Re: [RFC PATCH v2 4/8] sched/cpufreq_schedutil: split utilization signals

From: Juri Lelli
Date: Tue Dec 05 2017 - 10:27:00 EST


Hi,

On 05/12/17 15:17, Patrick Bellasi wrote:
> Hi Juri,
>
> On 04-Dec 11:23, Juri Lelli wrote:
> > From: Juri Lelli <juri.lelli@xxxxxxx>
> >
> > To be able to treat utilization signals of different scheduling classes
> > in different ways (e.g., CFS signal might be stale while DEADLINE signal
> > is never stale by design) we need to split sugov_cpu::util signal in two:
> > util_cfs and util_dl.
> >
> > This patch does that by also changing sugov_get_util() parameter list.
> > After this change, aggregation of the different signals has to be performed
> > by sugov_get_util() users (so that they can decide what to do with the
> > different signals).
>
> Did not tried myself, but to me it would be nice to have this patch
> squashed with the first one of this series. After all, looking at this
> one it seems that [RFC PATH 1/8] is just adding util_dl but it's not
> really using it the proper way.
>
> Here instead is where you better introduce two separate signals,
> tracked by struct sugov_cpu, and properly aggregate them.
>
> But perhaps that's just me being picky ;-)
>

Sure. It looked too invasive as a single patch to me. Also, I was trying
to follow the "one change one patch" rule. So, I'd keep them separate.
What others think?

Best,

- Juri