From owner-freebsd-current@FreeBSD.ORG Thu Sep 9 04:35:05 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 91F8316A4CE for ; Thu, 9 Sep 2004 04:35:05 +0000 (GMT) Received: from carver.gumbysoft.com (carver.gumbysoft.com [66.220.23.50]) by mx1.FreeBSD.org (Postfix) with ESMTP id 54B5643D5C for ; Thu, 9 Sep 2004 04:35:05 +0000 (GMT) (envelope-from dwhite@gumbysoft.com) Received: by carver.gumbysoft.com (Postfix, from userid 1000) id 4802F72DD4; Wed, 8 Sep 2004 21:35:05 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by carver.gumbysoft.com (Postfix) with ESMTP id 4350572DCB for ; Wed, 8 Sep 2004 21:35:05 -0700 (PDT) Date: Wed, 8 Sep 2004 21:35:05 -0700 (PDT) From: Doug White To: current@freebsd.org Message-ID: <20040908212739.O82773@carver.gumbysoft.com> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Subject: witness warnings in NFS X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 09 Sep 2004 04:35:05 -0000 Hey folks, If you are seeing lock warnings like this one, with the same (or close) file location and traceback, with WITNESS enabled, please let me know what the server and client OS versions and hardware are. I have one report that was a Solaris 10 server against a FreeBSD-CURRENT system running in vmware. I have a patch that quiets the warning and papers over the bug somewhat by faking an mbuf allocation failure as a packet parsing error. exclusive sleep mutex nfsd_mtx r = 0 (0xc08efb40) locked @ /usr/src/sys/nfsserver/nfs_serv.c:1103 [...] uma_zalloc_arg(c101f9a0,cc364a3c,2) at uma_zalloc_arg+0x41 nfsm_disct(cc364ac4,cc364ac8,14,e,46) at nfsm_disct+0xa2 nfsm_dissect_xx(14,cc364ac4,cc364ac8) at nfsm_dissect_xx+0x31 nfsrv_write(c14ff900,c19d8480,c13ec000,cc364ca8,cc364ca4) at nfsrv_write+0x1e2 The patch is at: http://people.freebsd.org/~dwhite/nfs_common.c.20040908.patch I'd be interested in knowing if someone with a way of creating an mbuf shortage condition can test this and see if they get EBADRPC errors. Thanks for any feedback! -- Doug White | FreeBSD: The Power to Serve dwhite@gumbysoft.com | www.FreeBSD.org