Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel
os
linux
drm/sched: Add warning for removing hack in drm_sched_fini()
The assembled developers agreed at the X.Org Developers Conference 2025
that the hack added for amdgpu in drm_sched_fini() shall be removed. It
shouldn't be needed by amdgpu anymore.
As it's unclear whether all drivers really follow the life time rule of
entities having to be torn down before their scheduler, it is reasonable
to warn for a while before removing the hack.
Add a warning in drm_sched_fini() that fires if an entity is still
active.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Philipp Stanner <phasta@kernel.org>
Link: https://patch.msgid.link/20251023123429.139848-2-phasta@kernel.org
+8
-1
+8
-1
drivers/gpu/drm/scheduler/sched_main.c
+8
-1
drivers/gpu/drm/scheduler/sched_main.c
···
1425
1425
struct drm_sched_rq *rq = sched->sched_rq[i];
1426
1426
1427
1427
spin_lock(&rq->lock);
1428
-
list_for_each_entry(s_entity, &rq->entities, list)
1428
+
list_for_each_entry(s_entity, &rq->entities, list) {
1429
1429
/*
1430
1430
* Prevents reinsertion and marks job_queue as idle,
1431
1431
* it will be removed from the rq in drm_sched_entity_fini()
···
1446
1446
* For now, this remains a potential race in all
1447
1447
* drivers that keep entities alive for longer than
1448
1448
* the scheduler.
1449
+
*
1450
+
* The READ_ONCE() is there to make the lockless read
1451
+
* (warning about the lockless write below) slightly
1452
+
* less broken...
1449
1453
*/
1454
+
if (!READ_ONCE(s_entity->stopped))
1455
+
dev_warn(sched->dev, "Tearing down scheduler with active entities!\n");
1450
1456
s_entity->stopped = true;
1457
+
}
1451
1458
spin_unlock(&rq->lock);
1452
1459
kfree(sched->sched_rq[i]);
1453
1460
}