strip the kernel version suffix from target directories, except for brcm-2.4 (the -2.4 will be included in the board name here). CONFIG_LINUX_<ver>_<board> becomes CONFIG_TARGET_<board>, same for profiles.

git-svn-id: svn://svn.openwrt.org/openwrt/trunk@8653 3c298f89-4303-0410-b956-a3cf2f4a3e73

1 files changed, 876 insertions, 0 deletions

diff --git a/target/linux/amazon/files/drivers/net/amazon_sw.c b/target/linux/amazon/files/drivers/net/amazon_sw.c
new file mode 100644
index 000000000..d19db6e36
--- /dev/null
+++ b/target/linux/amazon/files/drivers/net/amazon_sw.c
@@ -0,0 +1,876 @@
+/*
+ *   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.
+ */
+//-----------------------------------------------------------------------
+/*
+ * Description:
+ *	Driver for Infineon Amazon 3 port switch
+ */
+//-----------------------------------------------------------------------
+/* Author:	Wu Qi Ming[Qi-Ming.Wu@infineon.com]
+ * Created:	7-April-2004
+ */
+//-----------------------------------------------------------------------
+/* History
+ * Changed on: Jun 28, 2004
+ * Changed by: peng.liu@infineon.com
+ * Reason:	add hardware flow control (HFC) (CONFIG_NET_HW_FLOWCONTROL)
+ *
+ * Changed on: Apr 6, 2005
+ * Changed by: mars.lin@infineon.com
+ * Reason    : supoort port identification
+ */
+
+
+// copyright 2004-2005 infineon.com
+
+// copyright 2007 john crispin <blogic@openwrt.org>
+// copyright 2007 felix fietkau <nbd@openwrt.org>
+
+
+// TODO
+// 		port vlan code from bcrm target... the tawainese code was scrapped due to crappyness
+// 		check all the mmi reg settings and possibly document them better
+// 		verify the ethtool code
+//		remove the while(1) stuff
+// 		further clean up and rework ... but it works for now
+//		check the mode[]=bridge stuff
+//		verify that the ethaddr can be set from u-boot
+
+
+#ifndef __KERNEL__
+#define __KERNEL__
+#endif
+
+
+#if defined(CONFIG_MODVERSIONS) && !defined(MODVERSIONS)
+#define MODVERSIONS
+#endif
+
+#if defined(MODVERSIONS) && !defined(__GENKSYMS__)
+#include <linux/modversions.h>
+#endif
+
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/slab.h>
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/interrupt.h>
+#include <linux/mii.h>
+#include <asm/uaccess.h>
+#include <linux/in.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/ip.h>
+#include <linux/tcp.h>
+#include <linux/skbuff.h>
+#include <linux/in6.h>
+#include <linux/proc_fs.h>
+#include <linux/mm.h>
+#include <linux/ethtool.h>
+#include <asm/checksum.h>
+#include <linux/init.h>
+
+#include <asm/amazon/amazon.h>
+#include <asm/amazon/amazon_dma.h>
+#include <asm/amazon/amazon_sw.h>
+
+// how many mii ports are there ?
