Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17(1):164--182, January 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Other researchers have considered wait free and lock free implementations using instructions that are stronger than simple reads and writes. Implementations of various types of queues have been presented by Lamport [53] by Herlihy and Wing [38] by Israeli and Rappoport [41] by Wing and Gong [81, 82], and byMichael and Scott [64] Anderson and Woll [15] and Lanin and Shasha [56] present implementations for various set operations. Valois presents lock free implementations for various data structures, including queues, lists, trees, and dictionaries [78, 79, 80] Finally, Massalin and Pu have ....

J. Wing and C. Gong. Testing and verifying concurrent objects. Journal of Parallel and Distributed Computing, 17(2):164--182, December 1993.

....object under consideration to improve performance. These techniques also suffer from the complexity of reasoning about correctness. Some techniques use instructions stronger than normal reads and writes. Various concurrent queue implementations not relying on mutual exclusion fall in this category [67, 75, 99, 122, 167]. Valois proposed lock free implementations of common data structures such as queues, trees, and lists [164, 165] Massalin and Pu implemented an operating system using only lock free synchronization techniques [119] The terms non blocking synchronization and lock free synchronization have been ....

Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17(2):164--182, January/February 1993.

....components is itself linearizable and hence sequentially consistent. Linearizability offers the potential for scalability, as no central mechanism is required to enforce system sequencing. Several authors have identified linearizable implementations of data structures, notably queues and stacks[5, 9, 12, 23, 25]. These implementations are decentralised and have been demonstrated to be capable of providing scalable performance[9, 23] As an aside, the solution to the correctness problem proposed in [17] appears to be an application of the principle of Linearizability[12, 25] The semantics for concurrent ....

.... queues and stacks[5, 9, 12, 23, 25] These implementations are decentralised and have been demonstrated to be capable of providing scalable performance[9, 23] As an aside, the solution to the correctness problem proposed in [17] appears to be an application of the principle of Linearizability[12, 25]. The semantics for concurrent operations are defined by using a sequential semantics for operations on the data item combined with the assertion that these operations are linearizable[25] This provides a strong, and perhaps overly strict semantics. There is a well known tradeoff between ....

[Article contains additional citation context not shown here]

Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17:164--182, 1993. 13

....implementations of concurrent objects compose to form a correct system. The sequential consistency of an object implementation alone is insufficient to guarantee sequential consistency in a composite system[14, 15] The work of Herlihy and Wing has introduced the concept of Linearizability[14, 18]. Linearizability is a stronger condition than sequential consistency that has the property that linearizable objects compose to form linearizable systems. In section 5, the behaviour of the queue is analysed using linearizability and an outline of a linearizability proof for the queue is ....

....sequential consistency condition is itself inadequate for this purpose (section 3) as the sequential consistency of a component does not guarantee the sequential consistency of a system built from such components. An alternative model with well defined compositional properties is linearizability[14, 18]. In the literature at least one example has been found of an implementation that exhibits high performance and linearizable behaviour, but that this behaviour may fail under certain circumstances. To characterise this behaviour, the notion of ffl linearizability has been introduced[17] This ....

[Article contains additional citation context not shown here]

Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17:164--182, 1993.

.... [27] for material on CCS; Clarke et.al [12] and Manna and Pnueli [23, especially part II, pages 177 387] and [24] for a basic understanding of safety, liveness, and fairness properties and how to express them; Stirling [36, 37, 38, 39, 40] for links between CCS and process logics; Lynch and Wing [22, 21, 46] for some practical applications. 17 Summary of part I We have demonstrated the core of Demos, a typical process oriented discrete event simulation language, and given informal translations to two different process calculi, CCS and SCCS. Such translations can be used to reason formally about the ....

J. M. Wing and C. Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17:164--182, 1993.

....should return the same results as if occuring in some sequence, consistent with program order. It is known however that the sequential consistency of a concurrently accessed data item is itself insufficient to guarantee that a system composed with many such items is sequentially consistent[15, 24, . To restore this property, either a system must be collectively co ordinated[3, 2, or the concurrent accesses to data items must be implemented is such a way that these items naturally compose to form sequentially consistent systems. This second technique removes a potential source of ....

....] To date, work on concurrent data abstractions has tended to focus on performance issues. Whilst acknowledging and pursuing performance, this paper mainly addresses the problem of system building by considering the compositional properties of the queue implementation using linearizability. In [15, 24, , relatively inefficient queue implementations are presented that are linearizable. More recent work has identified more efficient implementations that it is claimed are linearizable[4, 22, This paper is a contribution to this literature, presenting the implementation of a highly scalable FIFO ....

[Article contains additional citation context not shown here]

Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17:164--182, 1993.

....in different orders. Deadlock avoidance can be awkward if processes must lock multiple data objects, particularly if the set of objects is not known in advance. A number of researchers have investigated techniques for implementing lock free concurrent data structures using software techniques [2, 4, 19, 25, 26, 32]. Experimental evidence suggests that in the absence of inversion, convoying, or deadlock, software implementations of lockfree data structures often do not perform as well as their locking based counterparts. This paper introduces transactionalmemory, a new multiprocessor architecture intended to ....

J. Wing and C. Gong. Testing and verifying concurrent objects. Journal of Parallel and Distributed Computing, 17(2), February 1993.

No context found.

Jeanette M. Wing and Chun Gong. Testing and Verifying Concurrent Objects. Journal of Parallel and Distributed Computing, 17(1):164--182, January 1993.

No context found.

J. Wing and C. Gong, "Testing and Verifying Concurrent Objects", Journal of Parallel and Distributed Computing, 17(2), 1993, pp. 164-182.

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