[concurrency-interest] question about spec for AtomicIntegerFieldUpdater

Doug Lea dl at cs.oswego.edu
Mon Dec 8 10:45:02 EST 2014


On 12/08/2014 10:33 AM, Josh Humphries wrote:

> For Power32, how do volatile semantics work for long and double values without
> word-tearing?

On Power32, there are ways to get 64bit atomic load and store, but
not those that work with 32bit-only CAS emulation. So if you use
CAS, you need more expensive lock-based forms.

>
> Also, I'm looking at the source for AbstractQueuedLongSynchronizer.setState (in
> Java SE 1.8.0_20):

Try our jsr166 sources from which you can see in CVS history that
the need for putVolatileLong even here has been forgotten and
re-remembered a few times :-)

-Doug





More information about the Concurrency-interest mailing list