Thanks for the report.
> _check_btrfs_filesystem: filesystem on /dev/sda5 is inconsistent (see
/lkp/benchmarks/xfstests/results//check.full)
> generic/088 0s
Ran tests again unfortunately I couldn't reproduce..
---------
generic/088 0s ... 0s
generic/089 21s ... 39s
generic/090 4s ... 4s
generic/091 58s ... 58s
generic/092 1s ... 0s
generic/096 [not run] xfs_io fzero failed (old kernel/wrong fs?)
generic/098 2s ... 2s
generic/100 18s ... 18s
generic/101 3s
generic/102 8s
generic/103 2s
----------
But a possible scenario .. in systems where rcu_callback itself
could be deferred it needs rcu_barrier(). Fixed this in V2, by
restoring rcu_barrier() and then grace period wait for devices
to close.
+mount: /dev/sda6 is already mounted or /fs/scratch busy
::
+umount: /dev/sda6: not mounted
This might be different issue, when this happens generally
wipefs all scratch pool devices helps.
Thanks, Anand
