Re-inventing a Great Idea

Abstract

As mentioned before, the concept of mutual exclusion has, ever since the early days of computing, been recognized as a central issue in taming the complexity brought about by multiprograms run on a shared installation. The reason why it became a central issue is that the primitive statements provided by actual machinery often were — and are — far too finegrained to make multiprogramming practically feasible. Thus, the mutual exclusion problem, i.e. the problem of how to build (arbitrary) coarsegrained atomic statements out of finer-grained ones, did become an urgent one. In its canonical form, the problem is:

Given a number of components, each of the form

$$*[ncs;cs] $$

synchronize them in such a way that at any moment in time, at most one component is engaged in its cs- fragment.