Compositional Symbolic Execution through Program Specialization

Scalability is a major challenge in symbolic execution. The large number of paths that need to be explored and the large size of the constraints that must be carried often compromise the effectiveness of symbolic execution for software testing in practice. Compositional symbolic execution aims to alleviate these scalability issues by executing the methods of a program separately, stowing their results in method summaries and using such summaries to incrementally execute the complete program. We present a novel compositional approach that leverages partial evaluation, a well-established technique that aims at automatically specializing a program with respect to some of its input. We report on its design and implementation in Symbolic PathFinder and on preliminary promising evaluation results.