pan/bi: Add some zero bytes after shaders on Bifrost

Bifrost will prefetch bytes after the end of shaders, so make sure
these bytes are allocated and zeroed.

Fixes GPU faults in Xonotic.

Suggested-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8428>

pan/bi: Add a define for the Bifrost shader prefetch size

Found by adding NOPs to the start of a shader and checking dmesg to
see at what sizes the GPU faulted trying to read the following
non-executable page.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8428>

radeonsi: Only set modifier creation function for GFX9+ & with kernel support.

Fixes: c786150dfa5 ("radeonsi: Add modifier support.")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3963
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8407>

panfrost: Skip an XFB test that's passing/failing randomly

transform_feedback.array_element.interleaved.triangles.mediump_mat2x4
seems to pass/fail randomly, skip it for now.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8449>

panfrost: Re-enable AFBC on 3D, 2D arrays

Things have now been fixed and AFBC on 3D/2D-arrays seems to work fine.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8449>

panfrost: Fix estimate_texture_payload_size() on Bifrost

Bifrost mandates manual stride usage.

Fixes: a3d2936a8e9e ("panfrost: The texture descriptor has a pointer to a trampoline")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8449>

panfrost: Pass the resource dimension to panfrost_compression_tag()

The reload surface logic creates 2D image views pointing to a specific
3D texture layer, but panfrost_compression_tag() cares about the resource
dimension, not the image view one.

Fixes: 4dd7991422ce ("panfrost: Add a pan_image_layout object")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8449>

panfrost: Get layer stride of level 0 on staging resources

Staging resources have one level, we shouldn't query the stride of
level > 0.

Fixes: 3c92abe35937 ("panfrost: Use panfrost_get_layer_stride() instead of open-coding it")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8449>

vulkan/overlay: fix vkGetInstanceProcAddr self-resolving

vkGetInstanceProcAddr(instance, "vkGetInstanceProcAddr") should return our
vkGetInstanceProcAddr not the next in the chain.

CC: mesa-stable
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8285>

docs/features: Fix missing close paranthesis

My bad. This is what I get for writing documentation at midnight.

Fixes: bd697652a73 ("docs: Document extensions exposing GL3.0")
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8429>

docs/features: Mark GL3.1 as done on Panfrost

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8429>

docs/panfrost: Update GL/ES versions for v5+

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8429>

docs: Add release note for Bifrost GL3.1

GL3.1 was already new for Midgard so we adjust the wording slightly.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8429>

Revert "ci: Disable the freedreno farm, which went down last night."

Networking broke on a reboot or something, fixed the broken setting and
it's back up.

This reverts commit ff67898dafaea82db669aef10ebb05e65eaa0921.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8454>

panfrost: Dual-source blending on Bifrost

Tested with the arb_blend_func_extended Piglit tests and Alacritty.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8455>

radv: Add Android module info to linker script.

The Android Vulkan loader needs this symbol, so the addition of the
linker script broke Vulkan for Android.

(For non-Android builds: I checked that having a non-existent symbol in
 the linker script works ok and doesn't put the symbol in the library)

Fixes: 41bb6459d3a ("radv: restrict exported symbols with static llvm")
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8437>

gallium/ntt: Take ownership of the NIR shader we're passed.

It makes no sense for the caller to keep it, since we've throughly changed
it to be suitable to NTT.  All callers just freed it afterward anyway.

Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8422>

mesa/st: Free the ARB_vp/fp nir-to-tgsi temporary tokens.

The driver interface doesn't take ownership of the TGSI tokens, so free
our temporary.

Fixes: cf3fc79cd0ab ("st/mesa: Replace mesa_to_tgsi() with prog_to_nir() and nir_to_tgsi().")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8422>

mesa/st: Free the NIR builtins TGSI tokens after passing to the driver.

The driver interface doesn't take ownership of the TGSI tokens, so free
our temporary.

Fixes: 57effa342b75 ("st/mesa: Drop the TGSI paths for PBOs and use nir-to-tgsi if needed.")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8422>

gallium/ntt: Fix leak of the per-instr liveness information.

