On 2017/7/12 0:34, Peter Zijlstra wrote:
> On Tue, Jul 11, 2017 at 06:09:27PM +0200, Frederic Weisbecker wrote:
>
>>>> - tick_nohz_idle_enter costs 7058ns - 10726ns
>>>> - tick_nohz_idle_exit costs 8372ns - 20850ns
>>>
>>> Right, those are horrible expensive, but skipping them isn't 'hard', the
>>> only tricky bit is finding a condition that makes sense.
>>
>> Note you can statically disable it with nohz=0 boot parameter.
>
> Yeah, but that's bad for power usage, nobody wants that.
>
>>> See Mike's patch: https://patchwork.kernel.org/patch/2839221/
>>>
>>> Combined with the above, and possibly a better condition, that should
>>> get rid of most of this.
>>
>> Such a patch could work well if the decision from the scheduler to not stop
>> the tick
>> happens on idle entry.
>>
>> Now if sched_needs_cpu() first allows to stop the tick then refuses it later
>> in the end of an idle IRQ, this won't have the desired effect. As long as
>> ts->tick_stopped=1,
>> it stays so until we really restart the tick. So the whole costly nohz
>> machinery stays on.
>>
>> I guess it doesn't matter though, as we are talking about making fast idle
>> entry so the
>> decision not to stop the tick is likely to be done once on idle entry, when
>> ts->tick_stopped=0.
>>
>> One exception though: if the tick is already stopped when we enter idle
>> (full nohz case). And
>> BTW stopping the tick outside idle shouldn't be concerned here.
>>
>> So I'd rather put that on can_stop_idle_tick().
>
> Mike's patch much predates the existence of that function I think ;-) But
> sure..
>
Okay, the difference is that Mike's patch uses a very simple algorithm to make
the decision.
/*
* delta is wakeup_timestamp - idle_timestamp
*/
update_avg(&rq->avg_idle, delta);
...
static void update_avg(u64 *avg, u64 sample)
{
s64 diff = sample - *avg;
*avg += diff >> 3;
}
While my proposal is trying to leverage the prediction functionality of the
existing idle menu
governor, which works very well for a long time.
I know the the code change is big and the running overhead is a bit higher than
rq->avg_idle, but
should we make a comparison for some typical workloads?
Thanks,
-Aubrey
