cdedd04640
panvk: Fix ls_tracker usage in cs_loop
...
LOAD_MULTIPLE can be emitted inside of a loop body. We need to WAIT if
that loads targets a register for which a load was not already in-flight
at the start of the loop body.
Technically we only have to emit the wait if the dst reg of a new load
is actually used inside the loop, but that would require separate
tracking of source regs used in the loop and is probably not worth
the effort for now.
Fixes: f75569734eboris.brezillon@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35436 >
2025-06-11 15:19:50 +00:00
a24abc3fbd
panvk: Move cs_wait_slot(s) and cs_flush_loads/stores
...
The following commit needs to flush loads from cs_loop_ functions.
Fixes: f75569734eboris.brezillon@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35436 >
2025-06-11 15:19:50 +00:00
a36e51af22
zink: remove fixup_driver_props
...
No longer needed after the codegen improvements
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:52 +00:00
4d1768aec5
zink/codegen: support double-loading dynamic properties arrays
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:52 +00:00
4eb50771e4
zink/codegen: add indentation to the mako templates
...
This regains us some much needed sanity
And it actually uncovered a bug in zink_instance.py:L306 (misplaced
%endif)
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:52 +00:00
a5e114ce97
zink: apply EXT-suffixed functions to their unsuffixed counterpart
...
...if they are promoted to core verbatim.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:51 +00:00
66f0a99c73
zink: apply core functions to extension functions
...
...if those core functions were promoted verbatim from the extension.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:51 +00:00
ef1ae1f24f
zink: make zink_device_info look for KHR versions of promoted exts
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:51 +00:00
bbfeb6cd87
zink: fix win32 detection in zink_device_info
...
str.find(substr) returns -1 if the substring's not in the string, and
bool(-1) returns True in Python, so the _WIN32 ifdef guard was actually
inserted everywhere.
Also, the win32 functions actually have "Win32" in their name, like
vkGetMemoryWin32HandleKHR
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35389 >
2025-06-11 14:49:50 +00:00
4dafd2e787
radeonsi: add a workaround for a DrawTransformFeedback issue on gfx11.5
...
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35344 >
2025-06-11 14:29:06 +00:00
d991810833
radeonsi/ci: update gfx8 failures
...
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35344 >
2025-06-11 14:29:06 +00:00
56ed949e08
radeonsi: don't clamp the shadow comparison value for nir_texop_lod on gfx8-9
...
It doesn't have the value. This fixes a crash on gfx8-9.
Cc: mesa-stable
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35344 >
2025-06-11 14:29:06 +00:00
595889018a
anv: implement VK_KHR_maintenance9
...
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com >
Reviewed-by: José Roberto de Souza <jose.souza@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35381 >
2025-06-11 14:09:56 +00:00
884df891d7
anv: allow device creation with no queue
...
Reviewed-by: José Roberto de Souza <jose.souza@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35381 >
2025-06-11 14:09:56 +00:00
1d8382b88e
brw: enable more lowering for bitfield manipulation at non 32bit sizes
...
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com >
Reviewed-by: José Roberto de Souza <jose.souza@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35381 >
2025-06-11 14:09:56 +00:00
d983280da9
vc4: free RA interference graph on failure
...
This fixes a leak in the driver.
Backport-to: 25.1
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com >
Reviewed-by: Jose Maria Casanova Crespo <jmcasanova@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35464 >
2025-06-11 13:29:35 +00:00
bacc0451bc
svga/ci: vmware farm is back; restore piglit job to automatic
...
Signed-off-by: Martin Krastev <martin.krastev@broadcom.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35459 >
2025-06-11 13:11:44 +00:00
7280e3b2a1
radeonsi/tests: update expected results
...
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:28 +00:00
3bcbd11a33
aco/isel: fix visit_tex handling of is_sparse
...
For cases when less than 4 components are read, the original code
would compute an incorrect dmask. eg: with a single component + is_sparse,
the dmask was 0x13:
- 0x 3 = coming from nir_def_components_read
- 0x10 = the sparse bit
While it should have at 2 bits set (1 for the color/depth, 1 for tfe).
This caused problem when expand_vector() used the dmask to generate
the final results, because the value for the sparse component was
read from the wrong index.
So after the call to emit_mimg() dmask needs to be adjusted
because the components will be stored in order, so if mask is 0x11
the tfe value would be stored at invalid index=5 (while it should
be at index=1).
This fixes KHR-GL46.sparse_texture_clamp_tests.SparseTextureClampLookupResidency_texture_2d_depth_component16
and KHR-GL46.sparse_texture2_tests.SparseTexture2Lookup_texture_2d_depth_component16
with ACO.
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:28 +00:00
4a84ebfcb1
ac/llvm: rework component trimming in visit_tex
...
The referenced commit was a step in the right direction, but not
complete.
ac_build_image_opcode returns a vec<4> or a struct<vec<4>, int>
so we can simplify visit_tex. We just need to map these 4/5 values
to the expected layout from NIR.
eg: depth + TFE would produces "<d, x, x, x>, t" so it has to be
transformed into <d, t>.
nir_texop_fragment_mask_fetch_amd + sparse doesn't exist, so it's
another opportunity for simplification.
This is required to get KHR-GL46.sparse_texture2_tests.SparseTexture2Lookup_texture_2d_depth_component16
working properly.
The same test fails with ACO so it probably needs a change in the
same area.
Fixes: c0ef2aa7f8marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:28 +00:00
1cc52dff05
radeonsi: allow sparse depth textures
...
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:28 +00:00
508bf2a490
radeonsi/tests: fix script handling of missing results
...
When running radeonsi-run-tests.py with the -t option, it's possible that
some suite won't generate any result so don't fail in this case.
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:28 +00:00
b153188f25
amd/ci: remove references to tests that don't exist anymore
...
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:27 +00:00
3d8fe9b36d
radeonsi: skip draws using invalid setup
...
Draws running with the secure bit set needs to have their destination
surfaces created with tmz to get the correct result.
Instead of asserting this fact, let just skip the invalid draws
completely.
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:27 +00:00
0e9ba3031e
radeonsi: allow msaa sparse textures on gfx10+
...
The hardware doesn't support the prt layouts, but we can use normal
layouts and ac_surface_addr_from_coord to determince which pages
need to be committed.
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:27 +00:00
ee6d3a593e
winsys/radeon: add surface_offset_from_coord
...
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35206 >
2025-06-11 12:11:27 +00:00
d6614099a0
vc4,v3d: update expected results
...
Add new failures/flakes.
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35462 >
2025-06-11 11:55:54 +00:00
1dfe97cc0b
llvmpipe: Make sure to not divide by zero on debug output
...
Check that lp->nr_fs_variants is not zero before dividing by it as done in a few
lines above.
CID: 1417662 Division or modulo by zero
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:34 +00:00
d4d37ead82
llvmpipe: Make sure to not divide by zero on debug output
...
Check that lp->nr_cs_variants is not zero before dividing by it as done in a few
lines above.
CID: 1531786 Division or modulo by zero
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:34 +00:00
383684596c
llvmpipe: Check for negative size before mapping the memory
...
CID: 1596479 Argument cannot be negative
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:34 +00:00
94f2d53586
llvmpipe: Close mem_fd or dmabuf_fd on error
...
The variable shadowing made the error path to leak these file descriptors.
CID: 1596485, 1596493 Logically dead code
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:34 +00:00
fb72e84c7a
llvmpipe: Do bitwise operation on unsigned constants
...
This fixes coverity issues while all compilers would do the right thing here.
CID: 1605222 Overflowed constant
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:34 +00:00
c9243262f0
llvmpipe: Check allocated memory instead of argument
...
This is likely a copy-paste error.
CID: 1616035 Dereference after null check
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:33 +00:00
09938eee85
llvmpipe: Asserts that the format at least has one non-void channel
...
As we are using its index as array index, avoid the use of the -1 index there.
CID: 1517247, 1517239 Negative array index read
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:33 +00:00
d2aade9184
llvmpipe: Avoid function name leak on fail paths
...
Free the memory on the fail path, also check early on allocation failure.
CID: 1644471 Resource leak
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35091 >
2025-06-11 11:03:33 +00:00
6e97df1d76
u_trace: Fix payload refcounting in u_trace_clone_append().
...
If the clone_append was to a chunk of the same u_trace that gets
process_chunk()ed after where we're cloning from, then the payloads would
have been unreffed in the previous chunk's cleanup_chunk().
Fixes use-after-frees with turnip gmem rendering that resulted in
corrupted payloads.
Fixes: 14e45cb21ehttps://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35379 >
2025-06-11 10:02:30 +00:00
bc2edf14d8
ac/nir: run nir_lower_vars_to_ssa after nir_lower_task_shader
...
nir_lower_task_shader does nir_lower_returns, so we need this if the
launch_mesh_workgroups was in control flow.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com >
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13326
Backport-to: 25.1
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35411 >
2025-06-11 09:01:39 +00:00
3b326abf7b
radv: add capture/replay for sparse buffers and descriptor buffer
...
Shouldn't be super useful in practice because the normal capture/replay
BDA path should also work.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35366 >
2025-06-11 07:31:29 +00:00
643e1c4395
radv: cleanup creating sparse buffers with capture/replay
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35366 >
2025-06-11 07:31:29 +00:00
74acae0ed8
radv: stop setting the address for capture/replay and non-sparse buffers
...
This doesn't do anything because for non-sparse buffers, a device
memory object must be bound to the buffer.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35366 >
2025-06-11 07:31:28 +00:00
ee200cc0d1
radv: stop using vk_common entrypoints when not necessary
...
For less indirections.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35359 >
2025-06-11 07:10:02 +00:00
7d2f20b2fb
radv: remove useless vk_common_entrypoints.h includes
...
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35359 >
2025-06-11 07:10:02 +00:00
f3578973d7
radv/meta: fix using the wrong pipeline layout for ASTC decoding
...
Cc: mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35359 >
2025-06-11 07:10:01 +00:00
2a58d7b333
zink: verify that mutable dmabufs are created with a format list
...
this is otherwise illegal
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35433 >
2025-06-10 23:12:22 +00:00
0d3dc92863
zink: emulated alpha formats do not require mutable
...
cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35433 >
2025-06-10 23:12:22 +00:00
760c96628d
ir3/legalize: don't allow end instruction as jump target on a5xx
...
From experiments it appears that (jp) flag on "end" instruction has no
effect (at least on a5xx). This means that we should insert "(jp)nop"
before it, which is what blob is doing too.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35331 >
2025-06-10 22:51:21 +00:00
e1d54be524
ci/lava: Avoid eval when generating env script
...
Remove use of `eval` when writing `dut-job-env-vars.sh`, as it's
unnecessary. The script only needs to declare variables, not evaluate
them.
Using `eval` introduces parsing issues when variables contain both
single and double quotes, such as in commit titles. Example:
https://gitlab.freedesktop.org/mesa/mesa/-/jobs/77995175#L3188
This job failed to parse `CI_COMMIT_TITLE` and `CI_MERGE_REQUEST_TITLE`
correctly due to mixed quoting in:
Revert "ci: disable Collabora's farm due to maintenance"
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35421 >
2025-06-10 18:19:21 -03:00
655cf2f553
ci: Simplify filter_env_vars using indirect expansion
...
Avoid parsing variable values manually by using `${!varname@A}`, which
returns the exact declaration as stored in the bash session.
Signed-off-by: Guilherme Gallo <guilherme.gallo@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35421 >
2025-06-10 18:19:21 -03:00
d77166edec
tu: Use a temporary descriptor to patch input attachments
...
The final descriptor is in write-combined memory, so avoid reading from
it. Instead modify a temporary descriptor on the stack.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35420 >
2025-06-10 20:25:54 +00:00
83212054e0
tu: Use cached coherent memory when possible for query pools
...
It's expected to read back the result on the CPU, so make this cached
when cached coherent memory is available.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35443 >
2025-06-10 20:12:11 +00:00
Christoph Pillmayer