mesa: upgrade 21.1.3 -> 21.1.4
Bug fixes
* [anv] GravityMark (benchmark) crashes on ANV
* [opengl] We happy few not being rendered correctly
* Factorio: GPU hang when opening machine inventory
* RuneScape on Mesa 21.1.1 (VEGA10) has bad performance and leaks memory
* Supraland: flickering black bars on ground
* radv: GPU hang in Cyberpunk 2077 on Big Navi
* [amdgpu][renoir][rx5500m]: [drm:0xffffffff8198ad5e] *ERROR* ring gfx timeout, signaled seq=10952, emitted seq=10954
* [spirv-fuzz] SPIR-V parsing FAILED: Invalid back or cross-edge in the CFG
* panfrost Mount and Blade: Warband (glitches)
* anv: dEQP-VK.robustness.robustness2*no_fmt_qual.null_descriptor.samples* fail
Changes
* util/fossilize_db: Pull seek into lock.
* util/fossilize_db: Split out reading the index.
* util/fossilize_db: Do not lock the fossilize db permanently.
* spirv: Fix handling of OpBranchConditional with same THEN and ELSE
* nir/opt_if: Don’t split ALU for single block infinite loops
* llvmpipe: Add handle export for resource_get_param
* i965: fix regression in pipe control on g45
* pick_status.json: Update to
977355c6e5efd781dde85d834172dd23cd4852f1
* pick_status.json: Update to
40b67a292297606f0a7576e3ef4087028d5edd17
* zink: do not unmap dt-buffers twice
* zink: drop repeated usage-bit
* zink: do not check buffer-format for usage-bits
* zink: remove unused moltenvk functions
* libgl-gdi: add missing include
* aux/trace: fix bool argument
* ci/windows: work around meson encoding issues
* intel/fs: Teach IR about EOT instruction writing the accumulator implicitly on TGL+.
* intel/fs: Fix synchronization of accumulator-clearing W/A move on TGL+.
* pan/mdg: Add a bundle ID to instructions
* pan/mdg: Reorder some code in mir_spill_register
* pan/mdg: Fill from TLS before spilling non-SSA nodes
* pan/mdg: Fix reading a spilt register in the bundle it’s written
* pan/mdg: Add 16 bytes of padding to the end of shaders
* intel/nir: Fix txs for null surfaces
* spirv: Create acceleration structure and shader record variables
* nv50/ir: fix surface lowering when values get shared accross operations
* anv: Fix dynamic primitive topology for tess on Gfx7.x too
* anv: allocate bigger batches as we grow command buffers
* mesa: unreference zombie buffers when creating buffers to lower memory usage
* compiler/glsl: Use mutex lock while freeing up mem_ctx
* anv: fix dynamic primitive topology for tess
* zink: handle custom border color without matching wrap mode case
* radeonsi: skip instance_count==0 draws on <= GFX9
* radeonsi: disable ngg culling on llvm < 12
* radv: reject binding buffer/image when the device memory is too small
* radv: always decompress both aspects of a depth/stencil image
* amd/addrlib: remove Meson debug message()
* radeon/vcn/enc: Add missing line to HEVC SPS header code
* util: add work around for the game We Happy Few
* ac/nir: Update TCS output barriers with nir_var_mem_shared.
* radv/llvm: Emit s_barrier at the beginning of NGG non-GS shaders.
* aco/gfx10: NGG zero output workaround for conservative rasterization.
* aco/gfx10: Emit barrier at the start of NGG VS and TES.
Signed-off-by: Andreas Müller <schnitzeltony@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
