The read-copy-update mechanism for supporting real-time applications on shared-memory multiprocessor systems with Linux

Abstract

Read-copy update (RCU) is a synchronization mechanism in the Linux TM kernel that provides significant improvements in multiprocessor scalability by eliminating the writer-delay problem of readers-writer locking. RCU implementations to date, however, have had the side effect of expanding non-preemptible regions of code, thereby degrading real-time response. We present here a variant of RCU that allows preemption of read-side critical sections and thus is better suited for real-time applications. We summarize priority-inversion issues with locking, present an overview of the RCU mechanism, discuss our counter-based adaptation of RCU for real-time use, describe an additional adaptation of RCU that permits general blocking in readside critical sections, and present performance results. We also discuss an approach for replacing the readers-writer synchronization with RCU in existing implementations. INTRODUCTIONIn this paper we focus on environments in which real-time applications are running on shared-memory multiprocessor systems with the Linux** operating system. Such environments require both realtime response and multiprocessor scalability. Realtime response means that the hardware and the operating system perform within real-time constraints; that is, the response times to certain events are subject to operational deadlines. Multiprocessor scalability means that the system can process growing amounts of work when the level of multiprocessing is proportionally increased. Tech-

Mapping of the synchronization mechanisms of the Linux kernel to the response-time analysis model

“…However, as it uses some features of the architectures of current processors such as the atomicity of operations with pointers aligned in memory, the RCU allows a writer and multiple readers in a critical section, concurrently. Thus it achieves a better performance when compared with the read-write spinlocks [3]. This Section presents the mapping of the Linux kernel mechanisms, listed in Sections 3 and 4, to the abstractions used in the response-time analysis, described in Section 2.…”

Brief announcement

“…Instances of this class are used to obtain efficient data structures such as stacks [16], queues [15], or hash tables [7]. The read-copyupdate (RCU) [8] synchronization mechanism employed by the Linux kernel is also an instance of this pattern.…”

User-Level Implementations of Read-Copy Update

“…The performance benefits of RCU are due to the fact that rcu_read_lock() and rcu_-read_unlock() are exceedingly fast. In fact, Appendix D2 in the Supplementary Material shows how these two primitives can incur exactly zero overhead, as they do in server-class Linux-kernel builds [13].…”

“…Distinct grace periods may overlap, either partially or completely. Any time period that includes a grace period is by definition itself a grace period [13,14]. Each grace period is guaranteed to complete as long as all read-side critical sections are finite in duration; thus even a constant flow of such critical sections is unable to extend an RCU grace period indefinitely.…”