[cpp-threads] RE: volatile, memory models, threads

[Boehm, Hans]

[Doug wrote:]
> But given the discussions about qualifiers etc, it doesn't 
> seem to me that we can revive this proposal. The main problem 
> is that we only care about and can give semantics for simple 
> scalar types.
> We don't know what to make of the others, and other parts of 
> the C and C++ specs don't seem to give us room to do anything 
> reasonable.
> 
I didn't yet completely arrive at the same conclusion, though you may be
right.

I don't think there's a fundamental problem with having volatile imply
ordering.  The issue is what to do with atomicity for objects that don't
have size and/or alignment to support hardware atomicity.

I have assumed that it's reasonable that accesses to only certain scalar
types are atomic for volatile objects.  For the rest, volatile implies
ordering/visibility constraints, but not atomicity.  That may be
controversial.

I don't know whether we have to address misaligned scalars.  I guess
they still exist in standard code with standard ABIs on M68Ks, and maybe
a few other older architectures?

Hans