arXivDaily arXiv每日学术速递 周一至周五更新
​通知:联邦节假日调休:哥伦布日 / 原住民日(Columbus Day / Indigenous Peoples' Day),今日(13日)无更新。
arXiv 2404.16297cs.SEcs.AI

当模糊测试遇上LLM:挑战与机遇

When Fuzzing Meets LLMs: Challenges and Opportunities

School of Software, Tsinghua University(清华大学软件学院);National University of Defense Technology(国防科技大学)

Yu Jiang, Jie Liang, Fuchen Ma, Yuanliang Chen, Chijin Zhou, Yuheng Shen, Zhiyong Wu, Jingzhou Fu, Mingzhe Wang, ShanShan Li, Quan Zhang

AI总结:

本文识别了LLM辅助模糊测试面临的五大挑战,并基于顶级会议论文验证其普遍性,进而提出可操作建议并在DBMS模糊测试中初步验证了其有效性。

摘要:

模糊测试是一种广泛使用的缺陷检测技术,已通过大型语言模型(LLMs)取得了进展。尽管具有潜力,LLMs在模糊测试中面临特定挑战。在本文中,我们识别了LLM辅助模糊测试的五个主要挑战。为了支持我们的发现,我们回顾了顶级会议的最新论文,证实这些挑战普遍存在。作为补救,我们提出了一些可操作的建议,以帮助改进LLM在模糊测试中的应用,并对DBMS模糊测试进行了初步评估。结果表明,我们的建议有效解决了所识别的挑战。

英文摘要:

Fuzzing, a widely-used technique for bug detection, has seen advancements through Large Language Models (LLMs). Despite their potential, LLMs face specific challenges in fuzzing. In this paper, we identified five major challenges of LLM-assisted fuzzing. To support our findings, we revisited the most recent papers from top-tier conferences, confirming that these challenges are widespread. As a remedy, we propose some actionable recommendations to help improve applying LLM in Fuzzing and conduct preliminary evaluations on DBMS fuzzing. The results demonstrate that our recommendations effectively address the identified challenges.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