rocket_reset() drops the in-flight job's runtime PM reference with pm_runtime_put_noidle(), which requests nothing, so the core does not suspend until something else asks and its power domain does not cycle. On RK3576 the domain's power-on resets the NPU bus interface; without it the IOMMU stops answering and the job after a timeout fails.
Use pm_runtime_put_autosuspend(), as the completion path does. On a ROCK 4D the job after a timeout goes from 0 to 128 of 128 channels correct. Link: https://lore.kernel.org/all/[email protected]/ Fixes: 0810d5ad88a1 ("accel/rocket: Add job submission IOCTL") Signed-off-by: Jiaxing Hu <[email protected]> Tested-by: Igor Paunovic <[email protected]> # RK3588, three cores, induced reset, JOB_TIMEOUT_MS=2 --- drivers/accel/rocket/rocket_job.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/drivers/accel/rocket/rocket_job.c b/drivers/accel/rocket/rocket_job.c index bcafa89ba..996890025 100644 --- a/drivers/accel/rocket/rocket_job.c +++ b/drivers/accel/rocket/rocket_job.c @@ -443,12 +443,12 @@ rocket_reset(struct rocket_core *core, struct drm_sched_job *bad) /* * No handler is running now, but we might still have stuck jobs. Let's - * make sure the PM counters stay balanced by manually calling - * pm_runtime_put_noidle(). + * make sure the PM counters stay balanced by putting the reference the + * job took, and request idle while doing it so the core can suspend. */ scoped_guard(mutex, &core->job_lock) { if (core->in_flight_job) - pm_runtime_put_noidle(core->dev); + pm_runtime_put_autosuspend(core->dev); iommu_detach_group(NULL, core->iommu_group); -- 2.43.0
