Re: [PATCH v5 09/17] sched/deadline: Reset the deadline and the runtime with the parameters
From: sashiko-bot
Date: Thu Nov 19 2026 - 14:48:49 EST
Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Medium] sched/deadline: donor runtime and deadline retained after PI
deboost
- [Medium] sched/deadline: uncanceled dl_timer in __dl_clear_params()
--
commit 05620e4d8280739bac7b4e66a17979a460a0143c
Author: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
sched/deadline: Reset the deadline and the runtime with the parameters
This commit resets a deadline entity's dynamic deadline and runtime in
__dl_clear_params() and when a PI-boosted task without its own deadline
parameters leaves SCHED_DEADLINE in switched_from_dl().
> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 71b1c155c2424..d6d2657dcddd6 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -3569,6 +3569,16 @@ static void switched_from_dl(struct rq *rq, struct task_struct *p)
> if (task_on_rq_queued(p) && p->dl.dl_runtime)
> task_non_contending(&p->dl, false);
>
> + /*
> + * A task with no parameters of its own was boosted into the class.
> + * Its deadline and runtime are the donor's. Reset them, so that its
> + * next entry into the class starts a new period.
> + */
> + if (!p->dl.dl_runtime) {
> + p->dl.deadline = 0;
> + p->dl.runtime = 0;
> + }
> +
[Severity: Medium]
Can a task still retain the donor's runtime and deadline after a PI deboost
when p->dl.dl_runtime is non-zero?
If a non-deadline task is PI-boosted into dl_sched_class and sched_setattr()
then switches it to SCHED_DEADLINE while still boosted, prev_class and
next_class in __sched_setscheduler() are both &dl_sched_class.
switched_from_dl() and switched_to_dl() are skipped, and __setparam_dl()
sets p->dl.dl_runtime without clearing the donor's p->dl.deadline or
p->dl.runtime (and if p is sleeping, skipping switched_to_dl() also skips
add_rq_bw(&p->dl, &rq->dl), later triggering the running_bw > this_bw
warning and underflow in __sub_rq_bw()). When rt_mutex_setprio() later
deboosts p, it remains in dl_sched_class and restores p->dl.pi_se = &p->dl
without ENQUEUE_REPLENISH. Because the donor's deadline is still in the
future, enqueue_dl_entity() skips setup_new_dl_entity() and p runs unboosted
on the donor's remaining runtime and deadline.
Similarly, if a task with non-zero p->dl.dl_runtime (either a task waiting
for its inactive_timer before 0-lag time that gets PI-boosted, or a boosted
SCHED_DEADLINE task that switches to a non-deadline policy via
__setscheduler_dl_pi()) is replenished from a donor while boosted,
switched_from_dl() on deboost sees p->dl.dl_runtime != 0, skips this reset,
and passes the donor-derived deadline and runtime to task_non_contending().
> @@ -3984,7 +3994,8 @@ bool __checkparam_dl(const struct sched_attr *attr)
> }
>
> /*
> - * This function clears the sched_dl_entity static params.
> + * Clear the static parameters of @dl_se. Also reset its deadline and its
> + * runtime, so that its next enqueue starts a new period.
> */
> static void __dl_clear_params(struct sched_dl_entity *dl_se)
> {
> @@ -3995,6 +4006,9 @@ static void __dl_clear_params(struct sched_dl_entity *dl_se)
> dl_se->dl_bw = 0;
> dl_se->dl_density = 0;
>
> + dl_se->deadline = 0;
> + dl_se->runtime = 0;
> +
> dl_se->dl_throttled = 0;
> dl_se->dl_yielded = 0;
> dl_se->dl_non_contending = 0;
[Severity: Medium]
Can clearing dl_se->deadline, dl_se->runtime, and dl_se->dl_throttled here
without canceling dl_se->dl_timer leave a stale replenishment timer queued
from the old period?
When a task is throttled with its 0-lag time earlier than
dl_next_period(dl_se) (for example after sched_yield() with remaining
runtime, or with a constrained deadline where dl_deadline < dl_period) and
switches out of SCHED_DEADLINE, __dl_clear_params() runs at the 0-lag time
while dl_se->dl_timer is still armed for the old period end. If the task
re-enters SCHED_DEADLINE before that timer fires, enqueue_dl_entity() sees
dl_se->deadline == 0 and starts a new period without canceling dl_timer
(since dl_se->dl_throttled was cleared). When the task throttles in its new
period, start_dl_timer() sees hrtimer_is_queued(&dl_se->dl_timer) and skips
arming the timer for the new period:
kernel/sched/deadline.c:start_dl_timer() {
...
if (!hrtimer_is_queued(timer)) {
if (!dl_server(dl_se))
get_task_struct(dl_task_of(dl_se));
hrtimer_start(timer, act, HRTIMER_MODE_ABS_HARD);
}
return 1;
}
When the old dl_timer expires at the previous period boundary,
dl_task_timer() replenishes and unthrottles the task at the wrong time,
allowing it to exceed its admitted bandwidth reservation.
--
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1795115075.git.mathieu.desnoyers@xxxxxxxxxxxx?part=9