Re: sched/fair: Kernel panics in pick_next_entity

From: Mike Galbraith
Date: Thu Oct 03 2024 - 05:32:01 EST


On Thu, 2024-10-03 at 06:41 +0200, Mike Galbraith wrote:
> On Wed, 2024-10-02 at 15:31 -0700, Benjamin Segall wrote:

> > Whether dequeue_entity + enqueue_entity is better or worse than
> > requeue_delayed_entity (+break), I really don't know.
>
> Hm, I'd say requeue_delayed_entity() not only fits better, it using
> less lines gives it an extra brownie point.

Probable not worth any churn or effort, but it is an option.

sched: Clean up sched_delayed handling in unthrottle_cfs_rq()

requeue_delayed_entity() achieves and documents in one line what a less
clear preparatory dequeue facilitates over several, so use it instead,
and remove the superfluous comment.

Signed-off-by: Mike Galbraith <efault@xxxxxx>
---
kernel/sched/fair.c | 9 +++------
1 file changed, 3 insertions(+), 6 deletions(-)

--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -6058,12 +6058,9 @@ void unthrottle_cfs_rq(struct cfs_rq *cf
for_each_sched_entity(se) {
struct cfs_rq *qcfs_rq = cfs_rq_of(se);

- /* Handle any unfinished DELAY_DEQUEUE business first. */
- if (se->sched_delayed) {
- int flags = DEQUEUE_SLEEP | DEQUEUE_DELAYED;
-
- dequeue_entity(qcfs_rq, se, flags);
- } else if (se->on_rq)
+ if (se->sched_delayed)
+ requeue_delayed_entity(se);
+ if (se->on_rq)
break;
enqueue_entity(qcfs_rq, se, ENQUEUE_WAKEUP);