Compressed writeback is currently charged as F2FS_WB_CP_DATA through
WB_DATA_TYPE(..., fio->compressed_page). That classification keeps
compressed cluster writeback under the checkpoint-guaranteed path for
POR safety after fsync/sync.
When such a compressed write bio fails,
f2fs_compress_write_end_io() only records -EIO in the inode mapping.
Ordinary F2FS_WB_CP_DATA writeback also stops checkpoint on writeback
failure, but compressed writeback does not. Checkpoint can therefore
continue after a failed checkpoint-guaranteed compressed write and later
persist metadata that points to compressed data blocks whose writeback
failed.
Stop checkpoint on failed compressed write IO the same way as other
F2FS_WB_CP_DATA writeback.
Fixes: 4c8ff7095bef ("f2fs: support data compression")
Cc: stable@kernel.org
Signed-off-by: Wenjie Qi <qiwenjie@xiaomi.com>
---
v2:
- rewrite the changelog around the existing checkpoint-guaranteed
compressed writeback semantics
- no code changes
fs/f2fs/compress.c | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/fs/f2fs/compress.c b/fs/f2fs/compress.c
index caf522d667d6..9b1501004456 100644
--- a/fs/f2fs/compress.c
+++ b/fs/f2fs/compress.c
@@ -1488,8 +1488,11 @@ void f2fs_compress_write_end_io(struct bio *bio, struct folio *folio)
f2fs_is_compressed_page(folio));
int i;
- if (unlikely(bio->bi_status != BLK_STS_OK))
+ if (unlikely(bio->bi_status != BLK_STS_OK)) {
mapping_set_error(cic->inode->i_mapping, -EIO);
+ if (type == F2FS_WB_CP_DATA)
+ f2fs_stop_checkpoint(sbi, true, STOP_CP_REASON_WRITE_FAIL);
+ }
f2fs_compress_free_page(page);
--
2.43.0