Buffer Overflow Vulnerability Detection based on Format-Matching on Source Level

Buffer overflow has become the most common software vulnerability, which seriously restricts the development of the software industry. It's very essential t o find out an effective method to detect this kind of software bugs accurately. In this paper, we design an improved buffer overflow detection system. At first, our system preprocesses the source code to add some auxiliary detection symbols. Then, it scans the source code by a static detector, which uses the identifier for auxiliary detection and combines with a dynamic detection method to improve the recognition accuracy and detection capability. Finally, we make a comparison between our system and the original detection system. To assess the usefulness of this approach, several experiments are performed on a simulation system, and we can draw a conclusion that our system performs better than other detection software. The method proposed in this paper is of the important application value and can improve detection accuracy.