patch-2.4.21 linux-2.4.21/drivers/net/ether00.c

Next file: linux-2.4.21/drivers/net/ewrk3.c
Previous file: linux-2.4.21/drivers/net/eth16i.c
Back to the patch index
Back to the overall index

diff -urN linux-2.4.20/drivers/net/ether00.c linux-2.4.21/drivers/net/ether00.c
@@ -0,0 +1,1001 @@
+/*
+ *  drivers/net/ether00.c
+ *
+ *  Copyright (C) 2001 Altera Corporation
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ */
+
+/* includes */
+
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/sched.h>
+#include <linux/netdevice.h>
+#include <linux/skbuff.h>
+#include <linux/etherdevice.h>
+#include <linux/module.h>
+#include <linux/tqueue.h>
+#include <linux/mtd/mtd.h>
+#include <asm/arch/excalibur.h>
+#include <asm/arch/hardware.h>
+#include <asm/irq.h>
+#include <asm/io.h>
+#include <asm/sizes.h>
+
+#include <asm/arch/ether00.h>
+#include <asm/arch/tdkphy.h>
+
+
+MODULE_AUTHOR("Clive Davies");
+MODULE_DESCRIPTION("Altera Ether00 IP core driver");
+MODULE_LICENSE("GPL");
+
+static long base=0x60000000;
+static int irq=0x1;
+static int phy_irq=0x2;
+MODULE_PARM(base,"l");
+MODULE_PARM(irq,"i");
+MODULE_PARM(phy_irq,"i");
+
+#define TX_TIMEOUT  (400*HZ/1000)
+#define PKT_BUF_SZ 1540 /* Size of each rx buffer */
+
+
+#undef  DEBUG
+#define DEBUG(x)
+
+#define __dma_va(x) (unsigned int)((unsigned int)priv->dma_data+(((unsigned int)(x))&(EXC_SPSRAM_BLOCK0_SIZE-1)))
+#define __dma_pa(x) (unsigned int)(EXC_SPSRAM_BLOCK0_BASE+(((unsigned int)(x))-(unsigned int)priv->dma_data))
+
+#define ETHER00_BASE	0
+#define	ETHER00_TYPE
+#define ETHER00_NAME "ether00"
+#define MAC_REG_SIZE 0x400 /* size of MAC register area */
+
+
+
+/* typedefs */
+
+/* The definition of the driver control structure */
+
+#define RX_NUM_BUFF     10
+#define RX_NUM_FDESC    10
+#define TX_NUM_FDESC    10
+
+struct tx_fda_ent{
+	FDA_DESC  fd;
+	BUF_DESC  bd;
+	BUF_DESC  pad;
+};
+struct rx_fda_ent{
+	FDA_DESC  fd;
+	BUF_DESC  bd;
+	BUF_DESC  pad;
+};
+struct rx_blist_ent{
+	FDA_DESC  fd;
+	BUF_DESC  bd;
+	BUF_DESC  pad;
+};
+struct net_priv
+{
+	struct net_device_stats stats;
+	struct sk_buff* skb;
+	void* dma_data;
+	struct rx_blist_ent*  rx_blist_vp;
+	struct rx_fda_ent* rx_fda_ptr;
+	struct tx_fda_ent* tx_fdalist_vp;
+	struct tq_struct  tq_memupdate;
+        unsigned char   rx_disabled;
+	unsigned char   memupdate_scheduled;
+	unsigned char   queue_stopped;
+	spinlock_t dma_lock;
+        unsigned int tx_head;
+        unsigned int tx_tail;
+};
+
+static const char vendor_id[2]={0x07,0xed};
+
+#ifdef ETHER00_DEBUG
+
+/* Dump (most) registers for debugging puposes */
+
+static void dump_regs(struct net_device *dev){
+	struct net_priv* priv=dev->priv;
+	unsigned int* i;
+
+	printk("\n RX free descriptor area:\n");
+
+	for(i=(unsigned int*)priv->rx_fda_ptr;
+	    i<((unsigned int*)(priv->rx_fda_ptr+RX_NUM_FDESC));){
+		printk("%#8x %#8x %#8x %#8x\n",*i,*(i+1),*(i+2),*(i+3));
+		i+=4;
+	}
+
+	printk("\n RX buffer list:\n");
+
+	for(i=(unsigned int*)priv->rx_blist_vp;
+	    i<((unsigned int*)(priv->rx_blist_vp+RX_NUM_BUFF));){
+		printk("%#8x %#8x %#8x %#8x\n",*i,*(i+1),*(i+2),*(i+3));
+		i+=4;
+	}
+
+	printk("\n TX frame descriptor list:\n");
+
+	for(i=(unsigned int*)priv->tx_fdalist_vp;
+	    i<((unsigned int*)(priv->tx_fdalist_vp+TX_NUM_FDESC));){
+		printk("%#8x %#8x %#8x %#8x\n",*i,*(i+1),*(i+2),*(i+3));
+		i+=4;
+	}
+
+	printk("\ndma ctl=%#x\n",readw(ETHER_DMA_CTL(dev->base_addr)));
+	printk("txfrmptr=%#x\n",readw(ETHER_TXFRMPTR(dev->base_addr)));
+	printk("txthrsh=%#x\n",readw(ETHER_TXTHRSH(dev->base_addr)));
+	printk("txpollctr=%#x\n",readw(ETHER_TXPOLLCTR(dev->base_addr)));
+	printk("blfrmptr=%#x\n",readw(ETHER_BLFRMPTR(dev->base_addr)));
+	printk("rxfragsize=%#x\n",readw(ETHER_RXFRAGSIZE(dev->base_addr)));
+	printk("tx_int_en=%#x\n",readw(ETHER_INT_EN(dev->base_addr)));
+	printk("fda_bas=%#x\n",readw(ETHER_FDA_BAS(dev->base_addr)));
+	printk("fda_lim=%#x\n",readw(ETHER_FDA_LIM(dev->base_addr)));
+	printk("int_src=%#x\n",readw(ETHER_INT_SRC(dev->base_addr)));
+	printk("pausecnt=%#x\n",readw(ETHER_PAUSECNT(dev->base_addr)));
+	printk("rempaucnt=%#x\n",readw(ETHER_REMPAUCNT(dev->base_addr)));
+	printk("txconfrmstat=%#x\n",readw(ETHER_TXCONFRMSTAT(dev->base_addr)));
+	printk("mac_ctl=%#x\n",readw(ETHER_MAC_CTL(dev->base_addr)));
+	printk("arc_ctl=%#x\n",readw(ETHER_ARC_CTL(dev->base_addr)));
+	printk("tx_ctl=%#x\n",readw(ETHER_TX_CTL(dev->base_addr)));
+}
+#endif /* ETHER00_DEBUG */
+
+
+static int ether00_write_phy(struct net_device *dev, short address, short value)
+{
+	volatile int count = 1024;
+	writew(value,ETHER_MD_DATA(dev->base_addr));
+	writew( ETHER_MD_CA_BUSY_MSK |
+		ETHER_MD_CA_WR_MSK |
+		(address & ETHER_MD_CA_ADDR_MSK),
+		ETHER_MD_CA(dev->base_addr));
+
+	/* Wait for the command to complete */
+	while((readw(ETHER_MD_CA(dev->base_addr)) & ETHER_MD_CA_BUSY_MSK)&&count){
+		count--;
+	}
+	if (!count){
+		printk("Write to phy failed, addr=%#x, data=%#x\n",address, value);
+		return -EIO;
+	}
+	return 0;
+}
+
+static int ether00_read_phy(struct net_device *dev, short address)
+{
+	volatile int count = 1024;
+	writew( ETHER_MD_CA_BUSY_MSK |
+		(address & ETHER_MD_CA_ADDR_MSK),
+		ETHER_MD_CA(dev->base_addr));
+
+	/* Wait for the command to complete */
+	while((readw(ETHER_MD_CA(dev->base_addr)) & ETHER_MD_CA_BUSY_MSK)&&count){
+		count--;
+	}
+	if (!count){
+		printk(KERN_WARNING "Read from phy timed out\n");
+		return -EIO;
+	}
+	return readw(ETHER_MD_DATA(dev->base_addr));
+}
+
+static void ether00_phy_int(int irq_num, void* dev_id, struct pt_regs* regs)
+{
+	struct net_device* dev=dev_id;
+	int irq_status;
+
+	irq_status=ether00_read_phy(dev, PHY_IRQ_CONTROL);
+
+	if(irq_status & PHY_IRQ_CONTROL_ANEG_COMP_INT_MSK){
+		/*
+		 * Autonegotiation complete on epxa10db. The mac doesn't
+		 * twig if we're in full duplex so we need to check the
+		 * phy diagnostic register and configure the mac accordingly
+		 */
+	  if(ether00_read_phy(dev, PHY_DIAGNOSTIC)&PHY_DIAGNOSTIC_DPLX_MSK){
+		  int tmp;
+		  tmp=readl(ETHER_MAC_CTL(dev->base_addr));
+		  writel(tmp|ETHER_MAC_CTL_FULLDUP_MSK,
+			 ETHER_MAC_CTL(dev->base_addr));
+		}
+                else
+		  {
+		    int tmp;
+		    tmp=readl(ETHER_MAC_CTL(dev->base_addr));
+		    writel(tmp&(~ETHER_MAC_CTL_FULLDUP_MSK),
+			   ETHER_MAC_CTL(dev->base_addr));
+
+		  }
+	}
+
+	if(irq_status&PHY_IRQ_CONTROL_LS_CHG_INT_MSK){
+
+		if(ether00_read_phy(dev, PHY_STATUS)& PHY_STATUS_LINK_MSK){
+			/* Link is up */
+			netif_carrier_on(dev);
+		}else{
+			netif_carrier_off(dev);
+		}
+	}
+
+}
+
+static void setup_blist_entry(struct sk_buff* skb,struct rx_blist_ent* blist_ent_ptr){
+	/* Make the buffer consistent with the cache as the mac is going to write
+	 * directly into it*/
+	blist_ent_ptr->fd.FDSystem=(unsigned int)skb;
+	blist_ent_ptr->bd.BuffData=(char*)__pa(skb->data);
+	consistent_sync(skb->data,PKT_BUF_SZ,PCI_DMA_FROMDEVICE);
+        /* align IP on 16 Byte (DMA_CTL set to skip 2 bytes) */
+	skb_reserve(skb,2);
+	blist_ent_ptr->bd.BuffLength=PKT_BUF_SZ-2;
+	blist_ent_ptr->fd.FDLength=1;
+	blist_ent_ptr->fd.FDCtl=FDCTL_COWNSFD_MSK;
+	blist_ent_ptr->bd.BDCtl=BDCTL_COWNSBD_MSK;
+}
+
+
+static int ether00_mem_init(struct net_device* dev)
+{
+	struct net_priv* priv=dev->priv;
+	struct tx_fda_ent *tx_fd_ptr,*tx_end_ptr;
+	struct rx_blist_ent* blist_ent_ptr;
+	int i;
+
+	/*
+	 * Grab a block of on chip SRAM to contain the control stuctures for
+	 * the ethernet MAC. This uncached becuase it needs to be accesses by both
+	 * bus masters (cpu + mac). However, it shouldn't matter too much in terms
+	 * of speed as its on chip memory
+	 */
+	priv->dma_data=ioremap_nocache(EXC_SPSRAM_BLOCK0_BASE,EXC_SPSRAM_BLOCK0_SIZE );
+	if (!priv->dma_data)
+		return -ENOMEM;
+
+	priv->rx_fda_ptr=(struct rx_fda_ent*)priv->dma_data;
+	/*
+	 * Now share it out amongst the Frame descriptors and the buffer list
+	 */
+	priv->rx_blist_vp=(struct rx_blist_ent*)((unsigned int)priv->dma_data+RX_NUM_FDESC*sizeof(struct rx_fda_ent));
+
+	/*
+	 *Initalise the FDA list
+	 */
+	/* set ownership to the controller */
+	memset(priv->rx_fda_ptr,0x80,RX_NUM_FDESC*sizeof(struct rx_fda_ent));
+
+	/*
+	 *Initialise the buffer list
+	 */
+	blist_ent_ptr=priv->rx_blist_vp;
+	i=0;
+	while(blist_ent_ptr<(priv->rx_blist_vp+RX_NUM_BUFF)){
+		struct sk_buff *skb;
+		blist_ent_ptr->fd.FDLength=1;
+		skb=dev_alloc_skb(PKT_BUF_SZ);
+		if(skb){
+			setup_blist_entry(skb,blist_ent_ptr);
+			blist_ent_ptr->fd.FDNext=(FDA_DESC*)__dma_pa(blist_ent_ptr+1);
+			blist_ent_ptr->bd.BDStat=i++;
+			blist_ent_ptr++;
+		}
+		else
+		{
+			printk("Failed to initalise buffer list\n");
+		}
+
+	}
+	blist_ent_ptr--;
+	blist_ent_ptr->fd.FDNext=(FDA_DESC*)__dma_pa(priv->rx_blist_vp);
+
+	priv->tx_fdalist_vp=(struct tx_fda_ent*)(priv->rx_blist_vp+RX_NUM_BUFF);
+
+	/* Initialise the buffers to be a circular list. The mac will then go poll
+	 * the list until it finds a frame ready to transmit */
+	tx_end_ptr=priv->tx_fdalist_vp+TX_NUM_FDESC;
+	for(tx_fd_ptr=priv->tx_fdalist_vp;tx_fd_ptr<tx_end_ptr;tx_fd_ptr++){
+		tx_fd_ptr->fd.FDNext=(FDA_DESC*)__dma_pa((tx_fd_ptr+1));
+		tx_fd_ptr->fd.FDCtl=1;
+		tx_fd_ptr->fd.FDStat=0;
+		tx_fd_ptr->fd.FDSystem=0;
+		tx_fd_ptr->fd.FDLength=1;
+
+	}
+	/* Change the last FDNext pointer to make a circular list */
+	tx_fd_ptr--;
+	tx_fd_ptr->fd.FDNext=(FDA_DESC*)__dma_pa(priv->tx_fdalist_vp);
+
+	/* Point the device at the chain of Rx and Tx Buffers */
+	writel((unsigned int)__dma_pa(priv->rx_fda_ptr),ETHER_FDA_BAS(dev->base_addr));
+	writel((RX_NUM_FDESC-1)*sizeof(struct rx_fda_ent),ETHER_FDA_LIM(dev->base_addr));
+	writel((unsigned int)__dma_pa(priv->rx_blist_vp),ETHER_BLFRMPTR(dev->base_addr));
+
+        priv->tx_tail = priv->tx_head = (unsigned int) priv->tx_fdalist_vp;
+	writel((unsigned int) __dma_pa(priv->tx_head), ETHER_TXFRMPTR(dev->base_addr));
+
+	return 0;
+}
+
+
+void ether00_mem_update(void* dev_id)
+{
+	struct net_device* dev=dev_id;
+	struct net_priv* priv=dev->priv;
+	struct rx_blist_ent* blist_ent_ptr;
+	unsigned long flags;
+        int enable_rx = 0;
+
+	priv->tq_memupdate.sync=0;
+	priv->memupdate_scheduled=0;
+
+	/* Fill in any missing buffers from the received queue */
+	blist_ent_ptr=priv->rx_blist_vp;
+	while(blist_ent_ptr<(priv->rx_blist_vp+RX_NUM_BUFF)){
+		spin_lock_irqsave(&priv->dma_lock,flags);
+		/* fd.FDSystem of 0 indicates we failed to allocate the buffer in the ISR */
+		if(!blist_ent_ptr->fd.FDSystem){
