Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / net / ethernet / ibm / emac / core.c
diff --git a/kernel/drivers/net/ethernet/ibm/emac/core.c b/kernel/drivers/net/ethernet/ibm/emac/core.c
new file mode 100644 (file)
index 0000000..b9df0cb
--- /dev/null
@@ -0,0 +1,3100 @@
+/*
+ * drivers/net/ethernet/ibm/emac/core.c
+ *
+ * Driver for PowerPC 4xx on-chip ethernet controller.
+ *
+ * Copyright 2007 Benjamin Herrenschmidt, IBM Corp.
+ *                <benh@kernel.crashing.org>
+ *
+ * Based on the arch/ppc version of the driver:
+ *
+ * Copyright (c) 2004, 2005 Zultys Technologies.
+ * Eugene Surovegin <eugene.surovegin@zultys.com> or <ebs@ebshome.net>
+ *
+ * Based on original work by
+ *     Matt Porter <mporter@kernel.crashing.org>
+ *     (c) 2003 Benjamin Herrenschmidt <benh@kernel.crashing.org>
+ *      Armin Kuster <akuster@mvista.com>
+ *     Johnnie Peters <jpeters@mvista.com>
+ *
+ * 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.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/string.h>
+#include <linux/errno.h>
+#include <linux/delay.h>
+#include <linux/types.h>
+#include <linux/pci.h>
+#include <linux/etherdevice.h>
+#include <linux/skbuff.h>
+#include <linux/crc32.h>
+#include <linux/ethtool.h>
+#include <linux/mii.h>
+#include <linux/bitops.h>
+#include <linux/workqueue.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/of_irq.h>
+#include <linux/of_net.h>
+#include <linux/slab.h>
+
+#include <asm/processor.h>
+#include <asm/io.h>
+#include <asm/dma.h>
+#include <asm/uaccess.h>
+#include <asm/dcr.h>
+#include <asm/dcr-regs.h>
+
+#include "core.h"
+
+/*
+ * Lack of dma_unmap_???? calls is intentional.
+ *
+ * API-correct usage requires additional support state information to be
+ * maintained for every RX and TX buffer descriptor (BD). Unfortunately, due to
+ * EMAC design (e.g. TX buffer passed from network stack can be split into
+ * several BDs, dma_map_single/dma_map_page can be used to map particular BD),
+ * maintaining such information will add additional overhead.
+ * Current DMA API implementation for 4xx processors only ensures cache coherency
+ * and dma_unmap_???? routines are empty and are likely to stay this way.
+ * I decided to omit dma_unmap_??? calls because I don't want to add additional
+ * complexity just for the sake of following some abstract API, when it doesn't
+ * add any real benefit to the driver. I understand that this decision maybe
+ * controversial, but I really tried to make code API-correct and efficient
+ * at the same time and didn't come up with code I liked :(.                --ebs
+ */
+
+#define DRV_NAME        "emac"
+#define DRV_VERSION     "3.54"
+#define DRV_DESC        "PPC 4xx OCP EMAC driver"
+
+MODULE_DESCRIPTION(DRV_DESC);
+MODULE_AUTHOR
+    ("Eugene Surovegin <eugene.surovegin@zultys.com> or <ebs@ebshome.net>");
+MODULE_LICENSE("GPL");
+
+/* minimum number of free TX descriptors required to wake up TX process */
+#define EMAC_TX_WAKEUP_THRESH          (NUM_TX_BUFF / 4)
+
+/* If packet size is less than this number, we allocate small skb and copy packet
+ * contents into it instead of just sending original big skb up
+ */
+#define EMAC_RX_COPY_THRESH            CONFIG_IBM_EMAC_RX_COPY_THRESHOLD
+
+/* Since multiple EMACs share MDIO lines in various ways, we need
+ * to avoid re-using the same PHY ID in cases where the arch didn't
+ * setup precise phy_map entries
+ *
+ * XXX This is something that needs to be reworked as we can have multiple
+ * EMAC "sets" (multiple ASICs containing several EMACs) though we can
+ * probably require in that case to have explicit PHY IDs in the device-tree
+ */
+static u32 busy_phy_map;
+static DEFINE_MUTEX(emac_phy_map_lock);
+
+/* This is the wait queue used to wait on any event related to probe, that
+ * is discovery of MALs, other EMACs, ZMII/RGMIIs, etc...
+ */
+static DECLARE_WAIT_QUEUE_HEAD(emac_probe_wait);
+
+/* Having stable interface names is a doomed idea. However, it would be nice
+ * if we didn't have completely random interface names at boot too :-) It's
+ * just a matter of making everybody's life easier. Since we are doing
+ * threaded probing, it's a bit harder though. The base idea here is that
+ * we make up a list of all emacs in the device-tree before we register the
+ * driver. Every emac will then wait for the previous one in the list to
+ * initialize before itself. We should also keep that list ordered by
+ * cell_index.
+ * That list is only 4 entries long, meaning that additional EMACs don't
+ * get ordering guarantees unless EMAC_BOOT_LIST_SIZE is increased.
+ */
+
+#define EMAC_BOOT_LIST_SIZE    4
+static struct device_node *emac_boot_list[EMAC_BOOT_LIST_SIZE];
+
+/* How long should I wait for dependent devices ? */
+#define EMAC_PROBE_DEP_TIMEOUT (HZ * 5)
+
+/* I don't want to litter system log with timeout errors
+ * when we have brain-damaged PHY.
+ */
+static inline void emac_report_timeout_error(struct emac_instance *dev,
+                                            const char *error)
+{
+       if (emac_has_feature(dev, EMAC_FTR_440GX_PHY_CLK_FIX |
+                                 EMAC_FTR_460EX_PHY_CLK_FIX |
+                                 EMAC_FTR_440EP_PHY_CLK_FIX))
+               DBG(dev, "%s" NL, error);
+       else if (net_ratelimit())
+               printk(KERN_ERR "%s: %s\n", dev->ofdev->dev.of_node->full_name,
+                       error);
+}
+
+/* EMAC PHY clock workaround:
+ * 440EP/440GR has more sane SDR0_MFR register implementation than 440GX,
+ * which allows controlling each EMAC clock
+ */
+static inline void emac_rx_clk_tx(struct emac_instance *dev)
+{
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_440EP_PHY_CLK_FIX))
+               dcri_clrset(SDR0, SDR0_MFR,
+                           0, SDR0_MFR_ECS >> dev->cell_index);
+#endif
+}
+
+static inline void emac_rx_clk_default(struct emac_instance *dev)
+{
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_440EP_PHY_CLK_FIX))
+               dcri_clrset(SDR0, SDR0_MFR,
+                           SDR0_MFR_ECS >> dev->cell_index, 0);
+#endif
+}
+
+/* PHY polling intervals */
+#define PHY_POLL_LINK_ON       HZ
+#define PHY_POLL_LINK_OFF      (HZ / 5)
+
+/* Graceful stop timeouts in us.
+ * We should allow up to 1 frame time (full-duplex, ignoring collisions)
+ */
+#define STOP_TIMEOUT_10                1230
+#define STOP_TIMEOUT_100       124
+#define STOP_TIMEOUT_1000      13
+#define STOP_TIMEOUT_1000_JUMBO        73
+
+static unsigned char default_mcast_addr[] = {
+       0x01, 0x80, 0xC2, 0x00, 0x00, 0x01
+};
+
+/* Please, keep in sync with struct ibm_emac_stats/ibm_emac_error_stats */
+static const char emac_stats_keys[EMAC_ETHTOOL_STATS_COUNT][ETH_GSTRING_LEN] = {
+       "rx_packets", "rx_bytes", "tx_packets", "tx_bytes", "rx_packets_csum",
+       "tx_packets_csum", "tx_undo", "rx_dropped_stack", "rx_dropped_oom",
+       "rx_dropped_error", "rx_dropped_resize", "rx_dropped_mtu",
+       "rx_stopped", "rx_bd_errors", "rx_bd_overrun", "rx_bd_bad_packet",
+       "rx_bd_runt_packet", "rx_bd_short_event", "rx_bd_alignment_error",
+       "rx_bd_bad_fcs", "rx_bd_packet_too_long", "rx_bd_out_of_range",
+       "rx_bd_in_range", "rx_parity", "rx_fifo_overrun", "rx_overrun",
+       "rx_bad_packet", "rx_runt_packet", "rx_short_event",
+       "rx_alignment_error", "rx_bad_fcs", "rx_packet_too_long",
+       "rx_out_of_range", "rx_in_range", "tx_dropped", "tx_bd_errors",
+       "tx_bd_bad_fcs", "tx_bd_carrier_loss", "tx_bd_excessive_deferral",
+       "tx_bd_excessive_collisions", "tx_bd_late_collision",
+       "tx_bd_multple_collisions", "tx_bd_single_collision",
+       "tx_bd_underrun", "tx_bd_sqe", "tx_parity", "tx_underrun", "tx_sqe",
+       "tx_errors"
+};
+
+static irqreturn_t emac_irq(int irq, void *dev_instance);
+static void emac_clean_tx_ring(struct emac_instance *dev);
+static void __emac_set_multicast_list(struct emac_instance *dev);
+
+static inline int emac_phy_supports_gige(int phy_mode)
+{
+       return  phy_mode == PHY_MODE_GMII ||
+               phy_mode == PHY_MODE_RGMII ||
+               phy_mode == PHY_MODE_SGMII ||
+               phy_mode == PHY_MODE_TBI ||
+               phy_mode == PHY_MODE_RTBI;
+}
+
+static inline int emac_phy_gpcs(int phy_mode)
+{
+       return  phy_mode == PHY_MODE_SGMII ||
+               phy_mode == PHY_MODE_TBI ||
+               phy_mode == PHY_MODE_RTBI;
+}
+
+static inline void emac_tx_enable(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r;
+
+       DBG(dev, "tx_enable" NL);
+
+       r = in_be32(&p->mr0);
+       if (!(r & EMAC_MR0_TXE))
+               out_be32(&p->mr0, r | EMAC_MR0_TXE);
+}
+
+static void emac_tx_disable(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r;
+
+       DBG(dev, "tx_disable" NL);
+
+       r = in_be32(&p->mr0);
+       if (r & EMAC_MR0_TXE) {
+               int n = dev->stop_timeout;
+               out_be32(&p->mr0, r & ~EMAC_MR0_TXE);
+               while (!(in_be32(&p->mr0) & EMAC_MR0_TXI) && n) {
+                       udelay(1);
+                       --n;
+               }
+               if (unlikely(!n))
+                       emac_report_timeout_error(dev, "TX disable timeout");
+       }
+}
+
+static void emac_rx_enable(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r;
+
+       if (unlikely(test_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags)))
+               goto out;
+
+       DBG(dev, "rx_enable" NL);
+
+       r = in_be32(&p->mr0);
+       if (!(r & EMAC_MR0_RXE)) {
+               if (unlikely(!(r & EMAC_MR0_RXI))) {
+                       /* Wait if previous async disable is still in progress */
+                       int n = dev->stop_timeout;
+                       while (!(r = in_be32(&p->mr0) & EMAC_MR0_RXI) && n) {
+                               udelay(1);
+                               --n;
+                       }
+                       if (unlikely(!n))
+                               emac_report_timeout_error(dev,
+                                                         "RX disable timeout");
+               }
+               out_be32(&p->mr0, r | EMAC_MR0_RXE);
+       }
+ out:
+       ;
+}
+
+static void emac_rx_disable(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r;
+
+       DBG(dev, "rx_disable" NL);
+
+       r = in_be32(&p->mr0);
+       if (r & EMAC_MR0_RXE) {
+               int n = dev->stop_timeout;
+               out_be32(&p->mr0, r & ~EMAC_MR0_RXE);
+               while (!(in_be32(&p->mr0) & EMAC_MR0_RXI) && n) {
+                       udelay(1);
+                       --n;
+               }
+               if (unlikely(!n))
+                       emac_report_timeout_error(dev, "RX disable timeout");
+       }
+}
+
+static inline void emac_netif_stop(struct emac_instance *dev)
+{
+       netif_tx_lock_bh(dev->ndev);
+       netif_addr_lock(dev->ndev);
+       dev->no_mcast = 1;
+       netif_addr_unlock(dev->ndev);
+       netif_tx_unlock_bh(dev->ndev);
+       dev->ndev->trans_start = jiffies;       /* prevent tx timeout */
+       mal_poll_disable(dev->mal, &dev->commac);
+       netif_tx_disable(dev->ndev);
+}
+
+static inline void emac_netif_start(struct emac_instance *dev)
+{
+       netif_tx_lock_bh(dev->ndev);
+       netif_addr_lock(dev->ndev);
+       dev->no_mcast = 0;
+       if (dev->mcast_pending && netif_running(dev->ndev))
+               __emac_set_multicast_list(dev);
+       netif_addr_unlock(dev->ndev);
+       netif_tx_unlock_bh(dev->ndev);
+
+       netif_wake_queue(dev->ndev);
+
+       /* NOTE: unconditional netif_wake_queue is only appropriate
+        * so long as all callers are assured to have free tx slots
+        * (taken from tg3... though the case where that is wrong is
+        *  not terribly harmful)
+        */
+       mal_poll_enable(dev->mal, &dev->commac);
+}
+
+static inline void emac_rx_disable_async(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r;
+
+       DBG(dev, "rx_disable_async" NL);
+
+       r = in_be32(&p->mr0);
+       if (r & EMAC_MR0_RXE)
+               out_be32(&p->mr0, r & ~EMAC_MR0_RXE);
+}
+
+static int emac_reset(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       int n = 20;
+
+       DBG(dev, "reset" NL);
+
+       if (!dev->reset_failed) {
+               /* 40x erratum suggests stopping RX channel before reset,
+                * we stop TX as well
+                */
+               emac_rx_disable(dev);
+               emac_tx_disable(dev);
+       }
+
+#ifdef CONFIG_PPC_DCR_NATIVE
+       /*
+        * PPC460EX/GT Embedded Processor Advanced User's Manual
+        * section 28.10.1 Mode Register 0 (EMACx_MR0) states:
+        * Note: The PHY must provide a TX Clk in order to perform a soft reset
+        * of the EMAC. If none is present, select the internal clock
+        * (SDR0_ETH_CFG[EMACx_PHY_CLK] = 1).
+        * After a soft reset, select the external clock.
