一句话：提出 PyRTFuzz，融合生成式与变异式模糊测试，有效检测 Python 解释器与运行时库中的运行时缺陷。
Python 的广泛使用使其运行时系统的可靠性与安全性至关重要，但针对该语言运行时的自动化缺陷检测工具仍然稀缺。本文以工具演示形式提出 PyRTFuzz，一个面向 Python 运行时（包括语言解释器与运行时库）的实用模糊测试工具。该方法融合了基于生成的模糊测试与基于变异的模糊测试，以提升测试覆盖与缺陷发现能力。评估中，PyRTFuzz 发现了 61 个可利用缺陷，涵盖解释器核心以及多个运行时库，并展示了良好的可扩展性与成本效益。作者认为该工具有望用于大规模缺陷检测，并可推广到其他解释型语言的运行时系统。
Python's widespread usage underscores the critical need for the reliability and security of its runtime system. However, automated tools for detecting runtime bugs in this impactful language are scarce. This tool demo showcases PyRTFuzz, a practical fuzzing tool for testing the Python runtime which encompasses the language interpreter and runtime libraries. Leveraging a blend of generation- and mutation-based fuzzing. PyRTFuzz unveils 61 exploitable bugs, including those within the interpreter core and various runtime libraries. Demonstrating scalability and cost-effectiveness, PyRTFuzz shows promise for extensive bug detection, with potential applicability to the runtime of other interpreted languages.