radv: Make fill_buffer_shader non-static

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15648>

radv: Use VAs in radv_meta_buffer.c

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15648>

ci: Always install glslangValidator for build jobs

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15648>

pan/bi: Simplify bi_propagate_pass_flags

Worklist-based dataflow analysis is overkill, we can just do DFS.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16045>

meson: Ignore unused variables in release builds

Variables that are only used for assertions are considered unused in release
builds. Don't treat this as an error, since we build with -Werror even for
release in CI. This causes reasonable code to build and pass tests locally (and
therefore to be queued for merge by reasonable developers), but later fail in CI
due to a variable used only as an assertion. This pattern is common enough we
have an ASSERTED macro to workaround the behaviour, but failing a CI run to
have the author go back and add in the ASSERTED and re-queue later is a recipe
for frustration, wasted time, and wasted CI bandwidth.

Disable this behaviour to reduce CI friction.

In my view, sprinkling in ASSERTED clutters the code, rather than helps; I find
CI's insistence on doing so actively counterproductive. Developers are free to
continue doing so after this change. But this way CI won't fail merge requests
over it. After all, CI enforces policy, and we shouldn't have "mark variables
only used for assertions as ASSERTED" as policy. Let's pick our battles wisely
and improve CI's signal-to-noise ratio.

As an added benefit, this eliminates a class of defects where ASSERTED is used
incorrectly, e.g:

   c91e3c6a428 ("util: Should not use ASSERTED in util_thread_get_time_nano")
   3e22fc27af9 ("zink: remove incorrect ASSERTED macro")
   0d08ce287bf ("pan/bi: Remove dated ASSERTED properties")

Note that actual unused variables will be caught by debug builds. It is expected
that developers do debug builds locally before ramming code through CI, so that
should be caught.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Suggested-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Acked-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15582>

ac/gpu_info: remove old and unused fields from radeon_info

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac,radeonsi: rework how scratch_waves is used and move it to ac_gpu_info.c

The addition of the "compute" parameter is for a future change.

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: rename tbuffer -> buffer where tbuffer instructions are not generated

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: don't use tbuffer_store as a fallback for swizzled stores

This depends on the offset computation fix from:
   "ac/llvm: remove inst_offset parameter from ac_build_buffer_store_dword"

v2: The instruction type is changed to MUBUF, which requires us to clear
    DATA_FORMAT with ADD_TID_ENABLE.

Reviewed-by: Mihai Preda <mhpreda@gmail.com> (v1)
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> (v1)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove unused code from ac_llvm_build.c/h

One forward declaration didn't have an implementation.

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove immoffset parameter from ac_build_tbuffer_load_byte/short

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove inst_offset parameter from ac_build_buffer_load

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: use soffset correctly in ac_build_buffer_load

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove inst_offset parameter from ac_build_buffer_store_dword

There was a bug that inst_offset was added to soffset in one codepath and
to voffset in all other codepaths. The correct behavior is to add it
to voffset.

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove immoffset parameter from ac_build_tbuffer_load

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove immoffset parameter from ac_build_tbuffer_store

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: add AC_WAIT_EXP for ac_build_waitcnt

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac: add more non-shadowed registers to the lists

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/surface/tests: generalize and extend gfx10 tests

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

amd: document chips

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15966>

ac/llvm: remove LLVM pass ac_optimize_vs_outputs

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: assign param export indices before compilation

This moves the logic out of LLVM-specific codepaths.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

ac: add ac_nir_optimize_outputs, a NIR version of ac_optimize_vs_outputs

ac_optimize_vs_outputs is an LLVM IR pass, and it will be replaced by this.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: move si_shader_info::stage into si_shader_selector

This will help me see all places where we use "info", which will
be moved from si_shader_selector to shader variants.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: move most "info" fields from si_shader_selector into si_shader_info

It's where they should be, and future commits might require this.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: gather pipe_stream_output_info from NIR intrinsics

This stops pipe_stream_output_info from create_*s_state context functions
because NIR contains everything and can do more advanced shader linking
this way.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: apply key.ge.opt.kill_{outputs,pointsize,clipdistance} in NIR

This may be needed by ACO, but it doesn't do anything for LLVM yet other
than making the initial LLVM IR smaller.

It will be needed by a future commit, which rewrites ac_optimize_vs_outputs
in NIR, which relies on NIR matching the shader key.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: use nir_shader_get_entrypoint in si_nir_scan_shader

The typecast is needed because nir is const.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: add pipe_stream_output_info into si_shader_context

to reduce deltas for the next commit

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: scalarize IO instructions

The next commit will rely on this.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

radeonsi: run nir_io_add_const_offset_to_base for TES/TCS as late optimizations

Other stages don't have indirect indexing, so it's always const.
Doing it here should also remove dead load_const instructions.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14414>

anv: fix INTEL_DEBUG=sync

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 3684012770b2 ("anv: implement DEBUG_SYNC")
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16101>

radeonsi: ignore AMD_DEBUG=nodcc if the texture is imported

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: dump shaders after si_shader_binary_upload to fix printed LDS stat

lds_size is set by si_shader_binary_upload, so it always printed zero.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: don't treat skipped GS copy shader compilation as successful

We don't expect the GS copy shader to ever use the scratch buffer,
so we just don't compile the shader, but the problem is we set ok to
true anyway.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: handle an out-of-bounds sampler/image index better

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: fix ALPHA_IS_ON_MSB determination

based on PAL

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: initialize line stipple config registers

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: set PA_SC_EDGERULE according to bottom_edge_rule state

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: add an SQTT workaround for chips with disabled RBs

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: don't compile monolithic PS for dual source blending

The code was compiling monolithic PS if a shader output didn't
have a color buffer, but dual src blending never has a color buffer
for mrt1.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

radeonsi: set more state bits for disabled binning

based on PAL

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15965>

dzn: Fix the STATIC_ASSERT() in dzn_meta_blits_get_context()

We care about the object key size, not the size of a pointer.

Fixes: a012b219640 ("microsoft: Initial vulkan-on-12 driver")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16109>

dzn: Pass the dzn_event pointer to _mesa_hash_table_insert()

The key is supposed to be the dzn_event pointer, not the VkEvent
object, even if they do match in practice on 64bit builds.

Fixes: a012b219640 ("microsoft: Initial vulkan-on-12 driver")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16109>

dzn: Add missing VKAPI_{ATTR,CALL} specifiers to BeginCommandBuffer()

Fixes: a012b219640 ("microsoft: Initial vulkan-on-12 driver")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16109>

ci: skqp: upgrade to skqp based on Android CTS 11.0 (r7)

Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15987>

ci: skqp: patch skqp to report also GL results

Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15987>

anv: Clean up pipeline cache helpers a bit

Instead of having two different helpers, delete the pipeline_cache ones.
Also, instead of manually handling the cache == NULL case in every
vkCreateFooPipelines call, handle it inside the helpers.  This means
that BLORP can use them too by passing cache=NULL.

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

anv: Switch to the new common pipeline cache

This patch is intended to be somewhat minimal.  There's a lot of cleanup
work that can be done but we'll leave that to later patches.

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

anv: Rename a fail label in CreateDevice

The rest of them are labeled with the thing they need to destroy first,
not the thing that failed.

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

vulkan/pipeline_cache: Add helpers for storing NIR in the cache

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

vulkan: Add a common VkPipelineCache implementation

This is partly copied+pasted from ANV but is mostly new code with lots
of reference counting bugs fixed (I hope!).  The new cache caches
"object" which derive from a base vk_pipeline_class_object struct.  It
uses a kernel-style "ops" interface for virtual methods on these objects
to allow for easy destruction (when the reference count hits zero) as
well as serialization an deserialization interfaces.  This should allow
drivers to cache basically whatever they want without having to think
too hard about the details.

Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

util/set: Respect found in search_or_add_pre_hashed

Fixes: 491e7decad0f "util/set: add the found param to search_or_add"
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

util/blob: Add align helpers

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13184>

ac/surface: fix an addrlib race condition on gfx9

Addrlib calls GetMetaEquation, which generates and saves address equations
in a global table that is not thread safe.

Fixes: df2cbdd2e32 - amd/addrlib: expose DCC address equations to drivers
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6361

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16091>

nir: fix an uninitialized variable valgrind warning in nir_group_loads

pass_flags is only initialized for grouped loads, so change the order

Fixes: 33b4eb149ea - nir: add new SSA instruction scheduler grouping loads into indirection groups

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16090>

ac/surface: adjust gfx9.pitch[*] based on surf->blk_w

This is the same as 8275dc1ed57, but since gfx9.pitch[...] is
used for linear surfaces since 86262b6eac0 we need to update
it as well.

Fixes: 86262b6eac0 ("radeonsi,radv: fix usages of surf_pitch")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16053>

util: remove simple_list

Down to two invasive linked list implementations in Mesa!

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

llvmpipe: replace uses of simple_list.h with list.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

llvmpipe: cleanup trailing whitespace

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/draw: use list.h instead of simple_list.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/draw: remove trailing whitespace

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/util/u_cache: replace simple_list.h with list.h

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/util/u_cache.c: remove trailing whitespace

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

util/list.h: add a function to move an item in a list

This allows for a 1:1 replacement of simple_list move_to_head (though
I've tried to make this function more generally useful.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/driver_trace: remove simple_list.h

Which isn't actually used

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

gallium/rbug: replace simple_list.h with list.h

Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

r300: remove unused simple_list.h

It's unused, so let's not include it.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

winsys/radeon: remove unused simple_list.h

It's not used, so let's not include it.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15418>

zink: add synchronization for buffer clears

cc: mesa-stable

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16098>

zink: fix stride=0 for dynamic state vertex stride checking

this case shouldn't trigger the fully-baked fallback since it's
legal with dynamic state

Fixes: 1eada1b02d3 ("zink: selectively disable dynamic vertex stride")

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16098>

vulkan/runtime: repair thread safety in method vk_queue_start_submit_thread()

Accessing queue->submit.thread_run without holding lock vk_queue.submit.mutex.
Elsewhere, vk_queue.submit.thread_run is accessed with vk_queue.mutex
held 1 out of 2 times (1 of these accesses strongly imply that it is necessary).

Found by Coverity.
CID: 1503334

Signed-off-by: Illia Abernikhin <illia.abernikhin@globallogic.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15736>

ac/spm: setup write broadcasting correctly

Based on PAL's PerfExperiment::BuildGrbmGfxIndex method.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi: make sure profile_peak is used before using sqtt

Using sqtt without this profile can cause hangs.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi/sqtt: enable spm counters on gfx10+

Tested on:
  - navi10: L0 cache counter doesn't work (always 0)
  - sienna_cichlid: L0 doesn't work (always 0) and L1 isn't visible in RGP

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi/sqtt: setup spm bo

The hardware will write SPM counters to this bo.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi: add spm counters setup code

Based on radv_spm.c and PAL.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi: add helpers to deal with spm counters

Based on radv_perfcounter.c

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

radeonsi: make si_pc_emit_shaders non static

Also pass "cs" as a param to allow updating a specific cs.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15646>

dzn: Add KHR_draw_indirect_count support

Unfortunately it's not as simple as passing the indirect draw count
buffer to ExecuteIndirect. The compute job that populate the execute
buffer also needs to know the number of entries that need to be
patched. Instead of transitioning the indirect count buffer from
GENERIC_READ to INDIRECT_ARGUMENT we just keep at as a read-only
resource and copy the draw_count value to the exec buffer in the
compute job.

Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15914>

dzn: Allow multiDrawIndirect and drawIndirectFirstInstance

Those are already supported, let's just toggle the switch.

Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15914>

v3dv: duplicate key on hashtable insert

The key is created on stack, so as soon as the function returns this key
is lost, so the inserted key in the hashtable is invalid.

Rather, insert a duplicated version on heap.

This fixes a stack-buffer-overflow when running some Vulkan CTS tests.

Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16083>

gallium: move get_canonical_format hook to pipe_screen

pipe_context shouldn't have functions that return values because this
prevent multithreading.

Move this hook to pipe_screen.

Fixes: 606e42027e6 ("gallium: add hook on getting canonical format")
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16078>

dzn: Get rid of the render pass logic

The core provides generic render-pass -> dynamic-rendering wrappers,
so let's rely on them instead of implementing our own logic.

Suggested-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Add support for dynamic rendering

This is how D3D12 works, so nothing really complicated here. We just
add a bunch of states to the render state, patch the pipeline building
logic to support dynamic rendering, and add new functions to

Suggested-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Inline dzn_cmd_buffer_clear_attachments()

It's only called from dzn_CmdClearAttachments(), so let's move the
code there.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Move all render-related states to a substruct

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Set MAX_RTS to D3D12_SIMULTANEOUS_RENDER_TARGET_COUNT

D3D12_SIMULTANEOUS_RENDER_TARGET_COUNT == 8, so no functional change
here, it's just cleaner to use the D3D12 definition.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Pass an image view to dzn_cmd_buffer_clear_attachment()

Let the caller extract the image view from the attachment index, so
we can make this function framebuffer-agnostic and re-use it in the
dynamic rendering path.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15913>

dzn: Advertise KHR_shader_draw_parameters

We already pass those parameters to shaders, so let's advertise the
feature.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15912>

dzn: Update the draw_id sysval when issuing indirect draws

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15912>

dzn: Simplify the triangle_fan_rewrite_index shader

We don't need the first_vertex argument when turning non-indexed
triangle-fan draws into indexed triangle-list ones, the firstVertex
argument can simply be passed as a vertexOffset.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15912>

dzn: Fix exec_buf_stride calculation

The stride depends on the indirect draw arguments which are different if
triangle-fan emulation is required.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15912>

microsoft/spirv_to_dxil: Add a draw_id field to vertex_runtime_data

So we can support the draw_id system value.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15912>

radv: only apply enable_mrt_output_nan_fixup for 32-bit float MRTs

This is incorrect for 32-bit integer MRTs which are clamped to the
maximum value of the format, and returning 0 can break some shaders.

This fixes a rendering issue with RAGE2.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4329
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/16080>

docs: update calendar for 22.1.0-rc2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16089>

docs: update calendar for 22.1.0-rc1

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16089>

docs: update calendar and link releases notes for 22.0.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16089>

docs: Add sha256sums for 22.0.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16089>

docs: add release notes for 22.0.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16089>