In order to adapt to different vehicle densities, a path connectivity probability based cross-layer routing handoff mechanism is proposed under software defined vehicle ad hoc networks (SDVN). In this mechanism, we employ the derived path connectivity probability model to predict the density degree of a path, which will be used to decide whether to use base stations (BSs) assisted routing or V2V cross-layer routing. Particularly, in the process of performing V2V cross-layer routing, we utilize the information such as connectivity probability, transmission rate, and load to further improve the effectiveness of the proposed mechanism. Simulation results show that our proposed routing algorithm achieves significant gains in terms of delivery ratio and average delay, compared with several existing routing protocols.