net: Add Marvell Kirkwood gigabit ethernet driver

This patch adds a egiga driver for the Marvell Kirkwood SoC's.

Contributors:
Yotam Admon <yotam@marvell.com>
Michael Blostein <michaelbl@marvell.com

Reviewed-by: Ronen Shitrit <rshitrit@marvell.com>
Acked-by: Stefan Rose <sr@denx.de>
Signed-off-by: Prafulla Wadaskar <prafulla@marvell.com>
Signed-off-by: Ben Warren <biggerbadderben@gmail.com>

1 files changed, 664 insertions, 0 deletions

diff --git a/drivers/net/kirkwood_egiga.c b/drivers/net/kirkwood_egiga.c
new file mode 100644
index 0000000..b43bbf2
--- /dev/null
+++ b/drivers/net/kirkwood_egiga.c
@@ -0,0 +1,664 @@
+/*
+ * (C) Copyright 2009
+ * Marvell Semiconductor <www.marvell.com>
+ * Written-by: Prafulla Wadaskar <prafulla@marvell.com>
+ *
+ * (C) Copyright 2003
+ * Ingo Assmus <ingo.assmus@keymile.com>
+ *
+ * based on - Driver for MV64360X ethernet ports
+ * Copyright (C) 2002 rabeeh@galileo.co.il
+ *
+ * See file CREDITS for list of people who contributed to this
+ * project.
+ *
+ * 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., 51 Franklin Street, Fifth Floor, Boston,
+ * MA 02110-1301 USA
+ */
+
+#include <common.h>
+#include <net.h>
+#include <malloc.h>
+#include <miiphy.h>
+#include <asm/errno.h>
+#include <asm/types.h>
+#include <asm/byteorder.h>
+#include <asm/arch/kirkwood.h>
+#include "kirkwood_egiga.h"
+
+/*
+ * smi_reg_read - miiphy_read callback function.
+ *
+ * Returns 16bit phy register value, or 0xffff on error
+ */
+static int smi_reg_read(char *devname, u8 phy_adr, u8 reg_ofs, u16 * data)
+{
+	struct eth_device *dev = eth_get_dev_by_name(devname);
+	struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	struct kwgbe_registers *regs = dkwgbe->regs;
+	u32 smi_reg;
+	volatile u32 timeout;
+
+	/* Phyadr read request */
+	if (phy_adr == 0xEE && reg_ofs == 0xEE) {
+		/* */
+		*data = (u16) (KWGBEREG_RD(regs->phyadr) & PHYADR_MASK);
+		return 0;
+	}
+	/* check parameters */
+	if (phy_adr > PHYADR_MASK) {
+		printf("Err..(%s) Invalid PHY address %d\n",
+			__FUNCTION__, phy_adr);
+		return -EFAULT;
+	}
+	if (reg_ofs > PHYREG_MASK) {
+		printf("Err..(%s) Invalid register offset %d\n",
+			__FUNCTION__, reg_ofs);
+		return -EFAULT;
+	}
+
+	timeout = KWGBE_PHY_SMI_TIMEOUT;
+	/* wait till the SMI is not busy */
+	do {
+		/* read smi register */
+		smi_reg = KWGBEREG_RD(regs->smi);
+		if (timeout-- == 0) {
+			printf("Err..(%s) SMI busy timeout\n", __FUNCTION__);
+			return -EFAULT;
+		}
+	} while (smi_reg & KWGBE_PHY_SMI_BUSY_MASK);
+
+	/* fill the phy address and regiser offset and read opcode */
+	smi_reg = (phy_adr << KWGBE_PHY_SMI_DEV_ADDR_OFFS)
+		| (reg_ofs << KWGBE_SMI_REG_ADDR_OFFS)
+		| KWGBE_PHY_SMI_OPCODE_READ;
+
+	/* write the smi register */
+	KWGBEREG_WR(regs->smi, smi_reg);
+
+	/*wait till read value is ready */
+	timeout = KWGBE_PHY_SMI_TIMEOUT;
+
+	do {
+		/* read smi register */
+		smi_reg = KWGBEREG_RD(regs->smi);
+		if (timeout-- == 0) {
+			printf("Err..(%s) SMI read ready timeout\n",
+				__FUNCTION__);
+			return -EFAULT;
+		}
+	} while (!(smi_reg & KWGBE_PHY_SMI_READ_VALID_MASK));
+
+	/* Wait for the data to update in the SMI register */
+	for (timeout = 0; timeout < KWGBE_PHY_SMI_TIMEOUT; timeout++) ;
+
+	*data = (u16) (KWGBEREG_RD(regs->smi) & KWGBE_PHY_SMI_DATA_MASK);
+
+	debug("%s:(adr %d, off %d) value= %04x\n", __FUNCTION__, phy_adr,
+		reg_ofs, *data);
+
+	return 0;
+}
+
+/*
+ * smi_reg_write - imiiphy_write callback function.
