[email protected]# set interfaces ethernet eth2 address xxxx:xxxx:1400:1::/64 Error: xxxx:xxxx:1400:1::/64 is not a valid IP host Invalid value Value validation failed Set failed
On dummy interfaces it works flawlessly and Linux supports it as well.
[email protected]# show interfaces dummy dum0 | commands set address 'xxxx:xxxx:1400::/128'
Confirmed:
[email protected]# set interfaces dummy dum4444 address 2001:db8::/64 Error: 2001:db8::/64 is not a valid IP host
It is a valid IPv6 address - usage of all-zero is sometimes discouraged, but we should not prevent the user from selecting this address.
[email protected]# sudo ip -6 addr add dev dum0 2001:db8::/64 [email protected]# ip addr show dev dum0 9: dum0: <BROADCAST,NOARP,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default qlen 1000 link/ether e2:b3:40:aa:be:9c brd ff:ff:ff:ff:ff:ff inet6 2001:db8::/64 scope global valid_lft forever preferred_lft forever inet6 fe80::e0b3:40ff:feaa:be9c/64 scope link valid_lft forever preferred_lft forever
For dummy it works because it has the prefix /128 and it is allowed/validated https://github.com/vyos/ipaddrcheck/blob/c34294b7a18072c7bd4eaf8ec0f433bb374fad5c/src/ipaddrcheck.c#L577
I wonder if adding the is_ipv6_cidr here https://github.com/vyos/ipaddrcheck/blob/c34294b7a18072c7bd4eaf8ec0f433bb374fad5c/src/ipaddrcheck_functions.c#L109-L110 is a correct way to resolve it.
Or is it better to add an additional validator here https://github.com/vyos/vyos-1x/blob/b49997d81c07431038720bed0ac178d08c40fb62/interface-definitions/include/interface/address-ipv4-ipv6-dhcp.xml.i#L24-L27
This comment in ipaddrcheck explains why I disallowed all-zero host part addresses: https://github.com/vyos/ipaddrcheck/blob/current/src/ipaddrcheck_functions.c#L337-L346
As per https://www.rfc-editor.org/rfc/rfc4291#section-2.6.1 ,
The "subnet prefix" in an anycast address is the prefix that identifies a specific link. This anycast address is syntactically the same as a unicast address for an interface on the link with the interface identifier set to zero. Packets sent to the Subnet-Router anycast address will be delivered to one router on the subnet. All routers are required to support the Subnet-Router anycast addresses for the subnets to which they have interfaces. The Subnet-Router anycast address is intended to be used for applications where a node needs to communicate with any one of the set of routers.
So, according that RFC (which is a part of the standards track and is not obsoleted by anything yet), all-zero addresses do have an intended special use.
From this point of view, we must not accept them as interface addresses.
However, I'm curious if other platforms do allow that use or if anyone actually implements the "subnet router" mechanism.
Cisco and Arista support such addresses as interface addresses, and the anycast bit doesn't preclude unicast usage, so we should allow that.