1.0.6-2 segfaults on Trusty when performing multiple disk IO threads

[Jeff Lane ]

When running this command using 1.0.6-2 on Trusty, I get segfaults until I remove the disk IO threads. The maximum threads I can run is 7 on my machine, which has 16 cores (HT enabled).

Each time, this appears in syslog:
Mar 31 20:22:31 optimistic-tank kernel: [16787.596152] traps: stressapptest[11794] general protection ip:404904 sp:7fff60c72f00 error:0 in stressapptest[400000+47000]

and attached is some output that shows more about what's going on.

The command used on this failing machine is:
sudo stressapptest -v 20 -s 300 -f //temp/disk_stress_data-1-1427835829 --filesize 402653184 -f //temp/disk_stress_data-2-1427835829 --filesize 402653184 -f //temp/disk_stress_data-3-1427835829 --filesize 402653184 -f //temp/disk_stress_data-4-1427835829 --filesize 402653184 -f //temp/disk_stress_data-5-1427835829 --filesize 402653184 -f //temp/disk_stress_data-6-1427835829 --filesize 402653184 -f //temp/disk_stress_data-7-1427835829 --filesize 402653184 -f //temp/disk_stress_data-8-1427835829 --filesize 402653184 -f //temp/disk_stress_data-9-1427835829 --filesize 402653184 -f //temp/disk_stress_data-10-1427835829 --filesize 402653184 -f //temp/disk_stress_data-11-1427835829 --filesize 402653184 -f //temp/disk_stress_data-12-1427835829 --filesize 402653184 -f //temp/disk_stress_data-13-1427835829 --filesize 402653184 -f //temp/disk_stress_data-14-1427835829 --filesize 402653184 -f //temp/disk_stress_data-15-1427835829 --filesize 402653184 -f //temp/disk_stress_data-16-1427835829 --filesize 402653184

I pulled the latest version from google, whcih is 1.0.7, and compiled it locally. 1.0.7 was able to successfullly run using all 16 IO Threads with no segfaulting. The test still failed, for some reason, but at least it is not segfaulting any longer.

[Jeff Lane  (bladernr)]

Interesting... unable to recreate this now... weird :/ it was 100% recreatable when I filed this.