Re: [PATCH v12 3/4] cpuidle: Export the next timer/tick expiration for a CPU

[Ulf Hansson]

On Tue, 26 Mar 2019 at 11:36, Rafael J. Wysocki <rafael@xxxxxxxxxx> wrote:
>
> On Mon, Mar 25, 2019 at 3:24 PM Ulf Hansson <ulf.hansson@xxxxxxxxxx> wrote:
> >
> > On Mon, 25 Mar 2019 at 13:21, Rafael J. Wysocki <rjw@xxxxxxxxxxxxx> wrote:
> > >
> > > On Wednesday, February 27, 2019 8:58:35 PM CET Ulf Hansson wrote:
> > > > To be able to predict the sleep duration for a CPU that is entering idle,
> > > > knowing when the next timer/tick is going to expire, is extremely useful.
> > > > Both the teo and the menu cpuidle governors already makes use of this
> > > > information, while selecting an idle state.
> > > >
>
> [cut]
>
> > >
> > > > +
> > > >       if (cpuidle_state_is_coupled(drv, index))
> > > >               return cpuidle_enter_state_coupled(dev, drv, index);
> > > >       return cpuidle_enter_state(dev, drv, index);
> > >
> > > Also I would clear next_hrtimer here to avoid dragging stale values
> > > around.
> >
> > Right, I can do that.
> >
> > However, at least in my case it would be an unnecessary update of the
> > variable, as I am never in a path where the value can be "stale".
>
> It easily can AFAICS.  After all, cpu_power_down_ok() need not run on
> the same CPU that is setting next_hrtimer here.

That's correct.

>
> > Even if one theoretically could use a stale value, it's seems likely to not
> > be an issue, don't you think?
>
> That would be because of the locking in the ->enter() callback I
> suppose?  But is it actually universally guaranteed that setting
> next_hrtimer will never be reordered with acquiring the lock?

In the PSCI case and for those CPUs that shares the same genpd
governor (even hierarchically), then yes.

Unfortunate, I haven't been able to explore this in that great detail
for other legacy ARM32 platforms, so maybe it's just better to play
safe, as you suggest and avoid a stale value.

>
> Also, there is some overhead to be avoided if cpu_power_down_ok()
> checked the next_hrtimer of the other CPUs against 0 explicitly, isn't
> it?

In regards to overhead and when using genpd for CPUs, there are a
couple of things I have in mind that we could try to improve. Yes,
checking for next_hrtimer against 0 could be one thing to consider.

Kind regards
Uffe