review.tizen.org Git - platform/upstream/mesa.git/log

ci: polish deqp-runner a bit

Plus cosmetics adjustments to pass more of shellcheck.

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci/freedreno: fix the a530_piglit job and switch to Weston

If we count devices which running a530 with mainline kernel and Mesa,
it's probably mostly phones and tablets running on Wayland. Adapt to it.

Fixes: 83c2b26acf1e ("ci/freedreno: Switch the piglit job to using a deqp-runner suite.")

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci/freedreno: update a530 flakes, fails and skips

Add multiple skips for the:
- KHR-GLES31.core.pixelstoragemodes.teximage2d
- KHR-GLES31.core.pixelstoragemodes.teximage3d

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8837

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci/freedreno: a530 behaves stable in 6.3

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci: drop overriding new a530 firmware due to preemption issues with older kernel

Fixes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8825

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci/freedreno: do not restrict to 2 cpus on a530

Not needed for modern kernels (6.3+ for sure).

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

ci: bump kernel to the 6.3, support HDK 888 based on sm8350

- uprev to kernel 6.3-rc6
- add sm8350 support (needed for HDK 888)
- add missing patch for Asurada functionality
- enable CONFIG_ARM_SMMU_QCOM, since we dropped specific patch
worarounding need for it

Acked-by: Guilherme Gallo <guilherme.gallo@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22255>

tu/perfetto: s/MRTs/attachment_count/ in traces.

MRTs usually means >1 color buffers, so seeing "2" is surprising when
you're expecing just color and depth. Makes the output look more Vulkan.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22278>

tu/perfetto: Drop unused arg to send_descriptors().

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22278>

tu/perfetto: Use tu_CmdBeginDebugUtilsLabelEXT as a stage event in perfetto.

This lets zink mark points of interest (particularly its barriers and
blits) with some useful data, for presenting in perfetto traces.

Closes: #8487
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22278>

tu/perfetto: Clean up an extra token paste to just use the arg being passed.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22278>

tu/perfetto: Refactor code out of the macro, to stage_end.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22278>

rusticl/event: drop work item before updating status

This fixes some CTS compiler tests where they relied on the cl_kernel
object to be released in time so it can recompile a program without
throwing CL_INVALID_OPERATION due to still having active kernel objects.

Fixes: 47a80d7ff4f ("rusticl/event: proper eventing support")
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22510>

symbol_table: Prehash the key on insert, and reuse the entry on shadowing.

Mostly saves computing the hash twice, but while we're here there's no
need for shadowing to walk the table again.

Release Mesa build runtime of
KHR-Single-GL46.arrays_of_arrays_gl.SizedDeclarationsPrimitive -4.19869%
+/- 3.20231%

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22451>

symbol_table: Don't bother resetting the key on popping scope.

If you made a same-name symbol, then its name was just the name from the
parent scope anyway.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22451>

symbol_table: Don't maintain the HT as we're destroying the table.

Release Mesa build runtime of
KHR-Single-GL46.arrays_of_arrays_gl.SizedDeclarationsPrimitive -5.05801%
+/- 3.41206% (n=12)

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22451>

symbol_table: Store the symbol name in the same allocation as the symbol entry.

Saves an extra malloc. Release Mesa build runtime of
KHR-Single-GL46.arrays_of_arrays_gl.SizedDeclarationsPrimitive (which is a
lot of GLSL 4.60 builtin vars symbol table setup) -5.15821% +/- 3.19636%
(n=13).

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22451>

ci/amd: update checksums after DXVK 2.1 update

Heaven got fixed with DXVK 2.1.

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22447>

ci/dxvk: uprev to 2.1

Major uprev, requires VK 1.3 now.

Acked-by: Martin Roukala <martin.roukala@mupuf.org>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22447>

ci/amd: 4/5 runners TPad-C13 runners are online, restore most of the tests

Partial revert of 590959057c2b ("ci/amd: raven is currently downgraded
to 2 machines only, adapt")

Test which remains disabled: radeonsi-raven-va:amd64 (VAAPI testing).

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22447>

anv: enable blorp query reset for performance queries

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Felix DeGrood <felix.j.degrood@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22480>

ci: Drop executable permissions from backend compiler wrapper script

Not needed anymore, since the frontend scripts just source the backend
script now.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22438>

ci: Remove shebang from backend compiler wrapper script

Not needed anymore, since the frontend scripts just source the backend
script now.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22438>

ci: Use set -e in frontend compiler wrapper scripts.

I doubt the shebang line in the backend script has any effect now,
since the frontend scripts just source it directly.

v2:
* Use "set -e" instead of adding -e to shebang (Eric Engestrom)

v3:
* Apply to the clang wrapper scripts as well (Eric Engestrom)

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22438>

ci: Explicitly test for meson feature checks in compiler wrapper

The previous indirect method was more complicated and still error prone.

v2:
* Use "grep -E" (Eric Engestrom)
* Exclude spaces and slashes in the grep pattern, to avoid accidentally
matching across unrelated compiler arguments.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22438>

util: Delete Offset() macro from u_memory.h

Reviewed-by: Giancarlo Devich <gdevich@microsoft.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22494>

spirv/tests: Add test for single-block loop

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22509>

dzn: Remove #if D3D12_SDK_VERSION blocks now that 610 is required

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22504>

microsoft/clc: Remove #if D3D12_SDK_VERSION blocks now that 610 is required

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22504>

d3d12: Remove #if D3D12_SDK_VERSION blocks now that 610 is required

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22504>

meson: keep Mako version checking in accord with build msg

Fixes: 52194ae4df1 ("meson: Ensure that mako is >= 0.8.0")
Signed-off-by: Luc Ma <luc@sietium.com>
Reported-by: Terry Zhang <terry@sietium.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22499>

radeonsi/vcn: engage all jpeg engines on gfx940 for mjpeg decode

engage all available jpeg engines to decode mjpeg video frames

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22388>

radeonsi/vcn: enable RGBP format on gfx940 jpeg

decode output format conversion to RGBP is supported on gfx940 jpeg.

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22388>

frontends/va: add support for RGBP rt_format

check and enable RGBP rt_format and support VaImage creation for the RGBP/ARGB.

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22388>

util/format: add planar3 r8_g8_b8_unorm pipe format

add pipe format to represent three plane 8bit RGB format

Signed-off-by: Sathishkumar S <sathishkumar.sundararaju@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22388>

docs: add INTEL_MEASURE=cpu

Update documentation about INTEL_MEASURE=cpu

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

iris: Enable INTEL_MEASURE=cpu

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

anv: Enable INTEL_MEASURE=cpu

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

intel: INTEL_MEASURE cpu mode

INTEL_MEASURE normally measures timing of GPU events. However, it
is sometimes useful to instead measure when these gfx API calls
were requested of the driver. INTEL_MEASURE cpu can be used in
in conjunction with other driver debug capabilities, like
INTEL_DEBUG=pc for analyzing stalls/flushes or when debugger is
attached, to track which frame you're currently on or where in
the frame you're at.

Initial commit, without plumbing into anv/iris.

"INTEL_MEASURE=cpu" will collect a cpu timestamp for each
INTEL_MEASURE event instead of GPU timestamps.

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

anv: split INTEL_MEASURE multi events

Measure performance of each draw separately in multi_draw event.
Previously, we measured duration of the sum of all draws launched
per multi_draw. This should provide more detailed data for
multi_draws.

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

anv: Add flush reason to NEEDS_END_OF_PIPE_SYNC

cs_stall gets inserted if both flushes and invalidates are required.
This cs_stall reason was not called out explicitly, until now.

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

anv: Add flush reasons to raytracing flushes

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

anv: Add END_OF_PIPE_SYNC reporting to INTEL_DEBUG=pc

Reviewed-by: Mark Janes <markjanes@swizzler.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21505>

radeonsi/vcn: add an exception of field case for h264 decoding

This is an error case where the I field has no reference, however
the reference list contains some reference, which resulting in
not be able to find its reference, and it is a correct behavior,
should not enter the correction code. Just add this exception to
prevent such a case, if more exceptions found, they will
be treated indiviually in the future.

Cc: mesa-stable
closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8740

Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Signed-off-by: David (Ming Qiang) Wu <David.Wu3@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22485>

llvmpipe: fix some corner cases with line rendering

Some comparisons were wrong depending on swapped endpoints and
the bottom_edge_rule (if the endpoint was exactly touching
pixel center), hence the code assuming a line endpoint which
should be drawn according to diamond exit rules would already be
drawn (so not adjusting the endpoint) when in reality it was not,
in which case the line would end up one pixel too short.

Note that this is still not fully correct - the logic as such
should be correct now, however these comparisons can give wrong
results due to float math vs. fixed point planes (an endpoint very
close to a pixel center might be exactly at pixel center in fixed
point), but this should do for now. (Also, the logic is still
completely wrong for state trackers not using half_pixel_center
setting, but this is only really a concern for dx9 state tracker.)

Reviewed-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22481>

llvmpipe: minor cleanups in line rendering code

Use float constants, some code deduplication, no functional change.

Reviewed-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22481>

v3d: set depth compare function correctly

Depth compare function must be set to the configured one only when
compare mode is enabled; otherwise it must be configured to never.

v2 (Eric):
- Handle V3D < 4.0 case

CC: mesa-stable
Reviewed-by: Eric Engestrom <eric@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22470>

zink: use correct primitives for passthrough gs with tess

When tessellation is used the geometry shader needs to take the
tessellator output primitive as input.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Fixes: 90a8525d703 ("zink: handle edgeflags")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22498>

zink: don't emulate edgeflags for patches

Patches are used for tessellation, edgeflags don't work with
tessellation.

Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Fixes: 90a8525d703 ("zink: handle edgeflags")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22498>

radv: tidy up dirtying RBPLUS state in radv_bind_dynamic_state()

To dirty driver specific states at the same place.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22472>

radv: re-emit the guardband state when related PSO are bound

If one of the related states is static, the guardband state needs to be
re-emitted when a graphics pipeline is bound.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8828
Fixes: 40d8df72808 ("radv: emit the guardband state separately from the scissor state")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22472>

r600: fix refcnt imbalance related to r600_set_vertex_buffers()

For instance, this issue is triggered with: "piglit/bin/useprogram-flushverts-2 -auto -fbo" or
"piglit/bin/primitive-restart-draw-mode line_loop -auto"
while setting GALLIUM_REFCNT_LOG=refcnt.log.

Fixes: 27dcb466293f ("gallium: add take_ownership param into set_vertex_buffers to eliminate atomics")

Signed-off-by: Patrick Lerda <patrick9876@free.fr>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22395>

meson: ensure i915 Gallium driver includes Intel sources

Ensure builds flag the use of Intel sources when the i915 Gallium driver
is configured (`-Dgallium-drivers=i915`). Otherwise, a build may fail if
other Intel-based configuration options are not enabled:

./src/gallium/winsys/i915/drm/meson.build:21:0: ERROR: Unknown variable "libintel_common".

Signed-off-by: James Knight <james.d.knight@live.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22490>

intel: split URB space between task and mesh proportionally to entry sizes

Improves performance by 0.5-2.5% in vk_meshlet_cadscene
depending on the model.

Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22445>

radv: try to keep HTILE compressed for READ_ONLY_OPTIMAL layout

It should be handled like DEPTH_STENCIL_READ_ONLY_OPTIMAL.

This fixes an issue with VRS attachment because HTILE was considered
disabled for READ_ONLY_OPTIMAL but there is no reasons to disable it
as long as the image is only used as a depth/stencil attachment.

Otherwise, when HTILE is disabled, VRS rates are ignored.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8675
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22468>

broadcom/ci: deduplicate script definition

It's always going to be the same for all the jobs in our farm.

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Juan A. Suarez <igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22457>

vulkan: ignore rasterizationSamples when the state is dynamic

Fixes: 1deb83fb86d ("vulkan: Add more dynamic multisample states")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22444>

ci: update shebang to make it more portable

Doesn't matter for now, but it doesn't cost anything to make it portable :)

Suggested-by: David Heidelberg <david.heidelberg@collabora.com>
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

ci: bump tags

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

ci: start documenting which image tags need to be bumped

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

ci: move deqp-egl instead of copying it

To make sure we can't accidentally run a random leftover platform.

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

ci: add -android suffix for android build of deqp-egl

Makes it clearer which platform is being run.

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

ci: stop removing -x11 suffix for x11 build of deqp-egl

Makes it clearer which platform is being run.

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Helen Koike <helen.koike@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22450>

intel/vec4: force exec_all on float control instruction

Applying the same rule as the fs backend so that generation code
doesn't assert.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: daa8003e45 ("intel/fs: use nomask for setting cr0 for float controls")
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22473>

glthread: disable by default with fewer than 4 (big) CPUs

with a low CPU count or low count of "big" CPUs, glthread hurts perf
much more than it helps, so it should be disabled

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21926>

driconf: rework glthread enablement

this splits out glthread enablement into 3 settings:
- driver
- app
- user

which can then be modified with more granularity

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21926>

util/cpu: add big.LITTLE cpu detection

it's useful for drivers to be able to detect big.LITTLE topology in
order to make decisions about how to handle workloads (specifically as
it relates to threading)

Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21926>

docs/asahi: Add hardware glossary

Explains some of hardware units referenced throughout the driver.

