A model highlighting the security of operating systems

The major thrust toward providing secure computing facilities has gone into the design of, or models for, new operating systems. Work directed toward securing current systems has, for the most part, taken the form of penetration attempts. Penetration efforts have led several authors to identify generic weaknesses, but grouping by weakness has not led to formal methods. An approach showing greater promise in identifying trouble spots, as well as characterizing existing operating systems in a more general sense, lies in forming graph models in which nodes are program modules or data structures, and arcs are access or shared resource synchronization paths. A given system should be capable of reduction to a graph of this sort by appropriate analysis of its load modules.