Re: Ethernets on SCC2 and SCC3 on mpc8xx

["Joakim Tjernlund" <Joakim.Tjernlund@lumentis.se>]

[-- Attachment #1: Type: text/plain, Size: 323 bytes --]

Hi Jan

I have a version that supports multiple enet's in one driver.
It is a bit specific for our board and contains a lot more:
- run time settable MAC and MTU.
- optimized RX interrupt handling(up to 20% better throuput)
- improved handling of received faulty frames.

    I am attaching the driver. Good luck.

   Jocke

[-- Attachment #2: enet.c --]
[-- Type: application/octet-stream, Size: 34299 bytes --]

/*
 * Ethernet driver for Motorola MPC8xx.
 * Copyright (c) 1997 Dan Malek (dmalek@jlc.net)
 *
 * I copied the basic skeleton from the lance driver, because I did not
 * know how to write the Linux driver, but I did know how the LANCE worked.
 *
 * This version of the driver is somewhat selectable for the different
 * processor/board combinations.  It works for the boards I know about
 * now, and should be easily modified to include others.  Some of the
 * configuration information is contained in "commproc.h" and the
 * remainder is here.
 *
 * Buffer descriptors are kept in the CPM dual port RAM, and the frame
 * buffers are in the host memory.
 *
 * Right now, I am very watseful with the buffers.  I allocate memory
 * pages and then divide them into 2K frame buffers.  This way I know I
 * have buffers large enough to hold one frame within one buffer descriptor.
 * Once I get this working, I will use 64 or 128 byte CPM buffers, which
 * will be much more memory efficient and will easily handle lots of
 * small packets.
 *
 */
#include <linux/config.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/string.h>
#include <linux/ptrace.h>
#include <linux/errno.h>
#include <linux/ioport.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/skbuff.h>
#include <linux/spinlock.h>

#include <asm/8xx_immap.h>
#include <asm/pgtable.h>
#include <asm/mpc8xx.h>
#include <asm/bitops.h>
#include <asm/uaccess.h>

#include <asm/commproc.h>

/*
 *				Theory of Operation
 *
 * The MPC8xx CPM performs the Ethernet processing on SCC1.  It can use
 * an aribtrary number of buffers on byte boundaries, but must have at
 * least two receive buffers to prevent constant overrun conditions.
 *
 * The buffer descriptors are allocated from the CPM dual port memory
 * with the data buffers allocated from host memory, just like all other
 * serial communication protocols.  The host memory buffers are allocated
 * from the free page pool, and then divided into smaller receive and
 * transmit buffers.  The size of the buffers should be a power of two,
 * since that nicely divides the page.  This creates a ring buffer
 * structure similar to the LANCE and other controllers.
 *
 * Like the LANCE driver:
 * The driver runs as two independent, single-threaded flows of control.  One
 * is the send-packet routine, which enforces single-threaded use by the
 * cep->tx_busy flag.  The other thread is the interrupt handler, which is
 * single threaded by the hardware and other software.
 *
 * The send packet thread has partial control over the Tx ring and the
 * 'cep->tx_busy' flag.  It sets the tx_busy flag whenever it's queuing a Tx
 * packet. If the next queue slot is empty, it clears the tx_busy flag when
 * finished otherwise it sets the 'lp->tx_full' flag.
 *
 * The MBX has a control register external to the MPC8xx that has some
 * control of the Ethernet interface.  Information is in the manual for
 * your board.
 *
 * The RPX boards have an external control/status register.  Consult the
 * programming documents for details unique to your board.
 *
 * For the TQM8xx(L) modules, there is no control register interface.
 * All functions are directly controlled using I/O pins.  See commproc.h.
 */


/* The transmitter timeout
 */
#define TX_TIMEOUT	(1*HZ)

/* Define COPY_SMALL_FRAMES if you want to save buffer memory for small packets 
 * at a small performance hit */
#ifdef COPY_SMALL_FRAMES
  #define RX_COPYBREAK (256-16+1) /* dev_alloc_skb() adds 16 bytes for internal use */
#endif
/* The number of Tx and Rx buffers.  These are allocated from the page
 * pool.  The code may assume these are power of two, so it is best
 * to keep them that size.
 * We don't need to allocate pages for the transmitter.  We just use
 * the skbuffer directly.
 */

#ifdef CONFIG_ENET_BIG_BUFFERS
  #define RX_RING_SIZE		64
  #define TX_RING_SIZE		64	/* Must be power of two for this to work */
#else
  #define RX_RING_SIZE		16
  #define TX_RING_SIZE		16	/* Must be power of two for this to work */
#endif
#define TX_RING_MOD_MASK	(TX_RING_SIZE-1)

#define CPM_ENET_RX_FRSIZE	1552 /* must be a multiple of cache line */
#if CPM_ENET_RX_FRSIZE % L1_CACHE_LINE_SIZE != 0
    #error CPM_ENET_RX_FRSIZE must be a multiple of L1 cache size
#endif


/* The CPM stores dest/src/type, data, and checksum for receive packets.
 */
#define MAX_MTU 1514  /* this is data size */
#define MIN_MTU 46    /* this is data size */
#define SCC_MAX_MTU   (((CPM_ENET_RX_FRSIZE-ETH_HLEN-8)/4)*4) /* must be a multiple of 4 */

#define PKT_MAXBUF_SIZE   (MAX_MTU+ETH_HLEN+4) /* allow fullsized pppoe packets and VLAN */
#define PKT_MINBUF_SIZE   (MIN_MTU+ETH_HLEN+4)
#define PKT_MAXBLR_SIZE	  SCC_MAX_MTU


