Source MAC address per IP packet

Albrecht,

For what practical, benevolent purpose would you ever need to do this!?

hat trick

What about replaying network traces to test (possibly modified) packet filter rules? What about generating specifically formatted packets to test such rules where they try to match MAC addresses (whatever good that would do)? What about testing layer 2 switch MAC table overflow characteristics?

best regards Patrick

Aside from all the others suggested, the most obvious would be if you are implementing a transparent layer 2 switch/bridge.

DS

On 11.08.2008 15:44 David Schwartz wrote

s/transparent/intransparent/ ? Otherwise I don't get what you are pointing at.

Arnold

As far as I know, in some the whole packet is generated in memory (or in NIC memory) and then sent, in others the source address is stuffed in as it is transmitted.

You might look at older ones from the days when hardware was more expensive. (Eight bit ISA days.)

You should be working at the hardware level, though, not while running a networking OS. (DOS makes it easy, especially if you never return.)

Otherwise, just write the source address register before sending each packet. (Again, working directly with the hardware registers, not through a networking OS.

You want the data sheet for the individual controller chip, not the whole card.

-- glen

Am Sun, 10 Aug 2008 10:51:49 -0700 schrieb hat-trick:

What about /usr/src/linux/if_ether.h ? (struct ethhdr)

You can set the MAC address to what ever you want.

cheers