+			struct sk_buff *skb;
+			skb=dev_alloc_skb(PKT_BUF_SZ);
+			blist_ent_ptr->fd.FDSystem=(unsigned int)skb;
+			if(skb){
+				setup_blist_entry(skb,blist_ent_ptr);
+                                enable_rx = 1;
+			}
+			else
+			{
+                                /*
+                                 * reschedule the clean up, since we
+                                 * didn't patch up all the buffers
+                                 */
+
+	                         if(!priv->memupdate_scheduled){
+				   schedule_task(&priv->tq_memupdate);
+				   priv->memupdate_scheduled=1;
+				 }
+		                 spin_unlock_irqrestore(&priv->dma_lock,flags);
+				 break;
+			}
+		}
+		spin_unlock_irqrestore(&priv->dma_lock,flags);
+		blist_ent_ptr++;
+	}
+
+	if(enable_rx){
+	  if (!priv->rx_disabled){
+	        priv->rx_disabled = 0;
+		writel(ETHER_RX_CTL_RXEN_MSK,ETHER_RX_CTL(dev->base_addr));
+	  }
+	}
+}
+
+
+static void ether00_int( int irq_num, void* dev_id, struct pt_regs* regs)
+{
+	struct net_device* dev=dev_id;
+	struct net_priv* priv=dev->priv;
+
+	unsigned int   interruptValue;
+
+        int enable_tx = 0;
+	struct tx_fda_ent *fda_ptr;
+	struct sk_buff* skb;
+
+	interruptValue=readl(ETHER_INT_SRC(dev->base_addr));
+
+	if(!(readl(ETHER_INT_SRC(dev->base_addr)) & ETHER_INT_SRC_IRQ_MSK))
+	{
+		return;		/* Interrupt wasn't caused by us!! */
+	}
+
+	if(readl(ETHER_INT_SRC(dev->base_addr))&
+	   (ETHER_INT_SRC_INTMACRX_MSK |
+	    ETHER_INT_SRC_FDAEX_MSK |
+	    ETHER_INT_SRC_BLEX_MSK)) {
+		struct rx_blist_ent* blist_ent_ptr;
+		struct rx_fda_ent* fda_ent_ptr;
+		struct sk_buff* skb;
+
+		fda_ent_ptr=priv->rx_fda_ptr;
+		spin_lock(&priv->dma_lock);
+		while(fda_ent_ptr<(priv->rx_fda_ptr+RX_NUM_FDESC)){
+			int result;
+
+			if(!(fda_ent_ptr->fd.FDCtl&FDCTL_COWNSFD_MSK))
+			{
+				/* This frame is ready for processing */
+				/*find the corresponding buffer in the bufferlist */
+				blist_ent_ptr=priv->rx_blist_vp+fda_ent_ptr->bd.BDStat;
+				skb=(struct sk_buff*)blist_ent_ptr->fd.FDSystem;
+
+				/* Pass this skb up the stack */
+				skb->dev=dev;
+				skb_put(skb,fda_ent_ptr->fd.FDLength);
+				skb->protocol=eth_type_trans(skb,dev);
+				skb->ip_summed=CHECKSUM_UNNECESSARY;
+				result=netif_rx(skb);
+				/* Update statistics */
+				priv->stats.rx_packets++;
+				priv->stats.rx_bytes+=fda_ent_ptr->fd.FDLength;
+
+				/* Free the FDA entry */
+				fda_ent_ptr->bd.BDStat=0xff;
+				fda_ent_ptr->fd.FDCtl=FDCTL_COWNSFD_MSK;
+
+				/* Allocate a new skb and point the bd entry to it */
+				blist_ent_ptr->fd.FDSystem=0;
+				skb=dev_alloc_skb(PKT_BUF_SZ);
+				if(skb){
+					setup_blist_entry(skb,blist_ent_ptr);
+
+				}
+				else if(!priv->memupdate_scheduled){
+					int tmp;
+					/* There are no buffers at the moment, so schedule */
+					/* the background task to sort this out */
+					schedule_task(&priv->tq_memupdate);
+					priv->memupdate_scheduled=1;
+					printk(KERN_DEBUG "%s:No buffers",dev->name);
+					/* If this interrupt was due to a lack of buffers then
+					 * we'd better stop the receiver too */
+					if(interruptValue&ETHER_INT_SRC_BLEX_MSK){
+						priv->rx_disabled=1;
+						tmp=readl(ETHER_INT_SRC(dev->base_addr));
+						writel(tmp&~ETHER_RX_CTL_RXEN_MSK,ETHER_RX_CTL(dev->base_addr));
+						printk(KERN_DEBUG "%s:Halting rx",dev->name);
+					}
+
+				}
+
+			}
+			fda_ent_ptr++;
+		}
+		spin_unlock(&priv->dma_lock);
+
+		/* Clear the  interrupts */