/* The CPM buffer descriptors track the ring buffers.  The rx_bd_base and
 * tx_bd_base always point to the base of the buffer descriptors.  The
 * cur_rx and cur_tx point to the currently available buffer.
 * The dirty_tx tracks the current buffer that is being sent by the
 * controller.  The cur_tx and dirty_tx are equal under both completely
 * empty and completely full conditions.  The empty/ready indicator in
 * the buffer descriptor determines the actual condition.
 */
struct scc_enet_private {
	/* The saved address of a sent-in-place packet/buffer, for skfree(). */
	struct	sk_buff* tx_skbuff[TX_RING_SIZE];
	ushort	skb_cur;
	ushort	skb_dirty;

	/* CPM dual port RAM relative addresses.
	*/
	cbd_t	*rx_bd_base;		/* Address of Rx and Tx buffers. */
	cbd_t	*tx_bd_base;
	cbd_t	*cur_rx, *cur_tx;		/* The next free ring entry */
	cbd_t	*dirty_tx;	/* The ring entries to be free()ed. */
	scc_t	*sccp;

	/* Virtual addresses for the receive buffers because we can't
	 * do a __va() on them anymore.
	 */
	void    *rx_vaddr[RX_RING_SIZE];
	struct	net_device_stats stats;
	uint	tx_full;
        spinlock_t lock;
        char    scc_num;
        int     cpm_cr_ch;
};

int __init scc_enet_init_params(int scc_num, int cpmvec, int proff, int cpm_cr_ch);
int __init cpm_enet_init_scc_pins(int scc_num);
int __init cpm_enet_init_scc_clock(int scc_num);

static int scc_enet_open(struct net_device *dev);
static int scc_enet_start_xmit(struct sk_buff *skb, struct net_device *dev);
static int scc_enet_rx(struct net_device *dev);
static void scc_enet_interrupt(void *dev_id, struct pt_regs *regs);
static int scc_enet_close(struct net_device *dev);
static struct net_device_stats *scc_enet_get_stats(struct net_device *dev);
static void set_multicast_list(struct net_device *dev);
static void cpm_enet_enable_xmit(int);

/*
 * Called from net/core/dev.c: int dev_open(struct net_device *dev)
 *
 * If CONFIG_LUMENTISCP defined: it will be called twice, once for
 * 'eth0' and once for 'eth1'.
 *
 * If CONFIG_LUMENTISUP defined: it will be called just once.
 * I.e. for'eth0'.
 */

static int
scc_enet_open(struct net_device *dev)
{
	struct	scc_enet_private *cep = (struct scc_enet_private *)dev->priv;

	/* I should reset the ring buffers here, but I don't yet know
	 * a simple way to do that.
	 */
	netif_start_queue(dev);

	/* enable the transmit and receive processing.
	*/
	cep->sccp->scc_gsmrl |= (SCC_GSMRL_ENR | SCC_GSMRL_ENT);

	return 0;					/* Always succeed */
}

static int
scc_enet_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct scc_enet_private *cep = (struct scc_enet_private *)dev->priv;
	volatile cbd_t	*bdp;
	volatile  scc_enet_t *ep = (scc_enet_t *)dev->base_addr;
	/* Fill in a Tx ring entry */
	bdp = cep->cur_tx;

	if (skb->len > (ep->sen_maxflr - 4)){
		printk("eth: TX frame is too long");
	}

	/* Clear all of the status flags.
	 */
	bdp->cbd_sc &= ~BD_ENET_TX_STATS;

	/* If the frame is short, tell CPM to pad it.
	if (skb->len <= ETH_ZLEN)
		bdp->cbd_sc |= BD_ENET_TX_PAD;
	else
		bdp->cbd_sc &= ~BD_ENET_TX_PAD;
	*/

	/* Set buffer length and buffer pointer.
	*/
	bdp->cbd_datlen = skb->len;
	bdp->cbd_bufaddr = __pa(skb->data);

	/* Save skb pointer.
	*/
	cep->tx_skbuff[cep->skb_cur] = skb;

	cep->stats.tx_bytes += skb->len;
	cep->skb_cur = (cep->skb_cur+1) & TX_RING_MOD_MASK;
	
	/* Push the data cache so the CPM does not get stale memory
	 * data.
	 */
	flush_dcache_range((unsigned long)(skb->data),
					(unsigned long)(skb->data + skb->len));

	spin_lock_irq(&cep->lock);

	/* Send it on its way.  Tell CPM its ready, interrupt when done,
	 * its the last BD of the frame, and to put the CRC on the end.
	 */
	bdp->cbd_sc |= (BD_ENET_TX_READY | BD_ENET_TX_INTR | BD_ENET_TX_LAST | BD_ENET_TX_TC | BD_ENET_TX_PAD);

	dev->trans_start = jiffies;

	/* If this was the last BD in the ring, start at the beginning again.
	*/
	if (bdp->cbd_sc & BD_ENET_TX_WRAP)
		bdp = cep->tx_bd_base;
	else
		bdp++;

	if (bdp->cbd_sc & BD_ENET_TX_READY) {
		netif_stop_queue(dev);
		cep->tx_full = 1;
	}

	cep->cur_tx = (cbd_t *)bdp;

	spin_unlock_irq(&cep->lock);

	return 0;
}

static void
scc_enet_timeout(struct net_device *dev)
{
	struct scc_enet_private *cep = (struct scc_enet_private *)dev->priv;

	printk("%s: transmit timed out.\n", dev->name);
	cep->stats.tx_errors++;
#ifndef final_version
	{
		int	i;
		cbd_t	*bdp;
		printk(" Ring data dump: cur_tx %p%s cur_rx %p.\n",
		       cep->cur_tx, cep->tx_full ? " (full)" : "",
		       cep->cur_rx);
		bdp = cep->tx_bd_base;
		for (i = 0 ; i < TX_RING_SIZE; i++, bdp++)
			printk("%04x %04x %08x\n",
			       bdp->cbd_sc,
			       bdp->cbd_datlen,
			       bdp->cbd_bufaddr);
		bdp = cep->rx_bd_base;
		for (i = 0 ; i < RX_RING_SIZE; i++, bdp++)
			printk("%04x %04x %08x\n",
			       bdp->cbd_sc,
			       bdp->cbd_datlen,
			       bdp->cbd_bufaddr);
	}
#endif
	if (!cep->tx_full)
		netif_wake_queue(dev);
}

/* The interrupt handler.
 * This is called from the CPM handler, not the MPC core interrupt.
 */
static void
scc_enet_interrupt(void *dev_id,  struct pt_regs *regs)
{
	struct	net_device *dev = dev_id;
	volatile struct	scc_enet_private *cep;
	volatile cbd_t	*bdp;
	ushort	int_events;
	int	must_restart;

	cep = (struct scc_enet_private *)dev->priv;

	/* Get the interrupt events that caused us to be here.
	*/
	int_events = cep->sccp->scc_scce;
	cep->sccp->scc_scce = int_events;
	must_restart = 0;

	/* Handle receive event in its own function.
	*/
	if (int_events & SCCE_ENET_RXF)
		scc_enet_rx(dev_id);

	/* Check for a transmit error.  The manual is a little unclear
	 * about this, so the debug code until I get it figured out.  It
	 * appears that if TXE is set, then TXB is not set.  However,
	 * if carrier sense is lost during frame transmission, the TXE
	 * bit is set, "and continues the buffer transmission normally."
	 * I don't know if "normally" implies TXB is set when the buffer
	 * descriptor is closed.....trial and error :-).
	 */

	/* Transmit OK, or non-fatal error.  Update the buffer descriptors.
	*/
	if (int_events & (SCCE_ENET_TXE | SCCE_ENET_TXB)) {
	    spin_lock(&cep->lock);
	    bdp = cep->dirty_tx;
	    while ((bdp->cbd_sc&BD_ENET_TX_READY)==0) {
		if ((bdp==cep->cur_tx) && (cep->tx_full == 0))
		    break;
		if (bdp->cbd_sc & (BD_ENET_TX_HB | BD_ENET_TX_LC | BD_ENET_TX_RL | BD_ENET_TX_UN | BD_ENET_TX_CSL | BD_ENET_TX_RCMASK)) {
			if (bdp->cbd_sc & BD_ENET_TX_HB)	/* No heartbeat */
				cep->stats.tx_heartbeat_errors++;
			if (bdp->cbd_sc & BD_ENET_TX_LC)	/* Late collision */
				cep->stats.tx_window_errors++;
			if (bdp->cbd_sc & BD_ENET_TX_RL)	/* Retrans limit */
				cep->stats.tx_aborted_errors++;
			if (bdp->cbd_sc & BD_ENET_TX_UN)	/* Underrun */
				cep->stats.tx_fifo_errors++;
			if (bdp->cbd_sc & BD_ENET_TX_CSL)	/* Carrier lost */
				cep->stats.tx_carrier_errors++;
			
			/* Check retry counter, i.e. collision counter */
			if (bdp->cbd_sc & BD_ENET_TX_RCMASK){
				/* Note that counter cannot go higher than 15 */
				cep->stats.collisions+=(bdp->cbd_sc & BD_ENET_TX_RCMASK)>>2;
			}
			
			/* No heartbeat or Lost carrier are not really bad errors.
			 * The others require a restart transmit command.
			 */
			if (bdp->cbd_sc &
			    (BD_ENET_TX_LC | BD_ENET_TX_RL | BD_ENET_TX_UN)) {
				must_restart = 1;
				cep->stats.tx_errors++;
			}
		}

		cep->stats.tx_packets++;

		/* Free the sk buffer associated with this last transmit.
		*/
		dev_kfree_skb_irq(cep->tx_skbuff[cep->skb_dirty]);
		cep->skb_dirty = (cep->skb_dirty + 1) & TX_RING_MOD_MASK;

		/* Update pointer to next buffer descriptor to be transmitted.
		*/
		if (bdp->cbd_sc & BD_ENET_TX_WRAP)
			bdp = cep->tx_bd_base;
		else
			bdp++;

		/* I don't know if we can be held off from processing these
		 * interrupts for more than one frame time.  I really hope
		 * not.  In such a case, we would now want to check the
		 * currently available BD (cur_tx) and determine if any
		 * buffers between the dirty_tx and cur_tx have also been
		 * sent.  We would want to process anything in between that
		 * does not have BD_ENET_TX_READY set.
		 */

		/* Since we have freed up a buffer, the ring is no longer
		 * full.
		 */
		if (cep->tx_full) {
			cep->tx_full = 0;
			if (netif_queue_stopped(dev))
				netif_wake_queue(dev);
		}

		cep->dirty_tx = (cbd_t *)bdp;
	    }

	    if (must_restart) {
		volatile cpm8xx_t *cp;

