摘要
针对当前分布式模糊测试任务分配中,大量测试用例重复执行导致测试效率低的问题,提出基于状态聚类的分布式模糊测试任务分配策略。通过大量随机测试获得目标程序的初始状态空间;根据路径前缀将初始状态空间划分为相互独立的状态子空间,将其对应的任务分发到各测试节点;利用路径约束控制变异的方向,保证各测试节点探测的状态空间相互独立。根据该方法,设计实现一个高效的分布式模糊测试系统。分别在libtiff库和CGC测试集上进行实验,实验结果表明,该系统能够有效减少测试用例的重复执行次数,提高模糊测试效率,有机会发现更多crash。
In the current distributed fuzzing test task assignment,the repeated execution of a large number of test cases leads to low test efficiency.To solve the problem,a distributed fuzzing test task assignment strategy based on state clustering was proposed.The initial state space of the target program was obtained through a large number of random tests.The initial state space was divided into independent state subspaces according to the path prefix,and the corresponding tasks were distributed to the test nodes.The path constraint control was utilized.The direction of the mutation ensured that the state space detected by each test node was independent of each other.According to this method,an efficient distributed fuzzing test system was designed and implemented.Results of experiments on the libtiff library and CGC test set show that the system can effectively reduce the number of repeated executions of test cases,improve the efficiency of fuzzing test,and it has the opportunity to find more crashes.
作者
邓一杰
刘克胜
赵军
常超
朱凯龙
DENG Yi-jie;LIU Ke-sheng;ZHAO Jun;CHANG Chao;ZHU Kai-long(College of Electronic Engineering,National University of Defense Technology,Hefei 230000,China)
出处
《计算机工程与设计》
北大核心
2019年第9期2428-2434,2501,共8页
Computer Engineering and Design
基金
国家重点研发计划“网络空间安全”重点专项基金项目(2017YFB0802905)
关键词
分布式
模糊测试
聚类
状态空间聚类
路径约束
distributed
fuzzing test
clustering
state space clustering
path constraint
