A Return-Value-Unchecked Vulnerability Detection Method Based on Property Graph

Abstract

Traditional static analysis methods for binary software vulnerability detection are used only to make use of a single aspect of the target software, so it is difficult to obtain the hidden global properties and relationships which leads to low detection accuracy and high rate of false positives. To improve the effectiveness of the binary software static vulnerability detection, this paper proposes a fusion method for binary software vulnerability detection which first represents the binary software as a single property graph and then the vulnerability is modeled and detected based on this property graph. Because property graph includes integrated information such as the relations between function calls, control flow, data flow relationship and so on, researchers can model vulnerability more easily and accurately. It can detect unknown vulnerabilities accurately and effi-ciently. The experiments of prototype system show that this method can effectively detect Return-Value-Unchecked Vulnerability in binary software.