J. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, May 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....consider several aspects of memory management. No node should be reclaimed and then later re allocated while some other process is traversing this node. This can be solved for example by careful reference counting. We have selected to use the lock free memory management scheme invented by Valois [19] and corrected by Michael and Scott [ 10] which makes use of the FAA and CAS atomic synchronization primitives. To insert or delete a node from the list we have to change the respective set of next pointers. These have to be changed consistently, but not necessary all at once. Our solution is to ....

....be deleted as well, as illustrated in Figure 4. There are several solutions to this problem. One solution is to use the CAS2 operation as it can change two pointers atomically, but this operation is not available in any existing multiprocessor system. A second solution is to insert auxiliary nodes [19] between each two normal nodes, and the latest method introduced by Harris [4] is to use one bit of the pointer values as a deletion mark. On most modem 32 bit systems, 32 bit values can only be located at addresses that are evenly dividable by 4, therefore bits 0 and 1 of the address are always ....

J. D. VALOIS. Lock-Free Data Structures. PhD. The- sis', Rensselaer Polytechnic Institute, Troy, New York, 1995.

.... overhead of a software implemented multi word CAS) counters and bit fields with widths up to 8 bytes, stacks, and FIFO queues [Tre86, MS96] Valois introduced a lock free singly linked list design supporting insertions and deletions anywhere in a list, as well as several other data structures [Val95a, Val95b] These designs also work with just CAS. However, Greenwald [Gre99] has criticized them for being quite complex, difficult to get right, and computationally expensive. Most of the algorithms for lock free data structure synchronization that have been developed recently assume ....

John D. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, May 1995.

....list. Stone describes a non blocking algorithm based on a circular queue. Massalin and Pu [8] present a non blocking array based queue which requires the double compare andswap atomic primitive that is available only on some members of the Motorola 68000 family of processors. Valois in [12] presents a non blocking queue algorithm together with several other non blocking data structures, his queue is an array based one. Michael and Scott in [10] presented a nonblocking queue based on a singly link list, which is the most eOEcient and scalable non blocking algorithm compared with the ....

J. D. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Department of Computer Science, 1995.

....guaranteed to complete operations in a nite number of steps [44] As we are more concerned with the eciency of synchronization algorithms, waitfreedom is often too expensive to implement for synchronization. On the other hand, lock free algorithms with careful implementation can be quite ecient [124]. In [75] the authors described an algorithm for constructing busy wait synchronization that aims to reduce lock contentions. It is well known that conservative parallel simulation cannot beat the critical path [50] To make things worse, it is considered as impractical to identify and therefore ....

J. D. Valois. Lock-free data structures. PhD thesis, May 1995. Rensselaer Polytechnic Institute.

....sequentialize the access. Furthermore locks have the problem that if the process with the lock is blocked (e.g. swapped out by the OS or dies) then all processes can become blocked. To avoid this problem many non blocking (or lock free) implementations of data structures have been suggested [1, 2, 9, 10, 17, 18, 24, 25]. As with the versions that use This work was supported in part by the National Science Foundation under grants CCR 9706572 and CCR 0085982. Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made ....

.... there has been considerable work on non blocking (or lock free) data structures [10] which only require that some user s request will complete in a bounded number of steps (although any particular user can be delayed indefinitely) Examples of non blocking data structures work includes [1, 2, 9, 10, 11, 17, 18, 24, 25]. Most of these implementations still fully sequentialize access to the data structure. Moreover, they often require unbounded memory 9 , or the use of atomic operations on two or more words of memory (such as a double compare andswap or transactional memory [12, 20] Such operations are ....

J. D. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Troy, NY, 1995.

....the overhead of a softwareimplemented multi word CAS) counters and bitfields with widths up to 8 bytes, stacks, and FIFO queues. 21, 18] Valois introduced a lock free single linked list design supporting insertions and deletions anywhere in a list, as well as several other data structures [23, 22]. These designs also work with just CAS. However, Greenwald [6] has criticized them for being quite complex, difficult to get right, and computationally expensive. Most of the algorithms for lock free data structure synchronization that have been developed recently assume availability of a ....

John D. Volois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, May 1995.