+#define AMAZON_SW_INT_NO 2
+
+#define ETHERNET_PACKET_DMA_BUFFER_SIZE 1536
+
+/***************************************** Module Parameters *************************************/
+char mode[] = "bridge";
+module_param_array(mode, charp, NULL, 0);
+
+static int timeout = 1 * HZ;
+module_param(timeout, int, 0);
+
+int switch_init(struct net_device *dev);
+void switch_tx_timeout(struct net_device *dev);
+
+struct net_device switch_devs[2] = {
+	{init:switch_init,},
+	{init:switch_init,}
+};
+
+int add_mac_table_entry(u64 entry_value)
+{
+	int i;
+	u32 data1, data2;
+
+	AMAZON_SW_REG32(AMAZON_SW_ARL_CTL) = ~7;
+
+	for (i = 0; i < 32; i++) {
+		AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) = 0x80000000 | 0x20 | i;
+		while (AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) & (0x80000000)) {};
+		data1 = AMAZON_SW_REG32(AMAZON_SW_DATA1);
+		data2 = AMAZON_SW_REG32(AMAZON_SW_DATA2);
+		if ((data1 & (0x00700000)) != 0x00700000)
+			continue;
+		AMAZON_SW_REG32(AMAZON_SW_DATA1) = (u32) (entry_value >> 32);
+		AMAZON_SW_REG32(AMAZON_SW_DATA2) = (u32) entry_value & 0xffffffff;
+		AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) = 0xc0000020 | i;
+		while (AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) & (0x80000000)) {};
+		break;
+	}
+	AMAZON_SW_REG32(AMAZON_SW_ARL_CTL) |= 7;
+	if (i >= 32)
+		return -1;
+	return OK;
+}
+
+u64 read_mac_table_entry(int index)
+{
+	u32 data1, data2;
+	u64 value;
+	AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) = 0x80000000 | 0x20 | index;
+	while (AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) & (0x80000000)) {};
+	data1 = AMAZON_SW_REG32(AMAZON_SW_DATA1) & 0xffffff;
+	data2 = AMAZON_SW_REG32(AMAZON_SW_DATA2);
+	value = (u64) data1 << 32 | (u64) data2;
+	return value;
+}
+
+int write_mac_table_entry(int index, u64 value)
+{
+	u32 data1, data2;
+	data1 = (u32) (value >> 32);
+	data2 = (u32) value & 0xffffffff;
+	AMAZON_SW_REG32(AMAZON_SW_DATA1) = data1;
+	AMAZON_SW_REG32(AMAZON_SW_DATA2) = data2;
+	AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) = 0xc0000020 | index;
+	while (AMAZON_SW_REG32(AMAZON_SW_CPU_ACTL) & (0x80000000)) {};
+	return OK;
+}
+
+u32 get_mdio_reg(int phy_addr, int reg_num)
+{
+	u32 value;
+	AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) = (3 << 30) | ((phy_addr & 0x1f) << 21) | ((reg_num & 0x1f) << 16);
+	while (AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) & (1 << 31)) {};
+	value = AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) & 0xffff;
+	return value;
+}
+
+int set_mdio_reg(int phy_addr, int reg_num, u32 value)
+{
+	AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) = (2 << 30) | ((phy_addr & 0x1f) << 21) | ((reg_num & 0x1f) << 16) | (value & 0xffff);
+	while (AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) & (1 << 31)) {};
+	return OK;
+}
+
+int auto_negotiate(int phy_addr)
+{
+	u32 value = 0;
+	value = get_mdio_reg(phy_addr, MDIO_BASE_CONTROL_REG);
+	set_mdio_reg(phy_addr, MDIO_BASE_CONTROL_REG, (value | RESTART_AUTO_NEGOTIATION | AUTO_NEGOTIATION_ENABLE | PHY_RESET));
+	return OK;
+}
+
+/*
+     In this version of switch driver, we split the dma channels for the switch.
+     2 for port0 and 2 for port1. So that we can do internal bridging if necessary.
+     In switch mode, packets coming in from port0 or port1 is able to do Destination 
+     address lookup. Packets coming from port0 with destination address of port1 should 
+     not go to pmac again. The switch hardware should be able to do the switch in the hard 
+     ware level. Packets coming from the pmac should not do the DA look up in that the
+     desination is already known for the kernel. It only needs to go to the correct NIC to 
+     find its way out.
+  */
+int amazon_sw_chip_init(void)
+{
+	u32 tmp1;
+	int i = 0;
+
+	/* Aging tick select: 5mins */
+	tmp1 = 0xa0;
+	if (strcmp(mode, "bridge") == 0) {
+		// bridge mode, set militarised mode to 1, no learning!
