I am aware that duirng the intial SYN, SYN-ACK, ACK (connection establishment) phase, both TCP ends advertize TCP options (or extended features) each would like to use for the connection. Ultimately both ends end up using TCP options commonly supported and advertized by both.
My doubt is simple - Does the TCP negotiation happen only at the connection establishment phase? Is it possible for one end to stop using a previously advertised option and make room for another (specifically an option that is set (for the first time) on TCP by the application by issuing a setsockopt on the connected socket?
Could you elaborate more on this.
Once a connection is established, how would the re-negotiation be accomplished?
TCP implementations would be using special options meant only for use in the three-way handshake to turn on the corresponding feature.
E.g -
To enable Selective Acknowledgment, the option sent in the SYN segments is SACK PERMITTED. This, if successfully negotiated translates to SACK blocks in non-SYN segments.
Similarly, the usage of Big Window option requires special Big Window permitted option in the SYN segments. If successfully negotiated this translates to the actual Big Window option.
So can you, if possible explain to me as to how would options be turned off during the connection lifetime (in the non-SYN segments)