Re: [PATCH v10 25/40] sched/deadline: Reset the donor's runtime at a deboost in the class

From: sashiko-bot

Date: Fri Nov 20 2026 - 16:26:22 EST


Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Medium] sched/deadline: dl_deboost() starts period for sleeping
constrained task
--

commit 280104d364cfcdeb61df32b73bafd44ef91c8ae1
Author: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>

sched/deadline: Reset the donor's runtime at a deboost in the class

This commit records whether a boosted deadline entity had its runtime
refilled from its donor's parameters in dl_donor_refill, and starts a new
period from the entity's own reservation when deboosting within
SCHED_DEADLINE if that flag is set.

> diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c
> index a566e9252a418..2912d13b60d60 100644
> --- a/kernel/sched/deadline.c
> +++ b/kernel/sched/deadline.c
> @@ -93,12 +93,62 @@ static inline bool is_dl_boosted(struct sched_dl_entity *dl_se)
> {
> return pi_of(dl_se) != dl_se;
> }
> +
> +/*
> + * Record that @dl_se had its runtime refilled from the parameters of its
> + * donor, if it is boosted.
> + */
> +static inline void dl_note_donor_refill(struct sched_dl_entity *dl_se)
> +{
> + if (is_dl_boosted(dl_se))
> + dl_se->dl_donor_refill = 1;
> +}
> +
> +static void task_non_contending(struct sched_dl_entity *dl_se, bool dl_task);
> +
> +/*
> + * End the boost of @dl_se, whose task stays in the class. If the boost
> + * refilled its runtime from the donor's parameters, start a new period
> + * now. The next enqueue is too late for a task blocked on another
> + * rt_mutex: the chain walk copies its deadline into its waiter node
> + * first. Without a refill, the task keeps the runtime and the deadline
> + * of its own reservation.
> + *
> + * The caller holds the rq lock and the pi_lock of the task, and has
> + * updated the rq clock.
> + */
> +void dl_deboost(struct sched_dl_entity *dl_se)
> +{
> + struct rq *rq = rq_of_dl_se(dl_se);
> +
> + if (dl_se->dl_donor_refill) {
> + dl_se->deadline = rq_clock(rq) + dl_se->dl_deadline;
> + dl_se->runtime = dl_se->dl_runtime;

[Severity: Medium]
Does starting a new period in dl_deboost() while the task is asleep cause
a constrained-deadline task (dl_deadline < dl_period) to lose runtime or
be throttled when it later wakes up?

If a deadline waiter times out on an rt_mutex held by a sleeping
constrained-deadline task with dl_donor_refill set (and not blocked on
another rt_mutex), dl_deboost() sets dl_se->deadline = t_deboost +
dl_deadline and dl_se->runtime = dl_runtime while the task sleeps. When
the task later wakes up at t_wake:

- if t_deboost < t_wake <= t_deboost + dl_deadline, dl_entity_overflow()
is true because dl_se->runtime == dl_runtime while remaining time to
deadline is less than dl_deadline, so update_dl_entity() calls
update_dl_revised_wakeup() and scales down dl_se->runtime for the time
spent sleeping since t_deboost;
- if t_deboost + dl_deadline < t_wake < t_deboost + dl_period,
enqueue_dl_entity() -> dl_check_constrained_dl() sees deadline < t_wake
< dl_next_period(dl_se) and throttles the waking task until t_deboost +
dl_period, even though it did not run during that period.

Later in this series, commit "sched/deadline: Check the runtime of a task
whose parameters change" avoids starting a period at deboost for
dl_new_params when the task is neither queued nor pi-blocked, and commit
"sched/deadline: Boost an owner that cannot run on its own reservation"
restores the task's saved reservation when dl_boost_held is set, but the
dl_donor_refill path in dl_deboost() still starts a period while asleep.

--
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1795207232.git.mathieu.desnoyers@xxxxxxxxxxxx?part=25