Rainer Orth <[EMAIL PROTECTED]> wrote:
> Joerg Schilling writes:
>
> > Don't forget the missing support for cheap DCF77 receivers.
> >
> > This needs driver support (e.g. save kernel hr time triggered on external
> > status interrupts) and user level support.
>
> Both user level and kernel support are available with the PARSE driver and
> STREAMS modules in both xntp3 and ntp4. The devices work without the
> STREAMS module, but precision is probably better with it. Unfortunately,
I am not sure about the quality of this solution.
I did look at available solutions in 1993 and everything was a desaster except
for a zs driver addition that implements the kernel part. What Solaris currently
implements is not useful as it includes an auto-drift compensation without
filtering and unaware for the missing 59th second pulse.
My user space code implements a filter that I did improve over many years.
It allows to throw away bad samples and avoids fast drifts while still
being able to capture the clock for less than 10 ms error in less than 10
minutes.
As my code controls nsec_per_tick in the kernel, it allows to have less than
5 ms error after one day without radio receiption.
Jörg
--
EMail:[EMAIL PROTECTED] (home) Jörg Schilling D-13353 Berlin
[EMAIL PROTECTED] (uni)
[EMAIL PROTECTED] (work) Blog: http://schily.blogspot.com/
URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
