Deloxide: Low-overhead real-time deadlock detection and visualization framework for Rust

Context: Concurrent software is prone to deadlocks, a class of defects that Rust’s memory safety guarantees cannot prevent. Existing deadlock detection tools fall into two categories with a fundamental trade-off; they are either passive monitors that miss non-deterministic concurrency bugs or rely on heavyweight analysis techniques that are impractical in production. Objectives: This paper presents Deloxide, a real-time deadlock detection and visualization framework designed to bridge this gap for Rust. We aim for it to remain viable in production environments. Methods: We introduce a novel dual detection architecture that combines low-overhead Wait-For Graph (WFG) analysis with predictive Lock Order Graph (LOG) monitoring. To remain viable in production environments, an “Optimistic Fast Path” semantic is used. Furthermore, the proposed framework incorporates a probabilistic stress testing engine using component-based delay injection. The system provides a stateless visualization pipeline that transforms complex traces into interactive and shareable diagnostics. Results: Empirical results demonstrate that stress testing with component-based strategy achieves a 99.6% detection rate for latent deadlocks compared to 62.5% for baseline. The “Optimistic Fast Path” limits the synchronization delay to 1.07x the baseline cost per lock acquisition in default settings. A real-world ray tracing case study confirms its competitive performance and superior deterministic stability in production workloads. Conclusion: Deloxide provides a real-time deadlock detection and visualization framework designed to bridge the gap between passive monitors and heavyweight analysis for Rust.