一种计算缺陷贡献率的代码缺陷定位规则

Research on codes defect localization rules by calculating defect contribution rate

为了提高基于覆盖率的缺陷定位方法的效率和准确率,将执行轨迹中被失效执行所覆盖的基本块(Cfailure)、被全部执行所覆盖的基本块(C-all)和未被任何执行所覆盖的基本块(C-non)这三个小概率事件进行了分析,指出这些小概率事件的出现包含大量信息,据此提出三条缺陷检测新规则。然后利用执行轨迹所收集的覆盖率度量值计算缺陷贡献率CDCR,给出了缺陷贡献率算法(Codes-TCC)。研究主要创新在于新规则中贡献率公式加入权重系数且算法实现层层递归,可精确定位缺陷位置;除此之外,规则中加入排除无关基本块的优化,很大地提高了缺陷定位的效率。最后在西门子和space程序集上将Codes-TCC缺陷定位结果与其他四种经典的基于覆盖率的缺陷定位算法进行对比实验,证实了该方法已达到期望的目的。

To improve the efficiency and accuracy of the defect localization based on coverage, this paper focused on the re- search of three type code blocks which had small probability such as blocks covered by the failure executions, blocks covered by all the cases and blocks not covered by any cases. This paper argued that these small probability events contain more infor- mation, and proposed three rules for defect localization. It calculated the codes defect contribution rate （CDCR） based on col- lected execution coverage metrics and presented related CDCR algorithm. The innovation of the research was that put the weight coefficient into the CDCR formula and a reeursive algorithm. It added some rules to remove the irrelevant blocks, which could greatly improve the efficiency of the defect location. In order to show feasibility of the methods, also conducted a case study using the Siemens and space suite, compared and the results of Codes-TCC with other four defect location algorithms based on coverage in the Siemens and space suite. The experimental results show that the Codes-TCC is better than other four algorithms in efficiency and accuracy of the defect localization.