Android binary file vulnerability detection method and system

Abstract

The invention belongs to the technical field of Android binary file vulnerability excavation, and discloses an Android binary file vulnerability detection method and system. The method comprises the steps that the collection of code coverage rate information is achieved by using a pile pitching technology based on LLVM to be adopted as judging criteria of measuring the feed sufficiency in a genetic algorithm, and seeds are selected to perform random variation of the next generation; a testing case generation model is constructed by using an RNN network in machine learning, and testing cases obtained by training using a large number of files of a certain format are used for generating the model to automatically generate new testing cases. By means of the method, a Fuzzing test can be conducted on binary files such as Android system libraries and the like, the coverage rate of initial codes can be effectively improved by using a machine learning technology to perform training to obtain the testing case generation model to generate initial testing cases, the needed variation and execution time for searching for some routes is saved, and the efficiency of Fuzzing is improved.

Description

technical field [0001] The invention belongs to the technical field of exploiting vulnerabilities of Android binary files, and in particular relates to a method and system for detecting vulnerabilities of Android binary files. Background technique [0002] At present, the existing technologies commonly used in the industry are as follows: [0003] Fuzzing is a black-box testing technique that tests the robustness and security of programs by continuously generating a large amount of malformed test data. The core of this technology is test case generation technology, good test case generation technology can guarantee higher code coverage and test efficiency. The test case generation technology in Fuzzing can be divided into two types: generation type and variation type. The generation type is to directly generate test cases according to preset rules (such as the format of files and protocols), while the variation type is to generate test cases through the given The seed tes...

AI Technical Summary

Problems solved by technology

MFFA, released in 2015, is a tool for fuzzing testing of the libstagefright library in the Android system, but the framework only uses a simple random mutation strategy to generate test cases, generating a large number of wrongly formatted and invalid test cases, making the entire The fuzzing process spends a lot of time on meaningless execution and mutation

[0007] (1) The code coverage rate caused by the simple random mutation strategy in Fuzzing technology is not high;

[0008] (2) At present, most of the Fuzzing tools on the Android platform are aimed at the communication mechanism between upper-layer applications and components, and simply test the system calls, but lack the Fuzzing tools for the Android system framework and system class library

Android does not support the SysV IPC function due to customization reasons, and cannot directly use shared memory directly like using functions such as shmget on linux

Images