[mvapich-discuss] Win_allocate -> MPIR_Shmem_Bcast_MV2 failing

[Mingzhe Li]

Hi Jeff,

Thanks for verifying that.

Just for everybody's information, this issue is resolved by setting
MV2_USE_SLOT_SHMEM_COLL=0.

Regards,
Mingzhe

On Wed, Apr 9, 2014 at 12:22 PM, Jeff Hammond <jeff.science at gmail.com>wrote:

> Sorry for the delay.  Thanks for your help.  Indeed it fixes my error.
>
> Best,
>
> Jeff
>
> [jhammond at b17 tests]$ /home/jhammond/MPI/gcc482-mv2trunk/bin/mpiexec
> -n 4 ./test_malloc_group
> ARMCI Group test starting on 4 procs
>  + Creating odd/even groups
>  + Performing group allocation
>
> test_malloc_group:80078 terminated with signal 11 at PC=3336689a31
> SP=7fffa021da98.  Backtrace:
> /lib64/libc.so.6(memcpy+0xe1)[0x3336689a31]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(mv2_shm_bcast+0x2ab)[0x2b92d6a86fbb]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPIR_Shmem_Bcast_MV2+0x186)[0x2b92d6a6ed06]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(+0xbb327)[0x2b92d69f8327]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPIDI_CH3U_Win_allocate+0x79)[0x2b92d69e7869]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPID_Win_allocate+0x9f)[0x2b92d69ee4ef]
>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(PMPI_Win_allocate+0x221)[0x2b92d6ad6a01]
> ./test_malloc_group[0x402eb6]
> ./test_malloc_group[0x402a66]
> ./test_malloc_group[0x401e0a]
> /lib64/libc.so.6(__libc_start_main+0xfd)[0x333661ed1d]
> ./test_malloc_group[0x401bc9]
>
>
> ===================================================================================
> =   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
> =   PID 80078 RUNNING AT b17
> =   EXIT CODE: 1
> =   CLEANING UP REMAINING PROCESSES
> =   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
>
> ===================================================================================
>
>
> [jhammond at b17 tests]$ /home/jhammond/MPI/gcc482-mv2trunk/bin/mpiexec
> -env MV2_USE_SLOT_SHMEM_COLL=0 -n 4 ./test_malloc_group
> ARMCI Group test starting on 4 procs
>  + Creating odd/even groups
>  + Performing group allocation
>  + Freeing group allocation
>  + Freeing group
>  + done
>
>
> On Fri, Apr 4, 2014 at 1:44 PM, Mingzhe Li <li.2192 at osu.edu> wrote:
> > Hi Jeff,
> >
> > Just want to follow up with you about that issue. Did you have chance to
> try
> > that runtime parameter?
> >
> > Thanks,
> > Mingzhe
> >
> >
> > On Fri, Mar 28, 2014 at 6:01 PM, Mingzhe Li <li.2192 at osu.edu> wrote:
> >>
> >> Hi Jeff,
> >>
> >> We took a look at this issue and had a quick fix for that. Could you try
> >> setting runtime parameter MV2_USE_SLOT_SHMEM_COLL=0?
> >>
> >> Mingzhe
> >>
> >>
> >> On Thu, Mar 27, 2014 at 5:13 PM, Jeff Hammond <jeff.science at gmail.com>
> >> wrote:
> >>>
> >>> This is an ARMCI-MPI test that succeeds with MPICH, MVAPICH2
> >>> non-ch3:psm, CrayMPI and mostly recently, OpenMPI.
> >>>
> >>> I do not need this for NWChem but I'd like it to work so that I can
> >>> say that ARMCI-MPI works on Qlogic IB.
> >>>
> >>> Thanks,
> >>>
> >>> Jeff
> >>>
> >>> [jhammond at blogin2 tests]$
> >>> /home/jhammond/MPI/gcc482-mv2trunk/bin/mpiexec -n 4
> >>> ./test_malloc_group
> >>> ARMCI Group test starting on 4 procs
> >>>  + Creating odd/even groups
> >>>  + Performing group allocation
> >>>
> >>> test_malloc_group:117675 terminated with signal 11 at PC=3b6ea89a31
> >>> SP=7fff4a9df1c8.  Backtrace:
> >>> /lib64/libc.so.6(memcpy+0xe1)[0x3b6ea89a31]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(mv2_shm_bcast+0x2ab)[0x2ba692f30fbb]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPIR_Shmem_Bcast_MV2+0x186)[0x2ba692f18d06]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(+0xbb327)[0x2ba692ea2327]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPIDI_CH3U_Win_allocate+0x79)[0x2ba692e91869]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(MPID_Win_allocate+0x9f)[0x2ba692e984ef]
> >>>
> >>>
> /home/jhammond/MPI/gcc482-mv2trunk/lib/libmpich.so.12(PMPI_Win_allocate+0x221)[0x2ba692f80a01]
> >>> ./test_malloc_group[0x402eb6]
> >>> ./test_malloc_group[0x402a66]
> >>> ./test_malloc_group[0x401e0a]
> >>> /lib64/libc.so.6(__libc_start_main+0xfd)[0x3b6ea1ed1d]
> >>> ./test_malloc_group[0x401bc9]
> >>>
> >>>
> >>>
> ===================================================================================
> >>> =   BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
> >>> =   PID 117675 RUNNING AT blogin2
> >>> =   EXIT CODE: 1
> >>> =   CLEANING UP REMAINING PROCESSES
> >>> =   YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
> >>>
> >>>
> ===================================================================================
> >>>
> >>> HOW TO REPRODUCE
> >>>
> >>> git clone git://git.mpich.org/armci-mpi.git && cd armci-mpi && git
> >>> checkout mpi3rma && ./autogen.sh && mkdir build && cd build &&
> >>> ../configure CC=mpicc && make check
> >>>
> >>>
> >>>
> >>> --
> >>> Jeff Hammond
> >>> jeff.science at gmail.com
> >>>
> >>> _______________________________________________
> >>> mvapich-discuss mailing list
> >>> mvapich-discuss at cse.ohio-state.edu
> >>> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
> >>>
> >>>
> >>
> >
>
>
>
> --
> Jeff Hammond
> jeff.science at gmail.com
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20140409/0bc95615/attachment.html>