Autodafé: an Act of Software Torture
Autodafé: an Act of Software Torture
复制标题
DOI：
--
复制
发表时间：
2005
期刊：
影响因子：
--
通讯作者：
Martin Vuagnoux
中科院分区：
文献类型：
--
作者：
Martin Vuagnoux
自动化漏洞搜索工具已导致发现此类漏洞的速度大幅提高。一种特定的搜索技术是故障注入，即将随机数据插入到输入文件、缓冲区或协议分组中，并结合对存储器违规的系统监控。即使这些工具允许发现许多漏洞，它们仍然非常原始；尽管它们的效率很低，但它们是有用的，因为现代软件中此类漏洞的密度非常高。本文提出了一种新的缓冲区溢出发现技术，它使用了一种更全面和更可靠的方法。这种技术被称为：通过标记加权攻击来模糊，是一种专门的故障注入，不需要为被监控的程序编写源代码或专门的编译。为了证明这项技术的有效性，开发了一个名为Autodafe的工具。它允许自动检测数量可观的缓冲区溢出漏洞。
Automated vulnerability searching tools have led to a dramatic increase of the rate at which such flaws are discovered. One particular searching technique is fault injection i.e. insertion of random data into input files, buffers or protocol packets, combined with a systematic monitoring of memory violations. Even if these tools allow to uncover a lot of vulnerabilities, they are still very primitive; despite their poor efficiency, they are useful because of the very high density of such vulnerabilities in modern software. This paper presents an innovative buffer overflow uncovering technique, which uses a more thorough and reliable approach. This technique, called: Fuzzing by Weighting Attacks with Markers, is a specialized kind of fault injection, which does not need source code or special compilation for the monitored program. As a proof of concept of the efficiency of this technique, a tool called Autodafe has been developed. It allows to detect automatically an impressive number of buffer overflow vulnerabilities.