+ *
+ * Returns 0 if write succeed, -EINVAL on bad parameters
+ * -ETIME on timeout
+ */
+static int smi_reg_write(char *devname, u8 phy_adr, u8 reg_ofs, u16 data)
+{
+	struct eth_device *dev = eth_get_dev_by_name(devname);
+	struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	struct kwgbe_registers *regs = dkwgbe->regs;
+	u32 smi_reg;
+	volatile u32 timeout;
+
+	/* Phyadr write request*/
+	if (phy_adr == 0xEE && reg_ofs == 0xEE) {
+		KWGBEREG_WR(regs->phyadr, data);
+		return 0;
+	}
+
+	/* check parameters */
+	if (phy_adr > PHYADR_MASK) {
+		printf("Err..(%s) Invalid phy address\n", __FUNCTION__);
+		return -EINVAL;
+	}
+	if (reg_ofs > PHYREG_MASK) {
+		printf("Err..(%s) Invalid register offset\n", __FUNCTION__);
+		return -EINVAL;
+	}
+
+	/* wait till the SMI is not busy */
+	timeout = KWGBE_PHY_SMI_TIMEOUT;
+	do {
+		/* read smi register */
+		smi_reg = KWGBEREG_RD(regs->smi);
+		if (timeout-- == 0) {
+			printf("Err..(%s) SMI busy timeout\n", __FUNCTION__);
+			return -ETIME;
+		}
+	} while (smi_reg & KWGBE_PHY_SMI_BUSY_MASK);
+
+	/* fill the phy addr and reg offset and write opcode and data */
+	smi_reg = (data << KWGBE_PHY_SMI_DATA_OFFS);
+	smi_reg |= (phy_adr << KWGBE_PHY_SMI_DEV_ADDR_OFFS)
+		| (reg_ofs << KWGBE_SMI_REG_ADDR_OFFS);
+	smi_reg &= ~KWGBE_PHY_SMI_OPCODE_READ;
+
+	/* write the smi register */
+	KWGBEREG_WR(regs->smi, smi_reg);
+
+	return 0;
+}
+
+/* Stop and checks all queues */
+static void stop_queue(u32 * qreg)
+{
+	u32 reg_data;
+
+	reg_data = readl(qreg);
+
+	if (reg_data & 0xFF) {
+		/* Issue stop command for active channels only */
+		writel((reg_data << 8), qreg);
+
+		/* Wait for all queue activity to terminate. */
+		do {
+			/*
+			 * Check port cause register that all queues
+			 * are stopped
+			 */
+			reg_data = readl(qreg);
+		}
+		while (reg_data & 0xFF);
+	}
+}
+
+/*
+ * set_access_control - Config address decode parameters for Ethernet unit
+ *
+ * This function configures the address decode parameters for the Gigabit
+ * Ethernet Controller according the given parameters struct.
+ *
+ * @regs	Register struct pointer.
+ * @param	Address decode parameter struct.
