Prioritizing refactorings for security-critical code

摘要

It is vitally important to fix quality issues in security-critical code as they may be sources of vulnerabilities in the future. These quality issues may increase the attack surface if they are not quickly refactored. In this paper, we use the history of vulnerabilities and security bug reports along with a set of keywords to automatically identify a project's security-critical files based on its source code, bug reports, pull-request descriptions and commit messages. After identifying these security-related files, we estimate their risks using static analysis to check their coupling with other project components. Then, our approach recommends refactorings to prioritize fixing quality issues in these security-critical files to improve quality attributes and remove identified code smells. To find a trade-off between the quality issues and security-critical files, we adopted a multi-objective search strategy. We evaluated our approach on six open source projects and one industrial system to check the correctness and relevance of the refactorings targeting security critical code. The results of our survey with practitioners supports our hypothesis that quality and security need to be considered together to provide relevant refactoring recommendations.