The loop through all waiting locks in recover_waiters can potentially be
long, so we should schedule explicitly.
Signed-off-by: NDavid Teigland <teigland@redhat.com>
Signed-off-by: NSteven Whitehouse <swhiteho@redhat.com>