[Devel] [PATCH vz9 v3 1/2] dm-qcow2: fix qcow2 decompression

Alexander Atanasov alexander.atanasov at virtuozzo.com
Wed Feb 19 13:47:31 MSK 2025


Two problems - using wrong size type:

- process_compressed_read used blk_status_t for ret which is u8
so it can not process negative error codes and larger sizes
properly - fix this by changing it to int

- decompression fails for images with size larger than 2GB
due to wrong return type of parse_compressed_l2 - the calculations
are performed in properly in u64 type but on return result is truncated
Fix this by changing the type to loff_t

https://virtuozzo.atlassian.net/browse/VSTOR-98868
Signed-off-by: Alexander Atanasov <alexander.atanasov at virtuozzo.com>
---
 drivers/md/dm-qcow2-map.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

patch sent in series as next is dependant

diff --git a/drivers/md/dm-qcow2-map.c b/drivers/md/dm-qcow2-map.c
index 54bb8bd713be..ccc8d4484049 100644
--- a/drivers/md/dm-qcow2-map.c
+++ b/drivers/md/dm-qcow2-map.c
@@ -1729,7 +1729,7 @@ static loff_t parse_l1(struct qcow2 *qcow2, struct qcow2_map *map,
 	return ret;
 }
 
-static int parse_compressed_l2(struct qcow2 *qcow2, struct qcow2_map *map,
+static loff_t parse_compressed_l2(struct qcow2 *qcow2, struct qcow2_map *map,
 			       struct qio **qio, bool write, u64 entry)
 {
 	u8 offset_bits = 62 - (qcow2->hdr.cluster_bits - 8);
@@ -3660,7 +3660,7 @@ static void process_compressed_read(struct qcow2 *qcow2, struct list_head *read_
 {
 	struct qcow2_bvec *qvec;
 	struct qio_ext *ext;
-	blk_status_t ret;
+	int ret;
 	void *buf = NULL, *arg;
 	struct qio *qio;
 	bool for_cow;
@@ -3698,7 +3698,7 @@ static void process_compressed_read(struct qcow2 *qcow2, struct list_head *read_
 
 		ret = extract_one_compressed(qcow2, buf, qvec,
 				    ext->zdata_off, qio->ret, arg);
-		if (ret)
+		if (ret <= 0)
 			goto err;
 
 		for_cow = op_is_write(qio->bi_op);
-- 
2.43.0



More information about the Devel mailing list