NetBSD Problem Report #10361

Received: (qmail 442 invoked from network); 14 Jun 2000 06:19:27 -0000
Message-Id: <200006140619.XAA13395@quatre.pa.dec.com>
Date: Tue, 13 Jun 2000 23:19:26 -0700
From: murray@pa.dec.com (Hal Murray)
Sender: murray@pa.dec.com
Reply-To: hmurray@megapathdsl.net
To: gnats-bugs@gnats.netbsd.org
Subject: "last message repeated" logic mangles /var/log/messages
X-Send-Pr-Version: 3.95

>Number:         10361
>Category:       kern
>Synopsis:       "last message repeated" logic mangles /var/log/messages
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    mycroft
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Jun 14 06:20:00 +0000 2000
>Closed-Date:    Fri Apr 04 22:04:54 +0000 2003
>Last-Modified:  Fri Apr 04 22:04:54 +0000 2003
>Originator:     Hal Murray
>Release:        1.4Z
>Organization:
        Systems Research Center, Compaq Computer Corporation
>Environment:
Alpha and rarely on i386
System: NetBSD mckinley 1.4Z NetBSD 1.4Z (MIATA) #4: Fri Jun 9 04:01:32 PDT 2000
 murray@mckinley:/usr/src/sys/arch/alpha/compile/MIATA alpha


>Description:

Here is a sample:

Jun 13 23:09:55 mckinley /netbsd: tlp0: receive ring overrun
Jun 13 23:10:14 mckinley ntpd[177]: synchronisation lost
Jun 13 23:09:55 mckinley last message repeated 301 times
Jun 13 23:10:26 mckinley /netbsd: rrun                          <==
Jun 13 23:10:26 mckinley /netbsd: tlp0: receive ring overrun
Jun 13 23:10:26 mckinley last message repeated 301 times
Jun 13 23:11:08 mckinley /netbsd: tlp0: receive error: CRC error
Jun 13 23:11:29 mckinley /netbsd: rrun                          <==
Jun 13 23:11:29 mckinley /netbsd: tlp0: receive ring overrun
Jun 13 23:11:29 mckinley last message repeated 301 times


>How-To-Repeat:

        I get them with nasty network tests. 


>Fix:
        ??

>Release-Note:
>Audit-Trail:

From: cgd@sibyte.com (Chris G. Demetriou)
To: murray@pa.dec.com
Cc: gnats-bugs@netbsd.org, netbsd-bugs@netbsd.org
Subject: Re: kern/10361: "last message repeated" logic mangles /var/log/messages
Date: 14 Jun 2000 11:05:28 -0700

 murray@pa.dec.com (Hal Murray) writes:
 > Jun 13 23:09:55 mckinley /netbsd: tlp0: receive ring overrun
 > Jun 13 23:10:14 mckinley ntpd[177]: synchronisation lost
 > Jun 13 23:09:55 mckinley last message repeated 301 times
 > Jun 13 23:10:26 mckinley /netbsd: rrun                          <==
 > Jun 13 23:10:26 mckinley /netbsd: tlp0: receive ring overrun
 > Jun 13 23:10:26 mckinley last message repeated 301 times
 > Jun 13 23:11:08 mckinley /netbsd: tlp0: receive error: CRC error
 > Jun 13 23:11:29 mckinley /netbsd: rrun                          <==
 > Jun 13 23:11:29 mckinley /netbsd: tlp0: receive ring overrun
 > Jun 13 23:11:29 mckinley last message repeated 301 times

 Uh, it sounds more to me like your message buffer is getting
 truncated, and the first thing in it (i.e. that hasn't been pushed off
 the beginning) is 'rrun'.

 note the consistency; i wouldn't expect that from a random bug.

 how big is your message buffer, and what does 'dmesg' say after one of
 these tests is over?



 cgd
State-Changed-From-To: open->feedback 
State-Changed-By: bjh21 
State-Changed-When: Sat Dec 23 09:46:27 PST 2000 
State-Changed-Why:  
cgd's message is clearly a request for feedback. 

From: Hal Murray <murray@pa.dec.com>
To: gnats-bugs@netbsd.org
Cc: murray@pa.dec.com
Subject: Re: kern/10361: "last message repeated" logic mangles /var/log/messages
Date: Wed, 31 Jan 2001 01:51:23 -0800

 I'm running similar tests on 1.5 systems with a 65K message buffer.
 I haven't seen any troubles.

 I think my initial problem report should probably be changed from 
 kern to doc, but I'm not sure where the documentation should go.

 I had to search the source code to find the parameter in order to 
 make the buffer bigger. 

 options(4) describes many/most of the kernel parameters.  Perhaps 
 MSGBUFSIZE should be added to it. 


 Perhaps another gnat should be opened for all the printfs that can 
 happen too-fast in drivers when they get overloaded.

 I think that all the drivers that can print too much under heavy 
 load should be fixed so that printout gets moved to background and 
 only prints one message (with the number of packets dropped) every 
 N seconds.

 This would probably be reasonably easy to implement - it just takes 
 a macro to bump a counter and a subroutine to poll the counters. 

 There seems to be the right sort of printout-limiting logic in the 
 code that prints "WARNING: mclpool limit reached; increase NMBCLUSTERS"
 (It doesn't include the number of failed attempts.)
State-Changed-From-To: feedback->open 
State-Changed-By: fair 
State-Changed-When: Thu Mar 22 15:15:06 PST 2001 
State-Changed-Why:  
Feedback has been provided - time to analyze... 
State-Changed-From-To: open->feedback 
State-Changed-By: mycroft 
State-Changed-When: Thu Apr 4 21:12:25 PST 2002 
State-Changed-Why:  
Should be fixed in tulip.c 1.107. 


Responsible-Changed-From-To: kern-bug-people->mycroft 
Responsible-Changed-By: mycroft 
Responsible-Changed-When: Thu Apr 4 21:12:25 PST 2002 
Responsible-Changed-Why:  
State-Changed-From-To: feedback->closed 
State-Changed-By: perry 
State-Changed-When: Fri Apr 4 14:04:13 PST 2003 
State-Changed-Why:  
It appears that the user resolved his problem. 
The suggested MSGBUFSIZE documentation for options(4) was added long ago, 
too. 
>Unformatted:

NetBSD Home
NetBSD PR Database Search

(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.