This paper deals with the practical precoding design for a two-hop downlink with MIMO amplify-and-forward relaying. A novel scheme of codebook-based joint precoding at both the base station and the relay station is first presented, in which a distributed codeword selection is proposed to concurrently choose two joint precoders such that the feedback delay is considerably reduced. Then, the joint codebook design in such a joint precoding system is analyzed, revealing that independent codebook designs at the base station and relay station using the conventional Grassmannian subspace packing method is able to guarantee that the overall performance improves with either size of two codebooks. The excellent performances of the proposed scheme, in terms of both the sum rate and the BER, are finally confirmed with computational simulations.