[PATCH] mm: mlock: avoid folio_within_range() on KSM pages

[PATCH] mm: mlock: avoid folio_within_range() on KSM pages

[Hugh Dickins]

Since mm-hotfixes-stable commit dc68badcede4 ("mm: mlock: update
mlock_pte_range to handle large folio") I've just occasionally seen
VM_WARN_ON_FOLIO(folio_test_ksm) warnings from folio_within_range(),
in a splurge after testing with KSM hyperactive.

folio_referenced_one()'s use of folio_within_vma() is safe because
it checks folio_test_large() first; but allow_mlock_munlock() needs
to do the same to avoid those warnings (or check !folio_test_ksm()
itself? or move either check into folio_within_range()? hard to tell
without more examples of its use).

Fixes: dc68badcede4 ("mm: mlock: update mlock_pte_range to handle large folio")
Signed-off-by: Hugh Dickins <hughd@google.com>
---
 mm/mlock.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/mm/mlock.c b/mm/mlock.c
index aa44456200e3..086546ac5766 100644
--- a/mm/mlock.c
+++ b/mm/mlock.c
@@ -346,6 +346,10 @@ static inline bool allow_mlock_munlock(struct folio *folio,
 	if (!(vma->vm_flags & VM_LOCKED))
 		return true;
 
+	/* folio_within_range() cannot take KSM, but any small folio is OK */
+	if (!folio_test_large(folio))
+		return true;
+
 	/* folio not in range [start, end), skip mlock */
 	if (!folio_within_range(folio, vma, start, end))
 		return false;
-- 
2.35.3

Re: [PATCH] mm: mlock: avoid folio_within_range() on KSM pages

[Yin Fengwei]

Hi Huge,

On 10/24/23 14:38, Hugh Dickins wrote:
> Since mm-hotfixes-stable commit dc68badcede4 ("mm: mlock: update
> mlock_pte_range to handle large folio") I've just occasionally seen
> VM_WARN_ON_FOLIO(folio_test_ksm) warnings from folio_within_range(),
> in a splurge after testing with KSM hyperactive.
> 
> folio_referenced_one()'s use of folio_within_vma() is safe because
> it checks folio_test_large() first; but allow_mlock_munlock() needs
> to do the same to avoid those warnings (or check !folio_test_ksm()
> itself? or move either check into folio_within_range()? hard to tell
> without more examples of its use).
Checking folio_test_large() here looks fine to me now. If KSM could support
large folio in the future (Not sure whether this will happen in the future),
we could revise.

> 
> Fixes: dc68badcede4 ("mm: mlock: update mlock_pte_range to handle large folio")
> Signed-off-by: Hugh Dickins <hughd@google.com>
Thanks a lot for catching this issue and fixing it.

Reviewed-by: Yin Fengwei <fengwei.yin@intel.com>