From info@maroufi.net  Tue May  6 12:28:26 2008
Return-Path: <info@maroufi.net>
Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34])
	by hub.freebsd.org (Postfix) with ESMTP id A0D71106564A
	for <FreeBSD-gnats-submit@freebsd.org>; Tue,  6 May 2008 12:28:26 +0000 (UTC)
	(envelope-from info@maroufi.net)
Received: from srv1.maruwebsrv1.de (maruweb.de [217.11.53.11])
	by mx1.freebsd.org (Postfix) with ESMTP id 6569F8FC19
	for <FreeBSD-gnats-submit@freebsd.org>; Tue,  6 May 2008 12:28:25 +0000 (UTC)
	(envelope-from info@maroufi.net)
Received: from merlin.maroufi (p5B2A39F9.dip0.t-ipconnect.de [91.42.57.249])
	by srv1.maruwebsrv1.de (Postfix) with ESMTP id AF4BB680009B
	for <FreeBSD-gnats-submit@freebsd.org>; Tue,  6 May 2008 14:06:48 +0200 (CEST)
Received: from dagda.maroufi (dagda.maroufi [192.168.0.50])
	by merlin.maroufi (Postfix) with ESMTP id 811DF2AB
	for <FreeBSD-gnats-submit@freebsd.org>; Tue,  6 May 2008 14:06:47 +0200 (CEST)
Received: from dagda.maroufi (rene@localhost.maroufi [127.0.0.1])
	by dagda.maroufi (8.14.1/8.14.1) with ESMTP id m46C6l8E000796
	for <FreeBSD-gnats-submit@freebsd.org>; Tue, 6 May 2008 14:06:47 +0200 (CEST)
Received: (from rene@localhost)
	by dagda.maroufi (8.14.1/8.14.1/Submit) id m46C6kde002426
	for FreeBSD-gnats-submit@freebsd.org; Tue, 6 May 2008 14:06:46 +0200 (CEST)
Message-Id: <20080506120646.GA25433@dagda.maroufi>
Date: Tue, 6 May 2008 14:06:46 +0200
From: Rene Maroufi <info@maroufi.net>
To: FreeBSD-gnats-submit@freebsd.org
Subject: clock is to fast

>Number:         123462
>Category:       i386
>Synopsis:       clock is too fast
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-i386
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Tue May 06 12:30:03 UTC 2008
>Closed-Date:    Mon Oct 24 15:33:34 UTC 2011
>Last-Modified:  Mon Oct 24 15:33:34 UTC 2011
>Originator:     Bourne-again Superuser
>Release:        FreeBSD 7.0-RELEASE i386
>Organization:
>Environment:
System: FreeBSD silvana.maroufi 7.0-RELEASE FreeBSD 7.0-RELEASE #0: Sun Feb 24 19:59:52 UTC 2008 root@logan.cse.buffalo.edu:/usr/obj/usr/src/sys/GENERIC i386


	Pentium III 850 MHz, Intel i815 Chipset
	
>Description:
	I installed FreeBSD 7.0 Release on an old Pentium III with 
	i815 Chipset. I noticed that date shows the wrong time. I
	correct this with ntpdate. Short time later time is wrong
	again, clock is far in the future. I booted the system with 
	acpi off. Clock is too fast too, but not so extreme like with 
	acpi on. This Problem happens only with this Hardware. Maybe 
	its a problem with the i815 Chipset. OpenBSD has on the same 
	Hardware, the same Problem. See:
	http://cvs.openbsd.org/cgi-bin/query-pr-wrapper?full=yes&numbers=5796

>How-To-Repeat:
	Install FreeBSD 7.0 on a Motherboard with i815 Chipset. Watch the clock.
	
>Fix:

	unknown
	
>Release-Note:
>Audit-Trail:

From: Bruce Cran <bruce@cran.org.uk>
To: bug-followup@FreeBSD.org, info@maroufi.net
Cc:  
Subject: Re: i386/123462: clock is to fast
Date: Thu, 08 May 2008 09:18:08 +0100

 Could you post the output from the kern.timecounter sysctl please?
 That will let us see which timecounter has been chosen along with other 
 information which will be useful to diagnose the problem.
 
 -- 
 Bruce

From: Rene Maroufi <info@maroufi.net>
To: Bruce Cran <bruce@cran.org.uk>
Cc: bug-followup@FreeBSD.org
Subject: Re: i386/123462: clock is to fast
Date: Thu, 8 May 2008 10:32:37 +0200

 On Thu, May 08, 2008 at 09:18:08AM +0100, Bruce Cran wrote:
 > Could you post the output from the kern.timecounter sysctl please?
 > That will let us see which timecounter has been chosen along with other 
 > information which will be useful to diagnose the problem.
 
 sysctl kern.timecounter
 kern.timecounter.tick: 1
 kern.timecounter.choice: TSC(800) ACPI-safe(850) i8254(0)
 dummy(-1000000)
 kern.timecounter.hardware: ACPI-safe
 kern.timecounter.nsetclock: 2
 kern.timecounter.ngetmicrotime: 30316
 kern.timecounter.ngetnanotime: 0
 kern.timecounter.ngetbintime: 0
 kern.timecounter.ngetmicrouptime: 7634
 kern.timecounter.ngetnanouptime: 45
 kern.timecounter.ngetbinuptime: 523
 kern.timecounter.nmicrotime: 632
 kern.timecounter.nnanotime: 24
 kern.timecounter.nbintime: 656
 kern.timecounter.nmicrouptime: 875
 kern.timecounter.nnanouptime: 6
 kern.timecounter.nbinuptime: 11193
 kern.timecounter.stepwarnings: 0
 kern.timecounter.tc.i8254.mask: 4294967295
 kern.timecounter.tc.i8254.counter: 2344
 kern.timecounter.tc.i8254.frequency: 1193182
 kern.timecounter.tc.i8254.quality: 0
 kern.timecounter.tc.ACPI-safe.mask: 16777215
 kern.timecounter.tc.ACPI-safe.counter: 16350933
 kern.timecounter.tc.ACPI-safe.frequency: 3579545
 kern.timecounter.tc.ACPI-safe.quality: 850
 kern.timecounter.tc.TSC.mask: 4294967295
 kern.timecounter.tc.TSC.counter: 1559205353
 kern.timecounter.tc.TSC.frequency: 846194682
 kern.timecounter.tc.TSC.quality: 800
 kern.timecounter.smp_tsc: 0
 

From: Bruce Cran <bruce@cran.org.uk>
To: Rene Maroufi <info@maroufi.net>
Cc: bug-followup@FreeBSD.org
Subject: Re: i386/123462: clock is to fast
Date: Thu, 08 May 2008 10:20:33 +0100

 The ACPI-safe timecounter doesn't work on some systems, and you need to 
 tell FreeBSD to use one of the other time sources. Could you try 
 enabling the i8254 by putting the following in /etc/sysctl.conf and 
 rebooting:
 
 kern.timecounter.hardware=i8254
 
 If that doesn't work you might want to try using the TSC instead.
 
 -- 
 Bruce
 
State-Changed-From-To: open->feedback 
State-Changed-By: linimon 
State-Changed-When: Thu May 8 19:40:39 UTC 2008 
State-Changed-Why:  
Note that submitter has been asked for feedback. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=123462 

From: Rene Maroufi <info@maroufi.net>
To: Bruce Cran <bruce@cran.org.uk>
Cc: bug-followup@FreeBSD.org
Subject: Re: i386/123462: clock is to fast
Date: Fri, 9 May 2008 10:48:21 +0200

 On Thu, May 08, 2008 at 10:20:33AM +0100, Bruce Cran wrote:
 > The ACPI-safe timecounter doesn't work on some systems, and you need to 
 > tell FreeBSD to use one of the other time sources. Could you try 
 > enabling the i8254 by putting the following in /etc/sysctl.conf and 
 > rebooting:
 > 
 > kern.timecounter.hardware=i8254
 > 
 > If that doesn't work you might want to try using the TSC instead.
 
 I tested it with kern.timecounter=i8254 and with kern.timecounter=TSC,
 but both are to fast to. It's not extreme like with acpi, but both to
 fast to. Both are 1 or 2 Seconds to fast per Minute:
 ntpdate ntp
  9 May 10:26:57 ntpdate[776]: step time server 192.168.0.20 offset
  -0.690100 sec
 
 ntpdate ntp
  9 May 10:36:08 ntpdate[801]: step time server 192.168.0.20 offset
  -9.297237 sec
 
 sysctl kern.timecounter.hardware
 kern.timecounter.hardware: i8254
 
 -- 
 Ren Maroufi
 info@maroufi.net

From: Bruce Cran <bruce@cran.org.uk>
To: bug-followup@FreeBSD.org, info@maroufi.net
Cc:  
Subject: Re: i386/123462: clock is to fast
Date: Fri, 09 May 2008 22:26:46 +0100

 ntp is designed to compensate for that sort of drift, but it should be 
 run continuously: the best setup is to run ntpdate once on boot to make 
 any large adjustments but also to run ntpd so it can track and correct 
 any drift.
 
 There's a guide to configuring ntpd in the handbook at 
 http://www.freebsd.org/doc/en/books/handbook/network-ntp.html
 
 -- 
 Bruce Cran

From: Rene Maroufi <info@maroufi.net>
To: Bruce Cran <bruce@cran.org.uk>
Cc: bug-followup@FreeBSD.org
Subject: Re: i386/123462: clock is to fast
Date: Fri, 9 May 2008 23:56:54 +0200

 On Fri, May 09, 2008 at 10:26:46PM +0100, Bruce Cran wrote:
 > ntp is designed to compensate for that sort of drift, but it should be 
 > run continuously: the best setup is to run ntpdate once on boot to make 
 > any large adjustments but also to run ntpd so it can track and correct 
 > any drift.
 
 No, I tested this with ntpd too, but ntpd can't correct the time. The
 time difference between the real time and the PC clock grows with ntpd
 too.
 
 Greetings
 Rene
 -- 
 Ren Maroufi
 info@maroufi.net
State-Changed-From-To: feedback->open 
State-Changed-By: vwe 
State-Changed-When: Sat May 10 18:54:29 UTC 2008 
State-Changed-Why:  

Feedback has been provided. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=123462 

From: Rene Maroufi <info@maroufi.net>
To: Rui Paulo <rpaulo@FreeBSD.org>
Cc: freebsd-i386@FreeBSD.org, bug-followup@FreeBSD.org
Subject: Re: i386/123462: clock is to fast
Date: Sat, 10 May 2008 20:42:50 +0200

 On Sat, May 10, 2008 at 05:15:41PM +0100, Rui Paulo wrote:
 > Rene Maroufi wrote:
 > >On Sat, May 10, 2008 at 11:50:19AM +0100, Rui Paulo wrote:
 > >>Can you boot with "kern.hz=100" and check whether or not it makes a 
 > >>difference?
 > >
 > >There is no sysctl called kern.hz:
 > >sysctl kern.hz
 > >sysctl: unknown oid 'kern.hz'
 > >
 > >Greetings
 > >Ren Maroufi
 > 
 > This is not a sysctl, it's a tunable. Escape to the boot loader prompt 
 > and type:
 > > set kern.hz=100
 > > boot
 
 It makes no difference. With kern.timecounter.hardware: i8254 and acpi
 enabled and no running ntpd:
 
 10 May 19:14:16 ntpdate[805]: adjust time server 192.168.0.20 offset
 -0.243301 sec
 
 10 May 20:35:02 ntpdate[968]: step time server 192.168.0.20 offset
 -77.874950 sec
 
 Thats 77 seconds in 81 minutes difference.
 
 Regards
 Ren
 -- 
 Ren Maroufi
 info@maroufi.net

From: Bruce Cran <bruce@cran.org.uk>
To: bug-followup@FreeBSD.org, info@maroufi.net
Cc:  
Subject: Re: i386/123462: clock is too fast
Date: Fri, 16 May 2008 11:52:07 +0100

 It seems strange that even other time sources within the computer are 
 drifting such a lot.  The only other idea I have is whether the bus 
 signals are running too quickly - is the computer overclocked at all?
 
 -- 
 Bruce Cran

From: Alexander <freebsd@nagilum.org>
To: bug-followup@freebsd.org
Cc: info@maroufi.net
Subject: Re: i386/123462: clock is too fast
Date: Sat, 24 May 2008 19:34:10 +0200

 Have you ever deleted the drift file between those tests?
 If you try a new timer mode you should also delete the drift file (as  
 the drift will be different..) and then restart the ntpd.
 Or as a short receipe:
 - stop ntpd
 - remove the drift file
 - set the new timer mode
 - run ntpdate
 - run ntpd
 
 
 
 
 ----------------------------------------------------------------
 cakebox.homeunix.net - all the machine one needs..

From: Rene Maroufi <info@maroufi.net>
To: Alexander <freebsd@nagilum.org>
Cc: bug-followup@freebsd.org
Subject: Re: i386/123462: clock is too fast
Date: Sun, 25 May 2008 21:28:33 +0200

 On Sat, May 24, 2008 at 07:34:10PM +0200, Alexander wrote:
 > Have you ever deleted the drift file between those tests?
 
 No, but I fix the problem in another way. It's old Hardware, I donate it
 to a school. I waste to much time for this problem. Sorry for not
 debugging anymore.
 
 Regards
 Ren
 -- 
 Ren Maroufi
 info@maroufi.net
State-Changed-From-To: open->closed 
State-Changed-By: remko 
State-Changed-When: Mon Oct 24 15:33:32 UTC 2011 
State-Changed-Why:  
The hardware is no longer there, the submitter can no longer provide 
feedback, so unless someone experiences this as well, we cannot properly 
resolve this. 

http://www.freebsd.org/cgi/query-pr.cgi?pr=123462 
>Unformatted:
