Improving the Preciseness of Dependence Analysis Using Exception Analysis

Abstract

Program dependence analysis is an analysis technique to identify and determine various program dependences in program source codes. It is an important approach for testing, understanding, maintaining programs. Exception handling mechanism in modern programming languages is a technology that can improve software robustness, but failures to account for the effects of exception handling mechanism in dependence analysis can result in incorrect analysis result. In this paper, we extend previous work on dependences analysis using exception analysis to analyze the dependences of C++ programs with exception handling constructs. The improved exception control flow graph we propose, IECFG, is a novel representation of C++ program's intra-functional control flow. The IECFG is more precise than traditional CFG in representing the programs with exception handling constructs. We introduce how to construct the program dependence graph (PDG) and the system dependence graph (SDG) based on the IECFG. With the help of this representation, we can obtain more precise information than most previous methods we know