gnrc/ipv6_auto_subnets: add option to derive subnet prefix from EUI

[benpicco]

Contribution description

This adds the gnrc_ipv6_auto_subnets_eui pseudo-module.

We have a number of wireless 802.15.4 nodes that each have a set of wired nodes connected locally.

[6LBR] - - - - [6LR]----[Node A]
   ¦             |
   ¦             \------[Node B]
   ¦
    \ - - - - - [6LR]---[Node C]
                  |
                  \-----[Node D]

To limit communication over the wireless link (The gnrc_ipv6_auto_subnets sync algorithm creates O(n) messages and does not deal well with lost messages) an alternative scheme of assigning subnet prefixes is needed.

Since this is a local sensor network, we can just choose a sufficiently small upstream prefix (e.g. fd12::/16) and create the downstream network by filling in the remaining bits with the EUI of the downstream wired interface.

Testing procedure

We have an upstream router that distributes the fd12::/16 prefix on the wireless network:

make PREFIX_CONF=static IPV6_PREFIX="fd12::/16" -C examples/gnrc_border_router all term

A routing node that receives that prefix will create subnets for it's downstream interfaces based on their EUI

Iface  5  HWaddr: 12:84:0C:87:1F:B7 
          L2-PDU:1500  MTU:1500  HL:64  RTR  
          RTR_ADV  Source address length: 6
          Link type: wired
          inet6 addr: fe80::1084:cff:fe87:1fb7  scope: link  VAL
          inet6 addr: fd12:1284:c87:1fb7:1084:cff:fe87:1fb7  scope: global  VAL
          inet6 group: ff02::2
          inet6 group: ff02::1
          inet6 group: ff02::1:ff87:1fb7
          
Iface  6  HWaddr: 79:3B  Channel: 26  NID: 0x23  PHY: O-QPSK 
          Long HWaddr: A6:C9:C2:7B:CC:D1:79:3B 
           State: IDLE 
          ACK_REQ  L2-PDU:102  MTU:1280  HL:64  RTR  
          6LO  IPHC  
          Source address length: 8
          Link type: wireless
          inet6 addr: fe80::a4c9:c27b:ccd1:793b  scope: link  VAL
          inet6 addr: fd12::a4c9:c27b:ccd1:793b  scope: global  VAL
          inet6 group: ff02::2
          inet6 group: ff02::1
          inet6 group: ff02::1:ffd1:793b

A node on the downstream (wired) network is able to ping the 6LBR:

2024-12-13 14:13:08,806 # Iface  6  HWaddr: E2:F7:1F:FD:29:80 
2024-12-13 14:13:08,807 #           L2-PDU:1500  MTU:1500  HL:64  Source address length: 6
2024-12-13 14:13:08,807 #           Link type: wired
2024-12-13 14:13:08,808 #           inet6 addr: fe80::e0f7:1fff:fefd:2980  scope: link  VAL
2024-12-13 14:13:08,809 #           inet6 addr: fd12:1284:c87:1fb7:e0f7:1fff:fefd:2980  scope: global  VAL
2024-12-13 14:13:08,810 #           inet6 group: ff02::1
2024-12-13 14:13:08,810 #           inet6 group: ff02::1:fffd:2980
2024-12-13 14:13:08,811 #           inet6 group: ff02::1a          

2024-12-13 14:13:16,428 # ping 2001:db8:1::1
2024-12-13 14:13:16,434 # 12 bytes from 2001:db8:1::1: icmp_seq=0 ttl=63 time=5.035 ms
2024-12-13 14:13:17,429 # 12 bytes from 2001:db8:1::1: icmp_seq=1 ttl=63 time=4.978 ms
2024-12-13 14:13:18,432 # 12 bytes from 2001:db8:1::1: icmp_seq=2 ttl=63 time=3.025 ms
2024-12-13 14:13:18,433 # 
2024-12-13 14:13:18,434 # --- 2001:db8:1::1 PING statistics ---
2024-12-13 14:13:18,436 # 3 packets transmitted, 3 packets received, 0% packet loss
2024-12-13 14:13:18,437 # round-trip min/avg/max = 0.978/3.012/5.035 ms

Issues/PRs references

[mguetschow]

I don't have enough knowledge on 6lo to judge the impact and fitness of this approach, but one minor comment below.

[fabian18]

Code looks fine. As auto_subnets is your creation I trust you testing.
You may squash and I will approve

[benpicco]

Actually we can do a little better and preserve the unused bits e.g. when combining a /8 with a EUI-48 from an ethernet interface:

fd00::/8 dev #6 
fdd2:7c7c:6d94:4100::/56 dev #5 

[fabian18]

Ok, I dont't wanna start bikeshedidng now, but since you did some changes with new_prefix_len, Is this something sensible to return from _init_sub_prefix{_eui}?