microsoft/compiler: Lower fquantize2f16

As far as I can't tell, there's no native operation doing this
equivalent of fquantize2f16. Let's lower this operation to

   if (val < MIN_FLOAT16)
      return -INFINITY;
   else if (val > MAX_FLOAT16)
      return -INFINITY;
   else if (fabs(val) < SMALLER_NORMALIZED_FLOAT16)
      return 0;
   else
      return val;

which matches the definition of OpQuantizeToF16:

"
If Value is an infinity, the result is the same infinity.
If Value is a NaN, the result is a NaN, but not necessarily the same NaN.
If Value is positive with a magnitude too large to represent as a 16-bit
floating-point value, the result is positive infinity. If Value is negative
with a magnitude too large to represent as a 16-bit floating-point value,
the result is negative infinity. If the magnitude of Value is too small to
represent as a normalized 16-bit floating-point value, the result may be
either +0 or -0.
"

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16959>

util/u_debug: Use 'initialized' instead of 'first'

Using 'initialized' to guard the one-time init, means it can be set to
false as part of .bss instead setting 'first' to true in .data.  This
is more efficient and works at .ctor time.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16952>

zink: add new flakes

these take a while to run and I don't want anyone failing ci

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16957>

crocus: Enable PIPE_CAP_LEGACY_MATH_RULES.

Now that TTN hooks this up to use_legacy_math_rules, we can flip the
switch and gallium nine can get the desired behavior from the hardware
instead of emitting math workarounds.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

iris: Enable PIPE_CAP_LEGACY_MATH_RULES.

Now that TTN hooks this up to use_legacy_math_rules, we can flip the
switch and gallium nine can get the desired behavior from the hardware
instead of emitting math workarounds.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

Revert "nine: Disable optional use of TTN when MUL_ZERO_WINS is available."

This reverts commit 7f01299c40ef5d301bbbb99fc6a6420aa27a65dc.

Now that I've got it hooked up to use_legacy_math_rules on the NIR side
and made sure that NIR frontends on drivers with
PIPE_CAP_LEGACY_MATH_RULES handle it, we should be able to enable this
again.

Fixes: #5678
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

gallium: Rename PIPE_CAP_TGSI_LEGACY_MATH_RULES to drop "TGSI"

Now NIR drivers can implement it as well.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

tgsi_to_nir: Set use_legacy_math_rules for TGSI_PROPERTY_LEGACY_MATH_RULES.

These control the same behavior, now that we've clarified what the flags
do.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

nouveau/nir: Implement mul_zero_wins behavior for use_legacy_math_rules.

This is the same flag TGSI sets for LEGACY_MATH_RULES.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: M Henning <drawoc@darkrefraction.com>
Tested-by: Mobin Aydinfar <mobin@mobintestserver.ir>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

r600/sfn: Implement 0*x=0 behavior for use_legacy_math_rules.

The TGSI backend chooses these opcodes for LEGACY_MATH_RULES, so do the
same thing here for ARB programs which want the same behavior.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

nir_to_tgsi: Set LEGACY_MATH_RULES cap for use_legacy_math_rules shaders.

This should help get correct math for ARB_fp/vp after the NTT transition,
and will be used for wine nine shortly.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

gallium: Rename MUL_ZERO_WINS to LEGACY_MATH_RULES.

This is a clearer name for what it does than MUL_ZERO_WINS, and matches up
to the new name in shader_info.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

nir: Rename is_arb_asm to use_legacy_math_rules and document its meaning.

On iris and crocus, this flag is used to set "alt mode" math on the shader
as a whole.  Some other drivers have a similar mode for DX9/ARB-program
behavior, so document what it does so we can start using it.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16176>

zink: update tu a630 baseline

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16956>

zink: fix 32bit bo rewriting

this was correct for 64bit loads and manually converted 32bit loads (e.g., bindless),
but it was broken for the case where 64bit was not supported, as the offset wasn't
being correctly adjusted

break out the offset division to hopefully make this a little clearer

Fixes: 150d6ee97e3 ("zink: move all 64-32bit shader load rewriting to nir pass")

