On Sep 1, 2008, at 7:36 PM, <stacieweber99_at_[hidden]> <stacieweber99_at_[hidden]
> wrote:
> I am testing LAM.
> make -k check MODES="tcp"
>
> make[3]: Entering directory `/home/stacie/Documents/LAM-MPI/
> lam-7.1.4/lamtests-7.1.4/lam/dynamic'
> mpirun -x TEST -ssi cr none -s h N -np 3 -ssi rpi tcp /home/stacie/
> Documents/LAM-MPI/lam-7.1.4/lamtests-7.1.4/lam/dynamic/./
> spawn_appschema
>
> I noticed that you fixed a similar error in 7.1.3 for Fedora.
>
> I am using UBUNTU 8.04.
> Is it possible there is a problem with UBUNTU?
The problem wasn't Fedora specific, but Fedora was the first to
complain (in recent versions of the Linux kernel, a field that
shouldn't need to be used because the length of the message was 0 was
being checked to see if it was a valid pointer). There might be a
similar problem somewhere else in the code that you're hitting -- hard
to tell, and I don't have a Ubuntu box to test with.
> I found this UBUNTU patch, but don't really know if it will help.
>
> http://patches.ubuntu.com/by-release/extracted/ubuntu/m/mpich/1.2.7-8/13_upstream_patchlevel1.dpatch
Unlikely, as that patch is for the MPICH implementation of the MPI
standard, not LAM/MPI.
Unfortunately, there is currently no staff in place to work on LAM/
MPI. The development team has moved on to working on Open MPI, which
has become a mature product. I'm busy trying to finish my
dissertation and its unlikely I'll have time to work on this issue.
We'd be happy to accept a patch that fixes this problem, of course.
Brian
--
Brian Barrett
LAM/MPI Developer
Make today a LAM/MPI day!
|
