Device tree aware EMAC driver

Based on BenH's earlier work, this is a new version of the EMAC driver
for the built-in ethernet found on PowerPC 4xx embedded CPUs.  The
same ASIC is also found in the Axon bridge chip.  This new version is
designed to work in the arch/powerpc tree, using the device tree to
probe the device, rather than the old and ugly arch/ppc OCP layer.

This driver is designed to sit alongside the old driver (that lies in
drivers/net/ibm_emac and this one in drivers/net/ibm_newemac).  The
old driver is left in place to support arch/ppc until arch/ppc itself
reaches its final demise (not too long now, with luck).

This driver still has a number of things that could do with cleaning
up, but I think they can be fixed up after merging.  Specifically:
	- Should be adjusted to properly use the dma mapping API.
Axon needs this.
	- Probe logic needs reworking, in conjuction with the general
probing code for of_platform devices.  The dependencies here between
EMAC, MAL, ZMII etc. make this complicated.  At present, it usually
works, because we initialize and register the sub-drivers before the
EMAC driver itself, and (being in driver code) runs after the devices
themselves have been instantiated from the device tree.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

20 files changed, 6507 insertions, 69 deletions

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 9dc4a80a6ac..cfa97f3fbeb 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -1247,75 +1247,8 @@ config IBMVETH
 	  <file:Documentation/networking/net-modules.txt>. The module will
 	  be called ibmveth.
 
-config IBM_EMAC
-	tristate "PowerPC 4xx on-chip Ethernet support"
-	depends on 4xx && !PPC_MERGE
-	help
-	  This driver supports the PowerPC 4xx EMAC family of on-chip
-          Ethernet controllers.
-
-config IBM_EMAC_RXB
-	int "Number of receive buffers"
-	depends on IBM_EMAC
-	default "128"
-
-config IBM_EMAC_TXB
-	int "Number of transmit buffers"
-	depends on IBM_EMAC
-	default "64"
-
-config IBM_EMAC_POLL_WEIGHT
-	int "MAL NAPI polling weight"
-	depends on IBM_EMAC
-	default "32"
-
-config IBM_EMAC_RX_COPY_THRESHOLD
-	int "RX skb copy threshold (bytes)"
-	depends on IBM_EMAC
-	default "256"
-
-config IBM_EMAC_RX_SKB_HEADROOM
-	int "Additional RX skb headroom (bytes)"
-	depends on IBM_EMAC
-	default "0"
-	help
-	  Additional receive skb headroom. Note, that driver
-	  will always reserve at least 2 bytes to make IP header
-	  aligned, so usually there is no need to add any additional
-	  headroom.
-	  
-	  If unsure, set to 0.
-
-config IBM_EMAC_PHY_RX_CLK_FIX
-	bool "PHY Rx clock workaround"
-	depends on IBM_EMAC && (405EP || 440GX || 440EP || 440GR)
-	help
-	  Enable this if EMAC attached to a PHY which doesn't generate
-	  RX clock if there is no link, if this is the case, you will 
-	  see "TX disable timeout" or "RX disable timeout" in the system
-	  log.
-	  
-	  If unsure, say N.
-
-config IBM_EMAC_DEBUG
-	bool "Debugging"
-	depends on IBM_EMAC
-	default n
-
-config IBM_EMAC_ZMII
-	bool
-	depends on IBM_EMAC && (NP405H || NP405L || 44x)
-	default y
-
-config IBM_EMAC_RGMII
-	bool
-	depends on IBM_EMAC && 440GX
-	default y
-		
-config IBM_EMAC_TAH
-	bool
-	depends on IBM_EMAC && 440GX
-	default y
+source "drivers/net/ibm_emac/Kconfig"
+source "drivers/net/ibm_newemac/Kconfig"
 
 config NET_PCI
 	bool "EISA, VLB, PCI and on board controllers"
diff --git a/drivers/net/Makefile b/drivers/net/Makefile
index 2ab33e8b915..9cbef582987 100644
--- a/drivers/net/Makefile
+++ b/drivers/net/Makefile
@@ -5,6 +5,7 @@
 obj-$(CONFIG_E1000) += e1000/
 obj-$(CONFIG_E1000E) += e1000e/
 obj-$(CONFIG_IBM_EMAC) += ibm_emac/
+obj-$(CONFIG_IBM_NEW_EMAC) += ibm_newemac/
 obj-$(CONFIG_IXGBE) += ixgbe/
 obj-$(CONFIG_IXGB) += ixgb/
 obj-$(CONFIG_IP1000) += ipg.o
diff --git a/drivers/net/ibm_emac/Kconfig b/drivers/net/ibm_emac/Kconfig
new file mode 100644
index 00000000000..f61c48047dc
--- /dev/null
+++ b/drivers/net/ibm_emac/Kconfig
@@ -0,0 +1,70 @@
+config IBM_EMAC
+	tristate "PowerPC 4xx on-chip Ethernet support"
+	depends on 4xx && !PPC_MERGE
+	help
+	  This driver supports the PowerPC 4xx EMAC family of on-chip
+          Ethernet controllers.
+
+config IBM_EMAC_RXB
+	int "Number of receive buffers"
+	depends on IBM_EMAC
+	default "128"
+
+config IBM_EMAC_TXB
+	int "Number of transmit buffers"
+	depends on IBM_EMAC
+	default "64"
+
+config IBM_EMAC_POLL_WEIGHT
+	int "MAL NAPI polling weight"
+	depends on IBM_EMAC
+	default "32"
+
+config IBM_EMAC_RX_COPY_THRESHOLD
+	int "RX skb copy threshold (bytes)"
+	depends on IBM_EMAC
+	default "256"
+
+config IBM_EMAC_RX_SKB_HEADROOM
+	int "Additional RX skb headroom (bytes)"
+	depends on IBM_EMAC
+	default "0"
+	help
+	  Additional receive skb headroom. Note, that driver
+	  will always reserve at least 2 bytes to make IP header
+	  aligned, so usually there is no need to add any additional
+	  headroom.
+
+	  If unsure, set to 0.
+
+config IBM_EMAC_PHY_RX_CLK_FIX
+	bool "PHY Rx clock workaround"
+	depends on IBM_EMAC && (405EP || 440GX || 440EP || 440GR)
+	help
+	  Enable this if EMAC attached to a PHY which doesn't generate
+	  RX clock if there is no link, if this is the case, you will
+	  see "TX disable timeout" or "RX disable timeout" in the system
+	  log.
+
+	  If unsure, say N.
+
+config IBM_EMAC_DEBUG
+	bool "Debugging"
+	depends on IBM_EMAC
+	default n
+
+config IBM_EMAC_ZMII
+	bool
+	depends on IBM_EMAC && (NP405H || NP405L || 44x)
+	default y
+
+config IBM_EMAC_RGMII
+	bool
+	depends on IBM_EMAC && 440GX
+	default y
+
+config IBM_EMAC_TAH
+	bool
+	depends on IBM_EMAC && 440GX
+	default y
+
diff --git a/drivers/net/ibm_newemac/Kconfig b/drivers/net/ibm_newemac/Kconfig
new file mode 100644
index 00000000000..0d3e7380bad
--- /dev/null
+++ b/drivers/net/ibm_newemac/Kconfig
@@ -0,0 +1,63 @@
+config IBM_NEW_EMAC
+	tristate "IBM EMAC Ethernet support"
+	depends on PPC_DCR && PPC_MERGE
+	help
+	  This driver supports the IBM EMAC family of Ethernet controllers
+	  typically found on 4xx embedded PowerPC chips, but also on the
+	  Axon southbridge for Cell.
+
+config IBM_NEW_EMAC_RXB
+	int "Number of receive buffers"
+	depends on IBM_NEW_EMAC
+	default "128"
+
+config IBM_NEW_EMAC_TXB
+	int "Number of transmit buffers"
+	depends on IBM_NEW_EMAC
+	default "64"
+
+config IBM_NEW_EMAC_POLL_WEIGHT
+	int "MAL NAPI polling weight"
+	depends on IBM_NEW_EMAC
+	default "32"
+
+config IBM_NEW_EMAC_RX_COPY_THRESHOLD
+	int "RX skb copy threshold (bytes)"
+	depends on IBM_NEW_EMAC
+	default "256"
+
+config IBM_NEW_EMAC_RX_SKB_HEADROOM
+	int "Additional RX skb headroom (bytes)"
+	depends on IBM_NEW_EMAC
+	default "0"
+	help
+	  Additional receive skb headroom. Note, that driver
+	  will always reserve at least 2 bytes to make IP header
+	  aligned, so usually there is no need to add any additional
+	  headroom.
+
+	  If unsure, set to 0.
+
+config IBM_NEW_EMAC_DEBUG
+	bool "Debugging"
+	depends on IBM_NEW_EMAC
+	default n
+
+# The options below has to be select'ed by the respective
+# processor types or platforms
+
+config IBM_NEW_EMAC_ZMII
+	bool
+	default n
+
+config IBM_NEW_EMAC_RGMII
+	bool
+	default n
+
+config IBM_NEW_EMAC_TAH
+	bool
+	default n
+
+config IBM_NEW_EMAC_EMAC4
+	bool
+	default n
diff --git a/drivers/net/ibm_newemac/Makefile b/drivers/net/ibm_newemac/Makefile
new file mode 100644
index 00000000000..0b5c9951276
--- /dev/null
+++ b/drivers/net/ibm_newemac/Makefile
@@ -0,0 +1,11 @@
+#
+# Makefile for the PowerPC 4xx on-chip ethernet driver
+#
+
+obj-$(CONFIG_IBM_NEW_EMAC) += ibm_newemac.o
+
+ibm_newemac-y := mal.o core.o phy.o
+ibm_newemac-$(CONFIG_IBM_NEW_EMAC_ZMII) += zmii.o
+ibm_newemac-$(CONFIG_IBM_NEW_EMAC_RGMII) += rgmii.o
+ibm_newemac-$(CONFIG_IBM_NEW_EMAC_TAH) += tah.o
+ibm_newemac-$(CONFIG_IBM_NEW_EMAC_DEBUG) += debug.o
diff --git a/drivers/net/ibm_newemac/core.c b/drivers/net/ibm_newemac/core.c
new file mode 100644
index 00000000000..653bfdc291e
--- /dev/null
+++ b/drivers/net/ibm_newemac/core.c
@@ -0,0 +1,2907 @@
+/*
+ * drivers/net/ibm_newemac/core.c
+ *
+ * Driver for PowerPC 4xx on-chip ethernet controller.
+ *
+ * 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/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 <asm/processor.h>
+#include <asm/io.h>
+#include <asm/dma.h>
+#include <asm/uaccess.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");
+
+/*
+ * PPC64 doesn't (yet) have a cacheable_memcpy
+ */
+#ifdef CONFIG_PPC64
+#define cacheable_memcpy(d,s,n) memcpy((d),(s),(n))
+#endif
+
+/* 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_NEW_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 (net_ratelimit())
+		printk(KERN_ERR "%s: %s\n", dev->ndev->name, error);
+}
+
+/* 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
+
+/* 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_TBI ||
+		phy_mode == PHY_MODE_RTBI;
+}
+
+static inline int emac_phy_gpcs(int phy_mode)
+{
+	return  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);
+	dev->no_mcast = 1;
+	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);
+	dev->no_mcast = 0;
+	if (dev->mcast_pending && netif_running(dev->ndev))
+		__emac_set_multicast_list(dev);
+	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);
+	}
+
+	out_be32(&p->mr0, EMAC_MR0_SRST);
+	while ((in_be32(&p->mr0) & EMAC_MR0_SRST) && n)
+		--n;
+
+	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)
+{
+	struct emac_regs __iomem *p = dev->emacp;
+	u16 gaht[4] = { 0 };
+	struct dev_mc_list *dmi;
+
+	DBG(dev, "hash_mc %d" NL, dev->ndev->mc_count);
+
+	for (dmi = dev->ndev->mc_list; dmi; dmi = dmi->next) {
+		int bit;
+		DBG2(dev, "mc %02x:%02x:%02x:%02x:%02x:%02x" NL,
+		     dmi->dmi_addr[0], dmi->dmi_addr[1], dmi->dmi_addr[2],
+		     dmi->dmi_addr[3], dmi->dmi_addr[4], dmi->dmi_addr[5]);
+
+		bit = 63 - (ether_crc(ETH_ALEN, dmi->dmi_addr) >> 26);
+		gaht[bit >> 4] |= 0x8000 >> (bit & 0x0f);
+	}
+	out_be32(&p->gaht1, gaht[0]);
+	out_be32(&p->gaht2, gaht[1]);
+	out_be32(&p->gaht3, gaht[2]);
+	out_be32(&p->gaht4, gaht[3]);
+}
+
+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 || ndev->mc_count > 32)
+		r |= EMAC_RMR_PMME;
+	else if (ndev->mc_count > 0)
+		r |= EMAC_RMR_MAE;
+
+	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 Rx 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);
+
+	DBG2(dev, "__emac4_calc_base_mr1" NL);
+
+	switch(tx_size) {
+	case 4096:
+		ret |= EMAC4_MR1_TFS_4K;
+		break;
+	case 2048:
+		ret |= EMAC4_MR1_TFS_2K;
+		break;
+	default:
+		printk(KERN_WARNING "%s: Unknown Rx 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;
+	u32 r, mr1 = 0;
+
+	DBG(dev, "configure" NL);
+
+	if (emac_reset(dev) < 0)
+		return -ETIMEDOUT;
+
+	if (emac_has_feature(dev, EMAC_FTR_HAS_TAH))
+		tah_reset(dev->tah_dev);
+
+	DBG(dev, " duplex = %d, pause = %d, asym_pause = %d\n",
+	    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;
+
+	/* Check for full duplex */
+	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.address);
+
+			/* Put some arbitrary OUI, Manuf & Rev IDs so we can
+			 * identify this GPCS PHY later.
+			 */
+			out_be32(&p->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) {
+			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))
+		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);
+	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 = 10;
+	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;