Mesa 24.0.3 Release Notes / 2024-03-13

Mesa 24.0.3 is a bug fix release which fixes bugs found since the 24.0.2 release.

Mesa 24.0.3 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 24.0.3 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

77aec9a2a37b7d3596ea1640b3cc53d0b5d9b3b52abed89de07e3717e91bfdbe  mesa-24.0.3.tar.xz

New features

  • None

Bug fixes

  • v3d: Line rendering broken when smoothing is enabled

  • DR crashes with mesa 24 and rusticl (radeonsi)

  • RADV: GPU crash when setting ‘RADV_DEBUG=allbos’

  • [intel] mesa ftbfs with time_t64

  • [radv] Crash when VkGraphicsPipelineCreateInfo::flags = ~0u

  • Gen4 assertion `force_writemask_all’ failed.

  • [radv] Holographic projection texture glitch in Rage 2

  • [build failure] [armhf] - error: #error “_TIME_BITS=64 is allowed only with _FILE_OFFSET_BITS=64”

  • RustiCL: Callbacks are not called upon errors

  • MTL: regressions in vulkancts due to BO CCS allocations

  • zink: spec@ext_external_objects@vk-image-overwrite fail

Changes

Boyuan Zhang (1):

  • meson: bump the minimal required vdpau version to 1.4

Caio Oliveira (1):

  • intel/compiler: Fix SIMD lowering when instruction needs a larger SIMD

Chia-I Wu (1):

  • aco: fix nir_op_pack_32_4x8 handling

Christian Gmeiner (1):

  • etnaviv: Fix how we determine the max supported number of varyings

Corentin Noël (1):

  • zink: Return early if the file descriptor could not have been duplicated/acquired

Daniel Schürmann (1):

  • radv: fix initialization of radv_shader_layout->use_dynamic_descriptors

Danylo Piliaiev (1):

  • tu: Fix dynamic state not always being emitted

David Heidelberg (6):

  • drm-shim: Avoid invalid file and time bits combination

  • ci/intel: decompose anv-tgl-test so we can specify custom devices for TGL

  • ci/intel: add acer-cp514-2h-11{30,60}g7-volteer

  • ci/intel: move machine definition to the intel-tgl-skqp job

  • ci/intel: split asus-cx9400-volteer into acer-cp514-2h-11{30,60}g7-volteer

  • intel/tools: avoid invalid time and file bits combination

David Rosca (1):

  • radeonsi/vcn: Use temporal_layer_index to select temporal layer

Eric Engestrom (6):

  • docs: add sha256sum for 24.0.2

  • .pick_status.json: Update to 7792ee1c15379d95ccb20ce34352473f2bb2bfbd

  • .pick_status.json: Update to f3fe1f2f18d7ccc8a7cf85cd88c4bdf426445702

  • .pick_status.json: Update to e1afffe7fa7bd8e1cd1f7e58cfa2f33faf889628

  • .pick_status.json: Mark a367cd49314a993d09168e790d3090a2303a48d9 as denominated

  • .pick_status.json: Update to 9a57b1df5395bbcaa6f48ea851860bedc7ceefb9

Eric R. Smith (1):

  • panfrost: protect alpha calculation from accessing non-existent component

Faith Ekstrand (4):

  • nvk: Return os_page_size for minMemoryMapAlignment

  • nvk: Document the register name for the helper load workaround

  • nvk: Always wait for the FALCON in set_priv_reg

  • nvk: Disable the Out Of Range Address exception

Felix DeGrood (1):

  • driconf: add SotTR DX12 to Intel XeSS workaround

Friedrich Vock (3):

  • radv/rt: Handle monolithic pipelines in capture/replay

  • radv: Set SCRATCH_EN for RT pipelines based on dynamic stack size

  • radv/rt: Fix frontface culling with emulated RT

Georg Lehmann (6):

  • aco: create pseudo instructions with correct struct

  • aco/post-ra: rename overwritten_subdword to allow additional uses

  • aco/post-ra: assume scc is going to be overwritten by phis at end of blocks

  • aco: store if pseudo instr needs scratch reg

  • aco/post-ra: track pseudo scratch sgpr/scc clobber

  • aco/ssa_elimination: check if pseudo scratch reg overwrittes regs used for v_cmpx opt

Gert Wollny (2):

  • zink: use only ZINK_BIND_DESCRIPTOR

  • zink/nir-to-spirv: Make sure sampleid for InterpolateAtSample is int

Ian Romanick (1):

  • i915: Fix value returned for PIPE_CAP_MAX_TEXTURE_CUBE_LEVELS

Jesse Natalie (3):

  • wgl: Check for stw_device->screen before trying to destroy it

  • wgl: Initialize DEVMODE struct

  • nir_lower_tex_shadow: For old-style shadows, use vec4(result, 0, 0, 1)

Job Noorman (1):

  • ir3: fix alignment of spill slots

Jonathan Gray (1):

  • intel/dev: update DG2 device names

Jose Maria Casanova Crespo (1):

  • ci: Adds /usr/local/bin to PATH at piglit-traces.sh

José Roberto de Souza (1):

  • iris/xe: Consider pat_index while unbinding the bo

Juan A. Suarez Romero (2):

  • v3d: add load_fep_w_v3d intrinsic

  • v3d: fix line coords with perspective projection

Karol Herbst (1):

  • rusticl/event: we need to call the CL_COMPLETE callback on errors as well

Kenneth Graunke (2):

  • intel/brw: Allow CSE on TXF_CMS_W_GFX12_LOGICAL

  • iris: Fix tessellation evaluation shaders that use scratch

Konstantin Seurer (2):

  • radv/rt: Use doubles inside intersect_ray_amd_software_tri

  • radv/rt: Fix raygen_imported condition

Lionel Landwerlin (3):

  • anv: fix non matching image/view format attachment resolve

  • anv: fix incorrect ISL usage in buffer view creation

  • anv/iris/blorp: use the right MOCS values for each engine

Mike Blumenkrantz (16):

  • zink: apply all storage memory masks to control barriers if no modes are specified

  • zink: emit SpvCapabilityImageMSArray for ms arrayed storage images

  • zink: null out bo usage when allocating from slab

  • zink: fix unsynchronized read-mapping of device-local buffers

  • zink: force max buffer alignment on return ptrs for mapped staging buffers

  • zink: fix stencil-only blitting with stencil fallback

  • vulkan/dispatch_table: add an uncompacted version of the table

  • zink: use uncompacted vk_dispatch_table

  • egl/dri2: use the right egl platform enum

  • zink: stop enabling EXT_conservative_rasterization

  • zink: fix PIPE_CAP_MAX_SHADER_PATCH_VARYINGS

  • zink: call CmdSetRasterizationStreamEXT when using shader objects

  • nvk: bump NVK_PUSH_MAX_SYNCS to 256

  • util/blitter: iterate samples in stencil_fallback

  • mesa: fix CopyTexImage format compatibility checks for ES

  • driconf: add radv_zero_vram for Crystal Project (1637730)

Oskar Viljasaar (1):

  • compiler/types: Fix glsl_dvec*_type() helpers

Patrick Lerda (2):

  • r300: fix constants_remap_table memory leak

  • radeonsi/gfx10: fix main_shader_part_ngg_es memory leak

Pierre-Eric Pelloux-Prayer (1):

  • radeonsi: try to disable dcc if compute_blit is the only option

Rhys Perry (1):

  • aco: don’t combine linear and normal VGPR copies

Robert Beckett (1):

  • vulkan/wsi: fix force_bgra8_unorm_first

Rohan Garg (1):

  • anv, blorp: Set COMPUTE_WALKER Message SIMD field

Samuel Pitoiset (5):

  • radv: fix conditional rendering with direct mesh+task draws and multiview

  • radv: fix conditional rendering on compute queue on GFX6

  • radv: add missing conditional rendering for indirect dispatches on GFX6

  • radv: enable radv_zero_vram for RAGE2

  • util/u_debug: fix parsing of “all” again

Simon Ser (1):

  • egl/wayland: ensure wl_drm is available before use

Tapani Pälli (4):

  • iris: make sure aux is disabled for external objects

  • anv: make sure aux is disabled for memory objects

  • hasvk: make sure aux is disabled for memory objects

  • crocus: make sure aux is disabled for memory objects

Vasily Khoruzhick (4):

  • lima: ppir: always use vec4 for output register

  • lima: ppir: use dummy program if FS has empty body

  • lima: gpir: abort compilation if load_uniform instrinsic src isn’t const

  • lima: update expected CI failures

Yiwei Zhang (1):

  • venus: fix ffb batch prepare for a corner case and avoid a memcpy UB

qbojj (1):

  • vulkan: Fix calculation of flags in vk_graphics_pipeline_state_fill