UNIFUZZ is a platform for using and evaluating fuzzers conveniently. UNIFUZZ consists of the following main components:
- Useable Fuzzers
- Benchmark
- Seeds
- Metrics for Evaluating Fuzzers
The corresponding paper will appear on USENIX Security 2021 paper pdf.
For citing this paper:
@inproceedings{unifuzz-li,
title={{UNIFUZZ}: A Holistic and Pragmatic Metrics-Driven Platform for Evaluating Fuzzers},
author={Li, Yuwei and Ji, Shouling and Chen, Yuan and Liang, Sizhuang and Lee, Wei-Han and Chen, Yueyao and Lyu, Chenyang and Wu, Chunming and Beyah, Raheem and Cheng, Peng and Lu, Kangjie and Wang, Ting},
booktitle={Proceedings of the 30th {USENIX} Security Symposium},
year={2021},
}
Currently we provide Dockerfile for 35 usable fuzzers.
https://github.com/unifuzz/dockerized_fuzzing
Currently we provide 20 real-world programs for fuzzing evaluation.
https://github.com/unifuzz/unibench
In addition, we also provide Dockerfile for building these programs under different fuzzer instrumentations.
https://github.com/unifuzz/unibench_build
This repo consists of the seeds used in the evaluations of UNIFUZZ paper.
https://github.com/unifuzz/seeds
https://github.com/unifuzz/metrics
Due to space limitaion of the published paper, we present some supplementary results of UNIFUZZ paper here.