Page MenuHomePhabricator

Segfault in refreshLinks job
Open, Needs TriagePublic

Description

Hi, I saw this:

Oct 29 08:09:34 mwtask141 kernel: [124675.764099] php[988430]: segfault at 7fd0377c2cf0 ip 00007fd0377c2cf0 sp 00007fd032ffec08 error 14 in libnss_files-2.31.so[7fd0377d3000+3000]
Oct 29 08:09:34 mwtask141 kernel: [124675.764128] Code: Unable to access opcode bytes at RIP 0x7fd0377c2cc6.
Oct 29 08:09:34 mwtask141 jobrunner[589261]: 2022-10-29T08:09:34+0000 ERROR: Runner loop 0 process in slot 0 gave status '139':
Oct 29 08:09:34 mwtask141 jobrunner[589261]: php /srv/mediawiki/w/maintenance/runJobs.php --wiki='constantnoblewiki' --type='refreshLinks' --maxtime='30' --memory-limit='500M' --maxjobs 1 --result=json STDERR:
Oct 29 08:09:34 mwtask141 jobrunner[589261]: #011Segmentation fault


.....



Oct 29 08:22:08 mwtask141 kernel: [125430.159906] php[1007797]: segfault at 7f94e9b7fcf0 ip 00007f94e9b7fcf0 sp 00007f94e55fec08 error 14
Oct 29 08:22:08 mwtask141 kernel: [125430.159919] Code: Unable to access opcode bytes at RIP 0x7f94e9b7fcc6.
Oct 29 08:22:08 mwtask141 jobrunner[589261]: 2022-10-29T08:22:08+0000 ERROR: Runner loop 0 process in slot 0 gave status '139':
Oct 29 08:22:08 mwtask141 jobrunner[589261]: php /srv/mediawiki/w/maintenance/runJobs.php --wiki='constantnoblewiki' --type='refreshLinks' --maxtime='60' --memory-limit='500M' --maxjobs 1 --result=json STDERR:
Oct 29 08:22:08 mwtask141 jobrunner[589261]: #011Segmentation fault

I'm not exactly sure how to reproduce it though.

Running mediawiki 1.38 and php 7.4 (on debian bullseye).

Event Timeline

Is it happening repeatedly? Or just seen once?

Seems it's happening repeatedly. Although we have securelinkfixer installed and domain.php is a large file (not sure if securelinkfixer is causing it but I was told that it could be by @Universal_Omega). We have the job using memory limit of 500 in the jobrunner. Not sure if it needs more memory or something.