With the switch to common dynamic state tracking, something got lost
that made the scissors not always be emitted when they are not dynamic
and the pipeline is marked dirty.
Since both viewport and scissors make use of each other to calculate
their values, just stick the scissor emit in the same if block as
viewport for now.
I'd rather have them decoupled, and at least the Vulkan CTS didn't
complain when I tried it, but I don't know what other effects that
may have, especially when it comes to the guardband.
Fixes a bunch of tests under
dEQP-VK.pipeline.*.multisample.misc.*
Fixes: 7d25c04236 ("anv: Switch to using common dynamic state tracking")
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17964>
We used the result type of lod and txs for sampler declarations,
which broke following instructions that are not lod and txs.
Use the sampler type from TGSI if it's present, not the result type
of lod and txs.
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17692>
The goal here is to allow zink and ANGLE to switch sRGB on and off, but
keep UBWC enabled for most textures.
This doesn't quite work yet for zink on gbm, where zink asks the device if
it supports a UBWC-modifier image with unorm+sRGB mutable formats (yes),
but then doesn't pass the unorm+sRGB format list to vkImageCreate(), which
then asserts because it couldn't make a general
VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT UBWC image.
The format list here comes from Danylo's testing in
https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17278#note_1447812
Reduces frame times in Genshin Impact GLES AoV 2.02063% +/- 0.207575% (n=5)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17900>
"If viewFormatCount is zero, pViewFormats is ignored and the image is
created as if the VkImageFormatListCreateInfo structure were not included
in the pNext chain of VkImageCreateInfo."
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17900>
For external image imports, we'd lose the mutable image format list,
causing turnip to get angry that we try to do UBWC despite not having a
UBWC-compatible format list.
Cc: mesa-stable
Fixes: 28ee911ad6 ("zink: handle mutable swapchain images with dmabuf")
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17900>
clc_compiler_test.cpp:1322:67: error: non-constant-expression cannot be narrowed from type 'double' to 'float' in initializer list
log(0.0f) / log(2), log(1.0f) / log(2), log(2.0f) / log(2), log(3.0f) / log(2)
clc_compiler_test.cpp:2306:25: error: non-constant-expression cannot be narrowed from type 'std::vector<unsigned int>::size_type' (aka 'unsigned long long') to 'unsigned int' in initializer list
CompileArgs args = { inout.size(), 1, 1 };
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17926>
The driver should still gather XFB info even if all XFB outputs are
dead, otherwise the pipeline can't find the streamout shader.
RADV should use vk_spirv_to_nir() at some point to reduce code
duplication during SPIRV->NIR compilation.
This fixes new dEQP-VK.transform_feedback.simple.*.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17939>
These jobs were taking more than 20 minutes to run, slowing the pipeline
down.
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17935>
As we have some free, and then we can reassign two Stoney Ridge boards
to reduce the duration for some jobs that are taking too long currently.
Also, these machines are much faster so we are able to run all of the
Piglit tests and a bigger portion of the VA tests.
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17935>
Using AMD_DEBUG=tmz to enable tmz without the application asking
for it is useful for testing. Since SCANOUT isn't set on Wayland,
use RENDER_TARGET instead as a heuristic to try to allocate color
framebuffer as tmz.
With this change we can enable tmz in x11 and wayland.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11449>
This check is an infinite source of issues when testing protected
content, so make it opt-in (eg for developers of applications dealing
with protected content buffers).
On AMD and Intel importing a protected buffer as non-protected will cause
the display to be incorrect (= garbage).
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11449>
Newcomers may want to test traces locally and understand how the replay
works.
Acked-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17347>
It seems like VKCTS's startup time is close to the current console
activity timeout (2 minutes...) on the Steam Deck, leading to spurious
timeouts...
Let's bump the timeout by 50%!
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17829>
The test can take a little longer than 60s, which leads to spurious
timeouts and thus unstable CI pipelines.
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17829>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17829>
v2:
- Move the other 2 `sample_texture` tests to the flake list (Ishi)
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17829>
We were writing descriptors into si_context and then copying them into
the command buffer. Just write them into the command buffer directly.
Also set the pointer to VBO descriptors right after them.
When we start a new command buffer or we finish blitting, we no longer
restore precomputed VBO descriptors. Instead, we just reupload them again.
It's a compromise to have the common path simpler and faster (maybe).
This removes a lot of stuff. Now the VBO descriptor upload path looks
very similar to the display list path.
There was an accidental hidden optimization that is now documented as
"last_const_upload_buffer".
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17933>
While this is nice to have, it doesn't include VBO descriptors in user
SGPRs, and we need to remove it, so that we can simplify the VBO code.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17933>
This is more straightforward. Also, radeon_add_to_buffer_list makes
writing VBO descriptors into the command buffer slower after that code
is reordered in following commits. This seems to be the only way that
isn't slower.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17933>
We can't invalidate CCU if there is any dirty data that hasn't been
flushed yet. In the case where we clear depth, we know that the depth
attachment itself isn't dirty but there may be dirty data from other
renderpasses. Therefore we need to flush before invalidating depth.
Fixes: 487aa80 ("tu: Rewrite flushing to use barriers")
Closes: #6987
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17940>
