A stochastic model predictive control framework is presented in this paper for nonlinear affine system with inputs constraints. The proposed MPC can achieve asymptotic stability in probability while ensuring safe guarantee. We first introduce the concept of stochastic control Lyapunov-barrier function and provide a method to construct control Lyapunov-barrier function by combining control Lyapunov function and control barrier functions. The unconstrained CLF is obtained from its corresponding semi-linear system through dynamic feedback linearization. Based on the constructed CLBF, we utilize MPC framework to deal with constraints and ensure feasibility. The proposed CLBF based MPC is validated via a wheeled mobile robot example.