Re: [PATCH 06/16] sched: Take cpuset_mutex for a deadline policy change only with pi
From: sashiko-bot
Date: Wed Nov 18 2026 - 15:17:58 EST
Thank you for your contribution! Sashiko AI review found no new issues
in this patch, but noticed 1 related pre-existing issue it does not
introduce:
- [High] sched/deadline: bandwidth leak in __sched_setscheduler() on
TASK_DEAD task [1]
--
commit eb6bc35bf21202354f98c206678837ac0c913d54
Author: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
sched: Take cpuset_mutex for a deadline policy change only with pi
This commit avoids taking cpuset_mutex in __sched_setscheduler() when pi
is unset, and retries under the runqueue lock if the target task entered
SCHED_DEADLINE after the unlocked policy check.
> diff --git a/kernel/sched/syscalls.c b/kernel/sched/syscalls.c
> index cb115aa9b0b33..9b04e77297aeb 100644
> --- a/kernel/sched/syscalls.c
> +++ b/kernel/sched/syscalls.c
> @@ -561,9 +561,11 @@ int __sched_setscheduler(struct task_struct *p,
>
> /*
> * SCHED_DEADLINE bandwidth accounting relies on stable cpusets
> - * information.
> + * information. Do not take cpuset_mutex when pi is unset: the
> + * caller is normalize_rt_tasks(), which holds tasklist_lock and
> + * cannot sleep.
> */
> - if (dl_policy(policy) || dl_policy(p->policy)) {
> + if (pi && (dl_policy(policy) || dl_policy(p->policy))) {
> cpuset_locked = true;
> cpuset_lock();
> }
> @@ -643,7 +645,13 @@ int __sched_setscheduler(struct task_struct *p,
> }
> }
>
> - /* Re-check policy now with rq lock held: */
> + /*
> + * The caller gave no policy (SETPARAM_POLICY): the request was
> + * derived from an unlocked read of p->policy above. If the task
> + * changed policy since, re-derive the request from scratch, and
> + * release cpuset_mutex: the re-derived request re-decides whether
> + * it is needed.
> + */
> if (unlikely(oldpolicy != -1 && oldpolicy != p->policy)) {
> policy = oldpolicy = -1;
> task_rq_unlock(rq, p, &rf);
> @@ -654,6 +662,18 @@ int __sched_setscheduler(struct task_struct *p,
> goto recheck;
> }
>
> + /*
> + * The caller gave an explicit policy, so the request needs no
> + * re-derivation (oldpolicy is -1 above). But the decision to
> + * take cpuset_mutex used the same unlocked read: the task
> + * entered SCHED_DEADLINE since, and the change below would take
> + * it out of the class. Retry to take the mutex.
> + */
> + if (unlikely(pi && !cpuset_locked && task_has_dl_policy(p))) {
> + task_rq_unlock(rq, p, &rf);
> + goto recheck;
> + }
> +
> oldprio = p->prio;
>
> newprio = __normal_prio(policy, attr->sched_priority, attr->sched_nice);
[Severity: High]
This problem wasn't introduced by this patch, but can
__sched_setscheduler() leak deadline bandwidth and cpuset deadline task
counters if called on a zombie task [1]?
A task that has reached TASK_DEAD and been deactivated from its runqueue
remains reachable by PID until reaped. If sched_setattr() sets
SCHED_DEADLINE on it, sched_dl_overflow() adds the requested bandwidth to
dl_b->total_bw and rq->dl.extra_bw, and switched_to_dl() increments
cs->nr_deadline_tasks via inc_dl_tasks_cs() and adds the bandwidth to
rq->dl.this_bw:
kernel/sched/deadline.c:switched_to_dl() {
...
inc_dl_tasks_cs(p);
/* If p is not queued we will update its parameters at next wakeup. */
if (!task_on_rq_queued(p)) {
add_rq_bw(&p->dl, &rq->dl);
return;
}
...
}
Because a TASK_DEAD task never wakes or runs again and release_task() does
not release deadline bandwidth or decrement cs->nr_deadline_tasks, the
bandwidth and cpuset task count are leaked until reboot, blocking cpuset
changes and eventually causing __dl_overflow() to reject all new
SCHED_DEADLINE admissions with -EBUSY.
[1] https://sashiko.dev/#/bug/linux-6b520332-dcc7-4d6b-b09a-5ec47df1d3cc
--
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1795031605.git.mathieu.desnoyers@xxxxxxxxxxxx?part=6