Assessing Common Software Vulnerabilities in Undergraduate Computer Science Assignments

Abstract

As the demand for secure coding education grows, there is a need for improvements in how secure coding is taught and in preparing students to develop more secure software. As time in a Computer Science classroom is finite, educational efforts should be placed on targeting the most common types of vulnerabilities to better prepare students to avoid common security pitfalls in coding. Existing research in this area mainly focuses on developing vulnerability detection tools rather than analyzing the types of commonly produced vulnerabilities by students. Limited research exists in determining common student-produced vulnerabilities, and the available studies differ from the types of vulnerabilities that are researched in vulnerability detection literature. Our research works to further establish the types of vulnerabilities produced by students by using a static analysis tool on assignment code submissions in an undergraduate Programming II (CS2) course. We present our findings on what types of vulnerabilities are commonly produced by students and contrast them with what is commonly researched in the literature. We find there is little overlap between the vulnerability types reported by our study and other studies in the research area. This research has potential implications for secure coding education in a Computer Science curriculum. Further work should be done to establish the contexts in which specific vulnerability types are more likely to be produced and how to best teach students to avoid producing these vulnerabilities.