From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jeff Layton Subject: Re: [PATCH 0/4] nfsd: add support for NFSv4 callbacks over IPv6 Date: Wed, 17 Jun 2009 15:01:47 -0400 Message-ID: <20090617150147.39303bec@tlielax.poochiereds.net> References: <1245262541-9362-1-git-send-email-jlayton@redhat.com> <20090617184701.GD24040@fieldses.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Cc: linux-nfs@vger.kernel.org, chuck.lever@oracle.com To: "J. Bruce Fields" Return-path: Received: from mx2.redhat.com ([66.187.237.31]:51252 "EHLO mx2.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758609AbZFQTBw (ORCPT ); Wed, 17 Jun 2009 15:01:52 -0400 In-Reply-To: <20090617184701.GD24040@fieldses.org> Sender: linux-nfs-owner@vger.kernel.org List-ID: On Wed, 17 Jun 2009 14:47:01 -0400 "J. Bruce Fields" wrote: > On Wed, Jun 17, 2009 at 02:15:37PM -0400, Jeff Layton wrote: > > This patchset is a first pass at adding IPv6 callback channel support > > for knfsd. The set is fairly straightforward, but it does require a > > number of changes to server side NFSv4 related structs that store > > addresses in places that are only suitable for IPv4 addresses. > > > > I've tested this by having Linux and OpenSolaris clients mount the > > server over an IPv6 socket, get a delegation and ensure that the server > > can recall that delegation. It all seems to work as expected. IPv4 > > callbacks also seem to continue to work correctly. > > > > This patchset does change the some of the new 4.1 functions > > (nfsd4_exchange_id in particular). Those changes are untested as of yet > > but I'll see if I can do so if the approach in this set seems > > reasonable. > > Thanks, just two quick comments: > > 1. I've been working on the callback code, so check that this > applies against for-2.6.31 (at > git://linux-nfs.org/~bfields/linux.git). Sorry, I should have made it clear in the original email. This patchset applies cleanly to your master branch. I just checked and it also applies cleanly to your for-2.6.31 branch. > 2. Any IP address checks in setclientid or exchange_id are > probably bogus; the former should be gone, the latter may still > be there but there'll be a patch queued up for 2.6.31 to fix > that. > There are definitely still IP address comparisons in those functions, though I confess I didn't look closely at the logic. My main concern with this set was to move the IPv4 specific address containers to sockaddr_storage, and to make the comparisons work for alternate address families. I tried not to alter the logic however. Thanks, -- Jeff Layton