panfrost_invert_swizzle produces a (one-sided) inverse, that is a
function g such that g(f(x)) = x. However, it is actually used in
some cases where we want a two-sided inverse, where we also have
f(g(x)) = x. An example is its use to pre-swizzle border color
values so that a later pass will un-swizzle them. If the swizzle
is not one-to-one this two sided inverse does not exist. However,
we can do better than the original code, e.g. for an RRR1 swizzle
inverse produced was originally B000, which when applied on the wrong
side results in BBB1 as output, whereas R000 would produce the
desired RRR1 output. Using the first valid component we see, rather
than the last one, is thus usually better.
The "correct" solution is to re-write all the code that uses
an inverse to handle non-unique inverses. But frankly these uses only
crop up in fairly niche cases like tests, and it's probably not worth
spending a lot of effort to deal with these edge cases when this
patch fixes most of them.
Fixes some failing piglit ext_framebuffer_multisample tests.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31060>
Found with a full run, fraction runs seems to not trigger it so far.
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/31089>
Panthor support was added in Linux 6.10 and Panfrost V10
in Mesa, enabling new GPUs on Rockchip's RK3588. Add CI
jobs to test Mali V10.
Signed-off-by: Vignesh Raman <vignesh.raman@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30883>
Create a new stage called arm-postmerge and move the full and manual
jobs over there, to avoid entanglement with the pre-merge jobs.
Signed-off-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30784>
After the transition to vk_meta, all clear tests seem to pass.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29451>
There's no real blocker here, so let's just allow it.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29451>
Use vk_meta_resolve_rendering() to handle render-pass resolves.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29451>
Forgot to remove those two entries when merging previous MR.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Fixes: 091df61138 ("panvk: Skip blend descriptors when no fragment shader is present")
Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30740>
Basically what 004e0eb3ab ("panfrost: use RGB1 component ordering
for R5G6B5 pixel formats") was doing in the gallium driver, but
applied to panvk this time.
Fixes: 004e0eb3ab ("panfrost: use RGB1 component ordering for R5G6B5 pixel formats")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30685>
Found with a full run without fraction, seems to have been present for a
while.
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/30293>
We already support everything needed for maintenace3.
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/29410>
Turns out the current approach makes implementation of advanced features
like update-after-bind or shader modules quite challenging. Instead of
adding hacks all over the place to support these features, let's use
the Valhall descriptor model.
Each shader now gets its own descriptor tables, which are fed by pilot
shaders copying the descriptors used by the shader from the descriptor
sets currently bound the command buffer.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Co-developped-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29654>
It was previously possible to overflow the data_size calculation (for
example with width=4096, height=4096, array_size=256 and any format)
This was then causing GetImageMemoryRequirements2 to return a size of 0.
Fix "dEQP-VK.pipeline.*.render_to_image.core.2d_array.huge.width_height_layers.r8g8b8a8_unorm" failures.
Signed-off-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29670>
Not that I really care about fixing copies now that vk_meta_copy is on
its way, but it fixes OOB accesses causing new crashes after the
panvk_mempool changes.
Fixes: f73ae1a6b5 ("panvk: Implement vkCmdCopyImageToBuffer()")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29670>
d91d2c275e ("panfrost: change default rounding mode for samplers")
changed the default coordinate rounding mode to nearest-even, which is
what we should use in Vulkan. But it turns out the CTS picks a value
that's too close to pixel edge on right_edge tests, which causes a
precision loss on coordinate rounding and leads to an off-by-one
on these tests.
Flag those as failing until the CTS is fixed to take subpixel precision
into account when selecting the pixel coordinate offset.
See https://gitlab.khronos.org/Tracker/vk-gl-cts/-/issues/5168.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29670>
And get it tested by CI.
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29302>
The hardware doesn't prevent us from mapping random 2DArray layers
as cube/cube arrays. The only restriction we have is on the number
of layers we pass (must be a multiple of 6 for cube arrays).
This patch makes the surface emission logic cube-agnostic, and
moves the cube face -> surface index conversion logic one layer
up to simplify things.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28909>
Stop passing panvk_varyings_info around and emit varying attributes
when building the pipeline. All the command buffer logic has to do is
allocate varying memory, and emit the attribute buffer descriptors
pointing to these buffers. We also keep the buffer index fixed to keep
things simple, when a buffer is missing, it will simply be filled with
a zero-sized/NULL entry.
Note that we store the buffer stride information in panvk_pipeline_shader
to prepare the transition to vk_shader.
As a bonus, this simplification seems to fix a few CTS failures.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28927>
Mark "dEQP-VK.pipeline.monolithic.vertex_input.misc.stride_change_vert_frag"
as failing.
Reason for failure is unknown for now, will need investigation at a
later time.
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/28939>
Apparently something fixed the copy tests that were failing.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Acked-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28811>
We regressed a pipeline.monolithic test when switching to dynamic
rendering and this went unnoticed because
dEQP-VK.pipeline.monolithic.* is not run.
With this addition, execution time is too long to fit in a single job,
so we split VK tests in 2 and only run half of the tests, which should
be okay as long as we keep the tests manual.
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28417>
