Commit graph

71220 commits

Author SHA1 Message Date
Mike Blumenkrantz
308744e789 zink: mark dirty_gfx_stages using util function
more extensible for mesh

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37358>
2025-09-14 07:56:05 -04:00
Mike Blumenkrantz
dce29ae35f Revert "util/vbuf: stop nooping set_vertex_buffers calls"
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
This reverts commit cccca8d140.

did not mean for this to merge

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37353>
2025-09-13 11:18:09 -04:00
Mike Blumenkrantz
cccca8d140 util/vbuf: stop nooping set_vertex_buffers calls
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
this sucks, but if the gallium expectation is that changing vertex state
always requires a vb update, vbuf can no longer deduplicate repeated
calls

Fixes: 1638d486ff ("gallium/u_threaded,st/mesa: add a merged set_vertex_elements_and_buffers call")

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:31 +00:00
Mike Blumenkrantz
d63374498d zink: eliminate even more calls to sync functions
many times (especially in viewperf), vertex buffers and draw buffers
have no prior access, which means incurring a heavy call to the barrier
function which will just return a no-op anyway

instead, have an inline helper to no-op these cases and just add the
expected access

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:30 +00:00
Mike Blumenkrantz
58ed06ad68 zink: only remove buffer deferred sync on release
this is otherwise just a huge amount of overhead whenever
descriptors are unbound

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:29 +00:00
Mike Blumenkrantz
1abc337e94 zink: delete some function decls that no longer exist
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:28 +00:00
Mike Blumenkrantz
5ef8453228 zink: use memcpy for vbo bind
there's no more refcounting here so this is fine

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:28 +00:00
Mike Blumenkrantz
aee0379414 zink: rescope some zink_set_vertex_buffers_internal variables
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:27 +00:00
Mike Blumenkrantz
f8c96df9d2 zink: move vbo unbind to bind_vertex_state
set_vertex_buffers is required to be called after the bind_state call,
which means some code can be moved out of the higher frequency call site

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:27 +00:00
Mike Blumenkrantz
06a8e7b87e zink: move zink_bind_vertex_elements_state() to zink_context.c
needed in order to handle unbinding

no functional changes

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:26 +00:00
Mike Blumenkrantz
f99503faa2 zink: use velems buffer count in blitter instead of gfx mask
small simplification

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:26 +00:00
Mike Blumenkrantz
a730c83dd0 zink: split update_res_bind_count
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:26 +00:00
Mike Blumenkrantz
c96bb8d1c4 zink: ALWAYS_INLINE zink_set_vertex_buffers_internal
otherwise the specialization is useless

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:25 +00:00
Mike Blumenkrantz
26f4cc39b5 zink: add a fastpath for nooping vertex and draw buffer barriers
even doing this mechanism inside the barrier function doesn't yield
anywhere near the same gains (~15-20% in viewperf catia), mostly for
the draw buffers, so I'd assume it's just added register pressure
from the huge draw function

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37277>
2025-09-13 12:20:25 +00:00
Mike Blumenkrantz
efd8e46891 zink: support NV_representative_fragment_test
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37081>
2025-09-13 11:28:44 +00:00
Mike Blumenkrantz
34641b44c6 mesa: support GL_NV_representative_fragment test
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37081>
2025-09-13 11:28:44 +00:00
Robert Mader
5cc0d1b681 kms-dri-sw: Implement create_mapped()
Following the example of dri_sw_winsys, in order to support
importing client-allocated (u)dmabufs into GL renderers as well
as KMS.
This crucially allows Wayland compositors running on vkms + llvmpipe
to behave much more like on HW GPUs/display engines, making CI testing
more robust and intuitive.

Signed-off-by: Robert Mader <robert.mader@collabora.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37305>
2025-09-13 06:47:58 +00:00
Robert Mader
306feaf3a4 sw_winsys: Add winsys_handle to displaytarget_create_mapped
The main intention is to allow to pass dmabuf handles, however as its
type differs depending on platform, just pass the whole whandle.

Also make use of it in llvmpipe_resource_from_handle(), in preparation
for the next commit.

Signed-off-by: Robert Mader <robert.mader@collabora.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37305>
2025-09-13 06:47:58 +00:00
Corentin Noël
63cd410c29 virgl: Skip resource destruction only when there are actually needed references
Suggested by Jan Alexander Steffens in https://gitlab.freedesktop.org/mesa/mesa/-/issues/13844#note_3087461

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/13844
Fixes: 6c1c116a0f
cc: mesa-stable

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37346>
2025-09-13 00:44:49 +00:00
Mike Blumenkrantz
c02d5d79b7 zink: add another tu flake
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Mike Blumenkrantz
69f90bfe29 zink: add some pre-checks before calling query update/suspend/resume
these trigger lots of context switches which impact perf

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Mike Blumenkrantz
9d19d7f6de zink: move draw state flag resets into their blocks
incredibly, this shows up in perf

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Mike Blumenkrantz
74b7b5b661 zink: simplify index type access to normal array
the sizes are 1/2/4, so just use that array index

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Mike Blumenkrantz
c852c25cdc zink: always use vkCmdBindVertexBuffers2
this cuts out a small amount of overhead in drivers that wrap
vkCmdBindVertexBuffers onto vkCmdBindVertexBuffers2

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Mike Blumenkrantz
07a7d3627f zink: store last index buffer
this allows skipping some draw-time ops when the same index buffer
is used repeatedly

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37318>
2025-09-12 22:09:50 +00:00
Martin Roukala (né Peres)
03239662c3 zink/ci: raise the job timeout from 5 to 8 minutes
Suggested-by: Daniel Stone <daniels@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37328>
2025-09-12 21:46:55 +00:00
Georg Lehmann
2ac5641473 shader_info: remove gl_subgroup_size enum
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37258>
2025-09-12 21:05:17 +00:00
Georg Lehmann
8d7b1498cc rusticl: switch to new subgroup size info
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37258>
2025-09-12 21:05:17 +00:00
Georg Lehmann
1b5c78d1f7 radeonsi: switch to new subgroup size info
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37258>
2025-09-12 21:05:17 +00:00
Georg Lehmann
04d3b3bde5 mesa,glsl,spirv: set new subgroup size info
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37258>
2025-09-12 21:05:17 +00:00
Georg Lehmann
95c2a65662 nir: remove unused shader_info param in nir_create_shader
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37258>
2025-09-12 21:05:17 +00:00
Ian Romanick
1eb4a2f5cd iris: Limit max_shader_buffer_size to INT32_MAX
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
GLSL arrays are sized by signed integers. Other places in the Mesa treat
this value is int32_t. Having the value larger than (uint32_t)INT32_MAX
causes assertion failures in many piglit tests.

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Closes: #13873
Fixes: 2f8b8649f0 ("iris: Increase max_shader_buffer_size to max_buffer_size")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37282>
2025-09-12 20:45:39 +00:00
Mike Blumenkrantz
2d7dd41506 tc: eliminate refcounting for set_shader_buffers
this is no longer necessary

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37310>
2025-09-12 15:47:34 +00:00
Mike Blumenkrantz
9f764486cb cso: unbind vertex buffers when unbinding context
cc: mesa-stable

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37276>
2025-09-12 15:06:15 +00:00
Christian Meissl
31a627bdc2 panfrost: take reference from pool used for allocation
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Fixes bo leaks caused by wrong ref counting when
using temporary sampler views.
The leak was observed when playing videos using the
DRM_FORMAT_MOD_VENDOR_MTK tiling format
through gstreamer.

Fixes: 0795f3d7e4 ("panfrost: Add a pool to sampler_view")
Signed-off-by: Christian Meissl <meissl.christian@gmail.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37325>
2025-09-12 11:44:46 +00:00
Rob Clark
52e02c2b0b freedreno/a6xx: Fix snorm rounding
See util_format_r8g8b8a8_snorm_pack_rgba_float()

Fixes the remaining failures in images/clFillImage/test_cl_fill_images.

Signed-off-by: Rob Clark <rob.clark@oss.qualcomm.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37319>
2025-09-12 10:10:29 +00:00
Mike Blumenkrantz
e11be7549d zink: more vvl exceptions
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37324>
2025-09-12 04:05:28 +00:00
Trigger Huang
430b85adbb radeonsi: Fix u_log_ctx for aux_context recreation
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
When recreating aux_context (typically following a GPU reset event),
the u_log_context must be properly initialized based on the aux_debug
enablement state, just like the logic in the function
radeonsi_screen_create_impl() during aux_context creation. Otherwise,
wrong u_log_context setting will lead to Mesa crashes. For example, in
si_blit_decompress_color(), it will check sctx->log to decide whether
to call u_log_printf(), which ultimately causes a crash.

Signed-off-by: Trigger Huang <Trigger.Huang@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37296>
2025-09-12 03:45:35 +00:00
Karol Herbst
be7afedba9 rusticl/mesa: fix NULL pointer access in set_constant_buffer_stream
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Fixes: b3133e250e ("gallium: add pipe_context::resource_release to eliminate buffer refcounting")
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37311>
2025-09-12 02:01:59 +00:00
Mike Blumenkrantz
f419a96d76 zink: use implicit offsets for function temp variables in ntv
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
this is also apparently illegal

Fixes: 0b7611824a ("zink: use implicit stride in ntv for temp vars"
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37312>
2025-09-12 00:52:46 +00:00
Gert Wollny
b26c68e40a r600/sfn: Filter lowering of b2f32(comp(x,y)) for 64 bit sources
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Instead of skipping later, we can filter these instructions right away.

Fixes: 51d8ca2dff ("r600/sfn: optimize comparison results")

Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37298>
2025-09-11 23:00:21 +00:00
Mike Blumenkrantz
32b4412d54 zink: update gfx pipeline less frequently
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37308>
2025-09-11 18:21:40 +00:00
Mike Blumenkrantz
b497a81dd1 zink: make find_completed_batch_state() only return state for COPY_ONLY ctx
COPY_ONLY has no descriptor support, so these states cannot be recycled
like the others

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37306>
2025-09-11 17:54:22 +00:00
Mike Blumenkrantz
13ea671b68 zink: simplify state iterating in find_completed_batch_state()
this is a bit easier to read

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37306>
2025-09-11 17:54:22 +00:00
Mike Blumenkrantz
a6918f1bed zink: don't init batch descriptors for copy contexts
these are never used and waste memory

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37306>
2025-09-11 17:54:22 +00:00
Eric Engestrom
2fe2ab2bd0 zink+nvk/ci: update test expectations
Some checks are pending
macOS-CI / macOS-CI (dri) (push) Waiting to run
macOS-CI / macOS-CI (xlib) (push) Waiting to run
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37302>
2025-09-11 16:02:38 +00:00
Eric Engestrom
d195e64278 zink+turnip/ci: update test expectations
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37302>
2025-09-11 16:02:38 +00:00
Eric Engestrom
4db3851b5a zink+lvp/ci: document recent flakes
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37302>
2025-09-11 16:02:38 +00:00
Eric Engestrom
ce39a7e2a8 zink+radv/ci: document recent flakes
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37302>
2025-09-11 16:02:38 +00:00
Eric Engestrom
bcb66ddd28 zink+radv/ci: sort vangogh flakes
removes noise out of the next commit

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/37302>
2025-09-11 16:02:38 +00:00