NetBSD Problem Report #37717
From oster@scrooge.localdomain Tue Jan 8 15:53:27 2008
Return-Path: <oster@scrooge.localdomain>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
by narn.NetBSD.org (Postfix) with ESMTP id 82F1C63BB70
for <gnats-bugs@gnats.NetBSD.org>; Tue, 8 Jan 2008 15:53:27 +0000 (UTC)
Message-Id: <20080108145528.58DB5113E32@scrooge.localdomain>
Date: Tue, 8 Jan 2008 08:55:28 -0600 (CST)
From: oster@netbsd.org
Reply-To: oster@netbsd.org
To: gnats-bugs@NetBSD.org
Subject: raidclose() no longer called on shutdown...
X-Send-Pr-Version: 3.95
>Number: 37717
>Category: kern
>Synopsis: raidclose() is no longer called on shutdown
>Confidential: no
>Severity: critical
>Priority: high
>Responsible: ad
>State: closed
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Tue Jan 08 15:55:00 +0000 2008
>Closed-Date: Thu Jan 24 19:33:27 +0000 2008
>Last-Modified: Thu Jan 24 19:33:27 +0000 2008
>Originator: oster@netbsd.org
>Release: NetBSD 4.99.48
>Organization:
>Environment:
System: NetBSD 4.99.48 (RAIDFRAME.ddbLD) #6: Mon Jan 7 17:01:28 CST 2008
oster@quad:/u1/devel/current/src/sys/arch/i386/compile/RAIDFRAME.ddbLD
Architecture: i386
Machine: i386
>Description:
Sometime in the last little while it appears that RAIDframe
RAID sets are always marked 'dirty' after a reboot, even if they were
'clean' before. It has not been determined when this issue started,
exactly.
In a nutshell, rf_netbsdkintf.c:raidclose() is no longer being called
when a filesystem on a RAID set is unmounted -- e.g. unmounting /
(/dev/raid0a) no longer calls raidclose(). With raidclose() not being
called, the "parity is clean" bit never gets set in the component
labels.
>How-To-Repeat:
raidctl -vP raid0
raidctl -s raid0
observe that parity status is "clean".
shutdown -r now
raidctl -s raid0
observe that partiy status is "DIRTY", when it really should be
"clean".
>Fix:
Something is no longer calling raidclose(). Find said
something, and bat it around the ears until it calls raidclose() at
the appropriate times. Please.
>Release-Note:
>Audit-Trail:
From: Juergen Hannken-Illjes <hannken@eis.cs.tu-bs.de>
To: gnats-bugs@netbsd.org
Cc: kern-bug-people@netbsd.org, gnats-admin@netbsd.org,
netbsd-bugs@netbsd.org
Subject: Re: kern/37717: raidclose() no longer called on shutdown...
Date: Tue, 8 Jan 2008 17:26:14 +0100
On Tue, Jan 08, 2008 at 03:55:00PM +0000, oster@netbsd.org wrote:
> >Number: 37717
> >Category: kern
> >Synopsis: raidclose() is no longer called on shutdown
Since vmlocking2 was merged a root file system containing the root device
vnode doesn't get closed because its v_usecount is 2 instead of 1 when
VOP_CLOSE() gets called from ffs_unmount().
Only need to find the bogus reference...
--
Juergen Hannken-Illjes - hannken@eis.cs.tu-bs.de - TU Braunschweig (Germany)
From: Antti Kantee <pooka@cs.hut.fi>
To: gnats-bugs@netbsd.org
Cc:
Subject: Re: kern/37717: raidclose() no longer called on shutdown...
Date: Tue, 8 Jan 2008 18:37:10 +0200
On Tue Jan 08 2008 at 17:26:14 +0100, Juergen Hannken-Illjes wrote:
> On Tue, Jan 08, 2008 at 03:55:00PM +0000, oster@netbsd.org wrote:
> > >Number: 37717
> > >Category: kern
> > >Synopsis: raidclose() is no longer called on shutdown
>
> Since vmlocking2 was merged a root file system containing the root device
> vnode doesn't get closed because its v_usecount is 2 instead of 1 when
> VOP_CLOSE() gets called from ffs_unmount().
>
> Only need to find the bogus reference...
The extra ref needs to be located if it's being leaked ...
... but more importantly, checking refcounts in VOP_CLOSE is wrong.
A vnode's refcount has nothing to do with open/close. I'd move closing
the b/cdev to spec_inactive.
--
Antti Kantee <pooka@iki.fi> Of course he runs NetBSD
http://www.iki.fi/pooka/ http://www.NetBSD.org/
"la qualité la plus indispensable du cuisinier est l'exactitude"
From: Greg Oster <oster@cs.usask.ca>
To: Juergen Hannken-Illjes <hannken@eis.cs.tu-bs.de>
Cc: gnats-bugs@netbsd.org
Subject: Re: kern/37717: raidclose() no longer called on shutdown...
Date: Tue, 08 Jan 2008 19:37:15 -0600
Juergen Hannken-Illjes writes:
> On Tue, Jan 08, 2008 at 03:55:00PM +0000, oster@netbsd.org wrote:
> > >Number: 37717
> > >Category: kern
> > >Synopsis: raidclose() is no longer called on shutdown
>
> Since vmlocking2 was merged a root file system containing the root device
> vnode doesn't get closed because its v_usecount is 2 instead of 1 when
> VOP_CLOSE() gets called from ffs_unmount().
I've been consistently seeing rootvp->v_usecount at 6 before the call to
VOP_CLOSE()... Havn't been able to figure out what's going on yet...
Later...
Greg Oster
Responsible-Changed-From-To: kern-bug-people->ad
Responsible-Changed-By: ad@netbsd.org
Responsible-Changed-When: Wed, 09 Jan 2008 11:43:09 +0000
Responsible-Changed-Why:
My recent changes include vnode reference counting.
From: Andrew Doran <ad@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc:
Subject: PR/37717 CVS commit: src
Date: Thu, 24 Jan 2008 17:32:58 +0000 (UTC)
Module Name: src
Committed By: ad
Date: Thu Jan 24 17:32:58 UTC 2008
Modified Files:
src/sys/dev: cons.c
src/sys/fs/adosfs: advfsops.c
src/sys/fs/cd9660: cd9660_vfsops.c
src/sys/fs/filecorefs: filecore_vfsops.c
src/sys/fs/hfs: hfs_subr.c hfs_vfsops.c
src/sys/fs/msdosfs: msdosfs_vfsops.c
src/sys/fs/ntfs: ntfs_vfsops.c
src/sys/fs/ptyfs: ptyfs_subr.c
src/sys/fs/puffs: puffs_node.c
src/sys/fs/sysvbfs: sysvbfs_vfsops.c
src/sys/fs/tmpfs: tmpfs_subr.c
src/sys/fs/udf: udf_vfsops.c
src/sys/kern: tty_ptm.c vfs_subr.c vfs_subr2.c vfs_syscalls.c
src/sys/miscfs/genfs: layer_subr.c
src/sys/miscfs/specfs: spec_vnops.c specdev.h
src/sys/nfs: nfs_subs.c
src/sys/rump/librump/rumpkern: rump.c vfs.c
src/sys/sys: vnode.h
src/sys/ufs/ext2fs: ext2fs_vfsops.c ext2fs_vnops.c
src/sys/ufs/ffs: ffs_vfsops.c
src/sys/ufs/lfs: lfs_vfsops.c
src/sys/ufs/mfs: mfs_vfsops.c
src/sys/ufs/ufs: ufs_vnops.c
src/usr.bin/fstat: ptyfs.c
src/usr.bin/systat: bufcache.c
src/usr.sbin/pstat: pstat.c
Log Message:
specfs changes for PR kern/37717 (raidclose() is no longer called on
shutdown). There are still problems with device access and a PR will be
filed.
- Kill checkalias(). Allow multiple vnodes to reference a single device.
- Don't play dangerous tricks with block vnodes to ensure that only one
vnode can describe a block device. Instead, prohibit concurrent opens of
block devices. As a bonus remove the unreliable code that prevents
multiple file system mounts on the same device. It's no longer needed.
- Track opens by vnode and by device. Issue cdev_close() when the last open
goes away, instead of abusing vnode::v_usecount to tell if the device is
open.
To generate a diff of this commit:
cvs rdiff -r1.64 -r1.65 src/sys/dev/cons.c
cvs rdiff -r1.46 -r1.47 src/sys/fs/adosfs/advfsops.c
cvs rdiff -r1.53 -r1.54 src/sys/fs/cd9660/cd9660_vfsops.c
cvs rdiff -r1.45 -r1.46 src/sys/fs/filecorefs/filecore_vfsops.c
cvs rdiff -r1.7 -r1.8 src/sys/fs/hfs/hfs_subr.c
cvs rdiff -r1.13 -r1.14 src/sys/fs/hfs/hfs_vfsops.c
cvs rdiff -r1.58 -r1.59 src/sys/fs/msdosfs/msdosfs_vfsops.c
cvs rdiff -r1.60 -r1.61 src/sys/fs/ntfs/ntfs_vfsops.c
cvs rdiff -r1.12 -r1.13 src/sys/fs/ptyfs/ptyfs_subr.c
cvs rdiff -r1.9 -r1.10 src/sys/fs/puffs/puffs_node.c
cvs rdiff -r1.21 -r1.22 src/sys/fs/sysvbfs/sysvbfs_vfsops.c
cvs rdiff -r1.43 -r1.44 src/sys/fs/tmpfs/tmpfs_subr.c
cvs rdiff -r1.33 -r1.34 src/sys/fs/udf/udf_vfsops.c
cvs rdiff -r1.22 -r1.23 src/sys/kern/tty_ptm.c
cvs rdiff -r1.317 -r1.318 src/sys/kern/vfs_subr.c
cvs rdiff -r1.12 -r1.13 src/sys/kern/vfs_subr2.c
cvs rdiff -r1.341 -r1.342 src/sys/kern/vfs_syscalls.c
cvs rdiff -r1.24 -r1.25 src/sys/miscfs/genfs/layer_subr.c
cvs rdiff -r1.111 -r1.112 src/sys/miscfs/specfs/spec_vnops.c
cvs rdiff -r1.33 -r1.34 src/sys/miscfs/specfs/specdev.h
cvs rdiff -r1.196 -r1.197 src/sys/nfs/nfs_subs.c
cvs rdiff -r1.29 -r1.30 src/sys/rump/librump/rumpkern/rump.c \
src/sys/rump/librump/rumpkern/vfs.c
cvs rdiff -r1.185 -r1.186 src/sys/sys/vnode.h
cvs rdiff -r1.127 -r1.128 src/sys/ufs/ext2fs/ext2fs_vfsops.c
cvs rdiff -r1.79 -r1.80 src/sys/ufs/ext2fs/ext2fs_vnops.c
cvs rdiff -r1.218 -r1.219 src/sys/ufs/ffs/ffs_vfsops.c
cvs rdiff -r1.252 -r1.253 src/sys/ufs/lfs/lfs_vfsops.c
cvs rdiff -r1.85 -r1.86 src/sys/ufs/mfs/mfs_vfsops.c
cvs rdiff -r1.162 -r1.163 src/sys/ufs/ufs/ufs_vnops.c
cvs rdiff -r1.3 -r1.4 src/usr.bin/fstat/ptyfs.c
cvs rdiff -r1.20 -r1.21 src/usr.bin/systat/bufcache.c
cvs rdiff -r1.105 -r1.106 src/usr.sbin/pstat/pstat.c
Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.
State-Changed-From-To: open->feedback
State-Changed-By: ad@narn.netbsd.org
State-Changed-When: Thu, 24 Jan 2008 18:35:00 +0000
State-Changed-Why:
Should be fixed, I'll test later.
From: Greg Oster <oster@cs.usask.ca>
To: gnats-bugs@NetBSD.org
Cc:
Subject: Re: kern/37717 (raidclose() is no longer called on shutdown)
Date: Thu, 24 Jan 2008 13:28:12 -0600
ad@NetBSD.org writes:
> Synopsis: raidclose() is no longer called on shutdown
>
> State-Changed-From-To: open->feedback
> State-Changed-By: ad@narn.netbsd.org
> State-Changed-When: Thu, 24 Jan 2008 18:35:00 +0000
> State-Changed-Why:
> Should be fixed, I'll test later.
Looks fixed to me. Thanks!
Later...
Greg Oster
State-Changed-From-To: feedback->closed
State-Changed-By: ad@narn.netbsd.org
State-Changed-When: Thu, 24 Jan 2008 19:33:27 +0000
State-Changed-Why:
Hallelujah, praise the lord, it's 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.