UCL logo

UCL Discovery

UCL home » Library Services » Electronic resources » UCL Discovery

Soundness of data flow analyses for weak memory models

Alglave, J; Kroening, D; Lugton, J; Nimal, V; Tautschnig, M; (2011) Soundness of data flow analyses for weak memory models. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) , 7078 LNCS 272 - 288. 10.1007/978-3-642-25318-8_21.

Full text not available from this repository.


Modern multi-core microprocessors implement weak memory consistency models; programming for these architectures is a challenge. This paper solves a problem open for ten years, and originally posed by Rinard: we identify sufficient conditions for a data flow analysis to be sound w.r.t. weak memory models. We first identify a class of analyses that are sound, and provide a formal proof of soundness at the level of trace semantics. Then we discuss how analyses unsound with respect to weak memory models can be repaired via a fixed point iteration, and provide experimental data on the runtime overhead of this method. © 2011 Springer-Verlag.

Title:Soundness of data flow analyses for weak memory models
UCL classification:UCL > School of BEAMS > Faculty of Engineering Science > Computer Science

Archive Staff Only: edit this record