Re: wonky time when upgrading to potato

[Michal Jaegermann <michal@ellpspace.math.ualberta.ca>]

Paul Slootman wrote:
> 
> On Mon 31 Jul 2000, Michal Jaegermann wrote:
> 
> > Much more probable is that one of kernels in question was compiled
> > with an RTC support and the other without ....
>
> No, that's not the problem here. If it was, he wouldn't have made the
> statement "... time seemed to jump around randomly...".

This "... time seemed to jump around randomly..." is exactly what I have
seen, for example, on UP1000 with a kernel compiled without RTC support.
If you will read with hwclock in a loop some of returned values will
be correct and some will jump around.  How often one or another seems
to depend on a phase of a moon.

> 
> Please try it for yourself :-) It was hashed out here in the list
> I think almost a year ago. It is definitely an incompatibility in
> the potato libs which simply won't work on a 2.0 kernel as far as
> time is concerned.

I have no idea about libs in potato but I have seen "jumping time"
happening too many times.  In particular in many situations updating
'clock' (later 'hwclock') without touching anything else made a huge
difference.

  Michal