The IPC RX hardirq handler matches consumers under a spinlock and
allocates rx_msg buffers. On PREEMPT_RT these spinlocks become sleeping
locks and the allocation may sleep, neither of which is allowed in true
hardirq context, resulting in "sleeping function called from invalid
context" splats.
IRQF_ONESHOT makes genirq keep the primary handler in hardirq even when
forced threading is enabled, so on PREEMPT_RT the handler cannot be
threaded. Drop the flag so the primary handler is threaded on PREEMPT_RT
and the IPC RX path runs in a context where sleeping is allowed. On the
MSI interrupt chip (IRQCHIP_ONESHOT_SAFE) the flag was stripped anyway,
so non-RT behaviour is unchanged.
Fixes: 85c9cc2d25 ("accel/ivpu: Use threaded IRQ for IPC callback processing")
Cc: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Cc: Karol Wachowski <karol.wachowski@linux.intel.com>
Cc: dri-devel@lists.freedesktop.org
Reviewed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20260617092031.3016582-1-karol.wachowski@linux.intel.com
The .driver_data member of the various struct pci_device_id arrays were
initialized by list expressions. This isn't easily readable if you're
not into PCI. Using the PCI_DEVICE macro and named initializers is more
explicit and thus easier to parse. Also skip explicit assignments of 0
(which the compiler then takes care of).
This change doesn't introduce changes to the compiled pci_device_id
arrays. Tested on x86 and arm64.
v2:
- gma500: fix coding style in cast
Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@baylibre.com>
Reviewed-by: Patrik Jakobsson <patrik.r.jakobsson@gmail.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patch.msgid.link/20260504150537.2136760-2-u.kleine-koenig@baylibre.com
The current DSI mode configuration disables horizontal front porch
(HFP) and back porch (HBP) transmission using
MIPI_DSI_MODE_VIDEO_NO_HFP and MIPI_DSI_MODE_VIDEO_NO_HBP.
However, the SN65DSI83/84 bridge relies on receiving full horizontal
timing information over DSI in order to correctly reconstruct the
LVDS output timings. When HFP and HBP are not transmitted, the bridge
cannot recreate the required timing parameters, resulting in unstable
or missing display output on some panels.
The SN65DSI84 datasheet also describes transmission of complete
horizontal timing information as part of the DSI video stream.
Remove MIPI_DSI_MODE_VIDEO_NO_HFP and MIPI_DSI_MODE_VIDEO_NO_HBP
so that HFP and HBP timing information is transmitted to the bridge.
Reference:
SN65DSI84 Datasheet, Section 7.4.7 (page 20):
https://www.ti.com/lit/ds/symlink/sn65dsi84.pdf
Tested-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Tested-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Sudarshan Shetty <tessolveupstream@gmail.com>
Reviewed-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Link: https://patch.msgid.link/20260527085736.920390-2-tessolveupstream@gmail.com
[Luca: cleaned up datasheet URL]
Signed-off-by: Luca Ceresoli <luca.ceresoli@bootlin.com>
Currently we have not explicitly defined the behaviour when the width
and height of the writeback framebuffer do not match the width and
height of the attached CRTC.
Allow the driver to perform a scale operation to fit the content, so
systems with scaling blocks in the writeback path can easily expose them.
A partially filled buffer should never be returned, so failure remains
an option for all the systems that don't have a scaler, or when scaler
limits are exceeded.
Cropping would make less sense, as the region to crop is underspecified
without some other properties to define an origin.
Signed-off-by: Derek Foreman <derek.foreman@collabora.com>
Link: https://patch.msgid.link/20260602191754.79047-1-derek.foreman@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Daniel Stone <daniels@collabora.com>
Dispatching IPC callbacks from system_percpu_wq adds scheduling latency
that is neither bounded nor predictable, which hurts job completion
turnaround. Handle them from a threaded IRQ instead: the hard-IRQ
handler drains the IPC FIFO and wakes the thread, which runs the
callback consumers such as job-done processing.
Job resource teardown can trigger IOMMU unmapping and context teardown,
which is too slow to run from the IRQ thread. Defer it to a dedicated
WQ_UNBOUND | WQ_MEM_RECLAIM workqueue via a per-device lockless list.
UNBOUND keeps the long-running cleanup off the percpu workers and
MEM_RECLAIM guarantees forward progress because the work frees buffer
objects. The runtime PM reference taken at submission is released only
after cleanup completes, otherwise runtime suspend could race the
pending work and deadlock.
Because cleanup is now asynchronous, userspace that rapidly recycles
file descriptors or command queues can momentarily observe stale
per-context resources and fail with -EMFILE or -EBUSY. Flush the
cleanup work once and retry before giving up.
Reviewed-by: Andrzej Kacprowski <andrzej.kacprowski@linux.intel.com>
Signed-off-by: Karol Wachowski <karol.wachowski@linux.intel.com>
Link: https://patch.msgid.link/20260611055201.948726-1-karol.wachowski@linux.intel.com
Triggering kswap at an order higher than beneficial_order makes little
sense, as the driver has already indicated the optimal order at which
reclaim is effective. Similarly, issuing direct reclaim or triggering
kswap at a lower order than beneficial_order is ineffective, since the
driver does not benefit from reclaiming lower-order pages.
As a result, direct reclaim should only be issued with __GFP_NORETRY at
exactly beneficial_order, or as a fallback, direct reclaim without
__GFP_NORETRY at order 0 when failure is not an option.
Signed-off-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Christian Koenig <christian.koenig@amd.com>
Reviewed-by: Andi Shyti <andi.shyti@linux.intel.com>
Link: https://patch.msgid.link/20260612225757.3844350-1-matthew.brost@intel.com
Commit
28c5bf2876 ("drm/sched: Disallow initializing entities with no schedulers")
failed to notice clearing of entity->rq in drm_sched_entity_init() is now
redundant and can be removed.
Given that entity->rq can now never be NULL, we also remove two impossible
checks, from drm_sched_entity_kill() and drm_sched_entity_flush()
respectively.
Similarly, we can also remove the !entity->rq check in
drm_sched_job_init(). And for the better, given that the error message, if
it ever triggered, would have dereferenced the yet un-initialized job->
sched (only initialized later in drm_sched_job_arm()). This appears to
have been theoretically broken ever since commit
56e449603f ("drm/sched: Convert the GPU scheduler to variable number of run-queues")
.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Cc: Christian König <christian.koenig@amd.com>
Cc: Danilo Krummrich <dakr@kernel.org>
Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Philipp Stanner <phasta@kernel.org>
Signed-off-by: Philipp Stanner <phasta@kernel.org>
Link: https://patch.msgid.link/20260602153339.43453-1-tvrtko.ursulin@igalia.com
The colorspace property's documentation states that BT2020_RGB and
BT2020_YCC are equivalent, and the output format depends on the driver.
Now that there is a "color format" property that userspace can use to
explicitly set a format, update the colorspace docs to mention this.
The behaviour here is not changed for userspace that doesn't know about
the color format property yet, as the color format property defaults to
"AUTO", where the choice of output format is left up to drivers.
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-24-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
The bridge chain format selection behaviour was, until now,
undocumented. With the addition of the "color format" DRM property, it's
not sufficiently complex enough that documentation is warranted,
especially for driver authors trying to do the right thing.
Add a high-level overview of how the process is supposed to work, and
mention what the display driver is supposed to do if it wants to make
use of this functionality.
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-23-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
With the "color format" property, the bridge chain format selection has
gained increased complexity. Instead of simply finding any sequence of
bus formats that works, the bridge chain format selection needs to pick
a sequence that results in the requested color format.
Add KUnit tests for this new logic. These take the form of some pleasant
preprocessor macros to make it less cumbersome to define test bridges
with a set of possible input and output formats.
The input and output formats are defined for bridges in the form of
tuples, where the first member defines the input format, and the second
member defines the output format that can be produced from this input
format. This means the tests can construct scenarios in which not all
inputs can be converted to all outputs.
Some tests are added to test interesting scenarios to exercise the bus
format selection in the presence of a specific color format request.
Furthermore, tests are added to verify that bridge chains that end in an
HDMI connector will always prefer RGB when the color format is
DRM_CONNECTOR_COLOR_FORMAT_AUTO, as is the behaviour in the HDMI state
helpers.
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-21-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
Switch between requested color formats by setting the right bus formats,
configuring the VO GRF registers, and setting the right output mode.
To do this, the encoder's atomic_check queries the bus format of the
first bridge, which was determined by the bridge chain recursive format
selection. Pick the input format if it's !FIXED, otherwise, pick the
output format.
The previously unused GRF register color format defines are redone as
well. Both RK3588 and RK3576 use the same defines; it didn't look like
this as there was a typo in the previously (unused) definition.
Reviewed-by: Cristian Ciocaltea <cristian.ciocaltea@collabora.com>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-16-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
drm_hdmi_connector_mode_valid assumes modes are only valid if they work
with RGB. The reality is more complex however: YCbCr 4:2:0
chroma-subsampled modes only require half the pixel clock that the same
mode would require in RGB.
This leads to drm_hdmi_connector_mode_valid rejecting perfectly valid
420-only or 420-also modes.
Fix this by checking whether the mode is 420-capable first. If so, then
proceed by checking it with DRM_OUTPUT_COLOR_FORMAT_YCBCR420 so long as
the connector has legalized 420, otherwise error out. If the mode is not
420-capable, check with RGB as was previously always the case.
Fixes: 47368ab437 ("drm/display: hdmi: add generic mode_valid helper")
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-9-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
The drm_bridge_funcs atomic_get_output_bus_fmts operation should be the
same for likely every HDMI connector bridge, unless such an HDMI
connector bridge has some special hardware restrictions that I cannot
envision yet.
To avoid code duplication and standardize on a set of media bus formats
that the HDMI output color formats translate to, add a common helper
function that implements this operation to the drm bridge helpers.
The function returns a list of output bus formats based on the HDMI
bridge's current output bits-per-component, and its bitmask of supported
color formats.
To guard against future expansion of DRM_OUTPUT_COLOR_FORMAT outgrowing
the hweight8 call, add a BUILD_BUG_ON statement where it's used that
checks for DRM_OUTPUT_COLOR_FORMAT_COUNT. The justification for not
using hweight32 in all cases is that not all ISAs have a popcount
instruction, and will benefit from a smaller/faster software
implementation that doesn't have to operate across all bits.
The justification for not defining an hweight_color depending on the
value of DRM_OUTPUT_COLOR_FORMAT_COUNT is that this count enum value is
only known at compile time, not at preprocessor time.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-7-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
The new DRM color format property allows userspace to request a specific
color format on a connector. In turn, this fills the connector state's
color_format member to switch color formats.
Make drm_bridges consider the color_format set in the connector state
during the atomic bridge check. Call into the connector function to get
the connector state's connector color format. For bridge connectors
including an HDMI bridge, this will make use of whatever the HDMI
implementation set as output formats, and AUTO will never be part of the
rejection logic.
Reject any output bus formats that do not correspond to the requested
color format. DRM_CONNECTOR_COLOR_FORMAT_AUTO is always accepted as a
matching color format for a bus format, meaning that non-HDMI bridge
chains will end up picking the first bus format choice that works, as
has already been the case previously.
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@oss.qualcomm.com>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-6-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
Add a new general DRM property named "color format" which can be used by
userspace to request the display driver to output a particular color
format.
Possible string values for the new enum property are:
- "AUTO" (setup by default, driver internally picks the color format)
- "RGB"
- "YUV 4:4:4"
- "YUV 4:2:2"
- "YUV 4:2:0"
Drivers should advertise from this list the formats they support in an
optimistic best-case scenario. EDID data from the sink can then be used
in the kernel's atomic check phase to restrict this set of formats, as
well as by userspace to make a correct choice in the first place.
Co-developed-by: Werner Sembach <wse@tuxedocomputers.com>
Signed-off-by: Werner Sembach <wse@tuxedocomputers.com>
Co-developed-by: Andri Yngvason <andri@yngvason.is>
Signed-off-by: Andri Yngvason <andri@yngvason.is>
Signed-off-by: Marius Vlad <marius.vlad@collabora.com>
Reviewed-by: Maxime Ripard <mripard@kernel.org>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-3-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
Remove unnecessary SIGNAL_TYPE_HDMI_TYPE_A check that was performed in the
drm_mode_is_420_only() case, but not in the drm_mode_is_420_also() &&
force_yuv420_output case.
