gallium: plumb resolve attachments through from frontends -> pipe_framebuffer_state

some drivers may find this useful

Reviewed-by: Adam Jackson <ajax@redhat.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18695>

radv/ci: add missing expected failures with RADV_PERFTEST=gpl on GFX1100

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21334>

hasvk: use Vulkan runtime's robust buffer access

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21338>

anv: use Vulkan runtime's robust buffer access

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21338>

frontends/va: Map VAEncCodedBufferType buffer as PIPE_MAP_READ

VAEncCodedBufferType is used for reading back encoded data.
Mapping it for read instead of write speeds up reading
the data on CPU.

On radeonsi this will result in VRAM copy to staging buffer
in cached GTT, making the CPU read much faster.

Signed-off-by: David Rosca <nowrep@gmail.com>
Reviewed-by: Ruijing Dong <ruijing.dong@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20376>

fix: clover/llvm: replace llvm::None with std::nullopt for LLVM 17+

llvm::None was deprecated and builds started failing with

  error: ‘None’ is not a member of ‘llvm’

Instead of using the temporarily available include in ADT which would
add a deprecation warning to the build, directly replace llvm::None with
the recommended std::nullopt

This change takes only effect with LLVM 17 or newer.

Reference: https://github.com/llvm/llvm-project/blob/d4f38ef288c3a4cf2318182c8585a5c7e760877a/llvm/include/llvm/ADT/None.h
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21125>

ir3: Consider dst type in ubo_vec4 to ldc lowering

The dst type could be either 16b or 32b.

Fixes validation failure in dEQP-VK.subgroups.* tests which deal with
16b types.

 validation fail: (type_size(instr->cat6.type) <= 16) == !!((instr->dsts[0])->flags & IR3_REG_HALF)
   -> for instruction: MESA: info: 0023:0000:000:        ldc.offset0.base0 hssa_23 (wrmask=0x3), ssa_1, ssa_22

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21316>

pvr: Add support for VK_ATTACHMENT_LOAD_OP_LOAD.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21280>

pvr: Upload spm load programs to device.

The programs are currently unused but will be needed for the spm
background object load op.

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21279>

pvr: Add SPM load usc empty programs

Signed-off-by: Karmjit Mahil <Karmjit.Mahil@imgtec.com>
Reviewed-by: Frank Binns <frank.binns@imgtec.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21279>

iris: Drop iris_cache_flush_for_render

Before dropping this function, handle the two callers of this function:

* The call in iris_blorp.c is redundant. The required cache flushes are
  already handled by the callers of blorp functions. Delete this.

* The call in iris_resolve.c is still providing a benefit because it
  calls iris_emit_buffer_barrier_for internally. Inline the needed
  barrier.

Cc: 23.0 <mesa-stable>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21303>

iris: Flush caches for aux-mode changes more often

Memory accesses can get corrupted when there's a disagreement between:
* the aux-mode of existing cache lines for a surface and
* the aux-usage in that surface's RENDER_SURFACE_STATE object

We have already prevented hardware from seeing this conflict for
rendering operations, but due to how the L3 is shared among multiple
clients in gfx12 (e.g., sampler engine, render engine, etc.), we need to
expand the scope of the existing solution. Now, before any access of a
compressible resource, we make sure to flush the prior aux-mode from the
caches.

The majority of changes here refactor things for use in a new function,
flush_previous_aux_mode. The remaining change calls that function from
within iris_resource_prepare_access.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6558
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7625
Cc: 23.0 <mesa-stable>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21303>

iris: Update comment in iris_cache_flush_for_render

Update the comment to reflect the fact that iris no longer switches
between CCS_E and CCS_D.

Cc: 23.0 <mesa-stable>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21303>

meson: Combine duplicated c_args and cpp_args

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19185>

meson: Split c_cpp_args from pre_args

pre_args should not include compiler options

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19185>

meson: drop `TODO: opengl`, it's done

Suggested-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

meson: only build glsl when needed

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

meson: only build libglsl_util when needed

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

meson: only build the loader when needed

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

meson: only build mapi when needed

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

meson: move float64_glsl_file one meson.build up

anv uses it.

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19600>

lavapipe: enable linear filtering for depth formats

this seems to work according to cts

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20859>

intel: Implement Wa_16011448509

"Use 3DSTATE_CONST command for individual shaders instead of
3DSTATE_CONST_ALL COMMAND"

On gen 12.0 platforms, 3DSTATE_CONSTANT_ALL command is not processed
correctly in certain cases.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
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/21301>

zink: block LINEAR filtered blits for zs formats

this is illegal, and the u_blitter path has to be taken to guarantee
enough accuracy that the strictest piglit tests pass

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21314>

Revert "zink: always use NEAREST for zs blits"

This reverts commit 067545eb9af79414054e4bef3ab48af003758663.

this is good enough for CI but not enough for anholt's piglit tests

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21314>

vk/runtime: turn vk.xml extension requirements into asserts

More specifically, turn
  <extension name="VK_KHR_foo" requires="VK_KHR_bar">
into
  assert(!ext->KHR_foo || ext->KHR_bar);

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21261>

vk/runtime: keep track of supported instance extensions

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21261>

vk/util: keep track of extension requirements

Signed-off-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21261>

dzn: Enable KHR_storage_buffer_storage_class

It's required for VK1.1 and Mesa's SPIR-V parser handles it for us.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21261>

intel/fs: bound subgroup invocation read to dispatch size

This is to avoid out of bound register accesses (potentially leading
to hangs) when the dispatch size is smaller than when is reported in
the NIR subgroup_size.

v2: Implement bounding with a mask (since workgroup sizes are powers of 2) (Faith)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 530de844ef4d ("intel,anv,iris,crocus: Drop subgroup size from the shader key")
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21282>

zink: enable renderpass optimizations by default for selected drivers

CI has been running this for months on turnip, so let's give it some
broader testing

also add a ZINK_DEBUG=norp for testing

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21296>

aco: Don't use vcmpx with DPP.

V_CMPX+DPP returns 0 with reads from disabled lanes, unlike V_CMP+DPP (RDNA3 ISA doc, 7.7)

Fixes: baab6f18c91 ("aco: Optimize branching sequence during SSA elimination.")
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20537>

nir/from_ssa: Use more helpers in resolve_parallel_copies

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21299>

nir: Fix typos in the from-SSA pass comments

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21299>

nir: Fix merge_set_dump() to compile again

This #if 0'd debug code has been broken since -Werror=vla was added.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21299>

nir: Print divergence information for registers as well as SSA defs

This patch causes us to print "con" and "div" for registers as well as
SSA defs.  We print it on both register declarations, and destinations.
The latter isn't strictly necessary, but it is handy to be able to see
e.g. a convergent value being assigned to a divergent register without
having to constantly refer back to definitions that might be much
earlier in the program.  I originally printed it for sources as well,
but that got to be a bit wordy, so I dropped that.

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21299>

meson: remove unused USE_FOO_ASM defines

The usage of these defines was removed, so let's remove the definitions
as well.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21212>

anv/blorp: use existing function to convert the op to a string

Signed-off-by: Rohan Garg <rohan.garg@intel.com>
Reviewed-by: José Roberto de Souza <jose.souza@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21283>

isl: fix some documentation

Signed-off-by: Rohan Garg <rohan.garg@intel.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21283>

anv: reuse the VK_IMAGE_ASPECT_PLANES_BITS_ANV macro

Signed-off-by: Rohan Garg <rohan.garg@intel.com>
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21283>

anv: drop unused headers

Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21283>

aco: new 16bit VOP3 opcodes can use opsel

No Foz-DB changes on gfx11.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20705>

ci/venus: Skip tests risking out of memory issues

These two dEQP tests:
 * dEQP-VK.api.object_management.max_concurrent.graphics_pipeline
 * dEQP-VK.api.object_management.max_concurrent.compute_pipeline

Require a lot of memory and might kill crosvm or the renderer server because of EOM
issues. Skip them for now.

Acked-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21306>

radv: ignore registering pipeline libaries with SQTT

They aren't executable pipelines and they might not contain all
shader stages.

This fixes a crash when generating RGP captures with GPL.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21235>

zink: remove incorrect trailing comma

A trailing comma in a list is not valid JSON. Let's drop it.

Fixes: f7b2dbb2bdc ("zink: relax bresenhamLines requirement for non-strictLine drivers")
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21211>

zink: correct companies in requirements

These contributors don't work for Zink itself, but various companies
who's working on Zink. Add the correct ones.

Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21211>

ci: Enable the hasvk Vulkan driver in the fedora-release job

To match the Fedora packaging.

While at it, sort the Vulkan drivers alphabetically.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

ci: Enable i915 Gallium driver in fedora-release job

To match the Fedora packaging.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

ci: Re-enable intel-clc in fedora-release job

Fedora 36 has new enough llvm-spirv-translator.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

ci: Update Fedora image to 36

Fedora 34 is EOL.

With Fedora 37, ccache hangs in a futex syscall in CI for some reason,
so let's go with 36 for now.

v2:
* No need to add -Wno-error=stringop-overflow anymore.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

clover: Reserve vector memory in make_text_section

This isn't strictly required, but it works around
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=100366 , and it might
avoid a memory reallocation.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

anv/grl: Use union for reinterpreting integer as float

Fixes strict aliasing violations flagged by GCC 12:

../src/intel/vulkan/grl/include/GRLOCLCompatibility.h: In function ‘float as_float(uint32_t)’:
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h:182:13: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
  182 |     return *reinterpret_cast<float*>(&i);
      |             ^~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h: In function ‘float3 as_float3(int3)’:
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h:187:13: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
  187 |     return *reinterpret_cast<float3*>(&i3);
      |             ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h:187:13: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h: In function ‘float4 as_float4(int4)’:
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h:192:13: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]
  192 |     return *reinterpret_cast<float4*>(&i4);
      |             ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../src/intel/vulkan/grl/include/GRLOCLCompatibility.h:192:13: warning: dereferencing type-punned pointer will break strict-aliasing rules [-Wstrict-aliasing]

Fixes: 5f948503e40c ("anv: Import GRL")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21236>

zink: set ZINK_DESCRIPTORS=db for radv jobs

there's no testing for this in ci, and radv has a unique heap/memory
setup which catches allocation regressions

Reviewed-by: Martin Roukala <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20968>

anv: enable task redistribution

Disabling is no longer needed after "intel/compiler/mesh: use
slice id of task urb handles in mesh shaders".

This reverts commit 4eaecd79657f0cd99f6536fd55a7a14958d6624b.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7141

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21007>

intel/compiler/mesh: use slice id of task urb handles in mesh shaders

When mesh shader is spawned on a different slice than the originating
task shader, then input task urb handle can come from a different
slice, so masking this information off will load data from the current
slice, instead of the one where real data are.

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21007>

anv: implement VK_EXT_vertex_input_dynamic_state

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21026>

anv: move 3DSTATE_VERTEX_ELEMENT emission to dynamic path

Prep work for VK_EXT_vertex_input_dynamic_state

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21026>

anv: remove copied information from runtime graphics state

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21026>

anv: fix how unset gl_Viewport & gl_Layer are handled in mesh case

See also: c6f69eea6ac ("anv/pipeline: Properly handle unset gl_Layer and gl_ViewportIndex")

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17620>

intel/compiler: replace gl_Layer & gl_ViewportIndex by 0 in fs if ms doesn't write it

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17620>

radv: disable DCC for mipmaps on GFX11

It seems broken but can't really figure out why and DCC levels aren't
interleaved on GFX11. Skipping DCC initialization for levels seems to
also fix it but seems safer to disable completely, as a hotfix.

Fixes DCC issues with Hi-Fi Rush, Sonic Frontiers, Hogwarts Legacy
and probably more.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8230
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21186>

ci/etnaviv: Update deqp xfails for gc2000.

More of a mixed bag here than gc7000, but at least the status is stable
in the last 3 nightly runs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21294>

ci/etnaviv: Drop stale xfails from gc7000.

These have passed in 3/3 of the last nightly runs.  Congrats, etnaviv!

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21294>

ci/radv: Add a skip for navi21-llvm for a test that consistently timeouts.

I had seen this in my last round of work, but forgot to git add the skips
file.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21294>

zink: verify compressed format layer count when creating surfaces

this is illegal for some cases, and mesa/st is equipped to handle
per-layer copying here

fixes #8283

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: don't handle mutable init on surface creation with tc enabled

using the cmdbuf during this call is illegal and causes desync, thus
the initialization has to be deferred until the surface is bound

fixes #7579

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: const-ify a surface param

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: break out pipe_surface init for new surface creation

no functional changes

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: account for null surface when trying to retain clears on fb bind

not currently possible but will be soon

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: only try for a fb rebind if fb binds exist in rebind_image()

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

zink: fix indentation of rebind_image()

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21292>

crocus: switch gen4/5 tiling flags to follow suggestions.

Fixes: 6043f66dd332 ("crocus: disable Y tiling for render targets properly.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21300>

broadcom/ci: mark test as flaky

It passed once here: https://gitlab.freedesktop.org/gdevi/mesa/-/jobs/36317041

Signed-off-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21295>

Revert "freedreno/a5xx: Fix clip_mask"

This reverts commit 2dfebf34874b5365156d254c2c1ba2ecc5262deb.

It causes GPU hangs in piglit tests like
spec@glsl-1.20@execution@clipping@vs-clip-vertex-enables, for reasons I'm
totally unclear on.  The commit was not necessary, because the frontend
lowering already handles disabled clip planes by storing 0.0 to the
corresponding clipdist array element in that shader variant.  Add a note
to that effect.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21298>

ci/freedreno: Update manual-run xfails for a530.

While I'm having a hard time stabilizing most of the test list on this HW
due to the clip-enable GPU hangs leaking into random other tests, these
have been consistent in the last 4 runs.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21298>

d3d12: Don't clear d3d12_shader_key

The shader key structure is quite large and memsetting it to zero to be
able to create or often simply find an existing shader is responsible
for a large portion of CPU usage during benchmarks.

