Share on Facebook Tweet on Twitter Share on LinkedIn Share by email
RADISH: Always-On Sound and Complete Race Detection in Software and Hardware

Joseph Devietti, Benjamin P. Wood, Karin Strauss, Luis Ceze, Dan Grossman, and Shaz Qadeer

Abstract

Data-race freedom is a valuable safety property for multithreaded programs that helps with catching bugs, simplifying memory consistency model semantics, and verifying and enforcing both atomicity and determinism. Unfortunately, existing software-only dynamic race detectors are precise but slow; proposals with hardware support offer higher performance but are imprecise. Both precision and performance are necessary to achieve the many advantages always-on dynamic race detection could provide.

To resolve this trade-off, we propose RADISH, a hybrid hardware-software dynamic race detector that is always-on and fully precise. In RADISH, hardware caches a principled subset of the metadata necessary for race detection; this subset allows the vast majority of race checks to occur completely in hardware. A flexible software layer handles persistence of race detection metadata on cache evictions and occasional queries to this expanded set of metadata. We show that RADISH is correct by proving equivalence to a conventional happens-before race detector.

Our design has modest hardware complexity: caches are completely unmodified and we piggy-back on existing coherence messages but do not otherwise modify the protocol. Furthermore, RADISH can leverage type-safe languages to reduce overheads substantially. Our evaluation of a simulated 8-core RADISH processor using PARSEC benchmarks shows runtime overheads from negligible to 2x, outperforming the leading software-only race detector by 2x-37x.

Details

Publication typeInproceedings
Published inProceedings of the 39th International Symposium on Computer Architecture
PublisherIEEE
> Publications > RADISH: Always-On Sound and Complete Race Detection in Software and Hardware