-
magges released this
2026-08-07 05:07:27 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r227503.ea1f62447af3-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
ea1f62447af3 - Package:
mesa-git_26.3.0-devel+r227503.ea1f62447af3-1_amd64.deb
Mesa upstream changes since
c7919be1fa9d- ea1f62447af ci/debian: Install spirv-llvm-translator instead of building it
- f5b6dfb260b intel/executor: Add 256_hw_regs.lua example
- 99bd1b1d89f intel/executor: Allow hw_regs based on intel_register_blocks_supported()
- f07fbfdbd31 intel/common: Add intel_register_blocks_supported()
- 011cf0d7759 intel/executor: Set state for hw_regs param
- 5eed5423c70 intel/common: Rework intel_register_blocks()
- 89d39f7fe06 intel: Move brw_register_blocks to common intel_register_blocks
- 4dba68207f4 intel/executor: Use hw_regs param to adjust internal/rsvd reg num
- d4c93387cfa intel/executor: Add hw_regs shader parameter
- a6bfa41effb svga: fix vertex buffer stride (#42)
- 6a5c13b3602 kraid: Add opt_promote_consts
- a962b84f2f6 gen_calendar_entries: fix datetime for newer Python
- 396ab8173b4 nir: use fadd_rtne in lower_round_even
- 5cc44a04f65 zink: don't use nir_lower_dround_even
- cdb092de762 r600/sfn: handle nir_op_fadd_rtne
- ff691e20c7b gallivm: handle nir_op_fadd_rtne
- 5b7b6cf9255 nir_to_tgsi: handle nir_op_fadd_rtne
- 97e6d3458ff microsoft/compiler: handle nir_op_fadd_rtne
- bb4143b3988 nak: handle nir_op_fadd_rtne
- 394075ccc2b jay: handle nir_op_fadd_rtne
- aceef72d4a7 brw: handle nir_op_fadd_rtne
- 56a008c697c elk: handle nir_op_fadd_rtne
- fae626a82e4 nir: add fadd_rtne opcode
- c208cb879ea docs: Add calendar entries for 26.3 release candidates.
- 499e898a361 iris: implement pipe_screen::semaphore_create
- afa5f341b9c ci: build drm-shim in the arm test builds
- 5da48049efd nouveau/drm-shim: use a 64 bit constant for the VRAM BAR size
- 80be7e22191 panfrost/drm-shim: go through uintptr_t when casting a query pointer
- 0d54c473ca8 drm-shim: forward the glibc time64 entry points on 32-bit
- 12af35ca13b gallium/va: fix build-order edge for libva win32 headers on Windows
- 768bbc5f20b d3d12: return av1 sequence header for encoder
- 3e25cedac17 tgsi: remove more unused opcodes
- 7e5b708afd4 tgsi: remove LOG/EXP opcodes
- 4c5ca0b7c0f tgsi: remove DST opcode
- abe7765dc01 tgsi: remove LIT opcode
- 6c225b48291 anv/brw: fixup per shader compile options
- dd44b7617a4 vulkan/runtime: discard disk cache items when deserialization failed
- 4139bf24d71 vulkan/runtime: remove assert on shader deserialization
- 04b2ce01545 nir: reuse copy propagation dynarray storage
- e654fd5ba13 nir: reuse copy propagation hash table storage
- 0fb27ae937f util/hash_table: add clone_into helper
- 0c006c56248 radv: remove deprecated RADV_DEBUG=nocompute
- 52b18c2d507 radv: remove deprecated RADV_PERFTEST options
- 524a1411b67 rusticl: add basic support for cl_ext_float_atomics
- 43e7c2cdeff clc: add flags for fp32 atomics
- 410a764236d util/dynarray: outline the growth slow path
- 8ca152c8cdb panvk/ci: Add a Kraid job on G925
- 69683217138 panvk/ci: Reduce panfrost-g925-vk:arm64 parallelism
- 4b3a40c4c2a nir/range_analysis: skip caching unsupported UUB ALU ops
- 4317b2391da amd: add gfx1171
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-08-04 05:06:45 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r227281.c7919be1fa9d-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
c7919be1fa9d - Package:
mesa-git_26.3.0-devel+r227281.c7919be1fa9d-1_amd64.deb
Mesa upstream changes since
0baeca38c72a- c7919be1fa9 venus: remove in-tree venus-protocol
- cd4581e393a venus: migrate to build with venus-protocol subproject
- d3a6025659d subprojects: pull in venus-protocol
- 371cad66708 ci/deqp-runner: Bump to 0.23.3 for CL CTS fix
- 6dae3eeeab5 ci: Skip CL_DEVICE_TYPE_CPU tests
- e46a8367d4a ci: Move OpenCL-CTS SPIR to global skips list
- a05eacc9aed anv/rt: Update quads during update BVH pass
- b460cff22a5 svga: Pass buffer handle for fd handles for resource_from_handle()
- 4a66f5a9178 iris: Add function to compute and emit engine async threads limits
- bc2865d8ae0 blorp: Program of async thread limits when VRT is disabled in runtime
- b5d4780f0b8 iris: Program of async thread limits when VRT is disabled in runtime
- 1cbbac0ccbe anv: Program of async thread limits when VRT is disabled in runtime
