Mesa 25.2.5 Release Notes / 2025-10-15

Mesa 25.2.5 is a bug fix release which fixes bugs found since the 25.2.4 release.

Mesa 25.2.5 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 25.2.5 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: bb6243e7a6f525febfa1e6ab50827ca4d4bfdad73812377b0ca9b6c50998b03e  mesa-25.2.5.tar.xz
SHA512: 29e61b5ecb467a706e3279c0e79ddd8d55109c08f7856d35c4042f518a70622fb19cdd208a82317654e0396835cb3117b756a96d9a0693bfa33730a50bbbd1d0  mesa-25.2.5.tar.xz

New features

  • None

Bug fixes

  • anv: Age of Wonders 4 corruption on a Arc b580

  • nvk commets cause problems with kepler

  • Confidential issue #14013

  • radv: meta pipeline cache appears to be broken

  • A bunch of CTS tests are failing on Gfx12.0 trying to use the blitter with TILE_X

  • [ANV] Bunch of tests in dEQP-VK.pipeline.*.render_to_image.*3d.*2d_compatible failing on gen9/11

  • bisected build failure in clc_helpers.ccp with llvm 22

Changes

Ashley Smith (2):

  • panfrost,mesa: Fix versions for EXT_shader_realtime_clock

  • panfrost,mesa: Fix versions for EXT_shader_clock

Benjamin Cheng (2):

  • vulkan/query_pool: Store video encode feedback

  • radv: Output requested encode query results only

Calder Young (1):

  • isl: Fix noncoherent framebuffer fetch when base_level != 0

Daniel Schürmann (1):

  • nir/divergence: don’t assume that load_sample_positions_amd is always uniform

Danylo Piliaiev (1):

  • tu: Prevent GPU hang with occlusion query + certain depth state

David Rosca (2):

  • radeonsi/vpe: Fix transfer function mapping to vpelib

  • frontends/va: Fix parsing VP9 frame header

Dmitry Baryshkov (1):

  • glx: provide glx.pc

Eric Engestrom (4):

  • docs: add sha sum for 25.2.4

  • .pick_status.json: Update to 282e8285f1b0e2808dc2416f7274a85e39828cdf

  • ci-tron: bump job template commit to get cached job templates

  • .pick_status.json: Mark 6f241b7f6afefe38d4d7610cf0237c91ca3806b0 as denominated

Erik Faye-Lund (1):

  • aux/pp: release correct sampler-views

Faith Ekstrand (1):

  • nil: Add a GOB_TYPE_MODIFIER_INFOS table

Georg Lehmann (2):

  • aco/optimizer: fix incorrect operand order assumption for neg(mul) opt

  • aco: fix global_atomic_swap offset overflow check

Hans-Kristian Arntzen (1):

  • radv: Actually fail custom border color sampler creation.

Iago Toral Quiroga (1):

  • panfrost: fix swapped stats for varing and position shaders

José Roberto de Souza (3):

  • anv/allocator: Change some parameters and variables from 32bit to 64bits

  • anv/allocator: Don’t call anv_block_pool_map() with an offset that includes start_offset

  • anv/allocator: Subtract start_offset in chunk_offset

Karol Herbst (2):

  • libagx: fix heap argument type in libagx_draw_robust_index

  • clc: Fix createDiagnostics for LLVM-22

Lionel Landwerlin (3):

  • vulkan/render_pass: fixup renderpasses barriers for 2D views of 3D images

  • anv: fix companion usage for emulated image

  • anv: fix query copy with shaders

Lorenzo Rossi (1):

  • nvk: Fix QMD buffer length on upload

Lucas Stach (1):

  • etnaviv: fix YUV tiler blits

Mauro Rossi (1):

  • amd: require LLVM when amd-use-llvm is enabled

Mike Blumenkrantz (7):

  • zink: stop using vk lazy allocations / transient attachments

  • zink: strip dmabuf bind flags when creating transient image

  • zink: always add mutable to transient surface creation when needed

  • zink: only add mutable bind for transient surfaces when necessary

  • zink: disable msrtss handling when blitting

  • glsl: fix gl_ViewID_OVR type to uint

  • mesa: copy NumSamples in reuse_framebuffer_texture_attachment

Natalie Vock (2):

  • vulkan/bvh: Mark instances with NAN AABBs as inactive

  • radv/bvh: Encode empty AS bounds as NaN

Olivia Lee (1):

  • vtn_bindgen2: use anonymous namespace to avoid name collisions

Pierre-Eric Pelloux-Prayer (1):

  • tc: prevent flush of incomplete batches

Rob Clark (1):

  • freedreno: Disable explicit sync heuristic for Xwayland

Sagar Ghuge (1):

  • Revert “intel: Always set Cube Face Enables for all surfaces.”

Samuel Pitoiset (5):

  • radv: reserve more CS space when executing DGC calls

  • radv: fix expected disk cache size for meta shaders

  • vulkan/runtime: fix memleak when creating ETC pipelines

  • radv/rt: fix memory leak in lower_rt_instructions_monolithic()

  • radv: fix shaders memleak when importing pipeline binaries with GPL

Simon Perretta (1):

  • pvr: amend num temps calculation when wg_size is not provided

Tapani Pälli (3):

  • intel/blorp: add restriction for gfx12

  • iris: add a check if blorp can support blitter copy

  • anv: add cs stall for any pipe control on compute

TellowKrinkle (1):

  • hk: Enable caching on memory marked with HOST_CACHED_BIT

Yiwei Zhang (2):

  • panvk: fix to clear FPK with incompatible blend modes

  • panvk: allow panvk_pool_alloc_mem to use full slab_size