Worked on the DragonOS-Community/DragonOS repository to deliver an optimization for SpinLock atomic memory ordering, targeting improved concurrency performance and correctness in system-level code. The approach involved refining atomic operations in Rust, specifically using Acquire on successful compare_exchange, Relaxed on failure, Release on unlock, and Relaxed for is_locked, which reduced synchronization overhead while maintaining safety. This change enhanced the scalability and predictability of core synchronization primitives under high contention. Demonstrated advanced skills in Rust, concurrent programming, and system programming, focusing on low-level concurrency control to make synchronization more efficient and reliable in the core paths of the operating system.