+        */
+       if (emac_has_feature(dev, EMAC_FTR_460EX_PHY_CLK_FIX)) {
+               if (dev->phy_address == 0xffffffff &&
+                   dev->phy_map == 0xffffffff) {
+                       /* No PHY: select internal loop clock before reset */
+                       dcri_clrset(SDR0, SDR0_ETH_CFG,
+                                   0, SDR0_ETH_CFG_ECS << dev->cell_index);
+               } else {
+                       /* PHY present: select external clock before reset */
+                       dcri_clrset(SDR0, SDR0_ETH_CFG,
+                                   SDR0_ETH_CFG_ECS << dev->cell_index, 0);
+               }
+       }
+#endif
+
+       out_be32(&p->mr0, EMAC_MR0_SRST);
+       while ((in_be32(&p->mr0) & EMAC_MR0_SRST) && n)
+               --n;
+
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_460EX_PHY_CLK_FIX)) {
+               if (dev->phy_address == 0xffffffff &&
+                   dev->phy_map == 0xffffffff) {
+                       /* No PHY: restore external clock source after reset */
+                       dcri_clrset(SDR0, SDR0_ETH_CFG,
+                                   SDR0_ETH_CFG_ECS << dev->cell_index, 0);
+               }
+       }
+#endif
+
+       if (n) {
+               dev->reset_failed = 0;
+               return 0;
+       } else {
+               emac_report_timeout_error(dev, "reset timeout");
+               dev->reset_failed = 1;
+               return -ETIMEDOUT;
+       }
+}
+
+static void emac_hash_mc(struct emac_instance *dev)
+{
+       const int regs = EMAC_XAHT_REGS(dev);
+       u32 *gaht_base = emac_gaht_base(dev);
+       u32 gaht_temp[regs];
+       struct netdev_hw_addr *ha;
+       int i;
+
+       DBG(dev, "hash_mc %d" NL, netdev_mc_count(dev->ndev));
+
+       memset(gaht_temp, 0, sizeof (gaht_temp));
+
+       netdev_for_each_mc_addr(ha, dev->ndev) {
+               int slot, reg, mask;
+               DBG2(dev, "mc %pM" NL, ha->addr);
+
+               slot = EMAC_XAHT_CRC_TO_SLOT(dev,
+                                            ether_crc(ETH_ALEN, ha->addr));
+               reg = EMAC_XAHT_SLOT_TO_REG(dev, slot);
+               mask = EMAC_XAHT_SLOT_TO_MASK(dev, slot);
+
+               gaht_temp[reg] |= mask;
+       }
+
+       for (i = 0; i < regs; i++)
+               out_be32(gaht_base + i, gaht_temp[i]);
+}
+
+static inline u32 emac_iff2rmr(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       u32 r;
+
+       r = EMAC_RMR_SP | EMAC_RMR_SFCS | EMAC_RMR_IAE | EMAC_RMR_BAE;
+
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+           r |= EMAC4_RMR_BASE;
+       else
+           r |= EMAC_RMR_BASE;
+
+       if (ndev->flags & IFF_PROMISC)
+               r |= EMAC_RMR_PME;
+       else if (ndev->flags & IFF_ALLMULTI ||
+                        (netdev_mc_count(ndev) > EMAC_XAHT_SLOTS(dev)))
+               r |= EMAC_RMR_PMME;
+       else if (!netdev_mc_empty(ndev))
+               r |= EMAC_RMR_MAE;
+
+       if (emac_has_feature(dev, EMAC_APM821XX_REQ_JUMBO_FRAME_SIZE)) {
+               r &= ~EMAC4_RMR_MJS_MASK;
+               r |= EMAC4_RMR_MJS(ndev->mtu);
+       }
+
+       return r;
+}
+
+static u32 __emac_calc_base_mr1(struct emac_instance *dev, int tx_size, int rx_size)
+{
+       u32 ret = EMAC_MR1_VLE | EMAC_MR1_IST | EMAC_MR1_TR0_MULT;
+
+       DBG2(dev, "__emac_calc_base_mr1" NL);
+
+       switch(tx_size) {
+       case 2048:
+               ret |= EMAC_MR1_TFS_2K;
+               break;
+       default:
+               printk(KERN_WARNING "%s: Unknown Tx FIFO size %d\n",
+                      dev->ndev->name, tx_size);
+       }
+
+       switch(rx_size) {
+       case 16384:
+               ret |= EMAC_MR1_RFS_16K;
+               break;
+       case 4096:
+               ret |= EMAC_MR1_RFS_4K;
+               break;
+       default:
+               printk(KERN_WARNING "%s: Unknown Rx FIFO size %d\n",
+                      dev->ndev->name, rx_size);
+       }
+
+       return ret;
+}
+
+static u32 __emac4_calc_base_mr1(struct emac_instance *dev, int tx_size, int rx_size)
+{
+       u32 ret = EMAC_MR1_VLE | EMAC_MR1_IST | EMAC4_MR1_TR |
+               EMAC4_MR1_OBCI(dev->opb_bus_freq / 1000000);
+
+       DBG2(dev, "__emac4_calc_base_mr1" NL);
+
+       switch(tx_size) {
+       case 16384:
+               ret |= EMAC4_MR1_TFS_16K;
+               break;
+       case 4096:
+               ret |= EMAC4_MR1_TFS_4K;
+               break;
+       case 2048:
+               ret |= EMAC4_MR1_TFS_2K;
+               break;
+       default:
+               printk(KERN_WARNING "%s: Unknown Tx FIFO size %d\n",
+                      dev->ndev->name, tx_size);
+       }
+
+       switch(rx_size) {
+       case 16384:
+               ret |= EMAC4_MR1_RFS_16K;
+               break;
+       case 4096:
+               ret |= EMAC4_MR1_RFS_4K;
+               break;
+       case 2048:
+               ret |= EMAC4_MR1_RFS_2K;
+               break;
+       default:
+               printk(KERN_WARNING "%s: Unknown Rx FIFO size %d\n",
+                      dev->ndev->name, rx_size);
+       }
+
+       return ret;
+}
+
+static u32 emac_calc_base_mr1(struct emac_instance *dev, int tx_size, int rx_size)
+{
+       return emac_has_feature(dev, EMAC_FTR_EMAC4) ?
+               __emac4_calc_base_mr1(dev, tx_size, rx_size) :
+               __emac_calc_base_mr1(dev, tx_size, rx_size);
+}
+
+static inline u32 emac_calc_trtr(struct emac_instance *dev, unsigned int size)
+{
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               return ((size >> 6) - 1) << EMAC_TRTR_SHIFT_EMAC4;
+       else
+               return ((size >> 6) - 1) << EMAC_TRTR_SHIFT;
+}
+
+static inline u32 emac_calc_rwmr(struct emac_instance *dev,
+                                unsigned int low, unsigned int high)
+{
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               return (low << 22) | ( (high & 0x3ff) << 6);
+       else
+               return (low << 23) | ( (high & 0x1ff) << 7);
+}
+
+static int emac_configure(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       struct net_device *ndev = dev->ndev;
+       int tx_size, rx_size, link = netif_carrier_ok(dev->ndev);
+       u32 r, mr1 = 0;
+
+       DBG(dev, "configure" NL);
+
+       if (!link) {
+               out_be32(&p->mr1, in_be32(&p->mr1)
+                        | EMAC_MR1_FDE | EMAC_MR1_ILE);
+               udelay(100);
+       } else if (emac_reset(dev) < 0)
+               return -ETIMEDOUT;
+
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+               tah_reset(dev->tah_dev);
+
+       DBG(dev, " link = %d duplex = %d, pause = %d, asym_pause = %d\n",
+           link, dev->phy.duplex, dev->phy.pause, dev->phy.asym_pause);
+
+       /* Default fifo sizes */
+       tx_size = dev->tx_fifo_size;
+       rx_size = dev->rx_fifo_size;
+
+       /* No link, force loopback */
+       if (!link)
+               mr1 = EMAC_MR1_FDE | EMAC_MR1_ILE;
+
+       /* Check for full duplex */
+       else if (dev->phy.duplex == DUPLEX_FULL)
+               mr1 |= EMAC_MR1_FDE | EMAC_MR1_MWSW_001;
+
+       /* Adjust fifo sizes, mr1 and timeouts based on link speed */
+       dev->stop_timeout = STOP_TIMEOUT_10;
+       switch (dev->phy.speed) {
+       case SPEED_1000:
+               if (emac_phy_gpcs(dev->phy.mode)) {
+                       mr1 |= EMAC_MR1_MF_1000GPCS | EMAC_MR1_MF_IPPA(
+                               (dev->phy.gpcs_address != 0xffffffff) ?
+                                dev->phy.gpcs_address : dev->phy.address);
+
+                       /* Put some arbitrary OUI, Manuf & Rev IDs so we can
+                        * identify this GPCS PHY later.
+                        */
+                       out_be32(&p->u1.emac4.ipcr, 0xdeadbeef);
+               } else
+                       mr1 |= EMAC_MR1_MF_1000;
+
+               /* Extended fifo sizes */
+               tx_size = dev->tx_fifo_size_gige;
+               rx_size = dev->rx_fifo_size_gige;
+
+               if (dev->ndev->mtu > ETH_DATA_LEN) {
+                       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+                               mr1 |= EMAC4_MR1_JPSM;
+                       else
+                               mr1 |= EMAC_MR1_JPSM;
+                       dev->stop_timeout = STOP_TIMEOUT_1000_JUMBO;
+               } else
+                       dev->stop_timeout = STOP_TIMEOUT_1000;
+               break;
+       case SPEED_100:
+               mr1 |= EMAC_MR1_MF_100;
+               dev->stop_timeout = STOP_TIMEOUT_100;
+               break;
+       default: /* make gcc happy */
+               break;
+       }
+
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_set_speed(dev->rgmii_dev, dev->rgmii_port,
+                               dev->phy.speed);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_set_speed(dev->zmii_dev, dev->zmii_port, dev->phy.speed);
+
+       /* on 40x erratum forces us to NOT use integrated flow control,
+        * let's hope it works on 44x ;)
+        */
+       if (!emac_has_feature(dev, EMAC_FTR_NO_FLOW_CONTROL_40x) &&
+           dev->phy.duplex == DUPLEX_FULL) {
+               if (dev->phy.pause)
+                       mr1 |= EMAC_MR1_EIFC | EMAC_MR1_APP;
+               else if (dev->phy.asym_pause)
+                       mr1 |= EMAC_MR1_APP;
+       }
+
+       /* Add base settings & fifo sizes & program MR1 */
+       mr1 |= emac_calc_base_mr1(dev, tx_size, rx_size);
+       out_be32(&p->mr1, mr1);
+
+       /* Set individual MAC address */
+       out_be32(&p->iahr, (ndev->dev_addr[0] << 8) | ndev->dev_addr[1]);
+       out_be32(&p->ialr, (ndev->dev_addr[2] << 24) |
+                (ndev->dev_addr[3] << 16) | (ndev->dev_addr[4] << 8) |
+                ndev->dev_addr[5]);
+
+       /* VLAN Tag Protocol ID */
+       out_be32(&p->vtpid, 0x8100);
+
+       /* Receive mode register */
+       r = emac_iff2rmr(ndev);
+       if (r & EMAC_RMR_MAE)
+               emac_hash_mc(dev);
+       out_be32(&p->rmr, r);
+
+       /* FIFOs thresholds */
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               r = EMAC4_TMR1((dev->mal_burst_size / dev->fifo_entry_size) + 1,
+                              tx_size / 2 / dev->fifo_entry_size);
+       else
+               r = EMAC_TMR1((dev->mal_burst_size / dev->fifo_entry_size) + 1,
+                             tx_size / 2 / dev->fifo_entry_size);
+       out_be32(&p->tmr1, r);
+       out_be32(&p->trtr, emac_calc_trtr(dev, tx_size / 2));
+
+       /* PAUSE frame is sent when RX FIFO reaches its high-water mark,
+          there should be still enough space in FIFO to allow the our link
+          partner time to process this frame and also time to send PAUSE
+          frame itself.
