Return-Path: linux-nfs-owner@vger.kernel.org Received: from fieldses.org ([174.143.236.118]:59535 "EHLO fieldses.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756561Ab3ETVZF (ORCPT ); Mon, 20 May 2013 17:25:05 -0400 Date: Mon, 20 May 2013 17:25:03 -0400 From: "J. Bruce Fields" To: "Myklebust, Trond" Cc: "linux-nfs@vger.kernel.org" Subject: Re: [PATCH 6/6] NFSv4.1: Enable state protection Message-ID: <20130520212502.GJ14677@fieldses.org> References: <1369070227-40671-1-git-send-email-Trond.Myklebust@netapp.com> <1369070227-40671-2-git-send-email-Trond.Myklebust@netapp.com> <1369070227-40671-3-git-send-email-Trond.Myklebust@netapp.com> <1369070227-40671-4-git-send-email-Trond.Myklebust@netapp.com> <1369070227-40671-5-git-send-email-Trond.Myklebust@netapp.com> <1369070227-40671-6-git-send-email-Trond.Myklebust@netapp.com> <20130520204009.GH14677@fieldses.org> <1369084243.52168.1.camel@leira.trondhjem.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: <1369084243.52168.1.camel@leira.trondhjem.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Mon, May 20, 2013 at 09:10:46PM +0000, Myklebust, Trond wrote: > On Mon, 2013-05-20 at 16:40 -0400, J. Bruce Fields wrote: > > On Mon, May 20, 2013 at 01:17:07PM -0400, Trond Myklebust wrote: > > > Use the EXCHGID4_FLAG_BIND_PRINC_STATEID exchange_id flag to enable > > > stateid protection. This means that if we create a stateid using a > > > particular principal, then we must use the same principal if we > > > want to change that state. > > > > Note that knfsd ignores this--its EXCHANGE_ID will always return with > > the flag unset regardless of what the client requests. My understanding > > is that that's legal ("Whether a bit is set or cleared on the arguments' > > flags does not force the server to set or clear the same bit on the > > results' side.") > > > > (Definitely not opposed to implementing it, just haven't gotten around > > to it.) > > Right, but the point here is that the server is allowed to set > EXCHGID4_FLAG_BIND_PRINC_STATEID without the client requesting it, so > the client is supposed to always be ready for that situation. > > This patch series is therefore more about ensuring that the Linux client > is spec compliant. Sure, makes sense. --b.