[cpp-threads] RE: "Agenda" for august 23-25 concurrency meeting
Doug Lea
dl at cs.oswego.edu
Wed Sep 6 17:30:27 BST 2006
Alexander Terekhov wrote:
> On 9/2/06, Herb Sutter <hsutter at microsoft.com> wrote:
>> P.S.: For the examples I just gave below, assume transitivity doesn't
>> hold, and
>> then the question: What if I used locks to protect x and y?
>
> Then it's all fine and dandy. Lack of remote write atomicity is not an
> issue if you use locks to protect x and y.
This sounded familiar, but it took me a few days before I recalled
where I saw a proof of (essentially) this a decade ago. See
http://hal.ccsd.cnrs.fr/docs/00/07/41/23/PDF/RR-2557.pdf
From Causal Consistency to Sequential Consistency in Shared Memory Systems
M Raynal, A Schiper - Proc. 15th Int. Conf. FST&TCS
for this and some other observations about relationships between
sequential consistency and a version of causal consistency
that I'm still not sure is the same as Herb is proposing.
-Doug
More information about the cpp-threads
mailing list