radv: disable dual source blending in more situations

According to PAL, there is more restrictions that RADV doesn't have.

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/19278>

radeonsi: handle sqtt pipeline in shader prefetch

When sqtt is enabled, the shader code lives in the pipeline bo,
not in the shader bo.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi: store the shader gpu adress in si_shader

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/sqtt: simplify condition to determine if sqtt is on

We don't need to load screen->debug_flags because sctx->thread_trace
is already telling us if sqtt is enabled.
Furthermore we can perform this check only for GFX9 because sqtt
isn't supported  currently on older chips.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi: simplify si_prefetch_shaders

Since 93cd96b5237 the only used value of si_L2_prefetch_mode
was PREFETCH_ALL so we can remove some dead code in si_prefetch_shaders.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/tests: update expected results for polaris11

Using the latest released LLVM version 15.0.1.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/tests: update expected results for navi10

Using the latest released LLVM version 15.0.1.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/tests: update expected results for navi21

Using the latest released LLVM version 15.0.1.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/tests: simplify the script

And use exit(0) on success, exit(1) on failure.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/tests: handle flakes file for all test suite

And a generic flakes file that applies to all GPUs.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/sqtt: allow to disable spm counters

And fix rgp capture on GFX9 (where spm was disabled by default).

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/sqtt: re-export shaders in a single bo

RGP expects a pipeline's shaders to be all stored sequentially, eg:

  [vs][ps][gs]

As such, it assumes a single bo is dumped to the .rgp file, with
the following info:
  * va of the bo
  * offset to each shader inside the bo

For radeonsi, the shaders are stored individually, so we may have
a big gap between the shaders forming a pipeline => we can produce
very large file because the layout in the file must match the one
in memory (see the warning in ac_rgp_file_write_elf_text).

This commit implements a workaround: gfx shaders are re-exported as a
pipeline.

To update the shader address, a new state is created (sqtt_pipeline),
which will overwrite the needed _PGM_LO_* registers.

This reduces DeuxEX rgp captures from 150GB+ to less than 100MB.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

radeonsi/sqtt: notify sqtt when we skip a flush

Otherwise we may not produce any data because
si_handle_thread_trace is never called.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18865>

util: #include "c11/threads.h" instead of #include "util/u_cpu_detect.h"  u_format.c

As it does not access any symbols in util/u_cpu_detect.h but accessed symbols in "c11/threads.h"

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19266>

util: Move UTIL_MAX_CPUS from u_thread.h to u_cpu_detect.h

UTIL_MAX_CPUS is not used by u_thread.* anymore after commit
"util: replace UTIL_MAX_CPUS by util_cpu_caps.num_cpu_mask_bits"

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19266>

util: Merge duplicated function prototype in u_thread.h that named with util_barrier*

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19266>

util: Remove non used UTIL_MAX_L3_CACHES in u_thread.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19266>

pvr: Change pvr_get_hw_clear_color() to pack based on accum formats.

`pvr_get_hw_clear_color()` now packs formats based on the accum
format which is how the hw deals with the formats internally and
might not line up exactly with the vk format representation.

E.g. R5G6B5_UNORM_PACK16 uses the U8 accum format so the USC will
internally use 3 bytes (1 per component) to deal with it instead
of 2 bytes which the vk format specifies. On USC EMITPIX, the PBE
will pack the results to 2 bytes using PACKMODE_R5G6B5 resulting
in the final value being in the vk format representation.

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/19218>

pvr: Add PBE accum format size in bytes.

This will be used later on to implement
vkCmdClearAttachments() where we'll need to pick the
appropriate clear color shader based on the dwords used.

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/19218>

docs: vmware -> VMware

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: healthcheck -> health check

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: dependant -> dependent

We generally prefer US English in the docs, so let's correct this.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: virtualisation -> virtualization

We generally perfer US English spelling in our docs, so let's fix up
this one word.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: Linux fbdev -> Linux Framebuffer

It's either "fbdev" or "Linux Framebuffer", not "Linux fbdev". Since we
can spare the extra letters here, let's spell it out.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: servo -> Servo

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: skqp -> SkQP

This seems to be the official styling of the name.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: lava -> LAVA

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: hw -> HW

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: xyzw -> XYZW

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: Gitlab -> GitLab

While we're at it, also fixup a single case of "Ci" -> "CI"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: meson -> Meson

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: ArchLinux -> Arch Linux

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: autotools -> Autotools

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: ppir -> PPIR

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: lima -> Lima

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: vulkan -> Vulkan

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: llvmpipe -> LLVMpipe

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: nouveau -> Nouveau

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: radeon -> Radeon

Let's also do radeonsi -> RadeonSI while we're at it.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: adreno -> Adreno

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: freedreno -> Freedreno

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: lavapipe -> Lavapipe

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: perfetto -> Perfetto

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: rgb -> RGB

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: gpu -> GPU

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: cpu -> CPU

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: pci -> PCI

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: xorg -> X.Org

Also one case of Xorg

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: irc -> IRC

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: ip -> IP

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: use consistent casing for dEQP

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell Haswell in camel-case

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell VirGL in camel-case

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell API all-caps

While we're at it, drop an incorrect apostrophe.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently use CMake

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell OK in allcaps

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell URL in allcaps

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell ANV in allcaps

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently spell DRI in all-caps

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: consistently upper-case VC4, V3D etc

This is how we're doing this elsewhere in the docs, and upper-casing is
a clear indication that this is a name.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: XOR'd -> XORed

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: eg. -> e.g.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: use the official spelling of SHA-1

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: spell out "distribution"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: spell out "environment variable"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: spell out "SHA256 checksum"

This isn't such a magical word, no need to quote it.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: spell out "build directory"

This is a bit less cryptic than "builddir", and it's the same wording as
the meson docs uses.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "Copr"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "Gouraud"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "specifies"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "semantic"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "irrespective"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "passes"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

docs: correct spelling of "addition"

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19191>

gallivm: Fix LLVM optimization with the new pass manager

The previous list of passes contained several errors: "constprop" does not
exist anymore and a trailing ',' is not allowed. This made LLVMRunPasses
fail, but the error is silently ignored. Thus none of the listed passes
ran at all.

https://reviews.llvm.org/D85159 suggests "InstSimplify really should be
used anywhere ConstProp is being used" so replace constprop with
instsimplify and remove the trailing comma.

By enabling pass logging with
LLVMPassBuilderOptionsSetDebugLogging(opts, true),
the difference is visible. Before:

Running pass: AlwaysInlinerPass on [module]
Running analysis: InnerAnalysisManagerProxy<llvm::FunctionAnalysisManager, llvm::Module> on [module]
Running analysis: ProfileSummaryAnalysis on [module]
Running pass: CoroConditionalWrapper on [module]
Running pass: AnnotationRemarksPass on fs_variant_partial (1162 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_partial
Running pass: AnnotationRemarksPass on fs_variant_whole (1110 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_whole

After:

