The runtime vk_android.h header has proper android detection inside, so
no need to wrap it with redundant android detection. Meanwhile, the enum
VK_EXTERNAL_MEMORY_HANDLE_TYPE_ANDROID_HARDWARE_BUFFER_BIT_ANDROID is
defined in the vulkan_core.h, so no need Android wrap either.
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Acked-by: Rob Clark <robclark@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36151>
Otherwise we can have a case where binning VS uses more consts than
full VS (when safe variant is used for full VS), that will result in
a rendering issue because SP_VS_CONST_CONFIG.CONSTLEN is shared between
full and binning VS in PROGRAM_CONFIG state and gets the value from the
full VS.
There are two alternative solutions that can allow binning VS to always
use maximum constlen:
- Move constlen emission to per-XS config. This interferes
PROGRAM_CONFIG state which uploads consts and does SP_UPDATE_CNTL.
Consts would need to be uploaded after constlen is defined, while
SP_UPDATE_CNTL must be done before per-XS state is emitted.
Also having SP_UPDATE_CNTL in a draw state that is always DIRTY
isn't great.
Something didn't work out on A6XX, so this idea was dropped.
- Emit constlen again in VS_BINNING draw state. This seem to work
but also likely an undefined behaviour since constlen is changed
after some consts are uploaded.
Cc: mesa-stable
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36203>
We need the job artifacts in all cases to gather performance data.
Since commit b723bc80d2, we were only saving them on failures.
Fixes: b723bc80d2 ("ci/lava: inherit .piglit-traces-test in .lava-piglit-traces and deduplicate configs")
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36195>
After commit 545727f97c (ci/android: Move ANDROID_CTS_MODULES to build
script, 2025-06-24) the comment about ANDROID_CTS_MODULES in
lvp-android-angle-android-cts-include.txt has become inaccurate.
Update the instructions to reflect the latest status.
Fixes: 545727f97c
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36187>
Check if filesystems are mounted before attempting to mount them,
and use mkdir -p to prevent errors if directories already exist.
Signed-off-by: Vignesh Raman <vignesh.raman@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35961>
The `.gitlab-ci/tests/` directory contains LAVA-specific tests, not
general test code. To avoid confusion with `.gitlab-ci/test/`, move it
to `.gitlab-ci/lava/`.
Signed-off-by: Vignesh Raman <vignesh.raman@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35961>
Thanks to the optimizations in the previous commits and further
increasing tests_per_group, we can now run more tests in a618-vk
while also reducing its parallelism.
Dropping `DEQP_FRACTION: 2` and only increasing the `fraction` of
the basic test set to 3 results in broader overall coverage.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36027>
The a618 jobs run on two different 8 thread devices, which is the optimal
value for FDO_CI_CONCURRENT on most of their jobs.
Signed-off-by: Valentine Burley <valentine.burley@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36027>
Cause of the following error on x86 (32-bit):
src/gbm/libgbm.so.1.0.0: unknown symbol exported: __deregister_frame_info_bases
src/gbm/libgbm.so.1.0.0: unknown symbol exported: __register_frame_info_bases
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36182>
For import, requested size might be smaller than the whole blob mem
size. Currently, venus virtgpu backend tracks the initial import size
and sanity checks with later re-import size. So let's add a way to
request whole blob mem size
No need vtest backend changes since dma_buf import is not supported.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36128>
Prepare for adopting common runtime so that we don't leave unused vars
after dropping codes from vn_AllocateMemory. It's trying to minimize
code changes in the venus adoption since that'd be huge.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36128>
Instead of failing the import, drop the mappable bit if blob size is
smaller than the requested alloc size, which defers the error to the
intended vkMapMemory api call if the client app requests Vulkan mapping
from the imported external memory.
Normally this won't occur if the app obeys the spec with a properly
implemented blob mem allocator. However, legacy allocators like minigbm
virtgpu_virgl backend could allocate via virgl w/o knowing the real blob
size from the guest side. The unsatified cases are external gralloc
images, which won't be mapped in any meaningful way after being
imported. This is to prepare dropping the force_unmappable workaround,
and to further prepare adopting common Android runtime.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36128>
I keep reaching for this helper but it doesn't exist. So I fixed that.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36142>
I'm guessing the hardware needs to prefetch the whole sampler heap, so if we're
not gonna use it, let's omit it. I don't know if this helps, but it can't hurt.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36127>
