Detecting State Inconsistency Bugs in DApps via On-Chain Transaction Replay and Fuzzing

Abstract

Decentralized applications (DApps) consist of multiple smart contracts running on Blockchain. With the increasing popularity of the DApp ecosystem, vulnerabilities in DApps could bring significant impacts such as financial losses. Identifying vulnerabilities in DApps is by no means trivial, as modern DApps consist of complex interactions across multiple contracts. Previous research suffers from either high false positives or false negatives, due to the lack of precise contextual information which is mandatory for confirming smart contract vulnerabilities when analyzing smart contracts. In this paper, we present IcyChecker, a new fuzzing-based framework to effectively identify State inconsistency (SI) Bugs - a specific type of bugs that can cause vulnerabilities such as re-entrancy, front-running with complex patterns. Different from prior works, IcyChecker utilizes a set of accurate contextual information for contract fuzzing by replaying the on-chain historical transactions. Besides, instead of designing specific testing oracles which are required by other fuzzing approaches, IcyChecker implements novel mechanisms to mutate a set of fuzzing transaction sequences, and further identify SI bugs by observing their state differences. Evaluation of IcyChecker over the top 100 popular DApps showed it effectively identifies a total number of 277 SI bugs, with a precision of 87%. By comparing IcyChecker with other state-of-the-art tools (i.e., Smartian, Confuzzius, and Sailfish), we show IcyChecker not only identifies more SI bugs but also with much lower false positives, thanks to its integration of accurate on-chain data and unique fuzzing strategies. Our research sheds light on new ways of detecting smart contract vulnerabilities in DApps.