+
+          Here is the worst case scenario for the RX FIFO "headroom"
+          (from "The Switch Book") (100Mbps, without preamble, inter-frame gap):
+
+          1) One maximum-length frame on TX                    1522 bytes
+          2) One PAUSE frame time                                64 bytes
+          3) PAUSE frame decode time allowance                   64 bytes
+          4) One maximum-length frame on RX                    1522 bytes
+          5) Round-trip propagation delay of the link (100Mb)    15 bytes
+          ----------
+          3187 bytes
+
+          I chose to set high-water mark to RX_FIFO_SIZE / 4 (1024 bytes)
+          low-water mark  to RX_FIFO_SIZE / 8 (512 bytes)
+        */
+       r = emac_calc_rwmr(dev, rx_size / 8 / dev->fifo_entry_size,
+                          rx_size / 4 / dev->fifo_entry_size);
+       out_be32(&p->rwmr, r);
+
+       /* Set PAUSE timer to the maximum */
+       out_be32(&p->ptr, 0xffff);
+
+       /* IRQ sources */
+       r = EMAC_ISR_OVR | EMAC_ISR_BP | EMAC_ISR_SE |
+               EMAC_ISR_ALE | EMAC_ISR_BFCS | EMAC_ISR_PTLE | EMAC_ISR_ORE |
+               EMAC_ISR_IRE | EMAC_ISR_TE;
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+           r |= EMAC4_ISR_TXPE | EMAC4_ISR_RXPE /* | EMAC4_ISR_TXUE |
+                                                 EMAC4_ISR_RXOE | */;
+       out_be32(&p->iser,  r);
+
+       /* We need to take GPCS PHY out of isolate mode after EMAC reset */
+       if (emac_phy_gpcs(dev->phy.mode)) {
+               if (dev->phy.gpcs_address != 0xffffffff)
+                       emac_mii_reset_gpcs(&dev->phy);
+               else
+                       emac_mii_reset_phy(&dev->phy);
+       }
+
+       return 0;
+}
+
+static void emac_reinitialize(struct emac_instance *dev)
+{
+       DBG(dev, "reinitialize" NL);
+
+       emac_netif_stop(dev);
+       if (!emac_configure(dev)) {
+               emac_tx_enable(dev);
+               emac_rx_enable(dev);
+       }
+       emac_netif_start(dev);
+}
+
+static void emac_full_tx_reset(struct emac_instance *dev)
+{
+       DBG(dev, "full_tx_reset" NL);
+
+       emac_tx_disable(dev);
+       mal_disable_tx_channel(dev->mal, dev->mal_tx_chan);
+       emac_clean_tx_ring(dev);
+       dev->tx_cnt = dev->tx_slot = dev->ack_slot = 0;
+
+       emac_configure(dev);
+
+       mal_enable_tx_channel(dev->mal, dev->mal_tx_chan);
+       emac_tx_enable(dev);
+       emac_rx_enable(dev);
+}
+
+static void emac_reset_work(struct work_struct *work)
+{
+       struct emac_instance *dev = container_of(work, struct emac_instance, reset_work);
+
+       DBG(dev, "reset_work" NL);
+
+       mutex_lock(&dev->link_lock);
+       if (dev->opened) {
+               emac_netif_stop(dev);
+               emac_full_tx_reset(dev);
+               emac_netif_start(dev);
+       }
+       mutex_unlock(&dev->link_lock);
+}
+
+static void emac_tx_timeout(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       DBG(dev, "tx_timeout" NL);
+
+       schedule_work(&dev->reset_work);
+}
+
+
+static inline int emac_phy_done(struct emac_instance *dev, u32 stacr)
+{
+       int done = !!(stacr & EMAC_STACR_OC);
+
+       if (emac_has_feature(dev, EMAC_FTR_STACR_OC_INVERT))
+               done = !done;
+
+       return done;
+};
+
+static int __emac_mdio_read(struct emac_instance *dev, u8 id, u8 reg)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r = 0;
+       int n, err = -ETIMEDOUT;
+
+       mutex_lock(&dev->mdio_lock);
+
+       DBG2(dev, "mdio_read(%02x,%02x)" NL, id, reg);
+
+       /* Enable proper MDIO port */
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_get_mdio(dev->zmii_dev, dev->zmii_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_get_mdio(dev->rgmii_dev, dev->rgmii_port);
+
+       /* Wait for management interface to become idle */
+       n = 20;
+       while (!emac_phy_done(dev, in_be32(&p->stacr))) {
+               udelay(1);
+               if (!--n) {
+                       DBG2(dev, " -> timeout wait idle\n");
+                       goto bail;
+               }
+       }
+
+       /* Issue read command */
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               r = EMAC4_STACR_BASE(dev->opb_bus_freq);
+       else
+               r = EMAC_STACR_BASE(dev->opb_bus_freq);
+       if (emac_has_feature(dev, EMAC_FTR_STACR_OC_INVERT))
+               r |= EMAC_STACR_OC;
+       if (emac_has_feature(dev, EMAC_FTR_HAS_NEW_STACR))
+               r |= EMACX_STACR_STAC_READ;
+       else
+               r |= EMAC_STACR_STAC_READ;
+       r |= (reg & EMAC_STACR_PRA_MASK)
+               | ((id & EMAC_STACR_PCDA_MASK) << EMAC_STACR_PCDA_SHIFT);
+       out_be32(&p->stacr, r);
+
+       /* Wait for read to complete */
+       n = 200;
+       while (!emac_phy_done(dev, (r = in_be32(&p->stacr)))) {
+               udelay(1);
+               if (!--n) {
+                       DBG2(dev, " -> timeout wait complete\n");
+                       goto bail;
+               }
+       }
+
+       if (unlikely(r & EMAC_STACR_PHYE)) {
+               DBG(dev, "mdio_read(%02x, %02x) failed" NL, id, reg);
+               err = -EREMOTEIO;
+               goto bail;
+       }
+
+       r = ((r >> EMAC_STACR_PHYD_SHIFT) & EMAC_STACR_PHYD_MASK);
+
+       DBG2(dev, "mdio_read -> %04x" NL, r);
+       err = 0;
+ bail:
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_put_mdio(dev->rgmii_dev, dev->rgmii_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_put_mdio(dev->zmii_dev, dev->zmii_port);
+       mutex_unlock(&dev->mdio_lock);
+
+       return err == 0 ? r : err;
+}
+
+static void __emac_mdio_write(struct emac_instance *dev, u8 id, u8 reg,
+                             u16 val)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 r = 0;
+       int n, err = -ETIMEDOUT;
+
+       mutex_lock(&dev->mdio_lock);
+
+       DBG2(dev, "mdio_write(%02x,%02x,%04x)" NL, id, reg, val);
+
+       /* Enable proper MDIO port */
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_get_mdio(dev->zmii_dev, dev->zmii_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_get_mdio(dev->rgmii_dev, dev->rgmii_port);
+
+       /* Wait for management interface to be idle */
+       n = 20;
+       while (!emac_phy_done(dev, in_be32(&p->stacr))) {
+               udelay(1);
+               if (!--n) {
+                       DBG2(dev, " -> timeout wait idle\n");
+                       goto bail;
+               }
+       }
+
+       /* Issue write command */
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               r = EMAC4_STACR_BASE(dev->opb_bus_freq);
+       else
+               r = EMAC_STACR_BASE(dev->opb_bus_freq);
+       if (emac_has_feature(dev, EMAC_FTR_STACR_OC_INVERT))
+               r |= EMAC_STACR_OC;
+       if (emac_has_feature(dev, EMAC_FTR_HAS_NEW_STACR))
+               r |= EMACX_STACR_STAC_WRITE;
+       else
+               r |= EMAC_STACR_STAC_WRITE;
+       r |= (reg & EMAC_STACR_PRA_MASK) |
+               ((id & EMAC_STACR_PCDA_MASK) << EMAC_STACR_PCDA_SHIFT) |
+               (val << EMAC_STACR_PHYD_SHIFT);
+       out_be32(&p->stacr, r);
+
+       /* Wait for write to complete */
+       n = 200;
+       while (!emac_phy_done(dev, in_be32(&p->stacr))) {
+               udelay(1);
+               if (!--n) {
+                       DBG2(dev, " -> timeout wait complete\n");
+                       goto bail;
+               }
+       }
+       err = 0;
+ bail:
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_put_mdio(dev->rgmii_dev, dev->rgmii_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_put_mdio(dev->zmii_dev, dev->zmii_port);
+       mutex_unlock(&dev->mdio_lock);
+}
+
+static int emac_mdio_read(struct net_device *ndev, int id, int reg)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int res;
+
+       res = __emac_mdio_read((dev->mdio_instance &&
+                               dev->phy.gpcs_address != id) ?
+                               dev->mdio_instance : dev,
+                              (u8) id, (u8) reg);
+       return res;
+}
+
+static void emac_mdio_write(struct net_device *ndev, int id, int reg, int val)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       __emac_mdio_write((dev->mdio_instance &&
+                          dev->phy.gpcs_address != id) ?
+                          dev->mdio_instance : dev,
+                         (u8) id, (u8) reg, (u16) val);
+}
+
+/* Tx lock BH */
+static void __emac_set_multicast_list(struct emac_instance *dev)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       u32 rmr = emac_iff2rmr(dev->ndev);
+
+       DBG(dev, "__multicast %08x" NL, rmr);
+
+       /* I decided to relax register access rules here to avoid
+        * full EMAC reset.
+        *
+        * There is a real problem with EMAC4 core if we use MWSW_001 bit
+        * in MR1 register and do a full EMAC reset.
+        * One TX BD status update is delayed and, after EMAC reset, it
+        * never happens, resulting in TX hung (it'll be recovered by TX
+        * timeout handler eventually, but this is just gross).
+        * So we either have to do full TX reset or try to cheat here :)
+        *
+        * The only required change is to RX mode register, so I *think* all
+        * we need is just to stop RX channel. This seems to work on all
+        * tested SoCs.                                                --ebs
+        *
+        * If we need the full reset, we might just trigger the workqueue
+        * and do it async... a bit nasty but should work --BenH
+        */
+       dev->mcast_pending = 0;
+       emac_rx_disable(dev);
+       if (rmr & EMAC_RMR_MAE)
+               emac_hash_mc(dev);
+       out_be32(&p->rmr, rmr);
+       emac_rx_enable(dev);
+}
+
+/* Tx lock BH */
+static void emac_set_multicast_list(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       DBG(dev, "multicast" NL);
+
+       BUG_ON(!netif_running(dev->ndev));
+
+       if (dev->no_mcast) {
+               dev->mcast_pending = 1;
+               return;
+       }
+       __emac_set_multicast_list(dev);
+}
+
+static int emac_resize_rx_ring(struct emac_instance *dev, int new_mtu)
+{
+       int rx_sync_size = emac_rx_sync_size(new_mtu);
+       int rx_skb_size = emac_rx_skb_size(new_mtu);
+       int i, ret = 0;
+       int mr1_jumbo_bit_change = 0;
+
+       mutex_lock(&dev->link_lock);
+       emac_netif_stop(dev);
+       emac_rx_disable(dev);
+       mal_disable_rx_channel(dev->mal, dev->mal_rx_chan);
+
+       if (dev->rx_sg_skb) {
+               ++dev->estats.rx_dropped_resize;
+               dev_kfree_skb(dev->rx_sg_skb);
+               dev->rx_sg_skb = NULL;
+       }
+
+       /* Make a first pass over RX ring and mark BDs ready, dropping
+        * non-processed packets on the way. We need this as a separate pass
+        * to simplify error recovery in the case of allocation failure later.
+        */
+       for (i = 0; i < NUM_RX_BUFF; ++i) {
+               if (dev->rx_desc[i].ctrl & MAL_RX_CTRL_FIRST)
+                       ++dev->estats.rx_dropped_resize;
+
+               dev->rx_desc[i].data_len = 0;
+               dev->rx_desc[i].ctrl = MAL_RX_CTRL_EMPTY |
+                   (i == (NUM_RX_BUFF - 1) ? MAL_RX_CTRL_WRAP : 0);
+       }
+
+       /* Reallocate RX ring only if bigger skb buffers are required */
+       if (rx_skb_size <= dev->rx_skb_size)
+               goto skip;
+
+       /* Second pass, allocate new skbs */
+       for (i = 0; i < NUM_RX_BUFF; ++i) {
+               struct sk_buff *skb = alloc_skb(rx_skb_size, GFP_ATOMIC);
+               if (!skb) {
+                       ret = -ENOMEM;
+                       goto oom;
+               }
+
+               BUG_ON(!dev->rx_skb[i]);
+               dev_kfree_skb(dev->rx_skb[i]);
+
+               skb_reserve(skb, EMAC_RX_SKB_HEADROOM + 2);
+               dev->rx_desc[i].data_ptr =
+                   dma_map_single(&dev->ofdev->dev, skb->data - 2, rx_sync_size,
+                                  DMA_FROM_DEVICE) + 2;
+               dev->rx_skb[i] = skb;
+       }
+ skip:
+       /* Check if we need to change "Jumbo" bit in MR1 */
+       if (emac_has_feature(dev, EMAC_APM821XX_REQ_JUMBO_FRAME_SIZE)) {
+               mr1_jumbo_bit_change = (new_mtu > ETH_DATA_LEN) ||
+                               (dev->ndev->mtu > ETH_DATA_LEN);
+       } else {
+               mr1_jumbo_bit_change = (new_mtu > ETH_DATA_LEN) ^
+                               (dev->ndev->mtu > ETH_DATA_LEN);
+       }
+
+       if (mr1_jumbo_bit_change) {
+               /* This is to prevent starting RX channel in emac_rx_enable() */
+               set_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags);
+
+               dev->ndev->mtu = new_mtu;
+               emac_full_tx_reset(dev);
+       }
+
+       mal_set_rcbs(dev->mal, dev->mal_rx_chan, emac_rx_size(new_mtu));
+ oom:
+       /* Restart RX */
+       clear_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags);
+       dev->rx_slot = 0;
+       mal_enable_rx_channel(dev->mal, dev->mal_rx_chan);
+       emac_rx_enable(dev);
+       emac_netif_start(dev);
+       mutex_unlock(&dev->link_lock);
+
+       return ret;
+}
+
+/* Process ctx, rtnl_lock semaphore */
+static int emac_change_mtu(struct net_device *ndev, int new_mtu)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int ret = 0;
+
+       if (new_mtu < EMAC_MIN_MTU || new_mtu > dev->max_mtu)
+               return -EINVAL;
+
+       DBG(dev, "change_mtu(%d)" NL, new_mtu);
+
+       if (netif_running(ndev)) {
+               /* Check if we really need to reinitialize RX ring */
+               if (emac_rx_skb_size(ndev->mtu) != emac_rx_skb_size(new_mtu))
+                       ret = emac_resize_rx_ring(dev, new_mtu);
+       }
+
+       if (!ret) {
+               ndev->mtu = new_mtu;
+               dev->rx_skb_size = emac_rx_skb_size(new_mtu);
+               dev->rx_sync_size = emac_rx_sync_size(new_mtu);
+       }
+
+       return ret;
+}
+
+static void emac_clean_tx_ring(struct emac_instance *dev)
+{
+       int i;
+
+       for (i = 0; i < NUM_TX_BUFF; ++i) {
+               if (dev->tx_skb[i]) {
+                       dev_kfree_skb(dev->tx_skb[i]);
+                       dev->tx_skb[i] = NULL;
+                       if (dev->tx_desc[i].ctrl & MAL_TX_CTRL_READY)
+                               ++dev->estats.tx_dropped;
+               }
+               dev->tx_desc[i].ctrl = 0;
+               dev->tx_desc[i].data_ptr = 0;
+       }
+}
+
+static void emac_clean_rx_ring(struct emac_instance *dev)
+{
+       int i;
+
+       for (i = 0; i < NUM_RX_BUFF; ++i)
+               if (dev->rx_skb[i]) {
+                       dev->rx_desc[i].ctrl = 0;
+                       dev_kfree_skb(dev->rx_skb[i]);
+                       dev->rx_skb[i] = NULL;
+                       dev->rx_desc[i].data_ptr = 0;
+               }
+
+       if (dev->rx_sg_skb) {
+               dev_kfree_skb(dev->rx_sg_skb);
+               dev->rx_sg_skb = NULL;
+       }
+}
+
+static inline int emac_alloc_rx_skb(struct emac_instance *dev, int slot,
+                                   gfp_t flags)
+{
+       struct sk_buff *skb = alloc_skb(dev->rx_skb_size, flags);
+       if (unlikely(!skb))
+               return -ENOMEM;
+
+       dev->rx_skb[slot] = skb;
+       dev->rx_desc[slot].data_len = 0;
+
+       skb_reserve(skb, EMAC_RX_SKB_HEADROOM + 2);
+       dev->rx_desc[slot].data_ptr =
+           dma_map_single(&dev->ofdev->dev, skb->data - 2, dev->rx_sync_size,
+                          DMA_FROM_DEVICE) + 2;
+       wmb();
+       dev->rx_desc[slot].ctrl = MAL_RX_CTRL_EMPTY |
+           (slot == (NUM_RX_BUFF - 1) ? MAL_RX_CTRL_WRAP : 0);
+
+       return 0;
+}
+
+static void emac_print_link_status(struct emac_instance *dev)
+{
+       if (netif_carrier_ok(dev->ndev))
+               printk(KERN_INFO "%s: link is up, %d %s%s\n",
+                      dev->ndev->name, dev->phy.speed,
+                      dev->phy.duplex == DUPLEX_FULL ? "FDX" : "HDX",
+                      dev->phy.pause ? ", pause enabled" :
+                      dev->phy.asym_pause ? ", asymmetric pause enabled" : "");
+       else
+               printk(KERN_INFO "%s: link is down\n", dev->ndev->name);
+}
+
+/* Process ctx, rtnl_lock semaphore */
+static int emac_open(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int err, i;
+
+       DBG(dev, "open" NL);
+
+       /* Setup error IRQ handler */
+       err = request_irq(dev->emac_irq, emac_irq, 0, "EMAC", dev);
+       if (err) {
+               printk(KERN_ERR "%s: failed to request IRQ %d\n",
+                      ndev->name, dev->emac_irq);
+               return err;
+       }
+
+       /* Allocate RX ring */
+       for (i = 0; i < NUM_RX_BUFF; ++i)
+               if (emac_alloc_rx_skb(dev, i, GFP_KERNEL)) {
+                       printk(KERN_ERR "%s: failed to allocate RX ring\n",
+                              ndev->name);
+                       goto oom;
+               }
+
+       dev->tx_cnt = dev->tx_slot = dev->ack_slot = dev->rx_slot = 0;
+       clear_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags);
+       dev->rx_sg_skb = NULL;
+
+       mutex_lock(&dev->link_lock);
+       dev->opened = 1;
+
+       /* Start PHY polling now.
+        */
+       if (dev->phy.address >= 0) {
+               int link_poll_interval;
+               if (dev->phy.def->ops->poll_link(&dev->phy)) {
+                       dev->phy.def->ops->read_link(&dev->phy);
+                       emac_rx_clk_default(dev);
+                       netif_carrier_on(dev->ndev);
+                       link_poll_interval = PHY_POLL_LINK_ON;
+               } else {
+                       emac_rx_clk_tx(dev);
+                       netif_carrier_off(dev->ndev);
+                       link_poll_interval = PHY_POLL_LINK_OFF;
+               }
+               dev->link_polling = 1;
+               wmb();
+               schedule_delayed_work(&dev->link_work, link_poll_interval);
+               emac_print_link_status(dev);
+       } else
+               netif_carrier_on(dev->ndev);
+
+       /* Required for Pause packet support in EMAC */
+       dev_mc_add_global(ndev, default_mcast_addr);
+
+       emac_configure(dev);
+       mal_poll_add(dev->mal, &dev->commac);
+       mal_enable_tx_channel(dev->mal, dev->mal_tx_chan);
+       mal_set_rcbs(dev->mal, dev->mal_rx_chan, emac_rx_size(ndev->mtu));
+       mal_enable_rx_channel(dev->mal, dev->mal_rx_chan);
+       emac_tx_enable(dev);
+       emac_rx_enable(dev);
+       emac_netif_start(dev);
+
+       mutex_unlock(&dev->link_lock);
+
+       return 0;
+ oom:
+       emac_clean_rx_ring(dev);
+       free_irq(dev->emac_irq, dev);
+
+       return -ENOMEM;
+}
+
+/* BHs disabled */
+#if 0
+static int emac_link_differs(struct emac_instance *dev)
+{
+       u32 r = in_be32(&dev->emacp->mr1);
+
+       int duplex = r & EMAC_MR1_FDE ? DUPLEX_FULL : DUPLEX_HALF;
+       int speed, pause, asym_pause;
+
+       if (r & EMAC_MR1_MF_1000)
+               speed = SPEED_1000;
+       else if (r & EMAC_MR1_MF_100)
+               speed = SPEED_100;
+       else
+               speed = SPEED_10;
+
+       switch (r & (EMAC_MR1_EIFC | EMAC_MR1_APP)) {
+       case (EMAC_MR1_EIFC | EMAC_MR1_APP):
+               pause = 1;
+               asym_pause = 0;
+               break;
+       case EMAC_MR1_APP:
+               pause = 0;
+               asym_pause = 1;
+               break;
+       default:
+               pause = asym_pause = 0;
+       }
+       return speed != dev->phy.speed || duplex != dev->phy.duplex ||
+           pause != dev->phy.pause || asym_pause != dev->phy.asym_pause;
+}
+#endif
+
+static void emac_link_timer(struct work_struct *work)
+{
+       struct emac_instance *dev =
+               container_of(to_delayed_work(work),
+                            struct emac_instance, link_work);
+       int link_poll_interval;
+
+       mutex_lock(&dev->link_lock);
+       DBG2(dev, "link timer" NL);
+
+       if (!dev->opened)
+               goto bail;
+
+       if (dev->phy.def->ops->poll_link(&dev->phy)) {
+               if (!netif_carrier_ok(dev->ndev)) {
+                       emac_rx_clk_default(dev);
+                       /* Get new link parameters */
+                       dev->phy.def->ops->read_link(&dev->phy);
+
+                       netif_carrier_on(dev->ndev);
+                       emac_netif_stop(dev);
+                       emac_full_tx_reset(dev);
+                       emac_netif_start(dev);
+                       emac_print_link_status(dev);
+               }
+               link_poll_interval = PHY_POLL_LINK_ON;
+       } else {
+               if (netif_carrier_ok(dev->ndev)) {
+                       emac_rx_clk_tx(dev);
+                       netif_carrier_off(dev->ndev);
+                       netif_tx_disable(dev->ndev);
+                       emac_reinitialize(dev);
+                       emac_print_link_status(dev);
+               }
+               link_poll_interval = PHY_POLL_LINK_OFF;
+       }
+       schedule_delayed_work(&dev->link_work, link_poll_interval);
+ bail:
+       mutex_unlock(&dev->link_lock);
+}
+
+static void emac_force_link_update(struct emac_instance *dev)
+{
+       netif_carrier_off(dev->ndev);
+       smp_rmb();
+       if (dev->link_polling) {
+               cancel_delayed_work_sync(&dev->link_work);
+               if (dev->link_polling)
+                       schedule_delayed_work(&dev->link_work,  PHY_POLL_LINK_OFF);
+       }
+}
+
+/* Process ctx, rtnl_lock semaphore */
+static int emac_close(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       DBG(dev, "close" NL);
+
+       if (dev->phy.address >= 0) {
+               dev->link_polling = 0;
+               cancel_delayed_work_sync(&dev->link_work);
+       }
+       mutex_lock(&dev->link_lock);
+       emac_netif_stop(dev);
+       dev->opened = 0;
+       mutex_unlock(&dev->link_lock);
+
+       emac_rx_disable(dev);
+       emac_tx_disable(dev);
+       mal_disable_rx_channel(dev->mal, dev->mal_rx_chan);
+       mal_disable_tx_channel(dev->mal, dev->mal_tx_chan);
+       mal_poll_del(dev->mal, &dev->commac);
+
+       emac_clean_tx_ring(dev);
+       emac_clean_rx_ring(dev);
+
+       free_irq(dev->emac_irq, dev);
+
+       netif_carrier_off(ndev);
+
+       return 0;
+}
+
+static inline u16 emac_tx_csum(struct emac_instance *dev,
+                              struct sk_buff *skb)
+{
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH) &&
+               (skb->ip_summed == CHECKSUM_PARTIAL)) {
+               ++dev->stats.tx_packets_csum;
+               return EMAC_TX_CTRL_TAH_CSUM;
+       }
+       return 0;
+}
+
+static inline int emac_xmit_finish(struct emac_instance *dev, int len)
+{
+       struct emac_regs __iomem *p = dev->emacp;
+       struct net_device *ndev = dev->ndev;
+
+       /* Send the packet out. If the if makes a significant perf
+        * difference, then we can store the TMR0 value in "dev"
+        * instead
+        */
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4))
+               out_be32(&p->tmr0, EMAC4_TMR0_XMIT);
+       else
+               out_be32(&p->tmr0, EMAC_TMR0_XMIT);
+
+       if (unlikely(++dev->tx_cnt == NUM_TX_BUFF)) {
+               netif_stop_queue(ndev);
+               DBG2(dev, "stopped TX queue" NL);
+       }
+
+       ndev->trans_start = jiffies;
+       ++dev->stats.tx_packets;
+       dev->stats.tx_bytes += len;
+
+       return NETDEV_TX_OK;
+}
+
+/* Tx lock BH */
+static int emac_start_xmit(struct sk_buff *skb, struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       unsigned int len = skb->len;
+       int slot;
+
+       u16 ctrl = EMAC_TX_CTRL_GFCS | EMAC_TX_CTRL_GP | MAL_TX_CTRL_READY |
+           MAL_TX_CTRL_LAST | emac_tx_csum(dev, skb);
+
+       slot = dev->tx_slot++;
+       if (dev->tx_slot == NUM_TX_BUFF) {
+               dev->tx_slot = 0;
+               ctrl |= MAL_TX_CTRL_WRAP;
+       }
+
+       DBG2(dev, "xmit(%u) %d" NL, len, slot);
+
+       dev->tx_skb[slot] = skb;
+       dev->tx_desc[slot].data_ptr = dma_map_single(&dev->ofdev->dev,
+                                                    skb->data, len,
+                                                    DMA_TO_DEVICE);
+       dev->tx_desc[slot].data_len = (u16) len;
+       wmb();
+       dev->tx_desc[slot].ctrl = ctrl;
+
+       return emac_xmit_finish(dev, len);
+}
+
+static inline int emac_xmit_split(struct emac_instance *dev, int slot,
+                                 u32 pd, int len, int last, u16 base_ctrl)
+{
+       while (1) {
+               u16 ctrl = base_ctrl;
+               int chunk = min(len, MAL_MAX_TX_SIZE);
+               len -= chunk;
+
+               slot = (slot + 1) % NUM_TX_BUFF;
+
+               if (last && !len)
+                       ctrl |= MAL_TX_CTRL_LAST;
+               if (slot == NUM_TX_BUFF - 1)
+                       ctrl |= MAL_TX_CTRL_WRAP;
+
+               dev->tx_skb[slot] = NULL;
+               dev->tx_desc[slot].data_ptr = pd;
+               dev->tx_desc[slot].data_len = (u16) chunk;
+               dev->tx_desc[slot].ctrl = ctrl;
+               ++dev->tx_cnt;
+
+               if (!len)
+                       break;
+
+               pd += chunk;
+       }
+       return slot;
+}
+
+/* Tx lock BH disabled (SG version for TAH equipped EMACs) */
+static int emac_start_xmit_sg(struct sk_buff *skb, struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int nr_frags = skb_shinfo(skb)->nr_frags;
+       int len = skb->len, chunk;
+       int slot, i;
+       u16 ctrl;
+       u32 pd;
+
+       /* This is common "fast" path */
+       if (likely(!nr_frags && len <= MAL_MAX_TX_SIZE))
+               return emac_start_xmit(skb, ndev);
+
+       len -= skb->data_len;
+
+       /* Note, this is only an *estimation*, we can still run out of empty
+        * slots because of the additional fragmentation into
+        * MAL_MAX_TX_SIZE-sized chunks
+        */
+       if (unlikely(dev->tx_cnt + nr_frags + mal_tx_chunks(len) > NUM_TX_BUFF))
+               goto stop_queue;
+
+       ctrl = EMAC_TX_CTRL_GFCS | EMAC_TX_CTRL_GP | MAL_TX_CTRL_READY |
+           emac_tx_csum(dev, skb);
+       slot = dev->tx_slot;
+
+       /* skb data */
+       dev->tx_skb[slot] = NULL;
+       chunk = min(len, MAL_MAX_TX_SIZE);
+       dev->tx_desc[slot].data_ptr = pd =
+           dma_map_single(&dev->ofdev->dev, skb->data, len, DMA_TO_DEVICE);
+       dev->tx_desc[slot].data_len = (u16) chunk;
+       len -= chunk;
+       if (unlikely(len))
+               slot = emac_xmit_split(dev, slot, pd + chunk, len, !nr_frags,
+                                      ctrl);
+       /* skb fragments */
+       for (i = 0; i < nr_frags; ++i) {
+               struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[i];
+               len = skb_frag_size(frag);
+
+               if (unlikely(dev->tx_cnt + mal_tx_chunks(len) >= NUM_TX_BUFF))
+                       goto undo_frame;
+
+               pd = skb_frag_dma_map(&dev->ofdev->dev, frag, 0, len,
+                                     DMA_TO_DEVICE);
+
+               slot = emac_xmit_split(dev, slot, pd, len, i == nr_frags - 1,
+                                      ctrl);
+       }
+
+       DBG2(dev, "xmit_sg(%u) %d - %d" NL, skb->len, dev->tx_slot, slot);
+
+       /* Attach skb to the last slot so we don't release it too early */
+       dev->tx_skb[slot] = skb;
+
+       /* Send the packet out */
+       if (dev->tx_slot == NUM_TX_BUFF - 1)
+               ctrl |= MAL_TX_CTRL_WRAP;
+       wmb();
+       dev->tx_desc[dev->tx_slot].ctrl = ctrl;
+       dev->tx_slot = (slot + 1) % NUM_TX_BUFF;
+
+       return emac_xmit_finish(dev, skb->len);
+
+ undo_frame:
+       /* Well, too bad. Our previous estimation was overly optimistic.
+        * Undo everything.
