Automated program debugging for multiple bugs based on semantic analysis

Fault locating is a time-consuming process. In a previous paper, Liu proposed an algorithm named bounded debugging via multiple predicate switching (BMPS)[1], which try to find a successful execution trace by switching outcomes of multiple predicates. Substantively, BMPS focuses on the program faults which are caused by control flow. However, this kind of faults represent only a small fraction. In this paper, we present an algorithm combining BMPS with a semantic based debugging method, which is aimed at locating more than control flow related faults. The semantic based debugging algorithm generates a sequence of equations from the execution trace of a failed test case, and give a minimum faulty program segment according to solutions of the equations. Our algorithm can locate multiple faults in the program one by one through an iterative and interactive process. Moreover, an optimization based on use-define chain is applied to BMPS for improving efficiency of the algorithm, as well as some other methods. To evaluate out approach, we conduct experiment on Siemens suite. The result indicates that our method has significant improvement on both accuracy and efficiency.