zink: fix texture barriers for real this time

according to spec, this is supposed to handle fragment shader fetch
from previous draw output, not color output readback from previous
color output write

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10706>

zink: break out draw dispatch into separate functions

this is much easier to read and is going to greatly simplify the eventual
multidraw implementation which will be dropped in

also it allows moving conditionals outside of loops to very slightly improve
drawoverhead performance (with multidraw)

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10701>

zink: remove compute cruft from resource mapping

this used to sync the compute batch, now it does nothing

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10707>

zink: add debug assert to verify that zink_clear_framebuffer() is accurate

this should never be called for buffers that don't have pending clears

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10705>

zink: break zs clear loop once both bits are set when beginning renderpass

no point in looping this any more than we have to

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10705>

zink: handle in-renderpass clears in fb_clears_apply_internal()

this isn't used yet, but it may as well be handled in case that changes

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10705>

zink: only set layer info for samplerviews if there are multiple layers

fixes a bunch of validation errors

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10703>

zink: simplify samplerview surface creation

instead of manually creating the ivci, there's already a util function for
that which will handle everything

also only set layer info if there are multiple layers

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10703>

zink: add a target param to create_ivci()

allow creating views of different types than the base resource

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10703>

zink: clamp 3D surface viewtype to 2D only in the create_surface hook

this is only applicable to things that go through the hook, not for internal
surfaces

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10703>

zink: make ZINK_INLINE_UNIFORMS more standardized in function

Reviewed-by: Joshua Ashton <joshua@froggi.es>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10696>

android: nir: add nir_lower_fragcolor.c to Makefile.sources

Fixes the following building error:

FAILED: out/target/product/x86_64/obj/SHARED_LIBRARIES/gallium_dri_intermediates/LINKED/gallium_dri.so
...
ld.lld: error: undefined symbol: nir_lower_fragcolor
>>> referenced by pan_assemble.c:81 (external/mesa/src/gallium/drivers/panfrost/pan_assemble.c:81)

Cc: 21.0 21.1 <mesa-stable@lists.freedesktop.org>
Fixes: 1fd356302590 ("nir: add lowering pass for fragcolor -> fragdata")
Acked-by: Rob Clark <robclark@freedesktop.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10712>

broadcom/compiler: change register allocation policy for accumulators

The current policy is to always favor accumulators if possible, however,
this is not always optimal.

Particularly, accumulators play a crucial role in enabling QPU instruction
merges, since these are limited to both the ADD and the ALU instructions
addressing at most 2 physical registers. For 2-src instructions, this means
that to be able to merge we need them to address at least 2 accumulators.

While favoring accumulators does help the case for instruction merges in
general, it is risky to assign accumulators to variables that have
long life spans. Doing so will make the accumulator unavailable for
any other instructions during that life span, and since we only have a few
accumulators, we can quickly run out and losing our capacity to merge
instructions for large parts of the qpu program.

On the other hand, we also want to avoid the extreme case were we keep
allocating physical registers to the point we run out, even if we have
accumulators available, since accumulators have additional restrictions
and may not be suitable for everything.

This change continues the policy of favoring accumulators, but it only
does so if the life span of the temps is short, to ensure that we can
recycle accumulators often across instructions and avoid running out
for sections of the QPU code, unless we are already running out of
physical registers.

total instructions in shared programs: 13654647 -> 13336921 (-2.33%)
instructions in affected programs: 11015919 -> 10698193 (-2.88%)
helped: 39758
HURT: 17325
Instructions are helped.

total threads in shared programs: 412046 -> 412038 (<.01%)
threads in affected programs: 16 -> 8 (-50.00%)
helped: 0
HURT: 4
Threads are HURT.

total uniforms in shared programs: 3745726 -> 3746003 (<.01%)
uniforms in affected programs: 17296 -> 17573 (1.60%)
helped: 76
HURT: 99
Uniforms are HURT.

total max-temps in shared programs: 2364430 -> 2359942 (-0.19%)
max-temps in affected programs: 109117 -> 104629 (-4.11%)
helped: 2893
HURT: 772
Max-temps are helped.

