sigio in ubuntu_stress_smoke_tests failed on X-lowlatency with node spitfire

[Po-Hsu Lin]

Issue found on with 4.4.0-204.236 - lowlatency on node spitfire in Intel cloud

Reproduce rate: 2/2

 sigio FAILED
 stress-ng: debug: [80678] 128 processors online, 128 processors configured
 stress-ng: info: [80678] dispatching hogs: 4 sigio
 stress-ng: debug: [80678] cache allocate: default cache size: 49152K
 stress-ng: debug: [80678] starting stressors
 stress-ng: debug: [80678] 4 stressors started
 stress-ng: debug: [80682] stress-ng-sigio: started [80682] (instance 3)
 stress-ng: debug: [80679] stress-ng-sigio: started [80679] (instance 0)
 stress-ng: debug: [80680] stress-ng-sigio: started [80680] (instance 1)
 stress-ng: debug: [80681] stress-ng-sigio: started [80681] (instance 2)
 stress-ng: debug: [80679] stress-ng-sigio: exited [80679] (instance 0)
 stress-ng: debug: [80680] stress-ng-sigio: exited [80680] (instance 1)
 stress-ng: debug: [80678] process [80679] terminated
 stress-ng: debug: [80678] process [80680] terminated
 stress-ng: debug: [80678] process [80681] (stress-ng-sigio) terminated on signal: 11 (Segmentation fault)
 stress-ng: debug: [80678] process [80681] terminated
 stress-ng: debug: [80678] process [80682] (stress-ng-sigio) terminated on signal: 11 (Segmentation fault)
 stress-ng: debug: [80678] process [80682] terminated
 stress-ng: info: [80678] unsuccessful run completed in 5.92s
 stress-ng: fail: [80678] sigio instance 2 corrupted bogo-ops counter, 2093457 vs 0
 stress-ng: fail: [80678] sigio instance 2 hash error in bogo-ops counter and run flag, 2817875306 vs 0
 stress-ng: fail: [80678] sigio instance 3 corrupted bogo-ops counter, 1810644 vs 0
 stress-ng: fail: [80678] sigio instance 3 hash error in bogo-ops counter and run flag, 20429112 vs 0
 stress-ng: fail: [80678] metrics-check: stressor metrics corrupted, data is compromised
 info: 5 failures reached, aborting stress process

 [ 1355.641466] stress-ng[80682]: segfault at 7fbe9f57dc80 ip 00007fbe9f372f1f sp 00007fbe9f57dc80 error 6 in ld-2.23.so[7fbe9f35b000+26000]
 [ 1355.641699] stress-ng[80681]: segfault at 7fbe9f57dc80 ip 00007fbe9f372f1f sp 00007fbe9f57dc80 error 6 in ld-2.23.so[7fbe9f35b000+26000]

[Colin Ian King (colin-king)]

I'm having problems reproducing this on my H/W. Can an instance on spitfire be set up for me so I can debug this further?

[Kleber Sacilotto de Souza (kleber-souza)]

Issue also found with xenial/linux-hwe 4.15.0-143.147~16.04.3.

[Colin Ian King (colin-king)]

@Kleber, do you have the logs for the failure in #2?

[Colin Ian King (colin-king)]

..and what machine/arch is it failing on for the failure in comment #2 ? more would be really helpful.

[Colin Ian King (colin-king)]

Urgh, was a subtle race condition on pending SIGIO once the stressor had completed and yet it was reporting an issue via the logging mechanism that triggered another SIGIO and this stacked up a bit.

Fix committed:
https://kernel.ubuntu.com/git/cking/stress-ng.git/commit/?id=364a8cc7c18a75d97f5c4096cce2a2ddfa20ed04

[Kleber Sacilotto de Souza (kleber-souza)]

SIGIO completed successfully with xenial/linux-hwe 4.15.0-143.147~16.04.3, both generic and lowlatency.

Thanks, Colin!