David S. Miller wrote:
HP printers had a different TCP bug :-)
The workaround for that is enabled by setting tcp_retrans_collapse to
zero.
If you resegment the queue during retransmits the old printer TCP
stacks refuse to take the retransmitted data.
The 16-bit signed window bug exists in some of the older DOS TCP
stacks and I would not eliminate the possibility that some people use
that code base in embedded applications.
So, it was "ok" to default to something that wouldn't interoperate well with
arguably broken HP printers, but it is not "ok" to default to something that
wouldn't interoperate well with broken DOS stacks?!?
Didn't Jesse say something about the 2.4 stack being "ok" wrt growing the window
beyond 32767 bytes? 2.4 has had lots of exposure right? Isn't that something
of an existence proof that the issue with older DOS TCP stacks is today a non-issue?
rick jones
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
