Cooperative concurrency for a multicore world

Abstract

Developing reliable multithreaded software is notoriously difficult, due to the potential for unexpected interference between concurrent threads. Even a familiar construct such as "x++" has unfamiliar semantics in a multithreaded setting, where it must in general be considered a non-atomic read-modify-write sequence, rather than a simple atomic increment. Understanding where thread interference may occur is a critical first step in understanding or validating a multithreaded software system. © 2012 Springer-Verlag.