docs/features: Mark more TBO exts done on panfrost

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Advertise OES_standard_derivatives

Has been implemented for ages since it's part of ES3, but we missed the
CAP needed for the ES2 extension.

Closes: #4203
Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Bump advertised ESSL feature level

Needed to run more dEQP tests.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Bump max SSBO count

There's no actual architectural limit, this is totally arbitrary, but
fixes some NotSupported errors in dEQP-GLES31.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Advertise SAMPLE_SHADING

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Assert on indirect compute shaders

This case was overlooked. Much simpler than indirect graphics calls, can
come as a follow on to the other indirect work.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Remove stale TODOs

The answer is no, we do not. Putting them in the primary batch allows
proper intermixing with graphics work. There's no harm for pure compute
loads, either.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Simplify bind_compute_state

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/{mdg, bi}: Lower load_sample_pos

Conceptually same lowering as the DDK, although we're missing a number
of relevant compiler optimizations so the generated code is awful.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/{mdg, bi}: Lower load_helper_invocation

Passes dEQP-GLES31.functional.shaders.helper_invocation.* on both
architectures

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement coverage mask updates

Impressive how simple this looks after dozens of patches of
sysval-related prepwork :)

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Decouple sysval loading from NIR

Useful for builder-based sysvals.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement nir_intrinsic_load_sample_positions_pan

Just copy the FAU result.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement load_sample_mask_in

Preloaded.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Fix gl_SampleID read

Fixes dEQP-GLES31.functional.shaders.sample_variables.sample_id.default_framebuffer

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Lower ifind_msb

We have ufind_msb efficient, implement in terms of that.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement ufind_msb

Lowered to #(sz - 1) - clz(x), taking advantage of the machine's
8-bit and 16-bit variants of clz and the widening on the second argument
of ISUB to implement neatly in two instructions.

Note that in NIR, ufind_msb can take any integer type but always output
i32.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement bitfield_reverse

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Support bit_count()

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Add uclz() support

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Lower bitfield inserts/extracts

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Implement texture gathers

Passes all of dEQP-GLES31.functional.texture.gather.*

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/bi: Remove redundant TEXC opcode check

Already checked in emit_tex

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Lower stores from helpers

Required for correct behaviour of SSBOs in fragment shaders on Midgard.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Stub load_barycentric_sample

Now parity with Bifrost for not failing over these intrinsics.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Lower ufind_msb, poorly

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Implement uclz

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Rename bitcount8 to popcnt, fixing the unit

Still doesn't seem to work correctly for negative values.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Lower bitfield instructions

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Remove unused pack_unorm_4x8 lowering

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/mdg: Assert on bad 64-bit swizzle in disassembly

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Add MULTISAMPLED sysval

To be used with Bifrost coverage mask updates.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Overhaul sysval handling

Don't preassign.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Implement get_sample_position

Easy now that we have the table.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Advertise MSAA 8x and 16x

On Bifrost. Could support Mali T760+ with additional debug.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Ensure open_device has pandecode initialized

Requires reordering some calls.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Use sample location LUT

Fixes multisampling issues on Bifrost.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Upload sample positions on device init

Needed to correctly implement multisampling on Bifrost, and used in
gl_SamplePosition  implementation on Midgard. These are architecturally
invariant, so it's fine to hard code.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Set sample count/pattern for tiler FBD

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Remove batch_is_scanout

Unused since last commit.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Remove PAN_REQ_DEPTH_WRITE

No consumers, garbage collect the whole thing.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Remove PAN_REQ_MSAA

Neglible win for draw time overhead.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Don't use PAN_REQ_MSAA in SFBD

We can also generalize a bit to eliminate the state dependence.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Don't set REQ_MSAA in pan_mfbd

No remaining consumer.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Generalize MSAA handling

This uses the framebuffer MSAA state directly and will generalize to
MSAA 8x/16x when somebody gets around to it.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Set tiler descriptor sampler pattern

Bifrost requires this to match the framebuffer descriptor's sample
pattern, but we were leaving the default (single-sampled)
unconditionally, leading to undefined behaviour.

It is unknown if this fixes any user-visible bugs, but without the
commit the descriptor is likely out-of-spec.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: 21.0 <mesa-stable@lists.freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Add panfrost_sample_pattern helper

We always use rotated grids to preserve current behaviour.

Cc stable as it is required for the next commit. If deemed too invasive,
a simpler fix could be backported.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Cc: 21.0 <mesa-stable@lists.freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Respect info.fs.uses_sample_shading

Now that this is a link-time property.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Refactor sample shading state

Moves the decision to evaluate per-sample to link time instead of
draw-time, with an explicit property. No functional change.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Push sample positions sysval for Midgard

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Add sample positions sysval

For Midgard. On Bifrost, the hardware pushes this directly to FAU.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

panfrost: Preload sample mask if needed

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/decode: Only print local storage for vertex jobs

It's convenient to group this with the framebuffer, but the other fields
are unused by the hardware for vertex jobs. They _are_ used for tiler
jobs.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

pan/decode: Cleanup sample locations decode

We know what this is now. I opted to leave it in ~fixed-point format to
avoid bikeshedding over precision.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

nir: Add sample_positions_pan intrinsic

Facilites the gl_SamplePosition lowering on Bifrost, where the sample
positions are accessed directly in a packed in-memory format.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8774>

iris: Make a pin_scratch_space() helper

We need to (re-)pin the scratch buffer in four different places, and
it's going to get slightly more complicated on future platforms.  So,
make a helper function, allowing us to add the complexity in one spot.

Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9023>

zink: enable KHR_shader_draw_parameters on Vulkan <1.2

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: do not enable extensions that are now core

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: fix type annotations

mypy complains about this

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: validate has_properties and has_features

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: perform basic validation in zink_device_info

Check for existence of extension and its type

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: make zink_device_info accept vk.xml

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: introduce notion of non-standard extensions

this is for the MoltenVK extensions, especially "VK_MVK_moltenvk", which
right now is reserved in the registry. Making it non-standard skips all
the validations.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: more validation in zink_instance

the MVK check is a workaround, since VK_MVK_moltenvk is not an official
VK extension per se - the next patch will introduce nonstandardness to
Extension.

Two new validations are added by this patch:
1. extension type (non-instance extensions are rejected)
2. existence of specified instance functions

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

zink/codegen: introduce ExtensionRegistry

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9021>

radv/winsys: set use_global_list inside the critical section

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/9006>

radv: only make the WSI images resident if the global BO list is used

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4270
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4286
Fixes: 96b03aaa175 ("radv: use the global BO list from the winsys")
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/9006>

aco: use VCC as regular SGPR pair on GFX10

There is no need to reserve it for special purposes, only.

Totals from 139391 (100.00% of 139391) affected shaders (Navi10):
VGPRs: 4738296 -> 4738156 (-0.00%); split: -0.01%, +0.00%
SpillSGPRs: 16188 -> 14968 (-7.54%); split: -7.60%, +0.06%
CodeSize: 294204472 -> 294118048 (-0.03%); split: -0.04%, +0.01%
MaxWaves: 2119584 -> 2119619 (+0.00%); split: +0.00%, -0.00%
Instrs: 56075079 -> 56056235 (-0.03%); split: -0.05%, +0.01%
Cycles: 1757781564 -> 1755354032 (-0.14%); split: -0.16%, +0.02%
VMEM: 52995887 -> 52996319 (+0.00%); split: +0.07%, -0.07%
SMEM: 9005338 -> 9004858 (-0.01%); split: +0.16%, -0.17%
VClause: 1178436 -> 1178331 (-0.01%); split: -0.02%, +0.01%
SClause: 2403649 -> 2404542 (+0.04%); split: -0.14%, +0.18%
Copies: 3447073 -> 3432417 (-0.43%); split: -0.66%, +0.23%
Branches: 1166542 -> 1166422 (-0.01%); split: -0.11%, +0.10%
PreSGPRs: 4229322 -> 4235538 (+0.15%)
PreVGPRs: 3817111 -> 3817040 (-0.00%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: don't abort() if disassembly fails

We used that to catch assembly errors in the past,
but now, there are too many hardware features we
use in ACO that are not supported by the LLVM disassembler,
that it is not really suited anymore as a debugging tool.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: check get_reg_specified() on register hints

This ensures that max_used_sgpr is adjusted accordingly.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: also consider VCC in get_reg_specified()

This allows split_vector and others to keep their VCC position.

Totals from 4573 (3.28% of 139391) affected shaders (Navi10):
CodeSize: 54292268 -> 54289324 (-0.01%); split: -0.03%, +0.03%
Instrs: 10327645 -> 10326941 (-0.01%); split: -0.04%, +0.04%
Cycles: 744410748 -> 744034732 (-0.05%); split: -0.07%, +0.02%
VMEM: 749093 -> 749092 (-0.00%); split: +0.00%, -0.00%
SMEM: 269306 -> 269322 (+0.01%)
SClause: 358746 -> 358744 (-0.00%)
Copies: 826051 -> 823910 (-0.26%); split: -0.55%, +0.29%
Branches: 355074 -> 356493 (+0.40%); split: -0.01%, +0.41%

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: don't decrease the vgpr_limit when encountering bpermute

Instead we recalculate vgpr_limit on demand, depending on
the number of needed shared VGPRs.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: refactor GPR limit calculation

This patch delays the calculation of GPR limits in order to
precisely incorporate extra registers (VCC etc.) and shared VGPRs.

Additionally, the allocation granularity is used to set the config.
This has some effect on the reported SGPR stats.

Totals (Navi10):
SGPRs: 6971787 -> 17753642 (+154.65%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: change gpr_alloc_granule to full alignment

This also switches the alloc_granule of Tonga and Iceland
to 96, so that the calculation is consistent.
Also changes the granularity for RDNA to 16 to keep
better stats with the upcoming patch.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

aco: fix shared VGPR allocation on RDNA2

VGPRs are now allocated in blocks of 8 normal
or 16 shared VGPRs, respectively.

Fixes: 14a5021aff661a26d76f330fec55d400d35443a8 ('aco/gfx10: Refactor of GFX10 wave64 bpermute.')

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8921>

zink: VK_KHR_draw_indirect_count is a device extension

this fixes some testcases on CI.

Fixes: 1c01ad1b804 ("zink: add KHR_draw_indirect_count detection")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8991>

radv: emit pipeline bind markers for SQTT

I suspect this marker to be useful for correlating pipeline shaders.

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/8995>

zink: fix streamout for tess stage

the tess shader needs to actually emit xfb stuff in order for it to work

Fixes: 2891e0b74e6 ("zink: pull xfb info from tess shader when applicable")

Reviewed-by: Hoe Hao Cheng <haochengho12907@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9013>

wgl: Disable automatic use of layered drivers with LIBGL_ALWAYS_SOFTWARE

Fixes: 8955980f ("gallium/targets/libgl-gdi: prefer d3d12 driver")
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Prodea Alexandru-Liviu <liviuprodea@yahoo.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8865>

d3d12: Fail screen creation if a shader validator is needed and can't be created

Also fail screen creation if experimental shader models are requested, but can't be enabled

Fixes: 2ea15cd6 ("d3d12: introduce d3d12 gallium driver")
Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Tested-by: Prodea Alexandru-Liviu <liviuprodea@yahoo.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4022
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8865>

wgl: Add a loop for screen creation with an ordered list of fallbacks

Fixes: 8955980f ("gallium/targets/libgl-gdi: prefer d3d12 driver")
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Prodea Alexandru-Liviu <liviuprodea@yahoo.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4022
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8865>

wgl: Refactor screen creation to a function

Fixes: 8955980f ("gallium/targets/libgl-gdi: prefer d3d12 driver")
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Prodea Alexandru-Liviu <liviuprodea@yahoo.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4022
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8865>

pan/bi: Fix empty shader handling

Fixes INSTR_INVALID_ENC fault on dEQP-GLES31.functional.compute.basic.empty

Fixes: bfcdc8f1747 ("pan/bi: Add some zero bytes after shaders on Bifrost")
Signed-off-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/9011>

pan/bi: Fix jumps to terminal block again

New scheduler broke this. We need to shuffle some code around so we do
the lower pre-schedule instead of post-schedule (no clauses to work
with).

Fixes: 77933d16d8c ("pan/bi: Switch to new scheduler")
Reported-by: Icecream95 <ixn@disroot.org>
Signed-off-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/9011>

panfrost: Fake shader images for bifrost+deqp

Signed-off-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/9011>

ci: Disable scons-win64 job

It's failed for almost a month, so right now it's mostly noise and a
waste of CI resources.

It can easily be re-enabled by an MR which makes it pass again.

Reviewed-by: Eric Anholt <eric@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8976>

radv: Ignore WC flags for VRAM.

Otherwise there might be buffers for which we don't have a type.

Fixes: 7262c743dc8 ("radv: Determine memory type for import based on fd.")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/4280
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8996>

zink: support SO_OVERFLOW pipe query types

this is really just not what we want to be doing. vulkan has no method for
doing on-gpu checks for xfb overflow, instead providing the info as two values
for the user to do with as they will, forcing us to gpu stall any time we
need to interact with these queries

for the ANY variant of the query, we need to create even more xfb query pools,
since we now need to be monitoring all available vertex streams for overflows

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8992>

zink: put SO_OVERFLOW queries on the primgen list

these need to know if xfb was active during the query

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8992>

zink: break out cpu query reading for qbos into separate function

we're going to need this more than once

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8992>

zink: make the xfb_query_pool into an array

we'll need to potentially be observing all streams to handle the
query types from ARB_transform_feedback_overflow_query

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8992>

zink: always use query->type for starting/stopping xfb queries

we're going to be seeing some overlap here

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8992>

pan/bi: Skip ATEST for colour blit shaders

Small win.

Signed-off-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/9002>

panfrost: Pass is_blit flag around

There are blit shader specific optimizations available.

Signed-off-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/9002>

zink: use gallium api to copy to display-target

This allows us to avoid us to avoid forcing linear and host-visible
display-targets.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>

zink: ignore irrelevant bind-flags

We don't need to create display-targets for shared or scanout, becuase
we never even see those in the sw-winsys case.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>

zink: limit host-visible bind-flags

The only type that should really require to be host-visible is the
display-target, and that's just because of our silly flush_frontbuffer
implementation.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>

zink: don't always require linear display-targets

We only need these display-targets to be linear in the case of a
software winsys. In the DRM case, they can be tiled without issues.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>

zink: do not use extra staging resource unless needed

The reason we check for staging-resources here is really because they
are the only images guaranteed to be host-visible.

But on UMA architectures, it's quite likely to have memory that is
*both* host-visible *and* device-local, so let's see what we found
instead.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>

zink: drop extra set of parens

We don't need to be doubly sure here, we can just use a single set of
parents instead.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8858>