Overview
Syllabus
Intro
Existing direction: Hybrid fuzzing
Obstacle: Hybrid fuzzing is not efficient
Reason: Hybrid fuzzing is not incremental
Observation: Constraints are discarded directly
Observation: Seeds provide limited guidance for further input generation
Motivating example
Goal: Incremental hybrid fuzzing
Challenge: How to preserve the constraints?
Intuition: Interpret the input search space
Existing challenges to generate inputs
Solution 1: prune the infeasible paths for constraint solver
Solution 2: accelerate solving speed for feasible path
Evaluation
Bug detection
How can guided constraint solving improve efficiency?
Key takeaway
Taught by
IEEE Symposium on Security and Privacy