		/* Some transmit errors cause the transmitter to shut
		 * down.  We now issue a restart transmit.  Since the
		 * errors close the BD and update the pointers, the restart
		 * _should_ pick up without having to reset any of our
		 * pointers either.
		 */
		cp = cpmp;
		cp->cp_cpcr =
                    mk_cr_cmd(cep->cpm_cr_ch, CPM_CR_RESTART_TX) | CPM_CR_FLG;
		while (cp->cp_cpcr & CPM_CR_FLG);
	    }
	    spin_unlock(&cep->lock);
	}

	/* Check for receive busy, i.e. packets coming but no place to
	 * put them.  This "can't happen" because the receive interrupt
	 * is tossing previous frames.
	 */
	if (int_events & SCCE_ENET_BSY) {
		cep->stats.rx_dropped++;
		printk("CPM ENET ethernet: BSY can't happen.\n");
	}

	return;
}

/* During a receive, the cur_rx points to the current incoming buffer.
 * When we update through the ring, if the next incoming buffer has
 * not been given to the system, we just set the empty indicator,
 * effectively tossing the packet.
 */
static int
scc_enet_rx(struct net_device *dev)
{
	struct	scc_enet_private *cep;
	volatile cbd_t	*bdp;
	struct	sk_buff *skb, *skb_tmp;
	ushort	pkt_len;

	cep = (struct scc_enet_private *)dev->priv;

	/* First, grab all of the stats for the incoming packet.
	 * These get messed up if we get called due to a busy condition.
	 */
	bdp = cep->cur_rx;

	for (;;) {
		if (bdp->cbd_sc & BD_ENET_RX_EMPTY)
			break;
#define RX_BD_ERRORS (BD_ENET_RX_LG | BD_ENET_RX_SH | BD_ENET_RX_NO | BD_ENET_RX_CR | BD_ENET_RX_OV | BD_ENET_RX_CL)
		if(bdp->cbd_sc & RX_BD_ERRORS){ /* Receive errors ? */
			cep->stats.rx_errors++;
			if (bdp->cbd_sc & (BD_ENET_RX_LG | BD_ENET_RX_SH)) /* Frame too long or too short. */
				cep->stats.rx_length_errors++;
			if (bdp->cbd_sc & BD_ENET_RX_NO)	/* Frame alignment */
				cep->stats.rx_frame_errors++;
			if (bdp->cbd_sc & BD_ENET_RX_CR)	/* CRC Error */
				cep->stats.rx_crc_errors++;
			if (bdp->cbd_sc & BD_ENET_RX_OV)	/* FIFO overrun */
				cep->stats.rx_fifo_errors++;
			if (bdp->cbd_sc & BD_ENET_RX_CL)        /* Late collision */
				cep->stats.collisions++;
		} else {
			
			/* Process the incoming frame.
			 */
			cep->stats.rx_packets++;
			pkt_len = bdp->cbd_datlen;
			cep->stats.rx_bytes += pkt_len;
			pkt_len -= 4; /* The packet length includes FCS, but we don't want to
				       * include that when passing upstream as it messes up
				       * bridging applications. */
#ifdef COPY_SMALL_FRAMES 
			/* Allocate the next buffer now so we are sure to have one when needed
			 * This does 16 byte alignment, exactly what we need(L1_CACHE aligned). */
			if(pkt_len < RX_COPYBREAK){
				skb_tmp = __dev_alloc_skb(pkt_len + 2, GFP_ATOMIC | GFP_DMA);
				skb_reserve(skb_tmp, 2); /* make the IP header L1 cache aligned */
			} else
#endif
				skb_tmp = __dev_alloc_skb(CPM_ENET_RX_FRSIZE, GFP_ATOMIC | GFP_DMA);
			
			if (skb_tmp == NULL) {
				printk("%s: Memory squeeze, dropping packet.\n", dev->name);
				cep->stats.rx_dropped++;

			} else {
				skb = cep->rx_vaddr[bdp - cep->rx_bd_base];
#ifdef COPY_SMALL_FRAMES 
				if(pkt_len < RX_COPYBREAK) {
					typeof(skb) skb_swap = skb;
					memcpy(skb_put(skb_tmp, pkt_len), skb->data, pkt_len);
					/* swap the skb and skb_tmp */
					skb = skb_tmp;
					skb_tmp = skb_swap;
				} 
				else 
#endif
				{
					skb_put(skb, pkt_len);  /* Make room */
					bdp->cbd_bufaddr = __pa(skb_tmp->data);
					cep->rx_vaddr[bdp - cep->rx_bd_base] = skb_tmp;
				}
				invalidate_dcache_range((unsigned long) skb_tmp->data,
							(unsigned long) skb_tmp->data + CPM_ENET_RX_FRSIZE);
				skb->dev = dev;
				skb->protocol=eth_type_trans(skb, dev);
				netif_rx(skb);
			}
		}
		
		/* Clear the status flags for this buffer.
		 */
		bdp->cbd_sc &= ~BD_ENET_RX_STATS;
		
		/* Mark the buffer empty.
		 */
		bdp->cbd_sc |= BD_ENET_RX_EMPTY;
		
		/* Update BD pointer to next entry.
		 */
		if (bdp->cbd_sc & BD_ENET_RX_WRAP)
			bdp = cep->rx_bd_base;
		else
			bdp++;
		
