Assisted Identification of Mode of Operation in Binary Code with Dynamic Data Flow Slicing

摘要

Verification of software security properties, when conducted at the binary code level, is a difficult and cumbersome task. This paper is focused on the reverse engineering task that needs to be performed prior to any thorough analysis. A previous line of work has been dedicated to the identification of cryptographic primitives. Relying on the techniques that have been proposed, we devise a semi-automated solution to identify modes of operation. Our solution produces a concise representation of the data transfers occurring within a cryptographic scheme. Inspired by program slicing techniques, we extract from a dynamic data flow a slice defined as the smallest subgraph that is distance preserving for the set of cryptographic parameters. We apply our solution to several modes of operation including CBC, CTR, HMAC and OCB. For each of them, we successfully obtain a complete and readable representation. Moreover, we show with an example that our solution can be applied on non standard schemes to quickly discover security flaw.