Re: TCP zero window probing
From: P Gentry (rdgentry1_at_cablelynx.com)
Date: 3 Apr 2004 12:18:32 -0800
Louis Laborde <email@example.com> wrote in message news:<406CA9A8.firstname.lastname@example.org>...
> It seems that to probe a closed receive window,
> linux TCP implementation sends an empty segment
> with its sequence number set to SND.UNA-1.
> I was wondering if this was compliant with RFC
> 793 which does not seem to describe precisely
> what such a probe should contain.
Hit the send button too soon. Sorry for following up my own post --
so quickly to boot!
In rfc 1122 I found the following when searching SND.UNA :
Here are some detailed error corrections and notes on the
Event Processing section of RFC-793.
(g) Check ACK field, ESTABLISHED state, p. 72: The ACK is a
duplicate if SEG.ACK =< SND.UNA (the = was omitted).
Similarly, the window should be updated if: SND.UNA
SEG.ACK =< SND.NXT.
And a much fuller discussion in section:
220.127.116.11 When to Send Data
Guess now I'll have to go looking up "SND.UNA" -- the others are a bit
more obvious in meaning/use.
email above disabled