LigRE: Reverse-engineering of control and data flow models for black-box XSS detection

摘要

Fuzz testing consists of automatically generating and sending malicious inputs to an application in order to hopefully trigger a vulnerability. In order to be efficient, the fuzzing should answer questions such as: Where to send a malicious value? Where to observe its effects? How to position the system in such states? Answering such questions is a matter of understanding precisely enough the application. Reverseengineering is a possible way to gain this knowledge, especially in a black-box harness. In fact, given the complexity of modern web applications, automated black-box scanners alternatively reverse-engineer and fuzz web applications to detect vulnerabilities. We present an approach, named as LigRE, which improves the reverse engineering to guide the fuzzing. We adapt a method to automatically learn a control flow model of web applications, and annotate this model with inferred data flows. Afterwards, we generate slices of the model for guiding the scope of a fuzzer. Empirical experiments show that LigRE increases detection capabilities of Cross Site Scripting (XSS), a particular case of web command injection vulnerabilities.