Say the PK of table B is a FK, which references the PK of table A. (and assume EntityA and EntityB are mapped to table A and table B respectively)

If EntitityA and EntityB are persisted (and PK is generated for EntityA) the provider should pick up the PK from EntityA and use it for EntityB as well.

But EclipseLink needs two separate PKs to be generated. That means we have to generate another PK which is definitely equal to the PK of the former according to the relationship.

Why EclipseLink needs duplicate information? Am I missing anything, please explain?

I don't know anything about EclipseLink or why you post a question about it in a forum about Hibernate. In any case, the Hibernate documentation has an example how to do this with standard JPA annotations. See the example with a @OneToOne mapping in http://docs.jboss.org/hibernate/stable/ ... identifier