blockdev: fix drive-backup transaction endless drained section

drive_backup_prepare() does bdrv_drained_begin() in hope that
bdrv_drained_end() will be called in drive_backup_clean(). Still we
need to set state->bs for this to work. That's done too late: a lot of
failure paths in drive_backup_prepare() miss setting state->bs. Fix
that.

Fixes: 2288ccfac96281c316db942d10e3f921c1373064
Fixes: https://gitlab.com/qemu-project/qemu/-/issues/399
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20210608171852.250775-1-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

diff --git a/blockdev.c b/blockdev.c
index f08192dedace8244866c15117d38f2777a7e4c2d..094c0859624cc4157fab1c84e821541f1d9d9a75 100644 (file)
--- a/blockdev.c
+++ b/blockdev.c
@@ -1714,6 +1714,7 @@ static void drive_backup_prepare(BlkActionState *common, Error **errp)
     aio_context = bdrv_get_aio_context(bs);
     aio_context_acquire(aio_context);
 
+    state->bs = bs;
     /* Paired with .clean() */
     bdrv_drained_begin(bs);
 
@@ -1813,8 +1814,6 @@ static void drive_backup_prepare(BlkActionState *common, Error **errp)
         }
     }
 
-    state->bs = bs;
-
     state->job = do_backup_common(qapi_DriveBackup_base(backup),
                                   bs, target_bs, aio_context,
                                   common->block_job_txn, errp);