power-management/rtc fails due to missing file

Is this the power-management/rtc test?

what happens if you run

sudo hwclock -r

?

-0.500808 seconds

Interesting, so why does it time out?

can you confirm it's the power-management/rtc test that's giving you trouble? I may be looking at the wrong one.

confirmed...power-management/rtc "waiting for clock tick timed out: no such file or directory"

This is very odd, all the test does is "sudo hwclock -r", which you already confirmed works. Plus the last report (#4) says no such file or directory... what is going on :)

This bug is affecting testing across the board, but it seems related to something that's borking the rtc prior to the test; I created a minimal whitelist:

miscellanea/submission-resources
rtc
__power-management__
power-management/rtc

running this through checkbox-gui passes the test.

This was of course after I ran just power-management/rtc using plainbox -i, to try to narrow it down to a gui-specific problem, which it turns out not to be.

Hi Chris, I've been unable to recreate this bug under observable conditions, can you tell me which whitelist you used to get this error? it seems related to some test that runs before rtc and borks it, but short of running one of my longer (>200 tests) whitelists, I haven't had luck pinpointing the borky test. Maybe the whitelist you're using will be shorter and easier to work with.

This problem in itself is not that bad, the main issue is that the suspend tests depend on this, so they often get blocked.

This problem is intermittent; I haven't seen any reports recently but I was reviewing some old submissions and they do have it.

We could remove the suspend test dependency on this,

Looks like a dupe of: https://bugs.launchpad.net/fwts/+bug/1333569

According to that bug, rtc gets borked if the system hibernated. So the sequence of events that may cause this is:

- Due to whitelist ordering, the system hibernates before running rtc.
- Tests are run twice. Thus, the hibernate test from the first test run borks rtc, and the second rtc run will fail.

I'll try to reproduce this a bit more, but we can try to ensure that this rtc test comes early in the whitelist and be aware that due to a kernel bug, it can't work if the system has hibernated in the past.

Were you ever able to reproduce consistently with the whitelist?

Not yet, as I have no system with Ubuntu desktop that is capable of hibernating :( I'll do an install as soon as possible and report on the results.

OK, I confirmed this happens because of bug 1333569.

For CDTS, the only thing we can do is check all our whitelists and ensure the rtc test runs *before* any hibernate sequences.

If someone runs CDTS twice without rebooting, though, the problem will still occur and we have no way of fixing this in CDTS, other than documenting the bug and advising people to reboot before rerunning an entire test suite (or be aware that rtc may fail mysteriously).

Yay, at least one whitelist has hibernate_advanced before power-management/rtc. I'll submit a merge request fixing this.