zink: set needs_mesa_flush_wsi for venus

venus relies on wsi_image_create_info and wsi_memory_signal_submit_info.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14872>

zink: always chain wsi_image_create_info for scanout images

Chaining wsi_image_create_info tells the drivers that the image can use
VK_IMAGE_LAYOUT_PRESENT_SRC_KHR layout.  We still use
wsi_image_create_info::scanout to indicate whether this is legacy
scanout or uses modifiers.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14872>

zink: set dma-buf bit for shared resources

Set the dma-buf bit when supported.  This is done because
zink_resource_get_handle exports dma-bufs when WINSYS_HANDLE_TYPE_FD is
requested.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14872>

agx: Call nir_lower_discard_if

We still need to implement discard itself, but this means we don't need
to worry about discard_if. This compiles down to the same idiom as
the vendor compiler (Metal) generates

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14217>

nir,zink: Make lower_discard_if a common pass

This pass (lowering discard_if to control flow and unconditional
discard) was originally written for Zink, but is useful for hardware
that lacks conditional discard instructions like AGX. In theory AGX
could implement a conditional discard with CSEL, but the vendor
compiler uses a lowering like this one. Since I like not writing code,
I'd like to use the pass that's already in tree.

v2: Don't preserve dominance (Jason). Assert we don't see demotes or
terminates (Jason). Add Mike's ack.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14217>

glx: Use the new no-error driver interface

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12474>

egl: Use the new no-error driver interface

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12474>

dri: Epoch how no-error context creation works

The bug here is that the DRI context "flags" are intended to alias the
GLX context flag values, and they don't, DRI's no-error flag is GLX's
reset-isolation flag. GLX (and EGL!) treat no-error as a context
attribute, and reset isolation predates Mesa's no-error implementation
by several years. The GL_KHR_no_error spec does describe it as a
"context flag", though, so maybe that's why we do it as a (DRI) context
flag.

In order to unalias these we need a new contract with the loader. We
remove the old __DRI_NO_ERROR extension, and add a new
__DRI_RENDERER_HAS_CONTEXT_NO_ERROR value to query. Loaders can key on
that to know to pass no-error-ness through as a context attribute,
matching the GLX/EGL calling convention. We go ahead and define
__DRI_CTX_FLAG_RESET_ISOLATION as well, and update the drivers to refuse
it since we don't support it yet.

This means mismatched drivers/loaders will not be able to create
no-error contexts. Too bad. If you want performance that badly you can
build both things at once.

Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12474>

meson: add LIBGL_DRIVERS_PATH to the devenv

This allows using built dri OpenGL drivers with the `meson devenv`
command

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14826>

meson: add support for `meson devenv` with vulkan

Meson devenv is a feature added in meson 0.58 (thus the features is
version guarded) that allows creating a shell environment with
environment variables automatically setup for running the project inside
the build dir. Some variables (such as LD_LIBRARY_PATH and PATH) are set
automatically, others must be added by the project.

For vulkan is is relativley simple, we create a new, uninstalled, icd
file for each driver and set the VK_ICD_FILENAMES variable
appropriately. This can be used with:

```sh
meson devenv -C $builddir
```

then, vulkan applications will automatically use the uninstall vulkan
driver, no need to install.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14826>

vulkan/util: simplify multialloc init

The syntax we're using doesn't work when included into C++ sources. So
let's make it C++ compabible.

It turns out, nobody needs this extra definition which is what's causing
issues. Let's just make the initializer trivial without casting the
struct.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14850>

vulkan/util: Add explicit casts to make c++ happy

We're about to need including this header from a C++ source, so let's
add some explicit casts for C++ compatibility.

In one case we can make things a bit cleaner by moving the
char-pointer-ism to the place that needs it, so let's clean that up
while we're at it.

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14850>

vulkan/util: Add extern "C" to allow inclusion from c++

We're about to need including this header from a C++ source, so let's
wrap the whole thing in extern "C".

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14850>

turnip: Expose VK_KHR_shader_non_semantic_info

This is entirely implemented in the SPIR-V frontend.

Relevant CTS tests:
dEQP-VK.spirv_assembly.instruction.compute.non_semantic_info.*

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14829>

turnip: Implement VK_KHR_zero_initialize_workgroup_memory

Moved nir_lower_compute_system_values to lower
load_local_invocation_index which could be emitted by
nir_zero_initialize_shared_memory.

Relevant CTS tests:
dEQP-VK.compute.zero_initialize_workgroup_memory.*

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14829>

turnip: Expose VK_EXT_image_robustness

VK_EXT_image_robustness is a strict subset of VK_EXT_robustness2
so we could just expose it.

Relevant CTS tests: dEQP-VK.robustness.image_robustness.*

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14829>

turnip: Use the shared helpers to expose 1.3 core extensions/limits

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14829>

turnip/doc: Update turnip extension list

We were missing VK_EXT_subgroup_size_control and
VK_EXT_extended_dynamic_state2.

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14829>

iris: Make an iris_foreach_batch macro that skips unsupported batches

IRIS_BATCH_BLITTER isn't supported prior to Tigerlake; in general,
batches may not be supported on all hardware.  In most cases, querying
them is harmless (if useless): they reference nothing, have no commands
to flush, and so on.  However, the fence code does need to know that
certain batches don't exist, so it can avoid adding inter-batch fences
involving them.

This patch introduces a new iris_foreach_batch() iterator macro that
walks over all batches that are actually supported on the platform,
while skipping the others.  It provides a central place to update should
we add or reorder more batches in the future.

Fixes various tests in the piglit.spec.ext_external_objects.* category.

Thanks to Tapani Pälli for catching this.

Fixes: a90a1f15 ("iris: Create an IRIS_BATCH_BLITTER for using the BLT command streamer")
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14834>

llvmpipe/triangle: don't store area in fixed_position.

It doesn't get used again, just extract the sign and move on.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

llvmpipe: just move opaque alpha lookup closer to use.

Saves looking this up before checking the variant.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

llvmpipe: refactor lp_rast_shader_inputs.

viewport_index won't be >= 16
layer should be < 2048
view_index should be < 2048

this leaves the last 64-bits as padding, which something
expects, but not having to write to it means we have to write
less memory every triangle.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

llvmpipe/setup: remove opaque from setup triangle

This isn't used in the rast side of things, so just pass it out here.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

llvmpipe: inline retry_triangle_ccw

This reduces some of the overheads in the callstack here.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

llvmpipe: optimise triangle setup a bit.

the bboxpos = bbox copy was visible overhead on perf traces,
but we don't need to really do it.

Extract all the things from bbox needed early, then don't copy
it just overwrite it.

use boolean to fix power build.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14846>

turnip: Do not use hw binning if tiles per pipe are over the limit

Otherwise GPU would hang.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5951

Freedreno commit as a reference:
39d00722b22a0059bbc58f0158a22f384519cd39

Fixes VK cts tests on a618 if their memory limit is raised to 1024 MB:
 dEQP-VK.pipeline.render_to_image.core.2d_array.huge.width_height.r8g8b8a8_unorm_d16_unorm
 dEQP-VK.pipeline.render_to_image.core.2d_array.huge.width_height.r8g8b8a8_unorm_s8_uint
 dEQP-VK.pipeline.render_to_image.core.2d_array.huge.width_height.r8g8b8a8_unorm_d24_unorm_s8_uint
 dEQP-VK.pipeline.render_to_image.core.2d_array.huge.width_height.r8g8b8a8_unorm_d32_sfloat_s8_uint
 dEQP-VK.pipeline.render_to_image.core.cube.huge.width_height.r8g8b8a8_unorm
 dEQP-VK.pipeline.render_to_image.core.cube.huge.width_height.r8g8b8a8_unorm_d16_unorm
 dEQP-VK.pipeline.render_to_image.core.cube.huge.width_height.r8g8b8a8_unorm_s8_uint
 dEQP-VK.pipeline.render_to_image.core.cube.huge.width_height.r8g8b8a8_unorm_d24_unorm_s8_uint
 dEQP-VK.pipeline.render_to_image.core.cube_array.huge.width_height.r8g8b8a8_unorm
 dEQP-VK.pipeline.render_to_image.core.cube_array.huge.width_height.r8g8b8a8_unorm_d24_unorm_s8_uint

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Tested-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14849>

turnip: Add TU_GMEM envvar to test different gmem sizes

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14849>

venus: Advertise VK_EXT_extended_dynamic_state support

Signed-off-by: Omar Akkila <omar.akkila@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14866>

venus: Implement VK_EXT_extended_dynamic_state commands

This implements hooks for the following commands:

- vkCmdBindVertexBuffers2
- vkCmdSetCullMode
- vkCmdSetDepthBoundsTestEnable
- vkCmdSetDepthCompareOp
- vkCmdSetDepthTestEnable
- vkCmdSetDepthWriteEnable
- vkCmdSetFrontFace
- vkCmdSetPrimitiveTopology
- vkCmdSetScissorWithCount
- vkCmdSetStencilOp
- vkCmdSetStencilTestEnable
- vkCmdSetViewportWithCount

Signed-off-by: Omar Akkila <omar.akkila@collabora.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14866>

docs: Update d3d12 features

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: ARB_gpu_shader_fp64

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Handle structs in TCS variants

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Handle structs in GS variants

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Set lower full fp64 compiler options flag when needed

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Lower [de]construction of doubles via math ops into pack/unpack ops

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Add int64 support

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Get OPTIONS1

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Add a driver version to the screen to be used for workarounds

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Cache a modifyable copy of the nir options in d3d12_screen

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Use the right constant for GS varying limits

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Use a constant define for max anisotropy

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

d3d12: Update max input, output, and varying caps

The simple-varyings piglit test attempts to use GL_MAX_VARYING_FLOATS
varyings, PLUS one additional vector for position (which is not used
as input to the PS). "Reserve" that additional position vector by
removing it from the max varyings and max PS inputs.

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Handle structs in I/O signatures

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Map I/O base locations to input IDs

When dealing with a vertex input that takes multiple rows, the value of
nir_intrinsic_base points to a driver-location-based index, but we need
to emit a location-based index (or more specifically, an index that
increments once per input, not once per register). Add a mapping to
the module of base -> ID.

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Process signatures before the shader code

This lets us set up some metadata based on I/O vars without having
to do multiple passes over them.

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Handle I/O vars larger than a vec4

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Lower 64bit I/O to 32 and then run lower_pack

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Handle b2f64

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Set dx11_1_double_extensions flag for dfma/ddiv

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Fix dxil_nir_lower_double_math_instr pass for vectors

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Fix make_double and split_double to respect swizzles

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Fix splitdouble struct name

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: It's possible to have doubles without int64

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Add never-supported double ops to lower_doubles bitmask

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Only treat tess level location as special if it's a patch constant

Fixes: a550c059 ("microsoft/compiler: For load_input from DS, use loadPatchConstant")
Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Only prep phis for the current function

Fixes: 41af9620 ("microsoft/compiler: Emit all NIR functions into the DXIL module")
Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

microsoft/compiler: Lower mul_2x32_64

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Reviewed-By: Sil Vilerino <sivileri@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14837>

ci: Uprev vulkan-cts to 1.2.8.0

This brings in some interesting new vulkan tests and fixes for the
spurious KHR-GL TF failures.  Also, reduces the runtime of
dEQP-GLES31.functional.ssbo.layout.random.all_shared_buffer.36 so that it
should stop timing out.

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

llvmpipe: Disable an assertion that may not be quite right.

It triggered on uprevving VK-GL-CTS, and @airlied says it's tripped
apparently spuriously before.  There seems to be some interesting logic
behind it, so leave the big comment for whoever can revisit the issue some
day.

Acked-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

ci/i915: Update rendering hash for plot3d trace.

Its rendering changed slightly at some point, but it's fine.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

ci/freedreno: Reduce concurrency for a618 vk_full.

This ran into OOM-kills with the CTS uprev.  Looking at caselists at the
time of fail, some had 500MB of system memory used by the CTS (mostly
spirv string codegen), plus whatever BOs were allocated, and the lazors
are only 4GB it looks like.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

ci/turnip: Extend the full-vk-run job timeouts.

Between adding features and increased test coverage, we're hitting the
1-hour job limit.  !13441 tried to increase the full run timeout for LAVA,
but by having not bumped the gitlab-ci timeout value it ended up just
letting the job keep running in LAVA after gitlab had given up on it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

ci/freereno: Reduce run-by-default a630-vk coverage.

In the autotune merge, we added another 1/15th run of a configuration
knob, thinking that was small enough to be in the noise.  But actually the
main run is only 1/9th, so another 1/15th took us from nearly hitting the
job runtime target, to totally missing it.  Crank things back down to keep
MRs flowing.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13779>

r300: Simplify DCE by assuming all output writes are used.

No change on shader-db.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14847>

r300: Set up shadow sampler lowering in precompiles.

Otherwise you end up lowering all shadow samples to a MOV dst
temp[0].0000, which is pretty silly.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14847>

r300: Fix missing \n in an error message.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14847>

zink: add synchronization for conditional render buffer

doesn't seem to do anything on any drivers I've tested, but maybe it's
needed somewhere

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14853>

zink: add VK_BUFFER_USAGE_CONDITIONAL_RENDERING_BIT_EXT for query binds

required by spec

cc: mesa-stable

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14853>

aco: don't encode src2 for v_writelane_b32_e64

Encoding src2 doesn't cause issues for print_asm() because we have a
workaround there, but it does for RGP and it seems the developers are not
interested in fixing it.

https://github.com/GPUOpen-Tools/radeon_gpu_profiler/issues/61

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Tested-by: Tatsuyuki Ishi <ishitatsuyuki@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14832>

aco: add test for optimizations with casts

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco: fix neg(mul)/abs(mul) optimization with different bit-size

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco: don't combine add/mul of different bit-size

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco: don't apply omod/clamp of different bit-size

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco: don't combine fneg/fabs of different bit-size

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco/tests: implement sub-dword program inputs

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

aco/tests: add a bunch more building helpers

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14810>

panfrost: Fix texel interleave flag on Valhall

Interleave mode specified per-plane on Valhall. The texture descriptor proper
merely has a flag specifying whether planes are somehow interleaved
(u-interleaved, AFBC, or block compressed formats) or whether they are all
linear (and uncompressed).

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14851>

panfrost: Add remaining ZS/CRC XML

Flesh out the ZS/CRC XML, adding fields required for AFBC. Valhall allows AFBC
compressing stencil buffers independent of depth buffers, which is a new feature
since Bifrost. That results in a shuffling of the descriptor.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14851>

panfrost: Add Valhall Plane Descriptor XML

This looks superficially like the Bifrost "Surface" descriptor, but it
additionally specifies the in-memory representation of blocks (clumps). If I
understand correctly, decompression is controlled by the plane descriptor,
rather than the texture descriptor level. This is a bit more flexible than
Bifrost.

Once the new fields here are wired up to Mesa, my
dEQP-GLES2.functional.texture.* failures should go away... I hope!

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14851>

panfrost: Fix alignments on Valhall

Otherwise we get DATA_INVALID_FAULT trying to run even trivial null jobs. For
each descriptor, set the correct alignment.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14851>

panfrost: Remove blend shader return value on v9

Removed since there's a new ABI for blend shaders. Even if we always write 0,
it's better not to pack this at all, and to denoise the dumps.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14851>

v3d/drm-shim: remove drm-shim driver

After starting to use a new version of the simulator, it got
outdated.

We made some initial effort to update it, but it was not
working. Taking into account that no one is using it, it is better to
just remove it.

We keep the noop drm drivers, as they could have some value for
developers that doesn't have access to the v3dv3 simulator.

Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14682>

radeonsi: allocate protected buffer only if required

protected buffer allocations need to be made if the context is secure

Signed-off-by: Shirish S <shirish.s@amd.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14848>

radeonsi: limit loop unrolling for LLVM < 13

Without this change LLVM 12 hits this error:

"""
LLVM ERROR: Error while trying to spill SGPR0_SGPR1 from class SReg_64:
Cannot scavenge register without an emergency spill slot!
"""

when running glcts KHR-GL46.arrays_of_arrays_gl.AtomicUsage test.

Fixes: 9ff086052ab ("radeonsi: unroll loops of up to 128 iterations")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14848>

radv: stop setting streamout state when a new pipeline is bound

It's required to have a valid graphics bound pipeline with XFB when
vkCmdBeginTransformFeedbackKHR() is called. This removes extra work
when binding a pipeline.

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/14610>

broadcom/compiler: allow ldunifa with read-only SSBOs

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14830>

broadcom/compiler: fix offset alignment for ldunifa when skipping

The intention was to align the address to 4 bytes (32-bit), not
16 bytes.

Fixes: bdb6201ea1 ("broadcom/compiler: use ldunifa with unaligned constant offset")

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14830>

docs: update calendar for 22.0.0-rc1

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14845>

docs: reset new_features.txt

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14843>

llvmpipe: ci updates

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14835>

llvmpipe: disable PIPE_SHADER_CAP_FP16_CONST_BUFFERS

this cap is broken

cc: mesa-stable

fixes:
GTF-GL46.gtf21.GL2Tests.glGetUniform.glGetUnifor

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14835>

zink: disable PIPE_SHADER_CAP_FP16_CONST_BUFFERS

this cap is broken

cc: mesa-stable

fixes:
GTF-GL46.gtf21.GL2Tests.glGetUniform.glGetUniform

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14835>

VERSION: bump version for 22.0 release

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14840>

radv: Handle SDMA for padding.

Also assert that nobody actually needs to chain an SDMA IB because we have
not implemented non-PKT3 chaining.

Fixes: ef40f2ccc29 ("radv/amdgpu: Fix handling of IB alignment > 4 words.")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/5923
Tested-by: Mike Lothian <mike@fireburn.co.uk>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14781>

intel: Add missing dep of gen_*_header.py on utils.py.

Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14725>

freedreno/isaspec: Add missing dep of encode.py/decode.py calls on isa.py

Fixes: #5921
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14725>

anv: Add experimental support for VK_NV_mesh_shader

Enable setting ANV_EXPERIMENTAL_NV_MESH_SHADER=1 environment variable.

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/13662>

intel/dev: Enable Mesh Shading for DG2

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/13662>

anv: Put first few push constants directly into Task/Mesh InlineData

Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13662>

anv: include ClipDistance array in mesh shader per-vertex output

Signed-off-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13662>