Re: [PATCH net-next] ipv6: RTAX_FEATURE_ALLFRAG causes inefficient TCP segment sizing

[Eric Dumazet <eric.dumazet@gmail.com>]

On Wed, 2012-04-25 at 11:20 +0200, Tore Anderson wrote:
> * Maciej Żenczykowski
> 
> >> But we chose to _not_ decrease mtu and adhere to the specs.
> >
> > I get that we _choose_ to behave such, and I agree this adheres to
> > specs.
> 
> "Chose" (past), not "choose" (present). ;-)
> 
> This patch does not make this choice. This patch merely fixes a bug in
> the implementation of the choice that was made a long time ago.
> 
> > But I'm not convinced that (even though this is allowed per RFC) this
> > is the right choice.
> 
> That is a different issue entirely, but I don't disagree with you. A
> "min_pmtu" sysctl or something like that would be useful.
> 
> > Also note that IPv6 prefers to see fragmentation happen at the end
> > hosts, and not at the routers.
> > Although of course it doesn't treat a tunnel end point as a router.
> 
> Actually, in IPv6, fragmentation *must* be performed by end hosts,
> routers (including tunnel end points) *cannot* fragment.
> 
> However, the use case for the allfrag feature is not handling tunnels,
> but IPv4<->IPv6 translation. The issue is that a IPv6 host may very 
> well
> receive an ICMPv6 Packet Too Big indicating a PMTU of <1280 that was
> originally transmitted by an IPv4 router (as an ICMPv4 Need To 
> Fragment)
> and underwent translation to IPv6.
> 
> In this case, the IPv6 node does not need to reduce the PMTU to <1280
> (Linux does not), but it is not invalid to have a <1280 MTU link in the
> IPv4 internet either, so something else must be done for the
> communication to work. The solution is then to include the IPv6 
> Fragment
> extension header, so that the translator have a suitable Identification
> value to copy into the translated IPv4 header, and may therefore clear
> the Don't Fragment flag, so that the IPv4 router will fragment the
> packet as it is forwarded onto the low-MTU link.
> 
> In case you're interested, I have a slide deck below that explains the
> use case for IPv4<->IPv6 translation. Slide 25 is about the particular
> corner case where the allfrag feature is necessary. URL:
> 
> http://fud.no/talks/20120417-RIPE64-The_Case_for_IPv6_Only_Data_Centres.pdf


Hmm, but what if we change linux to choice a) instead of b) ?

That is, not cap mtu to minimum value 1280 (and not use anymore
RTAX_FEATURE_ALLFRAG) : dst_allfrag() would be always false.

In this case, do we still need to send the frag header ?

I ask this because some TSO6 implementations probably dont cope very
well with this added header (untested path)