On 2025-01-30 16:03:09 [+0100], Christophe Leroy wrote:
> Le 30/01/2025 à 15:54, Sebastian Andrzej Siewior a écrit :
> > On 2025-01-06 10:49:19 [+0530], Shrikanth Hegde wrote:
> > > --- a/arch/powerpc/kernel/interrupt.c
> > > +++ b/arch/powerpc/kernel/interrupt.c
> > > @@ -25,6 +25,10 @@
> > > unsigned long global_dbcr0[NR_CPUS];
> > > #endif
> > > +#if defined(CONFIG_PREEMPT_DYNAMIC) &&
> > > defined(CONFIG_HAVE_PREEMPT_DYNAMIC_KEY)
> > > +DEFINE_STATIC_KEY_TRUE(sk_dynamic_irqentry_exit_cond_resched);
> > > +#endif
> >
> > I am uncertain here: Do you need to DEFINE it? It is set by the sched
> > core which also defines it. It should be same thing after all, right?
>
> As far as I can see it is not handled by sched core.
…
> It is in common entry but arm64 and powerpc don't use common entry.
Okay. So it is defined in the generic part (so you have to define it on
power) but it is used by sched-core (search for
preempt_dynamic_enable(irqentry_exit_cond_resched);
).
It might make sense to define it there (in the sched part where it is
used) but this is out of scope here, it just confused me :)
Thanks for the pointer.
> Christophe
Sebastian
