Cyber-physical systems integrate cyber capabilities (e.g., communications and computing) with physical devices (e.g., sensors, actuators and control processing units). Many of these systems support safety-critical applications such as electric power grids, water distribution systems and transportation systems. Failures of these systems can cause irreparable damage to equipment and injury or death to humans. While most of the efforts to protect the systems have focused on reliability, there are urgent concerns regarding malicious attacks. Trusted computing is a security paradigm that enables platforms to enforce the integrity of execution targets (code and data). However, protection under this paradigm is restricted to static threats.This chapter proposes a dynamic framework that addresses runtime integrity threats that target software programs in cyber-physical systems. It is well known that the attack surface of a multi-functional program (Swiss-army knife) can be much larger than the sum of the surfaces of its single-function components (e.g., the composition of programs that are secure in isolation is not necessarily secure). The proposed framework addresses this issue using calibration techniques that constrain the functionality of programs to the strict specifications of the cyber-physical application, thus steering execution flow away from the attack surface. Integrity is assured by verifying the calibration, while the burden of validation rests with system designers. The effectiveness of the approach is demonstrated by presenting a prototype for call integrity.