Re: [PATCH] hrtimer: reset .hres_active and .online at appropriate points
From: Koichiro Den
Date: Sat Jan 18 2025 - 02:37:25 EST
On Thu, Jan 16, 2025 at 10:24:07AM GMT, Thomas Gleixner wrote:
> On Thu, Jan 16 2025 at 10:16, Thomas Gleixner wrote:
>
> > On Fri, Dec 20 2024 at 22:44, Koichiro Den wrote:
> >> Reset these fields at the appropriate points. Ideally, .hres_active
> >> would be reset in tick_cpu_dying() as underlying clockevent is shut down
> >> there. However, since these operations occur in the atomic AP section
> >> with interrupts disabled in any case, this patch resets .hres_active to
> >> 0 in hrtimers_cpu_dying() for simplicity.
> >
> > It does not matter where it is reset and from a hrtimer point of view
> > the dying() callback is the appropriate place.
>
> That said. The change is actually incomplete as the other fields which
> are related to the per CPU base state are not reset either.
>
Thank you for the feedback and handling it.
-Koichiro Den