An Observational Approach to Defining Linearizability on Weak Memory Models - Archive ouverte HAL Access content directly
Conference Papers Year : 2017

An Observational Approach to Defining Linearizability on Weak Memory Models

(1) , (2)
1
2
John Derrick
  • Function : Author
  • PersonId : 1024703
Graeme Smith
  • Function : Author
  • PersonId : 1024704

Abstract

In this paper we present a framework for defining linearizability on weak memory models. The purpose of the framework is to be able to define the correctness of concurrent algorithms in a uniform way across a variety of memory models. To do so linearizability is defined within the framework in terms of memory order as opposed to program order. Such a generalisation of the original definition of linearizability enables it to be applied to non-sequentially consistent architectures. It also allows the definition to be given in terms of observable effects rather than being dependent on an understanding of the weak memory model architecture. We illustrate the framework on the TSO (Total Store Order) weak memory model, and show that it respects existing definitions of linearizability on TSO.
Fichier principal
Vignette du fichier
446833_1_En_8_Chapter.pdf (273.38 Ko) Télécharger le fichier
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-01658421 , version 1 (07-12-2017)

Licence

Attribution - CC BY 4.0

Identifiers

Cite

John Derrick, Graeme Smith. An Observational Approach to Defining Linearizability on Weak Memory Models. 37th International Conference on Formal Techniques for Distributed Objects, Components, and Systems (FORTE), Jun 2017, Neuchâtel, Switzerland. pp.108-123, ⟨10.1007/978-3-319-60225-7_8⟩. ⟨hal-01658421⟩
56 View
33 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More