Precise guidance to dynamic test generation
Dynamic symbolic execution has been shown an effective technique for automated test input generation. However, its scalability is limited due to the combinatorial explosion of the path space. We propose to take advantage of data flow analysis to better perform dynamic symbolic execution in the context of generating test inputs for maximum structural coverage. In particular, we utilize the chaining mechanism to (1) extract precise guidance to direct dynamic symbolic execution towards exploring uncovered code elements and (2) meanwhile significantly optimize the path exploration process. Preliminary experiments conducted to evaluate the performance of the proposed approach have shown very encouraging results.