From owner-freebsd-current  Tue Dec 26 19:09:11 1995
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id TAA28449
          for current-outgoing; Tue, 26 Dec 1995 19:09:11 -0800 (PST)
Received: from haywire.DIALix.COM (news@haywire.DIALix.COM [192.203.228.65])
          by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id TAA28439
          for <freebsd-current@freebsd.org>; Tue, 26 Dec 1995 19:09:02 -0800 (PST)
Received: (from news@localhost) by haywire.DIALix.COM (sendmail) id LAA08742 for freebsd-current@freebsd.org; Wed, 27 Dec 1995 11:08:57 +0800 (WST)
Received: from GATEWAY by haywire.DIALix.COM with netnews
	for freebsd-current@freebsd.org (problems to: usenet@haywire.dialix.com)
To: freebsd-current@freebsd.org
Date: 27 Dec 1995 11:08:52 +0800
From: peter@haywire.dialix.com (Peter Wemm)
Message-ID: <4bqdc4$8h2$1@haywire.DIALix.COM>
Organization: DIALix Services, Perth, Australia.
References: <199512261708.LAA14134@miller.cs.uwm.edu>
Subject: Re: Tick, tock, adjust the clock
Sender: owner-current@freebsd.org
Precedence: bulk

james@miller.cs.uwm.edu (Jim Lowe) writes:

>There seems to be a problem with the Pentium clock on my mainboard 
>with FreeBSD-current (as of a few weeks ago).   It doesn't seem to
>keep correct time.  I didn't have this problem when I ran a 486 system.
>I know for US$10 or a subscription to Sports Illustrated, I can get a
>football watch that keeps better time than my computer :-)!

>The mainboard is an ASUS P55TP4XE, but I also had the same problem
>with a SuperMicro mainboard.  I run xtnpd which adjusts the clock
>at a fairly regular interval:

>...
>Dec 26 02:54:02 miller-genuine-draft xntpd[75]: time reset (step) -1.803289 s
>Dec 26 03:01:45 miller-genuine-draft xntpd[75]: time reset (step) 1.913048 s
>Dec 26 03:08:09 miller-genuine-draft xntpd[75]: time reset (step) 0.465431 s
>Dec 26 03:14:19 miller-genuine-draft xntpd[75]: time reset (step) -1.514804 s
>Dec 26 03:19:52 miller-genuine-draft xntpd[75]: time reset (step) 0.452551 s
>Dec 26 03:26:16 miller-genuine-draft xntpd[75]: time reset (step) 0.373672 s
>...

>These step adjustments are extremley annoying to programs that run
>and clock things in the 10ms range.  The clock jumps forward and
>backward like a jumping bean.  If I discontinue running xntpd my
>time adjustment problems go away, but then my clock doesn't keep
>correct time.

>Any ideas or fixes?  Any good starting places to start hacking away to fix
>this?

I've had the same problem for quite some time..  On all the
FreeBSD machines I have access to, the xntpd oscilates very very badly
(like you've shown) and eventually logs "not logging any more time
steps" or something like that.

The really depressing part, is the SVR4 machines (you know, the ones
with the horrible 10ms clock resolution) sitting right next to the
FreeBSD machines with their super-high-res clocks are locking right in
and staying very stable (xntpd getting to 1024 poll and very low
dispersion), while the FreeBSD machines are constantly wobbling all
over the place. :-(

I've tried lots of silly things like chopping the xntpd precision down
to -6 like the SVR4 boxes, but it still doesn't help. :-(  On the SVR4
version, the xntpd PLL/VCO/whatever_its_called locks in, and on
FreeBSD it does not.  I do not understand the xntpd mechanisms, so I
have no idea what to try and change.

(just about to try making tickadj more coarse (5us -> 40us) in case
the clock drift is too fast for adjtime() to keep up.  The SVR4
machines are running at 40us. I have no idea what to expect as a result.)

-Peter

>Thanks for your help,
>	
>	-Jim