Perhaps he is referring to C99 7.14.1.1.5 "If the signal occurs other than as the result of calling the abort or raise function, the behavior is undefined if the signal handler refers to any object with static storage duration other than by assigning a value to an object declared as volatile sig_atomic_t, ..." The same part of C11 is worded as "If the signal occurs other than as the result of calling the abort or raise function, the behavior is undefined if the signal handler refers to any object with static or thread storage duration that is not a lock-free atomic object other than by assigning a value to an object declared as volatile sig_atomic_t" If a reduction, buffer copy, or similar is occurring inside a signal handler, it's violating the letter of the standard. As a practical matter, it can be made to work with careful attention to memory barriers. On Fri, Sep 21, 2012 at 3:24 PM, Jeff Hammond <[email protected]> wrote:

Why can't you just be explicit and say what the problem is?

Just show us some code that is compliant with both MPI-3.0 and C99 that does not work.

Jeff

On Fri, Sep 21, 2012 at 3:15 PM, N.M. Maclaren <[email protected]> wrote:

...

On Sep 21 2012, Jeff Hammond wrote:

...

Can you send me a link to a page that describes the C89 memory model? It would help me to write better C+Pthread code if I had a memory model to rely upon instead of nonportable inline assembly.

It's there in the standard. In C99, which is more readily accessible and very similar to C90, the relevant wording is mainly in 5.1.2.3, 6.5 and 6.5.2.2.

Regards, Nick Maclaren.

_______________________________________________ mpi-forum mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

-- Jeff Hammond Argonne Leadership Computing Facility University of Chicago Computation Institute [email protected] / (630) 252-5381 http://www.linkedin.com/in/jeffhammond https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond _______________________________________________ mpi-forum mailing list [email protected] http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum