Re: [PATCH v5 3/4] sched, psi: Don't account irq time if sched_clock_irqtime is disabled

[Yafang Shao]

On Tue, Dec 3, 2024 at 6:15 PM Michal Koutný <mkoutny@xxxxxxxx> wrote:
>
> On Tue, Dec 03, 2024 at 11:01:41AM GMT, Michal Koutný <mkoutny@xxxxxxxx> wrote:
> > On Fri, Nov 08, 2024 at 09:29:03PM GMT, Yafang Shao <laoar.shao@xxxxxxxxx> wrote:
> > > sched_clock_irqtime may be disabled due to the clock source. When disabled,
> > > irq_time_read() won't change over time, so there is nothing to account. We
> > > can save iterating the whole hierarchy on every tick and context switch.
> > >
> > > Signed-off-by: Yafang Shao <laoar.shao@xxxxxxxxx>
> > > Acked-by: Johannes Weiner <hannes@xxxxxxxxxxx>
> > > ---
> > >  kernel/sched/psi.c | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > Reviewed-by: Michal Koutný <mkoutny@xxxxxxxx>
>
> On second thought, similar guard may be useful in psi_show() too. Since
> there's a difference between zero pressure and unmeasured pressure (it'd
> fail with EOPNOTSUPP).

I'll update the psi_show() function in the next version.

>
> (How common is it actually that tsc_init fails?)

This is not a common scenario, but it can occur randomly across a large fleet.

--
Regards
Yafang