Hybrid Concurrent Objects

Abstract

This chapter focuses on hybrid implementations of concurrent objects. Roughly speaking, “hybrid” means that lock-based code and mutex-free code can be merged in the same implementation. After defining the notion of a hybrid implementation, this chapter presents hybrid implementations of concurrent objects, where each implementation has its own features. The chapter presents also the notion of an abortable object and shows how a starvation-free implementation of a concurrent object can be systematically obtained from an abortable non-blocking version of the same object.