Signed-off-by: i509VCB <git@i509.me>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22200>

nouveau: nouveau_copy_buffer can deal with user_ptrs just fine

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

nouveau: allow to enable SVM without having to enable CL

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

nv50/ir: ignore CL system values

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

rusticl: add support for fine-grained system SVM

At the moment it's an all or nothing. A driver supporting fine-grained
system SVM can enable it in order to get full SVM support.

Lower levels could be emulated by userptrs and placing the bo at the same
locations in the GPU's VM as well, but that would require reworking quite
a bit on the drivers side.

For now supporting mmu_notifiers on the kernel side is the only way of
getting SVM support with Rusticl.

The only driver having the gallium bits wired up atm is Nouveau, but I
suspect it shouldn't be all to hard for iris and radeonsi as well.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

rusticl/mem: add get_parent helper

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

rusticl/memory: Rework mapping of memory located in system RAM

The old code didn't consider a few cases where we could just map resources
located in system RAM.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

rusticl/context: add helper to get the max mem alloc size for all devices

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

gallium: correctly name the flags of svm_migrate

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19712>

zink: remove screen param from zink_prune_query()

no longer used

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22493>

zink: always defer query pool deletion

this feels dumb, but I can't think of a simpler way to do it that
would more accurately handle deletion while also guaranteeing
pool longevity

Fixes: 7da78ffb697 ("zink: create/use query pools dynamically")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22493>

zink: pass ctx through query destroy paths

no functional changes

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22493>

zink: remove atomics from zink_query

this is never accessed from threads anymore and
hasn't been for a long time

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22493>

frontends/va: report min width and min height values if available

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8756
Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Reviewed-by: Boyuan Zhang <Boyuan.Zhang@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22422>

radeonsi: return min width and min height video cap values

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Reviewed-by: Boyuan Zhang <Boyuan.Zhang@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22422>

gallium/pipe: add min width and min height video cap enums

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Reviewed-by: Boyuan Zhang <Boyuan.Zhang@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22422>

v3d: add flake spec@ext_framebuffer_blit@fbo-sys-sub-blit

https://gitlab.freedesktop.org/mesa/mesa/-/jobs/39905522

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22486>

d3d12: Support separate front/back stencils

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3833
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22454>

d3d12: Add ID3D12GraphicsCommandList8 to the context

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22454>

d3d12: Update PSO creation to use CreatePipelineState

This uses CD3DX12_PIPELINE_STATE_STREAM3 from
d3dx12_pipeline_state_stream.h and gives us access to newer D3D12
features.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22454>

d3d12: Query device for D3D12_FEATURE_D3D12_OPTIONS14

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22454>

d3d12: Update and require DirectX-Headers 1.610.0

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22454>

vulkan/runtime: discard unused graphics stages in libraries

Anv is trying to rely on the stages put into the library graphics
state.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22460>

mesa/st/program: don't init xfb info if there are no outputs

this is almost certainly a failure case, but drivers still shouldn't
get xfb info if there are no outputs

affects:
spec@glsl-1.50@execution@interface-blocks-api-access-members

cc: mesa-stable

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22448>

llvmpipe: code clean-ups in llvmpipe_get_query_result_resource()

Replace 'unsignalled' with 'ready' to simplify logic.
Remove needless !! in !!(a > b) expressions.
Remove some casting. Use MIN2() macro. Add const qualifiers.
Declare loop vars in loops.

Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22464>

llvmpipe: remove lp_setup_alloc_triangle()'s unneeded tri_size param

This returned size was never used by the callers.

Signed-off-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22464>

docs: add missing MESA_VK_WSI_HEADLESS_SWAPCHAIN variable

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22478>

amd: Rename INDIRECT_BUFFER_CIK to just INDIRECT_BUFFER.

This packet is supported on GFX6 too, its name should relect that.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22406>

radv: Chain command buffers on GFX6 in radv_queue.

Also don't check whether chaining is enabled in radv_queue, the
winsys will take care of that anyway.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22406>

radv: Use IB BOs (chaining) by default on GFX6.

GFX6 supports IB chaining since the PFP firmware version 20.
Note that the very first amdgpu firmware for GFX6 already had
version 29, so we can assume that all GPUs supported by RADV
have this feature.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22406>

radv: Disallow IB2 on GFX6 when using draw_indirect_multi.

GFX6 has the same problem as GFX7 here.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22406>

radv/amdgpu: Remove unnecessary assertions from chaining.

These used to guard against chaining on GFX6 and on HW IP types
that don't support chaining, but these things are now guarded
elsewhere and these assertions are no longer necessary.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/22406>