Re: [PATCH v2 1/4] mm/migrate_device: avoid leaking folio references during THP splitting
From: Kiryl Shutsemau (Meta)
Date: Tue Nov 17 2026 - 05:34:13 EST
On Mon, Nov 16, 2026 at 07:44:42PM +0100, David Hildenbrand (Arm) wrote:
> On 11/16/26 07:34, Usama Arif wrote:
> > migrate_vma_split_unmapped_folio() takes an extra folio reference because
> > split_huge_pmd_address(..., true) drops one when it splits a PMD.
> >
> > However, migrate_vma only holds mmap_lock for reading. A concurrent
> > MADV_DONTNEED can clear the migration PMD, or split it into PTE migration
> > entries, before the split helper takes the PMD lock. The helper then does
> > nothing and leaves the extra reference behind. folio_split_unmapped()
> > fails with -EAGAIN due to the unexpected reference, and the reference is
> > never dropped, so the source THP is leaked.
> >
> > If a THP is faulted in at the same address inside that window it is
> > worse: freeze = true replaces the new folio's mappings with migration
> > entries that nothing will ever remove, and drops a reference the caller
> > never took.
> >
> > An existing migration PMD is expanded into PTE migration entries even
> > with freeze = false. Use that mode and remove the extra folio_get(). The
> > reference already held by migration keeps the source folio alive whether
> > or not its PMD migration entry still exists.
> >
> > Reported-by: Zi Yan <ziy@xxxxxxxxxx>
> > Closes: https://lore.kernel.org/all/DLZ48KMS7YG8.3BTM0UUU8CTU2@xxxxxxxxxx/
> > Fixes: 4265d67e405a ("mm/migrate_device: add THP splitting during migration")
> > Cc: stable@xxxxxxxxxxxxxxx
> > Signed-off-by: Usama Arif <usama.arif@xxxxxxxxx>
> > ---
> > mm/migrate_device.c | 7 +------
> > 1 file changed, 1 insertion(+), 6 deletions(-)
> >
> > diff --git a/mm/migrate_device.c b/mm/migrate_device.c
> > index 9e19fadfa65c6..3286bb8a4b73e 100644
> > --- a/mm/migrate_device.c
> > +++ b/mm/migrate_device.c
> > @@ -920,12 +920,7 @@ static int migrate_vma_split_unmapped_folio(struct migrate_vma *migrate,
> > unsigned long flags;
> > int ret = 0;
> >
> > - /*
> > - * take a reference, since split_huge_pmd_address() with freeze = true
> > - * drops a reference at the end.
> > - */
> > - folio_get(folio);
> > - split_huge_pmd_address(migrate->vma, addr, true);
> > + split_huge_pmd_address(migrate->vma, addr, false);
> > ret = folio_split_unmapped(folio, 0);
> > if (ret)
> > return ret;
>
> Oh, that means we can remove the parameter from split_huge_pmd_address no? Best
> as a cleanup patch once this is in for-next-fixes.
Yeah, that's would be great.
> Reviewed-by: David Hildenbrand (Arm) <david@xxxxxxxxxx>
The patch looks good to me too:
Reviewed-by: Kiryl Shutsemau (Meta) <kas@xxxxxxxxxx>
--
Kiryl Shutsemau / Kirill A. Shutemov