1
Mesa 23.0.2 Release Notes / 2023-04-07
2
======================================
4
Mesa 23.0.2 is a bug fix release which fixes bugs found since the 23.0.1 release.
6
Mesa 23.0.2 implements the OpenGL 4.6 API, but the version reported by
7
glGetString(GL_VERSION) or glGetIntegerv(GL_MAJOR_VERSION) /
8
glGetIntegerv(GL_MINOR_VERSION) depends on the particular driver being used.
9
Some drivers don't support all the features required in OpenGL 4.6. OpenGL
10
4.6 is **only** available if requested at context creation.
11
Compatibility contexts may report a lower version depending on each driver.
13
Mesa 23.0.2 implements the Vulkan 1.3 API, but the version reported by
14
the apiVersion property of the VkPhysicalDeviceProperties struct
15
depends on the particular driver being used.
22
1b7d3399fc6f16f030361f925d33ebc7600cbf98094582f54775b6a1180529e7 mesa-23.0.2.tar.xz
35
- allwinner a64: DRM_IOCTL_MODE_CREATE_DUMB failed: Cannot allocate memory after some time of apps usage
36
- mesa: index buffer leaking
37
- RadeonSI: null dereference in amdgpu_cs_add_buffer, potential refcount mismatch, running BeyondAllReason
38
- eglCreateImageKHR, error: EGL_BAD_ALLOC (0x3003), message: "createImageFromDmaBufs failed" on AMD multi-gpu with explicit format modifiers
39
- libgrl.a installed but not used?
40
- radv: crash compiling UE5 lumen hardware RT shader
41
- aco: unused vtmp_in_loop
42
- radv,nir: dEQP-VK.ray_query.builtin.rayqueryterminate.* failures
43
- glsl compiled error when the RHS of operator \`>>` is int64_t by enabling GL_ARB_gpu_shader_int64 extension
44
- QPainter fails to render multiple shapes with a brush set since Mesa 23.0
45
- eglSwapBuffers blocks in wayland when it's wl_surface_frame event is stolen.
46
- plasmashell sometimes hangs with mesa_glthread
47
- pps_device.h:23:11: error: ‘uint32_t’ does not name a type
53
Alyssa Rosenzweig (1):
55
- panfrost: Don't round up Midgard polygon list BOs
59
- radv/rt: place any-hit scratch vars after intersection scratch vars
60
- radv/rt: Fix any_hit scratch variables.
64
- aco: drop leftover variable
68
- docs: Add sha256 sum for 23.0.1
69
- .pick_status.json: Update to 5e9ea154849ebe19b96a243b9716bd264545bf8b
70
- .pick_status.json: Update to 2cc9364c2061fe4a66398678cabf61e3464e4d77
71
- .pick_status.json: Update to 7abd8c499b4388f1e2947d78c59aafa0b52c254a
72
- .pick_status.json: Update to debc543904235ed7dfc14217bf43f33f9f768e5a
76
- turnip: Don't push inline uniform buffer contents outside constlen.
80
- v3d: fix \`dirty` bitset being too small to accept V3D_DIRTY_SSBO
84
- lima: don't use resource_from_handle while creating scanout
88
- crocus: don't quantize the clear value
92
- intel/fs: Fix register coalesce in presence of force_writemask_all copy source writes.
96
- r600/sfn: fix container allocators
98
Iago Toral Quiroga (1):
100
- vulkan/wsi/display: set pDisplay to NULL on error
104
- intel/vk/grl: don't install libgrl.a
108
- nir/algebraic: insert patterns inside optimizations list
109
- nir/algebraic: remove duplicate bool conversion lowerings
113
- vulkan/wsi/display: do not dereference a NULL pointer
117
- nir: Propagate alignment when rematerializing cast derefs
118
- microsoft/compiler: Fix setting bit 31 in feature flags
122
- iris: Retry DRM_IOCTL_I915_GEM_EXECBUFFER2 on ENOMEM
123
- Revert "iris: Avoid abort() if kernel can't allocate memory"
127
- nvc0: enable fp helper invocation memory loads on Turing+
128
- rusticl/util: extract offset_of macro
129
- rusticl/icd: Make it work in case Rustc shuffles struct around
131
Konstantin Seurer (3):
133
- radv/sqtt: Skip dumping pipeline libraries
134
- radv/rt: Handle load_constant instructions when inlining shaders
135
- nir/lower_shader_calls: Remat derefs before lowering resumes
139
- amd/common: fix a typo
141
Lionel Landwerlin (12):
143
- intel/fs: don't SEND messages as partial writes
144
- intel/fs: fix nir_opt_peephole_ffma max vec assumption
145
- intel/fs: fixup sources number from opt_algebraic
146
- intel/fs: fix subgroup invocation read bounds checking
147
- anv: fix incorrect utrace bo release
148
- anv: implement recommended flush/wait of AUX-TT invalidation
149
- iris: implement recommended flush/wait of AUX-TT invalidation
150
- anv: take care of maxStorageBufferRange being uint32_t
151
- isl: update max buffer size for SKL+
152
- intel/perf: fix OA format selection on MTL
153
- intel/fs: don't consider fixup_nomask_control_flow SENDs predicate
154
- intel/fs: UNDEF fixup_nomask_control_flow temp register
158
- aco_validate: allow for wave32 in p_dual_src_export_gfx11
162
- anv,hasvk: remove stale comments
166
- vbo: fix current attribs not updating gallium vertex elements
168
Mike Blumenkrantz (9):
170
- aux/trace: delete GALLIUM_TRACE_NIR log message
171
- zink: track tc fences better
172
- zink: double check layouts for possible feedback loop images
173
- zink: handle swapchain creation failure less lazily
174
- aux/trace: fix GALLIUM_TRACE_NIR handling
175
- llvmpipe: fix linear fs analysis with nonzero fs outputs
176
- llvmpipe: fix handling of unused color attachments
177
- zink: block resolves where src extents > dst extents
178
- zink: flag vertex buffers for rebind after vstate draws
182
- r600: fix typo that could lead to a possible crash
184
Pierre-Eric Pelloux-Prayer (5):
186
- yegl/wayland: fix glthread deadlocks
187
- Revert "driconf: add a workaround for plasmashell freezing"
188
- radeonsi: don't use alignment_log2 of imported buffers
189
- mesa: fix invalid index_bo refcounting
190
- util/vbuf: fix index_bo leak
194
- ac/llvm: respect channel_type when ac_build_buffer_load
195
- ac/llvm: add missing type convert for nir_load_buffer_amd
199
- aco: fix nir_var_shader_out barriers for task shaders
203
- gallium/pipe: change PIPE_DEFAULT_DECODER_FEEDBACK_TIMEOUT_NS to 1 second
205
Sviatoslav Peleshko (1):
207
- glsl: Fix codegen for constant ir_binop_{l,r}shift with mixed types
211
- anv: use primitive ID override when shader does not supply it
215
- glsl: allow 64-bit integer on RHS of shift
216
- util/00-mesa-defaults: add Akka Arrh workaround
220
- ac/nir: Store only lowest 8 bits for task draw ring DWORD3.
221
- aco: Consider p_cbranch_nz as divergent branch too.
222
- aco: Don't remove exec writes that also write other registers.
226
- pps: Fix build errors.
230
- zink: fix \`final_hash` update in \`zink_gfx_program_update`
231
- mesa: correctly allocate space for converted primtives
232
- nir: handle primitives with adjacency
233
- nir: avoid generating conflicting output variables