Daniel Lehmann, University of Stuttgart; Johannes Kinder, Bundeswehr University Munich; Michael Pradel, University of Stuttgart
WebAssembly is an increasingly popular compilation target designed to run code in browsers and on other platforms safely and securely, by strictly separating code and data, enforcing types, and limiting indirect control flow. Still, vulnerabilities in memory-unsafe source languages can translate to vulnerabilities in WebAssembly binaries. In this paper, we analyze to what extent vulnerabilities are exploitable in WebAssembly binaries, and how this compares to native code. We find that many classic vulnerabilities which, due to common mitigations, are no longer exploitable in native binaries, are completely exposed in WebAssembly. Moreover, WebAssembly enables unique attacks, such as overwriting supposedly constant data or manipulating the heap using a stack overflow. We present a set of attack primitives that enable an attacker (i) to write arbitrary memory, (ii) to overwrite sensitive data, and (iii) to trigger unexpected behavior by diverting control flow or manipulating the host environment. We provide a set of vulnerable proof-of-concept applications along with complete end-to-end exploits, which cover three WebAssembly platforms. An empirical risk assessment on real-world binaries and SPEC CPU programs compiled to WebAssembly shows that our attack primitives are likely to be feasible in practice. Overall, our findings show a perhaps surprising lack of binary security in WebAssembly. We discuss potential protection mechanisms to mitigate the resulting risks.
Open Access Media
USENIX is committed to Open Access to the research presented at our events. Papers and proceedings are freely available to everyone once the event begins. Any video, audio, and/or slides that are posted after the event are also free and open to everyone. Support USENIX and our commitment to Open Access.