- b683b5a33e7 anv: fixup the logic dealing with STATE_BYTE_STRIDE
- 0200462978a svga: fix shader leakage for emulated code path
- 65ac5d30143 wgl: Add driconf for XELive to force GDI compat
- 2eee75cfc7a mesa: Fix shared state bookkeeping for dynamic share list changes (wglShareLists)
- a2feaa67a07 intel/gen: decode type of src1 in basic 2 source after setting IMM
- 2de8a9eecb8 radv: drop inlined functions before cleanup passes
- e592c4d2c16 kraid: Report shader statistics
- 35981890340 kraid/v9: Fix up MOV.v2i16
- 242b2ae5891 compiler/rust/bitset: Add a BitSet::len() method
- 324182df510 pan/bi: Expose va_compute_alu_bound()
- 19f087e1e89 kraid/ir: Add is_tls flags to OpLoad and OpStore to track scratch
- b0f956d1053 kraid/model: Add a way to get at the pan_model
- 4e9b3ac69e1 kraid/model: Add an exececution unit query
- d9c56009c1d kraid/isa: Add execution units to InstructionInfo
- c9e4f184e59 pan/bi: Use kraid for OpenCL kernels when PAN_USE_KRAID=cs
- 42787ec9adc pan/bi/nir: FLOG/FEXP.f16 don't exist prior to v12
- bab5285218a pan/bi/nir: Scalarize the rest of nir_op_b(any|all)_*
- 7138a767a04 pan/bi: Promote UBO loads to FAU in NIR for Kraid
- bc001e82256 kraid: Allow MESA_SHADER_KERNEL for barriers
- e4aa1406389 kraid: Add ICMP.v4[su]8
- a0da1e96d85 kraid/v9: Apply i_as_u() to OpIMul::dst_type
- 69aa35ea77d kraid/v9: FADD_IMM doesn't support source modifiers
- 7eddee2c12d kraid/nir: Handle 8 and 16-bit load_push_constant
- 99b67912e60 kraid/legalize: Legalize 64-bit loads of high-word FAUs
- 4ba5f7ca3f4 kraid/legalize: Add a LegalizeFAU::new() helper
- 0815c20329e kraid: Use nir_shader_compiler_options directly
- 8be93459a68 compiler/rust/nir: Implement Send+Sync for nir_shader_compiler_options
- f5547ac5615 nir: Add load_shared_base_ptr to divergence analysis
- 84acd8488ad anv: Return early in anv_reloc_list_append() when relocs are not in use
- 30471fca566 anv: Fix segfault when binding shaders to anv_cmd_buffer before emit any instruction
- 5b08f536c7f anv: Copy anv_reloc_list::uses_relocs when clonning reloc list
- 85da60d843e panfrost: Use pan_get_compiler_flags()
- 3de2aa683a7 panvk: Pass pan_get_compiler_flags() to the disk cache
- cfd8a0d14e1 pan: Add a pan_get_compiler_flags() helper
- ff70d1c1e32 pan: Rework PAN_USE_KRAID
- 4fa60540d36 vtn: don't fail at uniformity decorations on variables and OpBufferPointer
- 311e001b789 nir/tests: add more load/store vectorizer tests
- 728b81fd3ca nir/load_store_vectorize: recreate entry key when adding from predecessor
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-08-01 05:07:02 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r227199.0baeca38c72a-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
0baeca38c72a - Package:
mesa-git_26.3.0-devel+r227199.0baeca38c72a-1_amd64.deb
Mesa upstream changes since
32b22499f488- 0baeca38c72 glx/apple: silence OpenGL deprecation warnings in libglx
- 5513651f32d CODEOWNERS: Update CODEOWNERS to add myself for src/glx/apple
- 3e3631d98d0 brw: Enforce Gfx9 restriction 3-source destination must be GRF
- 0b3ba7edae3 brw: Enforce undocumented Gfx11 and Gfx12 restriction on src mods w/ accumulators
- ed4bb05db77 brw/validate: Gfx11 can't have accumulator src0 in 3-src instructions
- 707cc8f5ab9 svga: fix shared memory index for svga driver
- 6c306fd75a9 d3d12: Disable vao fast path for AMD
- cc82e157111 pan/compiler/stats: Fix ALU not being used in instruction bounds
- 15c7cc550a0 pan/rusticl: Update the last OpenCL conformance version passed value
- af80126eedc pvr: increase maxImageArrayLayers value
- cece4db9407 pvr: increase maxPerStageDescriptorUniformBuffers value
- 7dd7147b9cd pvr: re-enable VK_KHR_shader_expect_assume
- 71366c8cddb pvr: re-enable VK_EXT_texel_buffer_alignment
- 096b8f2c092 pvr: re-enable VK_KHR_format_feature_flags2
- 271ad6c5cdc radeonsi: emit MESHLET registers for mesh shaders in gfx10_emit_shader_ngg
- 08ca8475fae radv: use a compute copy for large GTT/host copies on dGPUs
- b3f90cea886 pan: check full fb coverage on has_partial_tiles
- 46d668f2139 ci/build: Increase timeout for debian-build-x86_64
- edaa36d969a radv: replace PS_STATE_LINE_RAST_MODE with SMOOTH_LINES, make it accurate
- ede920a2681 radv: simplify code determining whether smooth lines may be enabled
- 62d59663e3b radv: don't lower for smooth lines if the raster prim is never lines
- 64df6b11b9a radv: fix incorrect and suboptimal point size output removal
- 0ea2191d70b radv: optimize layer_id with nir_opt_varyings
- b50298b50e0 radv: always remove PSIZ and LAYER as varyings
- b284173af5a nir/lower_sysvals_to_varyings: add lowering to intrinsics instead of variables
- 3336634fff0 nir/remove_outputs: don't remove LAYER as varying unconditionally
- 4c8e720c62a kk: Support VK_EXT_custom_border_color and VK_EXT_border_color_swizzle
- 52c5c8d3d0b lavapipe: Embed bindless sampler views by value
Downloads
- Repo commit:
-
2026-07-31 05:06:15 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r227171.32b22499f488-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
32b22499f488 - Package:
mesa-git_26.3.0-devel+r227171.32b22499f488-1_amd64.deb
Mesa upstream changes since
829e6d16301d- 32b22499f48 glx/apple: implement swap control on top of kCGLCPSwapInterval
- 98c5fe707c2 glx: add __glXForceEnableExtension()
- ea60532e251 llvmpipe: don't create a screen when the process is not allowed to JIT
- 2f891050788 anv/rt: separate copies from bvh update
- a62bd97754e jay: copy resource_intel surface handle value
- 209e68250df rusticl: Stop taking references to temporaries
- 1cc52dd0edd rusticl: Use unsafe(no_mangle)
- 692d57c2cba etnaviv/ci: update expectations after cmdstream reservation fixes
- 6ab174a3bc9 etnaviv: move sampler source update before draw space reservation
- 120476415f4 etnaviv: reserve state emission space early in draw_vbo
- 0833edefbea etnaviv: move index buffer handling in draw_vbo after derived state handling
- fb4cad4c05e etnaviv: clean up index buffer handling code a bit
- 7296f9af84c brw: Remove obsolete interface to get number of used registers
- a888a12174b anv: Use new interface to get number of register blocks
- 92055cbfd4e brw: Use new interface to get number of register blocks
- b2ddf668bac blorp: Use new interface to get number of register blocks
- 213c7f116bc iris: Use new interface to get number of register blocks
- e5bee43e29a brw: New interface to get number of grf register blocks
- fa0183466d7 radv/nir/opt_tid_function: try to use amd permute intrinsics
- 2b85a2daa33 radv/nir/opt_tid_function: add option to create standard quad ops
- d0e757a011f aco: implement permute_x16 and dpp8_swizzle
- 7301879254f nir: add amd dpp8 and permlanex16 intrinsics
- d06485a1465 nir: mark some AMD specific shuffles as subgroup ops
- dbbfbcc8d6b nv30: add CI results for NV34
- 7e2bdd8e407 etnaviv: Set up TFB buffer address and size at bind time
- 5a102f4ecee etnaviv: Coalesce TFB buffer state emission
- 7b8ff5692ba etnaviv: Drop stale stream output relocs in etna_update_hwxfb(..)
- 6bdcd282afc virtio/magma-gpu: continue with the renaming
- 84c7d18a4d0 zink/ci: Move glx@glx-make-current to flakes
- cb3bdc1bfd8 zink: Use ShaderLayer capability for gl_Layer when available
- 201cbc48803 tu: Merge disable_write_for_rp from secondary to primary
- 318240f418c kk: Fix up M5 robustness2 behavior when LOD is out of bounds
- ae84646e30c tu: Fix tu_event not being reset on creation
- 97d586cc664 zink/ci: switch zink/anv jobs to surfaceless+i915
- 803731f7ed4 freedreno: Add chip support for a722
- 4b37c2f5cb5 v3dv: advertise VK_EXT_device_memory_report
- e0ad113ed02 v3dv: emit device memory reports in v3dv_bo_alloc/free
- 292bc2c8d86 v3dv: store obj_type and obj_handle in v3dv_bo
- 276e2dcf3c5 v3dv: track self imported BOs
- 6552160da51 v3dv: introduce a unique per-device bo identifier
- a49a56ac125 v3dv: store shared data owner type and handle
- e79dfca9f9f glthread: fix check for unroll draws using user VBOs when the ctx supports GLES
- fc8dc63cc6d pan/nir: Fix header static inline function
- 395de865be7 poly: Fix comment after moving passthrough_gs
- d4d8b05477a hk: xfb: Avoid assertions in nir_slot_num_components
- 6b9c7505fb2 mailmap: update my email
- d796224b559 CODEOWNERS: update panfrost maintainers
- 41737d1e287 nvk/ci: Update expectations
- fb434489516 nvk: Optimize varyings
- 751cd2e11f0 nak: Move shader I/O lowering out of nak_postprocess_nir
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-28 05:06:48 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r226883.829e6d16301d-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
829e6d16301d - Package:
mesa-git_26.3.0-devel+r226883.829e6d16301d-1_amd64.deb
Mesa upstream changes since
827bf4790160- 829e6d16301 nir/opt_vectorize: Remove the phis that were combined
- 4f4b73dfe58 nir/opt_dead_write_vars: remove some shared writes after barriers
- 60654ceb913 vtn/opencl: fix libclc needing fp16 lowering to fp32
- c5f10a8501c anv: Enable the temporary RE engine workaround for several titles
- 1f9ac2ca8e7 anv: Add drirc option to workaround broken RE engine shader
- d6bd961b8dc brw: Add workaround pass for broken RE engine shader
- 43cd2c27da1 anv: Add drirc to emulate Xe2+ active thread barriers on Gfx125
- 46e5f2695f5 brw: Add pass to emulate Xe2+ active thread barriers on Gfx125
- 4034d758812 nir: Make it easier to define private driver values in SLM
- 39c018b9af8 nir/builder: Add nir_halt_if helper
- 367226bfd0e nir/builder: add deref/var atomic helpers
- c276444b719 asahi: ci: Add failing b10g11r11 and e5b9g9r9 copy tests
- aa4a378f301 panfrost: ci: Drop no longer failing VK.wsi.xcb.present_timing test
- c523afe2b0b asahi: ci: Drop no longer failing VK.wsi.xcb.present_timing test
- 28f0804ea03 brw: Match fill/spill optimization scratch accesses by logical offset
- 0989349030b virtio: magma-gpu-rs: accept a null device in virtgpu_kumquat_finish
- ad0cdded0e4 radv: fix RADV_PERFTEST=nircache enablement
- 3cd07aacb9e brw: add missing INTEL_FS_CONFIG_PER_PRIMITIVE_REMAPPING handling
- d5df9b5f385 jay: pep8
- 90893959c32 jay: rework last-use fields
- 7f83394368b jay/schedule: ignore liveness post-RA
- 3f18df3915a jay: Implement FB write payloads with multiple file swaps
- 32782ea3dde jay: Avoid propagating some immediates into uniform flags
- a5be5bede3b jay: Run nir_opt_dead_cf after lowering SIMD width and subgroup sizes
- 9fa2ceafbb6 jay: Delete load_subgroup_size translation
- 772de70801e jay: Return multiple shader stats
- 9bac8177ef4 jay: Compile multiple SIMD variants
- 57e1cd55ae6 jay: Rewrite SIMD selection
- 2e2118ccc7f brw: Refactor INTEL_SIMD_DEBUG helpers that take a dynamic SIMD index
- a4f4c22669f jay: Stop using brw_nir_apply_key
- 6d24a2c18f0 jay: Split jay_process_nir into 3 pieces
- 42158c6a893 jay: Call brw_nir_apply_key later
- 69ce95a0b03 brw: Lower subgroup masks in brw_preprocess_nir
- 9f5a38e8e30 jay: align demote handling with IGC
- d00974fdccd jay: allow sr0.2 in the IR
- aa057f07cab jay/performance: fix accumulator latency reporting
- 6adf20088c6 jay/schedule: try fewer schedules
- 273d55838b6 jay: handle VS urb input handle
- b1fcf08c69b jay: deal with unreachable loop back edges
- 78678041308 jay: fix null RT setting
- 65cf5f713ce jay: lower load_max_polygon_intel trivially
- 6641c551773 jay: implement robust_ubo_ranges
- 7942c27445a jay: remove bogus restriction
- b7382b3eee6 jay: lower nir_intrinsic_load_sample_pos_from_id
- fb7231fac2a jay: implement nir_intrinsic_load_sample_positions_intel
- dacc6b87d0e jay: enable nir_opt_uniform_atomics
- daa7d8d0890 jay: implement nir_intrinsic_last_invocation
- 05a3592b03e jay: lower scan/reduce in NIR
- a097b310673 jay: implement intrinsics for scan/reduce
- 88f54655ca2 jay/to_binary: simplify region handling expression
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-25 05:05:57 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r226774.827bf4790160-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
827bf4790160 - Package:
mesa-git_26.3.0-devel+r226774.827bf4790160-1_amd64.deb
Mesa upstream changes since
d49a00bdf15f- 827bf479016 brw: MAC has the same commutivity rules as MUL
- c4206d2b720 brw/xe2: Use accumulator for SIMD16 USUB_SAT / ISUB_SAT lowering
- 603a410af35 kk: Remove redundant render pass descriptor load op change
- 0dbf521d381 panvk: enable Vulkan 1.1 and subgroups on Bifrost
- 8b03ae787f4 pan/bi: set branch reconvergence at loop exit points
- f11071ce865 docs: advertise VK_KHR_multiview support for Bifrost
- 1b0f0e0a83c rusticl/memory: return 0 for CL_IMAGE_SLICE_PITCH also for 2d images
- aefa0ce06e0 freedreno/a6xx: Add SUBPASS_FENCE
- 33643dab065 freedreno: Add SUBPASS_FENCE, SLICE_SUBPASS_FENCE
- 2b6602eb4b2 freedreno: Split gen7 and gen8 event table
- 1c4b24a0e08 tu+freedreno: Add asserts for unsupported events
- b0960060487 freedreno/a6xx: Use a2d for z24s8 blits
- 5f59db1e03d freedreno/a6xx: Handle blit mask
- 46986ea2f72 freedreno/a6xx: Rework emit_blit_setup() params
- 1f95996ac08 freedreno/a6xx: Fix a2d_blt_cntl setup
- e5fbf9097b7 freedreno/a6xx: Fix ms resolve blits
- 9a61ecbb2a3 freedreno: Fix separate-stencil tracking
- d0a72a261e1 intel/gen: print sr0 subregisters
- 566b3f752ba intel/anv: fix potential overflow before widen
- 6ccebfb0444 radv: reduce PS_STATE_PS_ITER_MASK to 8 bits
- 30bff6e7b7a radv: get FRONT_FACE from a user SGPR if it's dynamic but constant
- edeb6dd258b radv: fix determining the static raster prim for FS barycentrics and front_face
- 9a77ee629d7 radv: fix determining the dynamic raster prim for FS barycentrics
- 6b534309ead radv: fix determining the raster prim for line mode
- c61fdeb64c4 radv: fix determining the raster prim for guardband
- 254943add03 nv30: reorder binding steps to match draw module intended behavior
- 07b4da6e186 nv30: give more control to draw module for drawing vertex shaders
- b84ccacd138 panfrost: restore max retry count for bifrost LCRA spills
- fd811cfd9b7 util: Remove string shader statistic support
- b277a4925a1 nir: Account for cmat memory accesses in copy_prop_vars
- a1d7f1770dc intel/compiler: Fix typo in comment (dst -> src)
- 1c3d4617f6f intel/compiler: Delete useless statement
- 37c0e7d8a59 egl: rewrite dispatchstubs generator to be more efficient
- 9f955af118d turnip/kgsl: close the dma-buf fd of our own allocations
- 5f378efa318 intel/compiler: Avoid rechecking first character of string_view
- 46927ba03a2 nir/divergence: consider nir_var_mem_constant uniform
- 2c3c8b7fc26 nir: fix divergence for deref_cast
- 95cf4cca42b st: invalidate shader constants on clip plane state changes
- 0bc2efab05c egl: avoid calling loader_get_driver_for_fd with fd = -1
- 282d791a69b kk: Determine supported samplers by Metal limit
- bf70ab68a21 tu: Fix subsampled metadata and blit emission for separate stencil
- 0409984f27a lavapipe: add tensorlayout and blockloads NV_cooperative_matrix2 support
- 694c8ba1b0e lvp: add support for load_ssbo_address to vulkan lowering.
- aab1472f78b gallivm: implement load_ssbo_address intrinsic
- 06a120d42b0 nir: move load_ssbo_address to taking variable size src0
- 7105f165a21 radv: add support for coopmat2 cooperativeMatrixBlockLoads
- 737562cc29f nir: add block load decode support
- a964a8e7832 vtn: add decode option for block loads
- 02de2842022 nir: add tensor load decode function pointer.
- 6cad4ee1837 radv: add support for coopmat2 tensorAddressing
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-22 05:06:33 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r226607.d49a00bdf15f-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
d49a00bdf15f - Package:
mesa-git_26.3.0-devel+r226607.d49a00bdf15f-1_amd64.deb
Mesa upstream changes since
a11215b318f6- d49a00bdf15 r300: fixes for UB around left shifts
- 77251a488e9 nir/unlower_io_to_vars: stop sorting variables by location
- 2064e2ea68f zink: assign the IO slot map in location order
- 5a16d024bff zink: add is_unassigned_slot() helper
- 3d6080f9a8f llvmpipe: don't rely on variable list order for per primitive outputs
- 90c66c69767 llvmpipe: emit FS input vertex attributes in driver location order
- efd508cf90d util: add workaround for Riddick using round() in glsl 1.20
- 6b89caf0e02 tu: AVOID_CCU should avoid depth CCU if custom resolve writes D/S
- 02035be13a3 tu: Custom resolve should always use AVOID_CCU layout
- be8bcf120e3 ethosu: Emit explicit convolution activation ranges
- 873f0b2a41a teflon: Compute explicit convolution activation ranges
- 7814b5a17d7 etnaviv/ml: Reject clipped convolution activations
- ce99a3e8a85 gallium: Add explicit ML convolution activation range
- ea77d4e2618 intel/dev: Clamp PTL+ CS workgroup threads to 32
- 20b39ff044a nir/meson: Add idep_nir_headers as a _libnir dependency
- 61aaad5a1ad virtio: magma-gpu-rs: add android to supported systems
- 935b49fcea9 kk: Advertise proper rendering granularity
- d486e4e8a10 anv/nir_lower_descriptor_heap: zero sampler_index
- cf3da2abcb0 kraid: Add fsin/fcos and wire them up with tests
- 23f8fde15a8 kraid: Add OpFLogD and wire it up with tests
- 8e2f0d37109 kraid: Add OpFAddLScale with tests
- 5de611d64e6 kraid: Add FExp.f32 and wire it up
- 375c80d2dc3 kraid/builder: Add fadd and fma helpers
- db6651fd2e8 kraid: Add From for SrcRef
- 53a453b3ef7 vulkan/runtime: Fix bad assumption in GetPipelineBinaryDataKHR
- 5767e2f13f7 jay: add and use jay_ugpr_per_gpr
- 285c4c1ff01 jay: fix MAD restrictions
- a9ff6349292 nir: allow 64-bit scalar indices
- 8b79064907b v3d: save fragment constants on sand8/sand30 blit
- 8a6f0fca20c util/mesa: add workaround to zero invalidated buffers
- 38169ede9b2 broadcom/ci: updated expected results
- 3a475cd2287 tu: Propagate vp state for binInfo
- 998c1ae11d6 mr-label-maker: Label src/vulkan/runtime/vk_video as video
- 83f41ac6e40 aco: use s_pack_hh with shifted constant when s_pack_hl is unavailable
- 817eb9548fd brw: Make brw_geometry_stage_dispatch_width available to C code
- 489976f9e69 brw: Skip load_subgroup_size for known sizes when lowering ray queries
- 7d68e402c5d jay: Optimize load_subgroup_id to 0 for single thread workgroups
- cff3cb2606c jay: Use a switch in jay_nir_lower_simd
- 8673b8322dc jay: Make simd_width not a pointer in jay_nir_lower_simd
- 91c5e970385 jay: Don't call brw_nir_apply_key twice for geometry shaders
- 73dd0ca0793 brw: Drop brw_nir_optimize call from brw_nir_apply_key
- 3eefda3b558 iris: Drop limit_trig_input_range key field, repack keys
- 73b612fb508 intel: Make limit_trig_input_range a compiler-wide setting
- c85e3522b59 panvk/ci: Update expectations for g610-vk-asan following VK CTS uprev
- 35a89677c45 vulkan/video: Fix coding H265 SPS pcm block sizes, inter ref pic set and lt refs
- 90e7282527a vulkan/video: Fix coding H265 ref pic list modification lists
- 772cb76cc80 venus: relax SYNC_FD fence export requirement
- 6aded3b9d65 venus: migrate fence over to VN_SYNC_TYPE_SYNC
- d921c271c8b venus: prepare for VN_SYNC_TYPE_SYNC
- cfe2a5987f4 venus/virtgpu/vtest: drop initial value from sync reset
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-19 05:06:21 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r226497.a11215b318f6-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
a11215b318f6 - Package:
mesa-git_26.3.0-devel+r226497.a11215b318f6-1_amd64.deb
Mesa upstream changes since
09180f0eed60- a11215b318f kk: call opt_licm to lighten loop executions
- db128e2f552 Uprev VVL to 6727ae4734bf9e1e6f6954254bbc382687a3866d
- 879365ec9f0 vulkan/runtime: fixup vk_shader leak on RT group recompile
- cf0588a5b83 vulkan: Implement a late triangles -> quads pass
- 580f5a48ead vulkan: Make prefix_scan a shared helper
- e17cd2442ec vulkan: Add quad node
- 279fb55074d vulkan: Add a helper for querying the IR leaf node size
- 91d07a204bb vulkan: Remove some unnecessary guards
- 4b9c74d0af5 tu: Don't truncate subdraw size
- ac6e355ca7e freedreno/a6xx: Don't trucate subdraw size
- f773b1f3a7e ac/llvm: remove the optimization barrier from ac_build_ballot
- cf67c0a7e04 ac/llvm: switch to amdgcn.ballot intrinsic
- 81dca3bf64c nv30: fix another left shift cannot be represented in type 'int'
- 51a7e972353 nv30: fix 1 << 31 issues
- 4a2c1806061 anv/brw: add a compiler parameter to limit VF payload delivery
- e6adbc1cd36 anv/brw: get rid of push constant promotion limitation
- b1febd32bef intel/compiler: lower load_input in vertex shaders to load_urb if too large
- a9ce0bb99ff jay: Move VS URB setup back to jay_nir.c
- e1871005c69 brw/nir: augment load_attribute_payload_intel to express vector layout
- 4083db5c003 brw/nir: Prepare URB lowering code for VERT_ATTRIB_*
- d6ac7b917de brw/nir: Don't set ACCESS_CAN_REORDER/NON_WRITEABLE for VS inputs
- f6177e000ed brw/nir: avoid generating invalid load_input
- 71b0d055583 brw: move variable off the stack
- e498f8e77ea brw/jay: move vs component packing to common code
- 3060ae5b625 brw: Rework URB lowering max_push_bytes to be an URB read length
- 4ce4feff89b nv30: fix an issue when this push buffer is NULL
- 80e1e584d73 kraid: Validate swizzle invariants
- b3f0b3809f0 kraid: Align hw_test compilation
- f95a0665e30 kraid/hw_tests: Use Src::swizzle() for filling out sources
- 0ef5d20261c kraid: Handle single-word swizzles in Src::swizzle()
- 354c8b0f1af kraid/swizzle: Allow composing word with byte swizzles
- b5e0a9ee995 kraid/copy-prop: Handle zero swizzles
- 0e0524d1afb kraid: Handle Swizzle::ZERO in Src::word()
- 185814b6a7c kraid/swizzle: Allow SwizzleByte/Word::Zero.modify
- b1eb5f0cd44 kraid/swizzle: Add Swizzle::ZERO and an is_zero() helper
