Directed Fuzzing Based on Dynamic Taint Analysis for Binary Software
Bo Wu, Bin Zhang, Shameng Wen, Meng Jun Li, Quan Zhang, Chao Jing Tang · Applied Mechanics and Materials · 2014
Traditional Fuzzing is simple and easy to deploy but inefficient due to different inputs usually execute the redundant path. In this paper, we put forward a binary-oriented Fuzzing technique based on input format analysis and dynamic taint analysis, which can detect vulnerability more efficient than traditional Fuzzing method. We implemented a prototype system called Smart and Directed Fuzz (SDFuzz), which first searches the locations where interested functions are called, then uses dynamic taint analysis technique to classify input data into safety-related data and safety-unrelated data, finally mutates safety-related data to direct the test procedure. The evaluation shows that our method can be used to detect vulnerabilities in binary software efficiently.