In this session we begin with modelling the attack surface of Java deserialization, which often leads to remote code execution (RCE), by showcasing vulnerabilities we found in modern and widely used applications and frameworks. We extend existing research about risks of deserialization broadening the attack surface. After a live demo of getting a Meterpreter shell in a modern Java endpoint setup we delve into the exploitation styles for this vulnerability to lay the foundation of the first of three key takeaways for the attendees:
The first key takeaway is identification of test types that should be executed during a dynamic assessment of an application in order to find this kind of vulnerability. This includes analyzing the deserialization interface and using blackbox tests to create payloads with gadgets matching the application’s classpath to verify the RCE. Discussion extends to cover indirect deserialization interfaces that use non-binary data formats, such as XML-based interfaces, which can also act as a driver for deserialization within the application.
The next key takeaway covers the realm of static code analysis (SAST). We present code patterns security reviewers should look for when doing whitebox assessments of applications or frameworks. This is especially interesting for code offering dynamic functionality including AOP, generic mappings, reflection, interceptors, etc. - all of which have a high probability of including code that can facilitate as deserialization gadgets and thus help the attackers in exploiting deserialization vulnerabilities. In this section we present the techniques used to find the vulnerabilities within the popular frameworks showcased during the live demo at the session’s start.
Finally we conclude with tips on implementing different techniques of hardening measures for applications offering deserialisation interfaces (either direct binary deserialization interfaces or indirect XML-based ones) to give the attendees the third key takeaway: protecting applications properly. This includes ways to verify data integrity prior to deserialization and ways to properly inspect the data before it’s handled by the Java deserialization process.
--
This talk was presented by Christian Schneider & Alvaro Muñoz at the OWASP BeNeLux Day 2016.