Luke Dramko (Carnegie Mellon University), Claire Le Goues (Carnegie Mellon University), Edward J. Schwartz (Carnegie Mellon University)

Decompilers help reverse engineers analyze software at a higher level of abstraction than assembly code. Unfortunately, because compilation is lossy, traditional decompilers, which are deterministic, produce code that lacks many characteristics that make source code readable in the first place, such as variable and type names. Neural decompilers offer the exciting possibility of statistically filling in these details. Unfortunately, existing work in neural decompilation suffers from substantial limitations that preclude its use on real code, such as the inability to provide definitions for user-defined composite types. In this work, we introduce Idioms, a simple, generalizable, and effective neural decompilation approach that can finetune any LLM into a neural decompiler capable of generating the appropriate user-defined type definitions alongside the decompiled code, and a new dataset, Realtype, that includes substantially more complicated and realistic types than existing neural decompilation benchmarks. We show that our approach yields state-of-the-art results in neural decompilation. On the most challenging existing benchmark—EXEBENCH—our model achieves 54.4% accuracy vs. 46.3% for LLM4Decompile and 37.5% for Nova; on REALTYPE, our model performs at least 95% better.

View More Papers

Beyond Conventional Triggers: Auto-Contextualized Covert Triggers for Android Logic...

Ye Wang (Department of Electrical Engineering and Computer Science, Institute for Information Sciences, The University of Kansas), Bo Luo (Department of Electrical Engineering and Computer Science, Institute for Information Sciences, The University of Kansas), Fengjun Li (Department of Electrical Engineering and Computer Science, Institute for Information Sciences, The University of Kansas)

Read More

Analysing Privacy Risks in Children’s Educational Apps in Australia

Sicheng Jin (University of New South Wales), Rahat Masood (University of New South Wales), Jung-Sook Lee (University of New South Wales), Hye-Young (Helen) Paik (University of New South Wales)

Read More

A Deep Dive into Function Inlining and its Security...

Omar Abusabha (Sungkyunkwan University, South Korea), Jiyong Uhm (Sungkyunkwan University, South Korea), Tamer Abuhmed (Sungkyunkwan University, South Korea), Hyungjoon Koo (Sungkyunkwan University, South Korea)

Read More