Mesa 26.0.2 Release Notes / 2026-03-12

Mesa 26.0.2 is a bug fix release which fixes bugs found since the 26.0.1 release.

Mesa 26.0.2 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 26.0.2 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.

SHA checksums

SHA256: 973f535221be211c6363842b4cce9ef8e9b3e1d5ea86c5450ca86060163c7346  mesa-26.0.2.tar.xz
SHA512: 0a7b9fc9b09e40345cc22d246dc1656900d74754c093882f6a39623af17fddc5f4a0c7938207c784ccf7306c5ed497be6a02c36f95c6548e01a2faa085e04c35  mesa-26.0.2.tar.xz

New features

  • None

Bug fixes

  • 26.0.1 fails to build: `create_context.c: error: ‘struct glx_screen’ has no member named ‘frontend_screen’`

  • A770: Counter-Strike 2 visual glitches (regression)

  • Bisected regression: Assertion texObj->pt == view->texture failed.

  • Kodi regression with panthor >= 1.7 after updating to Linux 7.0-rc1

  • MDK2 HD (opengl) has most elements rendered as black

  • Mesa 25.3 amdgpu memory issue

  • OpenGL 4.1 VRAM Memory Leak with setting uniform variables

  • Panfrost Bifrost compiler assertion failure: wrong vectorization in bi_alu_src_index (Mesa 26.0.0)

  • RADV: RDNA4 visual corruption in DX11 (DXVK) – Mafia III character model glitches, AMDVLK renders correctly (9070XT)

  • [radeonsi] Regression: GL_FEEDBACK returns 0.0 for X-coordinates (Legacy GL)

  • glsl: spec@glsl-es-1.00@linker@glsl-mismatched-uniform-precision-unused broken

  • ir3: ir3_get_predicate() vs &ctx->build

  • r300 , regression , bisected : Glitches with Sauerbraten

  • r300: HiZ related dEQP failures

Changes

Anna Maniscalco (1):

  • zink: don’t care about generated gs output primitive

Benjamin Cheng (1):

  • radeonsi/vcn: Use full pitch for pre-encode input

Boris Brezillon (1):

  • pan/kmod: Allow mmap() on foreign buffers

Caio Oliveira (4):

  • spirv: Refactor ALU opcode translation to take bit sizes

  • spirv: Pull constant source fixup to the existing loop

  • spirv: Fix spec constant to handle Select for non-native floats

  • nir: Fix constant folding for iadd_sat

Christoph Pillmayer (2):

  • pan/bi: Fix coupling spill placement

  • pan/bi: Move FAUs to memory for memory phis

Connor Abbott (4):

  • tu: Use HW offset 0 in 3d loads/clears with FDM

  • ir3: Fix constlen trimming when more than one stage is trimmed

  • tu: Set polygon mode when blitting

  • tu: Fix setting will_be_resolved with MSRTSS

Danylo Piliaiev (2):

  • tu: Store gmem attachments after custom resolve in dyn RP

  • tu: Don’t read .patch_input_gmem of unused attachment

David Rosca (1):

  • vl: Also disable MPEG2 Main profile when mpeg12 decode is disabled

Eric Engestrom (3):

  • docs: add sha sum for 26.0.1

  • fixup! docs: add release notes for 26.0.1

  • .pick_status.json: Update to 73dba1e15173ff6109925de9615f9d9f5cccc2be

Eric R. Smith (1):

  • pco: fix a typo in the check for optimization looping

Erik Faye-Lund (1):

  • gallium/dri: set LIBVA_DRIVERS_PATH in devenv

Faith Ekstrand (3):

  • etnaviv: Call lower_bool_to_int32 not to_bitsize

  • nir/lower_bool_to_bitsize: Make all bN_csel sources match

  • pan/bi: Be more careful about bit sizes in b2f lowering

Georg Lehmann (3):

  • ci: disable debian-ppc64el and debian-s390x

  • aco/insert_fp_mode: don’t skip setting round for fract

  • nir/opt_algebraic: fix frsq clamp pattern

Ian Romanick (5):

