Re: Broadcast traffic on vlans leaks into the parent interface on NetBSD-5.1

[Robert Elz <kre%munnari.OZ.AU@localhost>]

    Date:        Fri, 7 Dec 2012 17:48:12 -0800
    From:        jnemeth%victoria.tc.ca@localhost (John Nemeth)
    Message-ID:  <201212080148.qB81mC7k004269%vtn1.victoria.tc.ca@localhost>

I am not sure why we're having this discussion, but ...

  | } claim it yet, but that doesn't mean that the server cannot make use of it.

  |      No, it can't.  Obtaining an address is a four step process.

Yes, I know.

  | The first step, the client broadcasts a DISCOVER request.  The second step,
  | all servers that receive the message respond with an OFFER.

Yes, but in order to offer an address, they have to make a temporary
assignment of the address to the client.  (For example, during this
period, the same address cannot be offered to a different client.)
The client doesn't know it has been done, but it has.  Depending on
the servers, and their sync methods, and the nature of the address
being assigned, they may all offer the same address, or all offer
different ones, or some combination of that if there are more than
2 servers, but none of that makes any difference.  If an address has
been offered, then it has been assigned to the client for at least a
few seconds (until the client refuses an offer by accepting a different
one, or fails to accept anything).

In that period, the server is able (if its implementation permits it
to do it successfully) to send to the temporarily assigned address if
it chooses to.

I don't think there's much point in continuing this discussion further,
it has drifted way off the original point (which is either bugs in the wm
driver, or the appropriateness of using BPF at all for this purpose).
You can believe me, or not, as you choose - of if you like, you can ask
Ted or Ralph, or one of the other DHCP big names if you want further
reassurance.

kre