From mid@auk.cx Mon Dec 13 05:43:26 1999
Date: Mon, 13 Dec 1999 05:32:55 +0000 (UTC)
From: Adam Fritzler <mid@auk.cx>
Subject: Re: tm380tr on madge smart 16/4 pci ringnode



The MAC address is stored in the EEPROM.  The board does work just fine
with the tms380tr driver, except for the (fairly minor) MAC address
problem.  I don't know how Madge attached the eeprom (I'm still waiting on
cards from Magde so I can figure this out).  But I do know how to fake it.

In tms380tr_read_addr(), you'll want to comment out the code thats there
and just do this:

static void tms380tr_read_addr(struct net_device *dev, unsigned char *Address)
{
	Address[0] = 0x00; /* replace with your MAC, or this one... */
	Address[1] = 0x55;
	Address[2] = 0x00;
	Address[3] = 0x71;
	Address[4] = 0x2a;
	Address[5] = 0xed;
	return;
}



In tms380tr_init_opb(), you can tell the driver what address to set the
card to (this should really be made into an laa= option like the ibmtr
driver has).  It looks something like this...

/* Add the 'struct net_device *dev' parameter! (to prototype too!) */
static void tms380tr_init_opb(struct net_local *tp, struct net_device *dev)
{
        unsigned long Addr;
        unsigned short RplSize    = RPL_SIZE;
        unsigned short TplSize    = TPL_SIZE;
        unsigned short BufferSize = BUFFER_SIZE;
	int i;	

        tp->ocpl.OPENOptions     = 0;
        tp->ocpl.OPENOptions    |= ENABLE_FULL_DUPLEX_SELECTION;
        tp->ocpl.FullDuplex      = 0;
        tp->ocpl.FullDuplex     |= OPEN_FULL_DUPLEX_OFF;

	/* Set Node Address manually -- ADD THIS PART */
	for (i=0;i<6;i++)
		tp->ocpl.NodeAddr[i] = ((unsigned char *)dev->dev_addr)[i];

        /* Fixme: If mac address setable:
         * for (i=0; i<LENGTH_OF_ADDRESS; i++)
         *      mac->Vam->ocpl.NodeAddr[i] = mac->CurrentAddress[i];
         */

        tp->ocpl.GroupAddr       = 0;
        tp->ocpl.FunctAddr       = 0;

	/* ... */
}

Additionally, change the call in tms380tr_chipset_init() to be
tms380tr_init_opb(tp, dev).  It needs the net_device to get the address
from.  

Change those parts, and you should end up with the address you put in
tms380tr_read_addr(). Should.  I wrote the above without actually testing
anything, so no guarentees.  I know its possible (its been done before).  
I think I will in the next version of the driver (which will have some
relatively major changes), add an laa= module option or something so that
it can be specified.  Of course, I also expect to have a Madge card to
test with by then, so real support of those cards should be there.  

af


---
  Adam Fritzler
  { mid@auk.cx, afritz@iname.com}
    http://www.auk.cx/~mid/
  "You may call me Lor." -- Lor