From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH] IPv6: optionaly validate RAs on raw sockets Date: Wed, 11 Jul 2007 13:56:46 -0700 (PDT) Message-ID: <20070711.135646.115909857.davem@davemloft.net> References: <200707111919.11746@auguste.remlab.net> Mime-Version: 1.0 Content-Type: Text/Plain; charset=iso-8859-1 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: dlstevens@us.ibm.com, vladislav.yasevich@hp.com, netdev@vger.kernel.org, yoshfuji@linux-ipv6.org To: rdenis@simphalempin.com Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:43014 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1758161AbXGKU4X convert rfc822-to-8bit (ORCPT ); Wed, 11 Jul 2007 16:56:23 -0400 In-Reply-To: <200707111919.11746@auguste.remlab.net> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org =46rom: R=E9mi_Denis-Courmont Date: Wed, 11 Jul 2007 19:19:11 +0300 > Le mercredi 11 juillet 2007, David Stevens a =E9crit : > > That sounds like a good idea to me (FWIW), > > though I also still think a simple raw-socket > > application would do it just fine, possibly with > > no kernel modification at all. > > But since the kernel wouldn't be maintaining > > the DNS info, which was my real objection to the > > original version, netlink would work well too. >=20 > One remaining corner case is NFS/IPv6 root, whereby userland won't ha= ve=20 > a chance to start before the network, and hence may miss the solicite= d=20 > RA. Or would it? By default, the next unsolicited RA can be anytime=20 > from now to after 10 minutes, so that's not sufficient. I wouldn't=20 > personnaly care, but... We already have cases like that with network device firmware that has to be loaded in from the filesystem in userspace, and the answer is to use a properly populated initrd. Same goes for things like this. That's the fact of life these days, like it or not.