In this letter, we study the reconfigurable intelligent surfaces (RIS) aided full-duplex (FD) communication system. By jointly designing the active beamforming of two multi-antenna sources and passive beamforming of RIS, we aim to maximize the energy efficiency of the system, where extra self-interference cancellation power consumption in FD system is also considered. We divide the optimization problem into active and passive beamforming design subproblems, and adopt the alternative optimization framework to solve them iteratively. Dinkelbach's method is used to tackle the fractional objective function in active beamforming problem. Penalty method and successive convex approximation are exploited for passive beamforming design. Simulation results show the energy efficiency of our scheme outperforms other benchmarks.