On Partial Wait-Freedom in Transactional Memory

Abstract

Transactional memory (TM) is a convenient synchronization tool that allows concurrent threads to declare sequences of instructions on shared data as speculative transactions with "all-or-nothing" semantics. It is known that dynamic transactional memory cannot provide wait-free progress ensuring that every transaction commits in a finite number of its own steps. In this paper, we explore the costs of providing wait-freedom to only a subset of transactions. We require that read-only transactions commit in the wait-free manner, while updating transactions are guaranteed to commit only if they run in the absence of concurrency. We show that this kind of partial wait-freedom, combined with attractive requirements like read invisibility or disjoint-access parallelism, incurs considerable complexity costs.