rcu_find_next_bit
(cpu) = rcu_find_next_bit((rnp), 0, (mask)); \
(cpu) = rcu_find_next_bit((rnp), (cpu) + 1 - (rnp->grplo), (mask)))