P. A. Buhr, M. Fortier, and M. H. Co#n. Monitor classification. ACM Computing Surveys, 27(1):63--107, Mar. 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....version of) Abadi and Leino s proof system allows to factor out the proof obligations from the structural part of the proof. In [23] the logic is implemented in the Lego theorem prover. A survey about monitors in general, including proof rules for various monitor semantics, can be found in [13]. Formal semantics of Java, including multithreaded execution, and its virtual machine in terms of abstract state machines is given in [38] A structural operational semantics of multithreaded Java can be found in [14] Future work Based on the proof theory presented, we are currently developing ....

P. A. Buhr, M. Fortier, and M. H. Co#n. Monitor classification. ACM Computing Surveys, 27(1):63--107, Mar. 1995.

....Armed with this meta lock, we now proceed to implement the monitor operations: lock, unlock, wait, and notify. Because the meta lock arbitrates access among contending threads, we can implement monitor operations using a number of different data structures and offer a variety of 11 semantics [7]. We have chosen an implementation that uses a simple linked list of lock records and gives equal preference to awakened waiters and newly arrived threads. 4.3 Locking and unlocking objects Acquiring and releasing the monitor lock of an object corresponds to entering and exiting the object s ....

....two potential shortcomings of our simple linked list data structure: lack of fairness and long searches through queues. First, consider fairness. Motivated by Buhr et al. who classify and compare a spectrum of monitor styles that offer different trade offs between performance and fairness [7], we programmed a version that gives preference to awakened waiters (so called priority non blocking monitors ) To provide this preference, we replaced the single queue with three queues, holding entering, waiting, and awakened threads, respectively. Now it is possible to find and give ....

Peter A. Buhr, Michel Fortier, and Michael H. Coffin. Monitor Classification. ACM Computing Surveys, 27(1), p. 63-107, March 1995.

....is achieved with guard conditions that block the executing process until the boolean conditions of the guard are satisfied. No other granularity of mutual exclusion and no other synchronization mechanism is available. Orca s notion of guards is similar to the notion of an automatic signal monitor [Buhr et al. 1995] . However, since the classical notion of a monitor implies a central resource and no support for replication, Orca s objects are fundamentally different from monitors in their implementation. Orca s objects are abstract data types with support for mutual exclusion and declarative guard based ....

....code to maintain correctness when accessing shared data. Also, the programmer can reason about the synchronization behaviour of a shared data object by inspecting the static guard conditions instead of having to consider the run time behaviour of a traditional monitor s signal and wait operations [Buhr et al. 1995] . Furthermore, the shared data object model automatically maintains data coherence while transparently replicating and migrating shared data as needed to improve the locality of memory accesses. The automatic mutual exclusion, synchronization, and data coherence of the shared data object model ....

P.A. Buhr, M. Fortier, and M.H. Coffin. Monitor Classification. ACM Computing Surveys, 27(1):63--107, March 1995.

....the readers writers problem. Therefore, it needs to reconsider two subproblems in the view of implementation. Usually, the reader writer lock supported in operating system or user defined library is usually implemented with four procedures: EnterReader, EnterWriter, ExitReader and ExitWriter 6 [13, 17, 2]. Those procedures represent entrance and exit of readers and writers respectively as shown in Figure 4. Scheduling of readers and writers is done at the points executing those procedures. We define those points executing procedures as scheduling points and assume that they are atomic. That is, ....

P. A. Buhr, M. Fortier, and M. H. Coffin. Monitor classification. ACM Comput. Surveys, 27(1):64--107, 1995.

....priority task. Condition variables and their associated internal queues of waiting tasks are also a fundamental part of monitors [23] Signalling a condition variable makes the task at the head of the queue eligible to run. The next task to execute in the monitor depends on the kind of monitor [10]. For this discussion it is sufficient to look CHAPTER 3. REAL TIME SCHEDULING 40 at the major division of monitors into priority and no priority monitors 4 . Priority monitors give preference to tasks that have already waited to enter the monitor and no priority monitors do not. Hence, for a ....

Peter A. Buhr, Michel Fortier, and Michael H. Coffin. Monitor classification. ACM Computing Surveys, 27(1):63--107, March 1995.

No context found.

Buhr, P. A., Fortier, M., e Coffin, M. H, "Monitor Classification", ACM 177 Computing Surveys, Vol. 27, No. 1, pp. 63-108, Maro de 1995.

No context found.

Buhr P. A. and Fortier M., "Monitor Classification", ACM Computing Surveys, Vol. 27 No. 1, pp63-107, March 1995

No context found.

Buhr P. A. and Fortier M., "Monitor Classification", ACM Computing Surveys, Vol. 27 No. 1, pp63-107, March 1995

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC