Mesa 23.1.7 Release Notes / 2023-09-06

Mesa 23.1.7 is a bug fix release which fixes bugs found since the 23.1.6 release.

Mesa 23.1.7 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 23.1.7 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

409641eadf0ed1c7794797a6f5a0b0195b5580b282166e5ec5629c6bcda6acd3  mesa-23.1.7.tar.xz

New features

  • None

Bug fixes

  • [Google][Rex][anv] GLES dEQP test fails in anv when run via ANGLE-on-Venus on ChromeOS ARCVM.

  • VAAPI on VCN: bad stream may crash whole gfx system

  • VAAPI: AMDGPU crash on RX 6900 XT on corrupted video

  • aco: Assertion when compiling CP2077 shader

  • anv: flakiness on tgl+ with samplemask handling

  • [RADV] Dead by Daylight memory leak (shader-related?) on 23.1.6

  • gpu hang on DG2 when running KHR-GLES31.core.texture_cube_map_array.image_op_tess*

  • KHR-GLES31.core.texture_cube_map_array.image_op_tessellation_evaluation_sh fail on GFX12+

  • radv: commit 81641b01555faa4dd1dfc7de2513ad8d63e77ab7 leaded to artifacts in Quake II RTX

  • [radv] Colors are distorted in Cyberpunk 2077 with ray tracing enabled

  • Forza Horizon 5 stuttering since mesa 23.1.4 / 9b008673 revert as a FIX

  • glCopyTexSubImage2D is very slow on Intel

  • [RADV] red and pink tinted shadows in Overwatch 2 on 7900 XTX

Changes

Alyssa Rosenzweig (1):

  • nir/passthrough_gs: Fix array size

Chia-I Wu (1):

  • ac/surface: limit RADEON_SURF_NO_TEXTURE to color surfaces

Corentin Noël (1):

  • virgl: Do not expose EXT_texture_mirror_clamp when using a GLES host

Daniel Schürmann (1):

  • nir/opt_move: fix handling of if-condition

David Heidelberg (1):

  • ci: disable Material Testers.x86_64_2020.04.08_13.38_frame799.rdc trace

David Rosca (1):

  • Revert “radeonsi/vcn: add an exception of field case for h264 decoding”

Dmitry Baryshkov (1):

  • tu: Pass real size of prime buffers to allocator

Emma Anholt (2):

  • freedreno/a3-5xx: Don’t try to emit ISAM for SSBO loads.

  • disk_cache: Disable the “List” test for RO disk cache.

Eric Engestrom (22):

  • docs: add sha256sum for 23.1.6

  • .pick_status.json: Update to 9d442b459a43264c28994e65dcaacfe303feedd1

  • .pick_status.json: Mark aebe58458611e0bb585a5bce8e16c1175783f3cc as denominated

  • .pick_status.json: Update to 61dd0ff97d8d670e1b7dfea8a37e73b67a7f5ff3

  • .pick_status.json: Update to 5fa9f842b09694a91310640ae949327a74d1bcd2

  • vc4: drop duplicate .lower_ldexp

  • zink: fix format in zink_make_{image,texture}_handle_resident()

  • .pick_status.json: Mark 04ba4059b7243fb9258df12154ce1e7a01e019c8 as denominated

  • .pick_status.json: Mark fe4208ed4c9a08c7963caa287ae34e9357de68af as denominated

  • v3dv: fix VK_PIPELINE_ROBUSTNESS_{BUFFER,IMAGE}_BEHAVIOR_DEVICE_DEFAULT_EXT copy/paste typo

  • v3dv: fix copy/pasted type of `sample`

  • v3dv: fix shader stage name in error message

  • v3d/qpu: fix type of function argument

  • .pick_status.json: Update to 939845e424da7b2433902a9954cb6b615cd06d41

  • .pick_status.json: Mark 5d19a0a19ba2ffe50a7d05fcf0ff17e0078c2d8e as denominated

  • .pick_status.json: Mark 9865e5dff49395543da4331a943ba5a03ce6a413 as denominated

  • .pick_status.json: Update to 3a949de28c42d8714320e56bd99168148503da7d

  • .pick_status.json: Mark f97e38f9cdabf6e1cfab11ebaabf193492bd403c as denominated

  • .pick_status.json: Mark fa6562b239f00f9f72c988459e252bdee072fd73 as denominated

