On Thu 02-04-20 09:38:54, David Hildenbrand wrote:
> > + /*
> > + * Once we unlock here, the zone cannot be grown anymore, thus if an
> > + * interrupt thread must allocate this early in boot, zone must be
> > + * pre-grown prior to start of deferred page initialization.
> > + */
> > + pgdat_resize_unlock(pgdat, &flags);
> > +
> > /* Only the highest zone is deferred so find it */
> > for (zid = 0; zid < MAX_NR_ZONES; zid++) {
> > zone = pgdat->node_zones + zid;
> > @@ -1809,11 +1816,9 @@ static int __init deferred_init_memmap(void *data)
> > */
> > while (spfn < epfn) {
> > nr_pages += deferred_init_maxorder(&i, zone, &spfn, &epfn);
> > - touch_nmi_watchdog();
> > + cond_resched();
>
> I do wonder if this change is strictly required in this patch (IOW, if
> we could keep calling touch_nmi_watchdog() also without holding a spinlock)
Exactly. I would go with your patch on top.
> Anyhow, it's the right thing to do.
>
> > }
> > zone_empty:
> > - pgdat_resize_unlock(pgdat, &flags);
> > -
> > /* Sanity check that the next zone really is unpopulated */
> > WARN_ON(++zid < MAX_NR_ZONES && populated_zone(++zone));
> >
> > @@ -1855,17 +1860,6 @@ deferred_grow_zone(struct zone *zone, unsigned int order)
> >
> > pgdat_resize_lock(pgdat, &flags);
> >
> > - /*
> > - * If deferred pages have been initialized while we were waiting for
> > - * the lock, return true, as the zone was grown. The caller will retry
> > - * this zone. We won't return to this function since the caller also
> > - * has this static branch.
> > - */
> > - if (!static_branch_unlikely(&deferred_pages)) {
> > - pgdat_resize_unlock(pgdat, &flags);
> > - return true;
> > - }
> > -
> > /*
> > * If someone grew this zone while we were waiting for spinlock, return
> > * true, as there might be enough pages already.
> >
>
>
> I think we should also look into cleaning up deferred_grow_zone( next),
> we still have that touch_nmi_watchdog() in there. We should rework
> locking. (I think Michal requested that as well)
>
> For now, this seems to survive my basic testing (RCU stalls gone)
>
> --
> Thanks,
>
> David / dhildenb
>
--
Michal Hocko
SUSE Labs
