Research On Static Detection Method Of Memory Leak Based On Defect Pattern

With the continuous enlargement of software scale and its increasing complexity,how to detect the potential defect of software becomes the focus of the current research.For memory leak detection,it is mostly static detect by analyzing the source code to find potential defects.However,the static detection needs a long time to detect program,and its false positives rate is too high.Therefore,the study of high efficiency and accuracy of memory leak detection method is necessary,which has strong research value and practical significance.A memory leak detection method based on defect pattern under static detection is designed for memory leak detection.Based on the analysis of the existing memory leak defect model,the memory operation behavior(memory allocation,memory release and memory transfer)existing in the memory leak defect mode is summarized,and the corresponding memory operation behavior state machine model is designed.The method uses the fuzzy matching algorithm based on Regular Expression to match the memory behavior in the code to be detected,and then detecting the state transition of the memory operation behavior state machine to realiz the static detection of memory leak.At the same time,in order to improve the efficiency of detection and reduce the problem that function is repeated in function call point,a function summary based on memory behavior is proposed.And the corresponding function summary algorithm and function summary update algorithm are designed.On the basis of detecting the same open source code,experiments by comparing with the common memory leak detection tools analyze the detection efficiency,such as detecting time,false positives and false reporting rate.Experimental results show that the memory leak detection method based on defect pattern has better detection spend,and lower false positives rate.