An Empirical Study of Smart Contract Decompilers

【Author】 Liu, Xia; Hua, Baojian; Wang, Yang; Pan, Zhizhong

【Source】2023 IEEE INTERNATIONAL CONFERENCE ON SOFTWARE ANALYSIS, EVOLUTION AND REENGINEERING, SANER

【影响因子】

【Abstract】Smart contract decompilers, converting smart contract bytecode into smart contract source code, have been used extensively in many scenarios such as binary code analysis, reverse engineering, and security studies. However, existing studies, as well as industrial engineering practices, all assumed that smart contract decompilers are reliable and trustworthy, to generate correct and semantically equivalent source code from binaries. Unfortunately, whether such an assumption truly holds in practice is still unknown. In this paper, we conduct, to the best of our knowledge, the first and most comprehensive large-scale empirical study of smart contract decompilers, to gain an understanding of the reliability, limitations, and remaining research challenges of state-of-the-art smart contract decompilation tools. We first designed and implemented a software prototype SOLINSIGHT, then used it to study 5 state-of-the-art smart contract decompilers. We obtained important findings and insights from empirical results, such as: 1) we proposed 3 root causes leading to decompiler failures; 2) we revealed 2 reasons hurting performance; 3) we identified 3 root causes affecting decompilation effectiveness; 4) we proposed a measurement metric for completeness; and 5) we investigated the resilience of contract decompilers against program transformations. We suggest that: 1) decompiler builders should enhance decompilers in terms of effectiveness, performance, and completeness; and 2) security researchers should select appropriate decompilers based on the suggestions in this study. We believe these findings and suggestions will help decompiler builders, contract developers, and security researchers, by providing better guidelines for contract decompiler studies.

【Keywords】Empirical study; Smart contracts; Decompilation

【发表时间】2023

【收录时间】2023-07-24

【文献类型】实证数据

【主题类别】

区块链技术-核心技术-智能合约