In zswap_writeback_entry(), after we get a folio from
__read_swap_cache_async(), we grab the tree lock again to check that the
swap entry was not invalidated and recycled. If it was, we delete the
folio we just added to the swap cache and exit.
However, __read_swap_cache_async() returns the folio locked when it is
newly allocated, which is always true for this path, and the folio is
ref'd. Make sure to unlock and put the folio before returning.
This was discovered by code inspection, probably because this path
handles a race condition that should not happen often, and the bug would
not crash the system, it will only strand the folio indefinitely.
Fixes: 04fc7816089c ("mm: fix zswap writeback race condition")
Cc: [email protected]
Signed-off-by: Yosry Ahmed <[email protected]>
---
mm/zswap.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/mm/zswap.c b/mm/zswap.c
index 8f4a7efc2bdae..00e90b9b5417d 100644
--- a/mm/zswap.c
+++ b/mm/zswap.c
@@ -1448,6 +1448,8 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
if (zswap_rb_search(&tree->rbroot, swp_offset(entry->swpentry)) != entry) {
spin_unlock(&tree->lock);
delete_from_swap_cache(folio);
+ folio_unlock(folio);
+ folio_put(folio);
return -ENOMEM;
}
spin_unlock(&tree->lock);
--
2.43.0.429.g432eaa2c6b-goog
On Thu, Jan 25, 2024 at 08:51:27AM +0000, Yosry Ahmed wrote:
> In zswap_writeback_entry(), after we get a folio from
> __read_swap_cache_async(), we grab the tree lock again to check that the
> swap entry was not invalidated and recycled. If it was, we delete the
> folio we just added to the swap cache and exit.
>
> However, __read_swap_cache_async() returns the folio locked when it is
> newly allocated, which is always true for this path, and the folio is
> ref'd. Make sure to unlock and put the folio before returning.
>
> This was discovered by code inspection, probably because this path
> handles a race condition that should not happen often, and the bug would
> not crash the system, it will only strand the folio indefinitely.
>
> Fixes: 04fc7816089c ("mm: fix zswap writeback race condition")
> Cc: [email protected]
> Signed-off-by: Yosry Ahmed <[email protected]>
Ouch, good catch.
Acked-by: Johannes Weiner <[email protected]>
On Thu, Jan 25, 2024 at 12:51 AM Yosry Ahmed <[email protected]> wrote:
>
> In zswap_writeback_entry(), after we get a folio from
> __read_swap_cache_async(), we grab the tree lock again to check that the
> swap entry was not invalidated and recycled. If it was, we delete the
> folio we just added to the swap cache and exit.
>
> However, __read_swap_cache_async() returns the folio locked when it is
> newly allocated, which is always true for this path, and the folio is
> ref'd. Make sure to unlock and put the folio before returning.
>
> This was discovered by code inspection, probably because this path
> handles a race condition that should not happen often, and the bug would
> not crash the system, it will only strand the folio indefinitely.
>
> Fixes: 04fc7816089c ("mm: fix zswap writeback race condition")
> Cc: [email protected]
> Signed-off-by: Yosry Ahmed <[email protected]>
> ---
> mm/zswap.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/mm/zswap.c b/mm/zswap.c
> index 8f4a7efc2bdae..00e90b9b5417d 100644
> --- a/mm/zswap.c
> +++ b/mm/zswap.c
> @@ -1448,6 +1448,8 @@ static int zswap_writeback_entry(struct zswap_entry *entry,
> if (zswap_rb_search(&tree->rbroot, swp_offset(entry->swpentry)) != entry) {
> spin_unlock(&tree->lock);
> delete_from_swap_cache(folio);
> + folio_unlock(folio);
> + folio_put(folio);
> return -ENOMEM;
> }
> spin_unlock(&tree->lock);
> --
> 2.43.0.429.g432eaa2c6b-goog
>
Oof. Yeah this is probably rare IRL (that looks like a very specific
race condition), and the symptoms are rather subtle (no kernel crash).
LGTM.
Reviewed-by: Nhat Pham <[email protected]>