dzn: Support device factories in addition to global device creation

Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18306>

dzn: Store the d3d12 module reference on the instance

Acked-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18306>

zink: set default spec constant value to 1

this is less illegal

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19116>

zink: move spec constant emission to the types/consts block

spec constants are used as constants, so emitting them in the instructions
block can create mismatched ordering of the constant usage with other types

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19116>

intel/compiler: run nir_opt_idiv_const before nir_lower_idiv

Integer div lowering can potentially create a lot of code that is
not removed later on. Running const lowering pass first can be used
to eliminate that code.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19157>

radv: drop the ES vertex count requirement on GFX11

Ported from RadeonSI.

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/19155>

radv: do not use memory for waiting for cache flushes on GFX11

There is a different mechanism with an internal counter.
Ported from RadeonSI.

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/19155>

radv: use correct VGT_TESS_DISTRIBUTION settings on GFX11

Ported from RadeonSI.

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/19155>

radv: set SAMPLE_MASK_TRACKER_WATERMARK to 15 on GFX11

Ported from RadeonSI.

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/19155>

radeonsi,radv/ci: Increase coverage

12 more machines with Stoney Ridge have been added to Collabora's lab.

Use the new ones for testing more of RadeonSI, and repurpose the freed
ones for more RADV coverage.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19161>

ci: Run RadeonSI jobs on changes to src/amd/ci

Now that RadeonSI jobs are defined in src/amd/ci, make sure that those
jobs are run on changes to that directory.

Fixes: a1d34d8c9577 ("ci/amd: move radeonsi in amd directory")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19161>

ac/nir/ngg: use nir_load_provoking_vtx_in_prim_amd in ngg lower

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19166>

nir,ac/llvm,radv: add nir_intrinsic_load_provoking_vtx_in_prim_amd

For radeonsi which load this from arg.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19166>

radv/ci: run two jobs in parallel to reduce the execution time

This should bring the overall execution time at around 15 minutes \o/,
but it exposes more issues in the driver:

 - [...].linear_interpolation.no_offset_8_samples (#7220)
 - Timeouts on some dedicated_alloc.image.full.1048577.* tests
 - Timeout on memory.pipeline_barrier.host_read_host_write

More performance improvements should come later, as the load average
remains in the [12, 16] range when running on my machine on a navi22
gets me a 31 load average and a lower execution time by a couple of
minutes.

References: #7220
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19111>

radv/ci: bump the job count on navi21

We got shiny new hosts for our navi21 which sport twice as many cores
as the previous hosts. Let's try to keep them fed!

Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19111>

r600/sfn: Initialize TestInstrFromString member m_instr_factory.

Fix defect reported by Coverity Scan.

Uninitialized pointer field (UNINIT_CTOR)
uninit_member: Non-static class member m_instr_factory is not
initialized in this constructor nor in any functions that it calls.

Fixes: e840645bb79 ("r600/sfn:explicitly initialize the memory pool")
Suggested-by: Gert Wollny <gert.wollny@collabora.com>
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19098>

util: Move the function implementation from u_thread.h to u_thread.c

Move the platform dependent functions from u_thread.h to u_thread.c,
so that avoid pull unneeded headers when using u_thread.h, including the big windows.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: Remove include of windows.h in rwlock.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: Remove include of windows.h in memstream.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

llvmpipe: #include <windows.h> in lp_rast.c

It's accessed GetExitCodeThread

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: #include <windows.h> in disk_cache_os.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

c11: Getting the macro guard of HAVE_PTHREAD in c11/threads.h to be consistence with util/u_uthread.h

The macro guard of #if defined(_WIN32) && !defined(__CYGWIN__) is comes from yohhoy's implementation
and that's not consistence with util/u_uthread.h, this caused it's hard to understand.
Now we change the behavior that's always rely on how meson detecting HAVE_PTHREAD.

So we always disable detecting of threads on Win32 as it's always included in the kernel32 library to
avoid detecting pthreads in mingw.

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: Remove the guard WIN32_WINNT >= 0x0600 in u_thread.h

We already require Windows Vista or later before this commit;
No need check WIN32_WINNT >= 0x0600

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

meson/mesa/wide: define WIN32_LEAN_AND_MEAN with pre_args

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

c11: Remove the code for Windows XP

We already require Windows Vista or later before this commit;
So the code for Windows XP can be removed.

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: #include "util/macros.h" in u_cpu_detect.h for ATTRIBUTE_CONST

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

util: #include "c11/threads.h" when necessary in simple_mtx.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18425>

anv: adding parsetab.py to the .gitignore for grl

This file is autogenerated by python's ply package and is used during
compilation to cache tables built by it for quicker parsing.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19153>

anv: added proper handling for input argument in intel_clc

That was previously listed on the getopt_long struct but not actually
being used. This makes intel_clc argument processing easier as now
all of its arguments are handled with getopt and anything after the
special argument '--' is passed along to clang to form the final build
command.

Thanks to Dylan Baker for help with changes to the meson file.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19153>

anv: fixing typo on description of output flag for intel_clc

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19153>

anv: add missing separator to help for intel_clc

intel_clc relies on the special argument '--' for getopt to be given so
it knows when to start expecting purely input files or clang arguments.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19153>

anv: reword info flag in intel_clc's getopt to avoid clash

The info keyword was using the same short description that
was listed for input files on the struct for long_options.

Rewording it to 'v' and 'verbose' to be more in line with
expectations.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19153>

mesa: Use util_call_once_data instead of mtx_t init_once_lock

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

egl/dri2: Use util_call_once_data to replace usage of mtx_t glFlushMutex

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

egl/wgl: Use util_call_once_data to replace usage of mtx_t glFlushMutex

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

gallium/auxiliary: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

mapi: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER in stub.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

mapi: Use util_call_once to init exec_mem and mutex instead _MTX_INITIALIZER_NP

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_M in u_debug_refcnt.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER in u_debug_memory.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER in u_debug_stack_android.cpp

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER in u_debug_symbol.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Call function mtx_init on exit_mutex to replace the usage of _MTX_INITIALIZER_NP in u_queue.c

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

util: Remove the need of _MTX_INITIALIZER_NP by using simple_mtx_t/SIMPLE_MTX_INITIALIZER in u_debug_stack.c

This removed the thread-unsafe function initialize_backtrace_mutex.
And remove the _MTX_INITIALIZER_NP that's doesn't belong to c11

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

meson: add dep_valgrind dependency for 'dxil_compiler'

The error when include simple_mtx.h in glsl_types.h:
../src/microsoft/compiler/dxil_function.c
In file included from ../src/compiler/glsl_types.h:37,
                 from ../src/compiler/nir_types.h:36,
                 from ../src/compiler/nir/nir.h:44,
                 from ../src/microsoft/compiler/dxil_signature.h:28,
                 from ../src/microsoft/compiler/dxil_module.h:37,
                 from ../src/microsoft/compiler/dxil_function.h:48,
                 from ../src/microsoft/compiler/dxil_function.c:24:
../src/util/simple_mtx.h:37:12: fatal error: valgrind.h: No such file or directory
   37 | #  include <valgrind.h>

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19154>

d3d12: Release sharing contract in flush_frontbuffer

Just a leak I happened to notice.

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

d3d12: Use ID3D12DeviceConfiguration for root signature serialization

If we were able to create a device factory from a DLL-local D3D12Core,
we want to use that one to do root signature serialization, instead of
going to the globally-configured D3D12Core.

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

d3d12: Attempt to use the Agility SDK

The ideal would be to use a D3D12Core.dll next to the GL DLL, via
ID3D12SDKConfiguration1, but this requires a new D3D12.dll to be
present on the machine, which only very new OSes will currently have.
For older OSes, we can at least try to use ID3D12SDKConfiguration to
set a global Agility SDK override, but that only works in developer
mode and requires the DLL to be at a relative path from the app EXE,
so only try that if we have environment variables telling us to do so.

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

d3d12: Support device factories in addition to global device creation

As explained in a comment, a device factory allows modification of state
like debug layer enablement without touching global state / triggering
device removal on a singleton device, and can even create devices that are
not singletons if the driver supports that.

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

d3d12: Store the d3d12 module reference on the screen

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

d3d12: Move some things from screen late-init to early-init

The GLSL singleton ref was too late, and we could end up with an un-paired
dec-ref if screen init failed. Also the transfer slab init was in the wrong
init function, so we'd re-init that struct if the screen was re-initialized.

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

deps: Update DirectX-Headers req to 1.606.4

Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18305>

anv: enable rayTraversalPrimitiveCulling feature

It is mandatory if VK_KHR_ray_query is supported.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19152>

anv: support VK_PIPELINE_CREATE_RAY_TRACING_SKIP_*

VK_PIPELINE_CREATE_RAY_TRACING_SKIP_AABBS_BIT_KHR and
VK_PIPELINE_CREATE_RAY_TRACING_SKIP_TRIANGLES_BIT_KHR, when specified,
make TraceRay behave as if the corresponding shader flags were set, but
without affecting the value of IncomingRayFlags in shaders.

v2 (Lionel): Improve comments

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19152>

ac/llvm: add missing nir_intrinsic_bindless_image_atomic_{fmin,fmax}

Fix various atomic operations.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19134>

rusticl/api: Interpret `lengths` as a slice of Option<NonZeroUsize>

So far `lengths` as been interpreted as a slice of usize. However, 0 is specified as a sentinel
value signalling that the corresponding string is nul terminated. Since checking for sentinel
values is frequently forgotten, Option types should be preferred if possible.

Option<NonZeroUsize> is layout compatible with usize. The None variant is then represented as 0,
which is exaclty what we need here.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Don't check the program source for nul bytes needlessly

At this point we already know that the Vec we've constructed cannot contain a nul byte. Avoid the
needless scan.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Shrink unsafe block

There's only two unsafe operations left here, so shrink the unsafe block to make them obvious.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Factor pushing to the Vec out of the if

Both branches do the same here.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Fix creating a program if a nul byte is within the specified string length

The code assumed that if the length of a string was specified and greater than zero, the string
would not contain a nul byte.
Legal or not, there are apparently applications which violate that assumption. Since the spec
doesn't say anything about this case, take the likely most compatible route and treat a nul byte as
terminating the string early.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7408
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Drop UTF-8 conversion of input source

The spec doesn't define any charset for the source code. While the vast majority of inputs are
likely to be ASCII (which a subset of UTF-8), it is better not to make assumptions.
As a nice side effect this should be a minor speedup.

`CString`s can't currently be pushed to, so use a `Vec` as intermediate.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

rusticl/api: Use iterators in create_program_with_source

This replaces several instances of unsafe pointer arithmetic and dereferencing with a single unsafe
creation of a slice, which we then use normal iterators on.
The spec mandates that a null pointer is to be interpreted as if a slice filled with zeros had been
given. That case is represented by an infinite iterator returning only zero.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18978>

r300: fix generation of invalid swizzles in regalloc

rc_adjust_channels is inteded for moving the swizzles to a new channels
when rewriting the writemask of an instruction. However for readers one
needs to keep the swizzles in the old channels but rather convert to the
new values, so use the proper helper rc_rewrite_swizzle.

With the new swizzle fixed, we should properly detect that it would be
invalid and thus we can select the proper register class to prevent the
writemask rewrite in the regalloc.

Documentation was added to rc_adjust_channels to make it more clear what
it actually does.

Fixes a bunch of dEQP tests.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7521
Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Reviewed-by: Filip Gawin <filip@gawin.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19158>

nir2dxil: Use native helper lane intrinsic on SM >= 6.6

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19170>

aco: Pre-split result of bvh64_intersect_ray_amd.

Avoids  later moves with extractions from the vector.

Reduces VALU operation in the raytrace loop by ~6%, increasing
the RT performance in Q2RTX on a 6800 XT by about ~1.3%.

Suggested by Georg.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19148>

util/indicies: convert u_unfilled_gen.py to write a file

Which avoids meson needing to wrap the generator to capture the output,
and makes it faster

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

util/indicies: use itertools.product in u_unfilled_gen.py

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

util/indicies: simplify some of u_indicies_gen.py

v2:
  - Give variable more descriptive name to avoid shadowing

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
and avoid shadowing

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

util/indicies: write a file with u_indicies_gen.py

This avoids meson creating a wrapper to redirect stdout, and makes the
generator faster

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

util/indicies: use itertools.product instead of deeply nested loops

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

util/indicies: move common static functions to private header

less duplication is good

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19058>

docs: Extend calendar entries for 22.2 by 2 releases.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19172>

docs: update calendar and link releases notes for 22.2.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19172>

docs: Add sha256 sum for 22.2.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19172>

docs: add release notes for 22.2.2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19172>

frontends/va: fixed an av1 decoding image corruption issue.

problem:
When playing back some clips with loop restoration parameters
enabled, the display image could be corrupted.

solution:
correct loop restoration unit size logic in vaapi interface.

CC: 22.2 <mesa-stable>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Signed-off-by: Ruijing Dong <ruijing.dong@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19146>

zink: add ci flake

fixes #7442

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19160>

tu: Clean up variable usage in `tu6_draw_common`

Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19107>

tu: Only write `A6XX_PC_PRIMITIVE_CNTL_0` if changed

Increases the score in the `draw` test in `vkoverhead` to 71809
from 67170 on a HDK 888.

Signed-off-by: Mark Collins <mark@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19107>

gallium: Default to PIPE_CAP_MIXED_COLORBUFFER_FORMATS

There are three tiers of drivers:

* Drivers that support MRT and support mixed colorbuffer formats. All modern
  hardware fits this as it becomes a spec requirement.
* Drivers that do not support MRT. Then this CAP is a no-op, so we might as well
  set it by default even here (this commit trivially enables the CAP for lima,
  vc4, etanviv).
* Drivers that support MRT but do not support mixed colorbuffer formats! Very
  little hardware fits this category as it doesn't suffice for MRT in most APIs.
  Unfortunately we have a few drivers that are in this category, preventing us
  from bulldozing the CAP altogether.

Given that the CAP only exists for a few legacy drivers, default it to being
enabled to avoid new drivers falling into the trap of forgetting to enable it.
Failing to set this CAP causes failures in
dEQP-GLES3.functional.fbo.completeness.*

Drivers which still do not set this CAP: nv30, r300 (older than r500), virgl in
some cases. r300/r400 is due to a hardware requirement as Emma points out.

v2: Advertise the cap on lima like the commit message claims (delete the special
case).

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19079>

zink: Remove duplicate variable zm.

Fix defect reported by Coverity Scan.

Evaluation order violation (EVALUATION_ORDER)
write_write_typo: In
zm = zm = create_shader_module_for_stage_optimal(ctx, screen, prog->shaders[i], prog, i, state),
zm is written twice with the same value.

Fixes: 325c703624a ("zink: add 'optimal_keys' handling for shader keys")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19086>

pvr: Remove outdated comments.

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/19110>

pvr: Handle pipeline barrier vk_sync.

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/19110>

pvr: Fix overflow before widen warning for rgn headers size calculation.

This fixes CID 1515968.

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/19136>

pvr: Fix possible seg fault on csb copy.

When copying the secondary command buffer's deferred control stream
to the main stream we have to first allocate space in the main
stream. In case the allocation failed we were attempting to
memcpy() to a NULL destination causing a NULL dereference.

This fixes CID 1515977.

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/19136>

freedreno/a6xx: Drop max_scissor tracking

We don't actually really use this on a6xx, since SQE can early-exit IB2
when there are no more remaining primitives in the bin.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/a6xx: Update caps

We should be doing all the 64b lowering.. I think that should be enough
to get us at least glsl400.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/a6xx: Fix primitives-generated query

RBBM_PRIMCTR_7 is pre-clipped, whereas RBBM_PRIMCTR_8 is after clipping.
I believe we want pre-clipping, and this is what tu does.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/a6xx: Fix MAX_GEOMETRY_OUTPUT_VERTICES cap

Limited by the size of PC_PRIMITIVE_CNTL_5.GS_VERTICES_OUT

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/ir3: Lower all the 64b

Just need to enable some existing lowering.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/ir3: Fix clipvertex with GS+tess

If we have both GS and tess, GS is the stage we should run lower_clip_vs
on.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

freedreno/a6xx: Remove unused param

Minor cleanup, make it more clear that we aren't actually emitting in
IB2.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

mesa/st: ARB_vertex_attrib_64bit depend on glsl>=410

It is possible to expose ARB_gpu_shader_fp64 without supporting
ARB_vertex_attrib_64bit.  The supported GLSL version should tell
us if 64b vertex attribs are also supported or not.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

docs: Update freedreno features

Another manual re-sync with reality.

There are a few extension, notably GL_ARB_texture_compression_bptc,
which are exposed but mesamatrix doesn't show as supported.  I can't
really figure out why.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19100>

driconf: add a workaround for Cossacks 3

Wine issue: https://bugs.winehq.org/show_bug.cgi?id=44560

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19147>

driconf: add a workaround for a cursor issue with kwin_wayland

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7351

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19147>

docs: correct spelling of "freedreno"

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19139>

docs: correct spelling of "traveling"

While "travelling" isn't incorrect, we generally prefer US english in the
docs, where "traveling" is the common form.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19139>

docs: correct spelling of "correctly"

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19139>

docs: ie. -> i.e.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19139>