+		tmp1 |= 0xC00;
+	} else {
+		// enable learning for P0 and P1,
+		tmp1 |= 3;
+	}
+
+	/* unknown broadcast/multicast/unicast to all ports */
+	AMAZON_SW_REG32(AMAZON_SW_UN_DEST) = 0x1ff;
+
+	AMAZON_SW_REG32(AMAZON_SW_ARL_CTL) = tmp1;
+
+	/* OCS:1 set OCS bit, split the two NIC in rx direction EDL:1 (enable DA lookup) */
+#if defined(CONFIG_IFX_NFEXT_AMAZON_SWITCH_PHYPORT) || defined(CONFIG_IFX_NFEXT_AMAZON_SWITCH_PHYPORT_MODULE)
+	AMAZON_SW_REG32(AMAZON_SW_P2_PCTL) = 0x700;
+#else
+	AMAZON_SW_REG32(AMAZON_SW_P2_PCTL) = 0x401;
+#endif
+
+	/* EPC: 1 split the two NIC in tx direction CRC is generated */
+	AMAZON_SW_REG32(AMAZON_SW_P2_CTL) = 0x6;
+
+	// for bi-directional 
+	AMAZON_SW_REG32(AMAZON_SW_P0_WM) = 0x14141412;
+	AMAZON_SW_REG32(AMAZON_SW_P1_WM) = 0x14141412;
+	AMAZON_SW_REG32(AMAZON_SW_P2_WM) = 0x28282826;
+	AMAZON_SW_REG32(AMAZON_SW_GBL_WM) = 0x0;
+
+	AMAZON_SW_REG32(AMAZON_CGU_PLL0SR) = (AMAZON_SW_REG32(AMAZON_CGU_PLL0SR)) | 0x58000000;
+	// clock for PHY
+	AMAZON_SW_REG32(AMAZON_CGU_IFCCR) =	(AMAZON_SW_REG32(AMAZON_CGU_IFCCR)) | 0x80000004;
+	// enable power for PHY
+	AMAZON_SW_REG32(AMAZON_PMU_PWDCR) = (AMAZON_SW_REG32(AMAZON_PMU_PWDCR)) | AMAZON_PMU_PWDCR_EPHY;
+	// set reverse MII, enable MDIO statemachine
+	AMAZON_SW_REG32(AMAZON_SW_MDIO_CFG) = 0x800027bf;
+	while (1)
+		if (((AMAZON_SW_REG32(AMAZON_SW_MDIO_CFG)) & 0x80000000) == 0)
+			break;
+	AMAZON_SW_REG32(AMAZON_SW_EPHY) = 0xff;
+
+	// auto negotiation
+	AMAZON_SW_REG32(AMAZON_SW_MDIO_ACC) = 0x83e08000;
+	auto_negotiate(0x1f);
+
+	/* enable all ports */
+	AMAZON_SW_REG32(AMAZON_SW_PS_CTL) = 0x7;
+	for (i = 0; i < 32; i++)
+		write_mac_table_entry(i, 1 << 50);
+	return 0;
+}
+
+static unsigned char my_ethaddr[MAX_ADDR_LEN];
+/* need to get the ether addr from u-boot */
+static int __init ethaddr_setup(char *line)
+{
+	char *ep;
+	int i;
+
+	memset(my_ethaddr, 0, MAX_ADDR_LEN);
+	for (i = 0; i < 6; i++) {
+		my_ethaddr[i] = line ? simple_strtoul(line, &ep, 16) : 0;
+		if (line)
+			line = (*ep) ? ep + 1 : ep;
+	}
+	printk("mac address %2x-%2x-%2x-%2x-%2x-%2x \n", my_ethaddr[0], my_ethaddr[1], my_ethaddr[2], my_ethaddr[3], my_ethaddr[4], my_ethaddr[5]);
+	return 0;
+}
+
+__setup("ethaddr=", ethaddr_setup);
+
+static void open_rx_dma(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	struct dma_device_info *dma_dev = priv->dma_device;
+	int i;
+
+	for (i = 0; i < dma_dev->num_rx_chan; i++)
+		dma_dev->rx_chan[i].control = 1;
+	dma_device_update_rx(dma_dev);
+}
+
+#ifdef CONFIG_NET_HW_FLOWCONTROL
+static void close_rx_dma(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	struct dma_device_info *dma_dev = priv->dma_device;
+	int i;
+
+	for (i = 0; i < dma_dev->num_rx_chan; i++)
+		dma_dev->rx_chan[i].control = 0;
+	dma_device_update_rx(dma_dev);
+}
+
+void amazon_xon(struct net_device *dev)
+{
+	unsigned long flag;
+	local_irq_save(flag);
+	open_rx_dma(dev);
+	local_irq_restore(flag);
+}
+#endif
+