This change is more surgical about what, when, and how things get
cleared.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21247>

nir: Check sampler_binding is valid when lowering tex shadow

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21247>

d3d12: Add unions to encompass shader key stage vars, use in hashing

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21247>

d3d12: Use varying comparison function for TESS stage key compare

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21247>

util/disk_cache: Handle OS'es without d_type in struct dirent

Needed to build on Solaris

Fixes: f58e6fee745 ("util/disk_cache: delete more cache items in one go when full")
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21110>

zink: always use NEAREST for zs blits

LINEAR is illegal, and swapping filtering at this point still
seems to be conformant given questionable GL spec requirements for
LINEAR zs filtering

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21203>

radv/rra: Hide deferred accel struct data destruction behind an env var

Keeping around copies of the BVHs in CPU memory can cause issues with
Applications creating a large amount of acceleration structures (Control).
This commit adds back the old path of copying acceleration structures
while still keeping the deferred, possibly more accurate path around.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20700>

radv/rra: Find copy memory index when initializing the trace state

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20700>

anv/hasvk: handle a SAMPLED_READ/STORAGE_READ access flags

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21277>

zink: free resource objects' views array during destruction

since the array is not ralloc managed, it has be explicitly freed.

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21285>

zink: stop creating pipeline library cache for non-optimal_key drivers

currently the nin-optimal codepath doesn't use this at all and this just
leaks memory due to not handling non-tcs generated shaders.

Fixes: 487ac6dbd60 ("zink: implement cross-program pipeline library sharing")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21285>

zink: allocate program shader caches from the program's mem ctx

these will now be freed when the program itself is freed.

cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21285>

zink: fix compute shader leaks

Compute program owns the nir and zink shaders now and must free them
too when destroyed.

Fixes: 4cb4bb555e7 ("zink: create compute programs from compute shaders directly")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21285>

ci/mingw: drop useless -Wno-error flags

Will prevent merging incorrect code like in #8260

Reviewed-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21244>

radv: fix a hang with binning on CHIP_RENOIR

Using (6, 16) for (context_states_per_bin, persistent_states_per_bin)
causes gpu hang in RDR2 benchmark on CHIP_RENOIR.  Follow radeonsi and
use (3, 8) instead.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21254>

virgl: enable timer queries only if host supports it

Timer queries is tied to GL_ARB_timer_query/GL_EXT_disjoint_timer_query
support on the host. We have a flag that detects this so lets use it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20380>

pan/decode: Add support for decoding CSF

Add support to pandecode for Mali architecture v10, featuring the new command
stream frontend (CSF). This replaces the "job chain" with a new Command
Execution Unit (CEU) that runs a domain-specific assembly language. That
requires us to refactor pandecode substantially, splitting out JM-only code from
shared JM/CSF common code, and adding new CSF-only decode routines to
disassemble and interpret CSF command streams and pretty-printing the
data structures hit.

This is of course impossible to do properly, since the CEU is pretty easily
Turing-complete and hence subject to the halting problem. But we implement some
simple heuristics to follow jumps that are just good enough for the simple
command streams emitting by both the DDK and Panfrost.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20837>

panfrost: Fix some fields in v10.xml

Correct some errors from the file's initial check in, as we're about to add
corresponding pandecode changes for the file.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20837>

asahi: Implement indirect draws

Passes dEQP-GLES31.functional.draw_indirect.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21273>

asahi: Refactor index buffer upload for indirect

We need to avoid the reference to draws->count when indirect draws are used, as
it is not available CPU side.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21273>

asahi: Remove default=true on index list values

These will cause issues with indirect draws.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21273>

panvk: Disable SNORM rendering

Driver isn't ready for this yet. 7f98a9ba2bd ("panfrost: Implement
GL_EXT_render_snorm on Bifrost+") caused piles of tests to go from NotSupported
-> Fail, so let's functionally revert that.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21257>

panvk: Fix varying linking

Since 2316b80d77e ("panfrost: Don't use nir_variable to link varyings"), we can
only get correct type information from the fragment shader inputs (not the
vertex shader output). Fixes piles of CTS regressions.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21257>

panvk: Take lock when tracing

We're not supposed to call the GENX(pandecode_jc) routines (e.g.
pandecode_jc_v7), since it's an internal interface that expects the caller to
take a lock first. Instead we're supposed to call the non-GenXML pandecode_jc
entrypoint which does the locking properly. Fixes assertion failures when
tracing with recent pandecode:

deqp-vk: ../src/util/simple_mtx.h:142: simple_mtx_assert_locked: Assertion `mtx->val' failed.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21257>

rusticl/program: enable spirv

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19008>