Fixes: 34cc6a804ec9 ("gallium: Add a nir-to-TGSI pass.")
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8422>

r600: enable fp64 lowering to softemu with NIR

The NIR code path does not (yet) take hardware into account
that actually supports fp64 in hardware.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

r600: enable support for 64 bit DIVMOD when NIR is used

The glsl lowering seems to be broken, but the nir lowering pass handles
this correctly.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

r600/sfn: tie in 64 lowering code

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

r600/sfn: add lowering passes to get 64 bit ops lowered to 32 bit vec2

The lower_double and lower_int64 don't lower all 64 bit IO ops and merging
to and splitting fromn 64 bit values. So here goes a bunch of lowering
passes that takes care of this and also of merging IO that might have been
split.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

r600/sfn: Add number for source components for split_y

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

r600/sfn: C++ lower-instruct implementation

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7824>

panfrost: Enable ARB_texture_buffer_object

Tested with the arb_texture_buffer_object and arb_texture_buffer_range
piglit tests.

Exposes OpenGL 3.1 by default on T760+

    $ PAN_MESA_DEBUG= glxinfo | grep "GL version"
    OpenGL version string: 3.1 Mesa 21.0.0-devel

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

panfrost: Fix textureSize for buffer textures

Suggested-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

panfrost: Support buffer sampler views

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

panfrost: Make the width argument to panfrost_new_texture 32 bits

This is needed to represent a width of exactly 65536, which is
required by ARB_texture_buffer_object.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

st/mesa: Use samplers for buffer textures if requested

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

docs: Mention PIPE_CAP_TEXTURE_BUFFER_SAMPLER

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

gallium: Add new cap PIPE_CAP_TEXTURE_BUFFER_SAMPLER

This indicates whether a driver wants samplers for buffer textures as
well as normal textures.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8300>

aco: Wait for stores when NGG or legacy VS can finish early.

When there are no param exports in an NGG (or legacy VS) shader,
the NO_PC_EXPORT=1 is set, which means PS waves can launch before
the current stage finishes.

If the current stage has any stores, we need to make sure to wait for
those before we allow PS waves to start, so that PS can read what
these instructions stored.

Fossil DB results on Navi 10:
Totals from 45 (0.03% of 136420) affected shaders:
CodeSize: 87224 -> 87404 (+0.21%)
Instrs: 16750 -> 16795 (+0.27%)
Cycles: 69580 -> 69760 (+0.26%)
VMEM: 8022 -> 8167 (+1.81%)

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7868>

aco: Note if rasterization can start early.

When there are no param exports in an NGG (or legacy VS) shader,
the NO_PC_EXPORT=1 is set by RADV, which means PS waves can launch
before the current stage finishes.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7868>

aco/ra: fix infinite recursion in get_reg_simple() with subdword registers

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Fixes: f8c7661ecaa782fdde105a4bf756023eb88ea780 ('aco: try to better align 8+ dword SGPR vectors')
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

aco: simplify and fix operand/definition sizes

These are mainly needed for constant propagation
and subdword register allocation.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

radv: enable .lower_ineg

We already emit ineg as isub most of the time.

The results are a bit mixed, but shouldn't really make a difference.
A couple of additional copies are needed as isub writes scc.

Totals from 5975 (4.29% of 139391) affected shaders:
CodeSize: 31508648 -> 31509264 (+0.00%); split: -0.00%, +0.00%
Instrs: 6073379 -> 6073531 (+0.00%); split: -0.00%, +0.00%
Cycles: 47186280 -> 47187116 (+0.00%); split: -0.00%, +0.00%
VMEM: 2528515 -> 2529139 (+0.02%); split: +0.03%, -0.01%
SMEM: 596842 -> 596924 (+0.01%); split: +0.02%, -0.00%
SClause: 280596 -> 280594 (-0.00%)
Copies: 288554 -> 288669 (+0.04%); split: -0.00%, +0.04%
PreSGPRs: 240390 -> 240397 (+0.00%)
PreVGPRs: 349630 -> 349749 (+0.03%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

radv: don't lower_pack() after load-store-vectorization

Totals from 7 (0.01% of 139391) affected shaders:
CodeSize: 282900 -> 283324 (+0.15%); split: -0.01%, +0.16%
Instrs: 45287 -> 45338 (+0.11%); split: -0.01%, +0.12%
Cycles: 11496332 -> 11510396 (+0.12%); split: -0.00%, +0.12%
VMEM: 2355 -> 2335 (-0.85%)
Copies: 15506 -> 15561 (+0.35%)

