CAREER: A Framework For Customizable Program Flow Analysis

Rensselaer Polytechnic Institute, Troy NY

Investigators

Abstract

CCF-0642911 CAREER: A Framework for Customizable Program Flow Analysis Ana Milanova Today's software systems are large, complex and costly and research on program flow analysis for software productivity and software quality is increasingly relevant. Tools based on flow analysis can aid and improve software tasks such as development, testing, program understanding, verification, debugging, reverse engineering, and restructuring. This wide variety of tasks demands a wide variety of customized flow analyses with different degrees of precision and cost. This research builds technology that will allow specifying and developing customized flow analyses that support software tasks on large and complex software systems. There are three contributions. First, we develop the theory of contextual set constraints, a flexible declarative formalism for specifying flow analyses. Second, we design and implement a novel engine for resolution of set constraint which takes advantage of efficient Binary Decision Diagram (BDD) technology for set representation and set manipulation. Third, we build an analysis framework based on contextual set constraints and efficient constraint resolution. The framework is used to specify and develop new flow analyses that target long-standing problems related to software security and software dependability.