"Wolfgang S. Rupprecht" <wolfgang+gnus20040319T115736@dailyplanet.dontspam.wsrcc.com> writes:

> gdt@ir.bbn.com (Greg Troxel) writes:
> > Linux has a mechanism that marks an arp entry fresh when a TCP packet
> > is received that acks something, as that validates that the arp entry
> > at least caused the packet to get delivered.
> 
> Thats an interesting idea but it has to be done carefully when one is
> dealing with proxy-arps and wireless access points.  A mobile user
> might have moved out of range of one AP and onto another.  The TCP
> stream would still be following them, but "their" MAC address has
> changed.  (Or perhaps the linux folks are counting on the AP's sending
> gratuitous arps when the users shift from one to the other???)

The MAC address doesn't change when you change APs, just which AP
bridges the Ethernet onto 802.11.

My comment, which was perhaps too terse, was that when you get an ack
for new data, you have evidence that the data you just sent actually
got where it was going (assuming no spoofing).  You don't know that
you sent it to the 'right' address, defined by what you'd get from a
new request.


-- 
        Greg Troxel <gdt@ir.bbn.com>