SSABuilder::alloc_ssa() is now only for scalars. We intoduce
SSABuilder::alloc_ssa_vec() which handles the vector case the way
alloc_ssa previously did. This matches the split used in SSAValueAllocator
We're about to drop Copy from SSARef, which makes it a lot more annoying
to deal with. SSAValue will remain Copy though, so we want to start
using it instead of SSARef where possible.
Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34794>
Code that might have an invalid SSAValue is encouraged to use
Option<SSAValue> instead of NONE, which is now the same size as a u32
and provides more type safety.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34794>
Required extensions to support wsi have been relaxed earlier for sw blit
path. So the renderer extensions enablement is fixed to be passive based
on the renderer side correspondingly.
Test: emulate to drop dma-buf and modifier support from host anv driver,
and confirm wsi via venus works with sw blit fallback and device
creation no longer returns VK_ERROR_FEATURE_NOT_PRESENT.
Fixes: 06f5d1a105 ("venus: expose WSI on renderer without dma-buf support")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34827>
Without a type, we can't really interpret the data. Currently, it just
returns None in the presence of modifiers so it's okay. Also, all of
the callers of this helper today do so on the source of an OpPrmt which
doesn't support source modifiers.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34818>
This makes more sense as we don't want any complex logic around source
modifiers or any of that. We just want to handle Zero and Imm32 in the
same case. Also, explicitly assert that modifiers are None, which is
more clear anyway.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34818>
Those platforms requires aux map with 1MB alignment, for slab that
means that any buffer needs to have size of multiple of 1MB what
causes a lot of memory to be wasted causing it to run out of memory
when running multiple GPU applications.
Fixes: ea18572ff2 ("anv: Add support for ANV_BO_ALLOC_AUX_CCS in anv_slab_bo")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34817>
v10 have 96 and v12+ have 128, not the opposite.
Fixes: 811525b543 ("pan/genxml: Build libpanfrost_decode for v12")
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34815>
history:
- 8d83ffe531 ("radv/ci: Add coverage for ETC2 emulation.")
var added to the jobs as `radv_require_etc2`, but not passed to duts
- 567c32b55c ("radv, drirc: rename radv_require_{etc2,astc}")
var renamed to `vk_require_etc2`, except in the jobs
- 5f177018f7 ("radv/ci: re-enable ET2C emulation testing on non-native GPUs")
var renamed in the jobs, but still not passed to duts
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34796>
Emitting compute dispatches on SDMA just hangs. It might be needed
to switch to gang submit for these to work but fixing the GPU hang is
more important for now.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34805>
The dEQP-VK.ray_tracing_pipeline.* ones might need a bigger net to catch
them all, but I didn't want to prematurely ignore too many tests, so for
now I'm just dumping a big list.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34811>
../../../../../../../mesa/src/compiler/nir/nir_lower_gs_intrinsics.c: In function ‘nir_lower_gs_intrinsics’:
../../../../../../../mesa/src/compiler/nir/nir_lower_gs_intrinsics.c:523:93: warning: ‘state’ may be used uninitialized [-Wmaybe-uninitialized]
523 | state.decomposed_primitive_count_vars[i] = state.decomposed_primitive_count_vars[0];
| ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~
../../../../../../../mesa/src/compiler/nir/nir_lower_gs_intrinsics.c:464:17: note: ‘state’ declared here
464 | struct state state;
| ^~~~~
It's always initialized by the first iteration of the loop, but GCC
doesn't seem to know that.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34785>
