[PATCH v5 1/1] userfaultfd: fix a crash in UFFDIO_MOVE when PMD is a migration entry

[PATCH v5 1/1] userfaultfd: fix a crash in UFFDIO_MOVE when PMD is a migration entry

[Suren Baghdasaryan]

When UFFDIO_MOVE encounters a migration PMD entry, it proceeds with
obtaining a folio and accessing it even though the entry is swp_entry_t.
Add the missing check and let split_huge_pmd() handle migration entries.
While at it also remove unnecessary folio check.

Fixes: adef440691ba ("userfaultfd: UFFDIO_MOVE uABI")
Reported-by: syzbot+b446dbe27035ef6bd6c2@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/68794b5c.a70a0220.693ce.0050.GAE@google.com/
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Cc: stable@vger.kernel.org
---
Applies to mm-unstable after reverting older v4 [1] version.

Changes since v4 [1]
- Removed extra folio check, per David Hildenbrand

[1] https://lore.kernel.org/all/20250806220022.926763-1-surenb@google.com/

 mm/userfaultfd.c | 17 ++++++++++-------
 1 file changed, 10 insertions(+), 7 deletions(-)

diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
index 5431c9dd7fd7..aefdf3a812a1 100644
--- a/mm/userfaultfd.c
+++ b/mm/userfaultfd.c
@@ -1826,13 +1826,16 @@ ssize_t move_pages(struct userfaultfd_ctx *ctx, unsigned long dst_start,
 			/* Check if we can move the pmd without splitting it. */
 			if (move_splits_huge_pmd(dst_addr, src_addr, src_start + len) ||
 			    !pmd_none(dst_pmdval)) {
-				struct folio *folio = pmd_folio(*src_pmd);
-
-				if (!folio || (!is_huge_zero_folio(folio) &&
-					       !PageAnonExclusive(&folio->page))) {
-					spin_unlock(ptl);
-					err = -EBUSY;
-					break;
+				/* Can be a migration entry */
+				if (pmd_present(*src_pmd)) {
+					struct folio *folio = pmd_folio(*src_pmd);
+
+					if (!is_huge_zero_folio(folio) &&
+					    !PageAnonExclusive(&folio->page)) {
+						spin_unlock(ptl);
+						err = -EBUSY;
+						break;
+					}
 				}
 
 				spin_unlock(ptl);
-- 
2.50.1.703.g449372360f-goog

Re: [PATCH v5 1/1] userfaultfd: fix a crash in UFFDIO_MOVE when PMD is a migration entry

[David Hildenbrand]

On 07.08.25 22:04, Suren Baghdasaryan wrote:
> When UFFDIO_MOVE encounters a migration PMD entry, it proceeds with
> obtaining a folio and accessing it even though the entry is swp_entry_t.
> Add the missing check and let split_huge_pmd() handle migration entries.
> While at it also remove unnecessary folio check.
> 
> Fixes: adef440691ba ("userfaultfd: UFFDIO_MOVE uABI")
> Reported-by: syzbot+b446dbe27035ef6bd6c2@syzkaller.appspotmail.com
> Closes: https://lore.kernel.org/all/68794b5c.a70a0220.693ce.0050.GAE@google.com/
> Signed-off-by: Suren Baghdasaryan <surenb@google.com>
> Reviewed-by: Peter Xu <peterx@redhat.com>
> Cc: stable@vger.kernel.org
> ---
> Applies to mm-unstable after reverting older v4 [1] version.
> 
> Changes since v4 [1]
> - Removed extra folio check, per David Hildenbrand
> 
> [1] https://lore.kernel.org/all/20250806220022.926763-1-surenb@google.com/
> 
>   mm/userfaultfd.c | 17 ++++++++++-------
>   1 file changed, 10 insertions(+), 7 deletions(-)
> 
> diff --git a/mm/userfaultfd.c b/mm/userfaultfd.c
> index 5431c9dd7fd7..aefdf3a812a1 100644
> --- a/mm/userfaultfd.c
> +++ b/mm/userfaultfd.c
> @@ -1826,13 +1826,16 @@ ssize_t move_pages(struct userfaultfd_ctx *ctx, unsigned long dst_start,
>   			/* Check if we can move the pmd without splitting it. */
>   			if (move_splits_huge_pmd(dst_addr, src_addr, src_start + len) ||
>   			    !pmd_none(dst_pmdval)) {
> -				struct folio *folio = pmd_folio(*src_pmd);
> -
> -				if (!folio || (!is_huge_zero_folio(folio) &&
> -					       !PageAnonExclusive(&folio->page))) {
> -					spin_unlock(ptl);
> -					err = -EBUSY;
> -					break;
> +				/* Can be a migration entry */
> +				if (pmd_present(*src_pmd)) {
> +					struct folio *folio = pmd_folio(*src_pmd);
> +
> +					if (!is_huge_zero_folio(folio) &&
> +					    !PageAnonExclusive(&folio->page)) {
> +						spin_unlock(ptl);
> +						err = -EBUSY;
> +						break;
> +					}

Acked-by: David Hildenbrand <david@redhat.com>

-- 
Cheers,

David / dhildenb