Fuzzing With Optimized Grammar-Aware Mutation Strategies

Fuzzing is a widely used technique to discover vulnerabilities in software. However, for programs requiring highly structured inputs, the byte-based mutation strategies existing fuzzers have difficulties generating valid inputs. To resolve this challenge, Grammar-Based (GBF) utilizes grammar specifications generate new Some GBFs perform based on Abstract Syntax Trees (ASTs), which can inputs conforming grammars. neglect using feedback optimize strategies, and blindly without considering effectiveness of those In paper, we use power schedule subtree pool strategies. Specifically, first translate input files into ASTs, extract subtrees from ASTs pool. Then, AST nodes probabilistic model. That is, adaptively determine time budget mutating an node. Finally, replace along with their ones select We implement fuzzing tool demonstrate our The experiment results show that method outperforms state-of-the-art methods efficiency.