Previous work on the semantics of relaxed shared-memory concurrency has only considered the case in which each load reads the data of exactly one store. In practice, however, multiprocessors support mixed-size accesses, and these are used by systems software and (to some degree) exposed at the C/C++ language level. A semantic foundation for software therefore has to address them.
We investigate the mixed-size behaviour of ARMv8 and IBM POWER architectures and implementations: by experiment, by developing semantic models, by testing the correspondence between these, and by discussion with ARM and IBM staff. This turns out to be surprisingly subtle, and on the way we have to revisit the fundamental concepts of coherence and sequential consistency, which change in this setting. In particular, we show that adding a memory barrier between each instruction does not restore sequential consistency. We go on to extend the C/C++11 model to support non-atomic mixed-size memory accesses, and prove the standard compilation scheme from C11 atomics to POWER remains sound.
This is a necessary step towards semantics for real-world shared-memory concurrent code, beyond litmus tests.
Thu 19 JanDisplayed time zone: Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna change
14:20 - 16:00
|Mixed-size Concurrency: ARM, POWER, C/C++11, and SC|
Shaked Flur University of Cambridge, Susmit Sarkar University of St. Andrews, UK, Christopher Pulte University of Cambridge, Kyndylan Nienhuis University of Cambridge, Luc Maranget INRIA Rocquencourt, Kathryn E. Gray University of Cambridge, Ali Sezgin University of Cambridge, Mark Batty University of Kent, Peter Sewell University of Cambridge
|Dynamic Race Detection For C++11|
|Serializability for Eventual Consistency: Criterion, Analysis and Applications|
Lucas Brutschy ETH Zurich, Dimitar Dimitrov ETH Zurich, Switzerland, Peter Müller ETH Zurich, Martin Vechev ETH ZurichPre-print
|Thread Modularity at Many Levels: a Pearl in Compositional Verification|