Marek Olšák [Mon, 7 Jun 2021 13:26:55 +0000 (09:26 -0400)]
mesa: use atomics instead of mutexes for refcounting texture objects
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11339>
Marek Olšák [Mon, 7 Jun 2021 13:58:48 +0000 (09:58 -0400)]
cso: don't look up a sampler CSO if the last one is identical
This is benefical when sampler states are identical often, and detrimental
if they are not. The average case seems to be in favor of this.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Mon, 7 Jun 2021 13:58:05 +0000 (09:58 -0400)]
cso: update max_sampler_seen only once in cso_set_samplers
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 20:27:36 +0000 (16:27 -0400)]
cso: disallow NULL sampler state templates in cso_single_sampler
No code passes NULL into this except the single place that is updated.
Let's remove the if. The next commit depends on this.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Mon, 7 Jun 2021 10:55:12 +0000 (06:55 -0400)]
st/mesa: remove the sampler min_lod/max_lod value swap
It might not be needed at all.
v2: fix v3d assertion failure
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 10:08:07 +0000 (06:08 -0400)]
st/mesa: add a mechanism to bypass atomics when binding sampler views
This is the same mechanism we already use for buffers. The code is mostly
copied from there. See the big comment for explanation.
This will be very effective when take_ownership is added into pipe_context
::set_sampler_views because that and this commit together will eliminate
atomics for sample views almost entirely.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 09:23:30 +0000 (05:23 -0400)]
st/mesa: sink refcounting from st_get_sampler_views into st_sampler_view.c
Do it deeper in the sampler view return path. It minimizes the deltas
for the next commit.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 09:35:31 +0000 (05:35 -0400)]
st/mesa: remove the const qualifier for a few st_sampler_view instances
It will be needed by following commits.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 08:48:47 +0000 (04:48 -0400)]
st/mesa: return sview from st_update_single_texture via return value, not param
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 08:42:17 +0000 (04:42 -0400)]
st/mesa: read Target only once in st_update_single_texture
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 08:42:03 +0000 (04:42 -0400)]
st/mesa: sink _mesa_get_samplerobj into st_update_single_texture
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 02:35:09 +0000 (22:35 -0400)]
st/mesa: don't memset the sampler view array, don't init trailing slots to NULL
No functional change.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 02:10:12 +0000 (22:10 -0400)]
st/mesa: don't track FS sampler views for bitmap/drawpix in st_context
Just query the sampler views from the update function. This will help
optimize sampler view updates.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Marek Olšák [Sun, 6 Jun 2021 02:10:12 +0000 (22:10 -0400)]
st/mesa: don't track VS sampler views for st_draw_feedback in st_context
Just query the sampler views from the update function. This will help
optimize sampler view updates.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11428>
Rob Clark [Fri, 25 Jun 2021 19:45:03 +0000 (12:45 -0700)]
turnip: Add CrOS Gralloc support
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11612>
Rob Clark [Fri, 25 Jun 2021 18:24:59 +0000 (11:24 -0700)]
turnip: Fix AcquireImageANDROID() handle type
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11612>
Rob Clark [Thu, 24 Jun 2021 18:11:08 +0000 (11:11 -0700)]
turnip: Use drmIoctl()
Replace open-coded ioctl with drmIoctl() to get restart on interrupted
system calls.
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11612>
Lionel Landwerlin [Sat, 19 Jun 2021 10:14:16 +0000 (13:14 +0300)]
anv: allocate bigger batches as we grow command buffers
This is the first time we see an application running out of mmap().
We essentially allocate too many batches (+65k) and end up not being
able to mmap them, at which point we can't mmap anything anymore and
things go sideways.
This change allocates bigger batch BOs as we grow an existing command
buffer. This drastically reduces the number of BOs we need to allocate
(the benchmark that reported the issue now reaches a max of ~630 BOs,
instead of reaching 65k and failing previously).
v2: Track the total batch size of command buffers (Jason)
Just give 0 for batch_len to i915 (Jason)
v3: Fix indentation (Jason)
v4: Drop uncessary reshuffling of error labels (Jason)
v5: Remove empty lines (Marcin)
v6: Limit BO growing to chunks of 16Mb (Jason)
v7: Add assert on initial size (Jason)
v8: Add define for max size (Jason)
v9: Fixup v7 assert for non softpin platforms (Lionel)
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4956
Cc: mesa-stable
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11482>
Caio Marcelo de Oliveira Filho [Wed, 23 Jun 2021 22:58:40 +0000 (15:58 -0700)]
anv: Advertise VK_KHR_shader_subgroup_uniform_control_flow
This is always valid for Intel, so we don't need to do anything
different if the Execution Mode is being used.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11476>
Caio Marcelo de Oliveira Filho [Wed, 23 Jun 2021 22:57:31 +0000 (15:57 -0700)]
spirv: Support SPV_KHR_subgroup_uniform_control_flow
There's no SPIR-V Capability associated, so check in the Execution
Mode. For now, don't keep track of whether a shader uses uniform
control flow in the shader_info, we can add that when/if a driver
actually need that information.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11476>
Caio Marcelo de Oliveira Filho [Wed, 9 Jun 2021 22:13:06 +0000 (15:13 -0700)]
spirv: Update headers and metadata from latest Khronos commit
This corresponds to
f95c3b3761ee1b1903f54ae69b526ed6f0edc3b9
("Merge pull request #219 from cmarcelo/SPV_EXT_shader_atomic_float16_add")
in https://github.com/KhronosGroup/SPIRV-Headers.
Acked-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11476>
Caio Marcelo de Oliveira Filho [Wed, 23 Jun 2021 22:28:40 +0000 (15:28 -0700)]
nir: Add test to check edge case in Split ALU optimization
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11476>
Caio Marcelo de Oliveira Filho [Fri, 18 Jun 2021 19:24:11 +0000 (12:24 -0700)]
nir/opt_if: Don't split ALU for single block infinite loops
Some infinite loop cases were already covered by other
restrictions (e.g. if the loop had a body), but the case with a single
block in the loop body wasn't yet.
This prevents an infinite loop when optimizing the shader in
dEQP-VK.reconvergence.subgroup_uniform_control_flow_ballot.compute.nesting2.3.2
and various others reconvergence tests.
Fixes:
0881e90c099 ("nir: Split ALU instructions in loops that read phis")
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> [v1]
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11476>
Miguel Gomez [Thu, 3 Jun 2021 14:44:30 +0000 (16:44 +0200)]
i965: Prevent invalid framebuffer usage
When a surfaceless context is in use, driDrawablePriv might be NULL, so needs to
be checked before calling dri2InvalidateDrawable. Same for read calls.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Eleni Maria Stea <estea@igalia.com>
See https://gitlab.freedesktop.org/mesa/mesa/-/issues/778
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11459>
Philippe Normand [Sun, 11 Apr 2021 12:27:56 +0000 (13:27 +0100)]
i915: Prevent invalid framebuffer usage
When a surfaceless context is in use, driDrawablePriv might be NULL, so needs to
be checked before calling dri2InvalidateDrawable. Same for read calls.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Eleni Maria Stea <estea@igalia.com>
Closes https://gitlab.freedesktop.org/mesa/mesa/-/issues/778
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11459>
Matt Turner [Wed, 23 Jun 2021 20:19:14 +0000 (13:19 -0700)]
amd/ci: Use MESA_VK_IGNORE_CONFORMANCE_WARNING to reduce warnings
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11563>
Matt Turner [Wed, 23 Jun 2021 20:22:03 +0000 (13:22 -0700)]
ci: Unify on MESA_VK_IGNORE_CONFORMANCE_WARNING
Move and rename warn_non_conformant_implementation() to common location
of src/vulkan/util/vk_util.c as vk_warn_non_conformant_implementation().
In freedreno/ci, move MESA_VK_IGNORE_CONFORMANCE_WARNING to common
location of .baremetal-deqp-test-freedreno-vk.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11563>
Mike Blumenkrantz [Tue, 22 Jun 2021 16:53:15 +0000 (12:53 -0400)]
anv: VK_EXT_multi_draw implementation
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11531>
Mike Blumenkrantz [Tue, 22 Jun 2021 18:42:48 +0000 (14:42 -0400)]
anv: unify some draw state vertex constant emission
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11531>
Mike Blumenkrantz [Fri, 7 May 2021 13:41:45 +0000 (09:41 -0400)]
zink: repack zink_context struct a bit
Reviewed-by: Joshua Ashton <joshua@froggi.es>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11593>
Connor Abbott [Thu, 24 Jun 2021 16:30:24 +0000 (18:30 +0200)]
freedreno, tu: Set SP_XS_PVT_MEM_HW_STACK_OFFSET
Theoretically this register should only be used when function calls in
the shader are used, which we don't support. But with the default value
of 0 it seems like pvtmem doesn't work on a650. Just set it to the total
per-SP size, effectively leaving no space for the return-address stack,
like the blob does.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4949
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11581>
Connor Abbott [Fri, 25 Jun 2021 13:06:12 +0000 (15:06 +0200)]
freedreno/a6xx: Make SP_XS_PVT_MEM_HW_STACK_OFFSET non-inline
Otherwise we can't use the helper to pack it as it collides with the
function in a6xx-pack.xml.h.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11581>
Danylo Piliaiev [Wed, 16 Jun 2021 11:43:19 +0000 (14:43 +0300)]
ir3: add ldg.a,stg.a which allow complex in-place offset calculation
The full form for ldg.a/stg.a offset is:
g[reg_address + reg_offset << (imm_shift + 2) + imm_offset << 2]
where imm_shift is in [0, 3] and imm_offset is in [0, 3]
a6xx blob was found to produce a bit simplier offset calculations
for TES/TCS shaders in GTA V:
[c002000a_03c14215] ldg.a.f32 r2.z, g[r1.y+((r2.z+1)<<2)], 3;
[c0020004_01c14609] ldg.a.f32 r1.x, g[r1.y+((r1.x+3)<<2)], 1;
Our new syntax:
stg.a.u32 g[r2.x+(r1.x+1)<<2], r5.x, 1
stg.a.u32 g[r2.x+r1.x<<4+3<<2], r5.x, 1
ldg.a.f32 r1.w, g[r1.y+(r1.w+1)<<2], 3
ldg.a.f32 r1.w, g[r1.y+r1.w<<5+2<<2], 3
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11431>
Danylo Piliaiev [Fri, 18 Jun 2021 15:09:01 +0000 (18:09 +0300)]
freedreno/isa: add uoffset type to print positive-only offsets
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11431>
Danylo Piliaiev [Wed, 16 Jun 2021 18:31:03 +0000 (21:31 +0300)]
freedreno/computerator: pass iova of buffer to const register
The syntax is:
@buf 32 (c2.x)
The "(c2.x)" is optional.
This makes possible to test stg, ldg, and global atomics.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11431>
Samuel Iglesias Gonsálvez [Fri, 25 Jun 2021 08:14:21 +0000 (10:14 +0200)]
anv: do not dereference VkPipelineMultisampleStateCreateInfo always
VkGraphicsPipelineCreateInfo.pMultisampleState is a pointer to a
VkPipelineMultisampleStateCreateInfo structure, and is ignored if the
pipeline has rasterization disabled.
Fixes a crash in one CTS tests that checks this.
Signed-off-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11601>
Danylo Piliaiev [Wed, 23 Jun 2021 10:55:06 +0000 (13:55 +0300)]
turnip: early exit in tu6_draw_common to save cpu cycles
Improves Zink + drawoverhead perf up to 4%
Before:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 3981
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 3977
After:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 4136
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 4163
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11556>
Danylo Piliaiev [Tue, 22 Jun 2021 14:24:43 +0000 (17:24 +0300)]
turnip: do not re-emit same vs params
Improves drawoverhead perf through Zink up to 260%
Before:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 1518
After:
1, DrawElements ( 1 VBO| 0 UBO| 0 ) w/ no state change, 3981
This brings it close to Freedreno, which has around 4300.
In vkQuake vs params re-emission now occurs in 0.23% of draw calls.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11556>
Jason Ekstrand [Tue, 20 Oct 2020 21:36:00 +0000 (16:36 -0500)]
intel/genxml: Remove old scratch fields on GFX version 12.5
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 21:11:45 +0000 (16:11 -0500)]
anv: Add support for scratch on XeHP
Rework:
* Jordan: Handle per_thread_scratch==0 in anv_scratch_pool_get_surf
* Jordan: Update subslices in anv_scratch_pool_alloc
* Jason: Clean up the patch a bit
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 19:27:58 +0000 (14:27 -0500)]
iris: Add support for scratch on XeHP
Rework:
* Jordan: Handle prog_data->total_scratch==0 in iris_upload_compute_walker
* Jordan: Resolve iris_get_scratch_space conflict with
e2c5ef6cd65
* Jordan: Rebase on
4256f7ed584. broken
* Ken: Mostly fixed the rebase
* Jordan: Fix two small compilation issues
* Jordan: Rebase on Ken's ("iris: Make a pin_scratch_space() helper")
* Lionel: Fix a few bugs with scratch handles
* Jason: Tidy the patch up a bit
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Wed, 21 Oct 2020 02:29:50 +0000 (21:29 -0500)]
iris: Add a MEMZONE_BINDLESS and uploader
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 21:34:25 +0000 (16:34 -0500)]
intel/genxml: Add new ScratchSpaceBuffer fields on GFX version 12.5
Rework:
* Jordan: Fix MEMZONE_BINDER_START detection
* Jordan: Bump the IRIS_BINDLESS_SIZE to 8M
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 22:47:17 +0000 (17:47 -0500)]
intel/fs: Implement load/store_scratch on XeHP
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 22:42:21 +0000 (17:42 -0500)]
intel/fs: Implement spilling on XeHP
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 18:23:31 +0000 (13:23 -0500)]
intel/isl: Add support for scratch buffers
XeHP adds support for a new surface type for scratch. It's similar to
SURFTYPE_STRBUF in that it's a 2D array-of-struct format but the one
key difference is that the U coordinate is computed automatically based
on the thread ID and only the V coordinate is provided in the dataport
message.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Jason Ekstrand [Tue, 20 Oct 2020 18:09:23 +0000 (13:09 -0500)]
intel/genxml: Add SURFTYPE_SCRATCH on GFX version 12.5
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11582>
Emma Anholt [Thu, 24 Jun 2021 20:37:52 +0000 (13:37 -0700)]
Revert "freedreno: Cooperate with tc to stop checking the BC for resource_busy()."
This reverts commit
5cb043cf82af4af88dbfe83ce1eff4cc41e2a766.
While the numbers were impressive for drawoverhead, it comes at the cost
of additional flushes, which for gmem access (what we actually care about
most!) would greatly increase the actual cost to render. Also, gl_driver2
overhead is increased, probably due to spending time in the kernel for the
flushes.
