On Tue, Nov 27, 2001 at 04:16:56PM -0000, David Laight wrote:
> Could depend on the way the ethernet controller (and more probably the PHY)
> handle reflecting tx traffic back to the rx side.
> 
> Traditionally the transciever would generate the 'rx active' signal on the
> AUI cable - but the chiptset would ignore any associated data (lack of
> the signal generates a 'carrier loss' error on the transmit).
> 
> However I've seen one ethernet controller/PHY pair that would receive
> it's own transmittions (provided the destination address matched) -
> but only at 10M, at 100M it didn't!
> 
> Now the general requirment on an ethernet MAC driver is that it generate
> rx events for broadcast/multicast traffic, this usually has to be done
> within the driver code.  But I dunno if netbsd expects this!
> (any driver that conforms to the DLPI - as per the X/Open doc has to
> behave that way.)

Sure NetBSD does it, depend on how the hardware behaves (if the hardware
does it, no need to NetBSD to do it). This is what the IFF_SIMPLEX flag
is for.

--
Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
--