High-Performance Synchronization for User-Level Applications

点击翻译按钮获取中文摘要

Multiprocessor systems use synchronization primitives to coordinate theactivities of multiple threads of control. Spin locks in particular arewidely used in multiprocessor operating systems and in scientificappliactions. With the proliferation of multiprocessor servers, theselocks have come to be widely used in commercial applications as well.The multiprogrammed nature of server workloads, however, requires thatthreads be able to "time out" and abandon an attempt to acquire a lock.Timeout is easy on small machines, which can use traditional"test-and-set" spin locks, but these locks do not scale to largemachines. The principal alternative--scheduler-based locks such asthose provided by Java--has also proven to be prohibitively expensive.The proposed research aims to address the cost of user-levelsynchronization in multiprocessor servers by means of two principaltechniques: (1) the incorporation of timeout in scalable queue-basedlocks, and (2) the development of practical mechanisms for theconstruction of lock-free data structures. These techniques will beevaluated with respect to each other and to existing techniques, andwill be incorporated into threaded run-time systems such as the JavaVirtual Machine.