NetBSD Problem Report #15600
Received: (qmail 3710 invoked from network); 13 Feb 2002 22:03:50 -0000
Message-Id: <200202132203.g1DM3lj01674@night-porter.duskware.de>
Date: Wed, 13 Feb 2002 23:03:47 +0100 (MET)
From: Martin Husemann <martin@duskware.de>
Reply-To: martin@duskware.de
To: gnats-bugs@gnats.netbsd.org
Subject: mfs on swap, when swap is over NFS, panics kernel
X-Send-Pr-Version: 3.95
>Number: 15600
>Category: port-sparc
>Synopsis: mfs on swap, when swap is over NFS, panics kernel
>Confidential: no
>Severity: critical
>Priority: medium
>Responsible: port-sparc-maintainer
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Wed Feb 13 22:04:00 +0000 2002
>Closed-Date: Wed Mar 31 10:26:12 +0000 2004
>Last-Modified: Wed Mar 31 10:26:12 +0000 2004
>Originator: Martin Husemann
>Release: NetBSD 1.5ZA
>Organization:
>Environment:
System: NetBSD night-porter.duskware.de 1.5ZA NetBSD 1.5ZA (PORTER) #0: Tue Feb 12 21:59:35 CET 2002 martin@insomnia.duskware.de:/usr/src/sys/arch/i386/compile/PORTER i386
Architecture: i386
Machine: i386
>Description:
I have a diskless sparc (JavaStation Krups) and tried to speed up some things
via mounting /tmp on mfs. I used this /etc/fstab:
emmas:/exp3/sun / nfs rw 0 0
/swap none swap sw
swap /tmp mfs rw,-s=20m 0 0
I made sure swap is already configured and then said
mount /tmp
The kernel imediately caugt a trap (data access violation, have to provide
details and backtrace when I have access to that machine again).
So I checked if this is machine dependend and tried the same on an i386
machine, with the following fstab:
night-porter:/usr/exp/hosts/star / nfs rw 0 0
/swap none swap sw
swap /tmp mfs rw,-s=2m 0 0
Everything worked:
[root@evening-star] /etc # mount
night-porter.duskware.de:/usr/exp/hosts/star on / type nfs
mfs:251 on /tmp type mfs (local)
[root@evening-star] /etc # swapctl -l
Device 1K-blocks Used Avail Capacity Priority
/swap 10240 0 10240 0% 0
The sparc runs a kernel from sources cvs updated Feb 13, 2002, the i386 one
day older sources.
>How-To-Repeat:
s.a.
>Fix:
n/a
>Release-Note:
>Audit-Trail:
From: martin@netbsd.org
To: gnats-bugs@NetBSD.ORG
Cc:
Subject: Re: port-sparc/15600
Date: 14 Feb 2002 09:24:47 -0000
Here is the crash info (10 finger copy):
data fault: pc=0xe8052f6c addr=0x80 sfsr=326<PERR=0,LVL=3,AT=1,FT=1,FAV,OW>
panic: kernel fault
Stopped in pid 171 (mount_mfs) at cpu_Debugger+0x4: jmpl [%o7+0x8],%g0
db> tr
mem_access_fault4m(0xf60724b0, 0x326, 0x80, 0xf61efc88, 0xf094b600, 0x28) at mem_access_fault4m+0x304
memfault_sun4m(0x80, 0x0, 0xf607d000, 0x0, 0x21, 0x3) at memfault_sun4m+0xe4
lookup(0x0, 0x0, 0x0, 0xf61efe88, 0xf61ee00, 0xf6089400) at lookup+0x6a0
namei(0x0, 0x4900084, 0x16b28, 0xe8031970, 0x0, 0x47e6c) at namei+0x15c
sys_statfs(0xf6088b18m 0xf61eff28, 0xf61eff20, 0xe8057890, 0x3, 0xe7fff3ff8) at sys_statfs+0x2c
syscall(...)
...
Martin
From: matthew green <mrg@eterna.com.au>
To: martin@duskware.de
Cc: gnats-bugs@gnats.netbsd.org
Subject: re: port-sparc/15600: mfs on swap, when swap is over NFS, panics kernel
Date: Fri, 15 Feb 2002 17:28:12 +1100
FWIW, i have been running a diskless netbsd/sparc with mfs /tmp for
*years*. the remaining machine (a classic) is running a 13 month
old -current...so perhaps it's new breakage.
From: "Valeriy E. Ushakov" <uwe@ptc.spbu.ru>
To: gnats-bugs@netbsd.org
Cc: Martin Husemann <martin@duskware.de>, matthew green <mrg@eterna.com.au>
Subject: Re: port-sparc/15600
Date: Mon, 8 Apr 2002 20:51:49 +0400
This seem to be a generic problem, not port-sparc specific.
If I enable MFS in KRUPS config, it fails just as descibed in the PR.
But when I enable FFS as well - it works just fine.
What's interesting is that the two kernels only differ in few bytes in
data segment:
text data bss dec hex filename
1301456 85584 139388 1526428 174a9c netbsd options MFS only
1301456 85592 139388 1526436 174aa4 netbsd options MFS and FFS
So my educated guess is that some #if defined(FFS) shall really be
#if defined(FFS) || defined(MFS).
-uwe
From: "Valeriy E. Ushakov" <uwe@ptc.spbu.ru>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: port-sparc/15600
Date: Mon, 8 Apr 2002 21:31:48 +0400
So the two kernels only differ in that the MFS+FFS one has ffs_vfsops
in its vfs_list_initial[] in ioconf.c. Manually changing the order of
mfs_vfsops and ffs_vfsops in vfs_list_initial[] so that mfs_vfsops
comes first still produces a working kernel.
-uwe
State-Changed-From-To: open->closed
State-Changed-By: mrg
State-Changed-When: Wed Mar 31 10:25:44 UTC 2004
State-Changed-Why:
this is a duplicate of 6061 which also has a complete analysis of
the problem...
>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.