PSOFuzz: Fuzzing Processors with Particle Swarm Optimization

• DOI: 10.48550/arxiv.2307.14480
• Publication Date: 2023-01-01
• Authors: Chen Chen; Vasudev Gohil; Rahul Kande; Ahmad‐Reza Sadeghi; Jeyavijayan Rajen...
• Keywords: Fuzz testing; Speedup; Vulnerability

Hardware security vulnerabilities in computing systems compromise the defenses of not only hardware but also software running on it. Recent research has shown that fuzzing is a promising technique to efficiently detect such large-scale designs as modern processors. However, current techniques do adjust their strategies dynamically toward faster and higher design space exploration, resulting slow vulnerability detection, evident through low coverage. To address this problem, we propose PSOFuzz, which uses particle swarm optimization (PSO) schedule mutation operators generate initial input programs with objective detecting quickly. Unlike traditional PSO, finds single optimal solution, use modified PSO computes solution for selecting required explore new regions hardware. We challenge inefficient seed generation by employing PSO-based generation. Including these optimizations, our final formulation outperforms fuzzers without PSO. Experiments show PSOFuzz achieves up 15.25$\times$ speedup detection 2.22$\times$ coverage compared state-of-the-art simulation-based fuzzer.