HiFuzz: Hierarchical Reinforcement Learning for Semantic-Aware and Adaptive CPU Fuzzing
This paper introduces HiFuzz, a hierarchical reinforcement learning framework for processor verification that replaces mutation with a structured generation process and integrates coverage reward mechanism and semantic-aware basic block encoder.
The authors propose a hierarchical reinforcement learning framework for processor verification, replacing mutation with a structured generation process and integrating coverage reward mechanism and semantic-aware basic block encoder.
Before reading this…
Applications
- →Processor verification
To understand this paper, make sure you know these concepts first:
- Understanding of processor architecture and verification, reinforcement learning conceptsfind papers →
Abstract
More Like ThisModern processor verification struggles to reach deep architectural states due to the inefficiencies of traditional mutation-based fuzzing. We propose HiFuzz, a novel hierarchical reinforcement learning framework that replaces mutation with a structured, two-layer generation process: a Program Agent for global layout and a Basic Block Agent for precise instruction filling. To overcome reward sparsity, HiFuzz integrates an adaptive coverage reward mechanism and a semantic-aware basic block encoder providing intrinsic feedback. Extensive evaluations on three real-world RISC-V cores demonstrate that HiFuzz significantly outperforms state-of-the-art fuzzers in coverage and bug detection.