Abstraction-driven Concolic Testing

ion-driven Concolic Testing‹ Przemys law Daca, Ashutosh Gupta, and Thomas A. Henzinger 1 IST Austria, Austria 2 Tata Institute for Fundamental Research, India Abstract. Concolic testing is a promising method for generating test suites for large programs. However, it suffers from the path-explosion problem and often fails to find tests that cover difficult-to-reach parts of programs. In contrast, model checkers based on counterexample-guided abstraction refinement explore programs exhaustively, while failing to scale on large programs with precision. In this paper, we present a novel method that iteratively combines concolic testing and model checking to find a test suite for a given coverage criterion. If concolic testing fails to cover some test goals, then the model checker refines its program abstraction to prove more paths infeasible, which reduces the search space for concolic testing. We have implemented our method on top of the concolic-testing tool Crest and the model checker CpaChecker. We evaluated our tool on a collection of programs and a category of SvComp benchmarks. In our experiments, we observed an improvement in branch coverage compared to Crest from 48% to 63% in the best case, and from 66% to 71% on average. Concolic testing is a promising method for generating test suites for large programs. However, it suffers from the path-explosion problem and often fails to find tests that cover difficult-to-reach parts of programs. In contrast, model checkers based on counterexample-guided abstraction refinement explore programs exhaustively, while failing to scale on large programs with precision. In this paper, we present a novel method that iteratively combines concolic testing and model checking to find a test suite for a given coverage criterion. If concolic testing fails to cover some test goals, then the model checker refines its program abstraction to prove more paths infeasible, which reduces the search space for concolic testing. We have implemented our method on top of the concolic-testing tool Crest and the model checker CpaChecker. We evaluated our tool on a collection of programs and a category of SvComp benchmarks. In our experiments, we observed an improvement in branch coverage compared to Crest from 48% to 63% in the best case, and from 66% to 71% on average.