venus: drop VkQueueBindSparse

Sparse binding is already disabled with fence feedback enabled by
default due to the difficulty getting both to work.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: remove filtering external semaphores for QueueSubmit

We now require VK_EXTERNAL_SEMAPHORE_FEATURE_IMPORTABLE_BIT for
external semaphores.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: require exportable/importable bit for ext semaphores sync fd

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: require exportable bit for ext fence sync fd

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: require importable external semaphores for WSI

Semaphore sync fd import required for WSI to skip scrubbing
the wsi/external wait semaphores.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: refactor VK_KHR_synchronization2 ext sync fd requirements

In preparation for adding additional dependency requirements for
external sync fd support.

Move vn_physical_device_init_external_* so external sync fd support can
be retrieved earlier. Then move sync2 disabling to
vn_physical_device_get_passthrough_extensions and 1.3 downgrading
to vn_physical_device_init_properties.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

venus: refactor out vn_feedback_event_cmd_record2

Refactor out vn_feedback_event_cmd_record2 into a shared functon
that downgrades pipeline barrier commands for non synchronization2
commands.

Signed-off-by: Juston Li <justonli@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20088>

gbm/backend: fix gbm compile without dri

Commit introduces a fix that allows for gbm to be built with an empty
backend. There are situation especially in a Yocto/OE cross compilation
environment where you want to build with an empty backend. The particular
situation is as such:

The mesa-gl recipe is the preferred provider for virtual/libgbm, virtual/libgl,
virtual/mesa, etc... But the x11 DISTRO_FEATURE in't included this leads to build
errors such as:

| /../../../ld: src/gbm/libgbm.so.1.0.0.p/main_backend.c.o: in function `find_backend':
| backend.c:(.text.find_backend+0xa4): undefined reference to `gbm_dri_backend'
| /../../../ld: src/gbm/libgbm.so.1.0.0.p/main_backend.c.o:(.data.rel.ro.builtin_backends+0x4):
                undefined reference to `gbm_dri_backend'
| collect2: error: ld returned 1 exit status

Issue should be replicable by setting -Ddri3=disabled and -Dgbm=enabled

Add fix to bypasses compilation issue by excluding gbm dri backend. If
HAVE_DRI || HAVE_DRIX not specified.

Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Signed-off-by: Vincent Davis Jr <vince@underview.tech>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20447>

dzn: Fix clear bind flag logic

This is the patch I had meant to merge

Fixes: 8b79e6fb ("dzn: No need to add another bind flag for clears if one is already present")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20361>

dzn/ci: Add image test group, which is all passing now

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20361>

dzn/ci: Remove flakes/fails that don't hit anymore

The fails weren't causing pipeline failures because the tests just
aren't run, thanks to the dEQP fraction.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20361>

ci/windows: Download updated WARP 1.0.4 package

Acked-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20361>

mesa: Trivially advertise NV_generate_mipmap_sRGB

... if you have EXT_framebuffer_sRGB. The extension just relaxes an
error check that we're already not performing, and sRGB rendering
implies sRGB texture support, and mipmap generation would need it to be
a valid render format. So advertise it if EXT_framebuffer_sRGB works.

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19476>

mesa: Fix extension table formatting

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19476>

radv: Use the new NIR builder tex helpers for meta

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19480>

nir/builder: Add some texture helpers

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19480>

nir: Add more opcodes to nir_tex_instr_is_query()

Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19480>

glx: Harmonize glXCreateGLXPixmap with glXCreatePixmap

This was open-coding some things, and not perfectly. Likewise for the
destroy path.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Fix drawable type inference in visual/fbconfig setup

This was just all kinds of wrong. Note that the comment implying "this
is a workaround for old XQuartz" is on the "not apple" side of the
ifdef. Anyway. xserver didn't start sending GLX_DRAWABLE_TYPE in the
fbconfig until:

    commit 8cde0af3c57f0375ba8ba77af9fdf74b79d9496d
    Author: Kristian Høgsberg <krh@redhat.com>
    Date:   Wed Apr 2 19:06:40 2008 -0400

        Send the GLX_EXT_texture_from_pixmap attributes to the client.