+int switch_open(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	if (!strcmp(dev->name, "eth1")) {
+		priv->mdio_phy_addr = PHY0_ADDR;
+	}
+	open_rx_dma(dev);			
+
+#ifdef CONFIG_NET_HW_FLOWCONTROL
+	if ((priv->fc_bit = netdev_register_fc(dev, amazon_xon)) == 0) {
+		printk("Hardware Flow Control register fails\n");
+	}
+#endif
+
+	netif_start_queue(dev);
+	return OK;
+}
+
+int switch_release(struct net_device *dev)
+{
+	int i;
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	struct dma_device_info *dma_dev = priv->dma_device;
+
+	for (i = 0; i < dma_dev->num_tx_chan; i++)
+		dma_dev->tx_chan[i].control = 0;
+	for (i = 0; i < dma_dev->num_rx_chan; i++)
+		dma_dev->rx_chan[i].control = 0;
+
+	dma_device_update(dma_dev);
+
+#ifdef CONFIG_NET_HW_FLOWCONTROL
+	if (priv->fc_bit) {
+		netdev_unregister_fc(priv->fc_bit);
+	}
+#endif
+	netif_stop_queue(dev);
+
+	return OK;
+}
+
+
+void switch_rx(struct net_device *dev, int len, struct sk_buff *skb)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+#ifdef CONFIG_NET_HW_FLOWCONTROL
+	int mit_sel = 0;
+#endif
+	skb->dev = dev;
+	skb->protocol = eth_type_trans(skb, dev);
+
+#ifdef CONFIG_NET_HW_FLOWCONTROL
+	mit_sel = netif_rx(skb);
+	switch (mit_sel) {
+	case NET_RX_SUCCESS:
+	case NET_RX_CN_LOW:
+	case NET_RX_CN_MOD:
+		break;
+	case NET_RX_CN_HIGH:
+		break;
+	case NET_RX_DROP:
+		if ((priv->fc_bit)
+			&& (!test_and_set_bit(priv->fc_bit, &netdev_fc_xoff))) {
+			close_rx_dma(dev);
+		}
+		break;
+	}
+#else
+	netif_rx(skb);
+#endif
+	priv->stats.rx_packets++;
+	priv->stats.rx_bytes += len;
+	return;
+}
+
+int asmlinkage switch_hw_tx(char *buf, int len, struct net_device *dev)
+{
+	struct switch_priv *priv = dev->priv;
+	struct dma_device_info *dma_dev = priv->dma_device;
+
+	dma_dev->current_tx_chan = 0;
+	return dma_device_write(dma_dev, buf, len, priv->skb);
+}
+
+int asmlinkage switch_tx(struct sk_buff *skb, struct net_device *dev)
+{
+	int len;
+	char *data;
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+
+	len = skb->len < ETH_ZLEN ? ETH_ZLEN : skb->len;
+	data = skb->data;
+	priv->skb = skb;
+	dev->trans_start = jiffies;
+
+	if (switch_hw_tx(data, len, dev) != len) {
+		dev_kfree_skb_any(skb);
+		return OK;
+	}
+
+	priv->stats.tx_packets++;
+	priv->stats.tx_bytes += len;
+	return OK;
+}
+
+void switch_tx_timeout(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	priv->stats.tx_errors++;
+	netif_wake_queue(dev);
+	return;
+}
+
+void negotiate(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	unsigned short data = get_mdio_reg(priv->mdio_phy_addr, MDIO_ADVERTISMENT_REG);
+
+	data &= ~(MDIO_ADVERT_100_HD | MDIO_ADVERT_100_FD | MDIO_ADVERT_10_FD | MDIO_ADVERT_10_HD);
+
+	switch (priv->current_speed_selection) {
+	case 10:
+		if (priv->current_duplex == full)
+			data |= MDIO_ADVERT_10_FD;
+		else if (priv->current_duplex == half)
+			data |= MDIO_ADVERT_10_HD;
+		else
+			data |= MDIO_ADVERT_10_HD | MDIO_ADVERT_10_FD;
+		break;
+
+	case 100:
+		if (priv->current_duplex == full)
+			data |= MDIO_ADVERT_100_FD;
+		else if (priv->current_duplex == half)
+			data |= MDIO_ADVERT_100_HD;
+		else
+			data |= MDIO_ADVERT_100_HD | MDIO_ADVERT_100_FD;
+		break;
+
+	case 0:					/* Auto */
+		if (priv->current_duplex == full)
+			data |= MDIO_ADVERT_100_FD | MDIO_ADVERT_10_FD;
+		else if (priv->current_duplex == half)
+			data |= MDIO_ADVERT_100_HD | MDIO_ADVERT_10_HD;
+		else
+			data |=	MDIO_ADVERT_100_HD | MDIO_ADVERT_100_FD | MDIO_ADVERT_10_FD	| MDIO_ADVERT_10_HD;
+		break;
+
+	default:					/* assume autoneg speed and duplex */
+		data |= MDIO_ADVERT_100_HD | MDIO_ADVERT_100_FD | MDIO_ADVERT_10_FD | MDIO_ADVERT_10_HD;
+	}
+
+	set_mdio_reg(priv->mdio_phy_addr, MDIO_ADVERTISMENT_REG, data);
+
+	/* Renegotiate with link partner */
+
+	data = get_mdio_reg(priv->mdio_phy_addr, MDIO_BASE_CONTROL_REG);
+	data |= MDIO_BC_NEGOTIATE;
+
+	set_mdio_reg(priv->mdio_phy_addr, MDIO_BASE_CONTROL_REG, data);
+
+}
+
+
+void set_duplex(struct net_device *dev, enum duplex new_duplex)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	if (new_duplex != priv->current_duplex) {
+		priv->current_duplex = new_duplex;
+		negotiate(dev);
+	}
+}
+
+void set_speed(struct net_device *dev, unsigned long speed)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	priv->current_speed_selection = speed;
+	negotiate(dev);
+}
+
+static int switch_ethtool_ioctl(struct net_device *dev, struct ifreq *ifr)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	struct ethtool_cmd ecmd;
+
+	if (copy_from_user(&ecmd, ifr->ifr_data, sizeof(ecmd)))
+		return -EFAULT;
+
+	switch (ecmd.cmd) {
+	case ETHTOOL_GSET:
+		memset((void *) &ecmd, 0, sizeof(ecmd));
+		ecmd.supported = SUPPORTED_Autoneg | SUPPORTED_TP | SUPPORTED_MII |	SUPPORTED_10baseT_Half | SUPPORTED_10baseT_Full |
+						SUPPORTED_100baseT_Half | SUPPORTED_100baseT_Full;
+		ecmd.port = PORT_TP;
+		ecmd.transceiver = XCVR_EXTERNAL;
+		ecmd.phy_address = priv->mdio_phy_addr;
+
+		ecmd.speed = priv->current_speed;
+
+		ecmd.duplex = priv->full_duplex ? DUPLEX_FULL : DUPLEX_HALF;
+
+		ecmd.advertising = ADVERTISED_TP;
+		if (priv->current_duplex == autoneg && priv->current_speed_selection == 0)
+			ecmd.advertising |= ADVERTISED_Autoneg;
+		else {
+			ecmd.advertising |=	ADVERTISED_10baseT_Half | ADVERTISED_10baseT_Full |
+				ADVERTISED_100baseT_Half | ADVERTISED_100baseT_Full;
+			if (priv->current_speed_selection == 10)
+				ecmd.advertising &=	~(ADVERTISED_100baseT_Half | ADVERTISED_100baseT_Full);
+			else if (priv->current_speed_selection == 100)
+				ecmd.advertising &=	~(ADVERTISED_10baseT_Half | ADVERTISED_10baseT_Full);
+			if (priv->current_duplex == half)
+				ecmd.advertising &=	~(ADVERTISED_10baseT_Full | ADVERTISED_100baseT_Full);
+			else if (priv->current_duplex == full)
+				ecmd.advertising &=	~(ADVERTISED_10baseT_Half | ADVERTISED_100baseT_Half);
+		}
+		ecmd.autoneg = AUTONEG_ENABLE;
+		if (copy_to_user(ifr->ifr_data, &ecmd, sizeof(ecmd)))
+			return -EFAULT;
+		break;
+
+	case ETHTOOL_SSET:
+		if (!capable(CAP_NET_ADMIN)) {
+			return -EPERM;
+		}
+		if (ecmd.autoneg == AUTONEG_ENABLE) {
+			set_duplex(dev, autoneg);
+			set_speed(dev, 0);
+		} else {
+			set_duplex(dev, ecmd.duplex == DUPLEX_HALF ? half : full);
+			set_speed(dev, ecmd.speed == SPEED_10 ? 10 : 100);
+		}
+		break;
+
+	case ETHTOOL_GDRVINFO:
+		{
+			struct ethtool_drvinfo info;
+			memset((void *) &info, 0, sizeof(info));
+			strncpy(info.driver, "AMAZONE", sizeof(info.driver) - 1);
+			strncpy(info.fw_version, "N/A", sizeof(info.fw_version) - 1);
+			strncpy(info.bus_info, "N/A", sizeof(info.bus_info) - 1);
+			info.regdump_len = 0;
+			info.eedump_len = 0;
+			info.testinfo_len = 0;
+			if (copy_to_user(ifr->ifr_data, &info, sizeof(info)))
+				return -EFAULT;
+		}
+		break;
+	case ETHTOOL_NWAY_RST:
+		if (priv->current_duplex == autoneg	&& priv->current_speed_selection == 0)
+			negotiate(dev);
+		break;
+	default:
+		return -EOPNOTSUPP;
+		break;
+	}
+	return 0;
+}
+
+
+
+int mac_table_tools_ioctl(struct net_device *dev, struct mac_table_req *req)
+{
+	int cmd;
+	int i;
+	cmd = req->cmd;
+	switch (cmd) {
+	case RESET_MAC_TABLE:
+		for (i = 0; i < 32; i++) {
+			write_mac_table_entry(i, 0);
+		}
+		break;
+	case READ_MAC_ENTRY:
+		req->entry_value = read_mac_table_entry(req->index);
+		break;
+	case WRITE_MAC_ENTRY:
+		write_mac_table_entry(req->index, req->entry_value);
+		break;
+	case ADD_MAC_ENTRY:
+		add_mac_table_entry(req->entry_value);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+
+/*
+    the ioctl for the switch driver is developed in the conventional way
+    the control type falls into some basic categories, among them, the 
+    SIOCETHTOOL is the traditional eth interface. VLAN_TOOLS and  
+    MAC_TABLE_TOOLS are designed specifically for amazon chip. User 
+    should be aware of the data structures used in these interfaces. 
+*/
+int switch_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
+{
+	struct data_req *switch_data_req = (struct data_req *) ifr->ifr_data;
+	struct mac_table_req *switch_mac_table_req;
+	switch (cmd) {
+	case SIOCETHTOOL:
+		switch_ethtool_ioctl(dev, ifr);
+		break;
+	case SIOCGMIIPHY:			/* Get PHY address */
+		break;
+	case SIOCGMIIREG:			/* Read MII register */
+		break;
+	case SIOCSMIIREG:			/* Write MII register */
+		break;
+	case SET_ETH_SPEED_10:		/* 10 Mbps */
+		break;
+	case SET_ETH_SPEED_100:	/* 100 Mbps */
+		break;
+	case SET_ETH_SPEED_AUTO:	/* Auto negotiate speed */
+		break;
+	case SET_ETH_DUPLEX_HALF:	/* Half duplex. */
+		break;
+	case SET_ETH_DUPLEX_FULL:	/* Full duplex. */
+		break;
+	case SET_ETH_DUPLEX_AUTO:	/* Autonegotiate duplex */
+		break;
+	case SET_ETH_REG:
+		AMAZON_SW_REG32(switch_data_req->index) = switch_data_req->value;
+		break;
+	case MAC_TABLE_TOOLS:
+		switch_mac_table_req = (struct mac_table_req *) ifr->ifr_data;
+		mac_table_tools_ioctl(dev, switch_mac_table_req);
+		break;
+	default:
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+struct net_device_stats *switch_stats(struct net_device *dev)
+{
+	struct switch_priv *priv = (struct switch_priv *) dev->priv;
+	return &priv->stats;
+}
+
+int switch_change_mtu(struct net_device *dev, int new_mtu)
+{
+	if (new_mtu >= 1516)
+		new_mtu = 1516;
+	dev->mtu = new_mtu;
+	return 0;
+}
+
+int switch_hw_receive(struct net_device *dev, struct dma_device_info *dma_dev)
+{
+	u8 *buf = NULL;
+	int len = 0;
+	struct sk_buff *skb = NULL;
+
+	len = dma_device_read(dma_dev, &buf, (void **) &skb);
+
+	if (len >= 0x600) {
+		printk("packet too large %d\n", len);
+		goto switch_hw_receive_err_exit;
+	}
+
+	/* remove CRC */
+	len -= 4;
+	if (skb == NULL) {
+		printk("cannot restore pointer\n");
+		goto switch_hw_receive_err_exit;
+	}
+	if (len > (skb->end - skb->tail)) {
+		printk("BUG, len:%d end:%p tail:%p\n", (len + 4), skb->end, skb->tail);
+		goto switch_hw_receive_err_exit;
+	}
+	skb_put(skb, len);
+	skb->dev = dev;
+	switch_rx(dev, len, skb);
+	return OK;
+  
+  switch_hw_receive_err_exit:
+	if (skb)
+		dev_kfree_skb_any(skb);
+	return -EIO;
+}
+
+int dma_intr_handler(struct dma_device_info *dma_dev, int status)
+{
+	struct net_device *dev;
+
+	dev = switch_devs + (u32) dma_dev->priv;
+	switch (status) {
+	case RCV_INT:
+		switch_hw_receive(dev, dma_dev);
+		break;
+	case TX_BUF_FULL_INT:
+		netif_stop_queue(dev);
+		break;
+	case TRANSMIT_CPT_INT:
+		netif_wake_queue(dev);
+		break;
+	}
+	return OK;
+}
+
+/* reserve 2 bytes in front of data pointer*/
+u8 *dma_buffer_alloc(int len, int *byte_offset, void **opt)
+{
+	u8 *buffer = NULL;
+	struct sk_buff *skb = NULL;
+	skb = dev_alloc_skb(ETHERNET_PACKET_DMA_BUFFER_SIZE);
+	if (skb == NULL) {
+		return NULL;
+	}
+	buffer = (u8 *) (skb->data);
+	skb_reserve(skb, 2);
+	*(int *) opt = (int) skb;
+	*byte_offset = 2;
+	return buffer;
+}
+
+int dma_buffer_free(u8 * dataptr, void *opt)
+{
+	struct sk_buff *skb = NULL;
+	if (opt == NULL) {
+		kfree(dataptr);
+	} else {
+		skb = (struct sk_buff *) opt;
+		dev_kfree_skb_any(skb);
+	}
+	return OK;
+}
+
+int init_dma_device(_dma_device_info * dma_dev)
+{
+	int i;
+	int num_tx_chan, num_rx_chan;
+	if (strcmp(dma_dev->device_name, "switch1") == 0) {
+		num_tx_chan = 1;
+		num_rx_chan = 2;
+		dma_dev->priv = (void *) 0;
+	} else {
+		num_tx_chan = 1;
+		num_rx_chan = 2;
+		dma_dev->priv = (void *) 1;
+	}
+
+	dma_dev->weight = 1;
+	dma_dev->num_tx_chan = num_tx_chan;
+	dma_dev->num_rx_chan = num_rx_chan;
+	dma_dev->ack = 1;
+	dma_dev->tx_burst_len = 4;
+	dma_dev->rx_burst_len = 4;
+	for (i = 0; i < dma_dev->num_tx_chan; i++) {
+		dma_dev->tx_chan[i].weight = QOS_DEFAULT_WGT;
+		dma_dev->tx_chan[i].desc_num = 10;
+		dma_dev->tx_chan[i].packet_size = 0;
+		dma_dev->tx_chan[i].control = 0;
+	}
+	for (i = 0; i < num_rx_chan; i++) {
+		dma_dev->rx_chan[i].weight = QOS_DEFAULT_WGT;
+		dma_dev->rx_chan[i].desc_num = 10;
+		dma_dev->rx_chan[i].packet_size = ETHERNET_PACKET_DMA_BUFFER_SIZE;
