Make hardware address fallback more robust

[r0man]

The following code returns nil on my Linux machine, because the network
interface for "localhost", the loopback device, doesn't have a hardware
address.

(-> (InetAddress/getLocalHost)
     NetworkInterface/getByInetAddress
    .getHardwareAddress)

This patch uses the same fallback strategy that is used when an
UnknownHostException exception happens.

[maxcountryman]

Thanks!

[r0man]

@maxcountryman Could you also cut a release, please? I came accross this issue while trying to use https://github.com/xsc/kithara, and I thing @xsc would also be interested in this patch. Thanks!

[maxcountryman]

@r0man I think I need to change how we look for hardware addresses. Do you mind telling me if this var is nil on your machine with the loopback interface?

(import '[java.net InetAddress NetworkInterface])
(def hardware-address
  (some->> (NetworkInterface/getNetworkInterfaces)
               enumeration-seq
               (remove #(.isLoopback %))
               (map #(.getHardwareAddress %))
               (filter identity)
               first))

[r0man]

@maxcountryman This returns the hardware address of my docker0 device, which is up but not running.

docker0: flags=4099<UP,BROADCAST,MULTICAST>  mtu 1500
        inet 172.17.0.1  netmask 255.255.0.0  broadcast 0.0.0.0
        ether 02:42:42:61:cb:85  txqueuelen 0  (Ethernet)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10<host>
        loop  txqueuelen 1  (Local Loopback)
        RX packets 16532  bytes 2021646 (1.9 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 16532  bytes 2021646 (1.9 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

wlp4s0: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 192.168.20.105  netmask 255.255.255.0  broadcast 192.168.20.255
        inet6 fe80::baf6:b1ff:fe18:1e3f  prefixlen 64  scopeid 0x20<link>
        ether b8:f6:b1:18:1e:3f  txqueuelen 1000  (Ethernet)
        RX packets 100165  bytes 125363331 (119.5 MiB)
        RX errors 0  dropped 0  overruns 0  frame 997
        TX packets 84754  bytes 8177259 (7.7 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0
        device interrupt 17  

Maybe filter also by isUp? This returns the hardware address of my wifi device wlp4s0.

(def hardware-address
  (some->> (NetworkInterface/getNetworkInterfaces)
               enumeration-seq
               (remove #(.isLoopback %))
               (filter #(.isUp %))
               (map #(.getHardwareAddress %))
               (filter identity)
               first))

[maxcountryman]

@r0man the Erlang implementation doesn't seem concerned with whether the interface is running or not, only that it has a non-null MAC and that the MAC isn't 0.0.0.0.0.0. I think it wouldn't be unreasonable to filter by .isVirtual as well.

But I worry that by adding filters we risk failing to find a valid MAC more frequently than we should. (It's worth pointing out that the Erlang implementation does not fallback to a random array of bytes.)

[maxcountryman]

@r0man what about something like this, which closely mirrors the Erlang implementation?

(def ifs (->> (NetworkInterface/getNetworkInterfaces)
              enumeration-seq
              (map #(.getHardwareAddress %))
              (filter identity)
              (remove #(every? zero? %))))

This will return any interface which has a non-null and non-zeroed MAC address. If we can make a case for applying further filters, as you suggest above, I'm open to that possibility as well provided we can ensure it doesn't mean we hit the SecureRandom fallback more frequently.

[r0man]

@maxcountryman Sounds good. Thanks for looking into this again!