James R. Leu wrote:
> Between you and Ben Greear the linux kernel will have every possible
> scheme for sending packets to your self.
> 
> I still think my work on this (Virtual routing and forwarding:
> http://linux-vrf.sf.net/) is the less perverted(*) then the work that either
> you or Ben have come up with.  Plus it has other applications besides
> just being able to send packets to your self.
> 
> * in terms of the concept, not necessarily the actual implementation.

>>It'd be nice to have the module hardwire the arps itself, but this was
>>quickest.  Patch welcome.

It's likely that with my patch you wouldn't have to hard-wire arps at
all.  The primary thing that my patch does is to let a machine answer
arps from a local interface (over the external interface).

Then routing to self can happen by simply(?) binding to the local IP
of your choice and using policy-based routing to route correctly.
(You can loop-back through a router with this patch, for example.)

So, maybe both patches are useful together....

I can't find where I posted my patch last time, so it is
attached again for reference.  It contains a typo-fix in a comment
that may be worthy of inclusion by itself some day :)
Also, when nettool (ethtool) becomes generic, the ioctl code can be
configured through the nettool api, so that new ioctl will go a way.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear