diff --git a/docs/relnotes.rst b/docs/relnotes.rst index 154d2e933d0..deba9327fcd 100644 --- a/docs/relnotes.rst +++ b/docs/relnotes.rst @@ -3,6 +3,7 @@ Release Notes The release notes summarize what's new or changed in each Mesa release. +- :doc:`23.2.0 release notes ` - :doc:`23.1.3 release notes ` - :doc:`23.1.2 release notes ` - :doc:`23.0.4 release notes ` @@ -394,6 +395,7 @@ release notes, or in the `old docs`_. :maxdepth: 1 :hidden: + 23.2.0 23.1.3 23.1.2 23.0.4 diff --git a/docs/relnotes/23.2.0.rst b/docs/relnotes/23.2.0.rst new file mode 100644 index 00000000000..8394843d5be --- /dev/null +++ b/docs/relnotes/23.2.0.rst @@ -0,0 +1,5211 @@ +Mesa 23.2.0 Release Notes / 2023-07-29 +====================================== + +Mesa 23.2.0 is a new development release. People who are concerned +with stability and reliability should stick with a previous release or +wait for Mesa 23.2.1. + +Mesa 23.2.0 implements the OpenGL 4.6 API, but the version reported by +glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) / +glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used. +Some drivers don't support all the features required in OpenGL 4.6. OpenGL +4.6 is **only** available if requested at context creation. +Compatibility contexts may report a lower version depending on each driver. + +Mesa 23.2.0 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. + +SHA256 checksum +--------------- + +:: + + TBD. + + +New features +------------ + +- VK_EXT_attachment_feedback_loop_dynamic_state on RADV + +- extendedDynamicState3SampleLocationsEnable on RADV + +- VK_EXT_dynamic_rendering_unused_attachments on RADV + +- VK_EXT_mesh_shader on lavapipe + +- OpenGL 3.1 on Asahi + +- OpenGL ES 3.0 on Asahi + +- VK_KHR_fragment_shader_barycentric on RADV/GFX10.3+ + +- VK_KHR_ray_tracing_pipeline on RADV/GFX10.3+ + +- VK_EXT_depth_bias_control on RADV + +- VK_EXT_fragment_shader_interlock on RADV/GFX9+ + +- VK_EXT_pipeline_robustness on RADV + + + +Bug fixes +--------- + +- GTF-GL46.gtf21.GL.build.CorrectFull_vert regressed on intel platforms +- radeonsi: Deadlock when creating a new GL context in parallel with linking a shader on another GL context +- robustness2 raygen tests intermittently fail in Intel Mesa CI +- glthread: huge performance regression +- DirectX games do not launch on Intel HD Graphics 4000 (IVB GT2) [bisected] +- [Vega 64] Newer Mesa-git revisions past 283be8ac3b8610a77b28ebe9e44b946b979f0381 crash the system when accessing hardware accelerated apps +- Docs: Imagination driver not have documentation in https://docs.mesa3d.org/ +- Unigine Heaven broken on Navi 21 since https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22846 +- [bisected] amdgpu graphics acceleration causing system crashes on 22f3bcfb5a33 or later +- anv: GPU hangs on MSAA tests with Angle +- AMD OpenGL texture corruption and crashing regression in java app +- The filenumber part of the #line preprocessor directive is ignored for multiline error messages +- r300: channel merging missed case for mad + mov +- radv: incorrect RTE rounding in corner cases +- Confidential issue #4103 +- r600 regression +- clc: compiler_test gets built even if unit tests are disabled via -Dbuild-tests=false +- x11 swapchain fails to check for DRI3 PixmapFromBuffers error +- msys2: build fails with error: conflicting types for '_glapi_add_dispatch' +- [bisected][anv] newly enabled test (dEQP-VK.api.info.format_properties.g8_b8r8_2plane_420_unorm) failing +- deqp gles3.1 gpu hangs on DG2 A380 when running zink +- [amd/drm-shim] build issue on ppc64 +- radeonsi: bogus advertisement for encode/decode support for 10 bit h264 video +- radeon: Basphemous graphical glitch +- aztec ruins gl benchmark slow to compile shaders on intel +- anv: false cacheline flushing or insufficient buffer alignment on at least ADL +- macOS build error timespec +- intel: Borderlands 2 misrendering with ZINK with OpTerminateInvocation instruction +- gallium: Error path in st_create_context_priv leads to segfault +- [Vulkan][TGL] vkCmdCopyQueryPoolResults failed to write buffer after compute shader write with overlap +- r600: GPU hang on The Long Dark on R600/R700 +- Add iris and crocus to features.txt +- r600: Segfault on glxgears and almost every OpenGL applications on RV770 (regression) +- R9 280 - Broken font rendering in Godot Engine (GLES2) - Radeonsi +- radeonsi gcn1 regression +- ANV crashes on init on 32 bit builds +- eglCreateImageKHR should throw a error when called with anything but EGL_NO_CONTEXT +- virgl: Stack overflow in virgl_bind_sampler_states on hosts with more than 32 samplers +- [TGL] regression with r11_g11_b10 formats when running through virgl +- anv: incorrect vkGetPhysicalDeviceImageFormatProperties2KHR success +- r300: reconstruct ARR in shaders from wined3d +- ci: enable pre-merge testing for Zink/RADV +- rusticl: segmentation fault when enabling for llvmpipe and zink with the driver set to zink +- unify load_ubo_dxil and load_ubo_vec4 +- Move \`lower_loads_and_stores_to_dxil` bit-size lowering logic to \`nir_lower_mem_access_bit_sizes` +- Intel drivers fail to link with -Dxmlconfig=disabled +- mesa: Remove dynamic dispatch stubs +- radv: regression UE5.2 nanite d3d12 vertex explosion +- [ANV/DG2] nvpro-samples/vk_raytracing_tutorial_KHR/ray_tracing_reflections crash +- validate_intrinsic_instr: Assertion \`dst' failed. +- anv: fails to build on aarch64 +- radv: crash/freeze/assert with raytracing and Elden Ring 1.09 +- Elden Ring freeze when summoning cooperator with Raytracing +- gc_alloc_size returns unaligned allocations +- Rusticl OpenCL: panicked at 'called \`Option::unwrap()` on a \`None` value' && void SPIRVKernelParser::applyDecoration(uint32_t, const spv_parsed_instruction_t*): Assertion \`c.first == id' failed. +- Using a \`NULL` pointer as \`bitmap` in \`glBitmap` leads to \`GL_OUT_OF_MEMORY` while creating display-list ("glNewList -> glBitmap") +- Bitwise and with constant 31 removed on width argument to BitFieldSExtract, causing incorrect result on RADV ACO +- Rusticl OpenCL: Simple SYCL / DPC++ program hangs indefinitely at rusticl::api::event::wait_for_events() +- radeonsi: Metro Last Light Redux graphical glitches +- radv: VK_KHR_fragment_shader_barycentric support +- freedreno/a6xx: assert(state->view_rsc_seqno[i] == seqno) failed with sway +- radv: Trackmania 2020 crashing on mesa-git +- radv crashes when using vertex format VK_FORMAT_B10G11R11_UFLOAT_PACK32 +- changes in commit e4b6a0a82457b3ef40c5857412e20bc344ff302c leads to GPU hang +- radv,aco: In the game Rise of the Tomb Raider on RDNA 3 GPUs appeared artifacts after commit 290c3d360e5a6f5226c062d6a9267629adb1060e +- CI: Linux CI jobs naming +- docs.mesa3d.org contrast is bad in dark mode +- iris now requires Linux v5.3 +- Wolfenstein II: The New Colossus vsync off crash +- Surfaceless mode ES2.0 number of vertices limitation +- freedreno: firefox crashes on video playback +- radv: unaligned vertex input regression +- rusticl build error: error[E0308]: mismatched types on some archs +- GetInternalformativ with GL_TEXTURE_2D and GL_CLEAR_TEXTURE incrrectly returns GL_INVALID_ENUM +- radeonsi: texturing is broken on R9.270x since eaf98b14220d8cbc186d67a929254acc3e7de41a +- aco: KHR-GL46.shader_image_load_store.basic-allTargets-atomic asserts on Bonaire +- Firefox / VA-API / H.264 decoding artifacts on AMD RX 6600 / Fedora 37 +- AMD/RX 6600 - VA-API video output is corrupted if decoded surfaces are exported by vaExportSurfaceHandle and then quickly returned to ffmpeg/va-api decoder and reused +- Pixel Game Maker MV - Elfin Force +- Anvil - Vulkan CTS tests fail if has_context_isolation set to false +- anv: binding table pool leak or overly cached +- [regression/bisected] Bone wireframes are no longer rendered correctly in Blender with RadeonSi/Vega +- [nine, radeonsi] Texture missing in Heroes of Might and Magic 5 +- [REGRESSION] Crash in \`loader_dri3_wait_gl()` due to \`dri3_front_buffer(draw) == NULL` +- r600: Visual glitches on The Long Dark with the NIR backend +- some piglit tests seg-fault if -Dgles1=disabled is set +- anv: Tom Clancy's Rainbow Six Siege [DX11: Image Corruption(FIXED)/Vulkan: crash on lauch] +- glSpecializeShaderARB works on SPIR-V compiled with shader compilers ca. 2021, but fails for SPIR-V generated with current compilers +- ANV: Vulkan driver regression in clearing Depth/Stencil +- radv: Sample rate shading broken in AC:Valhalla +- FTBFS: gallivm: src/gallium/auxiliary/gallivm/lp_bld_init.c:45:10: fatal error: llvm-c/Transforms/Scalar.h: No such file or directory (Legacy Pass Manager removed in LLVM 17) +- intel: workaround mechanism initialized before device revision (stepping) is available +- radv/rt: crash compiling Unity Enemies Demo RT pipelines +- gfxbench4/5 crashing on android +- mesa 23.0.3 build failure +- gallium-aux msan failure on Debian bookworm +- radv: Battlefield {1,5} hangs on RX 7900 XTX +- radv: graphical artifacts in MSFS running via DXVK on RX7900XT +- vulkancts regressions on bdw +- [BDW] intel/blorp: MCS partial resolve produces unexpected result +- Compile failure v23.0.0 - error: implicit declaration of function +- ci/radv: Stop setting MESA_SPIRV_LOG_LEVEL +- "frontends/va: report min width and min height values if available" broke VA-API tests on amd-raven +- [radeonsi] flickering debug chunk border lines in Minecraft +- nine: Lower alpha test in shader key? or require PIPE_CAP_ALPHA_TEST? +- radv, radeonsi: Rogue Legacy 2 alpha-to-coverage rendering issues +- [r600, TURKS] R600: Unsupported instruction: vec1 32 ssa_1 = intrinsic image_samples (ssa_0) on spec@arb_shader_texture_image_samples@compiler@fs-image-samples.frag (23.1.0-rc4) +- anv: Overwatch 2 hangs GPU with GPL enabled +- Penumbra: Overture ingame enviroment not displaying on Proton version +- Penumbra: Overture ingame enviroment not displaying on Proton version +- nir: 'base' may be used uninitialized +- vulkan/device_select: no way to select between GPUs of the same model due to bugs +- radv: 7900 XTX hair flickering/rendering issues in VaM +- radv: cache crashing +- nouveau: Regression in arb_transform_feedback_overflow_query-basic from multithreading patches +- regression in aco,ac/llvm,radv,radeonsi: handle ps bc optimization in nir for radv +- radeonsi: vaapi: \`width >2880 && width % 64 != 0` results in wrong width in h265 stream +- [regression] iris: unable to use driver as secondary GPU (primary AMD GPU) +- iris: steam doesn't render on dg2 +- [llvm 16+] [microsoft-clc] opencl-c-base.h does not exist +- Vulkancts clipping / tesselation tests trigger gpu hang on DG2 +- Swaped fields in picture in vlc and mythtv if hw accel is on +- radeonsi: glGetGraphicsResetStatusEXT keeps returning GL_INNOCENT_CONTEXT_RESET after a GPU reset +- WGL: Assert assigns dwThreadId variable +- Intel/anv: Modifier problems running gamescope embedded +- R600: drop TGSI code path +- r600_shader.c:193 r600_pipe_shader_create - translation from TGSI failed ! +- nine regression with r600 (bisected) +- [ACO] [RADV] Flickering squares in some areas in The Last of Us Part 1 (with workaround) +- radv: Jedi Fallen Order flickering & blocky plants +- qemu 7.2.0-rc4 with sdl output crashes with assert +- nouveau: NV50 (NVAC) broken in latest master +- [dozen]: [vkcube] force closing on WSL2 +- rusticl failed to build with rust-bindgen 0.65.0 +- nine: strange color or transparency of trees when called DrawIndexedPrimitive? +- Regression, Bisected: glsl: Delete the lower_tess_level pass breaks r600 tesselation +- vkcts-navi21-valve failing often with GCVM_L2_PROTECTION_FAULT_STATUS:0x00X00830 +- ci/radv: Remove vkctx-navi21-llvm-valve job? +- Deep Rock Galactic GPU freeze (AMD, DX11 DXVK Proton) +- radv: Resident Evil 4 Chainsaw Demo GPU hang with Navi 24 +- radv: Gotham Knights GPU hang with Navi 24 +- SPIR-V error "Invalid back or cross-edge in the CFG" +- SPIR-V parsing FAILED: Loop breaks can only break out of the inner most nested loop level +- ci: a618 traces performance broken +- aco: s_load_dword with negative soffset cause GPU hang +- d3d12: Attempting to display a framebuffer through GDI with low bpc produces on-screen corruption +- piglit.spec.ext_image_dma_buf_import.ext_image_dma_buf_import crash shutting down +- overlay layer: unable to launch titles on steam +- radv/zink: spec@ext_texture_integer@multisample-fast-clear gl_ext_texture_integer +- ci: a530-gl with 6.3 kernel +- a530: hangs with newer firmware version on db820c (apq8096) +- tu: debug marker support +- VAAPI: Wrong H.264 playback on RX 6900 XT and RX 6700 XT (all Sienna?) +- radv: possibly not setting state dirty bits correctly +- RADV: VRS attachment not working in specific scenario +- VAAPI/AMD: videos less than 64 pixels in width or height are decoded to black +- d3d12: DirectX doesn't support seperate stencil functions for front and back face + + +Changes +------- + +Adam Jackson (1): + +- egl: Clear EGL_WINDOW_BIT for non-double-buffered EGLConfigs + +Alan Previn (2): + +- drm-uapi: bump headers (except AMD) +- iris: Add GET_PARAM for protected context capability support + +Alejandro Piñeiro (12): + +- v3dv/pipeline: don't prepack up early-z configuration +- v3d: use more an auxiliar devinfo +- v3d: remove v3d_create_texture_shader_state_bo +- v3d: remove v3d_tfu_supports_tex_format +- v3d: remove v3d_get_internal_type_bpp_for_output_format +- broadcom/compiler: return NULL if we fail to register allocate +- v3d: assert if v3d_compile returns NULL +- broadcom/compiler: disable tmu pipelining when needed +- broadcom/compiler: clarify use of QFILE_VPM +- v3dv: refactor copy_image_to_buffer_blit +- v3dv: add a linear images to buffer copy codepath +- v3dv/device: update conformanceVersion + +Aleksey Komarov (2): + +- pan/va: Fix MUX.v2i16 and MUX.v4i8 description +- pan/va: fix typo in IADD_IMM.i32 description + +Alexander von Gluck IV (1): + +- egl/haiku: Fix potential crash if double buffering is disabled + +Alyssa Rosenzweig (283): + +- gallium: Add u_default_get_sample_position +- zink: Use u_default_get_sample_position +- panfrost: Use u_default_get_sample_position +- freedreno: Use u_default_get_sample_position +- d3d12: Use u_default_get_sample_position +- nir: Add more system values for lowering XFB +- pan/bi: Don't set has_fsub +- asahi: Fix disk cache disable with AGX_MESA_DEBUG +- asahi: Minify width/height in create_surface +- asahi: Don't use depth/stencil staging blits +- asahi: Identify XML for barycentric coordinates +- asahi: Track write to separate stencil +- agx: Handle splits of uniforms +- agx: Fix abs/neg propagation into fcmpsel +- agx/lower_zs_emit: Fix progress returning +- agx: Handle linear 2D array textureSize() +- asahi: Explicitly ban MSAA, compression with linear +- asahi: Use 2D array staging resources for cube/3D +- asahi: Compress more texture targets +- agx: Remove bogus assert +- asahi: Use u_default_get_sample_position +- agx: Defeature fsub +- asahi: Use device_load shift for VBO loads +- agx: Fix packing for iadd with shift +- asahi: Rename no colour output to tag write disable +- asahi: Copy resources if needed to shadow +- agx: Don't wait at the end of the shader +- asahi: Bind staging resources as RENDER_TARGET +- agx/lower_address: Add helper to match multiplies +- agx/lower_address: Match multiplies, not only shifts +- agx: Ensure load_frag_coord has the right sizes +- agx: Rework z/s emit +- agx: Validate that collect sources are the same size +- agx: Lower I/O to scalar later +- asahi: Shrink disk cache size of push ranges +- asahi: Bump MAX_PUSH_RANGES to the worst-case +- asahi: Implement transform feedback +- asahi: Fix depth load/store flags +- nir: Add nir_alu_src_as_uint helper +- pan/bi: Use nir_alu_src_as_uint +- agx: Use nir_alu_src_as_uint +- nir: Model AGX-specific multiply-shift-add +- agx: Handle imadshl_agx, imsubshl_agx +- agx: Fix packing of imsub instructions +- agx: Optimize multiplies +- zink: Always set a blend state for shader-db +- ail: Handle larger block sizes +- nir: Allow adding descriptions to ALU opcodes +- nir: Make ALU descriptions machine-readable +- docs: Include ALU opcode descriptions +- nir: Add nir_foreach_phi(_safe) macro +- nir: Use nir_foreach_phi(_safe) +- dxil: Use nir_foreach_phi_safe +- ac/llvm: Use nir_foreach_phi +- nir: Use nir_block_last_phi_instr more +- nir: Add unified atomics +- nir: Add pass to lower atomics to unified +- agx: Use unified atomics +- pan/bi: Use unified atomics +- pan/mdg: Fix icky formatting +- pan/mdg: Use unified atomics +- gallivm: Use unified atomics +- ntt: Use unified atomics +- ac/llvm: Don't handle atomic derefs +- ac/llvm: Use unified atomics +- aco,radv: Use unified atomics +- zink: Use unified atomics +- ir3: Use unified atomics +- nir: Handle unified atomics in simple cases +- nir/lower_task_shader: Handle unified atomics +- nir/lower_io: Handle unified atomics +- nir/lower_ssbo: Handle unified atomics +- nir/opt_uniform_atomics: Handle unified atomics +- nir/validate: Handle unified atomics +- radv: Constify radv_device_supports_etc +- radv: Use common GetPhysicalDeviceFeatures2 +- r600: Use unified atomics +- lvp: Use common GetPhysicalDeviceFeatures2 +- tu: Use common GetPhysicalDeviceFeatures2 +- agx: Lower legacy atomics sooner +- pan/mdg: Lower legacy atomics earlier +- panvk: Lower legacy atomics earlier +- tu: Lower legacy atomics earlier +- v3dv: Lower legacy atomics earlier +- lavapipe: Lower legacy atomics sooner +- glsl/nir: Produce unified atomics +- nir/lower_atomics_to_ssbo: Produce unified atomics +- nir/lower_printf: Produce unified atomic +- mesa/st: Produce unified atomics +- vtn: Produce unified atomics +- intel: Produce unified atomics +- ac: Produce unified atomic +- treewide: Stop lowering legacy atomics +- nir: Drop nir_lower_legacy_atomics +- ntt: Stop handling legacy atomics +- nir: Drop legacy atomics in simple cases +- nir/lower_io: Drop legacy atomics +- nir/lower_task_shader: Drop legacy atomics +- nir/validate: Drop legacy atomics +- nir/opt_load_store_vectorize: Reclaim ATOMIC +- nir/opt_uniform_atomics: Drop legacy atomics +- nir: Remove legacy atomics +- nir: Drop unused name from nir_ssa_dest_init +- nir: Drop unused argument from nir_ssa_dest_init_for_type +- nir: Remove stale TODOs +- nir: Fix incorrect comment +- util: Add common hex dump utility +- asahi: Use common hexdump utility +- pan/decode: Use common hexdump +- CODEOWNERS: Update panfrost +- gallium: Drop Asahi-as-a-swrast hack +- asahi: Drop Asahi-as-a-swrast hack +- nir: Document extra image source +- nir: Add image_texel_address intrinsics +- nir: Add pass to lower image atomics +- pan/bi: Fix atomic exchange on Valhall +- pan/bi: Use nir_lower_image_atomics_to_global +- pan/mdg: Use nir_lower_image_atomics_to_global +- gallium: Add pipe_image_view::single_layer_view +- mesa/st: Set pipe_shader_image::single_layer_view +- dxil: Rely on scoped_barrier +- treewide: Avoid nir_lower_regs_to_ssa calls +- nir/opt_barriers: Add a default callback +- agx: Use common combine_all_barriers callback +- nir: Drop stale comments +- zink: Switch to scoped barriers +- panfrost/ci: Skip Piglit tests known to crash +- panfrost/ci: Skip hanging test +- nir: Add intrinsics for multisampling on AGX +- nir/builder: Add nir_replicate helper +- treewide: Use nir_replicate +- pan/lower_framebuffer: Use nir_replicate +- radv/query: Use nir_trim_vector +- intel/blorp: Use nir_trim_vector +- nir/print: Print locations for geometry shader inputs +- gallium: Add util_image_to_sampler_view helper +- panfrost: Use util_pipe_image_to_sampler_view +- nir: Add and use nir_tex_src_ssa +- treewide: Use nir_tex_src_for_ssa +- treewide: Use nir_trim_vector more +- agx: Set support_16bit_alu +- agx: Constant fold when optimizing int64 +- agx: Use textures_used, not num_textures +- asahi: Add passes to lower MSAA +- asahi: Add passes to lower sample intrinsics +- asahi: Add alpha-to-coverage (and alpha-to-one) lowering +- agx: Assert that sample shading is lowered +- asahi: Set uses_sample_shading for background program +- asahi: Plumb API sample mask into shaders +- asahi: Plumb ppp_multisamplectl into shaders +- agx: Model both sources of sample_mask +- agx: Plumb in nir_intrinsic_load_sample_mask_in +- agx: Handle sample_mask_agx +- agx: Enable tag writes when sample mask written +- agx: Lower discard in NIR +- asahi,agx: Call lower_discard_zs_emit in the driver +- agx: Split iter and iterproj instructions +- agx: Model interpolation for iter instructions +- agx: Handle centroid and sample interpolation +- asahi: Lower MSAA +- asahi: Use nonempty tib for MSAA +- agx: Emit shader info late +- asahi: Advertise GL 3.1 +- agx: Stop bit-inexact conversion propagation +- asahi: Add ASAHI_MESA_DEBUG=nowc flag +- asahi: Extract transition_resource helper +- asahi: Decompress writable images +- asahi: Decompress with format reinterpretation +- asahi: Remove stale comments +- pan/mdg: Drop lower_locals_to_regs call +- lima: Drop lower_locals_to_regs call +- ir2: Drop lower_locals_to_regs call +- nir: Add AGX atomic intrinsics +- agx: Refactor expressions in agx_nir_lower_address +- agx: Fold addressing math into atomics +- nir/builder: Add steal_tex_src helper +- nir/lower_tex: Use nir_steal_tex_src +- agx: Use common nir_steal_tex_src +- nir: Add interleave_agx instruction +- vtn: Handle atomic counter semantics +- ir3: Drop reference to unsupported intrinsic +- ttn: Emit scoped barriers when needed +- ntt: Use scoped barriers +- ac/llvm: Drop memory_barrier_buffer impl +- glsl: Assume use_scoped_barrier +- vtn: Assume use_scoped_barrier +- nir: Assume use_scoped_barrier +- ttn: Assume use_scoped_barrier +- treewide: Remove use_scoped_barrier +- nir/tests: Use scoped barriers internally +- nir: Remove handling for non-scoped barriers +- radeonsi: Scan for scoped barriers +- nir: Remove non-scoped barriers +- iris: Don't use STREAMING_LOAD without SSE +- nir/builder: Add ubitfield_extract_imm helper +- agx: Implement bitfieldExtract natively +- asahi: Use bitfield_extract for texture lowering +- nir: Remove integer and 64-bit modifiers +- aco: Drop NIR parallel copy handling +- nir: Add discard_agx intrinsic +- agx: Update explanation of sample_mask behaviour +- agx: Fix discards +- agx: Extract coordinate register size calculation +- agx: Recollect stored vectors at their use +- agx: Add loop header? flag +- agx: Validate predecessor information +- agx/lower_parallel_copy: Lower 64-bit copies +- agx: Implement vector live range splitting +- nir/lower_bool_to_int32: Fix progress reporting +- nir/lower_locals_to_regs: Add bool bitsize knob +- gallivm: Use NIR_PASS macros +- nir: Add pixel_coord, frag_coord_zw intrinsics +- nir: Add lower_frag_coord_to_pixel_coord pass +- pan/bi: Use lower_frag_coord_to_pixel_coord +- agx: Use nir_lower_frag_coord_to_pixel_coord +- asahi: Use txf for background program +- nir/lower_blend: Optimize masked out RTs +- nir: Add nir_builder_create returning nir_builder +- nir: Use nir_builder_create +- treewide: Use nir_builder_create more +- treewide: Remove unused builders +- nir: Add nir_foreach_function_impl helper +- nir: Convert to nir_foreach_function_impl +- nir/validate: Assert txf(_ms) matches dimension +- nir: Add nir_lower_robust_access pass +- broadcom/compiler: Use nir_lower_robust_access +- broadcom/compiler: Remove v3d_nir_lower_robust_access +- broadcom/compiler: Remove unused #define +- broadcom/compiler: Use nir_steal_tex_src +- nir: Add b32fcsel_mdg opcode for Midgard +- pan/mdg: Optimize b32csel(inot) in NIR +- pan/mdg: Type CSEL with a NIR pass +- pan/mdg: Lower isub in common code +- pan/mdg: Constant fold after algebraic_late +- pan/mdg: Add is_ssa helper +- pan/mdg: Fix IR from scheduling conditions +- pan/mdg: Fix 2-const CSEL at block beginning +- pan/mdg: Fix temp count calculation +- pan/mdg: Lower special reads better +- pan/mdg: Reset predicate.exclude while scheduling +- pan/mdg: Copy-prop even with swizzle restrictions +- pan/mdg: Propagate modifiers in the backend +- nir: Rename load/store_reg -> load/store_register +- nir: Rename nir_reg_{src,dest} -> nir_register_{src,dest} +- agx: Add algebraic opt to help with discard lowering +- agx: Smarten discard_agx -> sample_mask lowering +- asahi: Strip ? in GenXML +- asahi: Rename 'Render Target' to 'PBE' +- asahi: Identify PBE::sRGB flag +- asahi: Remove ; in perf_debug_ctx +- agx: Use nir_opt_shrink_stores +- agx: Use nir_opt_shrink_vectors +- agx: Assert that barriers are not used in the preamble +- asahi: Assert we don't transition shared resources +- asahi: Fix scissor_culls_everything check +- asahi: Use ralloc harder +- asahi: Take ownership of compute shader NIR +- agx: Don't leak ssa_to_reg_out +- asahi: Use txf_ms for MSAA background programs +- nir: Fix breaking in nir_foreach_phi(_safe) +- vulkan: Add vk_index_type_to_bytes helper +- lavapipe: Use vk_index_type_to_bytes +- v3dv: Use vk_index_type_to_bytes +- rogue: Remove commented convert_from_ssa call +- nir: Add intrinsics for register access +- nir: Add helpers for walking register uses +- nir: Add pass for trivializing register access +- nir: Add legacy data structures & helpers +- nir: Add new version of lower_regs_to_ssa +- nir: Produce intrinsics in lower_{phis,ssa_defs}_to_regs +- nir: Add intrinsics version of locals_to_regs +- nir: Add lower_vec_to_regs pass +- gallium: Return SSA values from TTN ALU helpers +- gallium: Convert TTN to register intrinsics +- mesa: Simplify ptn_log() a bit +- mesa: Return SSA defs from PTN ALU helpers +- mesa: Convert PTN to register intrinsics +- nir/lower_shader_calls: Convert to register intrinsics +- nir: Remove nir_lower_regs_to_ssa +- nir: Remove nir_register-based unit tests +- gallivm: Switch to reg intrinsics +- pan/mdg: Ingest new-style registers +- panfrost: Fix transform feedback on v9 + +Amber (3): + +- turnip: fix buffer markers using wrong addresses +- ir3, freedreno: implement GL_ARB_shader_draw_parameters +- freedreno: implement GL_ARB_indirect_parameters + +Andres Calderon Jaramillo (1): + +- r600: Report multi-plane formats as unsupported + +Andres Gomez (3): + +- .mailmap: add an alias for Miguel Casas-Sanchez +- .mailmap: add an alias for Clayton Craft +- .mailmap: add an alias for Christian Gmeiner + +André Almeida (2): + +- radv: debug: Update decode ring umr command +- radv: Search for guilty contexts at radv_check_status + +Antonio Gomes (3): + +- rusticl: Move nir compilation to Program +- rusticl: Drop some Kernel data and have a NirKernelBuild ref instead +- rusticl: Drop Program::kernel_count + +Asahi Lina (33): + +- asahi: Identify ZS resolve bits (tentative) +- asahi: Broadcast Z for all components on texture fetch +- asahi: Enable 2xMSAA (for deqp) +- asahi: Add batch state debugging +- asahi: Fix batch writer tracking for null batches +- asahi: Clear batch->resolve on agx_batch_init +- asahi: Assert that freed BOs have no pending writers +- asahi: Fix batch writer_syncobj cleanup +- asahi: Implement memory_barrier +- asahi: Implement create_fence_fd and fence_server_sync +- asahi: Make framebuffer texture barriers a no-op +- asahi: Disable tilebuffer write masking optimization +- asahi: Add missing stdbool include to lib/hexdump.h +- asahi: Fix check for sprite coord mode in agx_bind_rasterizer_state +- asahi: Add some more system registers +- asahi: Partially identify some missing index list stuff +- asahi: Lazily initialize batch state on first draw +- asahi: Make bo->writer_syncobj atomic +- ail: Implement multisampling for compression meta calculation +- asahi: Use ail_can_compress() in agx_compression_allowed() +- ail: Add MSAA tests +- asahi: Use os_dupfd_cloexec() instead of dup() +- asahi: Fix memory leak in agx_nir_lower_sysvals() +- asahi: Do not leak meta shader NIR +- asahi: Revert "Advertise ARB_texture_barrier" +- asahi: Disable PIPE_CAP_SURFACE_SAMPLE_COUNT +- asahi: Pass through surface sample count +- asahi: match_soa: Treat offsets as signed +- asahi: Identify the separate varying count fields +- asahi: Gather flat/linear shaded input info from uncompiled FS +- asahi: Fix type confusion for fragment shader keys +- asahi: Add flat/linear shaded varyings mask to the VS shader key +- asahi: Arrange VS varyings in the correct order + +Axel Davy (17): + +- frontend/nine: Fix missing clamping of pointsize for ff +- frontend/nine: Apply writemask to pointsize +- frontend/nine: fix fog key overflow +- frontend/nine: fix wfog +- frontend/nine: Fix num_textures count +- frontend/nine: Drop max_ps_const_f +- frontend/nine: Implement alpha test backup support +- frontend/nine: Implement backup support for pointsize +- frontend/nine: Improve VS_WINDOW_SPACE_POSITION fallback +- frontend/nine: Print warning incomplete position_t support +- frontend/nine: Enforce legacy pow behaviour +- frontend/nine: Get rid of INTERPOLATE_COLOR +- frontend/nine: initialize force_color_in_centroid +- docs/gallium: Clarify PIPE_CAP_CLIP_PLANES +- frontend/nine: Implement backup support for clip planes +- frontend/nine: Fix shader cap test for POSITIONT +- frontend/nine: Add debug driconf var force_features_emulation + +Bas Nieuwenhuizen (8): + +- radv: Reserve space for indirect descriptor set address writes. +- radv: Reserve space in the ACE pre/postambles. +- radv: Add stricter space checks. +- radv: Add asserts in radeon_emit{,_array}. +- radv: Move all the dirty flags from TES binding to TCS binding. +- amd/drm-shim: Add vangogh entry. +- amd/drm-shim: Add raphael&mendocino, polaris12 and gfx1100. +- amd/drm-shim: Update docs for more devices. + +Benjamin Lee (1): + +- intel: Fix stack overflow in intel_dump_gpu + +Billy Laws (1): + +- wgl: Fix depth/stencil image support when using zink kopper + +Blisto (1): + +- driconf: set vk_x11_strict_image_count for Wolfenstein II + +Boris Brezillon (4): + +- panfrost: Check blend enabled state in pan_allow_forward_pixel_to_kill() +- renderonly: Fix potential NULL deref in the error path +- renderonly: Make sure we reset scanout on error in create_kms_dumb_buffer_for_resource() +- winsys/panfrost: Make sure we reset scanout on error in create_kms_dumb_buffer_for_resource() + +Boyuan Zhang (2): + +- frontends/va: add default intra idr period +- radeonsi: disable H264HIGH10 profile + +Brian Paul (5): + +- llvmpipe: remove lp_setup_alloc_triangle()'s unneeded tri_size param +- llvmpipe: code clean-ups in llvmpipe_get_query_result_resource() +- lavapipe: clean-ups in lvp_GetQueryPoolResults() +- lavapipe: clean-ups in lvp_physical_device_get_format_properties() +- lavapipe: asst. clean-ups in lvp_execute.c + +Caio Oliveira (54): + +- spirv/tests: Add test for single-block loop +- spirv: Output spirv2nir tool result to stdout +- spirv: Add --optimize flag to spirv2nir tool +- spirv: Rework structured control flow handling +- spirv: Do more on spirv2nir --optimize +- spirv: Use NIR_PASS for spirv2nir --optimize +- spirv: Extract vtn_handle_debug_text() helper +- spirv: Fix gl_spirv_validation when OpLine with strings is present +- spirv: Improve the 'ID is the wrong kind of value' error messages +- mesa/spirv: Provide more specific error message for glSpecializeShader() +- spirv: Validate Dim of OpTypeSampledImage and OpSampledImage +- spirv: Assert sampler_dim is valid when building nir_tex_instr +- nir/print: Print 0 when mem_modes or resource_intel have no values +- nir/print: Do not print raw values +- spirv: Add workaround for OpImageQueryLevels with Multi-sampled images +- compiler/types: Make key in subroutine_name more effective +- r600/sfn: Fix warning about overloads hiding virtual functions +- spirv: Refactor and rename scope translation helper +- spirv: Use vtn_translate_scope for OpReadClockKHR +- intel/compiler: Refactor dump_instruction(s) +- intel/compiler: Remove unused functions and declarations +- compiler/types: Be consistent when naming array element/size +- compiler/types: Tidy up the asserts in get_*_instance functions +- compiler/types: Use hash table pre-hashed functions for type caching +- microsoft/clc: Add unreachable() to fix 'may be unitialized' warning +- compiler: Move from nir_scope to mesa_scope +- compiler: Add mesa_scope_name() function +- nir/print: Use mesa_scope_name() function to print scopes +- intel/compiler: Move brw_kernel.c to the intel_clc target +- compiler/clc: Rename the internal library from libclc to libmesaclc +- compiler/clc: Move related NIR passes to the common mesa clc +- compiler: Move spirv into a module of its own +- nir/print: Print whether the shader is internal or not +- intel/compiler: Respect NIR_DEBUG_PRINT_INTERNAL flag +- meson: Explicitly add "check : false" to a couple instances of run_command +- vulkan: Add NV suffix to VK_NV_cooperative_matrix feature names +- vulkan: Update XML and headers to 1.3.255 +- nir: Allow nir_gather_ssa_types() to ignore regs instead of assert +- nir/print: Improve NIR_PRINT=print_consts by using nir_gather_ssa_types() +- nir/print: Make NIR_DEBUG=print_consts behavior the default +- nir: Make a const-friendly way to get the offset_src and arrayed_io_src from intrinsic +- nir: Extract logic to get dest and srcs types from intrinsic +- nir/print: Use src_type when printing consts in SSA uses +- nir/print: Print more representations in load_const +- nir/print: Use symbols % for SSA and @ for intrinsic +- nir/print: Use \`bN` instead of \`block_N` for identifying basic blocks +- nir/print: Use BITSIZExELEMENTS for SSA sizes +- nir/print: Align instructions around \`=` +- nir/print: Rename print_tabs() to print_indentation() and use it more +- nir/print: Don't use comment syntax for deref_cast properties +- nir/print: Use \`//` for comments +- nir/print: Use 4-space indentation +- nir/print: Print div/con annotation first +- nir/print: Reformat the preds/succs block information + +Caleb Cornett (3): + +- d3d12: Fix Xbox GDK build errors +- wgl: Add BITMAPV5HEADER to stw_gdishim.h +- d3d12: Fix Xbox frame scheduling for interval != 1 + +Charmaine Lee (6): + +- translate: do not clamp element index in generic_run +- svga: set PIPE_CAP_VERTEX_ATTRIB_ELEMENT_ALIGNED_ONLY for VGPU10 device +- mesa/main: fix distance attenuation calculation in ffvertex +- svga: fix shader type after ntt +- svga: fix compute shader type after ntt +- svga: lower images before ntt + +Chia-I Wu (23): + +- drm-shim: apply file overrides for open +- amd/drm-shim: add amdgpu drm-shim +- hasvk: Refactor Android externalFormat handling in CreateYcbcrConversion +- hasvk/android: Use VkFormat for externalFormat +- hasvk: Use the common vk_ycbcr_conversion object +- vulkan: make sure vk_image::format is never UNDEFINED +- vulkan: make sure vk_image_view::format is never UNDEFINED +- vulkan: rename vk_image::ahardware_buffer_format +- vulkan: define inline stubs when android api level < 26 +- vulkan: add vk_ahb_format_to_image_format +- anv,hasvk,radv: do not fall back to AHARDWAREBUFFER_FORMAT_BLOB +- vulkan: add vk_image_format_to_ahb_format +- anv,hasvk: android ahb is not always exportable +- radv: improve externalMemoryFeatures for android ahb +- amd/drm-shim: add raven2 +- ac/surface: print tile_swizzle as well +- radv: do not use a pipe offset for aliased images +- aco: fix alignment check in emit_load +- ac, radeonsi: add and use ac_get_ps_iter_mask +- radv: fix gl_SampleMaskIn for sample shading +- radv: fix msaa feedback loop without tc-compat cmask +- radv: fix non-square compressed image copy on gfx9 +- radv: disable calibrated timestamps on raven/raven2 + +Christian Gmeiner (31): + +- etnaviv: Add util_blitter_save_so_targets(..) call +- etnaviv: nir: improve uniform usage for ALU opc +- etnaviv: correct number of instructions in dump_shader_info(..) +- etnaviv: move printing of final shader out of etna_link_shaders(..) +- etnaviv: nir: do not call nir_lower_idiv(..) unconditionally +- etnaviv: make wider use of DBG_ENABLED(..) +- ci: add debian-arm32-asan +- ci/etnaviv: add asan run +- etnaviv: Add support for conditional rendering +- etnaviv: add support for performance warnings +- mesa/arbprog: fix compile errors +- etnaviv: remove tgsi remains +- etnaviv: drop usage of tgsi_swizzle_names +- etnaviv: remove not used tgsi includes +- ci/etnaviv: update ci expectation +- ir3/analyze_ubo_ranges: Move IR3_DBG_NOUBOOPT check +- etnaviv: nir: call nir_remove_dead_variables(..) before linking setup +- etnaviv: linker: add fallback lookup to VARYING_SLOT_BFC[n] +- nir: add helper to clear all pass_flags +- nir/lower_amul: make use nir_shader_clear_pass_flags(..) +- etnaviv: make use nir_shader_clear_pass_flags(..) +- etnaviv: nir: do a late nir_opt_cse run +- docs: mark OES_texture_half_float done on etnaviv +- etnaviv: support OES_texture_half_float_linear +- ci/etnaviv: update ci expectation +- docs: update etnaviv extensions +- etnaviv: linker: handle scenario where there are FS inputs without matching VS output +- etnaviv: linker: clean up etna_link_shader(..) +- nir: rename intrinsic to have a more generic nameing +- nir: rename has_txs to has_texture_scaling +- nir/lower_tex: optimize offset lowering for has_texture_scaling + +Christopher Snowhill (2): + +- Corrects log print to produce hexadecimal base output +- intel: Sync xe_drm.h + +Collabora's Gfx CI Team (4): + +- Uprev Piglit to 79a084c56b6dd79f7c3a97b57a72963121ebb1e6 +- Uprev Piglit to 536975d94a40cf76a69fcfa786c2513eccd0c989 https://gitlab.freedesktop.org/mesa/piglit/-/compare/79a084c56b6dd79f7c3a97b57a72963121ebb1e6...536975d94a40cf76a69fcfa786c2513eccd0c989 +- Uprev Piglit to d8c08d123fadb986e9a8a7887b922ff63fcff52e https://gitlab.freedesktop.org/mesa/piglit/-/compare/536975d94a40cf76a69fcfa786c2513eccd0c989...d8c08d123fadb986e9a8a7887b922ff63fcff52e +- Uprev Piglit to 5036601c43fff63f7be5cd8ad7b319a5c1f6652c + +Connor Abbott (38): + +- tu: Don't override depth for GMEM +- tu: Don't pre-shift depth and stencil pitch +- freedreno/fdl: Don't pre-shift image view pitch +- freedreno/fdl: Expose view offset +- tu: Add 3D GMEM load path +- tu: Use dirty bit for scissor state +- tu: Precompute maximum views across all subpasses +- tu: Merge RB_DEPTH_CNTL and RB_STENCIL_CONTROL drawstates +- tu: Make dynamic viewport and scissor count more accurate +- freedreno/a6xx: Document per-view viewport in GRAS_SU_CNTL +- tu: Parse fragment density map attachment info +- tu: Implement sampling the fragment density map +- tu/cs: Add support for CS patching +- tu: Add core FDM patchpoint infrastructure +- ir3: Record whether a shader writes gl_ViewportIndex +- tu: Implement FDM viewport patching +- tu: Implement FDM scaled loads/stores +- nir, ir3: Add option to use unscaled FragCoord for input attachments +- tu, ir3: Handle FDM shader builtins +- tu/autotune: Always prefer GMEM with fragment density maps +- tu: Don't allow importing/exporting subsampled images with modifiers +- tu: Expose VK_EXT_fragment_density_map +- util/bitset: Add some extra functions +- vk/graphics_state: Remove vk_subpass_info +- vk/graphics_state: Add feedback_loop_input_only +- vk/graphics_state: Add VI_BINDINGS_VALID state +- vk/graphics_state: Fix some assertions when copying state +- vk/graphics_state: Add helpers for pre-baking state +- radv: Fix radv_pipeline_is_blend_enabled +- vk/graphics_state: Track attachment count as state +- vulkan: Fix renderpass flags with driver-specific renderpass +- vk/graphics_state: Don't track each vertex input field +- tu: Don't use A6XX_PC_PRIMITIVE_CNTL_0::TESS_UPPER_LEFT_DOMAIN_ORIGIN +- freedreno/a6xx: Fix name of A6XX_PC_PRIMITIVE_CNTL_0::TESS_UPPER_LEFT_DOMAIN_ORIGIN +- tu: Split pipeline struct into different types +- tu: Rewrite to use common Vulkan dynamic state +- tu: Use common dirty tracking for PC_PRIMITIVE_CNTL_0 +- freedreno/regs: Document a7xx CP_FIXED_STRIDE_DRAW_TABLE + +Constantine Shablia (3): + +- anv: move get_features after get_device_extensions (ugly diff) +- panvk: use common vkGetPhysicalDeviceFeatures2 +- v3dv: use common vkGetPhysicalDeviceFeatures2 + +Constantine Shablya (7): + +- vulkan: add common implementation of vkGetPhysicalDeviceFeatures2 +- vulkan: introduce supported_features parameter to vk_physical_device_init +- anv: switch to using the common vkGetPhysicalDeviceFeatures2 +- vulkan: inline vk_get_physical_device_features into vk_common_GetPhysicalDeviceFeatures2 +- vulkan: put interesting code before boring code +- vulkan: put TEMPLATE_H before TEMPLATE_C +- vulkan: rename vk_physical_device_features.py to vk_physical_device_features_gen.py + +Corentin Noël (16): + +- ci: Uprev crosvm and virglrenderer +- nir: Propagate the type sampler type change to the used variable. +- build-crosvm: Use the pkg-config crate 0.3.27 +- util: Use the gcc_struct attribute for packed structures in mingw +- ci: Bump base tag to rebuild piglit +- ci: uprev virglrenderer and crosvm +- gallium: Incorporate the device release in dri_destroy_screen_helper +- gallium: Rename dri_destroy_screen_helper into dri_release_screen +- pipe-loader: Document the behavior regarding screen creating failures +- pipe-loader: Do not destroy the winsys on screen creation failure +- gallium: Only call dri_init_options when the screen is actually created +- gallium: Use the common destroy function on screen initialization failure +- gallium: Rename dri_init_screen_helper into dri_init_screen +- compiler: Allow the explicit_stride of aoa types to be zero +- nir/split_64bit_vec3_and_vec4: Use the right number of components +- ci: Uprev virglrenderer + +Daniel Schürmann (59): + +- radv/rt: fix total stack size computation +- radv/rt: properly destroy radv_ray_tracing_lib_pipeline on error +- radv/rt: rename radv_ray_tracing_module -> radv_ray_tracing_group +- radv/rt: add shader stage indices to radv_ray_tracing_group +- radv/rt: replace uses of pGroups with radv_ray_tracing_group +- radv/rt: remove merged VkRayTracingShaderGroupCreateInfoKHR +- vulkan/pipeline_cache: replace raw data objects on cache insertion of real objects +- vulkan/pipeline_cache: use vk_pipeline_cache_insert_object() to replace raw data objects +- radv: add padding to radv_shader_binary_legacy +- vulkan/pipeline_cache: expose vk_raw_data_cache_object +- radv/pipeline_cache: add NIR caching capabilities +- radv/rt: expose radv_parse_rt_stage() +- radv/rt: introduce struct radv_ray_tracing_stage +- radv/rt: retain parsed NIR shaders in radv_ray_tracing_lib_pipeline +- radv/rt: use precompiled stages to create RT shader +- radv/rt: refactor compute_rt_stack_size() to use radv_ray_tracing_stage information +- radv/rt: remove merged VkPipelineShaderStageCreateInfo +- radv/rt: Fix and improve VkPipelineCreationFeedback +- radv/rt: change base of radv_ray_tracing_lib_pipeline to radv_compute_pipeline +- radv/rt: unify radv_ray_tracing_lib_pipeline and radv_ray_tracing_pipeline +- radv/rt: unify radv_rt_pipeline_create() and radv_rt_pipeline_library_create() +- radv/rt: refactor radv_rt_pipeline_compile() +- radv/rt: use vk_multialloc for radv_ray_tracing_pipeline +- radv/rt: store stack_sizes per stage instead of per group +- vulkan/pipeline_cache: don't log warnings for internal caches +- vulkan/pipeline_cache: don't log warnings for client-invisible caches +- radv: add remaining RT shader args for separate compilation +- nir,amd: add nir_intrinsic_store_[scalar|vector]_arg_amd to overwrite inputs +- nir: add nir_intrinsic_resume_shader_address_amd +- aco: implement nir_intrinsic_load_resume_shader_address_amd +- aco: implement select_program_rt() +- radv/rt: adjust shared_size when lowering hit_attribs +- radv/rt: extend radv_pipeline_group_handle with shader VAs +- radv/shader_info: add RT stages to radv_get_user_data_0() +- radv/rt: implement radv_nir_lower_rt_abi to lower RT shaders for separate compilation +- radv/rt: implement radv_rt_nir_to_asm() +- radv/rt: change RT main shader to MESA_SHADER_INTERSECTION +- radv/rt: replace pCreateInfo with VkPipelineCreateFlags in rt_variables +- radv/rt: pass radv_ray_tracing_pipeline to RT shader creation +- radv/rt: add and use specialized cache search/insert functions +- radv/rt: reference library shaders during radv_rt_fill_stage_info() +- radv/rt: don't write cache hit feedback per stage. +- radv/rt: create compile_rt_prolog() function +- radv/rt: set up RT shader args for separate compilation +- radv/rt: adjust lower_rt_instructions() for shader functions [disables RT] +- aco: adjust RT prolog for shader functions [disables RT] +- radv/rt: separate shader compilation +- radv/debug: dump ray tracing shaders in case of a hang +- radv/rt: use priorities to select the next shader +- radv/rt: remove now dead code +- radv: reference pipeline cache object in radv_pipeline +- aco/assembler: align resume shaders with cache lines +- aco/assembler: align loops if it reduces the number of cache lines +- aco/assembler: change prefetch mode on GFX10.3+ during loops if beneficial +- vulkan/pipeline_cache: add 'skip_disk_cache' option +- radv/meta: disable disk cache for meta shaders +- radv: migrate radv_shader hash to BLAKE3 +- amd: move end-of-code marker padding to ACO. +- amd: Do shader binary alignment for prefetch at memory allocation time. + +Daniel Stone (9): + +- wsi/wayland: Support VK_KHR_present_wait +- ci/zink: Disable Freedoom trace on ANV +- ci: Respect $HTTP_PROXY for ci_run_n_monitor +- ci: Elaborate causes for job retries +- ci: Don't retry manual or scheduled jobs +- ci: Extend a618_vk_full runtime +- CI: Re-enable freedreno CI +- ci/fdno: Pause a660 testing +- Revert "ci/fdno: Pause a660 testing" + +Danylo Piliaiev (38): + +- freedreno: Early exit in device matching if id doesn't have chip_id +- ir3/a7xx: NOPs may have some no-op bits set +- ir3/a7xx: Add new lock/unlock CS instructions +- ir3/a7xx: Add new form of stg.a/ldg.a addressing +- ir3/a7xx: Add STSC definition +- ir3: Document that stc has higher DST upper bound than we defined +- ir3/a7xx: Document "alias" instruction +- ir3: documents (ss) flag for cat7 instructions +- tu: Create drm fd per logical device +- tu: Move VMA heap to the logical device +- tu: Re-enable bufferDeviceAddressCaptureReplay +- freedreno/perfcntrs: Link with libfreedreno_common +- freedreno: Decouple GPU gen from gpu_id/chip_id +- freedreno,ir3: Don't call fd_dev_64b more than necessary +- freedreno/decode: Correctly handle chip_id +- tu: Add missing dbg reg stomping to tu_CmdBeginRendering +- tu: Fix zombie VMAs array not initialized when first BOs may be freed +- freedreno/regs: Print xml validation error if validation fails +- freedreno/rnn: Fix addvariant being set effectively once +- freedreno/rnn: Make addvariant work for fields in the same reg +- freedreno/rnn: Take into account array's variant for regs +- freedreno/regs: Change a7xx regs to have open range for generation +- freedreno/regs: More CP commands are the same on a7xx as on a6xx +- freedreno/regs: Document CP_MEM_TO_SCRATCH_MEM +- freedreno/regs: Document a7xx CP_MODIFY_TIMESTAMP +- freedreno/regs: Clarify polling on a7xx for CP_WAIT_REG_MEM/CP_COND_WRITE5 +- freedreno/regs: Add a7xx pseudo-regs to CP_SET_PSEUDO_REG +- freedreno/regs: a7xx has a new source type CP_REG_TEST +- freedreno/regs: Add 2 new a7xx modes to CP_COND_REG_EXEC +- freedreno/regs: Add some new a7xx events +- freedreno/regs: Add more a7xx regs and reg fields +- freedreno/regs: Fix a7xx SP_FS_PREFETCH definition +- freedreno/regs: Generate per-gen reg usage tables +- freedreno/regs: Define usage for all a6xx/a7xx regs +- tu: Allow reg stomping of compute related registers +- tu: Use reg usage tables for stale reg dbg option +- freedreno/regs: Properly document a7xx CP_EVENT_WRITE, CP_WAIT_TIMESTAMP +- freedreno/regs: Document a7xx CP_BV_BR_COUNT_OPS + +Dave Airlie (127): + +- radeonsi/ac: move some vcn defines to common +- radv/video: add missing gfx family +- radv: set a video decode ip block in physical device. +- radv/winsys: handle encoder queue padding/submits. +- radv/video: add a video addr gfx mode +- radv/video: fix dpb surface programming +- radv/video: start adding gfx11 vcn decoder +- lp_jit: use pipe max for the lp_jit texture levels. +- gallivm: consolidate draw/lp texture type. +- gallivm: consolidate llvmpipe/draw sampler types. +- gallium: consolidate jit image types between draw/llvmpipe +- gallivm: reorder some texture/image members. +- vulkan/cmd_queue: handle beta extensions. +- vulkan: write beta extensions into generator scripts. +- draw: align common members in jit context structs. +- llvmpipe: refactor fs/cs jit structure members. +- gallivm: refactor common resources out of contexts +- gallivm/draw/llvmpipe: consolidate the sampler/image dynamic state fns +- gallivm: add common code for sample/image tracking. +- llvmpipe: move to common sampler/image binding code +- draw: move to use common sampler/image binding code +- llvmpipe/cs: refactor cs generator args to use an enum +- gallivm/draw: refactor vertex header jit type out +- llvmpipe: convert a bunch of shader_type ifs to switches. +- llvmpipe/cs: start making variant generator less compute specific +- llvmpipe/cs: support passing a csctx instead of using implicit one +- lavapipe: add lavapipe specific shader stages define. +- lvp: explictly skip compute shader stage. +- gallivm: fix whitespace in get_deref_offset +- gallivm/nir: refactor the local invocation index calc. +- lvp: use stage mask +- lvp: use stage iterator macros instead of explicit loops +- ci: reenable lavapipe +- radv/video: add missing space checks for video. +- radv/video: use correct h264 levels +- radv/video: fix h264/265 dpb usage. +- radv/video: add missing offset to the dpb binding. +- radv/video: rework stream handle generation. +- radv/video: fix some whitespace. +- radv/video: add debug flag to enable dpb image array on newer GPUs. +- radv/video: fix physical device format property count. +- vk/video: add a common function to get block alignments for profiles +- radv: align video images internal width/height inside the driver. +- anv/video: move format properties to outarray. +- radv/meta: fix uninitialised stack memory usage. +- gallium: add task/mesh shader query types to stats interface. +- gallium: expand pipe_grid_info to handle task/mesh. +- gallium: add a new PIPE_SHADER_MESH_TYPES +- freedreno: don't report task/mesh. +- gallium: add task/mesh shader entrypoints in context +- iris: don't return shader params for task/mesh. +- crocus: don't report mesh/task limits +- radeonsi: don't report shader params for task/mesh +- svga: don't report mesh/task shader limits +- d3d12: don't report mesh/task limits +- gallium/cso: add task/mesh shaders to the cso cache +- gallium/nir/tgsi: add various support for task/mesh bits +- lavapipe: when in doubt, swizzle the swizzle +- lavapipe: fix pipeline sanitizing. +- lavapipe: fix indentation whitespace +- draw: add mesh shader infrastructure +- draw: move draw_vertex_info and draw_prim_info to public header. +- draw: add a mesh primitive assembler. +- draw: add mesh pipeline middle end. +- draw: add support for per primitive aos emission +- gallivm: add support for payload access +- gallivm/nir: add launch mesh workgroups +- gallivm/nir: add a mesh interface and vert/prim count setting. +- gallivm/nir: call task shader lowering. +- gallivm/nir: add support for mesh shader outputs. +- llvmpipe: resize arrays to handle mesh shaders. +- llvmpipe: start adding task/mesh support. +- llvmpipe: bump dirty tracker to 64-bits. +- llvmpipe: add dirty bits for mesh and task shaders. +- llvmpipe: add debug bit for mesh shaders +- llvmpipe: add query support for task/mesh shaders +- llvmpipe: bind task/mesh resources and dirty bits +- gallivm/cs: add payload ptr to the cs thread data. +- llvmpipe/cs: add task/mesh shader support to compute shader builder. +- llvmpipe/cs: add multiple stride indirect to fill_grid_info. +- llvmpipe: add mesh shader drawing. +- llvmpipe: enable task/mesh shader support. +- lavapipe: handle some mesh shader stage differences. +- lavapipe: add mesh query support +- lavapipe: add support for task/mesh shader stages in various places +- lavapipe: add execution backends for mesh shader draw apis +- lavapipe: enable task/mesh shaders. +- docs: update docs for lavapipe mesh shading +- llvmpipe: emit fences for barrier. +- lavapipe: don't remove queue family barriers. +- gallivm/nir: fix shuffleup tests. +- draw: rename jit to vs_jit in lots of places. +- draw/tess: drop unused tgsi bits. +- gallium/tgsi/draw/softpipe: remodel shader const/buffer bindings. +- draw: refactor resources to use arrays instead of explicit structs. +- draw: add a max stage define and use it in a few places +- draw: repack some members of context. +- radv/video: convert video format properties to an outarray +- radv/video: convert session memory requirements to outarray. +- radv/video: don't supply an 8-bit format for a 10-bit dpb. +- radv/video: rework h265 reference frame bindings. +- radv/video: fix hevc st rps programming +- radv/video: fix hevc scaling lists. +- lavapipe: ignore another yuv format. +- radv/video: report bad profile operation if h264 profile isn't supported. +- radv/video: fix hevc scaling list order. +- radv/video: program frame number correctly. +- radv/video: program hevc max dec pic buffering correctly +- radv/video: restrict the number of IBs on video related queues. +- ac/radeonsi: add av1 defaults header file from radeonsi +- radv/video: drop incorrect defines for uapi ones. +- lavapipe: check sampler pointer before deref +- draw/gs: handle extra shader outputs in geometry. +- lavapipe: expose subgroups in mesh/task shaders. +- gallivm: store thread id in separate values. +- gallivm: convert block_id to discrete values. +- gallivm: convert grid_size to discrete values. +- gallivm: make block_size use discrete values. +- clc: llvm 17 requires opaque pointers. +- gallium/va: fix superres av1 decoding. +- llvmpipe/linear: don't allow linear path for shader output with location frac +- llvmpipe/linear: refactor linear samplers into templated code. +- llvmpipe/linear/tgsi: calculate num_texs properly for nir. +- llvmpipe/linear: add sample routines for swapping r/b channels +- llvmpipe/linear: add support for sampling when cbuf order is different. +- llvmpipe/linear: add support for rgba color buffers. +- ci: update fails for fixed tests due to llvmpipe linear changes. + +David (Ming Qiang) Wu (1): + +- radeonsi/vcn: add an exception of field case for h264 decoding + +David Heidelberg (125): + +- ci/amd: 4/5 runners TPad-C13 runners are online, restore most of the tests +- ci/dxvk: uprev to 2.1 +- ci/amd: update checksums after DXVK 2.1 update +- ci: bump kernel to the 6.3, support HDK 888 based on sm8350 +- ci/freedreno: do not restrict to 2 cpus on a530 +- ci: drop overriding new a530 firmware due to preemption issues with older kernel +- ci/freedreno: a530 behaves stable in 6.3 +- ci/freedreno: update a530 flakes, fails and skips +- ci/freedreno: fix the a530_piglit job and switch to Weston +- ci: polish deqp-runner a bit +- ci: uninstall libdrm from the GL and VK containers +- ci: do not retry on forks to get the upstream kernel and rootfs +- ci/mold: bump to 1.11.0 +- ci: add Adreno 660 on sm8350 chipset (HDK 888) +- ci/lava: implement fastboot support +- ci/lava: add support for HDK 888 firmware +- ci: add a660 firmware into rootfs +- pvr: drop unused variable +- ci/dzn: add flaking test +- ci/skqp: replace license with SPDX and extract the used branch +- ci/skqp: update to the Android CTS 12.1_r5 version +- mesa/main: drop unused variable +- nir/lower_io_to_vector: initialize base +- panvk: clear dangling pointers +- ci: uprev kernel to 6.3.1 with fixed patch for Adreno SMMU +- util/tests: adjust for new gtest +- gtest: Update to 1.13.0 +- ci/skqp: handle all warnings printed with clang >= 14 +- panvk: drop path from panvk_physical_device struct +- venus: drop unused sem_feedback_count from vn_queue_bind_sparse_submit_batch +- ci/broadcom: skip timeouting ssbo.layout.3_level_array.std430.mat4 on RPi4 +- ci/venus: add recent flakes +- ci/freedreno: add recent a630 flake +- ci/v3d: add flaking opengl 1.1@depthstencil-default_fb-drawpixels-float-and-ushort +- ci/amd: re-enable VA-API testing +- ci/rules: radeonsi VAAPI rules should include also VA-API targets +- ci: update libva to 2.18.1 +- ci/gtest: improve the runner script +- ci/amd: update VA-API expectations +- ci/amd: add radeonsi-raven-va-full job to cover all VA-API tests +- ci/gtest-runner: fix results reporting +- ci/venus: add missing flakes +- ci/crosvm: update cmdline options +- docs: update crosvm networking options +- ci/radv: add another raven flake dEQP-VK.draw.dynamic_rendering.primary_cmd_buff.linear_interpolation +- ci/v3dv: add often timeouting ssbo.layout.3_level_array.std140.column_major_mat4 +- r300: workaround GCC 12+ warning, declare NULL value as unreachable +- docs: use meson instead invoking ninja directly +- ci/freedreno: disable 3 jobs to match our farm 3 devices down +- ci/freedreno: rename piglit job to respresent the real testing it does +- ci: move from pkg-config to pkgconf +- ci: use meson setup and meson install instead of meson and invoking ninja directly +- ci: bump libdrm from 2.4.110 to 2.4.114 present in Debian 12 +- ci: install stock android-libext4-utils (available in 12, bookworm) +- ci: bump gfxreconstruct revision up to compatible version with Debian 12 +- ci: libwayland from 1.18 to 1.21 and wayland protocols from 1.24 to 1.31 +- ci: VVL uprev (temporary until new release will be published) +- ci: bump from Debian 11 (bullseye) to 12 (bookworm) +- ci/apitrace: install win64 apitrace only on x86_64 +- ci/crosvm: install libelogind0 and sysvinit-core for poweroff functionality +- ci: add clang-15 and clang++-15 wrapper script +- ci/skqp: skqp can't live with compiler named clang-15, provide symlink +- ci: drop gallium-aux test on msan builds, renable freedreno +- ci/mingw: disable as it's broken +- ci/venus: add fail after CI uprev to the Debian 12 +- ci/virpipe: add flakes introduced with CI uprev to Debian 12 +- ci/zink: disable flaking anv traces +- ci: enable shellcheck on whole .gitlab-ci +- ci: disable bogus GCC warning with -Warray-bounds +- ci: do not fail when SHA1 impl. produce stringop-overreads warning +- ci/lavapipe: document subgroups.shuffle.compute.subgroupshuffleup_double_constant crash +- ci/lavapipe: zink failures +- ci/llvmpipe: document intel_shader_atomic_float_minmax@execution@ssbo-atomic* +- bin/ci: mention requirements.txt +- gitlab: add template for merge requests +- ci/zink: add KHR-GL46.limits.max_fragment_interpolation_offset flake +- ci/amd: previously missed raven flake +- ci/panfrost: add largest possible eglcreatepbuffersurface and then glclear flake +- gitlab: prefill MR template with first multiline commit message +- ci: bump Alpine to 3.18 +- ci/ccache: recent ccache changed a output a bit, adapt script +- ci: rename x86 and amd64 to x86_64, armhf to arm32, and i386 to x86_32 +- ci: use bash arrays in Fedora script + shebang change +- ci/fedora: re-enable ccache +- traces: update sir-f720 trace expectations for zink on anv and freedreno +- ci: missed variable inside the big rename and split ARCH and DEBIAN_ARCH +- ci: fix KVM module modprobe code +- ci: explicitely state BUILDTYPE +- ci: rename S3 artifacts according to scheme mesa-$arch-$config-$buildtype +- ci: rename MINIO to S3 +- ci: rename MINIO_HOST variable to S3_HOST +- ci: replace MINIO_RESULTS_UPLOAD with S3_RESULTS_UPLOAD +- ci: remove BUILD_PATH, always use S3_ARTIFACT_NAME +- ci/lava: rename rest local MINIO\_ variables to S3\_ +- ci/android: remove the artifact file just as we unpack it +- ci: valve and freedreno farm is down +- ci/windows: move microsoft farm rules +- ci/etnaviv: if farm is down, we expect no manual jobs can be triggered +- ci/amd: hide vaapi job dependent on Collabora farm when it's down +- ci/crocus: depend on state of the Anholt farm +- ci: implement farms handling trough files inside .ci-farms +- ci/docs: fixup incorrect spacing around console block +- ci/panfrost: switch panfrost-g52-piglit-gles2 from X to XWayland +- ci/fastboot: use gzipped Image to avoid compressing on the runner +- ci/microsoft: uploading artifacts gets stuck currently (retried) +- ci/microsoft: rename manual rules according to rest introduced rules +- ci: create manual farm rules +- ci/traces: guard DXVK and VK behind VK_DRIVER +- ci/apitrace: include version with LTO enabled +- ci/traces: print version of apps used for replaying traces +- ci: when touching farms, never run manual jobs +- ci/microsoft: partly revert rename from container-rules to manual-rules +- ci/x86: Build ANGLE for testing layering on VK drivers. +- ci/amd: switch all possible jobs from X11 to Wayland +- ci/freedreno: switch a630_{piglit,skqp} and a618_gl to Weston +- ci/freedreno: re-enable a530 as it's now stable with multiple skips +- ci/freedreno: document number of a630 devices available +- ci/freedreno: add KHR-GL46.buffer_storage flakes +- ci/freedreno: add execution@varying-struct-copy-return-vs flake +- ci/container: add weston into Vulkan container +- ci/container: we need to keep the wine inside +- ci/traces: switch from xvfb to Weston XWayland +- ci/freedreno: another batch of a530 flakes +- ci: add quirk for GitLab assuming changes is always true for scheduled runs +- ci/microsoft: when re-enabling Windows Farm, always run the container + +David Redondo (1): + +- egl/wayland: fix oob buffer access during buffer_fds clean up + +David Rosca (1): + +- radeonsi: Use DIV_ROUND_UP instead of ALIGN_POT + +Diederik de Haas (1): + +- treewide: spelling fixes + +Dmitry Baryshkov (2): + +- freedreno/registers: updte HDMI registers to include CEC details +- freedreno/registers: add bitfield for DSI wide bus enablement + +Dmitry Osipenko (4): + +- iris/bufmgr: Use intel_ioctl() helper for GEM_SET_TILING +- intel/dev: Use intel_ioctl() helper for GEM_SET_TILING +- anv: Use intel_ioctl() helper for GEM_SET_TILING +- hasvk: Use intel_ioctl() helper for GEM_SET_TILING + +Dmitry Rogozhkin (1): + +- meson/vaon12: fix driver file name for mingw build + +Donald Robson (2): + +- pvr: Move heap initialisation out of pvr_winsys_helper. +- pvr: Rename rogue_fw.xml -> rogue_kmd_stream.xml. + +Dor Askayo (2): + +- meson: add feature option for use of system Clang headers at runtime +- ci: Disable "opencl-external-clang-headers" when "microsoft-clc" is enabled + +Dr. David Alan Gilbert (4): + +- rusticl/screen: Wrap get_timestamp +- rusticl/device: Stash timestamp availability +- rusticl/api: Implement get_{device_and\_}host_timer +- rusticl/api: Wire up CL_DEVICE_PROFILING_TIMER_RESOLUTION + +Dylan Baker (26): + +- docs: add release notes for 23.0.1 +- docs: Add sha256 sum for 23.0.1 +- docs: add release notes for 23.0.2 +- docs: Add sha256 sum for 23.0.2 +- docs: add release notes for 23.0.3 +- docs: Add sha256 sum for 23.0.3 +- docs: update calendar for 23.0.1 +- docs: update calendar for 23.0.2 +- docs: update calendar for 23.0.3 +- docs: add release notes for 23.0.4 +- docs: Add sha256 sum for 23.0.4 +- docs: update calendar for 23.0.4 +- intel/tools/error2aub: Fix potential out of bounds read +- meson: Key whether to build batch decoder on expat +- bin/pick: fix issue where None for nomination_type could fail +- bin/pick: use lineboxes to make the UI clearer +- bin/pick: Add support for adding notes on patches +- bin/pick-ui: use asyncio.new_event_loop +- meson: Add back execmem option as a deprecated option +- VERSION: update to 23.2.0-rc1 +- docs: Update