It is well-known that any finite $p$-group in which all its abelia  subgroups are cyclic is either a cyclic group or a generalized quaterniin group. 

What can be said about $p$-groups in which every normal abelian subgroup is cyclic?