Re: [PATCH 2/3] rcu: Defer RCU kthreads wakeup when CPU is dying

[Hillf Danton]

On Tue, 19 Dec 2023 00:19:15 +0100 Frederic Weisbecker <frederic@xxxxxxxxxx>
> +static void swake_up_one_online(struct swait_queue_head *wqh)
> +{
> +	int cpu = get_cpu();
> +
> +	/*
> +	 * If called from rcutree_report_cpu_starting(), wake up
> +	 * is dangerous that late in the CPU-down hotplug process. The
> +	 * scheduler might queue an ignored hrtimer. Defer the wake up
> +	 * to an online CPU instead.
> +	 */

But why is scheduler having any interest selecting a dying CPU for
adding a hrtimer on at the first place?

> +	if (unlikely(cpu_is_offline(cpu))) {
> +		int target;
> +
> +		target = cpumask_any_and(housekeeping_cpumask(HK_TYPE_RCU),
> +					 cpu_online_mask);
> +
> +		smp_call_function_single(target, swake_up_one_online_ipi,
> +					 wqh, 0);
> +		put_cpu();
> +	} else {
> +		put_cpu();
> +		swake_up_one(wqh);
> +	}
> +}