....to complete operations in a finite number of steps [3] As we are more concerned with the efficiency of synchronization algorithms, wait freedom is often too expensive to implement for synchronization. On the other hand, lock free algorithms with careful implementation can be quite efficient [13]. In [8] the authors described an algorithm for constructing busy wait synchronization that aims to reduce lock contentions. Particularly, in our model of conservative parallel simulation, since the scheduler needs a snapshot of all incoming channels of an LP to calculate its safe time, the ....

J. Valois. Lock-free data structures. Ph.D Thesis, Rensselaer Polytechnic Institute, May 1995.

....in a structure called memory mapping database. Since more than one activity can operate over this structure, it must support concurrent operations. This can either be implemented by short, atomic instructions disabling interrupts or nonblocking fully preemptable synchronization as described in [Val95] In the following, we present a structure for a 64 bit version of the L4 kernel on Alpha [Sch96] that allows to use the technique of hierarchical address space management under real time constraints. Finally, we give some measurements to prove the ability of our technique. 2 Objects and ....

John .D Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Troy New York, 5 1995.

....among these iterations is exploited. Lock Free Task Queue Since the task queues are accessed many times (at least once for each scheduling event) the overhead of queue operations needs to be small. IML implements the task queue as a concurrent access non blocking stack based on Valois algorithm [16]. 2 The enabling factor was the IA32 instruction lock CMPXCHG8B, which performs an atomic compare and exchange operation on an eight byte quantity aligned at a four byte boundary. Fig. 3(a) and Fig. 3(b) illustrate enqueue and dequeue operations of the pointer P2, respectively. In the enqueue ....

J. D. Valois. Lock-Free Data Strucutres. PhD thesis, Rensselaer Polytechnic Institute, 1995.

....paradigm. The nanothreads runtime system uses distributed run queues to assist locality based scheduling and a central run queue used for load balancing purposes [13] In order to alleviate the overhead of queuing we implemented the run queues as non blocking stacks which allow concurrent updates [15]. The run queues are coupled in a one to one manner with memory pools used for recycling thread contexts. The memory pools were also implemented as non blocking stacks which take advantage of reusing cached thread contexts [10] Spin locks were not employed in order to avoid introducing hot spots ....

J. Valois. Lock-Free Data Structures. PhD Dissertation, Rensselaer Polytechnic Institute, 1995.

....among these iterations is exploited. Lock Free Task Queue Since the task queues are accessed many times (at least once for each scheduling event) the overhead of queue operations needs to be small. IML implements the task queue as a concurrent access non blocking stack based on Valois algorithm [13]. 2 The enabling factor was the IA32 instruction lock CMPXCHG8B, which performs an atomic compare and exchange operation on an eight byte quantity aligned at a four byte boundary. Fig. 3(a) and Fig. 3(b) illustrate enqueue and dequeue operations of the pointer P2, respectively. In the enqueue ....

J. D. Valois. Lock-Free Data Strucutres. PhD thesis, Rensselaer Polytechnic Institute, 1995.

....also writes its requests for processors in the appropriate field of the shared arena, when these requests vary during its execution life. Another augmentation to the nanothreading library is the addition of non blocking synchronization primitives for enqueuing and dequeuing user level threads[17]. An alternative approach would be the use of locks. However, this approach has proven to be quite inefficient in the presence of frequent, undesired EV preemptions by the kernel. If an EV gets preempted while the user level thread running on top of it holds a queue lock, all enqueuing and ....

J. Valois, Lock-Free Data Structures, PhD Dissertation, Rensselaer Polytechnic Institute, 1995.

....locations are universal atomic primitives. At least one of the first two is implemented in most commodity microprocessors. In this paper, we evaluate three non blocking and linearizable algorithms that implement concurrent FIFO queues. They are attributed to Prakash, Lee and Johnson [24] Valois [27] and Michael and Scott [21] respectively. All three algorithms use compare and swap to atomically update the queue and share a common technique for concurrent updates. The algorithms maintain pointers to the two ends of the queue. An enqueuing dequeuing process reads the pointer to the tail head ....

J. Valois, Lock-Free Data Structures, PhD Dissertation, Rensselaer Polytechnic Institute, 1995.

....solutions to more general problems with any number of either writers or readers. However, parallelism among concurrent processes is reduced especially within multiprocessor environments. As discussed in the literature, there are other negative effects of employing lock based mutual exclusion [23, 18, 39]. An alternative solution is to avoid mutual exclusion, allowing concurrent reading while shared data is being updated [21, 23, 31] Such algorithms are referred to as being lock free or non blocking because shared data may be accessed, without being locked, by the reader and the writer at any ....

Valois, J. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, 1995.

....Many of the papers present their algorithms withoutexperimental results. In general, the performance of non blockingalgorithms resulting from general methodologies is acknowledged to be significantly inferior to that of the corresponding lock based implementations. Prakash et al. 19, 21] Valois [24, 25], and Michael and Scott [17] proposed non blocking implementations of concurrent link based queues. Treiber [22] proposed a non blocking implementation of concurrent link based stacks. Valois [26] proposed a non blocking implementation of linked lists. Anderson and Woll [3] proposed a non blocking ....

J. D. Valois. Lock-Free Data Structures. Ph. D. dissertation, Rensselaer Polytechnic Institute, May 1995.

....locations are universal atomic primitives. At least one of the first two is implemented in most commodity microprocessors. In this paper, we evaluate three non blocking and linearizable algorithms that implement concurrent FIFO queues. They are attributed to Prakash, Lee and Johnson [21] Valois [24] and Michael and Scott [18] respectively. All three algorithms use compare and swap to atomically update the queue and share a common technique for concurrent updates. The algorithms maintain pointers to the two ends of the queue. An enqueuing dequeuing process reads the pointer to the tail head ....

J. Valois, Lock-Free Data Structures, PhD Dissertation, Rensselaer Polytechnic Institute, 1995.

....the data structure and determines if there is another operation in progress. If so it tries to complete the ongoing operation and then takes another snapshot of the data structure. Otherwise it tries to complete its own operation. The process keeps trying until it completes its operation. Valois [41, 42] presented a list based non blocking queue algorithm that avoids the contention caused by the snapshots of Prakash et al. s algorithm and allows more concurrency by keeping a dummy node at the head (dequeue end) of a singly linked list, thus simplifying the special cases associated with empty and ....

J. D. Valois. Lock-Free Data Structures. Ph. D. dissertation, Rensselaer Polytechnic Institute, May 1995.

No context found.

J.D. Valois. Lock-Free Data Structures. PhD. Thesis, Rensselaer Polytechnic Institute, Troy, New York, 1995.

....that more than one process can see this simultaneously) To ensure that only one 3 process can do this, Compare Swap is used at line 4 in the algorithm to turn off the claim bit. 4 Conclusion Although the original algorithms were formally verified as well as actually implemented and tested [3], the error described here was not detected. Formal verification did not identify the error because the assumption was made, in order to simplify the proofs, that memory was not reused; hence, the error could not occur. The error was not discovered during implementation because, like many bugs in ....

J. D. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Department of Computer Science, 1995.

....a process, and Reclaim returns a cell no longer being used to the set of free cells. For brevity, we describe only a very simple implementation of this object, in which free cells must all be of the same size. are kept on a simple list. Much more elaborate schemes are possible; in particular, in [28] we show how to extend these ideas to implement a lock free buddy system which provides management of variable sized cells. We keep cells which are not in use on a free list. New cells are allocated by removing them from the front of the list, and cells are reclaimed by putting them back on the ....

....with similar data structures that use spin locks. The most time consuming operation is most likely performing a SafeRead on each cell as we traverse the list; it would be useful to have this operation implemented in hardware. Preliminary performance analysis of these algorithms can be found in [28]; however, more work remains to be done in order to quantitatively determine the performance tradeoffs between algorithms such as these and more traditional methods using mutual exclusion. We are currently examining the performance of these algorithms and data structures experimentally. ....

J. D. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Department of Computer Science, 1995.

No context found.

J. Valois. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, May 1995.

No context found.

VALOIS, J. D. Lock-Free Data Structures. PhD thesis, Rensselaer Polytechnic Institute, Department of Computer Science, 1995.

No context found.

J. D. Valois, Lock-free data structures, Ph.D. dissertation, Rensselaer Polytechnic Institute, Troy, New York, 1995.

No context found.

Valois, J.D. Lock-Free Data Structures. Ph.D. Thesis. Rensselaer Polytechnic Institute, Troy, New York, May 1995.

No context found.

J. D. Valois. Lock-Free Data Structures. Ph.D. thesis, Rensselaer Polytechnic Institute, Troy, NY, 1995.

First 50 documents

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