jlindgren@espus.com (Jon Lindgren) writes:
> I've recently started using NTP 4.0.99i on an SS5, running
> NetBSD-1.4.2.  I've recompiled the kernel to include pps support.  I'm
> using a gpsclock as a primary source of NTP info.

Ah, I just tried this the other day too!  I was having a blast single
stepping through ntp with gdb trying to figure out what was going on.
Then I tried to exit gdb.  The kernel hung and I got to watch a 37 gig
fsck.  I decided I really didn't want to jeopardize that computer's
filesystem for a bit of fun with PPS.

> With the patches supplied by www.gpsclock.com, I'm running into some
> problems.  The offending call seems to be the time_pps_kcbind call, which
> returns "bad address."  The same code proportedly works under FreeBSD and
> Linux.

I read somewhere that linux and freebsd use a later version of the PPS
spec.  Doing a quick diff of the original and later PPS specs might
point you at some minor change that is causing these problems.

-wolfgang
-- 
       Wolfgang Rupprecht <wolfgang+gnus@dailyplanet.wsrcc.com>
		    http://www.wsrcc.com/wolfgang/
DGPS signals via the Internet  http://www.wsrcc.com/wolfgang/gps/dgps-ip.html