> It doesn't make sense for the driver to have different modes for IPv4 
 > and IPv6. But there are two things we need to look at with some care:
 >   - we issue a DL_INFO_ACK separately for the IPv4 and IPv6 ill. How 
 > should IP behave if the two would resulting in different POINTOPOINT etc 
 > settings? Assert?

Personally, I'd expect it to fail the mismatching plumb and syslog()
something.

 >   - we currently do not set IFF_BROADCAST on an IPv6 interface 
 > (presumably the logic is that IPv6 doesn't offer a broadcast service on 
 > that interface even though the datalink provides a broadcast service.) 
 > Would anything break if we changed this?

Before getting that far, I think we neet to bolt down what the flag is
supposed to mean.  In general, I'd expect IP (logical/physical) interface
flags to tell applications/consumers about the IP (logical/physical)
interface capabilities, not the capabilities of the underlying datalink.
(Otherwise, what do flags like IFF_IPV6 ormean at all?)  So, having
IFF_BROADCAST clear on an IPv6 interface seems appropriate.

-- 
meem
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to