These questions arose out of my careful reading of ticket 302, so it would make sense if you were willing to read them, since I think they are related.

Jeff

On Fri, Dec 16, 2011 at 10:38 AM, Jeff Squyres <jsquyres@cisco.com> wrote:

+1 / good points.

Will you make a proposal about this?

On Dec 16, 2011, at 10:44 AM, Jeff Hammond wrote:

> Are MPI_Initialized and MPI_Query_thread guaranteed to be thread-safe
> by all implementations? If these MPI calls cannot be made in a fully
> multithreaded context, how can a multithreaded library determine that
> it can make MPI calls in an unrestricted fashion, i.e. determine if
> MPI was initialized with MPI_THREAD_MULTIPLE? Must the multithreaded
> library mutex around these two MPI calls and assume that
> MPI_THREAD_SERIALIZED is always available?
>
> Can MPI_Is_thread_main be called from a thread that is not main when
> only MPI_THREAD_FUNNELED is supported? I recall that the funneled
> mode stipulates that only the main thread can make MPI calls and there
> is no text excepting this routine from that constraint.
>
> Thanks,
>
> Jeff
>
> --
> Jeff Hammond
> Argonne Leadership Computing Facility
> University of Chicago Computation Institute
> jhammond@alcf.anl.gov / (630) 252-5381
> http://www.linkedin.com/in/jeffhammond
> https://wiki-old.alcf.anl.gov/index.php/User:Jhammond

> _______________________________________________
> mpi-forum mailing list
> mpi-forum@lists.mpi-forum.org
> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

--
Jeff Squyres
jsquyres@cisco.com
For corporate legal information go to:
http://www.cisco.com/web/about/doing_business/legal/cri/

_______________________________________________
mpi-forum mailing list
mpi-forum@lists.mpi-forum.org
http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

--
Jeff Hammond
Argonne Leadership Computing Facility
University of Chicago Computation Institute
jhammond@alcf.anl.gov / (630) 252-5381
http://www.linkedin.com/in/jeffhammond
https://wiki-old.alcf.anl.gov/index.php/User:Jhammond