一句话：提出 PeachCI，一个面向基于生成协议模糊测试的可扩展持续集成平台，解耦 fuzzer 与目标协议并集成监控与可视化。
当前协议模糊测试的持续集成主要面向变异式 fuzzer，忽视了基于生成的 fuzzer；且常将 fuzzer 与目标协议打包进同一 Docker 镜像并使用简单前端，难以与企业级监控和数据可视化平台集成。本文设计并实现了 PeachCI，一个面向基于生成协议模糊测试的可扩展持续集成平台。PeachCI 将 fuzzer 与目标协议解耦，各自封装在独立 Docker 镜像中，通过容器间通信执行模糊测试；平台以 Pit 文件为输入，集成 Prometheus 监控系统与 Grafana 可视化平台，并支持并行模糊测试。其使用分支覆盖率作为关键评估指标，便于比较不同基于生成的 fuzzer 的有效性。实验结果表明，PeachCI 能有效评估不同 fuzzer 的性能，并为网络协议模糊测试提供有力支持。
Continuous integration (CI) of protocol fuzzing is important to enhance the efficiency of fuzzing. However, current CI architecture only focus on mutation-based fuzzers, which neglects generation-based fuzzers. Even worse, it make fuzzers and target protocol into a single Docker image and use a simple front-end, which suffers the practical integration with enterprise-level monitoring and data visualization platforms. In this paper, we designed and implemented a scalable continuous integration platform for generation-based protocol fuzzing, PeachCI. PeachCI decoupling the fuzzer from the target protocol, with each component encapsulated in its own Docker image, enabling fuzzing through inter-container communication. Furthermore, the platform integrates with the widely-used systems and service monitoring system Prometheus, and the data visualization platform Grafana, using Pit files as inputs and displaying the results through Grafana. It supports parallel fuzzing and uses branch coverage as a key evaluation metric, which allows for comparison of the effectiveness of various generation-based fuzzers. The experimental results demonstrate that PeachCI is an effective method for evaluating the performance of different fuzzers and provides robust support for the network protocol fuzzing.
@inproceedings{paperbot8959,
title = {PeachCI: Scalable Continuous Integration of Generation-Based Protocol Fuzzing},
author = {Wanli Chen and Yuanliang Chen and Fuchen Ma and Ruikang Peng and Qi Xu and Yu Jiang and Qiang Fu and Heyuan Shi},
booktitle = {FSE},
year = {2025},
doi = {10.1145/3696630.3728612}
}