In this paper, a two-loop control strategy is proposed for the fast and accurate contour tracking of the planar biaxial servo systems with typical uncertainties and physical constraints. In the inner loop, a constrained adaptive robust contour tracking controller is designed and implemented in high sampling rate to minimize the contour tracking error of the system computed with respect to the global task frame in the presence of different types of disturbances, uncertainties and physical constraints. In the outer loop, an efficient feedrate optimization algorithm is synthesized and implemented online at low sampling rate to generate the replanned feedrate profile so that the total time for the tracking of the entire contour is minimized while without violating physical constraints. It is theoretically shown that the resulting closed-loop system can track any feasible desired contours with a guaranteed tracking time and steady-state contour tracking accuracy without violating the physical constraints. The proposed controller is also implemented on a linear motor driven industrial biaxial gantry system. Experimental results confirm that the proposed approach can indeed achieve faster tracking speed without any compromise on the achievable contour tracking accuracy and constraint violations when compared to the existing adaptive robust contour tracking algorithms.