+        */
+       while (slot != dev->tx_slot) {
+               dev->tx_desc[slot].ctrl = 0;
+               --dev->tx_cnt;
+               if (--slot < 0)
+                       slot = NUM_TX_BUFF - 1;
+       }
+       ++dev->estats.tx_undo;
+
+ stop_queue:
+       netif_stop_queue(ndev);
+       DBG2(dev, "stopped TX queue" NL);
+       return NETDEV_TX_BUSY;
+}
+
+/* Tx lock BHs */
+static void emac_parse_tx_error(struct emac_instance *dev, u16 ctrl)
+{
+       struct emac_error_stats *st = &dev->estats;
+
+       DBG(dev, "BD TX error %04x" NL, ctrl);
+
+       ++st->tx_bd_errors;
+       if (ctrl & EMAC_TX_ST_BFCS)
+               ++st->tx_bd_bad_fcs;
+       if (ctrl & EMAC_TX_ST_LCS)
+               ++st->tx_bd_carrier_loss;
+       if (ctrl & EMAC_TX_ST_ED)
+               ++st->tx_bd_excessive_deferral;
+       if (ctrl & EMAC_TX_ST_EC)
+               ++st->tx_bd_excessive_collisions;
+       if (ctrl & EMAC_TX_ST_LC)
+               ++st->tx_bd_late_collision;
+       if (ctrl & EMAC_TX_ST_MC)
+               ++st->tx_bd_multple_collisions;
+       if (ctrl & EMAC_TX_ST_SC)
+               ++st->tx_bd_single_collision;
+       if (ctrl & EMAC_TX_ST_UR)
+               ++st->tx_bd_underrun;
+       if (ctrl & EMAC_TX_ST_SQE)
+               ++st->tx_bd_sqe;
+}
+
+static void emac_poll_tx(void *param)
+{
+       struct emac_instance *dev = param;
+       u32 bad_mask;
+
+       DBG2(dev, "poll_tx, %d %d" NL, dev->tx_cnt, dev->ack_slot);
+
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+               bad_mask = EMAC_IS_BAD_TX_TAH;
+       else
+               bad_mask = EMAC_IS_BAD_TX;
+
+       netif_tx_lock_bh(dev->ndev);
+       if (dev->tx_cnt) {
+               u16 ctrl;
+               int slot = dev->ack_slot, n = 0;
+       again:
+               ctrl = dev->tx_desc[slot].ctrl;
+               if (!(ctrl & MAL_TX_CTRL_READY)) {
+                       struct sk_buff *skb = dev->tx_skb[slot];
+                       ++n;
+
+                       if (skb) {
+                               dev_kfree_skb(skb);
+                               dev->tx_skb[slot] = NULL;
+                       }
+                       slot = (slot + 1) % NUM_TX_BUFF;
+
+                       if (unlikely(ctrl & bad_mask))
+                               emac_parse_tx_error(dev, ctrl);
+
+                       if (--dev->tx_cnt)
+                               goto again;
+               }
+               if (n) {
+                       dev->ack_slot = slot;
+                       if (netif_queue_stopped(dev->ndev) &&
+                           dev->tx_cnt < EMAC_TX_WAKEUP_THRESH)
+                               netif_wake_queue(dev->ndev);
+
+                       DBG2(dev, "tx %d pkts" NL, n);
+               }
+       }
+       netif_tx_unlock_bh(dev->ndev);
+}
+
+static inline void emac_recycle_rx_skb(struct emac_instance *dev, int slot,
+                                      int len)
+{
+       struct sk_buff *skb = dev->rx_skb[slot];
+
+       DBG2(dev, "recycle %d %d" NL, slot, len);
+
+       if (len)
+               dma_map_single(&dev->ofdev->dev, skb->data - 2,
+                              EMAC_DMA_ALIGN(len + 2), DMA_FROM_DEVICE);
+
+       dev->rx_desc[slot].data_len = 0;
+       wmb();
+       dev->rx_desc[slot].ctrl = MAL_RX_CTRL_EMPTY |
+           (slot == (NUM_RX_BUFF - 1) ? MAL_RX_CTRL_WRAP : 0);
+}
+
+static void emac_parse_rx_error(struct emac_instance *dev, u16 ctrl)
+{
+       struct emac_error_stats *st = &dev->estats;
+
+       DBG(dev, "BD RX error %04x" NL, ctrl);
+
+       ++st->rx_bd_errors;
+       if (ctrl & EMAC_RX_ST_OE)
+               ++st->rx_bd_overrun;
+       if (ctrl & EMAC_RX_ST_BP)
+               ++st->rx_bd_bad_packet;
+       if (ctrl & EMAC_RX_ST_RP)
+               ++st->rx_bd_runt_packet;
+       if (ctrl & EMAC_RX_ST_SE)
+               ++st->rx_bd_short_event;
+       if (ctrl & EMAC_RX_ST_AE)
+               ++st->rx_bd_alignment_error;
+       if (ctrl & EMAC_RX_ST_BFCS)
+               ++st->rx_bd_bad_fcs;
+       if (ctrl & EMAC_RX_ST_PTL)
+               ++st->rx_bd_packet_too_long;
+       if (ctrl & EMAC_RX_ST_ORE)
+               ++st->rx_bd_out_of_range;
+       if (ctrl & EMAC_RX_ST_IRE)
+               ++st->rx_bd_in_range;
+}
+
+static inline void emac_rx_csum(struct emac_instance *dev,
+                               struct sk_buff *skb, u16 ctrl)
+{
+#ifdef CONFIG_IBM_EMAC_TAH
+       if (!ctrl && dev->tah_dev) {
+               skb->ip_summed = CHECKSUM_UNNECESSARY;
+               ++dev->stats.rx_packets_csum;
+       }
+#endif
+}
+
+static inline int emac_rx_sg_append(struct emac_instance *dev, int slot)
+{
+       if (likely(dev->rx_sg_skb != NULL)) {
+               int len = dev->rx_desc[slot].data_len;
+               int tot_len = dev->rx_sg_skb->len + len;
+
+               if (unlikely(tot_len + 2 > dev->rx_skb_size)) {
+                       ++dev->estats.rx_dropped_mtu;
+                       dev_kfree_skb(dev->rx_sg_skb);
+                       dev->rx_sg_skb = NULL;
+               } else {
+                       memcpy(skb_tail_pointer(dev->rx_sg_skb),
+                                        dev->rx_skb[slot]->data, len);
+                       skb_put(dev->rx_sg_skb, len);
+                       emac_recycle_rx_skb(dev, slot, len);
+                       return 0;
+               }
+       }
+       emac_recycle_rx_skb(dev, slot, 0);
+       return -1;
+}
+
+/* NAPI poll context */
+static int emac_poll_rx(void *param, int budget)
+{
+       struct emac_instance *dev = param;
+       int slot = dev->rx_slot, received = 0;
+
+       DBG2(dev, "poll_rx(%d)" NL, budget);
+
+ again:
+       while (budget > 0) {
+               int len;
+               struct sk_buff *skb;
+               u16 ctrl = dev->rx_desc[slot].ctrl;
+
+               if (ctrl & MAL_RX_CTRL_EMPTY)
+                       break;
+
+               skb = dev->rx_skb[slot];
+               mb();
+               len = dev->rx_desc[slot].data_len;
+
+               if (unlikely(!MAL_IS_SINGLE_RX(ctrl)))
+                       goto sg;
+
+               ctrl &= EMAC_BAD_RX_MASK;
+               if (unlikely(ctrl && ctrl != EMAC_RX_TAH_BAD_CSUM)) {
+                       emac_parse_rx_error(dev, ctrl);
+                       ++dev->estats.rx_dropped_error;
+                       emac_recycle_rx_skb(dev, slot, 0);
+                       len = 0;
+                       goto next;
+               }
+
+               if (len < ETH_HLEN) {
+                       ++dev->estats.rx_dropped_stack;
+                       emac_recycle_rx_skb(dev, slot, len);
+                       goto next;
+               }
+
+               if (len && len < EMAC_RX_COPY_THRESH) {
+                       struct sk_buff *copy_skb =
+                           alloc_skb(len + EMAC_RX_SKB_HEADROOM + 2, GFP_ATOMIC);
+                       if (unlikely(!copy_skb))
+                               goto oom;
+
+                       skb_reserve(copy_skb, EMAC_RX_SKB_HEADROOM + 2);
+                       memcpy(copy_skb->data - 2, skb->data - 2, len + 2);
+                       emac_recycle_rx_skb(dev, slot, len);
+                       skb = copy_skb;
+               } else if (unlikely(emac_alloc_rx_skb(dev, slot, GFP_ATOMIC)))
+                       goto oom;
+
+               skb_put(skb, len);
+       push_packet:
+               skb->protocol = eth_type_trans(skb, dev->ndev);
+               emac_rx_csum(dev, skb, ctrl);
+
+               if (unlikely(netif_receive_skb(skb) == NET_RX_DROP))
+                       ++dev->estats.rx_dropped_stack;
+       next:
+               ++dev->stats.rx_packets;
+       skip:
+               dev->stats.rx_bytes += len;
+               slot = (slot + 1) % NUM_RX_BUFF;
+               --budget;
+               ++received;
+               continue;
+       sg:
+               if (ctrl & MAL_RX_CTRL_FIRST) {
+                       BUG_ON(dev->rx_sg_skb);
+                       if (unlikely(emac_alloc_rx_skb(dev, slot, GFP_ATOMIC))) {
+                               DBG(dev, "rx OOM %d" NL, slot);
+                               ++dev->estats.rx_dropped_oom;
+                               emac_recycle_rx_skb(dev, slot, 0);
+                       } else {
+                               dev->rx_sg_skb = skb;
+                               skb_put(skb, len);
+                       }
+               } else if (!emac_rx_sg_append(dev, slot) &&
+                          (ctrl & MAL_RX_CTRL_LAST)) {
+
+                       skb = dev->rx_sg_skb;
+                       dev->rx_sg_skb = NULL;
+
+                       ctrl &= EMAC_BAD_RX_MASK;
+                       if (unlikely(ctrl && ctrl != EMAC_RX_TAH_BAD_CSUM)) {
+                               emac_parse_rx_error(dev, ctrl);
+                               ++dev->estats.rx_dropped_error;
+                               dev_kfree_skb(skb);
+                               len = 0;
+                       } else
+                               goto push_packet;
+               }
+               goto skip;
+       oom:
+               DBG(dev, "rx OOM %d" NL, slot);
+               /* Drop the packet and recycle skb */
+               ++dev->estats.rx_dropped_oom;
+               emac_recycle_rx_skb(dev, slot, 0);
+               goto next;
+       }
+
+       if (received) {
+               DBG2(dev, "rx %d BDs" NL, received);
+               dev->rx_slot = slot;
+       }
+
+       if (unlikely(budget && test_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags))) {
+               mb();
+               if (!(dev->rx_desc[slot].ctrl & MAL_RX_CTRL_EMPTY)) {
+                       DBG2(dev, "rx restart" NL);
+                       received = 0;
+                       goto again;
+               }
+
+               if (dev->rx_sg_skb) {
+                       DBG2(dev, "dropping partial rx packet" NL);
+                       ++dev->estats.rx_dropped_error;
+                       dev_kfree_skb(dev->rx_sg_skb);
+                       dev->rx_sg_skb = NULL;
+               }
+
+               clear_bit(MAL_COMMAC_RX_STOPPED, &dev->commac.flags);
+               mal_enable_rx_channel(dev->mal, dev->mal_rx_chan);
+               emac_rx_enable(dev);
+               dev->rx_slot = 0;
+       }
+       return received;
+}
+
+/* NAPI poll context */
+static int emac_peek_rx(void *param)
+{
+       struct emac_instance *dev = param;
+
+       return !(dev->rx_desc[dev->rx_slot].ctrl & MAL_RX_CTRL_EMPTY);
+}
+
+/* NAPI poll context */
+static int emac_peek_rx_sg(void *param)
+{
+       struct emac_instance *dev = param;
+
+       int slot = dev->rx_slot;
+       while (1) {
+               u16 ctrl = dev->rx_desc[slot].ctrl;
+               if (ctrl & MAL_RX_CTRL_EMPTY)
+                       return 0;
+               else if (ctrl & MAL_RX_CTRL_LAST)
+                       return 1;
+
+               slot = (slot + 1) % NUM_RX_BUFF;
+
+               /* I'm just being paranoid here :) */
+               if (unlikely(slot == dev->rx_slot))
+                       return 0;
+       }
+}
+
+/* Hard IRQ */
+static void emac_rxde(void *param)
+{
+       struct emac_instance *dev = param;
+
+       ++dev->estats.rx_stopped;
+       emac_rx_disable_async(dev);
+}
+
+/* Hard IRQ */
+static irqreturn_t emac_irq(int irq, void *dev_instance)
+{
+       struct emac_instance *dev = dev_instance;
+       struct emac_regs __iomem *p = dev->emacp;
+       struct emac_error_stats *st = &dev->estats;
+       u32 isr;
+
+       spin_lock(&dev->lock);
+
+       isr = in_be32(&p->isr);
+       out_be32(&p->isr, isr);
+
+       DBG(dev, "isr = %08x" NL, isr);
+
+       if (isr & EMAC4_ISR_TXPE)
+               ++st->tx_parity;
+       if (isr & EMAC4_ISR_RXPE)
+               ++st->rx_parity;
+       if (isr & EMAC4_ISR_TXUE)
+               ++st->tx_underrun;
+       if (isr & EMAC4_ISR_RXOE)
+               ++st->rx_fifo_overrun;
+       if (isr & EMAC_ISR_OVR)
+               ++st->rx_overrun;
+       if (isr & EMAC_ISR_BP)
+               ++st->rx_bad_packet;
+       if (isr & EMAC_ISR_RP)
+               ++st->rx_runt_packet;
+       if (isr & EMAC_ISR_SE)
+               ++st->rx_short_event;
+       if (isr & EMAC_ISR_ALE)
+               ++st->rx_alignment_error;
+       if (isr & EMAC_ISR_BFCS)
+               ++st->rx_bad_fcs;
+       if (isr & EMAC_ISR_PTLE)
+               ++st->rx_packet_too_long;
+       if (isr & EMAC_ISR_ORE)
+               ++st->rx_out_of_range;
+       if (isr & EMAC_ISR_IRE)
+               ++st->rx_in_range;
+       if (isr & EMAC_ISR_SQE)
+               ++st->tx_sqe;
+       if (isr & EMAC_ISR_TE)
+               ++st->tx_errors;
+
+       spin_unlock(&dev->lock);
+
+       return IRQ_HANDLED;
+}
+
+static struct net_device_stats *emac_stats(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       struct emac_stats *st = &dev->stats;
+       struct emac_error_stats *est = &dev->estats;
+       struct net_device_stats *nst = &dev->nstats;
+       unsigned long flags;
+
+       DBG2(dev, "stats" NL);
+
+       /* Compute "legacy" statistics */
+       spin_lock_irqsave(&dev->lock, flags);
+       nst->rx_packets = (unsigned long)st->rx_packets;
+       nst->rx_bytes = (unsigned long)st->rx_bytes;
+       nst->tx_packets = (unsigned long)st->tx_packets;
+       nst->tx_bytes = (unsigned long)st->tx_bytes;
+       nst->rx_dropped = (unsigned long)(est->rx_dropped_oom +
+                                         est->rx_dropped_error +
+                                         est->rx_dropped_resize +
+                                         est->rx_dropped_mtu);
+       nst->tx_dropped = (unsigned long)est->tx_dropped;
+
+       nst->rx_errors = (unsigned long)est->rx_bd_errors;
+       nst->rx_fifo_errors = (unsigned long)(est->rx_bd_overrun +
+                                             est->rx_fifo_overrun +
+                                             est->rx_overrun);
+       nst->rx_frame_errors = (unsigned long)(est->rx_bd_alignment_error +
+                                              est->rx_alignment_error);
+       nst->rx_crc_errors = (unsigned long)(est->rx_bd_bad_fcs +
+                                            est->rx_bad_fcs);
+       nst->rx_length_errors = (unsigned long)(est->rx_bd_runt_packet +
+                                               est->rx_bd_short_event +
+                                               est->rx_bd_packet_too_long +
+                                               est->rx_bd_out_of_range +
+                                               est->rx_bd_in_range +
+                                               est->rx_runt_packet +
+                                               est->rx_short_event +
+                                               est->rx_packet_too_long +
+                                               est->rx_out_of_range +
+                                               est->rx_in_range);
+
+       nst->tx_errors = (unsigned long)(est->tx_bd_errors + est->tx_errors);
+       nst->tx_fifo_errors = (unsigned long)(est->tx_bd_underrun +
+                                             est->tx_underrun);
+       nst->tx_carrier_errors = (unsigned long)est->tx_bd_carrier_loss;
+       nst->collisions = (unsigned long)(est->tx_bd_excessive_deferral +
+                                         est->tx_bd_excessive_collisions +
+                                         est->tx_bd_late_collision +
+                                         est->tx_bd_multple_collisions);
+       spin_unlock_irqrestore(&dev->lock, flags);
+       return nst;
+}
+
+static struct mal_commac_ops emac_commac_ops = {
+       .poll_tx = &emac_poll_tx,
+       .poll_rx = &emac_poll_rx,