  • brw: Don’t mark_invalid in update_for_reads for non-VGRF destination

  • brw: Use brw_reg_is_arf in update_for_reads

  • brw: Also check for ADDRESS file in update_for_reads

  • brw/algebraic: Don’t optimize SEL.L.SAT or SEL.G.SAT

  • elk/algebraic: Don’t optimize SEL.L.SAT or SEL.G.SAT

Icenowy Zheng (1):

  • pvr: only specially handle gfx subcmd for BeginQuery

Iván Briano (1):

  • anv: don’t try to fast clear D/S with multiview

Jesse Natalie (1):

  • d3d12: Fix importing external resources

Job Noorman (2):

  • ir3: update context builder after ir3_get_predicate

  • ir3: don’t predicate vote_all/vote_any

Jose Maria Casanova Crespo (3):

  • v3d: flush write jobs before BO replacement in DISCARD_WHOLE path

  • vc4: flush write jobs before BO replacement in DISCARD_WHOLE path

  • v3d: reject fast TLB blit when RT formats don’t match

Karol Herbst (2):

  • nir: fix nir_alu_type_range_contains_type_range for fp16 to int

  • nir: fix nir_round_int_to_float for fp16

Lionel Landwerlin (2):

  • anv: add missing handling for attachment locations in secondaries

  • anv: dirty all push constant stages in simple shader

Lucas Fryzek (5):

  • drisw: Properly mark shmid as -1 when alloc fails

  • x11: Add helper util to check for xshm support

  • egl/dri: Check that xshm can be attached

  • glx: Check that xshm can be attached

  • vulkan/wsi: Check that xshm can be attached

Luigi Santivetti (1):

  • zink: fix format conversion logic for the alpha emulation case

Marek Olšák (1):

  • ac: set the correct number of Z planes for ALLOW_EXPCLEAR

Mary Guillemard (1):

  • vulkan: Do not override the shader_flags in case of no task shader

Mel Henning (1):

  • driconf: force_vk_vendor on No Man’s Sky + NVK

Mike Blumenkrantz (4):

  • zink: add TRANSFER_WRITE -> HOST_READ sync to end of batch

  • st/bitmap: only release YUV samplerviews

  • radv: fix multiview fast clears

  • egl/device: fix the fix for explicit sw rejection in non-sw EGL_PLATFORM=device

Patrick Lerda (1):

  • r600: fix cs atomic operations when the shader is called multiple times

Pavel Ondračka (3):

  • r300: copy target when merging alpha output instruction

  • r300: disable HiZ for PIPE_FUNC_ALWAYS

  • r300: disable clip-discard watermark for triangles

Pierre-Eric Pelloux-Prayer (2):

  • frontends/va: fix undefined ref error

  • mesa: don’t wraparound st_context::work_counter

Rhys Perry (2):

  • aco: perform dce for blocks skipped for process_block()

  • nir/range_analysis: set deleted key

Sagar Ghuge (1):

  • anv: Fix Wa_14021821874, Wa_14018813551, Wa_14026600921

Samuel Pitoiset (4):

  • radv: fix copying images with different swizzle modes on SDMA7

  • radv: fix a GPU hang with PS epilogs and secondary command buffers

  • radv: fix local invocation index for mesh/task and quad derivatives on GFX12

  • radv: fix missing L2 cache invalidation with streamout on GFX12

Tapani Pälli (2):

  • intel/dev: update mesa_defs.json from workaround database

  • anv: add handling for Wa_14026600921

Timothy Arceri (5):

  • glsl: relax precision matching on unused uniforms ES

  • glsl: add workaround for MDK2 HD

  • mesa/st: use same path for setting state ref locations

  • st/glsl_to_nir: update state var locations earlier

  • glx: guard glx_screen frontend_screen member

Yiwei Zhang (2):

  • pan: fix to not clear out of bitset range

  • lvp: avoid advertising dmabuf support for kms_swrast