Automatic Discovery of API-Level Vulnerabilities
Seshia, Sanjit A.
Reps, Thomas W.
Bryant, Randal E.
University of Wisconsin-Madison Department of Computer Sciences
MetadataShow full item record
A system is vulnerable to an API-level attack if its security can be compromised by invoking an allowed sequence of operations from its API. We present a formal framework to model and analyze APIs, and develop an automatic technique based upon bounded model checking to discover API-level vulnerabilities. If a vulnerability exists, our technique produces a trace of API operations demonstrating an attack. Two case studies show the efficacy of our technique. In the first study we present a novel way to analyze print f-family format-string attacks as MI-level attacks, and implement a tool to discover them automatically. In the second study, we model a subset of the IBM Common Cryptographic Architecture MI, a popular cryptographic key-management API, and automatically detect a previously known vulnerability.