panfrost: Avoid minimum stack allocations

If stack isn't used, don't allocate it - simple as that.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6373>

panfrost: Introduce invisible pool

Whereas the main batch->pool is CPU read/write, the new
batch->invisible_pool is not. This enables GPU-internal structures that
the CPU must allocate from a pool dynamically but does not read,
corresponding to the BO_INVISIBLE create flag.

The use case is speeding up varying allocation by skipping the
CPU-side mmap/munmap.

We simultaneously half the pool's minimal allocation to avoid negatively
affecting memory usage.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6373>

panfrost: Pre-allocate memory for pool

Seeing random BO creates in the middle of command stream generation is
annoying, let's preallocate memory so CPU profiling is less random.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6373>

iris: Fix headerless sampler messages in compute shaders with preemption

We were failing to set the "Headerless Message for Preemptable Contexts"
bit in SAMPLER_MODE in the compute context.  Other drivers use a single
hardware context, so setting it on the render engine was sufficient to
flip it in both pipelines.  But iris uses a separate hardware context
for compute, so we were only getting these set for the render context.

Thanks to Jason Ekstrand for catching this bug.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6380>

aco: fix file leak in ra_fail()

Fixes: c2b1978aa47 ("aco: rework the way various compilation/validation errors are reported")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6399>

nir/opt_large_constants: Fix a type/deref_type typo

Fixes: df9596353a60 "nir/large_constants: Handle incomplete derefs"
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6403>

gitlab: ask for more detailed info about GPU

PCIID is only printed by lspci/lshw on demand.

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6345>

radv: fix null memcpy and zero-sized malloc

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

radv: align pipeline cache entry and header sizes

Fixes UBSan error:
../src/amd/vulkan/radv_pipeline_cache.c:603:42: runtime error: member access within misaligned address 0x00000152ebcc for type 'struct cache_entry', which requires 8 byte alignment

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

radv: don't pass null to _mesa_sha1_update

This could then pass NULL to memcpy. Fixes UBSan error:
../src/util/sha1/sha1.c:140:8: runtime error: null pointer passed as argument 2, which is declared to never be null

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

nir: fix memory leak in nir_cf_list_clone

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

nir: fix potential left shift of a negative value

Fixes UBSan error:
src/compiler/nir/nir_constant_expressions.c:36573:32: runtime error: left shift of negative value -1

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

util: add a alignof() macro

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6206>

gitlab: ask inxi output to be in code blocks

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Acked-by: Danylo Piliaiev <danylo.piliaiev@globallogic.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6365>

radv,aco: report ACO errors/warnings back via VK_EXT_debug_report

To help developers.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6318>

aco: rework the way various compilation/validation errors are reported

The upcoming change will allow to report all ACO errors (or warnings)
directly to the app via VK_EXT_debug_report. This is similar to what
we already do for reporting various SPIRV->NIR errors.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6318>

aco: rename DEBUG_VALIDATE to DEBUG_VALIDATE_IR

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6318>

anv: null check for buffer before reading size

v2: move assignment under existing check (Nanley)

Fixes: b9a05447a19 ("anv: dynamic vertex input binding stride and size support")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6369>

util: Explicitly call the unpack functions from inside bptc pack/unpack.

We were calling the table-based unpack functions from inside the pack and
unpack table's methods, so if anything included these pack functions (such
as a call to a table-based pack function), you'd pull in all of unpack as
well.

By calling them explicitly, we save some overhead in these functions
(switch statement, address math on the zero x,y arguments) anyway.

Reviewed-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6307>

util: Expose rgba unpack/fetch functions as external functions as well.

Like we just did for pack functions for freedreno, it will be useful to be
able to pick out a specific rgba unpack function instead of going through
the table.

Reviewed-by: Jonathan Marek <jonathan@marek.ca>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6307>

iris: Drop buffer support in resource_from_handle

The callers don't seem to pass targets of PIPE_BUFFER. Stop nesting an
if-else block by dropping support for this target.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

iris: Add and use iris_resource_configure_main

Reduce duplication of ISL surface creation code for main surfaces. This
is similar to iris_resource_configure_aux.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

iris: Move size/offset calculations out of configure_aux

Not all callers want the result of the offset and size calculations.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

iris: Drop old comment on clear color BO allocation

Iris hasn't used a clear color BO on gen9 since commit 1cd13ccee7b.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

iris: Drop unused resource allocation optimization

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

iris: Drop iris_resource_alloc_separate_aux

When this function was called, the res->mod_info->aux_usage parameter
was ISL_AUX_USAGE_NONE. As a result, no aux buffer was ever allocated.

Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6311>

egl: drop another indentation level in _eglFindDisplay() by inverting an if

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6361>

egl: drop an indentation level in _eglFindDisplay() by replacing break/if with a goto

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6361>

gitlab-ci: fix quoting of variables passed down to bare-metal runners

Eric Anholt identified the issue when merging one of my MRs: the
variable contained words in '`' backticks, which caused them to be
executed by the bare metal runner's shell.

Quote the value printed using bash's shell expansion feature to make
sure anything in the future will be properly quoted.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6389>

gitlab-ci: fix testing whether a variable with a given name is set or not

The previous code considered unset variables the same as set-but-empty;
sometimes setting a variable as something empty is meaningful, so let's
pass them through properly.

Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6389>

docs: update calendar and link releases notes for 20.1.6

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6394>

docs: add release notes for 20.1.6

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6394>

v3d/compiler: add v3dv_prog_data_size helper

Main use case is to help to implement Vulkan PipelineCache, as we are
serializing/deserializing the prog_data too.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6078>

compiler/types: Allow interfaces in get_explicit_type_for_size_align

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6210>

nir/large_constants: Handle incomplete derefs

This pass works entirely with variables, all we have to do is ignore any
derefs we see which we can't chase back to the variable.  The one
interesting case we have to handle is if we have a complex use of a
deref_var.  In that case, we have to flag it non-constant.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6210>

nir/find_array_copies: Handle cast derefs

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6210>

nir/builder: Add a nir_iand_imm helper

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6210>

nir: Initialize nir_ssa_def::live_index

Previously, this was left uninitialized.  Let's initialize it to an
obviously bogus value so we notice if anyone ever tries to use stale
liveness data.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6210>

clover/nir: Use the correct address mode for shared

Shared memory needs to have 64-bit pointers but we want 32-bit offsets
most of the time.  This is exactly what 32bit_offset_as_64bit is for.

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

clover/nir: Stop computing the global address format twice

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

clover/nir: Stop setting ubo_addr_format

We unconditionally set constant_as_global = true so we should never get
UBO access out of spirv_to_nir.

Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nv50/ir/nir: support load_work_dim

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nvc0: handle nr being 0 in nvc0_set_global_bindings

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nv50/ir/nir: fix global_atomic_comp_swap

Fixes: 20d0ae464c4a ("nv50/ir: implement global atomics and handle it for nir")
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

clover/nir: support int64 atomics if the device supports it

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nv50/ir/nir: assert on unknown alu ops

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nve4: fix uploading unaligned sized input buffers

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Pierre Moreau <dev@pmoreau.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

clover/nir: Call vars_to_explicit_types for shared memory

This is required for shared memory buffers declared with an explicit
size inside the kernel.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

nir/glsl: Add glsl_get_cl_type_size_align helper

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6367>

freedreno/ir3: Fix assertion failures dumping CS high full regs.

The *2 here would bump into the *2 in regset, causing assertion failures
dumping CS programs.  Just set the mergedregs flag on a6xx, and don't
duplicate the mergedregs logic.  If you're dealing with new HW where we
don't know if mergedregs is set, you may need to tweak the flag during
disasm setup for the stats to make sense.

Fixes: f7bd3456d79a ("freedreno: deduplicate a3xx+ disasm")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6323>

freedreno/cffdec: When .mergedregs is set, don't count half regs.

This matches what ir3.c does in the mergedregs case: just count max full
reg used.  This flag is unset so far, but will be soon and keeps our
output comparable between blob and freedreno.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6323>

freedreno/ir3: Fix compiler warning from the setjmp fails path.

The TRY() macro doesn't call the contents if we fail to set up
setjmp/longjmp.

Fixes: 3d6e4a201af0 ("freedreno/decode: try harder to not crash in disasm")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6323>

nir/lower_input_attachments: Support loading layer id via gl_ViewIndex

This is required on adreno when the special multiview mode is switched
on.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

tu: Use an input for the layer when lowering input attachments

Also remove a hack that's no longer needed. This should fix input
attachments with layered rendering.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

radv: Use an input for the layer when lowering input attachments

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

nir/lower_input_attachments: Support loading layer id as an input

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

nir/lower_input_attachments: Refactor to use an options struct

While we're at it, fold the details of how to load the fragcoord into
load_fragcoord().

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

nir/spirv: Add the option to keep ViewIndex as an input

This is necessary for fragment shaders on adreno.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5719>

Revert "gitlab-ci: reuse container_post_build when building the test images"

ccache is installed as ephemeral. Hence, the build will fail if used
by the post_build script after removing it.

This reverts commit 40ab6d77c06fe5c0882fe3320583a6a2b1355eab.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6391>

freedreno/a6xx: fix occlusion query with more than one tile

We need to emit epilogue after each tile, not just after the last tile.

Fixes: 13fc03f4c0e ("freedreno/a6xx: Avoid stalling for occlusion queries")
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6376>

freedreno/a6xx: only generate streamout for draw pass shader

Previously we were doing this both for draw and binning pass.. resulting
that the stateobj contained first the incorrect state (based on binning
pass shader with varyings DCEd) followed by the correct state.

Also, in the streamout case we should link binning pass VS against the
real FS, rather than the dummy FS, so that OUTLOC's match (since same
streamout stateobj is used for both draw and binning pass).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6376>

freedreno/registers: SC_WAIT_WC is not a6xx

I think this is probably only a2xx, but it was masking
WRITE_PRIMITIVE_COUNTS on a6xx.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6376>

nir/opt_remove_phis: optimize out phis with undef

This removes some phis that loop unrolling can create.

fossil-db (Navi):
Totals from 349 (0.26% of 135946) affected shaders:
SpillSGPRs: 52 -> 41 (-21.15%)
CodeSize: 3179968 -> 3169920 (-0.32%); split: -0.33%, +0.01%
Instrs: 626965 -> 624712 (-0.36%); split: -0.37%, +0.01%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6348>

egl/x11_dri3: implement EGL_KHR_swap_buffers_with_damage

Passes all of `dEQP-EGL.functional.swap_buffers_with_damage.*`:

    Passed:        36/54 (66.7%)
    Failed:        0/54 (0.0%)
    Not supported: 18/54 (33.3%)
    Warnings:      0/54 (0.0%)
    Waived:        0/54 (0.0%)

The "not supported" ones are the `preserve_buffer_*` tests, which is not
supported on X11/DRI3.

Cc: 20.2 <mesa-stable>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3030
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6132>

egl/x11_dri3: enable & require xfixes 2.0

Cc: 20.2 <mesa-stable>
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Acked-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6132>

ci: Run deqp-gles2 on RadeonSI

We have one more DUT in the Collabora lab, use it to run GLES2 tests.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6385>

pan/mdg: Fix auxiliary load/store swizzle packing

It needs to respect the existing swizzle, as well as the type size.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

pan/mdg: Fix printing of r26 ld/st sources post-RA

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

pan/mdg: Identify barrier out-of-order field

All barrier types are implied regardless, this is an optimization for
out-of-order.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

pan/mdg: Handle 32-bit offsets from store_shared

Implicitly zero-extended for the 64-bit address.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

pan/mdg: Ensure barrier op is set on texture

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Fixes: 92c808cd475 ("pan/mdg: eliminate references to ins->texture.op")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

panfrost: Fix shared memory size computation

Based on core count. Also, avoid some of the more complex programming
and stick to powers-of-two for now.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

panfrost: Fix WRITES_GLOBAL bit

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Fixes: dce7722ef89 ("panfrost: Handle writes_memory correctly")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6321>

android: freedreno: Another build fix

During build on Android 10, build error occurred:

'''
[ 26% 456/1718] Gen Header: libfreedreno_registers_32 <= a3xx.xml.h
FAILED: out/target/product/pinephone/gen/STATIC_LIBRARIES/libfreedreno_registers_intermediates/registers/adreno/a3xx.xml.h
/bin/bash -c "PATH=/usr/bin:\$PATH python3 external/mesa3d/src/freedreno/registers/gen_header.py external/mesa3d/src/freedreno/registers/adreno/a3xx.xml > out/target/product/pinephone/gen/STATIC_LIBRARIES/libfreedreno_registers_intermediates/registers/adreno/a3xx.xml.h"
Traceback (most recent call last):
  File "external/mesa3d/src/freedreno/registers/gen_header.py", line 470, in <module>
    main()
  File "external/mesa3d/src/freedreno/registers/gen_header.py", line 446, in main
    xml_file = sys.argv[2]
IndexError: list index out of range
'''

Align build rules with meson fixes it.

Fixes: 62ebd342 ("freedreno/registers: split header build into subdirs")
Signed-off-by: Roman Stratiienko <r.stratiienko@gmail.com>
Acked-by: Rob Clark <robdclark@gmail.com>
Acked-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6170>

radv: restrict exported symbols with static llvm

Like the gallium --version-script magic but for radv.

The long term goal is to make LLVM support optional, remove it even, so
let's keep the hunk in an if block.

v2: fold if checks (Eric)
v3 (Tomeu): Remove spaces within [] (Dylan)

Signed-off-by: Emil Velikov <emil.velikov@collabora.com>
Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl> (v1)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6203>

freedreno: Destroy syncobj too when destroying fence

This is a missing part of freedreno EXT_semaphore support,
recently merged as part of series adding EXT_external_objects. This hunk was
actually lost due to a mistake on my side while doing a rebase.

Reviewed-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6383>

gitlab-ci: reorder container_post_build call for arm64_test image

And remove a redundant call to ccache --show-stats

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6343>

gitlab-ci: reuse container_post_build when building the test images

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6343>

panfrost: Hoist assert from bind to create

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Remove useless comment

Less than useless, actually - it's not even factually correct.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Hoist add_fbo_bo call

Move it to batch initialization, to take it out of the clear/draw path.
This should make sense - as soon as the batch is created, we need to
hold the refference. Functionally there shouldn't be unused batches
floating around anyway.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Drop QUADS primitive convert

Even Bifrost retains QUADS support.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Drop panfrost_invalidate_frame

Only used to initialize active_queries once at context initialization,
then unneeded.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Drop ZSA null checks in draws

Likewise.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Drop rasterizer null checks in draw calls

rasterizer must be non-null when draw_vbo is called (indeed, we assert
this there), so anything in the per-draw graphics path can safely assume
we have a rasterizer object bound.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

panfrost: Decontextualize rasterizer

Tightens the scope.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6327>

radv: track and report if a logical device is lost

This currently covers two situations where it's obvious that
the GPU hung:

1) when wait-of-idle doesn't finish in a finite time
2) when a CS submission is cancelled by the kernel

There is still probably some other situations that aren't yet handled.

According to the Vulkan spec, some operations should return
VK_ERROR_DEVICE_LOST when the corresponding logical device is
known to be lost.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/5878>

ci: Update kernel used in LAVA to 5.8-based drm-misc

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6371>

ci: Fix URL to imagediff page in traces dashboard

It has been updated recently.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Reviewed-By: Rohan Garg <rohan.garg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6185>

ci: Test Panfrost with more traces

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Reviewed-By: Rohan Garg <rohan.garg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6185>

ci: Split traces.yml file per driver

As drivers have been tested with more and more traces, the yml file is
becoming a bit unwieldy. As more drivers are going to be tested with
traces, and more traces will be used, split them in per-driver files so
the size stays manageable.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Reviewed-By: Rohan Garg <rohan.garg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6185>

llvmpipe/nir: lower some more packing at nir level.

Fixes:
dEQP-VK.glsl.builtin.function.pack_unpack.*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

llvmpipe: lower mul 2x32_64

Fixes:
dEQP-VK.glsl.builtin.function.integer.umulextended*
dEQP-VK.glsl.builtin.function.integer.imulextended*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

llvmpipe: lower cs local index from id

Fixes:
dEQP-VK.glsl.atomic_operations.*_shared

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

gallivm/nir: split tex/sampler indirect offsets

vulkan has these separate, should be fine for GL as well as
the values will be the same anyways.

Fixes:
dEQP-VK.binding_model.shader_access.secondary_cmd_buf.uniform_texel_buffer.*
dEQP-VK.binding_model.descriptorset_random.sets4.noarray.ubolimitlow.sbolimitlow.sampledimglow.outimgtexlow.noiub.uab.vert.noia*
dEQP-VK.binding_model.descriptor_copy.graphics.uniform_texel_buffer.*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

gallivm/nir: fixup compact TCS variable storage.

This fixes a lot of tessellation shaders, since tess factors
get emitted with spir-v now.

Fixes a bunch of:
dEQP-VK.tessellation*
along with 7000 others.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

gallium/nir/tgsi: add support for compact variables

vulkan tessellation uses compact variables for tess factors,
and trips over in the nir->tgsi info gathering, add support
to the info extraction for compact inputs and outputs.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

llvmpipe: lower uadd_carry/usub_borrow.

Fixes:
dEQP-VK.glsl.builtin.function.integer.uaddcarry.*
dEQP-VK.glsl.builtin.function.integer.usubborrow.*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

gallivm/sample: fix lod query on array textures.

The lod query doesn't take a layer, but the code tries to use one,
detect lodq and don't use a layer in those cases.

There appears to be no GL tests for this behaviour, but the vulkan
CTS hits it.

Fixes:
dEQP-VK.glsl.texture_functions.query.texturequerylod.*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>

gallivm/nir: add imod support

Just adds support for the imod instruction

Fixes:
dEQP-VK.glsl.operator.binary_operator.mod.*

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6339>