../../../../../../../mesa/src/compiler/nir/nir_opt_shrink_vectors.c: In function ‘shrink_dest_to_read_mask’:
../../../../../../../mesa/src/compiler/nir/nir_opt_shrink_vectors.c:140:36: warning: writing 16 bytes into a region of size 15 [-Wstringop-overflow=]
140 | swizzle[first_bit + i] = i;
| ~~~~~~~~~~~~~~~~~~~~~~~^~~
../../../../../../../mesa/src/compiler/nir/nir_opt_shrink_vectors.c:138:18: note: at offset [1, 15] into destination object ‘swizzle’ of size 16
138 | uint8_t swizzle[NIR_MAX_VEC_COMPONENTS] = { 0 };
| ^~~~~~~
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34785>
In function ‘SHA1Update’,
inlined from ‘SHA1Pad’ at ../../../../../../../mesa/src/util/sha1/sha1.c:157:2,
inlined from ‘SHA1Final’ at ../../../../../../../mesa/src/util/sha1/sha1.c:168:2:
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: warning: ‘SHA1Transform’ reading 64 bytes from a region of size 1 [-Wstringop-overread]
135 | SHA1Transform(context->state, (uint8_t *)&data[i]);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: note: referencing argument 2 of type ‘const uint8_t[64]’ {aka ‘const unsigned char[64]’}
../../../../../../../mesa/src/util/sha1/sha1.c: In function ‘SHA1Final’:
../../../../../../../mesa/src/util/sha1/sha1.c:55:1: note: in a call to function ‘SHA1Transform’
55 | SHA1Transform(uint32_t state[5], const uint8_t buffer[SHA1_BLOCK_LENGTH])
| ^~~~~~~~~~~~~
In function ‘SHA1Update’,
inlined from ‘SHA1Pad’ at ../../../../../../../mesa/src/util/sha1/sha1.c:159:3,
inlined from ‘SHA1Final’ at ../../../../../../../mesa/src/util/sha1/sha1.c:168:2:
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: warning: ‘SHA1Transform’ reading 64 bytes from a region of size 1 [-Wstringop-overread]
135 | SHA1Transform(context->state, (uint8_t *)&data[i]);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: note: referencing argument 2 of type ‘const uint8_t[64]’ {aka ‘const unsigned char[64]’}
../../../../../../../mesa/src/util/sha1/sha1.c: In function ‘SHA1Final’:
../../../../../../../mesa/src/util/sha1/sha1.c:55:1: note: in a call to function ‘SHA1Transform’
55 | SHA1Transform(uint32_t state[5], const uint8_t buffer[SHA1_BLOCK_LENGTH])
| ^~~~~~~~~~~~~
In function ‘SHA1Update’,
inlined from ‘SHA1Pad’ at ../../../../../../../mesa/src/util/sha1/sha1.c:160:2,
inlined from ‘SHA1Final’ at ../../../../../../../mesa/src/util/sha1/sha1.c:168:2:
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: warning: ‘SHA1Transform’ reading 64 bytes from a region of size 8 [-Wstringop-overread]
135 | SHA1Transform(context->state, (uint8_t *)&data[i]);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../../../../../mesa/src/util/sha1/sha1.c:135:25: note: referencing argument 2 of type ‘const uint8_t[64]’ {aka ‘const unsigned char[64]’}
../../../../../../../mesa/src/util/sha1/sha1.c: In function ‘SHA1Final’:
../../../../../../../mesa/src/util/sha1/sha1.c:55:1: note: in a call to function ‘SHA1Transform’
55 | SHA1Transform(uint32_t state[5], const uint8_t buffer[SHA1_BLOCK_LENGTH])
| ^~~~~~~~~~~~~
Reaching this code is impossible for the SHA1Update() calls in SHA1Pad().
Use assume() to inform the compiler of this.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34785>
If the render pass writes depth (with direction) but doesn't set
direction on the GPU, the LRZ buffer cannot be used in subsequent
render passes because the direction information is lost.
Fixes rendering in "Elite Dangerous" game.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34577>
- LOAD results in reusing LRZ from previous renderpasses.
- CLEAR clears LRZ.
- DONT_CARE doesn't clear and doesn't load, so LRZ should be
disabled, that was also the initial idea in code, but conditions
didn't work out and are now fixed.
Fixes glcts tests:
KHR-GL46.packed_depth_stencil.blit.depth24_stencil8
KHR-GL46.packed_depth_stencil.blit.depth32f_stencil8
KHR-GLES3.packed_depth_stencil.blit.depth24_stencil8
KHR-GLES3.packed_depth_stencil.blit.depth32f_stencil8
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34577>
