This really shouldn't happen. There are two possibilities for why this
is happening - one is that there is a bug in LAM. The other is that
you are violating the MPI standard, causing LAM to do something stupid.
Without knowing what your code looks like, I really can't offer advice
on how to fix the issue.
Do you have a small example application that causes these valgrind
errors?
Brian
On Nov 19, 2003, at 9:20 AM, Michael Sabielny wrote:
> Hi,
>
> during a valgrind debug session with a small parallel program I
> noticed this
> message from valgrid:
>
> ==13513== Source and destination overlap in memcpy(0x435feebc,
> 0x435feebc, 8)
> ==13513== at 0x40025BD5: memcpy (mac_replace_strmem.c:95)
> ==13513== by 0x82716E4: lam_memcpy (in
> /home/tems/concept.f90/be/pconcept.be)
>
> I am using a self compiled version of LAM, configured with these
> options:
> ./configure --prefix=/opt/intel/compiler70/lam \
> --with-prefix-memcpy \
> --with-purify \
> --with-trillium \
> --with-rsh=rsh \
> --with-tm=/opt/pbs
>
> As you see LAM uses the prefix-memcpy-feature as well as the purify
> feature
> for usage of LAM together with valgrind. Does this message point to a
> possible bug or is it a wanted behavior?
>
>
> Regards,
> Michael
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
Brian Barrett
LAM/MPI developer and all around nice guy
Have a LAM/MPI day: http://www.lam-mpi.org/
|