+       .peek_rx = &emac_peek_rx,
+       .rxde = &emac_rxde,
+};
+
+static struct mal_commac_ops emac_commac_sg_ops = {
+       .poll_tx = &emac_poll_tx,
+       .poll_rx = &emac_poll_rx,
+       .peek_rx = &emac_peek_rx_sg,
+       .rxde = &emac_rxde,
+};
+
+/* Ethtool support */
+static int emac_ethtool_get_settings(struct net_device *ndev,
+                                    struct ethtool_cmd *cmd)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       cmd->supported = dev->phy.features;
+       cmd->port = PORT_MII;
+       cmd->phy_address = dev->phy.address;
+       cmd->transceiver =
+           dev->phy.address >= 0 ? XCVR_EXTERNAL : XCVR_INTERNAL;
+
+       mutex_lock(&dev->link_lock);
+       cmd->advertising = dev->phy.advertising;
+       cmd->autoneg = dev->phy.autoneg;
+       cmd->speed = dev->phy.speed;
+       cmd->duplex = dev->phy.duplex;
+       mutex_unlock(&dev->link_lock);
+
+       return 0;
+}
+
+static int emac_ethtool_set_settings(struct net_device *ndev,
+                                    struct ethtool_cmd *cmd)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       u32 f = dev->phy.features;
+
+       DBG(dev, "set_settings(%d, %d, %d, 0x%08x)" NL,
+           cmd->autoneg, cmd->speed, cmd->duplex, cmd->advertising);
+
+       /* Basic sanity checks */
+       if (dev->phy.address < 0)
+               return -EOPNOTSUPP;
+       if (cmd->autoneg != AUTONEG_ENABLE && cmd->autoneg != AUTONEG_DISABLE)
+               return -EINVAL;
+       if (cmd->autoneg == AUTONEG_ENABLE && cmd->advertising == 0)
+               return -EINVAL;
+       if (cmd->duplex != DUPLEX_HALF && cmd->duplex != DUPLEX_FULL)
+               return -EINVAL;
+
+       if (cmd->autoneg == AUTONEG_DISABLE) {
+               switch (cmd->speed) {
+               case SPEED_10:
+                       if (cmd->duplex == DUPLEX_HALF &&
+                           !(f & SUPPORTED_10baseT_Half))
+                               return -EINVAL;
+                       if (cmd->duplex == DUPLEX_FULL &&
+                           !(f & SUPPORTED_10baseT_Full))
+                               return -EINVAL;
+                       break;
+               case SPEED_100:
+                       if (cmd->duplex == DUPLEX_HALF &&
+                           !(f & SUPPORTED_100baseT_Half))
+                               return -EINVAL;
+                       if (cmd->duplex == DUPLEX_FULL &&
+                           !(f & SUPPORTED_100baseT_Full))
+                               return -EINVAL;
+                       break;
+               case SPEED_1000:
+                       if (cmd->duplex == DUPLEX_HALF &&
+                           !(f & SUPPORTED_1000baseT_Half))
+                               return -EINVAL;
+                       if (cmd->duplex == DUPLEX_FULL &&
+                           !(f & SUPPORTED_1000baseT_Full))
+                               return -EINVAL;
+                       break;
+               default:
+                       return -EINVAL;
+               }
+
+               mutex_lock(&dev->link_lock);
+               dev->phy.def->ops->setup_forced(&dev->phy, cmd->speed,
+                                               cmd->duplex);
+               mutex_unlock(&dev->link_lock);
+
+       } else {
+               if (!(f & SUPPORTED_Autoneg))
+                       return -EINVAL;
+
+               mutex_lock(&dev->link_lock);
+               dev->phy.def->ops->setup_aneg(&dev->phy,
+                                             (cmd->advertising & f) |
+                                             (dev->phy.advertising &
+                                              (ADVERTISED_Pause |
+                                               ADVERTISED_Asym_Pause)));
+               mutex_unlock(&dev->link_lock);
+       }
+       emac_force_link_update(dev);
+
+       return 0;
+}
+
+static void emac_ethtool_get_ringparam(struct net_device *ndev,
+                                      struct ethtool_ringparam *rp)
+{
+       rp->rx_max_pending = rp->rx_pending = NUM_RX_BUFF;
+       rp->tx_max_pending = rp->tx_pending = NUM_TX_BUFF;
+}
+
+static void emac_ethtool_get_pauseparam(struct net_device *ndev,
+                                       struct ethtool_pauseparam *pp)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       mutex_lock(&dev->link_lock);
+       if ((dev->phy.features & SUPPORTED_Autoneg) &&
+           (dev->phy.advertising & (ADVERTISED_Pause | ADVERTISED_Asym_Pause)))
+               pp->autoneg = 1;
+
+       if (dev->phy.duplex == DUPLEX_FULL) {
+               if (dev->phy.pause)
+                       pp->rx_pause = pp->tx_pause = 1;
+               else if (dev->phy.asym_pause)
+                       pp->tx_pause = 1;
+       }
+       mutex_unlock(&dev->link_lock);
+}
+
+static int emac_get_regs_len(struct emac_instance *dev)
+{
+               return sizeof(struct emac_ethtool_regs_subhdr) +
+                       sizeof(struct emac_regs);
+}
+
+static int emac_ethtool_get_regs_len(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int size;
+
+       size = sizeof(struct emac_ethtool_regs_hdr) +
+               emac_get_regs_len(dev) + mal_get_regs_len(dev->mal);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               size += zmii_get_regs_len(dev->zmii_dev);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               size += rgmii_get_regs_len(dev->rgmii_dev);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+               size += tah_get_regs_len(dev->tah_dev);
+
+       return size;
+}
+
+static void *emac_dump_regs(struct emac_instance *dev, void *buf)
+{
+       struct emac_ethtool_regs_subhdr *hdr = buf;
+
+       hdr->index = dev->cell_index;
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4SYNC)) {
+               hdr->version = EMAC4SYNC_ETHTOOL_REGS_VER;
+       } else if (emac_has_feature(dev, EMAC_FTR_EMAC4)) {
+               hdr->version = EMAC4_ETHTOOL_REGS_VER;
+       } else {
+               hdr->version = EMAC_ETHTOOL_REGS_VER;
+       }
+       memcpy_fromio(hdr + 1, dev->emacp, sizeof(struct emac_regs));
+       return (void *)(hdr + 1) + sizeof(struct emac_regs);
+}
+
+static void emac_ethtool_get_regs(struct net_device *ndev,
+                                 struct ethtool_regs *regs, void *buf)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       struct emac_ethtool_regs_hdr *hdr = buf;
+
+       hdr->components = 0;
+       buf = hdr + 1;
+
+       buf = mal_dump_regs(dev->mal, buf);
+       buf = emac_dump_regs(dev, buf);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII)) {
+               hdr->components |= EMAC_ETHTOOL_REGS_ZMII;
+               buf = zmii_dump_regs(dev->zmii_dev, buf);
+       }
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII)) {
+               hdr->components |= EMAC_ETHTOOL_REGS_RGMII;
+               buf = rgmii_dump_regs(dev->rgmii_dev, buf);
+       }
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH)) {
+               hdr->components |= EMAC_ETHTOOL_REGS_TAH;
+               buf = tah_dump_regs(dev->tah_dev, buf);
+       }
+}
+
+static int emac_ethtool_nway_reset(struct net_device *ndev)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       int res = 0;
+
+       DBG(dev, "nway_reset" NL);
+
+       if (dev->phy.address < 0)
+               return -EOPNOTSUPP;
+
+       mutex_lock(&dev->link_lock);
+       if (!dev->phy.autoneg) {
+               res = -EINVAL;
+               goto out;
+       }
+
+       dev->phy.def->ops->setup_aneg(&dev->phy, dev->phy.advertising);
+ out:
+       mutex_unlock(&dev->link_lock);
+       emac_force_link_update(dev);
+       return res;
+}
+
+static int emac_ethtool_get_sset_count(struct net_device *ndev, int stringset)
+{
+       if (stringset == ETH_SS_STATS)
+               return EMAC_ETHTOOL_STATS_COUNT;
+       else
+               return -EINVAL;
+}
+
+static void emac_ethtool_get_strings(struct net_device *ndev, u32 stringset,
+                                    u8 * buf)
+{
+       if (stringset == ETH_SS_STATS)
+               memcpy(buf, &emac_stats_keys, sizeof(emac_stats_keys));
+}
+
+static void emac_ethtool_get_ethtool_stats(struct net_device *ndev,
+                                          struct ethtool_stats *estats,
+                                          u64 * tmp_stats)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       memcpy(tmp_stats, &dev->stats, sizeof(dev->stats));
+       tmp_stats += sizeof(dev->stats) / sizeof(u64);
+       memcpy(tmp_stats, &dev->estats, sizeof(dev->estats));
+}
+
+static void emac_ethtool_get_drvinfo(struct net_device *ndev,
+                                    struct ethtool_drvinfo *info)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+
+       strlcpy(info->driver, "ibm_emac", sizeof(info->driver));
+       strlcpy(info->version, DRV_VERSION, sizeof(info->version));
+       snprintf(info->bus_info, sizeof(info->bus_info), "PPC 4xx EMAC-%d %s",
+                dev->cell_index, dev->ofdev->dev.of_node->full_name);
+       info->regdump_len = emac_ethtool_get_regs_len(ndev);
+}
+
+static const struct ethtool_ops emac_ethtool_ops = {
+       .get_settings = emac_ethtool_get_settings,
+       .set_settings = emac_ethtool_set_settings,
+       .get_drvinfo = emac_ethtool_get_drvinfo,
+
+       .get_regs_len = emac_ethtool_get_regs_len,
+       .get_regs = emac_ethtool_get_regs,
+
+       .nway_reset = emac_ethtool_nway_reset,
+
+       .get_ringparam = emac_ethtool_get_ringparam,
+       .get_pauseparam = emac_ethtool_get_pauseparam,
+
+       .get_strings = emac_ethtool_get_strings,
+       .get_sset_count = emac_ethtool_get_sset_count,
+       .get_ethtool_stats = emac_ethtool_get_ethtool_stats,
+
+       .get_link = ethtool_op_get_link,
+};
+
+static int emac_ioctl(struct net_device *ndev, struct ifreq *rq, int cmd)
+{
+       struct emac_instance *dev = netdev_priv(ndev);
+       struct mii_ioctl_data *data = if_mii(rq);
+
+       DBG(dev, "ioctl %08x" NL, cmd);
+
+       if (dev->phy.address < 0)
+               return -EOPNOTSUPP;
+
+       switch (cmd) {
+       case SIOCGMIIPHY:
+               data->phy_id = dev->phy.address;
+               /* Fall through */
+       case SIOCGMIIREG:
+               data->val_out = emac_mdio_read(ndev, dev->phy.address,
+                                              data->reg_num);
+               return 0;
+
+       case SIOCSMIIREG:
+               emac_mdio_write(ndev, dev->phy.address, data->reg_num,
+                               data->val_in);
+               return 0;
+       default:
+               return -EOPNOTSUPP;
+       }
+}
+
+struct emac_depentry {
+       u32                     phandle;
+       struct device_node      *node;
+       struct platform_device  *ofdev;
+       void                    *drvdata;
+};
+
+#define        EMAC_DEP_MAL_IDX        0
+#define        EMAC_DEP_ZMII_IDX       1
+#define        EMAC_DEP_RGMII_IDX      2
+#define        EMAC_DEP_TAH_IDX        3
+#define        EMAC_DEP_MDIO_IDX       4
+#define        EMAC_DEP_PREV_IDX       5
+#define        EMAC_DEP_COUNT          6
+
+static int emac_check_deps(struct emac_instance *dev,
+                          struct emac_depentry *deps)
+{
+       int i, there = 0;
+       struct device_node *np;
+
+       for (i = 0; i < EMAC_DEP_COUNT; i++) {
+               /* no dependency on that item, allright */
+               if (deps[i].phandle == 0) {
+                       there++;
+                       continue;
+               }
+               /* special case for blist as the dependency might go away */
+               if (i == EMAC_DEP_PREV_IDX) {
+                       np = *(dev->blist - 1);
+                       if (np == NULL) {
+                               deps[i].phandle = 0;
+                               there++;
+                               continue;
+                       }
+                       if (deps[i].node == NULL)
+                               deps[i].node = of_node_get(np);
+               }
+               if (deps[i].node == NULL)
+                       deps[i].node = of_find_node_by_phandle(deps[i].phandle);
+               if (deps[i].node == NULL)
+                       continue;
+               if (deps[i].ofdev == NULL)
+                       deps[i].ofdev = of_find_device_by_node(deps[i].node);
+               if (deps[i].ofdev == NULL)
+                       continue;
+               if (deps[i].drvdata == NULL)
+                       deps[i].drvdata = platform_get_drvdata(deps[i].ofdev);
+               if (deps[i].drvdata != NULL)
+                       there++;
+       }
+       return there == EMAC_DEP_COUNT;
+}
+
+static void emac_put_deps(struct emac_instance *dev)
+{
+       of_dev_put(dev->mal_dev);
+       of_dev_put(dev->zmii_dev);
+       of_dev_put(dev->rgmii_dev);
+       of_dev_put(dev->mdio_dev);
+       of_dev_put(dev->tah_dev);
+}
+
+static int emac_of_bus_notify(struct notifier_block *nb, unsigned long action,
+                             void *data)
+{
+       /* We are only intereted in device addition */
+       if (action == BUS_NOTIFY_BOUND_DRIVER)
+               wake_up_all(&emac_probe_wait);
+       return 0;
+}
+
+static struct notifier_block emac_of_bus_notifier = {
+       .notifier_call = emac_of_bus_notify
+};
+
+static int emac_wait_deps(struct emac_instance *dev)
+{
+       struct emac_depentry deps[EMAC_DEP_COUNT];
+       int i, err;
+
+       memset(&deps, 0, sizeof(deps));
+
+       deps[EMAC_DEP_MAL_IDX].phandle = dev->mal_ph;
+       deps[EMAC_DEP_ZMII_IDX].phandle = dev->zmii_ph;
+       deps[EMAC_DEP_RGMII_IDX].phandle = dev->rgmii_ph;
+       if (dev->tah_ph)
+               deps[EMAC_DEP_TAH_IDX].phandle = dev->tah_ph;
+       if (dev->mdio_ph)
+               deps[EMAC_DEP_MDIO_IDX].phandle = dev->mdio_ph;
+       if (dev->blist && dev->blist > emac_boot_list)
+               deps[EMAC_DEP_PREV_IDX].phandle = 0xffffffffu;
+       bus_register_notifier(&platform_bus_type, &emac_of_bus_notifier);
+       wait_event_timeout(emac_probe_wait,
+                          emac_check_deps(dev, deps),
+                          EMAC_PROBE_DEP_TIMEOUT);
+       bus_unregister_notifier(&platform_bus_type, &emac_of_bus_notifier);
+       err = emac_check_deps(dev, deps) ? 0 : -ENODEV;
+       for (i = 0; i < EMAC_DEP_COUNT; i++) {
+               of_node_put(deps[i].node);
+               if (err)
+                       of_dev_put(deps[i].ofdev);
+       }
+       if (err == 0) {
+               dev->mal_dev = deps[EMAC_DEP_MAL_IDX].ofdev;
+               dev->zmii_dev = deps[EMAC_DEP_ZMII_IDX].ofdev;
+               dev->rgmii_dev = deps[EMAC_DEP_RGMII_IDX].ofdev;
+               dev->tah_dev = deps[EMAC_DEP_TAH_IDX].ofdev;
+               dev->mdio_dev = deps[EMAC_DEP_MDIO_IDX].ofdev;
+       }
+       of_dev_put(deps[EMAC_DEP_PREV_IDX].ofdev);
+       return err;
+}
+
+static int emac_read_uint_prop(struct device_node *np, const char *name,
+                              u32 *val, int fatal)
+{
+       int len;
+       const u32 *prop = of_get_property(np, name, &len);
+       if (prop == NULL || len < sizeof(u32)) {
+               if (fatal)
+                       printk(KERN_ERR "%s: missing %s property\n",
+                              np->full_name, name);
+               return -ENODEV;
+       }
+       *val = *prop;
+       return 0;
+}
+
+static int emac_init_phy(struct emac_instance *dev)
+{
+       struct device_node *np = dev->ofdev->dev.of_node;
+       struct net_device *ndev = dev->ndev;
+       u32 phy_map, adv;
+       int i;
+
+       dev->phy.dev = ndev;
+       dev->phy.mode = dev->phy_mode;
+
+       /* PHY-less configuration.
+        * XXX I probably should move these settings to the dev tree
+        */
+       if (dev->phy_address == 0xffffffff && dev->phy_map == 0xffffffff) {
+               emac_reset(dev);
+
+               /* PHY-less configuration.
+                * XXX I probably should move these settings to the dev tree
+                */
+               dev->phy.address = -1;
+               dev->phy.features = SUPPORTED_MII;
+               if (emac_phy_supports_gige(dev->phy_mode))
+                       dev->phy.features |= SUPPORTED_1000baseT_Full;
+               else
+                       dev->phy.features |= SUPPORTED_100baseT_Full;
+               dev->phy.pause = 1;
+
+               return 0;
+       }
+
+       mutex_lock(&emac_phy_map_lock);
+       phy_map = dev->phy_map | busy_phy_map;
+
+       DBG(dev, "PHY maps %08x %08x" NL, dev->phy_map, busy_phy_map);
+
+       dev->phy.mdio_read = emac_mdio_read;
+       dev->phy.mdio_write = emac_mdio_write;
+
+       /* Enable internal clock source */
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_440GX_PHY_CLK_FIX))
+               dcri_clrset(SDR0, SDR0_MFR, 0, SDR0_MFR_ECS);
+#endif
+       /* PHY clock workaround */
+       emac_rx_clk_tx(dev);
+
+       /* Enable internal clock source on 440GX*/
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_440GX_PHY_CLK_FIX))
+               dcri_clrset(SDR0, SDR0_MFR, 0, SDR0_MFR_ECS);
+#endif
+       /* Configure EMAC with defaults so we can at least use MDIO
+        * This is needed mostly for 440GX
+        */
+       if (emac_phy_gpcs(dev->phy.mode)) {
+               /* XXX
+                * Make GPCS PHY address equal to EMAC index.
+                * We probably should take into account busy_phy_map
+                * and/or phy_map here.
+                *
+                * Note that the busy_phy_map is currently global
+                * while it should probably be per-ASIC...
+                */
+               dev->phy.gpcs_address = dev->gpcs_address;
+               if (dev->phy.gpcs_address == 0xffffffff)
+                       dev->phy.address = dev->cell_index;
+       }
+
+       emac_configure(dev);
+
+       if (dev->phy_address != 0xffffffff)
+               phy_map = ~(1 << dev->phy_address);
+
+       for (i = 0; i < 0x20; phy_map >>= 1, ++i)
+               if (!(phy_map & 1)) {
+                       int r;
+                       busy_phy_map |= 1 << i;
+
+                       /* Quick check if there is a PHY at the address */
+                       r = emac_mdio_read(dev->ndev, i, MII_BMCR);
+                       if (r == 0xffff || r < 0)
+                               continue;
+                       if (!emac_mii_phy_probe(&dev->phy, i))
+                               break;
+               }
+
+       /* Enable external clock source */
+#ifdef CONFIG_PPC_DCR_NATIVE
+       if (emac_has_feature(dev, EMAC_FTR_440GX_PHY_CLK_FIX))
+               dcri_clrset(SDR0, SDR0_MFR, SDR0_MFR_ECS, 0);
+#endif
+       mutex_unlock(&emac_phy_map_lock);
+       if (i == 0x20) {
+               printk(KERN_WARNING "%s: can't find PHY!\n", np->full_name);
+               return -ENXIO;
+       }
+
+       /* Init PHY */
+       if (dev->phy.def->ops->init)
+               dev->phy.def->ops->init(&dev->phy);
+
+       /* Disable any PHY features not supported by the platform */
+       dev->phy.def->features &= ~dev->phy_feat_exc;
+       dev->phy.features &= ~dev->phy_feat_exc;
+
+       /* Setup initial link parameters */
+       if (dev->phy.features & SUPPORTED_Autoneg) {
+               adv = dev->phy.features;
+               if (!emac_has_feature(dev, EMAC_FTR_NO_FLOW_CONTROL_40x))
+                       adv |= ADVERTISED_Pause | ADVERTISED_Asym_Pause;
+               /* Restart autonegotiation */
+               dev->phy.def->ops->setup_aneg(&dev->phy, adv);
+       } else {
+               u32 f = dev->phy.def->features;
+               int speed = SPEED_10, fd = DUPLEX_HALF;
+
+               /* Select highest supported speed/duplex */
+               if (f & SUPPORTED_1000baseT_Full) {
+                       speed = SPEED_1000;
+                       fd = DUPLEX_FULL;
+               } else if (f & SUPPORTED_1000baseT_Half)
+                       speed = SPEED_1000;
+               else if (f & SUPPORTED_100baseT_Full) {
+                       speed = SPEED_100;
+                       fd = DUPLEX_FULL;
+               } else if (f & SUPPORTED_100baseT_Half)
+                       speed = SPEED_100;
+               else if (f & SUPPORTED_10baseT_Full)
+                       fd = DUPLEX_FULL;
+
+               /* Force link parameters */
+               dev->phy.def->ops->setup_forced(&dev->phy, speed, fd);
+       }
+       return 0;
+}
+
+static int emac_init_config(struct emac_instance *dev)
+{
+       struct device_node *np = dev->ofdev->dev.of_node;
+       const void *p;
+
+       /* Read config from device-tree */
+       if (emac_read_uint_prop(np, "mal-device", &dev->mal_ph, 1))
+               return -ENXIO;
+       if (emac_read_uint_prop(np, "mal-tx-channel", &dev->mal_tx_chan, 1))
+               return -ENXIO;
+       if (emac_read_uint_prop(np, "mal-rx-channel", &dev->mal_rx_chan, 1))
+               return -ENXIO;
+       if (emac_read_uint_prop(np, "cell-index", &dev->cell_index, 1))
+               return -ENXIO;
+       if (emac_read_uint_prop(np, "max-frame-size", &dev->max_mtu, 0))
+               dev->max_mtu = 1500;
+       if (emac_read_uint_prop(np, "rx-fifo-size", &dev->rx_fifo_size, 0))
+               dev->rx_fifo_size = 2048;
+       if (emac_read_uint_prop(np, "tx-fifo-size", &dev->tx_fifo_size, 0))
+               dev->tx_fifo_size = 2048;
+       if (emac_read_uint_prop(np, "rx-fifo-size-gige", &dev->rx_fifo_size_gige, 0))
+               dev->rx_fifo_size_gige = dev->rx_fifo_size;
+       if (emac_read_uint_prop(np, "tx-fifo-size-gige", &dev->tx_fifo_size_gige, 0))
+               dev->tx_fifo_size_gige = dev->tx_fifo_size;
+       if (emac_read_uint_prop(np, "phy-address", &dev->phy_address, 0))
+               dev->phy_address = 0xffffffff;
+       if (emac_read_uint_prop(np, "phy-map", &dev->phy_map, 0))
+               dev->phy_map = 0xffffffff;
+       if (emac_read_uint_prop(np, "gpcs-address", &dev->gpcs_address, 0))
+               dev->gpcs_address = 0xffffffff;
+       if (emac_read_uint_prop(np->parent, "clock-frequency", &dev->opb_bus_freq, 1))
+               return -ENXIO;
+       if (emac_read_uint_prop(np, "tah-device", &dev->tah_ph, 0))
+               dev->tah_ph = 0;
+       if (emac_read_uint_prop(np, "tah-channel", &dev->tah_port, 0))
+               dev->tah_port = 0;
+       if (emac_read_uint_prop(np, "mdio-device", &dev->mdio_ph, 0))
+               dev->mdio_ph = 0;
+       if (emac_read_uint_prop(np, "zmii-device", &dev->zmii_ph, 0))
+               dev->zmii_ph = 0;
+       if (emac_read_uint_prop(np, "zmii-channel", &dev->zmii_port, 0))
+               dev->zmii_port = 0xffffffff;
+       if (emac_read_uint_prop(np, "rgmii-device", &dev->rgmii_ph, 0))
+               dev->rgmii_ph = 0;
+       if (emac_read_uint_prop(np, "rgmii-channel", &dev->rgmii_port, 0))
+               dev->rgmii_port = 0xffffffff;
+       if (emac_read_uint_prop(np, "fifo-entry-size", &dev->fifo_entry_size, 0))
+               dev->fifo_entry_size = 16;
+       if (emac_read_uint_prop(np, "mal-burst-size", &dev->mal_burst_size, 0))
+               dev->mal_burst_size = 256;
+
+       /* PHY mode needs some decoding */
+       dev->phy_mode = of_get_phy_mode(np);
+       if (dev->phy_mode < 0)
+               dev->phy_mode = PHY_MODE_NA;
+
+       /* Check EMAC version */
+       if (of_device_is_compatible(np, "ibm,emac4sync")) {
+               dev->features |= (EMAC_FTR_EMAC4 | EMAC_FTR_EMAC4SYNC);
+               if (of_device_is_compatible(np, "ibm,emac-460ex") ||
+                   of_device_is_compatible(np, "ibm,emac-460gt"))
+                       dev->features |= EMAC_FTR_460EX_PHY_CLK_FIX;
+               if (of_device_is_compatible(np, "ibm,emac-405ex") ||
+                   of_device_is_compatible(np, "ibm,emac-405exr"))
+                       dev->features |= EMAC_FTR_440EP_PHY_CLK_FIX;
+               if (of_device_is_compatible(np, "ibm,emac-apm821xx")) {
+                       dev->features |= (EMAC_APM821XX_REQ_JUMBO_FRAME_SIZE |
+                                         EMAC_FTR_APM821XX_NO_HALF_DUPLEX |
+                                         EMAC_FTR_460EX_PHY_CLK_FIX);
+               }
+       } else if (of_device_is_compatible(np, "ibm,emac4")) {
+               dev->features |= EMAC_FTR_EMAC4;
+               if (of_device_is_compatible(np, "ibm,emac-440gx"))
+                       dev->features |= EMAC_FTR_440GX_PHY_CLK_FIX;
+       } else {
+               if (of_device_is_compatible(np, "ibm,emac-440ep") ||
+                   of_device_is_compatible(np, "ibm,emac-440gr"))
+                       dev->features |= EMAC_FTR_440EP_PHY_CLK_FIX;
+               if (of_device_is_compatible(np, "ibm,emac-405ez")) {
+#ifdef CONFIG_IBM_EMAC_NO_FLOW_CTRL
+                       dev->features |= EMAC_FTR_NO_FLOW_CONTROL_40x;
+#else
+                       printk(KERN_ERR "%s: Flow control not disabled!\n",
+                                       np->full_name);
+                       return -ENXIO;
+#endif
+               }
+
+       }
+
+       /* Fixup some feature bits based on the device tree */
+       if (of_get_property(np, "has-inverted-stacr-oc", NULL))
+               dev->features |= EMAC_FTR_STACR_OC_INVERT;
+       if (of_get_property(np, "has-new-stacr-staopc", NULL))
+               dev->features |= EMAC_FTR_HAS_NEW_STACR;
+
+       /* CAB lacks the appropriate properties */
+       if (of_device_is_compatible(np, "ibm,emac-axon"))
+               dev->features |= EMAC_FTR_HAS_NEW_STACR |
+                       EMAC_FTR_STACR_OC_INVERT;
+
+       /* Enable TAH/ZMII/RGMII features as found */
+       if (dev->tah_ph != 0) {
+#ifdef CONFIG_IBM_EMAC_TAH
+               dev->features |= EMAC_FTR_HAS_TAH;
+#else
+               printk(KERN_ERR "%s: TAH support not enabled !\n",
+                      np->full_name);
+               return -ENXIO;
+#endif
+       }
+
+       if (dev->zmii_ph != 0) {
+#ifdef CONFIG_IBM_EMAC_ZMII
+               dev->features |= EMAC_FTR_HAS_ZMII;
+#else
+               printk(KERN_ERR "%s: ZMII support not enabled !\n",
+                      np->full_name);
+               return -ENXIO;
+#endif
+       }
+
+       if (dev->rgmii_ph != 0) {
+#ifdef CONFIG_IBM_EMAC_RGMII
+               dev->features |= EMAC_FTR_HAS_RGMII;
+#else
+               printk(KERN_ERR "%s: RGMII support not enabled !\n",
+                      np->full_name);
+               return -ENXIO;
+#endif
+       }
+
+       /* Read MAC-address */
+       p = of_get_property(np, "local-mac-address", NULL);
+       if (p == NULL) {
+               printk(KERN_ERR "%s: Can't find local-mac-address property\n",
+                      np->full_name);
+               return -ENXIO;
+       }
+       memcpy(dev->ndev->dev_addr, p, ETH_ALEN);
+
+       /* IAHT and GAHT filter parameterization */
+       if (emac_has_feature(dev, EMAC_FTR_EMAC4SYNC)) {
+               dev->xaht_slots_shift = EMAC4SYNC_XAHT_SLOTS_SHIFT;
+               dev->xaht_width_shift = EMAC4SYNC_XAHT_WIDTH_SHIFT;
+       } else {
+               dev->xaht_slots_shift = EMAC4_XAHT_SLOTS_SHIFT;
+               dev->xaht_width_shift = EMAC4_XAHT_WIDTH_SHIFT;
+       }
+
+       DBG(dev, "features     : 0x%08x / 0x%08x\n", dev->features, EMAC_FTRS_POSSIBLE);
+       DBG(dev, "tx_fifo_size : %d (%d gige)\n", dev->tx_fifo_size, dev->tx_fifo_size_gige);
+       DBG(dev, "rx_fifo_size : %d (%d gige)\n", dev->rx_fifo_size, dev->rx_fifo_size_gige);
+       DBG(dev, "max_mtu      : %d\n", dev->max_mtu);
+       DBG(dev, "OPB freq     : %d\n", dev->opb_bus_freq);
+
+       return 0;
+}
+
+static const struct net_device_ops emac_netdev_ops = {
+       .ndo_open               = emac_open,
+       .ndo_stop               = emac_close,
+       .ndo_get_stats          = emac_stats,
+       .ndo_set_rx_mode        = emac_set_multicast_list,
+       .ndo_do_ioctl           = emac_ioctl,
+       .ndo_tx_timeout         = emac_tx_timeout,
+       .ndo_validate_addr      = eth_validate_addr,
+       .ndo_set_mac_address    = eth_mac_addr,
+       .ndo_start_xmit         = emac_start_xmit,
+       .ndo_change_mtu         = eth_change_mtu,
+};
+
+static const struct net_device_ops emac_gige_netdev_ops = {
+       .ndo_open               = emac_open,
+       .ndo_stop               = emac_close,
+       .ndo_get_stats          = emac_stats,
+       .ndo_set_rx_mode        = emac_set_multicast_list,
+       .ndo_do_ioctl           = emac_ioctl,
+       .ndo_tx_timeout         = emac_tx_timeout,
+       .ndo_validate_addr      = eth_validate_addr,
+       .ndo_set_mac_address    = eth_mac_addr,
+       .ndo_start_xmit         = emac_start_xmit_sg,
+       .ndo_change_mtu         = emac_change_mtu,
+};
+
+static int emac_probe(struct platform_device *ofdev)
+{
+       struct net_device *ndev;
+       struct emac_instance *dev;
+       struct device_node *np = ofdev->dev.of_node;
+       struct device_node **blist = NULL;
+       int err, i;
+
+       /* Skip unused/unwired EMACS.  We leave the check for an unused
+        * property here for now, but new flat device trees should set a
+        * status property to "disabled" instead.
