Statically detecting likely buffer overflow vulnerabilities 论文

摘要

Buffer overflow attacks may be today's single most important security threat. This paper presents a new approach to mitigating buffer overflow vulnerabilities by detecting likely vulnerabilities through an analysis of the program source code. Our approach exploits information provided in semantic comments and uses lightweight and efficient static analyses. This paper describes an implementation of our approach that extends the LCLint annotation-assisted static checking tool. Our tool is as fast as a compiler and nearly as easy to use. We present experience using our approach to detect buffer overflow vulnerabilities in two security-sensitive programs.