net.ipv6.conf.INT.accept_ra_rt_info_max_plen defaults to 0

net.ipv6.conf.INT.accept_ra_rt_info_max_plen defaults to 0

[Pekka Savola]

Hello,

As reported and discussed below and in 
<http://www.mail-archive.com/debian-ipv6@lists.debian.org/msg03753.html>, 
to accept "route information option" on Linux, you have to manually 
change accept_ra_rt_info_max_plen sysctl value.  Other implementations 
that I know of accept this by default.

Is it intentional that Route Information options are ignored by 
default?

I suspect not -- if so, the default value should be (IMHO) 64 or if 
that's disagreeable, 48.

Even if this is intentional, I think the intentions should be 
revisited.

---------- Forwarded message ----------
Date: Mon, 2 Feb 2009 17:24:00 +0000
From: Teran McKinney <sega01@gmail.com>
Reply-To: radvd Development Discussion <radvd-devel-l@litech.org>
To: radvd Development Discussion <radvd-devel-l@litech.org>
Subject: Re: [radvd-devel-l] Linux and specific routes

Hey,

I have finally obtained my answer. I ended up reading RFC 4191,
glancing at RFC 2461, and looking at Linux's source code, but found
what I was looking for. It turns out that "route information option"
is more precise than "specific routes", so I googled for it and found
that Linux supports it just fine; all you have to do is set
accept_ra_rt_info_max_plen to the maximum prefix length you want to
accept. <http://www.mail-archive.com/debian-ipv6@lists.debian.org/msg03753.html>
is very helpful.

Cheers and thanks,
Teran

On Sun, Dec 21, 2008 at 18:27, Arnaud Ebalard <arnaud.ebalard@eads.net> wrote:
>
> "Teran McKinney" <sega01@gmail.com> writes:
>
>> I added the patch, but it does not seem to affect specific routes. I
>> was only using one router at first, but tried with an additional
>> router and had no luck. Specific routes just aren't been added :-(.
>> This configuration seems to work fine if manually added, so I'm not
>> sure what is wrong.
>
> If you compiled the kernel with the support and enabled the /proc
> entries to activate the feature, then it looks like a bug. Maybe you
> could drop a mail to the author of the patch.
>
> Cheers,
>
> a+
>
> --
> radvd-devel-l mailing list  :  radvd-devel-l@litech.org
> http://lists.litech.org/listinfo/radvd-devel-l
>

-- 
radvd-devel-l mailing list  :  radvd-devel-l@litech.org
http://lists.litech.org/listinfo/radvd-devel-l

Re: net.ipv6.conf.INT.accept_ra_rt_info_max_plen defaults to 0

[David Miller]

From: Pekka Savola <pekkas@netcore.fi>
Date: Tue, 3 Feb 2009 06:38:49 +0200 (EET)

> As reported and discussed below and in
> <http://www.mail-archive.com/debian-ipv6@lists.debian.org/msg03753.html>,
> to accept "route information option" on Linux, you have to manually
> change accept_ra_rt_info_max_plen sysctl value.  Other
> implementations that I know of accept this by default.
>
> Is it intentional that Route Information options are ignored by default?
>
> I suspect not -- if so, the default value should be (IMHO) 64 or if
> that's disagreeable, 48.
> 
> Even if this is intentional, I think the intentions should be revisited.

It looks very intentional, both via the code and it's documentation
in ip-sysctl.txt

It seems that it is disabled like this by default when
accept_ra_rtr_pref is enabled, and that seems pretty reasonable to me.

I'm sure whoever made that decision didn't do so on a whim and had
a very good reason for it.

Re: net.ipv6.conf.INT.accept_ra_rt_info_max_plen defaults to 0

[Teran McKinney]

Hey,

I personally think that 64 would be a reasonable default route length,
but am currently setting it to 120 in Icadyptes to be safe (probably
closer to excessive). More so, I'm just happy that it is supported,
even though it is disabled by default. Few people use it in the first
place, and I don't see what harm it could bring other than alternative
perspectives on potential security issues that already exist in NDP.

@Pekka: Thanks for bringing this up. I think a short note in
radvd.conf.example might help, but I should do a blog post so that
Google is more useful if you aren't searching with exactly the right
terminology.

Thanks,
Teran

On Tue, Feb 3, 2009 at 05:36, David Miller <davem@davemloft.net> wrote:
> From: Pekka Savola <pekkas@netcore.fi>
> Date: Tue, 3 Feb 2009 06:38:49 +0200 (EET)
>
>> As reported and discussed below and in
>> <http://www.mail-archive.com/debian-ipv6@lists.debian.org/msg03753.html>,
>> to accept "route information option" on Linux, you have to manually
>> change accept_ra_rt_info_max_plen sysctl value.  Other
>> implementations that I know of accept this by default.
>>
>> Is it intentional that Route Information options are ignored by default?
>>
>> I suspect not -- if so, the default value should be (IMHO) 64 or if
>> that's disagreeable, 48.
>>
>> Even if this is intentional, I think the intentions should be revisited.
>
> It looks very intentional, both via the code and it's documentation
> in ip-sysctl.txt
>
> It seems that it is disabled like this by default when
> accept_ra_rtr_pref is enabled, and that seems pretty reasonable to me.
>
> I'm sure whoever made that decision didn't do so on a whim and had
> a very good reason for it.
>

Re: net.ipv6.conf.INT.accept_ra_rt_info_max_plen defaults to 0

[Pekka Savola]

Let me add to what DaveM said:

On Sun, 15 Feb 2009, Teran McKinney wrote:
>> It seems that it is disabled like this by default when
>> accept_ra_rtr_pref is enabled, and that seems pretty reasonable to me.

accept_ra_rtr_pref is about router preferences. 
accept_ra_rt_info_max_plen is about more specific routes.  They are 
defined in the same RFC, and probably many implementers implement them 
at the same time.  However, there is no logical connection between the 
two, and making the default value of one depend on the default value 
of the other doesn't seem to make sense to me.

Even if we didn't enable accept_ra_rt_info_max_plen by default, I 
don't think it should depend on the value of accept_ra_rtr_pref.

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings