fix handling else-breaks in merge_terminators
|
||
- gallium,st/mesa: Add and set pipe_image_view::is_2d_view_of_3d
|
||
- lavapipe: Implement VK_EXT_image_2d_view_of_3d with sparse textures
|
||
- lavapipe: Do not adjust imageGranularity for different block sizes
|
||
- radv: Work around broken terrain in Warhammer III
|
||
- radv: Initialize sqtt state before meta state
|
||
- radv: Workaround apps using ray tracing when it is unsupported
|
||
- lavapipe: Fix report_ray_intersection affecting terminated rays
|
||
- lavapipe: Do not return in report_ray_intersection
|
||
- radv: Fix report_ray_intersection affecting terminated rays
|
||
- lavapipe: Implement clustered reductions
|
||
- lavapipe: Implement VK_KHR_shader_subgroup_rotate
|
||
- Revert "lavapipe: Do not return in report_ray_intersection"
|
||
- lavapipe: Set minAccelerationStructureScratchOffsetAlignment to 8
|
||
- radv/meta: Do not pass NULL to vk_texcompress_astc_finish
|
||
- nir/print: Fix the alignment of 8-bit definitions
|
||
- radv: Remap 10 and 12 bit formats to 16 bit formats
|
||
- radv: Move ac_addrlib to the physical device
|
||
|
||
Kovac, Krunoslav (4):
|
||
|
||
- amd/vpelib: DPP starting changes
|
||
- amd/vpelib: MPC refactoring HW registers
|
||
- amd/vpelib: Refactor MPC registers
|
||
- amd/vpelib: Reuse existing float to reg format conversion
|
||
|
||
Lars Harrison (3):
|
||
|
||
- Provide error message on unhandled VK structs
|
||
- Always initialize gfxstream vulkan
|
||
- Fix coherent memory allocation to use device
|
||
|
||
Lars-Ivar Hesselberg Simonsen (12):
|
||
|
||
- panfrost: Move ForEachMacros into panfrost
|
||
- pan/genxml: Add pan_pack_nodefaults()
|
||
- panfrost: Fix near/far depth clip
|
||
- panfrost: Enable support for depth clamping
|
||
- panvk: Fix paths that assume a fragment shader
|
||
- panvk: Fix use of cs_sync64_wait in cmd_wait_event
|
||
- panvk: Map device memory for PANVK_DEBUG=trace
|
||
- panvk: Use the passed Vertex Attribute divisor
|
||
- panvk: Only set index buffer size for DrawIndexed
|
||
- panvk: Add get_tiler_idvs_flags
|
||
- panvk: Split panvk_cmd_draw into prepare_draw
|
||
- panvk: Add support for Draw[Indexed]Indirect
|
||
|
||
Leder, Brendan Steve (4):
|
||
|
||
- amd/vpelib: Fix output_ctx gamma curve cache bug
|
||
- amd/vpelib: Fix backend bug for multiple instances
|
||
- amd/vpelib: Generalize visual confirm handling
|
||
- amd/vpelib: Update chip headers
|
||
|
||
Lepton Wu (1):
|
||
|
||
- egl/android: Fix wrong pipe format for RGB_565
|
||
|
||
Lin, Ricky (4):
|
||
|
||
- amd/vpelib: Added JFIF format to RGB output side
|
||
- amd/vpelib: Use VPE_IP_LEVEL_1_0 for VPE IP 6.1.3
|
||
- amd/vpelib: Remove extra collaborate sync commands in IB
|
||
- amd/vpelib: Increase the CD field in vpe descriptor programming
|
||
|
||
LingMan (1):
|
||
|
||
- nak/hw_tests: Use f32::to_bits()
|
||
|
||
Lionel Landwerlin (139):
|
||
|
||
- anv: split instance from anv_device.c
|
||
- anv: split physical_device from anv_device.c
|
||
- anv: split events from anv_device.c
|
||
- anv: split buffer from anv_device.c
|
||
- anv: split sampler from anv_device.c
|
||
- anv: split buffer view from anv_image.c
|
||
- anv: split image view from anv_image.c
|
||
- isl: account for protection in base usage checks
|
||
- anv: properly flag image/imageviews for ISL protection
|
||
- anv: propagate protected information for blorp operations
|
||
- anv: stop using 3DSTATE_WM::ForceThreadDispatchEnable
|
||
- anv: fix check on pipeline mode to track buffer writes
|
||
- u_trace: add mako exception printouts
|
||
- u_trace: rework tracepoint argument declaration
|
||
- u_trace: remove timestamp reference in allocations
|
||
- u_trace: add support for indirect data
|
||
- anv/hasvk: add indirect tracepoint arguments
|
||
- u_trace: add csv output
|
||
- util/debug: update parse_enable_string to deal with +all/-all
|
||
- anv: better signal new frames to utrace
|
||
- vulkan/runtime: allow null/empty debug names
|
||
- anv: reuse object string for RMV token
|
||
- anv: prevent asserts with debug printf in internal shaders
|
||
- anv: remove some unused includes
|
||
- anv: reuse cs_prog_data pointer
|
||
- intel-clc: missing printf lowering
|
||
- anv: add missing MEDIA_STATE_FLUSH for internal shaders
|
||
- anv: ensure max_plane_count is at least 1
|
||
- genxml: unify some bits between Gfx8/Gfx11/Gfx12.5
|
||
- intel/mi_builder: enable control flow API on Gfx9+
|
||
- anv/blorp: force CC_VIEWPORT reallocation when programming 3DSTATE_VIEWPORT_STATE_POINTERS_CC
|
||
- anv: limit some state dirtying after blorp/simpler-shaders
|
||
- brw/rt: fix ray_object_(direction|origin) for closest-hit shaders
|
||
- intel/nir: remove load_global_const_block_intel intrinsic
|
||
- vulkan/runtime: fix GetBufferMemoryRequirements2 for maintenance4
|
||
- anv: fix extended buffer flags usages
|
||
- anv: remove unused macro
|
||
- anv: only set 3DSTATE_CLIP::MaximumVPIndex once
|
||
- anv: add assert to detect problematic instruction merges
|
||
- anv: reuse common pipeline state for compute push allocations
|
||
- anv: simplify loading driver internal constants
|
||
- anv: move lowering of descriptor intrinsics to apply_layout
|
||
- nir: remove unused intel intrinsics
|
||
- brw: remove unused prog_data field
|
||
- anv: optimize CLIP::MaximumVPIndex setting
|
||
- anv: move conditional render predicate after gfx_flush_state
|
||
- anv: don't miss workaround for indirect draws
|
||
- anv: optimize STATE_BYTE_STRIDE emission
|
||
- anv: explicitly disable BT pool allocations at device init
|
||
- anv: always use workaround_address, not workaround_bo
|
||
- anv: order data in wa_bo to leave wa_addr last
|
||
- nir/divergence: add missing load_constant_base_ptr
|
||
- brw: switch mesh/task URB fence prior to EOT to GPU
|
||
- brw: add a comment what Gfx12.5 URB fences
|
||
- anv: fix utrace compute timestamp reads on Gfx20
|
||
- iris: fix utrace compute end timestamp reads on Gfx20
|
||
- brw: align spilling offsets to physical register sizes
|
||
- anv: selectively disable binding table usage on Gfx20
|
||
- brw: remove (load|store)_raw_intel
|
||
- brw: fix TGM messages to use cmask lsc opcodes
|
||
- clc: find opencl headers from the installed llvm/clang location
|
||
- brw: use a builder of the size of the physical register for uniforms
|
||
- brw: fix vecN rebuilds
|
||
- brw: fix virtual register splitting to not go below physical register size
|
||
- anv: fix missing tracking for alpha-to-coverage runtime changes
|
||
- anv: Only flush render target cache when detecting RT changes
|
||
- iris: ensure null render target for specific cases
|
||
- brw: move null_rt control up a layer
|
||
- brw: disable null_rt only if color output does not affect other outputs
|
||
- anv: add missing pipeline instance multiplier
|
||
- brw: make sampler message emission more generic
|
||
- brw: fix mask componentation for 16-bit sampler returns
|
||
- brw: remove EOT handling from sampler messages
|
||
- intel/decoder: constify functions not modifying instructions/fields
|
||
- intel/decoder: add filter feature
|
||
- intel/decoder: split state tracking handlers from printing ones
|
||
- intel/decoder: decode the 8 BLEND_STATEs
|
||
- anv: avoid setting up a null RT unless needed
|
||
- anv: implement VK_KHR_dynamic_rendering_local_read
|
||
- anv: expose VK_EXT_pipeline_protected_access
|
||
- zink: avoid host transfer usage with sparse
|
||
- blorp: convert fast clear color for unsupported formats
|
||
- anv: limit render target cache flushing due to color output remapping
|
||
- isl: fix comment typo
|
||
- isl/tests: add a space in memcpy test error printouts
|
||
- isl/tests: fixup multi-tile testing
|
||
- isl/tests: fix coordinates for comparison scan
|
||
- isl/tests: rename ytile to be consistent with other names
|
||
- isl/tests: bump coordinate types to 32bits
|
||
- isl/tests: use modulo for more variability per pixel
|
||
- isl/tests: fix inclusive coordinate checking
|
||
- isl/tests: add TileX memcpy testing
|
||
- isl/tests: rename tile_(width|height) variables
|
||
- isl/tests: split linear & tiled buffer sizes
|
||
- anv: limit 22018402687 to impacted platforms
|
||
- anv: consolidate pre/post draw workaround in helpers
|
||
- anv: optimize WA 16011107343/22018402687
|
||
- isl: remove duplicated copy for tileX/TileY
|
||
- isl/tests: add more coordinates for full tiles testing
|
||
- isl/tests: rename span variable to xt_sub_range_alignment
|
||
- isl: Tile W memcpy support
|
||
- iris: delete stencil mapping support
|
||
- isl: add support of aux disable bit on hiz
|
||
- anv: use stage mask to deduce cs/pb-stall requirements
|
||
- brw: remove unused prototype
|
||
- brw: delay printf lowering
|
||
- brw: avoid clashing nested loop indices
|
||
- brw: remove rebuild single element special case
|
||
- brw: fix mesh fence emission
|
||
- brw/nir: rework inline_data_intel to work with compute
|
||
- anv: stop using a binding table entry for gl_NumWorkgroups
|
||
- elk: Don't apply discard_if condition opt if it can change results
|
||
- isl: fix range_B_tile end_tile_B value
|
||
- isl/tests: add range_B_tile test
|
||
- anv: allow subresource queries on non-linear images
|
||
- anv: drop non host memory types for host-transfer on non-rebar
|
||
- anv: disable pat compression for host images
|
||
- anv: wrap binding address setting
|
||
- anv: add a host map of image for host image copy usage
|
||
- anv: factor out sanitizing mmap offset code
|
||
- anv: add VK_EXT_host_image_copy support
|
||
- anv: fix binding table entry count for compute shaders
|
||
- anv: fix missing inline parameter emission
|
||
- anv: reemit push constants on pipeline changes
|
||
- anv: use UINT32_MAX to be consistent
|
||
- anv: avoid companion usage on RCS
|
||
- anv: avoid L3 fabric flush in pipeline barriers
|
||
- anv: update some of the indirect invalidations
|
||
- vulkan/runtime: fix allocation failure handling
|
||
- anv: fix even set/reset on blitter engine
|
||
- anv: add texture cache inval after binding pool update
|
||
- anv: fix indentation
|
||
- anv: move pipe control debug to anv_util.c
|
||
- anv: fix extent computation in image->image host copies
|
||
- anv: update shader descriptor resource limits
|
||
- brw: allocate physical register sizes for spilling
|
||
- anv: fix incorrect aspect flag for depth/stencil formats
|
||
- anv: fix missing push constant reallocation
|
||
- anv: prevent access to destroyed vk_sync objects post submission
|
||
|
||
Louis-Francis Ratté-Boulianne (18):
|
||
|
||
- gallium: add PIPE_CAP_ASTC_DECODE_MODE and expose extension
|
||
- panfrost: add support for ASTC decode mode extension
|
||
- pan/bi: implement nir_intrinsic_load_draw_id
|
||
- panfrost: use special DrawID register on CSF-based GPUs
|
||
- panfrost: properly lower DrawID sysval on v9 GPUs
|
||
- panfrost: initialize all UBO and vertex buffer descriptors
|
||
- pan/cs: only flush block instrs if there is a pending \`if` block
|
||
- pan/cs: add block to handle registers backup in exception handler
|
||
- pan/genxml: enable decoding for SET_EXCEPTION_HANDLER
|
||
- pan/genxml: small fix to FINISH_FRAGMENT tracing
|
||
- panfrost: add debug tracing for CSF context initialization
|
||
- panfrost: create a backend-specific hook for \`emit_fbds`
|
||
- panfrost: add debug flag to enable extra CS checks
|
||
- panfrost: add support for incremental rendering in CSF
|
||
- panfrost: properly update data size of AFBC-packed resources
|
||
- panfrost: disable CRC for AFBC-packed resources
|
||
- panfrost: check the right discard property when selecting CRC target
|
||
- panfrost: properly align CRC buffer size for prefetching
|
||
|
||
Lu Yao (1):
|
||
|
||
- ac/radeonsi: compute htile for tile mode RADEON_SURF_MODE_1D on GFX6-8
|
||
|
||
Lucas Fryzek (13):
|
||
|
||
- llvmpipe: Implement EGL_ANDROID_native_fence_sync
|
||
- lavapipe: Implement VK_KHR_external_*_fd
|
||
- venus/ci: Update skip tests to prevent timeouts
|
||
- lp: only map dt buffer on import from dmabuf
|
||
- egl/surfaceless: Add error checking to create image
|
||
- llvmpipe: Add android platform integration
|
||
- vulkan/runtime: Use DMA_BUF to import anb
|
||
- lavapipe: Add android platform integration
|
||
- docs/android: Add llvmpipe instructions
|
||
- lavapipe: Add AHB extension
|
||
- drisw: Copy entire buffer ignoring damage regions
|
||
- egl/dri/wl: Move swrast damage region from put to swap
|
||
- lp: Only close udmabuf handle if its valid
|
||
|
||
Lucas Stach (18):
|
||
|
||
- etnaviv: drm: use COARSE clock for BO cache timing
|
||
- etnaviv: drm: use COARSE clock for timeouts when possible
|
||
- etnaviv: hwdb: add COMPUTE_ONLY cap
|
||
- etnaviv: properly set PIPE_CAP_GRAPHICS
|
||
- etnaviv: emit SAMPLER_LOG_SIZE on sampler state changes
|
||
- etnaviv: wire up missing gl_VertexID bits
|
||
- etnaviv: allow shader machine code dumps in release builds
|
||
- etnaviv: singlethread shader variant compiles when dumping shaders
|
||
- etnaviv: emit all PA shader attributes
|
||
- etnaviv: support more VS outputs on halti5 GPUs
|
||
- etnaviv: fix total varying count assertion
|
||
- etnaviv: validate number of VS outputs against GPU limit
|
||
- etnaviv: limit number of varyings to fit into VS outputs
|
||
- gbm: mark surface buffers as explicit flushed
|
||
- etnaviv: re-emit uniforms on sampler view changes when txs is used
|
||
- etnaviv: flush shader caches when UBO content is modified
|
||
- etnaviv: Update headers from rnndb
|
||
- etnaviv: set PE_COLOR_FORMAT_OVERWRITE when no color target is active
|
||
|
||
Luigi Santivetti (4):
|
||
|
||
- pvr: fix calculation for textures z position fractional part
|
||
- pvr: fix when to emit ppp state uniform base address
|
||
- pvr: for fragment stages only do not insert a barrier
|
||
- pvr: really free memory in subpass render init
|
||
|
||
M Henning (15):
|
||
|
||
- nak: Don't mix up two types of barrier
|
||
- nak: Rename num_barriers to num_control_barriers
|
||
- nouveau/codegen: Switch to derivative intrinsics
|
||
- nv50/peephole: merge/split can swap word order
|
||
- nv50_ir_from_nir: Use getFile() more often
|
||
- nv/codegen: Use nir_lower_mem_access_bit_sizes
|
||
- nv/codegen: Limit MemoryOpt to input/output/const
|
||
