NetBSD Problem Report #40375

From www@NetBSD.org  Mon Jan 12 14:40:22 2009
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by narn.NetBSD.org (Postfix) with ESMTP id 0B21763B909
	for <gnats-bugs@gnats.netbsd.org>; Mon, 12 Jan 2009 14:40:22 +0000 (UTC)
Message-Id: <20090112144021.BF3FA63B8C9@narn.NetBSD.org>
Date: Mon, 12 Jan 2009 14:40:21 +0000 (UTC)
From: mm_lists@pulsar-zone.net
Reply-To: mm_lists@pulsar-zone.net
To: gnats-bugs@NetBSD.org
Subject: netbsd-5: Process locks in vnode wchan when writing to LFS
X-Send-Pr-Version: www-1.0

>Number:         40375
>Category:       kern
>Synopsis:       netbsd-5: Process locks in vnode wchan when writing to LFS
>Confidential:   no
>Severity:       serious
>Priority:       low
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Jan 12 14:45:00 +0000 2009
>Last-Modified:  Sat Jan 17 08:45:01 +0000 2009
>Originator:     Matthew Mondor
>Release:        netbsd-5
>Organization:
>Environment:
NetBSD hal.xisop 5.0_BETA NetBSD 5.0_BETA (GENERIC_MM) #1: Thu Jan  1 22:39:44 EST 2009  root@hal.xisop:/usr/obj/sys/arch/i386/compile/GENERIC_MM i386
>Description:
Since I was told that LFS was supposed to be more stable nowadays
than it was when I used it in the 1.5 days, I thought I'd give it
a try.

So I created a large fresh LFS file system (16GB), mounted it and
attempted a local mirror (on LAN) pkgsrc CVS checkout.  Less than
halfway the cvs process was permanently locked into vnode wchan.
>How-To-Repeat:
See above.
>Fix:
Unknown, but it looks like a race condition.

>Audit-Trail:
From: David Holland <dholland-bugs@netbsd.org>
To: mm_lists@pulsar-zone.net
Cc: kern-bug-people@netbsd.org, gnats-admin@netbsd.org,
	netbsd-bugs@netbsd.org, gnats-bugs@netbsd.org
Subject: Re: kern/40375: netbsd-5: Process locks in vnode wchan when
	writing to LFS
Date: Mon, 12 Jan 2009 16:36:30 +0000

 On Mon, Jan 12, 2009 at 02:45:00PM +0000, mm_lists@pulsar-zone.net wrote:
  > Since I was told that LFS was supposed to be more stable nowadays
  > than it was when I used it in the 1.5 days, I thought I'd give it
  > a try.

 Unfortunately, it's pretty much broke again in 5.0, due to the
 introduction of real fine-grained locking.

 For now, try 4.0_STABLE; there's also some talk of making an
 lfs-renovation branch so it'll be usable again in 6.0.

 -- 
 David A. Holland
 dholland@netbsd.org

From: "Matthew Mondor" <mm_lists@pulsar-zone.net>
To: David Holland <dholland-bugs@netbsd.org>
Cc: kern-bug-people@netbsd.org, netbsd-bugs@netbsd.org, gnats-bugs@netbsd.org
Subject: Re: kern/40375: netbsd-5: Process locks in vnode wchan when
Date: Mon, 12 Jan 2009 18:57:04 -0500 (EST)

 On Monday, 12 Jan 2009 11:36:32
 David Holland <dholland-bugs@netbsd.org> wrote:

 > Unfortunately, it's pretty much broke again in 5.0, due to
 > theintroduction of real fine-grained locking.

 > For now, try 4.0_STABLE; there's also some talk of making
 > anlfs-renovation branch so it'll be usable again in 6.0.

 Thanks for confirming this.

 The test was only to verify if it worked on netbsd-5, and to file a
 PR if it didn't, since no existing PR was obvious about it
 -- 
 Matthew Mondor

From: Andrew Doran <ad@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: kern-bug-people@netbsd.org, gnats-admin@netbsd.org,
	netbsd-bugs@netbsd.org, mm_lists@pulsar-zone.net,
	tech-kern@netbsd.org
Subject: Re: kern/40375: netbsd-5: Process locks in vnode wchan when writing to LFS
Date: Fri, 16 Jan 2009 09:23:37 +0000

 On Mon, Jan 12, 2009 at 04:40:07PM +0000, David Holland wrote:

 >  For now, try 4.0_STABLE; there's also some talk of making an
 >  lfs-renovation branch so it'll be usable again in 6.0.

 I will vehemently oppose any such work if the "resurrection" is to be
 anything like revivesa, where the patient is hooked back up to life support
 and left to remain in a persistent vegetative state.

 Our LFS implementation is utterly rotten and it would serve the _project_
 and its _users_ better if we invested time in another solution like ZFS or
 hammer. In fact, we are very close to a working port of ZFS.

 I do think that a complete rewrite of LFS would be a very interesting
 project and would support that, but it would be something done out of
 personal interest.

 Andrew,

From: David Holland <dholland-bugs@netbsd.org>
To: Andrew Doran <ad@netbsd.org>
Cc: gnats-bugs@NetBSD.org, kern-bug-people@netbsd.org,
	gnats-admin@netbsd.org, netbsd-bugs@netbsd.org,
	mm_lists@pulsar-zone.net, tech-kern@netbsd.org
Subject: Re: kern/40375: netbsd-5: Process locks in vnode wchan when
	writing to LFS
Date: Sat, 17 Jan 2009 08:44:25 +0000

 On Fri, Jan 16, 2009 at 09:23:37AM +0000, Andrew Doran wrote:
  >>  For now, try 4.0_STABLE; there's also some talk of making an
  >>  lfs-renovation branch so it'll be usable again in 6.0.
  > 
  > I will vehemently oppose any such work if the "resurrection" is to be
  > anything like revivesa, where the patient is hooked back up to life support
  > and left to remain in a persistent vegetative state.

 I think I know how to fix the locking problems for good, although I
 need to dig around some to be sure it'll really work. I'm not
 convinced any of the other problems are really all that serious by
 comparison.

 -- 
 David A. Holland
 dholland@netbsd.org

NetBSD Home
NetBSD PR Database Search

(Contact us) $NetBSD: query-full-pr,v 1.39 2013/11/01 18:47:49 spz Exp $
$NetBSD: gnats_config.sh,v 1.8 2006/05/07 09:23:38 tsutsui Exp $
Copyright © 1994-2007 The NetBSD Foundation, Inc. ALL RIGHTS RESERVED.