[M3devel] Proposed Atomic Word.T interfaces.

[hendrik at topoi.pooq.com]

On Sat, Dec 19, 2009 at 06:30:47PM -0500, Tony Hosking wrote:
> Here is my updated proposal for Modula-3 Atomic Word.T.
> These are based on what we can have implemented from libatomic_ops, a prototype implementation for the C++ memory model standardization process.  At the moment they are not gcc intrinsics, but I expect to see that coming as the standard progresses.
> 
> INTERFACE Atomic;
> 
> IMPORT Word;
> 
> TYPE T = Word.T;
> 
> TYPE
>   Order = { Relaxed, Release, Acquire, AcquireRelease, Sequential };
>   (* "Relaxed": The operation does not order memory.
> 
>      "Release": Performs a release operation on the affected memory locations,
>      thus making regular memory writes visible to other threads through the
>      variable to which it is applied.
> 
>      "Acquire": Performs an acquire operation on the affected memory
>      locations, thus making regular memory writes in other threads released
>      through the atomic variable to which it is applied, visible to the
>      current thread.

It's not quite clear to me what these Orders mean.  In particular,
I don't see a clear model for the interaction of 'visibility', 'memory 
writes' and 'threads'.  I can certainly find a ways to understand these 
words, but I can't be sure I understand them the same way as they 
are intended.

Perhaps I lack context that is provided somewhere.

-- hendrik