Subject: Re: NTP and PPS_SYNC option
To: Simon Burge <>
From: john heasley <>
List: tech-kern
Date: 12/15/2005 18:12:16
Fri, Dec 16, 2005 at 12:46:26PM +1100, Simon Burge:
> john heasley wrote:
> > I do not think I fully understand the PPSAPI either.  Looking at com(4),
> > and the RFC, it seems to exist for userland; to make PPS events available
> > to userland and provide the ability to adjust PPS parameters, including
> > which source (if any) calls hardpps().
> That's as I understand it too.
> Back to your original question, ntpd will use it to get pretty much the
> exact time of the PPS signal instead of just relying on the "relatively"
> approximate normal timestamp from the driver, and thus will be more
> accurate.  Adding PPSAPI support shouldn't add to much to your driver.

Thanks.  Guess I should stop being lazy about it.

> > the clock on this particular MB is heinous.  A reboot usually looses at
> > least a few seconds.  To the point that, in order to recover in any
> > reasonable timeframe, we added a wrapper that takes the timestamp from
> > the reference clock, calls settimeofday(), then starts ntpd.
> Cute.  I wonder if you can get the same effect playing around with the
> "burst" and/or "iburst" server keywords?

Hrm, we have not experimented with those knobs.  I'll try after the current
stuff has had an opportunity to run for 24hrs.  Reducing minpoll to 4
definitely improved the situation.