freedreno/drm/virtio: Guest side fence waits

Now that fd_fence and fd_submit_fence are unified, we can wait on the
fence fd rather than needing to poll the host.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm/virtio: More tracepoints

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno: Limit cached staging BOs to PIPE_MAP_READ

WC is fine for staging upload buffers, and will let us hit the default
heap.  Probably introducing a separate cached-coherent heap is overkill,
and given that staging uploads are common (but downloads uncommon, out-
side of dEQP/piglit), it is better to just ask for WC.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm/virtio: Remove prealloc_list

Now that we are allocating cmdstream from the heap, we shouldn't need
this.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Add sub-allocator

Add a heap that we can use for allocations of small mappable buffers.
This avoids the churn of mmap/unmap, which is especially expensive in
a VM.  It also allows packing more smaller allocations together in a
page, which is useful for PIPE_BUFFERs (which are also mappable).

This avoid jank caused by the overhead of setting up or tearing down
guest mappings when running in a VM.  And also significantly reduces
the # of BOs referenced on a submit.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

util/vma: Add "nospan"

Add a way to request that allocations do not span a specified PoT
boundary.  This can be used, for ex, to ensure that allocations do
not span 4GB boundaries to work around hw/fw bugs.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Merge in_fence_fd's

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Move no_implicit_sync accounting

No need to duplicate this.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Drop fd_bo_fence

Remove the last fence-like-struct that isn't fd_fence.  Now that
fences are refcnt'd we can just attach them directly to BOs.  This
will simplify busy-ness tracking for suballoc BOs.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Re-enable submit fence merging

Now that we (a) create the fence ourselves, returning a reference from
fd_submit_flush(), and (b) refcnt the fences, we can now merge submits
with on out-fence by just attaching the same fence to all the submits.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Simplify deferred submit flushing

Once we are sharing an out-fence for multiple deferred/merged submits
the complicated logic in fd_pipe_sp_flush() no longer works (or makes
sense), so just simplify all the paths into a single helper.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Return fence from submit flush

This moves away from embedding the submit fence inside the pipe fence,
which lets us start refcnt'ing the fence.  This will enable several
cleanups and improvements:

1. Get rid of fd_bo_fence, and just have fd_bo hold pending fd_fence
   refs instead, which will be needed for cpu_prep implementation of
   sub-allocated buffers.
2. For merged submits, we can just return a new reference to an
   existing fence.

Note that this temporarily defeats submit-merging, which will be
fixed (and improved) in a following commit.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno: Rename fd_fence -> fd_pipe_fence

So we don't have namespace clashes in the next commit.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Combine fd_fence and fd_submit_fence

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Replace shared/nosync fields

These are redundent with the alloc_flags.  A following commit will map
alloc_flags directly to bo heap that should be used, and adding the
FD_BO_SHARED bit when a BO is exported neatly avoids returning it to
the pool to be potentially erroneously reused.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Reorg BO alloc flags

And drop the unneeded _FD_BO_RING flag.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Invert BO destruction

Sub-alloc'd heap BOs will need to do something a bit more special, ie.
not immediately cleanup fences (because those will be needed to know
when the BO is actually idle and vma node can be released), not tear
down the mmap (because the BO doesn't own it), and not close the handle
(also because the BO doesn't own it).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Shuffle around BO init

Suballocated heap BOs will bypass the rest of bo_new() (as they skip the
BO cache, don't get an entry in the handle table, etc).  So move more of
the initialization into fd_bo_init_common().

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Drop pipe ref to device

The context and screen already holds refs to the device.  But since
fd_bo_fence holds a ref to the pipe, anyone anywhere leaking a single
bo with unwaited fences could cause the device to leak indirectly.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Get rid of fd_bo_del_locked()

This moves the table_lock into a small critical section in the BO delete
path when we are actually removing the entries from the handle/name
tables, so finally table_lock isn't the big-bo-lock!

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Rework batch BO delete

Get rid of the per-device table of deferred handles, and instead keep
this on the stack so that no additional locking is needed.  This will
simplify getting rid of table_lock in the bo delete path.  For the BO
cache, add fd_bo_del_list_nocache() which works like fd_bo_del_array()
except that it bypasses returning BOs to the cache.  This gets rid of
the BO cache calls to private fd_bo fxns (which is useful now that the
bo_del/close_handles() dance has become a bit more complicated).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Start prying apart table_lock

We want to get back to the point where table_lock is only doing what
it's named for (ie. protecting handle/name tables).  This is the first
step.  Add new locks to protect fencing and the bo caches.  A single
global lock for fences, to avoid atomic instructions in the loop per
submit bo attaching fences.

This couldn't be broken appart into different commits, otherwise there
would have been an ABBA deadlock in BO cache's call to fd_bo_state()
to check if a BO is idle.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Simplify cleanup_fences()

To prepare for moving fence state out from under table_lock, simplify
cleanup_fences() and stop trying to re-use it in the bo destruction
path (which otherwise wouldn't need the fence lock, since there is
no other owner of the bo ref at that point, and simply needs to drop
pipe refs).

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Add bo list iterator macros

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Add BO cache stats

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Allow submit merging with explicit sync

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Move no_implicit_sync to base

No need to duplicate between backends.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno/drm: Remove cpu_fini

It has been unused since nearly forever.

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

freedreno: Update obsolete comment

Signed-off-by: Rob Clark <robdclark@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20263>

venus: fix signaled fence creation under multiple timeline

Fixes: 3beb4f055e17 ("venus: allow renderer submissions on per-context virtgpu rings")

Signed-off-by: Yiwei Zhang <zzyiwei@chromium.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20366>

asahi: Implement anisotropy

Passes KHR-GLES3.texture_filter_anisotropic.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Identify seamful cube map bit

Fixes

   dEQP-GLES2.functional.texture.mipmap.cube.basic.linear_nearest

when run with a GLES2 version.

We wire up seamless cube maps for GLES3+ only, working around an obscure
mesa/st limitation. See 6148e3aae7e ("mesa: Fix
ctx->Texture.CubeMapSeamless") for the full context.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Set flatshading controls appropriately

Fixes dEQP-GLES3.functional.rasterization.flatshading.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Identify XML for more flatshading controls

Names from PowerVR <3

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Enable VS_INSTANCEID and VERTEX_ELEMENT_INSTANCE_DIVISOR caps

These two should be functional at this point. The latter is required
(and quite well tested) by Darwinia.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Implement depth and stencil export

For gl_FragDepth, passes dEQP-GLES3.functional.shaders.fragdepth.* and piglit
fragdepth_gles2.

For stencil export, passes piglit glsl-fs-shader-stencil-export.

For gl_FragDepth together with stencil export, passes
dEQP-GLES3.functional.fbo.blit.*

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

agx: Implement depth and stencil export

Lower FRAG_RESULT_DEPTH and FRAG_RESULT_STENCIL writes to a combnied zs_emit
instruction with a multisampling index. To be used in the following commit.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Allow texturing S8 portion of combined Z/S

Comes up in gles3.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Put meta shader keys into the meta shader itself

The hash table needs a key pointer with at least the lifetime of the
hash entry, which the key pointer we get does not have (since it is
stack-allocated by agx_build_meta). Copy it into the shader struct
itself and use that for the hash table.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Identify & disable triangle merging for shaders using derivatives

It seems triangle merging is incompatible with calculating derivatives
along primitive edges correctly. Take the appropriate NIR shader info
flags in the compiler and pass them down as a flag to the driver, so it
can set the disable triangle merging flag (formerly called "lines or
points").

TODO: Is this what macOS does when you set a sample mask there (which
apparently fixes the same bug on the Darwinia Metal backend)? Do we
also need to set this when sample masks are used?

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Fixes Darwinia and dEQP2 projected tests.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Fix include guard comment on decode.h

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Allocate enough push ranges for the worst possible case

We need one for every possible sysval, plus up to 16 VBOs.

Fixes plasma-systemmonitor.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Fix off-by-one assert in agx_create_vertex_elements

Needed to fix plasma-systemmonitor.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Consider shared resources always valid

Shared resources are in an unknown state, so we have to assume they
may contain valid data.

Fixes Xorg+Kwin random full-screen garbage on mouse cursor present.

However, we still get black flashing in some circumstances (that was
garbage flashing before this commit), which indicates that something
is reading an uninitialized/fresh shared framebuffer and expecting
valid data (prior screen contents?) TBD.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Ensure we always have space for stream links + padding

It seems VDM can overread (maybe it does blockwise transfers?), so we
always need to leave space at the end of encoder buffers to avoid
faults.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Don't crash on VBOs without resources

There's a Firefox QuakeJS heisenbug that triggers this sometimes, by
binding vertex attributes with no buffer...

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

asahi: Do not leak staging resources for PIPE_MAP_READ

We were only dereferencing the resource in the PIPE_MAP_WRITE path of
agx_transfer_unmap, which means that read-only transfers leaked the
staging resource/BO. Always free the staging resource unconditionally.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

agx: Fix packing of extension for block image stores

Probably impossible to hit in practice but let's get it right. Found when
forcing RA to use the upper half of the reg file.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

agx: Coalesce i2i16 and u2u16

Extract out the code for unpack_64_2x32_split_x and use it for other integer
downcasts too to coalesce out a move. Pointless, but I wanted to have a little
RA fun after getting stencil export working.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

ail: Assert that the mip level is in bounds

This preempts possible out-of-bounds accesses and later asserts when
trying to get the tile size.

Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

ail: Fix tile sizes

Fixes dEQP-GLES3.functional.texture.filtering.2d.sizes.3x7_nearest_mipmap_linear.

Tested for all sizes 1..256x1..256.

Tested-by: Asahi Lina <lina@asahilina.net>
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

ail: Fix typo

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

ail: Always allocate the full miptree

Layer strides are based on the full miptree, and even for single-layer
images macOS always allocates a full one (possibly relevant for
compression). Make sure we do the same, regardless of how many mip
levels the user asked for.

Fixes Darwinia.

Reviewed-by: Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
Signed-off-by: Asahi Lina <lina@asahilina.net>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

docs/features: Mark ARB_occlusion_query2 as done on asahi

Part of GLES.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

docs/features: Mark panfrost supporting aniso

On Mali-G72+ (except for some buggy early G72 models).

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

nir: Add store_zs_agx intrinsic

Will be used for frag depth/stencil export with multisampling.

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20365>

ci: Lima farm maintenance

General lab maintenance and addressing
https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20051

Signed-off-by: Erico Nunes <nunes.erico@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20343>

r600/sfn: Fix test compilation with -fno-rtti

79ca456b4837b3bc21cf9ef3c03c505c4b4909f6 reintroduced the use
of dynamic_cast<> in r600/sfn tests.  This breaks compilation with
-fno-rtti, as required to build against the LLVM configuration
recommended upstream.  Use static_cast<> instead to fix this.

Fixes: #7820
Signed-off-by: Michał Górny <mgorny@gentoo.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20239>

r600/sfn: Fix test for possible dest slot

Fixes Coverity CID 1517721

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20367>

r600/sfn: don't use a return register when shader doesn't read it

This reduces register pressure.

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20367>

iris: Limit DG2 modifiers to DG2

These DG2 modifiers should be disabled on other devices.

Signed-off-by: Jianxun Zhang <jianxun.zhang@intel.com>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20363>

venus: rename sync_queue -> timeline

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: support vtest+multiple_timelines and enable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: add supports_multiple_timelines capset and enable

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: support external fencing with multiple timelines

Without globalFencing, exportable fences and semaphores must instead
have their proxy vn_renderer_sync installed in the same renderer
ring_idx as ther last queue submissions to ensure they signal after
all work previously submitted to the same ring_idx. Exportable
fences/semaphores with a temporary (imported) payload don't need a proxy
vn_renderer_sync, since they already have a `poll()`able fd available.

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: place wsi submissions on corresponding ring_idx

With implicit fencing, the image has a fence that blocks scanout until
rendering is complete. virtgpu doesn't support implicit fencing yet, but
Sommelier (a VM Wayland compositor) does the wait by exposing the bo as
a GEM handle and waiting on all fences in userspace with a
DRM_IOCTL_VIRTGPU_WAIT before issuing the wl_surface commit.
During vkQueueSubmit involving wsi images, we follow with an empty
renderer submission on the corresonding ring_idx to install a fence
on the appropriate virtgpu fence context after the last rendering
submission.

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: remove workaround for empty renderer submissions

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: allow renderer submissions on per-context virtgpu rings

For submissions to renderers that support multiple timelines, put
them on the virtgpu fencing timeline (dma fence context) specified by
the VkQueue's bound ring_idx. CPU-sync'd renderer submissions
can be sent in the same manner by using ring_idx = 0.

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: bind VkQueue to available ring_idx at creation

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: use upstream DRM_IOCTL_VIRTGPU_CONTEXT_INIT

And remove SIMULATE_CONTEXT_INIT and PARAM_MAX_SYNC_QUEUE_COUNT now that
we expect guest kernel support for CONTEXT_INIT with standard support
for up to 64 rings.

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: update venus-protocol for multiple timelines

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

venus: formatting fixes

Signed-off-by: Ryan Neph <ryanneph@google.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19691>

CI: Overwrite valve infra's registry

This is a temporary workaround wihle we wait for valve-infra to sort
out this issue.

It is safe to do because registry.freedesktop.org is a mirror of
harbor.freedesktop.org.

Related: https://gitlab.freedesktop.org/mesa/mesa/-/issues/7913
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20362>

util: Fixes thread safety of DEBUG_GET_ONCE_*_OPTION macros

Pick DEBUG_GET_ONCE_BOOL_OPTION as a example:
The intention of DEBUG_GET_ONCE_BOOL_OPTION are returned the same value across
thread, before this commit, on different thread call the function generated by
DEBUG_GET_ONCE_BOOL_OPTION may return different value if called setenv in the
middle of debug_get_bool_option, so use debug_get_option_cached along with
new exposed function debug_parse_bool_option to solve this issue

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19554>

util: Add function debug_get_option_cached os_get_option_cached

This is used to fixes DEBUG_GET_ONCE_*_OPTION macros latter

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19554>

util: Improve os_get_android_option to be thread-safe

struct hash_table is not thread-safety, need guard by mutex,
but with thread local storage, we can simplify the code and also
got the thread safety without the need of mutex.
Another advantage is by using thread local storage, os_get_android_option
will have the same actions like getenv does, that it's not cached the
value, each call will access the property_get, like getenv will be affected
by putenv

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19554>

util: Improve prototype of debug_get_num_option and debug_get_flags_option

Getting debug_get_num_option to return int64_t, as long under 64 bit Linux are 64 bit size,
 so using fixed int64_t for cross platform consistence, as long under win32 is 32 bit size.

Getting DEBUG_GET_ONCE_FLAGS_OPTION to return uint64_t to getting it to be
consistence with debug_get_flags_option.

DEBUG_GET_ONCE_NUM_OPTION is not accessed in codebase, so add unittest for it, it maybe
used in future, remove it is not consistence

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19554>

util: Indent u_debug.c with 3 space instead of tab

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19554>

glsl: fixes -Werror,-Wunused-but-set-variable for clang-15 in glcpp-parse.y and glsl_parser.yy

error messages:
src/compiler/glsl/glcpp/glcpp-parse.c:1691:9: error: variable 'glcpp_parser_nerrs' set but not used [-Werror,-Wunused-but-set-variable]
    int yynerrs = 0;
        ^

src/compiler/glsl/glsl_parser.cpp:2370:9: error: variable '_mesa_glsl_nerrs' set but not used [-Werror,-Wunused-but-set-variable]
    int yynerrs = 0;
        ^

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

glsl: Fixes indent issue after replace tab with 3 space by tools in glcpp-parse.y

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

glsl: replace tab with 3 space in glcpp-parse.y

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

glsl: Fixes ident issue in glsl_parser.yy and update editorconfig for it

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Acked-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

panvk: Fixes -Werror,-Wunused-but-set-variable for clang-15 in panvk_descriptor_set.c

../../src/panfrost/vulkan/panvk_descriptor_set.c:67:13: error: variable 'dynoffset_idx' set but not used [-Werror,-Wunused-but-set-variable]
   unsigned dynoffset_idx = 0, img_idx = 0;
            ^

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

loader: fixes -Werror,-Wdeprecated-non-prototype for clang-15 in pipe-loader

There are multiple error messages, show one of them:
../../src/gallium/auxiliary/pipe-loader/pipe_loader_sw.c:219:54: error: passing arguments to a function without a prototype is deprecated in
all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
         sdev->ws = sdev->dd->winsys[i].create_winsys(drisw_lf);
                                                     ^

Signed-off-by: Yonggang Luo <luoyonggang@gmail.com>
Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19875>

d3d12/ci: Update expectations

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Swizzle SRV access to stencil

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Pass input clip size to shader compilation

For cases where both clip and cull are used, and a shader has both
inputs and outputs that can contain them, we need metadata to tell
us where the clip array ends and the cull array begins, since they
get combined into CLIP location registers. For outputs, this is in
the nir info, but for input we pass it in a sideband channel.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Don't add UAV flag to DSV textures

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Use ResolveSubresourceRegion to support non-averaging resolve modes

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Use driver-implemented triangle fans if available

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

dzn: Get options15

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

microsoft/compiler: Handle cull distance starting fractional with no previous clip

This can happen if the clip distance was declared, but was discarded as an
unused variable.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

microsoft/compiler: Sort all user varyings before any sysvals

User varyings are linked by both name and register. The name is based
on how many *variables* are before it in final driver_location sort
order, not necessarily how many registers are before it.

