NetBSD Problem Report #15519
Received: (qmail 12587 invoked from network); 7 Feb 2002 10:29:48 -0000
Message-Id: <200202071029.g17ATXo28070@Plectere.com>
Date: Thu, 7 Feb 2002 02:29:33 -0800 (PST)
From: paul@Plectere.com
Reply-To: paul@Plectere.com
To: gnats-bugs@gnats.netbsd.org
Subject: support for chroot'd "ntpd" incomplete
X-Send-Pr-Version: 3.95
>Number: 15519
>Category: kern
>Synopsis: support for chroot'd "ntpd" incomplete
>Confidential: no
>Severity: serious
>Priority: high
>Responsible: manu
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Thu Feb 07 10:30:00 +0000 2002
>Closed-Date: Mon Mar 04 06:04:04 +0000 2002
>Last-Modified: Mon Mar 04 06:04:04 +0000 2002
>Originator: Paul Shupak
>Release: NetBSD 1.5ZA -- current, today
>Organization:
>Environment:
kernel w/ "pseudo-device clockctl" & "options NTP"
System: NetBSD svcs 1.5ZA NetBSD 1.5ZA (PCOM-$Revision: 1.24 $) #185: Wed Feb 6 22:35:17 PST 2002 paul@svcs:/usr/src/sys/arch/i386/compile/PCOM i386
Architecture: i386
Machine: i386
>Description:
The ioctl call made in libc's ntp_adjtime() is actually unimplimented.
>How-To-Repeat:
verify by runing a kernel with "options NTP" and
"pseudo-device clockctl" and watch time `jump' instead
of `skew'. Or, run ntpdc's "kerninfo" command and look at
error messages ( EINVAL returned from ntp_adjtime() ) and/or
/var/log/messages.
>Fix:
--- /sys/dev/clockctl.c Sun Dec 9 22:08:54 2001
+++ clockctl.c Wed Feb 6 23:19:37 2002
@@ -33,6 +33,8 @@
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: clockctl.c,v 1.4 2001/12/09 16:10:43 manu Exp $");
+#include "opt_ntp.h"
+
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
@@ -125,7 +127,7 @@
struct clockctl_ntp_adjtime_args *args =
(struct clockctl_ntp_adjtime_args *)data;
- (void*)ntp_adjtime1(SCARG(args.uas,tp), &error);
+ (void) ntp_adjtime1(SCARG(args,uas.tp), args, &error);
return (error);
}
#endif /* NTP */
>Release-Note:
>Audit-Trail:
Responsible-Changed-From-To: kern-bug-people->manu
Responsible-Changed-By: lukem
Responsible-Changed-When: Sun Feb 24 19:10:01 PST 2002
Responsible-Changed-Why:
manu is responsible for this code
State-Changed-From-To: open->feedback
State-Changed-By: manu
State-Changed-When: Mon Feb 25 13:26:04 PST 2002
State-Changed-Why:
submitted patch was comitted
From: manu@netbsd.org (Emmanuel Dreyfus)
To: paul@Plectere.com (Paul Shupak)
Cc: gnats-bugs@netbsd.org
Subject: Re: kern/15519
Date: Sat, 2 Mar 2002 00:01:11 +0100
> Once you enable the code, immediately the compile dies -- wrong number
> of arguments to the called function. Looks like either a typo or porting
> change ( from some other BSD maybe ) that went uncompleted.
No, the code have been developped on NetBSD, so it is not a porting
issue.
Additionnaly, there is another bug: we do not copyin() the struct ntv
(first arg to ntp_adjtime1()), and it is a userland address. It seems it
just work on i386, but on macppc, it immediatly panics the kernel.
And we don't handle correctly the return value. I've committed a fix to
theses two problems, please let me know if things are still okay for
you.
In fact theses problems reminds me some bug fixes I did. I'm now
convinced that I forgot to commit some work on sys/dev/clockctl.c a few
weeks ago.
--
Emmanuel Dreyfus
manu@netbsd.org
From: Paul Shupak <paul@Plectere.com>
To: manu@netbsd.org
Cc: gnats-bugs@netbsd.org
Subject: Re: kern/15519
Date: Sun, 3 Mar 2002 14:35:20 -0800 (PST)
I think the return/errno issue is somewhat of a red-herring; The
return values for ntp_adjtime other than zero are inconsistent across
different platforms, so most programs only check for zero/non-zero or
zero/less-than-zero ( i.e. what would be "TIME_OK" in <sys/timex.h> ) :
Because the return was always zero AND the kernel clock/PLL code really
was working correctly, the userland code never had a problem with not
receiving notification of errors ( which hadn't actually occurred ).
The only code, in the NetBSD tree, which seems to use the leap-seconds'
return values was in the ref_clock in ntpd.
Anyway... Everything seems to be working just fine (after three
days on five machines) and certainly the original bug report (kern/15519)
should be closed.
Thanks again,
Paul Shupak
>From manu@netbsd.org Sat Mar 2 01:18:53 2002
>To: paul@Plectere.com (Paul Shupak)
>In-Reply-To: <200203020023.g220NdM00504@Plectere.com>
>Subject: Re: kern/15519
>From: manu@netbsd.org (Emmanuel Dreyfus)
>Date: Sat, 2 Mar 2002 10:18:34 +0100
>MIME-Version: 1.0
>Content-Type: text/plain; charset=ISO-8859-1
>X-Mailer: MacSOUP 2.3
>Content-Transfer-Encoding: 8bit
>X-MIME-Autoconverted: from quoted-printable to 8bit by Plectere.com id g229Ir410409
>
>> Sorry, I feel quite stupid for not even considuring the "copyin".
>> Anyway,o far the machines I've tried seem to act identically ( one
>> Athalon/Via-KX, one Celeron/i810 and a large memory PIII/Via-69x ).
>
>They are all i386, this is why. It seems okay to handle directly data
>outside kernel space on this arch...
>What surprises me a lot is that you did not have any problem with the
>return value missing (it was copied to errno, and ntp_adjtime was always
>returning 0)
>
>Anyway, thank you for digging this out. Feel free to send something to
>gnats saying that we can close the PR once you'll have tested it for a
>few days. And thank you also for using this feature.
>
>--
>Emmanuel Dreyfus.
>Vous avez deplacé la souris.
>Windows NT doit maintenant redemarrer pour valider les modifications.
>manu@netbsd.org
>
State-Changed-From-To: feedback->closed
State-Changed-By: manu
State-Changed-When: Sun Mar 3 22:02:36 PST 2002
State-Changed-Why:
The problem is fixed
>Unformatted:
(Contact us)
$NetBSD: query-full-pr,v 1.51 2026/08/10 02:28:17 riastradh Exp $
$NetBSD: gnats_config.sh,v 1.10 2026/05/13 22:00:09 riastradh Exp $
Copyright © 1994-2026
The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.