It doesn't currently
> T/TCP expects to see data, but looking at the comments in the file
> tcp_input.c from 2.1.86 I see that it opens up the way for a denial of
> service attack. Will this option be implemented in future versions of the
> kernel?
You cant do it when using syn cookies, you risk a major DoS attack and
the sender is violating RFC793 by sending data into a possibly closed
window. Thats one of the man reasons T/TCP is doomed to stay experimental.
> If there isn't a patch available, what would be the best way of going
> about queueing the data, storing it in a seperate queue in the sock
> structure and then using the data when ready?
Probably to limit it to about 4K and to queue it in the data quue for
the socket. If you fix tcp_rcv for those cases to drop into the data
path not blindly flush the frame then the right things will (I think)
occur elsewhere.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu