Andrew Case (Volexity)
Arghya Das
Seung-Jong Park
Ram Ramanujam
Golden Richard III, Ph.D. (Louisiana State University)

Abstract

Memory forensics is now a standard component of digital forensic investigations and incident response handling, since memory forensic techniques are quite effective in uncovering artifacts that might be missed by traditional storage forensics or live analysis techniques. Because of the crucial role that memory forensics plays in investigations and because of the increasing use of automation of memory forensics techniques, it is imperative that these tools be resilient to memory smear and deliberate tampering. Without robust algorithms, malware may go undetected, frameworks may crash when attempting to process memory samples, and automation of memory forensics techniques is difficult. In this paper we present Gaslight, a powerful and flexible fuzz-testing architecture for stress-testing both open and closed-source memory forensics frameworks. Gaslight automatically targets critical code paths that process memory samples and mutates samples in an efficient way to reveal implementation errors. In experiments we conducted against several popular memory forensics frameworks, Gaslight revealed a number of critical previously undiscovered bugs.