Hi Michael.
There is a last thing you may do.
Try to print the info flag within the source code and outside. In mpi.h there
is the list of errors and you will know more ... if is an MPI problem or the
public property of the octave_comm holding the pointer MPI_COMM_WORLD.
You may start with hello world .. it will be fast to print the error.
If you will have always MPI_SUCCESS the problem is only within octave-comm
class that needs probably a destructor.
C++ is not my field too, but on the other hand I am learning a lot from my
mistakes.
Thanks a lot
Riccardo
--- Dom 29/11/09, Michael Creel <[email protected]> ha scritto:
Da: Michael Creel <[email protected]>
Oggetto: Re: example octave_comm_test updated
A: "Riccardo Corradini" <[email protected]>
Cc: "octave-forge list" <[email protected]>
Data: Domenica 29 novembre 2009, 08:50
Hi Riccardo,
I have tried it using Open MPI 1.3.3 compiled from source, and I get
the same results as using Open MPI 1.3.2 from Ubuntu repos. I don't
think that the difference in Open MPI versions can be the cause,
because it has been stable for a while for this simple stuff on the
common architectures. Using the versions of MPI_* in SVN, both
helloworld.m and mc_example2.m do what they're expected to, but
segfault when finishing up. Using the old versions of MPI_*, from
before the communicator stuff, there is no segfault.
My conclusion is that there is some problem in the communicator
functions. I'll try to figure it out, but this is not an area I know
about.
Cheers, M.
On Sat, Nov 28, 2009 at 8:16 PM, Riccardo Corradini
<[email protected]> wrote:
>
> Michael,
> I have no problems with helloworld but I use 1.3.3 stable openmpi try to
> compile it form source ( look at my explanations in the package on configure
> options) I had no problems with all the examples.
> http://www.open-mpi.org/software/ompi/v1.3/downloads/openmpi-1.3.3.tar.bz2
> I will tell you more about my configuration when I will be back in my office
> on Monday.
> Bests
> Riccardo
>
> --- Sab 28/11/09, Michael Creel <[email protected]> ha scritto:
>
> Da: Michael Creel <[email protected]>
> Oggetto: Re: example octave_comm_test updated
> A: "Riccardo Corradini" <[email protected]>
> Cc: "Jaroslav Hajek" <[email protected]>, "octave-forge list"
> <[email protected]>
> Data: Sabato 28 novembre 2009, 16:10
>
> Hi Riccardo,
> I tried running the latest helloworld.m which uses the new comm stuff, and I
> get the same errors on exit. I'm posting the output below, in case it's
> helpful. Please let me know if I can provide any other useful information.
> Cheers, Michael
>
> mich...@yosemite:~/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/inst$
> mpirun -np 2 octave -q --eval helloworld
> We are at rank 0 that is master etc..
> Greetings from process: 1!
> *** glibc detected *** octave: double free or corruption (fasttop):
> 0x0000000002c24440 ***
> ======= Backtrace: =========
> /lib/libc.so.6[0x7f576b757dd6]
> /lib/libc.so.6(cfree+0x6c)[0x7f576b75c70c]
> /usr/lib/libstdc++.so.6(_ZNSsD1Ev+0x39)[0x7f576bf930c9]
> /lib/libc.so.6(exit+0xe2)[0x7f576b71ac12]
> /usr/lib/octave-3.2.2/liboctinterp.so(octave_main+0xe1c)[0x7f57720a49ac]
> /lib/libc.so.6(__libc_start_main+0xfd)[0x7f576b700abd]
> octave[0x400879]
> ======= Memory map: ========
> 00400000-00401000 r-xp 00000000 08:04 142133
> /usr/bin/octave-3.2.2
> 00600000-00601000 r--p 00000000 08:04 142133
> /usr/bin/octave-3.2.2
> 00601000-00602000 rw-p 00001000 08:04 142133
> /usr/bin/octave-3.2.2
> 02351000-02c34000 rw-p 00000000 00:00 0
> [heap]
> 418a3000-418a5000 rwxp 00000000 00:0f 1382
> /dev/zero
> 7f5758000000-7f5758021000 rw-p 00000000 00:00 0
> 7f5758021000-7f575c000000 ---p 00000000 00:00 0
> 7f575ca88000-7f575ca9a000 r-xp 00000000 08:04 288805
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Finalize.oct
> 7f575ca9a000-7f575cc9a000 ---p 00012000 08:04 288805
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Finalize.oct
> 7f575cc9a000-7f575cc9c000 r--p 00012000 08:04 288805
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Finalize.oct
> 7f575cc9c000-7f575cc9d000 rw-p 00014000 08:04 288805
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Finalize.oct
> 7f575cc9d000-7f575ccd0000 r-xp 00000000 08:04 288809
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Send.oct
> 7f575ccd0000-7f575ced0000 ---p 00033000 08:04 288809
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Send.oct
> 7f575ced0000-7f575ced3000 r--p 00033000 08:04 288809
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Send.oct
> 7f575ced3000-7f575ced4000 rw-p 00036000 08:04 288809
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Send.oct
> 7f575ced4000-7f575ceef000 r-xp 00000000 08:04 288803
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_size.oct
> 7f575ceef000-7f575d0ef000 ---p 0001b000 08:04 288803
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_size.oct
> 7f575d0ef000-7f575d0f1000 r--p 0001b000 08:04 288803
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_size.oct
> 7f575d0f1000-7f575d0f2000 rw-p 0001d000 08:04 288803
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_size.oct
> 7f5762be5000-7f5762c00000 r-xp 00000000 08:04 285390
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_rank.oct
> 7f5762c00000-7f5762e00000 ---p 0001b000 08:04 285390
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_rank.oct
> 7f5762e00000-7f5762e02000 r--p 0001b000 08:04 285390
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_rank.oct
> 7f5762e02000-7f5762e03000 rw-p 0001d000 08:04 285390
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/MPI_Comm_rank.oct
> 7f5762e03000-7f5762e1d000 r-xp 00000000 08:04 288820
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/octave_comm_make.oct
> 7f5762e1d000-7f576301d000 ---p 0001a000 08:04 288820
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/octave_comm_make.oct
> 7f576301d000-7f576301f000 r--p 0001a000 08:04 288820
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/octave_comm_make.oct
> 7f576301f000-7f5763020000 rw-p 0001c000 08:04 288820
> /home/michael/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/src/octave_comm_make.oct
> 7f5765697000-7f5765699000 r-xp 00000000 08:04 548
> /lib/libutil-2.10.1.so
> 7f5765699000-7f5765898000 ---p 00002000 08:04 548
> /lib/libutil-2.10.1.so
> 7f5765898000-7f5765899000 r--p 00001000 08:04 548
> /lib/libutil-2.10.1.so
> 7f5765899000-7f576589a000 rw-p 00002000 08:04 548
> /lib/libutil-2.10.1.so
> 7f576589a000-7f57658e7000 r-xp 00000000 08:04 140193
> /usr/lib/openmpi/lib/libopen-pal.so.0.0.0
> 7f57658e7000-7f5765ae7000 ---p 0004d000 08:04 140193
> /usr/lib/openmpi/lib/libopen-pal.so.0.0.0
> 7f5765ae7000-7f5765ae8000 r--p 0004d000 08:04 140193
> /usr/lib/openmpi/lib/libopen-pal.so.0.0.0
> 7f5765ae8000-7f5765aea000 rw-p 0004e000 08:04 140193
> /usr/lib/openmpi/lib/libopen-pal.so.0.0.0
> 7f5765aea000-7f5765b0d000 rw-p 00000000 00:00 0
> 7f5765b0d000-7f5765b50000 r-xp 00000000 08:04 140194
> /usr/lib/openmpi/lib/libopen-rte.so.0.0.0
> 7f5765b50000-7f5765d50000 ---p 00043000 08:04 140194
> /usr/lib/openmpi/lib/libopen-rte.so.0.0.0
> 7f5765d50000-7f5765d51000 r--p 00043000 08:04 140194
> /usr/lib/openmpi/lib/libopen-rte.so.0.0.0
> 7f5765d51000-7f5765d53000 rw-p 00044000 08:04 140194
> /usr/lib/openmpi/lib/libopen-rte.so.0.0.0
> 7f5765d53000-7f5765d55000 rw-p 00000000 00:00 0
> 7f5765d55000-7f5765de2000 r-xp 00000000 08:04 140189
> /usr/lib/openmpi/lib/libmpi.so.0.0.0--------------------------------------------------------------------------
> mpirun noticed that process rank 1 with PID 2599 on node yosemite exited on
> signal 6 (Aborted).
> --------------------------------------------------------------------------
> 2 total processes killed (some possibly by mpirun during cleanup)
> panic: Segmentation fault -- stopping myself...
> attempting to save variables to `octave-core'...
> panic: Aborted -- stopping myself...
> attempting to save variables to `octave-core'...
> panic: attempted clean up apparently failed -- aborting...
> mich...@yosemite:~/Desktop/of/octave/trunk/octave-forge/extra/openmpi_ext/inst$
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev
