RE: 6MAN: Adoption call on draft-hinden-6man-rfc1981bis-01

["Templin, Fred L" <[e-mail]>]

Hi Bob,

> -----Original Message-----
> From: Bob Hinden [mailto:[e-mail]]
> Sent: Thursday, February 04, 2016 11:16 AM
> To: Templin, Fred L
> Cc: Bob Hinden; Fred Baker (fred); IPv6 List
> Subject: Re: 6MAN: Adoption call on draft-hinden-6man-rfc1981bis-01
> 
> Fred,
> 
> > On Feb 4, 2016, at 10:58 AM, Templin, Fred L <[e-mail]> wrote:
> >
> > Hi Fred,
> >
> >> -----Original Message-----
> >> From: Fred Baker (fred) [mailto:[e-mail]]
> >> Sent: Thursday, February 04, 2016 10:50 AM
> >> To: Templin, Fred L
> >> Cc: Bob Hinden; 6man WG
> >> Subject: Re: 6MAN: Adoption call on draft-hinden-6man-rfc1981bis-01
> >>
> >>
> >>> On Feb 4, 2016, at 10:37 AM, Templin, Fred L <[e-mail]> wrote:
> >>>
> >>> Your text could be modified by simply adding the word "accurate" as follows:
> >>>
> >>> "It defines a method for Packetization Layer Path MTU Discovery (PLPMTUD)
> >>>   designed for use in environments in which delivery of accurate ICMP messages
> >>>   to a host are not assured."
> >>
> >> So you added the word "accurate". Anything else?
> >>
> >> My question would be what makes an ICMP "accurate", and how the host might or might not determine accuracy. As near as I can
> tell,
> >> RFC 1981 says that if a host receives an ICMP Packet Too Big, it should make the packet smaller. If the ICMP has been spoofed,
> unless
> >> the spoofer has done something really egregious, I don't know how the receiving host would know that.
> >
> > ICMP spoofing is exactly the point. But, it is not up to the host to determine
> > whether an ICMP is accurate - it is up to the host to determine whether it is
> > in an environment in which delivery of accurate ICMP messages is not assured.
> >
> >> Given that, I don't think the wording change helps.
> >
> > Without "accurate", the text does not address failure mode 2).
> 
> How is a host supposed to know if it is an “accurate ICMP message”?

Repeating what I said with a necessary correction:

  "But, it is not up to the host to determine
    whether an ICMP is accurate - it is up to the host to determine whether it is
    using a path for which delivery of accurate ICMP message is not assured."

> This is starting to read to me like creating a new specification, not advancing the one we have now.

Not at all. You were only too happy to accept Fred Baker's suggestion; I
have simply modified the aspects of his suggestion that were lacking
and/or inaccurate.

Thanks - Fred
[e-mail]

> Bob