<div dir="ltr">Hi Chris,<div style> Thanks, that did the trick. And it's good to know the mpich2-1.2.1 no longer allows in-place AllGather for some of my own work.</div><div style><br></div><div style>Cheers,</div><div style>
-Mark</div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Apr 5, 2013 at 9:35 AM, Christopher Daley <span dir="ltr"><<a href="mailto:cdaley@flash.uchicago.edu" target="_blank">cdaley@flash.uchicago.edu</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div bgcolor="#FFFFFF" text="#000000">
<div>Hi Mark, <br>
<br>
We fixed this issue a while back.<br>
<br>
You can apply the following patch<br>
<a href="http://flash.uchicago.edu/%7Ecdaley/FLASH3.2_patches/fix_mpi_calls/fix_mpi_calls.diff" target="_blank">http://flash.uchicago.edu/~cdaley/FLASH3.2_patches/fix_mpi_calls/fix_mpi_calls.diff</a><br>
<br>
but a better solution is to update to the latest version of FLASH.<br>
<br>
Chris<div><div class="h5"><br>
<br>
On 04/05/2013 11:27 AM, Mark L Richardson wrote:<br>
</div></div></div><div><div class="h5">
<blockquote type="cite">
<div dir="ltr">Hello Klaus,
<div> This is the same Sedov build that runs fine on
Lonestar.</div>
<div><br>
</div>
<div>The line appears to be in mpi_amr_morton_limits:</div>
<div><br>
</div>
<div>
<div>
call MPI_Allgather(morton_limits(1,1,1,mype+1),
6*4,MPI_INTEGER, & </div>
<div> &
morton_limits,6*4,MPI_INTEGER, & </div>
<div> & MPI_COMM_WORLD,ierror)</div>
<div><br>
</div>
<div>Thanks for your help,</div>
<div> -Mark</div>
<div><br>
</div>
</div>
</div>
<div class="gmail_extra"><br>
<br>
<div class="gmail_quote">On Fri, Apr 5, 2013 at 7:43 AM, Klaus
Weide <span dir="ltr"><<a href="mailto:klaus@flash.uchicago.edu" target="_blank">klaus@flash.uchicago.edu</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div>On Fri, 5 Apr 2013, Mark L Richardson wrote:<br>
<br>
> [cli_0]: aborting job:<br>
> Fatal error in PMPI_Allgather:<br>
> Invalid buffer pointer, error stack:<br>
> PMPI_Allgather(975): MPI_Allgather(sbuf=0xd665eb0,
scount=24, MPI_INTEGER,<br>
> rbuf=0xd665eb0, rcount=24, MPI_INTEGER,
MPI_COMM_WORLD) failed<br>
> PMPI_Allgather(950): Buffers must not be aliased<br>
><br>
><br>
> Has anyone dealt with this before?<br>
<br>
</div>
Mark,<br>
<br>
You seem to have somewhere in the source a call to Allgather
that<br>
improperly uses the same buffer as the first and 4th actual
argument.<br>
<br>
Try to locate this call. I assume it is in user-provided
code; if it<br>
was in a part of the FLASH code that gets exercised under
normal<br>
circumstances, we should have already run into it and fixed
it.<br>
<span><font color="#888888"><br>
Klaus<br>
</font></span></blockquote>
</div>
<br>
<br clear="all">
<div><br>
</div>
-- <br>
Mark Richardson, <br>
<a href="mailto:Mark.L.Richardson@asu.edu" target="_blank">Mark.L.Richardson@asu.edu</a>
<div>Ph.D. Candidate: Astrophysics<br>
ISTB4 623<br>
School of Earth and Space Exploration<br>
Arizona State University<br>
<a href="tel:480%20318-4449" value="+14803184449" target="_blank">480 318-4449</a><br>
<a href="http://www.public.asu.edu/%7Emlricha4" target="_blank">www.public.asu.edu/~mlricha4</a><br>
</div>
</div>
</blockquote>
<br>
</div></div></div>
</blockquote></div><br><br clear="all"><div><br></div>-- <br>Mark Richardson, <br><a href="mailto:Mark.L.Richardson@asu.edu" target="_blank">Mark.L.Richardson@asu.edu</a><div>Ph.D. Candidate: Astrophysics<br>ISTB4 623<br>
School of Earth and Space Exploration<br>Arizona State University<br>480 318-4449<br><a href="http://www.public.asu.edu/~mlricha4" target="_blank">www.public.asu.edu/~mlricha4</a><br></div>
</div>