Design of an Automatically Generated Retargetable Decompiler

Design of an Automatically Generated Retargetable Decompiler

Paper in proceedings outside WoS and Scopus

Accurate program decompilation is one of the most difficult tasks of reverse engineering. Currently, there exist several single-purpose decompilers targeted on a particular platform (e.g. Intel x86 architecture and Microsoft Windows OS) and on a particular language. These tools are always hand-written by the author from scratch. This paper presents a concept of a retargetable reverse compiler (i.e. a decompiler). This tool translates platform-specific binary applications into a high-level language (HLL) representation. A Python-like language was chosen as the target language, but the decompiler can be easily extended to other target languages. Our unique solution is automatically generated from the target platform description. It exploits the architecture description language ISAC for describing the target platform and the LLVM Compiler System as the core of the decompiler. The proof of our concept is presented on a Sony PlayStation Portable (PSP) handheld game console. As can be seen from the experimental results, we are able to automatically generate a decompiler producing a highly readable HLL code for this platform while preserving the functional equivalency with the original application.

Accurate program decompilation is one of the most difficult tasks of reverse engineering. Currently, there exist several single-purpose decompilers targeted on a particular platform (e.g. Intel x86 architecture and Microsoft Windows OS) and on a particular language. These tools are always hand-written by the author from scratch. This paper presents a concept of a retargetable reverse compiler (i.e. a decompiler). This tool translates platform-specific binary applications into a high-level language (HLL) representation. A Python-like language was chosen as the target language, but the decompiler can be easily extended to other target languages. Our unique solution is automatically generated from the target platform description. It exploits the architecture description language ISAC for describing the target platform and the LLVM Compiler System as the core of the decompiler. The proof of our concept is presented on a Sony PlayStation Portable (PSP) handheld game console. As can be seen from the experimental results, we are able to automatically generate a decompiler producing a highly readable HLL code for this platform while preserving the functional equivalency with the original application.

decompilation, reverse engineering, malware, LLVM, Lissom, ISAC

decompilation, reverse engineering, malware, LLVM, Lissom, ISAC

ĎURFINA, L.; KŘOUSTEK, J.; ZEMEK, P.; KOLÁŘ, D.; HRUŠKA, T.; MASAŘÍK, K.; MEDUNA, A.

2012

10.12.2011

North Atlantic University Union

Puerto De La Cruz, Tenerife

978-1-61804-056-5

2nd European Conference of COMPUTER SCIENCE (ECCS'11)

199

204

6

http://hdl.handle.net/