Re: [tcpm] Definition of SMSS/RMSS in RFC5681 are inaccurate?!

[Joe Touch <[e-mail]>]

Hi, all,

On 7/26/2010 10:06 AM, Ethan Blanton wrote:
> Alexander Zimmermann spake unto us the following wisdom:
>>> Yes, RFC5681 is correct,
>>
>> Sorry, but I disagree. I cite the sentence in question again:
>>
>> "The size does not include the TCP/IP headers and options."
>>
>> The "and" in the sentence is IMO wrong or at least very very misleading.
>> Either we write: "The size does include ONLY the fixed TCP/IP headers."
>> or we write "The size does not include the TCP/IP options"
>
> I believe (as one of the authors) that 5681 is correct.  Neither the
> fixed TCP header size nor the option space are included in this
> calculation for the purposes of 5681.

draft-ietf-tcpm-tcpmss states that the MSS option includes the fixed TCP 
and IP headers and NOT the options.

I think Alex did find a bug here (I had commented on his misquoted 
earlier post), as per below... it's not in BOTH descriptions, though.

In 5681, RMSS is in error as follows:

    RECEIVER MAXIMUM SEGMENT SIZE (RMSS): The RMSS is the size of the
       largest segment the receiver is willing to accept.  This is the
       value specified in the MSS option sent by the receiver during
       connection startup.  Or, if the MSS option is not used, it is 536
       bytes [RFC1122].  The size does not include the TCP/IP headers and
       options.

If RMSS is ever the value in the MSS option, it would include the fixed 
headers but not the options.

 > It is concerned only with data
> bytes sent in-band.  If SMSS is used differently elsewhere, that is
> unfortunate, but it does not affect the definition within 5681.

Only RMSS is in error; as you note, SMSS is correct.
5681 says SMSS is based on RMSS (where "based on" isn't explained):

    SENDER MAXIMUM SEGMENT SIZE (SMSS): The SMSS is the size of the
       largest segment that the sender can transmit.  This value can be
       based on the maximum transmission unit of the network, the path
       MTU discovery [RFC1191, RFC4821] algorithm, RMSS (see next item),
       or other factors.  The size does not include the TCP/IP headers
       and options.

SMSS clearly must omit all TCP and IP headers, because it is the value 
that updates cwnd.

So, IMO, yes, there's a bug, but it's in RMSS's description. There may 
be a useful clarification that defines SMSS in terms of RMSS explicitly 
as well.

Joe