On Tue, 2007-07-17 at 16:32 +0100, Kieran Mansley wrote: > My guess is that some other stack is being a bit unusual by not > advertising window space in its SYN packets, and lwIP is failing to > reply because it is simple and doesn't have special cases for stuff like > this. Am I right?
Having had a look at the RFCs they are a bit ambiguous. On one hand they say that the window field in the header refers to data octets (which suggests that SYNs, FINs etc, which are not data, are not limited by the window). On the other hand it mentions a set of packets that must be accepted even if you have a zero window and SYNs and FINs aren't in that list. Assuming I'm right about lwIP being responsible for not sending a SYN- ACK, I think we should relax it a little and allow it to send SYN and FIN packets with no available window space. It's not forbidden by the RFC, there seems to be some ambiguity there, and it has the benefit of allowing the connection to continue with little chance of a downside. My guess is that other stacks do this too (anyone want to confirm this of things like Linux/Windows)? It's also not a new query, see this from 1986: http://www- mice.cs.ucl.ac.uk/multimedia/misc/tcp_ip/8601.mm.www/0047.html and Vint Cerf agrees with me! Awesome: http://www- mice.cs.ucl.ac.uk/multimedia/misc/tcp_ip/8601.mm.www/0052.html Thanks Kieran _______________________________________________ lwip-users mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/lwip-users
