Re: [PATCH/RFC] sched: Remove SYSTEM_RUNNING checks from cond_resched*()

[Oleg Nesterov <oleg@redhat.com>]

On 07/08, Anton Vorontsov wrote:
>
> but I'm curious if removing the checks makes sense
> nowadays.
...
> --- a/kernel/sched.c
> +++ b/kernel/sched.c
> @@ -6560,8 +6560,7 @@ static void __cond_resched(void)
>
>  int __sched _cond_resched(void)
>  {
> -	if (need_resched() && !(preempt_count() & PREEMPT_ACTIVE) &&
> -					system_state == SYSTEM_RUNNING) {
> +	if (need_resched() && !(preempt_count() & PREEMPT_ACTIVE)) {
>  		__cond_resched();
>  		return 1;
>  	}

and, with CONFIG_PREEMPT preempt_schedule() does not check system_state,
so it looks really strange cond_resched() does check SYSTEM_RUNNING.



debug_smp_processor_id() looks strange too:
	
	/*
	 * It is valid to assume CPU-locality during early bootup:
	 */
	if (system_state != SYSTEM_RUNNING)
		goto out;

this doesn't look right, smp_init() is called before we set
SYSTEM_RUNNING.

Hmm, and

	/*
	 * Kernel threads bound to a single CPU can safely use
	 * smp_processor_id():
	 */
	if (cpumask_equal(&current->cpus_allowed, cpumask_of(this_cpu)))
		goto out;

perhaps this should use PF_THREAD_BOUND ?

Oleg.