So we can remove this default from the fbconfig path. But we preserve it
for the GLXGetVisualConfigs path, because that is specified not to send
GLX_DRAWABLE_TYPE.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Inline a few single-use constant strings into their user

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Move 1.2 GLXPixmap code into glx_pbuffer.c

We're going to rename this file to something like glx_drawable.c
eventually, but for right now let's just get all the create/destroy calls
in the same place.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Check for initial "glX" first in glXGetProcAddress

This makes us match eglGetProcAddress, otherwise you scan the entire GLX
function table on every glXGetProcAddress.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Reflow MakeContextCurrent a little

First, move a few early-out checks up since they don't need to take the
GLX lock. Second, move garbage collecting deleted contexts up to
immediately after they are unbound. This fixes a memory leak, albeit a
difficult one to hit, in the case where you switch away from a deleted
context but switching to the new one errors out. In that case we would
leak the deleted context, since it's been unbound from all threads and
there's no longer an XID for it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Clean up some funny business from context bind/unbind

We always fully unbind the old context before binding the new one, so
there's no point in passing both contexts to both the unbind and then
the bind.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Remove support for glXGetDriverConfig for old drivers

gallium implements __DRI_CONFIG_OPTIONS version 2.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Remove a can't-happen NULL check

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx: Remove glx_context::screen

We're already keeping a back pointer to the glx_screen which already
tracks this, it's not important enough to keep the screen number in two
places.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

glx/dri3: Simplify protocol version tracking

This is really just a single elaborate capability check, so stash a
boolean in the display state for it.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20549>

asahi: Handle sampler->compare_mode

Instead of smashing unconditionally to 1. Not sure if this fixes anything but it
gets rid of an unknown at least. Possibly slightly faster.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20561>

aco: Pass correct number of coords to Vega 1D LOD instruction.

If we pass a physical 2D texture descriptor we should also pass 2
coords. Otherwise it just uses the random content in the second
register which ends up funny sometimes.

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20696>

nir/nir_opt_move: fix ALWAYS_INLINE compiler error

Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Closes: #6825
Fixes: f1d20ec6 ("nir/nir_opt_move: handle non-SSA defs ")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17439>

radv: fix re-emitting RB+ when the non-compacted color format changes

If the previously emitted graphics pipeline uses the value A for
col_format_non_compacted and the new bound graphics pipeline uses B.

At bind time, radv_cmd_state::col_format_non_compacted will be set to
B and the rbplus flag will be dirtied. But if there is no draws and a
new graphics pipeline is bound with the same value as A, the next
draw will emit the rbplus state with B instead of A.

This can be basically triggered with meta operations after drawing
because the driver saves/restores the bound pipeline.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/8073
Fixes: 11469f7553d ("radv: copy the non-compacted color format at pipeline bind time")
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20692>

radv: fix hashing pipeline keys if RADV_PERFTEST=ngg_streamout is used

On GFX10-10.3, RADV_PERFTEST=ngg_streamout can be used to replace the
legacy path but it should be in the pipeline key. Otherwise, it's
possible to load legacy shaders from the cache instead of NGG.

Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20668>

zink/ci: relocate radv testing from radv's gitlab-ci.yml

Zink testing should happen in the layered-backends stage, not the AMD
one.

Suggested-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20677>

zink/ci: add a fail to the VG flake list

It seems to have happened only once so far, but let's be proactive as
an unexpected pass is fatal.

Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20677>

zink/ci/radv: remove a test from the fails list

It seems `spec@ext_texture_array@fbo-depth-array stencil-draw` got
fixed last week, so let's update the expectations.

Acked-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Acked-by: Daniel Stone <daniels@collabora.com>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20677>

ci/piglit: compress results.csv before uploading it to GitLab

The results.csv file of a full piglit run is about 6 MB.

Given how seldomly this file is being used, and the fact that it cannot
be viewed directly in gitlab's artifact page anyway.

Let's compress the file using zstd, and enjoy a ~90% reduction in size
at the cost of probably less than 500ms of compression time on a slow
device, and 55ms on the CI machines in the valve farm.

Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20669>

ci/deqp-runner: compress results.csv before uploading it to GitLab

The results.csv file can get ridiculously big for a vkcts run (up to
135MB). Given how seldomly this file is being used, and the fact that
it cannot be viewed directly in gitlab's artifact page anyway.

Let's compress the file using zstd, and enjoy a ~95% reduction in size
at the cost of probably less than 1 second of compression time on even
the slowest of the devices in CI (which would use sharing), and about
150ms on the CI machines in the Valve farm.

Suggested-by: Daniel Stone <daniels@collabora.com>
Acked-by: David Heidelberg <david.heidelberg@collabora.com>
Signed-off-by: Martin Roukala (né Peres) <martin.roukala@mupuf.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20669>

Revert "ci: Collabora's LAVA lab for maintance"

Farm up again

This reverts commit 900ccec7e8fa809ef0a9154db4d0ecec3fe26e9c.

Signed-off-by: Sergi Blanch Torne <sergi.blanch.torne@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20633>

ci: disable Collabora's LAVA lab for maintance

This is to inform you of some planned downtime in the LAVA lab as follows:

    Start: 2023-01-13 08:00 GMT
    End: 2023-01-13 12:00 GMT

Signed-off-by: Sergi Blanch Torne <sergi.blanch.torne@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20632>

radv: fix missing implementation of creating images from swapchains

These pNext structs are part of VK_KHR_swapchain which is core Vulkan
1.1 but they were missing. Loosely based on ANV.

Passed WSI CTS.

Cc: 22.3 mesa-stable
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7797
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20641>

intel/utrace: document tracepoints

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

intel/ds: remove unused trace point

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

intel/ds: add INTEL_GPU_TRACEPOINT envvar to toggle tracepoints

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

anv: add support for command buffer tagging in traces

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

vulkan/debug_utils: copy debug util labels

The copied structures currently can point to freed memory.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

util/u_trace: add support for variable length trace points

Use case being :

struct tp {
   ...
   char string[0];
};

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

intel/ds: move event_id access to perfetto lambda

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

intel/ds: track secondary cmdbuffers in perfetto

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

anv: record secondaries' traces into primaries

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

pps: print out message when we get the first counters

Mostly for debug purposes, show the first visible counter timestamp.

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16655>

agx: Peephole select after opt_preamble

Reduces control flow in Dolphin uber shaders, which saves us a few cycles.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20597>

nir/peephole_select: Allow load_preamble

load_preamble is intended to be almost free (costing at most a move), and it
does not have special bounds checking requirement, so it's ok to select with it.
With this, drivers that use nir_opt_preamble together with a late call to
peephole_select can optimize sequences like:

   if (x) {
      <uniform-on-uniform calculation>
   } else {
      <different uniform-on-uniform calculation>
   }

to simply

   bcsel(x, <uniform register 0>, <uniform register 1>)

rather than emitting needless control flow / branching over some moves.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20597>

zink: simplify get_slot_components() for xfb emission

multiplying by the array size is always wrong for this case, and not
doing so allows for some simplification and better inlining, though
the output results are identical

the one corner case is clip/cull distance, which need special handling
since they're arrays with vec semantics

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20678>

nir/lower_bool: ntt: Generate a good opcode for bcsel

This is heavily copy-pasted from a patch of Ian Romanick, including the
commit message.

Previously, this pass always generated fcsel for bcsel.  This was the
only place that generate fcsel, so various drivers assumed (and needed!)
that src0 was a Boolean with 0.0 or 1.0 as the only values.

Specifically, many DX9 / GL_ARB_vertex_program platforms lack a CMP
instruction in vertex shaders.  In those cases, they would use LRP to
implement fcsel.  The bummer is that many plaforms have a real fcsel
instruction, and those platforms would benefit from other places
generating that opcode.

Instead of leaving assumptions in drivers about the sources of an opcode
that they can't really support, allow them to control the way the
lowering pass translates bcsel.  Two flags are used to control this:

- If the driver sets has_fused_comp_and_csel in nir_options, fcsel_gt
  will be used.  Since the Boolean value is 0.0 or 1.0, this is
  equivalent to fcsel.

- If the parameter has_fcsel_ne is set, fcsel will be used.  This is the
  old path.

- Otherwise, the lowering pass assumes we're on a crufty, old DX9 vertex
  program, and it emits flrp.

With this, the assumptions about src0 of fcsel in NTT can be removed.
If a platform can't handle fcsel, it should ensure that the lowering
pass won't generate it.

No change in shader-db.

Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Reviewed-by: Emma Anholt <emma@anholt.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20162>

i915: Enable generation of fcsel_gt and fcsel_ge opcodes

DX9 PS 1.x / GL_ARB_fragment_program shaders that have been converted to
GLSL are littered with patterns like

    ps_r1.x = ((ps_t0.y >= 0.0) ? ps_r1.x : ps_c1.y);

This is because CMP is a fundamental opcode in those earlier shading
languages.  i915 supports this opcode natively, but there's no way to
get it directly into the backend.  Instead, NIR and NTT generate some
combination of fcsel and sge and hope for the best.

i915
total instructions in shared programs: 49032 -> 48897 (-0.28%)
instructions in affected programs: 4173 -> 4038 (-3.24%)
helped: 39
HURT: 0

total temps in shared programs: 2795 -> 2790 (-0.18%)
temps in affected programs: 22 -> 17 (-22.73%)
helped: 5
HURT: 0

total const in shared programs: 4976 -> 4967 (-0.18%)
const in affected programs: 203 -> 194 (-4.43%)
helped: 9
HURT: 0

GAINED: shaders/trine/fp-13.shader_test FS

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20162>

r300: Enable generation of fcsel_gt and fcsel_ge opcodes

RV530:
total instructions in shared programs: 135032 -> 134833 (-0.15%)
instructions in affected programs: 15654 -> 15455 (-1.27%)
helped: 133
HURT: 13

total temps in shared programs: 17051 -> 17045 (-0.04%)
temps in affected programs: 202 -> 196 (-2.97%)
helped: 11
HURT: 5

RV370:
total instructions in shared programs: 84439 -> 84378 (-0.07%)
instructions in affected programs: 2519 -> 2458 (-2.42%)
helped: 49
HURT: 7

total temps in shared programs: 12399 -> 12380 (-0.15%)
temps in affected programs: 322 -> 303 (-5.90%)
helped: 20
HURT: 10

GAINED: shaders/godot3.4/22-69.shader_test FS

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20162>

nir/lower_int_to_float: Add support for i32csel opcodes

These lower naturally to the corresponding fcsel opcodes.

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20162>

ntt: Add support for fcsel_gt and fcsel_ge opcodes

These match the TGSI CMP opcode very nicely.  Every driver that uses NTT
for its fragment shader path should be able to enable these opcodes now.

Reviewed-by: Emma Anholt <emma@anholt.net>
Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20162>

docs: reset new_features.txt

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20680>

VERSION: bump to 23.1.0-devel for further development

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20680>

intel/genxml: Drop CACHE_MODE_SS definition.

This is a global register which isn't settable by userspace contexts.
It also shouldn't appear in any of our aubinator decodes from error
states or aub dumps, as no userspace batch should be setting it.

So it's not very valuable to have here.  Just makes us think we can
set it.  Plus, a lot of the field definitions changed a bunch, and
would need updating.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20627>

radv: Fix deadlock in radv_rmv_log_event_create

Fixes: 8d0e6c0 ("radv: Add RMV tracing utilities")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20636>

dzn: Declare debug only root_dwords as ASSERTED

Such variable is only used in assert(), so declare it as ASSERTED.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7885
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7234

Signed-off-by: GH Cao <driver1998.ms@outlook.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20676>

anv: Add a tracepoint for the fallback implicit sync wait path.

If you're here, you'd really like to know.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20658>

perfetto: Add the intel data sources to system.cfg.

Now I don't need to remember different perfetto setups depending on which
system I'm profiling.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20658>

docs/perfetto: Fix the name of the i915 render stages data sources.

Fixes: ef27399bca95 ("docs: update perfetto with the latest status")
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20658>

ci: build hasvk if we're building anv

!19355 should never have happened, but we didn't bother to add build
coverage for hasvk when we split it out from anv.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19356>

vulkan/rmv: Only trace on the n-th frame

This makes it stop dumping on every n-th frame. It also uses a simple
increment, since the function is locked.

Fixes: defed48 ("vulkan: Add common RMV tracing infrastructure")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20635>

radv/rmv: Capture names of pipelines but not command buffers

For command buffers, only scratch/upload bos are logged which cannot be attributed to command buffer handles.

Fixes: 5611ab25 ("radv: Add RMV tracing layer")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20647>

radv/rmv: Fix resource ids for name tokens

Fixes: 5611ab25 ("radv: Add RMV tracing layer")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20647>

hud: fix values printing

Oops...

Fixes: 595079c37cc ("hud: extract float printf modifer selection logic to helper")
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20643>

util: add a return value to util_sprintf

The regular sprintf is expected to return the number of char writter,
so let's do the same in our version.

Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20643>

nv30: add nv49 results

Acked-by: Karol Herbst <kherbst@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19696>

r300: don't cache abs in fragment shader

Currently this ends up with unneeded move.

fixes: 63f353b45605379b8b81754543907a6b0fc3afdd

Reviewed-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20625>

zink: fix PIPE_RESOURCE_PARAM_NPLANES with format modifier

Some format modifiers change the number of planes used by an image.
For instance AMD DCC modifiers uses 2 or 3 planes. However the
format modifier was ignored in the PIPE_RESOURCE_PARAM_NPLANES
get_param hook.

Fix this by using get_dmabuf_modifier_planes() instead of
util_format_get_num_planes().

This fixes wlroots-based compositors under zink.

Signed-off-by: Simon Ser <contact@emersion.fr>
Fixes: c025cb9ee9d7 ("zink: fix dmabuf plane returns")
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Reviewed-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20395>

gen_release_notes: include links in relnotes.rst when generating the new release note

This is required to allow the docs to build, which in turn is required
if we want to allow merge requests against release (staging) branches.

Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20460>

gen_release_notes: allow using the script from another checkout

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20459>

gen_release_notes: avoid crashing when none of the commits mention closing an issue

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20459>

gen_release_notes: stop the script if we can't generate the release note

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20459>

frontends/va/postproc: yuv422 to nv12

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Tested-by: Suresh Guttula <suresh.guttula@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19915>

gallium/auxiliary/vl: compute shaders for progressive yuv

v2: Add a one line offset to the compute shader, to get the
correct output, as suggested by Suresh <Suresh.Guttula@amd.com>.
v3: Add `FALLTHROUGH` to fix a compilation error

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Tested-by: Suresh Guttula <suresh.guttula@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19915>

frontends/va/postproc: default to weave when deinterlacing

Signed-off-by: Thong Thai <thong.thai@amd.com>
Tested-by: Marcus Seyfarth <m.seyfarth@gmail.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19915>

gallium/auxiliary/vl: fix scale and translate parameters

Fixes the scale and translate portion of the code to allow for
scale and crop to work properly.

Signed-off-by: Thong Thai <thong.thai@amd.com>
Reviewed-by: Leo Liu <leo.liu@amd.com>
Tested-by: Marcus Seyfarth <m.seyfarth@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19915>

docs: utilities: Update list of development utilities

Explicitly mention supported APIs and add GFXReconstruct for
Vulkan tracing and debugging.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20474>

docs: debugging: Fix path to dlist.c

This file has now long being moved to another directory.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20474>

radv/gfx11: increase radeon_check_space for occlusion query begin

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20652>

dri: Free the already allocated optionCache and optionInfo on failure

These fields are allocated as they have to be taken into account for initScreen.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20663>

dri: Do not free the given screen in initScreen implementation

The given screen is already freed by the caller in case a NULL-pointer is
returned by the implementation.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20663>

gbm: Avoid leaks on screen creation failures

Some of the code paths were not freeing the allocated strings,
also remove the unused ret variable as we are always returning -1 on failure.

Signed-off-by: Corentin Noël <corentin.noel@collabora.com>
Reviewed-by: Simon Ser <contact@emersion.fr>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20667>

lima/ci: add deqp-egl run to the deqp suite

Both wayland and X backends are covered by headless weston, with
X enabled through weston Xwayland.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Acked-by: Vasily Khoruzhick <anarsoul@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20392>

lima/ci: change lima jobs to use deqp-runner suite

Align it with the state of other drivers' ci and make it
easier to enable more tests later.

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20392>

r600: Request that state validation is emitted for all changed states

Restricting the state change only to newly used states results in problems,
because SSBOs, Images, and the framebuffers make use of the same limited
set of resources, and not properly unbinding un-used resoureces leads to
invalid rendering and GPU hangs.

Fixes: aaa4b0e6
   st/mesa: move check_program_state code into _mesa_update_state

v2: use new cap name and switched meaning

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7969

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20493>

gallium: Add cap to request state validation for all dirty state

With aaa4b0e6 state validation is no longer called for all changed states,
but only for states that will be active with a new shader program.
Not all drivers support this and might prefer if the state validation
is emitted for all states that might be changed. So add a cap that the
driver can signal one or the other preference, and default to the new
behavior.

Fixes: aaa4b0e6
   st/mesa: move check_program_state code into _mesa_update_state

v2: - Rename cap and and invert its meaning, query the cap
      only once and store it in st, handle the mask update
      when updating the shader i.e. not in st_validate_state (Marek)

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20493>

docs: move note on other drivers to the bottom

This was always meant to be at the bottom of the page. To reduce the
risk of more driver-specific environment variables being added below,
let's add a horizontal rule to mark the difference. This should make it
more clear that this paragraph doesn't belong to the previous heading.

Fixes: c70c5ecd2e7 ("docs: move generic gallium envvars to root doc")
Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20644>

ac/nir/ngg: implement 16bit output streamout

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20350>

ac/nir/ngg: refine nogs outputs handling

Gather outputs in advance to save both output data and type. Output data
is used for streamout and gfx11 param export. Output type is used for
streamout latter.

The output info will also be used for nir vertex export in the future.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Singed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20350>

ac/nir/ngg: fix gs 16bit output uninitialized channel when gfx11

Fixes: abe2e99e9e5 ("ac/nir/ngg: gs support 16bit outputs")
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20350>

ac/nir/ngg: fix gs store output for no param offset slot when gfx11

When slot has no param offset, we should not emit store output for
them on gfx11.

Fixes: abe2e99e9e5 ("ac/nir/ngg: gs support 16bit outputs")
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20350>

ac/nir/ngg: always reset output when gs emit vertex

Follow the spec, all outputs even not this stream need to be
reset after emit vertex.

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20350>