As the real-time demand for persistent storage access in fields such as autonomous driving and industrial robotics increases, the existing Linux block storage subsystem lacks design guarantees of temporal determinism, making it challenging to meet these capability requirements. This paper analyzes the complex Linux kernel block storage I/O stack and identifies the key factor affecting its real-time performance—head-of-line blocking(HOL)—and proposes the real-time block storage model RTB2S. The key insight in RTB2S is to resolve the HOL blocking issue by submitting requests with real-time requirements to the vicinity of the front of the device queue through multi-mapped queues and request scheduling mechanisms. Prototype experiments indicate that RTB2S effectively reduces the tail latency of real-time requests and achieves a soft real-time performance with a tail latency of P99.99 within 500μs without sacrificing throughput, significantly outperforming existing schedulers. The RTB2S design is also versatile and can be extended to various operating system block storage access designs to enhance their real-time capabilities.