Re: [RFC][PATCH 2/5] mutex: Use p->oncpu for the adaptive spin

[Oleg Nesterov <oleg@redhat.com>]

On 12/16, Peter Zijlstra wrote:
>
> @@ -3867,52 +3866,29 @@ int mutex_spin_on_owner(struct mutex *lo
>  	 * DEBUG_PAGEALLOC could have unmapped it if
>  	 * the mutex owner just released it and exited.
>  	 */
> -	if (probe_kernel_address(&owner->cpu, cpu))
> +	if (probe_kernel_address(&owner->oncpu, oncpu))
>  		return 0;
>  #else
> -	cpu = owner->cpu;
> +	oncpu = owner->oncpu;
>  #endif
>
> -	/*
> -	 * Even if the access succeeded (likely case),
> -	 * the cpu field may no longer be valid.
> -	 */
> -	if (cpu >= nr_cpumask_bits)
> -		return 0;
> -
> -	/*
> -	 * We need to validate that we can do a
> -	 * get_cpu() and that we have the percpu area.
> -	 */
> -	if (!cpu_online(cpu))
> +	if (!oncpu)
>  		return 0;
>
> -	rq = cpu_rq(cpu);
> -
> -	for (;;) {
> -		/*
> -		 * Owner changed, break to re-assess state.
> -		 */
> -		if (lock->owner != owner) {
> -			/*
> -			 * If the lock has switched to a different owner,
> -			 * we likely have heavy contention. Return 0 to quit
> -			 * optimistic spinning and not contend further:
> -			 */
> -			if (lock->owner)
> -				return 0;
> -			break;
> -		}
> -
> -		/*
> -		 * Is that owner really running on that cpu?
> -		 */
> -		if (task_thread_info(rq->curr) != owner || need_resched())
> +	while (lock->owner == owner && owner->oncpu) {

It seems, this owner->oncpu dereference needs probe_kernel_address() too.
Of course, only in theory.

OTOH, I do not understand why we need the first "if (!oncpu)" check
(and "int oncpu" at all).

Oleg.