[eepro100] i82559 w/o EPROM.

[Donald Becker]

On Thu, 7 Dec 2000, Bjorn Eriksson wrote:

> Date: Thu, 7 Dec 2000 18:15:59 +0100
> From: Bjorn Eriksson <mdeans@algonet.se>
> To: eepro100@scyld.com
> Subject: Re: [eepro100] i82559 w/o EPROM.
> 
> 
> > >  I'm trying to get the eepro100.c driver running on my embedded
> > > board. This board has been designed without any non-volatile
> > > memory. Now, what shall I do 'bout the MAC adress? Any chance
> > > this will ever work without a serial EPROM?
> 
> > You will have to either add the EEPROM to your design, or modify
> > the driver to hard-wire the transceiver and station address information.
> 
>  I figured I'd look around some first before I hard-wire them adresses this
> my self. Has anyone brought this up on the list before? [I've downloaded and
> searched the archives w/o success.]

The MAC address isn't really the problem.  You can use
  ifconfig ... hw ether 00:11:22:33:44:55

The unavailable issue is that the driver reads the transceiver type and
MII address from the EEPROM, and without this info the driver will not work.

>  I'm a bit surprised to learn there are two addresses associated with each
> MAC.

I'm not certain what you mean by this.  Only 3Com has two slots for MAC
("station") addresses.  The first is the 3Com assigned permanent address.
The second is the station address that is actually used, which is usually
the same as the first.

>  Note: The comments in the source file suggested to me that only I/O access
> is supported (which is fine with me) but I had to #defined USE_IO_OPS to get
> that behaviour.

The i82559 has timing related bugs when used with PCI memory space accesses.
The previous generations seemed to work fine with either I/O or memory
access.

Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993