On Mon, 6 May 2002, Garrett Wollman wrote:
> <<On Mon, 6 May 2002 17:26:20 -0500 (CDT), Mike Silbersack <[EMAIL PROTECTED]> said:
>
> > Is doing this wise? I have this nagging feeling that randomizing (or
> > zeroing on each new connection) the timestamp would degrade its usefulness
> > for PAWS checks and the like. (Don't ask me how, I haven't thought it
> > through fully.)
>
> I don't think so, because the timestamps, as currently specified, are
> only meaningful within the context of a single connection. See
> sections 1.2, 4.3, and 4.2 of RFC 1323. The PAWS mechanism requires
> only that timestamps used by each connection be monotone increasing
> with respect to Sequence Number Arithmetic. RFC 1323 does require
> (section 4.2.2) that the clock be between 1 ms and 1 s in period,
> which I think we already violate on some platforms, although not
> seriously; there probably should be a pre-computed (global) scaling
> factor as well.
>
> -GAWollman
I looked over both our and Linux's tcp stack to double-check, and it
appears that my memory was faulty. You are correct, no PAWS checks are
done during TIME_WAIT recycling. Initializing to zero is probably the
best idea; getting fancy with random starts doesn't really help anything.
Mike "Silby" Silbersack
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-net" in the body of the message
