[mpich-discuss] MPICH-related runtime error

[Dave Goodell]

After just skimming that code right now, it looks like you are  
probably passing the same buffer to your send and recv calls.  This  
isn't permitted by the MPI Standard.

In the short term you could probably reconfigure MPICH2 without  
assertions enabled, via --enable-fast=ndebug.

-Dave

On Dec 1, 2009, at 11:22 AM, Judy Karpen wrote:

> Hi,
> I am running MPICH2v1.2 on a Mac with OS10.5.8 and the Intel Fortran  
> compiler. I am trying to run a 1D hydro-AMR code that requires MPI  
> although it only runs on a single processor (for reasons that are  
> not relevant here). It compiled OK, but I am getting the following  
> error when it runs:
>
>> Assertion failed in file /Users/jtk/Downloads/mpich2-1.2/src/mpi/ 
>> coll/helper_fns.c at line 337: 0
>> memcpy argument memory ranges overlap, dst_=0x100802730  
>> src_=0x100802730 len_=4
>>
>> internal ABORT - process 0
>> rank 0 in job 8  gs67-hsdjkmac.gsfc.nasa.gov_63272   caused  
>> collective abort of all ranks
>>   exit status of rank 0: killed by signal 9
>
> I tracked down the code subroutine where it is dying, and determined  
> that it is calling the helper_fns.c program (which is part of the  
> MPICH library) when an MPI send/receive is invoked. Unfortunately I  
> am only a Fortran programmer not C, so the error message isn't  
> telling me much. If anyone has an idea of what could cause such an  
> error, please let me know. Thanks for your help!
>
>
> Judy Karpen
> Code 674
> NASA/Goddard Space Flight Center
> Greenbelt, MD 20771
> email: judy.karpen at nasa.gov
>
> "For every problem there is one solution that is simple, neat, and  
> wrong" -- H.L. Mencken
>
>
>
>
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss