On 07/28/10 12:28 PM, Daniel Gavelle wrote:
Erik,
The ttl check is on the receiver. All of RFC 4861 says that a ND
message (NS, NA, RS, RA, redirect) with ttl != 255 must be discarded
by the receiver. The multi-hop DAD use of NS/NA in 6lowpan-nd is an
exception to that.
Some stacks have the rule that any message with a TTL of 255 should
never be forwarded on transmit.
FWIW That doesn't seem to have any basis in an RFC.
In any case, your comments are not about the security issues with the
ttl check, but about the ttl used to transmit the multihop dad. Those
are two orthogonal issues.
Also a TTL of 255 will remain in the
network for a long time if there is a loop. I think it would be better
if the multi-hop messages were sent with a smaller TTL (e.g. 64) rather
than 255. The exception on receive you mention would still be required.
That we could fix by introducing a protocol constant for the multihop
DAD hoplimit and referring to it in the spec.
Is 64 a reasonable upper limit on the diameter for a 6lowpan?
Erik
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
