Re: [PATCH v2 2/2 sched_ext/for-6.12] sched_ext: Use sched_clock_cpu() instead of rq_clock_task() in touch_core_sched()
From: David Vernet
Date: Fri Aug 30 2024 - 14:02:13 EST
On Fri, Aug 30, 2024 at 07:54:41AM -1000, Tejun Heo wrote:
> Since 3cf78c5d01d6 ("sched_ext: Unpin and repin rq lock from
> balance_scx()"), sched_ext's balance path terminates rq_pin in the outermost
> function. This is simpler and in line with what other balance functions are
> doing but it loses control over rq->clock_update_flags which makes
> assert_clock_udpated() trigger if other CPUs pins the rq lock.
>
> The only place this matters is touch_core_sched() which uses the timestamp
> to order tasks from sibling rq's. Switch to sched_clock_cpu(). Later, it may
> be better to use per-core dispatch sequence number.
>
> v2: Use sched_clock_cpu() instead of ktime_get_ns() per David.
>
> Signed-off-by: Tejun Heo <tj@xxxxxxxxxx>
> Fixes: 3cf78c5d01d6 ("sched_ext: Unpin and repin rq lock from balance_scx()")
> Cc: David Vernet <void@xxxxxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Acked-by: David Vernet <void@xxxxxxxxxxxxx>
Attachment:
signature.asc
Description: PGP signature