+ */
+static void set_access_control(struct kwgbe_registers *regs,
+				struct kwgbe_winparam *param)
+{
+	u32 access_prot_reg;
+
+	/* Set access control register */
+	access_prot_reg = KWGBEREG_RD(regs->epap);
+	/* clear window permission */
+	access_prot_reg &= (~(3 << (param->win * 2)));
+	access_prot_reg |= (param->access_ctrl << (param->win * 2));
+	KWGBEREG_WR(regs->epap, access_prot_reg);
+
+	/* Set window Size reg (SR) */
+	KWGBEREG_WR(regs->barsz[param->win].size,
+			(((param->size / 0x10000) - 1) << 16));
+
+	/* Set window Base address reg (BA) */
+	KWGBEREG_WR(regs->barsz[param->win].bar,
+			(param->target | param->attrib | param->base_addr));
+	/* High address remap reg (HARR) */
+	if (param->win < 4)
+		KWGBEREG_WR(regs->ha_remap[param->win], param->high_addr);
+
+	/* Base address enable reg (BARER) */
+	if (param->enable == 1)
+		KWGBEREG_BITS_RESET(regs->bare, (1 << param->win));
+	else
+		KWGBEREG_BITS_SET(regs->bare, (1 << param->win));
+}
+
+static void set_dram_access(struct kwgbe_registers *regs)
+{
+	struct kwgbe_winparam win_param;
+	int i;
+
+	for (i = 0; i < CONFIG_NR_DRAM_BANKS; i++) {
+		/* Set access parameters for DRAM bank i */
+		win_param.win = i;	/* Use Ethernet window i */
+		/* Window target - DDR */
+		win_param.target = KWGBE_TARGET_DRAM;
+		/* Enable full access */
+		win_param.access_ctrl = EWIN_ACCESS_FULL;
+		win_param.high_addr = 0;
+		/* Get bank base */
+		win_param.base_addr = kw_sdram_bar(i);
+		win_param.size = kw_sdram_bs(i);	/* Get bank size */
+		if (win_param.size == 0)
+			win_param.enable = 0;
+		else
+			win_param.enable = 1;	/* Enable the access */
+
+		/* Enable DRAM bank */
+		switch (i) {
+		case 0:
+			win_param.attrib = EBAR_DRAM_CS0;
+			break;
+		case 1:
+			win_param.attrib = EBAR_DRAM_CS1;
+			break;
+		case 2:
+			win_param.attrib = EBAR_DRAM_CS2;
+			break;
+		case 3:
+			win_param.attrib = EBAR_DRAM_CS3;
+			break;
+		default:
+			/* invalide bank, disable access */
+			win_param.enable = 0;
+			win_param.attrib = 0;
+			break;
+		}
+		/* Set the access control for address window(EPAPR) RD/WR */
+		set_access_control(regs, &win_param);
+	}
+}
+
+/*
+ * port_init_mac_tables - Clear all entrance in the UC, SMC and OMC tables
+ *
+ * Go through all the DA filter tables (Unicast, Special Multicast & Other
+ * Multicast) and set each entry to 0.
+ */
+static void port_init_mac_tables(struct kwgbe_registers *regs)
+{
+	int table_index;
+
+	/* Clear DA filter unicast table (Ex_dFUT) */
+	for (table_index = 0; table_index < 4; ++table_index)
+		KWGBEREG_WR(regs->dfut[table_index], 0);
+
+	for (table_index = 0; table_index < 64; ++table_index) {
+		/* Clear DA filter special multicast table (Ex_dFSMT) */
+		KWGBEREG_WR(regs->dfsmt[table_index], 0);
+		/* Clear DA filter other multicast table (Ex_dFOMT) */
+		KWGBEREG_WR(regs->dfomt[table_index], 0);
+	}
+}
+
+/*
+ * port_uc_addr - This function Set the port unicast address table
+ *
+ * This function locates the proper entry in the Unicast table for the
+ * specified MAC nibble and sets its properties according to function
+ * parameters.
+ * This function add/removes MAC addresses from the port unicast address
+ * table.
+ *
+ * @uc_nibble	Unicast MAC Address last nibble.
+ * @option      0 = Add, 1 = remove address.
+ *
+ * RETURN: 1 if output succeeded. 0 if option parameter is invalid.
+ */
+static int port_uc_addr(struct kwgbe_registers *regs, u8 uc_nibble,
+			int option)
+{
+	u32 unicast_reg;
+	u32 tbl_offset;
+	u32 reg_offset;
+
+	/* Locate the Unicast table entry */
+	uc_nibble = (0xf & uc_nibble);
+	/* Register offset from unicast table base */
+	tbl_offset = (uc_nibble / 4);
+	/* Entry offset within the above register */
+	reg_offset = uc_nibble % 4;
+
+	switch (option) {
+	case REJECT_MAC_ADDR:
+		/*
+		 * Clear accepts frame bit at specified unicast
+		 * DA table entry
+		 */
+		unicast_reg = KWGBEREG_RD(regs->dfut[tbl_offset]);
+		unicast_reg &= (0xFF << (8 * reg_offset));
+		KWGBEREG_WR(regs->dfut[tbl_offset], unicast_reg);
+		break;
+	case ACCEPT_MAC_ADDR:
+		/* Set accepts frame bit at unicast DA filter table entry */
+		unicast_reg = KWGBEREG_RD(regs->dfut[tbl_offset]);
+		unicast_reg &= (0xFF << (8 * reg_offset));
+		unicast_reg |= ((0x01 | (RXUQ << 1)) << (8 * reg_offset));
+		KWGBEREG_WR(regs->dfut[tbl_offset], unicast_reg);
+		break;
+	default:
+		return 0;
+	}
+	return 1;
+}
+
+/*
+ * port_uc_addr_set - This function Set the port Unicast address.
+ */
+static void port_uc_addr_set(struct kwgbe_registers *regs, u8 * p_addr)
+{
+	u32 mac_h;
+	u32 mac_l;
+
+	mac_l = (p_addr[4] << 8) | (p_addr[5]);
+	mac_h = (p_addr[0] << 24) | (p_addr[1] << 16) | (p_addr[2] << 8) |
+		(p_addr[3] << 0);
+
+	KWGBEREG_WR(regs->macal, mac_l);
+	KWGBEREG_WR(regs->macah, mac_h);
+
+	/* Accept frames of this address */
+	port_uc_addr(regs, p_addr[5], ACCEPT_MAC_ADDR);
+}
+
+/*
+ * kwgbe_init_rx_desc_ring - Curve a Rx chain desc list and buffer in memory.
+ */
+static void kwgbe_init_rx_desc_ring(struct kwgbe_device *dkwgbe)
+{
+	volatile struct kwgbe_rxdesc *p_rx_desc;
+	int i;
+
+	/* initialize the Rx descriptors ring */
+	p_rx_desc = dkwgbe->p_rxdesc;
+	for (i = 0; i < RINGSZ; i++) {
+		p_rx_desc->cmd_sts =
+			KWGBE_BUFFER_OWNED_BY_DMA | KWGBE_RX_EN_INTERRUPT;
+		p_rx_desc->buf_size = PKTSIZE_ALIGN;
+		p_rx_desc->byte_cnt = 0;
+		p_rx_desc->buf_ptr = dkwgbe->p_rxbuf + i * PKTSIZE_ALIGN;
+		if (i == (RINGSZ - 1))
+			p_rx_desc->nxtdesc_p = dkwgbe->p_rxdesc;
+		else {
+			p_rx_desc->nxtdesc_p = (struct kwgbe_rxdesc *)
+				((u32) p_rx_desc + KW_RXQ_DESC_ALIGNED_SIZE);
+			p_rx_desc = p_rx_desc->nxtdesc_p;
+		}
+	}
