In this paper, an iterative design of feedforward and feedback controller is proposed for a building temperature regulation application. An iterative learning control (ILC) strategy is utilized to eliminate the influence of repetitive disturbance and an iteratively tuned feedback controller is applied to reduce the influence of non-repetitive disturbance components. With constraints from input saturation, the controller design is formulated into a constrained optimization problem. To reduce the complexity of the problem, an alternating procedure is proposed to solve the optimization problem, separating the original problem into two parts, namely, constrained ILC problem and iterative feedback tuning problem. The proposed algorithm is demonstrated by simulations on a four-room testbed system.