- nv/codegen: Vector load ssbo/global/shared/scratch
|
||
- nv/codegen: Vector store ssbo/global/shared/scratc
|
||
- nv/codegen: Modify nir pass order
|
||
- nv/codegen: Use nir_opt_load_store_vectorize
|
||
- nv/codegen: Stop overaligning var_function_temp
|
||
- nak: Phi coalescing via biased register coloring
|
||
- nvk/cmd_buffer: Pass count to set_root_array
|
||
- nvk: Fix invalidation of NVK_CBUF_TYPE_DYNAMIC_UBO
|
||
|
||
Maaz Mombasawala (4):
|
||
|
||
- svga: Validate surface during copy check
|
||
- svga: Ensure an active context exists for a vmw_screen
|
||
- svga: Introduce userspace managed surfaces
|
||
- svga: Support older define commands for userspace surfaces
|
||
|
||
Mahmood - Zer0xFF (1):
|
||
|
||
- d3d12: Fix Xbox GDK Build
|
||
|
||
Marcin Radomski (7):
|
||
|
||
- Add missing includes
|
||
- Enable VK_EXT_image_drm_format_modifier in cerealgenerator.py
|
||
- Enable VK_EXT_image_drm_format_modifier on Linux guests
|
||
- vkAllocateMemory: support DRM format modifiers for dmabuf memory
|
||
- vkGetPhysicalDeviceImageFormatProperties2: support drm modifiers
|
||
- vkCreateInfo: support VkImageDrmFormatModifierListCreateInfoEXT
|
||
- Add mesa3d_platforms Soong config flag
|
||
|
||
Marek Olšák (189):
|
||
|
||
- nir/opt_algebraic: use fmulz for fpow lowering to fix incorrect rendering
|
||
- nir: add nir_intrinsic_load_per_primitive_input, split from io_semantics flag
|
||
- nir/opt_varyings: improve convergent input handling to fix data corruption
|
||
- amd: expose nir_io_mix_convergent_flat_with_interpolated
|
||
- ac,radeonsi: set 16-bit flags in io_options optimally
|
||
- nir/opt_vectorize_io: optionally don't vectorize IO with different types
|
||
- ac/surface/gfx12: fix setting tile_swizzle
|
||
- radeonsi/gfx12: fix a GPU hang due to an invalid packet with window rectangles
|
||
- radeonsi: ensure TC_L2_dirty is set if we don't sync after internal SSBO blits
|
||
- radeonsi: don't set TC_L2_dirty in si_query_hw_get_result_resource
|
||
- radeonsi: fix buffer coherency issues on gfx6-8,12 due to missing PFP->ME sync
|
||
- radeonsi/gfx12: fix register programming to fix GPU hangs
|
||
- radeonsi/gfx11: prepare for CWSR
|
||
- radeonsi: program OREO_MODE optimally on gfx11.x and gfx12
|
||
- radeonsi/gfx12: fix VS output corruption with streamout
|
||
- ac/surface/gfx12: turn off HiZ for pre-production samples
|
||
- radeonsi: strengthen the condition determining OREO_MODE
|
||
- nir: add ACCESS_KEEP_SCALAR, preventing vectorization
|
||
- util: move util_lower_clearsize_to_dword here
|
||
- radeonsi: reject insert/extract opcodes in si_vectorize_callback
|
||
- radeonsi: test more alignment cases in si_test_dma_perf
|
||
- radeonsi: add correctness tests for the clear/copy_buffer compute shader
|
||
- radeonsi: minor changes at the beginning of si_compute_clear_copy_buffer
|
||
- radeonsi: implement optimized unaligned clear/copy_buffer compute shader
|
||
- radeonsi: align waves to 256B clear/copy area for the clear/copy_buffer shader
|
||
- ac/nir: import the clear/copy_buffer compute shader from radeonsi
|
||
- ac/nir: adjust performance-related decisions for clear/copy_buffer shader
|
||
- ac/nir: adjust gfx11 tuning for the compute blit
|
||
- amd: update addrlib
|
||
- ac/nir/tess: don't allocate LDS for HS inputs that are passed via VGPRs
|
||
- radeonsi: pass TCS inputs_read mask to LS output lowering on GFX9 + monolithic
|
||
- ac: fix WAVES_PER_SH value for gfx12
|
||
- ac/nir/meta: move the "skip compute if no DCC image stores" condition to common
|
||
- ac,radeonsi: update comments related to the L2 cache, use "L2", not "TC"
|
||
- radeonsi/gfx12: remove CP DMA workarounds because CP DMA is never used on gfx12
|
||
- radeonsi/gfx12: disallow DCC for protected content
|
||
- radeonsi/gfx11: disable RB+ when blending
|
||
- radeonsi: use better OREO_MODE programming
|
||
- radeonsi: tweak si_test_dma_perf for better experience
|
||
- radeonsi: add a new PM4 helper radeon_event_write
|
||
- radeonsi: remove SI_CONTEXT_VGT_STREAMOUT_SYNC, emit it directly
|
||
- radeonsi: add ACQUIRE_MEM, RELEASE_MEM PWS packet helpers
|
||
- radeonsi: add si_cp_acquire_mem helper and clean up its usage for gfx6-9
|
||
- radeonsi: add gfx10+ support into si_cp_acquire_mem
|
||
- radeonsi: use the correct cs for SQTT barriers for gfx9
|
||
- radeonsi: add si_cp_pfp_sync_me
|
||
- radeonsi: remove unused size parameter from get_cache_policy
|
||
- radeonsi: don't use get_cache_policy in si_launch_grid_internal_ssbos
|
||
- radeonsi: move get_cache_policy into si_cp_dma.c
|
||
- radeonsi: don't use SI_COHERENCY_NONE in si_compute_shorten_ubyte_buffer
|
||
- radeonsi: don't use SI_COHERENCY_CB_META
|
||
- radeonsi: don't use SI_COHERENCY_CP
|
||
- radeonsi: remove all SI_COHERENCY_* flags except SI_COHERENCY_SHADER
|
||
- radeonsi: remove enum si_coherency
|
||
- radeonsi: don't use si_get_flush_flags in si_launch_grid_internal_ssbos
|
||
- radeonsi: move enum si_cache_policy and si_get_flush_flags into si_cp_dma.c
|
||
- radeonsi: remove L2_STREAM enum
|
||
- radeonsi: replace enum si_cache_policy with cp_dma_has_L2()
|
||
- radeonsi: inline si_get_flush_flags
|
||
- radeonsi: split si_launch_grid_internal to start preparing for barrier rework
|
||
- radeonsi: move compute barrier calls up one level
|
||
- radeonsi: move barrier code out of si_launch_grid_internal_ssbos/images
|
||
- radeonsi: skip no-op si_compute_blit()
|
||
- radeonsi: only set need_check_render_feedback if binding textures for PS
|
||
- radeonsi: split image saving/binding/restoring from si_launch_grid_internal_images
|
||
- radeonsi: do init/deinit for multi-dispatch compute blits only once
|
||
- radeonsi: don't flag both VS and PS partial flush in si_barrier_before_internal_op
|
||
- radeonsi: remove CP DMA code for GDS & L2 prefetch in the clear/copy_buffer path
|
||
- radeonsi: use si_barrier_{before,after}_internal_op for CP DMA
|
||
- radeonsi: remove SI_OP_SYNC_CPDMA_BEFORE, always sync CP DMA
|
||
- radeonsi: fold si_improve_sync_flags logic into si_barrier_before_internal_op
|
||
- radeonsi: fix waiting for PS to finish if there are no CB/DB bindings on gfx6-8
|
||
- radeonsi/gfx12: fix shader uploads via CP DMA
|
||
- ac/nir/meta: tune clear/copy_buffer performance for gfx6-10.3
|
||
- radeonsi: fix broken tessellation in Unigine Heaven
|
||
- radeonsi: remove SI_OP_CS_IMAGE
|
||
- radeonsi: inline si_launch_grid_internal_images without the barriers
|
||
- radeonsi: move barriers out of si_launch_grid_internal_ssbos
|
||
- radeonsi: remove unnecessary barriers from clear_buffer_rmw and clear_dcc_msaa
|
||
- radeonsi: add barrier helpers for simple internal buffer ops
|
||
- radeonsi: move barriers out of si_compute_clear_copy_buffer & si_cp_dma_*
|
||
- radeonsi: move barriers out of si_clear_buffer
|
||
- radeonsi: move barriers out of si_copy_buffer
|
||
- radeonsi: move barriers out of si_compute_shorten_ubyte_buffer
|
||
- radeonsi: remove SI_OP_SKIP_CACHE_INV_BEFORE
|
||
- radeonsi: remove SI_OP_SYNC_AFTER
|
||
- radeonsi: always pass flags=0 into si_barrier_after_internal_op/simple_buffer_op
|
||
- radeonsi: always wait for idle before get_query_result_resource
|
||
- radeonsi: remove unused SI_OP_IS_NESTED
|
||
- radeonsi: remove SI_OP_SYNC_BEFORE* flags
|
||
- radeonsi: change the flags parameter of si_compute_blit to bool fail_if_slow
|
||
- radeonsi: remove unused flags and user_flags params from clear/copy functions
|
||
- radeonsi: enable conditional rendering for si_compute_shorten_ubyte_buffer
|
||
- radeonsi: replace SI_OP_CS_RENDER_COND_ENABLE with bool render_condition_enable
|
||
- radeonsi: remove SI_CP_DMA_CLEAR_METHOD, call si_cp_dma_clear_buffer directly
|
||
- radeonsi: remove unused buffer_subdata fallback from si_clear_buffer
|
||
- radeonsi: rename "cache_flush" -> "barrier"
|
||
- radeonsi: move most barrier code into a new file si_barrier.c
|
||
- radeonsi: update remaining comments related to the L2 cache, use "L2", not "TC"
|
||
- radeonsi: rename TC_L2_dirty -> L2_cache_dirty
|
||
- radeonsi: check sctx->flags inside si_emit_barrier_direct
|
||
- radeonsi: rewrite si_memory_barrier to generate fewer barrier flags
|
||
- radeonsi: rename si_context::flags -> barrier_flags
|
||
- radeonsi: move initial framebuffer barrier code into si_barrier.c
|
||
- radeonsi: restructure fb_barrier code in si_launch_grid
|
||
- radeonsi: move CB synchronization into si_fb_barrier_after_rendering
|
||
- radeonsi: move DB synchronization into si_fb_barrier_after_rendering
|
||
- radeonsi: add si_fb_barrier_before_rendering
|
||
- radeonsi: rename SI_CONTEXT_* flags to SI_BARRIER_* flags
|
||
- radeonsi: consolidate code around unsetting barrier_flags in emit_barrier
|
||
- radeonsi: remove an obsolete comment about SMEM stores
|
||
- radeonsi: skip CB_META/DB_META events when they are redundant on gfx10-11
|
||
- radeonsi: deduplicate how GLM flush flags are set for gfx10-11
|
||
- radeonsi: merge the CB/DB conditional blocks in gfx10_emit_barrier
|
||
- mesa,glsl,gallium: add GL_OVR_multiview
|
||
- radeonsi: check and update compute_is_busy in get_reduced_barrier_flags
|
||
- radeonsi: don't sync VS and PS if they are idle
|
||
- radeonsi: count VS/PS/CS/L2 flushes in get_reduced_barrier_flags
|
||
- radeonsi: deduplicate code emitting VGT_FLUSH/PIPELINESTAT events
|
||
- radeonsi: don't sync CS and PS before rendering if there are no FBO attachments
|
||
- radeonsi: remove CB sync after FMASK and DCC decompression
|
||
- radeonsi: wait for idle after end_query in si_test_blit_perf
|
||
- radeonsi: don't pad esgs_vertex_stride if it's 0
|
||
- radeonsi: adjust GFX12 checks in si_compute.c
|
||
- radeonsi: clean up and make corrections to si_create_fmask_expand_cs
|
||
- radeonsi/aco: fix asm dumps to debug output via radeonsi_debug_disassembly=true
|
||
- radeonsi: move si_execute_clears barrier code into separate functions
|
||
- radeonsi: move barriers out of si_execute_clears
|
||
- radeonsi: execute clears at resource allocation using compute instead of gfx
|
||
- radeonsi: remove barriers around clears using aux_context.compute_resource_init
|
||
- radeonsi: don't use VS/PS/CS partial flushes if we use a TS event
|
||
- radeonsi: clean up set_log_context code for all aux contexts
|
||
- radeonsi: use ACO on GFX11.5 with LLVM 18 or older to work around GPU hangs
|
||
- radeonsi: don't insert any barrier after the copy for PIPE_MAP_READ
|
||
- radeonsi: rename hw_level -> view_level
|
||
- radeonsi: remove the make_texture_descriptor indirect function call
|
||
- nir/opt_shrink_vectors: shrink memory loads, not just IO
|
||
- ac: make sure VEGA20 and MI200 version ranges don't overlap with other chips
|
||
- ac/nir: set .image_dim and .image_array for all opcodes
|
||
- ac/llvm: use LLVM processor gfx942 for GFX940 when it's available
|
||
- ac/llvm: don't use the 64-bit umul_hi workaround with LLVM 19.1
|
||
- ac/gpu_info: print 32bpp modifiers
|
||
- st/mesa: copy some TES shader info fields to TCS
|
||
- nir/opt_vectorize_io: fix skipped output vectorization if inputs were vectorized
|
||
- nir/opt_vectorize_io: fix stack buffer overflow with 16-bit output stores
|
||
- nir/opt_load_store_vectorize: add entry::num_components
|
||
- nir: add hole_size parameter into the vectorize callback
|
||
- nir: reject unsupported component counts from all vectorize callbacks
|
||
- nir/opt_load_store_vectorize: allow overfetching, merge overfetched loads
|
||
- nir/opt_load_store_vectorize: allow a 4-byte hole between 2 loads
|
||
- nir/opt_load_store_vectorize: vectorize load_smem_amd
|
||
- gallium/u_threaded: fix crash in tc_create_image_handle due to resource == NULL
|
||
- radeonsi: set the valid buffer range for bindless image buffers
|
||
- nir/opt_varyings: assign locations of no_varying IO for TCS outputs only
|
||
- nir/opt_varyings: pack TCS inputs with cross-invocation access together
|
||
- ac/nir: rewrite ac_nir_lower_ps epilog to fix dual src blending with mono PS
|
||
- nir: add shader_info::tess::tcs_same_invocation_inputs_read(_indirect)
|
||
- mesa_interface: remove unused stuff
|
||
- mesa_interface: replace opaque __DRIscreen with struct dri_screen everywhere
|
||
- mesa_interface: replace opaque __DRIcontext with struct dri_context everywhere
|
||
- mesa_interface: replace opaque __DRIdrawable with struct dri_drawable everywhere
|
||
- mesa_interface: remove redundant __DRI*Rec types
|
||
- mesa_interface: rename __DRIconfig to struct dri_config
|
||
- mesa_interface: rename __DRIimage to struct dri_image
|
||
- nir: add nir_gather_tcs_info, new gathering/analysis pass
|
||
- nir: rename load_cull_small_primitives -> triangles, add load_cull_small_lines
|
||
- nir: rename load_cull_small_primitive_precision -> triangle, add line_precision
|
||
- nir: add cull_triangles\_, cull_lines\_ prefixes to viewport_xy_scale_and_offset
|
||
- ac/nir,radeonsi: use load_cull_small_lines_enabled_amd
|
||
- ac/nir,radeonsi: use load_cull_small_line_precision_amd
|
||
- ac/nir,radeonsi: use load_cull_line_viewport_xy_scale_and_offset_amd
|
||
- radeonsi: rewrite how small prim precision is passed to culling code
|
||
- radeonsi: simplify util_rast_prim_is_lines_or_triangles
|
||
- radeonsi/gfx11: fix Z corruption for Blender
|
||
- amd: move Tonga and Iceland TC-compat HTILE workarounds to ac_gpu_info.c
|
||
- radeonsi/gfx12: fix AMD_DEBUG=nodcc not working
|
||
- glsl: lower IO in the linker if enabled, don't lower it later
|
||
- nir: rename nir_io_glsl_lower_derefs -> nir_io_has_io_intrinsics
|
||
- nir: print interp_mode better
|
||
- nir/lower_io: change INTERP_MODE_NONE to SMOOTH when NONE means SMOOTH
|
||
- nir: add nir_clear_divergence_info, use it in nir_opt_varyings
|
||
- gallium: set proper type for pipe_shader_state::ir::nir
|
||
- st/mesa: fix incorrect types of shader CSOs
|
||
- amd/ci: adjust stoney traces checksums
|
||
- glsl: fix accidentally disabling nir_opt_varyings for all drivers
|
||
- Revert "amd/ci: adjust stoney traces checksums"
|
||
- st/mesa: implement key->persample_shading for lowered IO
|
||
- nir/lower_clip_disable: handle non-scalar store intrinsics
|
||
- radeonsi: fix gl_FrontFace elimination when one side is culled
|
||
|
||
Mark Burton (1):
|
||
|
||
- gallivm: Fix compilation errors when using LLVM 13.
