An Approach of Data-Flow Analysis Based on Exception Propagation Analysis

Abstract

Exception handling is a technology that tests and handles exception. Exception propagation induces a control flow other than the main control flow, so it changes the data flows of programs. For data flow analysis of C++ programs to be correct and precise, the flows induced by exception propagation must be properly analyzed. Firstly, based on analyzing the exception handling mechanism and the effects of exception propagation on data flow analysis, the paper proposes a precise and efficient representation for C++ programs with exception handling constructs—control flow graph that contains exception propagation. The control flow graph can represent precisely the implicit control flow for a raised exception and exception propagation path. Then this paper proposes an efficient method to analyze the data flow of the programs that contain exception propagations, and some efficient algorithms are also presented. This method overcomes the limitations of previous incorrect analysis due to failing to account for the effects of exception propagation, and also provides a basis for automatic data flow analysis that contains exception propagation. Finally, it validates the usability of the method by a case study. The method can provide related informations for software engineering tasks such as structure testing, regression testing and program slicing.