+	dkwgbe->p_rxdesc_curr = dkwgbe->p_rxdesc;
+}
+
+static int kwgbe_init(struct eth_device *dev)
+{
+	struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	struct kwgbe_registers *regs = dkwgbe->regs;
+
+	/* setup RX rings */
+	kwgbe_init_rx_desc_ring(dkwgbe);
+
+	/* Clear the ethernet port interrupts */
+	KWGBEREG_WR(regs->ic, 0);
+	KWGBEREG_WR(regs->ice, 0);
+	/* Unmask RX buffer and TX end interrupt */
+	KWGBEREG_WR(regs->pim, INT_CAUSE_UNMASK_ALL);
+	/* Unmask phy and link status changes interrupts */
+	KWGBEREG_WR(regs->peim, INT_CAUSE_UNMASK_ALL_EXT);
+
+	set_dram_access(regs);
+	port_init_mac_tables(regs);
+	port_uc_addr_set(regs, dkwgbe->dev.enetaddr);
+
+	/* Assign port configuration and command. */
+	KWGBEREG_WR(regs->pxc, PRT_CFG_VAL);
+	KWGBEREG_WR(regs->pxcx, PORT_CFG_EXTEND_VALUE);
+	KWGBEREG_WR(regs->psc0, PORT_SERIAL_CONTROL_VALUE);
+	/* Disable port initially */
+	KWGBEREG_BITS_SET(regs->psc0, KWGBE_SERIAL_PORT_EN);
+
+	/* Assign port SDMA configuration */
+	KWGBEREG_WR(regs->sdc, PORT_SDMA_CFG_VALUE);
+	KWGBEREG_WR(regs->tqx[0].qxttbc, QTKNBKT_DEF_VAL);
+	KWGBEREG_WR(regs->tqx[0].tqxtbc, (QMTBS_DEF_VAL << 16) | QTKNRT_DEF_VAL);
+	/* Turn off the port/RXUQ bandwidth limitation */
+	KWGBEREG_WR(regs->pmtu, 0);
+
+	/* Set maximum receive buffer to 9700 bytes */
+	KWGBEREG_WR(regs->psc0,	KWGBE_MAX_RX_PACKET_9700BYTE
+			| (KWGBEREG_RD(regs->psc0) & MRU_MASK));
+
+	/*
+	 * Set ethernet MTU for leaky bucket mechanism to 0 - this will
+	 * disable the leaky bucket mechanism .
+	 */
+	KWGBEREG_WR(regs->pmtu, 0);
+
+	/* Assignment of Rx CRDB of given RXUQ */
+	KWGBEREG_WR(regs->rxcdp[RXUQ].rxcdp, (u32) dkwgbe->p_rxdesc_curr);
+	/* Enable port Rx. */
+	KWGBEREG_WR(regs->rqc, (1 << RXUQ));
+
+#if (defined (CONFIG_MII) || defined (CONFIG_CMD_MII)) \
+	 && defined (CONFIG_SYS_FAULT_ECHO_LINK_DOWN)
+	u16 phyadr;
+	miiphy_read(dev->name, 0xEE, 0xEE, &phyadr);
+	if (!miiphy_link(dev->name, phyadr)) {
+		printf("%s: No link on %s\n", __FUNCTION__, dev->name);
+		return -1;
+	}
+#endif
+	return 0;
+}
+
+static int kwgbe_halt(struct eth_device *dev)
+{
+	struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	struct kwgbe_registers *regs = dkwgbe->regs;
+
+	/* Disable all gigE address decoder */
+	KWGBEREG_WR(regs->bare, 0x3f);
+
+	stop_queue(&regs->tqc);
+	stop_queue(&regs->rqc);
+
+	/* Enable port */
+	KWGBEREG_BITS_RESET(regs->psc0, KWGBE_SERIAL_PORT_EN);
+	/* Set port is not reset */
+	KWGBEREG_BITS_RESET(regs->psc1, 1 << 4);
+#ifdef CONFIG_SYS_MII_MODE
+	/* Set MMI interface up */
+	KWGBEREG_BITS_RESET(regs->psc1, 1 << 3);
+#endif
+	/* Disable & mask ethernet port interrupts */
+	KWGBEREG_WR(regs->ic, 0);
+	KWGBEREG_WR(regs->ice, 0);
+	KWGBEREG_WR(regs->pim, 0);
+	KWGBEREG_WR(regs->peim, 0);
+
+	return 0;
+}
+
+static int kwgbe_send(struct eth_device *dev, volatile void *dataptr,
+		      int datasize)
+{
+	struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	struct kwgbe_registers *regs = dkwgbe->regs;
+	struct kwgbe_txdesc *p_txdesc = dkwgbe->p_txdesc;
+
+	if ((u32) dataptr & 0x07) {
+		printf("Err..(%s) xmit dataptr not 64bit aligned\n",
+			__FUNCTION__);
+		return -1;
+	}
+	p_txdesc->cmd_sts = KWGBE_ZERO_PADDING | KWGBE_GEN_CRC;
+	p_txdesc->cmd_sts |= KWGBE_TX_FIRST_DESC | KWGBE_TX_LAST_DESC;
+	p_txdesc->cmd_sts |= KWGBE_BUFFER_OWNED_BY_DMA;
+	p_txdesc->cmd_sts |= KWGBE_TX_EN_INTERRUPT;
+	p_txdesc->buf_ptr = (u8 *) dataptr;
+	p_txdesc->byte_cnt = datasize;
+
+	/* Apply send command using zeroth RXUQ */
+	KWGBEREG_WR(regs->tcqdp[TXUQ], (u32) p_txdesc);
+	KWGBEREG_WR(regs->tqc, (1 << TXUQ));
+
+	/*
+	 * wait for packet xmit completion
+	 */
+	while (p_txdesc->cmd_sts & KWGBE_BUFFER_OWNED_BY_DMA) {
+		/* return fail if error is detected */
+		if (p_txdesc->cmd_sts & (KWGBE_UR_ERROR | KWGBE_RL_ERROR)) {
+			printf("Err..(%s) in xmit packet\n", __FUNCTION__);
+			return -1;
+		}
+	};
+	return 0;
+}
+
+static int kwgbe_recv(struct eth_device *dev)
+{
+	volatile struct kwgbe_device *dkwgbe = to_dkwgbe(dev);
+	volatile struct kwgbe_rxdesc *p_rxdesc_curr = dkwgbe->p_rxdesc_curr;
+	volatile u32 timeout = 0;
+
+	/* wait untill rx packet available or timeout */
+	do {
+		if (timeout < KWGBE_PHY_SMI_TIMEOUT)
+			timeout++;
+		else {
+			debug("%s time out...\n", __FUNCTION__);
+			return -1;
+		}
+	} while (p_rxdesc_curr->cmd_sts & KWGBE_BUFFER_OWNED_BY_DMA);
+
+	if (p_rxdesc_curr->byte_cnt != 0) {
+		debug("%s: Received %d byte Packet @ 0x%x (cmd_sts= %08x)\n",
+			__FUNCTION__, (u32) p_rxdesc_curr->byte_cnt,
+			(u32) p_rxdesc_curr->buf_ptr,
+			(u32) p_rxdesc_curr->cmd_sts);
+	}
+
+	/*
+	 * In case received a packet without first/last bits on
+	 * OR the error summary bit is on,
+	 * the packets needs to be dropeed.