|
||
|
||
Mark Collins (2):
|
||
|
||
- freedreno/registers: Document TPL1_2D_SRC_CNTL register
|
||
- tu: Improve 2D buffer-to-image copies for A7XX
|
||
|
||
Martin Krastev (12):
|
||
|
||
- svga/ci: enable vmware farm
|
||
- svga/ci: change DNS server for vmware jobs
|
||
- svga/ci: triage unexpected piglit pass
|
||
- svga/ci: update FORCE_KERNEL_TAG
|
||
- svga/ci: triage unexpected piglit passes
|
||
- svga/ci: disable vmware-qemu-traces jobs
|
||
- svga/ci: disable vmware farm
|
||
- svga/ci: enable vmware farm
|
||
- svga/ci: triage piglit tessellation failures
|
||
- svga/ci: disable vmware farm
|
||
- svga/ci: enable vmware farm
|
||
- svga/ci: disable vmware farm
|
||
|
||
Martin Roukala (né Peres) (9):
|
||
|
||
- radv/ci: update the flakes lists
|
||
- zink/ci: update the flakes list for ga106
|
||
- radv/ci: document more vkcts flakes
|
||
- zink/ci: document more flakes
|
||
- radv/ci: run vkcts-navi31-valve pre-merge
|
||
- freedreno/ci: fix the stage of the a750 jobs
|
||
- radv/ci: document more vkcts flakes
|
||
- zink/ci: document more radv flakes
|
||
- zink/ci: bump the polaris10 timeout to 30 minutes
|
||
|
||
Mary Guillemard (61):
|
||
|
||
- panvk: Update CI fails
|
||
- docs: Add panvk in VK_EXT_private_data to features.txt
|
||
- panvk: Implement CmdPushDescriptorSetWithTemplate2KHR
|
||
- panvk: Fix image support in vertex jobs
|
||
- panvk: Emit viewport when state is NULL
|
||
- panvk: Implement CmdBindDescriptorSets2KHR, CmdPushConstants2KHR and CmdPushDescriptorSet2KHR
|
||
- panfrost: Add B8G8R8A8_SNORM in panfrost_pipe_format
|
||
- panvk: Pass attrib_buf_idx_offset to desc_copy_info
|
||
- panvk: Fix NULL deref on model name when device isn't supported
|
||
- panvk: Skip blend descriptors when no fragment shader is present
|
||
- bi: Add bi_foreach_ssa_dest
|
||
- bi: Use bi_foreach_src in bi_foreach_ssa_src
|
||
- bi: Do not mark tex ops as skip when dest is used by control flow
|
||
- bi: Use nir_opt_load_store_vectorize
|
||
- bi: Use nir_opt_shrink_stores
|
||
- bi: Add loop_header in bi_block
|
||
- bi: Rewrite dead code elimination
|
||
- panvk/ci: Update G52 fail list to represent current state
|
||
- panvk: Properly propagate helper invocations requirement
|
||
- panvk: Expose trivial google extensions
|
||
- panvk: Fix viewport calculation
|
||
- nak: Use F2FP for nir_op_pack_half_2x16_split on SM86+
|
||
- panvk: Close batch in CmdBeginRendering if already open
|
||
- panvk: Manually copy dynamic state in meta_gfx
|
||
- panvk: Simplify meta ctx save/restore
|
||
- panvk: Save and restore vertex buffer in gfx_meta
|
||
- panvk: Remove unused vk_vertex_input_state in graphics_save_ctx
|
||
- panvk: Extend the descriptor lowering pass to support Valhall
|
||
- panvk: Extend the shader logic to support Valhall
|
||
- nak: Fix wrong encoding for F2FP cbuf and immediate forms
|
||
- panfrost: Update panfrost_drm.h
|
||
- panfrost: Update panthor_drm.h
|
||
- pan/kmod: Add timestamp uapi support
|
||
- panfrost: Move occlusion query allocation to create_query
|
||
- panfrost: Implement timer queries
|
||
- docs: Mark timer queries as done on panfrost
|
||
- panvk: Add atomic modifiers for r32 based formats
|
||
- panvk: Reject SNORM only for color attachment and blend
|
||
- panvk: Return proper buffer format feature flags
|
||
- pan/lib: Simplify pan_blitter
|
||
- panvk: Include new failures from VKCTS 1.3.9.1
|
||
- panvk: Ensure to clear dirty dynamic state in panvk_cmd_draw
|
||
- mr-label-maker: Add panvk
|
||
- pan/va: Define the LEA_BUFFER instruction
|
||
- panvk/ci: Update Mali-G52 status with full VKCTS
|
||
- panvk/ci: Update to run full CTS on G52
|
||
- panvk/ci: Make panfrost-g52-vk run at premerge
|
||
- egl: Support NV_context_priority_realtime
|
||
- panfrost: Update panthor_drm.h with latest
|
||
- pan/kmod: Add priority query uapi support
|
||
- panfrost: Implement context priority on v10
|
||
- panvk/ci: Increase panfrost-g52-vk fraction
|
||
- panvk: Cleanup rw_nc pool in panvk_device_cleanup_mempools
|
||
- hk: Implement global priority extensions
|
||
- panvk: Define primitive size for RUN_TILER/RUN_IDVS
|
||
- winsys/nouveau: Rework to use u_pipe_screen_lookup_or_create
|
||
- winsys/nouveau: Reformat to stop relying on tabs
|
||
- nir: Move atomic_op_to_alu to common code
|
||
- hk: Fill deviceUUID
|
||
- panvk: Ensure that render_info is not null in force_fb_preload
|
||
- bi: Execute nir_opt_algebraic after nir_lower_pack
|
||
|
||
Matt Coster (3):
|
||
|
||
- pvr: Fix ds subtile alignment NULL pointer dereference
|
||
- pvr: Fix reordering of sub-cmds when performing ds subtile alignment
|
||
- pvr: Use vk_buffer_view base
|
||
|
||
Matt Turner (14):
|
||
|
||
- intel/clc: Free parsed_spirv_data
|
||
- intel/clc: Free disk_cache
|
||
- intel/brw: Use REG_CLASS_COUNT
|
||
- intel/elk: Use REG_CLASS_COUNT
|
||
- docs: Drop references to LIBGL_DRIVERS_PATH
|
||
- util: Add ATTRIBUTE_OPTIMIZE(flags)
|
||
- util: Force emission of stack frame in stack unit test
|
||
- nir: Skip opt_if_merge when next_if has block ending in a jump
|
||
- nir/tests: Add tests for opt_if_merge
|
||
- anv: Set shader_spilling_rate=15 by default
|
||
- anv: Set shader_spilling_rate=11
|
||
- anv: Align anv_descriptor_pool::host_mem
|
||
- vulkan: Avoid pointer aliasing
|
||
- nir: Get correct number of components
|
||
|
||
Mauro Rossi (2):
|
||
|
||
- nvk: Fix regression observed on Kepler
|
||
- android: gbm: split the DRI backend into separate library
|
||
|
||
Maíra Canal (2):
|
||
|
||
- v3d, vc4: remove unused functions
|
||
- v3d: Don't use performance counters names array with an older kernel
|
||
|
||
Michael Catanzaro (1):
|
||
|
||
- Disk cache should consistently create parent directories when needed
|
||
|
||
Michael Cheng (2):
|
||
|
||
- anv: Remove extra hdc_flush from Perfetto
|
||
- anv: move trace logic to batch_emit_pipe_control_write
|
||
|
||
Michel Dänzer (10):
|
||
|
||
- Revert "radeonsi: remove CB sync after FMASK and DCC decompression"
|
||
- radeonsi: Revert to GLSL_SAMPLER_DIM_2D in si_create_fmask_expand_cs
|
||
- util/mesa-db: Make mesa_db_lock robust against signals
|
||
- util/mesa-db: Recreate files if header load or index update fails
|
||
- util/mesa-db: Reserve hash table for total number of index entries
|
||
- util/mesa-db: Use single read for whole index
|
||
- util: Use persistent array of index entries
|
||
- util/mesa-db: Further simplify mesa_db_compact
|
||
- Revert "util/mesa-db: Further simplify mesa_db_compact"
|
||
- Revert "util: Use persistent array of index entries"
|
||
|
||
Michel Zou (1):
|
||
|
||
- ac/gpu_info: Fix missing prototype mingw error
|
||
|
||
Mike Blumenkrantz (389):
|
||
|
||
- zink: move some flags from driver_workarounds to separate struct
|
||
- zink: hash in some driver workaround flags to shader cache
|
||
- zink: add lavapipe to can_do_invalid_linear_modifier list
|
||
- zink: use local screen variable in zink_prep_fb_attachment
|
||
- zink: use GENERAL layout for depth attachments on some drivers
|
||
- zink: use blake3 instead of sha1 for program cache
|
||
- gallium: install gallium-$version.so to libdir
|
||
- ci: prune dri from LD_LIBRARY_PATH
|
||
- dril: rework config creation
|
||
- llvmpipe: handle vma allocation failure
|
||
- llvmpipe: only use vma allocations on linux
|
||
- glx: delete tests
|
||
- st/interop: return unsupported without screen hooks
|
||
- dri: fix kmsro define
|
||
- glx: tabs -> spaces
|
||
- glx: delete __GLXDRIdisplay
|
||
- glx: simplify zink init variables
|
||
- glx: rework __glXInitialize
|
||
- glx: delete dri3_display::loader_extensions
|
||
- glx: move has_multibuffer to base glx_screen struct
|
||
- glx: move driwindows_display::event_base to driwindows_context
|
||
- glx: rework/simplify drisw zink handling
|
||
- glx: move dri2Hash to base glx_display
|
||
- glx: tweak dri2 init ordering
|
||
- glx: move dri2 init checks to separate function
|
||
- glx: move dri2 loader extensions to static vtable
|
||
- glx: delete all driver-specific display handling
|
||
- glx: use base screen has_multibuffer for drisw
|
||
- glx: move drisw dri3 handling for zink out to glxInitialize
|
||
- glx: consolidate some LIBGL_KOPPER_DISABLE checks
|
||
- glx: unhackify inferring zink from dri3
|
||
- Revert "vl/dri3: use loader's dri3 init code and delete everything else"
|
||
- loader: move some common dri3 functions out of dri3 loader
|
||
- meson: move glx subdir after gallium build
|
||
- loader: split out dri3 into subdir
|
||
- dri: declare IMAGE_DRIVER symbols PUBLIC
|
||
- glx: inline IMAGE_DRIVER functions
|
||
- dri: declare DRI_CORE functions public
|
||
- loader/dri3: inline DRI_CORE
|
||
- glx: inline DRI_CORE functions
|
||
- glx: inline createNewScreen3
|
||
- glx: delete __DRImesaCoreExtension usage
|
||
- glx: inline DRI_SWRAST
|
||
- dri: delete __DRImesaCoreExtensionRec::createNewScreen
|
||
- glx: inline DRI_DRI2
|
||
- dri: delete driCreateNewScreen2
|
||
- dri: declare DRI_KOPPER as PUBLIC
|
||
- glx: use local var for LIBGL_KOPPER_DISABLE check
|
||
- glx: switch kopper check for setting up swapinterval/bufferage functions
|
||
- glx: delete DRI_KOPPER
|
||
- glx: rename driswCopySubBuffer
|
||
- drisw: declare DRI_COPY_SUB_BUFFER as PUBLIC
|
||
- glx: inline DRI_COPY_SUB_BUFFER
|
||
- dri: delete __DRI_COPY_SUB_BUFFER
|
||
- glx: delete releaseTexBuffer
|
||
- glx: delete the only users of __DRItexBufferExtension::setTexBuffer
|
||
- dri: delete __DRItexBufferExtension::setTexBuffer
|
||
- dri: declare DRI_TEX_BUFFER functions public
|
||
- glx: delete check for texbuffer extension
|
||
- glx: inline DRI_TEX_BUFFER
|
||
- loader/dri3: delete texBuffer struct member
|
||
- glx: delete DRI_TEX_BUFFER handling
|
||
- kopper: increment lastStamp during invalidate
|
||
- dri: add kopper stubs to avoid build failures
|
||
- dri: make DRI2_RENDERER_QUERY interface PUBLIC
|
||
- glx: inline __DRI2_RENDERER_QUERY
|
||
- dri: delete __DRI2_RENDERER_QUERY interface
|
||
- glx: always expose GLX_ARB_context_flush_control
|
||
- dri: make DRI_FLUSH public
|
||
- kopper: reuse dri2 invalidate
|
||
- loader/dri3: inline DRI2_FLUSH
|
||
- glx: inline DRI2_FLUSH
|
||
- loader/dri3: delete DRI2_FLUSH remnants
|
||
- dri: unify DRI2_CONFIG_QUERY interfaces
|
||
- dri: make DRI2_CONFIG_QUERY public
|
||
- dri: move swapinterval functions from loader to dri frontend
|
||
- dri: remove __DRI2configQueryExtension param from swapinterval functions
|
||
- glx: inline DRI2_CONFIG_QUERY
|
||
- loader/dri3: delete DRI2_CONFIG_QUERY remnants
|
||
- dri: make DRI2_THROTTLE public
|
||
- glx: inline DRI2_THROTTLE
|
||
- dri: delete DRI2_THROTTLE interface
|
||
- dri: make DRI_INTEROP public
|
||
- glx: inline DRI_INTEROP
|
||
- glx: delete checks for interop support
|
||
- glx: delete more DRI_INTEROP checks
|
||
- glx: call dri interop functions directly
|
||
- glx: stop binding DRI_INTEROP
|
||
- glx: expose GLX_MESA_gl_interop for drisw
|
||
- glx: simplify DRI_IMAGE checks
|
||
- dri: make DRI_IMAGE public
|
||
- gallium: move loader_dri_create_image to dri frontend
|
||
- dri: use image loader function directly in dri_create_image_with_modifiers
|
||
- loader/dri3: inline DRI_IMAGE usage
|
||
- loader/dri: delete DRI_IMAGE remnants
|
||
- glx: delete DRI_IMAGE remnants
|
||
- loader/dri3: delete loader_dri3_extensions
|
||
- glx: move up glx_driver enum declaration
|
||
- glx: include src/gallium for apple
|
||
- dri: link with libloader
|
||
- dri: only link_whole once
|
||
- kopper: check swapchain size after possible loader image resize
|
||
- egl/dri2: delete unused function
|
||
- egl: inline DRI_CORE usage
|
||
- gbm: inline DRI_CORE usage
|
||
- egl/gbm: delete remnants of DRI_CORE
|
||
- dri: publicize dri_query_compatible_render_only_device_fd
|
||
- gbm: inline DRI_MESA
|
||
- egl: inline DRI_MESA
|
||
- gbm: delete DRI_MESA remnants
|
||
- egl: delete DRI_MESA remnants
|
||
- egl: inline DRI_KOPPER
|
||
- egl/gbm: remove DRI_KOPPER remnants
|
||
- dri: make driSWRastQueryBufferAge public
|
||
- egl: inline DRI_SWRAST
|
||
- egl/gbm: rip out DRI_SWRAST
|
||
- egl: collapse driCreateNewDrawable conditional
|
||
- egl: simplify a non-swrast conditional
|
||
- egl/drm: always expose EXT_buffer_age
|
||
- gbm: inline DRI_IMAGE_DRIVER
|
||
- egl/gbm: delete DRI_IMAGE_DRIVER remnants
|
||
- dri: store dmabuf import caps to dri_screen on create
|
||
- dri: check for dmabuf_import at top of dri2_from_dma_bufs
|
||
- gbm: inline dri2_from_dma_bufs and delete now-redundant check in dmabuf import
|
||
- egl: simplify MESA_drm_image enablement
|
||
- dri: make a few more functions public
|
||
- egl: inline simple DRI_IMAGE usage
|
||
- gbm: inline simple DRI_IMAGE usage
|
||
- egl: swap DRI_IMAGE checks for dmabuf/modifier support for driver check
|
||
- egl/android: check PIPE_CAP_NATIVE_FENCE_FD directly
|
||
- egl/wayland: delete check for DRI_IMAGE blitImage in drm init
|
||
- egl: inline DRI_IMAGE compression modifier handling
|
||
- egl: replace display_name checks with swrast checks
|
||
- egl: always check dri caps for EGL_WL_bind_wayland_display support
|
||
- gbm: check for dmabuf import on device creation
|
||
- gbm: use new dmabuf import flag to replace DRI_IMAGE checks
|
||
- gbm: always pass through images to unmap
|
||
- gbm: check for compression modifiers on device create
|
||
- gbm: use driver check for dmabuf export
|
||
- gbm: delete DRI_IMAGE remnants
|
||
- egl: move some extension enablement out DRI_IMAGE conditional
|
||
- egl: delete last DRI_IMAGE check
|
||
- egl: delete DRI_IMAGE remnants
|
||
- egl: delete local_buffers
|
||
- egl/x11: replace DRI_DRI2 check with fd check
|
||
- egl: delete DRI_DRI2 remnants
|
||
- gallium: make some sw screen create functions public
|
||
- pipe-loader: fix driconf memory management
|
||
- egl: inline CONFIG_OPTIONS usage
|
||
- egl: delete DRI_CONFIG_OPTIONS remnants
|
||
- egl: further simplify driver loading
|
||
- egl: add a bool to indicate swrast but not kms_swrast
|
||
- gbm: inline DRI_FLUSH
|
||
- gbm: delete DRI_FLUSH remnants
|
||
- egl: use a non-kms_swrast check for swapbuffers DRI2_FLUSH
|
||
- egl: use a swrast check for a swapbuffers DRI2_FLUSH case
|
||
- egl: delete a DRI2_FLUSH check in dri2 swapbuffers
|
||
- egl: replace another DRI2_FLUSH check with non-kms_swrast check in dri2_wait_client
|
||
- egl: inline DRI2_FLUSH usage
|
||
- egl: replace another DRI2_FLUSH check with a non-kms_swrast check in surface query
|
||
- egl: replace a DRI2_FLUSH check with swrast check in x11_copy_buffers
|
||
- egl/drm: replace a DRI2_FLUSH check with a non-kms_swarst check
|
||
- egl/wayland: replace a couple DRI2_FLUSH checks with non-kms_swrast checks
|
||
- egl: delete DRI2_FLUSH remnants
|
||
- dri: only add DRI_BUFFER_DAMAGE if the driver supports set_damage_region
|
||
- egl: simplify DRI_BUFFER_DAMAGE checks
|
||
- dri: make set_damage_region public
|
||
- egl: add a util to get the pipe_screen
|
||
- egl: use pipe_screen check for damage region, simplify buffer_damage checks
|
||
- egl: inline DRI_BUFFER_DAMAGE
|
||
- egl: delete DRI_BUFFER_DAMAGE remnants
|
||
- egl: always advertise KHR_context_flush_control
|
||
- egl: delete DRI_FLUSH_CONTROL remnants
|
||
- egl: inline DRI2_CONFIG_QUERY
|
||
- egl: delete DRI_CONFIG_QUERY remnants
|
||
- egl: inline DRI_TEX_BUFFER
|
||
- egl: delete DRI_TEX_BUFFER remnants
|
||
- dri: make DRI2_FENCE public
|
||
- egl: delete checks for DRI2_FENCE
|
||
- egl: simplify native fence check
|
||
- egl: don't check for create_fence_fd method when importing a fence
|
||
- egl: inline DRI2_FENCE
|
||
- egl: delete DRI2_FENCE remnants
|
||
- egl: always advertise MESA_gl_interop, inline calls
|
||
- egl: delete DRI2_INTEROP remnants
|
||
- dri: make DRI2_BLOB public
|
||
- egl: inline DRI2_BLOB
|
||
- egl: delete DRI2_BLOB remnants
|
||
- egl/android: replace DRI_MUTABLE_RENDER_BUFFER_DRIVER check with non-kms_swrast check
|
||
- egl: delete DRI_MUTABLE_RENDER_BUFFER_DRIVER remnants
|
||
- egl: move multibuffers check into platform_x11
|
||
- egl: simplify multibuffers check
|
||
- dril: always take the egl init path
|
||
- llvmpipe: init fd=-1 to avoid closing fd=0 on failure
|
||
- egl: fix zink init
|
||
- glx: replace a straggler DRI_CONFIG_OPTIONS usage
|
||
- dri: delete unused DRI_MESA queryCompatibleRenderOnlyDeviceFd method
|
||
- dri: delete __DRIdri2Extension interface
|
||
- dri: delete a couple unused functions
|
||
- dri: delete some dri_screen cruft
|
||
- dri: consolidate a bunch of dri_screen init code
|
||
- dri: delete all the extra dri2 extension handling
|
||
- egl/dri: move some screen accessors back to dri frontend
|
||
- glx: stop checking dri extensions
|
||
- glx: deduplicate and move driverName to base glx_screen struct
|
||
- glx: store the DRIscreen to the base glx_screen
|
||
- glx: set drisw glx_driver more usefully
|
||
- glx: use glx_driver to determine drisw extension enablement
|
||
- glx: add a glx_screen member to indicate GLX_EXT_texture_from_pixmap
|
||
- glx: unify extension binding
|
||
- dri: delete dri_screen::extensions accessor
|
||
- dri: delete dri_screen::extensions
|
||
- dri: fix kms_swrast screen fail
|
||
- dri: delete more dead code
|
||
- dri: pass through a type enum for creating screen instead of driver_extensions
|
||
- kopper: inline DRI_IMAGE usage
|
||
- egl: stop loading driver extensions
|
||
- glx: stop loading driver extensions
|
||
- gbm: stop loading driver extensions
|
||
- dri: massively prune the dri target
|
||
- dri: delete a bunch of unused interfaces
|
||
- egl/wayland: bail on zink init in non-sw mode if extension check fails
|
||
- egl: fix fd passing on init with zink
|
||
- zink: fix partial update handling
|
||
- egl: make dmabuf/modifier display bools a bit more precise/intelligible
|
||
- egl/glx: make detecting explicit modifier support more explicit
|
||
- dri: set __DRI_IMAGE_ERROR_BAD_PARAMETER if driver doesn't support dmabuf import
|
||
- egl: initialize dri error code variables
|
||
- kopper: reuse drisw_update_tex_buffer to delete some code
|
||
- dri: merge in loader_dri3
|
||
- kopper: reuse dri3 functions
|
||
- dri: rename and move handle_in_fence to dri_helpers
|
||
- kopper: reuse dri_image_fence_sync
|
||
- dri: break out get_pixmap_buffer
|
||
- kopper: reuse loader_dri3_get_pixmap_buffer
|
||
- glx: always set swapInterval in drisw
|
||
- egl: check x11 multibuffer support before screen create
|
||
- dri: pass has_multibuffers through from the loader
|
||
- dri: further collapse dri_screen creation
|
||
- dri: guard dmabuf support by loader's dmabuf support
|
||
- dri: collapse some modifier/dmabuf checks
|
||
- kopper: call kopper_create_drawable directly
|
||
- kopper: delete some redundant code
|
||
- dri: rework drawable creation
|
||
- glx: simplify kopper_get_buffer_age
|
||
- glx: flatten out getBufferAge a bit
|
||
- glx: move/add __DRIdrawable struct to __GLXDRIdrawable
|
||
- glx: use __GLXDRIdrawable more broadly instead of display type drawables
|
||
- glx: unify dri bind_context
|
||
- glx: unify dri unbind context
|
||
- glx: unify dri destroy context
|
||
- glx: tweak some drisw context create code
|
||
- glx: tweak some dri3 context create code
|
||
- glx: unify dri create_context_attribs
|
||
- glx: unify renderer query hooks
|
||
- glx/dri2: strdup driver name
|
||
- glx/dri3: strdup existing driverName instead of fetching it again
|
||
- glx: unify dri get_driver_name
|
||
- glx: unify dri screen vtable
|
||
- glx: unify bind_tex_image
|
||
- glx: delete private DRIscreen pointers
|
||
- glx: move driver_configs to base glx_screen
|
||
- glx: rework screen destroy
|
||
- glx: move base screen destroy to glx_screen_cleanup
|
||
- glx/dri3: delete driScreenRenderGPU
|
||
- glx: unify more extension setup
|
||
- glx: determine is_direct_capable from driver type
|
||
- glx: inline __GLXDRIscreen onto base glx_screen
|
||
- glx: unify dri screen init
|
||
- zink: bail on choose_pdev immediately if no devices are available
|
||
- st/pbo: reject vs/fs pbo ops if rowstride < width
|
||
- device_select: reorder CreateInstance to have shorter failure paths
|
||
- device_select: shortcut EnumeratePhysicalDevice* for count-only calls
|
||
- device_select: disable reordering with zink on xwayland
|
||
- zink: delete hackaround for disabling wsi extensions
|
||
- egl: add DRI_USE_INVALIDATE with non-sw kopper
|
||
- egl: really fix kopper fd passing
|
||
- egl/wayland: split out dmabuf extension stuff for kopper
|
||
- egl: pass real value for software to dri2_setup_device()
|
||
- egl/wayland: do display gpu fd dance for non-sw zink init
|
||
- zink: don't skip cbuf store ops if resolve is set
|
||
- tc: set resolve on renderpass info if blit terminates the renderpass
|
||
- egl/android: call dri2_display_create() like every other driver
|
||
- lavapipe: VK_EXT_pipeline_robustness
|
||
- device-select: block xserver+zink reordering via xcb
|
||
- ci: add a660 flake
|
||
- dril: add zink stub
|
||
- egl: unify LIBGL_KOPPER_DRI2 checks
|
||
- egl/x11: pretend kopper is software with LIBGL_KOPPER_DRI2
|
||
- llvmpipe: export dmabuf caps for kms_swrast
|
||
- zink: update profile with missing extensions
|
||
- revert part of 94e470a32d214052d89f519a07566d8098d05d28
|
||
- egl: use more precise conditional for passing fd through to dri screen create
|
||
- dril: use the super fallback path for software fallback
|
||
- vk: rename DGC feature to DGC-NV
|
||
- zink: create a surface for resolve resources
|
||
- zink: use tc renderpass optimizing to inline resolves with dynamic render
|
||
- tc: break out blit enqueue to separate function
|
||
- tc: optimize out tracked winsys resolves
|
||
- dril: also create double-buffered configs in swrast fallback
|
||
- meson: delete dri3 build option
|
||
- meson: require dri3 modifiers
|
||
- ci: fix trace update script
|
||
- ci: add --pipeline-url to trace update script
|
||
- zink: delete erroneous kopper assert
|
||
- zink: set resolve layout when inferring scanout resolves
|
||
- zink: stop leaking inferred resolve surfaces
|
||
- zink: fix sparse bo deallocation
|
||
- va: flush_resource before presentation
|
||
- vl/winsys: move dri3_get_screen_for_root to common
|
||
- va/vl: add kopper support
|
||
- gallium: add PIPE_CAP_MULTIVIEW
|
||
- zink: support GL_OVR_multiview
|
||
- mesa: support OVR_multiview2
|
||
- zink: support OVR_multiview2
|
||
- docs: OVR_multiview features
|
||
- lavapipe: VK_EXT_pipeline_protected_access
|
||
- lavapipe: VK_KHR_global_priority
|
||
- egl: replace fd check with swrast check for dri2
|
||
- llvmpipe: bump max point size to 256
|
||
- egl/x11: fix dri3 connect conditional for kopper without modifiers
|
||
- lavapipe: bump mipmapPrecisionBits
|
||
- zink: add screen create to match windows LUID
|
||
- vl: rename/ifdef the x11 kopper functionality
|
||
- vl: add Windows kopper support
|
||
- va: support zink on Windows
|
||
- mesa: fix sample count handling for MSRTT
|
||
- mesa: OVR_multiview_multisampled_render_to_texture
|
||
- anv: fix video profile lists
|
||
- trace: add get_device_reset_status
|
||
- gallium: delete pipe_screen::is_compression_modifier
|
||
- egl/kopper: hook up EGL_EXT_surface_compression on wayland
|
||
- zink: hook up compression control extensions
|
||
- zink: store compression rate info for formats
|
||
- zink: implement compression control
|
||
- zink: move x8 format function to zink_format.c
|
||
- zink: fix zink_format_is_voidable_rgba_variant()
|
||
- zink: add X format variants for 30-bit formats
|
||
- zink: add some other missing X format emulation variants
|
||
- zink: add adl flake
|
||
- vk/image: fix view creation for planar video aspects
|
||
- zink: only clamp samplerview aspect for zs formats
|
||
- zink: fix planar image creation with mutable
|
||
- zink: correctly handle disjoint image creation
|
||
- zink: slightly refactor image tiling inference
|
||
- zink: swap sampler conversion conditional with yuv check
|
||
- zink: stop creating useless sampler conversions
|
||
- zink: move some image ici init up
|
||
- zink: check HAVE_LIBDRM for xf86drm.h include
|
||
- zink: only use DISJOINT for planar images
|
||
- zink: zink_modifier_prop -> zink_modifier_props
|
||
- zink: wrap all access to format_props and modifier_props
|
||
- zink: init format props dynamically
|
||
- zink: fix some indentation
|
||
- zink: delete redundant HIC usage check
|
||
- zink: add HIC back if suboptimal check fails
|
||
- zink: simplify some confusing modifier code
|
||
- zink: stop passing modifier pointers internally
|
||
- zink: rework modifier selection (again)
|
||
- zink: rework HIC image creation checks
|
||
- vk/runtime: add EXT DGC layout handling
|
||
- lavapipe: EXT DGC
|
||
- ci: bump gl cts versions
|
||
- ci: bump VVL to current week
|
||
- zink: revert compression control handling
|
||
- vl: add pipe_video_buffer::flags to sync up with pipe_resource::flags
|
||
- util/vbuf: delete/fix broken incompatible stride calc
|
||
- zink: block all 2d view creation with sparse
|
||
- zink: block dmabuf fallback into optimal tiling
|
||
- zink: assert images aren't created with dmabuf export and optimal tiling
|
||
- zink: also init format props when getting modifier props
|
||
- zink: clamp out dmabuf exports from optimal tiling images
|
||
- zink: fix external_only reporting for dmabuf formats
|
||
- zink: block srgb with winsys imports
|
||
- anv: add VK_FORMAT_G10X6_B10X6R10X6_2PLANE_420_UNORM_3PACK16 to modifier exceptions
|
||
- gallium: rework vbuf alignment pipe caps
|
||
- util/vbuf: rename/consolidate some bools
|
||
- gallium/sw: make llvmpipe the "default" sw driver
|
||
- util/framebuffer: add viewmask compare for fb equal
|
||
- iris: assert that viewmask is 0
|
||
- gallium: delete duplicated viewmask member in draw info
|
||
- llvmpipe: expose GL multiview extensions
|
||
- zink: further improve image usage detection
|
||
- zink: rewrite the high-IQ image tiling selection loop
|
||
- device-select: only try wayland/x11 if the required vars are set
|
||
- va: fail context create if driver does not support video
|
||
- vdpau: fail context create if driver does not support video
|
||
- zink: stop leaking precompiled generated tcs
|
||
- zink: add VVL for RADV jobs
|
||
|
||
Mike Lothian (1):
|
||
|
||
- Revert "device_select: shortcut EnumeratePhysicalDevice* for count-only calls"
|
||
|
||
Mitchell Kember (1):
|
||
|
||
- [fxbug.dev] Migrate bug numbers
|
||
|
||
Mohamed Ahmed (6):
|
||
|
||
- nvk: Use stride in the explicit modifier case for linear images
|
||
- nil: Expose px_to_B helpers to C
|
||
- nil: Add level_layer_size_B() helper
|
||
- nvk: Add host copy functions
|
||
- nvk: Block off non-2D DRM format modifier images
|
||
- nvk: Enable VK_EXT_host_image_copy
|
||
|
||
Nanley Chery (49):
|
||
|
||
- anv: Drop flush from unused depth workaround
|
||
- intel/blorp: Use WA helpers for depth pipecontrol
|
||
- anv,iris: Use WriteImmediate instead of Z flush for WA
|
||
- anv: Batch MCS and CCS aux-op flushes
|
||
- anv: Add want_hiz_wt_for_image()
|
||
- iris: Add and use want_hiz_wt_for_res
|
||
- intel: Enable more LOD0 HIZ+CCS fast clears
|
||
- intel/blorp: Allow LOD0 fast-clears with HiZ WT
|
||
- intel: Adjust partial depth fast clear checks
|
||
- intel: Use a simpler workaround for HiZ WT fast-clears
|
||
- iris: Invalidate state cache for some depth fast clears
|
||
- intel: Add and use isl_get_sampler_clear_field_offset
|
||
- intel/blorp: Simplify depth clear value updates
|
||
- iris: Move a HIZ_CCS_WT fast-clear flush higher up
|
||
- intel: Move depth clear value writes to drivers
|
||
- intel/isl: Fix packing of SINT formats
|
||
- intel: Don't use HW clear color conversion on gfx11
|
||
- intel/blorp: Don't use clear color conversion on gfx12
|
||
- anv,iris: Skip tex invalidate for clear conversion
|
||
- anv,hasvk: Add and use set_image_clear_color()
|
||
- anv,hasvk: Always use BLORP_BATCH_NO_UPDATE_CLEAR_COLOR
|
||
- iris: Always use BLORP_BATCH_NO_UPDATE_CLEAR_COLOR
|
||
- iris: Skip some fast-clears even on color changes
|
||
- intel: Drop BLORP_BATCH_NO_UPDATE_CLEAR_COLOR
|
||
- anv: Pass the VkClearDepthStencilValue for clears
|
||
- anv: Add and use anv_image_hiz_clear_value()
|
||
- intel: Support any depth fast-clear value on Xe2
|
||
- intel/isl: Always set EnableUnormPathInColorPipe
|
||
- anv,iris: Pack depth pixels into initialized arrays
|
||
- intel/isl: Reduce miptail slot usage to allow CCS
|
||
- intel/isl: Disable 3D Ys/Yf miptails for CCS
|
||
- anv: Change params of anv_can_fast_clear_color_view
|
||
- anv: Move and rename anv_can_fast_clear_color_view
|
||
- anv: Move exec_ccs_op and exec_mcs_op higher up
|
||
- anv: Support fast clears in anv_CmdClearColorImage
|
||
- intel: Avoid no-op calls to anv_image_clear_color
|
||
- anv: Require compression for fast-clears on gfx20+
|
||
- anv: Use image formats when copying to/from buffers
|
||
- intel/blorp: Choose some copy formats independently
|
||
- intel/blorp: Use original surface format for some copies
|
||
- anv: Prevent clear color modifier corruption with views
|
||
- anv: Delete stale comment for BLORP clear color addr
|
||
- anv: Add an array of view formats to anv_image
|
||
- anv: Prepare dmabufs for clear color arrays
|
||
- anv: Move code out of loop in anv_CmdClearColorImage
|
||
- anv: Refactor clear color loading functions
|
||
- anv: Access more colors in fast_clear_memory_range
|
||
- anv: Load fast clear colors more often
|
||
- anv: Allow more fast clear colors for layouts
|
||
|
||
Neha Bhende (1):
|
||
|
||
- dri: fix macro name check to detect svga driver
|
||
|
||
Nikita Popov (1):
|
||
|
||
- gallium: Don't pass avx512er and avx512pf features on LLVM 19
|
||
|
||
Oskar Rundgren (1):
|
||
|
||
- pvr: improve buffer copy format selection
|
||
|
||
PODISHETTY KUMAR (1):
|
||
|
||
- Revert "gfxstream: add VK_EXT_vertex_attribute_divisor"
|
||
|
||
Patel, Utpal (2):
|
||
|
||
- amd/vpelib: Add input pixel format support
|
||
- amd/vpelib: Add resource function hooks for checking support
|
||
|
||
Patrick Lerda (8):
|
||
|
||
- iris: fix indirect draw refcnt imbalance
|
||
- i915: fix vertex atan regression
|
||
- iris: fix iris_ensure_indirect_generation_shader() memory leak
|
||
- r600/sfn: fix class Shader object m_register_allocations memory leak
|
||
- r600/sfn: fix class Shader object last_alu_with_indirect_reg memory leak
|
||
- r600: fix spec ext_packed_depth_stencil getteximage
|
||
- glsl: fix gl_nir_validate_intrastage_interface_blocks() memory leak
|
||
- r600: fix sfn_nir_legalize_image_load_store cubearray behavior
|
||
|
||
Paulo Zanoni (33):
|
||
|
||
- intel/genxml: add the BLT and COMP_CTX0 versions of the TR-TT registers
|
||
- anv/trtt: make genX(init_trtt_context_state) a little more compact
|
||
- anv/trtt: check the return value of anv_trtt_init_context_state()
|
||
- anv/trtt: submit a separate batch in anv_trtt_init_context_state()
|
||
- anv/trtt: make all contexts have the same TR-TT programming
|
||
- anv/trtt: fix the process of picking device->trtt.queue
|
||
- anv/trtt: don't just crash when we can't find device->trtt.queue
|
||
- anv/xe: try harder when the vm_bind ioctl fails
|
||
- anv: don't expose the compressed memory types when DEBUG_NO_CCS
|
||
- anv: disable CCS for Source2 games on Xe2
|
||
- intel: fix compute SLM sizes on Xe2 and newer
|
||
- anv: be consistent regarding non-render engines on i915.ko
|
||
- isl: don't assert(num_elements > (1ull << 27))
|
||
- anv: fix compute engines when using ANV_QUEUE_OVERRIDE
|
||
- anv/trtt: mark vk_sync_get_value()'s value as defined for Valgrind
|
||
- anv: remove duplicate pipe_control workaround
|
||
- anv: remove another copy of the texture cache pipe_control workaround
|
||
- anv/trtt: set every entry to NULL when we create an L2 table
|
||
- anv/trtt: extract anv_trtt_first_bind_init()
|
||
- anv/trtt: inline anv_trtt_init_queues_state()
|
||
- anv/trtt: ensure all L3 entries are NULL-bound during init
|
||
- anv/trtt: unset trtt->l3_addr if initialization fails
|
||
- anv/trtt: fix error handling when adding binds
|
||
- anv/trtt: don't submit empty batches when there are no binds to do
|
||
- anv/trtt: decrement trtt->timeline_val when submission fails
|
||
- anv/trtt: delay batch garbage collection
|
||
- anv/trtt: remove TODO comment regarding the reloc list
|
||
- anv/trtt: convert anv_trtt_bind arrays to util_dynarray
|
||
- anv/trtt: extract anv_trtt_first_bind_init_queue()
|
||
- anv/trtt: fix the creation of sparse buffers of size 2^32 on 32bit systems
|
||
- anv/trtt: remove useless VK_RESULT checks
|
||
- brw: don't emit instruction to add zero in spilling code
|
||
- brw: add a NOP in between WHILE instructions on LNL
|
||
|
||
Pavel Ondračka (28):
|
||
|
||
- r300: bias presubtract fix
|
||
- ttn: switch to derivative intrinsics
|
||
- r300: fix RGB10_A2 CONSTANT_COLOR blending
|
||
- mesa: implement GL_FRAMEBUFFER_BLEND query
|
||
- r300/ci: add more flakes
|
||
- ttn: use nir_shader_get_entrypoint in nir_shader_gather_info
|
||
- ttn: unconditionally call nir_shader_gather_info
|
||
- r300: remove nir_opt_load_store_vectorize
|
||
- r300/ci: update fails list with a recent fix
|
||
- r300: move HyperZ blacklist to driconf
|
||
- r300: use fake ADD instead of MOV for negative KIL source fixup
|
||
- r300: preserve NaNs and denormals for MIN/MAX/CMP/CND
|
||
- r300: get rid of unneeded CMP before KIL
|
||
- r300: remove gl_ClipVertex early
|
||
- r300/ci: share some common variables
|
||
- r300/ci: add deqp and piglit on RV410
|
||
- r300: add .clang-format file for the compiler
|
||
- r300: opt in to clang-format CI enforcement for the compiler
|
||
- r300/compiler: reformat using default mesa .clang-format rules
|
||
- r300/ci: update ci expectations after piglit uprev
|
||
- nir/nir_group_loads: reduce chance of max_distance check overflow
|
||
- r300/ci: add new RV410 flakes
|
||
- ci: disable ondracka farm
|
||
- Revert "ci: disable ondracka farm"
|
||
- r300: add switch to support IEEE and FF math opcodes
|
||
- r300: remove wrong Unigine Sanctuary driconf override
|
||
- r300: add driconf math mode override for Unigine Tropics and Oilrush
|
||
- r300/ci: fails update after recent piglit uprev
|
||
|
||
Peyton Lee (1):
|
||
|
||
- radeonsi: correct map and unmap function timing
|
||
|
||
Pierre-Eric Pelloux-Prayer (26):
|
||
|
||
- amd: use a valid size for ac_pm4_state allocation
|
||
- radeonsi: reject modifiers with DCC when NO_EXPORTED_DCC is used
|
||
- radeonsi: consider PIPE_BIND_LINEAR when filtering modifiers
|
||
- radeonsi: consider DBG(NO_TILING) when filtering modifiers
|
||
- egl,gbm,glx: fix log message spam
|
||
- radeonsi: don't always update shader coherency draw call counter
|
||
- gallium/ddebug: implement modifiers function
|
||
- gallium/ddebug: implement set_inlinable_constants
|
||
- gallium/ddebug: fix sampler_states handling
|
||
- gallium: add pipe_screen::get_driver_pipe_screen
|
||
- radeonsi: add si_screen helper
|
||
- kopper, dri: remove trace_screen_unwrap
|
||
- frontends/va: honor DRI_PRIME for VA_DISPLAY_WAYLAND
|
||
- radeonsi/tests: allow to run radeonsi-run-tests.py with AMD_DEBUG
|
||
- radeonsi: include the shader key in the sqtt pipeline hash
|
||
- radeonsi: check the pointer before adding an offset
|
||
- radeonsi/sqtt: don't store the offsets in the pipeline
|
||
- radeonsi/sqtt: use XXH64_update
|
||
- radeonsi/sqtt: don't leak the pipeline bo
|
||
- radeonsi/sqtt: don't store pipeline->bo in shader->bo
|
||
- radeonsi/gfx12: fill missing dcc tiling info
|
||
- radeonsi: fix radeon_canonicalize_bo_flags domain handling
|
||
- ac/surface: fix determination of gfx12_enable_dcc
|
||
- nir: skip offset=0 in nir_io_add_const_offset_to_base
|
||
- glsl: use nir_io_add_const_offset_to_base in gl_nir_opts
|
||
- radeonsi/ci: mark \*.tessellation_shader_tessellation.max_in_out_attributes as fixed
|
||
|
||
Pohsiang (John) Hsu (1):
|
||
|
||
- d3d12: fix incorrect memset in d3d12_video_encoder_references_manager_hevc
|
||
|
||
Qiang Yu (29):
|
||
|
||
- radeonsi: lower subgroup ops after wave size is known
|
||
- ac/llvm: build wqm for quad intrinsics only when fragment shader
|
||
- ac,radv,radeonsi: stop using quad vote any/all when llvm
|
||
- gallium: add caps for KHR_shader_subgroup
|
||
- mesa: add KHR_shader_subgroup extension
|
||
- mesa: implement KHR_shader_subgroup Get* values
|
||
- mesa: add spirv GroupNonUniform* capbilities support
|
||
- glsl: add KHR_shader_subgroup extensions
|
||
- glsl: add builtin variables for KHR_shader_subgroup
|
||
- glsl: remove unused builtin function macros
|
||
- glsl: use generic convertion code for some intrinsics
|
||
- glsl: add KHR_shader_subgroup_basic builtin functions
|
||
- glsl: prepare vote function args for subgroup usage
|
||
- glsl: add KHR_shader_subgroup_vote builtin functions
|
||
- glsl: prepare ballot function args for subgroup usage
|
||
- glsl: add KHR_shader_subgroup_ballot builtin functions
|
||
- glsl: add KHR_shader_subgroup_shuffle builtin functions
|
||
- glsl: add KHR_shader_subgroup_shuffle_relative builtin functions
|
||
- glsl: add KHR_shader_subgroup_arithmetic builtin functions
|
||
- glsl: add KHR_shader_subgroup_clustered builtin functions
|
||
- glsl: add KHR_shader_subgroup_quad builtin functions
|
||
- glsl: remove unused fields in gl_shader
|
||
- glsl: set shader_info.subgroup_size for KHR_shader_subgroup
|
||
- radeonsi: remove NULL check in si_determine_wave_size
|
||
- radeonsi: consider both stages to determine merged shader wave_size
|
||
- radeonsi: use wave64 for KHR_shader_subgroup enabled shader
|
||
- radeonsi: enable KHR_shader_subgroup
|
||
- nir: add skip_lower_packing_ops shader compile option
|
||
- ac: do not lower some ops in nir_lower_packing
|
||
|
||
Rebecca Mckeever (7):
|
||
|
||
- panvk: Move the VkEvent logic to the jm folder and make it per-arch
|
||
- panvk: Add VkEvent support to the CSF backend
|
||
- panvk/csf: Set and clear vb.dirty flag
|
||
- panvk/csf: Split tiler flags initialization into two steps
|
||
- panvk: Track VkFormats for depth and stencil attachments
|
||
- panvk: Move rendering info to panvk_rendering_state struct
|
||
- panvk/csf: Implement vkCmdExecuteCommands
|
||
|
||
Renato Pereyra (1):
|
||
|
||
- egl: Refactor init of CPU tracing in EGL for Android
|
||
|
||
Rhys Perry (88):
|
||
|
||
- nir/instr_set: combine XXH32 calls
|
||
- nir/instr_set: stop sorting phi sources
|
||
- nir/instr_set: hash tex sources commutatively
|
||
- aco/gfx11.5: workaround export priority issue
|
||
- aco: add tests for export priority issue
|
||
- aco/gfx11.5: skip dealloc_vgprs for stages with exports
|
||
- aco: add export instructions to should_form_clause
|
||
- aco: form export clauses
|
||
- aco: move s_setprio to before NGG exec initialization
|
||
- aco: consider exec empty after divergent continue then divergent break
|
||
- aco: add struct and helpers for exec potentially empty
|
||
- aco: combine DPP into v_cvt_f16_f32
|
||
- aco: fix validation of v_s\_ opcodes
|
||
- aco: don't transform v_interp_p2_f32 with constant into fma
|
||
- docs: update ACO_DEBUG documentation for scheduler options
|
||
- docs: update ACO_DEBUG documentation for perfwarn
|
||
- nir/opt_uniform_atomics: require block index metadata
|
||
- aco: split selection_control_remove into rarely_taken and never_taken
|
||
- aco: only remove branch jumping over SMEM/barrier if it's never taken
|
||
- aco: completely skip branches if they're never taken
|
||
- aco: set prefer_remove for gfx9- too
|
||
- aco: calculate indices from dominance tree
|
||
- aco: use dominance helpers
|
||
- aco: preserve SSA in try_eliminate_scc_copy
|
||
- aco: validate temporary reachability
|
||
- aco: split insert_wait_states into two
|
||
- aco: move insert_delay_alu to after insert_NOPs
|
||
- aco: forget valu delays after certain s_waitcnt_depctr/LDSDIR
|
||
- aco: ignore exec and literals when mitigating VALUMaskWriteHazard
|
||
- aco: also consider VALU reads for VALUMaskWriteHazard
|
||
- aco: don't consider sa_sdst=0 before SALU write to fix VALUMaskWriteHazard
|
||
- aco: check SALU writing lanemask later for VALUMaskWriteHazard
|
||
- aco: preserve bitsets after a lane mask is written
|
||
- aco/tests: add more VALUMaskWriteHazard tests
|
||
- aco/ra: fix sub-dword get_reg_specified in some cases
|
||
- aco/ra: use DefInfo for get_reg_specified
|
||
- aco: have get_subdword_definition_info update DefInfo
|
||
- aco: fix printing of d16 MIMG dmask
|
||
- nir/opt_if: fix fighting between split_alu_of_phi and peel_initial_break
|
||
- nir/opt_loop: skip peeling if the break is non-trivial
|
||
- nir/opt_loop: skip peeling if the loop ends with any kind of jump
|
||
- nir/tests: add some loop peeling tests
|
||
- nir: skip opt_loop_peel_initial_break if continue block only has phis
|
||
- radv/rt: align constant data by 64 when inlining shaders
|
||
- radv/rt: don't split array/struct payload variables
|
||
- aco: do not use inline constants for 16-bit pseudo scalar trancendentals
|
||
- aco/tests: update assembler tests for llvm
|
||
- aco: workaround hazards in emit_long_jump
|
||
- nir/opt_loop: rematerialize header block derefs in their use blocks
|
||
- nir/tests: test opt_loop_peel_initial_break with derefs in header block
|
||
- aco: stop using instructions in ra_ctx::vectors
|
||
- aco: create vector affinities for phi operands
|
||
- aco: fix is_vector_intact for GFX11 BVH
|
||
- nir/divergence_analysis: disable phi undef optimization by default
|
||
- aco: ensure phis uniformized by divergence analysis are SGPR
|
||
- aco: skip uniformization of certain merge phis
|
||
- radv: fix output statistic for fragment shaders
|
||
- radv: optimize VS input load components to constants earlier
|
||
- radv: use explicitly sized types for some radv_shader_info members
|
||
- ac/llvm: cast to integer after derivative intrinsics
|
||
- aco,radv,radeonsi: add aco_shader_info::ps::has_prolog
|
||
- aco: split CounterMap off from VGPRCounterMap
|
||
- aco: minor CounterMap::operator== fix
|
||
- aco: workaround VALUReadSGPRHazard
|
||
- aco/tests: add tests for VALUReadSGPRHazard
|
||
- nir: fix shfr constant folding with zero src2
|
||
- nir/algebraic: fix shfr optimization with zero src2
|
||
- aco: add waitcnt build helper
|
||
- aco: fix printing of block_kind_discard_early_exit
|
||
- aco: insert NOP before dealloc_vgpr in the insert_NOPs pass
|
||
- aco: don't emit early exit over dealloc_vgprs
|
||
- aco: wait for scratch stores to complete before dealloc_vgprs
|
||
- aco: don't byte align global VMEM loads if it might be unsafe
|
||
- nir/algebraic: add bit-size check to extract_u8 pattern
|
||
- aco/gfx11: fix v1b=p_extract(src, 0, 16, 0)
|
||
- aco: check for SDWA before applying extract to lshl/cvt_f32
|
||
- aco: disallow p_extract(,,32,)
|
||
- aco: handle SGPR limitations when applying extract
|
||
- aco: add a bit more p_extract/p_insert validation
|
||
- aco: shrink code size of some p_extract
|
||
- aco: allow applying sign-extended sel to p_extract more often
|
||
- aco: add and use apply_extract_twice helper
|
||
- aco: combine extracts with sub-dword definitions
|
||
- aco: apply extract to p_extract_vector
|
||
- aco: apply extract to v_cvt_f32_ubyte0
|
||
- nir/algebraic: fix iabs(ishr(iabs(a), b)) optimization
|
||
- nir/algebraic: check bit sizes in lowered unpack(pack()) optimization
|
||
- nir/lcssa: fix premature exit of loop after rematerializing derefs
|
||
|
||
Rob Clark (80):
|
||
|
||
- freedreno/a6xx: Implement reg stomper support
|
||
- freedreno/a7xx: Fix GRAS_UNKNOWN_80F4 writes
|
||
- freedreno/cffdec: Fix a7xx CP_EVENT_WRITE decoding
|
||
- tu/drm/virtio: Add missing a7xx case
|
||
- freedreno/drm: Handle a7xx case
|
||
- freedreno: Move GENX/CALLX magic to common
|
||
- freedreno: Extract out common UBWC helper
|
||
- freedreno: Extract out shared LRZFC layout helpers
|
||
- freedreno/a6xx: Allocate lrcfc when needed for direction tracking
|
||
- freedreno/a6xx: Refactor CP_EVENT_WRITE emit
|
||
- freedreno/a6xx: Rework CCU_CNTL emit for a7xx
|
||
- freedreno/a6xx: Initial a7xx support
|
||
- tu: Fix issues with 16k (or larger) page sizes
|
||
- freedreno/drm/virtio: Fix issues with 16k (or larger) page sizes
|
||
- tu: Random non-consequential fixes
|
||
- subprojects: perfetto uprev to v47.0
|
||
- egl: Fix surfaceless + modifiers
|
||
- gallium: Add option to not add version to libgallium filename
|
||
- freedreno/computerator: Use CHIP variant reg builders
|
||
- tu: Add helper to calculate layer address
|
||
- tu: Use CHIP variant reg builders
|
||
- freedreno/registers: Deprecate non-variant reg builders
|
||
- tu: Quiet vla warnings
|
||
- freedreno/a6xx: Fix arb_texture_view regression
|
||
- freedreno/decode: Dump register values on LRZ_CLEAR
|
||
- freedreno/a6xx: Fix LRZ
|
||
- freedreno: Re-enable LRZ for a7xx
|
||
- freedreno/a6xx: Cleanup setup_slices()
|
||
- nir/opt_loop: Don't peel initial break if loop ends in break
|
||
- freedreno/a6xx: Sysmem clear fixes
|
||
- freedreno/drm: Fix ring_heap flags
|
||
- freedreno/crashdec: Fix fault address handling
|
||
- freedreno/register: A couple pm4 updates
|
||
- freedreno/ir3: Fix GPU name in disasm test
|
||
- freedreno/ir3: Add half-FLUT cases
|
||
- freedreno/ir3: Fix cat1 parser ambiguity vs FLUTs
|
||
- freedreno/ir3: Add more cat1 float-immed cases
|
||
- freedreno/a6xx: Only emit VFD/PC_POWER_CNTL for a6xx
|
||
- freedreno/a6xx: De-open-code VFD_MODE_CNTL
|
||
- freedreno/a6xx: Cleanup WFIs around RB_CCU_CNTL
|
||
- freedreno/a6xx: Set bin size per bin
|
||
- freedreno/a6xx: Move PC_TESSFACTOR_ADDR emit
|
||
- freedreno/a6xx: Move static regs to preamble IB
|
||
- freedreno/a6xx: Move more state emit per-bin
|
||
- freedreno/a6xx: Emit CP_SET_AMBLE packets
|
||
- freedreno/drm: Add preemption support
|
||
- freedreno: Balance out u_blitter cb0 save/restore
|
||
- freedreno/a6xx: Add some missing a7xx bits
|
||
- freedreno: Assert we aren't writing to 0x0
|
||
- freedreno/decode: Fix UBO decode on a7xx
|
||
- freedreno/ir3: Avoid draw/grid time input iteration
|
||
- freedreno/ir3: Add assert about const emit
|
||
- freedreno/ir3+tu: Convert driver-params to structs
|
||
- freedreno/ir3: Indentation fix
|
||
- freedreno/a6xx: Fix double SP_MODE_CONTROL emit
|
||
- freedreno/a6xx: Move tess-bo emit
|
||
- freedreno/ir3: Track # of app UBOs
|
||
- freedreno/ir3: Fix need_driver_params for UBO case
|
||
- freedreno: Rework indirect compute param emit
|
||
- freedreno/a6xx: Add support to load driver-params via UBO
|
||
- freedreno/a6xx: Fix color_cache_size
|
||
- freedreno/a6xx: Don't open-code INVALID_REG
|
||
- freedreno/a6xx: Add VPC hardware workaround for a750
|
||
- freedreno/a6xx: Add missing GRAS_SU_DEPTH_CNTL
|
||
- freedreno/ir3: Create UBO variables for driver-UBOs
|
||
- nir/lower_amul: Fix ASAN error
|
||
- freedreno/a6xx: Support variable wg size
|
||
- ir3/ra: Better CL/kernel support
|
||
- freedreno/computerator: Make shader show up in devcore/etc
|
||
- freedreno/ir3: Do not propagate away a widening move
|
||
- ir3: Fix binning pass driver UBOs
|
||
- ir3: Don't lower VS driver params to UBO
|
||
- freedreno/a6xx: Don't try resolve blits
|
||
- freedreno/a6xx: Stop exposing MSAA image load/store harder
|
||
- freedreno: Fix tile-per-pipe debug overrides
|
||
- freedreno/a6xx: Fix MSAA depth on a6xx gen1
|
||
- freedreno/ci: Garbage collect some obsolete xfails
|
||
- util/primconvert: Avoid OoB with improbable draws
|
||
- freedreno/a6xx: Don't check dst coords
|
||
- freedreno/a6xx: Random whitespace fix
|
||
|
||
Robert Mader (5):
|
||
|
||
- egl: Stop requiring texture_3D_image for EGL 1.5
|
||
- perfetto: Add Panfrost data sources to system.cfg
|
||
- egl: Add more errors cases during context creation
|
||
- egl: Stop requiring EXT_create_context_robustness for EGL 1.5
|
||
- v3d: Support SAND128 base modifier
|
||
|
||
Rohan Garg (23):
|
||
|
||
- anv: use the WA infrastructure when emitting WA 16013994831
|
||
- anv: program a custom byte stride on Xe2 for indirect draws
|
||
- anv,iris: prefix the argument format with XI for a upcoming refactor
|
||
- anv: refactor indirect draw support into it's own function
|
||
- anv: dispatch indirect draws with a count buffer through the XI hardware on ARL+
|
||
- anv: migrate indirect mesh draws to indirect draws on ARL+
|
||
- vtn: handle SpvOpExtInstWithForwardRefsKHR for non semantic instructions
|
||
- anv: enable KHR_shader_relaxed_extended_instruction
|
||
- anv: prefetch samplers when dispatching compute shaders
|
||
- iris,anv: simplify and inline sampler count calculations
|
||
- anv: enable the VK_KHR_shader_relaxed_extended_instruction feature
|
||
- hasvk: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- intel/compiler: version can never be above 11 due to the previous check
|
||
- intel/compiler: use the correct cache enum for loads and stores
|
||
- hk: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- nvk: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- radv: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- tu: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- v3dv: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- lvp: enable VK_KHR_shader_relaxed_extended_instruction
|
||
- docs: update feature matrix for VK_KHR_shader_relaxed_extended_instruction
|
||
- intel/brw: lower math op regions for Xe2+
|
||
- anv: Xe2+ doesn't need the special flush for sparse
|
||
|
||
Rohit Athavale (1):
|
||
|
||
- d3d12: Fix typo in copyright comments
|
||
|
||
Roland Scheidegger (1):
|
||
|
||
- llvmpipe: Fix type mismatch when storing residency info
|
||
|
||
Romaric Jodin (1):
|
||
|
||
- perfetto: fix perfetto counter following perfetto v46.0
|
||
|
||
Roy Chan (3):
|
||
|
||
- amd/vpelib: Optimize the CPU usage by caching all the LUT configs
|
||
- amd/vpelib: fix zero input handling
|
||
- amd/vpelib: Add documentation
|
||
|
||
Ruijing Dong (8):
|
||
|
||
- radeonsi/vcn: correct a typo in a variable
|
||
- radeonsi/vcn: add new function for obu_header
|
||
- frontends/va: check av1 enc hdr metadata
|
||
- radeonsi/vcn: input av1 hdr metadata
|
||
- frontends/va: rework VAConfigAttribEncPackedHeaders query
|
||
- radeonsi/vcn: add HDR metadata obu in av1enc
|
||
- frontends/va: reset roi number
|
||
- radeonsi/vcn: qp map IB package sent by default
|
||
|
||
Russell Greene (2):
|
||
|
||
- wsi/wayland/perfetto: supply presentation clock to perfetto
|
||
- perfetto: fix macos compile
|
||
|
||
Sagar Ghuge (8):
|
||
|
||
- intel/compiler: Ray query requires write-back register
|
||
- intel/compiler: Adjust trace ray control field on Xe2
|
||
- intel/compiler: Fix indirect offset in GS input read for Xe2+
|
||
- anv: Reduce clear color state alignment to 64B
|
||
- iris: Reduce clear color state alignment to 64B
|
||
- intel: uncached L1 to fix memory barrier issue in RT shader
|
||
- anv: Track all the descriptor sets
|
||
- intel: Switch to COMPUTE_WALKER_BODY
|
||
|
||
Sai Teja (2):
|
||
|
||
- ci: Don't run vulkan jobs for gbm changes
|
||
- ci: Disable angle jobs for GL changes
|
||
|
||
Sai Teja Pottumuttu (2):
|
||
|
||
- docs: Add required dependencies to meson.rst
|
||
- docs: Match README installation guide with meson.rst
|
||
|
||
Sam Lantinga (1):
|
||
|
||
- util: Fixed crash in HEVC encoding on 32-bit systems
|
||
|
||
Samuel Pitoiset (233):
|
||
|
||
- radv: stop re-initializing HTILE for non-compressed to compressed transitions
|
||
- amd/drm-shim: add missing GC_11_5_0 family
|
||
- amd/drm-shim: add GFX1150 support
|
||
- ci: add GFX1150 support to radv-fossils
|
||
- radv: fix handling view index from device index
|
||
- radv: cleanup robustness with vk_pipeline_robustness_state_fill()
|
||
- radv: add a helper to set shader stage key robustness info
|
||
- radv: add support for dynamic vertex input state with DGC
|
||
- radv: allow VK_EXT_vertex_input_dynamic_state with DGC
|
||
- radv/meta: simplify radv_meta_blit2d_normal_dst()
|
||
- radv/meta: rework creating blit pipelines
|
||
- radv/meta: simplify initializing DCC comp-to-single pipelines
|
||
- radv/meta: remove unused parameter from some blit init functions
|
||
- radv/meta: create blit2d layouts on-demand
|
||
- radv/meta: create itob layouts on-demand
|
||
- radv/meta: create btoi layouts on-demand
|
||
- radv/meta: create itoi layouts on-demand
|
||
- radv/meta: create cleari layouts on-demand
|
||
- radv/meta: simplify initializing bufimage pipelines
|
||
- radv/meta: create DGC prepare pipeline on-demand
|
||
- aco: fix bogus assert in RT prolog on GFX11+
|
||
- radv: add a small helper to determine the index type with DGC
|
||
- radv: pass a dgc_cmdbuf parameter to more dgc helpers
|
||
- radv: refactor the DGC helpers to determine cmdbuf size
|
||
- radv: add a small comment describing the DGC implementation
|
||
- radv/winsys: pad gfx and compute IBs with only one NOP
|
||
- radv: pad GFX preambles IBs with only one NOP
|
||
- radv: use a sized NOP packet for the DGC preamble
|
||
- radv/amdgpu: fix CS padding for non-GFX/COMPUTE queues
|
||
- radv: remove redundant PA_SU_PRIM_FILTER_CNTL in the GFX preamble
|
||
- radv: stop emitting DB_RENDER_OVERRIDE in the GFX preamble
|
||
- radv: stop emitting PA_SC_CLIPRECT_RULE in the GFX preamble
|
||
- radv: stop emitting DB_COUNT_CONTROL in the GFX preamble
|
||
- vk/render_pass: add Mesa-specific flag for dynamic rendering
|
||
- radv: use the Mesa-specifc dynamic rendering flag for meta operations
|
||
- radv: add radv_shader_info::ps::uses_fbfetch_output
|
||
- radv: add support for input attachment indices with DRLR
|
||
- radv: fix DRLR with subpass input attachments and feedback loops
|
||
- radv/amdgpu: do not check that a CS is aligned if no padding is added
|
||
- radv/amdgpu: assert that the DGC IB VA is correctly aligned
|
||
- radv: optimize NOPs padding with DGC
|
||
- radv/ci: update flakes lists for NAVI21/VANGOGH
|
||
- radv/ci: enable RADV_PERFTEST=transfer_queue on GFX9+
|
||
- radv: remove RADV_DEBUG=nogsfastlaunch2
|
||
- radv: rework computing the DGC cmdbuf layout
|
||
- ac: add ac_gfx103_get_cu_mask_ps()
|
||
- ac,radeonsi,radv: add common GFX preambles
|
||
- radv: prepare for specialized DGC shaders
|
||
- radv: add a pointer to the DGC layout in dgc_cmdbuf
|
||
- radv: specialize dispatch DGC token
|
||
- radv: specialize draw DGC token
|
||
- radv: specialize index buffer DGC token
|
||
- radv: specialize pipeline DGC token
|
||
- radv: specialize VBO DGC token
|
||
- radv: simplify allocating push constants with DGC
|
||
- radv: specialize push constant DGC token
|
||
- radv: specialize indirect command layout stride for DGC
|
||
- radv: remove useless check about non-indexed draws and DGC
|
||
- radv: move emitting the compute pipeline with DGC
|
||
- radv: stop passing the upload offset to dgc_emit_bind_pipeline()
|
||
- radv: add a helper to store data to the DGC upload space
|
||
- radv: move emitting VBOs with DGC
|
||
- radv: specialize push constant stages with DGC
|
||
- radv: allow VK_EXT_legacy_vertex_attributes with DGC
|
||
- radv: fix emitting DGC indirect draws with drawid/base_instance
|
||
- radv: return early when the vertex input state is dynamic
|
||
- radv: only initialize per-attribute vertex input state when needed
|
||
- radv: add radv_vs_input_state::attrib_index_offset
|
||
- radv: bind the static vertex input state from the pipeline
|
||
- radv: rename radv_vs_input_state to radv_vertex_input_state
|
||
- radv: pass the vertex shader to radv_write_vertex_descriptors()
|
||
- radv: pass more VBO info as parameters to the DGC prepare shader
|
||
- radv: stop storing DGC info as part of the VBO descriptors
|
||
- radv: specialize dispatch initiator for compute/task shaders with DGC
|
||
- radv: store the indirect compute pipeline VA to dgc_cmdbuf
|
||
- radv: use compute pipeline metadata for non-indirect DGC pipelines
|
||
- radv: fix missing attribute offset with dynamic VS inputs and DGC
|
||
- radv: specialize one more VBO parameter with DGC
|
||
- radv: pass use_per_attribute_vb_descs in radv_dgc_params
|
||
- radv: add a helper to write a single vertex descriptor with DGC
|
||
- radv: add a helper to get VBO info for emitting descriptors
|
||
- radv: compute rsrc3 for VBO descriptors in NIR for DGC
|
||
- radv: use radv_vbo_info for emitting descriptors with DGC
|
||
- radv: emit all VBO descriptors in the prepare DGC shader
|
||
- radv: pass vb_desc_usage_mask in radv_dgc_params
|
||
- radv: stop relying on the VS for emitting VBO descriptors in DGC
|
||
- radv: fix allocating sparse descriptor buffers in the 32-bit addr space
|
||
- radv: disable shaders linking with ESO when nextStage of VS/TES isn't present
|
||
- Revert "radv: specialize push constant stages with DGC"
|
||
- vulkan: Update XML and headers to 1.3.294
|
||
- vulkan: skip the disk cache when disableInternalCache is true
|
||
- radv: rework helpers for serialize/deserialize shaders
|
||
- radv: store the SHA1 RT stage to the shaders cache
|
||
- radv: store the number of RT stages per pipeline to the shaders cache
|
||
- radv: store whether a RT pipeline is a library to the shaders cache
|
||
- radv: make pipeline hashing functions non-static
|
||
- radv: disable the in-memory cache when disableInternalCache is true
|
||
- radv: add initial support for pipeline binaries
|
||
- radv: add support for capturing pipeline binaries
|
||
- radv: add support for importing pipeline binaries
|
||
- radv: advertise VK_KHR_pipeline_binary
|
||
- radv: rework emitting indirect compute pipelines with DGC
|
||
- radv: cleanup some functions that emit shader user SGPRs
|
||
- radv: store the indirect descriptor sets VA to the descriptor state
|
||
- radv: rename radv_emit_descriptor_pointers() to radv_emit_descriptors_per_stage()
|
||
- radv: unify emitting non-indirect/indirect descriptor sets
|
||
- radv: make the helper that uploads indirect descriptors non-static
|
||
- radv: stop allocating upload space for indirect descriptors with DGC
|
||
- radv: use radv_upload_indirect_descriptor_sets() in DGC
|
||
- radv: update PGM register for TES+GS compiled separately with ESO
|
||
- radv: precompute more PGM registers for all stages
|
||
- radv: use radv_get_user_sgpr_loc() more
|
||
- radv,radeonsi: remove remaining occurrences of TCS epilog
|
||
- aco,radv,radeonsi: move has_epilog to the fragment shader info
|
||
- radv: merge radv_emit_epilog() with radv_emit_ps_epilog_state()
|
||
- radv/rt: skip shaders cache for pipelines created with the capture/replay flag
|
||
- radv: fix copying the pipeline binary key
|
||
- radv: compute shader hash for shaders created without internal cache
|
||
- radv: enable more properties with VK_KHR_maintenance5
|
||
- radv: fix lowering the view index to an input varying for FS
|
||
- radv,aco: fix legacy vertex attributes when offset >= stride on GFX6-7
|
||
- radv: fix lowering VS inputs when offset >= stride on GFX6-7
|
||
- radv: add missing cache flushes for DRLR feedback loops
|
||
- radv: update the stencil layout for DRLR feedback loops
|
||
- radv/ci: stop skipping few tests that hang only with LLVM
|
||
- radv: stop reporting VKCTS conformance on GFX11+
|
||
- amd/addrlib: remove bogus assert in HwlComputeSlicePipeBankXor()
|
||
- radv/ci: stop skipping dEQP-VK.api.command_buffers.many_indirect_disps_on_secondary
|
||
- zink/ci: update expected list of failures since recent piglit uprev
|
||
- radv: stop emulating GS invocations for legacy GS on RDNA1-2
|
||
- radv: only export KHR_video_maintenance1 with KHR_video_queue
|
||
- radv: fix assigning mesh shader outputs when clip/cull distances are read in FS
|
||
- radv: emit all shader related user SGPR states in one place
|
||
- radv: make sure to re-emit shader query state when a task shader is bound
|
||
- radv: use only one user SGPR for all NGG state
|
||
- radv: rename shader_query_state to task_state
|
||
- radv: introduce dirty flags for shaders state
|
||
- aco: fix descriptor leaking when printing assembly with CLRX
|
||
- radv: move emitting some RT user SGPRs when the RT pipeline is emitted
|
||
- radv: move updating compute scratch for RT when stack size is emitted
|
||
- radv: do not keep executable info when compiling shaders for ESO
|
||
- vulkan: Update XML and headers to 1.3.296
|
||
- radv: simplify determining conformant products
|
||
- ac/surface: add RADEON_SURF_VIEW_3D_AS_2D_ARRAY for GFX9+
|
||
- radv: implement 2D views of 3D images using 2D_ARRAY descriptors on GFX9+
|
||
- radv: advertise sampler2DViewOf3D
|
||
- zink/ci: update expected list of failures for VANGOGH
|
||
- ci: uprev VKCTS to 1.3.9.2
|
||
- radv/amdgpu: do not use a constant value for the IB size in dwords
|
||
- radv/amdgpu: add assertions to check the IB size
|
||
- radv: use base mip level from vk_image_view
|
||
- radv: stop passing redundant parameters to radv_image_view_make_descriptor()
|
||
- radv: remove redundant assertions about image views
|
||
- radv: stop passing image create flags to radv_image_view_init()
|
||
- zink/ci: update expected list of failures on NAVI31
|
||
- radv: do not expose NV DGC extensions on GFX6-7
|
||
- radv: remove GFX6-7 DGC support completely
|
||
- radv: update image view extent for non-compressed views earlier on GFX10+
|
||
- radv: fix image view descriptors for samplers on GFX6-8
|
||
- radv: remove RADV_MAX_DRM_DEVICES
|
||
- radv: squash radv_get_memory_fd() with radv_GetMemoryFdKHR()
|
||
- radv: remove RADV_THREAD_TRACE_TRIGGER completely
|
||
- zink/ci: update list of expected failures for NAVI10
|
||
- vulkan: fix merging bind sparse submits together
|
||
- radv: fix conditional rendering with DGC preprocessing on compute
|
||
- radv: fix generating the global key for pipeline binaries
|
||
- radv: move radv_compact_spi_shader_col_format() to radv_cmd_buffer.c
|
||
- radv: add a helper to bind the color output state
|
||
- radv: add PKT3_INDIRECT_BUFFER_BYTES in the DGC path
|
||
- radv: implement IB chaining for DGC when it's executed on compute
|
||
- radv/amdgpu: simplify cs_execute_ib()
|
||
- radv/amdgpu: remove unused code about external IBs in the submit path
|
||
- radv: rename 'gfx' to 'main' in the DGC path
|
||
- radv: fix returning non-zero captured address without binding
|
||
- radv: use app names instead of exec name for shader based drirc workarounds
|
||
- radv: stop recomputing the viewport xform for guarband/viewport
|
||
- radv: optimize breaking batch when CB_TARGET_MASK change
|
||
- radv: use radv_normalize_blend_factor() more
|
||
- radv do not force-disable hierarchical stencil testing
|
||
- radv: fix emitting DB_RENDER_OVERRIDE on GFX12
|
||
- radv: track more redundant DB related registers
|
||
- radv: regroup and emit all DS related states in the same function
|
||
- ci: uprev vkd3d-proton to 65b81403435576d882d9141ae3eb4a29373fba0e
|
||
- radv: fix enabling/disabling user sample locations
|
||
- ci: uprev vkd3d-proton to 59d6d4b5ed23766e69fe252408a3401d2fd52ce8
|
||
- radv: fix initializing the HTILE buffer on transfer queue
|
||
- radv: do not use MRT counters for images created for db capture&replay
|
||
- radv: capture shader statistics when RGP is enabled
|
||
- radv: fix emitting NGG culling state for ESO
|
||
- radv: simplify determining if a graphics pipeline uses NGG culling
|
||
- radv: do no emit PA_SC_CONSERVATIVE_RASTERIZATION_CNTL in the preamble on GFX12
|
||
- amd: do not emit PA_SU_PRIM_FILTER_CNTL in the common GFX preamble
|
||
- ci: uprev VKCTS to 1.3.10.0
|
||
- amd/descriptors: set fmask_tile_swizzle for TC-compat CMASK images on GFX8
|
||
- radv: set missing FMASK surface counters for MSAA MRTs
|
||
- radv: fix considering NGG culling for depth-only rendering
|
||
- radv: remove useless check about gl_Position as PS inputs for NGGC
|
||
- radv: simplify determining if dual-source blending is enabled
|
||
- Revert "radv: advertise sampler2DViewOf3D"
|
||
- Revert "radv: implement 2D views of 3D images using 2D_ARRAY descriptors on GFX9+"
|
||
- Revert "ac/surface: add RADEON_SURF_VIEW_3D_AS_2D_ARRAY for GFX9+"
|
||
- radv: implement VK_EXT_device_generated_commands
|
||
- radv: advertise VK_EXT_device_generated_commands on GFX8+
|
||
- radv: track more redundant raster related registers
|
||
- radv: regroup and emit all raster related states in the same function
|
||
- ac/spm: do not abort when the SPM BO is too small
|
||
- radv: resize the SPM bo when it's too small
|
||
- radv: fix wrong index in radv_skip_graphics_pipeline_compile()
|
||
- ac/nir: cull triangles/lines when all W positions are zero/NaN
|
||
- radv: fix configuring the memory violation exception for the compute stage
|
||
- docs: add missing RADV_TRAP_HANDLER
|
||
- radv: add an option to configure the trap handler exceptions
|
||
- radv/ci: skip dEQP-VK.api.command_buffers.many_indirect_disps_on_secondary
|
||
- radv: add missing L2 non-coherent image case for mipmaps with DCC/HTILE on GFX11
|
||
- ac: add ac_gpu_info::has_trap_handler_support
|
||
- radv: check for has_trap_handler_support instead of asserting
|
||
- radv: only emit the TBA/TMA registers on GFX8
|
||
- radv: fix dumping the faulty shader detected by the trap handler on GFX9+
|
||
- radv,aco: save SQ_WAVE_GPR_ALLOC from the trap handler
|
||
- radv: add a struct that describes the trap handler layout
|
||
- aco: fix reading registers from the trap handler shader
|
||
- aco: add support for the trap handler shader on GFX9-GFX10.3
|
||
- radv: set missing shader info values for the trap handler
|
||
- radv,aco: dump all SGPRS from the trap handler
|
||
- aco: fix validation for VOP1 instructions without any dest/src
|
||
- aco: clear the current wave exception in the trap handler
|
||
- aco: add support for the trap handler shader on GFX11
|
||
- aco: use inlined constant offsets for storing SGPRs in the trap handler
|
||
- radv: cleanup tools related resources when destroying logical device
|
||
- radv: save the trap handler report in the HOME directory
|
||
- radv: mark some GFX6-7 GPUs as Vulkan 1.3 conformant
|
||
- radv: fix ignoring src stage mask when dst stage mask is BOTTOM_OF_PIPE
|
||
- radv: add a new drirc option to disable DCC for mips and enable it for RDR2
|
||
|
||
Satadru Pramanik (1):
|
||
|
||
- Update lp_bld_misc.cpp to support llvm-19+.