Without further knowledge if YCbCr 4:2:0 is supported outside of HDMI,
there is no reason to use RGB when the display
reports drm_mode_is_420_only() even on a non HDMI connection.
This patch also moves both checks in the same if-case. This eliminates an
extra else-if-case.
Signed-off-by: Werner Sembach <wse@tuxedocomputers.com>
Signed-off-by: Andri Yngvason <andri@yngvason.is>
Tested-by: Andri Yngvason <andri@yngvason.is>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
Signed-off-by: Nicolas Frattaroli <nicolas.frattaroli@collabora.com>
Link: https://patch.msgid.link/20260609-color-format-v17-1-35739b5782cc@collabora.com
Signed-off-by: Daniel Stone <daniels@collabora.com>
V3D 3.3 (Broadcom BCM7268) and V3D 4.1 (Broadcom BCM7278) has had no
in-tree userspace since Mesa dropped support in 2024, on the grounds
that those generations were no longer being tested. The situation in
the kernel is similar: the maintainers don't have this hardware, the
hardware is not available for purchase, and there is no known user of
these GPUs.
With no userspace left to drive it and no known users, maintaining the
ver <= 41 code paths is a cost without a benefit, considering that these
paths are not being exercised on real hardware.
As a first step toward removal, emit a deprecation warning at probe for
V3D versions earlier than or equal to 4.1. The hardware remains
functional for now; this only warns. If any real user appears and
explain its use-case, support can be retained.
Schedule the removal of V3D 3.3 and 4.1 support to the next kernel
release.
Link: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/25851
Reviewed-by: Florian Fainelli <florian.fainelli@broadcom.com>
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com>
Link: https://patch.msgid.link/20260606185616.694188-2-mcanal@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Currently, v3d_submit_jobs() arms and pushes each job one at a time,
wiring dependencies between consecutive jobs after each push. If
drm_sched_job_add_dependency() fails midway, the already-pushed jobs are
scheduler-owned and will be submitted to the GPU for execution, even though
the subsequent jobs won't be submitted.
This breaks the atomicity of the submissions, as only some of the jobs
from a submission would be submitted to the hardware, while the other part
fails.
Restructure v3d_submit_jobs() into three phases: (1) arm all jobs belonging
to a given submission, (2) wire inter-job dependencies, and (3) push all
jobs to the scheduler unconditionally. Phase (2) can fail; on failure,
it marks every armed job finished fence with an error, so that run_job()
callbacks skip hardware execution.
This guarantees that every armed job is always pushed, either to run
or to be skipped, and it also ensures the atomicity of a submission.
Suggested-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-12-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
v3d_submit_process_post_deps() looks up the out_sync syncobj via
drm_syncobj_find(), and if userspace passes a non-zero handle that doesn't
refer to a valid syncobj, the lookup silently returns NULL and the
post-deps step skips publishing the submission's last fence to it. The
ioctl still returns success, leaving userspace to wait on a invalid
syncobj.
Instead of silently ignoring an invalid non-zero out_sync, move the syncobj
lookup to the submission and make it fail with -ENOENT up front, mirroring
the syncobj validation already done for in_sync. Now,
v3d_submit_process_post_deps() only does the fence replacement.
Note that the lookup is skipped when the multi-sync extension is in use,
since args->out_sync is unused in that case.
To keep cleanup symmetric on error paths, convert the function
v3d_put_multisync_post_deps() into a single function that releases the
references that were acquired but never published for both single-sync
and multi-sync.
Suggested-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-11-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
v3d_attach_fences_and_unlock_reservation() does three different things:
(1) attaches the submission's last fence to every BO, (2) releases
drm_exec, and (3) replaces the userspace out_sync syncobjs. Decouple
these three behaviors into different functions, so that each function
has a more self-contained behavior.
v3d_submit_jobs() now invokes the three steps explicitly, which makes the
submission sequence self-documenting and keeps each helper self-contained.
No functional change; just code consolidation.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-10-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Restructure the CPU ioctl so that all job types, including indirect CSD,
use a single struct v3d_submit chain and a single DRM exec context.
