On Sun, Apr 20, 2025 at 12:35 AM Kairui Song <ryncsn@xxxxxxxxx> wrote: > > On Sun, Apr 20, 2025 at 12:32 AM David Hildenbrand <david@xxxxxxxxxx> wrote: > > > > On 19.04.25 18:25, David Hildenbrand wrote: > > > > Oh, re-reading the condition 3 times, I realize that the sanity check is wrong ... > > > > diff --git a/mm/memory.c b/mm/memory.c > > index 037b6ce211f1f..a17eeef3f1f89 100644 > > --- a/mm/memory.c > > +++ b/mm/memory.c > > @@ -3789,7 +3789,7 @@ static bool __wp_can_reuse_large_anon_folio(struct folio *folio, > > > > /* Stabilize the mapcount vs. refcount and recheck. */ > > folio_lock_large_mapcount(folio); > > - VM_WARN_ON_ONCE(folio_large_mapcount(folio) < folio_ref_count(folio)); > > + VM_WARN_ON_ONCE(folio_large_mapcount(folio) > folio_ref_count(folio)); > > Ah, now it makes sense to me now :) > > Thanks for the quick response. > > > > > if (folio_test_large_maybe_mapped_shared(folio)) > > goto unlock; > > > > Our refcount must be at least the mapcount, that's what we want to assert. > > > > Can you test and send a fix patch if that makes it fly for you? > > Sure I'll keep the testing, I think it will just fix it, I have a few > WARN_ON_FOLIO reports all reporting mapcount is smaller than refcount. Hi David, I'm no longer seeing any warning after this, it fixed the problem well.
