首页 | 本学科首页   官方微博 | 高级检索  
     

一种基于污点分析的文件型软件漏洞发现方法
引用本文:刘智,张小松. 一种基于污点分析的文件型软件漏洞发现方法[J]. 小型微型计算机系统, 2012, 33(1): 42-48
作者姓名:刘智  张小松
作者单位:电子科技大学计算机科学与工程学院,成都,611731
基金项目:保密通信国防科技重点实验室基金项目(9140C1104020903)资助
摘    要:基于黑盒测试思想的Fuzzing是漏洞分析的主要方法,但效率较低且不能分析未知格式.基于污点分析,提出一种针对文件型软件的漏洞发现新方法.利用污点分析寻找输入中能导致执行流到达脆弱点的字节,再改变这些字节产生新输入;同时根据污点信息产生特征码.利用插桩工具实现了原型系统,对三个真实漏洞进行了测试.实验结果表明该方法能有效发现漏洞,生成的测试用例远小于Fuzzing,且不依赖输入格式;特征码的误报率漏报率均较低.

关 键 词:漏洞  污点分析  插桩  依赖性分析  特征码

New Approach of Vulnerability Finding for File-reading Applications Based on Taint Analysis
LIU Zhi , ZHANG Xiao-song. New Approach of Vulnerability Finding for File-reading Applications Based on Taint Analysis[J]. Mini-micro Systems, 2012, 33(1): 42-48
Authors:LIU Zhi    ZHANG Xiao-song
Affiliation:(Department of Computer Science & Engineering,University of Electronic Science and Technology of China,Chengdu 611731,China)
Abstract:Black-box based fuzzing is the major technique of discovering vulnerability,but it is quite ineffective and cannot analyze unknown input formats.Using taint analysis,we propose a new method of discovering vulnerability targeted at file-reading applications.It locates bytes in input that direct execution to vulnerable points,which will be mutated to generate new test cases;signatures are generated via vulnerability information.We implemented a proof-of-concept system with instrumentation tools and evaluated three real-world vulnerabilities.Experiment results show our approach is able to effectively find vulnerabilities with much fewer test cases,being independent of input formats,and signatures have low false positives and false negatives.
Keywords:vulnerability  taint analysis  instrumentation  dependency analysis  signature
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号