Remarks on global hypoellipticity