Vector-clock based partial order reduction for JPF

Abstract

Java Pathfinder (JPF) employs a dynamic partial order reduction based on sharing and state hashing to reduce the schedules in concurrent systems. That partial order reduction is believed to be complete in the new version of JPF using search global IDs (SGOIDs) but does miss behaviors when SGOIDs are not employed. More importantly, it is not clear how such a dynamic partial order reduction, with or without SGOIDs, compares to other dynamic partial order reductions based on persistent sets, sleep sets, or clock vectors. In order to understand JPF's native dynamic partial order reduction better, this paper discusses an implementation of Flanagan and Goidefroid's clock vector partial order reduction in JPF. Then, the performance of JPF's native dynamic partial order reduction and the clock vector partial order reduction in JPF using SGOIDs will be compared in an effort to understand JPF's dynamic partial order reduction more fully. It was discovered that a clock vector POR always performs better in terms of runtime on the benchmarks chosen, and sometimes even better in terms of memory.