+	 */
+	if ((p_rxdesc_curr->cmd_sts &
+		(KWGBE_RX_FIRST_DESC | KWGBE_RX_LAST_DESC))
+		!= (KWGBE_RX_FIRST_DESC | KWGBE_RX_LAST_DESC)) {
+
+		printf("Err..(%s) Dropping packet spread on"
+			" multiple descriptors\n", __FUNCTION__);
+
+	} else if (p_rxdesc_curr->cmd_sts & KWGBE_ERROR_SUMMARY) {
+
+		printf("Err..(%s) Dropping packet with errors\n",
+			__FUNCTION__);
+
+	} else {
+		/* !!! call higher layer processing */
+		debug("%s: Sending Received packet to"
+			" upper layer (NetReceive)\n", __FUNCTION__);
+
+		/* let the upper layer handle the packet */
+		NetReceive((p_rxdesc_curr->buf_ptr + RX_BUF_OFFSET),
+			(int)(p_rxdesc_curr->byte_cnt - RX_BUF_OFFSET));
+	}
+	/*
+	 * free these descriptors and point next in the ring
+	 */
+	p_rxdesc_curr->cmd_sts =
+		KWGBE_BUFFER_OWNED_BY_DMA | KWGBE_RX_EN_INTERRUPT;
+	p_rxdesc_curr->buf_size = PKTSIZE_ALIGN;
+	p_rxdesc_curr->byte_cnt = 0;
+
+	dkwgbe->p_rxdesc_curr = p_rxdesc_curr->nxtdesc_p;
+	return 0;
+}
+
+int kirkwood_egiga_initialize(bd_t * bis)
+{
+	struct kwgbe_device *dkwgbe;
+	struct eth_device *dev;
+	int devnum;
+	char *s, buf[NAMESIZE * 2];
+	u8 used_ports[MAX_KWGBE_DEVS] = CONFIG_KIRKWOOD_EGIGA_PORTS;
+
+	for (devnum = 0; devnum < MAX_KWGBE_DEVS; devnum++) {
+		/*skip if port is configured not to use */
+		if (used_ports[devnum] == 0)
+			continue;
+
+		if (!(dkwgbe = malloc(sizeof(struct kwgbe_device))))
+			goto error1;
+
+		memset(dkwgbe, 0, sizeof(struct kwgbe_device));
+
+		if (!(dkwgbe->p_rxdesc =
+		      (struct kwgbe_rxdesc *)memalign(PKTALIGN,
+						KW_RXQ_DESC_ALIGNED_SIZE
+						* RINGSZ + 1)))
+			goto error2;
+
+		if (!(dkwgbe->p_rxbuf = (u8 *) memalign(PKTALIGN, RINGSZ
+							* PKTSIZE_ALIGN + 1)))
+			goto error3;
+
+		if (!(dkwgbe->p_txdesc = (struct kwgbe_txdesc *)
+		      memalign(PKTALIGN, sizeof(struct kwgbe_txdesc) + 1))) {
+			free(dkwgbe->p_rxbuf);
+		      error3:
+			free(dkwgbe->p_rxdesc);
+		      error2:
+			free(dkwgbe);
+		      error1:
+			printf("Err.. %s Failed to allocate memory\n",
+				__FUNCTION__);
+			return -1;
+		}
+
+		dev = &dkwgbe->dev;
+
+		/* must be less than NAMESIZE (16) */
+		sprintf(dev->name, "egiga%d", devnum);
+
+		/* Extract the MAC address from the environment */
+		switch (devnum) {
+		case 0:
+			dkwgbe->regs = (void *)KW_EGIGA0_BASE;
+			s = "ethaddr";
+			break;
+		case 1:
+			dkwgbe->regs = (void *)KW_EGIGA1_BASE;
+			s = "eth1addr";
+			break;
+		default:	/* this should never happen */
+			printf("Err..(%s) Invalid device number %d\n",
+				__FUNCTION__, devnum);
+			return -1;
+		}
+
+		while (!eth_getenv_enetaddr(s, dev->enetaddr)) {
+			/* Generate Ramdom MAC addresses if not set */
+			sprintf(buf, "00:50:43:%02x:%02x:%02x",
+				get_random_hex(), get_random_hex(),
+				get_random_hex());
+			setenv(s, buf);
+		}
+
+		dev->init = (void *)kwgbe_init;
+		dev->halt = (void *)kwgbe_halt;
+		dev->send = (void *)kwgbe_send;
+		dev->recv = (void *)kwgbe_recv;
+
+		eth_register(dev);
+
+#if defined(CONFIG_MII) || defined(CONFIG_CMD_MII)
+		miiphy_register(dev->name, smi_reg_read, smi_reg_write);
+		/* Set phy address of the port */
+		miiphy_write(dev->name, 0xEE, 0xEE, PHY_BASE_ADR + devnum);
+#endif
+	}
+	return 0;