	}
	cep->cur_rx = (cbd_t *)bdp;

	return 0;
}

static int
scc_enet_close(struct net_device *dev)
{
	struct	scc_enet_private *cep = (struct scc_enet_private *)dev->priv;
	/* Don't know what to do yet.
	*/
	cep->sccp->scc_gsmrl &= ~(SCC_GSMRL_ENR | SCC_GSMRL_ENT);
	netif_stop_queue(dev);

	return 0;
}

static struct net_device_stats *scc_enet_get_stats(struct net_device *dev)
{
	struct scc_enet_private *cep = (struct scc_enet_private *)dev->priv;

	return &cep->stats;
}

static int change_mtu(struct net_device *dev, int new_mtu)
{
	volatile scc_enet_t *ep;

	if ((new_mtu < MIN_MTU ) || (new_mtu >  MAX_MTU))
	       return -EINVAL;
	ep = (scc_enet_t *)dev->base_addr;
	
	ep->sen_maxflr = new_mtu + ETH_HLEN + 4;   /* maximum frame length register */
	dev->mtu = new_mtu;
	printk("ENET: change_mtu: %d\n",new_mtu);
	return 0;
}

static int set_mac_address(struct net_device *dev, void *addr)
{
	volatile scc_enet_t *ep;
	unsigned char	    *eap;
	struct sockaddr     *mac = addr;
	int	i;

	if(netif_running(dev))
	       return -EBUSY;
	
	/* Get pointer to SCC area in parameter RAM. */
	ep = (scc_enet_t *)dev->base_addr;
	eap = (unsigned char *)&(ep->sen_paddrh);
	/* cli();  */
	for (i=5; i>=0; i--)
		*eap++ = dev->dev_addr[i] = mac->sa_data[i];
	/* sti(); */
	return 0;
}

/* Set or clear the multicast filter for this adaptor.
 * Skeleton taken from sunlance driver.
 * The CPM Ethernet implementation allows Multicast as well as individual
 * MAC address filtering.  Some of the drivers check to make sure it is
 * a group multicast address, and discard those that are not.  I guess I
 * will do the same for now, but just remove the test if you want
 * individual filtering as well (do the upper net layers want or support
 * this kind of feature?).
 */

static void set_multicast_list(struct net_device *dev)
{
	struct	scc_enet_private *cep;
	struct	dev_mc_list *dmi;
	u_char	*mcptr, *tdptr;
	volatile scc_enet_t *ep;
	int	i, j;
	cep = (struct scc_enet_private *)dev->priv;

	/* Get pointer to SCC area in parameter RAM.
	*/
	ep = (scc_enet_t *)dev->base_addr;

	if (dev->flags&IFF_PROMISC) {
		cep->sccp->scc_pmsr |= SCC_PMSR_PRO;
	} else {

		cep->sccp->scc_pmsr &= ~SCC_PMSR_PRO;

		if (dev->flags & IFF_ALLMULTI) {
			/* Catch all multicast addresses, so set the
			 * filter to all 1's.
			 */
			ep->sen_gaddr1 = 0xffff;
			ep->sen_gaddr2 = 0xffff;
			ep->sen_gaddr3 = 0xffff;
			ep->sen_gaddr4 = 0xffff;
		}
		else {
			/* Clear filter and add the addresses in the list.
			*/
			ep->sen_gaddr1 = 0;
			ep->sen_gaddr2 = 0;
			ep->sen_gaddr3 = 0;
			ep->sen_gaddr4 = 0;

			dmi = dev->mc_list;

			for (i=0; i<dev->mc_count; i++, dmi = dmi->next) {
				
				/* Only support group multicast for now.
				*/
				if (!(dmi->dmi_addr[0] & 1))
					continue;

				/* The address in dmi_addr is LSB first,
				 * and taddr is MSB first.  We have to
				 * copy bytes MSB first from dmi_addr.
				 */
				mcptr = (u_char *)dmi->dmi_addr + 5;
				tdptr = (u_char *)&ep->sen_taddrh;
				for (j=0; j<6; j++)
					*tdptr++ = *mcptr--;

				/* Ask CPM to run CRC and set bit in
				 * filter mask.
				 */
                                cpmp->cp_cpcr = mk_cr_cmd(cep->cpm_cr_ch, CPM_CR_SET_GADDR) | CPM_CR_FLG;
				/* this delay is necessary here -- Cort */
				udelay(10);
				while (cpmp->cp_cpcr & CPM_CR_FLG);

			}
		}
	}
}

int __init scc_enet_init(void)
{
	bd_t *bd = (bd_t *)__res;
	/*
	 * Always initialize eth0 on SCC1.
	 */
	scc_enet_init_params(0, CPMVEC_SCC1, PROFF_SCC1, CPM_CR_CH_SCC1);
	
	if(bd->board_type ==  CU_BOARDTYPE || bd->board_type ==  CU2_BOARDTYPE){
		/*
		 * On a CP board initialize eth1 on SCC2.
		 */
		scc_enet_init_params(1, CPMVEC_SCC2, PROFF_SCC2, CPM_CR_CH_SCC2);
		if(bd->board_type ==  CU2_BOARDTYPE){
			scc_enet_init_params(3, CPMVEC_SCC4, PROFF_SCC4, CPM_CR_CH_SCC4);
		}
	}
	return 0;
}


