More segfaults because "inter" is NULL

[PherricOxide]

We need to find the underlying cause of the interface variable being null. Checking for NULL everywhere is just going to cover it up. This is probably the same underlying issue as #33.

Thread [1] 4793 [core: 0] (Suspended : Signal : SIGSEGV:Segmentation fault) 
honeyd_ether_cb() at honeyd.c:560 0x80522c9 
arp_discover() at arp.c:218 0x80784d7   
arp_request() at arp.c:314 0x80788e9    
honeyd_deliver_ethernet() at honeyd.c:587 0x80523b7 
honeyd_delay_cb() at honeyd.c:703 0x80527de 
honeyd_delay_packet() at honeyd.c:835 0x8052de1 
honeyd_ip_send() at honeyd.c:912 0x8053149  
icmp_send() at honeyd.c:1,616 0x8054d18 
icmp_error_send() at honeyd.c:1,651 0x8054f01   
udp_recv_cb() at honeyd.c:2,513 0x8057661   
honeyd_dispatch() at honeyd.c:2,823 0x805848e   
honeyd_delay_cb() at honeyd.c:768 0x8052a6e 
honeyd_delay_packet() at honeyd.c:835 0x8052de1 
honeyd_input() at honeyd.c:3,056 0x8059059  
honeyd_recv_cb() at honeyd.c:3,211 0x80596b4    
0x17fe64    
pcap_dispatch() at 0x182668 
interface_recv() at interface.c:516 0x8077234   
event_base_loop() at 0x13fce9   
event_loop() at 0x140a37    
<...more frames...> 

[PherricOxide]

I believe this was a side effect of using the 'default' profile with a TCP reset action set. Honeyd was attempting to reply to broadcast packets and couldn't figure out what interface to use when trying to send something to 255.255.255.255. Hasn't been seen again since we changed the usage of the default profile.