Bottom-Up Context-Sensitive Pointer Analysis for Java']Java

This paper describes a new bottom-up, subset-based, and context-sensitive pointer analysis for Java. The main novelty of our technique is the constraint-based handling of virtual method calls and instantiation of method summaries. Since our approach generates polymorphic method summaries, it can be context-sensitive without reanalyzing the same method multiple times. We have implemented this algorithm in a tool called Scuba, and we compare it with k-CFA and k-obj algorithms on Java applications from the DaCapo and Ashes benchmarks. Our results show that the new algorithm achieves better or comparable precision to k-CFA and k-obj analyses at only a fraction of the cost.