Abstract

The computational cost and precision of a shape style heap analysis is highly dependent on the way method calls are handled. This paper introduces a new approach to analyzing method calls that leverages the fundamental object-oriented programming concepts of encapsulation and invariants. The analysis consists of a novel partial context-sensitivity heuristic and a new take on cutpoints that, in practice, provide large improvements in interprocedural analysis performance while having minimal impacts on the precision of the results. The interprocedural analysis has been implemented for .Net bytecode and an existing abstract heap model. Using this implementation we evaluate both the runtime cost and the precision of the results on a number of well known benchmarks and real-world programs. Our experimental evaluations show that, despite the use of partial context sensitivity heuristics, the static analysis is able to precisely approximate the ideal analysis results. Further, the results show that the interprocedural analysis heuristics and the approach to cutpoints used in this work are critical in enabling the analysis of large real-world programs, over 30K bytecodes in less than 65 seconds and using less than 130 MB of memory, and which could not be analyzed with previous approaches.