kern/95532: if_gif has artificial limitations

[Daniel O'Connor]

>Number:         95532
>Category:       kern
>Synopsis:       if_gif has artificial limitations
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Sun Apr 09 07:00:30 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     Daniel O'Connor
>Release:        FreeBSD 4.9-STABLE i386
>Organization:
>Environment:
System: FreeBSD metatron.gsoft.com.au 4.9-STABLE FreeBSD 4.9-STABLE #3: Fri Nov 28 12:06:51 CST 2003 doconnor at cain.gsoft.com.au:/usr/obj/usr/src/sys/METATRON i386


>Description:
I have ADSL through the ISP Internode [Australia] and it uses a bridged ethernet configuration - this is pretty standard, however we also have a class C routed to our connection and that is done using an IP over IP tunnel (ie gif).

The thing is that if I configure gif like so..
gifconfig gif0 150.101.23.134 203.16.215.227
ifconfig gif0 up mtu 1452

and do
route add default -iface gif0

all the machines on our LAN except the gateway itself work as expected. Howver the gateway doesn't work as when it makes outbound connections that aren't explicitly bound to an IP they go out with a from address of 0.0.0.0 which the ISP just drops (I presume).

I can configure gif0 like so..
ifconfig gif0 203.31.81.9/32 192.168.254.123 mtu 1452 up

(203.31.81.0/24 is our class C, and 203.31.81.9 is a spare IP - the routers normal IP is 203.31.81.1) everything works from including the gateway.

The 192.168.254.123 is a bogus address I had to use because gif won't accept only one address.. ie I would think this would be more correct..
ifconfig gif0 203.31.81.9/32 mtu 1452 up

Having a work around is OK but it would be nice to know if it's possible to fix (since I imagine the restriction on point to point interfaces is done in the network stack), if there are any other [better] work arounds, and if it is still a problem in 6.0.

Thanks

>How-To-Repeat:
>Fix:

	


>Release-Note:
>Audit-Trail:
>Unformatted: