Formal method provides a theoretical tool for security protocol analysis, but the theoretical security is not equivalent to the actual security. A verified protocol standard may not meet the required security properties when converted into a concrete program. Hence, a formal verification method for detecting semantic logic errors in security protocol code is proposed. By automatically abstracting the C source code of the protocol into Pi calculus model, protocol security properties are verified based on the Pi calculus. Finally, the correctness of the scheme transformation is proved and the validity of the method is verified by a Kerberos protocol instance code. C2P tools implemented can help protocol developers to detect semantic logic errors in code.