A survey of automated root cause analysis of software vulnerability

Abstract

In recent years, many researches on automatic exploit generation and automatic patch techniques have been published. Typically, in the CGC (Cyber Grand Challenge) competition hosted by DARPA, a hacking competition was held between machines to find vulnerabilities, automatically generate exploits and automatically patch them. In the CGC competition, they implemented themselves to work on their own platform, allowing only 7 system calls. However, in a real environment, there are much more system calls and the software works on complicated architecture. In order to effectively apply the vulnerability detection and patching process to the actual real environment, it is necessary to identify the point causing the vulnerability. In this paper, we introduce a method to analyze root cause of vulnerabilities divided into three parts, fault localization, code pattern similarity analysis, and taint analysis.