this avoids a number of cases where a shader was reading more components
from an input than an output was providing. functionally there was never
any issue as these read components were subsequently rewritten to use
constant data, but the read itself is a spec violation
shrinking can't be done in finalize, however, as that enables the frontend
to optimize vertex states, which seems like a good thing but ends up being
a bad thing since it may or may not be consistent across frontends and I
don't wanna deal with having to reorder i/o locations in unintuitive ways
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25433>
Either none push consts are loaded via preamble or all of them even if
shader uses a single push const.
Fixes misrenderings with Zink. Zink unconditionally declares push
consts, regardless of whether they are actually used in the shader.
Fixes: a5f0f7d4b1
("turnip,ir3: Implement A7XX push consts load via preamble")
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25638>
It's a common case for Zink which has to declare push consts in
pipeline layout, even if they are not actually used in shaders, due
to the compatibility rules.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25641>
It can be beneficial to move phi nodes, even though they can often be coalesced.
Model this cost so nir_opt_preamble can make good decisions about hoisting phi
nodes (and by extension, if-statements) into the preamble.
At this point in the series, this has no effect, but it will avoid certain
shader-db regressions associated with the nir_opt_preamble changes later in the
series.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24011>
Speculating these loads is safe, but nir_opt_preamble doesn't know that. Set the
ACCESS bits appropriately to let it know.
This will avoid any code gen regression from the nir_opt_preamble change.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24011>
The way we init render pass related structures is dangerous with when
structs are not zero initialized - too easy to miss a field. There
were already at least two issues with it.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25592>
First, we need to give the parent_instr field a unique name to be able to
replace with a helper. We have parent_instr fields for both nir_src and
nir_def, so let's rename nir_src::parent_instr in preparation for rework.
This was done with a combination of sed and manual fix-ups.
Then we use semantic patches plus manual fixups:
@@
expression s;
@@
-s->renamed_parent_instr
+nir_src_parent_instr(s)
@@
expression s;
@@
-s.renamed_parent_instr
+nir_src_parent_instr(&s)
@@
expression s;
@@
-s->parent_if
+nir_src_parent_if(s)
@@
expression s;
@@
-s.renamed_parent_if
+nir_src_parent_if(&s)
@@
expression s;
@@
-s->is_if
+nir_src_is_if(s)
@@
expression s;
@@
-s.is_if
+nir_src_is_if(&s)
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/24671>
A descriptor set is internally reserved for descriptor set dynamic
offset which might not be used by an applications which otherwise
requires an extra descriptor set. This driconf option allows making
that trade-off by dropping support for dynamic offsets in exchange
for an extra descriptor set which means 5 usable descriptor sets on
A6XX and 8 on A7XX.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25534>
Allows for the descriptor set count to vary at runtime depending on
the specific GPU to allow for 7 usable descriptor sets on A7XX with
one reserved for dynamic offsets.
Passing VK-CTS: dEQP-VK.binding_model.*
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25534>
A7XX increases the maximum amount of descriptor sets from 5 to 8,
the amount of descriptor sets was added as a property of the
A6xxGPUInfo to accommodate it changing across GPUs.
Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25534>
This test checks the driver's reported conformance version against the
version of the CTS we're running. This check fails every few months
and everyone has to go and bump the number in every driver.
Running this check only makes sense while preparing a conformance
submission, so skip it in the regular CI.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25519>
New push consts loading consist of:
- Push consts are set for the entire pipeline via HLSQ_SHARED_CONSTS_IMM
array which could fit up to 256b of push consts.
- For each shader stage that uses push consts READ_IMM_SHARED_CONSTS
should be set in HLSQ_*_CNTL, otherwise push consts may get overwritten
by new push consts that are set after the draw.
- Push consts are loaded into consts reg file in a shader preamble via
stsc at the very start of the preamble.
OPC_PUSH_CONSTS_LOAD_MACRO is used instead of directly translating NIR
intrinsic into stsc because: we don't want to teach legalize pass how
to set (ss) between stores and loads of consts reg file, don't want for
stsc to be reordered, etc.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25086>
Previously the idea was for each generation to have unique list
of feature flags, now it makes more sense for new generation
to have only new flags defined and "inherit" older gen flags.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25086>
Virtio backend backend was not updated because it was not
compiled in CI and not compiled locally.
Fixes: 7f59e37233
("tu/perfetto: Allow gpu time to be passed into tu_perfetto_submit")
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25531>
Raw GPU time is retrieved via kgsl_cmdbatch_profiling_buffer,
offseted GPU time is retrieved via KGSL_PERFCOUNTER_GROUP_ALWAYSON.
This allows to calculate GPU time offset for each submission and
synchronize CPU/GPU time domains.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12805>
In preparation to support perfetto on KGSL, on KGSL GPU time is
retrieved on submission and requires minimal post-processing.
Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12805>
