egl: allow INVALID format for linux_dmabuf

As per
https://gitlab.freedesktop.org/wayland/wayland-protocols/commit/fb9b2a87317c77e26283da5f6c9559d709f6fdcd,
the compositor may advertise DRM_FORMAT_MOD_INVALID as a supported
modifier. This patch makes mesa recognize this fact and allow
linux_dmabuf usage with the INVALID modifier in this case.

In case the driver doesn't support modifiers, we can still use
linux-dmabuf protocol instead of the legacy wl_drm interface to create
wl_buffers. This will help compositors to handle these buffers better.

In this commit, the INVALID modifier is allowed to be added to the list
of supported modifiers, and create_wl_buffer will be able to use
linux_dmabuf with an INVALID modifier if the compositor advertised it as
supported.

Signed-off-by: Ivan Molodetskikh <yalterz@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2147>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/2147>

lima: add RGBA5551 and RGBA4444 formats

We also need to set channel_layout in pp_frame reg (previously known as
foureight) depending on cbuf format.

Reviewed-by: Andreas Baierl <ichgeh@imkreisrum.de>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3972>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3972>

ci: Add a disabled-by-default job for GLES3 testing on db410c.

Now that we have 7 (soon 8) boards available, there's capacity to be
testing GLES 3.0.  However, due to (it looks like) buffer overflows in the
driver, we end up with flaky test results: 1/60 jobs spuriously failed,
and another 6/60 jobs reported flakes.  At 6 jobs per pipeline, that's way
too high of a failure rate to enable for non-freedreno developers.  Leave
the job present but disabled so that we can do manual test runs for
regressions.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3661>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3661>

ci: Switch testing on db410c over to LAVA.

This should get us better stability of the db410c boards by having a
smaller per-board software stack, with no disks involved (just initramfs).
Additionally, the new cluster is 7 (soon 8) db410cs, while currently the
docker cluster only has 1/4 of its db410cs still running.

Unfortunately, we have to prepare the fastboot boot image during the ARM
drivers build stage, because LAVA relies on publicly available URLs for
the images to load into the bootloaders of the boards, and the only thing
we have for that is gitlab's artifacts.

Note that this testing relies on the boards being freshly flashed with the
linaro v136 firmware to pick up the initramfs size fixes and to stop the
boot at fastboot.

Reviewed-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3661>

r600/sfn: Don't try to catch exceptions, the driver doesn't throw any

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3974>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3974>

r600/sfn: Use static_cast when type is already known

In all these cases the type was tested before based, so don't use
dynamic_casts.

Closes #2566

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Mauro Rossi <issor.oruam@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3974>

r600/sfn: Avoid using dynamic_cast to identify type

v2: Fix typo (maurossi)

Related: #2566

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Mauro Rossi <issor.oruam@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3974>

docs/features: add v3d driver

Now that we bumped the GLES version to 3.1, it makes even more sense
to include the driver here.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2507
Reviewed-by: Jose Maria Casanova <jmcasanova@igalia.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3810>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3810>

aco: pass vars by const &

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3935>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3935>

Revert "gitlab-ci: disable panfrost runners"

The infrastructure issues, caused by building electrical works gone
wrong, have been fixed, and the Panfrost LAVA runners are available
again.

This reverts commit a86662c44d7cb2541c3f613805533064219ad11f.

Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4019>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4019>

aco: Minor optimization in spill_ctx constructor

'register_demand' is passed by value and only copied once; consider moving it to avoid unnecessary copies

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3968>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3968>

radv: update entrypoints generation from ANV

It's a massive rework loosely based on ANV. This introduces separate
dispatch tables for the instance, physical device and device objects.

This will help for implementing internal driver layers for SQTT.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3930>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3930>

radv/sqtt: add support for GFX10

All SQTT registers were moved to privileged space on GFX10, to emit
them we need a workaround with COPY_DATA.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4018>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4018>

ac/registers: add definitions for thread trace on GFX10

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/4018>

radv/sqtt: update SPI_CONFIG_CNTL.EXP_PRIORITY_ORDER value

It should be 3.

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/4018>

radv/sqtt: do not assume that the number of shader engines is 4

It's not always 4, for example on RAVEN there is only one.

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/4018>

radv/rgp: adjust trace memory/shader clocks to fix frame duration

To report microseconds instead of clocks.

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/4018>

mesa/st: fix formats required for EXT_texture_norm16

Earlier commit did not take in to account that lists required for
rendering and texturing are parsed separately. This commit simply
removes formats added to the other list.

