Hi, Brian,

On 08/19/2014 09:09 PM, Brian E Carpenter wrote:
>> On 08/19/2014 07:58 PM, Brian E Carpenter wrote:
>>>> It does seem kind of silly that we say "you must support MTU >= 1280 to run
>>>> IPv6" and then allow PTB packets with an MTU < 1280. Any reason we can't
>>>> simply say that PTB packets < 1280 are invalid?
>>> Because of SIIT, that is equivalent to saying that the minimum IPv4
>>> MTU is now 1260. That might be a discussion worth having, but 576 has
>>> been around for a long time.
>>
>> Not sure what you meant about 576... that we can assume that to be a
>> minmum MTU, 
> 
> Yes, that hasn't changed since RFC 791 

Not really. For v4, 576 is the "minimum reassembly buffer size" (you're
guaranteed to the remote host can reassemble a datagram of that size),
not the minimum MTU.

The IPv4 minimum MTU is actually 68 bytes. While working on RFC5927, I
recall that at least OpenBSD enforced a lower limit as low as (around)
296 bytes, since that accommodated some radio links that were known to
be in use at the time...


> Maybe we consider it acceptable that SIIT will break on paths that
> include a shorter-than-Ethernet link MTU. But we need to make that
> statement explicit.

Or just update SIIT along with deprecating the generation of atomic
fragments. -- At the end of the day, not that long ago there were at
least a handful of implementations that didn't react to ICMPv6 PTB<1280
as required by RFC2460. Some we might already have scenarios in which
the SIIT device expects the host to generate atomic fragments, but it
doesn't.

Thanks!

Best regards,
-- 
Fernando Gont
SI6 Networks
e-mail: [email protected]
PGP Fingerprint: 6666 31C6 D484 63B2 8FB1 E3C4 AE25 0D55 1D4E 7492




_______________________________________________
OPSEC mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/opsec

Reply via email to