Parameter Sensitive Pointer Analysis for Java

Abstract

Pointer analysis is an important tool for program analysis and program optimization procedures used in modern compilers and bug checkers. This paper introduces parameter-sensitivity, a new methodology that improves pointer analysis for object-oriented programming languages such as Java. In this approach, we construct a method context for a callee method by a list of objects including both the receiver object and the objects that are passed as parameters at a call site. We believe that such a context is able to represent a significant portion of the current abstract state before the method invocation. The algorithm has been implemented in the Soot framework, and it is evaluated on the benchmarks from DaCapo, compared with the standard object-sensitive pointer analysis algorithms. The preliminary result demonstrates that our parameter-sensitive pointer analysis can achieve better precision with less time consumption than the standard k-object-sensitive algorithms.