total spills in shared programs: 5727 -> 5746 (0.33%)
spills in affected programs: 221 -> 240 (8.60%)
helped: 1
HURT: 2

total fills in shared programs: 13121 -> 13139 (0.14%)
fills in affected programs: 466 -> 484 (3.86%)
helped: 1
HURT: 2

total sfu-stalls in shared programs: 33432 -> 34491 (3.17%)
sfu-stalls in affected programs: 18219 -> 19278 (5.81%)
helped: 4459
HURT: 5087
Inconclusive result

total inst-and-stalls in shared programs: 13688079 -> 13371412 (-2.31%)
inst-and-stalls in affected programs: 11030017 -> 10713350 (-2.87%)
helped: 39630
HURT: 17429
Inst-and-stalls are helped.

total nops in shared programs: 335753 -> 333708 (-0.61%)
nops in affected programs: 112659 -> 110614 (-1.82%)
helped: 8726
HURT: 7383
Inconclusive result

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10686>

lavapipe: Fix lvp_pipeline_compile's nir_xfb_info leak.

v2: Cleanup, as suggested by Mike Blumenkrantz.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10688>

lavapipe: Fix lvp_execute_cmds' pipe_stream_output_target leak.

v2: Unbind SO targets, as suggested by Mike Blumenkrantz.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10688>

nv50/ir: Initialize CodeEmitterNV50 member progType.

Make progType a constructor argument.

Fix defect reported by Coverity Scan.

Uninitialized scalar field (UNINIT_CTOR)
uninit_member: Non-static class member progType is not initialized
in this constructor nor in any functions that it calls.

Suggested-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10562>

nv50/ir: Initialize BuildUtil member tail.

Fix defect reported by Coverity Scan.

Uninitialized scalar field (UNINIT_CTOR)
uninit_member: Non-static class member tail is not initialized in
this constructor nor in any functions that it calls.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10561>

nvc0: Remove unnecessary bsp_bo NULL check.

Fix defect reported by Coverity Scan.

Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking bsp_bo suggests that it may be
null, but it has already been dereferenced on all paths leading
to the check.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10564>

nv50/ir: Initialize Graph::Node member tag.

Fix defect reported by Coverity Scan.

Uninitialized scalar field (UNINIT_CTOR)
uninit_member: Non-static class member tag is not initialized in
this constructor nor in any functions that it calls.

Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10563>

zink: never use LINEAR for VK_EXT_4444_formats

we shouldn't be reading these back, and some drivers lie about what features
are supported in linear tiling anyway

Reviewed-by: Joshua Ashton <joshua@froggi.es>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10702>

gallium: Remove unused st_visual::render_buffer

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10675>

nv50: Fix use of initializers on older compilers

Fixes the following error on Ubuntu 18.04 with GCC 7.

src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp:1448:72:
sorry, unimplemented: non-trivial designated initializers not supported
info_out->prop.cp.gmem[gmemSlot++] = {.valid = 1, .slot = i};

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Fixes: f451854f3 ("nv50: add remapping of buffers/images into unified space")
Signed-off-by: Ernst Sjöstrand <ernstp@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10668>

freedreno/a5xx: SP_BLEND_CNTL has per-mrt blend enable bit

Blending in SP_BLEND_CNTL is not a binary flag but the same
mask as in RB_BLEND_CNTL. It is a per-mrt enable bit for blending.

Copied form a6xx, on a5xx it should be have the same since it seems
to have the same structure layout.

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10682>

turnip,freedreno/a6xx: SP_BLEND_CNTL has per-mrt blend enable bit

Blending in SP_BLEND_CNTL is not a binary flag but the same
mask as in RB_BLEND_CNTL. It is a per-mrt enable bit for blending.

Example SP_BLEND_CNTL produced by blob on a630 and different MRT
blendings:
  SP_BLEND_CNTL: { UNK8 | 0x6 }
  SP_BLEND_CNTL: { ENABLED | UNK8 | 0xe }
(Decoded before this commit)

Fixes mis-rendering with D3D11 game "Spelunky 2".

Signed-off-by: Danylo Piliaiev <dpiliaiev@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10682>

lavapipe: fix fsum with swizzle

We can do stuff like this:

vec1 32 ssa_207 = fsum3 ssa_209.xxx

In this case, we'd end up not swizzling in get_alu_src, and reading
components out-of-bounds, which LLVM isn't very happy about, and thus
takes punitive actions, in the form of a segfault.

We don't want that, and we already know from the opcode what the
component counts should be here.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10690>

pan/bi: Add divergent intrinsic lowering pass

This is undefined behaviour in the hardware but perfectly legal NIR, so
lower to an if ladder predicated on the lane ID (the blob's preferred
strategy).

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10022>

pan/bi: Add "lanes per warp" accessor

Varies by architecture version.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10022>

pan/bi: Map load_subgroup_invocation to FAU

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10022>

pan/bi: Drop load_sampler_lod_parameters_pan

Only used for an erratum workaround on Mali T720.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10022>

nir/divergence_anlysis: Add intrinsics for Bifrost

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10022>

radv: fix extending the dirty bits to 64-bit

New dynamic states added for VK_EXT_extended_dynamic_state2 causes
GPU hangs with vkd3d-proton.

Fixes: 7bdd569d7e1 ("radv: extend the dirty bits to 64-bit")
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/10693>

panfrost: Use common blend lowering

Contains a number of bugfixes.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10601>

nir: Add blend lowering pass

This pass was originally developed for Panfrost, where it passes the
relevant dEQP tests. Upstreaming so it can be extended and then shared
with:

* Asahi, for blending
* Zink, for logic ops
* Lavapipe, for advanced blending

Note that using this with MRT in a fragment shader (as non-panfrost
drivers will) has not yet been tested. Logic ops with integer
framebuffers are probably todo. It's been enough for Panfrost, will
suffice for ES2 on Asahi, and provides an upstream base for kusma's work
on advanced blending, so overall the merge is a net benefit.

v2: Remove bogus assert that the format layout is PLAIN. We need to
render R11G11B10, which Mesa reports as layout OTHER. The code is still
correct.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com> [v1]
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10601>

pan/lower_framebuffer: Fix bitsize mismatch

No idea why this never manifested before.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10601>

zink: add a pipe_context::clear_buffer hook

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10679>

mesa: add an extension MESA_bgra

This GLES extension allows to combine the formats BGR and BGRA
as host-side formatsto be combined with the  internal formats
RGB8/SRGB8 and RGBA8/SRGB8_ALPHA8 respectively.

This extension is of interest to support a subset of OpenGL in
virtualized environments where the host only supports GLES.

Initial mesa/glformat.c patch: rohan.garg@collabora.com

v2: - Correct names for ClearTexture calls
    - Add BGR(A)_EXT tokens
    - Add format check for BGR_EXT
      (All Adam Jackson)

v3: Fix ordering in extension table (Marge)

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Adam Jackson <ajax@redhat.com> (v2)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10613>

panfrost: Use pan_blit() when PAN_MESA_DEBUG=panblit

Hook-up support for native blits. We keep using u_blitter by default
for now.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Extend pan_blitter to support blit/resolve operations

Right now the lib is just used for TB preloads. Let's make it generic
to support actual blits.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Stop assuming the viewport will always cover the framebuffer

That's not necessarily true for actual blits, so let's pass the min/max
coords to pan_blitter_emit_viewport().

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Don't select the blit shader fragout type twice

The type has already been selected when forging the key, no need to do
it again pan_blitter_emit_bifrost_blend().

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Shrink the number of args passed to prepare_{bifrost,midgard}_rsd()

There's clearly no need to pass RTs/ZS views around, pass only what we
really need.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Rename pan_preload_emit_varying()

This function will be used for blits too.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Make pan_preload_emit_viewport() applicable to blits

We will emit one viewport and attach it to several DCDs for multi-layer
blits. Make that possible by adjusting the prototype and rename the
function along the way.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Make pan_preload_emit_*_textures() applicable to blits

The texture descriptors will be emitted once and re-used in several
DCDs when blitting more than one layer. Rename the functions and make
them return a GPU pointer instead of filling the DCD directly.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Stop assigning ->position in pan_preload_emit_varying()

It doesn't make sense to have it done in pan_preload_emit_varying().

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Make pan_preload_emit_*_sampler() applicable to blits

The sampler descriptors will be emitted once and re-used in
several DCDs for the multi-layer blit case. We will also need
to select the filter. Let's adjust the pan_preload_emit_*_sampler()
functions to support that and rename them along the way.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Get rid of the vertex_count arg in pan_preload_emit_varying()

We always pass 4 anyway.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Pack pan_blit_surface fields

It should speed up a bit hash calculation and allow us to add extra
info without changing the structure size.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Fix blit shader names

first should be initialized to true if we want to get rid of the leading
';' in the shader name.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

panfrost: Pass an image view to panfrost_estimate_texture_payload_size()

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10548>

gallium/tools: add option to use Meld for diffing

Add option -m/--meld to tracediff.sh to use Meld
<https://meldmerge.org/> instead of sdiff for comparing
traces.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: improve pointer type tracking in parse.py

In our simplistic model of assigning types to pointer, we treat
return values specially because their "type" can't be known
easily before their first use. Improve the "ret" handling by
removing one from their count when we reassign the type to
something else.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: implement "high-level" overview mode option in dump scripts

As per the suggestion in #4609, implement mode/option -M/--method-only
which only prints call method names, for quick overview of what is
happening in the trace. The same option can be used with both
dump.py and tracediff.sh.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: improve tracediff.sh argument handling

Implement better argument parsing/handling in tracediff.sh
so that the options passed to dump.py and sdiff are not required
to be positional.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: use left-column output mode of sdiff in tracediff.sh

Enabling --left-column option for sdiff inside tracediff.sh we
get output that only shows the differing lines on the right column.

This, when used with the -N/--named option should make many
trace diffs more understandable.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: implement 'named' pointers option in dump.py

As suggested by zmike in issue #4609, raw pointer values in
are problematic for diffing state dumps. In attempt to remedy
this, we implement an option --named/-N in dump.py that
generates symbolic names for all pointers based on the context.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: implement better suppression of variants

Previously some variants (such as execution time and call number
were suppressed in tracediff.sh via a sed script. It makes sense
to implement an option to leave out such variants to begin with
in dump.py, so let's do so and use it.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: improve option handling in dump_state.py

Previously we inherited some options from parse.py, but
that made no sense for some of the options that are not
needed for dump_state.py (such as --plain, as we output
only JSON format text.)

So, remove the inherit and implement filename argument
here independantly.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

gallium/tools: clean up tracediff.sh a bit

Clean up tracediff.sh shellscript slightly, partially in preparation
for upcoming changes. Also add a signal handler for removing the
created temporary directory and files if we are interrupted.

Also, use the new --plain option of dump.py to suppress output
of ANSI color/formatting codes, so we do not have to post-filter
them later.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10648>

pps: Documentation

Add documentation for Perfetto and Gfx-pps, together with some perfetto
config files to use as a starting point.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Signed-off-by: Rob Clark <robdclark@chromium.org>
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: John Bates <jbates@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9652>

pps: Gfx-pps config tool

Add helpful tool to query pps capabilites such as supported devices,
counters and counter groups, and to dump counter values to stdout.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: John Bates <jbates@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9652>

pps: Gfx-pps v0.3.0

Add the gfx-pps backbone in `src/pps`.

v2: Simplify supported drivers creation.
v3: No default getter is provided for counters.
v4: Open DRM device in read/write mode.
v5: Wait for datasource to be started.
v6: Set FIFO scheduler while sampling counters.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: John Bates <jbates@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9652>

util: Perfetto SDK v15.0

Add Perfetto SDK v15.0 as a dependency using a meson wrap.

Signed-off-by: Antonio Caggiano <antonio.caggiano@collabora.com>
Acked-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Rob Clark <robdclark@chromium.org>
Reviewed-by: John Bates <jbates@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9652>

util/hash_table: _mesa_hash_table_create_u32_keys()

Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10673>

v3d: re-enable GLSL loop unrolling

Disabling it that made us start to fail register allocation for a
few tests. Re-enable it until we figure out what is causing that.

Fixes: ca9e0871fb294 ('v3d: enable NIR loop unrolling')

Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10684>

r600/sfn: lower to scalar with filter applied

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9943>

gallium: pass lower_to_scalar_filter to lower_to_scalar pass

v2: Fix the callback name

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9943>

nir: Add filter callback for lower_to_scalar to the options

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9943>

gallivm: handle texture arrays in non-fragment shaders with lod.

We have to unwind the lod into the scalar path correctly.

Fixes a crash with renderdoc demo

Fixes: e168d148d76d ("gallivm/nir: handle non-uniform texture offsets")
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10676>

zink: start using dynamic front face state

this doesn't actually do anything other than test the codepath since
the hashed pipeline state will still change on every frontface change,
but it's a start

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: hook up dynamic dsa states

this is pretty straightforward, just adding some extra calls

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: return current pipeline object if state hasn't changed

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: flag pipeline for change more often when vbos change without dynamic state

this is always going to be bad for perf, but we still need to force updates anytime
we know for sure they'll be needed

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: split vertex state pipeline hashing into its own value

by creating a separate hash value for vertex state, changes to the vbos and
bindings requires massively less overhead as it doesn't require the rest of
the pipeline state to be rehashed

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: move more vertex state stuff into the hw state

this simplifies a lot of the hashing since only the vertex state pointer
needs to be hashed

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10508>

zink: check descriptor layout support before creating it

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: use explicit types during descriptor updates

less derefs

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: abstract descriptor pool usage for programs

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: abstract descriptor usage for programs

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: abstract descriptor init

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: unify pipeline layout creation functions

these codepaths are almost identical

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

zink: split off descriptor layout from descriptor pools

in theory we could have multiple pools per layout, but really it just makes
more sense to make the layouts persistent since they're more likely to be
reused

32bit compile fixes by Witold Baryluk

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9965>

panfrost: Enable 16-bit support on Bifrost

Remove the PAN_MESA_DEBUG=fp16 flag that was hiding it.

Skip two buggy dEQP tests. See linked discussion. We'll need to make
sure this gets sorted out before submitting conformance, but I don't see
a test with a fix in the pipeline as valid reason to hold back valid
code.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Enable mediump BLEND lowering

Other lowerings will wait until we iron out various missing features.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Garbage collect bifrost_nir.h

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Copyprop constants

Needed for constant folding to be effective. But don't copyprop into
instructions already reading from FAU, that will just end up adding more
moves!

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Fix int<-->float size converts

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Enable NIR vectorization

We don't vectorize transcendentals, since those are scalar only in
hardware. Also don't vectorize a few places where impedance mismatches
between NIR and the hardware make handling vectors infeasible for now.

Signed-off-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Temporarily switch back to 0/~0 bools

Keeps things simpler while debugging vectorization woes.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Handle make_vec with 1-bit bools

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Adapt branching for 1-bit bools

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/bi: Change swizzled scalars to identity

Allows packing for things like IADD.v2s16

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

panfrost: Fix typo handling blend types

This was right in my head.

Fixes: 93a176b6cff ("panfrost: Key blend shaders to the input types")
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

pan/mdg: Model blend shader interference

Backport of 4439757db23 ("pan/bi: Use the interference mechanism
to describe blend shader reg use") to Midgard.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/9239>

mesa: s/malloc/calloc/ to silence a warning

gcc 11 warns:

[846/1506] Compiling C object src/mesa/libmesa_common.a.p/main_shaderapi.c.o
In function ‘shader_source’,
    inlined from ‘_mesa_ShaderSource_no_error’ at ../src/mesa/main/shaderapi.c:2137:4:
../src/mesa/main/shaderapi.c:2095:25: warning: ‘*offsets_10 + _130’ may be used uninitialized [-Wmaybe-uninitialized]
 2095 |    totalLength = offsets[count - 1] + 2;

I can't really see how it's getting to that conclusion, but allocating
`offsets` with calloc is both natural to do here and guarantees
initialization.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10671>

format/fxt1: Clean up fxt1_variance's argument list

gcc 11 warns:

../src/util/format/u_format_fxt1.c:940:22: warning: ‘fxt1_variance.constprop’ accessing 128 bytes in a region of size 64 [-Wstringop-overflow=]
  940 |    int32_t maxVarR = fxt1_variance(NULL, &input[N_TEXELS / 2], n_comp);

But, suspiciously, if you inline fxt1_variance the warning goes away.
What's happening is that the 2nd arg is uint8_t[N_TEXELS][MAX_COMP], so
it looks like we're passing too small of an array in since gcc knows
that `input` is also [N_TEXELS][MAX_COMP]. Fair enough. Fix the
signature to reflect what's actually going on, and remove some unused
arguments while we're at it.

Reviewed-by: Matt Turner <mattst88@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10671>

radv: advertise VK_EXT_extended_dynamic_state2

This only implements dynamic primitive restart enable, depth bias
enable and rasterizer discard enable. I leave logic op and patch
control points for later.

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/10340>

radv: implement dynamic rasterizer discard enable

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/10340>

radv: implement dynamic primitive restart enable

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/10340>

radv: implement dynamic depth bias enable

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/10340>