ARP behavior in FreeBSD vs Linux

[Sam Leffler]

Pieter de Boer wrote:
> Chuck Swiger wrote:
> 
>>> In contrast, on Linux (by default), it
>>> responds as long as the target IP address in ARP Request matches with
>>> any "local" IP address on the system, which is not necessarily an IP
>>> address assigned to the interface through which the ARP request is
>>> received.
>>
>> This sounds like "proxy ARPing" is enabled by default on your 
>> particular flavor of Linux.  I don't think they all do that, 
>> hopefully, any more than ipforwarding should be enabled by default 
>> just because a machine has two NICs.
> 
> What Motonori Shindo described is actually the default behaviour for 
> Linux kernels (at least my 2.6.8-kernel does it by default). It could be 
> seen as a sort of proxy-arp, but only for the host itself, not other 
> systems. Let me try to describe when it happens. Say you have 
> 192.168.42.42 bound on eth0 and have eth1 connected to some ethernet 
> LAN. When a host on that eth1-connected LAN sends an 'arp who-has 
> 192.168.42.42', a Linux system will answer that arp-request with it's 
> eth1 MAC-address, although the IP-address is bound on eth0 and the arp 
> request comes in on eth0. FreeBSD obviously doesn't do this.
> 
>>> Is there any advantage/disadvantage in ARP implementation on FreeBSD
>>> over that of Linux? Thanks.
> 
> I was unhappily surprised by this 'feature'. I find it pretty 
> counter-intuitive. I expect two interfaces to be seperated inside a 
> kernel, but Linux more or less binds them together. Incoming traffic on 
> the 'wrong' interface will gladly be accepted, too. This broke things 
> for me, because I didn't want to have that certain IP-address accessible.
> 
> That said, this happens only when you have two interfaces connected to 
> the same subnet, which is a bit evil anyhow. It may be beneficial for 
> Linux to do things this way, perhaps for redundancy-purposes (two 
> interfaces, one IP-address, IP reachable over both interfaces, when one 
> fails, the other takes over.. no idea if that works out-of-the-box).
> 

The linux design philosophy, based on postings from various 
implementors, is that ip addresses are bound to a host, not to a 
particular interface.  I believe the arp behaviour reflects this.

	Sam