Re: acceptance-system-centos job failure running or1k_sim

[John Snow]

On 2/11/21 5:55 AM, Philippe Mathieu-Daudé wrote:
> Hi,
>
> I got the following error with the or1k_sim image
> (acceptance-system-centos job):
> https://gitlab.com/philmd/qemu/-/jobs/1022309593
>
> (09/55) tests/acceptance:BootLinuxConsole.test_or1k_sim:  INTERRUPTED
> 22:50:22 DEBUG| CPU: OpenRISC-13 (revision 8) @20 MHz
> ...
> 22:50:23 DEBUG| console [ttyS0] enabled
> 22:50:23 DEBUG| console [ttyS0] enabled
> 22:50:23 DEBUG| bootconsole [uart0] disabled
> 22:50:23 DEBUG| bootconsole [uart0] disabled
> 22:50:23 DEBUG| Freeing unused kernel memory: 1176K
> 22:50:23 DEBUG| This architecture does not have kernel memory protection.
> 22:51:53 ERROR| RuntimeError: Test interrupted by SIGTERM
>
> Not reproducible, so probably bad luck.
>
> Still, since the artifact logs are removed after 2 days, I attached
> the debug.log to this mail.
>
> Note, earlier failures were displayed on the console which is
> archived by Gitlab so we can still access them, example:
> https://gitlab.com/qemu-project/qemu/-/jobs/793741261#L322
>
> Regards,
>
> Phil.

Hi. from the gitlab output:

 (09/55) 
tests/acceptance/boot_linux_console.py:BootLinuxConsole.test_or1k_sim: 
INTERRUPTED: Test interrupted by SIGTERM\nRunner error occurred: Timeout 
reached\nOriginal status: ERROR\n{'name': 
'09-tests/acceptance/boot_linux_console.py:BootLinuxConsole.test_or1k_sim', 
'logdir': 
'/builds/philmd/qemu/build/tests/results/job-2021-02-10T22.50-0d5d33e/... (90.74 
s)


"INTERRUPTED" for Avocado means that the test reached the avocado test 
runner's timeout, right? (IIRC, it's 90 seconds.)


From the debug log, it looks like we hung here:

22:51:53 ERROR|   File "build/tests/acceptance/boot_linux_console.py", 
line 54, in wait_for_console_pattern
22:51:53 ERROR|     vm=vm)
22:51:53 ERROR|   File 
"build/tests/acceptance/avocado_qemu/__init__.py", line 130, in 
wait_for_console_pattern
22:51:53 ERROR|     _console_interaction(test, success_message, 
failure_message, None, vm=vm)
22:51:53 ERROR|   File 
"build/tests/acceptance/avocado_qemu/__init__.py", line 82, in 
_console_interaction
22:51:53 ERROR|     msg = console.readline().strip()
22:51:53 ERROR|   File "/usr/lib64/python3.6/socket.py", line 586, in 
readinto
22:51:53 ERROR|     return self._sock.recv_into(b)


Looks like we were waiting for some pattern to show up and it didn't 
arrive within 90 seconds.

The pattern we wanted here was 'QEMU advent calendar'

It looks like the image only got this far:

22:50:23 DEBUG| console [ttyS0] enabled
22:50:23 DEBUG| console [ttyS0] enabled
22:50:23 DEBUG| bootconsole [uart0] disabled
22:50:23 DEBUG| bootconsole [uart0] disabled
22:50:23 DEBUG| Freeing unused kernel memory: 1176K
22:50:23 DEBUG| This architecture does not have kernel memory protection.

before it stalled for at least 30 seconds when it was then canceled.

--js