xref: /aosp_15_r20/external/mesa3d/docs/relnotes/23.1.2.rst (revision 6104692788411f58d303aa86923a9ff6ecaded22)

Mesa 23.1.2 Release Notes / 2023-06-08
======================================

Mesa 23.1.2 is a bug fix release which fixes bugs found since the 23.1.1 release.

Mesa 23.1.2 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 23.1.2 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
---------------

::

    60b1f3adb1561830c158bf3c68508943674fb9d69f384c3c7289694385ab5c7e  mesa-23.1.2.tar.xz


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

- None


Bug fixes
---------

- Rusticl OpenCL: Simple SYCL / DPC++ program hangs indefinitely at rusticl::api::event::wait_for_events()
- radeonsi: Metro Last Light Redux graphical glitches
- ANV: Vulkan driver regression in clearing Depth/Stencil
- freedreno/a6xx: assert(state->view_rsc_seqno[i] == seqno) failed with sway
- radv crashes when using vertex format VK_FORMAT_B10G11R11_UFLOAT_PACK32
- changes in commit e4b6a0a82457b3ef40c5857412e20bc344ff302c leads to GPU hang
- iris now requires Linux v5.3
- Wolfenstein II: The New Colossus vsync off crash
- Surfaceless mode ES2.0 number of vertices limitation
- radv: unaligned vertex input regression
- glSpecializeShaderARB works on SPIR-V compiled with shader compilers ca. 2021, but fails for SPIR-V generated with current compilers
- [regression/bisected] Bone wireframes are no longer rendered correctly in Blender with RadeonSi/Vega


Changes
-------

Blisto (1):

- driconf: set vk_x11_strict_image_count for Wolfenstein II

Caio Oliveira (2):

- spirv: Extract vtn_handle_debug_text() helper
- spirv: Fix gl_spirv_validation when OpLine with strings is present

Chia-I Wu (4):

- radv: do not use a pipe offset for aliased images
- aco: fix alignment check in emit_load
- radv: fix msaa feedback loop without tc-compat cmask
- radv: fix non-square compressed image copy on gfx9

Christian Gmeiner (1):

- etnaviv: nir: do not call nir_lower_idiv(..) unconditionally

Dave Airlie (10):

- radv/video: fix h264/265 dpb usage.
- radv/video: add missing offset to the dpb binding.
- radv/video: fix physical device format property count.
- radv/meta: fix uninitialised stack memory usage.
- lavapipe: when in doubt, swizzle the swizzle
- lavapipe: fix pipeline sanitizing.
- llvmpipe: emit fences for barrier.
- lavapipe: don't remove queue family barriers.
- gallivm/nir: fix shuffleup tests.
- lavapipe: ignore another yuv format.

Emma Anholt (1):

- tu: Ignore unused shader stages in pipeline library creation.

Eric Engestrom (7):

- docs/relnotes: add sha256sum for 23.1.1
- .pick_status.json: Update to f85c286f51bab7d476111b781798ddb445c3f2d5
- .pick_status.json: Mark c9288808db08fd804c68750d7400c01e550c70d0 as denominated
- .pick_status.json: Update to b222b34e789812301f4f7b79911f9edc599937e2
- .pick_status.json: Update to 476022cbe1d1a06e43aacd81448a7a22c6331b1f
- .pick_status.json: Update to 75b63a5d3e4c0816b55deaf8d105d28bfebea98b
- docs: clear new_features.txt

Erik Faye-Lund (2):

- zink: compute correct location for line-smooth gs
- zink: do not lower line-smooth for non-lines

Francisco Jerez (1):

- anv: Fix calculation of guardband clipping region.

Friedrich Vock (2):

- nir: Rematerialize derefs in use blocks before repairing SSA
- nir: Remove unnecessary assert in nir_before_src

Gert Wollny (1):

- virgl: Fix IB upload when a start >0 is given

Hans-Kristian Arntzen (1):

- Fix DGC bug where indirect count > maxSequencesCount.

Jesse Natalie (1):

- nir_opt_algebraic: Don't shrink 64-bit bitwise ops if pack_split is going to be lowered

José Roberto de Souza (1):

- intel: Fix support of kernel versions without DRM_I915_QUERY_ENGINE_INFO

Julia Tatz (1):

- zink: fix layout(local_size_variable) for vk1.3+

Karol Herbst (2):

- rusticl: bump rust req to 1.60
- rusticl/event: flush queues from dependencies

Kenneth Graunke (1):

- intel/compiler: Fix a fallthrough in components_read() for atomics

Konstantin Seurer (2):

- gallivm: Fix gather/scatter types for newer llvm
- gallivm: Cast read_first_invocation source to an int

Lionel Landwerlin (2):

- anv: fix push range for descriptor offsets
- anv: fix null descriptor handling with A64 messages

Lucas Stach (1):

- etnaviv: update derived state after forced commandstream flush

Marek Olšák (2):

- radeonsi: re-enable fp16_rtz for compute blits to fix PBO tests on gfx11
- radeonsi: don't use SET_SH_REG_INDEX on gfx7-9

Matt Turner (1):

- u_format: Use memcpy to avoid unaligned accesses

Mike Blumenkrantz (5):

- zink: don't remove psiz from linked shaders if the consumer reads it
- zink: don't propagate psiz in quads emulation gs
- lavapipe: fix DS3 min sample setting
- lavapipe: don't pass indirect info in streamout draws
- lavapipe: handle index buffers with offsets for indirect draws

Patrick Lerda (4):

- mesa: fix refcnt imbalance related to _mesa_delete_semaphore_object()
- mesa/st: fix refcnt imbalance related to st_feedback_draw_vbo()
- mesa/st: fix buffer overflow related to set_program_string()
- r600: fix r600_draw_vbo() buffer overflow

Pierre-Eric Pelloux-Prayer (1):

- Revert "gallium/u_threaded: buffer sharedness tracking"

Qiang Yu (1):

- radeonsi: fix uses_instanceid for merged mono shader stage

Rhys Perry (1):

- radv: correctly skip vertex loads with packed formats

Rob Clark (4):

- freedreno: Add extra assert
- freedreno/batch: Add driver-thread assert
- freedreno/a6xx: Directly invalidate on samp view update
- freedreno/a6xx: Use idalloc for samp/view seqno's

Rohan Garg (4):

- iris: correctly set alignment to next power of two for struct size
- anv: enable single texel alignment
- anv: fix incorrect asserts when combining CPS and per sample interpolation
- hasvk: enable single texel alignment

Sagar Ghuge (1):

- anv: Set CS stall bit during HIZ_CCS_WT surface fast clear

Samuel Pitoiset (3):

- radv: fix copying 2D to 3D images
- amd: fix 64-bit integer color image clears
- radv: initialize the device cache UUID even if on-disk cache is disabled

Tatsuyuki Ishi (1):

- radv: Make shader related destruction happen before hw_ctx.

Timothy Arceri (1):

- util: add Jamestown+ workaround

Vinson Lee (4):

- freedreno/a6xx: Fix memory leak on error path.
- nv50: Fix memory leak in error path
- dzn: Fix qpool->queries_lock double lock
- vulkan/wsi: Remove duplicate NULL check

Víctor Manuel Jáquez Leal (1):

- vulkan: complete the usage flags for video layouts

Yonggang Luo (1):

- virgl: Array cmdline on stack should initialized to 0