[PATCH v3] mm/shmem, swap: fix race of truncate and swap entry split

[Kairui Song <ryncsn@gmail.com>]

From: Kairui Song <kasong@tencent.com>

The helper for shmem swap freeing is not handling the order of swap
entries correctly. It uses xa_cmpxchg_irq to erase the swap entry, but
it gets the entry order before that using xa_get_order without lock
protection, and it may get an outdated order value if the entry is split
or changed in other ways after the xa_get_order and before the
xa_cmpxchg_irq.

And besides, the order could grow and be larger than expected, and cause
truncation to erase data beyond the end border. For example, if the
target entry and following entries are swapped in or freed, then a large
folio was added in place and swapped out, using the same entry, the
xa_cmpxchg_irq will still succeed, it's very unlikely to happen though.

To fix that, open code the Xarray cmpxchg and put the order retrieval
and value checking in the same critical section. Also, ensure the order
won't exceed the end border, skip it if the entry goes across the
border.

Skipping large swap entries crosses the end border is safe here.
Shmem truncate iterates the range twice, in the first iteration,
find_lock_entries already filtered such entries, and shmem will
swapin the entries that cross the end border and partially truncate the
folio (split the folio or at least zero part of it). So in the second
loop here, if we see a swap entry that crosses the end order, it must
at least have its content erased already.

I observed random swapoff hangs and kernel panics when stress testing
ZSWAP with shmem. After applying this patch, all problems are gone.

Fixes: 809bc86517cc ("mm: shmem: support large folio swap out")
Cc: stable@vger.kernel.org
Signed-off-by: Kairui Song <kasong@tencent.com>
---
Changes in v3:
- Rebased on top of mainline.
- Fix nr_pages calculation [ Baolin Wang ]
- Link to v2: https://lore.kernel.org/r/20260119-shmem-swap-fix-v2-1-034c946fd393@tencent.com

Changes in v2:
- Fix a potential retry loop issue and improvement to code style thanks
  to Baoling Wang. I didn't split the change into two patches because a
  separate patch doesn't stand well as a fix.
- Link to v1: https://lore.kernel.org/r/20260112-shmem-swap-fix-v1-1-0f347f4f6952@tencent.com
---
 mm/shmem.c | 45 ++++++++++++++++++++++++++++++++++-----------
 1 file changed, 34 insertions(+), 11 deletions(-)

diff --git a/mm/shmem.c b/mm/shmem.c
index ec6c01378e9d..6c3485d24d66 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -962,17 +962,29 @@ static void shmem_delete_from_page_cache(struct folio *folio, void *radswap)
  * being freed).
  */
 static long shmem_free_swap(struct address_space *mapping,
-			    pgoff_t index, void *radswap)
+			    pgoff_t index, pgoff_t end, void *radswap)
 {
-	int order = xa_get_order(&mapping->i_pages, index);
-	void *old;
+	XA_STATE(xas, &mapping->i_pages, index);
+	unsigned int nr_pages = 0;
+	pgoff_t base;
+	void *entry;
 
-	old = xa_cmpxchg_irq(&mapping->i_pages, index, radswap, NULL, 0);
-	if (old != radswap)
-		return 0;
-	free_swap_and_cache_nr(radix_to_swp_entry(radswap), 1 << order);
+	xas_lock_irq(&xas);
+	entry = xas_load(&xas);
+	if (entry == radswap) {
+		nr_pages = 1 << xas_get_order(&xas);
+		base = round_down(xas.xa_index, nr_pages);
+		if (base < index || base + nr_pages - 1 > end)
+			nr_pages = 0;
+		else
+			xas_store(&xas, NULL);
+	}
+	xas_unlock_irq(&xas);
+
+	if (nr_pages)
+		free_swap_and_cache_nr(radix_to_swp_entry(radswap), nr_pages);
 
-	return 1 << order;
+	return nr_pages;
 }
 
 /*
@@ -1124,8 +1136,8 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, uoff_t lend,
 			if (xa_is_value(folio)) {
 				if (unfalloc)
 					continue;
-				nr_swaps_freed += shmem_free_swap(mapping,
-							indices[i], folio);
+				nr_swaps_freed += shmem_free_swap(mapping, indices[i],
+								  end - 1, folio);
 				continue;
 			}
 
@@ -1191,12 +1203,23 @@ static void shmem_undo_range(struct inode *inode, loff_t lstart, uoff_t lend,
 			folio = fbatch.folios[i];
 
 			if (xa_is_value(folio)) {
+				int order;
 				long swaps_freed;
 
 				if (unfalloc)
 					continue;
-				swaps_freed = shmem_free_swap(mapping, indices[i], folio);
+				swaps_freed = shmem_free_swap(mapping, indices[i],
+							      end - 1, folio);
 				if (!swaps_freed) {
+					/*
+					 * If found a large swap entry cross the end border,
+					 * skip it as the truncate_inode_partial_folio above
+					 * should have at least zerod its content once.
+					 */
+					order = shmem_confirm_swap(mapping, indices[i],
+								   radix_to_swp_entry(folio));
+					if (order > 0 && indices[i] + (1 << order) > end)
+						continue;
 					/* Swap was replaced by page: retry */
 					index = indices[i];
 					break;

---
base-commit: 24d479d26b25bce5faea3ddd9fa8f3a6c3129ea7
change-id: 20260111-shmem-swap-fix-8d0e20a14b5d

Best regards,
-- 
Kairui Song <kasong@tencent.com>