Path-sensitive alias analysis for data flow testing

This paper describes a new approach to performing data flow testing in the presence of aliasing, and a tool based on this method. The technique is based on the observation that, under certain reasonable assumptions, one can statically determine which variables are aliased whenever control reaches a given program point via a particular path. Furthermore, one can group together paths that behave similarly with respect to aliasing and represent them by regular expressions. The resulting test requirements demand that the test data execute representatives of particular sets of paths between variable definitions and uses.