drawoverhead's win came from the increased flushing causing the GPU to
start processing the buffers sooner on everything but test 1, which
already had some incremental flushing happening. That was certainly not
intended by the change.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11599>
Mike Blumenkrantz [Sat, 24 Apr 2021 17:16:20 +0000 (13:16 -0400)]
zink: flush pending clears if a resource is bound as a descriptor
ensure that resource data is in sync with the expected command stream
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Fri, 16 Apr 2021 14:27:54 +0000 (10:27 -0400)]
zink: only queue deferred descriptor layout change on first bind or change
otherwise, the layout is either already correct or already queued
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Fri, 16 Apr 2021 14:23:16 +0000 (10:23 -0400)]
zink: improve automatic layout transitions for sampler+image descriptors
binding new images may require auto-transitions for the other pipeline, and
unbinding an image requires that any persistent sampler binds have their layouts
updated
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Fri, 16 Apr 2021 14:21:39 +0000 (10:21 -0400)]
zink: use bind counts to more accurately determine image descriptor's exact layout
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Fri, 9 Apr 2021 14:30:49 +0000 (10:30 -0400)]
zink: only do deferred image barriers if layout changes
otherwise these should be handled by apps using glMemoryBarrier
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Mon, 10 May 2021 20:51:49 +0000 (16:51 -0400)]
zink: split deferring of barriers to image and buffer functions
make this more explicit
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Mike Blumenkrantz [Wed, 7 Apr 2021 14:59:41 +0000 (10:59 -0400)]
zink: break out image descriptor layout into util function
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11590>
Erik Faye-Lund [Thu, 24 Jun 2021 20:45:20 +0000 (22:45 +0200)]
ci/windows: fix zink msvc build-rules
When I added the build-rules for zink to the windows build, I
accidentally added it to .windows-test-rules instead of
.windows-build-rules. This seems to trigger a build-error if we trigger
*just* a test due to a zink-change, but not a build.
Hopefully this fixes the problems Mike has had with Zink CI recently.
Fixes:
a426d7c2643 ("ci/windows: enable msvc builds of zink")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11597>
Enrico Galli [Tue, 25 May 2021 21:41:33 +0000 (14:41 -0700)]
microsoft/spirv_to_dxil: Add drive_location assignment
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Tue, 8 Jun 2021 01:07:46 +0000 (18:07 -0700)]
microsoft/compiler: Switch io sort to use nir_sort_variables_with_modes
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Thu, 10 Jun 2021 20:46:15 +0000 (13:46 -0700)]
nir: Add modes filter to nir_sort_variables
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Jason Ekstrand [Fri, 6 Nov 2020 23:19:53 +0000 (15:19 -0800)]
nir: Add a function for sorting variables
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Mon, 7 Jun 2021 22:52:29 +0000 (15:52 -0700)]
util: Add qsort_r/s args adapter for MSVC and BSD/macOS
MSVC's qsort_s behaves similarly to sort_r. Unfortunately, qsort_s's
compare function has the "context"/"args" as its first argument. BSD's
qsort_r has a different order than GNU's qsort_r. Finally, C11 added
qsort_s's which look like GNU's gsort_r.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Tue, 15 Jun 2021 04:02:24 +0000 (21:02 -0700)]
util: Add simple test for util_qsort_r
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Jason Ekstrand [Tue, 10 Nov 2020 17:00:54 +0000 (11:00 -0600)]
util: Add an implementation of qsort_r for non-GNU platforms
Android and MSVC don't have qsort_r() so let's provide a util wrapper
that uses the old qsort and thread-local storage. We use C++ for this
because thread_local is built into C++11 and we can't rely on C11
everywhere.
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Tue, 25 May 2021 21:40:09 +0000 (14:40 -0700)]
d3d12, microsoft/compiler: Moving driver_location allocation to compiler
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Tue, 11 May 2021 01:00:57 +0000 (18:00 -0700)]
d3d12, microsoft/compiler: Switching semantic names to TEXCOORD
Switching from GENERIC/VARYING to TEXCOORD. This change better aligns
with other dxil generators (SPIRV-Cross, Tint).
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Enrico Galli [Thu, 24 Jun 2021 02:25:14 +0000 (19:25 -0700)]
microsoft/compiler: Remove de-duplication of arbitrary semantic names
Since DXC doesn't perform de-duplication for arbitrary semantic names,
and the DXIL validator checks against this behavior. We need to remove
the de-duplication.
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10989>
Niklas Haas [Mon, 26 Apr 2021 11:40:38 +0000 (13:40 +0200)]
vulkan/wsi/wayland: implement the full format table
This code was written but (rightfully) commented out, because it
referenced non-existing vulkan formats.