- 2c9c834f3ca kraid/swizzle: Add a SwizzleWord::modify() helper
- bb033618bbf kraid/swizzle: Implement Debug for Swizzle
- 0342e51ce44 kraid/swizzle: Fix Swizzle ToString crashing
- 1c6e7e82912 anv: fix 2d-array to 3d blits
- eb79e323da0 jay/schedule: add post-RA scheduling
- d9bfcdbcd32 jay/schedule: schedule for latency
- 78aca455872 jay/schedule: honour nosched within jay_schedule
- 296c1c62da0 jay/schedule: rename schedule_pressure -> schedule
- c4fd5d40a80 jay/schedule: add latency scheduler
- 234c264a708 jay/schedule: don't reorder timestamps
- 21db08c7c45 jay/schedule: refactors for forward scheduling
- 9cc49b00e69 jay/schedule: add a non-pressure mode
- 28bf12c95c7 jay/schedule: consider multiple scheduler phases
- 718414b4cd8 jay/schedule: add phi src special handling
- 47f1ed9c3cc jay/schedule: generalize choose_inst
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-16 05:06:15 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.3.0-devel+r226338.09180f0eed60-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
09180f0eed60 - Package:
mesa-git_26.3.0-devel+r226338.09180f0eed60-1_amd64.deb
Mesa upstream changes since
f768c98f3181- 09180f0eed6 zink: fix swap interval changes being dropped
- c6b5729a783 brw: Ignore source modifiers in needs_dst_copy
- 7a2bdd35b2b jay: Various geometry shader fixes.
- 9067042b28d jay: Use new lane_id fn for load_subgroup_invocation.
- 372c609834c kk: Move nir_lower_io_vars_to_temporaries to early msl_preprocess_nir
- 48297c83446 intel: Fix operator precedence in intel_simd_overridden
- 7635d56fba9 intel: Fix DEBUG_FS_SIMD mask
- 779cdb8eaaf freedreno/a6xx: Fix query barriers for gen8+
- 6f1739bf665 docs: update calendar for 26.2.0-rc1
- 1aad4c49094 docs: add sha sum for 26.1.5
- df23a992449 docs: add release notes for 26.1.5
- 3863fe02eeb docs: update calendar for 26.1.5
- 304e1b1d453 people: update my email address
- 40b056708b6 people: update Natalie's email
- d021138b410 anv: hide Intel vendor ID for Cyberpunk
- 1be67a7d22c nir/opt_vectorize: handle commutative ALU sources
- 05f9f41fc29 nv50_ir_ra: align B96 spill slots to vec4
- f449a3ae575 freedreno/a6xx: Split VS driver params
- df806d864a8 freedreno/a6xx: Split driver UBO upload and emit
- cbf6697c550 freedreno: Remove VS driver param patching
- 03bd1ff924b kraid/nir: Implement 64-bit integer comparisons
- 9a18efb87ee kraid: Add OpICmpMulti and plumb it through
- 48c8b0b1e5d kraid: Rework comparison folding
- e906eb87cd1 kraid/nir: Implement 64-bit integer down-conversions
- 54b8ab7e28b pan/nir: Cast 64-bit comparison results to 32 bits
- 259470dbad8 pan/nir: Use a single ALU op switch in lower_bool_to_bitsize
- ed41cf233f1 mediafoundation: fix shadow variable in stats_buffer_manager.cpp, also remove using namespace std to avoid collision
- 6e786bed043 mediafoundation: fix drain/flush when running ProcessDX12Context outside of m_lock, and allocator/buffer pool release.
- 0b3f2666029 mesa: fix variable shadowing
- fb7300c6da2 v3dv: record the no-op fragment shader with its Vulkan stage bit
- 1d9201e377e v3dv: record the multiview geometry shader with its Vulkan stage bit
- 771dfe55fcb v3dv: use the binning VS variant for the binning VPM config
- 9d534e38a2e v3dv: make room in the descriptor map for the no-sampler entries
- 6895c0736ac docs: reset new_features.txt
- 06272731414 VERSION: bump to 26.3
- 7aeab89318c pipe: Remove pipe_video_codec::chroma_format
- be00f53d4d5 broadcom/compiler: abort 2-thread spill loops over the best result so far
- 9510af092a6 broadcom/ci: unskip CTS tests that are no longer slow
- 6f7939f517c broadcom/compiler: drop V3D_DEBUG=opt_compile_time
- 7d9ee7960c3 broadcom/compiler: don't leak the compile on assembly allocation failure
- b44e72654c4 broadcom/compiler: pick the 2-thread compile strategy by register pressure
- 7ce6634c87e broadcom/compiler: move the 2-thread strategies to compile_2t_strategies
- a2fe9573a13 broadcom/compiler: consolidate the compile-strategy logging
- 91ab53b4b90 broadcom/compiler: support probing a compile's pre-spill register pressure
- b35772803b9 broadcom/compiler: split v3d_nir_to_vir_finish out of v3d_nir_to_vir
- 3ec0d524dde broadcom/compiler: reduce the compile-strategy fallback ladder
- 91b68c0bc32 jay: add geometry shader support.
