Re: Objection to draft-ietf-6man-rfc4291bis-07.txt - /63 and /65 RAs on linux

[Alexandre Petrescu <[e-mail]>]

Le 02/03/2017 à 09:14, Erik Kline a écrit :
>
>
> On 2 March 2017 at 17:11, Alexandre Petrescu
> <[e-mail] <mailto:[e-mail]>> wrote:
>
>
>
>     Le 01/03/2017 à 20:53, Mikael Abrahamsson a écrit :
>
>         On Wed, 1 Mar 2017, Alexandre Petrescu wrote:
>
>             For information, today on linux a /65 RA is called "illegal"
>             and a
>             /63 RA makes a Host to _not_ configure an address.  What linux
>             wants is a 64 and nothing else.  This is not normal, and
>             it's not
>             programmer's fault.
>
>
>         Could you please elaborate on the last sentence. What do you mean by
>         "not normal"?
>
>
>     It is not normal for linux kernel to complain about a plen==65 in the RA
>     - it should not complain.  Do you think it is normal?
>
>     It is not normal for a linux kernel to _not_ form an IPv6 address when
>     it receives a plen==63.  It would have been normal to append the IID to
>     it.  Do you think it is abnormal for a linux kernel to form an IPv6
>     address when it receives a plen==63?
>
>
>
> These statements have zero basis in fact.  The Linux kernel is behaving
> according to all currently available (proposed) standards.

Well... do you have a standard for IPv6 over cellular links?   Or IPv6 
over USB?  I dont.  And I consider linux to act wrongly when being sent 
an RA plen==65 on these links and refusing them as "illegal".

I think it should not refuse it on Ethernet either, but there you have a 
standard RFC2464.  And for that I agree with you.  Even though it should 
be perfectly fine to use that plen==65 to have an entry in the rt table, 
and use a manually configured address in that plen.

I dont have a standard that tells that the computer is forbidden from 
introducing a 0 between 63 and 64 such that the plen==63 plus IID==64 
make for 128.

Alex