Re: [tcpm] ICMPv6 Error Handling at TCP draft-fujisaki-tcpm-icmpv6-reaction-00.txt

Hi,

 | >Anyway, it doesn't matter which specific ICMPv6 code falls into hard
 | >or soft here.
 | 
 | Exactly. That's why it doesn't make sense to do the mapping from 
 | ICMPv6 to ICMPv4. 

What he want to say is the classification of ICMPv6 is further
discussion. He did not say the mapping does not make sense.

RFC1122 is valid spec, and some systems implement RFC1122 TCP reaction
in SYN-SENT state. We want IPv6 nodes to react similarly.

 | The ICMP attacks draft is aiming at Informational, too. Yet virtually 
 | all vendors have implemented most of it. Nobody ever bothered whether 
 | it was Std track or Informational.

Is it in reverse order, isn't it?  Many vendors implement not to
disconnect established TCP session because it would have security
risks. And then ICMP attacks draft was written, as far as I read ML
archives.

 | If your specific problem arises from use of IPv6, then you can send 
 | the vendor (Microsoft?) the soft errors draft, and even direct them 
 | to the v6fix site. If that's not enough, I guess the only thing left 
 | is to have their clients tell them that it really sucks to have to 
 | wait dozens of seconds to browse a web page.

Sorry for repeatedly, but we think soft errors draft does not provide
enough information to implement the TCP reaction. We are afraid every
system react differently when it receives ICMP errors.

Yours Sincerely,
--
Tomohiro Fujisaki

_______________________________________________
tcpm mailing list
tcpm@ietf.org
https://www1.ietf.org/mailman/listinfo/tcpm