+		dma_dev->rx_chan[i].control = 0;
+	}
+	dma_dev->intr_handler = dma_intr_handler;
+	dma_dev->buffer_alloc = dma_buffer_alloc;
+	dma_dev->buffer_free = dma_buffer_free;
+	return 0;
+}
+
+int switch_set_mac_address(struct net_device *dev, void *p)
+{
+	struct sockaddr *addr = p;
+	memcpy(dev->dev_addr, addr->sa_data, dev->addr_len);
+	return OK;
+}
+
+
+int switch_init(struct net_device *dev)
+{
+	u64 retval = 0;
+	int i;
+	int result;
+	struct switch_priv *priv;
+	ether_setup(dev);			/* assign some of the fields */
+	printk("%s up using ", dev->name);
+	dev->open = switch_open;
+	dev->stop = switch_release;
+	dev->hard_start_xmit = switch_tx;
+	dev->do_ioctl = switch_ioctl;
+	dev->get_stats = switch_stats;
+	dev->change_mtu = switch_change_mtu;
+	dev->set_mac_address = switch_set_mac_address;
+	dev->tx_timeout = switch_tx_timeout;
+	dev->watchdog_timeo = timeout;
+
+	SET_MODULE_OWNER(dev);
+
+	dev->priv = kmalloc(sizeof(struct switch_priv), GFP_KERNEL);
+	if (dev->priv == NULL)
+		return -ENOMEM;
+	memset(dev->priv, 0, sizeof(struct switch_priv));
+	priv = dev->priv;
+	priv->dma_device = (struct dma_device_info *) kmalloc(sizeof(struct dma_device_info), GFP_KERNEL);
+	if ((dev - switch_devs) == 0) {
+		sprintf(priv->dma_device->device_name, "switch1");
+	} else if ((dev - switch_devs) == 1) {
+		sprintf(priv->dma_device->device_name, "switch2");
+	}
+	printk("\"%s\"\n", priv->dma_device->device_name);
+	init_dma_device(priv->dma_device);
+	result = dma_device_register(priv->dma_device);
+
+	/* read the mac address from the mac table and put them into the mac table. */
+	for (i = 0; i < 6; i++) {
+		retval += my_ethaddr[i];
+	}
+	/* ethaddr not set in u-boot ? */
+	if (retval == 0) {
+		dev->dev_addr[0] = 0x00;
+		dev->dev_addr[1] = 0x20;
+		dev->dev_addr[2] = 0xda;
+		dev->dev_addr[3] = 0x86;
+		dev->dev_addr[4] = 0x23;
+		dev->dev_addr[5] = 0x74 + (unsigned char) (dev - switch_devs);
+	} else {
+		for (i = 0; i < 6; i++) {
+			dev->dev_addr[i] = my_ethaddr[i];
+		}
+		dev->dev_addr[5] += +(unsigned char) (dev - switch_devs);
+	}
+	return OK;
+}
+
+int switch_init_module(void)
+{
+	int i = 0, result, device_present = 0;
+
+	for (i = 0; i < AMAZON_SW_INT_NO; i++) {
+		sprintf(switch_devs[i].name, "eth%d", i);
+
+		if ((result = register_netdev(switch_devs + i)))
+			printk("error %i registering device \"%s\"\n", result, switch_devs[i].name);
+		else
+			device_present++;
+	}
+	amazon_sw_chip_init();
+	return device_present ? 0 : -ENODEV;
+}
+
+void switch_cleanup(void)
+{
+	int i;
+	struct switch_priv *priv;
+	for (i = 0; i < AMAZON_SW_INT_NO; i++) {
+		priv = switch_devs[i].priv;
+		if (priv->dma_device) {
+			dma_device_unregister(priv->dma_device);
+			kfree(priv->dma_device);
+		}
+		kfree(switch_devs[i].priv);
+		unregister_netdev(switch_devs + i);
+	}
+	return;
+}
+
+module_init(switch_init_module);
+module_exit(switch_cleanup);
+
+MODULE_LICENSE("GPL");
+MODULE_AUTHOR("Wu Qi Ming");