Mesa 22.3.4 Release Notes / 2023-01-26

Mesa 22.3.4 is a bug fix release which fixes bugs found since the 22.3.3 release.

Mesa 22.3.4 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 22.3.4 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

37a1ddaf03f41919ee3c89c97cff41e87de96e00e9d3247959cc8279d8294593  mesa-22.3.4.tar.xz

New features

  • None

Bug fixes

  • anv: Performance issue with Vulkan on Wayland KWin

  • VAAPI HEVC encode broken since 22.3

  • GPU HANG: ecode 12:1:859ffffb (Resetting rcs0 for stopped heartbeat on rcs0) - reproducible

  • zink: src/gallium/auxiliary/pipebuffer/pb_slab.c:138: Assertion failed: `heap < slabs->num_heaps`

  • [zink] Assertion `heap < slabs->num_heaps’ failed on Pascal (bisected)

  • [RADV] Incorrect copies to/from compressed textures with mipmaps

  • mesa_glthread=true and probably ANY id Tech 3 engine games, offroad…

  • radeonsi: VRAM Leak/abnormally high usage in Minecraft mod pack

  • libmesa_util depends on gallium

  • debug build compilation failed: inlining failed in call to ‘always_inline’ ‘src_is_ssa’: indirect function call with a yet undetermined callee

  • radv: Missing implementation of VkImageSwapchainCreateInfoKHR and VkBindImageMemorySwapchainInfoKHR

Changes

Alejandro Piñeiro (1):

  • vulkan/wsi: check if image info was already freed

Alyssa Rosenzweig (5):

  • pan/bi: Fix incorrect compilation of fsat(reg.yx)

  • panfrost: Enable NV_primitive_restart on Valhall

  • panfrost: Fix logic ops on Bifrost

  • panfrost: Fix clears with conditional rendering

  • mesa: Set info.separate_shader for ARB programs

Bas Nieuwenhuizen (1):

  • aco: Pass correct number of coords to Vega 1D LOD instruction.

Danylo Piliaiev (1):

  • tu/kgsl: do not use kgsl_command_object::offset

Dave Airlie (5):

  • lavapipe: fix descriptor set layout reference counting in layout merge

  • lavapipe: fix merged layout ref counting

  • llvmpipe: unref images correctly in the fragment shader path

  • vk_cmd_queue: add a callback to free push descriptors set khr

  • lavapipe: fix one more descriptor set reference issue

Emma Anholt (5):

  • docs/perfetto: Fix the name of the i915 render stages data sources.

  • zink: Add missing Flat decorations on some inputs.

  • zink: Fix validation failure for maxLod < minLod.

  • zink: Fix up mismatches of memory model vs addressing model.

  • zink: Re-emit the SpvBuiltInSampleMask access chain each load.

Eric Engestrom (4):

  • docs/relnotes: add sha256sum for 22.3.3

  • .pick_status.json: Update to 9f51340b99a39be80bf1bc6d4be15a87161c7a6b

  • .pick_status.json: Mark e527f686caaf688f0a185e2641fc96413dd502be as denominated

  • gen_release_notes: include links in relnotes.rst when generating the new release note

Erik Faye-Lund (2):

  • zink: fix depth-clip disable cap

  • radeonsi: respect smoothing_enabled

Francisco Jerez (2):

  • intel/fs/gfx12: Ensure that prior reads have executed before barrier with acquire semantics.

  • intel/fs: Fix src and dst types of LOAD_PAYLOAD ACP entries during copy propagation.

Friedrich Vock (1):

  • radv/rt: Divide by the correct workgroup size

Gert Wollny (2):

  • r600/sfn: Fix readport check

  • virgl: drop the separable flag for cases that can’t be handled

Italo Nicola (1):

  • panfrost: fix off-by-one when exporting format modifiers

Faith Ekstrand (1):

  • gallium,util: Pull u_indices and u_primconvert back into gallium

Jonathan Gray (1):

  • egl/dri2: avoid undefined unlocks

Julia Tatz (3):

  • zink: zink_heap isn’t 1-to-1 with memoryTypeIndex

  • zink: trival renames heap_idx -> memoryTypeIndex

  • zink: correct sparse bo mem_type_idx placement

Kenneth Graunke (1):

  • intel/compiler: Drop redundant 32-bit expansion for shared float atomics

Lionel Landwerlin (4):

  • nir/divergence: add missing RT intrinsinc handling

  • nir/lower_io: fix bounds checking for 64bit_bounded_global

  • intel/fs: avoid cmod optimization on instruction with different write_mask

  • vulkan/wsi/wayland: improve same gpu detection

Marcin Ślusarz (1):

  • intel/compiler/mesh: handle const data in task & mesh programs

Marek Olšák (4):

  • glthread: fix an upload buffer leak

  • util: fix util_is_vbo_upload_ratio_too_large

  • mesa: allow GL_UNSIGNED_INT64_ARB as vertex format for ARB_bindless_texture

  • glthread: handle GL_*_ARRAY in glEnable/Disable

Maíra Canal (1):

  • v3dv: remove unused clamp_to_transparent_black_border property

Mike Blumenkrantz (7):

  • zink: store drm format as internal_format for imported resources

  • zink: handle modifier nplanes queries correctly for planar formats

  • zink: preserve present resources during async presentation

  • zink: delete need_blend_constants

  • zink: don’t use ds3 blend states without color attachments

  • zink: use actual swapchain object for surface comparison

  • zink: fix VK_DYNAMIC_STATE_LINE_WIDTH usage

Paulo Zanoni (4):

  • anv: don’t leave undefined values in exec->syncobj_values

  • anv: check the return value of anv_execbuf_add_bo_bitset()

  • hasvk: don’t leave undefined values in exec->syncobj_values

  • hasvk: check the return value of anv_execbuf_add_bo_bitset()

Pierre-Eric Pelloux-Prayer (4):

  • vbo: remove bogus assert

  • vbo: lower VBO_SAVE_BUFFER_SIZE to avoid large VRAM usage

  • glthread: fix glArrayElement handling

  • radeonsi/gfx11: fix ge_cntl programming

Rhys Perry (1):

  • radv/gfx11: increase radeon_check_space for occlusion query begin

Rob Clark (1):

  • freedreno: Fix tracking of enabled SSBOs

Rose Hudson (1):

  • radeonsi: report 0 block size for Polaris HEVC encoding

Samuel Pitoiset (5):

  • radv: fix missing implementation of creating images from swapchains

  • radv: fix setting MAX_MIP for BC views

  • radv: fix buffer to image copies with BC views on the graphics queue

  • radv: fix creating BC image views when the base layer is > 0

  • radv: fix ignoring graphics shader stages that don’t need to be imported

Simon Ser (1):

  • zink: fix PIPE_RESOURCE_PARAM_NPLANES with format modifier

Tapani Pälli (3):

  • anv: add restrictions for 3DSTATE_RASTER::AntiAliasingEnable

  • hasvk: add restrictions for 3DSTATE_RASTER::AntiAliasingEnable

  • iris: add restrictions for 3DSTATE_RASTER::AntiAliasingEnable

t0b3 (1):

  • nir/nir_opt_move: fix ALWAYS_INLINE compiler error