Research on Modularization-based Code Reuse Technology in Software System Development

In this paper, we deeply analyze the dynamic characteristics of modern ROP and its variant attacks when they occur through attack replication. Based on its statistical and structural characteristics, this paper proposes a set of efficient and accurate detection techniques, including the multi-dimensional fusion of execution flow filtering method ROPMFilter and graph neural network-based ROP attack detection method ROPGMN. On this basis, the protocol-related code is integrated into the code generator, and the user interface is appropriately modified to construct a modularized code reuse generator. Conduct real application evaluation tests and find that the length of PICs generated by the algorithm in this paper is not over-inflated. Evaluating the performance of program execution, it is found that the accuracy rate, false positive rate, and false negative rate of this paper’s model in security evaluation are 97.1%, 0.42%, and 0.54%, respectively, and the performance overhead of the program is 0%, and the space required in the actual operation is very little. The designed algorithm not only realizes code reuse, but also improves the efficiency of attack detection. This paper provides new ideas and methods for code reuse and ROP attack detection.