Document Type

Technical Report

Publication Date

Spring 4-1-1976


In an earlier paper, the authors have defined type 1 and type 2 data flow anomalies to be, respectively, the reference to an undefined variable and the definition of a variable without subsequent reference. It is not difficult to devise search techniques to detect such anomalies when the anomalous data flow is contained in a single procedure. When the data flow crosses procedure boundaries, however, many difficulties may arise. In this paper, we carefully define the conditions under which inter-procedural anomalies occur. We also show how algorithms currently used in global program optimization can easily be adapted to yield highly efficient algorithms for the detection of such inter-procedural anomalies.