vulkan/util: generate vk_dispatch_table that combines all dispatch tables

Zink uses this, as it doesn't need to differentiate all the entrypoints.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11045>

radeon/vcn: enable parsing support for st_rps_bits

Set st_rps_bits in hevc message buffer and set corresponding flag to indicate
that st_rps_bits will be used for parsing the short_term_ref_pic_set structure.

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10889>

frontends/vdpau: disable UseStRpsBits for vdpau hevc

vdpau interface doesn't provide st_rps_bits, it uses NumDeltaPocsOfRefRpsIdx
instead. So disabling the flag to indicate st_rps_bits will not be used.

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10889>

frontends/va: get st_rps_bits from VA pic param hevc

Get st_rps_bits from VAPictureParameterBufferHEVC, and set the flag that
indicates st_rps_bits will be used for parsing the short_term_ref_pic_set
structure

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10889>

vl: add st_rps_bits for HEVC decode

st_rps_bits is used for accelorater to skip parsing the short_term_ref_pic_set
structure, which is needed for HEVC decode.

Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10889>

travis: Download XQuartz from GitHub.

Bintray was sunsetted on May 1st, 2021.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11026>

util/prim_restart: revert part of bad fix

When drawing using util_translate_prim_restart_ib, zink explicitly
ignores pipe_draw_start_count_bias::start, because
util_translate_prim_restart_ib used to create a new index-buffer without
padding at the start.

This makes a lot of sense, because creating a padded index buffer is
just wasteful.

So let's walk back on the choice of starting to pad the output buffer.

Fixes: 1272c2e0524 ("util/prim_restart: fix util_translate_prim_restart_ib")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4851
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11059>

zink: add support for string-markers

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11004>

zink: untangle have_EXT_debug_utils and ZINK_DEBUG_VALIDATION

EXT_debug_utils is useful for more than just validation, so let's
untangle these a bit.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11004>

zink: implement half-float packing

This cap isn't really optional for drivers that implement GL SPIR-V
support, so let's just implement it.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4846
Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11057>

zink: mark 2d-arrays as cube-compatible

OpenGL allows reinterpreting 2D array textures as cubemaps,
so we need to always set the cube-compatible flag for them.

This fixes a piglit test on Lavapipe.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10913>

zink: only mark resources as cube-compatible if supported

We need to check if the number of samples are valid for the image
first if we've going to set the cube-compatible bit.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10913>

lavapipe: do not interpret cube-compatible as cubemap

Now that we handle the cubemap-array views properly, we no longer need
to do this. It's allowed in Gallium to create cubemap views of 2d
arrays, so this should work fine.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10913>

lavapipe: handle cube-array image-views

It seems this was missed in the past.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10913>

venus: fix mismatched bo mmap_size for export and multiple imports

Spec requires apps to use the size returned from
vkGetAndroidHardwareBufferPropertiesANDROID for AHB import, which
includes dedicated image/buffer import and non-dedicated buffer import.

Spec requires venus to use the size from image and buffer memory
requirement for dma_buf fd import if it's dedicated. If not dedicated,
the actual payload size should be used.

For AHB export allocation of VkImage, the app provided size will be 0,
and internally we must use the size from image memory requirement.

For AHB export allocation of VkBuffer, the app provided size comes from
buffer memory requirements and it can be smaller than the actual AHB
size due to page alignment. Internally that's the size we must use.

For AHB import, app must use the size from AHB prop query. Internally,
we have to override that with the size from memory requirement if the
import operation is dedicated to a VkImage or a VkBuffer. If not
dedicated, the actual payload size should be used.

The not working scenario is:
1. App creates an AHB backed VkBuffer, and the exported AHB size is
   larger than the buffer memory requirement (very common).
2. App imports the AHB without a dedicated VkBuffer. Then the entire
   AHB payload will be imported and the mmap_size might increase.

Test: dEQP-VK.api.external.memory.android_hardware_buffer.*

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/11055>

venus: refactor for property query of dma_buf fd

With a TODO to route host storage size instead of that from the guest.

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/11055>

venus: fix misaligned bo_flags between import and query

For importing an dma_buf fd, export info is not required. Leaving the
bo_flags missing VIRTGPU_BLOB_FLAG_USE_SHAREABLE bit as well as the
VIRTGPU_BLOB_FLAG_USE_CROSS_DEVICE bit. Upon querying fd properties,
DMA_BUF handle type will be specified which generates a new bo_flags
not matching the prior one.

This patch aligns the above 2 bits for the dma_buf import path.

Test: vkGetAndroidHardwareBufferPropertiesANDROID should succeed with
      a valid AHB with AHARDWAREBUFFER_FORMAT_BLOB format.

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/11055>

venus: rename dmabuf to dma_buf when it represents a type

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/11055>

svga: fix texture rectangle sampling when no sampler view declaration is defined

 It is valid to not have a sampler view declaration for the corresponding
 sampler in a TGSI shader, and hence we should not rely on the sampler view
 declaration to determine if we need to adjust the unnormalized coordinates
 for texture rectangle sampling.

 This patch is to prep for tgsi shaders that are translated from nir which
 in many cases do not issue sampler view declarations.

 Fixes: 584b1070372a0e ("st/mesa: Drop the TGSI paths for drawpixels and use nir-to-tgsi")

Reviewed-by: Neha Bhende <bhenden@vmware.com
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11011>

svga: Use shader_key info to declare resources if TGSI shader is missing it

Sometimes, TGSI shader doesn't have SVIEW declaration if it is not
utilize in shader. In such cases, declare those resources with the
help of information stored in shader key.

Fixes: 584b1070372a0e ("st/mesa: Drop the TGSI paths for drawpixels and use nir-to-tgsi")

Tested with piglit, gleretrace

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11011>

svga: Add target and sampler_return_type info into shader key

Fixes: 584b1070372a0e ("st/mesa: Drop the TGSI paths for drawpixels and use nir-to-tgsi")

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11011>

anv_android: Add missing type

Fixes: 2797c75426cf ("anv: fix AHB leak upon exportable allocation")
Signed-off-by: Roman Stratiienko <r.stratiienko@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11068>

venus: ignore pInheritanceInfo when we should

pInheritanceInfo should be ignored when the command buffer is primary.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4850
Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11067>

venus: remember cmd buffer level and queue family

We need them for wsi queue ownership transfer and others.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11067>

venus: add struct vn_command_buffer_builder

We are going to remember more states.

Signed-off-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11067>

lavapipe: add more format mappings for vertex buffer formats

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10957>

mesa: don't save/restore VAO NumUpdates and IsDynamic to fix update tracking

NumUpdates is used to indicate whether a VAO is static or dynamic, but if
we restored an older value, it could incorrectly indicate that it's not
dynamic.

This fixes a small performance issue with torcs.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

mesa: don't call FLUSH_VERTICES in glPopClientAttrib

These states don't affect draws, so any enqueued Begin/End vertices don't
need flushing.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

mesa: optimize unreferencing VBOs in glPopClientAttrib

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

mesa: optimize glPush/PopClientAttrib for GL_CLIENT_VERTEX_ARRAY_BIT

This improves performance in torcs by 6%.

The idea is to skip saving and restoring vertex attribs and bindings that
have never been changed.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

mesa: don't call _mesa_set_draw_vao in glPushClientAttrib

Pushing states doesn't affect draws.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

mesa: move _mesa_copy_vertex_attrib/buffer functions to their only use

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10994>

aux/indices: employ Delete The Code methodology

the util functions will assert for unknown prim types, so the switches
are redundant now

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10965>

aux/indices: break out index count conversion into separate function

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10965>

aux/indices: break out index size conversion to separate function

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10965>

aux/indices: break out primitive type conversion to separate function

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10965>

llvmpipe: ci updates

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11049>

gallivm: fix oob imageLoad with formats that have <4 components

it's expected that these loads will have full alpha

fixes spec@arb_shader_image_load_store@invalid

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11049>

aux/trace: avoid deadlock in screen::flush_frontbuffer hook

this hook usually requires mapping a resource, which will trigger a
trace dump internally, which requires locking the output mutex, so
finish the trace dump before calling it

Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11052>

aco: Eliminate SALU comparison when SCC can be used instead.

For example:

s0, scc = s_and_u32 ...
scc = s_cmp_eq_u32 s0, 0
p_cbranch_sccz

is turned into:

s0, scc = s_and_u32 ...
p_cbranch_sccnz

Fossil DB results on Sienna Cichlid:

Totals from 85267 (56.91% of 149839) affected shaders:
CodeSize: 202539256 -> 202237268 (-0.15%)
Instrs: 38964493 -> 38888996 (-0.19%)
Latency: 750062328 -> 749913450 (-0.02%); split: -0.02%, +0.00%
InvThroughput: 167408952 -> 167405157 (-0.00%)

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

aco: Use s_cbranch_vccz/nz in post-RA optimization.

A simple post-RA optimization which takes advantage of the
s_cbranch_vccz and s_cbranch_vccnz instructions.

It works on the following pattern:

vcc = v_cmp ...
scc = s_and vcc, exec
p_cbranch scc

The result looks like this:

vcc = v_cmp ...
p_cbranch vcc

Fossil DB results on Sienna Cichlid:

Totals from 4814 (3.21% of 149839) affected shaders:
CodeSize: 15371176 -> 15345964 (-0.16%)
Instrs: 3028557 -> 3022254 (-0.21%)
Latency: 21872753 -> 21823476 (-0.23%); split: -0.23%, +0.00%
InvThroughput: 4470282 -> 4468691 (-0.04%); split: -0.04%, +0.00%

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

aco: Introduce a new, post-RA optimizer.

This commit adds the skeleton of a new ACO post-RA optimizer,
which is intended to be a simple pass called after RA, and
is meant to do code changes which can only be done
after RA.

It is currently empty, the actual optimizations will be added
in their own commits. It only has a DCE pass, which deletes
some dead code generated by the spiller.

Fossil DB results on Sienna Cichlid:

Totals from 375 (0.25% of 149839) affected shaders:
CodeSize: 2933056 -> 2907192 (-0.88%)
Instrs: 534154 -> 530706 (-0.65%)
Latency: 12088064 -> 12084907 (-0.03%); split: -0.03%, +0.00%
InvThroughput: 4433454 -> 4432421 (-0.02%); split: -0.02%, +0.00%
Copies: 81649 -> 78203 (-4.22%)

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

aco: New writeout overloads for the test framework.

These will be used by future tests.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

aco: Add Operand(Temp, PhysReg) constructor.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

aco: Don't DCE instructions that write non-temps, eg. exec.

No Fossil DB changes.
This commit makes DCE usable after RA.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7779>

Revert "ci: Configure DUTs for max performance"

This reverts commit eef5409df4c3f1192687eaa6157875e4650c1206.

I suspect this has caused a lot of the CI instability today -- some flakes
were already added, but the a630-traces job is still flaking.  Revert
until a fix makes it stable.

Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11024>

radv: fix generating hang reports if mutable descriptors are used

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/11031>

v3dv: remove unused v3dv_zs_buffer_from_vk_format

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11050>

mesa: fix error set for glCompressedTexSubImage calls

Desktop and ES expect a different error code here when dealing with
generic compressed format tokens. This fixes failures with upcoming
new tests for compressed texture related API calls.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11009>

mesa/st: Don't assert !unify_interfaces in the passthrough edge flags case

Technically, this is correct, and I think it's the right long-term
solution.  However, Iris has code that undoes the damage caused by
nir_lower_passthrough_edgeflags (see iris_fix_edge_flags in
iris_program.c), so it should be safe to do the lowering here.

I'm not marking this as closing mesa#4838 because I think we should move
the Iris code up to here, make a different version of the NIR pass, or
something different... to properly fix this problem.  In the mean time,
this gets a bunch of tests to stop crashing. :)

Fixes: a76ec17f12c ("mesa/st: Fix iris regression with clip distances.")
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11013>

turnip: Make sure that SNORM blits don't clamp ambiguous -1.0 values.

The CTS expects that some paths transfer SNORM data exactly, but the HW
will clamp 0x80 values to 0x81 in the process.  We can treat snorm as
unorm, though, and get working compression without the clamping happening.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10735>

turnip: Reorganize copy_format()'s switch statement.

Now that we need FALLTHROUGH macros we weren't saving much by falling
through, and things were weirdly ordered anyway with depth intermixed with
color formats and the default case tucked in the middle of the switch
statement.  Replace with pretty obvious ordering of normal color, planar
color, depth, then default.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10735>

nv50/ir: don't optimize shl(mul_hi, a) to mul_hi

Fixes shader_storage_buffer_object.advanced-unsizedArrayLength* CTS tests

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11021>

nv50/ir: when constant folding shl(mul, a) we need to copy muls type

Also clear subop while at it as we might convert a shl.wrap to mul.

Signed-off-by: Karol Herbst <kherbst@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11021>

panfrost: Performance configuration

Configure a panfrost performance structure with tables of categories and
counters for the current product id. An array for storing counter values
read from the GPU is also managed by this structure. A generic read
function can be used to retrieve the value of a counter from the conter
values array.

v2: Generate tables instead of calling register functions.
v3: Simplify counter read function and `pan_gen_perf.py` write method.
v4: Accumulate counter values from all cores.
v5: Wrap `STATIC_ASSERT`s within unused functions.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10844>

panfrost: Counter definitions

Add Mali events XML files generated with Panfrost HWC helper.
https://gitlab.freedesktop.org/fahien/panfrost-hwc-helper/

v2: Restore license headers.
v3: Fix shader core and memory system counter offsets.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10844>

st/mesa: fix clearing of 1D array textures

st_gl_texture_dims_to_pipe_dims isn't usable in this case since its
API doesn't match the usage so fix up the dimensions manually.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3646

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10523>

ci/freedreno: Skip Portal 2 trace on a630, due to flakiness

Sometimes the reticle is missing. Skip it for now to keep the number of
pipeline failures due to flakes low.

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11033>

radv: fix missing default state for DB_DFSM_CONTROL

Fixes: 69ae02151d7 ("radv: remove DFSM")
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/11028>

zink: improve unsupported feature warning message

this should print the device name and the missing features in order to
be less opaque

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10959>

zink: use actual const for const offset

When we emit constants, we don't know what type they'll be used as, so
we just emit them as uint, and then bitcast them to whatever we need.

But this isn't a good idea for ConstOffset, which needs to actually be a
const value, and not a const value bitcasted. So we sadly have to
open-code the const emitting here to avoid the problem.

Fixes: e963d35efe1 ("zink: use ConstOffset for nir_tex_src_offset")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4831
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11032>

v3dv: add v3dv_GetImageSparseMemoryRequirements back

This one is not implemented in the common dispatch handler in terms
of its KHR_get_memory_requirements2 version, so the driver needs
to implement it.

Fixes: d87afc1acc ('v3dv: implement VK_KHR_get_memory_requirements2')
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11038>

docs: drop clayton from intel-ci notice

As per his request.

Reviewed-by: Clayton Craft <clayton@craftyguy.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10975>

docs: update another IRC reference

In the last round of IRC link updates, #freedesktop hadn't moved yet. Now
it has, so let's update the reference.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Clayton Craft <clayton@craftyguy.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10975>

docs: update another IRC reference

I didn't initially update this one because craftguy hadn't moved to OFTC
when I wrote the last patch. But now he has, so let's also update this
reference.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Clayton Craft <clayton@craftyguy.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10975>

Partial revert of "ci: Add a manual job for tracking the performance of Freedreno"

This reverts commit 8e470457de5b6e7903ceea0efaa99cb6e8b611f0.

Drop that jobs, as it's sometimes causing a gitlab-ci.yml parse error
that isn't readily reproducible:

Found errors in your .gitlab-ci.yml:

'a630-profile-traces' job needs 'arm_test' job but it was not added to the pipeline
'a630-profile-traces' job needs 'meson-arm64' job but it was not added to the pipeline

Signed-off-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11030>

v3dv: trivially handle VK_STRUCTURE_TYPE_EXPORT_MEMORY_ALLOCATE_INFO_KHR

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11002>

v3dv: implement VK_KHR_dedicated_allocation

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11002>

v3dv: keep track of whether an image may be backed by external memory

Such images will always require dedicated allocations.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11002>

v3dv: implement VK_KHR_get_memory_requirements2

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11002>

Revert "util: disable glthread in CSGO"

This reverts commit 40c93e2f455d788cc920f395b899e70f730cf5a3.

The trust factor issue has now been resolved [1]

[1] https://github.com/ValveSoftware/csgo-osx-linux/issues/2630

Reviewed-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/10948>

virgl:Fix the leak of hw_res used as fence

Fence destroy hw_res not dec reference.Lead leak.
Call virgl_drm_resource_reference() to release hw_res
instead of calling virgl_hw_res_destroy() directly.

Fixes: c54fb6ef3d8 ("virgl: Don't destroy resource while it's in use.")

Signed-off-by: cheyang <cheyang@bytedance.com>
Reviewed-by: Chia-I Wu <olvaffe@gmail.com>
Reviewed-by: Lepton Wu <lepton@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11008>

gallium/u_vbuf: add a fast path to skip refcounting for uploaded user buffers

This improves performance by 23% with radeonsi and Ryzen 3900X running
the game torcs.

All vertex buffers must be user buffers to get this.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10992>

radeonsi: clamp clear_buffer values using new util helper

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10822>

zink: clamp clear_buffer values

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10822>

gallium/aux: add helper for pre-clamping clear_buffer value to dword

copied from radeonsi

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10822>

lavapipe: also ignore multiple pipeline barriers in succession

this just flushes repeatedly, which is pointless

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10762>

lavapipe: skip "pipeline barriers" if they're first or last in a cmdbuf

barriers just trigger a full flush, which is unnecessary if it's:
* the first cmd in a cmdbuf, as the previous cmdbuf will have already
  flushed and so this is flushing nothing
* the last cmd in a cmdbuf, as there will be a flush immediately after
  returning from this function

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10762>

llvmpipe: stop accessing pipe_resource::screen internally

this breaks driver tracing

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11017>

llvmpipe: store a screen pointer in resource struct

directly accessing the pipe_resource::screen pointer crashes trace,
so it's necessary to have an unwrapped pointer available if such things
are needed

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11017>

llvmpipe: only dump tgsi shaders if they're actually tgsi shaders

this crashes otherwise

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11016>

llvmpipe: split out scene surface init into separate function

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11015>

llvmpipe: split out scene surface info into separate struct

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11015>

lavapipe: moar @optimize

without enough nir optimizing, some shaders fail to reduce as they should,
resulting in the wrong output

fixes dEQP-VK.graphicsfuzz.spv-stable-mergesort-flatten-selection-dead-continues

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11019>

llvmpipe: remove clamping to [0,1] for tri offset

this was added in 315f8daab159 without much detail, then refactored
a bunch of times, and there's even now been a comment added questioning
whether it's correct: it does not appear to be.

fixes lavapipe cts cases:
dEQP-VK.draw.inverted_depth_ranges.depthclamp_deltaone_bias_clamp_neg
dEQP-VK.draw.inverted_depth_ranges.nodepthclamp_deltaone_bias_clamp_neg

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11012>

ci/freedreno: Turn off default a530 quick_gl testing, do full quick_shader.

The quick_gl set is too unstable -- even when I switched to a consistent
set of tests, and added lots of flakes, I keep getting new ones as some
test (unclear which, but it's like 7-8 minutes into the run) kills other
innocent ones.  Until we get per process pagetables, disable this testing
by default.

However, now that we've freed up a board that was doing quick_gl, we have
time to do all of quick_shader so that piglit uprevs don't reshuffle the
test list and expose new failures.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/11018>

venus: advertise VK_ANDROID_external_memory_android_hardware_buffer

Besides advertising AHB support, for Android, we do not advertise
KHR_external_memory_fd, EXT_external_memory_dma_buf or
EXT_image_drm_format_modifier so that AHB remains the sole channel for
external resource sharing.

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/10960>

venus: support AHB external format for sampler YCbCr conversion

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/10960>

venus: implement vn_GetMemoryAndroidHardwareBufferANDROID

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/10960>

venus: implement AHB allocation and import (part 2/2)

TODO left to fix plane count > 1 case for external memory.

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/10960>

venus: implement AHB allocation and import (part 1/2)

This patch refactors the struct look up logic for memory allocation, and
it prepares the necessary info for ahb allocation and import.

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/10960>

venus: refactor device memory fd import

This patch wraps up fd import logic, which will be used later by ahb
import. This patch also removes a redundant check in need_bo.

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/10960>

venus: implement image creation for ahb handle type

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/10960>

venus: prepare image creation helpers for AHB

Store image create info for deferred creation of AHB image.

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/10960>

venus: fix AHB image format properties query

1. bail early if there's no compatiable AHB format
2. check against the corresponding drm format modifier

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/10960>

venus: fix vn_GetAndroidHardwareBufferPropertiesANDROID

Append a VK_FORMAT_FEATURE_MIDPOINT_CHROMA_SAMPLES_BIT bit to pass cts.

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/10960>

venus: complete the format conversion between AHB and Vulkan

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/10960>

venus: tiny refactor of vn_android_get_gralloc_buffer_info

Return false if the queried drm format modifier is invalid.

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/10960>

radv: move all game workarounds to drirc

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/10978>

util/drirc: use application_name_match for the SotTR RADV workaround

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/10978>