ci: add xorg to the x86_test-vk container

Only the AMD video drivers for xorg are added since there are no other
expected users by now.

v2:
  - Remove the start/stop logic from the x.sh script. We don't care
    about stopping since that's already managed by gitlab-ci (Emma).

v3:
  - Remove mistakenly added ".gitlab-ci/common/start-x.sh"
    script (Martin).

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Martin Peres <martin.peres@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10856>

intel/compiler: balanced tileY/linear friendly LID order for CS

Fixes perf regression introduced from tileY LID order for CS
shaders that access both textures and buffers. Walks LIDs in
X-major fashion, but with blocks of height 4. This maps LIDs per
HW thread for SIMD8/16/32 as (2x4/4x4/8x4), which is always good
for tileY resources and usually good for linear resources.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10733>

intel/compiler: tileY friendly LID order for CS

Computer shaders that access tileY resources (textures) benefit
from Y-locality accesses. Easiest way to implement this is walk
local ids in Y-major fashion, instead of X-major fashion. Y-major
local ids will reduce partial writes and increase cache locality
for tileY accesses since tileY resources cachelines progress in
Y direction.

Improves performance on TGL:
  Borderlands3.dxvk-g2  +1.5%

Y-major can introduce a performance drop on CS that use mixture
of buffers and images. This should be fixed in next commit.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10733>

intel/compile: refactor DERIVATIVE_GROUP logic

Minor changes to logic to make following changes easier.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10733>

intel/compiler: Use switch for DERIVATIVE_GROUP logic

Switch statement is more readable.

Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10733>

iris: Check data alignment for copy_mem_mem

Check both source and destination offsets are aligned to 4. This
patch fixes dEQP-GLES{2|3}.functional.buffer.write.random.* tests
failures on guest side while trying to copy small (<16b) buffers
via glBufferSubData() with offset which isn't aligned to 4.

Fixes: 9b1b9714 ("iris: Use MI_COPY_MEM_MEM for tiny resource_copy_region calls.")
Cc: mesa-stable@lists.freedesktop.org
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/10910>

aux/vbuf: prevent uint underflow and assert if no vbs are dirty

if this mask is 0, there is nothing to do here

Fixes: e73bf3b805d ("gallium: add start_slot parameter to set_vertex_buffers")

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10929>

pps: fix a missing include in Intel pps driver

In file included from ../src/intel/ds/intel_pps_perf.cc:8:
../src/intel/ds/intel_pps_perf.h:34:78: error: ‘string’ in namespace ‘std’ does not name a type

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10934>

ir3: Only use per-wave pvtmem layout for compute

The blob seems to do this since a630, and it fixes
spec@glsl-1.30@execution@fs-large-local-array on a650.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10922>

ir3: Call nir_lower_wrmask() again after lowering scratch

I forgot that after rebasing on large_consts support that this is now
called after the first time nir_lower_wrmask is called and can generate
partial writemasks that need to be lowered. While we're here, also call
the main optimization loop if things are lowered to scratch because it
generates address arithmetic that may need to be cleaned up.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10922>

panfrost/ci: Disable G72 jobs for now

There's a bug in ATEST (and LD_TILE?) handling affecting v6, manifesting
as gpu sched timeouts in dmesg and ultimately as random dEQP flakes.
Until this can be sorted out and the G72 jobs are proven to be stable,
we can't have them in the critical merge path.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10923>

panfrost/ci: Disable GLES2 jobs when we run GLES3

I've been meaning to make this change for a while, but I've been holding
out in case dEQP-GLES2 on T860/Bifrost catches a driver issue missed by
dEQP-GLES3 and also missed by T720's dEQP-GLES2 job. Given that hasn't
happened (indeed, dEQP-GLES3 is an effective super set of dEQP-GLES2),
let's save the cycles and skip the redundant jobs.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10923>

lima: implement alpha test

As [1] suggests, 3 lower bits in rsw->multi_sample is alpha test
compare func and bits 16-23 are reference value - uint8,
0.0 is 0, 0.5 is 0x80, 1.0 is 0xff.

Just as on Panfrost, enabled alpha test needs early-Z and pixel kill
to be disabled.

