From owner-freebsd-fs@FreeBSD.ORG  Tue Aug  2 16:50:15 2005
Return-Path: <owner-freebsd-fs@FreeBSD.ORG>
X-Original-To: fs@freebsd.org
Delivered-To: freebsd-fs@FreeBSD.ORG
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 7DCB316A41F
	for <fs@freebsd.org>; Tue,  2 Aug 2005 16:50:15 +0000 (GMT)
	(envelope-from rick@snowhite.cis.uoguelph.ca)
Received: from mailhub.cs.uoguelph.ca (mailhub.cs.uoguelph.ca [131.104.96.75])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 2F35D43D46
	for <fs@freebsd.org>; Tue,  2 Aug 2005 16:50:14 +0000 (GMT)
	(envelope-from rick@snowhite.cis.uoguelph.ca)
Received: from snowhite.cis.uoguelph.ca (snowhite.cis.uoguelph.ca
	[131.104.48.1])
	by mailhub.cs.uoguelph.ca (8.13.1/8.13.1) with ESMTP id j72GoDg8006585; 
	Tue, 2 Aug 2005 12:50:13 -0400
Received: (from rick@localhost)
	by snowhite.cis.uoguelph.ca (8.9.3/8.9.3) id MAA45945;
	Tue, 2 Aug 2005 12:50:42 -0400 (EDT)
Date: Tue, 2 Aug 2005 12:50:42 -0400 (EDT)
From: rick@snowhite.cis.uoguelph.ca
Message-Id: <200508021650.MAA45945@snowhite.cis.uoguelph.ca>
To: kris@obsecurity.org
X-Scanned-By: MIMEDefang 2.44
Cc: fs@freebsd.org, openbsd-nfsv4@sfobugs.org
Subject: Re: FreeBSD6.0-BETA1 panics
X-BeenThere: freebsd-fs@freebsd.org
X-Mailman-Version: 2.1.5
Precedence: list
List-Id: Filesystems <freebsd-fs.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-fs>,
	<mailto:freebsd-fs-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-fs>
List-Post: <mailto:freebsd-fs@freebsd.org>
List-Help: <mailto:freebsd-fs-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-fs>,
	<mailto:freebsd-fs-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Tue, 02 Aug 2005 16:50:15 -0000

> Didn't help.  Running two cvs updates over NFS and one locally over
> ufs (all against the same repo) again caused a deadlock in a few
> seconds:
[good diagnostic stuff snipped]

Ok, the race is obvious (although the fix not quite so:-). It is specific
to the FreeBSD port. Most of my development has been done under OpenBSD
and it doesn't require the vnode to be locked for a VOP_GETATTR(). To
"fix" this for FreeBSD, I put a little glue routine called nfsrvn_getattr()
in that locks the vnode before the VOP_GETATTR() call. Unfortunately this
caused the race to hit (and probably several others at other nfsrvn_getattr()
calls).

I think the fix is probably to move the call down until after the vnode
that was looked up has been unlocked, but it will take a little time.

For the OpenBSDers: This race won't be a problem in OpenBSD3.7.

Thanks yet again for the testing and I'll let you know when I have
something for you to try, rick