Re: [PATCH v1] cpuidle: Do not return from cpuidle_play_dead() on callback failures
From: Peter Zijlstra
Date: Fri Nov 15 2024 - 12:40:38 EST
On Fri, Nov 15, 2024 at 02:25:23PM +0100, Rafael J. Wysocki wrote:
> > I mean, something like:
> >
> > if (drv->states[i].enter_dead && !drv->states[i].enter_dead(dev, i))
> > panic("Dead CPU walking...");
> >
> > is 'fun' but not very useful.
>
> The panic would be hard to debug if it ever triggers I'm afraid and
> there is the fallback to HLT in the caller.
I was being facetious, removing the return value and simply calling them
all in reverse order is fine.