>> If some nastie sent a packet with source= P::c, destination P::d, you
>> might end up in an endless ping-pong loop of ICMP error messages, or...?
>> (does the "no icmp error sent due to received icmp error" prevent this?)
>
>Yes, it is. Additionally, even if we did not have the rule, we would
>not see such infinite iteration of icmp6 errors, because the source
>address of the error packet from Router B would P::b, which is
>a perfectly legal address. We would only see one more error.
to illustrate the situation better, it should be like this.
(correct me if I'm wrong)
itojun
without the behavior change on the draft:
router A -> router B src=P::c, dst=P::d, TTL=x (original packet)
router B -> router A src=P::c, dst=P::d, TTL=x - 1 (original packet)
.... (forward until TTL reaches 0)
with the behavior change on the draft:
router A -> router B src=P::c, dst=P::d (original packet)
(router B detects the anomaly on forwarding, and suppress
the packet forwarding and raise address unreachable)
router B -> router A src=P::b, dst=P::c, icmp6 address unreachable
(there'll be no response, end)
--------------------------------------------------------------------
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]
--------------------------------------------------------------------