The correct vulkan analogs come in the form _PACK16 and _PACK32,
respectively, which have their component ordering in high->low bit
order. This is compatible with the DRM format ordering on little endian
systems, so we can just implement them all in the obvious way. On big
endian systems, the _PACKN formats have no analogs, because the
corresponding DRM formats are specified as always being little endian.
I implemented all of the formats I could, while excluding those that
don't map to vulkan at all (e.g. BGRA1010102), and commenting out the
ones that require an extra extension (e.g. ARGB4444) because they
probably need to be checked for extension presence first.
List slightly re-ordered to make it more neatly organized. Tested
working with the 2101010 and 565 variants. Was not able to test the 4444
or 1555 formats, but I verified the channel order was correct so they
should be working in theory.
Signed-off-by: Niklas Haas <git@haasn.xyz>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9737>
Alexey Nurmukhametov [Wed, 23 Jun 2021 16:38:06 +0000 (19:38 +0300)]
tu/kgsl: Fix file descriptor double close
tu_kgsl.c: tu_enumerate_devices closed fd previously closed by
tu_physical_device_init function.
Move out the fd closing from tu_physical_device_init function because
they do not belong to it.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11561>
Erik Faye-Lund [Wed, 23 Jun 2021 04:40:21 +0000 (06:40 +0200)]
ci/windows: enable msvc builds of zink
This depends on both the Vulkan SDK and the Vulkan Runtime, so let's
install those first.
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11424>
Erik Faye-Lund [Tue, 22 Jun 2021 10:42:50 +0000 (12:42 +0200)]
ci/windows: work around meson encoding issues
It turns out, some combinations of Meson and MSVC doesn't handle UTF-8
symbols in source-files as gracefully as we'd like. Luckily, there's a
work-around; forcing UTF-8 as the default-encoding.
Please see this Meson ticket for details:
https://github.com/mesonbuild/meson/issues/8263
Cc: mesa-stable
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11424>
Erik Faye-Lund [Thu, 24 Jun 2021 11:30:20 +0000 (13:30 +0200)]
zink: cast pointers to uintptr_t
The way p_atomic_cmpxchg is implemented on Windows really wants us to
use integers instead of things like pointers. So let's use uintptr_t
here. It seems cleaner to use that than the value zero for the last
argument, because it shows the intent of the code more clearly.
Fixes:
d80d9e1c935 ("zink: queue v3.0")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11424>
Erik Faye-Lund [Tue, 22 Jun 2021 14:01:49 +0000 (16:01 +0200)]
aux/trace: fix bool argument
The second argument to debug_get_bool_option is a boolean, not a
pointer. Passing a NULL-pointer there triggers a warning, which we treat
as an error on MSVC:
---8<---
../src/gallium/auxiliary/driver_trace/tr_screen.c(1071): error C2220:
the following warning is treated as an error
../src/gallium/auxiliary/driver_trace/tr_screen.c(1071): warning C4047:
'function': 'bool' differs in levels of indirection from 'void *'
../src/gallium/auxiliary/driver_trace/tr_screen.c(1071): error C4024:
'debug_get_bool_option': different types for formal and actual parameter 2
---8<---
Fixes:
c1270d4845d ("aux/trace: add screen deduplication for zink+lavapipe tracing")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11424>
Emma Anholt [Wed, 23 Jun 2021 21:37:55 +0000 (14:37 -0700)]
turnip: Disable buffer texturing on 422 formats.
Fixes:
dEQP-VK.api.info.format_properties.g8b8g8r8_422_unorm
dEQP-VK.api.info.format_properties.b8g8r8g8_422_unorm
and part of:
dEQP-VK.api.info.format_properties.g8_b8_r8_3plane_420_unorm
dEQP-VK.api.info.format_properties.g8_b8r8_2plane_420_unorm
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11562>
Emma Anholt [Wed, 23 Jun 2021 18:06:49 +0000 (11:06 -0700)]
ci/turnip: Document create_instance_device_intentional_alloc_fail's fail.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11562>
Emma Anholt [Mon, 21 Jun 2021 22:37:36 +0000 (15:37 -0700)]
turnip: Use vk_startup_errorf() in more startup paths.
This does the logging for you.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11562>
Emma Anholt [Mon, 21 Jun 2021 22:37:15 +0000 (15:37 -0700)]
turnip: Link more MRs and issues related to our xfails.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11562>
Emma Anholt [Wed, 23 Jun 2021 19:15:02 +0000 (12:15 -0700)]
freedreno/ir3: Report RA failure with mesa_loge().
This is a major failure that should never happen (if we had spilling
support), don't hide the log behind DEBUG builds.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11562>
Erik Faye-Lund [Thu, 24 Jun 2021 15:56:08 +0000 (17:56 +0200)]
iris/ci: disable amly jobs
These runners are having stability issues, and is causing a lot of
merge-requests to fail without reason. Let's disable them until the
runners have gotten some stability fixes in place.
See this discussion for details:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/3437#note_971266
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11580>
Daniel Schürmann [Wed, 23 Jun 2021 14:29:21 +0000 (16:29 +0200)]
aco/ra: handle copies of copies better
Instead of adding a second copy, just redirect
the existing copy.
No fossil-db changes.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11571>
Daniel Schürmann [Wed, 23 Jun 2021 13:53:43 +0000 (15:53 +0200)]
aco/ra: handle copies of definition registers
Previously, it could happen that a parallelcopy of
a definition was inserted before the instruction.
Fixes Rage 2 with GFX7.
No fossil-db changes.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11571>
Jason Ekstrand [Wed, 23 Jun 2021 20:15:03 +0000 (15:15 -0500)]
anv: Claim to be a discrete GPU if has_lmem
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Jordan Justen [Thu, 3 Oct 2019 00:39:27 +0000 (17:39 -0700)]
intel/dev: Set has_local_mem for DG1
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Sagar Ghuge [Fri, 17 Apr 2020 01:05:23 +0000 (18:05 -0700)]
anv: Allocate scratch and workaround BO in local memory
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Sagar Ghuge [Fri, 3 Apr 2020 03:16:08 +0000 (20:16 -0700)]
anv: Allocate BO in appropriate region
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Sagar Ghuge [Wed, 1 Apr 2020 01:26:20 +0000 (18:26 -0700)]
anv: Wrapper around I915_GEM_CREATE_EXT_MEMORY_REGIONS
v2 (Jordan Justin):
- add anv_gem_stubs.c impl
v3 (Jason Ekstrand):
- Use the upstream uAPI
- Rework the interface a bit
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Sagar Ghuge [Wed, 1 Apr 2020 01:23:25 +0000 (18:23 -0700)]
anv: Query memory region info
Create additional memory type with DEVICE_LOCAL_BIT if we have local
memory region aviable.
v2 (Jason Ekstrand):
- Don't leak mem_regions if the second ioctl fails
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Rafael Antognolli [Fri, 3 Jan 2020 18:42:10 +0000 (10:42 -0800)]
iris: Map with WC on non-LLC platforms.
Reworks:
* Jordan: Required rework following
f62724ccacf ("iris: Pick a single mmap mode (WB/WC) at BO allocation time")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Rafael Antognolli [Mon, 29 Jul 2019 17:59:01 +0000 (10:59 -0700)]
iris/bufmgr: Add flag to allocate from local memory.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Rafael Antognolli [Mon, 29 Jul 2019 16:54:17 +0000 (09:54 -0700)]
iris/bufmgr: Add new set of buckets for local memory.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Rafael Antognolli [Thu, 6 Jun 2019 17:37:02 +0000 (10:37 -0700)]
iris/bufmgr: Query memory region info.
