NetBSD Problem Report #45571

From nakayosh@leto.eonet.ne.jp  Sat Nov  5 13:39:27 2011
Return-Path: <nakayosh@leto.eonet.ne.jp>
Received: from mail.netbsd.org (mail.netbsd.org [204.152.190.11])
	by www.NetBSD.org (Postfix) with ESMTP id 7484B63B8DB
	for <gnats-bugs@gnats.NetBSD.org>; Sat,  5 Nov 2011 13:39:27 +0000 (UTC)
Message-Id: <20111105133924.2CA2E1B727A@ae0000-mailauth12.eo.k-opti.ad.jp>
Date: Sat,  5 Nov 2011 22:39:24 +0900 (JST)
From: nakayosh@leto.eonet.ne.jp
Reply-To: nakayosh@leto.eonet.ne.jp
To: gnats-bugs@gnats.NetBSD.org
Subject: lfs_vnops.c: redundant unlock and lock
X-Send-Pr-Version: 3.95

>Number:         45571
>Category:       kern
>Synopsis:       lfs_vnops.c: redundant unlock and lock
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Nov 05 13:40:04 +0000 2011
>Last-Modified:  Sat Nov 05 18:45:01 +0000 2011
>Originator:     NAKAJIMA Yoshihiro
>Release:        NetBSD-current/20111104
>Organization:
>Environment:
System: NetBSD asura 5.99.56 NetBSD 5.99.56 (ASURA) #5: Fri Nov 4 21:22:11 JST 2011 nakayosh@asura:/usr/src/sys/arch/amd64/compile/ASURA amd64
Architecture: x86_64
Machine: amd64
>Description:
	There are redundant unlock and lock in lfs_vnops.c.
	I think they should be removed in 1.204 or 1.208.
>How-To-Repeat:
>Fix:
Date: Sat Nov  5 22:33:48 JST 2011
diff -u src/sys/ufs/lfs/lfs_vnops.c.ORIG src/sys/ufs/lfs/lfs_vnops.c
--- src/sys/ufs/lfs/lfs_vnops.c.ORIG	2011-09-20 23:01:33.000000000 +0900
+++ src/sys/ufs/lfs/lfs_vnops.c	2011-11-05 22:30:42.000000000 +0900
@@ -1,4 +1,4 @@
-/*	$NetBSD: lfs_vnops.c,v 1.238 2011/09/20 14:01:33 chs Exp $	*/
+/*	$NetBSD$	*/

 /*-
  * Copyright (c) 1999, 2000, 2001, 2002, 2003 The NetBSD Foundation, Inc.
@@ -2315,9 +2315,6 @@
 		busypg = NULL;
 		if (check_dirty(fs, vp, startoffset, endoffset, blkeof,
 				ap->a_flags, 0, &busypg) < 0) {
-			mutex_exit(vp->v_interlock);
-
-			mutex_enter(vp->v_interlock);
 			write_and_wait(fs, vp, busypg, seglocked, NULL);
 			if (!seglocked) {
 				mutex_exit(vp->v_interlock);

>Audit-Trail:
From: David Holland <dholland-bugs@netbsd.org>
To: gnats-bugs@NetBSD.org
Cc: 
Subject: Re: kern/45571: lfs_vnops.c: redundant unlock and lock
Date: Sat, 5 Nov 2011 18:40:05 +0000

 On Sat, Nov 05, 2011 at 01:40:04PM +0000, nakayosh@leto.eonet.ne.jp wrote:
  > @@ -2315,9 +2315,6 @@
  >  		busypg = NULL;
  >  		if (check_dirty(fs, vp, startoffset, endoffset, blkeof,
  >  				ap->a_flags, 0, &busypg) < 0) {
  > -			mutex_exit(vp->v_interlock);
  > -
  > -			mutex_enter(vp->v_interlock);
  >  			write_and_wait(fs, vp, busypg, seglocked, NULL);
  >  			if (!seglocked) {
  >  				mutex_exit(vp->v_interlock);
  > 

 Almost certainly... but there's some (probably small) chance that's a
 deliberate hack to avoid some deadlock case, so I'd better look into
 it more closely.

 -- 
 David A. Holland
 dholland@netbsd.org

Home	PR Database Search