On Thu, 2020-11-12 at 19:31 +0000, Valentin Schneider wrote:
> One thing I don't get: that trace shows refcount_dec_and_test()
> (kernel/sched/core.c:2263) happening before the wait_for_completion(). It's
> not the case in the below trace.
Yes, that is normal. Sometimes, the decoding is a bit off not sure because of
some debugging options like KASAN obscures it.
> a) Do you also get this on CONFIG_PREEMPT=y?
I don't know. None of the systems here has that, but I could probably try.
> b) Could you try the below?
Let me run it and report.
>
> ---
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index 02076e6d3792..fad0a8e62aca 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -1923,7 +1923,7 @@ static int migration_cpu_stop(void *data)
> else
> p->wake_cpu = dest_cpu;
>
> - } else if (dest_cpu < 0) {
> + } else if (dest_cpu < 0 || pending) {
> /*
> * This happens when we get migrated between migrate_enable()'s
> * preempt_enable() and scheduling the stopper task. At that
> @@ -1933,6 +1933,17 @@ static int migration_cpu_stop(void *data)
> * more likely.
> */
>
> + /*
> + * The task moved before the stopper got to run. We're holding
> + * ->pi_lock, so the allowed mask is stable - if it got
> + * somewhere allowed, we're done.
> + */
> + if (pending && cpumask_test_cpu(task_cpu(p), p->cpus_ptr)) {
> + p->migration_pending = NULL;
> + complete = true;
> + goto out;
> + }
> +
> /*
> * When this was migrate_enable() but we no longer have an
> * @pending, a concurrent SCA 'fixed' things and we should be
>