/* Initialize the CPM Ethernet on SCC.  If EPPC-Bug loaded us, or performed
 * some other network I/O, a whole bunch of this has already been set up.
 * It is no big deal if we do it again, we just have to disable the
 * transmit and receive to make sure we don't catch the CPM with some
 * inconsistent control information.
 */
int __init scc_enet_init_params(int scc_num, int cpmvec, int proff, int cpm_cr_ch)
{
	struct net_device *dev;
	struct scc_enet_private *cep;
	int i, j, k;
	unsigned char	*eap;
	bd_t		*bd;
	volatile	cbd_t		*bdp;
	volatile	cpm8xx_t	*cp;
	volatile	scc_t		*sccp;
	volatile	scc_enet_t	*ep;
	volatile	immap_t		*immap;

	cp = cpmp;	/* Get pointer to Communication Processor */

	immap = (immap_t *)(mfspr(IMMR) & 0xFFFF0000);	/* and to internal registers */

	bd = (bd_t *)__res;

	/* Allocate some private information.
	*/
	cep = (struct scc_enet_private *)kmalloc(sizeof(*cep), GFP_KERNEL);
	if (cep == NULL)
		return -ENOMEM;

	__clear_user(cep,sizeof(*cep));
	spin_lock_init(&cep->lock);

	/* Create an Ethernet device instance.
	*/
	dev = init_etherdev(0, 0);

	/* Get pointer to SCC area in parameter RAM.
	*/
        ep = (scc_enet_t *)(&cp->cp_dparam[proff]);

	/* And another to the SCC register area.
	*/
        sccp = (volatile scc_t *)(&cp->cp_scc[scc_num]);
	cep->sccp = (scc_t *)sccp;		/* Keep the pointer handy */
        cep->scc_num = scc_num;
        cep->cpm_cr_ch = cpm_cr_ch;

	/* Disable receive and transmit in case EPPC-Bug started it.
	*/
	sccp->scc_gsmrl &= ~(SCC_GSMRL_ENR | SCC_GSMRL_ENT);

	/* Init physical pins
	 */
	cpm_enet_init_scc_pins(scc_num);

	/* Configure Serial Interface clock routing.
	 */
	cpm_enet_init_scc_clock(scc_num);


	/* Manual says set SDDR, but I can't find anything with that
	 * name.  I think it is a misprint, and should be SDCR.  This
	 * has already been set by the communication processor initialization.
	 */

	/* Allocate space for the buffer descriptors in the DP ram.
	 * These are relative offsets in the DP ram address space.
	 * Initialize base addresses for the buffer descriptors.
	 */
	i = m8xx_cpm_dpalloc(sizeof(cbd_t) * RX_RING_SIZE);
	ep->sen_genscc.scc_rbase = i;
	cep->rx_bd_base = (cbd_t *)&cp->cp_dpmem[i];

	i = m8xx_cpm_dpalloc(sizeof(cbd_t) * TX_RING_SIZE);
	ep->sen_genscc.scc_tbase = i;
	cep->tx_bd_base = (cbd_t *)&cp->cp_dpmem[i];

	cep->dirty_tx = cep->cur_tx = cep->tx_bd_base;
	cep->cur_rx = cep->rx_bd_base;

	/* Issue init Rx BD command for SCC.
	 * Manual says to perform an Init Rx parameters here.  We have
	 * to perform both Rx and Tx because the SCC may have been
	 * already running.
	 * In addition, we have to do it later because we don't yet have
	 * all of the BD control/status set properly.
	 */

	/* Initialize function code registers for big-endian.
	*/
	ep->sen_genscc.scc_rfcr = SCC_EB;
	ep->sen_genscc.scc_tfcr = SCC_EB;

	/* Set maximum bytes per receive buffer.
	 * This appears to be an Ethernet frame size, not the buffer
	 * fragment size.  It must be a multiple of four.
	 */
	ep->sen_genscc.scc_mrblr = PKT_MAXBLR_SIZE;

	/* Set CRC preset and mask.
	*/
	ep->sen_cpres = 0xffffffff;
	ep->sen_cmask = 0xdebb20e3;

	ep->sen_crcec = 0;	/* CRC Error counter */
	ep->sen_alec = 0;	/* alignment error counter */
	ep->sen_disfc = 0;	/* discard frame counter */

	ep->sen_pads = 0x8888;	/* Tx short frame pad character */
	ep->sen_retlim = 15;	/* Retry limit threshold */

	ep->sen_maxflr = PKT_MAXBUF_SIZE;   /* maximum frame length register */
	ep->sen_minflr = PKT_MINBUF_SIZE;  /* minimum frame length register */

	ep->sen_maxd1 = PKT_MAXBLR_SIZE;	/* maximum DMA1 length */
	ep->sen_maxd2 = PKT_MAXBLR_SIZE;	/* maximum DMA2 length */

	/* Clear hash tables.
	*/
	ep->sen_gaddr1 = 0;
	ep->sen_gaddr2 = 0;
	ep->sen_gaddr3 = 0;
	ep->sen_gaddr4 = 0;
	ep->sen_iaddr1 = 0;
	ep->sen_iaddr2 = 0;
	ep->sen_iaddr3 = 0;
	ep->sen_iaddr4 = 0;
	/* Set Ethernet station address.
	 *
	 * If we performed a MBX diskless boot, the Ethernet controller
	 * has been initialized and we copy the address out into our
	 * own structure.
	 *
	 * All other types of boards supply the address in the board
	 * information structure, so we copy that into the controller.
	 */
	eap = (unsigned char *)&(ep->sen_paddrh);
	for (i=5; i>=0; i--)
		*eap++ = dev->dev_addr[i] = bd->bi_enetaddr[i];

	ep->sen_pper = 0;	/* 'cause the book says so */
	ep->sen_taddrl = 0;	/* temp address (LSB) */
	ep->sen_taddrm = 0;
	ep->sen_taddrh = 0;	/* temp address (MSB) */

	/* Now allocate the host memory pages and initialize the
	 * buffer descriptors.
	 */
	bdp = cep->tx_bd_base;
	for (i=0; i<TX_RING_SIZE; i++) {

		/* Initialize the BD for every fragment in the page.
		*/
		bdp->cbd_sc = 0;
		bdp->cbd_bufaddr = 0;
		bdp++;
	}

	/* Set the last buffer to wrap.
	*/
	bdp--;
	bdp->cbd_sc |= BD_SC_WRAP;

	bdp = cep->rx_bd_base;
	k = 0;
	/* Initialize the BDs. */
	for (j=0; j < RX_RING_SIZE; j++) {
		struct	sk_buff * skb = __dev_alloc_skb(CPM_ENET_RX_FRSIZE, GFP_ATOMIC|GFP_DMA);

		invalidate_dcache_range((unsigned long) skb->data,
					(unsigned long) skb->data + CPM_ENET_RX_FRSIZE);
		bdp->cbd_sc = BD_ENET_RX_EMPTY | BD_ENET_RX_INTR;
		bdp->cbd_bufaddr = __pa(skb->data);
		cep->rx_vaddr[k++] = skb;
		bdp++;
	}
	/* Set the last buffer to wrap.
	*/
	bdp--;
	bdp->cbd_sc |= BD_SC_WRAP;

	/* Let's re-initialize the channel now.  We have to do it later
	 * than the manual describes because we have just now finished
	 * the BD initialization.
	 */
        cp->cp_cpcr = mk_cr_cmd(cep->cpm_cr_ch, CPM_CR_INIT_TRX) | CPM_CR_FLG;
	while (cp->cp_cpcr & CPM_CR_FLG);

	cep->skb_cur = cep->skb_dirty = 0;

	sccp->scc_scce = 0xffff;	/* Clear any pending events */

	/* Enable interrupts for transmit error, complete frame
	 * received, and any transmit buffer we have also set the
	 * interrupt flag.
	 */
	sccp->scc_sccm = (SCCE_ENET_TXE | SCCE_ENET_RXF | SCCE_ENET_TXB);

	/* Install our interrupt handler.
	*/
        cpm_install_handler(cpmvec, scc_enet_interrupt, dev);

	/* Set GSMR_H to enable all normal operating modes.
	 * Set GSMR_L to enable Ethernet to MC68160.
	 */
	sccp->scc_gsmrh = 0;
	sccp->scc_gsmrl = (SCC_GSMRL_TCI | SCC_GSMRL_TPL_48 | SCC_GSMRL_TPP_10 | SCC_GSMRL_MODE_ENET);

	/* Set sync/delimiters.
	*/
	sccp->scc_dsr = 0xd555;

	/* Set processing mode.  Use Ethernet CRC, catch broadcast, and
	 * start frame search 22 bit times after RENA.
	 */
	sccp->scc_pmsr = (SCC_PMSR_ENCRC | SCC_PMSR_NIB22);
#ifdef CONFIG_SCC_ENET_FULL_DUPLEX 
	sccp->scc_pmsr |= SCC_PMSR_LPB | SCC_PMSR_FDE;
	printk("Enet %d: Full duplex enabled\n", scc_num);
#endif
	/* It is now OK to enable the Ethernet transmitter.
	 * Unfortunately, there are board implementation differences here.
	 */
        cpm_enet_enable_xmit(scc_num);


#if defined(CONFIG_RPXLITE) || defined(CONFIG_RPXCLASSIC)
	/* And while we are here, set the configuration to enable ethernet.
	*/
	*((volatile uint *)RPX_CSR_ADDR) &= ~BCSR0_ETHLPBK;
	*((volatile uint *)RPX_CSR_ADDR) |=
			(BCSR0_ETHEN | BCSR0_COLTESTDIS | BCSR0_FULLDPLXDIS);

#endif


	dev->base_addr = (unsigned long)ep;
	dev->priv = cep;
#if 0
	dev->name = "CPM_ENET";
#endif

	/* The CPM Ethernet specific entries in the device structure. */
	dev->open = scc_enet_open;
	dev->hard_start_xmit = scc_enet_start_xmit;
	dev->tx_timeout = scc_enet_timeout;
	dev->watchdog_timeo = TX_TIMEOUT;
	dev->stop = scc_enet_close;
	dev->get_stats = scc_enet_get_stats;
	dev->set_multicast_list = set_multicast_list;
	dev->set_mac_address = set_mac_address;
	dev->change_mtu = change_mtu;

	printk("%s: CPM ENET Version 0.3, ", dev->name);
	for (i=0; i<5; i++)
		printk("%02x:", dev->dev_addr[i]);
	printk("%02x\n", dev->dev_addr[5]);

	return 0;
}

/* Cookbook style from the MPC860 manual.....
 * Not all of this is necessary if EPPC-Bug has initialized
 * the network.
 * So far we are lucky, all board configurations use the same
 * pins, or at least the same I/O Port for these functions.....
 * It can't last though......
 */

int __init cpm_enet_init_scc_pins(int scc_num)
{
    volatile immap_t *immap = (immap_t *)IMAP_ADDR;

    switch (scc_num) {
    case 0: /* SCC1 */
        /* Configure port A pins for Txd and Rxd.
         */
        immap->im_ioport.iop_papar |= (PA1_ENET_RXD | PA1_ENET_TXD);
        immap->im_ioport.iop_padir &= ~(PA1_ENET_RXD | PA1_ENET_TXD);
        immap->im_ioport.iop_paodr &= ~PA1_ENET_TXD;

        /* Configure port C pins to enable CLSN and RENA.
         */
        immap->im_ioport.iop_pcpar &= ~(PC1_ENET_CLSN | PC1_ENET_RENA);
        immap->im_ioport.iop_pcdir &= ~(PC1_ENET_CLSN | PC1_ENET_RENA);
        immap->im_ioport.iop_pcso |= (PC1_ENET_CLSN | PC1_ENET_RENA);

        /* Configure port A for TCLK and RCLK.
         */
        immap->im_ioport.iop_papar |= (PA1_ENET_TCLK | PA1_ENET_RCLK);
        immap->im_ioport.iop_padir &= ~(PA1_ENET_TCLK | PA1_ENET_RCLK);
        break;

    case 1: /* SCC2 */
        /* Configure port A pins for Txd and Rxd.
         */
        immap->im_ioport.iop_papar |= (PA2_ENET_RXD | PA2_ENET_TXD);
        immap->im_ioport.iop_padir &= ~(PA2_ENET_RXD | PA2_ENET_TXD);
        immap->im_ioport.iop_paodr &= ~PA2_ENET_TXD;

        /* Configure port C pins to enable CLSN and RENA.
         */
        immap->im_ioport.iop_pcpar &= ~(PC2_ENET_CLSN | PC2_ENET_RENA);
        immap->im_ioport.iop_pcdir &= ~(PC2_ENET_CLSN | PC2_ENET_RENA);
        immap->im_ioport.iop_pcso |= (PC2_ENET_CLSN | PC2_ENET_RENA);

        /* Configure port A for TCLK and RCLK.
         * eth1 on CP board: TCLK: CLK3    0x0400
         *                   RCLK: CLK4    0x0800
         */
        immap->im_ioport.iop_papar |= (PA2_ENET_TCLK | PA2_ENET_RCLK);
        immap->im_ioport.iop_padir &= ~(PA2_ENET_TCLK | PA2_ENET_RCLK);
        break;

    case 3: /* SCC4 */
        /* Configure port D pins for Txd and Rxd.
         */
        immap->im_ioport.iop_pdpar |= (PD4_ENET_RXD | PD4_ENET_TXD);
	/* Make sure PDDIR is zero otherwise RXD and TXD wont work on PD pins! */
        immap->im_ioport.iop_pddir &= ~(PD4_ENET_RXD | PD4_ENET_TXD);

        /* Configure port C pins to enable CLSN and RENA.
         */
        immap->im_ioport.iop_pcpar &= ~(PC4_ENET_CLSN | PC4_ENET_RENA);
        immap->im_ioport.iop_pcdir &= ~(PC4_ENET_CLSN | PC4_ENET_RENA);
        immap->im_ioport.iop_pcso |= (PC4_ENET_CLSN | PC4_ENET_RENA);

        /* Configure port A for TCLK and RCLK.
         * eth2 on CP board: TCLK: CLK7    0x8000
         *                   RCLK: CLK8    0x4000
         */
        immap->im_ioport.iop_papar |= (PA4_ENET_TCLK | PA4_ENET_RCLK);
        immap->im_ioport.iop_padir &= ~(PA4_ENET_TCLK | PA4_ENET_RCLK);
        break;
    default:
        break;
    }

    return 0;
}

int __init cpm_enet_init_scc_clock(int scc_num)
{
    volatile cpm8xx_t *cp = cpmp;

    /* First, clear all SCC bits to zero, then set the ones we want.
     */
    switch (scc_num) {
    case 0:
        cp->cp_sicr &= ~SICR_ENET_MASK_1;
        /*
         * 'First ethernet, eth0:
         * TCLK: CLK1    RCLK: CLK2
         */
        cp->cp_sicr |= SICR_ENET_CLKRT_1;  /* 0x0000002C */
        break;
    case 1:
        /*
         * 'Second ethernet, eth1:
         * TCLK: CLK3    RCLK: CLK4
         */
        cp->cp_sicr &= ~SICR_ENET_MASK_2; /* 0x0000ff00 */
        cp->cp_sicr |= SICR_ENET_CLKRT_2; /* 0x00003E00 */
        break;
    case 3:
        /*
         * eth2:
         * TCLK: CLK7    RCLK: CLK8
         */
        cp->cp_sicr &= ~SICR_ENET_MASK_4; /* 0xff000000 */
        cp->cp_sicr |= SICR_ENET_CLKRT_4; /* 0x3E000000 */
        break;
    default:
        break;
    }
    return 0;
}

static void cpm_enet_enable_xmit(int scc_num)
{
    volatile immap_t *immap = (immap_t *)IMAP_ADDR;
    volatile cpm8xx_t *cp = cpmp;

    switch (scc_num) {
    case 0: /* SCC1 */
	cp->cp_pbpar |= PB1_ENET_TENA;
	cp->cp_pbdir |= PB1_ENET_TENA;
        break;
    case 1: /* SCC2 */
	cp->cp_pbpar |= PB2_ENET_TENA;
	cp->cp_pbdir |= PB2_ENET_TENA;
        break;
    case 3: /* SCC4 */
        immap->im_ioport.iop_pdpar |= PD4_ENET_TENA;
	/* Make sure PDDIR is zero otherwise TENA wont work on PD pins! */
        immap->im_ioport.iop_pddir &= ~PD4_ENET_TENA;
        break;
    default:
        break;
    }
}