On Wed, Jan 11, 2006 at 05:16:41PM -0500, Chapman Flack wrote:
> I suppose just about anything that supports paging and unaligned
> access, a combination that had slipped my mind for a moment.
> Thanks for reminding me - looks like the assumption of atomicity
> is a good one to avoid unless the alignment situation is very
> clearly pinned down, and not worth the trouble for my purposes.

True, it is reasonably safe to assume that a reader will either
get the old value of the new one if you write without holding
the relevant lock.

	David

-- 
David Laight: david@l8s.co.uk