netdev - Re: [PATCH net 0/2] net: ipv6: seg6: headroom fixes

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <26689.1580905297@dooku.sandelman.ca>
Date:   Wed, 05 Feb 2020 13:21:37 +0100
From:   Michael Richardson <mcr@...delman.ca>
To:     Alexander Aring <alex.aring@...il.com>
cc:     netdev@...r.kernel.org
Subject: Re: [PATCH net 0/2] net: ipv6: seg6: headroom fixes

Alexander Aring <alex.aring@...il.com> wrote:
    > This patch series fixes issues which I discovered while implementing RPL
    > source routing for 6LoWPAN interfaces. 6LoWPAN interfaces are using a MTU
    > of 1280 which is the IPv6 minimum MTU. I suppose this is the right fix to
    > do that according to my explanation that tunnels which acting before L3
    > need to set this headroom. So far I see only segmentation route is affected
    > to it. Maybe BPF tunnels, but it depends on the case... Maybe a comment
    > need to be added there as well to not getting confused. If wanted I can
    > send another patch for a comment for net-next or even net? May the
    > variable should be renamed to l2_headroom?

I had discussed this with Alex over the past few days.
I had not looked closely at the code during that discussion, and maybe my
comments in chat were wrong.  So these patches don't look right to me.

I think that the issue we have here is that things are big vague when it
comes to layer-2.5's, and fatter layer-3s.  Maybe this is well established in
lore...

My understanding is that headroom is a general offset, usually set by the L2
which tells the L3/L4 how much to offset in the SKB before the ULP header is
inserted.   TCP/UDP/SCTP/ESP need to know this.

MPLS is a layer-2.5, and so it quite weird, because it creates a new L2
which lives upon other L2 and also other L3s.

Segment routing, and RPL RH3 headers involve a fatter L3 header.

Of course, one could mix all of these things together!

--
]               Never tell me the odds!                 | ipv6 mesh networks [
]   Michael Richardson, Sandelman Software Works        | network architect  [
]     mcr@...delman.ca  http://www.sandelman.ca/        |   ruby on rails    [

Download attachment "signature.asc" of type "application/pgp-signature" (488 bytes)