Distributed Algorithms: A Case Study of the Java Memory Model

Gespeichert in:
Bibliographische Detailangaben
Titel: Distributed Algorithms: A Case Study of the Java Memory Model
Autoren: Glavan, Paola, Botinčan, Matko
Verlagsinformationen: 2007.
Publikationsjahr: 2007
Schlagwörter: Abstract state machines, Distributed algorithms, Java memory model
Beschreibung: The goal of this paper is to give a mathematically precise specification of the Java memory model and discuss its interpretation in the ASM context. We have refactored the original specification in order to clearly stipulate how it conditions the behavior of the environment. We show how each thread in a multithreaded Java program can be seen as an ordinary interactive small-step algorithm, and, consequently, how the Java program gives rise to a distributed ordinary interactive small-step ASM. Due to rather relaxed conditions on the environment imposed by the Java memory model, runs of such ASM may, however, exhibit behavior that is impossible to be observed in sequentially consistent settings. We hope that notions of run and environment capturing this kind of behavior will serve as a helpful insight for the theory of distributed algorithms developed so far.
Publikationsart: Conference object
Zugangs-URL: http://ikt.hia.no/asm07/
Dokumentencode: edsair.dris...01492..d9d5fe0f6254f6686a43842cb2e683ef
Datenbank: OpenAIRE
Beschreibung
Abstract:The goal of this paper is to give a mathematically precise specification of the Java memory model and discuss its interpretation in the ASM context. We have refactored the original specification in order to clearly stipulate how it conditions the behavior of the environment. We show how each thread in a multithreaded Java program can be seen as an ordinary interactive small-step algorithm, and, consequently, how the Java program gives rise to a distributed ordinary interactive small-step ASM. Due to rather relaxed conditions on the environment imposed by the Java memory model, runs of such ASM may, however, exhibit behavior that is impossible to be observed in sequentially consistent settings. We hope that notions of run and environment capturing this kind of behavior will serve as a helpful insight for the theory of distributed algorithms developed so far.