+		writel(ETHER_INT_SRC_INTMACRX_MSK | ETHER_INT_SRC_FDAEX_MSK
+		       | ETHER_INT_SRC_BLEX_MSK,ETHER_INT_SRC(dev->base_addr));
+
+	}
+
+	if(readl(ETHER_INT_SRC(dev->base_addr))&ETHER_INT_SRC_INTMACTX_MSK){
+
+	  /* Transmit interrupt */
+
+	  fda_ptr=(struct tx_fda_ent*) priv->tx_tail;
+
+          /* free up all completed frames */
+
+	  while(!(FDCTL_COWNSFD_MSK&fda_ptr->fd.FDCtl) && fda_ptr->fd.FDSystem){
+	    priv->stats.tx_packets++;
+	    priv->stats.tx_bytes+=fda_ptr->bd.BuffLength;
+	    skb=(struct sk_buff*)fda_ptr->fd.FDSystem;
+	    dev_kfree_skb_irq(skb);
+	    fda_ptr->fd.FDSystem=0;
+	    fda_ptr->fd.FDStat=0;
+	    fda_ptr->fd.FDCtl=0;
+	    fda_ptr = (struct tx_fda_ent *)__dma_va(fda_ptr->fd.FDNext);
+	    enable_tx = 1;
+	  }
+	  priv->tx_tail = (unsigned int) fda_ptr;
+
+	  if(priv->queue_stopped && enable_tx){
+	    priv->queue_stopped=0;
+	    netif_wake_queue(dev);
+	  }
+
+	  /* Clear the interrupt */
+	  writel(ETHER_INT_SRC_INTMACTX_MSK,ETHER_INT_SRC(dev->base_addr));
+	}
+
+	if (readl(ETHER_INT_SRC(dev->base_addr)) & (ETHER_INT_SRC_SWINT_MSK|
+						    ETHER_INT_SRC_INTEARNOT_MSK|
+						    ETHER_INT_SRC_INTLINK_MSK|
+						    ETHER_INT_SRC_INTEXBD_MSK|
+						    ETHER_INT_SRC_INTTXCTLCMP_MSK))
+	{
+		/*
+		 *	Not using any of these so they shouldn't happen
+		 *
+		 *	In the cased of INTEXBD - if you allocate more
+		 *      than 28 decsriptors you may need to think about this
+		 */
+		printk("Not using this interrupt\n");
+	}
+
+	if (readl(ETHER_INT_SRC(dev->base_addr)) &
+	    (ETHER_INT_SRC_INTSBUS_MSK |
+	     ETHER_INT_SRC_INTNRABT_MSK
+	     |ETHER_INT_SRC_DMPARERR_MSK))
+	{
+		/*
+		 * Hardware errors, we can either ignore them and hope they go away
+		 *or reset the device, I'll try the first for now to see if they happen
+		 */
+		printk("Hardware error\n");
+	}
+}
+
+static void ether00_setup_ethernet_address(struct net_device* dev)
+{
+	int tmp;
+
+	dev->addr_len=6;
+	writew(0,ETHER_ARC_ADR(dev->base_addr));
+	writel((dev->dev_addr[0]<<24) |
+		(dev->dev_addr[1]<<16) |
+		(dev->dev_addr[2]<<8) |
+		dev->dev_addr[3],
+		ETHER_ARC_DATA(dev->base_addr));
+
+	writew(4,ETHER_ARC_ADR(dev->base_addr));
+	tmp=readl(ETHER_ARC_DATA(dev->base_addr));
+	tmp&=0xffff;
+	tmp|=(dev->dev_addr[4]<<24) | (dev->dev_addr[5]<<16);
+	writel(tmp, ETHER_ARC_DATA(dev->base_addr));
+	/* Enable this entry in the ARC */
+
+	writel(1,ETHER_ARC_ENA(dev->base_addr));
+
+	return;
+}
+
+
+static void ether00_reset(struct net_device *dev)
+{
+	/* reset the controller */
+	writew(ETHER_MAC_CTL_RESET_MSK,ETHER_MAC_CTL(dev->base_addr));
+
+	/*
+	 * Make sure we're not going to send anything
+	 */
+
+	writew(ETHER_TX_CTL_TXHALT_MSK,ETHER_TX_CTL(dev->base_addr));
+
+	/*
+	 * Make sure we're not going to receive anything
+	 */
+	writew(ETHER_RX_CTL_RXHALT_MSK,ETHER_RX_CTL(dev->base_addr));
+
+	/*
+	 * Disable Interrupts for now, and set the burst size to 8 bytes
+	 */
+
+	writel(ETHER_DMA_CTL_INTMASK_MSK |
+	       ((8 << ETHER_DMA_CTL_DMBURST_OFST) & ETHER_DMA_CTL_DMBURST_MSK)
+	       |(2<<ETHER_DMA_CTL_RXALIGN_OFST),
+	       ETHER_DMA_CTL(dev->base_addr));
+
+
+	/*
+	 * Set TxThrsh - start transmitting a packet after 1514
+	 * bytes or when a packet is complete, whichever comes first
+	 */
+	 writew(1514,ETHER_TXTHRSH(dev->base_addr));
+
+	/*
+	 * Set TxPollCtr.  Each cycle is
+	 * 61.44 microseconds with a 33 MHz bus
+	 */
+	 writew(1,ETHER_TXPOLLCTR(dev->base_addr));
+
+	/*
+	 * Set Rx_Ctl - Turn off reception and let RxData turn it
+	 * on later
+	 */
+	 writew(ETHER_RX_CTL_RXHALT_MSK,ETHER_RX_CTL(dev->base_addr));
+
+}
+
+
+static void ether00_set_multicast(struct net_device* dev)
+{
+	int count=dev->mc_count;
+
+	/* Set promiscuous mode if it's asked for. */
+
+	if (dev->flags&IFF_PROMISC){
+
+		writew( ETHER_ARC_CTL_COMPEN_MSK |
+			ETHER_ARC_CTL_BROADACC_MSK |
+			ETHER_ARC_CTL_GROUPACC_MSK |
+			ETHER_ARC_CTL_STATIONACC_MSK,
+			ETHER_ARC_CTL(dev->base_addr));
+		return;
+	}
+
+	/*
+	 * Get all multicast packets if required, or if there are too
+	 * many addresses to fit in hardware
+	 */
+	if (dev->flags & IFF_ALLMULTI){
+		writew( ETHER_ARC_CTL_COMPEN_MSK |
+			ETHER_ARC_CTL_GROUPACC_MSK |
+			ETHER_ARC_CTL_BROADACC_MSK,
+			ETHER_ARC_CTL(dev->base_addr));
+		return;
+	}
+	if (dev->mc_count > (ETHER_ARC_SIZE - 1)){
+
+		printk(KERN_WARNING "Too many multicast addresses for hardware to filter - receiving all multicast packets\n");
+		writew( ETHER_ARC_CTL_COMPEN_MSK |
+			ETHER_ARC_CTL_GROUPACC_MSK |
+			ETHER_ARC_CTL_BROADACC_MSK,
+			ETHER_ARC_CTL(dev->base_addr));
+		return;
+	}
+
+	if(dev->mc_count){
+		struct dev_mc_list *mc_list_ent=dev->mc_list;
+		unsigned int temp,i;
+		DEBUG(printk("mc_count=%d mc_list=%#x\n",dev-> mc_count, dev->mc_list));
+		DEBUG(printk("mc addr=%02#x%02x%02x%02x%02x%02x\n",
+			     mc_list_ent->dmi_addr[5],
+			     mc_list_ent->dmi_addr[4],
+			     mc_list_ent->dmi_addr[3],
+			     mc_list_ent->dmi_addr[2],
+			     mc_list_ent->dmi_addr[1],
+			     mc_list_ent->dmi_addr[0]);)
+
+		/*
+		 * The first 6 bytes are the MAC address, so
+		 * don't change them!
+		 */
+		writew(4,ETHER_ARC_ADR(dev->base_addr));
+		temp=readl(ETHER_ARC_DATA(dev->base_addr));
+		temp&=0xffff0000;
+
+		/* Disable the current multicast stuff */
+		writel(1,ETHER_ARC_ENA(dev->base_addr));
+
+		for(;;){
+			temp|=mc_list_ent->dmi_addr[1] |
+				mc_list_ent->dmi_addr[0]<<8;
+			writel(temp,ETHER_ARC_DATA(dev->base_addr));
+
+			i=readl(ETHER_ARC_ADR(dev->base_addr));
+			writew(i+4,ETHER_ARC_ADR(dev->base_addr));
+
+			temp=mc_list_ent->dmi_addr[5]|
+				mc_list_ent->dmi_addr[4]<<8 |
+				mc_list_ent->dmi_addr[3]<<16 |
+				mc_list_ent->dmi_addr[2]<<24;
+			writel(temp,ETHER_ARC_DATA(dev->base_addr));
+
+			count--;
+			if(!mc_list_ent->next || !count){
+				break;
+			}
+			DEBUG(printk("mc_list_next=%#x\n",mc_list_ent->next);)
+			mc_list_ent=mc_list_ent->next;
+
+
+			i=readl(ETHER_ARC_ADR(dev->base_addr));
+			writel(i+4,ETHER_ARC_ADR(dev->base_addr));
+
+			temp=mc_list_ent->dmi_addr[3]|
+				mc_list_ent->dmi_addr[2]<<8 |
+				mc_list_ent->dmi_addr[1]<<16 |
+				mc_list_ent->dmi_addr[0]<<24;
+			writel(temp,ETHER_ARC_DATA(dev->base_addr));
+
+			i=readl(ETHER_ARC_ADR(dev->base_addr));
+			writel(i+4,ETHER_ARC_ADR(dev->base_addr));
+
+			temp=mc_list_ent->dmi_addr[4]<<16 |
+				mc_list_ent->dmi_addr[5]<<24;
+
+			writel(temp,ETHER_ARC_DATA(dev->base_addr));
+
+			count--;
+			if(!mc_list_ent->next || !count){
+				break;
+			}
+			mc_list_ent=mc_list_ent->next;
+		}
+
+
+		if(count)
+			printk(KERN_WARNING "Multicast list size error\n");
+
+
+		writew( ETHER_ARC_CTL_BROADACC_MSK|
+			ETHER_ARC_CTL_COMPEN_MSK,
+			ETHER_ARC_CTL(dev->base_addr));
+
+	}
+
+	/* enable the active ARC enties */
+	writew((1<<(dev->mc_count+2))-1,ETHER_ARC_ENA(dev->base_addr));
+}
+
+
+static int ether00_open(struct net_device* dev)
+{
+	int result,tmp;
+	struct net_priv* priv;
+
+	if (!is_valid_ether_addr(dev->dev_addr))
