dev test in ubuntu_stress_smoke_test hang with G-5.8 lowlatency

[Po-Hsu Lin]

Issue found on Groovy 5.8.0-42.47 lowlatency kernel with node "spitfire" on Intel cloud.

Reproduce rate: 2/2

The ubuntu_stress_smoke_test hang with the dev test on it:

 dentry STARTING
 dentry RETURNED 0
 dentry PASSED
 dev STARTING
 Timer expired (2100 sec.), nuking pid 15918
                ERROR ubuntu_stress_smoke_test.stress-smoke-test ubuntu_stress_smoke_test.stress-smoke-test timestamp=1612262460 localtime=Feb 02 10:41:00 Test timeout expired, rc=15

However, this issue didn't happen on this node with Groovy 5.8 generic kernel.

[Po-Hsu Lin (cypressyew)]

Log for spitfire with 5.8 lowlatency kernel

[Po-Hsu Lin (cypressyew)]

Log for spitfire with 5.8 generic kernel

[Po-Hsu Lin (cypressyew)]

dmesg log on node spitfire.

[Po-Hsu Lin (cypressyew)]

Re-test on node spitfire with Bionic 4.15 lowlatency kernel.

With Test suite HEAD SHA1: 9963f45

This test is still expiring with the dev subtest
03:12:08 DEBUG| [stdout] dev STARTING
03:44:48 INFO | Timer expired (2100 sec.), nuking pid 16216

[Colin Ian King (colin-king)]

[ 491.663188] INFO: task stress-ng-dev:3184 blocked for more than 120 seconds.
[ 491.663193] Tainted: G W 4.15.0-135-lowlatency #139-Ubuntu
[ 491.663194] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 491.663196] stress-ng-dev D 0 3184 3164 0x00000004
[ 491.663202] Call Trace:
[ 491.663223] __schedule+0x250/0x8a0
[ 491.663237] ? ioapic_irq_get_chip_state+0x69/0x80
[ 491.664327] schedule+0x33/0x90
[ 491.664335] synchronize_irq+0x74/0xb0
[ 491.664341] ? wait_woken+0x80/0x80
[ 491.664349] uart_port_shutdown+0x4c/0x50
[ 491.664352] uart_tty_port_shutdown+0x47/0xb0
[ 491.664358] tty_port_shutdown+0x80/0x90
[ 491.664362] tty_port_close+0x3d/0x70
[ 491.664365] uart_close+0x33/0xa0
[ 491.664369] tty_release+0x124/0x610
[ 491.664377] __fput+0xe4/0x220
[ 491.664382] ____fput+0xe/0x10
[ 491.664389] task_work_run+0x7f/0xa0
[ 491.664395] exit_to_usermode_loop+0xc0/0xd0
[ 491.664399] do_syscall_64+0x121/0x130
[ 491.664405] entry_SYSCALL_64_after_hwframe+0x41/0xa6
[ 491.664411] RIP: 0033:0x7fc4abd669c7
[ 491.664414] RSP: 002b:00007fc4a7fcdbf0 EFLAGS: 00000293 ORIG_RAX: 0000000000000003
[ 491.664417] RAX: 0000000000000000 RBX: 0000000000000004 RCX: 00007fc4abd669c7
[ 491.664419] RDX: 0000000000000000 RSI: 000000000000540d RDI: 0000000000000004
[ 491.664420] RBP: 00007fc4a7fcde90 R08: 0000000000000000 R09: 0000000000000001
[ 491.664422] R10: fffffffffffffea1 R11: 0000000000000293 R12: 000055dfb749b2a8
[ 491.664424] R13: 0000000000000004 R14: 0000000000000004 R15: 00007fff992be370

[Colin Ian King (colin-king)]

same issue on 4.15.0-134-lowlatency

[Colin Ian King (colin-king)]

same issue on 4.15.0-121-lowlatency

[Colin Ian King (colin-king)]

and on 4.15.0-101-lowlatency.

So this is not a regression but a ttyS* lockup on low latency kernels.

[Po-Hsu Lin (cypressyew)]

Retested on this node and now it has passed with this test.
Thanks

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This bug is missing log files that will aid in diagnosing the problem. While running an Ubuntu kernel (not a mainline or third-party kernel) please enter the following command in a terminal window:

apport-collect 1914221

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[Po-Hsu Lin (cypressyew)]

Remove the blocker tag as it's not a regression.