基于程序特征谱整数溢出错误定位技术研究  

Research on Spectra-Based Integer Bug Localization

作　　者：惠战伟[1,2] 黄松[1,2] 嵇孟雨[1,2] 

机构地区：[1]解放军理工大学指挥自动化学院,南京210007 [2]全军军事训练软件测评中心,南京210007

出　　处：《计算机学报》2012年第10期2204-2214,共11页Chinese Journal of Computers

基　　金：国家"八六三"高技术研究发展计划项目基金(2009AA01Z402);江苏省自然科学基金(BK2012059;BK2012060)资助~~

摘　　要：随着软件业的飞速发展,人们对软件质量的要求也越来越高.整数溢出错误以其高危性和隐蔽性成为影响软件安全性和可靠性的重要因素之一.如何准确定位整数溢出错误是软件安全领域研究的热点.论文改进了现有错误定位模型,构建了整数溢出错误定位模型INTRank.实验结果表明:基于INTRank模型的语句可疑度估计方法可以较为准确地计算语句可疑度,使得程序员能够按照基于语句可疑度的优先级顺序检查源代码,找出导致整数溢出错误的原因,同时本文方法具有较低的漏报率.With the development of the software industry, people require higher quality of software. Integer bug is considered to be one of the main factors for the lacking of safety and reliability in computer systems, since it is elusive and of high risk. How to locate the integer bug accu rately is a hotspot in the field of software security testing. We propose a fault localization model INTRank, which improves the traditional model. The experiment result shows that the INTRank based approach could estimate the suspicious score of every statement under test in the program more accurately than the traditional ones. And programmers can examine the code based on the suspicious score of the priority to find out the root cause of the integer bug. Meanwhile our approach can reduce the false negative effectively.

关 键 词：整数溢出错误定位 程序特征谱 定义使用对覆盖 分支覆盖 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]