State-space exploration for concurrent algorithms under weak memory orderings

Abstract

Several concurrent implementations of familiar data abstractions such as queues, sets, or maps typically do not follow locking disciplines, and often use lock-free synchronization to gain performance. Since such algorithms are exposed to a weak memory model, they are notoriously hard to get correct, as witnessed by many bugs found in published algorithmsWe outline a technique for analyzing correctness of concurrent algorithms under weak memory models, in which a model checker is used to search for correctness violations. The algorithm to be analyzed is transformed into a form where statements may be reordered according to a particular weak memory ordering. The transformed algorithm can then be analyzed by a model-checking tool, e.g., by enumerative state exploration. We illustrate the approach on a small example of a queue, which allows an enqueue operation to be concurrent with a dequeue operation, which we analyze with respect to the RMO memory model defined in SPARC v9.