io/ompio: fix the get_byte_offset code

[edgargabriel]

fix the MPI_File_get_byte_offset bug in ompio reported by @dalcini reported on the mailing list

bot:milestone:v2.0.0
bot:label:bug

I leave it up the release managers to decide whether it truly comes in the 2.0.0 or the 2.0.1 release.

[mellanox-github]

Test PASSed.
See http://bgate.mellanox.com/jenkins/job/gh-ompi-release-pr/1736/ for details.

[ibm-ompi]

Build Failed with XL compiler! Please review the log, and get in touch if you have questions.

[jjhursey]

Ignore the IBM-CI (XL Compiler) bot - it was a local issue I'm working on resolving.

[jjhursey]

bot:retest:

[mellanox-github]

Test PASSed.
See http://bgate.mellanox.com/jenkins/job/gh-ompi-release-pr/1739/ for details.

[hppritcha]

what's the corresponding commit/pr that went in to master?

[edgargabriel]

On master it was commit d6af544
It was pr #1750

[hppritcha]

@edgargabriel please assign a reviewer

[edgargabriel]

@vvenkates27 would you mind reviewing this code? The ompi-tests/mpi2basic-tests/filetests contains a testcase for this particular scenario.

[jsquyres]

@vvenkates27 Will you be able to review this in the near future?

[mellanox-github]

Test FAILed.
See http://bgate.mellanox.com/jenkins/job/gh-ompi-release-pr/1792/ for details.

[edgargabriel]

Hm, does not look like the failed mellanox build has anything to do with my stuff

taskset -c 16,17 timeout -s SIGSEGV 10m /var/lib/jenkins/jobs/gh-ompi-release-pr/workspace-2/ompi_install1/bin/mpirun -np 4 -bind-to core -mca btl_openib_if_include mlx4_0:1 -x MXM_RDMA_PORTS=mlx4_0:1 -x UCX_NET_DEVICES=mlx4_0:1 -x UCX_TLS=rc,cm -mca pml yalla /var/lib/jenkins/jobs/gh-ompi-release-pr/workspace-2/ompi_install1/thread_tests/thread-tests-1.1/overlap 8
23:18:46 Time per iteration on each process (ms)
23:18:46 Time    Compute time    Comm time
23:18:47 [jenkins01:7761 :0] Caught signal 11 (Segmentation fault)
23:18:47 ==== backtrace ====
23:18:47  2 0x000000000005a42c mxm_handle_error()  /hpc/local/benchmarks/hpc-stack-gcc-Monday/src/install/mxm-master/src/mxm/util/debug/debug.c:641
23:18:47  3 0x000000000005a59c mxm_error_signal_handler()  /hpc/local/benchmarks/hpc-stack-gcc-Monday/src/install/mxm-master/src/mxm/util/debug/debug.c:616
23:18:47  4 0x0000003d690329a0 killpg()  ??:0
23:18:47  5 0x0000000000003637 opal_datatype_is_contiguous_memory_layout()  /var/lib/jenkins/jobs/gh-ompi-release-pr/workspace-2/ompi/mca/pml/yalla/../../../../opal/datatype/opal_datatype.h:217
23:18:47  6 0x0000000000005760 mca_pml_yalla_irecv()  /var/lib/jenkins/jobs/gh-ompi-release-pr/workspace-2/ompi/mca/pml/yalla/pml_yalla.c:348
23:18:47  7 0x000000000009bf61 PMPI_Irecv()  /var/lib/jenkins/jobs/gh-ompi-release-pr/workspace-2/ompi/mpi/c/profile/pirecv.c:78
23:18:47  8 0x0000000000400ef0 main()  ??:0
23:18:47  9 0x0000003d6901ed1d __libc_start_main()  ??:0
23:18:47 10 0x0000000000400b69 _start()  ??:0

_EDIT_ Added verbatim block

[jsquyres]

@jladd-mlnx Can you have a look?

[vvenkates27]

I have reviewed this patch, overall it looks good to me. I haven't had a chance to test and verify this as couldn't access ompi-tests. Now that I have access. I can verify this by tonight.

[opoplawski]

mpi4py tests are happy now with this.

[vvenkates27]

I now tested MPI_File_get_byte_offset with and without gaps in file-view with this branch. And all the filetest from ompi-tests also pass on my local build. So +1 verified from my side.

[jsquyres]

bot:retest
👍

[mellanox-github]

Test PASSed.
See http://bgate.mellanox.com/jenkins/job/gh-ompi-release-pr/1794/ for details.

[jsquyres]

Per discussion with @hppritcha earlier today, this is good to go.