| The use of libraries in software development can not only reduce the difficulty of software development to a large extent,but also effectively improve the efficiency of software development.However,when analyzing software programs,in order to achieve the purpose of accurately analyzing program behaviors Therefore,static data flow analysis tools have to add a large number of logically complex library functions to the analysis process,which greatly increases the cost of analysis.Currently constructing summaries for library functions is a common way to speed up program analysis.The summary refers to the analysis results of the program modules that can be reused.By modeling the data flow of the library function in advance,it is no longer necessary to add the library function code to the analysis process for analysis when analyzing the application.The same analysis effect can be obtained by loading the library summary on demand.Stub Droid is the first tool to automatically generate data flow summary,but it is found that Stub Droid only models the data dependencies of the methods in the known parts of the library functions,and does not process the unknown parts in the library functions,such as callbacks.Since the data flow summary of the callback function cannot be obtained,the library code needs to be analyzed again when the summary is used for analysis.This paper proposes a summarization technology that integrates conditional reachability analysis to improve Stub Droid's traditional problem of ignoring the processing of unknown component callback functions when generating summaries based on component-level analysis,and uses tree-adjoining-language(TAL)to build summaries further The conditional reachability relationship that includes callbacks,supplements the vacancy of callback functions in the data dependency summary,further improves the accuracy and completeness of the summary,and avoids the need to analyze library functions again when using the summary for program analysis. |
PDF Full Text Request