ci: build piglit inside baremetal and LAVA's rootfs

v2:
  - Updated the ci-fairy commit to use.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: move API specification to driver instead of test suite

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: move general build commands to their own section

The STRIP_CMD env var is used in several build scripts, not just by
dEQP.

Also, we want to have the rootfs destination directory created so it
also exists for builds that only move directory trees in there.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: add Vulkan piglit traces jobs and remove tracie ones

v2:
  - Rename *-piglit-traces jobs with *-traces.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> [v1]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: only modify LD_LIBRARY_PATH when running the piglit cmd

Doing it earlier may interfere with the optional command wrapper.

v2:
  - Fixed typo (Eric).

Suggested-by: Michel Dänzer <mdaenzer@redhat.com>
Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: add piglit replay jobs and remove tracie ones

Along the way, modify the piglit run script and refactor the way
piglit jobs are generated.

v2:
  - Squashed the commit to remove tracie jobs (Eric).

v3:
  - Extend information in the comments about the need to use a
    running X server for replaying with Vulkan (Tomeu).
  - Do actually fail if the upload doesn't work (Tomeu).

v4:
  - Rename *-piglit-traces jobs with *-traces.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> [v3]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: specify MinIO's host URL in a global variable

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: build piglit in the Vulkan testing image

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: update piglit's version so it features replayer

We also need that summary's aggregate command can use JUnit as backend,
a cross-compilation fix for its CMakeList.txt, a fix for the JUnit
backend and a small change in replayer's compare output.

v2:
  - Updated the piglit commit so it features another couple of needed
    fixes.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com> [v1]
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: spread the usage of the FDO_UPSTREAM_REPO variable

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: install ci-fairy in the testing images

v2:
  - Updated the ci-fairy commit to use.

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: use ephemeral packages when building the build-base image

v2:
  - Update tag for the android_build image (Michel).

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com> [v1]
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

ci: specify source and build directories with CMake

The -S and -B flags were officially introduced in CMake 3.13.

Avoids the following warning:

"
CMake Warning:
  No source or binary directory provided.  Both will be assumed to be the
  same as the current working directory, but note that this warning will
  become a fatal error in future CMake releases.
"

Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6388>

spirv2nir: Add --opengl (-g) argument for OpenGL SPIR-V

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8156>

docs/egl.rst: switch true→enabled

The old "true/false" values for some of the configuration options have been
deprecated in favour of enabled/disabled.

Signed-off-by: Trevor Woerner <twoerner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8173>

ci: Bump deqp to current vulkan-cts-1.2.5.0 also in the Lava jobs

The tag update was forgotten after e384476d1e9 ("ci: Bump deqp to
current vulkan-cts-1.2.5.0").

Noticeably, this introduces 2 more failures in the panfrost-t860 job:
  - dEQP-GLES3.functional.shaders.matrix.inverse.dynamic.lowp_mat2_float_vertex
  - dEQP-GLES3.functional.shaders.matrix.inverse.dynamic.mediump_mat2_float_vertex,Fail

Fixes: e384476d1e9 ("ci: Bump deqp to current vulkan-cts-1.2.5.0")
Signed-off-by: Andres Gomez <agomez@igalia.com>
Reviewed-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8198>

doc/features: mark off tessellation for zink

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8199>

aco: create VMEM clauses slightly more aggressive

Totals from 3325 (2.39% of 139391) affected shaders (NAVI10):
SGPRs: 331528 -> 331056 (-0.14%); split: -0.14%, +0.00%
VGPRs: 306164 -> 337764 (+10.32%); split: -0.02%, +10.34%
CodeSize: 38843180 -> 38865388 (+0.06%); split: -0.04%, +0.10%
MaxWaves: 18908 -> 17028 (-9.94%); split: +0.01%, -9.95%
Instrs: 7423908 -> 7427934 (+0.05%); split: -0.06%, +0.12%
Cycles: 527411756 -> 526388408 (-0.19%); split: -0.21%, +0.02%
VMEM: 1148421 -> 992660 (-13.56%); split: +0.10%, -13.67%
SMEM: 227337 -> 232380 (+2.22%); split: +2.26%, -0.04%
VClause: 146416 -> 111171 (-24.07%); split: -24.10%, +0.03%
SClause: 243674 -> 243689 (+0.01%); split: -0.00%, +0.01%
Copies: 663496 -> 660333 (-0.48%); split: -0.85%, +0.37%
Branches: 223725 -> 223721 (-0.00%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

aco: schedule position exports in the same pass as memory operations

No fossil-db changes.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

aco: fix def-use distance calculation when scheduling.

This change also increases the VMEM_MAX_MOVES
to mitigate some of the scheduling changes.

Totals from 34301 (24.61% of 139391) affected shaders:
SGPRs: 2515440 -> 2552304 (+1.47%); split: -1.25%, +2.71%
VGPRs: 1786676 -> 1794724 (+0.45%); split: -0.31%, +0.76%
CodeSize: 151079856 -> 151209828 (+0.09%); split: -0.06%, +0.15%
MaxWaves: 392454 -> 388966 (-0.89%); split: +0.39%, -1.28%
Instrs: 28870746 -> 28895907 (+0.09%); split: -0.09%, +0.17%
Cycles: 960450680 -> 961315796 (+0.09%); split: -0.09%, +0.18%
VMEM: 19027987 -> 19796223 (+4.04%); split: +7.49%, -3.45%
SMEM: 2434691 -> 2394829 (-1.64%); split: +2.80%, -4.43%
VClause: 551776 -> 543051 (-1.58%); split: -1.73%, +0.15%
SClause: 1230147 -> 1227637 (-0.20%); split: -1.40%, +1.20%
Copies: 1957640 -> 1963617 (+0.31%); split: -1.11%, +1.41%
Branches: 611747 -> 612504 (+0.12%); split: -0.11%, +0.23%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

aco: allow to schedule SALU/SMEM through exec changes

Totals from 16794 (12.05% of 139391) affected shaders (NAVI10):
SGPRs: 757760 -> 762048 (+0.57%); split: -0.39%, +0.95%
VGPRs: 402844 -> 402744 (-0.02%); split: -0.04%, +0.02%
CodeSize: 22290900 -> 22285068 (-0.03%); split: -0.06%, +0.04%
MaxWaves: 294163 -> 294222 (+0.02%); split: +0.03%, -0.01%
Instrs: 4190074 -> 4188513 (-0.04%); split: -0.08%, +0.04%
Cycles: 40685028 -> 40678640 (-0.02%); split: -0.03%, +0.02%
VMEM: 7711867 -> 7704315 (-0.10%); split: +0.28%, -0.38%
SMEM: 942472 -> 1007052 (+6.85%); split: +7.15%, -0.30%
VClause: 92990 -> 92974 (-0.02%); split: -0.03%, +0.01%
SClause: 263700 -> 263810 (+0.04%); split: -0.38%, +0.42%
Copies: 277467 -> 276988 (-0.17%); split: -0.37%, +0.20%
Branches: 45899 -> 45896 (-0.01%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

aco: make pred_by_exec_mask() accessible in other files

and rename to needs_exec_mask().

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

aco: don't emit parallelcopy when switching to WQM.

The reason was an RA bug which has been fixed a while ago.
This change fixes some register demand miscalculations.

Totals from 1013 (0.73% of 139391) affected shaders (NAVI10):
CodeSize: 6050408 -> 6047504 (-0.05%); split: -0.05%, +0.00%
Instrs: 1160533 -> 1159765 (-0.07%); split: -0.07%, +0.00%
Cycles: 8027212 -> 8024140 (-0.04%); split: -0.04%, +0.00%
VMEM: 296195 -> 296091 (-0.04%)
SMEM: 73003 -> 73011 (+0.01%); split: +0.05%, -0.04%
SClause: 37221 -> 37222 (+0.00%)
Copies: 70931 -> 70166 (-1.08%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7903>

zink: export tess shader pipe caps

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: handle partial writes to shader outputs

this is super gross. spirv doesn't provide any facility for doing per-component
writes, which means all components of a value must be written every time

to this end, we need to manually split both the src and dst composites and
do per-component access for each store in order to accurately handle both
non-sequential wrmasks (which could be handled by nir_lower_wrmasks, yes, but
we aren't using it) as well as partial wrmasks

see also mesa/mesa#4006

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: implement passthrough tcs shader injection

GL allows the pipeline to "infer" a tcs shader if a tes shader is bound using
API-specified default values for gl_TessLevelOuter and gl_TessLevelInner,
but VK requires that both shaders be explicitly present

to handle this, create a generic tcs which translates all vs outputs to
invocation-based arrays and copy the appropriate value to the expected tes
input array location. also emit the default inner/outer values as push constants
so we don't have to recompile the shaders whenever the api calls occur

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: add stubs for tess outer/inner level handling

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: add push constant handling to get_storage_class()

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: only run nir_lower_clip_halfz for last vertex processing stage

this lets us remove the calcs to un-convert POS during load

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: add handling for tcs and tes shader states

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: support PIPE_PRIM_PATCHES

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: set tess info in pipeline creation

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: pull xfb info from tess shader when applicable

if it's the last vertex stage then it does the xfb

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: set scoped barrier flag in nir options

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: set up ntv init for tess shaders

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: add handling for tess shader intrinsics

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: add ntv handling for tess shader i/o variables

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

zink: don't always run nir_lower_io_arrays_to_elements_no_indirects

this is automatically run for fs and vs, which is the only place we really
want it

Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8152>

radv: add code that checks if the extension table is sorted correctly

Ported from ANV.

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/8190>

radv: sort the extension table like Khronos

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/8190>

cso: inline cso_construct_key

The x86 asm is a lot shorter and the loop is unrolled.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

cso: remove context and delete_state pointers from all CSOs

We just need them per context, not per CSO. The new delete callback
replaces the per-CSO callbacks.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

st/mesa: don't make a local copy of blend color

This is perfectly safe and nothing bad can happen... and we have also CI.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

cso: don't pass blend_color through cso_context

It's never saved or restored. Redundant state changes are already
filtered out by mesa/main.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

gallium: inline pipe_depth_state to decrease DSA state size by 4 bytes

Depth and alpha states are now packed together, interleaved somewhat.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

gallium: inline pipe_alpha_state to enable better DSA bitfield packing

pipe_alpha_state and pipe_depth_state will be packed together
because they have only a few bitfields each. This will eventually
remove 4 bytes of padding in pipe_depth_stencil_alpha_state.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

gallium: pass pipe_stencil_ref by value (it has only 2 bytes)

This changes pipe_context::set_stencil_ref to pass the parameter by value.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7940>

radv: enable VK_EXT_line_rasterization on GFX9

It was disabled because some CTS failed but they pass now.

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/8189>

turnip: use ir3_compiler_destroy instead of ralloc_free

Fixes: c0f22c3d9406ef3 "freedreno/ir3: add ir3_compiler_destroy()"

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6808>

turnip/kgsl: support VK_KHR_performance_query

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6808>

turnip: enable VK_KHR_performance_query with new debug flag

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6808>

turnip: support multipass for performance query.

To support multipass, querying perf counters happens in several steps
below.

0) There's a scratch reg to set pass indices for perf counters query.
   Prepare cmd streams to set each pass index to the reg at device
   creation time. See tu_CreateDevice in tu_device.c
1) Emit command streams to read all requested perf counters at all
   passes in begin/end query with CP_REG_TEST/CP_COND_REG_EXEC, which
   reads the scratch reg where pass index is set.
2) Pick the right cs setting proper pass index to the reg and prepend it
   to the command buffer at each submit time.
3) If the pass index in the reg is true, then executes the command
   stream below CP_COND_REG_EXEC.

Would need to implement for kgsl in the future.

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6808>

turnip: Implement VK_KHR_performance_query

There are still some commands unimplemented yet.

- vkGetPhysicalDeviceQueueFamilyPerformanceQueryPassesKHR:
  The following patch supports this.

- vkAcquireProfilingLockKHR / vkReleaseProfilingLock
  This patch supports only monitoring perf counters for each submit.
  To reserve/configure counters across submits we would need a kernel
  interface to be able to do that.

Signed-off-by: Hyunjun Ko <zzoon@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6808>

panfrost: Fix panfrost_small_padded_vertex_count for 17 vertices

All odd numbers above 10 need to be rounded up to an even number, so
add one and mask off the least significant bit instead of maintaining
a list of special cases.

Fixes crashes in SuperTuxKart.

Cc: mesa-stable
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8191>

panfrost: Expose ARB_texture_filter_anisotropic on supported GPUs

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8149>

panfrost: Add a gpu_revision argument to panfrost_get_quirks

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8149>

panfrost: Set the anisotropy level when cso->max_anisotropy is set

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8149>

panfrost: Fix the Maximum anisotropy field in the XML

It needs a a minus(1) modifier.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8149>

panfrost: Fix LOD mode field on Bifrost

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8149>

pan/bi: Minor styling cleanup in disasm

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8139>

pan/bi: Remove all-0's termination condition

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8139>

pan/bi: Space out disassembly

Makes things much more legible with some "room to breathe".

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8139>

pan/bi: Allow toggling disassembly verbosity

Verbose mode is especially useful for debugging packing, but otherwise
just gets in the way I find.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8139>

intel/fs: don't spill a register, set by undef

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3941
Signed-off-by: Yevhenii Kolesnikov <yevhenii.kolesnikov@globallogic.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8185>

lvp: Fix vtn warnings about unsupported image read/write without format.

These are just warnings printed to the console and don't affect testcase
pass/fail, but clog up the deqp-runner job logs.

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8188>

ci/deqp: Make sure that we pull in all board-specific xfail/skip/flake files.

When introducing/removing these files, it's easy to forget to update the
yml to point to them.  Instead of requiring the separate update, just have
the runner script pick the right one from a single per-gpu variable.

As a result, we now pick up the new deqp-lvp-skips.txt that was added but
not conected.  This also required moving some bypass flakes from the
shared a630 flakes list to a separate list, which is a feature because now
we'd notice the introduction of flakes to the gmem path.

Fixes: ab79e6b8e336 ("ci: skip failing test on lavapipe")
Acked-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Reviewed-by: Christian Gmeiner <christian.gmeiner@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8147>

radv: Enable DCC in the GENERAL layout on GFX10+.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7004>

radv: Disable DCC explicitly for incompatible copies.

If we enable DCC for GENERAL we cannot set the layout to GENERAL to
disable DCC, so do it explicitly.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7004>

radv: Add option to disable DCC in renderpasses without layout.

If DCC is enabled for GENERAL then we cannot disable DCC by going
to the GENERAL layout.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7004>

radv: Never allow fast clears on DCC images that are not compressed.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7004>

radv: Don't skip layout transitions that only differ in render loop.

This can result in meaningful compression changes so we shouldn't skip.

Fixes: 66131ceb8bc "radv: Pass through render loop detection to internal layout decisions."
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7004>

radv: ignore the mutable bit for TC-compatible HTILE

All depth/stencil formats are incompatible each others, so the
mutable bit and the image format list can be ignored.

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/8126>

glsl: remove unused state variable

Because compile_shader function state variable
not determine whether the compilation is successful.

Signed-off-by: cheyang <cheyang@bytedance.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8178>

freedreno/a6xx: bump varyings limit

GL_MAX_VARYING_COMPONENTS is bumped to 124 since it should
not include the components of gl_Position. (Same as in blob)

GL_MAX_*_OUTPUT_COMPONENTS is bumped to 128, only
GL_MAX_GEOMETRY_INPUT_COMPONENTS is 64. (Same as in blob)

Per GL 3.2 spec the minimum of:
 - GL_MAX_GEOMETRY_OUTPUT_COMPONENTS is 128
 - GL_MAX_FRAGMENT_INPUT_COMPONENTS is 128
 - others is 64

Per ARB_tessellation_shader the minimum of:
 - GL_MAX_TESS_CONTROL_*_COMPONENTS to be 128
 - GL_MAX_TESS_EVALUATION_*_COMPONENTS to be 128

Allows passing of:
 gl-3.2-minmax
 arb_tessellation_shader-minmax

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7917>

freedreno/a6xx: Fix SP_HS_UNKNOWN_A831 value and document it

It appears that storage for varyings in a wave has an upper
limit of wavesize * max_a831 where max_a831 is 64.
Exceeding the limit seam to force gpu to reduce primitives
processed per wave, at least calculations make sense with
such interpretation.

With blob SP_HS_UNKNOWN_A831 never exceeds 64 and setting
it to 65 in freedreno leads to a hang.

On A630 tests (patch_size=3 + gl_Position + array of vec4)
have shown such relation:

| Num of vec4 | A831 | PC_HS_INPUT_SIZE |
|-------------|------|------------------|
| 1           | 0x10 | 0xc              |
| 2           | 0x14 | 0xf              |
| 3           | 0x18 | 0x12             |
| 4           | 0x1c | 0x15             |
| 5           | 0x20 | 0x18             |
| 6           | 0x24 | 0x1b             |
| 7           | 0x28 | 0x1e             |
| 8           | 0x2c | 0x21             |
| 9           | 0x30 | 0x24             |
| 10          | 0x34 | 0x27             |
| 11          | 0x38 | 0x2a             |
| 12          | 0x3c | 0x2d             |
| 13          | 0x3f | 0x30             |
| 14          | 0x40 | 0x33             |
| 15          | 0x3d | 0x36             |
| 16          | 0x3d | 0x39             |
| 17          | 0x40 | 0x3c             |
| 18          | 0x3f | 0x3f             |
| 19          | 0x3e | 0x42             |
| 20          | 0x3d | 0x45             |
| 21          | 0x3f | 0x48             |
| 22          | 0x3d | 0x4b             |
| 23          | 0x40 | 0x4e             |
| 24          | 0x3d | 0x51             |
| 25          | 0x3f | 0x54             |
| 26          | 0x3c | 0x57             |
| 27          | 0x3e | 0x5a             |
| 28          | 0x40 | 0x5d             |
| 29          | 0x3c | 0x60             |
| 30          | 0x3e | 0x63             |
| 31          | 0x40 | 0x66             |
|-------------|------|------------------|

Brief tests with high patch sizes also confirm that formula
matches blob behaviour.

A831 is not a limit for storage available for one thread, so
naming it as SP_HS_WAVE_INPUT_SIZE would make more sense.

Fixes: 47e2c195 "freedreno/a6xx: Program state for tessellation stages"

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7917>

ir3: Allow tesselation to use all 32 varying slots

POS, PSIZE, CLIP_DIST0, and CLIP_DIST1 have their own predefined
indices, map's size should take this into account.

Fixes: 9e063b01 "ir3: Switch tess lowering to use location"

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7917>

freedreno/a6xx: Fix assert which checks the count of shader outputs

The actual max count is 32 which corresponds to 128 output components.

Fixes: 2251a434 "freedreno/a6xx: Write multiple regs for SP_VS_OUT_REG and SP_VS_VPC_DST_REG"

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7917>

mesa: fix layered framebuffer attachment target check

Current code was making the layer target checking for depth and stencil
attachments as well while the check in spec is specified only for color.

Cc: mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3980
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8102>

lavapipe: add transform feedback support

This adds support for EXT_transform_feedback which can be used by
zink to enable more CI

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7981>

llvmpipe: handle SO statistics multi value query copy. (v2)

This is hard to abstract using the vulkan interface, so just
add support for copying both values in the llvmpipe backend
for the lavapipe frontend.

v2: use a loop

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7981>

llvmpipe: add support for vulkan streamout offset hook

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7981>

gallium: add an api to retrieve pipe offsets

This is needed to implement the vulkan transform feedback pause
resume functionality

Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7981>

d3d12: Fix memory leak if create_root_signature failed.

Fix defect reported by Coverity Scan.

Resource leak (RESOURCE_LEAK)
leaked_storage: Variable data going out of scope leaks the storage it points to.

Fixes: 2ea15cd661c ("d3d12: introduce d3d12 gallium driver")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Erik Faye-Lund <kusmabite@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8170>

d3d12: Fix memory leak if create_gfx_pipeline_state failed.

Fix defect reported by Coverity Scan.

Resource leak (RESOURCE_LEAK)
leaked_storage: Variable data going out of scope leaks the storage it points to.

Fixes: 2ea15cd661c ("d3d12: introduce d3d12 gallium driver")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8150>

panfrost: Fix CLAMP wrap mode

MALI_WRAP_MODE_CLAMP doesn't work fully on either GPU generation, so
use other wrap modes instead in some cases.

With nearest filtering, Midgard only clamps to the edge for two of the
edges, and uses the border colour for the other two. Using the clamp
mode on Bifrost causes broken rendering and/or GPU faults.

Fixes piglit test "texwrap" on both Midgard and Bifrost, and fixes
Chromium B.S.U. rendering on Bifrost.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8176>

panfrost: Fix precise occlusion queries on Bifrost

Bifrost doesn't need the sample count to be divided by four for
non-MSAA framebuffers.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8172>

docs/panfrost: Mention the IRC channel

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Suggested-by: Icecream95 <ixn@disroot.org>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8148>

docs/panfrost: Document building Panfrost

Pretty similar to anything else in Mesa, although kmsro is required and
LLVM is not.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8148>

docs/systems: Update Panfrost link

There's not much on the standalone site anymore (it dates back from
before Panfrost was upstreamed), let's just keep everything in Mesa.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8148>

docs: Update Panfrost in the source tree

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8148>

docs/panfrost: Fix comment about Lima

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Suggested-by: Vasily Khoruzhick <anarsoul@gmail.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8148>

vc4: Fix typos.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8120>

radv: remove useless push constants data when resolving ds attachments

Depth/stencil resolves are only allowed inside a subpass, which means
the offset is always 0 and the draw/dispatch covers the whole image.

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/8127>

spirv: Remove more dead variables

SPIR-V modules can have multiple shaders (including of the same
stage), but the global variables are all declared for the whole
module.  This can result in variables with same Binding but
incompatible types, so those need to be removed before we use.

Previously, a similar issue but with a narrower scope was fixed by
6775665e5ee ("spirv: Eliminate dead input/output variables after
translation.").

This patch depends on the previous patch that prevents variables used
only in pointer initializers to be considered dead.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3686
Fixes: 3a266a18 ("nir/spirv: Add support for declaring variables")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8133>

nir: Consider pointer initializers in nir_remove_dead_variables

Between the creation of a shader (from GLSL or SPIRV frontends) and
nir_lower_variable_initializers is called, variables may refer to
other variables for initialization.  Those referred variables need to
be kept alive, so consider that in the pass.

Fixes: 7acc81056f7 ("compiler/nir: Add support for variable initialization from a pointer")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8133>

nir: Remove unused parameter in remove_dead_var_writes

And destroy the live set earlier in nir_remove_dead_variables.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8133>

nir: Fix outdated name in comment

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8133>

spirv: Implement OpArrayLength for OpenGL

Uses same NIR intrinsic as glsl_to_nir.  Make it an option so it is
easy later to move Vulkan drivers incrementally to use it.

Fixes piglit test spec/arb_gl_spirv/execution/ssbo/unsized-array-length.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3691
Fixes: 15e43907 ("iris: Enable ARB_gl_spirv and ARB_spirv_extensions")
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8136>

zink: fix gl_SampleMaskIn handling

this is an array[1], so we need to do some extra legwork here even though
probably it'd work fine even without it

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7490>

zink: fix rp hash table

hash table keys for inserted items have to be valid memory ranges for the
lifetime of the corresponding entry, so using a stack-allocated key like this
is broken and doesn't accurately return the correct renderpass

Fixes a872f4636924: ("zink: cache render-passes")

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8011>