[EMAIL PROTECTED] wrote:
>>I think this is actually OK.   When the SYN arrives, it must be
>>accepted, as you quoted ...
>>      IP and higher layers (e.g., TCP, UDP) MUST
>>      continue to accept datagrams destined to a deprecated address
> 
> 
>       i think the sentence lacks "for existing connection" at the end.

Maybe not. I think the intention was for *our* node to not initiate
new connections using the deprecated address, not prevent other
nodes from doing so.

We could make it stricter of course, by not accepting even others
to initiate new connections. However, I believe this might be bad for
some protocols. Say, you are doing an FTP mget * from node X, which decides
to get a new RFC 3041 address in the middle of things. If I remember
correctly, FTP uses a new TCP session for each new file. Your FTP
session would not proceed if you refused to accept a new TCP session.

>>After that, TCP processes the incoming connection.   That makes existing
>>communications.   Then, when replying, the SYN+ACK can be sent from the
>>address to which it was sent (deprecated or not).   So, I doubt there's
>>actually a problem there.
> 
> 
>       TCP SYN is definitely an indication of a new connection.

Yes, but from the other end...

Jari


--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to