Re: I-D Action:draft-ietf-6man-exthdr-02.txt

[Suresh Krishnan <[e-mail]>]

Hi John,

On 11-03-15 04:15 PM, John Leslie wrote:
> Suresh Krishnan <[e-mail]> wrote:
>> On 11-03-15 09:25 AM, John Leslie wrote:
>>>>> But I don't see the equivalent of Section 4.2 of RFC 2460, specifying
>>>>> the TLV format.
>>>> The T is the "Next Header", the L is the "Hdr Ext Len" and V is the 
>>>> "Header Specific Data" as specified in the figure in Section 4 of the
>>>> draft.
>>> Well, of course "Next Header" _isn't_ the Type of this option (rather
>>> it's the Type of the next option).
>>>
>>> And the "Hdr Ext Len" isn't a particularly intuitive coding of Length
>>> either...
>> Yep. You are right on both counts, but I am not sure how we can change 
>> this. We cannot chain the headers without the T being in the *previous* 
>> header. I think the best we can do is to refrain from calling this TLV 
>> like you said.
> 
>    That would satisfy me...
> 
>    But, the 2460 section 4.2 TLV also defines four actions when a Type
> isn't recognized (skip and three cases of discard). exthdr-02 gives
> us no way of defining skip-vs-discard.
> 
>    I think some mention of this issue would be wise. It appears, at
> first blush, that you intend for a middlebox to skip over these
> prospective extension headers, whereas 2460 calls for dropping the
> packet.

Right. These behavior flags were added in version -01 of the draft (the 
previous version) due to a apparent consensus at the physical wg meeting 
but this change has been rolled back in version -02 due to lack of 
consensus in the mailing list.

> 
>    I'm not prepared to say what the resolution may be; I just think
> it deserves some text (even if only to say, "Beware!").

Let's see if we can make some progress at the next meeting.

Thanks
Suresh