- 476ae41b70d jay: rewrite flags
- 8f3c37380b8 jay: drop #include
- 4ab8dc33c1c jay: support as_uniform
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit:
-
2026-07-13 05:06:29 +02:00 | 0 commits to main since this releaseRepo Changes
- Repo commit:
ff36e283de4c2a19d352d9155cd5d689f97c3e19 - Release tag:
mesa-git-26.2.0-devel+r226085.f768c98f3181-1
Recent repo commits
ff36e28Harden mesa package for Ubuntu and Debian targets
Mesa Upstream
- Upstream:
https://gitlab.freedesktop.org/mesa/mesa.git - Ref:
main - Commit:
f768c98f3181 - Package:
mesa-git_26.2.0-devel+r226085.f768c98f3181-1_amd64.deb
Mesa upstream changes since
ac8773f05fd4- f768c98f318 freedreno/ci: document fixed tests
- deddba9add2 nvk/ci: document two vkd3d regressions
- 5da9e9183c4 nvk/ci: document fixed vkd3d tests
- 70dc0466fc2 nvk/ci: document fixed dEQP-VK test
- 58373d6addc venus: properly clean up driver internal sim syncobj allocs
- 584a0997c8e tu: Propagate allocation failures for tu_cs_* functions
- 9e6e6e5db7c pan/perf: fix building error due to 'Mali G1.xml' file name with space
- 02e187d684c rusticl: abort compilation if the nir shader is not fully linked
- 1fe24423672 rusticl: return Result instead of Option from convert_spirv_to_nir
- a21811f34b9 nir: add nir_shader_fully_linked helper
- 7b804b1a81f nir: use function foreach helpers inside nir_cleanup_functions
- 2e0ad862b58 nvk: Use ?: in nvk_physical_device_compiler_flags
- cb58a85cc74 nvk: Move ssbo_4b_align to drirc
- 41508406dc5 nvk: Don't re-align ssbo size/address
- a99a00044a6 nvk: Advertise minStorageBufferOffsetAlignment=4 for VKD3D
- 2d212623b38 kraid/nir: Implement nir_op_[iu]mul_2x32_64
- cb05eb808e9 kraid: Document a couple passes
- 4f74c7ad6c9 kraid: Optimize shift+logic ops
- 5ec3baf9427 kraid: Use the new NIR shift+logic ops
- 3c4c8247e96 nir: Add combined shift-logic ops for panfrost
- f92f07bf262 kraid/nir: Enable more 64-bit ops
- 08980957f98 kraid/nir: Implement 64-bit op_bitfield_select
- f0d4376c5ca kraid/copy-prop: Handle 64-bit OpShiftLop
- a16bbdc5146 kraid: Align registers based on the new ISA query
- a026a3e84ef kraid/isa,model: Add a op_srs_is_64bit() query
- 03696e9a6d7 Uprev VVL to e875181c4f0fab1bd0c4926c21a21de0bd967bb3
- 44be0a749ae pan: Use Kraid's NIR options when it's enabled
- a84cd8930c8 kraid: Expose our own NIR compiler options
- 935011f6807 pan: Move PAN_USE_KRAID into pan_compiler.c/h
- 9f836f3990a pan: Take a stage parameter to get_nir_shader_compiler_options()
- 93dc3517e87 panvk: simplify non_polygon calculation
- 74b5819d5ff gfxstream: Validate guest mapped-memory ranges in flush/invalidate
- b16af1f337c d3d12: Use res_device to import resource from different device via handle
- 63165596d2e d3d12: Fix NULL dereference check in d3d12_video_buffer_destroy
- 2a3dddbfff7 v3d: add support for array of textures blit with TFU
- cfe64a989b3 v3dv: enable VK_KHR_shader_maximal_reconvergence
- f365ae7bb6c gallivm: fix subtle filtering issue with different min/mag filter for cube maps
- 578bf3cf273 intel/gen: allow dead loops
- e6e3f33e46f intel/gen: Fix case where IF is a loop header
- 0e16ea6574c brw: Fix initializing matrix with uniform but non-constant values
- 0e79791fa5f nouveau: fix data race in nouveau_fence_ref
- ec2a3402c29 nak/hw_tests: add F2I test for NaN behavior
- 5e93b7bb9e9 nak/sm20: fix immediate encoding for F2I and F2F
- 8489c4fbb51 nvk: Convert draw/dispatch to device_address_commands
- 189da48917c nvk: Prepare indirect draws for 64-bit stride
- 9a029ea3839 nvk: Handle large indirect stride pre-Turing
- 43dafe15909 nvk: Replace nvk_addr_range with VkDeviceAddressRange
- 0e5b555e185 meson: ensure that libdrm auto-features match requirements
- 1a9f74263aa kk: Refactor encoder state
- 6d51a0420a8 radv: force late-Z with fragment shaders that use fbfetch
Mesa upstream changes truncated to the first 50 commits.
Downloads
- Repo commit: