Rigid-body reorientation under attitude constraints has been studied extensively, but few of them take into account the actuator faults which occur frequently in practice. Since the actuator faults may lead to performance degradation or even system instability, a control scheme with fault-tolerant ability is important to enhance safety and reliability. However, it is still an open problem to solve attitude constraints and actuator faults simultaneously in condition of model uncertainties. In this paper, a control algorithm combining the potential function and adaptive fault-tolerant control method is proposed to tackle this challenging problem. We parameterize the kinematics using the singularity-free unit-quaternion representation, and assume perfect measurement of full-state variables. A potential function is first developed based on the geometric properties of constrained zones, which can guide the rigid body to the desired orientation without causing unwinding, while ensuring the satisfaction of complex forbidden and mandatory attitude constraints. Based on the judiciously constructed potential function, an adaptive fault-tolerant control scheme with saturated virtual control is then proposed in the framework of backstepping. Lyapunov stability analysis shows that the derived controller can guarantee asymptotic convergence of both the gradient-related vector and angular rate to zero, despite the presence of actuator faults, unknown inertia, and external disturbances. This indicates that the proposed controller is capable of achieving safe, reliable, and high-precision attitude reorientation maneuvers under complex attitude constraints and adverse operational conditions. Finally, numerical simulations demonstrate the effectiveness of the proposed controller.