Now that v3d_get_cpu_indirect_csd_params() is a pure parser and the
submit helpers operate on struct v3d_submit, fold the indirect CSD path
into the standard flow by appending the CSD and CLEAN_CACHE jobs to the
same struct v3d_submit as the CPU job and locking the union of all jobs'
BOs under one drm_exec. This eliminates the second drm_exec, the nested
submission, and the conditional two-pass fence attachment that the CPU
ioctl previously required for the indirect CSD path.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-9-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Generalize the submission helpers so they act on a whole struct v3d_submit
(the entire job chain) rather than on individual jobs and a drm_exec. This
lets a submission of several chained jobs be locked, fenced, and finalized
as a single unit, and is the groundwork for collapsing the indirect CSD
path into one chain.
The following helpers were generalized:
- v3d_lookup_bos()
- v3d_lock_bo_reservations() (renamed to v3d_submit_lock_reservations()):
- v3d_attach_fences_and_unlock_reservation()
- v3d_setup_csd_jobs_and_bos()
Now, the locking helper now iterates over all jobs and locks the union of
their BOs under one DRM exec, using DRM_EXEC_IGNORE_DUPLICATES to tolerate
shared BO references. The fence-attach helper similarly walks every job
and attaches the chain's last fence to all touched BOs.
Also, v3d_submit_jobs() becomes the single submit-and-finalize entry
point and callers no longer need to open-code fence attachment,
reservation unlocking, etc.
Update CL/TFU/CSD/CPU ioctls to use the new helper signatures. The CPU
ioctl still uses two struct v3d_submit instances (one for the CPU job,
one for the indirect CSD jobs) and keeps its manual two-pass fence-attach
flow. Converting the indirect CSD path into the unified chain is done in
the next commit.
No functional change.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-8-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
v3d_get_cpu_indirect_csd_params() currently does double duty: it parses
the indirect CSD extension and, while still inside the extension parser,
also creates the CSD/clean jobs and locks their BOs through a separate
DRM exec context. This nested submission deviates from the standard flow
and makes it hard to fold the indirect CSD path into the unified submit
chain.
Stash the parsed drm_v3d_submit_csd args in struct v3d_indirect_csd_info
and have the parser only fill in the parameters. Then, move job creation
(v3d_setup_csd_jobs_and_bos()) into v3d_submit_cpu_ioctl(), where is the
proper place to create jobs.
No functional change, but prepares to move the CPU ioctl into the
unified submission chain.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-7-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
As the V3D driver grew with time, different types of submission were added
and the submission code grew more complex, but the driver stuck with the
same abstractions.
Nowadays, the submission ioctls don't submit a single job, but a
chain of jobs:
1. v3d_submit_cl_ioctl() submits a BIN job (optional), RENDER job
(mandatory), and a CLEAN_CACHE job (optional).
2. v3d_submit_csd_ioctl() submits a CSD, and a CLEAN_CACHE job.
3. v3d_submit_tfu_ioctl() submits a TFU job.
Therefore, each ioctl submits a chain of jobs in which each job depends on
the previous one. However, this concept is not well represented in software
at the moment.
To address this, introduce a new concept: the struct v3d_submit, which
groups the submission state and represents the submission chain formed by
an ordered array of jobs.
Add new helpers to allocate, add jobs to the chain and submit jobs to
the scheduler, all based on the new struct. Convert v3d_submit_cl_ioctl(),
v3d_submit_tfu_ioctl() and v3d_submit_csd_ioctl() to the new pattern. Each
ioctl now follows the same flow: add jobs -> attach perfmon -> lookup BOs
-> lock reservations -> submit chain -> attach fences -> put jobs.
The CPU ioctl is left on the old helpers for now; its indirect CSD path
requires some restructuring that will be addressed in the next few
commits.
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@igalia.com>
Link: https://patch.msgid.link/20260604-v3d-sched-misc-fixes-v4-6-c068f5bf5ccf@igalia.com
Signed-off-by: Maíra Canal <mcanal@igalia.com>
Eric Engestrom
Karol Wachowski
Uwe Kleine-König (The Capable Hub)
Sudarshan Shetty
Thorsten Blum
Alberto Ruiz
Derek Foreman
Nicolas Frattaroli
Matthew Brost
Ethan Nelson-Moore
Geert Uytterhoeven
Tvrtko Ursulin
Werner Sembach
Maíra Canal