From mboxrd@z Thu Jan 1 00:00:00 1970 From: "=?ISO-8859-1?Q?S=E9rgio_Gomes?=" Subject: Re: Routing header reversal in Linux kernel Date: Mon, 22 May 2006 15:43:32 +0000 Message-ID: <89fb8be00605220843u7c0869d6n4f9b004966db520@mail.gmail.com> References: <89fb8be00605220642l3730a60alf7015ee626b8ed69@mail.gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Return-path: Received: from wx-out-0102.google.com ([66.249.82.205]:45899 "EHLO wx-out-0102.google.com") by vger.kernel.org with ESMTP id S1750704AbWEVPnd convert rfc822-to-8bit (ORCPT ); Mon, 22 May 2006 11:43:33 -0400 Received: by wx-out-0102.google.com with SMTP id s6so873395wxc for ; Mon, 22 May 2006 08:43:32 -0700 (PDT) To: netdev@vger.kernel.org In-Reply-To: <89fb8be00605220642l3730a60alf7015ee626b8ed69@mail.gmail.com> Content-Disposition: inline Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org OK, I've simplified the problem greatly, please ignore my previous post= =2E I am no longer using the backdoor and I am following the RFC by using an authentication header between both machines. This, in theory, should allow for routing header reversal, i.e., when I send a packet from one machine with a routing header, the reply should follow the same route, in the opposite direction. But alas, the routing header is not being reversed in the 2nd machine, whether it be on a listening or on an active socket. I believe the implementation is therefore incorrect. However, I am by no means an expert, and may be overlooking some detail, like a /proc option or a later RFC that specifies the AH usage in more detail. Would anyone care to comment on this? On 5/22/06, S=E9rgio Gomes wrote: > Greetings! > > I am a final year student in Instituto Superior T=E9cnico, Lisbon, > Portugal, and part of my final project requires the use of IPv6 > routing headers. > > In particular, I need the route reversal to work. Now, as it didn't, = I > began analysing the source code, and found a comment, stating that it > was a known problem with the IPv6 standard (as I was aware) and that = a > way had been built around it via a constant, to avoid the whole AH > mess the RFC imposes. > > So I was hoping someone could help me with something. The problem is > that even if I remove that verification, it does not work -- > specifically, it looks as though the kernel is prepared to reverse > routing headers on a listening socket (i.e., a passive one), but not > on an active > socket (that is, one making the connection). My terminology is > probably wrong, but I hope you got my point. > > However, I have only looked into this now, so I barely understand the > IPv6 implementation in the Linux kernel. I was hoping that you would > be able to shed some light into the subject, namely: is it at all > possible to have a TCP socket initiate a connection, receive a SYN+AC= K > from the remote socket that includes a routing header, and reverse > this header for the remaining packets in the TCP connection? > > Thanks! > > S=E9rgio Gomes >