NetBSD Problem Report #8151

Received: (qmail 21038 invoked from network); 5 Aug 1999 21:27:58 -0000
Message-Id: <199908052127.RAA17020@hrothgar.gw.com>
Date: Thu, 5 Aug 1999 17:27:57 -0400 (EDT)
From: christos@zoulas.com
Reply-To: christos@zoulas.com
To: gnats-bugs@gnats.netbsd.org
Subject: System locks up with nfs loopback mounts
X-Send-Pr-Version: 3.95

>Number:         8151
>Category:       kern
>Synopsis:       System locks up with nfs loopback mounts
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Aug 05 14:35:01 +0000 1999
>Closed-Date:    
>Last-Modified:  
>Originator:     Christos Zoulas
>Release:        1.4
>Organization:
	Disorganization
>Environment:
System: NetBSD hrothgar.gw.com 1.3H NetBSD 1.3H (TAC-GENERIC) #2: Sun Nov 1 14:11:13 EST 1998 kim@hrothgar.gw.com:/net/hrothgar/src-1/NetBSD/cvsroot/src/sys/arch/i386/compile/TAC-GENERIC i386


>Description:
	Kernel dealocks/corrupts ffs filesystems when nfs is used as
	a loopback mount.
>How-To-Repeat:
	Mount localhost:/usr/src to /destdir
	cd /destdir, make build
>Fix:
	Don't use loopback mounts.
>Release-Note:
>Audit-Trail:

From: Izumi Tsutsui <tsutsui@ceres.dti.ne.jp>
To: gnats-bugs@netbsd.org
Cc: christos@zoulas.com, tsutsui@ceres.dti.ne.jp
Subject: kern/8151
Date: Fri, 6 Aug 1999 19:07:47 +0900 (JST)

 I saw the same problem on 1.3R. I guess this occured
 on only NFSv3, not NFSv2.

 When I tracked down this, "tcpdump -i lo0" said:

 23:54:08.081002 localhost.833822988 > localhost.nfs: 116 lookup [|nfs]
 23:54:08.081005 localhost.nfs > localhost.833822988: reply ok 236 lookup [|nfs]
 23:54:08.091000 localhost.833822989 > localhost.nfs: 112 lookup [|nfs]
 23:54:08.141003 localhost.nfs > localhost.833822989: reply ok 236 lookup [|nfs]
 23:54:08.141008 localhost.833822990 > localhost.nfs: 116 lookup [|nfs]
   :
 [snip]
   :
 23:54:10.501000 localhost.833823078 > localhost.nfs: 8316 write [|nfs]
 23:54:10.501007 localhost.833823079 > localhost.nfs: 116 commit [|nfs]
 23:54:10.501012 localhost.nfs > localhost.833823079: reply ok 152 commit [|nfs]
 23:54:10.511002 localhost.nfs > localhost.833823078: reply ok 160 write [|nfs]
 23:54:10.521000 localhost.833823080 > localhost.nfs: 8316 write [|nfs]
 23:54:10.521007 localhost.833823081 > localhost.nfs: 116 commit [|nfs]
 23:54:10.551000 localhost.nfs > localhost.833823081: reply ok 152 commit [|nfs]
 23:54:10.551003 localhost.833823082 > localhost.nfs: 116 commit [|nfs]
 23:54:10.551010 localhost.833823083 > localhost.nfs: 116 commit [|nfs]
 23:54:10.601000 localhost.833823080 > localhost.nfs: 8316 write [|nfs]
 23:54:10.771000 localhost.833823080 > localhost.nfs: 8316 write [|nfs]
 [... "write" rpc repeated and nfs hanged up]

 I guess that the kernel would hang up when it received
 the "commit" rpc contiguously (before replying the previous one).
 I have not tracked the sources, though.
 ---
 Izumi Tsutsui
 tsutsui@ceres.dti.ne.jp
>Unformatted:

Home	PR Database Search