FuzzFabric: A Scalable Grey-box Fuzzing Framework for Polymorphic Network Services

Bo Yu,Lei Zhou,Chengnuo Cai,Qiang Yang,Yaojia Yang,Runhao Liu,Lei Luo,Wenyu Long,Lianghua Gong

DOI: https://doi.org/10.1109/icws62655.2024.00139

2024-01-01

Abstract:Fuzzing network protocols is challenging due to their various factors including syntax, state transition, and even conformation of its executable software. To achieve better convenience of platform development and effectiveness of individual protocol testing, current grey-box fuzzing approaches generally focus on limited factors. However, it leaves poor performance on real-world services’s protocol evaluation since it requires significant additional adaptation efforts to accommodate the diversity of protocol software. To address this issue, we present FuzzFabric, a scalable grey-box fuzzing framework designed for polymorphic protocol services. FuzzFabric decouples existing fuzzing engines and modular programs the protocol processing adapters, and finally adaptive assembles fuzzing engine to evaluate different network protocol services. We implemented FuzzFabric on top of AFLNet, supporting 3 different protocol types and 4 instrumentation tools, and conducted experiments using 11 real-world protocol applications. The results demonstrate that FuzzFabric can effectively fuzz protocols in different forms with various instrumentation, validating its practical efficacy and scalability.