+        */
+       if (of_get_property(np, "unused", NULL) || !of_device_is_available(np))
+               return -ENODEV;
+
+       /* Find ourselves in the bootlist if we are there */
+       for (i = 0; i < EMAC_BOOT_LIST_SIZE; i++)
+               if (emac_boot_list[i] == np)
+                       blist = &emac_boot_list[i];
+
+       /* Allocate our net_device structure */
+       err = -ENOMEM;
+       ndev = alloc_etherdev(sizeof(struct emac_instance));
+       if (!ndev)
+               goto err_gone;
+
+       dev = netdev_priv(ndev);
+       dev->ndev = ndev;
+       dev->ofdev = ofdev;
+       dev->blist = blist;
+       SET_NETDEV_DEV(ndev, &ofdev->dev);
+
+       /* Initialize some embedded data structures */
+       mutex_init(&dev->mdio_lock);
+       mutex_init(&dev->link_lock);
+       spin_lock_init(&dev->lock);
+       INIT_WORK(&dev->reset_work, emac_reset_work);
+
+       /* Init various config data based on device-tree */
+       err = emac_init_config(dev);
+       if (err != 0)
+               goto err_free;
+
+       /* Get interrupts. EMAC irq is mandatory, WOL irq is optional */
+       dev->emac_irq = irq_of_parse_and_map(np, 0);
+       dev->wol_irq = irq_of_parse_and_map(np, 1);
+       if (dev->emac_irq == NO_IRQ) {
+               printk(KERN_ERR "%s: Can't map main interrupt\n", np->full_name);
+               goto err_free;
+       }
+       ndev->irq = dev->emac_irq;
+
+       /* Map EMAC regs */
+       if (of_address_to_resource(np, 0, &dev->rsrc_regs)) {
+               printk(KERN_ERR "%s: Can't get registers address\n",
+                      np->full_name);
+               goto err_irq_unmap;
+       }
+       // TODO : request_mem_region
+       dev->emacp = ioremap(dev->rsrc_regs.start,
+                            resource_size(&dev->rsrc_regs));
+       if (dev->emacp == NULL) {
+               printk(KERN_ERR "%s: Can't map device registers!\n",
+                      np->full_name);
+               err = -ENOMEM;
+               goto err_irq_unmap;
+       }
+
+       /* Wait for dependent devices */
+       err = emac_wait_deps(dev);
+       if (err) {
+               printk(KERN_ERR
+                      "%s: Timeout waiting for dependent devices\n",
+                      np->full_name);
+               /*  display more info about what's missing ? */
+               goto err_reg_unmap;
+       }
+       dev->mal = platform_get_drvdata(dev->mal_dev);
+       if (dev->mdio_dev != NULL)
+               dev->mdio_instance = platform_get_drvdata(dev->mdio_dev);
+
+       /* Register with MAL */
+       dev->commac.ops = &emac_commac_ops;
+       dev->commac.dev = dev;
+       dev->commac.tx_chan_mask = MAL_CHAN_MASK(dev->mal_tx_chan);
+       dev->commac.rx_chan_mask = MAL_CHAN_MASK(dev->mal_rx_chan);
+       err = mal_register_commac(dev->mal, &dev->commac);
+       if (err) {
+               printk(KERN_ERR "%s: failed to register with mal %s!\n",
+                      np->full_name, dev->mal_dev->dev.of_node->full_name);
+               goto err_rel_deps;
+       }
+       dev->rx_skb_size = emac_rx_skb_size(ndev->mtu);
+       dev->rx_sync_size = emac_rx_sync_size(ndev->mtu);
+
+       /* Get pointers to BD rings */
+       dev->tx_desc =
+           dev->mal->bd_virt + mal_tx_bd_offset(dev->mal, dev->mal_tx_chan);
+       dev->rx_desc =
+           dev->mal->bd_virt + mal_rx_bd_offset(dev->mal, dev->mal_rx_chan);
+
+       DBG(dev, "tx_desc %p" NL, dev->tx_desc);
+       DBG(dev, "rx_desc %p" NL, dev->rx_desc);
+
+       /* Clean rings */
+       memset(dev->tx_desc, 0, NUM_TX_BUFF * sizeof(struct mal_descriptor));
+       memset(dev->rx_desc, 0, NUM_RX_BUFF * sizeof(struct mal_descriptor));
+       memset(dev->tx_skb, 0, NUM_TX_BUFF * sizeof(struct sk_buff *));
+       memset(dev->rx_skb, 0, NUM_RX_BUFF * sizeof(struct sk_buff *));
+
+       /* Attach to ZMII, if needed */
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII) &&
+           (err = zmii_attach(dev->zmii_dev, dev->zmii_port, &dev->phy_mode)) != 0)
+               goto err_unreg_commac;
+
+       /* Attach to RGMII, if needed */
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII) &&
+           (err = rgmii_attach(dev->rgmii_dev, dev->rgmii_port, dev->phy_mode)) != 0)
+               goto err_detach_zmii;
+
+       /* Attach to TAH, if needed */
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH) &&
+           (err = tah_attach(dev->tah_dev, dev->tah_port)) != 0)
+               goto err_detach_rgmii;
+
+       /* Set some link defaults before we can find out real parameters */
+       dev->phy.speed = SPEED_100;
+       dev->phy.duplex = DUPLEX_FULL;
+       dev->phy.autoneg = AUTONEG_DISABLE;
+       dev->phy.pause = dev->phy.asym_pause = 0;
+       dev->stop_timeout = STOP_TIMEOUT_100;
+       INIT_DELAYED_WORK(&dev->link_work, emac_link_timer);
+
+       /* Some SoCs like APM821xx does not support Half Duplex mode. */
+       if (emac_has_feature(dev, EMAC_FTR_APM821XX_NO_HALF_DUPLEX)) {
+               dev->phy_feat_exc = (SUPPORTED_1000baseT_Half |
+                                    SUPPORTED_100baseT_Half |
+                                    SUPPORTED_10baseT_Half);
+       }
+
+       /* Find PHY if any */
+       err = emac_init_phy(dev);
+       if (err != 0)
+               goto err_detach_tah;
+
+       if (dev->tah_dev) {
+               ndev->hw_features = NETIF_F_IP_CSUM | NETIF_F_SG;
+               ndev->features |= ndev->hw_features | NETIF_F_RXCSUM;
+       }
+       ndev->watchdog_timeo = 5 * HZ;
+       if (emac_phy_supports_gige(dev->phy_mode)) {
+               ndev->netdev_ops = &emac_gige_netdev_ops;
+               dev->commac.ops = &emac_commac_sg_ops;
+       } else
+               ndev->netdev_ops = &emac_netdev_ops;
+       ndev->ethtool_ops = &emac_ethtool_ops;
+
+       netif_carrier_off(ndev);
+
+       err = register_netdev(ndev);
+       if (err) {
+               printk(KERN_ERR "%s: failed to register net device (%d)!\n",
+                      np->full_name, err);
+               goto err_detach_tah;
+       }
+
+       /* Set our drvdata last as we don't want them visible until we are
+        * fully initialized
+        */
+       wmb();
+       platform_set_drvdata(ofdev, dev);
+
+       /* There's a new kid in town ! Let's tell everybody */
+       wake_up_all(&emac_probe_wait);
+
+
+       printk(KERN_INFO "%s: EMAC-%d %s, MAC %pM\n",
+              ndev->name, dev->cell_index, np->full_name, ndev->dev_addr);
+
+       if (dev->phy_mode == PHY_MODE_SGMII)
+               printk(KERN_NOTICE "%s: in SGMII mode\n", ndev->name);
+
+       if (dev->phy.address >= 0)
+               printk("%s: found %s PHY (0x%02x)\n", ndev->name,
+                      dev->phy.def->name, dev->phy.address);
+
+       emac_dbg_register(dev);
+
+       /* Life is good */
+       return 0;
+
+       /* I have a bad feeling about this ... */
+
+ err_detach_tah:
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+               tah_detach(dev->tah_dev, dev->tah_port);
+ err_detach_rgmii:
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_detach(dev->rgmii_dev, dev->rgmii_port);
+ err_detach_zmii:
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_detach(dev->zmii_dev, dev->zmii_port);
+ err_unreg_commac:
+       mal_unregister_commac(dev->mal, &dev->commac);
+ err_rel_deps:
+       emac_put_deps(dev);
+ err_reg_unmap:
+       iounmap(dev->emacp);
+ err_irq_unmap:
+       if (dev->wol_irq != NO_IRQ)
+               irq_dispose_mapping(dev->wol_irq);
+       if (dev->emac_irq != NO_IRQ)
+               irq_dispose_mapping(dev->emac_irq);
+ err_free:
+       free_netdev(ndev);
+ err_gone:
+       /* if we were on the bootlist, remove us as we won't show up and
+        * wake up all waiters to notify them in case they were waiting
+        * on us
+        */
+       if (blist) {
+               *blist = NULL;
+               wake_up_all(&emac_probe_wait);
+       }
+       return err;
+}
+
+static int emac_remove(struct platform_device *ofdev)
+{
+       struct emac_instance *dev = platform_get_drvdata(ofdev);
+
+       DBG(dev, "remove" NL);
+
+       unregister_netdev(dev->ndev);
+
+       cancel_work_sync(&dev->reset_work);
+
+       if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+               tah_detach(dev->tah_dev, dev->tah_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_RGMII))
+               rgmii_detach(dev->rgmii_dev, dev->rgmii_port);
+       if (emac_has_feature(dev, EMAC_FTR_HAS_ZMII))
+               zmii_detach(dev->zmii_dev, dev->zmii_port);
+
+       busy_phy_map &= ~(1 << dev->phy.address);
+       DBG(dev, "busy_phy_map now %#x" NL, busy_phy_map);
+
+       mal_unregister_commac(dev->mal, &dev->commac);
+       emac_put_deps(dev);
+
+       emac_dbg_unregister(dev);
+       iounmap(dev->emacp);
+
+       if (dev->wol_irq != NO_IRQ)
+               irq_dispose_mapping(dev->wol_irq);
+       if (dev->emac_irq != NO_IRQ)
+               irq_dispose_mapping(dev->emac_irq);
+
+       free_netdev(dev->ndev);
+
+       return 0;
+}
+
+/* XXX Features in here should be replaced by properties... */
+static const struct of_device_id emac_match[] =
+{
+       {
+               .type           = "network",
+               .compatible     = "ibm,emac",
+       },
+       {
+               .type           = "network",
+               .compatible     = "ibm,emac4",
+       },
+       {
+               .type           = "network",
+               .compatible     = "ibm,emac4sync",
+       },
+       {},
+};
+MODULE_DEVICE_TABLE(of, emac_match);
+
+static struct platform_driver emac_driver = {
+       .driver = {
+               .name = "emac",
+               .of_match_table = emac_match,
+       },
+       .probe = emac_probe,
+       .remove = emac_remove,
+};
+
+static void __init emac_make_bootlist(void)
+{
+       struct device_node *np = NULL;
+       int j, max, i = 0, k;
+       int cell_indices[EMAC_BOOT_LIST_SIZE];
+
+       /* Collect EMACs */
+       while((np = of_find_all_nodes(np)) != NULL) {
+               const u32 *idx;
+
+               if (of_match_node(emac_match, np) == NULL)
+                       continue;
+               if (of_get_property(np, "unused", NULL))
+                       continue;
+               idx = of_get_property(np, "cell-index", NULL);
+               if (idx == NULL)
+                       continue;
+               cell_indices[i] = *idx;
+               emac_boot_list[i++] = of_node_get(np);
+               if (i >= EMAC_BOOT_LIST_SIZE) {
+                       of_node_put(np);
+                       break;
+               }
+       }
+       max = i;
+
+       /* Bubble sort them (doh, what a creative algorithm :-) */
+       for (i = 0; max > 1 && (i < (max - 1)); i++)
+               for (j = i; j < max; j++) {
+                       if (cell_indices[i] > cell_indices[j]) {
+                               np = emac_boot_list[i];
+                               emac_boot_list[i] = emac_boot_list[j];
+                               emac_boot_list[j] = np;
+                               k = cell_indices[i];
+                               cell_indices[i] = cell_indices[j];
+                               cell_indices[j] = k;
+                       }
+               }
+}
+
+static int __init emac_init(void)
+{
+       int rc;
+
+       printk(KERN_INFO DRV_DESC ", version " DRV_VERSION "\n");
+
+       /* Init debug stuff */
+       emac_init_debug();
+
+       /* Build EMAC boot list */
+       emac_make_bootlist();
+
+       /* Init submodules */
+       rc = mal_init();
+       if (rc)
+               goto err;
+       rc = zmii_init();
+       if (rc)
+               goto err_mal;
+       rc = rgmii_init();
+       if (rc)
+               goto err_zmii;
+       rc = tah_init();
+       if (rc)
+               goto err_rgmii;
+       rc = platform_driver_register(&emac_driver);
+       if (rc)
+               goto err_tah;
+
+       return 0;
+
+ err_tah:
+       tah_exit();
+ err_rgmii:
+       rgmii_exit();
+ err_zmii:
+       zmii_exit();
+ err_mal:
+       mal_exit();
+ err:
+       return rc;
+}
+
+static void __exit emac_exit(void)
+{
+       int i;
+
+       platform_driver_unregister(&emac_driver);
+
+       tah_exit();
+       rgmii_exit();
+       zmii_exit();
+       mal_exit();
+       emac_fini_debug();
+
+       /* Destroy EMAC boot list */
+       for (i = 0; i < EMAC_BOOT_LIST_SIZE; i++)
+               of_node_put(emac_boot_list[i]);
+}
+
+module_init(emac_init);
+module_exit(emac_exit);