In some cases where clip/cull distance are involved, it's possible
for one shader to write into a part of the cull distance that's
ignored by a downstream shader, but because linking is done by
*whole* register locations, and clip/cull can be combined using
*fractional* register locations, this is hard to detect. Since no
non-sysvals end up using fractional locations, just put all non-sysvals
first so they always generate the same semantic names for the same
register locations.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

microsoft/compiler: Make nir_var_to_dxil_sysvalue_type static

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

microsoft/compiler: Handle both input and output clip/cull distances

For clip/cull coming into a GS and being written, this pass was wrong
and would modify variable types incorrectly. Track both inputs and
outputs separately.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20346>

panfrost: Don't use texture format swizzles on v7

They're too restricted for AFBC. Fix up instead. There are two problems at play:

1. We can't just map the format swizzle to the pixel format ordering on v7,
   because the "reordered" values aren't allowed with compression.
2. We can't just compose the format swizzle with the API swizzle, because the
   composed swizzle is applied to the border colour, so we need to be able to
   apply an inverted swizzle to the border colour. That only works for bijective
   format swizzles.

Fortunately, there's a neat solution: decompose the format's swizzle into two
swizzles, the first mapping to a reordering that IS allowed for compression, and
the second a bijection. Then we use the allowed reordering when texturing, apply
the bijective swizzle to the API swizzle, and apply the inverse of the bijective
swizzle to the border colour. When we're sampling a border colour, what's now
happening mathematically is:

   (API swizzle o bijective swizzle)((bijective swizzle^-1)(border colour)) =
   (API swizzle o (bijective swizzle o bijective swizzle^-1))(border colour) =
   API swizzle(border colour)

which is exactly what we wanted.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20311>

panfrost: Allow swizzled AFBC on v9+

On v6 and earlier, the hardware supports arbitrary format swizzles for AFBC, so
there's no restriction on AFBC. On v8 and newer, the format swizzle gets applied
to the *decompressed* interchange format, so we can effectively support BGRA of
AFBC images without any special handling. (Confirmed working on v9. Obviously I
can't test on v8 but the expression is cleaner if we assume optimistically it's
like v9. Without hardware, we get to make that assumption :-p)

That just leaves v7 as the only architecture where format swizzles are
restricted for compression but there are no plane descriptor. Don't apply the
restriction to the newer parts.

This gets us AFBC of window surfaces on v9+. As the limiting case, fullscreen
glmark2-es2-wayland -btexture (1080p) in sway on Mali-G57 from 1300fps to
2353fps.

45% reduction in frame time is nothing to sneeze at.

Achoo.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20311>

panfrost: Introduce pan_afbc_mode

Introduce an enum to represent an AFBC compression mode. These modes are not
formats, on Valhall they are decoupled from the format. As such, it does not
make sense to use a pipe_format to represent them. Add an enum that we can use
in a straightforward way on Midgard and Bifrost to fallback for texture views,
and can map 1:1 to the Valhall hardware enum.

In addition to being less overloaded semantically, this lets -Wswitch kick in to
ensure that we handle all enums when translating. The straightforward
translation raises the following warnings:

../src/panfrost/lib/pan_cs.c:437:9: warning: enumeration value ‘PAN_AFBC_MODE_R5G5B5A1’ not handled in switch [-Wswitch]
  437 |         switch (panfrost_afbc_format(PAN_ARCH, format)) {
      |         ^~~~~~

...indicating that some formats were missed, leading to assertion fails "unknown
canonical AFBC format" when rendering RGB5A1, which dEQP-GLES31 does. Fixes
regressions in
dEQP-GLES31.functional.draw_buffers_indexed.random.max_required_draw_buffers.*
on Valhall.

Given how scarce v9 hardware is, that v10 isn't upstream yet, and the offending
code was merged a week ago, this should not have actually affected anyone. At
any rate, it's a good reminder we really do need CI for v9...

Fixes: 8e125b6c15b ("panfrost: Enable AFBC of more formats")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20311>

panfrost: Luminance-alpha AFBC unsupported on v7+

The L8_UNORM, A8_UNORM, and L8A8_UNORM v7 formats do not support AFBC,
regardless of swizzling. We're about to lift the restrictions on swizzling with
AFBC on v7, so we'll need to handle these cases explicitly to avoid using AFBC
in these cases.

Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20311>

panfrost: Align WSI strides for tiled AFBC

When calculating legacy WSI strides for tiled AFBC, we need to account for the
greater alignment requirement of tiled AFBC, or importing resources will fail
later.

Since tiled AFBC is only supported on v7 and later, and AFBC of window surfaces
isn't being used on Linux on v7 and later, this probably hasn't been hit in
practice. Probably.

We're about to fix AFBC of window surfaces so we need to fix this side first.

Fixes: 0255f554f39 ("panfrost: Advertise 16x16 tiled AFBC")
Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20311>