[1] http://web.archive.org/web/20171026123213/http://limadriver.org/Render_State/

Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10879>

ac: ac_gpu_info::has_vgt_flush_ngg_legacy_bug

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10911>

panfrost: Rename G52 product ID 0x7402 as G52r1

As pointed out by Robin Murphy, the new product ID belongs
to IP revision "G52r1".

Signed-off-by: Ezequiel Garcia <ezequiel@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10921>

gallium: remove structure u_resource

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove empty structure u_resource_vtbl

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove u_resource_vtbl::transfer_(un)map

This removes the call indirection.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: split transfer_(un)map into buffer_(un)map and texture_(un)map

The u_resource_vtbl indirection is going to be removed.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove u_resource_vtbl::resource_destroy

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

r600: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

svga: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

virgl: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

i915g: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

nouveau: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

r300: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

radeonsi: stop using u_resource_vtbl::resource_destroy

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove transfer_flush_region from u_resource_vtbl

It's only used by buffers and only zink uses it privately for textures too.
This is part of removing u_resource_vtbl.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove resource_get_handle from u_resource_vtbl

This is the initial step towards removing u_resource_vtbl.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

gallium: remove unused u_default_transfer_unmap

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

zink: don't set u_resource_vtbl

nothing uses it

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10659>

ci: more freedreno flakes

Acked-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10919>

freedreno: Implement TC resource_busy

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10883>

freedreno: Rename internal resource_busy

So I can re-use the name fd_resource_busy.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10883>

vc4: initialize array

This fixes a (rather false) error about accessing an array that it is
uninitialized.

Fixes: 7bc39c8418e ("vc4: Add a dump-the-surface-contents routine.")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4816
Cc: mesa-stable
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com>
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10918>

panfrost: Add PAN_MESA_DEBUG=dirty option

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Dirty track fragment images

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Dirty track viewport descriptor

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Dirty track textures/samplers

draws/second on drawoverhead test 4 is improved by 160% (not a typo)

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Dirty track RSDs

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Dirty all state when batch is set

State uploads are bound to the batch, so if the batch changes, we
invalidate all state. Theoretically this isn't totally optimal but in
practice it should be good enough.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Add the usual clean/dirty helpers

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

panfrost: Define dirty tracking flags

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10888>

frontends/va/picture:Fix wrong reallocation even surface is protected

This patch will avoid reallocation,if surface is already protected.
Fixing the comparision logic of boolean value(true \ flase) with
PIPE_BIND_PROTECTED.

Fixes: 81be8b3c2f2 ("va/picture: make sure destination buffer is protected if needed")

Signed-off-by: SureshGuttula <suresh.guttula@amd.corp-partner.google.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10916>

util/prim_restart: fix util_translate_prim_restart_ib

this was broken for the indirect case if the indirect draw count or
firstIndex was nonzero and also would rewrite the index buffer onto the
wrong offset of the dst buffer

Fixes: 0c85d6c523f ("gallium/util: factor out primitive-restart rewriting logic")
Fixes: 330d0607ed6 ("gallium: remove pipe_index_buffer and set_index_buffer")

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10909>

dri: don't call modifier interfaces when modifiers_count is 0

The wayland EGL platform sets the modifier count to 0 in some cases
to signal that modifiers should not be used, even if a list of modifiers
is present. The loader_dri_create_image helper didn't handle this case
properly and called the modifierful driver interface with a 0 modifier
count, leading to the obvious outcome of the driver being unable to
allocate an image.

Fixes: cb9ae4273d68 ("dri: add loader_dri_create_image helper")
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10903>

nir/lower_io_to_vector: fix per vertex io handling for arrays

The pass was processing the per vertex index from the wrong end
of the array deref chain.

Fixes: bcd14756eec8 ("nir/lower_io_to_vector: add flat mode")

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10798>

iris: Silence warnings about implicit enum type conversions

src/gallium/drivers/iris/iris_state.c: In function ‘iris_create_blend_state’:
src/gallium/drivers/iris/iris_state.c:1284:41: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1284 |          be.SourceBlendFactor           = src_rgb;
      |                                         ^
src/gallium/drivers/iris/iris_state.c:1285:41: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1285 |          be.SourceAlphaBlendFactor      = src_alpha;
      |                                         ^
src/gallium/drivers/iris/iris_state.c:1286:41: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1286 |          be.DestinationBlendFactor      = dst_rgb;
      |                                         ^
src/gallium/drivers/iris/iris_state.c:1287:41: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1287 |          be.DestinationAlphaBlendFactor = dst_alpha;
      |                                         ^
src/gallium/drivers/iris/iris_state.c:1308:28: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1308 |       pb.SourceBlendFactor =
      |                            ^
src/gallium/drivers/iris/iris_state.c:1310:33: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1310 |       pb.SourceAlphaBlendFactor =
      |                                 ^
src/gallium/drivers/iris/iris_state.c:1312:33: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1312 |       pb.DestinationBlendFactor =
      |                                 ^
src/gallium/drivers/iris/iris_state.c:1314:38: warning: implicit conversion from ‘enum pipe_blendfactor’ to ‘enum GFX125_3D_Color_Buffer_Blend_Factor’ [-Wenum-conversion]
 1314 |       pb.DestinationAlphaBlendFactor =
      |                                      ^

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

iris: Return correct enum names from fmt_swizzle

SCS_ONE and ISL_CHANNEL_SELECT_ONE (and SCS_ZERO and
ISL_CHANNEL_SELECT_ZERO) have the same values, so this doesn't matter in
practice.  However, some compiler warning options will cause warnings to
be generated.

src/gallium/drivers/iris/iris_state.c: In function ‘fmt_swizzle’:
src/gallium/drivers/iris/iris_state.c:2177:32: warning: implicit conversion from ‘enum GFX125_ShaderChannelSelect’ to ‘enum isl_channel_select’ [-Wenum-conversion]
 2177 |    case PIPE_SWIZZLE_1: return SCS_ONE;
      |                                ^~~~~~~
src/gallium/drivers/iris/iris_state.c:2178:32: warning: implicit conversion from ‘enum GFX125_ShaderChannelSelect’ to ‘enum isl_channel_select’ [-Wenum-conversion]
 2178 |    case PIPE_SWIZZLE_0: return SCS_ZERO;
      |                                ^~~~~~~~

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

gallium/dri: Allow use of R8G8_R8B8 for YUYV and G8R8_B8R8 for UYVY

v2: Add missing FALLTHROUGH. Caught by CI.

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

nir/lower_tex: Add support for lowering YUYV formats

v2: Rebase on bc438c91d9b ("nir/lower_tex: ignore texture_index if
tex_instr has deref src")

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

mesa: Add R8G8_R8B8 and G8R8_B8R8 formats

These aren't real formats.  They cannot be used with
RenderbufferObjects, Textures, or anything else visible through the GL
API.  The only purpose is to support YUYV textures imported through
EGL_image_external.  There is hardware that can sample from this
subsampled format, but the hardware does not do the colorspace
conversion automatically.

v2: Treat these formats as compressed.  This is a lie, but it seems to
be a less egregious lie than what I was doing before.  It also passes
'ninja test'.

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

intel/isl: Add mappings for PIPE_FORMAT_R8G8_R8B8_UNORM and PIPE_FORMAT_G8R8_B8R8_UNORM

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

gallium/dri: Add Y21x formats

v2: Add all the Y21x tests to the A530 expected fail list.  All of the
YUV image import tests fail on this platform, and nobody has been able
to investigate why.

v3: Update the comment describing the zeroed bits in Y212.  Suggested by
Emma.

v4: Add all Y21x test to the rpi3 expected fail list.

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

egl/dri2: Add Y21x formats

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

util/format: Add Y21x formats

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

gallium/dri: Add Y41x formats

v2: Don't leak __DRI_IMAGE_FOURCC_RGBA16161616 to applications.

v3: Fix typo in __DRI_IMAGE_FOURCC_RGBA16161616 table entry.

v4: Add the Y412 and Y416 tests to the A530 expected fail list.  Many
YUV image import tests fail on this platform, and nobody has been able
to investigate why.

v5: Update the comment describing the zeroed bits in Y412.  Suggested by
Emma.

v6: Add all Y41x test to the rpi3 expected fail list.

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

egl/dri2: Add Y41x formats

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

util/format: Add Y41x formats

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

nir/lower_tex: Add support for lowering Y41x formats

These are similar to AYUV, but the channel ordering is different... in
such a way that there's no RGBA format that will make the channels line
up right.

v2: Rebase on bc438c91d9b ("nir/lower_tex: ignore texture_index if
tex_instr has deref src")

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

mesa/st: Always call st_nir_lower_tex_src_plane if samplers were lowered

Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

egl/dri2: Rely on drm-uapi for DRM_FORMAT defines

All of these formats, and ones that will be added later in this series,
are in the copy of drm_fourcc.h in drm-uapi.

Suggested-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

dri: Fix typo before __DRI_IMAGE_COMPONENTS defines

s/are may be/might be/

Also add a missing comma.

v2: Also s/its/it's/.  Suggested by Nanley.

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

util/format: Delete trailing whitespace

The angry red marks in Emacs were annoying me.

Reviewed-by: Nanley Chery <nanley.g.chery@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9610>

iris: Don't advertise Y-tiled modifiers for scanout buffers on Gfx8

According to isl_gfx7.c:264, the display engine does not support Y
tiled buffers prior to Skylake.  But we exposed I915_FORMAT_MOD_Y_TILED
even when querying for a list of modifiers with PIPE_BIND_SCANOUT set,
which we can't support.  That led to crashes later when we tried to
create such an image, and isl rightly denied it.

Fixes crashes in wflinfo since c03e79d7831f, but the bug exists before
that and it's probably worth a stable backport even without that patch.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4815
Fixes: c03e79d7831 ("loader/dri: hook up createImageWithModifiers2")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10907>

i965: Don't advertise Y-tiled modifiers for scanout buffers on Gfx8-

According to isl_gfx7.c:264, the display engine does not support Y
tiled buffers prior to Skylake.  But we exposed I915_FORMAT_MOD_Y_TILED
even when querying for a list of modifiers with __DRI_IMAGE_USE_SCANOUT
set, which we can't support.  That led to crashes later when we tried
to create such an image, and isl rightly denied it.

This duplicates a bit of code from ISL, but the isl_gfx6_filter_tiling
function that we ought to use to filter things relies on surf_info,
which we don't have at this stage.  This is probably good enough.

Fixes crashes in wflinfo since c03e79d7831f, but the bug exists before
that and it's probably worth a stable backport even without that patch.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4815
Fixes: c03e79d7831 ("loader/dri: hook up createImageWithModifiers2")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10907>

ci: Add test which occasionally times out to lavapipe-vk skips

E.g:
  - https://gitlab.freedesktop.org/mesa/mesa/-/jobs/9881541
  - https://gitlab.freedesktop.org/mesa/mesa/-/jobs/9916369

Signed-off-by: Andres Gomez <agomez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10893>

isl: Work around NVIDIA and AMD display pitch requirements

In the case where we're rendering on the Intel GPU, but displaying
on an AMD and NVIDIA GPU, we need to follow their pitch requirements
for our linear scanout buffers.

Based on a patch by Lionel Landwerlin.

Closes: #4706
Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10895>

i915g: Fix undefined results for TGSI_OPCODE_KILL

The swizzle() requires a proper ureg with the 0/1 swizzle fields filled
in, or our ONE swizzles end up reading .x instead.  This meant that we
were killing based on the incoming value of R0.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Fix writing of undefined depth value if not writing any outputs.

This was showing up as flakes in CI on tests that didn't write any outputs
but did check the depth result.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Stop advertising support for indirect addressing in the FS.

This hardware can't do any form of indirect addressing.  The couple of new
Crashes are the backend falling over when faced with loops/ifs.

Fixes: 8a22064d316e ("i915g: Implement vertex textures.")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Add support for the .Absolute flag on TGSI srcs.

We don't have a way to ask TGSI to not have .Absolute, so lower it in the
backend.

Cc: mesa-stable
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Make the FS for compile failures write red instead of DIFFUSE.

