NetBSD Problem Report #42210

From www@NetBSD.org  Wed Oct 21 19:27:25 2009
Return-Path: <www@NetBSD.org>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by www.NetBSD.org (Postfix) with ESMTP id 31DEB63B902
	for <gnats-bugs@gnats.netbsd.org>; Wed, 21 Oct 2009 19:27:25 +0000 (UTC)
Message-Id: <20091021192721.693D163B8B6@www.NetBSD.org>
Date: Wed, 21 Oct 2009 19:27:21 +0000 (UTC)
From: pooka@iki.fi
Reply-To: pooka@iki.fi
To: gnats-bugs@NetBSD.org
Subject: rename EXDEV & abortop fail
X-Send-Pr-Version: www-1.0

>Number:         42210
>Category:       kern
>Synopsis:       rename EXDEV & abortop fail
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          closed
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Oct 21 19:30:00 +0000 2009
>Closed-Date:    Sun Jun 17 20:22:18 +0000 2012
>Last-Modified:  Sun Jun 17 20:22:18 +0000 2012
>Originator:     Antti Kantee
>Release:        
>Organization:
read it and weep inc.
>Environment:
>Description:
In a new and exciting episode of "I thought we fixed the last bug
from rename last week", plenty of drivers will happily fail VOP_RENAME
with EXDEV but neglect to call VOP_ABORTOP.  Notably, since one of the
vnodes might be of completely different type, a given driver
cannot decide to just not call abortop because it "knows" it's
not required.
>How-To-Repeat:

>Fix:
Is there some layering marvel preventing us from checking for
EXDEV before calling rename?  otherwise, have fun

>Release-Note:

>Audit-Trail:
From: David Holland <dholland-bugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/42210: rename EXDEV & abortop fail
Date: Thu, 22 Oct 2009 00:28:51 +0000

 On Wed, Oct 21, 2009 at 07:30:00PM +0000, pooka@iki.fi wrote:
  > In a new and exciting episode of "I thought we fixed the last bug
  > from rename last week", plenty of drivers will happily fail VOP_RENAME
  > with EXDEV but neglect to call VOP_ABORTOP.  Notably, since one of the
  > vnodes might be of completely different type, a given driver
  > cannot decide to just not call abortop because it "knows" it's
  > not required.

 Maybe we can abolish ABORTOP? Only onionfs uses it, but the purpose of
 the onionfs code in question is clear as mud...

  > Is there some layering marvel preventing us from checking for
  > EXDEV before calling rename?  otherwise, have fun

 I have been attempting to figure that one myself, and I don't know of
 any valid reason.

 -- 
 David A. Holland
 dholland@netbsd.org

From: Antti Kantee <pooka@iki.fi>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/42210: rename EXDEV & abortop fail
Date: Thu, 22 Oct 2009 14:39:13 +0300

 On Thu Oct 22 2009 at 00:30:04 +0000, David Holland wrote:
 > Maybe we can abolish ABORTOP? Only onionfs uses it, but the purpose of
 > the onionfs code in question is clear as mud...

 libp2k uses it.  Admittedly, it could do without it with more implicit
 magic, but I think there's enough implicit magic going on as-is.
 Feel free to look at p2k_node_lookup().

 No, I don't mind if you fix the componentname handling ;)

 >  > Is there some layering marvel preventing us from checking for
 >  > EXDEV before calling rename?  otherwise, have fun
 >  
 > I have been attempting to figure that one myself, and I don't know of
 > any valid reason.

 We really really need to come up with a set of tests which define the
 requirements for vfs & file systems.  Currently there are just too many
 butterfly effects ...

From: Antti Kantee <pooka@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/42210 CVS commit: src/sys/fs/puffs
Date: Thu, 19 Nov 2009 16:21:04 +0000

 Module Name:	src
 Committed By:	pooka
 Date:		Thu Nov 19 16:21:04 UTC 2009

 Modified Files:
 	src/sys/fs/puffs: puffs_vnops.c

 Log Message:
 Send VOP_ABORTOP() in case attempting cross-dev rename, part of
 PR kern/42210.  Also, fix a memory management error in said case.


 To generate a diff of this commit:
 cvs rdiff -u -r1.139 -r1.140 src/sys/fs/puffs/puffs_vnops.c

 Please note that diffs are not public domain; they are subject to the
 copyright notices on the relevant files.

From: Manuel Bouyer <bouyer@netbsd.org>
To: gnats-bugs@gnats.NetBSD.org
Cc: 
Subject: PR/42210 CVS commit: [netbsd-5] src/sys/fs/puffs
Date: Sat, 28 Nov 2009 16:00:17 +0000

 Module Name:	src
 Committed By:	bouyer
 Date:		Sat Nov 28 16:00:17 UTC 2009

 Modified Files:
 	src/sys/fs/puffs [netbsd-5]: puffs_vnops.c

 Log Message:
 Pull up following revision(s) (requested by pooka in ticket #1154):
 	sys/fs/puffs/puffs_vnops.c: revision 1.140
 Send VOP_ABORTOP() in case attempting cross-dev rename, part of
 PR kern/42210.  Also, fix a memory management error in said case.


 To generate a diff of this commit:
 cvs rdiff -u -r1.129.4.4 -r1.129.4.5 src/sys/fs/puffs/puffs_vnops.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->closed
State-Changed-By: dholland@NetBSD.org
State-Changed-When: Sun, 17 Jun 2012 20:22:18 +0000
State-Changed-Why:
EXDEV is detected in fs-independent rename code now, so this is
no longer an issue. The fact that libp2k uses ABORTOP is a bug
of sorts but we don't need to keep this PR open just because of
that.


>Unformatted:

Home	PR Database Search