Apple fdff::/64 Bug

[Fabian Bläse]

Hallo zusammen,

nach über 2 Jahren hat Apple tatsächlich mal auf meinen Bugreport reagiert.
Siehe unten, FYI.

Hab nochmal passend drauf geantwortet und halte euch auf dem laufenden.

Gruß
Fabian


Report:
> Summary: macOS doesn't correctly set routes for advertised prefixes, if they are announced with AdvDefaultLifetime set to zero (= not to be used as default route)
> 
> 
> Steps to Reproduce: Connect to a Router, which advertises ULA prefixes with AdvDefaultLifetime set to zero.
> 
> 
> Expected Results: macOS assigns an address for each advertiesed prefix to the corrosponding interface and sets a route to reach that subnet.
> (i.e. fd43:1:1:1::/64 is advertised -> fd43:1:1:1::/64 sould be added to the routing table)
> 
> Actual Results: macOS assigns an address for each advertiesed prefix to the corrosponding interface but doesn't sets a route to reach that subnet.
> If AdvDefaultLifetime is not zero, routes are correctly set (default and subnet specific route.

Reply:
> After reviewing your feedback, we need more information to continue our investigation.
> 
> Can you please explain your use-case in a bit more detail for us please?
> 
> The implementation is such that if the reachability of the router advertising the prefix is not known, the advertising prefixes are marked offlink.
> 
> For an RA with lifetime 0, there is no instance of router created on the stack and therefore reachability is an unknown. Even if we were to add such an instance, it will not be actively resolved.
> 
> If the intent is only to communicate on the LAN, the link local addressing should suffice.
> Why do we need ULA for that and why can’t just LLA addressing work?

-------------- nächster Teil --------------
Ein Dateianhang mit Binärdaten wurde abgetrennt...
Dateiname   : signature.asc
Dateityp    : application/pgp-signature
Dateigröße  : 833 bytes
Beschreibung: OpenPGP digital signature
URL         : <https://lists.freifunk.net/pipermail/franken-dev-freifunk.net/attachments/20191216/de520460/attachment.sig>