A Large-scale Parallel Fuzzing System

The application of parallelization to fuzzing can effectively improve test efficiency. But with the expansion of node scale, synchronization mechanism will become the bottleneck. To solve this problem, this paper presents a method which is suitable for large-scale parallelization to generate test cases. It simplifies the execution path of tree form into a one-dimensional array by preprocessing, which ensures validity and reduces processing time. This paper also designs and implements a parallel fuzzing system using this method. The system uses a polling mechanism to reduce repetitive tasks. A jump-oriented strategy is adopted to reduce redundancy when filtering crashes. At the end of this paper, the effectiveness of the system in improving the efficiency of fuzzing is further demonstrated through experiments.