When thinpool is suspended and sets fail_io, resume will report error
as below:
device-mapper: resume ioctl on vg-thinpool failed: Invalid argument
Thinpool also can't be removed if bio is in deferred list.
This can be easily reproduced using:
echo "offline" > /sys/block/sda/device/state
dd if=/dev/zero of=/dev/mapper/thin bs=4K count=1
dmsetup suspend /dev/mapper/pool
mkfs.ext4 /dev/mapper/thin
dmsetup resume /dev/mapper/pool
The root cause is maybe_resize_data_dev() will check fail_io and return
error before called dm_resume.
Fix this by adding FAIL mode check at the end of pool_preresume().
Fixes: da105ed5fd7e ("dm thin metadata: introduce dm_pool_abort_metadata")
Signed-off-by: Luo Meng <[email protected]>
---
drivers/md/dm-thin.c | 17 +++++++++++++----
1 file changed, 13 insertions(+), 4 deletions(-)
diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index dc271c107fb5..7a1df61ef083 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -3542,20 +3542,29 @@ static int pool_preresume(struct dm_target *ti)
*/
r = bind_control_target(pool, ti);
if (r)
- return r;
+ goto out;
r = maybe_resize_data_dev(ti, &need_commit1);
if (r)
- return r;
+ goto out;
r = maybe_resize_metadata_dev(ti, &need_commit2);
if (r)
- return r;
+ goto out;
if (need_commit1 || need_commit2)
(void) commit(pool);
- return 0;
+out:
+ /*
+ * When thinpool is PM_FAIL, it cannot be rebuilt if
+ * bio is in deferred list. Therefor need to return 0 and
+ * call pool_resume() to flush IO.
+ */
+ if (r && get_pool_mode(pool) == PM_FAIL)
+ r = 0;
+
+ return r;
}
static void pool_suspend_active_thins(struct pool *pool)
--
2.31.1
--
dm-devel mailing list
[email protected]
https://listman.redhat.com/mailman/listinfo/dm-devel
