intel/perf: add DG1 support

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: add RKL support

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/dev: identify rocketlake

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: break TGL perf configs in GT1/2

Programming and equations are different enough that we really need 2
files.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/dev: identify tigerlake

We'll need that to pick the right query sets between TGL/RKL/DG1.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: add async compute metrics

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: small ICL equation refactor

No functional changes.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: update files from IGT

IGT has received a bunch of updates, this is resyncing the files with
it.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: remove reordering script

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: reorder xml files

Make the file match the order of the ones from IGT (which have changed
because of python2->3 transition).

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: add reorder script

When transitioning the oa-*.xml files from Gputop to IGT, we also had
to deal with a python2->3 transition. Unfortunately the implementation
dependent hash table ordering leaked into the XML files and so things
changed quite a bit.

This script reorders things from the old to the new order in the
existing files.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: rename lkf into ehl

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

anv: remove unused query pool field

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

anv: fix layout comment

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Marcin Ślusarz <marcin.slusarz@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: switch query code to use query layout

That way we can describe new registers to that could be used both by
Anv & Iris/i965 without having to modifying code in multiple places.

v2: Do reverse order for begin queries so that we have MI_RPC as close
    as possible from the drawcall

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: add performance query layout using MI_SRM

For all generations supported we had a layout describing what register
to store to implement a MI_RPC replacement.

This is because, on Gen12 we need to snapshot OAG registers to get
correct values for the perf equations. There, the MI_RPC instruction
captures OAR register which do not have all the information we need.

v2: Fix commented code for debug (Marcin)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: query register descriptions

This will be useful when we implement queries using a series of MI_SRM
instead of MI_RPC.

Unfortunately on Gen12, the MI_RPC command sources values from the OAR
unit which has a similar series of register as the OAG unit but some
of the configuration of HW doesn't reach OAR so we have to snapshot
OAG manually instead.

v2: Fix comments
    Use const

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

intel/perf: prep work to enable new perf counters

Those are not part of the OA reports and need some additional
scaffolding. Those counters are only available when doing queries as
we need to emit MI_SRMs to record them.

Equations making use of those counters are not there yet, they will
come in a follow up commit updating a bunch of oa-*.xml files.

v2: Fix typo

v3: Use PERF_CNT_VALUE_MASK (Marcin)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

genxml: PERFCNT registers are available since HSW

We were using those registers on Gen7.5 in the GL driver already, we
just need them in Genxml for Anv too.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6518>

iris: don't emit IRIS_DIRTY_VF depending on trash in restart_index

The `restart_index` field can be uninitialized if `primitive_restart`
is false so we have to track `restart_index` changes
only if `primitive_restart` is true

Here is a valgrind warning:
Conditional jump or move depends on uninitialised value(s)
==52021==    at 0x6D44968: iris_update_draw_info (iris_draw.c:102)
==52021==    by 0x6D450B5: iris_draw_vbo (iris_draw.c:273)
==52021==    by 0x642FD8E: cso_multi_draw (cso_context.c:1708)
==52021==    by 0x5C434D3: st_draw_gallium (st_draw.c:271)
==52021==    by 0x5DF5F1B: _mesa_draw_arrays (draw.c:554)
==52021==    by 0x5DF68F7: _mesa_DrawArrays (draw.c:768)
==52021==    by 0x49011F2: stub_glDrawArrays (piglit-dispatch-gen.c:12181)
==52021==    by 0x11C611: piglit_display (shader_runner.c:4549)
==52021==    by 0x4994D83: process_next_event (piglit_x11_framework.c:137)
==52021==    by 0x4994E47: enter_event_loop (piglit_x11_framework.c:153)
==52021==    by 0x49939A4: run_test (piglit_winsys_framework.c:88)
==52021==    by 0x49821A9: piglit_gl_test_run (piglit-framework-gl.c:229)

v2: - don't propagate trash to state->cut_index
    (Kenneth Graunke <kenneth@whitecape.org>)

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Andrii Simiklit <andrii.simiklit@globallogic.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8409>

panfrost: Update ctx->batch when a fresh batch is requested

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Add a panfrost_compile_shader() helper

This deduplicates the

   if (pan_is_bifrost())
      return bifrost_compile_shader_nir();
   else
      return midgard_compile_shader_nir();

pattern.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Use dev->arch where appropriate

The architecture has already been extracted in panfrost_open_device()
don't do it again.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Rename and move pan_render_condition_check()

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Set attribs and attrib_bufs to NULL when attrib_count = 0

It's just easier to read pandecode traces when those pointers are set
to NULL for the attribute_count=0 case.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Only allocate the extra attribute buffer entry on Bifrost

Bifrost needs an empty attribute buffer entry to tell the prefecter it
should stop fetching attribute buffers, but Midgard doesn't have this
constraint. It's also useless to have 2 empty entries for the instance or
image case.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Don't memset the last attribute buffer entry twice

It's already done a few lines below.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Get rid of IS_BIFROST

Extract this information from dev->arch, and provide a helper to hide
this check.

Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Reviewed-by: Italo Nicola <italonicola@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

panfrost: Fix tiler job injection (again)

2f1947b39ca5 ("panfrost: Fix tiler job injection") had the tests
inverted: WRITE_VALUE jobs are only needed on Midgard, not Bifrost.

Cc: mesa-stable
Fixes: 2f1947b39ca5 ("panfrost: Fix tiler job injection")
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8808>

broadcom/compiler: improve generation of if conditions

Where it is safe to do so, avoid the generation of code to convert a
condition code into a boolean which is then tested to generate a
condition code. This is only done in uniform ifs, and only for condition
values that are SSA and only used once (in that if statement).

shader-db relative to MR 7726:

total instructions in shared programs: 8985667 -> 8974151 (-0.13%)
instructions in affected programs: 390140 -> 378624 (-2.95%)
helped: 810
HURT: 276
helped stats (abs) min: 1 max: 49 x̄: 17.77 x̃: 16
helped stats (rel) min: 0.10% max: 33.63% x̄: 7.97% x̃: 6.45%
HURT stats (abs)   min: 1 max: 46 x̄: 10.42 x̃: 10
HURT stats (rel)   min: 0.16% max: 21.54% x̄: 2.26% x̃: 2.03%
95% mean confidence interval for instructions value: -11.46 -9.75
95% mean confidence interval for instructions %-change: -5.76% -4.97%
Instructions are helped.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8709>

radeon: decrease the size of radeon_cmdbuf by switching prev fields to uint16

This also removes the 32-bit hole in radeon_cmdbuf.

Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8434>

radeonsi: tune NGG shader culling vertex threshold for each chip

These are based on my testing and estimation.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8434>

radeonsi: simplify the NGG culling condition in si_draw_vbo

Changes:

- disallow NGG culling for GS, fast launch for tess using template args
  (GS can't do NGG culling, tess can't do fast launch)

- skip checking current_rast_prim with tessellation
  (bake the condition into ngg_cull_vert_threshold)

- use only 1 vertex count threshold for enabling NGG shader culling
  to simplify it. I think it doesn't have a big impact. The threshold
  computation depends on more parameters than just fast launch.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8434>

radeonsi: set current_rast_prim at bind time for tess and GS

It doesn't have to be done in draw_vbo.

Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Zoltán Böszörményi <zboszor@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8434>

intel: combine common gather routines in INTEL_MEASURE

Anv and iris had separate, similar routines to gather intel_measure
timestamps.  Timestamps are now managed within intel_measure, allowing
those routines to be consolidated.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

intel: support secondary command buffers in INTEL_MEASURE

When a secondary command buffer is encountered, insert an event that
links to the new batch.

This commit leaves intel_measure timestamp buffer objects mmapped,
which is more efficient than mapping/unmapping several times.  With
the BOs mapped at all times, timestamp buffers can be managed directly
by intel_measure, where it will iterate over timestamps of linked
secondary buffers.

With timestamp buffers managed by intel_measure, a more efficient and
accurate check for render completion can be moved into intel_measure
from anv/iris.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

intel: stop tracking submission state in INTEL_MEASURE

With secondary command buffers, it is inconvenient to track whether a
batch has been submitted and needs to be gathered.  Instead, always
check for completed snapshots before destroying a command buffer.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

iris: add hooks to call INTEL_MEASURE

These hooks were written in the initial IRIS_MEASURE implementation.
Minor changes by Mark Janes <markjanes@swizzler.org> to adapt to the
INTEL_MEASURE reimplementation.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

iris: add a iris_context reference to iris_batch

This eliminates the need to use container_of in error handling code.
INTEL_MEASURE will need to access the iris context from each batch.

suggested-by: Kenneth Graunke <kenneth@whitecape.org>

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

iris: implement iris layer of INTEL_MEASURE

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

anv: add hooks to call INTEL_MEASURE

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

blorp: add hook for INTEL_MEASURE

Saves the snapshot type within the blorp parameters.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

anv: implement anv layer of INTEL_MEASURE

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

anv: enable timestamp for INTEL_MEASURE

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

intel: Print GPU timing data based on INTEL_MEASURE

This infrastructure collects GPU timestamps over common intervals, and
generates a CSV report to show how long rendering took.  The overhead
of collection is limited to the flushing that is required at the
interval boundaries for accurate timestamps.

By default, timing data is sent to stderr.  To direct output to a
file:
 INTEL_MEASURE=file=/tmp/measure.csv {workload}

To begin capturing timestamps at a particular frame:
 INTEL_MEASURE=file=/tmp/measure.csv,start=15 {workload}

To capture only 23 frames:
 INTEL_MEASURE=count=23 {workload}

To capture frames 15-37, stopping before frame 38:
 INTEL_MEASURE=start=15,count=23 {workload}

Designate an asynchronous control file with:
 INTEL_MEASURE=control=path/to/control.fifo {workload}

As the workload runs, enable capture for 5 frames with:

 $ echo 5 > path/to/control.fifo

Enable unbounded capture:

 $ echo -1 > path/to/control.fifo

and disable with:

 $ echo 0 > path/to/control.fifo

Select the boundaries of each snapshot with:
 INTEL_MEASURE=draw  : DEFAULT - Collects timings for every render
 INTEL_MEASURE=rt    : Collects timings when the render target changes
 INTEL_MEASURE=batch : Collects timings when batches are submitted
 INTEL_MEASURE=frame : Collects timings at frame boundaries

With INTEL_MEASURE=interval=5, the duration of 5 events will be
combined into a single record in the output.  When possible, a single
start and end event will be submitted to the GPU to minimize
stalling.  Combined events will not span batches, except in
the case of INTEL_MEASURE=frame.

Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7354>

radv: Enable modifiers with the WSI.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Enable DRM format modifiers on GFX9+.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Add image layout with drm format modifiers.

Half of it is passing the right modifier to ac_surface, the other half
is applying the offset/strides.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Add drm format modifier queries.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Add format modifier format queries.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Don't relayout images with modifiers.

The modifier should have been the exact layout of the image. Hence
we should not relayout the image according to the implicit modifier
on memory import (i.e. VkMemory allocation).

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Use the surface offset from ac_surface instead of a plane offset.

In preparation for doing this with modifiers in general.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

radv: Extract DCC format support handling.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

amd/common: Add modifier size helper.

For use in Vulkan.

Reviewed-By: Chad Versace <chad@kiwitree.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7667>

nir/from_ssa: don't check for interference within the same set

Signed-off-by: Yevhenii Kolesnikov <yevhenii.kolesnikov@globallogic.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8246>

nir/from_ssa: consider defs in sibling blocks

If def a and def b are in sibling blocks, the one with higher
parent_instr's index does not necessarily come after the other.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/3712
Fixes: 943ddb945877fc8 "nir: Add a better out-of-SSA pass"
Signed-off-by: Yevhenii Kolesnikov <yevhenii.kolesnikov@globallogic.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8246>

Scons: check for timespec_get on windows as well as unices

windows has a timespec_get function, and if we don't check that then it
will try to fall back to using clock_gettime, which windows doesn't
have.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8592>

vulkan: Drop the type_prefix parameter from gen_extensions

Now that all the drivers are converted, it's set to 'vk' by everyone so
there's no point in having the parameter.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Rework vk_device_init and friends

Now that all drivers are converted over, we can make a few changes.
First off, vk_device_init no longer takes two separate allocators
because we can assume that the parent instance is non-null and it can
pull the instance allocator from that.  Second, dispatch tables and the
instance extension table are no longer optional.  We leave the device
extension table optional for now because we don't do any verification at
vk_init_physical_device time and some drivers find it more convenient to
set the extensions later in their own physical_device_init for various
reasons.

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv,radv: Use common entrypoints for VK_KHR_deferred_operation

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Make the debug_report implementation internal

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Use vk_object_base::type for debug_report

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv: Use vk_object_base::type for debug_report

This is way better than the giant macro thing we had before.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Make vk_debug_report_callback derive from vk_object_base

Fixes: 51c6bc13ce3a "anv,vulkan: Implement VK_EXT_private_data"
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: Switch to the common VK_EXT_debug_report

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: Use common entrypoints for VK_EXT_private_data

Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: port to using common dispatch code.

I think this has the SQTT layer hooked in correctly, would
be good if somebody could confirm this.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: move to subclassed instance/physical_device structs

This moves to using the common base structs for these
two objects, this is prep work for the using the common
dispatch layer code.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: remove all entrypoint enabled debug option

This was useful back at the dawn of time when apps weren't as
well developed and layers mostly sucked. I don't think it's been
used in a quite a while so remove it, as the new dispatch
layer doesn't support it.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

radv: move queue object to a common base object

This is needed to use the new dispatch layer code.  While we're here, we
clean up the context on the error path.

Fixes: 9b1138e3f0e96 "radv: implement VK_EXT_private_data"
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: Switch to the common VK_EXT_debug_report

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: Use common entrypoints for VK_EXT_private_data

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: port to using common dispatch code.

This moves v3dv over to using the new common dispatch layer code.

v2 (Jason Ekstrand):
 - Remove some now dead function declarations

Acked-by: Iago Toral Quiroga <itoral@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: remove reference to v3dv_instance on v3dv_physical_device

As we already have a reference to vk_instance at vk_physical_device,
that we are setting when calling vk_physical_device_init.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: Drop v3dv_instance::app_info

There's an equivalent data structure in vk_instance.

Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

v3dv: move to subclassing instance/physical device

This moves to using the common base structs for these two objects, but
doesn't use any of the new features yet.

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

lavapipe: Drop some wrappers in favor of common code

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

turnip: Switch to the common VK_EXT_debug_report

Acked-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

turnip: Drop some legacy wrappers in favor of common code

Acked-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

turnip: Use common entrypoints for VK_EXT_private_data

Acked-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

turnip: Use the common dispatch framework

Acked-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

lavapipe: use common dispatch layer.

This moves lavapipe over to using the new common dispatch layer code.

v2 (Jason Ekstrand):
 - Less pointless messing around with python and meson

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

lavapipe: sort extensions in proper order.

Subsequent patch enforces this, so just do it in advance.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

lavapipe: add missing wsi entrypoint.

Add support for GetPhysicalDeviceSurfaceFormats2KHR which is helpful
to make the new dispatch code work.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

lavapipe: move to subclassing instance/physical device.

This doesn't use any of the features yet, just moves to subclassing.

Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv: Switch to the common VK_EXT_debug_report

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add a truly common VK_EXT_debug_report implementation

Now that we've got a common vk_instance, we can put the debug_report
stuff there and make it truly common.  For drivers to use this
implementation, they need to delete their own vk_debug_report_instance
and make sure everything references the common one.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan,anv: Move VK_KHR_copy_commands2 wrappers to common code

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan,anv: Move a bunch of trivial wrappers to common code

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv: Make Get*MemoryRequirements a wrapper

Instead of making Get*MemoryRequirements2 call the legacy version to
fill out the base struct, put everything in Get*MemoryRequirements2 and
make the old one a trivial wrapper.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan,anv: Add common entrypoints for VK_EXT_private_data

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan,anv: Move GetDeviceProcAddr to common code

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add framework for common entrypoints

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv: Use the common dispatch framework

This commit switches ANV to using the new common physical device and
instance base structs as well as the new dispatch framework.  This
should make code sharing between Vulkan drivers much easier.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

anv: Add a single anv_genX switch macro

This should make future platform enabling a good bit easier and more
reliable than having 3 of them.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add a common entrypoint table generator

This is based on the one in ANV but it's a bit different because it uses
vk_entrypoint_table instead of vk_dispatch_table.  This is because,
without knowing what extensions are implemented by a driver, we have to
generate a table with everything and sort it all out later.  We use the
same trick as ANV with weak function declarations to provide NULL values
for any entrypoints NOT implemented by the driver.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add common Get*ProcAddr implementations

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add entrypoint tables and related helpers

Entrypoint tables are distinct from dispatch tables because they are not
de-duplicated.  These are needed to make codegen easier.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add generators for instance trampoline functions

ANV needs these because we have a different dispatch table for each
hardware generation and Vulkan requires that the device entrypoints
returned from vkGetInstanceProcAddr work regardless of which device
they're used with.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>

vulkan: Add common instance and physical device structs

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/8676>