On Mon, Aug 03, 2026 at 07:19:32AM +0000, Harry Yoo wrote:
> On Thu, Jul 30, 2026 at 07:21:19PM +0800, [email protected] wrote:
> > Harry wrote:
> > > call_rcu() disables IRQs with local_irq_save() to protect its per-cpu
> > > data structures. Therefore, if IRQs are not disabled, they cannot be
> > > corrupted by reentrance into call_rcu(). So fall back to the deferred
> > > path only when !allow_spin && irqs_disabled().
> > >
> > > The RCU subsystem does not guarantee this contractually, and this
> > > optimization relies on RCU's implementation details. Ideally, it should
> > > be removed once call_rcu_nolock() is supported by the RCU subsystem.
> > >
> > > Link:
> > > https://lore.kernel.org/linux-mm/caadnvqkrvd5zsnekbzzu7w86ghbghuug2pvzpgztngns+fg...@mail.gmail.com
> > > Suggested-by: Alexei Starovoitov <[email protected]>
> > > Signed-off-by: Harry Yoo (Oracle) <[email protected]>
> > > ---
> > > mm/slub.c | 8 ++++++--
> > > 1 file changed, 6 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/mm/slub.c b/mm/slub.c
> > > index 2c5ba8c326e7..bb0f2975c95b 100644
> > > --- a/mm/slub.c
> > > +++ b/mm/slub.c
> > > @@ -6157,8 +6157,12 @@ bool __kfree_rcu_sheaf(struct kmem_cache *s, void
> > > *obj, unsigned int free_flags)
> > > if (likely(rcu_sheaf->size < s->sheaf_capacity)) {
> > > rcu_sheaf = NULL;
> > > } else {
> > > - if (unlikely(!allow_spin)) {
> > > - /* call_rcu() cannot be called in an unknown context */
> > > + /*
> > > + * With !allow_spin, we might have interrupted call_rcu()'s
> > > + * IRQ-disabled critical section. If IRQs are not disabled,
> > > + * we know that's not the case.
> > > + */
> > > + if (unlikely(!allow_spin && irqs_disabled())) {
> >
> > LGTM!
> > Reviewed-by: Shengming Hu <[email protected]>
>
> Thanks a lot, Shengming!
>
> > I initially wondered whether in_nmi() should also be checked here.
> > However, irqs_disabled() already covers the relevant reentrancy case,
> > since call_rcu() disables IRQs while modifying its per-CPU data.
>
> Exactly, that was my reasoning ;)
>
> As a side note there is an attempt to do this in call_rcu() instead of
> slab:
>
> https://lore.kernel.org/all/[email protected]
>
> Just FYI.
Why not both? ;-)
(Sorry, couldn't resist.)
Thanx, Paul