Running pass: AlwaysInlinerPass on [module]
Running analysis: InnerAnalysisManagerProxy<llvm::FunctionAnalysisManager, llvm::Module> on [module]
Running analysis: ProfileSummaryAnalysis on [module]
Running pass: CoroConditionalWrapper on [module]
Running pass: AnnotationRemarksPass on fs_variant_partial (1162 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_partial
Running pass: AnnotationRemarksPass on fs_variant_whole (1110 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_whole
Running analysis: InnerAnalysisManagerProxy<llvm::FunctionAnalysisManager, llvm::Module> on [module]
Running pass: SROAPass on fs_variant_partial (1162 instructions)
Running analysis: DominatorTreeAnalysis on fs_variant_partial
Running analysis: AssumptionAnalysis on fs_variant_partial
Running analysis: TargetIRAnalysis on fs_variant_partial
Running pass: EarlyCSEPass on fs_variant_partial (1111 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_partial
Running pass: SimplifyCFGPass on fs_variant_partial (961 instructions)
Running pass: ReassociatePass on fs_variant_partial (961 instructions)
Running pass: PromotePass on fs_variant_partial (897 instructions)
Running pass: InstCombinePass on fs_variant_partial (897 instructions)
Running analysis: OptimizationRemarkEmitterAnalysis on fs_variant_partial
Running analysis: AAManager on fs_variant_partial
Running analysis: BasicAA on fs_variant_partial
Running analysis: ScopedNoAliasAA on fs_variant_partial
Running analysis: TypeBasedAA on fs_variant_partial
Running analysis: OuterAnalysisManagerProxy<llvm::ModuleAnalysisManager, llvm::Function> on fs_variant_partial
Running pass: SROAPass on fs_variant_whole (1110 instructions)
Running analysis: DominatorTreeAnalysis on fs_variant_whole
Running analysis: AssumptionAnalysis on fs_variant_whole
Running analysis: TargetIRAnalysis on fs_variant_whole
Running pass: EarlyCSEPass on fs_variant_whole (1059 instructions)
Running analysis: TargetLibraryAnalysis on fs_variant_whole
Running pass: SimplifyCFGPass on fs_variant_whole (912 instructions)
Running pass: ReassociatePass on fs_variant_whole (912 instructions)
Running pass: PromotePass on fs_variant_whole (844 instructions)
Running pass: InstCombinePass on fs_variant_whole (844 instructions)
Running analysis: OptimizationRemarkEmitterAnalysis on fs_variant_whole
Running analysis: AAManager on fs_variant_whole
Running analysis: BasicAA on fs_variant_whole
Running analysis: ScopedNoAliasAA on fs_variant_whole
Running analysis: TypeBasedAA on fs_variant_whole
Running analysis: OuterAnalysisManagerProxy<llvm::ModuleAnalysisManager, llvm::Function> on fs_variant_whole

Fixes: 2037c34f245 ("gallivm: move to new pass manager to handle coroutines change.")
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19217>

aco: change thread_local memory resource to pointer

Apparently the TLS constructor doesn't work well if RADV
is instantiated multiple times and/or used by a program with
already existing threads.

Fixes: a128d444cbf4f74b0bdd4a8ae3f1cd2a52dee122 ('aco: use monotonic_buffer_resource for instructions')
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19219>

loader/dri3: Add DRI performance option to wait for next buffer on swap

For simple clients using the swap chain contention back pressure to regulate
their drawing and that don't query buffer age introduce a new DRI option with
which set to true (the default is false) we block the client until a new buffer
is available. This way we stall the client's execution until a new buffer is
available and the redrawing of the client starts only at this point and not
before.

The motivation for that is to reduce latency for clients that regulate their
drawing by swapchain contention back pressure. These clients draw whenever
possible and their drawing is implicitly stopping whenever we block. When we
block at the end of the swap and return only when a new buffer is available
the client can draw and we directly present. Otherwise the client would draw,
we block on the buffer becoming available, and only then show what the client
had drawn, usually one frame later.

Co-authored-by: Michel Dänzer <michel@daenzer.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14684>

loader/dri3: Simplify buffer age query

The ternary operator is hard to read. Instead initialize the return variable
to zero and adjust its value in an if-clause.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14684>

ci/linker: update mold to 1.6 and enable on s390x

Since 1.6 is s390x supported platform.

Signed-off-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19254>

vc4: don't use imprecise_32bit_lowering for idiv lowering

This is known to produce bogus results for certain combinations of
operands, so don't use it. See this issue for details:
https://gitlab.freedesktop.org/mesa/mesa/-/issues/6555

None of the shaders in shaderdb are affected by this change, as all of
the shaders containing an integer division require a higher GLSL version
than vc4 supports and are therefore all skipped.

This is port of the v3d change 73e8fc3efbfe4a46bfd5ca5b3269d06f086bc224
to vc4; we expect the impact to be the same as for v3d, based on testing
a custom shader.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18019>

vc4: don't lower uadd_carry

We can produce slightly better code for these in the backend, so
do that.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18019>

ac/nir/ngg,radv: use different counters for shader queries

VK_QUERY_TYPE_PRIMITIVES_GENERATED_EXT should count for each
stream.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7409
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19015>

nir,ac/nir/ngg,radv: split shader_query_enabled_amd

For used by different counter.

Vulkan:
1. VK_QUERY_PIPELINE_STATISTIC_GEOMETRY_SHADER_PRIMITIVES_BIT,
   sum generated primitives of all 4 streams when GS.
2. VK_QUERY_TYPE_PRIMITIVES_GENERATED_EXT, count generated
   primitives for all 4 streams when VS/TES/GS.
3. VK_QUERY_TYPE_TRANSFORM_FEEDBACK_STREAM_EXT, count generated
   and streamout primitives for all 4 streams when VS/TES/GS.

OpenGL:
1. GL_GEOMETRY_SHADER_PRIMITIVES_EMITTED_ARB, sum generated
   primitives for all 4 streams when GS.
2. GL_PRIMITIVES_GENERATED, count generated primitives for all 4
   streams when VS/TES/GS.
3. GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN, count streamout
   primitives for all 4 streams when VS/TES/GS.

pipeline_stat_query_enabled_amd is for Vulkan 1 and OpenGL 1.
xfb_query_enabled_amd is for Vulkan 2/3 and OpenGL 2/3.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19015>

radv: split active_pipeline_gds_queries

For different enabling of pipeline stat and prims gen.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19015>

radv: count gen_prims_queries_enabled

User can enable/disable multi VK_QUERY_TYPE_PRIMITIVES_GENERATED_EXT
queries with same or different index.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19015>

zink: delete unused zink descriptor mode

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: support multiple heaps per memory type

some allocations require a different memory heap even when using the
same memory bits, so allow iterating over heaps of the same memory type
to find the one that works

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: allocate/place memory using memoryTypeIndex directly

it's possible for drivers to have multiple heaps with identical flags,
so this will enable passing the heap that should actually be used for
allocation

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: allow direct memory mapping for any COHERENT+CACHED buffer

some drivers may provide this in heaps that get used by non-staging resources,
so avoid extra copies in that case

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: don't flag ubo0 as used in shaders with indirect ubo access

ubo0 is not a regular ubo, so avoid setting the hint here

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: decrement ZINK_FBFETCH_BINDING

there's no reason this shouldn't be contiguous with the other gfx stages

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19281>

zink: fix gpl compile optimization

this is supposed to create the non-optimized pipeline first and then
compile the optimized version in the background, not the other way around

facepalm.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19201>

lavapipe: fix BufferDeviceAddress usage

this is supposed to return the address at the start of the buffer,
not the address at the start of the memory allocation

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19274>

lavapipe: store binding offset for buffers

cc: mesa-stable

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19274>

crocus: fix off-by-one error when clearing stale syncobjs

This is the same fix as commit 1311eddd526 ("iris: fix off-by-one error
when clearing stale syncobjs"), but for crocus.

Signed-off-by: Nicholas Bishop <nicholasbishop@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19282>

clover/llvm: add kernel arg info for opaque pointers

With opaque pointers you can no longer the info from the pointertype,
so rebuild it.

v2: add a comment

Reviewed-by: Mihai Preda <mhpreda@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19103>

d3d12: Add cache managers for separate buffer usage cases

Mis-matched usage is a large percentage of buffer cache misses when
searching for applicable buffers. Separating these into separate
managers puts them into separate heaps and eliminates a significant
amount of CPU time spent searching.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19283>

freedreno/ir3: Switch to NIR for a5xx's vertex id lowering.

This was similarly done for a3xx & a4xx with 4bdd226ab61
("freedreno/ir3: Switch to NIR for a3xx/a4xx's vertex id lowering.")
already, yet a5xx was missed which I noticed as Minecraft window
titlebar text corruption on OnePlus 5T with Adreno 540.

Cc: mesa-stable
Signed-off-by: Jami Kettunen <jami.kettunen@protonmail.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19263>

crocus: Advertise PIPE_CAP_NATIVE_FENCE_FD

Enables EGL_ANDROID_native_fence_sync.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7253
Signed-off-by: Nicholas Bishop <nicholasbishop@google.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19235>

v3d/ci: Add a flake

I've seen this one flake, add it to flakes.txt to avoid blocking CI.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19259>