docs/relnotes/23.1.4.rst

Mesa 23.1.4 Release Notes / 2023-07-21
======================================

Mesa 23.1.4 is a bug fix release which fixes bugs found since the 23.1.3 release.

Mesa 23.1.4 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 23.1.4 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
---------------

::

    7261a17fb94867e3dc5a90d8a1f100fa04b0cbbde51d25302c0872b5e9a10959  mesa-23.1.4.tar.xz


New features
------------

- None


Bug fixes
---------

- radeonsi: Deadlock when creating a new GL context in parallel with linking a shader on another GL context
- robustness2 raygen tests intermittently fail in Intel Mesa CI
- glthread: huge performance regression
- DirectX games do not launch on Intel HD Graphics 4000 (IVB GT2) [bisected]
- anv: GPU hangs on MSAA tests with Angle
- AMD OpenGL texture corruption and crashing regression in java app
- The filenumber part of the #line preprocessor directive is ignored for multiline error messages
- Confidential issue #4103
- r600 regression
- [bisected][anv] newly enabled test (dEQP-VK.api.info.format_properties.g8_b8r8_2plane_420_unorm) failing
- deqp gles3.1 gpu hangs on DG2 A380 when running zink
- freedreno: firefox crashes on video playback
- radeon: Basphemous graphical glitch
- aztec ruins gl benchmark slow to compile shaders on intel
- anv: false cacheline flushing or insufficient buffer alignment on at least ADL
- [Vulkan][TGL] vkCmdCopyQueryPoolResults failed to write buffer after compute shader write with overlap


Changes
-------

Alyssa Rosenzweig (4):

- nir/lower_bool_to_int32: Fix progress reporting
- pan/mdg: Fix temp count calculation
- nir: Add nir_foreach_function_impl helper
- panfrost: Fix transform feedback on v9

Billy Laws (1):

- wgl: Fix depth/stencil image support when using zink kopper

Boris Brezillon (3):

- renderonly: Fix potential NULL deref in the error path
- renderonly: Make sure we reset scanout on error in create_kms_dumb_buffer_for_resource()
- winsys/panfrost: Make sure we reset scanout on error in create_kms_dumb_buffer_for_resource()

Corentin Noël (2):

- compiler: Allow the explicit_stride of aoa types to be zero
- nir/split_64bit_vec3_and_vec4: Use the right number of components

Dave Airlie (3):

- lavapipe: check sampler pointer before deref
- draw/gs: handle extra shader outputs in geometry.
- llvmpipe/linear: don't allow linear path for shader output with location frac

Eric Engestrom (11):

- docs/relnotes: add sha256sum for 23.1.3
- .pick_status.json: Update to 337908440eefc95abef9fa4464e29d299072df73
- .pick_status.json: Update to f9e6d35f65c6c5913d54ba5306af391a8701a68a
- asahi: drop unnecessary DRM_FORMAT_MOD_{LINEAR,INVALID} fallbacks
- .pick_status.json: Mark af895692b3b52599e456f14e08a3beaa32b5a4c8 as denominated
- .pick_status.json: Update to e67337bebf623e67d119f48c2d24f3b831482ede
- .pick_status.json: Update to ba0ceb3be33766e2d96d50be6827e709c25f57dc
- .pick_status.json: Update to 9bcdc45ee7e8c4ca0dbb18ba16077d474d5ebb31
- util/disk_cache: fix ~/.cache/ permissions
- .pick_status.json: Update to cfd24047a1d901d209b18828fd91f9ec36ef0fed
- .pick_status.json: Mark 79f7882fc604515302351282f5e01ec5907727f7 as denominated

Faith Ekstrand (1):

- nv50/ir: Support vector movs

Feng Jiang (2):

- frontends/va: Fix memory leak of decrypt_key
- virgl/video: Fix out-of-bounds access in fill_mpeg4_picture_desc()

Filip Gawin (1):

- crocus: Avoid fast-clear with incompatible view

Francisco Jerez (1):

- intel/gfx12.5: Enable L3 partial write merging for compressible surfaces among other cases.

Frank Binns (1):

- pvr: skip setting up SPM consts buffer when no const shared regs are used

Friedrich Vock (2):

- radv/rt: Enable exact on software intersection functions
- radv/rt: Miss rays that hit the triangle's v edge

Gert Wollny (2):

- r600/sfn: Don't clear clear group flag on vec4 that comes from TEX or FETCH
- r600: Pre-EG  - Set wrap texture modes to repeat when seemless cube is used

Hyunjun Ko (1):

- anv/video: fix to set U/V offset correctly.

Iago Toral Quiroga (1):

- broadcom/compiler: don't leak v3d_compile when finding a new best strategy

Iván Briano (1):

- anv: ensure CFE_STATE is emitted for ray tracing pipelines

Janne Grunau (1):

- st/mesa: Set gl_config.floatMode based on color_format

Jesse Natalie (1):

- d3d12: Fix indexing of local_reference_state

José Roberto de Souza (1):

- anv: Fix compute maximum number of threads value

Juston Li (1):

- radv: fix incorrect size for primitives generated query

Karmjit Mahil (2):

- pvr: Fix seg fault on empty descriptor set
- pvr: Fix dynamic offset patching

Karol Herbst (1):

- nvc0: backport fp helper invocation fix to 2nd gen Maxwell+

Kenneth Graunke (1):

- intel: Initialize FF_MODE2 on all Gfx12 platforms

Konstantin Seurer (2):

- gallivm: Clamp the texel buffer size
- nir/opt_dead_cf: Handle if statements ending in a jump correctly

Lina Versace (1):

- venus: Fix detection of push descriptor set

Lionel Landwerlin (6):

- anv: limit ANV_PIPE_RENDER_TARGET_BUFFER_WRITES to blorp operations using 3D
- anv: track buffer writes from shaders for query results writes
- anv: align buffers to a cache line
- anv: fix utrace batch allocation
- intel/fs: disable coarse pixel shader with interpolater messages at sample
- anv: fix utrace signaling with Xe

Marek Olšák (5):

- vbo: correctly restore _VaryingInputs for display list fast path
- glthread: determine global locking once every 64 batches to fix get_time perf
- mesa: fix 38% decrease in display list performance of Viewperf2020/NX8_StudioAA
- util/u_queue: fix util_queue_finish deadlock by merging lock and finish_lock
- lima: update CI fixes and flakes

Matt Turner (2):

- anv: Pipe anv_physical_device to anv_get_image_format_features2
- anv: Only expose video decode bits with KHR_video_decode_queue

Mike Blumenkrantz (9):

- nir/lower_tex: ignore saturate for txf ops
- zink: fix assert for inline uniform invalidation with generated gs bound
- zink: fix unbinding generated gs on real gs bind
- zink: get new bda when rebinding invalidated buffers
- zink: do initial program unref during program creation
- zink: fix separate shader program refcounting
- zink: don't destroy swapchain on initial CreateSwapchainKHR fail
- aux/trace: fix bindless texture dumping
- zink: fix batch disambiguation on first submit

Patrick Lerda (3):

- util/blitter: revert util_blitter_clear_buffer()
- radeonsi: fix refcnt imbalance related to util_blitter_save_fragment_constant_buffer_slot()
- panfrost: fix refcnt imbalance related to blitter

Rhys Perry (4):

- ac/llvm: fix wave32 ac_build_mbcnt_add with 64-bit mask
- ac/llvm: skip ballot zext for 32-bit dest with wave32-as-wave64
- radv: add conformant_trunc_coord to cache UUID
- nir/unsigned_upper_bound: fix phi(bcsel)

Rob Clark (4):

- freedreno/drm: Don't try to export suballoc bo
- freedreno: Handle export error handling
- freedreno: Add aux-context support
- freedreno: Reallocate on unshared export

Robert Beckett (1):

- winsys/panfrost: Fix a scanout resource leak

Rohan Garg (1):

- intel/perf: add perf query support for Intel Raptorlake

Samuel Pitoiset (3):

- radv/amdgpu: fix dumping cs with RADV_DEBUG=noibs
- radv/amdgpu: skip adding per VM BOs for sparse during CS BO list build
- radv/amdgpu: workaround a kernel bug when replacing sparse mappings

Semjon Kravtsenko (1):

- glx: Assign unique serial number to GLXBadFBConfig error

Shan-Min Chao (1):

- tu/kgsl: Fix memory overwrite with vkFlushMappedMemoryRanges when more than 1 range

SoroushIMG (1):

- pvr: Fix barrier insertion on merged subpasses

Sviatoslav Peleshko (1):

- glsl: Fix yylloc.source propagation in YYLLOC_DEFAULT

Timothy Arceri (1):

- glsl: call nir_opt_find_array_copies() when linking

Timur Kristóf (2):

- ac/nir/ngg: Call nir_convert_to_lcssa before divergence analysis.
- aco: Fix subgroup_id intrinsic on GFX10.3+.

Vinson Lee (1):

- tu: Fix missing unlock

Yiwei Zhang (3):

- venus: suballocate feedback slot with feedback buffer alignment
- venus: fix a cmd tmp storage leak
- venus: fix leaks from tracked present src images

Yonggang Luo (1):

- mapi: Fixes check_table.cpp for DrawArraysInstancedARB and DrawElementsInstancedARB