Quad ethernet card on AXi starts with identical MAC addresses on 5.2.1R

[Marius Strobl]

On Mon, Jul 19, 2004 at 08:41:27PM +0200, Joerg Wunsch wrote:
> As Ken Smith wrote:
> 
> > > > I am just wondering if that behavior is as intended.
> 
> > > Yes, unless you set local-mac-address? to true in OBP.
> 
> > I would be interested in knowing if this works.
> 
> Last time I tried it on FreeBSD, it didn't work.
> 
> > The pathway I followed seemed to dig the MAC address out of the
> > ID-PROM no matter what.  In particular hme_pci_attach() seems to
> > call OF_getetheraddr() which in turn seems to blindly dig the MAC
> > address out of the idprom.
> 
> I've also got that impression.
> 
> I'd also be interested in a way to dig out the MAC address of the card
> (which must be possible, somehow -- after all, there are four address
> labels on the NVRAM [or whatever it is] on the card itself), so it
> would be possible to e. g. use a QFE card (which once has been a huge
> investment) on a non-sparc64 FreeBSD system.  It seems that right now,
> the only dependency of the hme driver on sparc64 is the call to OFW to
> obtain the MAC address...
> 

That's correct but obtaining these addresses on platforms not having
Open Firmware isn't very nice. The problem is that the hme device is
function 1 of the PCI chip but the FCode containing the MAC address
is part of the EBus bridge which is function 0. So one would have to
either attach an own driver to the EBus bridge on platforms without
Open Firmware which is (only) used to obtain the MAC address or fake
bus tags etc. for function 0 in the hme driver. Reading the MAC
address from the FCode also has to be done manually.