v2 (Jason Ekstrand):
- Don't leak meminfo if the ioctl fails on the second pass
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Jordan Justen [Wed, 31 Jul 2019 22:18:59 +0000 (15:18 -0700)]
iris/bufmgr: Align vma addresses to 64K for local memory
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Jordan Justen [Thu, 3 Oct 2019 00:37:33 +0000 (17:37 -0700)]
intel/devinfo: Add has_local_mem
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Jason Ekstrand [Wed, 23 Jun 2021 19:21:16 +0000 (14:21 -0500)]
include/drm-uapi: bump headers
From drm-next at the following commit:
commit
2a7005c8a3982ba27fab237d85c27da446484e9c (HEAD)
Merge:
0666cba1f5b2b 47c65b3853f88
Author: Dave Airlie <airlied@redhat.com>
Date: Fri Jun 11 13:34:42 2021 +1000
Merge tag 'drm-intel-gt-next-2021-06-10' of...
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5599>
Simon Zeni [Mon, 21 Jun 2021 15:56:35 +0000 (11:56 -0400)]
radv: Implement VK_EXT_acquire_drm_display
Signed-off-by: Simon Zeni <simon@bl4ckb0ne.ca>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11014>
Simon Zeni [Mon, 21 Jun 2021 15:55:14 +0000 (11:55 -0400)]
vulkan/wsi: Implement VK_EXT_acquire_drm_display
Implements the two functions defined in the extension
VK_EXT_acquire_drm_display, vkAcquireDrmDisplayEXT and vkGetDrmDisplayEXT.
Signed-off-by: Simon Zeni <simon@bl4ckb0ne.ca>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Simon Ser <contact@emersion.fr>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11014>
Simon Zeni [Tue, 15 Jun 2021 15:05:05 +0000 (11:05 -0400)]
vulkan/wsi: add drm_fd param to wsi_display_get_connector
Modifies the signature on `wsi_display_get_connector` to retrieve the
connector of an arbitrary DRM FD instead of the one taken from the
wsi_display.
Signed-off-by: Simon Zeni <simon@bl4ckb0ne.ca>
Reviewed-by: Simon Ser <contact@emersion.fr>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11014>
Jason Ekstrand [Sat, 19 Jun 2021 02:55:08 +0000 (21:55 -0500)]
docs/isl: Improve the bit[6] swizzling section of the tiling chapter
Suggested-by: Luis Strano <luis.strano@intel.com>
Acked-by: Luis Strano <luis.strano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11479>
Jason Ekstrand [Fri, 18 Jun 2021 21:34:44 +0000 (16:34 -0500)]
isl,docs: Add a chapter on AUX state tracking
We also update and improve the docs in isl.h which get pulled into this
new chapter.
Acked-by: Luis Strano <luis.strano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11479>
Jason Ekstrand [Wed, 23 Jun 2021 21:48:36 +0000 (16:48 -0500)]
isl,iris: Move the extra_aux_surf logic into iris
This gets rid of the awkward interface for isl_surf_get_ccs_surf where
we passed it two aux surfaces and it was supposed to fill out the second
one based on whether or not the first one already had stuff in it.
Instead, we now pass it three well-labled surfaces: surf,
hiz_or_mcs_surf, and ccs_surf which have obvious meanings. This does
mean that iris has to carry a bit of logic and we have to flip
parameters around in all the callers. But the resulting interface is
much cleaner.
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11479>
Jason Ekstrand [Fri, 18 Jun 2021 22:12:32 +0000 (17:12 -0500)]
isl: Take a hiz_or_mcs_surf in isl_surf_supports_ccs
Whether or not a surface supports CCS on Tigerlake and later is
dependent not only on the main surface but also on the MCS or HiZ
surface, if any. We were doing some of these checks in
isl_get_ccs_surf based on the extra_aux parameter but not as many as we
probably should. In particular, we were really only checking HiZ
conditions and nothing for MCS. It also meant that, in spite of the
symmetry in names, the checks in isl_surf_get_ccs_surf were more
complete than in isl_surf_supports_ccs.
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11479>