The Art of Multiprocessor Programming by Maurice Herlihy & Nir Shavit

Author:Maurice Herlihy & Nir Shavit
Language: eng
Format: epub
ISBN: 9780123977953
Publisher: Elsevier Inc.
Published: 2012-06-01T16:00:00+00:00

Figure 11.6 The LockFreeExchanger<T> Class.

Notice that the algorithm allows the inserted item to be null, something used later in the elimination array construction. There is no ABA problem because the compareAndSet() call that changes the state never inspects the item. A successful exchange’s linearization point occurs when the second thread to arrive changes the state from WAITING to BUSY (Line 34). At this point both exchange() calls overlap, and the exchange is committed to being successful. An unsuccessful exchange’s linearization point occurs when the timeout exception is thrown.

The algorithm is lock-free because overlapping exchange() calls with sufficient time to exchange will fail only if other exchanges are repeatedly succeeding. Clearly, too short an exchange time can cause a thread never to succeed, so care must be taken when choosing timeout durations.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.