[cpp-threads] Somewhat relevant technical report

[Nick Maclaren]

> > As I see it, the only reason this might not be guaranteed is that
> > pthread_mutex_try_lock doesn't guarantee memory synchronization in
> > this case.
> >
> > It's unclear what that statement means.
> 
> To me it means exactly what it says. A failed pthread_mutex_trylock is a 
> no-op with respect to memory synchronization and visibility, a { return 
> false; }. 

Sigh.  No, it CLEARLY doesn't mean that.

It says that it is UNSPECIFIED, which is not the same as saying that
it is a null operation.  I agree that it means that memory MAY NOT
be synchronised - i.e. it MAY be implemented as:

    Synchronise state of mutex
    Attempt to get mutex
    Return if failed
    Synchronise memory
    Return

But it need not be.  It could be implemented as:

    Synchronise memory
    Attempt to get mutex
    Return

HOWEVER, Hans is right.  The pthread_mutex_lock did succeed (so memory
is synchronised), and the pthread_mutex_trylock is then temporarily
ordered with respect to that (because it fails).  Like Hans, I have
no idea what POSIX means - but, unlike him, I doubt that there is even
a consistent intent behind the specification.

I described some of the issues in a previous posting - the one that
referred to global, transitive or pairwise synchronisation - and that
posting is very relevant here.  Until and unless we can agree on the
basics of the model, there will be an indefinite stream of such, er,
cross-purpose debates.


Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email:  nmm1 at cam.ac.uk
Tel.:  +44 1223 334761    Fax:  +44 1223 334679