Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 828508 - dev-python/mpi4py-3.0.3: most tests segment fault
Summary: dev-python/mpi4py-3.0.3: most tests segment fault
Status: RESOLVED FIXED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: Current packages (show other bugs)
Hardware: All Linux
: Normal major (vote)
Assignee: Alex Fan
URL:
Whiteboard:
Keywords: PullRequest, TESTFAILURE
Depends on:
Blocks:
 
Reported: 2021-12-07 12:13 UTC by Alex Fan
Modified: 2022-01-08 12:45 UTC (History)
3 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments
build log (mpi4py-3.0.3:20211207-072806.log,32.38 KB, text/plain)
2021-12-07 12:14 UTC, Alex Fan
Details
mpi4py_test.out (mpi4py_test.out,59.84 KB, text/plain)
2021-12-07 12:14 UTC, Alex Fan
Details
emerge --info (emerge.info,5.48 KB, text/plain)
2021-12-07 12:15 UTC, Alex Fan
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Alex Fan archtester 2021-12-07 12:13:07 UTC
tested on sifive unmatched (riscv64). Compilation went fine, but almost all tests segfault.

seems it is not caused by oversubscribing. I append 
> import faulthandler
> faulthandler.enable()
in the first line and run mpiexec --oversubscribe --use-hwthread-cpus --mca btl tcp,self -n 1 python3.9 -B ./test/${test} -v for each test and obtained a full log in attachment mpi4py_test.out

some excerpt of segment fault message:

[unmatch8:908877] [[46553,0],0] ORTE_ERROR_LOG: Data unpack would read past end of buffer in file /var/tmp/portage/sys-cluster/openmpi-4.1.2/work/openmpi-4.1.2/orte/util/show_help.c at line 507
Fatal Python error: Segmentation fault

Current thread 0x0000003fa8d18010 (most recent call first):
  File "/var/tmp/portage/dev-python/mpi4py-3.0.3/work/mpi4py-3.0.3-python3_9/test/test_attributes.py", line 224 in <module>
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------
--------------------------------------------------------------------------
mpiexec noticed that process rank 0 with PID 0 on node unmatch8 exited on signal 11 (Segmentation fault).


Reproducible: Always
Comment 1 Alex Fan archtester 2021-12-07 12:14:02 UTC
Created attachment 757613 [details]
build log
Comment 2 Alex Fan archtester 2021-12-07 12:14:45 UTC
Created attachment 757614 [details]
mpi4py_test.out
Comment 3 Alex Fan archtester 2021-12-07 12:15:07 UTC
Created attachment 757615 [details]
emerge --info
Comment 4 Alex Fan archtester 2022-01-04 11:35:12 UTC
I reproduced this in amd64. Not sure if it is something with my environment or config, it fails on either mpich-3.4.3 or openmpi-4.1.2.ebuild.
Comment 5 Larry the Git Cow gentoo-dev 2022-01-08 12:41:56 UTC
The bug has been closed via the following commit(s):

https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=5d31c5144882b60ac02fc40de411e665a238c253

commit 5d31c5144882b60ac02fc40de411e665a238c253
Author:     Alex Fan <alex.fan.q@gmail.com>
AuthorDate: 2022-01-05 09:29:09 +0000
Commit:     Yixun Lan <dlan@gentoo.org>
CommitDate: 2022-01-08 12:41:00 +0000

    dev-python/mpi4py: add 3.1.3
    
    bump to EAPI 8, update upstream homepage,
    remove obsolete patches, and skip unstable openmpi spawn tests
    
    Closes: https://bugs.gentoo.org/828508
    Signed-off-by: Alex Fan <alex.fan.q@gmail.com>
    Signed-off-by: Yixun Lan <dlan@gentoo.org>

 dev-python/mpi4py/Manifest            |  1 +
 dev-python/mpi4py/metadata.xml        |  3 +-
 dev-python/mpi4py/mpi4py-3.1.3.ebuild | 63 +++++++++++++++++++++++++++++++++++
 3 files changed, 65 insertions(+), 2 deletions(-)