Fixes: de4eb9a3bb9 ("mesa/st: toggle EXT_texture_norm16 based on format support")
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3961>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3961>

lima: Add etc1 support

Layer stride has to be divided by 4. We also have to take care of
the array_size when returning the bo_size.

Drop the affected tests from the fails list.

Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Signed-off-by: Andreas Baierl <ichgeh@imkreisrum.de>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3946>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3946>

doc: Update features.txt for r600 with misc supported features

Update features.txt with misc supported features for r600,
as reported by glxinfo for Cypress XT [Radeon HD 5870].

Reviewed-By: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4010>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4010>

intel/tools/dump_gpu: fix getparam values

Don't return the pci_id for all params

Fixes: 76bf38eaf0b6 ("intel/tools/aub_dump: move aub file initialization to maybe_init()")
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3994>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3994>

meson: Enable -Wno-deprecated only for bison > 2.3.

Older versions of bison do not support the -W option.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2571
Fixes: 11a1cb2fa8da ("meson: Disable bison's -Wdeprecated since we still support old bison.")
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3993>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3993>

mesa: fix _mesa_draw_nonzero_divisor_bits to return nonzero divisors

The bitmask is _EffEnabledNonZeroDivisor, so no need to invert it before
returning.

Fixes: fd6636ebc06d (st/mesa: simplify determination whether a draw needs min/max index)
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <maraeo@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4009>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4009>

gitlab-ci: disable panfrost runners

They seem to be timing out.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4011>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/4011>

radv: fix size of sqtt_file_chunk_asic_info on 32-bit system

The struct is actually 716 bytes, but on 64-bit systems the compiler
aligns it to 720. Add padding to make sure it's always 720.

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2580
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2578
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3996>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3996>

radv: fix 32-bit build failure in radv_queue_internal_submit()

Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2580
Closes: https://gitlab.freedesktop.org/mesa/mesa/issues/2578
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/3996>

glsl: add some error checks to the nir uniform linker

These are optional for spirv but it shouldnt hurt to enable them.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: fix sampler index calculation in nir linker

Here we reset the counter to 0 for each shader stage not each program.
We also make add a flag to stop iterating over indices that have
already been processed.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: reset next_image_index count for each shader stage

This fixes the image index calculation in the nir linker. We need
to reset the counter to 0 for each shader stage not each program.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: fix resizing of the uniform remap table

In the NIR linker we were not resizing the remap table correctly
for explicit locations when it was needed.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: set the correct number of images in a shader

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: set the correct number of samplers in a shader

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

glsl: fix possible memory leak in nir uniform linker

Use UniformDataSlots for the context of UniformDataDefaults rather
than UniformStorage as in some cause UniformStorage may be NULL.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3992>

intel/compiler: Restrict cs_threads to 64

Our current GPGPU_WALKER code only supports up to 64 threads.

On HSW we could use up to 70 and TGL up to 112, but only if the walker
is adjusted so the width does not exceed 64. Work to support this is
in progress.

Previous to this change, we might try to downgrade to SIMD8 if the
SIMD16 shader spilled. Since HSW and TGL have the max number of
threads above 64, we would then try to emit an invalid GPGPU walker
command.

Fixes: 932045061b5 ("i965/cs: Emit compute shader code and upload programs")
Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Tested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>

st/va: remove unneeded code

No need to explicitly set the 10-bit buffer format as the correct
buffer format will be allocated later

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3998>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3998>

freedreno/ir3: add assert

Catch problems earlier when inputs are not setup correctly.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: fix assert with getinfo

Fixes:
dEQP-VK.glsl.texture_functions.query.texturesamples.sampler2dms_fixed_vertex

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: don't precolor unassigned inputs

Fixes crash seen in:
dEQP-VK.glsl.conversions.matrix_to_matrix.mat4_to_mat3x4_vertex

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: fix crash with samgq workaround

Need to list_delinit() before we clone the instruction to split it into
individual samgpN instructions, otherwise we get list corruption.

Tested-by: Eduardo Lima Mitev <elima@igalia.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: update SFU delay

1) emperically, 10 seems like a more accurate # than 4
2) push "soft" delay handling into ir3_delayslots(), as
   we should also be using it to calculate the costs
   that the schedulers use

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: track half-precision live values

In schedule live value tracking, differentiate between half vs full
precision.  Half-precision live values are less costly than full
precision.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: don't hide latency when there is none to hide

Current scheduler thresholds try to ensure there are warps available to
switch to when hiding texture fetch latency.  But if there is none to
hide, we should allow scheduler to use more registers to reduce nops.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: rewrite regmask to better support a6xx+

To avoid spurious sync flags, we want to, for a6xx+, operate in terms of
half-regs, with a full precision register testing the corresponding two
half-regs that it conflicts with.

And while we are at it, stop open-coding BITSET

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: remove regmask_set_if_not()

No longer used.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno: honor FD_MESA_DEBUG=nogrow

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/a6xx: enable SKIP_IB2_ENABLE properly

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/a6xx: don't emit YIELD packet

We don't implement the rest of this.. and it would probably cause bad
things when kernel gains support for preemption.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/a6xx: whitespace fix

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/a6xx: emit LRZ clear in sysmem too

Fixes rendering issues in manhattan with FD_MESA_DEBUG=nogmem

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/a6xx: remove unused param

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

freedreno/ir3: remove from_tgsi

No longer used, other than in ir3 cmdline compiler, where it can be
replaced with a local variable.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3989>

turnip: increase array sizes in tu_descriptor_map

Pending the descriptor rework, this allows running the follow test:
dEQP-VK.renderpass.suballocation.attachment_sparse_filling.input_attachment_127

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: fall back to sysmem when attachments don't fit into gmem

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: remove unnecessary fb size check

Framebuffer with 0 width or height is not valid.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: don't hardcode gmem base for input attachment

Newer a6xx no longer has programmable GMEM base, so we can't rely on the
kernel driver setting it to 0x100000 (GMEM base is 0 on such GPUs).

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: fix srgb MRT

Register packing macros makes this only set the first bit. Set to whole
dword to fix srgb for color attachments >0.

Fixes: 59f29fc8 ("turnip: Convert the rest of tu_cmd_buffer.c over to the new pack macros.")
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: fix hw binning + render_area offset interaction

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: minify image_view extent

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: remove unecessary MRT_CONTROL fill

Hardware won't use MRT_CONTROL after mrt_count

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: move some constant state to tu6_init_hw

Also remove duplicates.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: check the right alignment requirement on shader iova

I had some trouble because I assumed this was right, tested that the
alignment requirement is actually 16.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3979>

turnip: add r5g5b5a1_unorm/b5g5r5a1_unorm  formats

r5g5b5a1/b5g5r5a1 tiled/ubwc is the same as a1r5g5b5 (in memory), but
linear is read as 1_5_5_5 and written with 5_5_5_1 with swap.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3806>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3806>

turnip: rework format table to support r5g5b5a1_unorm/b5g5r5a1_unorm

These formats are an exception that can't be modeled in the current format
table. Switch to a table with only a single a6xx_format per vk format,
and deal with the exceptions separately (currently the only exception is
10_10_10_2_UNORM which has a different color format).

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3806>

util/format: add missing BC4/BC5 vulkan formats

Enables these formats for turnip.

Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3806>

panfrost: LogicOp fixes and non 8-bit format support

With the previous LogicOp commit almost half of the blend modes were
broken because the surplus bits were not cleared after an inot.

v2:
 - Remove u8 "fast path" as 8-bit is not well optimised yet
 - Don't mask for 32-bit formats as that triggers an assert

Fixes: 068806c9f6b ("panfrost: LogicOp support")
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3943>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3943>

nir: Allow nir_format conversions to work on 32-bit values

The constant has to changed to unsigned long long, as shifting a
32-bit value by 32 is undefined behaviour.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3943>

r600: add missing <array> include

Fixes error with clang/libc++:

../src/gallium/drivers/r600/sfn/sfn_emitaluinstruction.h:69:88: error: implicit instantiation of undefined template 'std::__1::array<unsigned char, 3>'
   bool emit_alu_op3(const nir_alu_instr& instr, EAluOp opcode, std::array<uint8_t, 3> reorder={0,1,2});

Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3967>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3967>

llvmpipe: add support for tessellation shaders

This adds the hooks between llvmpipe and draw to enable tessellation shaders.

It also updates the CI results and docs.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallium/nir/tgsi: only scan fragment shader inputs for usage_mask

The scanner doesn't work with tess shaders, but we don't need it for those,
in fact only frag shaders need it.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

draw: hook up final bits of tessellation

This hooks tessellation into various parts of draw, so the
tessellation shaders are used in the correct places as the
last shader of the pipeline.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

draw: add main tessellation code

This is the bulk of the llvm shader builders and tessellation
execution code.

TCS uses a coroutine launcher like compute shaders to handle
barriers. It executes 4-wide with one input vertex per lane.

Tessellation happens before the TES is run.

TES is just a 4-wide launcher, one per primitive is executed,
with one lane per tessellation coordinate input.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

draw: add JIT context/functions for tess stages.

This adds the initial draw_tess.h with a define needed
for the interfaces. TCS input array doesn't need to handle
patch inputs so can be smaller.

The TCS context has some dummy values to align the textures/images
properly.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallivm/nir: add tessellation i/o support.

This add support for the tessellation i/o callbacks.

Tessellation requires another level of indirect indexing,
and allows fetches from shader outputs.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallivm/tgsi/swr: add mask vec to the tcs store

For the nir paths we want to access the mask vector to only
store when the mask allows it.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallivm/nir: align store_var param order with load_var

This was ugly so align load/store to have mostly the same
parameter ordering

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallivm/nir: add support for tess system values

hooks up the tessellation specific system values in the NIR paths

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallivm/nir: split out 64-bit splitting code

This just lets it be reused for tess later.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

gallium/auxiliary: add the microsoft tessellator and a pipe wrapper.

This adds the same tessellator code that swr uses, swr should
move to using this copy, unfortunately that wasn't trivial on my first
look.

The p_tessellator wrapper wraps it in a form that is a useful interface
for draw.

Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3841>

radv: allow to capture SQTT traces with RADV_THREAD_TRACE=<start_frame>

This is pretty basic (and a bit crappy at the moment). I think we
might want some sort of overlay in the future and also be able to
trigger captures with F12 or whatever.

To record a capture, set RADV_THREAD_TRACE to something greater than
zero (eg. RADV_THREAD_TRACE=100 will capture frame #100). If the
driver didn't crash (or the GPU didn't hang), the capture file
should be stored in /tmp.

To open that capture, use Radeon GPU Profiler and enjoy your
profiling times with RADV! \o/

Note that thread trace support is quite experimental, only GFX9 is
supported at the moment, and a bunch of useful stuff are still missing
(shader ISA, pipelines info, etc). More is comming soon.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3900>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3900>

radv: add initial SQTT files generation support

SQTT is also a file format (.rgp extension) that can be consumed
by Radeon GPU Profiler for profiling purposes.

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/3900>

radv: emit thread trace markers after every draw/dispatch call

Thread trace markers (also called events in Radeon GPU Profiler)
should be emitted after every draw/dispatch calls to collect data.

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/3900>

radv: add initial SQ Thread Trace support for GFX9

SQTT is a hardware block that collects thread trace data (like
wave occupancy, timings, etc) for every draw/dispatch calls.

It's only supported on GFX9 at the moment but I will add other
generations support soon.

This is the first step towards profiling with RADV!

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/3900>

radv: add a small helper that allows to submit internal CS

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/3900>

ac/registers: add definitions for thread trace

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/3900>

ac: add more fields to ac_gpu_info

For RGP traces.

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/3900>

ci: Enable -Werror on meson-vulkan and meson-testing.

I want to make sure that I don't introduce warnings in turnip where we
have active work going on, and I also want to make sure that the drivers
we care about testing are warnings-clean.

As with the previous -Werror change, this is for CI only and doesn't
affect end-user builds.

Reviewed-by: Michel Dänzer <mdaenzer@redhat.com>
Reviewed-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3607>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3607>

aco: Fix signed-vs-unsigned warning.

The previous instance of this comparision was 1u to avoid the warning, fix
this one too.

Fixes: dba71de5c636 ("aco: only create parallelcopy to restore exec at loop exit if needed")
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3607>

gallium/u_vbuf: silence a warning by using unreachable

Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>

mesa: fix 11 warnings

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>

nir: fix 5 warnings

Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>

gallivm: fix 5 warnings

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>

nir: replace GCC unroll with an option that works on GCC < 8.0

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3970>

mesa: fix incorrect prim.begin/end for glMultiDrawElements

This has no effect on Gallium, but it affects tnl.

Cc: 19.3 20.0 <mesa-stable@lists.freedesktop.org>
Reviewed-by: Mathias Fröhlich <mathias.froehlich@web.de>
Tested-by: Marge Bot <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa: optimize glMultiDrawArrays, call Draw only once (v2)

v2: use the macros

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa: don't unroll glMultiDrawElements if one count is 0

let the driver skip or submit an empty draw call.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa: clean up glMultiDrawElements code, use alloca for small draw count (v2)

v2: use calloc, add reusable macros

Reviewed-by: Mathias Fröhlich <mathias.froehlich@web.de> (v1)
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com> (v1)
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa: move num_instances and base_instance out of _mesa_prim

They are never used by multi draws and internal draws.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa: remove redundant _mesa_prim::is_indexed

Instead, check (ib != NULL) like all other drivers.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

mesa/i965: remove _mesa_prim::indirect_offset

Only i965 was using it.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>

gallium/u_threaded: convert dividing by index_size to a bit shift

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3990>