From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: Linux IPv6 DAD not full conform to RFC 4862 ? Date: Wed, 09 Jan 2008 15:55:44 -0800 (PST) Message-ID: <20080109.155544.136768603.davem@davemloft.net> References: <20080109153656.GA16962@pingi.kke.suse.de> <20080109.153212.144388472.davem@davemloft.net> <20080110.084655.66182071.yoshfuji@linux-ipv6.org> Mime-Version: 1.0 Content-Type: Text/Plain; charset=iso-2022-jp Content-Transfer-Encoding: 7bit Cc: kkeil@suse.de, netdev@vger.kernel.org To: yoshfuji@linux-ipv6.org Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:53182 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1753148AbYAIXzo (ORCPT ); Wed, 9 Jan 2008 18:55:44 -0500 In-Reply-To: <20080110.084655.66182071.yoshfuji@linux-ipv6.org> Sender: netdev-owner@vger.kernel.org List-ID: From: YOSHIFUJI Hideaki / 吉藤英明 Date: Thu, 10 Jan 2008 08:46:55 +0900 (JST) > In article <20080109.153212.144388472.davem@davemloft.net> (at Wed, 09 Jan 2008 15:32:12 -0800 (PST)), David Miller says: > > > I question any RFC mandate that shuts down IP communication on a node > > because of packets received from remote systems. > > RFC4862 tell us that we SHOULD disable IP communication. > (IP means IPv6 here; IPv4 is out of scope.) > In IETF term, a SHOULD is almost a MUST. We are required to follow > unless we have very good reason to ignore it. A DoS by definition is a very good reason. > > If the TAHI test can trigger this, so can a compromised system on your > > network and won't that be fun? :-) > > So, I know the specification, but I have ignored it. > I think it is fine to implent in some way, but I do think we must have > a switch not to do this. Because of the above, the existing behavior must still stay the default. I hope this is your plan. By default Linux will not implement this SHOULD, it's a security issue. I more and more do not like these conformance tests, they leave no room whatsoever for handling bugs or ill-specified features in the specification.