|
||
|
||
Scott Moreau (2):
|
||
|
||
- egl: small refactor to avoid code duplication
|
||
- egl: hook up swap interval bits when using zink on wayland
|
||
|
||
Semenov Herman (Семенов Герман) (2):
|
||
|
||
- radv: fix memleaks in radv_sqtt_reloc_graphics_shaders()
|
||
- radv: fix memleaks in radv_init_shader_upload_queue()
|
||
|
||
Serdar Kocdemir (22):
|
||
|
||
- Reduce heap pressure on getPacketContents
|
||
- Use append to chain memory pointer info
|
||
- Add VK_EXT_debug_utils to host modules for codegen
|
||
- Add VK_EXT_validation_features to host modules
|
||
- Add support for VK_EXT_robustness2
|
||
- Add VK_EXT_metal_objects support
|
||
- Mark VK_MVK_moltenvk extension as supported on vulkan
|
||
- Add VK_EXT_debug_report extension support
|
||
- Remove direct usages of MoltenVK library functions
|
||
- Remove empty ifdef blocks after the codegen
|
||
- Show missing extension's name in the abort message
|
||
- Handle all extensions for structure size calculation
|
||
- Add extensions used in test applications
|
||
- Add error messages for unhandled extensions
|
||
- Add VK_EXT_external_memory_metal definitions
|
||
- gfxstream: Add VkPrivateDataSlot handle type
|
||
- gfxstream: Handle tmp folder explicitly on codegen
|
||
- gfxstream: Use KHR version of the line_rasterization extension
|
||
- Update decoder.py to use try_unbox on destroy calls
|
||
- gfxstream: Check metal extension for external memory
|
||
- gfxstream: Keep VK_EXT_line_rasterization for codegen
|
||
- gfxstream: Allow VK_KHR_line_rasterization
|
||
|
||
Sergi Blanch Torne (20):
|
||
|
||
- ci: disable Collabora's farm due to maintenance
|
||
- Revert "ci: disable Collabora's farm due to maintenance"
|
||
- ci: disable Collabora's farm due to maintenance
|
||
- Revert "ci: disable Collabora's farm due to maintenance"
|
||
- New DUT for Alder Lake
|
||
- New testing jobs anv-adl{,-full}
|
||
- New testing jobs anv-adl-angle{,-full}
|
||
- New testing jobs intel-adl-skqp
|
||
- New testing jobs zink-anv-adl{,-full}
|
||
- Uprev Piglit to e9ab30aeaed97b69868cf4d6d6a3f70f3b53c362
|
||
- ci: Disable gk20a-gles
|
||
- ci: disable Collabora's farm due to maintenance
|
||
- Revert "ci: disable Collabora's farm due to maintenance"
|
||
- ci: disable Collabora's farm due to maintenance
|
||
- ci: shellcheck requires better 'boolean or' syntax
|
||
- ci: run shellcheck when changes on scripts to check
|
||
- Revert "ci: disable Collabora's farm due to maintenance"
|
||
- WIP: Re-enable Comet Lake
|
||
- ci: disable Collabora's farm due to maintenance
|
||
- Nightly full job for a630-gles-asan
|
||
|
||
Sergio Lopez (2):
|
||
|
||
- hk: limit the number of free BOs in each cmd pool
|
||
- hk: allow overriding sysmem with an env var
|
||
|
||
Sergiu (1):
|
||
|
||
- Tweaking YCBCR conversion to return correct VK Format
|
||
|
||
Shih, Jude (1):
|
||
|
||
- amd/vpelib: Update Plane Descriptor Writer
|
||
|
||
Sid Pranjale (3):
|
||
|
||
- glx: add a missing nullptr check
|
||
- vulkan/util: add vk_format_has_float_depth()
|
||
- nvk: implement VK_EXT_depth_clamp_zero_one
|
||
|
||
Sil Vilerino (49):
|
||
|
||
- Revert "d3d12: Video Encode - Remove PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE as not supported" This reverts commit d6bb4ddc638f3ee37fbbe066c631dad80aaeb2d3. Fixes: d6bb4ddc638 ("d3d12: Video Encode - Remove PIPE_VIDEO_PROFILE_MPEG4_AVC_BASELINE as not supported")
|
||
- pipe: Add PIPE_BIND_VIDEO_DECODE_DPB/PIPE_BIND_VIDEO_ENCODE_DPB
|
||
- d3d12: Implement pipe_video_codec.create_dpb_buffer for AOT resources
|
||
- d3d12: Allow passing custom pipe_resource creation template/placed resource to d3d12_video_buffer_create_impl
|
||
- d3d12: Implement pipe_video_codec.create_dpb_buffer for texture array resources
|
||
- frontend/va: VaSyncSurface encoder check for surface feedback
|
||
- d3d12: Implement get_feedback_fence
|
||
- d3d12: Video Encode H264 to use direct DPB from frontend
|
||
- d3d12: Video Encode H264 - Support direct mmco operations
|
||
- d3d12: Video Encode HEVC to use direct DPB from frontend
|
||
- d3d12: Video Encode HEVC - Use VPS information from frontend, specifically for vps_max_dec_pic_buffering_minus1
|
||
- d3d12: Rename d3d12_video_encoder_config_dirty_flag_sequence_info to d3d12_video_encoder_config_dirty_flag_sequence_header
|
||
- pipe: Add pipe_h264_enc_pic_control.temporal_id
|
||
- d3d12: H264, HEVC, AV1 Store multiple rate control states for multiple temporal layers
|
||
- vl/vl_win32_screen_create_from_d3d12_device: Allow winsys to be injected
|
||
- vl/null_sw_winsys: Add extern declaration for C++ usage
|
||
- d3d12: Remove usage of WindowFromDC when building without d3d12 graphics
|
||
- d3d12: Video Encode HEVC - Store L0/L1 active nums at time to encode them in the headers
|
||
- d3d12: Video Encode H264 - Store L0/L1 active nums at time to encode them in the headers
|
||
- d3d12: Video Encode H264 - Workaround for D3D12 validation bug requiring pRefPicMarkingOperationsCommands for IDR frames
|
||
- pipe: Add PIPE_H26X_MAX_REFERENCES
|
||
- d3d12: Remove max_references usage from frontend to reserve barriers array
|
||
- d3d12: Fix d3d12_video_encoder_get_current_max_dpb_capacity to report actual max slots number in DPB
|
||
- d3d12: Use d3d12_video_encoder_get_current_max_dpb_capacity in tex array pool instead of separate constant
|
||
- d3d12: Use pipe_h264_enc_seq_param.max_num_ref_frames instead of pipe_video_codec.max_references in SPS
|
||
- Fix definitions of PIPE_BIND_VIDEO_DECODE/ENCODE_DPB as they conflict with PIPE_BIND_SCANOUT and PIPE_BIND_SHARED
|
||
- d3d12: Fix setting of direct_8x8_inference_flag in the SPS
|
||
- d3d12: Plumb constrained_intra_pred_flag and transform_8x8_mode_flag from pipe_h264_enc_pic_control
|
||
- d3d12: Set HEVC GOP state before calling d3d12_video_encoder_negotiate_requested_features_and_d3d12_driver_caps
|
||
- d3d12: Support for on demand vps, sps, pps, aud headers requests from frontend
|
||
- pipe: Remove renew_headers_on_idr, insert_aud_nalu unused params
|
||
- d3d12: Add HEVC support for on demand AUD NALUS
|
||
- pipe: Add HEVC encode range extension caps, SPS and PPS params
|
||
- d3d12: Add support for HEVC 4:4:4 video encode using AYUV format
|
||
- d3d12: Add support for HEVC range extension header parameters
|
||
- ci: Bump DirectX-Headers and Agility SDK dependencies to 1.614.1
|
||
- d3d12: Support create_dpb_buffer for non-opaque allocations when decoder/encoder supports it
|
||
- pipe: Add temporal_id to pipe_h26x_enc_dpb_entry
|
||
- d3d12: Support writing H264_SEI_SCALABILITY_INFO header on demand
|
||
- d3d12: Support writing H264 temporal scalability prefix slice NAL on demand
|
||
- d3d12: Report H264/HEVC support for 4 max temporal layers
|
||
- d3d12: H264 Encode - Set SPS.gaps_in_frame_num_value_allowed_flag=1 when num_temporal_layers > 1
|
||
- d3d12: d3d12_video_encoder_references_manager_h264 to use FrameDecodingOrderNumber as h264Pic->slice.frame_num
|
||
- d3d12: Plumb H264/HEVC temporal_id from pipe params
|
||
- d3d12: Fix HEVC chroma_format_idc, separate_colour_plane_flag and conf_win_*_offset for HEVC 444
|
||
- d3d12: Wrap d3d12_video_encode_get_hevc_codec_support around #if VIDEO_CODEC_H265ENC
|
||
- d3d12: fix incorrect memset in d3d12_video_encoder_references_manager_hevc::get_current_frame_picture_control_data
|
||
- d3d12: Fix video encode metadata snapshot at frame submission, must include changes done by DPB manager begin_frame method
|
||
- d3d12: Improve video encode intra-refresh logging
|
||
|
||
Simon Perretta (6):
|
||
|
||
- pvr: Add missing feature check in compute cmd stream
|
||
- pvr: Rescope msaa_mode
|
||
- pvr: Add missing include for FALLTHROUGH
|
||
- pvr: Inline xxhash header functions
|
||
- pvr: rename pds unused reg constant
|
||
- pvr: Make Vulkan driver depend only on nir headers, not library
|
||
|
||
SoroushIMG (4):
|
||
|
||
- pvr: fix intermediate size when configuring mrt setup
|
||
- pvr: cleanup 2101010 int format handling
|
||
- pvr: fix msaa setup for resolve TQs
|
||
- pvr: fix texture address offset when base level >0
|
||
|
||
Stéphane Cerveau (2):
|
||
|
||
- anv: inline query for vulkan video encode
|
||
- anv: check that inline query pool is VK_NULL_HANDLE
|
||
|
||
Surafel Assefa (1):
|
||
|
||
- vulkan: MESA_VK_ENABLE_SUBMIT_THREAD=0 disables threaded submit
|
||
|
||
Sushma Venkatesh Reddy (4):
|
||
|
||
- intel/clflush: Utilize clflushopt in intel_invalidate_range
|
||
- intel/brw: Fix DEBUG_OPTIMIZER
|
||
- intel/brw: Move get_nir_def() higher to avoid UNDEF
|
||
- intel/brw: Handle 16-bit sampler return payloads
|
||
|
||
Sviatoslav Peleshko (14):
|
||
|
||
- anv: Assert ray query BO actually exists
|
||
- brw,elk: Fix opening flags on dumping shader binaries
|
||
- anv: Release correct BO in anv_cmd_buffer_set_ray_query_buffer
|
||
- anv: Add full subgroups WA for the shaders with barriers in Breaking Limit
|
||
- brw: Fix mov cmod propagation when there's int signedness mismatch
|
||
- mesa: Reset vbo attributes after flushing them to Current in glPopAttrib
|
||
- anv: Update XeSS workaround executable names for Satisfactory 1.0
|
||
- intel/brw: Don't apply discard_if condition opt if it can change results
|
||
- intel/brw/gfx9: Implement WaClearArfDependenciesBeforeEot
|
||
- intel/elk/chv: Implement WaClearArfDependenciesBeforeEot
|
||
- intel/brw: List all instructions that have BranchCtrl bit
|
||
- intel/brw_asm: Add BranchCtrl support
|
||
- intel/elk: List all instructions that have BranchCtrl bit
|
||
- intel/elk_asm: Add BranchCtrl support
|
||
|
||
Tapani Pälli (34):
|
||
|
||
- intel/dev: update mesa_defs.json from workaround database
|
||
- blorp: support new flag for setting cps dependency
|
||
- anv: fix a cmd_buffer reference in simple shader
|
||
- anv: refactor cmd_buffer_flush_gfx_runtime_state for dirty state
|
||
- anv: implement workaround for Wa_18038825448
|
||
- intel/dev: update mesa_defs.json from workaround database
|
||
- gbm: depend on libdrm indepedent of dri2 setting
|
||
- anv: move setting 3DSTATE_CLIP::MaximumVPIndex from loop
|
||
- anv: check if RT writes are happening for HasWriteableRT
|
||
- anv: set correct miplevel for anv_image_hiz_op
|
||
- anv: change existing ICL workaround to depend on BLEND_STATE
|
||
- iris: use correct enum for aux state on depth fast clear
|
||
- mesa: check for ARB_ES3_compatibility in format checks
|
||
- anv: initialize pixel struct to zero when setting clear color
|
||
- iris: initialize pixel struct to zero when setting clear color
|
||
- glsl: set a version to standalone compiler context
|
||
- iris: fix issues with memory object updates via glBufferSubData
|
||
- blorp: assert that color depth is not 96 for Wa_16021021469
|
||
- anv: force StackIDControl value for Wa_14021821874
|
||
- drirc/anv: force_vk_vendor=-1 for Faaast Penguin
|
||
- anv: use mi_builder in CmdBeginTransformFeedbackEXT
|
||
- intel/genxml: introduce L3 Fabric Flush for gfx12
|
||
- intel/ds: add L3 fabric flush support
|
||
- anv: add plumbing/support for L3 fabric flush
|
||
- iris: add plumbing/support for L3 fabric flush
|
||
- iris: add depth, DC and L3 fabric flush for aux map invalidation
|
||
- anv: add depth, DC and L3 fabric flush for aux map invalidation
|
||
- drirc/anv: force_vk_vendor=-1 for Silent Hill 2
|
||
- intel/dev: update mesa_defs.json from workaround database
|
||
- anv: set StackIDControlOverride_RTGlobals for 2 workarounds
|
||
- mesa: fix DXT1 support with EXT_texture_compression_dxt1
|
||
- iris: implement VF_STATISTICS emit for Wa_16012775297
|
||
- anv: implement VF_STATISTICS emit for Wa_16012775297
|
||
- anv/iris: add note about Wa_18039438632 for RT flush after SBA
|
||
|
||
Tatsuyuki Ishi (2):
|
||
|
||
- util/u_queue: Replace relative time wait hack with u_cnd_monotonic
|
||
- radv: Enable descriptorBufferCaptureReplay.
|
||
|
||
Thomas Debesse (1):
|
||
|
||
- meson: fix linking gallium with xcb-xfixes when using x11
|
||
|
||
Thomas H.P. Andersen (8):
|
||
|
||
- nvk: EXT_post_depth_coverage
|
||
- nak: special case PhiDsts as not uniform
|
||
- docs: update for new extensions in nvk
|
||
- nvk: do not overwrite test cases
|
||
- dirconf: add a common vk_zero_vram
|
||
- nvk: handle driconf for zeroing vram
|
||
- driconf: set vk_zero_vram driconf for Path of Exile
|
||
- driconf: set vk_zero_vram driconf for X4 Foundations
|
||
|
||
Thomas Wagner (4):
|
||
|
||
- llvmpipe: Fix external memory object reference
|
||
- llvmpipe: Unmap DMA buffer to release all resources
|
||
- llvmpipe: properly save llvmpipe_memory_allocation in memory_object
|
||
- llvmpipe: Remove unused struct member
|
||
|
||
Thong Thai (1):
|
||
|
||
- ci: partially emulate cdna devices using lower image opcodes
|
||
|
||
Timothy Arceri (48):
|
||
|
||
- nir: create validate_tex_src_texture_deref() helper
|
||
- nir: add nir_tex_src_{sampler,texture}_deref_intrinsic
|
||
- glsl: make use of new tex src deref intrinsic
|
||
- nir: set disallow_undef_to_nan for legacy ARB asm programs
|
||
- glsl: fix glsl to nir support for lower precision builtins
|
||
- glsl: always copy bindless sampler packing constructors to a temp
|
||
- nir: make static assert more flexible
|
||
- nir/glsl: set deref cast mode during function inlining
|
||
- glsl: drop double support checks in helper
|
||
- glsl: store has implicit conversions bools in gl_shader
|
||
- glsl: apply implicit matching rules when linking
|
||
- glsl: fix function overload resolution at link time
|
||
- ci: bump piglit version
|
||
- nir/tests: make add_loop_terminators more flexible
|
||
- nir/tests: add basic terminator merge test
|
||
- ci: disable gimark trace
|
||
- mesa/glsl: add nir fields to the gl shader structs
|
||
- glsl: add gl_nir_validate_intrastage_interface_blocks()
|
||
- glsl: make gl_nir_validate_intrastage_arrays() more flexible
|
||
- glsl: add support for converting global instructions to NIR
|
||
- util/hash_table: add _mesa_string_hash_table_create() helper
|
||
- nir: add max_ifc_array_access field to vars
|
||
- nir: add nir_function_impl_clone_remap_globals()
|
||
- glsl: move _mesa_glsl_can_implicitly_convert() to linker_util.cpp
|
||
- nir: add function in/outs to variable modes
|
||
- nir: serialize nir_parameter type
|
||
- nir: store variable mode in nir_parameter
|
||
- nir: add implicit_conversion_prohibited field to nir_parameter
|
||
- glsl: add gl_nir_link_function_calls()
|
||
- glsl: add gl_nir_linker_size_arrays()
|
||
- nir: add nir_intrinsic_deref_implicit_array_length intrinsic
|
||
- nir: handle wildcard array deref
|
||
- glsl: add intrastage shader linking helpers for nir linker
|
||
- glsl: add lower_derivatives_without_layout() helper
|
||
- glsl: set blake3 hash in standalone scaffolding
|
||
- glsl: switch to a full nir based linker
|
||
- glsl: remove now unused linker code
|
||
- glsl: fix gl_{Clip,Cull}Distance error messages
|
||
- glsl/mesa: move uniform related shader structs to shader_types.h
|
||
- glsl: move resource_name_updated() to linker_util.cpp
|
||
- glsl/mesa: remove ir_uniform.h
|
||
- glsl: move error and warning helpers to util file
|
||
- glsl: move shader cache lookup call to st
|
||
- glsl: rename link_shaders() -> link_shaders_init()
|
||
- glsl: remove linker.cpp
|
||
- nir/glsl: set cast mode for image during function inlining
|
||
- nir/glsl: set deref cast mode for blocks during function inlining
|
||
- glsl/nir: fix function cloning at link time
|
||
|
||
Timur Kristóf (22):
|
||
|
||
- radv: Add gs/hs_inputs_read field for linked LS and ES.
