Re: [PATCH 2/2] sched/fair: Avoid calling sync_entity_load_avg() unnecessarily

[Viresh Kumar]

Hi Dietmar,

On 27-04-18, 10:30, Dietmar Eggemann wrote:
> I would suggest we leave the call to sync_entity_load_avg() in the slow path
> of strf() so that we're not forced to call it in
> find_energy_efficient_cpu().

Well, that's what I did in the very first attempt:

20180425051509.aohopadqw7q5urbd@vireshk-i7">https://lkml.kernel.org/r/20180425051509.aohopadqw7q5urbd@vireshk-i7

But then I realized that based on the current state of code, its
better to move it to find_idlest_cpu() instead.

I would be fine with sending the above patch instead (if Peter
agrees). Though it should be trivial to move it back to strq() in the
patch where you are adding call to find_energy_efficient_cpu().

I am fine with both the options though.

-- 
viresh