Mesa 26.0.4 Release Notes / 2026-04-01
Mesa 26.0.4 is a bug fix release which fixes bugs found since the 26.0.3 release.
Mesa 26.0.4 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 26.0.4 implements the Vulkan 1.4 API, but the version reported by
the apiVersion property of the VkPhysicalDeviceProperties struct
depends on the particular driver being used.
Bug fixes
- Accumulation of black squares with OpenGL applications at high resolutions (hiz-related)
- RADV: Invalid hitAttributeEXT value when using function-call RT pipelines
- Segmentation fault in gm200_validate_sample_locations with Firefox on GTX 1070 Ti (nouveau)
- Vulkan CTS regression bisected to 5d2c17a5fdce ("vtn: skip make-available/visible for shared")
- [anv] Intel ARC B390 | Horizon Forbidden West | DX12 | Flashing effects
- [radeonsi] Missing ground texture in Lethis Path of the Progress
- amdgpu reset/crash when simulating stereo camera
- building mesa_clc on ubutu-26.04 with gcc-16 fails link
- util: Build regression with MSYS2 MinGW-W64 x64 clang 21 on 26.0.0-rc3
- wsi: `assert(chain->dxgi);` may failed under venus for win32
Changes
Adam Simpkins (1):
- iris: fix a crash in disable_rb_aux_buffer
Alyssa Milburn (1):
- nv50,nvc0: Avoid uninitialized cbuf reads in blits
Alyssa Rosenzweig (1):
- nir: add nir_get_io_data_src
Dave Airlie (1):
- st/mesh: handle mesh shader point size
David Rosca (2):
- frontends/va: Fix leaking H264/5 PPS/SPS objects when decoder wasn't created
- frontends/va: Fix leaks when create_video_codec fails
Eric Engestrom (9):
- docs: add sha sum for 26.0.3
- .pick_status.json: Update to 48c086cb4203d1a8e7458e0d0a85cfffc5b4bfe5
- .pick_status.json: Mark 26b19e355fefcd6a8325924e6a391dd67a675c34 as denominated
- .pick_status.json: Mark 32a818d11d3d60ebbc23a62127e988d17e742b79 as denominated
- .pick_status.json: Mark d38916d673e6d2359e96fed45ebd83ca026dfcb5 as denominated
- .pick_status.json: Mark 384d12816459996fbac5722e9fdb29527662cafb as denominated
- ci: changing .gitlab-ci.yml itself also means the container jobs must exist
- .pick_status.json: Mark 538c3ee6c7a419d5c55bef2294ca10166f8d9af4 as denominated
- [26.0 only] venus/ci: mark a test as fixed
Eric Guo (1):
- panfrost: Fix NULL pointer dereference in panfrost_emit_images
Eric R. Smith (2):
- panfrost: fix texel buffer calculations
- panfrost: fix typos in architecture detection
Erik Faye-Lund (5):
- pan/genxml: remove non-existent YUV Enable for AFRC
- pan/lib: do not try to use stencil-aspect of color attachment
- pan/lib: set srgb-flag for afrc render-targets
- pan/lib: divide extent by tile-extend, not itself
- panvk: remove unused flag
Faith Ekstrand (4):
- nak: Report progress from nak_nir_rematerialize_load_const()
- nir: Consider if uses in nir_def_all_uses_*
- pan/bi: v2x16 conversions don't replicate
- pan/buffer: Add the offset to the size for buffer textures
Georg Lehmann (2):
- gallivm: don't optimize fadd(a, 0.0) with signed zero preserve
- nir/lower_non_uniform_access: fix fusing loops for same index but different array variable
Hyunjun Ko (1):
- anv: Add dummy workload for AV1 decode on affected platforms (Wa_1508208842)
Ian Romanick (2):
- brw/algebraic: Allow mixed types in saturate constant folding
- brw: Handle scalars and swizzles correctly in is_const_zero
Icenowy Zheng (8):
- vulkan/wsi/headless: properly use CPU images for CPU devices
- pco: fix encoding of fred's s0abs bit
- pvr: Align width for PBE write when creating linear image
- pvr: fix "obb" typo in oob_buffer_size when building vertex pds data
- pvr: save vertex attribute size for DMA checking
- pvr: move PVR_BUFFER_MEMORY_PADDING_SIZE definition to pvr_buffer.h
- pvr: consider the size of DMA request when setting msize of DDMADT
- pvr: fix dirty tracking for stencil ops
Iván Briano (2):
- anv: fix anv_is_dual_src_blend_equation
- brw: do not omit RT writes if dual_src_blend is on
Job Noorman (1):
- ir3/legalize: don't drop sync flags on removed predt/predf
Jose Maria Casanova Crespo (1):
- broadcom/common: fix V3D 7.1 TFU ICFG IFORMAT values
Juan A. Suarez Romero (1):
- vc4: fix unwanted buffer release on uploader
Lionel Landwerlin (3):
- anv: add an analysis pass to detect compute shaders clearing data
- anv: add drirc option to workaround missing application barriers on typed/untyped data
- brw: fence SLM writes between workgroups
Liviu Prodea (2):
- clc: Fix static link with clang>=22
- util: Fix use of undeclared identifier 'NULL' in src/util/os_misc.h when compiling with clang
Luigi Santivetti (2):
- pvr: expose partial usc mrt init routine
- pvr: keep compiler resources in sync with attachments
Marek Olšák (3):
- radeonsi: recompute IO bases after optimizations
- radeonsi: fix blits via util_blitter_draw_rectangle
- radeonsi: disable streamout queries for u_blitter
Mario Kleiner (1):
- dri: Fix "cosmetic" undefined behaviour warning for RGB[A]16_UNORM formats.
