A Theory-Oriented Introduction to Wait-Free Synchronization Based on the Adaptive Renaming Problem

The recent deployment of multiprocessors (such as multicores) as the mainstream computing platform has given rise to a new concurrent programming impetus. In such a context it becomes extremely important to be able to design shared objects that can cope with the net effect of asynchrony and process crashes. This paper is a theory-oriented introduction to wait-free synchronization for such systems. It uses the adaptive renaming problem as a paradigm to explain the difficulties and subtleties of synchronization in presence of process crashes. Renaming is one of the most famous coordination problems studied in distributed computability. It consists in assigning new names to processes in such a way that no two processes obtain the same new name and the new name space be as small as possible. The paper visits the problem by presenting three solutions. This paper, that has a strong survey/short tutorial flavor, can consequently be considered as an introduction to both there naming problem and progress conditions for synchronization in presence of process crashes in the context of multiprocessor systems.