RE: JavaMemoryModel: Unsynchronized read as an optimization

From: David Holmes (
Date: Mon Dec 18 2000 - 23:31:21 EST

> On 12/18/00, David Holmes <> wrote:
> > My understanding, and I'm sure Bill or Doug or someone,
> > will correct me if I am wrong, is that a variable accessed without
> > synchronization, that is not volatile, or accessed via a volatile or
> > ref, is allowed to take on any value including garbage.

Sorry Robert, my thinking is a little fuzzy today. The above is far too
sweeping as of course we have to ensure basic VM and type-safety.

I'm just trying to recall how it was anticipated that the reading of an
unrelated cache value would be avoided without forcing expensive cache
synchronization. ??? Maybe there was no way and it's just bad luck for those
(rare) architectures where it would be needed.

JavaMemoryModel mailing list -

This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:00:29 EDT