Mary Guillemard (5):
- nvk: Move viewport and scissor emit to their own function
- nvk: Broacast viewport0 and scissor0 in case of FSR on Turing
- nir/dead_cf: Add missing load_ssbo_ir3 handling
- nir/dead_cf: Add missing load_global_bounded handling
- nak: Do not allow load_helper_invocation reordering
Mike Blumenkrantz (3):
- ntv: always emit const coord components for fbfetch loads
- mesa/renderbuffer: always add PIPE_BIND_SAMPLER_VIEW to rendering textures
- llvmpipe: fix color fbfetch
Natalie Vock (1):
- vulkan: Bump MAX_ENCODE_PASSES
Nick Hamilton (1):
- pvr: Fix for multiple attachments being assigned to the same tile buffer.
Pavel Ondračka (5):
- r300: fix bias presubtract algebraic transformation
- r300: don't apply odd macroblock rounding to 3D textures
- r300: disable zmask clears for large surfaces
- r300: add shared HyperZ pipe-count helper
- r300: split large HiZ clears into multiple packets
Pierre-Eric Pelloux-Prayer (3):
- radeonsi: move spi_shader_*_format to si_shader_variant_info
- radeonsi: account for outputs_written when updating spi_shader_col_format
- gallium/u_blitter: add a new fs_color_clear variant
Radu Costas (1):
- pco: Amend errant nir_move_option
Rhys Perry (3):
- aco/tests: fix assembler tests with LLVM 22
- aco/tests: fix assembler/isel tests with LLVM 23
- radv: fix memory leak in radv_rt_nir_to_asm
Robert Mader (1):
- llvmpipe: Stop aligning height to raster block size for unbacked handles
Ryan Zhang (1):
- panvk: trivial fix to remove repeated assignment
Samuel Pitoiset (2):
- radv/amdgpu: free the VA range in case the BO allocation failed
- radv: emit BOP events after every draw to workaround a VRS bug on GFX12
Simon Perretta (1):
- pco: use vm/icm for tile buffer store coverage mask
Timothy Arceri (2):
- mesa: add force_explicit_uniform_loc_zero workaround
- util/driconf: add workarounds for Lethis - Path Of Progress
Valentine Burley (7):
- tu/drm/virtio: Add missing lock to virtio_bo_init_dmabuf
- tu/drm/virtio: Move set_iova into success path of virtio_bo_init_dmabuf
- tu/drm/virtio: Avoid freeing zombified tu_sparse_vma
- tu/drm/virtio: Do not free iova from heap for lazy BOs
- tu/drm/virtio: Fix GEM handle leak in tu_bo_init error path
- tu/drm/virtio: Fix GEM handle leak on failed dmabuf res_id lookup
- ci: Drop duplicate Intel shader-db run
Yiwei Zhang (3):
- venus: fix to relax the KHR_external_memory_fd requirement
- vulkan/wsi/win32: add wsi_win32_find_idle_image helper
- vulkan/wsi/win32: respect acquire timeout for sw wsi
emre (1):
- nvk: fix barrier cache invalidation
juntak0916 (1):
- nvk: fix BindImageMemory2 per-bind status result
kingstom.chen (1):
- radv/rt: only run move_rt_instructions() for CPS shaders
utzcoz (1):
- gfxstream: Fix vkSetDebugUtilsObjectNameEXT crash for unwrapped objects