From regnauld@td.rh.dk  Wed Feb 21 01:05:44 2001
Return-Path: <regnauld@td.rh.dk>
Received: from ns1-dmz.rh.dk (gw-ext.rh.dk [130.226.106.3])
	by hub.freebsd.org (Postfix) with ESMTP id 5BA2837B401
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 21 Feb 2001 01:05:43 -0800 (PST)
	(envelope-from regnauld@td.rh.dk)
Received: from ns1.int.rh.dk (ns1-108.rh.dk [130.226.108.194])
	by ns1-dmz.rh.dk (Postfix) with ESMTP id 759F0567D2
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 21 Feb 2001 10:05:41 +0100 (CET)
Received: from aylee.td.rh.dk (aylee.td.rh.dk [172.31.211.70])
	by ns1.int.rh.dk (8.9.3/8.9.3/mailhub-rh-dk-0.1) with SMTP id KAA02796
	for <FreeBSD-gnats-submit@freebsd.org>; Wed, 21 Feb 2001 10:05:41 +0100 (CET)
	(envelope-from regnauld@td.rh.dk)
Received: by aylee.td.rh.dk (Postfix, from userid 1001)
	id C4996B946; Wed, 21 Feb 2001 10:05:40 +0100 (CET)
Message-Id: <20010221090540.C4996B946@aylee.td.rh.dk>
Date: Wed, 21 Feb 2001 10:05:40 +0100 (CET)
From: regnauld@td.rh.dk
Reply-To: regnauld@td.rh.dk
To: FreeBSD-gnats-submit@freebsd.org
Cc:
Subject: system deadlock via loopback NFS (maybe jail)
X-Send-Pr-Version: 3.113
X-GNATS-Notify:

>Number:         25245
>Category:       kern
>Synopsis:       mounting NFS to/from same host + activity will eventually lock up all processes
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    freebsd-bugs
>State:          closed
>Quarter:        
>Keywords:       
>Date-Required:  
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Wed Feb 21 01:10:01 PST 2001
>Closed-Date:    Sun Feb 10 04:50:34 PST 2002
>Last-Modified:  Sun Feb 10 04:51:48 PST 2002
>Originator:     Phil Regnauld
>Release:        FreeBSD 5.0-CURRENT i386
>Organization:
Copenhagen Kingdom Hospital - Rigshospitalet
>Environment:
System: FreeBSD beurk.balder 4.2-STABLE FreeBSD 4.2-STABLE #0: Tue Jan 30 23:41:34 CET 2001     regnauld@beurk.balder:/var/obj/var/src/sys/BEURK  i386

	
>Description:
	When mounting an NFS exported FS where the client and the server are the same 
	(not explicitly using loopback, but using the ethernet IP), it is possible
	to lockup the system after some activity has occurred (usually some hours).

>How-To-Repeat:
	Example: (this has been repeatedly observed on a system running jail, but should
	be repeatable without).

	- export /src or some other directory containing with lots of data

	- mount myhost:/src /usr/src (for example)

	You then have a NFS mount of a local filesystem onto another mountpoint (very
	practical for jails with the ports collection)

	Go into that directory, and do for example a "make world".

	Processes will progressively hang until the machine will completely freeze
	(to the point where console switching is possible, but typing does not
	have any effect...)

	Has been observed with /ports nfs mounted to /usr/ports

>Fix:

	No known workaround for this particular setup...

>Release-Note:
>Audit-Trail:
State-Changed-From-To: open->feedback 
State-Changed-By: iedowse 
State-Changed-When: Sun Nov 18 17:59:34 PST 2001 
State-Changed-Why:  

I believe Matt Dillon fixed this - have you seen it occur with 
more recent sources? 

http://www.FreeBSD.org/cgi/query-pr.cgi?pr=25245 
State-Changed-From-To: feedback->closed 
State-Changed-By: iedowse 
State-Changed-When: Sun Feb 10 04:50:34 PST 2002 
State-Changed-Why:  

Submitter wasn't able to reproduce this on recent sources. 

http://www.FreeBSD.org/cgi/query-pr.cgi?pr=25245 
>Unformatted:
