Re: [PATCH v8 27/33] sched/deadline: Start a period at a deboost out of the class

From: sashiko-bot

Date: Thu Nov 19 2026 - 15:47:56 EST


Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Medium] sched/deadline: delayed bw release in switched_from_dl() at
0-lag time 0
- [Medium] sched/deadline: uncanceled donor timer in switched_from_dl()
when asleep
--

commit d30233a80ec3ae5aaad3cff8dd2909eea144f7c6
Author: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>

sched/deadline: Start a period at a deboost out of the class

This commit starts a new period using the task's own reservation parameters
in switched_from_dl() when a boosted task with dl_donor_refill set leaves
SCHED_DEADLINE, sharing the period reset logic with dl_deboost() via
dl_start_own_period().

> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index 77a1d1f463f6d..d284036ecb8a9 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -83,6 +83,16 @@ static inline int on_dl_rq(struct sched_dl_entity *dl_se)
> return !RB_EMPTY_NODE(&dl_se->rb_node);
> }
>
> +/*
> + * Start a period of the reservation of @dl_se now. The caller holds the rq
> + * lock and has updated the rq clock.
> + */
> +static inline void dl_start_own_period(struct sched_dl_entity *dl_se)
> +{
> + dl_se->deadline = rq_clock(rq_of_dl_se(dl_se)) + dl_se->dl_deadline;
> + dl_se->runtime = dl_se->dl_runtime;
> +}

[ ... ]

> @@ -4504,6 +4510,17 @@ void dl_clear_root_domains(void)
>
> static void switched_from_dl(struct rq *rq, struct task_struct *p)
> {
> + /*
> + * The boost of a task with parameters of its own ends, and the task
> + * leaves the class. If the boost refilled its runtime from the
> + * donor's parameters, its runtime and deadline are not those of its
> + * reservation. task_non_contending() would mix them with its own
> + * dl_runtime and dl_period. Start a period of its own instead, as
> + * dl_deboost() does. Its 0-lag time is then now, or has passed.
> + */
> + if (p->dl.dl_runtime && p->dl.dl_donor_refill)
> + dl_start_own_period(&p->dl);
> +
> /*
> * task_non_contending() can start the "inactive timer" (if the 0-lag
> * time is in the future). If the task switches back to dl before

[Severity: Medium]
Does task_non_contending() still arm a 0-ns inactive_timer instead of
releasing the bandwidth at once when dl_deadline == dl_period?

After dl_start_own_period(&p->dl), left equals dl_deadline and runtime
equals dl_runtime. For an implicit-deadline reservation (dl_deadline ==
dl_period), task_non_contending() computes zerolag_time == 0, which does
not take the immediate release branch:

kernel/sched/deadline.c:task_non_contending() {
...
} else if (left < 0 ||
dl_se->runtime > mul_u64_u64_div_u64(left, dl_se->dl_runtime,
dl_se->dl_period))
zerolag_time = -1;
else
zerolag_time = left -
mul_u64_u64_div_u64(dl_se->runtime, dl_se->dl_period,
dl_se->dl_runtime);

/*
* If the "0-lag time" already passed, decrease the active
* utilization now, instead of starting a timer
*/
if ((zerolag_time < 0) || hrtimer_active(&dl_se->inactive_timer)) {
...
}

Because zerolag_time is 0 rather than negative, task_non_contending() arms
inactive_timer for 0 ns and switched_from_dl() puts p on dl_departed_list,
keeping p's bandwidth in total_bw and its parameters uncleared until the
timer interrupt fires and inactive_task_timer() acquires the rq lock.

[Severity: Medium]
What happens if p is asleep with p->dl.dl_non_contending already set when
it is deboosted out of SCHED_DEADLINE?

If a boosted task with its own parameters was refilled from its donor and
then blocked, dequeue_dl_entity() -> task_non_contending(&p->dl, true)
already armed inactive_timer from the mixed donor and owner parameters and
set p->dl.dl_non_contending = 1. If the boost ends while p is still asleep
(for instance, when the donor waiter times out or is interrupted),
switched_from_dl() calls dl_start_own_period(&p->dl) here, but then skips
task_non_contending(&p->dl, false) in the !task_on_rq_queued(p) block
because p->dl.dl_non_contending is already set:

kernel/sched/deadline.c:switched_from_dl() {
...
if (!task_on_rq_queued(p)) {
if (p->dl.dl_non_contending)
sub_running_bw(&p->dl, &rq->dl);
sub_rq_bw(&p->dl, &rq->dl);
if (!p->dl.dl_non_contending)
task_non_contending(&p->dl, false);
}

if (p->dl.dl_non_contending) {
p->dl.dl_non_contending = 0;
dl_task_depart(p);
}
...
}

The old donor-mixed inactive_timer stays armed and p is placed on
dl_departed_list, keeping its bandwidth in total_bw until that timer fires.

--
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1795116790.git.mathieu.desnoyers@xxxxxxxxxxxx?part=27