docs/relnotes/23.0.4.rst

Mesa 23.0.4 Release Notes / 2023-05-30
======================================

Mesa 23.0.4 is a bug fix release which fixes bugs found since the 23.0.3 release.

Mesa 23.0.4 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 23.0.4 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
---------------

::

    ee7f2cad1c58f943ea3048e59871908900b0a656ebdcf6a309e60af4270792f6  mesa-23.0.4.tar.xz


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

- None


Bug fixes
---------

- radv: Battlefield {1,5} hangs on RX 7900 XTX
- radv: graphical artifacts in MSFS running via DXVK on RX7900XT
- [radeonsi] flickering debug chunk border lines in Minecraft
- vulkan/device_select: no way to select between GPUs of the same model due to bugs
- radv: cache crashing
- nouveau: Regression in arb_transform_feedback_overflow_query-basic from multithreading patches
- [llvm 16+] [microsoft-clc] opencl-c-base.h does not exist
- Vulkancts clipping / tesselation tests trigger gpu hang on DG2
- Swaped fields in picture in vlc and mythtv if hw accel is on
- WGL: Assert assigns dwThreadId variable
- nine regression with r600 (bisected)
- [ACO] [RADV] Flickering squares in some areas in The Last of Us Part 1 (with workaround)
- radv: Jedi Fallen Order flickering & blocky plants
- nouveau: NV50 (NVAC) broken in latest master
- rusticl failed to build with rust-bindgen 0.65.0
- aco: s_load_dword with negative soffset cause GPU hang
- [Intel TGL] Performance drop in 22.3.5 compared to 22.2.5


Changes
-------

Axel Davy (1):

- frontend/nine: Fix num_textures count

Charmaine Lee (2):

- translate: do not clamp element index in generic_run
- svga: set PIPE_CAP_VERTEX_ATTRIB_ELEMENT_ALIGNED_ONLY for VGPU10 device

Connor Abbott (1):

- tu: Don't override depth for GMEM

David Heidelberg (1):

- panvk: clear dangling pointers

Dylan Baker (17):

- docs: Add sha256 sum for 23.0.3
- bin/pick: fix issue where None for nomination_type could fail
- bin/pick: use lineboxes to make the UI clearer
- bin/pick: Add support for adding notes on patches
- bin/pick: Add a Read/Write lock class
- pick: Fix a lot of locking issues
- pick: Add fixes/reverts of applied patches as they're applied
- bin/pick-ui: use asyncio.new_event_loop
- wip: 23.0 fixes for pick-ui reworks
- .pick_status.json: Update to 41a36561491d0cd65c8813126db9ce41e4aff9c4
- fixup! pick: Fix a lot of locking issues
- Revert "anv: flag BO for write combine when CPU visible and potentially in lmem"
- .pick_status.json: Update to 218aa7794136c6a9bc55d5c8c31e0ba450cf0f0b
- intel/dev: create a helper dependency for libintel_dev
- .pick_status.json: Update to 5039b595477589a6e39bd6e5a390ed565b89a793
- .pick_status.json: Updates notes for 9fb9ae5ac69293ab60d5d1f7470d8f152ccde7f6
- .pick_status.json: Updates notes for a77c88f1e897d5ab5c046b35898eadbe806e2990

Emma Anholt (3):

- blob: Don't valgrind assert for defined memory if we aren't writing.
- vulkan: Handle alignment failure in the pipeline cache.
- vulkan: Actually increment the count of objects in GetPipelineCacheData.

Eric Engestrom (4):

- amd: fix buggy usage of unreachable()
- compiler: fix buggy usage of unreachable()
- pvr: fix buggy usage of unreachable()
- vk/util: fix buggy usage of unreachable()

Erik Faye-Lund (2):

- nir: fix constant-folding of 64-bit fpow
- llvmpipe: fixup refactor copypasta

Filip Gawin (1):

- nine: add fallback for D3DFMT_D16 in d3d9_to_pipe_format_checked

Friedrich Vock (2):

- radv/rmv: Fix import memory
- radv: Always call si_emit_cache_flush before writing timestamps

Gert Wollny (1):

- r600/sfn: fix cube to array lowering for LOD

Hans-Kristian Arntzen (1):

- wsi/x11: Fix present ID signal when IDLE comes before COMPLETE.

Iago Toral Quiroga (3):

- broadcom/compiler: fix v3d_qpu_uses_sfu
- broadcom/compiler: add a v3d_qpu_instr_is_legacy_sfu helper
- broadcom/compiler: fix incorrect check for SFU op

Jonathan Gray (1):

- intel/dev: remove dg2 0x5698 pci id

José Fonseca (1):

- wgl: Fix unintentional assignment on assert.

Karol Herbst (4):

- radeonsi: lower mul_high
- ac/llvm: support shifts on 16 bit vec2
- rusticl: don't set size_t-is-usize for >=bindgen-0.65
- nvc0: do not randomly emit fences.

Konstantin Seurer (1):

- nir/lower_fp16_casts: Fix SSA dominance

Lionel Landwerlin (5):

- isl: don't set inconsistent fields for depth when using stencil only
- isl: fix a number of errors on storage format support on Gfx9/12.5
- intel/fs: fix per vertex input clamping
- intel/fs: fix scheduling of HALT instructions
- anv: fixup workaround 16011411144

Lone_Wolf (3):

- compiler/clc: Fix embedded clang headers (microsoft-clc)  for LLVM 16+
- clc: Add clangASTMatchers to fix static llvm build of  microsoft-clc with LLVM 16+
- clc: Add clang frontendhlsl module to fix build of  microsoft-clc with llvm 16+

Luna Nova (3):

- device_select_layer: fix inverted strcmp in device_select_find_dri_prime_tag_default (v1)
- device_select_layer: apply DRI_PRIME even if default device is > 1 to match opengl behavior
- device_select_layer: pick a default device before applying DRI_PRIME

M Henning (1):

- nouveau/codegen: Check nir_dest_num_components

Marek Olšák (2):

- nir: fix 2 bugs in nir_create_passthrough_tcs
- radeonsi: call ac_init_llvm_once before any util_queue initialization

Mark Janes (2):

- util: add macro to support gcc/clang poison
- intel/dev: generate helpers to identify platform workarounds

Matt Turner (1):

- intel: Disable shader cache when executing intel_clc during the build

Matthieu Bouron (1):

- lavapipe: honor dst base array layer when resolving color attachments

Michel Zou (1):

- mesa/draw: fix -Wformat warning

Mike Blumenkrantz (25):

- zink: add z32s8 as mandatory GL3.0 profile attachment format
- zink: avoid zero-sized memcmp for descriptor layouts
- nir/lower_alpha_test: rzalloc state slots
- zink: emit demote cap when using demote
- util/debug: move null checks out of debug message macro
- zink: don't bitcast bool deref loads/stores
- drisw: don't leak the winsys
- zink: check for extendedDynamicState3DepthClipNegativeOneToOne for ds3 support
- draw: fix viewmask iterating
- gallivm: break out native vector width calc for reuse
- draw: fix robust ubo size calc
- zink: only unset a generated tcs if the bound tcs is the generated one
- zink: set depth dynamic state values unconditionally
- cso: unbind fb state when unbinding the context
- zink: don't init mutable resource bit for swapchain images
- zink: allow vk 1.2 timelineSemaphore feature if extension isn't supported
- zink: stringify unsupported prim restart log error
- zink: use an intermediate variable for binding ssbo slots
- zink: unbind the ssbo slot being iterated, not the index of the buffer
- zink: flush INDIRECT_BUFFER mem barrier for compute
- glsl/lower_samplers_as_deref: apply bindings for unused samplers
- zink: disable always zs feedback loop on radv
- zink: destroy current batch state after all other batch states
- zink: only add feedback loop usage bit if extension is supported
- zink: add feedback loop usage for swapchains

Nanley Chery (2):

- iris/bufmgr: Add and use zero_bo
- iris/bufmgr: Handle flat_ccs for BO_ALLOC_ZEROED

Patrick Lerda (7):

- lima: fix refcnt imbalance related to framebuffer
- r600/sfn: fix memory leak related to sh_info->arrays
- aux/draw: fix memory leak related to ureg_get_tokens()
- crocus: fix refcnt imbalance related to crocus_create_surface()
- r600: fix refcnt imbalance related to atomic_buffer_state
- radeonsi: set proper drm_amdgpu_cs_chunk_fence alignment
- crocus: fix scratch_bos memory leak

Qiang Yu (2):

- aco: fix nir_f2u64 translation
- ac/nir/cull: fix line position w culling

Rhys Perry (4):

- aco: remove SMEM_instruction::prevent_overflow
- aco: don't move exec reads around exec writes
- aco: don't move exec writes around exec writes
- aco/gfx11: fix VMEM/DS->VALU WaW/RaW hazard

Ruijing Dong (1):

- radeonsi/vcn: correct cropping for hevc case

Samuel Pitoiset (1):

- radv: fix usage flag for 3D compressed 128 bpp images on GFX9

SoroushIMG (3):

- zink: do not emit line stipple dynamic state when emulating
- zink: take location_frac into account in lower_line_smooth_gs
- zink: fix incorrect line mode check for bresenham

Tapani Pälli (3):

- isl: disable mcs (and mcs+ccs) for color msaa on gfxver 125
- iris: implement state cache invalidate for Wa_16013063087
- isl: fix layout for comparing surf and view properties

Thong Thai (3):

- mesa/main: rework locale setup/teardown
- util: check and initialize locale before using it
- tgsi: use locale independent float and double parsing

Timur Kristóf (2):

- aco: Disallow constant propagation on SOPP and fixed operands.
- aco: Initialize vcmpx field in get_cmp_info.

Yiwei Zhang (3):

- radv: respect VK_QUERY_RESULT_WAIT_BIT in GetQueryPoolResults
- pipe-loader: avoid undefined memcpy behavior
- lvp: avoid accessing member of NULL ptr for global entries

antonino (1):

- nir: don't create invalid inputs in \`nir_create_passthrough_gs`