Aggregate structure identification and its application to program analysis

In this paper, we describe an efficient algorithm for lazily decomposing aggregates such as records and arrays into simpler components based on the access patterns specific to a given program. This process allows us both to identify implicit aggregate structure not evident from declarative information in the program, and to simplify the representation of declared aggregates when references are made only to a subset of their components. We show that the structure identification process can be exploited to yield the following principal results: - A fast type analysis algorithm applicable to program maintenance applications such as date usage inference for the "Year 2000" problem. - An efficient algorithm for atomization of aggregates. Given a program, an aggregate atomization decomposes all of the data that can be manipulated by the program into a set of disjoint atoms such that each data reference can be modeled as one or more references to atoms without loss of semantic information. Aggregate atomization can be used to adapt program analyses and representations designed for scalar data to aggregate data. In particular, atomization can be used to build more precise versions of program representations such as SSA form or PDGs. Such representations can in turn yield more accurate results for problems such as program slicing.Our techniques are especially useful in weakly-typed languages such as Cobol (where a variable need not be declared as an aggregate to store an aggregate value) and in languages where references to statically-defined subranges of data such as arrays or strings are allowed.