Re: [PATCH v5 2/2] sched: make update_cpu_load_nohz() consider missed ticks in NOHZ_FULL

[Peter Zijlstra]

On Tue, Nov 10, 2015 at 09:36:02AM +0900, byungchul.park@xxxxxxx wrote:
> +++ b/kernel/sched/fair.c
> @@ -4419,10 +4419,11 @@ static void update_idle_cpu_load(struct rq *this_rq)
>  /*
>   * Called from tick_nohz_idle_exit() -- try and fix up the ticks we missed.
>   */
> -void update_cpu_load_nohz(void)
> +void update_cpu_load_nohz(int active)
>  {
>  	struct rq *this_rq = this_rq();
>  	unsigned long curr_jiffies = READ_ONCE(jiffies);
> +	unsigned long load = active ? weighted_cpuload(cpu_of(this_rq)) : 0;
>  	unsigned long pending_updates;
>  
>  	if (curr_jiffies == this_rq->last_load_update_tick)
> @@ -4433,10 +4434,11 @@ void update_cpu_load_nohz(void)
>  	if (pending_updates) {
>  		this_rq->last_load_update_tick = curr_jiffies;
>  		/*
> -		 * We were idle, this means load 0, the current load might be
> -		 * !0 due to remote wakeups and the sort.
> +		 * In the regular NOHZ case, we were idle, this means load 0.
> +		 * In the NOHZ_FULL case, we were non-idle, we should consider
> +		 * its weighted load.
>  		 */
> -		__update_cpu_load(this_rq, 0, pending_updates, 0);
> +		__update_cpu_load(this_rq, load, pending_updates, active);
>  	}
>  	raw_spin_unlock(&this_rq->lock);
>  }

Bah, so I did all the work to get the actual number of lost ticks in
there, only to _then_ find out that's mostly pointless :-)

The problem is update_idle_cpu_load() is called while idle (from another
CPU), so it still needs the whole jiffy based thing.

So I'll take this patch for now. Thanks.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/