an off-by-one error in dm_test_rtc_set_get()?

[Tom Rini]

On Wed, Oct 27, 2021 at 12:43:38PM +0800, Bin Meng wrote:
> Hi Simon,
> 
> gitlab reported the following test error below:
> 
> =================================== FAILURES ===================================
> __________________________ test_ut[ut_dm_rtc_set_get] __________________________
> test/py/tests/test_ut.py:43: in test_ut
> assert output.endswith('Failures: 0')
> E AssertionError: assert False
> E + where False = <built-in method endswith of str object at
> 0x7f3bb792dcb0>('Failures: 0')
> E + where <built-in method endswith of str object at 0x7f3bb792dcb0> =
> 'Test: dm_test_rtc_set_get: rtc.c\r\r\nexpected: 27/10/2021
> 03:38:15\r\r\nactual: 27/10/2021 03:38:14\r\r\ntest/dm/rtc...w, &cmp,
> 1): Expected 0x0 (0), got 0xffffffea (-22)\r\r\nTest:
> dm_test_rtc_set_get: rtc.c (flat tree)\r\r\nFailures: 1'.endswith
> ----------------------------- Captured stdout call -----------------------------
> =>
> 
> See https://source.denx.de/u-boot/custodians/u-boot-x86/-/jobs/341905
> 
> But the same branch same commit, azure test results passed:
> https://dev.azure.com/bmeng/GitHub/_build/results?buildId=460&view=results
> 
> It looks like the error is an off-by-one where actual time is 1 second
> behind the expected time?
> 
> expected: 27/10/2021 03:38:15
> actual: 27/10/2021 03:38:14
> 
> Is this a known issue?

Yes, which is why the test checks for a certain amount of "fuzz" around
the return value.  I've wondered about if we need to increase that value
slightly sometimes, or just live with hitting the re-run failed jobs
button on whatever CI system was a bit too slow sometimes.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <https://lists.denx.de/pipermail/u-boot/attachments/20211027/735dc1ef/attachment.sig>