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: avoid adding more 64bit alus during bo rewriting

nir_unpack_64_2x32 isn't supported, but split is, so use that

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: scalarize when rewriting explicit 64bit io

all of ntv requires scalarized io since the offsets are now array indices
instead of byte offsets, so enforce scalarization here to avoid breaking
the universe

Fixes: 150d6ee97e3 ("zink: move all 64-32bit shader load rewriting to nir pass")

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: rewrite 64bit shader i/o as 32bit

zink can't use lower_io, so this all has to be done manually and in
excruciating depth and detail

fixes (tu):
KHR-Single-GL46.arrays_of_arrays_gl.InteractionFunctionCalls2
KHR-GL46.gpu_shader_fp64.fp64.named_uniform_blocks
KHR-GL46.gpu_shader_fp64.fp64.varyings
KHR-GL46.vertex_attrib_binding.advanced-bindingUpdate
KHR-Single-GL46.enhanced_layouts.varying_array_components
KHR-Single-GL46.enhanced_layouts.varying_array_locations
KHR-Single-GL46.enhanced_layouts.varying_components
KHR-Single-GL46.enhanced_layouts.varying_locations
KHR-Single-GL46.enhanced_layouts.xfb_explicit_location
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.lines.highp_mat3x4
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.lines.mediump_vec3
dEQP-GLES3.functional.transform_feedback.basic_types.interleaved.points.mediump_vec4
dEQP-GLES3.functional.transform_feedback.basic_types.separate.points.highp_vec3

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: implement nir_op_unpack_64_2x32 in ntv

it's really annoying to not have the full suite of these, so just do it

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: improve/expand manual xfb emission

I left this semi-unfinished back when I discovered that I could blast
out xfb values inline with variable declarations, but this is not viable
for all scenarios, so it has to work and it has to be able to pass cts

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: if a variable is flagged as being xfb, treat it as manual xfb output

this allows passes to flag variables as "needs manual action" for further
handling later on

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: fix variable locations in manual xfb emission

the last output isn't always only consuming 1 slot, so ensure that the
xfb outputs begin at the appropriate place

cc: mesa-stable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: flag all shader i/o variables as !is_xfb

reclaiming this flag for internal use

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: fix ntv partial stores

the previous conditional here was stupid and wrong: it should be comparing
to see whether the writemask is the full mask of the type's size

cc: mesa-stable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: run more int64 passes during optimization loop if int64 isn't available

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: don't lower pack/unpack intrinsics

this just ends up propagating the bitsize that we don't want, making it
harder to remove them later

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: always scalarize pack/unpack alu ops

these are tricky

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: run nir_lower_alu_to_scalar in optimizer loop for 64bit lowering

ensure that pack/unpack ops aren't passing swizzles since those are hard
to implement

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: break out nir alu typing in ntv

this is reusable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

zink: flag optimization pass after inlining uniforms

this probably generated some pretty gnarly code

cc: mesa-stable

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16669>

intel/compiler: Fix A64 header construction with a uniform address

fs_visitor::assign_curb_setup() maps UNIFORM registers to HW regs,
and contains the following assert:

            assert(inst->src[i].stride == 0);

emit_a64_oword_block_header's striding tricks run afoul of this
restriction, by producing stride 1 values on a 64-bit UNIFORM source.

Work around this by copying the UNIFORM value to a VGRF first.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16938>

vulkan/wsi: Signal semaphores and fences from the dma-buf

Instead of attempting to signal based on the memory object, use the new
DMA_BUF_IOCTL_EXPORT_SYNC_FILE to get a sync_file for the dma-buf and
use that to signal the semaphore or fence.  Because this happens before
we transfer ownership back to the driver, the resulting sync_file should
only contain dma_fences from the compositor and/or display and shouldn't
be mixed up with the driver in any way.  This gives us a real semaphore
and fence (as opposed to the dummy objects we've used int the past)
without over-synchronization.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

vulkan/wsi: Set memory ownership after signaling fences and semaphores

This isn't a functional change today because the set of drivers which
use set_ownership and those that use signal_fence/semaphore_for_memory
are mutually exclusive.  It's important for the next commit, though.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

vulkan/wsi: Re-arrange QueueSubmit code in wsi_common_queue_present

Instead of treating the blit submit specially in the buffer_blit_queue
case, treat the dummy submit as special.  This lets us keep all the
handling of special-queue blits together.  It also means that the
wsi_memory_signal_submit_info gets chained into the final submit which
is what we want if we're to rely on it for implicit sync.  If we chain
it into the dummy submit, we'll implicit sync on all work previous to
the blit but not the blit.  This won't work if X11 or a Wayland
compositor is depending on that to synchronize the linear copy.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

vulkan/wsi: Reset fences earlier in wsi_common_queue_present

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

vulkan/wsi: Use a STACK_ARRAY for stage_flags

This avoids the heap allocation in the common case.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

dzn: Init sync types before wsi

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

turnip: Set supported_sync_types before wsi_init

The WSI code is about to start querying for available semaphore handle
types via GetPhysicalDeviceExternalSemaphoreProperties in wsi_init().
For drivers that use vk_sync, supported_sync_types needs to be
initialized before GetPhysicalDeviceExternalSemaphoreProperties is
called.  Really, wsi_init() should be the very last step of physical
device setup.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

turnip: Use the common QueuePresent implementation

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

turnip: Use the common AcquireNextImage implementation

The only reason for the wrapper was so that we could dummy signal the
semaphore and fence.  Now that the WSI code always dos this for us, we
can drop our wrapper.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

pvr: Use the common AcquireNextImage implementation

The only reason for the wrapper was so that we could dummy signal the
semaphore and fence.  Now that the WSI code always dos this for us, we
can drop our wrapper.

Reviewed-by: Rajnesh Kanwal <rajnesh.kanwal@imgtec.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

lavapipe: Use the common AcquireNextImage implementation

The only reason for the wrapper was so that we could dummy signal the
semaphore and fence.  Now that the WSI code always dos this for us, we
can drop our wrapper.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

radv: Drop create_sync_for_memory

Also, stop setting wsi_device::signal_semaphore/fence_with_memory
because those cause the WSI code to call the function we just dropped.
Since the core WSI code is now setting dummy syncs by default, we don't
need any of this anymore.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

v3dv: Use the common AcquireNextImage implementation

The only reason for the wrapper was so that we could dummy signal the
semaphore and fence.  Now that the WSI code always dos this for us, we
can drop our wrapper.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

panvk: Use the common AcquireNextImage implementation

The only reason for the wrapper was so that we could dummy signal the
semaphore and fence.  Now that the WSI code always dos this for us, we
can drop our wrapper.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

wsi: Always signal semaphores and fences in wsi_common_acquire_next_image

If the driver wants to do something special, it can reset the semaphore
or fence again and re-signal it.  Sure, that wastes a malloc/free but
this is the window-system path.  It'll be fine.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

vulkan/wsi: Fix a signal_semaphore_with_memory check

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

anv/wsi: Stop resetting semaphores

This will happen automatically when they're waited on by the dummy
submit in wsi_common_queue_present().

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4037>

intel: Set a more useful fake devinfo->gtt_size in no-hw mode

With the old value, anv didn't think that the hardware supported 48-bit
addresses, and hit this assert:

   assert(device->supports_48bit_addresses == !device->use_relocations);

The new value of 1ull << 48 is the one reported on my Icelake machine.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16933>

anv: Remove FS executables when applying the null FS optimization

If the executables are still hanging out,
anv_GetPipelineExecutableStatisticsKHR will try to dereference NULL
pointers in pipeline->shaders[MESA_SHADER_FRAGMENT].

At least in terms of fossil-db output, this matches the behavior from
before 73b3efcd59a.

Fixes: 73b3efcd59a ("anv: Handle the null FS optimization after compiling shaders")
Closes: #6590
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16898>

radv: Implement VK_EXT_non_seamless_cube_map.

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12729>

vulkan: Spec update to 1.3.217.

Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12729>

lima: implement lima-specific blitter

It should be more efficient than u_blitter since it skips
vertex shader stage. Also it's a prerequisite for supporting
MSAA since u_blitter can't do MSAA resolve for us.

Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16686>

compiler/types: Don't place members in the previous substruct's end padding

With the following structures :

  struct StructA
  {
     uint64_t value0;
     uint8_t  value1;
  };

  struct TopStruct
  {
     struct StructA a;
     uint8_t value3;
  };

Currently offsetof(struct TopStruct, value3) = 9. While the same code
on the CPU gives offsetof(struct TopStruct, value3) = 16.

This is impacting OpenCL kernels we're trying to use to build
acceleration structures.

v2: Add comment/link to some description of the alignment/size
    computation

Cc: mesa-stable
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16940>

.gitignore: Qualify the path for the ignored build directory.

This was origally added for scons, which is gone, but keep it around since
enough people use "build" for their meson builds that you probably
shouldn't add anything to git under that name.  The qualifying '/' is
needed because we have a .gitlab-ci/build/ directory where we do check in
code.

Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16827>

mesa/st: Abort the linking on driver link failure.

If the driver said it can't do the shader, then listen to it and don't ask
it to create the shaders anyway.  Fixes a bunch of spam on i915/r300 (with
!16878) about unsupported opcodes during dEQP runs.

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16895>

turnip: Reverse the order of walking pipes or tiles on odd rows.

This improves the cache locality compared to raster order.  Improves
gfxbench vk-5-normal perf by 3.3009% +/- 0.105934% (n=3).

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16917>

ci/turnip: add a link to VkPhysicalDeviceMemoryProperties failure

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16939>

ci/iris: Disable skqp until it can be stabilized.

It keeps blocking marge with flakes across many different tests.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16951>

util: Add unittest for timespec_get

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

util: Remove unused function timespec_passed

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

util: always enable timespec_test.cpp

c11/time.h always contains `struct timespec` now

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

util: Remove usage of designated initializers in timespec_test.cpp

As the default option for msvc 2019 does support designated initializers
```
../src/util/tests/timespec_test.cpp(302): error C7555: use of designated initializers requires at least '/std:c++20'
../src/util/tests/timespec_test.cpp(303): error C7555: use of designated initializers requires at least '/std:c++20'
../src/util/tests/timespec_test.cpp(312): error C7555: use of designated initializers requires at least '/std:c++20'
../src/util/tests/timespec_test.cpp(313): error C7555: use of designated initializers requires at least '/std:c++20'
```

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

util: Use timespec_get directly, it's always present

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

c11: Implement timespec_get on win32 properly when not available

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

c11: Implement c11/time.h with c11/impl/time.c

Create c11/time.h instead of put timespec_get in `c11/threads.h`

Creating impl folder is used to avoid `#include <time.h>` point the c11/time.h file

Detecting if `struct timespec` present with meson
Define TIME_UTC in `c11/time.h` instead `c11/threads.h`
Define `struct timespec` in `c11/time.h` when not present.
Implement timespec_get in c11/impl/time.c instead threads.h

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

c11/threads: Remove the detecting code for `HAVE_TIMESPEC_GET` in threads_win32.h

The meson already detecting it properly now, so remove the duplicated detecting code

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

meson: Fixes timespec_get detect code

Add functions_to_detect dictionary to records the paths to include

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15497>

d3d12: Properly set HS input control point count

Looks like some hardware needs this info in the shader to match the
topology. Since there's no spot in the shader info for it, we're
currently using the array size of the TCS input vars to store it.

Cc: mesa-stable
Reviewed-by: Paul Dodzweit <paul.dodzweit@amd.com>
Tested-by: Paul Dodzweit <paul.dodzweit@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16920>

nir/nir_lower_io: Optimize 32-bit inbounds access

Perform address calculation in 32 bits when
dealing with inbounds array derefs.

Closes: #6562
Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16729>

nir,spirv: Preserve inbounds access information

Preserving information about inbounds access and
the required bit size for the bounds will help
with avoiding 64-bit operations when lowering io.

Signed-off-by: Konstantin Seurer <konstantin.seurer@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16729>

dzn: Enable depth-bounds testing

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16948>

dzn: Use CreatePipelineState()

This way we can easily extend the logic to support new features, like
depth-bounds testing.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16948>

dzn: Drop an unused argument passed to dzn_graphics_pipeline_translate_vi()

The 'out' argument is not used, let's drop it.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16948>

dzn: Switch to ID3D12Device2

So we can use the CreatePipelineState method and unlock new features
like depth-bounds tests.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16948>

zink: always create descriptor templates

the 'notemplates' debug mode is somewhat misleading since there's no
uncached+notemplates mechanism, meaning that if the descriptor cache
explodes it'll still use templates for updating in the fallback path

Fixes: 4e3768914d4 ("zink: add ZINK_DESCRIPTORS env var to explicitly set a mode")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16927>

zink: fix buffer descriptor hashing

since all the buffers are now an array descriptor, hashing needs
to iterate over all the descriptors in order to be accurate

Fixes: a7327c7cac9 ("zink: implement indirect buffer indexing")

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16927>

zink: simplify some compiler bo code

this is a bit easier to read

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16927>

Init macOS GitHub Action

Closes #6517

Signed-off-by: Yurii Kolesnykov <root@yurikoles.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16875>

Remove Travis CI config

Closes #4919

Signed-off-by: Yurii Kolesnykov <root@yurikoles.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16875>

v3d: add ARB_polygon_offset_clamp extension support

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/16925>

panfrost: Constant stencil value tracking

If stencil is constant across the resource, then it can be treated as
if it was cleared.

Improves performance in applications which create a stencil buffer but
do not use it. Originally the same was done for depth to help some 2D
applications, but that gave mixed results so the patch was dropped.

v2: Don't do anything if a fragment job wouldn't be needed otherwise.
v3: Set stencil_value when a batch is cleared (Alyssa)
v4: Handle clears when the stencil is already known (Alyssa)
v5: Make sure shared resources are not used (Alyssa)

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16646>

radv: advertise VK_EXT_primitives_generated_query

Pass all dEQP-VK.transform_feedback.primitives_generated_query.*.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: implement VK_EXT_primitives_generated_query

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: add few helpers related to streamout

Streamout must be enabled for the PRIMITIVES_GENERATED query to work.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: flush the NGG query state when the argument is declared

When primitives generated query is used, the driver also needs to
emulate counting for NGG VS/TES.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

ac/nir/ngg: count the number of generated primitives for VS and TES

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: declare the NGG query argument for primitives generated query

To emulare counting for NGG VS/TES.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: add primitives_generated_query to the graphics pipeline key

Different shader variants will be used to not hurt when the features
aren't enabled.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: track if primitives generated query features are enabled

This will be used to conditionally declare the NGG query argument
for NGG VS/TES and also adjust the NGG shaders.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

radv: rename ngg_gs_state to ngg_query_state

This state will be used by NGG VS and NGG TES too.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15639>

lima/ci: Don't skip piglit opengl 1 tests

Some extensions and features we support only get covered in this space.
We still get a 10min runtime, so it makes sense to include them from
now on.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16928>

svga: fix shared memory index

State tracker only declares one shared memory for GLSL and all
references to the shared memory are with index 0.
So fix the shared memory index in the declaration to use 0 index.

Fixes spec@arb_compute_shader@execution@shared*

Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16930>

svga: fix reference to NULL offset for atomic counter

ntt does not supply an offset source for HWATOMIC.
So no need to reference the offset source for HWATOMIC.

Fixes spec@arb_compute_shader@execution@basic*

Reviewed-by: Neha Bhende <bhenden@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16930>

glsl: inline do_common_optimization() call

The function was previously a helper for when some drivers still
called the GLSL IR optimisations in a loop. No drivers do that
anymore.

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16924>

glsl: remove never true do_dead_code() parameter

Since we have now switched all drivers to using NIR and therefore
the NIR based uniform linker this param never needs to be set to
true so remove it.

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16924>

docs: link releases notes for 21.3.9

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16931>

docs Add sha256 sums for 21.3.9

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16931>

docs: add release notes for 21.3.9

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16931>

nir/deref: Handle RESTRICT for SSBO deref bindings

Tested-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: M Henning <drawoc@darkrefraction.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Cc: mesa-stable@lists.freedesktop.org
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16894>