On 07/05/2018 09:35 PM, Michael Davidsaver wrote: > Also, I though I had test coverage of this bug. That's actually how I > noticed it to begin with. But it seems my later change to allow for a > slow test runner also stopped testing readback of the 12/24 hour mode bit. > It just silently uses whichever it reads. I'll be re-issuing an updated > version which restores this check. Then you will be able to easily > see the effect of reverting 'timer: ds1338 persist 12-hour mode selection'.
I've posted revised versions of two of the three test patches #6 and #7 (which I've hopefully posted correctly...). #6 again tests for this issue. So you can demonstrate the problem by either applying 1,2,4-6, or just 1 and 6 to see that the issue is present in the original implementation. The test failure should be: > test_rtc_set: assertion failed (mode_expect == mode_actual): (1 == 0) Which shows that a write with 12 hour mode is read back as 24 hour mode. Similarly, omitting patch #5 will cause the tests added in #7 to fail.
signature.asc
Description: OpenPGP digital signature