The DIFFUSE thing would write an undefined result if the shader that
failed to compile didn't read VARYING_SLOT_COL0, causing flakes in CI.
This does make 2 piglit tests that were expecting red spuriously pass.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

ci/i915g: Introduce manual testing of i915g using anholt's runner.

Now you can mash the button to see how badly i915g's broken these days.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Fix dumping of the FS in batchbuffers.

We have to increment off of the packet header before disassembling (same
as i915c's copy of this code).

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Switch batchbuffer dumping to mesa_logi().

This makes I915_DUMP_CMD=1 actually do something on non-DDEBUG builds.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

i915g: Disable 3D-pipeline clears.

The 3D-pipeline fast clears try to emit FS constants before an FS is
necessarily bound, causing segfaults in dEQP.  Plus it flushes the whole
batchbuffer so it'll probably be slower anyway.

Fixes: 6358e6371b31 ("i915g: implement hw clear")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

mesa/st: Only use 16-bit ints or floats in the NIR path.

For i915g or other drivers using the draw path, llvmpipe (reasonably)
reports int16 support.  However, if we try to link and translate an int16
shader using st_glsl_to_tgsi, it will throw assertion failures.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10874>

vulkan: Avoid stomping array padding in the MemoryProperties wrapper.

The deqp test for it expects that the unused array elements are untouched,
so make sure they don't get replaced with random stack data.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10737>

venus: vn_GetAndroidHardwareBufferPropertiesANDROID (part 2/2)

Support VkAndroidHardwareBufferFormatPropertiesANDROID query.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

venus: vn_GetAndroidHardwareBufferPropertiesANDROID (part 1/2)

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

venus: add ahb image and buffer properties query support

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

venus: add vn_android_get_ahb_usage helper function

See "AHardwareBuffer Usage Equivalence" in the spec.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

venus: add AHB format and VkFormat conversion helper functions

See "AHardwareBuffer Format Equivalence" in the spec.

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

venus: clean up vn_android api names

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10847>

zink: fix provoking-vertex cap for quads

The way we've doing this currently preserves provoking vertex for quads,
so let's report that to the state-tracker.

This fixes a couple of piglits.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10905>

ir3: Use round-to-nearest-even for fquantize2f16

We're supposed to map a floating-point value too large to be represented
as fp16 to infinity, however round-to-zero naturally rounds it down to
the largest representable fp16 number instead. The blob emits a bunch of
fixup code to work around this, but instead we can just do what all the
other drivers seem to do and use round-to-nearest-even instead.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10897>

panfrost/ci: Add one more flake test for G72

dEQP-GLES31.functional.texture.gather.offset_dynamic.implementation_offset.2d_array.rgba8.size_pot.repeat_mirrored_repeat

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10901>

aco: remove condition operand from branch in invert block

As value numbering only handles logical blocks, this
could lead to invalid IR until insert_exec_mask().
No fossil-db changes.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10894>

radv/cmd_buffer: Fix warning by initializing instance count.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

nine: Fix uninitialized warning in texture9.c

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Axel Davy <davyaxel0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

gallivm: Fix a few uninitialized variable warnings.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

anv: Fix unused function warnings for memory range checks.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

gallium/tessellator: Fix uninitialized variable warnings.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

aco/util: Initialize IDSet::bits_set to zero.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

aco/insert_exec_mask: Fixed unused variable warning in release build.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10806>

aux/trace: add missing return value to trace output

trace_screen_resource_bind_backing() was missing printing of
the boolean return value. Add it.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10900>

radv: Fix compatible image handle type for dmabufs.

Reviewed-by: Joshua Ashton <joshua@froggi.es>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Fixes: 6c83e3ea98b7 ("radv: Add format modifier format queries.")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10891>

radv: fix heap indices when computing the budget

RADV_HEAP_* is the heap type, not the index.

Fixes dEQP-VK.info.device_memory_budget.

Fixes: 08d162f0b57 ("radv: expose 2/3rd of total memory as VRAM and 1/3rd as GTT on APUs")
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/10852>

Revert "radv: Do not access set layout during vkCmdBindDescriptorSets."

This was a temporary workaround for a Baldur's Gate 3 bug which
should now be fixed.

This reverts commit eb104e949ee4b7e0813d14f11a4a952bae48fe80.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3701
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/10839>