Re: [chrony-users] RTC Trimming Issues

[Ed W <lists@xxxxxxxxxxxxxx>]

On 02/11/2012 13:40, John.Florian@xxxxxxxx wrote:

> - I haven't seen any exhibit this after the clock is set once

This is where I think our situations differ.  I do have some that go wonky **again** even after I've set the clock.  Maybe we're setting the clock differently.  What method specifically do you use? The hwclock command, the BIOS, something else?

Or maybe, my situation is just like yours but in this same batch, I have a few boards that genuinely have bad batteries.

Yeah, I thought you meant it was the ones with bad batteries which were persisting with the problem - otherwise by definition they should have roughly accurate clocks after each reboot and probably you wouldn't have noticed they were badly out?

I guess with a bad battery you are booting from an uninitialised state each time?

Also you seem to confirm that a warm reboot correctly initialises the hardware, it's just the cold boot which is an issue.  This hints at some kernel driver failing to initialise something correctly.  I think if you have a single machine where it's reproducible, especially if you can clearly see the RTC driver failing to initialise in the dmesg logs, then you could take that to the kernel mailing list and get some help debugging the RTC driver?  Basically find a machine with dead battery where it's reproducible and don't fix the battery!

Please copy me in on any kernel correspondence?

Ed