If we get a folio from swap_cache_get_folio() successfully but encounter
a failure before the folio is locked, we will unlock the folio which was
not previously locked.
Put the folio and set it to NULL when a failure occurs before the folio
is locked to fix the issue.

Fixes: 058313515d5aa ("mm: shmem: fix potential data corruption during shmem swapin")
Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
---
 mm/shmem.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/mm/shmem.c b/mm/shmem.c
index 99327c30507c..980fa15f393e 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -2335,6 +2335,8 @@ static int shmem_swapin_folio(struct inode *inode, pgoff_t index,
 		 */
 		split_order = shmem_split_large_entry(inode, index, swap, gfp);
 		if (split_order < 0) {
+			folio_put(folio);
+			folio = NULL;
 			error = split_order;
 			goto failed;
 		}
-- 
2.30.0

On Thu, May 15, 2025 at 2:54 PM Kemeng Shi <shikemeng@huaweicloud.com> wrote:
>
> If we get a folio from swap_cache_get_folio() successfully but encounter
> a failure before the folio is locked, we will unlock the folio which was
> not previously locked.
> Put the folio and set it to NULL when a failure occurs before the folio
> is locked to fix the issue.
>
> Fixes: 058313515d5aa ("mm: shmem: fix potential data corruption during shmem swapin")
> Signed-off-by: Kemeng Shi <shikemeng@huaweicloud.com>
> Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
> ---
>  mm/shmem.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 99327c30507c..980fa15f393e 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -2335,6 +2335,8 @@ static int shmem_swapin_folio(struct inode *inode, pgoff_t index,
>                  */
>                 split_order = shmem_split_large_entry(inode, index, swap, gfp);
>                 if (split_order < 0) {
> +                       folio_put(folio);
> +                       folio = NULL;
>                         error = split_order;
>                         goto failed;
>                 }

Nice fix, I also noticed this and included the same fix in the swap
table series and forgot to split that out. We should merge this clean
fix first:

Reviewed-by: Kairui Song <kasong@tencent.com>