Re: [syzbot] [mm?] WARNING in __mod_zone_page_state

[Sebastian Andrzej Siewior]

On 2026-08-30 10:29:46 [-0700], Hugh Dickins wrote:
> Oh for the good old days when we were allowed to say preempt_disable()!

I'm sorry for taking this away from you.

> > --- a/mm/mlock.c
> > +++ b/mm/mlock.c
> > @@ -141,11 +141,16 @@ static struct lruvec *__munlock_folio(struct folio *folio, struct lruvec *lruvec
> >  
> >  munlock:
> >  	if (folio_test_clear_mlocked(folio)) {
> > -		__zone_stat_mod_folio(folio, NR_MLOCK, -nr_pages);
> > +		/*
> > +		 * This runs both with and without the lruvec lock held, and
> > +		 * mlock_drain_remote() reaches it fully preemptible, so use
> > +		 * the accessors that serialize themselves.
> 
> I'm very far from being a good advisor on PREEMPT_RT,
> but I think that comment about lruvec lock would be wrong there.

This change would eliminate the warning on !PREEMPT_RT and not cause any
further trouble for PREEMPT_RT.

Given that this it might mandate for

    WARN_ON_ONCE(IS_ENABLED(CONFIG_DEBUG_VM) &&
		!IS_ENABLED(CONFIG_PREEMPT_RT) &&
		!!irqs_disabled());

in __mod_zone_page_state() and other places where
preempt_disable_nested() was added. The counters do rely on disabled
interrupts not preemption.

> Hugh

Sebastian