This helper sanity-checks a VkVideoProfileInfoKHR to make sure it
represents a real profile corresponding to specific codec rules.
Reviewed-by: David Rosca <david.rosca@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37656>
this is still not technically correct, as I'm seeing some weird fails
on ANV, at least, and also it maybe needs a pass to strip InterpolateAtSample
(related?), but it is now actually enabling custom sample locations
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37618>
Xwayland seems to mix implicit and explicit sync, depending on client
app. This trips up the heuristic that disables implicit sync once it
starts seeing app using explicit sync. This is not typical behavior,
so add a driconf override to disable the heuristic.
Fixes: 137cd3b0fa ("freedreno/drm: Move no_implicit_sync accounting")
Cc: mesa-stable
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37703>
This function disabled HiZ support when it encountered LODs which did
not satisfy a restriction of ISL_AUX_OP_AMBIGUATE for gfx8-9. Now that
the previous commit avoids that auxiliary operation for those platforms,
it is not so useful. Replace it with a simple check of the aux-usage.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36383>
We disable HiZ for some LODs on gfx8-9 to comply with ambiguate
operation restrictions. Avoid this restriction by initializing HiZ to
the CLEAR state on those platforms. By doing this, an ambiguate will
never occur. Also, do this for ICL as an optimization.
We'll enable HiZ for all LODs on gfx8-9 in the next patch.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36383>
By dumping the contents of a HiZ buffer before and after fast-clearing,
I've observed that a zeroed HiZ block corresponds to the CLEAR state
until gfx12. The fast-clearing application was piglit's bin/hiz. I ran
this test on a couple bare metal platforms (ICL and BDW) and many
simulated ones (SKL, TGL, DG2, and LNL).
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36383>
For CCS_E on gfx12+, this will cause us to perform full resolves when
transitioning from undefined to a layout which does not support
compression. We don't currently perform such transitions because
compression is always enabled.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36383>
isl_aux_get_initial_state() will soon be used for non-zeroed CCS on
gfx9+. Update the function to avoid hitting an unreachable().
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36383>
RD dump generation can be expensive and can only be desired for some
specific part of execution. Trigger file mechanism helps with this to a
certain degree but is still somewhat inexact.
FD_RD_DUMP_SUBMITS environment variable can be used to specify ranges of
submit indices for which RD dumps of command streams should be generated.
FD_RD_DUMP_FRAMES environment variable can similarly be used to specify
ranges of frames under which RD dumps for submitted command streams should
be generated. Frame ranges only really work with Turnip since the frame
count data is available there.
Signed-off-by: Zan Dobersek <zdobersek@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37499>
Currently we wait until the second dword in feedback buffer changes
from 0 to 1, and then the rest of the feedback is read. There is no
guarantee that the rest of the feedback will be available, which can
cause bitstream size to be incorrectly returned as 0.
Add write memory command after encode, marking the query as available
to ensure the entire feedback buffer is ready.
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13601
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36772>
This is required for the next job, because we'll need a login shell
session to be able to use `ninja`, which the normal build jobs
automatically handle, but the custom docs job we had didn't.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37490>
With the migration concluded on !37510 of the lava-job-submitter (it is now an
independent project under the gfx-ci namespace), a lava requirements file was
not necessary anymore, but there was still one reference.
Signed-off-by: Sergi Blanch Torne <sergi.blanch.torne@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37713>
This header is only used in a single place, which will generate macros
based on these definitions. Let's remove the header, and just include
the separate files in pvr_csb.h instead.
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37675>
Signed-off-by: Frank Binns <frank.binns@imgtec.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Acked-by: Simon Perretta <simon.perretta@imgtec.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37683>
This can trigger an assert otherwise. The space reserved before
executing DGC IBs is an arbitrary number which should be large enough
in all cases.
Found this while implementing descriptor heap.
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37681>
Based on a patch by llyyr <llyyr.public@gmail.com>:
!36827 added the copy_sync_payloads function, but didn't enable use of
it in radv. This commit mirrors similar MRs for anv/panvk/nvk and uses
the common vk_drm_syncobj_copy_payloads function for copy_sync_payloads.
I'm not too familiar with radv internals, so there's potentially a good
reason why this isn't a good change. However, I've personally been using
this patch locally for around a month and have experienced no
regressions and around 8% uplift on vkmark test scores with a 6600 XT.
[vertex] device-local=true: 45110 -> 48489 (+7.5%)
[vertex] device-local=false: 17529 -> 17488 (-0.2%)
[texture] anisotropy=0: 44768 -> 48679 (+8.7%)
[texture] anisotropy=16: 44920 -> 48572 (+8.1%)
[shading] shading=gouraud: 44931 -> 48467 (+7.9%)
[shading] shading=blinn-phong-inf: 44849 -> 48740 (+8.7%)
[shading] shading=phong: 44695 -> 48645 (+8.8%)
[shading] shading=cel: 44809 -> 47938 (+7.0%)
[effect2d] kernel=edge: 45185 -> 47837 (+5.9%)
[effect2d] kernel=blur: 26919 -> 26762 (-0.6%)
[desktop] <default>: 40974 -> 44034 (+7.5%)
[cube] <default>: 45090 -> 49270 (+9.3%)
[clear] <default>: 41102 -> 44375 (+8.0%)
(https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37606)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37640>
This was getting quie big... also fix KEPLER_A missing compute channel.
This now make it possible to also specify by generation full name while
retaining compat with the old names.
Signed-off-by: Mary Guillemard <mary@mary.zone>
Reviewed-by: Mel Henning <mhenning@darkrefraction.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37700>
Here we write the binary search to avoid looping over the list
by keeping track of the mid point entries and moving directly
from the last test point.
Fixes: bf946bcc ("util: add range remap util")
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37386>
Commit 2aaa6ebb "build/amd: add amd-use-llvm build option"
allows to build radeonsi and radv without LLVM dependency
so LLVM constraint is applicable when amd-use-llvm=true
Fixes: 82047fa8 ("amd: drop support for LLVM 15, 16, 17")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37695>
We don't need to have this restriction, we just need to fence_flush()
before trying to access the fd. This path is exercised in CL CTS
gl-interop test if cl_gl_sharing is exposed.
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37696>
If we need to blit a resource to a buffer that can be exported (ie.
because it was suballocated, etc), always use the aux-context. This
can be called off of the driver thread, so it isn't thread-safe to
use the main context, even if it is not NULL.
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37696>
The layout_resource_for_handle() path is only intended for images/
textures. For buffers, call fdl_layout_buffer() directly.
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37696>
As long as fdl is happy to layout the imported buffer type, we shouldn't
restrict it further. The cl_gl_sharing extends import/export beyond
winsys buffers, and these restrictions were getting in the way.
Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37696>