|
||
- ac/nir/tess: Add tcs_inputs_read to LS output lowering.
|
||
- ac/nir/tess: Don't emit VS outputs that aren't read by TCS.
|
||
- ac/nir/esgs: Add gs_inputs_read to ES output lowering.
|
||
- ac/nir/esgs: Don't emit ES outputs that aren't read by GS.
|
||
- ac/nir: Add helper for I/O location mapping.
|
||
- ac/nir/tess: Map linked LS/HS I/O based on TCS input mask.
|
||
- ac/nir/esgs: Map linked ES/GS I/O based on GS input mask.
|
||
- ac/nir: Remove ac_nir_calc_io_offset function.
|
||
- radv: Stop assigning linked driver locations.
|
||
- nir: Remove unused nir_assign_linked_io_var_locations.
|
||
- ac/nir/tess: Simplify calculation of HS output LDS offset.
|
||
- ac/nir: Shorten the name of ac_nir_calc_io_offset_mapped.
|
||
- ac/nir/tess: Remove no_inputs_in_lds.
|
||
- ac/nir: Move varying cost functions from radeonsi to common code.
|
||
- radv: Add NIR XFB IO intrinsic info to legacy stages not just NGG.
|
||
- radv: Add new linking step and use nir_opt_varyings.
|
||
- radv: Determine linked I/O info after nir_opt_varyings.
|
||
- nir/opt_algebraic: Add various bitfield extract patterns.
|
||
- radv: Flush L2 cache for non-L2-coherent images in EndCommandBuffer.
|
||
- radv: Increase NGG culling PS param limit to 12 on GFX10.
|
||
- radv: Enable NGG culling by default on GFX10.
|
||
|
||
Tom Cherry (1):
|
||
|
||
- Do not pass std::shared_ptr when not transfering ownership
|
||
|
||
Tomeu Vizoso (2):
|
||
|
||
- etnaviv/ml: Fix includes
|
||
- etnaviv/nn: Fix use of etna_core_info
|
||
|
||
Tomson Chang (2):
|
||
|
||
- amd/vpelib: Fix color adjustment change detect hole
|
||
- amd/vpelib: Only update cached adjustment when it is valid
|
||
|
||
Trigger Huang (1):
|
||
|
||
- mesa: Fix AMD performance monitor implementation
|
||
|
||
Utku Iseri (1):
|
||
|
||
- st/mesa: clamp reported max lod bias
|
||
|
||
Valentine Burley (72):
|
||
|
||
- tu/rmv: Fix missing image bind logging for WSI images
|
||
- tu: Implement VK_EXT_device_address_binding_report
|
||
- tu/kgsl: Remove unused variable
|
||
- docs/features: Update features.txt for Turnip
|
||
- lavapipe: Block more formats from getting blit feature flags
|
||
- vulkan/format: Translate two 420_UNORM formats properly
|
||
- tu, fdl: Drop custom mapping for 1-plane YCbCr formats
|
||
- tu: Drop tu_vk_format_to_pipe_format helper function
|
||
- vulkan/wsi: Refactor can_present_on_device
|
||
- tu: Always report that we can present on kgsl
|
||
- freedreno/ci: Update expectations for Adreno 660
|
||
- freedreno/ci: Re-enable Adreno 660 for Vulkan pre-merge jobs
|
||
- android: Extract version from llvm-project instead of hardcoding it
|
||
- tu: Remove redundant bits override from lower_tex_ycbcr
|
||
- ci: Backport two fixes tor the Vulkan CTS
|
||
- freedreno/ci: Update expectations
|
||
- freedreno/ci: Skip tests that might time out during pre-merge jobs
|
||
- freedreno/ci: Move and document last remaining failures
|
||
- freedreno/ci: Re-enable Adreno 630 for Vulkan nightly jobs
|
||
- tu: Expose VK_KHR_workgroup_memory_explicit_layout
|
||
- freedreno/ci: Wire up skips for a630_vk_full
|
||
- nvk: Advertise VK_EXT_legacy_vertex_attributes
|
||
- llvmpipe: Only use udmabuf with libdrm
|
||
- vulkan, radv: Add new common vk_format_get_plane_width/height helpers
|
||
- tu: Use vk_format_get_plane_width/height to get the plane dimensions
|
||
- tu: Use vk_format_get_plane_count for tu6_plane_count
|
||
- util: Remove Vulkan-only formats from get_plane_width/height
|
||
- tu: Fix VK_EXT_extended_dynamic_state3 feature
|
||
- tu: Simplify VK_EXT_sample_locations SampleCounts assignment
|
||
- freedreno/ci: Uprev kernel for a630 runner
|
||
- freedreno/ci: Skip timing out test on a630
|
||
- freedreno/ci: Uprev kernel to 6.11
|
||
- freedreno/ci: Update expectations after Piglit uprev
|
||
- turnip/ci: Remove fixed test from a660 xfails
|
||
- freedreno/ci: Document some a630 EGL flakes
|
||
- ci: Add a new build-only job without libdrm
|
||
- freedreno/devices: Fix A740v3 from Quest 3
|
||
- freedreno/devices: Unify magic_regs for A740 and A32
|
||
- freedreno/devices: Document an alias for FD644
|
||
- egl: Fix missing libdrm check
|
||
- ci: Build zink and llvmpipe in debian-no-libdrm
|
||
- freedreno/ci: Document new flake
|
||
- tu: Streamline setting YCbCr feature bits
|
||
- tu: Be more consistent with using vk_format helpers
|
||
- tu: Use vk_format_get_plane_count instead of special casing
|
||
- tu: Fix conditional check in tu_image_view_init
|
||
- tu: Use vk_format_get_plane_format
|
||
- tu: Use existing helpers in tu_clear_blit
|
||
- tu: Introduce tu_aspects_to_plane helper function
|
||
- tu: Unify zeroing bufferFeatures
|
||
- tu: Set some feature bits earlier
|
||
- tu: Use v2 format feature flags
|
||
- tu: Update comments about the blob driver
|
||
- tu: Add an assert to tu6_plane_index
|
||
- util/format: Add new 10-bit P010 RGB/planar formats
|
||
- util/format: Add new 12-bit P012 RGB/planar formats
|
||
- lavapipe: Remap 10 and 12 bit formats to 16 bit formats
|
||
- nvk: Remap 10 and 12 bit formats to 16 bit formats
|
||
- hk: Remap 10 and 12 bit formats to 16 bit formats
|
||
- vulkan: Fix incorrect bpcs value for padded formats
|
||
- tu: Fix incorrect bpcs value for padded formats
|
||
- vulkan/format: Properly translate 10-bit and 12-bit formats
|
||
- vulkan/format: Update vk_format_from_pipe_format with additional formats
|
||
- ci: Add global ANGLE skips for its waiver
|
||
- freedreno/devices: Inline a690 quirk
|
||
- freedreno/devices: Document common name for a635 speedbins
|
||
- freedreno/devices: Enable 64-bit atomics on a735 and a740v3
|
||
- freedreno/devices: Establish a7xx sub-generations
|
||
- ir3: Don't lower to LCSSA before calling nir_divergence_analysis()
|
||
- intel/ci: Fix Alder Lake's configuration
|
||
- anv/ci: Remove additive blending fails on ADL
|
||
- lavapipe: Block YCbCr formats from getting blit feature flags
|
||
|
||
Vignesh Raman (19):
|
||
|
||
- ci/deqp-runner: unset CC for arm32 cross-compilation
|
||
- ci: uprev deqp-runner from 0.18.0 to 0.20.0
|
||
- ci: enable Mali V10 testing
|
||
- ci: use v6.11-rc5 kernel for Mali V10 testing
|
||
- ci: move mtl-fw.json to .gitlab-ci directory
|
||
- ci/deqp-runner: pass EXTRA_CARGO_ARGS
|
||
- ci: include exit code in test result output
|
||
- ci/lava: handle non-zero exit codes
|
||
- ci/lava: update unit tests
|
||
- ci/bare-metal: handle non-zero exit codes
|
||
- ci: add Gen10 Arm Mali firmware to rootfs
|
||
- ci/image-tags: add DEBIAN_PYUTILS_TAG
|
||
- ci/lava: set exit code in exception case
|
||
- ci: prepare-artifacts-python: copy if src dir exist
|
||
- ci: build KERNEL_IMAGE_BASE in before_script
|
||
- ci: move rules from rules-anchors to workflow
|
||
- ci/crosvm: Use default value for CROSVM_GALLIUM_DRIVER
|
||
- ci: rename FORCE_KERNEL_TAG to EXTERNAL_KERNEL_TAG
|
||
- Revert "ci: disable Collabora's farm due to maintenance"
|
||
|
||
Visan, Tiberiu (2):
|
||
|
||
- amd/vpelib: set the same range for clr adj
|
||
- amd/vpelib: Remove TODO comments and legacy check(#421)
|
||
|
||
Vitaliy Triang3l Kuzmin (2):
|
||
|
||
- r600: Add FMT_1_REVERSED to r600_formats.h
|
||
- r600: Add missing formats to r600_colorformat_endian_swap
|
||
|
||
Vlad Schiller (3):
|
||
|
||
- pvr: Handle VK_STRUCTURE_TYPE_EXPORT_MEMORY_ALLOCATE_INFO
|
||
- pvr: Handle VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO
|
||
- pvr: Minor code refactor
|
||
|
||
Vldly (1):
|
||
|
||
- freedreno: Fix resource tracking on repeated map with discard
|
||
|
||
Víctor Manuel Jáquez Leal (1):
|
||
|
||
- frontends/va: Don't return P010/P016 as surface formats when encoding
|
||
|
||
WANG Xuerui (2):
|
||
|
||
- meson: Force use of LLVM ORCJIT for hosts without MCJIT support
|
||
- meson: Additionally probe -mtls-dialect=desc for TLSDESC support
|
||
|
||
Wang, Kyle (1):
|
||
|
||
- amd/vpelib: Add missing copyrights
|
||
|
||
X512 (2):
|
||
|
||
- egl/haiku: fix double free of BBitmap
|
||
- egl/haiku: fix synchronization problems, add missing header
|
||
|
||
Yahan Zhou (30):
|
||
|
||
- Update vk.xml and code generator
|
||
- Fix testAndroidBaselineProfile2021Support for host swiftshader
|
||
- Add support for descriptor with inline uniform block
|
||
- Implement vkQueueSubmit2 in the same way as vkQueueSubmit
|
||
- Add some vk 1.3 extension strings
|
||
- Create sync device before calling goldfish_sync_queue_work
|
||
- Parse more ext structs in vkAllocateMemory and vkCreateBuffer
|
||
- Disable compatible handle types for semaphore timeline
|
||
- Fix dEQP-VK.wsi.android.maintenance1
|
||
- VkImageCreateInfo should ignore queeu family indices in some situation
|
||
- Use old en/decoding for vkQueueSignalReleaseImage
|
||
- Add VK_EXT_vertex_attribute_divisor support for guest ANGLE
|
||
- Declare AHB blob unsupported in Vulkan
|
||
- Add VK_EXT_fragment_density_map for Chrome
|
||
- Add snasphot support for vk coherent memory
|
||
- Snapshot vk image content in common situation
|
||
- Better track color buffer usage in Vulkan
|
||
- Make it not crash during vk snapshot load
|
||
- Fix corner cases where vk dependency graph breaks
|
||
- Handle dependency by VkMemoryDedicatedAllocateInfo
|
||
- Snapshot VkBuffer
|
||
- Add snapshot for command buffer
|
||
- Fix handle create mismatch for Vulkan snapshot
|
||
- Add snapshot support for vkBindImageMemory2
|
||
- Fix Vulkan snapshot with -guest-angle
|
||
- Revert^2 "Add snapshot support for vkBindImageMemory2"
|
||
- [Vulkan Snapshot] tryLock when an API might create extra handles
|
||
- Signal the fence after color buffer copying
|
||
- Revert "Signal the fence after color buffer copying"
|
||
- [Vulkan Snapshot] Add dependency graphics pipeline -> render pass
|
||
|
||
Yao Zi (1):
|
||
|
||
- panvk: Link with --build-id explicitly
|
||
|
||
YaoBing Xiao (1):
|
||
|
||
- vulkan/x11: use xcb_connection_has_error to check for failue
|
||
|
||
Yinjie Yao (2):
|
||
|
||
- radeonsi/vcn: Rename transform_skip_disabled and remove hardcoded value for VCN5
|
||
- radeonsi/vcn: Add support for QVBR rate control mode for VCN5
|
||
|
||
Yiwei Zhang (8):
|
||
|
||
- venus: only request ring thread prio for TLS ring
|
||
- venus: clarify wsi image ownership
|
||
- anv: improve vma usage for descriptor buffer
|
||
- ci: re-enable venus ci
|
||
- venus: fix a race condition between gem close and gem handle tracking
|
||
- Revert "meson: disallow Venus debug + LTO build via GCC"
|
||
- venus: workaround cacheline overflush issue on Intel JSL
|
||
- venus: avoid over-caching sfb cmds
|
||
|
||
Yunhyeok "Yune" Choi (1):
|
||
|
||
- glx: Getting rid of the double assignment in \`__glXWireToEvent`.
|
||
|
||
Zack Rusin (1):
|
||
|
||
- svga: Redo the way generated files are handled
|
||
|
||
Zan Dobersek (11):
|
||
|
||
- freedreno: fix a7xx perfcntr countables
|
||
- ir3: some 8-bit subgroup intrinsics must execute as 16-bit instructions
|
||
- nir/lower_vars_to_scratch: calculate threshold-limited variable size separately
|
||
- ir3: lower relevant 8-bit ALU ops in nir_lower_bit_size
|
||
- ir3: use fully-functional dp4acc when available
|
||
- ir3: indicate possible dword straddle for any multi-component pvtmem access
|
||
- ir3: improve validation, display for ldp instructions
|
||
- tu: enable shaderInt8 support
|
||
- tu: use instance indices in RD dump filenames
|
||
- freedreno: sanitize device names for config name usage
|
||
- zink: fix bo_export caching
|
||
|
||
Zhang He (1):
|
||
|
||
- iris, crocus: fix a typo and break comment line correctly
|
||
|
||
antonino (10):
|
||
|
||
- panfrost: rename \`PAN_GPU_INDIRECTS`
|
||
- panfrost: refactor draw logic to support native draw indirect
|
||
- panfrost: implement indirect draw for CSF-based GPUs
|
||
- pan/cs: add helpers to emit contiguous csf code blocks
|
||
- pan/cs: add helpers to generate csf loop
|
||
- panfrost: implement multi draw indirect for CSF-based GPUs
|
||
- panfrost: advertise \`PIPE_CAP_MULTI_DRAW_INDIRECT`
|
||
- mesa/main: expose \`EXT_multi_draw_indirect`
|
||
- pan/cs: add support for nop instruction
|
||
- pan/genxml: fix SET_EXCEPTION_HANDLER instruction definition
|
||
|
||
bbhtt (1):
|
||
|
||
- pipe_loader_drm: Fix virtgpu_drm header path
|
||
|
||
cheyang (1):
|
||
|
||
- Android15 support gralloc IMapper5
|
||
|
||
chyyran (1):
|
||
|
||
- spirv_to_dxil: add missing SPIR-V capabilities
|
||
|
||
dawnhan (1):
|
||
|
||
- vulkan/android: Add missing AHB formats
|
||
|
||
itycodes (1):
|
||
|
||
- intel: Fix a typo in intel_device_info.c:has_get_tiling
|
||
|
||
llyyr (1):
|
||
|
||
- vulkan/wsi/wayland: fix suboptimal flag being ignored with explicit sync
|
||
|
||
nyanmisaka (4):
|
||
|
||
- intel/pci_ids: Update ADL device names
|
||
- intel/pci_ids: Update RPL device names
|
||
- intel/pci_ids: Update TGL device names
|
||
- intel/pci_ids: Update DG1 device names
|
||
|
||
quic_lkondred (1):
|
||
|
||
- freedreno: Add support for Adreno 663 GPU
|
||
|
||
sergiuferentz (2):
|
||
|
||
- Adding support for VK_EXT_external_memory_host extension
|
||
- gfxstream: VulkanBatchedDescriptorSetUpdate toggled on caps on Guest
|
||
|
||
yinjiyao (2):
|
||
|
||
- frontends/va: check hevc enc hdr sei
|
||
- radeonsi/vcn: add HDR sei in hevc enc
|