release calendar for 23.2.0-rc1 +- .pick_status.json: Update to 6e87b277bde71e30c98ab9dda7bd2f2017b77ed5 +- .pick_status.json: Update to 27d30fe3c0e71efd90fcfe209d8515b195b0075f +- .pick_status.json: Update to 3a8aae9e6aa526367523c58dfe5046909776be74 +- .pick_status.json: Update to 59087003c4b7a4f5a6bf207f214a4c3443b9759f +- ci: mark passing zink and lima tests as expected + +Emma Anholt (122): + +- zink: Avoid infinite loop finding no var in update_so_info. +- ci/crocus: Update checksum for STK. +- symbol_table: Store the symbol name in the same allocation as the symbol entry. +- symbol_table: Don't maintain the HT as we're destroying the table. +- symbol_table: Don't bother resetting the key on popping scope. +- symbol_table: Prehash the key on insert, and reuse the entry on shadowing. +- tu/perfetto: Refactor code out of the macro, to stage_end. +- tu/perfetto: Clean up an extra token paste to just use the arg being passed. +- tu/perfetto: Use tu_CmdBeginDebugUtilsLabelEXT as a stage event in perfetto. +- tu/perfetto: Drop unused arg to send_descriptors(). +- tu/perfetto: s/MRTs/attachment_count/ in traces. +- anv: Only enable GPL if ANV_GPL=true, or if zink or DXVK are the engine. +- anv: Refactor repeated pipeline creation feedback output code. +- ci/lvp: Update sanctuary trace hash. +- ci/radv: Demote navi21 to manual until recent flakiness resolves. +- ci/zink+tu: Drop some intermittently failing a630 traces. +- ci/freedreno: Drop portal-2-v2 trace. +- ci/radv: Add known flakes for #8817 +- ci: Crank up the yamllint line length limit. +- ci/freedreno: Demote a530 to manual again. +- ci: Make a variable for the repeated rootfs directory name. +- ci: Add the Vulkan validation layer to amd64 rootfs builds. +- ci/zink: Re-enable traces now that !20319 has landed. +- ci: Move zink's validation layer setup to deqp-runner.sh. +- ci/zink: Enable the validation layer on the TGL GL46 run. +- blob: Don't valgrind assert for defined memory if we aren't writing. +- util/log: Fix log messages over 1024 characters. +- ci: Move some timeout xfails to skips. +- ci/deqp: Update to 1.3.5.1 and pull in additional bugfixes from main. +- ci/zink: Drop anv/lvp validation exceptions that should be fixed in the CTS. +- ci/valve: Add a workaround for finding libdrm on navi21s. +- ci/panfrost: Drop tex3d-maxsize on g52. +- ci/lima: Skip ppgtt_memory_alignment that flaked a job with the oomkiller. +- ci/crocus: Note a recent regression. +- ci/zink: Try to update TGL results for new MSAA behavior. +- vulkan: Handle alignment failure in the pipeline cache. +- vulkan: Actually increment the count of objects in GetPipelineCacheData. +- Revert "ci/zink: Try to update TGL results for new MSAA behavior." +- ci/zink: Update more xfails for tgl piglit. +- ci/zink+anv: Test piglit quick_gl pre-merge, dropping a few KHR-GL46 tests. +- ci/radeonsi: Mark glx-make-current as flaky. +- ci/radv: Disable flaky heaven d3d9 trace. +- ci/turnip: Drop an xfail from the full run for a recent fix. +- ci/turnip: Drop the IUB bug fallout flakes. +- mesa: Fix debug logging of fp compile compare func. +- mesa: Fix precompile of GLSL programs with shadow samplers. +- zink: Explain some of the current pathway for shadow sampling. +- zink: Fix silly void * type in rewrite_tex_dest. +- zink: Don't flag legacy_shadow_mask for RED-only reads in the shader. +- ci: Re-enable some piglit tests that should be fast enough post-uprev. +- ci/zink+anv: Skip a couple more long tests pre-merge. +- compiler: Update reference to name_for_stage func. +- nir: Add helpers for lazy var creation. +- drm-shim: Avoid assertion fail if someone does close(-1). +- glsl: Allow invariant flags on sysvals, such as gl_PointCoord. +- nir/lower_texcoord_replace: Flag SYSTEM_VALUE_POINT_COORD read when we load it. +- zink: Use PIPE_CAP_FS_POINT_IS_SYSVAL. +- mesa: Use find_state_var in lower_builtin. +- nir: Use find_state_var in lower_atomics_to_ssbo. +- nir,mesa: Add helpers for creating uniform state variables. +- mesa: Move ATI_fragment_shader fog code emit to a NIR lowering pass. +- mesa/ARB_fp: Drop an extra enum for fog mode. +- mesa/ARB_fp: Use the NIR pass for adding fog code instead of ARB instrs. +- mesa: Move ARB_vp position invariant option handling to NIR. +- mesa: Drop ARB program helper functions that are no longer used. +- mesa: Drop unused control flow instructions for ARB programs. +- mesa: Drop remaining unused ARB program instructions. +- mesa: Move st_prog_to_nir_postprocess out of prog_to_nir. +- mesa/ati_fs: Move sampler dim adjustment to a separate NIR pass. +- mesa/ati_fs: Move NIR translation to ATI_fs compile time. +- mesa/ati_fs: Move prog->SamplersUsed/TexturesUsed setup to EndFragmentShader. +- mesa: Use the NIR pass for fixed function fog. +- mesa/ffvs: Fix mvp_with_dp4 position transformation. +- mesa: Use shared NIR code for ARB_vp and FF VS position transformation. +- ci/freedreno: Update minetest hash. +- Revert "ci: disable anholt's farm" +- crocus: Fix regression from !20153 +- ci/crocus: Add a missing xfail. +- ci/turnip: Update full-run xfails. +- tu: Ignore unused shader stages in pipeline library creation. +- anv: Drop unused ALL_GRAPHICS_LIB_FLAGS. +- ci/crocus: Update trace hash for the neverball regression. +- ci/etnaviv: Update some xfails common between the last 3 nightly runs. +- v3d: Respect nir_intrinsic_store_output's write_mask. +- mesa: Emit full output write in st_pbo_create_vs(). +- mesa: Port the pbo.use_gs path to NIR and let it get used on NIR drivers. +- softpipe: Drop the use_tgsi debug flag. +- llvmpipe: Drop the LP_DEBUG=tgsi_ir debug option. +- virgl: Drop the VIRGL_DEBUG=use_tgsi debug var. +- r600: Drop docs for use_tgsi debug var. +- r300: Drop RADEON_DEBUG=use_tgsi. +- nouveau: Delete the NV50_PROG_USE_TGSI env var. +- svga: Switch to preferring NIR by default. +- nine: Drop the nir_vs/nir_ps env vars. +- gallium: Drop PIPE_SHADER_CAP_PREFERRED_IR. +- mesa/drawtex: Cut out the TGSI semantic translation. +- svga: Stop asserting that compute params are queried against TGSI. +- mesa: Always query our compute params against IR_NIR. +- mesa: Drop TGSI token handling +- mesa: Simplify st_get_nir_compiler_options(). +- mesa: Drop dead TGSI serialization prototypes. +- mesa/atifs: Rename the header guard. +- mapi: clang-format _glapi_add_dispatch(). +- mapi: Delete dynamic stub generation. +- mesa: Drop the function parameter spec from the remap table. +- mapi: Clean up mapi_stub struct. +- mesa: Drop the aliases from the remap table. +- mapi: Drop the unused_functions table. +- mapi: Delete execmem support code. +- intel: Count reads_remaining across all blocks. +- intel: Allocate the last_grf_write once per scheduler. +- intel: Reduce cost of resetting last_grf_write. +- ci/zink: Update current xfails on tgl. +- ci: Update to vulkan-cts-1.3.5.2 (and pull in some more fixes). +- ci: Drop skips for some previously-invalid CTS tests. +- ci: Drop some skips of GL CTS ArraysOfArrays tests. +- ci/anv: Make anv-manual-rules actually manual on anv-only changes. +- ci: Clean up .intel-rules definition. +- ci/amd: Report flakes to #amd-ci on OFTC. +- ci/anv: Add testing of the GLES CTS using ANGLE on TGL. +- ci/radv+radeonsi: Fix the combo rules to include core vulkan changes. +- ci/radv: Add testing of the GLES CTS using ANGLE on stoney. + +Eric Engestrom (125): + +- VERSION: bump to 23.2 +- docs: reset new_features.txt +- v3d: add flake spec@ext_framebuffer_blit@fbo-sys-sub-blit +- ci: stop removing -x11 suffix for x11 build of deqp-egl +- ci: add -android suffix for android build of deqp-egl +- ci: move deqp-egl instead of copying it +- ci: start documenting which image tags need to be bumped +- ci: bump tags +- ci: update shebang to make it more portable +- broadcom/ci: deduplicate script definition +- v3dv/ci: drop fixed failure from fails.txt +- amd: fix buggy usage of unreachable() +- compiler: fix buggy usage of unreachable() +- pvr: fix buggy usage of unreachable() +- vk/util: fix buggy usage of unreachable() +- util: enforce unreachable()'s argument being a literal string +- egl: inline driver.GetProcAddress() as it's always _glapi_get_proc_address() +- ci: rework vulkan validation layer build script +- v3d: document that \`V3D_DEBUG=shaderdb` is \*not* for shader-db +- v3d: fix tfu_supports_tex_format() param type, and document why +- v3d: fix various minor issues in gen_pack_header.py +- dzn: fix pointer type mismatch +- ci: bump bin/ci/ deps to support python 3.11 +- ci: drop GENERATE_ENV_SCRIPT +- ci: stop marking environment variable list as executable +- ci: replace write + cat with tee +- ci: disable anholt's farm +- ci: only execute capture-devcoredump.sh when it's present +- util/bitset: ensure the sets compared have the same size at compile time +- docs: add release notes for 23.1.0 +- docs: update calendar for 23.1.0 +- ci/b2c: increase timeout to 5 minutes +- ci/amd: don't override the b2c timeout in the steamdeck config +- ci/zink: add new zink-radv-navi10-valve flakes +- mailmap: update @mupuf's name +- docs: fix release date of 23.1.0 +- ci/zink: document new zink-radv-navi10-valve failures +- v3dv: fix align() computation for pixel formats with non-POT block sizes +- docs: update calendar for 23.1.1 +- docs: add release notes for 23.1.1 +- docs/relnotes: add sha256sum for 23.1.1 +- ci_run_n_monitor: add ability to specify the pipeline to use, instead of auto-detecting it +- ci/amd: move AMD-specific LD_PRELOAD to AMD config +- ci/amd: only define AMDGPU_GPU_ID for the duration of the call +- bin/ci: fix mistakenly hardcoded repo name in get_gitlab_project() +- ci/intel: reuse iris_file_list instead of copying its definition +- meson: simplify another "any of" check +- wsi/display: drop unused parameters from local functions +- ci: split clang-format list of folders for easier maintenance +- ci: show diff when clang-format check fails +- panfrost: fix formatting of a couple of files that were missed +- panfrost: rename \*.cc files to \*.cpp +- ci/zink+radv: fix flakes definition +- ci/zink+radv: mark all spec@arb_copy_image@arb_copy_image-targets* as flaky after getting a bunch more of them +- ci/zink+radv: document recent regressions +- ci: color the diff for clang-format +- meson: enable the clang-format target +- ci: use meson to run clang-format +- docs: document clang-format and how to use it +- docs/calendar: add 23.2 branchpoint and release candidates +- ci/zink+radv: mark flakes as such +- ci/radv: fix flakes definition +- ci/crocus: fix flakes definition +- ci/zink+anv: fix flakes definition +- ci/b2c: also detect non-soft GPU hangs with AMDGPU +- amd/ci: run gl(es) cts & piglit on radeonsi on vangogh +- ci/radv: update expectations +- ci/zink+radv: update expectations +- docs/relnotes/23.1.1: clear "new features" +- docs: add release notes for 23.1.2 +- docs/relnotes: add sha256sum for 23.1.2 +- docs: update calendar for 23.1.2 +- egl: return correct error for EGL_KHR_image_pixmap +- clang-format: add explanation for anyone reading .clang-format-include +- radv,aco: tweaks to get clang-format to print nicer code +- radv: reformat according to its .clang-format +- aco: reformat according to its .clang-format +- ci: enforce formatting for RADV & ACO +- radv: fix formatting +- Revert "ci: remove clang-format testing" +- asahi: drop unnecessary DRM_FORMAT_MOD_{LINEAR,INVALID} fallbacks +- ci: mark the valve farm as down +- docs/ci: fix command to disable/re-enable farms +- docs: add release notes for 23.1.3 +- docs/relnotes: add sha256sum for 23.1.3 +- docs: update calendar for 23.1.3 +- docs/coding-style: add example vim config for clang-format +- docs/coding-style: add example emacs config for clang-format +- docs/coding-style: add pre-commit hook fallback for clang-format +- v3dv: replace boolean and uint with bool and size_t +- amd/ci: add another dEQP-VK.multiview.renderpass2.multisample.* flake +- amd/ci: add another dEQP-VK.dynamic_rendering.primary_cmd_buff.basic.* flake +- ci: split valve farm in two +- util/disk_cache: fix ~/.cache/ permissions +- panfrost/ci: drop invalid skips that are already marked as known flakes +- intel/ci: fix skips definitions +- etnaviv/ci: fix skips definition +- zink/ci: fix skips & flakes for zink+radv on vangogh & navi10 +- docs/codingstyle: fix clang-format command +- vc4/ci: fix skipping of gles3 piglit tests +- v3dv/ci: fix skipping of vk tests +- v3dv/ci: skip more tests that are timing out +- virgl/ci: fix skips definition +- clang-format: add egl foreach macro +- clang-format: add wayland foreach macros +- egl: change a couple of clang-format settings +- egl: add a few trailing commas +- egl: protect the formatting in a couple of places +- egl: prevent clang-format from reordering some headers +- egl: re-format using clang-format +- clang-format: enforce formatting of egl +- add initial .git-blame-ignore-revs +- ci/zink+radv: document another flake +- ci/zink+radv: fix flake definition +- ci: document workflow rules +- ci: set priority:low tag only on non-Marge pipelines +- ci: fix .valve-farm-manual-rules +- ci: split farm rules out of test-source-dep.yml +- etnaviv/ci: drop duplicate line in etnaviv files list +- broadcom/ci: add the renderonly folder to things that can affect v3d & vc4 +- meson: clarify description of \`opengl` option +- meson: clarify what "off-screen rendering" means +- ci: avoid running hardware jobs if there are already trivial issues +- ci: avoid running hardware jobs if lint fails - now on LAVA too! +- ci: avoid running hardware jobs if lint fails - now on Windows too! + +Erico Nunes (5): + +- Revert "ci: disable lima farm, currently out-of-space, needs to be fixed" +- lima: fix stringop-overflow warning +- lima/ci: temporarily disable deqp-egl tests due to timeouts +- ci: temporarily disable lima farm +- ci: restore lima farm + +Erik Faye-Lund (143): + +- nir: remove nir_state_slot::swizzle +- glsl: remove ir_state_slot::swizzle +- docs: renderpass -> render pass +- docs: statechanges -> state changes +- docs: backfacing -> back-facing +- docs: codepath -> code-path +- docs: did't -> didn't +- docs: cma -> CMA +- docs: Anv -> ANV +- docs: perfetto -> Perfetto +- docs: use correct tick for "doesn't" +- docs: vlan -> VLAN +- docs: toplevel -> top-level +- docs: correct spelling of "source" +- docs: correct spelling of "tagged" +- docs: correct spelling of "frame" +- docs: sort extensions +- docs: add custom html theme +- docs: add bootstrap extension +- docs: translate admonitions into bootstrap alerts +- docs: remove support for old sphinx-versions +- docs: use custom html theme +- nir: clean up white-space in deref-printing +- mesa/main: clean up white-space in ffvertex_prog.c +- mesa/main: drop disasm-code from ffvertex_prog.c +- mesa/main: allow passing nir-shaders to st_program_string_notify +- mesa/main: make ffvertex output nir +- nir: fix constant-folding of 64-bit fpow +- docs: fix edit-links +- mesa/main: drop use_legacy_math_rules +- llvmpipe: fixup refactor copypasta +- docs: fixup About Mesa3D.org link +- docs/tgsi: fix up indent +- docs/tgsi: fix bad latex +- docs/tgsi: fixup bad latex +- docs/tgsi: wrap overly long lines +- docs/tgsi: use math-notations for conditionals +- docs/tgsi: do not use math-block for non-latex +- docs/tgsi: fixup latex for TEX and TEX2 +- docs/tgsi: use \\ll and \\gg for left and right shift +- aux/draw: check for lines when setting clipping-mode +- zink: fix bad indent +- zink: clean up tcs_vertices_out_word handling +- zink: do not open-code memcpy +- aco: use c++17 +- meson: remove needless c++17-overrides +- mesa/main: clean up white-space in ff_fragment_shader.cpp +- mesa/st: refactor st_translate_fragment_program +- mesa/st: allow using nir for ff-fragment shaders +- compiler/nir: move find_state_var to common code +- mesa/main: ff-fragshader to nir +- mesa/main: compile ff_fragment_shader as c-code +- mesa/program_cache: remove unused shader-cache functions +- panfrost: expose PIPE_CAP_POLYGON_OFFSET_CLAMP +- util: mark externally-unused functions as static +- nir: use more nir_fmul_imm +- nir: use more nir_fadd_imm +- nir: fsub -> fadd_imm +- nir: use more nir_ffma_imm variants +- nir: add nir_fsub_imm +- nir: use nir_fsub_imm +- radeonsi,radv: use nir_format_linear_to_srgb +- docs: explicitly mark extensions as obsolete +- docs: mark MESA_multithread_makecurrent as obsolete +- docs: mark MESA_shader_debug as obsolete +- docs: mark MESA_swap_frame_usage as obsolete +- docs: mark MESA_texture_array as obsolete +- docs: move obsolete extensions to their own list +- zink: update profiles schema +- zink: keep gl46_optimal extensions/features sorted +- zink: compute correct location for line-smooth gs +- zink: do not lower line-smooth for non-lines +- docs: increase contrast in dark-theme +- zink: update profiles schema +- d3d12, dozen: make sure we pass float to fge +- nir: use nir_i{ne,eq}_imm helpers +- nir: generate nir_{cmp}_imm variants +- nir: use generated immediate comparison helpers +- nir: add nir_[fui]gt_imm and nir_[fui]le_imm helpers +- nir: use new immediate comparison helpers +- mesa/st: use nir_imm_vec4 +- nir: use more imm-helpers +- nir: isub -> iadd_imm +- nir: use nir_imm_{true,false} +- nir: add and use nir_fdiv_imm +- nir: add and use nir_imod_imm +- nir: add missed nir_cmp_imm-helpers +- docs: upgrade bootstrap to 5.3.0 +- cso: use enum for render-conditions +- draw: use enum for tgsi-semantic +- draw: use uint32_t instead of uint +- draw: use enum for primitive-type +- draw: track vertices and vertex_ptr as byte-pointers +- draw: use stdint.h types +- cso: use unsigned instead of uint +- draw: match type of pipe_draw_start_count_bias::count +- draw: use unsigned instead of uint +- aux/indices: use stdint.h types +- draw/i915: move hwfmt array to i915 specific struct +- microsoft/compiler: use nir_imm_zero +- mesa/st: use nir_ineg +- vulkan: avoid needless constant-folding +- broadcom/compiler: use imm-helpers +- v3dv: use imm-helpers +- pan: use imm-helpers +- freedreno: use imm-helpers +- r600/sfn: use imm-helpers +- d3d12: use imm-helpers +- radeonsi: use imm-helpers +- vc4: use imm-helpers +- intel: use imm-helpers +- anv: use imm-helpers +- hasvk: use imm-helpers +- mesa/st: use imm-helpers +- amd: use imm-helpers +- etnaviv: use imm-helpers +- gallium: use imm-helpers +- nir: use imm-helpers +- math: fix indentation in m_matrix.[ch] +- math: remove unused defines +- math: drop MAT_[ST][XYZ] defines +- aux/trace: use stdint.h types +- pipebuffer: use unsigned instead of uint +- gallivm: use unsigned instead of uint +- aux/pp: use unsigned instead of uint +- aux/util: use enum for render-condition +- aux/util: match type of pipe_draw_start_count_bias::start/count +- aux/util: use enum for primitive-type +- aux/util: use unsigned instead of uint +- aux/util: use stdint.h types +- aux/util: uint -> unsigned +- tgsi: use enum instead of defines +- tgsi: use stdint.h types +- tgsi: use enum for tgsi-file type +- tgsi: use enum for property-name +- tgsi: use enum for shader-type +- tgsi: use enum for interpolate-mode +- tgsi: uint -> uint32_t +- tgsi: uint -> unsigned +- nir: constify intrin +- nir: use nir_intrinsic_get_var +- radv: do not rely on constant-folding +- nir: do not needlessly rely on optimizations + +Faith Ekstrand (8): + +- nouveau/nir: image_samples/size don't have coordinates +- vulkan: Document vk_physical_device::supported_features +- nir/opt_if: Use block_ends_in_jump +- nir: Add a reg_intrinsics flag to nir_convert_from_ssa +- nir/from_ssa: Make additional assumptions in coalescing +- nir/from_ssa: Support register intrinsics +- freedreno/ci: Update pixmark piano checksums +- nv50/ir: Support vector movs + +Felix DeGrood (19): + +- anv: disable reset query pools using blorp opt on MTL +- anv: Add END_OF_PIPE_SYNC reporting to INTEL_DEBUG=pc +- anv: Add flush reasons to raytracing flushes +- anv: Add flush reason to NEEDS_END_OF_PIPE_SYNC +- anv: split INTEL_MEASURE multi events +- intel: INTEL_MEASURE cpu mode +- anv: Enable INTEL_MEASURE=cpu +- iris: Enable INTEL_MEASURE=cpu +- docs: add INTEL_MEASURE=cpu +- intel/debug: Control start/stop frame of batch debug +- anv: Enable INTEL_DEBUG_BATCH_FRAME_START/_STOP +- iris: Enable INTEL_DEBUG_BATCH_FRAME_START/_STOP +- docs: Add INTEL_DEBUG_BATCH_FRAME_START/_STOP +- anv: fix INTEL_MEASURE on MTL +- anv: re-enable RT data in INTEL_MEASURE +- intel: refactor INTEL_MEASURE pointer dumping +- intel: batch consecutive dispatches into implicit renderpasses +- intel: Secondary CB print primary CB's renderpass +- anv: override vendorID for Cyberpunk 2077 + +Feng Jiang (3): + +- frontends/va: Fix memory leak of decrypt_key +- radeonsi/vcn: Remove unnecessary type conversion +- virgl/video: Fix out-of-bounds access in fill_mpeg4_picture_desc() + +Filip Gawin (5): + +- nine: add fallback for D3DFMT_D16 in d3d9_to_pipe_format_checked +- glx: fix build with APPLEGL +- ac/nir: fix slots in clamping legacy colors +- anv: allow intel_clflush_range only on igpu +- crocus: Avoid fast-clear with incompatible view + +Francisco Jerez (3): + +- anv: Fix calculation of guardband clipping region. +- intel/gfx12.5: Enable L3 partial write merging for compressible surfaces among other cases. +- anv: Swap ordering of memory types on non-LLC platforms to work around application bugs. + +Frank Binns (7): + +- pvr: add missing explicit check against VK_SUCCESS +- pvr: use util_dynarray_begin() in more places +- pvr: replace transfer EOT binary shaders with run-time compiled shaders +- pvr: fix typo in pvr_rt_get_region_headers_stride_size() +- pvr: fix array overflow in pvr_device_tile_buffer_ensure_cap() +- pvr: fix invalid read reported by valgrind +- pvr: skip setting up SPM consts buffer when no const shared regs are used + +Friedrich Vock (39): + +- radv/rmv: Fix creating RT pipelines +- radv/rmv: Fix import memory +- radv/rt: Plug some memory leaks during shader creation +- radv: Don't leak the RT prolog binary +- radv: Hash pipeline libraries separately +- radv: Always call si_emit_cache_flush before writing timestamps +- radv: Add driconf to always drain waves before writing timestamps +- nir: Rematerialize derefs in use blocks before repairing SSA +- nir: Remove unnecessary assert in nir_before_src +- radv: Disable capture/replay handles +- aco: Lower divergent bool phis iteratively +- radv: Always flush before writing acceleration structure properties +- aco: Reset scratch_rsrc on blocks without predecessors +- aco: Fix live_var_analysis assert +- aco: Fix assert in insert_exec_mask +- radv: Add driconf to force wave64 for RT +- radv: Add RADV_DEBUG=nort +- radv: Enable ray tracing pipelines by default +- radv: Add the BOs of all shaders in a RT pipeline +- radv: Add radv_shader_free_list +- radv: Move shader arena allocation to a separate function +- radv: Add option to allocate shaders in replayable VA range +- radv: Add utilities to serialize and deserialize shader allocation info +- radv: Add radv_shader_reupload +- radv: Break up radv_shader_nir_to_asm +- radv: Split up implementation of radv_shader_create +- radv: Add support for creating capture/replay shaders +- radv: Add radv_rt_capture_replay_handle +- radv/rt: Only compare the non-recursive capture/replay handle +- radv/rt: Associate capture/replay handles with stages +- radv/rt: Replay shader allocations according to capture/replay handle +- radv/rt: Rework radv_GetRayTracingCaptureReplayShaderGroupHandlesKHR +- radv: Re-enable RT pipeline capture/replay handles +- meson: Prefix Vulkan "Ray Tracing" summary with "Intel" +- radv/ci: Skip ray tracing tests on vangogh +- Revert "radv/rt: Enable RT pipelines on GFX10_3+ excluding vangogh" +- Revert "Revert "radv: Enable ray tracing pipelines by default"" +- radv/rt: Enable exact on software intersection functions +- radv/rt: Miss rays that hit the triangle's v edge + +Ganesh Belgur Ramachandra (5): + +- gallium/pipe: Add get_resources() to pipe_video_buffer +- gallium/vl: implementation for get_resources() +- nouveau: implementation for get_resources() +- d3d12: implementation for get_resources() +- frontends/va: use resources instead of views + +Georg Lehmann (47): + +- nir: lower ballot_bit_count_exclusive/inclusive to mbcnt_amd +- radv: use lower_ballot_bit_count_to_mbcnt_amd +- aco: Assert that operands have the same byte offset when reassigning split vectors +- aco: also reassign p_extract_vector post ra +- aco/vn: compare all valu modifers +- aco/optimizer: don't use pass_flags for mad idx +- aco/optimizer: copy pass flags for newly created valu instructions +- aco/assembler: support VOP3P with DPP +- aco/builder: support VOP3(P) with dpp +- aco: add assembler tests for VOP3(P) with DPP +- aco/ra: convert VOPC_DPP instructions without vcc to VOP3 +- aco: use VOP3+DPP +- aco: don't apply dpp if the alu instr uses the operand twice +- aco: emit_wqm on MIMG dst, not operands +- aco: introduce helper to swap valu operands with modifiers +- aco/gfx11: use fmamk/fmaak with opsel +- aco: add withoutVOP3 helper +- aco/ra: use smaller operand stride for VOP3P with DPP +- aco/ra: use fmac with DPP/opsel on GFX11 +- aco: add helper function for can_use_input_modifiers +- aco: use get_operand_size for dpp opt +- aco: use can_use_input_modifiers helper +- aco/optimizer: allow DPP to use VOP3 on GFX11 +- util: fix stack dynarray used by multiple tus +- nir/opt_if: use nir_alu_instr_is_comparison directly +- aco: cleanup v_cmp_class usage +- aco: p_start_linear_vgpr doesn't always need exec mask +- aco/ir: return true in hasRegClass for Operand(reg, rc) +- aco/statistics: improve v_fma_mix dual issuing detection +- aco: use v_add_f{16,32} with clamp for fsat +- aco: use v_fma_mix for f2f32 and f2f16 on gfx11 if wave64 +- aco: make validation work without SSA temps +- aco: move cfg validation to its own function +- aco: don't validate p_constaddr_addlo/p_resumeaddr_addlo operands +- aco: validate ir for prologs and after lower_to_hw_instr +- aco/opcodes: move v_cndmask_b32 back to the VOP2 list +- aco: remove v_cvt_pkrtz_f16_f32_e64 when it's actually VOP2 +- aco/opcodes: delete wrong comment copy pasted from NIR +- aco: use uses helpers for pk_fma opt +- aco: combine scalar mul+pk_add to pk_fma +- aco/gfx10+: use v_cndmask with literal for reduction identity +- nir: add single bit test opcodes +- nir/lower_bit_size: mask bitz/bitnz src1 like shifts +- aco: implement nir_op_bitz/bitnz +- nir/opt_algebraic: combine bitz/bitnz +- radv: set has_bit_test for aco +- aco/optimizer: delete s_bitcmp optimization + +George Ouzounoudis (1): + +- radv: small fix for VkDescriptorSetVariableDescriptorCountLayoutSupport + +Gert Wollny (93): + +- r600/sfn: Lower tess levels to vectors in TCS +- r600/sfn: make sure f2u32 is lowered late and correctly for 64 bit floats +- r600: remove TGSI code path +- r600/sfn: Add a type for address registers +- r600/sfn: don't track address registers in live ranges +- r600/sfn: Handle MOVA_INT in sfn assembler +- r600/sfn/tests: Cleanup and move some code around +- r600/sfn: Add address and index registers creation to ValueFactory +- r600/sfn: Rework query for indirect access in alu instr and opt +- r600/sfn: don't allow more than one AR per instruction +- r600: Allow both index registers for all CF types +- r600/sfn: Prepare uniforms and local arrays for better address handling +- r600/sfn: handle AR and IDX register in shader from string +- r600/sfn: add method to update indirect address to all instrution types +- r600/sfn: Add function to insert op in block +- r600/sfn: Update resource based instruction index mode check +- r600/sfn: Be able to track expected AR uses +- r600/sfn: AR and IDX don't need the write flag, but haev a parent +- r600/sfn: Add a RW get function of IF predicate access +- r600/sfn: Add interface to count AR uses in ALU op +- r600/sfn: Add pass to split addess and index register loads +- r600/sfn: Add function to check whether a group loads a index register +- r600/sfn: take address loads into account when scheduling +- r600/sfn: Add more tests and update to use address splits +- r600/sfn: Don't copy-propagate indirect access into LDS instr +- r600/sfn: Add test for multiple index load +- r600/sfn: set CF force flag always when starting a new block +- r600/sfn: Start a new ALU CF on index use, not on index emission +- r600/sfn: Add chip family to shader class +- r600/sfn: Add handling for R600 indirect access alias handling +- r600/sfn: Override Array access handling in backend assembler +- r600/sfn: Fix copy-prop with array access +- r600/sfn: scheduled instructions are always ready +- r600/sfn: Add more tests and update to use address splits +- r600/sfn: print failing block when scheduling fails +- r600/sfn: Can't use an indirect array access as source to AR load +- r600/sfn: factor out index loading for non-alu instructions +- r600/sfn: prepare for emitting AR loads +- r600/sfn: Tie in address load splitting +- r600+sfn: Assign ps_conservative_z and switch to NIR defines +- r600/sfn: assign window_space_position in shader state +- r600/sfn: Ass support for image_samples +- r600/sfn: fix cube to array lowering for LOD +- r600/sfn: Fix iterator use +- r600/sfn: move kill instruction test to alu instruction +- r600/sfn: add dependencies for kill instructions +- r600/sfn: move kill handling fully to scheduling +- r600/sfn: use correct FS output location if not all outputs are used +- virgl: Make query result resource as dirty before requesting result +- virgl: Add support for ARB_pipeline_statistics +- virgl/ci: uprev virglrenderer +- docs/features: fix empty line error +- virgl: Fix IB upload when a start >0 is given +- virgl: Submit drawid_offset if is not zero +- virgl: signal support for group vote and draw parameters +- virgl: enable ARB_gl_spirv +- features: Update virgl features +- ci: uprev virglrenderer to include changes needed for GL 4.6 support +- r600/sfn: assert that group barrier is not emitted in divergent code flow +- r600/sfn: Switch to scoped barriers +- util/driconf: pin minImageCount to three for "Path of Exile" +- r600/sfn: add read instruction for unused but required LDS op results +- r600/sfn: Don't rewrite TESS_EVAL inner tess level outputs +- r600/sfn: Add experimental support for load/store_global +- r600/sfn: Handle store_global when lowering 64 bit ops to vec2 +- r600/sfn: Handle load_global in 64 to vec2 lowering +- rusticl: compile r600 driver +- r600: fix handling of use_sb flag +- r600/sfn: move kill handling to fully scheduling +- 600/sfn: Trigger use of ACK for some barriers +- r600: Disable SB if we use the ariable length DOT +- r600/sfn: Silence warnings "overloaded-virtual" +- r600/sfn: Downgrade some error message to warning +- r600: Split tex CF only if written component is read +- r600/sfn: Don't deref unused group slots +- r600/sfn: on R600/R700 write a dummy pixel output if there is a gap +- r600/sfn: Clean up FS member initialization +- virgl: don't allow vertex input arrays on GLES hosts +- r600/sfn: Fix typo +- r600/sfn: drop use of nir source mods +- r600/sfn: allow source mods for per source with multi-slot ops +- r600/sfn: add source and dest mod info to opcode table +- r600/sfn: Implement source mod optimization in backend +- r600/sfn: Implement fsat for 64 bit ops +- r600/sfn: Add source mod propagation also to fp64 ops +- r600/sfn: Don't clear clear group flag on vec4 that comes from TEX or FETCH +- virgl/ci: Drop duplicate runs +- ci: Upref virglrenderer +- r600/sfn: Fix filling FS output gaps +- r600: Pre-EG - Set wrap texture modes to repeat when seemless cube is used +- r600/sfn: Be more conservative with AR re-use +- r600/sfn: Shorten array elements live range +- r600/sfn: remove debug output leftovers + +Giancarlo Devich (5): + +- d3d12: Update and require DirectX-Headers 1.610.0 +- d3d12: Query device for D3D12_FEATURE_D3D12_OPTIONS14 +- d3d12: Update PSO creation to use CreatePipelineState +- d3d12: Add ID3D12GraphicsCommandList8 to the context +- d3d12: Support separate front/back stencils + +Gregory Mitrano (2): + +- ac/sqtt: Add RGP Definitions for Mesh Shaders +- radv/sqtt: Add RGP Markers for Mesh Shaders + +Guilherme Gallo (29): + +- ci/lava: Move job definition stuff to another file +- ci/lava: Extract LAVA proxy and LAVAJob abstractions +- ci/lava: Use python-fire in job submitter +- ci/lava: Update LogFollower for better section handling and history +- ci/lava: Add a simple Structural Logger into submitter +- bin/ci: Add StructuredLogger to improve log handling +- ci/lava: Integrate StructuralLogger with AutoSaveDict +- ci/lava: Force use of UTC timezones +- ci/lava: Refactor LAVAJobSubmitter and add tests +- ci/lava: Use f-strings in job definition +- ci/lava: Skip regression test if LAVA log file is not present +- ci/freedreno: Fix a618-traces-performance rules +- ci/lava: Bypass arg list to print_log function +- ci/lava: Fix last section in job submitter +- ci: Use absolute paths in init-stage2.sh +- ci/lava: Add SSH support in rootfs +- ci/lava: Add SSH job definition +- ci/lava: Add bridge function for job definition +- ci/lava: Distinguish test suites in DUT vs Docker +- ci/lava: Only check for the first section marker +- ci/lava: Hide JWT block during YAML dump +- ci/lava: Tweak http-download timeout in SSH based jobs +- ci/lava: Raise the post test metadata gathering retry count +- ci/lava: Force LAVA panfrost jobs to use UART +- dzn: Skip a few deqp tests which are prone to timeout +- ci/lava: Renable SSH sessions for panfrost jobs +- ci/lava: Increase Docker action failure_retry counter +- ci/lava: Add LAVA SSH client container +- ci/lava: Use an alpine image for SSH client container + +Hans-Kristian Arntzen (5): + +- wsi/x11: Fix present ID signal when IDLE comes before COMPLETE. +- wsi/wayland: Simplify wait logic for present wait. +- wsi/wayland: Do not assert that all present IDs have been waited on. +- radv/amdgpu: Report 48-bit VAs in bo logs. +- Fix DGC bug where indirect count > maxSequencesCount. + +Harri Nieminen (11): + +- amd: fix typos +- amd: fix typos in code +- r300: fix typos +- radeonsi: fix typos +- r600: fix typos +- r600/sb: fix typo +- r600/sfn: fix typos +- r600/sfn: fix typos in code +- broadcom: fix typos +- egl: fix typos +- glx: fix typos + +Helen Koike (2): + +- ci: move .microsoft-farm-container-rules to test-source-dep.yml +- ci: remove unused tag DEBIAN_X86_64_TEST_IMAGE_PATH + +Hyunjun Ko (27): + +- intel/genxml: fix num bits of some MOCS fields +- intel/genxml: conform some fields to each other gen. +- intel/genxml: align some fields on gen9/11/12/125 with media driver. +- intel/genxml: add a command VD_CONTROL_STATE to gen12/125 +- util/vl: initialize data/end pointers. +- vulkan/video: add to parse h265 slice. +- vulkan/video: add h265 reference structures and relevant util functions. +- anv/image: Add a surface usage bit for video decoding +- anv/image: allocate mv storage buffers for h265 +- anv/image: allow VK_IMAGE_CREATE_ALIAS_BIT with a private binding. +- anv: add initial video decode support for h265 +- anv: support P010 format for video 10-bit hevc decoding +- anv/image: get width/height for each plane of a surface for video decoding. +- anv: support HEVC 10-bit decoding +- anv: enable the video h265 decode extension. +- anv/ci: Add tests for video formats to the failing tests. +- anv/video: move video requirements to outarray. +- vulkan/video: adds more conditions for setting loop_filter_across_slices_enable in h265 slice parsing. +- vulkan/video: move parsing longterm rps in h265 slice parsing. +- util/rbsp: keep track of removed bits for the emulation prevention three bytes. +- vulkan/video: consider removed bits when calculating the size of comsumed data. +- anv/video: fix to set U/V offset correctly. +- vulkan/video: keep delta weight and offsets of predicted weight tables in h265 slice parsing +- intel/genxml: changes the type for predicted weight to unsigned. +- anv: fix to set predicted weight tables correctly. +- anv/video: fix to support HEVC 10bit on some of 9th gens. +- anv: Adds a workaround for HEVC decoding on some old platforms. + +Iago Toral Quiroga (34): + +- broadcom/compiler: fix v3d_qpu_uses_sfu +- broadcom/compiler: add a v3d_qpu_instr_is_legacy_sfu helper +- broadcom/compiler: fix incorrect check for SFU op +- broadcom/compiler: fix incorrect ALU checks +- broadcom/compiler: return early for SFU op latency calculation +- broadcom/compiler: try harder to merge thread switch earlier +- broadcom/compiler: don't allocate undef to rf0 +- broadcom/compiler: move buffer loads to lower register pressure +- broadcom/compiler: increase peephole limit to 24 instructions +- broadcom/compiler: use unified atomics +- broadcom/compiler: skip jumps in non-uniform if/then when block cost is small +- v3dv: simplify too small Z viewport scale workaround +- v3dv: store slice dimensions in pixels +- v3dv: allow TFU transfers for mip levels other than 0 +- v3dv: align compressed image regions to block size +- broadcom/compiler: flag use of control barriers +- broadcom/compiler: use scoped barriers +- v3d: only warn about bining sync for indirect draw once +- v3dv: remove bogus viewport code +- v3dv: simplify scissor setup for negative viewport height +- broadcom/cle: fix up viewport offset packet definition for V3D 4.1+ +- v3d,v3dv: fix viewport offset for negative viewport center +- broadcom/compiler: only use last thread switch flag to detect final section +- nir/lower_tex: copy missing fields when creating copy of tex instruction +- nir/lower_tex: handle lower_tg4_offsets with lower_tg4_broadcom_swizzle +- broadcom/compiler: handle textureGatherOffsets +- v3dv: expose shaderImageGatherExtended +- v3dv: fix slice size for miplevels >= 2 +- v3dv: don't use the TLB path if we might be copying partial tiles +- v3dv: use div_round_up for division by block size +- v3dv: fix blit path for compressed image to buffer copies +- broadcom: use nir info to keep track of implicit sample shading +- broadcom/compiler: free defin and defout arrays if they already exist +- broadcom/compiler: don't leak v3d_compile when finding a new best strategy + +Ian Romanick (17): + +- intel/fs: Don't munge source order of 3-src instructions in opt_algebraic +- intel/fs: Fix handling of W, UW, and HF constants in combine_constants +- intel/fs: Allow HF const in MAD on Gfx12.5 if all sources are HF +- nir/algebraic: Fixup iadd3 related patterns +- intel/fs: Add constant propagation for ADD3 +- intel/eu/validate: Use a single macro define half_float_conversion cases +- intel/eu/validate: Add Gfx12.5 +- intel/eu/validate: Add some validation of ADD3 +- nir: Add optimization pass to reassociate some bfi instructions +- intel/fs: Use nir_opt_reassociate_bfi +- nir/algebraic: Lower some bfi with two constant sources +- intel/fs: Emit better code for bfi(..., 0) +- nir/algebraic: Optimize some u2f of bfi +- nir/algebraic: Simplify various trivial bfi +- intel/stub_gpu: Don't run program again after using GDB +- intel/fs: Constant propagate into SHADER_OPCODE_SHUFFLE +- intel/fs: Add missing newline + +Ikshwaku Chauhan (2): + +- radeonsi/gfx11: updated si_is_format_supported +- radeonsi/gfx11: updated vertex format changes + +Illia Abernikhin (3): + +- docs: add iris features to docs/features.txt +- docs: add crocus features to docs/features.txt +- docs: remove i965 features from docs/features.txt + +Illia Polishchuk (4): + +- glx: add fail check for current context in another thread +- drirc: add allow_sampled_tex_copy option +- nir: switch to a normal sampler for ARB program with not depth textures +- zink, drirc: Add Borderlands 2 workaround to fix spir-v 1.6 translated discard + +Italo Nicola (21): + +- egl: disable partial redraw when gallium hud is active +- egl: fix comments alignment +- freedreno: implement clear_render_target and clear_depth_stencil +- v3d: implement clear_render_target and clear_depth_stencil +- vc4: implement clear_render_target and clear_depth_stencil +- d3d12: fix clear_depth_stencil texture deref +- gallium: implement u_default_clear_texture +- gallium: use u_default_clear_texture where applicable +- gallium: rename util_clear_texture to util_clear_texture_sw +- mesa/st: use fallback path when pipe->clear_texture is not available +- rusticl: use fallback path when pipe->clear_texture is not available +- clover: use fallback path when pipe->clear_texture is not available +- gallium: cleanup util_blitter_clear_render_target +- gallium: remove PIPE_CAP_CLEAR_TEXTURE +- lima/ci: add some ARB_clear_texture piglit tests to lima-fails.txt +- d3d12/ci: add piglit arb_clear_texture-integer fail to CI expectations +- nir: add options to lower y_vu, yv_yu, yx_xvxu and xy_vxux +- gallium/st: add support for PIPE_FORMAT_NV21 and PIPE_FORMAT_G8_B8R8_420 +- mesa/main: add PIPE_FORMAT_YVYU and PIPE_FORMAT_R8B8_R8G8 +- mesa/main: add PIPE_FORMAT_VYUY and PIPE_FORMAT_B8R8_G8R8 +- freedreno/ci: add KHR-GL46.buffer_storage.map_persistent_flush to flakes + +Iván Briano (16): + +- anv: Remove dead parameters from copy_fast_clear_dwords +- anv: make anv_can_fast_clear_color_view more generally available +- anv: factor out code for ccs_op and mcs_op +- anv: expose some helper functions +- anv: support fast color clears on vkCmdClearAttachments +- anv: put EXT_mesh_shader behind an environment variable +- anv: enable graphics pipeline libraries by default +- hasvk: avoid assert due to unsupported format +- anv: enable the GPL feature based on whether the extension is supported +- vulkan/wsi: fix double free on error condition +- anv: do not explode on 32 bit builds +- anv: update conformanceVersion +- anv: flush data cache before emitting availability +- anv: ensure CFE_STATE is emitted for ray tracing pipelines +- iris: ensure mesh is disabled on context init +- anv: ensure mesh is disabled on context init + +James Glanville (7): + +- pvr: Improve support for image clears +- pvr: Fix vtxin special var allocation count +- pvr: Fix image to buffer copies +- pvr: Fix incorrect PBE packmode for S8_UINT +- pvr: Adjust clear's region clip words +- pvr: Fix seg fault on unused ds attachment +- pvr: Fix deferred_control_stream_flags + +James Knight (1): + +- meson: ensure i915 Gallium driver includes Intel sources + +Janne Grunau (3): + +- asahi: Fix typo in debug/error message helper macro +- asahi: Free low VA BOs correctly +- st/mesa: Set gl_config.floatMode based on color_format + +Jarred Davies (3): + +- pvr: Don't ralloc build context from compiler +- pvr: Use vk_device's enabled features struct +- pvr: Reduce free list initial size when multiple devices are created + +Jesse Natalie (133): + +- d3d12: Remove #if D3D12_SDK_VERSION blocks now that 610 is required +- microsoft/clc: Remove #if D3D12_SDK_VERSION blocks now that 610 is required +- dzn: Remove #if D3D12_SDK_VERSION blocks now that 610 is required +- util: Delete Offset() macro from u_memory.h +- d3d12: Respect buffer offsets for sampler views +- d3d12: Support blit texture uploads +- spirv2dxil: Lower quad ops in non-fragment/compute stages +- dzn: Remove driconf for quad ops in vertex stages +- dzn: Add physical device arg to format lookup +- dzn: Support dynamic depth bias via command list instead of PSO +- dzn: Use narrow quadrilateral lines when supported +- dzn: Support aniso-with-point-mip samplers +- dzn: Align-up heap sizes when allocating memory +- ci/windows: Update Agility SDK to 1.610.2 +- dzn: Use unrestricted copy alignments when available +- dzn: Handle opaque BC1 +- dzn: Handle depth bias for point fill mode emulation +- dzn: Re-design custom buffer descriptors +- ci/dzn: Run almost the full CTS +- dzn: Expose core VK1.1 extensions that aren't optional +- dzn: Expose core VK1.2 extensions that aren't optional +- meson: Don't use masm with VS backend +- spirv2dxil: Mark SSBO reads for bindless as CAN_REORDER +- microsoft/compiler: Unroll loops in opt passes +- dzn: Fix UBO descriptors pointing to the end of the buffer +- dzn: Hook up subgroup size to compute shader compilation +- dzn: Ensure sample-rate shading is factored into nir hash +- dzn: Use the nir hash as an input to the dxil hash +- dzn: Ensure subgroup size control is factored into pipeline hash +- dzn: Ensure bindless is factored into pipeline/nir hash +- dzn: Augment blit resolve to support min/max/sample-zero modes +- dzn: Support all available depth/stencil resolve modes +- dzn: Support separate depth/stencil resolves via blits +- dzn: Delete queue-level event waits +- ci/windows: Pick up WARP 1.0.6 NuGet with lots of dzn fixes +- dzn: Use A4B4G4R4 instead of B4G4R4A4 when available +- spirv2dxil: Lower large temps to scratch +- microsoft/compiler: Avoid integer divides by 0 +- dzn: Run nir_opt_remove_phis before nir_lower_returns +- dzn/ci: Remove 'exclude' for graphicsfuzz cases +- microsoft/compiler: Allocate space for I/O and viewID dependency tables before instruction processing +- microsoft/compiler: Do basic I/O analysis for dependency tables +- spirv2dxil: Support int64 and doubles +- d3d12: Convert from D3D shader model to Mesa shader model earlier +- dzn: Enable 64-bit ints and floats +- microsoft/compiler: Take inputs from callers before providing nir options +- microsoft/compiler: Enable packed dot product intrinsics for SM6.4+ +- dzn: Enable KHR_shader_integer_dot_product +- nir_lower_system_values: Add ASSERTED to assert-only variable +- nir: Load/store atomic op indices when lowering image intrinsics +- microsoft/compiler: Remove alu type info from store_dest() +- microsoft/compiler: Duplicate some SSA values to simplify SSA typing +- microsoft/compiler: Back-propagate type requirement information +- dxil: Use unified atomics +- vulkan: Win32 sync import/export support +- dzn: Don't zero an output struct that can have pNext +- dzn: Finish implementing KHR_synchronization2 +- dzn: Dedicated resource cleanup +- dzn: External Win32 memory extension +- dzn: External Fd memory extension +- dzn: Hook up win32 semaphore import/export +- dzn: Hook up fd semaphore import/export +- docs: Update list of extensions implemented by dzn +- glsl: Delete dead intrinsics +- microsoft/compiler: Better and simpler bitcast reduction +- dzn: Add a no-bindless debug flag +- dzn: Fix inverted assert +- dzn: Partial revert of 8887852d +- dzn: Don't expose copy queues +- dzn: Fix src/dest confusion for some non-bindless descriptor copies +- wsi/win32: Handle acquiring an image while one is already acquired +- nir_lower_returns: Optimize phis before beginning the pass +- nir: Add undef phi srcs when adding successors +- radv: Don't run opt_remove_phis before lower_returns +- dxil: Don't run opt_remove_phis before lower_returns +- ci/windows: Update WARP to 1.0.7 +- microsoft/compiler: Enable emitting type info for textures with <4 comps +- microsoft/compiler: Add a pass to assign image formats based on number of components +- spirv2dxil: Assign formats to image vars before lowering to bindless +- microsoft/compiler: Use image formats to determine texture types +- ci/windows: Update WARP to 1.0.7.1 +- nir_opt_algebraic: Don't shrink 64-bit bitwise ops if pack_split is going to be lowered +- nir: Add preserve_mediump as a shader compiler option +- microsoft/compiler: Always set support_16bit_alu +- microsoft/compiler: Handle mediump +- spirv2dxil: Enable mediump +- dzn: Don't lower away mediump +- microsoft/compiler: Fix the int->uint pass for arrayed I/O +- microsoft/compiler: Fix usage of type var in semantic asserts +- microsoft/compiler: Viewport/layer as input to GS/HS needs to set feature bit +- d3d12: Support PIPE_CAP_VS_LAYER_VIEWPORT +- dzn: Don't create D3D objects for secondary command buffers +- dzn: Fix incremental binding of VBs +- d3d12: Fully initialize UAV desc for null SSBOs +- dzn: Don't support VK R4G4B4A4_UNORM_PACK16 unless we have B4G4R4A4 +- nir_opt_constant_folding: Fix nir_deref_path leak +- nir: Add is_null_constant to nir_constant +- vtn: Set is_null_constant +- nir_split_struct_vars: Support more modes and constant initializers +- nir: Allow atomics as non-complex uses for var-splitting passes +- nir_lower_ubo_vec4: Delete an invalid assert +- nir_lower_mem_access_bit_sizes: Add a bit_size input to the callback +- nir_lower_mem_access_bit_sizes: Move options into a struct +- nir_lower_mem_access_bit_sizes: Support unaligned stores via a pair of atomics +- nir: Fix constant expression for unpack_64_4x16 +- nir: Optimize unpacking 16 bit values that were originally packed +- microsoft/clc: Try harder to optimize memcpys before lowering them +- microsoft/clc: Fix progress reporting for some lowering +- microsoft/compiler: Support vec/struct const vals +- microsoft/compiler: Improvements to constant -> shader_temp pass used for CL +- microsoft/compiler: Add some more lowering passes for derefs +- microsoft/compiler: Emit const accesses as load_deref +- microsoft/compiler: Use mem_constant instead of shader_temp for consts +- microsoft/compiler: Un-lower shared/scratch to derefs +- spirv2dxil: Don't lower shared/temp to explicit I/O +- microsoft/compiler: Support load_ubo_vec4 +- dxil: Don't generate load_ubo_dxil directly +- dxil: Delete load_ubo_dxil intrinsic +- microsoft/compiler: Don't lower bit sizes for movs +- microsoft/compiler: Don't over-align raw buffer load/store intrinsics +- dxil: Remove custom SSBO lowering +- nir_lower_returns: Mark assert-only var as ASSERTED +- dzn: Ignore export access parameters +- dzn: Inline D3D12 device creation in physical device creation +- dzn: Use common GetPhysicalDeviceFeatures2 +- dzn: Remove dynamic check for block-compressed support +- dzn: Fix multisample counts in device limits +- dzn: Align placed footprints used when copying linear <-> optimal for BC formats +- dzn: VK_EXT_external_memory_host +- radv: Fix label name +- microsoft/clc: Fix usage of nir_builder_at +- ci/windows: Re-enable Windows builds +- d3d12: Fix indexing of local_reference_state + +Jiadong Zhu (1): + +- ac: enable SHADOW_GLOBAL_CONFIG for preemptible ib + +Jianxun Zhang (8): + +- iris: Fix memory alignment when importing dmabuf (GFX12.5) +- include/uapi: Update drm_fourcc.h from drm kernel +- intel/isl: Add MTL RC CCS modifier into modifier info +- iris: Support I915_FORMAT_MOD_4_TILED_MTL_RC_CCS modifier +- intel/isl: Add MTL RC CCS CC modifier into modifier info +- iris: Support I915_FORMAT_MOD_4_TILED_MTL_RC_CCS_CC modifier +- intel/isl: Add MTL MC CCS modifier into modifier info +- iris: Support MTL modifier MC_CCS + +Jonathan Gray (1): + +- intel/dev: remove dg2 0x5698 pci id + +Jordan Justen (23): + +- intel/compiler/gfx12.5+: Lower 64-bit cluster_broadcast with 32-bit ops +- mesa/main: Exit early when trying to create an unsupported context API +- iris: Flush untyped dataport cache when HDC flush is requested on compute +- iris: Flush untyped dataport cache DC flush is requested on compute +- anv: Clear untyped dataport cache flush bit if not in GPGPU mode +- anv: Flush untyped dataport cache when HDC flush is requested on compute +- anv: Flush untyped dataport cache DC flush is requested on compute +- intel/devinfo: Add has_set_pat_uapi +- intel/devinfo: Define PAT indices used on MTL +- iris/bufmgr: Add iris_pat_index_for_bo_flags() +- iris/bufmgr: Skip bucket allocation if not using writeback cache PAT index +- iris: Map aux-map with WC on MTL+ (has_set_pat_uapi) +- drm-uapi/i915_drm.h: Update from drm-next (2023-06-09) +- iris: Use set PAT extension on BO creation for MTL +- anv: Use set PAT extension on BO creation for MTL +- intel/devinfo/i915: Set has_set_pat_uapi for MTL+ +- intel/genxml: Add COMPCS0 aux-table registers +- anv: Program compute aux-map base address during queue init +- anv: Use correct CCS0 aux-map register offset in pipe flush +- isl: Add ISL_SURF_USAGE_STREAM_OUT_BIT +- anv,iris,hasvk: Use ISL_SURF_USAGE_STREAM_OUT_BIT for setting stream-out MOCS +- isl/dev: Add uncached MOCS value +- isl: Set MOCS to uncached for MTL stream-out + +Joshua Ashton (8): + +- radv: Do not enable robustness for push constants with robustBufferAccess2 +- radv: Refactor buffer robustness to an enum +- radv: Rename radv_nir_compiler_options::robust_buffer_access to robust_buffer_access_llvm +- radv: Split and move buffer robustness to shader key +- radv: Rename radv_required_subgroup_info to radv_shader_stage_key +- radv: Implement VK_EXT_pipeline_robustness +- radv: Advertise VK_EXT_pipeline_robustness +- radv: Remove unused pipeline param from radv_generate_pipeline_key + +Joshua Watt (2): + +- drm-shim: Set file type in readdir() +- drm-shim: Use anonymous file for file override + +José Fonseca (2): + +- wgl: Fix unintentional assignment on assert. +- wgl: Remove needless \`if (1) { ... }`. + +José Roberto de Souza (29): + +- iris: Move i915 batch destroy logic to iris_i915_destroy_batch() +- iris: Initialize batch screen in iris_init_batch() +- iris: Move iris_batch i915 specific variables to union +- iris: Create, destroy and replace Xe engines +- iris: Implement batch_check_for_reset() in Xe kmd backend +- iris: Set priority to Xe engines +- iris: Fix close of exported bos +- intel/common: Add gt_id to intel_engine_class +- iris: Implement batch_submit() in Xe kmd backend +- iris: Fix vm bind of imported bos from other GPUs +- build: Add Iris and ANV to ARM's auto-generated drivers +- anv: Take into consideration physical device max heap size to set maxStorageBufferRange +- iris: Allow shared scanout buffer to be placed in smem as well +- iris: Add a function to return allocated bo mmap mode +- iris: Add function to return mmap mode for userptr bos +- iris: Add function to return mmap mode for aux map +- anv: Set memory types supported by Xe KMD +- anv: Fix ANV_BO_ALLOC_NO_LOCAL_MEM flag +- anv: Nuke ANV_BO_ALLOC_WRITE_COMBINE +- iris: Fix return of xe_batch_submit() when exec fails +- iris: Replace aperture_bytes by sram size in iris_resource_create_for_image() for PIPE_USAGE_STAGING +- intel: Fix support of kernel versions without DRM_I915_QUERY_ENGINE_INFO +- iris: Attach a dma-buf to bo flink +- iris: Implement external object implicit syncronization for Xe kmd +- anv: Fix compute maximum number of threads value +- anv: Fix some mismatches of canonical and regular addresses around anv_bo_vma_alloc_or_close() +- anv: Drop unnecessary intel_canonical_address() call around anv_address_physical() +- anv: Drop unnecessary intel_canonical_address() calls around bo->offset +- iris: Convert slab address to canonical + +Juan A. Suarez Romero (34): + +- v3d: set depth compare function correctly +- v3d: use primitive type to get stream output offset +- v3d/ci: annotate failure +- v3dv/ci: rename waiver test +- v3d: add support for ARB_texture_cube_map_array +- v3d/ci: enable glsl 1.30 and 1.40 piglit tests +- v3d: apply 1D texture miplevel alignment in arrays +- v3d/ci: update neverball-v2 trace reference +- vc4/ci: skip unsupported test versions +- vc4/ci: disable VC4 jobs +- v3d: add per hw-version caller macro +- v3d: upgrade V3D 4.1 to 4.2 version +- v3d: apply proper clamping when setting up RT +- v3d/ci: annotate failures +- vc4/ci: re-enable VC4 testing +- v3d: delay offset/counter values with primitive restart +- v3d/ci: run GPU piglit profile +- v3d/ci: make traces test mandatory +- v3d: enable NIR compact arrays +- vc4: set blit mask correctly +- vc4: call blit paths in chain +- vc4: allow tile-based blit for Z/S +- vc4: add specific stencil blit path +- v3d/v3dv/ci: adjust job fractions +- v3dv/vc4/ci: update expected results +- v3d/ci: update traces +- v3d: Z/S blit require Z/S formats +- broadcom/ci: update expected results +- v3d: handle samplerExternalOES +- broadcom/ci: update expected results +- gallium/util: fix color clamp for alpha-only formats +- v3d: clear alpha-only as red-only +- vc4/v3d/ci: update expected results +- v3d/ci: add new flake + +Julia Tatz (5): + +- zink: Implement PIPE_CAP_OPENCL_INTEGER_FUNCTIONS and PIPE_CAP_INTEGER_MULTIPLY_32X16. +- zink: Implement PIPE_CAP_RESOURCE_FROM_USER_MEMORY +- zink: fix layout(local_size_variable) for vk1.3+ +- zink/ci: update expected results +- aux/trace: fix (u)int dump + +Julia Zhang (1): + +- virgl: remove check of VIRGL_CAP_V2_UNTYPED_RESOURCE + +Juston Li (24): + +- venus: use pipelineCacheUUID for shader cache id +- venus: filter out queue familes with exclusive sparse binding support +- venus: add helper function support for VkBindSparseInfo +- venus: add back sparse binding support +- venus: enable sparse binding features +- venus: enable sparse binding properties +- venus: sync to latest protocol header from v1.3.252 +- venus: sync protocol for multiple extensions for zink +- venus: enable VK_EXT_non_seamless_cube_map +- venus: enable VK_EXT_dynamic_rendering_unused_attachments +- venus: enable VK_KHR_shader_clock +- venus: enable VK_EXT_border_color_swizzle +- venus: enable VK_EXT_fragment_shader_interlock +- venus: enable VK_EXT_shader_subgroup_ballot +- venus: enable VK_EXT_color_write_enable +- docs: venus: update extension support +- radv: fix incorrect size for primitives generated query +- venus: factor out flush barrier cmd +- venus: expose vn_feedback_buffer_create() +- venus: add query pool feedback cmds +- venus: track viewMask +- venus: track render pass +- venus: batch query feedback and defer until after render pass +- venus: use feedback for vkGetQueryPoolResults + +Karmjit Mahil (42): + +- pvr: Add missing includes in pvr_common.h +- pvr: Implement vkCmdUpdateBuffer(). +- pvr: Implement simple internal format v2 transfer paths. +- pvr: Add deferred RTA clears for cores without gs_rta_support. +- pvr: Finish pvr_perform_start_of_render_attachment_clear(). +- pvr: Collect vertex input data and fill info struct. +- pvr: Fix a comment in the PDS code +- pvr: Fix typo in PDS function name +- pvr: Add handling for missing entries in pvr_setup_vertex_buffers() +- pvr: Handle special built-in variable loading in vertex shader +- pvr: Add PVR_DW_TO_BYTES() +- pvr: Fix pvr_csb_bake() list return. +- pvr: Change push_constants_shader_stages to type pvr_stage_allocation +- pvr: Fix static assert check +- pvr: Fix unaligned VDMCTRL_PDS_STATE1 data address +- pvr: Don't advertise S8_UINT support +- pvr: Fix cs corruption in pvr_pack_clear_vdm_state() +- pvr: Add missing NULL checks in some vkDestroy...() functions +- pvr: Use original binding numbers instead of reassigning +- pvr: Remove custom status in command buffer +- pvr: Fix missing invalidation of the command buffer +- pvr: Fix possible allocation of 0 size +- pvr: Fix vk_free() in vkCreateRenderPass2() error path +- pvr: Use the suballocator for queries +- pvr: Add pvrsrvkm sync prim set bridge call +- pvr: Move pvrsrv sync prim code into new pvr_srv_sync_prim.{c,h} +- pvr: Use idalloc as the allocator for sync prims +- pvr: Handle barrier load and store flags. +- pvr: Fix typo causing seg faults copying immutable samplers +- pvr: Fix draw indirect page faults due to missing index list buffer +- pvr: Rename temps_count to pds_temps_count +- pvr: Fix PDS temps allocation on fragment stage +- pvr: HWRT creation simplifications. +- pvr: Dedup a check with pvr_is_render_area_tile_aligned() +- pvr: Remove outdated finishme +- pvr: Fix seg fault on empty descriptor set +- pvr: Fix dynamic offset patching +- pvr: Fix csb control stream extension +- pvr: Fix missing BITFIELD_BIT for winsys frag job flag +- pvr: Change winsys flag defines to bitfields +- pvr: Setup ZLS depth and stencil load/store separately +- docs: Add inital PowerVR driver documentation + +Karol Herbst (127): + +- rusticl: rework CLVec helper function to calculate bounds +- rusticl/mem: fix Mem::copy_rect +- rusticl/mem: replace buffer_offset_size with CLVec::calc_offset_size +- gallium: correctly name the flags of svm_migrate +- rusticl/context: add helper to get the max mem alloc size for all devices +- rusticl/memory: Rework mapping of memory located in system RAM +- rusticl/mem: add get_parent helper +- rusticl: add support for fine-grained system SVM +- nv50/ir: ignore CL system values +- nouveau: allow to enable SVM without having to enable CL +- nouveau: nouveau_copy_buffer can deal with user_ptrs just fine +- rusticl/event: drop work item before updating status +- rusticl: add create_pipe_box to better deal with pipe_box restrictions +- rusticl/mem: more region and origin validation +- radeonsi: lower mul_high +- ac/llvm: support shifts on 16 bit vec2 +- rusticl: don't set size_t-is-usize for >=bindgen-0.65 +- rusticl/device: improve advertisement of fp64 support +- rusticl/platform: make the initialization more explicit +- rusticl/platform: extract env variable parsing from Platform::init +- rusticl/platform: add RUSTICL_FEATURES boilerplate +- rusticl/device: allow enablement of fp64 via RUSTICL_FEATURES +- rusticl/program: rework dynamic Program state +- rusticl/program: use if let to get rid of an unwrap in build +- clc: free kernel args in clc_free_kernels_info +- rusticl/nir: finish blob after serializing +- nvc0: do not randomly emit fences. +- nv50/ir: Use unified atomics +- Reviewed-by: Nora Allen rusticl/platform: make the extension array a static +- rusticl/device: use PLATFORM_EXTENSIONS as a template for filling extensions +- rusticl/platform: advertise byte_addressable_store +- rusticl/device: split add_ext in fill_extensions +- rusticl: explicitly state supported SPIR-V extensions +- rusticl/platform: generate extension constants via macro +- rusticl/spirv: skip printing info messages +- rusticl/device: limit MAX_PARAMETER_SIZE to 32k +- rusticl/device: set preferred vector size of doubles if fp64 is enabled +- nv50/ir: convert to scoped_barrier +- doc/rusticl: add Rust Update Policy +- rusticl: bump rust req to 1.60 +- rusticl/event: flush queues from dependencies +- ci: add and use clippy for rusticl +- rusticl: fix clippy errors on image_slice_pitch change to usize +- clc: relax spec constant validation +- rusticl: add proc macro module for generating API stubs +- rusticl/icd: make release return nothing +- rusticl/icd: use new proc macros +- ac/llvm: support vec2 on b2i16 +- ac/llvm: replace MESA_SHADER_COMPUTE checks with gl_shader_stage_is_compute +- ac/llvm: set +cumode for radeonsi +- lp: align memory for long16 CL types +- rusticl/icd: fix ReferenceCountedAPIPointer::from_ptr for NULL pointers +- rusticl/api: remove some repr(C) +- rusticl/event: ensure even status is updated in order +- docs: improve OpenCL features +- rusticl/queue: overhaul of the queue+event handling +- rusticl: enforce using unsafe blocks in unsafe functions +- nv50/ir: use override +- nv50/ir: resolve -Woverloaded-virtual=1 warnings +- clc: add commment to clc_optional_features to ensure no padding exists +- rusticl/spirv: Key optional clc features when caching. +- clc: static assert that clc_optional_features has no padding +- nouveau: eliminate busy waiting on fences +- rusticl/device: add intel usm queries DPCPP cares about +- rusticl/device: sort cl_device_info queries +- rusticl/version: use cl_version instead of cl_uint and provide a From impl +- rusticl: advertize cl_khr_extended_versioning +- docs/cl: fix whitespace issues and add missing entries +- rusticl: advertize cl_khr_spirv_no_integer_wrap_decoration +- docs/cl: improve reporting of image features +- rusticl/mem: cache the pipe_format +- rusticl/mem: fix validation of packed image formats +- rusticl/format: pass order and type to rusticl_image_format directly +- rusticl/format: extract CL format to pipe format mapping into const function +- rusticl/format: extract required format checks into const functions +- rusticl/format: drop req_for_3d_image_write_ext +- rusticl/format: add required format table for CL2.0 +- rusticl/format: document cl to pipe format mapping +- rusticl/format: move format table generation into a macro +- rusticl/format: enable all trivial to support optional image formats +- clc: fix SPIRVMessageConsumer for NULL src +- clc: allow passing custom validator options +- rusticl/program: pass our max param size along to the spirv validator +- compiler/types: fix size of padded OpenCL Structs +- rusticl/device: rename doubles to fp64 and long to int64 +- rusticl: experimental support for cl_khr_fp16 +- rusticl: add ld_args_gc_sections +- rusticl: specify which symbols to export +- rusticl: stop linking with libgalliumvl +- rusticl/device: create helper context before loading libclc +- nir/load_libclc: run some opt passes for everybody +- docs: document CLC_DEBUG +- rusticl/program: add debugging for OpenCL C compilation +- rusticl/program: add debugging option to disable SPIR-V validation +- nvc0: fix printing shaders +- nv50/ir/nir: set numBarriers if we emit an OP_BAR +- rusticl: structurize and reorder mesa binding args +- rusticl: generate bindings for build-id stuff +- rusticl/meson: extract common bindgen rust args +- rusticl/mesa: create proper build-id hash for the disk cache +- rusticl: bump bindgen requirement +- rusticl/program: skip linking compiled binaries +- docs/rusticl: mark building section as such +- docs/rusticl: add Enabling section +- docs/cl: remove cl_khr_byte_addressable_store from extension list. +- docs/cl: move vec3 support under OpenCL C 1.1 +- docs/cl: timer sync is implemented +- docs: add missing get_compute_state_info documentation +- vtn: more CL subgroups +- clc: rework optional subgroup feature +- llvmpipe: report the proper subgroup size +- gallium: add simd_sizes to pipe_compute_state_object_info +- gallium: add get_compute_state_subgroup_size +- gallium: add PIPE_COMPUTE_CAP_MAX_SUBGROUPS +- iris: implement get_compute_state_subgroup_size +- rusticl/util: add an Iterator to iterate over set bits in an integer +- rusticl/util: add div_round_up +- rusticl/device: rework subgroups to subgroup_sizes +- gallium: change PIPE_COMPUTE_CAP_SUBGROUP_SIZE to a bitfield of sizes +- rusticl: deal with compute_param returning 0 +- rusticl: support subgroups +- nvc0: backport fp helper invocation fix to 2nd gen Maxwell+ +- rusticl/kernel: silence newer clippy warning +- rusticl: Replace &Arc with &Device +- rusticl/device: make it &'static +- api/icd: drop static lifetime from \`get_ref` return type +- nvc0: initial Ada enablement + +Kenneth Graunke (16): + +- intel/compiler: UNDEF comparisons with smaller than 32-bit +- intel/compiler: UNDEF SubgroupInvocation's register +- intel/compiler: Fold constants after distributing source modifiers +- nir: Add a variant of nir_lower_int64 for float conversions only +- intel/compiler: Postpone most int64 lowering to brw_postprocess_nir +- nir: Add find_lsb lowering to nir_lower_int64. +- intel/compiler: Fix 64-bit ufind_msb, find_lsb, and bit_count +- nir: Assert that we don't shrink bit-sizes in nir_lower_bit_size() +- intel/compiler: Fix a fallthrough in components_read() for atomics +- intel/genxml: Drop Tiled Resource Mode fields +- intel: Initialize FF_MODE2 on all Gfx12 platforms +- iris: Allocate coherent buffers for resources flagged as persistent/coherent +- isl: Don't set "Enable Unorm Path in Color Pipe" on Alchemist +- intel/genxml: Fix gen_sort_tags.py to handle mbz/mbo +- intel/genxml: Update RENDER_SURFACE_STATE Fields +- iris: Re-emit 3DSTATE_DS for each primitive (workaround 14019750404) + +Kiskae (1): + +- vulkan/wsi: check for dri3 buffer initialization failure + +Konrad Dybcio (2): + +- freedreno: Add some A6/7xx registers +- freedreno: Partially decode CP_PROTECT_CNTL + +Konstantin Kharlamov (1): + +- loader/dri3: temporarily work around a crash when front is NULL + +Konstantin Seurer (127): + +- nir/lower_fp16_casts: Fix SSA dominance +- nir/lower_io: Emit less iadd(x, 0) +- nir: Make rq_load committed src an index +- radv: Stop running constant folding during ray query lowering +- radv/ci: Test ray tracing pipelines +- gallium/nir: Handle unified atomics in nir_to_tgsi_info +- nir/inline_uniforms: Handle num_components > 1 +- nir/lower_shader_calls: Remat derefs earlier +- radv: Stop using radv_get_int_debug_option +- treewide: Add a .clang-format file +- amd: Use the Mesa base style +- asahi: Use the Mesa base style +- freedreno: Use the Mesa base style +- d3d12: Use the Mesa base style +- i915: Use the Mesa base style +- r600/sfn: Use the Mesa base style +- panfrost: Use the Mesa base style +- util/perf: Use the Mesa base style +- venus: Use the Mesa base style +- asahi: Reformat using the new style +- panfrost: Reformat using the new style +- gallivm: Fix gather/scatter types for newer llvm +- radv/rt: Fix pipeline libraries +- gallivm: Fix anisotropic sampling with num_mips=1 +- gallivm: Cast read_first_invocation source to an int +- llvmpipe: refactor out the pipe->lp_jit structure fillers. +- llvmpipe: Add lp_storage_image_format_supported +- llvmpipe: Add lp_storage_render_image_format_supported +- gallivm: Add lp_build_nir_sample_key +- gallivm: Add lp_img_op_from_intrinsic +- gallivm: Handle invalid image format/op combinations +- gallivm: Zero initialize param structs +- radv/rt: Do not guard the raygen shader +- radv/rt: Clear NIR metadata after lowering the ABI +- aco/rt: Do not initialize the next shader addr +- radv/ci: Test ray tracing on vkd3d-proton +- radv/rt: Stop forcing wave32 by setting compute_subgroup_size +- Revert "radv: Enable ray tracing pipelines by default" +- radv/rt: Enable RT pipelines on GFX10_3+ excluding vangogh +- radv: Move the shader type to radv_shader_info +- radv: Adjust the traversal shader description +- radv: Use get_shader_from_executable_index for executable properties +- radv: Implement executable properties for ray tracing stages +- radv: Use _mesa_shader_stage_to_string for executable name +- radv/rt: Store the prolog outside the shaders array +- radv: Call radv_pipeline_init_scratch per shader +- meson: Add a xcb-keysyms dependency +- vulkan: Common trace capturing infrastructure +- radv: Add radv_trace_mode +- vulkan/wsi/x11: Capture traces using a hotkey +- radv/rra: Use common trace trigger +- radv/rgp: Use common trace trigger +- vulkan/rmv,radv: Use common trace trigger +- docs: Update envvars used for tracing +- amd: Use nir\_ instead of nir_build\_ helpers +- microsoft: Use nir\_ instead of nir_build\_ helpers +- intel: Use nir\_ instead of nir_build\_ helpers +- freedreno: Use nir\_ instead of nir_build\_ helpers +- vtn: Use nir\_ instead of nir_build\_ helpers +- nir: Use nir\_ instead of nir_build\_ helpers +- nir/builder_opcodes: Remove nir_build\_ prefixed helpers +- util: Do not include immintrin.h in half_float.h +- radv/rt: Fix caching non-recursive stages +- radv/rt: Hash stages using radv_hash_shaders +- llvmpipe: Add BDA jit type helpers +- gallivm: Add missing includes +- gallivm: Add lp_descriptor struct +- gallivm: Expose lp_build_sample_soa_code +- llvmpipe: Add lp_build_sampler_soa_dynamic_state +- llvmpipe: Add lp_build_image_soa_dynamic_state +- gallivm: Add LP_IMG_OP_COUNT +- gallivm: Expose LP_MAX_TEX_FUNC_ARGS +- llvmpipe: Add LP_TOTAL_IMAGE_OP_COUNT +- gallivm: Expose lp_build_texel_type +- gallivm: Propagate vulkan resources +- gallivm: Clamp the texel buffer size +- llvmpipe: Pre compile sample functions +- gallivm: Add a function for loading vulkan descriptors +- gallivm: Implement vulkan UBOs +- gallivm: Implement vulkan SSBOs +- gallivm: Implement vulkan textures +- gallivm: Implement vulkan images +- llvmpipe: Disable the linear path when running vulkan +- lavapipe: Include llvmpipe +- lavapipe: Lower more texture OPs +- lavapipe: Make pipeline_lock generic for accessing the queue +- lavapipe: Rework descriptor handling +- lavapipe: Lower non uniform access +- lavapipe: EXT_descriptor_indexing +- llvmpipe: Use lp_jit_buffer_from_pipe_const in setup +- lavapipe: Make shader compilation thread safe +- zink: Increase ZINK_FBFETCH_DESCRIPTOR_SIZE to 280 +- zink/ci: Update lavapipe expectations +- venus/ci: Update fails +- lavapipe/ci: Update CI expectations for new extensions +- llvmpipe/ci: Update expectations +- nir: Add nir_builder_at +- radv: Use nir_builder_at +- asahi: Use nir_builder_at +- v3d: Use nir_builder_at +- glsl: Use nir_builder_at +- nir: Use nir_builder_at +- spirv: Use nir_builder_at +- freedreno: Use nir_builder_at +- gallium,st: Use nir_builder_at +- crocus: Use nir_builder_at +- etnaviv: Use nir_builder_at +- r600: Use nir_builder_at +- radeonsi: Use nir_builder_at +- vc4: Use nir_builder_at +- zink: Use nir_builder_at +- lavapipe: Use nir_builder_at +- microsoft: Use nir_builder_at +- panfrost: Use nir_builder_at +- intel: Use nir_builder_at +- nir/opt_dead_cf: Handle if statements ending in a jump correctly +- nir/builder_opcodes: Do not generate empty intrinsic indices +- amd: Move ac_hw_stage to its own file +- gallivm: Fix atomic_global types +- lavapipe: Set the descriptor count to what vkd3d-proton requires +- llvmpipe: Allow comparison sampling for float formats +- llvmpipe: Allocate more dummy sample functions for FORMAT_NONE +- llvmpipe,lavapipe: Relayout lp_descriptor +- lavapipe: Always advertise formatless storage image OPs +- nir/lower_shader_calls: Remat derefs after shader calls +- nir/opt_dead_cf: Run dead_cf_block while it makes progress +- nir/opt_dead_cf: Clarify comment + +Kurt Kartaltepe (1): + +- drirc: Set limit_trig_input_range option for Nier games + +Leo Liu (5): + +- radeonsi: create a new context for transcode with multiple video engines +- radeonsi/vcn: AV1 skip the redundant bs resize +- radeonsi: Remove redundant vcn_decode from info +- amd: Add vcn ip version info +- radeonsi: Use vcn version instead of CHIP family for VCNs + +Lina Versace (2): + +- venus: Advertise 1.3 in ICD file +- venus: Fix detection of push descriptor set + +LingMan (4): + +- rusticl: core: stop using cl_prop from the api module +- rusticl: drop CLProp implementation for String +- rusticl: drop cl_prop_for_type macro +- rusticl: fix UB in CLProp machinery + +Lionel Landwerlin (171): + +- docs: add missing MESA_VK_WSI_HEADLESS_SWAPCHAIN variable +- vulkan/runtime: discard unused graphics stages in libraries +- intel/vec4: force exec_all on float control instruction +- anv: enable blorp query reset for performance queries +- vulkan/overlay: deal with unknown pNext structures +- isl: don't set inconsistent fields for depth when using stencil only +- anv: introduce a base graphics pipeline object +- anv: move force shading rate writes checks +- anv: make input attachments available through bindless +- anv: move preprocessing of NIR right before compilation +- anv: add dynamic buffer offsets support with independent sets +- anv: implement VK_EXT_graphics_pipeline_library +- anv: Work around the spec question about pipeline feedback vs GPL. +- isl: fix a number of errors on storage format support on Gfx9/12.5 +- intel/nir: add options to storage image lowering +- anv: drop lowered storage images code +- anv: enable shaderStorageImageReadWithoutFormat on Gfx12.5+ +- anv: rework Wa_14017076903 to only apply with occlusion queries +- intel/tools: add ability to dump out raw kernels data +- nir/divergence: add missing load_global_constant_* intrinsics +- anv: fix anv_nir_lower_ubo_loads pass +- anv: enable shaderUniformBufferArrayNonUniformIndexing +- intel/fs: fix per vertex input clamping +- nir/lower_non_uniform_access: add get_ssbo_size handling +- intel/compiler: make uses_pos_offset a tri-state +- vulkan: bump headers to 1.3.249 +- spirv: update to latest headers +- spirv/nir: wire ray interection triangle position fetch +- intel/nir/rt: use a single load for instance leaf loading +- intel/nir/rt: wire position fetch intrinsic +- anv: implement VK_KHR_ray_tracing_position_fetch +- intel/fs: fix scheduling of HALT instructions +- anv: remove 48bit address space checks +- anv: avoid hardcoding instruction VA constant in shaders +- anv: link anv_bo to its VMA heap +- anv: make internal address space allocation more dynamic +- anv: increase instruction heap to 2Gb +- intel/fs: reduce register usage for relocated constants +- intel: enable protected context creation along with engines +- Revert "intel/compiler: make uses_pos_offset a tri-state" +- anv: fixup workaround 16011411144 +- intel/mi_builder: fixup tests for newer kernel uAPI +- intel: switch over to unified atomics +- spirv: fix argument to ray query intrinsic +- intel/devinfo: printout on stdout +- intel/devinfo: allow -p to take a pci-id in hexa +- intel/devinfo: call intel_device_info_init_was only once +- anv: put private binding BOs into execlists +- anv: mark images compressed for untracked layout/access +- gitlab-ci: add capture for i915 error state +- anv: defer binding table block allocation to when necessary +- anv: assume context isolation support +- anv: fix push descriptor deferred surface state packing +- intel/fs: fix size_read() for LOAD_PAYLOAD +- anv: move timestamp vfunc initialization to genX code +- anv: use COMPUTE_WALKER post sync field to track compute work +- iris: use COMPUTE_WALKER post sync field to track compute work +- intel/fs: make tcs input_vertices dynamic +- anv: implement EDS2.extendedDynamicState2PatchControlPoints +- iris: rework Wa_14017076903 to only apply with occlusion queries +- intel: add alignment helper for aux map +- iris: add a comment about aux-tt alignment requirements +- anv: update aux-tt alignment requirements for MTL +- intel: reduce minimum memory alignment on Gfx12.5 +- anv: further reduce pool alignments +- anv: opportunistically align VMA to 2Mb +- anv: update internal address space to have 4Gb of dynamic state +- anv: fix push range for descriptor offsets +- intel/fs: reuse descriptor helper +- intel/fs: lower get_buffer_size like other logical sends +- nir/lower_shader_calls: add ability to force remat of instructions +- nir: add a new intrinsic to describe resources accessed on intel +- nir: teach nir_chase_binding about resource_intel +- nir/opt_gcm: allow resource_intel to be moved anywhere +- intel/fs: add a pass to move resource_intel closer to user +- intel/fs: teach ubo range analysis pass about resource_intel +- intel/fs: keep track of new resource_intel information +- intel/fs: enable SSBO accesses through the bindless heap +- intel/fs: enable UBO accesses through bindless heap +- intel/fs: enable get_buffer_size on bindless heap +- intel/fs: enable extended bindless surface offset +- intel/fs: enable bindless sampler state offsets +- intel/fs: enable uniform block accesses through bindless heap +- intel/fs: try to rematerialize surface computation code +- anv: remove unused define +- anv: fix null descriptor handling with A64 messages +- anv: remove incorrect ifdef +- anv: bail flush_gfx_state when not gfx push constant is dirty +- anv: track pipeline in anv_cmd_pipeline_state +- anv: move pipeline active_stages to common structure +- anv: increase workaround BO so that we can hold a full 4Kb page of 0s +- anv: toggle extended bindless surface state on Gfx12.5+ +- docs/anv: some binding table explanations +- anv: add an option for using indirect descriptors +- anv: introduce a new descriptor set layout type +- anv: create a pool for indirect descriptors +- anv: reduce push constant size for descriptor sets +- anv: new structure to hold surface states +- anv: add a pass to partially lower resource_intel +- nir: expose a couple of address format add helpers +- anv: bound load descriptor mem better +- anv: prepare image/buffer views for non indirect descriptors +- anv: add support for direct descriptor in allocation/writes +- anv: add helpers to build pipeline bindings +- anv: handle null surface in the binding table with direct descriptors +- anv: factor out dynamic buffer bti emission +- anv: implement binding table emission for direct descriptors +- anv: simplify ycbcr bti computations +- anv: track descriptor data size +- anv: add direct descriptor support to apply_layout +- anv: bring back the max number of sets to 8 +- anv: descriptor binding for direct descriptors +- anv: ensure descriptor addresses are used with bindless stages +- anv: enable direct descriptors on platforms with extended bindless offset +- anv: add support for VK_EXT_dynamic_rendering_unused_attachments +- anv: remove unused functions +- intel/fs: fix a couple of descriptor mistakes +- intel/stub_gpu: add an option to launch valgrind +- intel/fs: fix pull-constant-load prior to gfx7 +- anv: allow binding tables allocations on compute only queues +- intel/nir: switch ray query state tracking to local variables uint16_t +- anv: add query tracepoints +- anv: deal with unsupported VkImageFormatListCreateInfo::pViewFormats +- anv: report max simd width only once for fragment shaders +- anv: always report all pipeline stats regardless of stages +- anv: only disable mesh when enabled at the VkDevice level +- anv: disable mesh/task for generated draws +- anv: fix incorrect batch for 3DSTATE_CONSTANT_ALL emission +- anv: limit ANV_PIPE_RENDER_TARGET_BUFFER_WRITES to blorp operations using 3D +- anv: factor out generation kernel dispatch into helper +- anv: add support for simple internal compute shaders +- anv: generalize internal kernel concept +- anv: add shaders for copying query results +- intel/ds: add query count in query tracepoints +- anv: enable CmdCopyQueryPoolResults to use shader for copies +- intel/fs: fix bindless/shared surface mistake +- intel/fs: print identation for control flow +- intel/fs: avoid reusing the VGRF for uniform load_ubo +- nir: add a new ubo uniform loading intrinsic for intel +- intel/fs: make use of load_ubo_uniform_block_intel +- nir: add a load_global_constant uniform intel variant +- intel/fs: handle load_global_constant_uniform_block_intel +- anv: avoid private buffer allocations in vkGetDeviceImageMemoryRequirementsKHR +- anv: add missing query clear flush for acceleration structure queries +- anv: track buffer writes from shaders for query results writes +- anv: change the way we clear pending query bits +- anv: fix pending query bits for compute only command buffers +- anv: tracking query buffer writes & query clears separately +- anv: switch copy query results with shaders from semaphore waits to flushes +- vulkan: registry/headers bump to 1.3.254 +- vulkan/runtime: add support for EXT_depth_bias_control +- anv: add VK_EXT_depth_bias_control support +- isl: assert on gfx6 condition that should not be met +- isl: assert on gfx7 condition that should not be met +- isl: assert on gfx8 condition that should not be met +- isl: add surface creation reporting mechanism +- anv: align buffers to a cache line +- anv: fix utrace batch allocation +- genxml: enable decoding on compute engine +- intel/aubinator_error_decode: add ccs support +- anv: look into batch bo reloc list looking for BOs to decode +- anv: implement storage image depth query using descriptor buffer read +- Revert "isl: Set Depth to array len for 3D storage images" +- docs/features: update anv entries +- intel/fs: disable coarse pixel shader with interpolater messages at sample +- nir/opt_shrink_vectors: enable sparse intrinsics shrinking +- docs/features: add more missing extensions +- docs/features: add hasvk entries +- zink: update profile vulkan version requirements +- zink: drop linear D32_SFLOAT_S8_UINT requirement +- anv: fix utrace signaling with Xe + +Liviu Prodea (1): + +- microsoft/clc: Don't build compiler test if build-tests is false + +Lone_Wolf (3): + +- compiler/clc: Fix embedded clang headers (microsoft-clc) for LLVM 16+ +- clc: Add clangASTMatchers to fix static llvm build of microsoft-clc with LLVM 16+ +- clc: Add clang frontendhlsl module to fix build of microsoft-clc with llvm 16+ + +Luc Ma (1): + +- meson: keep Mako version checking in accord with build msg + +Luca Bacci (1): + +- Add checks for NULL dxil_validator + +Luca Weiss (1): + +- freedreno: Enable A506 + +Lucas Fryzek (6): + +- broadcom: Add support for VK_FORMAT_A2R10G10B10_UNORM_PACK32 +- broadcom: Fix slice memory allocation logic for compressed textures +- v3d: Add support for ASTC texture compression +- v3dv: Update texture padding logic to match v3d changes +- mailmap: Add Lucas Fryzek to mailmap +- gallium: Remove \`PIPE_CAP_RGB_OVERRIDE_DST_ALPHA_BLEND` + +Lucas Stach (16): + +- etnaviv: update derived state after forced commandstream flush +- etnaviv: don't flush implicit flush resources when forced +- etnaviv: rs: flush TS cache before making configuration changes +- etnaviv: rs: unconditionally flush color and depth cache before using RS +- etnaviv: optimize transfer flushes +- etnaviv: query: move sample counter manipulation into query providers +- etnaviv: query: reset sample count on begin_query +- etnaviv: query: remove incorrect comment +- etnaviv: query: correct max number of perfmon samples +- etnaviv: query: correct max number of occlusion query samples +- etnaviv: query: optimize context flushes +- mesa/st: discard whole resource when mapping drawpixels texture +- etnaviv: only emit sampler config for changed samplers +- etnaviv: move resource level dimension members to make comments line up +- etnaviv: rs: fix multisampled blits +- etnaviv: blt: fix multisampled blits + +Luigi Santivetti (12): + +- pvr: use PVR_DW_TO_BYTES for stream_link_space calculation +- pvr: add GUARD_SIZE_DEFAULT for CDM and VDM control stream links 1 and 2 +- pvr: fixup stack overflow in {start,end}_sub_cmd +- pvr: introduce suballocator for internal allocations +- pvr: switch pvr_gpu_upload_* to use pvr_bo_suballoc +- pvr: switch pvr_cmd_buffer_alloc_mem to use pvr_bo_suballoc +- pvr: switch pvr_descriptor_set_create to use pvr_bo_suballoc +- pvr: switch pvr_clear to use pvr_bo_suballoc +- pvr: switch pvr_spm to use pvr_bo_suballoc +- pvr: fixup assert in pvr_cmd_buffer_alloc_mem +- pvr: fix division by block size in blit +- pvr: fixup transfer primary sub-command list + +Luna Nova (5): + +- device_select_layer: fix inverted strcmp in device_select_find_dri_prime_tag_default (v1) +- device_select_layer: apply DRI_PRIME even if default device is > 1 to match opengl behavior +- device_select_layer: pick a default device before applying DRI_PRIME +- device_select_layer: add MESA_VK_DEVICE_SELECT_DEBUG which logs why default selection was made +- device_select_layer: log selectable devices if MESA_VK_DEVICE_SELECT_DEBUG or DRI_PRIME_DEBUG are set + +Lynne (4): + +- radv/video: reject general unsupported video formats +- radv/video: reject non-8bit H264 +- radv/video: reject unsupported hevc profiles and bit depths +- anv_video: reject decoding of unsupported profiles and formats + +M Henning (12): + +- nvc0: Use nir in nvc0_program_init_tcp_empty +- nvc0: Use nir in nvc0_blitter_make_vp +- nv50,nvc0: Use nir in nv50_blitter_make_fp +- nv50,nvc0: Stop advertising TGSI by default +- nv50,nvc0: Use ttn for tgsi shaders by default +- gallium: Add pipe_shader_state_from_nir +- nouveau/codegen: Check nir_dest_num_components +- nv50/codegen: Set lower_uniforms_to_ubo +- nouveau/nir: Set isSigned on all atomic_imax/imin +- nv50,nvc0: Free nir from blitter fp shader +- nvc0: Free blitter->vp +- nv50: Fix return type of nv50_blit_is_array + +Marcin Ślusarz (14): + +- intel: split URB space between task and mesh proportionally to entry sizes +- anv: move nir_shader_gather_info to anv_pipeline_nir_preprocess +- intel/tools: decode ACTHD printed by newer kernels +- nir: extract try_lower_id_to_index_1d +- nir: use wg id to wg idx shortcut if two dims of num_workgroups are 1 +- nir: use constant components of num_workgroups in wg id to wg idx lowering +- nir: lower num_workgroups to constants +- intel/compiler: pass num_workgroups from task to mesh shaders +- nir: add cheap shortcut for wg id to wg idx lowering +- anv,intel/compiler: enable shortcut in wg id to wg idx lowering on >= gfx12.5 +- intel/compiler: simplify reading of gl_NumWorkGroups in task/mesh +- anv: fix how NULL buffer_view is handled in anv_descriptor_set_write_buffer_view +- anv: pass anv_surface_state using a pointer +- anv: limit stack usage for anv_surface_state + +Marek Olšák (167): + +- nir: fix 2 bugs in nir_create_passthrough_tcs +- nir: lower load_barycentric_at_offset in lower_wpos_ytransform +- nir: assign IO bases in nir_lower_io_passes +- nir: skip nir_lower_io_passes for compute shaders +- nir: extend nir_opt_fragdepth to handle lowered IO +- nir: handle more opcodes in nir_lower_io_to_scalar +- nir: handle all varying slots in gl_varying_slot_name_for_stage +- nir: don't remove dead IO variables in nir_lower_io_passes for st_link_nir +- nir: rework nir_lower_color_inputs to work with lowered IO intrinsics +- nir: return a status from nir_remove_varying whether it removed the instruction +- nir: remove an obsolete comment from nir_gather_xfb_info_from_intrinsics +- nir: add next_stage parameter to nir_slot_is_sysval_output to return better info +- nir: add next_stage parameter to nir_remove_varying +- nir: set uses_wide_subgroup_intrinsics for all shader stages +- venus: fix the RHEL8 build by using syscall for gettid +- nir: rename ACCESS_STREAM_CACHE_POLICY -> ACCESS_NON_TEMPORAL and document +- nir: add/update comments for gl_access_qualifier +- ac/surface: don't expose modifiers with DCC retiling if radeon_info forbids it +- ac/gpu_info: disable display DCC on Raphael and Mendocino to improve power usage +- radeon: add radeon_info parameter into radeon_winsys::surface_init +- radeonsi: do AMD_DEBUG=nodisplaydcc differently to also remove modifiers +- aco: don't treat ACCESS_NON_READABLE as ACCESS_COHERENT +- ac/llvm: don't treat ACCESS_NON_READABLE as ACCESS_COHERENT +- ac/llvm: rewrite and unify how GLC, DLC, SLC are set +- nir/lower_io: don't renumber VS inputs when not called from a linker +- ac/surface: fix address calculation for large images by using uint64_t +- radv: fix sparse image address calculation for large images by using uint64_t +- radv: fix SDMA image address calculation for large images by using uint64_t +- radeonsi: fix SDMA image address calculation for large images by using uint64_t +- radeonsi: fix image address calculation for large images by using uint64_t +- radeonsi: fix sparse image address calculation for large images by using uint64_t +- radeonsi: fix image size calculation in fast clear +- ac/surface: clean up and move the PIPE_CONFIG helper to ac_surface.c +- ac/surface: define LINEAR_PITCH_ALIGNMENT +- ac/surface: validate overridden pitch for all chips +- ac/surface: fix overridden linear pitch for CPU access +- ac/surface: add ac_surf_config::is_array +- amd/registers: update pitch definitions in descriptors +- mesa: fix a VBO buffer reference leak in _mesa_bind_vertex_buffer +- ac,radeonsi,winsyses: switch to SPDX-License-Identifier: MIT +- winsys/radeon: set has_image_opcodes to unbreak gfx6-7 +- winsys/radeon: fix the scratch buffer on gfx6-7 +- winsys/radeon: set more radeon_info fields +- ac/gpu_info: give has_msaa_sample_loc_bug a more accurate name +- ac/surface: move CB format translation helpers here +- ac/surface: move determing ADDR_FMT_* into a helper function +- ac/llvm: clean up translation of nir_intrinsic_load_invocation_id +- ac/llvm: clean up visit_load_local_invocation_index and visit_load_subgroup_id +- ac/llvm: use LLVM 0/1 constants from ac_llvm_context instead of LLVMConstInt +- radeonsi/gfx11: fix alpha-to-coverage with blending +- radeonsi: reorder code in si_texture_create_object as preparation for the future +- radeonsi: cosmetic changes in si_shader.h +- radeonsi: remove the gl_SampleMask FS output if MSAA is disabled +- radeonsi: don't enable WGP_MODE because of high cost of workgroup mem coherency +- radeonsi: move emitting draws states out of si_emit_all_states +- radeonsi/gfx11: use DISABLE_FOR_AUTO_INDEX to disable non-indexed prim restart +- radeonsi: reduce the supported compute grid size +- radeonsi: update test results and flakes +- radeonsi: re-enable fp16_rtz for compute blits to fix PBO tests on gfx11 +- amd/addrlib: switch the license to the SPDX identifier MIT +- amd/addrlib: add ADDR_FMT_BG_RG_16_16_16_16 +- ac/surface: fix is_linear for stencil-only surfaces +- ac/nir: handle DEPTH as PITCH in ac_nir_lower_resinfo +- radeonsi: implement setting a custom pitch to any multiple of 256B on gfx10.3+ +- radv: implement setting a custom pitch to any multiple of 256B on gfx10.3+ +- ac/surface: relax custom pitch requirements to any multiple of 256B on gfx10.3+ +- ac/surface: fix R32G3B32 image format regression for gfx6-8 +- ac/nir/ngg: always use load_initial_edgeflags_amd, choose the value in drivers +- amd: add radeon_info* into ac_llvm_context and radv_nir_compiler_options +- radeonsi: define si_shader_io_get_unique_index() values as SI_UNIQUE_SLOT_* +- radeonsi: remove gl_BackColor VS outputs on demand if color_two_side is disabled +- radeonsi: export non-zero edgeflags for GS and tess +- radeonsi/gfx11: extend DB_Z_INFO.NUM_SAMPLES programming to > GFX11 +- radeonsi: print shader-db stats with AMD_DEBUG=vs,ps,stats +- radeonsi: use nir_lower_alu_to_scalar correctly +- radeonsi: remove a useless depth texture function call in a fast color clear +- radeonsi: add a gfx11 version of si_decompress_textures, add assertions < GFX11 +- radeonsi: remove RADEON_FLAG_MALL_NOALLOC due to no use +- radeonsi: completely rewrite how VGT_SHADER_STAGES_EN is set +- radeonsi: unduplicate si_translate_format_to_hw +- radeonsi: decompress DCC for SDMA if we're really going to use SDMA +- radeonsi: increase SDMA gfx9+ limits +- radeonsi: split tracked_regs masks into context registers and other registers +- radeonsi: reorder and comment tracked registers +- radeonsi: move PA_CL_NGG_CNTL emission into rasterizer state +- radeonsi: always set sample locations even for 1x MSAA for simplicity +- radeonsi: adjust 16x EQAA sample locs to make PA_SU_PRIM_FILTER_CNTL immutable +- radeonsi: move PA_SU_SMALL_PRIM_FILTER_CNTL to the preamble when possible +- radeonsi: merge si_emit_msaa_sample_locs with si_emit_sample_locations +- radeonsi: rename the msaa_sample_locs state to sample locations +- radeonsi: optimize no-op primitive restart index changes thanks to index masking +- radeonsi: don't program COMPUTE_MAX_WAVE_ID (GDS register) on gfx6 +- radeonsi: add helpers to create and clone a sized pm4 state +- radeonsi: add a separate gfx10_init_gfx_preamble_state function +- radeonsi: don't set registers set by CLEAR_STATE in the preamble for gfx10-11 +- radeonsi: add a separate cdna_init_compute_preamble_state function +- radeonsi/ci: add gfx6 failures +- radeonsi: re-indent gfx10_create_sh_query_result_cs +- radeonsi: don't use SET_SH_REG_INDEX on gfx7-9 +- radeonsi: don't use SET_SH_REG_INDEX if the kernel doesn't use CU reservation +- amd: remove unused PKT0 definitions +- treewide: use uint64_t / (u)intptr_t in image address calculations +- amd: drop support for LLVM 11 +- amd: drop support for LLVM 12 +- amd: drop support for LLVM 13 +- amd: drop support for LLVM 14 +- mesa: fix glBitmap in display lists when width <= 0 || height <= 0 +- gallium/hud: append results to files instead of overwriting them +- radeonsi: don't convert L8A8 to R8A8 when blitting via compute to fix gfx7 +- amd: update SET_*_REG_PAIRS* documentation and remove radeon_info options +- amd: improve the IB parser, parse more packets +- amd: rename mid_command_buffer_preemption_enabled -> register_shadowing_required +- amd: increase the