+		return -EINVAL;
+
+	dev->base_addr=(unsigned int)ioremap_nocache(base,SZ_4K);
+
+	dev->irq=irq;
+
+	/* Allocate private memory */
+	dev->priv=kmalloc(sizeof(struct net_priv),GFP_KERNEL);
+	if(!dev->priv)
+		return -ENOMEM;
+	memset(dev->priv,0,sizeof(struct net_priv));
+	priv=(struct net_priv*)dev->priv;
+	priv->tq_memupdate.routine=ether00_mem_update;
+	priv->tq_memupdate.data=(void*) dev;
+	spin_lock_init(&priv->dma_lock);
+
+	/* Install interrupt handlers */
+	result=request_irq(dev->irq,ether00_int,0,"ether00",dev);
+	if(result)
+		goto open_err1;
+
+	result=request_irq(phy_irq,ether00_phy_int,0,"ether00_phy",dev);
+	if(result)
+		goto open_err2;
+
+	ether00_reset(dev);
+	result=ether00_mem_init(dev);
+	if(result)
+		goto open_err3;
+
+
+	ether00_setup_ethernet_address(dev);
+
+	ether00_set_multicast(dev);
+
+	result=ether00_write_phy(dev,PHY_CONTROL, PHY_CONTROL_ANEGEN_MSK | PHY_CONTROL_RANEG_MSK);
+	if(result)
+		goto open_err4;
+	result=ether00_write_phy(dev,PHY_IRQ_CONTROL, PHY_IRQ_CONTROL_LS_CHG_IE_MSK |
+				 PHY_IRQ_CONTROL_ANEG_COMP_IE_MSK);
+	if(result)
+		goto open_err4;
+
+	/* Start the device enable interrupts */
+	writew(ETHER_RX_CTL_RXEN_MSK
+//	       | ETHER_RX_CTL_STRIPCRC_MSK
+	       | ETHER_RX_CTL_ENGOOD_MSK
+	       | ETHER_RX_CTL_ENRXPAR_MSK| ETHER_RX_CTL_ENLONGERR_MSK
+	       | ETHER_RX_CTL_ENOVER_MSK| ETHER_RX_CTL_ENCRCERR_MSK,
+	       ETHER_RX_CTL(dev->base_addr));
+
+	writew(ETHER_TX_CTL_TXEN_MSK|
+	       ETHER_TX_CTL_ENEXDEFER_MSK|
+	       ETHER_TX_CTL_ENLCARR_MSK|
+	       ETHER_TX_CTL_ENEXCOLL_MSK|
+	       ETHER_TX_CTL_ENLATECOLL_MSK|
+	       ETHER_TX_CTL_ENTXPAR_MSK|
+	       ETHER_TX_CTL_ENCOMP_MSK,
+	       ETHER_TX_CTL(dev->base_addr));
+
+	tmp=readl(ETHER_DMA_CTL(dev->base_addr));
+	writel(tmp&~ETHER_DMA_CTL_INTMASK_MSK,ETHER_DMA_CTL(dev->base_addr));
+
+	return 0;
+
+ open_err4:
+	ether00_reset(dev);
+ open_err3:
+	free_irq(2,dev);
+ open_err2:
+	free_irq(dev->irq,dev);
+ open_err1:
+	iounmap((void*)dev->base_addr);
+	kfree(dev->priv);
+	return result;
+
+}
+
+
+static int ether00_tx(struct sk_buff* skb, struct net_device* dev)
+{
+	struct net_priv *priv=dev->priv;
+	struct tx_fda_ent *fda_ptr;
+	unsigned long flags;
+        int retcode = 0;
+
+	/*
+	 *	Find an empty slot in which to stick the frame
+	 */
+
+	spin_lock_irqsave(&priv->dma_lock,flags);
+
+	fda_ptr=(struct tx_fda_ent*) priv->tx_head;
+
+	priv->tx_head =(unsigned int) __dma_va(fda_ptr->fd.FDNext);
+
+	/* Write the skb data from the cache*/
+	consistent_sync(skb->data,skb->len,PCI_DMA_TODEVICE);
+	fda_ptr->bd.BuffData=(char*)__pa(skb->data);
+	fda_ptr->bd.BuffLength=(unsigned short)skb->len;
+	/* Save the pointer to the skb for freeing later */
+	fda_ptr->fd.FDSystem=(unsigned int)skb;
+	fda_ptr->fd.FDStat=0;
+	/* Pass ownership of the buffers to the controller */
+	fda_ptr->fd.FDCtl=1;
+	fda_ptr->fd.FDCtl|=FDCTL_COWNSFD_MSK;
+
+	fda_ptr=(struct tx_fda_ent*) priv->tx_head;
+
+	/* If the next buffer in the list is full, stop the queue */
+	if (fda_ptr->fd.FDSystem){
+		netif_stop_queue(dev);
+		priv->queue_stopped=1;
+	}
+
+	spin_unlock_irqrestore(&priv->dma_lock,flags);
+	return retcode;
+}
+
+static void ether00_tx_timeout(struct net_device* dev)
+{
+         /*
+          * Something really bad has happened here. This
+          * SHOULD never happen. Given that, it's difficult
+          * to know what to do to recover. For now we'll just
+          * count the error, and restart the queue.
+          */
+
+        struct net_priv *priv=dev->priv;
+
+	priv->stats.tx_errors++;
+
+	priv->queue_stopped=0;
+	netif_wake_queue(dev);
+
+}
+
+static struct net_device_stats *ether00_stats(struct net_device* dev)
+{
+	struct net_priv *priv=dev->priv;
+	return &priv->stats;
+}
+
+
+static int ether00_stop(struct net_device* dev)
+{
+	struct net_priv *priv=dev->priv;
+	int tmp;
+
+	/* Stop/disable the device. */
+	tmp=readw(ETHER_RX_CTL(dev->base_addr));
+	tmp&=~(ETHER_RX_CTL_RXEN_MSK | ETHER_RX_CTL_ENGOOD_MSK);
+	tmp|=ETHER_RX_CTL_RXHALT_MSK;
+	writew(tmp,ETHER_RX_CTL(dev->base_addr));
+
+	tmp=readl(ETHER_TX_CTL(dev->base_addr));
+	tmp&=~ETHER_TX_CTL_TXEN_MSK;
+	tmp|=ETHER_TX_CTL_TXHALT_MSK;
+	writel(tmp,ETHER_TX_CTL(dev->base_addr));
+
+	/* Free up system resources */
+	free_irq(dev->irq,dev);
+	free_irq(2,dev);
+	iounmap(priv->dma_data);
+	iounmap((void*)dev->base_addr);
+	kfree(priv);
+
+	return 0;
+}
+
+
+static void ether00_get_ethernet_address(struct net_device* dev)
+{
+	struct mtd_info *mymtd=NULL;
+	int i;
+	size_t retlen;
+
+	/*
+	 * For the Epxa10 dev board (camelot), the ethernet MAC
+	 * address is of the form  00:aa:aa:00:xx:xx where
+	 * 00:aa:aa is the Altera vendor ID and xx:xx is the
+	 * last 2 bytes of the board serial number, as programmed
+	 * into the OTP area of the flash device on EBI1. If this
+	 * isn't an expa10 dev board, or there's no mtd support to
+	 * read the serial number from flash then we'll force the
+	 * use to set their own mac address using ifconfig.
+	 */
+
+#ifdef CONFIG_ARCH_CAMELOT
+#ifdef CONFIG_MTD
+	/* get the mtd_info structure for the first mtd device*/
+	for(i=0;i<MAX_MTD_DEVICES;i++){
+		mymtd=get_mtd_device(NULL,i);
+		if(!mymtd||!strcmp(mymtd->name,"EPXA10DB flash"))
+			break;
+	}
+
+	if(!mymtd || !mymtd->read_user_prot_reg){
+		printk(KERN_WARNING "%s: Failed to read MAC address from flash\n",dev->name);
+	}else{
+		mymtd->read_user_prot_reg(mymtd,2,1,&retlen,&dev->dev_addr[5]);
+		mymtd->read_user_prot_reg(mymtd,3,1,&retlen,&dev->dev_addr[4]);
+		dev->dev_addr[3]=0;
+		dev->dev_addr[2]=vendor_id[1];
+		dev->dev_addr[1]=vendor_id[0];
+		dev->dev_addr[0]=0;
+	}
+#else
+	printk(KERN_WARNING "%s: MTD support required to read MAC address from EPXA10 dev board\n", dev->name);
+#endif
+#endif
+
+	if (!is_valid_ether_addr(dev->dev_addr))
+		printk("%s: Invalid ethernet MAC address.  Please set using "
+			"ifconfig\n", dev->name);
+
+}
+
+static int ether00_init(struct net_device* dev)
+{
+
+	ether_setup(dev);
+
+	dev->open=ether00_open;
+	dev->stop=ether00_stop;
+	dev->set_multicast_list=ether00_set_multicast;
+	dev->hard_start_xmit=ether00_tx;
+	dev->get_stats=ether00_stats;
+	dev->tx_timeout=ether00_tx_timeout;
+	dev->watchdog_timeo=TX_TIMEOUT;
+
+	ether00_get_ethernet_address(dev);
+
+	SET_MODULE_OWNER(dev);
+	return 0;
+}
+
+
+struct net_device ether00_dev={
+	init:ether00_init,
+	name:"eth%d",
+};
+
+
+static void __exit ether00_cleanup_module(void)
+{
+	unregister_netdev(&ether00_dev);
+}
+module_exit(ether00_cleanup_module);
+
+
+static int __init ether00_mod_init(void)
+{
+	int result;
+
+	result=register_netdev(&ether00_dev);
+	if(result)
+		printk("Ether00: Error %i registering driver\n",result);
+	return result;
+}
+
+module_init(ether00_mod_init);
+
+

FUNET's LINUX-ADM group, linux-adm@nic.funet.fi
TCL-scripts by Sam Shen (who was at: slshen@lbl.gov)