[cpp-threads] modes, pass 2

[Alexander Terekhov]

On 5/9/05, Doug Lea <dl at cs.oswego.edu> wrote:
[...]
> > Load Hit Load rule seems to prevent that; it was the last remaining piece of
> > the puzzle. :-) If isync is indeed a #loadload, 

AFAICS, it's not.

> > the isync+eieio combination
> > before RMW is a legitimate way to implement release as well.
> >
> 
> Not that I consider it authoritative, but this is my
> understanding as well.

Release needs a (leading) "export barrier". isync is known as 
"import barrier" (subsequent to RMW + control dependency).

http://www-106.ibm.com/developerworks/eserver/pdfs/archpub2.pdf

It's an acquire thing, not release.

regards,
alexander.