Incremental Hashing for Spin

This paper discusses a generalised incremental hashing scheme for explicit state model checkers. The hashing scheme has been implemented into the model checker Spin. The incremental hashing scheme works for Spin’s exhaustive and both approximate verification modes: bitstate hashing and hash compaction. An implementation has been provided for 32-bit and 64-bit architectures. We performed extensive experiments on the BEEM benchmarks to compare the incremental hash functions against Spin’s traditional hash functions. In almost all cases, incremental hashing is faster than traditional hashing. The amount of performance gain depends on several factors, though. We conclude that incremental hashing performs best for the (64-bits) Spin’s bitstate hashing mode, on models with large state vectors, and using a verifier, that has been optimised by the C compiler.