review.tizen.org Git - platform/upstream/mesa.git/log

dzn: Drop useless while(link_mask != 0)

That's not a `do {} while();`, just a simple `while() {}`, the second
while statement is a NOP.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Pass the maximum stream size to d3d12_pipeline_state_stream_new_desc()

This way we can use d3d12_pipeline_state_stream_new_desc() directly
without doing

   if (type == GRAPHICS)
      d3d12_gfx_pipeline_state_stream_new_desc()
   else
      d3d12_compute_pipeline_state_stream_new_desc()

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Use d3d12_compute_pipeline_state_stream_new_desc() in the compute path

Fixes: 9feda65a83c ("dzn: Use CreatePipelineState()")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Fix assertion in d3d12_pipeline_state_stream_new_desc()

Fixes: 9feda65a83c ("dzn: Use CreatePipelineState()")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Add missing D3D12_CACHED_PIPELINE_STATE to MAX_COMPUTE_PIPELINE_STATE_STREAM_SIZE

Fixes: 9feda65a83c ("dzn: Use CreatePipelineState()")
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Drop dzn_pipeline_cache.c

The core provides a conformant pipeline cache implementation, let's
use it.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Initialize UUIDs

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

dzn: Fix indentation

Replace tabs by spaces

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17140>

pytracediff: implement pager ('less') invocation internally

In order to get rid of the ntracediff.sh wrapper script, implement
invocation of 'less' internally, if the stdout is determined to
be a tty. Otherwise just print out normally.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

pytracediff: change how 'junk' calls are handled

Instead of discarding them at parsing phase, let the difflib
SequenceMatcher always ignore them, and optionally suppress
them from output if -I/--ignore-junk option is given.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

pytracediff: make -M ("method only") option print arguments for differing calls

Basically implement the last item on the original feature request list of #4609.

Example: ./pytracediff.py good.xml bad.xml -NM

Or suppress common calls completely via -C, e.g. -NC etc.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

pytracediff: add per-line difference highlighting for blocks

Highlight differing _lines_ in the differing blocks, with somewhat
different ANSI colors.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

gallium/tools: reimplement tracediff completely in Python

The limitations of current approach for Gallium trace diffing via
'tracediff.sh' are becoming all too apparent, as we are first dumping
both trace to text and performing plain line-based sdiff on them.

This obviously loses the context of calls and sometimes results in
confusing diffs when subsequent calls are similar enough. It also
prevents us from formatting the diff output in ways that would
benefit readability.

In attempt to rectify the situation, reimplement the diffing completely
in Python, using difflib and adding the necessary plumbing into the trace
model objects etc.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

gallium/tools: add option for ignoring junk calls in trace dumper

Previously tracediff.sh used postprocessing sed-script to remove unwanted
calls from the dump output. Instead of that, add option to parse.py to
ignore a list of calls at parsing phase. Currently this list is hardcoded
in parse.py.

Also clean up the trace model code and pointer tracking a bit to avoid
static state in Pointer class.

Signed-off-by: Matti Hamalainen <ccr@tnsp.org>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Dylan Baker <dylan@pnwbakers.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17135>

trace: Allow to control nir dumping via an environment variable.

As requested by Mike Blumenkrantz.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

trace: Bring state dump up to speed.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

trace: Parse character data more efficiently.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

trace: Dump NIR.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

lavapipe: Prevent mapping buffers beyond their size.

This was breaking trace driver serialization, since it relies upon the
transfer box to know what to serialize.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

trace: Fix framebuffer state serialization.

Ensure the argument name is serialized "state" not, "&tr_ctx->unwrapped_state".

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17107>

dzn: Hook-up device-lost detection

Provide a vk_device::check_status() implementation so the code can
call it at key moments to detect when a device-lost event (device-removed
in D3D12) has been received.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17249>

radv: Use vk_pipeline_hash_shader_stage()

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

radv: Kill unused fs_m local var in radv_create_shaders()

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

tu: Use vk_pipeline_hash_shader_stage()

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

v3dv: Use vk_pipeline_hash_shader_stage()

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

v3dv: Fix nir_shader leaks in v3dv_meta_{clear,copy}()

Reported-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

anv: Use vk_pipeline_hash_shader_stage()

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

vulkan: Add a vk_pipeline_hash_shader_stage() helper

All drivers implement some sort of shader hashing, but each of
them does it slightly differently. Let's provide a generic helper
to avoid new copies of the same logic and encourage new drivers
to use one of the already implemented function.

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

nir/serialize: Silence integer-overflow false positive

Use util_sign_extend() to silence the following integer-overflow
error.

src/compiler/nir/nir_serialize.c:1333:40: runtime error: left shift of 1000165000 by 13 places cannot be represented in type 'int'

Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

nir/serialize: Support texop >= 16

Extend the packed_instr struct to support texops above
nir_texop_fragment_fetch_amd.

Fixes: 603e6ba972da ("nir: add two new texture ops for multisample fragment color/mask fetches")
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17186>

broadcom/compiler: don't predicate postponed spills

The postponed spill is predicated using the condition from the
last write, but this is only correct if the register was only
written once in the TMU sequence, or if it is always written with
the same predication.

While we could try to track whether this is the case or not, it
would make the postponed spill path even more complex than it
already is, so let's just avoid predicating these. We are already
discouraging TMU spilling of registers in the middle of TMU
sequences, so this should not be a very common case.

Cc: mesa-stable
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17201>

broadcom/compiler: fix postponed TMU spills with multiple writes

If we are spilling a register that is used in the middle of a TMU
sequence, we postpone the spill until the TMU sequence finishes,
at which point we inject the spill and rewrite the original
instruction to write to the new temp.

However, this doesn't work if the register is written multiple
times during the TMU sequence. In that scenario, we need to ensure
that all writes are rewritten to use the new temp, not just the last
one.

Cc: mesa-stable
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17201>

v3dv: fix leak

Cc: mesa-stable
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17201>

zink: delete zink_resource_object::dt_has_data

this should be on the swapchain image

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: use kopper_displaytarget type directly

casting this all over is super annoying and unnecessary

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: acquire swapchain images on image map

this is a weird one

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: no-op pixmap frontbuffer flush calls

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: delete zink_resource_object::acquire

this was duplicated from the swapchain object for convenience, but really
it just leads to desync if a swapchain is shared, so use the swapchain
one instead

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: ensure pending present flushes are handled during frontbuffer flush

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: flag batch as having work during frontbuffer flush

avoids having the flush deferred

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: improve no-op acquire detection

the swapchain may have successfully completed the acquire in a previous
batch, in which case a new acquire is not needed

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: remove unnecessary code in kopper readback acquire

this used to be important but no longer is, so modify the loop
code to handle this case implicitly

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: remove broken kopper assert

in a single-buffered scenario this can and does happen

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: use better determination for handling swapchain acquire submits

if the swapchain has data, it has already passed through this function and
thus does not need to attempt to synchronize another swapchain acquire
for the same image

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: move 'acquired' flag onto swapchain struct

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: move to AoS for storing swapchain struct data

this is a bit less cumbersome

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: track whether current swapchain has data

duplicated from resource object since swapchains can be shared

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

zink: fix kopper_acquire return value

this should be a VkResult

Fixes: ab1941fc0e9 ("zink: handle zombie swapchains")

Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17149>

lima: wire up MSAA 4x support

Utgard supports MSAA 4x, so wire it up.

RSW bits were already REd by Luc, the only remaining part was storing
non-resolved buffers, reloading them (including for depth/stencil) and
doing MSAA resolve.

To store non-resolved buffer we need to set mrt_pitch and mrt_bits
registers in WB, and to resolve non-resolved buffer we need to reload
it into individual samples and then write out with mrt_bits = 0, it's
now done by lima blitter.

We also need to do resolve on transfer_map() of multi-sampled buffers,
so utilize u_transfer_helper for that.

As a side fix, it turns out that our wb_reg definition wasn't correct,
'zero' isn't always zero, it's set if we need to swap channels, and
it goes before mrt_bits. mrt_bits actually enables multiple MRTs,
so this commit renames 'zero' to 'flags' and changes its position.

If mrt_bits == 0 and MSAA is enabled, GPU does resolve
in place, to expose this functionality we set PIPE_CAP_SURFACE_SAMPLE_COUNT.

Fixes dEQP-GLES2.functional.multisample.*

Reviewed-by: Erico Nunes <nunes.erico@gmail.com>
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13963>

ci/turnip: Trim the a630 VK run a bit.

We have a lot of spilling coverage in a618 pre-merge, don't do it all (~2
minutes) here. Also, force-gmem touch testing should probably test less than
the default run does!

This should help make up for having added the tu-zink run.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17125>

ci/turnip: Test traces on turnip using zink.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17125>

ci/freedreno: Add vulkan+gl integration testing in piglit.

The libvulkan-dev was needed for building zink, which ended up turning on
the vulkan tests in piglit. Split them out here.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17125>

ci/traces: Drop ZINK_USE_LAVAPIPE ICD override.

Not set in the tree any more.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17125>

ci/traces: GC unused code for DXVK trace replay.

I haven't found any use of it since it was introduced, and it got in the
way of zink trace testing.

Reviewed-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17125>

lavapipe: always set point_tri_clip

this invokes GLES-compatible point clipping, which is more consistent
with vulkan expectations and fixes a number of zink tests

Acked-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17259>

lima,panfrost: Do not ralloc contexts against the screen

ralloc is not thread-safe. While a given context can only be accessed from a
single thread at once, multiple contexts can be created against the same screen
at once. The ralloc allocations against the shared screens will race. Depending
on the result of the race, the same block of memory can be returned as the two
new contexts in two different threads, causing a use-after-free when the context
is freed later.

We free the context explicitly when it's destroyed anyway. If screens are
getting destroyed without the contexts getting destroyed first, that's a state
tracker bug, not a Panfrost one.

This matches what Iris does.

Fixes crash in test_integer_ops.int_math on Panfrost.

Fixes: 0fcf73bc2d4 ("panfrost: Move to use ralloc for some allocations")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17234>

zink: remove swizzle from fbfetch lowering

I had this in at one point to fix something, but now it somehow just
breaks fbfetch instead of fixing anything

cc: mesa-stable

fixes:
dEQP-GLES31.functional.blend_equation_advanced*

Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17254>

glx/dri: Fix DRI drawable release at MakeCurrent time

We want to release the drawables of the context we're coming from, but
we were releasing them from the context we're switching to. This is
probably not a big deal normally because both contexts are likely to be
on the same display, which is all that driReleaseDrawables is really
sensitive to. But if the contexts are on different Displays this would
go quite wrong.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17176>

microsoft/clc: Add a unit test for unused image kernel args

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17246>

microsoft/clc: Remove dead image vars

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17246>

microsoft/clc: Fix test double free in the case of compilation failure

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17246>

microsoft/clc: Enable tests that pass on server 2022

Reviewed-by: Bill Kristiansen <billkris@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17246>

util/disk_cache: Implement disk_cache_get_function_identifier for Windows

Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17208>

intel/clc: enable fp16 & subgroups for GRL

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17253>

clc: add new feature options for intel_clc

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17253>

radv: vectorize nir_op_fabs

Totals from 4 (0.00% of 134913) affected shaders: (GFX10.3)
CodeSize: 37868 -> 36576 (-3.41%)
Instrs: 5332 -> 5169 (-3.06%)
Latency: 24452 -> 24174 (-1.14%)
InvThroughput: 9784 -> 9462 (-3.29%)
VClause: 54 -> 50 (-7.41%)
Copies: 520 -> 519 (-0.19%)
PreVGPRs: 266 -> 264 (-0.75%)

Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15176>

radv: vectorize nir_op_fdiv & nir_op_flrp & nir_op_ineg

These instructions are lowered to fmul/isub.
So, prevent scalarization.

Totals from 49 (0.04% of 134913) affected shaders: (GFX10.3)
VGPRs: 2576 -> 2568 (-0.31%)
SpillVGPRs: 1145 -> 1132 (-1.14%); split: -2.10%, +0.96%
CodeSize: 663968 -> 659376 (-0.69%); split: -1.08%, +0.38%
Scratch: 113664 -> 112640 (-0.90%)
Instrs: 110274 -> 109683 (-0.54%); split: -0.81%, +0.27%
Latency: 2904434 -> 2869588 (-1.20%); split: -1.64%, +0.44%
InvThroughput: 1414237 -> 1396600 (-1.25%); split: -1.69%, +0.44%
VClause: 2899 -> 2891 (-0.28%); split: -0.93%, +0.66%
SClause: 1520 -> 1537 (+1.12%); split: -0.07%, +1.18%
Copies: 28829 -> 28662 (-0.58%); split: -1.90%, +1.32%
Branches: 3560 -> 3564 (+0.11%)
PreVGPRs: 2550 -> 2427 (-4.82%)

Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15176>

radv: use callback for nir_lower_to_scalar

Now uses nir_lower_alu_width.
This avoids scalarization and re-vectorization of 16bit instructions.

Totals from 289 (0.21% of 134913) affected shaders: (GFX10.3)
VGPRs: 12864 -> 13072 (+1.62%); split: -0.50%, +2.11%
SpillSGPRs: 609 -> 505 (-17.08%)
SpillVGPRs: 946 -> 1145 (+21.04%)
CodeSize: 2537024 -> 2576976 (+1.57%); split: -0.10%, +1.67%
Scratch: 89088 -> 113664 (+27.59%)
MaxWaves: 7150 -> 7134 (-0.22%)
Instrs: 458352 -> 460830 (+0.54%); split: -0.45%, +0.99%
Latency: 6615279 -> 6844092 (+3.46%); split: -0.08%, +3.54%
InvThroughput: 1929504 -> 2044989 (+5.99%); split: -0.22%, +6.21%
VClause: 7186 -> 7338 (+2.12%); split: -0.08%, +2.20%
SClause: 13144 -> 13116 (-0.21%)
Copies: 46152 -> 50127 (+8.61%); split: -0.11%, +8.73%
Branches: 16530 -> 16572 (+0.25%); split: -0.02%, +0.27%
PreSGPRs: 14903 -> 14905 (+0.01%); split: -0.01%, +0.03%
PreVGPRs: 11806 -> 11730 (-0.64%); split: -1.83%, +1.19%

Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15176>

radv: don't lower vectorized instructions to 32bit

Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15176>

aco: correctly validate v_fma_mixhi_f16 register assignment

Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15176>

anv: disable injection of primitive shading rate for mesh

It's not needed and causes issues for mesh code (it doesn't
mark the output as per-primitive, which confuses brw_compute_mue_map)

Fixes many tests matching:
dEQP-VK.fragment_shading_rate.dynamic_rendering.*.ms

Fixes: 1542ab70eb4 ("anv: handle primitive shading rate for mesh")
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16196>

intel/compiler: adjust task payload offsets as late as possible

Otherwise passes which expect offsets to be in bytes (like
brw_nir_lower_mem_access_bit_sizes, called from brw_postprocess_nir)
may produce incorrect results.

Fixes 64-bit load/stores in task/mesh shaders.

Fixes: c36ae42e4cc ("intel/compiler: Use nir_var_mem_task_payload")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16196>

intel/common: allocate space for at least one task urb

Fixes: c93cbc77f78 ("intel/common: Add helper for URB allocation in Mesh pipeline")
Acked-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16196>

c11: reinstate the original license and authorship

The original code that was copied in was Boost licensed, so keep that
in. Since Yonggang Luo has code quite some work, keep their copyright
alongside the original one.

Fixes: b2ddec4e98f ("c11: Implement c11/time.h with c11/impl/time.c")
Fixes: e6392fcf3d8 ("c11: Move the implementation of threads.h into c source code")
Acked-by: Yonggang Luo <luoyonggang@gmail.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17232>

intel/blorp: Dirty depth bounds dynamic state bits after blorp

Blorp emits its own 3DSTATE_DEPTH_BOUNDS, so we'll have to re-emit the
expected state after that.

Signed-off-by: Sviatoslav Peleshko <sviatoslav.peleshko@globallogic.com>
Fixes: 56ef501e3aa4 ("blorp: disable depth bounds")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17205>

anv: Dirty all dynamic state bits when creating command buffer state

This makes sure that we'll handle situations when the new state has
the same value as the default one, so we won't dirty some bits, and
consequently will not emit necessary commands (e.g. 3DSTATE_DEPTH_BOUNDS).

Signed-off-by: Sviatoslav Peleshko <sviatoslav.peleshko@globallogic.com>
Fixes: 48229d11 ("anv: don't emit 3DSTATE_DEPTH_BOUNDS in pipeline batch")
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/6722
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17205>

intel: fix typos found by codespell

Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17191>

dzn: Enable the depthClamp feature

depthClampEnable is actually the case we support properly.

!depthClampEnable requires extra work to make sure the
depth clamping that's forced by D3D12 is inactive (setting the
viewport depth range to [0,1] and dealing with the actual range
at the shader level), and clamp the depth value read by the
fragment shader in that case. This will be addressed separately.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17231>

dzn: Enable shader{Clip,Cull}Distance

DXIL has clip/cull distance builtins too.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17231>

dzn: Enable dynamic indexing on all kind of descriptors

nir_to_dxil() supports it.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17231>

dzn: Advertise shaderImageGatherExtended support

nir_to_dxil() takes tg4 offsets into account.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17231>

dzn: Advertise anisotropic filtering support

We support it already, let's toggle the switch to expose it.

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17231>

nir/serialize: Put dest last in packed_instr::tex

packed_instr::tex::dest must be last to match the packed_instr::any::dest
position.

Fixes: 35655865cbde ("nir/serialize: pack instructions better")
Cc: stable
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17212>

anv: silence border color swizzle debug message

MESA-INTEL: debug: gfx11_CreateSampler: ignored VkStructureType 1000411001

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17245>

radv: dump UMR waves before UMR rings

Dumping UMR rings might be slow and dumping waves before would make it
more chance to dump them without reporting "No active waves".

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/17183>

radv: fix command line for dumping waves with UMR

GFXOFF must be disabled before dumping waves and re-enabled after.

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/17183>

radv: use RADEON_FLAG_VA_UNCACHED for the trace BO

Figured this while debugging a GPU hang with a simple CTS test. This
is to make sure data written by the CP are coherent on the CPU.

This also explains spurious GPU hang reports generated for Hitman 3
that made no sense without it. Now it's clear that this game hangs
after a DRAW_INDEX_INDIRECT packet.

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/17183>

radv: disable small primitive culling for user sample locations

The driver can't assume sample positions at (0.5, 0.5) when user
sample locations are used.

This doesn't fix anything in practice because NGGC is only enabled by
default on GFX10.3 and that extension is currently disabled on GFX10+,
but I would like to expose it at some point.

This fixes dEQP-VK.pipeline.*.sample_locations_ext.verify_location.*
(when the extension is enabled locally).

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17228>

v3dv: Implement VK_KHR_performance_query

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14061>

ci: Revert "CI: Lima farm is offline"

The lab is up and running again.

This reverts commit 686e20afcd32cb80b96984c8eb4428e777fbce97.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Acked-by: Vasily Khoruzhick <anarsoul@gmail.com>
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17240>

radeonsi: replace llvm gs input handle with nir lowering

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

ac/llvm: get back nir_intrinsic_load_gs_vertex_offset_amd

Will be used by radeonsi.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

ac/nir: add triangle_strip_adjacency_fix to gs input lower

From radeonsi.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

radeonsi: replace llvm es output with nir lowering

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

ac/nir: change es output lower param to esgs_itemsize

radeonsi may add extra dword to the stride, so let's pass it
directly.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

ac/nir: remove unused param num_reserved_es_outputs from gs input lower

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

radeonsi: set lds for gs/es to handle nir shared memory load/store

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>

ac/nir: skip gl_ViewportIndex and gl_Layer write in ES

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>