attribute ring size on gfx1103_r1 +- amd: don't set PA_RATE_CNTL because it has no effect +- amd: fix GPU cache sizes retrieved from the kernel +- amd: remove non-shadowed register tables +- amd: remove ac_check_shadowed_regs +- amd: add a new helper that prints all non-shadowed regs +- amd: update shadowed register tables for gfx11 +- amd: skip redundant PKT3_NUM_INSTANCES even with register shadowing +- amd: skip redundant INDEX_TYPE even with register shadowing +- radeonsi: set register_shadowing_enabled if AMD_DEBUG=shadowregs is set +- radeonsi/ci: add glx@glx-visuals-stencil to skips because it gets stuck often +- radeonsi: fix RB+ and gfx11 issues with framebuffer state +- radeonsi: change si_emit_derived_tess_state into a state atom +- radeonsi: shrink the last field of tcs_offchip_layout due to LDS limit +- radeonsi: don't do PFP_SYNC_ME before CP DMA and compute blits +- radeonsi: don't needlessly invalidate L0/L1 caches at the beginning of IBs +- radeonsi: add more variables into si_pm4_state and rework how it's created +- radeonsi: remove sscreen parameter from si_pm4_set_reg_idx3 +- radeonsi: set non-graphics uconfig registers first in the preamble +- radeonsi: handle demoted si_pm4_set_reg_idx3 as si_pm4_set_reg +- radeonsi: eliminate redundant compute SH register changes +- radeonsi: handle VGT_GS_OUT_PRIM_TYPE like a tracked register +- radeonsi: handle VGT_LS_HS_CONFIG like a tracker register +- radeonsi: handle GE_CNTL and IA_MULTI_VGT_PARAM as a tracked register +- radeonsi: remove gfx10 NGG streamout +- ci: remove clang-format testing +- intel/ci: disable iris-jsl-deqp because it always fails for an AMD MR +- radeonsi: move TCS.gl_PatchVerticesIn into the tcs_offchip_layout SGPR +- radeonsi: replace tcs_out_lds_layout with nearly identical tes_offchip_addr +- radeonsi: move the only tcs_out_lds_offsets field to vs_state_bits +- radeonsi: eliminate redundant TCS user data and RSRC2 register changes +- radeonsi/gfx11: use SET_*_REG_PAIRS_PACKED packets for pm4 states +- radeonsi: determine si_pm4_state::reg_va_low_idx automatically +- radeonsi: keep pipeline statistics disabled when they are not used +- radeonsi: don't do BREAK_BATCH for context regs with only 1 context per batch +- radeonsi: use si_pm4_create_sized for the shadowing preamble +- radeonsi: remove radeon_winsys::cs_set_preamble +- radeonsi: remove uses_reg_shadowing parameter from si_init_gfx_preamble_state +- radeonsi/gfx11: fix GLCTS with register shadowing by keeping the CS preamble +- radeonsi/gfx11: enable register shadowing by default +- radeonsi: reorder compute code to prepare for packed SET_SH_REG packets +- radeonsi/gfx11: use SET_SH_REG_PAIRS_PACKED for gfx by buffering reg writes +- radeonsi/gfx11: use SET_SH_REG_PAIRS_PACKED for compute by buffering reg writes +- radeonsi: clean up query functions, make them static, remove forward decls +- radeonsi: declare compiler[] and nir_options as pointers to reduce #includes +- radeonsi: clean up #includes +- Revert "egl: return correct error for EGL_KHR_image_pixmap" +- vbo: correctly restore _VaryingInputs for display list fast path +- radeonsi/gfx11: only use SET_*_PAIRS* packets on dGPUs +- radeonsi: fix gfx9 regression causing GPU hangs +- radeonsi/gfx11: fix a regression with PAIRS packets due to shader changes +- Revert "ac/nir/ngg: Follow intrinsic sources when analyzing before culling." +- glthread: determine global locking once every 64 batches to fix get_time perf +- mesa: fix 38% decrease in display list performance of Viewperf2020/NX8_StudioAA +- util/u_queue: fix util_queue_finish deadlock by merging lock and finish_lock + +Mark Collins (1): + +- ir3/a7xx: Add definitions for (last) src GPR attribute + +Mark Janes (8): + +- intel/dev: update mesa_defs.json from defect database +- intel/dev: report stepping for TGL systems +- intel/dev: switch defect identifiers to use lineage numbers +- isl: use generated workaround helpers for Wa_1806565034 +- iris: convert Wa_14010455700 to use workaround mechanism +- anv: convert Wa_14010455700 to use workaround mechanism +- intel: use generated helpers for Wa_1508744258 +- intel/dev: update mesa_defs.json from defect database + +Martin Roukala (né Peres) (26): + +- radv/ci: disable the vkcts-navi21-llvm-valve job +- radv/ci: document all the flakes we hit while I was away +- ci/b2c: allow not specifying a reboot condition +- radv/ci: only reboot on hangs for vkcts-navi10-valve +- zink/ci: document that some tests no longer fail +- zink/ci: mark 77 multisample-related tests as fixed +- radv/ci: document another vkcts flake on vega10 +- radv/ci: document a series of recent regressions +- zink/ci: document recent fixes on RADV +- zink/ci: document new flakes on RADV +- radv/ci: document more flakes for navi21 +- radv/ci: switch to b2c v0.9.10 +- ci/b2c: update to mesa-trigger:2023-03-08.1 +- zink/ci: add more QBO-related fails on RADV +- amd/ci: add another test to the vkcts-vega10 flake list +- zink/ci: remove spec@nv_shader_atomic_int64@* from the fail lists +- ci: bring back the valve farm +- ci/b2c: select the DUT to run on by name +- radv/ci: use the low-priority runners for vangogh jobs +- ci/b2c: change the default first-console-activity timeout to 2 minutes +- zink/ci: add more tests to the flake list of vangogh +- zink/ci: enable zink-radv-vangogh-valve for pre-merge testing +- Revert "ci: mark the valve farm as down" +- amd/ci: temporarily disable some manual jobs that take a long time to run +- zink/ci: remove 3 tests from the fails list +- Revert "amd/ci: temporarily disable some manual jobs that take a long time to run" + +Matt Coster (57): + +- pvr: Complete pvr_isp_ctrl_stream() +- pvr: Fully declare support for VK_EXT_private_data +- pvr: Remove false assumption from pvr_write_draw_indirect_vdm_stream() +- pvr: Fixup format features +- pvr: Unmap mapped memory on free +- pvr: Correctly validate PBE accum format +- pvr: Actually check for depth load when setting up load op constants +- pvr: Initialize aspect_mask when creating buffer views +- pvr: Correctly compile graphics pipelines without a fragment shader +- pvr: Fix off-by-one in pvr_cmd_buffer_upload_desc_set_table() assert +- pvr: Remove unneeded assert in pvr_get_hw_clear_color() +- pvr: Set output_offset correctly in pvr_clear_color_attachment_static() +- pvr: Return correct pbe_accum_format size for A2B10G10R10_UINT_PACK32 +- pvr: Remove bad assert in pvr_clear_attachments() +- pvr: Add PVR_DEBUG=vk_desc option to dump descriptor set layouts +- pvr: Simplify descriptor set layout dump separators +- pvr: Return VkResult from pvr_winsys_create() +- pvr: Propagate errors as VkResults from ioctls through winsys +- pvr: Fix incorrect error return in pvr_ctx_sr_programs_setup() +- pvr: Fix incorrect error handling in pvr_render_ctx_switch_init() +- pvr: Squeeze fd handling into winsys layer +- pvr: Drop pdevice from pvr_physical_device_get_supported_extensions() +- pvr: Rename primary_{device,fd,path} to display_* +- pvr: Use common physical device enumeration +- pvr: Assorted cleanup +- pvr: Return VkResult from winsys buffer_map operation +- pvr: Fix allocation scopes in vkCreateRenderPass2() code path +- pvr: Fix memory leaks on realloc failure in pvr_pipeline.c +- pvr: Correct error flow in pvr_graphics_pipeline_compile() +- pvr: Correct error flow in pvr_compute_pipeline_compile() +- pvr: Use correct surface for deferred RTA clear +- pvr: Rename shadowing loop variable in pvr_add_deferred_rta_clear() +- pvr: Do not free deferred pvr_transfer_cmd instances +- pvr: Fix out of range stream errors for geometry-only jobs on pvrsrvkm +- pvr: Reorder execution in pvr_cmd_buffer_end_sub_cmd() +- pvr: Fix page faults in occlusion query tests +- pvr: Fix rect splitting logic in pvr_unwind_rects() +- pvr: Use correct pbe format for VK_FORMAT_A8B8G8R8_UNORM_PACK32 +- pvr: Use common vkGetPhysicalDeviceFeatures2() implementation +- pvr: Fix segfault in pvr_physical_device_init() +- pvr: Move pvr_get_isp_num_tiles_xy() to rogue_hw_utils.h +- pvr: Use pvr_sub_cmd_event union members directly +- pvr: Add wait_on_previous_transfer flag to graphics subcommand +- pvr: Cleanup in pvr_process_cmd_buffer() +- pvr: Add pvr_image_view_get_image() +- pvr: Publicise some static functions from pvr_blit.c +- pvr: Rename ds_{image,iview} in pvr_gfx_sub_cmd_job_init() +- pvr: Implement ZLS subtile alignment +- pvr: Correct calculations in pvr_unwind_rects() +- pvr: Refactor pvr_unwind_rects() +- pvr: Allow S8_UINT to be used as a stencil attachment format +- pvr: Don't overwrite PDS vertex input flags +- pvr: Declare dependency on idep_mesautil +- pvr: Add support for sampler border colors +- pvr: Correctly read dynamic state setup during blend constant setup +- pvr: Advance entry pointer in pvr_setup_vertex_buffers() +- pvr: Rename transfer 3D heap to transfer frag heap + +Matt Turner (5): + +- intel: Disable shader cache when executing intel_clc during the build +- u_format: Use memcpy to avoid unaligned accesses +- meson: Remove reference to removed SWR driver +- anv: Pipe anv_physical_device to anv_get_image_format_features2 +- anv: Only expose video decode bits with KHR_video_decode_queue + +Matthieu Bouron (1): + +- lavapipe: honor dst base array layer when resolving color attachments + +Michael Tretter (2): + +- panfrost: remove BO from cache before closing GEM +- kmsro: assert that scanout refcount is larger than 0 + +Michel Dänzer (17): + +- ci: Explicitly test for meson feature checks in compiler wrapper +- ci: Use set -e in frontend compiler wrapper scripts. +- ci: Remove shebang from backend compiler wrapper script +- ci: Drop executable permissions from backend compiler wrapper script +- tgsi: Make ureg_DECL_output_masked definition match its declaration +- llvmpipe: Make lp_build_interp_soa declaration match its definition +- mesa/st: Make st_convert_image(_from_unit) declaration match definition +- vulkan: Fix GetPhysicalDeviceSparseImageFormatProperties definition +- anv/format: Fix GetPhysicalDeviceSparseImageFormatProperties definition +- vulkan: Fix GetPhysicalDeviceSparseImageFormatProperties definitions +- svga: Make vmw_svga_winsys_buffer_map definition match declaration +- svga: Make declaration of emit_input_declaration match definition +- clover/llvm: Use llvm::DataLayout::getABITypeAlign with LLVM >= 16 +- clover/llvm: Use std::nullopt already with LLVM 16 +- ci: Drop -Wno-error=array-bounds from fedora-release job +- ci: Upgrade fedora-release job to Fedora 38 +- ci: Enable rusticl in the fedora-release job + +Michel Zou (4): + +- vulkan/wsi: fix -Wnarrowing warning +- vk/entry_points:: fix mingw build +- mesa/draw: fix -Wformat warning +- util: reinstate ENUM_PACKED + +Mihai Preda (1): + +- nir: update nir->num_inputs, num_outputs in nir_recompute_io_bases() + +Mike Blumenkrantz (306): + +- mesa/st/program: don't init xfb info if there are no outputs +- zink: remove atomics from zink_query +- zink: pass ctx through query destroy paths +- zink: always defer query pool deletion +- zink: remove screen param from zink_prune_query() +- util/cpu: add big.LITTLE cpu detection +- driconf: rework glthread enablement +- glthread: disable by default with fewer than 4 (big) CPUs +- zink: move memoryTypeIndex selection down in general bo allocation +- zink: slightly rework memoryTypeIndex selection to pre-determine heap +- zink: restore BAR allocation failure demotion +- zink: make general bo allocation more robust by iterating +- zink: avoid zero-sized memcmp for descriptor layouts +- iris: use util_framebuffer_get_num_samples when setting ps dispatch samples +- nir/lower_alpha_test: rzalloc state slots +- zink: fix non-db bindless texture buffers +- util/blitter: fix line wrapping on error to avoid giving wrong line number +- glthread: add newline to env override +- zink: emit demote cap when using demote +- zink: only print copy box warning once per resource +- zink: hook up debug callback +- zink: use a perf_debug() macro for debug message logging of copy box warning +- util/debug: move null checks out of debug message macro +- zink: manually re-set framebuffer after msrtss replicate blit +- zink: handle 'blitting' flag better in msrtss replication +- zink: skip msrtss replicate if the attachment will be full-cleared +- zink: avoid recursion during msrtss blits from flushing clears +- zink: don't bitcast bool deref loads/stores +- zink: zink_shader_free -> zink_gfx_shader_free +- zink: split out generic shader destruction for reuse +- zink: always wait on precompile fence at start of zink_gfx_shader_free() +- zink: call zink_shader_free for compute shaders +- zink: add a util function for printing shaders +- zink: don't create separate shader dsls if there are no bindings +- drisw: don't leak the winsys +- zink: check for extendedDynamicState3DepthClipNegativeOneToOne for ds3 support +- mesa/st: try to block multisampled texsubimage from doing cpu writes +- mesa: fix ms fallback texture creation +- draw: fix viewmask iterating +- zink: use tes to generate tcs +- zink: hook up EXT_shader_object +- zink: wrap zink_shader_compile_separate() return +- zink: wrap return of compile_module() +- zink: make zink_shader_spirv_compile static +- zink: more zink_shader_object conversion +- zink: use zink_shader_object for precompiled separate shaders +- zink: minor whitespace cleanup +- zink: move separate shader dsl creation to compiler function +- zink: add a 'separate' flag to shader module compile to indicate separate shaders +- zink: run bo lowering passes for separate shader compile with uniform inlining +- zink: remove redundant compute program batch ref +- zink: use EXT_shader_object to (re)implement separate shaders +- zink: add validation exceptions for shader object extension enable +- zink: don't pin flush queue threads if no threads exist +- zink: add z32s8 as mandatory GL3.0 profile attachment format +- zink: add a driver workaround to disable background compiles +- nir/gs: fix array type copying for passthrough gs +- zink: fix array copying in pv lowering +- gallivm: break out native vector width calc for reuse +- llvmpipe: do late init for llvm builder +- zink: print the type of shader when dumping +- zink: use intermediate variable for separate shader descriptor update loop +- zink: use intermediate variable for separate shader db resize check +- zink: simplify separate shader prog init a little +- zink: streamline separate shader descriptor update +- zink: switch to a regular loop to wait on precompile shader fences +- zink: move some shader CSO functions around +- zink: assign separate shader prog stages from ctx->shader_stages +- zink: use a more standardized loop for initing separate shader program descriptors +- zink: move separate shader creation to shader CSO creation +- zink: handle all stages in fixup_io_locations() +- zink: fix longstanding TODO for generated tcs +- zink: use EXT_shader_object to implement generic separate shader precompile +- bump VVL to 1.3.248 +- zink: prune some validation errors from ci +- zink: break out VkImageViewUsageCreateInfo applying for reuse +- zink: reapply VkImageViewUsageCreateInfo when rebinding a surface +- zink: add a workaround for a nir_assign_io_var_locations bug +- zink: don't run update_so_info if shader has no outputs +- zink: add ZINK_DEBUG=noshobj to disable EXT_shader_object +- zink: rename 'separate' param in shader compilation to 'can_shobj' +- zink: explicitly block sample shading in the GPL precompile path +- zink: add zink_program::uses_shobj for managing shader object binds +- zink: use local screen var in zink_gfx_program_update_optimal() +- zink: deduplicate separable program replacement handling +- zink: delete redundant conditional +- zink: use zink_shader_object for zink_shader_module +- zink: use zink_destroy_shader_module() for compute to deduplicate code +- zink: store spirv onto zink_shader_object structs +- zink: allow zink_shader_module to be either a shobj or a mod using a bool +- zink: avoid accessing zink_gfx_program::modules during pipeline compile +- zink: add a union to zink_gfx_pipeline_cache_entry for gpl +- zink: use zink_shader_object for pipeline compiles from zink_gfx_program +- zink: make zink_shader_spirv_compile public +- zink: enable EXT_shader_object for generic precompiles +- draw: fix robust ubo size calc +- ci: disable all a306/a530/a630 jobs +- llvmpipe: fix native vector width init +- zink: update amdpro fails +- zink: add extendedDynamicState3DepthClipNegativeOneToOne to profile +- zink: only unset a generated tcs if the bound tcs is the generated one +- Revert "zink: don't create separate shader dsls if there are no bindings" +- zink: disable a630 traces +- zink: set depth dynamic state values unconditionally +- zink: null some descriptor buffer pointers during destruction +- zink: sync queries at the end of cmdbufs +- cso: unbind fb state when unbinding the context +- i915: use util_copy_framebuffer_state to set fb state +- i915: use util_unreference_framebuffer_state to unref fb state +- iris: use util_unreference_framebuffer_state to unref fb state +- softpipe: use util_unreference_framebuffer_state to unref fb state +- v3d: use util_unreference_framebuffer_state to unref fb state +- vc4: use util_unreference_framebuffer_state to unref fb state +- llvmpipe: use util_unreference_framebuffer_state to unref fb state +- svga: use util_unreference_framebuffer_state to unref fb state +- zink: move EXT_shader_object check to another place +- zink: break out optimal key handling into separate function +- zink: disable EXT_shader_object if !optimal_keys +- zink: add ZINK_DEBUG=optimal_keys +- gallium: pipe_rasterizer_state::point_tri_clip -> point_line_tri_clip +- aux/draw: guard_band_points_xy -> guard_band_points_lines_xy +- aux/draw: add guardband clipping for lines +- zink: don't init mutable resource bit for swapchain images +- zink: don't init mutable for swapchain src during blit +- tgsi_to_nir: handle PIPE_CAP_NIR_COMPACT_ARRAYS for clipdistance +- zink: allow vk 1.2 timelineSemaphore feature if extension isn't supported +- zink: stringify unsupported prim restart log error +- zink: delete persistent map tracking +- zink: add PERSISTENT for db buffer maps +- zink: delete unnecessary pipeline stage flags from inference +- zink: use an intermediate variable for binding ssbo slots +- zink: unbind the ssbo slot being iterated, not the index of the buffer +- zink: flush INDIRECT_BUFFER mem barrier for compute +- zink: disable batched unordered barries with ZINK_DEBUG=noreorder +- zink: block batching of unordered barriers if previous usage was write +- zink: fix uncached memory readback +- glsl/lower_samplers_as_deref: apply bindings for unused samplers +- vulkan/runtime: add VK_DYNAMIC_STATE_ATTACHMENT_FEEDBACK_LOOP_ENABLE_EXT +- zink: add ZINK_DEBUG=noopt +- zink: add ZINK_DEBUG=nobgc +- zink: make mesa_logw separate from perf_debug +- zink: add perf_debug for "interesting" shader compiles +- zink: set debug callback on context +- zink: bind bindless db set when updating separate shader db sets +- zink: compare desc set to detect bindless vars in separate shaders +- zink: adjust bindless texel buffer handle before indexing +- zink: block more flushes during unordered blits +- zink: also cache swapchain semaphores +- zink: disable always zs feedback loop on radv +- zink: add back some anv qbo flakes +- zink: disable have_EXT_vertex_input_dynamic_state without EDS2 +- zink: disable dynamic state exts if the previous ones aren't present +- zink: add some ci flakes +- zink: don't leak swapchain readback semaphores +- zink: destroy current batch state after all other batch states +- zink: reorder some native blit code +- zink: reject blits where src/dst is 3D and dst/src z!=0 +- zink: reorder some image copy code +- zink: ignore no-op image copies +- zink: only add feedback loop usage bit if extension is supported +- lavapipe: EXT_attachment_feedback_loop_layout_dynamic_state +- zink: slightly simplify bda allocation chaining +- zink: hook up some memory extensions +- zink: set higher prio on dedicated memory allocations +- zink: flag batch usage on swapchain images +- vulkan/wsi: add feedback loop usage to swapchain caps if supported +- zink: add feedback loop usage for swapchains +- vtn: add spirv index to type mismatch error for debugging +- vtn: print spirv id for type mismatch error +- vtn: print spirv ids for type mismatch in bcsel +- vtn: add more info to bitcast bit size error message +- zink: try update fb resource refs when starting new renderpass +- zink: add special-casing for (not) reordering certain image barriers +- zink: use batch usage function for a simple case +- zink: move zink_batch_state::submit_count to zink_batch_usage +- zink: move batch usage to substruct on zink_bo objects +- zink: track/check submit info on resource batch usage +- zink: disable unordered blits when swapchain images need aqcuire +- zink: explicitly disable reordering after restricted swapchain readback blits +- zink: explicitly disable promotion on images that are both unflushed and non-reorderable +- zink: flag 'has_work' on batch when promoting a cmd +- lavapipe: more correctly handle null pipeline states +- anv: more correctly handle null pipeline states +- vk/graphics_state: handle null pipeline state structs in creation +- zink: promote flushed clears to unordered cmdbuf when possible +- zink: also declare int size caps inline with signed int type usage +- zink: delete unnecessary bitcast in load_shared/scratch +- zink: use void return for store_dest +- zink: move get_alu_type() up in file +- zink: manually memcpy the spirv instruction buffer +- zink: write out register variables to a separate spirv buffer +- zink: dynamically emit non-bool register values using local_vars spirv buffer +- zink: store and use alu types for ntv defs +- zink: infer types from load_const instrs to avoid more bitcasts +- lavapipe: bump memory allocation heap to 3GiB +- lavapipe: report full memory in heap for 64bit processes +- lavapipe: EXT_memory_budget +- lavapipe: EXT_memory_priority +- lavapipe: store memory allocation size onto lvp_device_memory +- lavapipe: VK_EXT_pageable_device_local_memory +- zink: don't wait on queue thread if disabled +- zink: use the per-context track_renderpasses flag in more places +- zink: don't remove psiz from linked shaders if the consumer reads it +- zink: don't propagate psiz in quads emulation gs +- lavapipe: VK_EXT_dynamic_rendering_unused_attachments +- zink: require EXT_dynamic_rendering_unused_attachments for dynamic rendering +- zink: explicitly avoid ci errors due to unrecognized extensions in VVL +- vulkan: reorder vk_cmd_queue_entry +- vulkan/cmd_queue: allocate cmds based on the size of the cmd +- vulkan/cmd_queue: expose cmd sizes +- vulkan: use cmd size array for queued cmd allocations +- ci: uprev VVL to 1.3.251 +- lavapipe: fix DS3 min sample setting +- lavapipe: bump max push constant size +- lavapipe: stop setting patch vertices constantly +- lavapipe: don't pass indirect info in streamout draws +- draw: add (disabled) vertex dumping for non-linear emit +- lavapipe: fix memory budget reporting +- zink: also disable bg compile for compute with nobgc +- zink: hook up VK_EXT_attachment_feedback_loop_dynamic_state +- zink: use dynamic state for feedback loops when available +- zink: enable EXT_shader_object globally with have_EXT_attachment_feedback_loop_dynamic_state +- zink: add a ci flake +- lavapipe: pass list to cmdbuf exec, not cmdbuf +- lavapipe: add a mapping for BDA +- lavapipe: add a zeroed buffer that can be bound in place of an index buffer +- lavapipe: handle index buffers with offsets for indirect draws +- lavapipe: NV_device_generated_commands +- zink: combine some rast state draw conditionals +- zink: don't check prog->shaders when creating gfx pipeline +- zink: check for cached mem correctly when mapping buffer +- zink: remove assert for dt in zink_kopper_update +- zink: stop swizzling conditional render during batch flush +- zink: update some radv qbo fails +- radv: tweak gfx pipeline stage binding +- zink: only try to create srgb mutable images if the vk format is supported +- vk: make vk_format_map[] public +- radv: directly use vk_format_map for vertex input +- lavapipe: use PACKAGE_VERSION for cache uuid in release builds +- zink: massively shrink qbo size for timestamp queries +- zink: assert that ntv image creation isn't clobbering existing images +- zink: add some ntv asserts for ms txf +- zink: add a dgc debug mode for testing +- lavapipe: add version uuid to shader binary validation +- egl/dri2: trigger drawable invalidation from surface queries for zink +- zink: add some ci flakes +- zink: break out vk flag unrolling into util function +- zink: add mem debugging +- zink: remove redundant conditional in set_sampler_views +- zink: wrap format mismatch checks for blit/surface +- zink: add srgb mutable for all resources by default +- zink: drop dt checks for mutable format init +- zink: strip format list when disabling mutable during image creation +- dri3: only invalidate drawables on geometry change if geometry has changed +- zink: more anv ci flakes +- aux/trace: add methods for mesh shaders +- lavapipe: more fixes for sample shading +- lavapipe: fix shader binary binding with mesh shaders +- lavapipe: correctly update shader object per-stage push constant sizes +- zink: add COHERENT requirement for CACHED memory +- zink: ZINK_HEAP_HOST_VISIBLE_CACHED -> ZINK_HEAP_HOST_VISIBLE_COHERENT_CACHED +- zink: fix anv ci flake wildcarding +- aux/pipebuffer: add a return to pb_slabs_reclaim() +- aux/pipebuffer: add a return to pb_cache_release_all_buffers() +- zink: only retry bo allocation after reclaim if reclaims actually happened +- zink: fix ubo array sizing in ntv +- zink: acquire persistently bound swapchain descriptors before setting usage +- zink: recache present semaphores +- zink: always clamp NUM_QUERIES to 500 +- zink: radv vangogh ci updates +- radv: remove redundant intermediate variable in radv_is_mrt0_dual_src() +- radv: inline radv_can_enable_dual_src() +- zink: no-op redundant samplemask changes +- zink: force inlining for a bunch of functions +- zink: make invalidate_descriptor_state a ctx hook +- zink: specialize invalidate_descriptor_state hook for compact mode +- zink: clean up rp update tracking on dsa bind +- zink: use local screen var in blend state bind +- zink: track and apply ds3 states only on change +- zink: don't update tc info directly from cso binds +- zink: check sampler views pointer before loop +- zink: add fastpaths for no-op sampler/view rebinds +- nir/lower_tex: ignore saturate for txf ops +- radv: pre-init surface info +- ci: add a test-dozen-deqp flake +- lavapipe: handle multiview queries +- zink: fix assert for inline uniform invalidation with generated gs bound +- zink: fix unbinding generated gs on real gs bind +- zink: get new bda when rebinding invalidated buffers +- lavapipe: create a desc set for immutable sampler layouts +- lavapipe: split out descriptor stage setting +- lavapipe: EXT_descriptor_buffer +- lavapipe: VK_EXT_mutable_descriptor_type +- llvmpipe: flush/reference fs ubos on bind +- zink: do initial program unref during program creation +- zink: fix separate shader program refcounting +- docs: update lavapipe extensions +- zink: don't destroy swapchain on initial CreateSwapchainKHR fail +- aux/trace: fix bindless texture dumping +- vk/wsi/x11: move surface alpha check from get_caps to creation +- vk/wsi/x11: handle geometry updating more asynchronously +- vk/wsi/x11: stop roundtripping on presentation +- vk/wsi: unify dmabuf exporting +- vk/wsi: add error logging for syncfile import/export failures +- zink: fix anv ci flakes (for real this time) +- zink: fix batch disambiguation on first submit + +MouriNaruto (1): + +- dzn: Fix segmentation fault when Direct3D 12 user mode driver from at least one of GPUs is not available. + +MrRobbin (1): + +- zink: Move the workaround before the EDS setting. + +Mykhailo Skorokhodov (4): + +- mesa: Implement GL_CLEAR_TEXTURE flag +- mesa: Fallthrough GL_SRB_DECODE_ARB pname +- iris: Fix memory size with disabled resizable bar +- nir: Rematerialize derefs after opt_dead_cf + +Mykola Piatykop (1): + +- mesa: Fix use after free. + +Nanley Chery (26): + +- iris: Allocate ZEROED BOs for shared resources +- iris/bufmgr: Add and use zero_bo +- iris/bufmgr: Handle flat_ccs for BO_ALLOC_ZEROED +- intel/isl: Bump the MCS halign value for BDW+ +- iris: Add a barrier to iris_mcs_partial_resolve +- intel: Implement ISL_AUX_OP_AMBIGUATE for MCS +- iris: Enable MCS init with ISL_AUX_OP_AMBIGUATE +- anv: Drop the MCS initialization performance warning +- anv: Enable MCS init with ISL_AUX_OP_AMBIGUATE +- intel/blorp: Assert an 8bpp fast clear restriction +- iris: Init CCS_E to COMPRESSED_NO_CLEAR for XeHP +- intel/blorp: Use the depth copy format more on BDW+ +- intel/blorp: Add depth usage check for copy format +- intel/blorp: Change condition for CCS_E copy formats +- intel/blorp: Add and use blorp_copy_get_formats +- iris: Use known formats for tex_cache_flush_hack +- iris: Drop a GFX12_CCS_E check in can_fast_clear_color +- intel: Rename the GFX12_CCS_E aux-usage to FCV_CCS_E +- iris: Avoid extra CCS_E flushes for aux mode changes +- iris: Avoid FCV_CCS_E for shader image accesses +- iris: Assert against FCV_CCS_E for blitter writes +- intel/blorp: Avoid 32bpc fast clear sampling issue +- Revert "iris: Add missed tile flush flag" +- iris: Drop the RT flush for PIPE_BARRIER_TEXTURE +- iris: Drop GPGPU Tex Invalidate restriction for TGL+ +- isl: Add and use size and alignment calculators + +Oskar Rundgren (20): + +- pvr: Allow block compressed source blit +- pvr: Transfer PBE source snorm format should be signed +- pvr: Transfer PBE gamma is unset +- pvr: Transfer fix blit with multiple emits +- pvr: Transfer multiple emits clip rectangle +- pvr: Add back S8_UINT support +- pvr: Add PBE packmode for depth stencil formats +- pvr: Transfer add depth merge support for X8_D24 +- pvr: Transfer add s8_uint support +- pvr: PBE fix mesa pipe swizzle conversion +- pvr: Transfer ignore non zero stride for twiddled surface +- pvr: Transfer block compressed with 3d twiddled layout +- pvr: Transfer support flipped rectangle mapping +- pvr: Transfer remove byte unwind workaround +- pvr: fix texel unwind workaround mappings +- pvr: Transfer check valid source address mask +- pvr: Transfer optimisation remove unused features from API +- pvr: Transfer image to buffer dest rect +- pvr: Fix transfer image clearing PBE packmodes +- pvr: add block compressed formats blit support + +Patrick Lerda (22): + +- r600: fix refcnt imbalance related to r600_set_vertex_buffers() +- r600: fix refcnt imbalance related to evergreen_set_shader_images() +- lima: fix refcnt imbalance related to framebuffer +- r600/sfn: fix memory leak related to sh_info->arrays +- aux/draw: fix memory leak related to ureg_get_tokens() +- crocus: fix refcnt imbalance related to framebuffer +- crocus: fix refcnt imbalance related to crocus_create_surface() +- r600: fix refcnt imbalance related to atomic_buffer_state +- radeonsi: set proper drm_amdgpu_cs_chunk_fence alignment +- crocus: fix scratch_bos memory leak +- mesa: fix refcnt imbalance related to egl_image_target_texture() +- glthread: fix typo related to upload_vertices() +- mesa: fix refcnt imbalance related to _mesa_delete_semaphore_object() +- mesa/st: fix refcnt imbalance related to st_feedback_draw_vbo() +- mesa/st: fix buffer overflow related to set_program_string() +- r600: fix r600_draw_vbo() buffer overflow +- nouveau: fix nouveau_heap_destroy() memory leak +- r600: fix cayman_convert_border_color() swizzle behavior +- util/blitter: fix util_blitter_clear_buffer() refcnt imbalance +- util/blitter: revert util_blitter_clear_buffer() +- radeonsi: fix refcnt imbalance related to util_blitter_save_fragment_constant_buffer_slot() +- panfrost: fix refcnt imbalance related to blitter + +Paulo Zanoni (6): + +- iris: Store prime fd of external bos for Xe KMD +- iris: Add functions to import and export implicit sync state +- iris: Extend iris_bo_wait_syncobj() to wait on external implicit syncobj +- iris: Add iris_implicit_sync struct and functions to do implicit synchronization for Xe kmd +- iris: also avoid isl_memcpy_linear_to_tiled for Tile64 +- intel/isl: tile 64 calculations work with 1D surfaces + +Pavel Ondračka (32): + +- r300: fix unconditional KIL on R300/R400 +- r300: add CI list of known rv370 dEQP failures +- r300: remove simple duplicate ARL instructions +- r300: fuse ROUND and ARL to ARR +- r300: remove nir round lowering +- r300: enable PIPE_CAP_TGSI_TEXCOORD +- r300: fail linking instead of using dummy shaders +- CODEOWNERS: add r300 driver +- r300: move nir stuff to r300_nir file +- r300: move the ARL merging pass up in the opt loop +- r300: move the ROUND+ARL->ARR fusing to main optimization loop +- r300: optimize the load A0 pattern from wined3d +- r300: remove duplicate ARRs +- r300: be more agressive when merging A0 loads +- r300: remove unused SIN/COS lowering +- r300: remove unused SSG lowering +- r300: move CEIL lowering to NIR +- r300: remove unused FLR lowering +- r300: remove unused POW lowering +- r300: remove unused DST lowering +- r300: remove unused ROUND lowering +- r300: remove unused LIT lowering +- r300: remove unused opcodes from r300_tgsi_to_rc +- nir_opt_algebraic: don't use i32csel without native integer support +- r300: add partial CMP support on R5xx +- r300: properly count maximum used register index +- r300: lower undefs to zero +- r300: add some early safe bool lowering +- r300: remove most of backend contant folding +- r300: disable ntt regalloc for vertex shaders +- r300: assert that every writer has a reader +- r300: update RV370 failures + +Philipp Zabel (1): + +- etnaviv: fix segfault after compile failure + +Pierre-Eric Pelloux-Prayer (11): + +- amd: update amdgpu_drm.h +- amd: determine info->has_fw_based_shadowing +- radeonsi: implement fw based mcbp +- amd: update amdgpu_drm.h +- radeonsi: stop reporting reset to app once gpu recovery is done +- winsys/amdgpu: add a helper function to submit a no-op job +- winsys/amdgpu: use the no-op helper to detect if reset completion +- mesa: don't share reset status across contexts +- mesa: remove unused bools +- llvmpipe: only include old Transform includes when needed +- Revert "gallium/u_threaded: buffer sharedness tracking" + +Qiang Yu (119): + +- nir: add nir_load_barycentric_optimize_amd intrinsic +- radeonsi: implement nir_load_barycentric_optimize_amd +- ac/nir/ps: lower barycentric load when bc_optimize +- ac/nir/ps: add force lower barycentric load options +- ac/nir/ps: lower sample mask input when needed +- ac/llvm,radeonsi: lower ps color load in nir +- radeonsi: add si_nir_lower_ps_color_input +- radeonsi: add si_nir_emit_polygon_stipple +- radeonsi: handle lowered ps in scan_io_usage +- radeonsi: monolithic ps emit prolog in nir directly +- radeonsi: restructure mono merged shader build +- radeonsi: remove separate_prolog parameter +- radeonsi: add si_mark_divergent_texture_non_uniform +- ac/llvm,radeonsi: use texture non-uniform flag as waterfall switch +- nir,ac/llvm,radeonsi: replace nir_load_smem_buffer_amd with nir_load_ubo +- ac/llvm,radeonsi: lower nir_load_point_coord_maybe_flipped in nir +- ac,radv: move ps arg compation to common place +- aco: support 32bit address in nir_load_smem_amd +- nir: add missing image atomic_inc/dec_wrap intrinsic +- aco: implement nir_bindless_image_atomic_inc/dec_wrap +- aco: skip scratch buffer init when its arg is not used +- aco: fix nir_f2u64 translation +- nir: add nir_export_dual_src_blend_amd intrinsic +- aco: move create_fs_dual_src_export_gfx11 above +- aco: implement nir_export_dual_src_blend_amd +- ac/nir/ps: use nir_export_dual_src_blend_amd when aco +- ac/nir/ps: add no_color_export option +- aco: support nir_export_amd with ps targets +- aco,radv: lower outputs to exports when nir for monolithic ps +- ac/llvm: remove output variable declaration for radv ps +- radv: implement nir_load_barycentric_optimize_amd +- ac/nir/ps: remove used nir_variable if created +- aco,ac/llvm,radv,radeonsi: handle ps bc optimization in nir for radv +- aco,radv: remove unused aco compile options +- aco,radv: support symbol relocation in aco +- aco: get scratch addr from symbol for radeonsi +- aco: allow no export instruction for gfx10+ fs +- ac/nir/cull: fix line position w culling +- meson: build radeonsi with aco +- radeonsi: add aco debug option +- radeonsi: add use_aco field for struct si_shader +- radeonsi: add shader info for frag coord and sample pos read +- radeonsi: add shader info uses_sampleid +- radeonsi: pack spi ps input fixup to a function +- radeonsi: init spi ps input shader config when aco +- radeonsi: add a raw shader binary type +- ac/binary: pack prefech align code to a function +- radeonsi: support raw shader binary upload +- radeonsi: support print raw shader binary +- radeonsi: remove ps vgpr index save when args init +- tgsi_to_nir: call nir_lower_int64 when required +- ac/llvm,radeonsi: lower idiv in nir +- ac/llvm,radeonsi: lower fsin/fcos in nir +- ac/llvm,radeonsi: lower txf offset in nir +- ac/llvm,radeonsi: lower ineg in nir +- ac/llvm,radeonsi: lower some pack/unpack ops not supported by aco +- ac/llvm,radeonsi: lower nir_fpow for aco and llvm +- radeonsi: lower some 64bit ops aco does not support +- radeonsi: lower vector const to scalar at last for aco +- radeonsi: add has_non_uniform_tex_access shader info +- radeonsi: lower non uniform texture access when aco +- radeonsi: add initial aco compile code +- radeonsi: add symbols to si_shader_binary +- radeonsi: resolve aco scratch addr symbols +- radeonsi: adjust ps args for aco +- radeonsi: pass use_aco to ac_nir_lower_ps +- radeonsi: clamp shadow texture reference in nir for aco +- ac/llvm,radeonsi: enable lower_array_layer_round_even +- radeonsi: fixup sampler desc for tg4 in nir +- radeonsi: be able to use aco compiler for mono ps +- ac/llvm: remove the double frcp special handling +- radeonsi: fix aco compile for atomic ops +- ac/llvm: remove redundant nir_lower_legacy_atomics +- radeonsi: fix uses_instanceid for merged mono shader stage +- aco: implement two load lds ngg intrininsic for radeonsi +- aco,radv: remove unused aco_shader_info fields +- ac/nir/ngg: don't use 8bit alu ops +- aco: implement load buffer with ACCESS_USES_FORMAT_AMD +- aco/assembler: handle ds_(add|sub)_gs_reg_rtn encoding +- aco: use gds reg when ordered xfb counter add +- aco: implement nir_xfb_counter_sub_amd +- aco: implement nir_bindless_image_fragment_mask_load_amd +- aco: use ac_get_image_dim for array check when image intrinsic +- radeonsi: resolve lds ngg aco symbols +- radeonsi: add scratch offset vs args explicitly for aco +- ac/llvm,radeonsi: lower nir_load_gs_vertex_offset_amd in abi +- ac/llvm,radeonsi: lower nir_load_merged_wave_info_amd in abi +- ac/llvm,radeonsi: lower load_workgroup_num_input_(vertices|primitives) in abi +- ac/llvm,radeonsi: lower nir_load_initial_edgeflags_amd in abi +- ac/llvm,radeonsi: lower nir_load_packed_passthrough_primitive_amd in abi +- ac/llvm,radeonsi: lower nir_load_ordered_id_amd in abi +- ac/llvm,radeonsi: lower nir_load_ring_esgs_amd in abi +- nir,ac/llvm,radeonsi: replace nir_buffer_atomic_add_amd with ssbo atomic +- radeonsi: fill aco shader info for mono standalone vs +- radeonsi: calculate needed lds size when upload raw binary for vs +- radeonsi: use nir_umul_high for fast udiv +- radeonsi: always use scoped barrier +- ac/llvm: remove unused barrier implementation +- radeonsi: enable aco for mono standalone vs +- aco,radv: remove unused gs aco shader info +- ac/nir,radv: add 1 dword to LS/HS vertex stride +- ac/nir,radv: add 1 dword to ES/GS item size +- radeonsi: add scratch_offset arg for aco tcs +- radeonsi: lower nir_load_tess_rel_patch_id_amd in abi for aco +- ac/llvm,radeonsi: lower nir_load_ring_tess_offchip_amd in abi +- radeonsi: enable aco support for mono standalone tcs +- radeonsi: add scratch_offset arg for aco tes +- radeonsi: init tes aco shader info fields +- radeonsi: update lds size for tes +- radeonsi: enable aco support for standalone tes +- radeonsi: add scratch_offset arg for aco gs +- ac/llvm,radeonsi: lower nir_load_ring_gsvs_amd in abi +- radeonsi: enable aco for standalone gs +- radeonsi: enable aco support for gs copy shader +- radeonsi: add scratch_offset arg for aco cs +- ac/llvm,radeonsi: lower nir_load_user_data_amd in abi +- radeonsi: fix crash when AMD_DEBUG=cs,initnir +- radeonsi: enable aco support for compute shader +- ac/nir/ngg: fix ngg_gs_clear_primflags crash + +QwertyChouskie (1): + +- docs/features.txt(fix): mark VK_EXT_pipeline_robustness as supported on radv + +Rajnesh Kanwal (9): + +- pvr: Add support to process transfer and blit cmds +- pvr: Implement vkCmdCopyBufferToImage API. +- pvr: Implement vkCmdCopyImage2KHR API. +- pvr: Implement vkCmdBlitImage API. +- pvr: Implement vkCmdClearColorImage API. +- pvr: Implement vkCmdCopyImageToBuffer2 API. +- pvr: Implement vkCmdFillBuffer API. +- pvr: Implement vkCmdResolveImage2KHR API. +- pvr: Implement vkCmdClearDepthStencilImage API. + +Rhys Perry (77): + +- nir: add is_gather_implicit_lod +- vtn: set is_gather_implicit_lod +- aco: support implicit LOD for nir_texop_tg4 +- ac/llvm: support implicit LOD for nir_texop_tg4 +- aco: remove SMEM_instruction::prevent_overflow +- aco: use apply_nuw_to_ssa() with load_smem_amd +- ac/nir/ps: fix null export write mask miss set to 0xf +- aco: don't move exec reads around exec writes +- aco: don't move exec writes around exec writes +- radv: fix bc optimization with POS_W_FLOAT_ENA(1) +- aco/ra: create M0-affinities for s_sendmsg +- aco/gfx11: fix VMEM/DS->VALU WaW/RaW hazard +- amd/drm-shim: move device list to external file +- amd/drm-shim: add polaris10 +- amd/drm-shim: add vega10 +- amd/drm-shim: add navi10 +- aco: add get_op_fixed_to_def() helper +- aco: consider how definitions fixed to operands can change register demand +- nir/fold_16bit_tex_image: skip tex instructions with backend1 +- nir,vtn,aco,ac/llvm: make cube_face_coord_amd more direct +- ac/nir: add pass for lowering 1d/cube coordinates +- ac/nir: round layer in ac_nir_lower_tex +- radv,radeonsi: use ac_nir_lower_tex +- nir/lower_tex: remove lower_array_layer_round_even +- ac/nir: add fix_derivs_in_divergent_cf +- aco: remove unused RegType +- aco: let p_start_linear_vgpr take an operand +- aco: add MIMG_instruction::strict_wqm +- aco: implement strict_wqm_coord_amd +- aco: implement texture samples with strict WQM coordinates +- radv: use fix_derivs_in_divergent_cf +- aco/tests: improve performance of declaration parsing +- aco/tests: add fix_derivs_in_divergent_cf tests +- aco: fix update_alu(clear=true) for exports +- aco: use pass_flags to recover s_delay_alu cycles +- aco: insert s_delay_alu on the linear CFG +- aco: improve printing of s_delay_alu +- radv: allow wave32 for geometry shaders +- aco: fix has_color_exports=true for mrtz exports +- aco/tests: add discard export target tests +- aco: fix ds_sub_gs_reg_rtn validation +- radv: initialize aco_compiler_options::is_opengl +- radv: correctly skip vertex loads with packed formats +- aco: consider position/primitive exports around memory barriers +- ac/nir: use scoped barriers to finish stores before exports +- aco: remove memory_barrier_buffer implementation +- aco: mask bits source of s_bfe +- aco/tests: test that s_bfe bits is masked +- util: fix gc_alloc_size alignment +- util/tests: add gc_alloc_size alignment tests +- aco: run nir_lower_int64 after nir_opt_uniform_atomics +- ac: fix PIPE_FORMAT_R11G11B10_FLOAT DST_SEL_W +- radv: refactor CS subgroup size determination +- radv: use wave32 for small workgroups +- aco: don't try to form load+store clauses +- aco/gfx11: use s_clause with stores +- aco/gfx11: schedule for VMEM store clauses +- aco: don't set exec_hi for wave32 scan reductions +- amd/drm-shim: use fixed-width types +- nir/peephole_select: allow some invocation broadcast intrinsics +- aco: include helpers in emit_uniform_{reduce,scan} +- nir,aco: add INCLUDE_HELPERS index to reduce intrinsic +- nir/opt_intrinsic: optimize quad vote +- radv: use nir_opt_intrinsics +- aco,ac/llvm,ac/nir,vtn: unify cube opcodes +- nir: split nir_lower_mov64 +- radv: use nir_lower_conv64 +- radv: call nir_lower_int64 later +- radeonsi: use nir_lower_conv64 +- aco: remove 64-bit integer conversion opcodes +- ac/llvm: fix AC_TM_CHECK_IR +- radv: fix radv_get_ballot_bit_size with CS +- ac/llvm: fix wave32 ac_build_mbcnt_add with 64-bit mask +- ac/llvm: skip ballot zext for 32-bit dest with wave32-as-wave64 +- radv: add conformant_trunc_coord to cache UUID +- ac/nir: always round cube array layers +- nir/unsigned_upper_bound: fix phi(bcsel) + +Rob Clark (57): + +- freedreno/a6xx: Fix valid_format_cast logic for newer a6xx +- freedreno: Remove unused fd_batch_reset() +- freedreno: Inline single-caller helpers +- freedreno: Extra casting to make C++ happy +- freedreno/registers: C++ struct casting +- util/log: Add missing "const" +- freedreno/ir3: More perfetto tracing +- mesa/nir: Add some perfetto traces +- freedreno/perfetto: Add shader_id for compute stages +- freedreno: Add dirty state logging +- freedreno/a6xx: Pass ring to __ONE_REG() +- freedreno: Add more tracepoint fields +- freedreno: Fix resource tracking vs rebind/invalidate +- freedreno/a6xx: Change a618 tile_align_h back to 32 +- dri/android: Fix MSAA resolve +- Revert "ci: disable all a306/a530/a630 jobs" +- freedreno/a6xx: Rework set_bin_size() +- freedreno/a6xx+: Use template to handle a6xx vs a7xx differences +- freedreno/batch: Add helper to set fb state +- freedreno/a6xx: Move LRZ clear to blitter +- freedreno/a6xx: Add ctx->emit_sysmem() +- freedreno/a6xx: Simplify per-tile conditional IBs +- freedreno/a6xx: Switch to batch->cleared +- freedreno/a6xx: Split tile loads and clears +- freedreno/a6xx: Introduce batch subpasses +- freedreno/a6xx: Per-subpass LRZ +- freedreno/a6xx: New subpass on mid-frame clears +- freedreno/a6xx: Move LRZ clears to gmem +- freedreno/a6xx: Actually use LRZ for ms +- freedreno/a5xx+a6xx: Don't allocate LRZ for z32 +- tu: Move queue deletion to last +- mesa: Skip update_gl_clamp() if samplers need clamp +- freedreno/a6xx: Template specialization for draw type +- freedreno/a6xx: Template specialization for pipeline type +- freedreno/a6xx: Optimize max_indices calculation +- freedreno/batch: Move submit bo tracking to batch +- freedreno/drm: Don't try to export suballoc bo +- freedreno: Handle export error handling +- freedreno: Add aux-context support +- freedreno: Reallocate on unshared export +- freedreno/a6xx: Clean up open coded flushes +- freedreno/a6xx: Stop using fd_wfi() +- freedreno/a6xx: Add missing cap +- freedreno/a6xx: Fix xfb stream configuration +- freedreno/a6xx: Remove primitives_relocw() +- freedreno/a6xx: GL_ARB_transform_feedback_overflow_query +- freedreno/a6xx: Split primitives and pipeline-stats queries +- freedreno/a6xx: Handle nested pipeline stats queries +- freedreno: Handle compute queries +- freedreno/a6xx: GL_ARB_pipeline_statistics_query +- freedreno/a6xx: Enable gl46 +- freedreno: Add extra assert +- freedreno/batch: Add driver-thread assert +- freedreno/a6xx: Directly invalidate on samp view update +- freedreno/a6xx: Use idalloc for samp/view seqno's +- freedreno/fdperf: Use common device info helpers +- freedreno/drm/virtio: Trigger host side wait boost + +Robert Beckett (1): + +- winsys/panfrost: Fix a scanout resource leak + +Robert Mader (1): + +- egl/wayland: wait for compositor to release shm buffers + +Rohan Garg (32): + +- anv: use the workaround framework for WA 14013111325 +- hasvk: drop dead code +- iris: use the workaround framework for WA 14013111325 +- anv: use the common vulkan runtime to do the heavy lifting +- anv: drop duplicated nir_opt_dce passes +- intel: infer scalar'ness locally for brw_postprocess_nir +- intel: drop unused is_scalar function parameter in brw_nir_apply_key +- intel: update comments about non-existent function parameter +- intel: infer scalar'ness locally for brw_vectorize_lower_mem_access +- anv: drop duplicate checks when setting the compressed bit +- iris: correctly set alignment to next power of two for struct size +- ac/surface: make sure alignment is a POT +- freedreno: set alignment to next POT +- util: fix ROUND_DOWN_TO alignment type +- util: migrate alignment functions and macros to use ALIGN_POT +- util: revert back to ALIGN since it moved to util +- util: move pot functions to use existing macros +- anv: enable single texel alignment +- isl: add helper to check if aux usage is CCS_E +- anv: set aux usage to GFX12_CCS_E if a platform needs WA 14010672564 +- anv: limit non zero fast clear check to GFX12_CCS_E +- anv: fix incorrect asserts when combining CPS and per sample interpolation +- hasvk: enable single texel alignment +- anv: split ANV_PIPE_RENDER_TARGET_BUFFER_WRITES for finer grained flushing +- anv: move WA 1607854226 to use the WA infrastructure +- intel/compiler: construct masks instead of using magic values +- intel/compiler: reuse previously computed bitsize +- anv: retry batchbuffer submission with i915 +- iris: migrate WA 14013910100 to use the WA framework +- iris: migrate WA 14016118574 to use the WA framework +- iris: fix iris for WA 16013000631 +- intel/perf: add perf query support for Intel Raptorlake + +Roland Scheidegger (2): + +- llvmpipe: minor cleanups in line rendering code +- llvmpipe: fix some corner cases with line rendering + +Romain Failliot (1): + +- docs(fix): remove last ref to i965 in features.txt + +Ruijing Dong (18): + +- radeonsi/vcn: add macros used in av1 encoding +- radeonsi/vcn: enable 2 pass search center map +- radeonsi/vcn: enable swizzle mode in encoding ref frames. +- radeonsi/vcn: merge get_output_format_param function +- radeonsi/vcn: remove extra zero bytes from bitstream +- radeonsi/vcn: add av1 dpb variables and cdf table +- gallium/pipe: add av1 encoding data structure in pipe +- radeonsi/vcn: add av1 enc data structure +- radeonsi/vcn: add some av1 encoding function +- radeonsi/vcn: add av1 encoding ib packages and get_info +- frontends/va: adding va av1 encoding functions +- radeonsi/vcn: use PIPE_ENC_FEATURE enum +- frontends/va: define va av1 encoding caps +- radeonsi/vcn: correct cropping for hevc case +- radeonsi/vcn: fix decoding bs buffer alignement issue. +- gallium/pipe: add interface update_decoder_target +- radeonsi/vcn: apply update_decoder_target logic +- frontends/va: remove private member and update target buffer + +Ryan Houdek (1): + +- util: move check for AVX512 + +Ryan Neph (2): + +- virgl: add debug flag to force synchronous GL shader compilation +- virgl: check a debug option again at context creation + +Sagar Ghuge (19): + +- anv: Factor out code from anv_image_hiz_clear +- anv: Move and make anv_can_hiz_clear_ds_view non-static +- anv: Fast clear depth/stencil surface in vkCmdClearAttachments +- anv: Set CS stall bit during HIZ_CCS_WT surface fast clear +- iris: Set CS stall bit during HIZ_CCS_WT surface fast clear +- intel/genxml: Add CCS cache flush field to PIPE_CONTROL +- intel/genxml: Add Compute/Blitter CCS aux invalidation register +- anv: Add CCS cache flush bits to anv_pipe_bits +- anv: Fix AUX-TT invalidation +- anv: implement recommended flush/wait of AUX-TT invalidation on compute +- iris: Add CCS cache flush bits +- iris: Fix AUX-TT invalidation +- iris: implement recommended flush/wait of AUX-TT invalidation +- intel/ds: Track CCS cache flush bit +- iris: Use correct CCS0 aux-map register offset +- intel/genxml: Fix typo in CCS cache flush enable +- intel/genxml: Drop incorrect compute aux-inv register entry +- anv: Drop depth cache flush requirement after depth clear/resolve +- iris: Drop depth cache flush requirement after depth clear/resolve + +Samuel Pitoiset (192): + +- radv: fix detecting FMASK_DECOMPRESS/DCC_DECOMPRESS meta pipelines +- vulkan: ignore rasterizationSamples when the state is dynamic +- radv: try to keep HTILE compressed for READ_ONLY_OPTIMAL layout +- radv: re-emit the guardband state when related PSO are bound +- radv: tidy up dirtying RBPLUS state in radv_bind_dynamic_state() +- radv: disable fast-clears with CMASK for 128-bit formats +- radv: require DRM 3.27 +- radv/amdgpu: remove legacy code path for creating the BO list +- radv/amdgpu: remove legacy code for querying context status +- radv: do not allow 1D block-compressed images with (extended) storage on GFX6 +- radv: fix usage flag for 3D compressed 128 bpp images on GFX9 +- radv: wait for occlusion queries in the resolve query shader +- radv: delay enabling/disabling occlusion queries at draw time +- radv: track DB_COUNT_CONTROL changes to avoid context rolls +- radv: emit PIXEL_PIPE_STAT_CONTROL in the gfx preamble for GFX11 +- radv: use gfx_level in radv_flush_occlusion_query_state() +- radv: update binning settings to work around GPU hangs +- radv/ci: remove one expected test failure on PITCAIRN +- radv/amdgpu: fix adding continue preambles and postambles BOs to the list +- Revert "ci/radv: Demote navi21 to manual until recent flakiness resolves." +- radv: add the perf counters BO to the preambles BO list +- radv: do not overallocate the CS array during submissions +- ac/sqtt: add rgp_sqtt_marker_cb_id definition +- ac/sqtt: add a helper to get cmdbuf IDs per queue +- radv: reserve command buffer index for SQTT +- docs: rename ACO_DEBUG=noscheduling to ACO_DEBUG=nosched +- docs: add missing ACO_DEBUG=force-waitdeps +- radv: only enable extendedDynamicState3ConservativeRasterizationMode on GFX9+ +- ac/spm: introduce ac_spm_trace and ac_spm_get_trace() +- ac/spm: rename ac_spm_trace_data to ac_spm +- ac/sqtt: add a helper for adding clock calibration records +- ac/sqtt: add helpers for initializing ac_thread_trace_data +- ac/sqtt: initialize clock calibration/queue info/queue event records +- radv/sqtt: sample CPU/GPU clocks before starting the trace +- radv/sqtt: add support for queue info +- ac/sqtt: add new bits to rgp_sqtt_marker_barrier_end +- ac/sqtt: add missing EventUnknown to rgp_sqtt_marker_event_type +- ac/rgp: update SQTT_FILE_CHUNK_TYPE_API_INFO to minor version 2 +- ac/rgp: update SQTT_FILE_CHUNK_TYPE_ASIC_INFO to minor version 5 +- ac/sqtt: add ac_sqtt_se_is_disabled() helper +- ac/sqtt: add ac_sqtt_get_trace() helper +- radv: do not abort when the SQTT buffer resize failed +- ac/rgp: remove ac_thread_trace_data from ac_thread_trace +- ac,radv,radeonsi: rename thread_trace to sqtt everywhere +- ac/nir: fix 8-bit/10-bit PS exports clamping +- radv: enable RADV_THREAD_TRACE_CACHE_COUNTERS by default +- radv: fix dynamic depth clamp enable support +- radv: fix invalid type for usage in radv_get_buffer_memory_requirements() +- radv: fix fast-clearing images with VK_REMAINING_{ARRAY_LAYERS,MIP_LEVELS} +- radv: replace radv_get_layerCount by vk_image_subresource_layer_count() +- radv: replace radv_get_levelCount() by vk_image_subresource_level_count() +- radv/meta: rename dest to dst +- radv: disable RB+ blend optimizations on GFX11 when a2c is enabled +- radv: use vk_image::mip_levels instead of radv_image::info::levels +- radv: use vk_image::array_layers instead of radv_image::info::array_size +- radv: use vk_image::samples instead of radv_image::info::storage_samples +- radv: use vk_image::samples instead of radv_image::info::samples +- radv: use vk_image::extent instead of radv_image::info::{width,height,depth} +- radv: remove ac_surf_info from radv_image +- ac/spm: switch to SPM version 2.0 +- vulkan: Update XML and headers to 1.3.250 +- radv: implement VK_EXT_attachment_feedback_loop_dynamic_state +- radv: advertise VK_EXT_attachment_feedback_loop_dynamic_state +- spirv: ignore SpvDecorationInvariant warning on struct members +- radv/ci: stop setting MESA_SPIRV_LOG_LEVEL +- radv: reset the emitted VS prolog when a new vertex shader is bound +- radv: dirty the dynamic vertex input state only when needed +- radv: re-emit fragment shading rate state when PA_CL_VRS_CNTL changes +- radv: configure PA_CL_VRS_CNTL entirely from the cmd buffer +- radv: implement dynamic sample locations enable +- radv: handle NULL fragment shaders when recording cmdbuf +- radv: handle NULL fragment shaders when creating graphics pipelines +- radv: rework the checks for implicit exports with GPL +- radv: allow to determine NGG settings with a NULL fragment shader +- radv: stop compiling a noop FS when the application doesn't provide a FS +- radv: advertise VK_EXT_tooling_info +- radv: reset the emitted PS epilog when a new fragment shader is bound +- radv: remove unused pipeline param in radv_generate_ps_epilog_key() +- radv: stop using the pipeline for determining the null export workaround +- radv: fix emitting VRS state with a null fragment shader +- radv: fix resetting VRS if the graphics pipeline doesn't enable it +- radv: fix a sync issue with primitives generated query and NGG/legacy +- amd/drm-shim: add navi21 +- amd/drm-shim: add pitcairn +- amd/drm-shim: add bonaire +- amd/drm-shim: update README about which file to modify +- ci: build drm-shim in debian-testing +- ci,radv: use drm-shim instead of the null winsys for radv-fossils +- ci: stop using the hang-detection tool for vkd3d-proton +- ci: rework vkd3d-proton runner and fix detecting failures +- radv: reserve cmdbuf space in radv_flush_gfx2ace_semaphore() +- radv: bump the global VRS image size to maximum supported FB dimensions +- radv: disable IMAGE_USAGE_STORAGE with depth-only and stencil-only formats +- radv: remove useless check about USAGE_STORAGE for TC-compat HTILE +- nir: add nir_intrinsic_load_poly_line_smooth_enabled +- radeonsi: lower nir_intrinsic_load_poly_line_smooth_enabled_amd +- nir: lower smooth lines conditionally using the new intrinsic +- radv: track if the smoothLines features is enabled in the device +- radv: determine if smooth lines can be used in the pipeline key +- radv: declare a new user SGPR for the dynamic line rasterization mode +- radv: lower nir_intrinsic_load_poly_line_smooth_enabled_amd +- radv: add support for smooth lines +- radv: enable smoothLines +- radv: apply a bug workaround for smoothing on GFX6 +- radv: do not enable VRS flat shading if the VRS builtin is read +- zink/ci: update VANGOGH expected list of failures +- vulkan/pipeline_cache: remove a bogus assert when inserting objects +- zink/ci: skip arb_texture_buffer_object@texture-buffer-size-clamp* with RADV +- radv: fix copying 2D to 3D images +- ci: uprev vkd3d-proton to 2.9 +- amd: fix 64-bit integer color image clears +- radv: rework configuring VGT_SHADER_STAGES_EN +- radv/ci: update list of expected failures since Vulkan loader 1.3 +- radv/ci: skip tests that timeout since Vulkan loader 1.3 +- vulkan: Update XML and headers to 1.3.251 +- radv: advertise VK_EXT_dynamic_rendering_unused_attachments +- aco: remove nir_intrinsic_load_barycentric_at_sample occurences +- radv/ci: removed expected failures that are skipped now +- radv/nir: use ac_nir_unpack_arg() for packed shader input user SGPRS +- radv: introduce SHIFT/MASK for unpacking shader input args +- radv: regroup fragment shader user SGPRs emission +- radv: merge all FS user SGPRs into one using packed arguments +- spirv: add support for SpvCapabilityFragmentBarycentricKHR +- spirv,nir: add support for BaryCoord{NoPersp}KHR builtins +- spirv,nir: add support for SpvDecorationPerVertexKHR +- nir/lower_io: add nir_intrinsic_load_input_vertex to is_input() +- nir: print locations for per-vertex fragment shader inputs +- zink/ci: remove useless RADV_PERFTEST=gpl +- radv: initialize the device cache UUID even if on-disk cache is disabled +- nir: add nir_intrinsic_load_provoking_vtx_amd +- radv: add support for nir_intrinsic_load_provoking_vtx_amd +- radv: track if the rasterization primitive is known at compile time +- nir: add nir_intrinsic_load_rasterization_primitive_amd +- radv: add support for nir_intrinsic_load_rasterization_primitive_amd +- radv: handle per_vertex variables when gathering FS inputs +- radv: set ROTATE_PC_PTR for custom interpolations +- radv: configure RSRC1.LOAD_PROVOKING_VTX for the fragment shader +- radv: add a NIR pass that lower fragment shader barycentric intrinsics +- radv: gather info about nir_intrinsic_load_sample_positions_amd +- radv: advertise VK_KHR_fragment_shader_barycentric on GFX10.3+ +- radv: add a helper for emitting a null depth/stencil target +- radv: reset more DB registers when emitting a null ds target +- radv: emit DB_RENDER_CONTROL as part of the framebuffer +- radv: disable HTILE compression only when layouts are compressed +- radv/ci: update the list of expected failures on STONEY +- radv: gather info about load_poly_line_smooth_enabled +- radv: add a helper for forcing VRS 1x1 in some situations +- radv: do not force VRS 1x1 when smooth lines are enabled +- radv: fix smooth lines with graphics pipeline library +- radv: fix re-emitting some dynamic states when the previous FS is NULL +- radv: fix re-emitting early_z/late_z when the bound PS changes +- radv: reset some dynamic states when the fragment shader stage is unbound +- radv: remove unused radv_dgc_token struct +- radv: add dgc_emit_state() helper +- radv: add dgc_emit_push_constant() helper +- radv: add dgc_emit_vertex_buffer() helper +- radv: add dgc_emit_draw() helper +- radv: add dgc_emit_draw_indexed() helper +- radv: add dgc_emit_index_buffer() +- radv: do not use IB for the GFX preamble with RADV_DEBUG=noibs +- radv: use IB for the GFX preamble on GFX6 +- radv: reserve space for shadowed regs +- radv/amdgpu: fix a buffer overflow for submissions with RADV_DEBUG=noibs +- radv/amdgpu: remove useless assert in radv_amdgpu_winsys_cs_submit_internal() +- radv/amdgpu: add cs_execute_ib() for executing IBs +- radv: use cs_execute_ib() for GFX, MBCP and DGC IBs +- vulkan/runtime: call CmdSetDepthBias2EXT() from CmdSetDepthBias() +- radv: implement VK_EXT_depth_bias_control +- radv: advertise VK_EXT_depth_bias_control +- radv: implement padding cmdbuffer for DGC on GFX6 +- radv: enable NV_device_generated_commands on GFX6 +- radv: reserve more space in CS for SQTT +- radv/amdgpu: fix dumping cs with RADV_DEBUG=noibs +- radv/amdgpu: dump all cs with RADV_DEBUG=noibs +- radv: only dirty the index type when necessary with DGC +- radv: only dirty the active push constant stages with DGC +- radv: adjust alignment of the preprocess buffer with DGC +- radv/amdgpu: use the correct IB size when growing a CS with RADV_DEBUG=noibs +- radv/amdgpu: rework growing a CS with the chained IB path slightly +- radv/amdgpu: do not set the IB size when ending a CS with RADV_DEBUG=noibs +- radv/amdgpu: use the array of IB buffers for the chained IB path +- radv/amdgpu: use cs_finalize() when growing a CS +- radv/amdgpu: rename old_ib_buffers to ib_buffers +- radv/amdgpu: add a helper to get a new IB +- radv/amdgpu: skip adding per VM BOs for sparse during CS BO list build +- radv/amdgpu: workaround a kernel bug when replacing sparse mappings +- radv/amdgpu: add more small helpers for managing CS +- radv/amdgpu: add support for executing DGC cmdbuf with RADV_DEBUG=noibs +- radv: allow NV_device_generated_commands with RADV_DEBUG=noibs +- radv: stop emitting TILE_SURFACE_ENABLE for the ZRANGE_PRECISION workaround +- radv: inline more values in radv_emit_fb_ds_state() +- radv: emit PA_SC_SCREEN_SCISSOR_BR with the actual fb extent + +Sarah Walker (15): + +- pvr: Support single core transfer queue commands on multicore GPUs +- pvr: Implement pvr_pbe_setup_modify_defaults() +- pvr: Complete pvr_modify_command() +- pvr: Complete pvr_unwind_rects() +- pvr: Complete pvr_double_stride() +- pvr: Implement pvr_isp_scan_direction() +- pvr: Implement pvr_reroute_to_clip() +- pvr: Support ipf_creq_pf in pvr_isp_ctrl_stream() +- pvr: Complete pvr_3d_validate_addr() +- pvr: Support multiple sources per pass in TQ job submission +- pvr: Complete pvr_generate_custom_mapping() +- pvr: Fragment register fb_cdc_zls is feature dependent +- pvr: use pvr_csb_pack() to setup CR_FB_CDC_ZLS +- pvr: Rename heap reserved area to static data carveout +- pvr: Merge main and extension command streams + +Sathishkumar S (5): + +- util/format: add planar3 r8_g8_b8_unorm pipe format +- frontends/va: add support for RGBP rt_format +- radeonsi/vcn: enable RGBP format on gfx940 jpeg +- radeonsi/vcn: engage all jpeg engines on gfx940 for mjpeg decode +- frontends/va: return matching drm format for yuyv pipe format + +Semjon Kravtsenko (1): + +- glx: Assign unique serial number to GLXBadFBConfig error + +Sergi Blanch Torne (8): + +- ci: Move Vulkan CTS patches to their own directory +- ci: disable Collabora's LAVA lab for maintance +- Revert "ci: disable Collabora's LAVA lab for maintance" +- ci: Allow zink-radv jobs to be manual when uprev piglit +- ci: disable Collabora's LAVA lab for maintance +- Revert "ci: disable Collabora's LAVA lab for maintance" +- ci: disable Collabora's LAVA lab for maintance +- Revert "ci: disable Collabora's LAVA lab for maintance" + +Shan-Min Chao (1): + +- tu/kgsl: Fix memory overwrite with vkFlushMappedMemoryRanges when more than 1 range + +Sil Vilerino (29): + +- d3d12: Do not fail d3d12_screen creation if D3D12_FEATURE_D3D12_OPTIONS14 not available +- frontend/va: Support QVBR rate control mode +- frontend/va: Allow distinction for HRD params sent from app and frontend defaults +- frontend/va: Allow distinction for Min/MaxQP params sent from app and frontend defaults +- d3d12: Support QVBR rate control mode +- d3d12: Support rate control HRD and MaxFrameSize app params +- d3d12: Support QPMin/QPMax app params +- d3d12: Support PIPE_VIDEO_CAP_MIN_WIDTH/HEIGHT caps +- d3d12: Support PIPE_VIDEO_CAP_ENC_QUALITY_LEVEL +- frontend/va: Add VAProfileH264High10 +- frontend/va: Add H264 decode slice data +- d3d12: Use frontend H264 decode slice offsets and sizes instead of parsing buffer +- d3d12: Clean unused code for parsing slices +- frontends/va: Extend AV1 Encode params +- d3d12: AV1 Encode +- CI/windows: Update headers and Agility redist to 1.711.3-preview +- d3d12: Correct tx_mode_support reporting as specified in libva spec +- d3d12: Only set reduced_tx_set when supported by D3D12 caps (no libva caps for reduced_tx_set to map to) +- d3d12: Fix usage of D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG, was using D3D12_VIDEO_ENCODER_SUPPORT_FLAG wrongly instead +- frontend/va: Pass surf->fence in PIPE_VIDEO_ENTRYPOINT_ENCODE contexts for driver to wait on input surface pending work +- frontend/va: Add video processing async fence support +- d3d12: Video Decode - Implement get_decoder_fence and async queing +- d3d12: Apply style format to d3d12_video_dec.cpp +- d3d12: Video Decode - Sync 3D context copy with decode work for texture array case +- d3d12: Video Encode - GPU wait on input surface fence +- d3d12: Video Process - Implement get_processor_fence and async queing +- d3d12: Video Decode - Refactor and style fixes +- frontend/va: Fix vaSyncSurface and vaQuerySurface status for drivers not implementing get_processor_fence +- frontend/va: Remove fence_server_sync for surface in vlVaHandleVAProcPipelineParameterBufferType + +Simon Perretta (5): + +- pvr: Amend validation when checking multiple supported types +- pvr: Use movc for reading special registers +- pvr: Add support for generating transfer fragment programs +- pvr: Add support for generating transfer EOT programs +- pvr: Use driver vertex input data in the compiler + +Simon Ser (4): + +- wayland: generalize wayland-protocols code generation +- radv: advertise LINEAR filter support for multiplanar/subsampled +- vulkan/wsi/wayland: add 16-bit formats +- Update OpenGL headers + +SoroushIMG (8): + +- zink: do not emit line stipple dynamic state when emulating +- zink: take location_frac into account in lower_line_smooth_gs +- zink: fix incorrect line mode check for bresenham +- zink: refcount the correct query pool +- pvr: fix sync waiting while using pvrsrvkm +- pvr: fix infinite recursion in pvr_cmd_buffer_{start,end}_sub_cmd +- pvr: add missing frag to geom dependency for jobs targetting same render target +- pvr: Fix barrier insertion on merged subpasses + +SureshGuttula (2): + +- va/surface : Add Nv12 support for PRIME_2 imports +- radeonsi/vcn: update luma and chroma size + +Sviatoslav Peleshko (5): + +- isl: Check all channels in isl_formats_have_same_bits_per_channel +- anv: Handle UNDEFINED format in image format list +- anv: Improve image/view usage bits verification +- nir/lower_shader_calls: Fix cursor if broken after nir_cf_extract() call +- glsl: Fix yylloc.source propagation in YYLLOC_DEFAULT + +Tapani Pälli (30): + +- isl: disable mcs (and mcs+ccs) for color msaa on gfxver 125 +- iris: implement state cache invalidate for Wa_16013063087 +- anv: cleanup bitmask construction for PIPELINE_SELECT +- anv: implement state cache invalidate for Wa_16013063087 +- isl: fix layout for comparing surf and view properties +- egl/loader: move crtc resource infrastructure as common helper +- anv: handle missing astc for gfx125 in CreateImageView +- mesa: set a type for depth fallback texture +- intel/dev: provide helper to check if devinfo is ATS-M +- anv: add required invalidate/flush for Wa_14014427904 +- iris: add required invalidate/flush for Wa_14014427904 +- mesa: validate shader binary format in _mesa_spirv_shader_binary +- iris: make Wa_16013994831 to use intel_needs_workaround +- anv: make Wa_16013994831 to use intel_needs_workaround +- anv: remove BDW specific WA for CS stall enable +- intel/dev: add parentheses around intel_needs_workaround macro +- iris: use workaround framework for 1408224581, 14014097488 +- anv: use workaround framework for 1408224581, 14014097488 +- anv: wrap pipe control emission to a set of helper functions +- anv: implement flush part of emit_apply_pipe_flushes with helper +- anv: implement invalidate part of emit_apply_pipe_flushes with helper +- anv: convert genX_query pipe controls to use pc helper +- anv: change pipe controls in genX_state to use pc helper +- anv: change pipe control in genX_pipeline to use pc helper +- anv: change pipe controls in genX_gpu_memcpy to use pc helper +- anv: change pipe control in indirect draw gen to use pc helper +- anv: change most pipe controls in gfx8_cmd_buffer to use pc helper +- anv: convert most pc in genX_cmd_buffer to use pc helper +- isl: handle DRM_FORMAT_MOD_INVALID in isl_drm_modifier_has_aux +- intel/compiler: add more validation for acc register usage + +Tatsuyuki Ishi (17): + +- util: Add dedicated hex conversion functions and use it. +- util: Call mesa_bytes_to_hex directly instead of disk_cache_format_hex_id. +- util: Add a copy of BLAKE3 hash library. +- util/blake3: Patch with hidden visibility for asm symbols. +- util: Add mesa_blake3 wrappers. +- nir: Fix serializing pointer initializers. +- radv: Make shader related destruction happen before hw_ctx. +- radv: Add RGP barrier markers for render pass transition and copy. +- radv: Guard against misplaced RGP barrier markers. +- util/blake3: Add blake3_hash typedef. +- vulkan: Migrate shader module hash to BLAKE3. +- vulkan/pipeline_cache: Do not consume object passed into remove_object. +- vulkan/pipeline_cache: Move locking outside of remove_object. +- vulkan/pipeline_cache: Move cache_object_unref out of header. +- vulkan/pipeline_cache: Introduce weak reference mode. +- radv: Enable weak reference cache for device->mem_cache. +- zink/ci: Add ext_transform_feedback@api-errors to fail list. + +Teng, Jin Chung (1): + +- d3d12: HEVC Encode - Fix num_subregions_per_scanline rounding + +Thomas H.P. Andersen (30): + +- nir/nir_lower_wpos_center: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_wpos_ytransform: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_viewport_transform: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_var_copies: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_uniforms_to_ubo: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_two_sided_color: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_to_source_mods: Use the nir_shader_instructions_pass() helper +- nir/nir_lower_vec3_to_vec4: Use the nir_shader_instructions_pass() helper +- r600: remove unused code +- tgsi: delete unused functions +- aux: remove unused tgsi includes +- d3d12: remove unused tgsi includes +- etnaviv: remove unused tgsi includes +- freedreno: remove unused tgsi includes +- i915: remove unused tgsi includes +- llvmpipe: remove unused tgsi includes +- nouveau: remove unused tgsi includes +- r300: remove unused tgsi includes +- r600: remove unused tgsi includes +- radeonsi: remove unused tgsi includes +- softpipe: remove unused tgsi includes +- svga: remove unused tgsi includes +- v3d: remove unused tgsi includes +- vc4: remove unused tgsi includes +- virgl: remove unused tgsi includes +- zink: remove unused tgsi includes +- lavapipe: remove unused tgsi includes +- st: remove unused tgsi includes +- r600: tgsi cleanup +- tgsi: remove unused functions and structs + +Thong Thai (11): + +- gallium/pipe: add min width and min height video cap enums +- radeonsi: return min width and min height video cap values +- frontends/va: report min width and min height values if available +- mesa/main: rework locale setup/teardown +- util: check and initialize locale before using it +- tgsi: use locale independent float and double parsing +- frontends/va/config: add disable packed headers as valid config +- frontends/va/context: check min supported resolution when creating +- frontends/va/config: check for QVBR support when creating +- frontends/va/context: return error if context_id == 0 +- frontends/va: fix some coverity scan reported issues + +Tim Pambor (1): + +- virgl: Fix stack overflow in virgl_bind_sampler_states + +Timothy Arceri (30): + +- util: add Pixel Game Maker MV workaround +- util: add Jamestown+ workaround +- st/glsl: move linking code to the same st file +- glsl: call nir_opt_find_array_copies() when linking +- glsl: port lower_blend_equation_advanced() to nir +- glsl: call nir version of lower_blend_equation_advanced() +- glsl: remove old lower_blend_equation_advanced() code +- glsl: add some more c wrappers for string_to_uint_map +- mesa: add some new constants +- glsl: move some compiler code out of st +- glsl: move lowering linker code out of st +- glsl: port assign location code for VS inputs or FS outputs +- glsl: call assign_attribute_or_color_locations() in NIR linker +- glsl: remove unused buffer objects with packed layout +- glsl: remove unused system vars +- glsl: drop the dce of global vars from GLSL IR linker +- nir/glsl: add nir_var_declared_implicitly enum +- glsl: move disable_varying_optimizations_for_sso() to NIR linker +- glsl: remove the always_active_io flag from GLSL IR +- glsl: inline link_varyings() +- glsl: set last_vert_prog in the nir linker +- glsl: drop link_invalidate_variable_locations() +- glsl: move store_fragdepth_layout() to nir linker +- glsl: remove glsl ir optimisation loop from linker +- st/glsl: merge link_shader() into st_link_nir() +- st/glsl: merge st_link_glsl_to_nir() into st_link_nir() +- st/glsl: merge st_glsl_to_ir.cpp with st_glsl_to_nir.cpp +- glsl: remove dead varyings before assigning attr locations +- glsl: do vs attribute validation in NIR linker +- glsl: fix validation of ES vertex attribs + +Timur Kristóf (56): + +- radv/amdgpu: Remove unnecessary assertions from chaining. +- radv: Disallow IB2 on GFX6 when using draw_indirect_multi. +- radv: Use IB BOs (chaining) by default on GFX6. +- radv: Chain command buffers on GFX6 in radv_queue. +- amd: Rename INDIRECT_BUFFER_CIK to just INDIRECT_BUFFER. +- radv: Simplify IB2 workaround. +- radv: Remove IB2 workaround from mesh shader draws. +- radv: Enable IB2 workaround on all indirect draws. +- radv: Fix dword alignment in SDMA buffer copy. +- aco: Disallow constant propagation on SOPP and fixed operands. +- amd: Add and implement sendmsg_amd intrinsic. +- amd: Add and implement gs_wave_id sysval. +- amd: Move sendmsg defines to ac_shader_util. +- ac/llvm: Clarify arguments of ac_build_sendmsg. +- ac/nir: Use sendmsg in legacy GS lowering. +- ac/nir: Emit legacy GS DONE signal in NIR. +- ac/nir/ngg: Use sendmsg in NGG lowering. +- amd: Cleanup old GS intrinsics code. +- aco: Don't allow any VALU instruction to write m0. +- aco: Initialize vcmpx field in get_cmp_info. +- radv/amdgpu: Remove unused extra BO array. +- radv/amdgpu: Split radv_amdgpu_get_bo_list to smaller functions. +- radv/amdgpu: Pass preambles to get_bo_list. +- radv/amdgpu: Use STACK_ARRAY for IB array to reduce stack usage. +- radv: Move perf counter CS creation to where it's used. +- ac: Use const keyword for some function arguments. +- radv: Use const keyword more. +- radv: Emit primitive reset index with primitive restart enable. +- radv: Compute tess info when emitting patch control points. +- radv: Move ignore forced VRS code to more optimal place. +- radv: Set last_index_type in radv_before_draw. +- radv: Slight refactor to late_scissor_emission. +- radv: Move indirect check from index buffer emission to caller. +- radv: Move empty dynamic states check to caller. +- radv: Clear query dirty flags when flushing them. +- radv: Clarify gang submit terminology. +- radv: Use RESET_FILTER_CAM for some mesh shading draws. +- aco: Mark exec write used when it writes other registers. +- radv: Remove primitive reset index from late scissor workaround. +- radv: Leave primitive reset index at max on GFX8+. +- ac: Add ac_hw_stage enum. +- aco: Use ac_hw_stage instead of aco-specific HWStage. +- aco: Add hw_stage field to aco_shader_info. +- radeonsi: Set aco_shader_info::hw_stage +- radv: Set aco_shader_info::hw_stage +- aco: Use aco_shader_info::hw_stage instead of guessing. +- aco: Remove unneeded stage related info fields. +- ac/nir/ngg: Call nir_convert_to_lcssa before divergence analysis. +- ac/nir/ngg: Add upper limit to reusable uniforms. +- ac/nir/ngg: Follow intrinsic sources when analyzing before culling. +- ac/nir/ngg: Follow tex sources when analyzing before culling. +- radv: Refactor required subgroup size in pipeline key. +- radv: Use required subgroup info for graphics shaders. +- radv: Enable required subgroup size on mesh/task. +- aco: Add MESA_SHADER_KERNEL to instruction selection setup. +- aco: Fix subgroup_id intrinsic on GFX10.3+. + +Tony Wasserka (2): + +- aco/spill: Use arena allocator for next use distances +- aco/spill: Use arena allocator for spills + +Veerabadhran Gopalakrishnan (2): + +- radeonsi: return kernel queried video capability for HEVC and JPEG +- radeonsi: return kernel queried video capability for HEVC and JPEG + +Viktoriia Palianytsia (1): + +- iris,crocus: Add proper way of assigning num_levels value + +Vinson Lee (9): + +- r600/sfn: Initialize BlockScheduler member m_chip_family. +- freedreno/a6xx: Fix memory leak on error path. +- nv50: Fix memory leak in error path +- pvr: Fix signed comparison +- dzn: Fix qpool->queries_lock double lock +- tu: Fix missing unlock +- vulkan/wsi: Remove duplicate NULL check +- frontends/va: Fix missing unlock +- r600/sfn: Remove duplicate assignment + +Vitaliy Triang3l Kuzmin (27): + +- lavapipe: Fix vk_instance_init vk_error instance use-after-free +- radv: Fix vk_instance_init vk_error instance use-after-free +- radv: Move most of DB_SHADER_CONTROL to PS, more precise GFX11 blend WA +- docs/amd: Document Primitive Ordered Pixel Shading +- ac/nir: Support Primitive Ordered Pixel Shading in lower_ps +- aco: Support pops_exiting_wave_id PhysReg usage +- ac: Define POPS collision wave ID argument SGPR +- aco: Add s_wait_event argument bit definitions +- aco: Add Primitive Ordered Pixel Shading pseudo-instructions +- aco: Skip waitcnt insertion in the discard early exit block +- aco: Add Primitive Ordered Pixel Shading scheduling rules +- aco: Send MSG_ORDERED_PS_DONE where necessary +- aco: Add Primitive Ordered Pixel Shading waitcnt rules +- aco: Implement fragment shader interlock intrinsics +- radeonsi: Remove unconditional POPS_DRAIN_PS_ON_OVERLAP setting +- radv: Remove unconditional POPS_DRAIN_PS_ON_OVERLAP setting +- radv: Detect the use of Primitive Ordered Pixel Shading +- radv: Ensure 1x1 shading rate on GFX10.3 with interlock execution mode +- radv: Declare POPS collision wave ID shader argument +- radv: Enable POPS collision wave ID shader argument +- radv: Enable the null export workaround with POPS +- radv: Handle Primitive Ordered Pixel Shading in DB_SHADER_CONTROL +- ac/gpu_info: Check whether the device has the POPS missed overlap bug +- radv: Apply the POPS missed overlap hardware bug workaround +- radv: Disable VRS forcing with Primitive Ordered Pixel Shading +- zink/ci: Add broken fragment shader interlock test to RADV flakes +- radv: Enable VK_EXT_fragment_shader_interlock + +Víctor Manuel Jáquez Leal (1): + +- vulkan: complete the usage flags for video layouts + +Weibin Wu (1): + +- winsys/gdi: GDI B5G6R5 display target support + +Xi Ruoyao (1): + +- Revert "glx: Remove pointless GLX_INTEL_swap_event paranoia" + +Yiwei Zhang (40): + +- radv: respect VK_QUERY_RESULT_WAIT_BIT in GetQueryPoolResults +- venus: stop query experimental features +- venus: adopt venus protocol release +- meson/ci: promote virtio-experimental to virtio +- docs: update Virtio-GPU Venus driver page +- ci: carry venus-protocol 1.0 release patches in virglrenderer +- ci: uprev virglrenderer to drop venus release patches +- anv: apply ANV_BO_ALLOC_IMPLICIT_SYNC for external memory +- pipe-loader: avoid undefined memcpy behavior +- lvp: avoid accessing member of NULL ptr for global entries +- venus: bump ring space to 128K +- docs/venus: update vtest instructions +- radv: fix radv_emit_userdata_vertex for vertex offset -1 +- venus: silence -Wuninitialized +- venus: sync to latest protocol from header v1.3.248 +- venus: sync protocol for VK_EXT_image_2d_view_of_3d +- venus: enable VK_EXT_image_2d_view_of_3d +- docs/venus: advertise VK_EXT_image_2d_view_of_3d +- venus: temporarily disable VK_EXT_memory_budget +- venus: refactor vn_device_memory to track VkMemoryType +- venus: handle device memory report requests +- venus: emit device memory report for device memory events +- venus: enable VK_EXT_device_memory_report +- docs: update venus VK_EXT_device_memory_report support +- anv: avoid requiring ordered memory planes for explicit import +- venus: suballocate feedback slot with feedback buffer alignment +- venus: refactor ahb buffer mem type bits cache to be lazy +- venus: refactor buffer cache related bits +- venus: extend VkBuffer cache to cover concurrent sharing +- venus: fix a cmd tmp storage leak +- venus: fix leaks from tracked present src images +- venus: track pool in cmd and track device in pool +- venus: cmd to reuse alloc copy from cmd pool +- venus: refactor vn_cmd_add_query_feedback and miscs +- venus: cache query batches at cmd pool +- venus: refactor query batch handling +- venus: recheck valid bit after acquiring lock to init ahb mem type bits +- venus: handle query feedback creation failure +- venus: ensure consistency of query overflow behavior +- venus: add a missing barrier before copying query feedback + +Yogesh Mohan Marimuthu (2): + +- ac/gpu_info: num_cu = 4 and gfx11 enable dcc with retile +- ac/gpu_info: rearrange if checks for dcc config + +Yonggang Luo (121): + +- loader: Replace usage of mtx_t with simple_mtx_t in loader/loader_dri3_helper.c +- v3d: Replace usage of mtx_t with simple_mtx_t in v3d_simulator.c +- vc4: Replace usage of mtx_t with simple_mtx_t in vc4/vc4_simulator.c +- drm-shim: Replace usage of mtx_t with simple_mtx_t in drm_shim.c +- drm: Replace usage of mtx_t with simple_mtx_t in virgl/drm/virgl_drm_winsys.c +- drm: Replace usage of mtx_t with simple_mtx_t in drm/radeon_drm_winsys.c +- drm: Replace usage of mtx_t with simple_mtx_t in nouveau_drm_winsys.c +- hud: Replace usage of mtx_t with simple_mtx_t in hud_cpufreq.c +- hud: Replace usage of mtx_t with simple_mtx_t in hud_diskstat.c +- hud: Replace usage of mtx_t with simple_mtx_t in hud_nic.c +- hud: Replace usage of mtx_t with simple_mtx_t in hud_sensors_temp.c +- xlib: Replace usage of mtx_t with simple_mtx_t in xm_api.c +- rtasm: Trim trailing spaces and replace tab with 3 space +- rtasm: Replace usage of mtx_t with simple_mtx_t in rtasm_execmem.c +- nine: Replace usage of mtx_t with simple_mtx_t in nine_lock.c +- omx: Replace usage of mtx_t with simple_mtx_t in vid_omx_common.c +- vdpau: Replace usage of mtx_t with simple_mtx_t in htab.c +- c11: Remove _MTX_INITIALIZER_NP as it's not used anymore +- microsoft/compiler: Getting function impl to be consistence with decl in dxil_enums.* +- compiler: Getting shader_prim to be PACKED that consistence with pipe_prim_type +- compiler: Add SHADER_PRIM_COUNT to be SHADER_PRIM_MAX + 1 +- compiler: Rename shader_prim to mesa_prim and replace all usage of pipe_prim_type with mesa_prim +- docs: Update document about pipe_prim_type with mesa_prim +- util: Replace all usage of PIPE_TIMEOUT_INFINITE with OS_TIMEOUT_INFINITE +- r300: Replace usage of os_get_process_name with util_get_process_name in r300_chipset.c +- virgl: Array cmdline on stack should initialized to 0 +- virgl: Replace the usage of os_get_process_name with util_get_process_name +- compiler: Combine duplicated implementation of is_gl_identifier into glsl_types.h +- compiler: Move can_implicitly_convert_to helper to glsl module from glsl_types.h +- mesa, compiler: Move gl_texture_index to glsl_types.h +- compiler: Remove the need include "util/glheader.h" and "util/ralloc.h" in glsl_types.h +- compiler: Remove redundant struct glsl_type in nir_types.h +- vulkan: move nir_convert_ycbcr into vulkan runtime +- util: Remove redundant type cast in function align64 +- util: use uint32_t as the parameter of align function +- util: Do not use align as variable name +- compiler: use align instead glsl_align and remove glsl_align +- panfrost: Replace the usage of PIPE_BIND_* with PAN_BIND_* +- ac: Replace the usage of pipe_compare_func with compare_func +- dri: Replace usage of boolean/TRUE/FALSE with bool/true/false +- freedreno: Fixes error: passing argument 1 of ‘pthread_mutex_unlock’ from incompatible pointer type in tu_pipeline.c +- wsi: Fixes passing argument 1 of ‘mtx_unlock’ from incompatible pointer type +- c11: Improve timespec_get to support TIME_MONOTONIC TIME_ACTIVE TIME_THREAD_ACTIVE TIME_MONOTONIC_RAW +- c11: Improve mtx_timedlock to use timespec_get instead of time(NULL) +- c11: Implement os_time_get_nano with timespec_get(&ts, TIME_MONOTONIC) +- zink: Replace the usage of os_get_process_name with util_get_process_name +- dd: Replace the usage of os_get_process_name with util_get_process_name in dd_draw.c +- gallium: Remove unused os_process.h in gallium/auxiliary +- util: Fixes prototype of threads_timespec_compare +- mapi: Fixes check_table.cpp for DrawArraysInstancedARB and DrawElementsInstancedARB +- meson: Use consistence disabled/enabled comment for shared-glapi option +- mapi: Fixes non-constant-expression cannot be narrowed from type 'unsigned long' to 'unsigned int' in initializer list with clang +- meson: Guard the glsl tests that only working when OpenGL ES2 is enabled +- draw: Replace usage of boolean/TRUE/FALSE with bool/true/false in draw_pt_vsplit* +- draw: Replace usage of ubyte/ushort/uint with uint8_t/uint16_t/uint32_t in draw_pt_vsplit.c +- draw: Update the comment and function name to match the type +- vtn: Do not assign main_entry_point->impl twice +- nir: Add function nir_function_set_impl +- hud: Use bool/true/false to replace boolean/TRUE/FALSE in hud/hud_context.c +- gallium/draw: Replace the usage of ushort to uint16_t in files that can not found by tools +- llvmpipe: altivec.h inclusion in -std=c++98..11 causes bool to be redefined +- treewide: replace usage of boolean to bool +- treewide: style fixes after replace usage of boolean to bool +- treewide: Replace the usage of TRUE/FALSE with true/false +- treewide: Replace the usage of ubyte/ushort with uint8_t/uint16_t +- treewide: style fixes after replace the usage of ubyte/ushort with uint8_t/uint16_t +- util: Merge p_compiler.h into src/util/compiler.h +- util: include "util/compiler.h" instead of "pipe/p_compiler.h" +- mapi: Fixes compile error with build option "-D shared-glapi=disabled" +- mapi: Now _glapi_get_dispatch_table_size always equal to sizeof(struct _glapi_table) / sizeof(void \*) +- mapi: Hide OpenGL functions to be exported when shared-glapi is disabled +- ci: Testing -D shared-glapi=disabled with debian-clang-release +- d3d12: Fixes unused-variable compile error +- compiler: set alignment=1 by default for handling empty struct/interface in glsl_types.cpp +- util: Add function util_is_power_of_two_nonzero64 in bitscan.h +- util: use uint32_t instead of unsigned in bitscan.h +- util: Getting align and align64 consistence with ALIGN +- util: Replace the usage of redundant u_align_u32 with align and remove u_align_u32 +- util: Do not use align64 over unsigned int in register_allocate.c +- util: sizeof bucket are always 32bit width, use align instead align64 +- mapi: Style fixes in glapi/glapi_getproc.c +- mapi: Merge get_static_proc_address into _glapi_get_proc_address +- mapi: Remove dead struct _glapi_function in glapi/glapi_getproc.c +- nir: Split macro nir_foreach_function_with_impl out of nir_foreach_function_impl +- clang-format: Add nir_foreach_function_with_impl into src/.clang-format +- treewide: Switch to use nir_foreach_function_with_impl when possible +- clang-format: Add nir_foreach_function_impl into src/.clang-format +- gallium/auxiliary: Switch to use nir_foreach_function_impl +- asahi: Use nir_foreach_function_impl instead nir_foreach_function in function agx_nir_lower_zs_emit +- d3d12: Switch to use nir_foreach_function_impl +- glsl: Switch to use nir_foreach_function_impl from nir_foreach_function +- glsl: Remove the extra scope in gl_nir_link_uniforms.c +- crocus: Switch to use nir_foreach_function_impl +- intel/compiler: Switch to use nir_foreach_function_impl +- broadcom: replace redefined ALIGN() macro with common util functions +- util: Remove redundant defined(_WIN32) in u_string.h +- util: Remove redundant #if !defined(XF86_LIBC_H) in u_string.h +- nir: Strip the const modifier on nir_function * in nir_foreach_function_with_impl +- panfrost: Convert to use nir_foreach_function_with_impl in function midgard_compile_shader_nir +- panfrost: Convert to use nir_foreach_function_impl when possible +- mesa: Convert to use nir_foreach_function_impl +- llvmpipe: Convert to use nir_foreach_function_impl +- sfn: Convert to use nir_foreach_function_impl +- sfn: indent fixes after switch to use nir_foreach_function_impl +- compiler/clc: Switch to use nir_foreach_function_impl in function nir_lower_libclc +- dxil: Use nir_remove_non_entrypoints +- nir: Update the comment to call nir_remove_non_entrypoints directly +- glsl: Use nir_remove_non_entrypoints to simplify the code +- radv: Use nir_remove_non_entrypoints in radv_shader.c +- nir: Add nir_foreach_function_safe and use it +- pvr: Use alignas instead of ALIGN_ATTR and remove ALIGN_ATTR +- vc4: Convert to use nir_foreach_function_impl when possible +- v3d: Switch to use nir_foreach_function_impl +- broadcom: Switch to use nir_foreach_function_impl +- radeonsi: Use ALIGN_POT instead ALIGN_TO +- etnaviv: Convert to use nir_foreach_function_impl +- intel/vulkan: Convert to use nir_foreach_function_impl when possible +- iris: Convert to use nir_foreach_function_impl +- treewide: Remove all usage of nir_builder_init with nir_builder_create and nir_builder_at +- treewide: remove unused nir_builder +- nir: Remove nir_builder_init, it's not used anymore + +Zhang Ning (1): + +- lima: use u_pipe_screen_lookup_or_create in the renderonly path too + +Zhang, Jianxun (3): + +- intel/isl: Fix map between sRGB and linear formats +- anv: Support 1MB AUX mapping (MTL) +- anv: Remove alignment to aux ratio on size of main surface + +antonino (26): + +- zink: don't emulate edgeflags for patches +- zink: use correct primitives for passthrough gs with tess +- zink: add \`single_sample` to fs key +- zink: add to multisample field to \`zink_gfx_pipeline_state` +- zink: don't render with multisampling when it is disabled +- zink/ci: remove xt_framebuffer_multisample-interpolation fail +- zink: fix pv mode lowring index calculation +- zink: use ring buffer to preserve last element +- zink: fix exit condition on pv emulation loop +- zink: fix line strip offsets in pv mode emulation +- nir/zink: use sysvals in \`nir_create_passthrough_gs` +- zink: fix store subsitution in \`lower_pv_mode_gs_store` +- zink: set when pipeline dirty flag when multisample changes +- Revert "zink: set when pipeline dirty flag when multisample changes" +- Revert "zink/ci: remove xt_framebuffer_multisample-interpolation fail" +- Revert "zink: don't render with multisampling when it is disabled" +- Revert "zink: add to multisample field to \`zink_gfx_pipeline_state`" +- Revert "zink: add \`single_sample` to fs key" +- zink: take location_frac into account in pv emulation +- nir: use \`nir_variable_clone` in \`nir_create_passthrough_gs` +- nir: don't create invalid inputs in \`nir_create_passthrough_gs` +- zink: don't replace non generated gs +- nir: handle interface blocks in \`copy_vars` +- zink: handle interface blocks in \`copy_vars` +- nir: make var arrays large enough in \`nir_create_passthrough_gs` +- zink: don't create invalid inputs in \`zink_create_quads_emulation_gs` + +i509VCB (1): + +- docs/asahi: Add hardware glossary + +nihui (1): + +- panvk: port panvk_logi to vk_logi + +norablackcat (24): + +- rusticl: implement cl_khr_pci_bus_info +- docs/rusticl: add Contributing section +- rusticl/types add ::new for cl_dev_idp_accel_props +- rusticl/api: add integer_dot_product api +- rusticl/clc add integer_dot_prod feature macros +- rusticl/kernel: remove nir_lower_pack pass +- rusticl/device: add cl_khr_integer_dot_product ext +- rusticl/program: fix clippy cast to the same type +- rusticl/types: fix clippy new() not returning Self +- rusticl/screen: implement uuid wrapper funcs +- rusticl/device: implement cl_khr_device_uuid +- rusticl/screen: fix driver_uuid on non x86 +- rusticl: add cl_khr_create_command_queue +- docs/features update opencl extensions add rusticl +- docs: rusticl envvars list supported drivers +- rusticl/memory: fix clippy errors +- gallium: add PIPE_CAP_TIMER_RESOLUTION +- llvmpipe/screen: add PIPE_CAP_TIMER_RESOLUTION +- sofpipe/screen: add PIPE_CAP_TIMER_RESOLUTION +- crocus/screen: add PIPE_CAP_TIMER_RESOLUTION +- iris/screen: add PIPE_CAP_TIMER_RESOLUTION +- r600/pipe: add PIPE_CAP_TIMER_RESOLUTION +- radeonsi/get: add PIPE_CAP_TIMER_RESOLUTION +- zink/screen: add PIPE_CAP_TIMER_RESOLUTION + +xurui (6): + +- zink: Some return values of malloc should be checked +- zink: Use malloc instead of ralloc +- zink: Use malloc to allocate libs +- zink: Add some printfs when initialization fails +- zink: Free the cdt when an error occurs +- zink: The result should be assigned a value when returned