Demand-driven computation of interprocedural data flow

摘要: This paper presents a general framework for deriving demand-driven algorithms interprocedural data flow analysis of imperative programs. The goal is to reduce the time and/or space overhead conventional exhaustive by avoiding collection information that not needed. In our framework, demand modeled as set date queries. derived find responses these queries through partial reversal respective analysis. Depending on whether minimizing or primary concern, result caching may be incorporated in algorithm. Our applicable problems with finite domain set. If problem's functions are distributive, provide precise corresponding For monotone but non-distributive provided solutions only approximate. We demonstrate approach using example copy constant propagation.