  • .pick_status.json: Update to 318b83899056d02ee83817cf16201a128fb81443

  • .pick_status.json: Mark 83716b08cf103622829d84e673584781ee8850f6 as denominated

  • .pick_status.json: Mark 8b93fa149bfbe1ca1a59d89ec53d576aa4ede2ec as denominated

Faith Ekstrand (4):

  • nir: Fix nir_op_mov handling in nir_collect_src_uniforms

  • nir: Handle nir_op_mov properly in opt_shrink_vectors

  • nir: Don’t handle nir_op_mov in get_undef_mask in opt_undef

  • nir: Fix metadata in nir_lower_is_helper_invocation

Friedrich Vock (2):

  • nir/load_store_vectorize: Handle intrinsics with constant base

  • radv/rt: Pre-initialize instance address

Georg Lehmann (1):

  • aco: fix u2f16 with 32bit input

Helen Koike (1):

  • ci/android: remove strace output from cuttlefish-runner.sh

Jordan Justen (2):

  • intel/dev: Use RPL-U name on RPL-U devices

  • intel/dev: Add more RPL PCI IDs

Karol Herbst (3):

  • rusticl/kernel: optimize nir between lowering io and explicit types

  • nv50: limit max code uploads to 0x8000

  • rusticl/memory: do not verify pitch for IMAGE1D_BUFFER

Kenneth Graunke (1):

  • iris: Check prog[] instead of uncompiled[] for BLORP state skipping

Leo Liu (1):

  • radeonsi/vcn: fix the incorrect dt_size

Lionel Landwerlin (5):

  • intel/compiler: disable per-sample interpolation modes with non-per-sample dispatch

  • intel/compiler: fix dynamic alpha-to-coverage handling

  • intel/fs: implement dynamic interpolation mode for dynamic persample shaders

  • anv: add missing ISL storage usage

  • intel/nir: rerun lower_tex if it lowers something

Marek Vasut (1):

  • etnaviv: Fully replicate back stencil config

Matt Turner (1):

  • Revert “intel/fs: only avoid SIMD32 if strictly inferior in throughput”

Mike Blumenkrantz (3):

  • vk/graphics: fix CWE handling with DS3

  • zink: wait on async fence during ctx program removal

  • zink: don’t start multiple cache jobs for the same program

Paul Gofman (1):

  • driconf: add a workaround for Rainbow Six Extraction

Rhys Perry (5):

  • radv: disable 64-bit color attachments

  • radv: fix 128bpp comp-to-single clears

  • aco: fix p_bpermute_gfx6 with input at non-zero byte

  • aco/spill: skip p_branch in process_block

  • aco/spill: add all live-in to merge block spill candidates

Rohan Garg (3):

  • crocus: fix GFX_VERx10 macro

  • blorp: drop undefined macro

  • iris: migrate preemption streamwout wa to WA infra

Samuel Pitoiset (2):

  • Revert “radv/amdgpu: workaround a kernel bug when replacing sparse mappings”

  • Revert “radv/amdgpu: skip adding per VM BOs for sparse during CS BO list build”

Sviatoslav Peleshko (2):

  • dri: Use RGB internal formats for RGBX formats

  • intel/isl: Don’t over-allocate CLEAR_COLOR size to use whole cache line

Tapani Pälli (1):

  • mesa: fix some TexParameter and SamplerParameter cases

Tatsuyuki Ishi (1):

  • radv/amdgpu: Do not pass in a BO handle when clearing PRT VA region.

Timothy Arceri (1):

  • util: add radeonsi workaround for Nowhere Patrol

Timur Kristóf (5):

  • ac/nir: Add done arg to ac_nir_export_position.

  • ac/nir: Slightly refactor how pos0 exports are added when missing.

  • ac/nir/ngg: Wait for attribute stores before VS/TES/GS pos0 export.

  • ac/nir/ngg: Refactor mesh shader primitive export.

  • ac/nir/ngg: Wait for attribute ring stores in mesh shaders.

antonino (3):

  • vulkan/wsi: add `vk_wsi_force_swapchain_to_current_extent` driconf

  • drirc: enable `vk_wsi_force_swapchain_to_current_extent` for “The Talos Principle”

  • drirc: enable `vk_wsi_force_swapchain_to_current_extent` for “Serious Sam Fusion”