A bit of noise in some parallel-rdp shaders.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

radv: call nir_opt_algebraic_late() after lowering idiv for small bitsizes

This is needed because lower_idiv() introduces ineg again
which we'll remove next.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

radv: optimize idiv_const for small bitsizes

No fossil-db changes.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8425>

radv: fix color resolves if the dest image has DCC

Using the graphics resolve path when DCC is enabled should only be
a hint to avoid DCC fixup.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3388
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/8326>

radv: fixup DCC after color resolves using the compute path

If the dest image has DCC it should be re-initialized to the
uncompressed state.

Note that the driver always selects the graphics path if the dest
image has DCC, so this has no effect for 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/8326>

radv: decompress DCC for partial resolves using the compute path

Because DCC is re-initialized to the uncompressed state after the
resolve, so if the app does a partial resolve it should be
decompressed first.

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/8326>

radv: set depth to 1 for subpass resolves using the compute path

To match Vulkan convention.

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/8326>

aco: improve nir_op_vec with constant operands

Could still be improved a little. For example, 8-bit pack without
constants could be:
(s_pack_ll(x, z) & 0x00ff00ff) | ((s_pack_ll(y, w) & 0x00ff00ff) << 8)

fossil-db (Sienna):
Totals from 136 (0.10% of 139391) affected shaders:
CodeSize: 279776 -> 278144 (-0.58%)
Instrs: 50742 -> 50470 (-0.54%)
Cycles: 211560 -> 210472 (-0.51%)
SMEM: 3607 -> 3557 (-1.39%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8421>

wgl: Don't crash in stw_make_current if current framebuffer is NULL

Fixes #3878

Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7999>

radv: set invariantgeom for Shadow of the Tomb Raider

Work around flickering foliage on GFX10.3

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4064
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8104>

radv: add RADV_DEBUG=invariantgeom

This can be used to work around a common class of bugs appearing as
flickering.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8104>

nir/divergence_analysis: mark load_push_constant as uniform

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8439>

radv: mark some sparse texture CTS as expected failures on GFX9

It's likely a RADV bug but marking them as expected failures
to avoid more breakage in the meantime.

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/8442>

ci: mark some sparse tests as expected failures on Pitcairn (GFX6)

Two CTS bugs reported.

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/8436>

aco: fix inserting expcnt for MIMG on GFX6

MIMG VDATA has moved to its own operand.

Fixes: 962c917cea9 ("aco: move MIMG VDATA to its own operand")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8435>

vbo/dlist: free prim_store->prims when vbo_save is destroyed

Fixes: 83149e1c2d8 ("vbo/dlist: refactor prim_store/vertex_store allocations")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4067
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8410>

glcpp: disable 'windows' tests

these timeout a lot

Acked-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Acked-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8321>

frontends/omx: fix build warning

From code path, it's easy to see that the variable is only used
with condition, and the variable gets initialized with the same
condition before getting used, but we can initialize it when defined
to avoid the warning.

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8404>

nouveau: add drm-shim support

Seems to work reasonably well with shader-db's run.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8406>

nouveau: change fence destruction logic on screen destroy

With the drm shim, we can't actually wait for the hardware to do
anything. But why wait for it at all? We just need to make sure to
execute all the work and clean up any resources. Add a helper to do
that.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8406>

panfrost: Advertise ES3.0 on Bifrost

Doing so fixes our remaining dEQP-GLES2 failures.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

panfrost: Test GLES3 on Bifrost

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Tomeu Vizoso <tomeu.vizoso@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

panfrost: Disable AFBC of 3D, 2D arrays

These are broken at least on v7 and likely elsewhere. Until this can be
investigated, let's disable it so we don't break dEQP-GLES3.

Example of a failing test without this patch:

dEQP-GLES3.functional.texture.filtering.2d_array.sizes.128x32x64_nearest

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

panfrost: Promote 8b to 16b for blend descriptors

There is no 8-bit register_format applicable. Some lowering is needed
at the compiler level to make it work.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Lower 8bit fragment outputs to 16bit

Bifrost does not support passing fragment colors through 8 bit registers.

(Rewrote to use NIR helpers -Alyssa)

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Assert immediate indices fit

This would have caught the assortment of bugs fixed in the previous two
commits, and should help lint going forward.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Parametrize intrinsic immediate limits

Fixes issues on:

dEQP-GLES3.functional.transform_feedback.array_element.separate.points.lowp_mat2x3

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Use TEXC for indices >= 8

Otherwise it can't fit, fixes

   dEQP-GLES3.functional.texture.units.all_units.only_2d.*
   dEQP-GLES3.functional.texture.units.all_units.only_cube.*
   dEQP-GLES3.functional.texture.units.all_units.mixed.*

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Implement scalar i2i8/u2u8

Doesn't work for vectors though that's the vectorization branch's
problem to deal with now. Suffices for fragment output.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Allow passing thorugh 8-bit scalars

Needed for fragment output. Not nearly sufficient for actual 8-bit
types in e.g OpenCL.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Don't suppress Inf/NaN

glmark2 was fixed upstream and we don't have Bifrost trace testing
anyway. Fixes isinf/isnan tests.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Fix ATEST with pure integers

It doesn't matter what we pass due to a subtlety in the spec but the
assert is still wrong.

Fixes: 49f38aa9e79 ("pan/bi: Implement fragment_out by builder")
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

pan/bi: Pull out bi_dontcare helper

Where we need a power efficient encoding but don't care about the value
read, corresponding to rNULL in canonical assembly.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Tested-by: Maciej Matuszczyk <maccraft123mc@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8427>

intel/common: Build mi_builder_test for gen 12.5

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8424>

r600: Support TGSI_OPCODE_I64NEG

This is needed for removing the use of TGSI source modifiers on integers

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8401>

pan/bi: Allow non-terminal BLEND operations

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

pan/bi: Use the interference mechanism to describe blend shader reg use

Blend shader might clobber r0-r15, we must make sure the RA is aware
when compiling fragment shaders that have BLEND operations.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

pan/bi: Restrict registers to r0-r15 when compiling blend shaders

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

pan/bi: Port bi_collect_blend_ret_addr() to the new compiler infra

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

pan/bi: Pass bundle pointers to bi_pack_tuple()

So we have updated regs and can use that to collect blend return
addresses.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

pan/bi: Workaround BLEND precolour with explicit moves

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

panfrost: Set the RT index when emitting a Bifrost blend descriptor

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8417>

vulkan/overay: fix violation of VUID-VkDeviceCreateInfo-pNext-00373

We can't use VkDeviceCreateInfo::pEnabledFeatures when
VkPhysicalDeviceFeatures2 is present in pNext chain.

Fixes: ea7a6fa98055ad ("vulkan/overlay: add pipeline statistic & timestamps support")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8208>

nv50: use 2d blit when m2mf doesn't support the copy

Looks like m2mf bails if a line is >64k in width for tiled textures
(even if only a sub-section is copied as long as any part is beyond the
64k mark).

Fixes a number of GLES3 accuracy tests which made 8k-wide textures which
were read out as RGBA32_UINT, leading to problems.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8396>

nv50,nvc0: serialize between before/after using a zeta surface as color

This happens during "3d" blit operations, where we must reinterpret it
as color in order to support stencil/depth masking. However the hardware
isn't necessarily amused by this, esp when multiple draws are queued up.
Throw in serialize calls in order to get it to flush out previous draws.

This was noticeable in the test
dEQP-GLES3.functional.fbo.invalidate.sub.unbind_blit_msaa_stencil,
although 3d blit operation had to be forced on nvc0 where it's much
rarer.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8396>

nv50: fix instancing of client-side vertex buffers

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8396>

nv50/ir: allow a mov to emit directly to a shader output

Under some very rare circumstances, the OP_EXPORT will refer to a def
provided by a mov. When we then try to make the defining op write to the
export directly, it blows up. Reuse the existing setDst helper which
handles this and more for the long encoding.

Fixes dEQP-GLES3.functional.shaders.precision.int.highp_mul_vertex

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8396>

drisw: fix unused variables warnings

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8337>

llvmpipe: fix unused variables warnings

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8337>

nir: replace .lower_sub with .has_fsub and .has_isub

This allows a more fine-grained control about whether
a backend supports one of these instructions.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6597>

nir,vc4: Lower fneg to fmul(x, -1.0)

This patch also replaces lower_negate with lower_ineg / lower_fneg.

The fneg semantics have been clarified as of Version 1.5, Revision 1
of the SPIR-V specification, which means that the previous lowering
to fsub is not a viable solution anymore, and is replaced with
lowering to fmul(x, -1.0).

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6597>

pan/bi: Fix constant slot selection

The fau_index field contains the lower 4 bits of the 64bit constant,
which allows one to reuse the same clause constant slot from different
bundles if the upper 60 bits match. That doesn't work for constants
referenced from the same instruction or for constants referenced from
two instructions that are part of the same bundle though, since the
fau_index is shared in that case.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8418>

panfrost: Fix UBO count calculation on Bifrost

The Bifrost compiler relies on nir_lower_uniforms_to_ubo() to lower
uniforms to UBOs, and s->info.num_ubos is already incremented when a
UBO #0 is created.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8418>

vulkan/device-select: fix vkGetInstanceProcAddr self-resolving

vkGetInstanceProcAddr(instance, "vkGetInstanceProcAddr") should return our
vkGetInstanceProcAddr not the next in the chain.

CC: mesa-stable
Signed-off-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8286>

radv: mark VK_IMAGE_CREATE_SPARSE_RESIDENCY_BIT as unsupported on GFX6-7

This is only supported on GFX8+, this fixes a ton of CTS failures
on my Pitcairn (GFX6).

Fixes: af7fb4df50f ("radv: Add sparse image queries.")
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/8415>

aco: remove can_reorder semantic in get_sync_info_with_hack

No fossil-db changes.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8416>

aco: try to better align 8+ dword SGPR vectors

This doesn't have much of an effect, but it helps avoid a
pathological case for Assassin's Creed Valhalla and a RDR2 shader with a
future change.

fossil-db (Sienna):
Totals from 55074 (39.51% of 139391) affected shaders:
SGPRs: 3515076 -> 3567744 (+1.50%); split: -0.01%, +1.51%
CodeSize: 206942120 -> 206941868 (-0.00%); split: -0.00%, +0.00%
Instrs: 39625900 -> 39625837 (-0.00%); split: -0.00%, +0.00%
Cycles: 1640088780 -> 1640088828 (+0.00%); split: -0.00%, +0.00%

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4070
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8416>

mesa/st: fix regression for basic drivers

Drivers that doesn't support conditional rendering can't really require
respecting conditional rendering, so let's not ask for it to be
respected in the first place.

This fixes a problem where util_can_blit_via_copy_region started
unconditionally rejecting all blits that originate from
glBlitFramebuffer, even for drivers where this can't possibly be a
problem.

Fixes: 767f70dfe13 ("gallium/util: fix util_can_blit_via_copy_region for conditional rendering")
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8378>

mesa/teximage: show internal format when printing verbose api log

This would have saved me some time to determine that the problem was
coming from lack of RGBA32F / 8x MSAA support on nv50.

Signed-off-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/8405>

radv: enable DCC for layered color images on GFX10+

There is still some CTS failures on GFX9.

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/8368>

radv: do not use predication when the range doesn't cover the whole image

The predication is based on the mip level, so if the image has layers
and DCC is enabled, it should only be used if the range of layers
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/8368>

radv: clean up radv_decompress_dcc_compute()

Remove one old comment because it supports decompressing layers.

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/8368>