Static Buffer Overflow Detection for C/C++ Source Code Based on Abstract Syntax Tree

Buffer overflow vulnerability is currently one of the most serious software security vulnerabilities. Taking effective methods to detect buffer overflow vulnerabilities is of great significance to improve the robustness and security of software. After analyzing the principle of buffer overflow and various buffer overflow vulnerabilities, as well as comparing static with dynamic detection techniques, this paper presents a static detection method for buffer overflow based on abstract syntax tree, especially for vulnerabilities caused by string accessing of C source code. The proposed method first generates an abstract syntax tree for the source code by using the GCC compiler, and collects the characteristics of vulnerability by analyzing AST nodes. For the nodes associated with the buffer overflow, the method further gives a formal description of that characteristic so as to generate security attributed and constraint rules. By traversing the abstract syntax tree, the proposed method finally determines whether buffer overflow vulnerability exists in the source code. Experiments with 12 C programs were carried out on the Windows platform, and each program has been discovered at least one typical buffer overflow vulnerability. The experimental results highlight the feasibility and effectiveness of the proposed method, especially in the detecting out of range for a pointer, an array, and a structure object, as well as the buffer overflow vulnerability caused by some C library function calls.